Reversal of pulmonary veno-occlusive disease phenotypes by inhibition of the integrated stress response

Abstract

Pulmonary veno-occlusive disease (PVOD) is a rare form of pulmonary hypertension arising from EIF2AK4 gene mutations or mitomycin C (MMC) administration. The lack of effective PVOD therapies is compounded by a limited understanding of the mechanisms driving vascular remodeling in PVOD. Here we show that administration of MMC in rats mediates activation of protein kinase R (PKR) and the integrated stress response (ISR), which leads to the release of the endothelial adhesion molecule vascular endothelial (VE) cadherin (VE-Cad) in complex with RAD51 to the circulation, disruption of endothelial barrier and vascular remodeling. Pharmacological inhibition of PKR or ISR attenuates VE-Cad depletion, elevation of vascular permeability and vascular remodeling instigated by MMC, suggesting potential clinical intervention for PVOD. Finally, the severity of PVOD phenotypes was increased by a heterozygous BMPR2 mutation that truncates the carboxyl tail of the receptor BMPR2, underscoring the role of deregulated bone morphogenetic protein signaling in the development of PVOD.

Extended Data Fig. 1 |. Overexpression of Rad51 prevents MMC-mediated DNA damage, and MMC does not induce apoptosis.

(a) PMVECs (5 × 10³ cells) transfected with empty vector (mock) or Rad51 expression plasmid (Rad51) were subjected to the comet assay. The bar graph indicates the levels of DNA damage as a percentage of nuclei with damaged DNA out of total nuclei as mean ± SEM. Approximately one hundred nuclei per condition were examined. n = 5 independent samples. (b) PMVECs (1 × 10⁶ cells) were transfected with empty vector (mock) or Rad51 expression plasmid (+Rad51), followed by vehicle or MMC treatment for 14 h. Total cell lysates were subjected to immunoblot of VE-Cad, Rad51, γH2AX, and β-actin (control). (c) PMVECs were treated with vehicle (Veh), MMC for 14 h or 0.4 mM hydrogen peroxide (H₂O₂) for 2 h, followed by fluorescence staining of Annexin V (green) and DAPI (blue). Annexin V-positive cells’ fraction (%) is shown as mean ± SEM. Scale bar=10 mm Approximately one hundred cells per condition were examined. A similar result was obtained by flow cytometric analysis of FITC-Annexin V staining. The number of Annexin V-positive cells is shown as mean ± SEM. About 300 cells and 1.5 × 10⁶ cells per condition were analyzed by fluorescent staining and flow cytometry. Because of the autofluorescence of PMVECs, the signals of unstained PMVECs were used to set the gate for FITC-Annexin V-positive cells. A green rectangle indicates Annexin V-positive cells. Data are analyzed by one-way ANOVA with Tukey’s post-hoc test (c, d); two-way ANOVA With Tukey’s post-hoc test (a).

Extended Data Fig. 2 |. Transfer of VE-Cad and Rad51 to PMVECs damaged by MMC and transcriptional activation of ATF4 target genes in MMC-treated PMVECs.

(a) The levels of intracellular VE-Cad, Rad51, and β-actin (loading control) protein in the MMC-treated PMVECs (recipient cells; 1 × 10⁶ cells) after the incubation with the CM from PMVECs (1 × 10⁶ cells) treated with vehicle (CM^veh) or MMC (CM^MMC) for 14 h were examined by immunoblot analysis. (b) IF staining of PMVECs treated with CM^veh or CM^MMC (5 × 10³ cells) with an anti-VE-Cad antibody (green). Cell nuclei were stained with DAPI (blue). Scale bar=10 mm. (c) CM was collected from PMVECs following two h of MMC treatment and was immediately reintroduced to the donor cells’ CM after MMC was removed from the CM^MMC (−MMC, indicated in red), or it was added without removing MMC ( + MMC, indicated in blue). Duplicate samples were used. Non-specific IgG (IgG) was used as a negative control for IP. Subsequently, following a 14 h incubation period, the VRC amount in the donor cells was evaluated. (d) RNA-seq data (MMC 0 h vs 4 h) is plotted on a volcano map. X-axis and Y-axis indicate p-value (−Log₁₀) and fold change (FC) (Log₂). Annotated red dots indicate known ATF4 targets. n = 3 independent RNA samples per condition. Data are analyzed by one-way ANOVA with Tukey’s post-hoc test (c).

Extended Data Fig. 3 |. The analysis of apoptotic cells by Annexin V staining and PH phenotypes in Schistosomia-induced PH (Sch-PH).

