In the human pathogen Mycoplasma genitalium, homologous recombination is under the control of r , an alternative sigma factor that boosts the generation of genetic and antigenic diversity in the population. Under laboratory growth conditions, r activation is rare and the factors govern- ing its intermittent activity are unknown. Two r -regulated genes, rrlA and rrlB, showed to be important for recombination of homologous DNA sequences in this bacterium. Herein, we dem- onstrate that rrlA and rrlB code for two small proteins that participate in a feed-forward loop es- 20 20 sential for r function. In addition, we identify novel genes regulated by r and show that several non-coding regions, which function as a reservoir for the generation of antigenic diversity, are also activated by this alternative sigma factor. Finally, we reveal that M. genitalium cells can transfer DNA horizontally by a novel mechanism that requires RecA and is facilitated by r over- expression. This DNA transfer system is arguably fundamental for persistence of M. genitalium within the host since it could facilitate a rapid dissemination of successful antigenic variants within the population. Overall, these findings impose a novel conception of genome evolution, genetic variation and survival of M. genitalium within the host. Key words: mycoplasma, sigma factor, homologous recombination, antigenic variation, horizontal gene transfer 1. Introduction sequences. These chromosomal rearrangements are essentially facili- Pathogenic microorganisms have evolved sophisticated strategies to tated by homologous recombination and accordingly, RecA and other 1 4–8 evade or subvert the host immune system. Typically, extracellular important recombination enzymes play a fundamental role in AnV. bacteria modify their surface structures or control the expression of Likewise, the participation of proteins that regulate the capacity of their immunodominant proteins to avoid antibody recognition. These RecA to polymerize or load onto ssDNA in the generation of antigenic 9,10 two widespread strategies meant to survive and persist within the host variants is substantiated by several reports. Considering these stud- are known as antigenic variation (AnV) and phase variation (PhV), re- ies, RecA emerges as the primary target to control sequence variation spectively. The sexually transmitted pathogens Mycoplasma genita- of major surface antigens by homologous recombination. lium, Neisseria gonorrhoeae and Treponema pallidum,generate AnV Remarkably, despite RecA is critical for DNA repair and maintenance by means of programmed rearrangements of unique chromosomal of genome stability in bacteria, the sexually transmitted pathogens V C The Author(s) 2018. Published by Oxford University Press on behalf of Kazusa DNA Research Institute. This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (http://creativecommons.org/licenses/by-nc/4.0/), which permits non-commercial re-use, distribution, and reproduction in any medium, provided the original work is properly cited. For commercial re-use, please contact firstname.lastname@example.org 1 Downloaded from https://academic.oup.com/dnaresearch/advance-article-abstract/doi/10.1093/dnares/dsy011/4967678 by Ed 'DeepDyve' Gillespie user on 12 July 2018 2 S. Torres-Puig et al. aforementioned are devoid of a classical SOS system coordinating the needed. All primers used in this study are listed in Supplementary concerted activation of recA and other repair genes in response to Table S2. 7,11–14 DNA insults. In light of these data, it has been suggested that the participation of RecA in the generation of genetic variants imposes 2.2. DNA manipulation important restrictions as to the multifaceted mechanisms that bacteria exploit to regulate recombination. Standard techniques for cloning were performed as described in P140 (MgpB) and P110 (MgpC) are the major cytadhesins and Sambrook and Russell. Plasmid DNA was obtained using Fast the main surface antigens of M. genitalium. In its chromosome, this Plasmid Mini kit (5Prime). PCR products were puriﬁed from aga- bacterium comprises nine DNA repeats, designated as MgPar, that rose gels using Nucleospin Gel and PCR Clean-up kit (Macherey- contain sequences with homology to the MG_191 and MG_192 Nagel) and digested with the corresponding restriction enzymes 12,16 genes, which code for P140 and P110, respectively. (Fermentas) when necessary. Plasmids for M. genitalium transfor- Recombination between the cytadhesin genes and MgPar sequences mation were obtained using the GenElute HP Midiprep kit provides a virtually unlimited collection of genetic and antigenic vari- (Sigma). 17,18 ants. Moreover, recombination with particular MgPar regions leads to the expression of truncated P140 or P110 proteins, which 2.3. Mutant construction, transformation and screening evidences the existence of a subjacent and versatile mechanism of PhV. Remarkably, P140 and P110 variation is critical for survival A detailed explanation of the steps and methodology, including pri- 20–23 and persistence of M. genitalium within the host. Recently, we mers and plasmids, used to construct all the mutants created in this identiﬁed an alternative sigma factor, herein designated as r , that study is supplied as Supplementary Information. Transformation of controls the activation of homologous recombination in M. genita- M. genitalium was carried out as previously described. Screening 24 20 lium. Of note, r is a major determinant for the generation of ge- for mutants was performed using cell lysates as template for PCR or 24,25 20 netic variants of the cytadhesin genes. r regulates sequencing reactions. Cell lysates were obtained by centrifugation of transcription of recA (MG_339), ruvA (MG_358) and ruvB 5 ml cell cultures, disruption of pellets using 30 ml of Lysis Buffer (MG_359), plus several genes of unknown function. Bewilderingly, (Tris–HCl 0.1 M pH 8.5, Tween-20 0.05%, proteinase K 20 1 under laboratory growth conditions, the r regulon is only acti- 0.25 mg ml ) and incubation for 1 h at 37 C followed by inactiva- vated in a small subset of cells in response to unidentiﬁed factors. tion at 95 C for 10 min. These singularities impose single cell analyses as the only suitable techniques to study and characterize r activity. Of note, overex- 2.4. RNA extraction and transcriptional