Nucleic Acids Research

Pezo, Valérie;Martinez, Miguel Angel;Wain-Hobson, Simon

doi: 10.1093/nar/24.2.253pmid: 8628647

Abstract RNA hypermutagenesis results from cDNA synthesis in the presence of highly biased dNTP precursor concentrations and preferentially exploits human immunodeficiency virus type 1 (HIV-1) reverse transcriptase. Such reaction conditions slow down DNA synthesis, which might be conducive to strand transfer and deletion. This has been investigated. A 6 bp inverted repeat nested between 10 bp repeats was efficiently deleted at dCTP concentrations typically used. Inter- or intramolecular strand transfer between 10 bp repeated sequences separated by runs of templated G residues occurred, but at lower concentrations. If RNA hypermutagenesis of a sequence containing direct and inverted repeats is unavoidable, avian myeloblastosis virus (AMV) reverse transcriptase could be used, as strand transfer occurs with much diminished dCTP substrate dependence. Introduction Retroviral replication is an error prone process. Apart from the familiar point substitutions, frame-shift mutations and intramolecular deletions and insertions it is becoming increasingly apparent that complex mutations involving multiple strand transfer may also occur. The notorious examples involve proto-oncogene transduction, while others, less known but more frequent, may result in deletions with insertions and multiple insertions ( 1–8 ). In a single cycle reverse transcription assay deletions occurred frequently, but not exclusively, between short tracts of sequence identity ( 3 , 5 ). Inverted repeats were eliminated particularly efficiently ( 4 ). To this catalogue of replication errors must be added G→A hypermutation, which results from cDNA synthesis in the presence of highly biased intracellular dNTP pools ( 2 , 2–14 ). G→A hypermutation may be reproduced in a simple in vitro reaction using RNA, reverse transcriptase and highly biased [dCTP]/[dTTP] ratios ( 15 , 16 ) and forms the basis of a powerful method for in vitro protein evolution ( 17 ). Among the plethora of hypermutants sequenced ( 15 , 16 ) a few deletions were noted, most of which occurred within small runs of 2–3 bp homologous sequences. In the present study the in vitro hypermutation reaction has been adapted to specifically study the influence of dNTP pool imbalances on nascent DNA strand transfer. Materials and Methods The four RNA templates used are shown in Figure 1 . In all cases a 10 bp repeated sequence (DR1 and DR2) was separated by 1, 4 or 7 G residues or a GC inverted repeat. Single-stranded DNA oligonucleotides (72G1, 75G4, 78G7 and 81GC) were synthesized and rendered double-stranded by 20 rounds of PCR using the 19RT and 23R PCR primer pair and cloned into the pBluescript SK+ vector via Kpn I and Sac I restriction sites. Recombinants were verified by sequencing. RNA was made using T7 RNA polymerase and quantified as described ( 15 , 16 ). cDNA was synthesized using 0.5 pmol RNA with a 30-fold molar excess of human immunodeficiency virus type 1 (HIV-1) reverse transcriptase (RT) (5 U/reaction, ∼12–15 pmol; Boehringer), the 19RT primer and biased [dCTP]/[dTTP] ratios. Avian myeloblastosis virus (AMV) (Promega) and Moloney murine leukemia virus (MoMLV) (Gibco) RTs were also used at 5 U/reaction. After reverse transcription the cDNA was incubated with 1 µg DNase-free RNase (Boehringer) and 0.5 U RNase H (Boehringer) and concentrated on a Centricon 30 (Amicon). In order to recover sufficient material for analysis (template ∼0.5 pmol) the cDNA was PCR amplified under optimized conditions (2.5 mM Mg 2+ ) using the 19RT and 23R primer pair ( Fig. 1 ) and Taq polymerase (Roche). The amplification profile was classic apart from three initial cycles with an annealing temperature of 37°C followed by 12 cycles at 55°C. This was done to facilitate annealing of the 23R primer to hypermutated cDNA. PCR products were precipitated and 5′-end-labelled with [γ- 32 P]ATP and T4 polynucleotide kinase. Half of the reaction was electrophoresed through a 6% non-denaturing polyacrylamide gel, fixed in 10% acetic acid and dried. Band intensities were quantified using a Molecular Dynamics phosphorimager. PCR products were purified from a 10% native polyacrylamide gel, digested with Eco RI and Xba I and cloned into M13mp18 RF DNA. The template sequences were designed such that cloned undeleted PCR-amplified products from all four different templated reactions would yield a blue phenotype on X-gal plates, while frame-shift mutations and other deletions, including deletion between the direct repeats, would yield a white phenotype. Sequencing reactions were carried out with the Sequenase 2.0 kit (USB). Figure 1 View largeDownload slide Oligonucleotide sequences used in the in vitro hypermutagenesis assay. The sequences common to the four oligonucleotides are shown above while the differences are given underneath. Those used to produce double-stranded DNA and to amplify cDNA after reverse transcription, 19RT and 23R, are also given. 19RT was also used as the reverse transcription reaction primer. Kpn I and Sac I sites were used to clone into pBluescript SK+ while Xba I and Eco RI were used to clone the products into M13mp18. The template sequence was designed such that full-length and Xba I- and Eco RI-cleaved products from 72G1, 75G4, 78G7 and 81GC templated reactions would yield a blue phenotype on X-gal plates, while those resulting from deletion between the direct repeats would yield a white phenotype. The reading frame is indicated by dots under the first base of the codon. Figure 1 View largeDownload slide Oligonucleotide sequences used in the in vitro hypermutagenesis assay. The sequences common to the four oligonucleotides are shown above while the differences are given underneath. Those used to produce double-stranded DNA and to amplify cDNA after reverse transcription, 19RT and 23R, are also given. 19RT was also used as the reverse transcription reaction primer. Kpn I and Sac I sites were used to clone into pBluescript SK+ while Xba I and Eco RI were used to clone the products into M13mp18. The template sequence was designed such that full-length and Xba I- and Eco RI-cleaved products from 72G1, 75G4, 78G7 and 81GC templated reactions would yield a blue phenotype on X-gal plates, while those resulting from deletion between the direct repeats would yield a white phenotype. The reading frame is indicated by dots under the first base of the codon. Results The sequences were designed so as to allow reverse transcription to proceed to the end of DR1 unhindered by a dearth of dCTP. Depending upon the dCTP concentration the run of 1, 4 or 7 templated G residues would impose a pause in reverse transcription, probably allowing dissociation of the RT/template-primer complex. Fraying at the 3′-end of the primer would promote either intra- or interstrand transfer to DR2, offering the possibility of unimpeded reverse transcription up to the Eco RI site. Following PCR the proportion of deleted templates was quantitated by 5′-labelling of the extremities. The raw data for the HIV-1 RT hypermutagenesis reactions using three RNA templates (75G4, 78G7 and 81GC) are shown in Figure 2 . In all cases the proportion of the deleted form increased with decreasing dCTP concentration. Band intensities were quantified by phosphorimager, the proportion of deleted products being given in Figure 3 A. Fifty percent of the template with the inverted repeat (81GC) was deleted at a dCTP substrate concentration of ∼10 nM, with ∼3 nM for 78G7 and <0.1 nM for 75G4. Reverse transcription of template 72G1 did not result in any deletion whatsoever. The nature of the deletions were verified by sequencing cloned PCR products. In all cases they occurred precisely between DR1 and DR2. That no deletions were found at high dCTP concentrations rules out any contribution from PCR ( Figs 2 and 3 ). Efficient deletion of the inverted repeat was particularly worrying given that the 10–30 nM dCTP range is typical of most G→A hypermutagenesis reactions ( 15 , 16 ). As the sensitivity of different RTs to [dCTP]/[dTTP] changes is HIV-1 > AMV > MoMLV respectively ( 16 ), use of either of the latter two enzymes in the hypermutagenesis reaction might translate into less efficient strand transfer. Accordingly, reverse transcription of the same four templates was analysed under identical conditions but with the RTs from AMV ( Fig. 3 B) and MoMLV ( Fig. 3 C). For both RTs strand transfer and deletion between DR1 and DR2 occurred more frequently with the 81GC and 78G7 as opposed to the 75G4 and 72G1 templates. Perhaps the AMV enzyme could accommodate strand transfer a little better than the MoMLV RT, but given the experimental variation (±10%) the differences are probably not significant. For both enzymes the proportion of deleted product was less than for the corresponding reactions with HIV-1 RT. Approximately 8–10% of product remained undeleted following reverse transcription of the 81GC and 78G7 templates with HIV-1 RT, even at very low dCTP concentrations ( Fig. 3 A). In fact, dCTP concentrations below 1 µM are only nominal, being supplemented by contaminating dCTP from the three other dNTPs, particularly dTTP at 440 µM (data not shown). This explains why the proportion of deleted product never attained 100% ( Fig. 3 A). In order to link the biased [dCTP]/[dTTP] ratio to both strand transfer and G→A hypermutation some of the undeleted PCR material was cloned and sequenced. Only these full-length products encoded templated G residues and therefore could reveal the presence of G→A transitions. Sequences derived from blue recombinants (cloned undeleted 78G7, 10 pM dCTP reaction products using HIV-1 RT; Fig. 3 A) are summarized in Figure 4 . Those in Figure 4 A represent 39 clones encoding single point mutations. The distribution over the seven sites was non-random (χ 2 = 13.8, 6 degrees of freedom, P < 0.05), with positions G1 and G3 particularly being substituted. A further 18 recombinants derived from the same reaction encoded multiple G→A substitutions ( Fig. 4 B). Among these clones most of the G→A transitions occurred at G1, G6 and G7, again giving rise to a non-random distribution (χ 2 = 41.0, 6 degrees of freedom, P < 0.001). Two clones encoded a U→A substitution at position U8. This probably arose by a −1 dislocation of the primer strand with respect to the template between G5 and G7, misincorporation of a T opposite G5, G6 or G7, followed by relocation of the primer strand and subsequent elongation beyond the mismatched U-T pair. The multiply substituted clones clearly showed signs of increasing dCTP depletion as both the frequency of G→A substitutions, as well as those resulting from −1 nascent strand dislocation, increased towards the end of the run of seven G residues. This is best shown in Figure 4 C, which describes the probability of a substitution by A at i + 1 given a G→A transition at position i . Figure 2 View largeDownload slide Raw phosphorimager data for kinased PCR products following gel electrophoresis. All reactions were carried out using HIV-1 RT. ( A ) Reactions using RNA template 75G4. Lane p indicates reaction products resulting from cDNA synthesis with physiological [dCTP], i.e. 10 µM, while the dCTP concentration for reactions in lanes 1–5 was 10, 1 and 0.1 nM and 10 and 1 pM respectively. Lane m indicates the mock reaction products, i.e. without HIV-1 RT. After 15 cycles of PCR there was no signal, indicating that DNA contamination of the RNA template was negligible. ( B ) Reactions using RNA template 78G7. Lanes p and m are as above. The dCTP concentration for reactions in lanes 1–9 was 10, 8, 6, 5, 3, 1 and 0.1 nM and 10 and 1 pM respectively. ( C ) Reactions using RNA template 81GC. Lanes p and m are as above. The dCTP concentration for reactions in lanes 1–10 was 100, 50, 10, 8, 5, 3, 1 and 0.1 nM and 10 and 1pM respectively. [dATP] and [dGTP] were constant at physiological values ( 18 ), 40 and 20 µM respectively, while [dTTP] was kept at 10× the physiological value, i.e. 440 µM ( 18 ). Figure 2 View largeDownload slide Raw phosphorimager data for kinased PCR products following gel electrophoresis. All reactions were carried out using HIV-1 RT. ( A ) Reactions using RNA template 75G4. Lane p indicates reaction products resulting from cDNA synthesis with physiological [dCTP], i.e. 10 µM, while the dCTP concentration for reactions in lanes 1–5 was 10, 1 and 0.1 nM and 10 and 1 pM respectively. Lane m indicates the mock reaction products, i.e. without HIV-1 RT. After 15 cycles of PCR there was no signal, indicating that DNA contamination of the RNA template was negligible. ( B ) Reactions using RNA template 78G7. Lanes p and m are as above. The dCTP concentration for reactions in lanes 1–9 was 10, 8, 6, 5, 3, 1 and 0.1 nM and 10 and 1 pM respectively. ( C ) Reactions using RNA template 81GC. Lanes p and m are as above. The dCTP concentration for reactions in lanes 1–10 was 100, 50, 10, 8, 5, 3, 1 and 0.1 nM and 10 and 1pM respectively. [dATP] and [dGTP] were constant at physiological values ( 18 ), 40 and 20 µM respectively, while [dTTP] was kept at 10× the physiological value, i.e. 440 µM ( 18 ). A small number of white, rather than blue, plaques were found upon cloning the undeleted 78G7 DNA. Fourteen encoded a deletion of a single G within the run of seven G residues. As it is impossible to know where the deletion occurred and so a gap has been arbitrarily introduced at position G7 ( Fig. 4 D). Twelve clones carried one or two G→A substitutions in addition to deletion of a single G. Discussion The data illustrate that both G→A hypermutation and strand transfer may result from a common denominator, [dTTP]/[dCTP] biases. Presumably a pause in polymerization starting at the run of templated G residues resulting from the dearth of dCTP allows recycling of the RT and occasionally fraying of the template-nascent strand and either intra- or interstrand transfer between DR1 and DR2. Figure 3 View largeDownload slide Quantification of strand transfer between DR1 and DR2 by phosphorimager. ( A-C ) Reactions using HIV-1, AMV and MoMLV RTs respectively. Closed circles represent data for the 81GC template; closed squares, the 78G7 template; open squares the 75G4 template; open circles, the 72G1 template. The proportion of deleted product was calculated as the band intensity of the deleted product divided by the sum of the band intensities of the deleted and undeleted bands. The data for the HIV-1 reactions represent the mean of three reactions, while those for AMV and MoMLV are the means of two reactions. The error was ±10%. Figure 3 View largeDownload slide Quantification of strand transfer between DR1 and DR2 by phosphorimager. ( A-C ) Reactions using HIV-1, AMV and MoMLV RTs respectively. Closed circles represent data for the 81GC template; closed squares, the 78G7 template; open squares the 75G4 template; open circles, the 72G1 template. The proportion of deleted product was calculated as the band intensity of the deleted product divided by the sum of the band intensities of the deleted and undeleted bands. The data for the HIV-1 reactions represent the mean of three reactions, while those for AMV and MoMLV are the means of two reactions. The error was ±10%. HIV-1 RT can apparently continue reverse transcription after strand transfer more easily than either the AMV or MoMLV RTs, while the differences between the latter two enzymes are minimal. However HIV-1 RT is the enzyme of choice for hypermutagenesis, as it demonstrates the greatest sensitivity to dNTP biases. Thus during G→A hypermutagenesis reactions using this enzyme care must be exercised when using depleted dCTP concentrations, particularly below 10 nM. This is especially important if the target sequence harbours an inverted repeat. Above this concentration strand transfer, either intra- or intermolecular, between 10 bp direct repeats is tolerable. However, if a sequence proved particularly prone to internal deletions during HIV-1 RNA hypermutagenesis the AMV enzyme could be substituted, although the extent of G→A hypermutation would be reduced by a factor of two to three ( 16 ). The MoMLV enzyme is to be avoided, for it shows much reduced dCTP-dependent G→A hypermutation ( 16 ). Figure 4 View largeDownload slide Distribution of substitutions in undeleted cDNA products resulting from HIV-1 reverse transcription of the 78G7 template. In order to recover sufficient material cDNA from the 10 pM dCTP sample ( Fig. 2 B) was PCR amplified using 40 cycles and cloned into an M13mp18 vector. ( A ) Distribution of single base substitutions. ( B ) Distribution of multiple base substitutions. ( C ) Frequency (f) of substitution by A in the i + 1 position for G→A transitions of the i th site. ( D ) Sequences of single −1 frame-shift mutations. The arrows indicates the sense of cDNA synthesis. Undeleted products from the 72G1 and 75G4 templated 1 pM reactions using HIV-1 RT were cloned and sequenced. Of the 72G1 derived clones seven of eight encoded G→A substitutions, while of the seven 75G4 derived clones, six were mutated (one AGGG, three GGGA, one AGGA and one AAGA). PCR was not responsible for deletions between DR1 and DR2 (see Fig. 1 at high dCTP concentrations). Forty cycles of PCR were carried out on the mock reaction from the reverse transcription assay. The PCR products were cloned into M13mp18 and 20 recombinants sequenced. No base substitutions were found, indicating that all the substitutions noted above resulted from reverse transcription. Figure 4 View largeDownload slide Distribution of substitutions in undeleted cDNA products resulting from HIV-1 reverse transcription of the 78G7 template. In order to recover sufficient material cDNA from the 10 pM dCTP sample ( Fig. 2 B) was PCR amplified using 40 cycles and cloned into an M13mp18 vector. ( A ) Distribution of single base substitutions. ( B ) Distribution of multiple base substitutions. ( C ) Frequency (f) of substitution by A in the i + 1 position for G→A transitions of the i th site. ( D ) Sequences of single −1 frame-shift mutations. The arrows indicates the sense of cDNA synthesis. Undeleted products from the 72G1 and 75G4 templated 1 pM reactions using HIV-1 RT were cloned and sequenced. Of the 72G1 derived clones seven of eight encoded G→A substitutions, while of the seven 75G4 derived clones, six were mutated (one AGGG, three GGGA, one AGGA and one AAGA). PCR was not responsible for deletions between DR1 and DR2 (see Fig. 1 at high dCTP concentrations). Forty cycles of PCR were carried out on the mock reaction from the reverse transcription assay. The PCR products were cloned into M13mp18 and 20 recombinants sequenced. No base substitutions were found, indicating that all the substitutions noted above resulted from reverse transcription. Acknowledgements We would like to thank Henri Buc, Andreas Meyerhans, Jean-Pierre Vartanian and Lynn Ripley for helpful discussions. VP was supported by le Ministère de la Recherche et de l'Enseignement Supérieur and MAM by a post-doctoral fellowship from the European Union. This work was supported by grants from the Institut Pasteur and l'Agence Nationale pour la Recherche sur le SIDA (ANRS). References 1 Monk R.J.,  Malik F.G.,  Stokes D.,  Evans L.H.. ,  J. Virol. ,  1992, vol.  66 (pg.  3683- 3689) PubMed  2 Pathak V.K.,  Temin H.M.. ,  Proc. Natl. Acad. Sci. USA ,  1990, vol.  87 (pg.  6019- 6023) CrossRef Search ADS   3 Pathak V.K.,  Temin H.M.. ,  Proc. Natl. Acad. Sci. USA ,  1990, vol.  87 (pg.  6024- 6028) CrossRef Search ADS   4 Pathak V.K.,  Temin H.M.. ,  J. Virol. ,  1992, vol.  66 (pg.  3093- 3100) PubMed  5 Pusinelli G.A.,  Temin H.M.. ,  J. Virol. ,  1991, vol.  65 (pg.  4786- 4797) PubMed  6 Varela-Echavarria A.,  Garvey N.,  Preston B.D.,  Dougherty J.P.. ,  J. Biol. Chem ,  1992, vol.  267 (pg.  24681- 24688) PubMed  7 Mansky L.L.,  Temin H.M.. ,  J. Virol. ,  1994, vol.  68 (pg.  494- 499) PubMed  8 Mansky L.M.,  Temin H.M.. ,  J. Virol. ,  1995, vol.  69 (pg.  5087- 5094) PubMed  9 Vartanian J.P.,  Meyerhans A.,  Åsjö B.,  Wain-Hobson S.. ,  J. Virol. ,  1991, vol.  65 (pg.  1779- 1788) PubMed  10 Vartanian J.P.,  Meyerhans A.,  Sala M.,  Wain-Hobson S.. ,  Proc. Natl. Acad. Sci. USA ,  1994, vol.  91 (pg.  3092- 3096) CrossRef Search ADS   11 Johnson P.R.,  Hamm T.E.,  Goldstein S.,  Kitov S.,  Hirsch V.M.. ,  Virology ,  1991, vol.  185 (pg.  217- 228) CrossRef Search ADS PubMed  12 Gao F.,  Yue L.,  White A.T.,  Pappas P.G.,  Barchue J.,  Hanson A.P.,  Greene B.M.,  Sharp P.M.,  Shaw G.M.,  Hahn B.H.. ,  Nature (Lond.) ,  1992, vol.  358 (pg.  495- 499) CrossRef Search ADS   13 Perry S.T.,  Flaherty M.T.,  Kelley M.J.,  Clabough D.L.,  Tronick S.R.,  Coggins L.,  Whetter L.,  Lengel C.R.,  Fuller F.. ,  J. Virol. ,  1992, vol.  66 (pg.  4085- 4097) PubMed  14 Wain-Hobson S.,  Sonigo P.,  Guyader M.,  Gazit A.,  Henry M.. ,  Virology ,  1995, vol.  209 (pg.  297- 303) CrossRef Search ADS PubMed  15 Martinez M.A.,  Vartanian J.P.,  Wain-Hobson S.. ,  Proc. Natl. Acad. Sci. USA ,  1994, vol.  91 (pg.  11787- 11791) CrossRef Search ADS   16 Martinez M.,  Sala M.,  Vartanian J.P.,  Wain-Hobson S.. ,  Nucleic Acids Res. ,  1995, vol.  23 (pg.  2573- 2578) CrossRef Search ADS PubMed  17 Martinez M.A.,  Pezo V.,  Marliere P.,  Wain-Hobson S.. ,  EMBO J ,  1996  in press 18 Dahbo Y.,  Eriksson S.. ,  Eur. J. Biochem ,  1985, vol.  150 (pg.  429- 434) CrossRef Search ADS PubMed  © 1996 Oxford University Press

Pezo, Valérie; Martinez, Miguel Angel; Wain-Hobson, Simon

doi: N/Apmid: N/A

RNA hypermutagenesis results from cDNA synthesis in the presence of highly biased dNTP precursor concentrations and preferentially exploits human immunodeficiency virus type 1 (HIV-1) reverse transcriptase. Such reaction conditions slow down DNA synthesis, which might be conducive to strand transfer and deletion. This has been investigated. A 6 bp inverted repeat nested between 10 bp repeats was efficiently deleted at dCTP concentrations typically used. Inter- or intramolecular strand transfer between 10 bp repeated sequences separated by runs of templated G residues occurred, but at lower concentrations. If RNA hypermutagenesis of a sequence containing direct and inverted repeats is unavoidable, avian myeloblastosis virus (AMV) reverse transcriptase could be used, as strand transfer occurs with much diminished dCTP substrate dependence.

Schmitz, Jürgen; Prüfer, Dirk; Rohde, Wolfgang; Tacke, Eckhard

doi: N/Apmid: N/A

The 5′ untranslated leader (Ω sequence) of tobacco mosaic virus (TMV) genomic RNA was utilized as a translational enhancer sequence in expression of the 17 kDa putative movement protein (pr17) of potato leaf roll luteovirus (PLRV). In vitro translation of RNAs transcribed from appropriate chimeric constructs, as well as their expression in transgenic potato plants, resulted in the expected wild-type pr17 protein, as well as in larger translational products recognized by pr17-specific antisera. Mutational analyses revealed that the extra proteins were translated by non-canonical initiation at AUU codons present in the wild-type Ω sequence. In the plant system translation initiated predominantly at the AUU codon at positions 63–65 of the Ω sequence. Additional AUU codons in a different reading frame of the Ω sequence also showed the capacity for efficient translation initiation in vitro. These results extend the previously noted activity of the TMV 5′ leader sequence in ribosome binding and translation enhancement in that the TMV translation enhancer can mediate non-canonical translation initiation in vitro and in vivo.

