Mycoplasma Hyorhinis Infection Promotes NF-κB–Dependent Migration of Gastric Cancer Cells

It has been reported that more than 16% of new cancer cases worldwide are attributable to infection (1). The association between Helicobacter pylori and gastric cancer suggests that tumor formation could be initiated by persistent infection of the pathogenic microbes (2). Several other infectious organisms had been investigated for their roles in tumor development, including mycoplasmas (3, 4).

To date, at least 16 mycoplasma species have been isolated from human (3). Mycoplasma hyorhinis (M. hyorhinis), first isolated from the respiratory tracts of pigs (5), can cause polyserositis and arthritis in piglets (6). In addition, M. hyorhinis is one of common contaminants of mammalian cells in cultures, which affects the host metabolic pathways (7, 8). Several studies, including ours, have revealed an association between M. hyorhinis infection and malignancy (9–15). By the ELISA method, M. hyorhinis lipoprotein p37 was shown to be positive in sera of 36% men with benign prostatic hyperplasia and 52% with prostate cancer (10). We used a monoclonal antibody (mAb) PD4 against p37 to examine M. hyorhinis infection and found that the positive rate of M. hyorhinis in gastric cancer was higher than those in chronic superficial gastritis and gastric ulcer (12). However, the prevalence of M. hyorhinis infection in the general population is unknown. In addition, the effect of M. hyorhinis infection on patients' survival remains unclear.

M. hyorhinis could induce cell transformation (13) and increased the invasiveness of gastric cancer cells (11). It also promoted melanoma cell invasiveness (14). The percentage of CD133⁺ cells in human colorectal cancer cell lines was influenced by M. hyorhinis infection (15), indicating that M. hyorhinis affects the stemness of cancer cells.

Early studies demonstrated p37 as a major component of the high-affinity transport system of M. hyorhinis (16, 17), whereas biochemical characterization of this protein is still insufficient. Full-length p37 has a signal peptide of 23 amino acids (16, 17), which was found to be peculiar because of four phenylalanine residues and an untypical cleavage site (17). Following the cleavage site, the C-S-N motif fits the consensus sequence of bacterial lipoprotein and was thought to play a role in the membrane recruitment of M. hyorhinis (16, 17). p37 shares partial homology to the hemagglutinin protein of influenza A, a sialic acid–binding protein critical for viral entry (18), but p37 has no homology to mammalian proteins. Proinvasive function of p37 has been documented (18, 19), but the role of p37 in M. hyorhinis infection and mechanisms underlying M. hyorhinis– and p37-induced malignant phenotypes are unknown.

In this study, we demonstrated that M. hyorhinis infection in gastric cancer tissues predicts metastasis and poor survival. We further uncovered the roles of p37 as well as host Annexin A2 (ANXA2) and EGFR in M. hyorhinis infection and infection-induced cell migration.

The study with clinical samples was approved by the Medical Ethic Committee of Zhejiang Provincial People's Hospital. Two cohorts of formalin-fixed and paraffin-embedded (FFPE) gastric cancer tissues were obtained from archives of the Department of Pathology, Zhejiang Provincial People's Hospital (Hangzhou, Zhejiang, China): 339 collected during 1998 to 2004 (cohort 1), 88 collected during 2013 to 2014 (cohort 2). Written informed consents were obtained from all patients before operation. Specimens were diagnosed histopathologically and staged according to the TNM International Union against Cancer classification system. Data on patients of cohort 1, who had survived until the end of follow-up period, were censored at the date of last contact and their clinicopathologic characteristics were summarized in Fig. 1D. The tissue microarray was constructed as described previously (20). Immunohistochemistry (IHC) was performed to detect p37 in both cohorts with PD4 antibody. The degree of immunostaining was reviewed and scored independently by two pathologist (W. Chen and X. He) based on the intensity of staining: 0 (no staining), 1 (weak staining, light yellow), 2 (moderate staining, yellow brown), and 3 (strong staining, brown). Moderate and strong stainings were defined as M. hyorhinis–positive, whereas no and weak stainings were defined as negative.

