Human cytomegalovirus infection is correlated with enhanced cyclooxygenase-2 and 5-lipoxygenase protein expression in breast cancer

Breast cancer (BC) is the most common cancer in women worldwide and although implementation of novel therapies and early detection has improved quality of life and decreased mortality rates for some BC subtypes, the overall incidence of BC continues to increase (Jemal et al. 2011). Of high concern, the overall survival has not improved significantly albeit new therapies for patients with metastatic BC have been introduced during the past decades (https://​www.​kreftregisteret.​no/​globalassets/​publikasjoner-og-rapporter/​cancer-in-norway/​cancer_​in_​norway_​2009.​pdf), suggesting that alternative treatment options are urgently needed. Somatic mutations in breast cancer-associated genes are today considered to be one of the main causes and drivers of BC. In addition, epigenetic and genetic alterations in key breast cancer genes such as BRCA-1/2 and their downstream regulation of DNA repair mechanisms, or in genes with impact on epithelial cell proliferation, differentiation and migration are suggested to promote breast cancer carcinogenesis (Petrucelli et al. 1993). Emerging evidence also highlights a role of the tumor microenvironment to induce pre-malignancies and promote tumor progression. Stroma cells like specialized mesenchymal cells and myo-fibroblasts, endothelial cells and infiltrating leukocytes with capacity to produce inflammatory factors may promote breast cancer carcinogenesis and affect tumor progression and metastasis formation (Balkwill et al. 2005; Coussens and Werb 2002; de Visser et al. 2006). Inflammatory cells present in the tumor microenvironment support tumor growth and enhanced malignancy, by releasing cytokines, chemokines and growth factors. Their increased production of reactive oxygen species also further enhance inflammation, oxidative DNA damage and impair DNA repair mechanisms (Coussens and Werb 2002). Therefore, chronic inflammation is considered a risk factor for breast cancer development, and added inflammation to their revised version of the Hallmarks of Cancer (Hanahan and Winberg 2011).

Eicosanoids are potent inflammatory mediators that are produced from arachidonic acid by cyclooxygenases (COXs) and lipoxygenase (LO). Enhanced inflammation via COX-2 and 5-LO promotes tumorigenesis (Chen and Smyth 2011) and is associated with poor prognosis of several cancer forms (Dreyling et al. 1986; Hennig et al. 2002; Steele et al. 1999). Over-expression of COX-2 was observed in 40% of patients with invasive breast carcinoma and correlated with poor prognosis (Denkert et al. 2003; Howe 2007). Selective COX-2 inhibitors reduce the risk of breast cancer, suppress breast cancer cell migration and invasion, and exhibit strong anti-neoplastic effects in animal models (Wang and DuBois 2010; Harris et al. 2006, 2014). COX-2 inhibitors are, therefore, currently evaluated as potential new cancer therapies, especially in epithelial-derived malignancies. A randomised phase II study of 111 postmenopausal women with advanced breast cancer indicated a trend towards a longer duration of clinical benefit in the combination arm (median, 96.6 weeks vs 49 weeks) compared to the exemestane monotherapy arm (Dirix et al. 2008). Inhibition of 5-LO activity in several breast cancer cell lines also resulted in growth inhibition and enhanced apoptosis, but 5-LO inhibitors have not yet been evaluated in clinical trials in breast cancer patients (Avis et al. 2001).

Human cytomegalovirus (HCMV) is a ubiquitous herpesvirus that is proposed to be highly prevalent in different cancer forms including breast (Harkins et al. 2010; Taher et al. 2013, 2014), colon (Dimberg et al. 2013; Harkins et al. 2002; Tafvizi and Fard 2014) and prostate cancer (Samanta et al. 2003), glioblastoma multiforme (Cobbs et al. 2002; Lucas et al. 2011; Rahbar et al. 2015), medulloblastoma (Baryawno et al. 2011) and neuroblastoma (Wolmer-Solberg et al. 2013). HCMV is also present in sentinel lymph node metastases of breast cancer and in brain metastases originating from breast and colon cancer, while healthy tissues surrounding HCMV positive primary tumors are generally HCMV negative. In breast cancer, HCMV was detected in 92% of primary tumors, in 94% of sentinel lymph nodes, and in 99% of brain metastases originating from breast cancer (Taher et al. 2013, 2014). These observations suggest a concerning association between HCMV and malignant cell growth. However, it remains to be clarified whether these findings represent an epiphenomenon or indicate that HCMV infections induce tumor-promoting mechanisms of relevance for local tumor progression.

