SiaABCD – A threonine phosphorylation pathway that controls biofilm formation in 1 Pseudomonas aeruginosa 2

1University of Stuttgart, Institute of Biochemistry and Technical Biochemistry, Stuttgart, Germany 8 2Singapore Centre for Environmental Life Sciences Engineering, Nanyang Technological University, 9 Singapore 10 3Centre for Marine Bio-Innovation, School of Biotechnology and Biomolecular Sciences, University of New 11 South Wales, Sydney, New South Wales, Australia (apresent address: Bio Molecular Systems, Sydney, 12 Australia) 13 4 Konstanz Research School Chemical Biology, Departments of Chemistry and Biology, 14 University of Konstanz, D-78464 Konstanz, Germany 15 5The School of Biological Sciences, Nanyang Technological University, Singapore 16 6 the ithree Institute, The University of Technology Sydney, Sydney Australia 17 7NTU Institute of Structural Biology, Nanyang Technological University, Singapore 18 8 Antimicrobial Resistance Interdisciplinary Research Group, Singapore-MIT Alliance for Research and 19 Technology, Singapore 20


37
Biofilms are ubiquitous, multicellular structures embedded in a self-made matrix, that can occur attached 38 to a solid surface, floating at the gas-liquid interface as a pellicle and can be freely suspended in the liquid 39 phase as aggregates or flocs 1-3 . The ability to form and disperse from biofilms is a ubiquitous feature of 40 microorganisms and is achieved by differential regulation and production of extracellular polymeric 41 substances (EPS) such as polysaccharides, eDNA and proteinaceous surface adhesins 4-6 . 42 Matrix embedded aggregated cells embedded in biofilms are well known for their increased robustness 43 under stressful environmental conditions compared to their single-cell, planktonic counterparts, which 44 largely explains the evolutionary success of the biofilm lifestyle 7-11 . It is thus not surprising, that biofilm 45 formation can be triggered as an adaptive response to oxidative and nitrosative stress 12,13 and to the 46 presence of toxic compounds such as antibiotics 14,15 , surfactants 16,17 or primary fermentation 47 products 18,19 . As a consequence, biofilms enable bacteria to survive in health-care settings despite the use 48 of stringent hygiene regimes 20-22 and to establish chronic infections despite the host immune system and 49 therapeutic interventions [23][24][25][26][27][28] . Chronic infections are particularly problematic, as they are almost 50 impossible to eradicate with conventional therapies [29][30][31][32] . 51 The regulation of biofilm development and virulence traits is complex and can include multiple, highly 52 interconnected, signal transduction pathways 33,6,34-37 . In addition to quorum sensing (QS) 38,39 , nucleotide-53 based second messengers and changes in protein phosphorylation represent key mechanisms for the 54 regulation of these physiological changes [40][41][42][43] . Protein phosphorylation is most often mediated by two-55 component systems (TCSs) or chemosensory signalling pathways [44][45][46] . TCSs are typically composed of a 56 sensor kinase and a cognate response regulator that elicits a specific response upon its phosphorylation. 57 Most sensor kinases in bacteria belong to the family of histidine kinases, transferring a phosphoryl group 58 from a conserved histidine of its transmitter domain to a specific aspartate residue in the receiver domain 59 of the corresponding response regulator. This type of aspartate phosphorylation is labile and, thus, 60 phosphatase SiaA and its kinase counterpart SiaB. We further report crystal structures of the protein-85 phosphatase family-2C (PP2C) domain of SiaA and of the SiaC protein providing insights into their catalytic 86 mechanism and interaction. Our results strongly suggest that the SiaABCD pathway functions through a 87 partner switch mechanism, in which SiaC controls cellular aggregation by regulating SiaD activity, most 88 likely through direct protein-protein interaction. It further highlights the importance of this pathway in 89 the adaptation of P. aeruginosa to variable environmental conditions. We hypothesise that the regulation 90 of biofilm formation through SiaABCD plays an important role in the colonisation of new niches/habitats 91 and supports the persistence of cells under unfavourable conditions. 92

93
