A novel transcription factor PunR and Nac are involved in purine and purine nucleoside transporter punC regulation in E. coli

Many genes in bacterial genomes are of unknown function, often referred to as y-genes. Recently, novel analytic methods have divided bacterial transcriptomes into independently modulated sets of genes (iModulons). Functionally annotated iModulons that contain y-genes lead to testable hypotheses to elucidate y-gene function. Inversely correlated expression of a putative transporter gene, ydhC, relative to purine biosynthetic genes, has led to the hypothesis that it encodes a purine-related transporter and revealed a LysR-family regulator, YdhB, with a predicted 23-bp palindromic binding motif. RNA-Seq analysis of a ydhB knockout mutant conrmed the YdhB-dependent activation of ydhC in the presence of adenosine. The deletion of either the ydhC or the ydhB gene led to a substantially decreased growth rate for E. coli in minimal medium with adenosine as the nitrogen source, as well as with inosine or guanosine. Taken together, we provide clear evidence that YdhB activates the expression of the ydhC gene that encodes a novel purine transporter in E. coli. We propose that the genes ydhB and ydhC be re-named as punR and punC, respectively. from the beads, and overnight incubation at 65 degrees reversed the protein-DNA cross-link. RNA- and protein-free DNA samples were used to perform primer extension and second adaptor ligation with the following modications. The DNA samples, incubated for primer extension as described previously (9), were treated with the dA-Tailing Module and NEBNext Quick Ligation Module (New England Biolabs) for second adaptor ligation. The DNA sample, puried using the GeneRead Size Selection Kit (Qiagen), was enriched by polymerase chain reaction (PCR) using Phusion High-Fidelity DNA Polymerase (New England Biolabs). The amplied DNA samples were puried again with a GeneRead Size Selection Kit (Qiagen) and quantied using Qubit dsDNA HS Assay Kit (Life Technologies). The quality of the DNA sample was checked by running the Agilent High Sensitivity DNA Kit using an Agilent 2100 Bioanalyzer before sequencing using HiSeq 2500 (Illumina) following the manufacturer’s instructions. Each modied step was also performed following the manufacturer’s instructions. ChIP-exo experiments were performed in duplicate.


Introduction
At least one third of the genes in the growing number of available sequenced microbial genomes are still of unknown function. Furthermore, the functional annotations of a large portion of the remaining genes are based exclusively on sequence similarity to a relatively small pool of experimentally studied proteins.
Even the best-studied model bacterium, Escherichia coli K-12 MG165, is not an exception, despite the long-term effort to curate available knowledge bases. EcoCyc is the premier example of a curated knowledge-base that summarizes functional evidence for over 4,500 genes, including 1567 enzymes, 282 transporters, and 204 transcription factors (1). Previous efforts to construct a global functional atlas of E. coli proteins revealed that 1,431 of 4,225 (35%) protein coding genes were not functionally annotated as of 2019 (2,3) according to three criteria: (i) the gene name starts with 'y' (hence, it is a y-gene); (ii) it does not have a linked pathway in EcoCyc; and (iii) it has a 'predicted', 'hypothetical', or 'conserved' protein functional description in GenProtEC (4). The COMBREX knowledge base further reveals that experimental evidence is lacking for 44% of E. coli proteins (5).
The bioinformatic prediction and experimental con rmation of microbial gene functions is a challenging problem. Bioinformatic analyses for the prediction of gene regulation, gene clustering analysis, and metabolic pathway reconstruction lead to reliable hypotheses, followed by experimental veri cation (6)(7)(8). Such discovery efforts for speci c classes of gene functions, such as transcription factors (TFs), is bearing fruit (9,10), leading to a comprehensive list of 278 TFs in E. coli.
