Down-regulation of BCL2L13 renders poor prognosis in clear cell and papillary renal cell carcinoma

BCL2L13 belongs to the BCL2 super family, with its protein product exhibits capacity of apoptosis-mediating in diversified cell lines. Previous studies have shown that BCL2L13 has functional consequence in several tumor types, including ALL and GBM, however, its function in kidney cancer remains as yet unclearly. Multiple web-based portals were employed to analyze the effect of BCL2L13 in kidney cancer using the data from TCGA database. Functional enrichment analysis and hubs of BCL2L13 co-expressed genes in clear cell renal cell carcinoma (ccRCC) and papillary renal cell carcinoma (pRCC) were carried out on Cytoscape. Evaluation of BCL2L13 protein level was accomplished through immunohistochemistry on paraffin embedded renal cancer tissue sections. Western blotting and flow cytometry were implemented to further analyze the pro-apoptotic function of BCL2L13 in ccRCC cell line 786-0. BCL2L13 expression is significantly decreased in ccRCC and pRCC patients, however, mutations and copy number alterations are rarely observed. The poor prognosis of ccRCC that derived from down-regulated BCL2L13 is independent of patients’ gender or tumor grade. Furthermore, BCL2L13 only weakly correlates with the genes that mutated in kidney cancer or the genes that associated with inherited kidney cancer predisposing syndrome, while actively correlates with SLC25A4. As a downstream effector of BCL2L13 in its pro-apoptotic pathway, SLC25A4 is found as one of the hub genes that involved in the physiological function of BCL2L13 in kidney cancer tissues. Down-regulation of BCL2L13 renders poor prognosis in ccRCC and pRCC. This disadvantageous factor is independent of any well-known kidney cancer related genes, so BCL2L13 can be used as an effective indicator for prognostic evaluation of renal cell carcinoma.


Introduction
Kidney cancer remains one of the malignant tumors with the incidence increased notably in recent years [1]. It is composed of heterogeneous subtypes with histological and molecular abnormalities. Clear cell renal cell carcinoma (ccRCC) accounts for about 75% of all the cases, which is mainly characterized as constitutional chromosome 3p deletion, dysfunctional von Hippel-Lindau (VHL) gene and uncontrolled stabilization of hypoxia inducible factors (HIFs) by molecular features. Papillary renal cell carcinoma (pRCC), which occupies about 15% of the incidence, usually displays loss of chromosome Y, gain of chromosome 7 and/or chromosome 17. Moreover. different molecular characteristics are assigned to each pathological subtype, including both Type I and Type II pRCC [2,3]. Chromophobe renal cell carcinoma (chRCC) and some other rare types make up the rest.
Mutations of VHL have been proved to be one of the typical genetic causes of kidney cancer, which often lead to stabilization of HIF1α and HIF2α, creating an illusion of pseudohypoxia in stricken renal tissues [4]. The accumulated HIF1α-HIF1β and HIF2α-HIF1β dimers will cause an elevation in growth factors, including plateletderived growth factors (PDGFs) and vascular endothelial growth factors (VEGFs), which will prompt tumor angiogenesis [5,6]. The VHL-HIF-VEGF axis represents one of the canonical pathways for renal carcinogenesis, while some other gene mutations or epigenetic changes have also been reported [7,8], including mutations of polybromo-1 (PBRM1) [9], SET domain-containing 2 (SETD2) [10] and tuberous sclerosis complex 1/2 (TSC1/2) [11,12]. Nevertheless, nearly 40% of the resected kidney cancer patients are confronted with risk of recurrence, the effective predictive markers are absent so far [13].
BCL-2-like protein 13 (BCL2L13), also termed BCLrambo, is encoded by the BCL2L13 gene [14]. It has manifested that BCL2L13 participates in drug resistance in several tumors. For example, BCL2L13 was found elevated in tumors like glioblastoma (GBM) and childhood acute lymphoblastic leukemia (ALL) [15,16]. In GBM, BCL2L13 interacts with ceramide synthases 2 and 6 (CerS2/6), inhibiting the leakage of cytochrome c into cytoplasm [15]. Augmented BCL2L13, which takes part in L-asparaginase resistance, executes as an independent adverse prognostic factor in ALL [16]. On the other hand, BCL2L13 has been demonstrated to interact with adenine nucleotide translocator (ANT, encoded by SLC25A4), a component of mitochondrial permeability transition pore (MPTP), therefore promoting cytochrome c release from mitochondrial intermembrane space to cytoplasm, resulting in activation of the apoptotic caspase cascade [17].
The prognostic value of BCL2L13 in kidney cancer is still unclearly. In this study, we profiled BCL2L13 across 33 cancer types in the Cancer Genome Atlas (TCGA), and found that its mRNA expression is significantly reduced in ccRCC and pRCC. Moreover, low BCL2L13 expression correlates with lessened survival probability of kidney cancer. It poses an attractive hypothesis that BCL2L13 may be a promising prognosis marker for kidney cancer.