DNA was extracted from FFPE tissues (cohort 2) with the QIAamp DNA FFPE Tissue Kit from Qiagen. Alternatively, DNA was isolated from cells by the standard protocol. Quantitative PCR (qPCR) was performed with 30 ng (cell) or 100 ng (tissue) DNA using the StepOne system ABI and SYBR Green Real-Time PCR 2× premix Kit (Takara). The reaction programs and p37-specific primers (forward: 5′-TATCTCATTGACCTTGACTAAC-3′, reverse: 5′-ATTTTCGCCAATAGCATTTG-3′) were reported previously (21). Using DNA from M. hyorhinis–infected and –uninfected AGS cells as references, clinical samples with any recorded threshold cycle number (C_t) value were scored as positive. To compare M. hyorhinis DNA levels in cells, GAPDH (forward primer: 5′-TGAAGGTCGGAGTCAACGG-3′, reverse primer: 5′-CCTGGAAGATGGTGATGGG-3′) was amplified as control and the data were analyzed using the |$2^{ - \Delta \Delta \,C_{\rm t} }$| method. Primers were synthesized by Sangon.

Gastric cancer cell AGS was from the ATCC. Gastric cancer cell MGC803 and immortalized human gastric epithelia GES-1 were kept in Peking University Cancer Hospital and Institute (Beijing, China). Primary human fetal colon fibroblasts CC-18Co was gifted by Drs. Dajun Deng and Baozhen Zhang (Peking University Cancer Hospital and Institute). Human umbilical vein endothelial cell (HUVEC) was gifted by Dr. Chuanke Zhao (Peking University Cancer Hospital and Institute). Primary mouse embryonic epithelia (MEF) and primary mouse gastric epithelia (PMGE) were gifted by Ting Ma (Peking University Cancer Hospital and Institute). PMGE was isolated as previously reported (22). AGS, MGC803, and GES-1 cells were cultured in RPMI-1640 medium. MEF was cultured in DMEM. HUVEC, CC-18Co, and PMGE were cultured in EBM-2 medium. Media were supplemented with 5% to 10% FBS plus antibiotics from Invitrogen. Primary cells were used at passage 3 to 5, whereas cell lines were never used beyond passage 20. The mycoplasma test was performed biweekly by qPCR amplification of M. hyorhinis p37, Hochest 33258 staining, and PCR amplification of mycoplasma genome with a kit from HD Biosciences.

M. hyorhinis (ATCC 17981) was grown for 72 hours at 37°C in a modified Hayflick medium supplemented with 20% heat-inactivated FBS (23). The mycoplasmas were serially passaged for three times and harvested at the mid-exponential phase of growth by centrifugation for 20 minutes at 12,000 × g and stored at −80°C. Titer of M. hyorhinis was quantified as color change units (CCU) per milliliter (24). Cells at 80% confluence were serum starved for 24 hours before addition of 10⁵ CCU/mL M. hyorhinis, which was equal to 0.33 to 1 multiplicity of infection. Heat-inactivation of M. hyorhinis was reported previously (25).

Commercial antibodies were listed in Supplementary Materials and Methods. Anti-p37 mAb PD4 was generated and characterized previously (11, 12, 26). Polyclonal anti-p37 antibody was prepared by immunizing rabbit with GST-p37 fusion protein following the standard protocol. Bay 11–7082 was from Sigma-Aldrich. GST-ricin A chain was gifted by Dr. Xianping Wang (Institute of Biophysics, Chinese Academy of Sciences, Beijing, China). p37-2-23, ANXA2-2-26, and random control peptides were synthesized by SBS Bio. pcDNA3.0-ANXA2 was cloned in our laboratory. Deletion of N-terminus of ANXA2 was performed by PCR. ANXA2 cDNAs (full-length and truncated) were cloned into pET-28a-c and the recombinant proteins were prepared. pcDNA3.1-EGFR was gifted by Dr. Zhijie Chang (Tsinghua University, Beijing, China).

Of note, 96-well plates were seeded with cells (1 × 10⁴/well) or were immobilized with polypeptide for 24 hours, then cell ELISA or solid-phase binding was performed as described previously (27).

Cells (1 × 10⁶/10-cm plate) were infected with M. hyorhinis for 24 hours, harvested, fixed in 2% paraformaldehyde, stained with PD4 (2 μg/mL) at 4°C overnight and FITC-conjugated secondary antibody for 30 minutes, then were subjected to flow cytometry with BD FACSAria.

