A systems approach discovers the role and characteristics of seven LysR type transcription factors in Escherichia coli

Although Escherichia coli K-12 strains represent perhaps the best known model bacteria, we do not know the identity or functions of all of their transcription factors (TFs). It is now possible to systematically discover the physiological function of TFs in E. coli BW25113 using a set of synergistic methods; including ChIP-exo, growth phenotyping, conserved gene clustering, and transcriptome analysis. Among 47 LysR-type TFs (LTFs) found on the E. coli K-12 genome, many regulate nitrogen source utilization or amino acid metabolism. However, 19 LTFs remain unknown. In this study, we elucidated the regulation of seven of these 19 LTFs: YbdO, YbeF, YcaN, YbhD, YgfI, YiaU, YneJ. We show that: (1) YbdO (tentatively re-named CitR) regulation has an effect on bacterial growth at low pH with citrate supplementation. CitR is a repressor of the ybdNM operon and is implicated in the regulation of citrate lyase genes (citCDEFG); (2) YgfI (tentatively re-named DhfA) activates the dhaKLM operon that encodes the phosphotransferase system, DhfA is involved in formate, glycerol and dihydroxyacetone utilization; (3) YiaU (tentatively re-named LpsR) regulates the yiaT gene encoding an outer membrane protein, and waaPSBOJYZU operon is also important in determining cell density at the stationary phase and resistance to oxacillin microaerobically; (4) YneJ, re-named here as PtrR, directly regulates the expression of the succinate-semialdehyde dehydrogenase, Sad (also known as YneI), and is a predicted regulator of fnrS (a small RNA molecule). PtrR is important for bacterial growth in the presence of l-glutamate and putrescine as nitrogen/energy sources; and (5) YbhD and YcaN regulate adjacent y-genes on the genome. We have thus established the functions for four LTFs and identified the target genes for three LTFs.

www.nature.com/scientificreports/ Glu is also essential for tetrahydrofolate polyglutamylation 11 . Ptr is important for bacterial growth and for efficient DNA replication, transcription, and translation 12,13 and plays an important role in maintaining compact conformations of negatively charged nucleic acids 14 . Ptr is also involved in multiple antibiotic resistance mechanisms under stress conditions 15 . The puuA, puuD, puuE, and puuP genes in E. coli are induced by Ptr and regulated at the transcriptional level by the Ptr-responsive repressor PuuR 16,17 . The expression of sad (yneI) is induced by the addition of Ptr to the medium 18 ; however, it is not regulated by PuuR, and a transcriptional regulator for sad had not been described before this work. The predicted PtrR binding site in the sad promoter region was confirmed via an in vitro binding assay with the purified PtrR protein and using the ChIP-exo assay. We further compared the whole genome transcriptional response of the ptrR knockout and wild type E. coli strains to carbon/nitrogen starvation in the presence of Ptr and l-glutamate using RNA-Seq analyses, and the PtrR DNA-binding site was predicted for fnrS, encoding small regulatory RNA. The physiological roles of PtrR regulation and antibiotic resistance are discussed.

Results
An integrated systems-approach uncovers LysR-type transcription factors in E. coli K-12. Previously, we generated a list of candidate transcriptional factors (TFs) from uncharacterized genes ("y-genes") using a homology-based algorithm 19 . Among these TFs, it was predicted that YeiE, YfeR, YidZ, YafC, YahB, YbdO, YbeF, YbhD, YcaN, YdcI, YdhB, YeeY, YfiE, YgfI, YhaJ, YhjC, YiaU, YneJ, and YnfL belong to the LTF based on Hidden Markov Model. YdhB (re-named PunR) function had been shown to be an activator for the purine transporter, PunC 1,2 . Recently YdcI (Salmonella enterica) function was shown related to biofilm formation 20 . We further chose 7 uncharacterized LysR-type TFs (yLTFs), which have transcriptional responses (increased mRNAs) to the presence of a l-threonine (Thr) in M9 medium (iModulonDB, PRECISE2). l-threonine is an important source for l-serine, l-glycine, branched-chain amino acid biosynthesis, and formate, which is important for anaerobic respiration. To elucidate roles of each yLTF and genome-wide target genes, we performed gene expression profiling via RNA-Seq and ChIP-exo detection of the LTF DNA-binding and growth phenotyping. The overall workflow is shown in Fig. 1.

