Metabolic inhibitor screening identifies dihydrofolate reductase as an inducer of the tumor immune escape mediator CD24

Abstract

Immune checkpoint inhibitors have improved the clinical management of some cancer cases, yet patients still fail to respond to immunotherapy. Dysregulated metabolism is a common feature of many cancers, and metabolites are known to modulate functions in cancer cells. To identify potential metabolic pathways involved in anti-tumor immune response, we employed a metabolic inhibitor-based drug screen in human lung cancer cell lines and examined expression changes in a panel of immune regulator genes. Notably, pharmacologic inhibition of dihydrofolate reductase (DHFR) downregulated cancer cell expression of cluster of differentiation 24 (CD24), an anti-phagocytic surface protein. Genetic modulation of DHFR resulted in decrease of CD24 expression, whereas tetrahydrofolate, the product of DHFR, enhanced CD24 expression. DHFR inhibition and the consequent CD24 decrease enhanced T cell-mediated tumor cell killing, whereas replenishment of DHFR or CD24 partially mitigated the immune-mediated tumor cell killing that resulted from methotrexate treatment in cancer cells. Moreover, publicly available clinical data analyses further revealed the link between DHFR, CD24, and the antitumor immune response in lung cancer patients. Our study highlights a novel connection between folate metabolism and the anti-tumor immune response and partially interprets how DHFR inhibitors lead to clinical benefits when combined with cancer immunotherapy agents.

INTRODUCTION

The advent of immunotherapy has revolutionized the treatment of cancer, and immune checkpoint inhibitors (ICIs) are now used to treat a variety of cancer types (1). However, clinical response rates are below 30% in most solid tumors and many cancer patients are unresponsive to ICI treatments (2). For instance, non-small cell lung carcinoma (NSCLC) patients with KRas/LKB1 co-mutations are known to receive little therapeutic benefit from ICIs (3). Immune checkpoint molecules such as PD-1 and CTLA-4 are key targets in cancer immunotherapy (4). Emerging evidence suggest that cluster of differentiation 24 (CD24), an anti-phagocytic surface protein that signals through macrophage Siglec-10, is a promising target for cancer immunotherapy (5). However, the regulation mechanism of CD24 in human cancers is not fully understood.

Dysregulated cellular metabolism such as the Warburg Effect, is a hallmark of many cancers. In addition to providing materials for bioenergetics and biosynthesis, metabolites also serve as important co-factors to regulate cell signaling and gene expression (6). DHFR is an essential enzyme in de novo pyrimidine synthesis that catalyzes the conversion of dihydrofolate to tetrahydrofolate (THF). DHFR is commonly overexpressed in various cancer types to facilitate DNA replication and tumor proliferation (7). Pharmacological inhibitors of DHFR such as methotrexate and pemetrexed have been used in the clinic for many years to treat various cancers including NSCLC. Methotrexate is FDA-approved for use in acute lymphoblastic leukemia, breast cancer, osteosarcoma, head and neck cancer, advanced non-Hodgkin lymphoma, and lung cancer, while pemetrexed is approved for advanced NSCLC (8,9). Given the crucial metabolic role of DHFR in cancer and use of DHFR pharmacological inhibitors in the clinic, it is important to understand how drugs like methotrexate affect anti-tumor immune escape and patient response to subsequent ICI therapy.

Here, we screen for a novel metabolic enzyme involved in the gene expression of immune checkpoint (IC) regulators in NSCLC and uncover a unique connection between the enzyme DHFR and the innate immune checkpoint regulator CD24 in lung cancer.

RESULTS

Methotrexate decreases CD24 gene expression in lung cancer cells

To gain insight into the link between tumor metabolism and antitumor immunity, we examined the effect of metabolic pathway inhibition on the expression of immune checkpoint regulator genes. We employed a phenotypic drug screen using a customized metabolic inhibitor library targeting thirteen enzymes. These 13 enzymes were selected from a group of 219 enzymes involved in key metabolic pathways implicated in cancer. Furthermore, respective metabolites of the selected enzymes were not reported to regulate gene expression. The screen involved treating the A549 lung cancer cell line with two doses of each inhibitor targeting an individual metabolic enzyme and subsequently measuring gene expression of a panel of immune checkpoint regulators implicated in cancer (Table 1). Inhibition of metabolic enzymes using buthionine sulphoximine (BSO), methotrexate, or lometrexol resulted in significant gene expression decrease of multiple immune checkpoint regulators including PVR, CD24, and IDO1, respectively in a dose-dependent manner (Fig. 1A). Among these changes, decrease of CD24 gene and protein expression by time-dependent pharmacological inhibition of dihydrofolate reductase (DHFR) using methotrexate was further confirmed in A549 and H157 cell lines (Fig. 1B and 1C). These data suggest that DHFR activity is involved in induction of anti-phagocytic CD24 expression in lung cancer cells.

