Biosynthesis system of Synechan, a sulfated exopolysaccharide, in the model cyanobacterium Synechocystis sp. PCC 6803

Extracellular polysaccharides of bacteria contribute to biofilm formation, stress tolerance, and infectivity. Cyanobacteria, the oxygenic photoautotrophic bacteria, uniquely and widely have sulfated extracellular polysaccharides and they may utilize the polysaccharides for survival in nature. In addition, sulfated polysaccharides of cyanobacteria and other organisms have been focused as beneficial biomaterial. However, very little is known about their biosynthesis machinery and function in cyanobacteria. Here we found that the model cyanobacterium, Synechocystis sp. PCC 6803, formed bloom-like cell aggregates using sulfated extracellular polysaccharides (designated as synechan) and identified whole set of genes responsible for synechan biosynthesis and its transcriptional regulation, thereby suggesting a model for the synechan biosynthesis apparatus. Because similar genes are found in many cyanobacterial genomes with wide variation, our findings may lead elucidation of various sulfated polysaccharides, their functions, and their potential application in biotechnology.


Introduction 23
Bacterial extracellular polysaccharides establish biofilms for nutrient supply and stress 24 avoidance, and they sometimes support cellular activities such as motility and infectivity (Woodward 25 and Naismith, 2016). Generally, the polysaccharide chains consist of a few types of sugars (with or 26 without chemical modifications) and are anchored on cells (capsular polysaccharides, CPS) or exist as 27 nonanchored exopolysaccharides (EPS). Nonetheless, their molecular structures vary greatly, e.g., 28 branching schemes, sugar constituents, and modifications, and thus their physical properties also vary. 29 Bacterial extracellular polysaccharides and lipopolysaccharides are produced and exported via three 30 distinct pathways: the Wzx/Wzy-dependent pathway, ABC-dependent pathway, and synthase-dependent 31 pathway (Schmid et al., 2015). Every bacterium can produce several extracellular polysaccharides, and 32 production often depends on environmental conditions. Some extracellular polysaccharides have been 33 appropriated for use as biopolymers for food, cosmetics, medicine (Freitas et al., 2014, Lapasin and 34  Cells and CPS were removed from the culture by centrifugation, and EPS in the supernatant was 111 separated from "free" polysaccharide (PS) by membrane filtration followed by a second centrifugation 112 to remove residual cells. CPS was collected from the cell pellet after vortexing and centrifugation. G, 113 Sugar content of fractions from WT and Δslr5054. Error bars represent SD (n = 3, **P < 0.005). 114 Figure S1. Isolation of crude EPS and sugar analysis. formation by mutants after the second step of culture. 128 similar in the WT and ∆slr5054 (Fig. 1G). Then we performed Alcian blue staining to examine the 130 acidity of the EPS (Fig. S3). Generally, sulfated polysaccharides are stained at pH 0.5 condition, while 131 acidic polysaccharides, which contain sulfate groups and/or carboxylate groups (such as uronic acids 132 and carboxylate modification) are stained at pH 2.5 condition (Bellezza et al., 2006). The EPS from WT 133 was clearly stained under both pH conditions, strongly suggestive of the sulfate modification. 134 135 Gene cluster for the biosynthesis of viscous polysaccharides 136 slr5054 resides on a megaplasmid, pSYSM, in a large gene cluster (sll5042-60), which we 137 named xss (extracellular sulfated polysaccharide biosynthesis) ( Fig. 2A,   The microscopy images of isolated EPS from WT culture (Fig. 1g) stained with alcian blue at pH 2.5 145 (left) and pH 0.5 (right). 146 Figure 2 The xss gene cluster and phenotype of xss mutants. 149 A, The xss gene cluster. Red, polysaccharide biosynthesis genes; green, regulatory genes; black, genes 150 of unknown function. B, Bloom formation by the mutants carrying disruptions in the polysaccharide 151 biosynthesis xss genes. C, Total sugar content (µg glucose per 1 × 10 8 cells) of the EPS fraction from 152 mutants in b. Red bars, bloom-forming mutants; blue bars, non-bloom-forming mutants. Error bars 153 represent SD (WT grown at 20°C, n = 6; others, n = 3). Statistical significance was determined using 154 Welch's t test (*P < 0.05, **P < 0.005, ***P < 0.0005). D, Bloom formation by regulatory mutants, WT 155 grown at 20ºC, and OPX mutant (∆sll1581). E, Total sugar content of the EPS fraction from mutants in 156 d. Red bars, bloom-forming mutants; green bars, excess-bloom-forming mutants. F, A sheet of OX-xssR 157 cells was stripped off from the agar plate by tweezers. The culture temperature was 31°C unless 158 otherwise stated. phosphorelay system (xssR, xssS), and genes encoding several small proteins of unknown function 160 (Table S1, Fig. 2A). All genes except those of unknown function were disrupted individually with a 161 read-through cassette, and segregation was confirmed by colony PCR (Fig. S4). Bloom formation and 162 sugar content of the EPS fraction were reduced in many mutants (Fig. 2B, C). In particular, bloom 163 formation was completely abolished in ∆xssA, ∆xssB, ∆xssF, ∆xssH, ∆xssK, ∆xssM, ∆xssN, and ∆xssP, 164 in which EPS accumulation was also suppressed. Certain glycosyltransferase mutants (∆xssC, ∆xssG, 165 ∆xssI, ∆xssO) formed blooms but accumulated little EPS, and neither bloom formation nor EPS 166 accumulation was substantially altered in one sulfotransferase mutant (∆xssE). In general, the Wzx/Wzy 167 system in bacteria produces various EPS, lipopolysaccharides, and CPS through four steps: (i) 168 biosynthesis of a heterooligosaccharide repeat unit on a lipid linker at the cytoplasmic side of the plasma 169 membrane by a series of glycosyltransferases and modification enzymes, (ii) flip-out of the unit to the 170 periplasmic side by Wzx, (iii) polymerization by transfer of the nascent polysaccharide chain to the 171 repeat unit by Wzy, and (iv) export of the EPS chain through the periplasm and outer membrane via PCP

