KDELR2-KIF20A axis facilitates bladder cancer growth and metastasis by enhancing Golgi-mediated secretion

Bladder cancer (BCa) is a fatal form of cancer worldwide associated with a poor prognosis. Identifying novel drivers of growth and metastasis hold therapeutic potential for the disease. Transport homeostasis between the endoplasmic reticulum and Golgi and the secretion of matrix metalloproteinases (MMPs) mediated by Golgi have been reported to be closely associated with tumor progression. However, to date, mechanistic studies remain limited. Here, we identified KDELR2 as a potential risk factor with prognostic value in patients with BCa, especially those harbouring the KDELR2 amplification. In addition, we found that KDELR2 is a regulator of BCa cell proliferation and tumorigenicity based on bioinformatic analysis with functional studies. Mechanistically, we revealed that KDELR2 could regulate the expression of KIF20A, thus stimulating the expression of MMP2, MMP9 and MKI67. Functionally, the overexpression of KDELR2 and KIF20A markedly promoted proliferation, migration, and invasion in vitro and enhanced tumor growth in vivo, while knockdown of KDELR2 and KIF20A exerted the opposite effects. And the overexpression of KDELR2 also enhanced lymph node metastasis in vivo. Collectively, our findings clarified a hitherto unexplored mechanism of KDELR2-KIF20A axis in increasing Golgi-mediated secretion of MMPs to drive tumor progression in BCa.

chaperones from the Golgi complex to the ER. KDELRs play a key role in maintaining transport homeostasis between the Golgi and the ER. KDELR1-3 has different binding substrates, with KDELR2 being specific and the others being general [6]. Previous studies have shown that KDELR2 is involved in the regulation of cellular functions, including cell proliferation, survival, migration, cytoskeletal rearrangements, secretion and other biological activities [7], regulating the ER stress [8,9], cellular secretory traffic [10], and immune responses [11]. However, the roles of KDELRs family remain unclear, especially in tumor progression. The existing research state that human glioma tissues and cell lines highly express KDELR2 [12,13]. KDELR2 is a poor prognostic factor for glioblastoma and acts as a direct target of HIF-1α, a regulator of p-mTOR as well as the downstream protein in glioblastoma cells [12]. KDELR2 knockdown could inhibit the viability of glioma cells, promote cell cycle arrest at the G1 phase and induce apoptotic cell death by targeting CCND1 [13]. In melanoma cells, KDELR activity promoted invadopodia formation and extracellular matrix (ECM) degradation [14]. Overexpression of KDELR2 is sufficient to independently trigger lung cancer invasion and metastasis by an EMTindependent mechanism [15]. Increasing evidence has revealed that cellular KDELR2 has more complex roles in tumor progression. Nonetheless, to date, no study has demonstrated the putative role of KDELR2 and its function in the Golgi-mediated expression of proteins, as well as in the invasion and metastasis of BCa cells driven by matrix metalloproteinases (MMPs).
Comparatively, the pivotal role of the kinesin family member 20A (KIF20A) has been extensively explored in cancer. In 2015, upregulation of KIF20A was first identified in pancreatic cancer, and KIF20A inhibition appeared to significantly inhibit the proliferation of pancreatic cancer cells [16]. Subsequently, a prominent increase in KIF20A expression was observed in several malignancies, such as lung [17,18], breast [19], gastric [20], liver [21], bladder [22], and pancreatic cancers [23,24]. Moreover, overexpression of KIF20A was tightly associated with a poor prognosis and the clinical parameters in patients with gastric cancer [25]. Currently, although studies have reported that KIF20A induced the progression of BCa by promoting the growth, viability and metastasis of BCa cells [22], the specific regulatory and molecular mechanisms of KIF20A and its potential role in BCa tumorigenesis remains largely elusive.
Through the current study, we have provided mechanistic insights into the effect of KDELR2 activation on the behaviour of BCa cells. Our results revealed that KDELR2 positively modulates the expression of KIF20A, while knockdown of KDELR2 downregulated KIF20A and further attenuated the growth, migration, and invasion of BCa cells.