analysis pression of r induces a hyper-recombinogenic phenotype that is highly deleterious. Despite these hitches, we recently identiﬁed two Mycoplasma genitalium was grown to mid-exponential phase in 20 2 novel genes under the control of r , designated as recombination 25 cm tissue culture ﬂasks. Attached mycoplasmas were scrapped regulatory loci A and B (rrlA and rrlB), which are intimately related off in 1 ml of fresh SP-4, inoculated in two new 25 cm tissue culture to homologous recombination. Null mutants of these genes exhibit ﬂasks with fresh SP-4 medium and incubated at 37 C5% CO for severe recombination defects, similar to those observed in a r de- 6 h. Then, total RNA was extracted using the miRNeasy Mini Kit fective mutant, but the exact role of the RrlA and RrlB proteins in (Qiagen) following manufacturer’s instructions. Contaminant DNA M. genitalium remains obscure. was eliminated with the RNase-Free DNase Set (Qiagen). Overall, in the current study we provide further knowledge re- To conduct the RNAseq study, three independent biological re- garding the function of r in M. genitalium. We describe the whole peats of each strain were submitted to analysis. RNA libraries were regulon and identify several factors controlling r activation and prepared with TruSeq Stranded Total RNA Library Prep Kit modulating recombination in this bacterium. Moreover, we reveal a (Illumina) and analysed using a HiSeq 3000 System (Illumina) at the link between r and a mechanism of horizontal gene transfer Genomics Unit from Center for Genomic Regulation (CRG), (HGT) that is mediated by the recombination machinery of this bac- Barcelona. cDNA clusters were immobilized in sequencing lanes of terium. This novel activity associated to the r pathway is indepen- 2 50 reads. Prior to any data analysis, reverse and complementary dent of any known mobile genetic elements, which deﬁes traditional was computed for sequences coming from Read1 primer. Data HGT systems and might be a valuable tool for genome evolution and analysis and sequence alignment was performed using Bowtie2 adaptation of this small pathogen. tool in the End-to-End mode and Forward-Forward paired-ends. Sequences were piled up using SAMtools with no limit set to the number of sequences in the alignment. Counts in the different ORFs were performed with a standalone version of featureCounts pro- 2. Materials and methods gram without counting the multi-mapping reads and disabling 2.1. Bacterial strains, culture conditions and primers multi-overlapping reads. All M. genitalium strains were grown in SP-4 broth at 37 Cin a 5% Counted features were then submitted to the R/Bioconductor 30,31 CO atmosphere in tissue culture ﬂasks. SP-4 plates were prepared package DESeq2 for statistical analysis. DESeq2 analysis used a supplementing the medium with 0.8% agar (BD). Tetracycline parametric ﬁtType and a zero-mean normal prior on the non- 1 1 (3 mgml ), chloramphenicol (Cm) (17 mgml ) or puromycin (Pm) intercept coefﬁcients. Data were sorted by log2 fold change and sta- (3 mgml ) were added for mutant selection when necessary. All M. tistical signiﬁcance was set at P-value< 0.05. DESeq2 was chosen as genitalium strains used in this work are listed in Supplementary the RNAseq normalization method over other widely used proce- Table S1. Escherichia coli strain XL-1 Blue was used for cloning and dures, such as RPKM (Reads Per Kilobase per Million mapped plasmid propagation. The strain was grown in Luria Bertani (LB) or reads) or TC (Total Count), since a recent study has shown that 1 1 LB agar plates containing 100 lgml ampicillin, 40 lgml X-Gal DESeq2 normalization can maintain a reasonable false-positive rate 1 32 and 24 lgml Isopropyl b-D-1-thiogalactopyranoside (IPTG) when in different library sizes and widely different library compositions. Downloaded from https://academic.oup.com/dnaresearch/advance-article-abstract/doi/10.1093/dnares/dsy011/4967678 by Ed 'DeepDyve' Gillespie user on 12 July 2018 20 r activation in Mycoplasma genitalium 3 2.5. Primer extension 2.9. Mating of M. genitalium strains Primer extension analyses were performed with 20 lgof total Mycoplasma genitalium strains were grown separately in 75 cm tis- RNA as previously described. Total RNA was extracted from sue culture ﬂasks until mid-log phase. Then, cells were recovered in mid-log phase cultures using the RNAqueous kit (Life 10 ml of fresh SP-4 and passed through 0.45 mm ﬁlters. Four milli- Technologies) and treated using Turbo DNase (Life Technologies) litre of the cell suspension from each strain were mixed and incu- following manufacturer’s instructions. Fragments were analysed bated in a 75 cm tissue culture ﬂask with 12 ml of fresh SP-4 using PeakScanner v1.0 software (Applied Biosystems). At least without antibiotic selection. After 24 h of co-incubation, cells were two independent primer extension experiments were performed scrapped off in 1 ml of fresh SP-4 and 200 ml aliquots were seeded with each primer. on 0.9% SP-4 agar plates (8.5 cm diameter) supplemented with Cm 1 1 2 (34 mgml ) and Pm (3 mgml ) or used to inoculate 75 cm tissue culture ﬂasks with dual antibiotic selection. To exclude transforma- tion as a mechanism of transfer, the mating experiments were per- 2.6. Sequencing reactions V R formed in the presence of DNase. After 14 days, colonies were DNA sequencing reactions were performed using BigDye v3.1 picked up and screened for by PCR and sequencing of the resulting Cycle Sequencing kit using 2.5 ml of genomic DNA or M. genitalium amplicons. Mating efﬁciency was calculated by dividing the num- lysate, following manufacturer’s instructions. All reactions were ana- ber of double resistant colonies obtained on selective medium by lysed in an ABI PRISM 3130xl Genetic Analyzer at the Servei de the number of mycoplasma colonies obtained on non-selective me- Geno ` mica i Bioinforma ` tica (UAB). dium. For the liquid cultures, typically 14–16 days were needed to observe colonization of the ﬂask surface. Dual antibiotic selection was maintained during all the process. Mobilization of the antibi- 2.7. Quantitative assessment of the recombination otic markers in the cell pools was also assessed by PCR and se- capacity quencing analysis of the resulting amplicons. Recombination capacity was calculated using the transformation ef- ﬁciency by homologous recombination of a suicide plasmid as previ- ously described. Results presented in the manuscript correspond to 3. Results at least three independent biological repeats. 