SEARCH

SEARCH BY CITATION

Abstract

  1. Top of page
  2. Abstract
  3. Materials and Methods
  4. Results
  5. Discussion
  6. Acknowledgments
  7. Disclosure Statement
  8. References

Human leukocyte antigen-G (HLA-G) plays an important role in tumor cell escape. We investigated HLA-G expression and regulatory T cells (Tregs) infiltrates in patients with gastric cancer (GC), analyzed their relationship with clinicopathologic features, and characterized their role in tumor immune escape. We also investigated the plasma soluble HLA-G level and its potential in the diagnosis of GC. Effect of HLA-G on Tregs was further assessed by coculture experiments in vitro. Most interestingly, HLA-G positive expression was detected in GC tissues and it was significantly correlated with the presence of tumor-infiltrating Tregs. Patients with HLA-G positive expression or high Tregs had significantly poorer survival at 5 years after operation. Multivariate analysis indicated that HLA-G positive expression was an independent prognostic factor of GC. The coculture experiment showed overexpression of HLA-G in GC cell lines significantly enhanced the frequency of Tregs when GC cells were directly cocultured with human peripheral blood mononuclear cell. However, this effect disappeared when the indirect coculture system was applied. Some cytokines such as interleukin-6, interleukin-10, and tumor necrosis factor-α significantly changed in the coculture system. Moreover, plasma soluble HLA-G level in GC patients was higher than that in normal controls. Taken together, our results indicated that HLA-G expression was closely associated with tumor progression and involved in tumor evasion by increasing the frequency of infiltrating Tregs locally. Thus, HLA-G might be a promising predictor for disease prognosis and a possible novel target for immunotherapy in GC patients. (Cancer Sci 2011; 102: 1272–1280)

Gastric cancer (GC) is a major public health issue and the second leading cause of cancer-related death worldwide.(1) In spite of progression of comprehensive therapies, which have somewhat improved patients’ outcomes, the prognosis is still unsatisfying in the main. The overall 5-year survival rate after surgical resection is only approximately 40%.(2–4) Therefore, better definition of the pathogenesis of GC and exploration of novel biomarkers that could be used as possible therapeutic targets or prognostic predictors are urgent.

The human leukocyte antigen-G (HLA-G), a newly identified member of the non-classical MHC family, has very little sequence variability and expresses as seven isoforms, including four membrane-bound (HLA-G1 to HLA-G4) and three soluble (HLA-G5 to HLA-G7) forms.(5,6) Human leukocyte antigen-G has been involved in a broad spectrum of pathophysiological situations, such as oncology, autoimmunity, inflammation, infection and transplantation.(7) In tumor biology, HLA-G plays an important role in the escape of tumor cells from host immune surveillance by suppressing functions of various immune cells.(8) However, there have only been a few studies investigating the clinicopathologic and prognostic significance of HLA-G in GC.(9,10)

Previous clinical studies of tumor immunology indicated that HLA-G was closely related to regulatory T cells (Tregs).(11) During tumor development, immune cells infiltrated into tumors and made up a significant component of the multicellular cancer microenvironment, which participated in tumor formation and progression.(12,13) Regulatory T cells, one kind of tumor-infiltrating lymphocyte, played a crucial role in immunological self-tolerance when the immune system eradicated tumor cells.(14) High tumor-infiltrating Tregs were associated with dismal prognosis in a wide range of human malignant tumors.(15–17)

In this study, we evaluated HLA-G expression and Tregs in serial sections from GC tissues, and determined their correlation with clinicopathologic features and possible prognostic significance. In addition, we investigated the effects of HLA-G on Tregs in a GC cell line in vitro and analyzed the diagnostic significance of plasma soluble HLA-G (sHLA-G).