YbdO (CitR) regulatory effects are involved in citrate utilization and YbeF is involved in the flagella biosynthesis. The physiological function for LTFs CitR and YbeF was previously unknown. The
YbdO (CitR) ChIP-exo result detected the peak for DNA-binding upstream of citR-ybdNM ( Fig. 2A,B), that further was confirmed by the transcriptional analysis. The other ChiP-exo detected peaks for possible suggested genes regulation were not found in DEGs for citR deletion mutant strain. The citR (regulated by HNS) and ybdNM operons are conserved adjacent genes in Enterobacteriaceae (Fig. 2D) and the ybdN promoter was predicted to be regulated by FNR (EcoCyc). Further, gene expression profiling showed that citR deletion leads to strong upregulation of the ybdNM operon and the adjacent genes citCDEF. Interestingly that genome context analysis evolved that citR is conserved with the citrate lyase encoding genes cluster citCDEF. (Fig. 2C). Therefore, this result confirmed autoregulation of citR-ybdNM, and regulation for citrate lyase which is involved in anaerobic metabolism. From PRECISE2, upregulation of citC was detected when E. coli MG1655 strain was grown in M9 medium supplemented with Thr, suggesting that citrate utilization is important for the catabolizm of Thr. The YbdM has been predicted as yTF (Uniprot) and probably related to cit operon regulation. The DNA-binding site upstream of the citR gene was predicted by the phylogenetic footprinting method (Fig. S2), suggesting CitR is an autoregulator and ybdMN regulation.
To test the regulation of the citrate lyase encoding gene cluster (citEFG), the phenotype for growth in the M9 medium supplemented with citrate at low pH (pH 6.5) showed an effect of citR deletion on citrate utilization, suggesting artificial de-repression of citrate lyase ( Fig. 2E-F), but no difference for the growth in M9 medium supplemented with citrate was found at pH 7.5. Additionally, citR mutation decreases the E. coli BW25113 fitness phenotype for motility in LB medium and affects growth using glycolate as carbon source (Fitness Browser, fit. genomics.lbl.gov) and microaerobic utilization of formate as carbon source (Fig. 7).
CitR and YbeF are paralogs (identity 30%), presented in conserved gene clusters in Enterobacterial genomes with citrate lyase operon cit (citCDEFXGT, Figs. 2C, S3). Interestingly the transcriptomic analysis of the citR and ybeF deletion mutants showed that the expression level of lrhA was substantially decreased (− 4.3-fold and − 3.5 -fold, respectively, Table 1, Supplementary data), likely indicating that YbeF and CitR has affected lrhA regulation (CitR/YbeF DNA-binding was not detected/predicted). The FlhDC iModulon includes FlhDC regulated genes that were upregulated in the ybeF and citR deletion mutants (Fig. S4B). LrhA is an LTF repressor for flhDC, flagellar biosynthesis genes responsible for motility and chemotaxis 21 . The deletion of ybeF showed downregulation of lrhA and upregulation of flhC. The iModulon FlhDC (Fig. S4) was substantially upregulated in ybeF mutant strain, suggesting connection for YbeF transcriptional regulation and flagella biosynthesis.