(a) Schematic description of the experimental procedure. Lung samples of WT rats treated with vehicle or MMC were harvested on days 8, 16, and 24. IF staining of the apoptosis marker Annexin V (green) and CD31(red) for vascular endothelial cells. Hoechst (blue) is showing nuclei. Annexin V and CD31 stain images were merged and displayed in the “Merge” row. The intensity of the Annexin V signal in the CD31-positive endothelial layer was quantitated by ImageJ and shown as mean ± SEM. n = 4–7 independent samples per condition. White asterisks indicate pulmonary arteries. Scale bar= 10 μm. (b) Right ventricular (RV) systolic pressure (RVSP) and RV to left ventricular (LV)+septum weight ratio (RV/LV + S) were measured in control, Sch-PH, and CH-PH mice and shown as mean ± SEM. n = 5 independent samples per group. (c) The RVSP and RV/LV + S ratio in a vehicle or MCT-administered rats were measured and shown as mean ± SEM. n = 7–10 independent samples per group. Data are analyzed by two-sided unpaired student’s t-test (b, c); one-way ANOVA with Tukey’s post-hoc test (a).

Extended Data Fig. 4 |. Characterization of heterozygous BMPR2 mutant rats.

(a) The top 2 panels show BMPR2 gene sequencing results of E503fs (Δ2/+) rat and Q495fs (Δ26/+) rat, where the bases shown in red or in the red rectangle are mutations introduced. Blue letters show the protein sequence after the mutation. Asterisks indicate stop codons. (b) WT or Mut cells were treated with increasing concentrations of BMP4 for 2 h, and total cell lysates from 1 × 10⁶ cells were subjected to immunoblot with anti-phospho-Smad1/5/8 (p-Smad1/5/8), anti-total Smad1 (t-Smad1), anti-GAPDH (loading control), and anti-BMPR2(CTD) antibody. (c) WT or Mut cells (1 × 10⁶ cells) were treated with 100 pM of BMP4 for the indicated time (h), and total cell lysates were subjected to immunoblot with anti-p-Smad1/5/8, anti-t-Smad1, anti-GAPDH (loading control), and anti-BMPR2(CTD) antibody. Total RNAs isolated from the same cells were subjected to qRT-PCR of BMP-Smad1/5/8 target genes, Id3 and Smad6. Levels of Id3 and Smad6 mRNAs relative to GAPDH are shown by mean ± SEM. n = 3. Data are analyzed by one-way ANOVA with Tukey’s post-hoc test (c).

Extended Data Fig. 5 |. ISRIB does not affect the p-eIF2α amount, PH phenotypes in males and females are indistinguishable, physiological measurements of WT and Mut rats and lack of vascular remodeling and ISR activation in the heart and liver after MMC treatment.

(a) Lung lysates from vehicle-, MMC-, or MMC + ISRIB-treated WT and Mut rats were subjected to immunoblot analysis of p-eIF2α, t-eIF2α, and β-actin (loading control) in triplicate. n = 3 independent samples per condition. (b) p-eIF2α/t-eIF2α ratio is shown as mean ± SEM. n = 3 independent samples per condition. (c) RVSP and RV/LV + S ratio in female (orange) and male (blue) WT and Mut rats were measured 24 days after the administration of the vehicle of MMC and shown as mean ± SEM. M and F stand male and female. n = 3–5 independent samples per condition. The two-way analysis of variance was performed. (d) lung weight, lung weight/body weight ratio, % body weight change, liver weight/body weight ratio, heart weight/body weight ratio, and kidney weight/body weight ratio of WT or Mut rats subjected to ISRIB (24-days treatment) are shown as mean ± SEM. n = 4–44 independent samples per group. (e) Heart and liver were harvested from WT rats 24 days after injecting vehicle (saline) or MMC (3 mg/kg), followed by H&E staining. Arrows indicate vessels. Scale bar=10 μm. (f) Immunoblot analysis of ATF4, cardiac troponin T (cTnT; loading control for the heart), and β-actin (loading control for liver) using heart and liver lysates harvested from WT and Mut rats 24 days after the treatment with the vehicle of MMC. n = 3 independent samples per group. (g) Heart and liver lysates from WT rats treated with vehicle (24d) or MMC (8d or 24d) were subjected to anti-puromycin immunoblot and anti-GAPDH blot (loading control). The relative levels of puromycin-labeled proteins normalized by GAPDH were quantitated and shown as mean ± SEM. n = 4 independent samples. (h) Pulmonary vasculature of WT and Mut rats treated with vehicle or MMC with or without ISRIB was cast by Microfil on day 24 (d24), and holistic images of a lobe are shown as black and white images. Scale bar= 0.5 cm. The number of branches and junctions (per cm²) of distal pulmonary vessels was counted and shown as mean ± SEM. n = 3 independent samples per group. Data are analyzed by one-way ANOVA with Tukey’s post-hoc test (b, c, d, g, h).