A close link between HCMV infections and inflammation has been established in the literature. Thus, latent HCMV can be reactivated by inflammation and the virus further promotes inflammation by inducing expression of both COX-2 and 5-LO (Benard et al. 2014; Qiu et al. 2008; Hooks et al. 2006). Results from a mouse model based on virus-induced adenocarcinoma (Bongers et al. 2010), confirmed production of inflammatory factors such as IL-10, TGF-β, IL-1β, IL-8, IL-6, MIP-1α, MIP-1β, and RANTES (Bodaghi et al. 1998; Kotenko et al. 2000; Reeves et al. 2005; Söderberg-Nauclér et al. 1997) to be involved. We earlier showed that COX-2 is almost exclusively expressed by HCMV-infected cells in medulloblastoma and that COX-2 inhibition acts to inhibit HCMV, which led to decreased tumor growth in an animal model (Baryawno et al. 2011; Schroer and Shenk 2008). In addition to inducing inflammation, HCMV may directly modulate tumor cells by affecting intracellular pathways involved in cell cycle regulation, epigenetic regulation of gene expression, cellular invasion, angiogenesis, immune evasion, and apoptosis (Cinatl et al. 2004a, b; Slinger 2010; Vossen et al. 2002), all highly relevant in tumor biology.

The main objective of this study was to investigate a potential association between HCMV infection and simultaneous expression of COX-2 and 5-LO in BC. We examined the expression of HCMV proteins, COX-2 and 5-LO, in tissue samples obtained from BC and in adjacent normal breast tissues and investigated whether the activity level of HCMV was associated with inflammatory markers and impaired clinical outcome. In additional in vitro experiments, we assessed whether HCMV could affect the expression of 5-LO and COX-2 in well-established BC cell lines.

The presence of HCMV, as well as other herpesviruses in different types of cancers has been studied by several groups with variable results (Cianchi et al. 2006; Rahbar et al. 2017). Several studies have reported a very high prevalence of HCMV in breast cancer and linked the virus activity levels to overall survival (Taher et al. 2013; Tsai et al. 2007). Consistent with these observations, we found HCMV protein expression in all breast cancer tissue specimens examined. Moreover, HCMV IE protein expression was extensive (defined as > 50% positive cells) in 72% of breast cancer samples, but only in 8% of adjacent non tumor tissues. These observations give support for a potential role of HCMV in breast cancer progression (Kumar et al. 2018; Rahbar et al. 2017). As it seems, HCMV infections can provide direct effects on tumor cells by expression of viral proteins that affect tumor biology-relevant mechanisms and by virus-induced inflammation. In fact, HCMV seems to contribute to all established Hallmarks of Cancer (Soroceanu and Cobbs 2011).

In support of this statement, we detected extensive expression of the inflammatory factors COX-2 and 5-LO in infiltrating breast cancer (58%, 53%, respectively), but less pronounced activity of these inflammatory mediators in adjacent non-tumor breast tissues (8% and 31%, respectively). These inflammatory factors were expressed in the tumor and in inflammatory cells. We also found a concordant extensive expression of COX-2, 5-LO, and HCMV-IE in breast cancer, suggesting that HCMV activity on tumor cells may be linked to COX-2 and 5-LO expression. This hypothesis was further supported by in vitro experiments demonstrating virus-induced transcript of both COX-2 and 5-LO and COX-2 protein expression in HCMV-infected MDA-231 and MCF-7 cells. However, we did not observe higher expression of COX-2 or 5-LO transcripts in HCMV-infected MDA-231 cells, which may be explained by the already high expression levels of COX-2 proteins in the more aggressive cell line MDA-231 that was originally established from a triple negative breast cancer (TNBC). Previous studies have reported high expression levels of COX-2 in TNBCs (Chikman et al. 2014; Mosalpuria et al. 2014), and we recently reported that HCMV protein expression was high in TNBC (Rahbar et al. 2017). Our results confirm previous studies demonstrating increased levels of COX-2, 5-LO, and their metabolites prostaglandins and leukotrienes in HCMV-infected cells both in vitro and in vivo (Maussang et al. 2009; Qiu et al. 2008; Hooks et al. 2006; Bongers et al. 2010).

