The renal inflammatory network of nephronophthisis

Abstract

Renal ciliopathies are the leading cause of inherited kidney failure. In autosomal dominant polycystic kidney disease (ADPKD), mutations in the ciliary gene PKD1 lead to the induction of CCL2, which promotes macrophage infiltration in the kidney. Whether or not mutations in genes involved in other renal ciliopathies also lead to immune cells recruitment is controversial. Through the parallel analysis of patients’ derived material and murine models, we investigated the inflammatory components of nephronophthisis (NPH), a rare renal ciliopathy affecting children and adults. Our results show that NPH mutations lead to kidney infiltration by neutrophils, macrophages and T cells. Contrary to ADPKD, this immune cell recruitment does not rely on the induction of CCL2 in mutated cells, which is dispensable for disease progression. Through an unbiased approach, we identified a set of inflammatory cytokines that are upregulated precociously and independently of CCL2 in murine models of NPH. The majority of these transcripts is also upregulated in NPH patient renal cells at a level exceeding those found in common non-immune chronic kidney diseases. This study reveals that inflammation is a central aspect in NPH and delineates a specific set of inflammatory mediators that likely regulates immune cell recruitment in response to NPH genes mutations.

Introduction

The primary cilium is a solitary antenna-like structure protruding from the apical surface of renal epithelial tubular cells, which is believed to act as complex macromolecular sensors and signal transducers (1,2).

Renal ciliopathies are inherited disorders caused by mutations in genes encoding proteins that localize to primary cilia (3,4). These renal ciliopathies share common features, including tubular dilation, interstitial fibrosis and loss of tubular cell differentiation. Yet, the spectrum of renal ciliopathies encompasses genetically distinct and phenotypically heterogeneous diseases that manifest from early-childhood to late-adulthood eventually leading to kidney failure. Autosomal dominant polycystic kidney disease (ADPKD), the most common ciliopathy manifesting in adulthood, is characterized by the development of multiple renal cysts resulting in progressive kidney enlargement and kidney function decline. Most ADPKD cases are caused by inactivating mutations in PKD1, which encodes polycystin 1 (PC1) that localizes to the ciliary membrane. In contrast, nephronophthisis (NPH), the second most prevalent renal ciliopathy, is a rare autosomal recessive disorder that manifests in children and young adults by enuresis due to impaired urine concentrating ability and progressive kidney function decline. Although some cysts are observed in NPH, cyst burden remains marginal compared to ADPKD, and the kidneys mostly appear small and fibrotic. NPH is a genetically heterogeneous disorder caused by mutations in >20 genes identified so far (5). NPHP1 mutations are the most common genetic cause, accounting for 40–50% of identified causative mutations in NPH (6). Most of the proteins encoded by NPHP genes assemble in functional modules that control cilia morphology and gate protein entry and exit to and from the cilia (7,8). Yet, how disruption of these complexes translates into kidney damage is poorly understood.

Primary cilia of renal tubular cells are required for proper kidney development and maintenance (9). Yet, none of the mutations reported in renal ciliopathies abolish ciliogenesis. Although renal tubular cells remain ciliated in renal ciliopathies, they may display either increased or decreased cilia length (10,11). It is believed that the genetic defects involved in renal ciliopathies adversely affect cilia composition and signaling, which results in modifications of tubular cell behavior that, in turn, alter kidney morphology and function. Even outside cilia, NPHP regulates cellular junctions and epithelialization, which may also contribute to NPH-associated tubular defects (12).

Lessons gathered from acquired chronic kidney disease (CKD) models indicate that progressive renal scarring involves interplay between tubular, immune and mesenchymal cells that prompts immune cell infiltration, fibroblast activation and nephron loss (13–15).

Concordant evidence has pinpointed an important role of renal inflammation in the progression of ADPKD. Indeed, macrophages infiltrate the kidney of ADPKD patients and mice and promote cyst growth (16). Cilia signaling seem to be directly involved in this process as cilia ablation prevents macrophage recruitment in an orthologous mouse model of ADPKD (17). Mechanistically, cilia positively regulate the expression of the macrophage chemoattractant CCL2 in Pkd1-deficient tubular cells both in vivo and in vitro. Consistently, kidney-specific disruption of Ccl2 or inhibition of the CCL2 receptor reduces cyst burden in Pkd1 mutant mice (17,18).

In contrast to ADPKD, the role of immune cells in NPH has received little attention. Recent evidence suggests that renal inflammation may be involved in the disease. First, the inactivation of TLR2, a critical receptor of the immune system, prevents renal damage in an orthologous mouse model of a rare form of NPH caused by a mutation in Glis2/Nphp7 (Glis2^lacZ/lacZ mice) (19,20). Besides, we previously reported that LKB1, a ciliary kinase involved in the control of cell size and metabolism, interacts with NPHP proteins including NPHP1 (17). In human, heterozygous mutation in STK11, the gene encoding LKB1, does not lead to NPH but to Peutz–Jeghers syndrome, characterized by benign tumors of the skin and intestine, as well as cancer. Homozygous mutations in LKB1 have not been reported in humans, whereas mice with bi-allelic inactivation of Lkb1 die at mid-gestation (21). Yet, mice bearing a selective inactivation of Lkb1 in renal tubules (Lkb1^ΔTub mice) develop a NPH-like phenotype recapitulating the consequences of NPHP1 loss in humans including impaired urine concentration, thickened tubular basement membranes and fibrosis (17). The relevance of this mouse model to human NPH is not only sustained by the interaction of LKB1 with NPH proteins but also by parallel functions of NPHP1 and LKB1 in tubular cells. Indeed, in vitro, both NPHP1 and LKB1 repress the expression of CCL2, whereas CCL2 upregulation and macrophage recruitment are observed in vivo at an early time point of the development of the disease in Lkb1^ΔTub mice (17). Importantly, CCL2 dependent recruitment of immune cells has been repeatedly shown to promote renal fibrosis in different acquired CKD models (15,21). In contrast to these observations made in NPH mice models, most of the histopathologic studies performed in human NPH did not report immune cells infiltration as a notable feature of the disease (5). Yet, these studies have limitations: they are rare, included a limited number of patients (≤9 patients) and lacked standardized quantitative assessment of renal inflammation.

Considering the paucity and the inconstancy of the data regarding renal inflammation in NPH, we decided to explore the interplay between immune and tubular cells in NPH through a translational approach combining data from transgenic animals and human patients.

