Down-regulation of RAB6KIFL/KIF20A, a Kinesin Involved with Membrane Trafficking of Discs Large Homologue 5, Can Attenuate Growth of Pancreatic Cancer Cell

Pancreatic ductal adenocarcinoma (PDAC) is the forth leading cause of cancer death in the Western world and shows one of the worst mortality rates among malignancies, with a 5-year survival rate of only 4% (1, 2). Approximately 30,700 patients are diagnosed with pancreatic cancer in the United States alone, and nearly 30,000 will die of the disease (3). Because most PDAC patients are diagnosed at an advanced stage, no available therapy is effective; surgical resection offers the only possibility for cure at present, but 80% to 90% of PDAC patients who undergo surgery relapse and die from the disease (1, 2). Some approaches in surgery and chemotherapy, including 5-fluorouracil or gemcitabine, with or without radiation, can improve patients' quality of life (1, 2), but those treatments have only a limited effect on long-term survival because PDACs are extremely aggressive and chemoresistant. Hence, at present the management of most patients is focused on palliative measures (1).

To overcome this nearly hopeless situation, development of novel molecular therapies for PDAC through identification of molecular targets is an urgent priority. Earlier we generated precise expression profiles of PDACs using genome-wide cDNA microarrays containing approximately 27,000 genes, in combination with laser microdissection to obtain pure populations of cancer cells for testing (4). Among the genes we identified as being overexpressed in PDACs was RAB6KIFL, which we investigated as a novel molecular target for this disease.

RAB6KIFL (KIF20A) belongs to the family of kinesin proteins, which are characterized by a conserved motor domain that binds to microtubules and couples ATP hydrolysis to generate mechanical force (5, 6). Kinesin family plays a variety of roles in cellular functions, being involved in diverse processes including formation of the mitotic spindle and chromosome partitioning, as well as intracellular movement of organelles and vesicles (5, 7). RAB6KIFL was first reported to play a role in the dynamics of the Golgi apparatus through direct interaction with Rab6 small GTPase (6). Recent reports (8, 9) have showed that RAB6KIFL accumulates in mitotic cells, where it localizes to the midzone of the spindle during anaphase and to the cleavage furrow and midbody during telophase; in other experiments, microinjection of antibodies to RAB6KIFL resulted in multinuclear cells, indicating an essential role of RAB6KIFL in cytokinesis. However, the significance of its overexpression in human carcinogenesis has not been investigated.

In the study reported here, we found evidence for an essential role of RAB6KIFL in pancreatic carcinogenesis. We also determined that discs, large homologue 5 (DLG5), a membrane-scaffold molecule, is a cargo protein of RAB6KIFL and plays a similarly important role in the viability of PDAC cells.

Cell Lines. PDAC cell line MIA-Paca2 and embryonic kidney cell line HEK293, both purchased from the American Type Culture Collection (ATCC, Rockville, MD), were grown in DMEM (Sigma-Aldrich Corp., St. Louis, MO). PDAC cell lines PK59 and KLM1 were provided by the Cell Resource Center for Biomedical Research, Tohoku University (Sendai, Japan) and maintained in RPMI 1640 (Sigma-Aldrich); both media were supplemented with 10% fetal bovine serum (Cansera International, Ontario, Canada) and 1% antibiotic/antimycotic solution (Sigma-Aldrich). Cells were maintained at 37°C in atmospheres of humidified air with 5% CO2.