Extended Data Fig. 6 |. Examining the pulmonary vascular permeability in vivo, the amount of Rad51 and VE-Cad in the plasma of WT and Mut rats after MMC treatment, ATF4 recruitment to the PKR gene upon MMC treatment is diminished by ISRIB, and administration of C16 prevents reduction of the distal pulmonary vessel density mediated by MMC.

(a) The permeability of pulmonary vasculature was assessed by injecting Evans Blue dye in WT and Mut rats administered with vehicle or MMC with or without ISRIB or C16. The lung was harvested on day 24, and the image was taken after the lung became translucent. Scale bar=0.5 cm. (b) The relative intensity of EB staining of the lung in WT and Mut rats was quantitated by ImageJ and shown as mean ± SEM. n = 4 independent samples. (c) The amount of VE-Cad and Rad51 in the plasma of WT and Mut rats after MMC treatment (24d) were measured by ELISA and plotted as mean ± SEM. n = 9 independent samples per group. (d) ChIP assay was performed in PMVECs (1 × 10⁶ cells) treated with vehicle, MMC, or MMC + ISRIB for four h with anti-ATF4 antibody, followed by PCR amplification of intron 1 of the PKR gene that contains three ATF4 consensus binding motifs. The result is shown as fold enrichment over the input sample as mean ± SEM. n = 6 independent samples. (e) The pulmonary vessels in WT rats treated with vehicle or MMC with or without C16 were cast with Microfil. After harvesting the lung, the image of the lobe was taken and presented after converting it to the black and white image. Distal pulmonary vessels in the area indicated by the orange box are magnified and shown below. Scale bar=0.5 cm. The number of branches and junctions (per cm²) of distal pulmonary vessels was counted and shown as mean ± SEM. n = 3 independent samples per group. Data are analyzed by one-way ANOVA with Tukey’s post-hoc test (b, c, d, e).

Extended Data Fig. 7 |. The pulmonary vascular phenotypes of two lines of BMPR2 mutant rats following MMC treatment are indistinguishable.

Representative H&E images of pulmonary vasculature exhibit similar phenotypes in E503fs (Δ2/+) and Q495fs (Δ26/+) rats 24 days after vehicle or MMC administration. Scale bar=10 μm.

Extended Data Fig. 8 |. PVOD phenotypes are evident as early as eight days after MMC administration, and delayed treatment with ISRIB attenuates ISR activation and vascular phenotypes.

(a) RVSP and RV/LV + S ratio in WT rats was measured eight days after the administration of vehicle or MMC and shown as mean ± SEM. n = 6 independent samples. (b) To examine the morphology of the pulmonary arteries (PAs) and veins (PVs), lungs were harvested from WT rats eight days after administering vehicle or MMC and subjected to H&E staining. An asterisk indicates the location of the vessels. Scale bar=10 mm. (c) WT rats administered with vehicle or MMC with or without 16-day (16d) or 8-day (8d) treatment with ISRIB were harvested on day 24, and the lung weight/body weight ratio was examined. n = 5–11 per condition. (d) The lung lysates of vehicle- or MMC-treated WT rats with or without 16-day or 8-day treatment with ISRIB were subjected to immunoblot of indicated proteins. β-actin is loading control. Quantitation of these blots is shown in Fig. 8e. n = 2–3 per condition. Data are analyzed by two-sided unpaired student’s t-test (a); one-way ANOVA with Tukey’s post-hoc test (c).

Extended Data Fig. 9 |. ISRIB promotes the proliferation of vascular endothelial cells in the MMC-treated rats.

Scheme of delayed ISRIB treatment and EdU staining (16d and 8d treatment). WT rats were administered with vehicle or MMC on day 0 and treated with ISRIB starting on day 8 for 8 days (8d) or 16 days (16d). The EdU staining (green) was performed on day 8 (d8) or day 24 (d24). The lung section was co-stained with the endothelial marker CD31 (red) and Hoechst (blue for nuclei staining). The EdU and CD31 stain images were merged and shown in the “Merge” row. The intensity of the EdU signal in the CD31-positive vascular endothelial layer was quantitated by ImageJ and shown as mean + SEM. n = 7–14 independent samples per condition. White asterisks indicate pulmonary arteries. Scale bar= 10 μm. Data are analyzed by one-way ANOVA with Tukey’s post-hoc test.

Extended Data Fig. 10 |. Release of VE-Cad and Rad51 into the plasma in patients with PVOD.