Results

NPH is associated with macrophage recruitment and enhanced CCL2 expression

To elucidate the contribution of immune cell recruitment in the phenotype of NPH, we first aimed to assess whether, similar to ADPKD, macrophage infiltration occurs in human NPH. Histology inspection of kidney tissues from NPH patients mostly bearing NPHP1 mutations revealed significant infiltration by immune cells (Fig. 1A). Immunolabelling identified macrophages (CD68-positive cells) as an important contingent of cells infiltrating the kidneys from NPH patients as compared with control individuals (Fig. 1B and C). As CCL2 has been shown to be the major chemokine responsible for macrophage recruitment in ADPKD, we analyzed primary urine-derived renal epithelial cells (UREC) of patients bearing NPHP1 mutations and controls, including both age-matched controls and healthy relatives. UREC derived from NPHP1 patients showed decreased expression of NPHP1 mRNA associated with increased expression of LCN2 transcript, a marker of tubular injury known to correlate with the progression of CKD (Supplementary Material, Fig. S1A and B). Tubular cells derived from NPHP1 patients also showed enhanced expression of CCL2 transcript (Fig. 1D and Supplemental Material, Fig. S1C). Consistently, we observed that NPHP1 patients display a higher urinary excretion rate of CCL2 than controls (Fig. 1E). Of note, in NPHP1 patients, CCL2 urinary excretion was not correlated with eGFR (Pearson correlation P = 0.218, R² = 0.183) and no difference was observed according to the patient genotype (homozygous deletion vs heterozygous compound mutations). Collectively, these data revealed that, similarly to ADPKD, NPH patients show increased renal tubular expression of CCL2 associated with macrophage recruitment.

Figure 1

Human NPH is associated with renal inflammation, macrophage infiltration and CCL2 induction. (A) Representative images of PAS-stained kidney biopsies from control and 3 NPH patients. Scale bar: 100 μm. (B) Representative images of CD68 (macrophages) immunostaining in kidney biopsies from control and three NPH patients. Scale bar: 100 μm. (C) Quantification of CD68-positive staining area in kidney sections from control and NPH patients. (D) CCL2 mRNA expression in primary UREC derived from urine from controls (controls) and NPHP1 patients (NPHP1). (E) Urinary CCL2 secretion in controls and NPHP1 patients. Filled circles indicate NPHP1 patients bearing homozygous deletion, whereas filled squares indicate those with heterozygous compound mutations. (C–E) Each dot represents one individual. Bars indicate mean. Mann–Whitney t test, ^***P < 0.001. AU: arbitrary unit.

Tubule-specific Ccl2 inactivation does not prevent NPH-like phenotype

Deletion of Lkb1 in the distal tubule of mouse kidney cells results in an NPH phenotype associated with CCL2 upregulation and macrophage recruitment (17,22). To determine if the deletion of Ccl2 together with Lkb1 ameliorates the NPH-like renal disease, we crossed tubule-specific Lkb1^Δtub mice with mice bearing Ccl2 floxed alleles and compared mice with distal tubular inactivation of Lkb1 alone (Lkb1^Δtub), Ccl2 alone (Ccl2^Δtub) or Lkb1 and Ccl2 (Lkb1^Δtub; Ccl2^Δtub) with littermates controls. At 10 weeks of age, quantitative RT-PCR revealed that Ccl2 inactivation in distal tubules drastically blunted Ccl2 induction in Lkb1-deficient kidneys (Fig. 2A). Macroscopic inspection revealed irregular kidneys in both Lkb1^Δtub and Lkb1^Δtub; Ccl2^Δtub mice associated with reduced kidney size, which was even more pronounced in Lkb1^Δtub; Ccl2^Δtub mice (Fig. 2B, C and Supplemental Material, Fig. S2A, B). Lkb1^Δtub and Lkb1^Δtub; Ccl2^Δtub mice displayed similar urine concentration defect and loss of kidney function (Fig. 2D-F). Renal histology revealed similar tubular and glomerular basement membranes thickening, interstitial inflammation, fibrosis and tubular dilation in Lkb1^Δtub and Lkb1^Δtub; Ccl2^Δtub mice. Consistently, Ccl2 inactivation neither affected the upregulation of the tubular injury marker Lcn2, nor the extracellular matrix deposition markers Col1a1 and Tgfb1, in Lkb1-deficient kidneys (Fig. 2G–L). Ccl2^Δtub animals had normal renal function and histology (Supplemental Material, Fig. S2C–J). Unexpectedly, contrary to what was observed in Pkd1-deficient mice (17,18), Ccl2 tubular inactivation did not prevent macrophage infiltration of Lkb1-deficient kidneys as judged by quantification of both F4/80 immunolabelling and Adgre1 transcript (Fig. 2M–O). Collectively, these results demonstrate that, whereas distal tubules represent the major source of renal CCL2 in Lkb1^Δtub mice, CCL2 induction in those cells does not mediate macrophage infiltration of the kidney nor kidney damage.

Figure 2

Tubule specific Ccl2 inactivation does not prevent renal lesion development nor macrophage infiltration of Lkb1-deficient kidneys. (A) Ccl2 mRNA expression in kidneys from control, Lkb1^ΔTub and Lkb1^ΔTub; Ccl2^ΔTub mice at 10 weeks. (B) Representative kidneys from control, Lkb1^ΔTub and Lkb1^ΔTub; Ccl2^ΔTub mice. Scale bar: 5 mm. (C) Kidney weight to body weight ratio (KW/BW). (D–E) Urinary flow rate (D) and urine osmolality (E). (F) Plasma BUN. (G–H) Representative PAS-stained kidney sections (G) and histology scores (H) from control, Lkb1^ΔTub and Lkb1^ΔTub; Ccl2^ΔTub mice. Scale bars: 5 mm (left panel), 100 μm (middle, right panel). (I) Lcn2 mRNA expression in kidneys from the same animals. (J) Representative sirius red stained kidney sections of the same animals Scale bars: 5 mm (left panel), 100 μm (right panel). (K–L) Col1a1 (K) and Tgfb1 (L) mRNA expression in kidneys from control, Lkb1^ΔTub and Lkb1^ΔTub; Ccl2^ΔTub mice. (M–N) Representative images (M) and quantification (N) of F4/80 (macrophages) immunostaining in the same animals. Scale bar: 100 μm. (O) Adgre1 mRNA expression in kidneys from control, Lkb1^ΔTub and Lkb1^ΔTub; Ccl2^ΔTub mice. (A, C–F, H–I, K–L, N–O) Each dot represents one individual mouse. All mice were 10 weeks old. Bars indicate mean. Kruskal–Wallis test, ^*P < 0.05, ^**P < 0.01, ^***P < 0.001.

