Differential RNA Sequencing Implicates Sulfide as the Master Regulator of S⁰ Metabolism in Chlorobaculum tepidum and Other Green Sulfur Bacteria

ABSTRACT

The green sulfur bacteria (Chlorobiaceae) are anaerobes that use electrons from reduced sulfur compounds (sulfide, S⁰, and thiosulfate) as electron donors for photoautotrophic growth. Chlorobaculum tepidum, the model system for the Chlorobiaceae, both produces and consumes extracellular S⁰ globules depending on the availability of sulfide in the environment. These physiological changes imply significant changes in gene regulation, which has been observed when sulfide is added to Cba. tepidum growing on thiosulfate. However, the underlying mechanisms driving these gene expression changes, i.e., the specific regulators and promoter elements involved, have not yet been defined. Here, differential RNA sequencing (dRNA-seq) was used to globally identify transcript start sites (TSS) that were present during growth on sulfide, biogenic S⁰, and thiosulfate as sole electron donors. TSS positions were used in combination with RNA-seq data from cultures growing on these same electron donors to identify both basal promoter elements and motifs associated with electron donor-dependent transcriptional regulation. These motifs were conserved across homologous Chlorobiaceae promoters. Two lines of evidence suggest that sulfide-mediated repression is the dominant regulatory mode in Cba. tepidum. First, motifs associated with genes regulated by sulfide overlap key basal promoter elements. Second, deletion of the Cba. tepidum 1277 (CT1277) gene, encoding a putative regulatory protein, leads to constitutive overexpression of the sulfide:quinone oxidoreductase CT1087 in the absence of sulfide. The results suggest that sulfide is the master regulator of sulfur metabolism in Cba. tepidum and the Chlorobiaceae. Finally, the identification of basal promoter elements with differing strengths will further the development of synthetic biology in Cba. tepidum and perhaps other Chlorobiaceae.

IMPORTANCE Elemental sulfur is a key intermediate in biogeochemical sulfur cycling. The photoautotrophic green sulfur bacterium Chlorobaculum tepidum either produces or consumes elemental sulfur depending on the availability of sulfide in the environment. Our results reveal transcriptional dynamics of Chlorobaculum tepidum on elemental sulfur and increase our understanding of the mechanisms of transcriptional regulation governing growth on different reduced sulfur compounds. This report identifies genes and sequence motifs that likely play significant roles in the production and consumption of elemental sulfur. Beyond this focused impact, this report paves the way for the development of synthetic biology in Chlorobaculum tepidum and other Chlorobiaceae by providing a comprehensive identification of promoter elements for control of gene expression, a key element of strain engineering.

INTRODUCTION

The green sulfur bacteria (Chlorobiaceae) are a family of anaerobic photoautotrophic sulfur oxidizers. Chlorobaculum tepidum, the model system for this family, both produces and consumes extracellular S⁰ globules depending on the availability of sulfide in the environment (1). Recently, Hanson et al. showed that biogenic S⁰ globules could serve as the sole photosynthetic electron donor for growth of Cba. tepidum and that cell-S⁰ contact was required for that growth (2). Expanding upon that work, Marnocha et al. showed that cell-S⁰ contact was dynamic, with large populations of unattached Cba. tepidum cells growing at rates similar to those of attached cells (3). As polysulfides were detected in supernatants from cultures producing and consuming S⁰, Marnocha et al. proposed a model whereby polysulfides act as soluble intermediates in the formation and degradation of S⁰ globules and could feed unattached cells (3). Those two studies have significantly increased our understanding of how Cba. tepidum interacts with S⁰. However, many issues remain, in particular, how S⁰ metabolism is regulated and what proteins play a role in cell-S⁰ attachment.

The analysis of the Cba. tepidum genome showed that it encoded few recognizable transcriptional regulators, leading the authors to conclude that Cba. tepidum employs little transcriptional regulation (4). However, Eddie and Hanson observed a complex transcriptional response following the addition of sulfide to Cba. tepidum growing on thiosulfate (5). The transcript abundance of a two-gene operon (Cba. tepidum 1276 [CT1276]-CT1277) was significantly increased in the presence of sulfide, with the protein encoded by CT1277 belonging to the helix-turn-helix xenobiotic response element (HTH XRE) protein superfamily, indicating that transcriptional regulators with limited functional annotation may contribute to gene expression regulation in Cba. tepidum (5). While that study identified numerous transcriptionally regulated genes for the transition from thiosulfate to sulfide as a primary electron donor, the characteristics of the Cba. tepidum transcriptome during growth on S⁰ as a sole electron donor have not been documented.

Here, RNA sequencing (RNA-seq) and differential RNA-seq (dRNA-seq) were used to characterize the transcriptome of Cba. tepidum growing on biogenic S⁰ and to globally identify transcript start sites (TSS) active during growth on sulfide, biogenic S⁰, or thiosulfate as the sole electron donor. RNA-seq data suggest that the most dynamic changes in transcript abundance occur in response to sulfide and that the majority of genes differentially expressed in response to sulfide are downregulated. TSS positions were used to identify putative promoter elements and, in combination with RNA-seq data, were used to identify DNA motifs that were conserved across homologous Chlorobiaceae promoters. The positions of the discovered motifs relative to TSS and basal promoter elements suggest that, in agreement with the RNA-seq data, repression is the dominant mode of transcriptional regulation for genes involved in sulfur metabolism. In support of this observation, deletion of CT1277 from the Cba. tepidum genome led to overexpression of the sulfide:quinone oxidoreductase (SQR) CT1087 (6) during growth on thiosulfate, suggesting that the CT1277 gene product acts as a repressor.

RESULTS

The transcriptome of Cba. tepidum grown with S⁰ as the sole electron donor.

RNA-seq analysis of transcriptomes in Cba. tepidum cultures growing on thiosulfate and biogenic S⁰ revealed similar expression profiles for growth on the two substrates. A total of 28,183,709 reads were uniquely aligned to the genome for the thiosulfate library, while the S⁰ library had 37,350,690 uniquely aligned reads. Despite the tight correlation of expression results between thiosulfate and S⁰, 120 genes were differentially expressed (see Materials and Methods), with 55 genes being downregulated on S⁰ relative to thiosulfate and 65 genes being upregulated on S⁰ (see Table S1 in the supplemental material). Comparing the S⁰ transcriptome with the previously published sulfide transcriptome (5) identified 106 differentially expressed genes (see Materials and Methods), with 35 genes upregulated on sulfide relative to S⁰ and 71 genes downregulated on sulfide (Table S1).

