Mutational analyses reveal PLP-independent functions at PipY, the cyanobacterial paradigm for pyridoxal-phosphate binding proteins

9 Pyridoxal-phosphate binding proteins (PLPBP) are involved in the homeostasis of B 6 vitamers and 10 amino/keto acids , s hare a high degree of sequence conservation and are represented in all three 11 domains of life. Despite the obligate presence of the catalyst cofactor PLP, attempts to show enzymatic 12 activity have been unsuccessful. Instead, evidence of RNA binding activity has been provided for 13 several members of the family. Here we use PipY, one of the few PLBPB members studied so far, as 14 a model system to address the phenotypic impact in the cyanobacterium Synechococcus elongatus of 15 mutations K26A, P63L and R210Q, which respectively prevent PLP binding or are equivalent to those 16 conferring B6-dependent epilepsy in humans with a recessive inheritance pattern. We found that while 17 mutation K26A at the PLP -binding residue abrogated all phenotypes associated to PipY 18 overexpression and toxicity, P63L and R210Q behaved as dominant gain-of-function mutations that 19 inhibited bacterial growth. We provide in vivo evidence of PipY performing PLP -independent 20 functions, in which mutant variant PipYK26A but not PipYP63L or PipYR210Q would be defective. A 21 model integrating our observations with previous data from other organims and PLPBP variants is 22 discussed. 23

Synechococcus, PipX, B6 homeostasis, RNA binding, PLPHP 24

25 The Pyridoxal phosphate (PLP)-binding proteins (COG0325/PLPBP) are involved in vitamin B6 26 and amino acid homeostasis (Labella et al. 2017; Ito et al. 2019; Vu et al. 2020; Ito 2022; Tremiño et 27 al. 2022). PLPBP/PLPHP (for PLP homeostasis proteins) family members are found in all kingdoms 28 of life (Prunetti et al. 2016; Farkas and Fitzpatrick 2024) exemplified by the proteins YBL036C (yeast), 29 YggS (Gram-negative bacteria), YlmE (Gram -positive bacteria), PipY (cyanobacteria) and PLPHP 30 (previously called PROSC in humans or plants). They are all single domain proteins exhibiting the 31 type-III fold of PLP-holoenzymes (Schneider et al. 2000; Eswaramoorthy et al. 2003; Tremiño et al. 32 2017; Tremiño et al. 2018; He et al. 2022) whose deficiency alters the B6 pool in the different systems 33 studied (Darin et al. 2016; Prunetti et al. 2016; Ito et al. 2019; Johnstone et al. 2019; Ito et al. 2020; 34 Vu et al. 2020). In humans, the severity of PLPBP mutations causing vitamin B6-dependent epilepsy 35 appears to correlate with decreased ability of the PLPHP variants to bind the PLP cofactor (Johnstone 36 et al. 2019) although only some of th e pathogenic mutation s found in patients complement the 37 pyridoxine-sensitive phenotype of the E. coli yggS null mutant (Darin et al. 2016) . In zebrafish the 38 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted December 30, 2025. ; https://doi.org/10.64898/2025.12.29.696868doi: bioRxiv preprint 2 complete absence of PLPHP triggers epilepsy seizures and, importantly, early death, indicating the 39 essentiality of the protein in development (Johnstone et al. 2019) and further suggesting an important 40 regulatory complexity. 41 PipY from the cyanobacterium Synechococcus elongatus PCC7942 (hereafter S. elongatus) is one 42 of the best -characterized PL PBP members and can be regarded as a paradigm for these proteins 43 (Labella et al. 2017; Cantos et al. 2019; Tremiño et al. 2022; Llop, Labella, et al. 2023). In S. elongatus, 44 pipY null mutants display PLP-related phenotypes including increased sensitivity to pyridoxine and to 45 the antibiotics D-cycloserine and β-chloro-D-alanine (Labella et al. 2017). In addition, recombinantly 46 produced PipY has been used to mimick the effect of selected pathogenic mutations on protein function 47 (Tremiño et al. 2017) . Fig. 1 illustrates the structural parallelism between cyanobacterial PipY and 48 human PLPHP, highlighting the PipY residues Lys26, Pro63 and Arg210, whose mutations are 49 analysed in this work. 50 Despite the obligate presence of the catalyst cofactor PLP, no enzymatic activity has been 51 associated to the PLPBP family and the current view is that they would have purely regulatory 52 functions as components of signal transduction pathways (Tremiño et al. 2022; Graziani et al. 2024) . 53 In bacteria PLPBP genes cluster with cell division and cell wall biosynthesis ( dcw, in Gram-positive 54 bacteria and mycobacteria), PLP salvage, surface motility, secretion, amino acid metabolism, and 55 translation genes (Prunetti et al. 2016; Tremiño et al. 2022). In cyanobacterial genomes the sepF gene 56 (involved in cell division and restricted to gram -positive bacteria and cyanobacteria) is typically 57 located downstream pipY. I n S. elongatus and in most cyanobacteria pipY is part of a bicistronic 58 pipXpipY operon with relatively short or non -existent intergenic distances suggestive of tight co -59 regulation and functional interaction s between their products in this bacterial phylum (Labella et al. 60 2017; Cantos et al. 2019). 61 While deciphering the mechanisms of action and regulatory connections of PLPBP members 62 remains very challenging, the inferred functional association between PipY with PipX provides a 63 unique opportunity to investigate PLPBP functions in the context of a signalling pathway that is highly 64 conserved in an important biological group of organisms. PipX, a hallmark of cyanobacteria, is a small 65 protein involved in metabolic and environmental signalling that form complexes with other regulatory 66 proteins, transcriptional regulators, or translation -related factors (Burillo et al. 2004; Espinosa et al. 67 2006; Espinosa et al. 2007; Llácer et al. 2010; Laichoubi et al. 2011; Espinosa et al. 2018; Labella et 68 al. 2020; Jerez et al. 2021; Jerez et al. 2024; Salinas et al. 2024) . PipX stabilizes the conformation of 69 the transcriptional regulator NtcA and helps local recruitment of RNA polymerase (Forcada-Nadal et 70 al. 2025) in response to nitrogen limitation (Espinosa et al. 2014; Giner-Lamia et al. 2017). PipX also 71 interacts with the essential ribosome -assembly GTPase EngA (YphC/Der/YfgK) (Jerez et al. 2021; 72 Llop, Bibak, et al. 2023). Both PipX and PipY regulate the levels of a common set of transcripts and 73 afect the susceptibility to PLP-targeting antibiotics D-cycloserine and β-chloro-D-alanine, further 74 suggesting their involvement in the same genetic pathway(s) (Labella et al. 2017). 75 Neither PipX or PipY are required for growth of S. elongatus under standard laboratory conditions 76 (Espinosa et al. 2009; Labella et al. 2017; Cantos et al. 2019), although their overexpression or certain 77 gain-of-function mutations at PipX prevent growth (Laichoubi et al. 2012; Labella et al. 2017; Jerez 78 et al. 2021; Llop, Labella, et al. 2023) . Overexpression of either PipX or PipY also induce a specific 79 stress program termed chlorosis or bleaching (Jerez et al. 2021; Llop, Bibak, et al. 2023) , a complex 80 adaptative response by which they degrade their light -harvesting antenna, the phycobilisome , and 81 reduce their chlorophyll content (Schwarz and Grossman 1998; Spät et al. 2018; Forchhammer and 82 Schwarz 2019). PipY overexpression also confers specific phenotypic traits in S. elongatus including 83 increase of cell length, decrease of cell viability and photosynthesis activity, accompanied by a 84 dramatic and unprecedented accumulation of giant polyphosphate (hereafter polyP) granules (Labella 85 et al. 2017; Llop, Labella, et al. 2023). 86 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted December 30, 2025. ; https://doi.org/10.64898/2025.12.29.696868doi: bioRxiv preprint 3 The aim of this work was to gain insights into yet unknown functions of the PLPBP family by 87 taking advantage of the already existing structural and functional information o n PipY. Here we 88 address the phenotypic effects conferred in S. elongatus or Escherichia coli strains by PipY variants 89 expressing point mutations K26A, P63L and R210Q that together prevent PLP binding or mimic 90 pathogenic changes found in the orthologous human protein. 91

