PD-L1 expressed from tumor cells promotes tumor growth and invasion in lung cancer via modulating TGF-β1/SMAD4 expression.

BACKGROUND: Programmed death ligand-1 (PD-L1) has a known association with the prognosis of human cancers because of its ability to alter tumor immune surveillance via its interaction with PD-1. We questioned whether expression of PD-L1 in tumor cells could directly promote tumor growth and invasiveness in non-small cell lung cancer (NSCLC).
METHODS: Real-time reverse transcription-polymerase chain reaction (RT-PCR) was performed to evaluate PD-L1 messenger RNA (mRNA) expression in lung tumors. The prognostic value of PD-L1 mRNA was assessed by Cox regression model. Transcriptional regulation of PD-L1 by human papillomavirus (HPV) 16/18 E6 oncoprotein or by epidermal growth factor receptor (EGFR) mutation in lung cancer cells was examined by Western blot and luciferase reporter assay. The cell growth and invasion were evaluated by colony formation, soft agar growth, and Boyden chamber assay.
RESULTS: The PD-L1 mRNA levels showed a positive association with HPV 16/18 E6 oncoprotein and with EGFR mutation in 223 surgically resected NSCLC patients. The prognostic significance of PD-L1 was more commonly observed in patients with high PD-L1/E6 positive and high PD-L1/EGFR mutant tumors. Mechanistically, upregulation of PD-L1 transcription by E6 or mutant EGFR occurred largely through the ERK-C/EBPβ-TLR4-NF-κB cascade. PD-L1 promotes the efficacy of colony formation, soft agar growth, and cell invasion. PD-L1 upregulates BAG-1 to reduce transforming growth factor (TGF)-β1 expression, and the decrease in SMAD4 because of TGF-β1 occurs through the p53/microRNA (miR)-224 axis. The decreases in TGF-β1 and SMAD4 are responsible for PD-L1-mediated cell invasiveness.
CONCLUSION: Induction of PD-L1 by E6 oncoprotein or mutant EGFR through the ERK-C/EBPβ-TLR4-NF-κB cascade may promote tumor growth and invasiveness in NSCLC because of decreasing TGF-β1 and SMAD4 expression.

INTRODUCTION

Programmed death‐1 (PD‐1) is a costimulatory molecule that provides an inhibitory signal in T cell activation. It has two identified ligands, PD‐L1 (B7‐H1) and PD‐L2 (B7‐DC), which are cell‐surface glycoproteins belonging to the B7 family. , , Mounting evidence indicates that PD‐L1 is associated with poor outcomes in several different human cancers, including renal cell, , gastric, bladder, ovarian, , , melanoma, skin, hepatocellular, , , pancreatic, colorectal cancers, and non–small cell lung cancers (NSCLC). , ,

Increases in PD‐L1 expression occur in response to several different pathologies. For example, the loss of phosphatase and tensin homolog (PTEN) increases PD‐L1 expression in glioma, colorectal cancer, and lung squamous cell carcinoma. Similarly, PD‐L1 upregulation in lung cancer occurs through the MEK/ERK, nuclear factor (NF)‐κB, epidermal growth factor receptor (EGFR), and EML4‐ALK signaling pathways, and its expression is higher in patients with mutant EGFR than with wild‐type EGFR. , Enhanced expression of PD‐L1 has also been reported in cervical intraepithelial neoplasia and cervical cancers, and its correlation with high risk human papillomavirus (HPV) infection. PD‐L1 is now suggested as a biomarker of productive HPV infection of invasive squamous cervical cancer cells. , Our preliminary data seemed to support these findings that PD‐L1 expression was higher in Taiwanese NSCLC patients with HPV16/18 E6‐tumors than with E6‐negative tumors. In addition, EGFR mutations were more frequently observed in NSCLC patients with E6‐positive tumors than with E6‐negative tumors. However, the precise signaling pathway(s) involved in regulation of PD‐L1 expression in lung cancer, either by HPV infection or by EGFR mutation, remains to be elucidated although the review literatures have been documented some related signaling pathways involved in regulation of PD‐L1. ,

In the present study, we hypothesized that PD‐L1 highly expressed from E6‐positive and/or EGFR‐mutated tumor cells, not from immune cells, could directly promote tumor growth and invasion in NSCLC, and consequently to result in patients with poorer outcomes.

