If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
Corresponding author. Address for correspondence: Tatsuro Okamoto, MD, PhD, Department of Surgery and Science, Graduate School of Medical Sciences, Kyushu University 3-1-1, Maidashi, Higashi-ku, Fukuoka 812-8582, Japan.
The programmed death ligand 1(PD-L1)/programmed cell death protein 1 (PD-1) pathway is one of the most important checkpoint pathways for mediating tumor-induced immune suppression through T-cell exhaustion. Recently, targeted therapies using monoclonal antibodies against components of this pathway have been shown to reduce tumor burden in patients with non–small cell lung cancer (NSCLC). However, the prognostic significance of PD-L1 expression is controversial and the precise mechanisms of PD-L1 gene activation in lung cancer have yet to be clarified.
Methods
We investigated copy number alterations (CNAs) in the PD-L1 gene by real-time PCR in 94 surgically resected lung cancer samples to find possible associations between PD-L1 CNA and lung cancer biology. Janus kinase 2 gene (JAK2) CNA and its influence on the PD-L1/PD-1 pathway were also assessed.
Results
Five samples were shown to have PD-L1 gene amplification, whereas 89 samples did not. The patients with PD-L1 amplification had worse prognoses than did those without PD-L1 amplification. Genetic amplification of the PD-L1 gene was correlated with JAK2 gene amplification. The lung cancer cell line HCC4006 was found to harbor both JAK2 and PD-L1 amplification. Flow cytometry analyses revealed the level of PD-L1 protein expression to be higher in HCC4006 cells than in other NSCLC cell lines. Expression of the PD-L1 protein was significantly reduced by the JAK2 inhibitor TG-101348 and the signal transducer and activator of transcription 3 (STAT-3) inhibitor BP-1-102, but not by the STAT1 inhibitor fludarabine.
Conclusions
Our data suggest that expression of PD-L1 protein is upregulated by the simultaneous amplification of the PD-L1 and JAK2 genes through JAK-STAT signaling in NCSLC.
Lung cancer is one of the main causes of cancer death in Japan and in Western countries. The main reasons for the poor prognosis of patients with lung cancer are the biologically malignant potential of lung cancer and the insufficient efficacy of anticancer chemotherapy for lung cancer. Immunotherapy has been considered a promising modality for anticancer treatment and has been vigorously investigated over the past three decades. Because tumors usually express proteins with qualities and degrees of expression different from those of other cells in peripheral tissues, the host immune system is able to recognize the proteins and attack tumor cells to eliminate the unsuitable cells.
Despite the many credible reports of the antitumor function of the immune system in in vitro experiments and animal models, only small benefits in clinical settings have been reported. One exception is the recent success achieved through suppression of immune checkpoint pathways, such as the cytotoxic T-lymphocyte–associated protein 4 and programmed cell death protein 1 (PD-1) pathways.
The programmed death ligand 1 (PD-L1)–PD-1 axis is an inhibitory system that suppresses the effector functions of activated T cells, especially in the setting of chronic infection, and functions to inhibit autoimmune responses.
Tumor cells, which can express PD-L1, are believed to exploit this checkpoint mechanism to evade the anticancer immune response in peripheral tissues. When PD-1 expressed on T cells is ligated to PD-L1, the two molecules form a microcluster that recruits the SHP-2 phosphatase to the cell membrane through the immunoreceptor tyrosine-based switch motif.
SHP-2 phosphatase then dephosphorylates various signaling molecules, such as CD3ζ, which acts downstream of the T-cell receptor. As a result, cytokine production and proliferation of T cells is significantly inhibited, thus reducing the cytotoxicity of T cells toward cancer.
PD-L1 has been reported to be highly expressed in various cancers. High expression of PD-L1 has been found to be negatively correlated with patient prognosis in a variety of carcinomas.
Multiple conflicting results have been reported for the significance of PD-L1 in lung cancer, however, and the prognostic significance of PD-L1 expression in lung cancer is still debated.
High expression of PD-L1 in lung cancer may contribute to poor prognosis and tumor cells immune escape through suppressing tumor infiltrating dendritic cells maturation.
