Overexpression of the human cytomegalovirus UL111A is correlated with favorable survival of patients with gastric cancer and changes T-cell infiltration and suppresses carcinogenesis

Gastric cancer (GC) remains an important type of cancer worldwide; more than 1,000,000 new cases were reported in 2018, with an estimated 783,000 deaths, making GC the fifth most frequently diagnosed cancer and the third leading cause of cancer-related deaths worldwide (Bray et al. 2018). Adjuvant chemotherapy after curative resection for GC has been verified to have survival benefits, but unfortunately, disease prognosis remains poor (Mari et al. 2000; Cunningham et al. 2006). The 5-year overall survival (OS) for GC is generally 25–30% (Allemani et al. 2015). Thus, gaining a better understanding of tumorigenesis and developing new diagnostic and therapeutic strategies for GC are still urgently needed to improve clinical prognosis for these patients.

Previous studies have shown that human cytomegalovirus (HCMV) is related to various cancers such as glioblastoma, medulloblastoma, colorectal cancer, breast cancer, prostate cancer, and lymphoma(Cobbs et al. 2002; Baryawno et al. 2011; Harkins et al. 2002, 2010; Samanta et al. 2003; Mehravaran et al. 2017). However, the role of HCMV infection in tumors is still controversial (Herbein 2018). On one hand, HCMV infection can promote tumor development (Mossman et al. 2013; Teo et al. 2017), whereas, on the other hand, some reports also indicated that HCMV infection can inhibit the transformation process in cancer cells (Oberstein and Shenk 2017; Kumar et al. 2016). Our group previously confirmed that HCMV infection is associated with the development of GC (Zhang et al. 2017; Jin et al. 2014), but the mechanism underlying its effect on the occurrence and development of GC is still not clear.

HCMV belongs to the Betaherpesvirinae subfamily and contains a double-stranded DNA genome, 236 kbp in size (Dioverti and Razonable 2016). The UL111A gene was first identified in the complete DNA sequence of AD169 (Chee et al. 1989) and is the only gene in the HCMV genome that encodes a cellular cytokine homolog (McSharry et al. 2012). The cmvIL-10 transcript was originally discovered by two research groups, expressed during the productive phase of infection in MRC-5 cells infected with the Towne strain and HEL299 cells infected with the AD169 strain (Kotenko et al. 2000; Lockridge et al. 2000). The LAcmvIL-10 transcript was initially identified during latent HCMV infection in human granulocyte–macrophage progenitors (GMPs) infected with the Toledo strain (Jenkins et al. 2004). Subsequent analysis revealed that the LAcmvIL-10 transcript is also expressed in productively infected human foreskin fibroblasts (HFFs) (Jenkins et al. 2008a, b). In addition to cmvIL-10 and LAcmvIL-10 transcripts, five other transcripts have also been reported in AD169-infected MRC-5 cells, but their biological function was not reported (Yi-Ling et al. 2008). However, to date, there has been no research regarding the function of UL111A in GC. Therefore, extensive work on the UL111A gene is essential.

In this study, we investigated the expression profiles of UL111A in GC tissues. We also investigated the clinical significance of the UL111A protein in GC patients. The effects of LAcmvIL-10 and cmvIL-10 on GC cell growth and metastasis were also assessed by overexpression experiments in vitro. This study thus provides a detailed understanding of UL111A and its functions as a novel therapeutic target for GC.

In this study, we reported for the first time that UL111A is expressed in GC tissues. Previous studies have suggested that viral interleukin 10 (vIL-10) can be detected in the peripheral blood of healthy blood donors, which suggests that it might play a key role in sensing or modifying the host environment during latency (Young et al. 2017). We then uncovered that LAcmvIL-10 might be the major transcript of UL111A in GC tissues compared with the cmvIL-10 transcript. In addition, an unspliced isoform was detected most often in these tissues. This is the first report that shows the expression profiles of UL111A in GC tissues. Previous studies have analyzed the functions of cmvIL-10, which was first identified during lytic infection (Kotenko et al. 2000; Lockridge et al. 2000). However, there are few studies showing the biological function of LAcmvIL-10, which is expressed during latency. Studies have suggested that LAcmvIL-10 plays a role in regulating cellular miRNA expression and that this affects the levels of specific secreted cellular proteins (Fu et al. 2014).

Our IHC results also indicated that UL111A was completely distributed in the cytoplasm of GC epithelial cells and that patients with high UL111A expression had better OS and DFS than those with low UL111A expression. Remarkably, our study is the first to reveal the predictive value of UL111A with respect to postoperative ACT. Specifically, patients with high UL111A protein expression levels had superior survival after postoperative ACT.