Regulation of the sulfated EPS biosynthesis 191
The sensory histidine kinase mutant ∆xssS accumulated a much larger amount of EPS than 192 WT, whereas mutants of the cognate response regulator xssR and transcriptional regulator xssQ had a 193 null phenotype with regard to both bloom formation and EPS accumulation (Fig. 2D, E, Table S2). The Notably, the OX-xssR and OX-xssQ strains formed sticky, non-motile, biofilm-like colonies on agar 202 plates that could be picked by tweezers (Fig. 2F). Using real-time quantitative PCR (qPCR), we compared gene expression in the xss cluster for WT, ΔxssS, 208 and ΔxssQ (Fig. 3A). Expression of five genes (xssA, xssB, xssE, xssN, xssP) was very low in ∆xssQ 209 compared with WT, whereas that of xssF, xssH, and xssK was not substantially affected. These results 210 suggested that XssQ transcriptionally activates genes encoding sulfotransferases and certain 211 glycosyltransferases but not genes for polymerization and export via the Wzx/Wzy system. qPCR  genes down-regulated in ΔxssQ and up-regulated in ΔxssS were mostly xss genes. In detail, the regulated 233 genes were xssA-E and xssL-P, which were roughly consistent with the qPCR analysis. We conclude 234 that xssA-E and xssL-P were specifically regulated by XssS/XssR/XssQ. In a previous report, xssA-xssE 235 and xssL-xssP were up-regulated at low temperature in another substrain of Synechocystis 6803 (Kopf 236 et al., 2014b). To test this in our substrain, we measured the sulfated EPS accumulation of WT culture 237 at 20°C, and it was 3.1-fold greater than that at normal growth temperature (31°C; Fig. 2D, E and Table  238 S2). This result suggests that XssS/XssR/XssQ is a unique temperature sensor for xss gene expression. 239 We aligned nucleotide sequences near the transcription start sites of the regulated genes (xssA, 240 xssE, xssL, xssN, and xssP) to find the consensus sequences for XssQ binding ( roughly fits with the gene number, i.e., eight glycosyltransferase genes and two sulfotransferase genes. 274 We speculated that the overaccumulation of EPS in ∆xssS reflects the true product of the xss cluster. The 275 EPS from WT may contain a considerable amount of unrelated polysaccharides, which were erroneously 276 recovered together with the xss product. Here, the sulfated EPS produced by the xss cluster in 277 Synechocystis 6803 was designated "Synechan".

