Activity-Based Protein Profiling IdentifiesKlebsiella pneumoniaeSerine Hydrolases with Potential Roles in Host-Pathogen Interactions

20 21 Klebsiella pneumoniae is a normal resident of the human gastro-intestinal tract and an opportunistic, 22 critical priority pathogen that can cause a variety of severe systemic infections. Due to emerging 23 multi-drug resistance of this pathogen, the discovery and validation of novel targets for the 24 development of new treatment options is an urgent priority. Here, we explored the family of serine 25 hydrolases, a highly druggable and functionally diverse enzyme family which is uncharacterized in 26 K. pneumoniae. Using functionalized covalent fluorophosphonate inhibitors as activity-based probes 27 we identified 10 serine hydrolases by mass spectrometry-based activity-based protein profiling, 7 of 28 which were previously uncharacterized. Functional validation using transposon mutants deficient in 29 either of the putative lysophospholipase PldB, esterase YjfP and patatin-like phospholipase YchK 30 revealed severe growth defects in human colonic organoid co-culture models and reduced virulence 31 during Galleria mellonella infection. Mutants deficient in the PldB and YjfP, but not YchK show 32 increased susceptibility to killing by complement and the antimicrobial peptide antibiotic polymyxin 33 B, suggesting a role in maintaining cell envelope integrity. Biochemical characterization and 34 structural analysis of recombinant YjfP suggest this protein is a deacetylase. This study gives 35 important insights into the molecular mechanisms underlying virulence and cell physiology of K. 36 pneumoniae at the host-pathogen interface and it positions PldB, YjfP and YchK as potential 37 antimicrobial or anti-virulence target candidates, inhibition of which might synergize with existing 38 antibiotics and human immune defenses. 39 .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 28, 2024. ; https://doi.org/10.1101/2024.05.28.596221doi: bioRxiv preprint 40

41 42 The Gram-negative enterobacterium Klebsiella pneumoniae is an opportunistic pathogen that 43 naturally resides in the gastrointestinal tract of healthy humans and animals but can cause a variety of 44 severe extra-intestinal infections, including urinary tract, bloodstream, and lung infections 1. This 45 bacterium is also known for its ability to produce extended-spectrum beta-lactamases (ESBLs) and 46 carbapenemases enzymes that confer resistance to many antibiotics, including the carbapenems, 47 which are often used as last-resort antibiotics 2,3. This makes the infections caused by K. pneumoniae 48 difficult to treat and increases the risk of mortality 4. Carbapenemase-producing K. pneumoniae are 49 critical priority pathogens on the list of multi-drug resistant (MDR) pathogens by the World Health 50 Organization (WHO). Therefore, research and development of new therapeutics need to be 51 prioritized to prevent entry into a post-antibiotic era 5. Beyond multi-drug resistance the emergence 52 and global spread of hypervirulent K. pneumoniae strains, which are common causes of liver 53 abscesses and other systemic infections are of great concern 6,7. 54 55 In the light of the spread of multi-drug resistant strains of this pathogen, it is essential to understand 56 the molecular factors that contribute to its pathogenicity and validate their potential as drug targets. 57 One particularly attractive family of putative target enzyme are the serine hydrolases (SHs), a large 58 and diverse enzyme class with roles in various cellular processes, from metabolism, to signaling, and 59 regulation of gene expression, and can be putative drug targets for a variety of diseases 8-10. Due to 60 their conserved mechanism and active site architecture, many SHs share reactivity towards the same 61 active site-directed inhibitors, such as fluorophosphonates. This can be exploited by a 62 chemoproteomic technique called activity-based protein profiling (ABPP) that uses functionally-63 tagged active site-directed covalent enzyme inhibitors (activity-based probes, ABPs) to detect active 64 enzyme species under physiological conditions in complex samples of interest 11-14. Fluorescent 65 fluorophosphonate (FP)-probes can be used to detect SHs in gel-based ABPP, whereas FP-biotin 66 allows for pull-down and identification of targets by liquid chromatography-mass spectrometry (LC-67 MS) (Schematically illustrated in Fig. 1A, B ) . I n t h i s w a y , S H a c t i v i t i e s h a v e b e e n p r o f i l e d i n 68 diverse biological samples including animal tissues15, human cell lines 16, and more recently bacterial 69 pathogens including Staphylococcus aureus 17, S. epidermidis 18, Mycobacterium tuberculosis 19-21, 70 and Vibrio cholerae 22, the gut commensal Bacteroides thetaiotaomicron 23 and archaea 24. In our 71 recent chemoproteomic study on S. aureus , we identified a family of ten largely uncharacterized 72 fluorophosphonate-binding serine hydrolases FphA-J several of which have roles in pathogenesis 17 73 and bacterial stress responses 25. These enzymes can be targeted by specific small molecule inhibitors 74 and probes holding promise as anti-virulence 17 and pathogen-specific imaging targets26. 75 Surprisingly, very few serine hydrolases of K. pneumoniae have been described or functionally 76 characterized. A mutant of the periplasmic HtrA-like serine protease DegP showed higher 77 susceptibility to complement-mediated killing and reduced levels of capsule polysaccharide 27. 78 However, it remains unclear how this protease mediates these effects. In E. coli , the two HtrA-serine 79 proteases DegP and DegS have described functions in extracytoplasmic protein quality control and 80 stress responses (reviewed in28). 81 Another serine hydrolase, the phospholipase Tle1 has been described as a secreted effector of the 82 type 6 secretion system (T6SS) in a hypervirulent K. pneumoniae strain29,30. Due to toxic effects of 83 T6SS effectors on both bacterial competitors and host tissues, the K. pneumoniae T6SS is important 84 for both long-term colonization of the gut 30 as well as intestinal barrier dysfunction and bacterial 85 translocation from the gut31. 86 .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 28, 2024. ; https://doi.org/10.1101/2024.05.28.596221doi: bioRxiv preprint Here, we performed a systematic analysis of serine hydrolase activities in K. pneumoniae using a 87 broad-spectrum serine hydrolase probe. Seven out of 10 identified hydrolases were poorly annotated 88 and previously uncharacterized. Functional characterization using transposon mutants in 89 colonization/infection assays involving HT29-MTX cells, human-derived 2D colonic organoids, and 90 Galleria mellonella infection suggest crucial roles for at least three of these hydrolases during 91 infection and/or colonization. Our data suggest that the cell membrane-associated putative 92 lysophospholipase PldB and esterase YjfP contribute to cell envelope integrity and contribute to 93 resistance to antimicrobial peptides and complement, whereas the putatively secreted patatin-like 94 phospholipase YchK may contribute to K. pneumoniae virulence by degrading mucus or host-cell 95 derived phospholipids. Biochemical and structural analysis of recombinant YjfP suggest that this 96 enzyme functions as a deacetylase. 97 98 .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 28, 2024. ; https://doi.org/10.1101/2024.05.28.596221doi: bioRxiv preprint 99

100 101 Bacterial strains and culture conditions 102 Strains of K. pneumoniae MKP103 and its isogenic mutants are summarized in Supplementary 103 Table 1. All strains were routinely cultured on Blood agar or Luria-Bertani broth (LB broth). The 104 bacterial strains were incubated at 37°C, and liquid cultures were aerated by shaking at 180 rpm, 105 except when intended for co-incubation with mammalian cells or organoids (described in their 106 respective sections below). 107 108 Bacterial labeling with fluorophosphonate-tetramethylrhodamine (FP-TAMRA) 109 After overnight growth on either an agar plate or in liquid culture, as indicated, the bacteria were 110 adjusted to the desired density in LB/MH broth and added to microtubes in a final volume of 50-100 111 μ L. ActivX™ TAMRA-FP (ThermoFisher Scientific) (at a concentration of 1 μ M) was added from 112 100X stock solutions in DMSO and the cells were incubated for 60 minutes at 37 °C and 300 RPM. 113 Following probe labeling, bacterial suspensions were transferred to 2 mL screw-cap tubes filled with 114 30-50 µL of 4x SDS-Loading buffer (comprising 40% glycerol, 240 mM Tris/HCl at pH 6.8, 8% 115 SDS, 0.04% bromophenol blue, and 5% beta-mercaptoethanol) and 60-100 µL of 0.1 mm glass 116 beads, and then lysed using bead-beating (3×30s, 6500 rpm, with 60s pause in-between) (Precellys® 117 Evolution homogenizer (Bertin Technologies) and centrifuge samples at 6000g for 5 min at 4°C to 118 remove the debris. 119 120 SDS-PAGE analysis of fluorescently labeled proteins 121 After adding 4x SDS sample buffer, the samples were subjected to boiling at 95 °C for 10 minutes 122 and subsequently separated using SDS-PAGE gel electrophoresis. The resulting gels were scanned 123 for fluorescence scanning in the Cy3 (532nm) channel, utilizing the Amersham™ Typhoon™ 5 124 imaging system (cytiva). 125 126 FP -biotin labeling of K. pneumoniae and sample preparation for mass-spectrometry 127 K. pneumoniae MKP103 cultures were grown on a blood plate for 24 hours and resuspended to an 128 OD600 ~20 in 3 mL MH broth. For each biological replicate, 1 mL aliquots were transferred to a 1.5 129 mL tube and either FP-Biotin (3 μ M) or DMSO was added, and the cells were then incubated for 60 130 min at 37°C and 700 rpm before samples were centrifuged at 4,500 ×g for 5 minutes at 4°C, and the 131 supernatant was removed. The cell pellets were resuspended in 1.2 mL RIPA Lysis buffer (50 mM 132 Tris, 150 mM NaCl, 0.1% SDS, 0.5% sodium deoxycholate, 1% Triton X-100) in 2 mL screw-cap 133 tube filled with ca. 200 µL of 0.1 mm glass beads and lysed by bead-beating (3×30s, 6500 rpm, with 134 60s pause in-between, performed two to three times with one min interval on ice). After 135 centrifugation for 5 minutes at 10,000 ×g at 4°C, the protein concentration in the supernatant was 136 adjusted to 1.0 mg/mL, and the proteins were stored at -20°C until the sample preparation. 137 For each sample, 50 μ L of streptavidin magnetic beads were washed twice with 1 mL of 138 RIPA lysis buffer. The streptavidin beads were then incubated with 1 mg of protein from each 139 sample in an additional 500 μ L of RIPA lysis buffer at 4°C overnight on a rotator set at 18 RPM. 140 After enrichment, the beads were collected using a magnetic rack and washed with RIPA lysis buffer 141 twice (1 mL, 2 minutes at room temperature), followed by a wash with 1 M KCl (1 mL, 2 minutes at 142 room temperature), a wash with 0.1 M Na 2CO3 (1 mL, ~10 seconds), a wash with 2 M urea in 10 143 mM Tris-HCl (pH 8.0) (1 mL, ~10 seconds), and two washes with RIPA lysis buffer (1 mL per wash, 144 2 minutes at room temperature). After the final wash, the beads were transferred to fresh protein Lo-145 .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 28, 2024. ; https://doi.org/10.1101/2024.05.28.596221doi: bioRxiv preprint Bind tubes with 1 mL of RIPA lysis buffer and washed three times with 500 µL of 4M Urea in 50 146 mM Ammonium bicarbonate (Ambic) with shaking for 7 minutes each time to remove nonspecific 147 enrichments. Lastly, the beads were washed three times with 500 µL of 50 mM Ambic with shaking 148 for 7 minutes, changing the tube between these washes. 149 For on-bead digestion, 150 μ L of 50mM Ambic, 3 μ L of 1mM CaCl2, 0.75 μ L of 1M 150 DDT, 4.5 μ L of 500mM IAA, and 6 μ L of MS-grade trypsin solution were added to the protein Lo-151 Bind tube, and samples were incubated at 37 °C overnight with a shaker running at 800 rpm. After 152 digestion, tryptic peptide digests were separated, and beads were washed with 70 µL of 50 mM 153 Ambic. For each sample, 20 μ L of formic acid was added to the combined eluates. The samples were 154 stored at -20°C until LC-MS/MS analysis. The sample preparation was conducted using the same 155