3.1. Elucidation of the whole r regulon The recent identiﬁcation and initial characterization of the r regulator 2.8. Phase contrast and fluorescence microscopy of M. genitalium was not accompanied by a comprehensive transcrip- Mycoplasma genitalium was grown in ﬁltered (0.22 mm) SP-4 medium tional study. Herein, we conducted a genome-wide RNA-Seq analysis on IBIDI chamber slides for 16 h, washed once with 1 PBS and visu- to identify genes controlled by r -dependent promoters. To this end, alized on a Nikon Eclipse TE 2000-E microscope. All strains were we compared RNA samples from strains lacking or overexpressing r grown and visualized under the same conditions. Phase contrast and to those of the wild-type strain. We found that transcription of up to TRITC epiﬂuorescence images were captured with a Digital Sight DS- thirteen genes increased signiﬁcantly upon r overexpression SMC Nikon camera controlled by NIS-Elements BR software. Images (Table 1). In keeping with previous data, we observed activation of were analysed using Image J software and GDSC plug-in. Fluorescence recA, ruvAB and the recombination regulatory loci rrlAB. intensities were determined by quantifying gray levels of individual Additionally, we identiﬁed three novel genes controlled by r -depen- cells in binary images using Image J software. dent promoters: MG_285, MG_286 and MG_412. While the MG_285 Table 1. Differentially transcribed genes upon overexpression (Up) or deletion (DMG_428) of r Mean transcript fold-increase New locus Old locus Annotation Function Up r DMG_428 MG_RS01295 MG_220 rrlA Sigma accessory protein 56.57 0.63 MG_RS02205 MG_358 ruvA Recombination and repair 20.7 0.98 MG_RS02550 MG_428 rpo20 Alternative sigma factor 20.6 0.06 MG_RS02065 MG_339 recA Recombination and repair 19.6 0.79 MG_RS02210 MG_359 ruvB Recombination and repair 18.7 0.99 MG_RS01710 MG_286 HP Unknown 17.7 0.89 MG_RS01705 MG_285 HP Unknown 16.3 0.93 MG_RS02495 MG_414 HP Unknown 15.4 0.79 MG_RS02200 – rrlB Sigma accessory protein 12.4 0.90 MG_RS02370 MG_389 HP Unknown 9.7 0.90 MG_RS00050 MG_010 HP Replication (putative) 3.8 0.84 MG_RS02490 MG_412 pstS Phosphate binding lipoprotein (putative) 2.8 1.0 MG_RS02375 MG_390 sunT Peptide secretion (putative) 2.4 0.91 – – ncRNA-1 Unknown 5.4 0.80 – – ncRNA-2 Unknown 20.56 0.88 – – ncRNA-3/4 Unknown 86.82 0.02 Differential gene expression compared with the WT strain and judged based on the common arbitrary 2-fold cutoff. Downloaded from https://academic.oup.com/dnaresearch/advance-article-abstract/doi/10.1093/dnares/dsy011/4967678 by Ed 'DeepDyve' Gillespie user on 12 July 2018 4 S. Torres-Puig et al. and MG_286 genes code for two polypeptides with unknown function, identiﬁed two independent TSS. As expected, all the identiﬁed TSSs were the protein encoded by the MG_412 gene shows sequence homology to preceded by r promoter sequences (Fig. 1B), which show the consen- 0 0 the substrate binding subunit of phosphate transport systems. sus sequence 5 -TTGTCA-N18/19-ATTWAT-3 in M. genitalium. Of note, we found that several transcripts originating from chromo- Remarkably, the chromosome of Mycoplasma pneumoniae car- somal regions without annotated genes were markedly upregulated ries an orthologue of the MG_428 gene and accordingly, r -regu- upon r overexpression (Table 1 and Fig. 1A). These transcripts com- lated promoters of M. genitalium are conserved in this close related 24 20 prise DNA sequences with homology to the cytadhesin genes of M. gen- species. In keeping with this idea, we identiﬁed a putative r pro- italium. The transcript designated as ncRNA-1 lies within the MgPar-4 moter in the upstream region of the MPN404 gene, the ortholog of 17 20 region of the G37 chromosome. It comprises the repeat G of P140 MG_285. Likewise, we found a putative r promoter in the up- and the large KLM repeat of P110, both in the antisense orientation. stream region of MPN143, which codes for a protein of 175 amino Transcripts ncRNA-2 and ncRNA-3/4 lie within the MgPar-5 region, acids exclusive of M. pneumoniae. The MPN143 gene is located im- which is located immediately downstream of the cytadherence operon. mediately downstream of the cytadherence genes (MPN141 and ncRNA-2 comprises the repeats B and EF of P140, plus a putative novel MPN142), a chromosomal location that is similar to the putative ORF located at the 5 end of the transcript and herein designated as novel orf192.1 identiﬁed in M. genitalium (Supplementary Fig. S2). orf192.1. Upregulation of ncRNA-3/4 is particularly apparent and encompasses two short transcripts located immediately upstream of the 3.3. Expression of RrlA, RrlB and ORF192.1 in single EF repeat of the MgPar-5. In agreement with the intermittent activation cells of r , these two short transcripts were still detected in the WT strain To support and expand our transcriptional data, we monitored the (Fig. 1A). In contrast, transcription of ncRNA-3/4 was downregulated expression of RrlA and RrlB in single cells using ﬂuorescent protein by 50-fold in strains lacking r (Table 1 and Fig. 1A). fusions. Similarly, we used the same approach to test whether the pu- tative novel ORF identiﬁed within the ncRNA-2 transcript, orf192.1, 3.2. Identification of novel r -regulated promoters was expressed in M. genitalium. For this purpose, we obtained mu- Primer extension analyses established the presence of r -dependent tants carrying transcriptional fusions of rrlA, rrlB or orf192.1 to the transcriptional start sites (TSS) within the upstream regions of the mcherry ﬂuorescent marker at their respective native loci, which MG_285 and MG_412 genes (Supplementary Fig. S1). Similarly, we were designated as RrlA:Ch, RrlB:Ch and ORF192.1:Ch. As previ- 20 24 identiﬁed TSSs upstream of the three upregulated regions without anno- ously observed for other r -regulated genes, only a small subset tated genes (Supplementary Fig. S1). For the ncRNA-3/4 transcript, we of cells displayed mCherry ﬂuorescence (Table 2 and Fig. 2). Figure 1. Chromosomal localization of the ncRNAs activated by r . (A) Normalized relative coverage of the RNA-Seq data for the WT strain, a mutant overex- pressing r , and the MG_428 and