Semiquantitative reverse transcription–PCR for RAB6KIFL and DLG5. Purification of PDAC cells and normal pancreatic ductal epithelium were described previously (4); RNAs from the purified cell populations were subjected to two rounds of amplification by T7-based in vitro transcription (Epicentre Technologies, Madison, WI). From the three human pancreatic cancer cell lines listed above, total RNAs were extracted using Trizol reagent (Invitrogen Life Technologies, Carlsbad, CA) according to the manufacturer's recommendations, treated with DNase I(Roche Diagnostic, Mannheim, Germany), and reversely transcribed to single-stranded cDNAs using oligodeoxythymidylic acid primer with Superscript II reverse transcriptase (Invitrogen). We prepared appropriate dilutions of each single-stranded cDNA for subsequent PCR amplification by monitoring β-actin (ACTB) as quantitative control. The primer sequences were 5′-CATCCACGAAACTACCTTCAACT-3′ and 5′-TCTCCTTAGAGAGAAGTGGGGTG-3′ for ACTB; 5′-GTACCAACCAGGAAAATCAG-3′ and 5′-TGTCTGA GTATTGCATCCTG-3′ for RAB6KIFL; and 5′-GGGTCCTCCTTAGGCTCTGT-3′ and 5′-GCCGTGAGTGCTGTAATCAA-3′ for DLG5. All reactions involved initial denaturation at 94°C for 2 minutes followed by 21 cycles (for ACTB) or 28 cycles (for RAB6KIFL and DLG5) at 94°C for 30 seconds, 58°C for 30 seconds, and 72°C for 1 minute, on a GeneAmp PCR system 9700 (PE Applied Biosystems, Foster, CA).

Northern Blot Analysis. Human multiple-tissue Northern blots (BD Bioscience, Palo Alto, CA) were hybridized with an [α-³²P]dCTP-labeled, 1170-bp PCR product of RAB6KIFL that was prepared as a probe by reverse transcription–PCR (RT-PCR) using primers 5′-CAGACACAGGCCTTGATGATG-3′ and 5′-AGTGTCTGAGTATTGCATCCTGG-3′. Prehybridization, hybridization, and washing were done according to the supplier's recommendations. The blots were autoradiographed with intensifying screens at −80°C for 7 days.

Antibody to RAB6KIFL Protein. A cDNA fragment encoding the 430 N-terminal amino acids of RAB6KIFL was amplified by PCR using primers 5′-CCGGAATTCATGTCGCAAGGGATCCTTTC-3′ and 5′-ATTTGCGGCCGCATCAAATGTTTCCTGCTTC-3′, which contained EcoRI and NotI restriction sites indicated by the first and second underlines, respectively. The product was cloned into pET28a vector (Novagen, Madison, WI) to produce a fusion protein, bearing an N-terminal 6-histidine tag, which was used to immunize rabbits; the resulting polyclonal antibody was affinity purified using Affi-gel 10 (Bio-Rad Laboratories, Hercules, CA) conjugated with the 6-histidine fused protein.

Immunohistochemical Staining. Paraffin-embedded tissue sections were obtained from specimens that had been resected during surgery at the Department of Surgery, Osaka Medical Center for Cancer and Cardiovascular Diseases. In conformity with the principles of the Declaration of Helsinki, informed consent had been obtained from each patient. Tissue sections from normal pancreas were purchased from Biochain (Hayward, CA). The sections were deparaffinized and autoclaved at 108°C for 15 minutes. After endogenous peroxidase activity was quenched by incubation for 30 minutes in 0.33% hydrogen peroxide diluted in methanol, the samples were incubated with fetal bovine serum for blocking and with anti-RAB6KIFL polyclonal antibody at room temperature for 1 hour and washed with PBS. Immunodetection was achieved with peroxidase-labeled anti-rabbit immunoglobulin (Dako Cytomation, Carpinteria, CA). Finally, the reactants were developed with 3,3′-diaminobenzidine (Dako) and the sections were counterstained with hematoxylin.

Immunoprecipitation and Mass Spectrometric Analysis for RAB6KIFL-Associated Complexes. To isolate proteins that associated directly with RAB6KIFL, we did immunoprecipitation experiments using the anti-RAB6KIFL antibody. PDAC cell lines KLM1 and PK59, which overexpressed RAB6KIFL, were lysed in lysis buffer [50 mmol/L Tris-HCl (pH 8.0), 150 mmol/L NaCl, 0.5% NP40, Protease Inhibitor Cocktail Set III (Calbiochem, San Diego, CA)]. Equal amounts of total proteins were incubated at 4°C for1hour with 2 μg of anti-RAB6KIFL polyclonal antibody or a rabbit immunoglobulin G (Santa Cruz Biotechnologies, Santa Cruz, CA). Immunocomplexes were incubated with protein G-Sepharose (Zymed Laboratories, South San Francisco, CA) for 1 hour and washed with lysis buffer. Coprecipitated proteins were separated in 5% to 20% gradient SDS-PAGE and stained by silver-staining kits (Wako, Osaka, Japan). Bands that differentiated proteins precipitated with anti-RAB6KIFL polyclonal antibody from those precipitated with control immunoglobulin G were excised, digested in gel with trypsin, and analyzed for peptide-mass fingerprints using an AXIMA-CFR +6 mass spectrometer (Shimadzu Corporation, Tsukuba, Japan). Peptide masses were searched with 10-ppm mass accuracy, and protein database searches were done using the database-fitting program IntelliMarque (Shimadzu).