MATERIALS AND METHODS

Study population

Lung tumor specimens were collected from 223 patients with primary lung cancer. All patients were admitted to the Department of Thoracic Surgery at Taichung Veteran's General Hospital, Taiwan, between 1998 and 2008. All patients were asked to submit a written informed consent based on a biology study approved by the institutional review board. Tumor types and stages were histologically determined according to the World Health Organization (WHO) (1981) classification.

Quantitative real‐time reverse transcription‐polymerase chain reaction

Total RNA (5 μg) was used in complementary (c)DNA synthesis with random primers using Superscript III reverse transcriptase (Applied Biosystems). Relative messenger (m)RNA expression was calculated with the comparative C t method (ΔΔC t). 18S ribosomal ribonucleic acid (r)RNA was used for normalization.

Plasmids and transfection

All transfection experiments were performed with TransFast transfection reagents (Promega), in accordance with the manufacturer's protocols.

Doubling time

TL1 and TL2 cells (103/ml were transfected with two doses of E6 small interfering (si)RNA (1 and 5 μg) and A549 and TL4 cells were transfected with two doses of E6 expression plasmid (1 and 5 μg). These transfected cells were seeded in a 35‐mm dish and cultured for 24, 48, 72, 96, 120, and 144 hours, and then the cell number at each culture time point was counted for calculation of the doubling time.

Soft agar colony formation assay

Cells (3000 per well) were cultured on a 6‐well plate containing 1% base agar and 0.35% top agar in the medium described above and incubated at 37°C for 21 days.

Boyden chamber assay

A Boyden chamber with a pore size of 8 μm (Falcon) was used for the invasion assay. Each condition was assayed in triplicate.

Statistical analysis

Statistical analysis was performed using the SPSS statistical software program (Version 11.0 SPSS). The association between PD‐L1 and clinicopathologic variables was analyzed using the Pearson χ2 or Fisher's exact test, as appropriate. p values of <0.05 were considered statistically significant.

RESULTS

PD‐L1 mRNA expression is positively correlated with HPV16/18 E6 oncoprotein expression and EGFR mutation in NSCLC patients

PD‐L1 mRNA levels in lung tumors were evaluated by real‐time polymerase chain reaction (PCR). The data on E6 oncoprotein expression and EGFR mutation were collected from previous studies , and our own database. We used the median value of PD‐L1 mRNA levels in this study population as a cutoff point to divide patients into low and high PD‐L1 subgroups. High PD‐L1 mRNA expression occurred more common in females, nonsmokers, patients with adenocarcinoma, and late‐stage patients when compared with their counterparts (Table 1). High PD‐L1 mRNA expression was more frequently observed in patients who were E6‐positive or had mutant EGFR than in patients who were E6‐negative or had wild‐type EGFR (Table 1). These results were consistent with previous studies indicating that E6 expression and EGFR mutation were more common in females, nonsmokers, and patients with adenocarcinoma than in males, smokers, and patients with squamous cell carcinoma. , These findings, therefore, indicate that E6 or EGFR mutations might contribute to increases in PD‐L1 expression in NSCLC.

TABLE 1

Correlation between PD‐L1 mRNA expression and clinicopathologic features in lung cancer patients

Variables	N	PD‐L1		p
		Low	High
Age, y				0.160
<67	119	54 (45)	65 (55)
≥67	104	57 (55)	47 (45)
Gender				<0.001
Female	73	22 (30)	51 (70)
Male	150	89 (59)	61 (41)
Smoking				0.003
No	121	49 (41)	72 (59)
Yes	102	62 (61)	40 (39)
Type				<0.001
AD	116	40 (34)	76 (66)
SQ	107	71 (66)	36 (34)
Stage				<0.001
I	82	48 (59)	34 (41)
II	47	33 (70)	14 (30)
III	94	30 (32)	64 (68)
HPV E6 protein				<0.001
Negative	163	97 (60)	66 (40)
Positive	60	14 (23)	46 (77)
EGFR mutation				0.018
Negative	163	89 (55)	74 (45)
Positive	60	22 (37)	38 (63)

AD: adenocarcinoma; SQ: squamous cell carcinoma

High PD‐L1 mRNA levels are associated with poorer overall survival and relapse free survival in NSCLC patients

