Cathepsin S Cleavage of Protease-Activated Receptor-2 on Endothelial Cells Promotes Microvascular Diabetes Complications

Abstract

Endothelial dysfunction is a central pathomechanism in diabetes-associated complications. We hypothesized a pathogenic role in this dysfunction of cathepsin S (Cat-S), a cysteine protease that degrades elastic fibers and activates the protease-activated receptor-2 (PAR2) on endothelial cells. We found that injection of mice with recombinant Cat-S induced albuminuria and glomerular endothelial cell injury in a PAR2-dependent manner. In vivo microscopy confirmed a role for intrinsic Cat-S/PAR2 in ischemia-induced microvascular permeability. In vitro transcriptome analysis and experiments using siRNA or specific Cat-S and PAR2 antagonists revealed that Cat-S specifically impaired the integrity and barrier function of glomerular endothelial cells selectively through PAR2. In human and mouse type 2 diabetic nephropathy, only CD68(+) intrarenal monocytes expressed Cat-S mRNA, whereas Cat-S protein was present along endothelial cells and inside proximal tubular epithelial cells also. In contrast, the cysteine protease inhibitor cystatin C was expressed only in tubules. Delayed treatment of type 2 diabetic db/db mice with Cat-S or PAR2 inhibitors attenuated albuminuria and glomerulosclerosis (indicators of diabetic nephropathy) and attenuated albumin leakage into the retina and other structural markers of diabetic retinopathy. These data identify Cat-S as a monocyte/macrophage-derived circulating PAR2 agonist and mediator of endothelial dysfunction-related microvascular diabetes complications. Thus, Cat-S or PAR2 inhibition might be a novel strategy to prevent microvascular disease in diabetes and other diseases.

Keywords: albuminuria; diabetic nephropathy; endothelial cells; macrophages.

Figure 1.

Extrinsic and intrinsic Cat-S triggers endothelial cell injury and microvascular permeability through PAR2 in vivo. (A) C57BL/6 mice were injected with a single dose of 10 µg Cat-S ± pretreatment with two different Cat-S antagonists. Urinary albumin-to-creatinine (A/C) ratio was quantified at several time points as indicated. Data are means±SEMs; P<0.05 versus baseline. (B) At 24 hours, transmission EM of glomeruli showed massive endothelial cell swelling with vacuolization in kidneys from Cat-S–treated mice compared with healthy kidneys, which was prevented by pretreatment with both Cat-S inhibitors. (C and D) Also, PAR2 inhibition and lack of the Par2 gene had the same protective effect on albuminuria and glomerular ultrastructure. (E and F) FITC dextran leakage observed by intravital microscopy was used as a marker of microvascular permeability in the postischemic (ischemia-reperfusion) cremaster muscle of wild-type and Par2-deficient mice. Data are means±SEMs of four mice in each group. Representative images of FITC dextran leakage from postcapillary venules are shown. Original magnification, ×40. *P<0.05 versus PBS/vehicle group; **P<0.01 versus PBS/vehicle group.

Figure 2.

Cat-S specifically induces ED through PAR2 in vitro. (A and B) In vitro ECIS studies with GEnCs. (A) GEnC monolayers were exposed to increasing doses of Cat-S, and cell capacitance at 40 kHz was determined over a period of 9 hours. Note the dose-dependent increase that occurs very quickly on Cat-S exposure. (B) Cat-S–induced increase of cell capacitance was reversed by RO5461111. Graphs are readings of single experiments representative of at least three experiments for each condition. (C) GEnC monolayers were imaged by scanning EM after treatment as indicated. Representative images are shown. Note that either Cat-S (RO5461111) or PAR2 inhibition protects GEnCs from the Cat-S–induced monolayer disintegration. (D) Cat-S–induced reactive oxygen species (ROS) production in GEnCs was determined by electron spin resonance. A PAR2-activating peptide (AP) served as a positive control. (E) Transwell endothelial cell monolayer permeability assays with FITC albumin. Data represent FITC fluorescence in the lower well 1 hour after stimulation with Cat-S and/or PAR2 inhibitor. Note that the Cat-S effects are reversed by a PAR2 inhibitor. *P<0.05 versus control; ^#P<0.05 versus PAR2-AP or Cat-S. (F and G) Cell culture supernatant of GEnCs exposed to two different concentrations of Cat-S was quantified for (F) floating cells or (G) annexin positivity by flow cytometry as indicated. The cells had been pretreated with siRNA with either scrambled sequence or specific sequences for PAR1, PAR2, PAR3, and PAR4. Data are means±SEMs of three identical experiments. *P<0.05 versus medium.