Early NPH renal disease in mice is associated with a prominent immune signature

So far, our findings revealed macrophage infiltration in NPH but, as opposed to ADPKD, this recruitment appears independent of tubular CCL2. To approach the mechanisms underlying renal inflammation in NPH in an unbiased manner, we compared microarray-based transcriptomes from Lkb1^ΔTub and Glis2^lacZ/lacZ kidneys (17,19) with controls. Both mouse models recapitulated the features of human NPH: polyuria followed by progressive interstitial fibrosis, tubular basement membrane thickening, tubular dilations and immune cell infiltration (17,19,23). Transcriptome datasets were obtained at an early time point of disease development (5 and 4 weeks, respectively) when histological lesions were sparse and renal function decline mild (17,19). Gene overlap analysis identified 1262 genes that were commonly deregulated in the two models (Fig. 3A, B and Supplemental Material, Table S3). Gene set enrichment analysis (GSEA) of these genes identified a total of 72 enriched pathways [false discovery rate (FDR) < 0.05]. Among those pathways, eight were downregulated, consisting mostly of metabolic processes, whereas 64 biological processes were upregulated (Supplemental Material, Tables S4–5). Among the latter, the GSEA revealed a high enrichment in biological processes linked to immune response and inflammation (Fig. 3C). Network analysis showed a marked association of the common regulated genes with immune pathways (Fig. 3D and Supplemental Material, Fig. S3). This unbiased analysis identified renal inflammation as an early and prominent phenomenon in the course of NPH mouse models.

Figure 3

Comparative renal transcriptome analysis in Glis2 mutant mice and kidney-specific Lkb1 deficient mice. (A) Venn diagram showing intersection between Glis2^lacZ/lacZ (left; 23 957 genes) and Lkb1^ΔTub (right; 1991 DEGs according to FDR < 0.05) datasets. Red numbers: upregulated; blue numbers: downregulated. (B) Jointly up- and downregulated genes in Lkb1^ΔTub mice (Lkb1 dataset) and Glis2^lacZ/lacZ mice (Glis2 dataset). Genes with FDR < 0.01 are represented. Pearson correlation R² = 0.56, P < 0.0001. (C) GSEA on the 1262 common regulated genes revealed 72 pathways overrepresented (FDR < 0.05); the 20 most significantly upregulated are presented. See also Supplemental Material, Tables S4 and 5. (D) Summarized subnetwork representation of significantly enriched Biological Processes GO terms (MSigDB v7.1, C5-BP) linked to immune/inflammatory pathways derived from the GSEA of the 1262 common regulated genes in Glis2^lacZ/lacZ and Lkb1^ΔTub kidneys. Nodes are connected according to a similarity score > 0.5, and pathway clusters are hand-annotated according to the most representative GO-BP terms. The full network representation is available in Supplemental Figure S3.

Lkb1 tubular inactivation activates multiple chemotactic pathways and drives the recruitment of distinct immune cell populations independently of CCL2

To identify specific cytokines that might illuminate CCL2 independent immune cell recruitment in NPH, we matched the common upregulated genes between Lkb1^ΔTub and Glis2^lacZ/lacZ kidneys with UniProt database. This analysis retrieved 17 pro-inflammatory cytokines, other than CCL2 (Supplemental Material, Table S6) that have been implicated in macrophage, neutrophil, T cell and/or dendritic cell chemotaxis. Quantifying these transcripts in kidneys from Lkb1^Δtub, Lkb1^Δtub; Ccl2^Δtub and control mice, we observed that Ccl2 disruption had no impact on the upregulation of these immune mediators in Lkb1 deficient kidneys (Fig. 4A and Supplemental Material, Fig. S4). In line with the different immune cell populations attracted by these cytokines, immunolabelling demonstrated enhanced neutrophils (Ly-6B.2-positive cells) and T cells (CD3-positive cells) recruitment in both kidneys from Lkb1^Δtub and Lkb1^Δtub; Ccl2^Δtub mice as compared with control mice (Fig. 4B–F). These results uncover a CCL2-independent inflammatory network in NPH-like renal disease.

Figure 4

NPH cytokine signature and immune cell infiltrations occurred independently of CCL2 in Lkb1 deficient mice. (A) Heatmap showing Z-scores computed on mRNA expression measured by qPCR of the indicated cytokines in kidneys from 10-week old control, Lkb1^ΔTub and Lkb1^ΔTub; Ccl2^ΔTub mice. See also Supplemental Figure S4. (B–C) Representative images (B) and quantification (C) of Ly-6B.2 (neutrophils) immunostaining of kidney sections from control, Lkb1^ΔTub and Lkb1^ΔTub; Ccl2^ΔTub mice at 10 weeks. Scale bar: 100 μm. (D–E) Representative images (D) and quantification (E) of CD3 (T lymphocytes) immunostaining of kidney sections from the same animals. Scale bar: 100 μm. (F) Cd3 mRNA expression in kidneys from the same animals. (B, D–E) Each dot represents one individual mouse. Bars indicate mean. Kruskal–Wallis test, ^*P < 0.05, ^***P < 0.001.

The identified cytokine signature is not a general feature of renal ciliopathies

At this point, our findings identified a set of pro-inflammatory cytokines that is expressed early on in the course of NPH renal disease in mice. We next asked whether the observed immune signature was consistent across renal ciliopathies. To address this question, we compared Lkb1^Δtub mice, mimicking human NPH, to an orthologous model of ADPKD, in which post-natal inactivation of Pkd1 (Pkd1^Δtub mice) results in slow-onset PKD, recapitulating human disease. We focused on early time point of disease development (5 and 10 weeks, respectively). At 5 weeks of age, Lkb1^Δtub mice present scarce renal lesions, whereas Pkd1^Δtub kidneys showed mild tubular dilations 6 weeks after the induction of Pkd1 recombination, leading to kidney enlargement (Fig. 5A and B). Quantitative RT-PCR analysis confirmed a marked upregulation of the NPH inflammatory signature in the kidneys from Lkb1^Δtub mice at 5 week of age (Fig. 5C and Supplemental Material, Fig. S5). In contrast, although Pkd1-deficient kidneys displayed a slight but significant induction of CCL2 (Supplemental Material, Fig. S5A), most of the other cytokines precociously triggered by Glis2 or Lkb1 deletion were not induced 6 weeks after the induction of Pkd1 recombination (Fig. 5C and Supplemental Material, Fig. S5). Reanalyzing RNA-seq data from a study involving Pkd2^Δtub mice, another orthologous rodent model of ADPKD (24), further confirmed these results (Supplemental Material, Fig. S6 and Supplemental Material, Table S7).

