Skip to main page content
Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
Filters applied. Clear all
. 2018 Dec;6(12):1459-1471.
doi: 10.1158/2326-6066.CIR-18-0086. Epub 2018 Sep 12.

Stromal Fibroblasts Mediate Anti-PD-1 Resistance via MMP-9 and Dictate TGFβ Inhibitor Sequencing in Melanoma

Affiliations
Free PMC article

Stromal Fibroblasts Mediate Anti-PD-1 Resistance via MMP-9 and Dictate TGFβ Inhibitor Sequencing in Melanoma

Fei Zhao et al. Cancer Immunol Res. .
Free PMC article

Abstract

Although anti-PD-1 therapy has improved clinical outcomes for select patients with advanced cancer, many patients exhibit either primary or adaptive resistance to checkpoint inhibitor immunotherapy. The role of the tumor stroma in the development of these mechanisms of resistance to checkpoint inhibitors remains unclear. We demonstrated that pharmacologic inhibition of the TGFβ signaling pathway synergistically enhanced the efficacy of anti-CTLA-4 immunotherapy but failed to augment anti-PD-1/PD-L1 responses in an autochthonous model of BRAFV600E melanoma. Additional mechanistic studies revealed that TGFβ pathway inhibition promoted the proliferative expansion of stromal fibroblasts, thereby facilitating MMP-9-dependent cleavage of PD-L1 surface expression, leading to anti-PD-1 resistance in this model. Further work demonstrated that melanomas escaping anti-PD-1 therapy exhibited a mesenchymal phenotype associated with enhanced TGFβ signaling activity. Delayed TGFβ inhibitor therapy, following anti-PD-1 escape, better served to control further disease progression and was superior to a continuous combination of anti-PD-1 and TGFβ inhibition. This work illustrates that formulating immunotherapy combination regimens to enhance the efficacy of checkpoint blockade requires an in-depth understanding of the impact of these agents on the tumor microenvironment. These data indicated that stromal fibroblast MMP-9 may desensitize tumors to anti-PD-1 and suggests that TGFβ inhibition may generate greater immunologic efficacy when administered following the development of acquired anti-PD-1 resistance.See related Spotlight on p. 1444.

Conflict of interest statement

The authors declare no potential conflicts of interest.

