Oncogene (2012) 31, 4472–4483 & 2012 Macmillan Publishers Limited All rights reserved 0950-9232/12 www.nature.com/onc ORIGINAL ARTICLE DNAJB6 chaperones PP2A mediated dephosphorylation of GSK3b to downregulate b-catenin transcription target, osteopontin

A Mitra1, ME Menezes1, LK Pannell1, MS Mulekar2, RE Honkanen3, LA Shevde1 and RS Samant1

1Department of Oncologic Sciences, Mitchell Institute, University of South Alabama, Mobile, AL, USA; 2Department of Mathematics and Statistics, University of South Alabama, Mobile, AL, USA and 3Department of Biochemistry and Molecular Biology, University of South Alabama, Mobile, AL, USA

Elevated levels of the oncoprotein, osteopontin (OPN), Introduction are associated with poor outcome of several types of including melanoma. We have previously reported The codes for more than 40 distinct an important involvement of DNAJB6, a member of heat- heat-shock 40 (HSP40) (DNAJ) members (Mitra shock protein 40 (HSP40) family, in negatively impacting et al., 2009). A typical DNAJ protein is characterized by tumor growth. The current study was prompted by our a J domain, a glycine-Phenylalanine (G/F) domain and observations reported here which revealed a reciprocal a C-terminal domain. Depending on the presence of relationship between DNAJB6 and OPN in melanoma these domains, this family is divided into Type I, Type II specimens. The ‘J domain’ is the most conserved domain and Type III (also known as DNAJ A, B and C) types of HSP40 family of . Hence, we assessed the (Ohtsuka and Hata, 2000). The J domain consists of functional role of the J domain in activities of DNAJB6. highly conserved Histidine-Proline-Aspartic acid (HPD) We report that the J domain of DNAJB6 is involved in tri-peptide and, is the most conserved domain of this mediating OPN suppression. Deletion of the J domain family and has been implicated in providing the renders DNAJB6 incapable of impeding and substrate specificity to the individual members (Kelley, suppressing OPN. Our mechanistic investigations reveal 1998). The importance of this family in critical that DNAJB6 binds HSPA8 (heat-shock cognate protein, biological activities has been highlighted in the recent HSC70) and causes dephosphorylation of glycogen past (Edwards and Munger, 2004; Tsai et al., 2006; synthase kinase 3b (GSK3b) at Ser 9 by recruiting protein Dhennin-Duthille et al., 2011). Noteworthy among them phosphatase, PP2A. This dephosphorylation activates are the recruitment of , participation in guiding GSK3b, leading to degradation of b-catenin and subse- misfolded proteins to degradation, cell-cycle control and quent loss of TCF/LEF (T cell factor1/lymphoid enhancer regulation of protein kinases (Kelley, 1998; Vos et al., factor1) activity. Deletion of the J domain abrogates 2008; Kampinga and Craig, 2010). DNAJB6, the focus assembly of this multiprotein complex and renders GSK3b of this study, has been reported to reduce protein inactive, thus, stabilizing b-catenin, a transcription co- aggregates in Huntington’s, Parkinson’s diseases, pla- activator for OPN expression. Our in-vitro and in-vivo cental development and neural stem cells (Hunter et al., functional analyses show that silencing OPN expression in 1999; Chuang et al., 2002; Dai et al., 2005; Watson et al., the background of deletion of the J domain renders the 2007, 2009; Zhang et al., 2008; Dey et al., 2009). We resultant tumor cells less malignant despite the presence of have previously demonstrated the role of DNAJB6 in stabilized b-catenin. Thus, we have uncovered a new reducing tumorigenicity and metastatic potential (Mitra mechanism for regulation of GSK3b activity leading to et al., 2008). Our studies have also shown that DNAJB6 inhibition of Wnt/b-catenin signaling. can cause reversal of epithelial-mesenchymal transition Oncogene (2012) 31, 4472–4483; doi:10.1038/onc.2011.623; (EMT) by inhibiting Wnt/b-catenin signaling (Mitra published online 23 January 2012 et al., 2010). Melanoma is a highly invasive and metastatic type of Keywords: DNAJB6; cancer; SPP1; OPN; GSK3b; skin cancer. Cure rates drop precipitously once it has PP2A metastasized. Many distinct signaling mechanisms involving molecules such as b-Raf, AKT, P16 and P53 have been implicated to be involved in melanoma progression (Rodolfo et al., 2004; Lopez-Bergami et al., 2008; Martin et al., 2009). Osteopontin (OPN), is an RGD domain-containing SIBLING family glyco- Correspondence: Professor RS Samant, Department of Oncologic protein that has diverse roles in promoting malignancy Sciences, Mitchell Cancer Institute, University of South Alabama, of a multiple tumor types and is advocated to be 1660 Spring Hill Avenue, Mobile, AL 36604, USA. E-mail: [email protected] involved in melanoma progression and metastasis Received 5 July 2011; revised 28 November 2011; accepted 29 November (Buback et al., 2009; Shevde et al., 2010). We have 2011; published online 23 January 2012 previously shown that OPN expression is increased with DNAJB6 multiprotein complex with GSK3b regulates OPN A Mitra et al 4473 progression and gain of metastatic potential by melano- assessed DNAJB6 and OPN expression from melanoma mas (Riker et al., 2008; Metge et al., 2010). OPN has been specimens and found that OPN expression was sig- implicated to be involved in promoting immune evasion nificantly upregulated in stage III and IV metastatic by tumor cells, in EMT and in inhibiting (Wai melanomas (Figure 1a) and in the same set of samples, and Kuo, 2004; Chakraborty et al., 2006; Alonso et al., DNAJB6 expression was compromised (Figure 1b). Of 2007; Saika et al., 2007; Tuck et al., 2007; Chiodoni et al., those tumor specimens that expressed lower DNAJB6 2010). Elevated levels of OPN are associated with poor than normal controls, about 70% expressed higher OPN outcome in a variety of cancers such as , than normal controls (exact binomial test, P ¼ 0.0032). mesothelioma, colon cancer and ovarian cancer (Furger These observations strongly imply an inverse relation- et al., 2001; Yeatman and Chambers, 2003; Greillier et al., ship between expression of DNAJB6 and OPN in 2008; Shih Ie and Davidson, 2009). In this study, we metastatic melanomas and prompted us to study the assessed DNAJB6 and OPN expression from melanoma mechanism of regulation of OPN by DNAJB6. Inde- specimens and found an inverse relationship between the pendently, estimation of cellular levels of DNAJB6 and transcript levels of these two . Here, we present that OPN from a panel of melanoma cells with increasing the J domain of DNAJB6 has a critical role in mediating malignant potential in comparison with normal human its activity. Specifically, the interaction of DNAJB6 with epidermal melanocytes revealed an inverse pattern of HSPA8 (heat-shock cognate protein, HSC70) is mediated expression and corroborated with observations made via the J domain and is responsible for the protein from melanoma specimens (Supplementary Figure 1). phosphatase 2 (PP2A)-driven dephosphorylation of gly- cogen synthase kinase 3b (GSK3b)atSerine9.This The J domain of DNAJB6 is involved in mediating OPN dephosphorylation activates GSK3b causing downregula- suppression tion of TCF/LEF (T cell factor1/lymphoid enhancer The J domain of HSP40 proteins is critical in mediating factor1) signaling leading to reduced OPN expression. We many of their known functions. To investigate if the J demonstrate that this downregulation is one of the domain of DNAJB6 has a role in regulating OPN, we important mechanisms of DNAJB6-mediated inhibition generated two mutant DNAJB6 cDNA constructs. One of malignancy. capable of coding for DNAJB6 devoid of the J domain (referred to as DJ) and the other that introduced mutations in the most conserved HPD tri-peptide of Results the J domain, changing them to three alanines (AAA; referred to as HPDMUT) (Figure 2, schematic of Loss of DNAJB6 in aggressive melanoma is inversely DNAJB6). Serum-free conditioned media from the related with gain of OPN stable expressors of wild-type DNAJB6 in MDA-MB- Elevated levels of OPN in tumor tissue and serum are 435 cells (435-DNAJB6), the corresponding vector associated with increased invasion and metastasis of control (435-V), 435-DNAJB6-DJ or 435-DNAJB6- melanoma (Denhardt, 2005; Zhou et al., 2005; Buback HPDMUT were assessed for OPN levels. While serum- et al., 2009). We had previously noted a possible free conditioned media from 435-V showed copious negative regulatory effect of DNAJB6 on OPN tran- amounts of OPN, OPN was undetectable in 435-DNAJB6. script levels (Mitra et al., 2008). To directly test the However, 435-DNAJB6-DJ and 435-DNAJB6-HPDMUT clinical relevance of this observation in melanoma, we showed higher levels of OPN, suggesting that the ability