MG_191 null mutants, respectively. Only the chromosome section corresponding to the cytadherence operon and its flanking regions is shown. For each strain, the upper and lower panels depict the coverage for the þ and – strands, respectively. Different ORFs are coloured according to the different COG families (Supplementary Table S3). (B) Alignment of the upstream regions of the identified ncRNAs. Conserved bases corresponding to the -35 and -10 elements recognized by r are highlighted in bold and experimentally identified TSS are underlined. Downloaded from https://academic.oup.com/dnaresearch/advance-article-abstract/doi/10.1093/dnares/dsy011/4967678 by Ed 'DeepDyve' Gillespie user on 12 July 2018 20 r activation in Mycoplasma genitalium 5 Figure 2. Single cell analysis of RrlA, RrlB and ORF192.1 expression in different mutant backgrounds. Each row contains a series of three fluorescence micros- copy images corresponding to the phase contrast, the TRITC channel and the resulting overlay of the different mutants analysed. White arrows indicate the presence of mCherry fluorescent cells in strains where fluorescence is rare. Scale bar is 10 mm. Remarkably, the percentage of ﬂuorescent cells observed for each from the complemented mutants were non-ﬂuorescent. This was un- target protein was relatively diverse, which is likely due to a different expected because transcriptional fusions of the MG_427 gene to the strength of the transcriptional response plus speciﬁc differences in mcherry marker at its native locus indicate that the MG_427 protein turnover. promoter is constitutive (Supplementary Fig. S3). Altogether, these In agreement with the r -controlled expression of rrlA and rrlB, ﬁndings suggest that transcription of the target genes is still regulated we found that deletion of the MG_428 gene completely abrogated in the complemented strains. Therefore, either the expression of mCherry ﬂuorescence in the population (Table 2, Fig. 2 and rrlAB is under the control of additional factors beyond r or the Supplementary Fig. S3). Next, we reintroduced the MG_428 gene by activity of r is regulated at the post-transcriptional level in 20 20 transposon delivery (Tnr ) to the mcherry mutants lacking r to M. genitalium. assess complementation. Transcription of the transposon-encoded 24,33 copy of the MG_428 gene was driven by the MG_427 promoter, 3.4. RrlA and RrlB are necessary for the activation of which according to our transcriptional data is strong and not regu- the r -regulon lated by r . Expression of the ectopic copy of the MG_428 gene reestablished mCherry ﬂuorescence to all mutants (Fig. 2 and Aiming to better understand the role of rrlA and rrlB in the recombi- Supplementary Fig. S3). In particular, the presence of cells with de- nation pathway, we deleted one of these genes from strains carrying tectable levels of RrlA and RrlB increased by 25- and 18-fold, respec- MG_428- or recA-mcherry fusions at their respective native loci. In 20 20 tively, as compared with the parental strains where r was both mutant backgrounds, r :Ch and RecA:Ch, the absence of rrlA expressed from its native locus (Table 2). Yet, the majority of cells or rrlB completely abrogated the presence of ﬂuorescent cells Downloaded from https://academic.oup.com/dnaresearch/advance-article-abstract/doi/10.1093/dnares/dsy011/4967678 by Ed 'DeepDyve' Gillespie user on 12 July 2018 6 S. Torres-Puig et al. Table 2. Quantitative data of mCherry fluorescence 3.6. Role of RrlA and RrlB in r stability To get further knowledge on the role of RrlAB in the activation of Strain % of ﬂuorescent Average ﬂuorescence the r regulon, we deleted the rrlA or rrlB genes from the RecA:Ch cells intensity per cell6 SD Tnr :YFP reporter strain. This strain was chosen because it allows MG427:Ch 99.48 33.196 17.81 the study of r expression and activity at the same time. As de- ORF192.1:Ch 0.54 39.976 25.34 scribed earlier, mCherry ﬂuorescence was not observed when RrlA RrlA:Ch 0.32 14.626 10.46 or RrlB were absent (Table 2 and Fig. 4). Concurrently, the presence RrlA:Ch DMG_428 0 – of cells with YFP ﬂuorescence was substantially reduced and cells RrlA:Ch DMG_428 Tnr 8.13 38.926 41.64 with intense YFP ﬂuorescence were no longer observed (Fig. 5). RrlB:Ch 1.91 12.926 9.24 These data reinforce the role of RrlAB in r activation and suggest RrlB:Ch DMG_428 0 – that these two small proteins likely stabilize r . On the other hand, RrlB:Ch DMG_428 Tnr 34.17 58.526 44.22 RrlA or RrlB overexpression using the TnrrlA and TnrrlB minitrans- RecA:Ch 0.66 19.976 16.48 posons did not modify the percentage of RecA-mCherry or r -YFP RecA:Ch DrrlA 0– RecA:Ch DrrlB 0– ﬂuorescent cells (Table 2 and Supplementary Fig. S5). Remarkably, RecA:Ch DMG_390 0.51 25.846 21.81 the majority of clones (70%) recovered from these transformation RecA:Ch DMG_414 0.58 22.806 11.76 experiments showed a non-adherent phenotype, which is indicative RecA:Ch DrrlA TnrrlA 0.22 ND of an increased frequency of generation of phase variants. RecA:Ch DrrlB TnrrlB 0.24 ND RecA:Ch Tnr :YFP 3.21 57.306 62.76 3.7. r overexpression reestablishes recombination to RecA:Ch Tnr :YFP DrrlA 0– 20 - - RecA:Ch Tnr :YFP DrrlB 0– rrlA and rrlB mutants RecA:Ch Tnr :YFP TnrrlA 2.60 ND On the basis of our ﬁndings, the participation of RrlAB in the homol- RecA:Ch Tnr :YFP TnrrlB 2.85 ND ogous recombination pathway is expected to be indirect. To support r :Ch 0.46 6.506 4.09 20 this notion, we overexpressed r in mutants lacking RrlA, RrlB or r :Ch DrrlA 0– RecA using the Tnr minitransposon (Fig. 6). As expected, in agree- r :Ch DrrlB 0– 20 ment with the central role of RecA in the recombination pathway, r :Ch DrrlA TnrrlA 0.13 ND r overexpression did not restore recombination to recA mutants. r :Ch DrrlB TnrrlB 0.20 ND 20 20 In contrast, r overexpression reestablished recombination to the r :Ch Tnr :YFP 3.02 29.116 16.42 rrlB mutant, though the recombination levels were moderate as com- SD: standard deviation; ND: not determined. pared with the wild-type strain. The recombination defects were still pronounced in the rrlA mutants overexpressing r , but unlike in the parental strain, some recombination events could be detected. (Table 2, Fig. 3 and Supplementary Fig. S4). In contrast, deletion of other genes under the control of r such as MG_390 or MG_414, 3.8. r overexpression facilitates HGT had