ROCK1/2 signaling contributes to corticosteroid-refractory acute graft-versus-host disease

Abstract

Patients with corticosteroid-refractory acute graft-versus-host disease (aGVHD) have a low one-year survival rate. Identification and validation of novel targetable kinases in patients who experience corticosteroid-refractory-aGVHD may help improve outcomes. Kinase-specific proteomics of leukocytes from patients with corticosteroid-refractory-GVHD identified rho kinase type 1 (ROCK1) as the most significantly upregulated kinase. ROCK1/2 inhibition improved survival and histological GVHD severity in mice and was synergistic with JAK1/2 inhibition, without compromising graft-versus-leukemia-effects. ROCK1/2-inhibition in macrophages or dendritic cells prior to transfer reduced GVHD severity. Mechanistically, ROCK1/2 inhibition or ROCK1 knockdown interfered with CD80, CD86, MHC-II expression and IL-6, IL-1β, iNOS and TNF production in myeloid cells. This was accompanied by impaired T cell activation by dendritic cells and inhibition of cytoskeletal rearrangements, thereby reducing macrophage and DC migration. NF-κB signaling was reduced in myeloid cells following ROCK1/2 inhibition. In conclusion, ROCK1/2 inhibition interferes with immune activation at multiple levels and reduces acute GVHD while maintaining GVL-effects, including in corticosteroid-refractory settings.

Fig. 1. Neutrophils and monocytes are increased in aGVHD lesions and blood of SR-aGVHD patients.

a Representative immunohistochemistry staining (brown) for CD3, CD14, and myeloperoxidase (MPO) in intestinal tissue biopsies derived from patients with SR-aGVHD or patients with steroid responsive aGVHD, Scale bar: 100 µm. b Quantification of CD3⁺, CD14⁺ or MPO⁺ cells in intestinal tissue biopsies derived from patients with SR-aGVHD (n = 12 for CD3⁺ and CD14⁺, n = 9 for MPO⁺) versus responsive GVHD (n = 7 for CD3⁺ and CD14⁺, n = 10 for MPO⁺). Each dot represents one patient. Percentage represents positive cells / all cells (36 high power fields analyzed per sample) are shown. Plots show mean +/− SD and P-value was calculated by two-tailed unpaired student t-test. c Expression intensity of indicated markers on stochastically selected leukocytes (defined as single/CD45⁺) from healthy donors (HD, n = 12), steroid-responsive aGVHD patients (n = 10) and SR-aGVHD patients (n = 9) visualized by UMAP. d FlowSOM clustered leukocyte subsets are overlaid on the UMAP plot for each condition separately. Leukocytes were derived from healthy donors (HD, n = 12), steroid-responsive aGVHD patients (n = 10) and SR-aGVHD patients (n = 9) visualized by UMAP. e–j Frequency and absolute counts of neutrophils, monocytes, and lymphocytes from steroid-responsive GVHD (n = 10) and SR-aGVHD patients (n = 9). Each dot represents one patient. Plots show mean +/− SD and P-value was calculated by Mann–Whitney test. k Absolute counts of leukocytes from steroid-responsive GVHD (n = 10) and SR-aGVHD patients (n = 9). Each dot represents one patient, plots show mean +/− SD and P-value was calculated by Mann–Whitney test. l The scatter plot shows MPO⁺ cells relative to all cells in percent derived from the peripheral blood of patients with steroid-responsive aGVHD (n = 7) or SR-aGVHD (n = 12). Each dot represents one patient. Plots show mean +/− SD and P-value was calculated two-tailed unpaired student t-test.

Fig. 2. ROCK1/2 is increased in patients with aGVHD and in mice after allogenic BMT.