Figure 5

The early inflammatory signature of NPH is not a common feature of renal ciliopathies. (A) Representative PAS-stained kidney sections from 5-week old male Lkb1^ΔTub and 10-week old male Pkd1^ΔTub mice. Scale bars: 100 μm. (B) KW/BW from 5-week old control and Lkb1^ΔTub animals and 10-week old control and Pkd1^ΔTub mice. (C) Heatmap showing Z-scores computed on mRNA expression measured by qPCR of the indicated cytokines in kidneys from 5-week old control and Lkb1^ΔTub mice, and 10-week old control and Pkd1^ΔTub mice. See also Supplemental Figure S5. (D) Representative images of F4/80 (macrophages) immunostaining of kidney sections from 5-week old control and Lkb1^ΔTub mice and 10-week old control and Pkd1^ΔTub mice. Scale bar: 100 μm. (E) Adgre1 mRNA expression in the same mice. (F–G) Representative images (F) and quantification (G) of Ly-6B.2 (neutrophils) immunostaining of kidney sections from 5-week old control and Lkb1^ΔTub mice and 10-week old control and Pkd1^ΔTub mice. Scale bar: 100 μm. (H) Cd3 mRNA expression in kidneys from the same animals. (B, E, G–H) Each dot represents one individual mouse. Bars indicate mean. Mann–Whitney test, ^**P < 0.01, ^***P < 0.001. All animals were males.

Immunostaining and qPCR analysis demonstrated an infiltration of Lkb1-deficient kidneys by neutrophils, macrophages and T cells. In contrast, only rare and focal macrophage infiltrates surrounding expanding cysts were observed in Pkd1^Δtub mice (Fig. 5D–H). Collectively, these data pinpoint important divergence in the inflammatory processes triggered by Pkd1 or Lkb1 deletion. Although the inactivation of the former produced tubular dilations associated with focal and CCL2-dependent macrophage recruitment around expanding cysts, the latter is characterized by the upregulation of multiple cytokines driving kidney infiltration by neutrophils, macrophages and T cells.

Identifying the inflammatory network associated with loss of NPHP1 function in humans

Considering the distinct immune cell populations infiltrating the kidneys of Lkb1^Δtub mice, we aimed to determine if such phenomena were also found in NPH patients. CD15 and CD3 immunostaining revealed increased neutrophils and T cells recruitment to the renal parenchyma of NPH patients compared with controls (Fig. 6A and B). In NPH patients, neutrophil infiltrates correlated with the severity of renal lesions evaluated by the interstitial fibrosis and tubular atrophy (IFTA) score (Pearson correlation P = 0.02, R² = 0.552), whereas no correlation was observed with T cells infiltrate (Pearson correlation P = 0.69, R² = 0.027). In contrast, these immune cell populations were less abundant in kidney biopsies from patients suffering from acquired non-immune CKD caused by diabetes or hypertension (Fig. 6A and B), supporting the notion that a specific inflammatory network delineates human NPH.

Figure 6

The immune signature of murine NPH is also prominent in human NPH. (A) Representative images and quantification of CD15 (neutrophils) immunostaining of kidney biopsies from control, NPH and CKD patients. Scale bar: 100 μm. Asterisk: unspecific tubular staining, black arrow: intravascular neutrophils in a glomerulus, red arrow, interstitial neutrophils. (B) Representative images and quantification of CD3 (T lymphocytes) immunostaining of kidney biopsies from control, NPH and CKD patients. Scale bar: 100 μm. (C-L) CCL5 (C), CCL19 (D), CX3CL1 (E), CXCL1 (F), CXCL10 (G), CXCL16 (H), CXCL17 (I), IL1RN (J), IL33 (K), LGALS9 (L) mRNA expression expressed in Log10 in primary UREC from controls (Controls), NPHP1 patients (NPHP1) and non-ciliopathy CKD patients (CKD). (A–L) Each dot represents one individual. Bars indicate mean. (A–B) Kruskal–Wallis test, ^***P < 0.001. (C–L) One-way ANOVA followed by Tukey–Kramer test, ^*P < 0.05, ^**P < 0.01, ^***P < 0.001.

To determine if the cytokine network that we identified as features of NPH-like mouse models was relevant to human NPH, we quantified the mRNA abundance of its components in UREC from NPHP1 patients and controls. As there are no human orthologs for CCL6, 9 and 12, we focused our analysis on the 14 other inflammatory mediators. Indeed, we found eight transcripts upregulated in human tubular cells with NPHP1 mutations (CCL5, CXCL1, CXCL10, CXCL16, CXCL17, CX3CL1, IL1RN and LGALS9), whereas two were not differentially regulated (CCL19 and IL33) and four were not detected (CXCL9, CXCL12, CXCL14, IL34) (Fig. 6C–L). No difference of these cytokines mRNA expression was observed between age-matched controls and relatives with normal kidney function (Supplemental Material, Fig. S7). In addition, the mRNA level of CXCL1, CXCL17, IL1RN and LGALS9 were significantly higher in UREC derived from NPHP1 patients than from patients suffering from other CKD. Of note, CXCL17 mRNA was significantly correlated with eGFR of NPHP1 patients (Supplemental Material, Table S8). CX3CL1, CXCL10 and CXCL16 show the same trend, while not reaching statistical significance (Fig. 6C–L). Although the molecular mechanisms leading to the upregulation of cytokine expression remain unclear, our in vitro data demonstrate that renal tubular cells are the source of cytokine expression in NPH. These data reveal that, in addition to previously described CCL2 upregulation and macrophage infiltration, NPH is characterized by a complex and specific cytokine signature, which is associated with kidney infiltration by neutrophils and T cells.

