A polysaccharide deacetylase enhances bacterial adhesion in high ionic strength environments

The adhesion of organisms to surfaces in aquatic environments provides a diversity of benefits such as better access to nutrients or protection from the elements or from predation. Differences in ionic strength, pH, temperature, shear forces, and other environmental factors impact adhesion and organisms have evolved various strategies to optimize their adhesins for their specific environmental conditions. We know essentially nothing about how bacteria evolved their adhesive mechanisms to attach efficiently in environments with different physico-chemical conditions. Many species of Alphaproteobacteria, including members of the order Caulobacterales, use a polar adhesin, called holdfast, for surface attachment and subsequent biofilm formation in both freshwater and marine environments. Hirschia baltica, a marine member of Caulobacterales, produces a holdfast adhesin that tolerates a drastically higher ionic strength than the holdfast produced by its freshwater relative, Caulobacter crescentus. In this work, we show that the holdfast polysaccharide deacetylase HfsH plays an important role in adherence in high ionic strength environments. We show that deletion of hfsH in H. baltica disrupts holdfast binding properties and structure. Increasing expression of HfsH in C. crescentus improved holdfast binding in high salinity, whereas lowering HfsH expression in H. baltica reduced holdfast binding at high ionic strength. We conclude that HfsH plays a role in modulating holdfast binding at high ionic strength and hypothesize that this modulation occurs through varied deacetylation of holdfast polysaccharides.

biofilm formation in both freshwater and marine environments. Hirschia baltica, a marine member 23 of Caulobacterales, produces a holdfast adhesin that tolerates a drastically higher ionic strength 24 than the holdfast produced by its freshwater relative, Caulobacter crescentus. In this work, we The development of adhesives that perform well on wet surfaces has been a challenge for 35 centuries, yet this problem has been solved multiple times during the evolution of sessile aquatic 36 organisms. These organisms derive multiple benefits from their adhesion to surfaces in aquatic 37 environments such as increased access to nutrients, aerated water, and protection from 38 predation. Aquatic environments can differ in ionic strength, pH, temperature, and shear forces, 39 requiring the evolution of environment-optimized adhesion strategies. For example, mussels, a 40 diverse group of bivalve mollusk species, can attach to surfaces in freshwater, brackish waters, 41 and marine habitats, suggesting a successful evolution of adhesion mechanisms adapted to 42 different ionic environments. 1, 2 Both marine and freshwater mussels produce a fibrous polymeric 43 adhesin structure called the byssus for surface attachment. 1, 2 Mussel byssus-mediated adhesion 44 is one of the best characterized systems for how adhesins interact with wet surfaces in both low 45 and high ionic strength environments. 2, 3

46
The byssus adhesin contains more than 15 mussel foot proteins (Mfps). 3 Mfp-3 and Mfp-47 5 contain several 3,4-dihydroxyphenyl-L-alanine (DOPA) residues, a post-translational 48 hydroxylation of tyrosine that promotes byssal adhesiveness and cohesiveness through hydrogen 49 bonding and oxidative cross-linking. [2][3][4] Mfp-3 and Mfp-5 are also rich in lysine residues, which are 50 frequently adjacent to DOPA residues on the protein backbone. 1, 3 Marine mussel Mfps have more 51 DOPA residues than freshwater species (11 -30% mol in marine vs 0.1 -0.6% mol in freshwater), 52 which is hypothesized to contribute to overcoming the binding inhibition posed by high ionic 53 strength in marine waters. [5][6][7][8] The synergy between the DOPA and lysine residues is thought to 54 improve adhesion in marine environments by displacing hydrated salt ions from the surface and 55 increasing electrostatic interactions. 1-3 Despite the impressive progress in understanding the 56 mechanistic basis for mussel adhesion in different ionic strength environments, the lack of a 57 genetic system has made it difficult to study the evolution of those mechanisms.

