PAI-1 interaction with sortilin-related receptor 1 is required for lung fibrosis

Abstract

Mutation studies of plasminogen activator inhibitor 1 (PAI-1) have previously implied that PAI-1 promotes lung fibrosis via a vitronectin-dependent (VTN-dependent) mechanism. In the present study, employing 2 distinct murine fibrosis models and VTN-deficient mice, we found that VTN is not required for PAI-1 to drive lung scarring. This result suggested the existence of a profibrotic interaction involving the VTN-binding site on PAI-1 with an unidentified ligand. Using an unbiased proteomic approach, we identified sortilin-related receptor 1 (SorLA) as the most highly enriched PAI-1 binding partner in the fibrosing lung. Investigating the role of SorLA in pulmonary fibrosis demonstrated that deficiency of this protein protected against lung scarring in a murine model. We further found that SorLA is required for PAI-1 to promote scarring in mice, that both SorLA and PAI-1 protein levels are increased in human idiopathic pulmonary fibrosis (IPF) explants, and that these proteins are associated in IPF tissue. Finally, confocal microscopy showed that expression of SorLA in CHO cells increased cellular uptake of PAI-1, and these proteins colocalized in the cytoplasm. Together, these data elucidate a mechanism by which the potent profibrotic mediator PAI-1 drives lung fibrosis and implicate SorLA as a potential therapeutic target in IPF treatment.

Keywords: Aging; Fibrosis; Plasmin; Protein traffic; Pulmonology.

Figure 1. The profibrotic effects of PAI-1 following AEC2 injury are independent of VTN.

(A) Schematics of murine fibrosis models. In the targeted AEC2 injury model, DT (12.5 μg/kg) was administered for 14 days to (i) DTR+ mice, (ii) DTR+:PAI-1^–/– mice, (iii) DTR+:VTN^–/– mice, and (iv) DTR+:PAI-1^–/–:VTN^–/– mice. A control cohort of WT (DTR–) mice treated with DT was also included in the study protocol. In the single-dose bleomycin model, bleomycin was administered (2.5 U/kg in 50 μL by the oropharyngeal route) on day 0 to (i) WT mice, (ii) PAI-1^–/– mice, (iii) VTN^–/– mice, and (iv) double-knockout PAI-1^–/–:VTN^–/– mice. A group of uninjured WT mice were included as a negative control. (B) In the targeted AEC2 injury model, mice were weighed at regular intervals. (C and E) Lungs were harvested on day 21 (D21) and analyzed for hydroxyproline content. (D and F) BAL samples were obtained on D21 and assayed for total PAI-1 levels. Results in B–F are reported as the mean concentration ± SEM. n = 6–7 (B), n = 7–11 (C), n = 5–11 (D), n = 5–11 (E), and n = 5–8 (F). Representative data are displayed from 1 of 3 (A and B) and 1 of 2 (D) experiments. Significant P values are shown from 2-way ANOVA with Tukey’s multiple-comparison test. (G) H&E- and Picrosirius red–stained D21 lung sections from a representative animal in the targeted AEC2 injury model. Scale bars: 180 μm.

Figure 2. Reconstitution of PAI-1^–/–:VTN^–/– mice with PAI-1_RR restores pulmonary fibrosis in a VTN-independent manner.

(A) Schematics of murine fibrosis models. In the targeted AEC2 injury model, DT (10.0 μg/kg) was administered for 14 days to DTR+ and DTR+:PAI-1^–/–:VTN^–/– mice. On day 11 (D11), DT-injured DTR+:PAI-1^–/–:VTN^–/– mice received either i.p. recombinant PAI-1_RR (deficient antiprotease activity but intact VTN binding) at 100 μg twice daily or an equivalent volume of PBS. Control cohorts included DT-injured DTR+ mice and uninjured DTR+ mice (both without PAI-1_RR reconstitution). In the single-dose bleomycin model, bleomycin was administered (2.5 U/kg) to double-knockout PAI-1^–/–:VTN^–/– mice. Beginning on D11, PAI-1^–/–:VTN^–/– mice were treated with either recombinant PAI-1_RR or PBS. Control cohorts of mice receiving PBS in place of PAI-1_RR included bleomycin-injured and uninjured WT mice. (B and C) Lungs were harvested on D21 and analyzed for hydroxyproline content. (D) Harvested lungs were inflating fixed, sectioned, and stained with H&E (top panels) and Picrosirius red (bottom panels). Scale bars: 180 μm. Results in B and C are reported as the mean concentration ± SEM. n = 7–10 (B), n = 11–15 (C). Representative data are displayed from 1 of 3 (B) and 1 of 2 experiments (C). Significant P values are shown for comparisons performed using 2-way ANOVA with Tukey’s multiple-comparison test.

Figure 3. PAI-1 binds to SorLA as revealed by SPR analysis.

