Loss of SMAD4 Is Associated With Poor Tumor Immunogenicity and Reduced PD-L1 Expression in Pancreatic Cancer

Abstract

Transforming Growth Factor β (TGFβ) is a key mediator of immune evasion in pancreatic ductal adenocarcinoma (PDAC), and the addition of TGFβ inhibitors in select immunotherapy regimens shows early promise. Though the TGFβ target SMAD4 is deleted in approximately 55% of PDAC tumors, the effects of SMAD4 loss on tumor immunity have yet to be fully explored. Using a combination of genomic databases and PDAC specimens, we found that tumors with loss of SMAD4 have a comparatively poor T-cell infiltrate. SMAD4 loss was also associated with a reduction in several chemokines with known roles in T-cell recruitment, which was recapitulated using knockdown of SMAD4 in PDAC cell lines. Accordingly, JURKAT T-cells were poorly attracted to conditioned media from PDAC cells with knockdown of SMAD4 and lost their ability to produce IFNγ. However, while exogenous TGFβ modestly reduced PD-L1 expression in SMAD4-intact cell lines, SMAD4 and PD-L1 positively correlated in human PDAC samples. PD-L1 status was closely related to tumor-infiltrating lymphocytes, particularly IFNγ-producing T-cells, which were more abundant in SMAD4-expressing tumors. Low concentrations of IFNγ upregulated PD-L1 in tumor cells in vitro, even when administered alongside high concentrations of TGFβ. Hence, while SMAD4 may have a modest inhibitory effect on PD-L1 in tumor cells, SMAD4 indirectly promotes PD-L1 expression in the pancreatic tumor microenvironment by enhancing T-cell infiltration and IFNγ biosynthesis. These data suggest that pancreatic cancers with loss of SMAD4 represent a poorly immunogenic disease subtype, and SMAD4 status warrants further exploration as a predictive biomarker for cancer immunotherapy.

Keywords: interferon γ (IFNγ); pancreatic ductal adenocarcinoma (PDAC); programmed death-ligand 1 (PD-L1); transforming growth factor β (TGFβ); tumor immunology; tumor mircorenvironment.

Figure 1

Tumors with loss of SMAD4 display reduced lymphocyte infiltration independent of neoadjuvant chemotherapy status. (A) Excisional biopsies from 36 PDAC patients were sectioned and stained either with H&E or via immunohistochemistry for SMAD4, the pan-leukocyte antigen CD45, or T-cell marker CD3 and representative images shown for each from either chemotherapy naïve patients (N=18) or patients who had received neoadjuvant Gemcitabine-based chemotherapy (N=18). (B, C) The percent of patients from either the chemo-naïve or neoadjuvant Gemcitabine group that was either SMAD4-expressing (SMAD4+) or SMAD4-non-expressing (SMAD4-). (D, E) The number of CD45+ or CD3+ cells per 40X field was quantified by three blinded investigators, related to chemotherapy status, and displayed as an individual value plot. Using these values, the number of CD45 positive cells was next related to SMAD4 status in either (F) the chemo naïve group or (G) the neoadjuvant Gemcitabine group and displayed as an individual value plot. (H, I) The number of CD3+ T-cells were quantified as described and related to SMAD4 status in either the chemo-naïve group or the neoadjuvant Gemcitabine group and displayed as an individual value plot. (*p < 0.05).

Figure 2

SMAD4 expression and increased T-cell infiltration predict for improved overall survival in PDAC patients. Kaplan-Meier curve indicating months of overall survival following surgical resection for a cohort of 36 patients arranged by: (A) SMAD4 status determined by immunohistochemistry, (B) SMAD4 status for only the patients that received no neoadjuvant therapy, (C) SMAD4 status for only the patients that received neoadjuvant Gemcitabine-based chemotherapy, (D) Patients above or below the median value for CD3+ T-cells per high power field, (E) T-cells status for only the patients that received no neoadjuvant therapy, (F) T-cells status for only the patients that received neoadjuvant Gemcitabine-based chemotherapy.

Figure 3

Loss of SMAD4 impairs CCL, CXCL, and IL-family cytokine synthesis in PDAC cells. An equal number of PANC-1 cells were seeded into 6-well plates and incubated either with a control siRNA (siControl) or siRNA against SMAD4 (siSMAD4). After 24 hours, cells were treated with either a saline vehicle or 10ng/mL of recombinant TGFβ1. Following another 24 hours, cells were incubated with a protein transport inhibitor for one hour, lysed, and 200μg of total cell lysate was evaluated by a high throughput proteome profiler array (ARY022B). Pixel density was evaluated using ImageJ, and samples normalized to the mean intensity of the reference spots for each blot minus the background density. Values are presented as fold change for (A) CCL family cytokines/chemokines, (B) CXCL and IL family cytokines/chemokines, (C) additional immunomodulatory proteins. (*p < 0.05).

