The acidic tumor microenvironment enhances PD-L1 expression via activation of STAT3 in MDA-MB-231 breast cancer cells

Sodium bicarbonate (NaHCO₃), lactic acid, sodium lactate and sodium oxamate were obtained from Sigma Aldrich (St. Louis, MO, USA). The antibodies used in this study were as follows: anti-PD-L1 (Abcam, Cambridge, UK); anti-LDH-A (Santa Cruz Biotechnology, Santa Cruz, CA, USA); anti-pY⁷⁰¹-STAT1, anti-STAT1, anti-pY⁷⁰⁵-STAT3, anti-pS⁷²⁷-STAT3, anti-STAT3 and anti-β-actin (Cell Signaling Technology, Danvers, MA, USA); and anti-HIF-1α (Abbkine, Wuhan, China).

The human breast cancer cell line MDA-MB-231 was obtained from the American Type Culture Collection (Manassas, VA, USA), and Dulbecco’s modified Eagle’s medium (DMEM), fetal bovine serum (FBS), and penicillin/streptomycin were obtained from Capricorn Scientific (Ebsdorfergrund, Germany). MDA-MB-231 cells were maintained in DMEM with 10% FBS and 1% penicillin/streptomycin and incubated in a humidified incubator (Vision Science, Seoul, Republic of Korea) containing 5% CO₂ at 37 °C.

To mimic extracellular acidosis, 1 M HCl and lactic acid were used as described previously [18]. In addition, 1 M NaOH and NaHCO₃ were used to neutralize or buffer acidic pH. The pH-adjusted culture medium was stabilized in a 5% CO₂ incubator prior to treatment. The pH of cell culture media was measured using a pH meter (Mettler Toledo, Columbus, OH, USA).

MDA-MB-231 cells were cultured in normoxia (21% O₂ and 5% CO₂) or hypoxia (2% O₂ and 5% CO₂) with/without oxamate for 36 h, and normoxic conditioned medium (NCM) and hypoxic conditioned medium (HCM) were collected. Heat-inactivated HCM was obtained by boiling HCM for 20 min at 100 °C to degrade secretory proteins in the conditioned medium, and the pH neutralized HCM was obtatined by adding 1 M NaOH to neutralize acidified HCM.

MDA-MB-231 cells were lysed using 1% Triton X-100 lysis buffer containing protease and phosphatase inhibitors (2 mM PMSF, 1 mM NaF, 2 mM EDTA, 0.5 mM Na₃VO₄ and 10 μg/mL leupeptin). Cell lysates were separated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis and transferred onto a nitrocellulose membrane (GE Healthcare Life Sciences, Chicago, IL, USA). The membranes were blocked with 5% skim-milk (LPS Solution, Daejeon, Repulic of Korea) for 1 h. The blocked membranes were incubated with primary antibodies (1:1000–2000) at 4 °C overnight followed by horseradish peroxidase-conjugated secondary antibodies (1:5000–10,000) at room temperature for 2 h. The membranes were incubated with ECL reagents (BioMax, Seoul, Republic of Korea) and developed on AGFA CP-BU New X-ray film (Agfa-Gevaert N.V., Mortsel, Belgium). Densitometric and statistical analysis of western blot bands are included in the Supplementary Fig. 5 and 6, and all western blot band images are included in the Supplementary Fig. 7–13.

Total RNA was extracted using the RNAiso Plus (total RNA extraction reagent; Takara, Shiga, Japan) and cDNA was synthesized using the ReverTra Ace qPCR RT Master Mix (TOYOBO, Osaka, Japan) according to the manufacturer’s protocols. To compare the mRNA expression of target genes, qPCR was performed using the BlasTaq qPCR MasterMix (Applied Biological Materials, Richmond, Canada), and the fluorescent signal of individual target genes was detected using a CFX Connect Real-Time PCR Detection System (Bio-Rad Laboratories, Hercules, CA, USA). The qPCR primers used in this study were as follows: PD-L1 forward, 5′- AAATGGAACCTGGCGAAAGC − 3′, and PD-L1 reverse, 5′- GATGAGCCCCTCAGGCATTT -3′; PD-L2 forward, 5′- GTCTTGGGAGCCAGGGTGAC − 3′, and PD-L2 reverse, 5′- TGAAAAGTGCAAATGGCAAGC -3′; GAPDH forward, 5′- CTGACTTCAACAGCGACACC − 3′, and GAPDH reverse, 5′- TAGCCAAATTCGTTGTCATACC − 3′.