The magnitude of transcript abundance changes was examined in detail across the substrates. Log₂ fold change values for genes with increased expression on sulfide relative to thiosulfate and S⁰ (15.2 and 15.7, respectively) were both much larger than that of S⁰ relative to thiosulfate (5.74). Similarly, the fold change values for genes with decreased expression on sulfide relative to thiosulfate and S⁰ were considerably lower than that of S⁰ relative to thiosulfate (−9.31 and −9.85 versus −2.30). This suggests that the most dynamic changes in transcript abundance occur in response to sulfide. These changes are clearly observed in genes encoding key components of the sulfur oxidation machinery of Cba. tepidum (7). For example, the thiosulfate oxidation (Sox) and S⁰ oxidation (Dsr) operons were strongly downregulated on sulfide (5) but were not differentially expressed between growth on S⁰ and growth on thiosulfate (Table S1).

Global identification of TSS and basal promoter elements in Cba. tepidum.

The dRNA-seq protocol used in this study produced 6.1 to 17 million reads per replicate (Table S2), with 2.0 to 6.7 million reads uniquely aligned to the genome for each replicate. Analysis of these data identified a total of 3,426 putative TSS across the results of growth on three electron donors in Cba. tepidum: 1,086 primary TSS (pTSS), 393 secondary TSS (sTSS), 583 antisense TSS (asTSS), 2,100 internal TSS (iTSS), and 71 orphan TSS (oTSS) (Fig. 1A; see also Table S3). Although the absolute numbers of putative TSS (2,458 for sulfide, 2,303 for S⁰, and 2,477 for thiosulfate) were similar for all of the conditions, there were condition-specific differences between the predicted TSS (Fig. 1B). However, there was little correlation between the condition-dependent TSS and changes in transcript abundance. A total of 1,417 of the predicted TSS (718 pTSS, 102 sTSS, 613 iTSS, 338 asTSS, and 33 oTSS) were found under all three conditions.

FIG 1.

(A and B) Venn diagrams for TSS data organized by TSS classification (A) or by growth condition (B). (C and D) Core promoter motifs identified by analyzing 50 bp upstream of all TSS: the RpoD motif (C) and an ECF factor-like motif (D). (E) TSS identified by dRNA-seq here (pink) closely match those previously reported (blue) for the sigA, csmB, csmC, and csmE genes and are preceded by RpoD motifs (yellow); see the text for references. (F) Extended −10 motif for Cba. tepidum RpoD promoters.

Two motifs were identified when a 50-bp sequence upstream of all TSS was analyzed by the MEME software package (Fig. 1C and D) (8). The most abundant motif (Fig. 1C) closely resembles consensus RpoD binding sequences, with the highest similarity to those of members of the Bacteroidetes, specifically, Flavobacterium spp. (TTG-N_17–23-TANNTTTG) (9). This would be expected, as RpoD protein sequences from Bacteroidetes and Cba. tepidum form clades together and away from other RpoD proteins (10). However, the −7 consensus sequence, TA[ATC][ATC][AT]T, is different than those of other published RpoD consensus binding sites (9), making the Cba. tepidum RpoD consensus binding sequence unique among the published consensus sequences reported to date. The RpoD motif was associated with 1,227 TSS (36% of all TSS) and with 64% of pTSS. Taken together, the data suggest that the primary sigma factor of Cba. tepidum is RpoD encoded by CT1551 (sigA).

The second motif (Fig. 1D) resembles σ⁷⁰ extracytoplasmic function (ECF) subfamily binding sequences, with the distinct AAC motif in the −35 consensus region (11). This motif closely resembles the promoter consensus sequences of σ^E (GG[AG][AC]C-N₁₈-[CG]GTTg) and σ^H ([CG]GGAAc-N₁₇-[CG]GTT[CG]) from Mycobacterium tuberculosis (12, 13) and that of σ^R (GGAAT-N₁₈-GTT) from Streptomyces coelicolor (14). As the Cba. tepidum genome encodes three σ⁷⁰ ECF factors (CT0278, CT0502, and CT0648), this motif may represent a consensus motif that all three ECF factors bind to, or it may represent the consensus sequence of a number of highly similar and yet distinct motifs that bind the ECF factors with different affinities. For example, σ^X and σ^W promoter sequences of Bacillus subtilis are highly similar, with some promoters binding both proteins whereas others bind one but not the other (15). Thus, this motif likely binds one or more ECF factors in Cba. tepidum. This ECF-like motif was associated with 182 TSS (5.3% of all TSS) and with 7.9% of pTSS.

Aside from RpoD and the three ECF sigma factors, Cba. tepidum encodes one additional sigma factor, CT1193 (rpoN), a σ⁵⁴ factor that controls expression of nitrogen-regulated genes, including the nif operon that encodes proteins for N₂ fixation in diverse bacteria. Phylogenetic footprinting of the nifH promoter across the Chlorobiaceae identified RpoN and NtrC binding motifs (data not shown) (16, 17). Using the FIMO search tool (18) in the MEME suite with the RpoN motif as a query returned one additional pTSS (with a false-discovery-rate [q] value of <0.05) preceding gene CT0644, which encodes an HSP20 family protein.

Four transcript start sites previously identified by primer extension analysis (sigA, csmB, csmC, and csmE) were captured in these data (Fig. 1E) (19, 20). While the sigA TSS predicted in this study was two base pairs upstream of the TSS identified by primer extension analysis and the csmE TSS was one base pair upstream, the csmB and csmC TSS were identical between the studies. All four promoters appear to be RpoD dependent, with the consensus motif, generated by WebLogo (21), closely resembling the RpoD motif identified by MEME with an extended −10 region (Fig. 1C and F). Chung et al. (20) showed that the csmCA genes were cotranscribed as a single unit from the csmC TSS. Additionally, the authors found a higher-abundance csmA transcript of approximately 350 bp, which they characterized as the processing product of the csmCA transcript. However, we found an RpoD-dependent pTSS and a sTSS 57 and 67 bp upstream of the csmA start codon, respectively (Fig. S1). The pTSS occurs 5 bp upstream of the predicted 5′ end of the csmA transcript predicted by Chung et al. (20). These TSS can explain the observations of Chung et al. without the need to invoke transcript processing.

