Figures
Abstract
Growing evidence indicates that Rab GTPases, key regulators of intracellular transport in eukaryotic cells, play an important role in cancer. We analysed the deregulation at the transcriptional level of the genes encoding Rab proteins and Rab-interacting proteins in bladder cancer pathogenesis, distinguishing between the two main progression pathways so far identified in bladder cancer: the Ta pathway characterized by a high frequency of FGFR3 mutation and the carcinoma in situ pathway where no or infrequent FGFR3 mutations have been identified. A systematic literature search identified 61 genes encoding Rab proteins and 223 genes encoding Rab-interacting proteins. Transcriptomic data were obtained for normal urothelium samples and for two independent bladder cancer data sets corresponding to 152 and 75 tumors. Gene deregulation was analysed with the SAM (significant analysis of microarray) test or the binomial test. Overall, 30 genes were down-regulated, and 13 were up-regulated in the tumor samples. Five of these deregulated genes (LEPRE1, MICAL2, RAB23, STXBP1, SYTL1) were specifically deregulated in FGFR3-non-mutated muscle-invasive tumors. No gene encoding a Rab or Rab-interacting protein was found to be specifically deregulated in FGFR3-mutated tumors. Cluster analysis showed that the RAB27 gene cluster (comprising the genes encoding RAB27 and its interacting partners) was deregulated and that this deregulation was associated with both pathways of bladder cancer pathogenesis. Finally, we found that the expression of KIF20A and ZWINT was associated with that of proliferation markers and that the expression of MLPH, MYO5B, RAB11A, RAB11FIP1, RAB20 and SYTL2 was associated with that of urothelial cell differentiation markers. This systematic analysis of Rab and Rab effector gene deregulation in bladder cancer, taking relevant tumor subgroups into account, provides insight into the possible roles of Rab proteins and their effectors in bladder cancer pathogenesis. This approach is applicable to other group of genes and types of cancer.
Citation: Ho JR, Chapeaublanc E, Kirkwood L, Nicolle R, Benhamou S, Lebret T, et al. (2012) Deregulation of Rab and Rab Effector Genes in Bladder Cancer. PLoS ONE 7(6): e39469. https://doi.org/10.1371/journal.pone.0039469
Editor: Wanjin Hong, Institute of Molecular and Cell Biology, Singapore
Received: January 27, 2012; Accepted: May 21, 2012; Published: June 19, 2012
Copyright: © 2012 Ho et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Funding: The work was supported by grants from the Centre national de la recherche scientifique (CNRS) and the Institut Curie. The Molecular Oncology team is supported by La Ligue Nationale Contre le Cancer (“Equipe labellisée”). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing interests: The authors have declared that no competing interests exist.
Introduction
Intracellular trafficking is an essential process in eukaryotic cells. It relies on vesicular or tubular transport carriers that shuttle between cell compartments facilitating the constant exchange of proteins and lipids. Many studies have highlighted its complexity and led to the identification of a large number of proteins involved in the different steps of intracellular transport, i.e. the formation of transport carriers from donor membranes, their movement along cytoskeletal tracks and their tethering/fusion with target membranes. Small GTPases of the Rab family have emerged as key regulators of these different steps. As with other GTPases, Rab proteins cycle between an inactive GDP (guanosine diphosphate)-bound form and an active GTP (guanosine triphosphate)-bound form. The active GTP-bound form of the Rab is membrane-bound whereas hydrolysis of the GTP to GDP results in its dissociation into the cytosol. These two cycles are controlled by a complex regulatory network of proteins that includes guanine nucleotide exchange factors (GEFs), GTPase activating proteins (GAPs) and guanine nucleotide dissociation inhibitors (GDI). In their active form Rab GTPases interact with a diverse range of effector proteins, such as molecular motors, lipid kinases, tethering factors and scaffolding proteins (see [1] for review).
Recent studies have found a role for a number of Rab proteins in human cancers. Several expression studies have suggested that they could play both an activating and an inhibiting role in tumor progression. RAB1A is overexpressed in tongue squamous cell carcinoma [2]. RAB3A is expressed in insulinoma, but not in normal pancreatic islet cells [3]. RAB11A and RAB20 expression is increased during skin carcinogenesis [4] and in exocrine pancreatic adenocarcinomas [5], respectively. By contrast, RAB37 is down-regulated in metastatic tumors of lung cancer [6]. Both RAB5A and RAB7 were shown to be up-regulated in autonomous thyroid adenomas, such an up-regulation being correlated with an accelerated thyroglobulin endocytosis and hormone production [7].
Several functional studies have confirmed the role of Rab proteins in cancer progression. RAB5A, overexpressed in hepatocellular carcinomas, seems to be determinant for liver cancer progression, as suggested by the finding that a dominant negative form of RAB5A attenuates EGF-mediated signalling and cell migration of a human hepatoma cell line [8]. Other results have shown that RAB23, amplified and overexpressed in diffuse-type gastric cancer, acts as an invasion mediator gene [9]. RAB25 plays a role in the development of both ovarian and breast cancers [10], [11]. However, an opposite role of RAB25 has also been documented, i.e. as a tumor suppressor gene for colon cancer [12]. In addition, some proteins involved in Rab cycle regulation have also been implicated in carcinogenesis. For example RIN1, coding for a RAB5 GEF, was shown to be a breast tumor suppressor gene [13], whereas TBC1D3B, coding for a RAB5 GAP, was shown to be an oncogene amplified in prostate cancer [14].
Urinary bladder cancer is the fourth most common cancer in both European and American men. In women, it is the ninth most common cancer in the USA and the 14th most common cancer in Europe [15], [16]. According to the stage, at first presentation, about 50% of bladder carcinoma are Ta tumors which are generally of low grade (Ta tumors are papillary tumors that do not invade beyond the basement membrane), 20% are T1 tumors (tumors which invade beyond the basement membrane but not the underlying muscularis propria) and 30% are muscle-invasive tumors (T2–4). Carcinoma in situ (Cis) consisting of flat, high-grade lesions not invading beyond the basement membrane are rarely found in isolation. Instead, Cis is predominantly encountered with other urothelial tumors. Clinical and molecular evidence suggest that bladder tumors arise and progress along two main pathways: the “Ta” pathway and the “carcinoma in situ” pathway. Ta tumors display a high recurrence rate (60%, [17]), but have a low probability (5–10%) of progressing to T1 tumors and then to muscle-invasive tumors (T2–4). By contrast, Cis often progress (in about 50% of cases), to T1 tumors and then to muscle-invasive tumors ([18] for review). The Ta pathway is characterized by a high frequency of activating mutations of the FGFR3 gene (encoding the tyrosine kinase fibroblast growth factor receptor 3). FGFR3 mutations are present in 70–75% of Ta tumors and absent from Cis. Their relatively low frequency in T1 (20%) and T2–4 tumors (10–15%) is consistent with the high rate of Cis progression and the low rate of Ta progression [19]–[24].
The goal of this work was to perform a systematic study to identify Rab proteins and Rab-interacting proteins whose expression is deregulated during the Ta and Cis pathways of bladder tumor pathogenesis at the transcriptomic level.
Results
For this study, we applied the strategy presented in Figure 1 (flow chart). Briefly: 1. an exhaustive list of genes coding for Rab and Rab-interacting was established from public databases, 2. deregulated genes (up- or down-regulated compared to normal urothelium) in each of the two pathways (the FGFR3-mutated tumor pathway and the FGFR3-non-mutated tumor pathway) were identified from two transcriptome datasets, 3. deregulated genes possibly because of tumor stroma or tumor – stroma interactions were identified and then omitted from further analysis (analysis of transcriptome data in bladder tumor cell lines compared to normal cells in culture (NHU)), 4. for each deregulated gene, a specific association either with the FGFR3-mutated tumor group or the non-mutated tumor group was investigated as well as an association with a differentiation or proliferation phenotype. Additionally, we investigated for each Rab cluster (consisting of a given Rab protein and its interacting partners) whether it could be associated with bladder cancer pathogenesis.
The first step is the identification through public data bases and expert knowledge of the genes of interest to study, here the Rabs and their effectors. The second step consists of selecting subgroups of tumors and analysing the expression of the different genes selected in the first step in these subgroups compared to the normal urothelium. The subgrouping here has been done taking into account the FGFR3 mutation status, the stage and the grade, separating the tumors into two pathways. A comparison of the expression observed in bladder cancer cell lines and in cultured normal human urothelial cells allowed discarding of genes for which the expression could be possibly due to the presence of stroma (in comparison to normal cells, upregulation in bladder tumors but not in bladder tumor cell lines). Different types of analysis were then performed on the selected deregulated genes: 1) a comparison of the expression in FGFR3-mutated tumors and FGFR3-non-mutated tumors allowed the identification of genes specifically deregulated in one of the two pathways of bladder cancer pathogenesis; 2) by grouping genes into cluster of genes (here the Rab clusters), we identified clusters with deregulated expression; 3) by analysing the possible correlation between the expression of the deregulated genes and the expression of proliferation or differentiation marker genes, we identified the deregulated genes associated with proliferation or differentiation.
