Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 7-Day Trial for You or Your Team.

Learn More →

Use of RDA analysis of knockout mice to identify myeloid genes regulated in vivo by PU.1 and C/EBPα

Use of RDA analysis of knockout mice to identify myeloid genes regulated in vivo by PU.1 and C/EBPα 3034–3043 Nucleic Acids Research, 1998, Vol. 26, No. 12  1998 Oxford University Press Use of RDA analysis of knockout mice to identify myeloid genes regulated in vivo by PU.1 and C/EBPα 1 2 2 Atsushi Iwama, Pu Zhang, Gretchen J. Darlington , Scott R. McKercher , Richard Maki and Daniel G. Tenen* Hematology/Oncology Division and Department of Medicine, Beth Israel Deaconess Medical Center and Harvard Medical School, Boston, MA 02215, USA, Departments of Pathology and Molecular and Human Genetics, Baylor College of Medicine, Houston, TX 77030, USA and The Burnham Institute, La Jolla, CA 92037, USA Received January 13, 1998; Revised and Accepted April 16, 1998 DDBJ/EMBL/GenBank accession nos AA720492–AA720501 ABSTRACT retinoic acid receptor α (RARα), promyelocytic leukemia zinc finger (PLZF), myeloid zinc finger protein-1 (MZF-1), early PU.1 and C/EBPα are transcription factors essential for response gene-1 (Egr-1), Wilms’ tumor suppressor gene (WT-1) normal myeloid development. Loss-of-function mutation and homeobox proteins (2), among which PU.1 and C/EBPα of PU.1 leads to an absolute block in monocyte/ have been shown to be indispensable for myeloid development by macrophage development and abnormal granulocytic gene targeting experiments (3–5). development while that of C/EBPα causes a selective PU.1 is a member of the Ets transcription family and is block in neutrophilic differentiation. In order to under- predominantly expressed in hematopoietic cells (6). PU.1 mRNA stand these phenotypes, we studied the role of PU.1 is expressed at low levels in multipotential CD34 cells, and is and C/EBPα in the regulation of myeloid target genes upregulated with myeloid and B cell differentiation (6–9). in vivo. Northern blot analysis revealed that mRNAs Transient transfection studies have shown that PU.1 regulates the encoding receptors for M-CSF, G-CSF and GM-CSF, promoters of a number of myeloid genes, such as CD11b, primary –/– were expressed at low levels in PU.1 fetal liver granule proteins (myeloperoxidase, neutrophilic elastase and compared with wild type. To identify additional myeloid proteinase-3), GM-CSF receptor, G-CSF receptor and M-CSF genes regulated by PU.1 and C/EBPα, we performed receptor (10–16). Several different loss-of-function experiments representational difference analysis (RDA), a PCR-based revealed that PU.1 is involved in myeloid and lymphoid subtractive hybridization using fetal livers from wild development (4,5,9,17–19). PU.1 knockout mice completely type and PU.1 or C/EBPα knockout mice. By introducing lack macrophages including osteoclasts, as well as B cells, and a new modification of RDA, that of tissue-specific gene show impaired granulopoiesis and T-cell development (4,5,20). suppression, we could selectively identify a set of –/– However, fetal liver cells from PU.1 mice do express mRNA differentially expressed genes specific to myeloid cells. for early myeloid genes (17). Differentially expressed genes included both primary C/EBPα is a member of the C/EBP family, which has a bZIP and secondary granule protein genes. In addition, eight structure. C/EBPα was originally characterized in liver and adipose novel genes were identified that were upregulated in tissues, and has been shown to regulate a number of hepatic and expression during myeloid differentiation. These adipocyte genes (21–23). Recently, the expression of C/EBPα was methods provide a general strategy for elucidating the shown to initiate with the commitment of multipotential precursors genes affected in murine knockout models. to the myeloid lineage, and be specifically upregulated during granulocytic differentiation (2,24,25). Transient transfection INTRODUCTION studies have shown that C/EBPα can regulate the promoters of a number of myeloid specific genes, such as G-CSF receptor (15), Transcription factors play a major role in cell differentiation, neutrophil elastase (12) and myeloperoxidase (11). C/EBPα including the development of specific hematopoietic lineages knockout mice die within 8 h of birth because they are unable to from stem cells (1,2). Mature myeloid cells, consisting of blood properly synthesize and mobilize glycogen and fat (26,27). They monocytes and tissue macrophages, as well as the neutrophilic also show a selective block in differentiation of neutrophils. and eosinophilic granulocytes, develop from a common myeloid Mature neutrophils and eosinophils are not observed in the blood precursor. However, the mechanism controlling the development or fetal liver of mutant mice and, instead, myeloid blasts are of common myeloid precursors as well as the transition from common precursors into unipotential granulocyte and monocyte observed. Like the PU.1 knockout animals, fetal liver cells from –/– clearly express mRNA for some myeloid genes (3). precursors has not been fully identified. So far several transcrip- C/EBPα tion factors has been implicated in myelopoiesis. These include Other hematopoietic lineages are not affected, including monocytes PU.1, basic leucine zipper (bZIP) CCAAT/enhancer binding and macrophages (3). These results strongly suggest a critical role protein family (C/EBP), acute myelogenous leukemia 1 (AML1), for C/EBPα in granulocytic differentiation. *To whom correspondence should be addressed at: Harvard Institutes of Medicine, Room 954, 77 Avenue Louis Pasteur, Boston, MA 02115, USA. Tel: +1 617 667 5561; Fax: +1 617 667 3299; Email: [email protected] 3035 Nucleic Acids Research, 1998, Vol. 26, No. 12 3035 Nucleic Acids Research, 1994, Vol. 22, No. 1 To understand the impaired myeloid development caused by England Biolabs, Beverly, MA). Fifteen micrograms of each loss-of-function mutation of PU.1 and C/EBPα, myeloid colony- RNA sample were resolved by agarose formaldehyde gel stimulating factor (CSF) receptors have been suggested as critical electrophoresis and transferred to Biotrans nylon membranes targets for PU.1 and C/EBPα (3,17,18). However, the loss-of- (ICN, Biomedicals, Inc., Costa Mesa, CA). The blots were function mutant mice of each CSF or CSF receptor do not show hybridized to [α- P]dCTP-labeled DNA fragments generated by defects in myeloid development as severe as those of PU.1 or RDA as described previously (8) and exposed for 1–2 days with C/EBPα knockout mice (28–31). It is likely that there exist an intensifying screen. To normalize the loading of RNA samples additional critical targets. In order to identify these target genes in each lane, the probe was removed and the blot was rehybridized for PU.1 and C/EBPα during myeloid development, we have to an [α- P]dCTP 3′-end-labeled 18S oligonucleotide (39). analyzed expression of presumptive myeloid target genes in vivo as well as performed representational difference analysis (RDA), cDNA synthesis a PCR-based subtractive hybridization using wild-type and Oligo(dT)-primed double-stranded cDNA was synthesized from knockout mice. In contrast with differential display, which 5–10 μg of poly(A) RNA using a cDNA synthesis system amplifies fragments from all represented mRNA species, RDA (GIBCO BRL, Grand Island, NY) according to the manufacturer’s eliminates those fragments present in both populations, leaving instructions. Tester and driver cDNA samples were synthesized only the difference (32,33). Recently, several groups successfully in parallel at the same time. identified differentially expressed genes using RDA (33–36). In this study, to focus on the differentially expressed genes of myeloid Representational difference analysis lineage, we tried several new modifications of the RDA procedure. Here we show the in vivo role of PU.1 and C/EBPα in RDA was performed using reagents as described (32–34). The regulating myeloid genes by northern blot analysis, and by using following oligonucleotides were synthesized and used for RDA: RDA combined with specific gene suppression, we identified a R-Bgl-24, 5′-AGCACTCTCCAGCCTCTCACCGCA-3′; R-Bgl- set of myeloid genes, the expression of which is missing or 12, 5′-GATCTGCGGTGA-3′; J-Bgl-24, 5′-ACCGACGTCGA- significantly decreased in the mutant fetal liver. These genes CTATCCATGAACA-3′; J-Bgl-12, 5′-GATCTGTTCATG-3′; included myeloid granule protein genes and eight novel myeloid N-Bgl-24, 5′-AGGCAACTGTGCTATCCGAGGGAA-3′; and genes which are new candidate targets for PU.1 and C/EBPα N-Bgl-12, 5′-GATCTTCCCTCG-3′. cDNA was digested with transcription factors. DpnII and ligated to the R-Bgl-12/24 adaptors. Amplicons were made by PCR amplification of the ligated DpnII cDNA fragments for 20 cycles using the R-Bgl-24 as a primer. Driver DNA was MATERIALS AND METHODS prepared by digesting amplicons with DpnII. Tester DNA was Mice, tissues and cells prepared by gel-purification of digested amplicons between 150 and 2000 bp followed by ligation to J-Bgl-12/24 adaptors. First Targeted disruption of C/EBPα and PU.1 was achieved by subtractive hybridization was performed using 400 ng tester and homologous recombination in embryonic stem cells and generation 40 μg driver (tester:driver = 1:100). An aliquot of the hybridization of mice from these cell lines as reported previously (5,26). Fetal mixture was amplified by PCR for 10 cycles using the J-Bgl-24 liver was obtained from embryonic day 19 fetuses. Purification of as a primer. The PCR products were then digested with mung fetal liver hematopoietic cells was performed by passing fetal bean nuclease (New England Biolabs) at 30C for 35 min and liver through 70 μm nylon mesh cell strainers (Becton Dickinson further amplified for 18 cycles. These PCR products are the first Labware, Franklin Lakes, NJ). Peritoneal exudate cells were difference products (DP1). The difference products were digested harvested by lavage with 10 ml PBS 20, 48 and 72 h after i.p. with DpnII and ligated to a new adaptor, N-Bgl-12/24 (after the injection of 1.5 ml 10% thioglycollate broth. Morphological first hybridization) or J-Bgl-12/24 (after the second hybridization), examination revealed that the cells consisted of ~ 15% monocytes/ and the procedure was repeated twice using tester:driver ratios of macrophages and 80% polymorphonuclear granulocytes 20 h after 1:800 and 1:4000–400 000 for the second and third round of injection, and 50 versus 45%, and 80 versus 15% at 48 and 72 h, hybridization, respectively. respectively. In some experiments (see Results), previously generated The murine lymphohematopoietic progenitor cell line EML cDNA fragments were suppressed by adding 150–300 ng of each was maintained in IMDM supplemented with 20% horse serum, DNA fragment without adaptors to each round of hybridization. glutamine, non-essential amino acids and 10% conditioned The suppression of liver genes was performed by adding 40 μg medium from BHK cells transfected with rat stem cell factor driver prepared from adult liver to each round of hybridization. cDNA (BHK-MKL cells) (37). To induce myeloid differentiation, The suppression of mature myeloid genes was performed by EML cells were cultured in IMDM supplemented with 10% adding 30 μg driver prepared from peritoneal exudate cells BHK-MKL conditioned medium, 5% WEHI-3 conditioned collected 20 or 72 h after i.p. injection of thioglycollate. –5 medium and 10 M all-trans retinoic acid (RA) for 72 h. Cells DNA sequences of novel cDNA clones identified by RDA have were then washed three times to remove RA and recultured in been submitted to the GenBank database (accession nos IMDM supplemented with murine GM-CSF (2.5 ng/ml) for AA720492–AA720501). indicated times (37). RESULTS RNA preparation and northern blot analysis Expression of transcription factors in mutant mice Total RNA was isolated by guanidium isothiocyanate extraction followed by CsCl gradient purification (38). Poly(A) RNA was To check the involvement of PU.1 or C/EBPα in the regulation purified from total RNA with oligo(dT) cellulose columns (New of transcription factors likely to play critical roles in myeloid 3036 Nucleic Acids Research, 1998, Vol. 26, No. 12 Figure 1. Northern blot analysis of mRNA of transcription factor genes. Total –/– –/– RNA (15 μg) from wild-type (lanes 1 and 3), C/EBPα (lane 2) and PU.1 (lane 4) day 19 fetal liver was analyzed using species- and gene-specific 5′ Figure 2. Northern blot analysis of mRNA of growth factor receptors in PU.1 cDNA probes of PU.1 and Spi-B (8), a 3′ cDNA probe of C/EBPα (26) and knockout mice. Total RNA (20 μg) from wild-type (lane 1), PU.1 +/– (lane 2), exon 3 of murine C/EBPε (41). –/– or PU.1 (lane 3) day 19 fetal liver was electrophoresed in 1% agarose/ formaldehyde gels, transferred to a nylon membrane and probed with murine cDNAs corresponding to M-CSF receptor (M-CSFr) (59); G-CSF receptor (G-CSFr) (60); GM-CSF receptor α (GM-CSFr) (61); erythropoietin receptor development, we analyzed the expression of PU.1, Spi-B, (EPOr) (62); and 18S oligonucleotide (39). C/EBPα and C/EBPε (40,41), all of which have been shown or postulated to play a role in myeloid development (2), in the mutant fetal livers (Fig. 1). PU.1 knockout mice were made by with previous reports indicating the presence of early myeloid disrupting the DNA binding domain, inserting the neo gene –/– gene expression in PU.1 fetal liver cells (17). GM-CSF within exon 5 (5), but no PU.1-related transcripts were detected –/– –/– receptor mRNA was not detectable in PU.1 fetal liver by in PU.1 fetal liver by a probe specific to the 5′ end of the PU.1 northern blot analysis, although it has been detected by RT–PCR cDNA (8), confirming that this is indeed a null phenotype. PU.1 –/– (17). These findings are consistent with promoter studies expression was decreased ~ 50% in C/EBPα fetal liver, demonstrating a functional PU.1 site in transient transfection consistent with the presence of a significant number of immature analysis (14–16). The levels of erythropoietin receptor mRNA myeloid cells in these animals (3). Spi-B is an Ets transcription –/– were not affected in PU.1 fetal livers compared with wild type, factor closely related to PU.1 but now known to be expressed in keeping with the lack of a consistent effect of PU.1 disruption primarily in B cells (8,42). Spi-B expression was slightly –/– on erythropoiesis (4,5). Although PU.1 and C/EBPα knockout decreased in C/EBPα fetal liver, but was undetectable in –/– mice have no detectable, or very low, levels of CSF receptor PU.1 fetal liver. C/EBPα expression was not affected in –/– expression, their defects in myeloid