120
The holdfast polysaccharide deacetylase HfsH is required for adhesion and biofilm 121 formation in H. baltica 122 A putative acetyltransferase HfsK, and a polysaccharide deacetylase, HfsH modulate C. 123 crescentus holdfast binding properties. 26,27 In C. crescentus, deacetylation of holdfast 124 polysaccharides is important for both the cohesiveness and adhesiveness of holdfast. 27 In our 125 previous work comparing H. baltica and C. crescentus holdfasts, 25 we showed that both species 126 use similar genes to synthesize, export, and anchor holdfast to the cell envelope. We identified 127 H. baltica genes that modify holdfast in C. crescentus, namely the putative acetyltransferase hfsK 128 (hbal_0069) and the polysaccharide deacetylase hfsH (hbal_1965; Fig. 2A). In C. crescentus, the 129 hfsK gene as well as its paralogs CC_2277 and CC_1244 are found outside the core hfs locus 130 ( Fig. 2A). Similar to C. crescentus, the H. baltica hfsK gene and its paralogs hbal_1607 and 131 hbal_1184 are also found outside the hfs locus ( Fig. 2A). BLAST analysis did not identify any 132 additional hfsK paralogs in H. baltica.

133
We showed that HfsK is not required for holdfast synthesis, anchoring to the cell envelope,

161
HfsH is a predicted carbohydrate esterase family 4 (CE4) enzyme. 27 The CE4 family of 162 polysaccharide deacetylases have five catalytic motifs for substrate and co-factor binding, as well 163 as those that participate directly in the catalytic mechanism, 32 which are all present in C.

164
crescentus HfsH, 27 and in H. baltica HfsH (Fig. S2D). In order to test if H. baltica HfsH is a holdfast 165 polysaccharide deacetylase, we engineered a point mutation in a key substrate-binding residue, 166 resulting in an amino acid change from aspartic acid to alanine at position 43 (D43A; Fig. S2D, 167 asterisk). We monitored for the presence of holdfast using fluorescence microscopy with AF488-

168
WGA. Introduction of D43A in the H. baltica hfsH gene phenocopied the hfsH deletion (Fig. 2E,   169 white arrows). We complemented the ∆hfsH mutant with a WT copy of hfsH, or with the point mutant hfsHD43A, expressed under the native promoter on a low copy replicating plasmid (pMR10).

171
Although WT hfsH and hfsHD43A were expressed similarly (Fig. S2E), complementation with the 172 WT allele restored AF488-WGA holdfast labeling in the ∆hfsH mutant background, while the point 173 mutant hfsHD43A did not (Fig. 2E). These results confirm that H. baltica HfsH is involved in holdfast 174 biogenesis and that D43 is important for its activity, similarly to C. crescentus HfsH. 27

175
The above results indicate that HfsH plays an important role in H. baltica holdfast 176 properties, including their anchoring to the cell envelope. We hypothesized that (1) H. baltica 177 hfsH produces a small amount of holdfast polysaccharide that is insufficient to anchor the cell to

189
WGA and collected images every 5 min for 12 h. We observed that H. baltica WT produced 190 holdfasts that were labelled with AF488-WGA (Fig. 3A, upper panels) and that the ∆hfsH mutant 191 initially produced holdfasts similarly to WT (Fig. 3A, lower panels). However, the holdfasts 192 produced by H. baltica ∆hfsH appeared more diffuse compared to WT over time (Fig. 3A and 193 Movie S1A-B). These results show that H. baltica ∆hfsH produces holdfast material, indicating 194 that HfsH is not essential for holdfast synthesis.

195
In order to test how holdfast produced by H. baltica ∆hfsH interacts with a glass surface,

196
we performed time-lapse microscopy using a microfluidic device with a low flow rate (1.4 µl/min).

197
We injected exponential-phase cells mixed with AF488-WGA into the microfluidic chamber, turned 198 off the flow, and imaged the cells every 20 sec for 5.5 h. In the microfluidic chamber, we observed 199 that H. baltica WT cells arrived at the glass surface and produced holdfasts, allowing them to 200 remain bound to the surface (Fig. 3B, upper panels and Movie S1C). In contrast, H. baltica ∆hfsH 201 produced holdfasts that did not remain cohesive on the glass surface and instead formed thread-202 like fibers (Fig. 3B, lower panel and Movie S1D). These results indicate that HfsH in H. baltica is 203 not required for holdfast synthesis, but is essential for maintenance of holdfast cohesive

216
We next labelled holdfasts with AF488-WGA to analyze the effect of varying HfsH 217 expression on H. baltica holdfast production. We induced the expression of HfsH for 2 h in H.