Schmitz, Jürgen;Prüfer, Dirk;Rohde, Wolfgang;Tacke, Eckhard

doi: 10.1093/nar/24.2.257pmid: 8628648

Abstract The 5′ untranslated leader (Ω sequence) of tobacco mosaic virus (TMV) genomic RNA was utilized as a translational enhancer sequence in expression of the 17 kDa putative movement protein (pr17) of potato leaf roll luteovirus (PLRV). In vitro translation of RNAs transcribed from appropriate chimeric constructs, as well as their expression in transgenic potato plants, resulted in the expected wild-type pr17 protein, as well as in larger translational products recognized by pr17-specific antisera. Mutational analyses revealed that the extra proteins were translated by non-canonical initiation at AUU codons present in the wild-type Ω sequence. In the plant system translation initiated predominantly at the AUU codon at positions 63–65 of the Ω sequence. Additional AUU codons in a different reading frame of the Ω sequence also showed the capacity for efficient translation initiation in vitro . These results extend the previously noted activity of the TMV 5′ leader sequence in ribosome binding and translation enhancement in that the TMV translation enhancer can mediate non-canonical translation initiation in vitro and in vivo . Introduction Translational efficiencies of eukaryotic mRNAs are influenced by various factors, such as primary (5′-cap) and secondary (hairpin) structures, the sequence context of the start codon or upstream regulatory elements, such as enhancer sequences or small upstream open reading frames (uORFs) ( 1–7 ). While for their specific interaction with ribosomes and for start codon recognition prokaryotic mRNAs make use of the Shine-Dalgarno sequence ( 8 ), the lack of a corresponding sequence in eukaryotic mRNAs upstream of the start codon has led to various models for pre-initiation complexes binding to the RNA 5′-end and then scanning along the mRNA for recognition of the translational start codon(s) ( 7 , 9 , 10 ). An additional facet of eukaryotic mRNA translation has come from the identification of 5′ untranslated sequences which largely enhance translation. Such regulatory translational enhancer sequences have been primarily documented to exist in the 5′ leader sequences of RNAs from plant and animal viruses, such as potato virus X, rous sarcoma virus, brome mosaic virus and tobacco mosaic virus (TMV) ( 11 , 12 ). In the case of the TMV translational enhancer (Ω) sequence (consisting of the 5′-terminal 68 nt) it has been proposed that the absence of extended secondary structures in this region causes the increase in translational efficiency ( 13 ). In fact, a detailed analysis of the TMV (strain U1) Ω sequence pointed to the importance of the primary structure by identifying two elements, a direct repeat of 8 nt and a CAA-rich region, as being responsible for translation enhancement ( 14 ). In line with previous observations that the Ω sequence is capable of promoting binding of two ribosome molecules (disome formation) when elongation is blocked in the presence of sparsomycin ( 15 , 16 ), it was proposed that the core regulatory elements of the Ω sequence allow specific binding of a protein factor(s) required for efficient initiation ( 14 ). In the disome complex one of the two ribosomes occupies the AUG start codon of the replicase gene and the second was postulated to bind further upstream in the Ω sequence (TMV strain SPS) at an AUU codon in position 14 (AUU 14 ). Translation initiation at this AUU codon was proposed to occur ( 16 ), but with appropriate chimeric constructs consisting of the TMV (strain U1) translational enhancer in-frame with the AUG start codon of a reporter gene putative initiation at the corresponding AUU codon (AUU 15 ) did not contribute to increased reporter gene activity ( 14 ). Initiation at non-AUG codons was originally proposed from experiments using synthetic oligonucleotides ( 17 ). Furthermore, usage of AUU as a translational start was postulated for human mitochondrial mRNA ( 18 ), but the first evidence for involvement of AUU as a start codon was described for the Escherichia coli gene encoding initiation factor IF3 ( 19 ). Since then further evidence for eukaryotic translation initiation at AUU and other codons has accumulated for animal and plant systems ( 10 , 20–22 ). Here we show that the TMV translational enhancer sequence can promote alternative translation initiation at AUU codons. Materials and Methods Construction of plasmids for plant transformation The potato leaf roll luteovirus (PLRV) pr17 (ORF4) gene was amplified by PCR from clone pCPL1 ( 23 ). The primers were designed to give unique restriction sites for Spe I and Xba I at the 5′- and 3′-ends respectively. A plasmid previously constructed for high level expression of the PLRV capsid protein CP (ORF3) controlled by the Ω sequence and the 35S promoter of cauliflower mosaic virus (CaMV) ( 24 ) was cut with Spe I and Xba I to remove the ORF3 N-terminal sequence. Subsequently the amplified ORF4 fragment was cut with Spe I and Xba I and cloned into the linearized plasmid pRT17/NIV. A Hind III fragment isolated from pRT17/NIV was cloned into the binary vector pBIN19 ( 25 ). Plasmids containing the ORF4 expression cassette were designated p17/NIV. Transformation of Solanum tuberosum and Western blot analysis P17/NIV was transformed into Agrobacterium tumefaciens strain LBA 4404 ( 26 ) and stable transformation of Solanum tuberosum var. Desirée was performed according to published procedures ( 27 ), with the resulting agrobacteria carrying plasmid p17/NIV. Western blot analysis of regenerated plants was carried out as described in Tacke et al. ( 28 ). Plasmid construction for in vitro analysis of the Ω sequence A Hin dIII fragment comprising the Ω sequence and ORF4 was isolated from plasmid p17/NIV and cloned into plasmid pSP64 (Promega) under the control of the SP6 promoter (pS17N). Mutations in the Ω sequence were carried out by PCR using synthetic oligonucleotides (synthesized on a DNA/RNA synthesizer 392; Applied Biosystems, Darmstadt, Germany) and plasmid pS17N as the template. The upstream primer was located 5′ of the SP6 promoter (position 2105 of pSP64), comprising a unique Ssp I restriction site. This oligonucleotide was combined with a set of downstream primers complementary to the Ω sequence and bearing different point mutations and a Ksp I restriction site. Amplified fragments covered the SP6 promoter and the mutated Ω sequence. These PCR fragments were subsequently cut with Ssp I and Ksp I and cloned into pS17N. Plasmids containing mutated forms of the Ω sequence were sequenced on a DNA Sequencer 373A (Applied Biosystems). In vitro transcription/translation All pSP64-based plasmids were linearized downstream of ORF4 with Eco RI prior to in vitro transcription with SP6 RNA polymerase in the presence of the cap analogue m 7 GpppG ( 29 ). RNAs were translated either in a wheat germ extract or rabbit reticulocyte lysate (Amersham Buchler) in the presence of [ 35 S]methionine under conditions recommended by the supplier. In vitro products were analysed on 12.5% SDS-polyacrylamide gels and detected by fluorography ( 30 ). Figure 1 View largeDownload slide Western blot analysis of potato plants transformed with ORF4 from potato leaf roll virus. ( A ) Schematic representation of the expression cassette used for high level expression of ORF4 in planta . Two AUU codons (AUU 15 , AUU 63 ) of the Ω sequence which are located in-frame with the ORF4 start codon are indicated by open boxes. Mutation of AUU 15 and AUU 63 to AUG in constructs p17/NI and p17/NIII are indicated by arrowheads. In construct p17/NIII the first 59 nt of the Ω sequence were deleted to avoid initiation upstream of AUG 63 . Construct p17 without the Ω sequence was taken as a control, indicating the molecular weight of wild-type pr17. ( B ) Western analysis of protein extracts from two independent transgenic potato lines transformed with construct p17/NIV (loaded in lanes indicated with C1), p17/NI (C2) and p17/NIII (C3). In C4 protein extracts of plants expressing ORF4 without the Ω sequence are loaded. Proteins were separated by PAGE and processed for immunological detection as described before ( 28 ). Figure 1 View largeDownload slide Western blot analysis of potato plants transformed with ORF4 from potato leaf roll virus. ( A ) Schematic representation of the expression cassette used for high level expression of ORF4 in planta . Two AUU codons (AUU 15 , AUU 63 ) of the Ω sequence which are located in-frame with the ORF4 start codon are indicated by open boxes. Mutation of AUU 15 and AUU 63 to AUG in constructs p17/NI and p17/NIII are indicated by arrowheads. In construct p17/NIII the first 59 nt of the Ω sequence were deleted to avoid initiation upstream of AUG 63 . Construct p17 without the Ω sequence was taken as a control, indicating the molecular weight of wild-type pr17. ( B ) Western analysis of protein extracts from two independent transgenic potato lines transformed with construct p17/NIV (loaded in lanes indicated with C1), p17/NI (C2) and p17/NIII (C3). In C4 protein extracts of plants expressing ORF4 without the Ω sequence are loaded. Proteins were separated by PAGE and processed for immunological detection as described before ( 28 ). Results Analysis of ORF4 transgenic plants Potato leaf discs were transformed with construct p17/NIV ( Fig. 1 A) and transgenic lines carrying two or more copies of the transgene were recovered. Western blot analysis of extracts from all independent transformants detected the wild-type pr17 and an additional immunoreactive protein (pr17/n) with an apparent molecular weight of 24 kDa ( Fig. 1 B, C1). This pr17/n protein was not detected in PLRV-infected plants ( 28 ) nor in transgenic plants expressing ORF4 without the Ω sequence ( Fig. 1 A and B, C4). Reckoning and sequencing of the transgenes from a potato line containing two transgene copies revealed identical sequences for the transcribed and translated regions (data not shown). Together with the fact that a single copy line established at later stages also showed the same two immunoreactive proteins and that, moreover, transgenic lines expressing ORF4 without translational enhancer did not show the larger immunoreactive protein pr17/n, these data indicate that formation of pr17/n was possibly a result of alternative translation initiation at a non-AUG codon in the Ω wild-type sequence, thereby giving rise to an N-terminally elongated protein. Figure 2 View largeDownload slide Influence of a putative stem-loop structure on translation initiation. ( A ) Part of the Ω sequence and a potential stem-loop structure 3 nt downstream of AUU 63 in the multiple cloning site is shown. The AUG start codon of ORF4 and two potential sites of translation initiation (AUU 15 and AUU 63 ) in the Ω sequence are marked by open boxes. In construct pS17D the stem-loop was deleted. Arrowheads indicate the mutation of AUU 15 , AUU 63 and of the ORF4 AUG start codon in constructs pS17N15 and pS17N63. The 5′ located 59 nt of the Ω sequence were deleted in construct pS17N63. Construct pS17 without the Ω sequence served as a control for the synthesis of wild-type pr17. ( B and C ) In vitro translation of chimeric RNAs. RNAs from all constructs were translated in vitro using a reticulocyte lysate (B) or a wheat germ extract (C) and separated by PAGE. Control lanes are designated rl (reticulocyte lysate) or wg (wheat germ extract) respectively and represent in vitro translation products in the absence of externally added RNAs. C5–C9 correspond to the constructs shown in (A). Figure 2 View largeDownload slide Influence of a putative stem-loop structure on translation initiation. ( A ) Part of the Ω sequence and a potential stem-loop structure 3 nt downstream of AUU 63 in the multiple cloning site is shown. The AUG start codon of ORF4 and two potential sites of translation initiation (AUU 15 and AUU 63 ) in the Ω sequence are marked by open boxes. In construct pS17D the stem-loop was deleted. Arrowheads indicate the mutation of AUU 15 , AUU 63 and of the ORF4 AUG start codon in constructs pS17N15 and pS17N63. The 5′ located 59 nt of the Ω sequence were deleted in construct pS17N63. Construct pS17 without the Ω sequence served as a control for the synthesis of wild-type pr17. ( B and C ) In vitro translation of chimeric RNAs. RNAs from all constructs were translated in vitro using a reticulocyte lysate (B) or a wheat germ extract (C) and separated by PAGE. Control lanes are designated rl (reticulocyte lysate) or wg (wheat germ extract) respectively and represent in vitro translation products in the absence of externally added RNAs. C5–C9 correspond to the constructs shown in (A). Inspection of the transgene sequence revealed that two AUU codons (AUU 15 and AUU 63 ) of the Ω sequence were in-frame with the ORF4 AUG start codon ( Fig. 1 A). To assess the size of a protein that would initiate in the Ω sequence two constructs were synthesized by site-directed mutagenesis in which the pr17 start codon was converted to GCG and the AUU codons AUU 15 and AUU 63 of the TMV Ω sequence were mutated to AUG 15 and AUG 63 respectively (constructs p17/NI and p17/NIII; Fig. 1 A). Both constructs were used for transformation of S.tuberosum and protein extracts from leaves of regenerated plants were subjected to Western blot analysis ( Fig. 1 B). Plants transformed with construct p17/NI expressed a protein larger than pr17/n ( Fig. 1 B, C2), whereas p17/NIII transgenic plants showed a protein corresponding in size to pr17/n, as detected in p17/NIV transgenic plants ( Fig. 1 B, C3). It appears that expression of p17/NI and p17/NIII in transgenic plants ( Fig. 1 B, C2 and C3) resulted in much higher protein levels as compared with p17 transgenic plants ( Fig. 1 B, C4). As p17 and p17/NIII did not contain the Ω sequence, this observation was explained by the unfavourable context of the pr17 initiator codon (GGAA AUG UCA). These data provided the first evidence that translation initiation can occur in the Ω sequence and suggested a preferential translation initiation at AUU 63 . A further, more detailed analysis of potential translation start codons was carried out in vitro . A stem-loop structure does not contribute to translation initiation in the Ω sequence The 5′ leader of construct p17/NIV consisted of the TMV Ω sequence and an additional 63 nt derived from the multiple cloning site. Due to the cloning strategy part of this cloning site was inversely repeated, allowing the formation of a stable stem-loop structure ( Fig. 2 A). This stem-loop is located 3 nt downstream of codon AUU 63 and could have made a substantial contribution to the signal for translation initiation. In order to investigate the possible effect of this stem-loop on translation efficiency a Hin dIII fragment released from construct p17/NIV and comprising the complete 5′ leader and ORF4 (pr17) sequence was cloned under the control of the SP6 promoter into vector pSP64 (construct pS17N). Furthermore, the stem-loop was deleted to yield plasmid pS17D. RNAs from both constructs were transcribed in vitro and translated in a rabbit reticulocyte lysate, as well as in a wheat germ extract. Translation in both cell-free systems resulted in expression of wild-type pr17 and two additional proteins, one identical in size to the pr17/n protein detected in planta ( Fig. 2 B, C; see above). The two additional proteins corresponded in size to polypeptides synthesized by initiation at the AUG 15 or AUG 63 codons respectively of RNAs from constructs pS17N15 and pS17N63 ( Fig. 2 ). Thus in vitro expression of construct pS17N permitted translation initiation at more than one non-AUG codon, probably at AUU 15 and AUU 63 . In vitro translation of construct pS17D also showed expression of two additional proteins. It was concluded, therefore, that the putative stem-loop structure was not necessary in vitro for translation initiation in the Ω sequence ( Fig. 2 B and C). The expression level for the two extra proteins differed in the animal (reticulocyte lysate) and plant (wheat germ extract) in vitro translation systems. The 24 kDa protein was more prominently expressed in the wheat germ system, which thereby reflected the actual in planta situation ( Fig. 1 B). The reticulocyte lysate predominantly expressed the 26 kDa protein. Therefore, the wheat germ system was selected for further in vitro experiments. Translation in the wheat germ system initiates at AUU 63 of the TMV Ω sequence In vitro expression of the chimeric Ω-ORF4 construct (constructs pS17N and pS17D) indicated that translation initiation occurred at two non-AUG codons of the Ω sequence upstream of the ORF4 AUG start codon. In addition, translation of pS17N15 and pS17N63 RNAs provided circumstantial evidence for initiation at codons AUU 15 and AUU 63 respectively. Further analyses were directed at unequivocally identifying the non-AUG initiator codon in the Ω sequence utilized in planta . The most likely non-AUG codon recognized by the plant ribosomal initiation complex was AUU 63 , as transgenic plants transformed with construct p17/NIII (AUG 63 ) expressed a protein corresponding in size to pr17/n. The AUU 63 codon of plasmid pS17D was mutated to AGG 63 in order to inhibit translation initiation at this codon (construct pS17D3; Fig. 3 A). In fact, in vitro -translated RNA of plasmid pS17D3 did not result in a product corresponding in size to pr17/n, demonstrating that in vitro translation initiated at AUU 63 of the Ω sequence ( Fig. 3 B, C12). Based on the results of Gordon ( 21 ) and Peabody ( 22 ), AUU 63 was further mutated to ACG or CUG (constructs pS17D1 and pS17D2 respectively; Fig. 3 A). These codons are known to permit translation initiation with high efficiency in mammalian cells and plant protoplasts. Similar results were obtained with the mutated AUU 63 codon, as in vitro translation of pS17D1 and pS17D2 RNAs in the wheat germ system allowed expression of pr17/n by initiation at both ACG 63 and CUG 63 ( Fig. 3 B). The potential mechanism by which translation initiated at both AUU 63 (to yield pr17/n) and at the wild-type ORF4 AUG start codon (pr17 synthesis) was examined by converting AUU 63 into AUG 63 (pS17D4; Fig. 3 B, C13). The fact that AUG 63 directed almost exclusive synthesis of pr17/n, with scarcely any pr17 formed, was taken as an indication that in the wild-type situation internal initiation at the pr17 AUG, as opposed to AUU 63 initiation, occurred by a leaky scanning mechanism. Influence of AUU flanking sequences on translation initiation The flanking sequences at the AUU 63 codon largely conformed to the consensus context for plant AUG initiation codons ( Fig. 4 A). To further analyse the influence of bases neighbouring AUU 63 several point mutations were introduced into this region ( Fig. 4 A). Single point mutations did not alter translation efficiency at AUU 63 (the total amount of protein synthesized from construct pS17A3 RNA and loaded in lane C16 is lower as compared with total protein in the other lanes). Even the replacement of a purine by a pyrimidine at the mutation-sensitive position −3 did not inhibit expression of pr17/n (construct pS17A6, Fig. 4 , C19). Only when the entire context of the AUU 63 codon was disrupted, as in pS17A7, expression of pr17/n was reduced ( Fig. 4 , C20). On the other hand, adaptation of the flanking sequences according to the consensus sequence did not increase translation initiation at AUU 63 as compared with the wild-type sequence ( Fig. 4 B, C6, C14 and C15). These results indicate that the AUU 63 flanking sequences have only a minor effect on pr17/n translation efficiency in vitro . Figure 3 View largeDownload slide Mutational analysis of the AUU 63 translation initiation codon of the Ω sequence. ( A ) The Ω sequence and the start codon of ORF4 are shown. Based on construct pS17D four different point mutations were introduced in the AUU 63 codon to create constructs pS17D1–pS17D4. ( B ) PAGE analysis of in vitro translation products from RNAs of constructs ps17D (C6) and ps17D1–pS17D4 (C10–C13). Lanes indicated with C9, C7 and C8 contain in vitro translation products from RNAs of constructs pS17, pS17N15 and pS17N63 (see Fig. 2 ). Figure 3 View largeDownload slide Mutational analysis of the AUU 63 translation initiation codon of the Ω sequence. ( A ) The Ω sequence and the start codon of ORF4 are shown. Based on construct pS17D four different point mutations were introduced in the AUU 63 codon to create constructs pS17D1–pS17D4. ( B ) PAGE analysis of in vitro translation products from RNAs of constructs ps17D (C6) and ps17D1–pS17D4 (C10–C13). Lanes indicated with C9, C7 and C8 contain in vitro translation products from RNAs of constructs pS17, pS17N15 and pS17N63 (see Fig. 2 ). Interaction of a triple AUU block with translation initiation at AUU 63 As the flanking sequences exhibited little activity in modulating the efficiency of translation initiation at codon AUU 63 , sequences located further upstream of AUU 63 (positions 44–58 of the Ω sequence) were examined for their influence on translation initiation. An element composed of three AUU codons separated from each other by one codon (‘triple AUU block’) is located 4 nt upstream of AUU 63 in a different reading frame ( Fig. 5 A). Simultaneous mutation of all three AUU codons to ACU slightly increased expression of pr17/n ( Fig. 5 B, C21), whereas a point mutation of the central AUU codon to ACU had no effect on translation initiation at AUU 63 ( Fig. 5 B, C22). Thus the triple AUU block in the wild-type Ω sequence obviously decreased translation initiation at AUU 63 to some extent. Figure 4 View largeDownload slide Influence of flanking sequences on translation initiation. ( A ) The flanking sequences of the AUU 63 codon are underlined. Two lines of mutations were carried out by disrupting or adapting the flanking sequences according to the consensus sequence for plant translation initiation codons ( 2 ). Single base substitutions are indicated by arrows. ( B ) PAGE analysis of in vitro translation products from RNAs of constructs pS17A1–pS17A7 (C14–C20). As a negative control the wheat germ extract incubated without external RNA was loaded (wg). In vitro translation products of construct pS17D, p17N63 and pS17 RNAs (see Fig. 2 ) were loaded in lanes indicated with C6, C8 and C9 respectively. Figure 4 View largeDownload slide Influence of flanking sequences on translation initiation. ( A ) The flanking sequences of the AUU 63 codon are underlined. Two lines of mutations were carried out by disrupting or adapting the flanking sequences according to the consensus sequence for plant translation initiation codons ( 2 ). Single base substitutions are indicated by arrows. ( B ) PAGE analysis of in vitro translation products from RNAs of constructs pS17A1–pS17A7 (C14–C20). As a negative control the wheat germ extract incubated without external RNA was loaded (wg). In vitro translation products of construct pS17D, p17N63 and pS17 RNAs (see Fig. 2 ) were loaded in lanes indicated with C6, C8 and C9 respectively. This observation could have resulted from translation initiation at the triple AUU block, thereby competing for ribosomal initiation complex formation with codon AUU 63 . To test this possibility the triple AUU block was placed in-frame with ORF4 by the insertion of 2 nt upstream of AUU 63 ( Fig. 5 A, pS17C3). In vitro translation of the frame-shift mutant RNA resulted in a double band at 24 kDa, indicating that translation initiation had taken place at the triple AUU block as well as at AUU 63 ( Fig. 5 B, C23). Another frame-shift mutant was created to unequivocally demonstrate translation initiation at the triple AUU block ( Fig. 6 A, pS17B1). In this frame-shift mutation the triple AUU block was placed in-frame with ORF4 by inserting 2 nt into the multiple cloning site such that AUU 15 as well as AUU 63 were out-of-frame with respect to ORF4. In vitro translation of construct pS17B1 RNA resulted in expression of pr17 and a second protein with an apparent molecular weight of 25 kDa, as expected from translation initiation at the triple AUU block ( Fig. 6 B, C24). Thus the negative regulatory effect of the triple AUU block on translation initiation at the AUU 63 codon was due to competition for the scanning complex and initiation complex formation. The proteins translated from the triple AUU block (in the wild-type construct) would have calculated molecular weights of 2 kDa and were, therefore, not visible by SDS-PAGE analysis. Figure 5 View largeDownload slide Translation initiation at a triple AUU block ofthe Ω sequence. ( A ) The AUG codon of ORF4 and AUU 15 , as well as AUU 63 , are highlighted by blue boxes, indicating the same reading frame. Different reading frames are represented by different colours. The triple AUU block is located 4 nt upstream of AUU 63 , marked by yellow boxes. Point mutations and insertion of nucleotides are indicated by arrows. The altered reading frame of the triple AUU block in construct pS17C3 is represented by blue boxes. In the same construct AUU 15 is in the third reading frame, shown by a pink box. ( B ) PAGE analysis of in vitro translation products from RNAs of constructs pS17C1–pS17C3 (C21–C23). The controls pS17D, pS17N 63 and pS17 (see Fig. 2 ) were loaded in lanes indicated with C6, C8 and C9. Figure 5 View largeDownload slide Translation initiation at a triple AUU block ofthe Ω sequence. ( A ) The AUG codon of ORF4 and AUU 15 , as well as AUU 63 , are highlighted by blue boxes, indicating the same reading frame. Different reading frames are represented by different colours. The triple AUU block is located 4 nt upstream of AUU 63 , marked by yellow boxes. Point mutations and insertion of nucleotides are indicated by arrows. The altered reading frame of the triple AUU block in construct pS17C3 is represented by blue boxes. In the same construct AUU 15 is in the third reading frame, shown by a pink box. ( B ) PAGE analysis of in vitro translation products from RNAs of constructs pS17C1–pS17C3 (C21–C23). The controls pS17D, pS17N 63 and pS17 (see Fig. 2 ) were loaded in lanes indicated with C6, C8 and C9. An additional frame-shift mutant was constructed as a negative control with all AUU codons of the Ω sequence out-of-frame with ORF4 ( Fig. 6 A, construct pS17B2). Translation of RNA from this construct showed an additional protein with an apparent molecular weight of 22 kDa ( Fig. 6 B, C25). The extra protein had a smaller apparent molecular weight than pr17/n and was possibly synthesized by initiation within the multiple cloning site ( Fig.6 A). A GUG codon, the most likely initiator codon in this region, was mutated to GAG (pS17B3). Absence of the extra protein confirmed that initiation on pS17B2 RNA had occured at the GUG codon ( Fig. 6 B, C26). Whether GUG and the triple AUU block direct translation initiation in planta remains to be determined. Discussion Potato plants transformed with PLRV ORF4 under the translational control of the TMV Ω sequence expressed two immunoreactive proteins, wild-type pr17 and mutant protein pr17/n. We were able to show that initiation at the internally located translational start codons proceeded by leaky scanning of pre-initiation complexes and that a non-canonical translation mechanism was responsible for pr17/n formation by alternative translation initiation at a non-AUG codon of the TMV translational enhancer. In planta and during in vitro translation in a cell-free plant system (wheat germ) initiation occured efficiently at the ORF4 AUG start codon, as well as some 25 codons upstream at AUU 63 of the Ω sequence. When AUU 63 was replaced by AUG 63 (construct p17/NIII) a protein corresponding in size to pr17/n was expressed in transgenic plants, but mutation of AUU 63 to AGG 63 prevented expression of this N-terminally elongated pr17 (pr17/n). Figure 6 View largeDownload slide Translation initiation at a GUG codon of the 5′ untranslated leader. ( A ) Part of the Ω sequence and multiple cloning site are shown. Differently coloured boxes represent different reading frames. Blue indicates the reading frame of ORF4. Insertion or mutation of single bases are indicated by arrows. The frame-shift of the triple AUU block and initiation codons in constructs pS17B1 and pS17B2 are indicated by different colours. ( B ) PAGE analysis of in vitro -translated products from constructs pS17B3, pS17B1 and pS17B2 were loaded in lanes C25, C24 and C26 respectively. Controls pS17N15, pS17N63 and pS17 (see Fig. 2 ) were loaded in lanes C7, C8 and C9 respectively. Figure 6 View largeDownload slide Translation initiation at a GUG codon of the 5′ untranslated leader. ( A ) Part of the Ω sequence and multiple cloning site are shown. Differently coloured boxes represent different reading frames. Blue indicates the reading frame of ORF4. Insertion or mutation of single bases are indicated by arrows. The frame-shift of the triple AUU block and initiation codons in constructs pS17B1 and pS17B2 are indicated by different colours. ( B ) PAGE analysis of in vitro -translated products from constructs pS17B3, pS17B1 and pS17B2 were loaded in lanes C25, C24 and C26 respectively. Controls pS17N15, pS17N63 and pS17 (see Fig. 2 ) were loaded in lanes C7, C8 and C9 respectively. In vitro translation of ORF4 under the control of the Ω sequence resulted in expression of three proteins instead of the two detected in transgenic plants. This was observed both in the reticulocyte lysate and wheat germ extract: translation initiated additionally at AUU 15 , as is obvious from a mutant RNA in which AUU 15 had been replaced by AUG 15 . Differences in the expression patterns for both cell-free systems could probably reflect conditions of the in vitro translation systems which allow translation initiation at a non-AUG codon upstream of AUU 63 not recognized in planta . In addition, it is noteworthy that the animal and plant in vitro systems show different affinities for the two codons AUU 15 and AUU 63 . The fact that AUU 15 is predominantly used by the reticulocyte lysate for translation intiation does not reflect preferences of the animal system for a different consensus context of this AUU start codon, as the flanking sequences for AUU 15 and AUU 63 are identical. Although artefacts of the conditions of the in vitro translation cannot be excluded, animal-specific protein factors may be involved in mRNA interaction and specific recognition of the first initiator codon, a phenomenon recently discussed in detail for eukaryotic gene expression ( 7 ). Further analyses of the Ω sequence focused on elements contributing to translation initiation at AUU 63 . Optimal initiation of protein biosynthesis depends on the sequence context for the start codon ( 1–4 ), which is different in plant and animal consensus sequences. However, in both systems positions −3 and +4, with reference to the +1 adenosine of the AUG start codon, require purine residues for efficient translation initiation ( 31 ). According to Cavener and Ray ( 32 ) the flanking sequences of mono-and dicotyledonous plants differ substantially. As the experiments described here were carried out in a wheat germ system we cannot exclude that in S.tuberosum the point mutations in the flanking sequences would exert a more prominent effect on translational efficiency. The data presented here on the AUU flanking sequences confirm their importance for optimal protein initiation, but mutation of the entire consensus sequence did not completely inhibit translation initiation. While mutation of a purine to a pyrimidine residue at position −3 did not apparently alter initiation efficiency at AUU 63 , the triple AUU block preceding this codon reduces its efficiency in initiation. As was shown by site-directed mutagenesis, the triple AUU block may itself interact with the scanning complex, forming initiation complexes and thereby competing with AUU 63 . In fact, leaky scanning is obviously the mechanism by which recognition of start codons occurs in the TMV Ω sequence. When AUU 63 was mutated to AUG 63 expression of pr17 at the AUG of ORF4 was barely detectable, indicating that the canonical AUG start codon at position 63 was now almost exclusively used for formation of initiation complexes. The translation initiation at AUU codons described here is a novel feature of the Ω sequence, in addition to its function as a translational enhancer. Previously a detailed analysis of the Ω sequence had identified two motifs necessary for translation enhancement, a (CAA) n region and a direct repeat of 8 nt ( 12 ). Both the AUU 15 and AUU 63 codons are part of this direct repeat ACA AUU AC. Based on the observation of disome formation in the Ω sequence, translation initiation at AUU 15 was proposed ( 16 , 33 ) as contributing to enhancement of translation. However, mutation of AUU 15 to CUU 15 in the 5′ located direct repeat or the introduction of two stop codons further downstream demonfstrated that translation initiation at AUU 15 did not contribute to enhancement by the Ω sequence ( 14 , 34 ). The stop codons were introduced in a construct where the downstream direct repeats (comprising AUU 63 and part of the triple AUU block) were deleted. Thus a potential contribution to enhancement by translation initiation at AUU 63 could not be assessed. Alternative translation initiation at AUU 63 of the TMV Ω sequence, as well as leaky scanning and initiation at the canonical ORF4 AUG start codon from the identical mRNA, resulted in expression of two proteins. Such bifunctional mRNAs are known for a number of other viruses and eucaryotic mRNAs ( 6 , 20 ) and may lead to N-terminally altered proteins with modified functions, as for example in the expression of two N-terminally different serine-threonine protein kinases encoded by the mouse pim-1 oncogene or the translation of three proteins (C′, C and Y) from Sendai virus RNA by exploiting an ACG and two different AUG codons ( 35 , 36 ). Our results indicate that translation initiation at AUU 63 of the Ω sequence takes place with high efficiency in planta, with both pr17/n and wild-type pr17 accumulating to similar levels in transgenic plants. The question remains whether the TMV Ω sequence directs expression of two N-terminally different proteins from TMV RNA. In TMV RNA the Ω sequence is followed by the polymerase gene and initiation at AUU 63 of TMV strain U1 would extend the viral polymerase by only two amino acid residues. The Ω sequences of other TMV strains (U2, L and Dahlemense) are slightly different and the AUU codons corresponding in position to AUU 63 of TMV U1 are not in-frame with the polymerase gene. Hence, eventual expression of an N-terminally modified polymerase protein would not be a conserved feature of different TMV strains. As an alternative to the production of an N-terminally modified viral replicase the small uORFs starting at AUU codons of the Ω sequence could represent a regulatory mechanism for TMV gene expression, as they would decrease the number of ribosomes initiating at the AUG start codons of the polymerase gene, either by direct competition through the formation of initiation complexes or as a consequence of poor re-initiation of eukaryotic ribosomes subsequent to termination at the uORF stop codons. The efficiency of uORF translation may be regulated during TMV replication by interaction of this sequence with protein factors of the host cell, like eIF-2, which is involved in initiation site recognition and stabilization of tRNA-mRNA interactions ( 37 ). However, virus encoded proteins, like CaMV trans -activator protein ( 38 ), may also function in modulation of translational efficiency and it remains to be determined whether TMV proteins make use of this mechanism for regulation of TMV gene expression during the late stages of replication, when genomic TMV RNA is preferentially assembled into progeny virus particles. Acknowledgements The technical assistance of Alice Kaufmann and Dieter Becker is gratefully acknowledged. This work was in part supported by the Deutsche Forschungsgemeinschaft through grant Ro 300/63 to WR. References 1 Kozak M.. ,  Cell ,  1986, vol.  44 (pg.  283- 292) CrossRef Search ADS PubMed  2 Lütcke H.A.,  Chow K.C.,  Mickel F.S.,  Moss K.A.,  Kern H.F.,  Scheele G.A.. ,  EMBO J. ,  1987, vol.  6 (pg.  43- 48) PubMed  3 Grünert S.,  Jackson R.J.. ,  EMBO J. ,  1994, vol.  13 (pg.  3618- 3630) PubMed  4 Boeck R.,  Kolakofsky D.. ,  EMBO J. ,  1994, vol.  13 (pg.  3608- 3617) PubMed  5 Kozak M.. ,  Proc. Natl. Acad. Sci. USA ,  1986, vol.  83 (pg.  2850- 2854) CrossRef Search ADS   6 Gallie D.R.. ,  Annu. Rev. Plant Physiol. Plant Biol. ,  1993, vol.  44 (pg.  77- 105) CrossRef Search ADS   7 McCarthy J.E.G.,  Kollmus H.. ,  Trends Biochem. Sci. ,  1995, vol.  20 (pg.  191- 197) CrossRef Search ADS PubMed  8 Shine J.,  Dalgarno L.. ,  Proc. Natl. Acad. Sci. USA ,  1974, vol.  71 (pg.  1342- 1346) CrossRef Search ADS   9 Kozak M.. ,  Cell ,  1980, vol.  22 (pg.  7- 8) CrossRef Search ADS PubMed  10 Kozak M.. ,  Microbiol Rev. ,  1983, vol.  47 (pg.  1- 45) PubMed  11 Smirnyagina E.V.,  Morozov S.Y.,  Rodionova N.P.,  Miroshnichenko N.A.,  Solovyev A.G.,  Fedorkin O.N.,  Atabekov J.G.. ,  Biochimie ,  1991, vol.  73 (pg.  587- 598) CrossRef Search ADS PubMed  12 Gallie D.R.,  Sleat D.E.,  Watts J.W.,  Turner P.C.,  Wilson T.M.A.. ,  Nucleic Acids Res. ,  1987, vol.  15 (pg.  8693- 8711) CrossRef Search ADS PubMed  13 Lawson T.G.,  Ray B.K.,  Dodds J.T.,  Grifo J.A.,  Abramson R.D.,  Merrick W.C.,  Betsch D.F.,  Weith H.L.,  Thach R.E.. ,  J. Biol. Chem. ,  1986, vol.  261 (pg.  13979- 13989) PubMed  14 Gallie D.R.,  Walbot V.. ,  Nucleic Acids Res. ,  1992, vol.  20 (pg.  4631- 4638) CrossRef Search ADS PubMed  15 Konarska M.,  Filipowicz W.,  Domdey H.,  Gross J.. ,  Eur. J. Biochem ,  1981, vol.  114 (pg.  221- 227) CrossRef Search ADS PubMed  16 Tyc K.,  Konarska M.,  Gross J.,  Filipowicz W.. ,  Eur. J. Biochem. ,  1984, vol.  140 (pg.  503- 511) CrossRef Search ADS PubMed  17 Both G.W.,  Furuichi Y.,  Muthukrishnan S.,  Shatkin A.J.. ,  J. Mol. Biol. ,  1976, vol.  104 (pg.  637- 658) CrossRef Search ADS PubMed  18 Montoya J.,  Ojala D.,  Attardi G.. ,  Nature ,  1981, vol.  290 (pg.  465- 470) CrossRef Search ADS PubMed  19 Sacerdot C.,  Fayat G.,  Dessen P.,  Springer M.,  Plumbridge J.A.,  Grunberg-Manago M.,  Blanquet S.. ,  EMBO J. ,  1982, vol.  1 (pg.  311- 315) PubMed  20 Rohde W.,  Gramstat A.,  Schmitz J.,  Tacke E.,  Prüfer D.. ,  J. Gen. Virol. ,  1994, vol.  75 (pg.  2141- 2149) CrossRef Search ADS PubMed  21 Gordon K.,  Fütterer J.,  Hohn T.. ,  Plant J. ,  1992, vol.  2 (pg.  809- 813) PubMed  22 Peabody D.S.. ,  J. Biol. Chem. ,  1989, vol.  264 (pg.  5031- 5035) PubMed  23 Tacke E.,  Sarkar S.,  Salamini F.,  Rohde W.. ,  Arch. Virol. ,  1989, vol.  105 (pg.  153- 163) CrossRef Search ADS PubMed  24 Tacke E.,  Kull B.,  Prüfer D.,  Reinold S.,  Schmitz J.,  Salamini F.,  Rohde W..  Bills D.D.,  Kung S.D.. ,  Viral Pathogenesis and Disease Resistance ,  1995 River Edge World Scientific Publishing  in press 25 Bevan M.. ,  Nucleic Acids Res. ,  1984, vol.  12 (pg.  8711- 8721) CrossRef Search ADS PubMed  26 Hoekema A.,  Hirsch P.,  Hooykaas P.,  Schilperoort R.A.. ,  Nature ,  1983, vol.  303 (pg.  179- 180) CrossRef Search ADS   27 Horsch R.B.,  Fry J.E.,  Hoffmann N.C.,  Eichholtz D.,  Rogers S.G.,  Fraley R.T.. ,  Science ,  1985, vol.  227 (pg.  1229- 1231) CrossRef Search ADS PubMed  28 Tacke E.,  Schmitz J.,  Prüfer D.,  Rohde W.. ,  Virology ,  1993, vol.  197 (pg.  274- 282) CrossRef Search ADS PubMed  29 Melton D.A.,  Krieg P.A.,  Rebagliati M.R.,  Maniatis T.,  Zinn K.,  Green M.R.. ,  Nucleic Acids Res. ,  1984, vol.  12 (pg.  7035- 7056) CrossRef Search ADS PubMed  30 Bonner W.M.,  Laskey R.A.. ,  Eur. J. Biochem. ,  1974, vol.  46 (pg.  83- 88) CrossRef Search ADS PubMed  31 Kozak M.. ,  J. Cell Biol. ,  1989, vol.  108 (pg.  229- 241) CrossRef Search ADS PubMed  32 Cavener D.R.,  Ray S.C.. ,  Nucleic Acids Res. ,  1991, vol.  19 (pg.  3185- 3192) CrossRef Search ADS PubMed  33 Filipowicz W.,  Haenni A.L.. ,  Proc. Natl. Acad. Sci. USA ,  1979, vol.  76 (pg.  3111- 3115) CrossRef Search ADS   34 Gallie D.R.,  Sleat D.E.,  Watts J.W.,  Turner P.C.,  Wilson T.M.A.. ,  Nucleic Acids Res. ,  1988, vol.  16 (pg.  883- 893) CrossRef Search ADS PubMed  35 Saris C.L.M.,  Domen J.,  Berns A.. ,  EMBO J. ,  1991, vol.  10 (pg.  655- 664) PubMed  36 Curran J.,  Kolakofsky D.. ,  EMBO J. ,  1988, vol.  7 (pg.  245- 251) PubMed  37 Donahu T.F.,  Cigan A.M.,  Pabich E.K.,  Valavicius B.C.. ,  Cell ,  1988, vol.  54 (pg.  621- 632) CrossRef Search ADS PubMed  38 De Tapia M.,  Himmelbach A.,  Hohn T.. ,  EMBO J. ,  1993, vol.  12 (pg.  3305- 3314) PubMed  © 1996 Oxford University Press