STAT3 was overexpressed with pcDNA3.1-GFP and pcDNA3.1-wild type STAT3-GFP vectors using Lipofectamine 3000 (Invitrogen, Waltham, MA, USA) for 48 h. STAT3 was knocked down with human si-STAT3 (SI02662338; Qiagen, Hilden, Germany) and a negative control si-RNA (1,027,280; Qiagen) using RNAiMax (Invitrogen) for 48 h. Transfection experiments were performed according to the manufacturer’s protocol (https://​www.​thermofisher.​com).

Human breast tissues, provided by Seoul National University Hospital with approval from the SNUH Institutional Review Board (IRB; IRB No. 1904–141-1029), were fixed in 4% paraformaldehyde for 24 h at 4 °C and embedded in paraffin. The paraffin blocks were cut into 4-μm-thick sections, and the paraffin slides were deparaffinized with xylene and hydrated with graded ethanol. Antigen was unmasked by boiling the sections in 100 mM citrate buffer (pH 6.0) for 10 min and endogenous peroxidase activity was blocked with 3% hydrogen peroxide for 10 min. The slides were blocked in 5% normal goat serum (Vector Laboratories, Burlingame, CA, USA) for 1 h and incubated with primary antibodies (diluted 1:100–200 in antibody diluent solution; Life Technologies, Carlsbad, CA, USA) overnight at 4 °C. The following day, sections were incubated with biotin-conjugated secondary antibodies (diluted 1:100–200 in antibody diluent solution; Vector Laboratories) for 1 h and incubated with avidin-biotin complex reagent for 1 h. The slides were visualized using DAB staining solution (Dako, Santa Clara, CA, USA) followed by counterstaining with hematoxylin solution (Dako). Images of stained sections were obtained using the LAS microscope (Leica Microsystem, Wetzlar, Germany).

The public datasets used in this study were GSE12276, GSE27830, GSE5460, GSE54002, and GSE36295 (https://​www.​ncbi.​nlm.​nih.​gov/​geo/​). Gene set enrichment analysis (GSEA) was perform to analyze enrichment plots of tumor hallmark gene sets of breast cancer patients (https://​www.​gsea-msigdb.​org) by dividing the PD-L1 high and low expression groups in the GSE12276, GSE27830 and GSE5460 data sets. PD-L1 and LDH-A mRNA expression in human normal breast and breast cancer tissues were comparatively analyzed using the GSE54002 and GSE36295 data sets.

Statistical analysis of all data was performed using Microsoft Excel 2016 software (Microsoft Corp., Redmond, WA, USA) and GraphPad Prism 5 (GraphPad Software Inc., La Jolla, CA, USA). Results are presented as the mean ± standard deviation of at least three independent experiments. Unpaired Student’s t-test was used for analyzing the data, p-values of < 0.05, < 0.01, and < 0.001 were considered statistically significant.

To confirm the relationship between tumor acidosis and PD-L1 expression in breast cancer cells, we adjusted the pH of the medium using HCl or NaOH (Supplementary Fig. 1A). Protein and mRNA levels of PD-L1 were significantly increased at the acidic pH (Fig. 1A–C). When PD-L2 expression by tumor acidosis was confirmed, PD-L1 expression increased approximately 3–4-fold, while PD-L2 increased up to 1.5-fold; PD-L1 expression was higher than that of PD-L2 (Supplementary Fig. 1B). A time course analysis revealed that the protein level of PD-L1 was significantly increased after 12 h by acidic pH, and the mRNA level was significantly increased after 6 h. These results indicate that the protein and mRNA levels of PD-L1 are significantly increased by acidic pH in MDA-MB-231 cells.