218
baltica hfsH PCu-hfsH using 0 to 250 µM CuSO4, and visualized holdfasts of planktonic cells with 219 AF488-WGA by fluorescence microscopy. Addition of CuSO4 to H. baltica WT and H. baltica 220 hfsH with empty vector controls had no effect on cell anchoring or holdfast surface adhesion 221 (Fig. 4C,upper and middle panels

242
4D, upper panels, blue arrows). At 50 µM CuSO4, we observed a partial restoration of holdfast 243 adhesiveness and cohesiveness as cells were able to stay attached to the surface for longer after 244 re-initiation of the flow, however holdfast adhesiveness was still impaired as shed holdfasts could 245 be easily washed off the surface (Fig. 4D,

264
We analyzed the level of HfsH expression by western blot analysis and found that both HfsHHB 265 and HfsHCC were equally expressed from these promoters (Fig. S4).

285
Since only half of H. baltica hfsH PhfsHB-hfsHCC cells had faint AF488-WGA labelling and 286 surface binding properties were not fully restored to WT levels ( Fig. 5C, right panels), we 287 hypothesized that holdfasts produced by this mutant may have been shed into the medium.

288
Therefore, we tested whether the holdfast produced by this strain can bind to a glass surface by 289 incubating cells on a coverslip at room temperature for 1 h. After incubation, unattached cells 290 were washed off and AF488-WGA was added to label any holdfast bound to the coverslip. As 291 expected, C. crescentus ∆hfsH cross-complemented with PhfsCC-hfsHHB produced holdfasts that 292 bound to the coverslip and anchored the cells to the surface, like WT (Fig. 5D, left panels). We 293 observed that holdfasts produced by H. baltica ∆hfsH PhfsHB-hfsHCC were not able to anchor the 294 cells to the glass surface, although these holdfasts were able to bind to the coverslip (Fig. 5D, 295 right panels). These results imply that expression of HfsHCC from the H. baltica hfsH promoter 296 was sufficient for restoration of holdfast surface binding by H. baltica, but insufficient to maintain 297 interactions with the cell body. In addition, overexpression of either HfsHCC or HfsHHB in C.

318
We next examined the effect of cross-complementation with HfsHCC in H. baltica. We had 319 observed that expression of HfsHCC in H. baltica ∆hfsH from PhfsHB failed to restore holdfast binding, but its overexpression using Pxyl restored surface binding to the level observed in the 321 WT ( Fig. 5A-B). Therefore, we tested how holdfasts purified from H. baltica hfaB overexpressing

356
baltica holdfasts contain galactose monosaccharides. 25 In order to gain insights into how HfsH 357 modifies holdfast properties, we analyzed its impact on these holdfast components.

358
In order to test whether holdfast thiols are present in the deacetylase mutant H. baltica 359 hfsH, we co-labeled exponential-phase cells with both AF488-WGA (green, GlcNAc) and AF594 360 conjugated to maleimide (AF594-Mal), which reacts with free thiols molecules (red). As expected,

433
How holdfast interacts with surfaces remains unclear, but an electrostatic mechanism has 434 been suggested. 23, 25 C. crescentus holdfast binding is affected by pH and NaCl. 23 The mechanism 435 by which NaCl disrupts electrostatic interactions between holdfast components and glass 436 surfaces is unclear. High ionic strength has been shown to reduce the radius of the electrostatic 437 force on a surface, which would lower the likelihood that holdfast polysaccharides are able to 438 interact with the surface. 16 It is also known that increasing ionic strength has no effect on holdfast 439 that are already attached to a surface, 25 suggesting that high ionic strength only impairs the initial 440 interactions between the holdfast and the surface before a permanent bond is established. In