development are more severe PU.1 fetal liver, but it is very hard to detect a difference in than those of loss-of-function mutant mice of CSFs or CSF hematopoietic cells because of the high C/EBPα expression by receptors (28–31). This suggests that there are additional genes hepatocytes and adipocytes. Another myeloid specific C/EBP regulated by PU.1 and C/EBPα whose altered expression lead to the transcription factor, C/EBPε, which is critical for terminal observed phenotype. Therefore, we performed RDA to identify myeloid maturation (41), was not expressed in both mutant fetal these additional genes regulated by these transcription factors. livers. These findings are consistent with the lineage-specific expression of Spi-B and C/EBPε in B cell and granulocyte lineages, respectively, and suggest that they might be regulated by Identification of differentially expressed genes between PU.1 and/or C/EBPα. Alternatively, the knockout cells might be C/EBPα +/+ and –/– fetal livers by RDA blocked in their differentiation and do not become mature enough As noted above, C/EBPα-deficient mice show a selective block to express Spi-B or C/EBPε. in differentiation of neutrophils. Other hematopoietic lineages, including monocytes, are not affected (3). To identify C/EBP- mRNA expression of myeloid CSF receptors in mutant mice α-regulated genes during neutrophilic differentiation, we performed We have previously shown that G-CSF receptor mRNA is RDA using C/EBPα-deficient mice. The cDNA in which the selectively downregulated in C/EBPα knockout mice by northern differentially expressed cDNAs are to be found is called ‘tester’ blot analysis, whereas M-CSF receptor and GM-CSF receptor cDNA, and the reference cDNA is called ‘driver’ cDNA. The mRNA levels are not impaired (3). This suggests that impaired cDNA from each population is digested with a restriction G-CSF signaling might be in part responsible for the selective endonuclease, ligated to adaptors, and then amplified by PCR. block of neutrophilic differentiation. To determine the role of The products of amplification are called amplicons. To isolate –/– myeloid CSF receptors in the defect found in PU.1 mice, we amplicons unique to the tester cDNA, tester amplicon was ligated –/– analyzed their expression in PU.1 fetal liver by northern blot to new adaptors and hybridized to an excess of driver amplicon. analysis (Fig. 2). The expression of M-CSF receptor and G-CSF PCR with primers for the new adaptors preferentially amplifies receptor mRNA was markedly decreased but detectable, consistent tester–tester homoduplexes. This process is repeated several 3037 Nucleic Acids Research, 1998, Vol. 26, No. 12 3037 Nucleic Acids Research, 1994, Vol. 22, No. 1 which has been previously characterized in transient transfection AB studies as one of the targets for C/EBPα (12). DP2 (1:800) contained more genes than DP3, and most of them were distinct from DP3 except for contrapsin and protein C. Interestingly, DP2 contained three additional differentially expressed genes, but the expression of the remaining seven genes showed no difference between C/EBPα +/+ and –/– fetal liver or was unexpectedly –/– upregulated in C/EBPα fetal liver (Table 1). These results indicate that although RDA could selectively amplify differentially expressed genes, to keep high specificity, it requires a high stringency, which causes a limit on the number of genes amplified. Table 1. DNA fragments generated by RDA, C/EBPα +/+ minus –/– with different stringencies Figure 3. (A) Alkaline agarose gel analysis of second strand cDNA synthesis. Gene (accession no.) Expression α- P-labeled second strand cDNAs made from wild-type (lane 1) or E19 fetal liver BM Adult –/– C/EBPα (lane 2) day 19 fetal liver mRNA were electrophoresed on a 1.4% liver li C/EBPα +/+ –/– alkaline agarose gel, and the dried gel was exposed to X-ray film. (B) Agarose electrophoresis of difference products generated by RDA. Amplicon from DP3 (tester:driver = 1:400 000 –/– wild-type day 19 fetal liver cDNA (lane 2), amplicon from C/EBPα day 19 fetal liver cDNA (lane 3), first difference product (DP1) with tester to driver C/EBPα (M62362) + – + + ratio of 1:100 (lane 4), DP2 with tester to driver ratio of 1:800 (lane 5), DP3 with Neutrophilic elastase (U04962) 2+ – 3+ – tester to driver ratio of 1:4000 (lane 6) and 1:400 000 (lane 7) and DNA Contrapsin (X55147) 2+ – – 3+ molecular markers (lanes 1 and 8) were electrophoresed on a 2.0% agarose gel. Protein C (D10445) 5+ + – 6+ Unknown 1 + – – + DP2 (tester:driver = 1:800) times with increasing driver-to-tester ratios until only fragments Contrapsin (X55147) unique to the tester remain (32,33). Protein C (D10445) Since C/EBPα mutant mice die soon after birth, we used day Haptoglobin (S67972) 2+ – + 2+ 19 fetal liver as material for RDA. Because C/EBPα also regulates transcription of hepatocyte- and adipocyte-specific Apolipoprotein A-I (L04151) 4+ + – 3+ genes, these genes as well as myeloid-specific genes were Eosinophil chemotactic factor candidates for identification from fetal liver by RDA. Poly(A) (X15313) + – 3+ – RNA was purified from fetal livers of day 19 embryos and Pref-1 (L12721) + 5+ – – double-stranded cDNA was synthesized using an oligo(dT) Unknown 2 + + – + primer. After ligating adaptors to DpnII-digested cDNAs, the Unknown 3 + + – + tester and driver amplicons were generated by PCR amplification. Unknown 4 + + – 2+ It is important that preparations to be subtracted are as similar as Unknown 5 + + – + possible as shown in Figure 3. If the quality of cDNA and Unknown 6 2+ 2+ – 3+ amplicon varies between subtracted populations, this can result in amplification of false positives. The tester was subtracted with the Relative expression was evaluated by northern blot analysis. Evaluation of mRNA driver, and the difference was selectively amplified by PCR. This levels are consistent for each gene but cannot be compared among genes. process was repeated three times until the difference products Genes differentially expressed. (DPs) showed clear bands with little background visible by ethidium bromide staining (Fig. 3B). The third subtraction was performed with different hybridization ratios. The third difference Suppression of liver genes during RDA subtractive product (DP3) with a higher stringency (tester:driver = 1:400 000) hybridization showed fewer bands than that with lower stringency (1:4000), –/– indicating that more genes were suppressed in the third subtraction As shown in Table 1, RDA using C/EBPα fetal liver amplified by increasing ratios of tester to driver (Fig. 3B). After digesting liver genes more than myeloid genes, possibly because day 19 fetal with DpnII, DP2 (1:800) and DP3 (1:400 000) were separated on liver contains more liver RNA than myeloid RNA. To preferentially an agarose gel and each band was excised out and subcloned. The amplify myeloid genes, we modified the RDA technique. First of subcloned inserts were used in northern blot analysis as probes to all, we enriched hematopoietic cells by using cell strainers, nylon check mRNA expression of identified genes. mesh devices with 70 μm pore size which select for cell size. We Table 1 shows the profile of genes identified by RDA in this could enrich hematopoietic cells ~ 3–4-fold by passing fetal livers screening. DP3 with a higher stringency (1:400 000) contained through cell strainers (data not shown). We then prepared C/EBPα, demonstrating that the RDA procedure was selecting liver-specific DNA fragments generated in the previous RDA differentially expressed genes. The northern blot analysis revealed (Table 2) and driver from adult liver, and added either of them in that all genes were truly differentially expressed, i.e. expressed in the hybridization mixture to suppress the amplification of +/+ –/– the C/EBPα fetal liver, but not or at lower levels in C/EBPα liver-specific genes. We used a high stringency of 1:400 000 for fetal liver (Table 1). As expected, DP3 contained both myeloid- the third round of subtractive hybridization. The DP3 showed specific and liver-specific genes, including neutrophilic elastase, several clear bands on an agarose gel, and the profile of bands was 3038 Nucleic Acids Research, 1998, Vol. 26, No. 12 AB C –/– Figure 4. Northern blot analysis of mRNA of novel genes generated by RDA. (A) Total RNA (15 μg) from wild-type (lane 1) and C/EBPα (lane 2) day 19 fetal liver, bone marrow (lane 3), peritoneal exudate cells 48 h after thioglycollate stimulation (lane 4), spleen (lane 5), thymus (lane 6) and adult liver (lane 7). (B) Total –/– RNA (15 μg) from wild-type (lanes 1 and 3) and PU.1 (lanes 2 and 4) day 19 fetal liver treated (lanes 1 and 2) and untreated (lanes 3 and 4) with cell strainers, bone marrow (lane 5), peritoneal exudate cells 48 h after thioglycollate stimulation (lane 6), spleen (lane 7), thymus (lane 8) and adult liver (lane 9). (C) Poly(A) RNA –/– (3 μg) from wild-type (lane 1) and C/EBPα (lane 2) day 19 fetal liver, bone marrow (lane 3), peritoneal exudate cells 20 h (lane 4) and 72 h (lane 5) after thioglycollate stimulation, spleen (lane 6), thymus (lane 7) and adult liver (lane 8). DNA fragments generated by RDA were used as probes. very similar with each type of suppression. Nucleotide sequence monocytes/macrophages and lymphoid cells in the blood or fetal analysis of DP3 revealed that the suppression of liver-specific genes liver, and die from septicemia within 2 days of birth. However, worked very well (Table 2); liver genes were dramatically antibiotic-treated mice could survive for 2 weeks and show the suppressed during the subtractive hybridization. DP3 with specific development of normal appearing T cells and a few cells with the suppression by liver-specific DNA fragments contained only two characteristics of neutrophils (5). To identify PU.1-regulated liver-specific genes, and DP3 with suppression by the driver genes during myeloid differentiation, we performed RDA using prepared from adult liver contained no liver-specific genes. The day 19 fetal liver of PU.1-deficient mice. We prepared cDNAs successful suppression of liver genes led to the amplification of more from whole fetal liver and enriched hematopoietic cells by a cell myeloid genes. Many genes for primary and secondary granule size selection using cell strainers, and compared the profile of proteins of neutrophils were identified, including myeloperoxidase amplified genes. Because PU.1 is not expressed in hepatocytes, and neutrophilic elastase, targets for C/EBPα identified by only myeloid and lymphoid genes were expected to be identified transient transfection assays (11,12) (Table 2). In addition to from the fetal liver by RDA. known myeloid genes, five novel genes were identified (C/Edp Table 3 shows the profile of genes identified by RDA. We used 1–5, Table 2). C/Edp 2 and 3 showed high nucleotide sequence a high stringency of 1:400 000 for the third round of subtractive similarity to human neutrophil collagenase (79%) and human hybridization. As expected, most of the genes contained in DP3 ficolin (72%), respectively, suggesting that C/Edp 2 and 3 are were myeloid-specific genes, including those specific to the putative murine homologues of these genes. The other genes neutrophil and/or monocyte/macrophage lineage, and others showed no significant similarity to any known genes. Northern were lymphoid genes. There was no significant difference in the blots revealed that all of these unknown genes were differentially profile between DP3 from whole fetal liver and purified expressed and were preferentially expressed in the BM and/or hematopoietic cells, suggesting that this procedure is very peritoneal exudate cells consisting of mature neutrophils and sensitive, but we could amplify different genes by using different monocytes (Fig. 4A). These results demonstrate that liver genes materials. We identified five unknown genes (Pdp 1 and Pdp could be successfully suppressed by using specific gene fragments 3–6). They showed no significant similarity to any known genes. or adult liver driver, and this suppression facilitates the amplification Interestingly, two of them were the same genes as those identified of myeloid-specific genes. by RDA using C/EBPα knockout mice. Northern blots revealed that all of these unknown genes were differentially expressed and Identification of differentially expressed genes between also preferentially expressed in the BM and/or peritoneal exudate PU.1 +/+ and –/– fetal livers by RDA cells (Fig. 4B). Therefore, RDA is sensitive and specific enough PU.1-deficient mice show impaired myeloid and lymphoid to identify the difference in a small subpopulation from materials development (4,5). The mutant mice lack mature neutrophils, comprised of heterogenous cell populations. 