previously described32. 156 Liquid chromatography–mass spectrometry analysis 157 Sample cleanup and concentration were performed using OMIX C18 tip (A5700310, Varian). For the 158 LC-MS analysis, the concentrated samples were dissolved in 15 µl of 0.1% formic acid. Then, 0.5 µg 159 of peptides from each sample were injected for analysis. Peptide mixtures containing 0.1% formic 160 acid were loaded onto EASY-nLC1200 system (Thermo Fisher Scientific) with a C18 column (2 µm, 161 100 Å, 50 µm, 50 cm), and fractionation was performed using a 5-80% acetonitrile gradient in 0.1% 162 formic acid at a flow rate of 300 nL/min for 60 minutes. The fractionated peptides were analyzed 163 using a Orbitrap Exploris 480 mass spectrometer (Thermo Fisher Scientific). Data acquisition was 164 carried out in a data-dependent mode employing a Top20 method. Raw data was processed using 165 Proteome Discoverer 3.1 software with the CHIMERYS, and the fragmentation spectra were 166 matched against the ( Klebsiella pneumoniae subsp. pneumoniae KPNIH1, Taxonomy ID: 1087440) 167 database. A peptide mass tolerance of 10 ppm and a fragment mass tolerance of 0.02 Da were 168 employed during the search. Peptide ions were filtered using a false discovery rate (FDR) set at 5% 169 for accurate peptide identifications. To ensure precision at least three biological replicates were 170 conducted for all samples. Statistical analysis was conducted using Perseus software (version 2.03.0) 171 33. Potential contaminants, reverse hits, and proteins identified by side only were excluded. The 172 intensities from label-free quantification were converted using a log2 transformation. We applied 173 imputation based on a normal distribution (width, 0.3; down-shift, 1.8) to handle missing values. The 174 p-values were determined through a two-sided, two-sample t-test. Proteins identified as significantly 175 enriched serine hydrolases had at least a threefold change (equivalent to a log2 fold-change of 1.58) 176 and a minimum P value of 0.05 (corresponding to a −log10 value of 1.30). The mass spectrometry 177 proteomics data have been deposited to the ProteomeXchange Consortium via the PRIDE 34 partner 178 repository with the dataset identifier PXD052404. 179 180 181 Bioinformatic analyses 182 The 10 putative serine hydrolases were analyzed using both uniport 35 and the web-based InterPro 183 tool ( https://www.ebi.ac.uk/interpro/) to predict proteins structures. The subcellular localization of 184 serine hydrolases in K. pneumoniae was predicted with the web-based tools PSORTb v3.0 185 (https://www.psort.org/psortb/index.html)36. Furthermore, homology of K. pneumoniae serine 186 hydrolases was identified by querying the full-length protein sequences against the non-redundant 187 protein sequences database for gut commensal bacteria and Homo sapiens using Blastp with an E 188 value cutoff of 10-10 for all identified homologs. 189 190 191 192 .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 28, 2024. ; https://doi.org/10.1101/2024.05.28.596221doi: bioRxiv preprint Growth analyses 193 Growth curves were monitored using 96-well microtiter plates. Overnight cultures of K. pneumoniae 194 MKP103 WT and its transposon mutants were diluted 1:100 in fresh medium or in medium 195 supplemented with polymyxin B at a concentration of 8 µg/mL. A 200 µL aliquot of these diluted 196 cultures was transferred to each well of the 96-well plate as the initial culture. The plates were then 197 incubated at 37°C, and the optical density at 600 nm (OD600) was measured every 10 minutes using 198 a Synergy H1 Hybrid Reader (BioTek) or a Bioscreen plate reader. The area under growth curve 199 (AUC) was determined by using GraphPad Prism 9 for polymyxin B treatment. 200 201 Co-incubation of HT29-MTX cells and K. pneumoniae 202 The HT29-MTX human colorectal adenocarcinoma cell line 37 was grown in Dulbecco’s modified 203 Eagle medium (DMEM, supplied by Thermo Fisher) supplemented with 10% heat-inactivated (HI) 204 fetal calf serum (FCS). For coculture experiments, HT-29-MTX cells were seeded at a density of 205 approximately 1 x 10 5 cells per well in 48-well plates and mid-log phase K. pneumoniae MKP103 206 and SH-deficient mutants (with OD600 nm of 0.4) in FBS-free DMEM were added to each well at a 207 multiplicity of infection (MOI) of 20. After centrifugation at 200 x g for 5 minutes, the plates were 208 incubated at 37°C with 5% CO 2 for 2, 4, 6, and 8 hours and samples were collected from the media 209 above host cells for counting the colony-forming units per milliliter (CFU/mL). All experiments were 210 performed in duplicate and repeated independently at least three times. 211 212 Organoid line and growth conditions 213 The clonal human-derived colonic organoid cell line Pt15-70206 was used to grow colonic 214 epithelium, following the protocol described by Vonk et al. 38 . Briefly, colonic organoid stocks were 215 thawed, and the organoids were cultured in Matrigel domes with medium containing 15% Advanced 216 DMEM/F12, 1x Glutamax, 100 U/mL Penicillin-Streptomycin, 10 mmol/L HEPES (all Invitrogen), 217 25% Rspo1 Conditioned Medium, 10% Noggin Conditioned Medium, 0.5nM Wnt Surrogate-FC 218 Fusion Protein (ImmunoPrecise-UCN001), 2% B27 (Invitrogen), 1.25 mM N-acetylcysteine, 10 mM 219 Nicotinamide, 3μ M p38 inhibitor SB202190 (all Sigma-Aldrich), 50 ng/mL mEGF (Invitrogen), and 220 0.5 μ M A83-01 (Tocris). 221 After an average of 4 passages, the organoids were disrupted and used to seed transwells 222 (corning, 3470-clear) to generate confluent colonic epithelium. During the first 24 hours in 223 transwells, the organoid culture medium was supplemented with 10nM rock inhibitor Y-27632 224 (Sigma-Aldrich). Once the Trans-epithelial Electrical Resistance (TEER) surpassed 100 Ω /cm2, the 225 culture medium was replaced with differentiation medium 39, excluding Wnt Surrogate-Fc Fusion 226 Protein, nicotinamide, and p38 inhibitor. Differentiation medium was refreshed 24 hours before the 227 adding bacteria for co-culturing, in absence of Penicillin-Streptomycin. 228 229 Co-incubation of organoids and K. pneumoniae 230 Overnight K. pneumoniae MKP103 and its SH-deficient transposon mutants were diluted 1/100 in 231 fresh medium and cultured in LB at 37°C until reaching the exponential growth phase with an optical 232 density at 600nm of 0.3-0.4. Subsequently, the bacterial suspension was washed in PBS and 233 resuspended in PBS to an OD600 of 0.4. The bacterial suspension was further diluted in organoid 234 medium to achieve~2.5× 10 6 CFU/mL when adding them to the 2D organoid monolayers. The 235 transwells containing the organoids and bacteria were spun down at 250 x g for 2 minutes to ensure 236 cell contact. Finally, the co-culture was incubated at 37°C with 5% CO 2 for 2, 4, 6, 8, and 16 hours 237 and samples were collected from the apical side supernatant for counting the CFUs. To evaluate 238 bacterial replication on the epithelial surface, the monolayer tissue and membrane were removed 239 from the insert and dissociated using 2 mm glass beads in ice-cold PBS through 30 s of vortexing. 240 .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 28, 2024. ; https://doi.org/10.1101/2024.05.28.596221doi: bioRxiv preprint Alternatively, after removal from the insert, the cells were lysed using 0.2% Triton X-100. The 241 resulting PBS suspension was serially diluted and plated on LB agar. The agar plates were incubated 242 overnight at 37°C, and CFUs were counted. In parallel, samples were taken from the basolateral 243 compartment, serially diluted, and plated to assess any breach of the monolayer. The CFUs were 244 manually enumerated and expressed as CFU/ml. The epithelial cells were examined for damage with 245 the EVOS FL Auto 2 fluorescent microscopes (ThermoFisher Scientific) at 16 hours. 246 247 Visualization of 2D organoid cultures 248 Organoid 2D cultures were fixed with Methacarn fixation (Methanol-based Carnoy fixation), which 249 causes protein denaturation40, consisting of 60% methanol, 30% chloroform, and 10% glacial acetic 250 acid, for two hours. After washing away the fixative with 100% methanol, the samples were 251 rehydrated using different ethanol dilutions (100%, 90%, 80%, 70%, 60%, 40%, 20%), with each 252 step lasting ten minutes. The filter of the transwells, with the organoids, was separated from the insert 253 and blocked and permeabilized with a blocking buffer containing 2% Bovine Serum Albumin (BSA) 254 (Serva) and 0.1% saponin (Sigma Aldrich) before incubating with the primary and secondary 255 antibodies at 4°C in dark. The primary antibodies used were Anti-MUC1 21D4 (Sigma Aldrich, 05-256 652), Anti-MUC2 (Abcam, ab76774), Anti-MUC13 (generous gift by Karin Strijbis), Anti-Sox9 257 (Merck Millipore, AB5535), Anti-Villin 1D2C3 (Santa Cruz, sc-58897), and Anti-lysozyme (Dako, 258 A099). Then, organoids were stained with Phalloidin Atto-647 (65906-10NMOL) and DAPI and 259 imaged with an SP5 II confocal microscope (Leica TCS). 260 261 Membrane permeabilization 262 Bacteria cultured overnight were diluted 1/100 in fresh medium and grown to an OD600 of 0.4–0.5 263 at 37°C with shaking. The bacteria were washed with RPMI with 0.05% human serum albumin 264 (HSA, Sanquin, The Netherlands) through centrifugation at 10,000×g for 2 minutes and resuspended 265 to an OD600 of 0.5 in RPMI. For the assay, bacteria were adjusted to an OD600 of 0.05 and mixed 266 with 10 % normal human serum (NHS) (serum was prepared as described in 41) in the presence of 1 267 µM SYTOX Green Nucleic Acid stain (ThermoFisher) in RPMI. This mixture was incubated at 37°C 268 with shaking. Fluorescence was measured using a CLARIOstar microplate reader (Labtech) at an 269 excitation wavelength of 490–14 nm and an emission wavelength of 537–30 nm. 270 271 Protein Expression and Purification 272 The full length yjfP, yqiA, and pldB (Uniprot 35 IDs A6THA0, A6TE20, and A6TGK5 ) DNA 273 sequences were codon optimized for E. coli expression (IDT and TISIGNER 42) and synthesized by 274 IDT with overhangs for ligation-independent cloning 43. The synthesized DNAs were cloned into a 275 modified pET28a-LIC vector incorporating an N-terminal His 6-tag and a 3C protease cleavage site. 276 Plates were incubated at 37 °C overnight and single colonies were used to inoculate a starter culture 277 of 5 mL LB supplemented with 50 µg/mL kanamycin. The starter culture was used to inoculate 1 L 278 of LB media supplemented with 50 µg/mL kanamycin and incubated at 37°C while shaking at 279 180 rpm until the OD 600nm reached 0.5. The cultures were incubated at 18 °C while shaking at 180 280 rpm and recombinant protein expression was induced with 1 mM IPTG. Cell pellets were harvested 281 the following day by centrifugation at 5,000 × g for 30 minutes. Cell pellets were either stored at -282 20°C or lysed right away for protein purification. 