Beyond PD-L1/PD-1: expression of PVR/TIGIT and TNFRSF14/BTLA axes in response to Helicobacter pylori and in gastric cancer lesions

Abstract

Gastric cancer (GC) has a high mortality rate worldwide, mainly due to late detection and lack of effective treatments. Immune checkpoint (ICs) inhibitors have revolutionized cancer treatment; however, a substantial percentage of patients with GC do not respond to this treatment, which may be related to aspects underlying the conformation of the immune microenvironment during tumor development and progression. In this study, we assessed the role of Helicobacter pylori, the most recognized GC risk factor, in the concomitant induction of PD-L1, B7-H3, PVR, TNFRSF14, Gal-3, and Gal-9 in vitro. We also analyzed the expression of selected IC ligand-receptor axes in human GC lesions. The mRNA relative expression of PD-L1, PVR, and TNFRSF14 was significantly upregulated in human cell lines challenged with the bacterium. This was CagPAI dependent and further enhanced by PBMCs. In GC lesions, mRNA levels in the PVR/TIGIT and TNFRSF14/BTLA axes were higher than in the PD-L1/PD-1 axis. When the protein levels of these ICs axes were determined by flow cytometry on the different cell subpopulations of the tumor microenvironment, Tregs emerged as the lymphoid subpopulation contributing the most to the expression of PD-1, TIGIT, and BTLA receptors; their ligands were mainly expressed by GC cells. Our results suggest that, in addition to the PD-L1/PD-1 axis, PVR/TIGIT and TNFRSF14/BTLA axes may play a role in establishing an immunosuppressive microenvironment during GC development and progression. This study also highlights the potential relevance of Tregs as active players in shaping and maintaining an immunosuppressive milieu in GC lesions, in part through expression of ICs.

Introduction

Gastric cancer (GC) has a high incidence globally.¹ Despite its great molecular heterogeneity, therapeutic approaches are uniform and their clinical benefit remain limited.² In most cases, GC is diagnosed at advanced stages when clinical interventions are likely to fail.³ These factors, together, contribute to explaining why GC is the fourth most deadly type of cancer worldwide.¹ More than 90% of GC are adenocarcinomas,⁴ classified as intestinal, diffuse, and mixed subtypes according to the Laurén histological system.⁵

The association between Helicobacter pylori infection and GC development is well established.^6–8  H. pylori is a type I carcinogen, Gram-negative, helix-shaped, flagellated bacterium that usually colonizes early in life the hostile environment of the stomach. Despite the strong immune reaction generated against this bacterium, H. pylori can persist lifelong in the absence of treatment by modulating and evading the immune response, which leads to the establishment of a chronic infection that profoundly changes the gastric microenvironment.^9,10  H. pylori is genetically diverse and one of its most prominent virulence factors is the cytotoxin-associated gene A (CagA) protein. The gene cagA is part of a genomic region called the Cag pathogenicity island (CagPAI), which also encodes the type IV secretion system (T4SS) that injects CagA into the cytoplasm of gastric epithelial cells.¹¹ Given the variety of cellular changes that it triggers once translocated into the cells, CagA is considered an oncoprotein.^12,13 CagA-positive strains are usually associated with a more robust inflammatory response and, consequently, with greater risk of GC.^12,14 Nonetheless, the clinical outcome of H. pylori infection depends on a complex interaction of bacterial, genetic, and environmental factors, such as host-immune response, lifestyle, and co-infections. There is still much to be elucidated about the role of this bacterium in the pathogenesis of GC, especially regarding the immune response it triggers and how this contributes to creating an immune microenvironment prone to GC development and progression.

Inflammation is an intrinsic feature of cancer.^15,16 In fact, malignant tumor growth is severely compromised if neoplastic cells are not immersed in an appropriate microenvironment where they coexist with immune and other nonimmune stromal cells. The so-called tumor microenvironment (TME) is the result of a co-evolution process in which, as a tumor develops, the architecture of the tissue microenvironment gradually changes.^15–17 Many factors influence this very complex interplay between the (pre)neoplastic and the non-neoplastic cells, thus leading to tumors with diverse immune contexts. An overwhelming body of evidence shows the critical role of the immune microenvironment in cancer development and progression.^18,19 Thus, several immunotherapeutic strategies have been developed to treat cancer. One of the most promising approaches is the pharmacological blockade of immune checkpoints (ICs) (ie immune checkpoint inhibitors [ICIs]).

The US Food and Drug Administration (FDA) has approved the use of ICIs, namely anti–PD-L1/PD-1 Abs, to treat advanced GC.^20,21 Unfortunately, only 11% to 24% of patients with GC benefit from these therapies.²² Although PD-L1/PD-1 is the most extensively studied ICs axis, many other ICs have been described,^23–25 and their clinical significance in (pre)-GC is undetermined. Presumably, these inhibitory molecules exert redundant and/or complementary roles.^26,27 Therefore, it is necessary to study other ICs in cancer to understand their biological roles and clinical implications.