The transcriptional regulation of purine/pyrimidine biosynthesis and the salvage regulon are coordinated by the transcriptional regulator PurR in Enterobacteria. We previously identi ed groups of genes, each of which independently but co-ordinately modulated under different conditions, called iModulons (11).
iModulons differ from regulons in that they are computed as independent source signals in a set of transcriptomic data sets (a dynamic measure). In contrast, regulons are de ned by TF binding sites in the promoters of the regulated genes (a static measure). iModulons showed considerable overlap with previously characterized regulons (11). In some cases, iModulons split such regulons into two dynamically regulated sets of genes, thus, creating two iModulons corresponding to a single regulon.
One example highlighted in this study is the PurR-1 iModulon, which contains genes related to purine biosynthesis and utilization, as a part of the PurR regulon ( Fig. 1) (11). Genes encoding the purine biosynthetic pathway (pur) and the transporters (ghxP, xanP) were found to be repressed in the presence of exogenous adenine, whereas two genes were upregulated in the presence of adenine: add, encoding adenosine deaminase and involved in purine catabolism/salvage, and ydhC, encoding a putative transporter of unknown function from the Drug:H + Antiporter-1 (DHA1) family (TCDB family ID 2.A.1.2).
The expression of these genes was inversely correlated with purine biosynthesis gene expression (11). In particular, the add gene is directly activated by the ribose-responsive transcriptional regulator RbsR, while the transcription of purHD is repressed by RbsR (12), suggesting the dependency of purine biosynthesis on intracellular ribose concentration.
The purine and pyrimidine de novo biosynthetic group of genes is directly controlled by PurR; however, the regulatory mechanism for ydhC was not reported. YdhC had previously been detected as an arabinose e ux transporter in rich medium (13), suggesting a relationship to pentose e ux and possibly the nutrient uptake from the rich medium.
Here, we identify a LysR-family regulator and propose that YdhB directly controls the transcription of YdhC in response to adenine. We also provide evidence that YdhC is a purine and purine nucleoside transporter ( Fig. 1) and that YdhC is important for the resistance to sulfonamides. Machine learningbased analysis of RNA-Seq data was a useful tool for studying the co-expression of genes and identi ed the ydhBC genes as targets for investigation. Transcriptional activation of ydhC by the nitrogen assimilation control regulator Nac was shown by RNA-Seq analysis of the nac mutant in the adenine supplemented medium. The work described here expands our understanding of how purine uptake for salvage and degradation versus biosynthesis is reciprocally regulated in E. coli.

Results
Systems analysis and the prediction of PunC function. The reconstruction of purine salvage and utilization under different conditions shows that adenosine/adenine and guanosine/guanine transporters are important as mediators of the uptake of purines when present in the medium as a supplement or substitute for a more traditional carbon or nitrogen source. Guanosine, inosine, cytidine, and thymidine (but not uridine, adenosine, and xanthosine) are transported by NupG (14). The pyrimidine nucleoside:H + symporter, NupC, has been characterized (14,15). NupG and NupC can recognize the nucleoside ribose moiety (14) and are regulated by the global carbon catabolite repressor protein, CRP, as well as pyrimidine sensor/regulators CytR and DeoR (16). The NupC and NupG transporters are activated by the presence of cytidine. But no difference for nupG expression noticed in the presence of adenine as supplement to M9 medium (ModulonDB). NupC and NupG were shown recently to be important ADP-glucose uptake transporters, and they are essential for the incorporation of extracellular ADP-glucose into glycogen during biosynthesis (16). The coordinate regulation of different purine/pyrimidine transporters is important for the utilization of nucleosides as carbon and nitrogen sources or nucleoside salvage under various starvation conditions (17).
Adenosine as a nitrogen source is utilized via adenosine deaminase (Add), which converts it to inosine (Fig. 1). PpnP is a broad speci city pyrimidine/purine nucleoside phosphorylase that produces hypoxanthine (a purine base) and D-ribose-1-phoshate from phosphate and inosine (the corresponding nucleoside), respectively (18). However, it also acts on uridine, adenosine, guanosine, cytidine, thymidine, and xanthosine as substrates.
