Ethics statement
All animal experimentation was conducted according to the National Institutes of Health guidelines for the housing and care of laboratory animals. All experiments were performed in accordance with institutional regulations after review and approval by the Animal Studies Committee at Washington University School of Medicine in St Louis, Missouri.
Bacterial strains
The UPEC strains used in this study were the human cystitis isolate UTI89 and derivative thereof: UTI89 attHK022::COM-GFP (UTI89-KanR)35, UTI89 pANT4 and UTI89 hlyA::KD13 (UTI89 ΔhlyA-KanR)22. For both mouse and in vitro infection, UTI89 strains were cultured statically in lysogeny broth (LB) at 37 °C overnight, subcultured 1:1,000 into fresh media and cultured statically at 37 °C for 18 h.
Mouse infections
Female C3H/HeN mice (Envigo) were 7–8 weeks old (‘juvenile’) at the time of the initial infection. A total of 108 c.f.u. of UTI89 were inoculated into the bladder of C3H/HeN mice by transurethral catheterization5,36. C3H/HeN mice develop chronic cystitis in an infection dose-dependent manner and this inoculum results in chronic cystitis in ~50% of mice6. To monitor infection outcomes, urine was collected. Persistent bacteriuria (104 c.f.u. ml−1) is defined as a specific and sensitive cut-off for detecting chronic cystitis5. Chronic cystitis during initial infection was defined as persistent high bacteriuria (>104 c.f.u. ml−1 urine) at every timepoint urine was collected (1, 3, 7, 10, 14, 21 and 28 d post-infection), while resolution of cystitis was defined as urine bacterial titre dropping below this cut-off in at least one timepoint.
At 4 weeks post-infection, all mice were treated with trimethoprim and sulfamethoxazole in the drinking water for 10 d (54 and 270 μg ml−1 water, respectively)6. Urine was collected weekly to confirm clearance of bacteriuria. Four weeks after the initiation of antibiotics, naïve, resolved and sensitized mice were used to isolate primary USCs or used for secondary infection assay. For the secondary infection, mice were challenged with 107 c.f.u. of bacteria inoculated into the bladders, then humanely euthanized at 6 h post-infection, and bacterial burdens were determined to assess acute outcomes.
Cell line culture
Human bladder carcinoma epithelial cells, designated 5637 (ATCC HTB-9) cells, were cultured in RPMI-1640 medium containing 10% FBS at 37 °C in the presence of 5% CO 2 .
Primary USC isolation and culture
Bladder tissue from juvenile, convalescent naïve, resolved and sensitized mice were isolated, bisected and incubated in stripping solution at 4 °C overnight. The urothelial cells were scraped off from the bladder tissue, spun down at 4 °C at 300 g for 5 min, resuspended in fresh collagenase IV solution and incubated with rocking at 37 °C for 20 min. The cells were disaggregated by gentle pipetting, filtered with a 100 μm strainer, then washed with washing media. The cells were cultured in matrigel (BD Biosciences) with 50% L-WRN CM containing 10 mM Y-27632 and 10 mM SB431542 (R&D System)9. Media were changed every 2 d and cells were passaged every 3 d (1:2–3 split). USCs were used for experiments after 10 passages to remove any remaining non-stem urothelial cells.
Differentiated urothelium culture on transwell
USCs were washed in PBS with 0.5 mM EDTA, trypsinized in 0.05% Trypsin and 0.5 mM EDTA for 1 min at 37 °C, dissociated by vigorous pipetting, filtered through a 40 μm cell strainer and resuspended in washing media. Transwells (Corning Costar, 3413) were coated in PBS with 1:40 Matrigel for 30 min at 37 °C. Then 3–4 × 104 USCs were seeded on the transwell insert, and 100 μl and 600 μl 50% CM containing 10 mM Y-27632 were added to the apical and basolateral compartments of the transwell, respectively.
TER measurements
Resistance of the urothelial multilayers was assessed by TER measurement using an epithelial volt-ohm metre (World Precision Instruments). The average value of triplicate measurements was multiplied by the area of the transwell membrane (0.33 cm2) to obtain a final value in ohm × cm2 (ref. 37).