Discussion

In sharp contrast with the abundant literature regarding the molecular mechanisms of disease progression in ADPKD, insights into the pathophysiology of NPH remain scarce. Despite a growing list of causative gene defects, it is unknown how dysfunction of ciliary NPHP complexes results in the unique renal manifestations of NPH. Although some data point to a defect in renal development, a prominent feature is severe progressive fibrosis, resulting in kidney failure during the second decade of life. The study of the molecular pathogenesis of NPH is hampered by the lack of orthologous mouse models recapitulating the fibrotic disease observed in most NPH patients. Indeed, neither Nphp1^−/− nor Nphp4^−/− mice develop renal fibrosis (25,26) and only few mouse models orthologous to rare forms of the disease phenocopy the human pathology (19,23,27). In addition, since genetic testing has become more widely available, kidney biopsies are less and less performed in NPH, limiting availability of kidney tissues from patients and insights into the disease.

Renal inflammation has emerged as an important mediator of fibrosis in acquired CKD (13). In ADPKD, CCL2-dependent macrophage recruitment promotes disease progression (17,18). Having previously shown that both NPHP1 and LKB1 repress CCL2 expression in vitro and that Lkb1 inactivation in mice results in an NPH-like phenotype preceded by CCL2 upregulation and macrophage recruitment (17), we first sought to determine if CCL2 induction and macrophage recruitment were also hallmarks of human NPH. In line with the data gathered from Lkb1^Δtub mice, we observed significant macrophage infiltration in kidney biopsies from NPH patients and enhanced CCL2 levels in the urine. We then assessed the role of CCL2 in the Lkb1^Δtub phenotype by inactivating Ccl2 specifically in tubular cells. Contrary to its effect in ADPKD models (17,18), Ccl2 invalidation had no impact on the NPH phenotype of Lkb1^Δtub mice and did not reduce macrophage infiltration, nor inflammatory cytokine expression in Lkb1-deficient kidneys. Strikingly, CCL2 inactivation even aggravated kidney size decrease and macrophage infiltration in Lkb1^Δtub mice, suggesting that CCL2 may even exert a limited protective effect on Lkb1-deficient kidneys.

Beyond ADPKD, CCL2 is instrumental to macrophage recruitment and kidney damage in a range of experimental renal diseases including unilateral ureteric obstruction, acute kidney injury, diabetic nephropathy, subtotal nephrectomy (28) or in response to renal infection (29). Thus, the fact that renal inflammation and macrophage recruitment in Lkb1^Δtub mice is independent of tubular CCL2 suggests that the mechanisms driving renal inflammation in NPH are distinct from those implicated in most kidney diseases. Of note, as we deleted CCL2 only in Lkb1-deficient distal tubules, we cannot formerly exclude that CCL2 induction in other cell types may be involved in disease progression. Yet, our results clearly rule out the contribution of CCL2 to immune cell recruitment and disease progression in Lkb1 deficient cells. To get further insight into the nature of renal inflammation in NPH, we analyzed the population of immune cells infiltrating the kidneys in human NPH and Lkb1^Δtub mice. In both case, macrophage infiltration was associated with the recruitment of T cells and neutrophils, by opposition to ADPKD where macrophages are the prominent drivers of inflammation (16–18). To identify the mediators driving early NPH pathology, we analyzed common regulated genes in the transcriptome of Glis2^lacZ/lacZ and Lkb1^Δtub kidneys, two genetically distinct mouse models of NPH. Unbiased pathway enrichment analysis of their common upregulated genes revealed a striking preponderance of inflammatory processes, pinpointing that early inflammation is a prominent feature of experimental NPH.

Indeed, we found a large number of soluble mediators of immune activation in urinary tubular cells derived from NPH patients. A subset of 7 proinflammatory cytokines were specifically upregulated in NPHP1 UREC and NPH mouse models as compared with urinary tubular cells derived from non-ciliopathy CKD patients: CX3CL1, CXCL1, CXCL10, CXCL16 and CXCL17, IL1RN and LGALS9. Thus, these cytokines appear more specific of the early renal inflammation observed in NPH.

CXCL1 induces the recruitment of various immune cell types, especially neutrophils, through its receptor CXCR2. The CXCL1/CXCR2/neutrophil axis plays an important role in the pathology of acute kidney inflammation in different mouse models (30–32). Interestingly, we observed concordant neutrophil infiltration in human NPH kidneys and mice, even at an early stage of the disease. In contrast, neutrophil recruitment does not occur in ADPKD (17). As neutrophils are known to promote renal fibrosis (33), it is plausible that CXCL1-mediated neutrophil recruitment contributes to phenotypic changes in NPH.

Our findings add to the recently identified expression of CXCL17 in the kidney (34); the last identified member of the CXC chemokine family in mammals (35,36). CXCL17 attracts professional antigen presenting cells (37), including macrophages, through its suspected receptor GPR35/CXCR8 (38), even though a second unidentified G protein-coupled receptor may mediate CXCL17-dependent signaling (39). Although very little is known about the role of CXCL17 in the kidney, single-cell RNA sequencing analysis revealed that its expression is enhanced at an early time point in murine tubulointerstitial fibrosis model (34). It is consistent with the increased levels of CXCL17 observed in interstitial pulmonary fibrosis, where it may recruit immune cells that in turn produce proinflammatory cytokines (39). Further research may determine if this axis could participate in the pathogenesis of NPH and its fibrotic features.

LGALS9, a mammalian β-galactoside binding lectin (Gal-9), was first isolated from murine embryonic kidney (40). LGALS9 is known to participate in numerous cellular processes including the induction of apoptosis of different immune cells, particularly cytotoxic T lymphocytes when bound to its surface receptor TIM3/HAVCR2 (41). The precise role of LGALS9 in the kidney has not been described so far; however, a recent study showed that anti-TIM3 antibody ameliorates kidney injury and decreased macrophage infiltration in ischemic mouse model (43). This is consistent with a concomitant study that showed increased levels of soluble TIM3 and soluble Gal-9 in the blood of patients with kidney transplantation-related renal dysfunction (43). In addition, upregulation of serum Gal-9 is closely related to glomerular filtration rate decrease in patients with type 2 diabetes (44). Thus, circulating Gal-9 and TIM3 may be useful biomarkers to monitor GFR decline in NPH. Further studies are needed to assess their function and interplay in the disease. Besides, Gal-9-dependent signaling may participate in switching macrophages from pro-inflammatory M1 phenotype to its anti-inflammatory M2 counterpart, raising the possibility that Gal-9/TIM3 is involved in the polarization of macrophages in NPH (45,46).