The PurR regulon is essential for purine biosynthesis when salvage is limited. PurR repression is activated by the presence of hypoxanthine or guanine, signalling that the purine cytoplasmic concentration is su cient for growth (17,19,20). The inverse correlation for the modulation of mRNA levels for PurR-regulated purine biosynthetic genes and those of the ydhC and add genes (21) has been demonstrated during growth in minimal medium with glucose and adenine as a supplement, suggesting a role in purine salvage (11). For PurR regulon related genes, the levels of mRNAs encoding proteins of purine biosynthesis and guanosine uptake (PurR-1 iModulon) decrease in the presence of adenine in contrast to the high increase of ydhC mRNA level (11,19), but no changes in mRNA level was found for nupC and nupG.
Under the conditions described here, extracellular adenosine or adenine is the only nitrogen source supporting the growth of E. coli, but it does not support growth of a ydhC deletion mutant strain. The hypoxanthine produced from adenosine is a signal for the PurR-mediated repression of purine biosynthesis. This proposed regulation of the PurR-1 iModulon and reconstruction of purine biosynthesis and purine uptake is shown in Fig. 1. The reconstruction of purine biosynthesis and regulation is important for the gene function hypothesis. Independent component analysis revealed the transcriptional regulation of the Pur-1 iModulon, consisting of the purB, purC, purD, purE, purF, purH, purK, purl, purM, purN, purT genes and those encoding the corresponding transporters, as well as the inverse regulation of the transcription of the ydhC and add genes in response to the presence of adenine in the growth medium (11). Thus, purine biosynthesis pathway genes and those encoding the guanine/hypoxanthine and xanthine transporters (ghxP, xanP), are repressed by PurR in response to the availability of cellular hypoxanthine (Fig. 1). Extracellular adenosine/adenine potentially gives rise to hypoxanthine due to uptake followed by the action of adenosine deaminase, and further, by the PpnP-catalyzed reaction ( Fig. 1), but the adenosine uptake transporter for nitrogen source utilization was not found.
We hypothesize that YdhC is an adenosine/adenine transporter, re-named PunC, because the presence of a purine source in the medium inhibits biosynthesis (via the PurR-dependent repression), while upregulating PunC. Although NupG is a known purine transporter, regulation by CRP and CytR suggests that NupG is essential for uptake when purine is to be used as a carbon source. NupG requires hydroxyl groups in the ribose moiety for substrate binding and is not utilized for adenine uptake. The punC (b1660) regulation by nitrogen assimilation protein (Nac) was detected by RNA-Seq according to published data (22) (ModulonDB). Nac activates the pathways for utilization of histidine, proline, urea, and alanine in Klebsiella pneumoniae, a related enterobacteria (23). The physiological function of PunC is likely to supplement E. coli growth with purine/purine nucleobase to be used as a nitrogen source or during purine salvage.
Prediction of PunR binding sites. The previously uncharacterized transcriptional regulator YdhB (renamed PunR here) belongs to the LysR family of bacterial transcription factors. The PunR gene is located in a conserved gene cluster with the divergently transcribed punC gene, and this arrangement is conserved in other Proteobacteria (Fig. S1). The punR promoter is predicted to be Sigma 24-dependent, and the punC promoter is Sigma 24-and Sigma 70-dependent (RegulonDB). To identify and characterize DNA binding sites of PunR in the E. coli genome, we utilized combined bioinformatic and experimental approaches. We applied a comparative genomic approach of phylogenetic footprinting (24) to predict the putative PunR-binding site in the common intergenic region between the punR and punC genes (Fig. S2). Alignment of the upstream regions for the punC genes from three groups of Enterobacteria revealed conserved motifs with a common 23-bp palindromic consensus for binding of the predicted PunR regulator (Fig. 2). A similar palindromic motif was also identi ed in the punC/punR intergenic region in Pseudomonas spp. (Fig. 2), con rming strong conservation of the predicted PunR binding site across gamma-proteobacteria. A common motif of these orthologous operators is an imperfect palindrome with consensus TsttwTCAAwAwwwTTGaaGGCA, where 's' is either G or C and 'w' is either A or T.
