The role of the tyrosine kinase Wzc (Sll0923) and the phosphatase Wzb (Slr0328) in the production of extracellular polymeric substances (EPS) by Synechocystis PCC 6803

Abstract Many cyanobacteria produce extracellular polymeric substances (EPS) mainly composed of heteropolysaccharides with unique characteristics that make them suitable for biotechnological applications. However, manipulation/optimization of EPS biosynthesis/characteristics is hindered by a poor understanding of the production pathways and the differences between bacterial species. In this work, genes putatively related to different pathways of cyanobacterial EPS polymerization, assembly, and export were targeted for deletion or truncation in the unicellular Synechocystis sp. PCC 6803. No evident phenotypic changes were observed for some mutants in genes occurring in multiple copies in Synechocystis genome, namely ∆wzy (∆sll0737), ∆wzx (∆sll5049), ∆kpsM (∆slr2107), and ∆kpsM∆wzy (∆slr2107∆sll0737), strongly suggesting functional redundancy. In contrast, Δwzc (Δsll0923) and Δwzb (Δslr0328) influenced both the amount and composition of the EPS, establishing that Wzc participates in the production of capsular (CPS) and released (RPS) polysaccharides, and Wzb affects RPS production. The structure of Wzb was solved (2.28 Å), revealing structural differences relative to other phosphatases involved in EPS production and suggesting a different substrate recognition mechanism. In addition, Wzc showed the ATPase and autokinase activities typical of bacterial tyrosine kinases. Most importantly, Wzb was able to dephosphorylate Wzc in vitro, suggesting that tyrosine phosphorylation/dephosphorylation plays a role in cyanobacterial EPS production.


| INTRODUC TI ON
Polysaccharide-based biopolymers can provide a diverse and powerful platform to deliver a wide range of biological and functional properties to the industrial toolbox. However, and despite the overwhelming diversity of polymers synthesized by microorganisms, bacterial polysaccharides are still underrepresented in the market (Roca, Alves, Freitas, & Reis, 2015). Most cyanobacterial strains produce extracellular polymeric substances (EPS), mainly composed of heteropolysaccharides that can remain attached to the cell surface (CPS-capsular polysaccharides) or be released into the environment (RPS-released polysaccharides) Rossi & De Philippis, 2016). The distinctive features of the cyanobacterial EPS, including their strong anionic nature, presence of sulfate groups, high variety of possible structural conformations, and amphiphilic behavior, make these polymers suitable for biotechnological and biomedical applications such as flocculating, gelifying, emulsifying, or suspending agents, rheology modifiers, therapeutic or drug delivery agents (Leite et al., 2017;. The faster growth and easier genetic manipulation of cyanobacteria compared to algae and plants, and the low-cost biomass production (owing to their photosynthetic metabolism) are additional competitive advantages for the implementation of cyanobacteria as cost-effective cell factories for the production of these biopolymers. For this purpose, a deeper knowledge on the cyanobacterial EPS biosynthetic pathways is required to optimize the production and engineer structural and compositional variants tailored for a given application.
The mechanisms involved in EPS production seem to be relatively conserved throughout bacteria, with the polymerization, assembly, and export of the polymers usually following one of three main mechanisms, namely the Wzy-, ABC transporter-, or synthase-dependent pathways (Schmid, Sieber, & Rehm, 2015).
However, in a phylum-wide analysis, we showed that most cyanobacteria harbor gene-encoding proteins related to the three pathways but often not the complete set defining a single pathway (Pereira, Mota, Vieira, Vieira, & Tamagnini, 2015), implying a more complex scenario than that observed for other bacteria.
This complexity is also evident in the physical organization of the genes, with multiple copies scattered throughout the genomes, either isolated or in small clusters (Pereira et al., 2015).
In this work, aiming at elucidating the process of EPS production in cyanobacteria, we generated an array of Synechocystis' mutants and characterized them in terms of growth, amount of EPS produced, and polymer composition. The results obtained demonstrate that Wzc (Sll0923) and Wzb (Sll0328) are involved in the production of EPS, influencing both the amount and the composition of polymer(s). The absence of both Wzc and Wzb seems to redirect RPS production toward an alternative route. We clarified the roles of both proteins through biochemical and structural analysis, providing the first insights into the molecular mechanisms of EPS production in this cyanobacterium.

| Cyanobacterial DNA extraction and recovery
Cyanobacterial genomic DNA was extracted using the Maxwell ® 16 System (Promega) except to use in Southern blot, for which the phenol/chloroform method previously described (Tamagnini, Troshina, Oxelfelt, Salema, & Lindblad, 1997) was preferred. Agarose gel electrophoresis was performed by standard protocols (Sambrook & Russell, 2001), and the DNA fragments were isolated from gels, enzymatic, or PCR reactions using the NZYGelpure purification kit (NZYTech).