* P < 0.05 * P < 0.01 * P < 0.05 * P < 0.05 * P < 0.01 0.70 0.9 * P < 0.05

0.65 0.8

0.60 0.7 DNAJB6 0.55 0.6 levels of OPN Relative transcript 0.50 0.5 Relative transcript levels of

Normal Normal Stage III Stage IV Stage III Stage IV Stage III A/B Stage III A/B Figure 1 Expression analysis of OPN and DNAJB6 in melanoma specimens. The Tissue Scan Melanoma qPCR Arrays (Origene Technologies) were queried using primer probes for (a) DNAJB6 and (b) OPN. Levels were normalized to human b-actin. The analysis includes 6 normal, 9 stage III, 6 stage IIIA/B and 22 stage IV specimens. *Indicates statistically significant differences.

Oncogene DNAJB6 multiprotein complex with GSK3b regulates OPN A Mitra et al 4474

MDA-MB-435 A375 Schematic of DNAJB6 protein

HPD

DNAJB6 1 326 J OPN (SFM) DNAJB6-ΔJ 70 326 HPD-AAA DNAJB6 DNAJB6-HPDMUT DNAJB6-ΔJ 1 326 β-tubulin

* P < 0.001 * P < 0.01 100 150 * P < 0.001 P < 0.001 P < 0.05 * 75 * * 100 P<0.001 50 expression

expression 50 Percent OPN 25 Percent OPN

0 0 v DNAJB6ΔJ HPDMUT v DNAJB6 ΔJ HPDMUT Figure 2 Deletion of the J domain renders DNAJB6 incapable of suppressing OPN. Schematic of DNAJB6 depicts the DJ and HPDMUT in the perspective of full-length wild-type DNAJB6. Western blot of secreted medium and transcript levels measured by qRT–PCR show that suppression of OPN expression is lost upon J domain mutation. Serum-free conditioned medium (SFM) from equal number of DNAJB6, DNAJB6-DJ and DNAJB6-HPDMUT expressors was harvested and the levels of OPN were compared with that of the vector-only (V) transfected control cells for (a) MDA-MB-435 and (b) A375 cells. Expression of DNAJB6, DNAJB6-DJ and DNAJB6-HPDMUT was confirmed by western blot analysis of 30 mg of total protein. b-Tubulin was used as loading control. mRNA levels of OPN were evaluated by qRT–PCR for vector, DNAJB6, DNAJB6-DJ and DNAJB6-HPDMUT expressors of (c) MDA-MB- 435 and (d) A375 cells. *Indicates statistically significant differences.

of DNAJB6 to suppress OPN is significantly compro- evaluated the impact of deletion of the J domain on the mised upon deletion of the J domain or the HPD-AAA ability of DNAJB6 to regulate the mesenchymal mutation (Figure 2a), both of which compromise the phenotype. Expression of DNAJB6 in mesenchymal- activity of J domain. Similar results are observed in like MDA-MB-435 cells changed the appearance of A375 human melanoma cells (Figure 2b). these cells in 2D culture from spindle (lanceolate)- Quantitative RT–PCR (qRT–PCR) analysis of OPN shaped cells to cells with more compact, cobblestone- transcript from the 435 cell lines revealed that there was like morphology. However, 435-DNAJB6-DJ cells and almost no detectable OPN transcript in the DNAJB6 435-DNAJB6-HPDMUT cells did not show any notice- expressors. Abrogating the J domain restored the OPN able change in 2D morphology (Figure 3a). transcript level to about 45% of the vector control The same sets of cells were evaluated for morphology (Figure 2c). In the A375 cells, there was about 75% in three-dimensional (3D) growth. Cells that have reduction in the OPN transcript levels in DNAJB6 undergone EMT are known to exhibit an invasive expressers; this was relieved by J domain deletion and phenotype in 3D matrix. These cells show invadopodia HPD mutation (Figure 2d). Furthermore, the evaluation and invade away from the central spherical growth. of activity of OPN promoter using luciferase reporter The vector-only expressing cells and 435-DNAJB6-DJ assay showed 50% suppression when co-transfected cells and 435-DNAJB6-HPDMUT showed highly inva- with DNAJB6 compared with the empty vector control. sive outgrowths in 3D culture. However, the 435- However, DNAJB6-DJ was unable to suppress OPN DNAJB6 cells maintained a compact spherical growth promoter activity (Supplementary Figure 2). These pattern resembling a more epithelial-like behavior results indicate the critical involvement of the J domain (Figure 3a). The observations made for A375 melano- of DNAJB6 in mediating suppression of OPN. ma cells and their DNAJB6 and DJ expressors for 3D growth corroborate these results (Supplementary The J domain is involved in mediating reversal of EMT by Figure 3a). DNAJB6 Based on these clear morphological differences, we OPN is a pro-malignancy factor and we found that the J evaluated mRNA expression of (KRT18) as domain is largely responsible for regulating OPN; hence, an epithelial marker and vimentin as a mesenchymal we speculated that the J domain of DNAJB6 may have a marker. These markers did not show noticeable changes; critical role in mediating its anti-malignant functions. the levels of mesenchymal transcription factor SLUG The phenomenon of EMT is closely linked with invasion also remained unaltered. However, there was a con- and metastasis. Since one of the known functions of sistent, dramatic upregulation (Po0.001) in the expres- DNAJB6 is to suppress the mesenchymal phenotype, we sion of mesenchymal transcription factor, TWIST in

Oncogene DNAJB6 multiprotein complex with GSK3b regulates OPN A Mitra et al 4475 2-D 3-D *P<0.001 5000 ΔJ HPDMUT 10 V

DNAJB6 + rec-OPN

KRT18Vimentin SLUG TWIST DNAJB6 expressor Fold change relative to 0.1 DNAJB6

J+OPNi J+OPNi + rec-OPN J  MUT HPD

Figure 3 Deletion of J domain abrogates capacity of DNAJB6 to promote epithelial phenotype. (a) Bright field photomicrographs of vector, DNAJB6, DNAJB6-DJ and DNAJB6-HPDMUT expressors of MDA-MB-435 were analyzed for their cell shape in adherent culture on tissue culture plastic (2D) (column 1) or 3D morphology in 3D Culture Matrix Basement Membrane Extract (3D) (column 2). Arrows indicate invasive growth in the 3D matrix. Simultaneously, 3D morphology of DNAJB6 expressors cultured in media supplemented with 100 ng/ml human recombinant OPN was evaluated (column 3, row 2). 3D growth of DNAJB6-DJ cells silenced for OPN expression (DNAJB6-DJ-OPNi) was compared with the same cells cultured in media supplemented with 100 ng/ml human recombinant OPN (columns 3 and 4, row 3). Experiments were performed in triplicate and repeated once. Scale bars correspond to 100 mM.(b) DNAJB6, DNAJB6-DJ and DNAJB6-HPDMUT expressors of MDA-MB-435 were analyzed for mRNA levels for epithelial marker KRT18, mesenchymal markers vimentin, SLUG and TWIST using qRT–PCR. The readings show the fold change in expression with respect to the DNAJB6 expressors. The Y axis is in log10 scale. *Indicates statistically significant differences.