475
All the plasmids and primers used in this study are listed in Table S1 and S2, respectively.

476
In-frame deletion mutants were obtained by double homologous recombination as previously 477 described 40 using suicide plasmids transformed into the H. baltica host strains by electroporation 41 478 followed by sacB sucrose selection. Briefly, genomic DNA was used as the template to PCR-479 amplify 500 bp fragments immediately upstream and downstream of the gene to be deleted. The 480 primers used for amplification were designed with 25 bp overlapping segments for isothermal 481 assembly 42 using the New England Biolabs NEBuilder tools for ligation into plasmid pNPTS139, 482 which was digested using EcoRV-HF endonuclease from New England Biolabs. pNPTS139-483 based constructs were transformed into -select E. coli for screening and sequence confirmation 484 before introduction into the host C. crescentus or H. baltica strains by electroporation. Introduction 485 of the desired mutation onto the C. crescentus or H. baltica genome was verified by sequencing.
For gene complementation, the pMR10 plasmid was cut with EcoRV-HF and 500 bp 487 upstream of the gene of interest containing the promoter, as well as the gene itself, were designed 488 using New England Biolabs NEBuilder tools and fragments were amplified and ligated into 489 plasmid pMR10 as described above. The pMR10-based constructs were transformed into -490 select E. coli for screening and sequence confirmation before introduction into the host C. This assay was performed as previously described 25 with the following modifications. For 507 short-term binding assays, exponential cultures (OD600 of 0.6 -0.8) were diluted to an OD600 of 0.4 508 in fresh marine broth, added to 24-well plates (1 ml per well), and incubated with shaking (100 509 rpm) at room temperature for 4 h. For biofilm assays, overnight cultures were diluted to an OD600 510 of 0.1, added to a 24-well plate (1 ml per well), and incubated at room temperature for 12 hours 512 distilled H2O to remove non-adherent bacteria, stained using 0.1% crystal violet (CV), and rinsed 513 again with dH2O to remove excess CV. The CV was dissolved with 10% (v/v) acetic acid and 514 quantified by measuring the absorbance at 600 nm (A600). Biofilm formation was normalized to 515 A600 / OD600 and expressed as a percentage of WT.    Rzepecki L, Waite J. (1993).

37
In order to test whether H. baltica HfsK is involved in holdfast anchoring, we spotted 38 exponentially growing cultures on a glass coverslip and incubated for 1 h at room temperature to 39 allow for binding to the coverslip. Unbound cells were removed by washing, and AF488-WGA was added to label holdfasts that remained attached to the coverslip. As a control, C. crescentus and 41 H. baltica WT cells were incubated with coverslips, and adherent holdfasts were labeled with 42 AF488-WGA (Fig. S1C). C. crescentus ∆hfsK holdfasts were bound to coverslips but appeared 43 to be spread over the surface, covering a greater area than WT and suggesting that they may be 44 less cohesive (Fig. S1C), in agreement with previous studies. 1 These holdfasts also failed to 45 anchor C. crescentus ∆hfsK cells to the surface (3% of WT, Fig. S1C). In comparison, mutants 46 with deletion of hfsK and its paralogs in H. baltica produced holdfasts that were bound to the glass 47 surface and formed rosettes similarly to WT (Fig. S1C right panel). Interestingly, deletion of the 48 H. baltica hfsK paralog hbal_1184 led to the generation of large cellular aggregates that formed 49 independently of holdfast biogenesis (Fig. S1D). These cells had morphological defects and were 50 surrounded by debris that may have resulted from cell lysis, indicating that Hbal_1184 is likely 51 involved in a different polysaccharide biosynthetic pathway that contributes to cellular viability.

52
We conclude that HfsK and its paralogs do not contribute to H. baltica holdfast binding properties 53 under our assay conditions ( Fig. S1A-C).

57
The bacterial strains used in this study are listed in Table S1

70
All the plasmids and primers used in this study are listed in Table S1 and S2, respectively.

71
In-frame deletion mutants were obtained by double homologous recombination as previously 72 described 6 using suicide plasmids transformed into the H. baltica host strains by electroporation 7 73 followed by sacB sucrose selection. Briefly, genomic DNA was used as the template to PCR-74 amplify 500 bp fragments immediately upstream and downstream of the gene to be deleted. The 75 primers used for amplification were designed with 25 bp overlapping segments for isothermal 76 assembly 8 using the New England Biolabs NEBuilder tools for ligation into plasmid pNPTS139, 77 which was digested using EcoRV-HF endonuclease from New England Biolabs. pNPTS139-78 based constructs were transformed into -select E. coli for screening and sequence confirmation 79 before introduction into the host C. crescentus or H. baltica strains by electroporation. Introduction 80 of the desired mutation onto the C. crescentus or H. baltica genome was verified by sequencing.