* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
Download Word
Paracrine signalling wikipedia , lookup
Interactome wikipedia , lookup
Lipid signaling wikipedia , lookup
Signal transduction wikipedia , lookup
Gene regulatory network wikipedia , lookup
Fatty acid metabolism wikipedia , lookup
Biochemical cascade wikipedia , lookup
Metalloprotein wikipedia , lookup
Western blot wikipedia , lookup
Endogenous retrovirus wikipedia , lookup
Evolution of metal ions in biological systems wikipedia , lookup
Point mutation wikipedia , lookup
Genetic code wikipedia , lookup
Protein–protein interaction wikipedia , lookup
Magnesium transporter wikipedia , lookup
Expression vector wikipedia , lookup
Protein structure prediction wikipedia , lookup
Artificial gene synthesis wikipedia , lookup
Biosynthesis wikipedia , lookup
Two-hybrid screening wikipedia , lookup
Proteolysis wikipedia , lookup
1 Evidence That COG0325 Proteins are involved in PLP Homeostasis 2 Laurence Prunettia, Basma El Yacoubia, Cara R. Schiavona, Ericka Kirkpatricka, Lili Huangb, 3 Marc Baillya*, Mona ElBadawi-Sidhuc, Katherine Harrisona, Jesse F. Gregory 3rd b, Oliver 4 Fiehnc, Andrew D. Hansond and Valérie de Crécy-Lagarda# 5 6 Department of Microbiology and Cell Science, Institute for Food and Agricultural Sciences and 7 Genetic Institute, University of Florida, Gainesville, Florida, USAa; Department of Food Science 8 and Human Nutrition, University of Florida, Gainesville, Florida, USAb; Department of 9 Molecular and Cellular Biology & Genome Center, University of California, Davis, California, 10 USAc; Department of Horticultural Sciences, University of Florida, Gainesville, Florida, USAd 11 12 * Current address: Merck, Palo Alto California, USA 13 L. P. and B. E.-Y. contributed equally to this work 14 #Address correspondence to Valérie de Crécy-Lagard, [email protected] 15 16 Running title: Function of E.coli yggS 17 18 Keywords 19 yggS; COG0325; vitamin B6; PLP protein; PROSC; pyridoxine toxicity; protein of unknown 20 function. 21 22 23 Subject category: Physiology and metabolism 24 Word count 25 Abstract= 195 26 Main text (excluding abstract, and references) =5695 27 28 Abbreviations 29 GSA: Glutamic 5-semialdehyde; P5C: Δ1-pyrroline-5 carboxylic acid;; THF: Tetrahydrofolate; 30 CH2-THF: 5,10-methylene-tetrahydrofolate; Gcv: Glycine-cleavage complex; PLP: Pyridoxal 5’- 31 phosphate; PL: pyridoxal; PN: pyridoxine; PM: pyridoxamine; PNP: Pyridoxine 5’-phosphate; 32 PROSC: PROline Synthase Co-transcribed homolog; ALR: alanine racemase; ODC: ornithine 33 decarboxylase; meth: methionine; dT: thymidine; Cm: chloramphenicol; Amp: ampicillin.; Kan: 34 kanamycin; Tet: tetracycline; aTet: anhydrotetracycline. 35 36 37 ABSTRACT 38 Pyridoxal 5'-phosphate (PLP) is an essential cofactor for nearly 60 Escherichia coli enzymes but 39 is a highly reactive molecule that is toxic in its free form. How PLP levels are regulated and how 40 PLP is delivered to target enzymes are still open questions. The COG0325 protein family 41 belongs to the fold-type III class of PLP enzymes and binds PLP but has no known biochemical 42 activity although it occurs in all kingdoms of life. Various pleiotropic phenotypes of the 43 Escherichia coli COG0325 (yggS) mutant have been reported, some of which were reproduced 44 and extended in this study. Comparative genomic, genetic and metabolic analyses suggest that 45 these phenotypes reflect an imbalance in pyridoxal 5'-phosphate (PLP) homeostasis. The E. coli 46 yggS mutant accumulates the PLP precursor pyridoxine 5'-phosphate (PNP), and is sensitive to 47 an excess of pyridoxine but not of pyridoxal. The pyridoxine toxicity phenotype is 48 complemented by the expression of eukaryotic yggS orthologs. It is also suppressed by the 49 presence of amino acids specifically isoleucine, threonine and leucine suggesting the PLP 50 dependent enzyme transaminase B (IlvE) is affected. These genetic results lay a foundation for 51 future biochemical studies of the role of COG0325 proteins in PLP homeostasis. 52 53 Introduction 54 Pyridoxal 5’-phosphate (PLP) is one of six interconvertible vitamin B6 species (pyridoxal or PL, 55 pyridoxine or PN, pyridoxamine or PM and their 5'-phosphate forms). Enzymes utilizing PLP as 56 cofactor are found in all organisms, catalyze diverse reactions including transamination, 57 decarboxylation, racemization and β- and - elimination, and are mostly associated with amino 58 acid metabolism (Percudani and Peracchi 2009). PLP-enzymes belong to seven structurally 59 distinct families (fold types I-VII) that probably arose independently (Christen and Mehta 2001; 60 Percudani and Peracchi 2003) and encompass 184 enzyme activities. PLP metabolism is 61 particularly relevant to human health because several disorders have been linked to PLP 62 deficiency (Clayton 2006; Halsted 2013; Paul et al. 2013). 63 While the biosynthesis, interconversion and salvage pathways for vitamin B6 species are 64 well characterized (Fitzpatrick et al. 2010; Mooney and Hellmann 2010; Herrero et al. 2011; 65 Mukherjee et al. 2011; Sang et al. 2011; Rueschhoff et al. 2013; Szydlowski et al. 2013), little is 66 known about the regulation of PLP synthesis or about the connection between PLP and general 67 metabolism (Shi et al. 2002; Chen and Xiong 2005; Titiz et al. 2006; Rueschhoff et al. 2013; 68 Vanderschuren et al. 2013). How PLP molecules are delivered to their target enzymes and how 69 the free/bound PLP pool is regulated are also poorly understood (di Salvo et al. 2012). PLP 70 reacts with apo-B6 enzymes by forming an aldimine linkage with the ε-amino group of the active 71 site lysine residue to produce the catalytically active holo-B6 enzyme forms. Alternatively, its 72 highly reactive 4′-aldehyde group can spontaneously form unwanted aldimines with the ε-amino 73 group of lysine residues of non-B6 proteins and with many other amines, and thiazolidine 74 adducts with sulfhydryl groups of molecules such as cysteine, potentially leading to enzyme 75 inactivation (Ohsawa and Gualerzi 1981; Dong and Fromm 1990; Mizushina et al. 2003; 76 Vermeersch et al. 2004) and accumulation of damaged metabolites. For example, PLP reacts 77 through a Knoevenagel condensation with Δ1-pyrroline-5-carboxylate (P5C) (Fig. 1) and Δ1- 78 piperidine-6-carboxylate (P6C) (intermediates in proline and lysine metabolism, respectively) 79 that accumulate in patients deficient in P5C-dehydrogenase and in α-aminoadipic 80 semialdehyde/P6C dehydrogenase, respectively, leading to PLP deficiency (Fig. 1) (Farrant et al. 81 2001; Clayton 2006; Mills et al. 2006). 82 The high chemical reactivity of PLP requires a tight control of the free PLP pool such 83 that an appropriate supply is available to apo-B6 enzymes while undesirable interactions are 84 minimized. It has been proposed that PLP availability is regulated through product feedback 85 inhibition and tight binding to PL kinase (PdxK), PNP oxidase (PdxH) as well as PLP synthase 86 (PdxJ) (Yang and Schirch 2000; Moccand et al. 2011; Ghatge et al. 2012). However, a precise 87 model of PLP homeostasis and how PLP is delivered to target enzymes is still lacking in any 88 organism and is required to understand and manage PLP-related diseases. 89 A candidate for a missing player in PLP homeostasis is the YggS/PROSC/YBL036C 90 family (COG0325). This family belongs to the fold-type III class, along with alanine racemases 91 and certain decarboxylases (Percudani and Peracchi 2009), and has been shown to bind PLP 92 (Eswaramoorthy et al. 2003). Crystal structures of the yeast and Escherichia coli COG0325 93 proteins have been determined [(Eswaramoorthy et al. 2003) and PDB id 1W8G]. Like the N- 94 terminal domain of alanine racemase (ALR) and ornithine decarboxylase (ODC), this protein 95 folds as a TIM barrel with a characteristic long N-terminal helix, and binds PLP in a similar 96 mode. Unlike the dimeric ALR, YBL036C was found to be monomeric, and it was recently 97 reported that the E. coli member of the family, YggS, has no racemase activity towards any of 98 the 20 protein amino acids or their D enantiomers (Ito et al. 2013). 