Full-Length Enrich c-DNA Libraries-Clear Cell-Renal Cell Carcinoma

Clear cell renal cell carcinoma (ccRCC), the most common subtype of RCC, is characterized by high metastasis potential and strong resistance to traditional therapies, resulting in a poor five-year survival rate of patients. Several therapies targeted to VEGF pathway for advanced RCC have been developed, however, it still needs to discover new therapeutic targets for treating RCC. Genome-wide gene expression analyses have been broadly used to identify unknown molecular mechanisms of cancer progression. Recently, we applied the oligo-capping method to construct the full-length cDNA libraries of ccRCC and adjacent normal kidney, and analyzed the gene expression profiles by high-throughput sequencing. This paper presents a review for recent findings on therapeutic potential of MYC pathway and nicotinamide N-methyltransferase for the treatment of RCC.


Introduction
Renal cell carcinoma (RCC) represents 3% of all human malignancies worldwide with an increasing incidence and accounts for 85% of renal cancers, resulting in close to 78,000 deaths annually [1][2][3][4][5]. The most common subtype, clear cell RCC (ccRCC), which originates from the proximal tubule epithelium, is mostly sporadic, unilateral, and unifocal [6]. RCC cells have a poor response to traditional therapies, such as the chemotherapy, hormonal therapy, and radiation therapy [7][8][9][10]. These properties result in a poor prognosis and low five-year survival rate of RCC patients.
Surgery remains to be the main therapeutics for treatment of RCC [11]. It has been demonstrated that the activation of hypoxia-inducible factor-α (HIF-α) turns on the genes, such as VEGF and PDGF, which are responsible for the progression of ccRCC, providing potential targets for advanced ccRCC [12].
Gene expression analysis appears to be an important tool for studying cancer pathogenesis and progression. Markers related to tumor proliferation, growth, angiogenesis, and loss of cell adhesion have been evaluated for their potential as prognostic factors. Numerous reports have investigated the differential gene expression profile between normal and tumor tissues using the high-throughput technologies, including cDNA microarray, cDNA subtraction, and serial analysis of gene expression [13][14][15]. Many studies have applied these methods to analyze the genome-wide changes in ccRCC, providing specific gene expression signature, the potential biomarkers, and prognostic factors [16][17][18][19][20][21], whereas Lenburg et al. pointed out a diverse gene expression results by microarray method published previously [22]. Recently, two international projects, The Cancer Genome Atlas (TCGA) and International Cancer Genome Consortium (ICGC), have been organized to provide the platform for the integration and comparison of cancer genomic abnormalities identified by independent research groups [23][24][25].
The oligo-capping method developed by Dr. Suzuki and Dr. Sugano at the Institute of Medical Science of Tokyo University was applied to construct the full-length enriched cDNA library for the analysis of transcriptional start sites of genes [26,27]. Yamada et al. utilized the oligocapping method to explore gene expression profiles of hepatoblastomas and the corresponding normal livers and identify a poor-prognostic indicator for hepatoblastomas [28]. 2 Journal of Oncology To investigate the differentially expressed genes in ccRCC, the full-length enriched cDNA libraries of ccRCC and normal kidney tissues were constructed and sequenced. We identified 383 differentially expressed genes, and further confirmed the differential expression of 20 ccRCC-associated genes, which have not been reported to be dysregulated in ccRCC. By functional network analysis using 383 differentially expressed genes, MYC pathway was identified to be activated in neoplastic disease [29]. By using MYC siRNA, the importance of MYC was further demonstrated in ccRCC cells.
Additionally, nicotinamide N-methyltransferase (NNMT) was identified as the ninth of 201 upregulated genes with 65.2-fold increase in full-length cDNA-enriched libraries of ccRCC tissues. We showed that NNMT contributes to MMP-2 expression to induce cell invasiveness via PI3K/AKT/ SP1 pathway, suggesting NNMT as a novel invasive-related gene for ccRCC. Our results elucidate the novel functions of NNMT in ccRCC and suggest that NNMT could be considered as a potential therapeutic target of ccRCC.

