Both Specific Endothelial and Proximal Tubular Adam17 Deletion Protect against Diabetic Nephropathy

ADAM17 is a disintegrin and metalloproteinase capable of cleaving the ectodomains of a diverse variety of molecules including TNF-α, TGF-α, L-selectin, and ACE2. We have previously demonstrated that renal ADAM17 is upregulated in diabetic mice. The role of endothelial (eAdam17) and proximal tubular (tAdam17) Adam17 deletion in renal histology, modulation of the renin angiotensin system (RAS), renal inflammation, and fibrosis was studied in a mouse model of type 1 Diabetes Mellitus. Moreover, the effect of Adam17 deletion in an in vitro 3D cell culture from human proximal tubular cells under high glucose conditions was evaluated. eAdam17 deletion attenuates renal fibrosis and inflammation, whereas tAdam17 deletion decreases podocyte loss, attenuates the RAS, and decreases macrophage infiltration, α-SMA and collagen accumulation. The 3D in vitro cell culture reinforced the findings obtained in tAdam17KO mice with decreased fibrosis in the Adam17 knockout spheroids. In conclusion, Adam17 deletion either in the endothelial or the tubular cells mitigates kidney injury in the diabetic mice by targeting different pathways. The manipulation of Adam17 should be considered as a therapeutic strategy for treating DN.


Introduction
Diabetic nephropathy (DN) is the leading cause of end-stage renal disease in developed countries. Early DN is associated with glomerular injury, hyperfiltration and microalbuminuria, both attributable to podocyte loss, and tubular and endothelial dysfunction [1,2]. DN progression is followed by proteinuria, extracellular matrix deposition, glomerulosclerosis, and tubulointerstitial fibrosis [3][4][5]. Macrophage infiltration in DN characteristically appears in both human and experimental models and it is associated with hypertrophy, hypercellularity, interstitial myofibroblast accumulation, and collagen-IV deposition in the glomeruli and tubulointerstitium [6]. Profibrotic growth factors, such as transforming growth factor β (TGF-β), have a key role in extracellular matrix accumulation in high-glucose environments [7,8].
Adam17 knockout (Adam17KO) mice presented reduced renal fibrosis and inflammation after acute kidney injury (AKI) induction or angiotensin II (ANG II) infusion [16,17]. Furthermore, the decrease in cleaved tumor necrosis factor alpha (TNF-α) in these mice correlated with renal reduction in pro-inflammatory markers and infiltrating macrophages and neutrophils [16]. ADAM17 expression is increased in endothelial and proximal tubular cells of DN patients [18]. However, the effect of the specific deletion of endothelial or tubular Adam17 on DN has not been previously studied. Here, we elucidate whether specific Adam17 deletion on endothelial or renal proximal tubular cells has a beneficial effect in type 1 diabetic mice. Moreover, we performed an in vitro 3D-cell culture of human proximal tubular cells incubated with high-glucose as a disease model for kidney fibrosis aimed to decipher the effect of Adam17 deletion against a fibrotic diabetic milieu.

Confirmation of Adam17 Deletion on Endothelial and Renal Proximal Tubular Cells
eAdam17 and tAdam17 deletions were confirmed by immunohistochemistry, β-galactosidase staining and Adam17 gene expression. eAdam17KO mice presented reduced ADAM17 expression in the endothelium of the kidney arteries as compared with wild-type mice. Within the kidney tubules, tAdam17KO mice presented reduced ADAM17 protein expression in the proximal tubular cells compared with wild-type mice (Supplementary Figure S1B).
In addition, β-galactosidase staining as control of recombination was performed. tAdam17 mice presented the ROSA26-driven LacZ-Cre reporter with β-galactosidase activity after a STOP codon flanked by loxP sites ( Figure S1C). Therefore, only if recombination takes place will positive staining for β-galactosidase be detected. Moreover, we also showed lower Adam17 gene expression in tAdam17KO mice as compared to WT mice (Supplementary Figure S1D).