To date, few studies have reported the induction of ICs in response to H. pylori or in GC tissue.^28,29 Most of these focused on characterizing PD-L1/PD-1 and none of the studies evaluated the concomitant expression of several ICs. Also, the implications of the differential induction of less-explored ICs in the development and progression of GC remain enigmatic. Given that GC arises in the context of chronic inflammation caused by H. pylori, unraveling the role of this bacterium in the induction of nonclassical ICs is important because it may contribute to molding the local immunological microenvironments thus favoring tumor development. Interestingly, recent studies have highlighted the relevance of H. pylori infection in shaping the immune response systemically because it can affect the response to anti–PD-1 therapy in non–small-cell lung cancer³⁰ and melanoma.³¹

Here, we studied the potential role of H. pylori infection in the concomitant induction of several ICs. For this purpose, the PD-L1/PD-1, B7-H3, PVR/TIGIT, TNFRSF14/BTLA, Gal-3/LAG3, and Gal-9/TIM3 ICs axes were analyzed in GC cell lines challenged with H. pylori. Based on these analyses, we also evaluated the mRNA and protein levels of selected ICs axes in different cell populations in human GC samples. Overall, we aimed to gain insight into the role of ICs axes other than PD-L1/PD-1 in the establishment of an immune microenvironment that is optimal for GC development and progression.

Materials and methods

Bacterial and cell culture

The 7.13 wild-type (7.13 wt), P12wt, and P12 CagPAI-deficient (P12ΔCagPAI-deficient) H. pylori strains were grown at 37 °C in a microaerophilic atmosphere (5% O₂, 10% CO₂, and 85% N₂) on Wilkins-Chalgren anaerobic agar (Oxoid) supplemented with human erythrocytes (10%), vancomycin (10 mg/L), trimethoprim (5 mg/L), cefsulodin (5 mg/L), and amphotericin B (5 mg/L). Bacterial suspensions with an OD₆₀₀ of 0.95 to 1.1 were prepared in Dulbecco’s PBS (Gibco). To confirm bacterial viability, we observed suspensions at 100× to verify H. pylori characteristic curved-shape morphology and corkscrew-like motility. We also conducted phenol red urease, catalase, and oxidase tests.

GC cell lines MKN74 (derived from a secondary intestinal-subtype gastric adenocarcinoma) and MKN45 (derived from a secondary diffuse-subtype gastric adenocarcinoma) were cultured in RPMI-1640 medium (Gibco) supplemented with 10% FBS, 100 units/mL penicillin, 100 μg/mL streptomycin, 0.25 μg/mL amphotericin B, and 2.5 μg/mL ciprofloxacin at 37 °C in a 5% CO₂ humidified atmosphere. PBMCs were isolated from healthy donors using the Ficoll-diatrizoate method, as described elsewhere.³²

GC cell lines with H. pylori were co-cultured at a multiplicity of infection (MOI) of 50 in RPMI-1640 medium (Gibco) supplemented with 100 μg/mL vancomycin. The MOI of 50 was selected based on previous experiments.³³ To ensure the MOI in each experiment, the number of cells per well was confirmed on the day of infection. Infection efficiency was validated with the observation of the hummingbird phenotype, as previously reported.³⁴ The PBMCs-to-cell lines ratio for co-culture was 4:1. All in vitro experiments were incubated at 37 °C in a 5% CO₂ humidified atmosphere. Supernatant of H. pylori–exposed PBMCs (ie the conditioned medium) was collected after a 48-h incubation with H. pylori, centrifuged at 9,390 x g for 10 min at 4 °C to eliminate cells, and stored at −80°C until used.

GC samples

Gastric tumor samples of approximately 0.5 cm³ were prospectively collected from 16 May 2022, to 28 February 2025 at Dr. Maximilano Peralta Hospital (Cartago, Costa Rica) from surgically resected tissue from 42 patients, aged 18 years or older, who had GC. For RNA extraction, samples were placed in a microcentrifuge tube with 0.8 mL of RNAlater solution (Thermo Scientific) and stored at −80°C. For flow cytometry, samples were collected in microcentrifuge tubes with PBS and immediately transported to the laboratory to be macerated and passed through 70-μm filters to separate cells. Cells were then washed with PBS, resuspended in 1 mL of FBS with 10% DMSO, and stored at −80°C. For bacterial culture, samples were collected in microcentrifuge tubes with a semisolid transport medium made with 7.5 g of semisolid Brucella agar (Oxoid), 6.3 g of blood heart infusion broth (Oxoid), and 1 vial of H. pylori selective supplement (Dent, Oxoid). Samples were immediately transported to the laboratory to be macerated with blood heart infusion broth. Macerated material was distributed on 4 culture plates: 2 plates with Columbia agar (Oxoid) supplemented with lysed horse blood (10%), vancomycin (10 mg/L), trimethoprim (5 mg/L), and polymyxin B (2,500 UI/L); and 2 plates of Wilkins-Chalgren anaerobic agar (Oxoid) supplemented with human erythrocytes (10%), vancomycin (10 mg/L), trimethoprim (5 mg/L), cefsulodin (5 mg/L), and amphotericin B (5 mg/L). Plates were cultivated at 37 °C in a microaerophilic atmosphere (5% O₂, 10% CO₂, and 85% N₂) for at least 48 h and until 10 d.

For immunohistochemistry (IHC), formalin-fixed and paraffin-embedded samples were obtained postoperatively from the Pathology Laboratory of the forementioned hospital. IHC for H. pylori was performed as previously reported.³⁵ We also collected patients’ serum samples to measure Abs to H. pylori by an in-house ELISA developed in our laboratory.³⁶  H. pylori infection status was reported as positive if culture, IHC, and/or serology were positive. Histopathological characteristics of the tumor samples, H. pylori infection status, and demographic characteristics of the patients are summarized in Table 1.

Table 1.

Patients’ demographic data.