Table 1.

Cancer-related metabolic targets and related inhibitors

Target enzymes	Genes	Pathway involved	Inhibitor	Ref.
Phosphofructokinase 2	PFKFB3	Glycolysis	3-(3-Pyridinyl)-1-(4-pyridinyl)-2-propen-1-one (3PO)	(17)
Diacylglycerol acyltransferase 1	DGAT1	Triglyceride synthesis	A922500	(29)
AICAR transformylase	ADE17	De novo purine biosynthesis	ATIC dimerization inhibitor	(22)
Dihydroorotate dehydrogenase	DHODH	De novo pyrimidine biosynthesis	Brequinar	(19)
Glutamate cysteine ligase	GCL	Glutathione biosynthesis	L-Buthionine-(S,R)-Sulfoximine (BSO)	(23)
Monoamine oxidase A	MAOA	Tryptophan metabolism	Clorgyline	(24)
GTP cyclohydrolase 1	GCH1	Non-essential amino acid synthesis	2,4-Diamino-6-hydroxypyrimidine (DAHP)	(26)
GAR transformylase	GART	De novo purine biosynthesis	Lometrexol	(28)
Malic enzyme 1	ME1	Anaplerosis	Malic enzyme 1 (ME1) inhibitor	(18)
Dihydrofolate reductase	DHFR	Thymidine synthesis	Methotrexate	(25)
Glucose-6-phosphate dehydrogenase	G6PD	Oxidative PPP	Polydatin	(21)
Lanosterol synthase	LSS	Cholesterol biosynthesis	RO 48–8071 fumarate	(20)
Delta(24)-sterol reductase	DHCR24	Cholesterol biosynthesis	Triparanol	(27)

Figure 1. Metabolic inhibitor screen identifies cancer-related enzymes including DHFR as potential immune gene regulators.

(A) Primary screen examining 12 known immune checkpoint (IC) regulators expression in lung cancer cells treated with inhibitors targeting 13 cancer-related metabolic enzymes. A549 cells were treated with reported working concentrations for 24 hours. Relative mRNA levels of IC regulators were measured via qRT-PCR with GAPDH as a control. (B and C) Effect of targeting DHFR using methotrexate on levels of CD24 gene (B) and CD24 protein (C). A549 (top panels) and H157 (bottom panels) cells were treated with methotrexate at 2 μM for 24 and 48 hours. CD24 mRNA levels and protein levels were determined by qRT-PCR and flow cytometry analysis. Representative FACS plots are shown on the right for (C). Data are mean ± SD from 3 replicates. Two-way ANOVA was used for statistics (ns: not significant; *: 0.01 < p < 0.05; **: p < 0.01).

DHFR and the metabolic product THF induce CD24 expression in cancer cells

To investigate whether DHFR contributes to CD24 gene expression, lung cancer cells were genetically modulated with DHFR and consequent changes in CD24 were monitored. We first assessed the impact of genetically target-downregulating DHFR on CD24 expression in lung cancer cells. Stable knockdown of DHFR significantly decreased CD24 at both gene and protein levels (Fig. 2A). We next determined whether DHFR overexpression could confer CD24 induction in lung cancer cells. We observed that transient overexpression of DHFR in A549 cells significantly enhances CD24 at both the RNA and protein level (Fig. 2B). Furthermore, supplementation with DHFR’s metabolic product, THF, also enhanced CD24 expression in lung cancer cell lines in a dose-dependent manner (Fig. 2C and 2D). These findings indicate that DHFR and its product THF are involved in regulation of CD24 expression in lung cancer.

Figure 2. Genetic modulation of DHFR alters expression of CD24 in lung cancer.