| Plasmid construction for Synechocystis transformation
Plasmid pDsll0923::Km r was kindly provided by F. Chauvat (Jittawuttipoka et al., 2013). The Synechocystis chromosomal regions flanking wzy (sll0737), wzx (sll5049), kpsM (slr2107), wzb (slr0328), or the last 78 bp of wzc (sll0923) were amplified by PCR using the specific oligonucleotide primers (Supporting Information Appendix S1 and Supporting Information Table S2). An overlapping region containing an XmaI restriction site was included in primers 5I and 3I for cloning purposes. For each gene, the purified PCR fragments were fused by "overlap-PCR" (Supporting Information Appendix S1). The resulting products were purified and cloned into the vector pGEM-T ® Easy (Promega), originating pGDsll0737, pGDsll5049, pGDslr2107, pGDslr0328, and pGDsll0923 Trunc . A selection cassette containing the nptII gene (conferring resistance to neomycin and kanamycin) was amplified from pKm.1 using the primer pair Km.KmScFwd/KmRev (Pinto et al., 2015) (Supporting Information Appendix S1 and Supporting Information Table S2) and digested with XmaI (Thermo Scientific). Subsequently, the purified selection cassette was cloned in the XmaI restriction site of the plasmids using the T4 DNA ligase (Thermo Scientific) to form pGDsll0737.Km, pG-Dsll5049.Km, pGDslr2107.Km, pGDslr0328.Km, or pGDsll0923 Trunc .
Km, respectively. The cassette containing the aadA gene (conferring resistance to streptomycin and spectinomycin) was obtained by digesting the plasmid pSEVA481 (Silva-Rocha et al., 2013) with PshAI and SwaI, and the cassette was cloned in the XmaI/SmaI site of pGDslr0328 and pGDsll0727 to form plasmids pGDslr0328.Sm and pGDsll0727.Sm.
The purified product was cloned in pGEM-T ® Easy after A-tailing and subsequently digested with XbaI and SpeI. The resultant DNA fragment was cloned into pSEVA351 previously digested with XbaI and SpeI, originating plasmid pS351sll0923. To obtain plasmid pS351sll0923 Trunc, a similar procedure was used but the PCR was performed using primer pair sll0923_compF1/Sll0923.RTrunc (Supporting Information Appendix S1 and Supporting Information Table S2), and the digested fragment was cloned into the XbaI site of pSEVA351. Since wzb is part of a four gene operon (slr0326-slr0329) (Kopf et al., 2014), a different approach was used. The P rnpB promoter (Huang, Camsund, Lindblad, & Heidorn, 2010) was obtained by digesting the BioBrick vector pSB1C3 with XbaI and SpeI and subsequently cloned in pSEVA351 previously digested with the same enzymes, originating plasmid pS351P rnpB . wzb was amplified using primer pair slr0328_compF/slr0328_compR, incorporating the synthetic RBS BBa_B0030 (Supporting Information Appendix S1 and Supporting Information Table S2). The purified product was cloned in pGEM-T ® Easy after A-tailing, digested with XbaI and SpeI and subsequently cloned in the SpeI site of pS351P rnpB generating plasmid pS351slr0328.
All constructs were verified by sequencing (StabVida) before transformation of Synechocystis.

| Generation of Synechocystis mutants
Synechocystis was transformed with integrative plasmids using the procedure described previously (Williams, 1988). Briefly, Synechocystis cultures were grown until OD 730 around 0.5, cells were harvested by centrifugation and resuspended in one-tenth volume of BG11. One hundred microliter of cells were incubated with 6-20 µg/ml plasmid DNA for 5 hr before spread onto Immobilon™-NC membranes (0.45 µm pore size, Millipore) resting on solid BG11 plates at 30°C under continuous light. After 24 hr, the membranes were transferred to selective plates containing 10 µg/ml of kanamycin or 2.5 µg/ml of spectinomycin. Transformants were observed after 1-2 weeks.
For complete segregation, colonies were grown at increasing antibiotic concentrations. Nonintegrative plasmids were transformed into Synechocystis by electroporation, as previously described (Chiaramonte, Giacometti, & Bergantino, 1999;Ludwig, Heimbucher, Gregor, Czerny, & Schmetterer, 2008). In this case, cells were washed with HEPES buffer, 1 mM pH 7.5. Afterwards, cells were resuspended in 1 ml HEPES and 60 μl were mixed with 1 μg of DNA and electroporated with a Bio-Rad Gene Pulser TM , at a capacitor of 25 μF. The resistor used was 400 Ω for time constant of 9 ms with an electric field of 12 kV/cm. Immediately after the electric pulse, the cells were resuspended in 1 ml BG11 and spread onto the Immobilon™-NC membranes as described above.
After 24 hr, the membranes were transferred to selective plates containing 10 µg/ml of chloramphenicol before grown at increasing antibiotic concentrations.
RNA extraction, quantification, quality/integrity assessment were carried out as previously described (Pinto, Pacheco, Ferreira, Moradas-Ferreira, & Tamagnini, 2012). The absence of genomic DNA contamination was checked by PCR using primers for rnpB amplification and the following profile: 5 min at 95°C followed by 30 cycles of 30 s at 95°C, 30 s at 56°C, and 30 s at 72°C, and a final extension at 72°C for 7 min. After synthesis with random primers as described previously (Pinto et al., 2012), cDNAs were used as template in PCR amplifications with the oligonucleotide primers listed in Supporting Information Table S2. The PCR profile used was: 5 min at 95°C followed by 30 cycles of 40 s at 95°C, 40 s at 54°C, and 40 s at 72°C, and a final extension at 72°C for 7 min. A control PCR was performed for rnpB amplification as described above. Band intensities were estimated by ImageJ software (Schneider, Rasband, & Eliceiri, 2012).