SiaABCD is involved in biofilm formation during growth on SDS and glucose 94 Cells of P. aeruginosa preferentially grow as suspended biofilms in minimal medium with glucose 95 (aggregates up to 400 µm in size) or SDS (macroscopic aggregates in the mm range) 16,64 . To investigate 96 whether all genes of the siaABCD operon 70 are involved in biofilm formation under these conditions, we 97 tested growth of individual mutant strains in microtiter plates ( Figure 1AB). 98 As shown previously, growth on SDS led to the formation of macroscopic aggregates in the parental strain 99 but not in the ΔsiaA and ΔsiaD mutants 70 . Likewise, no macroscopic aggregates were observed in cultures 100 of ΔsiaC, whereas ΔsiaB showed increased aggregation. The aggregation was inversely correlated with the 101 optical density (OD600) of the surrounding medium: the OD600 in cultures of strains ΔsiaA, ΔsiaC and ΔsiaD 102 was higher (darker colour in normalised pictures; Figure 1B) than that of the parental strain and the ΔsiaB 103 mutant. No macroscopic aggregates were found during growth with glucose for all strains. However, the 104 OD600 of the culture medium followed a similar pattern as for those grown on SDS. We therefore 105 quantified the attached biofilms in glucose cultures ( Figure 1C). The parental strain formed significantly 106 more biofilm (2.3 -2.8 fold) compared to ΔsiaA, ΔsiaC and ΔsiaD mutants after 15 h. In contrast, 107 inactivation of siaB strongly increased biofilm formation by 16.2 fold. Overall, these results revealed that 108 attached biofilm formation during growth on glucose follows the same pattern as the formation of 109 suspended aggregation observed for cultures grown on SDS. (1,300; left panel of pictures); imaging at higher magnification (10,000; right panels) showed that the 116 multicellular aggregates of different sizes are arranged as three-dimensional structures that projected 117 Figure 1: (A) Growth of the P. aeruginosa strains on 22 mM glucose or 3.5 mM SDS in 12 well microtiter plates after 18 h, 30°C, and shaking at 200 rpm. Images were acquired using a flat-bed scanner (Umax Powerlook), normalised by the "match colour" function of Photoshop (Adobe Photoshop CS5), and converted to greyscale. (B) Images of different OD600 derived from a serial diluted culture of the parental strain grown on glucose. (C) Quantification of attached biomass using a crystal violet (CV) staining assay after growth with glucose for 15 h at 30°C and shaking at 200 rpm. Data are presented as the mean value of four biological replicates and the error bar represents the corresponding standard deviation. Statistical differences between strains are indicated (*p < 0.003, *p < 0.009) and were calculated using a two-sided, unpaired student t-test (α = 0.05, n = 4). above the substratum. These multicellular aggregates were observed also for the ΔsiaA, ΔsiaC and ΔsiaD 118 mutants at lower magnification (1,300), but their abundance was noticeably lower than for the parental 119 strain. Higher magnification (10,000) showed that these areas were less structured in those mutant 120 strains compared to the parental strain. However, the ΔsiaB mutant formed much larger, much more 121 densely packed aggregates than the parental strain. Light diffraction analysis (LDA) can be used to 122 determine the size distribution of aggregates in liquid cultures 64 and a corresponding analysis in our 123 present study demonstrated that cultures of the parental strain were dominated by 10 -200 µM particles 124 (≥ 77% of the total bio-volume of all particles). Particles < 10 µm, which include the single cells, 125 represented only a minor fraction (≤ 4%) and particles >200 µm were present in both biological replicates 126 (7% -28%). In line with SEM results, the distribution of particles in cultures of the ΔsiaA, ΔsiaC and ΔsiaD 127 mutants were strongly shifted towards smaller sizes. We consistently observed a substantial decrease in 128 the number of particles >200 µm and a concomitant increase in the bio-volume of particles <10 um for 129 ΔsiaA (≥ 8.4 fold), ΔsiaC (≥ 15 fold) and ΔsiaD (≥ 10.8 fold). In contrast, cultures of ΔsiaB were dominated 130 by particles >200 µm (≥ 77%) and almost no particles < 10 µm were detected (≥ 1%). 131 132 133 Figure 2: Scanning electron microscopy (SEM) and laser diffraction analysis (LDA) demonstrated the different aggregation phenotypes of P. aeruginosa wildtype and mutant strains during growth with glucose. Triplicate cultures of strains were grown with 22 mM glucose in 50 mL cultures shaking at 200 rpm at 30°C. After 5.5 h incubation, the replicate samples were pooled and used for SEM. The white arrows (left panel) highlight representative regions showing aggregated cells, which were then examined at higher magnification (right panel). The samples were also analyzed using a particle size analyzer (SALD 3101, Shimadzu; see Material and Methods) in the range 0.5 -3000 µm diameter. The data obtained from two independent replicates (A and B) is presented as percentages of the total bio-volume of particles distributed over three different size ranges (single cells, microscopic aggregates and macroscopic aggregates) relative to the total bio-volume of all particles detected. The data are the mean value of triplicate measurements with the corresponding technical error (see Supplementary for details).