Increasing concentrations of PAI-1_WT or PAI-1_AK were flowed over SPR flow cells coupled with soluble SorLA (A) or the VSP10 domain (B). Data were analyzed by equilibrium analysis by fitting the association data to a first-order process to determine R_eq. The data were normalized to R_max and plotted as R_eq/R_max versus PAI-1 concentrations. Data were fit to a single binding site by nonlinear regression analysis. (C) Inhibition of PAI-1 (250 nM) binding to SorLA in the presence of increasing concentrations of somatomedin B (SMB) domain of VTN.

Figure 4. PAI-1_WT binds to SorLA in lung tissue homogenates from patients with IPF.

(A) Fibrotic lung tissue obtained from explants at the time of transplant were homogenized in binding buffer. Each sample (200 μg) was incubated with either uncoated magnetic streptavidin-Sepharose beads or beads coated with biotin-tagged PAI-1_WT. Beads were collected, washed, and proteins were eluted with SDS loading buffer. The initial homogenate (input) and the eluted proteins (Beads, PAI-1-Beads) were separated by SDS-PAGE, blotted, and stained with an anti-SorLA antibody. Data are displayed as a representative gel. (B) Equal quantities of protein from homogenized IPF or normal control lung tissue were separated by SDS-PAGE and analyzed by Western blotting for SorLA and αSMA (normalized to vinculin, n = 13). (C) Quantification of SorLA and αSMA in fibrotic tissue. (D) Active PAI-1 levels measured by ELISA (normalized to total lung protein concentration, n = 13). (E–K) Immunofluorescent costaining of PAI-1 (Akoya Opal 520, green) and SorLA (Akoya Opal 690, red) in normal (E and F) and IPF (G–K) human lung tissue sections. (I–K) Enlargement of panel H with (I) DAPI and PAI-1, (J) DAPI and SorLA, and (K) merged. Scale bars: 20 μm. Data are represented as mean ± SEM. Significant P values are shown for comparisons performed using a parametric 2-tailed t test (Panel C, D) or Mann–Whitney U nonparametric 2-tailed t test.

Figure 5. SorLA-deficient and SorLA-heterozygous mice are protected from fibrosis following lung injury.

(A–C) Single-dose bleomycin (2.5 U/kg in 50 μL) was administered on day 0 (D0) to the following littermate cohorts: (i) SorLA^+/+, (ii) SorLA^+/–, and (iii) SorLA^–/– mice. Control littermates (mix of SorLA^+/+, SorLA^+/–, SorLA^–/–) were uninjured and served to establish baseline lung collagen content. (A) Mice were weighed at regular intervals (n = 7–8/group) and (B and C) lungs and BAL were collected on D21 for hydroxyproline analysis and measurement of PAI-1 concentration (n = 7–8). Representative data from 1 of 3 experiments are shown (B), and results are reported as the mean ± SEM. P values are displayed for comparisons performed using 2-way ANOVA with Tukey’s post hoc multiple-comparison test. (D) Lung sections obtained on D21 were stained with H&E (left panel) and Picrosirius red (right panel) and representative images are shown. Scale bars: 180 μm. Single-dose bleomycin was administered (2.5 U/kg in 50 μL) on D0 to the following littermate cohorts: (i) PAI-1^–/–:SorLA^+/+ and (ii) PAI-1^–/–:SorLA^–/–. On D11, subsets of mice from each genotype were administered either i.p. recombinant PAI-1_WT at 100 μg twice daily or an equivalent volume of PBS. On D21, lungs were analyzed for hydroxyproline content. (E) Mean hydroxyproline values in each group (n = 17–24/group). (F) Delta in hydroxyproline in PAI-1^–/–:SorLA^–/– mice treated with/without PAI-1_WT and in PAI-1^–/–:SorLA^+/+ mice treated with/without PAI-1_WT. Data are reported as mean ± SEM. P values are shown from 2-way ANOVA with Tukey’s multiple-comparison test (B and C), grouped 2-way ANOVA with Šidák’s test for multiple comparisons (E), and an unpaired t test (F). NS, not significant.

Figure 6. PAI-1 colocalizes with SorLA in cells.

CHO cells were transfected overnight with a SorLA-GFP expression construct and then incubated with 100 nM Alexa Fluor 594–labeled PAI-1 for 1 hour. (A) Representative confocal images showing DAPI (blue), phalloidin (white), SorLA (green), and PAI-1 (red). (B) Merged image shows colocalization (yellow) of PAI-1 and SorLA in cells (n = 10, MOC = 0.71 ± 0.03). Rectangular panels below (x–z) and to the right (y–z) of the merged image are orthogonal views showing SorLA and PAI-1 overlap. (C) Mean PAI-1 in SorLA positive (+) versus SorLA negative (–) cells (n = 10, P = 0.0002). Data shown as mean ± SEM. Scale bars: 10 μm for images and 2 μm for z-planes.