| Growth measurements
Growth measurements were performed by monitoring the Optical Density (OD) at 730 nm (Anderson & McIntosh, 1991) using a Shimadzu UVmini-1240 (Shimadzu Corporation) and determining the chlorophyll a content as described previously (Meeks & Castenholz, 1971). Data were statistically analyzed as described below.

| Analysis of total carbohydrates, RPS and CPS
Total carbohydrates and RPS contents were determined as described previously . For the quantification of CPS, 5 ml of dialyzed cultures was centrifuged at 3,857 g for 15 min at room temperature, resuspended in water, and boiled for 15 min at 10°C to detach the CPS from the cells' surface. After new centrifugation as described above, CPS were quantified from the supernatants using the phenol-sulfuric acid method (Dubois, Gilles, Hamilton, Rebers, & Smith, 1956). Total carbohydrate, RPS, and CPS were expressed as mg per L of culture or normalized by optical density. Data were statistically analyzed as described below.
To determine the RPS' monosaccharidic composition, dialyzed cultures were centrifuged at 3,857 g for 10 min at room temperature to remove the cells and the supernatant was further centrifuged at 75,000 g for 1 hr 30 min at 15°C to remove LPS prior to lyophilization. 2-5 mg of isolated RPS was hydrolyzed with 1 ml of 2 M trifluoroacetic acid (TFA) at 120ºC for 1 hr. Samples were analyzed by ion exchange chromatography using a Dionex ICS-2500 ion chromatograph with an ED50 pulsed amperometric detector using a gold working electrode (Dionex) and a Carbopac PA1 column (Dionex).
The eluents used were (A) MilliQ-grade water, (B) 0.185 M sodium hydroxide solution, and (C) 0.488 M sodium acetate solution. The gradient consisted of a first stage with 84% solution A, 15% solution B, and 1% solution C (for 7 min); a second stage with 50% solution B and 50% solution C (for 9 min); and a final stage with 84% solution A, 15% solution B, and 1% solution C (for 14 min). The flow rate was 1 ml/min.

| LPS extraction and analysis
LPS extraction was performed according to (Simkovsky et al., 2012) with some modifications. Briefly, 25 ml of samples was collected by centrifugation at 3,857 g for 15 min at room temperature, washed once in BG11, and incubated on ice for 30 min in stripping buffer (sucrose 15% (w/v), Tris-HCl 50 mM, and Na 2 EDTA 25 mM). After centrifugation at 3,802 g for 10 min at 4°C, the supernatant was transferred to a new tube and centrifuged at 16,170 g for 2 min at
Negative staining was performed on cells mounted on formvar/carbon film-coated mesh nickel grids (Electron Microscopy Sciences) with 1% Ruthenium Red.
After centrifugation at 35,000 g for 30 min at 4°C, His6-Wzb was purified using HisTrap affinity columns (GE Healthcare). Samples were loaded in Wzb-binding buffer (50 mM HEPES pH 8.0, 300 mM NaCl, 20 mM imidazole, 0.5% Triton X-100), and bound proteins were eluted using a step gradient in which imidazole increased up to 500 mM. Samples were concentrated and diafiltered with HEPES 50 mM pH 8.0 or further purified by size exclusion chromatography (SEC), using a Superpose12 10/300 column (GE healthcare) and Wzb SEC buffer (50 mM HEPES pH 8.0 and 100 mM NaCl).
For His6-Wzc, cultures were subjected to a 30-min cold shock before induction with 0.5 mM IPTG and incubation at 20°C overnight. Cells were disrupted using a French Press (Thermo Electron Corporation) at 30 Kpsi in Wzc lysis buffer (50 mM HEPES pH 8.0, 100 mM NaCl, 0.2 mg/ml lysozyme, 10 µg/ml DNase, 1 mM MgCl 2, and 1 mM PMSF). After centrifugation at 30,000 g for 30 min at 4°C to remove cell debris, the membrane fraction was collected at 200,000 g for 1 hr 10 min at 4ºC and resuspended in Wzc buffer A (50 mM HEPES pH 8.0, 100 mM NaCl). Triton X-100 was added to a final concentration of 10%, and samples were incubated for 1 hr at 4ºC with orbital shaking to maximize protein solubilization. His6-Wzc was purified as described above, except that the composition of the Wzc binding buffer was 50 mM HEPES pH 8.0, 0.10 M NaCl, 20 mM imidazole, and 0.2% Triton X-100. Pooled fractions containing His6-Wzc were concentrated and dialyzed (Dialysis Membrane, 25 kDa MWCO,Spectra/Por) against buffer A supplemented with 1 mM n-Dodecyl-b-maltoside (DDM) ON at 4°C. Protein was further purified by SEC using a HiPrep 16/60 Sephacryl S-300 High Resolution column (GE healthcare) with Wzc SEC buffer (50 mM HEPES pH 8.0, 100 mM NaCl, 1 mM DDM, and 2.5% glycerol). The concentrations of the purified His-tagged protein solutions were determined by BCA colorimetric assay (Thermo Scientific) using bovine serum albumin as standard.
Reactions were initiated by adding 0.2 or 0.5 µg of purified His6-Wzb. The concentration of pNP formed was estimated using a molar extinction coefficient of 18,000 M −1 cm −1 (Ferreira et al., 2007;Preneta et al., 2002). Data were statistically analyzed as described below.