The OPX protein for synechan biosynthesis 280
There is no candidate gene in the xss-carrying plasmid for the OPX protein of the Wzx/Wzy 281 system, whereas sll1581, an OPX homolog, was found on the main chromosome. Disruption of sll1581 282 (Δsll1581) abolished bloom formation and EPS accumulation (Fig. 2D, E). Thus, the chromosomal OPX 283 protein Sll1581 (XssT) appears to serve as the outer-membrane exporter for synechan. Interestingly, 284   Summarizing these data, we propose models for synechan biosynthesis apparatus including 302 OPX and temperature-responsive regulation (Fig. 4A, B and Fig. S5). The model of the Xss apparatus 303 fits well with the known Wzx/Wzy-dependent apparatus represented by xanthan biosynthesis in 304 Xanthomonas campestris (Katzen et al., 1998). The eight glycosyltransferases including XssP (the 305 priming glycosyltransferase) produce oligosaccharide repeat unit of eight sugars, which is consistent 306 with the sugar composition of synechan. These findings suggest that the xss cluster on the pSYSM 307 plasmid harbors a whole set of genes for synechan biosynthesis except the OPX gene (xssT on the main 308 chromosome). Notably, the cluster harbors two sulfotransferase genes, which have not been found to 309 Sulfotransferases, XssA and XssE, belong to distinct subfamilies of bacterial sulfotransferases. We 325 found many sulfotransferase genes in various cyanobacterial genomes by Pfam search (PF00685, 326 PF03567, PF13469). They are mostly found in gene clusters for putative extracellular polysaccharide 327 biosynthesis (Wzx/Wzy-type and ABC-type) (Fig. S8). It should be noted that they are more or less 328 partial as a cluster for extracellular polysaccharide biosynthesis system, whereas the xss cluster appears XssS/XssR/XssQ is found near the gene cluster for sulfated EPS biosynthesis with sulfotransferases in many 351 cyanobacteria (Fig. S9, Table S3). Consensus sequences are also found in upstream of some genes in the 352 cluster, suggesting that the XssS/XssR/XssQ system may operate universally for induction of sulfated 353 EPS production under certain environmental conditions such as cold temperature.
Acidic polysaccharides containing uronic acids and other carboxylic groups are common in 355 bacteria, but sulfated polysaccharides are produced exclusively by cyanobacteria (Pereira et al., 2009). 356 To speculate on the physiological significance of sulfated polysaccharides, we summarized the 357 distribution of sulfotransferase genes in cyanobacteria of various habitats (Table S3)  for bloom formation or EPS production (Jittawuttipoka et al., 2013), probably because the parent strain 379 did not produce discernable amount of EPS like our nonmotile strain. Similarly, the deletion mutant of 380 sll5049 (xssH) did not show any defect in EPS or CPS accumulation, though related mutants (∆sll0923 381 for a second PCP-2a) were shown to be depleted slightly of both CPS and EPS (Pereira et al., 2019). 382 These results contrast with our null phenotype of ∆sll5052 (xssK) and ∆sll5049 (xssH), probably because 383 of the difference in the parent strains. In addition, we found that our ∆sll0923 did not show any defect 384 in the bloom formation. On the other hand, disruption of sigF (slr1564) for a sigma factor of global cell 385 surface regulation increased three to four fold accumulation of sulfated EPS (Flores et al., 2019a, Flores 386 et al., 2019b. The proteome analysis of ∆sigF revealed many (more than 160) proteins except for any 387 Xss proteins were up-regulated, leaving the sulfated EPS biosynthesis pathway elusive. The sugar 388 composition of the sulfated EPS of their WT is similar to our WT, although the composition of EPS of 389 ∆sigF was different for WT or synechan from our ∆xssS. 390 To get insights into the difference in bloom formation between the motile and nonmotile 391 substrains, we compared transcription data (Table S4). It is evident that many xss genes on the plasmid 392 are expressed several times higher in the motile substrain than the nonmotile one except for xssT on the 393 main chromosome, despite that the nucleotide sequence of the xss gene cluster was completely 394 conserved between them. This fact suggests a possibility that another mechanism besides the 395 XssS/XssR/XssQ contributes to the difference between the substrains. For example, the plasmid copy 396 number of pSYSM may be higher in the motile substrain than the nonmotile substrain. The plasmid 397 function is often affected depending on variations in the main chromosome (Vial and Hommais, 2020). 398 Anyway, it is very important to select the parent strain depending on the research purpose. 399 The cyanobacterial bloom rapidly accumulates in populations of cyanobacterial cells floating 400 on the water surface, which often produce potent cyanotoxins (hepatotoxins, neurotoxins, etc.) (Merel 401 et al., 2013). Blooms are thought to be supported mainly by cellular buoyancy due to intracellular 402  (Beard et al., 2002, Walsby, 1994. 407 Moreover, recent studies suggested that extracellular polysaccharides are also important for the bloom 408 formation (Chen et al., 2019). Some papers reported that the cells without gas vesicle can form blooms

