Close Association between Altered Urine–Urothelium Barrier and Tertiary Lymphoid Structure Formation in the Renal Pelvis during Nephritis

Significance Statement

Tertiary lymphoid structures (TLSs) form in organs exhibiting chronic inflammation. Immune responses to infections, autoimmune responses, or allergic responses may trigger TLS development. We observed urine–urothelium barrier alterations in the renal pelvis of a mouse model and in human subjects with chronic nephritis, with or without infection. Furthermore, urine leaked from the renal lumen into the parenchyma in nephritic mice. This stimulated the production of cytokines/chemokines in renal stromal cells, resulting in TLS development termed “urinary tract–associated lymphoid structures” (UTALSs). UTALS development correlated with chronic nephritis in humans and mice regardless of UTI, thus possibly revealing a novel pathologic mechanism. Elucidation is crucial for better understanding of noninfectious chronic nephritis.

Visual Abstract

Abstract

Background

Kidneys with chronic inflammation develop tertiary lymphoid structures (TLSs). Infectious pyelonephritis is characterized by renal pelvis (RP) inflammation. However, the pathologic features of TLSs, including their formation and association with non-infectious nephritis, are unclear.

Methods

RPs from humans and mice that were healthy or had non-infectious chronic nephritis were analyzed for TLS development, and the mechanism of TLS formation investigated using urothelium or lymphoid structure cultures.

Results

Regardless of infection, TLSs in the RP, termed urinary tract–associated lymphoid structures (UTALSs), formed in humans and mice with chronic nephritis. Moreover, urine played a unique role in UTALS formation. Specifically, we identified urinary IFN-γ as a candidate factor affecting urothelial barrier integrity because it alters occludin expression. In a nephritis mouse model, urine leaked from the lumen of the RP into the parenchyma. In addition, urine immunologically stimulated UTALS-forming cells via cytokine (IFN-γ, TNF-α) and chemokine (CXCL9, CXCL13) production. CXCL9 and CXCL13 were expressed in UTALS stromal cells and urine stimulation specifically induced CXCL13 in cultured fibroblasts. Characteristically, type XVII collagen (BP180), a candidate autoantigen of bullous pemphigoid, was ectopically localized in the urothelium covering UTALSs and associated with UTALS development by stimulating CXCL9 or IL-22 induction via the TNF-α/FOS/JUN pathway. Notably, UTALS development indices were positively correlated with chronic nephritis development.

Conclusions

TLS formation in the RP is possible and altered urine–urothelium barrier–based UTALS formation may represent a novel mechanism underlying the pathogenesis of chronic nephritis, regardless of urinary tract infection.

Systemic lymphoid tissues (LTs) of mammals are classified into primary LT, secondary LT, and tertiary lymphoid structures (TLSs) on the basis of their localization, function, or manner of development.¹ Bone marrow and thymus constitute primary LTs, whereas spleen, lymph nodes, and mucosa-associated LTs (MALTs) constitute secondary LTs. TLSs are found in systemic organs and their formation is related to chronic inflammation associated with cancer, and autoimmune disease (AID) or allergic disease.^2,3 Recent studies have emphasized the importance of TLSs found in the kidney parenchyma of elderly individuals and patients with nephritis.^4,5 Moreover, TLS development may lead to CKD,⁵ whose global prevalence has increased and evolved into a serious public health challenge often associated with ESKD.⁶ In particular, stromal immunofibroblasts,⁷ which produce chemokines (CCL20 and CXCL13), may play a principal role in renal TLS formation by attracting immune cells in situ.^5,8,9

Kidneys are linked to the outer environment via the urinary tract (UT). The renal pelvis (RP), a border structure that exists between the kidney and the UT, receives urine. Because a portion of the RP penetrates deep into the kidney, the RP is intricately connected to the renal parenchyma via connective tissues. Primarily, UT infections that disrupt the barrier function of the urothelium lining the RP cause inflammatory lesions in the RP and renal parenchyma, leading to pyelonephritis.¹⁰ The mortality rate associated with severe pyelonephritis exceeds 20%.¹¹ Renal parenchymal inflammation caused by noninfectious inflammatory conditions such as AID may cause glomerular and tubulointerstitial inflammation.¹² Although MALTs do not form in the RP under noninfectious conditions, the possibility of their formation nevertheless exists because the urothelium and RP are intricately connected, similar to the intricate connection between the epithelium and parenchyma in organs with MALT. Nevertheless, MALT lymphoma can develop in the human RP.¹³ Age-related TLS formation may be observed in the urinary bladder even under noninfectious conditions.^14(preprint) These clinical findings suggest that lymphoid structure (LS) formation in the UT, including the RP, is possible.

Here we aimed to examine LS formation in the RP under normal and disease conditions in both humans and mice, and to investigate other factors that may be involved in LS development to demonstrate the pathologic association between noninfectious chronic nephritis and LS formation in the RP.

Methods

Human Sample Analysis

We obtained 10% neutral buffered formalin (NBF)–fixed, paraffin-embedded, normal kidney tissues (two White women, 33 and 34 years), and kidney samples of infectious pyelonephritis (White man, 39 years) and noninfectious chronic nephritis (two White men, 69 and 52 years and an Asian man, 53 years) from KAC Inc. (Kyoto, Japan). Urine was collected from healthy human volunteers from Hokkaido University (n=10, average age =27 years) or patients with AID-related nephritis at Kochi University (n=12, average 47 years). Written, informed consent from the donors, next of kin, or persons exercising parental authority for minors was obtained. The study of human samples was approved by the Ethics Committee of Kochi University (approval no. 24–134). Analyses of human samples were approved by the Ethics Committee of the Faculty of Veterinary Medicine and Research Center for Zoonosis Control, Hokkaido University (approval no. 29–5). The collection of samples from human subjects was performed in accordance with the Declaration of Helsinki and the Declaration of Istanbul.

Paper urinalysis was performed using Multistix PRO11 (Siemens Healthcare, Tokyo, Japan). Paper urinalysis did not indicate remarkable differences in pH (6.5 versus 7.0) or gravity (1.02 versus 1.02) of the pooled urine between healthy controls and patients with nephritis. The urine samples were analyzed using a Human ProQuantum Immunoassay Kit for IFN-γ and TNF-α (#A35576, #A35601; Thermo Fisher Scientific, Waltham, MA).

Animals and Sample Preparation

C57BL/6N of both sexes, female MRL/MpJJmsSlc (MRL/MpJ) and MRL/MpJJmsSlc-lpr/lpr (MRL/lpr), and male BXSB/MpJ and BXSB/MpJ-Yaa, were purchased from Japan SLC Inc. (Shizuoka, Japan) and used at the ages of 2, 3, 9, and 12 months (C57BL/6N); 3 months (MRL/lpr); and 6 months (all strains). Female B6.MRLc1 were maintained in our laboratory¹⁵ and used at the age of 9 months. Animal experimentation procedures were approved by the Institutional Animal Care and Use Committee of the Faculty of Veterinary Medicine, Hokkaido University (approval no. 16–0124). Investigators adhered to the Guidelines for the Care and Use of Laboratory Animals of Hokkaido University, Faculty of Veterinary Medicine. All animal experimental protocols were approved by the Association for Assessment and Accreditation of Laboratory Animal Care International. All mice used in this study were euthanized by cutting the femoral artery under deep anesthesia (0.3-mg/kg medetomidine, 4.0-mg/kg midazolam, and 5.0-mg/kg butorphanol). For 2-month-old C57BL/6N, unilateral ureter obstruction (UUO) for 1 day or its equivalent sham operation was performed under deep anesthesia. The number of animals used for each experiment is described in each figure legend. After euthanasia, spleen–to–body weight ratio (SPW/BW) was calculated, and serum double-stranded DNA autoantibody levels were measured using an ELISA kit (#82639; Immuno-Biologic Laboratories, Gunma, Japan). Kidneys were collected, cut into four transverse quarter sections, and fixed in either 4% paraformaldehyde (PFA)/0.1 M phosphate buffer (PB) for histologic analysis or 2.5% glutaraldehyde (GTA) for scanning electron microscopy (SEM) analysis. A part of the kidney was stored in RNAlater (Thermo Fisher Scientific) for gene expression analysis.