Putative sulfide operator sequence 1 (PSOS-1) is associated with sqr and putative regulatory genes.

Cba. tepidum displays a robust transcriptional response depending on the reduced sulfur compound utilized for growth. Binning genes by expression pattern, e.g., all genes with increased transcript abundance on sulfide relative to thiosulfate, and then analyzing regions upstream of the associated pTSS did not produce any significant motifs associated with the promoter regions (data not shown). Therefore, we turned to conservation across Chlorobiaceae genomes, i.e., phylogenetic footprinting, to identify putative regulatory motifs associated with sulfur-regulated genes that are shared between Cba. tepidum and other Chlorobiaceae (Table 1).

TABLE 1.

Chlorobiaceae genomes used for phylogenetic footprinting

Phylogenetic footprinting identified an unknown motif (putative sulfide operator sequence 1 [PSOS-1]) in the promoter of CT1277 (Fig. 2A), encoding a putative transcriptional regulator that was highly upregulated on sulfide compared to thiosulfate (5) and S⁰ (Fig. 2). PSOS-1 was found associated with the pTSS for the two bona fide SQRs, CT0117 and CT1087, via FIMO searches (6, 18). CT0117 is a low-sulfide-concentration-adapted SQR, while CT1087 is adapted for high sulfide concentrations (6). Phylogenetic footprinting recovered PSOS-1 motifs associated with sqr genes that encode sulfide:quinone oxidoreductases across the Chlorobiaceae, including all CT1087 homologues (Fig. 2B) and a subset of CT0117 homologues (Fig. 2C). A consensus motif for Cba. tepidum PSOS-1 sites was constructed from CT1277, CT0117, and CT1087 sites (Fig. 2E) and was used as the seed in a FIMO search against the Cba. tepidum genome. In addition to returning the input loci, PSOS-1 was predicted to occur near 156 TSS. The only other TSS with a q value of <0.05 was a pTSS for CT0742 (Fig. 2D), a putative transmembrane protein in the TauE-like family of anion transporters. In Neptuniibacter caesariensis, TauE has been proposed as a sulfite transporter (22). Based on our TSS data, it appears that CT0743 is cotranscribed with CT0742. The CT0743 gene product is a hypothetical protein containing a domain of unknown function. Chlorobium ferrooxidans is a member of the Chlorobiaceae that cannot oxidize reduced sulfur compounds and instead oxidizes iron (23). PSOS-1 was searched against the Chl. ferrooxidans genome via FIMO, and no significant PSOS-1 positions were returned.

FIG 2.

Identification of putative sulfide operator sequence 1 (PSOS-1). Promoter regions for orthologs of CT1277 (A), CT1087 (B), and CT0117 (C) were extracted from all Chlorobiaceae genomes. RpoD (yellow) and PSOS-1 (blue) motifs were discovered by promoter analysis. TSS mapped in Cba. tepidum are shown in pink. (D) The Cba. tepidum consensus motif was searched against the Cba. tepidum genome, which returned an additional TSS associated with PSOS-1. (E) The PSOS-1 consensus motif for the three Cba. tepidum sites. (F) Log₂ fold changes in transcript abundance on sulfide relative to S⁰ of genes associated with PSOS-1 with ribosomal protein genes as a comparator as described by Eddie and Hanson (5).

In Cba. tepidum, the putative CT1277 regulator appears to be expressed from an RpoD pTSS (Fig. 2A). In the CT1277 promoter, the PSOS-1 motif overlapped the +1 site and the first two base pairs of the −6 box. The positioning of PSOS-1 relative to the −6 box was conserved for all genomes analyzed except Chlorobium phaeobacteroides BS1, where it was found largely in the spacer sequences between the −6 and −33 boxes, with partial overlap of the −6 box (Fig. 2A). In Chlorobium chlorochromatii, the PSOS-1 motif was discovered in the promoter of Cag_0887, a hypothetical protein encoded upstream of the CT1277 homologue Cag_0886; Cag_0887 to Cag_0885 are likely transcribed as a single unit. The CT1087 and CT0117 PSOS-1 motifs were positioned near the pTSS for each gene (Fig. 2B and C). The positioning of PSOS-1 within homologous CT1087 promoters was variable, with PSOS-1 overlapping the −33 box of two of the promoters and the −6 box of the other two. PSOS-1 overlapped the −6 box and extended upstream into the spacer sequence of the six homologous CT0117 promoters. However, in the Chloroherpeton thalassium promoter, PSOS-1 overlapped the −6 box and extended downstream, and in the Chlorobium limicola promoter, it overlapped the −33 box. Given that the PSOS-1 motif overlaps the RpoD binding motif or the +1 site or both, the data suggest that this motif binds a negative repressor, as occlusion of the RpoD binding site, or the +1 site, would likely interfere with transcription initiation. The PSOS-1 sequence appears to be unique in that it does not match any characterized motifs present in the CollecTF, Prodoric, and RegTransBase databases (24–27).

Expression levels of the components of the PSOS-1 regulon were variable between growth on sulfide and S⁰ (Fig. 2F). CT1277 and CT1087 were both significantly upregulated on sulfide relative to thiosulfate and S⁰, while CT0117 was downregulated on both electron donors relative to sulfide (Fig. 2F) (5). CT0742 and CT0743 did not change in expression significantly between the growth conditions and displayed expression profiles similar to those seen with ribosomal genes.

PSOS-2 is associated with sox, dsr, and CT2230 TSS.