3039 Nucleic Acids Research, 1998, Vol. 26, No. 12 3039 Nucleic Acids Research, 1994, Vol. 22, No. 1 Table 2. DNA fragments generated by RDA, C/EBPα +/+ minus –/– with the immature myeloid cells, i.e. from myeloblasts to band cells, but suppression of liver-specific genes not in mature myeloid cells including peritoneal exudate cells (reviewed in 43). Liver genes were suppressed by adult liver Gene (accession no.) driver as described above. We used lower stringencies, 1:40 000 for the third round of subtractive hybridization, because the expression Suppression with liver gene fragments ( ) of differentially expressed genes specific to immature myeloid Lactoferrin (D88510) cells was expected to be weaker than before. As shown in Table 4, Myeloid bectenecin (U95002) we obtained quite a different profile of genes. Although we failed Lipocortin I (M24554) to suppress gelatinase B and Pdp 4 with specific DNA fragments, Neutrophil gelatinase associated lipocalin (X81627) suppression with mature myeloid cDNA and other specific DNA Eosinophil chemotactic factor (X15313) fragments worked well. We isolated a new DpnII fragment of Neutrophilic elastase (U04962) MPO, eosinophil peroxidase (EPO), proteinase-3 and gelatinase C/Edp 4 (Pdp 3) B. They are myeloid granule proteins and are expressed in immature myeloid cells (43,44). Others were a Kupffer cell-specific C/Edp 5 (Pdp 4) b gene, a B-cell gene and three unknown genes. The unknown Haptoglobin (M96827) genes showed no significant similarity to any known genes. C4 complement protein (M11789) Northern blots revealed that C/Edp 6 and 7 were differentially Suppression with adult liver cDNA expressed (Fig. 4C), but unknown 7 was not (data not shown). Lactoferrin (D88510) Because the expression of C/Edp 6 and 7 was relatively weak, Myeloid bectenecin (U95002) poly(A) RNA northerns were required for identification. C/Edp Lipocortin I (M24554) 6, 7 and Pdp 4 were not expressed in peritoneal exudate cells at Neutrophil gelatinase associated lipocalin (X81627) all (Fig. 4B and C). These results demonstrate that suppression of Eosinophil chemotactic factor (X15313) mature myeloid genes facilitates the amplification of genes preferentially expressed in immature myeloid cells. Myeloperoxidase (X15313) Gelatinase B (D12712) Stefin 1 (M92417) Table 3. DNA fragments generated by RDA, PU.1 +/+ minus –/– C/Edp 1 (AA720492) C/Edp 2 (AA720493) C/Edp 3 (AA720494) Gene (accession no.) C/Edp 4 (Pdp 3; AA720498) Whole fetal liver C/Edp 5 (Pdp 4; AA720499) Lysozyme M (M21050) Bacteria binding macrophage receptor (U18424) DNA fragments previously generated by RDA (contrapsin, protein C, haptoglobin, Complement subcomponent C1q α-chain (X58861) apolipoprotein A-I) were used for suppression. Lactoferrin (D88510) This haptoglobin fragment is a different one from that used for suppression. Eosinophil chemotactic factor (X15313) C/Edp, Pdp: Unknown differentially expressed gene isolated by RDA, C/EBP Ig λ-chain (M30387) +/+ minus –/– (C/Edp) and PU.1 +/+ minus –/– (Pdp). Pdp 1 (AA720497) Pdp 3 (C/Edp 4; AA720498) Pdp 4 (C/Edp 5; AA720499) Pdp 5 (AA720500) Suppression of mature myeloid genes leads to the amplification of immature myeloid-specific genes Pdp 6 (AA720501) Enriched fetal liver hematopoietic cells The absence of C/EBPα and PU.1 causes a block at an early stage Lysozyme M (M21050) of myeloid differentiation. The critical targets responsible for this Bacteria binding macrophage receptor (U18424) differentiation block are expected to be also expressed at early Complement subcomponent C1q α-chain (X58861) stage. To focus on the early targets during myeloid differentiation, Lactoferrin (D88510) we used fetal liver hematopoietic cells enriched by cell size Myeloperoxidase (X15313) selection, and performed suppression of mature myeloid genes. Myeloid bectenecin (U95002) We prepared driver amplicons from peritoneal exudate cells collected 20 and 72 h after i.p. injection of thioglycollate. The gp91phox (U43384) former cells consisted of ~ 80% of neutrophils, while the latter LAPTm 5 (U51239) consisted of 80% of monocytes/macrophages. Driver from MHC class II H2–1A-α (M11357) peritoneal exudate cells 20 h after stimulation was added to the Ig λ-chain (M30387) hybridization mixture of C/EBPα +/+ and –/– to suppress Pdp 1 (AA720497) neutrophilic genes, and both drivers were added to that of PU.1 Pdp 6 (AA720501) +/+ and –/– to suppress both neutrophilic and monocyte/macrophage genes. We also performed specific gene suppression using DNA fragments from our novel RDA clones and myeloid granule C/Edp, Pdp: unknown differentially expressed gene isolated by RDA, C/EBPα proteins (Table 4), because most of them are expressed only in +/+ minus –/– (C/Edp) and PU.1 +/+ minus –/– (Pdp). 3040 Nucleic Acids Research, 1998, Vol. 26, No. 12 Table 4. DNA fragments generated by RDA with the suppression of mature myeloid cDNAs C/EBPα +/+ minus –/– PU.1 +/+ minus –/– a a Myeloperoxidase (X15313) myeloperoxidase (X15313) Eosinophil peroxidase (D78353) gelatinase B (D12712) Proteinase-3 (U43525) Kupffer cell receptor (D88577) procathepsin E (X97399) C/Edp 6 (AA720495) Pdp 4 (AA720499) C/Edp 7 (AA720496) Unknown 7 Drivers prepared from peritoneal exudate cells and previously generated DNA fragments (C/Edp 1–3, Pdp 1–6, myeloperoxidase, neutrophilic elastase, lactoferrin, gelatinase B, myeloid bectenecin, neutrophil gelatinase associated lipocalin, lipocortin I and eosinophil chemotactic factor) were used for suppression. This MPO fragment is different from that in Tables 2 and 3. Expression of myeloid granule protein genes in mutant mice In this study, we identified many myeloid granule protein genes, including primary granule protein genes (myeloperoxidase, neutrophilic elastase and proteinase-3); secondary granule protein genes [lactoferrin, neutrophil gelatinase associated lipocalin (NGAL), putative murine homologue of neutrophilic collagenase C/Edp 2, gelatinase B and myeloid bectenecin]; and lysozyme M, Figure 5. Northern blot analysis of mRNA of myeloid granule protein genes. –/– –/– Total RNA (15 μg) from wild-type (lanes 1 and 3), C/EBPα (lane 2), PU.1 which is localized in both primary and secondary granules. –/– (lane 4) day 19 fetal liver, and wild-type (lanes 5 and 7), C/EBPα (lane 6) and Among them, myeloperoxidase and neutrophilic elastase have –/– PU.1 (lane 8) newborn lung. DNA fragments generated by RDA were used been characterized as common targets for PU.1 and C/EBPα by as probes. transient transfection assays (11,12). To determine the in vivo role of PU.1 and C/EBPα in the regulation of myeloid granule protein genes, we analyzed their expression by northern blotting. The generated by treatment with high concentrations of retinoic acid. expression of many of them was very low or undetectable in both These myeloid progenitors differentiate into neutrophils and mutant fetal livers in vivo (Fig. 5). Specifically, myeloperoxidase macrophages in response to GM-CSF, but still neutrophilic and proteinase-3 were expressed at very low levels in both mutant differentiation is blocked around the promyelocyte to myelocyte –/– fetal livers, NGAL and C/Edp2 at low levels in PU.1 fetal liver, stages and only few mature neutrophils could be observed. –/– and lysozyme M at low levels in C/EBPα fetal liver (Fig. 5). Differentiated neutrophilic cells appeared on day 3 after treatment Although mRNAs encoding myeloid granule proteins are confined with GM-CSF (blasts 14.3%, promyelocytes 53.0%, myelocytes/ to myeloid cells, lysozyme M is abundantly expressed also in metamyelocytes 24.5% and monocytes/macrophages 8.0%), non-hematopoietic tissues, particularly in the lung (45), in which reached a peak on day 6 (blasts 6.0%, promyelocytes 28.0%, C/EBPα is highly expressed (46). Interestingly, the expression of myelocytes/metamyelocytes 47.5%, band/segmented cells 8.5% –/– lysozyme was markedly downregulated in C/EBPα newborn and monocytes/macrophages 9.7%), and then decreased, while –/– lung but not in PU.1 newborn lung (Fig. 5). These findings macrophages gradually increased and dominated on day 10 suggest the critical role of PU.1 and C/EBPα in the regulation of (blasts 3.5%, promyelocytes 19.7%, myelocytes/metamyelocytes myeloid granule protein genes in vivo. 8.1% and monocytes/macrophages 69.8%). As shown in Figure 6, the expression of C/Edp 1–3 and Pdp 3 and 4 were strongly induced during myeloid differentiation, and downregulated on Myeloid-specific expression of novel genes day 10. The 2.0 kb transcript of Pdp 6 was weakly induced, while To clarify the lineage-specific expression of novel genes, we the 1.0 kb transcript, which is a minor transcript in peritoneal analyzed their expression in lymphohematopoietic tissues. Most exudate cells (Fig. 4B), was strongly upregulated. The expression of them were preferentially expressed in the bone marrow or of Pdp 5 was upregulated 3 days after GM-CSF stimulation and peritoneal exudate cells, but not in spleen, thymus or adult liver maintained during differentiation. The analysis of the expression (Fig. 4A and B), suggestive of their preferential expression in of these novel myeloid genes in other hematopoietic cell lines myeloid cells. revealed that they were expressed in myeloid cells but not in T We further analyzed their expression during the myeloid cells, B cells or erythroid cells (data not shown). Therefore, RDA differentiation of EML cells. EML is a stem cell factor-dependent selectively amplified differentially expressed genes which are lymphohematopoietic progenitor cell line immortalized by a preferentially expressed during myeloid differentiation. The retroviral vector harboring a dominant-negative retinoic acid expression of C/Edp 6 and 7 was not detected in EML cells or receptor (37). Myeloid differentiation is suppressed in EML cells, other hematopoietic cell lines (data not shown); therefore, they but common progenitors for neutrophils and macrophages are may be preferentially expressed in fetal liver. 3041 Nucleic Acids Research, 1998, Vol. 26, No. 12 3041 Nucleic Acids Research, 1994, Vol. 22, No. 1 amplifies only the difference. RDA is sensitive enough to isolate genes expressed in only a very small percentage of cells (32,33). Therefore, this technique was suitable for our cloning approach using fetal livers as materials in which myeloid cells compose only a small percentage of the total cell population. Our data demonstrated that this procedure truly amplified differentially expressed genes, and was able to amplify genes expressed at low levels as well. Most of the genes identified were expressed in the fetal liver at much lower levels than in the bone marrow or peritoneal exudate cells (Fig. 4). However, because so –/– many liver genes are differentially expressed in C/EBPα fetal liver, we amplified more liver genes than myeloid genes (Table 1). To suppress the amplification of liver genes, we first tried suppression by liver-specific DNA fragments. Suppression of expected difference products by specific DNA fragments has been reported to facilitate the amplification of new gene fragments (32,33). We prepared liver-specific DNA fragments generated in the previous RDA, and added these to the hybridization mixture. This suppression worked well, but other liver genes were still amplified (Table 2). To get complete suppression of liver genes and preferentially amplify myeloid genes, we prepared driver amplicon from adult liver and added it into each round of subtractive hybridization. As shown in Tables 2 and 4, liver genes were completely suppressed and this facilitated Figure 6. Northern blot analysis of mRNA of novel genes identified by RDA the amplification of myeloid genes. This modification was also during myeloid differentiation of EML cells. Total RNA (15 μg) from before successfully applied to suppression of mature myeloid genes to –5 stimulation (lane 1), after treatment with RA (10 M) and IL-3 for 3 days amplify immature myeloid genes (Table 4). Our data demonstrate (lane 2), after 1 day (lane 3), 3 days (lane 4), 6 days (lane 5) and 10 days (lane 6) of culture in the presence of GM-CSF. The same probes as in Figure 4 were used. that genes expressed in a certain cell population or at a specific stage of differentiation could be completely suppressed by the appropriate driver, and this suppression facilitates the amplification of differentially expressed genes in other cell populations or differentiation stages. RDA combined with this kind of gene DISCUSSION suppression would be helpful to focus on genes specific to a certain cell population in materials consisting of heterogenous cells. The receptors for the myeloid colony-stimulating factors, Although RDA is an effective technique, it still has some M-CSF, GM-CSF and G-CSF have been proposed to be critical limitations. First, our data clearly showed that some of the targets for the impaired myeloid development in PU.1 and differentially expressed genes are lost during the repeated C/EBPα mutant mice (3,17,18). We have previously shown by subtractive hybridization by increasing the stringencies (Table 1 northern blot analysis that G-CSF receptor mRNA is remarkably and Fig. 3B). Difference products with a low stringency could downregulated in C/EBPα knockout mice, whereas mRNAs for contain more differentially expressed genes but many more false M-CSF receptor and GM-CSF receptor are not impaired (3), positives as well. On the other hand, difference products with a suggesting that impaired G-CSF signaling might be responsible high stringency limit the number of fragments generated. This for selective block of neutrophilic differentiation. On the other problem could partially be resolved by suppression of expected hand, M-CSF receptor mRNA was undetectable by RT–PCR –/– difference products by specific DNA fragments or additional analysis of differentiated PU.1 ES cells (17,18). In this study, –/– drivers as we performed in this study. Secondly, RDA preferentially we analyzed the expression of myeloid CSF receptors in PU.1 amplifies genes with significant differences in expression. Most fetal liver by northern blot analysis, and noted that the expression of the differentially expressed genes identified were not expressed of all three is markedly decreased (Fig. 2). However, at least in the mutant fetal liver. Only several genes were still expressed M-CSF receptor and G-CSF receptor are still expressed at low –/– in mutant fetal liver at low levels. Decreasing the stringencies was levels, suggesting that PU.1 fetal liver cells could express at not as effective (Table 1). New modifications will be needed to least low levels of myeloid CSF receptors. No complementation amplify genes with small differences. A minor limitation is that the assays to rescue the defects by using myeloid CSF receptor transgenes have been reported. Therefore, the role of myeloid technique tends to isolate small portions of the full length cDNA. CSF receptors in both mutant mice still remains to be defined. In By RDA using PU.1 and C/EBPα knockout mice, we identified addition, the myeloid defects of PU.1 or C/EBPα knockout mice many differentially expressed genes, including myeloid- and do not completely match those of loss-of-function mutant mice of liver-specific genes. The expression of several liver genes have each CSF or CSF receptor (28–31). These findings suggest the already been shown to be downregulated in the fetal and newborn presence of additional genes regulated by PU.1 and C/EBPα. liver of C/EBPα knockout mice (26,27). We identified six In this study, we extended the studies of CSF receptor expression additional liver genes which are differentially expressed, and they and identified additional genes regulated by PU.1 and C/EBPα. We are presumably new targets for C/EBPα in hepatocytes. employed RDA, a PCR-based subtractive hybridization. RDA Interestingly, we happened to find that the expression of pref-1, eliminates those fragments present in both populations and a pre-adipocyte transmembrane protein, is upregulated in C/EBP- 3042 Nucleic Acids Research, 1998, Vol. 26, No. 12 –/– α fetal liver. During adipocyte differentiation, pref-1 is an important role of these transcription factors in the regulation reported to be downregulated, while C/EBPα is upregulated of lysozyme M expression. Moreover, we found that lysozyme M –/– (23,47). Our finding suggests that C/EBPα negatively regulates expression was impaired in the C/EBPα newborn lung. the expression of pref-1, and pref-1 might be a new direct target Lysozyme is expressed in type II alveolar pneumocytes and for C/EBPα in adipocytes. This finding also suggests the alveolar macrophages in rodent