This work identified a specific network of commonly regulated cytokines that represent plausible mediators of immune cell recruitment to NPH kidneys beyond CCL2. We identified inflammation as the predominant signature in two independent models of NPH and we found that this holds true in human patients. As most of the NPH patients studied bear NPHP1 mutations, we cannot exclude that these findings are specific to NPHP1 patients. Using a non-orthologous mouse model of NPH on the basis of genetic disruption of Lkb1 may be a limitation of the study. Yet, we confirmed the inflammatory signature identified in Lkb1^ΔTub mice in one of the rare murine orthologous model of NPH leading to renal fibrosis (Glis2^lacZ/lacZ mice) as well as in UREC derived from NPHP1 patients. Overall, the similar inflammatory profiles in NPH-like mouse models and in tubular cells derived from NPH patients strengthen our view that Lkb1^ΔTub mice represent a faithful model to study the pathophysiology of NPH. The parallelism between the inflammatory pathways activated in the early diseased kidneys from Lkb1^ΔTub or Glis2^lacZ/lacZ mice in one hand, and those enriched in the UREC from NPH patients in the other hand, along with the protective effect of TLR2/MYD88 inhibition in Glis2^lacZ/lacZ mice, suggest causality. However, we cannot exclude that chemokine expression and immune cell recruitment are innocent bystanders of an undefined process leading to kidney damage in NPH. Certain immune cells may also be protective, such as T lymphocytes in the context of ADPKD (47).

This work establishes renal inflammation as a prominent feature of human NPH and identifies specific mediators of this process that are common to NPH mouse models and patients. Addressing the precise function of these mediators in NPH in the future will help to characterize the underlying processes responsible for renal deterioration in this orphan disease and to evaluate inflammation as a potential therapeutic target.

Materials and Methods

Human kidney tissue specimens

Renal kidney tissues from patients suffering from juvenile NPH were collected from 1973 to 1998. Ineligible kidneys donated for transplantation or kidney from patient suffering from renal disease other than NPH with minimal histological lesions and collected over the same time period were used as controls. Renal biopsies from patients suffering from diabetic nephropathy or hypertensive nephrosclerosis were used to control for non-specific, CKD-related inflammation. IFTA score was evaluated on renal biopsy by a trained pathologist in a blinded fashion. NPH and CKD patient groups were matched according to their IFTA score. Patient data are listed in Supplemental Material, Table S1.

Isolation of UREC

Urine samples were collected from NPHP1 patients, healthy relatives and unrelated controls as well as CKD patients recruited at Necker Hospital (Paris, France) after written informed consent of the donor. Inclusion criteria for affected patients were to suffer from NPH with known genetic diagnosis. The relatives were the healthy relatives (father/mother/sister) of an included patient. Healthy controls were free from any CKD or with normal renal function. CKD patients were suffering from CKD other than renal ciliopathy. Patient data are listed in Supplemental Material, Table S1.

Urine-derived renal tubular epithelial cells (primary UREC) were isolated and cultured as previously described (49) with some modifications. Briefly, after centrifugation and washing steps, urine-derived cells were initially cultured for 4 days at 37°C in primary medium containing Dulbecco’s Modified Eagle Medium: Nutrient Mixture F-12 supplemented with 10% fetal bovine serum (16000-036, Gibco), 10% Penicillin–Streptomycin (15140-122, Gibco), 10% amphotericin B (15 290 026, Gibco) and 1× REGM™ SingleQuots™ kit (CC-4127, Lonza) to enhance cell survival and adherence. At day 4, primary medium was replaced by a growth medium containing REBM™ (Basal Medium, CC-3191, Lonza) supplemented with 2% fetal bovine serum, 10% penicillin–streptomycin, 10% amphotericin B, 1× REGM™ SingleQuots™ kit and 10 ng/ml rhEGF (R&D system) and was then changed every 2 days. At 80% confluence (7–30 days), 2 × 10⁴ cells were seeded for 7–24 days until confluence on 12 well plates (353 043, Dutscher) for RNA analysis.

Mice

Mice were housed in a specific pathogen-free facility, fed ad libitum and housed at constant ambient temperature in a 12-h day/night cycle. Breeding and genotyping were done according to standard procedures.

Lkb1^ΔTub mice were previously described (17). Ccl2-RFP^flox/flox (B6.Cg-Ccl2tm1.1Pame/J, stock number: 016849, C57BL/6 J genetic background) were purchased from The Jackson Laboratories and were backcrossed for two generations with Lkb1^ΔTub mice. The progeny was then intercrossed to generate mice with tubule-specific Lkb1 knockout (further referred to as Lkb1^ΔTub), Ccl2 knockout (further referred to as Ccl2^ΔTub) and Lkb1; Ccl2 double knockout (further referred to as Lkb1^ΔTub; Ccl2^ΔTub) on a common mixed C57BL/6 J; C57BL/6 N genetic background. Littermates lacking KspCre transgene were used as controls. Experiments were conducted on both females and males.

Pkd1^flox/flox mice (B6.129S4-Pkd1tm2Ggg/J, stock number: 010671, 129.B6 mixed genetic background) were purchased from The Jackson Laboratories and were crossed to Pax8rtTA (Tg(Pax8-rtTA2S^*M2)1Koes) and TetOCre (Tg(tetO-cre)1Jaw) mice to generate an inducible tubule-specific Pkd1 knockout (further referred to as iPkd1^ΔTub) as previously described (17). To induce recombination, mice received doxycycline (Abcam, ab141091) in drinking water (2 mg/ml with 5% sucrose, protected from light) from postnatal day 28 (P28) to P42. Littermates lacking either TetOCre or Pax8rtTA were used as controls. Experiments were conducted on males.

Quantitative PCR

Total RNAs were obtained from human UREC or mouse kidneys using RNeasy Mini Kit (Qiagen) and reverse transcribed using SuperScript II Reverse Transcriptase (Life Technologies) or High Capacity cDNA Reverse Transcription Kit (Applied Biosystems) according to the manufacturer’s protocol. Quantitative PCR were performed with iTaq™ Universal SYBR® Green Supermix (Bio-Rad) on a CFX384 C1000 Touch (Bio-Rad). Hprt, Ppia, Rpl13, Sdha and Tbp were used as normalization controls (50). Each biological replicate was measured in technical duplicates. The primers used for qRT-PCR are listed in Supplemental Material, Table S2.