Studies on tumor immunity have provided a wealth of information about tumor immune surveillance status and immunotherapy options (Fridman et al. 2017; Gajewski et al. 2013). Therefore, we investigated the correlation between UL111A protein levels and CD3⁺, CD4⁺, CD8⁺, and Foxp3⁺ infiltrating T-cell density (cells/mm²). High UL111A protein expression was positively correlated with CD4⁺, CD8⁺, and Foxp3⁺ T cells and patients with high-density infiltrating T cells had significantly longer survival than those with the corresponding low-density cells, indicating that higher UL111A protein might result in the recruitment of these infiltrating T cells. Previous evidence demonstrated that tumor-infiltrating lymphocytes, as well as Foxp3+ T cells, correlate with good prognosis in GC (Lee et al. 2008, 2018; Feichtenbeiner et al. 2013). However, the mechanism underlying this association between UL111A and immune context requires further investigation. cmvIL-10 consists of three exons that encode a protein of 175 amino acids; however, it shares only 27% amino acid homology with human IL-10 (hIL-10), so its binding capacity to IL-10R1 is stronger than that of hIL-10 (Jones et al. 2002). Some studies have shown that IL-10 plays a role in immunosuppression (Saraiva and O’Garra 2010; Ouyang et al. 2011). Similar to hIL-10, cmvIL-10 also has several immunosuppressive properties including the ability to decrease major histocompatibility complex (MHC) class I and MHC class II expression on monocytes, inhibit the production of pro-inflammatory cytokines by lipopolysaccharide-stimulated peripheral blood mononuclear cells (PBMCs), monocytes, and monocyte-derived DCs through the IL-10 receptor, decrease matrix metalloproteinase activity in endothelial cells and cytotrophoblasts, and impair endothelial cell migration and cytotrophoblast invasiveness in vitro (Slobedman et al. 2009; Spencer et al. 2002; Spencer 2006; Raftery et al. 2004; Yamamoto-Tabata et al. 2004). In addition, through surface IL-10R expression in MDA-MB-231 and MCF-7 breast cancer cells, cmvIL-10 protein was found to promote their proliferation, migration, and metastatic potential (Valle Oseguera and Spencer 2014; Bishop et al. 2015). LAcmvIL-10, alternatively spliced from the viral UL111A gene, consists of 139 amino acids, has a C-terminal truncation, and can be identified during both latent and productive infection (Jenkins et al. 2004; Jenkins et al. 2008a, b). Similarly, like cmvIl-10, LAcmvIL-10 can also suppress MHC class II expression by primary human myeloid progenitor cells and monocytes, but the mechanism might differ from that of cmvIL-10 (Jenkins et al. 2008a, b; Poole et al. 2014; Cheung et al. 2009). Unlike cmvIL-10, LAcmvIL-10 does not trigger the phosphorylation of Stat3, and its ability to downregulate MHC-II was not blocked by neutralizing antibodies to hIL-10R, suggesting that LAcmvIL-10 either does not engage cellular IL-10R or that it utilizes it in a manner different from that utilized by cmvIL-10 (Jenkins et al. 2008a, b).

The modulation of secreted cellular proteins by HCMV during latent infection and productive infection has been reported. During latency, LAcmvIL-10 modulates cellular IL-10 and CCL8 secretion through changes in the cellular microRNA hsa-miR-92a (Poole et al. 2014). The authors also showed that human cytomegalovirus latency alters the cellular secretome, which can first result in the recruitment of CD4⁺ T cells and then the inhibition of their antiviral effector functions, thereby aiding the maintenance of latent infection in the face of the host immune response (Mason et al. 2012). As is known, both CD4⁺ and CD8⁺ T cells play critical roles in the cell-mediated control of HCMV infection and clinical disease (Sylwester et al. 2005). Previous studies also suggested that latency-associated UL111A could promote MHC class II downregulation and inhibit CD4 + T cell activation (Cheung et al. 2009). Using a UL111A-deletion virus, one study found that the role of this gene is to downregulate the CD4⁺ T cell recognition of latently infected cells, which probably enhances the capacity of HCMV to persist in a latent state within the human host (Cheung et al. 2009). Therefore, we considered that this mechanism might also exist in GC patients, which needs to be further explored. Still, our data demonstrated that the UL111A, which encodes cmvIL-10 and LAcmvIL-10, was shown to effectively inhibit the malignancy of GC cells. Recently, one report suggested that cmvIL-10 rapidly and independently augments NK cell cytotoxicity through most activating receptors, which may benefit HCMV persistence or contribute to NK cell function against viral infections and cancers (Holder and Grant 2019). Therefore, we suppose that the function of UL111A on GC may be due to innate immune NK cells that play a part of the function, which requires further in vivo experiments to determine.

Next, we further detected UL111A transcripts in GC cell lines infected by the AD169 and Merlin virus (supplementary Fig. 4). As previously reported in MCF-7 and MDA-MB-231 breast cancer cell lines, cmvIL-10 protein showed a tumorigenic effect (Valle Oseguera and Spencer 2014; Bishop et al. 2015). Other researchers also found that cmvIL-10 protein could enhance the migration of glioma cancer stem cells (Dziurzynski et al. 2011). However, we found that exogenous cmvIL-10 had no significant effect on GC cell lines (supplementary Fig. 5), whereas the overexpression of LAcmvIL-10 and cmvIL-10 significantly inhibited the growth and malignancy of GC cells. The mechanism underlying this association between LAcmvIL-10 or cmvIL-10 and such anti-tumor effects still needs further investigation. Our study also has some limitations. First, the number of samples for detecting UL111A transcripts in GC tissues is too few and there may be other undetected transcripts. Second, considering the mechanism of tumor immunity, the correlation between the expression of markers such as PD-L1, PD-1 or CTLA-4 and the expression level of UL111A protein may require further study. Third, we carried out the present study at the in vitro level. However, the oncological properties of UL111A-deletion virus on GC cells and experiments at the animal level remain to be further explored and verified.

In summary, our study showed that high UL111A expression changes T-cell infiltration and is associated with better patient survival. Most importantly, high UL111A expression was found to be an independent prognostic factor of cumulative survival and might predict superior response to postoperative ACT in patients with GC. However, further prospective studies as well as more detailed mechanistic studies are needed.