435-DNAJB6-DJ cells and in 435-DNAJB6-HPDMUT colony formation assay) of 435-DNAJB6-DJ and 435- cells (Figure 3b). DNAJB6 with 435-V. While 435-DNAJB6 showed To test if the mesenchymal-like phenotype was due to about 40% suppression of colony formation, this the inability to suppress OPN, we stably silenced OPN attribute was lost upon deletion of the J domain from 435-DNAJB6-DJ cells using shRNA NM-636 (Figure 4a). However, DNAJB6-DJ-OPNi cells showed (Shevde et al., 2006) to generate DNAJB6-DJ-OPNi suppression of colony formation comparable to 435- cells. Unlike the highly invasive nature of 435-DNAJB6- DNAJB6 transfected with a scrambled control, further DJ observed in 3D culture, the DNAJB6-DJ-OPNi cells attesting the importance of OPN regulation downstream showed a compact spherical growth pattern resembling of DNAJB6 (Supplementary Figure 4a). a more epithelial-like behavior (Figure 3a), with We tested the in-vivo attributes of malignancy by noticeable downregulation of TWIST transcript and evaluating the growth of these cells as xenografts in a concomitant upregulation of KRT18 transcript athymic mice. 435-V, 435-DNAJB6-DJ and 435- (Supplementary Figure 3b). The involvement of OPN DNAJB6-HPDMUT show a rapid tumor take rate. All was further confirmed by our observation that the animals in these groups showed tumors at day 7 and a DNAJB6-DJ-OPNi cells show highly invasive growth in rapid tumor growth thereafter. The 435-DNAJB6 3D upon addition of exogenous OPN, which phenoco- tumors were slow to pick up and exhibited limited pies the change in 3D morphology observed for 435- growth (Po0.001) (Figure 4b); this strongly supports DNAJB6 upon exogenous addition of OPN (Figure 3a). our contention that the J domain of DNAJB6 is Hence, we further evaluated the role of the J domain important for its negative impact on tumor growth. in regulation of malignant properties by DNAJB6 by a EMT is implied to have a role in affecting invasion series of in-vitro and in-vivo assays. and metastasis. Figure 4c shows that the efficiency of formation of lung metastasis from 435-DNAJB6 The J domain is critical for restriction of attributes of was significantly lower compared with that of 435-V, malignancy by DNAJB6 435-DNAJB6-DJ and 435-DNAJB6-HPDMUT, further To assess the contribution of the J domain on the ability confirming the critical role of the J domain in the of DNAJB6 to restrict malignant behavior in vitro,we anti-malignant activity of DNAJB6. Comparison of compared the anchorage-independent growth (soft agar the 435-V with 435-DNAJB6-DJ or the 435-DNAJB6-

Oncogene DNAJB6 multiprotein complex with GSK3b regulates OPN A Mitra et al 4476 * P < 0.001 10.0 V DNAJB6 * P < 0.01 Δ 120 7.5 J HPDMUT 90 5.0

60 +/- S.E.M.

SEM 2.5

30 Mean Tumor Diameter

% of colonies +/- 0.0 0 0 5 10 15 20 25 30 V DNAJB6 ΔJ Days elapsed

* P< 0.05

30 * P< 0.005

20 * P< 0.005 SEM 10

No. of Metastases +/- 0 V DNAJB6 ΔJ HPDMUT Figure 4 J domain of DNAJB6 has a role in negatively influencing malignant attributes. (a) Ability of anchorage-independent growth of DNAJB6, DNAJB6-DJ expressing MDA-MB-435 cells was compared with the empty vector-only transfectants using soft agar colony formation assay. Experiment was performed in duplicate and repeated once. The error bars represent s.e.m. (b) In-vivo tumor growth ability of the DNAJB6, DNAJB6-DJ and DNAJB6-HPDMUT expressors of MDA-MB-435 was compared with the empty vector-only control using athymic mice. Tumor growth was measured as mean tumor diameter. Each experimental group had eight mice and repeated once. Error bars represent s.e.m. Significant differences were noted in the groups V vs DNAJB6 (Po0.001), DNAJB6 vs DJ(Po0.001) and DNAJB6 vs HPDMUT (Po 0.001). (c) To evaluate metastatic ability, primary tumors were surgically removed by survival surgery under anesthesia. After 7 weeks, mice were euthanized and lung metastases were counted. *Indicates statistically significant differences.

HPDMUT cells shows that there is no statistically were unaltered in A375-DNAJB6-DJ relative to the significant difference in the metastasis. When similarly vector control cells (Figure 5c). Correspondingly, unlike tested, the DNAJB6-DJ-OPNi cells showed suppressed the DNAJB6 expressors, A375-DNAJB6-DJ lacked tumor growth and pulmonary metastasis compared with the ability to suppress the TCF/LEF transcription the scrambled control (Supplementary Figures 4b–d). (Figure 5d). These compelling results prompted us to explore the It is well documented that phosphorylation of b- molecular mechanism of J domain-dependent regulation catenin by GSK3b targets it for degradation (Behrens of OPN. OPN has been reported to be a transcription et al., 1998; Hinoi et al., 2000). However, phosphoryla- target of b-catenin (TCF/LEF) transcription (El-Tanani tion of GSK3b itself at Ser 9 residue (P-Ser 9-GSK3b) et al., 2001). Hence, we evaluated effects of the J domain dictates its activity; the dephosphorylated form is the deletion on Wnt/b-catenin signaling. active form responsible for phosphorylation of b-catenin (Miller and Moon, 1996; Orford et al., 1997). Western Deletion of J domain upregulates TCF/LEF activity and blot analysis showed that P-GSK3b-Ser 9 levels were renders DNAJB6 incapable of dephosphorylating GSK3b undetectable in DNAJB6 expressors of 435 and A375 We assessed the levels of b-catenin in 435-DNAJB6 and relative to the vector. However, DNAJB6-DJ expressors 435-DNAJB6-DJ. We found that 435-DNAJB6 did not showed P-GSK3b-Ser 9 levels, which are comparable to show detectable levels of b-catenin. However, 435- the respective vector controls. Levels of total GSK3b DNAJB6-DJ showed levels of b-catenin comparable to were unaltered (Figures 5a and b). These observations 435-V (Figure 5a). DNAJB6 significantly downregulates indicate that DNAJB6 expression causes dephosphor- the b-catenin-dependent signaling through TCF/LEF ylation of Ser 9 of GSK3b, and the J domain is essential transcription factors (Mitra et al., 2010). To test if the J for mediating this activity. domain has a role in this activity, we assessed the luciferase reporter activity of the TOPFlash reporter DNAJB6 interacts with HSPA8 (consisting of seven tandem TCF/LEF binding sites) in To get more insight into the mechanism by which 435-DNAJB6, 435-DNAJB6-DJ and 435-V. The results DNAJB6 regulates the phosphorylation status of showed that 435-DNAJB6 suppressed the TCF/LEF GSK3b, we performed a mass-spectrometric analysis activity by about 50%; however, DNAJB6-DJ failed of the co-immunoprecipitated proteins from 435- to exhibit such suppression (Figure 5b). Similarly in DNAJB6 expressing cells. The results revealed HSPA8 A375, DNAJB6 expressors showed very low levels (HSC70) as an interactor of DNAJB6 and it was absent of b-catenin, in contrast to levels of b-catenin that in co-immunoprecipitate from DNAJB6-DJ expressing

Oncogene DNAJB6 multiprotein complex with GSK3b regulates OPN A Mitra et al 4477 V DNAJB6 J V DNAJB6 J

-catenin

P-GSK3-ser9

GSK3

-actin

MDA-MB-435 A375

* P < 0.001 * P < 0.01 150 * P < 0.05 200 * P < 0.05 150 100 100 50 50 Percent activity Percent activity