In vivo Binding Between RAB6KIFL and DLG5. The entire coding sequence of RAB6KIFL cDNA was amplified by RT-PCR with primers 5′-CCGGAATTCATGTCGCAAGGGATCCTTTC-3′ (forward) and 5′-ATTTGCGGCCGCCAGTACTTTTTGCCAAA-3′ (reverse), and the product was inserted into the EcoRI and NotI sites of pCAGGS for expressing a FLAG-tagged protein (10). The entire coding sequence of DLG5 cDNA was amplified by RT-PCR with primers 5′-CGCGGATCCACATGCCCTCAGACTCAGAAAG-3′ (forward) and 5′-ATTTGCGGCCGCCAGAGCGGGCAGGCTGGAATCC-3′ (reverse), and that product was inserted into the BamHI and NotI sites ofpcDNA3.1(+)/myc-His A (Invitrogen) for expressing a myc-tagged protein. FLAG-tagged RAB6KIFL and myc-tagged DLG5 were transiently cotransfected into HEK293 cells. The transfected cells were lysed as described above and immunoprecipitated with 5 μg of agarose-conjugated mouse anti-FLAG antibody (Sigma-Aldrich) or mouse anti-myc antibody (Santa Cruz). To examine interaction of FLAG-RAB6KIFL with myc-DLG5, we analyzed the immune complexes by Western blotting with rabbit anti-FLAG and anti-myc antibodies.

Analysis of RAB6KIFL and DLG5 by Immunofluorescence. PK59 cells were fixed with 4% paraformaldehyde, then permeabilized with 0.1% Triton X-100 in PBS for 1 minute at room temperature, covered with blocking solution (3% bovine serum albumin/PBS) for 30 minutes at room temperature, and finally incubated for 60 minutes at room temperature with rabbit anti-RAB6KIFL polyclonal antibody and FITC-conjugated mouse anti-α-tubulin antibody (Sigma-Aldrich) in blocking solution. After washing with PBS, the cells were stained by Alexa594-conjugated anti-rabbit secondary antibody (Molecular Probes, Eugene, OR) for 60 minutes at room temperature and washed with PBS. Each specimen was mounted with VECTASHIELD (Vector Laboratories Inc., Burlingame, CA) containing 4′,6-diamidino-2-phenylindole (DAPI) and visualized with spectral confocal scanning systems (Leica, Bensheim, Germany). To determine whether RAB6KIFL and DLG5 were colocalized, myc-tagged DLG5 was transfected into KLM1 cells, which endogenously express a high level of RAB6KIFL. After the cells were fixed as described above, they were incubated with rabbit anti-RAB6KIFL and mouse anti-myc antibodies (Santa Cruz), washed with PBS, and stained by Alexa594-conjugated anti-rabbit and Alexa488-conjugated anti-mouse secondary antibodies (Molecular Probes). Each specimen was mounted with VECTASHIELD and visualized with spectral confocal scanning systems (Leica).