99 Both the broad phylogenetic distribution and the pleiotropic phenotypes linked to the 100 mutations of members of the COG0325 family are suggestive of a core conserved function. In E. 101 coli, the ΔyggS ΔglyA double mutant is not viable on LB medium (Nichols et al. 2011). GlyA 102 (SHMT, serine hydroxymethyltransferase) converts serine to glycine and in the process transfers 103 a hydroxymethyl group to tetrahydrofolate (THF) forming 5,10-methylene-tetrahydrofolate 104 (CH2-THF), the major source of C1 units in the cell (Green et al. 1996) (Fig. 1). GlyA, and the 105 glycine cleavage enzyme system (Gcv), another source of one carbon units, are key to the 106 biosynthesis of purines, thymidine, methionine, and lipids (Fig. 1). Mutants in the THF pathway 107 are sensitive to sulfonamides (targeting FolP) and trimethoprim (targeting FolA) (Nichols et al. 108 2011). Like mutants in the Gcv system, the ΔyggS strain is sensitive to sulfonamides and not to 109 trimethoprim while the serine hydroxymethyltransferase mutant (ΔglyA::KanR) is sensitive to 110 trimethoprim and not to sulfonamides (Nichols et al. 2011). Recently, ΔyggS strains were shown 111 to display altered intracellular amino acid and acetyl coenzyme A (CoA) pools, and to excrete L- 112 valine in the culture medium while accumulating pyruvate, 2-ketobutyrate and 2-aminobutyrate 113 (Ito et al. 2013). Another connection with amino acid metabolism comes from Pseudomonas 114 aeruginosa, whose yggS homolog was proposed to be co-transcribed with proC, the gene 115 encoding P5C reductase that catalyzes the last step of proline synthesis (Fig. 1). This observation 116 gave the PROline Synthase Co-transcribed homolog (PROSC) name to the first studied members 117 of the family including the human homolog, which is expressed in all organs (De Wergifosse et 118 al. 1994). Finally, in yeast, the COG0325 protein YBL036C was induced three-fold in response 119 to the DNA-damaging agent, methyl methanesulphonate (MMS), implying involvement in 120 processes ensuring genetic integrity (Lee et al. 2007). 121 In this work, genetic, biochemical and comparative genomics approaches were combined 122 to show that COG0325 family proteins have a role in PLP homeostasis that could explain the 123 pleiotropic phenotypes of the yggS strain. 124 125 MATERIAL AND METHODS 126 Bioinformatic analyses. The BLAST tools (Altschul et al. 1997) and resources at NCBI 127 (http://www.ncbi.nlm.nih.gov/) were routinely used. Multiple sequence alignments were built 128 using Clustal Omega (Li et al. 2015) or Multalin (Corpet 1988). Protein domain analysis was 129 performed using the Pfam database tools (Finn et al. 2014). Analysis of phylogenetic distribution 130 and physical clustering was performed in the SEED database (Overbeek et al. 2014). Results are 131 available 132 (http://pubseed.theseed.org/SubsysEditor.cgi). The representative genome sets ( set 1 of ~1000 133 genomes and set 2 of ~1500 genomes) were chosen based on phylogenetic diversity 134 previously described (Dailey et al. 2015; Niehaus et al. 2015).The Ortho-MCL database (Chen et 135 al. 2006) was queried for the analysis of COG0325 orthologs in Eukarya and BLASTp (Altschul 136 et al. 1997) searches were performed specifically against sequences of archaeal genomes. 137 Physical clustering was analyzed with the SEED subsystem coloring tool or the SeedViewer 138 Compare Regions tool (Overbeek et al. 2014). The protein association network analysis was 139 performed on the STRING database (string-db.org/) (Szklarczyk et al. 2015). The Enzyme 140 Function Initiative-Enzyme Similarity Tool (EFI-EST) (http://enzymefunction.org/) was used to 141 extract a physical clustering network (Gerlt et al. 2015) as follows. The amino acid sequences of 142 proteins of the IPR011078 InterPro family was extracted to generate a sequence similarity 143 network with an original alignment score threshold of 30 and no restrictions for alignment in the “YggS_2015_Minimal” subsystem on the public SEED server as 144 lengths. A 65% sequence identity representative node network for IPR011078 was also edited in 145 Cytoscape (Shannon et al. 2003) to produce multiple networks with alignment score thresholds 146 of 40, 50, 60, 70, 80, 90 and 100 by deleting all edges with –log10 (E) values below those 147 thresholds. These seven networks were then used to generate genome neighborhood networks 148 using the EFI Genome Neighborhood Tool with default parameters. The top 11 Pfam protein 149 families in the genome neighborhood networks with alignment scores of 70 and 80 were 150 identified by restricting the network to Pfam protein families with ≥3,000 neighbors. These 151 specific networks were chosen because they are the ones that separate into the most clusters 152 before the network begins to disintegrate into mostly single nodes. The Interactive Tree of Life 153 v2 (ITOL) platform was used to build the gene distribution trees (http://itol.embl.de/index.shtml) 154 (Letunic and Bork 2011). Gene essentiality data was extracted from the Database of Essential 155 Genes (DEG) database (Luo et al. 2014) (http://www.essentialgene.org/). Structures were 156 visualized with the Protein Data Bank (PDB) tools (www.rcsb.org) (Berman et al. 2000). The 157 distribution 158 (http://bioinformatics.unipr.it/B6db) (Percudani and Peracchi 2009). The id-mapping tools of 159 Ecogene 160 (http://www.uniprot.org/) (Li et al. 2015) where routinely used..The positions of regulatory sites 161 for the PdxR family were extracted from RegPrecise 3.0 (http://regprecise.lbl.gov) (Novichkov et 162 al. 2013). The subcellular localization of plant COG0325 proteins was predicted using TargetP 163 (Emanuelsson et al. 2007). Prediction of E. coli promoters was done using RegulonDB (Salgado 164 et al. 2013). 165 Growth conditions and media. Bacteria were grown on Luria Bertani (LB) medium (BD 166 Diagnostics Systems) at 37 °C or on M9 minimal medium (Sambrook et al. 1989) unless of PLP enzymes in (http://www.ecogene.org/) E. coli (Zhou was and extracted Rudd from 2013) the B6 and database Uniprot 167 otherwise stated. Growth media were solidified with agar (15 g/l) (BD Diagnostics Systems) for 168 the preparation of plates. Transformation and P1 transductions were performed following 169 standard procedures (Moore 2011). The sensitivity to P1 phage of all recipient strains used was 170 verified. Anhydrotretracycline (aTet, 50 ng/ml), Ampicillin (Amp, 100 µg/ml), Kanamycin (Kan, 171 50 µg/ml), Spectinomycin (Sp, 50 µg/ml) and Chloramphenicol (Cm, 30 µg/ml) were used as 172 appropriate. ilvE avtA::KanR (LSP5001) and ilvE::KanR avtA (LSP5001) were grown on 173 VBE minimal medium 0.5% glucose (w/v) (Whalen and Berg 1982). 174 175 Strain and plasmid constructions. All strains and plasmids used in this study are listed in Table 176 S1 and all oligonucleotides in Table S2. Details of the constructions are described in the 177 supplemental methods. 178 179 Effect of PN on yggS. yggS (VDC6594) cells were grown overnight in 5 ml of M9 glucose 180 0.2% (w/v), diluted 500-fold in 5 ml of M9 medium and plated on M9 glucose 0.2%; 20 l drops 181 of PN at concentrations of 5.9 mM, 590 M, 295 M or 59M were set on the top of the agar 182 and plates were incubated overnight at 37 °C. For complementation experiments, the yggS 183 strain was transformed freshly for every experiment. The plasmids used are listed in Table S1. 