Experimental validation of the PunR binding site. We conducted ChIP-exo studies in M9 glucose medium supplemented with 2.5 mM adenosine to validate the predicted PunR operator and searched for additional candidate PunR binding sites in the E. coli genome (Fig. 2). The chromosomal punR gene was genetically modi ed to encode a Myc-tagged PunR. The E. coli strain was grown with glucose as the carbon source in M9 minimal media, and the ChIP-exo experiment was designed to pull down PunR which occupied an ~ 200 bp region upstream of punC. The PunR-protected region from E. coli corresponds to the PunR binding site predicted by phylogenetic footprinting (Fig. 2).
We further conducted uorescent polarization (FP) assays to con rm PunR binding to its predicted 23-bp DNA motif. The puri ed PunR_His protein (40% purity) or refolded protein was incubated in the assay mixture with a uorescently labelled DNA fragment (as described in Materials and Methods), predicted by phylogenetic footprinting. PunR (0-60 nM) was incubated with a 29-bp DNA fragment containing the predicted PunR-binding site plus a 10 mM MgSO 4 supplement for 1 hour at room temperature in the presence and the absence of adenine. The FP signals at different concentrations of PunR are shown in Fig. 3, but the binding for the refolded PunR had not been detected. PunR binding in the presence, but not the absence, of 0.35 mM adenine was measured using the uorescence polarization assay. The PunR binding constant K d = 27 nM, for binding to the DNA, was calculated using Prism 7.
RNA-Seq analysis of the PunR regulon in adenosine or adenine supplemented media. RNA-Seq analysis of the punR mutant of E. coli MG1655 (WT) in M9 with glucose as the carbon source (to exclude NupG overproduction), supplemented with 2.5 mM adenosine, showed downregulation of the punC gene. The punR deletion strain of E. coli MG1655 showed a lower expression level of punC (Table 1), while the level of other mRNAs was not changed, suggesting that PunR is a speci c activator for punC expression. The location of the PunR binding site upstream of the predicted punC promoter agrees with the PunRdependent activation of punC (Fig. S2). Supplementation of M9 medium with adenosine increased punC mRNA levels only in the presence, but not in the absence, of PunR. medium and growth with adenosine supplementation as the sole nitrogen source. The iModulon involved in motility and agellar synthesis and regulated by the minor sigma factor FliA was downregulated in the WT strain (Fig. S3), but not in the punR mutant strain. FliA regulates the motility and bio lm formation in E. coli and is regulated by FlhDC (11). The hDC operon is subject to regulation at the transcriptional level by more transcriptional regulatory proteins than any other known operon in E. coli (25).
The upregulation of punC has been shown by RNA-Seq analysis according to published data (22) for the E. coli MG1655 (WT) in M9 minimal medium in the presence, but not in the absence of adenine (iModulonDB) and revealed that punC regulation is Nac-dependent (additionally to PunR). The punC expression pro le analysis revealed upregulation during growth in M9 medium supplemented with 10 mM adenine in the presence of Nac, but not in the absence (Fig. 4A), suggesting that both regulators are essential for the punC transcriptional activation. The volcano plot for the differentially expressed genes compared to the nac mutant is shown (Fig. 4B).
Growth in M9 medium with adenosine or 2-deoxyadenosine as the nitrogen source and with glucose or glycerol as the carbon source. Two E. coli strains, MG1655 and BW25113, have been tested with glucose as the carbon source, when adenosine was added to the M9 medium. The difference between two strains has been shown by the iModulons analysis (11). Under nitrogen limiting conditions, growth of E. coli BW25113 was stimulated, but no growth enhancement was observed for the punR or punC mutants (Fig. 5A), but no difference in M9 medium with NH 4 Cl was detected (Fig. 5B). The punR mutant of E. coli MG1655 has a similar growth phenotype under the same conditions (Fig. 5C). When the nupG mutant was tested with the same M9 medium (2.5 mM adenosine as the nitrogen source) no growth phenotype was observed (data not shown).