University of California Santa Cruz (UCSC) genome browser and UCSC Xena
Protein-protein interaction (PPI) network analysis of BCL2L13 was performed using UCSC genome browser (https:// genome. ucsc. edu), which offers visualized interconnection of the concerned genes [18,19]. BCL2L13 mRNA and exon expression were completed on UCSC Xena platform (http:// xena. ucsc. edu), complied with data from TCGA [20].

Catalogue of Somatic Mutations in Cancer (COSMIC) and Open Targets Platform
The most frequent somatic mutations in ccRCC and pRCC tissues were queried by COSMIC cancer browser (http:// cancer. sanger. ac. uk), showing the top 20 candidate genes curated from published research for each cancer type [21]. Heritable kidney cancer-predisposing syndrome related genes are searched from Open Targets Platform (https:// www. targe tvali dation. org). The overall target-disease association score is the harmonic sum aggregated from genetics, genomics, drugs, animal models and text mining data [22].

Cytoscape and Search Tool for the Retrieval of Interacting Genes (STRING)
Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analysis and hubs of BCL2L13 coexpressed genes were accomplished within Cytoscape, with the plug-in ClueGO, CluePedia, MCODE and CytoHubba, which realized by the grid topology algorithm density of maximum neighborhood component (DMNC) [29][30][31]. STRING database (https:// string-db. org/) was combined for sifting the hub genes [32].

Transient transfection and apoptosis induction
Full-length of human BCL2L13 was cloned into eukaryotic S-tag fusion expression vector pcDNA3-S-tag via EcoRI/XhoI (TaKaRa) restriction sites. The recombinant plasmids were then harvested from Escherichia coli DH-5α. Transient transfection was performed with Lipofectamine 2000 (Invitrogen, 11668019) according to the manufacturer's instructions.

Apoptotic assay
The optical microscope (Olympus, CKX53) and microcamera (TUCSEN, DigiRetina 16, TCapture version 5.1) were used to capture the morphological changes of 786-0 cells after treatments. For quantitative assessment, 786-0 cells were harvested after treatments and stained with both FITC Annexin V and PI (BD Pharmingen, 556547), then subjected to flow cytometry (Beckman Coulter, CytoFLEX). Annexin V + / PI − , Annexin V + /PI + and Annexin V − /PI + cells were used to represent the types of early apoptotic, late apoptotic and dead cells, respectively.

Statistical analysis
IBM SPSS statistical software (version 24.0) was used for statistical analysis of the experimental data. All other statistical methods come with the web tools by default. Welch's test was conducted for differential expression of functional mRNA in UCSC Xena. Poisson test was adopted to measure the correlation of a given gene in Open Targets Platform. Transcripts per million (TPM) values and Student's t tests were employed to calculate the significance of gene expression divergence between categories in UALCAN. The linear dependence of targeted gene pair was evaluated through Spearman's correlation coefficient and Pearson's correlation coefficient in cBioPortal. Log-rank test was implemented in both UAL-CAN and cBioPortal for comparison of survival curves, which displayed as Kaplan-Meier plot. P < 0.05 is considered significant in this entry (*, P < 0.05; **, P < 0.01; ***, P < 0.001).