Cox regression analysis indicated that shorter overall survival (OS) and relapse free survival (RFS) periods were associated with tumors expressing high levels of PD‐L1 mRNA when compared with tumors expressing low levels of PD‐L1 mRNA (Table 2). By contrast, no prognostic value was observed for E6 and EGFR mutation on OS and RFS in this study population. Interestingly, the combination of PD‐L1 expression and either E6 expression or mutant EGFR also showed a prognostic significance in terms of OS and RFS in patients with high PD‐L1/E6 positive and high PD‐L1/EGFR mutant tumors, when compared with low PD‐L1/E6 negative or low PD‐L1/EGFR wild‐type tumors as the reference (Table 3). These results suggest that PD‐L1 expression may confer poorer outcomes in patients with NSCLC and especially in those with E6‐positive or EGFR mutant tumors.

TABLE 2

Cox regression analysis for the influence of HPV16/18 E6, PD‐L1 and combined effects on OS and RFS in lung cancer patients

		OS					RFS
	N	Median, mo	Survival rate at 5 y (%)	HR a	95% CI	p	Median, mo	Survival rate at 5 y (%)	HR a	95% CI	p
E6 protein
Negative	163	30.6	39.9	1			23.9	28.0	1
Positive	60	26.4	18.3	1.32	0.92–1.91	0.130	16.8	15.0	1.39	0.97–1.97	0.066
PD‐L1
Low	111	60.8	53.2	1			37.7	36.9	1
High	112	18.9	15.2	2.69	1.83–3.93	<0.001	13.8	13.4	2.01	1.40–2.85	<0.001
PD‐L1/HPV E6
−/−	92	60.8	52.2	1			40.0	38.0	1
−/+	13	‐	53.8	1.06	0.52–2.15	0.860	78.7	46.2	0.77	0.34–1.70	0.521
+/−	71	19.5	23.9	1.41	1.15–1.74	0.001	14.1	16.9	1.32	1.08–1.61	0.026
+/+	47	25.9	8.5	2.19	1.48–3.26	<0.001	14.8	6.4	1.84	1.29–2.38	0.001

All HR were adjusted for age, gender, smoking status, stage, and type.

TABLE 3

Cox regression analysis for the influence of EGFR mutation, PD‐L1 and combined effects on OS and RFS in lung cancer patients

	N	OS					RFS
		Median, mo	Survival rate at 5 y (%)	HR a	95% CI	p	Median, mo	Survival rate at 5 y (%)	HR a	95% CI	p
EGFR mutation
Negative	163	27.8	34.4	1			19.1	25.8	1
Positive	60	30.6	33.3	0.98	0.65–1.47	0.927	25.9	23.3	0.89	0.61–1.30	0.568
PD‐L1
Low	111	60.8	53.2	1			37.7	36.9	1
High	112	18.9	15.2	2.69	1.83–3.93	<0.001	13.8	13.4	2.01	1.40–2.85	<0.001
PD‐L1/EGFR mutation
−/−	89	60.8	50.6	1			37.7	37.1	1
−/+	22	71.0	63.6	0.66	0.28–1.52	0.331	41.2	36.4	0.81	0.41–1.58	0.543
+/−	74	17.7	14.9	1.30	1.15–1.61	0.016	12.3	12.2	1.53	1.25–1.87	0.01
+/+	38	25.9	15.8	1.63	1.32–2.01	0.011	12.2	15.8	1.74	1.09–2.76	0.019

All HR were adjusted for age, gender, smoking status, stage, and type.

PD‐L1 promotes in vitro cell proliferation, colony formation, invasion, and soft agar growth

Two high PD‐L1‐expressing cell lines (HPV16 E6‐positive TL1 and TL2) and two low PD‐L1‐expressing cell lines (HPV16 E6‐negative A549 and TL4) were collected and used for PD‐L1 manipulation using a small hairpin (sh)RNA and an expression vector of PD‐L1. PD‐L1 knockdown in TL1 and TL2 cells prolonged their doubling time and decreased their ability for colony formation, soft agar growth, and cell invasion (Figure 1(a),(b)). PD‐L1‐overexpressing A549 and TL4 cells, by contrast, showed reduced doubling time and enhanced colony formation, soft agar growth, and cell invasion (Figure 1(c),(d)). These results indicate that PD‐L1 expression may be responsible for cell proliferation, colony formation, invasion, and soft agar growth in HPV16 E6‐postive or ‐negative lung cancer cells.