a The volcano plot shows significant protein abundance differences of activated kinases isolated from patients without aGVHD (n = 6) and patients with SR-aGVHD (n = 6). Kinases were enriched by kinase-bead based selection as reported by others. The graph shows in red the kinases that were significantly more abundant in patients with SR-aGVHD (t-test, permutation-based FDR < 0.05). b The scatter plot shows the quantification (fold change of MFI) of ROCK1/2 expression in PBMCs of healthy controls versus SR-aGVHD patient samples. Each data point represents an individual patient and values were normalized to the mean of the control group. The plot show mean +/− SD and the P-value was calculated using a two-tailed unpaired student t-test. c Representative histograms of ROCK1/2 expression in PBMCs of healthy controls and SR-aGVHD patients as indicated. d The scatter plot shows the quantification (fold change of MFI) of ROCK1/2 expression in CD14⁺ cells of healthy controls versus SR-aGVHD patient samples. Each data point represents an individual patient and values were normalized to the mean of the control group. The plot show mean +/− SD and the P-value was calculated using a two-tailed unpaired student t-test. e Representative histograms of ROCK1/2 expression in CD14⁺ cells of healthy controls and GVHD patients as indicated. f The scatter plot shows the quantification (fold change of MFI) of ROCK1/2 expression in CD11c⁺CD11b⁺ DCs isolated from the spleen of mice on day 14 after syngeneic HCT or allogeneic HCT as indicated. The experiment was performed three times, each data point represents one mouse and values were normalized to the mean of the control group. The plot show mean +/− SD and the P-value was calculated using a two-tailed unpaired student t-test. g Representative histograms of ROCK1/2 expression in CD11c⁺CD11b⁺ DCs. h The scatter plot shows the quantification (fold change of MFI) of ROCK1/2 expression in CD11b⁺ F4/80⁺ macrophages isolated from the spleen of mice on day 14 after syngeneic HCT or allogeneic HCT as indicated. The experiment was performed two times, each data point represents one mouse and values were normalized to the mean of the control group. The plot show mean +/− SD and the P-value was calculated using a two-tailed unpaired student t-test. i Representative histograms of ROCK1/2 expression in CD11b⁺ F4/80⁺ macrophages.

Fig. 3. ROCK1/2-inhibition reduces aGVHD in mice and causes downregulation of genes related to IL6 JAK STAT3 signaling in CD11b⁺ cells.

a–k Mice underwent allo-HCT to induce aGVHD. Mice in the ROCK1/2-inhibitor group were treated intraperitoneal with 8 mg/kg ROCK1/2-inhibitor (dissolved in 100 µl PBS) from day 4 to day 13, mice in the vehicle group were treated intraperitoneal with an equal volume PBS. a Survival of mice was plotted by using the Kaplan–Meier method and compared by using a log-rank (Mantel Cox) test. Depicted are 10 mice in the vehicle group and 11 mice in the ROCK1/2-inhibitor group. The experiment was performed twice. b–d Histological aGVHD scoring of liver (b), colon (c) and small intestine (d) of mice on day 14 after allo-HCT that received either ROCK1/2-inhibitor or vehicle treatment. Depicted are 10 mice in each group. The plots show mean +/− SD and the P-value was calculated using a two-tailed unpaired student t-test. e The scatter plot shows the percentage of CD11b⁺ cells of all CD45⁺ cells in the lamina propria of small intestine on day 14 after allo-HCT, which equals 10 days of ROCK1/2-inhibitor or vehicle treatment. Depicted are 10 mice each group. The experiment was performed twice. The plot show mean +/− SD and the P-value was calculated using a two-tailed unpaired student t-test. f Principal Component (PC) Analysis based on gene expression, determined by mRNA-microarray of CD45⁺CD11b⁺ cells that were FACS-isolated from the spleen of mice on day 14 after allo-HCT, which equals 10 days of ROCK1/2-Inhibitor or vehicle treatment. Depicted are four mice in the ROCK1/2-inhibitor group (red) and three mice in the vehicle group (blue). g The heatmap shows downregulated genes involved in IL6_JAK_STAT3_signaling. Color scale represents the row-wise z-score intensity. * indicates significant changes in gene expression. h The scatter plot shows the MFI of Il-6 expression in CD11c⁺CD11b⁺ DCs isolated from the spleen on day 14 after allo-HCT. The experiment was performed three times, each data point represents one mouse and values were normalized to the mean of the control group. The plot show mean +/− SD and the P-value was calculated using a two-tailed unpaired student t-test. i Representative histograms of Il-6 expression in CD11c⁺CD11b⁺ DCs isolated from the spleen on day 14 after allo-HCT. j The scatter plot shows the MFI of Tnf expression in CD11c⁺CD11b⁺ DCs isolated from the spleen on day 14 after allo-HCT. The experiment was performed three times and each data point represents one mouse. The plot show mean +/− SD and the P-value was calculated using an two-tailed unpaired student t-test. k Representative histograms of Tnf expression in CD11c⁺CD11b⁺ DCs isolated from the spleen on day 14 after allo-HCT.