The PD-L1 gene is located in the 9p24.1 region; it is 322 kb downstream of JAK2, which is activated at the cell membrane in response to various cytokine signals and in turn activates signal transducer and activator of transcription (STAT) through phosphorylation.
Activated STAT then translocates to the nucleus, where it controls gene expression through binding to the promoter regions of certain genes. JAK-STAT signaling is considered important in PD-L1 upregulation in response to inflammatory cytokines, such as interferon gamma (IFNγ). Among the members of the STAT family, STAT1 and STAT3 have been reported to upregulate PD-L1 as a target gene.
Programmed cell death-ligand 1 expression in surgically resected stage I pulmonary adenocarcinoma and its correlation with driver mutations and clinical outcomes.
On the other hand, amplification of the 9p24.1 region, which contains PD-L1 and JAK2, was recently reported to be evident in a certain type of Hodgkin's lymphoma [HL] and primary mediastinal B-cell large cell lymphoma, and both genes were positively correlated with PD-L1 protein expression.
Integrative analysis reveals selective 9p24.1 amplification, increased PD-1 ligand expression, and further induction via JAK2 in nodular sclerosing Hodgkin lymphoma and primary mediastinal large B-cell lymphoma.
These results imply that enhancement of JAK-STAT signaling through amplification of the JAK2 gene, in addition to amplification of the PD-L1 gene itself, further enhances PD-L1 protein expression; however, there is no definitive molecular evidence supporting the significance of coamplification of the two genes.
PD-L1 expression and transport to the plasma membrane is controlled by several molecular events: the innate response, such as oncogenic signals, and the adaptive response, such as cytokine stimulation (innate immune resistance).
Therefore, we carefully examined the expression of PD-L1 in each type of cancer and each case. Here we report that the level of PD-L1 expression in non–small cell lung cancer (NSCLC) did not prove to have significant prognostic value among the cases we examined, as compatible with some previous reports.
High expression of PD-L1 in lung cancer may contribute to poor prognosis and tumor cells immune escape through suppressing tumor infiltrating dendritic cells maturation.
However, five cases had amplification of the 9p21 region, including both PD-L1 and JAK2, which was correlated with poor prognosis. This amplification was associated with a strong cellular response to tumor necrosis factor-α (TNF-α) and IFNγ stimulation in terms of surface expression and total expression of PD-L1 through JAK-STAT signaling.
Methods
Patients
We examined 94 cases of patients with NSCLC for whom surgical specimens and genomic DNA of both tumor and adjacent normal tissue were available. All the patients underwent surgery in our department at Kyushu University Hospital from October 2001 to March 2006. All tumor samples were diagnosed as NSCLC by pathologists. The histological classification of the tumor cell type was based on the World Health Organization histological classification of lung tumors. Our institutional review board approved this retrospective study (ID number: 26-34).
Immunohistochemistry
Tumor sections were assessed immunohistochemically using rabbit IgG polyclonal antibodies against PD-L1 (Lifespan Biosciences). Briefly, 4-μm sections were deparaffinized in xylene and dehydrated in an ethanol series. For antigen retrieval, slides were immersed in 10 mM sodium citrate (pH 6.0) and autoclaved (120°C for 15 minutes). The sections were incubated for 30 minutes in 0.3% hydrogen peroxidase in absolute methanol to deactivate endogenous peroxidases. After nonspecific binding of antibodies had been blocked, the specimens were incubated at room temperature with primary antibodies against PD-L1 for 60 minutes. Immunohistochemical (IHC) staining was performed using an EnVision system and DAB kits (Dako). Two investigators (including one general pathologist) who were blinded to any information about the samples evaluated the levels of expression of PD-L1. The expression of PD-L1 was quantified according to the intensity of the staining (0, negative; 1, very weak; 2, moderate; and 3, strong expression).
Prognostic impact of programmed cell death 1 ligand 1 expression in human leukocyte antigen class I–positive hepatocellular carcinoma after curative hepatectomy.
The level of PD-L1 expression seen in normal lung tissue was scored as 0 or 1. In the normal tissue samples, PD-L1 expression in alveolar type II epithelial cells and Clara cells was scored as 1. The low-expression group was defined as consisting of samples with scores of 0 and 1, and the high-expression group was defined as containing samples with scores of 2 and 3.
