Mesangial cell integrin αvβ8 provides glomerular endothelial cell cytoprotection by sequestering TGF-β and regulating PECAM-1

Abstract

Integrins are heterodimeric receptors that regulate cell adhesion, migration, and apoptosis. Integrin αvβ8 is most abundantly expressed in kidney and brain, and its major ligand is latent transforming growth factor-β (TGF-β). Kidney αvβ8 localizes to mesangial cells, which appose glomerular endothelial cells and maintain glomerular capillary structure by mechanical and poorly understood paracrine mechanisms. To establish kidney αvβ8 function, mice with homozygous Itgb8 deletion (Itgb8(-/-)) were generated on outbred and C57BL/6 congenic backgrounds. Most Itgb8(-/-) mice died in utero, and surviving Itgb8(-/-) mice failed to gain weight, and rarely survived beyond 6 weeks. A renal glomerular phenotype included azotemia and albuminuria, as well as increased platelet endothelial cell adhesion molecule-1 (PECAM-1) expression, which was surprisingly not associated with conventional functions, such as endothelial cell hyperplasia, hypertrophy, or perivascular inflammation. Itgb8(-/-) mesangial cells demonstrated reduced latent TGF-β binding, resulting in bioactive TGF-β release, which stimulated glomerular endothelial cell apoptosis. Using PECAM-1 gain and loss of function strategies, we show that PECAM-1 provides endothelial cytoprotection against mesangial cell TGF-β. These results clarify a singular mechanism of mesangial-to-endothelial cell cross-talk, whereby mesangial cell αvβ8 homeostatically arbitrates glomerular microvascular integrity by sequestering TGF-β in its latent conformation. Under pathological conditions associated with decreased mesangial cell αvβ8 expression and TGF-β secretion, compensatory PECAM-1 modulation facilitates glomerular endothelial cell survival.

Figure 1

Integrin β8 is expressed in a mesangial distribution. Unfixed frozen sections from Itgb8^+/+ (A, +/+) and Itgb8^−/− (B, −/−) mouse kidneys were permeabilized in 0.2% Triton X-100, blocked with goat serum, followed by incubation with rabbit anti-human β8 integrin IgG, then Alexa 488-conjugated goat anti-rabbit IgG (green), and TOTO-3 nuclear counterstain (blue). Slides were viewed by confocal microscopy at 1000X magnification. Digital images were processed with Molecular Devices deconvolution software v9.1 and Adobe Photoshop v7.0.

Figure 2

Itgb8^−/− mice have extrarenal phenotypes. A:Itgb8 genotypes were established by PCR analysis. Representative Itgb8^+/+ (+/+) and Itgb8^−/− (−/−) mice on outbred (B) and C57BL/6 congenic (C) backgrounds. Body weights (D), kidney weights (E) and kidney:body weight ratios (F) are shown from 4-week-old outbred male mice with Itgb8^+/+ (+/+), Itgb8^+/− (+/−), and Itgb8^−/− (−/−) genotypes (n = 4 to 8 per group). *P < 0.05 compared to +/+ group.

Figure 3

Itgb8^−/− mice have glomerular dysfunction. A: Spot urine samples were assayed for albumin from 4-week-old and 12-week-old Itgb8^+/+ (+/+) and Itgb8^−/− (−/−) outbred and C57BL/6 congenic mice by enzyme-linked immunosorbent assay methods. N = 6 mice per group. B: Blood urea nitrogen (BUN) was measured in 4-week-old Itgb8^+/+ (+/+), Itgb8^+/− (+/−), and Itgb8^−/− (−/−); and 12-week-old Itgb8^+/+ (+/+), Itgb8^+/− (+/−), and Itgb8^−/− (−/−) mice with a Beckman-Coulter LX20 analyzer. N = 3 to 9 mice per group. *P < 0.01 compared to age-matched +/+ mice. Paraffin sections from 12-week-old outbred Itgb8^+/+ (C) and Itgb8^−/− (D) mice were stained with PAS. Images are viewed using light microscopy, original magnifications, 1000. E–H: Transmission electron micrographs of glomeruli from 12-week-old Itgb8^+/+ and Itgb8^−/− kidneys. E: Glomerulus from wild-type (+/+) kidney, original magnification, 4,400. F: Glomerulus from wild-type (+/+) kidney, original magnification, 15,000. G: Glomerulus from Itgb8^−/− (−/−) kidney, original magnification, 4,400. H: Glomerulus from Itgb8^−/− (−/−) kidney, original magnification, 15,000. I: Slit diaphragms, defined as discrete areas of separation between podocyte foot processes, were tabulated from electron micrographs of Itgb8^+/+ and Itgb8^−/− glomeruli. Data are expressed as number of slit diaphragms per μm glomerular basement membrane length. MC, mesangial cells; Arrows, glomerular basement membrane; Arrowhead, normal-appearing, fenestrated glomerular endothelial cells; asterisks indicate normal podocyte foot processes; double asterisks indicate, areas of foot process fusion.