0 0 V DNAJB6 ΔJ V DNAJB6 ΔJ Figure 5 J domain is necessary for the negative impact of DNAJB6 on Wnt/b-catenin signaling. Expressors for DNAJB6 and DNAJB6-DJ were probed for levels of GSK3b, b-catenin and P-GSK3b-Ser 9 western blot analyses for (a) MDA-MB-435 and (b) A375. The levels were compared with respective vector-only controls. b-Actin was used as a loading control. TCF/LEF-driven transcription activity was measured as an indicator of Wnt/b-catenin signaling using TOPFlash luciferase reporter assay in DNAJB6, DNAJB6-DJ expressors of (c) MDA-MB-435 and (d) A375. The assay was performed in triplicate and repeated once. The luciferase activity readings were normalized with activity from a co-transfected b-gal expressing control. The error bars represent s.e.m. *Indicates statistically significant differences. cells (Supplementary Figures 5a and b). To confirm this dently from the total protein of DNAJB6 expressors. result, we transiently co-expressed cDNAs of HSPA8 As seen in Figure 6d, both PP2A and GSK3b could with HA epitope-tagged DNAJB6 and DNAJB6-DJin co-immunoprecipitate HSPA8. Independently, we also COS7 cells. Immunoprecipitation was performed using verified the interaction of GSK3b with DNAJB6 anti-HA antibody and was analyzed by western blot using mammalian two-hybrid system (Figure 6b(II)). for the presence of HSPA8. The results demonstrated DNAJB6 interacted with GSK3b resulting in twofold that DNAJB6 was able to co-immunoprecipitate increase in the pGL4.31-luciferase reporter activity. HSPA8; however, DNAJB6-DJ failed to bind HSPA8 Similarly, HSPA8 interacted with GSK3b resulting in (Figure 6a). little over twofold increase in the reporter activity The DNAJB6–HSPA8 interaction was also confirmed (Figure 6b(II)). Additionally, silencing DNAJB6 expres- using mammalian two-hybrid system. DNAJB6 inter- sion from MCC013 cells resulted in elevated levels of acted with HSPA8 resulting in about sixfold activation P-GSK3b-Ser 9, which also reflected in elevated levels of in the pGL4.31-luciferase reporter activity; however, b-catenin transcription target OPN (Supplementary DNAJB6-DJ failed to interact with HSPA8 as seen by Figure 6b). Taken together, the results of the protein lack of activation of the reporter (Figure 6b(I)). interaction studies indicate that DNAJB6 and HSPA8 HSPA8 has been implicated to interact with PP2A complex chaperones interaction of PP2A and GSK3b. (Polanowska-Grabowska et al., 1997; Whalen et al., 2005). Hence, we hypothesized that DNAJB6–HSPA8- Okadaic acid treatment renders DNAJB6 incapable of PP2A form a molecular complex that binds to and dephosphorylating GSK3b dephosphorylates GSK3b at Ser 9. We used size Protein phosphatase 2 is a multimeric enzyme with exclusion chromatography to separate multiprotein structural subunit A, catalytic subunit B and regulatory complexes. Western blot analysis of the eluted fractions subunit C. Each of the subunits is coded by multiple revealed co-elution of DNAJB6–HSPA8-PP2A and genes and hence there could be several PP2 holoenzymes GSK3b in fractions 13–15, which corresponded to (Sablina and Hahn, 2007; Eichhorn et al., 2009). approximate molecular weight range of 700–1000 kDa Okadaic acid is a known inhibitor of protein phospha- (Figure 6c). Similarly, co-elution of DNAJB6, HSPA8, tase with a relatively high specificity toward PP2 GSK3b and PP2A was detected in the melanoma cell (Fernandez et al., 2002). To test if the reduction in P- line MCC013 (derived from primary tumor, which still GSK3b-Ser 9 and subsequent degradation of b-catenin retains noticeable DNAJB6 expression; Supplementary upon DNAJB6 expression was dependent on the Figure 6a). phosphatase activity of PP2A, DNAJB6 expressing cells To confirm the interactions further, immunoprecipi- were treated with 20 nM okadaic acid (concentration tation of PP2A and GSK3b was performed indepen- relatively specific to PP2A) for 8 h duration. As seen in

Oncogene DNAJB6 multiprotein complex with GSK3b regulates OPN A Mitra et al 4478 * P < 0.001 * P < 0.001 HSPA8 750 IP: HA (DNAJB6 & DNAJB6ΔJ) (I) 500 IB: HSPA8 250

IB: DNAJB6 % activation 0 V DNAJB6 J IB: DNAJB6-J * P < 0.001  300 GSK3 * P < 0.01 ~ 700 K.Da (II) 200 Mol. wt 100

Input 3 4 567891011121314151617181920 % activation 0 DNAJB6 V DNAJB6 HSPA8

PP2A

HSPA8 I.P. PP2AGSK3 C Lysate IB: HSPA8 GSK3B IB: PP2A

IB: GSK3

Figure 6 DNAJB6 forms a multiprotein complex with HSPA8 and GSK3b.(a) Binding of DNAJB6 and DNAJB6-DJ with HSPA8 was determined by transient expression in COS7 cells. HA-epitope tag was used for the immunoprecipitation of DNAJB6 and DNAJB6-DJ. The precipitate was analyzed by western blot for the presence of HSPA8. (b) Mammalian two-hybrid assay was performed for verifying interactions between (I) HSPA8 and DNAJB6 and DNAJB6-DJ (II) GSK3b with DNAJB6 and HSPA8. Experiments were performed in quadruplicate and repeated once. The error bars represent s.e.m. *Indicates statistically significant differences. (c) Fast protein liquid chromatography separation of 3 mg total protein from DNAJB6 expressors of MDA-MB-435 was performed by size exclusion chromatography on a pre-calibrated Superose 6 HR 10/30 column. Eluted fractions were analyzed by western blotting for the presence of DNAJB6, PP2A, HSPA8 and GSK3b.(d) Immunoprecipitate of PP2A and GSK3b from DNAJB6 expressing MDA-MB-435 is analyzed for the presence of HSPA8. ‘C’ represents isogenic antibody þ beads control.

Figure 7a, okadaic acid treatment resulted in increased cellular homeostasis. Aberrant activation of Wnt/b- expression of P-GSK3b-Ser 9 and b-catenin. catenin signaling can be oncogenic (Lucero et al., 2010). Real-time PCR done with cDNA generated from Multiple reports document deregulation of the Wnt okadaic acid-treated 435-DNAJB6 cells showed in- pathway in melanoma clinical specimens (Pecina-Slaus creased levels of OPN transcript but no change in et al., 2007; Zuidervaart et al., 2007; Moore et al., 2008). b-catenin transcript (Figure 7b). Cumulatively, these The Wnt pathway is intricately regulated by a complex observations suggest that okadaic acid inhibition of PP2 interplay of a variety of Wnt ligands, receptors and activity led to increased P-Ser 9-GSK3b levels rendering secreted inhibitors leading to modulation of intracellular it inactive. This increased the stability of b-catenin, signaling components (Weeraratna, 2005). Wnt ligands which was reflected in the elevated b-catenin protein such as Wnt1 and Wnt3A signal via b-catenin to promote levels and is evident by a concomitant increase in OPN melanocyte differentiation and tumor development transcript levels. Additionally, silencing PP2A using (O’Connell and Weeraratna, 2009). Critical protein siRNA in 435-DNAJB6 cells showed increased P- members of the Wnt/b-catenin pathway, such as adeno- GSK3b-Ser 9 with concomitant elevation in b-catenin matous polyposis coli, ICAT, LEF1 and b-catenin are protein levels as detected by western blot analysis. The modified in melanomas leading to activation of this involvement of PP2A was also more evident when signaling (Larue and Delmas, 2006). However, tumor siRNA silencing of PP1 (protein phosphatase 1) from microenvironment alterations of secreted inhibitors of the same cells failed to show any increase in b-catenin Wnt pathway and complex interplay of canonical, non- protein (Figure 7c). A concomitant upregulation in canonical Wnt signaling results in context-dependent TCF/LEF transcription in the PP2A silenced cells was effects of Wnt pathway in cancers such as melanoma evident as a twofold increase in the TOPFlash reporter (Lucero et al., 2010; Menezes et al., 2011). activity (Figure 7d). Cumulatively, these observations The role of HSP40 family members in regulating a confirm the involvement of PP2A in removal of the variety of cellular signaling is emerging. Notably, tumor phosphate from P-GSK3b-Ser 9 and converting it to the suppressor hTid1 (DNAJA3) has been reported to be active form. associated with adenomatous polyposis coli, together with a complex of HSP70, HSC70, Actin, Dvl and Axin in normal colon epithelium. This complex is not Discussion associated with adenomatous polyposis coli’s involve- ment in b-catenin degradation; however, it is implicated Wnt/b-catenin signaling is a critical and highly regulated in maintenance of the polarity of colonic epithelial cells developmental pathway, which is equally important for (Kurzik-Dumke and Czaja, 2007; Kurzik-Dumke et al.,

