Background
Urothelial carcinoma of the bladder is a common malignancy with 165,000 estimated global human deaths per year [
1]. This cancer is heterogeneous and histologically classified into low-grade, superficial type and high-grade, invasive type. Low-grade tumors are more common (approximately 70%) and are usually associated with a favorable prognosis. In contrast, high-grade, muscle-invasive urothelial carcinoma (iUC) is less prevalent (approximately 30%) but has a high risk of death from distant metastasis [
2]. Cisplatin-based chemotherapy is the standard first-line treatment of metastatic iUC and provides overall survival benefits [
3]. Meanwhile, up to two-thirds of patients cannot tolerate the regimens due to nephrotoxicity [
4]. There have been little advances in the treatment of iUC in the past 20 years. Thus, better management of iUC is urgently required.
Our understanding of various cancers has benefited greatly from comparative oncology using relevant animal models, especially mouse models. Several experimentally induced mouse models of bladder cancer have been established, including carcinogen-based models, engraftment models, and genetically engineered models. However, in contrast to many other cancer types, relatively few mouse models of bladder cancer have been described, and in particular, few models display muscle-invasive or metastatic cancer phenotypes [
5]. Therefore, more suitable iUC animal models are crucial.
Naturally occurring iUC in pet dogs closely resembles human iUC with regard to clinical signs, histopathology, heterogeneity, metastatic behavior, disease progression, and response to cisplatin-based chemotherapy [
6,
7]. Bladder carcinoma makes up approximately 2% of all spontaneous cancers in dogs, which is similar to its incidence in humans [
1,
7]. However, unlike humans, the vast majority of canine bladder carcinomas are high-grade iUC (> 90% of cases); low-grade, superficial tumors are uncommon in dogs. Metastases of lymph nodes or distant regions (e.g., lungs, liver, kidneys) have been reported in approximately 15% of dogs at diagnosis and in 40–50% at death [
6]. Moreover, clinical trials of canine iUC can be conducted in a relatively short period because dogs have a shorter life span than humans. Thus, dogs with iUC offer an ideal model to evaluate novel anti-cancer drugs. However, canine iUC has not been well characterized at the molecular level. Only one study, based on microarray analysis, has shown similarities in gene expression patterns between dogs and humans with iUC [
8]. Although this previous study revealed the involvement of certain genes (TP53 and EGFR), other pathways related to carcinogenesis were not mentioned. To our knowledge, no detailed gene expression analyses using RNA sequencing (RNA-Seq) in canine iUC have been reported to date. RNA-Seq allows more sensitive detection of transcripts than microarrays in both dogs and humans [
9,
10], contributing to a better understanding of molecular pathways in iUC.
In the present study, we performed a comprehensive gene expression analysis of canine iUC by RNA-Seq to identify key molecular pathways in canine iUC. The findings from dogs with iUC have the potential to improve our understanding of progression, diagnosis, and treatment of iUC in humans.
Discussion
In this study, we applied a transcriptome sequencing approach to identify gene expression characteristics of canine iUC. We detected 2531 DEGs, and hierarchical clustering of these DEGs showed a clear separation of normal and tumor tissues. Using genes that were previously shown to divide the canine iUC into subgroups [
8], the present study also identified two distinct subgroups within dogs with iUC. Furthermore, pathway analysis revealed a number of upstream regulators of the DEGs in canine iUC, providing insights into disease biology and potential targets for therapeutic intervention.