Histopathology

Histologic sections of human kidney were stained using Masson’s trichrome (MT) or periodic acid–Schiff (PAS), whereas mouse kidneys were stained using hematoxylin and eosin (H&E), MT, PAS, or silver impregnation. In both human and mouse kidneys, immunohistochemistry (IHC) and immunofluorescence (IF) were performed as previously described.¹⁶ Information regarding antigen retrieval and antibodies used is summarized in Supplemental Table 1. Stained sections were examined using a BZ-X710 microscope (Keyence, Osaka, Japan) and converted to virtual slides using NanoZoomer 2.0-RS (Hamamatsu Photonics, Shizuoka, Japan). For mice, in situ hybridization (ISH) was performed using RNAscope (R) 2.5 HD Reagent Kit–BROWN (#322300; Advanced Cell Diagnostics, Newark, CA) and HybEZ II Hybridization System with EZ-Batch Slide System (Advanced Cell Diagnostics) as previously described.¹⁷ The probes used in this study are listed in Supplemental Table 2.

For localization analysis of LSs formed in the RP, semiserial sections were obtained from the cranial to caudal poles of a 6-month-old male C57BL/6N kidney at 70-μm intervals and stained using H&E. These sections were converted into virtual slides using NanoZoomer 2.0-RS and observed using the NDP.view2 program (Hamamatsu Photonics). LSs in the abdominal, central, and dorsal RPs in each section were counted (Supplemental Figures 1 and 2A).

Histoplanimetry

The glomerular area and the number of CD3⁺ or B220⁺ cells in the glomerular sections obtained from a 6-month-old mouse kidney were quantified via IHC using the BZ-H3C analysis application. Digital images were obtained using the BZ-X710 microscope. For glomerular lesions (GLs), the number of positive cells was divided by the glomerular area to obtain the density of positive cells in the glomerulus. For TLS formation, digital images of kidney sections stained for CD3 or B220 in IHC were obtained using NDP.view2 at 0.36-fold magnification. The digital images were converted to binary images, and the area of clusters positive for CD3 or B220 in the areas of kidney sections without LSs formed in the RP was measured using ImageJ (National Institutes of Health, Bethesda, MD). For tubulointerstitial lesions (TILs), images of tubulointerstitial regions that did not contain glomeruli were obtained with NDP.view2, converted to binary images, and then integrated density (IntDen) of CD3⁺ or B220⁺ reactions in the unit area was quantified via ImageJ. The number of CD3⁺ or B220⁺ cells in the examined RP area was quantified in 9-month-old C57BL/6N and B6.MRLc1.

For collagen type XVII α 1 chain (COL17A1) IHC, images of LSs formed in the mouse RPs were randomly obtained by NDP.view2, converted to binary images, and then IntDen of COL17A1⁺ reactions in urothelium was quantified using ImageJ. These values were divided by the IntDen of nuclei stained by hematoxylin in the examined area of LSs formed in the RP or the examined LS area.

SEM

GTA-fixed mouse kidneys were postfixed with 1% OsO₄ at 4°C for 1 hour, washed with 0.1 M PB, and treated with 1% tannic acid at 4°C for 1.5 hours. After washing with 0.1 M PB, the specimens were dehydrated via an ascending alcohol gradient, transferred into 3-methylbutyl acetate, and finally dried (HCP-2 critical point dryer; Hitachi, Tokyo, Japan). The dried samples were fixed on aluminum stubs with double-faced adhesive tabs. Surface treatment with a platinum layer was performed with ion sputtering (E-1030; Hitachi), and the samples were observed via SEM (SU8000; Hitachi).

Mimic LS Culture and Analysis

The spleen of a 6-month-old mouse was collected and minced in a culture dish. Next, 10 ml of 1× PBS (#164–25511; FUJIFILM Wako, Osaka, Japan) was added, after which the sample was filtered using 100-μm cell strainers (#352360; Corning Inc., Corning, NY) and centrifuged at 1000 rpm for 5 minutes. After supernatant aspiration, cells were resuspended, incubated with 10 ml of 0.017 M Tris buffer containing 0.75% ammonium chloride (pH 7.65) for 5 minutes, and washed with 10 ml of PBS. This solution was filtered using 40-μm cell strainers (#352340; Corning Inc.) and centrifuged at 1000 rpm for 5 minutes. After washing twice with PBS, the collected cells were resuspended in RPMI1640 with l-glutamine (#189–02025; FUJIFILM Wako) without antibiotics and harvested in 24-well culture plates (#92424; Techno Plastic Products, Trasadingen, Switzerland).

The cells were defined as mimic LSs (mLSs) containing immune and stromal cells. Furthermore, some mouse spleen cells were similarly collected without ammonium chloride treatment, and then T, B, and CD11b⁺ cells were separately isolated using magnetic-activated cell sorting–based commercial kits (EasySep, #ST-19851RF, #ST-19854RF, #ST-18970RF; VERITAS, Santa Clara, CA). mLSs or separated cells were stimulated using recombinant human COL17A1 (rhCOL17A1; #MBS1265558; MyBioSource, San Diego, CA) or BSA (#013–27054; FUJIFILM Wako). Endotoxin levels in rhCOL17A1 were <1.0 endotoxin unit per 1.0 μg of protein as determined using a limulus amebocyte lysate test (MyBioSource). Urine was collected from five MRL/MpJ or MRL/lpr at 6 months of age via compressive urination, after which samples from each strain were pooled and centrifuged at 1500 rpm for 5 minutes. Paper urinalysis was performed using Multistix PRO11. The urine supernatant was filtered using a 0.22-μm syringe filter (#SLPES2522S; Hawach Scientific, Shaanxi, China) and used for mLS stimulation. mLS or isolated cell experiments were performed using 2.5×10⁷ cells per ml and 500 μl of medium per well or 2.5×10⁷ cells per ml and 500 μl of medium per well, respectively.

After rhCOL17A1 or BSA treatment, mLS cell viability was analyzed via the CellTiter 96 AQueous One Solution Assay (#G3582; Promega, Madison, WI). After stimulation with rhCOL17A1 or urine, the culture supernatant was collected, centrifuged at 2000 rpm for 5 minutes, and analyzed using mouse ELISA kits for IL-6 (#27768; Immuno-Biologic Laboratories), IFN-γ, or TNF-α (#630–44701 or #634–44721; FUJIFILM Wako), and CXCL9 or CXCL13 (#MCX900 or #MCX130; R&D Systems, Minneapolis, MN). Cells were then collected for quantitative PCR (qPCR). For IF, the stimulated mLSs were embedded in iPGell (#PG20–1; GenoStaff, Tokyo, Japan), fixed with 4% PFA/0.1 M PB, and paraffin embedded. Histologic sections were stained according to appropriate antigen retrieval conditions and antibody dilutions (Supplemental Table 1).