very little or no impact on RecA expression (Table 2). Previously, we described that cells with an active r pathway were Reintroduction of the rrlA or rrlB genes by transposon delivery often observed in pairs in single cell analyses. On the basis of this (TnrrlA or TnrrlB) reestablished the presence of ﬂuorescent cells to observation, we wondered whether the activation of r -recombina- their respective mutant backgrounds (Table 2, Fig. 3 and tion could be associated to a mechanism of HGT. To ascertain this Supplementary Fig. S4). Hence, our data indicate an important role question, we mixed strains carrying two different antibiotic resis- for RrlAB in the activation of r -dependent recombination. tance markers in the presence of high concentrations of DNase and assessed the generation of double resistant mutants. In the ﬁrst at- 3.5. Positive autoregulation by r tempts, we mixed strains where r expression was driven from its Next, we wondered whether r expression was autoregulated. To as- natural locus. We used combinations of different antibiotic resis- certain this question, we delivered an ectopic copy of the MG_428 gene tance markers and strains, but we repeatedly failed to isolate double fusedtothe eyfp ﬂuorescent marker under the control of a constitutive resistant mutants. Then, we initiated experiments using strains over- 20 20 20 20 promoter of M. genitalium (Tn r :YFP) into the r :Ch and RecA:Ch expressing r by means of the Tnr minitransposon (Fig. 7). This mutants. In both cases, we observed a 6-fold increase in the percentage time, we successfully isolated mutants resistant to Cm and Pm upon of cells displaying mCherry ﬂuorescence (Table 2 and Fig. 4). In addi- incubation of a strain carrying the cat and tetM markers, and a tion, the average mCherry ﬂuorescence intensity of the population was strain carrying the pac gene. The calculated mating efﬁciency was 20 8 incremented 3–6-fold by the presence of the Tnr :YFP transposon 1.16 0.2 10 transconjugants per viable cell, which is in agree- 34,35 (Table 2). Altogether, these data demonstrate the existence of a positive ment with the transfer efﬁciencies reported in other mollicutes. feedback during the activation of the r pathway. The presence of the cat and pac markers in the isolated mutants was On the other hand, despite the majority of cells carrying the r : conﬁrmed by PCR and sequencing (Supplementary Fig. S6). Of YFP fusion showed yellow ﬂuorescence (74.96 2.3%), cells without note, we did not detect the presence of the tetM marker in the trans- ﬂuorescence were also present (Fig. 4). Moreover, YFP ﬂuorescence conjugant cells, suggesting the transfer of the cat gene from the intensity was exceptionally heterogeneous and we observed individ- strain overexpressing r (donor) to the Pm resistant strain (recipi- ual cells (2%) with an intensity 10–20-fold higher than the average ent). The location of the cat gene in the recipient strain was the of the population (Fig. 5). Of note, in the RecA-mCherry back- same as in the donor strain, suggesting that the cat marker was inte- ground, we observed a strong correlation between mCherry positive grated into the chromosome by homologous recombination. cells and cells with intense YFP ﬂuorescence (Fig. 4), suggesting that Similarly, we also observed the mobilization of the pac gene from a 20 20 20 r is more stable coinciding with the activation of the r regulon. donor strain overexpressing r to a recipient strain carrying the Downloaded from https://academic.oup.com/dnaresearch/advance-article-abstract/doi/10.1093/dnares/dsy011/4967678 by Ed 'DeepDyve' Gillespie user on 12 July 2018 20 r activation in Mycoplasma genitalium 7 Figure 3. Single cell analysis of RecA and r expression in different mutant backgrounds. Each row contains a series of three fluorescence microscopy images corresponding to the phase contrast, the TRITC channel and the resulting overlay of the different mutants analysed. White arrows indicate the presence of mCherry fluorescent cells in strains where fluorescence is rare. Scale bar is 10 mm in all images. cat marker (data not shown). In addition, we also found that recipi- 4. Discussion ent strains receiving the ectopic MG_428 gene, responsible for r In M. genitalium, homologous recombination is controlled by the al- overexpression, become donor strains capable to transfer selectable 20 24 ternative sigma factor r . Under laboratory growth conditions, markers (Fig. 7B). r activity is only observed in a small subset of cells and so far, it is On the other hand, we found that the absence of RecA or the ex- unknown whether this intermittent activation is elicited by means or pression of a truncated r protein in the donor strain, prevented transcriptional or post-transcriptional factors. Herein, we assessed the isolation of transconjugant strains (Fig. 7A). Similarly, heat in- the expression in single cells of a r -YFP fusion protein under the activation of the donor strains prevented the isolation of double re- control of a constitutive, endogenous promoter of M. genitalium.We sistant mutants. Moreover, we could not isolate double resistant found that r levels were highly heterogeneous in the population mutants upon incubation of the recipient strains with chromosomal and r expression could not be detected in numerous cells. DNA from potential donors or plasmid DNA carrying different Although we cannot rule out the existence of transcriptional factors antibiotic markers. Altogether, these data indicate that activation of controlling r -activation, our results demonstrate a conspicuous r facilitates cell-to-cell transfer of DNA sequences by homologous post-transcriptional regulation of r function. On the other hand, recombination by an uncharacterized mechanism of HGT of M. we found a direct correlation between the presence of rrlAB, two genitalium. Downloaded from https://academic.oup.com/dnaresearch/advance-article-abstract/doi/10.1093/dnares/dsy011/4967678 by Ed 'DeepDyve' Gillespie user on 12 July 2018 8 S. Torres-Puig et al. 20 20 Figure 4. Single cell analysis of RecA and r expression upon r overexpression. Each row contains a series of four fluorescence microscopy images corre- sponding to the phase contrast, the TRITC channel, the eYFP channel and the resulting overlay of the different mutants analysed. Yellow arrows point to cells showing