Oncogene DNAJB6 multiprotein complex with GSK3b regulates OPN A Mitra et al 4479

10

1 OPN 0.1 P-GSK3β-Ser9 0.01 GSK3β 0.001 β-catenin

Fold change in 0.0001 Transcript Abundance * P<0.005 DNAJB6 0.00001 β-actin V V+OA DNAJB6

DNAJB6+OA 435-DNAJB6

I II Control si PP2AControl si PP1 P<0.05 250 * PP2A PP1 200 P-GSK3 P-GSK3 ser9 ser9 150

GSK3 GSK3 100

50

-catenin -catenin Percent Activation 0 Control siPP2A -actin -actin

Figure 7 Activity of PP2A is necessary for regulation of Wnt/b-catenin by DNAJB6. (a) DNAJB6 expressors of MDA-MB-435 and corresponding vector-only control were independently treated with 20 nM okadaic acid (OA). Western blot analysis was performed from 40 mg of total protein to determine the levels of P-GSK3b-Ser 9 and b-catenin. b-Actin levels were used to ensure equal loading. (b) cDNA generated from 20 nM OA treated 435-DNAJB6 and corresponding vector control cells was compared with vehicle (DMSO)- treated controls for transcript levels of OPN using qRT–PCR. Glyceraldehyde 3-phosphate dehydrogenase was used as normalization control. Experiment was performed in duplicate and repeated once. The error bars represent s.e.m. The bars represented are in comparison with the untreated vector control levels set at 1 for ease of comparison. The Y axis is in log10 scale. *Indicates statistically significant differences. (c) I: PP2A expression was silenced from 435-DNAJB6 expressors using siRNA transfection. Total protein (30 mg) was resolved using SDS–PAGE and probed for levels of PP2A, b-catenin and P-GSK3b-Ser 9. Levels of b-actin were determined to confirm comparable loading. II: PP1 expression was silenced from 435-DNAJB6 expressors using siRNA transfection. Total protein (30 mg) was resolved using SDS––PAGE and probed for levels of PP1, b-catenin. Levels of b-actin were determined to confirm comparable loading. (d) TOPFlash reporter activity was evaluated in 435-DNAJB6 expressors silenced for PP2A expression in comparison with the control siRNA. Experiment was performed in triplicate and repeated once. The error bars represent s.e.m. *Indicates statistically significant differences.

2008). Htid1 is a Type I HSP40 protein whereas tively expressed member of HSP/HSC70 family. We also DNAJB6 belongs to another distinct type (Type II). demonstrate that the Ser 9 phosphorylation status of The core J domain of the HSP40 family members has GSK3b is dependent on formation of this complex. similarity but each of the members perform a distinct GSK3b is a highly regulated kinase that regulates a function (Mitra et al., 2009). In this work, we have critical signaling node of Wnt, Hedgehog and SNAIL demonstrated that DNAJB6 significantly downregulates signaling (Zhou and Hung, 2005; Wang et al., 2011). expression of the secreted oncoprotein, OPN. We also Our work shows that DNAJB6 expression keeps GSK3b show that this activity of DNAJB6 is mediated through in an activated state by mediating Ser 9 dephosphoryla- its J domain. Though J domains, with conserved HPD tion. DNAJB6 performs this task through its J domain. amino acid trio, are conserved throughout the HSP40 We also see that the J domain is involved in recruiting family of proteins the actual roles played by these HSPA8. Hence, it is tempting to propose that by domains for the respective HSP40 vary significantly and binding to HSPA8, DNAJB6 recruits PP2A and GSK3b are dependent on the sequence that is flanking the HPD to facilitate Ser 9 dephosphorylation of GSK3b. motif (Hennessy et al., 2005). These domains bind to the OPN has been suggested to be a target of active Wnt/ specific member(s) of DNAK (HSP/HSC70) family of b-catenin signaling (El-Tanani et al., 2006). The role of proteins and this association decides recruitment of OPN in modulation of the malignant properties of appropriate client proteins for a variety of downstream cancer cells is well established (Furger et al., 2001). processing or for targeting proteins to destruction (Fan Results reported here by us and reports by others et al., 2003; Kampinga and Craig, 2010). Thus, the document that OPN expression is significantly elevated DNAJ family of proteins has a critical role in regulation in melanomas (Riker et al., 2008; Das et al., 2009; Metge of several critical cellular processes in association with et al., 2010). Our nude mice xenograft studies demon- HSP70 family of proteins (Qiu et al., 2006). Our data strate that suppression of OPN is responsible for the show that DNAJB6 interacts with HSPA8, a constitu- reduced malignant attributes of DNAJB6 expressors.