The results obtained at each step of the analyses are summarized in Figure 2.
The results of each analysis step are shown in the flow chart presented in Figure 1.
Definition of the list of genes to be analysed
The first part of this work consisted of establishing a list of genes coding for Rab proteins and Rab-interacting proteins, including activating GEF and inactivating GAP proteins (Figure 3). 61 human RAB genes were found. The list of Rab-interacting proteins was established by a literature search to gather papers that have reported the identification and/or the characterization of proteins that directly interact with Rab GTPases, using different approaches (two-hybrid assays, GST-pull down, coimmunoprecipitation, etc.). This list comprised 217 proteins: 23 GEFs, 20 GAPs and 174 effector proteins. We added to this list two genes coding for the two GDI (GDP Dissociation Inhibitor) proteins (GDI1 and 2) that are common to all Rab GTPases. We also included four genes coding for proteins involved in post-translational modification and membrane association of all newly-synthesized Rabs: CHM and CHML coding for Rab escort proteins (REP) and the two isoforms of the Rab geranylgeranyl transferase (RABGGT A and B). This made a total of 284 Rabs and Rab-interacting proteins (Figure 2).
Rab GTPases cycle between an active GTP-bound form and an inactive GDP-bound form. Rab activation is mediated by a guanine exchange factor (GEF). The hydrolysis of bound GTP is catalyzed by a GTPase activating protein (GAP) resulting in the inactivation of the Rab protein. In its active form, the Rab is associated with membranes and can interact with a variety of effector proteins. In its inactive form the protein is cytosolic and is in complex with a GDP dissociation inhibitor (GDI) protein.
The list of genes analysed in this study and the list of papers in which they were originally described are shown in Table 1 and Table S1. Figure 4 illustrates the cluster of proteins shown to interact with RAB27A/B (as an example) and Figure S1 illustrates all Rab clusters analysed; Rab proteins are in yellow, GEFs in green, GAPs in light red and effector proteins in blue.
Example of the Rab27 cluster. The Rab27 cluster is comprised of the two RAB27 isoforms (RAB27A and RAB27B), the GEF MADD, the GAP TBC1D10A and 12 effector proteins. The other Rab and Rab-interacting proteins are shown in supplementary Figure S1.
Model of bladder cancer pathogenesis used in this study
Two main progression pathways have been so far identified in bladder cancer, the Ta pathway characterized by a high frequency of FGFR3 mutation and the carcinoma in situ (Cis) pathway where no or infrequent FGFR3 mutations have been identified. In this study we therefore considered two pathways: the FGFR3-mutated tumor pathway and the FGFR3-non-mutated tumor pathway (Figure 5). The FGFR3-mutated tumor pathway comprised the TaG1 and TaG2 FGFR3-mutated tumors, the T1 FGFR3-mutated tumors and the muscle-invasive FGFR3-mutated tumors (T2–4 tumors). We analysed two sets of bladder tumors (n = 152 in the first data set and n = 75 in the second data set). The number of FGFR3-mutated TaG3 was too small to identify them as a separate group (2 tumors in the first data set and 1 tumor in the second data set), but they could not either be included in the TaG1/TaG2 group due to their different clinical and molecular characteristics so they were not considered in the analysis. The FGFR3-non-mutated tumor pathway comprised the TaG3 FGFR3-non-mutated tumors, the T1 FGFR3-non-mutated tumors and the muscle-invasive FGFR3-non-mutated tumors. Transcriptomic data from Cis tumors were not available in our series. We used FGFR3-non-mutated TaG3 tumors instead of Cis. These tumors share several properties with Cis as they progress to an invasive stage [25] with a high probability (45%, [26]). In addition, both lesions are microscopically very similar at individual cell examination. TaG1/TaG2 tumors not mutated for FGFR3 (9 samples in the first set of data, 2 samples in the second set of data) were not considered in the analysis as they do not fit in the Cis pathway: indeed these tumors are of low grade and they rarely progress to T1 and then to muscle-invasive tumors [26].
Identification of up- or down- regulated genes
In order to identify genes up- or down-regulated during bladder cancer progression, the two pathways of the “FGFR3 model” were analyzed separately. We used the three groups, TaG3, T1, and T2–4, previously defined for the FGFR3-non-mutated pathway and the three groups, TaG1G2, T1, and T2–4 for the FGFR-mutated tumor pathway (Figure 5). For each tumor group, the expression of the Rab and Rab-interacting protein genes listed in Table S1 was compared to their expression in the normal urothelium (obtained without stroma) group using the statistical SAM test. The results were filtered with the following thresholds: Fold Change (FC) >1.5 for up-regulation (and <0.667 ( = 1/1.5) for down-regulation) and qValue <5%. We first analysed a collection of samples containing 4 normal urothelium and 141 tumor samples (see Material and Methods) using the Affymetrix HG U133 Plus 2.0 DNA microarrays. 269 of the 284 genes listed in Table S1 were present in these microarrays. For the FGFR3-non-mutated tumor pathway, 19 genes were found to be down-regulated and 12 genes up-regulated (Table 2); for the FGFR-mutated tumor pathway, 21 genes were found to be down-regulated and 4 genes up-regulated (Table 3) (summary in Figure 2).
The “FGFR3 model” of bladder cancer progression distinguishes a FGFR3-non-mutated tumor pathway and a FGFR3-mutated tumor pathway. Cis: Carcinoma in situ. Stages were defined by the 1997 TNM classification and grades by the 1973 World Health Organization classification (see Material and Methods for reference).
We then analysed a second set of tumors to verify the results obtained. This set, collected independently from the first set, was analysed with the Affymetrix HG U95A/U95Av2 DNA microarrays and contained 5 normal urothelium and 72 tumor samples (see Material and Methods). Only 165 of the 284 genes listed in Table S1 (58 %) were present in these microarrays. Among the 31 genes found to be up- or down-regulated in the FGFR3-non-mutated tumor pathway (Table 2), 17 were present on the Affymetrix HG U95A/U95Av2 DNA microarrays: CASP1, CAV1, CD2AP, EEA1, ICA1, ITGA5, MICAL2, PIGR, RAB11A, RAB14, RAB31, RAB4A, RABAC1, STXBP1, TBC1D30, TBC1D4 and ZWINT. Among the 25 genes found to be up- or down-regulated in the FGFR3-mutated tumor pathway (Table 3), 14 were present on the Affymetrix U95A/U95Av2 DNA microarrays: CAV1, EEA1, ICA1, ITGA5, PIGR, RAB11FIP2, RAB14, RAB27A, RAB27B, RAB9A, RABGAP1L, SDC1, TBC1D30 and UNC13B. For these genes, 56 % of the results obtained with the first set of data were validated with the second set of data with at least one threshold passed: FC>1.5 (or <0.667) and/or qValue <5% (Table S2). The other results were not significant. Of note, no discordant result between the two sets of data was found.
Expression analyses of upregulated genes in vitro
A tumor is composed of both tumor and stromal cells. As the transcriptomic analysis is performed with this mix of cells, up-regulation of a given gene could then be due to the presence of stroma (genes up-regulated in the stroma or in the tumor cells due to stromal cell – tumor cell interaction).
To address this issue, we analyzed in normal human urothelial (NHU) cells grown in culture under regenerating and differentiating conditions [27], [28] and in 7 established bladder cancer cell lines (without stromal cells) the expression of the 13 genes found to be up-regulated in the FGFR3-non-mutated and mutated tumor pathways: CAV1, ITGA5, KIF20A, LEPRE1, MICAL1, MICAL2, RAB23, RAB31, RABAC1, SDC1, STXBP1, TMEM22 and ZWINT (Table 2 and Table 3). For most of the genes, we found at least one cancer cell line in which its expression was at least two-fold higher than in cultured normal urothelial cells. This was not the case for MICAL1, RABAC1 and SDC1 (Figure 6). The overexpression of MICAL1 and RABAC1 is likely due to the presence of stromal cells in the tumor as MICAL1 is highly expressed in different types of hematopoietic cells (dendritic cells, mast cells and NK cells) and RABAC1 in mast cells (data not shown). The overexpression of SDC1 could be due to stromal-epithelial interaction. MICAL1, RABAC1 and SDC1 were excluded for further analysis (see summary in Figure 2).