SiaABCD regulates biofilm formation as a response to carbon availability 134 Biofilm formation is a dynamic process that varies according to the time point of sampling and the carbon 135 source used. Thus, we monitored the impact of the SiaABCD pathway during biofilm development and on 136 growth with various carbon sources (Figure 3, Figure S1). In these experiments, the parental strain always 137 exhibited the same growth pattern, irrespective of the carbon source used: i) a steady increase in biofilm 138 formation while the carbon source was in excess; ii) a decrease in biofilm formation upon exhaustion of 139 the carbon source; and iii) an increase in optical density in the culture supernatant that lagged behind 140 biofilm formation, but continued to increase even after the cells experienced carbon starvation. 141 The ΔsiaA (Figure 3), ΔsiaC ( Figure S1) and ΔsiaD ( Figure S1) mutants showed a similar growth pattern, 142 however, with a strong decrease in biofilm formation and mild increase in the maximal OD600 at the end 143 of the experiment. In contrast, the ΔsiaB mutant predominantly grew as a biofilm with generally lower 144 planktonic growth compared to all other strains. Biofilms of ΔsiaB dispersed upon glucose and succinate 145 limitation, but not when grown on ethanol or 2,3-butanediol. However, it is important to note that 146 cultures of the parental strain, ΔsiaB and ΔsiaC growing on 2,3-butanediol were still not carbon-limited 147 by the end of the experiment (< 0.6 mM). In contrast, carbon depletion and dispersal was observed for all 148 other strains and conditions. 149 150 151 152 Figure 3: Quantification of formation of attached biofilms and of growth in suspension by the P. aeruginosa strains, as followed in liquid cultures grown with 22 mM glucose, 20 mM succinate, 40 mM ethanol or 20 mM 2,3-butanediol. Individual 24 well microtiter plates were used to quantify attached biofilms by crystal violet (CV) staining (grey bars) as well as the growth in the supernatant as OD600 (black dots); substrate concentrations were also determined (open circles). For crystal violet (CV) staining, the error bars represent the standard deviation of four biological replicates. The data of OD600 and substrate concentration represent the mean value from the same quadruplicates but quantified from pooled (1:1:1:1 [v/v/v/v]) samples using a photometer, the GO assay kit and specific HPLC methods, respectively (see Methods section for details). Cultures were incubated at 30°C and 200 rpm shaking. 153 154 155 Figure 4: Biochemical characterization of the purified phosphatase activity of the C-terminal part (SiaA-PP2C; amino acids 386-663) of the SiaA protein (Genbank: NP_248862) with p-nitrophenyl phosphate (pNPP) as substrate. Phosphatase activity was measured as increase in absorbance at 405 nm. Absorbance readings were converted to M production formation using a generated standard curve for the determination of specific activities. The buffer used contained 150 mM NaCl, 20 mM Tris-HCl pH 7.5 and incubations were performed at 37°C. For enzyme kinetics, a higher buffering capacity of 100 mM Tris-HCl pH 7.5 was used. SiaA shows Mn 2+ -dependent phosphatase activity with pNPP 156 The SiaA protein is predicted to be a PP2C-like protein phosphatase (PPM-type) 157 (http://www.pseudomonas.com/feature/show/?id=103077&view=functions). A distinctive feature of 158 these phosphatases is their dependency on Mn 2+ or Mg 2+ ions as cofactors and serine and/or threonine 159 residues in their target protein (http://www.ebi.ac.uk/interpro/entry/IPR001932). To characterize the 160 activity of the SiaA protein, we purified the C-terminal part of SiaA (amino acids 386-663; SiaA-PP2C; 161 Genbank: NP_248862) including the putative PP2C-like phosphatase domain (amino acids 453-662) from 162 the parental strain using Escherichia coli as a heterologous host by metal-affinity chromatography ( Figure  163 4A; Figure S2). The activity of the purified SiaA-PP2C was tested for its metal dependency using p-164 nitrophenyl phosphate (pNPP). SiaA-PP2C showed strong phosphatase activity in the presence of 20 mM 165 of Mn 2+ (1.658 ± 0.3766 kU mg -1 ), lower activity in the presence of 20 mM Co 2+ (0.6747 ± 0.2932 kU mg -1 ) 166 and no detectable activity with similar amounts of Mg 2+ , Ca 2+ or Zn 2+ ( Figure 4B). Subsequently, kinetic 167 parameters for SiaA-PP2C at various Mn 2+ concentrations were derived from nonlinear regression to a 168 Michaelis-Menten model ( Figure 4C, Table 4E). From experiments with 20 mM Mn 2+ , a maximum velocity 169 of (Vmax) of ~2.5 ± 1.01 kU mg -1 , a substrate affinity (KM) of ~6.254 ± 1.01 mM and a turnover number 170 (kcat) of ~8.389 ± 0.5142 min -1 were calculated.. 