Figures

Figure 1.
Figure 1.. TGFβ inhibition augments anti–CTLA-4 immunotherapy in an autochthonous BRAFV600EPTEN-/- melanoma model.
A. Mice were treated with either anti–CTLA-4 (purple, 100 μg i.p.) or IgG isotype control (black, 100 μg i.p.) every 3 days when tumors reached 60–80 mm3. Left: tumor volumes monitored every 3 days. Right: representative tumor photos. 6 mice/group. Representative of 3 independent experiments. B. Left: resected BrafV600EPten-/- melanoma tissues analyzed for TGFβ1 expression by IHC and whole tissue Western blots. IHC representative of 3 tumor specimens (40x). Right: Spearman correlation between tumor volume and TGFβ1/β-actin density ratios from Western blots. C. Phospho-SMAD2 (pSMAD2) Western blot analysis performed following TEW-7197 treatment (25 mg/kg daily p.o. x 2 weeks). tSMAD2: total SMAD2. Representative of 2 independent experiments. D. Left: mice were treated with TEW-7197 monotherapy (25 mg/kg p.o. daily). Tumor volumes were monitored every 3 days. 6 mice/group. Representative of 3 independent experiments. Center: final tumor weights. Right: representative photos of resected tumors. ns: non-significant. E. Left: mice were treated with either IgG isotype control/vehicle control, TEW-7197(25 mg/kg p.o. daily) monotherapy, or combination anti–CTLA-4 (100 μg i.p. every 3 days)/TEW-7197. Left: tumor volumes monitored every 3 days. 6 mice/group. Black arrow, treatment initiation. Representative of 2 independent experiments. Right top: lungs resected and enumerated for metastatic foci by staggered IHC. Right bottom: Kaplan-Meier survival plot. Data analyzed by the log-rank test. F. Left: flow cytometry on tumor-infiltrating lymphocytes (TILs) at the conclusion of the experiment. 5 tumors/group. Right: representative flow diagrams. See Supplementary Fig. S1. All data is mean±SEM. Significance calculated using the unpaired t-test or a one-way ANOVA. *p<0.05.
Figure 2.
Figure 2.. TGFβ inhibition suppresses the PD-1/PD-L1 signaling axis and fails to augment anti–PD-1 immunotherapy in an autochthonous BRAFV600EPTEN-/- melanoma model.
A. Mice treated with either anti–PD-1(250 μg i.p.) or IgG isotype control (250 μg i.p.) every 3 days±TEW-7197 (25 mg/kg po daily) when tumors reach 60–80 mm3. Tumor volumes monitored every 3 days. 6 mice/group. Representative of 2 independent experiments. B. Mice treated with either anti–PD-L1(200 μg i.p.) or IgG isotype control (200 μg i.p.) every 3 days ±TEW-7197(25 mg/kg p.o. daily). Tumor volumes monitored every 3 days. 6 mice/group. Representative of 2 independent experiments. C. Left: autochthonous and syngeneic transplant BrafV600EPten/– melanoma tissues were analyzed with CD8 immunofluorescence (10x) and PD-L1 IHC (20x). Representative of 3 tumors. Right top: flow cytometry of infiltrating CD45+CD3+CD8+ T cells in both autochthonous and syngeneic BrafV600EPten/– melanoma tissues. Gated on viable CD45+ cells. Right bottom: CD3+CD8+ T cells (% of CD45+ cells). D. Syngeneic BrafV600EPten/– melanoma model treated with either anti–PD-1 (250 μg i.p.), anti–CTLA-4(100 μg i.p.), or IgG isotype control (250 μg i.p.) every 3 days±TEW-7197 (25 mg/kg p.o. daily). Tumor volumes were monitored every 3 days. 7 mice/group. Representative of 2 independent experiments. E. Flow cytometry for TILs at the conclusion of the experiment in D. F. Mice from the autochthonous BrafV600EPten/– melanoma model were treated with TEW-7197 (25 mg/kg po daily x 2 weeks). Primary melanoma tissues were resected, scored (Top), and analyzed for PD-L1 expression by IHC (red, Bottom). Representative of 5 tumors (40x). IHC scores were calculated for 10 random fields in 5 tumors/group. G. Autochthonous BrafV600EPten/– melanoma model was treated with TEW-7197 (25 mg/kg p.o. daily x 2 weeks). Primary melanoma tissues were resected and flow cytometry performed to quantitate PD-L1 surface expression on CD45EpCAMCD90.2 cells. 3 tumors/group. Representative of 2 independent experiments. See Supplementary Fig. S2. All data is mean±SEM. Significance calculated using an unpaired t-test or a one-way ANOVA. *p<0.05, **p<0.005.