Cyanobacterial strains and cultures 427
The motile substrain PCC-P of the unicellular cyanobacterium Synechocystis sp. PCC 6803, 428 which exhibits phototaxis (Yoshihara et al., 2000) and forms bloom-like aggregates, was used as the 429 WT in this work. A non-motile glucose-tolerant substrain, which has been widely used for studies of 430 photosynthesis, was used for comparison (Chin et al., 2018). Cells were maintained in BG11 liquid 431 medium (Stanier et al., 1971) under continuous illumination with bubbling of 1% CO2 in air at 31°C, or 432 on 1.5% agar plates. White light of 30 μmol photons m −2 s −1 was generated by fluorescent lamps. Cell 433 density was monitored at 730 nm. 434 435

Construction of plasmids and mutants 436
Primers used are listed in Table S5. Plasmids and mutants were constructed as described 437 (Chin et al., 2018). In brief, the DNA fragments, antibiotic-resistance cassettes, the trc promoter, and 438 plasmid vectors were amplified by PCR using PrimeSTAR MAX DNA polymerase (Takara, Shiga, 439 Japan) and combined using the In-Fusion System (Takara). The resulting plasmid constructs were 440 confirmed by DNA sequencing. 441 Gene disruption was performed in two different ways. One method was replacement of a 442 large portion of a targeted gene(s) with an antibiotic resistance cassette. The other method was 443 replacement of the translation initiation codon with a stop codon. In both cases, the screening cassette 444 without the terminator was inserted in the direction of the targeted gene(s) to allow transcriptional 445 readthrough of the downstream gene(s). For overexpression, gene expression was constitutively driven 446 by the strong trc promoter in two ways: integration of a target gene with the strong trc promoter into a 447 neutral site near slr0846 or IS203c, or replacement of the target-gene promoter with the trc promoter. 448 Natural transformation and subsequent homologous recombination were performed as described (Chin 449 et al., 2018). The antibiotic concentration for the selection of transformants was 20 μg·mL -1 450 chloramphenicol, 20 μg·mL -1 kanamycin, and/or 20 μg·mL -1 spectinomycin. Complete segregation of

EPS fractionation 477
The fractionation method to isolate the crude EPS is shown in Fig. S1A. The viscous 478 materials including cells after the second step of culture were collected by filtration using a 1.0-μm pore 479 PTFE membrane (Millipore). The trapped materials were gently and carefully recovered from the 480 membrane using MilliQ water with the aid of flat-tip tweezers. The collected sample was vortexed and 481 then centrifuged at 20,000 × g for 10 min to remove cells. The supernatant constituted the crude EPS 482 that contained viscous EPS and possibly CPS. 483 The refined fractionation method to isolate EPS is shown in Fig. 1F. The entire culture at the 484 end of the first step, which did not contain gas bubbles, was first centrifuged at 10,000 × g for 10 min 485 to remove cells and CPS and then filtered through a 1.0-μm pore PTFE membrane. The trapped EPS 486 was carefully recovered as described above. The flowthrough of the filtration was regarded as free 487 polysaccharides, which were recovered by ethanol precipitation. CPS was released from the cell pellet 488 by vigorous vortexing with MilliQ water and recovered by centrifugation to remove cells (20,000 × g 489 for 10 min). 490 491

Sugar quantification 492
Total sugar was quantified using the phenol-sulfate method (DuBois et al., 1956). A 100-μL 493 aliquot of 5% (w/w) phenol was added to 100 μL of a sample in a glass tube and vortexed three times 494 for 10 s. Then, 500 μL of concentrated sulfuric acid was added, and the tube was immediately vortexed 495 three times for 10 s and then kept at 30°C for 30 min in a water bath. Sugar content was measured by 496 absorption at 487 nm using a UV-2600PC spectrophotometer (Shimadzu, Japan, Tokyo). Any 497 contamination of the BG11 medium was evident by slight background coloration. This background was 498 subtracted on the basis of the extrapolation of absorption at 430 nm, where the coloration due to sugars 499 was minimal. Glucose was used as the standard. Some EPS samples were highly viscous, so we vortexed 500 and sonicated them before measurement. Statistical significance was determined using Welch's t test. 501