HCMV-induced COX-2 expression is mediated by the constitutively active viral chemokine receptor homologue US28, which promotes inflammation, angiogenesis, and tumor formation (Maussang et al. 2009). US28 promotes angiogenesis and tumor formation via COX-2 induced production of IL-6, phosphorylation of STAT-3, and activation of nuclear factor-kappaB (Maussang et al. 2009). High COX-2 and 5-LO expression and production of inflammatory mediators will also attract inflammatory cells capable of releasing inflammatory factors and reactive oxygen species causing inflammation, oxidative DNA damage, and impaired DNA repair mechanisms (Coussens and Werb 2002). In a mouse model with targeted expression of US28 in colon, mice developed inflammation associated adenocarcinoma (Bongers et al. 2010). Consistent with an important role for the virus to induce COX-2 expression, COX-2 inhibition obstructs HCMV replication (Zhu et al. 2002) and reduces growth of HCMV positive tumors in animal models (Baryawno et al. 2011; Maussang et al. 2009). In addition, combined antiviral therapy targeting HCMV (Valganciclovir) and COX-2 (Celecoxib) prevents HCMV replication and PGE2 production, further reduced meduloblastoma and neuroblastoma cell proliferation in vitro (Baryawno et al. 2011; Wolmer-Solberg et al. 2013) and tumor size in meduloblastoma xenografts (Baryawno et al. 2011).

Taken together, these observations raise the question of a possible impact of HCMV on tumor progression through induced production of potent inflammatory factors such as COX-2 and 5-LO that through their effects on both tumor cells and the tumor microenvironment may enhance tumor malignancy grade and promote tumor progression.

Epidemiological studies in large patient cohorts have reported that the cancer risk and metastasis development is reduced in patients who are treated with non-selective COX inhibitors (Anderson et al. 2002; Burn et al. 2011; Xu 2002). The critical role of inflammatory COX-2 and 5-LO and their biological products, prostaglandin and leukotrienes in inflammation and their ability to stimulate signalling pathways contributing to angiogenesis, tumor growth, and invasiveness, further highlights them as potential targets for cancer therapy. Notably, inhibitors of 5-LO were efficiently used to block proliferation of breast cancer cells and a lipoxygenase inhibitor administrated to one breast cancer patient successfully reverted multiple brain metastasis (Flavin 2007; Hammamieh et al. 2007; Poeckel et al. 2006). In vitro experiments and animal models using specific COX-2 inhibitors show a reduction in proliferation and invasion of breast cancer cells and an effect on tumor development and growth (Bocca et al. 2011; Na et al. 2013; Silva et al. 2012). Consistent with these findings, specific or non-specific inhibition of COX-2 has successfully been used for prevention and therapy of breast cancer (Arun and Goss 2004). Selective COX-2 inhibitors appear to be more protective but in both cases, with a significant reduction in the breast cancer risk (Ashok et al. 2011). However, the inhibition of both COX-2 and 5-LO would have additive benefits by simultaneously targeting both pathways of arachidonic acid metabolism. Indeed, combined treatment with COX-2 and 5-LO inhibitors showed a stronger effect on tumor cell proliferation and induction of apoptosis in colon cancer cells (Che et al. 2016; Cianchi et al. 2006). However, these compounds may also involve a so far not considered anti-viral effect against HCMV to affect tumor growth (Schroer and Shenk 2008; Zhu et al. 2002). As selective COX-2 inhibitors may confer increased risk of cardiovascular diseases and stroke (Trelle et al. 2011; Back et al. 2012; Martinez-Gonzalez and Badimon 2007; Sibbald 2004), there is a need to develop a non-toxic drug with potential to inhibit COX-2 and 5-LO, with therapeutic action in human cancers (Gautam et al. 2017).

In conclusion, we detected higher grades of HCMV-IE, COX-2 and 5-LO in the majority of BC samples than in adjacent non-malignant tissue specimens and a significant correlation between extensive HCMV-IE, COX-2 and 5-LO protein levels in infiltrating BCs. These results suggest that inflammation driven by COX-2 and 5-LO in human BC might be induced by HCMV in some patients and promote tumor progression. Thus, we suggest that anti-viral therapy targeting HCMV and inhibitors targeting COX-2 and 5-LO should be evaluated as new therapeutic options in selected breast cancer patients.