92 Plasmid construction 93 The plasmids and primers used in this study are listed in Table 1 and S1, respectively. Cloning 94 procedures were carried out in Escherichia coli XL1-Blue. All constructs were verified by automated 95 Sanger sequencing. 96 Plasmids pUAGC1133, pUAGC1130, and pUAGC1129 carrying m utant alleles pipYK26A, 97 pipYP63L, and pipYR210Q, respectively, were generated by QuickChange site-directed mutagenesis using 98 pUAGC294 as template with primer pairs PipY-K26A-F/PipY-K26A-F, PipY-P63L-F/PipY-P63L-R 99 and PipY-R210Q-F/PipY-R210Q-R, respectively. 100 E. coli assays 101 Transformation of Escherichia coli XL1-Blue and MG1655 (Table 2) was performed by heat 102 shock, essentially as described in (Froger and Hall 2007) . 65 μL of competent cells were incubated 103 with 50 ng of the plasmids. After the heat shock, 1 mL LB was added cells were incubated at 37°C for 104 1 hour with agitation. An equivalent volume to 100 or 900 μL of XL1 -Blue and MG1655 cells, 105 respectively, were spread on LB agar (1.5% w/v) plates. Ampicillin was added to solid media at a 106 concentration of 75 μg/mL and, when indicated, the appropriate amount of IPTG. Pictures were taken 107 after 24- or 48-hours incubation at 37ºC. 108 Cyanobacteria culture conditions and strain generation 109 S. elongatus cultures were routinely grown in blue –green algae BG11 medium (BG110 110 supplemented with 17.5 mM sodium nitrate (NaNO₃) and 10 mM HEPES/NaOH (pH 7.8; (Rippka et 111 al. 1979)) at 30 °C under constant cool white fluorescent light, either in liquid cultures (150 rpm, 70 112 μmol photons m⁻²s⁻¹; mix of two clones) or on plates (50 μmol photons m⁻² s⁻¹; individual clones). 113 Solid media contained 1.5% (w/v) agar and 0.5 mM sodium thiosulfate (Na₂S₂O₃). When appropriate, 114 chloramphenicol (Cm, 3.5 μg/mL) or streptomycin (Sm, 15 μg/mL) were added. 115 For liquid growth, cultures of 30 mL were grown in BG11 using baffled flasks. Growth was 116 monitored by measuring the optical density at 750 nm (OD 750nm) in 1 mL samples using an Ultrospec 117 2100 Pro UV-Vis Spectrophotometer (Amersham Biosciences, Amersham, UK). All experiments were 118 performed on mid -exponential phase cultures (OD 750nm = 0.4 – 0.8). To overexpress proteins the 119 indicated concentration of isopropyl β-D-1-thiogalactopyranoside (IPTG) was added to media. 120 S. elongatus strains used in this study are listed in Table 2. Transformations were performed 121 essentially as described in (Taton et al. 2020) , and correct allele replacement verified by PCR. The 122 primer pairs used were inter2060 -1F/2059-Cm-1R and inter2060 -1F/PipX-5R-129 for pipY, PipX -123 126-F/2060-Cm-1R for pipXpipY, and 7942NSIA-F/ NSI-1R for NSI. 124 Determination of growth and pigment content 125 For growth measurements in flasks experiments, 25 mL of cultures were adjusted to an initial 126 optical density at 750 nm (OD 750) of 0.1 using a Ultrospec 2100 Pro UV -Vis Spectrophotometer 127 (Amersham Biosciences, Amersham, UK). OD750 was then recorded at 0, 24 and 72 h after addition of 128 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted December 30, 2025. ; https://doi.org/10.64898/2025.12.29.696868doi: bioRxiv preprint 4 50 µM IPTG. For drop -plate assays in solid media, 5 μL of culture samples adjusted to an OD 750 of 129 0.5 (and serial dilutions of 1/2, 1/10 and 1/100) were spotted on BG11 plates without or with 50 µM 130 IPTG. Plates were incubated for 5 days under standard conditions. 131 Cell staining and confocal microscopy parameters 132 To detect polyP granules in S. elongatus, 4′,6-diamidino-2-phenylindole (DAPI) staining was 133 used adapting the protocol described by (Llop, Labella, et al. 2023). Samples (1 mL) were fixed with 134 1% formaldehyde for 20 minutes at room temperature, washed once with deionized water, and frozen 135 at -20°C for 24 hours. Staining was performed using 0.25 mg/mL DAPI for 15 minutes at room 136 temperature in the dark, followed by three washes with deionized water. 137 Micrographs were taken using a Zeiss LSM800 confocal laser scanning microscope. 5 μL of the 138 stained samples were placed on 2% low-melting-point agarose pads. The microscopy settings were as 139 follows: Ex 640 nm/em 650+ nm for auto-fluorescence, ex 405 nm/Em 470–617 nm for polyP, and ex 140 405 nm/em 470 –617 nm for DAPI -control signals. Micrographs were coloured as red (auto -141 fluorescence) and light blue (polyP) to improve visualization contrast. 142 Protein extraction, immunodetection, and band quantification 143 For protein extraction, 10 mL samples of S. elongatus liquid cultures at mid -exponential phase 144 were collected at 0, 6 or 24 hours after addition of 1 mM IPTG. Cells were pelleted at 7300× g, flash 145 frozen in liquid nitrogen, and stored at −20 °C. Pellets were resuspended in 60 μL of lysis buffer (10 146 mM Tris/HCl pH 7.5, 0.5 mM EDTA, 1mM β-mercaptoethanol, 1 mM phenylmethylsulfonyl fluoride 147 (PMSF)), and cells were disrupted with 1 spoonful of 0.1 mm glass beads (≈30 µL) i n a high-speed 148 homogenizer Minibeadbeater (three pulses of 1 minute each, separated by 1 -minute intervals on ice; 149 speed of 5 m/s). Samples were centrifuged (5500× g for 5 min), and the supernatant fractions (crude 150 protein extracts) were transferred to a new tube. Protein concentrations were estimated via the Bradford 151

(Bradford, 1976) using the PierceTM detergent -compatible Bradford assay kit 152 (ThermoScientific, Waltham, MA, USA) in a VICTOR3TM 1420 Multilabel Plate Reader 153 (PerkinElmer). Protein extracts were stored at −20 °C until needed. 154 For immunodetection, 50 μg of total protein extracts were loaded into a sodium dodecyl sulphate 155 polyacrylamide gel (SDS -PAGE; 15% polyacrylamide). Electrophoresis was followed by 156 immunoblotting onto 0.2 μm polyvinylidene fluoride membranes (from GE Healthcare Technologies, 157 Inc., Chicago, IL, USA), and the membranes were subsequently blocked with 5% non -fat dry milk in 158 phosphate-buffered saline with 0.1% Tween 20 (PBS -Tween) for 1 hour at room temperature. 159 Membranes were then incubated overnight in PBS -Tween with 5% non -fat dried milk with a 1:300 160 dilution of PipY primary antibody (Labella et al. 2017) and for 1 h at room temperature with a 161 1:150,000 dilution of ECL rabbit IgG and an HRP -linked F(ab')2 fragment (from a donkey, GE 162 Healthcare). Visualization of bands was performed using SuperSignal WestFemto reagent (Thermo 163 Fisher Scientific, Waltham, MA, USA) in a Biorad ChemiDoc Imager using the automatic exposure 164 mode and avoiding pixel saturation. 