Oncogene DNAJB6 multiprotein complex with GSK3b regulates OPN A Mitra et al 4480 The inverse relationship of DNAJB6 and OPN expres- Materials and methods sion attests the relevance and importance of the described signaling in melanoma progression. Our Plasmids and constructs molecular signaling observations uncover the mechan- Details of DNAJB6 expressors are described previously (Mitra ism behind this inverse relationship. We observed et al., 2008). DNAJB6-DJ was generated by deletion of the J downregulation of DNAJB6 expression accompanying domain (amino acids 2–70) using inverse PCR. DNAJB6- HPDMUT was generated by specifically mutating each of the melanoma progression. DNAJB6 expression compro- HPD amino acids (#31–33) to alanine. TOPFlash is a TCF- mises stability of b-catenin and hence negatively LEF transcription activity luciferase reporter, which was regulates Wnt/b-catenin signaling. The OPN promoter generously gifted by Dr Randall Moon (Veeman et al., consists of many regulatory elements, among which the 2003). Full-length DNA constructs for GSK3b and HSPA8 TCF4/LEF1 has been reported to be binding partners of (Plasmid #14 753 and 19 541) were obtained from Addgene b-catenin. Our results using the TOPFlash reporter (Addgene Inc., Cambridge, MA, USA). DNAJB6, DNAJB6- consisting of tandem TCF/LEF sites in a reporter vector DJ, DNAJB6-HPDMUT and HSPA8 were PCR amplified and along with full-length DNAJB6 or DNAJB6-DJ indicate cloned into pACT (XbaI/KpnI). GSK3b (PCR amplified) and the role of the J domain in repressing the TCF/LEF HSPA8 were cloned into pBIND (XbaI/KpnI). reporter activity. Details of cell culture and collection of serum-free media, immunoblotting, immunoprecipitation and soft agar coloniza- Mesenchymal phenotype has been implied as one tion are described in Supplementary Material. of the major determinant of melanoma metastasis (Lin et al., 2010). Based on a high-throughput study of 34 vertical growth phase melanoma patients with 36 Mass spectrometry months follow-up, EMT-related expression, in- The immunoprecipitates were run on an SDS–PAGE and stained with Coomassie blue (Bio-Rad, Hercules, CA, USA). cluding OPN has been reported to be significantly The differentially stained bands were excised using a razor associated with metastasis development (Alonso et al., blade and the gel pieces were treated with trypsin. Electrospray 2007). Similarly, TWIST expression in melanoma cell tandem mass spectrometry was performed with a Q-TOF line has been suggested to have a role in EMT (Kushiro Ultima API mass spectrometer (Waters, Milford, MA, USA) and Nunez, 2011). Several reports have implicated equipped with a nanoflow electrospray. The resulting data files TWIST to be downstream target of Wnt signaling were searched using an in-house MASCOT search engine (Howe et al., 2003; Reinhold et al., 2006). Our study has (version 2.1.03; Matrix Science Ltd, London, UK). Ion scores not directly explored the relationship between OPN >35 (Po0.05) were considered significant. Only proteins and TWIST. However based on our results, it is matching at least two peptides in MASCOT were accepted. reasonable to posit that both OPN and TWIST, proteins that contribute to mesenchymal-like phenotype, are RNA isolation and real-time qRT–PCR downstream negatively regulated targets of DNAJB6 TRIzol reagent (Invitrogen, Carlsbad, CA, USA) was used to regulated Wnt/b-catenin signaling. isolate total RNA from cultured cells. RNA was treated with In summary, we propose that GSK3b is a client of DNase I (Promega Corp., Madison, WI, USA). cDNA DNAJB6–HSPA8 complex. This complex recruits PP2 synthesis was carried out using a cDNA synthesis kit (Applied Biosystems Inc., Foster City, CA, USA) with 1 mg total RNA activity and dephosphorylates GSK3b at Ser 9. The as the template and random primers. Real-time qRT–PCR specificity and activity of PP2A are highly regulated analysis was performed on the experimental mRNAs. The and accumulating evidence indicates that PP2A acts as a PCR primers and probes for DNAJB6, c-Myc, SPP1(OPN), tumor suppressor (Eichhorn et al., 2009) due to its Twist, SLUG, KRT18, Vimentin and endorse control gene involvement in regulation of a number of major glyceraldehyde 3-phosphate dehydrogenase were purchased signaling pathways, including Wnt, PI3K/Akt and from Applied Biosystems Inc. qRT–PCR was performed using c-Myc. However, direct substrates and intricate mechan- Bio-Rad iQ5 Real-Time Detection System (Bio-Rad). The istic details involved in modulation of these signaling DCT values of mRNAs from each sample were pathways have yet to be identified for the specific PP2A calculated by normalizing with internal control, glyceralde- complexes (Sablina et al., 2010). The regulatory subunit hyde 3-phosphate dehydrogenase and relative quantization values were plotted using Graphpad Prism (San Diego, CA, of PP2A, B56, has been shown to interact with USA). Real-time qRT–PCR analysis was performed on the adenomatous polyposis coli and is implicated in redu- experimental mRNAs in triplicate, and the experiment was cing the abundance of b-catenin and in inhibition of repeated once from an independent passage to confirm the transcription of b-catenin target genes (Seeling et al., findings. 1999). Independently, in Drosophila, it has been shown that the regulatory subunit of PP2A, PR55-a, controls Melanoma specimen study b-catenin phosphorylation, degradation and hence the The Tissue Scan Melanoma qPCR Array containing cDNAs Wnt/b-catenin signaling (Zhang et al., 2009). Our from normal skin, stage III and stage IV melanomas was findings reinforce the overall contribution of PP2A in obtained from Origene Technologies (Rockville, MD, USA) downregulating the Wnt/b-catenin signaling. However, and assessed for OPN and DNAJB6 transcript levels by we highlight a key role of the DNAJB6–HSPA8 qRT–PCR. Data were normalized to b-actin. complex in PP2A-dependent dephosphorylation of GSK3b, and thus define a novel mechanism for depho- 3D culture sphorylation of GSK3b and consequent regulation of 3D cultures were grown following the protocol by Debnath Wnt signaling. et al. (2003). Briefly, eight well-chambered coverglass slides

Oncogene DNAJB6 multiprotein complex with GSK3b regulates OPN A Mitra et al 4481 (Thermo Scientific, Waltham, MA, USA) were placed on ice Tumor growth and metastasis and coated with 50 ml of 3D Culture Matrix Basement Cells (107/ml) were injected into exposed axillary mammary fat Membrane Extract Reduced Growth Factor (phenol red free) pads of 6-week-old, female athymic mice (Harlan-Sprague- from Trevigen (Gaithersburg, MD, USA). The slide was Dawley, Indianapolis, IN, USA). Tumor size was measured incubated for 30 min in 37 1C incubator. Cells (5000/well) were weekly and mean tumor diameter was calculated as previously plated in growth media containing 2% reduced growth factor reported (Fillmore et al., 2009). For monitoring metastatic basement membrane extract (Trevigen). Human recombinant spread, tumors were surgically removed under aseptic condi- OPN (100 ng/ml) (R&D Systems, Minneapolis, MN, USA) was tions under anesthesia. After a period of about 6 weeks, mice supplemented for DJ þ OPNi cells. Media was changed every 4 were euthanized; lungs were removed, rinsed with phosphate- days and the slide was viewed and photographed using a Nikon buffered saline and fixed in diluted Bouin’s solution (20% microscope (Nikon, Tokyo, Japan) using the  20 objective. Bouin’s fixative in neutral buffered formalin) before quantifi- cation of surface metastasis. Eight mice were used per group Size exclusion chromatography (fast protein liquid and each experiment was repeated once. Animals were chromatography (FPLC)) maintained under the guidelines of the National Institutes of Cells grown as monolayers were lysed in NP-40 lysis buffer. Health and Institutional Animal Care and Use Committee of The clear lysate (1 ml) was separated using Superose 6 HR 10/ the University of South Alabama, Mobile. Food and water 30 size exclusion column (Amersham Biosciences, Piscataway, were provided ad libitum. NJ, USA) as described previously (Clark et al., 2008). Phosphate-buffered saline was used for elution, at a flow rate Statistics of 0.2 ml/min. Fractions (1 ml) were collected and analyzed by Statistical analyses were performed using GraphPad Prism 4 immunoblotting. (San Diego, CA, USA). Student’s t-test, Mann–Whitney, one- way analysis of variance or two-way analysis of variance was Mammalian two hybrid applied based on the statistical mandates or suggestions of CheckMate/Flexi Vector System (Promega Corp.) was used to each analysis. study protein–protein interaction. In brief, COS7 cells were maintained in 5% FBS-DMEM, seeded into 96-well plates 18 h before transfection. The various , pACT-DNAJB6, pACT-DNAJB6-DJ, pACT-HSPA8, pBIND-HSPA8, Conflict of interest pBIND-GSK3b or pBIND-vector and pGL4.31-luciferase reporter were combined in the molar ratio (1:1:2). Lipofecta- The authors declare no conflict of interest. mine 2000 (Invitrogen) was used as transfection reagent as per the manufacturer’s recommendations.

Luciferase assay Acknowledgements Cells were transfected using Lipofectamine 2000 (Invitrogen) as per the manufacturer’s instructions. Total protein was This study was supported by USPHS grants CA140472 (RSS) harvested and luciferase activity measured using a Turner 20/ and CA138850 (LAS). RSS is the recipient of the Mayer 20 luminometer (Turner Biosystems, Sunnyvale, CA, USA). Mitchell Award for Excellence in Cancer Research and The luciferase reading was normalized to the total protein acknowledges the support. We thank Dr Randall Moon concentration. Data are expressed as relative luciferase (HHMI University of Washington) for providing the TOP- activity. Flash reporter.

