Selective pharmacological inhibition of DDR1 prevents experimentally-induced glomerulonephritis in prevention and therapeutic regime

Selective pharmacological inhibition of DDR1 prevents experimentally-induced glomerulonephritis... Background: Discoidin domain receptor 1 (DDR1) is a collagen-activated receptor tyrosine kinase extensively implicated in diseases such as cancer, atherosclerosis and fibrosis. Multiple preclinical studies, performed using either a gene deletion or a gene silencing approaches, have shown this receptor being a major driver target of fibrosis and glomerulosclerosis. Methods: The present study investigated the role and relevance of DDR1 in human crescentic glomerulonephritis (GN). Detailed DDR1 expression was first characterized in detail in human GN biopsies using a novel selective anti- DDR1 antibody using immunohistochemistry. Subsequently the protective role of DDR1 was investigated using a highly selective, novel, small molecule inhibitor in a nephrotoxic serum (NTS) GN model in a prophylactic regime and in the NEP25 GN mouse model using a therapeutic intervention regime. Results: DDR1 expression was shown to be mainly limited to renal epithelium. In humans, DDR1 is highly induced in injured podocytes, in bridging cells expressing both parietal epithelial cell (PEC) and podocyte markers and in a subset of PECs forming the cellular crescents in human GN. Pharmacological inhibition of DDR1 in NTS improved both renal function and histological parameters. These results, obtained using a prophylactic regime, were confirmed in the NEP25 GN mouse model using a therapeutic intervention regime. Gene expression analysis of NTS showed that pharmacological blockade of DDR1 specifically reverted fibrotic and inflammatory gene networks and modulated expression of the glomerular cell gene signature, further validating DDR1 as a major mediator of cell fate in podocytes and PECs. Conclusions: Together, these results suggest that DDR1 inhibition might be an attractive and promising pharmaco- logical intervention for the treatment of GN, predominantly by targeting the renal epithelium. Keywords: Fibrosis, DDR1 inhibition, Glomerulosclerosis, CKD *Correspondence: marco.prunotto@roche.com Solange Moll and Yukari Yasui contributed equally to this work Present Address: Office of Innovation, Immunology, Infectious Diseases & Ophthalmology (I2O), Roche and Genentech Late Stage Development, 124 Grenzacherstrasse, 4070 Basel, Switzerland Full list of author information is available at the end of the article © The Author(s) 2018. This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creat iveco mmons .org/licen ses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creat iveco mmons .org/ publi cdoma in/zero/1.0/) applies to the data made available in this article, unless otherwise stated. Moll et al. J Transl Med (2018) 16:148 Page 2 of 20 the subtle podocyte staining detected with ISH, probably Background due to differential detection thresholds, showed a charac - Discoidin domain receptor 1 (DDR1) is a collagen-acti- teristic membranous localization of DDR1. DDR1 immu- vated receptor tyrosine kinase (RTK) [1] extensively nostaining was observed in the tubules of the cortex and implicated in diseases such as cancer [2], atherosclerosis of both outer and inner medulla, most likely in the distal [3] and fibrosis [3–5]. In humans, single nucleotide poly - nephron (Fig.  1b). Double immunostaining with specific morphisms of DDR1 are associated with susceptibility for tubular markers (Megalin, Calbindin, Tamm–Horsfall and disease progression of childhood IgA nephropathy Protein/Uromodulin and Aquaporin 2) on serial sections [6]. In multiple preclinical studies DDR1 has been shown confirmed DDR1 protein localization in the distal parts of to play a major role in the pathogenesis of fibrosis and the tubule (distal convoluted tubule, thick ascending limb glomerulosclerosis [7–13]. The evidence for a protective of the Henle’s loop/pars recta distal tubule, connecting role of DDR1 in glomerulosclerosis has been supported tubule and collecting duct) (Fig.  1c). No DDR1 staining so far by gene knockout (KO) experiments or the use of was detected in the interstitial space or in vessels, using antisense oligonucleotides (ASO), with both technolo- both ISH and IHC techniques. Taken together, these data gies relying on reduction of total DDR1 protein [1–5]. demonstrate that DDR1 expression in the normal human Although these techniques can demonstrate mechanism, kidney is exclusively limited to renal epithelial cells, PECs they have very limited translational potential. Knock-out and podocytes in the glomerulus and tubular cells in the mouse models mimic prophylactic regimens, which are distal part of the nephron. not applicable to patients suffering from a substantial loss of kidney function. ASOs, though tested in a therapeu- tic intervention regimen by one of the co-authors [10], DDR1 protein is abnormally induced during crescentic are predominantly cleared by the liver and kidneys, and glomerulonephritis in humans and is detected in cellular thus represent a non-preferred clinical scenario in renal- crescents impaired patients. Twenty-nine biopsies with cellular and/or fibrous cres - The present study includes results originally generated cents were obtained from patients with Goodpasture’s as part of a pharmaceutical program, aimed at the crea- syndrome (5 cases), ANCA-associated GN (7 cases), tion of a safe and effective DDR1 inhibitor to be used in IgA GN (12 cases) and lupus GN (5 cases: class III and patients affected by glomerulonephritis (GN) [14]. Firstly, IV according to the ISN/RPS classification) (Table  1: 14 we show in depth characterisation of DDR1 expression in males and 15 females, mean age 56 ± 21  years, mean normal human kidney and in renal biopsies from patients serum creatinine 199 ± 175  μmol/l, mean blood urea with crescentic GN, using a newly-developed highly spe- nitrogen (BUN) 11 ± 8 mg/dl, mean proteinuria 3 ± 3 g/l). cific anti-DDR1 antibody. These translational data were Twenty-five patients had biopsies performed for diag - essential to further enhance confidence in the initiation nosis of acute renal insufficiency and 4 for follow-up of and progression of a medicinal chemistry effort to gen - previously treated crescentic GN (patients #4, 25, 28 and erate an exquisitely selective and potent DDR1 inhibitor 29). Glomerulus number per biopsy varied from 7 to 54 (DDR1i). The characterisation of DDR1i in two differ - (mean 24 ± 15), globally sclerosed glomeruli from 0 to ent mouse models of GN [14], in both prophylactic and 23 (mean 2 ± 6) and crescents, either cellular or fibrous, therapeutic regimens, is presented here. Gene expres- from 1 to 30 (mean 9.7 ± 8.7). DDR1 immunostaining sion profiles of selective DDR1i-treated animals were also was evaluated in each crescentic glomerulus as posi- profiled in order to gain further knowledge regarding the tive or negative. DDR1 protein was detected in most of pathways and networks selectively modulated by drug the cellular crescents of all four types of crescentic GN targeting. The data suggest that DDR1 is an important (Table  1, Fig.  2A). DDR1 staining intensity was variable player in human GN and that its pharmacological inhibi- between crescents within the same biopsy and expression tion is translatable into a valid therapeutic intervention limited to a subset of cells forming the cellular crescent. tested in preclinical GN models. In addition to positive cellular crescents, DDR1 immu- noreactivity was detected in injured podocytes (Fig. 2A). DDR1 podocytes were observed in most of the glomer- Results uli, with or without crescent lesion. Furthermore, some of DDR1 is exclusively expressed in renal epithelium these podocytes appeared to adhere to both the glomeru- under physiological conditions lar basement membrane and the parietal basement mem- DDR1 mRNA and protein expression is restricted to brane, forming podocyte bridges between the glomerular the glomerular parietal epithelial cells (PECs) of the tuft and the Bowman’s capsule. All fibrous crescents were Bowman’s capsule and to podocytes and some tubules DDR1 negative. It should also be noted that DDR1 stain- (Fig.  1a). Immunohistochemistry (IHC) failed to reveal ing was increased in distal tubules and detected in some Moll et al. J Transl Med (2018) 16:148 Page 3 of 20 Fig. 1 Localisation of DDR1 in human control kidney. a DDR1 in situ hybridization (ISH) and immunohistochemistry (IHC) analyses in normal human kidney. (Scale bar = 100 μm). ISH higher magnification rectangle: DDR1 positive podocyte (black arrowheads), DDR1 positive podocytes (green arrowheads). b DDR1 immunostaining in the nomal human kidney with representative images of the cortex, outer medulla (outer and inner stripes) and inner medulla. Note the DDR1 membranous staining of the tubules. (Magnification ×125, scale bar 50 μm). c Serial sections of normal human kidney were double immunostained with DDR1 and specific tubular antibodies against Megalin, Calbindin, Tamm–Horsfall Protein ( THP or Uromodulin) or Aquaporin 2 (AQP2). c1 Representative micrographs of the cortex show DDR1 protein localization in the distal convoluted tubules (Calbindin+, Aquaporin 2−, Megalin−). Proximal tubules (Megalin+) are DDR1 negatively stained. Boxed areas are enlarged in the right side of the figure. c2 Representative micrographs of the medullar rays of the cortex show DDR1 protein localization in the connecting tubules (1: Calbindin+, Aquaporin 2+, THP−), distal convoluted tubules (2: Calbindin+, Aquaporin 2−, THP−), thick ascending limbs of the Henle’s loop/pars recta of the distal tubules (3: THP+, Calbindin−, Aquaporin 2−) and collecting ducts (4: Aquaporin 2+, Calbindin−, THP−). Boxed areas are enlarged in the right side of the figure. Magnification ×100, scale bar = 100 μm; Boxed areas Magnification ×250, scale bar = 50 μm cases, particularly SLE cases, in both distal and proximal glomerular capillary lumen and only 2–3 layers of cells tubules as depicted in Fig. 2A. within the Bowman’s space, DDR1 staining was detected To define more precisely the profile of DDR1 expres - in numerous cells of the glomerulus. These cells were sion during the sequence of morphological changes most likely injured podocytes, indicated by similar nestin observed in crescent formation, we performed IHC on immunostaining and by abnormal morphologic aspects lupus nephritis class IV-G (A/C) biopsies. Lupus nephri- (Fig.  2B, C line a). It should be noted that the DDR1 / tis biopsy samples can display glomerular lesions that are Nestin cells, localized anterior to the emerging cellullar morphologically heterogeneous, consistent with vari- crescent, were also slightly positive for the PEC markers ous stages of crescentic lesions. Furthermore, crescent CK 8–18 and CK 19 (Fig.  2C line a). At ther very initial formation results from the serial participation of several stage of the cellular crescent formation, characterized by different cell types, including macrophages, glomerular gaps in the glomerular capillary wall and plasma proteins parietal epithelial cells, glomerular visceral epithelial cells and cells within Bowman’s space, DDR1 staining was also (podocytes), renal progenitor cells and interstitial fibro - detected in injured podocytes, but with a weaker stain- blasts [15]. With this in mind, serial sections of lupus ing intensity (Fig.  2B, C line b). It should be noted that nephritis biopsies were immunostained with DDR1 the strongest DDR1 signal was observed in podocytes and four different cellular markers: the monocyte/mac - localized in front of the gap in the glomerular capillary rophage marker CD68, the podocyte marker nestin, wall, near the Bowman’s capsule or near the growing cel- + + and the PEC markers cytokeratin (CK) 8–18 and CK 19 lular crescent. Some of these DDR1 /Nestin cells, also (Fig.  2B, C). In the very initial stage of cellular crescent positive for CK 8–18, seemed to adhere to both the glo- formation, characterized by fibrin deposition within the merular basement membrane and the PECs, forming Moll et al. J Transl Med (2018) 16:148 Page 4 of 20 Table 1 Clinical, biological and  histo-pathological data from  the  29 patients with  crescentic glomerulonephritis and  DDR1 immunostaining evaluation of the crescents as positive or negative Patient nb Clinical data Histological data Diagnostic Age/gender SCreat (µmol/l) eGFR ml/ BUN mg/dl Proteinuria (g/l) Hematuria Glomeruli Cellular crescents Fibrous crescents TIF % min/1.73 m Total Globally DDR1+ DDR1− DDR1+ DDR1− sclerosed 1 Goodpasture 81/F 478 8 18.9 15 pos 8 0 5 0 0 0 25 2 85/F 776 4 32.7 3.68 pos 37 23 3 0 0 2 75 3 22/M 113 > 60 5.3 1.54 pos 9 0 3 0 0 0 0 4 21/M 196 38 10 0.64 pos 14 0 7 0 0 2 10 5 72/M 117 98 10 9 pos 42 0 3 0 0 1 0 6 ANCA 82/F 678 15 26.2 2.5 pos 54 20 0 27 0 3 0 7 74/M 312 16 25.3 0.58 pos 32 0 27 1 0 2 35 8 78/F 84 58 4.3 1.39 pos 9 2 0 1 0 0 0 9 66/F 165 29 ND 0.4 neg 27 0 11 4 0 2 55 10 66/M 311 19 ND 0.6 pos 17 0 7 0 0 0 20 11 70/M 202 28 12 2.09 pos 12 1 1 0 0 3 0 12 68/M 158 40 ND 1.8 pos 24 0 10 0 0 3 30 13 IgA 45/F 88 67 ND 7 neg 28 1 19 1 0 4 40 14 84/F 310 12 11 4.93 pos 7 0 2 0 0 1 10 15 56/M 94 > 60 3.3 0.3 pos 15 0 8 0 0 0 20 16 39/F 135 43 11 1.78 pos 21 8 5 0 0 0 50 17 43/F 56 > 60 4.8 0.6 pos 17 3 3 0 0 0 30 18 74/M 180 34 15 0.7 pos 14 1 0 0 0 1 50 19 19/F 62 > 60 4.5 3 pos 44 0 12 0 0 0 15 20 19/F 57 > 60 3 1.0 pos 34 0 2 0 0 0 10 21 23/M 91 > 60 5.9 0.5 pos 52 0 7 0 0 0 0 22 64/M 99 > 60 5.1 0.4 pos 8 0 4 0 0 0 45 23 72/F 250 17 ND 2 pos 12 0 4 0 0 1 30 24 64/F 108 47 12 1.88 pos 9 0 0 1 0 1 0 25 Lupus 37/F 159 36 8.5 8.0 pos 27 0 16 3 0 3 0 26 40/F 107 56 10.2 4.05 pos 19 0 6 0 0 4 30 27 44/M 180 38 ND 4.6 pos 14 0 3 0 0 4 40 28 51/M 97 > 60 7.6 5.0 pos 47 1 9 0 0 2 10 29 52/M 102 > 60 6.88 5.0 pos 45 2 20 0 0 6 20 Moll et al. J Transl Med (2018) 16:148 Page 5 of 20 (See figure on next page.) Fig. 2 DDR1 is highly induced in crescentic glomerulonephritis. A DDR1 expression in 4 different types of human crescentic glomerulonephritis. DDR1 immunostaining in human kidney biopsies from patients with 4 different types of crescentic glomerulonephritis (GN): Goodpasture’s syndrome (patient #1), ANCA-associated GN (patient #12), IgA GN (patient #19) and lupus GN class IV-G (A/C) (patient #29). DDR1 is expressed in the cellular crescents of all 4 types of crescentic GN with variability in the staining intensity and expression limited within crescents to a subset of cells. Injured podocytes and bridging podocytes (arrows) strongly express DDR1. Fibrous crescent (*) are DDR1 negative. Magnification ×200 for SLE, ×250 for ANCA, ×320 for Goopasture and IgA, Scale bar = 20 μm. B–C Detailed analysis of DDR1 expresion during crescent formation. Serial sections of human kidney biopsies from patients with lupus GN class IV-G (A/C) (patients #25 and #29) were immunostained with specific antibodies against CD68/PAS (Periodic Acid Schiff ), DDR1, nestin, cytokeratin 8/18 or cytokeratin 19 proteins. B Representative micrographs show expression and localization of each protein in the glomeruli. Boxed areas represent crescentic lesion and are enlarged in C. Lines a–d illustrate 4 different morphological stages of the crescent formation; a: early stage with fibrin within the glomerular capillary lumen and presence of 2–3 layers of cells in Bowman’s space; b: early stage with gap in the glomerular capillary wall, plasma proteins in Bowman’s space and cellular crescent formation; c: active hypercellular circumferential crescent compressing the glomerular tuft; d: advanced stage with fibrocellular crescent, capsular rupture and periglomerular inflammation. Line a = patient #25; lines b–d = patient #29. Fine arrows = injured podocytes; large arrows = bridging cells; * = fibrin and plasma proteins within the glomerular capillary lumen or in Bowman’s space. Magnification A line a: ×320, b and c: ×250, d: ×200; B line a: ×640, b and c: ×500 d: ×400; Scale bar = 20 μm podocyte bridges between the glomerular tuft and Bow- similarly to human GN, DDR1 appears to be predomi- man’s capsule (Fig.  2C line b). In a well formed cellular nantly expressed in epithelial cells in the NTS mouse crescent characterized by multiple layers of cells within model. the Bowman’s space, DDR1 expression was readily To address the functional significance of DDR1 in detected within a subset of cells of the crescent, mostly experimental crescentic GN, mice were pre-treated with located in the periphery of the glomerular tuft (Fig.  2B, two doses of a very selective DDR1 inhibitor (Roche-Chu- C, line c). These DDR1 cells positively immunostained gai DDR1i), followed by NTS administration. To unravel with both PEC markers CK 8/18 and CK 19, but were the specificity of DDR1 activation blockade, a second nestin negative. In a more advanced stage, characterized group was treated with the kinase inhibitor imatinib, by fibro-cellular crescent, Bowman’s capsule rupture and which was previously shown to elicit a marked renopro- periglomerular inflammation, DDR1 immunostaining tective effect in the NTS-induced model in Wistar-Kyoto was detected in the cellular compartment of the cres- rats [16]. All functional parameters were significantly cent but not in the fibrous area (Fig.  2B, C, line d). A improved with the high dose of DDR1 inhibitor. The low similar staining was observed with CK 8/18 and CK 19. dose was less efficient, whereas Imatinib failed to show a Finally, it should be noted that in severe necrotic lesions significant protection (Fig.  3d–g). Histological analyses of as observed in patient #6 and #8 with ANCA-associated kidney sections followed by semiquantitative histopatho- GN (Table 1), no DDR1 staining could be detected (Addi- logic evaluation with glomerular and tubulo-interstitial tional file  1: Figure S3). This result is not surprising given (TI) scores revealed that functional kidney protection in the degree of structural damage in these glomeruli. the high dose DDR1i group was paralleled by tissue pres- Taken together, our data suggest that DDR1 is de novo ervation of both glomerular and tubulo-interstitial scores expressed during crescent formation, predominantly (p = 0.02 and p = 0.01 respectively, Fig. 3h–i). In contrast, in injured podocytes, in bridging cells expressing both none of these findings were observed in the Imatinib- podocyte and PEC markers, and in a subset of PECs treated group (glomerular and tubulo-interstitial scores: forming the cellular crescent. p = 0.92 and p = 0.93 respectively). Concordant with glomerular histology, IHC staining with the PEC mark- DDR1 expression increases in experimentally‑induced ers desmin and CD44 revealed marked reductions in the crescentic glomerulonephritis and its pharmacological DDR1i HD group (Fig.  3h). Likewise, collagen type IV prophylactic inhibition confers morphological staining showed a marked reduction of immunoreactivity and functional protection in the interstitial areas in the DDR1i HD group, but not First, we confirmed that DDR1 expression is induced fol - in the DDR1i LD and Imatinib groups, which were both lowing NTS administration (Fig. 3a–c). Consistently, ISH comparable to the vehicle group. A similar trend was showed strong DDR1 upregulation in crescents, PECs, observed in the case of collagen type I (data not shown). and in the tubular structures, most of them display- Morphometric analysis of collagen type IV staining ing lesions characterized by flattening of the epithelium showed a marked and statistically significant reduction and dilation of the lumen (Fig.  3b). No DDR1 staining of immunoreactivity in the DDR1i HD group (33.1% ± 3.7 was detected in cells co-labeled with Acta2 (myofibro - versus 63.9% ± 1.7 for vehicle-treated group, p < 0.001) blasts) or Emr-1 (macrophages) (Fig.  3c), indicating that (Fig.  3j). No significant reduction in collagen type IV Moll et al. J Transl Med (2018) 16:148 Page 6 of 20 Moll et al. J Transl Med (2018) 16:148 Page 7 of 20 (See figure on next page.) Fig. 3 Pharmacological inhibition of DDR1 activation protects animals against NTS-induced crescentic glomerulonephritis. a Quantitative RT-PCR for Ddr1 mRNA on whole kidney lysate of control mice (Control) and mice injected with nephrotoxic serum and treated with vehicle (vehicle). b Representative Ddr1 in situ hybridization (ISH) performed on tissue harvested from mice 14 days after NTS injection. * = crescent c representative DDR1 ISH double labelling with alpha smooth muscle actin (Acta2) or EGF-like module-containing mucin-like hormone receptor-like 1 (Emr-1) in control mice (Control) and mice injected with nephrotoxic serum and treated with vehicle ( Vehicle). Arrows = cells labeled with Acta2 or Emr-1. d–g Body weight evolution (day 1, 4, 7 and 14) and renal function parameters (e–g blood urea nitrogen (BUN), serum creatinine and proteinuria) measured at sacrifice (day 14). h Representative histopathology with Hematoxylin and Eosin (H&E) and Periodic Acid Schiff staining (PAS) and immunohistochemistry for desmin, CD44, Collagen type IV and Ki67. i Glomerular or tubulo-interstitial ( TI) summary scores from semiquantitative histopathologic evaluation on H&E and PAS stained kidney sections respectively. j Morphometry analysis of collagen IV IHC. Statistically significant p value: p < 0.05 = *; p < 0.005 = **. Magnification ×200, scale bar 100 μm staining was observed in the other treated groups. Finally, of genes related to immune response, integrin pathway in agreement with the histology data, quantitation of activation and fibrotic processes (TGFβ pathway), all the cellular proliferation marker Ki67 showed a marked increased in the NTS group (Fig. 5a). In contrast, tissues reduction of PEC and tubulo-interstitial staining in the from mice treated with DDR1i HD showed a significant DDR1i HD and DDR1i LD groups (Fig. 3h). reduction in the expression of this set of genes (Fig.  5a). This effect was also observed, albeit to a lesser extent, in Selective DDR1 inhibition protects when tested using the DDR1i LD group, but was not in the Imatinib group. a therapeutic intervention regime Noteworthy, among the weakly perturbed gene signa- We then aimed to assess the effect of DDR1 inhibition tures, GSEA analyses revealed that the glomerular gene in the context of progressive glomerulosclerosis in the signature, driven by synaptopodin, podocalyxin, NPHS1 NEP25 mouse model [14]. Mice were treated with the and NPHS2 genes, was clearly preserved in the DDR1i Roche-Chugai DDR1i, Captopril or vehicle (Fig.  4a). HD group, whereas no protection was observed in the Histological and semi-quantitative analyses of glomeru- case of the Imatinib treated group (Fig. 5b). losclerosis in PAS sections and tubulo-interstitial fibro - Comparison of DDR1i GSEA results with a previous sis/inflammation in H&E sections of kidney tissue at datasets generated in the NTS mouse model treated day 15 showed a significant reduction of glomerular with other target inhibitors revealed a consistent pattern PAS positive area in both DDR1i and Captopril groups of biological processes associated with NTS injection (p < 0.001 and p < 0.0001 respectively) (Fig.  4B, C). A that were selectively reverted by DDR1i (Fig.  5c). These slight, but not significant, reduction in tubulointersti - include reactivation of genes involved in glomerular epi- tial damage was observed in the DDR1i-treated group, thelial cell processes and deactivation of genes related to whereas this reduction was significant in the Captopril wound healing, cell matrix, cell adhesion, and motility group (p < 0.001) (Fig.  4c). Analyses of several fibrosis and inflammation processes. Thus, to identify the main and inflammation markers by qRT-PCR in renal cortical driver genes specifically modulated by DDR1i treatment, tissues at day 15 showed that DDR1i induced a marked we conducted additional gene level analysis. These data reduction of alpha smooth muscle actin (p < 0.01), colla- revealed that among the statistically significantly changed gen type 1, TGF-β1 (p < 0.01) and of Ccl2 (p < 0.001) in a genes (n = 785), 175 were identified with a large effect range similar to that observed in Captopril-treated mice size (logFC ≥ 2) and reverted in the DDR1i group as com- (Fig. 4d). Captopril was more efficient in improving renal pared to the NTS group (Additional file  2: Table S3). The function (Fig. 4e, f )due to the fact that Captopril admin- highest fraction of perturbed genes (n = 151) was rep- istration started before disease initiation (preventive resented by genes exhibiting increased expression upon approach), whilst the DDR1 inhibitor was only adminis- NTS injection. Genes regulated by treatment with DDR1i tered during the progression of nephropathy (curative or were hallmarks in the induction of inflammatory pro - interventional approach). cesses, and regulation of cell matrix and adhesion mol- ecules, cell cycle, and apoptosis. DDR1 inhibition modulates specific gene networks An additional classification criteria (detectable expres - in experimentally‑induced crescentic glomerulonephritis sion in the human kidney epithelial cell line HKC8 [17] or In order to unveil the mode of action (MoA) of DDR1i, PEC [18] was adopted to further narrow down the gene mouse kidneys from the NTS experiment were subjected list (n = 45) to those mechanistically-related to DDR1. to gene expression profiling (GEP, GEO Accession Num - To remove model-specific gene perturbations and ber GSE104426). Unsupervised gene signature analysis identify DDR1 MoA, the differentially expressed genes and Gene Set Enrichment Analysis (GSEA) using Gene obtained in the NTS GEP analysis were compared to the Ontology processes as input gene sets showed networks genes significantly perturbed in the NEP25 mouse model Moll et al. J Transl Med (2018) 16:148 Page 8 of 20 Moll et al. J Transl Med (2018) 16:148 Page 9 of 20 Fig. 4 Treatment with DDR1 inhibitor in the NEP25 mouse model of glomerulosclerosis. a Schema of the experiment b Representative histopathology with Hematoxylin and Eosin (H&E) and Periodic Acid Schiff staining (PAS) in control, vehicle-, DDR1i- and Captopril treated groups at day 15. Magnification ×200, scale bar 100 μm. c Semi-quantificative analysis of glomerulosclerosis (glomerular PAS positive area) and tubulointerstitial lesions (tubulointerstitial damage) in control, vehicle-, DDR1i- and Captopril treated groups at day 15. ***p < 0.001, ****p < 0.0001; t-test and Mann–Whitney U test were used for the score of glomerular PAS positive area and tubulointerstitial damage respectively. d Quantitative RT-PCR for the fibrosis markers alpha smooth muscle actin (Acta2 mRNA), collagen type 1 (Col1a mRNA) and TGF-β1 ( TGF-β1 mRNA), and for the inflammation marker Ccl2 in NEP25 mice treated with DDR1 inhibitor (DDR1i), Captopril or vehicle and in control mice. *p < 0.05, **p < 0.01, ***p < 0.001; t-test. e Body weight evolution. f Renal function parameters (plasma creatinine and plasma Cystatin C measured at sacrifice) and urinary ACR (24-h urine collection from day 14 to 15 divided by creatinine concentration) in NEP25 mice treated with DDR1 inhibitor (DDR1i), Captopril or vehicle and in control mice (CTRL) Moll et al. J Transl Med (2018) 16:148 Page 10 of 20 (See figure on next page.) Fig. 5 Gene expression profile in NTS-treated mice. a Unsupervised gene signature analysis and GSEA using Gene Ontology processes as input genesets. b Glomerular cell signature with individual component genes. c Comparison of different target modulation in NTS. Heatmap showing differential modulation (by DDR1i or Target X) of the gene networks modulated by NTS induction compared to control. d Heatmap showing kinome expression modulation in the different treatment conditions Previously, DDR1 has been localized in rat kidney [20] that were reverted by DDR1i (n = 85). The final intersec - and our data extend that knowledge to humans providing tion revealed 30 genes in common to both models (Addi- further confidence for the target, indicating that DDR1 tional file 2: Table S4). expression is restricted to the epithelial compartment Finally, the unexpected lack of therapeutic effect of (with no DDR1 expression in vessels as reported with the Imatinib in the NTS model prompted us to conduct use of unselective antibodies [12]) and providing, for the a specific kinome analysis to identify potential differ - first time, a detailed analysis of the role of DDR1 in the ences in kinase modulation between DDR1i and Imatinib sequential morphological and cellular events leading to groups, which could expain the differential therapeu - crescent formation in humans. DDR1 was expressed (1) tic response. Best BUN-correlated kinases (with a cor- in most injured podocytes during the very initial stage relation coefficient better than 0.7) were retained and of the disease characterized by glomerular intracapillary plotted (Fig.  5d). These analyses revealed that 6 kinases immunoinflammatory processes, (2) in bridging cells co- were specifically upregulated by DDR1i HD treatment expressing podocyte and PEC markers located in front in comparison to Imatinib. In contrast, 19 kinases were of, or adjacent to, disruption of the integrity of the glo- downregulated. merular capillary wall, and (3) in a subset of activated/ Taken together, our data demonstrated that treatment injured PECs forming the cellular crescent, most of these with Roche-Chugai DDR1i reverts inflammatory and cells being located near the glomerular tuft. Based on this fibrotic genes and positively affects genes of the glomeru - expression analysis, it could be speculated that DDR1 lar cell signature particularly podocytes and with an evi- expressed during the initial stage of the crescentic disease dent effect on the expression of specific kinases. in podocytes and bridging cells is reparative but then, when expressed during the active and severe stage of the DDR1 inhibition decreases PEC pro‑inflammatory disease in proliferative PECs is deleterious, participating phenotype in vitro to excessive maladaptive repair processes. We then aimed to asses to characterize the impact of These findings, in addition to previous preclinical DDR1 inhibition on PEC phenotype using a murine evidence of the pathogenic role of DDR1 in crescentic PEC cell line [18]. Pre-treatment of PECs for 1  h with GN [6, 8–10, 12, 13, 21] induced our research group to DDR1i followed by stimulation with collagen type I for develop a highly selective DDR1i. Selectivity of DDR1 24  h blocked DDR1 phosphorylation in a dose depend- over the close analogue protein DDR2 is of utmost ent manner (Additional file  1: Figure S6). Treatment with importance to harness the therapeutic potential of DDR1 type I collagen alone induced a significant increase of the antagonism. DDR2 inhibition has in fact been associated matrix metalloproteinases Mmp14 and Mmp2, comple- with enhanced liver fibrosis (evidence generated by the ment component C3 and vascular cell adhesion molecule renowned group of prof. Scott Friedman [6]) and DDR2 1 (VCAM1) mRNA (Additional file 1: Figure S6B-E). Pre- activation has been shown to inhibit the development of treatment of the cells with DDR1i caused a dose-depend- experimental choroidal neovascularization and retinal ent inhibition of expression of these genes. neovascularization in mice [7]; thus, inhibition of DDR2 could translate into undesirable side-effects in patients. Discussion None of these undesirable side effects were noted in ani - The present study details DDR1 expression in human GN mals treated with the selective DDR1i. biopsies and demonstrates, for the first time using phar - Treatment with our selective DDR1i resulted in pre- macological intervention, that selective DDR1 inhibi- served renal function and structure, as suggested by tion translates into renal protection. These data confirm histology, IHC and morphometry analyses. Glomeru- the central role of DDR1 in glomerular diseases, previ- lar lesions were reduced, in particular the number of ously indicated using gene deletion [8, 12, 13, 19] or gene activated PECs as shown by reduced Desmin and CD44 silencing paradigms [10]. [22] positivity, both recognized markers of an activated Analysis of human renal biopsies with different types of PEC phenotype [22, 23]. Tubulo-interstitial inflamma - crescentic GNs shows DDR1 to be strongly expressed in tion and fibrosis were decreased and the architecture of injured podocytes and in PECs forming cellular crescents. Moll et al. J Transl Med (2018) 16:148 Page 11 of 20 Moll et al. J Transl Med (2018) 16:148 Page 12 of 20 renal epithelium was maintained. The use of a second DDR1 inhibition blocks activation of PEC in vitro. Taken model, the NEP25 mouse model [14], was prompted by together, the in  vivo and the in  vitro evidence seem to our interest to further assess the role of DDR1 in a injury suggested that PEC de-activation could be a key part of model where PEC-induced glomerulosclerosis is caused DDR1i MoA. by podocyte depletion in absence of intra-glomerular One of the major strengths of the present work is that inflammation, primary mesangiolysis and primary hyper - for the first time pharmacological intervention target - tension thus further strengthening the role of DDR1i in ing predominantly the renal epithelium can reduce both the protection of glomerular function also in a model inflammation and fibrosis. These effects seem to occur of focal segmental glomerulosclerosis. Though not as without directly affecting fibroblast to myofibroblast acti - impressive as in the case of NTS, where DDR1i was vation or the inflammatory component as DDR1 expres - administrated in a prevention mode, these findings sug - sion was not observed in the relevant cell types. We are gest that selective DDR1 inhibition is protective when however conscious that the lack of immunohistochemi- used in the context of an established lesion (therapeutic cal staining does not exclude these additional potential regime). effector cells, highly relevant to the two models studied. The choice of two different mouse models to investi - Our hypothesis is concordant with a recently published gate the relevance of DDR1 in GNs was consciously made study demonstrating that selective activation of EGFR to avoid a model-biased interpretation of the results. in renal epithelial cells (proximal tubules) is sufficient to Since the therapeutic regime was already probed, by one induce tubulo-interstitial fibrosis [24]. As shown by our of the co-authors, in the NTS model [19] we chose to molecular genomics analysis of NTS kidney samples, generate data in the alternative model of NEP25 [10] to DDR1 antagonism has a direct effect in the preserva - test if treatment could transfer to protection in a differ - tion of the podocyte network and in the de-activation of ent experimental setup. The NEP25 mouse model proved PECs. It is possible that DDR1 antagonism could reduce to be extremely robust and was characterized by uniform an excessive “response to injury” occurring within the glomerular injury and sclerosis induced by a simple pro- glomerular compartment in case of crescentic lesions, cedure, i.e., a single intravenous injection of LMB2 toxin. and having deleterious consequences on tubulo-intersti- Indeed the model perfectly fits a drug discovery program. tial compartment. Those speculations will have to be fur - Genomic analysis of the NTS experiment, performed ther addressed by deeper mechanistic studies critical to in the context of a drug discovery program, aiming to comprehend the role of DDR1 in glomerular repair and identify biomarkers exploitable in clinical studies allowed regeneration processes. us to suggest that the observed functional and organ pro- Moreover, we can’t ignore that our IHC data seem to tection is due to reduced inflammation and fibrosis. suggest a broader role for DDR1 in kidney. We have in To specifically decode DDR1 molecular MoA, we took fact observed a clear DDR1 tubular upregulation in advantage of data generated in the NTS mouse model human crescentic GN as well as in mouse models of with compounds modulating other pathways (not DDR1- glomerular disease. It is interesting to note that DDR1 dependent) and used them to filter and identify a consist - genetic ablation in other models of chronic, acute or ent pattern of biological processes associated with renal genetic renal disease, such as hypertensive nephropathy, injury and specifically reverted by DDR1 antagonism. Alport’s syndrome and unilateral ureteral obstruction, Gene and pathway level analysis identified DDR1 driver has been shown to be protective [3, 19]. Causal asso- genes associated with cell cycle and cell matrix and adhe- ciation between epithelial cell injury and renal function sion, some amongst them were expected for example dif- deterioration, paralleled by fibrosis might be an integral ferent collagen genes (Col1A1, Col1a2, Col3a1, Col6a2), part of the DDR1 MoA in GN. Certainly, this hypothe- or other extracellular matrix related proteins such as sis deserves further investigation using dedicated mouse TIMP1 and FN1, but also a marked reduction of LTBP2 models and a detailed IHC characterisation in particu- suggestive of a reduced state of activation of epithelial lar in the context of tubulopathies, acute renal ischemic cells. Interestingly, Cd44, a hallmark of activated PECs lesion or transplant reperfusion and/or in general in was observed among the top DDR1i modulated genes. other ischemic conditions. Surprisingly, unsupervised gene signature analysis iden- It should also be mentioned that when designing the tified podocyte-specific genes as being clearly protected preclinical experiments we sought to add control com- by DDR1i treatment. This is quite remarkable since GEP pounds to validate the models adopted. For that reason, were generated using whole kidney lysates. These molec - the control compounds (Imatinib and Captopril, respec- ular findings are concordant with the IHC results dem - tively used in the NTS and NEP25 models) were admin- onstrating PEC de-activation. To support that notion, istered in a prevention mode to insure better protective we also performed a series of experiments showing that performance and consequently better model validation. Moll et al. J Transl Med (2018) 16:148 Page 13 of 20 The doses of both compounds were selected using data approved if the ratios 260/280 and 260/230 were both reported in the literature for the selected models. Lack greater than 1.8. of efficacy of Imatinib was surprising since efficacy has DDR1 splice isoform specific qRT-PCR assays have been reported at the selected dose [25]. Theoretically this been developed for DDR1-variant 3 (DDR1_V3), DDR1- dose has in vivo exposure similar to the DDR1 HD group variant 4 (DDR1_V4) and DDR1-variant 5 (DDR1_V5), based on reported in vitro DDR1 potency [26] and in vivo DDR1-variant 6 (DDR1_V6) which specifically detect pharmacokinetic data (due the better bioavailability and one variant. Due to close sequence homology PCR assays reduced clearance of Imatinib compared to the DDR1 for DDR1-variant 1/2 (DDR1_V1/2), DDR1-variant 1/6 HD). The lack of efficacy of Imatinib, might suggest that (DDR1_V1/6) detect both isoform (see Additional file  3: selective inhibition of DDR1 and not of a broader panel for primer/probe sequences). All PCR assays were devel- of kinases might be a crucial element to achieve kidney oped using synthetic plasmids carrying DDR1 variant protection. Comparative kinome analysis between DDR1i specific sequences assuring comparable sensitivity, speci - and Imatinib extracted from our GEP indicates that lack ficity, PCR amplification efficiency (see Additional file  3: of efficacy of Imatinib might possibly due to antagonism for plasmid sequences). of kinases whose activation is key to confer protection to Reverse Transcription of RNA was accomplished using injury, such as STK16, INSSR or CDK7. This hypothesis the SuperScript III First-Strand Synthesis Kit (Invitrogen) might also suggest an interesting role for those kinases in according to the manufacturers’ protocol with a total GN. RNA input of 100  ng per reaction. cDNA samples were then diluted 1:3 in TE buffer and 14 cycles of pre-amplifi - cation were carried out using 2× TaqMan PreAmp Mas- Conclusion ter Mix (Applied Biosystems) and pooled DDR1 splice In conclusion, in the present work, pharmacological inhi- variant specific assays at a final concentration of 0.2× bition of DDR1 phosphorylation emerges as a unique and per assay. The following thermocycler program was used: differentiated pharmacological therapeutic intervention 95 °C for 10 min, followed by 14 cycles at 95 °C for 15 s