184 Cells transformed with pBAD18, pBAD24yggSEc (pBY291.3) were plated on M9 glucose 0.2% 185 (w/v) arabinose 0.2% (w/v), ampicillin 100 g/ml. Cells transformed with pBAD18 186 LOC100191932 were plate on M9 glucose 0.2% (w/v) in presence or in absence of arabinose 187 0.2% (w/v) ampicillin 100 g/ml. Finally, yggS were transformed with pBEY YBL036C 188 (pBEY329.12), pBEY279.1 (empty vector) in the presence or absence of aTet 50 ng/ml. 189 190 Bioscreen growth curves. Cells were grown at 37 °C overnight in M9 supplemented with 0.2% 191 glucose (w/v) and 1 mM glycine, starting from an optical density (OD) of 0.05 at 600 nm. Cells 192 were then diluted 50-fold into sterile 100-well honeycomb plates with cover (Labsystems) 193 containing the corresponding medium: M9 glucose 0.2% (w/v) supplemented or not with glycine 194 1 mM, LB supplemented or not with methionine (0.13 mM) and 0.16 mM thymidine (dT), M9 195 glucose 0.2% (w/v) and glycine 1 mM supplemented or not with 0.4% casamino acids, or with 196 0.13 mM methionine and 0.16 mM thymidine. The OD at 600 nm was measured using a 197 Labsystems Bioscreen C plate reader. Cells were grown at 37 °C with vigorous shaking. Time 198 points were recorded every 30 minutes. For growth in presence of 20 different amino acids, cells 199 were grown in M9 glucose 0.2% (w/v) supplemented with 1 mM glycine and 0.2 mg/ml or 0.1 200 mg/ml of the corresponding amino acids. All growth experiments were performed in 10 201 replicates and standard deviations (SD) were determined 202 203 Extraction and vitamin B6 analysis. B6 species (PLP, PNP, PMP, PL, PN and PM) were 204 extracted from wild-type and yggS pellets (1.0 ml culture of OD600=1.0) in 0.9 ml of 5% (w/v) 205 metaphosphoric acid and 0.1 ml of internal standard 4-deoxypyridoxine (4-dPN, 73 nmol/ml in 206 5% metaphosphoric acid). The suspension was vortexed and sonicated. After centrifugation at 207 10,000 g for 15 min, the supernatant was filtered and a 50-µl sample was taken for HPLC 208 analysis using fluorescence detection (excitation 328 nm, emission 393 nm). The separation was 209 performed on a Microsorb-MV C18 column (150 × 4.6 mm, 5 µm particle size) using a gradient 210 program described by Sampson et al with some modifications (Sampson and O'Connor 1989). 211 Mobile phase A (0.033 M phosphoric acid and 0.008 M 1-octanesulfonic acid, adjusted to pH 2.2 212 with KOH), B (0.033 M phosphoric acid, adjusted to pH 2.2 with KOH) and C (acetonitrile) 213 were used and the gradient program was as follows: 98% A and 2% C for 10 minutes; a linear 214 gradient to 78% B and 22% C for 8 minutes; a linear gradient to 98% A and 2% C for 2 min; 215 column equilibration for 5.0 min with 98% A and 2% C. Total running time was 25 min and the 216 flow rate was 1.2 ml/min. A post-column reagent (1.0 mg/ml sodium bisulfite in 1.0 M 217 potassium phosphate buffer adjusted to pH 7.5 with KOH) was used to enhance fluorescence of 218 PLP. All activities were measured at least three times and standard deviations were determined. 219 220 Chemicals. PNP was provided by the Vanderbilt Institute of Chemical Biology, Chemical 221 Synthesis Core, Vanderbilt University, Nashville, TN 37232‐0412. 222 223 Results 224 yggS is widely distributed and clusters strongly with genes in diverse metabolic and cellular 225 pathways. 226 To gain further insight in the role of the COG0325 family, a subsystem was constructed in the 227 SEED database to capture all COG0325 members (named YggS_2015_Minimal). The yggS gene 228 is widely distributed in Bacteria and Eukaryotes but is quite rare in Archaea where only 229 Methanosarcinales and Aciduliprofundum sp. MAR08-339 harbor yggS orthologs. Members of 230 this family are present almost universally among Bacteria with >90% of the bacterial species in 231 the SEED database containing at least one yggS homolog (see Subsystem and Fig. 2A). A few 232 Bacteria harbor yggS paralogs and these are scattered around the phylogenetic tree (Fig. 2A). 233 COG0325 genes are present in most (but not all) Eukaryotes including yeast, Caenorhabditis 234 elegans, fruit fly, Arabidopsis, Z. mays, zebrafish, chicken, and mammals including humans [see 235 group OG5_127174 in the OrthoMCL database]. 236 Physical clustering was first analyzed using the STRING and SEED databases and is 237 summarized in Table 1. The clustering between COG0325 genes and proC observed decades ago 238 (De Wergifosse et al. 1994) was confirmed; clustering between yggS and proC was observed in 239 ~20% of the genomes analyzed (3% in the 1000 genome set) (Table 1). However, our data 240 showed that the link between yggS and proline synthesis is not robust (Fig. S1A). First, although 241 proABC genes are quite often clustered, such clusters never include yggS and some Bacteria such 242 as Helicobacter pylori have proAB homologs but lack yggS homologs (Fig. S1A). Second, 243 several bacteria have two proC genes, one that clusters with proAB and another that clusters with 244 yggS (Fig. S1B). 245 Our analysis uncovered other physical clustering associations. The strongest association 246 was with cell division/cell wall related genes (Table 1). The yggS gene is often located near the 247 end of the well-known division and cell wall (dcw) cluster (Tamames et al. 2001) and also 248 clusters independently with two genes also often located at the end of the dcw cluster yggT 249 (YlmG) and sepF (Table 1). Physical clustering associations also linked YggS to PLP salvage 250 and to the PLP dependent enzyme GlyA (Table 1 and Fig. S2). Indeed, yggS genes were found 251 associated with pdxK, glyA and B6 transporter genes in operons predicted to be under the control 252 of various PdxR-type PLP-responsive transcription regulators (Jochmann et al. 2011; Suvurova 253 and Rodionov 2015; Tramonti et al. 2015) (Fig. S1B and Fig. S2). Finally, genes encoding 254 essential ATP-dependent enzymes (with essential lysines) were found to cluster strongly with 255 yggS (Table 1). Examples include the leucyl- and isoleucyl-tRNA synthetase genes that in 256 combination clustered with yggS in ~25-30% of the genomes. 257 258 Vitamin B6 homeostasis defects are observed in the yggS strain. 259 YggS is a PLP-binding protein (Eswaramoorthy et al. 2003), and our comparative genomic 260 analyses linked members of the YggS family to vitamin B6 synthesis (Table 1, Fig. S2). To 261 explore a potential role of this protein family in vitamin B6 homeostasis in vivo, we tested 262 whether the absence of yggS led to phenotypes in conditions of B6 excess. An E.coli K12 yggS 263 derivative VDC6594 was constructed and tested for response to B6 vitamers. Excess PN led to a 264 toxicity ring on minimal medium in the yggS but not in the wild-type strain (Fig. 3, Table 2 and 265 Table 3). PLP, PM or PL were not toxic in any background (data not shown). This PN toxicity 266 ring was very reproducible (Table 2) and was complemented by expressing the E. coli yggS gene 267 in trans (Fig. 3C) and suppressed by the presence of PL (Table 2). The presence of ring, and not 268 a halo, couldsuggest that suppressors appeared. To discriminate between the two hypotheses, we 269 re-isolated yggS cells from the inside or from the outside of the PN toxicity ring and both 270 retained the PN sensitivity phenotype (Fig. S3A). We also showed that overexpressing pdxK led 271 to PN toxicity in the WT strain (Fig. S3B), the absence of yggS did not seem to increase this 272 toxicity (data not shown). 273 The universality of the YggS function first reported by Ito et al. (Ito et al. 2013) was 274 confirmed here, as the PN toxicity phenotype was complemented by expressing the COG0325 275 gene from Z. mays, LOC100191932, in the presence of inducer (Fig. 3D). No complementation 276 was observed in absence of inducer (Fig. 3E). Similar results were found with the COG0325 277 gene from A. thaliana, At1g11930 (Table 2) but in addition it appeared that overexpression of 278 that gene was toxic (data not shown). Finally, it was found that expression of the yeast COG0325 279 gene, YBL036C, also complemented the PN toxicity phenotype (Fig. 3F). 280 As the growth phenotypes were consistent with a role of YggS in vitamin B6 281 homeostasis, we analyzed the B6 pools in the wild-type and yggS strains grown in M9 glucose 282 at the end of the exponential phase (Fig. 4A). The mutant accumulated PNP 71.64 ± 6.10 283 pmol/mg of proteins and no PNP is detected in the wild-type (Fig. 4B). No difference in the PLP 284 levels was observed between the yggS and the wild-type cells. Of note the total pool of PLP is 285 measured here not the free pool. Because the free PLP pool is only a small proportion of the total 286 pool (di Salvo et al. 2011), changes in the free pool will not be detected with this method, hence 287 it still to be determined if the absence of yggS affects the free PLP pool. Methods to measure the 288 PLP-ome or the PLP bound to proteins exists but they are all semi-quantitative (Whittaker et al. 289 2015). We still decided to compare the bound PLP (PLP to proteins) in the WT and yggS 290 strains using an antibody that detects PLP bound to proteins (Whittaker et al. 2015). The soluble 291 fractions of both strains were separated by 2D gels (Fig. S4A and B) and PLP-bound proteins 292 were detected using the anti-PLP antibody. 