As many genes related to thiosulfate oxidation (sox) and S⁰ oxidation (dsr and dsbE) were found to be downregulated on sulfide relative to S⁰ and thiosulfate (Fig. 3) (5), we searched the promoters of genes downregulated on sulfide for putative regulatory motifs. Only one RpoD TSS was observed in the sox operon immediately adjacent to soxJ (CT1015), supporting the assertion that the sox operon is transcribed as a single unit (7). Therefore, sequences upstream of the start codon of the first gene in the sox operon across the Chlorobiaceae were analyzed for motif discovery. Phylogenetic footprinting identified an unknown motif (putative sulfur operator sequence 2 [PSOS-2]) in the promoters analyzed (Fig. 3A). The sox PSOS-2 motif was searched against the Cba. tepidum genome. The top two positions occurring near TSS, excluding the pTSS for soxJ, were pTSS for dsrC-1 (CT0851) and CT2231. Phylogenetic footprinting of dsrC-1 and dsrC-2 (CT2250) and of orthologs across the Chlorobiaceae returned PSOS-2 sites in all dsrC promoters across the Chlorobiaceae (Fig. 3B). CT2231, encoding a hypothetical protein with no apparent homologues, appears to be cotranscribed with CT2230, encoding a putative outer membrane protein. As no TSS were found between these two genes, and as both are downregulated on sulfide relative to thiosulfate and S⁰ (Fig. 3) (5), sequences upstream of CT2230 homologues across the Chlorobiaceae and the CT2231 promoter were analyzed for motif discovery. PSOS-2 sites were found in all promoters (Fig. 3C). A consensus motif for PSOS-2 was constructed from soxJ, dsrC-1, dsrC-2, and CT2231 sites (Fig. 3E). In addition to returning the input loci, PSOS-2 was predicted to occur near 184 additional TSS. The only TSS with a q value of <0.05 were the pTSS for dsbE (CT1072; Fig. 3D) and an asTSS for CT1231, encoding a transposase and annotated as having an internal deletion (data not shown). PSOS-2 was searched via FIMO against the Chl. ferrooxidans genome, and no significant positions were identified.

FIG 3.

Identification of putative sulfide operator sequence 2 (PSOS-2). (A to C) Promoter regions for orthologs of the sox operon (A), dsrC (B), and CT2230 (C) were extracted from all Chlorobiaceae genomes. RpoD (yellow) and PSOS-2 (blue) motifs were discovered by promoter analysis. TSS mapped in Cba. tepidum are shown in pink. (D) The Cba. tepidum consensus motif was searched against the Cba. tepidum genome, which returned additional TSS associated with PSOS-2. (E) The PSOS-2 consensus motif for the four Cba. tepidum sites. (F) Log₂ fold change in transcript abundance on sulfide relative to S⁰ of genes associated with PSOS-2 with ribosomal protein genes as a comparator as described by Eddie and Hanson (5).

In Cba. tepidum, PSOS-2 was found to overlap the +1 site of the pTSS of soxJ and the −6 box of the RpoD motif predicted using the bulk TSS data (Fig. 3A). The positioning of PSOS-2 across sox promoters of the Chlorobiaceae was highly conserved, with all PSOS-2 sites found to overlap the −6 box and to extend downstream. PSOS-2 overlapped the +1 sites of both dsrC-1 and dsrC-2 and partially overlapped the −6 box of the RpoD motif (Fig. 3B). The position of PSOS-2 was variable across dsrC promoters, and yet it at least partially overlapped the −6 box in all promoters except that of Chl. phaeobacteroides, where it was found 6 bp downstream of the −6 box. PSOS-2 was found to overlap the +1 site of CT2231 and the last base pair of the −6 RpoD box (Fig. 3C). Positioning of PSOS-2 was fairly highly conserved across CT2230 promoters, with all but one site at least partially overlapping the −6 box. The PSOS-2 site in the dsbE promoter overlapped the −6 box and part of the RpoD spacer sequence (Fig. 3D). The PSOS-2 sequence appears to be unique in that it does not match any characterized motifs present in the CollecTF, Prodoric, and RegTransBase databases (data not shown) (24–27).

The components of the PSOS-2 regulon were downregulated on sulfide relative to S⁰ (Fig. 3F). The sox operon, CT2231-CT2230, dsbE, dsrC1A1, and dsrC2 are significantly downregulated on sulfide relative to S⁰ and thiosulfate (Table S1) (5). The first dsr cluster, dsrCABLEFH (CT0851-CT0857), appears to be a single transcriptional unit under the control of the dsrC1 RpoD pTSS. The second dsr cluster, dsrNCABLUEFHTMKJOPVW (CT2251-CT2238), appears to be broken into three transcriptional units: dsrN, dsrC2-P, and dsrVW. dsrN and dsrC2-P appear to be controlled by independent RpoD pTSS, while dsrVW appears to be controlled by an ECF factor pTSS (Table S3). This may explain why dsrVW is upregulated on sulfide relative to thiosulfate and S⁰ whereas the rest of the cluster is downregulated on sulfide relative to thiosulfate and S⁰ (Fig. 3F; Table S1) (5).

A CRP-like motif (CLM) is associated with psrABC, cydAB, and CT0729.

The putative polysulfide oxidoreductase complex, psrABC (CT0496-CT0494), is significantly upregulated on sulfide compared to S⁰ and thiosulfate (Fig. 4) (5). Phylogenetic footprinting identified two occurrences of a small motif (cyclic AMP receptor protein [CRP]-like motif [CLM]) upstream of the RpoD binding site in the Cba. tepidum and Cba. parvum promoters (Fig. 4A). The psrABC CLM was searched against the Cba. tepidum genome for positions that occurred within 200 bp upstream or 50 bp downstream of a TSS. Aside from returning the two psrABC sites, FIMO returned positions near the pTSS for CT1818 and CT0729 as the only positions with q values of <0.05. CT1818 and CT1819 encode cydAB, a terminal oxidase cytochrome bd complex that confers sulfide-resistant O₂-dependent respiration in Escherichia coli (28). CT0729 encodes a Nudix hydrolase domain protein. A consensus motif was generated from the psrABC, cydAB, and CT0729 sites and was used to search the Cba. tepidum genome (Fig. 4C). Significant positions (q values = <0.05) included pTSS for CT1089 and CT1061 and an oTSS at position 503855 (Fig. 4B). CT1089 and CT1088 encode the two subunits of ATP:citrate lyase. CT1061 is a hypothetical protein with a predicted steriodogenic acute regulatory protein-related lipid transfer (Start) domain found in polyketide cyclases and dehydrogenases. The sequence downstream of the oTSS was searched for open reading frames and for RNA families (29), but nothing of significance was found.