Cell Culture and Analysis

NIH3T3 cells (RIKEN Bioresource Center, Tsukuba, Japan) were maintained in DMEM (#041–29775; FUJIFILM Wako) supplemented with 10% FBS (#10099141; Thermo Fisher Scientific) and 1× penicillin/streptomycin (Thermo Fisher Scientific) until reaching 70% confluency. NIH3T3 cells were resuspended in DMEM without FBS and antibiotics and cultured in 24-well culture plates until 90% confluency. After 24 hours, the cells were stimulated using rhCOL17A1, BSA, or 25% MRL/MpJ or MRL/lpr urine diluted with DMEM for 3 hours as described for mLS culture. The culture supernatant was collected, centrifuged at 2000 rpm for 5 minutes, and analyzed using mouse ELISA kits for CXCL13 (#MCX130; R&D Systems). Cells were collected by scraping for qPCR. For IF, stimulated NIH3T3 cells were analyzed as described for mLSs.

Normal human urothelial cells (HUCs; #KP-4109; Kurabo, Osaka, Japan) were maintained in a UroLife Comp kit (#LUC-LL0071; Kurabo) until 100% confluency. HUCs were cultured in six-well culture plates (#92406; Techno Plastic Products) or on cover glasses placed within microplate wells for IF. The medium was changed to basal medium (UroLife BM; #LUC-LL0054; Kurabo) without growth factors. After 24 hours, HUCs were stimulated using 25% human urine diluted with basal medium for 3 hours or 1.0 ng of human recombinant IFN-γ or TNF-α (#285-IF or #210-TA; R&D Systems) for 24 hours. Cells were collected by scraping for qPCR. HUCs were fixed using methanol/acetone at 4°C for 30 minutes, and then IF was performed as previously described.¹⁸ Information regarding antibodies used is summarized in Supplemental Table 1.

RNA Analysis

Using a stereo microscope, the RP was isolated by microdissection from the stored mouse kidney section in RNAlater (Supplemental Figure 2B). Then, total RNA was purified from the mouse kidney section, separated mouse RP, and cultured mouse or human cells using TRIzol reagent (#15596018; Thermo Fisher Scientific). Purified total RNA was used as a template to synthesize cDNA using ReverTra Ace qPCR RT Master Mix with gDNA Remover (#FSQ-301; Toyobo, Osaka, Japan). Products amplified by PCR using GoTaq Green Master Mix (#M7121; Promega, Madison, WI) and gene-specific primers (Supplemental Table 2) were visualized with 2% agarose gel electrophoresis.

qPCR analysis was performed using the THUNDERBIRD SYBR qPCR Mix (#QPS-101; Toyobo), gene-specific primers (Supplemental Table 2), and a real-time thermal cycler (CFX Connect; BIO-RAD, Santa Rosa, CA). Cycle conditions are listed in Supplemental Table 2. Data were normalized to the values of actin β (Actb) or uroplakin 3 (Upk3) in mice, glyceraldehyde-3-phosphate dehydrogenase (GAPDH) in human cells, and to those of control expression using the δ-δ Ct method.

Microarray

Total RNA was purified from mouse tissues stored in RNAlater or cultured mLS cells using the RNeasy Mini Kit (#74104; Qiagen, Hilden, Germany). For kidneys, RNA from MRL/MpJ (n=7) and MRL/lpr (n=10) mice was pooled into one sample for each group. The RNA of mLS cells was also prepared from the PBS- or rhCOL17A1-treated group (n=3). RNA integrity was validated using an Agilent 2100 Bioanalyzer II (Agilent Technologies, Santa Clara, CA); complementary RNA was synthesized using a Low Input Quick Amp Labeling Kit (Agilent Technologies), whereas gene expression was analyzed using an Agilent Technologies Microarray Scanner and SurePrint G3 Mouse 8×60K ver.2.0 (Agilent Technologies). Obtained raw data were normalized using 75Percentile Shift (GeneSpring; Agilent Technologies). Minimum information about a microarray experiment–compliant dataset was deposited in the NCBI Gene Expression Omnibus and is accessible under the GEO Series accession numbers GSE151929 and GSE151930. STRINGs (https://string-db.org/) was used for gene ontology (GO) analysis. RP samples highly expressed urothelium markers, indicating their successful isolation from mouse kidneys (Supplemental Table 3).

Urothelium Integrity Analysis

Evans blue (5 mg/ml; #E2129; Merck, Kenilworth, NJ) was mixed with BSA (40 mg/ml), incubated for 30 minutes, and filtered using a 0.22-μm syringe filter to produce Evans blue–conjugated BSA (EBA) solution. The urethra of female mice was ligated under combination anesthesia (0.3 mg/kg medetomidine, 4.0 mg/kg midazolam, and 5.0 mg/kg butorphanol), and a 20 G feeding needle was inserted from the apex of the urinary bladder and fixed via ligation. After clamping one ureter with vascular clips (KN-353-AS-1–40 g; Natsume Seisakusho, Tokyo, Japan), 50 μl of EBA was injected into the urinary bladder through an inserted feeding needle. EBA injection to the RP was confirmed via color change in the other ureter (Supplemental Figure 3). After 5 minutes of contact time, the mice were euthanized, and 4% PFA/0.1 M PB was perfused from the left ventricle to the systemic circulation. Next, the kidneys were immediately collected and embedded in OCT compound (#4583; Sakura Finetek Japan, Tokyo, Japan) using liquid nitrogen. Frozen sections (4 µm) were prepared, fixed with 4% PFA/0.1 M PB for 5 minutes, washed with PBS and then distilled water, and finally air dried. The EBA fluorescence signal was observed using the BZ-X710 microscope and Cy5 filter. Obtained images were converted to binary images, and the relative area of EBA in RP to urothelium length along with the examined area was quantified using ImageJ.

Statistical Analysis

For two-group comparisons, Students t test or Mann–Whitney U test was used for microarray analysis and other experiments, respectively. For multiple comparisons, Dunnett’s test was performed when statistical significance was observed with the Kruskal–Wallis test. Spearman's correlation tests were used to analyze the correlation between two parameters. P values <0.05 were considered statistically significant.

Results

LSs Form in the Human RP

Patients diagnosed with either infectious pyelonephritis or noninfectious chronic nephritis showed severe TILs and GLs (Figure 1A). MT staining showed well-developed collagen fibers beneath the RP urothelium in the normal kidney (Figure 1B). In both patients, these fibers were reticularly distributed in the RP, whereas numerous infiltrating CD3⁺ T cells (Figure 1C), CD20⁺ B cells, and IBA1⁺ macrophages (Supplemental Figure 4) were found between fibers and in the urothelium. Type III collagen (COL3A1)⁺ reticular fibers were localized in all examined RPs (Figure 1D). Furthermore, several PNAd⁺ high endothelial venules (HEVs), entrances for immune cells, were observed in the RP of both patients (Figure 1E).

Figure 1.