Identification of KDELR2 upregulation is prognostically significant in patients with BCa
Three microarray datasets (GSE3167, GSE52519, GSE65635) were first downloaded from the GEO database to identify the candidate genes associated with the carcinogenesis and progression of BCa.  (Fig. 1A). Those genes that were not differentially expressed in BCa, such as EGFR, p53, and FGFR3, were excluded (data not showed). Surprisingly, the three datasets contained one overlapping co-dysregulation gene in BCa, KDELR2, as shown in the Venn diagram (Fig. 1B). Furthermore, the results revealed that KDELR2 was significantly elevated in BCa tissues relative to normal tissues (Fig. 1C). The ROC curves, based on the GSE3167 dataset, revealed that the level of KDELR2 could differentiate BCa from normal tissues with an AUC of 0.9101 (95% CI: 0.8389 to 0.9814, P < 0.0001) (sFig. 1A). Consequently, we investigated the changes in the expression of KDELR2 in TCGA database and assessed its prognostic correlation in patients with BCa. As expected, this upregulation of KDELR2 expression was also observed in TCGA-BLCA database (Fig. 1D). Correspondingly, the in-depth analysis revealed a significant positive correlation between the level of KDELR2 and the clinical stage of patients with BCa (Fig. 1E). The ROC curves showed that KDELR2 exhibited considerable diagnostic efficiency for BCa (P < 0.0001; AUC = 0.8286, 95% CI: 0.7572 to 0.8999) (sFig. 1B). Here, for the first time, Kaplan-Meier curves revealed that KDELR2 was statistically associated with the OS and DFS rates ( Fig. 1F-G). The survival time of patients with BCa with low KDELR2 expression was longer than that of those with high KDELR2 expression. Similar survival analysis outcomes were observed in the additional dataset GSE32894 (sFig. 1C). Next, KDELR2 was identified through multivariate Cox regression analysis to construct a predictive signature. The multivariate analysis showed that high KDELR2 expression was an independent predictor of poor OS (HR, 1.359; 95% CI, 1.003-1.84; p = 0.0481) (Fig. 1H) and DFS (HR, 1.394; 95% CI, 0.993-1.96; p = 0.0552) (sFig. 1D). Thus, we focused on this significant prognostic or clinical outcome-related gene in our further research.

KDELR2 functions as an oncogene in BCa cells
In addition to the data obtained from publicly available clinical cohorts, we further verified the aberrant expression of KDELR2 in 24 clinical samples. The qRT-PCR assay indicated that KDELR2 was upregulated in BCa compared to normal bladder tissues, which was consistent with the results of the bioinformatic analysis ( Fig. 2A). The ROC curves showed a similar high diagnostic efficiency of KDELR2 in the 24 clinical samples (sFig. 1E). Immunohistochemical analysis of the clinical samples revealed positive KDELR2 staining in the cytoplasm of tissues, indicating that KDELR2 is commonly overexpressed in BCa (Fig. 2B). In addition, higher expression levels of KDELR2 were detected in BCa cell lines (T24 and UMUC3) than in the urothelial cell line SV-HUC-1 (Fig. 2C, sFig. 1F), implying the potential role of KDELR2 in promoting bladder tumorigenesis. Subsequently, we evaluated the potential biological function of KDELR2 in BCa cells by introducing three independent siRNAs targeting KDELR2 into the UMUC3 and T24 cells. Interestingly, although all three siR-NAs significantly inhibited the mRNA levels of KDELR2, only siKDELR2-3 significantly reduced the protein expression (Fig. 2D). Thus, siKDELR2-3 was introduced into BCa cells to observe cell growth. We found that KDELR2 suppression significantly reduced cell viability (Fig. 2E). Invasion and metastasis are hallmarks of cancer cells. KDELR malfunctions have been reported to be associated with changes in ECM degradation and cellular adhesion [14,26]. We observed that the control cells had a high level of mobility and invasiveness than the siKDELR2 cells, indicating that siKDELR2 induced tumor suppression (Fig. 2F).
We also established exogenous KDELR2-overexpressing T24 and UMUC3 cell lines to further understand the oncogenic functions of KDELR2 (Fig. 2G). Increased KDELR2 expression significantly increased the proliferation, migration, and invasion of BCa cells ( Fig. 2H-I). Taken together, the results provide strong evidence that KDELR2 has oncogenic properties in vitro.