Tubular Adam17 Knockout Protects from DN Hyperfiltration
Increased blood glucose levels and decreased body weight (BW) were observed in all diabetic mice. Kidney hypertrophy was calculated by the ratio kidney weight (KW)/BW. Diabetic animals presented a higher KW/BW ratio than their controls in both animal models. Diabetic tAdam17KO mice presented a lower KW/BW ratio when compared with diabetic wild-type mice, suggesting a reduced hypertrophy (Table 1).
Diabetes was associated with increased urine albumin-to-creatinine ratio in both animal models (Supplementary Figure S2).

Diabetic Kidneys of Adam17-Deficient Mice at Either Proximal Tubular Cells or Endothelial Cells Are Protected from Matrix Mesangial Expansion
PAS-stained kidney samples revealed a significant increase in glomerular tuft area, matrix mesangial expansion and mesangial index in all diabetic wild-type mice. This increase was attenuated in diabetic knockout mice ( Figure 1A-G). Moreover, among diabetic animals, eAdam17KO mice exhibited a significantly decreased mesangial matrix expansion and mesangial index when compared with wild-type mice ( Figure 1B,C).

Adam17 Deletion in Proximal Tubular Cells Protects against Podocyte Loss
The podocyte number was significantly decreased in diabetic groups. However, diabetic tAdam17KO displayed lower podocyte loss in comparison with diabetic wild-type mice ( Figure 1H-J).

Adam17 Deletion in Proximal Tubular Cells Attenuates the Renin Angiotensin System
Circulating ACE2 enzymatic activity was increased in all diabetic mice. Interestingly, tAdam17 deletion in diabetic mice reduced circulating ACE2 activity when compared with diabetic wild-type mice (Figure 2A,B).
Renal ACE2 enzymatic activity was significantly increased in diabetic mice ( Figure 2C,D). eAdam17 deletion exacerbates renal ACE2 activity in such mice ( Figure 2C), while tAdam17 deletion attenuates the increase in renal ACE2 activity in diabetic mice ( Figure 2D).
We also analysed the gene expression of different RAS components. All diabetic mice presented upregulated Agt gene expression (Supplementary Figure S3). Regarding ANG II receptors, At1r gene expression was downregulated in eAdam17 diabetic mice ( Figure 2E). In contrast, At2r gene expression increased in diabetic wild-type mice of both animal models ( Figure 2G,H).

Adam17 Deletion Decreased Inflammation in DN
To evaluate whether deleting Adam17 in endothelial or proximal tubular cells reduces renal inflammation, we analysed the expression of different pro-inflammatory markers. Circulating TNF-α levels were significantly increased in diabetic wild-type mice in comparison with controls. Interestingly, among diabetic animals, both knockout mice exhibited significantly decreased TNF-α levels when compared with wild-type mice ( Figure 3A,B). Moreover, hyperglycaemia was accompanied by a significant increase in mRNA levels of Tnfα ( Figure 3C,D). Similarly, Tnfr1 expression was upregulated in all diabetic mice ( Figure 3E,F). Interestingly, diabetic eAdam17KO mice presented decreased Tnf-α ( Figure 3C) and increased Tnfr1 ( Figure 3E) gene expression when compared with eAdam17WT mice.  betic tAdam17KO displayed lower podocyte loss in comparison with diabetic wild-type mice ( Figure 1H-J).