References

Alonso SR, Tracey L, Ortiz P, Perez-Gomez B, Palacios J, Pollan M Clark DW, Mitra A, Fillmore RA, Jiang WG, Samant RS, Fodstad O et al. (2007). A high-throughput study in melanoma identifies et al. (2008). NUPR1 interacts with p53, transcriptionally regulates epithelial-mesenchymal transition as a major determinant of p21 and rescues breast epithelial cells from -induced metastasis. Cancer Res 67: 3450–3460. genotoxic stress. Curr Cancer Drug Targets 8: 421–430. Behrens J, Jerchow BA, Wurtele M, Grimm J, Asbrand C, Wirtz R Dai YS, Xu J, Molkentin JD. (2005). The DnaJ-related factor Mrj et al. (1998). Functional interaction of an axin homolog, conductin, interacts with nuclear factor of activated T cells c3 and mediates with beta-catenin, APC, and GSK3beta. Science 280: 596–599. transcriptional repression through class II histone deacetylase Buback F, Renkl AC, Schulz G, Weiss JM. (2009). Osteopontin and recruitment. Mol Cell Biol 25: 9936–9948. the skin: multiple emerging roles in cutaneous biology and Das S, Harris LG, Metge BJ, Liu S, Riker AI, Samant RS et al. (2009). . Exp Dermatol 18: 750–759. The hedgehog pathway transcription factor GLI1 promotes Chakraborty G, Jain S, Behera R, Ahmed M, Sharma P, Kumar V malignant behavior of cancer cells by up-regulating osteopontin. et al. (2006). The multifaceted roles of osteopontin in cell J Biol Chem 284: 22888–22897. signaling, tumor progression and angiogenesis. Curr Mol Med 6: Debnath J, Muthuswamy SK, Brugge JS. (2003). Morphogenesis 819–830. and oncogenesis of MCF-10A mammary epithelial acini grown Chiodoni C, Colombo MP, Sangaletti S. (2010). Matricellular in three-dimensional basement membrane cultures. Methods 30: proteins: from homeostasis to inflammation, cancer, and metastasis. 256–268. Cancer Metastasis Rev 29: 295–307. Denhardt D. (2005). Osteopontin expression correlates with melanoma Chuang JZ, Zhou H, Zhu M, Li SH, Li XJ, Sung CH. (2002). invasion. J Invest Dermatol 124: xvi–xviii. Characterization of a brain-enriched , MRJ, that inhibits Dey S, Banerjee P, Saha P. (2009). Cell cycle specific expression and Huntingtin aggregation and toxicity independently. J Biol Chem nucleolar localization of human J-domain containing co-chaperone 277: 19831–19838. Mrj. Mol Cell Biochem 322: 137–142.

Oncogene DNAJB6 multiprotein complex with GSK3b regulates OPN A Mitra et al 4482 Dhennin-Duthille I, Nyga R, Yahiaoui S, Gouilleux-Gruart V, Lopez-Bergami P, Fitchman B, Ronai Z. (2008). Under- Regnier A, Lassoued K et al. (2011). The tumor suppressor hTid1 standing signaling cascades in melanoma. Photochem Photobiol 84: inhibits STAT5b activity via functional interaction. J Biol Chem 289–306. 286: 5034–5042. Lucero OM, Dawson DW, Moon RT, Chien AJ. (2010). A re- Edwards KM, Munger K. (2004). Depletion of physiological levels of evaluation of the ‘oncogenic’ nature of Wnt/beta-catenin signaling the human TID1 protein renders cancer cell lines resistant to in melanoma and other cancers. Curr Oncol Rep 12: 314–318. apoptosis mediated by multiple exogenous stimuli. Oncogene 23: Martin MJ, Carling D, Marais R. (2009). Taking the stress out of 8419–8431. melanoma. Cancer Cell 15: 163–164. Eichhorn PJ, Creyghton MP, Bernards R. (2009). Protein phosphatase Menezes ME, Devine DJ, Shevde LA, Samant RS. (2011). Dickkopf1: 2A regulatory subunits and cancer. Biochim Biophys Acta 1795: a tumor suppressor or metastasis promoter? Int J Cancer 1–15. (doi:10.1002/ijc.26449). El-Tanani M, Barraclough R, Wilkinson MC, Rudland PS. (2001). Metge BJ, Liu S, Riker AI, Fodstad O, Samant RS, Shevde LA. Metastasis-inducing regulates the expression of the osteopontin (2010). Elevated osteopontin levels in metastatic melanoma gene by binding the transcription factor Tcf-4. Cancer Res 61: 5619– correlate with epigenetic silencing of breast cancer metastasis 5629. suppressor 1. Oncology 78: 75–86. El-Tanani MK, Campbell FC, Kurisetty V, Jin D, McCann M, Miller JR, Moon RT. (1996). through beta-catenin Rudland PS. (2006). The regulation and role of osteopontin in and specification of cell fate during embryogenesis. Genes Dev 10: malignant transformation and cancer. Cytokine Growth Factor Rev 2527–2539. 17: 463–474. Mitra A, Fillmore RA, Metge BJ, Rajesh M, Xi Y, King J et al. (2008). Fan CY, Lee S, Cyr DM. (2003). Mechanisms for regulation of Hsp70 Large isoform of MRJ (DNAJB6) reduces malignant activity of function by Hsp40. Cell Stress Chaperones 8: 309–316. breast cancer. Breast Cancer Res 10: R22. Fernandez JJ, Candenas ML, Souto ML, Trujillo MM, Norte M. Mitra A, Menezes ME, Shevde LA, Samant RS. (2010). DNAJB6 (2002). Okadaic acid, useful tool for studying cellular processes. induces degradation of beta-catenin and causes partial reversal of Curr Med Chem 9: 229–262. mesenchymal phenotype. J Biol Chem 285: 24686–24694. Fillmore RA, Mitra A, Xi Y, Ju J, Scammell J, Shevde LA et al. Mitra A, Shevde LA, Samant RS. (2009). Multi-faceted role of HSP40 (2009). Nmi (N-Myc interactor) inhibits Wnt/beta-catenin signaling in cancer. Clin Exp Metastasis 26: 559–567. and retards tumor growth. Int J Cancer 125: 556–564. Moore SR, Persons DL, Sosman JA, Bobadilla D, Bedell V, Smith DD Furger KA, Menon RK, Tuck AB, Bramwell VH, Chambers AF. et al. (2008). Detection of copy number alterations in (2001). The functional and clinical roles of osteopontin in cancer metastatic melanoma by a DNA fluorescence in situ hybridization and metastasis. Curr Mol Med 1: 621–632. probe panel and array comparative genomic hybridization: a Greillier L, Baas P, Welch JJ, Hasan B, Passioukov A. (2008). southwest oncology group study (S9431). Clin Cancer Res 14: for malignant pleural mesothelioma: current status. Mol 2927–2935. Diagn Ther 12: 375–390. O’Connell MP, Weeraratna AT. (2009). Hear the Wnt Ror: how Hennessy F, Nicoll WS, Zimmermann R, Cheetham ME, Blatch GL. melanoma cells adjust to changes in Wnt. Pigment Cell Melanoma (2005). Not all J domains are created equal: implications for Res 22: 724–739. the specificity of Hsp40-Hsp70 interactions. Protein Sci 14: Ohtsuka K, Hata M. (2000). Mammalian HSP40/DNAJ homologs: 1697–1709. cloning of novel cDNAs and a proposal for their classification and Hinoi T, Yamamoto H, Kishida M, Takada S, Kishida S, Kikuchi A. nomenclature. Cell Stress Chaperones 5: 98–112. (2000). Complex formation of adenomatous polyposis coli gene Orford K, Crockett C, Jensen JP, Weissman AM, Byers SW. (1997). product and axin facilitates glycogen synthase kinase-3 beta- Serine phosphorylation-regulated ubiquitination and degradation of dependent phosphorylation of beta-catenin and down-regulates beta-catenin. J Biol Chem 272: 24735–24738. beta-catenin. J Biol Chem 275: 34399–34406. Pecina-Slaus N, Zigmund M, Kusec V, Martic TN, Cacic M, Howe LR, Watanabe O, Leonard J, Brown AM. (2003). Twist is up- Slaus M. (2007). E-cadherin and beta-catenin expression patterns regulated in response to Wnt1 and inhibits mouse mammary cell in malignant melanoma assessed by image analysis. J Cutan Pathol differentiation. Cancer Res 63: 1906–1913. 34: 239–246. Hunter PJ, Swanson BJ, Haendel MA, Lyons GE, Cross JC. Polanowska-Grabowska R, Simon Jr CG, Falchetto R, Shabanowitz (1999). Mrj encodes a DnaJ-related co-chaperone that is J, Hunt DF, Gear AR. (1997). Platelet adhesion to collagen under essential for murine placental development. Development 126: flow causes dissociation of a phosphoprotein complex of heat-shock 1247–1258. proteins and protein phosphatase 1. Blood 90: 1516–1526. Kampinga HH, Craig EA. (2010). The HSP70 chaperone machinery: J Qiu XB, Shao YM, Miao S, Wang L. (2006). The diversity of the proteins as drivers of functional specificity. Nat Rev Mol Cell Biol DnaJ/Hsp40 family, the crucial partners for Hsp70 chaperones. Cell 11: 579–592. Mol Life Sci 63: 2560–2570. Kelley WL. (1998). The J-domain family and the recruitment of Reinhold MI, Kapadia RM, Liao Z, Naski MC. (2006). The Wnt- chaperone power. Trends Biochem Sci 23: 222–227. inducible transcription factor Twist1 inhibits chondrogenesis. J Biol Kurzik-Dumke U, Czaja J. (2007). Htid-1, the human homolog of the Chem 281: 1381–1388. Drosophila melanogaster l(2)tid tumor suppressor, defines a novel Riker AI, Enkemann SA, Fodstad O, Liu S, Ren S, Morris C et al. physiological role of APC. Cell Signal 19: 1973–1985. (2008). The gene expression profiles of primary and metastatic Kurzik-Dumke U, Horner M, Czaja J, Nicotra MR, Simiantonaki N, melanoma yields a transition point of tumor progression and Koslowski M et al. (2008). Progression of colorectal cancers metastasis. BMC Med Genomics 1: 13. correlates with overexpression and loss of polarization of expression Rodolfo M, Daniotti M, Vallacchi V. (2004). Genetic progression of of the htid-1 tumor suppressor. Int J Mol Med 21: 19–31. metastatic melanoma. Cancer Lett 214: 133–147. Kushiro K, Nunez NP. (2011). Ob/ob serum promotes a mesenchymal Sablina AA, Hahn WC. (2007). The role of PP2A A subunits in tumor cell phenotype in B16BL6 melanoma cells. Clin Exp Metastasis suppression. Cell Adh Migr 1: 140–141. 28: 877–886. Sablina AA, Hector M, Colpaert N, Hahn WC. (2010). Identification Larue L, Delmas V. (2006). The WNT/Beta-catenin pathway in of PP2A complexes and pathways involved in cell transformation. melanoma. Front Biosci 11: 733–742. Cancer Res 70: 10474–10484. Lin K, Baritaki S, Militello L, Malaponte G, Bevelacqua Y, Bonavida Saika S, Shirai K, Yamanaka O, Miyazaki K, Okada Y, Kitano A B. (2010). The role of B-RAF mutations in melanoma and the et al. (2007). Loss of osteopontin perturbs the epithelial-mesench- induction of EMT via dysregulation of the NF-kappaB/Snail/ ymal transition in an injured mouse lens epithelium. Lab Invest RKIP/PTEN circuit. Genes Cancer 1: 409–420. 87: 130–138.