Effect of Small Interfering RNA on Growth of PDAC Cells. To evaluate the biological functions of RAB6KIFL and DLG5 in PDAC cells, we used a psiU6BX3.0 vector for expression of short-hairpin RNA against the target gene, as described previously (11). The U6 promoter was cloned upstream of the gene-specific sequence (19-nucleotide sequence from the target transcript, separated from the reverse complement of the same sequence by a short spacer, TTCAAGAGA), with five thymidines as a termination signal and a neo-cassette for selection by Geneticin (Sigma). The target sequences for RAB6KIFL were 5′-TTGGCCAAGCCACACACAG-3′ (RAB6KIF-si1), 5′-GTTCTCAGCCATTGCTAGC-3′ (RAB6KIF-si2), 5′-GGCAGCATGTATTGCTGAG-3′ (RAB6KIF-si3), and 5′-GAAGCAGCACGACTTCTTC-3′ (for siEGFP as a negative control). The target sequences for DLG5 were 5′-GCTCATCATGAGTGAGCGT-3′ (DLG5-si1), 5′-GCTCAAGAGCAGCACATCT-3′ (DLG5-si2), and 5′-GATTCGGTGAGCCTGGCCT-3′ (DLG5-si3). Human PDAC cell lines MIA-Paca2 and PK59 were plated onto 10-cm dishes (5 × 10⁵ cells per dish), and transfected with psiU6BX vectors that included the target sequences for EGFP, RAB6KIFL, or DLG5, using LipofectAMINE 2000 (Invitrogen) for MIA-Paca2 and FuGENE6 (Roche) for PK59, according to the manufacturers' instructions. Cells were selected in medium containing 500 μg/mL of geneticin (Invitrogen) for 10 days and harvested after 48 hours for RT-PCR analysis of knockdown effects on RAB6KIFL and DLG5. Primers for these RT-PCR experiments were the same as those described above. After 10 days of incubation these cells were stained by Giemsa solution to assess colony formation, and cell numbers were assessed by 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) assay.

Inhibition of Endogenous RAB6KIFL Expression by Small Interfering RNA–Inducible Expression Vector. To down-regulate endogenous RAB6KIFL expression in PDAC cells in an inducible manner, we used a doxycycline-inducible small interfering RNA (siRNA) expression system. A pTER vector was kindly provided by Dr. van de Wetering at the Hubrecht Laboratory, Center for Biomedical Genetics, Utrecht, Netherlands (12). This vector had been designed as a doxycycline-regulated form of the RNA polymerase III H1 promoter to allow inducible knockdown of gene expression by short-hairpin RNAs. To construct the pTER-RAB6KIFL vector, we phosphorylated the oligonucleotide sequences of RAB6KIFL-si2 described above, as well as the negative control (siEGFP), by incubating the oligonucleotides with T4-polynucleotide kinase (Toyobo, Osaka, Japan) before ligation into the pTER vector. A tetracycline-repressor expression construct, pcDNA6/TR (Invitrogen), was used to generate clones expressing the tetracycline repressor (13). ThepcDNA6/TR vector was transfected into KLM1 cells, using Blasticidin (Sigma-Aldrich) to select stable transfectants. Clones that stably expressed the tetracycline repressor were subsequently transfected with pTER-RAB6KIFL-si2 or pTER-siEGFP, and selected using Zeocin (Invitrogen). In this way, we obtained derivative KLM1 cells containing pTER-RAB6KIFL-si2 (KLM1-siRAB6KIFL) or pTER-siEGFP (KLM1-siEGFP). After inducing the RAB6KIFL-specific siRNA by addition of doxycycline (1.0 μg/mL, Sigma-Aldrich), we analyzed selected clones by Western blotting for their ability to down-regulate RAB6KIFL, using anti-RAB6KIFL polyclonal antibody.

To evaluate the subcellular localization of RAB6KIFL-associated DLG5, we transfected a myc-tagged DLG5 expression construct into the KLM1-siRAB6KIFL cells, in which endogenous RAB6KIFL expression could be down-regulated in an inducible manner, and into the KLM1-siEGFP cells as a control. These cells were incubated with or without doxycycline for 3 to 5days. To analyze the subcellular localization of exogenous myc-DLG5 in these cells, immunostaining was done on day 5 as described above, using mouse anti-myc antibody.