BCL2L13 mRNA expression is significantly reduced in ccRCC and pRCC
BCL2L13 expresses anomalously in a variety of tumors, participating in tumor progression with its apoptosisregulating activity [33][34][35]. The mRNA expression of BCL2L13 was analyzed in a variety of tumors using the data from TCGA. Significant differences were found between quite a few cancer tissues and their normal counterparts, including BLCA, BRCA, CHOL, COAD, HNSC, chRCC, ccRCC, pRCC, LIHC, LUAD, LUSC, READ, THCA, STAD, UCEC (Additional file 1, Additional file 11). BCL2L13 mRNA is highly expressed in chRCC, but has no effect on prognosis. In contrast, significant reduction of BCL2L13 mRNA and exon expression were both found in ccRCC and pRCC primary tumors, when compared to corresponding normal tissues (Fig. 1). Further analysis in ccRCC and pRCC indicated that the reduced BCL2L13 mRNA is independent of patients' race, gender, age, lymph node metastasis status, clinical stages and tumor subtypes (Fig. 2, Additional file 1). Moreover, down-regulation of BCL2L13 actively impact on the patients' survival probability (see below). In other cancers not mentioned herein, the expression differences between tumor tissues and normal tissues were not significant, which might succumb to the corresponding normal tissues are scarcity in TCGA database.
The clinical proteomic tumor analysis consortium (CPTAC) was then engaged to analyze the protein expression level of BCL2L13 in renal cell carcinoma. The BCL2L13 protein expression is consistently lower in ccRCC patients compared to healthy crowd, which is independent of the tumor stage (Additional file 2). However, the relevant data is unavailable for pRCC.
Abnormal methylation of genes often plays an important role in cancer progression. DNA methylation of BCL2L13 coding regions was not altered in renal cell carcinoma samples (Fig. 1A). Moreover, BCL2L13 promoter methylation was elevated significantly in ccRCC, but not in pRCC (Additional file 3). Specifically, hypermethylation of BCL2L13 promoter in ccRCC is independent of patients' age, gender, race, clinical stages, lymph node metastasis status and tumor grade (Additional file 3).

Low expression of BCL2L13 has significant impact on survival probability in ccRCC and pRCC
To evaluate how BCL2L13 expression impact on patient survival, ccRCC and pRCC cases were grouped by their BCL2L13 mRNA levels. BCL2L13 low expression significantly correlated with poor prognosis in ccRCC (P = 0.0021, Fig. 3A), which was independent of patients' gender or tumor grade (Fig. 3C). In pRCC, similar results were observed, albeit to a lesser extent (P = 0.049, Fig. 3B). BCL2L13 expression levels had no obvious impact on patient survival of other cancer types aforementioned by UALCAN (Additional file 4).
Somatic mutations and gene CNAs could also influence the prognosis of cancer patients [36]. Through analysis using data from TCGA, BCL2L13 copy number altered only 0.2% in ccRCC with deep deletion, and 1.1% in pRCC with deep deletion and missense mutations (Additional file 5). Amino acid mutations of BCL2L13 were only found in pRCC cases, namely, Ser38 was mutated to Leu and Gly182 was mutated to Ser, in which S38L occurred mainly in the patients with shallow deletion of BCL2L13 copy number (Additional file 5). BCL2L13 maintained diploid status in ccRCC and pRCC, while the Because only a few patients carry BCL2L13 genetic changes, the impact of these changes on prognosis of ccRCC and pRCC is undeterminable due to limited patient numbers (Additional file 6).

