Molecular Basis of Acute Cystitis Reveals Susceptibility Genes and Immunotherapeutic Targets

Tissue damage is usually regarded as a necessary price to pay for successful elimination of pathogens by the innate immune defense. Yet, it is possible to distinguish protective from destructive effects of innate immune activation and selectively attenuate molecular nodes that create pathology. Here, we identify acute cystitis as an Interleukin-1 beta (IL-1β)-driven, hyper-inflammatory condition of the infected urinary bladder and IL-1 receptor blockade as a novel therapeutic strategy. Disease severity was controlled by the mechanism of IL-1β processing and mice with intact inflammasome function developed a moderate, self-limiting form of cystitis. The most severe form of acute cystitis was detected in mice lacking the inflammasome constituents ASC or NLRP-3. IL-1β processing was hyperactive in these mice, due to a new, non-canonical mechanism involving the matrix metalloproteinase 7- (MMP-7). ASC and NLRP-3 served as transcriptional repressors of MMP7 and as a result, Mmp7 was markedly overexpressed in the bladder epithelium of Asc -/- and Nlrp3 -/- mice. The resulting IL-1β hyper-activation loop included a large number of IL-1β-dependent pro-inflammatory genes and the IL-1 receptor antagonist Anakinra inhibited their expression and rescued susceptible Asc -/- mice from bladder pathology. An MMP inhibitor had a similar therapeutic effect. Finally, elevated levels of IL-1β and MMP-7 were detected in patients with acute cystitis, suggesting a potential role as biomarkers and immunotherapeutic targets. The results reproduce important aspects of human acute cystitis in the murine model and provide a comprehensive molecular framework for the pathogenesis and immunotherapy of acute cystitis, one of the most common infections in man. Trial Registration The clinical studies were approved by the Human Ethics Committee at Lund University (approval numbers LU106-02, LU236-99 and Clinical Trial Registration RTP-A2003, International Committee of Medical Journal Editors, www.clinicaltrials.gov).


Trial Registration
The clinical studies were approved by the Human Ethics Committee at Lund University (approval numbers LU106-02, LU236-99 and Clinical Trial Registration RTP-A2003, International Committee of Medical Journal Editors, www.clinicaltrials.gov).

Introduction
Acute cystitis is rapidly becoming a therapeutic enigma, as antibiotic resistance is reducing the options to a minimum [1][2][3][4].Fortunately, new insights are now making it possible to explore immune response modifiers as alternatives to antibiotics.Acute cystitis occurs predominantly in girls and women with normal urinary tracts and at least 60% of all females will report an episode during their lifetime [5][6][7].The recurrence rate is high, especially in a subset of patients, where severe, often recurrent cystitis episodes may cause chronic tissue damage and negatively impact the quality of life [8].In addition, acute cystitis patients pose a highly significant challenge to the health care system.This study addresses if immunotherapy might be a relevant complement to antibiotics, in this patient group.
The urinary bladder mucosa is often exposed to bacteria but does not always retaliate with full force.In patients with acute cystitis, infection triggers a rapid and potent innate immune and inflammatory response in the bladder mucosa and clinical symptoms include pain, urgency and frequency of urination [9][10][11][12].The molecular basis of these symptoms is not well understood, but bacterial interactions with the bladder epithelium have been shown to create inflammatory cascades [13][14][15], which also involve adjacent mucosal cells, such as mast cells and macrophages [16][17][18][19][20].In asymptomatic carriers, the mucosa is exposed to bacteria of lower virulence and the mucosa remains fairly unresponsive, despite the presence of large numbers of bacteria in the lumen [21][22][23][24].Asymptomatic bacteriuria (ABU) strains have evolved a mechanism to avoid elimination by the innate immune defense, through effects on RNA polymerase II and inhibition of host gene expression [22,25].It is therefore challenging to understand, at the molecular level, how a state of exaggerated mucosal inflammation can be generated specifically in acute cystitis patients.The specific molecular interactions that drive the transition from a homeostatic innate immune response to bladder disease remain unclear.
This study examined how innate immune response genes influence the outcome of bladder infection and the pathogenesis of acute cystitis.We identify acute cystitis as an IL-1β-driven, hyper-inflammatory disease [26,27], possibly related to other hyper-inflammatory disorders [28,29].Consistent with such a role, Il1b -/-mice were protected from infection and pathology.In contrast Asc -/-and Nlrp3 -/-mice developed progressive IL-1β-driven bladder inflammation and severe pathology, caused by a new, non-canonical IL-1β processing mechanism, involving the metalloproteinase MMP-7.We also identified the inflammasome constituents ASC (Apoptosis-associated speck-like protein containing a CARD) and NLRP-3 (NACHT, LRR and PYD domains-containing protein 3) as negative regulators of MMP7, explaining why MMP-7 is overexpressed in the mucosa of Asc -/-and Nlrp3 -/-mice and the resulting state of IL-1β hyper-activation.Using IL-1β and MMP-7 as targets for immunotherapy, we succeeded in protecting susceptible Asc -/-mice against acute cystitis, confirming the potential of immunotherapy for this indication.