171 SiaA is a Ser/Thr phosphatase that uses phosphorylated threonine residue 68 of SiaC as substrate 172 SiaC contains a phosphorylation site at threonine residue 68 (T68) 73 . To confirm this phosphorylation site 173 and to test whether it represents a substrate for the phosphatase domain SiaA-PP2C and the predicted 174 kinase SiaB (see section below), His6-tagged SiaC variants were purified from E. coli and ∆siaA cell lysates 175 For the purified P. aeruginosa SiaC (SiaC P ) variant however, the corresponding phosphorylated peptide 181 was found at a higher abundance than the non-phosphorylated peptide (7.8 x × 10 6 vs. 3.2 × 10 6 peak 182 area). Hence, these preparations contained relevant amounts of phosphorylated SiaC P (71.27% of total 183 peak area). The SiaC P preparations was tested as substrates for SiaA-PP2C and incubation, the protein 184 samples were separated by SDS-PAGE and the protein band representing the SiaC protein were excised 185 and analysed by PMS ( Figure 5A; Tables 3). Incubation with SiaA-PP2C resulted in > 10 fold increase in the 186 phosphorylated peptide (LLYLNTSSIK) compared to the control incubation without SiaA-PP2C added. 187 Hence, the PMS data obtained from these enzyme assays provide direct evidence that the phosphorylated 188 threonine at position 68 of SiaC is a target for SiaA phosphatase activity. 189 SiaB is a protein kinase that uses dephosphorylated SiaC as a substrate 190 Protein sequence analysis of SiaB ( Figure 5B) suggested that SiaB is a protein kinase. To test whether SiaC 191 is a target for SiaB activity, we purified a His6-tagged allele of SiaB from E. coli by metal-affinity 192 chromatography ( Figure S2) and performed ATP consumption assays (ADP-Glo™ Kinase Assay; Promega) 193 with SiaC as the substrate ( Figure 5C). Incubation of SiaB (0.5 µM) alone or with SiaC P (5 µM) did not 194 consume ATP, whereas incubations of SiaB with SiaC (5 µM) consumed 4.45 µM ATP. 195 To confirm that SiaB was specific towards position T68 of the SiaC protein, we performed the kinase assay 196 in the absence or presence of ATP and subsequently analysed the SiaC protein by PMS ( Figure 5A; Tables  197   S3). For SiaC incubations with SiaB in the absence of ATP, the non-phosphorylated peptide was > 1000 198 fold more abundant than the corresponding phosphorylated peptide (2.9 x 10 9 vs. 1.3 x 10 6 peak area). 199 Following incubation of SiaC with SiaB in the presence of ATP, the abundance of the phosphorylated 200 peptide increased (1.5 x 10 9 peak area) and the non-phosphorylated peptide decreased dramatically (2.7 201 x 10 8 peak area), demonstrating that T68 of SiaC constitutes a target for SiaB kinase activity. Notably, reactions in which SiaB and SiaC or SiaC P were incubated together with SiaA-PP2C (0.5 µM), the 205 consumption of ATP was increased (6.42 M and 6.57 M in the presence of 10 M ATP and to 12.01 µM 206 ) as identified by their parental (and by their fragment) masses without or with mass-shift from T68-phosphorylation, are indicated as peak area and number of peptide spectrum matches (PSMs). All incubations were performed in suitable buffer conditions for 2 h at 30°C and then separated by SDS-PAGE for PMS (see Supplementary file for details) (B) Homology model (C-score = -2.36, estimated TM-score = 0.44 ± 0.14 and estimated root-mean-square-mean (RMSD) = 10.4 ± 4.6 Å) of SiaB (PA0171) predicted using I-Tasser (https://zhanglab.ccmb.med.umich.edu/I-TASSER/) shown as cartoon with 80% transparent surface using the Pymol software (version 2.1.1). The predicted catalytic glutamic acid residue at position 61 of the putative protein kinase is highlighted (red sticks). (C) ATP consumption of SiaB kinase activity measured by the ADP-Glo™ assay kit. Assays were performed in the presence of 10 µM or 25 µM (orange bars) ATP. (D) Phosphate released via SiaA phosphatase activity as measured by the malachite green assay. All reactions were carried out with 0.5 M SiaA or B with 5 M SiaC in buffer containing 20 mM MgCl2 or MnCl2 (green bar), and 10 or 25 M ATP (orange bars). Data represent results from at least two independent replicates and error bars represent standard deviation of the mean. Results below the detection limit are indicated (n.d.). and 12.39 µM in the presence of 25 M of ATP, respectively) when observed after the reactions (Figure  207 5C). These experiments demonstrated that SiaA-PP2C was functional in the presence of Mg 2+ and that 208 SiaA-PP2C and SiaB were both enzymatically active on their respective substrates (SiaC P and SiaC, 209 respectively), catalysing a cycle of SiaC phosphorylation and de-phosphorylation. Activity of SiaA with 210 Mg 2+ ions was not observed with the artificial substrate pNPP ( Figure 4A). To test whether the metal-211 dependency of SiaA-PP2C is different using SiaC P as substrate compared to pNPP, we quantified activities 212 using a Malachite-Green phosphate assay ( Figure 5D). In reactions containing SiaA-PP2C and SiaC P , the 213 concentration of phosphate quantified in the presence of Mg 2+ (4.74 µM) was slightly higher than that 214 determined in the presence of Mn 2+ (4.52 µM), and both were strongly correlated with the amount of 215  Table S4. 