| Wzb crystallization, data collection, and processing
For the initial screening, His6-Wzb aliquots at 10 mg/ml were used.
The screening was performed in 24-well sitting-drop vapor diffusion plates at 20°C with several commercially available kits. A single hit was obtained with MembFac (Hampton Research) condition #22.
After optimization with solutions with varying pH and precipitant concentrations, the best diffracting crystals appeared between pH 6.2 and 7.0 and 1 M of ammonium sulfate. Crystals were subjected to a glycerol gradient up to 30% prior to being flash frozen in liquid nitrogen. A data set of a single diffracting crystal to 2.28 Å was determined at the ID23-2 beamline of the European Synchrotron Radiation Facility (λ = 0.873 Å; Grenoble, France). Diffraction images were processed with the XDS Program Package (Kabsch, 1993), and the diffraction intensities converted to structure factors in the CCP4 format (Bailey, 1994). A random 5% sample of the reflection data was flagged for R-free calculations (Brunger, 1992) during model building and refinement. A summary of the data collection and refinement statistics is presented in Table 2. Molecular replacement phases were generated with PhaserMR (McCoy et al., 2007), using as initial model the protein tyrosine phosphatase from Entamoeba histolytica (PDB entry 3IDO; (Linford et al., 2014)). The final models were obtained after further cycles of refinement and manual model building, carried out with PHENIX (Adams et al., 2010) and Coot (Emsley, Lohkamp, Scott, & Cowtan, 2010), respectively. Protein structure figures were generated with PyMol (Schrödinger, 2010).

| Phylogenetic analysis of Wzb
Synechocystis' Wzb sequence was used as query in blast searches against the PDB database (Jan 2017; https://www.rcsb.org/) (Berman et al., 2000) to retrieve homologs with 3D crystal structures available, using a significance cutoff of e-05. Sequences were aligned in MEGA6 (Tamura et al., 2013) using the ClustalW algorithm, and the phylogenetic tree was constructed in the same software by maximum likelihood, using the Jones-Taylor-Thornton (JTT) substitution model and a bootstrap of 500. A three-dimensional protein structure alignment was performed using representative LMW-PTP sequences from the eukaryote E. histolytica (PDB: 3ido; UniProt:C4LSE7), the Gram-negative bacteria E. coli (PDB: 2wja; UniProt:Q9X4B8) and the Gram-positive bacteria Staphylococcus aureus (PDB: 3rof; UniProt: P0C5D2), and the root mean square deviation (RMSD) was calculated as previously described (Krissinel & Henrick, 2004).
Membranes were then incubated with goat anti-mouse IgG-HRP (Santa Cruz Biotechnology) at a dilution of 1:5,000. Immunodetection was performed using the ECL TM Western blotting detection reagents (GE healthcare) or the WesternBright TM Quantum (Advansta) and a Chemi DocTM XRS+Imager (Bio-Rad). The relative signal intensity of the bands obtained by immunodetection was quantified using the Image Lab software (Bio-Rad). Data were statistically analyzed as described below.

| Wzc ATPase activity
The ATPase activity of His6-Wzc was determined using the ENLITEN ® ATP Assay System Bioluminescence Detection Kit for ATP Measurement (Promega). Samples were incubated in 25 mM Tris-HCl pH 7.0, 1 mM DTT, and 5 mM MgCl 2 at 30ºC for 45 min before reading luminescence using a Synergy TM 2 Multi-Mode Microplate Reader and Gen5 TM software (BioTek) with an integration time of 10 s. When necessary, His6-Wzc and His6-Wzc were inactivated by incubation at 95ºC for 5 min before adding ATP. Data were statistically analyzed as described below.

| Wzc autokinase activity
The autokinase activity of His6-Wzc was evaluated using everted membrane vesicles overexpressing the protein. For that, E. coli M15 (pREP4) cells harboring pQE-30::His6-Wzc or empty pQE-30 were grown as described above. Everted membrane vesicles were prepared from the E. coli cells following the protocol previously described (Rosen, 1986) and statistical analyses were performed as described below.