BCL2L13 doesn't correlate with the kidney cancer related genes or putative target miRNA in ccRCC and pRCC
Previous studies have found that several genes, including VHL, PBRM1 and TSC1, played an important role in tumorigenesis of kidney cancer [37]. To uncover the mechanisms underlying the aggravated poor prognosis drived by BCL2L13 downregulation in ccRCC and pRCC, the correlation between BCL2L13 and these genes were analyzed. BCL2L13 correlated weakly with the most frequently mutated VHL, PBRM1, SETD2 in ccRCC and VHL, KDM5C, SPEN in pRCC respectively (Fig. 4, Additional file 7). Hereditary kidney cancer patients occupy 5-8% in all the diagnosed ones, and commonly harbor some cancer predisposition genes, such as TSC1/2, CDKN1C and DIS3L2 [38,39]. BCL2L13 showed frail correlation with the top 5 genes TSC1/2, CDKN1C, VHL and DIS3L2, that associated with inherited kidney cancer-predisposing syndrome (Additional file 12, Fig. 4C) [40].
While only weak correlation was uncovered between BCL2L13 and the known kidney cancer related genes, we found that voltage-dependent L-type calcium channel subunit beta-1 (CACNB1) and numb-like protein (NUMBL) are the two BCL2L13 negatively correlated genes. Compared to the downregulation of BCL2L13, CACNB1 and NUMBL were both significantly upregulated in ccRCC and pRCC, and also affected the prognosis of these patients (Additional file 8).

BCL2L13 regulates metabolism pathway in ccRCC and pRCC
BCL2L13 has previously reported to participate in several physiological processes. For example, down-regulated BCL2L13 inclined to relieve brain injury induced by ischemia/reperfusion (I/R) [42]. Mitochondrial dynamics and biogenesis, which is also regulated by BCL2L13, could facilitate the browning process of preadipocytes  [44]. Moreover, low expression of BCL2L13, which was related to weakened oxidative phosphorylation and enhanced glycolysis, was often found in cancer cells [45,46]. Thus, the cellular functions of BCL2L13 might have latent impact on the poor prognosis of ccRCC and pRCC.
Co-expressed genes of BCL2L13 were investigated for comprehensive grasp [47]. Filtered by |Spearman's r|> 0.4, 519 and 1318 BCL2L13 co-expressed genes were found in ccRCC and pRCC respectively (Fig. 5A, Additional  file 14). KEGG pathway enrichment analysis manifested the physiological characteristics of these genes, including Huntington disease, citrate cycle and substance metabolism for ccRCC (Fig. 5B), and oxidative phosphorylation, citrate cycle, substance metabolism and Huntington disease for pRCC (Fig. 5C). No substantial differences were found between ccRCC and pRCC. While many of these pathways of BCL2L13 co-expressed genes are related to citrate cycle and substance metabolism, these data suggested that BCL2L13 regulated energy and metabolism might be involved in its prognostic role in ccRCC and pRCC patients. ccRCC was selected for further mechanism exploration, because of the more significant prognostic role of BCL2L13 low expression in this kind of cancer.

BCL2L13 has positive correlation with SLC25A4 (ANT) in ccRCC
The PPI network and hub genes were analyzed for BCL2L13 regulated metabolism, and a small-scale of PPI network was found based on this analysis (Fig. 6A). First, a PPI of BCL2L13 and HSP60 (heat shock protein 60, encoded by HSPD1) was found, because BCL2L13 is anchored on the outer mitochondria membrane, while HSP60 is distributed in the mitochondrial matrix, resulting in a fluorescence co-localization [17]. Second, BCL2L13 had PPI connections with UBC, GABARAPL2 and APP respectively, implying that it may engaged in mitophagy [48][49][50][51]. Third, BCL2L13 was reported to interact with ANT, the protein localized in mitochondrial inner membrane. DMNC, which can be employed for some covert hubs, was further used for the essential genes calculating in Cytoscape, for its higher hit rate to key nodes compared to general degree method [52]. Through this analysis, SLC25A4 was found to be one of the senior hub genes that mediate physiological activity of BCL2L13 (Fig. 6B).
ANT is in charge of exchanging ADP for ATP through MPTP, executing a fundamental role in mitochondrial respiration [53]. ANT could induce mitophagy independent of its ADP/ATP exchange activity, though inhibition or ablation of ANT culminate in disparate phenotypic mitophagy [54]. Studies have evinced that ANT substantially interacts with BCL2L13, mediating its pro-apoptotic activity, while further analysis also supported their correlation (Fig. 6C) [17,55,56].