165 Computational methods 166 To measure cell length and protein band intensities images were analyzed using ImageJ v1.54g. 167 The length was determined with the “Straight” tool of the program and the “Measure” button. The 168 Western blot specific bands were selected using the “Rectangle” function, and their corresponding 169 profiles were measured with the “Wand” tool. 170 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted December 30, 2025. ; https://doi.org/10.64898/2025.12.29.696868doi: bioRxiv preprint 5 To quantify the polyP signal, whole-cell fluorescence intensities from the polyP and DAPI-control 171 channels in confocal images were measured using ImageJ v1.54g. PolyP intensities were normalized 172 to the DAPI signal to account for differences in cell numbers and DAPI uptake. 173 To quantify E. coli colony sizes, Feret’s diameter was measured using ImageJ v1.54g. Images 174 typed as “8-bit” were thresholded by Otsu's method after background subtraction and subjected to a 175 “watershed” process to separate colonies. The resultant images were adjusted to an ellipse and the 176 particles filtered with a minimum size of 0.3 mm before measuring Feret’s diameter. 177 To perform statistical analyses, RStudio v2025.09.2 Build 418 was used (RStudio: Integrated 178 Development for R. RStudio 2020). 179 To generate graphical representations of protein structures PyMOL v1.7.1.7 was used (The 180 PyMOL Molecular Graphics System, Schrödinger, LLC). 181

182 PipY mutations P63L or R210Q, but not K26A, prevent growth of S. elongatus 183 In contrast to PipY deficiency, PipY overexpression results in rather dramatic phenotypes (Labella 184 et al. 2017; Llop, Labella, et al. 2023). Strain 1SPtrc-PipY, used in the previous studies, carries an extra 185 copy of pipY at the neutral site I (NSI) of the S. elongatus chromosome under the control of the IPTG-186 inducible promoter Ptrc. Tight control and inducible overexpression of PipY from the pipY gene is 187 provided by the LacI repressor, also encoded within the NSI (see Fig. 2a). 188 To determine the effect of relevant point mutations on PipY activity in S. elongatus we first 189 attempted construction of variants of strain 1SPtrc-PipY, so the phenotypic effects of wild type and 190 mutant PipY derivatives can be studied in parallel both at close to physiological levels and at very high 191 levels of expression. With this in mind we first generated mutations encoding K26A, P63L or R210Q 192 variants in plasmid pUAGC294, giving plasmids pUAGC1133, pUAGC1130 or pUAGC1129, 193 respectively (Table 1). To avoid possible interference and/or gene conversion between the wild type 194 and mutant alleles of pipY, we performed all experiments in pipY null strains. To explore a possible 195 contribution of PipX to the phenotypes studied here while strengthening the robustness of the data we 196 also obtained and studied pipXpipY derivatives in parallel by precisely replacing their ORF by the cat 197 (chloramphenicol-acetyltransferase) gene as described in (Labella et al. 2017). 198 The pipY and pipXpipY mutants were next used as recipients in independent transformations with 199 plasmids containing wild type, the three mutant alleles of pipY or, to provide a negative control for 200 PipY activity, carrying no pipY gene (plasmid pUAGC280). Surprisingly, while constructs encoding 201 PipY or PipYK26A yield thousands of streptomycin-resistant transformants, the plasmids encoding 202 PipYP63L or PipYR210Q produced just a few. In each case several transformants clones were PCR -203 analysed for the presence of the insert at NSI. For comparison the unmodified NSI was amplified from 204 the WT strain. 205 As shown in Fig. 2b (left panel), PCR products of the expected size were found when the 206 transformations involved 1SPtrc-PipY, 1SPtrc-PipYK26A or 1SPtrc constructs in both genetic 207 backgrounds. In contrast, the PCR products of some of the few transformants obtained for 1SPtrc-208 PipYP63L and 1SPtrc-PipYR210Q constructs were indistinguishable from those of the WT control strain 209 in the two genetic backgrounds (Fig. 2b, right panel). Since the unmodified NSI allele was the only 210 one detected in those transformants, the results indicate that, even at relatively low levels, expression 211 of PipYP63L or PipYR210Q, but not of PipYK26A, prevents the appearance of S. elongatus colonies, 212 independently of the pipX gene. Therefore, the results indicate that PipYP63L or PipYR210Q, but not 213 PipYK26A, behave as gain-of-function proteins that prevent growth of S. elongatus in our experimental 214 conditions. 215 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted December 30, 2025. ; https://doi.org/10.64898/2025.12.29.696868doi: bioRxiv preprint 6 Heterologous expression of PipY variants in E. coli parallels their behaviour in S. 216 elongatus 217 The failure to obtain viable transformants expressing the PipYP63L or PipYR210Q variants (Fig. 2) 218 raised questions on the mechanism(s) by which each of these proteins prevent S. elongatus growth. 219 Since PipY is not required for S. elongatus viability, it follows that mutations P63L or R210Q are 220 behaving as gain-of-function mutations and thus both PipYP63L and PipYR210Q must be interfering with 221 cellular processes required for culture growth. To distinguish if this interference affects signalling 222 pathways restricted to cyanobacteria or more universal processes conserved in phylogenetically distant 223 bacterial groups, we next investigated the effect of expressing each of the PipY variants in E. coli. 224 It is worth noting that all plasmids constructed in this work were obtained in E. coli XL1-Blue 225 and that this strain carries a lacIq gene to prevent expression of the cloned genes. Therefore, to unveil 226 a possible phenotypic impact of the PipY variants in E. coli the LacI repressor must be inactivated. To 227 this goal , plasmids pUAGC294, pUAGC1133, pUAGC1130 or pUAGC1129 were independently 228 introduced into E. coli XL1-Blue and their corresponding transformants were selected in parallel in 229 the absence or presence of the IPTG inducer. 230 As shown in Fig. 3a, multiple ampicillin -resistant transformants were obtained in all cases in 231 the absence of IPTG. However, colonies derived from the plasmid bearing the P63L variant were 232 noticeably smaller than those obtained with the other constructs (Fig. 3b), suggesting greater toxicity 233 for PipYP63L than for PipYR210Q. Since the latter protein is recovered at lower yield than the former 234 from E. coli extracts (Tremiño et al. 2017), differences in expression level may, at least in part, explain 235 their different toxicities. Upon IPTG addition (0.5 mM), the number of transformants remained 236 unchanged only for the construct expressing PipYK26A but decreased by 98% for PipY and by 100% 237 for both PipYP63L and PipYR210Q (Fig. 3c). Independent corroboration of these results was next provided 238 by transforming the same four plasmids into E. coli MG1655, a strain lacking the extra lacIq copy and 239 therefore unable to provide additional repression of the promoter driving expression of the PipY 240 variants (Fig. 3d). The substantially higher number of transformant obtained for PipYK26A relative to 241 PipY further supports that expression of PipY may also have a negative impact on E. coli viability that 242 can be attenuated by the K26A mutation. No colonies were obtained when the proteins expressed were 243 PipYP63L or PipYR210Q, confirming the high toxicity of both variants in E. coli. 244 Taken together, our results show that the expression of PipYR210Q and PipYP63L variants inhibit 245 growth in both S. elongatus and E. coli. Therefore, while the behaviour of PipYK26A is consistent with 246 loss-of-function (see also below), that of the “pathogenic” variants reveals gain -of-function in both 247 bacterial model systems. 