Pericyte phenotype switching alleviates immunosuppression and sensitizes vascularized tumors to immunotherapy in preclinical models

Abstract

T cell-based immunotherapies are a promising therapeutic approach for multiple malignancies, but their efficacy is limited by tumor hypoxia arising from dysfunctional blood vessels. Here, we report that cell-intrinsic properties of a single vascular component, namely the pericyte, contribute to the control of tumor oxygenation, macrophage polarization, vessel inflammation, and T cell infiltration. Switching pericyte phenotype from a synthetic to a differentiated state reverses immune suppression and sensitizes tumors to adoptive T cell therapy, leading to regression of melanoma in mice. In melanoma patients, improved survival is correlated with enhanced pericyte maturity. Importantly, pericyte plasticity is regulated by signaling pathways converging on Rho kinase activity, with pericyte maturity being inducible by selective low-dose therapeutics that suppress pericyte MEK, AKT, or notch signaling. We also show that low-dose targeted anticancer therapy can durably change the tumor microenvironment without inducing adaptive resistance, creating a highly translatable pathway for redosing anticancer targeted therapies in combination with immunotherapy to improve outcome.

Keywords: Cancer immunotherapy; Mouse models; Oncology; Pericytes; Therapeutics.

Figure 1. RGS5 expression levels regulate pericyte phenotype in vitro.

(A) Relative RGS5 mRNA expression in 10T1/2, RGS5myc overexpressing and Rgs5 knockdown cell lines (RGS5shRNA1/3). n = 3 replica. Data are represented as mean ± SEM. *P < 0.01; **P < 0.001; ***P < 0.0001, 1-way ANOVA. (B) Microscopic images depicting 10T1/2, RGS5 overexpressing, and Rgs5 knockdown cells. Scale bar: 50 μm. (C) Cell proliferation (48 hours) in parental and transfectant 10T1/2 cells. n = 5 replica. Data are represented as mean ± SEM. ***P < 0.0001, 1-way ANOVA. (D) FACS blots showing PI histograms with color-coded cell cycle phases. Green, G0/G1 phase; yellow, S phase; blue, G2/M phase. Quantitative analysis of cell-cycle progression. n = 3 experiments. Data are represented as mean ± SEM. *P = 0.02; **P = 0.001; ***P = 0.0001, 1-way ANOVA. (E) Western blot (WB) of phosphorylated/total FOXO3a and phosphorylated/total p27KIP proteins. Red box highlights results in RGS5 overexpression cells. Duplicates are shown for each marker. Relative phosphorylated and total FOXO3a expression were normalized to tubulin; relative phosphorylated p27KIP was normalized to total p27KIP expression and quantified. n = 3 experiments (2 replica each). Data are represented as mean ± SEM. tFOXOa: ****P < 0.0001; pFOXO3a: *P = 0.028; p-p27: *P = 0.020, **P = 0.006, 1-way ANOVA. (F) WB of contractile (CNN1, ACTG2) and synthetic (KLF4, CNX43) markers in correlation to Rho kinase activity (p-MLC). Red box highlights results in RGS5 overexpression cells. Duplicates are shown for each marker, and relative protein expression normalized to tubulin was quantified. n = 3 experiments (2 replica each). Data are represented as mean ± SEM. CNN1: ****P < 0.0001, **P = 0.011; ACTG2: ***P = 0.0001, **P = 0.012, *P = 0.047; KLF4: **P = 0.003; CNX43: ***P = 0.003; p-MLC: ****P < 0.0001, *P = 0.017, 1-way ANOVA. (G) WB of ROCK1 and ROCK2 proteins. Duplicates are shown for each marker. Relative protein expression normalized to tubulin was quantified. n = 3 experiments (2 replica each). Data are represented as mean ± SEM. **P = 0.005, Student’s t test. (H) Contractile and synthetic pericyte markers in relation to RGS5 high (Rgs5^hi) or low (Rgs5^low) expression.

Figure 2. Intratumoral RGS5 expression determines pericyte phenotype in vivo and blood vessel functionality.