Overexpression of RAB6KIFL in PDAC Cells. Among a number of genes that were overexpressed in pancreatic tumor cells on a genome-wide cDNA microarray analysis (4), we focused on RAB6KIF for this study. Our microarray data had shown strong overexpression of RAB6KIFL in all of the informative cases of PDAC examined (six of six informative cases showed >10-fold expression signal), and its overexpression was confirmed by RT-PCR in all of the nine microdissected PDAC cell populations examined (Fig. 1A). Northern blot analysis using an RAB6KIFL cDNA fragment as the probe identified a transcript of about 3.0 kb that was highly expressed in testis and at a very low level in bone marrow and thymus; no expression was observed in any other vital organs including lung, heart, liver, and kidney (Fig. 1B). Immunohistochemical analysis using a polyclonal antibody to RAB6KIFL showed strong signals in the cytoplasm and nuclei of PDAC cells in all of the PDAC tissue sections from five additional patients, although its presence was more predominant in cytoplasm than in nuclei. No staining was observed in normal pancreatic epithelia (Fig. 1C).

Variable Subcellular Localization of RAB6KIFL in PDAC Cells. We examined the subcellular localization of endogenous RAB6KIFL in more detail in PDAC cell line PK59 by immunofluorescence analysis (Fig. 2). In interphase cells, the staining pattern of RAB6KIFL in the cytoplasm overlapped with that of α-tubulin; RAB6KIFL was also detected in the nucleus. In metaphase cells possessing a short spindle, the two proteins colocalized in microtubules together with cytosolic staining. In cells at anaphase, RAB6KIFL staining was present in a short filamentous band extending across the spindle midzone. When cells were at telophase, RAB6KIFL was sharply concentrated in the midbody. On the other hand, the continuous presence of RAB6KIFL in the cytoplasm was observed in all cell cycle phases.

Effect of RAB6KIFL--Small Interfering RNAs on Growth of PDAC Cells. We constructed several siRNA expression vectors specific to RAB6KIFL tetracycline sequences and transfected them into MIA-Paca2 and PK59, PDAC cell lines that endogenously express high levels of RAB6KIFL mRNA. A knockdown effect was confirmed by RT-PCR when we used RAB6KIFL-si2 and -si3 constructs (Fig. 3A). Colony-formation assays (Fig. 3B) and MTT assays (Fig. 3C) using MIA-Paca2 revealed a drastic reduction in the number of cells transfected with RAB6KIFL-si2, compared with RAB6KIFL-si1 for which no knockdown effect was apparent. Similar effects were obtained with the PK59 cell line (data not shown). Although previous studies had shown that microinjection of anti-RAB6KIFL antibody into HeLa cells caused cells to become multinucleated (8, 9), knockdown of endogenous RAB6KIFL in PDAC cell lines did not induce multinuclear cells (data not shown), indicating that the growth suppression observed in PDACs was not simply associated with the failure of cytokinesis.

Identification of DLG5 as a Cargo of RAB6KIFL Protein. To investigate the biological functions of RAB6KIFL further, we carried out immunoprecipitation experiments to detect possible complexes of RAB6KIFL with other protein(s), using a polyclonal antibody to RAB6KIFL. Silver-stained immunoprecipitated fractions separated on SDS-PAGE gels showed that RAB6KIFL-associated precipitates contained at least two other proteins, of approximately 200 and 300 kDa, respectively, compared with results from a control sample (data not shown). Each protein was analyzed by a matrix-assisted laser desorption ionization-time of flight (MALDI-TOF) system after in-gel trypsin digestion; the 200-kDa protein was identified as discs, large homologue 5 (DLG5), and the other was hypothetical protein FLJ20035. Because the biological role of the latter is unknown, we investigated the biological function of RAB6KIFL through its interaction with DLG5.

To validate that interaction we did coimmunoprecipitation experiments. First, FLAG-tagged RAB6KIFL and myc-tagged DLG5 constructs were cotransfected into HEK293 cells and the cell lysates were immunoprecipitated using mouse anti-FLAG or anti-myc antibody. Immunoblotting of the precipitates using rabbit anti-FLAG and anti-myc antibodies revealed that FLAG-RAB6KIFL had coprecipitated with myc-DLG5 (Fig. 4A).