248 Lys26 is required for PipY overexpression phenotypes on growth, chlorosis , cell size 249 and overaccumulation of polyP 250 To gain insights into the factors affecting PipY overexpression phenotypes in S. elongatus, we 251 addressed the impact of mutation K26A on previously studied traits, which include growth, chlorosis, 252 cell size and overaccumulation of polyP (Llop, Labella, et al. 2023). The S. elongatus strains compared 253 were pipY 1SPtrc, pipY 1SPtrc-PipY, pipY 1SPtrc-PipYK26A, pipXpipY 1SPtrc, pipXpipY 1SPtrc-PipY, 254 and pipXpipY 1SPtrc-PipYK26A. 255 The impact on the different strains o f IPTG induction on growth and pigment composition was 256 followed by performing drop-plate assays or according to the visual appearance and absorbance spectra 257 of liquid cultures (Fig. 4). As expected, no differences in growth were found amongst otherwise 258 isogenic 1SPtrc-PipY and 1SPtrc strains in the absence of IPTG (left panel in Fig. 4a). However, 259 addition of 50 μM IPTG prevented growth of 1SPtrc-PipY but not of 1SPtrc-PipYK26A or 1SPtrc strains 260 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted December 30, 2025. ; https://doi.org/10.64898/2025.12.29.696868doi: bioRxiv preprint 7 (Fig. 4a, b), indicating that mutation K26A abrogates the inhibitory effects of PipY overexpression on 261 growth. 262 Next, we compared the dynamic of pigment loss induced by PipY overexpression in the different 263 strains. As previously shown, addition of IPTG to pipY 1SPtrc-PipY cultures triggered the yellow color 264 and the progressive decrease in phycocyanin and chlorophyll a peaks characteristic of chlorosis (Fig. 265 4c, left panels ). In contrast, pipY 1SPtrc-PipYK26A cultures maintained the normal green appearance 266 and pigment composition throughout the 48 hours experiment, exactly as the negative control strain 267 pipY 1SPtrc. 268 The impact of IPTG induction on cell size was subsequently analysed by confocal microscopy at 269 0 and 48 h after IPTG induction (Fig. 4d, S1, S2). Representative micrographs showed that cell 270 appearance and cell size were indistinguishable amongst the strains before IPTG addition (0), and that 271 only 1SPtrc-PipY cells were significantly longer in both backgrounds after IPTG addition (48 h; Fig. 272 4d). The lower autofluorescence of 1SPtrc-PipY cells paralleled the loss of photosynthetic pigments 273 shown in Fig. 4c. Scatter-plot representation of cell lengths for all six strains confirmed the significant 274 increase in 1SPtrc-PipY cells after IPTG induction (Fig. 4d, right). 275 The impact of IPTG induction on polyP accumulation, another relevant phenotype associated with 276 PipY overexpression (Llop, Labella, et al. 2023), was subsequently analysed by confocal microscopy 277 in the previous strains at 0 and 48 h after IPTG induction. Representative micrographs of DAPI-stained 278 cells for the same six strains and two timepoints are shown in Fig. 4 e. To quantify the polyP signal 279 avoiding possible differences in cell number per field , the total signal in the DAPI –polyP channel 280 (ex405, em470–617) was normalized against the corresponding DAPI channel (ex405, em410 –470). 281 No differences between IPTG treated or untreated cells were observed in strain 1SPtrc-PipYK26A (Fig. 282 4f), indicating that the K26A mutation abolished the over accumulation of polyP resulting from PipY 283 overexpression. 284 In summary, all four PipY overexpression-related phenotypic features analyzed here are abolished 285 by the K26A mutation, indicating that they require the PLP cofactor and/or the integrity of Lys26 at 286 PipY. 287 Inactivation of pipX affects overaccumulation of polyP but has no effect on other PipY 288 overexpression phenotypes. 289 Comparison between pipY and pipXpipY strains expressing PipY or PipYK26A showed no 290 phenotypic differences attributable to the presence or absence of pipX regarding growth, chlorosis or 291 cell size ( Fig. 4a,b,c,d ). However, although pipXpipY 1SPtrc-PipY cells showed significant 292 accumulation of polyP afer 48h of induction, it was not as large as in pipY 1SPtrc-PipY cells (Fig. 4e). 293 PolyP signals were at least 1.7 -fold lower from pipXpipY 1SPtrc-PipY than from pipY 1SPtrc-PipY 294 cells (Fig. 4f), thus indicating that the pipX gene is required for the very high levels of polyP associated 295 to PipY overexpression. 296 In summary, our results indicate that PipX plays a positive regulatory role in polyP accumulation, 297 suggesting that both PipX and PipY may be required to increase polyP synthesis under certain 298 environmental conditions. On the other hand, we cannot yet exclude regulation by PipX of the cellular 299 targets involved in the other phenotypic traits analysed here in S. elongatus cells or cultures. Since we 300 studied the effect of the presence or absence of pipX gene in conditions in which the PipX/PipY is 301 abnormally low, a regulatory role of PipX in the studied traits could have been missed, particularly if 302 it was a negative role. 303 Mutation K26A increase the levels of PipY in S. elongatus in a PipX independent 304 manner 305 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted December 30, 2025. ; https://doi.org/10.64898/2025.12.29.696868doi: bioRxiv preprint 8 Given that the phenotypes discussed above depend on PipY overaccumulation in S. elongatus 306 ((Labella et al. 2017; Llop, Labella, et al. 2023) ; Fig. 4), it was important to exclude the possibility 307 that mutation K26A impaired the stability of PipY therefore prevent ing accumulation of very high 308 levels of PipYK26A in the presence of IPTG. To investigate that possibility , samples of pipY 1SPtrc-309 PipY, pipXpipY 1SPtrc-PipY, pipY 1SPtrc-PipYK26A, and pipXpipY 1SPtrc-PipYK26A strains were taken 310 at 0, 6, and 24 h timepoints after IPTG induction, and subsequently analysed by Western blot with 311 anti-PipY (Figs. 5 and S3). 312 Although no differences between strains could be observed at timepoint 0, upon IPTG addition 313 the signal from the PipY band was higher at both 6 and 24 h timepoints for PipY K26A than for PipY 314 (Fig. 5a). Quantification of band signals confirmed a ≈ 2-fold increase in PipYK26A levels with respect 315 to PipY (Fig. 5b) regardless of the presence or absence of the pipX gene (Fig. 5c). 316 Therefore, the failure of PipYK26A to trigger any one of the PipY overexpression phenotypes in S. 317 elongatus analysed here is not the result of lower accumulation of PipYK26A in the presence of IPTG, 318 indicating that either the PLP cofactor or Lys26 itself is required for PipY -triggered growth arrest, 319 chlorosis, increased cell length and overaccumulation of polyP in S. elongatus. 320