MGC803 cells were homogenized in lysis buffer (50 mmol/L Tris-HCl pH 8.0, 150 mmol/L NaCl, 1% Triton X-100, 0.5 mmol/L DTT, 1 mmol/L PMSF, and 1x complete protease inhibitors) for 20 minutes. After centrifugation, supernatants were incubated with 2 μg purified GST-p37 or GST (glutathione S–transferase) protein plus 20 μL glutathione sepharose beads (GE Healthcare) at 4°C overnight. The precipitates were resolved by SDS-PAGE and stained by Coomassie brilliant blue G-250. Protein bands precipitated by GST-p37 were subjected to MALDI-TOF mass spectrometry analysis by the Central Laboratory of Peking University Cancer Hospital and Institute.

Cell migration assay was carried out as previously reported (11).

Cells grown on the coverslips were fixed with 4% paraformaldehyde for 30 minutes at 4°C, followed by permeabilization with 0.1% Triton X-100 in PBS for 5 minutes and blocked with 5% BSA at room temperature for 1 hour. Antibodies were then applied to the cells overnight at 4°C, followed by probing with secondary antibodies, counterstaining with DAPI, and mounting in 50% glycerol/PBS. To observe membrane localization of FITC-p37-2-23 peptide, cells were coincubated with peptide plus DiIC18(3)-DS (Beyotime) for 10 minutes. A Leica SP5 confocal system (Leica) was used to observe the localization of indicated proteins or peptides. Digital images were processed with Adobe Photoshop CS (Adobe Systems) by adjusting the linear image intensity display range.

Cells were plated in 6-well plates at 3 × 10⁵ cells per well and transfected with siRNAs plus siRNA mate (both from GenePharma). The target sequences of siRNAs were listed in Supplementary Materials and Methods. Plasmids were transfected into cells with Lipofectamine 2000 (Invitrogen).

Data analysis was performed using SPSS 13.0 (SPSS, Inc.). A standard χ² test was performed to assess the association between p37 and clinicopathologic characteristics. Survival curves were estimated using the Kaplan–Meier method and compared with the log-rank test. A two-tailed Student t test was used to determine the significance of differences. Values represent mean ± SD. A P value of less than 0.05 was considered statistically significant.

To determine the association between M. hyorhinis infection and clinicopathologic parameters in gastric cancer, we examined the p37 protein expression in two cohorts of samples by IHC. p37 exhibited punctuate distribution in the cytoplasm, especially in the compartment adjacent to cell membrane (Fig. 1A and Supplementary Fig. S1A). Immunofluorescence staining was also performed in a subset of samples to better visualize p37 in tissues (Fig. 1B and Supplementary Fig. S1B). Of note, 39.5% (134/339, cohort 1) and 44.3% (39/88, cohort 2) cases were positive for M. hyorhinis in IHC. To validate the reliability of IHC technique, we performed qPCR with DNA extracted from cohort 2 and found a positive rate of 46.5% (41/88). In addition, 31 of 41 of qPCR-positive samples were IHC-positive, whereas 39 of 47 of qPCR-negative samples were IHC-negative (Fig. 1C). Comparison of the results of two techniques confirmed the reliability of IHC (Fig. 1C), in which specificity was 83.0% and the sensitivity was 75.6%. As for cohort 1, although p37 expression exhibited no correlation with gender, age, tumor size, clinical stage, differentiation, or lymph node metastasis, significant correlations with blood vessel invasion and metastasis were observed (Fig. 1D). The Kaplan–Meier plotting of cohort 1 showed that patients positive for M. hyorhinis had poorer survival than those negative for M. hyorhinis (Fig. 1E), suggesting that M. hyorhinis infection might act as an unfavorable prognostic factor for patients with gastric cancer.