The expression of the 13 up-regulated genes (Table 2 and 3) (CAV1, ITGA5, KIF20A, LEPRE1, MICAL1, MICAL2, RAB23, RAB31, RABAC1, SDC1, STXBP1, TMEM22 and ZWINT) was measured by Affymetrix array in 7 bladder cancer cell lines (KK47, MGHU3, RT112, RT4, SCaBER, SD48 and T24) and normal human urothelial (NHU) cells grown in culture. The threshold for genes to be considered as up-regulated in tumor cell lines (2 fold the expression measured in NHU cells) is represented by a black line.
Genes specifically up- or down-regulated in each pathway
Among the genes found to be deregulated during bladder cancer progression (Table 2 and Table 3), 13 were common to both pathways, 10 down-regulated: EEA1, ICA1, MLPH, MYO5B, MYO5C, PIGR, RAB14, RPH3AL, SYTL2, TBC1D30 and 3 up-regulated: CAV1, ITGA5, MICAL1. In order to search for genes specifically deregulated in each pathway, we proceeded in two steps. 1. We selected the genes found significantly up- or down- regulated only within the FGFR3-non-mutated tumor pathway or within the FGFR3-mutated tumor pathway. 9 down-regulated and 8 up-regulated genes were selected for the FGFR3-non-mutated tumor pathway and 11 down-regulated genes were selected for the FGFR3-mutated tumor pathway (Table S3, left column). 2. We used the statistical SAM test to compare the expression of the selected genes in the tumor group of the same stage but of opposite FGFR3 mutation status (Table S3, right column). SYTL1 was significantly down-regulated in the T2–4 (FGFR3-non-mutated) tumor group whereas LEPRE1, MICAL2, RAB23 and STXBP1 were significantly up-regulated in the same tumor group (Table 4 and Figure 7; see summary in Figure 2). No gene was found to be specifically deregulated for the FGFR3-mutated tumors.
Expression of SYTL1, LEPRE1, MICAL2, RAB23 and STXBP1 measured by Affymetrix array in normal samples (n = 4) and FGFR3-mutated tumor samples (TaG1G2, n = 28; T1, n = 13; T2-4, n = 9), and FGFR3-non-mutated samples (TaG3, n = 3; T1, n = 25; and T2-4, n = 63). Are represented the 10th percentile (below bar), the 25th percentile (box bottom), the median (bar in the box), the 75th percentile (box top) and the 90th percentile (upper bar). The points represent the outlier samples. SYTL1 is down-regulated in FGFR3-non-mutated tumors, whereas LEPRE1, MICAL2, RAB23 and STXBP1 are up-regulated.
Analysis by clusters of genes
In the previous part of this study, we used the statistical SAM test to analyse separately the expression of each gene during bladder cancer progression. To investigate whether a Rab cluster (consisting of a given Rab protein and its interacting partners, see Figure 4 and Figure S1) could be specifically associated with bladder cancer progression, we used the statistical binomial test to determine whether the percentage of genes deregulated within a given Rab cluster was significantly greater than the percentage obtained by analyzing all the Rab and Rab-interacting protein genes. This test was applied for each Rab cluster within each tumor group and the results were filtered with the threshold pValue <1%. Interestingly, the Rab27 cluster passed this threshold for the T2–4 tumor groups of both the FGFR3-non-mutated and mutated tumor pathways (Table 5 and Table S4; see summary in Figure 2). In addition to RAB27A and RAB27B, the genes down-regulated in this cluster comprise GCC2, MLPH, RPH3AL, SYTL1 and SYTL2.
Genes associated with proliferation
In order to evaluate a possible association of the deregulated genes listed in Table 2 and Table 3 with cell proliferation, we calculated a Pearson correlation between their expression and that of the proliferation marker gene: MKI67 [29]. For this analysis, among the 6 groups constituted for this study, we worked with the two homogenous tumor groups with the higher number of samples: the Ta G1/G2 (FGFR3-mutated) tumor group (28 samples) and the stage T2–4 (FGFR3-non-mutated) tumor group (63 samples). We first noticed, as expected, that the expression of MKI67 was significantly higher in the T2–4 (FGFR3 non-mutated) group than in the Ta G1/G2 (FGFR3-mutated) group (about 3 fold; Student test, p = 8.8E-10). We chose to filter the Pearson correlation values with the threshold: |r| >0.479 (pVal <1%) for the Ta G1/G2 (FGFR3-mutated) tumor group and |r| >0.323 (pVal <1%) for the T2–4 (FGFR3-non-mutated) tumor group. The expression of a gene was considered to be correlated with the expression of MKI67 if the pValue <1% with both tumor groups. The expression of KIF20A and ZWINT were correlated, whereas the expression of MYO5C was inversely correlated, with the expression of MKI67 (Figure 8 and Table S5; see summary in Figure 2).
The Pearson correlation coefficient (r) between the expression of the deregulated genes and the expression of proliferation marker gene, MKI67, was calculated for FGFR3-mutated tumors in the TaG1G2 group (n = 28) and for FGFR3-non-mutated tumors in the T2–4 group (n = 63). The expression of the deregulated genes as a function of MKI67 expression is shown in TaG1G2 FGFR3-mutated tumors (upper figures) and in T2–4 non-mutated tumors (lower figures). Only the plots for the correlated genes are presented (p<1%, which corresponds to a correlation coefficient, |r| above 0.479 for the FGFR3-mutated tumor group and above 0.323 for the FGFR3-non-mutated tumor group).
Genes associated with differentiation
The same analysis was performed to evaluate the association of deregulated genes with the differentiation process of urothelial cells. The uroplakin genes UPK1A, UPK1B, UPK2, UPK3A and UPK3B, encoding for urothelium-specific markers [30]–[33] and the two genes GRHL3 [34] and FOXA1 [28], encoding for transcription factors, were used as markers of urothelial differentiation. We used the same two tumor groups as above. The results of the Pearson correlation were filtered with the same thresholds as above and we considered that the expression of a gene is correlated with the expression of another gene if the pValue was less than 1% for both tumor groups. The results are shown in Table 6 and Table S6. Seven genes were found to be correlated with at least one urothelial differentiation marker: ANKRD27, MLPH, MYO5B, RAB11A, RAB11FIP1, RAB20 and SYTL2. Two genes were found to be inversely correlated with at least one urothelial differentiation marker: CASP1 and RAB27A (see summary in Figure 2).
Gene expression in NHU cells in culture
Normal human urothelial cells (NHU cells) can grow in culture for a finite number of passages until they enter into senescence. Before senescence, differentiation can be induced by growing them in specific media. Here NHU cells, after passage, were either grown in non-differentiating conditions (control) or in differentiating conditions (in the presence of an EGFR (epidermal growth factor receptor) inhibitor and an activator of PPARγααμμ (peroxisome proliferator-activated receptor gamma)) [28]. In control conditions, cells reach confluency and become contact-inhibited, whereas in differentiating conditions they also stop growing but begin to express markers of urothelial terminal differentiation, such as the uroplakins (UPKs).
We used Affymetrix data of NHU cells in different conditions to look at the expression of the genes previously found to be associated in tumors with proliferation or differentiation markers. The NHU expression data were obtained at different times after passage: 6 hours, 1 day, 3 days and 6 days, with or without differentiating medium. As shown in Figure 9, the expression of KIF20A and ZWINT decreased in cultures treated with the differentiating medium, as well as in control cultures upon reaching confluence, indicating a link of these two genes with proliferation. In this model, MYO5C was not linked with proliferation, as observed in tumor expression data, but with differentiation (increased expression in the differentiating medium but not in the control medium). Among the 7 genes positively correlated with differentiation markers in the tumors, 6 were indeed induced by the differentiating medium (MLPH, MYO5B, RAB11A, RAB11FIP1, RAB20 and SYTL2) but not ANKDR27. Neither of the two genes inversely correlated with differentiation (CASP1 and RAB27A) were down-regulated upon differentiation in NHU cells.
The expression of the genes found to be associated with proliferation or differentiation in tumors were measured in normal urothelial cells (NHU) in culture at four different times after passage: 6 hours, 1 day, 3 days and 6 days in differentiating conditions (in the presence of an inhibitor of EGFR and an activator of PPARγααμμ in the NHU medium) or in non-differentiating conditions (NHU medium only).
