Edited by Norman R. Pace, University of Colorado, Boulder, CO, and approved October 8, 2009
↵1D.J. and C.M.S. contributed equally to this work. (received for review August 10, 2009)
Methanosarcina mazei and related mesophilic archaea are the only organisms fermenting acetate, methylamines, and methanol to methane and carbon dioxide, contributing significantly to greenhouse gas production. The biochemistry of these metabolic processes is well studied, and genome sequences are available, yet little is known about the overall transcriptional organization and the noncoding regions representing 25% of the 4.01-Mb genome of M. mazei. We present a genome-wide analysis of transcription start sites (TSS) in M. mazei grown under different nitrogen availabilities. Pyrosequencing-based differential analysis of primary vs. processed 5′ ends of transcripts discovered 876 TSS across the M. mazei genome. Unlike in other archaea, in which leaderless mRNAs are prevalent, the majority of the detected mRNAs in M. mazei carry long untranslated 5′ regions. Our experimental data predict a total of 208 small RNA (sRNA) candidates, mostly from intergenic regions but also antisense to 5′ and 3′ regions of mRNAs. In addition, 40 new small mRNAs with ORFs of ≤30 aa were identified, some of which might have dual functions as mRNA and regulatory sRNA. We confirmed differential expression of several sRNA genes in response to nitrogen availability. Inspection of their promoter regions revealed a unique conserved sequence motif associated with nitrogen-responsive regulation, which might serve as a regulator binding site upstream of the common IIB recognition element. Strikingly, several sRNAs antisense to mRNAs encoding transposases indicate nitrogen-dependent transposition events. This global TSS map in archaea will facilitate a better understanding of transcriptional and posttranscriptional control in the third domain of life.
Methanosarcina mazei strain Gö1 is a representative methane-producing archaeon of ecologic significance because of its role in biogenic methane production in various anaerobic habitats on Earth (1). The genome sequences of M. mazei and its close relatives Methanosarcina acetivorans and Methanosarcina barkeri have recently become available and have revealed an unexpected low proportion of coding region (74.2% in M. acetivorans, 75.15% in M. mazei, and 79.2% in M. barkeri) (2–4). The biochemical basis of methanogenesis has been analyzed in considerable detail (5, 6). In contrast, little is known about global regulatory networks that ensure survival in periods of nutrient starvation or stress in this important group of archaea. More than 50 predicted transcriptional regulators were annotated in the genome of M. mazei. Strikingly, most of them seem to be closely related to bacterial proteins (2), whereas the basic components of the archaeal transcription and translation machineries generally are more similar to those of eukaryotes (7). A recent genetic study (8) discovered the first global transcriptional regulator of M. mazei, the nitrogen regulator NrpR, which was experimentally demonstrated to globally repress transcription of nitrogen fixation and assimilation genes in response to the nitrogen source.
Besides transcriptional regulation, small regulatory RNAs (sRNAs) have increasingly been implicated in a variety of adaptive cellular responses to biotic and abiotic stresses and development in bacteria (9, 10). In comparison, there are few studies on the role of sRNAs in archaea. Computational and experimental analysis discovered C/D-box and H/ACA-box small nucleolar RNAs (snoRNAs) with predicted functions in ribosomal RNA maturation in Sulfolobus sulfataricus, S. acidocaldarius, Archaeaglobus fulgidus, and Pyrococcus (11–14), as well as stable antisense RNAs and sRNA species from intergenic regions (IGRs), which are likely to be involved in posttranscriptional control of gene expression (15–18). More recently, several dozen sRNAs were predicted in the high salt-adapted species Haloferax volcanii (19).
Although all previous experimental approaches in archaea relied on cDNA cloning followed by Sanger sequencing, high-throughput sequencing of cDNA [RNA-seq (20)] has now revolutionized transcriptome analysis and sRNA discovery in many organisms (21–23). In the present work we used a recently developed differential RNA-seq method selective for newly initiated transcripts. We present a genome-wide map of transcriptional start sites (TSS) and sRNA output in M. mazei and provide insight into transcriptional regulation in response to nitrogen in this organism.