KDELR2 is a positive regulator of MMP2, MMP9, MKI67 in BCa
We performed GSEA to establish the potential biological functions of KDELR2 using the LinkedOmics tool. KEGG pathway enrichment analysis revealed that KDELR2 was related to protein processing in the ER, phagosome, ECMreceptor interaction, focal adhesion, cell adhesion molecules (CAMs), and regulation of the actin cytoskeleton (sFig. 2A). GO analysis showed that KDELR2 was mainly involved in the biological processes of immune responses, collagen metabolic process, phagocytosis and in regulating the ER stress and Golgi vesicle transport (Fig. 3A). To identify the components that assist KDELR2-mediated proliferation, we investigated its correlation with MKI67. We observed that the mRNA levels of KDELR2 positively correlated with those of MKI67 (r = 0.2401, p < 0.001) ( Fig. 3B) in TCGA-BLCA dataset. In addition, we investigated some of the known proteases secreted by the Golgi, such as MMPs, to identify the components that assist KDELR2-mediated invasion. It has been well established that MMP2 [27] and MMP9 [28] enhance BCa invasion and metastasis. We observed a close correlation between KDELR2 and MMP2 (r = 0.2230, p < 0.001) (Fig. 3C). However, the correlation between KDELR2 and MMP9 was not statistically significant (r = 0.01280, p = 0.7966) (Fig. 3D). Next, we profiled the expression of these genes by RT-qPCR and western blot analysis after silencing or overexpressing KDELR2 in UMUC3 and T24 cells. The results revealed that altering KDELR2 regulated the expressions of MMP2, MMP9 and MKI67 in BCa cells ( Fig. 3E-H). Taken together, our data suggest that KDELR2 regulates the expression of MMP2, MMP9 and MKI67 in BCa cells.

Upregulation of KIF20A promotes tumor progression in BCa
Furthermore, we observed that KDELR2 might be involved in gene regulation through regulation of epigenetic, gene silencing, RNA splicing or mRNA processing (sFig. 2B). Thus, we sought to investigate the regulatory effects of KDELR2 in BCa, and performed further bioinformatics analysis based on data from the GEO database. Based on the median value of KDELR2 expression, every GSE expression matrix was divided into high and low groups, and then limma package was used to accessed top DEGs that related to KDELR2. |Foldchange|> 1.5 and p-value < 0.05 were considered statistically significant. As shown in the Venn diagram, we identified six commonly up-regulated DEGs (KIF20A, CDK1, CAPG, NOP10, PKM, IP6K2) in KDELR2 high expression group (Fig. 4A) and one commonly down-regulated DEG (CRP) in KDELR2 low expression group (Fig. 4B). To further analyse the clinical significance of these prospective candidates in BCa, we evaluated their differential expression by analysing the TCGA-BLCA dataset. Accordingly, KIF20A, CAPG, and CDK1 were dysregulated in BCa, but no significant difference was observed in the other genes (sFig 2C). Importantly, we profiled the expression of KIF20A, CAPG, CDK1 by RT-qPCR analysis after silencing or overexpressing of KDELR2 in UMUC3 and T24 cells. The results revealed that altering KDELR2 regulated the expression of KIF20A, but not CAPG and CDK1 (Fig. 4C-D). We also observed that the mRNA levels of KIF20A were positively related to those of KDELR2 in BCa tissues (r = 0.2975, p < 0.001) (Fig. 4E). And changes in protein levels of KIF20A was also observed in BCa cells after altering KDELR2 (Fig. 4F). In addition, the ROC curves revealed that up-regulated KIF20 possess high diagnostic efficiency in BCa (sFig 2D). Thus, we identified KIF20A as a potential regulatory target of KDELR2 for further study.
To explore the underlying role of KIF20A in BCa, its abnormal expression in clinical samples and BCa cells were also evaluated. The results revealed that KIF20A was highly expressed in clinical samples and BCa cells (Fig. 4G-H), which implied that an abnormality in KIF20A might induce potential oncogene function in BCa. Similar to KDELR2, exogenous KIF20A and three independent siRNAs targeting KDELR2 were introduced into UMUC3 and T24 cells. The efficiency was profiled by RT-qPCR and western blot analysis (sFig. 3A-D). As expected, KIF20A knockdown markedly impaired the growth of UMUC3 and T24 cells (Fig. 4I), while its overexpression significantly increased cell proliferation (sFig. 3E). Transwell assays were performed to confirm the role of KIF20A as a driver of invasion and metastasis. KIF20A silencing significantly reduced cellular mobility and invasiveness (Fig. 4J), while higher KIF20A expression increased the invasiveness of UMUC3 and T24 cells compared with the control group (sFig. 3F). Collectively, these results suggest that upregulation of KIF20A promotes tumor progression in BCa.