Characteristic	No. (%)
Sex
Female	23 (54.8)
Male	19 (45.2)
Laurén classification
Intestinal	17 (40.5)
Diffuse	18 (42.8)
Mixed	5 (11.9)
Other	2 (4.8)
Stage
I	17 (40.5)
II	11 (26.2)
III	14 (33.3)
H. pylori status
Positive	37 (88.1)
Negative	5 (11.9)

The Scientific Ethics Committee of the University of Costa Rica, the Central Scientific Ethics Committee of the Costa Rican Social Security (Caja Costarricense del Seguro Social), and the National Health Research Council (Consejo Nacional de Investigación en Salud) of Costa Rica approved the collection and use of these samples (permission no. R019-SABI-00215). Written informed consent was obtained from all patients. All these steps are accordance with Costa Rican Law 9234 (Ley Reguladora de Investigaciones Biomédicas).

RNA extraction, retrotranscription, and quantitative PCR

RNA from in vitro experiments and human gastric samples was isolated using Trizol reagent (Invitrogen) following the manufacturer’s instructions. cDNA synthesis was performed using the High-Capacity cDNA Reverse Transcription Kit (Applied Biosystems). Quantitative PCR (qPCR) was performed using the Maxima SYBR Green/ROX qPCR Master Mix kit (Thermo Scientific) with specific primers (Table S1) at 0.3 µM. For data normalization, housekeeping genes were chosen by NormFinder software (available at https://www.moma.dk/software/normfinder). For cell lines, NormFinder analysis was performed with the cycle threshold data from H. pylori–infected and noninfected cells from 2 biological replicates. For gastric tissue, NormFinder analysis was performed with cycle threshold data from tumor and adjacent-tumor samples from 3 patients. The genes with the best stability index were HRPT1 for the MKN74 cell line and gastric tissue, and RPS20 for the MKN45 cell line (Table S2). mRNA levels of each gene of interest in cell lines and human tissue samples were normalized with mRNA levels of the corresponding reference gene. The mRNA relative expression in the in vitro model with respect to uninfected cells was calculated using the comparative cycle threshold method.

Flow cytometry analysis

Cultured and patient cells were blocked with inactivated human serum and stained with fluorescent-conjugated specific Abs (Table S3). Cells populations were designated according to the flow chart shown in Figure S1 for cultured cells and Figure S2 for cells in patients’ samples.

Statistical analyses

Statistical analyses were performed in GraphPad Prism 7.02. To evaluate if data had a normal distribution, the Shapiro–Wilk test was performed. To compare between 2 groups, Student’s t test or the Mann–Whitney U test were used, according to data normality. To compare among multiple groups, 1-way ANOVA with a Dunnett’s or a Tukey’s test for multiple comparisons was performed when data had a normal distribution, and the Kruskal–Wallis test with a Dunn’s test for multiple comparisons was performed when data did not pass the normality test. When analyzing data with 2 independent categorical factors (ie time and bacterial strain), a 2-way ANOVA with a Tukey’s post hoc test was performed when data met the normality criteria; separate Kruskal–Wallis tests with a Dunn’s test for multiple comparisons were performed when data did not have a normal distribution. A significant relationship was established when P < 0.05.

The Cancer Genome Atlas database

The publicly available Cancer Genome Atlas (TCGA) database was accessed to compare IC mRNA relative gene expression in samples from patients with GC. This information was reported by the TCGA program based on data obtained from different high-throughput next-generation sequencing platforms, according to the involved center (https://www.cancer.gov/tcga).

Results

H. pylori induces mRNA relative expression of IC ligands in GC cell lines

MKN74 and MKN45 cell lines were co-cultured with H. pylori 7.13 wt to evaluate the kinetics of mRNA relative expression of different IC ligands, namely PD-L1, B7-H3, PVR, TNFRSF14, Gal-3, and Gal-9. When MKN74 cells were co-cultured with H. pylori 7.13 wt, the mRNA relative expression of PD-L1, PVR, TNFRSF14, and Gal-3 increased gradually as a function of postinfection time, reaching their top level with a significant upregulation at 24 h after infection (Figure 1A). In the MKN45 cell line, PD-L1, B7-H3, TNFRSF14, and Gal-3 mRNA relative expression was significantly increased in response to H. pylori 7.13 wt (Figure 1B). In this cell line, the kinetics of expression for TNFRSF14 showed a steady upregulation after 8 h after infection, and its expression was highest at 24 h, whereas PD-L1 expression reached its peak at 4 h after infection and remained significantly upregulated, but stable, at later time points. The comparison of the mRNA relative expression levels of the analyzed ICs between the 2 cell lines revealed a dissimilar induction at 24 h after infection (Figure 1C). Of note, TNFRSF14 stands out as the most clearly upregulated gene in both the MKN74 and MKN45 lines. Overall, we show that H. pylori infection induces the relative expression of several IC ligands in vitro and confirm the already documented upregulation of PD-L1 in response to this bacterium.^28,29

Figure 1.

mRNA relative expression of IC ligands in H. pylori–infected GC cell lines. (A and B) mRNA relative expression of PD-L1, B7-H3, PVR, TNFRSF14, Gal-3, and Gal-9 in MKN74 (A) and MKN45 (B) cell lines co-cultured with H. pylori 7.13 wt at different times. Statistically significant differences in mRNA relative expression between each time point of infection vs noninfected cells are shown. (C) MKN74 and MKN45 cells’ mRNA relative expression of IC ligands after 24 h of H. pylori 7.13 wt infection. Results are shown as the mean of 3 to 6 biological replicates. Normality was assessed using the Shapiro–Wilk test. Based on the distribution of the data, we applied either a 1-way ANOVA followed by a Dunnett’s post hoc test (for normally distributed data) or a Kruskal–Wallis test followed by a Dunn’s multiple-comparison test (for non-normally distributed data). Statistically significant differences in mRNA relative expression between each gene vs PDCD1L1 are shown. hpi, hours postinfection; Ni, not infected. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