We analyzed cDNA libraries derived from M. mazei cultures growing under different nitrogen availabilities [i.e., from cells growing exponentially under nitrogen sufficiency (NH4+; NS libraries) or under nitrogen-fixing conditions with molecular nitrogen as the sole nitrogen source (N2; NF libraries)]. For each of the 2 conditions, 2 differential cDNA libraries were constructed. The NS(−) and NF(−) libraries were generated from the original RNA pool and covered both primary (carrying 5′ triphosphate group) and processed (5′ monophosphate) transcripts; whereas in the cognate NS(+) and NF(+) libraries, newly initiated transcripts with primary 5′ end were enriched by enzymatic treatment. After library construction, ≈185,000 cDNAs from ammonium conditions (NS libraries) and ≈43,000 cDNAs from nitrogen-fixing conditions (NF libraries) were sequenced. The obtained cDNA reads varied in length from 1 to 280 bp, and disregarding cDNAs <18 bp, between 85% and 94% of them were unequivocally mapped to the M. mazei genome. The number of cDNA hits for each nucleotide position for both strands were calculated and the data visualized using the Integrated Genome Browser (Affymetrix). Of the cDNAs, 26.4% (NF) and 31.6% (NS) mapped to intergenic regions (IGRs) of the genome in the enriched (+) libraries, as compared with 21% and 28%, respectively, in the cognate nonenriched (−) libraries, indicating depletion of processed rRNAs and tRNAs in favor of IGR transcripts. Furthermore, a strong enrichment of cDNAs clustering toward the TSS of several mRNA genes analyzed in previous studies was observed [e.g., of grpE and glnA1 (24, 25)]; the 5′ ends of the grpE and glnA1 cDNA clusters exactly match the previously identified TSS (Table S1). Fig. 1 illustrates TSS identification and nitrogen source-dependent transcription regulation of glnA1 (encoding glutamine synthetase). Nitrogen-responsive regulation of glnA1 was previously shown by quantitative RT-PCR analysis (25) and is verified here by Northern blot probing (Fig. 1). A second, constitutive TSS associated with glnA1 was found upstream of the nitrogen-regulated TSS. Overall, 876 potential TSS were identified, confirming transcription (coverage of >5 cDNAs) for 586 of the originally annotated 3,371 ORFs. Inspection of cDNAs from noncoding regions identified 42 new ORFs encoding small proteins (30–99 aa) (Table S2), which had failed identification in the original genome sequence analysis owing to the criteria of automated annotation (2). Overall, the majority (60%) of all primary transcripts detected here are differentially transcribed in response to nitrogen; 287 transcripts were exclusively present under ammonium conditions and 253 exclusively under nitrogen fixation conditions (see Tables S2–S6).
Visualization of the clone distribution of cDNAs mapped to the chromosomal glnA1 region. An extraction of the screenshot of the Integrated Genome Browser (Affymetrix) of the mapped cDNAs is shown for nitrogen fixation (NF+) and ammonium conditions (NS+) of the M. mazei glnA1 region. The y axis indicates the relative score of clone numbers per nucleotide. The nitrogen-regulated TSS and a potential second TSS of glnA1 are indicated. The respective Northern blot analysis is shown (Right).
Binding boxes for general transcription factors can be predicted in an appropriate distance to the TSS of 53% [IIB recognition element (BRE)] and 75% (TATA-box) of all primary transcripts, with the consensus sequence depicted in Fig. S1. For ≈60% of all detected mRNAs a ribosome binding site with the consensus sequence AGGAGG was identified, consistent with previous computational predictions (26). Most interestingly, 521 mRNAs contained long 5′ UTRs up to 500 nt, with an average size of 150–200 nt (Table S3). Such long 5′ UTRs may be targeted by posttranscriptional regulators. Sixty-six mRNAs possess 5′ UTRs of <10 nt and are thus designated leaderless mRNAs (Table S4).