| Analysis of Wzc phosphorylation
Protein bands were excised from stained gels, and samples were processed for mass spectrometry analysis as previously described (Gomes et al., 2013;Osório & Reis, 2013). Briefly, protein spots were sequentially washed with ultrapure water, 50% acetonitrile in 50 mM ammonium bicarbonate (ABC) followed by dehydration with 100% acetonitrile. Afterward, protein spots were reduced with 25 mM dithiothreitol in 50 mM ABC, at 56ºC for 20 min and alkylated with 55 mM iodoacetamide in 50 mM ABC, for 20 min at room temperature in the dark, followed by the above described washing/dehydration procedures. In-gel protein enzymatic digestion was performed using trypsin in the presence of 0.01% surfactant (Promega) for 3 hr at 37ºC. Resulting peptides were extracted from gel plugs with 2.5%

| Statistical analysis
Data were statistically analyzed in GraphPad Prism v7 (GraphPad Software) using a one-way analysis of variance (ANOVA), followed by Tukey's multiple comparisons.
Although a ∆wzc mutant strain was already described, we generated a ∆wzc in our Synechocystis strain using the construct kindly provided by Jittawuttipoka et al. (2013) to avoid phenotypic changes arising from the use of different Synechocystis substrains (Trautmann, Voß, Wilde, Al-Babili, & Hess, 2012). In all cases, the mutant strains were fully segregated (Supporting Information Figure S7) and did not display significant growth differences compared to the wild type (Figures 1 and 2), indicating that the targeted genes are not essential in standard laboratory conditions.
With the exception of wzc Trunc , the chlorophyll a content showed a linear correlation with OD values, proving to be a good estimation of cell density/number (Figures 1 and 2). The amount of total carbohydrates, RPS, and CPS produced per liter of culture was measured and normalized by the OD value and chlorophyll a content.
These two approaches lead to congruent results and; therefore, only the production per OD value is shown (Figures 1c,d, Data are means ± SD. Statistical analysis performed using one-way analysis of variance (ANOVA), followed by Tukey's multiple comparisons, is presented for the last time point. Significant differences are identified: *(p ≤ 0.05), **(p ≤ 0.01), ***(p ≤ 0.001). Ns: no significant differences as the differences accumulate with the increase of the cell density of the cultures. The four mutants showed different amounts of CPS and/or RPS compared to the wild type, even if none of the mutations abolished CPS and/or RPS production. ∆wzc showed an approximately 20% and 17% decrease in the amount of RPS ( Figure 1d) and CPS (Figure 1e), respectively, confirming the role of Wzc in the production of these polymers (Jittawuttipoka et al., 2013). These authors reported a higher decrease in CPS and RPS production (about 50%). The difference to the amounts reported here is likely related to the Synechocystis genetic background used to generate the mutants (Trautmann et al., 2012), the growth conditions (e.g. cultivation of cells in medium with Na 2 CO 3 by Jittawuttipoka et al., 2013), and the experimental design (e.g. time course of the experiment and normalization of the data by amount of protein by Jittawuttipoka et al., 2013). ∆wzb exhibited 35% less RPS (Figure 2d) but no significant changes in the amount of CPS could be observed (Figure 2e), indicating that Wzb is only involved in RPS production. The double mutant ∆wzc∆wzb exhibited 18% decrease in CPS ( Figure 2e) and a 23% increase of RPS (Figure 2d).
The decrease of CPS displayed by the double mutant and the single mutant ∆wzc is consistent with the hypothesis that Wzb does not play a role in the regulation of CPS production. In contrast, the higher amount of RPS produced by the double mutant suggests that, in the absence of both proteins, the RPS production is likely to be diverted to an alternative route. Interestingly, the wzc Trunc mutant produced the same amount of RPS and 19% more CPS compared to the wild type (Figure 1d,e), suggesting that the C-terminal Y-rich region of this protein is only functionally relevant in the regulation of CPS production. The apparently con- It has been previously shown for other bacteria that the assembly of EPS and other surface polysaccharides, such as the O-antigen of LPS and S-layer glycans, follows similar mechanisms that may be functionally connected (Babu et al., 2011;Ristl et al., 2011;Simkovsky et al., 2012;Whitfield & Trent, 2014). However, for Synechocystis, no obvious differences were observed for the LPS profile and the cells' surface ultrastructure comparing the wild type and the mutants (Supporting Information Figures S8 and S9). The monosaccharidic composition of the RPS produced by the different strains was also evaluated. A considerable increase in the percentage of rhamnose was observed in all mutants (Table 1). In addition, the RPS produced by ∆wzb are composed by a smaller number of different monosaccharides. These results highlight that the differences induced by these mutations are not limited to the amount of the polymer but extend to their composition.