H. pylori–sustained induction of mRNA IC ligands relative expression is CagPAI dependent

To determine the role of CagPAI in the induction of IC ligands, the MKN74 cell line was co-cultured with P12wt or P12ΔCagPAI H. pylori strains, and mRNA relative expression of IC ligands was evaluated at 4, 12, and 24 h after infection. PD-L1, PVR, and TNFRSF14 mRNA relative expression was significantly upregulated at 24 h after infection in the cells co-cultured with the wt strain, but not with the CagPAI-deficient strain (Figure 2). At 12 h after infection, P12ΔCagPAI H. pylori induced a significant increase in the mRNA relative expression of B7-H3, PVR, TNFRSF14, and Gal-9, which did not persist because it was not observed at the next time point (Figure 2). Hence, this analysis revealed that the sustained induction of mRNA IC ligand relative expression in response to H. pylori is CagPAI dependent.

Figure 2.

mRNA relative expression of IC ligands in a GC cell line infected with a CagPAI-deficient H. pylori strain. mRNA relative expression of PD-L1, B7-H3, PVR, TNFRSF14, Gal-3, and Gal-9 at different time points in the MKN74 cell line co-cultured with H. pylori P12wt or P12ΔCagPAI strains. Results are shown as the mean of 3 biological replicates. Normality was assessed using the Shapiro–Wilk test. Based on the distribution of the data, either a 2-way ANOVA with a Tukey’s post hoc test (for normally distributed data) or separate Kruskal–Wallis tests with a Dunn’s multiple-comparison test for each time point (for non-normally distributed data) was applied. For each time point, statistically significant differences in mRNA relative expression between infected cells with different H. pylori strains vs not infected cells are shown. Hp, Helicobacter pylori; hpi, hours postinfection; Ni, not infected. *P < 0.05, **P < 0.01, ***P < 0.001.

H. pylori induction of protein IC ligands expression on GC cell lines is significantly increased in the presence of immune cells

Next, we assessed the expression at the protein level of IC ligands selected based on the mRNA results reported in the preceding subsections. For this purpose, MKN74 cell line cultures were exposed either to H. pylori 7.13 wt alone, PBMCs alone, or H. pylori 7.13 wt and PBMCs simultaneously for 48 h, and protein levels of PD-L1, PVR, and TNFRSF14 on the surface of MKN74 cells were assessed by flow cytometry (Figure 3 and Figure S3). When MKN74 cells were co-cultured with H. pylori 7.13 wt alone, protein levels of the 3 ligands tended to increase, although this was not statistically significant (Figure 3). However, when the cells were co-cultured with H. pylori 7.13 wt and PBMCs simultaneously, PD-L1, PVR, and TNFRSF14 protein levels were significantly increased (Figure 3 and Figure S3). In summary, H. pylori–driven induction of IC ligands in cancer cell lines at the protein level is further increased in the presence of immune cells.

Figure 3.

IC ligand surface protein levels in a H. pylori–infected GC cell line. Protein levels of PD-L1, PVR, and TNFRSF14 on the surface of MKN74 cells (CD45⁻, EpCAM⁺) after 48 h of exposure to H. pylori 7.13 wt alone, PBMCs alone, H. pylori 7.13 wt plus PBMCs, or H. pylori 7.13 wt plus PBMCs in conditioned medium (PBMCs-CM). Results are reported as the mean of 3 to 5 biological replicates. Normality was confirmed using the Shapiro–Wilk test. For each molecule, a 1-way ANOVA followed by a Tukey’s post hoc test was performed. Statistically significant differences in protein expression between the experimental conditions are shown. Hp, Helicobacter pylori; MFI, median fluorescence intensity; Ni, not infected. *P < 0.05, **P < 0.01.

The effect of immune cells in the upregulation of H. pylori–driven IC ligands on the surface of cancer cell lines is mediated by different mechanisms

To evaluate whether the upregulation in the protein level of IC ligands is mediated by interaction between cells or through secreted molecules, the conditioned medium of H. pylori–exposed PBMCs was added to co-cultures of MKN74 cells and H. pylori 7.13 wt. Adding the conditioned medium did not change PD-L1 and TNFRSF14 protein levels on the surface of H. pylori–challenged MKN74 cells (Figure 3). In contrast, PVR protein levels on the surface of MKN74 cells significantly increased when H. pylori–challenged cells were also exposed to the conditioned medium of H. pylori–exposed PBMCs (Figure 3); in fact, the protein level was comparable to MKN74 cells co-cultured with H. pylori and PBMCs simultaneously. These findings suggest the effect of immune cells in the upregulation of H. pylori–driven induction of IC ligand expression on the surface of cancer cell lines in some cases results from cell–cell interaction (eg PD-L1 and TNFRSF14), whereas in others, it is through secreted molecules (eg PVR).

mRNA levels of PVR/TIGIT and TNFRSF14/BTLA axes are higher than those of the PD-L1/PD-1 axis in human GC tissue