17 major protein spots were detected in both strains 293 and no major differences were observed (Fig. S4C and S4D). The absence of yggS does not have 294 a major impact on the bound PLP pool, but clearly the method is not quantitative enough to 295 detect small differences. 296 297 YggS is a monomeric PLP protein that has no direct effect on PdxH enzymatic activity. 298 The structure of E. coli YggS shows that PLP is covalently bound to lysine residue 36 299 through a Schiff base linkage (PDB: 1W8G). The role of PLP in YggS quaternary structure or 300 stability has yet to be determined. Recombinant YggS purified by Ni-affinity as described in the 301 supplemental methods and gel filtration chromatography migrated in SDS-PAGE as a single 302 band with a molecular mass of ~30 kDa (Fig. S5A). The monomeric state observed in the crystal 303 structure was confirmed by gel filtration chromatography, which indicated a native molecular 304 mass of 22.6 kDa (data not shown). The presence of PLP in the sample was detected by optical 305 spectroscopy (Fig. S5B); 32% of the purified YggS contained bound PLP. 306 Two forms of YggS were generated: apo-YggS with no detectable bound PLP, and native 307 holo-YggS in which PLP was reincorporated to the purified YggS (82% PLP). The holo- and 308 apo-YggS gave a symmetrical peak upon gel filtration that corresponded to the monomeric state 309 (data not shown). Thus, PLP had no effect on YggS oligomerization. The ability of apo-YggS to 310 bind PLP was confirmed by ITC with a Kd value was 0.37 ±0.49 (SD) M. 311 To determine whether YggS can bind PNP, a comparative analysis of thermal stability of 312 the purified YggS (with 32% of PLP binding sites occupied by PLP) was carried out by 313 thermofluor assay (Ericsson et al. 2006; Lavinder et al. 2009; Phillips and de la Pena 2011) in the 314 absence or presence of PNP or PLP (Fig. S5C and S5D). Only PLP was found to stabilize YggS 315 as the apparent Tm of the YggS was 86 °C with PLP compared to 78 °C without it. In presence of 316 PNP the apparent Tm of YggS was 78 °C. That PNP did not affect the thermal stability of YggS 317 even at high concentration (data not shown) indicates that YggS does not bind PNP. 318 PdxH (pyridoxine 5'-phosphate oxidase) mediates a key step in the PLP biosynthesis 319 pathway using PNP as substrate (Zhao and Winkler 1995; di Salvo et al. 2011) (Fig. 1). Free PLP 320 inhibits both human and E. coli PdxH and interacts with FMN when it is present (Musayev et al. 321 2003). (Fig. 4) A defect of the PNP oxidase activity could lead to the accumulation of PNP in 322 the yggS strain (Fig.4). This hypothesis was tested by measuring PNP oxidase specific activity 323 in the soluble fraction of the wild-type and the yggS strains as described in the supplemental 324 methods. No difference in PdxH specific activity was observed. The specific activities of 325 theyggS and wild-type were 2.9 ±0.02 (SD) and 2.11 ± 0.12 (SD) nmol h-1 mg-1. 326 327 328 329 Pleotropic effects of yggS depletion. Previous studies have linked the absence of yggS to amino acid pool imbalances, particularly the secretion of valine in E.coli MG1655 (Ito et al. 2013). 330 A full analysis of the intracellular and extracellular metabolite pools extracted from the 331 wild-type and yggS strains grown in minimal medium was therefore performed and the 332 metabolomes of the two strain compared by gas chromatography time-of-flight mass 333 spectrometry as described in the supplemental methods (Table S3 and Table S4). Among the 334 most striking differences observed were increases in putrescine, N-acetylputrescine, and 335 putrescine-related metabolites such as γ-aminobutyric acid (and its cyclized form γ- 336 butyrolactam) and 5’-deoxy-5’-thiomethyladenosine in the yggS strain (Table S3). Putrescine is 337 the major polyamine in E. coli (Cohen 1998) and the regulation of its catabolism is very 338 complex, being induced by different stresses including nitrogen limitation (Schneider et al. 339 2013); several enzymes in these pathways are PLP-dependent (Samsonova et al. 2003; Kurihara 340 et al. 2005; Samsonova et al. 2005). In the conditions used in our study, we did not observe 341 significant accumulation of valine in the yggS strain or valine release to the medium (Table S4, 342 and Table 3). Nor could valine secretion by the yggS mutant be detected by feeding the valine 343 requirements of the valine auxotrophs LSP5001 or 5002 (data not shown). 344 Another link between YggS, PLP, and amino acid metabolism was observed with the 345 synthetic lethality phenotype of the E. coli ΔyggS::CmR and ΔglyA::KanR alleles on the rich 346 medium LB (Nichols et al. 2011). GlyA is a PLP-dependent enzyme involved in glycine and 347 serine metabolism (Fig. 1). As this phenotype had only been observed in a high throughput 348 experiment, a yggSglyA derivative was constructed in a BW25113 background. The double 349 mutants did not grow on LB (Fig. S6 and Table 3) but grew on M9 glucose 0.2% (w/v) 350 supplemented with 1 mM glycine (Fig. 5A). The ΔglyA strain also showed a growth phenotype 351 in LB (Fig. S6) as previously reported (Nichols et al. 2011) but it was not as severe as the double 352 mutant. It was then shown that the LB sensitivity phenotype is caused by the presence of amino 353 acids in the medium, as addition of casamino acids 0.4% (w/v) (Fig. 5B and Table 3). Casamino 354 acids also affected growth of the glyA strain (VDC6664) but not of the yggS (VDC6594) or 355 wild-type (WT) (BW25113) strains (Fig. 5B). Growth was then followed with all 20 protein L- 356 amino acids (added individually in presence of 1 mM glycine). Out of these 20, threonine (0.84 357 mM) and alanine (2.24 mM) delayed growth of the double mutant compared to the glyA, yggS 358 and wild-type strains (Fig. S7A and S7B) even if the final growth rates were similar (~0.04 359 OD600nm.h-1). The growth defects with single amino acids were not as pronounced as with LB or 360 casamino acids, suggesting additive or synergistic effects. The addition of dT and methionine 361 also partially suppressed the phenotype (Fig. 5C). 362 Finally, because of these links with amino acid metabolism, we tested whether amino 363 acids affected the PN toxicity phenotype of the yggS. We found that the PN toxicity ring 364 observed in the yggS strain (Fig. 5D) was suppressed by the presence of casamino acids (Fig. 365 S8B and Table 3) and that out of the 20 amino acids tested individually, only leucine (Fig. 5D), 366 isoleucine (Fig. 5E) or threonine (Fig. 5F) suppressed the ring of toxicity (Table 3, Fig. S8). 367 Valine could not be tested because it is toxic. This suppression was not due to an increase growth 368 rate as adding L-leucine, L-isoleucine or L-threonine did not affect the growth rate of the yggS 369 mutant (data not shown). Thr is a precursor or Ile and the last step in the synthesis of Ile and Leu 370 is catalyzed by the PLP-dependent transaminase B or IlvE (Fig. 1). The levels of IlvE were 371 analyzed by Western blot and a slight but reproducible increase in levels of IlvE protein in the 372 yggS strain was observed (Fig. S9). 373 The strongest physical clustering observed in our comparative genomic analysis was 374 between yggS genes and cell division and murein synthesis genes (Table 1). Scanning electron 375 microscopy supported this functional link, indeed yggS glyA cells grown in minimal medium 376 were found to be 23% longer than wild-type cells and a few filaments were observed (Fig. S10 377 and Table 3). The morphology and length of yggS or glyA cells were similar to those of wild-type 378 cells and the addition of D-alanine (1 mM), PL (1 mM), or the combination of methionine (0.13 379 mM) and dT (0.16 mM) did not suppress the observed phenotype in the conditions tested (data 380 not shown). 