KDELR2 promotes BCa cell proliferation and metastasis in a KIF20A-mediated manner
To investigate the underlying mechanism by which KIF20A drives BCa progression, we tested the expression levels of MMPs after altering KIF20A. As expected, overexpression and knockdown of KIF20A significantly upregulated and downregulated the protein expression of MMP2 and MMP9 in BCa cells, respectively (Fig. 5A-B). To further explore whether KDELR2 exerts regulatory effects via KIF20A, functional rescue experiments were performed. We evaluated the expression of MMP2 and MMP9 upon knockdown of KIF20A in KDELR2amplified BCa cells. The results confirmed that silencing KIF20A in these cells reduced the levels of MMP2 and MMP9, thereby reducing its induction by KDELR2 (Fig. 5C). The CCK-8 assay results revealed that overexpression of KDERL2 promoted the proliferation of UMUC3 and T24 cells, which was nonetheless impaired by simultaneous knockdown of KIF20A (Fig. 5D). Transwell assays also revealed that KIF20A knockdown partially attenuated the effects of overexpression of KDELR2 on BCa cell metastasis compared with that of the controls (Fig. 5E), indicating that KDELR2 promotes BCa cell proliferation and metastasis in a KIF20A-mediated manner in vitro.

KDELR2 is important for BCa progression in vivo
To further confirm the aforementioned in vitro findings, T24 cells stably transfected with KDELR2 or control vector were subcutaneously injected into BALB/c nude mice. Supporting the results obtained in vitro, the xenograft experiments showed that the tumor volume and weight significantly increased in the KDELR2 overexpression group (Fig. 6A-D). As evidenced by IHC staining, the expression levels of KIF20A, MMP-2, MMP-9 and MKI67 increased significantly in tumor tissues in the KDELR2 overexpression group (Fig. 6E). Furthermore, T24 cells were injected into the footpads of nude mice to construct a lymph node metastasis model to confirm the role of KDELR2 in facilitating tumor metastasis (Fig. 6F). We found that the volumes of popliteal LNs increased in the stable KDELR2 overexpression group (Fig. 6G). HE staining of the popliteal LNs further indicated that upregulation of KDELR2 could promote lymphatic metastasis of BCa in vivo (Fig. 6H). We also observed a significant reduction in tumor volume and weight in the stable KDELR2 silence group (Fig. 6I-L). Additionally, overexpression of KIF20A resulted in an increased tumor growth, which also reversed KDELR2 silence-induced decreased tumor growth (Fig. 6I-L). Taken together, the present study indicated that KDELR2 facilitated the expression of KIF20A, MMP-2, MMP-9 and MKI67 and BCa growth and metastasis in vivo.