RP histopathology in human patients with infectious or noninfectious nephritis. (A and B) Histopathology of the (A) human kidney and (B) RP. Patients with pyelonephritis and chronic nephritis show (A) severe GLs and TILs and (B) cell infiltrations in the RP regardless of infection. Insets indicate the magnified square area. Lu, lumen of RP; MT staining. (C–E) Localization of (C) CD3⁺ T cells, (D) COL3A1⁺ fibers, and (E) PNAd⁺ HEVs in the RP of normal kidney donors and patients with nephritis. (C) Numerous CD3⁺ cells are localized in the lamina propria and urothelium in both patients. (D) Strong COL3A1⁺ reactions are observed in all RP samples. Arrows show PNAd⁺ HEVs, and insets indicate the magnified square area (D, E); IHC. Normal: female, 33 years. Pyelonephritis: male, 39 years. Chronic nephritis: male, 69 years. (F and G) Histopathology of the (F) human kidney and (G) RP in different samples of (A–E). Necropsy samples of human kidneys showing normal histopathology and moderate or severe tubulointerstitial cell infiltration in noninfectious chronic nephritis. Normal kidney exhibited scarce cell infiltrations in both the tubulointerstitium and RP. In the kidney showing moderate cell infiltration, mononuclear cells are scattered in the RP parenchyma beneath the urothelium. In the kidney showing severe lesions, severe cell infiltrations are also observed in RP; MT staining. Normal: female, 34 years. Moderate: male, 52 years. Severe: male, 53 years. Scale bars, 50 μm (A and F), 100 μm (B–E and G).

In the absence of infection, necropsy samples showing normal histology, moderate chronic nephritis, or severe chronic nephritis also revealed scarce, moderate, or abundant cell infiltration of RPs, respectively (Figure 1, F and G). These structures were termed UT-associated LSs (UTALSs) and the mechanism underlying their formation was investigated in subsequent experiments.

UTALSs Form in the RP of Healthy and Nephritis Mice

Histologic analysis using semiserial sections of 6-month-old C57BL/6N kidneys demonstrated UTALS localization in the deep portion of RP (Figure 2A, Supplemental Figures 1 and 2A). Gene expression analysis revealed that T and B cells, macrophages, and antigen-presenting cells (APCs) were abundantly localized in the RP (Figure 2B).

Figure 2.

Localization and cell composition of UTALSs in the mouse kidney. C57BL/6N (6 months) were used in (B–F). C57BL/6N and B6.MRLc1 (9 months) were used in (G and H). (A) UTALS localization considering RP morphology. Dotted straight lines indicate distance from the reference section (0 μm) at approximately 500-μm intervals. Red circles, UTALS. (B) Immune cell marker gene expression in the kidney and RP. aT, active T cells; aB, active B cells; Mφ, macrophages; panT, pan T cells; qPCR. (C) Representative histologic features of UTALS. Numerous cells infiltrate between connective fibers in RP; MT staining. (D) Large UTALS comprising CD3⁺ T cells, B220⁺ B cells, and IBA1⁺ macrophages. Hoechst, nuclear staining; IF. (E) Large UTALS comprising COL3A1⁺ reticular fibers. Dotted line indicates the border between Lu and urothelium; IHC. (F) Large UTALS comprising PNAd⁺ HEVs. Insets indicate the magnified square area (right panel); IHC. (G) UTALS in a spontaneous mouse model for chronic GN. Compared with C57BL/6N, B6.MRLc1 develops UTALSs with CD3⁺ T cells or B220⁺ B cells in the RP and manifests GN in CO. B220⁺ B cells are also observed in the urothelium (square and inset); PAS staining or IHC. (H) CD3⁺ T cell or B220⁺ B cell number in the examined RP area; histoplanimetry. Values are presented as the mean±SEM; n=7 (B); n=7 or 22 (C57BL/6N or B6.MRLc1) (H). Significant differences between kidney and RP or C57BL/6N and B6.MRLc1 are indicated by **P<0.01; Mann–Whitney U test. Scale bars, 50 μm (C and D), 10 μm (E), 100 μm (F and G). CO, cortex; IM, inner medulla; Lu, lumen of RP; OM, outer medulla; UE, urothelium.

Small UTALSs were composed of condensed collagen fibers, several CD3⁺ T cells, B220⁺ B cells, IBA1⁺ macrophages, Ly-6G⁺ granulocytes, and LYVE1⁺ lymphatic vessels (LVs) (Supplemental Figure 5). Large UTALSs comprised abundant immune cells—consisting of T cells, B cells, and macrophages similar to human UTALSs—occupying spaces between developed connective tissue fibers (Figure 2, C and D, Supplemental Figure 6A). Well-developed COL3A1⁺ fibers and PNAd⁺ HEVs were detected in the UTALSs (Figure 2, E and F). Additionally, large UTALSs consisted of several LVs and FOXP3⁺ regulatory T cells (Supplemental Figure 6B). Twelve-month-old C57BL/6N developed large UTALSs (Supplemental Figure 6, C–F), suggesting the effect of age-related factors on their development.

The other GN model, BXSB/MpJ-Yaa (Supplemental Figure 7), and B6.MRLc1, a mouse model of C57BL/6N-background spontaneous chronic GN,¹⁵ showed increased lymphocyte numbers in the RP (Figure 2, G and H).

UTALSs Develop in the RP of Severe Nephritis Mice

Next, we examined MRL/lpr, a representative severe immune-mediated nephritis model. Compared with control MRL/MpJ, MRL/lpr exhibited significantly higher histopathologic indices of GL, TIL, and TLS formation in the kidney (Supplemental Figure 8, A–C). MRL/lpr developed nephritis with large UTALSs that consisted of B cells, T cells, macrophages, APCs, LVs, and HEVs (Figure 3, A–D). Immune cell marker expression was significantly higher in the RP than in the MRL/MpJ kidney, and significantly higher in the kidney and RP of MRL/lpr compared with MRL/MpJ (Figure 3E). Furthermore, there was a significant correlation between GL, TIL, or TLS indices and Il2ra, Cd3e, or Ptprc expression in RPs (Table 1).

Figure 3.

Development of UTALSs in nephritis models. MRL/MpJ and MRL/lpr (6 months) were used. (A) RP histopathology, including UTALSs and the renal cortex. UTALS is more well-developed in MRL/lpr than in MRL/MpJ. MRL/lpr kidney showing severe GLs and TILs; MT staining. (B and C) Localization of B220⁺ B cells and CD3⁺ T cells in the RP and kidneys of (B) MRL/MpJ and (C) MRL/lpr. (C) MRL/lpr showing developed UTALSs with numerous positive cells in the RP and renal inflammation. Arrows and arrowheads indicate positive cells in the glomerulus and tubulointerstitial regions, respectively; IHC. (D) Localization of IBA1⁺ macrophages, MHCII⁺ APCs, LYVE1⁺ LVs, and PNAd⁺ HEVs in MRL/lpr UTALSs. Insets indicate the magnified square area. PNAd⁺ HEVs localization with RP (upper panel) and urothelium (lower panel); IHC. (E) Immune cell marker gene expression in the kidney and RP; qPCR. Values represent the mean±SEM; n=7 or 10 (MRL/MpJ or MRL/lpr). Significant differences between kidney and RP or MRL/MpJ and MRL/lpr are indicated by ^#P<0.05, ^##P<0.01, or **P<0.01; Mann–Whitney U test. Scale bars, 100 μm (A, RP; D, PNAd), 50 μm (A, cortex; B, C), 300 μm (D, IBA1, MHCII, LYVE1). IM, inner medulla; Lu, lumen of the RP.