283 A cell pellet was resuspended in lysis buffer (50 mM Tris pH 8.0, 300 mM NaCl, 50 mM 284 imidazole, 10% sucrose, 10% glycerol) and incubated on ice for 10 minutes with 400 µg lysozyme 285 and 200 µg DNase. The cells were subsequently lysed via sonication in an ice bath using Sonifier 286 Heat Systems Ultrasonics, for 5 min using a one second pulse mode. Lysate was cleared by 287 .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 28, 2024. ; https://doi.org/10.1101/2024.05.28.596221doi: bioRxiv preprint centrifugation at 15,000 x g for 30 minutes. The supernatant was loaded onto 2.0 mL Ni2+-NTA resin 288 which was pre-washed with lysis buffer. The resin-supernatant mixture was incubated at 4 °C room 289 for 30 min before the resin was washed in lysis buffer. The recombinant protein was eluted via 7 mL 290 of elution buffer (50 mM Tris pH 8.0, 300 mM NaCl, 300 mM imidazole, 10% sucrose, 10% 291 glycerol). For PldB and YjfP, these purified His6-tagged proteins were used for biochemical analysis 292 described below. To cleave the N-terminal His 6 tag for protein crystallization studies, the eluted 293 fractions of YqiA and YjfP incubated with 5 mM DTT and 3C protease overnight at 4 °C. 294 Recombinant proteins of YjfP and YqiA were finally purified by size exclusion chromatography (gel 295 filtration). Protein samples were concentrated using 10,000 MWCO spin columns to approximately 296 250 µL and injected onto a size exclusion column (Superdex 75 Increase 10/300 GL column, GE 297 Life Sciences) preequilibrated with size exclusion buffer (25 mM Tris pH 8.0, 150 mM NaCl). 298 Protein purity was assessed by SDS-PAGE and nanodrop. Samples were concentrated to ~10 mg/mL 299 and snap frozen in liquid nitrogen or used immediately for crystallization. 300 301 Biochemical assay of PldB, YjfP, and YqiA 302 Activity tests of purified serine hydrolases on different 4-methylumbelliferyl(MU) substrates were 303 carried out as previously described 44. Briefly, the reaction was carried out in PBS with 0.02 % v/v 304 Triton X-100 (Fisher Scientific, Fairlawn, NJ). 8 µL enzyme solution (0.625 nM His 6-PldB, 0.625 305 nM His6-YjfP, 62.5 nM YqiA (tag-free)) was mixed with 2 µL 5x 4-MU substrate stock solutions 306 (250 µM) to initiate reactions and then the fluorescence signal ( λ ex= 335 nm and λ em= 450 nm) was 307 monitored by Cytation 3 Multi-Mode Reader (BioTek, Winooski, VT) at 25 ºC for 1 hour. Turnover 308 rates of the 4-MU substrates in the linear phase of the reaction were calculated in GraphPad Prism9 309 as RFU/second. The reaction rates were corrected by subtraction of the background hydrolysis in the 310 absence of proteins for each substrate. The substrate preference profile was generated by dividing the 311 corrected reaction rates by enzyme concentrations in each assay. 312 313 Protein Crystallisation 314 YjfP and YqiA were crystallized using sitting drop vapour diffusion and screened for crystallization 315 against several broad screens. 0.2 µL of protein solutions at 10 mg/mL were mixed with 0.2 µL of 316 mother liquor. YjfP appeared as needle-like crystals in several different conditions, however the final 317 structure of YjfP was determined from cubic crystals grown in 0.1 M ammonium sulfate, 35 % w/v 318 PEG 8,000, 0.1 M sodium acetate pH 5.0. A YjfP crystal from this condition was soaked in mother 319 liquor supplemented with 25 % v/v ethylene glycol for one minute before freezing in liquid nitrogen 320 for analysis at the Australian Synchrotron. YqiA crystallized in several conditions that mostly 321 contained either calcium chloride or calcium acetate, a crystal grown in 0.3 M calcium chloride 322 hexahydrate, 10 % w/v PEG 6,000, 0.1 Tris pH 8.0 led to successful structure determination of YqiA. 323 This crystal was soaked in mother liquor supplemented with 25 % v/v glycerol for one minute, 324 before being frozen in liquid nitrogen. 325 326 Crystal data collection and Processing 327 Protein crystals were subject to x-ray diffraction at the Australian Synchrotron MX2 beamline 9. 328 Datasets were processed in XDS 10, before being merged and scaled using AIMLESS 11 in the CCP4 329 suite. Phases were determined by molecular replacement in Phenix Phaser 12 using an AlphaFold 330 model13 of YqiA or YjfP prepared with Phenix Process Predicted Model 14. The output model 331 underwent iterative building and refinement using COOT15 and Phenix16. 332 333 .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 28, 2024. ; https://doi.org/10.1101/2024.05.28.596221doi: bioRxiv preprint Infection of K. pneumoniae in G. mellonella larvae 334 Galleria mellonella larvae were obtained from Reptilutstyr AS (Norway). For infection experiments, 335 K. pneumoniae MKP103 and its transposon mutants were cultured overnight in LB broth and 336 harvested by centrifugation at 4,500 × g for 10 min at 4°C followed by one wash with PBS. The 337 bacteria were then diluted in PBS to achieve OD600 of 1, approximately 1 × 10 9 CFU/ml. Each 338 experimental group consisted of n=10 larvae. For infection, 10 µl of a bacterial suspension adjusted 339 to 1 × 10 5 CFU/ml in PBS was injected into right hid proleg using a 30G syringe microapplicator 340 (0.30 mm (30G) × 8 mm, BD Micro-Fine demi). A separate group of 10 larvae were injected with 10 341 µl of PBS to ensure that death was not due to injection trauma as negative control. The larvae were 342 placed in 9.2 cm Petri dishes and incubated at 37°C in the dark and survival was monitored for 5 343 days. Each experiment was independently replicated at least three times. 344 345 346 Statistical analysis 347 348 Graphs and statistical analyses were produced using GraphPad Prism 9. Unless specified otherwise, 349 all experiments were conducted using three independent biological replicates. Data for graphs are 350 shown as mean ± standard deviation (SD). To compare wild-type (WT) to mutant strains, either one-351 way or two-way ANOVA tests were utilized, followed by Dunnett’s multiple comparisons test, using 352 the WT strain as the control. Significance levels were denoted as follows: *P < 0.05, **P < 0.01, 353 ***P < 0.001, and ****P < 0.0001.354 .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 28, 2024. ; https://doi.org/10.1101/2024.05.28.596221doi: bioRxiv preprint 355

356 357 Identification serine hydrolases in K. pneumoniae 358 359 We first set out to determine global serine hydrolase (SH) activity profile in K. pneumoniae when 360 grown on blood agar using the fluorescent ABP FP-TMR. We selected strain MKP103, a derivative 361 of the multi-drug resistant clinical outbreak strain KPNIH1 for which the carbapenemase-gene has 362 been deleted45. KPNIH1 belongs to sequence type ST258 which can cause life-threatening infections 363 in hospitalized individuals and poses a considerable risk to spread multidrug resistance into the 364 community46 47. After labeling, the cells were lysed, and the labeled proteins were separated via 365 SDS-PAGE analysis and visualized by in-gel fluorescence scanning. The results revealed that several 366 bands were predominantly labeled at the 1 µM probe concentration ( Fig. 1C). To identify these SH 367 targets, we then used a biotiynylated probe (FP-biotin) with a chemoproteomic workflow using 368 streptavidin-enrichment and LC-MS/MS analysis (Fig. 1 Aiii, B). Using this approach, we identified 369 10 K. pneumoniae SHs that displayed significant enrichment (p-value 3.5-fold) 370 when compared to a control dataset that was not treated with FP-biotin (as shown in Fig. 1B. E and 371 detailed in Supplementary Table 2 and Extended Dataset 1, data are available from 372 ProteomeXchange with identifier PXD052404). Of the 10 identified SHs, 8 were predicted to possess 373 α ,β -hydrolase domains, while the remaining two were annotated as serine proteases with Trypsin-like 374 peptidase domains ( Table 1 ). Most of these enzymes are generally not well characterized in K. 375 pneumoniae and for many of these enzymes their cellular functions are ill defined. Of the 10 SHs, 376 only the HtrA-protease DegP and the carboxylesterase BioH enzymes have been studied functionally 377 in K. pneumoniae. The latter putatively hydrolyze the pimeloyl-acyl carrier protein methyl ester as 378 one of the first steps of biotin biosynthesis. 48 Sequence-based bioinformatic prediction of the 379 subcellular location of the 10 identified SH proteins using PSORTb v3.0 36 suggests that only the 380 patatin-like phospholipase YchK is presumably secreted, whereas the two HtrA-like proteases, DegP 381 and DegQ, were predicted to localize to the periplasmic. YjfP and YqiA were predicted to localize to 382 the cytoplasmic. YbIF and YcfA are unknown. PldB is on the cytoplasmic membrane, and BioH and 383 CatD to the cytosol. Except for DegP, DegQ, YchK and YqiA, most of the K. pneumoniae SHs 384 displayed limited or no homology with serine hydrolases from 19 other gut commensals and human 385 (Fig. 1E, Supplementary Table 3, Extended Dataset 2). 386 To functionally validate the identified target enzymes, we retrieved transposon mutant 387 strains with insertions in individual serine hydrolase genes ( yfbF, yjf P, yqi A, ych K, deg P, pldB, 388 degQ, and catD) from the Manoil Lab Transposon Mutant Library 45. The library did not contain 389 mutants in bioH or ycf P. At least for BioH with its presumable role in the biosynthesis of biotin, the 390 absence of viable transposon mutants might indicate essentiality. We first labelled the obtained 391 transposon mutants with FP-TMR labeled protein bands in the gel-based FP-TMR labeling. This 392 allowed us to confirm the identities of pldB, degP, ychK, and ybfF, but not degQ, yjfP, and ycfF, as 393 their corresponding transposon mutants showed similar FP-TMR profiles as the WT ( Fig. 2A). This 394 lack of labeling in the gel based ABPP assay suggesting that some of the identified enzymes may not 395 be abundant and can only be detected in the more sensitive MS assay. In addition, five fluorescent 396 protein bands could not be assigned to the MS identified serine hydrolases, suggesting that 397 differences in selectivity or cellular permeability of the biotinylated and fluorescent probes might 398 have prevented them from being detected in the proteomics approach ( Fig. 2B ). Based on this 399 outcome, a set of five targets were selected for further functional studies: YfbF, YchK, PldB, DegP 400 .