Adam17 Deletion in Proximal Tubular Cells Attenuates the Renin Angiotensin System
Circulating ACE2 enzymatic activity was increased in all diabetic mice. Interestingly, tAdam17 deletion in diabetic mice reduced circulating ACE2 activity when compared with diabetic wild-type mice (Figure 2A,B).  parison with controls. Interestingly, among diabetic animals, both knockout mice exhibited significantly decreased TNF-α levels when compared with wild-type mice ( Figure  3A,B). Moreover, hyperglycaemia was accompanied by a significant increase in mRNA levels of Tnfα ( Figure 3C,D). Similarly, Tnfr1 expression was upregulated in all diabetic mice ( Figure 3E,F). Interestingly, diabetic eAdam17KO mice presented decreased Tnf-α ( Figure 3C) and increased Tnfr1 ( Figure 3E) gene expression when compared with eAdam17WT mice. A similar pattern was observed for chemokine Ccl5 ((C-C motif) ligand 5) gene expression. While hyperglycaemia induced an increase in Ccl5 gene expression, Adam17 deletion in endothelial cells attenuates Ccl5 upregulation in diabetic mice ( Figure 3G).
Macrophage infiltration in the kidney was assessed by F4/80 protein detection. In both strains, F4/80 staining was increased in wild-type diabetic mice as compared to controls. Adam17 deletion either in endothelial or proximal tubular cells reduced macrophage infiltration in diabetic mice in comparison with diabetic wild-type mice ( Figure 4).
A similar pattern was observed for chemokine Ccl5 ((C-C motif) ligand 5) gene expression. While hyperglycaemia induced an increase in Ccl5 gene expression, Adam17 deletion in endothelial cells attenuates Ccl5 upregulation in diabetic mice ( Figure 3G).
Macrophage infiltration in the kidney was assessed by F4/80 protein detection. In both strains, F4/80 staining was increased in wild-type diabetic mice as compared to controls. Adam17 deletion either in endothelial or proximal tubular cells reduced macrophage infiltration in diabetic mice in comparison with diabetic wild-type mice ( Figure 4).

Diabetic Kidneys of Adam17 Deficient Mice in Either Proximal Tubular Cells or Endothelial Cells Are Protected from Renal Fibrosis
To establish whether Adam17 deletion ameliorates renal fibrosis, gene expression for different fibrotic markers related to EGFR signaling pathways was studied. Tgfα and Hbegf gene expression were analysed as substrates of the EGFR signaling pathway. In both strains, Tgfα gene expression was increased in all diabetic mice ( Figure 5A,B). However, no differences were observed between groups regarding Hb-egf gene expression (Supplementary Figure S4).

Diabetic Kidneys of Adam17 Deficient Mice in Either Proximal Tubular Cells or Endothelial Cells Are Protected from Renal Fibrosis
To establish whether Adam17 deletion ameliorates renal fibrosis, gene expression for different fibrotic markers related to EGFR signaling pathways was studied. Tgfα and Hb-egf gene expression were analysed as substrates of the EGFR signaling pathway. In both strains, Tgfα gene expression was increased in all diabetic mice ( Figure 5A,B). However, no differences were observed between groups regarding Hb-egf gene expression (Supplementary Figure S4).
For downstream effectors of the EGFR signaling pathway, we observed downregulated Tgfβ gene expression in diabetic mice except for the eAdam17KO mice. Interestingly, only eAdam17 deletion ameliorated the diabetes effect on renal Tgfβ gene expression ( Figure 5C,D). At the protein level, only diabetic wild-type mice increased TGF-β ( Figure 5E,F). Surprisingly, diabetic eAdam17KO mice presented lower TGF-β protein expression than control mice ( Figure 5E). Moreover, both endothelial and proximal tubular Adam17 deletion reduced or attenuated TGF-β production in diabetic mice.
An increase in Fn gene expression was seen in all diabetic mice and Fn gene expression was attenuated only in diabetic eAdam17KO mice ( Figure 5).
To gain an insight into the role of Adam17 deletion on renal fibrosis, we examined cortical α-SMA expression and collagen depositions in the studied groups. As depicted in Figure 5A-C, diabetes induced α-SMA accumulation in the interstitial compartment of both animal models. However, this effect was subtly lost on diabetic eAdam17 knockout mice ( Figure 6A). For downstream effectors of the EGFR signaling pathway, we observed downregulated Tgfβ gene expression in diabetic mice except for the eAdam17KO mice. Interestingly, only eAdam17 deletion ameliorated the diabetes effect on renal Tgfβ gene expression (Figure 5C,D). At the protein level, only diabetic wild-type mice increased TGF-β ( Figure  5E,F). Surprisingly, diabetic eAdam17KO mice presented lower TGF-β protein expression than control mice ( Figure 5E). Moreover, both endothelial and proximal tubular Adam17 deletion reduced or attenuated TGF-β production in diabetic mice. An increase in Fn gene expression was seen in all diabetic mice and Fn gene expression was attenuated only in diabetic eAdam17KO mice ( Figure 5).
To gain an insight into the role of Adam17 deletion on renal fibrosis, we examined cortical α-SMA expression and collagen depositions in the studied groups. As depicted in Figure 5A-C, diabetes induced α-SMA accumulation in the interstitial compartment of both animal models. However, this effect was subtly lost on diabetic eAdam17 knockout mice ( Figure 6A). In concordance with gene expression data, semiquantitative scoring of collagen deposition showed increased accumulation of collagen fibres in diabetic mice. In contrast to the gene expression profiles, collagen accumulation was downregulated by endothelial and proximal tubular Adam17 deletion in diabetic mice ( Figure 6D-F).
PI3K/Akt signaling pathway is the downstream mediator of the EGFR, important in glucose-induced pro-fibrotic responses in the kidney. While all diabetic mice exhibited a significantly increased pAkt/Akt ratio in renal tissue, eAdam17 deletion attenuated diabetes' effect on pAkt/Akt ratio ( Figure 6).