Oncogene DNAJB6 multiprotein complex with GSK3b regulates OPN A Mitra et al 4483 Seeling JM, Miller JR, Gil R, Moon RT, White R, Virshup DM. Watson ED, Geary-Joo C, Hughes M, Cross JC. (2007). The Mrj (1999). Regulation of beta-catenin signaling by the B56 subunit of co-chaperone mediates keratin turnover and prevents the formation protein phosphatase 2A. Science 283: 2089–2091. of toxic inclusion bodies in trophoblast cells of the placenta. Shevde LA, Das S, Clark DW, Samant RS. (2010). Osteopontin: Development 134: 1809–1817. an effector and an effect of tumor metastasis. Curr Mol Med 10: Watson ED, Mattar P, Schuurmans C, Cross JC. (2009). Neural 71–81. stem cell self-renewal requires the Mrj co-chaperone. Dev Dyn 238: Shevde LA, Samant RS, Paik JC, Metge BJ, Chambers AF, Casey G 2564–2574. et al. (2006). Osteopontin knockdown suppresses tumorigenicity of Weeraratna AT. (2005). A Wnt-er wonderland–the complexity of Wnt human metastatic breast carcinoma, MDA-MB-435. Clin Exp signaling in melanoma. Cancer Metastasis Rev 24: 237–250. Metastasis 23: 123–133. Whalen KA, de Jesus R, Kean JA, Schaffhausen BS. (2005). Genetic Shih Ie M, Davidson B. (2009). Pathogenesis of ovarian cancer: clues analysis of the polyomavirus DnaJ domain. J Virol 79: 9982–9990. from selected overexpressed genes. Future Oncol 5: 1641–1657. Yeatman TJ, Chambers AF. (2003). Osteopontin and colon cancer Tsai MF, Wang CC, Chang GC, Chen CY, Chen HY, Cheng CL et al. progression. Clin Exp Metastasis 20: 85–90. (2006). A new tumor suppressor DnaJ-like , Zhang W, Yang J, Liu Y, Chen X, Yu T, Jia J et al. (2009). HLJ1, and survival of patients with non-small-cell lung carcinoma. PR55 alpha, a regulatory subunit of PP2A, specifically regulates J Natl Cancer Inst 98: 825–838. PP2A-mediated beta-catenin dephosphorylation. J Biol Chem 284: Tuck AB, Chambers AF, Allan AL. (2007). Osteopontin over- 22649–22656. expression in breast cancer: knowledge gained and possible Zhang Y, Yang Z, Cao Y, Zhang S, Li H, Huang Y et al. (2008). The implications for clinical management. J Cell Biochem 102: 859–868. Hsp40 family chaperone protein DnaJB6 enhances Schlafen1 Veeman MT, Slusarski DC, Kaykas A, Louie SH, Moon RT. (2003). nuclear localization which is critical for promotion of cell-cycle Zebrafish prickle, a modulator of noncanonical Wnt/Fz signaling, arrest in T-cells. Biochem J 413: 239–250. regulates gastrulation movements. Curr Biol 13: 680–685. Zhou BP, Hung MC. (2005). Wnt, hedgehog and snail: sister pathways Vos MJ, Hageman J, Carra S, Kampinga HH. (2008). Structural and that control by GSK-3beta and beta-Trcp in the regulation of functional diversities between members of the human HSPB, HSPH, metastasis. Cell Cycle 4: 772–776. HSPA, and DNAJ chaperone families. Biochemistry 47: 7001–7011. Zhou Y, Dai DL, Martinka M, Su M, Zhang Y, Campos EI et al. Wai PY, Kuo PC. (2004). The role of Osteopontin in tumor metastasis. (2005). Osteopontin expression correlates with melanoma invasion. J Surg Res 121: 228–241. J Invest Dermatol 124: 1044–1052. Wang H, Brown J, Martin M. (2011). Glycogen synthase kinase 3: a Zuidervaart W, Pavey S, van Nieuwpoort FA, Packer L, Out C, Maat point of convergence for the host inflammatory response. Cytokine W et al. (2007). Expression of Wnt5a and its downstream effector 53: 130–140. beta-catenin in uveal melanoma. Melanoma Res 17: 380–386.

Supplementary Information accompanies the paper on the Oncogene website (http://www.nature.com/onc)

Oncogene