capable of preserving renal function in both prevention and 60 °C for 4 min. Pre-amplified cDNA products were and therapeutic regimens, acting on the renal epithelium diluted 1:5 in TE buffer. qPCR was performed using the and resulting in reduced inflammation and fibrosis in the 96.96 dynamic array (Fluidigm Corporation, CA, USA) surrounding interstitial space. Such an intervention para- following the manufacturer’s protocol (Fluidigm Quick digm, when available in the clinics, would provide a novel Reference Card, PN 68000130, Rev. B). Briefly, for each pharmacological tool for combination therapy with anti- sample a 5μl sample mix was prepared with 1× GE inflammatory, endothelial protective or myofibroblast Sample Loading Reagent (Fluidigm), 1× Taqman Gene blocking agents. Expression Mastermix (Applied Biosystems) and diluted, pre-amplified cDNA. For the assay mix 1× Assay Loading Reagent (Fluidigm) was mixed with each of the Taqman Methods Assays (final concentration: 10×), respectively. Priming DDR1 isoform RT‑PCR of the Fluidigm array with control line fluid and mixing RNA was purified from normal, ADPKD and ESRD tis - of sample and assay reagents was done with an IFC con- sue sections using a Tissue Lyser protocol (Qiagen) by troller. qPCR was performed using the BioMark Instru- adding 350  µl ‘RLT buffer’ plus 350  µl 70% ethanol and ment with the following cycling parameters: 95  °C for vortexing. The sample was then transferred to an RNeasy 10 min, followed by 40 cycles at 95 °C for 15 s and 60 °C column and centrifuged at 8000×g for 15 s at room tem- for 1 min. Data was collected and analyzed with the Real- perature. RNase free DNase 1 (80  µl) was added to the Time PCR Analysis Software (Fluidigm Corporation, CA, column and incubated at room temperature for 15  min. USA). Normalization was performed using geometric The column was then washed by adding 350 µl RLT buffer mean expression of housekeeping genes B2M, GAPDH, and centrifuged at 8000×g for 15 s at room temperature. GUSB: ∆Cq = Cq gene of interest—Cq geomean House- This was repeated with ‘RPE buffer’ (500 µl) twice where keepers. Normalized values were transformed into rela- the second spin was prolonged to 2  min. Following this tive expression levels using (POWER; -Cq) calculation. the RNaeasy column was transferred to a new micro- The sum of all DDR1 splice isoform expression levels was tube and centrifuged at 11,000×g for 1  min. The RNAe - set at 100% and DDR1 splice isoforms values calculated in asy tube was once again transferred to a new microtube % accordingly. DDR1 splice isoform variant 1 and variant and the RNA eluted by adding 50 µl H O and centrifuga- 2 isoform expression was calculated using by subtract- tion at 8000×g for 1 min. The elution wash was repeated ing results of DDR1-variant 6 (DDR1_V6) from DDR1- once and the eluate pooled. Absorbance was measured at variant 1/6 (DDR1_V1/6) to obtain variant DDR1-variant 230, 260 and 280 nm in a spectrophotometer and quality Moll et al. J Transl Med (2018) 16:148 Page 14 of 20 1 (DDR1_V1) results which were then used to calculate supplements were purchased from Thermo Fisher Sci - DDR1-variant 2 (DDR1_V2) results. entific (Waltham, MA). Eight-well Permanox Lab-Tek chamber slides were purchased from NUNC (Cat.# 177445). ImmPRESS Reagent (Anti-Rabbit Ig) (Cat. # DDR1 in situ hybridization on mouse and human kidney MP-7401) and ImmPACT DAB Peroxidase Substrate sections (Cat. # SK-4105) were purchased from Vector Labora- In situ hybridization (ISH) on 4  μm-thick mouse or tories (Burlingame, CA). A rabbit monoclonal antibody human kidney sections was performed using the against DDR1 (D1G6) was purchased from Cell Signal- RNAscope 2.5 HD Duplex Assay from Advanced Cell ing Technology (Danvers, MA) and a monoclonal anti- Diagnostics (California, USA) and the DDR1 probes body against β-actin was purchased from Sigma (St. according to the manufacturer’s instructions: mouse Louis, MO). Mayer’s hematoxylin (Cat.# HMM999) DDR1 (RNAscope Probe-Mm-Ddr1-C2) and human was purchased from Scytek and Crystal Mount Aque- DDR1 (RNAscope Probe- Hs-DDR1-C2, target region ous Mounting Medium (Cat.# C0612) from Sigma 285—2016, #593591-C2). Double ISH was performed (St. Louis, MO). Immunoblot analyses and cell immu- on mouse kidney sections using mouse DDR1 probe and nostaining were performed to assess antibody specific - mouse Acta2 (RNAscope Probe- Mm-Acta2-C2) or ity and sensibility. mouse Emr1 (RNAscope Probe- Mm-Emr1). Images For immunoblot analyses, cells were lysed in RIPA were acquired with an Olympus VS120 scanner (Olym- buffer (50  mM Tris–HCl, pH 7.4, 150  mM NaCl, 1% pus AG, Switzerland) equipped with a VC50 camera and NP-40, 0.25% sodium deoxycholate and 1.0  mM EDTA) 20× objective. supplemented with protease inhibitors on ice for 1  h. The cell lysates were cleared by centrifugation at 14,000g Generation of a selective anti‑DDR1 monoclonal antibody at 4 °C for 15 min, and protein concentration was deter- (DDR1 Rab‑819) for IHC analysis mined using the BCA kit (Pierce). 20–40  µg of protein New Zealand White (NZW) rabbits were used for immu- lysate were mixed with 1× reducing Laemmli SDS-sam- nization with extracellular domain (ECD) DDR1 protein ple buffer. After a brief centrifugation, the supernatants (R&D Bio-Tech, Zug, Switzerland). The immunization were boiled and resolved by reducing 7.5% SDS-PAGE, protocol included the repeated injection of immunogen followed by immunoblot analyses using anti-DDR1 anti- emulsified with CFA into the same animal rotating dif - body D1G6 or SC532. The blots were reprobed with anti - ferent application routes. 10  ml peripheral whole blood bodies against β-actin for loading control. For positive were collected and used for B cell isolation followed by control of DDR1 immunoreactivity, we used a lysate of B-cell culture. After total RNA isolation, cDNA was gen- human prostate cancer PC3 cells overexpressing DDR1b, erated by reverse transcription of the mRNA followed by which was generated in our laboratory. PCR amplification of the V regions of the rabbit B cells For immunostaining, CFPAC-1 or PANC-1 cells using appropriate primers. The DNA sequences encod - (5 × 10 /250  µl of complete medium) were seeded in ing the VHs and VLs were obtained by sequencing the 8-well chamber slides. Twenty-four hours later, the media PCR products. Prototype cDNA expression plasmids were removed and the cells were fixed with cold (− 20 °C) were used for the recombinant expression of the HC methanol. The cells were then washed (3×) with PBS and and LC of monoclonal rabbit antibodies that are then then incubated with 3% H O in PBS for 20 min at room 2 2 expressed transiently in HEK293 cells. After 7  days, the temperature (RT) followed by two washes with PBS. The culture supernatants were harvested, purified by Pro - cells were incubated with 2.5% normal horse serum for tein A column standard protocols and analyzed for anti- 20 min at RT. The serum was removed and the cells were body content and specificity on human pancreatic cancer incubated with either in house raised DDR1 antibody CFPAC-1 and PANC-1 cells. (1:50) or control rabbit IgG (1:50) diluted in PBS. After an overnight incubation at 4  °C, the cells were washed Anti‑DDR1 antibody specificity testing using DDR1 with PBS and incubated with the ImmPRESS Reagent or DDR2 expressing cancer cell lines for 30 min at RT. Detection and visualization of antibody Human pancreatic cancer CFPAC-1 and PANC-1 cells binding was assessed using ImmPACT DAB Peroxi- were a gift from Dr. Howard Crawford, University of dase chromogenic substrate, according to the manufac- Michigan. CFPAC-1 cells were maintained in IMDM turer instructions. The slides were counterstained with supplemented with 10% fetal bovine serum 1% l -glu- Mayer’s hematoxylin for 5  s, followed by differentiation tamine and 1% streptomycin/penicillin antibiotics, and with tap water for 1 min, and finally covered with a thin PANC-1 cells in DMEM High Glucose supplemented layer of Crystal Mount Aqueous Mounting Medium. with 10% fetal bovine serum 1% l -glutamine and 1% Stained cells were photographed using a Zeiss Axioplan 2 streptomycin/penicillin antibiotics. Culture media and Moll et al. J Transl Med (2018) 16:148 Page 15 of 20 microscope (Zeiss, Gottingen, Germany) equipped with alkaline-fast red enzyme system (DakoCytomation, a software-controlled digital camera (Axiovision; Zeiss). Glostrup, Denmark). For DDR1 expression experiments during crescent Human renal tissues formation, immunostaining was performed on serial Human renal tissue, fixed in formaldehyde and embed - sections of 2 lupus nephritis class IV-G (A/C) biopsies ded in paraffin, was selected from the files of the Service (patients #25 and #29) using DDR1 and each of these 4 of Pathology, University Hospital Geneva: five control different antibodies: mouse monoclonal anti-human normal renal tissues were obtained from patients with CD68 (DakoCytomation, Glostrup, Denmark) at a 1:100 nephrectomy performed for neoplasia, and 29 biopsy dilution, mouse monoclonal anti-human nestin (R&D specimens were obtained from patients with crescentic systems Bio-Techne, Minneapolis. MN) at a 1:750 dilu- glomerulonephritis (GN): 5 Goodpasture’s syndrome, tion, mouse monoclonal anti-human cytokeratin 8 7 ANCA-associated GN, 12 IgA GN, and 5 lupus GN. and 18 (BioGenex, Fremont, CA) at a 1:20 dilution and For all biopsy specimens, standard analysis using light mouse monoclonal anti-human cytokeratin 19 (Dako- microscopy, immunofluorescence (with anti-immuno - Cytomation, Glostrup, Denmark) at a 1:20 dilution. In globulin Ig A, G, M, and anti-complement C1q, C3, C4c, addition, some serial sections were immunostained with and C5b-9 antibodies), and electron microscopy were E-cadherin (Novocastra, Newcastle, UK) at a 1:20 dilu- performed. For all biopsy specimens, standard analyses tion. Briefly, 3  μm sections of paraffin-embedded kid - were performed. Each patient gave informed consent neys were submitted to the appropriate antigen retrieval before enrollment. The institutional ethical committee and incubated with each antibody 1 h at room tempera- board approved the clinical protocol (CEREH Number ture followed by the appropriate secondary antibody for 03-081). The research was performed according to the 30 min (room temperature) and then liquid diaminoben- Helsinki’s declaration principles. zidine substrate–chromogen system (DakoCytomation, Glostrup, Denmark). Immunohistochemistry on human tissues Counterstaining was performed using Mayer hematox- Immunohistochemistry was performed as follows: after ylin. For CD68 and E-cadherin, immunostaings, sections antigen heat retrieval, 3  μm sections of the formalde- were counterstained with Periodic Acid Schiff (PAS) hyde-fixed, paraffin-embedded biopsy specimens were staining. Stained sections were examined with a Zeiss incubated with our specifically in-house raised rabbit microscope. monoclonal anti-human DDR1 antibody at a 1:100 dilu- tion 1  h at room temperature followed by an anti-rabbit Pharmacokinetic studies antibody for 30  min (room temperature) and then liq- The route of administration and dosing-regimes for uid diaminobenzidine substrate–chromogen system in  vivo studies study were based on pharmacokinetic (DakoCytomation, Glostrup, Denmark). For tubular studies. Roche-Chugai DDR1i was formulated as a colocalization experiments in the normal kidney, double micro-suspension in a vehicle (7.5% gelatin and 0.9% immunostaining was performed on serial sections (3 μm sodium chloride) and administered by oral application thick) using DDR1 and each of these 4 different anti - at 10 ml/kg. The oral pharmacokinetics of Roche-Chugai bodies: mouse monoclonal anti-human Tamm–Horsfall DDR1i was investigated in male 129 Sv mice (Additional protein (Technically Speaking, Ontario, Canada) at a file  1: Figure S4). The compound was administered via 1:80 dilution, mouse monoclonal anti-human Calbindin oral gavage, at 200 mg/kg. Plasma samples (0.05 ml) were D (clone CB-955, Sigma-Aldrich, St. Louis, MO) at a collected at 1, 3, 6 and 24  h post-dose (n = 2 mice/time 1:800 dilution, rabbit polyclonal anti-human Aquaporin point). Concentrations in mouse plasma samples were 2 (Abcam, Cambridge, UK) at a 1:20 dilution and rab- determined using a high performance liquid chroma- bit polyclonal anti-human Megalin (anti-LRP2, Sigma- tography-tandem mass spectrometry (HPLC–MS/MS) Aldrich) at a 1:5000 dilution. Briefly, 3  μm sections method. of paraffin-embedded kidneys were submitted to the appropriate antigen retrieval and incubated with DDR1 Roche‑Chugai DDR1i in vivo pharmacokinetics at a 1:100 dilution 1  h at room temperature followed by and estimated pharmacodynamics an anti-rabbit antibody for 30  min (room temperature) Estimation of inhibition of DDR1-phosphorylation was and then liquid diaminobenzidine substrate–chromo- calculated from the observed plasma concentrations in gen system (DakoCytomation, Glostrup, Denmark). Sec- relationship to the in vitro binding IC50 for Roche-Chu- tions were then incubated with the adequate antibody gai DDR1i against human DDR1, accounting for mouse for 1  h at room temperature followed by the appropri- plasma protein binding of 98.9% (Additional file  1: Fig- ate second antibody for 30 min and then by phosphatase ure S4). Thus, inhibition of DDR1-phosporylation of Moll et al. J Transl Med (2018) 16:148 Page 16 of 20 > 90% in average over 24 h in vivo was estimated from the the seminal work of Morley and Wheeler in mice [28, observed plasma concentrations of Roche-Chugai DDR1i 29]. Briefly, sheeps were immunized by subcutaneous on daily treatment of 129 Sv mice with 200 mg/kg po. injection of mouse isolated glomeruli. The total dose of isolated glomeruli (600  μg) was divided to four injec- Kinase assays tions separated by a 1  week interval. 15  days later and The KINOMEscan screening platform (DiscoveRx Cor- under treatment with anti-allergic agent (Phenergan, SM poration, San Diego, USA DiscoverX s canMAX Kinase Wyeth, USA), sheep received another intravenous injec- Assay Panel) employing an active site-directed competi- tion of the same quantity of mouse glomeruli. One week tion binding assay was used to quantitatively measure later animals were sacrificed and serum was collected. interactions between the 2 test compounds Roche-Chu- Serum was then heat-inactivated at 56 °C for 30 min and gai DDR1i and Imatinib on 451 kinases and disease rel- filtered (pore size = 0.2  μm) and stored in 1  ml aliquots evant mutant variants. In brief, T7 kinase-tagged phage at − 20  °C. Next we passively induced glomerulonephri- strains were grown in parallel in 24-well blocks or 96-well tis by injecting mice intravenously with the serum col- blocks in an E. coli host derived from the BL21 strain. E. lected from sheep. We used four different doses (5, 10, coli were grown to log-phase and infected with T7 phage 15, 20 μl/g) for the four serum lots in order to validate the from a frozen stock (multiplicity of infection = 0.4) and serum and choose the appropriate dose. All serum lots incubated with shaking at 32 °C until lysis (90–150 min). were toxic at doses 10, 15 and 20  μl/g and non-nephro- The lysates were centrifuged (6000×g) and filtered toxic at 5 μl/g dose. All serums were specific against glo - (0.2  μm) to remove cell debris. The remaining kinases merular compartments (data not shown). One lot was were produced in HEK-293 cells and subsequently tagged found to induce severe proteinuria with typical histologi- with DNA for qPCR detection. Streptavidin-coated cal lesions at 10 μl/g; therefore this serum was used in the magnetic beads were treated with biotinylated small next protocol. molecule ligands for 30 min at room temperature to gen- erate affinity resins for kinase assays. The liganded beads Dose selection for the NTS experiment were blocked with excess biotin and washed with block- DDR1i doses selection was designed based on in  vitro ing buffer [SeaBlock (Pierce), 1% BSA, 0.05% Tween 20, pharmacology and bioavailability and clearance obtained 1  mM DTT] to remove unbound ligand and to reduce in the above mentioned pharmacokinetic studies. Two nonspecific phage binding. Binding reactions were doses were selected a dose covering 10× the in  vitro assembled by combining kinases, liganded affinity beads, IC50, resulting in the DDR1 HD dose (200  mg/kg) and and the 2 test compounds in 1× binding buffer (20% Sea - a second dose around 1× coverage over IC50, result- Block, 0.17× PBS, 0.05% Tween 20, 6  mM DTT). Test ing in the DDR1 LD dose (75 mg/kg). The Imatinib dose compounds were prepared as 40× stocks in 100% DMSO (60  mg/kg) was simply derived from the literature [25] and directly diluted into the assay. All reactions were per- and was added to the study as pure control since the formed in polypropylene 384-well plates in a final volume imatinib in vitro potency and in vivo exposure (data not of 0.02 ml. The assay plates were incubated at room tem - shown) are in the very same order of the DDR1i HD dose. perature with shaking for 1 h and the affinity beads were washed with wash buffer (1× PBS, 0.05% Tween 20). Animal treatment and nephrotoxic serum protocol The beads were then re-suspended in elution buffer (1× All mice were kept in well-controlled animal housing PBS, 0.05% Tween 20, 0.5  μM non-biotinylated affinity facilities and had free access to water and pellet food. ligand) and incubated at room temperature with shaking Animal procedures and protocols were in accordance for 30  min. The kinase concentration in the eluates was with the European Guidelines for the Care and use of measured by qPCR. The Selectivity Score (S-score) was Laboratory Animals and have been approved by the calculated for both compounds. The compounds were Inserm and UPMC ethical committees. screened at the concentrations requested, and results Glomerulonephritis was induced by retro-orbitaly injec- for primary screen binding interactions were reported as tion of decomplementated nephrotoxic serum (NTS). percent competition (% Competition). A total of 40 female mice 129/SV aged 3–6  months and weighting 18–25  g were used (Janvier, Le Genest-St-Isle, Nephrotoxic mouse model of crescentic France). The total number of mice was divided into five glomerulonephritis treatment groups as followed: 8 mice were injected with Preparation of nephrotoxic serum NTS and fed with vehicle, 8 mice were injected with NTS Decomplementated nephrotoxic serum (NTS) was pre- and fed with low dose DDR1i, 8 mice were injected with pared as previously described [27]. This protocol has NTS and fed with high dose DDR1i, 8 mice were injected been popularized by Salant and Cybulsky according to with NTS and fed with Imatinib and 8 mice were injected Moll et al. J Transl Med (2018) 16:148 Page 17 of 20 Immunohistochemistry and morphometric analysis on NTS with PBS and fed with vehicle. NTS was injected in mice mouse tissues (10  μl/gBW/day) during 3 consecutive days. Treatment On consecutive 4  μm thick slides, immunohistochemis- was started 1 day prior first injection of NTS or PBS. All try (IHC) was performed on the Ventana Discovery XT treatments were provided by oral gavage. The average immunostainer with anti-desmin rabbit polyclonal anti- food intake was controlled by weighing the food every body (Spring Bioscience, E2574, dilution 1:100), anti- 3 days. Mice were found to consume about 4 g/day/mouse CD44 rat monoclonal (BD Pharmingen, 550538, dilution which was similar to all groups. 