an intense YFP fluorescence. Red arrows point to mCherry fluorescent cells that show also intense YFP fluorescence. Scale bar is 10 mm in all images. Figure 5. Fluorescence intensity distribution in the population of different mutants carrying a r :YFP fusion. Distribution was obtained by analysing at least 150 YFP fluorescent cells from each strain. AU denotes arbitrary units. Figure 6. Recombination capacity of different mutants. Graphic showing the recombination capacity of different M. genitalium mutants. Bars represent the averages and the standard deviations of three independent biological 20 20 20 genes subject to r -regulation, and r activity. Moreover, r ex- repeats. The recombination capacity of the DrrlA, DrrlB, DrecA and DrecA pression was negligible when RrlAB were absent, indicating that Tnr mutants was below the limit of detection. Asterisks mark differences these small proteins act synergistically to stabilize and therefore pro- that are statistically significant compared with the WT. Statistical signifi- long the activity of r in M. genitalium. In keeping with this idea, cance was assessed using a standard Student’s t test (p< 0.05). Downloaded from https://academic.oup.com/dnaresearch/advance-article-abstract/doi/10.1093/dnares/dsy011/4967678 by Ed 'DeepDyve' Gillespie user on 12 July 2018 20 r activation in Mycoplasma genitalium 9 Figure 7. Schematic representation of representative mating experiments. Qualitative assessment of recovery of double resistant mutants in mating experi- ments using M. genitalium strains carrying different antibiotic gene markers. (A) Mating experiments using donor strains (depicted as white mycoplasma cells) R R carrying the chloramphenicol (Cm ) and tetracycline (Tet ) resistance markers and recipient strains (depicted as light gray mycoplasma cells) carrying the puro- R 20 20 mycin (Pm ) resistance marker. Donor strains overexpressing r are indicated as ‘Upr .’ One of the donor strains used, carried a deletion of the recA gene (recA ). Unsuccessful DNA transfers are indicated as empty, dash-lined mycoplasma cells. (B) Mating experiments using a previous transconjugant mutant car- rying the chloramphenicol and puromycin resistance markers as a donor strain (depicted as a light gray mycoplasma cell) and a recipient strain (depicted as a dark gray mycoplasma cell) carrying the tetracycline resistance marker. Selection with tetracycline and chloramphenicol results in the isolation of a new trans- conjugant strain bearing the corresponding antibiotic markers and overexpressing r . Selection with puromycin and tetracycline allows the isolation of an- other transconjugant strain with the corresponding antibiotic markers but without the ectopic copy of r . All the transconjugant strains were genotyped by PCR and sequencing. 20 S we demonstrate that r positively autoregulates its own activity, polymerase complex. This step is critical for r activity as it aids over- which reinforces the participation of RrlAB in a positive feed-back coming the low afﬁnity of the alternative sigma factor for the RNA po- loop enabling r -activation. lymerase core enzyme as compared with the primary sigma particle. Sigma factors direct the activity of the RNA polymerase complex to Alternatively, the activity of the RrlAB could be related to anti-sigma speciﬁc promoter sequences. When a bacterium expresses simulta- factors, which control the availability of alternative sigma factors. In neously more than one sigma factor, these transcription factors compete this case, RrlAB could hamper de formation of a r -RNA polymerase to bind to the RNA polymerase enzyme. Bacteria use diverse strategies complex, facilitating the activity of r . Neither RrlA nor RrlB shows to ensure the concurrent action of different sigma factors and to enforce sequence homology to known anti-sigma factors, but these regulatory 36 39 transcription transition from the primary to alternative sigma factors. proteins usually exhibit low sequence conservation. On the basis of this notion, we hypothesize that RrlAB could be impor- In this study, we demonstrate that strains overexpressing r can 70 20 tant to redirect transcription from r to r -dependent promoters in act as donor cells for HGT. In mycoplasmas, HGT has only been de- M. genitalium. Speciﬁcally, these two small proteins could facilitate scribed in some ruminant species that encode integrative conjugative 20 40,41 binding of r to the RNA polymerase complex, which in turn could elements (ICE) or ICE-like sequences in their chromosomes. prevent the rapid degradation of the otherwise free r particles. The Recent studies have documented the transfer of large chromosomal proposed role for RrlAB as sigma auxiliary proteins that assist r in regions between different mycoplasma strains by homologous re- the activation of the recombination regulon is not unprecedented. combination. A similar mechanism was described years ago in Accordingly, the positive effect of RrlAB on r function resembles the Spiroplasma citri, a species phylogenetically related to the mycoplas- activity of the curli ﬁmbriae formation regulator Crl of E. coli and mas. Barroso and coworkers reported the recombination-dependent 37,38 Salmonella typhimurium. Crl is a small protein that binds to the al- chromosomal transfer of genetic traits conferring antibiotic resis- S S 34 ternative sigma factor r and enhances the formation of a r -RNA tance between spiroplasma cells. Remarkably, these DNA transfer Downloaded from https://academic.oup.com/dnaresearch/advance-article-abstract/doi/10.1093/dnares/dsy011/4967678 by Ed 'DeepDyve' Gillespie user on 12 July 2018 10 S. Torres-Puig et al. Meeting of the American Society for Microbiology held in Boston, events were independent of known mobile genetic elements and re- Massachusetts, in June 2016. lied on stable cell membrane fusions. Genetic ﬂuxes in wall-less bac- teria are not extraordinary, as similar chimeric genomes obtained by membrane fusions have been observed in L-forms of different bacte- 42 Conflict of interest ria. As M. genitalium seems to be devoid of ICE or ICE-like ele- None declared. ments, gene transfer could be catalyzed by intrinsic non-mobilizable factors encoded in the chromosome. At the present time, it is un- known whether genes unrelated to recombination under the control 20 20 Supplementary