Materials and Methods

  1. Top of page
  2. Abstract
  3. Materials and Methods
  4. Results
  5. Discussion
  6. Acknowledgments
  7. Disclosure Statement
  8. References

Patients and follow up.  A total of 179 patients with primary GC were enrolled in this study. All underwent radical resection without any adjuvant therapy before surgery between May 2002 and June 2005 in the Department of General Surgery, Qilu Hospital, Shandong University (Jinan, China). The tumor specimens and corresponding adjacent normal gastric mucous tissues (>5 cm from the margin of the tumor) were collected and divided into two halves. Half of the specimens were fixed in formalin and embedded with paraffin, and the other half were stored at −80°C until use. The clinicopathologic characteristics of the participating patients are listed in Table 1. For sHLA-G evaluation, plasma from 58 GC patients before operation and 64 age-matched normal individuals was collected at the same department from May 2008 to August 2009. Our study was approved by the Ethics Committee of Qilu Hospital and signed informed consent was obtained from each patient.

Table 1.   Correlation of clinicopathologic variables with HLA-G expression and Tregs in primary GC tumors
VariablesNo. of casesHLA-G expressionP*TregsP*
NegativePositiveLowHigh
  1. *Chi-squared test. NOTE: The postoperative pathological staging was determined based on the International Union Against Cancer (IUCC) for GC. GC, gastric cancer; HLA-G, human leukocyte antigen-G; Tregs, regulatory T cells.

Gender
 Male13167640.89069620.573
 Female4823252325
Age
 ≤60 years10855530.83158500.446
 >60 years7135363437
Tumor size
 ≤5 cm10558470.11459460.088
 >5 cm7432423242
Differentiation
 Well3119120.31415160.875
 Moderate5227252626
 Poor9644525145
Borrmann-type
 I171160.1601160.312
 II3115161912
 III9049414446
 IV4115261823
Invasion depth
 T1141040.0031040.047
 T28451334935
 T34819292127
 T43310231221
Lymph node metastasis
 N05737200.33037200.046
 N16632342640
 N23914252019
 N31771098
Invaded adjacent organs
 No14080600.00179610.012
 Yes3910291326
Clinical stage
 I4030100.00030100.002
 II3421131618
 III4017232218
 IV6522432441

For patients’ follow-up, all patients were evaluated at the clinic every 3–6 months after discharge. The follow-up was completed in all 179 patients until December 2009, and the median follow-up period was 21 months (range, 4–79 months).

Tissue microarray and immunohistochemistry.  All paraffin-embedded tissue samples from GC patients were reviewed histologically by H&E staining, and representative tumor areas were premarked in blocks. Duplicates of 2 mm diameter cylinders were obtained from each block, and two tissue microarray blocks were reconstructed with each one containing 179 cylinders. Tissue microarray sections were used for immunohistochemistry.

Specimens were deparaffinized and rehydrated. After antigen retrieval, primary antibodies against HLA-G mAb 4H84 (Exbio, Prague, Czech Republic) and Foxp3 (Abcam, Cambridge, UK) were used at dilutions of 1:400 and 1:300, respectively, and incubated overnight at 4°C. Immunoglobulin G-conjugated HRP and 3,3-diaminobenzidine tetrahydrochloride (Vector Laboratories, Burlingame, CA, USA) were used to visualize antibody binding. First trimester cytotrophoblast sections were served as the positive control for HLA-G staining.

Human leukocyte antigen-G staining in GC and normal gastric mucous tissues was evaluated by three pathologists who had no knowledge of patients’ clinicopathologic factors and outcomes. Membrane or combined membrane and cytoplasmic expression of HLA-G was interpreted as positive and the immunoreactivity was further graded as follows by the percentage of positively stained cancer cells: −, 0%; +, 1–25%; ++, 26–50%; and +++, >50%. For the evaluation of Tregs, five independent microscopic fields, representing the densest lymphocytic infiltrates, were selected.