Our analysis predicted 208 candidate sRNAs from IGRs, including such that overlapped in antisense orientation with 5′ regions (n = 43) or 3′ regions (n = 5) of mRNAs, and were thus designated cis-encoded antisense RNAs (asRNAs). Strikingly, 36 sRNA candidates were exclusively detected under nitrogen sufficiency, whereas 99 were solely present under nitrogen-fixing conditions, indicating differential transcription in response to nitrogen (Table S5).
In addition, 40 sRNAs candidates located in IGRs contain very short ORFs potentially encoding peptides <30 aa, 15 of which are preceded by a consensus ribosome binding site. A high number of these small ORFs are conserved in other Methanosarcina strains (Table S6), and several exist in multiple homologues in the M. mazei genome. Most of the associated peptides >20 aa contained a transmembrane helix, indicating membrane association. For 16 candidates the flanking RNA region upstream of the ORF also showed high conservation within the Methanosarcina strains, suggesting a potential dual function as both mRNA and RNA regulator. Interestingly, 7 of the short ORF loci were associated with a cis-encoded antisense RNA overlapping the 5′ (n = 5) or 3′ (n = 2) ORF region (Fig. 2).
Transcription of spRNA09 and sRNA60. (A) Chromosomal localization; (B) cDNA clone distribution for nitrogen fixation (NF+) and ammonium conditions (NS+); (C) confirmatory Northern blot analysis; the calculated fold induction (NF vs. NS) is given below the blot.
For a randomly chosen ≈400-kb fragment of the chromosome (nt 11,700–407,000) all IGR transcripts identified by the pyrosequencing approach were subjected to confirmatory Northern blot analysis. Of 36 predicted transcripts (sRNAs and new ORFs), 16 could be verified (Table S7), indicating that a transcript has to be covered by at least 5 cDNAs to be detectable on Northern blots. Outside the above selected region, an additional 41 sRNA candidates with high cDNA coverage were analyzed by Northern blot and verified with few exceptions (Table S8). For >85% the transcript length determined was consistent with the length predicted by cDNA coverage; in several cases, however, the primary transcript seemed to be processed into 2 or 3 smaller products. In addition, the Northern blot analysis confirmed differential expression (≥1.5-fold regulation) of 18 sRNAs, as summarized in Tables S7 and S8 and exemplarily depicted in Fig. 3.
Expression of selected sRNA candidates in response to nitrogen. Total RNA was isolated from M. mazei grown under nitrogen fixation (NF) and ammonium conditions (NS) and subjected to Northern blot analysis. Blots showing stable sRNAs transcripts, the lower line the 5S rRNA transcripts of the respective RNA preparations. The calculated fold induction (NF vs. NS) of at least 2 independent experiments is given below the blots.
We inspected the 5′ regions including 500 bp upstream of TSS of sRNA candidate genes with significantly higher transcript levels under nitrogen limitation (cDNA libraries). Two contained the operator sequence of the global nitrogen repressor (NrpR) in their promoter regions (Fig. 4), overlapping TSS (sRNA154) or the BRE (sRNA159). In addition, a unique conserved putative binding site for a transcriptional regulator was identified in the promoter region of sRNA030, sRNA089, sRNA090, and sRNA106, consisting of 2 motifs located up- and downstream of the BRE and TATA-box (AGGAGGCA-N27-AAAGCTA). Genome-wide searches for this sequence motif identified 8 additional sRNA candidate genes covered by the cDNA libraries with a similar motif in their promoter region (see Fig. 5 for consensus sequence), 6 of which showed significantly higher transcript levels in the NF cDNA libraries, indicating association with nitrogen-responsive regulation.
Promoter region of sRNA154 (A) and sRNA159 (B). The NrpR binding boxes are indicated by gray boxes.
The exceptionally large number of insertion sequence (IS) elements and transposase genes in the M. mazei genome (2) has been considered evidence that the high content (30%) of M. mazei ORFs with closest homologues in the bacterial domain were acquired by lateral gene transfer. Our cDNA libraries covered 13 mRNAs of transposase genes, validating expression of these mobility loci. Intriguingly, the 5′ UTRs of 6 of these transposase genes overlap with an asRNA candidate. Northern blot analysis discovered differential transcription of several transposase asRNAs in response to nitrogen (e.g., asRNA036) (Figs. 3 and 6), indicating posttranscriptional (antisense-mediated) control of transposases and therefore nitrogen-dependent regulation of potential transposition events in M. mazei. Thus, it is tempting to speculate that lateral gene transfer via transposition events may occur in response to changes of environmental conditions.