DNA preparation
DNA was extracted as follows: frozen samples were incubated in lysis buffer (0.01 mol/L Tris-HCl, pH 8.0; 0.1 mol/L ethylenediaminetetraacetic acid, pH 8.0; 0.5% sodium dodecyl sulfate) containing proteinase K (100 mg/mL) at 37°C for 2 hours. The samples were extracted twice in phenol, then once in phenol/chloroform, and finally once in chloroform. After ethanol precipitation, the samples were dissolved in tris-ethylenediaminetetraacetic acid (0.01 mol/L Tris-HCl, pH 8.0; 0.01 mol/L ethylenediaminetetraacetic acid, pH 8.0).
TaqMan copy number assay
The tumor tissue samples were analyzed for PD-L1 copy number, and four lung adenocarcinoma cell lines were analyzed for PD-L1 and JAK2 copy number. Copy number was determined using the TaqMan Copy Number Assay kits specific to PD-L1 and JAK2 (Hs03704252_cn and Hs05090456_cn, respectively, Life Technologies). In brief, polymerase chain reaction (PCR) was performed with the TaqMan Genotyping Master Mix (Life Technologies) on a StepOnePlus Real-Time PCR System (Life Technologies). For each single-well reaction containing 10 ng of genomic DNA and 10 μL of TaqMan Genotyping Master Mix, a 1-μL TaqMan Copy Number Assay, which included forward primer, reverse primer, and FAM dye-labeled minor groove binder probe specific for the gene of interest, was run simultaneously with a 1-μL TaqMan Copy Number Reference Assay, which included forward primer, reverse primer, and a VIC dye-labeled TAMRA probe specific for RNase P according to the manufacturer’s instructions. The TaqMan Copy Number Reference Assay was used as an in-well copy number control. Four replicate reactions were performed on each sample.
Copy numbers were calculated with CopyCaller v2.0 Software (Life Technologies) using the ΔΔCt relative quantification method. In brief, a maximum likelihood algorithm was used to estimate the mean ΔCt expected for a copy number of 1 on the basis of the probability density distribution across all samples, and this parameter was used in subsequent copy number calculations for each given gene. This analytical method was used to calculate the relative copy number of a target gene normalized to RNase P, a reference of known copy number 2, without the use of a calibrator sample for each target gene.
With respect to the tumor tissue samples, adjacent normal tissue samples (n = 94) were also run, and all the samples (n = 188) were analyzed using an algorithm in which the most frequent copy number for both PD-L1 and JAK2 was adjusted to 2.0. Cell lines were analyzed using an algorithm in which the copy numbers for both PD-L1 and JAK2 in normal human umbilical vein endothelial cells (HUVECs) was adjusted to 2.0. A PD-L1 gene copy number of 3 or greater was defined as amplification positive, whereas a PD-L1 gene copy number less than 3 was defined as amplification negative. To examine the relationship between PD-L1 and JAK2 copy number in the same tumor samples, samples with PD-L1 amplification (five samples, ≥3 copies) and samples without PD-L1 amplification (five samples, mean 2 ± 0.02 copies) were selected and analyzed for both PD-L1 and JAK2 copy number.
Flow cytometry
Flow cytometry was performed by resuspending 1 x 105 cells in 20 μL of staining buffer, which consisted of Hank's balanced salt solution supplemented with 0.1% sodium azide and 0.1% fetal bovine serum (Nichirei), incubating the cells with 0.8 μg of antibody for 30 minutes at 4°C, washing them once in staining buffer, and resuspending them in a volume of 200 μL. The cells were stained with phycoerythrin-conjugated antibodies specific for PD-L1 (clone 29E.2A3) or an isotype control antibody (immunoglobulin G2b; BioLegend). After staining, 10,000 cells were analyzed with a BD FACSCalibur flow cytometer (BD Biosciences). FlowJo software (TreeStar) was used for selection of viable cells by forward and side scatter and subsequent generation of histograms and median fluorescence intensities. Flow cytometry analysis, including analysis of intracellular protein expression, was performed by fixing and permeabilizing the cells with fixation/permeabilization solution followed by permeabilization solution (eBioscience) and then staining them with phycoerythrin-conjugated anti–PD-L1 antibodies.