YcaN and YbhD regulators mutant characterization. The ybhD and ycaN mutant strains and WT
were collected at the late-exponential phase after growth in the M9 medium supplemented with Thr. The conserved gene cluster ycaN-ycaK-ycaM with adjacent ycaC and ycaD genes was detected in the Escherichia coli K12 and Shigella boydii genomes. We noticed that in the ycaN deletion mutant strain differentially expressed genes ycaC (downregulated) and ycaD, focA (encoding formate channel) (upregulated) are genes adjacent to ycaN. It is interesting that ycaK, ycaC, and ycaD are predicted to be regulated by Nac (belonging to LTF), as Nac function is nitrogen assimilation, and ycaC and ycaD are probably nitrogen assimilation function related (EcoCyc). In ycaN deletion mutant highly DEGs are the tnaAB genes, encoding l-tryptophanase and a tryptophan transporter, l-arginine degradation, astCADBE, an autoinducer-2 transport system (lsrABCD, lsrR), and the HTH-type transcriptional regulator, galS, the genes were strongly downregulated (Supplementary Data 1). The genes encod- www.nature.com/scientificreports/ ing amino acid metabolism (l-valine biosynthesis (IlvB, IlvN), threonine dehydrogenase (Tdh), transcriptional activator (TdcA), and glycolate utilization pathway (GlcC, GlcD, GlcF)) were additionally downregulated in the ycaN mutant. The ycaN deletion mutant phenotype microarray applied for 95 carbon sources had evolved the negative phenotype for formate utilization specifically suggesting YcaN dependent focA regulation involved formate utilization (Fig. 7). The ybhD gene is divergently oriented with respect to the conserved Enterobacterial gene cluster ybhHI and the putative hydrolase gene, ybhJ (Fig. S5A). YbhH is a 4-oxalomesaconate tautomerase homologous protein, and YbhI is a putative tricarboxylate transporter, homologous to 2-oxoglutarate/malate translocator, (id. 35%) (iModulonDB) (Fig. S5B). The YbhH encoding gene was strongly upregulated in a ybhD deletion mutant, as shown by transcriptomic analysis. ybhH and ybhI transcription is regulated by Nac (EcoCyc, iModulonDB) and the functional relation to nitrogen assimilation is suggested. The ybhD deletion strain growth phenotype in the M9 minimal medium with glycerol (carbon source), supplemented by l-malate was detected (Fig. S6).

YgfI (DhfA) regulation and glycerol and formate utilization. ChIP-exo assays previously detected
DhfA binding upstream of the dhaKLM operon encoding the dihydroxyacetone phosphotransferase (DHAK) (Fig. 3A-B). The transcriptional activation of DhaKLM is important for glycerol utilization and M9 supplemented by Thr or l-tryptophan ( Fig. 3B) (iModulonDB). DHAK in the E. coli MG1655 strain is involved in glycerol utilization (Uniprot). Accordingly, we decided to test the effect of the dhfA deletion on the growth on glucose (Fig. 3C-D) and glycerol (Fig. 3E). The resulting deficiency in growth on glycerol is potentially explained by dhaKLM as well as pflB, hycBCD, hycEFG, and adhB transcriptional DhfA activation, as shown by RNA-seq (Table 2), and the DhfA binding site was predicted upstream of those genes (Table S1). The DEGs detected by RNA-seq showed substantial downregulation of formate fermentation related genes (Fig. 3F) encoding pyruvate-formate lyase (pflB), fumarate reductase (frdABCD), formate hydrogenlyase (hycBCD, hycEFG), and the regulator (hycA), as well as hydrogenase encoding genes (hybABC, hybEF, and hybO) and the gene encoding protein involved in maturation of all hydrogenases isozymes (hypB) (Fig. 3F) and the dhaKLM operon ( Table 2, Supplementary Data). dhfA deletion had little effect on growth using glucose as the carbon source (Fig. 3C) and with Thr supplement (Fig. 3D) aerobically, but the dhfA strain had a low growth rate microaerobically without Thr supplement in minimal medium (Fig. S7).
The detected DhfA binding upstream of dhaKLM and the defect for the growth on glycerol (Fig. 3E) for the dhfA mutant confirm the DhfA-dependent transcriptional activation of DHAK, as DHAK is involved in glycerol    www.nature.com/scientificreports/ utilization 22 . We suggest re-naming YgfI to DhfA-dihydroxyacteone/formate utilization activator. The fermentation products during microaerobic growth on M9 glucose were detected in the Thr supplemented M9 medium. The analysis evolved the higher efflux of formate and acetate for the dhfAmutant strain ( Fig. 3G), suggesting that DhfA is important for formate utilization. The DhfA dependent activation of hyc and hyb operons was important the formate utilization microaerobically as was shown by the phenotype microarray analysis (Fig. 7).