The well-stablished role of H. pylori infection in gastric carcinogenesis prompted us to evaluate mRNA levels of the IC ligand-receptor axes that were induced in our in vitro model—namely PD-L1/PD-1, PVR/TIGIT, and TNFRSF14/BTLA—on 42 surgically resected human GC tissue samples. Interestingly, PVR and TNFRSF14 mRNA levels were markedly higher than PD-L1 levels (Figure 4A). Similarly, TIGIT and BTLA mRNA levels were higher than PD-1 levels (Figure 4B). An analysis of mRNA levels performed for each patient revealed a general trend indicating that TNFRSF14 and TIGIT are the most highly expressed IC ligand (Figure 4C) and receptor (Figure 4D), respectively, although interpatient variability in IC expression was evident. In fact, an exploratory analysis of the relative gene expression of PD-L1 (CD274), PVR, TNFRSF14, PD-1, TIGIT, and BTLA with data from patients with GC in the TCGA also showed heterogeneity (Figure 4E).

Figure 4.

IC mRNA levels in human GC tissue. (A) IC ligand PD-L1, PVR, and TNFRSF14 mRNA levels normalized with HPRT1 mRNA levels in human GC samples. (B) IC receptors PD-1, TIGIT, and BTLA mRNA levels normalized with HPRT1 mRNA levels in human GC samples. (C and D) Heat maps of IC ligands (C) and receptors (D) mRNA levels normalized with HPRT1 mRNA levels in human GC samples. (E) IC GC mRNA relative expression heat map from TCGA data. (A–D) n = 42 patients. (A and B) The normality Shapiro–Wilk test showed non-normally distributed data. A Kruskal–Wallis test followed by a Dunn’s multiple-comparison test was applied for each molecule. Statistically significant differences in mRNA expression between the experimental conditions are shown. *P < 0.05, ****P < 0.0001.

To evaluate how much of this variation was related to tumor pathological characteristics, IC mRNA levels were assessed according to Laurén classification of histologic tumor subtypes and TNM stage. These analyses revealed that PD-L1 mRNA levels were significantly higher in the mixed subtype compared with the intestinal-subtype, although few samples were classified as mixed (Figure 5A, left). More evident, however, was the significantly higher mRNA levels of TNFRSF14 (Figure 5E, left) and BTLA (ie the receptor of TNFRSF14) (Figure 5F, left) in diffuse and mixed subtypes compared with the intestinal subtype. mRNA levels of PVR (Figure 5C, left), PD-1 (Figure 5B, left), and TIGIT (Figure 5D, left) did not differ among Laurén histological subtypes. Finally, none of the ICs analyzed showed differential mRNA levels according to TNM stage (Figs 5A–F, right). We did not compare IC mRNA levels according to H. pylori infection status, because 88% of patients were positive (Table 1), which is in line with global statistics.^7,37 Therefore, the negative cohort is underrepresented, and this compromises statistical validity. In summary, the mRNA level of the PVR/TIGIT and TNFRSF14/BTLA axes is higher than of the PD-L1/PD-1 axis in human GC, but only the TNFRSF14/BTLA axis is seemingly correlated with pathological features of the tumor.

Figure 5.

IC mRNA levels in human GC tissue according to pathological features. (A–F) PD-L1 (A), PD-1 (B), PVR (C), TIGIT (D), TNFRSF14 (E), and BTLA (F) mRNA levels normalized with HPRT1 mRNA levels in human GC samples according to Laurén histological classification (on the left in each panel) and TNM stage (on the right in each panel). n = 42 patients. Normality was assessed using the Shapiro–Wilk test. Based on the distribution of the data, either a 1-way ANOVA followed by a Tukey’s post hoc test (for normally distributed data) or a Kruskal–Wallis test followed by a Dunn’s multiple-comparison test (for non-normally distributed data) was applied. Statistically significant differences in mRNA levels between the pathological features are shown. *P < 0.05, **P < 0.01.

IC protein levels are higher in tumor cells and Tregs in GC

Surgically resected human GC tissue samples were also analyzed by flow cytometry to evaluate the protein levels of ICs on the surface of different cell types found in the TME. Ligands were evaluated on tumor (EpCAM⁺) and myeloid (CD11b⁺) cells, and receptors were measured on the following lymphoid populations: CD8⁺ T cells, CD4⁺ cells, Tregs, NK cells, and NKT cells. Because both TNFRSF14 and BTLA can be expressed on tumor, myeloid, and lymphoid populations,^38,39 BTLA was also analyzed as a ligand and TNFRSF14 was also evaluated as a receptor (Figure S4). We found that tumor cells are the main cell type expressing the IC ligands PD-L1, PVR, TNFRSF14, and BTLA (Figure 6 and Figure S4A). With regard to IC receptors, Tregs was the cell subpopulation with the highest levels of PD-1, TIGIT, BTLA, and TNFRSF14, followed by NK/NKT cells (Figure 7 and Figure S4B). Hence, in human GC, the analyzed IC ligands and receptors are mainly expressed by tumor cells and Tregs, respectively.

Figure 6.

IC ligands surface protein levels in human GC tissue. (A–C) PD-L1 (A), PVR (B), and TNFRSF14 (C) protein levels in tumor cells (CD45⁻, EpCAM⁺) and myeloid cells (CD45⁺, EpCAM⁻, CD11b⁺) populations of human GC samples. Dot plots show mean fluorescence intensity (MFI) (left), and histograms illustrate the distribution of fluorescence intensity at the single-cell level for a representative sample (right). n = 42 patients. The normality Shapiro–Wilk test showed a non-normally distributed data. For each molecule, a Mann–Whitney test was applied. Statistically significant differences in protein levels between both cell types are shown. ****P < 0.0001.