225 The SiaA-PP2C domain is arranged as a tight dimer in the crystal asymmetric unit. The electron density 226 map of SiaA-PP2C allowed unambiguous tracing of residues 407 to 663 for monomer A and 409 to 663 for 227 monomer B. However, no clear electron density was present for residues 521-528 and 578-581 of chain 228 A and residues 522-527 and 577-579 of chain B, indicating a high degree of flexibility in these solvent-229 exposed loops regions. The SiaA-PP2C monomer adopts the canonical  fold first described for The core structure of the SiaA-PP2C monomer is a -sandwich composed of two antiparallel -sheets each 234 comprising five β-strands. The active site is at the apex of the  sheet structure ( Figure 6B). This core 235 structure is flanked on the N-terminal side by helices α1-α4 and on the C-terminal end by helices α5-α7, 236 forming a four-layered αββα structure. The two monomers interact extensively with each other through 237 residues projecting from helices α1, α2 and α3, forming a compact homodimer that buries a total solvent 238 accessible area of 1392 Å 2 ( Figure 6B and C), which is consistent with the observation of a dimer by size 239 exclusion chromatography (data not shown). The homodimer presents two concave surfaces that are 240 likely to accommodate the incoming substrates ( Figure 6D). Mycobacterium tuberculosis is shown in Figure 6F. The two Mg 2+ ions labelled as M1 and M2 directly 248 involved in the catalytic mechanism are coordinated by oxygen atoms from three evolutionary conserved 249 aspartate residues and occupy equivalent positions in SiaA-PP2C compared to the two Mn 2+ ions found in 250 the Rv1364C active site. Metal-coordinating residues are all conserved except C475 of SiaA-PP2C, which 251 interacts with metal ion M2 through its carbonyl oxygen ( Figure 6E and F). However, compared to 252 Rv1364C, which only has two bound metal ions, a third Mg 2+ ion (M3) was found in SiaA-PP2C coordinated 253 by D600 and with an incomplete octahedral coordination shell. D604 and D534 assist in stabilization of 254 M3 through formation of hydrogen bond with M3-coordinating water molecules. Given the structural 255 similarity with PP2C 74 , an SN2 mechanism is the most plausible whereby the water molecule bridging M1 256 and M2 ( Figure 6E) performs a nucleophilic attack of the phosphorus atom from the phosphate bound to 257 the threonine target residue. However, M3 could also play a direct role in the catalytic activity, as was 258 recently proposed for the human PPM1A protein, a negative regulator of cellular stress response 259 pathways 77,78 . We note that near the entrance of the active site, a hydrophobic pocket lined with residues 260 I554, V564, L603, F619 and A620 constitutes an attractive site for the design of inhibitors ( Figure 6D and 261 G). Another distinct feature of SiaA-PP2C is a flexible loop located between β7 and β8, whereas a Flap 262 subdomain, which has been reported to aid in substrate specificity, is present at the equivalent position 263 in human PPM1A and many other PP2C-type phosphatases 78 ( Figure 6H). and three α-helices ( Figure 7A). A hydrophobic pocket next to T68 is formed by residues L61 and L63 272 projecting from β4, W95 from β5, I71 and M74 from α2, L107 and F111 from α3 while L66 belongs to the 273 β4-β5 loop ( Figure 7B and C). Mycobacterium paratuberculosis (PDB accession code: 4QTP; Figure 7D) as the top hit with a Z score of 276 9.0, an amino-acid sequence identity of 10% and a RSMD of 2.8 Å over the -carbon atoms of 117 residues. 277 Despite sharing very low sequence identity, a closer examination of both structures reveals a conserved 278 overall topology with several unique features, especially at β3, β4 strands and α3 helix that are longer in 279 SpollAA compared to SiaC. These variations are probably related to differences in the ability to establish 280 protein-protein interactions. Nonetheless, their target phosphorylation sites, T68 for SiaC and S58 for 281 SpollAA, are both located at the N-terminus of helix α2. 282 To compare the phosphorylated and non-phosphorylated form SiaC, a molecular model was built by 283 adding a phosphate group to T68 ( Figure 7E). After energy minimization, no major structural difference 284 was observed between the SiaC structure and the predicted SiaC P structure, except that the N-terminal 285 half of α3 shifts slightly away from the hydrophobic pocket defined above (by distances of 1.2 Å for L107 286 SiaC ( Figure 7G and H). The binding of the two proteins was driven by both a decrease in enthalpy and 299 increase in entropy. The stoichiometry of 0.5 suggested that only one SiaC molecule binds to a dimer of 300 SiaB. Binding of SiaC to the second SiaB molecule of the dimer was not observed, even when an excess of 