Dowhan, Dennis H.; Muscat, George E. O.

doi: N/Apmid: N/A

The retinoid X receptors α-, β- and γ- (RXRs) share a highly conserved ‘C’ region or DNA binding domain (DBD). The conserved ‘DE’ region or ligand binding domain (LBD) of the RXRs is functionally complex, mediating dimerization and a ligand-dependent activation function (AF-2). The AB or N-terminal region of the RXRs is poorly conserved and encodes a ligand-independent activation function (AF-1). RXRγ mRNA is preferentially expressed in skeletal and cardiac muscle, however, cell-specific steroid receptor-mediated trans-activation is a poorly understood phenomenon. We utilized the GAL4 hybrid assay system and have demonstrated that RXRγ contains two functional domains in the AB and DE regions that activate transcription in a ligand-independent and -dependent manner respectively. The functions of the AB (AF-1) and DE (AF-2) domains were regulated by cAMP-dependent protein kinases, furthermore, the function of AF-2 in the LBD was activated by 8-Br-cAMP, independent of 9-cis-retinoic acid treatment. Deletion analysis demonstrated that the AF-1 of RXRγ, is located between amino acids 1 and 103 and contained multiple motifs that were targets of cAMP-dependent protein kinases. Transfection analyses in non-muscle and myogenic cells clearly demonstrated that: (i) the AF-1 of RXRγ functions in a muscle-specific manner and is required for optimal ligand-dependent trans-activation from an RXRE; (ii) RXRγ trans-activates more efficiently in a myogenic background.

Dowhan, Dennis H.;Muscat, George E. O.