321 In this work we have exploited the existing structural and functional information on the 322 cyanobacterial model protein PipY to gain insights into the PLPBP family by focussing on the 323 phenotypic effects conferred by PipY variants expressing point mutations that prevent PLP binding 324 (K26A) or that mimic pathogenic changes (P63L and R210Q) found in the orthologous human protein. 325 The apparent correlation between the severity of PLPBP mutations causing vitamin B6-dependent 326 epilepsy and the decreased ability of some of the corresponding PLPHP variants to bind the PLP 327 cofactor (Johnstone et al. 2019) speaks of the importance of the PLP cofactor for PL PBP function in 328 the homeostasis of B 6 vitamers. For instances, both PipYP63L and PLPHPP87L were respectively less 329 impaired than PipYR210Q and PLPHPR241Q in vitro (Tremiño et al. 2017; Tremiño et al. 2018) as well 330 as in E. coli, where PLPHPP87L but not PLPHP R241Q complemented the pyridoxine-sensitivity of the 331 yggS null mutant (Darin et al. 2016) . It is worth noting that while the P63L mutation did not impair 332 any of the parameters assayed at PipY, including protein yield, thermal stability, proper fold or PLP -333 content, the R210Q change resulted in a lower yield of the recombinant protein that, although well 334 folded, contained hardly any PLP and showed decreased stability at high temperature. Pro63/87 is 335 located at the tip of the turn between α3-β3, facing outwards and well away from the PLP cofactor, 336 while Arg210/241 is an invariant r esidue that establish contacts with the PLP cofactor (Fig. 1) . 337 Considering that proteins with obvious defects in PLP binding were less functional in vivo , the 338 prediction was that PipYK26A would be at least as affected as PipYR210Q when tested in S. elongatus. 339 Phenotypic analyses in S. elongatus indicated that mutation K26A caused complete loss of 340 function. Despite the high levels of PipYK26A protein achieved in S. elongatus (Fig. 5a, b) this protein 341 did not confer any of the phenotypes associated with PipY overexpression, or indeed any other of the 342 phenotypic features analysed in S. elongatus. Furthermore, while h eterologous expression of PipY 343 decreased viability in E. coli, expression of PipYK26A did not, indicating that the differential behaviour 344 of wild type and mutant proteins is maintained across bacterial species. Since the equivalent mutation 345 (K36A) at the E. coli homolog YggS also increases the stability of the protein (Tramonti et al. 2022), 346 our results emphasise the importance and common roles of this invariant lysine within bacterial 347 PLPBPs. Not surprisingly, PipX, a protein unique to cyanobacteria, appears to play no role in the 348 different stability of PipY and PipYK26A proteins (Fig. 5c). 349 Interestingly, the PipYR210Q variant, which is also impaired in PLP binding (Tremiño et al. 2017), 350 did not decrease PipY toxicity. On the contrary, it resulted toxic in both S. elongatus and E. coli (Figs. 351 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted December 30, 2025. ; https://doi.org/10.64898/2025.12.29.696868doi: bioRxiv preprint 9 2 and 3) and thus the opposite phenotypes caused by two mutations abolishing or drastically impairing 352 PLP binding argues against PipY toxicity requiring the presence of the PLP cofactor. It is thus likely 353 that toxicity depends on a particular protein conformation which could not be adopted by PipYK26A. 354 Since our results indicate that Lys26 must have additional functions beyond cofactor binding, it is 355 reasonable to think that PipY must interact with yet unknown targets in bacterial cells. In this context, 356 the simplest interpretation for the opposite phenotypes of PipYK26A and PipYR210Q is that Lys26, and 357 not PLP itself, was required for PipY toxicity and that mutation K26A would likely impair interactions 358 with those hypothetical targets while R210Q, as well as P63L, would favour it. 359 Since two mutations (P63L and R210Q) that have different effects on the affinity for PLP confer 360 the same gain-of-function phenotype, increasing PipY toxicity in both S. elongatus and E. coli cells 361 even when expressed at very low levels (Figs. 2 and 3), the PLP cofactor would not be directly involved 362 in PipY toxicity. Instead, the results suggested that PipY interacts with yet unknown targets in both 363 bacterial systems and that the PipYP63L and PipYR210Q variants do it to greater extent. 364 Since PipYP63L conferred higher toxicity than PipYR210Q in bacteria (Fig. 3), it appears that the 365 P63L mutation confers less severe loss-of-function and more severe gain-of-function phenotypes than 366 R210Q for respectively, PLP -dependent and PLP -independent activities. These results, in complete 367 agreement with the higher protein yield of PipYP63L found in E. coli (Tremiño et al. 2017), support the 368 idea that for PLP-independent functions, that is, toxicity, there would be two alternative conformations 369 of relevance, one represented by PipYK26A (inactive) and the other one represented by PipYP63L or 370 PipYR210Q (active). Taken together, the results suggest that in addition to vitamer B 6 homeostasis 371 PLPBP is involved in PLP-independent activities. 372 Concerning the nature of PipY interactants, a recent work shows that PLPBP members are RNA 373 binding proteins and that the apo -forms of YggS and human PLPHP bind RNA with much greater 374 affinity than the holo -forms or the YggS K36A variant, which adopts a holo -like conformation despite 375 the absence of PLP (Graziani et al. 2024). According to that work, PLP would have a role as an effector 376 molecule inhibiting the binding of RNA to PLPBPs (Graziani et al. 2024). It is that tempting to propose 377 that PipY toxicity is related to its ability to bind RNA and that the apo-form is involved. 378 A comparative analysis of the PipY structures from S. elongatus (PLP-bound PipY, PDB: 5NM8, 379 versus apo PipY, PDB: 5NLC) reveal changes associated with PLP binding at helices α1, α2, α6, and 380 α9, as well as the β5–α6, β6–α7, and β7–α8 loops (Fig. 6a,b,c; (Tremiño et al. 2017) ). Use of the 381 KVFinder web tool (https://kvfinder-web.cnpem.br/; (Guerra et al. 2020; Guerra et al. 2023)) detected 382 differences in the solvent -accessible PLP-binding cavity, with its volume increasing from 144.94 ų 383 to 161.57 ų upon PLP release. Thus, close and open conformations of the protein surface can be 384 respectively associated to the holo or apo forms. 385 A working model for PipY functions is schematically illustrated in Fig. 6d,e. We propose that like 386 YggSK36A, PipYK26A would adopt a holo-like conformation with low affinity for RNA and that in the 387 case of PipYR210Q, insufficient binding of the PLP cofactor would displace the equilibrium towards the 388 apo-form. Since PipYP63L is not impaired in PLP binding, displacement of the equilibrium towards the 389 apo-form of PipY by mutation P63L requires a different explanation. We speculate that Pro63 may 390 play a role in maintaining the holo-form by imposing rigidity to the α3-β3 loop. Based on the different 391 expression levels for PipYR210Q and PipYP63L (Tremiño et al. 2017) , the prediction is that in vivo 392 virtually all of PipYR210Q would be in the apo-form, while for the more abundant protein PipYP63L there 393 would still be enough amount of the minoritarian holo form to explain its function on vitamer B 6 394 homeostasis. In support of the model, the program GPSite (https://bio-web1.nscc-gz.cn/app/GPSite; 395 (Yuan et al. 2024)) predicts that mutation K26A, but not R210Q or P63L, impair RNA binding (Table 396 S2). 