Acute cystitis strains elicit an IL-1β response in human bladder epithelial cells
To address how infection creates a hyper-inflammatory state in patients with acute cystitis, we first infected the human bladder epithelial cell line HTB-9 in vitro and quantified inflammatory mediators in cell supernatants.We detected an increase in IL-1β secretion, four hours after infection with acute cystitis (CY) strains CY-17, CY-92, CY-132 or the uropathogenic Escherichia coli strain CFT073 (P < 0.001, compared to uninfected cells, two-tailed unpaired t-test).In contrast, the IL-1β response was low in cells infected with the ABU strain E. coli 83972, indicating a virulence-association (Fig 1A).IL-1β secretion was not detected in kidney epithelial cell supernatants after infection with the same strains, suggesting specificity for the bladder epithelium (Fig 1B).Western blot analysis confirmed that mature IL-1β was present in the supernatants of the infected HTB-9 cells (4 hours), as well as unprocessed pro-IL-1β and N-terminal fragment (Fig 1C).A rapid increase in IL-1β staining intensity was observed by confocal microscopy, in cells infected for one hour with 10 5 CFU/ml of CY-17, CY-92 and CFT073 compared to uninfected cells or cells infected with the ABU strain (Fig 1D and 1E, two-tailed unpaired t-test).This increase in cellular IL-1β levels was confirmed by Western blot analysis of whole cell extracts (Fig 1F and 1G).At this time (1 hour), low levels of pro-IL-1β were detected.There was no significant reduction in cell viability after one (! 95% viable) or four hours (! 90% viable), as quantified by PrestoBlue staining and no evidence of pyroptosis after one hour, when the increase in cellular IL-1β levels was detected (S1 Fig) .The presence of unprocessed pro-IL-1β in the 4 hour supernatant might be due to secretion of unprocessed IL-1β via exosomes [30].
To address if IL-1β activation is a characteristic of acute cystitis strains, we infected human bladder epithelial cells with an epidemiologically defined collection of pediatric acute cystitis isolates (n = 67, [31,32]).The majority of these strains (85%) triggered an IL-1β response > 5 pg/ml and 64% of those triggered a high response (40-1,000 pg/ml, Fig 1H).We also examined a collection of pediatric ABU strains (n = 62, [31,33]), which was obtained by screening infants and children in the same geographic area for bacteriuria in the absence of urinary tract infection (UTI) symptoms.In contrast to the CY strains, most of the ABU strains did not trigger a strong IL-1β response (61% < 5 pg/ml), resulting in significantly higher mean IL-1β concentrations in supernatants of bladder cells infected with the CY strains than the ABU strains (121.8 and 32.4 pg/ml respectively, P < 0.001, Fig 1H).To address if the secretion of IL-1β was influenced by bacterial hemolysin [19,20], IL-1β concentrations were examined as a function of hemolytic activity in 40 CY and 38 ABU strains (Fig 1I).There was no significant difference in IL-1β response between hemolysin positive and negative strains (P = 0.07, Mann Whitney unpaired test).In the cystitis subset, a significant difference between hemolysin positive and negative strains was observed, however (P = 0.01, Mann Whitney unpaired test), suggesting that hemolytic cell lysis may contribute to the IL-1β activating virulent phenotype of the acute The results suggest that the majority of acute cystitis strains activate an IL-1β response in human bladder epithelial cells.