Fibrotic Markers Expression in Adam17-Deleted HKC-8 Spheroids
To further delineate the effect of tAdam17 deletion against fibrosis, a 3D in vitro culture with HKC-8 cells resembling the human diabetic microenvironment was used.
Mature HKC-8 spheroids expressed proximal tubular transporters GLUT-1 and AQP1 keeping phenotypic characteristics (data not shown). In concordance with our results in mice, wild-type Adam17 spheroids incubated with HG for 72 h increased the expression of α-SMA and type-IV collagen. Interestingly, Adam17 deletion blocked the increase of α-SMA and type-IV collagen expression compared with wild-type cells ( Figure 7A-D).
In concordance with gene expression data, semiquantitative scoring of collagen deposition showed increased accumulation of collagen fibres in diabetic mice. In contrast to the gene expression profiles, collagen accumulation was downregulated by endothelial and proximal tubular Adam17 deletion in diabetic mice ( Figure 6D-F).
PI3K/Akt signaling pathway is the downstream mediator of the EGFR, important in glucose-induced pro-fibrotic responses in the kidney. While all diabetic mice exhibited a significantly increased pAkt/Akt ratio in renal tissue, eAdam17 deletion attenuated diabetes' effect on pAkt/Akt ratio ( Figure 6).

Fibrotic Markers Expression in Adam17-Deleted HKC-8 Spheroids
To further delineate the effect of tAdam17 deletion against fibrosis, a 3D in vitro culture with HKC-8 cells resembling the human diabetic microenvironment was used.
Mature HKC-8 spheroids expressed proximal tubular transporters GLUT-1 and AQP1 keeping phenotypic characteristics (data not shown). In concordance with our results in mice, wild-type Adam17 spheroids incubated with HG for 72 h increased the expression of α-SMA and type-IV collagen. Interestingly, Adam17 deletion blocked the increase of α-SMA and type-IV collagen expression compared with wild-type cells ( Figure  7A-D).

Discussion
We demonstrate that specific Adam17 deletion on endothelial cells prevents renal pro-inflammatory and pro-fibrotic events caused by DN, and specific Adam17 deletion on renal proximal tubular cells protect from diabetic pro-fibrotic events, podocyte loss, and attenuates renal RAS. Remarkably, this is the first evidence of the beneficial effect of Adam17 downregulation either in endothelial or renal proximal tubular cells in experimental DN (Figure 8).