Extracellular acidosis occurs in a hypoxic environment and/or via the Warburg effect in cancer [1, 2]. In these tumor microenvironments, cancer cells excessively convert pyruvate to lactic acid and ATP by overexpressing LDH-A to produce energy [19]. As a result, a large amount of lactic acid is released from the cells and the pH around them is lowered (Fig. 2A). We predicted that expression of PD-L1 would be increased in cancer cells as a result of lactic acid-induced acidosis. To mimic the lactic acid acidification caused by the Warburg effect and hypoxic environment in cancer, NCM obtained by the Warburg effect under normoxic conditions and HCM obtained by the Warburg effect under hypoxic conditions were prepared. Newly seeded cells were treated with these conditioned media (Fig. 2B). Under both the mildly and severely acidic conditions (NCM and HCM, respectively), protein and mRNA levels of PD-L1 were significantly increased by lactic acidosis, but neutralization of HCM by NaOH did not increase PD-L1 expression (Fig. 2C–E). We also confirmed that PD-L1 expression was increased following treatment with heat-inactivated HCM, suggesting that hypoxia-induced cytokines and/or chemokines did not affect the expression of PD-L1 in MDA-MB-231 cells (Supplementary Fig. 2A–C). As we expected PD-L1 expression to be increased by the Warburg effect and hypoxia-induced lactic acidosis, we directly treated MDA-MB-231 cells with lactic acid. Direct treatment with lactic acid resulted in a decrease in the pH of the medium, and increased protein and mRNA levels of PD-L1. However, sodium lactate, which did not regulate the pH of the medium, did not affect PD-L1 expression (Fig. 2F–H and Supplementary Fig. 2D, E). These results suggest that tumor-derived lactic acidosis is a major cause of the increased expression of PD-L1 in MDA-MB-231 cells.

When PD-L1 of cancer cells binds PD-1 of immune cells, immune cells become exhausted and cancer cells exhibit immune escape [6]. Therefore, targeting overexpressed PD-L1 in cancer cells may be a viable therapeutic strategy for cancer treatment. Since the expression of PD-L1 is increased by acidic pH, we wondered whether PD-L1 expression would be lowered if the acidic pH was restored to normal levels. The expression of PD-L1 was lowered when the acidic medium was replaced with fresh medium or the acidic medium was buffered with NaHCO₃ (Fig. 3A–C). The lactic acid-induced increase in PD-L1 expression was also reduced when the acidic medium was replaced with fresh medium or acidic medium buffered with NaHCO₃ (Fig. 3D–F). Based on these results, we suggest that the expression of acidosis-induced PD-L1 is attenuated when the acidic pH is adjusted to a normal pH in MDA-MB-231 cells.

We decided to further investigate the tumor acidosis-induced increase in PD-L1 expression. GSEA of public breast cancer patient databases confirmed that IL-6/JAK/STAT3 signaling was significantly correlated with PD-L1 expression (Fig. 4A–C). Previous studies have reported that activation of STAT3 is related to PD-L1 expression in various cancers, including breast cancer [11, 20]. Interestingly, when MDA-MB-231 cells were treated with pH-dependent medium in a time-dependent manner, pS-STAT3 was unchanged while pY-STAT3 was significantly increased under acidic conditions (Fig. 4D, E). NCM, HCM, and lactic acid also increased pY-STAT3 levels in a similar manner as PD-L1, but not when the acidic pH was neutralized with NaOH or sodium lactate (Supplementary Fig. 3A, B). The increased expression of pY-STAT3 induced by acidosis or lactic acidosis was attenuated by replacing or buffering the acidic pH, as with PD-L1 (Fig. 4F, G). In addition, when oxamate, an LDH inhibitor, was treated under hypoxic conditions, hypoxia-induced lactic acidosis was suppressed; oxamate pre-treated HCM did not increase pY-STAT3 and PD-L1 expression (Supplementary Fig. 3C-F). Since PD-L1 is a target gene of STAT3, we expected that inhibition of STAT3 in tumor acidosis would inhibit the expression of PD-L1. As expected, levels of PD-L1 and pY-STAT3 were not increased under acidic conditions when STAT3 was silenced by siRNA, or inhibited with the STAT3 inhibitor nifuroxazide (Fig. 4H and Supplementary Fig. 3G), and PD-L1 expression was increased when STAT3 was overexpressed (Fig. 4I). These results suggest that STAT3 activation by tumor acidosis is a major factor underlying increased PD-L1 expression.

To demonstrate the clinical relevance of PD-L1 expression and tumor acidosis, we determined the correlations among PD-L1, STAT3, and acidic microenvironment markers by analyzing breast cancer tissue. The expression of LDH-A, an enzyme important in the production of lactic acid, and hypoxia-inducible factor-1α (HIF-1α), a major marker of hypoxia, was increased in breast cancer cells compared with normal breast cells; the expression of PD-L1 and pY-STAT3 was also increased in breast cancer cells (Fig. 5A). Using the GEO databases, we confirmed that the expression of PD-L1 and LDH-A was higher in cancer cells than normal cells (Fig. 5B, C). These results suggest that tumor acidosis is positively correlated with PD-L1 expression.