397 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted December 30, 2025. ; https://doi.org/10.64898/2025.12.29.696868doi: bioRxiv preprint 10 Last but not least, t ranscriptomic and proteomic changes associated to PLPBP deficient cells 398 (Labella et al. 2017; Fux and Sieber 2020) support the recent proposal that PLPBP may be part of a 399 regulatory mechanism of translation linked to amino acid levels (Graziani et al. 2024) . Since PipX, a 400 multifunctional protein co -expressed with PipY , is also involved in carbon/nitrogen homeostasis, 401 transcription and translation regulation (Llácer et al. 2010; Espinosa et al. 2014; Labella et al. 2016; 402 Labella et al. 2017; Cantos et al. 2019; Jerez et al. 2021) , understanding the functions of the pipXY 403 operon in cyanobacteria should help address the universal functions of PLPBP. 404 Data availability statement 405 The original contributions presented in the study are included in the article/Supplementary 406 material. Further inquiries can be directed to the corresponding author. 407 Acknowledgments 408 The authors thank S. Bibak for Western blots, R. Cantos, T. Mata -Balaguer, P. Salina s, and L. 409 Fuertes-García for technical contributions or advice. 410 Study funding 411 This work was supported by grant PID2023 -149456NB-I00, funded by 412 MCIN/AEI/10.13039/501100011033 from the Spanish Government, and grants VIGROB23 -126 and 413 GRE20-04-C from the University of Alicante to A.C. 414 Conflict of interest 415 The authors declare that the research was conducted in the absence of any commercial or financial 416 relationships that could be construed as a potential conflict of interest. 417

418 Bullock WO, Fernandez JM, Short JM. 1987. XL1 -Blue—a high -efficiency plasmid transforming 419 recA Escherichia coli strain with β-galactosidase selection. Biotechniques. 5(3):376–379. 420 Burillo S, Luque I, Fuentes I, Contreras A. 2004. Interactions between the Nitrogen Signal 421 Transduction Protein PII and N -Acetyl Glutamate Kinase in Organisms That Perform Oxygenic 422 Photosynthesis. J Bacteriol. 186(11):3346–3354. doi:10.1128/JB.186.11.3346-3354.2004. 423 Cantos R, Labella JI, Espinosa J, Contreras A. 2019. The nitrogen regulator PipX acts in cis to prevent 424 operon polarity. Environ Microbiol Rep. 11(4):495–507. doi:10.1111/1758-2229.12688. 425 Darin N, Reid E, Prunetti L, Samuelsson L, Husain RA, Wilson M, El Yacoubi B, Footitt E, Chong 426 WK, Wilson LC, et al. 2016. Mutations in PROSC Disrupt Cellular Pyridoxal Phosphate Homeostasis 427 and Cause Vitamin-B6-Dependent Epilepsy. The American Journal of Human Genetics. 99(6):1325 –428 1337. doi:10.1016/j.ajhg.2016.10.011. 429 Espinosa J, Castells MA, Laichoubi KB, Contreras A. 2009. Mutations at pipX Suppress Lethality of 430 P II -Deficient Mutants of Synechococcus elongatus PCC 7942. J Bacteriol. 191(15):4863 –4869. 431 doi:10.1128/JB.00557-09. 432 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted December 30, 2025. ; https://doi.org/10.64898/2025.12.29.696868doi: bioRxiv preprint 11 Espinosa J, Forchhammer K, Burillo S, Contreras A. 2006. Interaction network in cyanobacterial 433 nitrogen regulation: PipX, a protein that interacts in a 2‐oxoglutarate dependent manner with PII and 434 NtcA. Mol Microbiol. 61(2):457–469. doi:10.1111/j.1365-2958.2006.05231.x. 435 Espinosa J, Forchhammer K, Contreras A. 2007. Role of the Synechococcus PCC 7942 nitrogen 436 regulator protein PipX in NtcA -controlled processes. Microbiology (N Y). 153(3):711 –718. 437 doi:10.1099/mic.0.2006/003574-0. 438 Espinosa J, Labella JI, Cantos R, Contreras A. 2018. Energy drives the dynamic localization of 439 cyanobacterial nitrogen regulators during diurnal cycles. Environ Microbiol. 20(3):1240 –1252. 440 doi:10.1111/1462-2920.14071. 441 Espinosa J, Rodríguez -Mateos F, Salinas P, Lanza VF, Dixon R, de la Cruz F, Contreras A. 2014. 442 PipX, the coactivator of NtcA, is a global regulator in cyanobacteria. Proceedings of the National 443 Academy of Sciences. 111(23):201404030–201404097. doi:10.1073/pnas.1404097111. 444 Eswaramoorthy S, Gerchman S, Graziano V, Kycia H, Studier FW, Swaminathan S. 2003. Structure 445 of a yeast hypothetical protein selected by a structural genomics approach. Acta Crystallogr D Biol 446 Crystallogr. 59(1):127–135. doi:10.1107/S0907444902018012. 447 Farkas P, Fitzpatrick TB. 2024. Two pyridoxal phosphate homeostasis proteins are essential for 448 management of the coenzyme pyridoxal 5′ -phosphate in Arabidopsis. Plant Cell. 36(9):3689 –3708. 449 doi:10.1093/plcell/koae176. 450 Forcada-Nadal A, Bibak S, Salinas P, Contreras A, Rubio V, Llácer JL. 2025. Structures of the 451 cyanobacterial nitrogen regulators NtcA and PipX complexed to DNA shed light on DNA binding by 452 NtcA and implicate PipX in the recruitment of RNA polymerase. Nucleic Acids Res. 53(4). 453 doi:10.1093/nar/gkaf096. 454 Forchhammer K, Schwarz R. 2019. Nitrogen chlorosis in unicellular cyanobacteria – a developmental 455 program for surviving nitrogen deprivation. Environ Microbiol. 21(4):1173–1184. doi:10.1111/1462-456 2920.14447. 457 Froger A, Hall JE. 2007. Transformation of Plasmid DNA into E. coli Using the Heat Shock Method. 458 Journal of Visualized Experiments.(6). doi:10.3791/253. 459 Fux A, Sieber SA. 2020. Biochemical and Proteomic Studies of Human Pyridoxal 5′ -Phosphate-460 Binding Protein (PLPBP). ACS Chem Biol. 15(1):254–261. doi:10.1021/acschembio.9b00857. 461 Giner-Lamia J, Robles-Rengel R, Hernández-Prieto MA, Muro-Pastor MI, Florencio FJ, Futschik ME. 462 2017. Identification of the direct regulon of NtcA during early acclimation to nitrogen starvation in the 463 cyanobacterium Synechocystis sp. PCC 6803. Nucleic Acids Res. 45(20):11800 –11820. 464 doi:10.1093/nar/gkx860. 465 Graziani C, Barile A, Parroni A, di Salvo ML, De Cecio I, Colombo T, Babor J, de Crécy‐Lagard V, 466 Contestabile R, Tramonti A. 2024. The ubiquitous pyridoxal 5′‐phosphate‐binding protein is also an 467 RNA‐binding protein. Protein Science. 33(12). doi:10.1002/pro.5242. 468 Guerra JV da S, Ribeiro Filho HV, Bortot LO, Honorato RV, Pereira JG de C, Lopes-de-Oliveira PS. 469 2020. ParKVFinder: A thread -level parallel approach in biomolecular cavity detection. SoftwareX. 470 12:100606. doi:10.1016/j.softx.2020.100606. 471 Guerra JVS, Ribeiro -Filho H V, Pereira JGC, Lopes -de-Oliveira PS. 2023. KVFinder -web: a web -472 based application for detecting and characterizing biomolecular cavities. Nucleic Acids Res. 473 51(W1):W289–W297. doi:10.1093/nar/gkad324. 474 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted December 30, 2025. ; https://doi.org/10.64898/2025.12.29.696868doi: bioRxiv preprint 12 Guyer MS, Reed RR, Steitz JA, Low KB. 1981. Identification of a Sex -factor-affinity Site in E. coli 475 as gamma delta. Cold Spring Harb Symp Quant Biol. 45(0):135 –140. 476 doi:10.1101/SQB.1981.045.01.022. 477 He S, Chen Y, Wang L, Bai X, Bu T, Zhang J, Lu M, Ha N-C, Quan C, Nam KH, et al. 2022. Structural 478 and Functional Analysis of the Pyridoxal Phosphate Homeostasis Protein YggS from Fusobacterium 479 nucleatum. Molecules. 27(15):4781. doi:10.3390/molecules27154781. 480 Ito T. 2022. Role of the conserved pyridoxal 5ʹ-phosphate-binding protein YggS/PLPBP in vitamin B6 481 and amino acid homeostasis. Biosci Biotechnol Biochem. 86(9):1183 –1191. 482 doi:10.1093/bbb/zbac113. 483 Ito T, Hori R, Hemmi H, Downs DM, Yoshimura T. 2020. Inhibition of glycine cleavage system by 484 pyridoxine 5′‐phosphate causes synthetic lethality in glyA yggS and serA yggS in Escherichia coli. Mol 485 Microbiol. 113(1):270–284. doi:10.1111/mmi.14415. 486 Ito T, Yamamoto K, Hori R, Yamauchi A, Downs DM, Hemmi H, Yoshimura T. 2019. Conserved 487 Pyridoxal 5’-Phosphate-Binding Protein YggS Impacts Amino Acid Metabolism through Pyridoxine 488 5’-Phosphate in Escherichia coli . Zhou N -Y, editor. Appl Environ Microbiol. 