Discussion
We demonstrate that specific Adam17 deletion on endothelial cells prevents renal pro-inflammatory and pro-fibrotic events caused by DN, and specific Adam17 deletion on renal proximal tubular cells protect from diabetic pro-fibrotic events, podocyte loss, and attenuates renal RAS. Remarkably, this is the first evidence of the beneficial effect of Adam17 downregulation either in endothelial or renal proximal tubular cells in experimental DN (Figure 8).  Hyperglycaemia stimulates cells to release pro-inflammatory cytokines into the kidney, leading to the recruitment of fibroblasts, which initiates fibrotic processes [19]. ADAM17 is a disintegrin and metalloprotease associated with the shedding of most pro-inflammatory and pro-fibrotic substrates [15,20,21]. Targeting the specific inflammatory and fibrotic pathways could be an effective approach for the management of DN.
Although paricalcitol, an analogue of the active form of vitamin D with ADAM17 inhibitory properties had shown a reduction in ACR levels in diabetic patients [22]; our study showed that Adam17 deletion either in endothelial or renal proximal tubular cells was unable to prevent albuminuria in the STZ model. The lack of effect on albuminuria in Adam17 knockout mice might be ascribed to the huge ACR variability among animals. This variability could be associated with the urine collection method by morning urine spots.
Regarding renal hypertrophy, we demonstrated that tAdam17 deletion in diabetic mice reduced the KW/BW ratio as previously shown with paricalcitol administration in non-obese diabetic (NOD) mice [11]. Taken together, ADAM17 modulation may exert a positive effect in the diabetic kidney by decreasing renal hypertrophy.
At the glomerular level, our histological analyses revealed glomerular alterations in diabetic mice due to an increased mesangial index, as previously shown [23][24][25]. However, in eAdam17 and tAdam17 deletion, diabetes-induced glomerular hypertrophy and mesangial matrix expansion were lost. These results suggested that tissue-specific deletion of Adam17 from either endothelial cells or proximal tubular cells protected glomeruli from diabetic deleterious effects, probably by decreasing macrophage infiltration in the glomeruli of diabetic animals [26,27].
We and others demonstrated podocyte loss in diabetic mice [1,28,29]. In this line, we observed a decrease in podocyte number in the glomeruli of all diabetic mice. Interestingly, Guo et al. described TNF-α as a key mediator of HG-activated macrophages to induce podocyte apoptosis [27]. This may explain why we observed a decrease in podocyte loss in diabetic tAdam17KO mice. Blocking renal Adam17 and consequently TNF-α signaling at a proximal tubular level might induce a tubular-glomerular feedback recovery that leads to podocyte protection during DN. In this line, Hasegawa et al. proved that molecular changes in proximal tubular cells affected contiguous cells including podocytes [30].
Serum ACE2 activity in diabetes has been widely described [1,31,32] and ADAM17 has been shown to induce this shedding [33]. Our study also showed increased circulating ACE2 activity in diabetic mice, which was reduced by Adam17 deletion in proximal tubular cells. In concordance, Riera et al. demonstrated that paricalcitol administration in NOD mice significantly decreased circulating ACE2 activity in diabetic mice [11]. These results suggest a possible crosstalk between both proteins in which ADAM17 in the proximal tubular cell may transduce cellular signals somehow resulting in increased circulating ACE2.
In concordance with previous studies [1,31], we observed that diabetic mice presented increased kidney ACE2 activity that could be a compensatory beneficial mechanism, counterbalancing the deleterious effect of ANG II accumulation. Interestingly, after eAdam17 deletion, a further increase in renal ACE2 activity was detected in diabetic mice in comparison with diabetic wild-type mice. These data suggested that because of diabetes, more ACE2 was produced at the tubular level and due to the deletion of eAdam17, less ACE2 shedding occurred from the endothelium, resulting in ACE2 accumulation in the renal cortex. In contrast, tAdam17 deletion attenuated the increase of renal ACE2 activity in diabetic mice. We surmise that tAdam17 deletion attenuates ACE2 shedding, leading to more ANG II degradation and therefore decreased activation of inflammatory signaling pathways in diabetic mice, preventing the increase in renal ACE2 activity.
ADAM17 upregulation leads to the activation of TNFR and EGFR signaling pathways, stimulating the shedding of its substrates and inducing a protein matrix accumulation, ending in renal inflammation and fibrosis [13,18,21]. We observed an increase in soluble TNF-α in diabetic wild-type mice that was reduced with a specific deletion of Adam17. These results suggested that blocking Adam17 at a renal level during DN is enough to prevent TNF-α shedding and attenuate renal inflammation. Decreased renal Tnf-α and increased Tnfr1 gene expression were found in diabetic eAdam17KO mice. These results may be due to a possible negative feedback, since the decrease in the release of TNF-α into the circulation could be well explained by the absence of ADAM17. In these circumstances, more TNFR1 may be needed to balance the downstream signaling pathway.
Omote et al. demonstrated increased renal macrophage infiltration in diabetic KK-Ay mice that was reduced by etanercept, a TNF-α inhibitor [34]. In line with these results, we observed increased macrophage infiltration in diabetic wild-type kidneys. Instead, both eAdam17 and tAdam17 deletion reduced F4/80 immunostaining in diabetic mice when compared with diabetic wild-type mice. Additionally, eAdam17KO mice were protected against the increase in Ccl5 gene expression. The attenuation of inflammatory markers suggests that Adam17 deletion ameliorates kidney inflammation by decreasing TNFR signaling.
ADAM17 releases the active ectodomains of EGFR ligands in injured proximal tubular cells. Accordingly, EGFR is persistently activated and induces the pro-inflammatory response favoring kidney fibrosis [16,35]. EGFR ligands are found to be low-affinity (TGFα, epiregulin, and amphiregulin) or high-affinity EGFR ligands (HG-EGF). However, only low-affinity EGFR ligands can be responsible for sustained profibrotic EGFR activation [35]. In concordance, we observed higher renal Tgfα gene expression but not increased Hb-egf gene expression in diabetic mice.
EGFR hyperactivation induces renal fibrosis by stimulating TGF-β signaling [36]. Activation of the TGF-β signaling pathway favors DN progression through enhancing synthesis of collagen, fibronectin, and laminin and blocking matrix metalloproteinase-mediated extracellular matrix degradation [37]. EGFR transactivation has been demonstrated to be required for TGF-β upregulation by high-glucose [38]. In concordance, we also observed increased Tgfβ and Fn gene expression and increased TGF-β and pAkt protein expression in diabetic mice. Overstreet et al. showed that TGF-β protein expression in tubular epithelial cells was decreased after the administration of erlotinib, an EGFR inhibitor [39]. We observed that only eAdam17 deletion prevents the activation of pAKT and the upregulation of Tgf-β gene expression in the diabetic renal cortex. However, both eAdam17 and tAdam17 deletion protects from TGF-β and collagen accumulation. In this line, immunofluorescence on HKC-8 spheroids showed increased accumulation of type-IV collagen when incubated in HG medium that was decreased with Adam17 deletion. These findings suggest that in Adam17-deleted spheroids, EGFR activation is reduced due to less shedding of the EGFR substrates. Hence, TGF-β signaling is downregulated, leading to less type-IV collagen production.
In an experimental model of ischemia-reperfusion injury, Kefaloyianni et al. demonstrated that tAdam17 deletion decreases α-SMA expression after kidney injury [35]. In this line, our results demonstrated that both eAdam17 and tAdam17 deletion prevent the accumulation of α-SMA induced by diabetes. Immunofluorescence on HKC-8 spheroids strengthen the fact that tAdam17 deletion exerts a higher effect on blocking kidney fibrosis.