PTGER2, encoding prostaglandin E
2 (PGE
2) receptor 2 (also known as EP2), was the most activated upstream regulator in canine iUC. In addition, we confirmed that EP2 protein was strongly expressed in tumor cells and surrounding inflammatory cells but not in normal epithelial cells. PGE
2 is a major arachidonic acid-derived metabolite that is abundantly produced by cyclooxygenase-2 (COX-2) upon inflammation [
27]. The biological effects of PGE
2 are mediated through four types of the receptor, EP1–4 [
28]. Upregulation of COX-2 accompanied by high levels of PGE
2 is common in a variety of tumor tissues [
29]. Previous studies have shown that elevated expression of COX-2 and increased levels of PGE
2 are observed in both human and canine bladder cancer [
30‐
35]. PGE
2–EP2 signaling promotes tumor development via multiple mechanisms, including stimulatory effects on cell proliferation, survival, invasion, angiogenesis, and metastasis [
36]. A recent study demonstrated that PGE
2 induces programmed cell death protein ligand 1 expression in tumor-associated macrophages and myeloid-derived suppressor cells, leading to the inhibition of anti-tumor immunity [
37]. Furthermore, PGE
2 is involved in resistance to chemotherapy by promoting repopulation of cancer stem cells in bladder cancer [
38]. In addition, COX-2 inhibitors consistently exert anti-tumor effects on bladder cancer in both dogs and humans [
7,
24,
39]. These findings suggest that the COX-2/PGE
2/EP2 pathway plays a role in tumorigenesis and progression of bladder cancer.
ERBB2 (also known as HER2) was the second most activated upstream regulator in canine iUC. HER2 is a membrane-bound receptor tyrosine kinase belonging to the epidermal growth factor receptor (EGFR) family and is implicated as an oncogene in many human cancers [
40]. Notably, HER2 is amplified and overexpressed in one-third of patients with breast cancer and is associated with poor prognosis [
41]. HER2 overexpression facilitates excessive dimer formation, driving oncogenic cell survival, proliferation, apoptosis suppression, and angiogenesis [
40,
42]. Thus, a number of clinical trials of HER2-targeted therapies in patients with breast cancer have been conducted, showing increased survival in both the metastatic and early-stage settings of the disease [
40,
43‐
45]. Recently, The Cancer Genome Atlas project reported mutation or amplification of ERBB2 in a subset of human iUC [
26]. Another study also identified activating ERBB2 mutations in the absence of gene amplification in human iUC [
46]. These observations suggest that HER2-targeted therapy might also be effective for the treatment of iUC. Clinical trials to evaluate the efficacy and safety of HER2-targeted drugs in patients with bladder cancer are ongoing [
47]. Our data indicate that HER2 is involved in disease progression and is a therapeutic target in canine iUC, as is the case with humans.
Pathways of several transcription factors (CCND1, FOXM1, E2F3, FOXO1, and JUN) were enriched in canine iUC. Of these, CCND1, which encodes cyclin D1, plays a critical role in the cell cycle machinery (e.g., G1-S transition) and is considered a promising target for cancer therapy [
48]. Proinflammatory and angiogenesis cytokines (IL6, IL1, CSF2, TNF, IFNG, IL22, HGF, and Vegf) were also enriched in canine iUC. Our data support that aberrant inflammation and neovascularization occur in the tumor environment, and that anti-inflammatory and anti-angiogenic drugs targeting these cytokines have therapeutic potential for canine iUC [
49]. Recently, immune-checkpoint inhibitors targeting the programmed death 1/programmed death-ligand 1 (PD-1/PD-L1), atezolizumab and nivolumab, have been approved by the Food and Drug Administration (FDA) in patients with advanced bladder cancer [
50]. Although our study did not detect PD-L1 (CD274) within the 2513 DEGs, immune-checkpoint inhibitors might be effective against canine iUC.
By contrast, this study predicted that the pathways of Irgm1, TP53, let-7, and ZFP36 are inhibited in canine iUC. Irgm1 is a member of the interferon-γ (IFN-γ)-inducible GTPase family and plays a role in innate immunity by regulating autophagy [
51]. Recent studies have reported that Irgm1 enhances tumorigenesis and metastasis of melanoma in mice [
52,
53]. High concentrations of IFN-γ are toxic to neighbor cells; however, Irgm1 protects immune cells that produce IFN-γ from autophagic cell death, thereby promoting the expansion of CD4
+ T cells [
54]. Although the significance of Irgm1 in canine iUC is unclear, inhibition of the pathway might prevent anti-tumor immunity from eliminating cancer cells through enhancement of IFN-γ-induced immune cell death.