381 Discussion 382 This study clearly connects YggS, the E. coli member of the COG0325 family, with B6 383 metabolism. First, comparative genomic associations firmly linked the COG0325 family to PLP 384 salvage (Fig. S2). Second, we confirmed that YggS binds PLP as previously reported (Ito et al. 385 2013), and determined the Kd value. We also showed that PLP binding does not affect the 386 quaternary structure of the protein. Finally, we show that the E. coli yggS strain accumulates 387 PNP and that high levels of exogenous PN (but not PL, PM or PLP) are toxic to the mutant 388 strain. In the toxicity tests, the formation of a toxicity ring was very reproducible, however the 389 transformedyggS cells could not be stored for more than a week at 4°C without losing the 390 phenotype. Moreover, bacterial growth was observed in the center of the ring of toxicity. We 391 showed that these were not suppressors but further work is needed to understand concentration 392 dependence of the PN sensitivity phenotype. 393 YggS is widely distributed among bacteria, fungi, and eukaryotes and the universality of 394 YggS function is strengthened by the fact that expressing orthologs from yeast and plants 395 complement the PN toxicity phenotype of the E. coli yggS mutant. However, both the cause of 396 PN toxicity and the exact role of YggS in vitamin B6 homeostasis remain to be elucidated. PN 397 neurotoxicity has been previously reported in mammals (rats, dogs and humans) (Schaeppi and 398 Krinke 1982; Albin et al. 1987; Perry et al. 2004) and we show also here that PN is toxic when 399 PdxK is overproduced in E.coli also suggestive of PNP toxicity (Fig. S3B). However, the 400 mechanism of this toxicity is unknown although it was suggested that PNP competitively inhibits 401 PLP-dependent enzymes in the brain (Albin et al. 1987). That PN (or PNP) could inhibit PLP 402 enzymes in E. coli in the absence of YggS fits with the suppression of the PN toxicity phenotype 403 by PL (if this allows to rebalance the PNP/PLP pools) or casamino acids (if this bypasses the 404 need for specific PLP enzymes). One candidate PLP enzyme is IlvE as adding Ile, Val or Thr 405 suppresses the phenotype and as -ketoglutarate a product of IlvE activity was found to be 406 reduced in the metabolomics analysis (Table S3). How YggS would protect from this PN toxicity 407 is not clear. Indeed, YggS does not protect by sequestering PNP as we showed that it does not 408 bind PNP. We found that PdxH specific activity was not affected by the absence of YggS so 409 other type of regulations must occur and further studies are required to elucidate the causes of 410 PNP increase. 411 As summarized Table 3, YggS depletion has a pleotropic effect on metabolism and 412 diverse phenotypes were observed. We propose that most of the phenotypes observed in E. coli 413 in the absence of yggS are caused by lower activities of PLP-dependent enzymes (Fig. 1). 414 Because PLP participates in nearly sixty reactions in E. coli (http://bioinformatics.unipr.it/B6db), 415 physiological studies in strains with defects in PLP metabolism are difficult to interpret, mainly 416 because: 1) several independent phenotypes are observed; 2) these phenotypes can be due to 417 primary and secondary effects; 3) causes cannot be distinguished from effects. Such difficulties 418 are why it took decades to decipher the molecular basis of phenotypes caused by the absence of 419 ridA that, possibly like yggS, perturbs whole cellular networks by inhibiting PLP enzymes 420 (Downs and Ernst 2015). In addition, even if the core function of YggS is conserved, as the 421 cross-kingdom complementations observed here and in previous studies suggest, the network 422 perturbations caused by its absence may vary from one organism to another. For example, YggS 423 is dispensable in most organisms studied so far but is essential in Pseudomonas aeruginosa 424 (Rusmini et al. 2014) and possibly in the pathogenic bacteria Haemophilus influenzae, 425 Helicobacter pylori, Streptococcus pneumoniae and Staphylococcus aureus (data extracted from 426 the DEG database (Luo et al. 2014)). These combinations of issues make the reproducibility and 427 interpretation of physiological and metabolomic data in mutants affected in PLP metabolism 428 very difficult, as discussed below. 429 According to Ito et al., the absence of yggS leads to a 10% decrease of CoA that triggers 430 accumulation of -ketobutyrate and an increase in the L-valine pool (Ito et al. 2013). These 431 results were not reproduced in our study as no significant increase in L-valine was observed and 432 the -ketobutyrate pool was found to decrease and not increase in most conditions. The 433 differences between the two studies could be due to the different genetic backgrounds 434 (BW25133 or MG1655), the different growth temperatures (30 °C or 37 °C) or to the growth 435 phase at which the analyses were made. Indeed, metabolite pools varied greatly with the stage of 436 growth (Table S3). Further analytical studies are required to reconcile these discrepancies. 437 The growth defect of the yggS glyA strain in the presence of exogenous amino acids is a 438 strong phenotype, but remains unexplained. The partial suppression of this phenotype by dT and 439 methionine suggests a link to THF metabolism that fits with the previously reported sensitivity to 440 sulfonamides of a yggS mutant (Nichols et al. 2011). Our data suggested a strong link of YggS to 441 amino acid metabolism, and indeed metabolomics analysis does reveal changes in levels between 442 wild type and yggS mutant for six amino acids. Finally, our metabolomics data suggested that 443 several PLP-dependent enzymes involved in the putrescine metabolism could have been affected 444 in the yggS strain. 445 Another phenotype observed was the cell division defect of the glyA yggS strain. This 446 could be due to lower levels of D-alanine if the PLP-dependent alanine racemase activity is 447 lowered in this background as shown in numerous studies on B6 limitation (Grogan 1988; Kim et 448 al. 2010). As the addition of D-alanine did not suppress the phenotype it could also have more 449 complex causes. Indeed, it was previously reported that cells containing a pdxH mutation placed 450 under PL limitation were elongated and contained nucleoids that could not segregate and this 451 phenotype was not suppressed by PN or D-alanine, suggesting that it was caused by an early cell 452 division defect (Lam and Winkler 1992). The strong clustering of yggS with genes involved in 453 cell division combined with the observed phenotypes point to a regulatory role in this process 454 that will require further investigation. 455 Finally, our comparative genomics data suggest that the strong linkage of yggS with proC 456 is almost certainly due to a function of proC not directly related to proline synthesis. We propose 457 a role of ProC in PLP metabolism, as in humans as the ProC deficiency depletes the PLP pool 458 through the formation of PLP-P5C adducts (Farrant et al. 2001)(Fig. 1). In Bacteria, ProC could 459 be a mechanism to detoxify excess P5C and we are currently exploring this hypothesis. 460 This work has laid the foundation to elucidate the molecular function of the COG0325 461 family. Possibilities include a carrier function to deliver PLP to the target enzymes or a 462 protective function so that PLP does not inactivate essential lysines in proteins. More generally, 463 like the predatory genetic studies of the rid family (Downs and Ernst 2015), our work 464 underscores the power of genetics to uncover, and to start unraveling, complex biochemical 465 processes. 466 467 Acknowledgments 468 We thank S. Shanker and the staff at UF ICBR for DNA sequencing and the ICBR 469 members Cecilia Silva-Sanchez and Sixue Chen for help with 2D-gel analyses. We gratefully 470 acknowledge the staff at UF ICBR for microscopy. We thank the Cambillau laboratory for the E. 471 coli yggS expression plasmid and Robert McKenna, Antonette Bennett, and the University Of 472 Florida Center Of Structural Biology for help with data acquisition and interpretation of the 473 thermal denaturation and titration calorimetry studies. This work was funded by US National 474 Science Foundation (NSF) grants MCB-1153413 and IOS-1025398. MB was a recipient of a 475 postdoctoral fellowship from the Human Frontier Scientific Program. 476 477 478 References 479 480 Albin R. L., Albers J. W., Greenberg H. S., Townsend J. B., Lynn R. B., Burke J. M., Jr., 481 Alessi A. G. 1987. Acute sensory neuropathy-neuronopathy from pyridoxine overdose. 