Discussion
Growing evidence indicates that Rab proteins and their effectors are involved in cancer progression, both as inhibitory and promoting factors. However, to our knowledge no systematic study has addressed their deregulation during cancer progression. In this study, we have identified several genes coding for Rabs and Rab-interacting proteins whose expression is deregulated during bladder cancer pathogenesis.
Pertinence of the two pathways model
Based on clinical and molecular evidence, two progression pathways have been described in bladder cancer, the Ta pathway presenting high frequency of FGFR3 mutation and the Cis pathway with no or low rate of FGFR3 mutation. In this study, a simple method, i.e. the presence or absence of FGFR3 mutation, was used to separate the tumors within the two pathways. This led to discard the TaG3 FGFR3-mutated tumors (4% of FGFR3-mutated cases in the first data set) and the TaG1, G2 FGFR3-non-mutated tumors (9% of FGFR3-non-mutated cases in the first data set), as in these two subgroups the FGFR3 mutation was likely not the correct parameter for classifying a tumor in one of the two pathways (see paragraph on “Model of bladder cancer pathogenesis used in this study” in the Results section). Beside FGFR3 mutation, two other molecular markers can be used to distinguish the two pathways of bladder tumor progression: the Cis signature [35] and the MRES (multiple regional epigenetic silencing) phenotype [36]. The classifications obtained by using the three different methods (Cis signature, MRES phenotype and FGFR3 mutations), although presenting an important overlap, display some differences ([36] and data not shown). A better characterization of the FGFR3 signaling pathway, as well as a better knowledge of the mechanism involved in the MRES phenotype and a detailed analysis of the precursor lesions will lead to a better definition of the two pathways of bladder tumor progression and possibly to the definition of additional pathways. It will be important to apply again the analyses described in this study once the different pathways will have been better characterized. Because of more homogeneous subgroups, this could lead to the identification of additional genes involved in particular subgroups of tumors.
Globally, more genes were found to be down-regulated (30) than up-regulated (13) (p = 2.16.10−5, binomial test). Interestingly, more genes were found to be up-regulated in the FGFR3-non-mutated tumor pathway (12 genes) as compared to the FGFR3-mutated tumor pathway (4 genes). This suggests the existence of two different mechanisms for tumor progression along the two pathways. LEPRE1, MICAL2, RAB23 and STXBP1 were specifically up-regulated in the muscle-invasive FGFR3-non-mutated tumors as compared to both normal urothelium and muscle-invasive FGFR3-mutated tumors, suggesting that these genes play an active role in tumor progression along the FGFR3-non-mutated tumor pathway. By contrast, SYTL1 was specifically down-regulated in the muscle-invasive FGFR3-non mutated tumors. Thus SYTL1 could play a role as an inhibitor of tumor progression along this pathway. Alternatively SYTL1 could be lost because it is not essential in these tumors, but is maintained in the FGFR3-mutated tumors where it is essential. Overall, these results suggest that even at the muscle-invasive stage, the tumors of the two pathways should be distinguishable.
Concerning MICAL2, which codes for a RAB1 effector [37], this gene has been previously shown to be up-regulated in prostate cancer and to be necessary for prostate cancer cell viability [38]. LEPRE1 encodes for an enzyme member of the collagen prolyl hydroxylase family. These enzymes are localized to the ER, and their activity is required for proper collagen synthesis and assembly. We indeed found a correlation between the expression of LEPRE1 and the expression of COL1A1 and COL1A2 (data not shown). The last gene, STXBP1, encodes for a RAB3 effector and regulates exocytosis in neurons and neuroendocrine cells through modulation of vesicle docking and membrane fusion [39]. This suggests a possible neuroendocrine differentiation in the FGFR3-non-mutated tumor pathway that would not be present in the FGFR3-mutated tumor pathway.
Genes deregulated in the TaG1G2 (FGFR3-mutated) group
Most of the deregulated genes in the FGFR3-mutated tumor pathway were found in this group. ANKRD27, RAB20 and RAB11FIP1 were correlated with UPK1A, UPK2, UPK3A, FOXA1 and GRHL3 (Table 6), suggesting that they are associated with urothelial cell differentiation. Even if the tumors of low grade (Ta)G1G2 still appeared well differentiated, we nevertheless observed a down-regulation of UPK1A, UPK3A and GRHL3 (not UPK2 and FOXA1, data not shown). ANKRD27 encodes for a RAB21 GEF [40], a Rab involved in the early endocytic pathway [41]. ANKRD27 also interacts with VAMP7 [42], RAB32 and RAB38 [43]. Less is known about RAB20 function, a Golgi-associated Rab overexpressed in exocrine pancreatic carcinoma [5]. RAB11FIP1 (also named RCP) is a RAB11A effector [44] involved in recycling processes from endosomes to the plasma membrane. The finding that RAB11FIP1 is likely to be involved in urothelial cell differentiation is consistent with the role of RAB11A in urothelial umbrella cells [45]. Nevertheless, RAB11FIP1 also interacts with RAB25 [44] that can act as a tumor suppressor or tumor promoter depending on the RCP expression status [46]. Of note, RAB25 was not found to be deregulated in this study. Finally, the deregulation of RAB27A/B suggests an alteration of the transport of lysosome-related organelles [47] in the tumors of the FGFR- mutated tumor pathway (see also below).
Genes deregulated in the T2–4 (FGFR3-non-mutated) group
Most of the deregulated genes in the FGFR3-non-mutated tumor pathway were found in this group. Among them, RAB11A expression was correlated with the expression of UPK1A and UPK2, two urothelium marker genes also down-regulated in this tumor group (data not shown). The down-regulation of RAB11A may thus be secondary to the loss of differentiation. This result is consistent with the finding that the exocytosis of the discoidal/fusiform vesicles in the most differentiated urothelial cells, the umbrella cells, is RAB11A-dependent [45].
The expression of ZWINT was found to be correlated with the expression of MKI67, a proliferation marker gene. Therefore the observed up-regulation of ZWINT could be due to a higher proliferation index for the muscle-invasive tumors. ZWINT has indeed been shown to play a role in the mitotic checkpoint [48], [49], an essential step for cell division. Moreover, a previous study has shown that a high expression of ZWINT is associated with a poor prognosis in pulmonary carcinomas [50], thus the up-regulation of ZWINT could also be associated with a poor prognosis for bladder cancer. The fact that ZWINT is up-regulated in the muscle-invasive tumors but not in the superficial tumors is concordant with this hypothesis.
In addition to RAB11A and ZWINT, the other deregulated genes are also of interest. Their deregulation could be associated with events occurring before the establishment of bladder cancer or could be a consequence of genomic instability resulting from the malignant process. Among them RAB23, found to be up-regulated in this study, has been shown to be over-expressed and/or activated in hepatocellular carcinoma [51]. It was also found to be up-regulated in mucosa of atrophic gastritis and intestinal metaplasia [52] and it is required for the invasion of diffuse-type gastric cancer [9]. One RAB23 function is to antagonize sonic hedgehog (Shh)-mediated signalling during mouse development [53], a pathway implicated in urothelial cell proliferation in mice [54]. Of note, RAB23 is highly expressed in muscle therefore, the up-regulation observed could be due to this stromal expression. Nevertheless it is also overexpressed in bladder cancer cell lines compared with normal urothelial cells grown in culture.
The down-regulation of RAB4A, encoding for a Rab that regulates fast recycling from early endosomes, and that of CD2AP, coding for a RAB4A effector involved in actin cytoskeleton regulation [55], points towards a general deregulation of the recycling pathway in T2–4 (FGFR3-non-mutated) tumors. The alteration of the fast recycling process was not found in the TaG1G2 (FGFR3-mutated) tumors. In these tumors RAB4A expression was not altered. The only RAB4A effector encoding gene whose expression was altered is RAB11FIP1, which is likely to be involved in urothelial cell differentiation (see above).
The possible role of the deregulation of intracellular trafficking of surface receptors in bladder cancer progression is supported by the finding that RAB31 is up-regulated. RAB31 regulates EGFR trafficking in A431 cells [56]. Interestingly, EGFR is overexpressed in urothelial tumors [57]. Of note, high expression of RAB31 is associated with a worse outcome in patients with breast cancer [58].