C‌CL2 ELISA

For the quantification of urinary CCL2 excretion, urine specimens were collected and centrifuged at 1500g for 10 min at 4°C within 4 h of collection. The supernatants were collected and stored at −80°C. Frozen aliquots of urine supernatants were thawed at room temperature immediately before the ELISA. The samples were used with a 2-fold dilution and were tested in duplicates. CCL2 levels in urine specimens were quantified using Human CCL2/MCP-1 Quantikine® ELISA Kit (R&D systems, DCP00) according to the manufacturer’s instructions. The plate was read using a Multiskan Sky plate reader. The optical densities were derived from 4-parameter logistic regression of the standard curve. Measurement of creatinine in urine was performed in the same samples using IDMS-standardized enzymatic method on C16000 Architect analyzer (Abbott Diagnostic). The results were normalized to the urinary creatinine level. Indeed, normalization by the urinary creatinine levels avoids the pitfall of concentration or dilution of urine.

Morphological analysis

Human kidney biopsies were fixed in alcohol formalin and acetic acid and paraffin-embedded, 4-μm sections were stained with periodic acid–Schiff (PAS). PAS-stained full size images were recorded using a whole-slide scanner Nanozoomer 560 (Hamamatsu) coupled to NDPview software (Hamamatsu).

Mouse kidneys were fixed in 4% paraformaldehyde, embedded in paraffin and 4-μm sections were stained with PAS or Picrosirius red. Stained full size images were recorded using a whole slide scanner Nanozoomer 2.0 (Hamamatsu) equipped with a 20×/0.75 NA objective coupled to NDPview software (Hamamatsu). Histology score was evaluated by an independent observer in a blinded fashion assessing the overall lesions of the whole-kidney section stained with PAS. Six scores were defined ranging from score 1 normal kidney architecture to score 6 associating tubular atrophy, tubular basement thickening and interstitial cell infiltration.

Urine and plasma analyses

8-h urine samples were obtained from mice housed in individual boxes without access to water and food. Body weight and urine excretion were measured. Urine osmolality was measured with a freezing point depression osmometer (Micro-Osmometer from Knauer or OSMOMAT 3000basic from Gonotec). Retro-orbital blood was collected from anaesthetized mice. Plasma blood urea nitrogen (BUN) was measured using urea kit (LT-UR; Labor&Technik, Eberhard Lehmann GmbH) according to the manufacturer’s instructions.

Immunohistochemistry

For human kidney sections

An automated IHC stainer BOND-III (Leica Biosystems) was used. Briefly, 4-μm sections of paraffin-embedded human kidney biopsies were submitted to the appropriate antigen retrieval. Then, sections were incubated with CD68 (Dako, M0814, 1:3000), CD15 (Beckam Coulter, NIM0165, 1:200) or CD3 (Dako, A0452, 1:200) antibodies. Peroxide blocking, post primary, DAB chromogen and hematoxylin counterstaining was performed automatically using Bond polymer refine detection kit (Leica Biosystems, DS9800). The degree of interstitial cell infiltration was determined using immunostaining targeting macrophages (CD68), neutrophils (CD15) and T lymphocytes (CD3). Slides were scanned with a whole-slide scanner Nanozoomer 560 (Hamamatsu). Randomly selected microscopic fields (×200) representative of the entire cortical surface were scored. The degree of cell infiltrate was quantified using ImageJ software. For CD3 and CD68, the area of DAB staining was measured after color deconvolution and intensity thresholding of the images and visualized as the ratio of DAB surface to cortical surface on each microscopic field. For CD15, the number of CD15-positive interstitial cells was quantified manually in a blinded fashion and was expressed as a ratio per mm².

For mouse kidney sections

Macrophage staining: 4-μm sections of paraffin-embedded mouse kidneys were incubated for 20 min at 95°C in citrate buffer (Zytomed, ZUCD28) followed by avidin/biotin blocking (Vector, SP-2001). Sections were incubated with F4/80 antibody (Clone Cl:A3-1, Bio-Rad, MCA497R, 1:100) followed by biotinylated antibody (Vector, BA-4001, 1:200), horseradish peroxidase (HRP)-labeled streptavidin (Southern Biotech, 7100-05, 1:2000) and 3–3′-diamino-benzidine-tetrahydrochloride (DAB, Dako, K3468) revelation.

Neutrophil staining: 4-μm sections of paraffin-embedded mouse kidneys were submitted to antigen retrieval for 20 min at 95°C in citrate buffer followed by incubation with Ly-6B.2 antibody (Abcam, ab53457, 1:100). Sections were incubated with HRP-labeled secondary antibody (Vector, PI-9400, 1:200) and DAB revelation.

T cell staining: 4-μm sections of paraffin-embedded mouse kidneys were incubated for 20 min at 95°C in Tris-EDTA pH 9 buffer followed by avidin/biotin blocking. Sections were incubated with CD3 antibody (Abcam, ab16669, 1:100) followed by biotinylated antibody (GE Healthcare, RPN1004V, 1:200), HRP-labeled streptavidin (Southern Biotech, 7100-05, 1:2000) and DAB revelation.

Full size images were recorded using a whole slide scanner Nanozoomer 2.0 coupled to NDPview software. Stained area was measured with ImageJ software from full size kidney images and visualized as the ratio of stained DAB surface to total kidney section area. For neutrophil quantification in Figure 5, because neutrophils are rare and focal in 5-week old mice, we counted manually the number of foci in whole kidney section. The number of foci per kidney section was scaled to the surface of the section. Foci were defined as 4 or more neutrophils surrounding a tubule. For all quantifications, glomerular and non-specific intra-tubular staining were removed from the analysis.