Figure 4

Itgb8^−/− mice display increased glomerular PECAM-1 expression. Frozen sections from 12-week-old Itgb8^+/+ (+/+) (A) and Itgb8^−/− (−/−) (B) kidneys were labeled for endothelial cells by blocking with MOM mouse blocking reagent (Vector, 1 hour), then incubating with rat anti-mouse PECAM-1 (1:200, 1 hour), followed by Texas red-conjugated goat anti-rat IgG (1:300, 1 hour). Slides were viewed by confocal microscopy, original magnification, 1000. (C) PECAM-1 expression was determined by quantitative RT-PCR in isolated glomeruli from Itgb8^+/+ (+/+) and Itgb8^−/− (−/−) kidneys. Data are expressed as relative PECAM-1 mRNA expression (PECAM-1 normalized to β-actin transcript content within the same sample). *P < 0.05 compared to +/+ group.

Figure 5

Increased PECAM-1 expression is not associated with GEC hyperplasia in Itgb8^−/− kidneys. A: GEC proliferation was assessed by co-localization of Ki-67-positive cells (red) within PECAM-1-stained capillaries (green) in a glomerulus from a 4-week-old Itgb8^−/− kidney. B: Quantitation of GEC from 60 glomeruli in three 4-week-old Itgb8^+/+ (+/+) and Itgb8^−/− (−/−) kidneys.

Figure 6

MC integrin β8 and GEC PECAM-1 are synergistically cytoprotective. A: Immunoblots from whole cell lysates (20 μg total protein) from mouse glomerular endothelial cells (GEC), kidney endothelial cells (KEC) from wild-type and PECAM-1^−/− mice, and GEC transfected with adenovirus (Ad) or PECAM-1 adenovirus (AdP). The PECAM-1 band appears as a doublet due either to the presence of more than one isoform or glycosylated and unglycosylated proteins. Blots were stripped and re-probed with anti-tubulin antibodies as a loading control (lower panel). B: Untransfected (mGEC), adenoviral vector-transfected (Ad), and PECAM-1 adenoviral vector-transfected (AdP) mouse GEC were incubated for 24 hours with serum-free DMEM, followed by conditioned media obtained from quiescent wild-type MC (+/+), Itgb8^−/− (−/−) MC or DMEM (D) only (16 hours, 37°C), and then evaluated for apoptosis by annexin V labeling. *P < 0.01 compared to D- and +/+ treated groups. C: Conditionally immortalized endothelial cells derived from PECAM-1^−/− or PECAM-1^+/+ mouse kidneys were grown to near confluence in complete media under permissive 31°C conditions, and then placed in serum-free DMEM (72 hours, 37°C). Cells were then washed with Hanks' solution, and incubated with conditioned media from wild-type (+/+) or Itgb8^−/− (−/−) MC (8 hours, 37°C) and then evaluated for apoptotic nuclear morphology by DAPI labeling by two observers blinded to experimental conditions. *P < 0.01 compared to other groups.