Genes deregulated in both Ta G1G2 (FGFR3-mutated) and T2–4 (FGFR3-non-mutated) groups
Amongst this group of genes, ITGA5 (up-regulated) encodes for the alpha 5 subunit of the alpha-5/beta-1 integrin that associates with RAB25 to promote invasive migration of ovarian carcinoma cells [59]. ITGA5 is down-regulated in transformed plasma cells compared to normal plasma cells [60] and in highly invasive potential breast cancer cell lines [61]. Another study indicates, on the contrary, that ITGA5 might be essential for breast cancer metastatic capacity [62]. We found an increase of ITGA5 in tumors independently of the stage (stage Ta or stage T2–4) or of the pathway of bladder cancer progression. This suggests a role other than migration/invasion for ITGA5 since Ta tumors do not have a high invasive potential.
The Rab27 cluster
When the 284 genes analysed in this study were grouped in Rab-Rab effector clusters, we noticed that the Rab27 cluster was significantly deregulated in both pathways. In addition to RAB27A/B, five genes coding for RAB27 interacting partners (GCC2, MLPH, RPH3AL, SYTL1 and SYTL2) were also down-regulated. In both pathways the deregulation of the Rab27 cluster is associated with the muscle-invasive tumors: RAB27B, MLPH, RPH3AL and SYTL2 with the T2–4 (FGFR3-mutated) tumors and GCC2, MLPH, RPH3AL, SYTL1 and SYTL2 with the T2–4 (FGFR3-non-mutated) tumors. Interestingly, MLPH and SYTL2 were found to be correlated with differentiation markers. RAB27B, even if it did not correlate with the urothelial differentiation markers used in this study, has been previously shown to be associated with fusiform vesicles in urothelial cells [63]. Therefore the association of the Rab27 cluster, of which all the genes were down-regulated, with muscle-invasive tumors could be linked to the loss of differentiation of muscle-invasive tumors. It is of interest to note that a recent study suggests that RPH3AL may have a role as a tumor suppressor gene in a minority of colorectal cancer cases [64].
Proposed functions for some of the deregulated genes
Three genes were found to be linked with proliferation: KIF20A, MYO5C (inversely correlated) and ZWINT (discussed above). Finding KIF20A was not surprising, as this protein plays an essential role in cytokinesis [65], [66]. A previous study has also documented an up-regulation of KIF20A in bladder cancer [67]. This independent result aids in the validation of our approach. Finding MYO5C was more surprising since no previous study has linked this actin-based motor protein with proliferation. In addition, we could not confirm an association of MYO5C with proliferation using the NHU (normal human urothelial) cell culture model.
On the other hand, since bladder tumors tend to be less differentiated while they progress, we expected to find many down-regulated genes linked to differentiation. Seven genes: ANKRD27, MLPH, MYO5B, RAB11A, RAB11FIP1, RAB20 and SYTL2 among 30 down-regulated genes were indeed found to be positively correlated with differentiation markers. Therefore the majority of genes (23) found to be down-regulated in this study are not linked to differentiation. Interestingly, 6 of 9 genes linked to differentiation were found to be deregulated in the FGFR3-mutated group and not in the FGFR3-non-mutated group. In addition, this deregulation was already apparent in tumors of low stage and low grade (TaG1G2). Therefore, although TaG1G2 FGFR3-mutated tumors are well differentiated, they already present a loss of expression of several genes associated with differentiation. It remains to know whether these Rab genes are simply secondary markers of differentiation or if they play a role in bladder cancer pathogenesis. We expected to find more genes linked to differentiation being lost in the FGFR3-non-mutated tumor pathway as these tumors are usually of high grade compared to the tumors of the FGFR3-mutated tumor pathway. This was actually not the case. This unexpected result could be due to some heterogeneity in FGFR3-non-mutated tumors, as heterogeneity decreases the number of identified deregulated genes. A better characterization of bladder tumors should lead to the identification of new genes specifically deregulated in new subgroups of tumors.
In conclusion, the strategy applied in this study can be easily applied to any set of genes and any tumor subgroups. It requires biologists, specialists of a given pathway/process, to select the most appropriate set of genes and to interpret the results. In many tumors, the normal counterpart of the transformed cell is not available or not known. In this case, one can compare tumors of different stages and grades. The next step will be to address the cellular functions of the genes deregulated at the transcriptome level and their possible implication in cancer pathogenesis.
Materials and Methods
Establishment of the list of genes
We first used the GeneCards database and found 61 human RAB genes indexed. Then using the RAB gene name and the different aliases listed as key search words we found in 185 articles indexed in PubMed a total of 217 proteins whose interaction with one or several Rab proteins has been published up to April 2010. After adding GDI1, GDI2, CHM, CHML, RABGGTA and RABGGTB, we worked with a total of 284 genes. We adopted HUGO gene nomenclature, except for CP110 since its Entrez Gene name had not been indexed in the HUGO database.
Normal and tumor samples
Tumors were staged according to the TNM classification [68] and graded according to the criteria recommended by the World Health Organization (1973 classification) [69]. FGFR3 mutation was assessed by direct sequencing or by the SNaPshot method [70].
Two sets of independent tumors obtained during transurethral resection or radical cystectomy [71] and normal urothelium samples obtained from organ donors [72] were used. The tumors were provided by Drs A.-C. Baglin and Y. Denoux (Tumorothèque, Service d'Anatomie et Cytologie Pathologiques, CMC Foch, Suresnes, France), Dr M.-J. Terrier-Lacombe (Service de Pathologie, Institut Gustave Roussy, Villejuif, France), and Pr K. Leroy (Plateforme de Ressources Biologiques, Pôle de Recherche Clinique, APHP, Hôpital Henri Mondor, Créteil, France).
The first set is composed of 4 isolated samples of normal urothelium and 152 tumor samples: 28 TaG1G2 (FGFR3-mutated), 2 TaG3 (FGFR3-mutated), 13 T1 (FGFR3-mutated), 9 T2–4 (FGFR3-mutated), 9 TaG1G2 (FGFR3-non-mutated), 3 TaG3 (FGFR3-non-mutated), 25 T1 (FGFR3-non-mutated) and 63 T2–4 (FGFR3-non-mutated). These samples were analyzed with Affymetrix HG U133 Plus 2.0 DNA microarrays. 269 of the 284 genes listed were found in these microarrays when using BrainArray Annotation ENTREZGENE [73]. F8A1, CCZ1, GMCL1, RAB13, RAB19, RAB41, RAB43, RAB44, RAB6C, RAB7B, RAB9B, REP15, RPH3A, TBC1D3B and YWHAQ were absent.
The second set is composed of 5 isolated samples of normal urothelium and 75 tumor samples: 18 TaG1G2 (FGFR3-mutated), 1 TaG3 (FGFR3-mutated), 5 T1 (FGFR3-mutated), 7 T2–4 (FGFR3-mutated), 3 TaG3 (FGFR3-non-mutated), 2 TaG1G2 (FGFR3-non-mutated), 7 T1 (FGFR3-non-mutated) and 32 T2–4 (FGFR3-non-mutated). These samples were analyzed with Affymetrix HG U95A/U95Av2 DNA microarrays. 165 of the 284 genes listed were present in custom chip definition of BrainArray ENTREZGENE.
We worked with 7 bladder cancer cell lines: KK47, MGHU3, RT112, RT4, SCaBER, SD48 and T24, and normal human urothelial (NHU) cells grown in culture following previous protocols [27], [28]. These cell lines were analyzed with Affymetrix Human exon 1.0stV2 microarrays. 281 of the 284 genes listed were present in remapped chip definition: CCZ1, F8A1 and RAB7B were absent. RT4, RT112 and T24 were obtained from DSMZ (Braunschweig, Germany), SCaBER from ATCC (Rockville, MD), KK47 and SD48 from the laboratory of D. Chopin (Hôpital Henri Mondor, Créteil, France) and MGHU3 from the laboratory of Y. Fradet (University of Laval, Québec, Canada). To verify the identity of the various cell lines used, we analyzed the genomic alterations with comparative genomic hybridization (CGH) arrays and FGFR3, TP53, HRAS, and KRAS mutations with the SNaPshot technique (for FGFR3) or classical sequencing for the other genes. CGH array profiles were found to be similar to published CGH profiles and/or consistent with previously reported genomic alterations, and mutations were found to be identical to the reported mutations (data not shown).
The NHU cell cultures used for the last part of the results were established and maintained in culture as described [27] and were induced to differentiate by treatment with an EGFR (epidermal growth factor receptor) inhibitor and an activator of PPARγ (peroxisome proliferator-activated receptor gamma) [28], [74]. Differentiated and untreated cultures were analyzed with Affymetrix HG U133 Plus 2.0 DNA microarrays.
RNA was prepared from the normal and tumor samples and cancer cell lines with the cesium chloride procedure [75].