FIG 4.

Identification of the CRP-like motif (CLM). Promoter regions for orthologs of the psrABC operon were extracted from all Chlorobiaceae genomes. (A) RpoD (yellow), ECF sigma factor (purple), and CLM (blue) motifs were discovered by promoter analysis. TSS mapped in Cba. tepidum are shown in pink. (B) The Cba. tepidum consensus motif was searched against the Cba. tepidum genome, which returned additional TSS associated with CLM. (C) Consensus motif derived from Cba. tepidum CLM sites. (D) Log₂ fold change in transcript abundance of genes associated with CLM; S⁰ relative to thiosulfate (white) and sulfide relative to S⁰ (gray).

In psrABC promoters of Cba. tepidum and Cba. parvum, the first CLM site occurred 49 bp upstream of the RpoD −6 box, while the second site occurred 114 bp upstream of the Cba. tepidum −6 box and 113 bp upstream of the Cba. parvum −6 box (Fig. 4A). In Cba. tepidum, the RpoD −6 box occurs 4 bp upstream of the +1 site. The CLM site for cydAB occurred 57 bp upstream of the TSS, in a position similar to that of the first psrABC site, which occurred 58 bp upstream of the psrABC TSS (Fig. 4B). The CLM site for CT1089 occurred 35 bp upstream of the TSS and 4 bp upstream of the −33 box, while both the CT1061 and orphan sites overlapped the RpoD spacer sequence and −6 box. The CT0729 CLM site occurred 10 bp downstream of the CT0729 TSS.

The CLM motif resembles the cyclic AMP receptor protein (CRP [TGTGA-N₆-TCACA]) and the fumarate nitrate reduction protein (FNR [TTGAT-N₄-ATCAA]) consensus binding motifs of E. coli without the spacer sequence (30, 31). The spacing of the motif sites (Fig. 4A) echoes that of CRP-activating promoters in E. coli with multiple CRP binding sites upstream of the RpoD binding site (30). In support of this, TOMTOM (27) matched the CLM consensus to the PrfA consensus motif from Listeria monocytogenes, a CRP-domain transcriptional activator. The PrfA consensus sequence, WTAACAWWTGTTAA (32), does not contain the N_4–6 spacer present in E. coli CRP/FNR domain binding sites, adding further support for the idea that the Cba. tepidum CLM may bind a CRP domain protein. Cba. tepidum encodes a single CRP/FNR domain protein, CT1719. However, no gene encoding a CRP/FNR domain protein was found in the Cba. parvum genome or in any other Chlorobiaceae genome examined.

While the components of the CLM regulon had variable expression on S⁰ relative to thiosulfate and sulfide, the directions of expression on S⁰ relative to sulfide and thiosulfate were similar for all components; psrABC and cydAB were both significantly downregulated on S⁰ relative to thiosulfate and sulfide, while CT1061 and CT0729 were upregulated on S⁰ relative to thiosulfate and sulfide (Fig. 4D). The CT1089-CT1088 operon was not differentially expressed under the different growth conditions.

CT1277 is required for transcriptional repression of CT1087.

Previous data showed that the CT1276-CT1277 cassette displayed increased transcript abundance in cells following sulfide addition after growth on thiosulfate (5). Given that CT1277 belongs to the HTH-XRE family of transcriptional regulators, the CT1277 gene was deleted from the Cba. tepidum genome to assess its role in sulfide-dependent gene regulation. Expression of CT1087 was monitored by quantitative reverse transcriptase PCR (qRT-PCR) in the wild-type strain and in two independently isolated ΔCT1277 strains (Fig. 5). CT1087 transcript abundance was significantly elevated (3.5-fold, P < 0.03) in both ΔCT1277 strains compared to the wild type during growth on thiosulfate. Transition to growth on sulfide from thiosulfate resulted in a 1.9-fold increase (P = 0.04) in CT1087 transcript abundance in the wild-type strain, confirming that its expression is sulfide dependent as previously reported (5, 6). Post-sulfide addition, CT1087 displayed 16.3-fold (P = 0.003) and 4.6-fold (P < 0.001) increases in transcript abundance relative to the wild type in strains ΔCT1277.6 and ΔCT1277.11, respectively. Thus, CT1087 expression was induced much more strongly (9.7-fold; P = 0.003) in the absence of CT1277. Increased expression of CT1087 after sulfide addition in the ΔCT1277 strains suggests the presence of a sulfide-dependent activator. Sequences up to 2,000 bp upstream of the CT1087, CT1277, and CT0117 promoters were analyzed for motif discovery, but no candidate activator motif was found. Altered expression of CT1087 in the ΔCT1277 strains indicates that CT1277 negatively regulates the transcription of CT1087 in the presence of an activator. No significant growth phenotype was observed for the ΔCT1277 strain compared to the wild-type strain (data not shown).

FIG 5.

CT1277 is a transcriptional repressor of CT1087. qRT-PCR of CT1087 in the Cba. tepidum wild-type strain (white), the ΔCT1277.6 mutant (dark gray), and the ΔCT1277.11 mutant (light gray) pre-sulfide addition and 40 min post-sulfide addition. Error bars are equal to 1 standard deviation. HS−, hydrogen sulfide.