Genetic control of acute cystitis in the murine UTI model
As IL-1β is processed by the inflammasome, we subsequently examined if mice with intact or defective inflammasome function develop acute cystitis.We infected mice with genetic defects affecting the NLRP-3 inflammasome: NLRP-3 deficient mice (Nlrp3 -/- [34]) or ASC deficient mice (Asc -/- [35]).In addition, Il1b -/- [36] and Casp1 -/- [37] mice were used and C57BL/6 WT mice were included as controls (Fig 2).For sample sizes and number of experiments, please see each figure legend and an overview in S1 Table .The mice were infected by intravesical inoculation with E. coli strains that triggered high IL-1β responses in human bladder epithelial cells, in vitro (CFT073, CY-17 or CY-92).Infected bladders were evaluated macroscopically, at sacrifice after 7 days and assigned a gross pathology score, defined by size, edema and hyperemia.Tissue pathology was further evaluated by hematoxylin and eosin (H&E) staining and immunohistochemistry of frozen tissue sections, and a histo-pathology score was assigned to each mouse.Histology was scored, independently, by two experienced researchers.The analysis was not blinded.Infection kinetics was followed in urine samples obtained after 6 and 24 hours, 3 and 7 days.
Major, genotype-specific differences in bladder inflammation and pathology were detected after seven days (Fig 2).Two disease end points were distinguished. 1. Severe, progressive cystitis in mice lacking ASC or NLRP-3, resembling chronic human disease.2. A moderate, selflimiting form of acute cystitis in C57BL/6WT mice with intact inflammasome function, resembling sporadic human cystitis.
Bladder pathology was accompanied by high bacterial counts in urine and bladder tissue (Fig 2D -2F).In Asc -/-mice, the neutrophil influx accelerated until day 7, indicating a loss of homeostatic control and progression to chronic inflammation.Infiltrating bacteria and neutrophil aggregates or micro-abscesses were detected in the mucosa of Asc -/-and Nlrp3 -/-mice, with  extensive sloughing of epithelial cells into the lumen.Bacteria were mainly localized along the mucosal surface, with no evidence of bacterial invasion (Fig 2G).This hyper-inflammatory phenotype was also observed after infection of Asc -/-mice with the acute cystitis strains CY-92 and CY-17, which triggered high IL-1β responses in vitro (S3A-S3D Fig) .In contrast, there was no disease phenotype in Asc -/-mice infected with the ABU strain E. coli 83972 or in C57BL/6 WT mice after 24 hours or 7 days (S3E Fig) .There was no evidence of kidney involvement or pathology in mice infected with CY-92 and CY-17, despite positive bacterial cultures from renal tissues.
In contrast, C57BL/6 WT mice infected with CFT073 showed moderate macroscopic evidence of acute cystitis including a small increase in size, edema and hyperemia compared to uninfected controls (Fig 2A and 2C, mean pathology score 1.5).The low level of edema was confirmed by histology, with no evidence of tissue damage (Fig 2B).Infection was accompanied by an increase in urine neutrophil numbers (Fig 2D ) and bacterial numbers reached a peak after 24 hours and then declined (Fig 2E and 2F).By immunohistochemistry, bacterial staining was weak and very few neutrophils were detected in the bladder mucosa (Fig 2G).
Il1b -/-mice showed an even more attenuated phenotype after infection with CFT073, consistent with a key role of IL-1β for bladder inflammation and pathology compared to uninfected bladders (Fig 2A  mean histo-pathology score 0.9, P = 0.003 compared to C57BL/6 WT mice).Il1b -/-mice had fewer infiltrating neutrophils and lower bacterial counts than the C57BL/6 WT mice on day seven (P < 0.001 and P < 0.05), (Fig 2D -2F).Furthermore, Casp1 -/-mice, which have a functional IL-1β deficiency due to defective IL-1β processing and secretion [37], did not develop acute cystitis (S4A-S4D Fig) .The bladders were enlarged and hyperemic, but there was no evidence of inflammatory changes or tissue damage.Casp1 -/-mice showed reduced IL-1β secretion and tissue retention of IL-1β (S4E-S4G Fig) .As a result, IL-1β dependent gene expression was low and Casp1 -/-mice showed a lack of inflammation in bladder tissues.Neutrophils and bacteria were present in urine but did not accumulate in the tissues and the mucosal morphology was intact (S4 Fig).
Infection was accompanied by strong mucosal IL-1β staining in bladder tissue sections in C57BL/6 WT mice, Asc -/-and Nlrp3 -/-mice after 24 hours (Fig 2H).Staining was mainly epithelial and was not seen in Il1b -/-mice or uninfected C57BL/6 WT mice.In parallel with the epithelial staining IL-1β was detected by ELISA in the urine of infected Asc -/-and Nlrp3 -/-mice, with lower levels in C57BL/6 WT mice.By Western blot analysis, bands of approximately 36 and 18 kDa were detected (S2B Fig).
These studies identify genetic determinants of host susceptibility to acute cystitis.Asc and Nlrp3 were defined as key resistance determinants and IL-1β activation as a crucial step in the pathogenesis of acute cystitis.
Bladder tissue structure in H&E-stained tissue sections.Bladders from Asc -/-and Nlrp3 -/-mice showed extensive edema, a loss of tissue structure and epithelial hypertrophy, compared to C57BL/6 WT and Il1b -/-mice or uninfected control mice.The histology score was assessed using H&E stained bladder sections based on neutrophil infiltration, tissue architecture and epithelial thickness on a scale of 0-10, where 10 is highest neutrophil infiltration, least preserved tissue architecture and maximum epithelial thickness).Individual histology

Asc and Nlrp3 control gene expression in infected bladders
To define the mechanism of bladder pathology, we extracted total bladder RNA from infected Asc -/-and Nlrp3 -/-mice with the highest pathology score after seven days and from C57BL/6 WT and Il1b -/-mice, with low pathology scores (Experiments 1 and 2 in S1 Table ) and from uninfected bladders.The RNA was amplified, hybridized onto Mouse Genome array strips, washed, stained and scanned using the GeneAtlas system.Significantly altered genes were identified, by comparing infected-to uninfected mice of the same genetic background (Pvalues < 0.05 and absolute fold change > 1.41) and sorted by relative expression using 2-way ANOVA [38].Heat-maps were constructed by Gitools 2.1.1 software and differentially expressed genes and regulated pathways were analyzed by Ingenuity Pathway Analysis software (see Materials and Methods).
We identified a set of strongly upregulated genes in Asc -/-and Nlrp3 -/-mice with the highest bladder pathology score, but not in C57BL/6 WT mice or Il1b -/-mice (2,228 specifically regulated genes).The heat map in Fig 3A illustrates the similarities in gene expression between 5/7 Asc -/-and 2/5 Nlrp3 -/-mice analyzed by this technology.Those mice also had high and comparable histology scores, defined by evaluation of the H&E-stained bladder tissue sections from the corresponding mice.To further understand the disease process, we identified the most strongly upregulated genes in these mice.Genes with a FC > 100 included metalloproteinase Mmp7, the neutrophil and monocyte chemoattractants Cxcl6 and Cxcl3, the genes encoding calprotectin S100a8 and a9 and the stefin gene Stfa1 (Fig 3B and S2 Table ).By analysis of top-scoring canonical pathways, these genes were shown to control granulocyte and leucocyte diapedesis and signaling, acute phase responses including IL-6 and IL-1β signaling, IL-1R expression and NF-κBsignaling and dendritic cell maturation (S5A Fig) .These genes and pathways were not significantly regulated in C57BL/6 WT or Il1b -/-mice, supporting a disease association.
The results identify IL-1β driven pro-inflammatory genes that are activated, exclusively in Asc -/-and Nlrp3 -/-mice with severe bladder pathology.This response was not detected in the kidneys of infected C57BL/6 WT mice (S5B Fig).