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 28, 2024. ; https://doi.org/10.1101/2024.05.28.596221doi: bioRxiv preprint and the uncharacterized esterase YjfP which lacks homologs outside of the Enterobacteriales and 401 humans (Fig. 1E, Supplementary Table 3) 402 As a first step, we investigated the growth of the selected transposon mutants in LB, a rich 403 liquid media and found that ychK, pldB, yjfP, degP and ybf F were all dispensable for growth ( Fig. 404 2C). Since the primary niche for K. pneumoniae as a commensal is the human gut, we next evaluated 405 growth fitness using in vitro models mimicking the host-interface of the gut. We adopted a co-culture 406 model with HT29-MTX cells, a mucus-rich goblet cell line 49. During an 8 h co-culture period with 407 HT29-MTX cells in DMEM/10% HI-FBS, three of the tested transposon mutants ( pldB, ychK, yjfP), 408 showed a notable decrease in fitness compared to the WT MKP103, degP:Tn and ybf F:Tn strains 409 (Fig. 2D). This defect was evident both as a delayed onset of replication in the co-culture model and 410 a reduced growth rate. In comparison, neither WT MKP103, nor any of the Tn-mutants was able to 411 g r o w i n D M E M (Fig. 2E ), suggesting that interactions with the host cells are crucial to sustain 412 growth in this model and may be key for understanding these phenotypes. 413 414 PldB, YchK, and YjfP are important for the initial stages of infection in the gut 415 To determine whether the lysophospholipase (PldB), patatin-like lipase (YchK), and putative esterase 416 (YjfP), play important roles in host-pathogen interactions in the gut, we used a more physiologically 417 relevant model system that better accounts for the complexity of human colonic epithelia. This model 418 is a 2D-co-culture model using a colonic organoid monolayer derived from human adult tissue stem 419 cell-derived organoids using a previously published protocol 38 ( Fig. 3A ). Unlike 3D organoids 420 which have enclosed apical sides, the monolayered organoids directly exposed their apical sides to 421 the culture medium, facilitating direct interactions between the epithelial surface and the 422 administered bacteria50,51. Cellular differentiation was assessed by immunofluorescence microscopy 423 and confirmed the presence of the main cell types characteristic of mature/differentiated human 424 colonic epithelium: mucin-1, mucin-2, and mucin-13 cells, (identified by anti-mucin antibodies, note 425 that mucin-13 was detected with a polyclonal antiserum that might also stain other mucins) ( Fig. 426 3B), enterocytes (recognized by anti-villin) and stem cells (anti-sox9 antibody). We also detected 427 lysozyme C (by anti-lysozyme C antibody) that is produced by Paneth cells, even though this cell 428 type is usually present in the small intestine52 (Fig. 3C). 429 To develop a colonization/infection model of K. pneumoniae in these 2D-colonic organoids, 430 we added bacteria to the apical side of the monolayers and assessed bacterial growth by determining 431 CFU numbers over time. We observed that the CFU counts of WT bacteria first dropped by an order 432 of magnitude after 2 hours of co-culture compared to the inoculum. This finding could be due to the 433 bactericidal activities of organoid-derived antimicrobial peptides. After 2 h WT bacteria started 434 growing exponentially and approached a plateau at 8-16 hours (Fig. 3D). Since the bacteria were not 435 able to grow in organoid media alone (Supplementary figure S1), we assume that access to host-cell 436 derived nutrients is necessary to sustain bacterial growth in the co-culture model. After 16 h of co-437 culture, bacteria have translocated to the basal side of the transwell chamber suggesting damage of 438 the epithelial barrier (Fig. 3F). 439 Upon co-culture with the organoid monolayer, the pldB:Tn, ychK:Tn, and yjfP:Tn mutants 440 showed fitness defects compared to WT bacteria. The initial drop in CFU at 2 h is more pronounced 441 in pldB:Tn, ychK:Tn and yjfP:Tn cells compared to WT (Fig. 3D) Furthermore, we observed that the 442 outgrowth of these transposon mutant cells in the culture supernatant was delayed, but once they 443 have resumed growth, the overall rate of growth of the transposon mutants was similar to that of WT 444 cells. ( Fig. 3D ). From 4 h of co-culture onwards, we also observed a lower number of bacteria 445 .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 28, 2024. ; https://doi.org/10.1101/2024.05.28.596221doi: bioRxiv preprint attached to the organoid monolayer for the three tested mutants compared to WT ( Fig. 3E). After 16 446 hours of culturing, the yjfP:Tn mutants showed similar numbers of bacteria at the basal side as the 447 WT despite reduced number of bacteria at the apical end. For pldB:Tn and ychK:Tn strains there was 448 a trend towards lower bacterial numbers in the basal compartment that may be indicative of reduced 449 or delayed translocation (Fig. 3F). 450 451 Sensitivity of serine hydrolases-deficient transposon mutants to cell envelope stressors 452 453 We hypothesized that the reduced CFU levels observed in the mutants compared to WT ( Fig. 3 ) 454 could be explained by an increased sensitivity to the organoid-produced antimicrobial peptides 455 (AMPs). To test this hypothesis, we evaluated the susceptibility of the different mutant strains to 456 polymyxin B, a membrane-destabilizing AMP that interacts with lipopolysaccharide (LPS) 53 . We 457 observed an increased susceptibility to polymyxin B for both pldB and yjf P as well as for the degP 458 transposon mutant, whereas the mutant deficient in the secreted patatin-like phospholipase ychK had 459 a similar susceptibility to the WT (Fig. 4 A,B). 460 In the Gram-negative cell envelope, LPS is not the only the target of AMPs like 461 polymyxin, but it also plays a crucial role in resistance to complement-mediated killing 54. A degP 462 mutant was previously reported to be more sensitive to complement-mediated killing than the parent 463 strain55. To test the effects of complement mediated killing, we treated each of the mutant lines with 464 serum and heat-inactivated serum in which components of complement are inactivated. We found 465 that serum induced higher killing effect in the pldB:Tn, yjfP:Tn and deg P:Tn mutants compared to 466 control or treatment with heat killed serum ( Fig. 4C, D) . These results suggest a role of YjfP and 467 PldB in shaping envelope integrity, which in turn may affect susceptibility to complement and 468 antimicrobial peptides. 469 470 Loss of PldB, YchK, and YjfP reduced K. pneumoniae virulence in G. mellonella larvae 471 472 The Galleria mellonella infection model ( Fig. 5A ) is a simple, cost-effective and easily accessible 473 animal model that allows us to assess the effects of serine hydrolases on virulence 56,57. We found 474 that upon infection with 10 5 CFU WT K. pneumoniae , approximately 75% of G. mellonella larvae 475 died within 2 days, whereas larvae infected with the pldB:Tn, ychK:Tn, and yjfP:Tn strains showed 476 significantly higher survival rates (p /i5 </i5 0.0001) (Fig. 5B) and approx. 65% of the larvae remained 477 alive even after 5 days. The ybfF:Tn and y qiA:Tn mutants showed similar virulence as the WT 478 K.pneumoniae MKP103 whereas , the degP:Tn mutant displayed reduced virulence compared to 479 WT. 480 However, phenotypic results achieved by testing of the transposon mutants might be 481 confounded by secondary mutations or polar effects. To rule out secondary mutations that may 482 account for these phenotypes, we retrieved additional mutants from the Manoil K. pneumoniae 3-483 allele transposon mutant library 45 representing individual clones where the transposons were 484 localized within the same gene, but at a different location. Assessment of the virulence of different 485 transposon mutants with insertions in pldB (n=3 available strains), yjfP (n=3) and ych K (n=2) 486 (Supplementary Figure S2 ) all showed similar phenotypes, suggesting that transposon insertion in 487 the respective gene accounts for the observed phenotype rather than secondary mutations. In 488 conclusion, assessment of virulence of the ychK, pldB, and yjfP mutants supports a role of these 489 enzymes in virulence of K. pneumoniae. 490 491 Biochemical and structure characteristics of YjfP and other serine hydrolases 492 493 .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 28, 2024. ; https://doi.org/10.1101/2024.05.28.596221doi: bioRxiv preprint Of the seven uncharacterized serine hydrolases that we identified in our ABPP proteomics studies, 494 we were able to purify YjfP, YqiA, and PldB ( Supplementary figure S3 ). To assess the substrate 495 preference of recombinantly expressed and purified YjfP, YqiA, and PldB, we examined their ability 496 to cleave various commercially available fluorogenic substrates ( Fig. 6 A, B, C ). Our results 497 indicated that YjfP effectively hydrolyzes saturated lipid esters with chain lengths from C2 to C4, 498 showing a strong preference for the short-chain C2 acetate substrate ( Fig. 6A ). Y qiA cleaved a 499 variety of artificial substrates ranging from C2 to C10, with the highest activity observed for the C7 500 heptanoate substrate ( Fig. 6B). PldB hydrolyzed saturated lipid esters ranging in chain length from 501 C2 to C8, showing a particularly strong preference for the C2 and C4 chain lengths (Fig. 6B). 502 We also determined the crystal structures of YjfP and Y qiA to 1.3 Å and 1.5 Å, respectively 503 (Fig. 6D, E, F, G, Suppl. Tables S4 and S5, Suppl. Figures S5, S6 ), but we were unable to 504 crystalize purified PldB protein. YjfP has the canonical α /β -hydrolase fold which is typical of serine 505 hydrolases58, with a catalytic triad conserved at residues Ser116, Asp187, and His221 ( Fig. 6D). An 506 8-stranded β -sheet is central to the α /β -hydrolase fold, sandwiched between α -helices 1 and 8, and 507 α -helices 2-7 (Supplementary Figure S5). Serine hydrolases often contain a flexible lid region 508 between α -helices 4 and 5 that mediates substrate specificity59. However, this is absent in YjfP , which 509 instead has a short loop connecting α -helices 4 and 5. Consistent with our gel filtration results 510 (Supplementary Figure S4 ), the asymmetric unit of the YjfP crystal contains a homodimer that 511 creates a closed cavity around the catalytic triad. Submission of the YjfP structure to PISA, an online 512 tool to analyse protein interfaces, predicts the dimer interface to be biologically relevant, scoring a 513 maximum CSS of 1.0 60. The dimer interface covers 11 % of each protein surface area and is 514 mediated mostly through β -strand 1 and α -helix 1. At the top of this interface, Asp188 and Arg220 515 from one chain form salt bridges with Arg220 and Asp188 from the second chain. These residues are 516 on the same loops as Asp187 and His221 from the catalytic triad, forming a roof above the putative 517 substrate binding site. Inside the substrate binding cavity there is discontinuous density, potentially 518 corresponding to a cleaved substrate that may have co-purified ( Fig. 6E). This density approaches a 519 hydrophobic face within the cavity that would not accommodate longer carbon chains than butyrate, 520 explaining the sharp drop-off in activity for C7 and longer chain substrates (Fig. 6A). 521 Similar to YjfP, YqiA has a canonical α /β hydrolase fold58, with a conserved catalytic triad 522 of residues Ser69, Asp147, and His172 ( Fig. 6F). However, there are several features in YqiA that 523 differ from YjfP. Firstly, the asymmetric unit contains a monomer, consistent with gel filtration 524