Discussion
Postoperative recurrence and distant metastasis remain a conundrum for BCa prognosis. Therefore, understanding the regulators of these processes is important to therapeutically prevent BCa metastasis. KDELR2 has been reported to be a robust and independent driver of lung cancer invasion and metastasis [15]. In the current study, we revealed that KDELR2 was highly expressed in BCa, especially advanced BCa tissues. In addition, we identified KDELR2 as a potential risk factor with prognostic value in patients with BCa, especially those harbouring the KDRLR2 amplification. Moreover, bioinformatic analysis with a gain-of-function invasion screen revealed that KDELR2 was a robust and independent driver of BCa cell proliferation and tumorigenicity. Our data indicated that KDELR2 could alter cellular proliferation and robustly promote invasion and metastasis. There has been a long-established connection between cancer outcomes and the transport homeostasis between the Golgi and the ER; however, mechanistic studies remain limited. After KDELR/ligand interaction, the active G α subunits activate their specific target protein kinases, which subsequently modulate gene transcription followed by regulation of the anterograde or retrograde trafficking [10,29]. Our results provide the first evidence that active KDELR2 regulates the transcription and expression of MMP2, MMP9 and MKI67. Previous reports showed that MMP secretion was modulated by alterations in KDELR2. RT-qPCR revealed that the MMPs 1a, 2 and 9 were transcriptionally upregulated upon KDELR2 overexpression; however, they did not show any change upon gene repression [15]. Interestingly, in this study, we observed alterations in mRNA and protein expression in both MMP2 and MMP9 upon KDELR2 overexpression or repression. Thus, the tumorpromotive function of KDELR2 on MMP2 and MMP9 might be tissue/cell-specific. Detailed parallel studies on different tissue/cell types are required to further elucidate the complex functionality.
Despite evidence highlighting the roles of KIF20A in various tumor cells and its importance in regulating tumorigenesis, the underlying mechanism in BCa remains largely elusive. KIF20A could promote the growth of colorectal cancer cells via the JAK/STAT3 signalling pathway [30]. Moreover, inhibiting KIF20A resulted in gastric cancer cell mitosis (G2/M phase) arrest and improved drug resistance to chemotherapy [20]. However, KIF20A overexpression conferred resistance to paclitaxel in breast cancer cases [9]. Collectively, the results demonstrated that KIF20A played a critical role in the proliferation, invasion and chemotherapeutic drug sensitivity of the tumor cells. Although abnormal expression of KIF20A has been reported in BCa, the present study is the first to reveal its specific mechanism in promoting tumor progression and regulating the expression of MMPs. Mechanistically, this study identified KIF20A as a regulatory target of KDELR2, in which KDELR2 promotes KIF20A expression and KIF20A enhances MMP2 and MMP9 expression to drive cellular invasion and metastasis. Therefore, our findings elucidate a hitherto unexplored mechanism of KDELR2 and KIF20A in BCa, linking a Golgi-ER traffic transport protein to a critical kinesin during cancer progression.
MMP2 and MMP9 are two major MMPs that play critical roles in the degradation of various ECM components. Both two MMPs are overexpressed in many cancers, which correlate with increased cancer cell invasion and metastases [31,32]. Interestingly, MMP2 and MMP9 have been studied extensively considering their prognostic values and pro-invasive functions in BCa [28,33]. The results of the current study showed that the expression levels of MMP2 and MMP9 correlated well with KIF20A and KDELR2 overexpression and BCa cell invasion. The experiments further revealed that both KIF20A and KDELR2 upregulated the transcription and expression of MMP2 and MMP9, further promoting the invasion of BCa cells. Nevertheless, our research has certain limitations. We conducted the survival analysis on KDELR2 using a reliable public database; however, additional studies are needed to evaluate the role of KDELR2 in patients with BCa. In subsequent work, we will enlarge the KDELR2 sample amount and supplement sufficient follow-up information of clinical patients. Furthermore, additional studies are needed for an in-depth investigation of the mechanism by which KDELR2 promotes the expression of KIF20A.

Conclusions
In summary, our study elucidated the individual roles of KDELR2 and KIF20A in driving BCa tumorigenesis and metastasis. In addition, our study revealed that the combined inhibition of KDELR2 and KIF20A could be a promising therapeutic option in patients with BCa. Our findings may have broader implications for targeted therapies that could be extended to other cancers.

Human BCa and paired adjacent specimens
Human BCa tumor samples were obtained from the Department of Urology, Union Hospital, Tongji Medical College, Wuhan, China. The study complied with the relevant ethical regulations regarding research involving human participants. Written informed consent was obtained from each participant, and the local ethics committee approved this study.

Cell proliferation assays
The cell proliferation assay was performed based on a method described previously [34]. Briefly, 2000 cells were plated in 96-well plates, and the cell proliferation rates were measured using the MTS method according to the manufacturer's instructions.