482 Neurology 37: 1729-1732. 483 Altschul S. F., Madden T. L., Schaffer A. A., Zhang J., Zhang Z., Miller W., Lipman D. J. 484 1997. Gapped BLAST and PSI-BLAST: a new generation of protein database search 485 programs. Nucleic Acids Res 25: 3389-3402. 486 487 Berman H. M., Westbrook J., Feng Z., Gilliland G., Bhat T. N., Weissig H., Shindyalov I. N., Bourne P. E. 2000. The Protein Data Bank. Nucleic Acids Res 28: 235-242. 488 Chen F., Mackey A. J., Stoeckert C. J., Jr., Roos D. S. 2006. OrthoMCL-DB: querying a 489 comprehensive multi-species collection of ortholog groups. Nucleic Acids Res 34: D363- 490 368. 491 492 493 494 495 496 Chen H., Xiong L. 2005. Pyridoxine is required for post-embryonic root development and tolerance to osmotic and oxidative stresses. Plant J 44: 396-408. Christen P., Mehta P. K. 2001. From cofactor to enzymes. The molecular evolution of pyridoxal-5'-phosphate-dependent enzymes. Chem Rec 1: 436-447. Clayton P. 2006. B6-responsive disorders: A model of vitamin dependency. J Inherit Metab Dis 29: 317-326. 497 Cohen S. S. 1998. A guide to the polyamines Oxford Univ. Press, New York. 498 Corpet F. 1988. Multiple sequence alignment with hierarchical clustering. Nucleic Aids Res 16: 499 10881-10890. 500 Dailey H. A., Gerdes S., Dailey T. A., Burch J. S., Phillips J. D. 2015. Noncanonical 501 coproporphyrin-dependent bacterial heme biosynthesis pathway that does not use 502 protoporphyrin. Proc Natl Acad Sci U S A 112: 2210-2215. 503 De Wergifosse P., Jacques B., Jonniaux J. L., Purnelle B., Skala J., Goffeau A. 1994. The 504 sequence of a 22.4 kb DNA fragment from the left arm of yeast chromosome II reveals 505 homologues to bacterial proline synthetase and murine alpha-adaptin, as well as a new 506 permease and a DNA-binding protein. Yeast 10: 1489-1496. 507 508 509 510 di Salvo M. L., Contestabile R., Safo M. K. 2011. Vitamin B(6) salvage enzymes: mechanism, structure and regulation. Biochim Biophys Acta 1814: 1597-1608. di Salvo M. L., Safo M. K., Contestabile R. 2012. Biomedical aspects of pyridoxal 5'phosphate availability. Front Biosci (Elite Ed) 4: 897-913. 511 Dong Q., Fromm H. J. 1990. Chemical modification of adenylosuccinate synthetase from 512 Escherichia coli by pyridoxal 5'-phosphate. Identification of an active site lysyl residue. J 513 Biol Chem 265: 6235-6240. 514 Downs D. M., Ernst D. C. 2015. From microbiology to cancer biology: the Rid protein family 515 prevents cellular damage caused by endogenously generated reactive nitrogen species. 516 Mol Microbiol 96: 211-219. 517 518 Emanuelsson O., Brunak S., von Heijne G., Nielsen H. 2007. Locating proteins in the cell using TargetP, SignalP and related tools. Nat Protoc 2: 953-971. 519 Ericsson U. B., Hallberg B. M., Detitta G. T., Dekker N., Nordlund P. 2006. Thermofluor- 520 based high-throughput stability optimization of proteins for structural studies. Anal 521 Biochem 357: 289-298. 522 Eswaramoorthy S., Gerchman S., Graziano V., Kycia H., Studier F. W., Swaminathan S. 523 2003. Structure of a yeast hypothetical protein selected by a structural genomics 524 approach. Acta Crystallogr D Biol Crystallogr 59: 127-135. 525 Farrant R. D., Walker V., Mills G. A., Mellor J. M., Langley G. J. 2001. Pyridoxal phosphate 526 de-activation by pyrroline-5-carboxylic acid:increased risk of vitamin b6 deficiency and 527 seizures in hyperprolinemia type II. J Biol Chem 276: 15107-15116. 528 Finn R. D., Bateman A., Clements J., Coggill P., Eberhardt R. Y., Eddy S. R., Heger A., 529 Hetherington K., Holm L.& other authors. 2014. Pfam: the protein families database. 530 Nucleic Acids Res 42: D222-230. 531 Fitzpatrick T. B., Amrhein N., Kappes B., Macheroux P., Tews I., Raschle T. 2007. Two 532 independent routes of de novo vitamin B6 biosynthesis: not that different after all. 533 Biochem J 407: 1-13. 534 535 Fitzpatrick T. B., Moccand C., Roux C. 2010. Vitamin B6 biosynthesis: charting the mechanistic landscape. Chembiochem 11: 1185-1193. 536 Gerlt J. A., Bouvier J. T., Davidson D. B., Imker H. J., Sadkhin B., Slater D. R., Whalen K. 537 L. 2015. Enzyme Function Initiative-Enzyme Similarity Tool (EFI-EST): A web tool for 538 generating protein sequence similarity networks. Biochim Biophys Acta 1854: 1019-1037. 539 Ghatge M. S., Contestabile R., di Salvo M. L., Desai J. V., Gandhi A. K., Camara C. M., 540 Florio R., González I. N., Parroni A.& other authors. 2012. Pyridoxal-5'-Phosphate 541 Is a slow tight binding inhibitor of E. coli pyridoxal kinase. PLoS ONE 7: e41680. 542 Green J. C., Nichols B. P., Matthews R. G. 1996. Folate biosynthesis, reduction, and 543 polyglutamylation. in Escherichia coli and Salmonella, Cellular and Molecular Biology 544 (ed. FC Neidhart), pp. 665–673. American Society for Microbiology, Washington, DC. 545 546 547 548 Grogan D. W. 1988. Temperature-sensitive murein synthesis in an Escherichia coli pdx mutant and the role of alanine racemase. Arch Microbiol 150: 363-367. Halsted C. H. 2013. B-Vitamin dependent methionine metabolism and alcoholic liver disease. Clin Chem Lab Med 51: 457-465. 549 Herrero S., Gonzalez E., Gillikin J. W., Velez H., Daub M. E. 2011. Identification and 550 characterization of a pyridoxal reductase involved in the vitamin B6 salvage pathway in 551 Arabidopsis. Plant Mol Biol 76: 157-169. 552 Ito T., Iimori J., Takayama S., Moriyama A., Yamauchi A., Hemmi H., Yoshimura T. 2013. 553 A conserved pyridoxal protein that regulates Ile and Val metabolism. J Bacteriol 195: 554 5439-5449. 555 Jochmann N., Gotker S., Tauch A. 2011. Positive transcriptional control of the pyridoxal 556 phosphate biosynthesis genes pdxST by the MocR-type regulator PdxR of 557 Corynebacterium glutamicum ATCC 13032. Microbiology 157: 77-88. 558 Kim J., Kershner J. P., Novikov Y., Shoemaker R. K., Copley S. D. 2010. Three 559 serendipitous pathways in E. coli can bypass a block in pyridoxal-5'-phosphate synthesis. 560 Mol Syst Biol 6: 436. 561 Kurihara S., Oda S., Kato K., Kim H. G., Koyanagi T., Kumagai H., Suzuki H. 2005. A 562 novel putrescine utilization pathway involves gamma-glutamylated intermediates of 563 Escherichia coli K-12. J Biol Chem 280: 4602-4608. 564 Lam H. M., Winkler M. E. 1992. Characterization of the complex pdxH-tyrS operon of 565 Escherichia coli K-12 and pleiotropic phenotypes caused by pdxH insertion mutations. J 566 Bacteriol 174: 6033-6045. 567 Lavinder J. J., Hari S. B., Sullivan B. J., Magliery T. J. 2009. High-throughput thermal 568 scanning: a general, rapid dye-binding thermal shift screen for protein engineering. J Am 569 Chem Soc 131: 3794-3795. 570 Lee M.-W., Kim B.-J., Choi H.-K., Ryu M.-J., Kim S.-B., Kang K.-M., Cho E.-J., Youn H.- 571 D., Huh W.-K.& other authors. 2007. Global protein expression profiling of budding 572 yeast in response to DNA damage. Yeast 24: 145-154. 573 574 Letunic I., Bork P. 2011. Interactive Tree Of Life v2: online annotation and display of phylogenetic trees made easy. Nucleic Acids Res 39: W475-478. 575 Li W., Cowley A., Uludag M., Gur T., McWilliam H., Squizzato S., Park Y. M., Buso N., 576 Lopez R. 2015. The EMBL-EBI bioinformatics web and programmatic tools framework. 577 Nucleic Acids Res. 578 Luo H., Lin Y., Gao F., Zhang C. T., Zhang R. 2014. DEG 10, an update of the database of 579 essential genes that includes both protein-coding genes and noncoding genomic elements. 580 Nucleic Acids Res 42: D574-580. 581 Mills P. B., Struys E., Jakobs C., Plecko B., Baxter P., Baumgartner M., Willemsen M. A. 582 A. P., Omran H., Tacke U.& other authors. 2006. Mutations in antiquitin in 583 individuals with pyridoxine-dependent seizures. Nat Med 12: 307-309. 584 Mizushina Y., Xu X., Matsubara K., Murakami C., Kuriyama I., Oshige M., Takemura M., 585 Kato N., Yoshida H.& other authors. 2003. Pyridoxal 5'-phosphate is a selective 586 inhibitor in vivo of DNA polymerase alpha and epsilon. Biochem Biophys Res Commun 587 312: 1025-1032. 588 589 Moccand C., Kaufmann M., Fitzpatrick T. B. 2011. It takes two to tango: defining an essential second active site in pyridoxal 5'-phosphate synthase. PLoS ONE 6: e16042. 