Gastric cancer cell lines, HLA-G transfection, human PBMC, coculture.  The study included human GC cell lines AGS, BGC-823, SGC-7901, MKN-45, MKN-28, and HGC-27. Cell lines AGS (poorly differentiated), BGC-823 (undifferentiated), SGC-7901 (moderately differentiated), MKN-45 (poorly differentiated), MKN-28 (moderately differentiated), and HGC-27 (undifferentiated) were purchased from the Culture Collection of the Chinese Academy of Sciences (Shanghai, China) and Riken Cell Bank (Tsukuba, Japan), and maintained in RPMI-1640 (Life Technologies, Paisley, UK) with 10% FBS and antibiotics.

For transfection treatment, SGC-7901 cells were transfected with HLA-G eukaryotic expression vector (pVITRO2-mcs vector-HLA-G) as SGC-7901-G, or pVITRO2-mcs control vector as SGC-7901-V, using Lipofectamine 2000 (Invitrogen, Carlsbad, CA, USA) according to the manufacturer’s instructions. Transfectants were screened with hygromycin B (Roche, Indianapolis, IN, USA).

Human peripheral blood mononuclear cells (PBMCs) were obtained from healthy adult volunteers and separated by Ficoll-Paque Plus (Sigma-Aldrich, St. Louis, MO, USA). For coculture, 5 × 104 SGC-7901, SGC-7901-V, or SGC-7901-G cells were seeded into each well of a 24-well plate. After 24 h, PBMCs (1 × 106) as the target cells were added directly into the wells for direct coculture or into the Boyden chamber insert (Millipore, Bedford, MA, USA) for indirect coculture. After coculture for 24 h, PBMCs were gently rinsed with PBS and harvested for analysis. All supernatants were collected after coculture.

RNA extraction and real-time fluorescent quantitative PCR.  Total RNA was isolated from GC cell lines using TRIzol (Invitrogen). Equal amounts of RNA, measured by spectrophotometer and RNA gel, were used for first-strand cDNA synthesis with SuperScript II (Invitrogen) in a 20 μL reaction. One microliter cDNA product was then subjected to reverse transcription PCR (RT-PCR) with Taq polymerase (Boehringer Mannheim, Mannheim, Germany). Real-time fluorescent quantitative PCR was carried out using a Light Cycler system (Roche Molecular Biochemicals, Mannheim, Germany) and 2× SYBR Premix Ex Taq (Takara, Shiga, Japan) was used to detect PCR products. The comparative threshold cycle (Ct) method (inline image) was used to analyze relative changes in gene expression.

Western blot analysis.  Tumor tissues, corresponding adjacent normal gastric mucous tissues, or GC cell lines, were resuspended in ice-cold radioimmunoprecipitation buffer for lysis. For protein assay, a BCA protein assay kit (Beyotime Biotechnology, Nanjing, China) was used. Approximately 30 μg protein of samples was loaded, separated on 9.0% Tris–glycine–SDS polyacrylamide gels, and then electroblotted onto PVDF membranes. After blocking, the membranes were incubated with HLA-G mAb 4H84 (1:400; Exbio) or anti-β-actin mouse mAb (1:500; Santa Cruz Biotechnology, Santa Cruz, CA, USA) overnight at 4°C. Membranes were further incubated with polyclonal goat anti-mouse HRP-conjugated secondary antibody (Santa Cruz Biotechnology) for 2 h at room temperature. Bands were detected by a chemiluminescence detection kit (Amersham Pharmacia Biotech, Piscataway, NJ, USA). The intensity of each band was quantified using Image J 1.32 software (National Institutes of Health, Bethesda, MD, USA).

Enzyme-linked immunosorbent assay (ELISA).  The sHLA-G in plasma and supernatants was determined by an ELISA kit (Exbio) that measured the sHLA-G level of both sHLA-G1 and HLA-G5. Supernatants from direct and indirect coculture were collected for cytokine detection. Interleukin (IL)-2, IL-4, IL-6, IL-8, IL-10, interferon-γ (IFN-γ), and tumor necrosis factor-α (TNF-α) were measured by ELISA kits (R&D Systems, Abingdon, UK) according to the manufacturer’s instructions.