Transcription of MM2686 and the corresponding cis-encoded asRNA36. cDNA clone distribution for nitrogen fixation (NF+) and ammonium (NS+) conditions of the M. mazei transposase gene (MM2686) region is shown for the forward and reverse strand.
Our TSS map covering ≈20% of all predicted ORFs of this ecologically important archaeal model organism M. mazei provides remarkable insight into general archaeal transcription and global regulation in response to nitrogen or general stress. The comprehensive information will facilitate future studies elucidating general aspects of the archaeal transcription and translational machinery (promoter elements, binding sites for transcriptional regulators, and mechanisms of translation initiation in methanoarchaea) and of regulation in response to (nitrogen) stress. Given the newly identified ORFs, the protein-coding gene number of M. mazei increased by ≈1.3% (from 3,371 previously annotated). Several unique features of the transcriptional and posttranscriptional regulation, as well as of the genome organization, of M. mazei have been discovered in this study.
First, the genome-wide identification of 876 TSS demonstrates that the vast majority of the covered mRNAs contained a long 5′ UTR (up to 500 nt) and a ribosome binding site. Long 5′ UTRs in methanoarchaea were completely unexpected, because previous genomic and experimental analyses had predicted archaeal mRNAs to either be leaderless or possess rather short 5′ UTRs (27). In addition, a recent systematic screen showed that the majority of haloarchaeal transcripts might be leaderless (28), and a very recent RNA-seq analysis (52) demonstrated that leaderless mRNAs are the rule in S. solfataricus P2. In contrast, the high number of long 5′ UTRs argue for extensive posttranscriptional regulation at 5′ UTRs in methanoarchaea (e.g., at the level of transcript stability) by regulatory proteins or RNAs, or through riboswitches. To date, evidence for regulatory functions of archaeal UTRs has been limited to translational regulation by 3′ UTR and short 5′ UTR in haloarchaea (29), and in silico prediction of 1 riboswitch candidate in Thermoplasma spp. resembling bacterial THI-element responsible for thiamine-mediated regulation (30).
Only <10% of the 5′ UTRs identified in M. mazei here are present upstream of bacterial-like genes assumed to have been acquired by lateral gene transfer (2), and almost all of them contain a BRE and a TATA-box in their promoters. Collectively, these findings strongly indicate that 5′ UTRs are specific for methanoarchaeal ORFs, and the few bacterial-like ORFs with long 5′ UTRs have been placed under the control of an archaeal promoter after acquisition.
Second, we identified 248 sRNA candidates including cis-encoded antisense RNAs from IGRs scattered across the genome. Of 77 randomly selected sRNA candidates, 57 have been verified on Northern blots (summarized in Fig. 7), allowing us to estimate a total number of ≈180 sRNAs on the basis of the empirical threshold (5 cDNAs per verifiable transcript). Although all genes coding for the core small nucleolar ribonucleoprotein (snoRNP) are present in M. mazei, and 34 snoRNAs were in silico predicted [University of California-Santa Cruz Archaeal Genome Browser (31)], only 1 of our sRNA candidates (spRNA26) was identified as a putative snoRNA using the Snoscan server 1.0 (11) and SnoReport (32); this seemingly low coverage of snoRNAs might be due to the detection limit or our protocol for cDNA library construction. Comparative genome analysis demonstrated that 30% of the identified asRNA and 21% of sRNA candidates in M. mazei were conserved in all 3 Methanosarcina species (Table S5). Less than 3% of the sRNAs candidates showed homology to bacterial IGRs, and only 2 asRNAs are antisense to mRNAs of bacterial-like ORFs in M. mazei, which is a striking result considering that ≈30% of the identified ORFs in M. mazei are proposed to be acquired by lateral gene transfer. Thus, it is tempting to speculate that the majority of sRNAs in M. mazei function as regulators of archaeal gene expression, unlike in other archaea in which ≈30% of these sRNA candidates were snoRNAs (16, 17).
