bioRxiv preprint doi: https://doi.org/10.1101/2020.10.22.350942; this version posted October 22, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity.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 available under aCC-BY-NC-ND 4.0 International license.

1 The PRC2-dependent epigenetic reprograming of the bladder epithelium 2 exacerbates urinary tract infections 3 4 Chunming Guo1,a, Mingyi Zhao1,a, Xinbing Sui2,a, Zarine Balsara2,a, Songhui Zhaia, Ping 5 Zhub and Xue Li*,a,c 6 7 a, Departments of Urology and Surgery, Boston Children’s Hospital, Harvard Medical 8 School, 300 Longwood Avenue, Boston, MA 02115, USA; 9 b, Guangdong Academy of Medical Sciences, Guangdong Cardiovascular Institute, 10 Guangdong Provincial People’s Hospital, Guangzhou, Guangdong 510100, China; 11 b, Samuel Oschin Comprehensive Cancer Institute, Departments of Medicine and 12 Biomedical Sciences, Cedars Sinai Medical Center, 8700 Beverly Blvd, Davis Building, 13 3094D/E, Los Angeles, CA 90048 14 1, Equal first author; 2, Equal second author; 15 *Correspondence: [email protected] (XL) or [email protected] (XL) 16 Tel: +1 3104239546 17 bioRxiv preprint doi: https://doi.org/10.1101/2020.10.22.350942; this version posted October 22, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity.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 available under aCC-BY-NC-ND 4.0 International license.

18 Mucosal imprint sensitizes recurrent urinary tract infections (UTIs), a significant health 19 and quality of life burden worldwide, which are associated with heightened inflammatory 20 host response, severe basal cell hyperplasia and impaired superficial cell differentiation. 21 Here, we show that bladder infections induce expression of Ezh2, the methyltransferase 22 of polycomb repressor complex 2 (PRC2)-dependent epigenetic gene silencing program. 23 In mouse models of UTIs, urothelium-specific inactivation of PRC2 reduces the urine 24 bacteria burden. The mutants exhibit a blunted inflammatory response likely due to the 25 diminished activity of NF-kB signaling pathway. PRC2 inactivation also improves 26 urothelial differentiation and attenuates basal cell hyperplasia phenotype. Moreover, the 27 Ezh2-specific small molecule inhibitors markedly improve disease outcomes of bladder 28 superinfection and chronic cystitis. Taken together, these findings suggest that the UTI- 29 induced epigenetic reprograming in the bladder urothelium likely contributes to the 30 mucosal imprint, and further suggest that targeting PRC2 methyltransferase offers a non- 31 antibiotic strategy to mitigate UTIs. 32

2 bioRxiv preprint doi: https://doi.org/10.1101/2020.10.22.350942; this version posted October 22, 2020. The copyright holder for this preprint (which 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 available under aCC-BY-NC-ND 4.0 International license.

33 Introduction 34 35 The ability to learn from the past experience offers an added advantage to defend against 36 pathogens upon subsequent encounters. For example, a local infection in plants results 37 in the epigenetic changes that lead to systemic acquired resistance when challenged by 38 the same or unrelated pathogens(1). Invertebrates also display an enhanced response to 39 secondary infection; such a memory-like feature persists and is transmissible to the next 40 generation(2, 3). Vertebrates have a highly specialized adaptive immune system, which 41 produces long-lasting immunological memory and enhanced responses to eliminate 42 pathogens. Emerging evidence suggests that the innate immune cells also display a long- 43 term enhanced inflammatory response, termed “trained immunity”(4), to challenge 44 infections, suggesting that mammalian cells may have retained the ancestral mechanism 45 of immunological memory to defend against infections(5, 6). The barrier epithelial cells 46 are often the first to encounter pathogens. However, whether mammalian epithelial cells 47 are able to retain the information of pathogen encounter and to respond accordingly 48 during secondary infections have been poorly understood. 49 The bladder urothelium is a specialized epithelial barrier that prevents entry of 50 irritants and pathogens in the urine. Unlike other epithelial barriers such as those of the 51 skin and intestine, mature bladder urothelium has a very low proliferation index(7). The 52 estimated urothelial cell turnover rate in rodents is once every 200 days. Nevertheless, 53 bladder urothelium has a remarkable regenerative capacity to maintain its barrier function 54 after injuries(8). For example, during urinary tract infections (UTIs), the quiescent 55 urothelium becomes hyperproliferative and regenerates the superficial or umbrella cells 56 within days to repair the damaged tissue(9-13). The rapid regenerative kinetics of the 57 bladder urothelium appears to be biologically necessary for the restoration of the barrier 58 function. Likewise, the exceedingly low turnover rate suggests that molecular changes of 59 the urothelial cells could persist throughout the lifetime of bladder urothelium. 60 UTIs caused by uropathogenic E. coli (UPEC) are among the most common 61 bacterial infections that afflict over 150 million people each year worldwide; and have the 62 estimated 2-3 billion dollars of annual costs in the United States alone(14, 15). Even after 63 appropriate antibiotic treatment, UTI patients have >25% chance of developing a

3 bioRxiv preprint doi: https://doi.org/10.1101/2020.10.22.350942; this version posted October 22, 2020. The copyright holder for this preprint (which 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 available under aCC-BY-NC-ND 4.0 International license.

64 recurrent infection in the following 6 months, suggesting that a prior history of bladder 65 infection is an independent risk factor for subsequent UTIs(16-18). Intriguingly, bacteria 66 induce bladder mucosal imprint, which sensitizes to recurrent and chronic diseases(19). 67 Patients with chronic and recurrent UTIs display pathological mucosal remodeling 68 including urothelial hyperplasia and a lack of terminal differentiation(20, 21). Analogous 69 to trained immunity, mucosal imprint results in a heightened inflammatory response to 70 bacterial infections(19). An exaggerated bladder inflammation, however, predisposes to 71 symptomatic chronic and recurrent infections(22, 23). When mice are experimentally 72 primed with either the bacterial or chemical injuries, these pretreated mice develop a 73 heightened inflammatory host response but poor disease outcomes upon reinfections(24, 74 25). Rather than a self-limiting acute cystitis, the primed mice exhibit severe urothelial 75 damage followed by chronic cystitis, pyelonephritis and mortality. Hence, the amplitude 76 of bladder inflammatory response to bacterial infections is closely linked to clinical 77 outcomes of UTIs from asymptomatic to life-threatening diseases(18). 78 Among the host defense strategies, exfoliation physically removes infected 79 urothelial cells and intracellular bacterial communities (IBCs)(26-28), thereby facilitating 80 bacterial clearance. However, this is a double-edged strategy as it damages the epithelial 81 barrier and exposes the underlying tissue to pathogens and irritants, which could seed 82 recurrent infections(29). In addition to exfoliation, the bladder urothelial cells are key in 83 detecting UPEC through receptors including Uroplakin Ia (Upk1a) and Toll-like receptor 84 4 (Tlr4)(30-32), which activate downstream effectors such as the nuclear factor kappa B 85 (NF-kB)-dependent inflammatory cytokines and chemokines. Along with other 86 inflammatory mediators, the bladder urothelial cells produce a burst of interleukin 1a (Il- 87 1a), Il-6 and Il-8 to recruit immune cells such as neutrophils that are critical in actively 88 clearing out the invading pathogens(31-33). Collectively, the bladder urothelium is the 89 essential epithelial barrier and a central player in orchestrating the immunological 90 responses to UTIs. 91 Exposure to pathogens may cause epigenetic remodeling of host cells(34-36). The 92 epigenetic deposits such as histone H3 lysine 27 trimethylation (H3K27me3) have the 93 potential to function as a molecular memory after pathogen clearance since they can 94 persist through multiple cell cycles and several generations(37-39). H3K27me3 is tightly

4 bioRxiv preprint doi: https://doi.org/10.1101/2020.10.22.350942; this version posted October 22, 2020. The copyright holder for this preprint (which 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 available under aCC-BY-NC-ND 4.0 International license.

95 linked to chromatin compaction and transcription silencing(40-42). Intriguingly, exposure 96 of human bladder urothelial carcinoma cells to UPEC in vitro induces expression of Ezh2, 97 the methyltransferase of polycomb repressor complex 2 (PRC2)(43), which catalyzes 98 formation of H3K27me3. We have previously reported that the PRC2-dependent 99 H3K27me3 modification plays a critical role in the bladder urothelium during embryonic 100 development(9). However, it’s potential role in mature urothelial cells during UTIs has not 101 been reported yet. Here, we provide evidence suggesting that UPEC induces Ezh2 in the 102 urothelial cells; and the PRC2-dependent epigenetic program is tightly associated with an 103 exaggerated inflammatory response and poor disease outcomes of UTIs. Moreover, we 104 show that targeting the PRC2 activity, both genetically and pharmacologically, 105 significantly mitigates severity of infections. These findings collectively suggest that the 106 UTI-induced mucosal imprint is in part mediated by the PRC2-dependent epigenetic 107 mechanism and further suggest that targeting the underlying epigenetic program offers 108 an alternative non-antibiotic adjuvant strategy to ameliorate UTIs. 109 110 Results 111 112 Bacterial infection induces Ezh2 gene expression 113 Wildtype C57Bl6 mice were infected by transurethral inoculation of human UPEC 114 UTI89 directly into the bladder(44). To identify candidate UTI-induced epigenetic 115 regulators, we performed whole transcriptome analysis of the bladder urothelium at one 116 day post inoculation (dpi, Figure 1A). Among the differentially expressed genes (DEGs) 117 between the UPEC-infected and vehicle controls, Ezh2 was significantly upregulated 118 (Table S1). Genes that encode the DEAD-box containing proteins Ddx26b and Ddx5, the 119 SWI/SNF complex protein Smarcad1 and bromodomain protein Baz2b were also 120 upregulated by UPEC infection. Several long intergenic noncoding RNAs including Xist, 121 Neat1, A330044P14Rik, Gm21781, Al480526 and Gm26905 were also significantly 122 induced. Other candidate epigenetic regulators including Chst2, Uchl1, Padi1, Padi2 and 123 Inmt were downregulated in the bladder urothelium after the infection. 124 We focused on the PRC2-dependent epigenetic repression program as it is known to 125 play a critical role in urothelial differentiation during bladder development(9). Quantitative

5 bioRxiv preprint doi: https://doi.org/10.1101/2020.10.22.350942; this version posted October 22, 2020. The copyright holder for this preprint (which 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 available under aCC-BY-NC-ND 4.0 International license.

126 analysis confirmed that bacterial infection resulted in a significant increase of Ezh2 127 (Figure 1B). Expression of other members of PRC2 complex including Ezh1, Eed and 128 Suz12 was not affected (Figures 1B and S1A). Ezh2 protein was barely detectable in the 129 adult bladder urothelium by immunohistochemistry underlying physiological condition 130 (Figure 1C, panel 1). In contrast, strong Ezh2 immunoreactivity was observed in the 131 bladder urothelium and lamina propria layers at 1dpi (Figure 1C, panel 2). Consistently, 132 strong H3K27me3 immunoreactivity was detected in the bladder urothelial and lamina 133 propria cells (Figure 1C, panels 6-7). Similar to an acute infection, the repeated bladder 134 infections also resulted in the markedly inductions of Ezh2 and H3K27me3 in the bladder 135 and moreover, Ezh2 induction lasted for more than 7 days (Figure 1C, 7xUTI, panels 4-5 136 and 9-10). Together, these findings suggest that bladder infections may lead to the PRC2- 137 dependent epigenetic remodeling by inducing expression Ezh2 methyltransferase. 138 139 Urothelium-specific Eed condition knockout blunts the proliferative response to 140 UTI via the p53 pathway 141 To examine a potential role of the PRC2-dependent epigenetic program in bladder 142 urothelium during UTI, we used urothelium-specific Eed conditional knockout mice, 143 EedcKO, in which PRC2-dependent methyltransferase activity mediated by both Ezh1 and 144 Ezh2 was eliminated from the bladder urothelium(9). An acute infection (1xUTI) resulted 145 in the loss of cytokeratin 20-positive (Krt20+) superficial cells, accompanied by an 146 increase in Krt5+ basal cells in both the mutants and controls (Figure S1B). Krt20+ 147 superficial cells were recovered within 3-7 days in both EedcKO and controls. While EedcKO 148 responded to the bladder infection by activating urothelial cell proliferation, the cell 149 proliferation rate in EedcKO mice was more than 2-fold less (<30%) than that of controls 150 (~60%) based on Ki67 staining (Figures 1D and E). Similarly, ~15% of control and ~7% 151 of EedcKO urothelial cells were positively labeled by a mitotic marker phospho-histone H3 152 (Figure S1C and D, p < 0.05). Collectively, these findings suggest that urothelium-specific 153 inactivation of PRC2 significantly blunts the hyper-proliferative response to bacterial 154 infection. 155 To examine the PRC2-dependent transcriptional program in the developing and 156 mature bladder urothelium, we profiled the urothelium-specific transcriptome using a

6 bioRxiv preprint doi: https://doi.org/10.1101/2020.10.22.350942; this version posted October 22, 2020. The copyright holder for this preprint (which 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 available under aCC-BY-NC-ND 4.0 International license.