590 591 592 593 594 595 Mooney S., Hellmann H. 2010. Vitamin B6: Killing two birds with one stone? Phytochemistry 71: 495-501. Moore S. D. 2011. Assembling new Escherichia coli strains by transduction using phage P1. Methods Mol Biol 765: 155-169. Mukherjee T., Hanes J., Tews I., Ealick S. E., Begley T. P. 2011. Pyridoxal phosphate: biosynthesis and catabolism. Biochim Biophys Acta 1814: 1585-1596. 596 Musayev F. N., Di Salvo M. L., Ko T. P., Schirch V., Safo M. K. 2003. Structure and 597 properties of recombinant human pyridoxine 5'-phosphate oxidase. Protein Sci 12: 1455- 598 1463. 599 Nichols R. J., Sen S., Choo Y. J., Beltrao P., Zietek M., Chaba R., Lee S., Kazmierczak K. 600 M., Lee K. J.& other authors. 2011. Phenotypic landscape of a bacterial cell. Cell 144: 601 143-156. 602 Niehaus T. D., Gerdes S., Hodge-Hanson K., Zhukov A., Cooper A. J., ElBadawi-Sidhu M., 603 Fiehn O., Downs D. M., Hanson A. D. 2015. Genomic and experimental evidence for 604 multiple metabolic functions in the RidA/YjgF/YER057c/UK114 (Rid) protein family. 605 BMC Genomics 16: 382. 606 Novichkov P. S., Kazakov A. E., Ravcheev D. A., Leyn S. A., Kovaleva G. Y., Sutormin R. 607 A., Kazanov M. D., Riehl W., Arkin A. P.& other authors. 2013. RegPrecise 3.0--a 608 resource for genome-scale exploration of transcriptional regulation in bacteria. BMC 609 Genomics 14: 745. 610 Ohsawa H., Gualerzi C. 1981. Structure-function relationship in Escherichia coli initiation 611 factors. Identification of a lysine residue in the ribosomal binding site of initiation factor 612 by site-specific chemical modification with pyridoxal phosphate. J Biol Chem 256: 4905- 613 4912. 614 Overbeek R., Olson R., Pusch G. D., Olsen G. J., Davis J. J., Disz T., Edwards R. A., 615 Gerdes S., Parrello B.& other authors. 2014. The SEED and the Rapid Annotation of 616 microbial genomes using Subsystems Technology (RAST). Nucleic Acids Res 42: D206- 617 214. 618 619 620 621 Paul L., Ueland P. M., Selhub J. 2013. Mechanistic perspective on the relationship between pyridoxal 5'-phosphate and inflammation. Nutr Rev 71: 239-244. Percudani R., Peracchi A. 2003. A genomic overview of pyridoxal-phosphate-dependent enzymes. EMBO Rep 4: 850-854. 622 Percudani R., Peracchi A. 2009. The B6 database: a tool for the description and classification 623 of vitamin B6-dependent enzymatic activities and of the corresponding protein families. 624 BMC Bioinformatics 10: 273. 625 Perry T. A., Weerasuriya A., Mouton P. R., Holloway H. W., Greig N. H. 2004. Pyridoxine- 626 induced toxicity in rats: a stereological quantification of the sensory neuropathy. Exp 627 Neurol 190: 133-144. 628 Phillips K., de la Pena A. H. 2011. The combined use of the Thermofluor assay and ThermoQ 629 analytical software for the determination of protein stability and buffer optimization as an 630 aid in protein crystallization. Curr Protoc Mol Biol Chapter 10: Unit10 28. 631 Rueschhoff E. E., Gillikin J. W., Sederoff H. W., Daub M. E. 2013. The SOS4 pyridoxal 632 kinase is required for maintenance of vitamin B6-mediated processes in chloroplasts. 633 Plant Physiol Biochem 63: 281-291. 634 Rusmini R., Vecchietti D., Macchi R., Vidal-Aroca F., Bertoni G. 2014. A shotgun antisense 635 approach to the identification of novel essential genes in Pseudomonas aeruginosa. BMC 636 Microbiol 14: 24. 637 Salgado H., Peralta-Gil M., Gama-Castro S., Santos-Zavaleta A., Muniz-Rascado L., 638 Garcia-Sotelo J. S., Weiss V., Solano-Lira H., Martinez-Flores I.& other authors. 639 2013. RegulonDB v8.0: omics data sets, evolutionary conservation, regulatory phrases, 640 cross-validated gold standards and more. Nucleic Acids Res 41: D203-213. 641 642 643 644 645 646 Sambrook J. E., Fritsch F., Maniatis T. 1989. Molecular cloning : A laboratory manual, 2nd edn. Cold Spring Harbor Laboratory Press, Cold Spring Harbor. Sampson D. A., O'Connor D. K. 1989. Response of B-6 vitamers in plasma, erythrocytes and tissues to vitamin B-6 depletion and repletion in the rat. J Nutr 119: 1940-1948. Samsonova N. N., Smirnov S. V., Altman I. B., Ptitsyn L. R. 2003. Molecular cloning and characterization of Escherichia coli K12 ygjG gene. BMC Microbiol 3: 2. 647 Samsonova N. N., Smirnov S. V., Novikova A. E., Ptitsyn L. R. 2005. Identification of 648 Escherichia coli K12 YdcW protein as a gamma-aminobutyraldehyde dehydrogenase. 649 FEBS Lett 579: 4107-4112. 650 Sang Y., Locy R. D., Goertzen L. R., Rashotte A. M., Si Y., Kang K., Singh N. K. 2011. 651 Expression, in vivo localization and phylogenetic analysis of a pyridoxine 5'-phosphate 652 oxidase in Arabidopsis thaliana. Plant Physiol Biochem 49: 88-95. 653 Schaeppi U., Krinke G. 1982. Pyridoxine neuropathy: correlation of functional tests and 654 neuropathology in beagle dogs treated with large doses of vitamin B6. Agents Actions 12: 655 575-582. 656 657 Schneider B. L., Hernandez V. J., Reitzer L. 2013. Putrescine catabolism is a metabolic response to several stresses in Escherichia coli. Mol Microbiol 88: 537-550. 658 Shannon P., Markiel A., Ozier O., Baliga N. S., Wang J. T., Ramage D., Amin N., 659 Schwikowski B., Ideker T. 2003. Cytoscape: a software environment for integrated 660 models of biomolecular interaction networks. Genome Res 13: 2498-2504. 661 Shi H., Xiong L., Stevenson B., Lu T., Zhu J. K. 2002. The Arabidopsis salt overly sensitive 4 662 mutants uncover a critical role for vitamin B6 in plant salt tolerance. Plant Cell 14: 575- 663 588. 664 Suvurova I. A., Rodionov D. A. 2015. MocR ASK RODIONOV. submitted. 665 Szklarczyk D., Franceschini A., Wyder S., Forslund K., Heller D., Huerta-Cepas J., 666 Simonovic M., Roth A., Santos A.& other authors. 2015. STRING v10: protein- 667 protein interaction networks, integrated over the tree of life. Nucleic Acids Res 43: D447- 668 452. 669 670 671 672 Szydlowski N., Burkle L., Pourcel L., Moulin M., Stolz J., Fitzpatrick T. B. 2013. Recycling of pyridoxine (vitamin B6) by PUP1 in Arabidopsis. Plant J 75: 40-52. Tamames J., Gonzalez-Moreno M., Mingorance J., Valencia A., Vicente M. 2001. Bringing gene order into bacterial shape. Trends Genet 17: 124-126. 673 Titiz O., Tambasco-Studart M., Warzych E., Apel K., Amrhein N., Laloi C., Fitzpatrick T. 674 B. 2006. PDX1 is essential for vitamin B6 biosynthesis, development and stress tolerance 675 in Arabidopsis. Plant J 48: 933-946. 676 Tramonti A., Fiascarelli A., Milano T., di Salvo M. L., Nogues I., Pascarella S., 677 Contestabile R. 2015. Molecular mechanism of PdxR - a transcriptional activator 678 involved in the regulation of vitamin B biosynthesis in the probiotic bacterium Bacillus 679 clausii. FEBS J 282: 2966-2984. 680 Vanderschuren H., Boycheva S., Li K. T., Szydlowski N., Gruissem W., Fitzpatrick T. B. 681 2013. Strategies for vitamin B6 biofortification of plants: a dual role as a micronutrient 682 and a stress protectant. Front Plant Sci 4: 143. 683 Vermeersch J. J., Christmann-Franck S., Karabashyan L. V., Fermandjian S., Mirambeau 684 G., Der Garabedian P. A. 2004. Pyridoxal 5'-phosphate inactivates DNA topoisomerase 685 IB by modifying the lysine general acid. Nucleic Acids Res 32: 5649-5657. 686 687 688 689 690 691 Whalen W. A., Berg C. M. 1982. Analysis of an avtA::Mu d1(Ap lac) mutant: metabolic role of transaminase C. J Bacteriol 150: 739-746. Whittaker M. M., Penmatsa A., Whittaker J. W. 2015. The Mtm1p carrier and pyridoxal 5'phosphate cofactor trafficking in yeast mitochondria. Arch Biochem Biophys 568: 64-70. Yang E. S., Schirch V. 2000. Tight binding of pyridoxal 5'-phosphate to recombinant Escherichia coli pyridoxine 5'-phosphate oxidase. Arch Biochem Biophys 377: 109-114. 692 Zhao G., Winkler M. E. 1995. Kinetic limitation and cellular amount of pyridoxine 693 (pyridoxamine) 5'-phosphate oxidase of Escherichia coli K-12. J Bacteriol 177: 883-891. 694 695 696 697 Zhou J., Rudd K. E. 2013. EcoGene 3.0. Nucleic Acids Res 41: D613-624. 698 Table 1. Summary of physical clustering analysis. Set 1: 1536 representative genome set. Set 699 2: 1000 representative genome set. Total: all 11411 organisms in PubSeed at the time of the 700 analysis. Genes encoding for: * PLP enzymes; proteins with 701 bind GTP or ATP. #no current name for this gene. & Essential lysines; proteins that $ 702 Gene %Set1 %Set2 %Total Function Pyrroline-5-carboxylate reductase (EC 1.5.1.2) SepF, FtsZ-interacting protein related to cell division Cell division initiation protein DivIVA COG0762, cell division protein YlmG/Ycf19 (putative), YggT family Septum formation protein Maf Cell division protein FtsZ (EC 3.4.24.