FIGURE 1

PD‐L1 expression may be responsible for cell proliferation, colony formation, invasion, and soft agar growth in lung cancer cells. (a) TL1 and (b) TL2 cells were transfected with shPD‐L1; (c) A549 and (d) TL4 cells were transfected with a PD‐L1 expression plasmid for 48 hours. PD‐L1 mRNA and protein expression in four lung cancer cell types were evaluated by RT‐PCR analysis and Western blotting. PD‐L1, programmed death ligand‐1

Induction of PD‐L1 transcription by the E6 oncoprotein occurs through the C/EBPβ‐TLR4‐NF‐κB cascade

We examined the possibility that an increase in PD‐L1 expression because of the E6 oncoprotein could occur through TLR4‐mediated NF‐κB activation. Real‐time PCR analysis indicated that PD‐L1 expression was correlated with TLR4 expression in a panel of lung cancer cell lines (Figure 2(a)). Western blotting showed that the expressions of PD‐L1 and TLR4 were concomitantly decreased by E6 knockdown in TL1 and TL2 cells, but were increased by E6 overexpression in A549 and TL4 cells (Figure 2(b)). An increase in PD‐L1 expression by ectopic E6 expression was almost completely eliminated by TLR4 silencing in A549 and TL4 cells (Figure 2(c)).

FIGURE 2

Induction of PD‐L1 transcription by E6 oncoprotein occurs predominantly through the C/EBPβ/TLR4/NF‐κB cascade. (a) The expression of PD‐L1 and TLR4 mRNA levels in different lung cancer cell lines was determined by real‐time PCR. The changes in PD‐L1 promoter activity (p1405) because of E6 manipulation in TL1 and TL4 cells were evaluated by the luciferase reporter assay. (b) Western blotting was used to evaluate the expression of E6, p53, TLR4, and PD‐L1 in E6‐knockdown TL1 and TL2 cells and in E6‐overexpressing A549 and TL4 cells. (c) A549 and TL4 cells were treated with LPS, E6 expression plasmid, and/or shTLR4 for 48 hours as indicated and then the expressions of E6, PD‐L1, and TLR4 were evaluated by Western blotting. (d) These three TLR4 promoters were transfected into four lung cancer cells as indicated to evaluate their promoter activity using a luciferase reporter assay. (e) PD‐L1 mRNA expression was evaluated by real‐time PCR in HPV16 E6‐positive SiHa and Caski cells and compared with HPV‐uninfected C33A cervical cancer cells (f) C33A cells were treated with LPS, E6 expression plasmid, E6/shCEBPβ, and E6/shTLR4 and then the expression of E6, C/EBPβ, TLR4, and PD‐L1 were evaluated by Western blotting. β‐Actin was used as a loading control. PD‐L1, programmed death ligand‐1

Fourteen C/EBPβ binding sites on the TLR4 promoter were predicted by a software analysis (http://www.genome.jp/tools/motif/, −1405/+1) (Figure 2(d), upper panel). Luciferase reporter assays indicated that the p1405 promoter activity was greater than the p530 and p180 promoter activity (Figure 2(d) upper panel). The p1405 promoter activity was dose‐dependently decreased and increased by E6 manipulation in TL1, TL2, A549, and TL4 cells (Figure 3(d) lower panel). Western blotting showed that the increases in C/EBPβ, TLR4, and PD‐L1 expression by ectopic E6 expression in A549 and TL4 cells were suppressed in a dose‐dependent manner by C/EBPβ silencing (Figure 2(e)). Similar findings were revealed in SiHa, Caski, and C33A cervical cancer cells subjected to the same treatments (Figure 2(f)). These results clearly indicate that induction of PD‐L1 transcription by the E6 oncoprotein occurs predominantly through the C/EBPβ‐TLR4‐NF‐κB cascade.