157 Translating Ribosome Affinity Purification (TRAP) strategy(45). Urothelium TRAP RNAs 158 were isolation from EedcKO and controls at embryonic day 18.5 (E18.5) and adult bladders 159 (Figure S1E). Transcriptomic analysis identified 197 differentially expressed genes 160 (DEGs, q<0.05, n=2 per group) at E18.5 (Figure S1F). Among the DEGs, 159 of them 161 were upregulated and 38 were downregulated. Consistent with the previous report(9), 162 expression of cell cycle inhibitors including Cdkn2a, Cdkn1a, En1, En2, Pax6 and Six3 163 were aberrantly upregulated in the urothelium of EedcKO mice (Figure S1F). Ingenuity 164 Pathway Analysis (IPA) of the DEGs revealed that the p53 molecular pathway was 165 significantly activated (Figure S1G). Among p53 gene targets, Cdkn1a, Bbc3, Trp53inp1 166 and TP73 were highly upregulated. Additional pathways including the Aryl hydrocarbon 167 receptor, G1/S checkpoint and glioblastoma multiforme signaling pathways were 168 upregulated while the cyclins and cell cycle regulation pathway was significantly 169 downregulated. Similar to the developing urothelium, Cdkn2a was also upregulated in 170 adult urothelium of EedcKO mice (Figure 1F). A total of 27 DEGs was observed from the 171 adult urothelium, which was much fewer than that from the embryonic urothelium, 172 consistent with the observation that the adult Eed knockout bladders were histologically 173 indistinguishable from littermate controls. Taken together, results from these 174 transcriptomic analyses suggest that the PRC2-dependent epigenetic program promotes 175 urothelial cell proliferation possibly via the p53 tumor suppressor pathway. 176 177 PRC2 heightens the inflammatory response to bladder infections 178 Bladder infections resulted in the immune cell infiltration, which was observed in all 179 three layers of the bladder including the urothelial, lamina propria and smooth muscle 180 layers (Figure 2A). After an acute bacterial infection, damage to the urothelium as 181 indicated by the loss of Krt20+ superficial cells and increase of Krt5+ basal cells were 182 apparent and comparable between EedcKO knockouts and controls (Figure S1B). IBCs 183 formed at 3hpi or 6hpi were also comparable between the knockouts and controls 184 (Figures S2A). Interestingly, the urine bacterial counts were lower in the knockouts 185 (Figure 2C). EedcKO bladders appeared to have fewer infiltrating immune cells when 186 compared to controls (Figure 2A). To quantify the difference, we analyzed bladder 187 inflammation using the criteria based on immune cell infiltration, edema, location of

7 bioRxiv preprint doi: https://doi.org/10.1101/2020.10.22.350942; this version posted October 22, 2020. The copyright holder for this preprint (which 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 available under aCC-BY-NC-ND 4.0 International license.

188 neutrophils infiltration and loss of urothelium(46). The overall inflammatory score of Eed 189 knockouts was approximately 2-fold less when compared to controls (Figure 2B). At 2 190 hour-post-inoculation (hpi), acute infection resulted in a rapid upregulation of 191 inflammatory regulators including Il-6, suppressor of cytokine signaling-3 (Socs3), C-C 192 motif chemokine ligand 2 (Ccl2)/monocyte chemoattractant protein 1 (MCP1), C-X-C 193 motif chemokine ligand 1 (Cxcl1)/neutrophil-activating protein 3 (NAP-3)/(KC) and 194 chemokine (C-X-C motif) receptor 2 (Cxcr2) in the bladder of both Eed knockouts and 195 controls. Consistent with the observation that EedcKO mice had lower inflammatory score, 196 expression of Il-6, Socs3, Ccl2, Cxcr2 and chemokine (C-C motif) receptor 2 (Ccr2) and 197 chemokine (C-X-C motif) receptor 2 (Cxcr2) were significantly lower in the knockouts than 198 in controls at 24hpi (Figure 2D). Expression of Cxcl1 was significantly lower at both 2hpi 199 and 24hpi in the mutants. 200 Bladder infection results in a rapid nuclear accumulation of NF-kB in superficial cells 201 and, subsequently, in intermediate and basal cells(30). NF-kB is a central transcription 202 regulator that activates expression of a number of gene targets including cytokines and 203 chemokines(47). Inhibitor of kB (IkB) binds and sequesters NF-kB in the cytoplasm, 204 thereby inhibiting the NF-kB-dependent inflammatory response(48). We found that 205 expression of IkBa was significantly upregulated in the bladder urothelium of Eed 206 knockout mice than in controls (Figure 3A). Consistently, anti-IkBa immunoreactivity was 207 markedly stronger in EedcKO mice compared to controls (Figure 3B). Chromatin 208 immunoprecipitation (ChIP) further showed that the occupancy of a repression histone 209 mark H3K27me3 was significantly reduced, while the epigenetic activation mark 210 H3K4me3 was significantly enriched at the IkBa promoter region of EedcKO mice (Figure 211 3C). Furthermore, while many superficial cells with strong nuclear NF-kB staining were 212 observed in controls, fewer were found in EedcKO mice (Figure 3D). Taken together, these 213 observations suggest that the crosstalk between PRC2 and NF-kB molecular pathways 214 promotes the inflammatory host response during bladder infections. 215 216 A history of UTI heightens the inflammatory host response 217 To examine the impact of a prior history of bladder infections, mice were inoculated 218 repeatedly for up to 7 times with a 7-day convalescence after each infection (Figure 4A).

8 bioRxiv preprint doi: https://doi.org/10.1101/2020.10.22.350942; this version posted October 22, 2020. The copyright holder for this preprint (which 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 available under aCC-BY-NC-ND 4.0 International license.

219 Mice developed bacteriuria after the first infection with approximately 106 colony forming 220 units (CFU) of bacteria in the urine at 1dpi (Figure 4B, 1xUTI, 1dpi, Ctr). The second 221 infection (2xUTI) resulted in a significantly lower urine bacterial count (~103 CFU, p<0.05). 222 However, additional reinfections (3xUTI-7xUTI) resulted no further decrease in bacteriuria, 223 similar to previous reports using different UPEC and mouse strains(49-51). unlike the 224 controls that were inoculated repeatedly with saline, histological analyses showed that 225 repeated infections increased the inflammatory score and caused severe urothelial 226 hyperplasia of the bladder (Figures 4C-E and S2B, Ctr). Expression of Krt20 and 227 Uroplakin 3a (Upk3a) were undetectable at 1, 3 and 7dpi after repeated infections 228 (Figures 4E and S2C, Ctr). The Krt5+ basal cells were expanded and occupied the entire 229 thickness of the bladder urothelium, strongly suggesting that the repeated infections 230 caused exfoliation of both intermediate and superficial cells. Collectively, these findings 231 suggest that a history of UTI heightens the inflammatory host response to reinfections 232 and causes pathological remodeling of the bladder urothelium, including basal cell 233 hyperplasia and a lack of terminal superficial cell differentiation. 234 235 Eed deletion improves outcomes for both acute and repeated infections 236 Compared to controls, urinalysis showed that the bacterial counts were significantly 237 lower in EedcKO mice at 1dpi after 1xUTI (Figure 4B). The same trend persisted during 238 repeated infections (2xUTI to 7xUTI). Urine bacteria were reduced to almost undetectable 239 in EedcKO mice at 1dpi and 7dpi during reinfections (2-7x UTIs). Eed knockout bladders 240 had significantly fewer infiltrating immune cells, less edema and milder urothelial 241 hyperplasia when compared with controls (Figures 4C-E and S2B-D). The overall 242 inflammatory scores were also lower in the knockouts than controls (Figure 4D). Urothelial 243 cell proliferation, as indicated by both pHH3 and Ki67 markers, was significantly 244 attenuated in EedcKO mice compared to controls (Figures 4F, S2C and D). Krt20 245 immunoreactivity was not detected at 1dpi and 3dpi in Eed knockouts (Figure 4E), 246 indicating a complete superficial cell exfoliation after repeated infections. However, 247 abundant Krt20+ superficial cells were observed at 7dpi in the knockouts but not in 248 controls (Figure 4E). Consistently, abundant Upk3a+ intermediate and superficial cells 249 were observed in EedcKO mice but not in controls (Figure S2C). Thus, genetic inactivation

9 bioRxiv preprint doi: https://doi.org/10.1101/2020.10.22.350942; this version posted October 22, 2020. The copyright holder for this preprint (which 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 available under aCC-BY-NC-ND 4.0 International license.

250 of PRC2 in the bladder urothelium improves outcomes of both the acute and repeated 251 bladder infections. 252 253 Eed deletion improves outcome of the C+UTI superinfection 254 Mice primed with cyclophosphamide (CPP) one day before the acute infection, the 255 CPP plus UTI (C+UTI) superinfection(24), have severe outcomes including high levels of 256 urine bacterial counts that persisted for more than four weeks (Figures 5A and S3A). The 257 superinfected bladders had apparent edema, urothelial damage and basal cell 258 hyperplasia with an inflammatory score of 5 (Figures 5B and C). Krt20 and Upk3a were 259 undetectable at 1dpi and the urothelium consisted entirely of Krt5+ basal cells (Figure 5D 260 and S3B). While Upk3a expression began to recover at 7dpi after the C+UTI 261 superinfection, expression of Krt20 was not detected before 14dpi in controls. High 262 degrees of basal cell proliferation were observed at all stages analyzed (Figures S3C and 263 D). 264 Compared to controls, the Eed knockout mice had significantly lower urine bacterial 265 burden at 1dpi; and bacteriuria was largely cleared within two weeks of the C+UTI 266 superinfection (Figure 5A). Inflammatory score was also lower in EedcKO mice (Figure 5B 267 and C). Exfoliation of Krt20+ superficial cells was observed in both the knockouts and 268 controls at 1dpi, implying that the knockout mice had sustained a similar level of the initial 269 urothelial damage. However, two weeks later at 14dpi, Krt20 expression began to recover 270 in the knockouts but not in controls (Figure 5D). Similarly, Upk3a expression was absent 271 from controls but was clearly detected in the knockouts (Figure S3B). Basal cell 272 hyperplasia was milder in the knockouts, which was coupled with a significantly lower cell 273 proliferation rate at all stages examined from 1dpi to 14dpi (Figures S3C and D). Together, 274 these findings suggest that conditional knockout of Eed in the bladder urothelium 275 dampens the inflammatory host response while improving bacterial clearance and 276 urothelial regeneration in bladder superinfection. 277 To better understand the PRC2-dependent inflammatory program, we performed 278 whole genome transcriptomic analysis of the urothelium of superinfected bladders at 1dpi. 279 We found that 124 genes were significantly downregulated and 62 genes were 280 significantly upregulated in EedcKO mice in comparison with controls (Figure S3E). Among

10 bioRxiv preprint doi: https://doi.org/10.1101/2020.10.22.350942; this version posted October 22, 2020. The copyright holder for this preprint (which 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 available under aCC-BY-NC-ND 4.0 International license.