Flow cytometry.  All fluorescence-conjugated antibodies used in this study were purchased from BD Bioscience (San Jose, CA, USA) except for phycoerythrin (PE)-conjugated anti-HLA-G antibody (Biolegend, San Diego, CA, USA) and anti-Foxp3 PE antibody (eBioscience, San Diego, CA, USA). After blocking with anti-Fc receptor (clone 2.4G2), cell surface markers were identified by specific mAbs conjugated with different fluorochromes. Cells were analyzed by flow cytometry using a flow cytometer (BD FACSCalibur, BD Biosciences). To analyze intracellular proteins, cells were first fixed and permeabilized, then stained with appropriate mAbs using a Cytofix/Cytoperm plus kit (eBioscience). Flow cytometric analysis was carried out using Flowjo software (Tree Star, San Carlos, CA, USA).

Statistical analysis.  Data was mostly presented as the mean ± SEM. The spss software package (version 13.0; SPSS, Chicago, IL, USA) was used for all statistical analysis. Correlations of HLA-G expression or tumor-infiltrating Tregs with clinicopathologic factors were examined by Chi-squared test. Spearman ρ coefficients test was carried out for the relationship between HLA-G expression and tumor-infiltrating Tregs. Survival curves were calculated by the Kaplan–Meier method and survival differences of subgroups were compared by the log-rank test. Cox regression multivariate analysis was carried out to identify independent prognostic factors of predicting survival. A non-parametric test (Kruskal–Wallis test) was applied to analyze the correlation of sHLA-G levels with clinicopathologic factors. Receiver-operating characteristic (ROC) analysis was used to assess the diagnostic capability of sHLA-G for GC. Other data from experiments were analyzed by Student’s t-test or anova wherever appropriate. Difference was considered significant when the P-value was < 0.05.

Results

  1. Top of page
  2. Abstract
  3. Materials and Methods
  4. Results
  5. Discussion
  6. Acknowledgments
  7. Disclosure Statement
  8. References

Human leukocyte antigen-G expression in primary GC lesions and involvement in GC development.  Human leukocyte antigen-G expression was detected in 89 (49.7%) tumor tissues, including 26 cases (14.5%) with weak expression (+), 35 cases (19.6%) with moderate expression (++), and 28 cases (15.6%) with strong expression (+++). However, HLA-G expression was not detected in corresponding adjacent normal gastric tissues and sections incubated with mouse IgG1 (Fig. 1A). In addition, Western blot analysis, testing HLA-G protein in the above tumor tissues, was highly consistent with those observed by immunohistochemistry (Fig. 1B,C).

image

Figure 1.  Correlations between human leukocyte antigen-G (HLA-G) expression in gastric cancer (GC) specimens and patient survival. (A) Immunohistochemical staining of HLA-G expression in normal and cancerous gastric tissues. Representative images show normal, cancer (negative and positive expression) tissues, and cancer tissue microarray. (B) Western blot analyses of HLA-G protein in normal and cancerous gastric tissues. Representative results are shown. (C) The relative quantitation of HLA-G protein. ***< 0.001 versus normal tissues. Data were compiled from three independent experiments in each condition.

Download figure to PowerPoint

Moreover, the association of HLA-G expression with clinicopathological parameters was further analyzed. It showed HLA-G expression was significantly correlated with tumor invasion depth (= 0.003), invaded adjacent organs (= 0.001), and clinical stages (< 0.001) (Table 1), which indicated that HLA-G was involved in tumor development.