Mechanism of atypical IL-1β processing in infected bladders
The Mmp7 gene, which encodes the matrix metalloproteinase (MMP)-7 [39] was strongly upregulated in Asc -/-and Nlrp3 -/-mice with a high histo-pathology score (Fig 3B).MMP-7 expression was therefore examined as a function of the histo-pathology score (Fig 4A).In Asc -/-and Nlrp3 -/-mice, Mmp7 expression showed a clear association to the overall bladder tissue pathology score and was not regulated in the Il1b -/-or C57BL/6 WT mice (Fig 4A).High MMP-7 protein expression was confirmed, by immunohistochemistry, in bladder tissue sections from Asc -/- and Nlrp3 -/-mice (Fig 4B).Importantly, staining was exclusively epithelial, with shedding of MMP-7 positive cells into the bladder lumen.Epithelial MMP-7 activation was detected as early as 24 hours after infection and importantly, MMP-7 showed no detectable co-localization with neutrophils in the mucosa or sub-mucosa (S6 Fig) To further evaluate the involvement of MMP-7 in acute cystitis, we infected Mmp7 -/-mice [40] with CFT073 and used Asc -/-mice as To address if the cleaved IL-1β fragments were biologically active, reaction mixtures containing pro-IL-1β and MMP-7 were collected after 30 minutes, when the mature product was detected by Western blot (Fig 4D).Human bladder epithelial cells were stimulated with the reaction mixture for one hour, and IL-1β activity was quantified, by measuring the prostaglandin E2 (PGE 2 ) response [41].The 30 minutes reaction mixture activated a dose-dependent PGE 2 response but recombinant MMP-7 and pro-IL-1β (280 and 840 ng/ml) alone had no effect (Fig 4E ).

NLRP-3 and ASC act as negative regulators of MMP7 expression
To understand the mechanism of increased MMP-7 expression in infected Asc -/-and Nlrp3 -/- mice, we examined if ASC and/or NLRP-3 may act as negative regulators of MMP7 expression.After infection of human bladder epithelial cells, with CY-17 and CY-92, we detected a significant increase in MMP-7 staining (confocal microscopy, Fig 5A and 5B).In contrast, ASC staining was reduced after infection with the virulent strains (P < 0.001) and NLRP-3 showed a weaker staining (P < 0.01).The MMP-7 response to infection and the decrease in ASC and NLRP-3 levels were confirmed by Western blot analysis (Fig 5C ).
ASC or NLRP-3 expression was subsequently inhibited by transfection of human bladder epithelial cells with ASC-or NLRP3-specific siRNAs and the effects on MMP- genotype.Means ± SEMs of 5 mice for Asc -/-mice and 2 Nlrp3 -/-mice.(C) Analysis of IL-1β, inflammasome activators and effectors in Asc -/- and Nlrp3 -/-mice, detecting massive over-expression compared to Il1b -/-and WT mice.Red = upregulated, blue = suppressed.The data set included gene expression profiles from 7 Asc -/-and 5 Nlrp3 -/-mice, and two each of the C57BL/6 WT and Il1b -/-controls.Uninfected control RNA of each genotype were used to define significantly regulated genes (! 2 mice per genotype).Histopathology scores and group numbers for individual mice (see also Experiments 1, 2 and 3 in S1 Table ).To address if infection with cystitis strains modifies the interaction of ASC and NLRP-3 in cells, co-immunoprecipitation was performed.ASC was shown to pull down NLRP-3 in nuclear extracts of uninfected cells but after infection, a reduction in ASC/NLRP-3 interaction was detected suggesting that a loss of ASC/NLRP-3 interaction in the nuclear compartment accompanies MMP7 activation (S8C Fig).
To determine if ASC and NLRP-3 interact with the MMP7 promoter, DNA fragments spanning the entire promoter were used as probes in electrophoretic mobility shift assays (EMSA) (S9A Fig) .A DNA fragment of 259 bp, adjacent to the transcription start site (P1, position -18/-276) was shown to interact with a nuclear protein extract from infected bladder cells, resulting in a significant band shift (Fig 5F and 5G).Specificity for ASC and NLRP-3 was confirmed by competition with specific antibodies (Fig 5G).In the absence of nuclear extract, the probe formed a single low molecular weight band, serving as a negative control.To confirm that ASC binds directly to the MMP7 promoter, recombinant ASC protein was incubated with the 259 bp DNA sequence and examined by EMSA.Strong dose-dependent binding of ASC to MMP7 promoter DNA was detected as a band shift, which was competitively inhibited by specific antibodies but not by the IgG isotype control (Fig 5H).Other MMP7 promoter sequences did not interact with ASC or NLRP-3 in this assay (S9A and S9B Fig) .The results suggesting that NLRP-3 and ASC act as negative regulators of MMP7 expression and identify an ASC binding site in MMP7 promoter DNA, adjacent to the transcription start site.