In vitro UPEC infection assay
When urothelium was fully differentiated (TER value >4,000 ohm × cm2), cultures were washed 3 times in warm DMEM/F12 media and infected with UPEC strains at multiplicity of infection 10. Transwells were then incubated at 37 °C for 30 min, changed to media containing 100 μg ml−1 gentamicin to clear the extracellular bacteria and cultured for an extended time. After infection, apical and basolateral media were spun down at 2,000 g at 4 °C for 5 min and used for LDH assay (TaKaRa, MK401). Transwells were washed with sterile PBS, then used for various analyses.
Whole-mount confocal staining
Differentiated urothelia on transwells were washed and fixed in PBS with 4% paraformaldehyde for 15 min and rinsed 3 times with PBS. Subsequently, 100 μl 0.2% Triton X was added for 10 min then dumped and the cells were incubated in 100 μl 2% BSA for blocking for 30 min. The samples were stained with primary antibody, mouse monoclonal anti-keratin 20 (Abcam, ab854, 1:200) and secondary antibody, Alexa Fluor 647 donkey anti-mouse IgG (Invitrogen, A-31571, 1:1,000), then further stained with Alexa Fluor 555 Phalloidin (ThermoFisher, A34055, 1:200) and 4′,6-diamidino-2-phenylindole (DAPI) (ThermoFisher, D1306, 1:1,000). For confocal microscopy,va ZEISS LSM880 laser scanning microscope with Airyscan was used. Fiji ImageJ and macro programme were used to automatically calculate urothelial cell surface area in z-stacked confocal images.
Histopathology and immunofluorescence
USCs or differentiated urothelia were fixed overnight in 10% formaldehyde at 4 °C. After washing in PBS, the fixed samples were pre-embedded into 2% agar, cut vertically, put in transwells side face up, embedded again in paraffin blocks and sectioned. The slides were stained for H&E and immunostained for selected antibodies. For immunofluorescence staining, slides were deparaffinized, hydrated, blocked with 10% heat-inactivated horse serum (HIHS) and 0.3% Triton X-100 in PBS, incubated with primary antibody in 1% HIHS and PBS overnight at 4 °C and secondary antibody in PBS for 30–60 min at room temperature6. The primary antibodies used were mouse monoclonal anti-keratin 20 (Abcam, ab854, 1:200), goat polyclonal anti-E-cadherin (R&D Systems, AF748, 1:500), goat polyclonal anti-uroplakin 3a (Santa Cruz, sc-15186, 1:500), mouse monoclonal anti-uroplakin 3a (Fitzgerald, 10R-U103a, 1:50), rabbit polyclonal anti-p63 (GeneTex, GTX102425, 1:1,000), rabbit monoclonal anti-K5 (Abcam, ab150074, 1:100) and mouse monoclonal anti-keratin 14 (Santa Cruz, sc-53253, 1:50). Alexa Fluor secondary antibodies and DAPI were used at 1:1,000 dilution. Further antibody information is provided in Supplementary Table 3. Samples were visualized on a Zeiss Axio Imager M2 Plus wide-field fluorescence microscope.
Scanning electron microscopy (SEM)
Differentiated urothelia were washed 3 times in PBS, fixed in EM fixative (2% paraformaldehyde, 2.5% glutaraldehyde in 1× PBS) for 1 h on ice and washed 3 times in PBS. Samples were then post-fixed in 1.0% osmium tetroxide, dehydrated in increasing concentrations of ethanol, then dehydrated at 31.1 °C and 1,072 p.s.i. for 16 min in a critical point dryer6. Samples were mounted on carbon tape-coated stubs and sputter-coated with gold/palladium under argon6, then imaged on a Zeiss Crossbeam 540 FIB-SEM.
RNA isolation and RT–qPCR
RNAs were extracted from USCs or differentiated urothelia using RNAeasy Plus mini kit (Qiagen) and reverse-transcribed with iScript Reverse Transcription Supermix (BioRad). We used 1 μl 12.5 ng μl−1 ccomplementary DNA with intron-spanning primers specific to each gene, and iQ SYBR Green Supermix was used according to the manufacturer’s instructions (BioRad). Sequences of the primers we used in this study are listed in Supplementary Table 4. Expression values were normalized to 18S, and relative expression compared to control was determined by the cycle threshold (ΔΔCt) method38. Each sample was run in triplicate, and average Ct values were calculated.