Human leukocyte antigen-G expression correlated with poor prognosis in GC patients.  At the time of the last follow-up, 116 patients died: 87 died of cancer-related causes, and 29 died of other causes. Tumor recurred within the follow-up period in 109 (60.9%) of the 179 patients, local recurrence in 58 patients, distant metastasis in 32 patients, both local recurrence and distant metastasis in 19 patients. The 5-year overall, cancer-specific, and disease-free survivals were 35.2%, 42.0%, and 39.1%, respectively. Kaplan–Meier analysis showed that patients with HLA-G expression had a significantly poorer overall, cancer-specific, and disease-free survival than those without HLA-G expression at 5 years after operation (Fig. 2A–C). To evaluate the robustness of the prognostic value of HLA-G, we further divided patients into four groups based on expression levels of HLA-G, and compared the survival differences between these groups. Our data indicated that the survival decreased as the HLA-G expression increased (Fig. 2D–F). Subsequently, Cox regression multivariate analysis was carried out to identify independently prognostic factors, and it showed that HLA-G expression still retained its significance as independently prognostic factors for unfavorable overall, cancer-specific, and disease-free survivals (Table 2).

image

Figure 2.  Prognostic significance of human leukocyte antigen-G (HLA-G) positive expression was assessed in patients with gastric cancer (GC). (A–C) Patients with HLA-G positive expression had a poorer prognosis in terms of overall, cancer-specific, and disease-free survival. (D–F) Survival curves were analyzed among groups with different HLA-G expression.

Download figure to PowerPoint

Table 2.   Univariate and multivariate analyses of prognostic variables
VariablesUnivariate analysisMultivariate analysis
P*PP95% CI*P*95% CI†P95% CI‡P
  1. *Overall survival; †disease-specific survival; ‡disease-free survival. CI, confidence interval; HLA-G, human leukocyte antigen-G; NA, not assessed; Tregs, regulatory T cells.

Gender0.9870.4890.987NANANANANANA
Age0.6930.7770.869NANANANANANA
Tumor size0.1000.1040.050NANANANANANA
Differentiation0.0590.0650.174NANANANANANA
Borrmann-type0.2490.3550.715NANANANANANA
Invasion depth<0.001<0.001<0.0010.739–1.3660.9750.721–1.3980.9820.733–1.4190.909
Lymph node metastasis<0.001<0.001<0.0010.932–1.5230.1620.933–1.5660.1510.813–1.4010.639
Invading adjacent organs<0.001<0.001<0.0011.073–3.0430.0261.154–3.7640.0151.020–3.2170.043
Clinical stages<0.001<0.001<0.0011.131–2.2520.0081.168–2.5310.0061.008–2.0900.045
Tregs0.0200.0380.0350.460–1.2270.2530.417–1.2020.2000.440–1.2140.226
HLA-G expression<0.001<0.001<0.0011.094–3.0400.0211.041–3.1920.0361.187–3.4450.010

Tumor-infiltrating Tregs expressed in HLA-G positive GC tissues and contributed to tumor prognosis.  Tumor-infiltrating Tregs were quantified by counting Foxp3+ positive cells in the same series of GC tissues and the staining intensity of intratumoral Tregs ranged broadly from 0.0 to 60.4 in a high microscopy field. Based on the median number (3.4) as the cut-off value, patients were classified into two groups: Tregs with low numbers, and Tregs with high numbers. As shown in Figure 3(A), sections with negative expression of HLA-G barely showed tumor-infiltrating Tregs, whereas sections with positive HLA-G expression showed large numbers of tumor-infiltrating Tregs. The correlation between HLA-G expression and tumor-infiltrating Tregs was analyzed in Table 3, which indicated a significant positive correlation.

image

Figure 3.  Relation between human leukocyte antigen-G (HLA-G) expression and tumor-infiltrating regulatory T cells (Tregs) in gastric cancer (GC) specimens and the correlation between tumor-infiltrating Tregs and patient survival. (A) Negative HLA-G expression was associated with low Tregs, whereas positive HLA-G expression was associated with high Tregs. Overall (B), cancer-specific (C), and disease-free (D) survival curves of GC patients were analyzed among groups with low and high Tregs, separately. ***< 0.001 versus low Tregs group.