lung (57), while C/EBPα mRNA –/– existence of other negatively regulated genes by C/EBPα or PU.1. is localized to type II pneumocytes (58) and C/EBPα mice To selectively identify these genes, reverse RDA, i.e. mutant fetal show hyperproliferation of type II pneumocytes (27). C/EBPα is liver minus wild-type fetal liver, would be an approach to be taken also expressed in activated macrophages (A.Iwama and in the next step. D.G.Tenen, unpublished data). Taken together with the fact that With some modifications of RDA, we could preferentially PU.1 is not expressed in type II pneumocytes, C/EBPα could be identify myeloid-specific genes. Because PU.1 knockout mice a major regulator for lysozyme M expression in the lung, show impaired development of neutrophils as well as monocytes particularly in type II pneumocytes. These findings indicate that in the fetal liver, many neutrophil-specific genes were identified C/EBPα plays a major role in the regulation of lysozyme M in from both PU.1 and C/EBPα knockout mice (Tables 1–4). non-hematopoietic cells, and suggest the possibility that C/EBPα Among known genes, primary granule protein genes myeloper- is a key transcription factor in the regulation of genes specific to oxidase and neutrophilic elastase are known as common targets type II pneumocytes, such as surfactant protein genes. for both PU.1 and C/EBPα (11,12), and another primary granule The expression analysis of transcription factors in vivo protein gene, proteinase-3, is a target for PU.1 (13). Myeloperoxi- indicated that C/EBPα does not affect the expression of PU.1, + –/– dase mRNA is expressed in CD34 multipotential cells (48,49) because the reduction of PU.1 mRNA in C/EBPα fetal liver is and at high levels in myeloid progenitors at the promyelocytic and likely to parallel the decrease in mature granulocytic cells. On the promonocytic stages of myeloid differentiation (50,51), while contrary, PU.1 might be important in the regulation of C/EBPα, neutrophilic elastase and proteinase-3 mRNAs are expressed in and it is possible that impaired granulopoiesis in PU.1 knockout the promyelocytic stage (52). Although myeloid progenitors are mice is caused by defective C/EBPα expression. The expression –/– present in both knockout mice (3,17), and monocytic cells are intact analysis of C/EBPα in the neutrophilic cells of PU.1 bone –/– in C/EBPα fetal liver (3), the expression of myeloperoxidase marrow would help to address this question. The specific absence –/– mRNA was significantly low in both mutant fetal livers (Fig. 5), of Spi-B mRNA only in PU.1 fetal liver is consistent with our indicating that both PU.1 and C/EBPα are critical for the previous data of its B-cell-specific expression (8), and the –/– transcription of myeloperoxidase in vivo. Neutrophilic elastase absence of C/EBPε mRNA in C/EBPα fetal liver as well as –/– mRNA was also missing and proteinase-3 mRNA was markedly PU.1 fetal liver confirms its granulocyte-specific expression downregulated in both mutant mice. They are also likely in vivo (40). Spi-B and C/EBPε might be regulated by PU.1 or both PU.1 targets of these two transcription factors. These findings suggest and C/EBPα in each cell lineage. cooperative regulation of myeloid primary granule genes by PU.1 We identified eight novel myeloid genes differentially expressed and C/EBPα in vivo. between wild-type and mutant fetal livers. Among them, C/Edp The expression of secondary granule protein genes were also 2 and 3 are likely to be the murine homologues of neutrophil undetectable in both mutant mice (Fig. 5), possibly because of the collagenase and ficolin, respectively. C/Edp 1 and Pdp 3, and Pdp lack of expressing cells, such as myelocytes, metamyelocytes and 1 and 6 seem to be different DpnII fragments from the same gene, band cells. Low levels of mRNA of NGAL and neutrophilic respectively, because they showed the same mRNA expression –/– collagenase (C/Edp2) were detected only in PU.1 fetal liver. profile (Figs 4A and B, and 6). Northern blot analysis showed that This might well represent the development of a few neutrophilic most of the novel genes were preferentially expressed in the bone –/– cells in PU.1 fetal liver, consistent with the reported development marrow or peritoneal exudate cells, but not in spleen, thymus or –/– of cells characteristic of neutrophils in PU.1 bone marrow (5). adult liver (Fig. 4A and B). They were undetectable or only Recently, CCAAT displacement protein has been reported to weakly expressed in other tissues, such as brain, heart, lung, repress the expression of secondary granule protein genes kidney, skeletal muscle and testis (data not shown), indicating that (53,54), but transcription factors that directly activate their they are hematopoietic-specific genes. Only Pdp 1 and 6 were transcription have not been well characterized. Because the expressed in adult liver (Fig. 4B), but they were not differentially –/– expression of PU.1 and C/EBPα is maintained during granulocytic expressed between wild type and C/EBPα fetal liver (data not differentiation, they are candidate regulators of secondary shown), suggesting that they are expressed in macrophages, granule expression, as is C/EBPε (40,41). including Kupffer cells, in the liver. Most of them were Lysozyme M is a myeloid granule protein localized in both upregulated during myeloid differentiation of the multipotential primary and secondary granules. Its expression is already detectable hematopoietic cell line, EML (Fig. 6), suggesting these genes are in myeloblasts and upregulated during myeloid differentiation, good candidate targets for PU.1 and C/EBPα. Although C/Edp 6 including both granulocytic and monocytic lineages (43). PU.1 is and 7 were differentially expressed between wild-type and reported to activate the myeloid-specific enhancer of the chicken mutant fetal liver, we could not detect any apparent expression in lysozyme gene (55), and C/EBPβ interacts with another enhancer the bone marrow, peritoneal exudate cells, adult liver or other and mediates lipopolysaccharide-induced expression of the chicken adult organs. It is possible that they are specifically expressed in lysozyme gene (56). Northern blot analysis showed that lysozyme the embryonic stage. –/– M expression was absent in PU.1 fetal liver and markedly Our data confirmed the critical role of PU.1 and C/EBPα in vivo –/– impaired in C/EBPα fetal liver (Fig. 5). In addition, lysozyme in the regulation of myeloid genes, including myeloid CSF M mRNA was immediately upregulated after induction of receptors and myeloid granule proteins. Using RDA combined C/EBPα expression in an immature hematopoietic cell line with specific gene suppression, we further identified novel (A.Iwama and D.G.Tenen, unpublished data). These data suggest myeloid genes, the expression of which are missing in the mutant 3043 Nucleic Acids Research, 1998, Vol. 26, No. 12 3043 Nucleic Acids Research, 1994, Vol. 22, No. 1 23 Cao,Z., Umek,R.M. and McKnight,S.L. (1991) Genes Dev., 5, 1538–1552. fetal livers. These novel genes are new candidate targets for PU.1 24 Scott,L.M., Civin,C.I., Rorth,P. and Friedman,A.D. (1992) Blood, 80, and C/EBPα. Characterization of their roles in myeloid 1725–1735. development as well as their transcriptional regulation in relation 25 Radomska,H.S., Huettner,C.S., Zhang,P., Cheng,T., Scadden,D.T. and to PU.1 and C/EBPα will be helpful in elucidating the mechanism Tenen,D.G. (1998) Mol. Cell. Biol., in press. of impaired myeloid development caused by loss-of-function 26 Wang,N.D., Finegold,M.J., Bradley,A., Ou,C.N., Abdelsayed,S.V., Wilde,M.D., Taylor,L.R., Wilson,D.R. and Darlington,G.J. (1995) Science, mutation of PU.1 and C/EBPα. Additional studies, including 269, 1108–1112. isolation and characterization of the promoter elements, will be 27 Flodby,P., Barlow,C., Kylefjord,H., Ahrlund-Richter,L. and required to establish whether these genes are directly or indirectly Xanthopoulos,K.G. (1996) J. Biol. Chem., 271, 24753–24760. regulated by PU.1 and C/EBPα. 28 Yoshida,H., Hayashi,S., Kunisada,T., Ogawa,M., Nishikawa,S., Okamura,H., Sudo,T. and Shultz,L.D. (1990) Nature, 345, 442–444. 29 Nishinakamura,R., Nakayama,N., Hirabayashi,Y., Inoue,T., Aud,D., ACKNOWLEDGEMENTS Mcneil,T., Azuma,S., Yoshida,S., Toyoda,Y., Arai,K., Miyajima,A. and Murray,R. (1995) Immunity, 2, 211–222. 30 Lieschke,G.J., Stanley,E., Grail,D., Hodgson,G., Sinickas,V., Gall,J.A.M., We thank Linda Clayton for her technical instructions on RDA, Sinclair,R.A. and Dunn,A.R. (1994) Blood, 84, 27–35. Schickwann Tsai and Arch Perkins for providing EML and 31 Liu,F., Wu,H.Y., Wesselschmidt,R.L., Kornaga,T. and Link,D.C. (1996) BHK-MKL cells, respectively, Kleanthis Xanthopoulos for the Immunity, 5, 491–501. murine C/EBPε cDNA probe, David Gonzalez for oligonucleotides, 32 Lisitsyn,N. and Wigler,M. (1993) Science, 259, 946–951. Dong-Er Zhang, Liqin Liu, Milton Datta and Hanna Radomska for 33 Hubank,M. and Schatz,D.G. (1994) Nucleic Acids Res., 22, 5640–5648. 34 Braun,B.S., Frieden,R., Lessnick,S.L., May,W.A. and Denny,C.T. (1995) useful discussions. Supported by National Institutes of Health Grants Mol. Cell. Biol., 15, 4623–4630. CA41456, DK48660 and AI30656, and Japan research foundation 35 Lerner,A., Clayton,L.K., Mizoguchi,E., Ghendler,Y., van Ewijk,W., for clinical pharmacology and a long-term fellowship from the Koyasu,S., Bhan,A.K. and Reinherz,E.L. (1996) EMBO J., 15, 5876–5887. Human Frontier Science Program to A.I. 36 Zheng,W. and Flavell,R.A. (1997) Cell, 89, 587–596. 37 Tsai,S., Bartelmez,S., Sitnicka,E. and Collins,S. (1994) Genes Dev., 8, 2831–2841. 38 Chirgwin,J.M., Przybyla,A.E., MacDonald,R.J. and Rutter,W.J. (1979) REFERENCES Biochemistry, 18, 5294–5299. 39 Rosenberg,H.F., Ackerman,S.J. and Tenen,D.G. (1990) Biotechniques, 8, 384. 1 Shivdasani,R.A. and Orkin,S.H. (1996) Blood, 87, 4025–4039. 40 Yamanaka,R., Kim,G.-D., Radomska,H.S., Lekstrom-Himes,J., Smith,L.T., 2 Tenen,D.G., Hromas,R., Licht,J.D. and Zhang,D.-E. (1997) Blood, 90, Antonson,P., Tenen,D.G. and Xanthopoulos,K.G. (1997) Proc. Natl. Acad. 489–519. Sci. USA, 94, 6462–6467. 3 Zhang,D.-E., Zhang,P., Wang,N.D., Hetherington,C.J., Darlington,G.J. and 41 Yamanaka,R., Barlow,C., Lekstrom-Himes,J., Castilia,L.H., Liu,P.P., Tenen,D.G. (1997) Proc. Natl. Acad. Sci. USA, 94, 569–574. Eckhaus,M., Decker,T., Wynshaw-Boris,A. and Xanthopoulos,K.G. (1997) 4 Scott,E.W., Simon,M.C., Anastai,J. and Singh,H. (1994) Science, 265, 1573–1577. Proc. Natl. Acad. Sci. USA, 94, 13187–13192. 5 McKercher,S.R., Torbett,B.E., Anderson,K.L., Henkel,G.W., Vestal,D.J., 42 Chen,H.M., Gonzalez,D.A., Radomska,H.S., Voso,M.T., Sun,Z., Zhang,P., Baribault,H., Klemsz,M., Feeney,A.J., Wu,G.E., Paige,C.J. and Maki,R.A. Zhang,D.-E. and Tenen,D.G. (1998) Gene, 207, 209–218. (1996) EMBO J., 15, 5647–5658. 43 Borregaard,N. and Cowland,J.B. (1997) Blood, 89, 3503–3521. 6 Klemsz,M.J., McKercher,S.R., Celada,A., Van Beveren,C. and Maki,R.A. 44 Gruart,V., Truong,M.J., Plumas,J., Zandecki,M., Kusnierz,J.P., Prin,L., (1990) Cell, 61, 113–124. Vinatier,D., Capron,A. and Capron,M. (1992) Blood, 79, 2592–2597. 7 Hromas,R., Orazi,A., Neiman,R.S., Maki,R., Van Beveren,C., Moore,J. 45 Cross,M., Mangelsdorf,I., Wedel,A. and Renkawitz,R. (1988) and Klemsz,M. (1993) Blood, 82, 2998–3004. Proc. Natl. Acad. Sci. USA, 85, 6232–6236. 8 Chen,H.M., Zhang,P., Voso,M.T., Hohaus,S., Gonzalez,D.A., Glass,C.K., 46 Antonson,P. and Xanthopoulos,K.G. (1995) Biochem. Biophys. Res. Zhang,D.-E. and Tenen,D.G. (1995) Blood, 85, 2918–2928. Commun., 215, 106–113. 9 Voso,M.T., Burn,T.C., Wulf,G., Lim,B., Leone,G. and Tenen,D.G. (1994) 47 Smas,C.M. and Sul,H.S. (1993) Cell, 73, 725–734. Proc. Natl. Acad. Sci. USA, 91, 7932–7936. 48 Berliner,N., Hsing,A., Graubert,T., Sigurdsson,F., Zain,M., Bruno,E. and 10 Pahl,H.L., Scheibe,R.J., Zhang,D.-E., Chen,H.M., Galson,D.L., Maki,R.A. Hoffman,R. (1995) Blood, 85, 799–803. and Tenen,D.G. (1993) J. Biol. Chem., 268, 5014–5020. 49 Strobl,H., Takimoto,M., Majdic,O., Fritsch,G., Scheinecker,C., Hocker,P. 11 Ford,A.M., Bennett,C.A., Healy,L.E., Towatari,M., Greaves,M.F. and and Knapp,W. (1993) Blood, 82, 2069–2078. Enver,T. (1996) Proc. Natl. Acad. Sci. USA, 93, 10838–10843. 50 Tobler,A., Miller,C.W., Johnson,K.R., Selsted,M.E., Rovera,G. and 12 Oelgeschlager,M., Nuchprayoon,I., Luscher,B. and Friedman,A.D. (1996) Koeffler,H.P. (1988) J. Cell Physiol., 136, 215–225. Mol. Cell. Biol., 16, 4717–4725. 51 Rosmarin,A.G., Weil,S.C., Rosner,G.L., Griffin,J.D., Arnaout,M.A. and 13 Sturrock,A., Franklin,K.F. and Hoidal,J.R. (1996) J. Biol. Chem., 271, Tenen,D.G. (1989) Blood, 73, 131–136. 32392–32402. 52 Fouret,P., du Bois,R.M., Bernaudin,J.F., Takahashi,H., Ferrans,V.J. and 14 Hohaus,S., Petrovick,M.S., Voso,M.T., Sun,Z., Zhang,D.-E. and Crystal,R.G. (1989) J. Exp. Med., 169, 833–845. Tenen,D.G. (1995) Mol. Cell. Biol., 15, 5830–5845. 53 Khanna-Gupta,A., Lawson,N., Zibello,T. and Berliner,N. (1996) Blood, 88, 15 Smith,L.T., Hohaus,S., Gonzalez,D.A., Dziennis,S.E. and Tenen,D.G. 630a. (1996) Blood, 88, 1234–1247. 54 Khanna-Gupta,A., Zibello,T., Kolla,S., Neufeld,E.J. and Berliner,N. (1997) 16 Zhang,D.-E., Hetherington,C.J., Chen,H.M. and Tenen,D.G. (1994) Blood, 90, 2784–2795. Mol. Cell. Biol., 14, 373–381. 55 Ahne,B. and Stratling,W.H. (1994) J. Biol. Chem., 269, 17794–17801. 17 Olson,M.C., Scott,E.W., Hack,A.A., Su,G.H., Tenen,D.G., Singh,H. and 56 Goethe,R. and Loc,P.V. (1994) J. Biol. Chem., 269, 31302–31309. Simon,M.C. (1995) Immunity, 3, 703–714. 57 Singh,G., Katyal,S.L., Brown,W.E., Collins,D.L. and Mason,R.J. (1988) 18 Henkel,G.W., McKercher,S.R., Yamamoto,H., Anderson,K.L., Am. Rev. Respir. Dis., 138, 1261. Oshima,R.G. and Maki,R.A. (1996) Blood, 88, 2917–2926. 58 Li,F., Rosenberg,E., Smith,C.I., Notarfrancesco,K., Reisher,S.R., 19 Scott,E.W., Fisher,R.C., Olson,M.C., Kehrli,E.W., Simon,M.C. and Shuman,H. and Feinstein,S.I. (1995) Am. J. Physiol., 269, L241–L247. Singh,H. (1997) Immunity, 6, 437–447. 59 Rothwell,V.M. and Rohrschneider,L.R. (1987) Oncogene Res., 1, 311–324. 20 Tondravi,M.M., McKercher,S.R., Anderson,K., Erdmann,J.M., Quiroz,M., 60 Fukunaga,R., Ishizaka-Ikeda,E., Seto,Y. and Nagata,S. (1990) Cell, 61, Maki,R. and Teitelbaum,S.L. (1997) Nature, 386, 81–84. 341–350. 21 Johnson,P.F., Landschulz,W.H., Graves,B.J. and McKnight,S.L. (1987) 61 Liboi,E., Jubinsky,P., Andrews,N.C., Nathan,D.G. and Mathey-Prevot,B. Genes Dev., 1, 133–146. 22 Costa,R.H., Grayson,D.R., Xanthopoulos,K.G. and Darnell,J.E.,Jr (1988) (1992) Blood, 80, 1183–1189. Proc. Natl. Acad. Sci. USA, 85, 3840–3844. 62 D’Andrea,A.D., Lodish,H.F. and Wong,G.G. (1989) Cell, 57, 277–285. http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Nucleic Acids Research Oxford University Press