(A) PNET from WT RIP1-Tag5 (WT), RIP1-Tag5 on a Rgs5-knockout background (Rgs5^KO or KO), and triple-transgenic RIP1-Tag5 × UbiCRGS5 × RGS5CreERT2 mice engineered to overexpress RGS5 (Rgs5^hi or HI) were analyzed at 27 weeks. Images depict vascular CD31 expression (red) and infused FITC-lectin (green) as surrogate markers for tumor perfusion; arrows indicate overlay (yellow). Quantification of overlay and CD31 vessel area. n = 3–4 mice. *P = 0.02; ****P < 0.0001, 1-way ANOVA. (B) Extravasation of 70 kD dextran (red, arrowheads) from blood vessels into tumor parenchyma as marker for vessel leakiness. n = 3 mice. Data are represented as mean ± SEM. *P = 0.04; **P = 0.0014, 1-way ANOVA. (C) Calponin (CNN1, red) expression in pericytes (NG2, green); arrows indicate overlay (yellow); quantification of overlay and frequency of NG2⁺ pericytes. n = 5 mice, *P = 0.0064; **P = 0.0008; ***P < 0.0001, 1-way ANOVA. (D) COLI (red) deposition around pericytes (ACTA2, green); brackets indicate width of COLI deposits. n = 3–5 mice. Data are represented as mean ± SEM. *P = 0.022; ***P = 0.0006; ****P < 0.0001, 1-way ANOVA. (E) VE-cadherin (CDH5, red, arrowheads) coverage of CD31 (blue) vessels, n = 4–5 mice. Data are represented as mean ± SEM. **P = 0.0023; ***P < 0.0001, 1-way ANOVA. (F) p-MLC (red) expression in pericytes (NG2, green); arrows indicate overlay (yellow), n = 3–4 mice. Data are represented as mean ± SEM. *P = 0.03; **P = 0.0003; ***P < 0.0001, 1-way ANOVA. (G) Fasudil treatment schematic of Rgs5^KO PNET mice and assessment of tumor perfusion at endpoint. CD31 (red) overlay with infused FITC-lectin (yellow) is highlighted by arrows. Perfusion was quantified in Rgs5^KO+fasudil (F) group in comparison with WT and Rgs5^KO (data from A, shadowed). n = 4–5 mice. Data are represented as mean ± SEM. ****P < 0.0001, 1-way ANOVA. Scale bars: 50 μm (A–D, F, G); 25 μm (E).

Figure 3. “Forced” pericyte maturation changes the tumor microenvironment and enhances immunotherapy.

(A) B16-OVA tumors in WT or Rgs5^KO (KO) mice. Overlay (yellow, arrows) of FITC-lectin (green) with CD31⁺ vessels. n = 5 mice. Data are represented as mean ± SEM. ***P < 0.0001, Student’s t test. (B) Hypoxy probe accumulation (red circles). n = 5–6 mice. Data are represented as mean ± SEM. *P = 0.01, Student’s t test. (C) M1/M2 macrophage ratio. n = 4–6 mice. Data are represented as mean ± SEM. **P = 0.01, Student’s t test. (D) Endogenous T cells. n = 7–8 mice. Data are represented as mean ± SEM. **P = 0.008; ***P = 0.0004, Student’s t test. (E) OT-I CD8⁺ T cells (CD45.1⁺TCRv2α⁺). n = 7–8 mice. Data are represented as mean ± SEM. *P = 0.024, Student’s t test. (F) Macrophage M1/ M2 ratio after OT-I transfer. n = 7–8 mice. Data are represented as mean ± SEM. *P = 0.014, Student’s t test. (G) ICAM (green) expression on vessels (red). n = 5–8 mice. Data are represented as mean ± SEM. *P = 0.013; **P = 0.0013; ****P < 0.0001, 1-way ANOVA. (H) B16-OVA mice, untreated or treated with adoptive transfer (arrow). Tumor growth, n = 4–6 mice. Data are represented as mean ± SEM. *P = 0.019 on day 16, WT+OT-I versus Rgs5^KO+OT-I, multiple unpaired t tests. **P = 0.012, Rgs5^KO versus Rgs5^KO+OT-I; ***P = 0.0067, WT+OT-I versus Rgs5^KO+OT-I, log-rank (Mantel-Cox) test. (I) Rgs5^KO mice, untreated or treated with adoptive OT-I cell transfers (arrows) or transfers combined with αCSF1R. n = 3–4 mice. Data are represented as mean ± SEM. *P = 0.013 Rgs5^KO+OT-I versus Rgs5^KO+OT-I+anti-CSF1R on day 15, multiple unpaired t tests. Scale bars: 50 μm (A); 100 μm (B); 25 μm (G).

Figure 4. Low-dose therapeutics mimic Rgs5 knockdown by inducing pericyte maturation.