281 the DEGs, a list of inflammatory markers including Cxcl2, Cxcl3, Ccl2, Ccl3, Il1β, Il11 and 282 Ptgs2 (aka Cox2) were significantly downregulated in EedcKO mice (Figure S3E), 283 consistent with the findings that the inflammatory score was lower. Furthermore, IPA 284 analysis revealed that molecular pathways such as the acute phase inflammation and Il- 285 8 neutrophil chemotaxis were significantly reduced (Figure S3F). Among the top predicted 286 upstream regulators, induces such as lipopolysaccharides (LPS), Il-1b, Tumor necrosis 287 factor alpha (Tnfa), Interferon gamma (Infg) and Il-6 were significantly suppressed while 288 inflammatory modulators including Kruppel-like factor 2 (KLF2)(52), Nr3c1 (aka. 289 Glucocorticoid receptor)(22), Zinc finger protein 36 (Zfp36)(53) and Heme Oxygenase 1 290 (Hmox1)(54) were significantly enhanced. Particularly noteworthy were suppression of 291 Cox2 and activation of glucocorticoid molecular pathways (Figure 5E); mice treated with 292 either Cox2 inhibitor or glucocorticoid agonist significantly reduces chronic cystitis(22, 23). 293 Collectively, these findings strongly suggest that the PRC2-dependent epigenetic 294 program in bladder urothelium exacerbates the inflammatory host response. 295 296 Small molecule Ezh2 inhibitors mitigate the C+UTI superinfection 297 To further explore the role of Ezh2 in bladder infection, we examined the potential 298 beneficial effects of Ezh2 small molecule inhibitors DZNep and EPZ005687, both of which 299 reduce H3K27me3 and tumor growth in vivo(55). Because of the long half-life of 300 H3K27me3 but short half-life of the inhibitors(39, 56), mice were pretreated with three 301 doses of the inhibitors to deplete H3K27me3 before bacterial inoculation and 2 extra 302 doses post-inoculation (Figure 6A). DZNep, a potent inhibitor of methyltransferases 303 including Ezh2(56), markedly reduced H3K27me3 in the bladder (Figure 6B). DZNep 304 treatment resulted in a mild but significant reduction of urine bacteria compared with the 305 vehicle treatment in the C+UTI superinfection (Figure 6C). Bacterial counts in the bladder 306 and kidneys were not affected by DZNep treatment (Figure S4A). Unlike the vehicle 307 controls, the DZNep-treated mice displayed fewer infiltrating neutrophils, mild basal 308 hyperplasia, blunted expression of Ezh2, and increased expression of Upk3a and Krt20 309 (Figures 6D-G and S4B). EPZ005687 is an Ezh2-specific inhibitor(57), which did not 310 affect UPEC growth or variability in vitro, nor the bacterial counts in the bladder or kidneys 311 (Figure S4C-E). It did, however, decrease modestly bacteriuria at 3dpi (Figure S4D).

11 bioRxiv preprint doi: https://doi.org/10.1101/2020.10.22.350942; this version posted October 22, 2020. The copyright holder for this preprint (which 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 available under aCC-BY-NC-ND 4.0 International license.

312 EPZ005687 treatment also decreased the NF-kB pathway activity as indicated by fewer 313 cells with nuclear p65 staining (Figure S4F) and a reduction of the inflammatory score 314 (Figures 6H and S4G). Expression of the urothelial cell differentiation marker Upk3a was 315 also preserved (Figure 6I), and concordantly urothelial proliferation was significantly 316 reduced in the EPZ005687-treated mice (Figure 6J). Collectively, these findings suggest 317 that Ezh2 inhibitors ameliorate effects of bladder superinfection. 318 319 Ezh2 small molecule inhibitors improve outcomes of chronic cystitis 320 Unlike C57Bl6 mice, C3H/HeN mice are susceptible to persistent bacteriuria or 321 chronic cystitis(19, 22, 23). We therefore used this mouse strain to examine whether Ezh2 322 inhibitors may decrease the risk of chronic infection (Figure 7A). DZNep reduced 323 H3K27me3 levels in the bladder of wildtype C3H/HeN mice (Figure 7B). However, it did 324 not have major impact on the urine bacterial counts in C3H/HeN mice (Figure S5A). 325 Nevertheless, neutrophil infiltration and inflammatory score were significantly lower in the 326 DZNep-treated mice (Figures 7C and D). Apparent bladder edema, neutrophil infiltration 327 and urothelial hyperplasia were found among the vehicle-treated controls at 10dpi (Figure 328 7C). These histological observations, however, were rarely detected in DZNep-treated 329 mice. Consistently, high levels of Upk3a expression were detected in DZNep-treated mice 330 but not vehicle-treated controls (Figure 7E). Similar to DZNep, EPZ005687 significantly 331 blunted the induction of Ezh2, reduced H3K27me3 levels and decreased the overall 332 inflammatory scores in C3H/HeN mice (Figures 7F-I and S5D). These mice also exhibited 333 a mild basal cell hyperplasia phenotype with a concordant increase of terminal 334 differentiation marker Upk3a. EPZ005687 alone had minimal effects on the bacterial 335 counts in the urine, bladder and kidneys, although bacteriuria was moderately lower at 336 1dpi (Figures S5B and C). Taken together, these findings suggest that Ezh2 inhibitors 337 decrease inflammation and increase urothelial cell differentiation in chronic cystitis. 338 339 Discussion 340 341 This study uncovers a previously unknown role of the PRC2-dependent epigenetic 342 repression program in regulating the inflammatory host response to bacterial infections in

12 bioRxiv preprint doi: https://doi.org/10.1101/2020.10.22.350942; this version posted October 22, 2020. The copyright holder for this preprint (which 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 available under aCC-BY-NC-ND 4.0 International license.

343 the bladder. Findings here suggest that the bacteria-induced epigenetic remodeling of the 344 bladder urothelium exacerbates inflammation and urothelial injury, thereby leading to poor 345 outcomes of bladder infections. Collectively, these findings highlight the possibility of 346 epigenetic basis of mucosal imprint and further suggest that targeting the underlying 347 PRC2-dependent epigenetic program offers a non-antibiotic strategy to ameliorate UTIs. 348 UPEC induce expression of a list of epigenetic regulators in the bladder. Among them, 349 Ezh2 is significantly upregulated in both the bladder urothelium and underlying 350 mesenchymal cells. By conditionally deleting Eed in the bladder urothelium, a strategy to 351 inactivate PRC2 methyltransferase activity mediated by both Ezh1 and Ezh2, our findings 352 strongly suggest that the PRC2-dependent epigenetic program in the urothelial cells plays 353 a critical role in facilitating the heightened inflammatory host response to bladder 354 infections. Ezh2 induction is tightly linked to the increased expression of inflammatory 355 cytokines and chemokine, severe basal cell hyperplasia and impairment of urothelial cell 356 terminal differentiation. While the underlying mechanism of PRC2-mediated inflammatory 357 response in the bladder remains to be defined, our findings suggest that it promotes the 358 NF-kB signaling pathway by epigenetic repression of the IkBa locus. Indeed, the UTI- 359 induced NF-kB activation, as indicated by NF-kB nuclear localization, is markedly 360 reduced in EedcKO mice. In addition to the PRC2-dependent histone methylation, Ezh2 361 may also function in the methyltransferase-independent manner to enhance NF-kB 362 activity(58). It is worth noting that NF-kB directly regulates Ezh2 expression in human 363 fibroblasts(59), suggesting a feedback regulation between Ezh2 and NF-kB molecular 364 pathways may exaggerate and extend inflammatory host response to bladder infections. 365 Importantly, the small molecule Ezh2 inhibitors significantly lessened bladder 366 inflammatory responses to the acute, chronic and super infections. The treated mice have 367 improved outcomes to bladder infections including decreased bacteriuria and enhanced 368 urothelial regeneration. While the design of this study could not differentiate the 369 prophylactic and therapeutic effects, these findings nevertheless provide the proof of 370 concept evidence that targeting the PRC2-dependent epigenetic program offers a non- 371 antibiotic adjuvant strategy to ameliorate UTIs. 372 Epigenetic activation has been linked to the trained immunity-like inflammatory 373 memory in epithelial cells including the skin and airway epithelial progenitors(60, 61).

13 bioRxiv preprint doi: https://doi.org/10.1101/2020.10.22.350942; this version posted October 22, 2020. The copyright holder for this preprint (which 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 available under aCC-BY-NC-ND 4.0 International license.

374 However, epigenetic repression in modulating inflammatory host response to bacterial 375 infections in the bladder has not been reported until now. Given the quiescent nature of 376 bladder urothelium and how the PRC-dependent H3K27me3 modification persists 377 through cell cycles and generations, we speculate that the PRC2-dependent epigenetic 378 reprograming will have a lasting impact on the pathophysiology of the bladder urothelium. 379 The virulence factor a-hemolysin is required in attenuating host Akt/protein kinase B 380 signaling pathway during UPEC infection (62) and, at the a-hemolysin-dependent manner, 381 UPEC also abrogate global histone H3 and H4 acetylation in Sertoli cells to promoter cell 382 survival(63). In addition to a-hemolysin, other pore-forming bacterial toxins have been 383 reported to mediate dephosphorylation of histone H3 and deacetylation of histone H4(34). 384 Hultgren and colleagues have previously shown that a primary bladder infection leaves 385 behind a molecular imprint on the bladder tissue that lasts for more than seven months 386 and, importantly, sensitize recurrent infections(19). Taken together, these findings 387 suggest, for the first time, that the PRC2-dependent epigenetic remodeling may contribute 388 to the bacteria-induced mucosal imprint to sensitize recurrent diseases. 389 Heightened inflammatory response to bladder reinfections has been previously 390 observed and largely attributed to the adaptive immunity(49-51). It is therefore a surprise 391 to find that the urothelial cells also contribute to the heighten inflammatory response. 392 Intriguingly, the PRC2-dependent inflammatory response in the urothelial progenitor cells 393 appears to be at the expense of terminal differentiation of superficial cells. Loss of the 394 superficial cell layer grants the invading pathogens access to deeper bladder tissue and 395 predisposes the host to chronic cystitis(29). On the one hand, the UTI-induced 396 immunological memory in urothelial cells accelerates bacterial clearance by a heightened 397 inflammatory response; on the other hand, it increases the risk of chronic infection due to 398 the impaired urothelial homeostasis. UPEC have evolved multiple strategies to subvert 399 the host’s defense(64), including intracellular localization(26, 27, 29), filamentation(65), 400 attenuation of cytokine production(66-68), killing of natural killer cells(69), limiting antigen 401 presentation(50), and promoting cell survival(63). Our findings here provide the initial 402 evidence suggesting that UPEC have also taken advantage of epigenetic remodeling of 403 the bladder urothelial cells to regulate host innate immunity to benefit the pathogens.

14 bioRxiv preprint doi: https://doi.org/10.1101/2020.10.22.350942; this version posted October 22, 2020. The copyright holder for this preprint (which 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 available under aCC-BY-NC-ND 4.0 International license.