We subsequently did immunocytochemical analysis to examine whether RAB6KIFL and DLG5 would be colocalized in PDAC cells. The myc-tagged DLG5 construct was transfected into KLM1 cells, in which RAB6KIFL was expressed at a high level (confirmed by Western blot analysis, data not shown). The subcellular localization of DLG5 was visualized with mouse anti-myc antibody and Alexa488-labeled secondary antibody; the same cells were also incubated with anti-RAB6KIFL polyclonal antibody and Alexa594-labeled secondary antibody. Exogenous myc-DLG5 was observed as granular signals in the cytoplasm that were concentrated under the plasma membrane of the cell-cell contact sites (Fig. 4B). Endogenous RAB6KIFL was present in both cytoplasm and nucleus and partially colocalized with myc-DLG5 at the cell-cell contact sites of the plasma membrane.

Effect of DLG5--Small Interfering RNAs on Growth of PDAC Cells. To investigate the biological significance of DLG5 expression in PDAC cells, we constructed plasmids expressing siRNAs specific to DLG5 and transfected them into PDAC cell lines to assess their knockdown effect on endogenous DLG5. MIA-Paca2 and PK59 cells, in which DLG5 and RAB6KIFL are both expressed at a high level, were transfected with one of three independent DLG5-specific siRNA-expressing vectors. RT-PCR experiments revealed that DLG5-si2 significantly suppressed endogenous expression of DLG5 48 hours after transfection (Fig. 5A). Because of this reduced expression, growth of MIA-Paca2 and PK59 cells transfected with DLG5-si2 was suppressed significantly; the colony-formation assay and MTT experiments revealed reduction in the numbers of colonies and of viable cells (Fig. 5B and C for MIA-Paca2; data not shown for PK59), indicating that DLG5 and RAB6KIFL are both associated with viability of PDAC cells.

Alteration in Trafficking of DLG5 after Knockdown of Endogenous RAB6KIFL. Because we had identified DLG5 as a candidate cargo molecule of the RAB6KIFL motor kinesin, we hypothesized that RAB6KIFL would be essential for membrane trafficking of this scaffold protein. To examine the effect of endogenous RAB6KIFL on the subcellular localization of DLG5, we generated KLM1 derivative cells (KLM1-siRAB6KIFL) in which RAB6KIFL was down-regulated by RAB6KIFL-si2 in a doxycycline-inducible siRNA expression system. Stable integration of the inducible siRNA vector in KLM1-siRAB6KIFL cells allowed specific down-regulation of endogenous RAB6KIFL by induction with doxycycline for 3 to 5 days (Fig. 6A). We transiently transfected myc-tagged DLG5 into KLM1 derivatives that had been cultured in medium with or without doxycycline for 3 days. Myc-tagged DLG5 was observed in cytoplasm but not at cell-cell contact sites in cells in which RAB6KIFL was knocked down by expression of siRNA (Fig. 6B). However, myc-DLG5 was localized in subcellular membranes at cell-cell contact sites in the cells without doxycycline or in the control KLM1 derivative cells (KLM1-siEGFP) containing the inducible siEGFP expression vector with doxycycline. The results indicated that DLG5 was not transported to the plasma membrane in cells in which RAB6KIFL was knocked down (KLM1-siRAB6KIFL); therefore RAB6KIFL is likely to play a critical role in trafficking of DLG5 to sites in which cells contact each other. Moreover, 1 week after incubation with doxycycline, we observed a significant reduction in the number of KLM1-siRAB6KIFL colonies compared with control cells (KLM1-siEGFR; Fig. 6C).

In this study we focused on RAB6KIFL, a member of the kinesin superfamily, which had been overexpressed in pancreatic cancers (PDACs) analyzed on a cDNA microarray. Among a number of overexpressed genes we identified by our genome-wide cDNA microarray (4), we selected RAB6KIFL for our detail analysis because its signal intensity of cancer cells in cDNA microarray was obvious (more than 10 times) comparing with that of normal ductal cells, and RAB6KIFL expression pattern in normal adult tissues was quite restricted according to our previous microarray data analyzing the gene expression profile of 29 normal human tissues (14). In selecting molecular targets against cancer cells, restricted expression pattern in normal adult tissue is an important factor, considering the side effect of the treatment targeting the candidates.