301 SiaC was injected ( Figure S4). In contrast, a 2SiaB:2SiaC complex was suggested by size exclusion As SiaA-PP2C targets SiaC P , a transient binding of these two proteins is likely to occur. To study the 311 interaction between SiaA and SiaC during the dephosphorylation reaction, we performed a MD simulation 312 on SiaA binding to SiaC and SiaC P . The SiaA-SiaC complex was predicted by molecular docking of SiaA-PP2C 313 dimer and SiaC P -SiaC using ZDOCK, followed by 60 ns of MD simulation ( Figure 8A and B). These 314 experiments demonstrated that the phosphorylated pT68 is required for productive binding to the 315 catalytic site of SiaA-PP2C, where it remains efficiently bound via the catalytic Mg 2+ ion M3 and the 316 bridging water molecules surrounding it ( Figure 8C-E). This result is consistent with the experimental 317 observation that only pT68 in SiaC P constitutes the substrate for SiaA-PP2C activity. In contrast, the non-318 phosphorylated T68 dissociated from the catalytic site in less than 15 ns of simulation, suggesting that 319 SiaC will quickly dissociate from SiaA upon dephosphorylation. 320 Based on the simulated model, several key interactions that could contribute to substrate specificity of 321 SiaA-PP2C were identified. First, R652 stabilizes the positively charged phosphate group of pT68 through 322 the formation of a salt bridge ( Figure 8F). In addition, the flexible loop between α5 and α6 of SiaA inserts 323 in a shallow pocket located between α2, α3, β4, β5 and β6 of SiaC P and establishes hydrophobic contacts 324 with L61, L66, I71, M74, M75, L78, W95, L107 and F111 through F612 ( Figure 8G). Notably, the salt bridge 325 identified between R618 and E609 of SiaA is likely to restrain the orientation of the flexible loop between 326 α5 and α6, which could favours interaction with SiaC P . The non-aggregative mutant ΔsiaA was recently 327 found to harbor deletion of two amino acids in the phosphatase domain of SiaA (G611F612; SiaA-PP2C*) 70 . 328 To gain a deeper insight into the underlying mechanism for the loss of function in SiaA-PP2C*, we 329 generated a homology model using SWISS-MODEL (https://swissmodel.expasy.org/) with the SiaA-PP2C 330 crystal structure as template (GMQE =0.98, QMEAN = -0.54). Notably, the deleted amino acids G611F612 331 map to the flexible loop between α5 and α6, and F612 is the essential residue for interaction with the 332 hydrophobic pocket of SiaC P (Figure 8H1-I2). Thus, it is possible that by being unable to form this 333 hydrophobic contact, SiaA-PP2C* cannot accommodate its substrate and is therefore unable to Overall, this highlights that the SiaABCD pathway represents an attractive target for signal interference 377 for therapeutic interventions. However, for the development of any signal interference drug, an in-depth 378 knowledge about the underlying molecular mechanisms of the target pathways is a prerequisite. Here, 379 we provide compelling evidence that the SiaABC proteins make use of a partner switch mechanism to 380 balance cellular aggregation in response to carbon availability and propose that SiaD activity is the final 381 target of this regulation. A partner switch system (PSS) typically involves a PP2C phosphatase (SiaA), an 382 anti-sigma factor with kinase activity (SiaB) and anti-sigma factor antagonist representing the target for 383 phosphorylation (SiaC). In line with such a mechanism, the phosphorylation status of the proposed anti-384 sigma factor antagonist SiaC is crucial for the switch to occur. While SiaC showed a strong binding to SiaB, 385 the binding affinity was significantly reduced after phosphorylation. Based on our bioinformatics 386 approach, the dissociation upon SiaB activity shares a similar mechanism of action to that of SpoIIAA and 387 SpoIIAB, where phosphorylation of the anti-sigma factor antagonist SpoIIAA only results in small structural 388 changes but leads to a rapid dissociation from the anti-sigma factor SpoIIAB due to electrostatic and steric 389 clashes 79 . Although SiaABC resembles the SpoIIAA-AB-sigma F and SpoIIE system, we propose that its role 390 is likely not the sequestration of a sigma factor, but rather to regulate c-di-GMP biosynthesis of SiaD. Our 391 proposal is based on recent studies that described partner switches that indeed regulate DGC activity. In 392 P. aeruginosa, the HptB-HsbR-HsbA system was shown to promote DGC activity of HsbD following its 393 interaction with the phosphorylatable anti-sigma factor antagonist HsbA, finally regulating biofilm 394 formation and swimming motility 107 . More recently, a PSS was shown to be involved in mixed-linkage ß-395 glucan synthesis in S. meliloti through the induction of c-di-GMP levels 108 . In this system, the Ser/Thr-396 specific phosphatase/kinase couple BgrU/BgrW controls the phosphorylation status of the BgrV protein 397 and hence controls the activity