YiaU (LpsR) regulatory network and yiaU mutant growth phenotype. ChIP-exo results show the
LpsR binding for the regulation of waaP (encoding lipopolysaccharide (LPS) biosynthesis glycero-d-mannoheptose kinase) (Fig. 4A). The results suggest that LpsR is important for LPS biosynthesis at specific conditions and the majority of DEGs encode the proteins involved in cell wall/membrane/envelope biogenesis, carbohydrate transport, energy production, and amino acid metabolism (Fig. 4B). Accordingly, gene expression for the lpsR mutant was lower for the genes from the operon waaPSBOJYZU, suggesting that LpsR is the LPS biosynthesis operon activator (Table 3); previously the regulator for waa operon was not known (ecocyc.org). Additionally, ChIP-exo detected binding for the genes encoding adenine transporter, adeP, suggests transcriptional regulation, and they were detected as DEGs for the yiaU mutant (Fig. 4D). We suggest re-name YiaU to LpsR-LPS biosynthesis regulator. The Biolog plates with the antibiotics were tested microaerobically in RPMI10LB medium for the possible LTF waa operon activation in response to the stress. LpsR was found essential for survival at high concentration of oxacillin microaerobically (Fig. 4C) in RPMI_10LB medium. WaaZ and WaaY were shown to be essential for survival with nafcillin in E. coli 23 . The difference for the growth in exponential/ log phase for the lpsR mutant strain had not been detected (Fig. 4D), although LpsR regulation had affected final OD 600 (stationary phase) in the M9 glucose medium. The phenotype (log phase) for the mutant in M9 with 0.3 M NaCl added was additionally detected (Fig. 4D). www.nature.com/scientificreports/ The LysR family representatives are known to regulate adjacent genes, and lpsR-yiaT are conserved in the bacterial genomes (Fig. 4E). The RNA-seq results for the lpsR deletion mutant strain showed upregulation of the yiaT gene, encoding a predicted outer membrane protein membrane anchor for the surface display for the proteins, homologue of MipA. MipA is an MltA (murein-degrading enzyme) interacting protein.

YneJ (PtrR) regulatory effects for sad and fnrS transcription and putrescine utilization.
Two distinct Ptr utilization pathways are known for E. coli (Fig. 5A). The first is catalyzed by the PuuA, PuuB, PuuC, and PuuD enzymes encoded by the puuP, puuA, puuDR, puuCB, puuE gene cluster and involves degradation of Ptr to γ-aminobutyric acid (GABA) via γ-glutamylated intermediates. The alternative pathway of Ptr degradation to GABA consists of PatA (Ptr aminotransferase) and PatD (γ-aminobutyraldehyde dehydrogenase) 24 . The PuuABCDE pathway is essential for Ptr utilization in E. coli using PuuP as the major Ptr transporter 25 . GABA is further utilized by two alternative 4-aminobutyrate aminotransferases (GABA-AT) encoded by gabT and puuE, and also two succinate semialdehyde dehydrogenases (SSADH) encoded by gabD and sad 26,27 .
We decided to characterize in detail the YneJ (re-named PtrR) by analyzing the PtrR ChIP-exo detected binding sites 1 . The ptrR gene is located in a conserved gene cluster with the divergently transcribed sad (yneI) gene, which encodes succinate semialdehyde dehydrogenase and yneH (glsB) glutaminase ( Fig. 6A and 6C), upregulated in the evolved yneJ mutant strain (iModulonDB) 28 . To identify and characterize DNA binding sites of PtrR in the E. coli genome we utilized a combined bioinformatics and experimental approach. First, we applied a comparative genomic approach of phylogenetic footprinting 27 to predict putative PtrR-binding sites in the common intergenic region of the ptrR and sad genes (Fig. 6C, Fig. S8). The ptrR/sad genes are conserved in several taxonomic groups including Escherichia/Salmonella/Shigella, Citrobacter, Enterobacter, and Klebsiella, as well as in Pseudomonas spp. In E. coli and closely related enterobacteria the sad gene belongs to the putative sad-yneH gene cluster, while in Enterobacter and Citrobacter the orthologous genes include an additional gene encoding the methyl-accepting chemotaxis protein I (serine chemoreceptor protein, Mcp) (Fig. 6C). The multiple sequence alignment of ptrR/sad upstream regions from E. coli and closely related enterobacteria (termed Group 1 species) contains a conserved 15-bp palindromic sequence with consensus TTCACnAATnGAGAA downstream   www.nature.com/scientificreports/ predicted sigma-E dependent promoter (Fig. 6A). We also analyzed upstream regions of ptrR orthologs in other enterobacterial genomes (Group 2 species), where the sad gene ortholog is absent and ptrR is co-localized with an uncharacterized MFS-family transporter gene. Further, we predicted two conserved DNA sites with similar consensus sequences located in their common intergenic region (Fig. 6C). We further confirmed the identified putative PtrR-binding site upstream of the sad genes in E. coli and conducted genome-wide mapping of other PtrR-binding sites using the ChIP-exo method. To identify in vivo PtrR binding sites, E. coli was grown under glucose as the carbon source in the M9 minimal media. A total of nine PtrR-binding sites were detected in these experiments. PtrR binds in the promoter regions of the fhuC, moeB, dhaK, fnrS, gltS/xanP, and ptrR/sad genes. The experimentally identified 50-bp PtrR-binding region at sad/ptrR genes contains the conserved palindromic DNA motif identified via phylogenetic footprinting (Fig. S8). Comparison of this DNA motif with eight other regions containing experimentally mapped PtrR-binding regions did not reveal significant sequence similarity except for the PtrR-binding area at fnrS, which shares a common consensus with the identified DNA motif at sad/ptrR. We created multiple alignments of the upstream DNA sequences of closely related species with the beginning of the E. coli gene for fnrS and these binding sites corresponded to the ChIP-exo protected areas. The binding sites TTC ACG AATCGaGAA, TTCtCGA TTC GTGAA, and TgaAtGcAaCGTcAA were predicted for ptrR, sad(yneI), and fnrS, respectively.