Use of RDA analysis of knockout mice to identify myeloid genes regulated in vivo by PU.1 and C/EBPα

Download PDF
10 pages

Loading next page...
 
/lp/oxford-university-press/use-of-rda-analysis-of-knockout-mice-to-identify-myeloid-genes-1tPIr52b5S

References (53)

Publisher
Oxford University Press
Copyright
© 1998 Oxford University Press
ISSN
0305-1048
eISSN
1362-4962
DOI
10.1093/nar/26.12.3034
Publisher site
See Article on Publisher Site

Abstract

3034–3043 Nucleic Acids Research, 1998, Vol. 26, No. 12  1998 Oxford University Press Use of RDA analysis of knockout mice to identify myeloid genes regulated in vivo by PU.1 and C/EBPα 1 2 2 Atsushi Iwama, Pu Zhang, Gretchen J. Darlington , Scott R. McKercher , Richard Maki and Daniel G. Tenen* Hematology/Oncology Division and Department of Medicine, Beth Israel Deaconess Medical Center and Harvard Medical School, Boston, MA 02215, USA, Departments of Pathology and Molecular and Human Genetics, Baylor College of Medicine, Houston, TX 77030, USA and The Burnham Institute, La Jolla, CA 92037, USA Received January 13, 1998; Revised and Accepted April 16, 1998 DDBJ/EMBL/GenBank accession nos AA720492–AA720501 ABSTRACT retinoic acid receptor α (RARα), promyelocytic leukemia zinc finger (PLZF), myeloid zinc finger protein-1 (MZF-1), early PU.1 and C/EBPα are transcription factors essential for response gene-1 (Egr-1), Wilms’ tumor suppressor gene (WT-1) normal myeloid development. Loss-of-function mutation and homeobox proteins (2), among which PU.1 and C/EBPα of PU.1 leads to an absolute block in monocyte/ have been shown to be indispensable for myeloid development by macrophage development and abnormal granulocytic gene targeting experiments (3–5). development while that of C/EBPα causes a selective PU.1 is a member of the Ets transcription family and is block in neutrophilic differentiation. In order to under- predominantly expressed in hematopoietic cells (6). PU.1 mRNA stand these phenotypes, we studied the role of PU.1 is expressed at low levels in multipotential CD34 cells, and is and C/EBPα in the regulation of myeloid target genes upregulated with myeloid and B cell differentiation (6–9). in vivo. Northern blot analysis revealed that mRNAs Transient transfection studies have shown that PU.1 regulates the encoding receptors for M-CSF, G-CSF and GM-CSF, promoters of a number of myeloid genes, such as CD11b, primary –/– were expressed at low levels in PU.1 fetal liver granule proteins (myeloperoxidase, neutrophilic elastase and compared with wild type. To identify additional myeloid proteinase-3), GM-CSF receptor, G-CSF receptor and M-CSF genes regulated by PU.1 and C/EBPα, we performed receptor (10–16). Several different loss-of-function experiments representational difference analysis (RDA), a PCR-based revealed that PU.1 is involved in myeloid and lymphoid subtractive hybridization using fetal livers from wild development (4,5,9,17–19). PU.1 knockout mice completely type and PU.1 or C/EBPα knockout mice. By introducing lack macrophages including osteoclasts, as well as B cells, and a new modification of RDA, that of tissue-specific gene show impaired granulopoiesis and T-cell development (4,5,20). suppression, we could selectively identify a set of –/– However, fetal liver cells from PU.1 mice do express mRNA differentially expressed genes specific to myeloid cells. for early myeloid genes (17). Differentially expressed genes included both primary C/EBPα is a member of the C/EBP family, which has a bZIP and secondary granule protein genes. In addition, eight structure. C/EBPα was originally characterized in liver and adipose novel genes were identified that were upregulated in tissues, and has been shown to regulate a number of hepatic and expression during myeloid differentiation. These adipocyte genes (21–23). Recently, the expression of C/EBPα was methods provide a general strategy for elucidating the shown to initiate with the commitment of multipotential precursors genes affected in murine knockout models. to the myeloid lineage, and be specifically upregulated during granulocytic differentiation (2,24,25). Transient transfection INTRODUCTION studies have shown that C/EBPα can regulate the promoters of a number of myeloid specific genes, such as G-CSF receptor (15), Transcription factors play a major role in cell differentiation, neutrophil elastase (12) and myeloperoxidase (11). C/EBPα including the development of specific hematopoietic lineages knockout mice die within 8 h of birth because they are unable to from stem cells (1,2). Mature myeloid cells, consisting of blood properly synthesize and mobilize glycogen and fat (26,27). They monocytes and tissue macrophages, as well as the neutrophilic also show a selective block in differentiation of neutrophils. and eosinophilic granulocytes, develop from a common myeloid Mature neutrophils and eosinophils are not observed in the blood precursor. However, the mechanism controlling the development or fetal liver of mutant mice and, instead, myeloid blasts are of common myeloid precursors as well as the transition from common precursors into unipotential granulocyte and monocyte observed. Like the PU.1 knockout animals, fetal liver cells from –/– clearly express mRNA for some myeloid genes (3). precursors has not been fully identified. So far several transcrip- C/EBPα tion factors has been implicated in myelopoiesis. These include Other hematopoietic lineages are not affected, including monocytes PU.1, basic leucine zipper (bZIP) CCAAT/enhancer binding and macrophages (3). These results strongly suggest a critical role protein family (C/EBP), acute myelogenous leukemia 1 (AML1), for C/EBPα in granulocytic differentiation. *To whom correspondence should be addressed at: Harvard Institutes of Medicine, Room 954, 77 Avenue Louis Pasteur, Boston, MA 02115, USA. Tel: +1 617 667 5561; Fax: +1 617 667 3299; Email: [email protected] 3035 Nucleic Acids Research, 1998, Vol. 26, No. 12 3035 Nucleic Acids Research, 1994, Vol. 22, No. 1 To understand the impaired myeloid development caused by England Biolabs, Beverly, MA). Fifteen micrograms of each loss-of-function mutation of PU.1 and C/EBPα, myeloid colony- RNA sample were resolved by agarose formaldehyde gel stimulating factor (CSF) receptors have been suggested as critical electrophoresis and transferred to Biotrans nylon membranes targets for PU.1 and C/EBPα (3,17,18). However, the loss-of- (ICN, Biomedicals, Inc., Costa Mesa, CA). The blots were function mutant mice of each CSF or CSF receptor do not show hybridized to [α- P]dCTP-labeled DNA fragments generated by defects in myeloid development as severe as those of PU.1 or RDA as described previously (8) and exposed for 1–2 days with C/EBPα knockout mice (28–31). It is likely that there exist an intensifying screen. To normalize the loading of RNA samples additional critical targets. In order to identify these target genes in each lane, the probe was removed and the blot was rehybridized for PU.1 and C/EBPα during myeloid development, we have to an [α- P]dCTP 3′-end-labeled 18S oligonucleotide (39). analyzed expression of presumptive myeloid target genes in vivo as well as performed representational difference analysis (RDA), cDNA synthesis a PCR-based subtractive hybridization using wild-type and Oligo(dT)-primed double-stranded cDNA was synthesized from knockout mice. In contrast with differential display, which 5–10 μg of poly(A) RNA using a cDNA synthesis system amplifies fragments from all represented mRNA species, RDA (GIBCO BRL, Grand Island, NY) according to the manufacturer’s eliminates those fragments present in both populations, leaving instructions. Tester and driver cDNA samples were synthesized only the difference (32,33). Recently, several groups successfully in parallel at the same time. identified differentially expressed genes using RDA (33–36). In this study, to focus on the differentially expressed genes of myeloid Representational difference analysis lineage, we tried several new modifications of the RDA procedure. Here we show the in vivo role of PU.1 and C/EBPα in RDA was performed using reagents as described (32–34). The regulating myeloid genes by northern blot analysis, and by using following oligonucleotides were synthesized and used for RDA: RDA combined with specific gene suppression, we identified a R-Bgl-24, 5′-AGCACTCTCCAGCCTCTCACCGCA-3′; R-Bgl- set of myeloid genes, the expression of which is missing or 12, 5′-GATCTGCGGTGA-3′; J-Bgl-24, 5′-ACCGACGTCGA- significantly decreased in the mutant fetal liver. These genes CTATCCATGAACA-3′; J-Bgl-12, 5′-GATCTGTTCATG-3′; included myeloid granule protein genes and eight novel myeloid N-Bgl-24, 5′-AGGCAACTGTGCTATCCGAGGGAA-3′; and genes which are new candidate targets for PU.1 and C/EBPα N-Bgl-12, 5′-GATCTTCCCTCG-3′. cDNA was digested with transcription factors. DpnII and ligated to the R-Bgl-12/24 adaptors. Amplicons were made by PCR amplification of the ligated DpnII cDNA fragments for 20 cycles using the R-Bgl-24 as a primer. Driver DNA was MATERIALS AND METHODS prepared by digesting amplicons with DpnII. Tester DNA was Mice, tissues and cells prepared by gel-purification of digested amplicons between 150 and 2000 bp followed by ligation to J-Bgl-12/24 adaptors. First Targeted disruption of C/EBPα and PU.1 was achieved by subtractive hybridization was performed using 400 ng tester and homologous recombination in embryonic stem cells and generation 40 μg driver (tester:driver = 1:100). An aliquot of the hybridization of mice from these cell lines as reported previously (5,26). Fetal mixture was amplified by PCR for 10 cycles using the J-Bgl-24 liver was obtained from embryonic day 19 fetuses. Purification of as a primer. The PCR products were then digested with mung fetal liver hematopoietic cells was performed by passing fetal bean nuclease (New England Biolabs) at 30C for 35 min and liver through 70 μm nylon mesh cell strainers (Becton Dickinson further amplified for 18 cycles. These PCR products are the first Labware, Franklin Lakes, NJ). Peritoneal exudate cells were difference products (DP1). The difference products were digested harvested by lavage with 10 ml PBS 20, 48 and 72 h after i.p. with DpnII and ligated to a new adaptor, N-Bgl-12/24 (after the injection of 1.5 ml 10% thioglycollate broth. Morphological first hybridization) or J-Bgl-12/24 (after the second hybridization), examination revealed that the cells consisted of ~ 15% monocytes/ and the procedure was repeated twice using tester:driver ratios of macrophages and 80% polymorphonuclear granulocytes 20 h after 1:800 and 1:4000–400 000 for the second and third round of injection, and 50 versus 45%, and 80 versus 15% at 48 and 72 h, hybridization, respectively. respectively. In some experiments (see Results), previously generated The murine lymphohematopoietic progenitor cell line EML cDNA fragments were suppressed by adding 150–300 ng of each was maintained in IMDM supplemented with 20% horse serum, DNA fragment without adaptors to each round of hybridization. glutamine, non-essential amino acids and 10% conditioned The suppression of liver genes was performed by adding 40 μg medium from BHK cells transfected with rat stem cell factor driver prepared from adult liver to each round of hybridization. cDNA (BHK-MKL cells) (37). To induce myeloid differentiation, The suppression of mature myeloid genes was performed by EML cells were cultured in IMDM supplemented with 10% adding 30 μg driver prepared from peritoneal exudate cells BHK-MKL conditioned medium, 5% WEHI-3 conditioned collected 20 or 72 h after i.p. injection of thioglycollate. –5 medium and 10 M all-trans retinoic acid (RA) for 72 h. Cells DNA sequences of novel cDNA clones identified by RDA have were then washed three times to remove RA and recultured in been submitted to the GenBank database (accession nos IMDM supplemented with murine GM-CSF (2.5 ng/ml) for AA720492–AA720501). indicated times (37). RESULTS RNA preparation and northern blot analysis Expression of transcription factors in mutant mice Total RNA was isolated by guanidium isothiocyanate extraction followed by CsCl gradient purification (38). Poly(A) RNA was To check the involvement of PU.1 or C/EBPα in the regulation purified from total RNA with oligo(dT) cellulose columns (New of transcription factors likely to play critical roles in myeloid 3036 Nucleic Acids Research, 1998, Vol. 26, No. 12 Figure 1. Northern blot analysis of mRNA of transcription factor genes. Total –/– –/– RNA (15 μg) from wild-type (lanes 1 and 3), C/EBPα (lane 2) and PU.1 (lane 4) day 19 fetal liver was analyzed using species- and gene-specific 5′ Figure 2. Northern blot analysis of mRNA of growth factor receptors in PU.1 cDNA probes of PU.1 and Spi-B (8), a 3′ cDNA probe of C/EBPα (26) and knockout mice. Total RNA (20 μg) from wild-type (lane 1), PU.1 +/– (lane 2), exon 3 of murine C/EBPε (41). –/– or PU.1 (lane 3) day 19 fetal liver was electrophoresed in 1% agarose/ formaldehyde gels, transferred to a nylon membrane and probed with murine cDNAs corresponding to M-CSF receptor (M-CSFr) (59); G-CSF receptor (G-CSFr) (60); GM-CSF receptor α (GM-CSFr) (61); erythropoietin receptor development, we analyzed the expression of PU.1, Spi-B, (EPOr) (62); and 18S oligonucleotide (39). C/EBPα and C/EBPε (40,41), all of which have been shown or postulated to play a role in myeloid development (2), in the mutant fetal livers (Fig. 1). PU.1 knockout mice were made by with previous reports indicating the presence of early myeloid disrupting the DNA binding domain, inserting the neo gene –/– gene expression in PU.1 fetal liver cells (17). GM-CSF within exon 5 (5), but no PU.1-related transcripts were detected –/– –/– receptor mRNA was not detectable in PU.1 fetal liver by in PU.1 fetal liver by a probe specific to the 5′ end of the PU.1 northern blot analysis, although it has been detected by RT–PCR cDNA (8), confirming that this is indeed a null phenotype. PU.1 –/– (17). These findings are consistent with promoter studies expression was decreased ~ 50% in C/EBPα fetal liver, demonstrating a functional PU.1 site in transient transfection consistent with the presence of a significant number of immature analysis (14–16). The levels of erythropoietin receptor mRNA myeloid cells in these animals (3). Spi-B is an Ets transcription –/– were not affected in PU.1 fetal livers compared with wild type, factor closely related to PU.1 but now known to be expressed in keeping with the lack of a consistent effect of PU.1 disruption primarily in B cells (8,42). Spi-B expression was slightly –/– on erythropoiesis (4,5). Although PU.1 and C/EBPα knockout decreased in C/EBPα fetal liver, but was undetectable in –/– mice have no detectable, or very low, levels of CSF