(A) RGS5 signaling and Rho kinase–activating effects of inhibitors (blue bars). (B) Relative RGS5 expression in 10T1/2 cells, 40 μM DAPT. n = 3 biological replica. Data are represented as mean ± SEM. ***P = 0.0001, Student’s t test. (C) Relative RGS5 expression in RIP1-Tag5 tumors treated with DAPT. n = 7–8 mice. Data are represented as mean ± SEM. *P = 0.025, Student’s t test. (D) Contractile markers (CNN1, ACTG2) and p-MLC in RGS5myc cells with increasing doses of trametinib. Quantification of 3 independent experiments. Data are represented as mean ± SEM. *P ≤ 0.04, **P ≤ 0.006, 1-way ANOVA (Kruskal-Wallis test). (E) Representative WB of CNN1, ACTG2, and p-MLC in RGS5myc cells with increasing doses of BEZ235 (left) and DAPT (right). The experiment was conducted twice. (F) RIP1-Tag5 mice untreated (U) or treated with trametinib (T), BEZ235 (B), or DAPT (D). FITC-lectin overlay (yellow) with CD31⁺ (red) vessels was quantified. n = 4–12. Data are represented as mean ± SEM. *P = 0.0173; **P = 0.0001; ***P < 0.0001, 1-way ANOVA. (G) Quantification of CNN1 expression (red) in relation to NG2⁺ pericytes (green). Arrows indicate overlay (yellow), n = 6–8. Data are represented as mean ± SEM. *P = 0.022, ***P < 0.0001, 1-way ANOVA. (H) COLI deposition (red) around NG2⁺ pericytes (green). Arrows indicate overlay of markers (yellow). n = 6–12. Data are represented as mean ± SEM. *P = 0.006; **P = 0.0006; ***P = 0.0005, 1-way ANOVA. (I) p-MLC expression in NG2⁺ pericytes (green). Arrows indicate overlay (yellow). n = 4–7. Data are represented as mean ± SEM. *P = 0.016; **P = 0.014; ***P = 0.0053, 1-way ANOVA. Scale bars: 100 μm (F); 50 μm (G–I).

Figure 5. Low-dose therapeutics improve effectiveness of anticancer immunotherapy.

(A) B16-OVA tumors untreated or treated with trametinib, BEZ235, or DAPT. Quantification of FITC-lectin (green) overlay (yellow, arrows) with CD31⁺ (red) blood vessels. n = 4–7 mice. Data are represented as mean ± SEM. ****P < 0.0001, 1-way ANOVA. (B) Quantification of hypoxy probe (red, circles). n = 4–9 mice. Data are represented as mean ± SEM. ****P < 0.0001, 1-way ANOVA. (C) Quantification of vascular (CD31⁺, red) ICAM (green) expression (yellow, arrows). n = 3–5 mice. Data are represented as mean ± SEM. *P = 0.042; **P = 0.034; ***P = 0.0033, 1-way ANOVA. (D) Tumors untreated or treated with trametinib, BEZ235, or DAPT. M1/M2 macrophage ratio. n = 3–4 tumors. Data are represented as mean ± SEM. *P = 0.034; **P = 0.008; *P = 0.0272 (DAPT), Student’s t test. (E) Quantification of OT-I T cells (CD45.1⁺TCRv2α⁺), following adoptive transfer, groups as in D. n = 5–7 mice. Data are represented as mean ± SEM. ***P = 0.004; ****P ≤ 0.0001, Student’s t test. (F) B16-OVA mice untreated or treated with drugs before OT-I cell transfers (arrows). Trametinib: n = 5–7, mean ± SEM. Tumor growth on days 17 and 21, *P = 0.012; ****P < 0.0001, multiple unpaired t tests. Survival: **P = 0.0039, log-rank (Mantel-Cox) test. BEZ235: n = 5 mice, mean ± SEM. Tumor growth on days 13 and 14. *P = 0.014; ***P < 0.0001, multiple unpaired t tests. Survival: **P = 0.0039 OT-I versus BEZ235+OT-I, log-rank (Mantel-Cox) test. DAPT: n = 5 mice, mean ± SEM. Tumor growth on day 16. *P = 0.04, multiple unpaired t tests. Survival: *P = 0.0276, log-rank (Mantel-Cox) test. Scale bars: 100 μm (A, B); 25 μm (C).

Figure 6. Tumor-vessel normalization following long-term treatment with low-dose therapeutics is highly sustainable.