doi: 10.1093/nar/24.2.264pmid: 8628649

Abstract The retinoid X receptors α-, β- and γ- (RXRs) share a highly conserved ‘C’ region or DNA binding domain (DBD). The conserved ‘DE’ region or ligand binding domain (LBD) of the RXRs is functionally complex, mediating dimerization and a ligand-dependent activation function (AF-2). The AB or N-terminal region of the RXRs is poorly conserved and encodes a ligand-independent activation function (AF-1). RXRγ mRNA is preferentially expressed in skeletal and cardiac muscle, however, cell-specific steroid receptor-mediated trans -activation is a poorly understood phenomenon. We utilized the GAL4 hybrid assay system and have demonstrated that RXRγ contains two functional domains in the AB and DE regions that activate transcription in a ligand-independent and -dependent manner respectively. The functions of the AB (AF-1) and DE (AF-2) domains were regulated by cAMP-dependent protein kinases, furthermore, the function of AF-2 in the LBD was activated by 8-Br-cAMP, independent of 9- cis -retinoic acid treatment. Deletion analysis demonstrated that the AF-1 of RXRγ, is located between amino acids 1 and 103 and contained multiple motifs that were targets of cAMP-dependent protein kinases. Transfection analyses in non-muscle and myogenic cells clearly demonstrated that: (i) the AF-1 of RXRγ functions in a muscle-specific manner and is required for optimal ligand-dependent trans -activation from an RXRE; (ii) RXRγ trans -activates more efficiently in a myogenic background. Introduction Retinoids play an important and fundamental role in development, differentiation and homeostasis ( 1–3 ). The effects of retinoids are mediated by two subgroups of the steroid receptor superfamily of nuclear receptors that bind specific DNA sequences, termed hormone response elements (HREs), and act as ligand-inducible transcriptional regulators ( 4 , 5 ). The two subgroups that mediate the effects of retinoids are the retinoic acid receptors (RARα, β, γ and various isoforms) that mediate transcriptional activation in response to both all- trans -retinoic acid (RA) or 9- cis -RA and retinoid X receptors (RXRα, β and γ), that mediate transcriptional activation in response to 9- cis -RA ( 3 , 6 ). The RXRs share similar structural domains with other members of the steroid receptor superfamily based on amino acid similarity ( 3 , 5 ). RXRs and other members of the steroid/thyroid nuclear receptor family share a highly conserved C region or DNA binding domain (DBD) and DE region or ligand binding domain (LBD) ( 3 ). The LBD of the three RXRs characterized to date is functionally complex, mediating ligand binding, dimerization and a ligand-dependent activation function (AF-2) responsible for ligand-mediated transactivation ( 7–11 ). The AB regions of the three RXRs show <40% homology ( 6 ), with RXRα and RXRγ containing an activation function (AF-1) responsible for ligand-independent transactivation ( 8 ). The functional properties of the N-terminal AB region of the steroid/thyroid receptors has developed into an area of increasing interest. Recent studies on the members of the steroid receptor gene family indicate that this region, depending on the receptor, may play an important role in DNA binding, transactivation, cell type- and/or promoter-specific regulation or interaction with the general transcription factor TFIIB ( 12 , and references therein). In terms of the retinoid receptors transactivation and promoter-specific regulation has been shown to be mediated by the different N-terminal regions of RXRs and RARs when linked to the DBD of the estrogen receptor ( 8 , 13 ) or GAL4 ( 14 ). Many of the nuclear receptors identified to date have been found to be phosphoproteins ( 15 ), including RARα, RARβ and RARγ ( 16–18 ). The protein kinase C- and cAMP-dependant protein kinase pathways have been implicated in regulation of retinoid-mediated transcription, which suggest that phosphorylation processes may be involved in regulating the function of retinoid receptors ( 19–21 ). In recent studies with RXRs and RARs it has been demonstrated that in both the presence and absence of ligand manipulation of the phosphorylation state of the cell with okadaic acid (OA), which inhibits protein phosphatases PP1 and PP2A ( 22 ), led to increases in transactivation of RXR-or RAR-responsive reporter genes in transient transfection experiments ( 23 , 24 ). In an effort to further investigate the role of phosphorylation in relation to RXRγ-mediated transcriptional activation we have studied the effects of 8-Br-cAMP [an activator of protein kinase A (PKA)] and OA on the AF-1 and AF-2 domains of RXRγ. Materials and Methods Plasmids The expression plasmids pGALO ( 25 ), pSG5 (Stratagene) and pSG5RXRγ ( 26 ) and reporter plasmids pBLCAT2 (ptkCAT) ( 27 ), pG18 2 -tkCAT ( 28 ) and G5E1b-CAT ( 29 ) have been described elsewhere. Generation of mRXRγ DNA fragments was performed by PCR with Pfu DNA polymerase, using the manufacturer's buffer. All PCR products were cloned into the Sma I site of pBS (Stratagene) and then isolated after Eco RI digestion. GAL-RXRγ was constructed by excising mRXRγ1l DNA from the pSG5-RXRγ vector, end filling with Klenow fragment and ligating into an end-filled Sal I site in pGALO. GAL-RXRγAB was constructed by PCR using oligonucleotides 173, 5′-GCGGAATTCACCATGTATGGAAATTATTCCCAC-3′, and 225, 5′-GCGGAATTCATAAGATGTGTTTCACCAGAGAC-3′, to generate the RXRγ AB region, which was ligated into the Eco RI site of pGALO. GAL-RXRγDE was generated by PCR using oligonucleotides 223, 5′-GCGGAATTCACCAAGCGGGAAGCTGTGC-3′, and 176, 5′-GCGGAATTCCTCAGGTGATCTGCAGTGGGGTCT-3′, to generate the RXRγ DE region, which was cloned into the Eco RI site of pGALO. GAL-RXRγABDE was constructed by PCR using oligonucleotides 173 (see above) and 174, 5′-GCGGAATTCCTCAATGTGTTTCACCAGAGAC-3′, to generate the RXRγ AB region and the RXRγ DE region (as described above). The RXRγ AB and DE regions were then ligated and cloned into Eco RI-digested pBS. Plasmids containing only one copy of the RXRγ AB and DE regions were then sequenced to determine the correct orientation and reading frame. A plasmid containing the RXRγ AB and DE regions [containing amino acids 1–138 and 205–463 of RXRγ separated by four amino acids (Glu-Glu-Phe-Thr)] in the correct orientation and reading frame was identified, subjected to partial Eco RI digestion and the RXRγ ABDE fragment isolated and ligated into the Eco RI site of pGALO. GAL-RXRγ1–103 and GAL-RXRγ1–43 were constructed by digesting the PCR-generated RXRγ AB fragment (see above) with Hin cII or Alu I to produce the DNA fragments corresponding to amino acids 1–103 and 1–43 respectively, which were then ligated into Eco RI/ Sma I-digested pGALO. GAL-RXR-γ1–77, GAL-RXRγ77–138 and GAL-RXRγ104–138 were constructed by digesting the PCR-generated RXRγ AB fragment (see above) with Nco I or Hin cII. The isolated DNA fragments were than end filled with Klenow fragment to produce the DNA fragments corresponding to amino acids 1–77, 77–138 and 104–138, which were then ligated into the end filled Nde I, Sma I and end filled Sal I sites of pGALO respectively. GAL-RXRγ44–138 was constructed by PCR using oligonucleotides 259, 5′-GCGGAATTCACCAGCTACACAGACACCCCAG-3′, and 225 (see above) to generate the DNA fragment corresponding to RXRγ amino acids 44–138, which was ligated into the Eco RI site of pGALO. GAL-RXRγ44–103 was constructed by digesting the PCR-generated RXRγ amino acids 44–138 fragment (see above) with Hinc II and cloning the DNA fragment corresponding to RXRγ amino acids 44–103 into Eco RI/ Sma I-digested pGALO. GAL-RXRγ44–77 and GAL-RXR-γ77–103 were constructed by digesting the DNA fragment corresponding to RXRγ amino acids 44–103 with Nco I. The isolated DNA fragments were then end filled with Klenow fragment to produce the DNA fragments corresponding to amino acids 44–77 an 77–103, which were then ligated into the end-filled Nde I and Sma I sites of pGALO respectively. RXRγΔLBD was constructed by PCR using oligonucleotides 173 (see above) and 262, 5′-GCGGAATTCCTCAACTGGCACATTCTGCCTCAC-3′, to generate the RXRγ DNA fragment corresponding to amino acids 1–229, which was ligated into the Eco RI site of pSG5. RXRγΔAB was constructed by PCR using oligonucleotides 263, 5′-GCGGAATTCATGACCAGCCCTGGGTCTCTGGTG-3′, and 176 (see above) to generate the RXRγ DNA fragment corresponding to amino acids 129–463, which was ligated into the Eco RI site of pSG5. All GALO/RXRγ constructs were sequenced to confirm the reading frame using the Pharmacia T7 sequencing kit (Uppsala, Sweden). Cell culture and transfection COS-1 cells were cultured for 24 h in Dulbecco's modified Eagle's medium (DMEM) containing 5% charcoal-stripped fetal calf serum (FCS) prior to transfection. Each 60 mm dish of COS-1 cells (60–70% confluence) was transiently transfected with 5 µg reporter plasmid DNA (G5E1b-CAT, pG182-tkCAT or ptkCAT) expressing chloramphenicol acetyltransferase (CAT), mixed with the appropriate amount of expression vector (1 µg for pSG5-RXRγ, -RXRγΔLBD, -RXRγΔAB or 3 µg for GAL/RXRγ chimeras) or pUC18/carrier plasmid in each transfection experiment by the DOTAP (Boehringer Mannheim)-mediated procedure as described previously ( 28 ). The DNA/DOTAP mixture was added to the cells in 6 ml fresh medium. After a period of 24 h fresh medium with or without 0.5 mM 8-Br-cAMP, 50 nM okadaic and/or 9- cis -RA (10 −7 M) was added to the cells. The cells were harvested for assay of CAT activity 24–72 h after the transfection period. Each transfection experiment was performed at least three times in order to overcome the variability inherent in transfections. Mouse myogenic C 2 C 12 cells ( 30 , 31 ) were grown in DMEM supplemented with 20% (v/v) FCS in 6% CO 2 . Prior to and during transfection this cell line was induced to biochemically and morphologically differentiate into multinucleated myotubes by serum withdrawal in ligand-deficient medium [DMEM supplemented with 2% (v/v) charcoal-stripped FCS]. Each 60 mm dish of myogenic C 2 C 12 cells (90–100% confluence) was transiently transfected as described above. After a period of 24 h fresh medium [DMEM supplemented with 2% (v/v) charcoal-stripped FCS] ± 9- cis -RA (10 −7 M) was added to the cells. The cells were harvested for assay of CAT activity 24 h after addition of fresh medium. Each transfection experiment was performed at least three times in order to overcome the variability inherent in transfections. CAT assays The cells were harvested, normalized for protein concentration and CAT activity measured as previously described ( 32 ). Aliquots of the cell extracts were incubated at 37°C with 0.1–0.4 mCi [ 14 C]chloramphenicol (ICN) in the presence of 5 mM acetyl CoA and 0.25 M Tris-HCl, pH 7.8. After a 1–4 h incubation period the reaction was stopped by addition of 1 ml ethyl acetate, which was used to extract the chloramphenicol and its acetylated forms. The extracted materials were analysed on Silica gel thin layer chromatography plates as described previously ( 32 ). Quantitation of CAT assays was performed by an AMBIS β-scanner. Figure 1 View largeDownload slide RXRγ has two regions involved in transactivation. The GAL fusion constructs containing RXRγ or various sub-domains of RXRγ (3 µg) were transfected together with the GAL reporter G5E1b-CAT (5 µg) into COS-1 cells in either the absence (−) or presence (+) of 9- cis -RA (10 −7 M) (see Materials and Methods for transient transfection and CAT assay details). A schematic diagram of the GAL fusion constructs (not to scale) and a representative autoradiogram of the CAT assay of these transfections is shown. The mean CAT activity values and standard deviations (bars) were derived from a triplicate experiment. Figure 1 View largeDownload slide RXRγ has two regions involved in transactivation. The GAL fusion constructs containing RXRγ or various sub-domains of RXRγ (3 µg) were transfected together with the GAL reporter G5E1b-CAT (5 µg) into COS-1 cells in either the absence (−) or presence (+) of 9- cis -RA (10 −7 M) (see Materials and Methods for transient transfection and CAT assay details). A schematic diagram of the GAL fusion constructs (not to scale) and a representative autoradiogram of the CAT assay of these transfections is shown. The mean CAT activity values and standard deviations (bars) were derived from a triplicate experiment. Results Autonomous activation functions of RXRγ To identify and further characterize the domains of RXRγ involved in transcriptional activation we utilized the GAL4 hybrid system, whereby putative activation domains are fused to the DBD of the yeast transcription factor GAL4 ( 33 , 34 ). If these regions encode modular activation domains they complement the GAL4 DBD (to produce a functional trans -activator) and induce transcription of the GAL-responsive reporter construct G5E1b-CAT, containing an E1b TATA box with five 17mer GAL4 binding sites linked to the CAT reporter ( 29 ). The system utilized an SV40 promoter expression vector pGALO ( 25 ) that contains a multiple cloning site downstream of the GAL4 DBD. We fused RXRγ and various domains (e.g. AB or DE regions) of RXRγ to the GAL4 DNA binding domain and examined the ability of these chimeras to regulate expression of the G5E1b-CAT reporter in COS-1 cells. The GAL-RXRγ chimera containing the full open reading frame (ORF) of RXRγ in the presence of 9- cis -RA activated transcription ∼4-fold above the control, pGALO (GAL4 DBD), and GAL-RXRγ in the absence of 9- cis -RA. This demonstrated that the GAL-RXRγ chimera conferred appropriate 9- cis -RA-dependent trans -activation via RXRγ to the GAL4 DBD ( Fig. 1 ). The GAL-RXR γ AB plasmid, which contains only the AB regions of RXRγ, with the DBD and LBD deleted, increased transcription of the reporter construct ∼14-fold over the GAL DBD alone, independent of 9- cis -RA treatment ( Fig. 1 ). This indicated that the AB region of RXRγ contained a ligand-independent transactivation function. GAL-RXRγDE, which contains the LBD of RXRγ with the AB and C regions deleted, activated transcription of the G5E1b-CAT reporter ∼8-fold in a 9- cis -RA-dependent manner. The GAL-RXR-γABDE plasmid (which lacks the C region of RXRγ, which contains the DBD), increased transcription ∼10-fold in the presence of 9- cis -RA. Deletion of only the C region of RXRγ increased 9- cis -RA-mediated activation by 2- to 4-fold in comparison with the full-length RXRγ linked to the GAL DBD (GAL-RXRγ. The lower activity of full-length RXRγ may be attributed to the presence of two DNA binding domains in the GAL-RXRγ chimeric protein, causing possible steric hindrance in either the DNA binding or transactivation function. Possible repression of chimeric constructs containing two DBDs with the GAL system have been previously reported ( 14 ). These experiments indicate that there are two domains involved in transactivation by RXRγ; AF-1 in the AB domain, which is ligand-independent, and the ligand-dependant AF-2 in the DE domain, which activates transcription in response to the ligand 9- cis -RA. The activity of the AF-1 and AF-2 domains of RXRγ are regulated by 8-Br-cAMP and OA Recent reports have indicated that transcriptional activation by RXRs can be regulated by phosphorylation ( 23 , 24 ). We therefore investigated the affect of 8-Br-cAMP (a stimulator of cAMP-dependent protein kinases) and OA (an inhibitor of serine-threonine protein phosphatases) on the AF-1 and AF-2 functions of RXRγ. In control studies (to examine non-specific effects of 8-Br-cAMP on the GAL4 hybrid system) when COS-1 cells were transfected with the reporter plasmid G5E1b-CAT the presence of 0.5 mM 8-Br-cAMP increased CAT expression by 1.4 ± 0.25-fold (data not shown). Furthermore, the presence of 0.5 mM 8-Br-cAMP increased CAT expression from the reporter in the presence of the GAL4 DBD (pGALO) by 2.4-fold ( Fig. 2 A). The ability of GAL-RXRγAB to trans -activate gene expression (in a ligand-independent manner) was increased ∼4.0-fold by 8-Br-cAMP treatment ( Fig. 2 A). GAL-RXRγDE increased expression of the reporter by 10-fold in the presence of 8-Br-cAMP (and more importantly, in the absence of the ligand 9- cis -RA). Activation of GAL-RXRγDE in the presence of 8-Br-cAMP was 1.2- to 2-fold greater than that seen in GAL-RXRγDE-transfected cells treated only with 9- cis -RA. Co-treatment of GAL-RXRγDE with 9- cis -RA and 8-Br-cAMP resulted in a similar activation of gene expression compared with treatment with 8-Br-cAMP. For GAL-RXRγABDE results similar to GAL-RXRγDE were observed when treated with either 8-Br-cAMP and/or 9- cis -RA. Figure 2 View largeDownload slide Phosphorylation activates the AF-1 and AF-2 domains of RXRγ. The GAL fusion constructs containing RXRγ or various sub-domains of RXRγ were transfected together with the GAL reporter G5E1b-CAT into COS-1 cells, as described in Figure 1 , in either the absence (−) or presence (+) of 9- cis -RA (10 −7 M), 0.5 mM 8-Br-cAMP ( A ) or 50 nM OA ( B ) (see Materials and Methods for transient transfection and CAT assay details). Representative autoradiograms for each CAT assay of these transfections is shown. The mean CAT activity values and standard deviations (bars) were derived from a triplicate experiment. Figure 2 View largeDownload slide Phosphorylation activates the AF-1 and AF-2 domains of RXRγ. The GAL fusion constructs containing RXRγ or various sub-domains of RXRγ were transfected together with the GAL reporter G5E1b-CAT into COS-1 cells, as described in Figure 1 , in either the absence (−) or presence (+) of 9- cis -RA (10 −7 M), 0.5 mM 8-Br-cAMP ( A ) or 50 nM OA ( B ) (see Materials and Methods for transient transfection and CAT assay details). Representative autoradiograms for each CAT assay of these transfections is shown. The mean CAT activity values and standard deviations (bars) were derived from a triplicate experiment. Activation of the AF-2 domain by 8-Br-cAMP (independent of ligand) was higher than activation mediated by 9- cis -RA and indicates that activation of AF-2 by phosphorylation is not dependant upon hormone binding. Therefore, this set of controlled experiments clearly illustrates that both the AF-1 and AF-2 functions of RXRγ are regulated either directly or indirectly by cAMP-dependent protein kinase cascades in a ligand-independent manner. The ability of GAL-RXRγAB to trans -activate gene expression (in a ligand-independent manner) was increased ∼5.0-fold by OA treatment ( Fig. 2 B). However, in control studies (to examine non-specific effects of OA on the GAL4 hybrid system) when COS-1 cells were transfected with the reporter plasmid G5E1b-CAT the presence of 50 nM OA increased CAT expression by 1.3 ± 0.14-fold (data not shown). Furthermore, 50 nM OA or OA + 9- cis -RA increased CAT expression from the reporter in the presence of the GAL4 DBD by 4.6- and 4-fold respectively. This suggested that the effects of OA on GAL-RXRγAB were not specific to RXR. GAL-RXRγDE increased expression of the reporter by 5-fold in the presence of OA (and more importantly in the absence of the ligand, 9- cis -RA). However, this was 2-fold less than that seen in cells treated only with 9- cis -RA and similar to the increased activity of the GAL 4 DBD after OA treatment. Co-treatment of GAL-RXRγDE with 9- cis -RA and OA resulted in a 4.5-fold greater activation of gene expression compared with treatment with either 9- cis -RA or OA. For GAL-RXRγABDE results similar to GAL-RXRγDE were seen when cells were treated with either OA and/or 9- cis -RA. Our controls demonstrated that it was difficult to interpret the specific effects of OA on RXRγ in the GAL4 hybrid system, because of the generalized effect of OA on the reporter construct and the GAL4 DBD. Therefore, we cannot determine or make firm statements about the role of serine-threonine phosphorylation in the activity of RXRγ, however, we note that OA can activate the LBD in a ligand-independent manner. The N-terminal AF-1 of RXRγ is located between amino acids 1 and 103 We have shown that the N-terminus or AB domain of RXRγ contains a ligand-independent activation function (AF-1) located between amino acids 1 and 138 ( Fig. 1 ). In order to further characterize the AF-1 region of RXRγ we have constructed various deletions of the AB region and fused these sub-domains to the GAL DBD ( Fig. 3 A). These constructs were transfected into COS-1 cells in the absence of ligand and assayed with respect to the ability to trans -activate the reporter ( Fig. 3 B). A construct (GAL-RXRγ1–103) containing the first 103 amino acids of the 138 amino acid AB region of RXRγ increased activation 10.3-fold above the control, pGALO (GAL-DBD) alone, and had similar activity to the entire AB domain of RXRγ (GAL-RXR-γAB). The plasmid GAL-RXRγ104–138 did not trans -activate gene expression in this assay system. This and the previous experiment demonstrate that amino acids 104–138 are not essential for activity of the AB region and do not contain an activation domain. The constructs GAL-RXRγ1–43, GAL-RXRγ44–77 and GAL-RXRγ77–103 all activated the reporter ∼2-fold, indicating that these sub-domains synergistically interacted within the context of the AB region to activate gene expression ∼10-fold. The plasmid GAL-RXRγ77–138 had similar activity to GAL-RXRγ77–103, further demonstrating that amino acids 104–138 do not contain an activation domain. The constructs GAL-RXR-γ44–103 and GAL-RXRγ44–138 activated gene expression 3- to 4-fold, indicating that the sub-domains between 44–77 and 77–103 could interact additively, but not synergistically. In summary, the data demonstrate that the AF-1 domain contains multiple motifs in the first 103 amino acids that function synergistically to activate transcription in a ligand-independent manner. Figure 3 View largeDownload slide AF-1 of RXRγ is located between amino acids 1 and 103. ( A ) A schematic diagram of the GAL fusion constructs containing various deletions of the AB region of RXRγ are shown. The names of the various constructs represent the first and last amino acid (aa) or the beginning/end of internal deletions in the AB region of RXRγ. ( B ) The GAL fusion constructs containing the AB region or various deletions of the AB region of RXRγ were transfected together with the GAL reporter G5E1b-CAT into COS-1 cells, as described in Figure 1 , in the absence of ligand (see Materials and Methods for transient transfection and CAT assay details). A representative autoradiogram of the CAT assay of these transfections is shown. The mean CAT activity values and standard deviations (bars) were derived from a triplicate experiment. Figure 3 View largeDownload slide AF-1 of RXRγ is located between amino acids 1 and 103. ( A ) A schematic diagram of the GAL fusion constructs containing various deletions of the AB region of RXRγ are shown. The names of the various constructs represent the first and last amino acid (aa) or the beginning/end of internal deletions in the AB region of RXRγ. ( B ) The GAL fusion constructs containing the AB region or various deletions of the AB region of RXRγ were transfected together with the GAL reporter G5E1b-CAT into COS-1 cells, as described in Figure 1 , in the absence of ligand (see Materials and Methods for transient transfection and CAT assay details). A representative autoradiogram of the CAT assay of these transfections is shown. The mean CAT activity values and standard deviations (bars) were derived from a triplicate experiment. We then examined the effect of 8-Br-cAMP and OA treatment on the various sub-domains of the AB region to identify potential regions that were targets of phosphorylation ( Fig. 4 ). In this experiment 8-Br-cAMP and OA treatment increased the ability of the GAL DBD to