404 In the repeated infection model, EedcKO mice appear to have partial resistance to 405 reinfections while maintaining urothelial tissue homeostasis (Figure 4). While the 406 underlying mechanism remains to be defined, we have noted that the aryl hydrocarbon 407 receptor (AhR) signaling pathway is significantly upregulated in Eed mutant urothelium 408 (Figure 1). AhR is a ligand-dependent transcription factor that is able to detect bacterial 409 pigments and regulate antibacterial defense(70). AhR regulates expression of several 410 LPS-responsive genes(71), and mitigates inflammatory response to primary LPS 411 challenge(72). On LPS rechallenge, however, the AhR-dependent mechanism promotes 412 an establishment of a disease tolerance state that resists bacterial infection and 413 preserves tissue homeostasis(72). These observations suggest an alternative 414 mechanism through which targeting PRC2 may mitigate symptoms of UTIs by improving 415 bladder urothelial homeostasis. 416 Injuries awaken the quiescent bladder urothelium, causing urothelial progenitor cell 417 proliferation and differentiation to repair the damage(8). UTI-induced Ezh2 expression 418 coincides with reactivation of the cell cycle of urothelial progenitor cells. The basal cell 419 proliferation rate is significantly reduced in the urothelium-specific Eed mutants during 420 acute, repeated and superinfections, consistent with the previous observation that Ezh2 421 expression is closely associated with cell proliferation during bladder urothelial 422 development(9). Pathway analysis of the differentially expressed genes in the bladder 423 urothelium between EedcKO mice and controls suggest that activity of the p53 tumor 424 suppression pathway is significantly elevated in EedcKO mice. Consistent with these 425 observations, conditional knockout of Kdm6a in the bladder urothelial cells, a histone 426 demethylase that catalyzes an opposing biochemical reaction of PRC2 methyltransferase, 427 results in a significant decrease in the p53 tumor suppressor pathway activity(73). 428 Moreover, loss-of-function of Kdm6a significantly increases bladder cancer risk(73). Thus, 429 the PRC2/Kdm6a-p53 axis regulates urothelial progenitor cell proliferation and 430 tumorigenesis. Collectively, the PRC2-dependent epigenetic program may play a dual 431 role in urothelial progenitor cells – it enhances inflammatory host response through the 432 NF-kB signaling pathway and, at the same time, controls urothelial regeneration through 433 the p53 tumor suppressor pathway by promoting proliferation of basal progenitor cells 434 and inhibiting differentiation of superficial cells.

15 bioRxiv preprint doi: https://doi.org/10.1101/2020.10.22.350942; this version posted October 22, 2020. The copyright holder for this preprint (which 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 available under aCC-BY-NC-ND 4.0 International license.

435 In summary, the findings here suggest that UTI induces the PRC2-dependent 436 epigenetic program in the bladder urothelium, which exaggerates inflammation and 437 proliferation at the expense of differentiation and tissue homeostasis, and further suggest 438 an non-antibiotic strategy to ameliorate UTIs by inhibiting the methyltransferase activity 439 of PRC2. This study also raises several questions regarding mucosal imprint in the 440 bladder. Future studies are needed to examine whether the immune and urothelial cells 441 may share the similar or divergent roles to modulate the immunity; whether and how the 442 UTI-induced epigenetic reprograming can be reversed to reduce the risk of bladder 443 diseases; and whether targeting Ezh2 enzymatic activity may reduce the risk and/or 444 mitigate UTIs. 445

16 bioRxiv preprint doi: https://doi.org/10.1101/2020.10.22.350942; this version posted October 22, 2020. The copyright holder for this preprint (which 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 available under aCC-BY-NC-ND 4.0 International license.

446 Materials and Methods 447 448 Mouse lines including EedF/F, UpkIICre and ShhCre have been described previously in 449 detail(9). CAG-TRAP mice were kindly provided by Dr. William T Pu, Boston Children's 450 Hospital and available at The Jackson Laboratory (Jackson Laboratory, 022367). 451 C57BL/6J mice (000664) were obtained from The Jackson Laboratory; And C3H/HeNCrl 452 mice (025) were obtained from Charles River. DZNep (1.5mg/kg, Selleckchem, S7120), 453 EPZ005687 (2.5mg/kg, APExBIO, A4171) and cyclophosphamide (CPP, 300mg/kg, 454 Sigma C7397) were administered to mice via intraperitoneal (i.p.) injection. Female mice 455 between 6-8 weeks of age were used in the bladder infections as previously 456 described(44). The uropathogenic Escherichia coli UTI89 strain was generously provided 457 by Dr. Scott Hultgren (Washington University, St. Louis, USA). For the C+UTI 458 superinfection model, mice received a single dose of cyclophosphamide (CPP, 300mg/kg, 459 Sigma C7397, i.p.) 1 day prior to UTI89 inoculation. All animal studies were performed 460 according to protocols reviewed and approved by the Institutional Animal Care and Use 461 Committee at Boston Children’s Hospital. 462 463 464

17 bioRxiv preprint doi: https://doi.org/10.1101/2020.10.22.350942; this version posted October 22, 2020. The copyright holder for this preprint (which 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 available under aCC-BY-NC-ND 4.0 International license.

465 Author contributions 466 XL, ZB and CG designed the research studies. CG, MZ, XS, ZB and SZ conducted 467 experiments and acquired data. XL, CG and MZ analyzed data. XL wrote the manuscript 468 with the help from CG. 469 470

18 bioRxiv preprint doi: https://doi.org/10.1101/2020.10.22.350942; this version posted October 22, 2020. The copyright holder for this preprint (which 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 available under aCC-BY-NC-ND 4.0 International license.

471 Acknowledgments 472 473 We thank all members of the X.L. lab, particularly Drs. Satoshi Kaneko, Chao Yang and 474 Christa Lam for technical supports and helpful discussions. We appreciate Drs. Rosalyn 475 Adam and Eric Greer for critical and insightful comments on the manuscripts. This work 476 was supported by NIH/NIDDK (1R01DK091645-01A1, X.L.), NIH/NCI (R21CA249701, 477 X.L) and NHLBI (1R01HL136921, X.L.). 478

19 bioRxiv preprint doi: https://doi.org/10.1101/2020.10.22.350942; this version posted October 22, 2020. The copyright holder for this preprint (which 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 available under aCC-BY-NC-ND 4.0 International license.

479 References: 480 481 1. Ramirez-Prado JS, Abulfaraj AA, Rayapuram N, Benhamed M, & Hirt H (2018) 482 Plant Immunity: From Signaling to Epigenetic Control of Defense. Trends Plant 483 Sci 23(9):833-844. 484 2. Agrawal AA, Laforsch C, & Tollrian R (1999) Transgenerational induction of 485 defences in animals and plants. Nature 401:60. 486 3. Little TJ, O'Connor B, Colegrave N, Watt K, & Read AF (2003) Maternal transfer 487 of strain-specific immunity in an invertebrate. Curr Biol 13(6):489-492. 488 4. Netea MG, Quintin J, & van der Meer JW (2011) Trained immunity: a memory for 489 innate host defense. Cell Host Microbe 9(5):355-361. 490 5. Lau CM & Sun JC (2018) The widening spectrum of immunological memory. 491 Current opinion in immunology 54:42-49. 492 6. Netea MG, Schlitzer A, Placek K, Joosten LAB, & Schultze JL (2019) Innate and 493 Adaptive Immune Memory: an Evolutionary Continuum in the Host's Response to 494 Pathogens. Cell Host Microbe 25(1):13-26. 495 7. Hicks RM (1975) The mammalian urinary bladder: an accommodating organ. 496 Biological reviews of the Cambridge Philosophical Society 50(2):215-246. 497 8. Balsara ZR & Li X (2017) Sleeping beauty: awakening urothelium from its 498 slumber. Am J Physiol Renal Physiol 312(4):F732-F743. 499 9. Guo C, Balsara ZR, Hill WG, & Li X (2017) Stage- and subunit-specific functions 500 of polycomb repressive complex 2 in bladder urothelial formation and 501 regeneration. Development 144(3):400-408. 502 10. Papafotiou G, et al. (2016) KRT14 marks a subpopulation of bladder basal cells 503 with pivotal role in regeneration and tumorigenesis. Nat Commun 7:11914. 504 11. Gandhi D, et al. (2013) Retinoid signaling in progenitors controls specification 505 and regeneration of the urothelium. Dev Cell 26(5):469-482. 506 12. Shin K, et al. (2011) Hedgehog/Wnt feedback supports regenerative proliferation 507 of epithelial stem cells in bladder. Nature 472(7341):110-114. 508 13. Mysorekar IU, Isaacson-Schmid M, Walker JN, Mills JC, & Hultgren SJ (2009) 509 Bone morphogenetic protein 4 signaling regulates epithelial renewal in the 510 urinary tract in response to uropathogenic infection. Cell Host Microbe 5(5):463- 511 475. 512 14. Flores-Mireles AL, Walker JN, Caparon M, & Hultgren SJ (2015) Urinary tract 513 infections: epidemiology, mechanisms of infection and treatment options. Nature 514 reviews. Microbiology 13(5):269-284. 515 15. Foxman B (2014) Urinary tract infection syndromes: occurrence, recurrence, 516 bacteriology, risk factors, and disease burden. Infect Dis Clin North Am 28(1):1- 517 13. 518 16. Hooton TM, et al. (1996) A prospective study of risk factors for symptomatic 519 urinary tract infection in young women. The New England journal of medicine 520 335(7):468-474. 521 17. Scholes D, et al. (2000) Risk factors for recurrent urinary tract infection in young 522 women. The Journal of infectious diseases 182(4):1177-1182.

20 bioRxiv preprint doi: https://doi.org/10.1101/2020.10.22.350942; this version posted October 22, 2020. The copyright holder for this preprint (which 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 available under aCC-BY-NC-ND 4.0 International license.

523 18. O'Brien VP, Hannan TJ, Schaeffer AJ, & Hultgren SJ (2015) Are you 524 experienced? Understanding bladder innate immunity in the context of recurrent 525 urinary tract infection. Current opinion in infectious diseases 28(1):97-105. 526 19. O'Brien VP, et al. (2016) A mucosal imprint left by prior Escherichia coli bladder 527 infection sensitizes to recurrent disease. Nature microbiology 2:16196. 528 20. Hansson S, et al. (1990) Follicular cystitis in girls with untreated asymptomatic or 529 covert bacteriuria. J Urol 143(2):330-332. 530 21. Schlager TA, LeGallo R, Innes D, Hendley JO, & Peters CA (2011) B cell 531 infiltration and lymphonodular hyperplasia in bladder submucosa of patients with 532 persistent bacteriuria and recurrent urinary tract infections. J Urol 186(6):2359- 533 2364. 534 22. Hannan TJ, Mysorekar IU, Hung CS, Isaacson-Schmid ML, & Hultgren SJ (2010) 535 Early severe inflammatory responses to uropathogenic E. coli predispose to 536 chronic and recurrent urinary tract infection. PLoS Pathog 6(8):e1001042. 537 23. Hannan TJ, et al. (2014) Inhibition of Cyclooxygenase-2 Prevents Chronic and 538 Recurrent Cystitis. EBioMedicine 1(1):46-57. 539 24. Lyon D, Howard EB, & Montgomerie JZ (1982) Increased severity of urinary tract 540 infection and bacteremia in mice with urinary bladder injury induced by 541 cyclophosphamide. Infect Immun 38(2):558-562. 542 25. Schwartz DJ, Conover MS, Hannan TJ, & Hultgren SJ (2015) Uropathogenic 543 Escherichia coli superinfection enhances the severity of mouse bladder infection. 544 PLoS Pathog 11(1):e1004599. 545 26. Mulvey MA, et al. (1998) Induction and evasion of host defenses by type 1- 546 piliated uropathogenic Escherichia coli. Science 282(5393):1494-1497. 547 27. Justice SS, et al. (2004) Differentiation and developmental pathways of 548 uropathogenic Escherichia coli in urinary tract pathogenesis. Proc Natl Acad Sci 549 U S A 101(5):1333-1338. 550 28. Nagamatsu K, et al. (2015) Dysregulation of Escherichia coli alpha-hemolysin 551 expression alters the course of acute and persistent urinary tract infection. Proc 552 Natl Acad Sci U S A 112(8):E871-880. 553 29. Mysorekar IU & Hultgren SJ (2006) Mechanisms of uropathogenic Escherichia 554 coli persistence and eradication from the urinary tract. Proc Natl Acad Sci U S A 555 103(38):14170-14175. 556 30. Liu Y, et al. (2015) Dual ligand/receptor interactions activate urothelial defenses 557 against uropathogenic E. coli. Scientific reports 5:16234. 558 31. Song J, et al. (2007) A novel TLR4-mediated signaling pathway leading to IL-6 559 responses in human bladder epithelial cells. PLoS Pathog 3(4):e60. 560 32. Samuelsson P, Hang L, Wullt B, Irjala H, & Svanborg C (2004) Toll-like receptor 561 4 expression and cytokine responses in the human urinary tract mucosa. Infect 562 Immun 72(6):3179-3186. 563 33. Agace WW, Hedges SR, Ceska M, & Svanborg C (1993) Interleukin-8 and the 564 neutrophil response to mucosal gram-negative infection. J Clin Invest 92(2):780- 565 785. 566 34. Hamon MA, et al. (2007) Histone modifications induced by a family of bacterial 567 toxins. Proc Natl Acad Sci U S A 104(33):13467-13472.