of the DGC BgrR. 398 We were unable to purify the SiaD protein in a soluble form to directly test the proposed interaction of 399 SiaD with SiaC and/or SiaC P . However, based on the similarities described above, we propose a model in 400 which both SiaA and SiaB balance the phosphorylation status of SiaC at position T68 that in turn directly 401 influences the activity of SiaD. In this scenario, the activation of SiaA shifts the equilibrium of SiaC/SiaC P 402 towards the phosphorylated form through SiaB activity. This would stimulate biosynthesis of c-di-GMP by 403 SiaD resulting in increased cellular aggregation, most likely through influencing the direct interaction of 404 SiaD with either SiaC and/or SiaC P . As a negative feedback response, SiaC P has an increased binding affinity 405 towards SiaA, which downregulates SiaD activity through the dephosphorylating of T68 and thereby 406 inversely affecting protein interactions. 407 We identified the PP2C domain of SiaA and the putative anti-sigma factor antagonist SiaC as prime targets 408 for the development of biofilm interference drugs. The PP2C-type phosphatase SiaA, while conserved in 409 prokaryotes and eukaryotes, is structurally unique. A Flap sub-domain, suggested of being involved in 410 substrate interaction and specificity, has been reported in human PPM1A and many other PP2C-type 411 phosphatases 78 . SiaA-PP2C lacks the Flap sub-domain and instead has a long loop between β7 and β8 that 412 does not seem to interact with SiaC P according to MD simulations. However, the flexible loop between α5 413 and α6 might aid in substrate specificity of SiaA by interacting with the hydrophobic pocket of SiaC near 414 the phosphorylation site. Although further studies are needed to confirm this finding, the fact that the 415 phosphatase domain of the mutant allele SiaA-PP2C* has a deletion in G611F612, of which the latter 416 residue seems to be crucial for the interaction with SiaC, is supportive of this hypothesis. Thus, the design 417 of drugs targeting the flexible loop in SiaA, could be exploited to interfere with SiaA/SiaC P interaction and 418 hence, block downstream induction of biofilm formation through the inhibition of SiaC P 419 dephosphorylation. SiaC also represents an exciting target for interference, as it is structurally most 420 similar to bacterial anti-sigma factor antagonists with little homology with human proteins. The 421 hydrophobic pocket next to the phosphorylation site of SiaC is of particular interest, as it seems to be 422 involved in the stability of the SiaA/SiaC interaction. In addition, a large patch of negatively charged 423 surface is present at α2 and α3 near the hydrophobic pocket ( Figure 7C), and could thus be useful for drug 424 design as potential anchor points. 425 The high-resolution structures for of these target proteins will allow the de novo design of small molecule 426 drugs or synthetic peptides. Using physiological mid-throughput assays based on biofilm formation, novel 427 drugs can also be screened for their ability to penetrate cells, which is of essential importance to reach 428 the proposed interference target sites. In this regard, it is interesting to note that the use of engineered 429 phages as delivery vehicles for protein and/or peptide-based drugs represents a suitable and highly 430 attractive approach to potentially overcome this obstacle [109][110][111][112][113] . As an alternative to the de novo design 431 approach, existing drug and small molecule libraries as well as synthetic peptide libraries could be 432 screened for activity against these targets. 433 In summary, the presented work provides a deeper understanding of the SiaABCD mediated regulation 434 by expanding the fundamental genetic and biochemical mechanisms that control cellular aggregation in 435 response to carbon availability and it potentially paves the way for the development of novel biofilm-436 interference drugs and combinatorial therapies, which are desperately needed to combat the rising 437 antibiotics crisis. to all particle in the sample. Two biological replicates were analysed and from each sample, the particle 499 size distribution was quantified with three subsequent measurements. Data are presented as the mean 500 value of technical triplicates from a single experiment with the error representing the technical variation. 501

Enzymatic assays 502