1:50), anti-collagen type IV rabbit polyclonal antibody (Millipore, AB756P, dilution 1:200) and anti-Ki67 rat Urine and plasma sample collection and analyses polyclonal antibody (eBioscience, 14-5698-82, dilution All mice were acclimated in metabolic cages for 24 h with 1:2000), as primary antibodies in a standard protocol. free access to food and tap water for 24-h urine collec- The Biotin-SP-conjugated AffiniPure donkey anti-rabbit tion. Proteinuria, expressed as grams of protein per mil- IgG (Jackson ImmunoResearch, 711-065-152, dilution limole of creatininuria, was assessed at day 14 using the 1:100) for desmin and collagen type IV or the Biotin- Pyrogallol Red method and utilizing a KONELAB auto- SP-conjugated AffiniPure donkey anti-rat IgG (Jackson mate (Thermo Scientific, Waltman, MA, USA). Blood ImmunoResearch, 712-065-153, dilution 1:100) for CD44 samples were collected on the day of sacrifice (day 14) in and Ki67 were used as secondary antibody in combina- EDTA tubes. Creatinine and BUN were assessed in blood tion with the Ventana DAB Ma p (05266360001) detec- plasma and measured using an enzymatic spectrophoto- tion kit. Slides were counterstained with hematoxylin. metric method and were expressed respectively as mg/dl Collagen type IV IHC slides were scanned with an Aperio and mmol/l. ScanScopeAT slidescanner. Quantitative morphom- etry analysis was performed with a rule set recognizing Animal sacrifice and tissue processing the percentage area of collagen type IV staining using All mice were sacrificed at day 14 post injection. Kidney the Definiens TissueStudio Version 3.51 software. All tissue was processed after normal saline perfusion as fol- images were taken at 20×. lowed. Left kidney: one half was fixed in formalin acetic acid (Formol 2%, acetic acid 5%, alcohol 75% and distilled NEP25 mouse model of glomerulosclerosis water 18%) for 12 h and then embedded in paraffin, and NEP25 mice were kindly provided from Prof. Matsusaka the other half was frozen (OCT-embedded). Right kid- in Tokai University. All mice were kept in well-controlled ney: one half was fixed in formalin acetic acid for 12  h animal housing facilities and had free access to water and then embedded in paraffin. The other half was cut in and pellet food. Animal procedures and protocols were two; 1/4 snap frozen, and the other 1/4 was kept in RNA in accordance with the Guidelines for the Care and Use later and frozen in liquid nitrogen. of Laboratory Animals at Chugai Pharmaceutical Co. Ltd. and approved by Institutional Animal Care and Use Histology and semi‑quantitative scoring analyses on NTS Committee. mouse tissues A total of 28 male mice (4 C57BL/6NCrlCrlj mice From kidney paraffin blocks, sections of 4  μm were pre - and 24 Nephrin-hCD25 (NEP25) transgenic mice) aged pared and stained with Hematoxylin and Eosin (H&E) or 10  weeks were used. One day before LMB2 (Anti-Tac Periodic Acid Schiff (PAS) and examined under a con - (Fv)-PE38) injection, NEP25 mice were divided into three ventional light microscope (Zeiss Axioskop). The fol - groups of 8 mice each based on bodyweight as followed: lowing glomerular findings were appreciated and scored: one group with DDR1i treatment, one group with Capto- glomerular hypertrophy, mesangial matrix expansion, pril treatment and one group receiving vehicle. 0.7  ng/g hypercellularity and hyperplasia of the parietal epithe- BW of LMB2 was intravenously injected in NEP25 mice lial cells and crescents; the following tubulo-interstitial to induce glomerulosclerosis (day 1). Captopril treat- findings were appreciated and scored: tubular degenera - ment (0.15  mg/mL in drinking water, approximatively tion/regeneration, tubular casts, interstitial inflammation 30  mg/kg/day) was started 1  day before LMB2 injection and interstitial fibrosis. For severity scoring, the follow - and given until sacrifice. DDR1i treatment (50  mg/kg/ ing scores were applied for each mouse tissue: 0 = none; day) and vehicle were started 7  days after LMB2 injec- 1 = minimal (approximately 1–20% of the kidney affected; tion. DDR1i and vehicle were provided by oral gavage. All 2 = slight (approximately 21–40% affected); 3 = moderate mice were acclimated in metabolic cages with free access (approximately 41–60% affected); 4 = marked (61–80% to food and tap water. Body weight was measured daily. affected); 5 = severe (approximately 81–100% affected). A 24-h urine was collected from day 14 to 15. Albumin Both glomerular and tubulo-interstitial summary scores and creatinine were measured in urine by Lbi s Mouse were built from the individual scores. Moll et al. J Transl Med (2018) 16:148 Page 18 of 20 Urinary Albumin Assay Kit using a TBA-120FR (Toshiba (Affymetrix Inc., Santa Clara, CA). Quantile normali - Medical Systems Corporation). Albumin-creatinine ratio zation using the Robust Multi-array Average (RMA) (ACR), expressed as mg of albuminuria/mg of creatin- method was applied to the raw individual microarray inuria, was assessed at day 15. All mice were sacrificed data set. The dataset was processed using the standard at day 15 post injection. It should be mentioned that Bioconductor affy package [30]. After RMA normali - one mouse in the DDR1i group was excluded from the zation, probes representing the same genes were col- analysis because congenital renal anomalies were found lapsed into a single value and standardized by taking the at autopsy (DDR1i group N = 7). Blood samples were col- mean value for each gene across the sample set. For each lected and mixed with heparin. Creatinine was measured treatment group (LD DDR1i, HD DDR1i and Imatinib) by HPLC and expressed as mg/dl. Cystatin C was meas- contrasts were calculated against the Control and the ured using Iatro CysC (LSI Medience Corporation) and resulting differentially expressed gene list were subject to expressed as mg/L. Both kidneys were removed and pro- Gene Set Enrichment Analysis [31]. cessed as followed: one-quarter of each kidney was fixed in either 10% Formalin Neutral Buffer Solution (for H&E Gene ontology signature enrichment staining) or Methyl Carnoy’s fixative (for PAS staining) Expression signals for each samples was tested for and then embedded in paraffin. Cortex of the rest kidney enrichment of gene ontology biological process terms was snap frozen in liquid nitrogen for mRNA analysis. [32] using an implementation of the Wilcoxxon’ test PAS stained sections were analyzed for quantifying glo- called bioQC [33]. The resulting enrichment scores were merular damage. PAS positive area was graded for each taken as a measure of the upregulation/downregulation glomerulus using a score of 0–4 as previously described of a specific biological process. [14]. Briefly, more than 50 of randomly selected glomeruli from each mouse were evaluated for glomerulosclerosis. In vitro PEC experiments Score 0 represents no lesion, whereas 1, 2, 3, and 4 rep- Mouse primary PEC were plated and rested overnight. resent mesangial matrix expansion, hyalinosis, or sclero- Cells were incubated in RPMI-1640 containing 1% serum sis, involving ≤ 25, 25% to ≤ 50, 50% to ≤ 75 and > 75% of and supplemented with different concentrations of DDR1 the glomerular tuft area, respectively. Tubulointerstitial inhibitor (0.01, 0.1 and 1  μM) for 1  h followed by addi- damage was graded for each mouse on H&E-stained sec- tional treatment with type I collagen 100  μg/mL (Nitta tions, using a score of 0–4. Score 0 represents no lesion, Gelatin, Japan). After 6 or 24 h of type I collagen stimu- whereas 1, 2, 3, and 4 represents lesions of tubulointersti- lation, cells were lysed for phospho DDR1 ELISA (Cell tial fibrosis/inflammation, involving ≤ 25, 25% to ≤ 50%, Signaling Technology) and total RNA isolation. Total 50 to ≤ 75%, and > 75% of the cortex tubulointerstitial RNA was isolated and amplified using an RNeasy Mini area, respectively. kit (Qiagen) and Transcriptor Universal cDNA Mas- Total RNA was isolated with RNeasy MINI kit (QIA- ter (Roche) according to manufacturer’s instructions. GEN, #74106) from frozen cortical kidney lysate and Quantitative RT-PCR was performed on a LightCycler cDNA was synthesized by reverse transcription. qRT- LC480 (Roche) for C3, Mmp2, Mmp14 and Vcam1 using PCR was performed using fibrosis markers Tgf -b1 the following primers: C3 (Mm01232779_m1, TaqMan (Mm01178820_m1, TaqMan Gene Expression Assays, Gene Expression Assays, Applied Biosystems), Mmp2 Applied Biosystems), Acta2 (Mm00725412_s1) and (Mm00439498_m1), Mmp14 (Mm00485054_m1), and Col1a1 (Mm00801666_g1) and of the inflammation Vcam1 (Mm01320970_m1). Relative gene expression was marker Ccl2 (Mm00441242_m1). calculated with the 2-∆Ct method using GAPDH as an endogenous control. Microarray analysis of gene expression on NTS and NEP25 mouse tissues Statistics Harvested tissues were lysed with RNeasy lysis buffer Values are expressed as mean ± SE M. Box-and-whisker immediately after treatment with MM or DM. Lysates plot represents quartiles. Data were analyzed using one- were sheared with QIAshredder spin columns and total way analysis of variance followed by protected least sig- RNA was extracted using RNeasy kits as per the man- nificant difference Fisher’s test of the Stat-view software ufacturer’s guidelines (QIAGEN GmbH, Hilden, Ger- package. Glomerular and tubular summary scores and many). Starting with 1 μg total RNA per sample, reverse collagen type IV quantitative morphometry data were transcription into cDNA and subsequent steps until analyzed using the Kruskal–Wallis test followed by p hybridization onto GeneChip oligonucleotide microar- value adjusted Dunett’s post hoc test. Data on in  vivo rays (Human Genome U133 plus 2.0) and scanning were NEP study were analysed using t-test with GraphPad conducted using the manufacturer’s kits and instructions Prism except for the score of tubulointerstitial damage Moll et al. J Transl Med (2018) 16:148 Page 19 of 20 manuscript and Prof. Matsusaka ( Tokai University Medical School, Isehara, which was analysed by Mann–Whitney U test. Data on Kanagawa, Japan) for kindly providing NEP25 mice. in vitro PEC study were analysed using Dunnett’s mul- tiple comparison test with GraphPad Prism. Results Competing interests At the time of the experiments L.B., I.F., G.G., S.U., T.B., R.G., M.T., H.R., A.A., J.F. with p < 0.05 were considered statistically significant. and M.P. were employees of Hoffmann-La Roche Ltd. Y.Y, T.M, H.S., A.M, N.F and M.K. are Chugai Pharma Ltd. employees., S.M., R.F., M.J.M and D.R.B have Additional files no conflict of interest. A.A and C.C. received a research grant support from Hoffmann-La Roche Ltd. Additional file 1. Additional figures. Availability of data and materials Additional file 2. Additional tables. All data gene expression data generated in NTS mouse model have been deposited on the GEO database (GEO Accession Number GSE104426). For the Additional file 3. Additional materials. other data and materials, please, contact author for data requests. Consent for publication Abbreviations All authors have given consent for publication. DDR1: Discoidin domain receptor 1; GN: glomerulonephritis; NTS: nephrotoxic serum; PEC: parietal epithelial cell; RTK: receptor tyrosine kinase; KO: knockout; Ethics approval and consent to participate ASO: antisense oligonucleotides; DDR1i: DDR1 inhibitor; IHC: immunohisto- Studies in NTS and NEP25 mouse models of GNs were approved by the local chemistry; CK: cytokeratin; GEP: gene expression profiling; GSEA: Gene Set appropriate ethical review boards. NTS animal authorisation was released Enrichment Analysis; MoA: mode of action; BUN: blood urea nitrogen. by the French minister of research (Authorisation Number C 75-20-01 of the 17.11.2014) and the NEP25 animal authorisation was released by Chugai Authors’ contributions institutional animal care and use committee, a committee accredited by the SM has generated all human data included in that manuscript analysed the Association for Assessment and Accreditation of Laboratory Animal Care data and co-written the manuscript with MP. YY has designed and conducted (Authorisation Number 12-093). the NEP25 in vivo experiment and analysed the data. TM has invented and provided the DDR1i used in the present study. FD has written part of the text Funding and provided critical review of the manuscript. NF has contributed to the No funding was requested to support that research work. entire experimental design of NEP25 in vivo study and examined the pathol- ogy of the NEP25 study. MK has generated the PEC in vitro experiment. AA has performed the in vivo NTS experiments HS has run the NEP25 gene expres- Publisher’s Note sion profile analysis. HS and AM have contributed to the entire experimental Springer Nature remains neutral with regard to jurisdictional claims in pub- design of NEP25 in vivo study and PEC in vitro study. SU has provided the lished maps and institutional affiliations. ISH characterisation of both NEP25 and NTS tissue samples. LB has run the entire gene expression profile analysis. TC has run the immunohistochemi- Received: 30 April 2018 Accepted: 23 May 2018 cal characterisation of DDR1 in renal biopsies. IF contributed to the entire experimental design of the study and provided valuable biomarker support to the present manuscript. GG has generated and provided the proprietary anti-DDR1 antibody. TB has generated the DDR1 isoform qRT-PCR data. RG has run the pharmacokinetic of compound A and Imatinib and designed the References doses for the in vivo NTS experiment and provided valuable support to data 1. Vogel W, et al. The discoidin domain receptor tyrosine kinases are acti- interpretation. DRB and RF have characterised and validated the proprietary vated by collagen. Mol Cell. 1997;1(1):13–23. anti-DDR1 antibody using specific cancer cell lines. HR has run the in vitro 2. Valiathan RR, et al. Discoidin domain receptor tyrosine kinases: new play- characterisation of DDR1i (binding, phosphorylation and selectivity screen). ers in cancer progression. Cancer Metastasis Rev. 2012;31(1–2):295–321. JF has contributed the pathological characterisation of the NTS in vivo experi- 3. Leitinger B. Discoidin domain receptor functions in physiological and ment. MJM has contributed critical reflection on the interpretation of the PEC pathological conditions. Int Rev Cell Mol Biol. 2014;310:39–87. in vitro experiments. CC contributed NTS experimental design and revised the 4. Borza CM, Pozzi A. Discoidin domain receptors in disease. Matrix Biol. manuscript. MP has designed the experiments, analysed the data and written 2014;34:185–92. the manuscript. All authors read and approved the final manuscript. 5. Yeh YC, Lin HH, Tang MJ. A tale of two collagen receptors, integrin beta1 and discoidin domain receptor 1, in epithelial cell differentiation. Am J Author details Physiol Cell Physiol. 2012;303(12):C1207–17. Department of Pathology, University Hospital of Geneva, Geneva, Swit- 6. Hahn WH, et al. Linkage and association study of discoidin domain recep- zerland. Research Division, Chugai Pharmaceutical Co., Ltd, Tokyo, Japan. tor 1 as a novel susceptibility gene for childhood IgA nephropathy. Int J 3 4 INSERM, UMR S 1155, Hôpital Tenon, 75020 Paris, France. Roche Pharma Mol Med. 2010;25(5):785–91. Research and Early Development, Roche Innovation Center Basel, Basel, 7. Kavvadas P, Dussaule JC, Chatziantoniou C. Searching novel diagnostic Switzerland. Roche Pharma Research and Early Development, Roche Innova- markers and targets for therapy of CKD. Kidney Int Suppl. 2014;4(1):53–7. tion Center Munich, Munich, Germany. Department of Pathology, College 8. Flamant M, et al. Discoidin domain receptor 1 null mice are protected of Medical Sciences, Nova Southeastern University, Fort Lauderdale, FL, USA. against hypertension-induced renal disease. JASN. 2006;17(12):3374–81. 7 8 Department of Pathology, Wayne State University, Detroit, MI, USA. Depart- 9. Guerrot D, et al. Discoidin domain receptor 1 is a major mediator of ment of Nephrology and Clinical Immunology, RWTH University, Aachen, Ger- inflammation and fibrosis in obstructive nephropathy. Am J Pathol. many. Present Address: Chugai Pharmabody Research Pte. Ltd., Singapore, 2011;179(1):83–91. Singapore. Present Address: Late Stage, AstraZeneca, Göteborgs, Sweden. 10. Kerroch M, et al. Protective effects of genetic inhibition of Discoidin Present Address: Office of Innovation, Immunology, Infectious Diseases & Domain Receptor 1 in experimental renal disease. Sci Rep. 2016;6:21262. Ophthalmology (I2O), Roche and Genentech Late Stage Development, 124 11. Alfieri C, et al. Discoidin domain receptor-1 and periostin: new players in Grenzacherstrasse, 4070 Basel, Switzerland. School of Pharmaceutical Sci- chronic kidney disease. Nephrol Dial Transplant. 2015;30(12):1965–71. ences, University of Geneva, Geneva, Switzerland. 12. Kerroch M, et al. Genetic inhibition of discoidin domain recep- tor 1 protects mice against crescentic glomerulonephritis. FASEB J. Acknowledgements 2012;26(10):4079–91. We acknowledge Andrea Araujo Del Rosario and Franziska Weibel for their 13. Gross O, et al. Loss of collagen-receptor DDR1 delays renal fibrosis in technical support, Faye Drawnel for language proofing of the original hereditary type IV collagen disease. Matrix Biol. 2010;29(5):346–56. Moll et al. J Transl Med (2018) 16:148 Page 20 of 20 14. Matsusaka T, et al. Genetic engineering of glomerular sclerosis in the 24. Overstreet JM, et al. Selective activation of epidermal growth factor mouse via control of onset and severity of podocyte-specific injury. JASN. receptor in renal proximal tubule induces tubulointerstitial fibrosis. FASEB 2005;16(4):1013–23. J. 2017;31(10):4407–21. 15. Jennette JC. Rapidly progressive crescentic glomerulonephritis. Kidney 25. Iyoda M, et al. Preventive and therapeutic effects of imatinib in Wistar- Int. 2003;63(3):1164–77. Kyoto rats with anti-glomerular basement membrane glomerulonephri- 16. Iyoda M, et al. Long- and short-term treatment with imatinib attenuates tis. Kidney Int. 2009;75(10):1060–70. the development of chronic kidney disease in experimental anti- 26. Day E, et al. Inhibition of collagen-induced discoidin domain receptor glomerular basement membrane nephritis. Nephrol Dial Transplant. 1 and 2 activation by imatinib, nilotinib and dasatinib. Eur J Pharmacol. 2013;28(3):576–84. 2008;599(1–3):44–53. 17. Moll S, et al. Epithelial cells as active player in fibrosis: findings from an 27. Salant DJ, Cybulsky AV. Experimental glomerulonephritis. Methods Enzy- in vitro model. PLoS ONE. 2013;8(2):e56575. mol. 1988;162:421–61. 18. Kabgani N, et al. Primary cultures of glomerular parietal epithelial cells or 28. Lloyd CM, et al. RANTES and monocyte chemoattractant protein-1 podocytes with proven origin. PLoS ONE. 2012;7(4):e34907. (MCP-1) play an important role in the inflammatory phase of crescentic 19. Dorison A, Placier S, Dubois Y, Chladichristos C, Rondeau E, Chatzianto- nephritis, but only MCP-1 is involved in crescent formation and interstitial niou C, Dussaule JC. Discoidin domain receptor 1 is a key mediator of fibrosis. J Exp Med. 1997;185(7):1371–80. ischemia-reperfusion induced injury. San Diego: American Society of 29. Morley AR, Wheeler J. Cell proliferation within Bowman’s capsule in mice. Nephrology; 2015. J Pathol. 1985;145(4):315–27. 20. Lee R, et al. Localization of discoidin domain receptors in rat kidney. 30. Gautier L, et al. affy—analysis of Affymetrix GeneChip data at the probe Nephron. Exp Nephrol. 2004;97(2):e62–70. level. Bioinformatics. 2004;20(3):307–15. 21. Rubel D, et al. Collagen receptors integrin alpha2beta1 and discoidin 31. Subramanian A, et al. Gene set enrichment analysis: a knowledge-based domain receptor 1 regulate maturation of the glomerular basement approach for interpreting genome-wide expression profiles. Proc Natl membrane and loss of integrin alpha2beta1 delays kidney fibrosis in Acad Sci USA. 2005;102(43):15545–50. COL4A3 knockout mice. Matrix Biol. 2014;34:13–21. 32. Osumi-Sutherland D, Ponta E, Courtot M, Parkinson H, Badi L. Cell, chemi- 22. Fatima H, et al. Parietal epithelial cell activation marker in early recurrence cal and anatomical views of the gene ontology: mapping to a roche of FSGS in the transplant. Clin J Am Soc Nephrol. 2012;7(11):1852–8. controlled vocabulary. CEUR Workshop Proc. 2015;1456:84–93. 23. Stamenkovic I, Skalli O, Gabbiani G. Distribution of intermediate fila- 33. Zhang JD, et al. Detect tissue heterogeneity in gene expression data with ment proteins in normal and diseased human glomeruli. Am J Pathol. BioQC. BMC Genomics. 2017;18(1):277. 1986;125(3):465–75. Ready to submit your research ? Choose BMC and benefit from: fast, convenient online submission thorough peer review by experienced researchers in your field rapid publication on acceptance support for research data, including large and complex data types • gold Open Access which fosters wider collaboration and increased citations maximum visibility for your research: over 100M website views per year At BMC, research is always in progress. Learn more biomedcentral.com/submissions http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Journal of Translational Medicine Springer Journals

