Genetics of Renal Tumors

Kidney and urinary tract cancers accounted for a total of 16936 cases and 6764 deaths in 2007 in Japan (Matsuda et al., 2012), which is roughly 2% of all cancers. Renal cell carcinoma (RCC) is the most common type of kidney cancer, and is classified into three major subtypes, clear cell RCC, papillary RCC and chromophobe RCC, representing 80, 10, and 5% of all RCCs, and the majority of renal tumors are sporadic although 2-4% are hereditary (Hagenkord et al., 2011).

also discuss a methodology for collecting information on multiple gene functions with a simple pathological system (Section 4).

Genes associated with renal tumors
While kidney cancer ranked 9th in 2002 in the European Union and the United States (Baldewijns et al., 2008), its mortality rate was not high in Japan (12th in 2002 and 2007: Matsuda et al., 2012). Although this difference could be attributable to risk factors such as smoking, hypertension and long-term dialysis, there might be a contribution of genes associated with the cancer. In spite that RCC shows a poor survival rate (less than 19%) for patients with metastasis, molecular pathological tests, such as those dividing good and poor prognosis groups, have not been established (Stewart et al., 2011). A lack of such effective tests may be one of the reasons why the mortality rate in Japan has been gradually increasing from 1.8% (2002) to 2.0% (2007).
A large majority of RCC cases are sporadic and only 2-4% are hereditary. There are cases where gene expression profiling cannot distinguish between them (Beroukhim et al., 2009), suggesting common genetic factors between them. Several genes are known to be associated with RCC, such as VHL, TSC1 and TSC2, which play different roles in the mechanism of cancer and so have different advantages in diagnostics/therapeutics. The information about genes can be categorized by the levels of genomics, transcriptomics, proteomics and others including metabolomics, and used to understand the mechanism of cancer, to support diagnostic or therapeutic processes. In this section, we focus on the roles and merits of these genes.

Genes associated with tumorigenesis
Since a majority of sporadic cancers originate from a recessive mutation that causes a loss of function of a particular type of gene, loss of heterozygosity (LOH) is an important step in the disabling of a functional gene (or a wild-type allele) to give a mutated and cancer phenotype. Such genes are termed tumor suppressor genes, and so far, more than 100 have been reported (Fearon, 2002;Polinsky, 2007). Among them, twenty well-characterized genes showed both familial and sporadic phenotypes (Sherr, 2004). Since a cancer phenotype can be revealed by morphological changes, growth stimulation, gaining immortality and/or others, there are quite a few functions associated with tumor suppressor genes. Thus, it is easier to examine tumorigenesis in association with genomic status, mutations and/or epigenetic modifications, by analyzing the loci specific to RCC.

VHL gene
The gene best known to be associated with RCC is the von Hippel-Lindau (VHL) gene, whose inactivation accounts for nearly 100% of hereditary cases and sporadic clear cell RCC cases (Baldewijns et al., 2008). This gene was found by positional cloning from the locus associated with the VHL disease, a familial syndrome accompanying cancer in the eye, brain, spinal cord, kidney, pancreas and adrenal glands. The VHL gene encodes the 30-kDa protein VHL, 213 amino acid residues long, and is implicated in the regulation of hypoxia-inducible factors (HIFs) (Maher et al., 2011). The VHL protein forms a complex with elongin B, elongin C and cullin-2, and the complex has ubiquitin ligase E3 activity and is involved in the ubiquitination and degradation of HIFα, the α subunits of transcription factors HIF-1 and HIF-2, which form a dimer with HIFβ and regulate the transcription of hypoxia-inducible genes such as those for VEGF (vascular endothelial growth factor), PDGF (platelet derived growth factor) and TGFα (transforming growth factor α) (Kondo and Kaelin, 2001;Kaelin, 2009; Fig. 1). However, the cancer found in VHL disease is sporadic and the lifetime risk of RCC in VHL disease patients is about 70% (Maher et al., 2011). So, it is reasonable to assume that additional genes are involved in RCC and the mutations in VHL are not the definitive cause of RCC, which is one of the reasons to explore new genes and genetic loci (see below). Meanwhile, the status of the VHL gene is important for the treatment of VHL disease and kidney cancer patients. HIFresponsive gene products, such as VEGF and PDGF, activate the angiogenesis of tumors and therefore are good therapeutic targets. Inhibitors of VEGF and PDGF, sunitinib and sorafenib, have been approved by the US Food and Drug Administration (Kaelin, 2009).