Experimental assessment of the computationally predicted PtrR DNA-binding site TTCtCGA TTC GTGAA in the sad promoter region has been facilitated using a PtrR-binding fluorescent polarization (FP) assay (Fig. 6B). The recombinant overproduced PtrR was obtained using a strain from the ASKA collection. The PtrR purification procedure is described in Supplemental materials. The binding of the purified refolded PtrR protein to synthetic DNA fragments containing the predicted PtrR-binding site was assessed using FP in the assay mixture containing 10 mM urea (Fig. 6B). Specific binding of PtrR to the DNA (5′-GGG TTC TCG ATT CGT GAA GGG-3′) was detected at 0.6 uM of PtrR in contrast to the negative control (PhrR) 29 . The fluorescent polarization assay showed binding for the PtrR to the predicted sad binding site and the addition of GABA leads to dissociation of PtrR from the fluorescently labeled DNA (Fig. 6B), suggesting a regulatory function for Ptr/GABA catabolism for energy. If the Ptr utilization pathway intermediate, GABA, accumulates, PtrR de-repress sad and fnrS. The upregulation for sad and glsB had been detected for the yneJ mutant strain for adapted E. coli MG1655 mutant (deletion menF-entC-ubiC) (Fig. 5D) 28 .
An E. coli BW25113 (WT) and ptrR mutant strain growth phenotype on glutamate as the nitrogen source in minimal medium (glycerol as the carbon source) has been detected for the growth. Cells showed a growth phenotype when the ptrR gene was deleted under the starvation conditions (Fig. S9); a decrease in the growth    www.nature.com/scientificreports/ rate was observed for the ptrR mutant. The mRNA level for sad was higher in the ptrR mutant at these conditions (Table S2, Table 4). The ptrR mutation led to the upregulation of 121 genes. The phenotype microarray using Biolog PM2A was tested under the microaerobic conditions for the ptrR mutant phenotype and the E. coli WT BW25113 strain. The phenotype using Glu as the energy/nitrogen source was minimal when D-glucosamine or dihydroxyacetone was the carbon source. The ptrR mutant defect in growth/respiration with glycine, l-ornithine, or gamma-hydroxybutyrate was observed using M9 medium with l-glutamate as the sole nitrogen source. Phenotypes for the ptrR mutant with d-tagatose, oxalomalic acid, gamma-hydroxybutyrate, glycine, and l-alaninamide were observed under the same conditions with Glu as a supplement (Fig. S10). The regulatory effect of PtrR during aerobic growth with putrescine/Glu as the nitrogen source was detected for M9 medium with glycerol as the primary carbon source. The growth of BW25113 (WT) as well as ptrR, yneH (glsB) null mutant strains are shown in Fig. 5B-C. The E. coli WT strain had a longer lag-phase compared to the ptrR mutant. A growth deficiency for a glsB mutant was observed under the same conditions, suggesting a functional relationship between GlsB (YneH) and Sad, encoding genes conserved in genome clusters with ptrR. The effect of a yneH deletion was substantial as cells approached the stationary phase.