Table 1.

Correlation between urinary tract–associated lymphoid structures development indices (immune cell markers) and pathologic parameters in mice

Disease Parameter		Indices for Urinary Tract–Associated Lymphoid Structures Development (Expression in Renal Pelvis)
		Immune Cell Marker
Category	Parameter	CD25 (Il2ra)	CD3 (Cd3e)	B220 (Ptprc)
GL	Glomerular area	0.679**	0.779**	0.718**
	CD3+ cell density	0.685**	0.681**	0.715**
	B220+ cell density	0.774**	0.769**	0.737**
TIL	CD3+ score	0.684**	0.765**	0.787**
	B220+ score	0.755**	0.801**	0.699**
TLS	% of CD3+ cluster	0.713**	0.848**	0.850**
	% of B220+ cluster	0.703**	0.873**	0.860**

Value=ρ (Spearman's rank correlation coefficient). MRL/MpJ (n=7), MRL/lpr (n=10). **P<0.01. mRNA expression was normalized to that of Actb.

UTALSs of MRL/lpr at 3 months were smaller than at 6 months but larger than in same-aged healthy mice (Supplemental Figures 6 and 9). Moreover, 3-month-old MRL/lpr manifested AID,¹⁶ suggesting the involvement of systemic immune factors in UTALS development.

Altered Urothelium Barrier of the RP in Mice and Humans Showing Nephritis

UTALSs are in proximity with the RP lumen through the urothelium. Although aging and AID are well-established factors that contribute to the development of TLS,^4,5 we hypothesized that urothelium barrier (UB) integrity in the RP is altered under nephritic conditions, resulting in subsequent leakage of urine into the RP that may in turn contribute to UTALS development. To assess this phenomenon, we examined UBs in the RP. After 1 day of UUO in young C57BL/6N (2-month-old), the RP showed mild cellular aggregation in the RP parenchyma with decreased staining intensity of tight junction (TJ) molecules such as occludin (OCLN) and ZO-1, suggesting an association between UTALS formation and altered UB; however, UTALS size was smaller compared with that in aged or AID mice (Supplemental Figures 6, 9, and 10). To clearly characterize the association between UB integrity and UTALS formation, subsequent experiments were performed using MRL/MpJ and MRL/lpr.

The urothelium surface in MRL/MpJ was wrinkled at low magnification and the cell surface was irregular; however, the RP wall was thicker in MRL/lpr than in MRL/MpJ, several epithelial cells were absent, and alterations to the spherical or domed shape were observed in the RP epithelium of MRL/lpr (SEM; Figure 4, A–C).

Figure 4.

Altered UB integrity in the RPs of mice and humans with nephritis. MRL/MpJ and MRL/lpr (6 months) were used in (A–E), (G), and (H). (A and B) Surface ultrastructure of the RP in (A) MRL/MpJ and (B) MRL/lpr. Arrowheads indicate RP walls and arrows indicate lack of urothelium in MRL/lpr mice. Dotted lines indicate the cutting surface of renal papilla. (C) Surface ultrastructure of the MRL/lpr RP. Well-formed (arrows) or distorted (arrowheads) spherical cells are observed in some parts of the urothelium (left panel), a large area of the urothelium is lacking (middle panel), and alterations to the domed shape are observed (right panel); SEM. (D) Upk3a and TJ-associated molecule gene expression, and expression of TJ-associated molecules normalized to Upk3a (urothelium marker) in the RP; qPCR. (E and F) TJ-associated molecule localization in the RP of (E) mice and (F) humans. (E) Arrows indicate the faint area of OCLN⁺ reaction, and arrowheads indicate invagination of the urothelium. The square and its insets indicate the urothelium. Dotted lines indicate the border between the urothelium and RP parenchyma. (F) Localization patterns of ZO-1⁺ or OCLN⁺ reactions are altered in the urothelium of both patients; IF. (G) Localization of EBA (red stain). Asterisks indicate the area of the urothelium from which EBA leaked into the RP parenchyma in MRL/lpr; fluorescence microscopy. (H) Quantification of EBA leakage in RP; histoplanimetry. Values are shown as the mean±SEM (D and H). n=7 or 10 (MRL/MpJ or MRL/lpr) (D); n=7 (H). **P<0.01; Mann–Whitney U test. Scale bars, 500 μm (A and B, low magnification), 20 μm (A and B, high magnification; C), 50 μm (E–G). BV, blood vessels; CO, cortex; IM, inner medulla; Lu, lumen of RP.

On the basis of microarray data (Supplemental Material), the expression of urothelium markers and TJ molecules decreased in the RP of MRL/lpr compared with that of MRL/MpJ (Supplemental Table 4). qPCR of selected genes identified using microarray confirmed that OCLN (Ocln) levels significantly decreased in the RP of MRL/Ipr compared with MRL/MpJ (Figure 4D).

In MRL/MpJ, ZO-1 and OCLN localized to the urothelium in the RP, because OCLN⁺ staining showed a linear pattern at the apical side (Figure 4E). However, in MRL/lpr, OCLN⁺ staining was faint and localized in the basal portion of the urothelium, whereas some parts of the urothelium invaginated toward the RP. Furthermore, some ZO-1⁺ staining patterns in the urothelium were discontinuous instead of linear. In human patients (Figure 4F), TJ localization patterns were altered, showing discontinuous ZO-1⁺ staining and OCLN⁺ cells that were localized to the basal side and other areas of the urothelium, compared with normal human kidneys. Retrograde EBA administered from the urinary bladder to RP (Supplemental Figure 3) was retained and observed in the urothelium of MRL/MpJ RP; however, EBA significantly leaked into some parts of the developing UTALS in the MRL/lpr RP (Figure 4, G and H).

Next, we assessed TNF-α and IFN-γ as candidates for UB defects on the basis of previous organ studies.^19–21 Urine TNF-α and IFN-γ were significantly upregulated in patients with noninfectious nephritis compared with healthy controls, whereas cultured HUCs expressed one and two types of TNF-α and IFN-γ receptor genes, respectively (Figure 5, A and B). OCLN expression in HUCs significantly decreased after stimulation with IFN-γ or urine from patients with nephritis (Figure 5, C and D). Although nonstimulated HUCs formed several clusters with linear localization of TJ proteins, these clusters significantly decreased in size and number after IFN-γ stimulation (Figure 5, E and F).

Figure 5.

Effect of cytokines on the UB integrity in cultured HUCs. (A) Determination of cytokine levels in human urine. HC, healthy control. Nephritis, noninfectious chronic nephritis; immunoassay. (B) Receptor gene expression for TNF-α and IFN-γ in HUCs. THP1 monocytes were used as positive control. M, size marker; RT-PCR. (C) mRNA expression of TJP1 (gene encoding ZO-1) and OCLN in cultured HUCs after cytokine stimulation for 24 hours; qPCR. (D) mRNA expression of TJP1 and OCLN in cultured HUCs after cytokine stimulation from pooled urine from (A) for 3 hours; qPCR. (E) TJ-associated molecules in cultured HUCs after IFN-γ stimulation for 24 hours. Inset indicates the magnified square area; IF. (F) ZO-1⁺OCLN⁺ cluster size and number in cultured HUCs after IFN-γ stimulation for 24 hours; morphometry. Values are expressed as the mean±SEM (A, C, D, and F). n=9 or 12 (HC or nephritis) (A); n=4 (C and D); 20 area (F). *P<0.05; **P<0.01; Mann–Whitney U test (A, D, and F). Significant differences compared with the PBS control (Cont) are indicated by *P<0.05; Dunnett’s test (C). Scale bars, 50 μm (E).