85(11):NA -NA. 489 doi:10.1128/AEM.00430-19. 490 Jerez C, Llop A, Salinas P, Bibak S, Forchhammer K, Contreras A. 2024. Analysing the 491 Cyanobacterial PipX Interaction Network Using NanoBiT Complementation in Synechococcus 492 elongatus PCC7942. Int J Mol Sci. 25(9):4702. doi:10.3390/ijms25094702. 493 Jerez C, Salinas P, Llop A, Cantos R, Espinosa J, Labella JI, Contreras A. 2021. Regulatory 494 Connections Between the Cyanobacterial Factor PipX and the Ribosome Assembly GTPase EngA. 495 Front Microbiol. 12(NA):781760-NA. doi:10.3389/fmicb.2021.781760. 496 Johnstone DL, Al-Shekaili HH, Tarailo-Graovac M, Wolf NI, Ivy AS, Demarest S, Roussel Y, Ciapaite 497 J, van Roermund CWT, Kernohan KD, et al. 2019. PLPHP deficiency: clinical, genetic, biochemical, 498 and mechanistic insights. Brain. 142(3):542–559. doi:10.1093/brain/awy346. 499 Labella JI, Cantos R, Espinosa J, Forcada-Nadal A, Rubio V, Contreras A. 2017. PipY, a Member of 500 the Conserved COG0325 Family of PLP -Binding Proteins, Expands the Cyanobacterial Nitrogen 501 Regulatory Network. Front Microbiol. 8(NA):1244. doi:10.3389/fmicb.2017.01244. 502 Labella JI, Cantos R, Salinas P, Espinosa J, Contreras A. 2020. Distinctive Features of PipX, a Unique 503 Signaling Protein of Cyanobacteria. Life. 10(6):79. doi:10.3390/life10060079. 504 Labella JI, Obrebska A, Espinosa J, Salinas P, Forcada -Nadal A, Tremiño L, Rubio V, Contreras A. 505 2016. Expanding the Cyanobacterial Nitrogen Regulatory Network: The GntR -Like Regulator PlmA 506 Interacts with the PII-PipX Complex. Front Microbiol. 7(NA):1677. doi:10.3389/fmicb.2016.01677. 507 Laichoubi KB, Beez S, Espinosa J, Forchhammer K, Contreras A. 2011. The nitrogen interaction 508 network in Synechococcus WH5701, a cyanobacterium with two PipX and two PII -like proteins. 509 Microbiology (NY). 157(4):1220–1228. doi:10.1099/mic.0.047266-0. 510 Laichoubi KB, Espinosa J, Castells MA, Contreras A. 2012. Mutational Analysis of the Cyanobacterial 511 Nitrogen Regulator PipX. Janssen PJ, editor. PLoS One. 7(4):e35845. 512 doi:10.1371/journal.pone.0035845. 513 Llácer JL, Espinosa J, Castells MA, Contreras A, Forchhammer K, Rubio V. 2010. Structural basis for 514 the regulation of NtcA-dependent transcription by proteins PipX and PII. Proceedings of the National 515 Academy of Sciences. 107(35):15397–15402. doi:10.1073/pnas.1007015107. 516 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted December 30, 2025. ; https://doi.org/10.64898/2025.12.29.696868doi: bioRxiv preprint 13 Llop A, Bibak S, Cantos R, Salinas P, Contreras A. 2023. The ribosome assembly GTPase EngA is 517 involved in redox signaling in cyanobacteria. Front Microbiol. 14(NA):1242616 -NA. 518 doi:10.3389/fmicb.2023.1242616. 519 Llop A, Labella JI, Borisova M, Forchhammer K, Selim KA, Contreras A. 2023. Pleiotropic effects of 520 PipX, PipY, or RelQ overexpression on growth, cell size, photosynthesis, and polyphosphate 521 accumulation in the cyanobacterium Synechococcus elongatus PCC7942. Front Microbiol. 522 14(NA):1141775-NA. doi:10.3389/fmicb.2023.1141775. 523 Moronta-Barrios F, Espinosa J, Contreras A. 2013. Negative control of cell size in the cyanobacterium 524 Synechococcus elongatus PCC 7942 by the essential response regulator RpaB. FEBS Lett. 587(5):504–525 509. doi:10.1016/j.febslet.2013.01.023. 526 Prunetti L, El Yacoubi B, Schiavon CR, Kirkpatrick E, Huang L, Bailly M, El Badawi -Sidhu M, 527 Harrison K, Gregory JF, Fiehn O, et al. 2016. Evidence that COG0325 proteins are involved in PLP 528 homeostasis. Microbiology (N Y). 162(4):694–706. doi:10.1099/mic.0.000255. 529 Rippka R, Deruelles J, Waterbury JB, Herdman M, Stanier RY. 1979. Generic Assignments, Strain 530 Histories and Properties of Pure Cultures of Cyanobacteria. Microbiology (N Y). 111(1):1 –61. 531 doi:10.1099/00221287-111-1-1. 532 RStudio: Integrated Development for R. RStudio. 2020. [accessed 2025 Jan 1]. 533 http://www.rstudio.com/. 534 Salinas P, Bibak S, Cantos R, Tremiño L, Jerez C, Mata T, Contreras A. 2024. Studies on the PII -535 PipX-NtcA Regulatory Axis of Cyanobacteria Provide Novel Insights into the Advantages and 536

of Two-Hybrid Systems for Protein Interactions. Int J Mol Sci. doi: 10.3390/ijms25105429 537 Schneider G, Käck H, Lindqvist Y. 2000. The manifold of vitamin B 6 dependent enzymes. Structure. 538 8(1):R1–R6. doi:10.1016/S0969-2126(00)00085-X. 539 Schwarz R, Grossman AR. 1998. A response regulator of cyanobacteria integrates diverse 540 environmental signals and is critical for survival under extreme conditions. Proceedings of the National 541 Academy of Sciences. 95(18):11008–11013. doi:10.1073/pnas.95.18.11008. 542 Spät P, Klotz A, Rexroth S, Maček B, Forchhammer K. 2018. Chlorosis as a Developmental Program 543 in Cyanobacteria: The Proteomic Fundament for Survival and Awakening. Molecular & Cellular 544 Proteomics. 17(9):1650–1669. doi:10.1074/mcp.RA118.000699. 545 Taton A, Erikson C, Yang Y, Rubin BE, Rifkin SA, Golden JW, Golden SS. 2020. The circadian clock 546 and darkness control natural competence in cyanobacteria. Nat Commun. 11(1):1688. 547 doi:10.1038/s41467-020-15384-9. 548 Tramonti A, Ghatge MS, Babor JT, Musayev FN, di Salvo ML, Barile A, Colotti G, Giorgi A, Paredes 549 SD, Donkor AK, et al. 2022. Characterization of the Escherichia coli pyridoxal 5′‐phosphate 550 homeostasis protein (YggS): Role of lysine residues in PLP binding and protein stability. Protein 551 Science. 31(11):e4471-NA. doi:10.1002/pro.4471. 552 Tremiño L, Forcada‐Nadal A, Contreras A, Rubio V. 2017. Studies on cyanobacterial protein PipY 553 shed light on structure, potential functions, and vitamin B 6 ‐dependent epilepsy. FEBS Lett. 554 591(20):3431–3442. doi:10.1002/1873-3468.12841. 555 Tremiño L, Forcada -Nadal A, Rubio V. 2018. Insight into vitamin B 6 -dependent epilepsy due to 556 PLPBP (previously PROSC) missense mutations. Hum Mutat. 39(7):1002 –1013. 557 doi:10.1002/humu.23540. 558 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted December 30, 2025. ; https://doi.org/10.64898/2025.12.29.696868doi: bioRxiv preprint 14 Tremiño L, Llop A, Rubio V, Contreras A. 2022. The Conserved Family of the Pyridoxal Phosphate-559 Binding Protein (PLPBP) and Its Cyanobacterial Paradigm PipY. Life. 12(10):1622. 560 doi:10.3390/life12101622. 561 Vu HN, Ito T, Downs DM. 2020. The Role of YggS in Vitamin B6 Homeostasis in Salmonella enterica 562 Is Informed by Heterologous Expression of Yeast SNZ3. Metcalf WW, editor. J Bacteriol. 563 202(22):NA-NA. doi:10.1128/JB.00383-20. 564 Yuan Q, Tian C, Yang Y. 2024. Genome-scale annotation of protein binding sites via language model 565 and geometric deep learning. Elife. 13. doi:10.7554/eLife.93695. 566 567 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted December 30, 2025. ; https://doi.org/10.64898/2025.12.29.696868doi: bioRxiv preprint 15 Figures 568 Figure 1. Conservation between cyanobacterial and human PLPBP orthologs and location of 569 relevant residues mutated in this work . a) Structure of PipY from S. elongatus (PDB ID: 5NM8), 570 colored according to the evolutionary conservation score of residues between PipY and its human 571 ortholog (PLPHP). Color coding was defined based on the conservation score of an automatic 572 ClustalW alignment of both protein sequences, as indicated in the legend. Re levant re sidues are 573 mapped as spheres. PLP is shown in stick representation and labeled, with C, O, N, and P atoms colored 574 yellow, red, blue, and orange, respectively. b, c) I-TASSER superimposition of the structural model 575 of PLPHP over PipY colored purple and orange, respectively, with the residues of interest and 576 secondary-structure elements indicated. 