Fig. 4. ROCK1/2-inhibition reduces stimulatory activity, co-stimulatory molecule expression, Il-6, iNOS, and migratory activity of DCs.

a–d Coculture of T cells and allogeneic BM-derived DCs was performed for 72 hours. DCs were activated with LPS for 24 h and when indicated thereafter treated with 30 µg/ml ROCK1/2-inhibitor for two hours prior to the coculture. a The scatter plot shows the proliferation of CD8⁺ T cells in response to coculture with allogeneic DCs. The experiment was performed three times, each point represents one mouse. The plot shows mean +/− SD and the P-value was calculated using a two-way ANOVA (Dunnett’s multiple comparison). b Representative histogram showing the proliferation of CD8⁺ T cells in response to coculture with allogeneic DCs. c, d The scatter plots show the fold change in the mean fluorescence intensity (MFI) of the surface marker CD86 (c) and CD80 (d) on CD11c⁺ cells incubated in a coculture with T cells. The experiment was performed three times, each point represents one mouse. The plots show mean +/− SD and the P-value was calculated using a two-tailed paired student t-test. e–g The scatter plots show the fold change of iNOS mRNA (e), Il-6 mRNA (f) and Il-10 (g) mRNA isolated from CD11c⁺ cells. BM-derived DCs were treated with LPS for two hours and exposed to ROCK1/2-Inhibitor or vehicle for two hours. The experiment was performed three times, each point represents a mouse, values of the experimental groups were normalized to vehicle control. The plots show mean +/− SD and the P-value was calculated using a two-tailed paired student t-test. h The scatter plot shows the migration of CD11c⁺ cells. BM-derived DCs were treated with LPS for two hours and exposed to ROCK1/2-inhibitor for two hours. Thereafter, they were incubated in a transwell chamber for five hours with a CXCL12 gradient and migration was assessed using FACS. The experiment was performed three times, each point represents one mouse. The plot shows mean +/− SD and the P-value was calculated using a two-tailed paired student t-test. i The scatter plot shows the fold change of MFI for phalloidin in CD11c⁺ cells. BM-derived DCs were treated with LPS for two hours and exposed to ROCK1/2-inhibitor for two hours. The experiment was performed three times, each point represents one mouse, values of the experimental group were normalized to vehicle control. The plot shows mean +/− SD and the P-value was calculated using a two-tailed paired student t-test. j The representative images show staining for phalloidin (green) in CD11c⁺ cells isolated from mice that had undergone allo-HCT. Mice were treated intraperitoneal with vehicle or ROCK1/2- inhibitor (8 mg/kg in 100 µl PBS) from day 4 to day 13. Images (z-stacks) were acquired by structured illumination microscopy (SIM). Superresolution images were processed with Zen Black software and visualized in 3D with Imaris software (maximum intensity projection). All scale bars are 2 µm in overviews and 1 µm in magnifications. k The scatter plot shows the Volume [µm³] of F-actin in CD11c⁺ cells isolated from mice that had undergone allo-HCT. Mice were treated intraperitoneal with vehicle or 8 mg/Kg ROCK1/2-inhibitor from day 4 to day 13. Cells were isolated from mice of two independent experiments and analyzed. Individual symbols represent measurements of cells isolated from the same animal. The plot shows mean +/− SD and P-value was calculated using an two-tailed unpaired student t-test. l Representative histogram showing phalloidin staining of F-actin in CD11c⁺ cells isolated from mice that had undergone allo-HCT as described in (k). m The scatter plot shows the MFI for phalloidin staining of F-actin in CD11c⁺ cells isolated from mice that had undergone allo-HCT. Mice were treated intraperitoneal with vehicle or ROCK1/2-inhibitor (8 mg/kg in 100 µl PBS) from day 4 to day 13. The experiment was performed twice, each point represents one mouse. The plot shows mean +/− SD and the P-value was calculated using an two-tailed unpaired student t-test. n Survival of mice undergoing allo-HCT to induce aGVHD. On the same day, BM-derived DCs were incubated with ROCK1/2-inhibitor or PBS for four hours, thereafter 1 × 10⁶ donor DCs were injected together with the graft. Depicted are 10 mice in each group from two individual experiments. Survival of mice was plotted by using the Kaplan-Meier method and compared by using a log-rank (Mantel Cox) test. o–q Histological aGVHD scoring of liver (o), colon (p) and small intestine (q) of mice on day 14 after allo-HCT as described in (n). The plots show mean +/− SD and the P-value was calculated using a two-way ANOVA (Dunnett’s multiple comparison).