Gene Expression Profiles of ccRCC and Normal Kidney
To construct the full-length enriched cDNA libraries from two tissues of ccRCC (Grade II/Stage I; Grade II/Stage II) and adjacent normal kidney, the oligo-capping method developed by Y. Suzuki and S. Sugano was performed. We successfully sequenced 19,425 and 12,400 colonies of ccRCC and normal kidney cDNA libraries, respectively. The sequencing results were blasted to the UniGene database of NCBI using the BLAST program. The blast result with the score more than 200 was denoted as a gene, and a total of 4,356 and 3,055 genes were identified in the fulllength enriched cDNA libraries from ccRCC and normal kidney tissues, respectively. The gene expression patterns of ccRCC and normal kidney were further compared to identify differentially expressed genes. By the definition that the genes with the colony number in ccRCC three-fold higher or lower than that in normal kidney were considered as the ccRCCassociated genes, 201-upregulated and 182-downregulated genes were identified [30].

Classification of ccRCC-Associated Genes by Cellular Function
The differentially expressed genes in ccRCC were classified into groups by their cellular functions according to gene ontology (GO) term in NCBI [30]. We observed that many metabolism-related genes were differentially expressed in ccRCC, such as NNMT, lysyl oxidase, transglutaminase 2, heme oxygenase 1, argininosuccinate synthetase, alcohol dehydrogenase 6, and phosphoenolpyruvate carboxykinase 1, indicating that the ccRCC cells might lose their normal functions as the kidney cells, and the metabolismrelated genes might play a role in cancer progression. Additionally, the differential expression of several signal transduction-related genes was found, including secreted frizzled-related protein 1, insulin-like growth factor-binding protein 3, regulator of G-protein signaling 1, and frizzledrelated protein, implying the activation of signaling pathways associated with the process of carcinogenesis. The genes involved in cellular proliferation were also found to be dysregulated. For instance, transforming growth factor (beta-induced), CD74 antigen, alpha-2-glycoprotein 1, inhibin, cyclin D1, and cell division cycle 42 were differentially expressed. The results might support the high proliferation ability of ccRCC cells. Interestingly, we also observed the altered expression of genes which function as cytoskeleton assembly, protease or protease inhibitor, such as gelsolin, MMP-2, cathepsin S, kininogen 1, alpha-1 antiproteinase member 5 and tissue inhibitor of metalloproteinase 3, providing the strong invasive activity of ccRCC cells.

Differential Expression of the Novel ccRCC-Associated Genes in ccRCC
The differential expression of the ccRCC-associated genes, which have not been reported to be dysregulated in ccRCC, was further studied in ccRCC tissue pairs by Q-PCR analysis (Table 1). Interestingly, the expression of the modulators of the Wnt signaling pathway, secreted frizzled-related protein 2 and 4, were significantly upregulated 34.9-, and 4.3-fold, respectively. In addition, the detoxification-related gene, glutathione S-transferase A3 was found to be significantly downregulated in ccRCC, indicating an elevated oxidative stress in ccRCC tissues. The overexpression of secreted protein acid rich in cysteine (SPARC) has been demonstrated in various cancers. Previous reports have shown that SPARC is involved in tumor development [31,32], and in this study we also observed the upregulation of SPARC expression in ccRCC, suggesting the important role of SPARC in ccRCC (Table 1).