-) SepF, FtsZ-interacting protein related to cell division UDP-N-acetylglucosamine--Nacetylmuramyl-(pentapeptide) pyrophosphoryl-undecaprenol Nacetylglucosamine transferase (EC 2.4.1.227) proC 21.7 3.0 18.4 sepF 22.8 0.8 24.3 divIVA 20.1 1.8 26.1 yggT 19.2 0.5 36.3 maf 2.0 1.5 0.9 ftsZ$ 17.3 0.1 29.0 sepF 22.8 0.8 24.3 murG 11.6 0.1 18.0 pilT$ 19.2 0.8 30.9 lsp 14.8 1.4 13.5 pdxR 1.2 0.4 0.8 pdxK$ 0.9 PLP 0.7 transporter# 0.3 0.5 Lipoprotein signal peptidase (EC 3.4.23.36) Predicted transcriptional regulator of pyridoxine metabolism Pyridoxal kinase (EC 2.7.1.35) 0.4 0.5 B6 transport protein yqgF 12.0 0.2 23.5 ileS&$ 16.7 0.3 25.6 leuS&$ 7.6 6.4 5.5 rluD 7.3 0.1 8.7 Twitching motility protein PilT YqgF, involved in pre-16S RNA processing Isoleucyl-tRNA synthetase (EC 6.1.1.5) Leucyl-tRNA synthetase (EC 6.1.1.4) Ribosomal large subunit pseudouridine synthase D (EC 4.2.1.70) Process Proline synthesis Cell division Cell division Cell division Cell division Cell division Cell division Cell wall synthesis Surface motility Secretion PLP salvage PLP salvage PLP salvage Translation Translation Translation Translation 703 704 rdgB 12.0 ND 23.1 yggW 12.4 0.6 23.3 yqgE 11.1 0.3 22.8 topoI 1.3 0.9 0.4 glyA* 0.8 0.1 0.5 aOT3* 0.3 0.5 0.3 Nucleoside 5-triphosphatase RdgB (dHAPTP, dITP, XTP-specific) (EC 3.6.1.15) Radical SAM family enzyme, similar to coproporphyrinogen III oxidase, oxygen-independent, clustered with nucleoside-triphosphatase RdgB UPF0301 protein YqgE DNA topoisomerase I (EC 5.99.1.2) Serine hydroxymethyltransferase (EC 2.1.2.1) Acetylornithine aminotransferase (EC 2.6.1.11) Nucleotide metabolism ? ? Replication Amino acid metabolism Amino acid metabolism 705 Table 2. Pyridoxine toxicity phenotypes on M9 glucose 0.2% (w/v). % Antibiotics Cm or Amp 706 were added when required; the three plasmids were used as controls and gave similar results. 707 *Arabinose 0.2 % was added. & plate was supplemented with aTet 50 ng/ml. Strain Wild-type BW25113 glyA (VDC6664) yggS pBAD18/pBAD24/pBAD33%* yggS pBAD24 yggSEC* yggS pBAD18: LOC100191932* yggS pBAD18: At1g11930 yggS pBEY YBL036C & yggS pBAD18 + PL 50 M* 708 709 710 Diameter of the zone of inhibition (cm) 0.1 mg/ml 1 mg/ml PN PN None None None None 4.15 ±0.05 5.25±0.1 None None None None None None None None None None 711 712 Table 3. Phenotypes discussed in this study. M9 Casa is M9 glucose 0.2% (w/v) supplemented 713 with 0.4% (w/v) casamino acids. M9 is M9 glucose 0.2% (w/v) supplemented with isoleucine 50 714 g/ml, leucine 30g/ml , or threonine 30g/ml. $metabolomics data presented in Table 1 and 715 Table S3. *in the conditions used in this study, no significant accumulation of valine was 716 observed in the yggS strain. nd is non determined; - : no phenotype observed; + : phenotype 717 observed. Leu: L-leucine; Ile: L-isoleucine; Thr: L-threonine. Phenotype PN sensitivity PN sensitivity on M9 Casa, M9 Leu, Ile or Thr Growth on LB Growth on M9 Casa Valine secretion Longer cells 718 719 720 glyA - yggS + yggS glyA + nd - nd + + nd - + + -$* - nd + 721 Figures legends 722 Figure 1. PLP synthesis and a subset of pathways influenced by PLP levels in E.coli. In E. 723 coli PLP can be synthesized via the de novo PdxFBAJ dependent pathway or salvaged using 724 PdxK, PdxY and PdxH (see review in (Fitzpatrick et al. 2007)). PLP is a cofactor for enzymes 725 involved in cell wall synthesis, amino acids, and one-carbon metabolism. P5C can sequester free 726 PLP. 727 Tetrahydrofolate; CH2-THF: 5,10-methylene-tetrahydrofolate; Gcv: Glycine-cleavage complex; 728 TA: L-threonine aldolase. P5C is formed spontaneously and reversibly from GSA. Dashed 729 arrows indicate that several enzymes are required for a conversion. Proteins in bold with an 730 asterisk are enzymes that require PLP. This figure is based on E. coli literature (see introduction) 731 except for formation of the P5C-PLP adduct that has been reported in human but not E. coli to 732 date Abbreviations for enzymes names are given in the text. 733 Figure 1. PLP synthesis and a subset of pathways influenced by PLP levels in E.coli. In E. 734 coli PLP can be synthesized via the de novo PdxFBAJ dependent pathway or salvaged using 735 PdxK, and PdxH (see review in (Fitzpatrick et al. 2007)). PLP is a cofactor for enzymes involved 736 in cell wall synthesis, amino acids, and one-carbon metabolism. P5C can sequester free PLP. 737 GSA: Glutamic 5-semialdehyde; P5C: Δ1-pyrroline-5 carboxylic acid; THF: Tetrahydrofolate; 738 CH2-THF: 5,10-methylene-tetrahydrofolate; Gcv: Glycine-cleavage complex; TA: L-threonine 739 aldolase; TyrB: aromatic amino acid transferase. P5C is formed spontaneously and reversibly 740 from GSA. Dashed arrows indicate that several enzymes are required for a conversion. Proteins 741 in bold with an asterisk are enzymes that require PLP. This figure is based on E. coli literature 742 (see introduction) except for formation of the P5C-PLP adduct that has been reported in human 743 but not E. coli to date Abbreviations for enzymes names are given in the text. GSA: Glutamic 5-semialdehyde; P5C: Δ1-pyrroline-5 carboxylic acid; THF: 744 Figure 2. Distribution of yggS genes among bacteria. (A) The tree was constructed in iTol 745 (itol.embl.de/). Tree branches are colored by phylum. The presence of yggS homologs are 746 displayed in rings around the tree. The presence of at least one yggS homolog in the specific 747 organism in the tree is shown by the innermost (blue) ring. The presence of two or three yggS 748 paralogs in one genome are shown by the middle (green) ring and the outer (orange) ring 749 respectively. (B) Physical clustering of yggS genes in Pseudomonas sp. GM79 and 750 Rhodopseudomonas palustris TIE-1 with PLP synthesis, salvage genes and PLP dependent 751 enzymes. The red arrows indicate predicted PdxR binding sites. 752 753 Figure 3. Toxic effect of PN on the E. coli yggS strain and the universality of YggS 754 function. The cells were plated as described in the material and methods section. (A) yggS 755 (VDC6594) lawn and drops of PN at concentrations of 5.9 mM (a), 590 M (b), 295 M (c) or 756 59M (d); PN at the concentrations (a) or (b) were used in experiments (B-F). (B) yggS 757 pBAD18 lawn; in presence of arabinose 0.2% (C) yggSin pBAD24yggSEc (pBY291.3) lawn in 758 presence of arabinose 0.2%; (D)yggS transformed with pBAD18 LOC100191932 (pBY298.7) 759 lawn in presence of arabinose 0.2%; (E) yggS transformed pBAD18 LOC100191932 in absence 760 of arabinose 0.2%; (F) yggS transformed with pBEY YBL036C (pBEY329.12) in presence of 761 aTet 50 ng/ml. The arrows indicate the presence of the ring of toxicity. 762 763 Figure 4. Analysis of B6 derivatives in wild-type and yggS E. coli. The B6 vitamers were 764 analyzed by HPLC with fluorometric detection. (A) B6 vitamer profile; 4-dPN is the internal 765 standard 4-deoxypyridoxine. (B) Determination of PNP, PMP, PLP, concentrations (pmol/mg 766 protein) in wild-type (dark gray) and yggS(light gray) strains. 767 768 Figure 5. Effects of amino acids in a yggS background. Wild-type (yellow), yggS 769 (VDC6594) (blue), glyA (orange), and yggS glyA (grey) strains were grown in the Bioscreen 770 C at 37 °C in different media. (M9 glucose 0.2% (w/v) supplemented with 1 mM glycine; 771 (M9 glucose 0.2% (w/v) supplemented with 1 mM glycine and 0.4% casamino acids; (C) LB 772 supplemented with methionine (0.13 mM) and 0.16 mM dT. The average of 10 independent 773 growth curves was plotted, and errors bars represent standard deviations. PN at the 774 concentrations 5.9 mM (a), 590 M (b) were used in experiments (D-F). (D) yggSin pBAD33 775 lawn on M9 supplemented with L-leucine 30 g/mL. (E) yggSin pBAD33 lawn on M9 776 supplemented with L-isoleucine 50 g/mL; (F) (E) yggSin pBAD33 lawn on M9 supplemented 777 with L-threonine 30 g/mL. Rings of toxicity were observed for yggSin pBAD33 lawn on M9 778 (Fig. S7A) for PN at concentrations of 5.9 mM (a), 590 M (b). M9 was supplemented with each 779 20 amino acids individually. Any other amino acid than L-leucine, L-isoleucine and L-threonine 780 gave the same phenotype as Fig. S7C. dT: thymidine; Met: methionine. 781 . 782 783 784