Human patients with PVOD, IPAH, and PAH with BMPR2 mutations (PAH-BR2^Mut) and non-PAH control individuals (n = 10 per group) were subjected to immunoblot with anti-VE-Cad, anti-Rad51, and anti-Transferrin (loading control) antibody. #1–10 indicate plasma samples from ten individuals. The relative amounts of VE-Cad and Rad51 normalized by transferrin are quantitated and shown as mean ± SEM. n = 10 independent samples per condition. Data are analyzed by one-way ANOVA with Tukey’s post-hoc test.

Fig. 1 |. E3 ubiquitin ligase FBH1 mediates RAD51 degradation upon MMC treatment.

a, PMVECs were transfected with siRNA targeting the expression of RAD51, FBH1, NEDD4–1 (NEDD4), NEDD4–2 (NEDD4L), SMURF1 or nonspecific control siRNA (siCtrl), and total cell lysates and total RNA were subjected to immunoblot analysis of RAD51 and GAPDH (loading control). RAD51 protein amount normalized to GAPDH values and the mRNA levels of RAD51 and E3 ubiquitin ligases normalized to GAPDH values are shown as mean ± s.e.m. n = 3 independent samples. b, PMVECs were transfected with empty vector or increasing amounts of the FBH1 expression plasmid, and total cell lysates were subjected to immunoblot analysis for FBH1, RAD51 and β-actin (loading control). The amount of FBH1 and RAD51 normalized to β-actin values is shown as mean ± s.e.m. n = 3 independent samples. c, Total cell lysates were prepared from vehicle- or MMC-treated PMVECs after transfection of siCtrl, siFbh1 or siRad51 and subjected to immunoblotting for RAD51, FBH1, γH2AX and GAPDH (loading control). Relative levels of the indicated proteins (normalized to GAPDH values) are shown as mean ± s.e.m. n = 3 independent samples. d, Time course changes in the amount of RAD51, FBH1 and VE-Cad following MMC treatment were examined by immunoblot. The amount of proteins normalized to β-actin values is shown as mean ± s.e.m. n = 3 independent samples. e, Cytoplasmic (Cyto) and nuclear (Nuc) fractions of PMVECs treated with vehicle (saline) or MMC for 14 h were subjected to immunoblot analysis for RAD51, FBH1, lamin A and C (loading control for the nuclear fraction) and β-actin (loading control for the cytoplasmic fraction). The relative amount of RAD51 and FBH1 in the cytoplasmic and nuclear compartments was quantitated and is shown as mean ± s.e.m. n = 4 independent samples. f, The alkaline comet assay was performed in PMVECs treated with vehicle (mock) or MMC for 14 h after the transfection of siCtrl, siFbh1 or siRad51. Representative confocal images are shown. White arrows indicate DNA with double-stranded breaks. The bar graph indicates levels of DNA damage as a percentage of nuclei with damaged DNA out of total nuclei as mean ± s.e.m. Approximately 100 nuclei per condition were examined. Scale bars, 25 μm. n = 8 independent samples. Samples for each independent experiment were prepared simultaneously, and gels were run in parallel (c,d,e). Data were analyzed by one-way ANOVA with Tukey’s or Sidak’s post hoc test (a) and by two-way ANOVA with Tukey’s post hoc test (c,e,f).

Fig. 2 |. MMC treatment increases permeability and impairs the endothelial barrier.

a, PMVECs transfected with siCtrl, siFbh1, siRad51 or siFbh1 and siRad51 were treated with vehicle (Veh; saline) or MMC for 14 h, followed by IF staining for VE-Cad and ZO-1 (green). Cell nuclei were stained with 4,6-diamidino-2-phenylindole (DAPI; blue). Scale bars, 25 μm. The intensity of the IF signal of VE-Cad (VE) and ZO-1 (ZO) relative to vehicle-treated, siCtrl-transfected cells is shown as mean ± s.e.m. n = 3 independent samples. b, PMVECs transfected with siCtrl, siFbh1 or siRad51 were treated with vehicle or MMC for 14 h and subjected to the TVP assay using molecular tracers: sodium fluorescein (SF), EB or FITC-dex. The concentration of the tracer is presented as fluorescence intensity (FCI) or the permeability coefficient (Ps); mean ± s.e.m. n = 3 independent samples. c, PMVECs transfected with siCtrl, siFbh1, siRad51 or siFbh1 and siRad51 were treated with vehicle or MMC and subjected to TEER measurements (Ω cm²) every hour. Results are shown as mean ± s.e.m. n = 3 independent samples. d, PMVECs were transfected with empty vector or the RAD51 or FBH1 expression plasmid. After treatment with vehicle or MMC, TEER was measured every hour, and relative TEER values are shown as mean ± s.e.m. n = 3 independent samples. All data were analyzed by two-way ANOVA with Tukey’s post hoc test.