Figure 3.
Figure 3.. TGFβ inhibition expands melanoma stromal fibroblasts.
Autochthonous BrafV600EPten/– melanoma model treated with TEW-7197 (25 mg/kg po daily x 2 weeks), and tumors resected for A. trichrome staining (blue: connective tissue), B. Vimentin IHC, and C. α-SMA IHC. All histology are representative of 3 tumors/group and ≥6 sections/tumor. D. Autochthonous BrafV600EPten/– melanoma model was treated with TEW-7197 (25 mg/kg p.o. daily x 2 weeks) versus a vehicle control, resected, and single-cell suspensions were generated. Flow cytometry used to quantitate CD45EpCAMCD90.2+ melanoma-associated fibroblasts (MAFs). 3 tumors/group. Representative of 2 independent experiments. E. Left: in vitro transwell migration assay. The BrafV600EPten/– melanoma cell line treated with TEW-7197 and BrafV600EPten/–/MAF chemotaxis measured. 5 wells/condition. Representative of 2 independent experiments. Right: in vitro proliferation assay. BrafV600EPten/–/MAFs treated with increasing concentrations of TEW-7197 and cellular proliferation was monitored by MTT assay. 3 wells/condition. Representative of 3 independent experiments. F. Schematic of the in vivo MAF proliferation assay. 4-HT: 4-hydroxytamoxifen, EdU: 5-ethynyl-2’-deoxyuridine DNA incorporation dye. G. Quantitation of EdU-FAM+CD45EpCAMCD90.2+ proliferating MAFs in vehicle control and TEW-7197-treated autochthonous BrafV600EPten/– melanomas. 4 mice/group. Representative of 2 independent experiments. See Supplementary Fig. S3 and S4. All data is mean±SEM. Significance calculated using an unpaired t-test or a one-way ANOVA. *p<0.05, **p<0.005. ns: non-significant.
Figure 4.
Figure 4.. MAFs suppress anti–PD-1 activity in the BrafV600EPten-/- melanoma model.
A. BrafV600EPten/– melanoma cells transplanted ± BrafV600EPten/–/MAFs (1:3 ratio) into syngeneic C57BL/6 mice and treated with anti–PD-1 (250 μg i.p. every 3 days) or IgG isotype control (250 μg i.p. every 3 days) when tumors reach 60–80 mm3. Left: tumor volumes monitored every 3 days. 6 mice/group. Representative of 2 independent experiments. Right, representative photos of resected primary melanomas. B. Right: flow cytometry for tumor-infiltrating CD3+CD8+ T cells from A. Right, representative flow diagrams. 5 tumors/group. C. IFNγ by TRP2-specific T cells at the conclusion of the experiment in A. Splenocytes harvested from 5 mice/group. Representative of 2 independent experiments. D. Dual IHC analysis of α-SMA (red) and CD8 (brown) in BrafV600EPten/–/MAF and BrafV600EPten/– melanoma tissues. 3 tumors/group. 10x images representative of 10 fields/tumor. E. RNAseq analysis of BrafV600EPten/–/MAF and BrafV600EPten/– melanoma tissues. Left: enrichment plot showing downregulation of IMMUNE_RESPONSE GO pathway from BrafV600EPten/–/MAF melanomas versus BrafV600EPten/– melanomas. Right: association of the most downregulated GO pathways in co-transplanted BRAFV600EPTEN/–/MAF melanomas with cellular immunity. NES: normalized enrichment score. All data is mean±SEM. Significance calculated using a one-way ANOVA. *p<0.05.
Figure 5.
Figure 5.. Fibroblast MMP-9 negatively regulates melanoma surface PD-L1 expression.
A. Left: PD-L1 expression from in vitro cocultures of BrafV600EPten/– melanoma cells±MAFs. Representative of 3 independent experiments. Right, flow diagram of BrafV600EPten/– PD-L1 expression from different conditions. Black: unstained control, Blue: +MAFs, Green: –MAFs. B. Left: PD-L1 expression of in vivo BrafV600EPten/– tumors±MAFs. 5 mice/group. Right, representative IHC of PD-L1 (40x). C. Mmp9 mRNA expression of implanted BrafV600EPten/– tumors±MAFs from RNAseq (Left), validated by qRT-PCR (Right). D. Mmp9 mRNA expression by BrafV600EPten/– tumor cells and MAFs when cultured separately or cocultured. 