Figure 7.

IC receptors surface protein levels in human GC tissue. (A–C) PD-1 (A), TIGIT (B), and BTLA (C) protein levels in populations of CD8⁺ T cells (CD45⁺, EpCAM⁻, CD3⁺, CD11b⁻, CD56⁻, CD8⁺), CD4⁺ T cells (CD45⁺, EpCAM⁻, CD3⁺, CD11b⁻, CD56⁻, CD4⁺, CD25⁻, CD127⁻), NK cells (CD45⁺, EpCAM⁻, CD3⁻, CD11b⁻, CD56⁺), NKT cells (CD45⁺, EpCAM⁻, CD3⁺, CD11b⁻, CD56⁺), and Tregs (CD45⁺, EpCAM⁻, CD3⁺, CD11b⁻, CD56⁻, CD4⁺, CD25⁺, CD127⁻) in human GC samples. Dot plots show mean fluorescence intensity (MFI) (left), and histograms illustrate the distribution of fluorescence intensity at the single-cell level for a representative sample (right). n = 42 patients. The normality Shapiro–Wilk test showed a non-normally distributed data. For each molecule, a Kruskal–Wallis test followed by a Dunn’s multiple-comparison test was applied. Statistically significant differences in protein levels between the cell types are shown. *P < 0.05, **P < 0.01, ****P < 0.0001.

Discussion

In this study, we assessed the differential expression of PD-L1/PD-1 and other unconventional IC axes in cell lines challenged with H. pylori and in human GC tissue samples. Our results reveal that mRNA levels of PVR/TIGIT and TNFRSF14/BTLA axes are higher than of the PD-L1/PD-1 axis, both in vitro and in vivo, and Tregs play a major role in the protein expression of ICs receptors in human GC lesions. These less-studied IC axes can exert inhibitory stimulus on lymphocytes, acting in redundance or complementation to PD-L1/PD-1.^26,27 Also, H. pylori–driven induction of nonclassical ICs may contribute to shaping the local gastric microenvironment toward a milieu more prone to GC development. In fact, the effect of the infection can go beyond the local microenvironment, as exemplified by recent studies showing that H. pylori seropositivity is associated with a decreased response to anti–PD-1 therapy in patients with non–small cell lung cancer and those with melanoma.^30,31

We found that H. pylori strains 7.13wt and P12wt, both CagPAI positive, induce a concomitant upregulation in the mRNA relative expression of several IC ligands in the MKN45 and MKN74 cell lines. H. pylori colonization of the gastric mucosa triggers an acute inflammation characterized by the recruitment and activation of several immune cells, which is further exacerbated by CagPAI-positive strains.^12,14 Because the immune response is unable to clear the bacterial infection, it evolves to a chronic and persistent inflammatory reaction.⁴⁰ Presumably, the observed overexpression of ICs is a consequence of this process.

Of note, the kinetics of mRNA relative expression of the studied IC ligands in H. pylori–challenged cell lines revealed that the highest level of induction occurs at late time points (eg 24 h after infection), which is consistent with the expected biological behavior of ICs.²⁵ Nevertheless, PD-L1 mRNA relative expression significantly increased much earlier, at 4 h after infection, and remained upregulated at later time points. Overall, the observed kinetics may reflect different functional categories of IC ligands, playing redundant or complementary roles, as previously conceptualized for ICs receptors.⁴¹ Also, some ligands (eg PVR and TNFRSF14) act as costimulators or co-inhibitors, depending on the receptors and type of lymphocytes they interact with.^38,39,42 Given these complex interactions, the concomitant expression of several ICs, as well as the time of induction, greatly influences the strength and polarity of the immune response. This, in turn, affects the conformation of the (pre)malignant microenvironment during GC development and progression, which, ultimately, could have an impact on the clinical outcome of anticancer immunotherapy. In line with this, it has been reported that the effectiveness of immunotherapy-induced antitumor response is time dependent.⁴³

The present study also revealed that the sustained induction of all the analyzed IC ligands was CagPAI dependent, because their mRNA relative expression returned to basal levels at 24 h after infection in cell lines challenged with a CagPAI-negative H. pylori strain. The CagPAI gene region comprises more than 30 genes, including the gene encoding CagA, an oncoprotein that exerts a variety of effects in the gastric microenvironment, including the regulation of the inflammatory response.⁴⁴ Consequently, the association between H. pylori CagPAI-positive strains and GC is well established.^12,14 In the present study, we did not assess whether the induction of ICs results from translocation of CagA itself, the interaction between components of the T4SS and cell surface proteins, or if it results from any other CagPAI-mediated mechanism. Nevertheless, several of the IC ligands we studied had a transient upregulation at 12 h after infection in response to the CagPAI-mutant strain, being particularly clear for PVR, TNFRSF14, and Gal-9. This observation suggests that H. pylori can induce these molecules through CagPAI-independent mechanisms, but only CagPAI components lead to a sustained upregulation. This is supported by previous reports showing the immune response against CagPAI-positive strains is more robust than for CagPAI-negative strains.⁴⁵

Our results show that PD-L1, PVR, and TNFRSF14 were the most upregulated IC ligands in H. pylori–challenged cell lines; of these 3, only PD-L1 had been previously linked to this bacterial infection.^28,29 We also found induction of other IC molecules, such as B7-H3, Gal-3, and Gal-9, in response to H. pylori infection, which is consistent with previous reports,^46,47 but at a much lower scale than PD-L1, PVR, and TNFRSF14. Interestingly, upregulation of Gal-3 and Gal-9 receptors (ie LAG3 and TIM3, respectively) is a general feature of immunosuppressive TMEs in many types of cancers, including GC.^48,49 Both LAG3 and TIM3 can interact with other ligands (not studied in this work) that can also lead to the activation of immune-suppressive signaling pathways. All this reflects that the induction of ICs in response to H. pylori infection is part of an intricate, heterogeneous, and dynamic series of immunological events occurring during gastric carcinogenesis. It also highlights the importance of analyzing the kinetics of expression of several of these molecules simultaneously to gain better insight into the complex dynamics taking place in the (pre)neoplastic lesions.