(Supplementary Figure S4), and PISA predicts there are no crystal contacts that are likely to 525 create a dimer60. Unlike YjfP, the YqiA active site is much more open despite the presence of a small 526 lid region. Within the oxyanion hole there is a chlorine atom and a hydrophobic face is formed 527 around the active site ( Fig. 6G ). This hydrophobic, open face likely directs substrate specificity 528 towards hydrophobic carbon chains. Since the lid region is small and removed from the active site, it 529 might not accommodate the 4MU leaving group of the fluorogenic substrates well, which might 530 explain the much lower activity rates of YqiA compared to YjfP. 531 Together, the substrate profiling and structural characterization of YjfP and YqiA suggest they 532 are short-chain esterases. the YjfP structure showed that an acetate group would fit well into the 533 small hydrophobic pocket near the active site Ser116, whereas chains longer than butyrate would not, 534 thus giving an explanation for the observed substrate selectivity (Fig. 6A). Our data suggest that YjfP 535 may function as a deacetylase.536 .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 28, 2024. ; https://doi.org/10.1101/2024.05.28.596221doi: bioRxiv preprint 537

538 539 Serine hydrolases play important roles in various biological processes of both bacteria and host cells 540 yet our understanding of these enzymes in K. pneumoniae remains limited. Using a chemical 541 proteomic approach, we have identified 10 previously uncharacterized serine hydrolases. Using 542 transposon mutants of the hydrolases PldB, YchK, and YjfP we observed growth defects in co-543 culture models with HT29-MTX cells, human colonic organoids and in a Galleria mellonella 544 infection model, suggesting putative roles in both gut colonization and infection. In the G. mellonella 545 infection model, alternative clones with transposon insertion in the same gene displayed the same 546 phenotype suggesting its specificity. 547 Based on previous annotation and bioinformatic analysis, the previously uncharacterized 548 PldB, is a lysophospholipase L2 located at the cytoplasmic membrane. YjfP is annotated as a 549 putatively cytosolic esterase. YchK which possesses a patatin-like phospholipase domain is predicted 550 to be secreted. Of note, despite similarities of the phenotypes in the complex organoid and G. 551 mellonella infection models, we observed some important differences that might give clues about the 552 underlying molecular functions of these enzymes. Both the pldB and yjfP-deficient strains showed an 553 increased susceptibility to AMPs, whereas the ychK mutant did not. 554 AMPs are short amphiphilic peptides that, in general, act by disrupting bacterial 555 membranes. Polymyxin B was shown to exert its membrane-destabilizing effect by binding to the 556 lipid A anchor of LPS in the outer membrane of Gram-negative bacteria 61. AMPs produced by 557 human cells, e.g., the defensins, are considered to act by similar mechanisms. In the human gut, the 558 specialized secretory Paneth cells in the small intestine produce alpha-defensins (as reviewed by 559 Bevins and Salzman62), whereas epithelial cells contribute through production of beta-defensins (see 560 review by Gallo and Hooper 63). In our colonic stem-cell derived organoid model, the presence of 561 Paneth cells was indirectly detected, and we therefore assume that bacteria in the organoid-model are 562 exposed to both alpha- and beta-defensins. Human beta-defensin 1 is expressed by HT29 cells 64 and 563 thus may affect bacterial growth in the HT29-MTX co-culture model. Of note, AMPs are also 564 important effectors of insect innate immunity (as reviewed by Stacek 65), and G. mellonella has been 565 shown to mount pathogen-specific AMP responses upon infection with diverse bacteria and fungi 66. 566 Thus, an increased susceptibility to AMPs could be detected in both organoid and HT29-MTX co-567 culture models as well as in the G. mellonella infection model. 568 The most pronounced increase in polymyxin B-sensitivity was observed for the degP-569 mutant, which is consistent with a previous study 27. This mutant, however, did not show any fitness 570 defect in the HT29-MTX model (and was hence not tested in organoids) and the reduction in 571 virulence in the G. mellonella model was less pronounced (and less stable across different clones) 572 compared to the pldB and yjf P mutants. This discrepancy could be explained by differences in the 573 susceptibility to polymyxin B compared to the AMPs that are endogenously produced in the co-574 culture and infection models. However, this outcome might also indicate that a reduced susceptibility 575 to AMPs might contribute to the phenotypes in the co-culture and infection models, but on its own is 576 not sufficient to explain them. It is possible that the putative deficiency in cell envelope integrity in 577 the pldB and yjf P mutants leads to multiple downstream effects that collectively contribute to 578 reduced virulence and fitness. 579 We can only speculate about by which mechanism PldB and YjfP may affect cell envelope 580 integrity, but roles in membrane remodeling, tailoring modifications of LPS cell wall components, or 581 posttranslational modifications of proteins that are involved in cell wall and LPS synthesis could be 582 plausible. Of note, secondary acylation of LPS in K. pneumoniae affects susceptibility to polymyxin 583 B and several other AMPs 67, suggesting that remodeling of the lipid A anchor through previously 584 .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 28, 2024. ; https://doi.org/10.1101/2024.05.28.596221doi: bioRxiv preprint unidentified hydrolases or transferases could affe ct cell membrane integrity and AMP susceptibility. 585 However, cleavage of lipid substrates with longer chain fatty acid esters does not appear a likely 586 physiological function of YjfP. The narrow substrate profile and structural characterization of this 587 enzyme suggest that is a deacetylase, that may act on a small molecule, peptide or protein. PldB, in 588 contrast accepts a broader range of substrates cleaving synthetic fluorogenic substrates up to C8 589 chain length. The interesting phenotypes associated with these SH mutants and the architecture of the 590 YjfP (and YqiA) active sites, warrant further investigation into the native substrates of each enzyme. 591 The pldB and yjfP mutant also showed increased sensitivity to complement killing. Since 592 complement activation involves pore formation in the outer membrane and complement resistance is 593 mediated by the protective effects of LPS or capsule polysaccharide 68, these results support the 594 notion that PldB and YjfP might also be partially responsible for the modulation of LPS production 595 and/or constructing the outer membrane. Recently, another chemoproteomic study identified 596 BT4193, a homolog of human dipeptidyl peptidase 4, in the Gram-negative gut commensal 597 Bacteroides thetaiotaomicron 23. Due to its periplasmic location, the authors suspected a role in cell 598 envelope integrity and observed that a BT4193 mutant displayed increased sensitivity for killing by 599 both vancomycin and polymyxin B. However, treatment of WT cells with a BT4193 inhibitor did not 600 phenocopy these deficiencies, so the authors concluded that the phenotype of the ΔBT4193 mutant 601 might be caused by a structural role of this protein rather than by its enzymatic function 23 . Whether 602 the putative role of PldB and YjfP on AMP and complement susceptibility will be due to its 603 enzymatic functions as a lipases or esterase or due to structural roles, remains to be determined. 604 In contrast to the cell-membrane associated putative lysophospholipase PldB and putatively 605 cytosolic deacetylase YjfP, the putative phospholipase YchK is predicted to be secreted. As may be 606 expected for a secreted phospholipase, the YchK-mutant did not show any phenotype related to 607 susceptibility to AMP and envelope integrity. Secreted lipases have been described as virulence 608 factors for a number of bacterial pathogens and their hydrolytic function mostly associated with the 609 destruction of defense barrier and toxicity 69-71. One interesting observation that we made is that K. 610 pneumoniae is unable to grow in organoid media alone or in the DMEM medium alone used in the 611 HT29-MTX culture. This means that to sustain growth in the organoid co-culture model, bacteria 612 need to access nutrients from the host cell layers. Both the colon organoid and HT29-MTX 613 monolayer produce a pronounced mucus layer. In the gut, the mucus layer, which is predominantly 614 composed of glycoproteins, is an essential part of human colonic epithelia and acts as the primary 615 defense against bacteria residing in the colon 72-74. In the human gut, the mucus layer of colonic 616 epithelia has a surface layer of phosphatidylcholine (PC) which contributes to barrier function and 617 immune regulation 75. Bacterial phospholipases in this layer can convert PC into lyso-PC, thereby 618 disturbing the integrity and protective function of the mucus structure and enabling damage to 619 epithelial cells 75-77. For the gastric pathogen H. pylori secreted lipases are responsible for breaching 620 a mucus-associated defensive phospholipid layer, allowing the pathogen to partially invade the 621 gastric mucus 78,79. A similar role is plausible for the putatively secreted patatin-like phospholipase 622 YchK of K. pneumoniae. Its hydrolytic activity could enable either direct access to phospholipid-623 derived nutrients or contributing to the break-down of defense barriers that might facilitate cell 624 invasion and extraction of nutrients. However, the origin of the surface layer of PC in the human 625 colon is unclear (Review by 75) and it might be derived from the ileum or jejunum. It thus remains to 626 be determined if a similar PC layer is in fact found in our organoid model. An alternative role for 627 YchK at the host-pathogen interface in the organoid model as well as in Galleria infection model 628 could be that it directly acts on host cell membranes. 629 Our results indicate that PldB, YjfP and YchK may be promising targets for further 630 evaluation as anti-virulence targets. One of the last-resort antibiotics for treatment of infections with 631 carbapenemase producing K. pneumoniae is the AMP colistin 80 . The emergence of colistin-632 .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 28, 2024. ; https://doi.org/10.1101/2024.05.28.596221doi: bioRxiv preprint resistance in the clinic is therefore of great concern 81 and strategies to reverse colistin resistance are 633 being explored 82. The sensitization to AMPs observed for pldB, yjfP and deg P mutants is a 634 significant finding and we suggest that inhibitors of either of these enzymes should be tested for 635 synergy with colistin and other AMP antibiotics. 636 Another potential drug target that was identified in our dataset but was not validated further, 637 is the carboxylesterase BioH. This enzyme has a putative role in the biosynthesis of biotin 48 and the 638 absence of viable mutants within the Manoil Lab transposon mutant library 45 suggests that it might 639 be essential. Of note, in Mycobacterium tuberculosis, biotin metabolism is already considered a 640 viable antibacterial drug target pathway 83 84, and efforts to target its three bioH isoenzymes are 641 underway85 . 642 For ABPP, enzyme targets are identified through covalent interaction with a small molecule probe, 643 thus focusing the set of targets that are often likely to be druggable and provides direct assays for 644 development of target-specific inhibitors. There is an increasing pool of serine-hydrolase-reactive 645 chemotypes that continue to produce target-specific inhibitors that can be used as tools to further 646 investigate biological function while also sometime serving as antimicrobial drug candidates. The 647 structural information of YjfP provided in this work may enable structure-based design of specific 648 YjfP inhibitors. Alternatively, inhibitors may be identified through traditional target-based screening 649 approaches44, or through competitive ABPP which simultaneously assesses the selectivity profile of 650 candidate inhibitors against all of the active serine hydrolase targets in a sample17,20,86-88. Importantly, 651 if performed on live cells, as in the current study, cell-based competitive ABPP, can also confirm the 652 accessibility of the targets to small molecule inhibitors17,86. 653 We believe that further functional characterizations of the uncharacterized serine hydrolases 654 identified will significantly enhance our understanding of K. pneumoniae pathogenesis and 655 colonization and provide the basis for their validation as anti-virulence and antibacterial drug 656 candidates that are urgently needed in times of emerging multidrug resistance of this critical priority 657 pathogen. 658 659 660 661