MET gene
The MET protooncogene was found in hereditary papillary RCC without mutations in the VHL gene (Schmidt et al., 1997). MET encodes a membrane receptor (MET) for hepatocyte growth factor (HGF). MET has tyrosine kinase activity, and HGF activates this kinase activity and initiates signaling for mitogenesis and migration (Fig. 1 (Parry et al., 2001) and therefore its role is not completely clear yet.

PBRM1 gene
Several genes, UTX (or KDM6A), JARID1C (or KDM5C) and SETD2, were found in close association with clear cell RCC by a recent technology of the next-generation sequencing (Dalgliesh et al., 2010). As these genes are related with the methylation status of lysine residues of hitone H3, further mutation studies were conducted to identify a SWI/SNF chromatin remodeling complex gene, PBRM1, to be frequently (over 40%) mutated in clear cell RCC (Varela et al., 2011). PBRM1 is mapped to chromosome 3p21 and encodes the BAF180 protein, a chromatin targeting subunit of a SWI/SNF chromatin remodeling complex, which regulates replication, DNA repair and cell proliferation/differentiation. Knock-down of this gene enhanced colony formation and migration of cancer cells, suggesting this gene to be a tumor suppressor gene. Further studies are needed to reveal a mechanism of cancer involving PBRM1 and to find its clinical application. From the analysis of their mutations, this gene is considered as a tumor suppressor gene (Sudarshan et al., 2007). Although the mechanism that leads FH alterations to cancer is not clearly understood, there is a link between fumarate dysregulation and impaired HIF hydroxylation (Isaacs et al., 2005).

Genes related to hereditary renal cancer syndromes
FLCN, on the other hand, is the gene responsible for Birt-Hogg-Dubé (BHD) syndrome, which is a rare autosomal dominant disease including kidney tumors, predominantly chromophobe RCC. Mutations in this gene were found in approximately 80% of BHD kindreds and loss of expression of this gene were frequently found in kidney tumors from BHD patients, suggesting this gene to be a tumor suppressor gene (Baldewijns et al., 2008).

Other genes
Several genes were recently implicated in association with RCC, including BAP1, SETD2 and NF2, by means of advanced technologies such as the next-generation sequencing, a microarraybased analysis and a mouse transgene analysis. NF2 was identified as a tumor suppressor gene by the analysis of knock-out mice (Morris and McClatchey, 2009). The mice developed kidney tumors in 6-10 months with characteristics of hyperactive epidermal growth factor receptor (EGFR) signaling. Merlin, the NF2 gene product, was implicated in suppressing tumorigenesis by inhibiting hyperactivated EGFR signaling.

Genes implicated in diagnostic markers and therapeutic targets
The recurrence of RCC is 20 to 40%, depending on the stage and grade of tumor (Chin et al., 2006). So, it is important to understand the genes (and their products) associated with progression/metastasis to predict the outcome of cancer. The classification of RCC subtypes is apparently not possible by a single marker, but could be done using combinations of markers such as vimentin, epithelial cell adhesion molecule (EpCAM), glutathione S-transferase α (GSTα), carbonic anhydrase II (CA II), cytokeratin 7 (CK7) and cluster of differentiation 10 (CD10) (Stewart et al., 2011).
Another pathway for potential makers is the mTOR pathway (Fig. 1). The main cascade of this pathway is PI3K/AKT/mTOR, which mediates signals by activating phosphoinositide 3-kinase (PI3K) through kinases such as receptor tyrosine kinases to generate phosphatidylinositol (3,4,5)-trisphosphate (PIP3), which further activates AKT via phosphorylation and phospho-AKT activates mTOR complex 1 (mTORC1) through inhibition of the TSC1/TSC2 complex (Allory et al., 2011). Then, mTORC1 phosphorylates proteins such as P70-S6 kinase and activates protein synthesis and cell proliferation. Importantly, HIF-1α expression is dependent on mTORC1 signaling (Toschi et al., 2008). Potential markers in this pathway include P70-S6 kinase, PTEN (a phosphatase that decreases PIP3) and phospho-AKT. We used a genome-subtraction technique, or the in-gel competitive reassociation method (Kiyama et al., 1995;Rodley et al., 2003), for cloning the sites of LOH that occurred in a RCC genome by subtracting normal DNA from cancer DNA of the same patient (Hatano et al., 2001). The minimum size of LOH (caused by hemizygous deletions) detected by this method was roughly 50 kb. This resolution was made possible by MseI, which recognizes TTAA, a sequence appearing frequently in human genomic DNA, and completely digests genomic DNA to sizes mostly below 1 kb. Such a high resolution has not been used even in recent genome-wide association studies (see Jacobs et al., 2012, for example). A total of 187 clones were mapped on the chromosomes and a total of 44 candidate regions, where at least two clones were mapped within 5 Mb, were selected and analyzed for mapping the sites of LOH in 61 cancer cases (Table 1). Among them, we found interesting LOH sites at 5q32-q34, 6q21-q22, 8p12 and 9p24, whose frequencies are relatively high among RCC and whose lengths are less than ~10 Mb (