Loading next page...
 
/lp/springer_journal/selective-pharmacological-inhibition-of-ddr1-prevents-experimentally-LovpRd8oWo
Publisher
Springer Journals
Copyright
Copyright © 2018 by The Author(s)
Subject
Biomedicine; Biomedicine, general; Medicine/Public Health, general
eISSN
1479-5876
D.O.I.
10.1186/s12967-018-1524-5
Publisher site
See Article on Publisher Site

Abstract

Background: Discoidin domain receptor 1 (DDR1) is a collagen-activated receptor tyrosine kinase extensively implicated in diseases such as cancer, atherosclerosis and fibrosis. Multiple preclinical studies, performed using either a gene deletion or a gene silencing approaches, have shown this receptor being a major driver target of fibrosis and glomerulosclerosis. Methods: The present study investigated the role and relevance of DDR1 in human crescentic glomerulonephritis (GN). Detailed DDR1 expression was first characterized in detail in human GN biopsies using a novel selective anti- DDR1 antibody using immunohistochemistry. Subsequently the protective role of DDR1 was investigated using a highly selective, novel, small molecule inhibitor in a nephrotoxic serum (NTS) GN model in a prophylactic regime and in the NEP25 GN mouse model using a therapeutic intervention regime. Results: DDR1 expression was shown to be mainly limited to renal epithelium. In humans, DDR1 is highly induced in injured podocytes, in bridging cells expressing both parietal epithelial cell (PEC) and podocyte markers and in a subset of PECs forming the cellular crescents in human GN. Pharmacological inhibition of DDR1 in NTS improved both renal function and histological parameters. These results, obtained using a prophylactic regime, were confirmed in the NEP25 GN mouse model using a therapeutic intervention regime. Gene expression analysis of NTS showed that pharmacological blockade of DDR1 specifically reverted fibrotic and inflammatory gene networks and modulated expression of the glomerular cell gene signature, further validating DDR1 as a major mediator of cell fate in podocytes and PECs. Conclusions: Together, these results suggest that DDR1 inhibition might be an attractive and promising pharmaco- logical intervention for the treatment of GN, predominantly by targeting the renal epithelium. Keywords: Fibrosis, DDR1 inhibition, Glomerulosclerosis, CKD *Correspondence: marco.prunotto@roche.com Solange Moll and Yukari Yasui contributed equally to this work Present Address: Office of Innovation, Immunology, Infectious Diseases & Ophthalmology (I2O), Roche and Genentech Late Stage Development, 124 Grenzacherstrasse, 4070 Basel, Switzerland Full list of author information is available at the end of the article © The Author(s) 2018. This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creat iveco mmons .org/licen ses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creat iveco mmons .org/ publi cdoma in/zero/1.0/) applies to the data made available in this article, unless otherwise stated. Moll et al. J Transl Med (2018) 16:148 Page 2 of 20 the subtle podocyte staining detected with ISH, probably Background due to differential detection thresholds, showed a charac - Discoidin domain receptor 1 (DDR1) is a collagen-acti- teristic membranous localization of DDR1. DDR1 immu- vated receptor tyrosine kinase (RTK) [1] extensively nostaining was observed in the tubules of the cortex and implicated in diseases such as cancer [2], atherosclerosis of both outer and inner medulla, most likely in the distal [3] and fibrosis [3–5]. In humans, single nucleotide poly - nephron (Fig.  1b). Double immunostaining with specific morphisms of DDR1 are associated with susceptibility for tubular markers (Megalin, Calbindin, Tamm–Horsfall and disease progression of childhood IgA nephropathy Protein/Uromodulin and Aquaporin 2) on serial sections [6]. In multiple preclinical studies DDR1 has been shown confirmed DDR1 protein localization in the distal parts of to play a major role in the pathogenesis of fibrosis and the tubule (distal convoluted tubule, thick ascending limb glomerulosclerosis [7–13]. The evidence for a protective of the Henle’s loop/pars recta distal tubule, connecting role of DDR1 in glomerulosclerosis has been supported tubule and collecting duct) (Fig.  1c). No DDR1 staining so far by gene knockout (KO) experiments or the use of was detected in the interstitial space or in vessels, using antisense oligonucleotides (ASO), with both technolo- both ISH and IHC techniques. Taken together, these data gies relying on reduction of total DDR1 protein [1–5]. demonstrate that DDR1 expression in the normal human Although these techniques can demonstrate mechanism, kidney is exclusively limited to renal epithelial cells, PECs they have very limited translational potential. Knock-out and podocytes in the glomerulus and tubular cells in the mouse models mimic prophylactic regimens, which are distal part of the nephron. not applicable to patients suffering from a substantial loss of kidney function. ASOs, though tested in a therapeu- tic intervention regimen by one of the co-authors [10], DDR1 protein is abnormally induced during crescentic are predominantly cleared by the liver and kidneys, and glomerulonephritis in humans and is detected in cellular thus represent a non-preferred clinical scenario in renal- crescents impaired patients. Twenty-nine biopsies with cellular and/or fibrous cres - The present study includes results originally generated cents were obtained from patients with Goodpasture’s as part of a pharmaceutical program, aimed at the crea- syndrome (5 cases), ANCA-associated GN (7 cases), tion of a safe and effective DDR1 inhibitor to be used in IgA GN (12 cases) and lupus GN (5 cases: class III and patients affected by glomerulonephritis (GN) [14]. Firstly, IV according to the ISN/RPS classification) (Table  1: 14 we show in depth characterisation of DDR1 expression in males and 15 females, mean age 56 ± 21  years, mean normal human kidney and in renal biopsies from patients serum creatinine 199 ± 175  μmol/l, mean blood urea with crescentic GN, using a newly-developed highly spe- nitrogen (BUN) 11 ± 8 mg/dl, mean proteinuria 3 ± 3 g/l). cific anti-DDR1 antibody. These translational data were Twenty-five patients had biopsies performed for diag - essential to further enhance confidence in the initiation nosis of acute renal insufficiency and 4 for follow-up of and progression of a medicinal chemistry effort to gen - previously treated crescentic GN (patients #4, 25, 28 and erate an exquisitely selective and potent DDR1 inhibitor 29). Glomerulus number per biopsy varied from 7 to 54 (DDR1i). The characterisation of DDR1i in two differ - (mean 24 ± 15), globally sclerosed glomeruli from 0 to ent mouse models of GN [14], in both prophylactic and 23 (mean 2 ± 6) and crescents, either cellular or fibrous, therapeutic regimens, is presented here. Gene expres- from 1 to 30 (mean 9.7 ± 8.7). DDR1 immunostaining sion profiles of selective DDR1i-treated animals were also was evaluated in each crescentic glomerulus as posi- profiled in order to gain further knowledge regarding the tive or negative. DDR1 protein was detected in most of pathways and networks selectively modulated by drug the cellular crescents of all four types of crescentic GN targeting. The data suggest that DDR1 is an important (Table  1, Fig.  2A). DDR1 staining intensity was variable player in human GN and that its pharmacological inhibi- between crescents within the same biopsy and expression tion is translatable into a valid therapeutic intervention limited to a subset of cells forming the cellular crescent. tested in preclinical GN models. In addition to positive cellular crescents, DDR1 immu- noreactivity was detected in injured podocytes (Fig. 2A). DDR1 podocytes were observed in most of the glomer- Results uli, with or without crescent lesion. Furthermore, some of DDR1 is exclusively expressed in renal epithelium these podocytes appeared to adhere to both the glomeru- under physiological conditions lar basement membrane and the parietal basement mem- DDR1 mRNA and protein expression is restricted to brane, forming podocyte bridges between the glomerular the glomerular parietal epithelial cells (PECs) of the tuft and the Bowman’s capsule. All fibrous crescents were Bowman’s capsule and to podocytes and some tubules DDR1 negative. It should also be noted that DDR1 stain- (Fig.  1a). Immunohistochemistry (IHC) failed to reveal ing was increased in distal tubules and detected in some Moll et al. J Transl Med (2018) 16:148 Page 3 of 20 Fig. 1 Localisation of DDR1 in human control kidney. a DDR1 in situ hybridization (ISH) and immunohistochemistry (IHC) analyses in normal human kidney. (Scale bar = 100 μm). ISH higher magnification rectangle: DDR1 positive podocyte (black arrowheads), DDR1 positive podocytes (green arrowheads). b DDR1 immunostaining in the nomal human kidney with representative images of the cortex, outer medulla (outer and inner stripes) and inner medulla. Note the DDR1 membranous staining of the tubules. (Magnification ×125, scale bar 50 μm). c Serial sections of normal human kidney were double immunostained with DDR1 and specific tubular antibodies against Megalin, Calbindin, Tamm–Horsfall Protein ( THP or Uromodulin) or Aquaporin 2 (AQP2). c1 Representative micrographs of the cortex show DDR1 protein localization in the distal convoluted tubules (Calbindin+, Aquaporin 2−, Megalin−). Proximal tubules (Megalin+) are DDR1 negatively stained. Boxed areas are enlarged in the right side of the figure. c2 Representative micrographs of the medullar rays of the cortex show DDR1 protein localization in the connecting tubules (1: Calbindin+, Aquaporin 2+, THP−), distal convoluted tubules (2: Calbindin+, Aquaporin 2−, THP−), thick ascending limbs of the Henle’s loop/pars recta of the distal tubules (3: THP+, Calbindin−, Aquaporin 2−) and collecting ducts (4: Aquaporin 2+, Calbindin−, THP−). Boxed areas are enlarged in the right side of the figure. Magnification ×100, scale bar = 100 μm; Boxed areas Magnification ×250, scale bar = 50 μm cases, particularly SLE cases, in both distal and proximal glomerular capillary lumen and only 2–3 layers of cells tubules as depicted in Fig. 2A. within the Bowman’s space, DDR1 staining was detected To define more precisely the profile of DDR1 expres - in numerous cells of the glomerulus. These cells were sion during the sequence of morphological changes most likely injured podocytes, indicated by similar nestin observed in crescent formation, we performed IHC on immunostaining and by abnormal morphologic aspects lupus nephritis class IV-G (A/C) biopsies. Lupus nephri- (Fig.  2B, C line a). It should be noted that the DDR1 / tis biopsy samples can display glomerular lesions that are Nestin cells, localized anterior to the emerging cellullar morphologically heterogeneous, consistent with vari- crescent, were also slightly positive for the PEC markers ous stages of crescentic lesions. Furthermore, crescent CK 8–18 and CK 19 (Fig.  2C line a). At ther very initial formation results from the serial participation of several stage of the cellular crescent formation, characterized by different cell types, including macrophages, glomerular gaps in the glomerular capillary wall and plasma proteins parietal epithelial cells, glomerular visceral epithelial cells and cells within Bowman’s space, DDR1 staining was also (podocytes), renal progenitor cells and interstitial fibro - detected in injured podocytes, but with a weaker stain- blasts [15]. With this in mind, serial sections of lupus ing intensity (Fig.  2B, C line b). It should be noted that nephritis biopsies were immunostained with DDR1 the strongest DDR1 signal was observed in podocytes and four different cellular markers: the monocyte/mac - localized in front of the gap in the glomerular capillary rophage marker CD68, the podocyte marker nestin, wall, near the Bowman’s capsule or near the growing cel- + + and the PEC markers cytokeratin (CK) 8–18 and CK 19 lular crescent. Some of these DDR1 /Nestin cells, also (Fig.  2B, C). In the very initial stage of cellular crescent positive for CK 8–18, seemed to adhere to both the glo- formation, characterized by fibrin deposition within the merular basement membrane and the PECs, forming Moll et al. J Transl Med (2018) 16:148 Page 4 of 20 Table 1 Clinical, biological and  histo-pathological data from  the  29 patients with  crescentic glomerulonephritis and  DDR1 immunostaining evaluation of the crescents as positive or negative Patient nb Clinical data Histological data Diagnostic Age/gender SCreat (µmol/l) eGFR ml/ BUN mg/dl Proteinuria (g/l) Hematuria Glomeruli Cellular crescents Fibrous crescents TIF % min/1.73 m Total Globally DDR1+ DDR1− DDR1+ DDR1− sclerosed 1 Goodpasture 81/F 478 8 18.9 15 pos 8 0 5 0 0 0 25 2 85/F 776 4 32.7 3.68 pos 37 23 3 0 0 2 75 3 22/M 113 > 60 5.3 1.54 pos 9 0 3 0 0 0 0 4 21/M 196 38 10 0.64 pos 14 0 7 0 0 2 10 5 72/M 117 98 10 9 pos 42 0 3 0 0 1 0 6 ANCA 82/F 678 15 26.2 2.5 pos 54 20 0 27 0 3 0 7 74/M 312 16 25.3 0.58 pos 32 0 27 1 0 2 35 8 78/F 84 58 4.3 1.39 pos 9 2 0 1 0 0 0 9 66/F 165 29 ND 0.4 neg 27 0 11 4 0 2 55 10 66/M 311 19 ND 0.6 pos 17 0 7 0 0 0 20 11 70/M 202 28 12 2.09 pos 12 1 1 0 0 3 0 12 68/M 158 40 ND 1.8 pos 24 0 10 0 0 3 30 13 IgA 45/F 88 67 ND 7 neg 28 1 19 1 0 4 40 14 84/F 310 12 11 4.93 pos 7 0 2 0 0 1 10 15 56/M 94 > 60 3.3 0.3 pos 15 0 8 0 0 0 20 16 39/F 135 43 11 1.78 pos 21 8 5 0 0 0 50 17 43/F 56 > 60 4.8 0.6 pos 17 3 3 0 0 0 30 18 74/M 180 34 15 0.7 pos 14 1 0 0 0 1 50 19 19/F 62 > 60 4.5 3 pos 44 0 12 0 0 0 15 20 19/F 57 > 60 3 1.0 pos 34 0 2 0 0 0 10 21 23/M 91 > 60 5.9 0.5 pos 52 0 7 0 0 0 0 22 64/M 99 > 60 5.1 0.4 pos 8 0 4 0 0 0 45 23 72/F 250 17 ND 2 pos 12 0 4 0 0 1 30 24 64/F 108 47 12 1.88 pos 9 0 0 1 0 1 0 25 Lupus 37/F 159 36 8.5 8.0 pos 27 0 16 3 0 3 0 26 40/F 107 56 10.2 4.05 pos 19 0 6 0 0 4 30 27 44/M 180 38 ND 4.6 pos 14 0 3 0 0 4 40 28 51/M 97 > 60 7.6 5.0 pos 47 1 9 0 0 2 10 29 52/M 102 > 60 6.88 5.0 pos 45 2 20 0 0 6 20 Moll et al. J Transl Med (2018) 16:148 Page 5 of 20 (See figure on next page.) Fig. 2 DDR1 is highly induced in crescentic glomerulonephritis. A DDR1 expression in 4 different types of human crescentic glomerulonephritis. DDR1 immunostaining in human kidney biopsies from patients with 4 different types of crescentic glomerulonephritis (GN): Goodpasture’s syndrome (patient #1), ANCA-associated GN (patient #12), IgA GN (patient #19) and lupus GN class IV-G (A/C) (patient #29). DDR1 is expressed in the cellular crescents of all 4 types of crescentic GN with variability in the staining intensity and expression limited within crescents to a subset of cells. Injured podocytes and bridging podocytes (arrows) strongly express DDR1. Fibrous crescent (*) are DDR1 negative. Magnification ×200 for SLE, ×250 for ANCA, ×320 for Goopasture and IgA, Scale bar = 20 μm. B–C Detailed analysis of DDR1 expresion during crescent formation. Serial sections of human kidney biopsies from patients with lupus GN class IV-G (A/C) (patients #25 and #29) were immunostained with specific antibodies against CD68/PAS (Periodic Acid Schiff ), DDR1, nestin, cytokeratin 8/18 or cytokeratin 19 proteins. B Representative micrographs show expression and localization of each protein in the glomeruli. Boxed areas represent crescentic lesion and are enlarged in C. Lines a–d illustrate 4 different morphological stages of the crescent formation; a: early stage with fibrin within the glomerular capillary lumen and presence of 2–3 layers of cells in Bowman’s space; b: early stage with gap in the glomerular capillary wall, plasma proteins in Bowman’s space and cellular crescent formation; c: active hypercellular circumferential crescent compressing the glomerular tuft; d: advanced stage with fibrocellular crescent, capsular rupture and periglomerular inflammation. Line a = patient #25; lines b–d = patient #29. Fine arrows = injured podocytes; large arrows = bridging cells; * = fibrin and plasma proteins within the glomerular capillary lumen or in Bowman’s space. Magnification A line a: ×320, b and c: ×250, d: ×200; B line a: ×640, b and c: ×500 d: ×400; Scale bar = 20 μm podocyte bridges between the glomerular tuft and Bow- similarly to human GN, DDR1 appears to be predomi- man’s capsule (Fig.  2C line b). In a well formed cellular nantly expressed in epithelial cells in the NTS mouse crescent characterized by multiple layers of cells within model. the Bowman’s space, DDR1 expression was readily To address the functional significance of DDR1 in detected within a subset of cells of the crescent, mostly experimental crescentic GN, mice were pre-treated with located in the periphery of the glomerular tuft (Fig.  2B, two doses of a very selective DDR1 inhibitor (Roche-Chu- C, line c). These DDR1 cells positively immunostained gai DDR1i), followed by NTS administration. To unravel with both PEC markers CK 8/18 and CK 19, but were the specificity of DDR1 activation blockade, a second nestin negative. In a more advanced stage, characterized group was treated with the kinase inhibitor imatinib, by fibro-cellular crescent, Bowman’s capsule rupture and which was previously shown to elicit a marked renopro- periglomerular inflammation, DDR1 immunostaining tective effect in the NTS-induced model in Wistar-Kyoto was detected in the cellular compartment of the cres- rats [16]. All functional parameters were significantly cent but not in the fibrous area (Fig.  2B, C, line d). A improved with the high dose of DDR1 inhibitor. The low similar staining was observed with CK 8/18 and CK 19. dose was less efficient, whereas Imatinib failed to show a Finally, it should be noted that in severe necrotic lesions significant protection (Fig.  3d–g). Histological analyses of as observed in patient #6 and #8 with ANCA-associated kidney sections followed by semiquantitative histopatho- GN (Table 1), no DDR1 staining could be detected (Addi- logic evaluation with glomerular and tubulo-interstitial tional file  1: Figure S3). This result is not surprising given (TI) scores revealed that functional kidney protection in the degree of structural damage in these glomeruli. the high dose DDR1i group was paralleled by tissue pres- Taken together, our data suggest that DDR1 is de novo ervation of both glomerular and tubulo-interstitial scores expressed during crescent formation, predominantly (p = 0.02 and p = 0.01 respectively, Fig. 3h–i). In contrast, in injured podocytes, in bridging cells expressing both none of these findings were observed in the Imatinib- podocyte and PEC markers, and in a subset of PECs treated group (glomerular and tubulo-interstitial scores: forming the cellular crescent. p = 0.92 and p = 0.93 respectively). Concordant with glomerular histology, IHC staining with the PEC mark- DDR1 expression increases in experimentally‑induced ers desmin and CD44 revealed marked reductions in the crescentic glomerulonephritis and its pharmacological DDR1i HD group (Fig.  3h). Likewise, collagen type IV prophylactic inhibition confers morphological staining showed a marked reduction of immunoreactivity and functional protection in the interstitial areas in the DDR1i HD group, but not First, we confirmed that DDR1 expression is induced fol - in the DDR1i LD and Imatinib groups, which were both lowing NTS administration (Fig. 3a–c). Consistently, ISH comparable to the vehicle group. A similar trend was showed strong DDR1 upregulation in crescents, PECs, observed in the case of collagen type I (data not shown). and in the tubular structures, most of them display- Morphometric analysis of collagen type IV staining ing lesions characterized by flattening of the epithelium showed a marked and statistically significant reduction and dilation of the lumen (Fig.  3b). No DDR1 staining of immunoreactivity in the DDR1i HD group (33.1% ± 3.7 was detected in cells co-labeled with Acta2 (myofibro - versus 63.9% ± 1.7 for vehicle-treated group, p < 0.001) blasts) or Emr-1 (macrophages) (Fig.  3c), indicating that (Fig.  3j). No significant reduction in collagen type IV Moll et al. J Transl Med (2018) 16:148 Page 6 of 20 Moll et al. J Transl Med (2018) 16:148 Page 7 of 20 (See figure on next page.) Fig. 3 Pharmacological inhibition of DDR1 activation protects animals against NTS-induced crescentic glomerulonephritis. a Quantitative RT-PCR for Ddr1 mRNA on whole kidney lysate of control mice (Control) and mice injected with nephrotoxic serum and treated with vehicle (vehicle). b Representative Ddr1 in situ hybridization (ISH) performed on tissue harvested from mice 14 days after NTS injection. * = crescent c representative DDR1 ISH double labelling with