Figure 7

Latent TGF-β binds to MC integrin β8. A–C: Unfixed cryosections from wild-type C57BL/6 mouse kidney were blocked with donkey serum, then incubated with goat anti-human LAP-β1 IgG and rabbit anti-β8 integrin IgG. Sections were washed with PBS, then incubated with Texas red-conjugated donkey anti-goat IgG and Alexa 488-conjugated donkey anti-rabbit IgG. Slides were mounted and viewed with a Leica confocal microscope, original magnification, 630. D–F: Human MC were grown to subconfluence on coverslips. Unfixed cells were labeled for latent TGF-β expression by initially incubating with goat anti-LTBP-1 antibodies (E). To determine integrin β8 expression, slides then were washed in serum-free DMEM, followed by fixation in paraformaldehyde, blocking with donkey serum, and incubation with rabbit anti-β8 integrin antibodies (D). Slides were then incubated with Texas red-conjugated donkey anti-goat and Alexa 488-conjugated donkey anti-rabbit antibodies, and viewed by confocal microscopy, original magnification, 1000. All digital images were analyzed with deconvolution software. Arrows denote regions of colocalization. G–H: Frozen sections from formaldehyde-fixed Itgb8^+/+ (G, +/+) or Itgb8^−/− (H, −/−) kidneys were probed for LAP-β1, as in B. MC derived from Itgb8^+/+ (I, +/+) or Itgb8^−/− (J, −/−) mice were labeled for LTBP-1 as in E.

Figure 8

PECAM-1-deficient GEC are susceptible to apoptosis from TGF-β released by Itgb8^−/− MC. A: Aliquots of conditioned media from Itgb8^+/+ or Itgb8^−/− MC were obtained over 48 hours and then incubated with plasminogen activator inhibitor-1 promoter/luciferase-expressing MLEC for 6 hours. TGF-β bioactivity was determined by luciferase luminescence as described in Methods. *P = 0.05 compared to the Itgb8^+/+ conditioned media group. B: GEC cultured on glass coverslips were incubated for 24 hours with serum-free DMEM in all groups, followed by conditioned media harvested from quiescent Itgb8^+/+ (+/+) or Itgb8^−/− (−/−) MC as in A, or DMEM supplemented with TGF-β1 (2.5 ng/ml) or DMEM only (16 hours, 37°C). GEC were then washed in PBS, fixed in paraformaldehyde, and incubated with rabbit monoclonal anti-phospho-Smad2/3 IgG (1 hour, at room temperature), followed by FITC-conjugated goat anti-rabbit IgG (1 hour,at room temperature). Cells were examined from six different fields and counted for nuclear staining of phospho-Smad2/3 as an index of TGF-β bioactivity (original magnification, 400). *P < 0.05 compared to Itgb8^+/+ conditioned media alone. C: PECAM-1-deficient GEC, adenoviral vector-transfected GEC (Ad), and PECAM-1 adenoviral vector-transfected GEC (AdP) were incubated for 24 hours with serum-free DMEM at 37°C, followed by conditioned media harvested as in A from Itgb8^+/+ (+/+) or Itgb8^−/− (−/−) MC (16 hours, 37°C). In some groups, conditioned media was supplemented with TGF-β1 (2.5 ng/ml) or neutralizing anti-TGF-β1 antibodies (10 μg/ml). GEC were evaluated for apoptosis by nuclear morphology of DAPI-stained cells by two observers blinded to experimental conditions. *P < 0.01 compared to other groups incubated with Itgb8^−/− conditioned media. **P < 0.05 compared to Itgb8^+/+ conditioned media alone.

Figure 9

Schematic model of MC integrin β8 effects on GEC. A: Baseline conditions, whereby integrin β8 serves as a reservoir for TGF-β by maintaining the cytokine in its latent conformation through binding of an RGD tri-peptide recognition sequence within LAP to integrin β8. B: Pathological conditions, such as chronic glomerular diseases, in which MC β8 expression is decreased or MMP-14 activity is stimulated. In the former situation, which is modeled with Itgb8^−/− mice, TGF-β is no longer secluded in its latent form, and TGF-β could then be released and activated by glomerular matrix proteases or GEC-secreted factors. Alternatively, TGF-β could be enzymatically activated by MMP-14, which is upregulated in MC undergoing myofibroblast differentiation. A glomerular capillary phenotype is minimized by GEC PECAM-1 induction, perhaps in response to TGF-β stimulation.C: Based on in vitro data with PECAM-1-deficient endothelial cells and conditioned media from MC, TGF-β released by Itgb8^−/− MC causes apoptosis in PECAM-1-null GEC. By extrapolation, a parallel in vivo scenario (diminished integrin β8 and GEC with insufficiently upregulated PECAM-1 expression) would result in a more severe renal phenotype characterized by GEC apoptosis. β8, integrin β8; GEC, glomerular endothelial cell; LTGF-β, latent TGF-β; MC, mesangial cell; Podo, podocyte; skull and crossbones represents an apoptotic GEC.