| Wzb has a classical LMW-PTP conformation
To better define the functional role of Synechocystis' Wzb in the production of EPS, we overexpressed this protein in E. coli and performed its purification. The presence of recombinant protein was confirmed by SDS-PAGE/Western blot (Supporting Information Figure S10).
The Wzb protein was crystallized and analyzed through X-ray diffraction. Data collection and refinement statistics are shown in Table 2. One monomer was present in the asymmetric unit, with the crystal lattice belonging to the P3 1 2 1 space group. Synechocystis' Wzb displays the structure of a low molecular weight protein tyrosine phosphatase (LMW-PTP) (Figure 3). The PTPases signature motif, the PTP loop, with the characteristic sequence C(X) 5 R(S/T) (cysteine 7 to serine 14) is located between the C-terminus of the β1 strand and the N-terminus of the α1 helix, at the N-terminus side of the protein (Zhang, 1998). As in other LMW-PTP proteins, the PTP loop is preceded by a valine residue (V6) (Kolmodin & Aqvist, 2001;Mukhopadhyay & Kennelly, 2011). Opposite and topping this loop, there is the DPYY loop, which contains the functionally essential aspartate residue (D124). Together, the DPYY and PTP loops constitute the enzyme´s active site (Standish & Morona, 2014;Zhang, 1998). The tyrosine residues of the DPYY loop form one of the walls of the active site pocket, with a histidine (H45) delineating the other side of the substrate entry site.
A sequence alignment was performed, and a phylogenetic tree was constructed to assess sequence similarities and conservation of residues with other LMW-PTP with solved structures.
Synechocystis' Wzb was shown to be more closely related to the TA B L E 1 Monosaccharidic composition of the released polysaccharides extracted from Synechocystis sp. PCC 6803 wild type, Δwzc, Δwzb, ∆wzc∆wzb, and wzc Trunc LMW-PTP of the unicellular eukaryote E. histolytica than of the LMW-PTPs of other bacteria (Supporting Information Figure S11).

| Wzc is a substrate of the Wzb phosphatase
Our phenotypic characterization of the Δwzb and wzc Trunc mutants raised the possibility that Wzb does not dephosphorylate Wzc. To solve this question, we first used a p-nitrophenyl phosphate hydrolysis assay to confirm that Wzb functions as a phosphatase (Figure 4), in agreement with our structural analysis ( Figure 3) and previous data (Mukhopadhyay & Kennelly, 2011). We then asked whether Synechocystis' Wzc is a substrate for Wzb. Importantly, mass spectrometry analysis of purified Wzc expressed in E. coli and solubilized with detergent revealed the presence of phosphorylated tyrosine residues. In particular, the C-terminal Y-rich peptide-740 YYNNRYYDR 748 -shows the presence of a phosphorylation in Y745 and a double phosphorylation at either the Y741 + Y746 pair or at Y745 + Y746 (Supporting Information Figure S12). We used this characteristic of Wzc and incubated the protein with Wzb, assessing the membrane protein phosphorylation load at different time points. In the presence of Wzb, a time-dependent dephosphorylation of Wzc was observed, with the signal of the anti-phophotyrosine antibody decreasing significantly after 2-hr incubation ( Figure 5).
By MS/MS analysis, we determined that Y745 is one of the residues that undergoes dephosphorylation, with the signal decreasing significantly after incubation of Wzb. In contrast, no Wzc dephosphorylation occurred in the absence of Wzb ( Figure 5).These results clearly establish that Wzb is able to interact in vitro with the C-terminal Yrich tail of Wzc, suggesting that in the cell, the phosphorylation state of Wzc is dependent on the activity of Wzb.

| Wzc has ATPase and autokinase activity
It was previously suggested that Wzc is a BY-kinase (Mijakovic et al., 2016;Pereira et al., 2013Pereira et al., , 2015. To demonstrate that this protein functions as a kinase, we investigated its ability to hydrolyze TA B L E 2 Crystallographic data collection, processing, and structure refinement statistics for Wzb Allowed regions (%) 6.33 Outliers ( ATP. ATPase activity was evaluated using a bioluminescence assay in which the signal intensity is proportional to the ATP concentration. A significant reduction of the signal intensity was observed for the reactions containing either Wzc or both Wzc and Wzb, compared to those in which Wzc was previously heat inactivated ( Figure 6a).
These results provide evidence that Wzc is capable of ATP binding and hydrolysis and that the presence of Wzb does not significantly impair this activity. To determine whether Wzc also displays autokinase activity, we tested its ability to autophosphorylate in the presence of ATP. For this, we used everted membrane vesicles (EMVs) of E. coli overexpressing Synechocystis' Wzc, which were pre-incubated with Wzb to generate dephosphorylated Wzc. To take into account possible differences in the amount of protein loaded, the signal obtained for the anti-phosphotyrosine antibody was normalized by that obtained for the 6x-His epitope tag antibody. The results revealed an ATP-dependent increase in the level of phosphorylated Y residues of Wzc over time (Figure 6b). To confirm that the immunoblot signal was due to the Synechocystis' Wzc and not the native E. coli Wzc,

EMVs isolated from cells transformed with the empty pQE-30 vector
were also included in the assay. Altogether, the ATPase activity and the increase in phosphorylation strongly suggest autokinase activity.