DISCUSSION

In this report, we provide global transcript abundance data for Cba. tepidum during growth on S⁰ as the sole electron donor and use these data to identify genes that were differentially expressed between growth on sulfide and growth on S⁰. Many of these genes encode key components of the sulfur oxidation machinery of Cba. tepidum. The most dynamic changes in gene expression with respect to growth on different reduced sulfur compounds were in response to sulfide. We also provide a global transcript start site map for Cba. tepidum. dRNA-seq identified 3,426 putative TSS across the results of growth on sulfide, thiosulfate, and S⁰, of which 1,086 were primary TSS. These data also include 71 orphan TSS that may control transcription of functional elements that were missed during genome annotation and provide evidence for antisense transcription in Cba. tepidum. Two basal promoter motifs were identified: an RpoD motif and an ECF sigma factor motif. Three putative regulatory motifs were discovered by phylogenetic footprint analysis of orthologous promoters across the Chlorobiaceae for genes that were differentially expressed during growth on reduced sulfur compounds in Cba. tepidum. Together, the data presented in this report provide a set of predictions for a mechanistic understanding of transcriptional regulation in different sulfur-dependent growth states in Cba. tepidum and in the members of the Chlorobiaceae as a whole.

TSS that are not associated with the RpoD motif or the ECF factor motif may not bind σ factors and therefore may not be bona fide TSS. In support of these possibilities, sequences (±50 bp) surrounding TSS associated with σ factor motifs were found to have significantly higher AT content than those sequences surrounding TSS without σ factor motifs (data not shown). Alternatively, these TSS may be associated with divergent σ factor binding sites that are not represented by the motifs discovered in this study (Fig. 1). Of the 2,049 TSS that do not have σ factor motifs, 1,733 (84.6%) are iTSS and 319 (15.6%) are pTSS, whereas of the 1,377 TSS with σ factor motifs, 367 (26.7%) are iTSS and 767 (55.7%) are pTSS. iTSS are enriched in TSS with no associated σ factor motif, suggesting that most iTSS in this study are not bona fide TSS. This is in contrast to a recent study in E. coli that found that most iTSS were associated with σ factor motifs (33). Indeed, the fraction of iTSS detected in this study, 61% of all TSS, is higher than that detected in other organisms: 37% in Escherichia coli, 36% in Campylobacter jejuni, and 18% in Helicobacter pylori (33–35). pTSS and sTSS occur at similar percentages across these organisms, while Cba. tepidum has fewer asTSS (17%) than others: 43% in E. coli, 48% in C. jejuni, and 41% in H. pylori. These variations may reflect differences in how more closely related organisms in the Gammaproteobacteria and Epsilonproteobacteria regulate transcription relative to the Chlorobiaceae. Future experiments will determine if the high percentage of iTSS in Cba. tepidum is characteristic of all Chlorobiaceae and if there is any transcriptional activity associated with iTSS-associated sequences that lack a recognizable sigma factor motif in Cba. tepidum.

A putative operator sequence, PSOS-1, was discovered in the promoters of the two SQRs CT0117 and CT1087, as well as in the promoters of the putative regulatory protein CT1277 and the TauE-domain protein CT0742 (Fig. 2). While CT1087 and CT1277 are highly upregulated on sulfide relative to S⁰ (Fig. 2) and thiosulfate (5), CT0117 is downregulated, and CT0742 and CT0743 are not differentially expressed. The expression data corresponding to growth on sulfide versus growth on S⁰ corroborate the results of previous studies that have observed downregulation of CT0117, and upregulation of CT1087 and CT1277, in response to high levels of sulfide (5, 6). Here, CT1277 was shown to negatively regulate CT1087 expression on thiosulfate and sulfide (Fig. 5). Therefore, PSOS-1 may represent the CT1277 binding motif, which would suggest that CT1277 is subject to autorepression. If PSOS-1 binds CT1277 and functions as an operator sequence, then promoters associated with PSOS-1 should be downregulated in response to sulfide due to increased CT1277 expression, as increased repressor concentrations would be expected to lead to increased repression of those loci; this is the case for CT0117 (Fig. 2). However, CT1087 expression increased in response to sulfide in the ΔCT1277 background, suggesting the presence of a sulfide-dependent activator (Fig. 5). As CT1277 is associated with PSOS-1, and shows an expression profile similar to that of CT1087, the results suggest that the activator acting on CT1087 also activates CT1277. Thus, whatever element activates CT1087 and CT1277 expression should be absent from the CT0117 promoter. This leads to a model where sulfide activates a sulfide-dependent activator and indirectly activates the PSOS-1-binding protein (CT1277) via the inferred activator (Fig. 6). CT1277 represses transcription of CT0117, CT1087, and CT1277, while the sulfide-dependent activator activates transcription of CT1087 and CT1277 but not that of CT0117. This may occur to fine tune expression of CT1277 and CT1087.

FIG 6.

Simplified model of the oxidative sulfur metabolism regulatory network in Cba. tepidum with nodes representing metabolites (HS⁻ and S⁰), putative regulatory signals (gray), genes downregulated on sulfide relative to S⁰ (pink), and genes upregulated on sulfide (HS⁻) relative to S⁰ (blue). Arrows represent activation, and bars represent repression.

The proposed models for CT1277 function are based on its association with the members of the HTH_XRE protein superfamily (NCBI accession no. cl22854) that are thought to be DNA binding transcriptional regulators. The notion that CT1277 and orthologs bind DNA was supported by a secondary structure prediction that identified three helical regions located at residue 36 to residue 46, residue 53 to residue 58, and residue 63 to residue 82 (see Fig. S2 in the supplemental material). These regions are annotated in multiple databases as homologous to sigma factor DNA binding regions, i.e., InterPro homologous family IPR013324-RNA polymerase sigma factor, region 3/4, and TIGR02937-sigma70-ECF (RNA polymerase sigma factor, sigma-70 family). DNA binding of these regions is conferred by a three-helix bundle as predicted for CT1277. The protein sequence alignment of Chlorobiaceae CT1277 orthologs also revealed six conserved cysteine residues (Fig. S2) (36). Four cysteine residues (Cys-105, Cys-108, Cys-124, and Cys-127) occur in -CXXC- motifs often associated with FeS cluster binding, while Cys-7 and Cys-151 are isolated. We hypothesize that these conserved cysteine residues may interact with sulfide, possibly affecting the formation or stability of an FeS cluster, giving the protein a mechanism for sulfide- or redox-mediated activation.