Cell culture
Among the lung cancer cell lines used in the present study, PC-9 and 11-18 were the kind gift of Dr. M. Takeshita, and HCC4006 was the kind gift of Dr. A. F. Gazdar. The HUVECs were a kind gift from Dr. T. Saito. No further authentication of the cell lines was performed. Epidermal growth factor receptor gene (EGFR) mutations were detected in HCC827, 11–18 (exon 21, L858R) and HCC4006, PC-9 (exon 19, deletion). All lung cancer cell lines were cultured in RPMI 1640 (Invitrogen) with 10% fetal bovine serum and 100 mg/mL streptomycin (Invitrogen). HUVECs were cultured in EGM-2 endothelial cell growth medium-2 from the corresponding BulletKit (Lonza). In some experiments, lung cancer cell lines were treated for 48 hours with 100 U/mL IFNγ, 500 U/mL TNF-α (R&D Systems), or control solution. These experiments were performed independently three times. In another experiment, these cell lines were treated for 48 hours with 200 or 500 nM TG-101348 (Selleck) or 0.5 or 5 μM BP-1-102 (Calbiochem). Five (TG-101348 treatment) or four (BP-1-102 treatment) independent experiments were performed.
Western blotting
JAK2 expression in lung cancer cell lines was evaluated by Western blotting using an anti-JAK2 rabbit monoclonal antibody (Cell Signaling Technology). Glyceraldehyde-3-phosphate dehydrogenase expression was also evaluated as a loading control in Western blot analysis.
Statistical analysis
All analyses were performed using JMP software (SAS Institute). Overall survival (OS) and recurrence-free survival (RFS) rates were calculated by the Kaplan–Meier method, with between-group differences compared using the log-rank test. The relationship between PD-L1 expression and PD-L1 copy number in NSCLC tissue samples was evaluated by the likelihood ratio test. Mean fluorescence intensity analyses using flow cytometry were performed by Dunnett’s test. A p value less than 0.05 was considered statistically significant.
SNP-CGH array
Whole genome single-nucleotide polymorphism–comparative genomic hybridization (SNP-CGH) array assays were performed to confirm gene copy numbers at the 9p24.1 locus, including those of the PD-L1 and JAK2 genes, from six NSCLC specimens: one of six was noncancerous tissue, and of the remaining five, one was cancerous tissue that was amplification negative for the PD-L1 and JAK2 genes (as determined by the TaqMan copy number assay) and four were amplification positive by the same assay, as described elsewhere.
The tissues were genotyped by using 1,140,419 autosomal SNPs (HumanOmni1-Quad BeadChip; Illumina Inc.) and copy number variation was analyzed with GenomeStudio V2009.1 (Illumina Inc.).
Results
Relationship between PD-L1 protein expression and prognostic significance in NSCLC
First, we assessed protein expression in surgical tumor samples through immunohistochemistry to investigate the prognostic significance of PD-L1 protein expression in patients with NSCLC (Figs. 1A and 1B). PD-L1 protein was shown to localize to both cell membranes and the cytosol in lung cancer samples, which is similar to the PD-L1 localization reported elsewhere for hepatic cell carcinoma.
Prognostic impact of programmed cell death 1 ligand 1 expression in human leukocyte antigen class I–positive hepatocellular carcinoma after curative hepatectomy.
In addition, PD-L1 was highly expressed in areas where lymphocytic infiltration was also observed; PD-L1 was also highly expressed in the infiltrating lymphocytes. We scored PD-L1 protein expression level on the basis of immunostaining signal intensity. We divided patients into two groups—a low-expression group (n = 18) with scores of 0 or 1 and a high-expression group (n = 76) with scores of 2 or 3—and compared outcomes and various clinicopathological factors between the groups. The high-expression group consisted of 56 patients with adenocarcinoma, 14 with squamous cell carcinomas, four with large cell carcinomas, one with adenosquamous cell carcinoma, and one with giant cell carcinoma. No significant differences were observed for the clinicopathological factors except for differentiation (Supplemental Table 1), as well as for OS and RFS between the low–PD-L1 and high–PD-L1 expression groups (Figs. 1B and 1C).