PtrR-dependent regulation during growth with Ptr/Glu as nitrogen sources and antibiotic resistance. The E. coli WT and ptrR mutant were grown aerobically in M9 medium with 20 mM Glu and Ptr as nitrogen sources and 0.2% glycerol. To determine the effect of the ptrR deletion mutation, the cells were collected at the log-phase, and total mRNA was purified (see Materials and methods). PuuR and PuuADE, SodB, and two copper related transport systems' mRNA levels increased in the ptrR mutant strain (Table 5). SodB (superoxide dismutase) mRNA was increased in ptrR mutant and SodB produced H 2 O 2 . The copA and cus operons are regulated by the CusSR and HprRS system. H 2 O 2. is the effector for HprRS and likely has a transcriptional effect for the copA and cus system. Antibiotic resistance induced by ptrR mutation in E. coli BW25113 was detected. We propose that PtrR negatively controls the FnrS small RNA, which is involved in regulation of MarA mRNA. MarA is a global regulator of E. coli genes involved in resistance to antibiotics, oxidative stress, organic solvents, and heavy metals 30 . We tested the ptrR mutant and wild type E. coli strains for antibiotic resistance using the Biolog plate 11C 31 . Since FnrS is under positive control of the global anaerobic regulator Fnr, E. coli was grown under microaerobic conditions. The ptrR mutant showed increased resistance to high concentrations of demeclocycline, a tetracycline group antibiotic, which survived after 42 h, in contrast to the wild type. We also detected the increased resistance of  www.nature.com/scientificreports/ the ptrR mutant to chlorotetracycline (another tetracycline analog) (Fig. S11). However, with other antibiotics tested, no significant difference in growth of the mutant and wild type strains was observed.

Discussion
In this study, we applied a systems approach to characterize the transcriptional responses of seven putative LTFs: YbdO (CitR), YbeF, YbhD, YcaN, YiaU (LpsR), YgfI (DhfA), and YneJ (PtrR) ( Table 6). The transcriptional response for the deletion of each LTF had been detected by RNAseq in M9 minimal medium supplemented by 7 mM l-threonine for all yTFs, except PtrR. The transcriptional analysis for the ptrR deletion mutant was detected in the M9 medium with Ptr and/or Glu as nitrogen source. For LTFs, conserved adjacent genes had been shown to be differentially expressed. For example, gene clusters encoding ybdNM and citrate utilization genes citCDEF (citrate lyase) were detected as transcriptionally regulated in the citR deletion mutant. The CitR DNA-binding site upstream of citR has been predicted and confirmed by ChIP-exo assay, suggesting autoregulation. CitR has been shown to be important for the growth in minimal medium supplemented by citrate at acidic conditions, suggesting citrate lyase regulation microaerobically. Additionally, flagella biosynthesis genes (FlhDC regulon) were differentially expressed in the mutant. The citR mutant in E. coli BW25113 fitness phenotype had been previously shown for motility in LB medium. Additionally, CitR is important for E. coli BW25113 fitness in minimal medium with glycolate as the carbon source (fit.genomics.lbl.gov), and D-glycine as the nitrogen source. The phenotype for formate and l-glutamate utilization citR deletion mutant was detected microaerobically (Fig. 7). YbeF is conserved in the gene cluster with citrate lyase encoding genes. The ybeF deletion leads to lrhA gene downregulation and upregulation of FlhDC regulated genes and, accordingly, the FlhDC iModulon. The YgfI (DhfA) DNA-binding site upstream of the dhaKLM operon for dihydroxyacetone phosphotransferase was detected by ChIP-exo. The transcriptome analysis shows that DhfA was important for regulation of dhaKLM, pflB, hycBCDEFG, narZYWV and adhB. The dhfA mutant growth phenotype using glycerol as a carbon source had been detected (Fig. 3C). The common DNA-binding motif upstream of the dhak, pflB, adhE, hycB, and narZ genes was found, but future experiments to confirm DhfA binding to the predicted DNA-binding sites are essential ( Table 6). The PflB and hycBCDEFG encoded hydrogenlyase are involved in pyruvate and Thr utilization as energy source (Fig. 3F). The dhfA mutant growth deficiency on glucose as the carbon source in minimal medium at microaerobic conditions had been shown, but supplementation by Thr reduced the growth phenotype (Fig. S7), suggesting DhfA dependent pflB and hycBCDEFG activation important for anaerobic metabolism. dhfA deletion mutant phenotype had been detected in microaerobic conditions for formate utilization in minimal medium after incubation at 37 °C for 24 h (Fig. 7).   