receptor PU.1 fetal liver. C/EBPα expression was not affected in –/– expression, their defects in myeloid development are more severe PU.1 fetal liver, but it is very hard to detect a difference in than those of loss-of-function mutant mice of CSFs or CSF hematopoietic cells because of the high C/EBPα expression by receptors (28–31). This suggests that there are additional genes hepatocytes and adipocytes. Another myeloid specific C/EBP regulated by PU.1 and C/EBPα whose altered expression lead to the transcription factor, C/EBPε, which is critical for terminal observed phenotype. Therefore, we performed RDA to identify myeloid maturation (41), was not expressed in both mutant fetal these additional genes regulated by these transcription factors. livers. These findings are consistent with the lineage-specific expression of Spi-B and C/EBPε in B cell and granulocyte lineages, respectively, and suggest that they might be regulated by Identification of differentially expressed genes between PU.1 and/or C/EBPα. Alternatively, the knockout cells might be C/EBPα +/+ and –/– fetal livers by RDA blocked in their differentiation and do not become mature enough As noted above, C/EBPα-deficient mice show a selective block to express Spi-B or C/EBPε. in differentiation of neutrophils. Other hematopoietic lineages, including monocytes, are not affected (3). To identify C/EBP- mRNA expression of myeloid CSF receptors in mutant mice α-regulated genes during neutrophilic differentiation, we performed We have previously shown that G-CSF receptor mRNA is RDA using C/EBPα-deficient mice. The cDNA in which the selectively downregulated in C/EBPα knockout mice by northern differentially expressed cDNAs are to be found is called ‘tester’ blot analysis, whereas M-CSF receptor and GM-CSF receptor cDNA, and the reference cDNA is called ‘driver’ cDNA. The mRNA levels are not impaired (3). This suggests that impaired cDNA from each population is digested with a restriction G-CSF signaling might be in part responsible for the selective endonuclease, ligated to adaptors, and then amplified by PCR. block of neutrophilic differentiation. To determine the role of The products of amplification are called amplicons. To isolate –/– myeloid CSF receptors in the defect found in PU.1 mice, we amplicons unique to the tester cDNA, tester amplicon was ligated –/– analyzed their expression in PU.1 fetal liver by northern blot to new adaptors and hybridized to an excess of driver amplicon. analysis (Fig. 2). The expression of M-CSF receptor and G-CSF PCR with primers for the new adaptors preferentially amplifies receptor mRNA was markedly decreased but detectable, consistent tester–tester homoduplexes. This process is repeated several 3037 Nucleic Acids Research, 1998, Vol. 26, No. 12 3037 Nucleic Acids Research, 1994, Vol. 22, No. 1 which has been previously characterized in transient transfection AB studies as one of the targets for C/EBPα (12). DP2 (1:800) contained more genes than DP3, and most of them were distinct from DP3 except for contrapsin and protein C. Interestingly, DP2 contained three additional differentially expressed genes, but the expression of the remaining seven genes showed no difference between C/EBPα +/+ and –/– fetal liver or was unexpectedly –/– upregulated in C/EBPα fetal liver (Table 1). These results indicate that although RDA could selectively amplify differentially expressed genes, to keep high specificity, it requires a high stringency, which causes a limit on the number of genes amplified. Table 1. DNA fragments generated by RDA, C/EBPα +/+ minus –/– with different stringencies Figure 3. (A) Alkaline agarose gel analysis of second strand cDNA synthesis. Gene (accession no.) Expression α- P-labeled second strand cDNAs made from wild-type (lane 1) or E19 fetal liver BM Adult –/– C/EBPα (lane 2) day 19 fetal liver mRNA were electrophoresed on a 1.4% liver li C/EBPα +/+ –/– alkaline agarose gel, and the dried gel was exposed to X-ray film. (B) Agarose electrophoresis of difference products generated by RDA. Amplicon from DP3 (tester:driver = 1:400 000 –/– wild-type day 19 fetal liver cDNA (lane 2), amplicon from C/EBPα day 19 fetal liver cDNA (lane 3), first difference product (DP1) with tester to driver C/EBPα (M62362) + – + + ratio of 1:100 (lane 4), DP2 with tester to driver ratio of 1:800 (lane 5), DP3 with Neutrophilic elastase (U04962) 2+ – 3+ – tester to driver ratio of 1:4000 (lane 6) and 1:400 000 (lane 7) and DNA Contrapsin (X55147) 2+ – – 3+ molecular markers (lanes 1 and 8) were electrophoresed on a 2.0% agarose gel. Protein C (D10445) 5+ + – 6+ Unknown 1 + – – + DP2 (tester:driver = 1:800) times with increasing driver-to-tester ratios until only fragments Contrapsin (X55147) unique to the tester remain (32,33). Protein C (D10445) Since C/EBPα mutant mice die soon after birth, we used day Haptoglobin (S67972) 2+ – + 2+ 19 fetal liver as material for RDA. Because C/EBPα also regulates transcription of hepatocyte- and adipocyte-specific Apolipoprotein A-I (L04151) 4+ + – 3+ genes, these genes as well as myeloid-specific genes were Eosinophil chemotactic factor candidates for identification from fetal liver by RDA. Poly(A) (X15313) + – 3+ – RNA was purified from fetal livers of day 19 embryos and Pref-1 (L12721) + 5+ – – double-stranded cDNA was synthesized using an oligo(dT) Unknown 2 + + – + primer. After ligating adaptors to DpnII-digested cDNAs, the Unknown 3 + + – + tester and driver amplicons were generated by PCR amplification. Unknown 4 + + – 2+ It is important that preparations to be subtracted are as similar as Unknown 5 + + – + possible as shown in Figure 3. If the quality of cDNA and Unknown 6 2+ 2+ – 3+ amplicon varies between subtracted populations, this can result in amplification of false positives. The tester was subtracted with the Relative expression was evaluated by northern blot analysis. Evaluation of mRNA driver, and the difference was selectively amplified by PCR. This levels are consistent for each gene but cannot be compared among genes. process was repeated three times until the difference products Genes differentially expressed. (DPs) showed clear bands with little background visible by ethidium bromide staining (Fig. 3B). The third subtraction was performed with different hybridization ratios. The third difference Suppression of liver genes during RDA subtractive product (DP3) with a higher stringency (tester:driver = 1:400 000) hybridization showed fewer bands than that with lower stringency (1:4000), –/– indicating that more genes were suppressed in the third subtraction As shown in Table 1, RDA using C/EBPα fetal liver amplified by increasing ratios of tester to driver (Fig. 3B). After digesting liver genes more than myeloid genes, possibly because day 19 fetal with DpnII, DP2 (1:800) and DP3 (1:400 000) were separated on liver contains more liver RNA than myeloid RNA. To preferentially an agarose gel and each band was excised out and subcloned. The amplify myeloid genes, we modified the RDA technique. First of subcloned inserts were used in northern blot analysis as probes to all, we enriched hematopoietic cells by using cell strainers, nylon check mRNA expression of identified genes. mesh devices with 70 μm pore size which select for cell size. We Table 1 shows the profile of genes identified by RDA in this could enrich hematopoietic cells ~ 3–4-fold by passing fetal livers screening. DP3 with a higher stringency (1:400 000) contained through cell strainers (data not shown). We then prepared C/EBPα, demonstrating that the RDA procedure was selecting liver-specific DNA fragments generated in the previous RDA differentially expressed genes. The northern blot analysis revealed (Table 2) and driver from adult liver, and added either of them in that all genes were truly differentially expressed, i.e. expressed in the hybridization mixture to suppress the amplification of +/+ –/– the C/EBPα fetal liver, but not or at lower levels in C/EBPα liver-specific genes. We used a high stringency of 1:400 000 for fetal liver (Table 1). As expected, DP3 contained both myeloid- the third round of subtractive hybridization. The DP3 showed specific and liver-specific genes, including neutrophilic elastase, several clear bands on an agarose gel, and the profile of bands was 3038 Nucleic Acids Research, 1998, Vol. 26, No. 12 AB C –/– Figure 4. Northern blot analysis of mRNA of novel genes generated by RDA. (A) Total RNA (15 μg) from wild-type (lane 1) and C/EBPα (lane 2) day 19 fetal liver, bone marrow (lane 3), peritoneal exudate cells 48 h after thioglycollate stimulation (lane 4), spleen (lane 5), thymus (lane 6) and adult liver (lane 7). (B) Total –/– RNA (15 μg) from wild-type (lanes 1 and 3) and PU.1 (lanes 2 and 4) day 19 fetal liver treated (lanes 1 and 2) and untreated (lanes 3 and 4) with cell strainers, bone marrow (lane 5), peritoneal exudate cells 48 h after thioglycollate stimulation (lane 6), spleen (lane 7), thymus (lane 8) and adult liver (lane 9). (C) Poly(A) RNA –/– (3 μg) from wild-type (lane 1) and C/EBPα (lane 2) day 19 fetal liver, bone marrow (lane 3), peritoneal exudate cells 20 h (lane 4) and 72 h (lane 5) after thioglycollate stimulation, spleen (lane 6), thymus (lane 7) and adult liver (lane 8). DNA fragments generated by RDA were used as probes. very similar with each type of suppression. Nucleotide sequence monocytes/macrophages and lymphoid cells in the blood or fetal analysis of DP3 revealed that the suppression of liver-specific genes liver, and die from septicemia within 2 days of birth. However, worked very well (Table 2); liver genes were dramatically antibiotic-treated mice could survive for 2 weeks and show the suppressed during the subtractive hybridization. DP3 with specific development of normal appearing T cells and a few cells with the suppression by liver-specific DNA fragments contained only two characteristics of neutrophils (5). To identify PU.1-regulated liver-specific genes, and DP3 with suppression by the driver genes during myeloid differentiation, we performed RDA using prepared from adult liver contained no liver-specific genes. The day 19 fetal liver of PU.1-deficient mice. We prepared cDNAs successful suppression of liver genes led to the amplification of more from whole fetal liver and enriched hematopoietic cells by a cell myeloid genes. Many genes for primary and secondary granule size selection using cell strainers, and compared the profile of proteins of neutrophils were identified, including myeloperoxidase amplified genes. Because PU.1 is not expressed in hepatocytes, and neutrophilic elastase, targets for C/EBPα identified by only myeloid and lymphoid genes were expected to be identified transient transfection assays (11,12) (Table 2). In addition to from the fetal liver by RDA. known myeloid genes, five novel genes were identified (C/Edp Table 3 shows the profile of genes identified by RDA. We used 1–5, Table 2). C/Edp 2 and 3 showed high nucleotide sequence a high stringency of 1:400 000 for the third round of subtractive similarity to human neutrophil collagenase (79%) and human hybridization. As expected, most of the genes contained in DP3 ficolin (72%), respectively, suggesting that C/Edp 2 and 3 are were myeloid-specific genes, including those specific to the putative murine homologues of these genes. The other genes neutrophil and/or monocyte/macrophage lineage, and others showed no significant similarity to any known genes. Northern were lymphoid genes. There was no significant difference in the blots revealed that all of these unknown genes were differentially profile between DP3 from whole fetal liver and purified expressed and were preferentially expressed in the BM and/or hematopoietic cells, suggesting that this procedure is very peritoneal exudate cells consisting of mature neutrophils and sensitive, but we could amplify different genes by using different monocytes (Fig. 4A). These results demonstrate that liver genes materials. We identified five unknown genes (Pdp 1 and Pdp could be successfully suppressed by using specific gene fragments 3–6). They showed no significant similarity to any known genes. or adult liver driver, and this suppression facilitates the amplification Interestingly, two of them were the same genes as those identified of myeloid-specific genes. by RDA using C/EBPα knockout mice. Northern blots revealed that all of these unknown genes were differentially expressed and Identification of differentially expressed genes between also preferentially expressed in the BM and/or peritoneal exudate PU.1 +/+ and –/– fetal livers by RDA cells (Fig. 4B). Therefore, RDA is sensitive and specific enough PU.1-deficient mice show impaired myeloid and lymphoid to identify the difference in a small subpopulation from materials development (4,5). The mutant mice lack mature neutrophils, comprised of heterogenous cell populations. 