M. hyorhinis could infect human gastric cancer cells and promoted cell invasion (11), and purified p37 alone was sufficient to promote cancer cell invasiveness (18, 19). However, the role of p37 in M. hyorhinis infection remained unclear. We found that live, but not heat-inactivated, M. hyorhinis bound to gastric cancer cells MGC803 and AGS in a time- and dose-dependent manner in the cell ELISA (Fig. 2A and Supplementary Fig. S2A). Purified GST-p37 bound to gastric cancer cells MGC803 and AGS, immortalized gastric epithelia GES-1, and primary cells MEF, PMGE, and CC-18Co; however, GST or GST-ricin A chain, which has similar isoelectric point to p37, had low binding activity (Fig. 2B and Supplementary Fig. S2B), indicating that GST moiety or electrostatic interaction plays minimal role in the binding of GST-p37 to cells. M. hyorhinis could infect both cancerous and noncancerous cells, as detected by qPCR amplification of p37 (Fig. 2C), but heat inactivation abolished this potential (Fig. 2C). A polyclonal anti-p37 antibody blocked the binding of M. hyorhinis to gastric cancer cells in qPCR, cell ELISA, and immunofluorescence staining assays (Fig. 2C–E). Similarly, M. hyorhinis infection of GES-1 and HUVEC cells was blocked by this antibody (Fig. 2C). Results of flow cytometry further validated M. hyorhinis infection and inhibition by anti-p37 in both cancer cells and primary cells (Fig. 2F and Supplementary Fig. S2C). M. hyorhinis promoted gastric cancer cell migration in a dose-dependent fashion (Fig. 2G), whereas heat inactivation or anti-p37 decreased M. hyorhinis–promoted cell migration (Fig. 2G). These results indicate that infection of mammalian cells by M. hyorhinis is p37 dependent.

p37 was reported to localize on the surface of mammalian cells (16), but the mechanism is elusive. Its N-terminal region (amino acids 2–23) was predicted to be hydrophobic (16), and has limited homology to proteins of other mycoplasma species (Supplementary Fig. S3). We found that GST-p37-2-23 bound to MGC803 and AGS cells, but GST-p37-Δ2-23 had no binding (Fig. 3A). The immunofluorescence assays revealed the colocalization between FITC-p37-2-23 and the membrane indicator DiIC18 (3)-DS (Fig. 3B). In addition, application of a synthesized p37-2-23 peptide blocked the M. hyorhinis infection of AGS, MGC803, GES-1, and PMGE cells, as evaluated by qPCR (Fig. 3C). These results suggest that p37-mediated M. hyorhinis infection requires its N-terminal region.

To identify the host-binding partner(s) of p37, we performed pull-down assay and identified a 36-kD protein precipitated by GST-p37 from MGC803 cell lysates, which was characterized as ANXA2 (Fig. 4A). Coimmunoprecipitation assay confirmed the binding of p37 to endogenous ANXA2, but not to annexin family members ANXA1 or ANXA4 (Fig. 4B). The immunofluorescence assay further showed the colocalization of p37 and ANXA2 in M. hyorhinis–infected AGS cells (Fig. 4C). Using pull-down assays, we found that p37 that lacked its N-terminal region failed to precipitate ANXA2, whereas this region alone was sufficient to interact with ANXA2 (Fig. 4D), suggesting that the N-terminal region of p37 mediates p37-ANXA2 interaction.

ANXA2 belongs to the annexin family of calcium-binding proteins (28). The unique N-terminal domains of annexins determine their distinct functions (28). The N-terminal domain of ANXA2 (amino acids 2–26) is conserved across vertebrates and mediates its interactions with other proteins (28). We found that His-ANXA2 lacking this domain failed to be precipitated by GST-p37 (Fig. 4D). Furthermore, the results of solid-phase–binding assay indicated that the N-terminal region of p37 directly interacted with the N-terminal domain of ANXA2 (Fig. 4E), and p37-2-23 peptide blocked interaction between GST-p37 and His-ANXA2 (Fig. 4F).

Next, we found that M. hyorhinis infection was blocked by an anti-ANXA2 antibody in the immunofluorescence staining assay and qPCR assay (Fig. 5A and B). Binding of GST-p37 to MGC803 and AGS cells was also reduced by anti-ANXA2 antibody (Supplementary Fig. S4A and S4B). Anti-ANXA2 antibody antagonized M. hyorhinis– and GST-p37–promoted cell migration (Fig. 5C and Supplementary Fig. S4C). In addition, when ANXA2 was knocked down by siRNAs targeting the noncoding regions (Fig. 5D), infection of M. hyorhinis and bindings of GST-p37 to cells were lowered (Fig. 5E, Supplementary Fig. S4D and S4E). Conversely, expression of ectopic ANXA2 potentiated M. hyorhinis infection (Fig. 5F). Moreover, knockdown of ANXA2 inhibited M. hyorhinis–promoted cell migration (Fig. 5G). These results indicate that ANXA2 is a host receptor mediating M. hyorhinis infection.