When the punC and add mutants of E. coli BW25113 were examined with 0.4% glycerol as the carbon source and 2.5 mM adenosine as the nitrogen source, the growth effect was substantial (the growth phenotypes are shown in Fig. 6A). The growth phenotype with 5 mM 2-deoxyadenosine for the punC mutant was determined, while the add deletion had no effect on growth. These effects are shown in Fig. 6D. We suggest that high concentrations of adenosine/2-deoxyadenosine can support metabolism via the pentose phosphate pathway under nitrogen/carbon starvation, and that PunC is the major transporter under these conditions.
Growth in M9 medium with guanosine (or inosine) and glutamate as nitrogen sources and glycerol as the carbon source. Supplementation of M9 minimal medium with glycerol as the sole carbon source and glutamate as the nitrogen source did not support growth of E. coli. The effect of either inosine or guanosine as supplement as the nitrogen sources was examined using the WT and isogenic add and punC deletion mutant strains, measuring growth when glycerol was the carbon source. The punC mutant showed a decrease in the growth rate compared to the WT strain under nitrogen starvation conditions when inosine or guanosine was present as an additional nitrogen source ( Fig. 6B and C). This suggested that the PunC transporter can take up, in addition to adenosine and 2-deoxyadenosine, inosine and guanosine. Thus, the PunC uptake transporter exhibits broad speci city for purine nucleosides as has been revealed by these phenotypic analyses under different conditions.
PunC speci city screening for carbon sources using Biolog plate 1 with NH 4 Cl as the nitrogen source.
The growth observed for wild type E. coli BW25113 and its isogenic punC mutant (WT) using different carbon sources, including adenosine or inosine, on Biolog plate 1 was measured using M9 minimal medium without another carbon source. The strains were grown overnight in LB and washed twice with M9 medium without the addition of a carbon source. The cultures were diluted in M9 medium without a carbon source, and 0.1 ml of each was added to the 96-well plates. Growth was detected with Omnilog, but no difference for adenosine or other carbon sources was detected for the mutant and WT strain under microaerobic conditions. The growth of the punC deletion mutant in minimal medium with adenosine as the carbon source was likely supported by NupG transporter.
PunC transporter speci city screening for antibiotic sensitivities using Biolog plates 11C and 12. We screened for the possible PunC-mediated transport of various antibiotics using the punC mutant compared to the E. coli BW25113 strain. For these purposes, we used Biolog plates 11C and 12. Each of these plates contained four different concentrations of 24 different antibiotics. The punC mutant and WT strains showed the same resistance for all of the included antibiotics except sulfonamides: sulfathiazole, sulfadiazine and sulfamethoxazole. The punC mutant effect has been shown at different sulfonamides concentration (Fig. 7). The increased resistance observed for the punC deletion mutant strain suggests that it exhibits speci city for sulfonamides. The sulfonamide group of antibiotics are structural analogs of para-aminobenzoic acid (PABA). Mercaptoguanine derivatives inhibit conversion of PABA and 6hydroxy-methyldihydropterin-PP to dihydropteroate (26). It is interesting that a distant punC paralogous gene in E. coli encodes Bcr -a bicyclomycin/sulfonamide resistance protein (27). Bcr has been shown to be involved in the export of L-cysteine (28). In contrast, PunC appears to be an adenine/2-deoxyadenosine/adenosine/guanosine/inosine uptake transporter with broad purine speci city, that may allow uptake of sulfamethoxazole and sulfothiazole, but not sulfamethazine. The analysis of E. coli inhibition by sulfonamides structural analogs shows the PunC speci city for different sulfonamides modi cations (Fig. 7).