FIGURE 3

An increase in C/EBPβ expression by the ERK signaling pathway enhances TLR4 expression and then upregulates PD‐L1 transcription. (a) TL1 cells were treated with a tyrosine kinase inhibitor (gefitinib), a MEK/ERK inhibitor (U0126), a PI3K/AKT inhibitor (LY294002), or an NF‐κB inhibitor (BAY) for 48 hours. the changes in expression of p‐ERK, ERK, p‐C/EBPβ (Thr 235), C/EBPβ, TLR4 and PD‐L1 in response to these inhibitors were evaluated by Western blotting. (b) A549 cells were treated with gefitinib, U0126, LY294002, BAY, and/or E6 expression plasmid for 48 hours and then the changes in expression of E6, p‐ERK, ERK, p‐C/EBPβ (Thr 235), C/EBPβ, TLR4, and PD‐L1 were evaluated by Western blotting. The ability of C/EBPβ to bind to the putative binding site on the TLR4 promoter was determined by ChIP analysis. (c) H1975 and H1650 cells were treated with gefitinib, U0126, LY294002, or BAY for 48 hours and then the changes in expression of EGFR, p‐ERK, ERK, p‐C/EBPβ (Thr 235), C/EBPβ, TLR4, and PD‐L1 were evaluated by Western blotting. (d) H1975 cells were transfected with shEGFR and A549 and TL4 cells were transfected with EGFR L858R expression plasmid for 48 hours and then the expressions of EGFR, p‐ERK, ERK, C/EBPβ, TLR4, and PD‐L1 were evaluated by Western blotting. β‐Actin was used as a loading control. PD‐L1, programmed death ligand‐1

Induction of PD‐L1 transcription by the E6 oncoprotein or by mutant EGFR may occur through the ERK signaling pathway

We examined the possibility that an increased expression of C/EBPβ and phosphorylated (p)‐C/EBPβ because of ERK activation could play a role in E6‐mediated TLR4 transcription. TL1 and E6‐overexpressing A549 cells were treated with inhibitors of the EGFR, ERK, PI3K/AKT, and NF‐κB signaling pathway. Western blotting indicated that the expressions of p‐ERK, p‐C/EBPβ (Thr 235), C/EBPβ, TLR4, and PD‐L1 were concomitantly decreased by treatment with an ERK inhibitor (U0126), and partially reduced by an EGFR inhibitor (gefitinib); however, these expressions were unchanged by treatment with inhibitors of PI3K/AKT (LY294002, LY) and NF‐κB (BAY) (Figure 3(a)). Similarly, the expressions of p‐ERK, p‐C/EBPβ (Thr 235), C/EBPβ, TLR4, and PD‐L1 were markedly decreased by U0126 in E6‐overexpressing A549 cells (Figure 3(b) left panel). Chromatin immunoprecipitation (ChIP) analysis further confirmed that C/EBPβ bound to the TLR4 promoter, but this binding to the TLR4 promoter was almost completely eliminated by U0126 treatment in E6‐overexpressing A549 cells (Figure 3(b) right panel). EGFR mutant H1975 and H1650 cells responded in a similar fashion to the TL1 and E6‐overexpressing A549 cells when subjected to the same treatments (Figure 3(c)). These results suggest that C/EBPβ expression mediated by the E6 oncoprotein or by mutant EGFR via the ERK signaling pathway may increase TLR4 expression, and in turn, upregulate PD‐L1 transcription.

Expression of p‐ERK, C/EBPβ, p‐C/EBPβ (Thr 235), and PD‐L1 was lower in EGFR‐knockdown H1975 cells than in H1975 cells transfected with non‐specific shRNA (NC). Conversely, these expressions of all four genes were elevated in A549 and TL4 cells transfected with the L858R mutant EGFR expression vector (Figure 3(d)). These results clearly indicate that induction of PD‐L1 transcription by the E6 oncoprotein or by mutant EGFR in lung cancer cells may occur through the ERK‐C/EBPβ‐TLR4‐NF‐κB cascade.

PD‐L1 promotes cell invasiveness and soft agar growth by decreasing transforming growth factor‐β1 expression