Sugar composition analysis 503
The collected EPS samples were dialyzed with MilliQ water and then freeze-dried for 3 days. For neutral sugars, the column was TSK-gel Sugar AXG (TOSOH, Japan) and the temperature was 70°C. 511 The mobile phase was 0.5 M potassium borate (pH 8.7) at 0.4 mL/min. Post-column labelling was 512 performed using 1% (w/v) arginine and 3% (w/v) boric acid at 0.5 mL/min, 150°C. For uronic acids, the 513 column was a Shimpack ISA-07 (Shimadzu) and the temperature was 70°C. The mobile phase was 1.0 514 M potassium borate (pH 8.7) at 0.8 mL/min. Post-column labelling was performed using 1% (w/v) 515 arginine and 3% (w/v) boric acid at 0.8 mL/min, 150°C. The detector was a RF-10AXL (Shimadzu), with 516 excitation at 320 nm and emission at 430 nm. The standard curves were prepared for each 517 monosaccharide with standard samples. 518 The SO4 2− content was determined by anion exchange column chromatography using the 519 ISC-2100 system (Thermo Fisher Scientific, USA, Massachusetts). The column was eluted via a 520 gradient of 0-1.0 M KOH. The separation column was IonPac ASI l-HC-4 μm (Thermo Fisher 521 Scientific). Electric conductivity was used for detection. 522 523

Alcian blue staining 524
The polysaccharides were stained with 1% Alcian blue 8GX (Merck) for 10 min in 3 % acetic 525 acid (pH 2.5) or 0.5 N HCl (pH 0.5) as previously described (Di Pippo et al., 2013). 526 qPCR 528 The qPCR was performed as described in our previous work (Maeda et al., 2018). Cells were 529 harvested by centrifugation at 5000 × g for 10 min at 4°C. Cell disruption and RNA extraction were 530 done using an RNeasy Mini kit for bacteria (Qiagen, Venlo, Netherlands). In addition, cells were 531 disrupted five times by mechanical homogenization with zirconia beads (0.1-mm diameter) in a 532 microhomogenizing system (Micro Smash MS-100, TOMY SEIKO, Tokyo, Japan) at 5,000 rpm for 40 533 s. For cDNA preparation, RNA was reverse-transcribed using random primers (PrimeScript RT reagent 534 kit with gDNA eraser, Takara). Real-time PCR was performed using the THUNDERBIRD SYBR qPCR 535 Mix (Toyobo) and the Thermal Cycler Dice Real Time System II (Takara). The transcript level in each 536 strain was normalized to the internal control (rnpB). The primers used are listed in Table S5. 537 538 EMSA (electrophoretic mobility shift assay) 539 The expression and purification of recombinant His-tagged proteins and EMSA were 540 performed as described in our previous works (Hirose et al., 2010, Maeda et al., 2014. In brief, His-541 tagged XssQ was expressed using pET28a vector system and E. coli C41(DE3) strain. The protein was 542 purified by Histrap HP column (Cytiva, Tokyo, Japan) and AKTA prime system (Cytiva). For probe and 543 native competitor, the upstream region of xssE was amplified with the primer set xssEup-1F/2R (total 544 251 bp). As a mutant competitor, the same region of the chemically synthesized DNA fragment 545 containing mutations in the two consensus sequences was used for amplification with the same primer 546 set as mention above. Labelling of the DNA probe, electrophoresis, and autoradiography were 547 performed as described (Midorikawa et al., 2009). We incubated the aliquots of the XssQ protein (0, 548 500, 1000, or 3000 ng/lane) with the radiolabeled probe for 30 min at room temperature. For competition, 549 3000 ng of XssQ was incubated with the probe and 20 pmol of unlabeled competitors (native or mutant). 550

RNA-seq analysis 552
RNA-seq analysis was performed as described in our previous work (Ohbayashi et al., 2016). 553 Total RNA of Synechocystis 6803 was extracted as described in the qPCR protocol. The contaminated 554 genome DNA was removed by TURBO DNA-free™ Kit (Thermo Fisher Scientific). 555 556