Activation of MYC Pathway in ccRCC
To identify the deregulated cellular pathways in ccRCC, functional network analysis was performed with the 383 ccRCC-associated genes by using Ingenuity Pathway Analysis, which has been used to identify significant pathways related to tumorigenesis in several tumors [33,34]. The list of the ccRCC-associated genes was uploaded to the Ingenuity system and used as focus genes to produce the biological networks. The P values were calculated to represent the statistical significance that the focus genes in a network are randomly found together. The results of Ingenuity Pathway Analysis revealed that 17 networks were statistically significant and contained more than 10 ccRCC-associated genes. The top-ranking biological functions of each network were assigned according to Ingenuity Pathways Knowledge Base. The five biological networks with the highest score were found to be functionally related to cancer progression, such as cell cycle, cellular growth, or cellular movement ( Table 2). Interestingly, the top functional network with the highest score was composed of 35 differentially expressed genes and was assembled around the oncogene, MYC, with a set of MYC-target genes. The crucial role of MYC has been demonstrated in numerous kinds of human tumors and the regulation of cancer-related gene expression [35]. The upregulation or amplification of MYC in ccRCC has been previously reported [36][37][38][39]. Recent reports showed that HIF-2α promotes the transcriptional activity of MYC, whereas HIF-1 suppresses it [40,41]. We further identified the MYC transcriptional targets among the 383 differentially expressed genes in ccRCC by using Ingenuity Pathways Knowledge Base. The differentially expressed genes with the connection to MYC labeled as expression (E), transcription (T), or protein-DNA interaction (PD) were considered as the MYC-target genes. As shown in Figure 1, 37 differentially expressed genes were found to be the MYC-target genes, suggesting that the MYC pathway members may be activated in ccRCC tissues, indicating that MYC pathway is activated and plays an important role in carcinogenesis of ccRCC.
Furthermore, the upregulated expression of MYC was examined in 44 ccRCC tissue pairs by Q-PCR analysis. The average expression level of MYC in ccRCC tissues was significantly higher than that in normal kidney tissues. The results of immunohistochemical staining analysis also demonstrated that 24 (96%) of 25 ccRCC tissues showed the positive staining for MYC, including 15 cases with 3+ staining, 7 cases with 2+ staining, and 2 cases with 1+ staining, but negative staining for nonneoplastic renal tubules. The expression levels of MYC were also examined in 786O, 769P, A498, ACHN, and Caki-1 cells, which were derived from patients with ccRCC, and the results revealed that all of these ccRCC cell lines displayed a significant upregulation of MYC as compared with that of a pool of 5 normal tissues. These results suggest that the expression of MYC is markedly increased in ccRCC tissues and cell lines [30].  Cancer, tumor morphology, cardiovascular system development and function The score represents the negative log of P value that indicates the likelihood of the focus genes in a network being found together due to chance. The number of focus genes reflects number of the differentially expressed genes of ccRCC in a network generated by IPA.  To investigate whether the expression of MYC pathway signature is enhanced due to MYC activation, the expression levels of the MYC-target genes, BCL2, CCND1, PCNA, PGK1, and VEGFA, which have been reported to be functionally related to cancer progression [42,43], were further investigated in 44 ccRCC tissues by Q-PCR analysis. As shown in Table 3, the significant upregulation of these MYC pathway members was demonstrated. By analysis of the Pearson correlation method, we observed that the expression levels of BCL2, CCND1, PCNA, PGK1, and VEGFA were correlated with the levels of MYC in the tumor tissues of ccRCC (Table 3). These results indicate that the MYC pathway signature is activated and is associated with MYC upregulation in ccRCC tissues.