21 bioRxiv preprint doi: https://doi.org/10.1101/2020.10.22.350942; this version posted October 22, 2020. The copyright holder for this preprint (which 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 available under aCC-BY-NC-ND 4.0 International license.

568 35. Pereira JM, Hamon MA, & Cossart P (2016) A Lasting Impression: Epigenetic 569 Memory of Bacterial Infections? Cell Host Microbe 19(5):579-582. 570 36. van der Heijden C, et al. (2018) Epigenetics and Trained Immunity. Antioxid 571 Redox Signal 29(11):1023-1040. 572 37. Zheng Y, Tipton JD, Thomas PM, Kelleher NL, & Sweet SM (2014) Site-specific 573 human histone H3 methylation stability: fast K4me3 turnover. Proteomics 574 14(19):2190-2199. 575 38. Zenk F, et al. (2017) Germ line-inherited H3K27me3 restricts enhancer function 576 during maternal-to-zygotic transition. Science 357(6347):212-216. 577 39. Zee BM, et al. (2010) In vivo residue-specific histone methylation dynamics. J 578 Biol Chem 285(5):3341-3350. 579 40. Cao R, et al. (2002) Role of histone H3 lysine 27 methylation in Polycomb-group 580 silencing. Science 298(5595):1039-1043. 581 41. Kuzmichev A, Nishioka K, Erdjument-Bromage H, Tempst P, & Reinberg D 582 (2002) Histone methyltransferase activity associated with a human multiprotein 583 complex containing the Enhancer of Zeste protein. Genes Dev 16(22):2893- 584 2905. 585 42. Muller J, et al. (2002) Histone methyltransferase activity of a Drosophila 586 Polycomb group repressor complex. Cell 111(2):197-208. 587 43. Ting K, et al. (2016) Uropathogenic E. coli (UPEC) Infection Induces Proliferation 588 through Enhancer of Zeste Homologue 2 (EZH2). PLoS One 11(3):e0149118. 589 44. Hung CS, Dodson KW, & Hultgren SJ (2009) A murine model of urinary tract 590 infection. Nat Protoc 4(8):1230-1243. 591 45. Zhou P, et al. (2013) Interrogating translational efficiency and lineage-specific 592 transcriptomes using ribosome affinity purification. Proc Natl Acad Sci U S A 593 110(38):15395-15400. 594 46. Hopkins WJ, Gendron-Fitzpatrick A, Balish E, & Uehling DT (1998) Time course 595 and host responses to Escherichia coli urinary tract infection in genetically 596 distinct mouse strains. Infect Immun 66(6):2798-2802. 597 47. Ghosh S & Karin M (2002) Missing pieces in the NF-kappaB puzzle. Cell 109 598 Suppl:S81-96. 599 48. Hinz M, Arslan SC, & Scheidereit C (2012) It takes two to tango: IkappaBs, the 600 multifunctional partners of NF-kappaB. Immunological reviews 246(1):59-76. 601 49. Thumbikat P, Waltenbaugh C, Schaeffer AJ, & Klumpp DJ (2006) Antigen- 602 specific responses accelerate bacterial clearance in the bladder. J Immunol 603 176(5):3080-3086. 604 50. Mora-Bau G, et al. (2015) Macrophages Subvert Adaptive Immunity to Urinary 605 Tract Infection. PLoS Pathog 11(7):e1005044. 606 51. O'Brien VP, Dorsey DA, Hannan TJ, & Hultgren SJ (2018) Host restriction of 607 Escherichia coli recurrent urinary tract infection occurs in a bacterial strain- 608 specific manner. PLoS Pathog 14(12):e1007457. 609 52. Das H, et al. (2006) Kruppel-like factor 2 (KLF2) regulates proinflammatory 610 activation of monocytes. Proc Natl Acad Sci U S A 103(17):6653-6658. 611 53. Moore MJ, et al. (2018) ZFP36 RNA-binding proteins restrain T cell activation 612 and anti-viral immunity. eLife 7.

22 bioRxiv preprint doi: https://doi.org/10.1101/2020.10.22.350942; this version posted October 22, 2020. The copyright holder for this preprint (which 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 available under aCC-BY-NC-ND 4.0 International license.

613 54. Sebastian VP, et al. (2018) Heme Oxygenase-1 as a Modulator of Intestinal 614 Inflammation Development and Progression. Frontiers in immunology 9:1956. 615 55. Pan M, et al. (2016) Regional glutamine deficiency in tumours promotes 616 dedifferentiation through inhibition of histone demethylation. Nat Cell Biol 617 18(10):1090-1101. 618 56. Kim KH & Roberts CW (2016) Targeting EZH2 in cancer. Nat Med 22(2):128- 619 134. 620 57. Knutson SK, et al. (2012) A selective inhibitor of EZH2 blocks H3K27 methylation 621 and kills mutant lymphoma cells. Nat Chem Biol 8(11):890-896. 622 58. Lee ST, et al. (2011) Context-specific regulation of NF-kappaB target gene 623 expression by EZH2 in breast cancers. Mol Cell 43(5):798-810. 624 59. Iannetti A, et al. (2014) Regulation of p53 and Rb links the alternative NF-kappaB 625 pathway to EZH2 expression and cell senescence. PLoS Genet 10(9):e1004642. 626 60. Naik S, et al. (2017) Inflammatory memory sensitizes skin epithelial stem cells to 627 tissue damage. Nature 550(7677):475-480. 628 61. Ordovas-Montanes J, et al. (2018) Allergic inflammatory memory in human 629 respiratory epithelial progenitor cells. Nature 560(7720):649-654. 630 62. Wiles TJ, Dhakal BK, Eto DS, & Mulvey MA (2008) Inactivation of host 631 Akt/protein kinase B signaling by bacterial pore-forming toxins. Molecular biology 632 of the cell 19(4):1427-1438. 633 63. Zhang Z, et al. (2016) Uropathogenic Escherichia coli Epigenetically Manipulate 634 Host Cell Death Pathways. The Journal of infectious diseases 213(7):1198-1207. 635 64. Olson PD & Hunstad DA (2016) Subversion of Host Innate Immunity by 636 Uropathogenic Escherichia coli. Pathogens 5(1). 637 65. Justice SS, Hunstad DA, Seed PC, & Hultgren SJ (2006) Filamentation by 638 Escherichia coli subverts innate defenses during urinary tract infection. Proc Natl 639 Acad Sci U S A 103(52):19884-19889. 640 66. Klumpp DJ, et al. (2001) Uropathogenic Escherichia coli potentiates type 1 pilus- 641 induced apoptosis by suppressing NF-kappaB. Infect Immun 69(11):6689-6695. 642 67. Hunstad DA, Justice SS, Hung CS, Lauer SR, & Hultgren SJ (2005) Suppression 643 of bladder epithelial cytokine responses by uropathogenic Escherichia coli. Infect 644 Immun 73(7):3999-4006. 645 68. Billips BK, et al. (2007) Modulation of host innate immune response in the 646 bladder by uropathogenic Escherichia coli. Infect Immun 75(11):5353-5360. 647 69. Gur C, et al. (2013) Natural killer cell-mediated host defense against 648 uropathogenic E. coli is counteracted by bacterial hemolysinA-dependent killing 649 of NK cells. Cell Host Microbe 14(6):664-674. 650 70. Moura-Alves P, et al. (2014) AhR sensing of bacterial pigments regulates 651 antibacterial defence. Nature 512(7515):387-392. 652 71. Kimura A, et al. (2009) Aryl hydrocarbon receptor in combination with Stat1 653 regulates LPS-induced inflammatory responses. The Journal of experimental 654 medicine 206(9):2027-2035. 655 72. Bessede A, et al. (2014) Aryl hydrocarbon receptor control of a disease tolerance 656 defence pathway. Nature 511(7508):184-190. 657 73. Kaneko S & Li X (2018) X chromosome protects against bladder cancer in 658 females via a KDM6A-dependent epigenetic mechanism. Sci Adv 4(6):eaar5598.

23 bioRxiv preprint doi: https://doi.org/10.1101/2020.10.22.350942; this version posted October 22, 2020. The copyright holder for this preprint (which 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 available under aCC-BY-NC-ND 4.0 International license.

659 Figure legends: 660 661 Figure 1. Bacterial infection induces Ezh2 gene expression to promote hyper- 662 proliferation of urothelial progenitor cells. (A) Schematic diagram of mouse models of 663 acute (1x) and repeated (2x-7x) bladder infections. (B) Quantitative RT-PCR analyses of 664 Ezh1 and Ezh2 mRNAs in the bladder urothelium at 24 hours after an acute infection. 665 Control mice (Ctr) were treated with saline (PBS). *, p<0.05, Student’s t-test. (C) Ezh2 666 and H3K27me3 immunostaining (red) of mouse bladders that have been infected either 667 acutely (1xUTI) or repeatedly (7xUTI). dpi, day-post-inoculation. Control mice were 668 treated with PBS. Blue, nuclear DAPI counter staining; LP, lamina propria; U, urothelium; 669 dashed line separates LP and U. (D and E) Ki67 immunostaining (red) of bladder sections 670 of urothelium-specific Eed conditional knockout mice (EedcKO) that are acutely infected 671 and analyzed at 1dpi. Littermate heterozygous mice are used as controls (Ctr). 672 Percentage of Ki67 positive (Ki67+) cells are quantified in D. *, p<0.05, Student’s t-test; 673 PBS group, n=3; 1xUTI group, n=3 (Ctr) and 5 (EedcKO). (F) Volcano blot of the 674 differentially expressed genes (DEGs, padj<0.05, n=2) in adult EedcKO urothelium. 675 676 Figure 2. Genetic inactivation of PRC2 significantly blunts inflammatory host 677 response to bladder infections. (A) Representative histological images of EedcKO 678 bladders one day post-inoculation (1xUTI and 1dpi) of UTI89 into the bladder. Control 679 (Ctr) mice were heterozygous littermates. Arrowheads, polymorphonuclear leukocytes; 680 LP, lamina propria; SM, smooth muscle; U, urothelium; dashed lines separate SM, LP 681 and U. (B) Inflammatory scores based on histological data as shown in A. *, p<0.05, 682 Student’s t-test. (C) Urine bacterial counts at 1dpi. *p<0.05, Mann-Whitney test. (D) 683 Quantitative RT-PCR analyses of expression of inflammatory markers in the bladder of 684 EedcKO mice at 2 hours and 24 hours post-inoculation. Heterozygous littermates were 685 used as controls (Ctr). Gapdh was used as an internal control. *, p<0.05, vs PBS with the 686 same genotype; #, p<0.05, vs control (Ctr) with the same treatments; Student’s t-test. 687 688 Figure 3. PRC2 enhances activity of the NF-kB signaling pathway in the bladder 689 urothelium during acute bacterial infection. (A) Quantitative analysis of IkBa and IkBb

24 bioRxiv preprint doi: https://doi.org/10.1101/2020.10.22.350942; this version posted October 22, 2020. The copyright holder for this preprint (which 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 available under aCC-BY-NC-ND 4.0 International license.