All patients provided written informed consent and the study was approved by the ethics committees of the different hospitals (Comité de Protection des Personnes de l'hôpital Henri Mondor, Comité de Protection des Personnes de Boulogne – Ambroise Paré and Comité de Protection des Personnes de Bicêtre). All analyses were performed on the basis of anonymized patient data.
Affymetrix microarray data
We used the Human Genome U95A, U95Av2, U133Plus2.0 and Human Exon 1.0st V2 arrays (Affymetrix). Details of the methods for RNA amplification, cDNA probe labeling and hybridization steps can be obtained from the Affymetrix web site. All gene expression data were normalized and summarized using RMA (Robust Multi-array Averaging) algorithm [76] with custom chip definition developed by Microarray Lab, BrainArray [73].
BrainArray annotation ENTREGENE (version 12, available at http://brainarray.mbni.med.umich.edu/Brainarray/Database/CustomCDF/CDF_download.asp#v12) providing one remapped probe set per gene according to National Center for Biotechnology Information (NCBI) Homo sapiens ENTREGENE build 36.1 was used. One log2-transformed signal value per gene was obtained. The all normalized microarray data sets are available online at http://microarrays.curie.fr/publications/UMR144/RabBladder/ (login: rabbld; password: 248ba64). Concerning the deposition in ArrayExpress, we do not expect to receive the accession numbers immediately and will notify PLoS ONE when we will receive them.
Data were generated by using R language environment (version 2.12.0, available at http://www.r-project.org/) and bioconductor packages (available at http://www.bioconductor.org/).
Method of analysis
We used the statistical SAM (Significance Analysis of Microarrays) test [77], adapted for Excel, to compare the transcriptional levels of the genes between two different groups. We used Log2 (data) and the following parameters: Two class unpaired; Analysis type (Standard (genes)); T test statistic; No median center the array; Number of permutations: 100; Automatic estimate for s0 factor for denominator; Imputation Engine: K-Nearest Neighbours Imputer; Number of Neighbours: 10; Random number seed: 1234567.
We used Microsoft Office Excel 2003 to calculate the Pearson correlation. Log2(data) was used. The significance (two-tailed probability) of the Pearson correlation was calculated using the p-Value calculator of danielsoper.com (http://www.danielsoper.com/statcalc/calc44.aspx).
The binomial test on R (binom.test) was used with the parameter: alternative = c(“greater”).
The Student test pValue was calculated with the following Excel parameters: two tails and different variance hypothesis.
Supporting Information
Figure S1.
Rab clusters. A Rab cluster is defined as a Rab and its interacting proteins: the GEFs (guanine nucleotide exchange factors), the GAPs (GTPase activating proteins) and the effector proteins. Rab proteins are in yellow, GEFs in green, GAPs in light red and effector proteins in blue.
https://doi.org/10.1371/journal.pone.0039469.s001
(PDF)
Table S1.
List of papers in which the Rab partners were originally described.
https://doi.org/10.1371/journal.pone.0039469.s002
(XLS)
Table S2.
Genes deregulated during bladder cancer pathogenesis: second set of data. Second set of data. Affymetrix HG U95A/U95Av2 DNA microarrays. Genes down- or up-regulated in the tumor samples. Left: FGFR3-non-mutated tumor groups. Right: FGFR3-mutated tumor groups. The results shown to pass the thresholds: FC>1.5 (or <0.667) and qValue <5% with the first set of data are in red (or green) or highlighted in red (or green). The highlighted values in red or green pass the thresholds FC>1.5 or FC <0.667 or qValue <5% with the second set of data also. The results highlighted in yellow pass the thresholds: FC>1.5 (or <0.667) and qValue <5% only with the second set of data.
https://doi.org/10.1371/journal.pone.0039469.s003
(PDF)
Table S3.
Genes deregulated only in FGFR-non-mutated (A) and mutated tumors (B). Genes up- (or down)-regulated that pass the thresholds: FC>1.5 (or <0.667) and qValue <5% only for the FGFR3-non-mutated (A) or FGFR3-mutated (B) tumor pathway. Left column: comparison with the normal urothelium samples. Right column: comparison with the tumor samples of same stage with the opposite FGFR3 mutation status. The results that pass the above thresholds are highlighted (red for up-regulation and green for down-regulation).
https://doi.org/10.1371/journal.pone.0039469.s004
(PDF)
Table S4.
Genes significantly up- or down-regulated in each Rab cluster for each tumor group of the two pathways. “D” (green) corresponds to a gene significantly down-regulated in the tumor samples, “U” (red) corresponds to a gene significantly up-regulated in the tumor samples. “% All genes” corresponds to the total number of genes up- or down-regulated within each tumor group. A binomial test was applied for each Rab cluster within each tumor group. The binomial test pValues <1% are highlighted in orange. “Absent” corresponds to a gene absent in the Affymetrix HG U133 Plus 2.0 DNA microarrays.
https://doi.org/10.1371/journal.pone.0039469.s005
(XLS)
Table S5.
Pearson correlation (r) (and pValue) between the expression of MKI67 and the expression of genes listed in left column. Pearson correlation (r) (and pValue) between the expression of MKI67 and the expression of genes listed in left column in 28 Ta G1G2 (FGFR3-mutated) tumor samples and 63 T2-4 (FGFR3-non-mutated) tumor samples. Are highlighted (in green or red) when |r| >0.479 for the Ta G1G2 (FGFR3-mutated) group (28 samples) and |r| >0.323 for the T2–4 (FGFR3-non-mutated) group (63 samples).
https://doi.org/10.1371/journal.pone.0039469.s006
(PDF)
Table S6.
Pearson correlation (r) (and pValue) between the expression of different urothelial differentiation markers genes and the expression of genes listed in left column. Pearson correlation (r) (and pValue) between the expression of FOXA1, GRHL3, UPK1A/B, UPK2, UPK3A/B and the expression of genes listed in left column in 28 Ta G1/G2 (FGFR3-mutated) tumor samples and 63 T2–4 (FGFR3-non-mutated) tumor samples. Are highlighted (red corresponds to a correlation, green corresponds to an inverse correlation) when the two correlation values pass the thresholds: |r| >0.479 (pValue <1%) for the Ta G1G2 (FGFR3-mutated) group (28 samples) and |r| >0.323 (pValue <1%) for the T2–4 (FGFR3-non-mutated) group (63 samples).
https://doi.org/10.1371/journal.pone.0039469.s007
(XLS)
Acknowledgments
We thank Nicolas Stransky, Céline Vallot, Isabelle Bernard-Pierrot, Sandra Rebouissou, Sabine Bardin and Andrew Lindsay for helpful suggestions during this work. We thank Christophe Battail, Nicolas Servant and Georges Lucotte for their help with bioinformatic data. We thank members of the bladder cancer CIT (« carte d'identité des tumeurs ») consortium (Pascale Maillé, Henri Mondor hospital, Pathology Department; Dimitri Vordos, Henri Mondor hospital, Urology Department; Mathilde Sibony, Tenon hospital, Pathology Department; Agnès Laplanche, IGR, INSERM; Yves Denoux, Foch hospital, Pathology Department; Yann Neuzillet, Foch hospital, Urology Department; Aurélien de Reyniès, La Ligue Nationale Contre Le Cancer) for their constant support. The transcriptomic data used in this study were obtained through a collaboration with the Ligue Nationale Contre le Cancer (“Carte d'identité des tumeurs” program) for the first set of samples, and with AstraZeneca (Stransky et al. 2006 and unpublished results) for the second set of samples.
Author Contributions
Conceived and designed the experiments: BG FR JS. Performed the experiments: JH LK. Analyzed the data: JH EC RN. Contributed reagents/materials/analysis tools: YA TL SB. Wrote the paper: JH BG FR.
References
- 1. Stenmark H (2009) Rab GTPases as coordinators of vesicle traffic. Nat Rev Mol Cell Biol 10(8): 513–25.
- 2. Shimada K, Uzawa K, Kato M, Endo Y, Shiiba M, et al. (2005) Aberrant expression of RAB1A in human tongue cancer. Br J Cancer 92(10): 1915–21.
- 3. Lankat-Buttgereit B, Fehmann HC, Hering BJ, Bretzel RG, Göke B (1994) Expression of the ras-related rab3a gene in human insulinomas and normal human pancreatic islets. Pancreas 9(4): 434–8.
- 4. Gebhardt C, Breitenbach U, Richter KH, Fürstenberger G, Mauch C, et al. (2005) c-Fos-dependent induction of the small ras-related GTPase Rab11a in skin carcinogenesis. Am J Pathol. 167(1): 243–53.