Figure 3.

Cathepsin S is expressed by macrophages infiltrating the human kidney. Cat-S immunostaining in human DN. Archived kidney biopsies were stained for Cat-S. Representative images are shown at original magnifications of ×100, ×200, and ×1000. (A) A nondiabetic control kidney shows strong Cat-S positivity in proximal tubules. At a magnification of ×1000, some positivity is noted in parietal epithelial cells as well as in podocytes in a cytoplasmic staining pattern. (B) In a patient with DN, Cat-S positivity localizes to infiltrating leukocytes inside the glomerulus. At a magnification of ×1000, positivity is noted in leukocytes within capillary lumen and mesangium as well as in GEnCs. (C) In a patient with advanced DN, Cat-S positivity localizes to interstitial cell infiltrates. (D) Dual staining for Cat-S (brown) and CD68 (red) identifies CD68+ macrophages as a source of intrarenal Cat-S expression. (E) In situ hybridization does not display any Cat-S mRNA in normal (panel 1) and diabetic glomeruli. In advanced DN, Cat-S mRNA was detected in interstitial cells that show a positive signal for CD68 (arrows). Original magnification, ×400.

Figure 4.

Cathepsin S is expressed by macrophages infiltrating the mouse kidney. Cat expression in kidneys of db/db mice. (A) mRNA expression levels of cysteine Cats in kidneys from 6-month-old nondiabetic and diabetic db/db mice; db/db mice were either sham operated (2K) or uninephrectomized (1K) at 6 weeks of age. Data are means±SEMs of 5–12 mice in each group, and the values given are normalized to 18S rRNA. (B) Western blot for Cat-S from kidney tissue obtained from the same mice. β-Actin is shown as a protein loading control. (C) Cat-S immunostaining on the same kidneys localizes Cat-S protein mostly to parietal and proximal tubular epithelial cells. (D and E) In situ hybridization of mouse kidneys for Cat-S colocalizes with EMR1 mRNA expression, indicating Cat-S expression in infiltrating mononuclear phagocytes. In contrast, no Cat-S signal is detectable in renal parenchymal cells. EMR1, epidermal growth factor-like module containing mucin-like hormone receptor 1; KO, knockout; WT, wild type. Original magnification, ×400. ^#P<0.05 versus wild type; *P<0.05 versus sham group.

Figure 5.

RO5461111 selectively inhibits Cat-S in vivo and prevents glomerulosclerosis in db/db mice. (A) db/db Mice were treated from months 4 to 6 of age with RO5461111, which significantly increased the spleen levels of the Cat-S substrate Lip10 compared with vehicle-treated mice of the same age. The graph represents the ratio of Western blot signals for p10 and the invariant chain Li. The effect is indicative of target inhibition. (B) Cat-S enzymatic activity in plasma was determined by a substrate assay in samples obtained at 16 and 24 weeks as indicated from vehicle–treated 1K db/db mice (black bars) and RO5461111–treated 1K db/db mice (white bar). (C) Kidney Cat-S protein levels were determined by Western blot. The respective β-actin expression served as protein loading control. (D) The respective quantification of the Western blot results is shown as means±SDs. (E) The spleen Cat-S mRNA expression levels were quantified by real–time RT-PCR and are expressed as the ratio to the respective 18s rRNA expression levels. Data in A, B, D, and E are means±SEMs from 8 to 12 mice in each group. Kidney sections from both treated and untreated mice were stained with (F and G) anti-CD31 to see GEnCs and quantification of CD31 signal in each glomeruli and (H) periodic acid–Schiff (PAS) reagent to (I) quantify glomerulosclerosis. Representative glomeruli are shown from each group at an original magnification of ×400; 50 glomeruli were scored by using a semiquantitative PAS scores ranging from zero to four. (J) Silver-stained sections were digitally scored as a marker of fibrous extracellular matrix. The graphs in G, I, and J illustrate the mean percentage of each score±SEM from all mice in each group. Note that uninephrectomy was associated with a shift toward higher scores as seen in vehicle-treated mice, but with RO5461111 treatment, the overall scores were significantly reduced. WT, wild type. *P<0.05 versus vehicle-treated group; **P<0.01 versus vehicle-treated group; ***P<0.001 versus vehicle-treated group; ^#P<0.05 versus the earlier time point.