690 mRNAs in the bladder urothelium of control and EedcKO mice without infection. Gapdh 691 was used as an internal control. *, p<0.05, Student’s t-test. (B) Immunostaining of IkBa 692 (red) and Krt5 (green) of bladder sections from EedcKO mice without bacterial infection. 693 (C) Chromatin immunoprecipitation (ChIP) assay of IkBa locus (schema) using 694 H3K27me3- and H3K4me3-specific antibodies. TSS, transcription start site. *, p<0.05, 695 Student’s t-test. (D) Immunostaining of NF-kB p65 subunit (red) and Krt5 (red) of bladder 696 sections from EedcKO mice at 1 hour post inoculation (hpi). White arrowheads, nuclear 697 localized p65. Ctr, heterozygous littermates; PBS, uninfected mice; Green, Krt5+ basal 698 cells; blue, DAPI counter staining; LP, lamina propria; U, urothelium; dashed lines 699 separate U from LP. 700 701 Figure 4. Genetic inactivation of PRC2 in the bladder urothelium attenuates 702 inflammatory host response to the repeated bladder infections. (A) Schematic 703 diagram of repeated infections. (B) Urine bacterial counts during repeated infections. 704 *p<0.05, Mann-Whitney test. (C) Representative images of histological staining of bladder 705 tissues from EedcKO mice received 7xUTI at 1 day-post-inoculation (dpi). Ctr, littermate 706 heterozygous control. (D) Inflammatory scores based on histological data as shown in C. 707 *p<0.05, Student’s t-test. (E) Immunostaining of Krt5 (red) and Krt20 (green) of bladder

708 sections from EedcKO mice after 7xUTI. Blue, DAPI counter staining; LP, lamina propria; 709 U, urothelium; dashed lines separate U from LP. (F) Analysis of pHH3+ and Ki67+ cells in 710 the bladder urothelium of EedcKO and control mice after 7xUTI. *p<0.05, Student’s t-test. 711 712 Figure 5. Inactivation of PRC2 in the bladder urothelium attenuates inflammatory 713 host response to the C+UTI superinfection. (A) Urine bacterial counts of mice after the 714 C+UTI superinfection. *p<0.05, Mann-Whitney test, dpi, day-post-inoculation. (B) 715 Representative histological images of bladder sections from EedcKO mice one day after 716 the C+UTI superinfection. Ctr, littermate heterozygous mice; LP, lamina propria; SM, 717 smooth muscle; U, urothelium; dashed lines separate SM, LP and U; arrowheads, 718 polymorphonuclear leukocytes. (C) Inflammatory scores based on histological data as 719 shown in B. *p<0.05, Student’s t-test. (D) Immunostaining of Krt5 (red) and Upk3a (green) 720 of bladder sections from EedcKO mice after the C+UTI superinfection. Blue, DAPI counter

25 bioRxiv preprint doi: https://doi.org/10.1101/2020.10.22.350942; this version posted October 22, 2020. The copyright holder for this preprint (which 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 available under aCC-BY-NC-ND 4.0 International license.

721 staining; LP, lamina propria; U, urothelium; dashed lines separate U from LP. (E) The top 722 predicted Ingenuity Pathway Analysis (IPA) upstream regulators of the differentially 723 expressed genes (padj<0.05, n=2) identified from the bladder urothelium of EedcKO mice 724 at 1dpi after the C+UTI superinfection. 725 726 Figure 6. Ezh2 small molecule inhibitors blunt inflammatory host response and 727 improve outcomes of the C+UTI superinfections. (A) Schematic diagram of the 728 experimental regimen using C57Bl/6 mice. (B) H3K27me3 immunostaining in the bladder 729 urothelium of mice treated with DZNep. PBS, vehicle control; blue, DAPI nuclear counter 730 staining. LP, lamina propria; U, urothelium; dashed lines separate U from LP. (C) Urine 731 bacterial counts at 1 dpi. *p<0.05, Mann-Whitney test. (D) Representative histological 732 images of bladder sections from C57Bl/6J mice 7 days after the superinfections. SM, 733 smooth muscle; arrowheads, polymorphonuclear leukocytes. (E-G) Immunostaining of 734 Ezh2 (E, purple), Krt20 and Krt5 (F, green and red, respectively) and Upk3a and Krt5 (G, 735 red and green, respectively) of the bladder urothelium of superinfected mice at 7dpi. (H) 736 Inflammatory scores of the superinfected mice treated with Ezh2-specific inhibitor 737 EPZ005687. *p<0.05, Student’s t-test, n=5. (I) Immunostaining of Upk3a (red) and Krt5 738 (green) of the bladder urothelium of superinfected mice at 7dpi. EPZ, EPZ005687; blue, 739 DAPI nuclear counter staining. (J) Percentage of Ki67+ proliferating cells in the bladder 740 urothelium of the EPZ005687-treated and super-infected mice. *p<0.05, Student’s t-test. 741 742 Figure 7. Ezh2 small molecule inhibitors blunt inflammatory host response and 743 improve outcomes of the chronic cystitis. (A) Schematic diagram of the experimental 744 regimen using C3H/HeN mice. (B) H3K27me3 immunostaining in the bladder urothelium 745 of mice at 10dpi. Blue, DAPI nuclear counter staining; LP, lamina propria; U, urothelium; 746 dashed lines separate U from LP. (C) Representative histological images of bladder 747 tissues from C3H/HeN mice at 10dpi. SM, smooth muscle. Arrowheads, 748 polymorphonuclear leukocytes. (D) Neutrophil infiltration and inflammatory score based 749 on histological data as shown in C. *p<0.05, Student’s t-test. (E) Krt5 (red) and Upk3a 750 (green) immunostaining of the bladder sections at 10dpi. (F) Inflammatory score of vehicle 751 (n=3) or EPZ005687-treated (n=3) C3H/HeN mice at 7dpi. *p<0.05, Student’s t-test. (G-I)

26 bioRxiv preprint doi: https://doi.org/10.1101/2020.10.22.350942; this version posted October 22, 2020. The copyright holder for this preprint (which 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 available under aCC-BY-NC-ND 4.0 International license.

752 Immunostaining of Ezh2 (G, purple), H3K27me3 (H, red) and Upk3a and Krt5 (I, green 753 and red, respectively) of the bladder urothelium of mice at 7dpi. 754

27 bioRxiv preprint doi: https://doi.org/10.1101/2020.10.22.350942; this version posted October 22, 2020. The copyright holder for this preprint (which 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 available under aCC-BY-NC-ND 4.0 International license.

1 Supplemental Materials and Methods 2 3 Bladder infections 4 Female mice between 6-8 weeks of age were used in the bladder infections as previously 5 described(Hung et al., 2009). The uropathogenic Escherichia coli UTI89 strain was 6 generously provided by Dr. Scott Hultgren (Washington University, St. Louis, USA). 7 Under isoflurane, 108 colony forming units (CFU, in 50µl PBS) of UTI89 stationary culture 8 were transurethrally inoculated into the bladder. Urine bacterial burdens were determined 9 by serially diluting urine samples and counting CFU. For the repeated infections, infected 10 mice were re-infected (up to 7 times) after 7 days of convalescence. For the C+UTI 11 superinfection model, mice received a single dose of cyclophosphamide (CPP, 300mg/kg, 12 Sigma C7397, i.p.) 1 day prior to UTI89 inoculation. 13 14 Histology, inflammation score and immunofluorescent staining 15 Histology and immunostaining were performed as previously described(Guo et al., 2017). 16 To quantify infiltration of neutrophils, we used four bladders per experimental group. 17 Three paraffin sections from each of the bladders were H&E stained for histological 18 analysis. To count infiltrating neutrophils, 200X microscopic fields of each section were 19 used to count neutrophils and total number of cells. Neutrophils were identified based on 20 their characteristic multilobed nucleus morphology. Percent of neutrophils of total cells 21 was calculated and analyzed using Student’s t-test and p<0.05 was considered significant. 22 Inflammatory score were determined using the described criteria(Hopkins et al., 1998). 23 Two investigators were Antibodies used for immunostaining included Ezh2 (1:100 Cell 24 signaling, 5246), H3K27me3 (1:100 Millipore 07-449), Ki67 (1:100 Abcam 16667), pHH3 25 (1:100 Millipore, 06-570), Krt5 (1:500 Abcam, ab53121; 1:500, Covance, SIG-3475), 26 Upk3a (1:200 provided by Dr. T.T. Sun), Krt20 (1:50, Dako, M7019), IkBa (1:100 Cell 27 signaling, 4814) and p65 (1:100, Santa Cruz, sc8008). Images were captured using a 28 ZEISS fluorescence microscope. 29 30 Chromatin immunoprecipitation assay (ChIP) bioRxiv preprint doi: https://doi.org/10.1101/2020.10.22.350942; this version posted October 22, 2020. The copyright holder for this preprint (which 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 available under aCC-BY-NC-ND 4.0 International license.

31 Chromatin immunoprecipitation was performed as previously described(Guo et al., 2017). 32 Briefly, micro-dissected urothelial cells from 20 bladders per group were crosslinked with 33 1% formaldehyde at room temperature for 15 minutes. Cells were then lysed and 34 sonicated to an average size of 500-1000 base pairs. Immunoprecipitation of chromatin 35 lysates was performed using 1μg per reaction of the following antibodies: anti-H3K27me3 36 (Millipore, 07-449), anti-H3K4me3 (Motif Active, 39159), or control IgG. 37 Immunoprecipitates were collected with Protein G Dynabeads (Invitrogen) and 38 protein/DNA crosslinks were reversed with 5M NaCl. DNA was purified and ChIP DNA 39 was analyzed by quantitative PCR using the following IkBa locus-specific primers, -10kb 40 region: forward, 5’TAA AGC ACA TGT GTG GGG TCT C; reverse, 5’ATC GAA CAT AGG 41 GCC ATG AGA G; -0.3kb region: forward, 5’CAT CGG AGA AAC TCC CTG CG; reverse, 42 5’CAA ACC AAA ATC GCC CGG TG. Each sample was normalized using input DNA. 43 44 RNA isolation, TRAP, RNA-seq and RT-qPCR 45 RNeasy RNA isolation kit (Qiagen) was used to isolate total RNA from micro-dissected 46 bladder urothelium. Briefly, mouse bladder was harvested and placed immediately in cold 47 autoclaved phosphate buffered saline (PBS). Under stereo dissecting microscope, the 48 bladder was bisected to expose the lumen. With one forceps holding and stabilizing the 49 bladder at the muscular layer, the urothelium was gently peeled off using another forceps. 50 This process usually took less than two minutes. Genomic DNA was removed using 51 gDNA Eliminator spin columns (Qiagen). For Translating Ribosome Affinity Purification 52 (TRAP) isolation, mutant (ShhCre/+; Eedf/f; TRAPf/f) and control (ShhCre/+; Eedf/+; TRAPf/f) 53 mouse bladders at embryonic 18.5 (E18.5) were micro-dissected and stored in -800C after 54 ten minutes of incubation with 100µM cycloheximide buffer. Pooled bladder samples 55 (100mg each) were used for TRAP RNA purification as described(Zhou et al., 2013). 56 Illumina 2500 was used for RNA-seq (20M, PE150). Cutoff for differentially expressed 57 genes (DEGs) was based on padj < 0.05. Pathway analysis was done using the Ingenuity 58 Pathway Analysis (IPA) software package (Qiagen). For quantitative real-time PCR (qRT- 59 PCR) experiments, mRNA was reverse-transcribed into cDNA using the SuperScript® III 60 Reverse Transcriptase Kit (Invitrogen). qRT-PCR analyses were performed using SYBR 61 Green (Roche) on an ABI-7500 detector (Applied BioSystems). Relative gene expression bioRxiv preprint doi: https://doi.org/10.1101/2020.10.22.350942; this version posted October 22, 2020. The copyright holder for this preprint (which 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 available under aCC-BY-NC-ND 4.0 International license.