BAG‐1 suppressed transforming growth factor (TGF)‐β1 expression in colorectal cancer cells, but BAG‐1 was upregulated by PD‐L1 in NSCLC cells. TGF‐β1 inhibited cell proliferation via upregulation of p21 expression. The TGF‐β1‐SMAD4 signaling pathway was involved in PD‐L1‐induced EGFR‐ tyrosine kinase inhibitors (TKI) resistance in NSCLC. Therefore, we hypothesized that TGF‐β1, p21, and SMAD4 could be involved in PD‐L1‐mediated cell invasiveness. High PD‐L1‐expressing TL1 and CL1‐5 cells were transfected with shPD‐L1 and/or co‐transfected with shTGF‐β1. Western blotting indicated that TGF‐β1, p21, and SMAD4 expressions were markedly increased, but VEGF‐C was decreased by PD‐L1‐knockdown in TL1 (Figure 4(a)) and CL1‐5 cells (Figure 4(b)). However, the increase in SMAD4 and p21 expression by PD‐L1 knockdown in both cell types was nearly completely and dose‐dependently eliminated by TGF‐β1 silencing (Figure 4(a),(b)). The abilities for invasion and soft agar growth were almost completely suppressed by PD‐L1 knockdown; however, the decrease in both abilities by PD‐L1 knockdown in TL1 and CL1‐5 cells was nearly completely reversed by TGF‐β1 silencing (Figure 4(c),(d)). These results clearly indicated that PD‐L1 promotes cell invasiveness and soft agar growth by decreasing TGF‐β1 expression.

FIGURE 4

PD‐L1 promotes cell invasiveness and soft agar growth by decreasing TGF‐β1 expression. (a) TL1 and (b) CL1‐5 cells were transfected with shPD‐L1 and/or shTGF‐β1 for 48 hours and then the expression of PD‐L1, TGF‐β1, SMAD4, VEGFC, and p21 was evaluated by Western blotting in both cell types. The representative soft agar growth on soft agar plates and invasiveness of cells in matrigel membranes in TL1 (C) and CL1‐5 (D) cells are shown as indicated

PD‐L1 upregulates BAG‐1 to reduce TGF‐β1 expression, and the decrease in SMAD4 because of TGF‐β1 occurs through the p53‐miR‐224 axis

We examined whether PD‐L1 could increase BAG‐1 expression and, thereby, downregulate TGF‐β1 expression. The decrease in SMAD4 expression by TGF‐β1 might occur through modulation of the p53‐ microRNA (miR)‐224 axis. Two low PD‐L1‐expressing TL4 and A549 cells were selected for overexpression of PD‐L1 and were co‐transfected with BAG‐1 shRNA (shBAG‐1). Western blotting indicated that BAG‐1 expression was markedly elevated by PD‐L1 overexpression in TL4 and A549 cells. BAG‐1 expression was dose dependently decreased by shBAG‐1 transfection in both PD‐L1‐overexpressing cell types (Figure 5(a) left panel). The expressions of TGF‐β1 and SMAD4 were almost completely suppressed by PD‐L1 overexpression; however, the expression of both molecules was rescued by BAG‐1 knockdown in PD‐L1‐overexpressing TL4 and A549 cells (Figure 5(a) left panel). The increase in TGF‐β1 expression by BAG‐1 knockdown in both cell types occurred at the transcriptional level (Figure 5(a) right panel). These results indicated that the decrease in TGF‐β1 by PD‐L1 may occur through increased BAG‐1 expression.

FIGURE 5

The decrease in TGF‐β1 and SMAD4 expression may be responsible for PD‐L1‐mediated cell invasion in lung cancer cells. (a) TL4 and A549 cells were transfected with PD‐L1 and/or shBAG‐1 for 48 hours and then the expression of PD‐L1, BAG‐1, TGF‐β1, and SMAD4 was evaluated by Western blotting. (b) H1975 and H1650 cells were transfected with TGF‐β1 expression plasmid and/or shp53 and then the expression of TGF‐β1, p53, and SMAD4 was evaluated by Western blotting. (c) H1355 and CL3 cells were transfected with shTGF‐β1 and/or p53 expression plasmid and then the expression of TGF‐β1, p53, and Smad4 was evaluated by Western blotting. The change in miR‐224 levels by ectopic TGF‐β1 expression and/or p53 silencing was evaluated by real‐time PCR. The change in invasion ability of H1975 and H1650 cells was determined by Boyden chamber assays. All experiments were performed independently and in triplicate

Low TGF‐β1‐expressing EGFR mutant H1975 and H1650 cells were selected to overexpress TGF‐β1 and were then co‐transfected with shp53. Western blotting showed that the expression of SMAD4 and p53 increased markedly in TGF‐β1‐overexpressing H1975 and H1650 cells (Figure 5(b) upper panel). The increase in SMAD4 because of TGF‐β1 overexpression was decreased by p53‐knockdown in TGF‐β1‐overexpressing H1975 and H1650 cells (Figure 5(b) upper panel). Moreover, miR‐224 expression levels were markedly decreased by TGF‐β1 overexpression, but miR‐224 levels were restored by p53‐knockdown in TGF‐β1‐overexpressing H1975 and H1650 cells (Figure 5(b) lower panel). Similar findings were observed in EGFR‐wild‐type H1355 and CL3 cells subjected to the same treatments (Figure 5(c)). These results indicated that a decrease in SMAD4 expression by TGF‐β1 may occur partially through the p53‐miR‐224 axis.