| D ISCUSS I ON
The results obtained in this work clearly show that Synechocystis' Wzc and Wzb are involved in the production of EPS. Importantly, the amount of total carbohydrates did not vary in ∆wzc and ∆wzb, consistent with an intracellular accumulation of the polysaccharides and supporting the idea that these proteins are mainly involved in the later steps of polymer assembly and export. Neither Wzc nor Wzb are essential for EPS production, as the polymer(s) is still produced in their absence. This is consistent with the hypothesis of functional redundancy, either owing to the existence of multiple copies for some of the EPS-related genes/proteins and/or a crosstalk between the components of the different assembly and export pathways (Pereira et al., 2015). This hypothesis also helps to explain the lack of phenotypes for ∆wzy (∆sll0737), ∆wzx (∆sll5049), ∆kpsM (∆slr2107), and ∆kpsM∆wzy (∆slr2107∆sll0737).
We have established that the Wzb phosphatase has a classical LMW-PTPs conformation and that Wzc has the functional properties of a BY-kinase, showing ATPase and autophosphorylation activity. We also provide evidence that Wzc is a substrate of Wzb and that the Wzb-mediated dephosphorylation occurs at the C-terminal Y-rich loop of Wzc. These findings fit well with the previously proposed roles for the two proteins and strongly indicate that tyrosine phosphorylation of Wzc and its dephosphorylation by Wzb is a mechanism of regulation of EPS production. Two features of the Wzb structure are particularly relevant for its functional properties: (a) it possesses the second Y of the DPYY motif (Y127) that is absent in most prokaryotic LMW-PTPs (Bohmer, Szedlacsek, Tabernero, Ostman, & Hertog, 2013;Lescop et al., 2006;Standish & Morona, 2014). While the first tyrosine usually interacts with the aromatic ring of the substrate, being crucial for the interaction between LMW-PTPs and the BY-kinases (Lescop et al., 2006;Standish & Morona, 2014), the second tyrosine seems to play a role in substrate specificity (Bucciantini et al., 1999;Raugei, Ramponi, & Chiarugi, 2002). (b) Synechocystis' Wzb possesses an aromatic residue-histidine (H45)-in the catalytic site instead of the hydrophobic leucine that occupies this position in most prokaryotic LMW-PTPs (Lescop et al., 2006;Linford et al., 2014). Based on these characteristics, Synechocystis Wzb can be classified as a class I or eukaryotic-like LMW-PTP and is therefore likely to present a substrate recognition mechanism different from that of E. coli Wzb, a class II or prokaryotic-like LMW-PTPs (Lescop et al., 2006). These differences may contribute to the higher substrate affinity of the F I G U R E 4 His6-Wzb phosphatase activity, expressed in µmoles per minute of p-nitrophenol (pNP), using two different amounts of protein and 10 mM of p-nitrophenyl phosphate (pNPP) substrate at pH 6.5. Experiments were performed in triplicate.
Data are mean ± SD. Statistical analysis was performed using oneway analysis of variance (ANOVA), followed by Tukey's multiple comparisons. Statistically significant differences are identified: **(p ≤ 0.01) and ***(p ≤ 0.001). Ns: no significant differences F I G U R E 5 In vitro dephosphorylation of His6-Wzc by His6-Wzb for 0-24 hr. Samples were analyzed by Western blot using an anti-phosphotyrosine antibody. Subsequently, membranes were stripped and reprobed with a 6x-His epitope tag antibody (loading control). Lanes 24 hr (-His6-Wzb) refer to samples in which His6-Wzc was incubated in the absence of His6-Wzb (control). Experiments were performed in triplicate. Statistical analysis was performed using one-way analysis of variance (ANOVA), followed by Tukey's multiple comparisons. Significant intensity differences between 0 hr and other time points are identified: *(p ≤ 0.05), **(p ≤ 0.01), ***(p < 0.001). Ns: no significant differences Synechocystis' Wzb compared to the E. coli homolog using pNPP as substrate (Lescop et al., 2006;Mukhopadhyay & Kennelly, 2011) and to the ability of Synechocystis' Wzb to dephosphorylate other substrates (Mukhopadhyay & Kennelly, 2011). In addition, since most of the phosphatases involved in the biosynthesis of capsules and/or EPS are class II LMW-PTPs, it is likely that the structural features of Synechocystis' Wzb reflect differences in EPS production.
Strikingly, although our results indicate that Wzb regulates the function of Wzc through dephosphorylation, the two proteins appear to have different roles in EPS production. Wzc plays a role in both CPS and RPS production, as shown here and previously (Jittawuttipoka et al., 2013), while Wzb participates in RPS production only. Our wzc Trunc mutant, which lacks the protein region where Wzb acts, provides some insights into these differences. The phenotype of wzc Trunc , together with the results obtained for ∆wzc and ∆wzb, supports previous indications that the production of CPS and RPS is, at least, divergent processes (Micheletti et al., 2008), relying on different proteins and/or different contributions of the same proteins (Jittawuttipoka et al., 2013). For example, it is possible that the assembly and export of CPS and RPS are regulated by different signals involving the Y-rich region of Wzc. This is true for E. coli K30 capsule (CPS-like) and K-12 colanic acid (RPS-like), with the production of the capsule depending on Wzc undergoing cycles of phosphorylation/dephosphorylation while the production of colanic acid being negatively regulated by phosphorylated Wzc (Paiment, Hocking, & Whitfield, 2002). However, it is important to keep in mind that the C-terminal Y-rich region is not universally present in the Wzc proteins of EPS-producing bacteria, including Xanthomonas campestris' GumC (Cuthbertson et al., 2009) and the Wzc homologs of the highly efficient EPS producer Cyanothece sp. CCY 0110 (Pereira et al., 2013).