Comparative microarray data analysis

Processing of data was carried out using R v3.6.0, RStudio and R package dplyr v1.0.1. Renal transcriptomic datasets from 5-weeks old Lkb1^ΔTub and 4 weeks old Glis2^lacZ/lacZ mouse models (17,19) were compared. Although transcriptomic dataset from Lkb1^ΔTub mice was established by our team (GSE86011) (17), we downloaded from the GEO database (https://www.ncbi.nlm.nih.gov/geo/) the expression matrix of mRNAs expressions in the kidneys of Glis2^lacZ/lacZ mice under the accession number GSE6113 (19). Probe IDs in the expression matrix were matched with the corresponding gene IDs in the lookup table (GPL2897) to identify the expression of each mRNA. To assess the common regulated genes between the two mouse models, Lkb1^ΔTub dataset was filtered out according to FDR (Benjamini–Hochberg procedure) < 0.05 to obtain a list of 1991 differentially expressed genes (DEGs). Then, Glis2^lacZ/lacZ dataset (23 957 genes) was matched with this DEGs list, and the genes whose expression varies in the same way were considered as the common regulated genes (Supplemental Material, Table S3).

GSEA, networking and visualization

Subsequently, GSEA approach was applied to the 1262 common regulated genes identified using GSEA software v4.0.0 (Broad Institute) set for 000 gene set permutations. The enrichment involved the ‘biological process’ classification of the Gene Ontology downloaded from MSigDB v7.1 (C5-BP). Up- and downregulated pathways are listed in Supplemental Material, Tables S4 and 5. Pathways with FDR < 0.05 were considered significant. R package ggplot2 was used to present the more significant up- and downregulated pathways. Cytoscape software v3.0 and Cytoscape applications EnrichmentMap v3.3.0 (edge cutoff threshold set to 0.5) and WordCloud v3.1.3 were used for pathway clustering and pathway network visualization (51). Pathway clusters have been hand-annotated.

Identification of common upregulated cytokines

To identify common upregulated cytokines, the 823 common upregulated genes were matched with lists of genes obtained from the UniProt database (https://www.uniprot.org/) using the following keywords: ‘cytokine’, ‘secreted’ and ‘immune/inflammation’. Genes that were linked to the three keywords are listed in Supplemental Material, Table S6.

Heatmaps of the expression of the identified cytokines from 5-week old Lkb1^ΔTub and 10-week old Pkd2^ΔTub were compared. The Pkd2^ΔTub log-normalized expression matrix was downloaded from GEO under the accession number GSE149739 (24). Differential expression testing between groups was performed using limma v3.42.2. Heatmaps were generated using the R package pheatmap v1.0.12 (note: heatmaps show Z-scores computed from the expression matrix for each gene, thus presenting expression levels with relatives equal means for each gene). Expression levels of the identified cytokines are listed in Supplemental Material, Table S7. Finally, the heatmap in Figures 4C and5C displays Z-scores computed on the expression levels of the identified cytokines, measured by qPCR. The corresponding qPCR results are listed in Supplemental Figures S4 and 5.

Statistical analysis

Data were expressed as means. Differences between groups were evaluated using Mann–Whitney test when only two groups were compared. When testing more comparisons, Kruskal–Wallis test was used when the groups do not meet normal distribution otherwise one-way analysis of variance (ANOVA) followed when significant (P < 0.05) by the Tukey–Kramer test was used. The statistical analysis was performed using GraphPad Prism V8 software. All image analysis and mouse phenotypic analysis were performed in a blinded fashion.

Study approval

The human kidney tissue specimens belonging to the Imagine Biocollection are declared to the French Minister of Research under the number DC-2020-3994 and approved by the French Ethics committee for research at Assistance Publique-Hôpitaux de Paris (CERAPHP) under the IRB registration number #00011928. The urine study was approved by the French National Committee for the Protection of Persons under the ID-RCB number 2016-A00541-50 and is kept in full accordance with the principles of the Declaration of Helsinki and Good Clinical Practice guidelines.

All animal experiments were conducted according to the guidelines of the National Institutes of Health Guide for the Care and Use of Laboratory Animals, as well as the German and French laws for animal welfare, and were approved by regional authorities (Regierungspräsidium Freiburg G-16/27 and Ministère de l’Enseignement, de la Recherche et de l’Innovation APAFIS#2020090715389782).

Acknowledgements

The authors thank Dr Alexandre Benmerah for critical appraisal of the manuscript.

We thank the technicians from the mouse histology and breeding facilities (S.F.R Necker INSERM US24, Paris, France) and the department of Pathology (Necker Hospital, Paris, France) for technical assistance. We are grateful to the patients and their families for their participation. We thank Pauline Krug, Olivia Boyer, Nathalie Biebuyck, Saoussen Krid, Marina Charbit, Romain Berthaud and Guillaume Lezmi (Pediatric Nephrology, Necker Hospital, AP-HP, Paris, France), Aurélie Hummel (Adult Nephrology, Necker Hospital, AP-HP, Paris, France), Amélie Lezmi Ryckewaert and Sophie Taque (Pediatric Hematology and Oncology, Hôpital Universitaire, Rennes, France), Odile Boespflug-Tanguy (Centre de Compétence des Leucodystrophies et Leucoencéphalopathies de Cause Rare, Pôle Femme et Enfant, Hôpital Estaing, Centre Hospitalier Universitaire de Clermont-Ferrand, Clermont-Ferrand, France), Jérôme Harambat and Brigitte Llanas (Department of Pediatrics, Bordeaux University Hospital, Bordeaux, France), Bruno Ranchin (Pediatric Nephrology, Centre Hospitalier Universitaire de Lyon, Bron, France), Elodie Merieau (Centre Hospitalier Régional Universitaire de Tours, Tours, France), Marc Fila (Centre Hospitalier Universitaire de Montpellier, Montpellier, France) and Tory Kalman (Semmelweis University, Budapest, Hungary) who helped the follow-up of patients. We thank Corinne Antignac and Laurence Heidet (Department of Genetics, Necker Hospital, AP-HP, Paris, France) for genetic diagnostic. We thank the Department of Clinical Research at Imagine Institute with the sponsorship team that facilitates and structures the set-up of the clinical research projects and the investigation team that prepares and ensures the follow-up of the clinical trials.

Conflict of Interest statement. None declared.

Funding

Public grant `RHU-C'ILL-LICO' overseen by the Agence Nationale de la Recherche (grant number: ANR-17-RHUS-0002 to M.Q., S.S. and A.V.); European Molecular Biology Organization (grant number: ALTF 927-2013 to F.B.); Deutsche Forschungsgemeinschaft (grant numbers: KU1504/7-1, KU1504/8-1 to E.W.K.); State funding from the Agence Nationale de la Recherche (grant numbers: ANR-19-CE14-0016 to F.B. and ANR-10-IAHU-01 to S.S.).