Therapeutic attenuation of the IL-1β response
To address if IL-1β serves as target for immunomodulatory therapy, we selected the most susceptible genotype (Asc -/-mice) for treatment with the IL-1 receptor antagonist (IL-1RA) Anakinra.A dose of 1 mg per mouse in 100 μl of PBS was given intra-peritoneally, 30 minutes before infection and daily after infection with E. coli CFT073 (Fig 6A).This dose was selected based on previous studies in murine models [43].
A dramatic therapeutic effect was observed compared to infected Asc -/-control mice.By macroscopic evaluation, the extent of edema, hyperemia and enlargement was reduced, resulting in a significantly lower pathology score (P < 0.001), (Fig 6B and 6C).By histology, a reduced inflammatory response was seen in the bladders of treated mice and mucosal pathology was inhibited compared to untreated controls that developed extensive bladder pathology (Fig 6D).Mucosal neutrophil infiltration, which accompanies pathology, was prevented and urine neutrophil numbers were low (Fig 6E).As a control for unspecific effects of Anakinra on the bacteria, CFT073 was grown in Luria-Bertani with or without 500 ng/ml of IL-1RA for 10 hours.No difference in bacterial growth rate was detected (S10 Fig).
We subsequently treated susceptible Asc -/-mice with the matrix metalloproteinase inhibitor (MMPI) Batimastat.The MMPI was given 30 minutes before infection and on days 0-2 and 4-6 after infection (0.5 mg in 100 μl of PBS i.p., Fig 6A).The MMPI had a significant protective effect (Fig 6B -6D), detected by macroscopic evaluation, resulting in a reduced pathology score (P = 0.002).By histology neutrophil infiltration was reduced (Fig 6D).As in the IL-1RA-treated compared to Asc -/-mice (n = 5 mice per group, means ± SEMs, ** P < 0.01, *** P < 0.001, two-tailed unpaired t-test).Scale bar = 1 mm.IL-1β levels were elevated in the urine of Asc -/-mice but not in Mmp7 -/-mice, as detected by ELISA.(D) Proteolytic cleavage of pro-IL-1β by MMP-7 in vitro, using purified enzyme and GST-tagged pro-IL-1β.The IL-1β fragments generated by proteolysis were 18 and 16 kDa, defined by Western blot using an antibody specific for the mature form of IL-1β.Recombinant mature IL-1β and GST-tagged pro-IL-1β were used as controls, as well as recombinant MMP-7.One representative experiment out of three, see also S7A As Batimastat is a broad metalloproteinase inhibitor, unspecific effects on other proteases might occur.Proteases inhibited by Batimastat other than MMP-7, were not transcriptionally regulated in any of the mice with acute cystitis or controls.MMP-15, which is not susceptible to Batimastat, was weakly activated in mice with bladder pathology (FC 2.0).These findings suggest that the therapeutic effect of Batimastat reflects inhibition of MMP-7.
The results confirm the importance of IL-1β and MMP-7 for the pathogenesis of acute cystitis and identify these molecules as functional targets for immunomodulatory therapy.Bacterial counts remained elevated in the IL-1RA and the MMPI treated mice in the absence of inflammation, suggesting that the treated Asc -/-mice might develop a condition more like asymptomatic bacteriuria than acute cystitis (Fig 6E).

IL-1β and MMP-7 responses in patients with acute cystitis
To examine the human relevance of the findings in the murine UTI model, we collected urine samples from patients with acute cystitis or ABU and quantified the IL-1β and MMP-7 levels, by ELISA ( Fig 7).Samples from patients with sporadic episodes of acute cystitis were collected at the time of diagnosis, defined by a positive dipstick, dysuria, urgency and frequency of urination but no fever (n = 9).Samples were also obtained from patients with ABU (n = 161), who carried the prototype ABU strain E. coli 83972, following therapeutic inoculation [21].The patients with ABU participated in a prospective study of E. coli 83972-inoculation with detailed monthly collection of symptom scores and urine samples.There were 20 patients with low symptom scores and 161 urine samples were obtained from this group.
We found elevated concentrations of IL-1β in patients with acute cystitis compared to the asymptomatic patient group (Fig 7A -7C) resulting in a means of 264.5 pg/ml and 1.5 pg/ml, respectively (P < 0.001).
In addition, all the patients with acute cystitis had positive MMP-7 levels, above the detection limit of 0.15 ng/ml (Fig 7D).In a subset of 28 ABU urine samples, the mean MMP-7 concentration was low, resulting in mean concentrations of 15.4 ng/ml and 4.3 ng/ml, respectively (P < 0.001).
The results show that patients with acute cystitis have more elevated concentrations of IL-1β and MMP-7 in urine, than patients with ABU, identifying IL-1β and MMP-7 as potential biomarkers of acute cystitis.

Discussion
Symptoms and disease are the price we pay for an efficient host defense against infection.As innate immune effectors are activated to clear tissues of bacteria, they may also cause inflammation, symptoms and tissue damage, especially if innate immune control is compromised.