In addition, our results show that Wzc and Wzb influence EPS composition. Three factors are likely to contribute to the monosaccharidic differences observed for the RPS of the mutants versus wild type: (a) Synechocystis' Wzc and Wzb interact with other proteins that may directly or indirectly affect monosaccharide and/or sugar nucleotide metabolism (Mukhopadhyay & Kennelly, 2011;Sato et al., 2007). In particular, Wzb substrates include proteins involved in the photosynthetic process (Mukhopadhyay & Kennelly, 2011) and changes in Wzb activity will likely affect central carbon metabolism. It is therefore not surprising that the RPS of ∆wzb displays the highest number of differences relative to wild type composition; (b) mutations may lead to differences in the affinity of the periplasm-spanning complex for monosaccharides. This explanation has been suggested for a mutant in the ATP-binding component (KpsT; Sll0982) of an EPS-related ABC transporter that also produces EPS enriched in rhamnose (Fisher et al., 2013). (c) The increase in the F I G U R E 6 In vitro evaluation of His6-Wzc ATPase (a) and autokinase (b) activities. (a) The ATPase activity was measured using the luciferase assay, being the intensity of the emitted light (Relative Luminescence Unit-RLU) proportional to the ATP concentration in the samples. The presence/absence of ATP, active His6-Wzc, His6-Wzc inactivated by heat, His6-Wzb and/ or Wzb inactivated by heat is indicated below. (b) To evaluate the autokinase activity, everted membrane vesicles of Escherichia coli cells containing pQE-30::His6-Wzc or pQE-30 were isolated, dephosphorylated for 0 or 6 hr with His6-Wzb, and dialyzed to remove the dephosphorylation buffer and most of the phosphatase, before incubation with or without 200 μM ATP for 0 hr, 10 min, 30 min, or 6 hr. The time of dephosphorylation and the presence/ absence ATP is indicated below. Samples were analyzed by Western blot using an anti-phosphotyrosine antibody. Subsequently, membranes were stripped and reprobed with a 6x-His epitope tag antibody (loading control). The ratio between the two signals (adjusted volume) is shown. Experiments were performed in triplicate. Data are means ± SD. Statistical analysis was performed using one-way analysis of variance (ANOVA), followed by Tukey's multiple comparisons. Statistically significant differences between inactivated Wzc in the presence of ATP (a) or dephosphorylation for 6 hr followed by incubation with ATP for 0 hr (b) and other conditions are identified: *(p ≤ 0.05), **(p ≤ 0.01), ***(p ≤ 0.001), ****(p ≤ 0.0001). Ns: no significant differences rhamnose content may be related to the genomic context of wzc, since this gene is present in the vicinity of slr0985 that encodes one of the enzymes involved in the dTDP-l-rhamnose biosynthesis, the dTDP-4-dehydrorhamnose 3,5-epimerase. Interestingly, the same is observed for kpsT (sll0982) and its cognate permease component KpsM (sll0977). In bacteria, l-rhamnose is a rare sugar that is most frequently found in the EPS and O-antigen of LPS and this genetic organization is not uncommon (Boels et al., 2004) (Giraud & Naismith, 2000;Roca et al., 2015). Further studies are necessary to discern the mechanistic role of Wzb and Wzc on the composition of EPS.
Overall, the results obtained in this study emphasize the complexity of EPS production in cyanobacteria, supporting previous hypothesis of functional redundancy and crosstalk between different pathways, and provide novel data on the specific function of two proteins-Wzc and Wzb-in Synechocystis. These findings provide a robust basis for future studies aiming at further elucidate cyanobacterial EPS production with the ultimate goal of establishing these organisms as sustainable platforms for the production of large amounts of these polymers and/or polymers with the desired characteristics for the industrial toolbox.

ACK N OWLED G M ENTS
This work was financed by FEDER-Fundo Europeu de Desenvolvimento Regional funds through the COMPETE 2020-Operational Programme for Competitiveness and Internationalisation

CO N FLI C T O F I NTE R E S T
The authors declare that they have no conflict of interest.

E TH I C S S TATEM ENT
None required.

DATA ACCE SS I B I LIT Y
The atomic coordinates of Wzb have been deposited in the Protein Data Bank, www.pdb.org (PDB ID 5O7B). All data supporting this study are provided as Supporting Information Appendix S1 accompanying this paper.