trans -activate gene expression by 2- and 2.8-fold respectively. Although all the GAL-AB sub-domain chimeras were activated by OA treatment, the level of induction was <3-fold, making it difficult to make firm statements about the effects of OA on the AB region. However, the effects of 8-Br-cAMP on the AB region were quite striking in comparison. GAL-RXRγ1–103 was 5.5-fold more active in the presence of 8-Br-cAMP. The plasmid GAL-RXRγ104–138 was only 2-fold more active in the presence of 8-Br-cAMP, hence, with respect to the 2-fold effect of 8-Br-cAMP on the GAL DBD, this and the previous experiment demonstrate that amino acids 104–138 are not essential for the effect of 8-Br-cAMP on activity of the AB region. The constructs GAL-RXRγ1–43, GAL-RXRγ44–77 and GAL-RXRγ77–103 were 5-, 3- and 5-fold respectively more active in the presence of 8-Br-cAMP. The plasmids GAL-RXR-γ77–138 and GAL-RXRγ77–103 were similarly activated by 8-Br-cAMP treatment and further demonstrate that amino acids 104–138 are not involved in AB region phosphorylation. The constructs GAL-RXRγ44–103 and GAL-RXRγ44–138 were 6-and 4-fold more active in the presence of 8-Br-cAMP. These experiments indicate that the AB regions contains multiple motifs in the first 103 amino acids that are targets of cAMP-dependent protein kinase cascades. The AB (AF-1) region of RXRγ is required for optimal ligand-dependant transactivation in muscle cells and functions in a cell-specific manner: RXRγ trans -activates more efficiently in a myogenic background We went on to examine the role of the AB region of RXRγ and the effect of phosphorylating agents in RXR-mediated transactiation from an optimal RXRE (G18) cloned into the heterologous herpes simplex virus thymidine kinase (tk) promoter ( 27 ) linked to the CAT gene in non-muscle and myogenic cells (pG18 2 -tkCAT). We transfected both cell types because RXRγ is selectively expressed in skeletal and cardiac muscle. The G18 RXRE utilized was derived from a RXRγ binding site selection experiment and was the optimal sequence with respect to binding of RXRγ homodimers and RXR-dependent trans -activation in vivo by 9- cis -RA ( 28 ). We investigated the ability of full-length RXRγ, RXRγ lacking the AB region (the RXRγΔAB construct contains amino acids 129–463) and RXRγ lacking the D/E region (the RXRγΔLBD construct contains amino acids 1–229) to transactivate an RXRE in non-muscle and muscle cells in the presence and absence of phosphorylating agents. We utilized the RXRγΔLBD construct as a control construct, since removal of the LBD of RXRs has been shown by Zhang et al. ( 10 ) to abolish receptor function (i.e. homodimerization, ligand binding and transactivation). Therefore, the use of plasmid RXRγΔLBD acts as a proper negative vector control and serves to highlight the contribution of endogenously expressed RXRs ( 35 ). In control studies when COS-1 cells were transfected with the reporter plasmid ptkCAT relative CAT activities in the presence of 0.5 mM 8-Br-cAMP or 50 nM OA compared with untreated cells were 2.1 ± 0.2 and 2.9 ± 0.18 respectively (data not shown). COS-1 cells ( Fig. 5 ) and C 2 C 12 muscle cells ( Fig. 6 ) were co-transfected with the expression vector RXRγΔLBD, RXR-γΔAB or RXRγ and the reporter plasmid pG18 2 -tkCAT. Transfection of reporter plasmid pG18 2 -tkCAT with an expression vector containing the RXRγΔLBD construct in the presence of 9- cis -RA and/or 8-Br-cAMP or OA (in COS-1 cells) trans -activated gene expression 1.6- to 2.3-fold and negligibly in COS-1 and C 2 C 12 cells respectively ( Figs 5 and 6 ). These experiments in COS-1 cells and C 2 C 12 cells verified the inability of RXR lacking the LBD to trans -activate gene expression. When cells were co-transfected with full-length RXRγ and the reporter pG18 2 -tkCAT addition of 9- cis -RA induced a 7.2-fold increase in CAT expression in COS-1 cells ( Fig. 5 ). After treatment with either 8-Br-cAMP or OA CAT expression was increased only 2.8- and 1.8-fold respectively. However, these increases in activation were not significant, as 8-Br-cAMP and OA stimulated CAT expression mediated by RXRγΔLBD ∼2-fold ( Fig. 5 ). Simultaneous 9- cis -RA + 8-Br-cAMP or 9- cis -RA + OA treatment stimulated CAT expression mediated by RXRγ 13.7- and 13.2-fold respectively ( Fig. 5 ). Whether this truly reflects a synergistic activation or simply a generalized/indirect increase mediated by the non-specific effects of 8-Br-cAMP and OA on transcription in COS-1 cells is unclear. Co-transfection with RXRγΔAB and the reporter pG182-tkCAT and addition of 9- cis -RA induced a 5.9-fold increase in CAT expression. Co-treatment with 9- cis -RA and 8-Br-cAMP or OA resulted in 14.2- and 8.8-fold increases respectively ( Fig. 5 ). These studies in COS-1 cells indicate that RXRγΔAB and RXRγ trans -activate gene expression in a similar manner. These results are in agreement with studies by Nagpal et al. ( 13 ), which showed that co-transfection of RXRγ with the AB domain removed (RXR-γΔAB) did not affect activation of an RXRE (DR-1) placed upstream of the tk promoter in COS-1 cells. Figure 4 View largeDownload slide Various deletions of the AB region of RXRγ are augmented by 8-Br-cAMP. The GAL fusion constructs containing the AB region or various deletions of the AB region of RXRγ were transfected together with the GAL reporter G5E1b-CAT into COS-1 cells, as described in Figure 1 , in the absence (−) or presence (+) of 0.5 mM 8-Br-cAMP or 50 nM OA (see Materials and Methods for transient transfection and CAT assay details). A representative autoradiogram of the CAT assay of these transfections is shown. The mean CAT activity values and standard deviations (bars) were derived from a triplicate experiment. Figure 4 View largeDownload slide Various deletions of the AB region of RXRγ are augmented by 8-Br-cAMP. The GAL fusion constructs containing the AB region or various deletions of the AB region of RXRγ were transfected together with the GAL reporter G5E1b-CAT into COS-1 cells, as described in Figure 1 , in the absence (−) or presence (+) of 0.5 mM 8-Br-cAMP or 50 nM OA (see Materials and Methods for transient transfection and CAT assay details). A representative autoradiogram of the CAT assay of these transfections is shown. The mean CAT activity values and standard deviations (bars) were derived from a triplicate experiment. Cell specificity has been found to play an important role in the activation functions (AFs) of the AB and DE domains in the estrogen, glucocorticoid and progesterone receptors and RARs ( 14 , 36 ). The ability of different AFs to function has been found to: (i) vary in relation to the cell line used; (ii) depend on the spatio-temporal expression pattern of the specific receptor. This indicates that cell-specific activation mechanisms are involved in functioning of the different AFs. The RXRγ isoform is preferentially/abundantly expressed in skeletal and cardiac muscle ( 6 , 6 ). Hence, we investigated whether the AF-1 and AF-2 domains of RXRγ activate/function in a cell-specific fashion. C 2 C 12 myogenic cells were co-transfected with receptor expression vector RXRγΔLBD, RXRγΔAB or RXRγ and the reporter plasmid pG182-tkCAT in the presence and absence of 9- cis -RA ( Fig. 6 ). As expected, RXRγΔLBD was unable to activate pG18 2 -tkCAT in the presence or absence of 9- cis -RA. However, full-length RXRγ produced a 19-fold induction of the pG18 2 -tkCAT reporter in a 9- cis -RA-dependent manner ( Fig. 6 ). This 19-fold induction of gene expression by RXRγ in muscle cells was significantly more efficient than the 7-fold induction by RXRγ in COS-1 cells ( Fig. 5 ). In contrast to the observations in COS-1 cells, activation of the reporter by the RXRγΔAB construct after 9- cis -RA treatment was significantly less than that mediated by the native full-length receptor; 11-fold (RXRγΔAB) versus 19-fold (RXRγ) respectively ( Fig. 6 ). Figure 5 View largeDownload slide The effect of 8-Br-cAMP and OA on RXR-mediated transcription. Different pSG5-RXRγ constructs, RXRγΔLBD containing amino acids 1–229 of RXRγ, RXRγΔAB containing amino acids 129–463 of RXRγ or full-length RXRγ (1 µg) were transfected into COS-1 cells together with the reporter pG18 2 -tkCAT (5 µg) containing an optimal RXRE inserted upstream of the herpes simplex virus thymidine kinase (tk) promoter linked to the CAT gene in either the absence (−) or presence (+) of 9- cis -RA (10 −7 M), 0.5 mM 8-Br-cAMP or 50 nM OA (see Materials and Methods for transient transfection and CAT assay details). Representative autoradiograms for each CAT assay of these transfections is shown. The mean CAT activity values and standard deviations (bars) were derived from a triplicate experiment. Figure 5 View largeDownload slide The effect of 8-Br-cAMP and OA on RXR-mediated transcription. Different pSG5-RXRγ constructs, RXRγΔLBD containing amino acids 1–229 of RXRγ, RXRγΔAB containing amino acids 129–463 of RXRγ or full-length RXRγ (1 µg) were transfected into COS-1 cells together with the reporter pG18 2 -tkCAT (5 µg) containing an optimal RXRE inserted upstream of the herpes simplex virus thymidine kinase (tk) promoter linked to the CAT gene in either the absence (−) or presence (+) of 9- cis -RA (10 −7 M), 0.5 mM 8-Br-cAMP or 50 nM OA (see Materials and Methods for transient transfection and CAT assay details). Representative autoradiograms for each CAT assay of these transfections is shown. The mean CAT activity values and standard deviations (bars) were derived from a triplicate experiment. Figure 6 View largeDownload slide The AB (AF-1) region of RXRγ is required for optimal ligand-de-pendent transactivation in muscle cells. Different pSG5-RXRγ constructs, RXRγΔLBD, RXRγΔAB or full-length RXRγ (1 µg) or carrier plasmid pUC18 were transfected into C 2 C 12 muscle cells together with the reporters ptkCAT or pG18 2 -tkCAT (5 µg) in either the absence (−) or presence (+) of 9- cis -RA (10 −7 M) (see Materials and Methods for transient transfection and CAT assay details). Representative autoradiograms for each CAT assay of these transfections is shown. The mean CAT activity values and standard deviations (bars) were derived from a triplicate experiment Figure 6 View largeDownload slide The AB (AF-1) region of RXRγ is required for optimal ligand-de-pendent transactivation in muscle cells. Different pSG5-RXRγ constructs, RXRγΔLBD, RXRγΔAB or full-length RXRγ (1 µg) or carrier plasmid pUC18 were transfected into C 2 C 12 muscle cells together with the reporters ptkCAT or pG18 2 -tkCAT (5 µg) in either the absence (−) or presence (+) of 9- cis -RA (10 −7 M) (see Materials and Methods for transient transfection and CAT assay details). Representative autoradiograms for each CAT assay of these transfections is shown. The mean CAT activity values and standard deviations (bars) were derived from a triplicate experiment These experiments indicate that both the AF-1 and AF-2 domains are required for optimal 9- cis -RA-dependent transactivation of RXRE-dependent reporters in C 2 C 12 muscle cells. Furthermore, and more importantly, these experiments clearly demonstrate that the AB region (AF-1) of RXRγ functions in a cell-specific manner. Discussion We have shown in these studies utilizing the GAL4 hybrid system that RXRγ contains two transactivation functions, AF-1 and AF-2. The first of these is located in the N-terminal AB region (AF-1) and is constitutively active, independent of 9- cis -RA, while the second is located in the DE (LBD) region (AF-2) and functions in a ligand-dependant manner. These experiments are in agreement with previous studies by Nagpal et al. ( 8 , 13 ), which identified two separate domains involved in transactivation by RXRγ, by linkage to the estrogen receptor DBD. Interestingly, the GAL-RXRγABCDE and GAL-RXRγABDE constructs were non-functional in the absence of the ligand, 9- cis -RA. This indicates that in the presence of the ligand-dependent AF-2 domain the function of AF-1 (AB region) is repressed in COS-1 cells. We examined the effect of 8-Br-cAMP and OA on GAL-RXRγ chimeras to assess the ability of phosphorylation to modulate trans -activation of the modular AF-1 and AF-2 domains of RXRγ in the GAL4 hybrid system. Although several studies have reported that protein kinase C- and cAMP-dependant protein kinase pathways are involved in retinoid-mediated transcription ( 19–21 ), our studies are the first to address specific activation of the AF-1 and AF-2 domains of RXRγ by phosphorylation. Our study has demonstrated that both the AF-1 and AF-2 domains of RXRγ are regulated either directly or indirectly by cAMP-dependent protein kinase cascades in a ligand-independent manner. By examination of the amino acid sequence of mRXRγ with respect to PKA consensus phosphorylation sites (RXS, RRXS or RXXS) we identified some putative PKA targets (e.g. RTLS, RVIT, RQRS, RAES and RSVS) in the AB and DE domains of the receptor, which we are evaluating by mutagenesis and transfection studies. Although OA activated the AF-1 (AB) and AF-2 (LBD) regions in the absence of ligand, the generalized effects of OA in the GAL hybrid system obscured the effects of this agent on receptor function. There are various mechanisms by which 8-Br-cAMP and OA could influence transcriptional activation mediated by the AF-1 and AF-2 domains of RXRγ. The AF-1 and AF-2 domains of RXRγ may be phosphorylated, with a consequent change in receptor conformation/activity, whereby the ability to interact with the transcriptional machinery and/or other accessory proteins is enhanced. The use of phosphorylation enhancing agents like 8-Br-cAMP and OA may also phosphorylate proteins in the transcriptional machinery (e.g. TF Il A-J and/or TAFs), other accessory proteins and/or other protein kinase pathways, resulting in a change in their activities. In our study deletion analysis of the AF-1 domain of RXRγ demonstrated that the AB region contained multiple motifs in the first 103 amino acids that function synergistically to activate transcription in a ligand-independent manner and are targets either directly or indirectly of cAMP-dependent protein kinase cascades. The AF-1 domain of RARβ has been analysed using the GAL4 hybrid system in P19 embryonal carcinoma cells and has been shown to be located in the first 32 amino acids of the AB region ( 14 ), which is in contrast to our findings with AF-1 of RXRγ. This indicates that the AF-1 domains of RXRγ and RARβ may function by different mechanisms. Transfection experiments in COS-1 cells revealed that simultaneous treatment with 9- cis -RA and OA or 8-Br-cAMP produced a synergistic activation of an optimal RXRE linked to tkCAT in an RXRγ-dependent manner. However, whether the effects of these agents were direct or indirect was masked by the effects of these phosphorylating agents on the basal reporter in the absence of functional receptor. During the course of this study two recent reports have indicated that OA is able to regulate the DNA binding activities and/or function (independent of ligand) of RXR and RAR ( 23 , 24 ). Differences between those studies and ours probably reflect the use of different response elements and/or cell lines, which have been demonstrated to influence receptor-mediated trans -activation. Important observations of these studies were the demonstration that: (i) RXRγ trans -activated more efficiently in a myogenic background; (ii) the AB region of RXRγ functions in a cell-specific manner and is required for optimal ligand-dependant transactivation of an RXRE in muscle cells. Specifically, full-length RXRγ produced a 7.2- and 19-fold induction of G18 RXRE linked to tkCAT in COS-1 and myogenic C 2 C 12 cells respectively. Furthermore, deletion of the AB region reduced trans -activation from 19- to 11-fold in myogenic cells, whereas in non-muscle cells the impact of this deletion on trans -activation was minimal (7- versus 5.9-fold). Nagpal et al. ( 13 ), using COS-1 cells and RXREs, demonstrated that deletion of the AB domain of RXRγ did not have any significant affect on the ability of the receptor to activate transcription. Our results correlate with the preferential expression of this isoform in skeletal and cardiac muscle. In COS-1 cells activity of GAL-RXRΔAB is contradictory to the similar ability of native RXRγ and RXRγΔAB to trans -activate gene expression (in accordance with Nagpal et al. ; 13 ). This discrepancy is probably a result of the different assay systems involved; the GAL4 hybrid sytem examines the ability of a modular domain to independently complement the GAL4 DBD and trans -activate gene expression, whereas, the other assay examines the function of the entire receptor. The results obtained with our RXRγ deletion constructs in COS-1 and C 2 C 12 muscle cells indicate that the two AFs of RXRγ function by different mechanisms, which could be explained by the presence or absence of cell-specific auxiliary/accessory factors needed for trans -activation in the cell lines tested. The presence of an AF-1 in RXRγ which preferentially activates transcription in a cell-specific manner suggests that RXRγ may function in a programmed spatio-temporal manner. Acknowledgements We sincerely thank Drs David J.Mangelsdorf and Ronald M.Evans for the murine RXRγ1 cDNA, Dr Pierre Chambon for the pSG5 expression vector containing murine RXRγ1 and Dr M.Klaus of Hoffmann-LaRoche Ltd for the 9- cis retinoic acid. This work was supported by the National Health and Medical Research Council (NHMRC) of Australia. We also wish to thank Michael Downes and Dr Amanda Carozzi for excellent technical assistance and helpful discussions. References 1 Sporn M.B.,  Roberts A.B.. ,  Cancer Res. ,  1983, vol.  43 (pg.  3034- 3040) PubMed  2 De-Luca L.M.. ,  FASEB J. ,  1991, vol.  5 (pg.  2924- 2933) PubMed  3 Leid M.,  Kastner P.,  Chambon P.. ,  Trends Biochem Sci. ,  1992, vol.  17 (pg.  427- 433) CrossRef Search ADS PubMed  4 Evans R.M.. ,  Science ,  1988, vol.  240 (pg.  889- 895) CrossRef Search ADS PubMed  5 Green S.,  Chambon P.. ,  Trends Genet. ,  1988, vol.  4 (pg.  309- 315) CrossRef Search ADS PubMed  6 Mangelsdorf D.J.,  Borgmeyer U.,  Heyman R.A.,  Zhou J.Y.,  Ong E.S.,  Oro A.E.,  Kakizuka A.,  Evans R.M.. ,  Genes Dev. ,  1992, vol.  6 (pg.  329- 344) CrossRef Search ADS PubMed  7 Forman B.M.,  Samuels H.H.. ,  Mol. Endocrinol. ,  1990, vol.  4 (pg.  1293- 1301) CrossRef Search ADS PubMed  8 Nagpal S.,  Friant S.,  Nakshatri H.,  Chambon P.. ,  EMBO J. ,  1993, vol.  12 (pg.  2349- 2360) PubMed  9 Durand B.,  Saunders M.,  Gaudon C.,  Roy B.,  Losson R.,  Chambon P.. ,  EMBO J. ,  1994, vol.  13 (pg.  5370- 5382) PubMed  10 Zhang X.-K.,  Salbert G.,  Lee M.-O.,  Pfahl M.. ,  Mol. Cell. Biol. ,  1994, vol.  14 (pg.  4311- 4323) CrossRef Search ADS PubMed  11 Leng X.,  Blanco J.,  Tsai S.Y.,  Ozato K.,  O'Malley B.O.,  Tsai M.-J.. ,  Mol. Cell. Biol. ,  1995, vol.  15 (pg.  255- 263) CrossRef Search ADS PubMed  12 Hadzic E.,  Desai-Yajnik V.,  Helmer E.,  Guo S.,  Wu S.,  Koudinova N.,  Casanova J.,  Raaka B.M.,  Samuels H.H.. ,  Mol. Cell. Biol. ,  1995, vol.  15 (pg.  4507- 4517) CrossRef Search ADS PubMed  13 Nagpal S.,  Saunders M.,  Kastner P.,  Durand B.,  Nakshatri H.,  Chambon P.. ,  Cell ,  1992, vol.  70 (pg.  1007- 19) CrossRef Search ADS PubMed  14 Folkers G.E.,  van der Leede B.-J. M.,  van der Saag P.T.. ,  Mol. Endocrinol. ,  1993, vol.  7 (pg.  616- 27) PubMed  15 Orti E.,  Bodwell J.E.,  Munck A.. ,  Endocrine Rev. ,  1992, vol.  13 (pg.  105- 28) 16 Gaub M.P.,  Rochette E.C.,  Lutz Y.,  Ali S.,  Matthes H.,  Scheuer I.,  Chambon P.. ,  Exp. Cell Res. ,  1992, vol.  201 (pg.  335- 46) CrossRef Search ADS PubMed  17 Rochette-Egly C.,  Gaub M.P.,  Lutz Y.,  Ali S.,  Scheuer I.,  Chambon P.. ,  Mol. Endocrinol. ,  1992, vol.  6 (pg.  2197- 2209) PubMed  18 Rochette-Egly C.,  Lutz Y.,  Saunders M.,  Scheuer I.,  Gaub M.P.,  Chambon P.. ,  J. Cell Biol. ,  1991, vol.  115 (pg.  535- 545) CrossRef Search ADS PubMed  19 Tahayato A.,  Lefebvre P.,  Formstecher P.,  Dautrevaux M.. ,  Mol. Endocrinol. ,  1993, vol.  7 (pg.  1642- 1653) PubMed  20 Huggenvik J.I.,  Collard M.W.,  Kim Y.-W.,  Sharma R.P.. ,  Mol. Endocrinol. ,  1993, vol.  7 (pg.  543- 550) PubMed  21 Rochette-Egly C.,  Oulad-Abdelghani M.,  Staub A.,  Pfister V.,  Scheuer I.,  Chambon P.,  Gaub M.-P.. ,  Mol. Endocrinol. ,  1995, vol.  9 (pg.  860- 871) PubMed  22 Bialojan C.,  Takai A.. ,  Biochem. J. ,  1988, vol.  256 (pg.  283- 290) CrossRef Search ADS PubMed  23 Matkovits T.,  Christakos S.. ,  Mol. Endocrinol. ,  1995, vol.  9 (pg.  232- 242) PubMed  24 Lefebvre P.,  Gaub M.-P.,  Tahayato A.,  Rochette-Egly C.,  Formstecher P.. ,  J. Biol. Chem. ,  1995, vol.  270 (pg.  10806- 10816) CrossRef Search ADS PubMed  25 Kato G.J.,  Barrett J.,  Villa G.M.,  Dang C.V.. ,  Mol. Cell. Biol. ,  1990, vol.  10 (pg.  5914- 5920) CrossRef Search ADS PubMed  26 Leid M.,  Kastner P.,  Lyons R.,  Nakshatri H.,  Saunders M.,  Zacharewski T.,  Chen J.,  Staub A.,  Garnier J.M.,  Mader S.,  Chambon P.. ,  Cell ,  1992, vol.  68 (pg.  377- 395) CrossRef Search ADS PubMed  27 Luckow B.,  SchYtz G.. ,  Nucleic Acids Res. ,  1987, vol.  15 pg.  5490  CrossRef Search ADS PubMed  28 Dowhan D.H.,  Downes M.,  Sturm R.A.,  Muscat G.E.O. ,  Endocrinology ,  1994, vol.  135 (pg.  2595- 2607) PubMed  29 Lillie J.W.,  Green M.R.. ,  Nature ,  1989, vol.  338 (pg.  39- 44) CrossRef Search ADS PubMed  30 Yaffe D.,  Saxel O.. ,  Nature ,  1977, vol.  270 (pg.  725- 727) CrossRef Search ADS PubMed  31 Yaffe D.,  Saxel O.. ,  Differentiation ,  1977, vol.  7 (pg.  159- 166) CrossRef Search ADS PubMed  32 Gorman C.M.,  Moffat L.F.,  Howard B.H.. ,  Mol. Cell. Biol. ,  1982, vol.  2 (pg.  1044- 1051) CrossRef Search ADS PubMed  33 Johnson S.,  Hopper J.. ,  Proc. Natl. Acad. Sci. USA ,  1982, vol.  79 (pg.  6971- 6975) CrossRef Search ADS   34 Laughon A.,  Gesteland R.. ,  Proc. Natl. Acad. Sci. USA ,  1982, vol.  79 (pg.  6827- 6831) CrossRef Search ADS   35 Mader S.,  Leroy P.,  Chen J.-Y.,  Chambon P.. ,  J. Biol. Chem. ,  1993, vol.  268 (pg.  591- 600) PubMed  36 Bocquel M.T.,  Kumar V.,  Stricker C.,  Chambon P.,  Gronemeyer H.. ,  Nucleic Acids Res. ,  1989, vol.  17 (pg.  2581- 2595) CrossRef Search ADS PubMed  37 Dolle P.,  Fraulob V.,  Kastner P.,  Chambon P.. ,  Mech. Dev. ,  1994, vol.  45 (pg.  91- 104) CrossRef Search ADS PubMed  © 1996 Oxford University Press

Graham, Sheila V.; Jefferies, David; Barry, J. David

doi: N/Apmid: N/A

The major surface antigen of procyclic and epimastigote forms of Trypanosoma congolense in the tsetse fly is GARP (glutamic acid/alanine-rich protein), which is thought to be the analogue of procyclin/PARP in Trypanosoma brucei. We have studied two T.congolense GARP loci (the 4.3 and 4.4 loci) whose transcription is α-amanitin sensitive. Whilst a transcriptional gap 5′ of the first GARP gene in the cloned region of the 4.4 locus could not be detected, such a gap was present in the 5′ flank of the first GARP gene in the 4.3 locus. We have located a GARP transcription start site and, using reporter gene constructs containing a putative GARP promoter region in transient transfection studies, we have demonstrated promoter activity for the test region in T.congolense. There are species-specific differences in sequences regulating expression of the two major surface antigens, GARP and procyclin/PARP: the GARP promoter is inactive in T.brucei while the procyclin/PARP promoter is inactive in T.congolense. We have defined the splice acceptor site for the 4.3 GARP gene by sequencing and by 5′ RT-PCR and demonstrated microheterogeneity in GARP polyadenylation by 3′ RT-PCR. It appears that some GARP and procyclin/PARP RNA processing signals, although similar, are also species-specific.