IC expression in cancer cells can be regulated by aberrant oncogenic and inflammatory signaling pathways, such as PI3K-AKT, ALK, and IFN.⁵⁰ In our experimental setting, adding PBMCs to the MKN74–H. pylori co-cultures reinforced the induction of PD-L1, PVR, and TNFRSF14 ligands on cultured cells, which is seemingly mediated via different mechanisms. More specifically, reinforcement in the expression of PD-L1 and TNFRSF14 was only seen when cancer cells interacted with PBMCs, suggesting cell–cell contact dependency, whereas PVR expression was equally upregulated in the presence of PBMCs and when only the conditioned medium from PBMCs previously exposed to H. pylori was added, indicating induction via secreted soluble factors. Heterotypic signaling plays a pivotal role in determining cellular function and this is the case for IC ligand-receptor interaction, whose impact on immune outcome depends on the balance between the strength of the inhibitory and activation signals received within a certain time window.⁵¹ Presumably, PBMCs’ strengthening of H. pylori–driven IC induction enhances the immune suppressive properties, thus contributing to the generation of a tumor-supportive milieu that may facilitate malignant transformation of the gastric epithelial cells.

When IC ligand-receptor pairs were measured in human GC samples, the mRNA levels of both PVR/TIGIT and TNFRSF14/BTLA axes were higher than that of the PD-L1/PD-1 axis. This observation is particularly interesting because it suggests other immunosuppressor molecules may be as relevant as PD-L1/PD-1 in shaping the TME. As mentioned, the concomitant expression of several IC axes may exert redundant or complementary effects.^26,27 A limitation of our study, however, is that we could not measure the relative abundance of these genes compared with the baseline condition, because of the lack of non-neoplastic tissue from healthy control individuals. Nevertheless, based on our observations, it is tempting to speculate that concomitant expression of ICs in the tumor milieu may contribute to explaining the poor response rate to anti–PD-L1/PD-1 in several cancer types, including GC. This could be the reason why the combination of IC blockers results in improved efficacy, as exemplified by the FDA-approved combined administration of anti–PD-1 with anti–CTLA-4 or anti-LAG3 Abs, both for treatment of patients with melanoma.^52,53 More recently, a study conducted with samples from patients with non–small cell lung cancer who had participated in the CITYSCAPE clinical trial,⁵⁴ coupled with a preclinical exploration, concluded that combination of anti-TIGIT Ab and anti–PD-L1 Ab results in improved progression-free and overall patient survival, compared with those treated with anti–PD-L1 alone.⁵⁵ In line with this, 2 clinical trials are currently evaluating clinical efficacy of anti-TIGIT and anti–PD-1 combinations in locally advanced unresectable or metastatic gastric, gastroesophageal junction, and esophageal adenocarcinoma (ClinicalTrials.gov identifiers NCT05329766 and NCT05568095). Taken together, these studies and our results suggest that the expression of immunoregulatory molecules varies according to the type of cancer and the benefit of combining ICIs depends on the immune context of each cancer type. These observations also highlight the relevance of further evaluating the concomitant expression of immunoregulatory molecules in cancer, particularly in GC.

We assessed the variability of the studied ICs among patients with GC. We observed marked interpatient variability, which was expected, given the etiological, histological, molecular, and immunological heterogeneity that characterizes GC.^56–58 This variability can have important repercussions in the clinical outcome of and treatment options for patients. Nevertheless, we found that several ICs were generally co-expressed in the same sample from a patient with GC, which, as mentioned, is also relevant because treatment with a combination of ICIs results in improved patient survival. Possibly, to obtain maximal benefit from anti–PD-L1/PD-1 therapies, other ICs molecules should be evaluated as predictive biomarkers. We also compared IC mRNA levels in GC tissue samples according to parameters of clinical relevance, such as the Laurén histological subtype and tumor stage. The TNFRSF14–BTLA pair was the only studied IC axis differentially expressed between intestinal and diffuse subtypes, even though Laurén histological GC subtypes have marked etiological, biological, and clinical differences.^56–58 More specifically, the gene expression of both TNFRSF14 and BTLA was higher in diffuse- than intestinal-subtype GC tissue samples, which is intriguing because the diffuse subtype is less associated with chronic inflammation than is the intestinal subtype.^59–61 There was no difference in the expression of any of the analyzed ICs in terms of GC different tumor stages, which is contrary to what has been observed in colorectal cancer.⁶²