Figure 6.

Cat-S inhibition protects the glomerular filtration barrier in db/db mice. (A) Urinary albumin-to-creatinine ratios were determined as a functional marker of the glomerular filtration barrier in 6-month-old 1K db/db mice with or without treatment. (B) RNA isolates from kidneys of both treated and untreated 6-month-old 1K db/db mice underwent quantitative real–time PCR for the podocyte–specific genes nephrin and podocin as indicated. (C) Renal sections from mice of both treated and untreated groups were stained for Wilms Tumor-1 (WT-1) and nephrin. Original magnification, ×400. (D) The graph shows the mean number of WT-1–positive cells in 15 glomeruli ±SEM in sections from 6-month-old 1K db/db mice with or without treatment. Note the potent effect of RO5461111 on the number of podocytes. Data in A–D are means±SEMs of 5–12 mice in each group. (E) Transmission EM of glomeruli from 1K db/db mice at 6 months displayed morphologic signs of glomerular endothelial damage, such as cytoplasmic swelling with whorling cytoplasmic inclusions and loss of fenestrations (arrows) in vehicle–treated 1K db/db mice (left panel). In addition, podocytes showed partial foot process effacement as a sign of podocyte injury (asterisks). Treatment with RO5461111 prevented these abnormalities (right panel). (F) Real–time RT-PCR for VCAM and eNOS on total renal mRNA as markers of ED. Data in B and E are expressed as means of the ratio of the specific mRNA versus those of 18S rRNA ±SEMs. VCAM, vascular cell adhesion molecule. *P<0.05 versus vehicle group; **P<0.01 versus vehicle group.

Figure 7.

Cat-S inhibition attenuates DR of db/db mice. Vascular leakage of serum albumin from isolectinB4–labeled blood vessels is reduced in (C) RO5461111–treated db/db mice compared with (A) vehicle-treated mice. B and D show albumin leakage at a higher magnification. Scale bar, 50 μm. (E and F) Vehicle–treated 1K db/db mice displayed increased glial fibrillary acidic protein (GFAP) expression in retinal astrocytes, which was reduced by RO5461111. (G and H) RO5461111 also restored retinal eNOS expression. (I–K) The respective quantitative analysis is shown below. Data represent means±SDs of five retina cross-sections per mouse and five mice per group. GCL, ganglion cell layer; INL, inner nuclear layer; IPL, inner plexiform layer; ONL, outer nuclear layer; OPL, outer plexiform layer. Scale bars, 100 μm. ***P<0.001 versus vehicle.

Figure 8.

PAR2 inhibition attenuates glomerulosclerosis and retinopathy in db/db mice. Male uninephrectomized db/db mice were treated with either vehicle or PAR2 inhibitor with or without RO5461111 from months 4 to 6 of age. (A) Kidney sections from untreated (left panel), PAR2 inhibitor (center panel), and PAR2 and Cat-S inhibitor–treated (right panel) mice were stained with anti-CD31 to see GEnCs, and (B) periodic acid–Schiff (PAS) -stained sections shown at an original magnification of ×400 were used for scoring of glomerulosclerosis (0, no lesion; 1, ≤25%; 2, ≤50%; 3, ≤75%; 4, ≤75% sclerosis). Data are mean scores of 50 glomeruli per mouse±SEM. (C–E) quantification of CD31 fluorescence intensity, average PAS score, and mean silver area in the kidneys of the above groups. (F and G) Retinas were prepared form the same mice, and albumin leakage was assessed by immunostaining. Data represent means±SDs of five retina cross-sections per mouse and five mice per group. GCL, ganglion cell layer; INL, inner nuclear layer; IPL, inner plexiform layer; ONL, outer nuclear layer; OPL, outer plexiform layer. Scale bars, 100 μm. *P<0.05 versus vehicle; **P<0.01; ***P<0.001. DAPI, 4′,6-diamidino-2-phenylindole.