Graham, Sheila V.;Jefferies, David;Barry, J. David

doi: 10.1093/nar/24.2.272pmid: 8628650

Abstract The major surface antigen of procyclic and epimastigote forms of Trypanosoma congolense in the tsetse fly is GARP (glutamic acid/alanine-rich protein), which is thought to be the analogue of procyclin/PARP in Trypanosoma brucei . We have studied two T.congolense GARP loci (the 4.3 and 4.4 loci) whose transcription is α-amanitin sensitive. Whilst a transcriptional gap 5′ of the first GARP gene in the cloned region of the 4.4 locus could not be detected, such a gap was present in the 5′ flank of the first GARP gene in the 4.3 locus. We have located a GARP transcription start site and, using reporter gene constructs containing a putative GARP promoter region in transient transfection studies, we have demonstrated promoter activity for the test region in T.congolense . There are species-specific differences in sequences regulating expression of the two major surface antigens, GARP and procyclin/PARP: the GARP promoter is inactive in T.brucei while the procyclin/PARP promoter is inactive in T.congolense . We have defined the splice acceptor site for the 4.3 GARP gene by sequencing and by 5′ RT-PCR and demonstrated microheterogeneity in GARP polyadenylation by 3′ RT-PCR. It appears that some GARP and procyclin/PARP RNA processing signals, although similar, are also species-specific. Introduction Gene organisation in parasitic protozoa of the order Kinetoplastida is unusual among eukaryotes in that genes are usually found grouped together in polycistronic transcription units ( 1 ). Each polycistronic transcription unit has a single 5′ promoter, and polycistronic pre-mRNAs are processed by the functionally-linked mechanisms of 5′ trans -splicing and 3′ polyadenylation ( 2–5 ). In many cases genes within a polycistronic transcription unit may be differentially expressed, and therefore regulation of gene expression must take place largely at the post-transcriptional level ( 6 ). Investigation of the mechanisms regulating gene expression in trypanosomes has focused on the African trypanosome Trypanosoma brucei ; very little is known about gene expression in other trypanosomatids and whether this is similar to or different from that in T.brucei . Another African trypanosome, Trypanosoma congolense (subgenus Nannomonas ), has a life cycle similar to that of T.brucei (subgenus Trypanozoon ), with two main phases, one in a mammalian host and the other in the insect vector, the tsetse fly. When T.brucei bloodstream forms enter the tsetse fly midgut, differentiation to the procyclic stage occurs ( 7 ). Concomitant with this transformation, parasites lose the variant surface glycoprotein (VSG) coat, continual switching of which allows antigenic variation in the bloodstream ( 8 ) replacing it with a new, densely packed surface coat composed of procyclin/PARP ( 9–12 ). Procyclin/PARP is retained on the surface of insect stage parasites (procyclic and epimastigote) until differentiation to the metacyclic stage occurs in the salivary glands of the fly ( 12 , 13 ). It has been suggested that acquisition of procyclin/PARP might serve to protect the procyclic form from the hostile environment of the tsetse fly midgut and may also have a role in directing differentiating parasites from the midgut to the salivary glands ( 14 , 15 ). The analogue of procyclin/PARP in T.congolense is GARP (glutamic acid/alanine-rich protein) ( 13 , 15 ). GARP displays several features similar to those of procyclin/PARP: the protein is acquired during differentiation from bloodstream to procyclic trypanosomes; it is abundant on the procyclic and epimastigote cell surface ( 13 ); it is acidic and glycosylated ( 13 , 15 ); and is attached to the cell membrane by a GPI anchor (Hulsmeier et al. , unpublished). However, GARP has almost twice the predicted polypeptide molecular mass for procyclin/PARP and has a very different amino acid sequence. DNA fragments homologous to two genomic loci containing at least one GARP gene each have been cloned and characterised ( 15 ) ( Fig. 1 ). Sequence analysis has shown that GARP genes display no homology to procyclin/PARP genes except for a 16 nt motif found in the 3′ untranslated (UTR) region ( 15 ). A procyclin/PARP promoter ( 16 , 17 ) and signals involved in regulating procyclin/PARP RNA processing have been characterised: regulation of procyclin/PARP RNA processing and translation may play a major role in control of procyclin/PARP expression ( 4 , 5 , 18–20 ). In T.brucei procyclin/PARP genes (and the VSG genes) are transcribed by an RNA polymerase I-like enzyme which is insensitive to α-amanitin. The existence of trans -splicing in trypanosomatids means mRNA transcription and capping are not coupled, permitting RNA polymerase I to transcribe protein-coding genes. We wished to begin to understand how regulation of expression of GARP is achieved as a model for understanding control of gene expression in another trypanosome species and to allow us to compare and contrast regulatory mechanisms in gene expression between T.congolense and T.brucei . Study of regulation of gene expression in T.congolense offers a major advantage over such studies in T.brucei since the entire T.congolense life cycle can be achieved in vitro ( 21–23 ). We describe here the identification and characterisation of sequences regulating expression of GARP genes, in particular a promoter for GARP gene expression. GARP transcription is α-amanitin-sensitive. This is the first trypanosomatid promoter identified which directs transcription by an RNA polymerase II-like enzyme. We present evidence that some GARP gene putative regulatory sequences are species-specific: they are not interchangeable with similar regions of sequence from the procyclic/PARP genes from T.brucei . Figure 1 View largeDownload slide Maps of the 5′ ends of the 4.4 and 4.3 GARP loci and the plasmid subclones derived from these. It is not known from which locus the cDNA clone P4 and the subclone pgarp3′utr are derived. Open boxes: GARP genes; horizontal black bars represent fragments of DNA homologous to the 5′ ends of both loci subcloned into pBluescript. Vertical dotted lines delineate the region of 96% homology between the two GARP loci. Abbreviations are: E, Eco RI; H, Hin dIII; Pv, Pvu II; B, Bam HI; Bs, Bst XI. Figure 1 View largeDownload slide Maps of the 5′ ends of the 4.4 and 4.3 GARP loci and the plasmid subclones derived from these. It is not known from which locus the cDNA clone P4 and the subclone pgarp3′utr are derived. Open boxes: GARP genes; horizontal black bars represent fragments of DNA homologous to the 5′ ends of both loci subcloned into pBluescript. Vertical dotted lines delineate the region of 96% homology between the two GARP loci. Abbreviations are: E, Eco RI; H, Hin dIII; Pv, Pvu II; B, Bam HI; Bs, Bst XI. Materials and Methods Trypanosomes Cloned lines of T.congolense TREU1457 and 1/148 were used in these studies as described previously ( 15 ). Procyclic populations were grown in Eagle's MEM supplemented with 2 mM Glutamax (Gibco-BRL) and 20% foetal calf serum (JRH Biosciences) at 27°C, in 5% CO 2 . Recombinant clones The plasmid subclones generated to perform this work are illustrated in Figure 1 , Figure 3 A and B and Figure 4 A. The routes of construction of each subclone and the specific oligonucleotides used to amplify the inserted DNA in p4.35′flank+SA ( Fig. 1 ) and the inserted DNA in p4.35′flank ( Fig. 4 A) are available upon request. All constructs are in pBluescript (KS − ) except p4.3garp3′flank which is in pGEM3. The GARP 3′UTR included in pgarp3′utr was derived from the P4 cDNA clone which was isolated from a different stock (1/148) of T.congolense than the genomic DNA-derived subclones (TREU 1457) used in transcriptional analysis. However, the 3′UTR and 620 bp of GARP downstream intergenic region were subsequently subcloned from λ4.3 to give p4.3garp3′flank, and sequencing showed a 96% identity between the sequence of the two UTRs derived from the different stocks. The recombinant plasmids for transient transfection studies were all derivatives of pJP44, a T.brucei expression construct, which contains, in a 5′ to 3′ direction, the PARP B promoter, a PARP splice acceptor site, a chloramphenicol acetyl transferase (CAT) reporter gene and the 3′ end of the PARP B α gene to provide polyadenylation signals ( 16 ). The PARP promoter region was the 278 bp Kpn I- Sma I fragment; the splice acceptor region was the 90 bp Sma I- Hin dIII fragment and the 3′UTR was the 360 Bam HI- Pst I fragment ( 16 ). For the GARP constructs the 1095 nt region containing the putative GARP promoter and splice acceptor region was amplified by PCR using the 5′E/Pv Bam HI oligonucleotide (5′-CGCGGATCCACTATCCTCCAACATGTG-3′) ( Fig. 4 A) and 3′5′congoprom oligonucleotide (5′-AGCTTCGTTGCACAATGTGTG-3′) ( Fig. 4 A) with Pfu DNA polymerase (Stratagene). The 3′UTR was the 465 bp Bam HI- Kpn I fragment at the 3′ end of the GARP cDNA clone P4 (the Kpn I site is in the plasmid polylinker downstream of the 3′ insertion site). Other plasmid clones used were L29, a T.congolense ribosomal protein cDNA clone (R. Bayne, unpublished) pPRO2001, a T.brucei procyclin/PARP cDNA clone ( 9 ), pTbαβ-T1, a T.brucei plasmid clone containing an α- and β-tubulin repeat unit ( 24 ), pR4 a T.brucei ribosomal DNA repeat unit ( 25 ) and pActine, containing a T.brucei actin gene ( 26 ). DNA sequence analysis Sequencing was carried out on denatured double-stranded plasmid DNA using the dideoxy chain termination method either conventionally (Sequenase kit: Amersham International) or by polymerase chain reaction cycle sequencing on an Applied Biosystems automated sequencer. Sequence for both strands of recombinant plasmids was obtained using the recommended primers for pBluescript or specific primers synthesised on an Applied Biosystems PCR-mate oligonucleotide synthesiser. Computer analysis was carried out using the GCG sequence analysis software package. Nuclear run-on analysis Preparation and storage of nuclei and run-on reactions were carried out exactly as described ( 25 , 27 ). Procyclic run-on reactions were at 27°C, using α-amanitin at a concentration of 500 µg/ml in methanol. Nuclei (10 9 /reaction) were pre-incubated with the drug in nuclei storage buffer for 10 min on ice ( 27 ). Hybridisations were carried out at 55°C in 3× SSC for 48 h and washes were to 0.1× SSC, 0.1% SDS at 65°C. Purification of DNA, RNA, Northern and Southern blotting and hybridisation Standard procedures were used for DNA preparation, gel electrophoresis and Southern blotting onto Hybond N membrane (Amersham International plc). Immobilisation of nucleic acids onto filters was by UV irradiation. RNA was prepared by lithium chloride/urea lysis of trypanosomes followed by phenol extraction ( 28 ). Following DNase I treatment for 1 h in the presence of 100 mM NaCl, 6 mM MgCl 2 and removal of the enzyme by phenol extraction, RNA was fractionated by electrophoresis on denaturing formaldehyde gels following denaturation of 5 µg total RNA by incubation for 10 min in the presence of 50% formamide, 2.2 M formaldehyde ( 29 ). RNA was Northern blotted directly onto Hybond-N membrane and immobilised on the filter by UV irradiation. Radiolabelled probes were prepared by either random hexanucleotide priming of restriction fragments separated by electrophoresis in low melting point gels ( 30 ) or by in vitro transcription of the CAT gene cloned into the vector pBluescript (Stratagene protocol handbook). Hybridisation with the random primed probes was carried out at 42°C in 50% formamide 5× SSC (1× SSC is 150 mM NaCl, 0.015 mM Na citrate), 5× Denhardt's solution, 0.1% SDS, 100 µg/ml herring sperm DNA and blots were washed to 3× SSC or 0.1× SSC, 0.1% SDS at 65°C. Hybridisation with the in vitro transcribed probes was carried out at 55°C in 50% formamide, 5× SET (1× SET is 150 mM NaCl, 10 mM Tris-HCl pH 7.5 and 1 mM EDTA), 5× Denhardt's solution, 50 µg/ml tRNA, 0.5% SDS and washed at 65°C in 0.1× SET, 0.1% SDS. Removal of hybridised probes was carried out as detailed in the Hybond protocol. Following removal of probes filters were autoradiographed to check that no residual hybridisation remained. RNase protection and in vitro transcription In vitro transcription was carried out using T3 and T7 RNA polymerases, and the pBluescript recombinant clone, p4.35′flank as described in the Stratagene pBluescript protocol handbook. RNase protection and fractionation of protected fragments was carried out by standard means ( 29 ). Total RNA (10 µg) was hybridised with radiolabelled RNA probes in 80% formamide, 40 mM PIPES, pH 6.4, 400 mM NaCl, 1 mM EDTA at 50°C overnight. Hybrids were digested with RNaseA (10 µg) and RNase T 1 (3 U) for 30 min at 30°C. Protected fragments were fractionated by electrophoresis in a 6% acrylamide, 7 M urea gel, followed by autoradiography. Transient transfection of trypanosomes Supercoiled, CsCl-purified plasmid DNA (5 µg for CAT assays or 50 µg for RNA extractions per transfection cuvette) was electroporated into procyclic culture cells exactly as described ( 16 , 31 , 32 ) with a single pulse of 1500 V, 25 µF capacitance from a BioRad Gene Pulser. Following electroporation parasites were transferred to 5 ml Eagle's MEM, 20% foetal calf serum per cuvette and cultured overnight (CAT assays) or 5 h (RNA extractions) at 27°C in 5% CO 2 . CAT reactions were for 2 h at 37°C and assays were by xylene extraction ( 32 , 33 ). Transfections were performed in replicate. RNA was prepared from transiently transfected cells by lysis in 3 M LiCl/6 M urea followed by phenol extraction ( 28 ). Prior to Northern blot analysis RNA was DNase I-treated by incubating up to 50 µg RNA in 100 mM NaCl, 50 mM Tris-HCl pH 8.0, 10 mM MgCl 2 , 40 U RNasin (Promega) and 100 µg/ml DNase I (RNase-free, Life Technologies) in a final volume of 100 µl for 1 h at 37°C followed by phenol/chloroform extraction and ethanol precipitation. PCR amplification Using a 5′ RACE kit purchased from Life Technologies, 5′ RT-PCR was carried out exactly as described in the protocol. The primer for first strand synthesis was GARPgsp1 (5′-GCAGTGTGACCGCCATTAAGTGTAG-3′) ( Fig. 4 A) which is homologous to sequences 52–27 bp downstream of the start codon for the GARP gene. The cDNAs were purified from primer and unincorporated nucleotides then tailed with an oligo-dC anchor. The first round of amplification was carried out with oligonucleotide GARPgsp2 (5′-CGTTGCACAATGTGTGAAGAGGAGC-3′) ( Fig. 4 A) which is homologous to sequences 62–88 upstream of the start codon for the GARP gene and the anchor primer supplied with the kit, that contains an oligo-dG anchor region attached to a universal amplification primer region. A second round of PCR was carried out using the oligonucleotide GARPgsp3 (5′-CAAGCAGCGAGCGTGGCG-3′) ( Fig. 4 A) which is homologous to sequences 103–120 upstream of the start codon for the GARP gene, and the universal amplification primer supplied with the kit. PCR amplification was performed for 35 cycles of 30 s at 94°C, 1 min at 55°C (first round of PCR) and 60°C (second round of PCR), 1 min at 70°C in a final volume of 50 µl containing 50 mM KCl, 10 mM Tris-HCl (pH 8.3), 2.5 mM MgCl 2 , 100 µg/ml BSA, 100 pmol of each primer. PCR products were resolved by gel electrophoresis in 1.5% agarose. PCR products were cloned into a T-vector system (Promega), and recombinant plasmids were sequenced using the dideoxy chain termination method (Sequenase kit: Amersham International). 3′ RT-PCR First strand cDNA was synthesised from total RNA isolated from procyclic T.congolense using reverse transcriptase and the oligo [dT]-anchor primer PWM5ANC (5′-CGGTGGCAGCAGCCAACTTTTTTTTTTTT-3′) ( 3 ). For determining the wild-type polyadenylation sites for GARP RNAs, cDNAs were PCR-amplified with oligonucleotides PWMEco (5′-CGAGAATTCGGTGGCAGCAGCCAACT-3′) an Eco RI-tailed anchor primer homologous to the oligo [dT]-anchor primer ( 3 ) and GARPSG1 (5′-CAGATGGTGCCCGTGCCGTGCTGAC-3′) located 80 nt 5′ of the GARP stop codon. One further round of amplification was performed with PWMEco and GARPSG2 (5′ GAGGCGGGATCCCCCAGCTCA 3′) located immediately 3′ of the GARP stop codon ( Fig. 4 B). For analysis of the polyadenylation site of CAT transcripts expressed from transiently transfected T.congolense, first strand CAT cDNAs were synthesised from total RNA isolated from T.congolense cells transiently transfected with p5′garpCAT3′garp, or p-CAT3′garp as a negative control, and were hybrid-selected prior to amplification, using CAT DNA fragments bound to nylon membrane, to exclude recombination with endogenous GARP transcripts. CAT/GARP chimaeric cDNAs were amplified using oligonucleotides PWMEco (5′-CGAGAATTCGGTGGCAGCAGCAACT-3′) ( 3 ) and CATSG4 (5′-GCCCGCCTGATGAATGCTCATCCGG-3′), 470 nt upstream from the CAT stop codon. A second round of amplification was carried out by nested PCR with oligonucleotide CATSG1 (5′-TGGCAGGGCGGGGGTAA-3′) 18 nt upstream from the CAT stop codon and PWM5Eco. PCR amplifications were performed for 35 cycles of 30 s at 94°C, 1 min at 60°C and 1 min at 70°C in a final volume of 50 µl containing 50 mM KCl, 10 mM Tris-HCl (pH 8.3), 2.5 mM MgCl 2 , 100 µg/ml BSA and 100 pmol of each primer. PCR products were resolved by gel electrophoresis in 1.5% agarose. Results GARP gene loci Previously we had isolated three lambda clones (λ4.3, λ4.4, λ4.5) containing sequences homologous to cDNA P4 which encodes the GARP protein ( 15 ). Two cDNA P4-homologous regions are contained in λ4.3, while λ4.4 and λ4.5 contain one such region each. Partial sequencing had shown that sequence identity was very high between the genes from the 4.3 and 4.4 loci, at least in the 3′ region of the open reading frame ( 15 ). Further sequence analysis has now demonstrated a high degree of sequence similarity between the 5′ ends of the GARP genes from each locus ( Fig. 5 ). Subsequent analysis of the genomic clones showed that λ4.5 was derived from the same locus as λ4.3 (data not shown). Partial maps of the 4.3 and 4.4 GARP loci are shown in Figure 1 . GARP transcription is sensitive to α-amanitin We synthesised 32 P-labelled nascent transcript probes by in vitro run-on using nuclei pre-incubated with α-amanitin for 10 min on ice ( Fig. 2 , + α-amanitin) or methanol, the solvent for the drug ( Fig. 2 , − α-amanitin). The nascent transcript probes were hybridised to identical Southern blots of restriction digests of L29, a T.congolense cDNA clone encoding a ribosomal protein (lanes 1 and 5) (R. Bayne, unpublished), the P4 GARP cDNA clone ( 15 ) (lanes 2 and 6), a T.brucei α- and β-tubulin DNA repeat unit ( 24 ) (lanes 3 and 7) and a T.brucei ribosomal DNA repeat unit ( 27 ) (lanes 4 and 8). The P4 GARP cDNA insert hybridised only with the probe synthesised in nuclei which had not been pre-incubated with α-amanitin ( Fig. 2 , lane 6). There was no detectable hybridisation when the cDNA insert was hybridised with transcripts from α-amanitin-treated nuclei ( Fig. 3 , lane 2). Hybridisation to T.brucei tubulin sequences was decreased also in the presence of the drug ( Fig. 3 , compare lanes 3 and 7) showing that α-amanitin was inhibiting RNA polymerase II. There was no decrease in RNA polymerase I transcription of ribosomal DNA (lanes 4 and 8) as expected. Transcription of T.congolense L29 ribosomal protein gene(s), however may be insensitive to the drug since we found consistently that transcription of L29 sequences was not completely inhibited by α-amanitin (lanes 1 and 5). Promoter localisation We performed nuclear run-on analysis to determine whether there was a transcriptional gap, and therefore a putative promoter located 5′ of the 4.3 or 4.4 GARP loci. The 32 P-labelled nascent transcripts run-on in nuclei isolated from procyclic T.congolense were hybridised to Southern blots of a Hin dIII digest of pE/Pv4.4 (the inserted DNA in this subclone is a Pvu II fragment containing the Eco RI- Pvu II fragment shown in Figure 1 but flanked 5′ with a Pvu II- Eco RI fragment from the very 3′ end of the left hand arm of the lambda clone. The DNA is inserted in the Hin cII site in pBluescript, thus the second Hin dII site is in the 5′ polylinker of the plasmid) ( Fig. 3 B and E) or to a Bst XI digest (there is a Bst XI site in the 5′ polylinker of the plasmid) of pE/Pv4.3 ( Fig. 3 A and C). For the 4.4 locus, hybridisation was detected to both Hin dIII fragments ( Fig. 3 F) indicating that transcription occurs across this region. The larger Hin dIII fragment contains 2960 bp of plasmid sequence plus 1070 bp of GARP-related sequence, and we cannot discount the possibility that there may be another gene 5′ of the GARP gene in the 4.4 locus but part of another transcription unit. In this case a small gap in transcription may be present within this fragment which would not be detected in this crude assay. For the 4.3 locus we detected hybridisation only to the larger Bst XI fragment (2.96 kb of plasmid DNA and 728 bp of GARP-related sequence) ( Fig. 3 D). The run-on probe did not hybridise to plasmid DNA alone (data not shown). Since we did not detect hybridisation to the smaller, upstream fragment (572 bp), a transcriptional gap must exist upstream of the 4.3 GARP locus, and it is reasonable to assume that transcription of the 4.3 GARP locus initiates, and a promoter for GARP gene transcription may be present, within the cloned region 5′ of the 4.3 GARP genes around or downstream of the Bst XI site. Figure 2 View largeDownload slide Nuclear run-on analysis of α-amanitin sensitivity of GARP transcription. + α-amanitin: Southern blot hybridised with 32 P-labelled nascent transcripts run-on in nuclei pre-incubated in the presence of 500 µg/ml α-amanitin for 10 min on ice.—α-amanitin: Southern blot hybridised with 32 P-labelled transcripts run-on in nuclei pre-incubated with 100 µl methanol, the solvent for the drug, for 10 min on ice. Lanes 1 and 4, Xho I digest of L29, a cDNA plasmid clone encoding a ribosomal protein (R. Bayne, unpublished); lanes 2 and 5; Pst I digest of P4 the GARP cDNA clone ( 15 ); lanes 3 and 6, Hin dIII digest of pTBα,β-T1 (an α-and β-tubulin repeat unit clone) ( 24 ); lanes 4 and 8, Pst I digest of pR4, a ribosomal DNA repeat unit clone ( 25 ). Hybridisations were in 3x SSC at 55°C and blots were washed to 0.1x SSC at 65°C. Identical quantities of DNA are loaded in each pair of tracks and identical autoradiograph exposures are shown. Figure 2 View largeDownload slide Nuclear run-on analysis of α-amanitin sensitivity of GARP transcription. + α-amanitin: Southern blot hybridised with 32 P-labelled nascent transcripts run-on in nuclei pre-incubated in the presence of 500 µg/ml α-amanitin for 10 min on ice.—α-amanitin: Southern blot hybridised with 32 P-labelled transcripts run-on in nuclei pre-incubated with 100 µl methanol, the solvent for the drug, for 10 min on ice. Lanes 1 and 4, Xho I digest of L29, a cDNA plasmid clone encoding a ribosomal protein (R. Bayne, unpublished); lanes 2 and 5; Pst I digest of P4 the GARP cDNA clone ( 15 ); lanes 3 and 6, Hin dIII digest of pTBα,β-T1 (an α-and β-tubulin repeat unit clone) ( 24 ); lanes 4 and 8, Pst I digest of pR4, a ribosomal DNA repeat unit clone ( 25 ). Hybridisations were in 3x SSC at 55°C and blots were washed to 0.1x SSC at 65°C. Identical quantities of DNA are loaded in each pair of tracks and identical autoradiograph exposures are shown. Figure 3 View largeDownload slide Nuclear run-on analysis of transcription of the 5′ ends of the 4.3 and 4.4 loci. ( A ) Map of the 5′ end of the 4.3 GARP locus. ( B ) Map of the 5′ end of the 4.4 GARP locus. The open boxes represent GARP coding regions. The horizontal thick black lines represent DNA homologous to the loci shown above subcloned into pBluescript (thin black lines). Abbreviations: E, Eco RI, Bs; Bst XI, Pv; Pvu II, H; Hin dIII. Numbers 1–4 refer to the digestion products of the subclones. P refers to the pBluescript plasmid portion of the subclones. ( C ) An ethidium bromide-stained gel of a Bst XI digest (there is a Bst XI site in the 5′ polylinker of the plasmid) of the plasmid clone pE/Pv4.3garp. ( D ) Southern blot of the gel in C hybridised with 32 P-labelled nascent transcripts synthesised in vitro in procyclic T.congolense nuclei. ( E ) An ethidium bromide-stained gel of a Hin dIII digest of the plasmid clone pE/Pv4.4garp (there is a Hin dIII site in the 5′ polylinker of the plasmid). ( F ) Southern blot of the gel in E hybridised with 32 P-labelled nascent transcripts synthesised in vitro in procyclic T.congolense nuclei. Hybridisations were in 3x SSC at 55°C and blots were washed to 0.1x SSC at 65°C. Figure 3 View largeDownload slide Nuclear run-on analysis of transcription of the 5′ ends of the 4.3 and 4.4 loci. ( A ) Map of the 5′ end of the 4.3 GARP locus. ( B ) Map of the 5′ end of the 4.4 GARP locus. The open boxes represent GARP coding regions. The horizontal thick black lines represent DNA homologous to the loci shown above subcloned into pBluescript (thin black lines). Abbreviations: E, Eco RI, Bs; Bst XI, Pv; Pvu II, H; Hin dIII. Numbers 1–4 refer to the digestion products of the subclones. P refers to the pBluescript plasmid portion of the subclones. ( C ) An ethidium bromide-stained gel of a Bst XI digest (there is a Bst XI site in the 5′ polylinker of the plasmid) of the plasmid clone pE/Pv4.3garp. ( D ) Southern blot of the gel in C hybridised with 32 P-labelled nascent transcripts synthesised in vitro in procyclic T.congolense nuclei. ( E ) An ethidium bromide-stained gel of a Hin dIII digest of the plasmid clone pE/Pv4.4garp (there is a Hin dIII site in the 5′ polylinker of the plasmid). ( F ) Southern blot of the gel in E hybridised with 32 P-labelled nascent transcripts synthesised in vitro in procyclic T.congolense nuclei. Hybridisations were in 3x SSC at 55°C and blots were washed to 0.1x SSC at 65°C. Figure 4 View largeDownload slide RNase protection analysis of the transcriptional start site for the 4.3 GARP gene locus. ( A ) Map of the 4.3 GARP gene locus and, below, the pBluescript subclone p4.35′flank, containing a PCR-amplified fragment stretching from the nt immediately 5′ of the splice acceptor site for the 5′-most 4.3 GARP gene to the Eco RI site 1.1 kb upstream. Primers used to amplify this fragment introduced a Bam HI site at the 5′ end and an Eco RI site at the 3′ end. These are shown in Figure 5 A as 5′E/Pv Bam HI and 3′E/Pv Eco RI which is equivalent to gsp2 but with a 5′ flanking EcoRI site. Open boxes: GARP coding regions, closed box: PCR-amplified insert in the subclone pB/Epcr5′4.3; B and E, Bam HI and Eco RI sites generated by PCR at the ends of the inserted sequence in pB/Epcr5′4.3; flags: bacterial promoters T7 and T3 flanking the polylinker sequence in the pBluescript subclone; arrowhead: major protected fragment. Abbreviations are exactly as for Figure 1 . ( B ) Sequencing gel showing fractionation by electrophoresis of RNase protected fragments. Products were run next to a sequencing ladder (ACGT) as a size marker. Lane 1, T.congolense total RNA hybridised with an antisense probe for the 5′ flank of the 4.3 GARP gene; lane 2, E.coli total RNA hybridised with the same probe; lane 3, T.congolense total RNA hybridised with a sense probe for the 5′ flank of the 4.3 GARP gene; lane 4, E.coli total RNA hybridised with the sense probe. M is a double-stranded DNA marker made by end-labelling a 1 kb ladder (Life technologies), and sizes are shown in bp. Abbreviations: T3, antisense RNA probe synthesised with the pB/Epcr54.3 plasmid template and T3 RNA polymerase; T7, sense RNA probe synthesised with the pB/Epcr54.3 plasmid template and T7 RNA polymerase; T.c., probe hybridised with total T.congolense RNA; E.c., probe hybridised with E.coli tRNA. Figure 4 View largeDownload slide RNase protection analysis of the transcriptional start site for the 4.3 GARP gene locus. ( A ) Map of the 4.3 GARP gene locus and, below, the pBluescript subclone p4.35′flank, containing a PCR-amplified fragment stretching from the nt immediately 5′ of the splice acceptor site for the 5′-most 4.3 GARP gene to the Eco RI site 1.1 kb upstream. Primers used to amplify this fragment introduced a Bam HI site at the 5′ end and an Eco RI site at the 3′ end. These are shown in Figure 5 A as 5′E/Pv Bam HI and 3′E/Pv Eco RI which is equivalent to gsp2 but with a 5′ flanking EcoRI site. Open boxes: GARP coding regions, closed box: PCR-amplified insert in the subclone pB/Epcr5′4.3; B and E, Bam HI and Eco RI sites generated by PCR at the ends of the inserted sequence in pB/Epcr5′4.3; flags: bacterial promoters T7 and T3 flanking the polylinker sequence in the pBluescript subclone; arrowhead: major protected fragment. Abbreviations are exactly as for Figure 1 . ( B ) Sequencing gel showing fractionation by electrophoresis of RNase protected fragments. Products were run next to a sequencing ladder (ACGT) as a size marker. Lane 1, T.congolense total RNA hybridised with an antisense probe for the 5′ flank of the 4.3 GARP gene; lane 2, E.coli total RNA hybridised with the same probe; lane 3, T.congolense total RNA hybridised with a sense probe for the 5′ flank of the 4.3 GARP gene; lane 4, E.coli total RNA hybridised with the sense probe. M is a double-stranded DNA marker made by end-labelling a 1 kb ladder (Life technologies), and sizes are shown in bp. Abbreviations: T3, antisense RNA probe synthesised with the pB/Epcr54.3 plasmid template and T3 RNA polymerase; T7, sense RNA probe synthesised with the pB/Epcr54.3 plasmid template and T7 RNA polymerase; T.c., probe hybridised with total T.congolense RNA; E.c., probe hybridised with E.coli tRNA. Transcription initiation site for the 4.3 locus 5′GARP gene Having located a transcriptional gap 5′ of the 4.3 5′-most GARP gene, we localised the transcription initiation site by RNase protection ( Fig. 4 ) and 5′ RT-PCR. For RNase protection analysis we used the recombinant pBluescript plasmid template, p4.35′flank. The inserted DNA in this subclone is homologous to the 1.1 kb of sequence flanking the splice acceptor site in λ 4.3 ( Fig. 5 A) and was amplified by PCR using primers 5′E/Pv Bam HI and 3′E/Pv Eco RI ( Fig. 5 A) to generate a Bam HI site at the 5′ end and an Eco RI site at the 3′ end. This subclone was used to synthesise in vitro, labelled sense transcripts (T7 RNA polymerase), or antisense transcripts (T3 RNA polymerase) homologous to the putative 5′ end of the GARP transcription unit. The sense and antisense radiolabelled transcripts were hybridised to either total T.congolense RNA or to Escherichia coli tRNA. Following RNase digestion of hybrids and fractionation of digestion products on a sequencing gel, a major protected fragment of around 440 bases and a minor product at around 330 bases were observed when the antisense T3 RNA polymerase-synthesised transcripts from p4.35′flank were hybridised with total T.congolense RNA ( Fig. 4 B, lane 1) but not when the same transcripts were hybridised with E.coli tRNA ( Fig. 4 B, lane 2). Hybridisation of T7 RNA polymerase-synthesised sense transcripts with either total T.congolense RNA ( Fig. 4 B, lane 3) or E.coli tRNA ( Fig. 4 , lane 4) gave no protected product, as expected. If the large protected fragment represents a primary transcript for the 5′ end of the 4.3 GARP locus then this places the transcription initiation site around 460 bp upstream of the start of the cDNA. The minor protected fragment may be a degradation product of the major fragment; it may represent transcription from a secondary promoter; or it may be a fragment from another GARP locus which we have not yet identified, that has a shorter region of homology with 5′ sequences of the 4.3 GARP locus. Next we used 5′RT-PCR to confirm this result and to locate the specific nucleotide(s) where transcription initiated. The oligonucleotide used to direct first strand cDNA synthesis was a 25mer homologous to nucleotides 50–25 3′ of the ATG start codon of the 4.3 GARP gene (GARPgsp1; Fig. 5 A). Two major products were obtained, one at around 130 nt corresponding to the spliced mature RNA and one at around 550 nt (data not shown). Two further oligonucleotides (GARPgsp2, gsp3) were used to prime two subsequent rounds of PCR, and these are indicated by long arrows below the sequence in Figure 5 A. A single major PCR product was obtained (around 450 bp) and amplified DNAs were cloned and sequenced. All five clones sequenced gave the same initiation site corresponding to a G residue 466 nt upstream of the cDNA start site indicated by the small arrowhead in Figure 5 A. There are no significant open reading frames within the entire 1.1 kb fragment 5′ of the cDNA start and stop codons are present in all frames. Figure 5 View largeDownload slide Sequence of the putative promoter region, 3′ UTR and a portion of the intergenic region downstream for the 4.3 GARP locus. ( A ) The putative promoter region from the Eco RI site 1.1 kb upstream of the cDNA start to the transcriptional start site indicated with a small arrowhead at 645 bp; the 5 UTR including the splice acceptor region which stretches from the square bracket at 834 bp to the splice acceptor site shown in bold type at 1088 nt; the cDNA start site indicated with a large arrowhead at 1110 bp, followed by the ATG start codon (boxed) at 1149 bp and 50 bp downstream of the GARP start codon. The Eco RI and Bst XI sites are underlined and the primers used to amplify the region 5′ of the splice acceptor site by PCR or the primers used in 5′ RT-PCR to locate the transcriptional start site are indicated by bold arrows below the sequence. The primer 3′E/Pv Eco RI exactly overlaps the primer gsp2 with an Eco RI site attached at the 5′ end. The region bearing 96% homology with the 4.4 locus starts at the square bracket and is indicated with a vertical bar to the right of the sequence. The remainder of the sequence 5 has 48% homology with the 4.4 locus. Polypyrimidine tracts of more than 5 nt length are overlined .