alpha smooth muscle actin (Acta2) or EGF-like module-containing mucin-like hormone receptor-like 1 (Emr-1) in control mice (Control) and mice injected with nephrotoxic serum and treated with vehicle ( Vehicle). Arrows = cells labeled with Acta2 or Emr-1. d–g Body weight evolution (day 1, 4, 7 and 14) and renal function parameters (e–g blood urea nitrogen (BUN), serum creatinine and proteinuria) measured at sacrifice (day 14). h Representative histopathology with Hematoxylin and Eosin (H&E) and Periodic Acid Schiff staining (PAS) and immunohistochemistry for desmin, CD44, Collagen type IV and Ki67. i Glomerular or tubulo-interstitial ( TI) summary scores from semiquantitative histopathologic evaluation on H&E and PAS stained kidney sections respectively. j Morphometry analysis of collagen IV IHC. Statistically significant p value: p < 0.05 = *; p < 0.005 = **. Magnification ×200, scale bar 100 μm staining was observed in the other treated groups. Finally, of genes related to immune response, integrin pathway in agreement with the histology data, quantitation of activation and fibrotic processes (TGFβ pathway), all the cellular proliferation marker Ki67 showed a marked increased in the NTS group (Fig. 5a). In contrast, tissues reduction of PEC and tubulo-interstitial staining in the from mice treated with DDR1i HD showed a significant DDR1i HD and DDR1i LD groups (Fig. 3h). reduction in the expression of this set of genes (Fig.  5a). This effect was also observed, albeit to a lesser extent, in Selective DDR1 inhibition protects when tested using the DDR1i LD group, but was not in the Imatinib group. a therapeutic intervention regime Noteworthy, among the weakly perturbed gene signa- We then aimed to assess the effect of DDR1 inhibition tures, GSEA analyses revealed that the glomerular gene in the context of progressive glomerulosclerosis in the signature, driven by synaptopodin, podocalyxin, NPHS1 NEP25 mouse model [14]. Mice were treated with the and NPHS2 genes, was clearly preserved in the DDR1i Roche-Chugai DDR1i, Captopril or vehicle (Fig.  4a). HD group, whereas no protection was observed in the Histological and semi-quantitative analyses of glomeru- case of the Imatinib treated group (Fig. 5b). losclerosis in PAS sections and tubulo-interstitial fibro - Comparison of DDR1i GSEA results with a previous sis/inflammation in H&E sections of kidney tissue at datasets generated in the NTS mouse model treated day 15 showed a significant reduction of glomerular with other target inhibitors revealed a consistent pattern PAS positive area in both DDR1i and Captopril groups of biological processes associated with NTS injection (p < 0.001 and p < 0.0001 respectively) (Fig.  4B, C). A that were selectively reverted by DDR1i (Fig.  5c). These slight, but not significant, reduction in tubulointersti - include reactivation of genes involved in glomerular epi- tial damage was observed in the DDR1i-treated group, thelial cell processes and deactivation of genes related to whereas this reduction was significant in the Captopril wound healing, cell matrix, cell adhesion, and motility group (p < 0.001) (Fig.  4c). Analyses of several fibrosis and inflammation processes. Thus, to identify the main and inflammation markers by qRT-PCR in renal cortical driver genes specifically modulated by DDR1i treatment, tissues at day 15 showed that DDR1i induced a marked we conducted additional gene level analysis. These data reduction of alpha smooth muscle actin (p < 0.01), colla- revealed that among the statistically significantly changed gen type 1, TGF-β1 (p < 0.01) and of Ccl2 (p < 0.001) in a genes (n = 785), 175 were identified with a large effect range similar to that observed in Captopril-treated mice size (logFC ≥ 2) and reverted in the DDR1i group as com- (Fig. 4d). Captopril was more efficient in improving renal pared to the NTS group (Additional file  2: Table S3). The function (Fig. 4e, f )due to the fact that Captopril admin- highest fraction of perturbed genes (n = 151) was rep- istration started before disease initiation (preventive resented by genes exhibiting increased expression upon approach), whilst the DDR1 inhibitor was only adminis- NTS injection. Genes regulated by treatment with DDR1i tered during the progression of nephropathy (curative or were hallmarks in the induction of inflammatory pro - interventional approach). cesses, and regulation of cell matrix and adhesion mol- ecules, cell cycle, and apoptosis. DDR1 inhibition modulates specific gene networks An additional classification criteria (detectable expres - in experimentally‑induced crescentic glomerulonephritis sion in the human kidney epithelial cell line HKC8 [17] or In order to unveil the mode of action (MoA) of DDR1i, PEC [18] was adopted to further narrow down the gene mouse kidneys from the NTS experiment were subjected list (n = 45) to those mechanistically-related to DDR1. to gene expression profiling (GEP, GEO Accession Num - To remove model-specific gene perturbations and ber GSE104426). Unsupervised gene signature analysis identify DDR1 MoA, the differentially expressed genes and Gene Set Enrichment Analysis (GSEA) using Gene obtained in the NTS GEP analysis were compared to the Ontology processes as input gene sets showed networks genes significantly perturbed in the NEP25 mouse model Moll et al. J Transl Med (2018) 16:148 Page 8 of 20 Moll et al. J Transl Med (2018) 16:148 Page 9 of 20 Fig. 4 Treatment with DDR1 inhibitor in the NEP25 mouse model of glomerulosclerosis. a Schema of the experiment b Representative histopathology with Hematoxylin and Eosin (H&E) and Periodic Acid Schiff staining (PAS) in control, vehicle-, DDR1i- and Captopril treated groups at day 15. Magnification ×200, scale bar 100 μm. c Semi-quantificative analysis of glomerulosclerosis (glomerular PAS positive area) and tubulointerstitial lesions (tubulointerstitial damage) in control, vehicle-, DDR1i- and Captopril treated groups at day 15. ***p < 0.001, ****p < 0.0001; t-test and Mann–Whitney U test were used for the score of glomerular PAS positive area and tubulointerstitial damage respectively. d Quantitative RT-PCR for the fibrosis markers alpha smooth muscle actin (Acta2 mRNA), collagen type 1 (Col1a mRNA) and TGF-β1 ( TGF-β1 mRNA), and for the inflammation marker Ccl2 in NEP25 mice treated with DDR1 inhibitor (DDR1i), Captopril or vehicle and in control mice. *p < 0.05, **p < 0.01, ***p < 0.001; t-test. e Body weight evolution. f Renal function parameters (plasma creatinine and plasma Cystatin C measured at sacrifice) and urinary ACR (24-h urine collection from day 14 to 15 divided by creatinine concentration) in NEP25 mice treated with DDR1 inhibitor (DDR1i), Captopril or vehicle and in control mice (CTRL) Moll et al. J Transl Med (2018) 16:148 Page 10 of 20 (See figure on next page.) Fig. 5 Gene expression profile in NTS-treated mice. a Unsupervised gene signature analysis and GSEA using Gene Ontology processes as input genesets. b Glomerular cell signature with individual component genes. c Comparison of different target modulation in NTS. Heatmap showing differential modulation (by DDR1i or Target X) of the gene networks modulated by NTS induction compared to control. d Heatmap showing kinome expression modulation in the different treatment conditions Previously, DDR1 has been localized in rat kidney [20] that were reverted by DDR1i (n = 85). The final intersec - and our data extend that knowledge to humans providing tion revealed 30 genes in common to both models (Addi- further confidence for the target, indicating that DDR1 tional file 2: Table S4). expression is restricted to the epithelial compartment Finally, the unexpected lack of therapeutic effect of (with no DDR1 expression in vessels as reported with the Imatinib in the NTS model prompted us to conduct use of unselective antibodies [12]) and providing, for the a specific kinome analysis to identify potential differ - first time, a detailed analysis of the role of DDR1 in the ences in kinase modulation between DDR1i and Imatinib sequential morphological and cellular events leading to groups, which could expain the differential therapeu - crescent formation in humans. DDR1 was expressed (1) tic response. Best BUN-correlated kinases (with a cor- in most injured podocytes during the very initial stage relation coefficient better than 0.7) were retained and of the disease characterized by glomerular intracapillary plotted (Fig.  5d). These analyses revealed that 6 kinases immunoinflammatory processes, (2) in bridging cells co- were specifically upregulated by DDR1i HD treatment expressing podocyte and PEC markers located in front in comparison to Imatinib. In contrast, 19 kinases were of, or adjacent to, disruption of the integrity of the glo- downregulated. merular capillary wall, and (3) in a subset of activated/ Taken together, our data demonstrated that treatment injured PECs forming the cellular crescent, most of these with Roche-Chugai DDR1i reverts inflammatory and cells being located near the glomerular tuft. Based on this fibrotic genes and positively affects genes of the glomeru - expression analysis, it could be speculated that DDR1 lar cell signature particularly podocytes and with an evi- expressed during the initial stage of the crescentic disease dent effect on the expression of specific kinases. in podocytes and bridging cells is reparative but then, when expressed during the active and severe stage of the DDR1 inhibition decreases PEC pro‑inflammatory disease in proliferative PECs is deleterious, participating phenotype in vitro to excessive maladaptive repair processes. We then aimed to asses to characterize the impact of These findings, in addition to previous preclinical DDR1 inhibition on PEC phenotype using a murine evidence of the pathogenic role of DDR1 in crescentic PEC cell line [18]. Pre-treatment of PECs for 1  h with GN [6, 8–10, 12, 13, 21] induced our research group to DDR1i followed by stimulation with collagen type I for develop a highly selective DDR1i. Selectivity of DDR1 24  h blocked DDR1 phosphorylation in a dose depend- over the close analogue protein DDR2 is of utmost ent manner (Additional file  1: Figure S6). Treatment with importance to harness the therapeutic potential of DDR1 type I collagen alone induced a significant increase of the antagonism. DDR2 inhibition has in fact been associated matrix metalloproteinases Mmp14 and Mmp2, comple- with enhanced liver fibrosis (evidence generated by the ment component C3 and vascular cell adhesion molecule renowned group of prof. Scott Friedman [6]) and DDR2 1 (VCAM1) mRNA (Additional file 1: Figure S6B-E). Pre- activation has been shown to inhibit the development of treatment of the cells with DDR1i caused a dose-depend- experimental choroidal neovascularization and retinal ent inhibition of expression of these genes. neovascularization in mice [7]; thus, inhibition of DDR2 could translate into undesirable side-effects in patients. Discussion None of these undesirable side effects were noted in ani - The present study details DDR1 expression in human GN mals treated with the selective DDR1i. biopsies and demonstrates, for the first time using phar - Treatment with our selective DDR1i resulted in pre- macological intervention, that selective DDR1 inhibi- served renal function and structure, as suggested by tion translates into renal protection. These data confirm histology, IHC and morphometry analyses. Glomeru- the central role of DDR1 in glomerular diseases, previ- lar lesions were reduced, in particular the number of ously indicated using gene deletion [8, 12, 13, 19] or gene activated PECs as shown by reduced Desmin and CD44 silencing paradigms [10]. [22] positivity, both recognized markers of an activated Analysis of human renal biopsies with different types of PEC phenotype [22, 23]. Tubulo-interstitial inflamma - crescentic GNs shows DDR1 to be strongly expressed in tion and fibrosis were decreased and the architecture of injured podocytes and in PECs forming cellular crescents. Moll et al. J Transl Med (2018) 16:148 Page 11 of 20 Moll et al. J Transl Med (2018) 16:148 Page 12 of 20 renal epithelium was maintained. The use of a second DDR1 inhibition blocks activation of PEC in vitro. Taken model, the NEP25 mouse model [14], was prompted by together, the in  vivo and the in  vitro evidence seem to our interest to further assess the role of DDR1 in a injury suggested that PEC de-activation could be a key part of model where PEC-induced glomerulosclerosis is caused DDR1i MoA. by podocyte depletion in absence of intra-glomerular One of the major strengths of the present work is that inflammation, primary mesangiolysis and primary hyper - for the first time pharmacological intervention target - tension thus further strengthening the role of DDR1i in ing predominantly the renal epithelium can reduce both the protection of glomerular function also in a model inflammation and fibrosis. These effects seem to occur of focal segmental glomerulosclerosis. Though not as without directly affecting fibroblast to myofibroblast acti - impressive as in the case of NTS, where DDR1i was vation or the inflammatory component as DDR1 expres - administrated in a prevention mode, these findings sug - sion was not observed in the relevant cell types. We are gest that selective DDR1 inhibition is protective when however conscious that the lack of immunohistochemi- used in the context of an established lesion (therapeutic cal staining does not exclude these additional potential regime). effector cells, highly relevant to the two models studied. The choice of two different mouse models to investi - Our hypothesis is concordant with a recently published gate the relevance of DDR1 in GNs was consciously made study demonstrating that selective activation of EGFR to avoid a model-biased interpretation of the results. in renal epithelial cells (proximal tubules) is sufficient to Since the therapeutic regime was already probed, by one induce tubulo-interstitial fibrosis [24]. As shown by our of the co-authors, in the NTS model [19] we chose to molecular genomics analysis of NTS kidney samples, generate data in the alternative model of NEP25 [10] to DDR1 antagonism has a direct effect in the preserva - test if treatment could transfer to protection in a differ - tion of the podocyte network and in the de-activation of ent experimental setup. The NEP25 mouse model proved PECs. It is possible that DDR1 antagonism could reduce to be extremely robust and was characterized by uniform an excessive “response to injury” occurring within the glomerular injury and sclerosis induced by a simple pro- glomerular compartment in case of crescentic lesions, cedure, i.e., a single intravenous injection of LMB2 toxin. and having deleterious consequences on tubulo-intersti- Indeed the model perfectly fits a drug discovery program. tial compartment. Those speculations will have to be fur - Genomic analysis of the NTS experiment, performed ther addressed by deeper mechanistic studies critical to in the context of a drug discovery program, aiming to comprehend the role of DDR1 in glomerular repair and identify biomarkers exploitable in clinical studies allowed regeneration processes. us to suggest that the observed functional and organ pro- Moreover, we can’t ignore that our IHC data seem to tection is due to reduced inflammation and fibrosis. suggest a broader role for DDR1 in kidney. We have in To specifically decode DDR1 molecular MoA, we took fact observed a clear DDR1 tubular upregulation in advantage of data generated in the NTS mouse model human crescentic GN as well as in mouse models of with compounds modulating other pathways (not DDR1- glomerular disease. It is interesting to note that DDR1 dependent) and used them to filter and identify a consist - genetic ablation in other models of chronic, acute or ent pattern of biological processes associated with renal genetic renal disease, such as hypertensive nephropathy, injury and specifically reverted by DDR1 antagonism. Alport’s syndrome and unilateral ureteral obstruction, Gene and pathway level analysis identified DDR1 driver has been shown to be protective [3, 19]. Causal asso- genes associated with cell cycle and cell matrix and adhe- ciation between epithelial cell injury and renal function sion, some amongst them were expected for example dif- deterioration, paralleled by fibrosis might be an integral ferent collagen genes (Col1A1, Col1a2, Col3a1, Col6a2), part of the DDR1 MoA in GN. Certainly, this hypothe- or other extracellular matrix related proteins such as sis deserves further investigation using dedicated mouse TIMP1 and FN1, but also a marked reduction of LTBP2 models and a detailed IHC characterisation in particu- suggestive of a reduced state of activation of epithelial lar in the context of tubulopathies, acute renal ischemic cells. Interestingly, Cd44, a hallmark of activated PECs lesion or transplant reperfusion and/or in general in was observed among the top DDR1i modulated genes. other ischemic conditions. Surprisingly, unsupervised gene signature analysis iden- It should also be mentioned that when designing the tified podocyte-specific genes as being clearly protected preclinical experiments we sought to add control com- by DDR1i treatment. This is quite remarkable since GEP pounds to validate the models adopted. For that reason, were generated using whole kidney lysates. These molec - the control compounds (Imatinib and Captopril, respec- ular findings are concordant with the IHC results dem - tively used in the NTS and NEP25 models) were admin- onstrating PEC de-activation. To support that notion, istered in a prevention mode to insure better protective we also performed a series of experiments showing that performance and consequently better model validation. Moll et al. J Transl Med (2018) 16:148 Page 13 of 20 The doses of both compounds were selected using data approved if the ratios 260/280 and 260/230 were both reported in the literature for the selected models. Lack greater than 1.8. of efficacy of Imatinib was surprising since efficacy has DDR1 splice isoform specific qRT-PCR assays have been reported at the selected dose [25]. Theoretically this been developed for DDR1-variant 3 (DDR1_V3), DDR1- dose has in vivo exposure similar to the DDR1 HD group variant 4 (DDR1_V4) and DDR1-variant 5 (DDR1_V5), based on reported in vitro DDR1 potency [26] and in vivo DDR1-variant 6 (DDR1_V6) which specifically detect pharmacokinetic data (due the better bioavailability and one variant. Due to close sequence homology PCR assays reduced clearance of Imatinib compared to the DDR1 for DDR1-variant 1/2 (DDR1_V1/2), DDR1-variant 1/6 HD). The lack of efficacy of Imatinib, might suggest that (DDR1_V1/6) detect both isoform (see Additional file  3: selective inhibition of DDR1 and not of a broader panel for primer/probe sequences). All PCR assays were devel- of kinases might be a crucial element to achieve kidney oped using synthetic plasmids carrying DDR1 variant protection. Comparative kinome analysis between DDR1i specific sequences assuring comparable sensitivity, speci - and Imatinib extracted from our GEP indicates that lack ficity, PCR amplification efficiency (see Additional file  3: of efficacy of Imatinib might possibly due to antagonism for plasmid sequences). of kinases whose activation is key to confer protection to Reverse Transcription of RNA was accomplished using injury, such as STK16, INSSR or CDK7. This hypothesis the SuperScript III First-Strand Synthesis Kit (Invitrogen) might also suggest an interesting role for those kinases in according to the manufacturers’ protocol with a total GN. RNA input of 100  ng per reaction. cDNA samples were then diluted 1:3 in TE buffer and 14 cycles of pre-amplifi - cation were carried out using 2× TaqMan PreAmp Mas- Conclusion ter Mix (Applied Biosystems) and pooled DDR1 splice In conclusion, in the present work, pharmacological inhi- variant specific assays at a final concentration of 0.2× bition of DDR1 phosphorylation emerges as a unique and per assay. The following thermocycler program was used: differentiated pharmacological therapeutic intervention 95 °C for 10 min, followed by 14 cycles at 95 °C for 15 s capable of preserving renal function in both prevention and 60 °C for 4 min. Pre-amplified cDNA products were and therapeutic regimens, acting on the renal epithelium diluted 1:5 in TE buffer. qPCR was performed using the and resulting in reduced inflammation and fibrosis in the 96.96 dynamic array (Fluidigm Corporation, CA, USA) surrounding interstitial space. Such an intervention para- following the manufacturer’s protocol (Fluidigm Quick digm, when available in the clinics, would provide a novel Reference Card, PN 68000130, Rev. B). Briefly, for each pharmacological tool for combination therapy with anti- sample a 5μl sample mix was prepared with 1× GE inflammatory, endothelial protective or myofibroblast Sample Loading Reagent (Fluidigm), 1× Taqman Gene blocking agents. Expression Mastermix (Applied Biosystems) and diluted, pre-amplified cDNA. For the assay mix 1× Assay Loading Reagent (Fluidigm) was mixed with each of the Taqman Methods Assays (final concentration: 10×), respectively. Priming DDR1 isoform RT‑PCR of the Fluidigm array with control line fluid and mixing RNA was purified from normal, ADPKD and ESRD tis - of sample and assay reagents was done with an IFC con- sue sections using a Tissue Lyser protocol (Qiagen) by troller. qPCR was performed using the BioMark Instru- adding 350  µl ‘RLT buffer’ plus 350  µl 70% ethanol and ment with the following cycling parameters: 95  °C for vortexing. The sample was then transferred to an RNeasy 10 min, followed by 40 cycles at 95 °C for 15 s and 60 °C column and centrifuged at 8000×g for 15 s at room tem- for 1 min. Data was collected and analyzed with the Real- perature. RNase free DNase 1 (80  µl) was added to the Time PCR Analysis Software (Fluidigm Corporation, CA, column and incubated at room temperature for 15  min. USA). Normalization was performed using geometric The column was then washed by adding 350 µl RLT buffer mean expression of housekeeping genes B2M, GAPDH, and centrifuged at 8000×g for 15 s at room temperature. GUSB: ∆Cq = Cq gene of interest—Cq geomean House- This was repeated with ‘RPE buffer’ (500 µl) twice where keepers. Normalized values were transformed into rela- the second spin was prolonged to 2  min. Following this tive expression levels using (POWER; -Cq) calculation. the RNaeasy column was transferred to a new micro- The sum of all DDR1 splice isoform expression levels was tube and centrifuged at 11,000×g for 1  min. The RNAe - set at 100% and DDR1 splice isoforms values calculated in asy tube was once again transferred to a new microtube % accordingly. DDR1 splice isoform variant 1 and variant and the RNA eluted by adding 50 µl H O and centrifuga- 2 isoform expression was calculated using by subtract- tion at 8000×g for 1 min. The elution wash was repeated ing results of DDR1-variant 6 (DDR1_V6) from DDR1- once and the eluate pooled. Absorbance was measured at variant 1/6 (DDR1_V1/6) to obtain variant DDR1-variant 230, 260 and 280 nm in a spectrophotometer and quality Moll et al. J Transl Med (2018) 16:148 Page 14 of 20 1 (DDR1_V1) results which were then used to calculate supplements were purchased from Thermo Fisher Sci - DDR1-variant 2 (DDR1_V2) results. entific (Waltham, MA). Eight-well Permanox Lab-Tek chamber slides were purchased from NUNC (Cat.