3039 Nucleic Acids Research, 1998, Vol. 26, No. 12 3039 Nucleic Acids Research, 1994, Vol. 22, No. 1 Table 2. DNA fragments generated by RDA, C/EBPα +/+ minus –/– with the immature myeloid cells, i.e. from myeloblasts to band cells, but suppression of liver-specific genes not in mature myeloid cells including peritoneal exudate cells (reviewed in 43). Liver genes were suppressed by adult liver Gene (accession no.) driver as described above. We used lower stringencies, 1:40 000 for the third round of subtractive hybridization, because the expression Suppression with liver gene fragments ( ) of differentially expressed genes specific to immature myeloid Lactoferrin (D88510) cells was expected to be weaker than before. As shown in Table 4, Myeloid bectenecin (U95002) we obtained quite a different profile of genes. Although we failed Lipocortin I (M24554) to suppress gelatinase B and Pdp 4 with specific DNA fragments, Neutrophil gelatinase associated lipocalin (X81627) suppression with mature myeloid cDNA and other specific DNA Eosinophil chemotactic factor (X15313) fragments worked well. We isolated a new DpnII fragment of Neutrophilic elastase (U04962) MPO, eosinophil peroxidase (EPO), proteinase-3 and gelatinase C/Edp 4 (Pdp 3) B. They are myeloid granule proteins and are expressed in immature myeloid cells (43,44). Others were a Kupffer cell-specific C/Edp 5 (Pdp 4) b gene, a B-cell gene and three unknown genes. The unknown Haptoglobin (M96827) genes showed no significant similarity to any known genes. C4 complement protein (M11789) Northern blots revealed that C/Edp 6 and 7 were differentially Suppression with adult liver cDNA expressed (Fig. 4C), but unknown 7 was not (data not shown). Lactoferrin (D88510) Because the expression of C/Edp 6 and 7 was relatively weak, Myeloid bectenecin (U95002) poly(A) RNA northerns were required for identification. C/Edp Lipocortin I (M24554) 6, 7 and Pdp 4 were not expressed in peritoneal exudate cells at Neutrophil gelatinase associated lipocalin (X81627) all (Fig. 4B and C). These results demonstrate that suppression of Eosinophil chemotactic factor (X15313) mature myeloid genes facilitates the amplification of genes preferentially expressed in immature myeloid cells. Myeloperoxidase (X15313) Gelatinase B (D12712) Stefin 1 (M92417) Table 3. DNA fragments generated by RDA, PU.1 +/+ minus –/– C/Edp 1 (AA720492) C/Edp 2 (AA720493) C/Edp 3 (AA720494) Gene (accession no.) C/Edp 4 (Pdp 3; AA720498) Whole fetal liver C/Edp 5 (Pdp 4; AA720499) Lysozyme M (M21050) Bacteria binding macrophage receptor (U18424) DNA fragments previously generated by RDA (contrapsin, protein C, haptoglobin, Complement subcomponent C1q α-chain (X58861) apolipoprotein A-I) were used for suppression. Lactoferrin (D88510) This haptoglobin fragment is a different one from that used for suppression. Eosinophil chemotactic factor (X15313) C/Edp, Pdp: Unknown differentially expressed gene isolated by RDA, C/EBP Ig λ-chain (M30387) +/+ minus –/– (C/Edp) and PU.1 +/+ minus –/– (Pdp). Pdp 1 (AA720497) Pdp 3 (C/Edp 4; AA720498) Pdp 4 (C/Edp 5; AA720499) Pdp 5 (AA720500) Suppression of mature myeloid genes leads to the amplification of immature myeloid-specific genes Pdp 6 (AA720501) Enriched fetal liver hematopoietic cells The absence of C/EBPα and PU.1 causes a block at an early stage Lysozyme M (M21050) of myeloid differentiation. The critical targets responsible for this Bacteria binding macrophage receptor (U18424) differentiation block are expected to be also expressed at early Complement subcomponent C1q α-chain (X58861) stage. To focus on the early targets during myeloid differentiation, Lactoferrin (D88510) we used fetal liver hematopoietic cells enriched by cell size Myeloperoxidase (X15313) selection, and performed suppression of mature myeloid genes. Myeloid bectenecin (U95002) We prepared driver amplicons from peritoneal exudate cells collected 20 and 72 h after i.p. injection of thioglycollate. The gp91phox (U43384) former cells consisted of ~ 80% of neutrophils, while the latter LAPTm 5 (U51239) consisted of 80% of monocytes/macrophages. Driver from MHC class II H2–1A-α (M11357) peritoneal exudate cells 20 h after stimulation was added to the Ig λ-chain (M30387) hybridization mixture of C/EBPα +/+ and –/– to suppress Pdp 1 (AA720497) neutrophilic genes, and both drivers were added to that of PU.1 Pdp 6 (AA720501) +/+ and –/– to suppress both neutrophilic and monocyte/macrophage genes. We also performed specific gene suppression using DNA fragments from our novel RDA clones and myeloid granule C/Edp, Pdp: unknown differentially expressed gene isolated by RDA, C/EBPα proteins (Table 4), because most of them are expressed only in +/+ minus –/– (C/Edp) and PU.1 +/+ minus –/– (Pdp). 3040 Nucleic Acids Research, 1998, Vol. 26, No. 12 Table 4. DNA fragments generated by RDA with the suppression of mature myeloid cDNAs C/EBPα +/+ minus –/– PU.1 +/+ minus –/– a a Myeloperoxidase (X15313) myeloperoxidase (X15313) Eosinophil peroxidase (D78353) gelatinase B (D12712) Proteinase-3 (U43525) Kupffer cell receptor (D88577) procathepsin E (X97399) C/Edp 6 (AA720495) Pdp 4 (AA720499) C/Edp 7 (AA720496) Unknown 7 Drivers prepared from peritoneal exudate cells and previously generated DNA fragments (C/Edp 1–3, Pdp 1–6, myeloperoxidase, neutrophilic elastase, lactoferrin, gelatinase B, myeloid bectenecin, neutrophil gelatinase associated lipocalin, lipocortin I and eosinophil chemotactic factor) were used for suppression. This MPO fragment is different from that in Tables 2 and 3. Expression of myeloid granule protein genes in mutant mice In this study, we identified many myeloid granule protein genes, including primary granule protein genes (myeloperoxidase, neutrophilic elastase and proteinase-3); secondary granule protein genes [lactoferrin, neutrophil gelatinase associated lipocalin (NGAL), putative murine homologue of neutrophilic collagenase C/Edp 2, gelatinase B and myeloid bectenecin]; and lysozyme M, Figure 5. Northern blot analysis of mRNA of myeloid granule protein genes. –/– –/– Total RNA (15 μg) from wild-type (lanes 1 and 3), C/EBPα (lane 2), PU.1 which is localized in both primary and secondary granules. –/– (lane 4) day 19 fetal liver, and wild-type (lanes 5 and 7), C/EBPα (lane 6) and Among them, myeloperoxidase and neutrophilic elastase have –/– PU.1 (lane 8) newborn lung. DNA fragments generated by RDA were used been characterized as common targets for PU.1 and C/EBPα by as probes. transient transfection assays (11,12). To determine the in vivo role of PU.1 and C/EBPα in the regulation of myeloid granule protein genes, we analyzed their expression by northern blotting. The generated by treatment with high concentrations of retinoic acid. expression of many of them was very low or undetectable in both These myeloid progenitors differentiate into neutrophils and mutant fetal livers in vivo (Fig. 5). Specifically, myeloperoxidase macrophages in response to GM-CSF, but still neutrophilic and proteinase-3 were expressed at very low levels in both mutant differentiation is blocked around the promyelocyte to myelocyte –/– fetal livers, NGAL and C/Edp2 at low levels in PU.1 fetal liver, stages and only few mature neutrophils could be observed. –/– and lysozyme M at low levels in C/EBPα fetal liver (Fig. 5). Differentiated neutrophilic cells appeared on day 3 after treatment Although mRNAs encoding myeloid granule proteins are confined with GM-CSF (blasts 14.3%, promyelocytes 53.0%, myelocytes/ to myeloid cells, lysozyme M is abundantly expressed also in metamyelocytes 24.5% and monocytes/macrophages 8.0%), non-hematopoietic tissues, particularly in the lung (45), in which reached a peak on day 6 (blasts 6.0%, promyelocytes 28.0%, C/EBPα is highly expressed (46). Interestingly, the expression of myelocytes/metamyelocytes 47.5%, band/segmented cells 8.5% –/– lysozyme was markedly downregulated in C/EBPα newborn and monocytes/macrophages 9.7%), and then decreased, while –/– lung but not in PU.1 newborn lung (Fig. 5). These findings macrophages gradually increased and dominated on day 10 suggest the critical role of PU.1 and C/EBPα in the regulation of (blasts 3.5%, promyelocytes 19.7%, myelocytes/metamyelocytes myeloid granule protein genes in vivo. 8.1% and monocytes/macrophages 69.8%). As shown in Figure 6, the expression of C/Edp 1–3 and Pdp 3 and 4 were strongly induced during myeloid differentiation, and downregulated on Myeloid-specific expression of novel genes day 10. The 2.0 kb transcript of Pdp 6 was weakly induced, while To clarify the lineage-specific expression of novel genes, we the 1.0 kb transcript, which is a minor transcript in peritoneal analyzed their expression in lymphohematopoietic tissues. Most exudate cells (Fig. 4B), was strongly upregulated. The expression of them were preferentially expressed in the bone marrow or of Pdp 5 was upregulated 3 days after GM-CSF stimulation and peritoneal exudate cells, but not in spleen, thymus or adult liver maintained during differentiation. The analysis of the expression (Fig. 4A and B), suggestive of their preferential expression in of these novel myeloid genes in other hematopoietic cell lines myeloid cells. revealed that they were expressed in myeloid cells but not in T We further analyzed their expression during the myeloid cells, B cells or erythroid cells (data not shown). Therefore, RDA differentiation of EML cells. EML is a stem cell factor-dependent selectively amplified differentially expressed genes which are lymphohematopoietic progenitor cell line immortalized by a preferentially expressed during myeloid differentiation. The retroviral vector harboring a dominant-negative retinoic acid expression of C/Edp 6 and 7 was not detected in EML cells or receptor (37). Myeloid differentiation is suppressed in EML cells, other hematopoietic cell lines (data not shown); therefore, they but common progenitors for neutrophils and macrophages are may be preferentially expressed in fetal liver. 3041 Nucleic Acids Research, 1998, Vol. 26, No. 12 3041 Nucleic Acids Research, 1994, Vol. 22, No. 1 amplifies only the difference. RDA is sensitive enough to isolate genes expressed in only a very small percentage of cells (32,33). Therefore, this technique was suitable for our cloning approach using fetal livers as materials in which myeloid cells compose only a small percentage of the total cell population. Our data demonstrated that this procedure truly amplified differentially expressed genes, and was able to amplify genes expressed at low levels as well. Most of the genes identified were expressed in the fetal liver at much lower levels than in the bone marrow or peritoneal exudate cells (Fig. 4). However, because so –/– many liver genes are differentially expressed in C/EBPα fetal liver, we amplified more liver genes than myeloid genes (Table 1). To suppress the amplification of liver genes, we first tried suppression by liver-specific DNA fragments. Suppression of expected difference products by specific DNA fragments has been reported to facilitate the amplification of new gene fragments (32,33). We prepared liver-specific DNA fragments generated in the previous RDA, and added these to the hybridization mixture. This suppression worked well, but other liver genes were still amplified (Table 2). To get complete suppression of liver genes and preferentially amplify myeloid genes, we prepared driver amplicon from adult liver and added it into each round of subtractive hybridization. As shown in Tables 2 and 4, liver genes were completely suppressed and this facilitated Figure 6. Northern blot analysis of mRNA of novel genes identified by RDA the amplification of myeloid genes. This modification was also during myeloid differentiation of EML cells. Total RNA (15 μg) from before successfully applied to suppression of mature myeloid genes to –5 stimulation (lane 1), after treatment with RA (10 M) and IL-3 for 3 days amplify immature myeloid genes (Table 4). Our data demonstrate (lane 2), after 1 day (lane 3), 3 days (lane 4), 6 days (lane 5) and 10 days (lane 6) of culture in the presence of GM-CSF. The same probes as in Figure 4 were used. that genes expressed in a certain cell population or at a specific stage of differentiation could be completely suppressed by the appropriate driver, and this suppression facilitates the amplification of differentially expressed genes in other cell populations or differentiation stages. RDA combined with this kind of gene DISCUSSION suppression would be helpful to focus on genes specific to a certain cell population in materials consisting of heterogenous cells. The receptors for the myeloid colony-stimulating factors, Although RDA is an effective technique, it still has some M-CSF, GM-CSF and G-CSF have been proposed to be critical limitations. First, our data clearly showed that some of the targets for the impaired myeloid development in PU.1 and differentially expressed genes are lost during the repeated C/EBPα mutant mice (3,17,18). We have previously shown by subtractive hybridization by increasing the stringencies (Table 1 northern blot analysis that G-CSF receptor mRNA is remarkably and Fig. 3B). Difference products with a low stringency could downregulated in C/EBPα knockout mice, whereas mRNAs for contain more differentially expressed genes but many more false M-CSF receptor and GM-CSF receptor are not impaired (3), positives as well. On the other hand, difference products with a suggesting that impaired G-CSF signaling might be responsible high stringency limit the number of fragments generated. This for selective block of neutrophilic differentiation. On the other problem could partially be resolved by suppression of expected hand, M-CSF receptor mRNA was undetectable by RT–PCR –/– difference products by specific DNA fragments or additional analysis of differentiated PU.1 ES cells (17,18). In this study, –/– drivers as we performed in this study. Secondly, RDA preferentially we analyzed the expression of myeloid CSF receptors in PU.1 amplifies genes with significant differences in expression. Most fetal liver by northern blot analysis, and noted that the expression of the differentially expressed genes identified were not expressed of all three is markedly decreased (Fig. 2). However, at least in the mutant fetal liver. Only several genes were still expressed M-CSF receptor and G-CSF receptor are still expressed at low –/– in mutant fetal liver at low levels. Decreasing the stringencies was levels, suggesting that PU.1 fetal liver cells could express at not as effective (Table 1). New modifications will be needed to least low levels of myeloid CSF receptors. No complementation amplify genes with small differences. A minor limitation is that the assays to rescue the defects by using myeloid CSF receptor transgenes have been reported. Therefore, the role of myeloid technique tends to isolate small portions of the full length cDNA. CSF receptors in both mutant mice still remains to be defined. In By RDA using PU.1 and C/EBPα knockout mice, we identified addition, the myeloid defects of PU.1 or C/EBPα knockout mice many differentially expressed genes, including myeloid- and do not completely match those of loss-of-function mutant mice of liver-specific genes. The expression of several liver genes have each CSF or CSF receptor (28–31). These findings suggest the already been shown to be downregulated in the fetal and newborn presence of additional genes regulated by PU.1 and C/EBPα. liver of C/EBPα knockout mice (26,27). We identified six In this study, we extended the studies of CSF receptor expression additional liver genes which are differentially expressed, and they and identified additional genes regulated by PU.1 and C/EBPα. We are presumably new targets for C/EBPα in hepatocytes. employed RDA, a PCR-based subtractive hybridization. RDA Interestingly, we happened to find that the expression of pref-1, eliminates those fragments present in both populations and a pre-adipocyte transmembrane protein, is upregulated in C/EBP- 3042 Nucleic Acids Research, 1998, Vol. 26, No. 12 –/– α fetal liver. During adipocyte differentiation, pref-1 is an important role of these transcription factors in the regulation reported to be downregulated, while C/EBPα is upregulated of lysozyme M expression. Moreover, we found that lysozyme M –/– (23,47). Our finding suggests that C/EBPα negatively regulates expression was impaired in the C/EBPα newborn lung. the expression of pref-1, and pref-1 might be a new direct target Lysozyme is expressed in type II alveolar pneumocytes and for C/EBPα in adipocytes. This finding also suggests the alveolar