data of r are involved in HGT. However, several r -regulated genes code for proteins with putative transmembrane domains that could Supplementary data are available at DNARES online. be implicated in the establishment of cell-to-cell contacts and ulti- mately facilitate DNA exchange between cells. The ability to ex- Funding change DNA through inter- and intra-chromosomal recombination increases the potential to generate genetic diversity and the versatility This work was supported by Grants BIO2013-48704-R and BFU2013-50176- of the M. genitalium chromosome. EXP from the Ministerio de Economı´a y Competitividad. It is expected the r recombination pathway will be triggered by factors positioned hierarchically upstream from RrlAB in the References cascade of r -activation. Remarkably, we have strong evidence that r activation is inhibited or severely impaired in non-adherent 1. Finlay, B.B. and McFadden, G. 2006, Anti-immunology: evasion of the strains. Indeed, transcriptional analysis of a MG_191 mutant re- host immune system by bacterial and viral pathogens, Cell, 124, 767–82. veals that the ncRNA-3/4 is downregulated 13-fold as compared 2. Van Der Woude, M.W. and Ba ¨ umler, A.J. 2004, Phase and antigenic vari- ation in bacteria, Clin. Microbiol. Rev., 17, 581–611. with the wild-type strain (Fig. 1A), which denotes a decreased r 3. Vink, C., Rudenko, G. and Seifert, H.S. 2012, Microbial antigenic varia- activity. Of note, the main cytadhesins of M. genitalium are recip- tion mediated by homologous DNA recombination, FEMS Microbiol. rocally stabilized and neither P140 nor P110 is expressed in the 19 20 Rev., 36(5), 917–48. MG_191 mutant. As the biological role of r seems intimately 4. Koomey, M., Gotschlich, E.C., Robbins, K., Bergstro ¨ m, S. and Swanson, related to the generation of antigenic variants, it is reasonable to J. 1987, Effects of recA mutations on pilus antigenic variation and phase think that r activity will be less relevant in strains lacking these transitions in Neisseria gonorrhoeae, Genetics, 117, 391–8. two immunodominant proteins. Of note, non-adherent mutants of 5. Mehr, I.J. and Seifert, H.S. 1998, Differential roles of homologous recom- M. genitalium arise spontaneously at relatively high-rates by means bination pathways in Neisseria gonorrhoeae pilin antigenic variation, 19,43 of an exquisite PhV mechanism. For this reason, it is thought DNA transformation and DNA repair, Mol. Microbiol., 30, 697–710. that non-adherent cells play an important role in infection and sur- 6. Centurion-Lara, A., LaFond, R.E., Hevner, K., et al. 2004, Gene conver- sion: a mechanism for generation of heterogeneity in the tprK gene of vival of M. genitalium within the host. The inhibition of the r Treponema pallidum during infection, Mol. Microbiol., 52, 1579–96. system in non-adherent mutants could represent an attempt to 7. Burgos, R., Wood, G.E., Young, L., Glass, J.I. and Totten, P.A. 2012, arrest this phenotype, which can often be reversed by homologous RecA mediates MgpB and MgpC phase and antigenic variation in recombination. Whether P140 or other adherence-related factors Mycoplasma genitalium, but plays a minor role in DNA repair, Mol. are directly involved in the activation of r , remains to be investi- Microbiol., 85, 669–83. gated. On the other hand, the prominent upregulation of the 8. Burgos, R. and Totten, P.A. 2014, Characterization of the operon encoding ncRNA-3/4 transcript as well as other non-coding RNAs by r is the holliday junction helicase RuvAB from Mycoplasma genitalium and its puzzling. In the case of the ncRNA-1, which carries homologous re- role in mgpB and mgpC gene variation, J. Bacteriol., 196, 1608–18. gions to the cytadhesins genes in the antisense orientation, we hy- 9. Stohl, E.A. and Seifert, H.S. 2001, The recX gene potentiates homologous pothesize that the activation of this non-coding RNA could recombination in Neisseria gonorrhoeae, Mol. Microbiol., 40, 1301–10. 10. Skaar, E.P., Lazio, M.P. and Seifert, H.S. 2002, Roles of the recJ and recN interfere with the translation of P140 and P110. Silencing of the genes in homologous recombination and DNA repair pathways of main cytadhesins could hinder the recognition of M. genitalium Neisseria gonorrhoeae, J. Bacteriol., 184, 919–27. cells by the host immune system during a coordinated response to 11. Black, C.G., Fyfe, J.A.M. and Davies, J.K. 1998, Absence of an SOS-like boost the generation of antigenic variants. system in Neisseria gonorrhoeae, Gene, 208, 61–6. Overall, in this work we have identiﬁed two factors that regulate 12. Fraser, C.M., Gocayne, J.D., White, O., et al. 1995, The minimal gene r activity and are key to generate AnV catalyzed by this alternative complement of Mycoplasma genitalium, Science, 270, 397–403. sigma factor. Moreover, the discovery of a form of HGT in 13. Fraser, C.M. 1998, Complete genome sequence of Treponema pallidum, M. genitalium opens a new avenue in the understanding of dissemi- the syphilis spirochete, Science, 281, 375–88. nation of effective antigenic variants and provides some important 14. Schook, P.O.P., Stohl, E.A., Criss, A.K. and Seifert, H.S. 2011, The clues as to the rapid emergence of antibiotic resistance in urogenital DNA-binding activity of the Neisseria gonorrhoeae LexA orthologue NG1427 is modulated by oxidation, Mol. Microbiol., 79, 846–60. pathogens. 15. Gruenig, M.C., Stohl, E.A., Chitteni-Pattu, S., Seifert, H.S. and Cox, M.M. 2010, Less is more: neisseria gonorrhoeae RecX protein stimulates recombination by inhibiting RecA, J. Biol. Chem., 285, 37188–97. Acknowledgements 16. Peterson, S.N., Bailey, C.C., Jensen, J.S., et al. 1995, Characterization of re- We are grateful to the staff of the Servei de Geno ` mica I Bioinforma ` tica (UAB) petitive DNA in the Mycoplasma genitalium genome: possible role in the gen- and the Genomics Unit (CRG) for performing Sanger sequencing and RNA- eration of antigenic variation, Proc. Natl. Acad. Sci. U. S. A., 92, 11829–33. Seq analysis, respectively. We are also very grateful to Dr Eva Yus (CRG) for 17. Iverson-Cabral, S.L., Astete, S.G., Cohen, C.R. and Totten, P.A. 2007, her valuable advice on performing RNA preparation and RNA-Seq analysis. mgpB and mgpC sequence diversity in Mycoplasma genitalium is gener- ST and CM want to acknowledge a PIF fellowship from the Universitat ated by segmental reciprocal recombination with repetitive chromosomal Auto ` noma de Barcelona. This work was presented in part at the ﬁrst Microbe sequences, Mol. Microbiol., 66, 55–73. Downloaded from https://academic.oup.com/dnaresearch/advance-article-abstract/doi/10.1093/dnares/dsy011/4967678 by Ed 'DeepDyve' Gillespie user on 12 July 2018 20 r activation in Mycoplasma genitalium 11 18. Ma, L., Jensen, J.S., Myers, L., et al. 2007, Mycoplasma genitalium: an ef- 31. Love, M.I., Huber, W. and Anders, S. 2014, Moderated estimation of fold ﬁcient strategy to generate genetic variation from a minimal genome, Mol. change and dispersion for RNA-seq data with DESeq2, Genome Biol., 15, Microbiol., 66, 220–36. 