Fig. 3 |. Interaction of RAD51 and VE-Cad in PMVECs.

a, Images of an in situ PLA using anti-RAD51 and anti-VE-Cad antibodies in PMVECs treated with vehicle or MMC for 14 h are shown in duplicate. Images of samples without primary antibodies (PLA probes only) are shown as a control (no Ab). PLA signal (red dots) represents the colocalization of VE-Cad and RAD51. DAPI was used to stain nuclei. PLA signal was quantitated and is shown as mean ± s.e.m. Scale bars, 25 μm. n = 6 independent samples. b, PMVECs treated with vehicle or MMC for 14 h were subjected to immunoprecipitation (IP) by nonspecific immunoglobulin (IgG) or an anti-RAD51 antibody, followed by immunoblot analysis for VE-Cad, RAD51 and β-actin (loading control) (IP). Total cell lysates without IP were also subjected to immunoblotting (input). The relative amount of the VRC is shown as mean ± s.e.m. n = 6 independent samples. c, PMVECs treated with vehicle or MMC for 14 h were subjected to IF staining for VE-Cad (red) and RAD51 (green). Cell nuclei were stained with DAPI (blue). Areas indicated by the white boxes are magnified and displayed below. Arrows indicate endothelial junctions where the colocalization of VE-Cad and RAD51 is found as yellow dots. The light blue signal is due to the nuclear localization of RAD51. n = 5 independent samples. Scale bars, 10 μm. d, A schematic representation of the VE-Cad cytoplasmic domain (Cyt; amino acids 621–789) and four deletion mutants (Prox, proximal; Mid, middle; Dist, distal; amino acids 780–784 deleted, Δ5) that are glutathione S-transferase (GST) fusion proteins. The amount of RAD51 pulled down with GST fusion proteins was analyzed by immunoblot. The ratio of RAD51 bound by GST fusion protein and the amount of GST fusion protein (RAD51/GST) is shown as a bar graph (mean ± s.e.m.) and summarized as + or −. The RAD51-interaction region and LC3 motifs are indicated with blue arrows and yellow boxes. n = 3 independent samples. aa, amino acids. e, Time course of the conversion of the LC3-I form of LC3 protein to the LC3-II form and β-actin (control) after MMC treatment was examined by immunoblot. The relative amount of LC3-II normalized to β-actin values is shown as mean ± s.e.m. n = 4 independent samples. Data were analyzed by two-sided unpaired Student’s t-test (b) and by one-way ANOVA with Tukey’s post hoc test (a,d,e).

Fig. 4 |. Secretion of the VRC from vascular endothelial cells.

a, CM was collected from PMVECs treated with vehicle or MMC for 14 h and subjected to immunoblotting with anti-VE-Cad, anti-RAD51 and anti-transferrin (TF) (loading control) antibodies, and data were quantified. Data are plotted as mean ± s.e.m. n = 3 independent samples. b, Time course changes in the amount of VE-Cad and RAD51 in the CM of PMVECs after MMC treatment (0–14 h) were measured by ELISA and are plotted as mean ± s.e.m. n = 3 independent samples. c, CM from PMVECs treated with vehicle or MMC for 14 h was subjected to IP with nonspecific IgG (control) or an anti-RAD51 antibody, followed by anti-VE-Cad (for VRC) and anti-RAD51 antibodies. Immunoblotting with an anti-TF antibody is shown as a loading control. Data are plotted as mean ± s.e.m. n = 4 independent samples. d, A scheme of MMC treatment in rats. Plasma samples of WT rats injected once with vehicle or MMC were collected 8, 16 and 24 d after injection and subjected to immunoblotting for VE-Cad and RAD51. Immunoblotting with an anti-TF antibody is shown as a loading control. Relative amounts of RAD51 and VE-Cad in the plasma of vehicle-treated or MMC-treated (days (d)8, 16 and 24) rats were quantitated and are plotted as mean ± s.e.m. Samples for each independent experiment were prepared simultaneously, and gels were run in parallel. n = 3 independent samples per group. PL, plasma. e, The plasma of vehicle or MMC-treated rats (day 24) was subjected to IP with an anti-RAD51 antibody, followed by immunoblot analysis with anti-VE-Cad (for the VRC) and anti-RAD51 antibodies to detect the interaction between these proteins. Immunoblotting with an anti-TF antibody is shown as a loading control. Data are plotted as mean ± s.e.m. n = 3 independent samples. f, CM from PMVECs treated with vehicle (CM^veh) and MMC for 14 h (CM^MMC) was supplemented into the culture medium of PMVECs pretreated with MMC for 2 h. Time course changes in permeability were measured using the TPV assay in triplicate with FITC-dex as a tracer. Results are shown as relative FCI and are presented as mean ± s.e.m. g, The amount of RAD51 and VE-Cad protein in PMVECs after incubating with CM^veh or CM^MMC for 14 h was quantitated by immunoblotting. Results are shown as mean ± s.e.m. n = 4 independent samples. Data were analyzed by two-sided unpaired Student’s t-test (a,c,e), by one-way ANOVA with Tukey’s post hoc test (b,d,f) and by two-way ANOVA with Tukey’s post hoc test (g).