62 levels were normalized to the internal control Gapdh. The following gene-specific primers 63 were used: 64 Gene Sequence 5’ à 3’ Direction Purpose

Ezh1 TCTTCCACGGCACCTATTTC Forward qRT-PCR TTCTGTGCAGGGTTCATGAG Reverse Ezh2 TCCCGTTAAAGACCCTGAATG Forward qRT-PCR TGAAAGTGCCATCCTGATCC Reverse Gapdh TTGTCTCCTGCGACTTCAAC Forward qRT-PCR GTCATACCAGGAAATGAGCTTG Reverse IkBa TCGCTCTTGTTGAAATGTGG Forward qRT-PCR TCATAGGGCAGCTCATCCTC Reverse IkBb CCACCCAAGAGATGCCTCAG Forward qRT-PCR GTGGGGTATGGCCATCATAGT Reverse Cdkn2a CTTGGTCACTGTGAGGATTCA Forward qRT-PCR GATCCTCTCTAGCCTCAACAAC Reverse Cdkn1a CTGGTGATCTGCTGCTCTTT Forward qRT-PCR CCCTGCATATACATTCCCTTCC Reverse En1 TTTCTGAGCCGTGGCTTATC Forward qRT-PCR ACAGTGAAGACACTTGGCTATT Reverse En2 TGGGACATTGGACACTTCTTC Forward qRT-PCR CAGAGCCCACAGACCAAATAG Reverse Pax6 AGTGAATGGGCGGAGTTATG Forward qRT-PCR GAACTGACACTCCAGGTGAAA Reverse Six3 AAGAACAGGCTCCAGCATCA Forward qRT-PCR ATCACATTCCGAGTCGCTGG Reverse IL6 TAGTCCTTCCTACCCCAATTTCC Forward qRT-PCR TTGGTCCTTAGCCACTCCTTC Reverse Socs3 GCGGGCACCTTTCTTATCC Forward qRT-PCR TCCCCGACTGGGTCTTGAC Reverse bioRxiv preprint doi: https://doi.org/10.1101/2020.10.22.350942; this version posted October 22, 2020. The copyright holder for this preprint (which 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 available under aCC-BY-NC-ND 4.0 International license.

Ccl2 GTGCTGACCCCAAGAAGGAA Forward qRT-PCR GTGCTGAAGACCTTAGGGCA Reverse Ccr2 TCTCTGCAAACAGTGCCCAG Forward qRT-PCR CAACCCAACCGAGACCTCTT Reverse Cxcl1 ATGGCTGGGATTCACCTCAA Forward qRT-PCR GTGTGGCTATGACTTCGGTTTG Reverse Cxcr2 GGGTCGTACTGCGTATCCTG Forward qRT-PCR AGACAAGGACGACAGCGAAG Reverse 65 bioRxiv preprint doi: https://doi.org/10.1101/2020.10.22.350942; this version posted October 22, 2020. The copyright holder for this preprint (which 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 available under aCC-BY-NC-ND 4.0 International license.

66 Figure S1. PRC2 controls urothelial progenitor cell proliferation during infection 67 and development. (A) RNA-seq normalized read counts of Ezh1, Ezh2, Eed and Suz12 68 mRNAs in adult urothelium at 1dpi. *p<0.05, Student’s t-test, n=2. (B) Krt5 (red) and Krt20 69 (green) immunostaining of the bladder sections of EedcKO mice after an acute infection at 70 1 day-post-inoculation (dpi). Ctr, littermate heterozygous; PBS, vehicle treatment. Blue, 71 DPAI counter staining; LP, lamina propria; U, urothelium. (C) Immunostaining of Krt5 72 (green) and pHH3 (red) of the bladder sections of EedcKO mice after an acute infection at 73 1dpi. (D) Percentage of pHH3+ proliferating cells in the bladder urothelium relative to total 74 cells. *p<0.05, Student’s t-test, n=3. (E) Schematic diagram of Translating Ribosome 75 Affinity Purification (TRAP) strategy to isolate urothelial cell-specific TRAP RNAs for RNA- 76 seq (Illumina 2500, 20M, pair end 150). mRNAs that were associated with polyribosomes 77 were co-immunoprecipitated by GFP antibody, which recognizes the GFP-L10A fusion 78 protein. The resulting TRAP RNAs were subjected to standard RNA-seq analysis. (F) 79 Volcano plot of the differentially expressed genes (DEGs) in EedcKO urothelium at 80 embryonic day 18.5 (E18.5). n=2, padj<0.05. (G) The top 10 signaling pathways that are 81 dysregulated in EedcKO urothelium based on the Ingenuity Pathway Analysis (IPA) of 82 DEGs shown in F. The p53 molecular pathway is significantly activated. 83 84 Figure S2. Genetic inactivation of PRC2 in the bladder urothelium improves 85 outcomes of the repeated infections. (A) A number of intracellular bacterial 86 communities (IBCs) observed in each bladder with 1xUTI at 3 or 6 hours post-inoculation. 87 (B) Representative histological images of bladder sections from mice that were inoculated 88 7 times (7xUTI) and analyzed at 1 day-post-inoculation (dpi) of the last inoculation. LP, 89 lamina propria; SM, smooth muscle; U, urothelium; dashed lines separate SM, LP and U; 90 arrowheads, polymorphonuclear leukocytes. (C) Krt5 (red) and Upk3a (green) 91 immunostaining of the bladder urothelium of mice with 7xUTI at 1, 3 and 7dpi. Blue, DAPI 92 counter staining. (D) Ki67 immunostaining (red) of adjacent sections shown in C. 93 94 Figure S3. Genetic inactivation of PRC2 in the bladder urothelium improves 95 outcomes of the C+UTI superinfection. (A) Schematic diagram of the C+UTI 96 superinfection regimen. (B) Krt5 (red) and Upk3a (green) immunostaining of the bladder bioRxiv preprint doi: https://doi.org/10.1101/2020.10.22.350942; this version posted October 22, 2020. The copyright holder for this preprint (which 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 available under aCC-BY-NC-ND 4.0 International license.

97 urothelium of mice that were C+UTI superinfected and analyzed at 1 and 7dpi. PBS, 98 phosphate-buffer saline inoculated mice at 7dpi were used as controls. (C and D) 99 Adjacent sections shown in A were stained with Ki67 antibody (red). Percentage of Ki67+ 100 cells were quantitatively analyzed and shown in D. p<0.05, Student’s t-test, n≥3. (E) 101 Volcano plot of the differentially expressed genes (DEGs) between EedcKO and control 102 urothelium at 1dpi of the C+UTI superinfection. n=2, padj<0.05. (F) Top ten canonical 103 pathways identified by Ingenuity Pathway Analysis (IPA) of the DEGs that are affected in 104 EedcKO urothelium at 1dpi of the C+UTI superinfection. 105 106 Figure S4. Ezh2 small molecule inhibitors improve outcomes of the C+UTI 107 superinfection. (A) Bacterial counts in the bladder and kidneys of C57Bl/6 mice after the 108 C+UTI superinfection with or without DZNep treatment. (B) Percentage of neutrophils 109 found in the bladder of C57Bl/6 mice at 10dpi of the C+UTI superinfection. Treatments 110 with DZNep and PBS vehicle control are indicated. p<0.05, Student t-test, n=3. (C) 111 EPZ005678 has no detectable effect on UTI89 survival or proliferation in vivo. n=6. (D) 112 Urine bacterial counts of C57Bl/6 mice after the C+UTI superinfection with or without 113 EPZ005678 treatment. *p<0.05, Mann-Whitney test. (E) Bacterial counts in the bladder 114 and kidneys of C57Bl/6 mice after the C+UTI superinfection with or without EPZ005678 115 treatment. (F) p65 immunostaining (red) the bladder sections that are treated with 116 EPZ005678 or vehicle controls. Blue, DAPI counter staining. Dash lines separates the 117 bladder urothelial and lamina propria layers. (G) Representative histological images of 118 bladder sections of wild type C57Bl/6 mice at 7dpi of the C+UTI superinfections. 119 Treatments with EPZ005687 and vehicle control are indicated. LP, lamina propria; U, 120 urothelium; SM, smooth muscle; dashed lines separate SM, U and LP layers; arrowheads, 121 polymorphonuclear leukocytes. 122 123 Figure S5. Ezh2 small molecule inhibitors improve outcomes of the chronic cystitis. 124 (A and B) Urine bacterial counts of C3H/HeN mice after an acute infection after DZNep 125 (A) or EPZ005687 (B) treatments. *p<0.05, Mann-Whitney test. (C) Bacterial counts in 126 the bladder and kidney. (D) Representative histological images of bladder sections of wild 127 type C3H/HeN mice at 7dpi. Treatments with EPZ005687 and vehicle control are bioRxiv preprint doi: https://doi.org/10.1101/2020.10.22.350942; this version posted October 22, 2020. The copyright holder for this preprint (which 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 available under aCC-BY-NC-ND 4.0 International license.

128 indicated. LP, lamina propria; U, urothelium; SM, smooth muscle; dashed lines separate 129 SM, U and LP layers; arrowheads, polymorphonuclear leukocytes. 130 bioRxiv preprint doi: https://doi.org/10.1101/2020.10.22.350942; this version posted October 22, 2020. The copyright holder for this preprint (which 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 available under aCC-BY-NC-ND 4.0 International license.

Figure 1

A 1x 2x 3x 4x 5x 6x 7x

31 5 7 days post inoculation (dpi) 31 5 7 21 B C 1xUTI 7xUTI PBS 1dpi 7dpi 7dpi 21dpi 1 2 3 4 5 3 Ctr UTI * 2 Ezh2 / DAPI U LP U LP U LP U LP U LP

1 6 7 8 9 10 mRNA fold change fold mRNA

0 Ezh1 Ezh2 U H3K27me3 / DAPI LP U LP U LP U LP U LP

D E F Pttg1ip Ki67/ DAPI up: 12 Ctr 1xUTI, 1dpi down: 15 Rpl26 cKO Eed Krt4 Rps3a1 44 Ctr 80 * * Myh11 LP U 60

(q value) Cbr2 40 10

Cells (%) Cells 22 Kctd12b + +

-log Actg2 Cyp1b1 cKO 20 Acta1

Cyp1a1 Cdkn2a Ki67 Eed Eno1b Actc1 Gm10020 0 2.3 LP U 1xUTIPBS -4 -1 1 4

log2 (fold change) bioRxiv preprint doi: https://doi.org/10.1101/2020.10.22.350942; this version posted October 22, 2020. The copyright holder for this preprint (which 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 available under aCC-BY-NC-ND 4.0 International license.

Figure 2

A Ctr, 1x UTI, 1dpi EedcKO, 1x UTI, 1dpi B 10X 10X Ctr (n=3) 4 cKO 2 Eed (n=7) 6 1 3 5 5 4 SM LP U SM LP U 3 1 2 3 4 5 6 2 *

1 Inflammatory score Inflammatory 0 1dpi

C D Ctr (n=3) EedcKO (n=4) Ctr (n=35) Il-6 8 * Socs3 5 Ccl2 cKO 60 * * Eed (n=15) 6 * 4 * 10 # * 10 40 * 3 9 * 4 * * 10 * 2 # 108 20 # * 2 1 107 change Fold 106 0 0 0 5 80 105 Ccr2 Cxcl1 Cxcr2 4 * 100 * * 104 60 # 3 CFU/ml urin e 10 3 * # 40 2 * 10 2 50 # 101 20 * Fold change change Fold 1 * 100 * # * 1dpi 0 0 0 PBS + - - + - - + - - + - - + - - + - - UTI - 2h 24h - 2h 24h - 2h 24h - 2h 24h - 2h 24h - 2h 24h bioRxiv preprint doi: https://doi.org/10.1101/2020.10.22.350942; this version posted October 22, 2020. The copyright holder for this preprint (which 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 available under aCC-BY-NC-ND 4.0 International license.