Fig. 5. ROCK1/2-inhibition reduces MHC-II expression, Il-6, iNOS, and migratory activity of macrophages.

a The scatter plot shows the fold change of MFI for MHC-II on F4/80⁺ CD11b⁺ cells on day 14 after allo-HCT. Mice underwent allo-HCT to induce aGVHD and were treated intraperitoneal with vehicle or ROCK1/2-inhibitor (8 mg/kg in 100 µl PBS) from day 4 to day 13. The experiment was performed three times, each point represents one mouse. The plot shows mean +/− SD and the P-value was calculated using a two-tailed unpaired student t-test. b Representative histogram showing MHC-II expression on F4/80⁺ CD11b⁺ cells. c–g BM-derived macrophages were stimulated with LPS for two hours and thereafter with ROCK1/2-inhibitor or vehicle for two hours. The scatter plots show the fold change of Il-6 mRNA (c), iNOS mRNA (d), Il-1β mRNA (e), Tnf mRNA (f) and Il-10 mRNA (g) isolated from BM-derived macrophages. Values of the experimental groups were normalized to control group. The experiment was performed three times, each point represents a mouse. The plot shows mean +/− SD and the P-value was calculated using a two-tailed paired student t-test. h The scatter plot shows the total count of macrophages migrated through a membrane towards MCP-1. BM-derived macrophages were pre-treated with LPS for two hours followed by ROCK1/2-inhibitor treatment or vehicle for two hours. Thereafter, they were incubated in a transwell-chamber for five hours with a MCP-1 gradient and migration was assessed by microscopy. The experiment was performed three times, each point represents one mouse. The plot shows mean +/− SD and the P-value was calculated using a two-tailed paired student t-test. i The scatter plot shows the fold change of MFI for phalloidin in BM-derived macrophages. Cells were treated with LPS for two hours and exposed to ROCK1/2-inhibitor or vehicle for two hours. The experiment was performed three times, each point represents a mouse. The plot shows mean +/− SD and the P-value was calculated using a two-tailed paired student t-test. j Quantification of ROCK1 in the LPS stimulated macrophage cell line RAW264.7 with ROCK1 knockdown or without (shRNA directed against Renilla luciferase) compared to BM-derived macrophages exposed to LPS and ROCK1/2-inhibitor or vehicle for two hours. The experiment was performed three times, each point represents a different passage of cells, values of the experimental groups were normalized to control group. The plot shows mean +/− SD and the P-value was calculated using a two-way ANOVA (Dunnett’s multiple comparison). k Representative western blot showing ROCK1 expression after ROCK1 knockdown or ROCK1/2-inhibitor treatment in RAW264.7 cells. The experimental design is the same as panel j. Uncropped western blots are shown in Source Data. l–n The scatter plots show the fold change of Il-6 mRNA (l), Il-1β mRNA (m) and iNOS mRNA (n) isolated from RAW264.7 cells with a ROCK1 knockdown or without (shRNA directed against Renilla luciferase) compared to BM-derived macrophages exposed to LPS and ROCK1/2-inhibitor or vehicle for two hours. The experiment was performed three times, each point represents a different passage of cells, and values of the experimental group were normalized to control group. The plot shows mean +/− SD and the P-value was calculated using a two-way ANOVA (Dunnett’s multiple comparison). o Survival of mice undergoing allo-HCT to induce aGVHD. On the same day, BM-derived macrophages were incubated with ROCK1/2-inhibitor or PBS for four hours, thereafter 7 × 10⁵ donor macrophages were injected together with the graft. Survival of mice was plotted by using the Kaplan–Meier method and compared by using a log-rank (Mantel Cox) test. Depicted are 10 mice in each group from two individual experiments. p–r Histological aGVHD scoring of liver (p), colon (q) and small intestine (r) of mice on day 14 after allo-HCT as described in (o). The plots show mean +/− SD and the P-value was calculated using a two-way ANOVA (Dunnett’s multiple comparison).

Fig. 6. ROCK1/2-inhibition alters the gene expression profile in vivo and reduces NF-κB and cofilin phosporylation in DCs and in macrophages in vitro.