Discussion
A new RNA-Seq data analytic approach was applied to study the PurR regulon and PurR-1 iModulon (see iModulonDB.org). The inverse relationship in the PurR-1 i-Modulon, between the expression of purine biosynthetic genes and punC, led to the hypothesis that punC encodes a purine transporter. We addressed this hypothesis and showed that: 1) expression of the putative transporter, PunC, is essential for adenosine, inosine, guanosine, and 2-deoxyadenosine uptake/growth under nitrogen (nitrogen/carbon) starvation conditions, but not for growth using adenosine as a carbon source in M9 medium; 2) the upregulation of the punC gene in the presence of adenine in M9 medium is the nitrogen assimilation regulatory protein, Nac, dependent. The analysis of RNA-Seq data for the nac mutant during growth in the medium supplemented with 10 mM adenine showed the activation of punC only in the presence of nac; PunC is additionally regulated by the LysR family HTH-type transcriptional regulator PunR, and the punR gene is conserved within the punC genome context. The punC transporter is upregulated under nitrogen starvation conditions, consistent with the sigma-24 (sigma-E) dependent promoter catalogued RegulonDB (29); and 3) The PunC dependent uptake of sulfathiazole and sulfamethoxazole was implied from the fact that the punC mutant strain has substantially increased resistance under microaerobic conditions compared to wild type E. coli BW25113.
The E. coli punC deletion mutant has a growth defect in a minimal medium with either 2-deoxyadenosine, adenosine, inosine, or guanosine as the nitrogen source. The predicted DNA-binding for PunR to the palindromic sequence between punR and punC, conserved in many Proteobacteria, serves as the PunR binding site. PunR binding was demonstrated for E. coli using ChIP-exo and a uorescent polarization assay. We suggest that PunR functions as an activating transcriptional regulator. This hypothesis is supported by the RNA-Seq analysis of the punR mutant and the fact that PunR binding has been detected in the presence, but not in the absence, of adenine by the FP method. The punC gene is present in many Proteobacterial genomes, including all Enterobacteria. The same is true for the punR and purR genes, which are conserved in the same gene contexts (Fig. S1). It is interesting that PunR had previously been shown to be essential for Yersinia pseudotuberculosis growth (30), suggesting that adenosine uptake may be a property of many Proteobacteria. The Yersinia pseudotuberculosis NupG transporter ortholog is absent, but the NupC ortholog is present.
PunR is also essential for growth during nitrogen starvation with purine as the sole nitrogen source and glycerol as the carbon source (Fig. 5). It is interesting that PunR had previously been shown to be essential for Yersinia pseudotuberculosis growth (30), and the NupG ortholog is absent in this bacteria, suggesting that PunC dependent adenosine/guanosine uptake may be a property of many Proteobacteria. A punC mutant was found to be more resistant to representative toxic nucleobase analogs, suggesting an uptake function.
The presence of adenosine as the sole nitrogen source can support E. coli growth, and it may be involved in acid resistance because the deletion of add attenuates growth in the presence of adenosine under acidic conditions (31). The hypothesis that PunC is involved in acid resistance in the presence of adenosine has been proposed, but under aerobic growth conditions at pH 5.5 with adenosine or inosine as a supplement, PunC dependent growth phenotypes have not been detected in the BW25113 strain (data not shown).
Adenine, adenosine, and deoxyadenosine can be converted to the guanine nucleotide via the salvage pathway and support growth in minimal medium (32). The presence of inosine or guanosine in the E. coli growth minimal medium increase the growth rate with a poor nitrogen source as with glutamate and glycerol (sole carbon source), and PunC is essential for the growth. The poor tness phenotype for an punC mutant, encoding a homologous transporter in Pseudomonas simiae WCS417 (66% identity with the E.coli protein), has been demonstrated during growth in minimal medium with glucose and adenine as the nitrogen source (33), but mildly important with adenosine as the nitrogen source ( t.genomics.lbl.gov), then adenine transport function for the Pseudomonas simiae PunC is suggested. The PunC broad speci city function in Klebsiella michiganensis M5al is supported by the strong tness phenotype for the punC mutant homolog (75% identity with the E. coli homolog). This mutant was found to have a strong negative tness during growth with any one of several purines and purine nucleosides as carbon sources: inosine, 2-deoxyinosine, 2-deoxyadenosine, and 2-deoxyadenosine 5-phosphate, as well as a mild negative effect with adenosine as the carbon source. All of these observations substantiate the main conclusion of this paper that PunC is an adenosine/inosine transporter in E. coli, but the PunC homologous transporter in Klebsiella michiganensis likely has broad speci city for 2-deoxy-inosine/2deoxyadenosine/2-deoxyadenosine-phosphate.