DISCUSSION

The clearance of viral infections depends on a functional T cell response. The activation and proliferation of T cells and the production of cytokines is now well documented to be negatively regulated by the PD‐1/PD‐L1 pathway, which then allows viral persistent infection. , Current evidence points to an important role for the PD‐1 pathway in inhibiting the function of virus‐specific CD8+ T cells in chronic viral infection involving several viruses, including human immunodeficiency virus, Epstein–Barr virus, and HPV. , , In the present study, we demonstrated that PD‐L1 expression was elevated in HPV16‐infected TL1 lung cancer cells (Figure 1(a)). Moreover, neutralization of endogenous PD‐L1 released from TL1 cells could restore PHA‐stimulated Th1 cytokine expression of peripheral blood mononuclear cells and cell‐mediated cytotoxicity (Figure S1). These results concur with previous reports suggesting that antibody blockade of PD‐L1‐mediated inhibition in vivo rapidly enhances T cell function and reduces viral replication.

The prognostic role of PD‐L1 in NSCLC remains equivocal. For example, PD‐L1 expression was associated with a favorable prognosis in pulmonary squamous cell carcinoma, but was associated with poor prognosis in NSCLC. , , , Consistently, high PD‐L1 mRNA levels were associated with poor outcomes, particularly in NSCLC patients with E6‐positive or EGFR‐mutated tumors (Table 2). In addition, high expression of PD‐L1 was associated with the presence of EGFR mutations in patients with NSCLC and was an independent negative prognostic factor for this disease. Disease‐free survival and disease‐specific survival were significantly poorer in patients with cervical squamous cell carcinoma and diffuse PD‐L1 expression when compared with patients with marginal PD‐L1 expression in their primary tumors. The present study from cell model experiments provides evidence to indicate that PD‐L1 may directly promote tumor cell growth and invasiveness, and thereby, confer poor prognosis in NSCLC, particularly in patients with E6‐positive and EGFR‐mutated tumors. The possible route for the mechanistic action for induction of PD‐L1 by E6 oncoprotein or mutant EGFR and promotion of tumor growth and invasion are presented in Figure 6.

FIGURE 6

The possible mechanistic action for induction of PD‐L1 by E6 oncoprotein or mutant EGFR and promotion of tumor growth and invasion in lung cancer. PD‐L1, programmed death ligand‐1

An association between HPV infection and EGFR mutation has been reported in Japanese and Taiwanese patients with NSCLC. , Moreover, the use of anti‐PD‐L1 monoclonal antibody plus an E6 vaccine has been shown to have substantial suppressive effects on tumor growth and invasion in immune deficiency nude mice that were treated with HPV16 E6‐positive TL1 cells. Therefore, we suggest that anti‐PD‐L1 immunotherapy may show great potential in improving clinical outcomes in patients with HPV‐infected and/or EGFR mutant NSCLC.

In summary, we have provided evidence that PD‐L1 expression may be induced by the E6 oncoprotein or EGFR mutation via the ERK‐C/EBPβ‐TLR4‐NF‐κB cascade and result in promotion of tumor growth, invasion, and poor outcomes in patients with NSCLC. Recently, MEK/ERK inhibitors have been selected for treatment of patients with NSCLC. , Therefore, we suggest that anti‐PD‐L1 immunotherapy combined with MEK/ERK inhibitors might have a clinical benefit in these patients, and especially those with HPV16/18 infection and/or EGFR mutation. PD‐L1 may represent a target for lung cancer immunotherapy and could serve as a potential biomarker to facilitate patient assignment to treatment, as well as aid in the determination of prognosis both before and after therapy.

CONFLICT OF INTEREST

The authors disclose no conflicts of interests.

Figure S1

Click here for additional data file.