Genomic map showing the localization of the verified IGR transcripts. The 400-kb fragment (nt 11,700–407,000) thoroughly analyzed is enlarged. Red, new ORFs; green, spRNAs; dark blue, asRNAs; and light blue, sRNAs.
The expression of 135 of the sRNA candidates is affected by nitrogen availability, as validated for 18 sRNAs on Northern blots. To date, there are very few examples for sRNAs indirectly involved in nitrogen regulation in bacteria (33). A single sRNA, GlmY, has been demonstrated to be transcribed by the σ54-RNA polymerase in Escherichia coli, but independently of the global nitrogen transcription regulator NtrC (34). Consistent with nitrogen-regulated transcription of ORFs by the global nitrogen regulator NrpR in M. mazei, 2 of the verified nitrogen-regulated sRNAs contained the NrpR operator in their promoters (8, 35). Furthermore, a new motif (box I), which seems to be associated with nitrogen-responsive regulation, was detected up- and downstream of the BRE and TATA-box of 12 sRNA candidates (Fig. 5). On the basis of our observation that the BRE of those sRNA candidates may have low affinity to transcription factor IIB (TFB), we hypothesize that a yet-unknown activator protein is binding to box I. Binding of this yet-to-be identified factor would then recruit the general transcription factors, TFB and TATA-box binding protein, and consequently RNA polymerase. Because the motif has not been identified in promoters of mRNA genes, it might be recognized by an sRNA-specific regulator. Overall, the strong conservation of the identified sRNA candidates in Methanosarcinales, and the large number of differentially transcribed sRNAs in response to nitrogen, argue for a prominent regulatory function of sRNAs in the nitrogen and in general stress responses of methanoarchaea. Thus, together with the predictions by Straub et al. (19) in Haloarchaea, sRNA-mediated regulation might play a much more prominent role in archaea than appreciated.
We identified a high number of asRNA candidates in opposite orientation to transposase transcripts in M. mazei, indicating that transposon mobility might be regulated by an RNA-antisense mechanism as previously demonstrated for the Tn10 and Tn30 transposons in E. coli (36, 37). Recently, 8 transposase asRNAs were identified in S. sulfataricus, suggesting that this archaeon is using a similar strategy to control mobility of its multiple insertion elements at the posttranscriptional level (17). Most strikingly, several of the transposon asRNAs identified here in M. mazei are differentially expressed in response to nitrogen, predicting a potential link between nitrogen availability and transposition events in M. mazei. Regulation of transposition—well studied in bacteria but much less so in archaea—was known to occur mostly in response to stresses (e.g., UV light or temperature) and to take place at the transcriptional and posttranscriptional level, as well as by modulating transposase activity (38). However, there was only 1 prominent example for direct control of transposase gene transcription in response to nutrient starvation, by σS in Pseudomonas putida (39). Given our observations, M. mazei seems to lend itself as a model to elucidate the predicted relationship of transposition events and nitrogen availability. Moreover, taking into account the fluctuations of nitrogen availability in multiple ecosystems, nitrogen-stimulated genetic exchange mediated by transposons might have pronounced ecologic consequence.
Finally, in recent years, small ORFs of a size too short for automated genome annotation (usually <150 nt) have been discovered at a staggering rate in various bacterial genomes, including E. coli (40), Pseudomonas aeruginosa (41), and marine cyanobacteria (42). In addition, many of the corresponding small peptides were experimentally validated in E. coli (43, 44). Our analysis in M. mazei identified small ORFs in 40 sRNA candidates (Table S6), the majority of which is conserved in the 3 Methanosarcina strains. Some of those sRNAs might have a dual function, given that the flanking 5′ RNA regions of the small ORFs also showed conservation. At present, no such small peptide has been demonstrated in M. mazei; consequently, it remains to be shown whether the oligopeptide and/or the noncoding part of the respective sRNA are functional.