It is well known that IC overexpression triggers an exhausted cell state on CD8⁺ T cells that inhibits their cytotoxic effect and, thus, they lose their capability to eliminate tumor cells.^51,63–65 ICs also impair the function of other lymphoid populations, such as NK cells⁶⁶ and CD4⁺ T cells.⁶⁷ Conversely, overexpression of these same molecules on Tregs increases their suppressive functions,^68,69 which ultimately favor tumor growth. Interestingly, in our human GC samples, Tregs were the immune cell subpopulation with the highest contribution to the protein expression of PD-1, TIGIT, and BTLA. Researchers who conducted single-cell RNA sequencing found that Tregs are abundant in gastric tumors, but the exhausted CD8⁺ T cells are scarce.⁷⁰ Collectively, our own and these findings suggest that, in GC, Tregs are active players in shaping and maintaining the tumor immune-suppressive context, possibly by several mechanisms including the expression of IC receptors on their surface. In fact, Guan et al.⁵⁵ showed that patients with cancer with high tumor density of Tregs benefited the most when treated with anti-TIGIT and anti–PD-L1 Abs. Shitara et al.⁷¹ reported that high Treg gene expression signature scores in gastroesophageal adenocarcinomas were associated with an increase in overall survival when patients received anti–PD-1 in combination with anti–CTLA-4, and Kwon et al. found that in patients with metastatic GC treated with pembrolizumab, the Tregs subpopulation was decreased in responders.⁷²

PVR and TNFRSF14 are recognized as ICs because both can exert coinhibitory roles on lymphoid cells by interacting with TIGIT and BTLA receptors, respectively. Additionally, they can stimulate different pathways, thus generating different functional effects, depending on the cells expressing them and the receptors they interact with.^38,39,42 In fact, TNFRSF14 has been regarded as a “molecular switch” because of its dual role, acting as ligand when expressed on tumor or myeloid cells or as receptor when expressed on lymphoid cells.³⁹ In the context of Ag presentation, the interaction of TNFRSF14, on CD4⁺ T cells, with BTLA, on APCs, induces Treg differentiation.^73,74 In our set of GC samples, TNFRSF14 was mainly expressed on Tregs, which is interesting considering that other members of the same family, such as TNFRSF1B, TNFRSF9, TNFRSF18, and TNFRSF25, have been associated with an increased regulatory function of Tregs.⁷⁵ For example, Zheng et al.⁷⁶ reported that TNFRSF9⁺ Tregs are one of the most influential cells in the TME of several types of cancer, including GC. Besides the role of TNFRSF14 on Treg differentiation, to our knowledge, no other roles in Treg functional properties have been attributed to this molecule. It is tempting to speculate that TNFRSF14 also works as a costimulatory receptor maintaining Treg suppressor functions, similar to other members of the TNF superfamily.

In conclusion, we show that, in addition to PD-L1/PD-1, the expression of other nonclassical IC axes is upregulated in response to H. pylori and in GC lesions. Our observations should serve as the basis for analyzing PVR/TIGIT and TNFRSF14/BTLA axes in greater detail in the context of GC development and progression. The clinical implications of these findings also warrant further investigations, especially when considering that, in some cancer types, combined ICI treatment yields greater clinical benefit than when ICIs are administered alone,^52,53 and H. pylori infection impairs the response to anti–PD-1 therapy.^30,31 Importantly, the abundance of Tregs in GC microenvironments⁷⁰ and their major contribution to the expression of ICs, as shown in our study, highlight the fact that exhausted CD8⁺ T cells are not the only feasible target of ICIs in cancer; this may change our view of how ICIs therapies should be tailored according to the tumor immune context. Although the current knowledge about the TME has translated into new therapies with important clinical benefit for many patients with different types of cancer, studies like ours reflect that the very intricate dynamics of the heterotypic interactions between components of such milieu is still a nebula we do not fully understand. It is indisputable that ICIs have come to revolutionize cancer treatment; however, we must look deeper into the main determinants underlying the conformation of the TME, because these may dictate what can be accomplished with this therapeutic strategy and what is the best target in each type of cancer.

Author contributions

Lucía Figueroa-Protti (Conceptualization [Equal], Formal analysis [Lead], Funding acquisition [Supporting], Investigation [Lead], Methodology [Lead], Project administration [Supporting], Validation [Supporting], Visualization [Lead], Writing—original draft [Equal]), Giovanna Mainieri-Breedy (Data curation [Equal], Methodology [Supporting], Resources [Supporting], Writing—review & editing [Equal]), Vanessa Ramírez-Mayorga (Resources [Supporting], Writing—review & editing [Equal]), Javier Mora (Methodology [Supporting], Resources [Supporting], Supervision [Supporting], Validation [Lead], Writing—review & editing [Equal]), Silvia Molina-Castro (Methodology [Supporting], Resources [Supporting], Supervision [Supporting], Validation [Lead], Writing—review & editing [Equal]), and Warner Alpízar-Alpízar (Conceptualization [Equal], Data curation [Equal], Funding acquisition [Lead], Methodology [Supporting], Project administration [Lead], Resources [Lead], Supervision [Lead], Writing—original draft [Equal])

Supplementary material

Supplementary material is available at ImmunoHorizons online.

Funding

This research was funded by the Vicerrectoría de Investigación of the University of Costa Rica (https://vinv.ucr.ac.cr/) through the projects 810-B9-108, led by L.F.-P., and 810-B9-473, led by W.A.-A. Funding was obtained from Red de Mujeres en Ciencias, Ingenierías y Humanidades of the University of Costa Rica to cover the open access charges.

The funders did not play any role in the study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Conflicts of interest

None declared.

Data availability

The data underlying this article will be shared on reasonable request to the corresponding author.