662 663 This work was funded by a Centre for New Antibacterial Strategies (CANS) starting-grant through 664 the Trond-Mohn Foundation to C.S.L. We are grateful for additional financial support through the 665 Aurora Outstanding Career Development program at UiT through C.S.L. The research stay of M. J. 666 U. at UMC Utrecht was generously supported by The National Graduate School in Infection Biology 667 and Antimicrobials (IBA). LC-MS/MS analyses were carried out at the UiT Proteomics and 668 Metabolomics Core Facility (PRiME). The PRiME is part of the National Network of Advanced 669 Proteomics Infrastructure (NAPI), funded by the Research C ouncil of Norway INFRASTRUKTUR-670 program (project number: 295910). We also thank Fernanda Paganelli for her support during the 671 initial establishment of the organoid model, Janetta Top, Rob Willems and Bart W. Bardoel for 672 helpful discussions, Coco Duizer for discussions in the lab, and Karin Strijbis for generously 673 providing the Anti-MUC13 antibody, all from UMC Utrecht. Some of this work was supported by 674 the New Zealand China - Maurice Wilkins Centre Collaborative Research Programme (to M.F.) and 675 was conducted during tenure of The Sir Charles Hercus Health Research Fellowship of the Health 676 Research Council of New Zealand (to M.F.). This research was undertaken in part using the MX2 677 beamline at the Australian Synchrotron, part of ANSTO, and made use of the ACRF detector. F.M. 678 .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 28, 2024. ; https://doi.org/10.1101/2024.05.28.596221doi: bioRxiv preprint was supported by the European Union’s Horizon 2020 research program 787 H2020-EU-ITN-EJD 679 (CORVOS #860044) 680 681

Virulence · Chemoproteomics · Organoids · Galleria mellonella · Bacterial colonization 682 ·Lipases · Deacetylase 683 Ethics declaration 684 685 The organoids used were obtained from the Foundation Hubrecht Organoid Biobank (Utrecht, The 686 Netherlands) under TC-Bio protocol number 14-008 and used according to informed consent. 687 688 Human blood was isolated after informed consent was obtained from all subjects in accordance with 689 the Declaration of Helsinki. Approval was obtained from the medical ethics committee of the UMC 690 Utrecht, The Netherlands 691 692 Competing interests 693 694 The authors declare no competing interests. 695 696 697 Author contributions 698 699 Conceptualization: C.S.L., M.J.U., Investigation: M.J.U., G.R., J.Z., T.U., L.v.E, P.H., F.M.; M.C.V., 700 Data analysis: M.J.U., G.R., J.Z., T.U., L.v.E, P.H., M.C.V., M.B., M.F., M.R.d.Z., M.J., C.S.L., 701 Supervision: C.S.L., M.J., M.R.d. Z., M.F., M. B., Funding acquisition: C.S.L., Writing – Original 702 manuscript draft: M.J.U. and C.S.L., Writing - Editing: All authors.703 .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 28, 2024. ; https://doi.org/10.1101/2024.05.28.596221doi: bioRxiv preprint 18 704

705 706 1. Martin, R.M. & Bachman, M.A. Colonization, Infection, and the Accessory Genome of 707 Klebsiella pneumoniae. Frontiers in Cellular and Infection Microbiology 8 (2018). 708 2. Navon-Venezia, S., Kondratyeva, K. & Carattoli, A. Klebsiella pneumoniae : a major 709 worldwide source and shuttle for antibiotic resistance. FEMS Microbiology Reviews 41, 710 252-275 (2017). 711 3. Gupta, A., Ampofo, K., Rubenstein, D. & Saiman, L. Extended Spectrum β Lactamase-712 producing Klebsiella pneumoniae Infections: a Review of the Literature. Journal of 713 Perinatology 23, 439-443 (2003). 714 4. Xu, L., Sun, X. & Ma, X. Systematic review and meta-analysis of mortality of patients 715 infected with carbapenem-resistant Klebsiella pneumoniae . Ann Clin Microbiol 716 Antimicrob 16, 18 (2017). 717 5. WHO (2017). Fact Sheet on Antimicrobial Resistance. https://www.who.int/news-718 room/detail/27-02-2017-who-publishes-list-of-bacteria-for-which-new-antibiotics-are-719 urgently-needed. 720 6. Shon, A.S., Bajwa, R.P.S. & Russo, T.A. Hypervirulent (hypermucoviscous) Klebsiella 721 pneumoniae. Virulence 4, 107-118 (2013). 722 7. Lam, M.M.C. et al. Population genomics of hypervirulent Klebsiella pneumoniae clonal-723 group 23 reveals early emergence and rapid global dissemination. Nature 724 Communications 9, 2703 (2018). 725 8. Kluge, A.F. & Petter, R.C. Acylating drugs: redesigning natural covalent inhibitors. 726 Current opinion in chemical biology 14, 421-427 (2010). 727 9. Thornberry, N.A. & Weber, A.E. Discovery of JANUVIA™ (Sitagliptin), a Selective 728 Dipeptidyl Peptidase IV Inhibitor for the Treatment of Type2 Diabetes. Current topics in 729 medicinal chemistry 7, 557-568 (2007). 730 10. Bachovchin, D.A. & Cravatt, B.F. The pharmacological landscape and therapeutic 731 potential of serine hydrolases. Nature Reviews Drug Discovery 11, 52-68 (2012). 732 11. van Kasteren, S.I., Florea, B.I. & Overkleeft, H.S. Activity-Based Protein Profiling: From 733 Chemical Novelty to Biomedical Stalwart. Methods in molecular biology (Clifton, N.J.) 734 1491, 1-8 (2017). 735 12. Cravatt, B.F. Activity-Based Protein Profiling – Finding General Solutions to Specific 736 Problems. Israel Journal of Chemistry 63, e202300029 (2023). 737 13. Killinger, B.J., Brandvold, K.R., Ramos-Hunter, S.J. & Wright, A.T. Chemoproteomic 738 Analyses by Activity-Based Protein Profiling. in Mass Spectrometry /i1Based Chemical 739 Proteomics 67-99 (2019). 740 14. Fang, H. et al. Recent advances in activity-based probes (ABPs) and affinity-based 741 probes (AfBPs) for profiling of enzymes. Chemical Science 12, 8288-8310 (2021). 742 15. van Rooden, E.J. et al. Chemical Proteomic Analysis of Serine Hydrolase Activity in 743 Niemann-Pick Type C Mouse Brain. Frontiers in Neuroscience 12(2018). 744 16. Jessani, N., Liu, Y., Humphrey, M. & Cravatt, B.F. Enzyme activity profiles of the 745 secreted and membrane proteome that depict cancer cell invasiveness. Proceedings of the 746 National Academy of Sciences 99, 10335-10340 (2002). 747 17. Lentz, C.S. et al. Identification of a S. aureus virulence factor by activity-based protein 748 profiling (ABPP). Nat Chem Biol 14, 609-617 (2018). 749 .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 28, 2024. ; https://doi.org/10.1101/2024.05.28.596221doi: bioRxiv preprint 19 18. Keller, L.J. et al. Characterization of Serine Hydrolases Across Clinical Isolates of 750 Commensal Skin Bacteria Staphylococcus epidermidis Using Activity-Based Protein 751 Profiling. ACS Infect Dis 6, 930-938 (2020). 752 19. Ortega, C. et al. Systematic Survey of Serine Hydrolase Activity in Mycobacterium 753 tuberculosis Defines Changes Associated with Persistence. Cell Chem Biol 23, 290-298 754 (2016). 755 20. Li, M. et al. Identification of cell wall synthesis inhibitors active against Mycobacterium 756 tuberculosis by competitive activity-based protein profiling. Cell Chem Biol 29, 883-757 896.e5 (2022). 758 21. Babin, B.M. et al. Identification of covalent inhibitors that disrupt M. tuberculosis growth 759 by targeting multiple serine hydrolases involved in lipid metabolism. Cell Chem Biol 29, 760 897-909.e7 (2022). 761 22. Hatzios, S.K. et al. Chemoproteomic profiling of host and pathogen enzymes active in 762 cholera. Nature Chem Biol 12, 268-274 (2016). 763 23. Keller, L.J. et al. Chemoproteomic identification of a DPP4 homolog in Bacteroides 764 thetaiotaomicron. Nat Chem Biol 19, 1469-1479 (2023). 765 24. Zweerink, S. et al. Activity-based protein profiling as a robust method for enzyme 766 identification and screening in extremophilic Archaea. Nature Communications 8, 15352 767 (2017). 768 25. Fellner, M. et al. Biochemical and Cellular Characterization of the Function of 769 Fluorophosphonate-Binding Hydrolase H (FphH) in Staphylococcus aureus Support a 770 Role in Bacterial Stress Response. ACS Infectious Diseases 9, 2119-2132 (2023). 771 26. Jo, J. et al. Development of Oxadiazolone Activity-Based Probes Targeting FphE for 772 Specific Detection of Staphylococcus aureus Infections. Journal of the American 773 Chemical Society 146, 6880-6892 (2024). 774 27. Cortés, G., de Astorza, B., Benedí, V.J. & Albertí, S. Role of the htrA gene in Kl ebsiella 775 pneumoniae virulence. Infect Immun 70, 4772-6 (2002). 776 28. Meltzer, M. et al. Structure, function and regulation of the conserved serine proteases 777 DegP and DegS of Escherichia coli. Res Microbiol 160, 660-6 (2009). 778 29. Russell, A.B. et al. Diverse type VI secretion phospholipases are functionally plastic 779 antibacterial effectors. Nature 496, 508-12 (2013). 780 30. Merciecca, T. et al. Role of Klebsiella pneumoniae Type VI secretion system (T6SS) in 781 long-term gastrointestinal colonization. Sci Rep 12, 16968 (2022). 782 31. Nakamoto, N. et al. Gut pathobionts underlie intestinal barrier dysfunction and liver T 783 helper 17 cell immune response in primary sclerosing cholangitis. Nat Microbiol 4, 492-784 503 (2019). 785 32. Uddin, M.J., Overkleeft, H.S. & Lentz, C.S. Activity-Based Protein Profiling in 786 Methicillin-Resistant Staphylococcus aureus Reveals the Broad Reactivity of a 787 Carmofur-Derived Probe. Chembiochem 24, e202300473 (2023). 788 33. Tyanova, S. et al. The Perseus computational platform for comprehensive analysis of 789 (prote)omics data. Nat Methods 13, 731-40 (2016). 790 34. Perez-Riverol, Y. et al. The PRIDE database resources in 2022: a hub for mass 791 spectrometry-based proteomics evidences. Nucleic acids research 50, D543-D552 792 (2022). 793 35. UniProt: the Universal Protein Knowledgebase in 2023. Nucleic Acids Res 51, D523-794 d531 (2023). 795 .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 28, 2024. ; https://doi.org/10.1101/2024.05.28.596221doi: bioRxiv preprint 20 36. Yu, N.Y. et al. PSORTb 3.0: improved protein subcellular localization prediction with 796 refined localization subcategories and predictive capabilities for all prokaryotes. 