Exploration of new genetic markers
Even though a number of genetic markers have been reported, they are not currently used for the diagnosis of RCC. As discussed in Section 1, this is because understanding a single gene or a few genes is not enough for a diagnosis of sufficient reliability. For diagnosing more complex and more specific states of diseases or disease phenotypes, groups of markers that are able to more accurately distinguish the phenotypes are needed. Such markers should be derived from the direct process of the disease and therefore would represent the signal transduction that occurs within the cell. There are several new technologies which might open the door to a more comprehensive understanding of RCC especially at the level of cellular signaling: array-based genome-wide association studies, microRNA (miRNA) studies and next-generation sequencing-based expression profiling.  Recent advances in high-resolution genomic arrays have enabled us to analyze 1,000 or more disease cases efficiently, and thus to give statistically significant loci associated with the diseases. Such an approach was applied to the study of RCC. A genome-wide association study based on more than 5,000 RCC cases revealed two loci, 2p21 and 11q13.3, to be associated with RCC susceptibility (Purdue et al., 2011). Although the authors claimed these sites to be previously unidentified, both of the loci were actually identified in 2001 (Hatano et al., 2001; Table 1). While the association is statistically significant, the frequencies among RCC cases are not very high (less than 20%), and therefore, it is doubtful that these sites alone can be used for diagnosis. The candidate genes in these loci which contribute to the association are EPAS1 encoding hypoxia-inducible factor-2α (HIF2α) and SCARB1 encoding a scavenger receptor. While HIF2α was known to be associated with RCC though it has not yet been used clinically, SCARB1 is new and its association with RCC may indicate a new signaling pathway. The arraybased genome-wide association technique was also applied to the study of copy-number variations (Krill-Burger et al., 2012).
The study of miRNA is rapidly providing as new information about disease phenotypes. MiRNA, a group of short non-coding RNA with lengths of 19-22 nucleotides, differs from mRNA in that it has a role in gene function, and, while information about mutations is important for mRNA, quantity is mostly emphasized for miRNA. So, while there are cases where mRNA bearing a mutation without a change in its quantity contributes to a disease phenotype, there would be few such cases for miRNA. Naturally, the linkage of a disease to a genomic location reveals in most cases a mutation in a gene. This may indicate that miRNA contributes to quantitative change as a group as a result of changes in transcriptional efficiency caused by alterations to the transcriptional machinery or genomic location/status, or by epigenetic modifications. In contrast to mRNA, however, the quantity of miRNA can be controlled rapidly and specifically, and thus, miRNA could be more advantageous for the rapid control of the amount of specific proteins, which is important in signal transduction. Next-generation sequencing technology was applied to genome-wide expression profiling of miRNA related to clear cell RCC (Osanto et al., 2012). By analyzing 22 RCCs, 100 miRNA differentially expressed between clear cell RCC and matched normal tissues were found. While the biological relevance of these novel miRNAs is unknown, they may be potential diagnostic markers or targets for therapeutics.