BCL2L13 overexpression promote apoptosis in ccRCC cells 786-0
786-0 cells were used to study the pro-apoptotic activity of BCL2L13. Compared to HEK (human embryonic kidney) 293 T cells, the expression of BCL2L13 is much lower in clear cell renal cell carcinoma cell line 786-0 ( Fig. 7A), consistent with the silico-based analyses. Moreover, the IHC results also indicated a low expression of BCL2L13 in tumor tissues (Fig. 7B). After transient transfection for 48 h, BCL2L13 induces apoptosis in 786-0 cells, characterized by increased cleavage of poly-(ADPribose) polymerase (PARP) and also the percentage of Annexin V/PI positive cells (Fig. 7C-D, Additional files 9, 10). BCL2/BCLx L /BCLw inhibitor ABT-263 was used to induce mitochondrial mediated apoptosis [57]. Under the treatment of ABT-263 on BCL2L13 overexpressed 786-0 cells, cleaved PARP and the proportion of Annexin-V/PI positive cells were more apparent (Fig. 7C-D, Additional files 9, 10). These data hint that BCL2L13 performs proapoptotic function in ccRCC cells.

Discussions and conclusions
BCL2L13 belongs to BCL2 protein family, and possesses complete BCL2 homology (BH) 1-4 domains. The BHNo domain embedded between BH regions and the C-terminal transmembrane motif endows BCL2L13 with some  non-canonical characteristics [58][59][60]. BCL2L13 was reported to induce mitochondria fragmentation that in favor of the caspase activation cascade [55]. In addition, BCL2L13 has been shown to accelerate mitophagy by binding to microtubule-associated protein 1 light chain 3 (MAP1LC3), suggesting a role as a mitophagy receptor for the quality control of mitochondria [60,61]. BCL2L13 is reported to promote apoptosis, but this function doesn't relate to any of its four BH domains [14,17]. The pathological action of BCL2L13 will be the next focal point in this specialism.
Declining BCL2L13-directed poor prognosis in ccRCC is independent of patients' gender or tumor grade. On the other hand, in ccRCC and pRCC, BCL2L13 has weak correlation with the genes mutated in kidney cancer or the genes associated with inherited kidney cancer predisposing syndrome [62,63], as well as the BCL2L13 related miRNA. That miRNA may not the proximate cause of reduced BCL2L13 in cancerous renal tissues. Although BCL2L13 does not show strong correlation with these genes or miRNA, SLC25A4 (ANT) exhibits high correlation with BCL2L13, supposed to play as one of the hubs involved in BCL2L13-mediated prognostic consequence of kidney cancer [17]. And the attenuated BCL2L13-ANT pathway may be a possible reason for poor prognosis of ccRCC, that is different from its performance in GBM cells [15].
In addition, less genetic alterations of BCL2L13 in ccRCC and pRCC are observed, complied with the data from TCGA, suggesting that BCL2L13 is relatively genetic stable in kidney cancer patients. Shallow deletion of the copy number in a fraction of patients may result in the low expression of BCL2L13 in pRCC, while promoter hypermethylation of BCL2L13 probably accounts for the low expression in ccRCC. Similar functions of BCL2L13 are presented in ccRCC and pRCC by KEGG pathway enrichment analysis. However, it remains an attractive hypothesis that BCL2L13 maintains a linear dose-effect relationship with its physiological activity.
Taken together, BCL2L13 may act as an independent and desirable prognostic marker for ccRCC and pRCC. In addition to that, the functions of CACNB1 and NUMBL seem antagonistic toward BCL2L13 activity, while further