550. 19. Burgos, R., Pich, O.Q., Ferrer-Navarro, M., Baseman, J.B., Querol, E. 32. Dillies, M.-A., Rau, A., Aubert, J., et al. 2013, A comprehensive evalua- and Pinol, J. 2006, Mycoplasma genitalium P140 and P110 cytadhesins tion of normalization methods for Illumina high-throughput RNA se- are reciprocally stabilized and required for cell adhesion and quencing data analysis, Brief. Bioinform., 14, 671–83. terminal-organelle development, J. Bacteriol., 188, 8627–37. 33. Zhang, W. and Baseman, J.B. 2014, Functional characterization of osmot- 20. Wood, G.E., Iverson-Cabral, S.L., Patton, D.L., Cummings, P.K., ically inducible protein C (MG_427) from Mycoplasma genitalium, Cosgrove Sweeney, Y.T. and Totten, P.A. 2013, Persistence, immune re- J. Bacteriol., 196, 1012–9. sponse, and antigenic variation of Mycoplasma genitalium in an experi- 34. Barroso, G. and Labarere, J. 1988, Chromosomal gene transfer in mentally infected pig-tailed macaque (Macaca nemestrina), Infect. Spiroplasma citri, Science, 241, 959–61. Immun., 81, 2938–51. 35. Dordet-Frisoni, E., Sagne ´ , E., Baranowski, E., et al. 2014, Chromosomal 21. Wood, G.E., Patton, D.L., Cummings, P.K., Iverson-Cabral, S.L. and transfers in mycoplasmas: when minimal genomes go mobile, mBio, 5, Totten, P.A. 2017, Experimental infection of pig-tailed macaques e01958–14. (Macaca nemestrina) with Mycoplasma genitalium, Infect. Immun., 85, 36. Osterberg, S., Peso-Santos, T.D. and Shingler, V. 2011, Regulation of al- e00738–16. ternative sigma factor use, Annu. Rev. Microbiol., 65, 37–55. 22. Ma, L., Mancuso, M., Williams, J.A., et al. 2014, Extensive variation and 37. Gaal, T., Mandel, M.J., Silhavy, T.J. and Gourse, R.L. 2006, Crl facili- rapid shift of the MG192 Sequence in Mycoplasma genitalium strains tates RNA polymerase holoenzyme formation, J. Bacteriol., 188, from patients with chronic infection, Infect. Immun., 82, 1326–34. 7966–70. 23. Ma, L., Jensen, J.S., Mancuso, M., Myers, L. and Martin, D.H. 2016, 38. Typas, A., Barembruch, C., Possling, A. and Hengge, R. 2007, Stationary Kinetics of genetic variation of the Mycoplasma genitalium MG192 gene phase reorganisation of the Escherichia coli transcription machinery by in experimentally infected chimpanzees, Infect. Immun., 84, 747–53. Crl protein, a ﬁne-tuner of rS activity and levels, EMBO J., 26, 1569–78. 24. Torres-Puig, S., Broto, A., Querol, E., Pinol, ~ J. and Pich, O.Q. 2015, A 39. Paget, M.S. 2015, Bacterial sigma factors and anti-sigma factors: struc- novel sigma factor reveals a unique regulon controlling cell-speciﬁc recom- ture, function and distribution, Biomolecules, 5, 1245–65. bination in Mycoplasma genitalium, Nucleic Acids Res., 43, 4923–36. 40. Tardy, F., Mick, V., Dordet-Frisoni, E., et al. 2015, Integrative conjuga- 25. Burgos, R. and Totten, P.A. 2014, MG428 is a novel positive regulator of tive elements are widespread in ﬁeld isolates of Mycoplasma Species path- recombination that triggers mgpB and mgpC gene variation in ogenic for ruminants, Appl. Environ. Microbiol., 81, 1634–43. Mycoplasma genitalium, Mol. Microbiol., 94, 290–306. 41. Dordet Frisoni, E., Marenda, M.S., Sagne ´ , E., et al. 2013, ICEA of 26. Sambrook, J., Fritsch, E.F. and Maniatis, T. 1989, Molecular cloning: a Mycoplasma agalactiae: a new family of self-transmissible integrative ele- laboratory manual. Cold Spring Harbor Laboratory Press: New York. ments that confers conjugative properties to the recipient strain, Mol. 27. Langmead, B. and Salzberg, S.L. 2012, Fast gapped-read alignment with Microbiol., 89, 1226–39. Bowtie 2, Nat. Methods, 9, 357–9. 42. Errington, J. 2013, L-form bacteria, cell walls and the origins of life, 28. Li, H., Handsaker, B., Wysoker, A., et al. 2009, The sequence align- Open Biol., 3, 120143. ment/map format and SAMtools, Bioinformatics, 25, 2078–9. 43. Mernaugh, G.R., Dallo, S.F., Holt, S.C. and Baseman, J.B. 1993, 29. Liao, Y., Smyth, G.K. and Shi, W. 2014, featureCounts: an efﬁcient gen- Properties of adhering and nonadhering populations of Mycoplasma geni- eral purpose program for assigning sequence reads to genomic features, talium, Clin. Infect. Dis., S1, S69–78. Bioinformatics, 30, 923–30. 44. Galperin, M.Y., Makarova, K.S., Wolf, Y.I. and Koonin, E.V. 2015, 30. Anders, S. and Huber, W. 2010, Differential expression analysis for se- Expanded Microbial genome coverage and improved protein family anno- quence count data, Genome Biol., 11, R106. tation in the COG database, Nucleic Acids Res., 43, D261–9. Downloaded from https://academic.oup.com/dnaresearch/advance-article-abstract/doi/10.1093/dnares/dsy011/4967678 by Ed 'DeepDyve' Gillespie user on 12 July 2018
DNA Research – Oxford University Press
Published: Apr 11, 2018
It’s your single place to instantly
discover and read the research
that matters to you.
Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.
All for just $49/month
Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly
Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.
Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.
Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.
All the latest content is available, no embargo periods.
“Hi guys, I cannot tell you how much I love this resource. Incredible. I really believe you've hit the nail on the head with this site in regards to solving the research-purchase issue.”Daniel C.
“Whoa! It’s like Spotify but for academic articles.”@Phil_Robichaud
“I must say, @deepdyve is a fabulous solution to the independent researcher's problem of #access to #information.”@deepthiw
“My last article couldn't be possible without the platform @deepdyve that makes journal papers cheaper.”@JoseServera