Fig. 5 |. MMC activates the ISR pathway via PKR induction.

a, ATF4 target genes differentially expressed in PMVECs treated with MMC for 0, 4 or 14 h are shown in a heatmap. n = 3 independent RNA samples per condition. Min, minimum; max, maximum. b, Total RNA isolated from the lungs of WT rats injected with MMC (day 24) was subjected to quantitative-PCR (RT–qPCR) for Atf4 mRNA, ATF4 target mRNA (Atf3 and Ppp1r15a (Gadd34)), mRNA encoding Eif2ak2 (PKR) and mRNA encoding Cdh5 (VE-Cad) (control for MMC treatment) in triplicate. Results were normalized to Gapdh mRNA values and are presented as mean ± s.e.m. c, ChIP assay in PMVECs treated with vehicle or MMC for 4 h. ChIP was performed with nonspecific IgG (control) or an anti-ATF4 antibody, followed by PCR amplification using primers (blue arrows) flanking the genomic region of the Eif2ak2 gene containing ATF4-binding motifs (yellow boxes). PCR results are shown as fold enrichment over the input sample as mean ± s.e.m, and the image of the PCR product is shown as an inset. n = 3 independent experiments. bs, binding site. d, Lung lysates of WT rats 24 d after vehicle or MMC administration were subjected to immunoblotting for the indicated proteins. The amount of the indicated proteins normalized to β-actin values, the ratio of p-eIF2α/t-eIF2α and p-PACT/t-PACT are shown as mean ± s.e.m. n = 3 independent samples per condition. e, Lungs from WT rats administered vehicle or MMC were harvested on day 24 and subjected to IF staining with anti-ATF4 (green), anti-p-PKR (green), anti-α-SMA (red), anti-CD31 (red), anti-RAD51 (green) and anti-FBH1 (green) antibodies, images of the PA and the PV were taken, and merged images are shown. Cell nuclei were stained with DAPI (blue). Asterisks indicate the location of the PA or the PV. Scale bars, 10 μm. f, RAD51 and VE-Cad in the plasma of WT rats 24 d after vehicle or MMC administration were detected by immunoblot. Plasma samples were also subjected to IP with nonspecific IgG (control) or an anti-RAD51 antibody, followed by immunoblotting with anti-VE-Cad (for the VRC), anti-RAD51 and anti-TF (loading control) antibodies. The relative amount of the VRC, VE-Cad and RAD51 normalized to TF values is shown as mean ± s.e.m. n = 4 per condition. PL, plasma. g, An in vivo puromycin-incorporation assay was performed using lung lysates of WT rats (n = 4 per condition) 24 d after vehicle or MMC administration. After normalizing the total protein amount, puromycin-labeled proteins were visualized by immunoblot with an anti-puromycin antibody and are presented in quadruplicate. The amount of puromycin-labeled proteins normalized to GAPDH values (loading control) is shown as mean ± s.e.m. Ctrl, control. h, The amount of the indicated proteins in the lung was compared between control mice and Sch-PH mice and CH-PH mice. Protein levels normalized to β-actin values (loading control) are shown as mean ± s.e.m. n = 4 per condition. i, The plasma of Sch-PH and CH-PH mice and their controls was subjected to immunoblotting with anti-VE-Cad and anti-RAD51 antibodies. An immunoblot with an anti-TF antibody is shown as a loading control. The plasma samples were also immunoprecipitated with an anti-RAD51 antibody, followed by immunoblotting with an anti-VE-Cad antibody to detect the VRC. The plasma of MMC-treated rats (MMC) was included as a control. The relative amount of the VRC is shown as mean ± s.e.m. n = 4 independent samples per condition. j, The amount of the indicated proteins in the lung was compared between WT rats administered vehicle (control) or MCT. Proteins normalized to β-actin values (loading control) are shown as mean ± s.e.m. n = 3 independent samples per condition. k, The plasma of WT rats administered vehicle (control) or MCT was subjected to immunoblotting with anti-VE-Cad and anti-RAD51 antibodies. An immunoblot with an anti-TF antibody is shown as a loading control. Plasma samples were immunoprecipitated with an anti-RAD51 antibody and immunoblotted with an anti-VE-Cad antibody to detect the VRC. The plasma of MMC-treated rats (MMC) is included as a control. The relative amount of the VRC is shown as mean ± s.e.m. n = 3 independent samples per condition. l, PMVECs (5 × 10³ cells) were pretreated with the PKR inhibitor C16 or the ISR inhibitor ISRIB for 15 min, followed by MMC exposure for 1–14 h. The TPV assay using FITC-dex as a tracer was used to examine time course changes in permeability. Results are presented as relative FCI; mean ± s.e.m. n = 4 independent samples. m, PMVECs were pretreated with MMC for 2 h, followed by treatment with the PKR inhibitor C16 or the ISR inhibitor ISRIB for 0–14 h. The TPV assay using FITC-dex as a tracer was used to examine time course changes in permeability. Results are presented as relative FCI; mean ± s.e.m. n = 4 independent samples. n, PMVECs treated with vehicle or MMC alone, MMC and ISRIB or C16 for 14 h were subjected to IF staining with an anti-VE-Cad antibody (red) and DAPI (blue) for nuclei. Scale bars, 10 μm. o, PMVECs treated with vehicle or MMC alone, or MMC and ISRIB or C16 for 14 h were subjected to TEM, and representative images are shown. The areas shown as yellow and pink boxes are magnified and shown on the right. Scale bars, 500 nm (left), 200 nm (middle), 100 nm (right). White dotted lines delineate cell morphology. AJs and TJs are marked by yellow and green arrows, respectively. Pink arrows highlight gaps between two neighboring cells. Samples for each independent experiment were prepared simultaneously, and gels were run in parallel (d,h–k). Data were analyzed by two-sided unpaired Student’s t-test (d,f,g), by one-way ANOVA with Tukey’s post hoc test (c,i,k–m) or Sidak’s post hoc test (b) and by two-way ANOVA with Tukey’s post hoc test (h,j).