Suppressing the Expression of MYC-Target Genes by MYC siRNA
We have demonstrated the correlation between the expression levels of MYC and the MYC-target genes in ccRCC tissues (Table 3). Therefore, whether the expression of the MYC-target genes is suppressed by the knockdown of MYC expression in ccRCC cells were investigated. The results of Q-PCR analysis revealed that treatment of MYC-specific siRNA significantly reduced expression levels of MYC-target genes BCL2, CCND1, PGK1, PCNA, and VEGFA in 769P cells [30]. It has been demonstrated that MYC is able to bind to specific DNA sequences (E-box motif, CACGTG) to function as a transcription factor [31]. Therefore, whether  MYC could be directly recruited to the promoter regions of these MYC-target genes was studied. The potential Ebox sites on the promoters of BCL2, CCND1, and PCNA were identified by blasting the sequence, CACGTG, with the potential promoter regions of BCL2, CCND1, and PCNA, and the primer pairs to specifically amplify the promoter regions containing the potential E-box sites were generated using Beacon Designer 4 program. The primers to amplify the E-box sites in the promoters of PGK1 and VEGFA have been described in previous studies. The Chromatin immunoprecipitation (ChIP) assay was carried out using these primers to amplify the potential MYC-binding regions in BCL2, CCND1, PCNA, PGK1, and VEGFA promoters. The endogenous MYC was observed to bind onto the promoter regions of BCL2, CCND1, PCNA, PGK1, and VEGFA [30]. These results indicate that MYC expression is able to enhance the expression levels of MYC-target genes by binding to their promoters in ccRCC cells. Previous studies have demonstrated that MYC plays a role in cellular transformation [29]. To address whether the expression of MYC is required for the malignant ability of ccRCC cells, the effects of the knockdown of MYC expression by RNAi technique was investigated in ccRCC cell lines, 769P and Caki-1 cells. As displayed in Figure 4(a), the expression level of MYC in 769P and Caki-1 cells were markedly suppressed by MYC-specific siRNA in a dose-dependent manner. The results of MTT assay revealed that the proliferation rate of 769P and Caki-1 cells was significantly reduced by knockdown of MYC expression in a dose-dependent manner (Figure 4(b)). Additionally, treatment with 20 nM of MYC-specific siRNA dramatically suppressed 95% and 91% of anchorage-independent growth ability in 769P cells and Caki-1 cells as shown by soft agar assay, respectively [30]. These results indicate that RNAi-mediated knockdown of MYC strongly suppresses the proliferation and anchorageindependent growth of ccRCC cells.
It has been shown that MYC promotes cell cycle progression of cancer cells to enhance an uncontrolled cellular proliferation in various malignances [43]. Therefore, we further investigate whether the knockdown of MYC expression has the capacity of arresting cell cycle progression in ccRCC cells. Cell cycle distribution of 769P cells was analyzed by PI staining and flow cytometry analysis after treated with MYC siRNA. The results showed a significant increase of G0/G1 phase population of MYC siRNA-transfected cells as compared with the control cells. Additionally, the population in the process of DNA synthesis (S phase) was markedly decreased in 769P cells treated with MYC-specific siRNA [30]. These results show that a reduction of MYC expression disrupted the G1/S phase transition of 769P cells.

Overexpression of NNMT in ccRCC Tissues
Among these differentially expressed genes identified by analysis of full-length cDNA libraries in ccRCC, NNMT was one of the most upregulated genes with 65.2-fold overexpression. Previous studies have demonstrated that NNMT is overexpressed in ccRCC tissues and has the potential to be a ccRCC biomarker [44,45]. To investigate whether NNMT is overexpressed in ccRCC tissue pairs, Q-PCR analysis was used to examine the mRNA levels of NNMT in these tissue pairs. The results showed that the expression of NNMT in 27 of 33 ccRCC tumors was highly upregulated (T/N > 3.0) by an average of 52.8-fold compared with that in adjacent normal kidney tissues (Figure 2(a)). The results of Western blot analysis also demonstrated a marked increase in NNMT protein levels in ccRCC tissues (Figure 2(b)). These results suggest that NNMT is overexpressed in ccRCC, and it was also reported that the overexpression of NNMT was found in colorectal cancer, papilloma thyroid cancer, and gastric cancer [46][47][48].