Figure 1Survival outcomes in patients with high and low PD-L1 expression. (A) Immunohistochemical staining for PD-L1 in sections of non–small cell lung cancer (NSCLC) tissue samples embedded in paraffin (original magnification ×400). (B) Hematoxylin and eosin staining of parallel sections. Arrows in both panels indicate NSCLC tissue. Recurrence-free survival (C) and overall survival (D) rates in patients with NSCLC and high (blue lines) and low (red lines) PD-L1 expression. Numbers of patients and p values are indicated.
To investigate PD-L1 gene amplification in lung carcinoma cells, we assessed the copy number of the PD-L1 gene using genomic DNA from the surgical samples through a real-time PCR assay (Fig. 2A). We divided the patients into two groups—the amplification group (n = 5) and the nonamplification group (n = 89)—on the basis of a PD-L1 gene copy number threshold of 3. The amplification group consisted of two patients with adenocarcinoma, one with squamous cell carcinoma, one with large cell carcinoma, and one with giant cell carcinoma. The PD-L1 amplification group was found to have poorer outcomes than did the nonamplification group in terms of both OS and RFS (Figs. 2B and 2C).
Figure 2Survival outcomes in patients with and without PD-L1 amplification. (A) Quantitative polymerase chain reaction–based DNA copy number analysis of PD-L1 in 94 samples of non–small cell lung cancer. Black box indicates samples with gene amplification. Recurrence-free survival (B) and overall survival (C) rates in patients with non–small cell cancer with (blue lines) and without (red lines) PD-L1 amplification. Numbers of patients and p values are indicated.
We then assessed JAK2 gene copy number in both the PD-L1 amplification group and the five members of the nonamplification group whose average PD-L1 copy number was 2 ± 0.02. As a result, JAK2 and PD-L1 copy numbers were shown to have a strong positive correlation (R
= 0.9615) (Fig. 3A). We used a SNP-CGH array to confirm copy number gain in NSCLC samples that were shown to have both PD-L1 and JAK2 amplification by PCR-based gene copy number assay. The SNP-CGH assay showed that all four samples harbored 9p24.1 locus amplification (one of the four samples is shown in Fig. 3B). These results suggest that a chromosomal region including the 9p24.1 locus that harbors PD-L1 and JAK2 is amplified in lung carcinoma cells in some cases.
Figure 3Coamplification of PD-L1 and JAK2 in non–small cell lung cancer (NSCLC). (A) Relationship between PD-L1 and JAK2 copy number in five samples with PD-L1 amplification and five samples without PD-L1 amplification. (B) Single-nucleotide polymorphism–comparative genomic hybridization analysis of DNA from a primary NSCLC tumor without PD-L1 amplification (upper panel) and from a primary NSCLC tumor with PD-L1 amplification (lower panel). Quantitative polymerase chain reaction–based DNA copy number analysis of PD-L1 (C) and JAK2 (D) in lung adenocarcinoma cell lines. Human umbilical vein endothelial cells were used as a copy number control (two copies). Results are representative of three independent experiments.
In our study, no correlation was observed between PD-L1 protein expression assessed by IHC staining and PD-L1 gene amplification in NSCLC specimens (Supplemental Table 2); four of the five samples that showed PD-L1 gene amplification did not have PD-L1 protein overexpression.
PD-L1 gene amplification in NSCLC cell lines
Next, we conducted a search for NSCLC cell lines harboring PD-L1 gene amplification using the same method. The HCC4006 cell line was found to have a significant amplification of the PD-L1 gene (copy number 3.74) (Fig. 3C). In addition, the JAK2 gene was found to be amplified (copy number 3.79 computational) (Fig. 3D). Thus, the 9p24.1 region, which contains both JAK2 and PD-L1, is amplified in HCC406 cells.
Relationship between PD-L1 gene amplification and protein expression in NSCLC cell lines
We then examined JAK2 protein expression levels in HCC4006, 11-18, PC-9, and HCC827 cells (lung adenocarcinoma cell lines with EGFR mutations) by Western blotting. We found that JAK2 is more highly expressed in HCC4006 cells than in the other cell lines (Fig. 4A). Levels of expression of PD-L1 protein in these cell lines were also examined by flow cytometry. In these experiments, we analyzed surface expression of PD-L1 using standard unfixed cells and total expression of PD-L1 in fixed and permeabilized cells. In addition, PD-L1 expression was examined in (1) untreated control cells, (2) TNF-α–treated cells, and (3) IFNγ-treated cells because both TNF-α and IFNγ have the potential to upregulate the expression of PD-L1.