For several of the key components of sulfide oxidation that are downregulated in response to sulfide, a putative operator sequence, PSOS-2, was discovered overlapping the TSS and/or the RpoD −6 box (Fig. 3). The positioning of this motif and the expression patterns of the associated transcriptional units strongly suggest that this motif functions as an operator sequence. As these genes are downregulated in response to sulfide relative to thiosulfate (5), and appear to also be downregulated in response to sulfide on S⁰ (Fig. 3), PSOS-2 may bind a sulfide-dependent repressor (Fig. 6). CT2230 is predicted to be a membrane transporter in the FadL family of outer membrane proteins and has been characterized in long-chain fatty acid transport in E. coli (37). As CT2230 is involved in the transport of hydrophobic molecules, and is predicted to be part of a sulfide-repressed regulon, it may be involved in transport of polysulfide or hydrophobic sulfur chains across the outer membrane.

The psrABC promoter was found to contain two CLM sites, with single CLM sites occurring near the TSS of cydAB, CT1089-CT1088, CT1061, and CT0729 (Fig. 4). In E. coli, CRP can function as both a repressor and an activator depending on its positioning relative to the TSS and sigma factor binding site (30). psrABC and cydAB displayed similar expression profiles in that they were downregulated on S⁰ relative to thiosulfate and sulfide (Fig. 4D). However, CT1089-CT1088 did not change significantly in expression between growth conditions, and CT1061 and CT0729 are both upregulated on S⁰ relative to sulfide. These expression patterns can be explained if the location of the CLM site is taken into account (Fig. 4A and B) and if S⁰ inhibits the activity of the CLM-binding protein. The CLM sequence likely functions as an activator for psrABC and cydAB, while it likely functions as a repressor for CT1061, the oTSS, and CT0729 in that it interferes with sigma factor binding or transcription elongation. The CLM site may not have a large effect on CT1089-CT1088 and could function as either a repressor or an activator (30). Thus, if S⁰ inhibits the function of the CLM, it would repress psrABC and cydAB, while activating CT1061 and CT0729 (Fig. 6).

Between the transcript abundance data (this study and reference 5) and the data from the three putative regulatory motifs discovered in sulfur-regulated genes, it appears that sulfide likely acts as a master regulator, activating proteins that bind PSOS-1, where CT1277 is the most likely candidate binding protein, PSOS-2, and the inferred but unidentified sulfide-dependent activator acting on CT1087 (Fig. 6). If S⁰ inhibits the activity of the CLM binding protein, where CT1719 is the most likely candidate binding protein, then the expression pattern of CLM-associated genes can be explained. As the three motifs identified here were found to be conserved between multiple Chlorobiaceae genomes, it suggests that the regulatory functions carried out by these motifs are conserved across genomes. Thus, we have identified genes important for growth on S⁰ relative to sulfide and thiosulfate, globally mapped TSS that were active during growth on these electron donors, identified basal promoter motifs, and discovered putative sulfur regulatory motifs. From these data, a model has emerged whereby sulfide is a master regulator of the metabolic state of Cba. tepidum, inducing the repression of genes important for thiosulfate and S⁰ oxidation and the activation of several key components for sulfide oxidation.

MATERIALS AND METHODS

RNA sequencing.

Cultures were grown under conditions of 20 μmol photon m⁻² s⁻¹ photosynthetically active radiation (PAR) in Pf-7 medium (6) at 47°C with thiosulfate as the sole electron donor or at 42°C with biogenic S⁰ as the sole electron donor. RNA was extracted using a NucleoSpin RNA kit (Macherey-Nagel), and rRNA was depleted using a MicrobExpress kit (Ambion) and treated with a TURBO DNA-free kit (Ambion) to remove residual genomic DNA (gDNA). RNA-seq libraries were constructed using a NuGEN Ovation kit (NuGEN) to convert 25 ng RNA to double-stranded cDNA, with subsequent fragmentation to approximately 100-bp fragments performed using an S2 Adaptive Focused Acoustic Disruptor (Covaris). Fragment size was confirmed by agarose gel electrophoresis on a 2% gel. A NuGEN Encore kit (NuGEN) was used to ligate adapters suitable for Illumina sequencing to the ends of these fragments. Libraries were sequenced on an Illumina HiSeq 1000 sequencer at the University of Delaware Sequencing and Genotyping Center.

RNA-seq analysis.

Reads were aligned to the Cba. tepidum genome using the Eland pipeline (Illumina). Custom Perl scripts (available upon request) were used to calculate the number of sequences that mapped to each annotated gene, noncoding RNA, and intergenic region of more than 50 bp, correcting for the length of the region. Data were normalized using quantile normalization as previously described (5). Levels of expression that differed for each pair of libraries were calculated using DESeq (38).

We previously reported the transcriptional response of Cba. tepidum to sulfide addition after growth on thiosulfate (5). The sulfide and S⁰ expression libraries were constructed with different kits and were sequenced at different times, so, in order to compare expression of Cba. tepidum on S⁰ to that on sulfide, fold changes on S⁰ relative to thiosulfate were divided by fold changes on sulfide relative to thiosulfate, giving a fold change value for each gene on sulfide relative to S⁰. JMP Pro (version 12.1.0) was used to create a box-and-whisker plot for the distribution of the fold change of growth on sulfide relative to growth on S⁰. Outliers, or those points outside 1.5× the interquartile range from the mean, were called as differentially expressed genes (see Table S1 in the supplemental material).

Differential RNA sequencing.