Dual function of an sRNA was originally identified in Staphylococcus aureus; RNAIII encodes a 26-aa delta-hemolysin peptide and also acts as a translational regulator of trans-encoded virulence and transcription factor mRNAs (45, 46). Moreover, the 5′ region of the ≈220-nt E. coli SgrS RNA encodes a functional peptide, whereas the 3′ region contains a regulatory RNA domain that targets ptsG mRNA by base pairing (47). Besides such potential dual functions, some of the short ORFs identified in M. mazei were associated with a cis-encoded asRNA overlapping the 5′ or 3′ region of the short ORF (Fig. 2). Because a subset of bacterial cis-encoded antisense RNAs promotes degradation and/or represses translation of mRNAs that are toxic at high levels [so-called type I toxin–antitoxins (48)], it is tempting to speculate that the identified mRNA–asRNA pairs in M. mazei are archaeal examples for type I toxin–antitoxins.
M. mazei strain Gö1 was grown under anaerobic conditions at 37 °C with an atmosphere of 80% N2 plus 20% CO2 in 70-mL sealed tubes in minimal medium that contained 150 mM methanol plus 40 mM acetate as carbon source and 10 mM ammonium chloride (N sufficiency); for nitrogen-fixing conditions the gas atmosphere served as sole nitrogen source (25). Cultures were grown until cells reached a turbidity of 0.18–0.21 (N fixation) or 0.5–0.6 (N sufficiency) at 600 nm, which corresponds to the respective midexponential growth phase. Cells were harvested at 4 °C and RNA isolated by phenol extraction followed by DNase I treatment (49). Before cDNA construction, differential transcription of the glnK1 gene was verified by quantitative RT-PCR as described previously (25).
cDNA libraries were prepared and analyzed on a Roche FLX sequencer as previously described (22). In addition, we used a unique treatment protocol to take into account the 5′ triphosphate end characteristic for primary transcripts (most mRNAs and sRNAs), whereas processed RNA including abundant rRNA and tRNA have 5′ monophosphate (5′P) ends. The details of this protocol, which includes treatment with Terminator 5′P-dependent exonuclease (Epicenter) to deplete processed RNAs, will be published elsewhere. For each library, graphs representing the number of mapped reads per nucleotides were calculated and visualized using the Integrated Genome Browser software from Affymetrix. TSS were detected by higher cDNA coverage of the 5′ end of a given RNA in the library constructed with nuclease-treated RNA. The cDNA sequencing raw data are available at our institution's web site (www.uni-kiel.de/mikrobio/data.html).
RNA was separated in 6% Tris-borate-EDTA polyacrylamide (PAA) gels containing 7 M urea, or in agarose gels. Ten micrograms (PAA gel) or 20 μg (agarose gel) RNA were loaded per lane; pUC Marker Mix 8 (Fermentas) served as a size marker. After separation, RNA was transferred onto HybondXL membranes (GE Healthcare) by electroblotting and cross-linked to the membrane. Membranes were prehybridized in Rapid-hyb buffer (GE Healthcare) at 42 °C, followed by hybridization with 10 pmol [γ-32P]-ATP end-labeled oligodeoxynucleotides (Table S9) for 2 h. After washing 3 times for 15 min in 5×, 1×, and 0.5× SSC–0.1% SDS solutions (42 °C), signals were visualized on a phosphorimager (FLA-5000 Series, Fuji) and quantified with AIDA software (Raytest).
This research was supported by the Deutsche Forschungsgemeinschaft as part of the priority program SPP 1258, “Sensory and Regulatory RNAs in Prokaryotes.”
Author contributions: J.V. and R.A.S. designed research; D.J., C.M.S., and C.E. performed research; J.V. contributed new reagents/analytic tools; D.J., C.M.S., J.T., and R.A.S. analyzed data; and R.A.S. wrote the paper.
The authors declare no conflict of interest.
This article is a PNAS Direct Submission.
This article contains supporting information online at www.pnas.org/cgi/content/full/0909051106/DCSupplemental.