The maximum likelihood phylogenetic tree was constructed using RAxML. The distinct genomic context of ptrR genes from two major tree branches (groups of PtrR orthologs) is shown by arrows with the following colors: black (ptrR regulator), yellow (sad for succinate semialdehyde dehydrogenase), green (yneH for glutaminase), pink (mcp for methyl-accepting chemotaxis protein), and blue (mfs for putative MFS-family transporter), while the predicted PtrR-binding sites are shown by black dots. Sequence logos of predicted DNA binding sites of PtrR from each of the two groups of analyzed species were constructed using WebLogo 2.0. www.nature.com/scientificreports/  www.nature.com/scientificreports/ We suggest re-name YiaU to LpsR, lipopolysaccharide biosynthesis regulator. ChIP-exo detected multiple LpsR DNA-binding sites (Fig. 4A). The RNAseq transcriptomic analysis and ChIP-Exo (LpsR DNA-binding) additionally detected direct regulation of the waaPSBOJYZU operon gltB and adeP genes, suggesting LpsR relationship to glutamate metabolism. lpsR deletion mutant transcriptomic analysis suggested the regulation of yiaT, the conserved adjacent gene to lpsR, which is divergently transcribed and yeaK, encoding deacylase for mischarged aminoacyl-tRNA (l-serine, l-threonine) ( Table 3, Fig. 4D). Thr supplement in minimal medium could lead to mischarged tRNA and LpsR is important for the yeaK regulation in the presence of Thr. The lpsR deletion leads to increased sensitivity to oxacillin in RPMI_10LB medium microaerobically and LpsR was found to have the effect on the biofilm formation previously 3 .

Scientific Reports
The ybhD adjacent gene ybhH conserved in enterobacterial genomes was upregulated in the ybhD deletion mutant. The ybhH, ybhI, and ybhD genes are conserved adjacent genes and potentially regulated by Nac (EcoCyc). YbhI is the putative tricarboxylate transporter, homologous to 2-oxoglutarate/malate translocator, (id. 35%) (Fig. S8) 32 . We detected the ybhD deletion strain growth phenotype in the M9 minimal medium with glycerol (carbon source), supplemented by l-malate (Fig. S9), but not in the absence of l-malate, suggesting that the de-repressed ybhI and ybhH genes are probably involved in l-malate utilization ( Table 6).
We detected that PtrR (YneJ) is the transcriptional regulator for Sad and the small RNA, FnrS. PtrR was predicted to be a repressor of the fnrS gene, encoding a small regulatory RNA (Table 6). We demonstrated PtrR binding to the predicted DNA-binding site. According to RNA-Seq data, PtrR is a repressor for sad under the nutrient limitation-stress conditions. The ptrR gene deletion effect was shown by growth phenotype (aerobically) and phenotype microarray data (micro-aerobically).
PtrR-mediated regulation appears to be important for Ptr utilization as an energy source. A pleiotropic effect of the PtrR-dependent regulation of the sad gene under nitrogen/carbon starvation has been investigated and discussed. The known stress/starvation sigma σ S -controlled csiD-ygaF-gabDTP region is related to GABA utilization, while Sad is important for Ptr utilization 25,[33][34][35] .
Extracellular Ptr alters the OmpF porin charge and pore size, resulting in partial pore closure and a consequent decrease in outer membrane permeability 15,36 . Our results demonstrated that PtrR is important for the growth of the E. coli BW25113 strain with Glu as the sole nitrogen source and glycerol as the carbon source and resistance to the tetracycline group of antibiotics (i.e., demeclocycline and chlortetracycline), but not to chloramphenicol, erythromycin, and other antibiotics. PtrR is potentially important for the regulation of the highly conserved, anaerobically induced small RNA-fnrS, which is likely important for regulation under anaerobic  www.nature.com/scientificreports/ growth conditions 37 . Interestingly, a ptrR mutant was shown previously to be resistant to bacteriophage lambda infection 38 and we found that PtrR is potentially related to tetracycline resistance 39 . ChIP-exo and RNAseq results have been analyzed, providing a hypothesis for the physiological functions of YneJ (tentatively re-named PtrR, putrescine related regulator), YgfI (tentatively re-named DhfA, dihydroxyacetone phosphotransferase and formate utilization activator), and YbdO (tentatively re-named CtrR, citrate utilization related regulator). The identification of the additional DNA-binding sites for YgfI, YcaN, YiaU, YbeF, YneJ by gSELEX (genomic SELEX) method in the presence of Thr in the minimal medium possibly will provide additional information about novel LTFs transcriptional regulatory network 6,40 . LTFs are not always expressed under laboratory growth conditions (for instance, see Ishihama et al. J. Bacteriol. 196, 2718-2727, 2014.