(A and B) Effect of DHFR knockdown (A) or overexpression (B) on CD24 gene expression and protein level. A549 cells were stably transduced with lentiviral DHFR shRNA for knockdown or transiently transfected with pLenti-V5-DHFR for overexpression. CD24 mRNA levels and protein amount on the cell surface were assessed by qRT-PCR and flow cytometry analysis, respectively. Knockdown or overexpression of DHFR was determined by DHFR qRT-PCR. (C and D) Effect of THF treatment on CD24 gene expression in lung cancer cells. A549 (C) and H157 (D) cells were treated with indicated concentrations of THF for 24 hours. CD24 mRNA levels were determined by qRT-PCR and normalized to GAPDH. Data are mean ± SD from 3 replicates. P values were determined by two-tailed unpaired Student’s t test (A-B) and one-way ANOVA (C-D) (ns: not significant; *: 0.01 < p < 0.05; **: p < 0.01).

Targeting DHFR in cancer cells enhances T cell-mediated tumor cell killing in part by signaling through CD24

We next examined whether DHFR is important for immune evasion in lung cancer. Through a co-culture model with lung cancer cells and human peripheral blood mononuclear cells (PBMCs) activated by anti-CD3 and anti-CD28 co-stimulation, we demonstrated that targeting DHFR with methotrexate results in enhanced immune-mediated tumor cell killing and reduced viability of A549 cells when co-cultured with activated T cells (Fig. 3A). In agreement, we observed that the treatment of methotrexate decreases CD24 gene expression in the presence of activated human PBMCs (Fig. 3B).

Figure 3. Methotrexate treatment attenuates CD24 expression and induces immune cell-mediated tumor cell killing.

(A) Effect of targeting DHFR using methotrexate on cancer cell viability in the presence and absence of activated primary human T cells. T cells in human peripheral blood mononuclear cells (PBMC) were activated by CD3/CD28/CD2 and IL2 followed by co-culture with A549 cancer cells for 48 hours treated with 2 μM of methotrexate. Immune cells were washed away, and cancer cell viability was measured by crystal violet assay. (B) CD24 levels were assessed by qRT-PCR in A549 cells co-cultured with primary activated T cells treated with methotrexate as described in (A). (C) Effect of methotrexate on T cell proliferation. Activated and CFSE-stained T cells in human PBMCs were cultured alone (left) or with A549 cancer cells (right) for 72 hours with or without 2 μM of methotrexate. T cell proliferation rate was determined by assessing CFSE stained cells through flow cytometry analysis. (D) Pretreatment of cancer cells with methotrexate enhances T cell-mediated immune cell killing. A549 cells were pretreated with 2 μM of methotrexate for 48 hours. A549 cells were washed and co-cultured with activated T cells as described in (A). Data are mean ± SD from 3 replicates. P values were determined by one-way ANOVA (*: 0.01 < p < 0.05; **: p < 0.01).

To further investigate whether the effect of methotrexate on immune-mediated tumor cell killing is through the inhibition of DHFR, specifically in cancer cells, we first examined the effect of methotrexate on T cell proliferation. Methotrexate enhanced T cell activity not when the drug was applied to T cells cultured alone but when these T cells were co-cultured with cancer cells (Fig. 3C). Moreover, pretreatment of cancer cells with methotrexate in the absence of T cells still enhanced T cell-mediated tumor cell killing (Fig. 3D). These data suggest that DHFR in cancer cells contributes to cancer immune evasion.

We next examined whether the DHFR-CD24 axis is involved in methotrexate-mediated immune-tumor cell killing. Forced expression of DHFR or CD24 significantly attenuated the methotrexate-induced tumor cell killing by T cells (Fig. 4A and 4B). These data suggest the methotrexate effect on T cell-mediated tumor cell killing occurs at least in part by inhibition of DHFR and the downstream effector CD24.

Figure 4. Overexpression of DHFR-CD24 in cancer cells enhances tumor immune evasion.

(A and B) Effect of DHFR (A) or CD24 (B) overexpression on methotrexate-mediated cancer cell killing in the presence of T cells. A549 cells transfected with vectors containing V5 tagged DHFR or CD24 were co-cultured with activated primary human T cells in the presence or absence of 2 μM of methotrexate for 48 hours. Cancer cell viability was assessed by crystal violet assay. The ectopic expression of DHFR and CD24 was analyzed by immunoblotting using anti- V5 antibody. Data are mean ± SD from 3 replicates. P values were determined by one-way ANOVA (ns: not significant; **: p < 0.01).