(B ) The 3′ UTR of the 5′-most GARP gene and a portion of the downstream intergenic region. The sequence begins at the TGA (shown in bold) stop codon for the GARP coding region and the Bam HI site used for subcloning the 3′ UTR of the GARP gene is overlined. The second primer used to amplify the GARP 3′ UTR by 3′ RT-PCR is indicated by a bold arrow below the sequence. Polyadenylation sites are indicated with arrowheads. The 16 nt motif conserved between PARP and GARP cDNAs is boxed. Polypyrimidine tracts in the 5 portion of the intergenic region are underlined and YAG trinucleotides are circled. The next open reading frame is 2.6 kb downstream. Figure 5 View largeDownload slide Sequence of the putative promoter region, 3′ UTR and a portion of the intergenic region downstream for the 4.3 GARP locus. ( A ) The putative promoter region from the Eco RI site 1.1 kb upstream of the cDNA start to the transcriptional start site indicated with a small arrowhead at 645 bp; the 5 UTR including the splice acceptor region which stretches from the square bracket at 834 bp to the splice acceptor site shown in bold type at 1088 nt; the cDNA start site indicated with a large arrowhead at 1110 bp, followed by the ATG start codon (boxed) at 1149 bp and 50 bp downstream of the GARP start codon. The Eco RI and Bst XI sites are underlined and the primers used to amplify the region 5′ of the splice acceptor site by PCR or the primers used in 5′ RT-PCR to locate the transcriptional start site are indicated by bold arrows below the sequence. The primer 3′E/Pv Eco RI exactly overlaps the primer gsp2 with an Eco RI site attached at the 5′ end. The region bearing 96% homology with the 4.4 locus starts at the square bracket and is indicated with a vertical bar to the right of the sequence. The remainder of the sequence 5 has 48% homology with the 4.4 locus. Polypyrimidine tracts of more than 5 nt length are overlined .(B ) The 3′ UTR of the 5′-most GARP gene and a portion of the downstream intergenic region. The sequence begins at the TGA (shown in bold) stop codon for the GARP coding region and the Bam HI site used for subcloning the 3′ UTR of the GARP gene is overlined. The second primer used to amplify the GARP 3′ UTR by 3′ RT-PCR is indicated by a bold arrow below the sequence. Polyadenylation sites are indicated with arrowheads. The 16 nt motif conserved between PARP and GARP cDNAs is boxed. Polypyrimidine tracts in the 5 portion of the intergenic region are underlined and YAG trinucleotides are circled. The next open reading frame is 2.6 kb downstream. GARP splice acceptor and polyadenylation sites The site for addition of the spliced leader sequence was predicted to be about 20 bp upstream of the start of the cDNA homologous to the GARP gene in the 4.4 locus ( 15 ). RNase protection studies and 5′RT-PCR indicated that a spliced leader addition sequence was located at the same distance upstream of the first GARP gene in the 4.3 locus (AG in bold type at position 1088 in Fig. 5 A) (data not shown). Sequencing upstream of the AG dinucleotide splice acceptor site for both loci revealed that there was a 258 bp region with 96% sequence identity between the two GARP gene loci (running 3′ from the square bracket in Fig. 5 A) 5′ of which the homology dropped to 48% identity. Since these homologous sequences contain the splice acceptor sites for the 5′ GARP genes in both loci it is probable that these regions contain the sequences necessary to direct trans -splicing of the GARP genes in T.congolense . We used 3′RT-PCR to determine where GARP transcripts were polyadenylated. Figure 5 B shows the 3′ UTR and some 350 bp of intergenic region downstream. First strand cDNA synthesis was primed with oligo [dT] then two nested oligonucleotides, the first homologous to a sequence at the 3′ end of the GARP coding region and the second homologous to a sequence at the 5′ end of the 3′ UTR of GARP ( Fig. 5 B) were used to prime two subsequent rounds of PCR. A single major PCR product was obtained and amplified DNAs were cloned and sequenced. We found a range of sites of polyadenylation within a 25 nt region in the 3′ UTR of GARP mRNAs which was located 468–493 nt downstream of the translation stop codon (arrowheads in Fig. 5 B). Sequencing of part of the intergenic region downstream revealed a distribution of sequence motifs (underlined and circled in Fig. 5 B) similar to those found for intergenic regions in T.brucei ( 4 ). Functional analysis of sequences controlling GARP gene expression by transient transfection of trypanosomes In order to test whether the GARP putative promoter region was functional and to determine whether sequences involved in regulating GARP gene expression in T.congolense could be recognised in T.brucei, we carried out a series of transient transfection assays ( Fig. 6 ). We compared the ability of the procyclin/PARP promoter and the putative GARP promoter to drive expression of a CAT reporter gene in the constructs shown in Figure 6 A and B. When we transiently transfected T.congolense cells with the construct p5′garpCAT3′garp (644 bp of sequence 5′ of the transcription start site and the entire putative splice acceptor site with the insert of pgarp3′utr as the 3′ UTR) we could not detect CAT activity, even with a range of protease inhibitors included in the transfection buffer and the cell lysis buffer, but CAT RNA was readily detectable ( Fig. 6 A, lane 4). This indicated that we had indeed identified a region of sequence 5′ of the first GARP gene in the 4.3 locus which could act as a promoter in this assay. Either CAT enzyme is highly unstable in T.congolense cells or the CAT RNA is not able to be translated. Thus, rather than assaying CAT enzyme activity we analysed the abundance of CAT transcripts in the transiently transfected cells. The GARP constructs numbered 1, 2 and 4 in Figure 6 A contained a GARP 3′ UTR isolated from cDNA P4. In T.brucei it has been shown that, in transient transfection experiments, inclusion of only the 3′ UTR of a procyclin/PARP cDNA downstream of the CAT gene is not sufficient to specify positionally accurate polyadenylation of CAT transcripts ( 4 , 5 , 20 ). However, it does allow polyadenylation of PARP transcripts but at a site around 100 bases 5′ of the site used in vivo ( 20 ). For T.congolense we found by 3′ RT-PCR that the 3′ UTR used in the transient transfection studies did direct polyadenylation but at a novel site 120 bases 5′ of the polyadenylation site used in the cDNA clone P4. Inclusion of a further 620 bp 3′ of the endogenous polyadenylation site which includes part of the intergenic region flanking the next GARP gene downstream had no effect on CAT RNA abundance in transiently transfected cells (data not shown), but still did not allow us to detect CAT enzyme activity. This intergenic region includes several sequences similar to the types of signal (an extensive polypyrimidine tract, an AG dinucleotide putative splice site flanked 3′ by a short polypyrimidine tract, Fig. 5 B) which have been shown to be important for regulation of polyadenylation in T.brucei ( 4 ). As negative controls for promoter activity we used the same plasmids from which we removed the promoter regions but retained the splice acceptor regions: for GARP this was the region downstream of the square bracket in Figure 5 A (p-CAT3′congo; Figure 6 A lane 2). For PARP the splice acceptor region was the Sma I -Hin dIII fragment from pJP44 (p-CAT3′parp; Fig. 6 B lane 6). No CAT transcripts were detected in either T.congolense or T.brucei cells transiently transfected with these constructs ( Fig. 6 C, lane 2, Fig. 6 D, lane 6). Next we asked if the procyclin/PARP promoter could direct CAT expression in transient transfection of T.congolense and conversely, if the GARP putative promoter region was operative in driving CAT gene expression in transient transfection of T.brucei ( Fig. 6 A construct 1, Fig. 6 B construct 5). Initial studies using p5′parpCAT3′parp to transiently transfect T.congolense and p5′garpCAT3′garp to transiently transfect T. brucei yielded no CAT RNA (data not shown). Similarly, Figure 6 C lane 1 shows that steady-state levels of CAT transcripts were not produced using the PARP promoter/splice acceptor site in p5′parpCAT3′garp in T.congolense transient transfection. Figure 6 D lane 5 shows that no CAT transcripts were detected in T.brucei cells transiently transfected with the GARP promoter construct p5′garpCAT3′parp. Finally, we had observed that the 3′ UTRs of GARP and procyclin/PARP transcripts shared a 16mer motif at approximately the same distance upstream of the poly(A) addition site ( 15 ). This suggested that in both species, a similar mechanism might operate to regulate procyclin/PARP and GARP gene expression mediated through the 3′ end of the mRNAs. It was therefore possible that sequences at the 3′ end of the gene were not entirely species-specific. To test this possibility we exchanged the 3′ UTRs of the GARP and procyclin/PARP cDNAs in the constructs p5′parpCAT3′parp and p5′garp-CAT3′garp to give p5′parpCAT3′garp and p5′garpCAT3′parp. The result shown in Figure 6 C, lane 3 indicates that replacement of the GARP gene 3′ UTR with the corresponding region of the procyclin/PARP gene does not allow efficient expression of CAT in transient transfection experiments in T.congolense . Similarly, Figure 6 D, lane 7 indicates that replacement of the procyclin/ PARP 3′ UTR by the GARP gene 3′ end does not allow efficient expression of CAT in transient transfection of T.brucei . For all these experiments, replicate transient transfection experiments were always very reproducible. We also assayed CAT activity in T.brucei, transiently transfected with the constructs shown in Figure 6 B and obtained results consistent with those obtained by measuring CAT RNA abundance in Figure 6 D. Rehybridisation of the Northern blots in Figure 6 C and D with a T.brucei actin probe showed that failure to detect CAT transcripts in tracks 1–3 and 5–7 was not due to lack of RNA in each track ( Fig. 6 E and F). These experiments demonstrate that in both flanks there are significant species-specific differences in sequences regulating gene expression in T.brucei and T.congolense . Figure 6 View largeDownload slide Northern blot analysis of CAT reporter gene expression from the GARP promoter. ( A ) Schematic illustrations of the GARP-based constructs: 1, p5′parpCAT3′garp; 2, p-CAT3′garp; 3, p5′garpCAT3′parp; 4, p5′garpCAT3′garp. ( B ) Schematic illustrations of the PARP-based constructs: 5, p5′garpCAT3′parp; 6, p-CAT3′parp; 7, p5′parpCAT3′garp; 8, p5′parpCAT3′parp. Abbreviations: prom, promoter; sa, splice acceptor region; CAT, chloramphenicol acetyl transferase gene; 3′, 3′ UTR. ( C ) Northern blot analysis of total RNA isolated from 10 7T.congolense cells/lane transiently transfected with 50 µg/10 8 cells of the DNA constructs schematically illustrated on the left in A. ( D ) Northern blot analysis of total RNA isolated from 10 7T.brucei cells/lane transiently transfected with 50 µg/10 8 cells of the DNA constructs schematically illustrated on the left in B. Both Northern blots were hybridised with a 32 P-labelled CAT antisense RNA in 50% formamide, 5x SET (1xSET is 150 mM NaCl, 10 mM Tris-HCl pH 7.5 and 1 mM EDTA), 5x Denhardt's solution, 50 µg/ml tRNA, 0.5% SDS at 55°C for 16 h. Blots were washed to 0.1x SET, 0.5% SDS. ( E ) and ( F ) Rehybridisation of the blots in C and D with a 32 P-labelled T.brucei actin probe prepared by random hexanucleotide priming of the gel-purified insert from pActine ( 26 ). The blots in C and D were stripped of probe and autoradiographed overnight to check that no hybridisation remained before rehybridisation. Hybridisation was in 50% formamide, 5x SSC, 0.1% SDS at 42°C for 16 h. The blot in E (a cross-species hybridisation) was washed to 3x SSC at 65°C while the blot in F (homologous hybridisation) was washed to 0.1x SSC at 65°C. Figure 6 View largeDownload slide Northern blot analysis of CAT reporter gene expression from the GARP promoter. ( A ) Schematic illustrations of the GARP-based constructs: 1, p5′parpCAT3′garp; 2, p-CAT3′garp; 3, p5′garpCAT3′parp; 4, p5′garpCAT3′garp. ( B ) Schematic illustrations of the PARP-based constructs: 5, p5′garpCAT3′parp; 6, p-CAT3′parp; 7, p5′parpCAT3′garp; 8, p5′parpCAT3′parp. Abbreviations: prom, promoter; sa, splice acceptor region; CAT, chloramphenicol acetyl transferase gene; 3′, 3′ UTR. ( C ) Northern blot analysis of total RNA isolated from 10 7T.congolense cells/lane transiently transfected with 50 µg/10 8 cells of the DNA constructs schematically illustrated on the left in A. ( D ) Northern blot analysis of total RNA isolated from 10 7T.brucei cells/lane transiently transfected with 50 µg/10 8 cells of the DNA constructs schematically illustrated on the left in B. Both Northern blots were hybridised with a 32 P-labelled CAT antisense RNA in 50% formamide, 5x SET (1xSET is 150 mM NaCl, 10 mM Tris-HCl pH 7.5 and 1 mM EDTA), 5x Denhardt's solution, 50 µg/ml tRNA, 0.5% SDS at 55°C for 16 h. Blots were washed to 0.1x SET, 0.5% SDS. ( E ) and ( F ) Rehybridisation of the blots in C and D with a 32 P-labelled T.brucei actin probe prepared by random hexanucleotide priming of the gel-purified insert from pActine ( 26 ). The blots in C and D were stripped of probe and autoradiographed overnight to check that no hybridisation remained before rehybridisation. Hybridisation was in 50% formamide, 5x SSC, 0.1% SDS at 42°C for 16 h. The blot in E (a cross-species hybridisation) was washed to 3x SSC at 65°C while the blot in F (homologous hybridisation) was washed to 0.1x SSC at 65°C. Discussion We have shown that transcription of GARP genes in T.congolense is sensitive to α-amanitin. We have identified a gap in transcription upstream of the 5′-most GARP gene in the 4.3 locus and localised a transcription initiation site for this gene. The putative promoter thus defined appears to be able to drive transcription of a CAT reporter gene when the gene is flanked 5′ by a GARP gene splice acceptor site and 3′ by a GARP gene 3′ UTR. This is the first report of the cloning and characterisation of a promoter for a gene in T.congolense and the first identified promoter in trypanosomes which directs RNA polymerase II-like transcription. The GARP putative promoter region has no significant homology with any other T.brucei promoter sequence and especially with the procyclin/PARP promoter. It is located much further upstream (504 bp) of the GARP start codon in T.congolense than the promoter reported for procyclin/PARP genes in T.brucei, which is around 86 bp upstream of the start codon of the first gene in the PARP A locus ( 16 ). The AG dinucleotide is located 60 nt upstream of the translation start codon for both the 4.3 and 4.4 GARP loci while for the 5′ procyclin/PARP gene in the PARP A locus the distance is only 30 nt ( 34 ). In many eukaryotes the splice acceptor site at an AG dinucleotide is preceded 5′ with a pyrimidine-rich tract and it has been shown that this is also the case for some kinetoplastid genes ( 2 , 3 , 35 , 36 ) including procyclin/PARP genes where there is a 26/29 pyrimidine tract very close to the splice acceptor site ( 34 ). We have not observed extensive polypyrimidine tracts within the putative splice acceptor regions for the GARP loci we have studied, although for the 4.3 locus the region 5′ of the splice acceptor site is 66% TC-rich over the first 100 nt. There is a 9-pyrimidine tract 17 nt upstream of the splice acceptor site and two further short pyrimidine tracts (>5 nt) 31 and 107 nt upstream (overlined, Fig. 5 A). Mutational analysis will be necessary to determine the importance of these sequences in trans -splicing in T.congolense . However, polypyrimidine motifs may not be entirely necessary since experiments with deletion mutations in the dihydrofolate reductase-thymidylate synthase/DST intergenic region of Leishmania major have shown that splice acceptor sites lacking a strong polypyrimidine tract immediately upstream can still be used efficiently ( 2 ). We have also mapped the polyadenylation sites for the GARP genes in T.congolense . While procyclin/PARP transcripts have a single major site of polyadenylation ( 4 , 5 ) GARP transcripts appear to be polyadenylated differentially over a region of 20 bases. We cannot rule out the possibility that the different sites are specific to different individual GARP genes. Such microheterogeneity has also been observed for genes in Leishmania ( 2 , 37–40 ) and T.brucei ( 3 , 41 ) but in the absence of information on other genes in T.congolense it is not possible to determine whether this is a feature of polyadenylation in this species or whether it is peculiar to GARP transcripts. Recent studies have indicated that accurate polyadenylation of transcripts from polycistronic transcription units in Kinetoplastida is dependent on sequence motifs, located downstream of the gene sequence, in the intergenic region ( 2–5 ). One study where the nt sequences from several intergenic regions in T.brucei were compared, revealed a similar organisation of related motifs at a fixed distance downstream of the polyadenylation sites for each gene ( 4 ). Of the four elements identified which were proposed potentially to contribute to specification of accurate polyadenylation three are also present in the intergenic region downstream of the 5′-most GARP gene in the 4.3 locus in T.congolense . These are (i) an intervening sequence of 80 nt between the poly(A) addition site and (ii) a polypyrimidine tract followed by the trinucleotide YAG, and (iii) a further polypyrimidine tract a short distance down-stream followed by another YAG sequence ( Fig. 4 B). Thus although our transient transfection experiments suggest that there are significant cross-species differences in sequences regulating gene expression in the two African trypanosomes, there may be conservation of intergenic region signals directing polyadenylation between the two species. Although we were unable to obtain and assay CAT activity from transiently transfected T.congolense, CAT transcripts were readily detectable. Either CAT enzyme is highly unstable in T.congolense cells, inactive in extracts, or the CAT RNA is not able to be translated. In the constructs we used initially, only the GARP 3 UTR was used to specify polyadenylation. CAT transcripts produced were polyadenylated but not at the wild-type site. It is possible that aberrantly polyadenylated transcripts could be inefficiently translated leading to undetectable levels of CAT enzyme activity. We did not know whether T. congolense followed the pattern for T.brucei and Leishmania, where a downstream splice acceptor site and pyrimidine-rich sequences downstream of the polyadenylation site are required for correct polyadenylation of transcripts (although our sequence analysis suggested this may be true). Therefore, we tested whether inclusion of a portion of the first intergenic region in the 4.3 GARP locus ( Fig. 4 B), including the sequence motifs similar to those necessary for accurate polyadenylation of procyclin/PARP transcripts in T.brucei, would result in our being able to detect CAT activity. However, when we included in our constructs the GARP 3′ UTR and a further 650 nt downstream (the insert in the construct pCG4.33garp, Fig. 1 ) we still obtained no CAT activity (data not shown). CAT expression must be blocked at translation or downstream in these cells. Our results indicate that, despite the assumed relatively close species relationship between T.congolense and T.brucei, sequences important in regulating expression of the major surface antigen of the procyclic form in these organisms are rather different. We found that the T.congolense GARP promoter was inactive in driving CAT expression in transient transfection assays in T.brucei where CAT constructs contained a GARP splice acceptor site but a PARP 3′ UTR. It is possible that in T.brucei the GARP promoter is active, but the GARP splice acceptor site inactive, leading to the production of unstable primary transcripts for CAT. Similarly, the apparent inactivity of the procyclin/PARP promoter in T. congolense could be due to the PARP splice acceptor region not being recognised. However, a construct with a GARP putative promoter region and a PARP splice acceptor site gave no CAT RNA or CAT activity in T.brucei (data not shown), suggesting that the source of the splice acceptor site used has no effect in these constructs. The fact that GARP and procyclin/PARP genes are transcribed by different RNA polymerases may be a more likely cause of promoter inactivity across species. CAT transcripts produced in our constructs were not polyadenylated at the correct site in either the GARP or PARP 3′ UTRs, but this is irrelevant in these experiments since the GARP and PARP promoters gave high levels of CAT RNA with these 3′ UTRs in T.congolense and T.brucei respectively. We observed that the 3′ UTR of procyclin/PARP and GARP transcripts are not interchangeable between species in transient transfection assays. We had noted previously that there existed a conserved 16 nt sequence motif situated at approximately the same distance with respect to the polyadenylation site for the GARP and procyclin/ PARP genes ( 15 ). In T.brucei while one study showed that the 16mer was necessary for efficient translation of PARP/procyclin mRNAs ( 18 ) another study, using a transient transfection approach, found that CAT transcripts whose truncated 3′ UTR lacked the 16mer seemed to be translated efficiently at least in procyclic cells ( 20 ). The 16mer may have a role in GARP gene expression but since the 3′UTRs of both procyclin/PARP and GARP genes are not recognised in the heterologous system, the conserved motif cannot be sufficient for any regulation of gene expression exerted by the 3′ end of the mRNAs. Finally, we have observed that GARP mRNA is readily detected in bloodstream form T.congolense although the protein is not produced (D. Jefferies, unpublished). This is a very different situation from that for procyclin/PARP where transcripts are barely detectable in bloodstream form T.brucei . Our observation indicates that, for GARP expression, life cycle stage-specific regulation must be achieved at the translational or post-translational level. This observation may help to explain why GARP is transcribed by an α-amanitin sensitive RNA polymerase. RNA polymerase II appears to transcribe genes whose expression is constitutive in the trypanosome life cycle as is the case for GARP in T.congolense . Procyclin/PARP transcription in regulated during the life cycle, at least partly ( 42 , 43 ), metacyclic VSG genes are truly transcriptionally regulated ( 44 ), while bloodstream VSG gene expression sites are transcriptionally regulated, especially during the bloodstream phase of the life cycle ( 45 ). All of these genes which encode major surface antigens are transcribed by RNA polymerase I ( 46 ), and this may be a crucial factor in singling out these genes for at least some degree of transcriptional regulation. Acknowledgements We thank Carole Ross (Centre for Tropical and Veterinary Medicine, Edinburgh) for provision of trypanosome stocks and cultures and for advice on culturing. This work was supported by the Wellcome Trust. J.D.B. is a Wellcome Trust Senior Lecturer. References 1 Clayton C.. ,  Prog. Nucleic Acid Res. Mol. Biol. ,  1992, vol.  43 (pg.  37- 65) PubMed  2 LeBowitz J.H.,  Smith J.,  Rusche L.,  Beverley S.M.. ,  Genes Dev. ,  1994, vol.  7 (pg.  996- 1007) CrossRef Search ADS   3 Matthews K.R.,  Tschudi C.,  Ullu E.. ,  Genes Dev. ,  1994, vol.  8 (pg.  491- 501) CrossRef Search ADS PubMed  4 Schürch N.,  Hehl A.,  Vassella E.,  Braun R.,  Roditi I.. ,  Mol. Cell. Biol. ,  1994, vol.  14 (pg.  3668- 3675) CrossRef Search ADS PubMed  5 Hug M.,  Hotz H.,  Hartmann C.,  Clayton C.. ,  Mol. Cell. Biol. ,  1994, vol.  14 (pg.  7428- 7435) CrossRef Search ADS PubMed  6 Graham S.V.. ,  Parasitol. Today ,  1995, vol.  11 (pg.  217- 223) CrossRef Search ADS PubMed  7 Vickerman K.. ,  Br. Med. Bull. ,  1985, vol.  41 (pg.  105- 114) PubMed  8 Cross G.A.M.. ,  Annu. Rev. Immunol. ,  1990, vol.  8 (pg.  83- 110) CrossRef Search ADS PubMed  9 Roditi I.,  Carrington M.,  Turner M.. ,  Nature ,  1987, vol.  325 (pg.  272- 274) CrossRef Search ADS PubMed  10 Roditi I.,  Schwarz H.,  Pearson T.W.,  Beecroft R.P.,  Liu M.K.,  Richardson J.P.,  Buhring H.J.,  Pleiss J.,  Bulow R.,  Williams R.O.,  Overath P.. ,  J. Cell Biol. ,  1989, vol.  108 (pg.  737- 746) CrossRef Search ADS PubMed  11 Mowatt M.R.,  Clayton C.E.. ,  Mol. Cell. Biol. ,  1987, vol.  7 (pg.  2838- 2844) CrossRef Search ADS PubMed  12 Richardson J.P.,  Beecroft R.P.,  Tolson D.L.,  Liu M.K.,  Pearson T.W.. ,  J. Cell Biol. ,  1988, vol.  108 (pg.  737- 746) 13 Beecroft R.P.,  Roditi I.,  Pearson T.W.. ,  Mol. Biochem. Parasitol. ,  1993, vol.  61 (pg.  285- 294) CrossRef Search ADS PubMed  14 Hehl A.,  Pearson T.,  Barry J.D.,  Braun R.,  Roditi I.. ,  Mol. Biochem. Parasitol. ,  1995, vol.  70 (pg.  45- 58) CrossRef Search ADS PubMed  15 Bayne R.A.L.,  Kilbride E.A.,  Lainson F.A.,  Tetley L.,  Barry J.D.. ,  Mol. Biochem. Parasitol. ,  1993, vol.  61 (pg.  295- 310) CrossRef Search ADS PubMed  16 Sherman D.R.,  Janz L.,  Hug M.,  Clayton C.. ,  EMBO J. ,  1991, vol.  10 (pg.  3379- 3386) PubMed  17 Brown S.D.,  Huang J.,  Van der Ploeg L.H.T.. ,  Mol. Cell. Biol. ,  1992, vol.  12 (pg.  2644- 2652) CrossRef Search ADS PubMed  18 Hehl A.,  Vassella E.,  Braun R.,  Roditi I.. ,  Proc. Natl Acad. Sci. USA ,  1994, vol.  91 (pg.  370- 374) CrossRef Search ADS   19 Vassella E.,  Braun R.,  Roditi I.. ,  Nucleic Acids Res. ,  1994, vol.  22 (pg.  1359- 1364) CrossRef Search ADS PubMed  20 Hug M.,  Carruthers V.B.,  Hartmann C.,  Sherman D.S.,  Cross G.A.M.,  Clayton C.. ,  Mol. Biochem. Parasitol. ,  1993, vol.  61 (pg.  87- 96) CrossRef Search ADS PubMed  21 Ross C.A.,  Gray M.A.,  Taylor A.M.,  Luckins A.G.. ,  Acta Trop. ,  1985, vol.  42 (pg.  113- 122) PubMed  22 Gray M.A.,  Ross C.A.,  Taylor A.M.,  Luckins A.G.. ,  Acta Trop. ,  1984, vol.  41 (pg.  343- 353) PubMed  23 Gray M.A.,  Ross C.A.,  Taylor A.M.,  Tetley L.,  Luckins A.G.. ,  Acta Trop. ,  1985, vol.  42 (pg.  99- 111) PubMed  24 Thomashow L.S.,  Milhausen M.,  Rutter W.J.,  Agabian N.. ,  Cell ,  1983, vol.  32 (pg.  35- 43) CrossRef Search ADS PubMed  25 Kooter J.M.,  Borst P.. ,  Nucleic Acids Res. ,  1984, vol.  12 (pg.  9457- 9472) CrossRef Search ADS PubMed  26 Ben Amar M.F.,  Pays A.,  Tebabi P.,  Dero B.,  Seebeck T.,  Steinert M.,  Pays E.. ,  Mol. Cell. Biol. ,  1988, vol.  8 (pg.  2166- 2176) CrossRef Search ADS PubMed  27 Kooter J.M.,  van der Spek H.J.,  Wagter R.,  d'Oliveira C.E.,  van der Hoeven F.,  Johnson P.J.,  Borst P.. ,  Cell ,  1987, vol.  51 (pg.  261- 272) CrossRef Search ADS PubMed  28 Auffray C.,  Rougeon F.. ,  Eur. J. Biochem. ,  1980, vol.  107 (pg.  303- 314) CrossRef Search ADS PubMed  29 Sambrook J.,  Fritsch E.F.,  Maniatis T.. ,  Molecular cloning: A laboratory manual ,  1989 Cold Spring Harbor, NY Cold Spring Laboratory Press 30 Feinberg A.P.,  Vogelstein B.. ,  Anal. Biochem. ,  1983, vol.  132 (pg.  6- 13) CrossRef Search ADS PubMed  31 Bellofatto V.,  Cross G.A.M.. ,  Science ,  1989, vol.  244 (pg.  1167- 1169) CrossRef Search ADS PubMed  32 Zomerdijk J.C.B.M.,  Ouellette M.,  Ten Asbroek A.L.M.A.,  Kieft R.,  Bommer A.M.M.,  Clayton C.E.,  Borst P.. ,  EMBO J. ,  1990, vol.  9 (pg.  2791- 2801) PubMed  33 Seed B.,  Sheen J.-Y.. ,  Gene ,  1988, vol.  67 (pg.  271- 277) CrossRef Search ADS PubMed  34 Huang J.,  Van der Ploeg L.H.T.. ,  EMBO J. ,  1991, vol.  10 (pg.  3877- 3885) PubMed  35 Curotto de Lafaille M.A.,  Laban A.,  Wirth D.F.. ,  Proc. Natl Acad. Sci. USA ,  1992, vol.  89 (pg.  2703- 2707) CrossRef Search ADS   36 Layden R.E.,  Eisen H.. ,  Mol. Cell. Biol. ,  1988, vol.  8 (pg.  1352- 1360) CrossRef Search ADS PubMed  37 Landfear S.M.,  Miller S.I.,  Wirth D.F.. ,  Mol. Biochem. Parasitol. ,  1986, vol.  21 (pg.  235- 245) CrossRef Search ADS PubMed  38 Kapler G.M.,  Coburn C.M.,  Beverley S.M.. ,  Mol. Cell. Biol. ,  1990, vol.  10 (pg.  1084- 1094) CrossRef Search ADS PubMed  39 Flinn H.M.,  Smith D.F.. ,  Nucleic Acids Res. ,  1992, vol.  20 (pg.  755- 762) CrossRef Search ADS PubMed  40 Ramamoorthy R.,  Donelson J.E.,  Paetz K.E.,  Maybodi M.,  Roberts S.C.,  Wilson M.E.. ,  J. Biol. Chem. ,  1992, vol.  267 (pg.  1888- 1895) PubMed  41 Tschudi C.,  Ullu E.. ,  EMBO J. ,  1988, vol.  7 (pg.  455- 463) PubMed  42 Pays E.,  Coquelet H.,  Tebabi P.,  Pays A.,  Jefferies D.,  Steinert M.,  Koenig E.,  Williams R.O.,  Roditi I.. ,  EMBO J. ,  1990, vol.  9 (pg.  3145- 3151) PubMed  43 Berberof M.,  Vanhamme L.,  Tebabi P.,  Pays A.,  Jefferies D.,  Welburn S.,  Pays E.. ,  EMBO J. ,  1995, vol.  14 (pg.  2925- 2934) PubMed  44 Graham S.V.,  Barry J.D.. ,  Mol. Cell. Biol. ,  1995, vol.  15 (pg.  5945- 5956) CrossRef Search ADS PubMed  45 Rudenko G.,  Blundell P.A.,  Taylor M.C.,  Kieft R.,  Borst P.. ,  EMBO J. ,  1994, vol.  13 (pg.  5470- 5482) PubMed  46 Chung H.-M.,  Lee M.G.-S.,  Van der Ploeg L.H.T.. ,  Parasitol. Today ,  1992, vol.  8 (pg.  414- 418) CrossRef Search ADS PubMed  © 1996 Oxford University Press