Surface expression of PD-L1 was found to be significantly higher in HCC4006 cells than in the other cell lines (11-18, PC-9, and HCC827, p < 0.05), and this expression was enhanced by the addition of TNF-α or IFNγ. Surface expression of PD-L1 was also significantly higher in TNF-α– or IFNγ-treated cells (Fig. 4B). Levels of total expression of PD-L1 were also significantly higher in HCC4006 cells than in the other cells, except for HCC827 (Fig. 4C). In addition, levels of total expression of PD-L1 were also significantly higher in TNF-α– and IFNγ-treated cells. These results suggested that amplification of the 9p24.1 region might play an important role in the high level of expression of PD-L1 both in the steady state and in the adaptive state stimulated by cytokines.
Figure 4JAK2 and PD-L1 expression in non–small cell lung cancer cell lines harboring EGFR gene mutations. (A) Western blot analysis of JAK2 in HCC4006, 11-18, PC-9, and HCC827 cells. Glyceraldehyde-3-phosphate dehydrogenase was analyzed as a loading control. Lung adenocarcinoma cell lines were treated for 48 hours with bovine serum albumin (BSA), tumor necrosis factor-α (500 U/mL), or interferon gamma (100 U/mL). Surface (B) and whole (C) expression of PD-L1 were evaluated by flow cytometry. Results are representative of three independent experiments. Asterisks indicate statistically significant differences between HCC4006 and other cell lines (*p < 0.05 versus BSA-treated condition, **p < 0.05 versus tumor necrosis factor-α–treated condition, ***p < 0.05 versus interferon gamma–treated condition). Daggers indicate statistically significant differences between the experimental and BSA-treated cells. (†p < 0.05 versus BSA-treated condition in each cell line; n.s. indicates not significant).
Involvement of JAK2-STAT signals in expression of PD-L1 in HCC4006 cells
We examined whether the JAK2-STAT signals were involved in the expression of PD-L1 in HC4006 cells with JAK2 and PD-L1 gene amplification. In this study, of the STATs functioning downstream of JAK2, we focused on STAT3, which has tumor-promoting effects when activated. We first examined the effects of JAK2 activation on the expression of PD-L1 in HCC4006 cells using the JAK2-specific inhibitor TG-101348.
Upon exposure to TG-101348, the level of total expression of PD-L1 was downregulated in a dose-dependent manner, with a significant difference at a concentration of 500 nM (Fig. 5A). The level of surface expression of PD-L1 was also significantly suppressed, even at an inhibitor concentration of 200 nM (Fig. 5B). Next, we examined the effects of STAT3 activation on the expression of PD-L1 in HCC4006 cells using the STAT3-specific inhibitor BP-1-102.
The addition of BP-1-102 significantly reduced the total amount of PD-L1 in cells but did not reduce the levels of PD-L1 at the cell membrane (Fig. 5C and 5D). In contrast, the STAT1-specific inhibitor fludarabine did not affect either total or cell surface expression of PD-L1 in HCC4006 cells (data not shown). These findings suggest that JAK2 signaling contributes to the enhanced expression of PD-L1 in lung cancer cell lines and that a downstream STAT3 signal is partially involved in determining the level of total expression of PD-L1.
Figure 5Effects of JAK2, STAT3, and STAT1 inhibitor on surface and total expression of PD-L1 in the lung adenocarcinoma cell line HCC4006. JAK2 inhibition suppresses surface and whole expression of PD-L1 in HCC4006 cells. HCC4006 cells were treated for 48 hours with the JAK2 pharmacological inhibitor TG101348 (200 nM or 500 nM) or DMSO. Surface (A) and total (B) expression of PD-L1 were evaluated by flow cytometry. Results are representative of five independent experiments. STAT3 inhibition suppresses total expression of PD-L1 in HCC4006 cells but has no effect on surface expression of PD-L1. HCC4006 cells were treated for 48 hours with the STAT3 pharmacological inhibitor BP-1-102 (0.5 μM or 5 μM) or DMSO. Surface (C) and total (D) expression of PD-L1 were evaluated by flow cytometry. Results are representative of four independent experiments. Asterisks indicate statistically significant differences between the experimental and DMSO-treated cells (*p < 0.05, **p < 0.01, ***p < 0.001).