Tyr23 phosphorylation of ANXA2 is associated with its membrane localization under stress conditions (29). Upon M. hyorhinis infection, Tyr23 phosphorylation of ANXA2 was upregulated, whereas total ANXA2 was unaltered (Fig. 6A). The membrane ANXA2 was increased and cytosolic ANXA2 was decreased when the cells were exposed to M. hyorhinis (Supplementary Fig. S5A). Treatment with GST-p37 could induce ANXA2 phosphorylation (Supplementary Fig. S5B), suggesting that p37 is critical for ANXA2 phosphorylation upon M. hyorhinis infection.

EGFR plays a role in regulating ANXA2 phosphorylation and localization (30, 31). By coimmunoprecipitation assay, we found that p37 interacted with both ANXA2 and EGFR (Fig. 6B). Colocalization of these proteins was observed in M. hyorhinis–infected cells (Fig. 6C). M. hyorhinis infection potentiated EGFR–ANXA2 association and EGFR preferentially interacted with phosphorylated ANXA2 (Fig. 6D). When EGFR was knocked down (Fig. 6E), M. hyorhinis–induced ANXA2 Tyr23 phosphorylation and membrane localization of ANXA2 were decreased (Fig. 6F and Supplementary Fig. S5C). M. hyorhinis infection was decreased by knockdown of EGFR, but was increased by expression of ectopic EGFR (Fig. 6G). Furthermore, M. hyorhinis–promoted migration was inhibited by knockdown of EGFR (Fig. 6H). Therefore, EGFR could facilitate M. hyorhinis infection likely by enhancing ANXA2 phosphorylation.

To explore the mechanisms of M. hyorhinis–promoted invasiveness, we performed microarray analysis and the NF-κB pathway was predicted to be activated in the infected cells (Supplementary Fig. S6A and S6B). Quantitative RT-PCR analysis showed increased expression in 5 of 6 of NF-κB target genes (IKBA, COX2, MMP1, PRDM1, SOCS2, MAP2K1) by M. hyorhinis infection (Fig. 7A). Consistently, S536 phosphorylation of NF-κB p65 was upregulated and p65 tended to be accumulated in the nuclei of M. hyorhinis–infected cells, supporting the activation of NF-κB (Fig. 7B and C). Paralleled with enhanced p65 phosphorylation, EGFR phosphorylation was elevated by M. hyorhinis infection (Fig. 7B). When ANXA2 or EGFR was knocked by siRNA, M. hyorhinis–induced Ser536 phosphorylation of p65 was attenuated (Fig. 7D), implying that NF-κB signaling is downstream of ANXA2 and EGFR in the context of M. hyorhinis infection. When cells were pretreated with the NF-κB signaling inhibitor Bay 11–7082, M. hyorhinis– and p37-induced cell migrations were significantly inhibited (Fig. 7E), suggesting that activated NF-κB is responsible for M. hyorhinis–induced cell invasiveness.