Chai, Sunghee; Alonso, Juan C.

doi: N/Apmid: N/A

The small subunit of the Bacillus subtilis bacteriophage SPP1 terminase (G1P) forms a sequence-specific nucleoprotein complex with the SPP1 non-encapsidated end (pacL site) during initiation of DNA encapsidation. Gel mobility shift assay was used to study the G1P-pacL interaction. Distamycin, a minor groove binder that induces local distortion of the DNA, inhibits G1P-pacL complex formation. The competition of G1P with distamycin for DNA binding at the pacL site is independent of the order of addition of the reactants. Other minor groove binders, such as spermine or Hoechst 33258, which do not distort DNA, failed to compete with G1P for pacL DNA binding. Cationic metals, which generate a repertoire of DNA structures different from that caused by the minor groove binders, can partially reverse the distamycin-induced inhibition of G1P binding to pacL DNA. The major groove binder methyl green, which does not distort sequence-directed bending of pacL DNA, competes with G1P for binding at the pacL site. Our data suggest that the natural sequence-directed bend that exists within the pacL site is the architectural element that facilitates assembly of a nucleoprotein complex and hence initiation of DNA encapsidation by bacteriophage SPP1.

Chai, Sunghee;Alonso, Juan C.

doi: 10.1093/nar/24.2.282pmid: 8628651

Abstract The small subunit of the Bacillus subtilis bacteriophage SPP1 terminase (G 1 P) forms a sequence-specific nucleoprotein complex with the SPP1 non-encapsidated end ( pac L site) during initiation of DNA encapsidation. Gel mobility shift assay was used to study the G 1 P- pac L interaction. Distamycin, a minor groove binder that induces local distortion of the DNA, inhibits G 1 P- pac L complex formation. The competition of G 1 P with distamycin for DNA binding at the pac L site is independent of the order of addition of the reactants. Other minor groove binders, such as spermine or Hoechst 33258, which do not distort DNA, failed to compete with G 1 P for pac L DNA binding. Cationic metals, which generate a repertoire of DNA structures different from that caused by the minor groove binders, can partially reverse the distamycin-induced inhibition of G 1 P binding to pac L DNA. The major groove binder methyl green, which does not distort sequence-directed bending of pac L DNA, competes with G 1 P for binding at the pac L site. Our data suggest that the natural sequence-directed bend that exists within the pac L site is the architectural element that facilitates assembly of a nucleoprotein complex and hence initiation of DNA encapsidation by bacteriophage SPP1. Introduction Initiation of packaging of double-stranded viral DNA concatemers involves specific interaction of the prohead with virus DNA in a process mediated by a phage-encoded DNA recognition and cleavage (terminase) protein (reviewed in 1 , 2 ). The terminase enzymes described so far, which are hetero-oligomers composed of a small and a large subunit, do not have a significant level of sequence homology (reviewed in 1 ). The role of the terminase small subunit is to specifically recognize the packaging initiation site ( cos or pac ). It is thought that the small terminase subunit forms a nucleoprotein structure that helps to position the terminase large subunit at cos or pac (reviewed in 1 ). The Bacillus subtilis bacteriophage SPP1 terminase enzyme is composed of a small (G 1 P) and a large (G 2 P) subunit which are the products of genes 1 and 2 respectively ( 3 ). G 1 P (estimated native molecular mass 190–210 kDa) is a three-domain protein (DNA binding and G 1 P-G 1 P and G 1 P-G 2 P interacting domains) ( 4 , 5 ). The N-terminal domain contacts DNA by a helix-turn-helix (HTH) motif, the central domain mediates the G 1 P-G 1 P contact, whereas an uncharacterized domain could mediate G 1 P-G 2 P hetero-oligomer formation. No apparent biological role can be assigned to the C-terminal region ( 5 ). The SPP1 pac region can be subdivided into three discrete sites ( pac L, pac C and pac R). The G 2 P cleavage site ( pac C) is located between the pac L and pac R sites ( 3–7 ; Fig. 1 ). G 1 P binds co-operatively to the encapsidated ( pac R) and non-encapsidated ( pac L) DNA ends and holds the two binding sites together in a DNA loop ( 4 ). DNase I footprinting experiments indicate that each G 1 P binding site contains two discrete binding domains, termed box a in pac L and box c in pac R ( 4 ; see Fig. 1 ). The pac L and pac R sites are separated from each other by a stretch of 140 bp ( 4 ). The center-to-center distance of these two non-adjacent sites is ∼204 bp ( 4 ; Fig. 1 ). The G 1 P recognition site at pac L is embedded in a sequence-directed DNA bend. The interaction of G 1 P with pac L DNA is observed only on one side of the double helix ( 4 ). The pac L site contains two directly repeated boxes (box a ), which are located four helical turns away from each other ( Fig. 1 ). G 1 P binding to pac L enhances DNA bending. Therefore, any interaction between G 1 P bound to pac L and G 1 P bound to pac R would give rise to a loop of 204 bp in length (or ∼20 turns of the DNA helix) ( 4 ). DNA loop formation mediated by G 1 P could distort the DNA within the loop and hence alter the binding characteristics of G 2 P to its asymmetric target site ( 4 ). Additional evidence for DNA looping is provided by alternating DNase I-hypersensitive and DNase I-resistant sites in the complexed DNA, which appear with an approximate periodicity of 10 bp (see Fig. 1 ). To address the nature of the interaction between G 1 P and the pac L site and to elucidate the influence of intrinsic DNA bending on this reaction several groove binders (GBs) were employed. Figure 1 View largeDownload slide Schematic representation of the SPP1 pac L, pac C and pac R sites. The thick line indicates the SPP1 pac region. The open bars denote the non-encapsidated phage genome end ( pac L), the packaging cleavage region ( pac C) and the encapsidated end ( pac R) respectively. The broken lines on the pac C bar denote the region of overlap. The directly repeated boxes a, b and c are depicted. The sites hypersensitive to DNase I digestion are denoted by vertical filled arrows. The vertical open arrows indicate the cleavage site at pac C. The distance in base pairs is indicated. Figure 1 View largeDownload slide Schematic representation of the SPP1 pac L, pac C and pac R sites. The thick line indicates the SPP1 pac region. The open bars denote the non-encapsidated phage genome end ( pac L), the packaging cleavage region ( pac C) and the encapsidated end ( pac R) respectively. The broken lines on the pac C bar denote the region of overlap. The directly repeated boxes a, b and c are depicted. The sites hypersensitive to DNase I digestion are denoted by vertical filled arrows. The vertical open arrows indicate the cleavage site at pac C. The distance in base pairs is indicated. The pyrrole amide antibiotic distamycin A (henceforth distamycin) binds to the minor groove of B-DNA in a non-covalent and a non-intercalative mode ( 8 ). Distamycin associates preferentially with a nucleotide sequence rich in adenosine (A) and thymine (T) base pairs. Distamycin binds to the minor groove of B-DNA and induces local structural distortion by bending the DNA helix and inducing conformational changes in the neighborhood ( 9 , 10 ). Electron microscopic ( 11 ), 1 H-NMR ( 12 ) and X-ray crystallographic ( 13 ) studies provide direct evidence of distamycin-induced changes of local DNA structures. Distamycin interferes with proteins that act upon DNA binding ( 14–16 ). The binding of distamycin to the minor groove of DNA can compete with protein binding. Alternatively, the binding of distamycin in the minor groove induces a change in the local conformation which diminishes binding affinity of the protein for the major groove. In this paper we present evidence that binding of G 1 P to the intrinsically bent locus within pac L is inhibited by the minor GB distamycin and the major GB methyl green. Distamycin competition for G 1 P binding to the pac L DNA segment is independent of the order of addition of the reactants. Other minor GBs that do not distort DNA upon binding (spermine and Hoechst 33258) do not compete with G 1 P for pac L binding. Cationic metals, which generate a different repertoire of DNA structures, can partially reverse the distamycin-induced inhibition of G 1 P binding to pac L DNA. It is likely, therefore, that the distamycin-induced conformational change diminishes the binding affinity of G 1 P for pac L DNA (the SPP1 non-encapsidated end). Materials and Methods Bacterial strains and plasmids Escherichia coli strain JM109 ( 17 ) and plasmid pBT397 ( 5 ) have been previously described. Enzymes and reagents SPP1 G 1 P was purified as previously described ( 5 ). Protein concentrations were determined by the method of Bradford ( 18 ), using bovine γ-globulin as a standard. The amount of G 1 P is expressed as mol protein protomers (predicted molecular mass of SPP1 G 1 P is 20.7 kDa). DNA restriction and modification enzymes and poly(dI-dC) were purchased from Boehringer Mannheim (Germany), distamycin and spermine from Sigma (USA), Hoechst 33258 from Polysciences Inc. (USA) and methyl green from ICN (Germany). All chemicals used were reagent grade and solutions were made in quartz-distilled H 2 O. [α- 32 P]dATP was from Amersham Buchler GmbH (Germany). Ultra pure acrylamide was from Serva (Germany). DNA manipulations Covalently closed circular plasmid DNA was purified by the SDS lysis method ( 19 ), followed by purification on a cesium chloride-ethidium bromide gradient. Gel-purified DNA fragments were end-labeled by filling in the restriction site with the large fragment of DNA polymerase I in the presence of [α- 32 P]dATP and dTTP, dCTP and dGTP. Analytical and preparative gel electrophoresis of plasmid DNA and restriction fragments was carried out either in 0.8% (w/v) agarose, Tris-acetate-EDTA, ethidium bromide horizontal slab gels or 4% (w/v) non-denaturing polyacrylamide, Tris-borate gels ( 19 ). The relative amounts of DNA present in any particular band in the autoradiograms was quantitatively scanned with a laser densitometer (LKB UltroScan XL). The linearity of the response with respect to DNA concentration was checked using autoradiograms at different exposure times. Quantitative scans were integrated using the LKB GelScan XL software package. The concentration of the 324 bp pac L DNA was determined using molar extinction coefficients of 6500/M/cm at 260 nm. Except for poly(dI-dC), the amount of DNA is expressed as mol DNA molecules. Gel shift assay The SPP1 271 bp Hpa II- Bsm I SPP1 pac L DNA fragment (obtained as a 324 bp segment) was labeled with [α- 32 P]dATP by filling in the ends ( Eco RI- Hin dIII) with the large fragment (Klenow) of DNA polymerase I. The unincorporated nucleotides were removed by gel filtration. In all conditions 23 pM α- 32 P-labeled DNA (324 bp pac L DNA) and 1 µg poly(dI-dC) were used per reaction mixture (20 µl). When required the pac L DNA was incubated with an excess of G 1 P (240 nM) for 10 min at 37°C. When indicated increasing concentrations of a GB or metal cations were added prior to or after G 1 P- pac L complex formation. The binding reactions were immediately subjected to 4% non-denaturing polyacrylamide gel electrophoresis (ndPAGE) and run at 2 V/cm at 4°C. The gels were dried prior to autoradiography. Results G 1 P binding to the pac L DNA is inhibited by distamycin Recently it has been shown that: (i) the SPP1 terminase small subunit (G 1 P) binds specifically and co-operatively to pac L (non-encapsidated end) and pac R (encapsidated end) sites; (ii) the pac L site contains an intrinsically bent sequence; (iii) G 1 P binding to the pac region enhances DNA looping between the pac L and pac R sites ( 4 ). To address whether the intrinsically bent pac L sequence plays a role in G 1 P- pac L interaction we have analyzed the influence of distamycin on G 1 P binding by gel shift assay. It has been reported that at low doses (500 nM-1 µM) distamycin selectively prevents DNA bending and does not abolish protein-DNA interactions, whereas high doses (1–200 µM) induce local structural distortions by bending the DNA helix ( 9 , 10 ). In a previous study we showed that G 1 P- pac DNA complexes remain in the well when a low ionic strength ndPAGE running buffer is used in the gel shift assay, whereas in high ionic strength ndPAGE a diffuse retarded G 1 P- pac DNA complex is observed ( 5 ). Under high ionic strength ndPAGE conditions, however, an ∼15-fold excess of G 1 P is required to saturate the DNA substrate when compared with the G 1 P- pac DNA complexes retained by the filter binding assay (see 4 ). Since a high ionic strength ndPAGE running buffer was used, we first set up the conditions of the gel shift assay to maximize complex formation using a fixed amount of labeled DNA fragment (23 pM) and purified G 1 P (240 nM). Figure 2 View largeDownload slide Effect of distamycin on pac L and on G 1 P- pac L DNA complex formation. ( A ) 32 P-Labeled 324 bp pac L DNA (23 pM) and 1 µg poly(dI-dC) were incubated with increasing concentrations of distamycin (Dis) for 20 min at room temperature (reaction mixture 20 µl) and then for 10 min at 37°C prior to 4% ndPAGE. ( B ) 32 P-Labeled 324 bp pac L DNA (23 pM) and 1 µg poly(dI-dC) were pre-incubated with increasing concentrations of distamycin for 20 min at room temperature before addition of G 1 P (240 nM). The reaction mixture (20 µl) was then incubated for 10 min at 37°C and subjected to 4% ndPAGE. Lanes 1 and 1′, DNA incubated without distamycin. Lanes 2–10, distamycin at: lane 2, 1 nM; lane 3, 10 nM; lane 4, 50 nM; lane 5, 100 nM; lane 6, 500 nM; lane 7, 1 µM; lane 8, 10 µM; lane 9, 50 µM; lane 10, 100 µM. Figure 2 View largeDownload slide Effect of distamycin on pac L and on G 1 P- pac L DNA complex formation. ( A ) 32 P-Labeled 324 bp pac L DNA (23 pM) and 1 µg poly(dI-dC) were incubated with increasing concentrations of distamycin (Dis) for 20 min at room temperature (reaction mixture 20 µl) and then for 10 min at 37°C prior to 4% ndPAGE. ( B ) 32 P-Labeled 324 bp pac L DNA (23 pM) and 1 µg poly(dI-dC) were pre-incubated with increasing concentrations of distamycin for 20 min at room temperature before addition of G 1 P (240 nM). The reaction mixture (20 µl) was then incubated for 10 min at 37°C and subjected to 4% ndPAGE. Lanes 1 and 1′, DNA incubated without distamycin. Lanes 2–10, distamycin at: lane 2, 1 nM; lane 3, 10 nM; lane 4, 50 nM; lane 5, 100 nM; lane 6, 500 nM; lane 7, 1 µM; lane 8, 10 µM; lane 9, 50 µM; lane 10, 100 µM. The SPP1 324 bp [ 32 P] pac L DNA, which is rich in dA + dT, migrates much slower than a 430 bp size marker DNA fragment in 4% ndPAGE at 4°C ( 4 ; Fig. 2 A). This anomalous mobility is greatly reduced at higher temperatures, implying the existence of sequence-directed curvature (see 4 ). To examine the influence of distamycin on the conformation of pac L DNA we incubated the 324 bp [ 32 P] pac L DNA (23 pM) and 1 µg poly(dI-dC) with varying concentrations of distamycin (1 nM-100 µM) and subjected the reaction mixture to ndPAGE. As revealed in Figure 2 A, distamycin influences the mobility of pac L DNA. Distamycin concentrations from 1 nM to 1 µM increase the mobility of the pac L DNA fragment. In the presence of 1 µM distamycin the 324 bp [ 32 P] pac L DNA segment migrated between the 341 and 380 bp size markers (‘reduced curvature’), but in the presence of higher concentrations of distamycin (5–100 µM) the DNA fragment again migrated slower than the 430 bp size marker DNA fragment ( Fig. 2 A). In the presence of 100 nM distamycin (∼13 distamycin molecules/bp pac L DNA) migration of the fragment was faster than in its absence, but to obtain the fastest migrating DNA segment 1 µM distamycin (134 distamycin molecules/bp pac L DNA) was needed. The absence of a large excess of non-specific competitor DNA [1 µg poly(dI-dC)] did not alter the pattern observed in Figure 2 A (data not shown). It is likely, therefore, that poly(dI-dC) does not titrate distamycin for pac L DNA. Pre-incubation of distamycin with the 324 bp [ 32 P] pac L DNA fragment (23 pM) reduced the ability of G 1 P (240 nM) to bind to its target DNA in a concentration-dependent manner ( Fig. 2 B). Inhibition of G 1 P binding to pac L DNA occurred at the same distamycin concentrations that modified or eliminated curvature of the naked DNA ( Fig. 2 A and B). As revealed in Figure 2 B, G 1 P- pac L DNA complex formation was completely blocked at 1 µM distamycin, while the first indication of inhibition of complex formation was seen at 500 nM distamycin. The amount of distamycin required to prevent 50% G 1 P- pac L DNA complex formation was ∼750 nM. The distamycin-induced inhibition of G 1 P- pac L DNA complex formation is independent of the order of addition of the reactants. The same results were obtained when the pac L DNA was pre-incubated with distamycin and G 1 P was added subsequently as when the G 1 P-DNA complex was pre-formed and then challenged with distamycin (data not shown). It could be hypothesized that distamycin inhibits G 1 P- pac L complex formation either by competing for binding to the minor groove of DNA or by altering the local DNA conformation of pac L DNA, which could diminish the binding affinity of G 1 P for DNA. These two possibilities were therefore analyzed. The minor groove binders spermine and Hoechst 33258 do not compete with G 1 P for binding to pac L DNA To analyze whether distamycin-induced inhibition of G 1 P binding to pac L DNA is due to an occupancy of the same site we used minor GBs that do not modify the structure of DNA upon binding, such as spermine ( 20 ) or Hoechst 33258 ( 21 ). The presence of increasing concentrations of spermine (100 nM-1 mM) or Hoechst 33258 (100 nM-10 µM) did not alter the migration of DNA in 4% ndPAGE at 4°C ( Fig. 3 A and data not shown). The presence of concentrations of Hoechst 33258 >10 µM produced cross-linked or precipitated material and the DNA did not enter the gel. Pre-incubation of pac L DNA (23 pM) and 1 µg poly(dI-dC) with increasing concentrations of spermine (100 nM-1 mM; Fig. 3 B) or of Hoechst 33258 (100 nM-10 µM ; data not shown) and subsequent addition of G 1 P (240 nM) did not affect the affinity of G 1 P for pac L DNA under our experimental conditions. The same effect was observed when pre-formed complexes of G 1 P- pac L DNA were challenged with increasing concentrations of spermine or Hoechst 33258 (data not shown). The presence of a large excess (up to 4 × 10 3 molecules) of spermine per G 1 P molecule did not affect its binding to pac L DNA. It seems, therefore, that the minor GBs spermine and Hoechst 33258 do not compete with G 1 P for binding to pac L DNA. Figure 3 View largeDownload slide Effect of spermine on pac L and on G 1 P -pac L DNA complex formation. ( A ) 32 P-Labeled 324 bp pac L DNA (23 pM) and 1 µg poly(dI-dC) were incubated with increasing concentrations of spermine (sp) for 5 min at 37°C (reaction mixture 20 µl) and subjected to 4% ndPAGE. Lanes 2–6, spermine at: lane 2, 100 nM; lane 3, 1 µM; lane 4, 10 µM; lane 5, 100 µM; lane 6, 1 µM. ( B ) 32 P-Labeled 324 bp pac L DNA (23 pM) and 1 µg poly(dI-dC) were incubated with increasing concentrations of spermine as in (A) (lanes 2–6) before addition of G 1 P (240 nM). The reaction mixture (20 µl) was then incubated for 10 min at 37°C and subjected to 4% ndPAGE. Lanes 1 and 1′, DNA not treated with spermine. Figure 3 View largeDownload slide Effect of spermine on pac L and on G 1 P -pac L DNA complex formation. ( A ) 32 P-Labeled 324 bp pac L DNA (23 pM) and 1 µg poly(dI-dC) were incubated with increasing concentrations of spermine (sp) for 5 min at 37°C (reaction mixture 20 µl) and subjected to 4% ndPAGE. Lanes 2–6, spermine at: lane 2, 100 nM; lane 3, 1 µM; lane 4, 10 µM; lane 5, 100 µM; lane 6, 1 µM. ( B ) 32 P-Labeled 324 bp pac L DNA (23 pM) and 1 µg poly(dI-dC) were incubated with increasing concentrations of spermine as in (A) (lanes 2–6) before addition of G 1 P (240 nM). The reaction mixture (20 µl) was then incubated for 10 min at 37°C and subjected to 4% ndPAGE. Lanes 1 and 1′, DNA not treated with spermine. Cationic metals partially compete with G 1 P binding to pac L DNA Certain divalent cationic metals, such as Ba 2+ , Co 2+ , Mn 2+ and Zn 2+ , can promote sequence-directed DNA bending ( 22 ). The fraction of molecules with a cation-induced bend is dependent on both the type and the concentration of cationic metal. MnCl 2 and BaCl 2 are most effective in inducing curvature at 50–100 mM and less effective at higher concentrations ( 22 ). To analyze whether cationic metals could affect G 1 P binding to pac L DNA by generating a different repertoire of bent structures we have used BaCl 2 and MnCl 2 (see 22 ). As revealed in Figure 4 , the addition of 40–160 mM BaCl 2 or MnCl 2 reduced the mobility of the intrinsically bent 324 bp [ 32 P] pac L (23 pM) DNA segment in 4% ndPAGE at 4°C, implying that curvature is increased in the presence of the divalent cations. In the presence of 160 mM BaCl 2 <3% of total DNA showed an anomalous mobility, but at the same concentration of MnCl 2 >99% of the molecules showed an anomalous mobility when compared with absence of the cationic metals (see Fig. 4 ). Pre-incubation of pac L DNA (23 pM) and 1 µg poly(dI-dC) with increasing concentrations (5–40 mM) of BaCl 2 ( Fig. 4 A) or MnCl 2 ( Fig. 4 B) and subsequent addition of G 1 P (240 nM) did not affect the affinity of G 1 P for pac L DNA. It is likely, therefore, that the metal ions at concentrations ranging between 5 and 40 mM do not have a direct effect on G 1 P or on G 1 P interaction with DNA. In the presence of 80 or 160 mM MnCl 2 , however, ∼5% of total DNA were freed from the G 1 P- pac L complex ( Fig. 4 B). Similar results were observed when other cationic metals (e.g. MgCl 2 ) that induce sequence-directed DNA bending were used (data not shown). It is likely, therefore, that metal ion-induced bending (see Fig. 4 A) can generate a DNA structure at pac L that is not recognized by G 1 P. Figure 4 View largeDownload slide Effect of cationic metals on pac L DNA and on G 1 P- pac L DNA complex formation. ( A ) (Left) 32 P-Labeled 324 bp pac L DNA (23 pM) and 1 µg poly(dI-dC) were incubated with increasing concentrations of BaCl 2 (Ba 2+ ) for 5 min at 37°C (reaction mixture 20 µl) and subjected to 4% ndPAGE. Lanes 2–7, BaCl 2 at: lane 2, 5 µM; lane 3, 10 µM; lane 4, 20 µM; lane 5, 40 µM; lane 6, 80 µM; lane 7, 160 µM. (Right) the 324 bp [ 32 P] pac L DNA (23 pM) and 1 µg poly(dI-dC) were pre-incubated with increasing concentrations of BaCl 2 before addition of G 1 P (240 nM). The reaction mixture (20 µl) was then incubated for 10 min at 37°C. Lanes 1 and 1′, DNA not treated with BaCl 2 . ( B ) (Left) 32 P-Labeled 324 bp pac L DNA (23 pM) and 1 µg poly(dI-dC) were pre-incubated with increasing concentrations of MnCl 2 (Mn 2+ ) for 10 min at 37°C (reaction mixture 20 µl). Lanes 2–7, MnCl 2 at: lane 2, 5 µM; lane 3, 10 µM; lane 4, 20 µM; lane 5, 40 µM; lane 6, 80 µM; lane 7, 160 mM. (Right) 32 P-Labeled 324 bp pac L DNA (23 pM) and 1 µg poly(dI-dC) were pre-incubated with increasing concentrations of MnCl 2 before addition of G 1 P (240 nM). The reaction mixture was then incubated for 10 min at 37°C and subjected to 4% ndPAGE. Lanes 1 and 1′, DNA not incubated with MnCl 2 . Figure 4 View largeDownload slide Effect of cationic metals on pac L DNA and on G 1 P- pac L DNA complex formation. ( A ) (Left) 32 P-Labeled 324 bp pac L DNA (23 pM) and 1 µg poly(dI-dC) were incubated with increasing concentrations of BaCl 2 (Ba 2+ ) for 5 min at 37°C (reaction mixture 20 µl) and subjected to 4% ndPAGE. Lanes 2–7, BaCl 2 at: lane 2, 5 µM; lane 3, 10 µM; lane 4, 20 µM; lane 5, 40 µM; lane 6, 80 µM; lane 7, 160 µM. (Right) the 324 bp [ 32 P] pac L DNA (23 pM) and 1 µg poly(dI-dC) were pre-incubated with increasing concentrations of BaCl 2 before addition of G 1 P (240 nM). The reaction mixture (20 µl) was then incubated for 10 min at 37°C. Lanes 1 and 1′, DNA not treated with BaCl 2 . ( B ) (Left) 32 P-Labeled 324 bp pac L DNA (23 pM) and 1 µg poly(dI-dC) were pre-incubated with increasing concentrations of MnCl 2 (Mn 2+ ) for 10 min at 37°C (reaction mixture 20 µl). Lanes 2–7, MnCl 2 at: lane 2, 5 µM; lane 3, 10 µM; lane 4, 20 µM; lane 5, 40 µM; lane 6, 80 µM; lane 7, 160 mM. (Right) 32 P-Labeled 324 bp pac L DNA (23 pM) and 1 µg poly(dI-dC) were pre-incubated with increasing concentrations of MnCl 2 before addition of G 1 P (240 nM). The reaction mixture was then incubated for 10 min at 37°C and subjected to 4% ndPAGE. Lanes 1 and 1′, DNA not incubated with MnCl 2 . BaCl 2 reverses the distamycin-induced inhibition of G 1 P-DNA interaction To examine whether distamycin-induced inhibition of G 1 P- pac L complex formation is due to a conformational change in the DNA we pre-incubated the 324 bp pac L DNA with distamycin and with increasing concentrations of BaCl 2 or MnCl 2 prior to addition of G 1 P. Figure 5 A shows 324 bp [ 32 P] pac L DNA (23 pM) which was first incubated with 5 µM distamycin and then with increasing concentrations of BaCl 2. The latter reactant caused little effect, if any, on mobility of the DNA fragment in concentrations ranging from 5 to 20 mM. The presence of 40–160 mM BaCl 2 changed the mobility of 324 bp [ 32 P] pac L DNA (see Fig. 4 A) and addition of distamycin enhanced such an effect (see Fig. 5 A). When the DNA was pre-incubated with 5 µM distamycin G 1 P was unable to form a complex with the 324 bp [ 32 P] pac L DNA (see Fig. 2 B), whereas when the DNA was pre-incubated with 5 µM distamycin and 5–40 mM BaCl 2 , G 1 P (240 nM) recovered its ability to bind the pac L DNA segment ( Fig. 5 B). The presence of BaCl 2 (20–40 mM) reversed the negative effect of distamycin on G 1 P-DNA complex formation and >70% of the DNA was complexed with G 1 P ( Fig. 5 B). At higher concentrations of BaCl 2 (80 mM) we did not observe reversal of distamycin-induced inhibition of the G 1 P-DNA interaction ( Fig. 5 B). Similar results were observed when MnCl 2 was used (data not shown). These results indicate that these metal ions at concentrations ranging between 5 and 40 mM enhance a specific structure(s) (‘active curvature’) on the DNA that reverses distamycin-induced inhibition of the G 1 P-DNA interaction, whereas other repertoires of structures fail to reverse the negative effect of distamycin on G 1 P binding to pac L DNA. Methyl green partially competes with G 1 P for binding to pac L DNA In a previous study we showed that G 1 P lacking the DNA binding HTH motif (G 1 P*) is able to interact with wild-type G 1 P, but fails to bind to the SPP1 pac region ( 5 ). It is thought that the HTH motif is the principal structural element of the terminase small subunit of many different phages (reviewed in 1 ). To investigate whether G 1 P- pac L interaction also occurs in the major groove of DNA we measured G 1 P-DNA complex formation in the presence of increasing concentrations of methyl green, which is a major GB (see 23 ). Figure 5 View largeDownload slide Effect of BaCl 2 on pac L and on G 1 P- pac L DNA complex formation. ( A ) 32 P-Labeled 324 bp pac L DNA (23 pM) and 1 µg poly(dI-dC) were pre-incubated with 5 µM distamycin for 20 min at room temperature. After subsequent addition of increasing concentrations of BaCl 2 (Ba 2+ ) for 10 min at 37°C (reaction mixture 20 µl) probes were subjected to 4% ndPAGE. Lanes 2–7, BaCl 2 at: lane 2, 5 mM; lane 3, 10 mM; lane 4, 20 mM; lane 5, 40 mM; lane 6, 80 mM; lane 7, 160 mM. ( B ) 32 P-Labeled 324 bp pac L DNA (23 pM) and 1 µg poly(dI-dC) were pre-incubated with 5 µM distamacyn for 20 min at room temperature and then with increasing concentrations of BaCl 2 as in (A) before addition of G 1 P (240 nM). The reaction mixture was then incubated for 10 min at 37°C and subjected to 4% ndPAGE. Lanes 1 and 1′, DNA not treated with BaCl 2 . Figure 5 View largeDownload slide Effect of BaCl 2 on pac L and on G 1 P- pac L DNA complex formation. ( A ) 32 P-Labeled 324 bp pac L DNA (23 pM) and 1 µg poly(dI-dC) were pre-incubated with 5 µM distamycin for 20 min at room temperature. After subsequent addition of increasing concentrations of BaCl 2 (Ba 2+ ) for 10 min at 37°C (reaction mixture 20 µl) probes were subjected to 4% ndPAGE. Lanes 2–7, BaCl 2 at: lane 2, 5 mM; lane 3, 10 mM; lane 4, 20 mM; lane 5, 40 mM; lane 6, 80 mM; lane 7, 160 mM. ( B ) 32 P-Labeled 324 bp pac L DNA (23 pM) and 1 µg poly(dI-dC) were pre-incubated with 5 µM distamacyn for 20 min at room temperature and then with increasing concentrations of BaCl 2 as in (A) before addition of G 1 P (240 nM). The reaction mixture was then incubated for 10 min at 37°C and subjected to 4% ndPAGE. Lanes 1 and 1′, DNA not treated with BaCl 2 . As revealed in Figure 6 A, the addition of 50 nM-50 µM methyl green did not modify mobility of the intrinsically bent SPP1 324 bp [ 32 P] pac L (23 pM) DNA segment in 4% ndPAGE at 4°C. Methyl green concentrations ranging from 50 nM to 1 µM did not alter the rate of G 1 P-DNA complex formation, but at 50 µM it decreased the DNA binding activity of G 1 P to pac L DNA ( Fig. 6 B). The amount of methyl green required to prevent 50% G 1 P- pac L DNA complex formation, compared with an untreated control, is <50 µM. Similar results were obtained when the pac L DNA was pre-incubated with methyl green and G 1 P was subsequently added ( Fig. 6 C; our unpublished results). Figure 6 View largeDownload slide Effect of methyl green on pac L and on G 1 P- pac L DNA complex formation. ( A ) 32 P-Labeled 324 bp pac L DNA (23 pM) and 1 µg poly(dI-dC) were incubated with increasing concentrations of methyl green (MG) for 10 min at 37°C (reaction mixture 20 µl) and subjected to 4% ndPAGE. Lanes 2–7, methyl green at: lane 2, 50 nM; lane 3, 100 nM; lane 4, 500 nM; lane 5, 1 µM; lane 6, 10 µM; lane 7, 50 µM. ( B ) 32 P-Labeled 324 bp pac L DNA (23 pM) and 1 µg poly(dI-dC) were pre-incubated with increasing concentrations of methyl green as in (A) for 10 min at 37°C before addition of G 1 P (240 nM). The reaction mixture (20 µl) was then incubated for 10 min at 37°C. ( C ) 32 P-Labeled 324 bp pac L DNA (23 pM) and 1 µg poly(dI-dC) were pre-incubated with (240 nM. G 1 P for 10 min at 37°C before addition of increasing concentrations of methyl green as in (A). Lanes 1 and 1′, DNA not treated with methyl green. Figure 6 View largeDownload slide Effect of methyl green on pac L and on G 1 P- pac L DNA complex formation. ( A ) 32 P-Labeled 324 bp pac L DNA (23 pM) and 1 µg poly(dI-dC) were incubated with increasing concentrations of methyl green (MG) for 10 min at 37°C (reaction mixture 20 µl) and subjected to 4% ndPAGE. Lanes 2–7, methyl green at: lane 2, 50 nM; lane 3, 100 nM; lane 4, 500 nM; lane 5, 1 µM; lane 6, 10 µM; lane 7, 50 µM. ( B ) 32 P-Labeled 324 bp pac L DNA (23 pM) and 1 µg poly(dI-dC) were pre-incubated with increasing concentrations of methyl green as in (A) for 10 min at 37°C before addition of G 1 P (240 nM). The reaction mixture (20 µl) was then incubated for 10 min at 37°C. ( C ) 32 P-Labeled 324 bp pac L DNA (23 pM) and 1 µg poly(dI-dC) were pre-incubated with (240 nM. G 1 P for 10 min at 37°C before addition of increasing concentrations of methyl green as in (A). Lanes 1 and 1′, DNA not treated with methyl green. Discussion Genetic evidence suggests that the terminase enzyme from the B.subtilis phage SPP1 is essential for recognition and cleavage at the packaging initiation region ( pac ) ( 3 ). The terminase enzyme is a hetero-oligomer composed of a small (G 1 P) and a large (G 2 P) subunit, which are the products of genes 1 and 2 respectively. The pac region consists of three discrete sites ( pac L, pac C and pac R) ( 4 ). The site of double-stranded DNA cleavage by G 2 P is called pac C ( 4 , 6 , 7 ), whereas the sites recognized by G 1 P in the encapsidated and non-encapsidated DNA strand are termed pac R (right) and pac L (left) respectively ( 4 ). In previous studies we have shown that G 1 P is an oligomer (190–210 kDa) with a ring-like structure in solution. We have shown that several G 1 P molecules specifically interact with an intrinsically bent region of pac L covering a tract of almost 100 bp and that the specific interaction between G 1 P and pac L occurs on only one side of the DNA double helix ( 4 ). Upon binding to pac DNA G 1 P induces DNA bending and looping between the pac L and pac R sites and binding of G 1 P to pac L and pac R DNA facilitates assembly of a higher order nucleoprotein structure ( 4 , 5 ). On the basis of these data we have hypothesized that the G 1 P- pac L nucleoprotein complex and the G 1 P- pac R complex, with subsequent looping of the intervening DNA, could direct the positioning of G 2 P and define the directionality of DNA packaging ( 4 ). The small subunit of the terminase enzyme from different bacteriophages interacts with DNA through a HTH DNA binding motif ( 1 ). In a previous study we postulated that G 1 P binding to the pac sites occurs in the major groove of B-DNA ( 3 , 5 ). In this study we show that methyl green, which is a major GB, affects the binding of G 1 P to pac L DNA. It is likely, therefore, that G 1 P interacts with DNA in the major groove. The DNA features recognized by G 1 P indicate that the protein requires a bent DNA with an ‘active’ phase for assembly of a specialized nucleoprotein structure. Our studies have demonstrated that distamycin alters the mobility of pac L and interferes with G 1 P DNA binding. In the presence of 1 µM distamycin G 1 P fails to bind to the ‘unbent’ pac L DNA ( Fig. 2 B). Distamycin concentrations ranging from 5 to 100 µM, which decreased the mobility of pac L DNA, inhibit binding of G 1 P to (‘distamycin-induced inactive bent’) pac L DNA ( Fig. 2 B). Inhibition of G 1 P binding is also observed after addition of distamycin to preformed G 1 P- pac L DNA complexes, suggesting that either binding of the antibiotic to the minor groove can displace G 1 P from the DNA or that upon binding of distamycin in the minor groove it induces a conformational change in DNA which diminishes the binding affinity of G 1 P for pac L DNA. The competition of distamycin for G 1 P binding to pac L is neither the result of inhibition of contacts of G 1 P with the minor groove nor a direct interaction of distamycin and G 1 P that interferes with binding of the protein to pac L. These conclusions are based, first, on the findings that other minor GBs that do not affect the mobility of pac L DNA in ndPAGE do not affect G 1 P- pac L complex formation and, second, that metal ions, which are agents known to promote DNA bending, can partially reverse distamycin-induced inhibition. Previous studies have demonstrated that distamycin interferes with the interaction of proteins that bind to the DNA major groove at concentrations comparable (1–2 µM) with those required to interfere with protein-DNA complex formation ( 24 , this work) or greater (20–200 µM) ( 15 ). The interference caused by distamycin with proteins that interact with DNA through the major groove could be a result of a DNA conformation change, rather than a direct impediment due to occupancy by the minor GB of the protein target site. It is likely, therefore, that distamycin can effectively displace G 1 P bound in the major groove of DNA. We show here that the specificity of G 1 P- pac L DNA binding is due not only to the sequence of its target site (box a ), but also to the local conformation at that site. This conclusion is based on the finding that BaCl 2 and MnCl 2 , which are agents known to promote DNA bending ( 22 ), can partially reverse (>70% of distamycin-induced inhibition at metal ion concentrations ranging from 20 to 40 mM) the inhibitory effect of distamycin. Higher concentrations of BaCl 2 or MnCl 2 , which promote DNA looping, also exerted a negative effect on G 1 P- pac L complex formation. It is likely that only a limited repertoire of DNA structures promoted by the metal ions could reverse distamycin-induced inhibition of G 1 P- pac L complex formation. The specific interaction of G 1 P with pac L occurs on only one face of the DNA double helix ( 4 ). It is likely that G 1 P deflects the pac L DNA towards itself upon binding and the net curvature of pac L reaches extremes when the deformation affects the same DNA face (‘active bend’) or opposite DNA faces (‘inactive bend’). Distamycin could affect both DNA faces or the opposite DNA face, generating a different trajectory of the DNA path. It is likely, therefore, that distamycin inhibits G 1 P- pac L complex formation by altering the conformation of pac L DNA, rather than by competing for binding to the minor groove (see 9 , 16 ). In summary, G 1 P seems to interact with pac L DNA via the major groove. Upon binding of G 1 P to box a , which is embedded in the intrinsically bent pac L DNA, ∼100 bp are wrapped around a multimeric G 1 P molecule ( 4 ). Distamycin, which is a drug that has been shown to remove DNA curvature ( 9 ), inhibits G 1 P- pac L complex formation. DNA curvature, which is particularly pronounced in the presence of divalent cations, reverses distamycin-induced inhibition of G 1 P- pac L complex formation. Acknowledgements This work was partially supported by a grant from Direción General de Investigación Cientifica y Técnica (PB93-0116) to JCA. We thank T.A.Trautner, F.Rojo and F.Weise for critical reading of the manuscript. References 1 Black L.W.. ,  Annu. Rev. Microbiol. ,  1989, vol.  43 (pg.  267- 292) CrossRef Search ADS PubMed  2 Murialdo H.. ,  Annu. Rev. Biochem. ,  1991, vol.  60 (pg.  125- 153) CrossRef Search ADS PubMed  3 Chai S.,  Bravo A.,  Lüder G.,  Nedlin A.,  Trautner T.A.,  Alonso J.C.. ,  J. Mol. Biol. ,  1992, vol.  224 (pg.  87- 102) CrossRef Search ADS PubMed  4 Chai S.,  Lurz R.,  Alonso J.C.. ,  J. Mol. Biol. ,  1995, vol.  252 (pg.  386- 398) CrossRef Search ADS PubMed  5 Chai S.,  Kruft V.,  Alonso J.C.. ,  Virology ,  1994, vol.  202 (pg.  930- 939) CrossRef Search ADS PubMed  6 Bravo A.,  Alonso J.C.,  Trautner T.A.. ,  Nucleic Acids Res. ,  1990, vol.  18 (pg.  2881- 2886) CrossRef Search ADS PubMed  7 Deichelbohrer I.,  Messer W.,  Trautner T.A.. ,  J. Virol. ,  1982, vol.  42 (pg.  83- 90) PubMed  8 Zimmer C.,  Wahnert U.. ,  Prog. Biophys. Mol. Biol. ,  1986, vol.  47 (pg.  31- 112) CrossRef Search ADS PubMed  9 Wu H.-M.,  Crothers D.M.. ,  Nature ,  1984, vol.  308 (pg.  509- 513) CrossRef Search ADS PubMed  10 Kopka M.L.,  Yoon C.,  Goodsell D.,  Pjura P.,  Dickerson R.E.. ,  Proc. Natl. Acad. Sci. USA ,  1985, vol.  82 (pg.  1376- 1380) CrossRef Search ADS   11 Griffith J.,  Bleyman M.,  Rauch C.A.,  Kitchin P.A.,  Englund P.T.. ,  Cell ,  1986, vol.  46 (pg.  717- 724) CrossRef Search ADS PubMed  12 Klevit R.R.,  Wemmer D.E.,  Reid B.R.. ,  Biochemistry ,  1986, vol.  25 (pg.  3296- 3303) CrossRef Search ADS PubMed  13 Coll M.,  Aymami J.,  van der Marel G.A.,  van Boom J.H.,  Rich A.,  Wang A.H.-J.. ,  Biochemistry ,  1989, vol.  28 (pg.  310- 320) CrossRef Search ADS PubMed  14 Bruzik J.P.,  Auble D.T.,  deHaseth P.L.. ,  Biochemistry ,  1987, vol.  26 (pg.  950- 956) CrossRef Search ADS PubMed  15 Broggini M.,  Ponti M.,  Ottlenghi S.,  D'Incalci M.,  Mongelli N.,  Mantovani R.. ,  Nucleic Acids Res. ,  1989, vol.  17 (pg.  1051- 1059) CrossRef Search ADS PubMed  16 Straney D.C.,  Crothers D.M.. ,  Biochemistry ,  1987, vol.  26 (pg.  1987- 1995) CrossRef Search ADS PubMed  17 Yanisch-Perron C.,  Vieira J.,  Messing J.. ,  Gene ,  1985, vol.  33 (pg.  103- 119) CrossRef Search ADS PubMed  18 Bradford M.M.. ,  Anal. Biochem. ,  1986, vol.  72 (pg.  248- 254) CrossRef Search ADS   Sambrook J.,  Maniatis T.,  Fritsch E.F.. ,  Molecular Cloning: A Laboratory Manual ,  1989 2nd Ed. Cold Spring Harbor, NY Cold Spring Harbor Laboratory Press 20 Schmid N.,  Behr J.-P.. ,  Biochemistry ,  1991, vol.  30 (pg.  4357- 4361) CrossRef Search ADS PubMed  21 Beerman T.A.,  McHugh M.M.,  Sigmund R.,  Lown J.W.,  Rao K.E.,  Bathini Y.. ,  Biochim. Biophys. Acta ,  1992, vol.  1131 (pg.  53- 61) CrossRef Search ADS PubMed  22 Laundon C.H.,  Griffith J.D.. ,  Biochemistry ,  1987, vol.  26 (pg.  3759- 3762) CrossRef Search ADS PubMed  23 Kumar K.A.,  Muniyappa K.. ,  J. Biol. Chem. ,  1992, vol.  267 (pg.  24824- 24832) PubMed  24 Dorn A.,  Affolter M.,  Muller M.,  Gehring W.J.,  Leupin W.. ,  EMBO J. ,  1992, vol.  11 (pg.  279- 286) PubMed  © 1996 Oxford University Press