Download figure to PowerPoint

Table 3.   Correlation between HLA-G expression and tumor-infiltrating Tregs in human gastric cancer
TregsHLA-GrP
++++++
  1. Immunoreactivity was graded by the percentage of positively stained cancer cells: −, 0%; +, 1–25%; ++, 26–50%; and +++, >50%. GC, gastric cancer; HLA-G, human leukocyte antigen-G; Tregs, regulatory T cells.

Low745760.575<0.001
High16212822

We also found that the number of tumor-infiltrating Tregs was significantly correlated with invasion depth (= 0.047), lymphatic metastasis (= 0.046), invaded adjacent organs (= 0.012), and clinical stages (= 0.002) (Table 1). Meanwhile, patients with high Tregs had significantly poorer overall, disease-specific, and disease-free survivals than the remaining cases at 5 years after operation (28.7%vs 41.3%, = 0.020, Fig. 3B; 35.7%vs 47.5%, = 0.038, Fig. 3C; 34.5%vs 43.5%, = 0.030, Fig. 3D, log-rank test), which showed that Tregs contributed to poor prognosis in GC patients.

Human leukocyte antigen-G expression in GC cell lines and transfection in SGC-7901 cell line.  Human leukocyte antigen-G expression was examined in a panel of GC cell lines (AGS, BGC-823, SGC-7901, MKN-28, MKN-45, and HGC-27) by real-time fluorescent quantitative PCR. Interestingly, HLA-G expression was weak in all of these cell lines, which differed from that observed in tumor tissues (Fig. 4A,B). These findings were confirmed by flow cytometric analysis (data not shown). To determine the function of HLA-G in GC cell lines, SGC-7901 cells were transfected with pVITRO2-mcs vector-HLA-G (SGC-7901-G) and pVITRO2-mcs vector (SGC-7901-V) as control. Human leukocyte antigen-G expression increased after transfection, further confirmed by real-time fluorescent quantitative PCR and Western blot analysis (Fig. 4C–F).

image

Figure 4.  Human leukocyte antigen-G (HLA-G) expression in various human gastric cancer (GC) cell lines and the level of HLA-G were analyzed after transfection. Analysis of HLA-G mRNA levels in human GC cell lines using RT-PCR (A) and real-time fluorescent quantitative PCR (B). Analysis of HLA-G mRNA levels after transfection using RT-PCR (C) and real-time fluorescent quantitative PCR (D). Analysis of HLA-G protein levels after transfection using Western blot analysis (E) and quantification (F). ***< 0.001. Data were compiled from three independent experiments in each condition. C, SGC-7901 cells alone without transfection; G, SGC-7901 cells transfected with HLA-G; P, positive control; V, SGC-7901 cells transfected with empty vector.

Download figure to PowerPoint

Human leukocyte antigen-G increased CD4+CD25+Foxp3+ Tregs and modulated cytokine production.  The transfected SGC-7901 cells were cocultured with PBMC directly and indirectly for 24 h, and CD4+CD25+Foxp3+ Tregs were analyzed by flow cytometry. There was a significant increase in the percentage of CD4+CD25+Foxp3+ Tregs after PBMC direct coculture with SGC-7901-G rather than with SGC-7901-V (Fig. 5A,B). However, we found that the percentage of CD4+CD25+Foxp3+ Tregs remained unchanged either in indirect coculture with SGC-7901-G or with SGC-7901-V (data not shown).

image

Figure 5.  Effects of human leukocyte antigen-G (HLA-G) on immunologically competent cell and cytokines after coculture. (A,B) Percentage of CD25+Foxp3+ regulatory T cells (Tregs) was analyzed on gated CD4+ cells. *< 0.05. (C,D) Surface expression of CD69 was analyzed on gated CD56+ cells. **< 0.01. (E,F) Soluble HLA-G (sHLA-G) levels in supernatants of direct and indirect cocultures were detected. ***< 0.001,*< 0.05. (G,H) Level of cytokines detected after direct and indirect coculture. *< 0.05. Data were compiled from three independent experiments in each condition. IFN, interferon; IL, interleukin; TNF, tumor necrosis factor.