Animal Models and Diabetes Induction
Experiments were performed on wild-type and endothelial or proximal-tubular Adam17KO male mice on a C57BL/6 background. Mice were housed in ventilated cages with full access to chow and water. Diabetes was induced in 12-week-old mice by two intraperitoneal injections of 150 mg/kg streptozotocin (STZ) (Sigma, St Louis, MO, USA) on 4 h-fasted mice in 2 consecutive weeks as previously described [28]. Induction of diabetes was confirmed by measuring non-fasting blood glucose (BG) with the ACCU-CHEK Compact System (Roche, Basel, Switzerland) on blood samples from the caudal vein. Mice were considered diabetic when BG > 13.88 mmol/L during the first 4 weeks after STZ administration.
Diabetic mice were followed for 19 weeks after diabetes induction. At the end of the follow-up, mice were sacrificed by terminal surgery as previously described [28]. Blood was extracted by cardiac puncture and serum was obtained by centrifugation at 6000 g for 10 min. Mice were perfused with cold PBS prior to kidney removal. Half of the right kidney was embedded in OCT (Sakura Finetek, Alphen aan den Rijn, Netherlands) medium and the other half was maintained in 10% formalin solution (Sigma-Aldrich, Saint Louis, MO, USA) for paraffin embedding. The remaining tissue was snap frozen on liquid nitrogen and kept at −80 • C for further analyses.