(B) Difference in gross bladder
This is exemplified here by acute cystitis, which is a common, mostly self-limiting infection except in a subset of patients, who develop severe, recurrent infections, suggesting increased susceptibility.This study proposes a new, genetic basis of susceptibility, exemplified by the disease phenotype in Asc -/-or Nlrp3 -/-mice or resistance in Il1b -/-mice that were protected from infection.The transition of the bladder mucosa from a homeostatic innate immune response to acute disease reflects the molecular control of IL-1β processing, through inflammasomedependent or non-canonical mechanisms (Fig 8), [44][45][46].The findings suggest that acute cystitis might resemble hyper-inflammatory disorders [28,47,48], where therapeutic efficacy of IL-1β inhibitors has been documented [49,50].The results provide a molecular context for acute cystitis and for the susceptibility to acute cystitis in patients with severe and chronic disease.
The severity of acute cystitis was clearly influenced by bacterial virulence as the acute cystitis strains activated IL-1β more efficiently than ABU strains.This comparison was especially valid, as the CY and ABU strains were isolated from the same pediatric population and geographic area, from children who either developed symptoms or were screened for asymptomatic carriage, by collection and culture of urine samples [31,33].The mechanism of IL-1β activation by the acute cystitis strains remains unclear, however.Schaale et al. have studied the IL-1β response of macrophages infected with UPEC strains CFT073 or UTI89 and reported that IL-1β activation and secretion is hemolysin-dependent in murine macrophages but not in human [20].Consistent with their studies, we saw high IL-1β responses to virulent and hemolysin positive CY strains, but we did not detect a direct association with hemolysin production, suggesting that additional features control the induction and secretion of IL-1β.Schaale at al. also pointed to the diversity among different UPEC strains, showing that some are able to boost the inflammasome while others may escape detection by not activating IL-1β.Nagamatsu et al. examined hemolysin and the IL-1β response to UTI89, by inactivating a two-component signal transduction system.The mutant induced significantly higher IL-1β responses than the WT strain, in a hemolysin-dependent manner [19].In addition, pyroptosis was linked to the presence of hemolysin, through activation of Caspase-1 and Caspase-4.In the present study, bacterial determinants of pathology were not identified but the CY isolates are being subjected to whole-genome sequence analysis for this purpose.
Paradoxically, Il1b -/-mice were resistant to infection, unlike the invasive enteropathogens Salmonella and Shigella, which are lethal for Il1b -/-mice [51,52].Internalization of uro-pathogens by cells in the bladder mucosa has been extensively studied and intracellular communities have been highlighted as a niche for bacterial persistence [53][54][55].Specific signaling pathways involved in bacterial uptake by bladder epithelial cells include the ubiquitin-proteasome machinery [56], and especially type 1 fimbriae have been identified as essential ligands [57].As Il1b -/-mice did not develop infection or mucosal inflammation, and proinflammatory genes were not expressed, the IL-1β response may help render the bladder mucosa susceptible to infection, possibly by enhancing bacterial growth [58] or tissue invasion.The ability to activate IL-1β production in host cells may therefore be a key to bacterial virulence, as suggested by the epidemiologic survey of strains used in the study.The findings add MMP-7 to the list of metalloproteinases (MMP-2, MMP-3 and MMP-9) that cleave pro-IL-1β or degrade IL-1β in other cell types [59].MMP-7 has also been shown to process and modulate the activity of anti-bacterial peptides produced by the Paneth cells in the mouse small intestine [60], where cryptdins played a protective role during Salmonella typhimurium- [61] or Chlamydia trachomatis infections [62].In that model, pro-inflammatory effects of MMP-7 were also detected in the intestinal mucosa [63].
ASC and NLRP-3 have recently been identified as transcriptional regulators of innate immune responses.ASC forms a complex with NF-κB and modifies NF-κB-dependent gene expression [64].NLRP-3 is involved in the T H 2 cell differentiation program and facilitates the binding of IRF-4 to DNA [65].Here, we identify ASC and NLRP-3 as negative regulators of MMP7 transcription, based on 1) inverse regulation of MMP-7 with ASC and NLRP-3 in cells infected with CY-17; 2) strongly upregulated MMP7 expression after transfection of human cells with specific siRNAs against ASC and NLRP3; 3) identification of a specific MMP7 promoter DNA sequence, to which nuclear proteins from infected cells bind; 4) inhibition by antibodies to ASC or NLRP-3 of the interaction between nuclear proteins and the MMP7 promoter; 5) binding of recombinant ASC to the MMP7 promoter.As HDAC6 was recently found to interact with NLRP-3 and to modulate its inflammasome function [66], we speculate that NLRP-3 may act as a co-repressor by binding to ASC and recruiting histone deacetylase (HDAC) to the MMP7 promoter, thereby generating a tight chromatin structure refractory to transcription.This was supported by evidence of a direct interaction between ASC and NLRP-3 in the nuclei of CY-17 infected cells.Future studies are required to address in greater detail the regulation of MMP7 expression by ASC and NLRP-3.
Acute cystitis is a handicap, socially and emotionally but despite its prevalence and importance for patients and society, acute cystitis is a poorly understood disease [6,67].Social and behavioral factors have been emphasized as a cause of recurrent infections and until recently, therapeutic options have included a variety of shorter or longer antibiotic regimens, many of which have been discontinued, due to resistance development.It comes as no surprise, that this highly painful condition has been the focus of various interventions in addition to antibiotic therapy.Deliberate establishment of competitive microflora has shown promising clinical effects [21,22] but novel, therapeutic approaches are needed in this large patient group.In this study, we show that acute cystitis is amenable to IL-1 receptor inhibition and/or MMP blockade.As IL-1RA is in clinical use, short-term immunotherapy might be a realistic option as an adjunct to antibiotics in acute cystitis patients.The identified molecular disease determinants may also be helpful to address the unmet need for diagnostic tools in this patient group.The frequency of genetic variants, such as ASC mutations, and their relevance to disease would be an interesting focus of prospective clinical studies.