3 independent replicates. E. PD-L1 expression of BRAFV600EPTEN/–±MAFs with the indicated concentration of MMP9i (MMP9 inhibitor I). 3 independent replicates. F. MMP-9 expression in stable cell lines BrafV600EPten/–-MAFNTC (NTC: non-targeted control) and BrafV600EPten/–-MAFMmp9KD (KD, Mmp9 knock-down). Left: Western blot. Right: qRT-PCR. G. PD-L1 expression by BrafV600EPten/–-melanoma cells following coculture with the BrafV600EPten/–-MAFNTC and the BrafV600EPten/–-MAFMmp9KD cell lines. Representative of 3 independent experiments. H. Tumor measurements from BrafV600EPten/– melanomas co-transplanted with either BrafV600EPten/–-MAFNTC or BrafV600EPten/–-MAFMmp9KD fibroblasts treated with either IgG control or anti–PD-1. 5–6 mice/group. I. Final tumor volumes from experiment in H. J. Flow cytometry for TILs performed at the conclusion of the experiment in H. See Supplementary Fig. S5. All data is mean±SEM. Significance calculated using the unpaired t-test or a one-way ANOVA. *p<0.05, ***p<0.0005.
Figure 6.
Figure 6.. Stromal expansion and TGFβ signaling activation is associated with anti–PD-1 resistance.
A. Autochthonous BrafV600EPten/– melanoma model treated with either anti–PD-1(250 μg i.p. every 3 days) or IgG isotype control (250 μg i.p. every 3 days). Tumor volumes monitored every 3 days. Tumors resected following escape and progression during anti–PD-1 therapy, total RNA isolated, and RNAseq analysis conducted. 3 age/sex-matched mice/group. B. Left: principal component analysis of gene expression in response to anti–PD-1 versus IgG isotype control. Center: enrichment plot showing enhancement of Extracellular Space GO pathway upon anti–PD-1 treatment. Right: individual collagen-expressing genes in tumors treated with anti–PD-1 versus IgG isotype control. NES: normalized enrichment score. C. Upregulated gene expression in 13 anti–PD-1 refractory advanced melanoma patients based on GO term enrichment analysis of an available RNAseq database (22). D. Genes associated with TGFβ signaling activation in anti–PD-1-treated melanoma patients in C (22). E. Left: enrichment plot showing enhancement of TGFβ gene expression targets in the BrafV600EPten/– melanoma model following escape from anti–PD-1 therapy. Right: serial biopsy Tgfb1 qRT-PCR analysis of the transgenic BrafV600EPten/– melanomas undergoing treatment with the IgG isotype control or anti–PD-1. 6 mice/group. See Supplementary Fig. S6. All data is mean±SEM. Significance calculated using the unpaired t-test. *p<0.05, ***p<0.0005.
Figure 7.
Figure 7.. Delayed dosing of TGFβ inhibitor TEW-7197 improves anti–PD-1 therapy in the transgenic BrafV600EPten-/- melanoma model.
A.Top: delayed dosing scheme of anti–PD-1 and TEW-7197. Bottom: tumor measurements before and after TEW-7197 initiation following previous anti–PD-1 therapy. 6 mice/group. B. Final tumor volume measurements from A. C. Flow cytometry for TILs performed at the conclusion of the experiment in A. D. IFNγ ELISPOT analysis of TRP2-specific T cells performed in A. Splenocytes harvested from 5 mice/group. Representative of 2 independent experiments. E. Quantitation of EdU-FAM+CD45EpCAMCD90.2+ proliferating MAFs in autochthonous BrafV600EPten/– melanomas following the indicated treatments as performed in A. 3 tumors/group. Representative of 2 independent experiments. F. Tumor tissues resected for trichrome staining and G. α-SMA IHC. All histology representative of 3 tumors/group and at least 6 sections/tumor. H. Mice treated as indicated by the experiment described in A. Primary melanoma tissues resected at the same time points as in A-D. Flow cytometry utilized to quantitate PD-L1 surface expression on CD45EpCAMCD90.2 cells. 3 tumors/group. See Supplementary Fig. S7. All data is mean±SEM. Significance calculated using a one-way ANOVA. *p<0.05.

Comment in

Similar articles

See all similar articles

Cited by 10 articles

See all "Cited by" articles

Publication types

MeSH terms

Feedback