RNA-seq and data analysis
Illumina cDNA libraries were generated using a modified version of the RNAtag-seq protocol39. Briefly, 1 μg of total RNA was fragmented, depleted of genomic DNA, dephosphorylated and ligated to DNA adaptors carrying 5’-AN 8 -3’ barcodes of known sequence with a 5’ phosphate and a 3’ blocking group. Barcoded RNAs were pooled and depleted of ribosomal RNA using the Ribo-Zero rRNA depletion kit (Illumina). cDNA libraries were generated by adding a second adaptor by template switching and PCR amplification with primers carrying Illumina P5 or P7 sequences, then the libraries were sequenced on the Illumina HiSeq 2500. Paired-end sequencing reads in a pool were demultiplexed on the basis of their associated barcode sequence using custom scripts (https://github.com/broadinstitute/split_merge_pl). Reads were then trimmed using cutadapt v1.6 and trimmed reads were aligned to the Mus musculus mm10 genome using tophat2 v2.0.11 and bowtie2 v2.2.2. Gene counts were conducted by HTSeq v0.6.0 and read counts were assigned to annotated transcripts using Salmon v0.8.27.
Read normalization and differential expression were conducted with DESeq2 v1.14.040. rlog transformations of DESeq-normalized reads were used for PCA plots. Fragments per kilobase of transcript per million mapped reads (FPKM) normalization of DEseq2 reads was used for z-score heat maps. TF expression was determined using DESeq2 FPKM-normalized values and a list of mouse TFs (n = 453) from HOCOMOCO v1141, a TF database of validated TF motifs. An adjusted P value cut-off of 0.05 was used and TF candidate expression was visualized using z-score heat maps. Statistically significant differences in gene expression were assessed by the Wald test, followed by multiple test correction using Benjamini-Hochberg false discovery rate (FDR), with adjusted P < 0.05 being considered significant. Pathway analyses were performed with ingenuity pathway analysis (IPA). Significance was determined by a right-tailed Fisher’s exact test, with P adj < 0.05 being considered significantly enriched pathways.
ATAC-seq and data analysis
Single cells (1–2 × 105) of naïve, resolved and sensitized USCs were used for nuclei preparation, and 50,000 nuclei were counted and transferred into 25 μl of 2× TD buffer. Omni-ATAC-seq reaction mix (25 μl) including TDE1 enzyme was added to 25 μl of 50,000 nuclei in 2× TD buffer, then the samples were incubated at 37 °C for 30 min (tapped every 10 min during the incubation in a heat block). Transposed DNA fragments were immediately purified using a MinElute PCR purification kit (Qiagen). ATAC-seq libraries were amplified by PCR amplification (10–12 cycles) with an initial 5 min extension at 72 °C and purified using AMPure XP beads (Beckman Coulter). The purified libraries were eluted with 20 μl of nuclease-free water, quantified using Qubit dsDNA HS assay kit (ThermoFisher), and their size distribution checked with a 4200 TapeStation (High Sensitivity D1000 ScreenTape and Reagents). Paired-end ATAC-seq libraries were sequenced on an Illumina NextSeq 500 (~350 million reads).
ATAC-seq analysis42 used the following tools and versions: Fastqc v0.11.5, Cutadapt v1.11, Samtools v1.5, Bowtie2 v2.3.0, picard v2.10.0, Macs2 v2.1.1.20160309 and bedtools v2.26.0. Sequencing reads were demultiplexed using sample-specific index sequences, quality checked with fastqc, trimmed using cutadapt and aligned to a reference mouse genome (mm10) using bowtie243. Picard was then used to remove secondary alignment, multiply mapped reads and PCR duplicated reads, and peak calling was done with MACS244. Irreproducible discovery rate (IDR) analysis with two replicates was performed following ENCODE’s guidelines45, and ATAC peaks with IDR < 0.05 were chosen as highly reproducible accessible chromatin regions for further analysis. The ATAC-seq signals were visualized on the WashU Epigenome Browser46 as fold change (FC) over background using bedGraph tracks generated using the MACS2 bdgcmp function.