DHFR, CD24, and immune infiltration are clinically linked in human lung cancer

We next demonstrated the clinical relevance of our finding using publicly available databases. We found that DHFR is associated with poor clinical outcome in lung cancer patients. Overall survival rates were significantly lower in the patient group with high DHFR expression compared to the group with low DHFR expression (Fig. 5A). According to the TIDE database, there exists a significant negative correlation between CD24 mRNA expression and tumor infiltrating cytotoxic T lymphocyte levels in lung cancer patients (Fig. 5B). Furthermore, CD24 mRNA levels were lower in lung adenocarcinoma patients who received therapy containing DHFR inhibitor pemetrexed compared to patients who did not receive DHFR inhibitor-containing therapy (Fig. 5C). These data are in support of our finding that DHFR contributes to CD24 expression which is associated with tumor immune evasion and poor prognosis.

Figure 5. DHFR and CD24 expression negatively correlates with patient outcome and antitumor immunity in patients with lung cancer.

(A) Kaplan-Meier plot of lung cancer patients stratified by DHFR tumor mRNA expression. Highest and lowest 20% of DHFR mRNA expression groups were used to estimate overall survival in TCGA lung adenocarcinoma (LUAD) or lung squamous cell carcinoma (LUSC) patients. DHFR high (n=193) and DHFR low (n=192). (B) Correlation analysis of CD24 mRNA expression z-scores and tumor infiltrating cytotoxic T lymphocyte level z-scores in patients with LUAD or LUSC. Data are obtained from TIDE database patient cohort GSE50081 (n=181). (C) CD24 mRNA expression in LUAD patients treated with non-DHFR inhibitor (n=467) and DHFR inhibitor pemetrexed (n=57). Clinical information and CD24 RSEM were downloaded from Firebrowse. The log-rank test (A), Pearson correlation (B) and Welch two sample t-test (C) were used for statistical analyses.

DISCUSSION

Metabolic reprogramming is a common feature of cancer and has been researched extensively in the context of tumor bioenergetics and biosynthesis. In recent years, dysregulated cellular metabolism has been linked to tumor immune evasion, but the detailed molecular mechanisms through which altered metabolic pathways in cancer cells contribute to immune escape are still not well-understood. Here we seek to better understand the unique connection between lung cancer metabolism and anti-tumor immunity by investigating how the targeting of metabolic enzymes commonly implicated in cancer could influence the expression of genes related to immune regulation. Our findings delineate that cancer cell DHFR, an enzyme important for folate metabolism, is involved in regulating the anti-tumor immune response in part through the regulation of CD24: an innate immune checkpoint protein expressed on breast and ovarian cancer cells which prevents phagocytosis by macrophages.

Our study provides evidence that CD24 regulation mediated by DHFR involves its enzyme activity since pharmacological DHFR inhibition and supplementation of the metabolic product, THF, resulted in decreased and increased expression of CD24, respectively. Metabolites like succinate and alpha-ketoglutarate are reported to affect gene expression through allosteric modulation of epigenetic factors (10). Here we show that the immunomodulatory effects of DHFR in cancer cells is mediated through its metabolic product, THF. Further detailed biochemical and biophysical studies are warranted to elucidate the mechanism by which the enzyme activity and related metabolite THF and the NADP+/NADPH ratio are involved in CD24 regulation process. In addition, a previous report suggests that oncogenic Ras, which is activated in approximately 30% of all cancers, is a suppressor of CD24 mRNA and protein (11). Since oncogenic Ras signaling is prevalent in NSCLC, it would be worthwhile to further investigate how crosstalk between oncogenic Ras signaling and DHFR metabolic activity in lung cancer work to dynamically regulate the expression of the immune factor CD24.

Our clinical correlation studies using publicly available databases demonstrate that DHFR-CD24 expression is associated with poor survival and inversely correlates with the percent of tumor-infiltrating lymphocytes in lung cancer patients (12). DHFR inhibitors such as methotrexate and pemetrexed are widely used in the clinic to treat lung and other types of cancers (13). A recent study suggests that the combination of pembrolizumab and pemetrexed results in significantly higher rates of response and longer progression-free survival in NSCLC patients (14). From a clinical perspective, our finding may provide a clue that explains the therapeutic advantage of the combination, showing that it may be in part due to the CD24 mitigation mediated by DHFR inhibition, which may provide an additive effect with pembrolizumab on enhancing anti-tumor immune response.