Taken together, the systems analysis of the E. coli BW25113 and MG1655 strains transcriptomic data, ChIPexo DNA-binding data for LTF, and the regulated biochemical pathway reconstruction and fitness/phenotype of the LTF deletion mutants strains produce fruitful hypotheses for the yTF function prediction that is important for TRN reconstruction in E. coli.

Methods
RNA sequencing. The wild type BW25113 strain was grown as a control for the isogenic ptrR mutant strain.
Pre-cultures were obtained by scraping frozen stocks and growing the cells in LB medium. Cells were washed twice with M9 medium and inoculated to an OD 600 of 0.05. The cells were collected at an OD 600 of 0.9 (only ycaN, yiaU, ybhD mutants and the WT control were collected at OD 600 of 2.0; late-exponential phase) and were harvested using the Qiagen RNA-protect bacteria reagent according to the manufacturer's specifications. Pelleted cells were stored at − 80 °C, and after cell resuspension and partial lysis, they were ruptured with a beat beater; the total RNA was extracted using a Qiagen RNA purification kit. After total RNA extraction and subsequent ribosomal RNA removal, the quality was assessed using an Aglient Bioanalyser using an RNA 6000 kit. The data processing is described in Supplemental materials.
Protein purification. The PrtR-producing strain was grown overnight, re-inoculated into 50 mL of the fresh medium, and induced with 0.6 mM IPTG after an OD 600 of 0.6 was reached. The cells were harvested after 4 h and lysed, the cell pellet was resuspended in the lysis buffer. Rapid purification of recombinant proteins on Ni-nitrilotriacetic acid-agarose minicolumns was performed. The protein was refolded on a mini-column, and the buffer was changed to a buffer containing 0.1 M Tris-HCl, 0.1 M NaCl, 10 mM urea.
Fluorescent polarization assay. The purified PtrR protein and 10 nM fluorescently labeled DNA fragment (5′-gggTTC TCG ATT CGT GAAggg-3′) were incubated in the assay mixture. The PtrR binding assay mixture (0.1 ml) contained Tris buffer, pH 7.5, 0.1 M NaCl, 10 mM MgSO 4 , 5 mg/ml sperm DNA and 1uM of the fluorescently labeled predicted PtrR binding DNA fragment as well as 0-0.6 mM GABA. Then the PtrR protein (0-1.5 uM) was added to the assay mixture, and it was incubated for 1 h at 30 °C in the presence or the absence of GABA.
Targeted high-performance liquid chromatography. For organic acid and carbohydrate detection, samples were collected after 4 h for every 30-45 min. The filtered samples were loaded onto a 1260 Infinity series (Agilent Technologies) high-performance liquid chromatography (HPLC) system with an Aminex HPX-87H column (Bio-Rad Laboratories) and a refractive index detector and HPLC was operated using ChemStation software. The HPLC was run with a single mobile phase composed of HPLC grade water buffered with 5 mM sulfuric acid (H 2 SO 4 ). The flow rate was held at 0.5 mL/minute, the sample injection volume was 10 uL, and the column temperature was maintained at 45 °C. The identities of compounds were determined by retention time comparison to standard curves of acetate, ethanol, glucose, lactate, pyruvate, formate and succinate. The peak area integration and resulting chromatograms were generated within ChemStation and compared to that of the standard curves to determine the concentration of each compound in the samples.
Phenotype microarray. The E. coli BW25113 wild type and ybdO, ygfI, ptrR mutant strains were grown overnight in M9 glucose medium, washed with M9 medium (PM1, PM2) or RPMI with 10% LB (RPMI_LB) for PM12B and inoculated as recommended to the Omnolog plates PM1, PM2, PM11C or PM12B, for the antibiotic resistance measurements at 37 °C. The experiments were repeated two times.