Candidate Molecules Associated with UTALS Development

Next, we examined the response of cells composing UTALSs after urine stimulation. First, we performed microarray analysis using the RP and kidneys of MRL/MpJ or MRL/lpr (Supplemental Material). Using MRL/MpJ data, we detected several immune-associated GO terms, and STRING analysis of upregulated gene sets in the RP showed a large cluster of chemokines and associated receptor genes (Supplemental Figure 11, A–C), which are crucial in attracting immune cells.^{4,5,7–9,14(preprint)} Microarray and qPCR data (Figure 6A) revealed that Ccl1, Ccl8, Cxcl9, and Cxcl13 were highly expressed with their receptor genes in the MRL/lpr RP; we therefore focused on them as candidates associated with UTALS development.

Figure 6.

Chemokine-expressing cells in the RPs of mice and humans with nephritis and cytokine/chemokine induction in cultured cells by urine stimulation. MRL/MpJ and MRL/lpr (6 months) were used in (A–E). (A) Gene expression of chemokine ligands or their receptors in the kidney and RP; qPCR. (B) Percentage of CD45⁺ immune cells and vimentin⁺ stromal cells in MRL/MpJ-derived mLSs; IF. (C) Culture supernatant levels and gene expression of cytokines and chemokines in MRL/MpJ-derived mLSs after stimulation with 25% MRL/lpr urine for 4 hours; ELISA and qPCR. (D) Localization of chemokine-expressing cells in the MRL/lpr RP. Insets indicate the magnified square area; ISH. (E and F) Localization of (E) chemokines, vimentin⁺ stromal cells, and B220⁺ B cells in the MRL/lpr RP or that of (F) chemokines and PDPN⁺ stromal cells in the RP of the normal human kidney and kidneys of patients with noninfectious chronic nephritis. Yellow arrows indicate colocalization of chemokines with vimentin or PDPN. Insets indicate the magnified square area. White dots indicate the border between the urothelium and RP parenchyma; IF. (G) Gene expression of cytokines and chemokines in NIH3T3 fibroblasts after stimulation with 25% urine from MRL/lpr for 4 hours; qPCR. (H) Culture supernatant CXCL13 levels and CXCL13 and vimentin expression in NIH3T3 fibroblasts after 25% MRL/lpr urine stimulation for 4 hours; ELISA and IF. Values are expressed as the mean±SEM; n=7 or 10 (MRL/MpJ or MRL/lpr, A), n=4 (C, G, and H). Significant differences between kidneys and RP or MRL/MpJ and MRL/lpr are indicated by ^#P<0.05 and ^##P<0.01 or *P<0.05 and **P<0.01. Significant differences compared with the PBS control (Cont) are indicated by *P<0.05 and **P<0.01; Mann–Whitney U test. Scale bars, 10 μm (B and H), 100 μm (D), 50 μm (E and F). Lu, lumen of RP.

We next stimulated mouse mLSs composed of 87% CD45⁺ immune cells, 8% vimentin⁺ stromal cells, and others (Figure 6B) with urine in vitro. We used 25% urine, a concentration that stimulated the expression of cytokines/chemokines without altering Actb expression. Among the examined cytokines or candidate CCLs/CXCLs, healthy C57BL/6- or MRL/MpJ-derived mLSs significantly induced Ifng, Tnf, Cxcl9, or Cxcl13 expression after urine administration (Supplemental Figure 12, A–C). Notably, MRL/MpJ-derived mLSs exhibited elevated expression of these cytokines/chemokines after stimulation with MRL/lpr urine (Figure 6C). CXCL9 and CXCL13 tended to be strongly induced by urine from MRL/lpr compared with that from MRL/MpJ (Supplemental Figure 12D).

In MRL/lpr RP, Cxcl9 and Cxcl13 mRNA was detected in interstitial stromal cells, and CXCL9⁺ or CXCL13⁺ cells were positive for vimentin, a fibroblast marker (Figure 6, D and E). In humans, CXCL9⁺ or CXCL13⁺ cells were scarce in the normal kidney, but several CXCL9⁺ or CXCL13⁺ cells showed colocalization with podoplanin (PDPN), a marker of stromal immunofibroblasts,⁷ in patients with noninfectious chronic nephritis (Figure 6F) and infectious pyelonephritis (Supplemental Figure 13). On the basis of these results, we stimulated NIH3T3 fibroblasts with MRL/lpr-derived urine. Urine stimulation specifically induced CXCL13 expression, and increased fibroblast size and CXCL13 positivity (Figure 6, G and H); similarly, MRL/MpJ urine also induced CXCL13 expression (Supplemental Figure 12D). These results indicate that the examined chemokines are expressed in UTALS-forming cells and contribute toward attracting immune cells in situ. Furthermore, there was a significant correlation between GL, TIL, or TLS indices and CXCL9 and CXCL13 expression in the RPs (Table 2).

Table 2.

Correlation between urinary tract–associated lymphoid structures development indices (chemokines) and pathologic parameters in mice

Disease Parameter		Indices for Urinary Tract–Associated Lymphoid Structures Development (Expression in Renal Pelvis)
		Chemokine
Category	Parameter	Cxcl9	Cxcl13
GL	Glomerular area	0.504	0.688**
	CD3+ cell density	0.757**	0.696**
	B220+ cell density	0.747**	0.623*
TIL	CD3+ score	0.639*	0.738**
	B220+ score	0.489	0.509*
TLS	% of CD3+ cluster	0.650**	0.629**
	% of B220+ cluster	0.486	0.591*

Value=ρ (Spearman's rank correlation coefficient). MRL/MpJ (n=7), MRL/lpr (n=10). **P<0.01; *P<0.05. mRNA expression was normalized to that of Actb.

Unique Role of a Collagen Family in UTALS Development

Given that histopathologic analysis suggested a strong association between collagen fibers and UTALS development, we next focused on collagen families. STRING analysis of upregulated gene sets in mouse RP showed a cluster of several collagen genes (Supplemental Figure 14). Microarray analysis results of collagen genes highly expressed in the RP compared with the kidney in both mouse strains are summarized in Figure 7A. Col17a1 was found highly expressed in the RPs (Figure 7B). COL17A1, also known as bullous pemphigoid antigen II (BP180), is expressed in the epidermis.²² Ectopic COL17A1⁺ staining was strong in well-developed UTALSs but faint in the urothelium covering RPs that did not develop UTALSs (Figure 7C). Histologically, COL17A1 expression indices in the urothelium were significantly and positively correlated with those of UTALS development in MRL/lpr (Figure 7D). COL17A1⁺ staining was also detected in basal cells of the urothelium in human RPs with nephritis, and its staining intensity was faint in normal kidneys but stronger in infected kidneys (Figure 7E).