Fig. 6 |. Inhibition of the ISR prevents MMC-induced PVOD in WT and Mut rats.

a, A scheme of ISRIB treatment in MMC-treated rats. Immunoblot analysis of the indicated proteins in total lung lysates from WT and Mut rats treated with vehicle (−) or MMC (+), with or without ISRIB. Amounts of the indicated proteins relative to β-actin values are shown as mean ± s.e.m. Samples for each independent experiment were prepared simultaneously, and gels were run in parallel. n = 3 independent samples. b, RVSP (mmHg), RV/LV + S ratio and LVSP (mmHg) in WT and Mut rats in control conditions, exposed to MMC (MMC) and treated with MMC and ISRIB. The results are shown as mean ± s.e.m. n = 8–34 independent samples. c, H&E staining of heart from WT and Mut rats treated with vehicle, MMC or MMC and ISRIB. R and L indicate the right and left ventricles, respectively. A black bar is used to indicate RV wall thickness. RV and LV wall thickness was measured with ImageJ at three different locations per sample and is shown as mean ± s.e.m. n = 6 independent samples. d, H&E staining of pulmonary vasculature (PA and PV), in WT and Mut rats treated with vehicle or MMC with or without ISRIB, was performed on day 24. The third column is a magnified image of the black rectangular area in the second column. The medial thickness of microvessels (<50 μm in diameter) and medium-sized vessels (50–80 μm in diameter) was quantitated and converted to the value relative to the medial thickness of vehicle-treated WT rats and is shown as mean ± s.e.m. Scale bars, 10 μm. n = 10 independent samples. e, The PA and the PV from WT and Mut rats 8, 16 or 24 d after vehicle or MMC administration were stained with anti-α-SMA (red) and anti-VE-Cad (green) antibodies for SMCs and endothelial cells, respectively. The fourth row represents 24 d of treatment with MMC and ISRIB. The asterisk indicates the location of the PA or the PV. Scale bars, 50 μm. IF signals of images on day 24 were quantitated and converted to the value relative to the signal intensity of vehicle-treated WT rats and are shown as mean ± s.e.m. n = 10–11 independent samples. f, mRNA levels of ATF4 target genes in the lungs of WT and Mut rats administered vehicle, MMC or MMC and ISRIB were analyzed by RT–qPCR and are shown as mean ± s.e.m. n = 3 independent samples. g, The amount of RAD51 and VE-Cad in the plasma of rats treated with MMC with or without ISRIB was quantitated by ELISA and is shown as mean ± s.e.m. n = 3 independent samples. All data were analyzed by two-way ANOVA with Tukey’s post hoc test (a–g).