macrophages in rodent lung (57), while C/EBPα mRNA –/– existence of other negatively regulated genes by C/EBPα or PU.1. is localized to type II pneumocytes (58) and C/EBPα mice To selectively identify these genes, reverse RDA, i.e. mutant fetal show hyperproliferation of type II pneumocytes (27). C/EBPα is liver minus wild-type fetal liver, would be an approach to be taken also expressed in activated macrophages (A.Iwama and in the next step. D.G.Tenen, unpublished data). Taken together with the fact that With some modifications of RDA, we could preferentially PU.1 is not expressed in type II pneumocytes, C/EBPα could be identify myeloid-specific genes. Because PU.1 knockout mice a major regulator for lysozyme M expression in the lung, show impaired development of neutrophils as well as monocytes particularly in type II pneumocytes. These findings indicate that in the fetal liver, many neutrophil-specific genes were identified C/EBPα plays a major role in the regulation of lysozyme M in from both PU.1 and C/EBPα knockout mice (Tables 1–4). non-hematopoietic cells, and suggest the possibility that C/EBPα Among known genes, primary granule protein genes myeloper- is a key transcription factor in the regulation of genes specific to oxidase and neutrophilic elastase are known as common targets type II pneumocytes, such as surfactant protein genes. for both PU.1 and C/EBPα (11,12), and another primary granule The expression analysis of transcription factors in vivo protein gene, proteinase-3, is a target for PU.1 (13). Myeloperoxi- indicated that C/EBPα does not affect the expression of PU.1, + –/– dase mRNA is expressed in CD34 multipotential cells (48,49) because the reduction of PU.1 mRNA in C/EBPα fetal liver is and at high levels in myeloid progenitors at the promyelocytic and likely to parallel the decrease in mature granulocytic cells. On the promonocytic stages of myeloid differentiation (50,51), while contrary, PU.1 might be important in the regulation of C/EBPα, neutrophilic elastase and proteinase-3 mRNAs are expressed in and it is possible that impaired granulopoiesis in PU.1 knockout the promyelocytic stage (52). Although myeloid progenitors are mice is caused by defective C/EBPα expression. The expression –/– present in both knockout mice (3,17), and monocytic cells are intact analysis of C/EBPα in the neutrophilic cells of PU.1 bone –/– in C/EBPα fetal liver (3), the expression of myeloperoxidase marrow would help to address this question. The specific absence –/– mRNA was significantly low in both mutant fetal livers (Fig. 5), of Spi-B mRNA only in PU.1 fetal liver is consistent with our indicating that both PU.1 and C/EBPα are critical for the previous data of its B-cell-specific expression (8), and the –/– transcription of myeloperoxidase in vivo. Neutrophilic elastase absence of C/EBPε mRNA in C/EBPα fetal liver as well as –/– mRNA was also missing and proteinase-3 mRNA was markedly PU.1 fetal liver confirms its granulocyte-specific expression downregulated in both mutant mice. They are also likely in vivo (40). Spi-B and C/EBPε might be regulated by PU.1 or both PU.1 targets of these two transcription factors. These findings suggest and C/EBPα in each cell lineage. cooperative regulation of myeloid primary granule genes by PU.1 We identified eight novel myeloid genes differentially expressed and C/EBPα in vivo. between wild-type and mutant fetal livers. Among them, C/Edp The expression of secondary granule protein genes were also 2 and 3 are likely to be the murine homologues of neutrophil undetectable in both mutant mice (Fig. 5), possibly because of the collagenase and ficolin, respectively. C/Edp 1 and Pdp 3, and Pdp lack of expressing cells, such as myelocytes, metamyelocytes and 1 and 6 seem to be different DpnII fragments from the same gene, band cells. Low levels of mRNA of NGAL and neutrophilic respectively, because they showed the same mRNA expression –/– collagenase (C/Edp2) were detected only in PU.1 fetal liver. profile (Figs 4A and B, and 6). Northern blot analysis showed that This might well represent the development of a few neutrophilic most of the novel genes were preferentially expressed in the bone –/– cells in PU.1 fetal liver, consistent with the reported development marrow or peritoneal exudate cells, but not in spleen, thymus or –/– of cells characteristic of neutrophils in PU.1 bone marrow (5). adult liver (Fig. 4A and B). They were undetectable or only Recently, CCAAT displacement protein has been reported to weakly expressed in other tissues, such as brain, heart, lung, repress the expression of secondary granule protein genes kidney, skeletal muscle and testis (data not shown), indicating that (53,54), but transcription factors that directly activate their they are hematopoietic-specific genes. Only Pdp 1 and 6 were transcription have not been well characterized. Because the expressed in adult liver (Fig. 4B), but they were not differentially –/– expression of PU.1 and C/EBPα is maintained during granulocytic expressed between wild type and C/EBPα fetal liver (data not differentiation, they are candidate regulators of secondary shown), suggesting that they are expressed in macrophages, granule expression, as is C/EBPε (40,41). including Kupffer cells, in the liver. Most of them were Lysozyme M is a myeloid granule protein localized in both upregulated during myeloid differentiation of the multipotential primary and secondary granules. Its expression is already detectable hematopoietic cell line, EML (Fig. 6), suggesting these genes are in myeloblasts and upregulated during myeloid differentiation, good candidate targets for PU.1 and C/EBPα. Although C/Edp 6 including both granulocytic and monocytic lineages (43). PU.1 is and 7 were differentially expressed between wild-type and reported to activate the myeloid-specific enhancer of the chicken mutant fetal liver, we could not detect any apparent expression in lysozyme gene (55), and C/EBPβ interacts with another enhancer the bone marrow, peritoneal exudate cells, adult liver or other and mediates lipopolysaccharide-induced expression of the chicken adult organs. It is possible that they are specifically expressed in lysozyme gene (56). Northern blot analysis showed that lysozyme the embryonic stage. –/– M expression was absent in PU.1 fetal liver and markedly Our data confirmed the critical role of PU.1 and C/EBPα in vivo –/– impaired in C/EBPα fetal liver (Fig. 5). In addition, lysozyme in the regulation of myeloid genes, including myeloid CSF M mRNA was immediately upregulated after induction of receptors and myeloid granule proteins. Using RDA combined C/EBPα expression in an immature hematopoietic cell line with specific gene suppression, we further identified novel (A.Iwama and D.G.Tenen, unpublished data). These data suggest myeloid genes, the expression of which are missing in the mutant 3043 Nucleic Acids Research, 1998, Vol. 26, No. 12 3043 Nucleic Acids Research, 1994, Vol. 22, No. 1 23 Cao,Z., Umek,R.M. and McKnight,S.L. (1991) Genes Dev., 5, 1538–1552. fetal livers. These novel genes are new candidate targets for PU.1 24 Scott,L.M., Civin,C.I., Rorth,P. and Friedman,A.D. (1992) Blood, 80, and C/EBPα. Characterization of their roles in myeloid 1725–1735. development as well as their transcriptional regulation in relation 25 Radomska,H.S., Huettner,C.S., Zhang,P., Cheng,T., Scadden,D.T. and to PU.1 and C/EBPα will be helpful in elucidating the mechanism Tenen,D.G. (1998) Mol. Cell. Biol., in press. of impaired myeloid development caused by loss-of-function 26 Wang,N.D., Finegold,M.J., Bradley,A., Ou,C.N., Abdelsayed,S.V., Wilde,M.D., Taylor,L.R., Wilson,D.R. and Darlington,G.J. (1995) Science, mutation of PU.1 and C/EBPα. Additional studies, including 269, 1108–1112. isolation and characterization of the promoter elements, will be 27 Flodby,P., Barlow,C., Kylefjord,H., Ahrlund-Richter,L. and required to establish whether these genes are directly or indirectly Xanthopoulos,K.G. (1996) J. Biol. Chem., 271, 24753–24760. regulated by PU.1 and C/EBPα. 28 Yoshida,H., Hayashi,S., Kunisada,T., Ogawa,M., Nishikawa,S., Okamura,H., Sudo,T. and Shultz,L.D. (1990) Nature, 345, 442–444. 29 Nishinakamura,R., Nakayama,N., Hirabayashi,Y., Inoue,T., Aud,D., ACKNOWLEDGEMENTS Mcneil,T., Azuma,S., Yoshida,S., Toyoda,Y., Arai,K., Miyajima,A. and Murray,R. (1995) Immunity, 2, 211–222. 30 Lieschke,G.J., Stanley,E., Grail,D., Hodgson,G., Sinickas,V., Gall,J.A.M., We thank Linda Clayton for her technical instructions on RDA, Sinclair,R.A. and Dunn,A.R. (1994) Blood, 84, 27–35. Schickwann Tsai and Arch Perkins for providing EML and 31 Liu,F., Wu,H.Y., Wesselschmidt,R.L., Kornaga,T. and Link,D.C. (1996) BHK-MKL cells, respectively, Kleanthis Xanthopoulos for the Immunity, 5, 491–501. murine C/EBPε cDNA probe, David Gonzalez for oligonucleotides, 32 Lisitsyn,N. and Wigler,M. (1993) Science, 259, 946–951. Dong-Er Zhang, Liqin Liu, Milton Datta and Hanna Radomska for 33 Hubank,M. and Schatz,D.G. (1994) Nucleic Acids Res., 22, 5640–5648. 34 Braun,B.S., Frieden,R., Lessnick,S.L., May,W.A. and Denny,C.T. (1995) useful discussions. Supported by National Institutes of Health Grants Mol. Cell. Biol., 15, 4623–4630. CA41456, DK48660 and AI30656, and Japan research foundation 35 Lerner,A., Clayton,L.K., Mizoguchi,E., Ghendler,Y., van Ewijk,W., for clinical pharmacology and a long-term fellowship from the Koyasu,S., Bhan,A.K. and Reinherz,E.L. (1996) EMBO J., 15, 5876–5887. Human Frontier Science Program to A.I. 36 Zheng,W. and Flavell,R.A. (1997) Cell, 89, 587–596. 37 Tsai,S., Bartelmez,S., Sitnicka,E. and Collins,S. (1994) Genes Dev., 8, 2831–2841. 38 Chirgwin,J.M., Przybyla,A.E., MacDonald,R.J. and Rutter,W.J. (1979) REFERENCES Biochemistry, 18, 5294–5299. 39 Rosenberg,H.F., Ackerman,S.J. and Tenen,D.G. (1990) Biotechniques, 8, 384. 1 Shivdasani,R.A. and Orkin,S.H. (1996) Blood, 87, 4025–4039. 40 Yamanaka,R., Kim,G.-D., Radomska,H.S., Lekstrom-Himes,J., Smith,L.T., 2 Tenen,D.G., Hromas,R., Licht,J.D. and Zhang,D.-E. (1997) Blood, 90, Antonson,P., Tenen,D.G. and Xanthopoulos,K.G. (1997) Proc. Natl. Acad. 489–519. Sci. USA, 94, 6462–6467. 3 Zhang,D.-E., Zhang,P., Wang,N.D., Hetherington,C.J., Darlington,G.J. and 41 Yamanaka,R., Barlow,C., Lekstrom-Himes,J., Castilia,L.H., Liu,P.P., Tenen,D.G. (1997) Proc. Natl. Acad. Sci. USA, 94, 569–574. Eckhaus,M., Decker,T., Wynshaw-Boris,A. and Xanthopoulos,K.G. (1997) 4 Scott,E.W., Simon,M.C., Anastai,J. and Singh,H. (1994) Science, 265, 1573–1577. Proc. Natl. Acad. Sci. USA, 94, 13187–13192. 5 McKercher,S.R., Torbett,B.E., Anderson,K.L., Henkel,G.W., Vestal,D.J., 42 Chen,H.M., Gonzalez,D.A., Radomska,H.S., Voso,M.T., Sun,Z., Zhang,P., Baribault,H., Klemsz,M., Feeney,A.J., Wu,G.E., Paige,C.J. and Maki,R.A. Zhang,D.-E. and Tenen,D.G. (1998) Gene, 207, 209–218. (1996) EMBO J., 15, 5647–5658. 43 Borregaard,N. and Cowland,J.B. (1997) Blood, 89, 3503–3521. 6 Klemsz,M.J., McKercher,S.R., Celada,A., Van Beveren,C. and Maki,R.A. 44 Gruart,V., Truong,M.J., Plumas,J., Zandecki,M., Kusnierz,J.P., Prin,L., (1990) Cell, 61, 113–124. Vinatier,D., Capron,A. and Capron,M. (1992) Blood, 79, 2592–2597. 7 Hromas,R., Orazi,A., Neiman,R.S., Maki,R., Van Beveren,C., Moore,J. 45 Cross,M., Mangelsdorf,I., Wedel,A. and Renkawitz,R. (1988) and Klemsz,M. (1993) Blood, 82, 2998–3004. Proc. Natl. Acad. Sci. USA, 85, 6232–6236. 8 Chen,H.M., Zhang,P., Voso,M.T., Hohaus,S., Gonzalez,D.A., Glass,C.K., 46 Antonson,P. and Xanthopoulos,K.G. (1995) Biochem. Biophys. Res. Zhang,D.-E. and Tenen,D.G. (1995) Blood, 85, 2918–2928. Commun., 215, 106–113. 9 Voso,M.T., Burn,T.C., Wulf,G., Lim,B., Leone,G. and Tenen,D.G. (1994) 47 Smas,C.M. and Sul,H.S. (1993) Cell, 73, 725–734. Proc. Natl. Acad. Sci. USA, 91, 7932–7936. 48 Berliner,N., Hsing,A., Graubert,T., Sigurdsson,F., Zain,M., Bruno,E. and 10 Pahl,H.L., Scheibe,R.J., Zhang,D.-E., Chen,H.M., Galson,D.L., Maki,R.A. Hoffman,R. (1995) Blood, 85, 799–803. and Tenen,D.G. (1993) J. Biol. Chem., 268, 5014–5020. 49 Strobl,H., Takimoto,M., Majdic,O., Fritsch,G., Scheinecker,C., Hocker,P. 11 Ford,A.M., Bennett,C.A., Healy,L.E., Towatari,M., Greaves,M.F. and and Knapp,W. (1993) Blood, 82, 2069–2078. Enver,T. (1996) Proc. Natl. Acad. Sci. USA, 93, 10838–10843. 50 Tobler,A., Miller,C.W., Johnson,K.R., Selsted,M.E., Rovera,G. and 12 Oelgeschlager,M., Nuchprayoon,I., Luscher,B. and Friedman,A.D. (1996) Koeffler,H.P. (1988) J. Cell Physiol., 136, 215–225. Mol. Cell. Biol., 16, 4717–4725. 51 Rosmarin,A.G., Weil,S.C., Rosner,G.L., Griffin,J.D., Arnaout,M.A. and 13 Sturrock,A., Franklin,K.F. and Hoidal,J.R. (1996) J. Biol. Chem., 271, Tenen,D.G. (1989) Blood, 73, 131–136. 32392–32402. 52 Fouret,P., du Bois,R.M., Bernaudin,J.F., Takahashi,H., Ferrans,V.J. and 14 Hohaus,S., Petrovick,M.S., Voso,M.T., Sun,Z., Zhang,D.-E. and Crystal,R.G. (1989) J. Exp. Med., 169, 833–845. Tenen,D.G. (1995) Mol. Cell. Biol., 15, 5830–5845. 53 Khanna-Gupta,A., Lawson,N., Zibello,T. and Berliner,N. (1996) Blood, 88, 15 Smith,L.T., Hohaus,S., Gonzalez,D.A., Dziennis,S.E. and Tenen,D.G. 630a. (1996) Blood, 88, 1234–1247. 54 Khanna-Gupta,A., Zibello,T., Kolla,S., Neufeld,E.J. and Berliner,N. (1997) 16 Zhang,D.-E., Hetherington,C.J., Chen,H.M. and Tenen,D.G. (1994) Blood, 90, 2784–2795. Mol. Cell. Biol., 14, 373–381. 55 Ahne,B. and Stratling,W.H. (1994) J. Biol. Chem., 269, 17794–17801. 17 Olson,M.C., Scott,E.W., Hack,A.A., Su,G.H., Tenen,D.G., Singh,H. and 56 Goethe,R. and Loc,P.V. (1994) J. Biol. Chem., 269, 31302–31309. Simon,M.C. (1995) Immunity, 3, 703–714. 57 Singh,G., Katyal,S.L., Brown,W.E., Collins,D.L. and Mason,R.J. (1988) 18 Henkel,G.W., McKercher,S.R., Yamamoto,H., Anderson,K.L., Am. Rev. Respir. Dis., 138, 1261. Oshima,R.G. and Maki,R.A. (1996) Blood, 88, 2917–2926. 58 Li,F., Rosenberg,E., Smith,C.I., Notarfrancesco,K., Reisher,S.R., 19 Scott,E.W., Fisher,R.C., Olson,M.C., Kehrli,E.W., Simon,M.C. and Shuman,H. and Feinstein,S.I. (1995) Am. J. Physiol., 269, L241–L247. Singh,H. (1997) Immunity, 6, 437–447. 59 Rothwell,V.M. and Rohrschneider,L.R. (1987) Oncogene Res., 1, 311–324. 20 Tondravi,M.M., McKercher,S.R., Anderson,K., Erdmann,J.M., Quiroz,M., 60 Fukunaga,R., Ishizaka-Ikeda,E., Seto,Y. and Nagata,S. (1990) Cell, 61, Maki,R. and Teitelbaum,S.L. (1997) Nature, 386, 81–84. 341–350. 21 Johnson,P.F., Landschulz,W.H., Graves,B.J. and McKnight,S.L. (1987) 61 Liboi,E., Jubinsky,P., Andrews,N.C., Nathan,D.G. and Mathey-Prevot,B. Genes Dev., 1, 133–146. 22 Costa,R.H., Grayson,D.R., Xanthopoulos,K.G. and Darnell,J.E.,Jr (1988) (1992) Blood, 80, 1183–1189. Proc. Natl. Acad. Sci. USA, 85, 3840–3844. 62 D’Andrea,A.D., Lodish,H.F. and Wong,G.G. (1989) Cell, 57, 277–285.

Journal

Nucleic Acids Research – Oxford University Press

Published: Jun 1, 1998

There are no references for this article.