(A) Eight-week treatment schematic in RIP1-Tag5 mice including 2-week priming and 6-week maintenance phase. (B) Representative images of untreated mice or mice treated with trametinib (0.02 mg/kg), BEZ235 (5 mg/kg), DAPT (5 mg/kg), or anti-VEGFR2 antibodies (V, DC101, 15 mg/kg). FITC-lectin (green) overlay (yellow, arrows) with CD31⁺ (red) blood vessels was quantified as surrogate marker for tumor perfusion. n = 3–8 mice. Data are represented as mean ± SEM. ****P < 0.0001 for all treatment groups compared with untreated. Quantification of CD31⁺ intratumoral blood vessels, *P = 0.031, NS, not statistically significant for trametinib, BEZ235, and DAPT treatments compared with untreated, 1-way ANOVA. Scale bar: 100 μm. (C) Representative images and quantification of overlay (yellow, arrows) of CNN1 (red) expressing NG2⁺ pericytes (green). n = 3–6 mice. Data are represented as mean ± SEM. **P = 0.009; ***P = 0.007; ****P < 0.0001. Quantification of NG2⁺ intratumoral pericytes: ****P < 0.0001, NS, not statistically significant for trametinib, BEZ235, and DAPT treatments compared with untreated, 1-way ANOVA. Scale bar: 50 μm. (D) Representative H&E images and quantification of percentage of RIP1-Tag5 tumors displaying an intact collagen capsule (dotted line). n = 3–6 mice. Data are represented as mean ± SEM. ****P < 0.0001, NS, not statistically significant for trametinib, BEZ235, and DAPT treatments compared with untreated, 1-way ANOVA. Scale bars: 200 μm (upper images and αVEGFR2); 50 μm (αVEGFR2 detail).

Figure 7. Pericyte phenotype switching is active in melanoma PDX and facilitates TIL tumor uptake.

(A) NSG mice were implanted with melanoma PDX and treated with 10 doses of trametinib (0.02 mg/kg, o.g.) or left untreated. Representative images of blood vessels (CD31⁺) and pericytes (ACTA2⁺) in untreated and trametinib-treated PDX melanoma tumors. Quantification of total CD31⁺ vessels (red), total ACTA2⁺ (green) pericytes, and ACTA2⁺ covered CD31⁺ blood vessels (yellow). n = 7 mice. Data are represented as mean ± SEM. *P = 0.0224, Student’s t test. Scale bar: 50 μm. (B) Representative images and quantification of total CNN1 (green) expression and CNN1⁺ (green) covered CD31⁺ blood vessels (red). n = 7 mice. Data are represented as mean ± SEM. **P = 0.0026; ****P < 0.0001, Student’s t test. Scale bar: 50 μm. (C) Quantification of tumor hypoxia in treatment groups (red hypoxy probe deposits). n = 5 mice. Data are represented as mean ± SEM. ***P = 0.0009, Student’s t test. Scale bar: 500 μm. (D) Treatment schematic of melanoma PDX tumor–bearing NSG mice with autologous TILs and time line for analysis. (E) Representative images depicting human CD3⁺ TIL infiltration (green) at weeks 1 or 3 into melanoma PDX left untreated or pretreated with 5 doses of trametinib before TIL transfer. Quantification of infiltrating TILs 1 week and 3 weeks after adoptive transfer into PDX mice. n = 3 mice. Data are represented as mean ± SEM. *P = 0.01, 1-way ANOVA. Scale bar: 50 μm.

Figure 8. Pericyte phenotype switching is inducible in human cancer and correlates with melanoma patient survival.

(A) Schematic of ex vivo organ slice culture; 1 to 2 mm² diameter/300 μm thick human meningioma sections in agarose were cultured on sponge material in media in 24-well plates, and vascular markers were analyzed by immunohistochemistry after 3 and 5 days in culture with or without trametinib. (B) Microscopic images of meningioma tumor slices cultured for 3 or 5 days with or without trametinib. CNN1 staining (red) depicts mature CNN1⁺ covered (yellow, arrows) and NG2⁺ (green) pericytes. Scale bar: 100 μm. (C) Quantification of total NG2 signals and CNN1 covered NG2⁺ pericytes in untreated meningioma slices (day 3, day 5) and slices incubated with 50 mM trametinib for 3 and 5 days (D3, D5). n = 3 patients. Data are represented as mean ± SEM. ****P < 0.0001, 1-way ANOVA. (D) Prognostic value of a contractile gene signature for disease progression in a metastatic melanoma patient cohort comparing top and bottom expression quartiles (n = 29 patients each). P = 0.01, log-rank test.