a The heatmap shows the expression of downregulated genes related to PI3K signaling in CD45⁺CD11b⁺ cells that were FACS-isolated from mice on day 14 after allo-HCT after 10 days of ROCK1/2-inhibitor or vehicle treatment. Color scale represents the row-wise z-score intensity. * Indicates significant changes in gene expression. b The heatmap shows the expression of genes involved in Tnf signaling of CD45⁺CD11b⁺ cells isolated as indicated in (a). Color scale represents the row-wise z-score intensity. * Indicates significant changes in gene expression. Red arrow indicates NF-κB1 gene. c–f The western blots show total and phosphorylated NF-κB of BM-derived DCs or macrophages incubated with LPS for two hours and thereafter with ROCK1/2-inhibitor for two hours. Samples derived from the same experiment and gels/blots were processed in parallel. Loading controls ran on the same blot. When applicable, data is normalized to control group, the experiments were performed three times, each point represents a mouse, values of the experimental group were normalized to control group. The plots show mean +/− SD and the P-value was calculated using a two-tailed paired student t-test. Uncropped western blots are shown in Source Data. c The bar graph shows pNF-κB/total NF-κB in BM-derived DCs. d Representative western blot image of pNF-κB and total NF-κB in BM derived DCs. e The bar graph shows pNF-κB/total NF-κB in BM-derived macrophages. f Representative western blot image of pNF-κB and total NF-κB in BM-derived macrophages. g–j The western blots show total and phosphorylated Cofilin of BM-derived DCs or macrophages incubated with LPS for two hours and thereafter with ROCK1/2-inhibitor for two hours. Samples derived from the same experiment and gels/blots were processed in parallel. Loading controls ran on the same blot. When applicable, data are normalized to control group, the experiments were performed three times, each point represents a mouse, values of the experimental group were normalized to control group. The plots show mean +/− SD and the P-value was calculated using a two-tailed paired student t-test. Uncropped western blots are shown in Source Data. g The bar graph shows pCofilin/total Cofilin in BM-derived DCs. h Representative western blot image of pCofilin and total Cofilin in BM-derived DCs. i The bar graph shows pCofilin/total Cofilin in BM-derived macrophages. j Representative western blot image of pCofilin and total Cofilin in BM-derived macrophages.

Fig. 7. ROCK1/2-inhibition allows for cytotoxicity and GVL effects in mice and has additive effects with JAK1/2-inhibition.

a The diagram shows the percentage of dead A20 cells that had been exposed to allogeneic T cells at different ratios (effector to target 1:1, 2:1, 4:1). Prior to the coculture, T cells were preincubated with ROCK1/2-inhibitor for two hours. T cells were activated with CD3/CD28 Dynabeads for 48 h. Cytotoxicty was measured by FACS after 20 h of coculture. Represented data are pooled from three independent experiments performed in triplicate. The plots show mean +/− SD and the P-value was calculated using an ordinary two-way ANOVA test (Sidaks multiple comparison). b Representative histograms showing the percentage of dead A20 cells of all A20 cells after exposure to allogeneic T cells as described in panel (a). c The scatter plot shows the fold change of granzyme B in CD3⁺ cells. T cells were activated with CD3/CD28 Dynabeads for 48 h and then exposed to ROCK1/2-inhibitor for two hours. The experiment was performed three times, each point represents a mouse, values of the experimental group were normalized to control group. The plots show mean +/− SD and the P-value was calculated using a two-tailed paired student t-test. d The scatter plot shows the fold change of perforin in CD3⁺ cells. The experimental design was the same as panel (c). e Survival of mice after allo-HCT that received WEHI-3B leukemia cells and BM cells alone or BM/T cells. ROCK1/2-inhibitor was injected from day 4 to day 13. Survival was plotted using the Kaplan–Meier method and compared by using a log-rank (Mantel Cox) test. The experiment was performed three times, each point represents a mouse (15 mice in each group). f Survival of mice after allo-HCT that received MLL^PTD/wt FLT3 ^ITD/wt AML cells and BM cells alone or BM/T cells. ROCK1/2-inhibitor was injected from day 4 to day 13. Survival was plotted using the Kaplan–Meier method and compared by using a log-rank (Mantel Cox) test. The experiment was performed three times, each point represents a mouse (15 mice in each group). g–i Histological aGVHD scoring of liver (g), colon (h) and small intestine (i) of mice on day 14 after allo-HCT that received either ROCK1/2-inhibitor, ruxolitinib, a combination of ROCK1/2-inhibitor and ruxolitinib or vehicle treatment as described before. The plots show mean +/_ SD and the P-value was calculated using a two-way ANOVA (Dunnett’s multiple comparison).

Fig. 8. Proposed mechanism of action of ROCK1 inhibition.

Acute GVHD pathophysiology (left panel) involves expansion of MPO positive neutrophils and upregulation of ROCK1 expression in myeloid cells. ROCK1/2-inhibition in mice leads to reduced IL-6 and TNF levels, thereby reducing aGVHD histopathology and improving survival. ROCK1/2-inhibition counteracts multiple pathogenic events in dendritic cells (blue panel) and macrophages (green panel) including TNF production by DCs and macrophages, antigen presentation and migration. Moreover, ROCK1/2-inhibition in macrophages or dendritic cells prior to transfer reduces aGVHD severity. On the signaling level ROCK1/2-inhibition interferes with the NF-κB pathway activity.