Taken together, the data analysis approach used here shows that PunC family representatives are regulated by PunR and Nac. The transcriptional analysis revealed that the putative transporter encoded by punC is upregulated by Nac and is correlated with PurR regulation. It is interesting that Nac also upregulates L-histidine biosynthesis genes and pili synthesis in M9 supplemented with adenine (Fig. 4).
The experimental veri cation of the novel PunC transporter function revealed that PunC has a broad speci city to purines and important under nitrogen limited conditions. punC is transcriptionally regulated by the LysR-family regulator PunR and Nac. Nac previously had been shown as adaptor to NtrC protein, probably PunR-Nac complex is essential for the activation. PunR is the transcriptional activator for the punC gene in the presence of adenine, con rmed by experimental veri cation using RNA-Seq, uorescent polarisation, and ChIP-exo. The screening of the data represented in the iModulonDB, RegulonDB, EcoCyc, Fitness Browser, and PubSEED databases/platforms will produce new hypothesis for y-gene functions.

Materials And Methods
Bacterial strains and growth conditions. The punR mutant of E. coli MG1655 strain was constructed as described in (9). The punR, add and punC mutant strains of E. coli BW25113 from the Keio single-gene knockout collection (34) or punR mutant of E. coli MG1655 and wild type were grown overnight in LB medium. The punR, add and punC gene deletions were con rmed by PCR using gene speci c primers. The cells were refreshed for 3 hours, washed with M9 salts medium and inoculated in the M9 medium with 0.4% glycerol or 0.2% glucose as the carbon source and lacking the usual a nitrogen source, but substituting it with 2.5 mM adenosine, 5 mM 2-deoxyadenosine or M9 supplemented by 5 mM of Lglutamate and 5 mM inosine or guanosine as nitrogen sources. The E. coli BW25113 wild type and punC mutant were grown overnight in LB medium, washed by M9 medium and inoculated as recommended to the Omnolog plates 11C and 12B for the antibiotic resistance measurements.  (36,37). The insoluble fraction was solubilized in 7 M urea and puri ed on a Ni-NTA minicolumn with At-buffer (50 mM Tris-HCl buffer, pH 8, 0.5 mM NaCl, 5 mM imidazole, and 0.3% Brij) with 7M urea. The PunR refolding was effected on the column in the At-buffer. The puri cation procedure has been described in detail (36).
Identi cation of putative PunR binding sites by comparative genomics. The potential PunR-binding sites were identi ed by a phylogenetic footprinting approach using multiple sequence alignments (Fig. S2).
Orthologs of the E. coli punC and punR genes in other proteobacteria, as well as multiple sequence alignments of orthologous upstream regions, were identi ed using the PubSEED comparative genomics platform (38). PunR-binding site sequence logos were constructed using the WebLogo tool (39).
RNA sequencing. RNA sequencing data were generated after growth under aerobic exponential growth conditions in M9 medium supplemented with adenosine. The wild type MG1655 strain was grown as a control for the isogenic punR mutant strain. Pre-cultures for the RNA sequencing experiments were started for the growing the cells in LB medium. Cells were then washed twice with M9 medium and inoculated at an OD 600 of 0.05. The cells were collected at an OD 600 of 0.6 and were harvested using the Qiagen RNA-protect bacterial reagent according to the manufacturer's speci cations. Pelleted cells were stored at -80 o C, and after cell resuspension and partial lysis, they were ruptured with a bead beater; the total RNA was extracted using a Qiagen RNA puri cation kit. After total RNA extraction, the quality was assessed using an Aglient Bioanalyser using an RNA 6000 kit after removal of ribosomal RNA. Paired-end strand speci c RNA sequencing libraries were prepared as described (9).