Generation of the Specific Endothelial or Tubular Adam17KO Mice
We conditionally deleted Adam17 in endothelial (eAdam17) or renal proximal tubular (tAdam17) cells. Adam17 flox/flox mice (kindly provided by Dr. Raines, Washington Uni-versity, Saint Louis, MO, USA) [40], tamoxifen-inducible platelet-derived growth factor (Pdgf)-iCreER mice (kindly provided by Dr. Fruttiger, University College London, London, UK) [41] and spontaneous phosphoenolpyruvate carboxykinase (Pepck)-CreER mice (kindly provided by Dr. Haase, Vanderbilt University, Nashville, TN, USA) [42] were previously described. The inducible and spontaneous Cre recombinase was fused to the PDGF or PEPCK promoter, allowing the generation of Adam17 deletion in endothelial or proximal tubular cells, respectively. The Pepck-CreER transgene contains a mutated version of the Pepck promoter, which reduces Pepck expression in the liver by 60% and increases Pepck expression in the kidney by 10-fold [43]. Adam17 flox/flox mice presented two loxP sites surrounding the exon 5 of the Adam17 gene. After recombination, the excision of the sequence results in a frame shift producing a nonsense protein [40] (Supplementary Figure S1). To induce Adam17 gene recombination in endothelial cells, five intraperitoneal doses of 0.1 mg/g body weight of tamoxifen were administrated to 10-week-old mice. Wild-type (WT) mice receiving tamoxifen were used as controls. These animals did not carry any loxP site surrounding the exon 5 of the Adam17 gene, so a recombination did not occur. Then, all WT mice expressed ADAM17 in all cell types.

Urine Albumin Creatinine Ratio
Urinary albumin excretion (UAE) was determined using the albumin-to-creatinine (ACR) ratio on morning spot urine collections obtained on the last week of the follow-up. Urinary albumin levels were measured by an ELISA kit (Albuwell M, Exocell, Philadelphia, PA, USA). Creatinine levels were measured by a colorimetric assay (Creatinine Companion, Exocell, Philadelphia, PA, USA). The albumin-to-creatinine ratio was calculated and expressed as µg Alb/mg Crea [11].

Immunohistochemistry on Paraffined-Embedded Tissue
Paraffin-embedded tissues were cut into 3 µm sections, deparaffined in xylene, and rehydrated through graded alcohols. Sections were stained with periodic acid-Schiff (PAS) for glomerular area and mesangial matrix expansion measurements as previously reported [1]. Twenty microphotographs of glomeruli were taken at ×400 magnification for each animal.
Binding of the antibodies was detected by oxidation of DAB using the Liquid DAB+ Substrate Chromogen System (Dako, Santa Clara, CA, USA). Samples were counterstained with hematoxylin and dehydrated through graded alcohols and preserved with DPX mounting media (Sigma-Aldrich, Saint Louis, MO, USA). Twenty microphotographs of glomeruli stained with anti-WT-1 were taken at ×400 magnification. Six microphotographs of renal cortex stained with anti-α-SMA or anti-F4/80 were taken at ×100 or ×200 magnification respectively. All analyses were performed on ImageJ software v1.51j8 (Bethesda, MD, USA).