For enzymatic activity measurements the purified SiaA-PP2C protein fragment of P. aeruginosa PAO1 was 503 used. The phosphatase activity of SiaA-PP2C was accessed using either para-nitrophenyl phosphate 504 (pNPP) or the purified SiaC P protein as the target substrate. For assays with pNPP as the substrate, 505 phosphatase activity was determined via the production of a colorimetric p-nitrophenolate product that to 80 L of the sample to stop the reaction. The mixture was then incubated at room temperature for 30 529 min for the generation of a green complex formed between Malachite Green, molybdate and free 530 orthophosphate. Measurements were taken at 620 nm using a microplate reader. Readings were then 531 converted to phosphate measurements using a standard curve. For ADP-Glo™ assays, 20 uL of the ADP-532 Glo™ reagent was added to 20 uL of the sample to stop the reaction and deplete remaining ATP. If enzyme 533 reactions were carried out in the absence of Mg 2+ , MgCl2 was supplemented to the mixture to a final 534 concentration of 20 mM. The mixture was incubated at room temperature for 40 min. A second kinase 535 detection reagent was added to the mixture to convert ADP consumed to ATP and for conversion of the 536 ATP signal to luminescence signals and further incubated for 45 min. Luminescence was measured using 537 a microplate reader and converted to ATP consumed in kinase reaction using a standard curve generated. 538 At least two independent replicates with technical duplicates were carried out for each enzymatic 539 reaction. 540

MS fingerprinting analysis of SiaC phosphorylation state 541
For assays with SiaC P as substrate, phosphatase activity was determined using the protein that was initially 542 produced and purified from strain ∆siaA (see supplementary material file for detailed information). To 543 elucidate the phosphorylation site in the SiaC protein, 20 µg of purified SiaC P was incubated in the absence 544 or presence of 20 µg of purified SiaA for 1 h at 37°C in a final volume of 40 µL assay solution (20 mM HEPES 545 [pH 7.5], 300 mM NaCl, 0.02 % β-mercaptoethanol, 0.1 mg mL -1 BSA, 5 % glycerol, 5 mM Mn 2+ , 1 mM 546 EDTA). After 2 h of incubation at 30°C, 20 µL were used for the separation of the two proteins using a 15 547 % SDS-PAGE gel. The SiaC protein band (17 kDa) was excised from the SDS-PAGE gel, and the gel sample 548 was submitted to shotgun peptide mass spectrometry (PMS) analysis, to identify SiaC and its 549 phosphorylation state at threonine residue 68 (T68) 73 . Therefore, the relative abundance of signals in the 550 PF-MS fragmentation-mass spectra for the specific peptide containing the phosphorylation site T68 (i.e., 551 peptide LLYLNTSSIK) in a non-phosphorylated state (parental fragmentation-mass spectrum for peptide 552 LLYLNTSSIK) and for mass-shifts indicating its phosphorylation (peptide LLYLNTSSIK with monoisotopic 553 mass-change, +79.96633 Da; average mass-change +79.9799 Da), was determined (see Supplementary  554 for more details). 555

Analytics of supernatants 556
For glucose quantification, the liquid cultures were filtered through a 0.22 µM PES filter (PN 4612, Pall) 557 and quantified using the GO assay kit (GAGO20, Sigma Aldrich). Briefly, one volume of sample or standard 558 solution (20 µL) was mixed with two volumes of assay reagent (40 µL) in a 96 well plate and incubated 559 statically at 37C for 30 min. Subsequently, the reaction was stopped by the addition of 12 N H2SO4 (40 560 µL) to the reaction mixture. Glucose concentrations were quantified at 540 nm using a microplate reader 561 (Infinite 200 pro, Tecan). Zero -80 µg mL -1 glucose standard solutions were used for the calibration curve. 562 Concentrations of succinate, ethanol and 2,3-butanediol were quantified by a HPLC method using an 563 Aminex HPX-87H 300 x 7.8 mm ion-exchange column (BioRad, Munich, Germany) heated to 60°C. The 564 eluent was 5 mM H2SO4, which was delivered to the column by a LC-10ATvp pump (Shimadzu, Munich, 565 Germany) at a flow rate of 0.6 mL min -1 . The eluent was continuously degassed with a DGU-20A3R 566 degassing unit (Shimadzu, Munich, Germany). Samples were injected using 10 µL with a 234 autosampler 567 (Gilson, Limburg-Offheim, Germany). Resolved compounds were analysed with a refractive index detector 568 (RID-10A, Shimadzu, Munich, Germany) and the data processed using the Shimadzu Lab solutions 569 software version 5.81. Concentrations were finally calculated from calibration curves of the corresponding 570 metabolite of interest. 571

Additional methods and procedures 575
Detailed methods for construction of the siaB and siaC mutant strains, the construction of plasmids, and 576 the production and purification, crystallization and/or shotgun peptide mass spectrometry, as well as the 577 procedure for the generation of homology models, molecular docking experiments, and procedures for 578 The authors further declare that the research was conducted in the absence of any commercial or financial 589 relationships that could be construed as a potential conflict of interest. Janosch Klebensberger would like 590 to thank Prof. Bernhard Hauer for his continuous support. We also acknowledge beam time allocation at 591 the SOLEIL synchrotron. The MD simulations for this article were performed on ASPIRE 1 of the National 592 Supercomputing Centre, Singapore (https://www.nscc.sg). 593