Identification of the Potential Role of NNMT in ccRCC
Interestingly, Wu et al. recently demonstrated that a correlation between NNMT and cancer cell migration has been reported in bladder cancer [49]. We further investigated the expression levels of genes related to cell invasiveness in HEK293/NNMT cells using Q-PCR analysis, and MMP-2 expression was found to be enhanced by NNMT overexpression. As shown in Figure 3(a), MMP-2 expression of HEK293/NNMT cells was enhanced 4-fold compared with that of HEK293/vector cells. The results of gelatin zymography and western blot analyses demonstrated a significant increase of MMP-2 protein level in HEK293/NNMT cells (Figure 3(b)). Moreover, by treating siRNA specific to NNMT gene, the expression of NNMT was significantly suppressed and was accompanied by a reduction of MMP-2 expression in HEK293/NNMT cells (Figure 3(c)). The effects of silenced NNMT expression on the MMP-2 expression were further investigated in ccRCC cell lines. Figure 4(a) indicated that 786O and Caki-1 cells (invasive ccRCC cell lines) had a higher NNMT and MMP-2 expression than those of 769P and Caki-2 cells (primary ccRCC cell lines). The results of siRNA treatment showed that the knockdown of NNMT expression resulted in the suppression of MMP-2 expression in 786O and Caki-1 cells (Figure 4(b)). These results suggest that NNMT expression is involved in MMP-2 expression.   Moreover, we examined whether the expression of NNMT and MMP-2 was correlated in ccRCC tissues, and the results indicated a positive correlation between the mRNA levels of NNMT and MMP-2 in ccRCC tissues based on the statistical analysis using the Pearson correlation method (r = 0.47). The results of immunohistochemistry analysis also demonstrated that NNMT was overexpressed in 76% of ccRCC tissues, whereas the expression of MMP-2 was upregulated in 58% of ccRCC tissues. By the analysis of the Pearson's chi-square test, the expression of NNMT and MMP-2 was observed to be significantly correlated in ccRCC tissues (Figure 5(a)), and the representative cases were shown in Figure 5(b). These results indicate that the expression of NNMT and MMP-2 is correlated in ccRCC tissues.
To study the crucial region in the MMP-2 promoter responding to NNMT-mediated MMP-2 expression, a 1,716 kb genomic fragment containing the upstream region of MMP-2 gene (WT) and a serial of successive 5 deletions (D1 to D7 constructs) were cloned into pGL3 luciferase reporter vector [50]. pGL3-MMP-2 promoter plasmid and pGL3 control reporter plasmid were transfected into cells, and the promoter activities were examined using the luciferase reporter assay. As shown in Figure 7(a), a 2-fold increase of the transcriptional activity of the full-length MMP-2 promoter in HEK293/NNMT cells was observed than that in control cells. Additionally, about 4-fold decrease of transcription activities of D6 and D7 mutants, but not D1-D5 mutants, was found as compared with that of the full-length MMP-2 promoter in HEK293/NNMT cells. It indicates that the SP1-binding elements in MMP-2 promoter are required for NNMT-mediated MMP-2 expression. The importance of SP1-binding elements was further examined by using a MMP-2 promoter construct with mutated SP1binding elements (bp −94 to −64). The results of luciferase reporter assay revealed that the mutation of SP1-binding elements caused a 4-fold reduction of MMP-2 promoter activity in HEK293/NNMT cells (Figure 7(b)). These results suggest that the SP1-binding region of MMP-2 promoter plays a crucial role for NNMT-mediated MMP-2 expression. To study whether SP1 plays a role in NNMT-mediated MMP-2 expression, siRNA specific to SP1 gene was used. As revealed in Figure 8(a), the expression of MMP-2 was markedly reduced by the treatment of SP1 siRNA in HEK293/NNMT, 786O. and Caki-1 cells. Moreover, mithramycin, which has been shown to prevent SP1 binding to gene promoters [51], was used to suppress SP1 transcriptional activity. The results showed that treating with mithramycin apparently suppressed the expression of MMP-2 in HEK293/NNMT, 786O, and Caki-1 cells (Figure 8(b)). It suggests that SP1 is an important factor for MMP-2 expression activated by NNMT.
Furthermore, CHIP assay was carried out to investigate the role of PI3K/AKT pathway in SP1-binding to MMP-2 promoter. The results demonstrated a 2.9-fold increase of SP1 association to the SP1 binding elements of the MMP-2 promoter in HEK293/NNMT cells compared with HEK293/vector cells (Figure 9). Treating with LY294002 or AKT inhibitor IV remarkably reduced the association between SP1 and the MMP-2 promoter in HEK293/NNMT cells (Figure 9). These results strongly  implicate that PI3K/AKT/SP1 pathway participates in MMP-2 promoter activation induced by NNMT expression.