Figure 7.

COL17A1 associated with UTALS development. MRL/MpJ and MRL/lpr (6 months) were used in (A–D). (A) Collagen family relative gene expression in the RP and kidney; microarray. (B) Col17a1 expression in the kidney and RP; qPCR. (C) COL17A1 localization in the RP. COL17A1⁺ reactions are observed in the urothelium covering mature UTALSs (arrows), but not immature urinary UTALSs (asterisks). Insets: magnified square area; IHC. (D) Histopathologic correlations of COL17A1 expression indices (x axis) in the urothelium with UTALS development indices including its nuclear number (y axis, upper panel) or area ratio of COL17A1⁺ reactions to UTALS (y axis, lower panel) in MRL/lpr; histoplanimetry. (E) Localization of COL17A1 in the human RP. COL17A1⁺ reactions are only observed in the patient urothelium; IHC. Values are shown as the mean±SEM (B); n=7 or 10 (MRL/MpJ or MRL/lpr). **P<0.01; Mann–Whitney U test. **P<0.01; Spearman's correlation test (D). Bars, 500 μm (C), 100 μm (E). Lu, lumen of RP.

Microarray analysis revealed that immune-associated GO terms were significantly upregulated in mLSs after stimulation by rhCOL17A1 (Supplemental Material, Table 3). Predicted promoter region analysis of mouse Col17a1 using JASPAR revealed the highest scores in the binding site of FOS or JUN—transcriptional regulators mediated by several cytokines like TNF-α. FOS and JUN localized to the nuclei of both COL17A1⁻ and COL17A1⁺ urothelium covering mouse UTALSs (Figure 8A). Furthermore, FOS exhibited nuclear translocation in HUCs after stimulation with urine from patients with nephritis (Figure 8B). TNF-α significantly increased COL17A1, FOS, and JUN expression in HUC (Figure 8C).

Table 3.

Gene ontology of mimic lympoid structures cells after stimulation with rhCOL17A1

Pathway Gene Ontology Identifier	Pathway Description	Gene Count	False Discovery Rate
0006952	Defense response	88	3.46×10−39
0002376	Immune system process	99	2.76×10−33
0006955	Immune response	75	4.18×10−33
0051707	Response to other organism	79	8.53×10−33
0034097	Response to cytokine	69	6.08×10−32
0006954	Inflammatory response	54	1.82×10−30
0009605	Response to external stimulus	102	3.59×10−30
0071345	Cellular response to cytokine stimulus	61	9.88×10−29
0050896	Response to stimulus	183	2.25×10−27
0006950	Response to stress	114	3.20×10−25

Ten upregulated GO terms are listed. Stimulation with 0.4 μg/ml rhCOL17A1 for 4 hours; microarray.

Figure 8.

Upstream and downstream events of COL17A1 in UTALS development. MRL/MpJ and MRL/lpr (6 months) were used in (A). C57BL/6N (6 months) mice were used in (D–H). (A) Localization of FOS, JUN, and COL17A1 in the RP. Nuclear localization of FOS and JUN (arrows) is observed in small and large UTALSs covering COL17A1⁻ and COL17A1⁺ urothelium in MRL/lpr. Lu, lumen of RP; IF. (B) Nuclear translocation of FOS after 24 hours of treatment with 25% human urine in HUCs; IF. (C) COL17A1, FOS, and JUN expression after 24 hours of treatment with 1.0 ng of TNF-α in HUCs; qPCR. (D–F) Assessment of (D) cell viability, (E) gene expression, and (F) IL-6 supernatant level of mLS cells after 24 hours (D) and 3 or 15 hours of treatment (E and F) with BSA or rhCOL17A1; (D) MTT assay; (E) qPCR; (F) ELISA. (G) Chemokine or upregulated gene expression in mLT cells after stimulation with 0.4 μg/ml rhCOL17A1 for 4 hours. Candidate chemokines or the top four upregulated genes are listed; microarray. (H) Il6, Cxcl9, and Il22 expression in B, T, and CD11b⁺ cells isolated from mLSs, and NIH3T3 fibroblasts after stimulation with 0.4 μg/ml rhCOL17A1 for 4 hours; qPCR. Values are shown as the mean±SEM (C–F and H); n=4 (C–E and H); n=4–7 (F); n=3 (G). Significant differences compared with the PBS control (Cont; C and G) or BSA (D–F and H) are indicated by *P<0.05 or **P<0.01; Mann–Whitney U test (C, D, and H), Dunnett’s test (E and F), or Students t test (H). Scale bars, 50 μm (A and B).

Next, we assessed the role of COL17A1 in UTALS development by stimulating mLSs with rhCOL17A1, which increased the viability of mLS cells (Figure 8D). Inflammatory gene expression (Ifng, Il1b, Il6) at 3 hours and IL-6 secretion at 3 and 15 hours in mLSs were significantly increased post rhCOL17A1 stimulation (Figure 8, E and F). For microarray analysis (Supplemental Material), Cxcl9 from among the chemokine candidates and immune response genes (Il22, Edn1, Cxcl2, Ptgs2) were significantly upregulated after rhCOL17A1 stimulation in mLSs, indicating that they act downstream of the COL17A1-related pathway (Figure 8G). Notably, Il6, Cxcl9, and Il22 expression was significantly upregulated in all examined cells separated from mLSs, CD11b⁺ cells (APC, monocytes) and NIH3T3 fibroblasts, and CD11b⁺ cells after 4 hours of rhCOL17A1 stimulation, respectively (Figure 8H).

Discussion

This study demonstrated that UTALSs form in the RPs of humans and mice with chronic nephritis and revealed a mechanism of immune-related kidney lesion induction, regardless of RP-related infections. Although LSs were not clearly developed in normal human kidney RPs, reticular fiber between LSs was observed,²³ indicating that RPs may provide a scaffold for immune cells to form LSs. Furthermore, UTALS development appears to require stimulation that alters local immune microenvironments, chemotactic factor–producing cells, and immune cell entry. As shown in C57BL/6N and a previous study,⁴ aging is a crucial factor for UTALS development.

Although pyelonephritis is primarily caused by UT infections,^10,24 our noninfectious, specific pathogen–free mice developed UTALS. MRL/lpr accumulate aberrant lymphocytes in systemic LSs owing to the lpr mutation in Fas, an apoptotic gene.²⁵ Therefore, this mutation and the subsequent inflammatory phenotype represents an established mechanism underlying LS development. In fact, a recent study using AID-prone (NZBxNZW)F1 also reported TLSs in the RPs,^4,26 and severe cell aggregations were observed in the RPs of humans with pyelonephritis.²⁷ In addition to AID and infection, our data suggest that urine is another unique factor contributing to UTALS formation in the RP. Nephritic urine contains bioactive molecules, including IFN-γ, TNF-α, and chemokines.^28–30 Notably, MRL/lpr urine stimulated cytokine/chemokine production in mLSs and CXCL13 in fibroblasts. Healthy C57BL/6N and MRL/MpJ urine also partially stimulated this production. Hence, the contact between RP cells and urine may play a crucial role in UTALS development via chemokine production, whereas change in the abundance of urinary bioactive molecules modifies the immunologic response of RP cells.