- 5. Amillet JM, Ferbus D, Real FX, Antony C, Muleris M, et al. (2006) Characterization of human Rab20 overexpressed in exocrine pancreatic carcinoma. Hum Pathol. 37(3): 256–63.
- 6. Wu CY, Tseng RC, Hsu HS, Wang YC, Hsu MT (2009) Frequent down-regulation of hRAB37 in metastatic tumor by genetic and epigenetic mechanisms in lung cancer. Lung Cancer 63(3): 360–7.
- 7. Croizet-Berger K, Daumerie C, Couvreur M, Courtoy PJ, van den Hove MF (2002) The endocytic catalysts, Rab5a and Rab7, are tandem regulators of thyroid hormone production. Proc Natl Acad Sci USA 99(12): 8277–82.
- 8. Fukui K, Tamura S, Wada A, Kamada Y, Igura T, et al. (2007) Expression of Rab5a in hepatocellular carcinoma: Possible involvement in epidermal growth factor signaling. Hepatol Res. 37(11): 957–65.
- 9. Hou Q, Wu YH, Grabsch H, Zhu Y, Leong SH, et al. (2008) Integrative genomics identifies RAB23 as an invasion mediator gene in diffuse-type gastric cancer. Cancer Res. 68(12): 4623–30.
- 10. Cheng KW, Lahad JP, Kuo WL, Lapuk A, Yamada K, et al. (2004) The RAB25 small GTPase determines aggressiveness of ovarian and breast cancers. Nat Med. 10(11): 1251–6.
- 11. Agarwal R, Jurisica I, Mills GB, Cheng KW (2009) The emerging role of the RAB25 small GTPase in cancer. Traffic 10(11): 1561–8.
- 12. Goldenring JR, Nam KT (2011) Rab25 as a tumour suppressor in colon carcinogenesis. Br J Cancer 104(1): 33–6.
- 13. Milstein M, Mooser CK, Hu H, Fejzo M, Slamon D, et al. (2007) RIN1 is a breast tumor suppressor gene. Cancer Res. 67(24): 11510–6.
- 14. Pei L, Peng Y, Yang Y, Ling XB, Van Eyndhoven WG, et al. (2002) PRC17, a novel oncogene encoding a Rab GTPase-activating protein, is amplified in prostate cancer. Cancer Res. 62(19): 5420–4.
- 15. Ferlay J, Parkin DM, Steliarova-Foucher E (2010) Estimates of cancer incidence and mortality in Europe in 2008. Eur J Cancer 46(4): 765–81.
- 16. Jemal A, Siegel R, Xu J, Ward E (2010) Cancer statistics, 2010. CA Cancer J Clin. 60(5): 277–300.
- 17. Jacobs BL, Lee CT, Montie JE (2010) Bladder cancer in 2010: how far have we come? CA Cancer J Clin. 60(4): 244–72.
- 18. Knowles MA (2008) Molecular pathogenesis of bladder cancer. Int J Clin Oncol. 13(4): 287–97.
- 19. Cappellen D, De Oliveira C, Ricol D, de Medina S, Bourdin J, et al. (1999) Frequent activating mutations of FGFR3 in human bladder and cervix carcinomas. Nat Genet. 23(1): 18–20.
- 20. Billerey C, Chopin D, Aubriot-Lorton MH, Ricol D, Gil Diez de Medina S, et al. (2001) Frequent FGFR3 mutations in papillary non-invasive bladder (pTa) tumors. Am J Pathol. 158(6): 1955–9.
- 21. van Rhijn BW, Vis AN, van der Kwast TH, Kirkels WJ, Radvanyi F, et al. (2003) Molecular grading of urothelial cell carcinoma with fibroblast growth factor receptor 3 and MIB-1 is superior to pathologic grade for the prediction of clinical outcome. J Clin Oncol. 21(10): 1912–21.
- 22. Kimura T, Suzuki H, Ohashi T, Asano K, Kiyota H, et al. (2001) The incidence of thanatophoric dysplasia mutations in FGFR3 gene is higher in low-grade or superficial bladder carcinomas. Cancer 92(10): 2555–61.
- 23. Zieger K, Dyrskjøt L, Wiuf C, Jensen JL, Andersen CL, et al. (2005) Role of activating fibroblast growth factor receptor 3 mutations in the development of bladder tumors. Clin Cancer Res. 11(21): 7709–19.
- 24. Tomlinson DC, Baldo O, Harnden P, Knowles MA (2007) FGFR3 protein expression and its relationship to mutation status and prognostic variables in bladder cancer. J Pathol. 213(1): 91–8.
- 25. Luis NM, López-Knowles E, Real FX (2007) Molecular biology of bladder cancer. Clin Transl Oncol. 9(1): 5–12.
- 26. Heney NM, Ahmed S, Flanagan MJ, Frable W, Corder MP, et al. (1983) Superficial bladder cancer: progression and recurrence. J Urol. 130(6): 1083–6.
- 27. Southgate J, Hutton KA, Thomas DF, Trejdosiewicz LK (1994) Normal human urothelial cells in vitro: proliferation and induction of stratification. Lab Invest. 71(4): 583–94.
- 28. Varley CL, Bacon EJ, Holder JC, Southgate J (2009) FOXA1 and IRF-1 intermediary transcriptional regulators of PPARgamma-induced urothelial cytodifferentiation. Cell Death Differ. 16(1): 103–14.
- 29. Gerdes J, Li L, Schlueter C, Duchrow M, Wohlenberg C, et al. (1991) Immunobiochemical and molecular biologic characterization of the cell proliferation-associated nuclear antigen that is defined by monoclonal antibody Ki-67. Am J Pathol. 138(4): 867–73.
- 30. Lobban ED, Smith BA, Hall GD, Harnden P, Roberts P, et al. (1998) Uroplakin gene expression by normal and neoplastic human urothelium. Am J Pathol. 153(6): 1957–67.
- 31. Olsburgh J, Weeks R, Selby P, Southgate J (2002) Human uroplakin lb gene structure and promoter analysis. Biochim Biophys Acta. 1576(1–2): 163–70.
- 32. Deng FM, Liang FX, Tu L, Resing KA, Hu P, et al. (2002) Uroplakin IIIb, a urothelial differentiation marker, dimerizes with uroplakin Ib as an early step of urothelial plaque assembly. J Cell Biol. 159(4): 685–94.
- 33. Olsburgh J, Harnden P, Weeks R, Smith B, Joyce A, et al. (2003) Uroplakin gene expression in normal human tissues and locally advanced bladder cancer. J Pathol. 199(1): 41–9.
- 34. Yu Z, Mannik J, Soto A, Lin KK, Andersen B (2009) The epidermal differentiation-associated Grainyhead gene Get1/Grhl3 also regulates urothelial differentiation. EMBO J. 28(13): 1890–903.
- 35. Dyrskjøt L, Kruhøffer M, Thykjaer T, Marcussen N, Jensen JL, et al. (2004) Gene expression in the urinary bladder: a common carcinoma in situ gene expression signature exists disregarding histopathological classification. Cancer Res. 64(11): 4040–8.
- 36. Vallot C, Stransky N, Bernard-Pierrot I, Hérault A, Zucman-Rossi J, et al. (2011) A novel epigenetic phenotype associated with the most aggressive pathway of bladder tumor progression. J Natl Cancer Inst. 103(1): 47–60.
- 37. Fischer J, Weide T, Barnekow A (2005) The MICAL proteins and rab1: a possible link to the cytoskeleton? Biochem Biophys Res Commun. 328(2): 415–23.
- 38. Ashida S, Furihata M, Katagiri T, Tamura K, Anazawa Y, et al. (2006) Expression of novel molecules, MICAL2-PV (MICAL2 prostate cancer variants), increases with high Gleason score and prostate cancer progression. Clin Cancer Res. 12(9): 2767–73.
- 39. Graham ME, Handley MT, Barclay JW, Ciufo LF, Barrow SL, et al. (2008) A gain-of-function mutant of Munc18-1 stimulates secretory granule recruitment and exocytosis and reveals a direct interaction of Munc18-1 with Rab3. Biochem J. 409(2): 407–16.
- 40. Zhang X, He X, Fu XY, Chang Z (2006) Varp is a Rab21 guanine nucleotide exchange factor and regulates endosome dynamics. J Cell Sci. 119(Pt 6): 1053–62.
- 41. Simpson JC, Griffiths G, Wessling-Resnick M, Fransen JA, Bennett H, et al. (2004) A role for the small GTPase Rab21 in the early endocytic pathway. J Cell Sci. 117(Pt 26): 6297–311.