# 177445). ImmPRESS Reagent (Anti-Rabbit Ig) (Cat. # DDR1 in situ hybridization on mouse and human kidney MP-7401) and ImmPACT DAB Peroxidase Substrate sections (Cat. # SK-4105) were purchased from Vector Labora- In situ hybridization (ISH) on 4  μm-thick mouse or tories (Burlingame, CA). A rabbit monoclonal antibody human kidney sections was performed using the against DDR1 (D1G6) was purchased from Cell Signal- RNAscope 2.5 HD Duplex Assay from Advanced Cell ing Technology (Danvers, MA) and a monoclonal anti- Diagnostics (California, USA) and the DDR1 probes body against β-actin was purchased from Sigma (St. according to the manufacturer’s instructions: mouse Louis, MO). Mayer’s hematoxylin (Cat.# HMM999) DDR1 (RNAscope Probe-Mm-Ddr1-C2) and human was purchased from Scytek and Crystal Mount Aque- DDR1 (RNAscope Probe- Hs-DDR1-C2, target region ous Mounting Medium (Cat.# C0612) from Sigma 285—2016, #593591-C2). Double ISH was performed (St. Louis, MO). Immunoblot analyses and cell immu- on mouse kidney sections using mouse DDR1 probe and nostaining were performed to assess antibody specific - mouse Acta2 (RNAscope Probe- Mm-Acta2-C2) or ity and sensibility. mouse Emr1 (RNAscope Probe- Mm-Emr1). Images For immunoblot analyses, cells were lysed in RIPA were acquired with an Olympus VS120 scanner (Olym- buffer (50  mM Tris–HCl, pH 7.4, 150  mM NaCl, 1% pus AG, Switzerland) equipped with a VC50 camera and NP-40, 0.25% sodium deoxycholate and 1.0  mM EDTA) 20× objective. supplemented with protease inhibitors on ice for 1  h. The cell lysates were cleared by centrifugation at 14,000g Generation of a selective anti‑DDR1 monoclonal antibody at 4 °C for 15 min, and protein concentration was deter- (DDR1 Rab‑819) for IHC analysis mined using the BCA kit (Pierce). 20–40  µg of protein New Zealand White (NZW) rabbits were used for immu- lysate were mixed with 1× reducing Laemmli SDS-sam- nization with extracellular domain (ECD) DDR1 protein ple buffer. After a brief centrifugation, the supernatants (R&D Bio-Tech, Zug, Switzerland). The immunization were boiled and resolved by reducing 7.5% SDS-PAGE, protocol included the repeated injection of immunogen followed by immunoblot analyses using anti-DDR1 anti- emulsified with CFA into the same animal rotating dif - body D1G6 or SC532. The blots were reprobed with anti - ferent application routes. 10  ml peripheral whole blood bodies against β-actin for loading control. For positive were collected and used for B cell isolation followed by control of DDR1 immunoreactivity, we used a lysate of B-cell culture. After total RNA isolation, cDNA was gen- human prostate cancer PC3 cells overexpressing DDR1b, erated by reverse transcription of the mRNA followed by which was generated in our laboratory. PCR amplification of the V regions of the rabbit B cells For immunostaining, CFPAC-1 or PANC-1 cells using appropriate primers. The DNA sequences encod - (5 × 10 /250  µl of complete medium) were seeded in ing the VHs and VLs were obtained by sequencing the 8-well chamber slides. Twenty-four hours later, the media PCR products. Prototype cDNA expression plasmids were removed and the cells were fixed with cold (− 20 °C) were used for the recombinant expression of the HC methanol. The cells were then washed (3×) with PBS and and LC of monoclonal rabbit antibodies that are then then incubated with 3% H O in PBS for 20 min at room 2 2 expressed transiently in HEK293 cells. After 7  days, the temperature (RT) followed by two washes with PBS. The culture supernatants were harvested, purified by Pro - cells were incubated with 2.5% normal horse serum for tein A column standard protocols and analyzed for anti- 20 min at RT. The serum was removed and the cells were body content and specificity on human pancreatic cancer incubated with either in house raised DDR1 antibody CFPAC-1 and PANC-1 cells. (1:50) or control rabbit IgG (1:50) diluted in PBS. After an overnight incubation at 4  °C, the cells were washed Anti‑DDR1 antibody specificity testing using DDR1 with PBS and incubated with the ImmPRESS Reagent or DDR2 expressing cancer cell lines for 30 min at RT. Detection and visualization of antibody Human pancreatic cancer CFPAC-1 and PANC-1 cells binding was assessed using ImmPACT DAB Peroxi- were a gift from Dr. Howard Crawford, University of dase chromogenic substrate, according to the manufac- Michigan. CFPAC-1 cells were maintained in IMDM turer instructions. The slides were counterstained with supplemented with 10% fetal bovine serum 1% l -glu- Mayer’s hematoxylin for 5  s, followed by differentiation tamine and 1% streptomycin/penicillin antibiotics, and with tap water for 1 min, and finally covered with a thin PANC-1 cells in DMEM High Glucose supplemented layer of Crystal Mount Aqueous Mounting Medium. with 10% fetal bovine serum 1% l -glutamine and 1% Stained cells were photographed using a Zeiss Axioplan 2 streptomycin/penicillin antibiotics. Culture media and Moll et al. J Transl Med (2018) 16:148 Page 15 of 20 microscope (Zeiss, Gottingen, Germany) equipped with alkaline-fast red enzyme system (DakoCytomation, a software-controlled digital camera (Axiovision; Zeiss). Glostrup, Denmark). For DDR1 expression experiments during crescent Human renal tissues formation, immunostaining was performed on serial Human renal tissue, fixed in formaldehyde and embed - sections of 2 lupus nephritis class IV-G (A/C) biopsies ded in paraffin, was selected from the files of the Service (patients #25 and #29) using DDR1 and each of these 4 of Pathology, University Hospital Geneva: five control different antibodies: mouse monoclonal anti-human normal renal tissues were obtained from patients with CD68 (DakoCytomation, Glostrup, Denmark) at a 1:100 nephrectomy performed for neoplasia, and 29 biopsy dilution, mouse monoclonal anti-human nestin (R&D specimens were obtained from patients with crescentic systems Bio-Techne, Minneapolis. MN) at a 1:750 dilu- glomerulonephritis (GN): 5 Goodpasture’s syndrome, tion, mouse monoclonal anti-human cytokeratin 8 7 ANCA-associated GN, 12 IgA GN, and 5 lupus GN. and 18 (BioGenex, Fremont, CA) at a 1:20 dilution and For all biopsy specimens, standard analysis using light mouse monoclonal anti-human cytokeratin 19 (Dako- microscopy, immunofluorescence (with anti-immuno - Cytomation, Glostrup, Denmark) at a 1:20 dilution. In globulin Ig A, G, M, and anti-complement C1q, C3, C4c, addition, some serial sections were immunostained with and C5b-9 antibodies), and electron microscopy were E-cadherin (Novocastra, Newcastle, UK) at a 1:20 dilu- performed. For all biopsy specimens, standard analyses tion. Briefly, 3  μm sections of paraffin-embedded kid - were performed. Each patient gave informed consent neys were submitted to the appropriate antigen retrieval before enrollment. The institutional ethical committee and incubated with each antibody 1 h at room tempera- board approved the clinical protocol (CEREH Number ture followed by the appropriate secondary antibody for 03-081). The research was performed according to the 30 min (room temperature) and then liquid diaminoben- Helsinki’s declaration principles. zidine substrate–chromogen system (DakoCytomation, Glostrup, Denmark). Immunohistochemistry on human tissues Counterstaining was performed using Mayer hematox- Immunohistochemistry was performed as follows: after ylin. For CD68 and E-cadherin, immunostaings, sections antigen heat retrieval, 3  μm sections of the formalde- were counterstained with Periodic Acid Schiff (PAS) hyde-fixed, paraffin-embedded biopsy specimens were staining. Stained sections were examined with a Zeiss incubated with our specifically in-house raised rabbit microscope. monoclonal anti-human DDR1 antibody at a 1:100 dilu- tion 1  h at room temperature followed by an anti-rabbit Pharmacokinetic studies antibody for 30  min (room temperature) and then liq- The route of administration and dosing-regimes for uid diaminobenzidine substrate–chromogen system in  vivo studies study were based on pharmacokinetic (DakoCytomation, Glostrup, Denmark). For tubular studies. Roche-Chugai DDR1i was formulated as a colocalization experiments in the normal kidney, double micro-suspension in a vehicle (7.5% gelatin and 0.9% immunostaining was performed on serial sections (3 μm sodium chloride) and administered by oral application thick) using DDR1 and each of these 4 different anti - at 10 ml/kg. The oral pharmacokinetics of Roche-Chugai bodies: mouse monoclonal anti-human Tamm–Horsfall DDR1i was investigated in male 129 Sv mice (Additional protein (Technically Speaking, Ontario, Canada) at a file  1: Figure S4). The compound was administered via 1:80 dilution, mouse monoclonal anti-human Calbindin oral gavage, at 200 mg/kg. Plasma samples (0.05 ml) were D (clone CB-955, Sigma-Aldrich, St. Louis, MO) at a collected at 1, 3, 6 and 24  h post-dose (n = 2 mice/time 1:800 dilution, rabbit polyclonal anti-human Aquaporin point). Concentrations in mouse plasma samples were 2 (Abcam, Cambridge, UK) at a 1:20 dilution and rab- determined using a high performance liquid chroma- bit polyclonal anti-human Megalin (anti-LRP2, Sigma- tography-tandem mass spectrometry (HPLC–MS/MS) Aldrich) at a 1:5000 dilution. Briefly, 3  μm sections method. of paraffin-embedded kidneys were submitted to the appropriate antigen retrieval and incubated with DDR1 Roche‑Chugai DDR1i in vivo pharmacokinetics at a 1:100 dilution 1  h at room temperature followed by and estimated pharmacodynamics an anti-rabbit antibody for 30  min (room temperature) Estimation of inhibition of DDR1-phosphorylation was and then liquid diaminobenzidine substrate–chromo- calculated from the observed plasma concentrations in gen system (DakoCytomation, Glostrup, Denmark). Sec- relationship to the in vitro binding IC50 for Roche-Chu- tions were then incubated with the adequate antibody gai DDR1i against human DDR1, accounting for mouse for 1  h at room temperature followed by the appropri- plasma protein binding of 98.9% (Additional file  1: Fig- ate second antibody for 30 min and then by phosphatase ure S4). Thus, inhibition of DDR1-phosporylation of Moll et al. J Transl Med (2018) 16:148 Page 16 of 20 > 90% in average over 24 h in vivo was estimated from the the seminal work of Morley and Wheeler in mice [28, observed plasma concentrations of Roche-Chugai DDR1i 29]. Briefly, sheeps were immunized by subcutaneous on daily treatment of 129 Sv mice with 200 mg/kg po. injection of mouse isolated glomeruli. The total dose of isolated glomeruli (600  μg) was divided to four injec- Kinase assays tions separated by a 1  week interval. 15  days later and The KINOMEscan screening platform (DiscoveRx Cor- under treatment with anti-allergic agent (Phenergan, SM poration, San Diego, USA DiscoverX s canMAX Kinase Wyeth, USA), sheep received another intravenous injec- Assay Panel) employing an active site-directed competi- tion of the same quantity of mouse glomeruli. One week tion binding assay was used to quantitatively measure later animals were sacrificed and serum was collected. interactions between the 2 test compounds Roche-Chu- Serum was then heat-inactivated at 56 °C for 30 min and gai DDR1i and Imatinib on 451 kinases and disease rel- filtered (pore size = 0.2  μm) and stored in 1  ml aliquots evant mutant variants. In brief, T7 kinase-tagged phage at − 20  °C. Next we passively induced glomerulonephri- strains were grown in parallel in 24-well blocks or 96-well tis by injecting mice intravenously with the serum col- blocks in an E. coli host derived from the BL21 strain. E. lected from sheep. We used four different doses (5, 10, coli were grown to log-phase and infected with T7 phage 15, 20 μl/g) for the four serum lots in order to validate the from a frozen stock (multiplicity of infection = 0.4) and serum and choose the appropriate dose. All serum lots incubated with shaking at 32 °C until lysis (90–150 min). were toxic at doses 10, 15 and 20  μl/g and non-nephro- The lysates were centrifuged (6000×g) and filtered toxic at 5 μl/g dose. All serums were specific against glo - (0.2  μm) to remove cell debris. The remaining kinases merular compartments (data not shown). One lot was were produced in HEK-293 cells and subsequently tagged found to induce severe proteinuria with typical histologi- with DNA for qPCR detection. Streptavidin-coated cal lesions at 10 μl/g; therefore this serum was used in the magnetic beads were treated with biotinylated small next protocol. molecule ligands for 30 min at room temperature to gen- erate affinity resins for kinase assays. The liganded beads Dose selection for the NTS experiment were blocked with excess biotin and washed with block- DDR1i doses selection was designed based on in  vitro ing buffer [SeaBlock (Pierce), 1% BSA, 0.05% Tween 20, pharmacology and bioavailability and clearance obtained 1  mM DTT] to remove unbound ligand and to reduce in the above mentioned pharmacokinetic studies. Two nonspecific phage binding. Binding reactions were doses were selected a dose covering 10× the in  vitro assembled by combining kinases, liganded affinity beads, IC50, resulting in the DDR1 HD dose (200  mg/kg) and and the 2 test compounds in 1× binding buffer (20% Sea - a second dose around 1× coverage over IC50, result- Block, 0.17× PBS, 0.05% Tween 20, 6  mM DTT). Test ing in the DDR1 LD dose (75 mg/kg). The Imatinib dose compounds were prepared as 40× stocks in 100% DMSO (60  mg/kg) was simply derived from the literature [25] and directly diluted into the assay. All reactions were per- and was added to the study as pure control since the formed in polypropylene 384-well plates in a final volume imatinib in vitro potency and in vivo exposure (data not of 0.02 ml. The assay plates were incubated at room tem - shown) are in the very same order of the DDR1i HD dose. perature with shaking for 1 h and the affinity beads were washed with wash buffer (1× PBS, 0.05% Tween 20). Animal treatment and nephrotoxic serum protocol The beads were then re-suspended in elution buffer (1× All mice were kept in well-controlled animal housing PBS, 0.05% Tween 20, 0.5  μM non-biotinylated affinity facilities and had free access to water and pellet food. ligand) and incubated at room temperature with shaking Animal procedures and protocols were in accordance for 30  min. The kinase concentration in the eluates was with the European Guidelines for the Care and use of measured by qPCR. The Selectivity Score (S-score) was Laboratory Animals and have been approved by the calculated for both compounds. The compounds were Inserm and UPMC ethical committees. screened at the concentrations requested, and results Glomerulonephritis was induced by retro-orbitaly injec- for primary screen binding interactions were reported as tion of decomplementated nephrotoxic serum (NTS). percent competition (% Competition). A total of 40 female mice 129/SV aged 3–6  months and weighting 18–25  g were used (Janvier, Le Genest-St-Isle, Nephrotoxic mouse model of crescentic France). The total number of mice was divided into five glomerulonephritis treatment groups as followed: 8 mice were injected with Preparation of nephrotoxic serum NTS and fed with vehicle, 8 mice were injected with NTS Decomplementated nephrotoxic serum (NTS) was pre- and fed with low dose DDR1i, 8 mice were injected with pared as previously described [27]. This protocol has NTS and fed with high dose DDR1i, 8 mice were injected been popularized by Salant and Cybulsky according to with NTS and fed with Imatinib and 8 mice were injected Moll et al. J Transl Med (2018) 16:148 Page 17 of 20 Immunohistochemistry and morphometric analysis on NTS with PBS and fed with vehicle. NTS was injected in mice mouse tissues (10  μl/gBW/day) during 3 consecutive days. Treatment On consecutive 4  μm thick slides, immunohistochemis- was started 1 day prior first injection of NTS or PBS. All try (IHC) was performed on the Ventana Discovery XT treatments were provided by oral gavage. The average immunostainer with anti-desmin rabbit polyclonal anti- food intake was controlled by weighing the food every body (Spring Bioscience, E2574, dilution 1:100), anti- 3 days. Mice were found to consume about 4 g/day/mouse CD44 rat monoclonal (BD Pharmingen, 550538, dilution which was similar to all groups. 