Figure 3

B Ctr EedcKO A Ctr (n=3) IκBα/DAPI IκBα/Krt5/DAPI IκBα/DAPI IκBα/Krt5/DAPI EedcKO (n=4) 2.0 1.5 *

1.0

0.5

mRNA fold change 0 IκBα IκBβ LP U LP U LP U LP U D PBS UTI, 1hpi p65/DAPI p65/DAPI p65/Krt5/DAPI C TSS -10kb -0.3kb IκBα

1 2 Ctr 4 6 * 3 4 LP U LP U LP U 2

2 H3K4me3

H3K27me3 1

* cKO 0 0 1 2 1 2 Eed Ctr (n=3) EedcKO (n=3)

LP U LP U LP U bioRxiv preprint doi: https://doi.org/10.1101/2020.10.22.350942; this version posted October 22, 2020. The copyright holder for this preprint (which 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 available under aCC-BY-NC-ND 4.0 International license.

Figure 4

cKO A C Ctr Eed 1x 2x 3x 4x 5x 6x 7x (UTI)

71 71 1dpi 7dpi 7xUTI, 1dpi Day post inoculation (dpi)

B Ctr (n>17) EedcKO (n>5) D Ctr (n=3) * EedcKO (n=4) 108 * 5 4 6 10 3 104 2 * CFU/ml urin e 1 102

Inflammatory score Inflammatory 0 1717171717171 7(dpi) 1dpi 1xUTI 2xUTI 3xUTI 4xUTI 5xUTI 6xUTI 7xUTI 7xUTI

E PBS 7xUTI 1dpi 3dpi 7dpi F Ctr (n=3) 10 EedcKO (n=3) 8 Ctr

6 cells (%) cells + 4 LP LP 2 * DAPI

pHH3 * U ULP LP U U 0 * 60

Krt20 / 1dpi 3dpi 7dpi

40

Krt5 / cells (%) cells cKO + 20 Eed Ki67 0 ** * 1 3 7(dpi) ULP ULP ULP ULP 7xUTI bioRxiv preprint doi: https://doi.org/10.1101/2020.10.22.350942; this version posted October 22, 2020. The copyright holder for this preprint (which 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 available under aCC-BY-NC-ND 4.0 International license.

Figure 5

Ctr (n=3) A cKO C 1010 * * Ctrl (n≥8) Eed (n≥5) EedcKO (n=4) 5 108 4 * 6 10 3 ** 104 2

CFU/ml urin e 1 102 Inflamatory score Inflamatory 0 100 1 3 7 1dpi 7dpi 14dpi 21dpi 28dpi C+UTI, dpi B Ctr EedcKO 1 10x 10x

2 4 6 5 SM LP U 3 SM ULP 1 2 3 4 5 6

D C+UTI PBS 1dpi 7dpi 14dpi E Top Upstream Regulator

-log10 (p value) -log10 (p value) 20 15 10 5 0 0 12 24 36 48 Ctr

Klf2 LPS Fas Il1β DAPI Tgfβ1 LP U ULP U LP U Nr3c1 Dusp1 Tnfα

Krt20 / α-Catenin Infγ Sftpa1 Pgdfα Zfp36 Hif1α

Krt5 / z-score cKO Il6 Hmox1 +6 -6 Gata6 Igf1

Eed U Cat Smad3 N/A U LP LP ULP ULP bioRxiv preprint doi: https://doi.org/10.1101/2020.10.22.350942; this version posted October 22, 2020. The copyright holder for this preprint (which 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 available under aCC-BY-NC-ND 4.0 International license.

Figure 6

A B C+UTI, 7dpi C PBS (n=10) + DZNep (n=12) C UTI superinfection, C57BL/6 mice PBS DZNep 109 CPP UTI 108 * 107 6

/ DAPI 10 5 Day -5 -3 0-1 1 3 5 7 10 104 103 DZNep (1.5mg/kg) or CFU/ml urin e 102 EPZ005687 (2.5mg/kg) 1 H3K27me3 LP 10 LP U U 100 1dpi D PBS DZNep 10x 10x

4 6 5 3 1 2 SM LP U SM LP U 1 2 3 4 5 6

E PBS DZNep F PBS DZNep G PBS DZNep Ezh2 / DAPI Krt20 / Krt5 DAPI LP LP U Upk3a / Krt5 DAPI U LP U LP U LP U LP U

H I C+UTI, 7dpi J C+UTI, Vehicle C+UTI, Vehicle C+UTI, EPZ005687 Vehicle EPZ C+UTI, EPZ005687 5 20 4 * 16 3 LP 12 / Krt5 DAPI

(7dpi) 2 8 1 Ki67+(%) 4 * 0 Upk3a U LP U Inflammatory score score Inflammatory 0 bioRxiv preprint doi: https://doi.org/10.1101/2020.10.22.350942; this version posted October 22, 2020. The copyright holder for this preprint (which 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 available under aCC-BY-NC-ND 4.0 International license.

Figure 7

A B D PBS (n=3) DZNep (n=4) PBS DZNep Chronic cystitis, C3H/HeNCrl mice 5 UTI 10dpi 10dpi 24 4

/ DAPI 18 3 12 Day -5 -3 0-1 1 3 5 10 2 * eutrophil (%) eutrophil 6 * 1 DZNep (1.5mg/kg) or LP N * * EPZ005687 (2.5mg/kg) H3K27me3 LP U 0 0 U U LP SM score Inflammatory

C PBS DZNep E Upk3a/Krt5/DAPI 10x 10x 10dpi 6 1 3 5 U PBS 2 4 LP SM LP U SM LP U 1 2 3 4 5 6 10dpi DZNep

LP U

F G H I Vehicle Vehicle EPZ005687 Vehicle EPZ005687 Vehicle EPZ005687 EPZ005687 5 4 3 2

* Ezh2 / DAPI LP LP LP 1 H3K27me3 / DAPI Upk3a / Krt5 DAPI U U LP U LP U U LP U Inflammatory Score Inflammatory 0 bioRxiv preprint doi: https://doi.org/10.1101/2020.10.22.350942; this version posted October 22, 2020. The copyright holder for this preprint (which 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 available under aCC-BY-NC-ND 4.0 International license.

Figure S1

B 1xUTI C PBS 1dpi 3dpi 7dpi PBS UTI (1dpi)

A LP

120 Ctr Ctr 1xUTI Ctr

90 U DAPI DAPI U * ULP ULP ULP LP ULP 60 Krt5 / Krt20 / UpkII

30 UpkII

Krt5 / Readcounts pHH3 / cKO

0 cKO

Ezh1 Ezh2 Eed Suz12 Eed Eed U U ULP ULP LP LP LP ULP U

E F G up:159 Cdkn2a -log (p value) 0 2 4 6 8 D 80 10 AAAA down:38 GFP p53 signaling 20 * GFP * GFP Cyclins and cell cycle regulation 60 Aryl hydrocarbon receptor signaling 15 AAAA G1/S checkpoint regulation anti-GFP, co-IP

Cell (%) Cell 10 Glioblastoma Multiforme signaling (q value) + 40

10 GADD45 signaling 5 Dmrta2 AAAA -log Transcriptional regulatory network z-score

pHH3 20 Zic1 Hbb-bt Glutathione-mediated detoxification 0 AAAA Ttr Pax6 Six3 +6 -6 UTIPBS En2 LPS/IL-1 mediated inhibition of RXR En1 0 Cdkn1a Stem cell differentiation N/A RNA-seq -6 -3 3 6 log2 (fold change) bioRxiv preprint doi: https://doi.org/10.1101/2020.10.22.350942; this version posted October 22, 2020. The copyright holder for this preprint (which 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 available under aCC-BY-NC-ND 4.0 International license.

Figure S2

B cKO A Ctr, 7xUTI, 1dpi Eed , 7xUTI, 1dpi WT (n=3) b cKO e f 80 Eed (n=3) 10x a c d LP U 60 SM SM LP U a b c d e f 40

20 60x

# IBC per bladder 0 3hrs 6hrs SM LP U SM LP U

C 7xUTI D 7xUTI PBS 1dpi 3dpi 7dpi PBS 1dpi 3dpi 7dpi Ctr Ctr LP LP LP LP LP DAPI

ULP ULP U U DAPI ULP U U U Upk3a / Ki67 / Krt5 / cKO cKO Eed Eed

ULP ULP ULP ULP ULP LP U ULP ULP bioRxiv preprint doi: https://doi.org/10.1101/2020.10.22.350942; this version posted October 22, 2020. The copyright holder for this preprint (which 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 available under aCC-BY-NC-ND 4.0 International license.

Figure S3

B PBS 1dpi 7dpi

A

CPP UTI Ctr DAPI ULP ULP U Day -5 -3 0-1 1 3 5 7 Upk3a / cKO Krt5 / Eed

U ULP LP ULP C C+UTI PBS 1dpi 3dpi 7dpi 14dpi D Ctr (n=3) cKO 100 Eed (n=3) Ctr 75

U LP ULP ULP ULP U *

DAPI 50

(% urothelium) (% +

Ki67 / 25

* Ki67 cKO * *

Eed 0 PBS 1 3 7 14 + U C UTI (dpi) ULP LP ULP ULP ULP

C+UTI (1dpi, EedcKO vs ctr) -log (p value) E F 10 up:62 Krt8 Serpina2n Top 10 Cannonical pathways 0 2.5 5.0 7.5 10.0 12.5 200 down:124 Agranulocyte adhesion and Diapedesis Granulocyte adhesion and Diapedesis Hepatic Fibrosis/Hepatic Stellate Cell activation ILK signaling

Krt18 Acute phase inflammation response signaling

(q value) 100 Reg3g HIF1α signaling 10 Serpine1 TGFβ signaling Myh11 Glucocorticoid receptor signaling -log Cxcl2 Osteoarthritis Pathway Ptgs2 Serpina3m z-score Arg1 Il1β Role of IL-17A in psoriasis Cdkn2a +6 -6 Hbb-bs Cxcl3 Upk3a Renal cell carcinoma signaling Ccl3 P2rx1 2.3 IL-8 signaling N/A -4 -1 1 4

log2 (fold change) bioRxiv preprint doi: https://doi.org/10.1101/2020.10.22.350942; this version posted October 22, 2020. The copyright holder for this preprint (which 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 available under aCC-BY-NC-ND 4.0 International license.

Figure S4

A B + C 7 C UTI, PBS 16 10 PBS (n=8) DZNep (n=10) + 6 C UTI, DZNep 10 12 105 40 CFU 4 10 30 8 103 2 10 20 * 101 CFU (x10^-2) 4 0 10 10 * Bladder Left kidney Right kidney 0 0.54 5.4 54

Neutrophil (%, 7dpi) (%, Neutrophil 0 U LP SM DMSO EPZ(mg/ml)

D Vehicle (n=8) EPZ005687 (n=10) E Vehicle (n=8) EPZ005687 (n=10) 109 * 109 108 108 107 107 106 106 105 105 104 104 103 103 CFU/ml Urin e 2 2 10 CFU per tissute 10 101 101 100 100 1dpi 3dpi 7dpi Bladder Left kidney

F before inoclation 1hpi of inoculation Vehicle 786500ZPE Vehicle 786500ZPE

U U U U U

p65 / DAPI LP LP LP LP

G C+UTI, Vehicle C+UTI, EPZ005687

3 1 4 10x 2 6 5

SM LP U 1 2 3 4 5 6

60x

SM LP U SM LP U bioRxiv preprint doi: https://doi.org/10.1101/2020.10.22.350942; this version posted October 22, 2020. The copyright holder for this preprint (which 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 available under aCC-BY-NC-ND 4.0 International license.

Figure S5

A 1xUTI - PBS (n=9-12) 1xUTI - DZNep (n=12) 109 * 108 107 106 105 104 CFU, urin e 103 102 101 100 1dpi 3dpi 7dpi

B Vehicle (n=20) EPZ005687 (n=7) C Vehicle (n=6) EPZ005687 (n=7)

10 9 * 109 10 8 108 10 7 107 10 6 106 uri ne 5 105 , 10 10 4 104 3 103 CFU 10 2 2 10 CFU per tissute 10 10 1 101 10 0 100 1dpi 3dpi 7dpi Bladder Left kidney

D 1xUTI, 7dpi, Vehicle 1xUTI, 7dpi, EPZ005687

2 3 4 6 10x 1

5 SM LP U SM LP U 1 2 3 4 5 6

60x

SM LP U SM LP U