Download figure to PowerPoint

Furthermore, CD56+ natural killer (NK) cell activation was analyzed by detecting CD69 expression of CD56+ NK cells. The activation of CD56+ NK cells was inhibited after PBMC direct coculture with SGC-7901-G, but it did not occur when cocultured with SGC-7901-V (Fig. 5C,D).

Interestingly, sHLA-G in supernatants of direct coculture with SGC-7901-G strikingly increased, compared to that with SGC-7901-V (Fig. 5E). It rose slightly in supernatants of indirect coculture with SGC-7901-G compared with SGC-7901-V, but sustained a very low level in both groups (Fig. 5F).

To evaluate the possible effects of HLA-G on immune responses, various cytokines of cocultured supernatants were analyzed by ELISA. Notably, levels of IL-6 and IL-10 were significantly higher in supernatants of direct coculture with SGC-7901-G. No significant differences were found between the two groups in terms of IL-2, IL-4, IL-8, or IFN-γ levels (Fig. 5G). The TNF-α level of the SGC-7901-G group was lower than that of the SGC-7901-V group. All cytokines remained unchanged after indirect coculture except that the TNF-α level was lower in the SGC-7901-G group (Fig. 5H).

Diagnostic significance of plasma sHLA-G for GC.  The sHLA-G level in GC patients was significantly higher than that in normal individuals (Fig. 6A). However, no significant correlation was observed between sHLA-G levels and clinicopathologic factors (data not shown). The sHLA-G level among patients with different clinical stages is shown in Figure 6(B). To determine the diagnostic significance of sHLA-G, ROC curve analysis was carried out, and it revealed that the area under the curve was 0.831 and the sHLA-G level at 101.37 U/mL was the clear cut-off value to distinguish GC patients from normal individuals. With this cut-off value, the sensitivity and specificity of sHLA-G for GC was 69.0% and 87.5%, separately (Fig. 6C).

image

Figure 6.  Plasma soluble human leukocyte antigen-G (sHLA-G) levels in plasma of samples. (A) Comparison of sHLA-G level between normal controls and gastric cancer (GC) patients. ***< 0.001 versus controls. (B) sHLA-G levels among patients with different clinical stages. (C) sHLA-G receiver-operating characteristic (ROC) curve. Curve analysis assessed the accuracy of plasma sHLA-G levels for diagnosis of GC.

Download figure to PowerPoint

Discussion

  1. Top of page
  2. Abstract
  3. Materials and Methods
  4. Results
  5. Discussion
  6. Acknowledgments
  7. Disclosure Statement
  8. References

Our study showed that HLA-G was expressed in tumor tissues and involved in tumor development in GC. Interestingly, the positive expression of HLA-G was significantly associated with tumor-infiltrating Tregs. Moreover, both HLA-G expression and tumor-infiltrating Tregs were significantly associated with tumor progression and metastasis. However, neither HLA-G expression nor tumor-infiltrating Tregs were detected in normal gastric mucous. These findings provided evidence that HLA-G played a role in local immune escape, probably by influencing the frequency of tumor-infiltrating Tregs.

We further investigated the prognostic value of HLA-G and tumor-infiltrating Tregs and found that patients with HLA-G expression or tumor-infiltrating Tregs had significantly poor overall, disease-specific, and disease-free survivals after operation. In the multivariate analysis, HLA-G expression maintained its value as an independent prognostic factor for unfavorable survival. These findings suggested that immunohistochemical detection of aberrant expression of HLA-G could be used as a prognostic marker of GC.