Soluble TNF-α ELISA
Serum TNF-α levels were measured using the Mouse TNF-α Quantikine ELISA Kit (R&D Systems, Minneapolis, MN, USA) according to the manufacturer's instructions. Next, 50 µL of mouse serum were incubated with the Assay Diluent for 2 h in microplates coated with a monoclonal antibody specific for mouse TNF-α. After washing unbound substances, an enzyme-linked polyclonal antibody specific for mouse TNF-α conjugated to HRP was added and incubated for 2 h. After washing, the plate was incubated with the Substrate Solution for 30 min. The enzyme reaction yields a blue product that turns yellow when the Stop Solution was added. Finally, optical density was determined using the Tecan Infinite 200 reader at 450 nm and 570 nm for wavelength correction. Results were expressed as pg/mL. To obtain the HKC-8 spheroids, the 3D Life dextran hydrogel kit (BioCat, Heidelberg, Germany) was functionalized with RGD peptide (BioCat, Heidelberg, Germany). The HKC-8 single cell suspension (around 57,600 cells) was mixed with the RGD-functionalized dextran hydrogel and crosslinked with PEG-based crosslinker as previously described [45].
Thirteen days post-seeding, the spheroids were incubated with either of the final concentrations of 35 mM D-glucose (HG, high-glucose), 5.5 mM D-glucose (LG, low-glucose), or 35 mM mannitol (M) (Sigma-Aldrich, Saint Louis, MO, USA) as an osmotic control for 72 h.
Adam17 deletion on HKC-8 cells was performed using the human TACE CRISPR/Cas9 KO plasmid (sc-400827, Santa Cruz Biotechnology, Dallas, TX, USA). The TACE CRISPR/Cas9 KO Plasmid consists of a pool of three plasmids, each encoding for a GFP protein, the Cas9 nuclease, and a target-specific 20 nt guide RNA designed for maximum knockout efficiency.
Positive transfected cells were selected by GFP+ sorting. Cells were resuspended in PBS + 1% FBS and filtered directly before sorting using cell strainer caps in 5 mL FACS tubes (Corning, Corning, NY, USA). Non-transfected cells were used as controls to estimate background signals. The cell sorting was performed using the FACSAria III sorter (BD Biosciences, San Jose, CA, USA). GFP+ cells were collected in 5 mL tubes with complete DMEM-F12 medium and centrifuged at 300 g for 5 min. After aspiration, cells were resuspended in complete DMEM-F12 medium and seeded in the single well of a 24-well plate.
Microscopical images from the middle of the spheroid core were acquired using a HC APO CS2 20x/0.75 IMM on a Leica SP8 inverted confocal microscope (Leica, Wetzlar, Germany). The 3D image stack was reconstructed using Imaris Software (Bitplane, Zürich, Switzerland). The experiment was repeated three times and eight to ten images were taken from each slide. The staining intensity was normalized to the spheroid volume.

Statistical Analyses
Statistical analyses between groups were performed by one-way ANOVA test (SPSS 22.0 software, Armonk, NY, USA). Non-parametric Kruskal-Wallis tests were performed between groups. Non-parametric Mann-Whitney tests were used for group-to-group comparisons. Data were expressed as mean ± SD. Significance was defined as p < 0.05.

Conclusions
In conclusion, this study shows that tissue-specific Adam17 deletion protects against renal inflammation and fibrosis. Diabetic mice lacking specific eAdam17 displayed attenuated renal inflammation and ameliorated renal fibrosis. Adam17 deletion in the renal proximal tubular cells of diabetic mice attenuated podocyte loss, RAS, renal inflammation and fibrosis in terms of macrophage infiltration, collagen and α-SMA accumulation. The study of fibrotic markers on a renal proximal tubular 3D in vitro cell culture reinforced the animal findings. All things considered, these results suggest that Adam17 should be considered as a therapeutic strategy for treating DN.