TP53 is one of the most important tumor-suppressor genes and plays a role in cell cycle arrest and apoptosis in response to DNA damage [
55]. In humans, about half of all tumor types including bladder cancer possess alterations in TP53 or its upstream/downstream genes [
26,
56]. In dogs, TP53 mutations are also observed in many types of tumors, such as lymphoma [
57], osteosarcoma [
58,
59], and mammary tumor [
60,
61]. In canine iUC, the presence of TP53 mutation is deduced by immunohistochemistry (IHC). The mutated p53 protein, with a prolonged half-life, is considered to result in positive immunoreactivity of neoplastic cell nuclei, and thus can be detected by IHC [
62]. However, TP53 mutations have not been directly confirmed in canine iUC by DNA sequencing analysis. Considering that the TP53 pathway was inhibited in this study, genetic and/or epigenetic abnormalities of TP53 may be present in canine iUC.
let-7 is a founding member of the microRNA (miRNA) family, postulated to function as a tumor suppressor gene by negatively regulating the post-transcriptional expression of multiple oncogenes including Ras, Myc, and Hmga2, as well as other cell cycle regulator genes in humans [
63‐
65]. Reduced expression of let-7 is often observed and associated with poor prognosis in many human cancers, particularly in lung carcinoma [
66]. Furthermore, let-7 has an attractive potential for the treatment of cancers; for example, intranasal let-7 administration reduced tumor formation in a mouse model of lung cancer [
67]. As the let-7 pathway was significantly inhibited while the Ras pathway was significantly activated in canine iUC (Additional file
7: Table S4), exogenous let-7 administration may restore the reduced expression of let-7 and may be a potent therapeutic strategy.
ZFP36 encodes a tandem Cys-Cys-Cys-His zinc finger protein (also known as tristetraprolin) that binds to and promotes degradation of AU-rich elements in the 3′-untranslated regions of certain mRNAs [
68]. COX-2 is a well-established target of ZFP36 [
69,
70], and in this study, we found that dogs with iUC exhibited suppression of the ZFP36 pathway. Thus, overexpression of COX-2 in canine iUC [
30,
31] could be attributed to the inhibition of the ZFP36 pathway and subsequent lack of rapid mRNA degradation.
Highly expressed genes present in canine iUC were associated with cancer and leukocyte migration, whereas downregulated genes were associated with muscle function. Decreased expression of muscle-related genes may result in a muscle-invasive phenotype. In addition, gene sets enriched in canine iUC were associated with mammary tumors, consistent with previous studies whereby signatures of human bladder cancer are similar to that of breast cancer [
26,
71].
Several biomarkers can be used as a screening test for bladder cancer in dogs as well as humans [
72]. Urinary bladder tumor antigen (BTA) test is an agglutination reaction that detects complexes of basement membrane proteins degraded by proteinases of urothelial neoplasms [
73]. The present study revealed that the mRNA expressions of matrix metalloproteinase 7 (MMP7), MMP9, MMP12, and MMP13 were significantly upregulated in canine iUC (Additional file
3: Table S2). A recent study observed significantly high rates of the BRAF V595E mutation (approximately 80%) in canine iUC and prostate cancer [
13]. In our study, this mutation was detected in 6/11 (54.5%) cases, although no clear separation was observed between dogs with or without BRAF mutation in a PCA plot. A possible reason may be that the sample size was too small to detect a significant correlation; therefore, more extensive studies are needed to clarify the role of BRAF mutation in canine iUC.
One limitation of this study is that the healthy control dogs were not well matched to the iUC cases by sex and breed, although there was no significant difference in the sex distribution between the two groups. In the pathway analysis of this study, sex hormone-associated genes were not observed in the top 500 upstream regulators of DEGs (Additional file
7: Table S4). Therefore, it appears that gender bias did not impact the RNA-Seq data. Since it was impossible to obtain fresh bladder tissues from various breeds matched to the clinical cases, all healthy control dogs were the same breed, and this could lead to erroneous results related to the expression of some genes that differ between breeds. Further studies will be necessary to examine breed differences in bladder gene expression.