Originally identified by its physical function in mediating maternal–fetal immune tolerance,(18,19) HLA-G has recently been found to be involved in immune escape.(13,20) It was preferentially detected in a wide range of malignancies such as colorectal cancer, esophageal cancer, chronic lymphoid leukemia, ovarian cancer, lung cancer, breast cancer, and renal cell carcinoma, which showed that HLA-G expression was related to tumorigenesis and progression.(7,11,20,21,22) Its aberrant expression in malignant cells may provide itself to escape from the host’s immunosurveillance, which is similar to the role of HLA-G antigens in the escape of trophoblasts from maternal allorecognition.(18,23–26)

More and more evidence suggests that the immunosuppressive network in the tumor microenvironment, mainly composed of Tregs and tumor-associated factors, is crucial to tumor progression. Regulatory T cells played an important role in impeding immune surveillance against cancer and hampering the development of effective antitumor immunity.(27) We analyzed changes of Tregs after PBMC coculture with the SGC-7901 cell line transfected with pVITRO2-mcs vector HLA-G. Data showed that the percentage of CD4+CD25+Foxp3+ Tregs significantly upregulated after direct coculture. However, Tregs were essentially unchanged after indirect coculture. It strongly indicated that the effect of HLA-G on Tregs in this GC cell line was probably through cell–cell direct contact. Based on the above analysis, our study found that HLA-G could increase the frequency of Tregs both in vivo and in vitro, which gave a novel viewpoint that HLA-G was involved in tumor evasion by influencing tumor-infiltrating Tregs. Furthermore, we found that activation of CD56+ NK cells was inhibited in the direct coculture system, but was unchanged in indirect coculture. The findings might be strengthened by evidence of the inhibition of NK cell cytolysis after HLA-G transfection in a similar scenario in hepatocellular carcinoma and non-small-cell lung cancer.(28,29)

Cytokine-mediated effects represent another molecular mechanism by which HLA-G can exert its immunosuppression. When PBMC were directly cocultured with SGC-7901-G, we found that levels of IL-6 and IL-10 were significantly increased, whereas the TNF-α level decreased. However, the TNF-α level decreased little in the indirect coculture. The data indicated that HLA-G affected some cytokine production, especially in the direct culture system. Cell–cell contact was crucial in this process.

In addition, GC patients showed a higher level of plasma sHLA-G than healthy individuals, which was consistent with our previous report on colorectal cancer.(30) However, the plasma sHLA-G level was not associated with clinicopathologic factors. The plasma sHLA-G level could be a highly specific and sensitive index to diagnose gastrointestinal cancers.

Our study showed that HLA-G expression in primary GC tissues was significantly related to survival and could be used as an independent prognostic factor. The concrete mechanism of HLA-G in immune escape might influence the frequency of local Tregs through changing the tumor microenvironment by cell–cell contact. A recent study indicated that the HLA-G-derived peptide HLA-G146-154 could be a promising candidate in peptide-based immunotherapy for HLA-A24+ renal cell carcinoma.(31) Thus, using HLA-G as a target molecule in specific immunotherapy against tumors should be considered as a new therapeutic strategy for the treatment of GC.

Acknowledgments

  1. Top of page
  2. Abstract
  3. Materials and Methods
  4. Results
  5. Discussion
  6. Acknowledgments
  7. Disclosure Statement
  8. References

The project was supported by the China National Natural Science Foundation Projects (Grant No. 30672010). The authors are grateful to Dr. Wei-Hua Yan (Taizhou Hospital, Taizhou, China) for providing the recombinant pVITRO2-mcs-HLA-G. The authors also wish to thank Yong-Mei Yang, A-Lei Feng,Yan-Feng Liu, and Shu-Hai Li for their technical guidance and data analysis from Qilu Hospital, Shandong University.

Disclosure Statement

  1. Top of page
  2. Abstract
  3. Materials and Methods
  4. Results
  5. Discussion
  6. Acknowledgments
  7. Disclosure Statement
  8. References

The authors have no conflict of interest to declare.

References

  1. Top of page
  2. Abstract
  3. Materials and Methods
  4. Results
  5. Discussion
  6. Acknowledgments
  7. Disclosure Statement
  8. References