To identify DARs, Diffbind v2.10.0 was used on IDR < 0.05 ATAC peaks, and Benjamini-Hochberg FDR with a cut-off <0.05 was used for statistical significance. Significant DARs (FDR < 0.05) were used for generating volcano plots and heat maps. GREAT16 (basal plus extension parameter) was used for GO pathway analysis. GREAT ranks results by binomial P value using a binomial test. Sensitized (FC > 1.5) and resolved-specific DARs (FC < −1.5) were separately analysed and the top 15 enriched pathways are shown in Fig. 3e–f.
WGBS and data analysis
Single cells (1–2 × 105) of USCs were treated with DNase I to remove trace DNA contamination from the Matrigel. Genomic DNA (gDNA) was prepared from the cells using DNeasy Blood & Tissue kit (Qiagen, 69504). Using 200 ng of gDNA and 0.4 ng lambda, DNA was bisulfite treated using EZ DNA Methylation-Direct kit (Zymo, D5020) and processed with xGen Methyl-Seq Library Prep kit (IDT, 10009824) to generate Illumina-compatible WGBS libraries. The libraries were sequenced on a NovaSeq S4 300XP (~300 million reads) by MGI institute.
WGBS analysis commands with specific parameters are detailed in the Code availability section. Briefly, fastqQC v0.11.8 was used to assess the quality of the raw reads. Subsequently, the paired-end reads were trimmed to remove adaptor sequences and low-quality reads with Cutadapt v1.18 and reassessed using FastqQC. The mouse reference genome mm10 was first bisulfite converted using Bismark v0.20.0. The paired-end reads were aligned to the mm10 bisulfite-converted genome and deduplicated using ‘deduplicate_bismark’. DNA methylation levels were calculated using ‘bismark_methylation_extractor’ and displayed in a methylC format on the WashU Epigenome Browser46. Bisulfite conversion was estimated using the conversion rate of cytosine to thymine in the lambda reference genome.
DMRs were identified with DSS v2.43.247 using a two-group comparison for biological replicates and called using ‘DMLtest’ and ‘callDMR’. A PCA plot of CpG methylation within DMRs was generated using Deeptools v3.3.0. Biological replicates were combined by merging fastq files between replicates and reprocessing using the steps previously described. CpG density was visualized using a 5x coverage cut-off and ggplot2 v3.3.6.
Sensitized-specific DMRs were defined as the overlapping regions between naïve vs sensitized and resolved vs sensitized DMRs. The percent methylation for sensitized-specific DMRs was visualized using the R package ‘ComplexHeatmap’. The DNA methylation over sensitized-specific hypo-DMRs were plotted using Deeptools and visualized using ggplot2. Overlapping regions between DMRs were identified and visualized using Intervene48 ‘Venn’ with default parameters. GREAT16 analysis on sensitized hypo-DMRs was performed as described in ATAC analysis. The sensitized hypo-DMRs were also analysed for genomic annotation using UCSC (https://genome.ucsc.edu/cgi-bin/hgTables) to download GENCODE M25 (https://www.gencodegenes.org/mouse/release_M25.html). The promoter was defined as 1 kb upstream of transcription start site. Genomic annotation priority was assigned in the following order: promoter, coding exon, 5’ UTR, 3’ UTR, intron and intergenic. DMRs were assigned to annotation if the DMR overlapped 20% of the annotation using BEDTools v2.27.1 intersect and were plotted using DNA methylation percent change between sensitized and resolved against the log 2 (FC) of associated genes between sensitized and resolved differentiated urothelia with or without infection.