Structure of Kank-family genes
The human Kank1 gene was found as a candidate tumor suppressor gene for renal tumors at 9p24, and encodes a protein containing ankyrin-repeats at the C-terminus and coiled-coil motifs near the N-terminus (Sakar et al., 2002). Based on domain and phylogenetic analyses, Kank2, Kank3 and Kank4 were found to form a family with Kank1 . Five repeats of the ankyrin-repeat motif comprise the basic structure of all Kank proteins (Fig. 3A). In addition, each Kank protein contains different combinations of four types of coiled-coil motifs. They also have a conserved region close to the N-terminus, named the KN-motif ; Fig. 3A), which contains a leucine-rich region and an arginine-rich region.  Fig. 3B). Kank1 regulates the Rac1-dependent formation of lamellipodia and the activity of RhoA, resulting in the inhibition of cell migration. This function is mediated through two binding partners of Kank1, 14-3-3 and IRSp53. Kank1 binds to the Akt-phosphrylation motif of 14-3-3θ, 14-3-3γ, 14-3-3η and 14-3-3ε. Interaction between these two proteins is enhanced by growth factors such as insulin and epidermal growth factor (EGF) . This interaction regulates the activation of RhoA through the PI3K/Akt signaling pathway. When a 14-3-3 binding motif is phosphorylated by Akt, 14-3-3 is separated from an activation complex for RhoA, and binds to Kank1 resulting in the inhibition of RhoA activities, and thereby decreases the formation of actin stress fibers and inhibition of cell migration . The coiled-coil domain of IRSp53, which is the site for the interaction with active Rac1, binds to Kank1. Endogenous Kank1 and IRSp53 are co-localized at the site of membrane protrusions such as lamellipodia, which are needed for cell migration. Overexpression of Kank1 inhibits the formation of lamellipodia induced by active Rac1 in NIH3T3 cells, and knockdown of Kank1 enhances the formation. Therefore, Kank1 negatively regulates membrane protrusions at the leading edge of cells, by inhibiting the association between active Rac1 and IRSp53 . Taken together, Kank1 regulates cell migration through inhibition of IRSp53 in Rac1 signaling and inactivation of RhoA activity through PI3K/Akt signaling (Fig. 3B). As the Kank1 locus shows loss of heterozygosity in RCC and the expression of the Kank1 gene is suppressed in RCC, Kank1 may contribute to the malignant transformation of cells such as metastasis. . Although the mechanism is still not clear, Kank1 may block cytokinesis by regulating Rho activity through the interaction with Daam1 (Fig. 3B). Therefore, it may reveal a new mechanism of regulation of cytokinesis and tumor suppression.

Kank-family genes and renal tumors
The According to studies to date, the Kank1 protein may act as a tumor suppressor through inhibition of cell migration and cell cycle. These functions are facilitated by several proteins interacting with Kank1, including 14-3-3, IRSp53, Kif21a and Daam1. Further studies of the interactions of these proteins will help us to understand clearly the role of Kank family proteins in tumorigenesis. Kank1 was found by a genome subtraction method among the genes at 9p24 susceptible to RCC (Sarkar et al., 2002). A devoted study revealed that Kank1 belongs to a fourmember family, has splice variants, and plays a role in cell migration, intracellular transport and cell division, suggesting that Kank1 has a kind of tumor suppressor function . In this section, the expression of the Kank1 protein in renal cancer specimens resected from RCC patients is indicated using immunohistochemical methods, and the relationship between the expression and tumor pathology, patient status, and clinical outcomes is examined.

Expression of Kank1 in RCC
We tried to find a RCC-related gene at 9p24, which lead to the discovery of Kank1. Of nine ESTs analyzed in the 9p24 region, only three (WI-17492, WI-12779 and WI-19184) were expressed in the kidney. The Kank1 gene was associated with WI-12779. This Kank1-associated EST lost its expression in six out of eight cancer cases. Kank1 expression was examined in 5 matched normal kidney and cancer pairs by Western blotting using an anti-Kank1 antibody, which was obtained as mentioned below. Reduced or loss of Kank1 expression in cancer was observed in all 5 cases.