Many recent studies investigating PD-L1 expression in various type of cancer have demonstrated that PD-L1 expression is correlated with disease outcome and has prognostic value.
With regard to lung cancer, some reports have also demonstrated that high levels of PD-L1 expression are associated with poor prognoses. Mu et al. investigated PD-L1 protein expression in 109 patients with NSCLC and demonstrated that high expression of PD-L1 protein is correlated with poor OS after surgery.
High expression of PD-L1 in lung cancer may contribute to poor prognosis and tumor cells immune escape through suppressing tumor infiltrating dendritic cells maturation.
Similarly, Azuma et al. reported that high expression of PD-L1 protein was associated with poor OS in their 164-patient cohort when the same rabbit polyclonal antibody as in the present study was used.
In contrast to the aforementioned reports, our study found no significant correlation between level of PD-L1 protein expression and outcome among patients with lung cancer, although a trend was seen in the Kaplan-Meier survival curves. One possible explanation for this discrepancy was that our data set was small, as a result of which statistical significance was not achieved. However, some other reports have shown the opposite result for prognosis among patients with NSCLC. Yang et al.
Programmed cell death-ligand 1 expression in surgically resected stage I pulmonary adenocarcinoma and its correlation with driver mutations and clinical outcomes.
demonstrated that high expression of PD-L1 was associated with better RFS among patients with stage I adenocarcinoma. They used a rabbit monoclonal antibody and considered only membrane staining to be expression positive. Velcheti et al.
investigated the relationship between PD-L1 expression and patient survival using a mouse monoclonal antibody in two relatively large cohorts. They found that high expression of PD-L1 was significantly associated with better OS in both cohorts. The relationship between level of PD-L1 protein expression and outcomes of patients with lung cancer remains to be elucidated.
We found that PD-L1 copy number was increased in some cases of NSCLC and that an increase in the copy number of the PD-L1 gene was accompanied by an increase in that of the JAK2 gene, which is located close to the PD-L1 gene in the 9p24.1 region. We confirmed amplification of the 9p24.1 region by SNP-CGH array. To our knowledge, ours is the first study to demonstrate simultaneous increases in the copy number of both genes in NSCLC. Green et al.
Integrative analysis reveals selective 9p24.1 amplification, increased PD-1 ligand expression, and further induction via JAK2 in nodular sclerosing Hodgkin lymphoma and primary mediastinal large B-cell lymphoma.
found amplification of 9p24.1 in nodular sclerosing HL and primary mediastinal large B-cell lymphoma (MLBCL). They also discovered that JAK2 inhibition decreased PD-L1 gene expression in HL cell lines. More recently, a study from the same group demonstrated that the JAK2 inhibitor fedratinib suppressed cellular proliferation in classical HL and MLBCL cell lines. This effect was accompanied by suppression of JAK2/STAT signaling and reduction of PD-L1 expression in a copy number–dependent manner. They also observed that fedratinib decreased tumor growth in murine xenograft models.
Integrative analysis reveals selective 9p24.1 amplification, increased PD-1 ligand expression, and further induction via JAK2 in nodular sclerosing Hodgkin lymphoma and primary mediastinal large B-cell lymphoma.
PD-L1 amplification was observed in 63% of primary MLBCLs (26 of 41). They considered tumors amplification positive when the PD-L1 DNA copy number was more than 2.2. In the present study, we set the cutoff line at 3.0 copies, and on the basis of this cutoff, we found only five NSCLC tumors that we considered to have PD-L1 gene amplification. Patients who had PD-L1 amplification had a poor RFS and a poor OS. Carcinoma cells harboring amplification of the 9p24.1 region may express high levels of PD-L1 more easily than does the nonamplified group when the local concentration of inflammatory cytokines or immune cell infiltration is increased, which might contribute to the development of cancer and, thus, might be correlated with increased risk for poor outcome.