797 Bioinformatics 26, 1608-1615 (2010). 798 37. Lesuffleur, T. et al. Differential expression of the human mucin genes MUC1 to MUC5 799 in relation to growth and differentiation of different mucus-secreting HT-29 cell 800 subpopulations. Journal of Cell Science 106, 771-783 (1993). 801 38. Vonk, A.M. et al. Protocol for Application, Standardization and Validation of the 802 Forskolin-Induced Swelling Assay in Cystic Fibrosis Human Colon Organoids. STAR 803 Protoc 1, 100019 (2020). 804 39. Zomer-van Ommen, D.D. et al. Comparison of ex vivo and in vitro intestinal cystic 805 fibrosis models to measure CFTR-dependent ion channel activity. J Cyst Fibros 17, 316-806 324 (2018). 807 40. Howat, W.J. & Wilson, B.A. Tissue fixation and the effect of molecular fixatives on 808 downstream staining procedures. Methods 70, 12-9 (2014). 809 41. Heesterbeek, D.A. et al. Bacterial killing by complement requires membrane attack 810 complex formation via surface-bound C5 convertases. Embo j 38(2019). 811 42. Bhandari, B.K., Lim, C.S. & Gardner, P.P. TISIGNER.com: web services for improving 812 recombinant protein production. Nucleic Acids Research 49, W654-W661 (2021). 813 43. Luna-Vargas, M.P. et al. Enabling high-throughput ligation-independent cloning and 814 protein expression for the family of ubiquitin specific proteases. J Struct Biol 175, 113-9 815 (2011). 816 44. Jo, J. et al. Development of Oxadiazolone Activity-Based Probes Targeting FphE for 817 Specific Detection of S. aureus Infections. bioRxiv, 2023.12.11.571116 (2023). 818 45. Ramage, B. et al. Comprehensive Arrayed Transposon Mutant Library of Klebsiella 819 pneumoniae Outbreak Strain KPNIH1. J Bacteriol 199(2017). 820 46. Chen, L., Mathema, B., Pitout, J.D., DeLeo, F.R. & Kreiswirth, B.N. Epidemic Klebsiella 821 pneumoniae ST258 is a hybrid strain. mBio 5, e01355-14 (2014). 822 47. Pitout, J.D., Nordmann, P. & Poirel, L. Carbapenemase-Producing Klebsiella 823 pneumoniae, a Key Pathogen Set for Global Nosocomial Dominance. Antimicrob Agents 824 Chemother 59, 5873-84 (2015). 825 48. Wang, L. et al. Structural insight into the carboxylesterase BioH from Klebsiella 826 pneumoniae. Biochem Biophys Res Commun 520, 538-543 (2019). 827 49. Lesuffleur, T. et al. Differential expression of the human mucin genes MUC1 to MUC5 828 in relation to growth and differentiation of different mucus-secreting HT-29 cell 829 subpopulations. J Cell Sci 106 ( Pt 3), 771-83 (1993). 830 50. In, J. et al. Enterohemorrhagic Escherichia coli reduce mucus and intermicrovillar 831 bridges in human stem cell-derived colonoids. Cell Mol Gastroenterol Hepatol 2, 48-832 62.e3 (2016). 833 51. Sato, T. et al. Long-term expansion of epithelial organoids from human colon, adenoma, 834 adenocarcinoma, and Barrett's epithelium. Gastroenterology 141, 1762-72 (2011). 835 52. Yu, S. et al. Paneth Cell-Derived Lysozyme Defines the Composition of Mucolytic 836 Microbiota and the Inflammatory Tone of the Intestine. Immunity 53, 398-416.e8 (2020). 837 53. Domingues, M.M. et al. Biophysical characterization of polymyxin B interaction with 838 LPS aggregates and membrane model systems. Biopolymers 98, 338-44 (2012). 839 .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 28, 2024. ; https://doi.org/10.1101/2024.05.28.596221doi: bioRxiv preprint 21 54. Merino, S., Camprubí, S., Albertí, S., Benedí, V.J. & Tomás, J.M. Mechanisms of 840 Klebsiella pneumoniae resistance to complement-mediated killing. Infect Immun 60, 841 2529-35 (1992). 842 55. Cortés, G., Astorza, B.d., Benedí, V.J. & Albertí, S. Role of the htrA Gene in Klebsiella 843 pneumoniae Virulence. Infection and Immunity 70, 4772-4776 (2002). 844 56. Ménard, G., Rouillon, A., Cattoir, V. & Donnio, P.-Y. Galleria mellonella as a Suitable 845 Model of Bacterial Infection: Past, Present and Future. Frontiers in Cellular and 846 Infection Microbiology 11 (2021). 847 57. Bruchmann, S., Feltwell, T., Parkhill, J. & Short, F.L. Identifying virulence determinants 848 of multidrug-resistant Klebsiella pneumoniae in Galleria mellonella. Pathog Dis 79 849 (2021). 850 58. Ollis, D.L. et al. The alpha/beta hydrolase fold. Protein Eng 5, 197-211 (1992). 851 59. Nardini, M., Lang, D.A., Liebeton, K., Jaeger, K.E. & Dijkstra, B.W. Crystal structure of 852 pseudomonas aeruginosa lipase in the open conformation. The prototype for family I.1 of 853 bacterial lipases. J Biol Chem 275, 31219-25 (2000). 854 60. Krissinel, E. & Henrick, K. Inference of macromolecular assemblies from crystalline 855 state. J Mol Biol 372, 774-97 (2007). 856 61. Trimble, M.J., Mlynár č ik, P., Kolá ř , M. & Hancock, R.E. Polymyxin: Alternative 857 Mechanisms of Action and Resistance. Cold Spring Harb Perspect Med 6(2016). 858 62. Bevins, C.L. & Salzman, N.H. Paneth cells, antimicrobial peptides and maintenance of 859 intestinal homeostasis. Nature Reviews Microbiology 9, 356-368 (2011). 860 63. Gallo, R.L. & Hooper, L.V. Epithelial antimicrobial defence of the skin and intestine. Nat 861 Rev Immunol 12, 503-16 (2012). 862 64. O'Neil, D.A. et al. Expression and regulation of the human beta-defensins hBD-1 and 863 hBD-2 in intestinal epithelium. J Immunol 163, 6718-24 (1999). 864 65. St ą czek, S., Cytry ń ska, M. & Zdybicka-Barabas, A. Unraveling the Role of 865 Antimicrobial Peptides in Insects. International Journal of Molecular Sciences 24, 5753 866 (2023). 867 66. Mak, P., Zdybicka-Barabas, A. & Cytry ń ska, M. A different repertoire of Galleria 868 mellonella antimicrobial peptides in larvae challenged with bacteria and fungi. 869 Developmental & Comparative Immunology 34, 1129-1136 (2010). 870 67. Clements, A. et al. Secondary acylation of Klebsiella pneumoniae lipopolys accharide 871 contributes to sensitivity to antibacterial peptides. J Biol Chem 282, 15569-77 (2007). 872 68. Doorduijn, D.J., Rooijakkers, S.H., van Schaik, W. & Bardoel, B.W. Complement 873 resistance mechanisms of Klebsiella pneumoniae. Immunobiology 221, 1102-9 (2016). 874 69. Schmiel, D.H., Wagar, E., Karamanou, L., Weeks, D. & Miller, V.L. Phospholipase A of 875 Yersinia enterocolitica Contributes to Pathogenesis in a Mouse Model. Infection and 876 Immunity 66, 3941-3951 (1998). 877 70. Sitkiewicz, I. et al. Emergence of a bacterial clone with enhanced virulence by 878 acquisition of a phage encoding a secreted phospholipase A2. Proceedings of the National 879 Academy of Sciences 103, 16009-16014 (2006). 880 71. Istivan, T.S. & Coloe, P.J. Phospholipase A in Gram-negative bacteria and its role in 881 pathogenesis. Microbiology 152, 1263-1274 (2006). 882 72. Braun, A., Stremmel, W. & Ehehalt, R. The Role of Phospholipids Within the Intestinal 883 Mucosal Barrier. Z Gastroenterol 49, A11 (2011). 884 .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 28, 2024. ; https://doi.org/10.1101/2024.05.28.596221doi: bioRxiv preprint 22 73. Swidsinski, A. et al. Mucosal flora in inflammatory bowel disease. Gastroenterology 122, 885 44-54 (2002). 886 74. Stremmel, W. et al. Mucosal protection by phosphatidylcholine. Dig Dis 30 Suppl 3, 85-887 91 (2012). 888 75. Ehehalt, R., Braun, A., Karner, M., Füllekrug, J. & Stremmel, W. Phosphatidylcholine as 889 a constituent in the colonic mucosal barrier—Physiological and clinical relevance. 890 Biochimica et Biophysica Acta (BBA) - Molecular and Cell Biology of Lipids 1801, 983-891 993 (2010). 892 76. Stremmel, W. et al. Phosphatidylcholine passes through lateral tight junctions for 893 paracellular transport to the apical side of the polarized intestinal tumor cell-line CaCo2. 894 Biochimica et Biophysica Acta (BBA) - Molecular and Cell Biology of Lipids 1861, 1161-895 1169 (2016). 896 77. Tannaes, T., Bukholm, I.K. & Bukholm, G. High relative content of lysophospholipids of 897 Helicobacter pylori mediates increased risk for ulcer disease. FEMS Immunology & 898 Medical Microbiology 44, 17-23 (2005). 899 78. Goggin, P.M. et al. Surface hydrophobicity of gastric mucosa in Helicobacter pylori 900 infection: effect of clearance and eradication. Gastroenterology 103, 1486-90 (1992). 901 79. Mauch, F., Bode, G., Ditschuneit, H. & Malfertheiner, P. Demonstration of a 902 phospholipid-rich zone in the human gastric epithelium damaged by Helicobacter pylori. 903 Gastroenterology 105, 1698-704 (1993). 904 80. Munoz-Price, L.S. et al. Clinical epidemiology of the global expansion of Klebsiella 905 pneumoniae carbapenemases. Lancet Infect Dis 13, 785-96 (2013). 906 81. van Duin, D. & Doi, Y. Outbreak of Colistin-Resistant, Carbapenemase-Producing 907 Klebsiella pneumoniae: Are We at the End of the Road? J Clin Microbiol 53, 3116-7 908 (2015). 909 82. Han, R. et al. The antihelminth drug rafoxanide reverses chromosomal-mediated colistin-910 resistance in Klebsiella pneumoniae. mSphere 8, e0023423 (2023). 911 8 3. Park, S.W. et al. Target-based identification of whole-cell active inhibitors of biotin 912 biosynthesis in Mycobacterium tuberculosis. Chem Biol 22, 76-86 (2015). 913 84. Salaemae, W., Booker, G.W. & Polyak, S.W. The Role of Biotin in Bacterial Physiology 914 and Virulence: a Novel Antibiotic Target for Mycobacterium tuberculosis . Microbiol 915 Spectr 4(2016). 916 85. Xu, Y. et al. Three enigmatic BioH isoenzymes are programmed in the early stage of 917 mycobacterial biotin synthesis, an attractive anti-TB drug target. PLoS Pathog 18, 918 e1010615 (2022). 919 86. Chen, L., Keller, L.J., Cordasco, E., Bogyo, M. & Lentz, C.S. Fluorescent Triazole Urea 920 Activity-Based Probes for the Single-Cell Phenotypic Characterization of Staphylococcus 921 aureus. 5699-5703 (Weinheim ; [New York] :, 2019). 922 87. Prothiwa, M. & Böttcher, T. Competitive profiling for enzyme inhibitors using chemical 923 probes. Methods Enzymol 633, 49-69 (2020). 924 88. Baggelaar, M.P. & Van der Stelt, M. Competitive ABPP of Serine Hydrolases: A Case 925 Study on DAGL-Alpha. Methods Mol Biol 1491, 161-169 (2017). 926 927 .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 28, 2024. ; https://doi.org/10.1101/2024.05.28.596221doi: bioRxiv preprint 23 928 Figure 1. Chemoproteomic identification of serine hydrolases in K. pneumoniae. A) Schematic drawing of the929 features of Activity-based probes (ABPs) (i) and the chemical structures of the ABPs FP-TAMRA (ii) and FP-biotin930 (iii) that were used in this study. B) Schematic overview of general workflow for activity- based protein profiling931 (ABPP) of serine hydrolases in live K. pneumoniae. Fluorophosphonate-based ABPs selectively bind to accessible932 and active serine hydrolase species. Fluorescently tagged ABPs allow their visualization by SDS-PAGE analysis and933 fluorescence scanning, whereas biotinylated probes are used for streptavidin-enrichment and identification via LC -934 MS/MS or they can be visualized using a fluorescent tag through SDS-PAGE and in-gel fluorescence analysis. C)935 SDS-PAGE analysis of K. pneumoniae MKP103 live cells labeled with FP-TMR for 60 min at 37 °C. The graph936 depicts fluorescent scans in the Cy3 (520nm) channel utilizing the Amersham™ Ty phoon™ 5 (cytiva) imaging937 system. D) Volcano plot of K. pneumoniae proteins identified by LC-MS/MS that were enriched after FP- biotin938 treatment compared to a vehicle-treated control dataset. A two-tailed two-sample t- test was conducted to compare939 cells labeled with DMSO and FP-biotin. Significantly enriched hits are highlighted with blue dots, significantly940 enriched 10 SHs are indicated with red dots, and enriched but not significant hits are marked with grey rectangles941 E) Heatmap of homologs of K. pneumoniae serine hydrolases across 20 representative gut commensal bacterial942 species and in humans (Chordata), as sourced from the Human Microbiome Project Reference Genomes for the943 Gastrointestinal Tract database using BLAST- P. In the heatmap, each filled cell indicates that the species has a944 homolog of the K. pneumoniae SHs, as determined by a threshold e-value of 1 × 10 –10; white background denotes945 the absence of a homolog.946 23 he tin ng ble nd - C) ph ng tin are tly . ial the s a tes g. .