CUT&RUN and data analysis
Single cells (0.2 × 106) of USCs were slightly crosslinked in 0.1% formaldehyde and fixed cell pellets were stored at −80 °C before use. H3K4Me3, H3K27Ac and H3K27Me3 CUT&RUN was performed using CUT&RUN assay kit (Cell Signaling, 86652), with a few modifications. Briefly, cells were attached to concanavalin A beads for each experiment. Cells were permeabilized with digitonin in the antibody binding buffer containing spermidine and protease inhibitors and then incubated with primary antibodies against H3K4Me3 (Cell Signaling, 9751, 1:50), H3K27Ac (Cell Signaling, 8173, 1:100) or H3K27Me3 (Cell Signaling, 9733, 1:50) at 4 °C overnight on a rotator. The beads–cells mixture was washed 3 times with digitonin buffer, resuspended in 50 µl pAG-MNase and incubated at 4 °C for 1 h on a rotator. Samples were digested in PCR tubes containing 150 µl cold digitonin buffer and 3 µl CaCl 2 at 4 °C for 30 min in a thermal cycler. Then, beads were transferred back to the microcentrifuge tubes, 150 µl of 1× STOP buffer was added and tubes were incubated at 37 °C for 10 min. Placing tubes on a magnetic rack, supernatants were collected to a new tube. Crosslinks were reversed by adding 3 µl 10% SDS solution and 2 µl 20 mg ml−1 proteinase K, then samples were incubated at 65 °C for 2 h. DNA from enriched chromatin samples were purified using DNA spin columns (Zymo, D4013). The sequencing library was prepared with Ultra II DNA Library Prep kit (NEB, E7645) following the manufacturer’s instructions, but reducing the anneal and extension time to 10 s during PCR enrichment of adaptor-ligated DNA.
A detailed list of commands and parameters can be found under Code availability. Briefly, fastqQC v0.11.9 was used to assess the read quality. Subsequently, the paired-end reads were trimmed with Cutadapt v1.9 and reassessed using fastqQC. Reads were then aligned using bowtie2 v2.3.4.149. Mitochondrial reads were removed using samtools v1.9 and deduplicated using Picard v2.8.1 MarkDuplicates. Uniquely mapped reads were extracted using samtools view. Peaks were called using MACS2 v2.1.1.20160309 ‘callpeak’: ‘-q 0.01’ for narrow peaks H3K4Me3 and H3K27Ac, and ‘-q 0.05 --broad’ for H3K27Me3. Encode-defined blacklisted regions were removed. For each histone modification, a consensus peak list was used to calculate the fraction of reads in peaks (FRIP). Reads then were converted to bigWig format using Deeptools and normalized using read coverage and FRIP score. The normalized biological replicates were combined using ucsc-bigwigmerge v377 and converted from bedGraph to bigWigs using kentUCS v334 and mm10 chromosome sizes from UCSC (http://hgdownload.cse.ucsc.edu/goldenPath/mm10/bigZips/mm10.chrom.sizes). To profile sensitized-specific DMR regions for other epigenetic modifications, the sensitized hypo-DMRs were overlapped with MASC2 narrow peaks for ATAC, H3K4Me3 and H3K27Ac, and broad peaks for H3K27Me3. The corresponding peak score was normalized using the FRIP scores and plotted using the R package ‘ComplexHeatmap’. Using the normalized bigWig tracks, ATAC, H3K4Me3, H3K27Ac and H3K27Me3 signals were plotted over the sensitized-specific hypo-DMRs using Deeptools. The normalized bigWig signals were used for the casp1 heat map.
Immunoblotting
Cells were lysed with cell lysis buffer (Cell Signaling, 9803S) according to the manufacturer’s instructions. Rapid Gold BCA Protein Assay kit was used to determine protein concentrations in the cell lysate, and equal amounts of protein were separated by SDS–PAGE and transferred to a nitrocellulose membrane. Membranes were incubated overnight with primary antibodies against caspase-1 (AdipoGen, AG-20B-0042-C100, 1:5,000) and β-actin (MA5-15739, Invitrogen, 1:10,000). HRP-linked secondary antibody (Cell Signaling, 7076S, 1:3,000) and ECL reagent (Amersham, RPN2209) were used to visualize protein bands.
Statistics and reproducibility
For representing images of confocal, immunofluorescence staining and SEM, 3–4 different cell lines (among J1-5, N1-4, R1-4 and S1-4) were stained and imaged. For western blot images, two different cell lines were tested. Statistics for plots/graphs were analysed in GraphPad Prism v8.4.3. Exact P values are indicated when significant (P < 0.05) (GraphPad Prism did not provide exact P value when P < 0.0001).
Reporting summary
Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.