Figure 4

JURKAT T-cells remain refractory from full activation when grown in conditioned media from SMAD4-deficient tumor cells. (A) PANC-1 tumor cells were incubated with either a control siRNA (siControl) or siRNA against SMAD4 (siSMAD4) and stimulated with 10ng/mL of recombinant TGFβ1 after 24 hours. Four hours after treatment, media was changed to serum-free DMEM and collected after another 24 hours. This media was supplemented with 10% FBS and 2μl/mL ImmunoCult Human CD3/CD28 T Cell Activator and administered to 1 million serum-starved JURKAT T-cells. After 24 hours, JURKAT cells were collected, incubated with a protein transport inhibitor for one hour, and analyzed for T-cell activation by flow cytometry for the activation markers CD69, IFNγ, or CD69 and IFNγ. (B) The modal expression of CD69 and IFNγ is displayed as a histogram plot. (C) Using the described gating, the relative percent of CD69+ and IFNγ+ events are plotted, as are the absolute number of each per 10,000 events (*p < 0.05). (D) MIA PaCa-2 tumor cells were incubated with either a siControl or siSMAD4, treated similarly, and media collected as described. This media was supplemented with 10% FBS and 2μl/mL ImmunoCult Human CD3/CD28 T Cell Activator, and administered to 1 million serum-starved JURKAT T-cells, which were analyzed by flow cytometry as described previously. (E) The modal expression of CD69 and IFNγ is displayed as a histogram plot. (F) Using the described gating, the relative percent of CD69+ and IFNγ+ events are plotted, as are the absolute number of each per 10,000 events. (*p < 0.05).

Figure 5

TGFβ/SMAD signaling downregulates PD-L1 expression in vitro, yet SMAD4-intact tumors have higher expression of PD-L1 in vivo. (A) Non-malignant HPNE cells and human PDAC cell lines PANC-1, MIA PaCa-2, CaPan-1, CaPan-2, BxPC-3, and AsPC-1 were lysed and analyzed for basal expression of SMAD4, PD-L1, and HLA-A,B,C by western blot. (B-D) PANC-1, BxPC-3, and AsPC-1 cells were incubated with either a saline vehicle or 10ng/mL of recombinant TGFβ1 and evaluated after 24 hours by western blot. (E) PANC-1 and MIA PaCa-2 cells were incubated with either a control siRNA (siControl) or siRNA against SMAD4 (siSMAD4), and after 24 hours, CD274 (PD-L1) mRNA was evaluated by qPCR. (F, G) PANC-1 and MIA PaCa-2 cells were incubated with either siControl or siSMAD4 and, after 24 hours, stimulated with 10ng/mL of recombinant TGFβ1. Cells were lysed after another 24-hour period and evaluated by western blot analysis. (H) BxPC-3 cells were transfected with a wild-type SMAD4 plasmid (SMAD4^WT). After 24 hours, cells were stimulated with 10ng/mL of recombinant TGFβ1 and evaluated by western blot analysis after another 24 hours. (I) Using the TCGA genomic database of pancreatic cancer patients (N=186), the 149 fully sequenced tumors were separated into two groups: those with no SMAD4 alteration (SMAD4 wild-type or WT), and those with presumptive SMAD4 loss via a known inactivating mutation, mRNA downregulation, and/or copy number deletion. We then compared the mRNA expression of CD274 in each group. (J) SMAD4 mRNA expression was plotted against that of CD274. All mRNA expression values are plotted in log scale and are displayed with the associated p and Spearmen (S) coefficient values. (K) Excisional biopsies from two cohorts of PDAC patients (N=44 and N=36, respectively) were sectioned and stained via immunohistochemistry for SMAD4 or PD-L1. (L-N) Patients were grouped as being either SMAD4-expressing (SMAD4+) or SMAD4-non-expressing (SMAD4-), and the percent of each group also positive for PD-L1 displayed as a pie chart. (*p < 0.05).

Figure 6

IFNγ overcomes the inhibitory effects of TGFβ/SMAD signaling on PD-L1 expression both in vivo and in vitro. (A) Excisional biopsies from 44 PDAC patients in cohort A and 36 PDAC patients in cohort B were sectioned and stained with H&E, lymphocytes quantified per 40X field, and arranged by PD-L1 status. (B-D) Using the TCGA genomic database of pancreatic cancer patients, CD274 mRNA expression was plotted against that of CD3E, CD3G, and IFNG. (E-G) Also, using the TCGA genomic database, IFNG mRNA expression was plotted against that of CD3E, CD3G, or SMAD4. (H) The 36 excisional PDAC specimens from cohort B were stained by immunohistochemistry for IFNγ, as well as dual-stained for either the duct marker CK19 and IFNγ, epithelial surrogate marker E-Cadherin and IFNγ, or the T-cell marker CD3 and IFNγ. The percent area positive for IFNγ was quantified as described and related to SMAD4 status. (I) The number of CD3+IFNγ+ cells were quantified per 40X field and arranged by both SMAD4 and PD-L1 status. (J) PANC-1 cells were incubated with 1ng/mL of recombinant IFNγ in the presence of increasing doses of recombinant TGFβ1, and PD-L1 expression evaluated by western blot after 24 hours. (K) PANC-1 cells were again incubated with either siControl or siSMAD4 and, after 24 hours, stimulated with 10ng/mL of recombinant TGFβ1. This experiment was also conducted in the presence of 1ng/mL recombinant IFNγ given concurrently with TGFβ1, and 24-hours after stimulation, cells were evaluated by western blot. (*p < 0.05).