Fig. 7 |. The PKR antagonist C16 prevents PVOD phenotypes.

a, A scheme of C16 treatment in WT rats. RVSP, RV/LV + S ratio and the lung/body weight ratio in vehicle- or MMC-exposed WT rats treated with or without C16 are shown. The results are shown as mean ± s.e.m. n = 6 independent samples. b, H&E and trichrome staining of pulmonary vasculature in WT rats treated with vehicle or MMC with or without C16. The third column is a magnified image of the black rectangular area in the second column. Scale bars, 10 μm. c, Lung lysates from WT rats injected with vehicle or MMC with or without C16 were subjected to immunoblot analysis for the indicated proteins. The relative amount of proteins, normalized to β-actin values (loading control), and the p-eIF2α/t-eIF2α ratio are shown as mean ± s.e.m. Samples for each independent experiment were prepared simultaneously, and gels were run in parallel. n = 2–3 independent samples per condition. d, The amount of RAD51 and VE-Cad in the plasma of rats treated with MMC with or without C16 was quantitated by ELISA and is shown as mean ± s.e.m. n = 3 independent samples. Data were analyzed by one-way ANOVA with Tukey’s post hoc test (a) and by two-way ANOVA with Tukey’s post hoc test (c,d).

Fig. 8 |. Delayed ISRIB treatment reverses PVOD phenotypes in rats.

a, Scheme of the delayed ISRIB treatment (16-d and 8-d treatment). RVSP, the RV/LV + S ratio and LVSP in vehicle- or MMC-exposed WT rats with or without delayed ISRIB treatment (16 d and 8 d) are shown. The results are shown as mean ± s.e.m. n = 6 independent samples. b, H&E and trichrome staining of the pulmonary vasculature in WT rats treated with vehicle or MMC with or without delayed ISRIB treatment (16 d and 8 d). The third column is a magnified image of the black rectangular area in the second column. Scale bars, 10 μm. c, IF staining of pulmonary vasculature (PA and PV) in WT rats treated with vehicle or MMC with or without delayed ISRIB treatment (16 d and 8 d) with an anti-VE-Cad (red) antibody, an anti-RAD51 (green) antibody and DAPI (for nuclei). Scale bars, 10 μm. Asterisks indicate the location of the PA or the PV. d, The amount of the indicated mRNA species relative to Gapdh mRNA in the lungs of rats treated with MMC with or without delayed ISRIB treatment was analyzed by RT–qPCR and is shown as mean ± s.e.m. n, at least six independent samples. e, The amount of the indicated proteins relative to GAPDH protein in the lungs of rats treated with MMC with or without delayed ISRIB treatment was analyzed by immunoblot analysis and is shown as mean ± s.e.m. n = 3 independent samples per condition. f, The amount of RAD51 and VE-Cad in the plasma of rats treated with MMC with or without delayed ISRIB treatment was quantitated by ELISA and is shown as mean ± s.e.m. n = 3 independent samples. g, Human lung samples from control individuals and patients with PVOD and IPAH were stained with anti-ATF4 (green), anti-p-PKR (green), anti-CD31 (red), anti-α-SMA (red) and anti-RAD51 (green) or anti-FBH1 (green) antibodies, images of the PA and the PV were taken, and merged images are shown. Cell nuclei were stained with DAPI (blue). Asterisks indicate the location of the PA or the PV. Scale bars, 10 μm. h, Concentrations (ng ml⁻¹) of RAD51 and VE-Cad in the plasma of control (non-PAH) individuals, patients with PVOD, patients with IPAH and patients with PAH-BR2^Mut were measured in duplicate by ELISA and are plotted as mean ± s.e.m. n = 10 independent samples per group. i, Plasma from control individuals and patients with PVOD, IPAH and PAH-BR2^Mut was subjected to IP with an anti-RAD51 antibody, followed by immunoblot analysis for VE-Cad (for the VRC) and RAD51. Plasma was also subjected to immunoblot analysis for RAD51 and TF (loading control). The relative amount of the VRC is shown as mean ± s.e.m. Samples for each independent experiment were prepared simultaneously, and gels were run in parallel. n = 5 independent samples per group. j, A schematic diagram of MMC-induced aberrant activation of the PKR–ISR pathway, and depletion of VE-Cad and RAD51 promotes the pathogenesis of PVOD, which is facilitated by deregulation of the BMPR2 signaling pathway. C16 and ISRIB are equally effective in attenuating MMC-induced PVOD phenotypes. All data were analyzed by one-way ANOVA with Tukey’s post hoc test (a,d–f,h,i).