We could not find any correlation between level of expression of PD-L1 protein, as assessed by IHC staining, and gene amplification in the present study (Supplemental Table 2). The discrepancy between our finding and those of others may be due to the fact that levels of the protein PD-L1 in tumor cells are affected by various factors, especially by immunoreactive circumstances (adaptive immune resistance) and oncogenic signaling molecules (innate immune resistance).
The heterogeneity of expression of PD-L1 protein within individual patients and even within individual tumors may also contribute to the discrepancy. On the basis of our in vitro assays, induction of PD-L1 protein by inflammatory cytokines was more efficient in cancer cells harboring PD-L1 amplification (Fig. 4B and 4C); thus, gene amplification may more accurately predict malignant behavior in tumor cells than does protein expression detected by IHC staining.
In the HCC4006 lung adenocarcinoma cell line, which harbors amplification of the 9p24.1 region, expression of PD-L1 was increased relative to the levels in lung adenocarcinoma cell lines without this amplification, whereas levels of total expression of PD-L1 were not significantly higher in HCC4006 cells than in HCC827 cells. This fact may due to the p53 mutation in HCC827 cells because deletion or silencing of PTEN, which is positively regulated by p53, enhances the level of expression of PD-L1 protein in glioblastomas.
In this study, we examined surface and total expression of PD-L1 separately; the former reflects membrane trafficking signals and the latter reflects protein synthesis. Moreover, surface and total expression of PD-L1 in HCC4006 cells were significantly reduced by treatment with TG-101348, an inhibitor of JAK2 (Fig. 5A and 5B). These results suggest that JAK2 activity not only leads to the transcriptional activation of PD-L1 through the classical JAK-STAT pathway but also controls the transport of PD-L1. Total expression of PD-L1 in whole cells was significantly inhibited by the STAT3 inhibitor BP-1-102; however, cell surface expression was not affected. These findings suggest that STAT3 activation may be involved in controlling total expression of PD-L1 as a downstream effector of JAK2 activation and that other effectors besides STAT3 may play a role in controlling PD-L1 protein expression at the cell surface, which may include control of PD-L1 transport to the cell surface. Involvement of STAT1 was ruled out because there was no effect of the STAT1 inhibitor fludarabine on PD-L1 expression, although this result should be interpreted carefully because fludarabine can also cause DNA damage.
Limitations of the present study include the small sample size; the size difference between the two groups, which led to difficulties in making statistical comparisons for OS and RFS; the fact that the definition of gene amplification has not yet been validated in other cohorts; and the fact that only one cell line was found to have PD-L1 gene amplification for in vitro inhibition assay. Further studies are warranted to confirm our results and to fully determine which STAT molecules are involved in controlling PD-L1 expression.
Conclusions
In conclusion, the present data demonstrate that there is a certain population of patients with primary NSCLC who harbor simultaneous amplification of the PD-L1 and JAK2 genes and that PD-L1 protein expression is synergistically upregulated by PD-L1 gene amplification and JAK2/STAT signaling.
Acknowledgments
We thank Mototsugu Shimokawa for a statistical review.
High expression of PD-L1 in lung cancer may contribute to poor prognosis and tumor cells immune escape through suppressing tumor infiltrating dendritic cells maturation.
Programmed cell death-ligand 1 expression in surgically resected stage I pulmonary adenocarcinoma and its correlation with driver mutations and clinical outcomes.
Integrative analysis reveals selective 9p24.1 amplification, increased PD-1 ligand expression, and further induction via JAK2 in nodular sclerosing Hodgkin lymphoma and primary mediastinal large B-cell lymphoma.
Prognostic impact of programmed cell death 1 ligand 1 expression in human leukocyte antigen class I–positive hepatocellular carcinoma after curative hepatectomy.
Disclosure: Dr. Okamoto received financial support through a grant-in-aid from the Japan Society for the Promotion of Science (24592095). The remaining authors declare no conflict of interest.
00019-2&id=gr1.jpg){kind=link}
00019-2&id=gr2.jpg){kind=link}
00019-2&id=gr3.jpg){kind=link}
00019-2&id=gr4.jpg){kind=link}
00019-2&id=gr5.jpg){kind=link}