Urine leakage from the RP lumen into the parenchyma through the urothelium is required to stimulate RP cells in vivo. Infections or other diseases may rupture urothelial TJs, composed of OCLN or ZO-1, thereby disrupting the UB.³¹ Our findings indicated that UB integrity was altered in RPs that form UTALSs, as demonstrated by the abnormal localization of TJ-associated molecules, especially decreased OCLN expression in nephritic mice. However, because altered UB integrity did not strongly induce well-developed UTALS formation as found in aged or nephritic mice (Supplemental Figure 10), we considered that the altered urine condition due to renal pathology is also important for UTALS development. Notably, IFN-γ levels were increased in nephritis urine, whereas OCLN levels were decreased by IFN-γ and urine from nephritis mice, and TJ-mediated cell cluster formation was inhibited by IFN-γ in HUCs. TLS formation in the urinary bladder due to altered UB integrity has been reported previously.^14(preprint) Leakage of ovo-albumin from the epithelium to the parenchyma due to inflammation and development of dry eye and decreased OCLN expression was also reported.³² Furthermore, proinflammatory cytokines reportedly alter epithelial barrier integrity in the intestinal and brain vessels.^19–21 These findings suggest a potential mechanism underlying UTALS formation, wherein altered UB integrity, caused by immune mediators from the RP and immunoreactive urine, lead to UTALS formation via leakage of urine from the RP lumen.

COL17A1 was ectopically²² localized to the urothelium to cover the UTALS and generated immunoreactive activity. Although there is no previous record of COL17A1 function in the urothelium, GO analysis of COL17A1-stimulated mLS suggests that COL17A1 plays a role in immunologic events. The FOS/JUN pathway, mediated by urinary immune mediators such as TNF-α, was also implicated in COL17A1 expression induction in the urothelium. Notably, nuclear localization of FOS/JUN was observed before COL17A1 expression in mouse RP urothelium. Furthermore, IL-22, which acts downstream of COL17A1, is an immunologic modulator between the epithelium and parenchyma of the skin, inducing antimicrobial peptides in keratinocytes.³³ IL-22 is also elevated in patients with UT infections and nephritis.^34,35 Furthermore, our data suggest that CD11b⁺ cells (APC, monocytes) produce IL-22 in response to COL17A1. Therefore, urine- and/or RP-derived molecules, particularly TNF-α, may activate the FOS/JUN pathway, increasing COL17A1 expression in RP parenchymal cells and subsequently inducing UTALS development via activation of downstream molecules such as IL-22.

RP stromal cells are involved in UTALS development by producing chemotactic factors. Several studies have indicated that CXCL13 is crucial for TLS development.^4,5,8,36 Previously, we also found that CXCL9- or CXCL13-expressing stromal cells in perivascular TLS formed in the MRL/lpr kidneys.³⁷ CXCLs were also considered important for UTALS development, as the expression of Cxcl9 and Cxcl13 and their receptor genes Cxcr3 and Cxcr5 was elevated in mouse RPs. Stromal cells expressing these chemokines localized to mouse and human UTALS, whereas mouse urine stimulated CXCL13 production in mLSs and fibroblasts, and CXCL9 production in mLSs. Additionally, COL17A1 exposure increased Cxcl9 expression in mLSs and NIH3T3 fibroblasts. CXCL9 and CXCL13 are strongly induced by IFN-γ and bind CXCR3 on either helper or effector T cells and innate-type lymphocytes and CXCR5 on B cells.^38,39 Specifically, PDPN⁺ chemokine-expressing stromal immunofibroblasts,⁷ including those in the human RP, are key modulators of UTALS and their phenotype is regulated by IL-22.⁷ Notably, the renal parenchyma is directly connected to RPs via connective tissues composed of stromal cells. Moreover, we revealed a positive correlation between UTALS development and both TIL and TLS formation in the kidneys.

Although altered local immune conditions due to aging, infection, or AID are crucial factors for inducing the development of UTALS, we propose another pathway of pathogenesis involving RP stromal cells in UTALS development and renal pathogenesis via the connective tissue pathway resulting in either TLSs or TILs (Supplemental Figure 15). Specifically, bioactive factors from the urine and RP alter the UB, allowing urine leakage from the RP lumen into its parenchyma, where it immunologically stimulates RP stromal cells to produce cytokine/chemokines that attract immune cells. COL17A1 induction in the urothelium appears to be involved in these processes, and pathologic crosstalk was observed between UTALSs and renal lesions stemming from TILs or TLSs. Therefore, UTALS formation caused by urine–UB alteration may represent an additional mechanism underlying the pathogenesis of chronic nephritis, regardless of infection.

Disclosures

All authors have nothing to disclose.

Funding

This work was supported by Japan Society for the Promotion of Science grants 19K22352 and 21H04751, Suzuken Memorial Foundation grant 18-057, Nakajima Foundation grant H31, and Suhara Memorial Foundation grant H30). The funders had no role in the study design, data collection, analysis, decision to publish, or preparation of the manuscript.

Data Sharing Statement

The microarray data are available in the Supplemental Material. The remaining data that support the findings of this study are available from the corresponding author upon reasonable request.

Supplemental Material

This article contains the following supplemental material online at http://jasn.asnjournals.org/lookup/suppl/doi:10.1681/ASN.2021040575/-/DCSupplemental.

Supplemental Data 1. Summary of genes expressed in the kidney and renal pelvis of MRL/MpJ and MRL/lpr mice (microarray analysis).

Supplemental Data 2. Summary of genes expressed in the murine mimic lymphoid structures (mLSs) after stimulation by human recombinant collagen type XVII α 1 (microarray analysis).

Supplemental Figure 1. Method used to localize the mouse lymphoid structures (LSs) formed in the renal pelvis (RP).

Supplemental Figure 2. Appearance of the mouse LSs formed in the RP and isolation of RP.

Supplemental Figure 3. In vivo methods for analyzing mouse urinary tract–associated lymphoid structures (UTALSs).

Supplemental Figure 4. Cell infiltration of the human renal pelvis with or without urinary tract infection.

Supplemental Figure 5. Histology and cells composing small UTALSs in mice.

Supplemental Figure 6. Cell composition and age-related changes in the UTALSs of C57BL/6N.

Supplemental Figure 7. Histology of UTALSs in spontaneous chronic GN mice.

Supplemental Figure 8. Indices for abnormalities of systemic immune conditions and renal histopathology in healthy and a noninfectious nephritis mouse model.

Supplemental Figure 9. Age-related changes in the UTALSs of MRL/lpr.

Supplemental Figure 10. Histopathologic changes in mouse RP after unilateral ureter obstruction (UUO).

Supplemental Figure 11. Chemokine family expression in mouse RP.

Supplemental Figure 12. Housekeeping gene and chemokine gene expression in cultured cells after stimulation by mouse urine.

Supplemental Figure 13. CXCL9- or CXCL13-expressing fibroblasts in the human RP showing infectious pyelonephritis.

Supplemental Figure 14. Collagen family expression in the mouse RP.

Supplemental Figure 15. Formation of UTALSs in the RP is associated with alterations in the urine–urothelium barrier.

Supplemental Table 1. Information regarding antibodies used in this study.

Supplemental Table 2. Information regarding primers and probes used in this study.

Supplemental Table 3. Summary of urothelium marker gene expression in the mouse kidney and renal pelvis.

Supplemental Table 4. Summary of gene expression for urothelium markers and tight junction molecules in the mouse renal pelvis.