Transwell assays
The migration assays were performed based on a method described previously [35]. Briefly, after pretreatment with serum-free medium for 24 h, 50,000 UMUC3 and T24 cells were seeded in the top chamber (8 μm pore size, Corning). After 12 h (T24) and 24 h (UMUC3), the cells on the lower surface were fixed in 100% methanol and stained with 0.05% crystal violet, photographed randomly (200x) and counted. A similar protocol was followed for the invasion assays, except that the cells were further required to invade through Matrigel (Thermo Fisher Scientific).

RNA isolation and qRT-PCR
Total RNA was extracted using the TRIzol reagent (Thermo, Massachusetts, USA) and reverse-transcribed as described previously [36]. The real-time PCR analysis was performed using SYBR Green mix (Thermo, Massachusetts, USA). Results were normalised to GAPDH in each sample. Gene primers were obtained from Sangon Biotech (Shanghai) and are listed:

Animal experiments
BALB/c nude mice (5-6 weeks old) were purchased from Beijing HFK Bioscience (Beijing, China). All animal care and experimental procedures were approved by the Institutional Animal Use and Care Committee of Tongji Medical College, Huazhong University of Science and Technology and were performed according to established guidelines. The study complied with the relevant ethical regulations pertaining to animal research.
For the tumor growth study, 5 × 10 6 T24 cells were injected subcutaneously into the flank of mice. Each group contained five mice. Tumor size was measured every three days. Four weeks later, the mice were sacrificed, and the weight and volume of the tumors were assessed.
For the popliteal lymph node (LN) metastasis assay, 10 mice were randomly assigned to two group. 5 × 10 5 lentivirus-transduced stably expressed KDELR2 or control T24 cells were inoculated into the footpads. After 5 weeks, the popliteal LNs were enucleated and measured. The tumor and LN volumes were measured using the following formula: length × width 2 × 0.52. Tumor specimens and popliteal LNs were fixed and embedded in paraffin for HE staining and immunohistochemical (IHC) analysis.

Bioinformatic analysis
The UCSC Xena browser (https:// xenab rowser. net/) was used to investigate the gene expression profiles and the correlation of between KDELR2 and KIF20A, MMP2, MMP9, and MKI67 in bladder urothelial carcinoma (BLCA) from The Cancer Genome Atlas (TCGA) database. The normalisation of gene expression was performed using the edgeR package. The gene expression profiles of KIF20A, CDK1, CAPG, NOP10, PKM, IP6K2, and CRP was obtained from the GEPIA [40]. TCGA-BLCA dataset contained 414 BCa tissue samples and 19 non-cancerous samples. The clinical information and pathology records of the patients with BCa were also obtained from TCGA-BLCA database and were used to evaluate the overall survival (OS) and disease-free survival (DFS) of the patients. The GSE32894 dataset [41], contained the clinical information and gene expression profile of 224 patients with BCa, which was also used for the survival analysis.
To identify the underlying biological functions of KDELR2, Gene Set Enrichment Analysis (GSEA) was performed in the LinkInterpreter module of the LinkedOmics tool [42]. GO assessed the biological processes (BP). The minimum number of genes (size) was set at 10 and simulations as 500. The P-value cut-off was set at 0.05.

Statistical analysis
GraphPad Prism 8.0 (GraphPad Software Inc.) was used to perform the statistical analyses. Results are presented as means ± standard error of the mean (SEM). Student's t-test or one-way analysis of variance (ANOVA) was used to evaluate the differences in KDELR2 gene expression between different subgroups of patients. In addition, the Pearson correlation coefficient and chi-squared test were performed to assess the associations between KDELR2 expression levels and clinicopathological characteristics. Receiver operating characteristic (ROC) curves were plotted, and the area under the curve (AUC) was calculated to analyse the diagnostic efficiency. The prognostic significance of KDELR2 to predict the OS and DFS of patients with BCa were estimated by the Kaplan-Meier method. Furthermore, the log-rank test was used to depict the OS or PFS distributions of the patients with different expression levels of KDELR2. A confidence threshold, p < 0.05, was considered statistically significant. *, p < 0.05; **, p < 0.01; ***, p < 0.001; ****, p < 0.0001.