2.
Immunohistochemical study of Kank1 expression in RCC and the relationship between its expression and clinical-pathological outcomes One hundred and five formalin-fixed paraffin-embedded slides including normal renal tubular cells and RCC were subjected to immunohistological staining for Kank1 with a monoclonal antibody. An anti-Kank1 (total Kank1) antibody was generated by a previously reported method . In brief, amino acids 406 to 580 of the Kank1 protein were fused in-frame with the glutathione S-transferase gene in the vector pGEX. After induction of the fusion protein in E. coli, it was purified and used to immunize mice. A mouse hybridoma cell producing an anti-Kank1 antibody was selected and amplified for further use.
The histological subtypes of RCC analyzed here were as follows; 92 clear cell RCCs, 11 papillary RCCs, 5 chromophobe RCCs and 7 other histological types. We compared all histological subtypes with clear cell RCC. The evaluation of positivity of staining was done by two independent examiners, who decided that the sample was positive when more than 30 % of cells were stained with the antibody, weakly positive (±) when 5 to 30 % cells were stained, and negative when less than 5 % cells were stained. There were no differences in the survival curves for clear cell RCC among the groups (Fig. 5). However, when the positivity rate was evaluated among the groups divided by the Furman nuclear grade, a highly malignant grade of clear cell RCC showed high Kank1 positivity (p < 0.05), while the others did not (

Meaning of Kank1 expression and clinical outcome
Many RCC cells showed inactivation of the Kank1 gene as shown here. This inactivation presumably occurs at the early stage of carcinogenesis in normal renal tubular cells. Because hemizygous methylation of Kank1 was observed in many cancer cells (Sarkar et al., 2002), inactivation of Kank1 could be caused in both alleles by an epigenetic modification such as methylation, rather than by mutations.
Concerning the genetic abnormality of RCC, mutations in the VHL gene are most prevalent especially in clear cell RCC (Arai and Kanai, 2011). While VHL mutations can be found quite often in sporadic clear cell RCC, they are not significant in other RCC histological subtypes or benign oncocytoma. VHL mutations affect the activation of hypoxia-inducible factors, and investigation of this pathway will contribute to a new molecular targeting therapy for RCC (Suwaki et al., 2011). The difference in VHL mutations among the RCC histological subtypes suggests a difference in carcinogenesis for each histological subtype, though the origin of the cancer is always a renal tubular cell.
Given that the alteration of Kank1 expression occurred at the early stage of carcinogenesis, our findings that Kank1 expression differed among the histological subtypes of RCC might reflect a difference in cancer development (Kim et al., 2005). In clear cell RCC, the loss of Kank1 expression occurred at a high rate in the lower grade tumors, and the expression was reoccurred as the malignant grade increased. Although the reason for this is not clear, it is presumed that epigenetic modifications such as methylation might have been removed when the malignant grade increased, and consequently, the expression reoccurred (Kisseljova and Kisseljov, 2005). There was no difference in Kank1 expression between the samples obtained from the groups of patients who survived or not ( Table 2). This may reflect the fact that histological grade does not necessarily contribute to clinical outcome, but clinical stage (i.e. the presence of metastasis) is more crucial to obtaining a good prognosis (RCC patients diagnosed at the early stage have more than a 90% five year survival rate) (Lane and Kattan, 2008). The discordance of T stage (tumor size) and the malignant grade on Kank1 expression could also be supposed for the same reason. A similar result was found for the expression of CDKN2A encoding a growth suppressor protein, which is located at 9p21 and close to Kank1 (9p24) (unpublished data). Although the loss of Kank1 expression resulted in increased proliferation and poor differentiation in in vitro study (Sarkar et al., 2002), our results about the in vivo expression of Kank1 in clinical cases proved that reduced expression does not necessarily reflect a high grade malignancy or poor clinical outcome. These contradictory experimental and clinical results are very interesting, because they suggest that malignant transformation of a normal renal tubular cell has many genetic alterations and clinical outcome is contributed to by many factors in RCC.