1:50), anti-collagen type IV rabbit polyclonal antibody (Millipore, AB756P, dilution 1:200) and anti-Ki67 rat Urine and plasma sample collection and analyses polyclonal antibody (eBioscience, 14-5698-82, dilution All mice were acclimated in metabolic cages for 24 h with 1:2000), as primary antibodies in a standard protocol. free access to food and tap water for 24-h urine collec- The Biotin-SP-conjugated AffiniPure donkey anti-rabbit tion. Proteinuria, expressed as grams of protein per mil- IgG (Jackson ImmunoResearch, 711-065-152, dilution limole of creatininuria, was assessed at day 14 using the 1:100) for desmin and collagen type IV or the Biotin- Pyrogallol Red method and utilizing a KONELAB auto- SP-conjugated AffiniPure donkey anti-rat IgG (Jackson mate (Thermo Scientific, Waltman, MA, USA). Blood ImmunoResearch, 712-065-153, dilution 1:100) for CD44 samples were collected on the day of sacrifice (day 14) in and Ki67 were used as secondary antibody in combina- EDTA tubes. Creatinine and BUN were assessed in blood tion with the Ventana DAB Ma p (05266360001) detec- plasma and measured using an enzymatic spectrophoto- tion kit. Slides were counterstained with hematoxylin. metric method and were expressed respectively as mg/dl Collagen type IV IHC slides were scanned with an Aperio and mmol/l. ScanScopeAT slidescanner. Quantitative morphom- etry analysis was performed with a rule set recognizing Animal sacrifice and tissue processing the percentage area of collagen type IV staining using All mice were sacrificed at day 14 post injection. Kidney the Definiens TissueStudio Version 3.51 software. All tissue was processed after normal saline perfusion as fol- images were taken at 20×. lowed. Left kidney: one half was fixed in formalin acetic acid (Formol 2%, acetic acid 5%, alcohol 75% and distilled NEP25 mouse model of glomerulosclerosis water 18%) for 12 h and then embedded in paraffin, and NEP25 mice were kindly provided from Prof. Matsusaka the other half was frozen (OCT-embedded). Right kid- in Tokai University. All mice were kept in well-controlled ney: one half was fixed in formalin acetic acid for 12  h animal housing facilities and had free access to water and then embedded in paraffin. The other half was cut in and pellet food. Animal procedures and protocols were two; 1/4 snap frozen, and the other 1/4 was kept in RNA in accordance with the Guidelines for the Care and Use later and frozen in liquid nitrogen. of Laboratory Animals at Chugai Pharmaceutical Co. Ltd. and approved by Institutional Animal Care and Use Histology and semi‑quantitative scoring analyses on NTS Committee. mouse tissues A total of 28 male mice (4 C57BL/6NCrlCrlj mice From kidney paraffin blocks, sections of 4  μm were pre - and 24 Nephrin-hCD25 (NEP25) transgenic mice) aged pared and stained with Hematoxylin and Eosin (H&E) or 10  weeks were used. One day before LMB2 (Anti-Tac Periodic Acid Schiff (PAS) and examined under a con - (Fv)-PE38) injection, NEP25 mice were divided into three ventional light microscope (Zeiss Axioskop). The fol - groups of 8 mice each based on bodyweight as followed: lowing glomerular findings were appreciated and scored: one group with DDR1i treatment, one group with Capto- glomerular hypertrophy, mesangial matrix expansion, pril treatment and one group receiving vehicle. 0.7  ng/g hypercellularity and hyperplasia of the parietal epithe- BW of LMB2 was intravenously injected in NEP25 mice lial cells and crescents; the following tubulo-interstitial to induce glomerulosclerosis (day 1). Captopril treat- findings were appreciated and scored: tubular degenera - ment (0.15  mg/mL in drinking water, approximatively tion/regeneration, tubular casts, interstitial inflammation 30  mg/kg/day) was started 1  day before LMB2 injection and interstitial fibrosis. For severity scoring, the follow - and given until sacrifice. DDR1i treatment (50  mg/kg/ ing scores were applied for each mouse tissue: 0 = none; day) and vehicle were started 7  days after LMB2 injec- 1 = minimal (approximately 1–20% of the kidney affected; tion. DDR1i and vehicle were provided by oral gavage. All 2 = slight (approximately 21–40% affected); 3 = moderate mice were acclimated in metabolic cages with free access (approximately 41–60% affected); 4 = marked (61–80% to food and tap water. Body weight was measured daily. affected); 5 = severe (approximately 81–100% affected). A 24-h urine was collected from day 14 to 15. Albumin Both glomerular and tubulo-interstitial summary scores and creatinine were measured in urine by Lbi s Mouse were built from the individual scores. Moll et al. J Transl Med (2018) 16:148 Page 18 of 20 Urinary Albumin Assay Kit using a TBA-120FR (Toshiba (Affymetrix Inc., Santa Clara, CA). Quantile normali - Medical Systems Corporation). Albumin-creatinine ratio zation using the Robust Multi-array Average (RMA) (ACR), expressed as mg of albuminuria/mg of creatin- method was applied to the raw individual microarray inuria, was assessed at day 15. All mice were sacrificed data set. The dataset was processed using the standard at day 15 post injection. It should be mentioned that Bioconductor affy package [30]. After RMA normali - one mouse in the DDR1i group was excluded from the zation, probes representing the same genes were col- analysis because congenital renal anomalies were found lapsed into a single value and standardized by taking the at autopsy (DDR1i group N = 7). Blood samples were col- mean value for each gene across the sample set. For each lected and mixed with heparin. Creatinine was measured treatment group (LD DDR1i, HD DDR1i and Imatinib) by HPLC and expressed as mg/dl. Cystatin C was meas- contrasts were calculated against the Control and the ured using Iatro CysC (LSI Medience Corporation) and resulting differentially expressed gene list were subject to expressed as mg/L. Both kidneys were removed and pro- Gene Set Enrichment Analysis [31]. cessed as followed: one-quarter of each kidney was fixed in either 10% Formalin Neutral Buffer Solution (for H&E Gene ontology signature enrichment staining) or Methyl Carnoy’s fixative (for PAS staining) Expression signals for each samples was tested for and then embedded in paraffin. Cortex of the rest kidney enrichment of gene ontology biological process terms was snap frozen in liquid nitrogen for mRNA analysis. [32] using an implementation of the Wilcoxxon’ test PAS stained sections were analyzed for quantifying glo- called bioQC [33]. The resulting enrichment scores were merular damage. PAS positive area was graded for each taken as a measure of the upregulation/downregulation glomerulus using a score of 0–4 as previously described of a specific biological process. [14]. Briefly, more than 50 of randomly selected glomeruli from each mouse were evaluated for glomerulosclerosis. In vitro PEC experiments Score 0 represents no lesion, whereas 1, 2, 3, and 4 rep- Mouse primary PEC were plated and rested overnight. resent mesangial matrix expansion, hyalinosis, or sclero- Cells were incubated in RPMI-1640 containing 1% serum sis, involving ≤ 25, 25% to ≤ 50, 50% to ≤ 75 and > 75% of and supplemented with different concentrations of DDR1 the glomerular tuft area, respectively. Tubulointerstitial inhibitor (0.01, 0.1 and 1  μM) for 1  h followed by addi- damage was graded for each mouse on H&E-stained sec- tional treatment with type I collagen 100  μg/mL (Nitta tions, using a score of 0–4. Score 0 represents no lesion, Gelatin, Japan). After 6 or 24 h of type I collagen stimu- whereas 1, 2, 3, and 4 represents lesions of tubulointersti- lation, cells were lysed for phospho DDR1 ELISA (Cell tial fibrosis/inflammation, involving ≤ 25, 25% to ≤ 50%, Signaling Technology) and total RNA isolation. Total 50 to ≤ 75%, and > 75% of the cortex tubulointerstitial RNA was isolated and amplified using an RNeasy Mini area, respectively. kit (Qiagen) and Transcriptor Universal cDNA Mas- Total RNA was isolated with RNeasy MINI kit (QIA- ter (Roche) according to manufacturer’s instructions. GEN, #74106) from frozen cortical kidney lysate and Quantitative RT-PCR was performed on a LightCycler cDNA was synthesized by reverse transcription. qRT- LC480 (Roche) for C3, Mmp2, Mmp14 and Vcam1 using PCR was performed using fibrosis markers Tgf -b1 the following primers: C3 (Mm01232779_m1, TaqMan (Mm01178820_m1, TaqMan Gene Expression Assays, Gene Expression Assays, Applied Biosystems), Mmp2 Applied Biosystems), Acta2 (Mm00725412_s1) and (Mm00439498_m1), Mmp14 (Mm00485054_m1), and Col1a1 (Mm00801666_g1) and of the inflammation Vcam1 (Mm01320970_m1). Relative gene expression was marker Ccl2 (Mm00441242_m1). calculated with the 2-∆Ct method using GAPDH as an endogenous control. Microarray analysis of gene expression on NTS and NEP25 mouse tissues Statistics Harvested tissues were lysed with RNeasy lysis buffer Values are expressed as mean ± SE M. Box-and-whisker immediately after treatment with MM or DM. Lysates plot represents quartiles. Data were analyzed using one- were sheared with QIAshredder spin columns and total way analysis of variance followed by protected least sig- RNA was extracted using RNeasy kits as per the man- nificant difference Fisher’s test of the Stat-view software ufacturer’s guidelines (QIAGEN GmbH, Hilden, Ger- package. Glomerular and tubular summary scores and many). Starting with 1 μg total RNA per sample, reverse collagen type IV quantitative morphometry data were transcription into cDNA and subsequent steps until analyzed using the Kruskal–Wallis test followed by p hybridization onto GeneChip oligonucleotide microar- value adjusted Dunett’s post hoc test. Data on in  vivo rays (Human Genome U133 plus 2.0) and scanning were NEP study were analysed using t-test with GraphPad conducted using the manufacturer’s kits and instructions Prism except for the score of tubulointerstitial damage Moll et al. J Transl Med (2018) 16:148 Page 19 of 20 manuscript and Prof. Matsusaka ( Tokai University Medical School, Isehara, which was analysed by Mann–Whitney U test. Data on Kanagawa, Japan) for kindly providing NEP25 mice. in vitro PEC study were analysed using Dunnett’s mul- tiple comparison test with GraphPad Prism. Results Competing interests At the time of the experiments L.B., I.F., G.G., S.U., T.B., R.G., M.T., H.R., A.A., J.F. with p < 0.05 were considered statistically significant. and M.P. were employees of Hoffmann-La Roche Ltd. Y.Y, T.M, H.S., A.M, N.F and M.K. are Chugai Pharma Ltd. employees., S.M., R.F., M.J.M and D.R.B have Additional files no conflict of interest. A.A and C.C. received a research grant support from Hoffmann-La Roche Ltd. Additional file 1. Additional figures. Availability of data and materials Additional file 2. Additional tables. All data gene expression data generated in NTS mouse model have been deposited on the GEO database (GEO Accession Number GSE104426). For the Additional file 3. Additional materials. other data and materials, please, contact author for data requests. Consent for publication Abbreviations All authors have given consent for publication. DDR1: Discoidin domain receptor 1; GN: glomerulonephritis; NTS: nephrotoxic serum; PEC: parietal epithelial cell; RTK: receptor tyrosine kinase; KO: knockout; Ethics approval and consent to participate ASO: antisense oligonucleotides; DDR1i: DDR1 inhibitor; IHC: immunohisto- Studies in NTS and NEP25 mouse models of GNs were approved by the local chemistry; CK: cytokeratin; GEP: gene expression profiling; GSEA: Gene Set appropriate ethical review boards. NTS animal authorisation was released Enrichment Analysis; MoA: mode of action; BUN: blood urea nitrogen. by the French minister of research (Authorisation Number C 75-20-01 of the 17.11.2014) and the NEP25 animal authorisation was released by Chugai Authors’ contributions institutional animal care and use committee, a committee accredited by the SM has generated all human data included in that manuscript analysed the Association for Assessment and Accreditation of Laboratory Animal Care data and co-written the manuscript with MP. YY has designed and conducted (Authorisation Number 12-093). the NEP25 in vivo experiment and analysed the data. TM has invented and provided the DDR1i used in the present study. FD has written part of the text Funding and provided critical review of the manuscript. NF has contributed to the No funding was requested to support that research work. entire experimental design of NEP25 in vivo study and examined the pathol- ogy of the NEP25 study. MK has generated the PEC in vitro experiment. AA has performed the in vivo NTS experiments HS has run the NEP25 gene expres- Publisher’s Note sion profile analysis. HS and AM have contributed to the entire experimental Springer Nature remains neutral with regard to jurisdictional claims in pub- design of NEP25 in vivo study and PEC in vitro study. SU has provided the lished maps and institutional affiliations. ISH characterisation of both NEP25 and NTS tissue samples. LB has run the entire gene expression profile analysis. TC has run the immunohistochemi- Received: 30 April 2018 Accepted: 23 May 2018 cal characterisation of DDR1 in renal biopsies. IF contributed to the entire experimental design of the study and provided valuable biomarker support to the present manuscript. GG has generated and provided the proprietary anti-DDR1 antibody. TB has generated the DDR1 isoform qRT-PCR data. RG has run the pharmacokinetic of compound A and Imatinib and designed the References doses for the in vivo NTS experiment and provided valuable support to data 1. Vogel W, et al. The discoidin domain receptor tyrosine kinases are acti- interpretation. DRB and RF have characterised and validated the proprietary vated by collagen. Mol Cell. 1997;1(1):13–23. anti-DDR1 antibody using specific cancer cell lines. HR has run the in vitro 2. Valiathan RR, et al. Discoidin domain receptor tyrosine kinases: new play- characterisation of DDR1i (binding, phosphorylation and selectivity screen). ers in cancer progression. Cancer Metastasis Rev. 2012;31(1–2):295–321. JF has contributed the pathological characterisation of the NTS in vivo experi- 3. Leitinger B. Discoidin domain receptor functions in physiological and ment. MJM has contributed critical reflection on the interpretation of the PEC pathological conditions. Int Rev Cell Mol Biol. 2014;310:39–87. in vitro experiments. CC contributed NTS experimental design and revised the 4. Borza CM, Pozzi A. Discoidin domain receptors in disease. Matrix Biol. manuscript. MP has designed the experiments, analysed the data and written 2014;34:185–92. the manuscript. All authors read and approved the final manuscript. 5. Yeh YC, Lin HH, Tang MJ. A tale of two collagen receptors, integrin beta1 and discoidin domain receptor 1, in epithelial cell differentiation. Am J Author details Physiol Cell Physiol. 2012;303(12):C1207–17. Department of Pathology, University Hospital of Geneva, Geneva, Swit- 6. Hahn WH, et al. Linkage and association study of discoidin domain recep- zerland. Research Division, Chugai Pharmaceutical Co., Ltd, Tokyo, Japan. tor 1 as a novel susceptibility gene for childhood IgA nephropathy. Int J 3 4 INSERM, UMR S 1155, Hôpital Tenon, 75020 Paris, France. Roche Pharma Mol Med. 2010;25(5):785–91. Research and Early Development, Roche Innovation Center Basel, Basel, 7. Kavvadas P, Dussaule JC, Chatziantoniou C. Searching novel diagnostic Switzerland. Roche Pharma Research and Early Development, Roche Innova- markers and targets for therapy of CKD. Kidney Int Suppl. 2014;4(1):53–7. tion Center Munich, Munich, Germany. Department of Pathology, College 8. Flamant M, et al. Discoidin domain receptor 1 null mice are protected of Medical Sciences, Nova Southeastern University, Fort Lauderdale, FL, USA. against hypertension-induced renal disease. JASN. 2006;17(12):3374–81. 7 8 Department of Pathology, Wayne State University, Detroit, MI, USA. Depart- 9. Guerrot D, et al. Discoidin domain receptor 1 is a major mediator of ment of Nephrology and Clinical Immunology, RWTH University, Aachen, Ger- inflammation and fibrosis in obstructive nephropathy. Am J Pathol. many. Present Address: Chugai Pharmabody Research Pte. Ltd., Singapore, 2011;179(1):83–91. Singapore. Present Address: Late Stage, AstraZeneca, Göteborgs, Sweden. 10. Kerroch M, et al. Protective effects of genetic inhibition of Discoidin Present Address: Office of Innovation, Immunology, Infectious Diseases & Domain Receptor 1 in experimental renal disease. Sci Rep. 2016;6:21262. Ophthalmology (I2O), Roche and Genentech Late Stage Development, 124 11. Alfieri C, et al. Discoidin domain receptor-1 and periostin: new players in Grenzacherstrasse, 4070 Basel, Switzerland. School of Pharmaceutical Sci- chronic kidney disease. Nephrol Dial Transplant. 2015;30(12):1965–71. ences, University of Geneva, Geneva, Switzerland. 12. Kerroch M, et al. Genetic inhibition of discoidin domain recep- tor 1 protects mice against crescentic glomerulonephritis. FASEB J. Acknowledgements 2012;26(10):4079–91. We acknowledge Andrea Araujo Del Rosario and Franziska Weibel for their 13. Gross O, et al. Loss of collagen-receptor DDR1 delays renal fibrosis in technical support, Faye Drawnel for language proofing of the original hereditary type IV collagen disease. Matrix Biol. 2010;29(5):346–56. Moll et al. J Transl Med (2018) 16:148 Page 20 of 20 14. Matsusaka T, et al. Genetic engineering of glomerular sclerosis in the 24. Overstreet JM, et al. Selective activation of epidermal growth factor mouse via control of onset and severity of podocyte-specific injury. JASN. receptor in renal proximal tubule induces tubulointerstitial fibrosis. FASEB 2005;16(4):1013–23. J. 2017;31(10):4407–21. 15. Jennette JC. Rapidly progressive crescentic glomerulonephritis. Kidney 25. Iyoda M, et al. Preventive and therapeutic effects of imatinib in Wistar- Int. 2003;63(3):1164–77. Kyoto rats with anti-glomerular basement membrane glomerulonephri- 16. Iyoda M, et al. Long- and short-term treatment with imatinib attenuates tis. Kidney Int. 2009;75(10):1060–70. the development of chronic kidney disease in experimental anti- 26. Day E, et al. Inhibition of collagen-induced discoidin domain receptor glomerular basement membrane nephritis. Nephrol Dial Transplant. 1 and 2 activation by imatinib, nilotinib and dasatinib. Eur J Pharmacol. 2013;28(3):576–84. 2008;599(1–3):44–53. 17. Moll S, et al. Epithelial cells as active player in fibrosis: findings from an 27. Salant DJ, Cybulsky AV. Experimental glomerulonephritis. Methods Enzy- in vitro model. PLoS ONE. 2013;8(2):e56575. mol. 1988;162:421–61. 18. Kabgani N, et al. Primary cultures of glomerular parietal epithelial cells or 28. Lloyd CM, et al. RANTES and monocyte chemoattractant protein-1 podocytes with proven origin. PLoS ONE. 2012;7(4):e34907. (MCP-1) play an important role in the inflammatory phase of crescentic 19. Dorison A, Placier S, Dubois Y, Chladichristos C, Rondeau E, Chatzianto- nephritis, but only MCP-1 is involved in crescent formation and interstitial niou C, Dussaule JC. Discoidin domain receptor 1 is a key mediator of fibrosis. J Exp Med. 1997;185(7):1371–80. ischemia-reperfusion induced injury. San Diego: American Society of 29. Morley AR, Wheeler J. Cell proliferation within Bowman’s capsule in mice. Nephrology; 2015. J Pathol. 1985;145(4):315–27. 20. Lee R, et al. Localization of discoidin domain receptors in rat kidney. 30. Gautier L, et al. affy—analysis of Affymetrix GeneChip data at the probe Nephron. Exp Nephrol. 2004;97(2):e62–70. level. Bioinformatics. 2004;20(3):307–15. 21. Rubel D, et al. Collagen receptors integrin alpha2beta1 and discoidin 31. Subramanian A, et al. Gene set enrichment analysis: a knowledge-based domain receptor 1 regulate maturation of the glomerular basement approach for interpreting genome-wide expression profiles. Proc Natl membrane and loss of integrin alpha2beta1 delays kidney fibrosis in Acad Sci USA. 2005;102(43):15545–50. COL4A3 knockout mice. Matrix Biol. 2014;34:13–21. 32. Osumi-Sutherland D, Ponta E, Courtot M, Parkinson H, Badi L. Cell, chemi- 22. Fatima H, et al. Parietal epithelial cell activation marker in early recurrence cal and anatomical views of the gene ontology: mapping to a roche of FSGS in the transplant. Clin J Am Soc Nephrol. 2012;7(11):1852–8. controlled vocabulary. CEUR Workshop Proc. 2015;1456:84–93. 23. Stamenkovic I, Skalli O, Gabbiani G. Distribution of intermediate fila- 33. Zhang JD, et al. Detect tissue heterogeneity in gene expression data with ment proteins in normal and diseased human glomeruli. Am J Pathol. BioQC. BMC Genomics. 2017;18(1):277. 1986;125(3):465–75. Ready to submit your research ? Choose BMC and benefit from: fast, convenient online submission thorough peer review by experienced researchers in your field rapid publication on acceptance support for research data, including large and complex data types • gold Open Access which fosters wider collaboration and increased citations maximum visibility for your research: over 100M website views per year At BMC, research is always in progress. Learn more biomedcentral.com/submissions

Journal

Journal of Translational MedicineSpringer Journals

Published: Jun 1, 2018

References

You’re reading a free preview. Subscribe to read the entire article.


DeepDyve is your
personal research library

It’s your single place to instantly
discover and read the research
that matters to you.

Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.

All for just $49/month

Explore the DeepDyve Library

Search

Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly

Organize

Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.

Access

Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.

Your journals are on DeepDyve

Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.

All the latest content is available, no embargo periods.

See the journals in your area

DeepDyve

Freelancer

DeepDyve

Pro

Price

FREE

$49/month
$360/year

Save searches from
Google Scholar,
PubMed

Create lists to
organize your research

Export lists, citations

Read DeepDyve articles

Abstract access only

Unlimited access to over
18 million full-text articles

Print

20 pages / month

PDF Discount

20% off