- 42. Burgo A, Sotirakis E, Simmler MC, Verraes A, Chamot C, et al. (2009) Role of Varp, a Rab21 exchange factor and TI-VAMP/VAMP7 partner, in neurite growth. EMBO Rep. 10(10): 1117–24.
- 43. Tamura K, Ohbayashi N, Maruta Y, Kanno E, Itoh T, et al. (2009) Varp is a novel Rab32/38-binding protein that regulates Tyrp1 trafficking in melanocytes. Mol Biol Cell. 20(12): 2900–8.
- 44. Hales CM, Griner R, Hobdy-Henderson KC, Dorn MC, Hardy D, et al. (2001) Identification and characterization of a family of Rab11-interacting proteins. J Biol Chem. 276(42): 39067–75.
- 45. Khandelwal P, Ruiz WG, Balestreire-Hawryluk E, Weisz OA, Goldenring JR, et al. (2008) Rab11a-dependent exocytosis of discoidal/fusiform vesicles in bladder umbrella cells. Proc Natl Acad Sci U S A 105(41): 15773–8.
- 46. Tang BL (2010) Is Rab25 a tumor promoter or suppressor–context dependency on RCP status? Tumour Biol. 31(4): 359–61.
- 47. Izumi T (2007) Physiological roles of Rab27 effectors in regulated exocytosis. Endocr J. 54(5): 649–57.
- 48. Starr DA, Saffery R, Li Z, Simpson AE, Simpson AE, et al. (2000) HZwint-1, a novel human kinetochore component that interacts with HZW10. J Cell Sci. 113(Pt 11): 1939–50.
- 49. Famulski JK, Vos L, Sun X, Chan G (2008) Stable hZW10 kinetochore residency, mediated by hZwint-1 interaction, is essential for the mitotic checkpoint. J Cell Biol. 180(3): 507–20.
- 50. Endoh H, Tomida S, Yatabe Y, Konishi H, Osada H, et al. (2004) Prognostic model of pulmonary adenocarcinoma by expression profiling of eight genes as determined by quantitative real-time reverse transcriptase polymerase chain reaction. J Clin Oncol. 22(5): 811–9.
- 51. Liu YJ, Wang Q, Li W, Huang XH, Zhen MC, et al. (2007) Rab23 is a potential biological target for treating hepatocellular carcinoma. World J Gastroenterol. 13(7): 1010–7.
- 52. Kim KR, Oh SY, Park UC, Wang JH, Lee JD, et al. (2007) [Gene expression profiling using oligonucleotide microarray in atrophic gastritis and intestinal metaplasia]. Korean J Gastroenterol. 49(4): 209–24.
- 53. Wang Y, Ng EL, Tang BL (2006) Rab23: what exactly does it traffic? Traffic 7(6): 746–50.
- 54. Shin K, Lee J, Guo N, Kim J, Lim A, et al. (2011) Hedgehog/Wnt feedback supports regenerative proliferation of epithelial stem cells in bladder. Nature 472(7341): 110–4.
- 55. Cormont M, Metón I, Mari M, Monzo P, Keslair F, et al. (2003) CD2AP/CMS regulates endosome morphology and traffic to the degradative pathway through its interaction with Rab4 and c-Cbl. Traffic 4(2): 97–112.
- 56. Ng EL, Ng JJ, Liang F, Tang BL (2009) Rab22B is expressed in the CNS astroglia lineage and plays a role in epidermal growth factor receptor trafficking in A431 cells. J Cell Physiol. 221(3): 716–28.
- 57. Bellmunt J, Hussain M, Dinney CP (2003) Novel approaches with targeted therapies in bladder cancer. Therapy of bladder cancer by blockade of the epidermal growth factor receptor family. Crit Rev Oncol Hematol. 46:
- 58. Kotzsch M, Sieuwerts AM, Grosser M, Meye A, Fuessel S, et al. (2008) Urokinase receptor splice variant uPAR-del4/5-associated gene expression in breast cancer: identification of rab31 as an independent prognostic factor. Breast Cancer Res Treat. 111(2): 229–40.
- 59. Caswell PT, Spence HJ, Parsons M, White DP, Clark K, et al. (2007) Rab25 associates with alpha5beta1 integrin to promote invasive migration in 3D microenvironments. Dev Cell. 13(4): 496–510.
- 60. Andrade VC, Vettore AL, Panepucci RA, Almeida MS, Yamamoto M, et al. (2010) Number of expressed cancer/testis antigens identifies focal adhesion pathway genes as possible targets for multiple myeloma therapy. Leuk Lymphoma 51(8): 1543–9.
- 61. Fang Z, Yao W, Xiong Y, Zhang J, Liu L, et al. (2010) Functional elucidation and methylation-mediated downregulation of ITGA5 gene in breast cancer cell line MDA-MB-468. J Cell Biochem. 110(5): 1130–41.
- 62. Valastyan S, Chang A, Benaich N, Reinhardt F, Weinberg RA (2010) Concurrent suppression of integrin alpha5, radixin, and RhoA phenocopies the effects of miR-31 on metastasis. Cancer Res. 70(12): 5147–54.
- 63. Chen Y, Guo X, Deng FM, Liang FX, Sun W, et al. (2003) Rab27b is associated with fusiform vesicles and may be involved in targeting uroplakins to urothelial apical membranes. Proc Natl Acad Sci U S A 100(24): 14012–7.
- 64. Goi T, Takeuchi K, Katayama K, Hirose K, Yamaguchi A (2002) Mutations of rabphillin-3A-like gene in colorectal cancers. Oncol Rep. 9(6): 1189–92.
- 65. Fontijn RD, Goud B, Echard A, Jollivet F, van Marle J, et al. (2001) The human kinesin-like protein RB6K is under tight cell cycle control and is essential for cytokinesis. Mol Cell Biol. 21(8): 2944–55.
- 66. Neef R, Grüneberg U, Barr FA (2005) Assay and functional properties of Rabkinesin-6/Rab6-KIFL/MKlp2 in cytokinesis. Methods Enzymol. 403: 618–28.
- 67. Lu Y, Liu P, Wen W, Grubbs CJ, Townsend RR, et al. (2010) Cross-species comparison of orthologous gene expression in human bladder cancer and carcinogen-induced rodent models. Am J Transl Res. 3(1): 8–27.
- 68. Sobin LH, Fleming ID (1997) TNM Classification of Malignant Tumors, fifth edition. Union Internationale Contre le Cancer and the American Joint Committee on Cancer. Cancer 1997 80(9): 1803–4.
- 69.
Mostofi FK, Sobin LH, Torloni H (1973) International Histological Clasification of Tumors, No 10: Histological Typing of Urinary Bladder Tumors World Health Organization, Geneva, 1973.
- 70. van Oers JM, Lurkin I, van Exsel AJ, Nijsen Y, van Rhijn BW, et al. (2005) A simple and fast method for the simultaneous detection of nine fibroblast growth factor receptor 3 mutations in bladder cancer and voided urine. Clin Cancer Res. 11(21): 7743–8.
- 71. Stransky N, Vallot C, Reyal F, Bernard-Pierrot I, de Medina SG, et al. (2006) Regional copy number-independent deregulation of transcription in cancer. Nat Genet. 38(12): 1386–96.
- 72. Diez de Medina SG, Chopin D, El Marjou A, Delouvée A, LaRochelle WJ, et al. (1997) Decreased expression of keratinocyte growth factor receptor in a subset of human transitional cell bladder carcinomas. Oncogene 14(3): 323–30.
- 73. Dai M, Wang P, Boyd AD, Kostov G, Athey B, et al. (2005) Evolving gene/transcript definitions significantly alter the interpretation of GeneChip data. Nucleic Acids Res. 33(20): e175.
- 74. Varley CL, Stahlschmidt J, Lee WC, Holder J, Diggle C, et al. (2004) Role of PPARgamma and EGFR signalling in the urothelial terminal differentiation programme. J Cell Sci. 117(Pt 10): 2029–36.
- 75. Chirgwin JM, Przybyla AE, MacDonald RJ, Rutter WJ (1979) Isolation of biologically active ribonucleic acid from sources enriched in ribonuclease. Biochemistry 18(24): 5294–9.
- 76. Bolstad BM, Irizarry RA, Astrand M, Speed TP (2003) A comparison of normalization methods for high density oligonucleotide array data based on variance and bias. Bioinformatics 19(2): 185–93.
- 77. Tusher VG, Tibshirani R, Chu G (2001) Significance analysis of microarrays applied to the ionizing radiation response. Proc Natl Acad Sci USA 98(9): 5116–21.