Bacterial strains and cell infection procedure
Cystitis (CY) and asymptomatic bacteriuria (ABU) strains were prospectively isolated during a study of childhood UTI in Göteborg, Sweden, using standard microbiological techniques [32, (D) MMP-7 concentrations were higher in urine samples from the patients with acute cystitis than in patients with long-term ABU (means ± SEMs, *** P < 0.001, two-tailed unpaired t-test).Urine samples were obtained from patients with sporadic acute cystitis at the time of diagnosis.The patients with ABU participated in a prospective study of therapeutic inoculation with E. coli 83972 and were subjected to long-term follow up [21].Multiple samples were obtained during asymptomatic carriage (3-14 samples per patient).33].The strain collection has been extensively studied and characterized with fimbrial genotype and phenotype, virulence factor expression, OKH antigen profiles and multilocus enzyme typing [31,68].The hemolytic activity was assessed with blood agar plates, where the hemolytic zone surrounding the central stab of bacteria is recorded.The phenotype has been compared to the hly genotype and found to be a very close fit.The UPEC strain, E. coli CFT073 (O6:K2:H1) [69] and the ABU strain E. coli 83972 (OR:K5:H-) [25] have been extensively characterized, including whole genome sequencing and were used as positive or negative controls.Bacteria were cultured on tryptic soy agar (TSA, 16 h, 37°C), harvested in phosphate-buffered saline (PBS, pH 7.2) and diluted as appropriate in RPMI without FCS.In the screen of IL-1β responses to acute cystitis or ABU strains, cells were exposed to 10 8 CFU/ml of bacteria with Gentamicin for 4 hours.In remaining experiments, cells were exposed to 10 5 CFU/ml for 1 hour or 4 hours without antibiotics.Overnight static cultures of E. coli CFT073, CY-17 or CY-92 or 83972 in Luria-Bertani (LB) broth were used for experimental infection.

Global gene expression
Total RNA was extracted from murine bladders or kidneys in RLT buffer with 1% β-Mercaptoethanol after disruption in a tissue homogenizer (TissueLyser LT, Qiagen) using Precellys Lysing kits (Bertin Technologies), with the RNeasy Mini Kit (Qiagen), 100 ng of RNA was amplified using GeneChip 3´IVT Express Kit, 6 μg of fragmented and labeled aRNA was hybridized onto Mouse Genome 430 PM array strips for 16 hours at 45°C, washed, stained and scanned using the Geneatlas system (all Affymetrix).All samples passed the internal quality controls included in the array strips (signal intensity by signal to noise ratio; hybridization and labeling controls; sample quality by GAPDH signal and 3'-5' ratio < 3).
Transcriptomic data was normalized using Robust Multi Average implemented in the Partek Express Software (Partek) [70,71].Fold change was calculated by comparing infected (7 days) to uninfected mice of the same genetic background.Significantly altered genes were sorted by relative expression (2-way ANOVA model using Method of Moments, Pvalues < 0.05 and absolute fold change > 1.41) [38].Heat-maps were constructed by Gitools 2.1.1 software.Differentially expressed genes and regulated pathways were analyzed by Ingenuity Pathway Analysis software (IPA, Ingenuity Systems, Qiagen).Qiagen's list of 84 key inflammasome genes was selected for analysis.
The microarray data are available in the NCBI's Gene Expression Omnibus repository (accession number GSE86096).
Mice were intravesically infected under Isofluorane anesthesia (10 8 CFU in 0.1 ml), through a soft polyethylene catheter (outer diameter 0.61 mm; Clay Adams).Animals were sacrificed under anesthesia; bladders and kidneys were aseptically removed and macroscopic pathology was documented by photography.Tissues were fixed with 4% paraformaldehyde or frozen for sectioning and RNA extraction.Viable counts in homogenized tissues (Stomacher 80, Seward Medical) were determined on TSA (37°C, overnight).Urine samples were collected prior to and at regular times after infection and quantitatively cultured.Neutrophils in uncentrifuged urine were counted, using a hemocytometer.
Gross pathology was scored based on the macroscopic appearance of the bladders at sacrifice.The score was based on edema, hyperemia and size, on a scale of 0-10, where 0 is unchanged compared to the uninfected controls and 10 is most edematous, most hyperemic and largest size.
Histology was scored using H&E stained bladder sections.The score was based on neutrophil infiltration, tissue architecture and epithelial thickness on a scale of 0-10, where 0 is unchanged compared to uninfected controls and 10 the highest neutrophil infiltration, most destroyed tissue architecture and maximum epithelial thickness.

Patients
Urine samples from patients with sporadic acute cystitis were obtained at two primary care clinics in Lund, Sweden.A diagnosis of acute cystitis was based on a urine dipstick analysis positive for bacteria and symptoms from the lower urinary tract, including frequency, dysuria and suprapubic pain.Midstream urine specimens were obtained at the time of diagnosis.
Patients with ABU were included in a placebo-controlled study of asymptomatic bacteriuria, following intravesical inoculation with E. coli 83972 [21].Briefly, E. coli 83972 bacteriuria was established by intravesical inoculation (10 5 CFU/ml in saline), daily for three days and the outcome was measured as the total number of UTIs during an optimal period of 12 months followed by a cross over to a similar period without E. coli 83972 bacteriuria.Urine samples were obtained for cytokine analysis during E. coli 83972 bacteriuria, with negative symptom scores [21].