In this study, we revealed that M. hyorhinis infection correlated with metastasis and poor survival of patients with gastric cancer in cohort 1, confirming the previous proposal that M. hyorhinis could be a risk factor for gastric cancer (9–15) We used both IHC and qPCR techniques to detect M. hyorhinis infection in FFPE gastric cancer tissues. By using qPCR and freshly isolated DNA samples, M. hominis infection was shown to be associated with prostate cancer (32). However, extracting mycoplasma DNA from formalin-fixed tissues is challenging. There would be a quick decline in the success of PCR for large fragment after storage in FFPE after 1 to 2 years (33). Moreover, formalin can introduce damages to AT-rich regions of DNA (34), whereas M. hyorhinis genome has an A-T content of more than 74% (35). For these reasons, it is hard to validate the IHC results by qPCR for samples of cohort 1, which were collected 10 to 16 years ago. This issue was resolved by using cohort 2, which was collected for less than 1 year. Statistical analysis of M. hyorhinis–positive rate and clinicopathologic factors indicated that results calculated from IHC of cohort 1 were consistent with those from qPCR of cohort 2 and IHC of cohort 2. By this approach, we indirectly validated the results of IHC-based survival prediction. However, we could not exclude the effects of recent mycoplasma colonization and/or antibiotics treatment on the results of IHC and qPCR. It should be pointed out that we previously found M. hyorhinis infection correlates with clinical stage and differentiation of tumor (12), which was not supported by the present study. The inconsistence could be resulted from difference in sample size and/or source of samples. Thus, multicentric studies using multiple techniques and more samples are required to further verify the association between M. hyorhinis infection and gastric cancer.

We found M. hyorhinis could infect both primary cells and cancerous cells, whereas results of qPCR and flow cytometry suggest that primary cells are less vulnerable to be infected by this microbe. We further identified ANXA2 as a host factor mediating M. hyorhinis infection. ANXA2 plays multiple roles in cancer development (28, 36, 37). Recent studies disclosed the relationship between ANXA2 and microbal infection. ANXA2 contributes to HPV16 infection (31), and is also involved in the formation of hepatitis C virus (HCV) replication complex on the lipid raft (38). Our findings demonstrate that ANXA2 contributes to M. hyorhinis infection through its interaction with p37 at their N-termini. Potential inhibitors targeting the binding interface between p37 and ANXA2 may provide therapeutic opportunities to combat M. hyorhinis infection.

ANXA2 phosphorylation at Tyr23 is associated with its cell surface translocation (29). In this study, Tyr23 phosphorylation was found to be enhanced and the ANXA2 on cell surface was increased upon M. hyorhinis infection in an EGFR-dependent manner. It is noteworthy that EGFR plays some roles in microbial infection. EGFR is a cofactor for HCV entry, whereas kinase inhibitors targeting EGFR reduce HCV infection (39). EGFR and HER2 function together as receptors for C. albicans (40). It is also a receptor for HCMV (41). Alternatively, EGFR signaling mediates infection-induced malignant phenotype. For example, Helicobacter pylori upregulates EGFR in AGS cells and promotes migration by activating the EGFR–PI3K pathway (42, 43). Our present results again underscore the essential role of EGFR in microbial infection. We noticed an increased ANXA2–EGFR interaction and EGFR preferentially interacts with phosphorylated ANXA2 in M. hyorhinis–infected cells, the mechanisms underlying these alterations need to be explored in future studies.

In addition, we showed that M. hyorhinis activates NF-κB signaling, which is required for M. hyorhinis–induced gastric cancer cell migration. Considering the critical role of the NF-κB pathway in inflammation-related carcinogenesis (44), our results provide another potential approach for counteracting M. hyorhinis-related malignant phenotypes.

No potential conflicts of interest were disclosed.

Conception and design: H. Duan, L. Qu, X. He, C. Shou

Development of methodology: H. Duan, L. Chen, L. Qu

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): H. Duan, M. Ye, X. He

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): H. Duan, L. Chen, L. Qu, S.W. Song, Y. Han, M. Ye, X. He

Writing, review, and/or revision of the manuscript: H. Duan, L. Qu, S.W. Song, M. Ye, X. He, C. Shou

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): L. Chen, H. Yang, W. Chen, C. Shou

Study supervision: L. Qu, W. Chen, C. Shou

The authors thank Dr. Dajun Deng, Dr. Bin Dong, Dr. Caiyun Liu, Ting Ma, Lin Meng, Dr. Zhihua Tian, Lixin Wang, Dr. Baozhen Zhang, Dr. Chuanke Zhao (all from Peking University Cancer Hospital and Institute), Dr. Zhijie Chang (Tsinghua University), and Dr. Xianping Wang (Institute of Biophysics, Chinese Academy of Sciences) for providing critical reagents or technical assistance. Transcript profiling: Genechip accession number is GSE52638.

This work was supported by National Natural Science Foundation of China (no. 91029713) and National Basic Research Program of China (no. 2010CB529303).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.