Involvement of MMP-2 in NNMT-Mediated Cell Invasiveness
Whether NNMT-mediated MMP-2 expression causes cell invasion was investigated by matrigel invasion assay. As The results of matrigel invasion assay indicated that the inhibition of MMP-2 activity significantly suppressed the invasive activity in HEK293/NNMT cells (Figure 10(b)). Additionally, the effects of NNMT siRNA on the invasiveness of ccRCC cell lines were studied. The results showed that treating with NNMT siRNA resulted in about 70% reduction of the invasive activity in 786O and Caki-1 cells (Figure 10(c)). Moreover, NNMT siRNA-suppressed invasiveness was restored by the treatment of MMP-2 active protein ( Figure 10(c)). These results indicate that MMP-2 activity is required for the invasiveness of NNMT-overexpressing cells.

Effects of NNMT shRNA on Tumor Growth and Metastasis of ccRCC Cells in NOD-SCID Mice
Whether the reduced NNMT expression inhibits the tumor growth and metastasis of ccRCC cells was investigated by the experiment of NOD-SCID mice. 786O cells with stably expressing NNMT or luciferase shRNA were generated by using puromycin selection after infected with the lentivirus containing NNMT or luciferase shRNA. As demonstrated by western blot analysis, NNMT shRNA markedly suppressed about 90% of NNMT and 70% of MMP-2 expression in 786O cells (Figure 11(a)). The invasive activity of 786O/NNMT shRNA was reduced to 25% of that of 786O/ control shRNA (Figure 11(b)). By subcutaneous injection into the hind limb of NOD-SCID mice, 786O/NNMT shRNA cells developed less tumor mass than 786O/luciferase shRNA cells in NOD-SCID mice (Figure 11(c)). The lung metastatic activity of 786O cells with or without NNMT knockdown was further studied by using intravenous injection into lateral tail vein of NOD-SCID mice. As shown  Figure 11: Effects of NNMT shRNA on the growth and lung metastasis of ccRCC cells in NOD-SCID mice. Western blot analysis (a) and matrigel invasion assay (b) were performed to examine NNMT and MMP-2 expression and invasive activities of 786O/NNMT shRNA and 786O/control shRNA cells. Actin was used as loading control. (c) 4 × 10 6 of 786O/NNMT shRNA or 786O/luciferase shRNA cells were s.c. injected into the hind limb of NOD-SCID mice (n = 6). Tumor volume was measured every three days. Data are presented as the mean of triplicate replications. S.D. is indicated by error bars. * * P < 0.001 versus control cells. (d) 1 × 10 6 of cells in 200 μL of serum-free DMEM medium were intravenously injected into the lateral tail vein of NOD-SCID mice (n = 6). After 6 weeks of injection, the mice were sacrificed, and the metastatic nodules of lungs were counted. Data are presented as the mean ± SD. * P < 0.05 versus control cells.
in Figure 11(d), the lung metastasis of 786O cells was significantly suppressed 90% by NNMT shRNA as compared to control shRNA. These results indicate that silenced NNMT expression inhibits tumor growth and lung metastasis of ccRCC cells.

Conclusions
Gene expression profiling is one of important methods for studying multifactor diseases and for indentifying the potential biomarkers as the target genes for cancer chemotherapy.
We apply the full-length enriched cDNA method to construct the first full-length enriched cDNA libraries of ccRCC and normal kidney tissues. The advantages of present method are (1) the number of upregulated or downregulated genes is obtained more precisely than that of microarray, and (2) the mutations such as point mutation, insertion or deletion, and SNP sites can be identified from the sequencing of fulllength enriched cDNA libraries.