HTB- 9
cells infected with CY-92, CY-17 and CY-132 (Western blot analysis of cell supernatants, 4 hours, one representative experiment of several repeats).(D) IL-1β staining of HTB-9 cells infected with CY-17, CY-92, CY-132, CFT073 or ABU compared to the background in uninfected cells (PBS).Scale bars = 20 μm (upper panel) and 10 μm (lower panel).One representative experiment is shown.(E) Quantification of the cellular IL-1β response to infection.Increase in total fluorescence intensity (open pin-hole) after subtraction of the background staining in uninfected cells (PBS) (means ± SEMs of 50 cells per sample, ** P < 0.01 and *** P < 0.001 compared to PBS, two-tailed unpaired t-test).One of three experiments is shown.(F, G) IL-1β response to infection (1 hour) quantified by Western blot of whole cell extracts.Quantification of integrated density relative to GAPDH normalized against the background of uninfected cells.One representative experiment of several repeats.(H) IL-1β response to an epidemiologically defined collection of pediatric acute cystitis strains (n = 67) compared to ABU strains (n = 62), obtained from children in the same geographic area.IL-1β was quantified in infected cell supernatants, by ELISA.Pie chart depicting the frequency of bacterial strains activating IL-1β responses: high (orange), intermediate (blue), low (purple) or negative (green).Histogram (inset) of the mean IL-1β response to CY versus ABU strains (means ± SEMs, *** P < 0.001, two-tailed Mann Whitney test).(I) IL-1β activation plotted against hemolytic activity in the collection of CY and ABU strains.No significant association was detected (n = 18-21, Hly+ versus Hly-, twotailed Mann Whitney test).doi:10.1371/journal.ppat.1005848.g001

Fig 3 .
Fig 3. Hyper-activation of IL-1β dependent gene expression and bladder pathology in Asc -/-and Nlrp3 -/-mice.Transcriptomic analysis of whole bladder RNA from infected mice (CFT073, 7 days), compared to uninfected controls of each genotype (cut off FC 1.41, P < 0.05).(A) Heatmap of regulated genes in Asc -/-and Nlrp3 -/-mice with the highest bladder pathology score, defined by neutrophil infiltration, loss of tissue structure and epithelial thickness in H&E stained bladder tissue sections.Scale FC -4 to 4, red = upregulated, blue = downregulated.A distinct gene set distinguished the Asc -/-and Nlrp3 -/-mice with a high histopathology score from C57BL/6 WT mice or Il1b -/-mice without pathology.(B) Top up-regulated genes in the pathology-associated gene set, compared to uninfected controls of each 7 expression were examined by confocal imaging (Fig 5D and S8A Fig).MMP-7 expression increased drastically in transfected and infected cells, where the expression of ASC or NLRP3 had been inhibited, but not in cells transfected with negative control siRNA (Fig 5D).Inhibition efficiency of ASC and NLRP-3 expression by specific siRNAs was confirmed by Western blot analysis.Infection of the cells with CY-17 caused a further decrease in ASC and NLRP-3 staining (Fig 5E, quantified in S8B Fig).Two protein bands where detected, one of 24 kDa, corresponding to the common ASC variant and one of 20 kDa (ASC-b), corresponding to an ASC variant that enhances IL-1β secretion in human promyelocytic leukemia cells (HL60) [42] (Fig 5E).

Fig 6 .
Fig 6.Acute cystitis immunotherapy, using an IL-1 receptor antagonist (IL-1RA) or an MMP inhibitor.(A) Overview of therapeutic regimen used to inhibit bladder pathology.Asc -/-mice were pre-treated with Anakinra (IL-1RA), 30 min before infection and daily after infection with E. coli CFT073 (1 mg in 100 μl of PBS i.p. per mouse) and sacrificed 7 days after infection.Alternatively, Asc -/-mice were pretreated with the matrix metalloproteinase inhibitor (MMPI) Batimastat, 30 min before infection and daily after infection with E. coli CFT073 (0.5 mg in 100 μl of PBS i.p. per mouse, except day 3) (n = 7 per treatment group, total of two experiments).(B) Difference in gross bladder

Fig 7 .
Fig 7. Elevated concentrations of IL-1β and MMP-7 in the urine of patients with acute cystitis.(A) IL-1β concentrations in urine samples from patients with acute cystitis (n = 9).(B) IL-1β concentrations in consecutive urine samples from patients with ABU, who were long-term asymptomatic carriers of E. coli 83972 [21] (means ± SEMs, 20 patients, 161 urine samples).Elevated levels of IL-1β in the cystitis patients compared to the ABU group.Pie chart (inset) depicts the distribution of IL-1β concentrations in each patient group.(C) Histogram compares IL-1β concentrations between the two patient groups (means ± SEMs, *** P < 0.001, two-tailed Mann Whitney test).

Fig 8 .
Fig 8. Models of IL-1β processing-cellular determinants and biological effects.(A) Caspase-1 dependent pro-IL-1β processing by the NLRP-3 inflammasome in mice with intact inflammasome function.The NLRP-3/ASC complex activates Caspase-1, which in turn cleaves pro-IL-1β and the secretion of mature IL-1β by infected cells.The production of MMP-7 is normally low, due to transcriptional repression by ASC and NLRP-3, bound to the MMP-7 promoter.(B) C57BL/6 WT mice develop a mild form of acute cystitis with transient inflammation.Il1b -/-mice were protected against infection and inflammation and Casp1 -/-