Future diagnostics for RCC
The lack of clinical impact of the current diagnostic markers for RCC apparently requires progress in methodology, biology and pathology (Stewart et al., 2011). The progress in methodology needs the quality of the methods to satisfy the specificity, stability and biological relevance of the markers for diagnosis. For this, sufficient numbers, tens to thousands, of markers would be needed and such markers could be obtained only through cellular signaling analyses. There are quite a number of potential protein and genetic markers for diagnosis and therapeutic targets of RCC based on the information of signal transduction (see Section 2), and more information would be added in the future. While sampling is easier for DNA and RNAbased assays, protein assays such as immunohistochemistry and more advanced massspectrometry techniques have problems of contamination and degradation/modification at sampling and processing. In immunohistochemisty, protein cross-linking at the preparation steps disturbs antibody binding. Sampling of homogenously expressed proteins is crucial for the stability of assays, but would not be possible for most sampling cases as the tissue itself is not homogenous. However, diagnosis even for such cases could be possible with markers sufficiently distinguishing heterogenously expressed proteins in different parts of the diseased tissue. In all cases, a statistical significance analysis should be included as a standard evaluation step for quality control of multi-marker systems such as DNA microarrays (Shi et al., 2010).
Biologically relevant markers will be made available in the future based on the analysis of signal transduction, because, as shown in Fig. 1 (Section 2), there are a number of markers available even within a single signaling pathway and there are sufficient numbers of different pathways affected by the disease, which will contribute to the stability of assays. As discussed, the VHL and mTOR pathways have drawn much attentions to prognosis/diagnosis and therapeutic targets for RCC, but there are more pathways such as the Myc and FLCN pathways and pathways related to VEGF, PDGF and TGFα, and some are specific to subtypes of RCC Meanwhile, pathologically relevant markers will also be made available in the future, although the situation is different from other technologies due to the technical limit in the number of markers to examine simultaneously.

A new fluorescence-based immunohistochemical technique
One obstacle to improving immunohistochemistry is the availability of markers. Immunostaining is a relatively simple technique and thus can be used in unequipped laboratories and hospitals, because the preparation, storage and handling of samples are relatively simple. However, ordinary immunostaining is based on single-dye (or singlemarker) colorimetric techniques such as the alkaline phosphatase-based method. This is because of a lack of multi-dye (or multi-marker) colorimetric techniques due to expensive devices and, especially, inavailability of stable fluorescent dyes. Fluorescent dyes have been used in many technologies although this has not happened yet in immunohistochemistry because of the lack of their sufficient stability. Stable fluorescent dyes are thus needed for progress in immunohistochemistry.
We reported applications of a new fluorescent dye, Fluolid, for DNA microarray assays and immunohistochemistry . Fluolid dyes, including Fluolid-Orange, show stability against heat and excess light compared with other dyes (Fig. 6) and thus can be stored for more than a year without losing fluorescence (data not shown). So, multi-color immunohistochemistry with stable fluorescent dyes will change the pathological diagnostics in several ways: long-term storage of stained sections, simultaneous multi-marker detection and handling of fluorescently stained sections. Heat and light stable fluorescent dyes will enable us to store fluorescently stained sections at room temperature for a long time, which will be important for follow-up studies by microdissection of specific regions.

Future therapeutics
As discussed in Section 4.1, future diagnosis will be based on sufficient numbers of protein markers possibly obtained from signal transduction pathways, which will give a statistically significant decision even for cases where no decisive markers, such as disease-causing mutations or constitutive active proteins, are available. In the case of future therapeutics, multiple targets will also be considered to be an effective strategy. Signal transduction-based targeted therapeutics have already been developed for some diseases and drugs such as imatinib or Gleevec/Glivec, a small molecule inhibitor against activated tyrosine kinase activity by the Bcr-Abl fusion gene used for the treatment of chronic myelogenous leukemia, are available (Radford, 2002). Other monoclonal antibody-based drugs such as trastuzumab or Herceptin, which blocks a growth factor receptor HER2/neu (c-erbB-2) to treat breast cancer, and panitumumab or Vectibix, which blocks HER1 to treat colorectal cancer, have been developed based on signal-transduction. Although these drugs are effective, continuous use will sometimes generate drug-resistant cancer (Schenone et al., 2011). So, treatment with multiple targeting drugs will be important in future therapeutics and the same is true for the matched diagnostics about multiple targets.