EXPERIMENTAL PROCEDURES

Reagents

RO 48–8071 fumarate, clorgyline hydrochloride, malic enzyme inhibitor ME1 were obtained from MedChemExpress. Lometrexol and BSO were from Cayman Chemical. A922500 and brequinar were from Selleck Chemicals. Polydatin was from Santa Cruz Biotechnology. All other metabolic inhibitors used in the study were from Sigma Aldrich. Tetrahydrofolate was obtained from Sigma-Aldrich. Lentiviral short hairpin RNA clone for DHFR was from Dharmacon (sense 5’-CCTGAGAAGAATCGACCTTTA-3’). pLenti6.3/V5-DEST vector encoding DHFR and pLX304-V5-CD24 are obtained from DNASU (HsCD00940668 and HsCD00442245). Primers for HLA-A (forward 5’-GAGGAGGAAGAGCTCAGATAGA-3’ and reverse 5’-GGCAGCTGTCTCACACTTTA-3’) and all other commercially available primers for quantitative RT-PCR were obtained from Integrated DNA Technologies. FITC-conjugated anti-human CD24 antibody (SN3/ab30350) was purchased from Abcam. Anti-V5 antibody (sc-81594) and anti-β-actin antibody (AC-74/A2228) are from Santa Cruz Biotechnology and Sigma-Aldrich, respectively. CellTrace^™ CFSE Cell Proliferation Kit (C34554A) is from Life Technologies.

Cell culture and lentiviral infection

A549 and H157 cell lines were obtained from American Type Culture Collection (ATCC) and cultured in RPMI 1640 medium with 10% fetal bovine serum (FBS). 293T cells were cultured in DMEM with 10% FBS. Lentivirus production, stable gene knockdown and overexpression in human cancer cells were performed as described (15,16).

Metabolic inhibitor-based screen

Screening was performed using a customized inhibitor library, which targets metabolic pathways implicated in cancer. 2×10⁵ A549 cells were seeded into 6-well plates and treated with vehicle or two doses of each inhibitor targeting an individual metabolic enzyme. The dosages for each of the 13 inhibitors were determined based on previous literatures (17–29). Expression levels of the 12 immune checkpoint regulator genes were assessed after 24 hours of inhibitor treatment via quantitative RT-PCR.

Quantitative RT-PCR and flow cytometry analysis

The total RNA of A549 and H157 were isolated using RNeasy kit (QIAGEN) followed by reverse transcription of RNA (Applied Biosystems). qRT-PCR was performed using SYBR Green Supermix (Bio-Rad). The CD24 protein level was assessed by flow cytometric staining with anti-human CD24-FITC antibody (BD FACSymphony) and data were analyzed using FlowJo software.

T cell-mediated tumor cell killing assay

Human peripheral blood mononuclear cells (PBMC) were cultured in ImmunoCult-XF T cell expansion medium with ImmunoCult Human CD3/CD28/CD2 T cell activator (STEMCELL Technologies) and 10 ng/mL of IL-2 for one week according to the manufacturer’s protocol. The experiments were performed in RPMI1640 medium with anti-CD3 antibody (500 ng/mL) and IL-2 (10 ng/mL). 2 × 10⁴ A549 cells were plated into a 12-well plate and then co-incubated for 48 hours with activated T cells in the presence or absence of 2 μM of Methotrexate. The cancer cells-to-activated T cells ratio of 1:10 was used. T cells and cell debris were removed and live cancer cells were stained with crystal violet and quantified by a spectrometer at OD 570 nm (30).

T cell proliferation assay

Human primary T cells were activated in T cell expansion medium using ImmunoCult Human CD3/CD28/CD2 T cell activator and 10 ng/mL of IL-2 for a week. To investigate the effect of MTX on T cell proliferation in co-culture or only t-cell culture, T cells were stained with 5 μM of CellTrace^™ CFSE dye and maintained in RPMI medium supplemented with human CD3/CD28/CD2 T cell activator and 10 ng/mL of IL-2 for three days. T cell proliferation rates were assessed by analyzing CFSE fluorescence intensity using a flow cytometer.

Bioinformatics and statistical analysis

Gene-expression data and clinical information from TCGA were downloaded from Broad GDAC Firehose. Correlation analysis between z scores of tumor infiltrated cytotoxic T lymphocytes and CD24 expression were obtained from Tumor Immune Dysfunction and Exclusion (TIDE) database (31). Statistical analyses and graphical presentations were performed using GraphPad Prism 9. Data with error bars represent mean ± SD from three replicates. Statistical analysis of significance was based on two-way ANOVA for Figure 1A, two-tailed unpaired Student’s t test for Figures 2A, 2B, and 3C, one-way ANOVA for Figures 3 and 4, log-rank test for Figure 5A, Pearson correlation for Figure 5B, and Welch two sample t-test for Figure 5C.

DATA AVAILABILITY

All data are contained within the article.