Recently, RAB6KIFL was documented to accumulate in the midzone of the spindle during anaphase and to the cleavage furrow and midbody during telophase (8). That information, combined with evidence that microinjection of anti-RAB6KIFL blocking antibody caused HeLa cells to become multinuclear, implied an important role of RAB6KIFL in cytokinesis (8). On the other hand, the continuous presence of RAB6KIFL in the cytoplasm in all cell cycle phases suggests functions in the cytoplasm in addition to cytokinesis and mitosis. In our experiments, down-regulation of endogenous RAB6KIFL in PDAC cells by siRNA resulted in drastic attenuation of cancer cell growth, although we did not find multinuclear cells in any PDAC samples we examined. Hence, we investigated a possible additional biological function of RAB6KIFL by searching for interacting protein(s).

Through MALDI-TOF analysis coupled with immunoprecipitation experiments, we identified a membrane-associated guanylate kinase protein, DLG5, as a candidate cargo molecule or binding partner of RAB6KIFL. RAB6KIFL possesses a cargo-binding domain within its carboxyl terminal coiled-coil region and is expected to carry molecules along microtubules as a motor kinesin (5). We showed in vivo interaction of RAB6KIFL and DLG5 and also documented partial colocalization at cell-cell contact sites by immunocytochemistry. RAB6KIFL and DLG5 were cotransactivated in some proportion of PDACs (data not shown), indicating that their interaction is likely to play a critical role in pancreatic carcinogenesis. DLG5 contains four PDZ domains (although other membrane-associated guanylate kinases contain one or three), an SH domain, and a GUK domain (15, 16). Membrane-associated guanylate kinases serve as scaffolding molecules for signal-transduction complexes by using multiple interactions among various protein modules to assemble receptors, adaptor proteins, and cytosolic signaling proteins at the cell membrane (15). Recently, DLG5 was shown to colocalize with vinexin and β-catenin, major adherence junction proteins, at sites of cell-cell connection (16). These three proteins could form a ternary complex in which DLG5 would link vinexin-β-catenin to cell-cell contact sites.

In the study reported here, endogenous RAB6KIFL coimmunoprecipitated with DLG5 and β-catenin (data not shown), and those results suggested that RAB6KIFL also forms a complex with DLG5 and β-catenin at sites of cell-cell contact. DLG scaffolding proteins are necessary for the assembly and organization of protein complexes at specialized membrane sites (17–19), and they interact with several proteins involved in the cell cycle and in tumorigenesis (20, 21). For example, DLG1 binds APC, PTEN, and several viral oncoproteins, and has been implicated in regulation of cell proliferation and polarity (22–24). However, how the locations of DLG5 and other DLGs are determined in the physiologic environment remains unclear. We have clearly showed here, by means of a doxycycline-inducible siRNA expression system, that accumulation of exogenous DLG5 at submembrane sites was inhibited when endogenous RAB6KIFL was down-regulated by a specific siRNA. Hence, we suggest that RAB6KIFL may function as part of the intracellular trafficking machinery for DLG5, transporting it to membrane sites in PDAC cells in which DLG5 can interact with other proteins including β-catenin. We also showed that knocking down endogenous DLG5 led to significant suppression of growth of PDAC cells. The data presented here imply to us that trafficking of DLG5 to cell-cell contact sites and its interaction with other molecules as a scaffold protein are highly important for pancreatic carcinogenesis and that RAB6KIFL is likely to play a critical role in the DLG5 trafficking system. Inhibition of this interaction or of membrane trafficking of DLG5 or inhibition of RAB6KIFL itself may provide a promising new approach to molecular therapy for PDACs.

Grant support: Japan Society for the Promotion of Science Research for the Future Program grant 00L01402 (Y. Nakamura).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

We thank Dr. Hao Yu and Ryo Ishimine for their technical assistance; Dr. Kioka at Kyoto University for kindly providing us with anti-sera to hDLG5; and Dr. van de Wetering for kindly providing us with pTER vector.