dRNA-seq libraries were constructed from four independent cultures grown on thiosulfate and on biogenic S⁰ and from those grown on thiosulfate, spiked with 1.6 mM sulfide, and harvested 30 min after sulfide addition. Cba. tepidum cultures were grown as previously described by Levy et al. (39) at 20 μmol photon m⁻² s⁻¹ PAR. Replicate 1.5-ml cell pellets were harvested by centrifugation after 20 h of growth (mid-log phase). Cell pellets were flash frozen in liquid nitrogen and were stored at −70°C. Cell pellets were thawed on ice, resuspended in 100 μl Tris-EDTA (TE) buffer (pH 8.0) with 1 μl Ready-Lyse lysozyme (Epicentre) (20 KU/μl), and incubated at room temperature for 30 min. Two replicate cell pellets were combined, and RNA was purified by the use of a NucleoSpin RNA kit (Macherey-Nagel). A 10-μg volume of RNA was treated with a Turbo DNA-free kit (Thermo Fisher) for gDNA removal and was then concentrated over RNA Clean & Concentrator-25 columns (Zymo Research). rRNA was depleted via the use of a MicrobExpress kit (Ambion), and samples were concentrated over Zymo-25 columns. Libraries were normalized by sigA copy number (5) prior to terminator 5′ phosphate-dependent exonuclease (TEX; Epicentre) treatment. Each replicate was split into two samples for TEX treatment: the first was treated with 1 U TEX in a 20-μl reaction mixture, while the second was incubated in the same buffer without TEX (35). Reactions were cleaned over RNA Clean & Concentrator-5 columns (Zymo Research). Samples were subsequently treated with 2 U tobacco acid pyrophosphatase (TAP; Epicentre) in a 50-μl total reaction volume followed by RNA concentration on Zymo-5 columns. 5′ adapters from a NEBNext Multiplex Small RNA Library Prep Set for Illumina kit (NEB) were ligated to the treated RNA samples using 2.5 μl of ligation enzyme mix in a 20-μl total reaction volume with final concentrations of 1 mM ATP and 12.5% polyethylene glycol 8000 (PEG 8000). RNA was subsequently fragmented for 2 min using a NEBNext Magnesium RNA fragmentation module (NEB) and was cleaned using Agencourt RNAClean XP beads (Beckman Coulter). 3′ ends were repaired via calf intestinal alkaline phosphatase (NEB) treatment and were cleaned over Zymo-5 columns. 3′ adapter ligation, first strand synthesis with indexed primers, and cDNA amplification (15 cycles) were completed using a NEBNext small RNA kit. cDNA reactions were purified via the use of Agencourt AMPure XP beads (Beckman Coulter). BluePippin was used for size selection (150 to 600 bp) prior to Illumina sequencing performed over two lanes (12 libraries pooled per lane) on a HiSeq 2000 sequencer at the University of Delaware Sequencing and Genotyping Center.

dRNA-seq analysis.

The RNA-seq analysis pipeline READemption (version 0.3.9) was used to align reads to the Cba. tepidum genome, and the output was passed to TSSpredator (version 1.06) for TSS identification (34, 40). In order for a TSS to be called as active under a specific condition, it needed to be detected in at least three of four replicates with an enrichment score of ≥2 and to meet the remaining parameters specified under the “very specific” setting. TSS were classified according to the definitions of Sharma et al. (35). Primary TSS (pTSS) are defined as representing TSS within 300 bp upstream of an annotated gene. If multiple TSS are present in this range, the pTSS is that with the highest enrichment value, and the others are secondary TSS (sTSS). Antisense TSS (asTSS) occur on the antisense strand internal to or within 100 bp downstream of an annotated feature. Internal TSS (iTSS) occur on the sense strand within an annotated gene. Orphan TSS (oTSS) do not fall into any of the other defined classes.

Motif discovery and analysis.

Basal promoter motifs were identified by analyzing 50 bp upstream of all TSS by the use of the MEME software suite (8). The outputs from multiple MEME runs using various parameter settings were pooled via the use of a custom R script (available upon request) to obtain estimates for the number of TSS associated with each motif. Promoter regions for orthologs of genes that are strongly regulated by sulfide in Cba. tepidum were extracted from all Chlorobiaceae genomes (Table 1). Sequences of up to 1,000 bp upstream of the start codon for each orthologous gene set were analyzed by MEME and/or DMINDA (8, 41). Motifs identified as described above were searched against the Cba. tepidum genome using FIMO to identify additional occurrences (18).

Deletion mutagenesis.

CT1277 was deleted from the Cba. tepidum genome using a counterselectable suicide vector that is to be fully described elsewhere (J. M. Hilzinger and T. E. Hanson, unpublished data). Briefly, a PCR product containing flanking DNA upstream and downstream of CT1277 was cloned into a mobilizable suicide vector with both antibiotic resistance and a counterselectable marker based on a vector used for gene deletions in Shewanella oneidensis MR-1 (42). The primers used for amplifying the flanking DNA were as follows: CT1277-5′UTR-F-EcoRV (GATATCTCATCCCACCTATGAGCA), CT1277-5′UTR-R (GTCGAGCGGAATGATGATCTTCATTCTAGATTTTTTTAGTCAGCAATT), CT1277-3′UTR-F (AATTGCTGACTAAAAAAAATCTAGATGAAGATCATCATTCCGCTCGAC), and CT1277-3′UTR-R-EcoRV (GATATCGTCGAGTTTGTTGAGGATAC).

Response to sulfide qRT-PCR.

Cba. tepidum wild-type and ΔCT1277 strains were grown in Pf-7 medium at 47°C with 20 μmol photon m⁻² s⁻¹ PAR. The absence of sulfide from cultures was verified by testing with CuCl₂ (2). Presulfide biomass was pelleted by centrifugation, flash frozen in liquid nitrogen, and stored at −70°C. Sulfide was added to a final concentration of 2 mM, and cultures were incubated at 47°C with 20 μmol photon m⁻² s⁻¹ PAR for 40 min. Postsulfide biomass was pelleted, flash frozen, and stored at −70°C. RNA was extracted and purged of residual genomic DNA as described for dRNA-seq. sigA and CT1087 mRNAs were reverse transcribed into cDNA using the SigA-R-RT and CT1087-R-RT primers (6) and a ProtoScript II cDNA synthesis kit (NEB) in the same reaction mixture. Negative controls lacking reverse transcriptase were performed to detect the presence of gDNA. The expression levels of sigA and CT1087 were determined using RealMasterMix SYBR ROX (5 Prime) and an ABI 7500 Fast real-time PCR system (Applied Biosystems). Genomic DNA standards were used to determine the efficiency of the SigA-RT and CT1087-RT primer sets and to quantify transcript abundance. CT1087 levels were normalized using sigA expression levels.

Data availability.

Data described in this paper have been deposited in the National Center for Biotechnology Information Sequence Read Archive affiliated with BioSample accession numbers SAMN07413950 for S⁰ and thiosulfate RNA-seq data and SAMN07413841 for all dRNA-seq data.