To ensure the quality of the compendium, genes shorter than 100 nucleotides and genes with under 10 fragments per million-mapped reads across all samples were removed before further analysis. Transcripts per million (TPM) were calculated by DESeq2 (v1.22.1). The nal expression compendium was log-transformed log 2 (TPM + 1) before analysis, erred to as log-TPM. Biological replicates with R 2 < 0.9 between log-TPM were removed to reduce technical noise.
ChIP-exo experiments. The strains harboring 8-myc were generated by a λ red-mediated site-speci c recombination system, targeting the C-terminal region as described previously (40). ChIP-exo experimentation was performed following the procedures previously described (10,41). To identify PunR binding sites for each strain, the DNA bound to PunR from formaldehyde cross-linked cells, collected after growth in M9 supplemented with adenosine, was isolated by chromatin immunoprecipitation (ChIP) with the antibodies that speci cally recognize the myc tag (9E10, Santa Cruz Biotechnology), and Dynabeads Pan Mouse IgG magnetic beads (Invitrogen) were added, followed by stringent washings as described previously (9). ChIP materials (chromatin-beads) were used to perform on-bead enzymatic reactions of the ChIP-exo method. Brie y, the sheared DNA of the chromatin-beads was repaired by the NEBNext End Repair Module (New England Biolabs), followed by the addition of a single dA overhang and ligation of the rst adaptor (5'-phosphorylated) using a dA-Tailing Module (New England Biolabs) and NEBNext Quick Ligation Module (New England Biolabs), respectively. Nick repair was performed by using the PreCR Repair Mix (New England Biolabs). Lambda exonuclease-and RecJf exonuclease-treated chromatin was eluted from the beads, and overnight incubation at 65 degrees reversed the protein-DNA cross-link. RNA-and protein-free DNA samples were used to perform primer extension and second adaptor ligation with the following modi cations. The DNA samples, incubated for primer extension as described previously (9), were treated with the dA-Tailing Module and NEBNext Quick Ligation Module (New England Biolabs) for second adaptor ligation. The DNA sample, puri ed using the GeneRead Size Selection Kit (Qiagen), was enriched by polymerase chain reaction (PCR) using Phusion High-Fidelity DNA Polymerase (New England Biolabs). The ampli ed DNA samples were puri ed again with a GeneRead Size Selection Kit (Qiagen) and quanti ed using Qubit dsDNA HS Assay Kit (Life Technologies). The quality of the DNA sample was checked by running the Agilent High Sensitivity DNA Kit using an Agilent 2100 Bioanalyzer before sequencing using HiSeq 2500 (Illumina) following the manufacturer's instructions. Each modi ed step was also performed following the manufacturer's instructions. ChIP-exo experiments were performed in duplicate.
PunR uorescent polarization assay. The recombinant PunR protein with N-terminal His 6 tag was overproduced in an E. coli strain from the ASKA collection (35). The PunR-producing strain was grown overnight, inoculated into a 50 ml culture, and induced with 0.6 mM IPTG after an OD 600 = 0.6 was reached. The cells were harvested after 4 hours and lysed as previously described (42). The puri ed protein and a 10 nM uorescently labelled DNA fragment (5'-AGG GGG-3') were incubated in the assay mixture. The PunR binding assay mixture (0.1 ml) contained Tris buffer, pH 7.5, 0.1 M NaCl, 0.5 mM EDTA, 10 mM MgSO 4 , 2 mM DTT, 5 µg/ml sperm DNA and 1µM of the uorescently labelled predicted