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 28, 2024. ; https://doi.org/10.1101/2024.05.28.596221doi: bioRxiv preprint 24 947 948 949 Figure 2. Functional validation of K. pneumoniae transposon mutant strains. A) FP-TMR labeling profiles of K.950 pneumoniae transposon mutant strains with insertions in the indicated serine hydrolase genes. Live cells were951 labelled with FP-TMR and lysed prior to SDS- PAGE analysis and fluorescence scanning. Arrowheads indicate952 labeled proteins disappearing in individual mutant strains (Red arrowhead: PldB, blue arrowhead: DegP, green953 arrowhead: YchK, and orange arrowhead: YbfF). The experiment was performed three times with similar results. B)954 Annotated FP-TAMRA profile of K. pneumoniae indicating four identified probe targets. C) Growth curves of K.955 pneumoniae WT and SH-deficient transposon mutants in LB media, n=3 biologically independent samples. D)956 Analysis of bacterial growth in HT29-MTX co-culture model. The graph shows bacterial CFUs (log10 CFU/mL) at957 different time points (0, 2, 4, 6, and 8 hours) after adding K. pneumoniae WT and SH-deficient transposon mutants958 to HT29-MTX cells at an MOI of 20. Statistical significance was assessed using two- way ANOVA test with post959 hoc Dunnet’'s multiple comparisons tests compared with wild type (*p < 0.05, **p < 0.01, ***p < 0.001). E)960 Growth curves of K. pneumoniae WT and SH-deficient transposon mutants in D MEM media without FBS. Growth961 curves in B-E show means ± standard deviation of n=3 independent biological culture replicates.962 24 K. re ate en ) K. D) at nts ost ) th es. .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 28, 2024. ; https://doi.org/10.1101/2024.05.28.596221doi: bioRxiv preprint 25 963 964 965 Figure 3. Assessment of SH-deficient K. pneumoniae mutants in 2D colonic organoid co-culture model . A)966 Schematic representation of the establishment and differentiation of 2D organoid cultures, including co- culturing967 with K. pneumoniae cultures. B) Sideview of Methacarn fixed organoid stained with Wheat Germ Agglutinin968 conjugated to Alexa fluorTM647 and anti-MUC antibodies combined with Alexa fluorTM488goat anti- mouse or969 Alexa fluorTM488 goat anti-rabbit. WGA (red) labels the cell layer as well as the mucus layer. (A) Anti- MUC1970 (green) binds to the MUC1 mucin, present in the organoid samples in a relatively low abundance. The side view971 shows co-localization of WGA and MUC1, mainly visible at the ends of the picture. (B) Anti-MUC2 and WGA co -972 25 A) ng in or C1 ew - .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 28, 2024. ; https://doi.org/10.1101/2024.05.28.596221doi: bioRxiv preprint 26 localize, visible in the middle part of the side view. (C) Anti-MUC13 (green) binds to the MUC13 mucin present in 973 these organoid samples. WGA and anti-MUC13 co-localize, visible primarily on the right end of the image, where 974 the red and green signal are at the same height. 40x magnification. C) Methacarn-fixed organoid cultures stained 975 with antibodies against villin, sox9 and lysozyme C. Scale bar to represent size. (A) Negative control stained with 976 only Phalloidin-iFluor 633 (red), Alexa fluorTM488 goat anti-mouse and Alexa fluorTM488 goat anti-rabbit (green) 977 and Alexa fluorTM405 goat anti-rabbit (blue). 10x magnification, 4x optical zoom. (B) Organoids stained with 978 Phalloidin-iFluor 633 to visualize the F-actin filaments of the cells (red) and anti-villin with Alexa fluorTM 488 goat 979 anti-mouse(green) to stain villin cells. 10x magnification, 5,2x optical zoom. (C) Organoids stained with Phalloidin-980 iFluor 633 to visualize the F-actin filaments of the cells (red) and anti-sox9 with Alexa fluorTM 405goat anti-rabbit 981 (blue) to stain stem cells. 10x magnification, 7,3x optical zoom. (D) Organoids stained with Phalloidin-iFluor 633 to 982 visualize the F-actin filaments of the cells (red) and DAKOA099 with Alexa fluorTM 488 goat anti-rabbit (green) to 983 stain lysozyme C which is secreted by Paneth cells. Yellow boxes show individual Paneth cells. 10x magnification, 984 5,5x optical zoom. D) Bacterial counts (log10 CFU/mL) in the apical part of organoid co-cultures measured at 0, 2, 985 4, 8, and 16 h following incubation with K. pneumoniae WT and SH-deficient mutants on a 2D organoid monolayer. 986 A two-way ANOVA test with Dunnett's multiple comparisons test was used to compare with the wild type (*p < 987 0.05, **p < 0.01, ***p < 0.001). E) Bacterial counts (log10 CFU/mL) for surface-attached bacteria to the organoid 988 co-cultures, were measured at 0, 2, 4, 8 and 16 h following incubation with K. pneumoniae WT and SH-deficient 989 mutants on a 2D organoid monolayer. A two-way ANOVA test with Dunnett's multiple comparisons test was used, 990 comparing mutants to the wild type, with significance levels noted as *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 991 0.0001. F) K. pneumoniae WT and SH-deficient mutants counts (log10 CFU/mL) in the apical part, surface-992 attached and the basolateral compartments 16 h post infection. LoD is the lower limit of detection level. Dunnett's 993 one-way ANOVA was used for comparison to the WT (**p < 0.01, ***p < 0.001, ****p < 0.0001, ns indicates no 994 significant difference).995 .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 28, 2024. ; https://doi.org/10.1101/2024.05.28.596221doi: bioRxiv preprint 27 996 997 Figure 4 . PldB, YjfP, and DegP confers resistance to envelope stressor polymyxin B and to complement998 killing. A) The growth of K. pneumoniae strains in absence (grey) or presence of 8 m g/ml polymyxin B (red)999 (mean/i1 ±/i1 s.e.m.; n /i1 =/i1 5 independent replicates). B) Quantificatio n of the area under the growth curve for1000 treatment with polymyxin B was normalized to the growth curve of untreated cultures. C) Inner membrane1001 permeabilization was evaluated in K. pneumoniae WT and its transposon mutants exposed to either 10% normal1002 human serum (NHS) or 10% heat- inactivated NHS (HiNHS). The bacteria were incubated at 37°C with 1 µM1003 SYTOX Green nucleic acid sta in and SYTOX fluorescence intensity was monitored at 1 min intervals for 120 min1004 using a microplate reader. The presented data are the mean ± standard deviation from two independent experiments.1005 D) SYTOX Green fluorescence intensity, measured after 120 min of exposure to 10% NHS as described in C. The1006 data are presented as mean ± standard deviate on from two independent experiments. The dashed line refers to1007 buffer permeabilization signal (RPMI + SytoxGreen ).1008 27 nt d) for ne al M in ts. he to ). .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 28, 2024. ; https://doi.org/10.1101/2024.05.28.596221doi: bioRxiv preprint 28 1009 1010 Figure 5 . Infection of G. mellonella with K. pneumoniae strains. A) Schematic diagram illustrating the1011 experimental process: Groups of G. mellonella larvae were injected with K. pneumoniae W T o r S H -deficient1012 transposon mutants and incubated at 37 °C for up to 5 days. Bacterial virulence was evaluated by monitoring of live1013 and dead larvae. B) Kaplan–Meier (KM) survival plots of G. mellonella larvae after inoculation of the K.1014 pneumoniae WT and SH-deficient mutants. Plots show an average of 3 independent experiments with 10 larvae per1015 group with mortality monitored daily for 5 day (N= 210). Larvae injected with PBS were used as negative control.1016 Mutants displaying a significant difference in survival compared to the WT were determined by the log- rank1017 (Mantel–Cox) test and are denoted (**** p < 0.0001, ** p < 0.05, ns = not significantly different). 1018 28 the ent ve K. er ol. nk .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 28, 2024. ; https://doi.org/10.1101/2024.05.28.596221doi: bioRxiv preprint 29 1019 Figure 6. Biochemical characterization of YjfP, YqiA, and PldB and structure of hydrolases YjfP and YqiA.1020 Normalized cleavage rate (, in relative fluorescent unit (RFU)/s per nM of protein) of 4-methylumbelliferone (4 -1021 M U ) s u b s t r a t e s b y Y j f P (A) , b y Y q i A (B), and by PldB. ( C). The presented bars represent the mean ± standard1022 deviation (n = 3 technical replicates). D) YjfP crystallised as a dimer (cyan and pink), where the dimer interface is1023 mediated by connections across β strands 1 and α helices 1 from each chain. Two salt bridges are formed at the top1024 of the dimer, creating a roof above a putative substrate binding cavity. This cavity contains the catalytic triad;1025 Ser116, Asp187, and His221. E) Interactions are shown between side chains of the catalytic triad (modelled in two1026 conformations). Displayed around Ser116 is the mFo-DFc map contoured at 3.5 for a putative ligand. This putative1027 ligand likely interacts with a hydrophilic portion of th e active site (Tyr44, Ser116, and Arg222 modelled in two1028 conformations). Part of the density is in the oxyanion hole, coordinated by backbone amides from Met117 and1029 Phe37. Beyond the oxyanion hole, several hydrophobic side chains are shown. F) Y qiA (beige) crystallised as a1030 monomer, where β strands 4 and 5 form a lid region. The catalytic triad (Ser69, Asp147, His172) is more exposed1031 than in the YjfP structure. G ) Interactions between side chains of the catalytic triad are shown. A chlorine atom1032 (green) is modelled in the oxyanion hole coordinated by backbone amides of Leu70 and Phe10. Several hydrophobic1033 side residues on α helix 4 and a loop between β strand 7 and α helix 6 are likely important for substrate recognition.1034 29 A. - rd is op ad; wo ive wo nd s a ed m bic n. .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 28, 2024. ; https://doi.org/10.1101/2024.05.28.596221doi: bioRxiv preprint 30 1035 Table 1. Overview of serine hydrolases identified in Klebsiella pneumoniae subsp. pneumoniae 1036 KPNIH1. Refer to Supplementary Dataset 1 and Supplementary Table 2. 1037 Gene name Functional Annotation (putative) (Pfam) Predicted location (Putative) (PSORTb) % Homology with H. sapiens protein MW [kDa] KPSH 1 degP Trypsin-like peptidase Periplasmic - 49.5 KPSH 2 ychK Patatin-like phospholipase Extracellular - 33.3 KPSH 3 ybfF AB_Hydrolase 6 Unknown - 28.5 KPSH 4 catD Hydrolase 4 Cytoplasmic - 27.3 KPSH 5 pldB Hydrolase 4 Cytoplasmic membrane 25% 38.2 KPSH 6 degQ Trypsin-like peptidase Periplasmic - 47.2 KPSH 7 yqiA UPF0227 Cytoplasmic - 21.5 KPSH 8 ycfP UPF0227 Cytoplasmic - 21.1 KPSH 9 bioH AB_Hydrolase 1 Cytoplasmic 30% 28.3 KPSH 10 yjfP Peptidase S9 Unknown - 26.5 1038 1039 1040 .CC-BY-NC-ND 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 28, 2024. ; https://doi.org/10.1101/2024.05.28.596221doi: bioRxiv preprint