577 Figure 2. Constructs to overexpress PipY derivatives in S. elongatus and segregation analysis. a) 578 Schematic representation of the NSI at the S. elongatus genomic region and of the constructs to express 579 the PipY derivatives under the control of the Ptrc promoter in pipY or pipXpipY null backgrounds. 580 Primer positions for PCR analysis are depicted by blue arrows, with the size of products on the right. 581 b) Representative agarose gels showing PCR products obtained in the indicated strains and 582 backgrounds after up to ten selective transfers, with the expected amplicon depicted by arrows on the 583 right. M, 100 bp size marker. The number of colonies obtained in two independent transformations 584 was >103 for PipY or PipY K26A, and <15 for PipY P63L or PipYR210Q, with no significative differences 585 between both backgrounds. * refers to the WT, K26A, P63L, and R210Q versions of PipY. 586 Figure 3. Effects of PipY variants in E. coli colony yield. Results of transformations with 50 ng of 587 plasmids expressing the indicated PipY variants under the Ptrc promoter. a) XL1-blue strain (1/10 588 dilution). Insets show a 3× zoom of the indicated regions. b) Quantification of the Feret’s diameter of 589 colonies, normalized to those of the PipY strain. The median, the interquartile range (box), and outliers 590 (dots) are shown, along with the number of colonies measured. c) XL1-blue strain (1/10 dilution) in 591 the presence of 0.5 mM IPTG. d) MG1655 strain (9/10 dilution). The number of colonies obtained is 592 indicated inside each plate. Data is based on two independent experiments. 593 Figure 4. Effects of K26A mutation or pipX inactivation on PipY overexpression phenotypes in 594 S. elongatus. Strains Ptrc (-) and derivatives expressing PipY (Y) or PipYK26A (K26A) in the indicated 595 backgrounds, grown in BG11 and, when indicated, supplemented with 50 µM IPTG. a) Drop-plate 596 assay (5 μL; initial OD 750nm= 0.5) with the dilutions indicated at the top. b) Growth curves (mean ± 597 standard deviation). c) Whole absorbance spectra, indicating the peaks of phycocyanin (PC) and 598 chlorophyll a (Chl a), and i nsets showing cultures after 48 h of IPTG induction. d) Representative 599 confocal micrographs of autofluorescence signal (+40% brightness, –40% contrast; left) and scatterplot 600 of cells length (n = 180 cells; right). Means are indicated by black lines and numbers inside the graphs. 601 e) Representative confocal micrographs of DAPI-polyP channel. f) Ratio of the polyP signal (48 / 0 h) 602 referred to the Ptrc strain (mean ± standard deviation). Pairwise comparisons using the Mann–Whitney 603 U test with Bonferroni correction were performed. Significance levels were denoted as p ≤ 0.01 (**) 604 or ≤0.001 (***). Effect size r was interpreted as large according to Cohen’s thresholds (r > 0.5). Data 605 from two independent experiments are presented. 606 Figure 5. Effects of K26A mutation or pipX inactivation on PipY levels. a) Representative 607 immunodetections (α-PipY) and loading control (LC) of Ptrc strains expressing PipY (Y) or PipYK26A 608 (K26A) in the indicated null mutant grew in the presence of 1 mM of IPTG. b) Ratio of PipYK26A/PipY 609 protein levels, normalized to loading control. c) pipXpipY/pipX ratio at 24 h. Data are presented as 610 means and error bars (standard deviation) from two biological replicates. 611 Figure 6. PLP driven changes and model for PipY functions based on alternative conformations. 612 a) Superposition of PipY-apo (blue) onto PipY-PLP (yellow). b) PipY–PLP. c) PipY-apo. The dashed 613 ellipses mark solvent-accessible regions. d) Schematic representation of the holo and apo forms bound 614 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted December 30, 2025. ; https://doi.org/10.64898/2025.12.29.696868doi: bioRxiv preprint 16 to their respective ligands, with indication of functions. e) Predicted effect of the indicated mutations 615 on protein conformation. See text for additional details. 616 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted December 30, 2025. ; https://doi.org/10.64898/2025.12.29.696868doi: bioRxiv preprint preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted December 30, 2025. ; https://doi.org/10.64898/2025.12.29.696868doi: bioRxiv preprint 1 Tables 1 Table 1. Plasmids. 2 Plasmid Description, Relevant Characteristics Source or Reference pUAGC127 pipY replaced with cat, ApR CmR (Labella et al. 2017) pUAGC128 pipXpipY replaced with cat, ApR CmR (Labella et al. 2017) pUAGC280 (cs3 lacI Ptrc) into NSI, ApR SmR (Moronta-Barrios et al. 2013) pUAGC294 (cs3 lacI Ptrc:pipY) into NSI, ApR SmR (Labella et al. 2017) pUAGC1129 (cs3 lacI Ptrc:pipY628cga>caa) into NSI, ApR SmR This work pUAGC1130 (cs3 lacI Ptrc:pipY199ccc>ctg) into NSI, ApR SmR This work pUAGC1133 (cs3 lacI Ptrc:pipY76aag>gcg) into NSI, ApR SmR This work 3 4 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted December 30, 2025. ; https://doi.org/10.64898/2025.12.29.696868doi: bioRxiv preprint 2 Table 2. Strains 1 Strain Genotype, Relevant Characteristics Source or

E. coli XL1-Blue recA1 endA1 gyrA96 thi-1 hsdR17 supE44 relA1 lac [F′ proAB lacIqZ∆M15 Tn10 (TetR)] (Bullock et al. 1987) E. coli MG1655 F- λ- ilvG- rfb-50 rph-1 (Guyer et al. 1981) WT Synechococcus elongatus PCC7942 Pasteur Culture Collection pipY ΔpipY::cat, CmR (Labella et al. 2017) pipXY ΔpipXpipY::cat, CmR (Labella et al. 2017) pipY 1SPtrc ΔpipY::cat NSI::(cs3 lacI Ptrc), SmR This work pipY 1SPtrc-PipY ΔpipY::cat NSI::(cs3 lacI Ptrc:pipY), CmR SmR (Labella et al. 2017) pipY 1SPtrc-PipYK26A ΔpipY::cat NSI::( (cs3 lacI Ptrc:pipY76aag>gcg), CmR SmR This work pipY 1SPtrc-PipYP63L ΔpipY::cat NSI::(cs3 lacI Ptrc:pipY199ccc>ctg), CmR SmR This work pipY 1SPtrc-PipYR210Q ΔpipY::cat NSI::(cs3 lacI Ptrc:pipY628cga>caa), CmR SmR This work pipXpipY 1SPtrc ΔpipXpipY::cat NSI::(cs3 lacI Ptrc), CmR SmR This work pipXpipY 1SPtrc-PipY ΔpipXpipY::cat NSI::(cs3 lacI Ptrc:pipY), CmR SmR (Labella et al. 2017) pipXpipY 1SPtrc-PipYK26A ΔpipXpipY::cat NSI::(cs3 lacI Ptrc:pipY76aag>gcg), CmR SmR This work pipXpipY 1SPtrc-PipYP63L ΔpipXpipY::cat NSI::(cs3 lacI Ptrc:pipY199ccc>ctg), CmR SmR This work pipXpipY 1SPtrc-PipYR210Q ΔpipXpipY::cat NSI::(cs3 lacI Ptrc:pipY628cga>caa), CmR SmR This work 2 preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted December 30, 2025. ; https://doi.org/10.64898/2025.12.29.696868doi: bioRxiv preprint preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted December 30, 2025. ; https://doi.org/10.64898/2025.12.29.696868doi: bioRxiv preprint preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted December 30, 2025. ; https://doi.org/10.64898/2025.12.29.696868doi: bioRxiv preprint preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted December 30, 2025. ; https://doi.org/10.64898/2025.12.29.696868doi: bioRxiv preprint preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted December 30, 2025. ; https://doi.org/10.64898/2025.12.29.696868doi: bioRxiv preprint preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted December 30, 2025. ; https://doi.org/10.64898/2025.12.29.696868doi: bioRxiv preprint preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for thisthis version posted December 30, 2025. ; https://doi.org/10.64898/2025.12.29.696868doi: bioRxiv preprint