CRYAB) Associates with the Cadherin/Catenin Adherens Junction and Impairs NPC Progression-Associated Properties

Oncogene (2012) 31, 3709–3720 & 2012 Macmillan Publishers Limited All rights reserved 0950-9232/12 www.nature.com/onc ORIGINAL ARTICLE Tumor suppressor Alpha B-crystallin (CRYAB) associates with the cadherin/catenin adherens junction and impairs NPC progression-associated properties

Z Huang1, Y Cheng1,PMChiu1, FMF Cheung2, JM Nicholls3, DL-W Kwong1, AWM Lee4, ER Zabarovsky5,6, EJ Stanbridge7, HL Lung1 and ML Lung1

1Department of Clinical Oncology and Center for Cancer Research, University of Hong Kong, Hong Kong (SAR), PR China; 2Department of Clinical Pathology, Pamela Youde Nethersole Eastern Hospital, Hong Kong (SAR), PR China; 3Department of Pathology, University of Hong Kong, Hong Kong (SAR), PR China; 4Department of Clinical Oncology, Pamela Youde Nethersole Eastern Hospital, Hong Kong (SAR), PR China; 5Department of Microbiology, Tumor and Cell Biology, Department of Clinical Science and Education, Sodersjukhuset, Karolinska Institute, Stockholm, Sweden; 6Laboratory of Structural and Functional Genomics, Engelhardt Institute of Molecular Biology, Russian Academy of Sciences, Moscow, Russia and 7Department of Microbiology and Molecular Genetics, University of California, Irvine, CA, USA

Alpha B-crystallin (CRYAB) maps within the nasophar- Introduction yngeal carcinoma (NPC) tumor-suppressive critical region 11q22-23 and its downregulation is significantly asso- Nasopharyngeal carcinoma (NPC) is a malignancy ciated with the progression of NPC. However, little is associated with multiple genetic alterations. Loss of known about the functional impact of CRYAB on NPC heterozygosity on chromosome 11q is frequently ob- progression. In this study we evaluated the NPC tumor- served in NPC (Hui et al., 1996; Lo et al., 2000). suppressive and progression-associated functions of Previously, through a monochromosome transfer ap- CRYAB. Activation of CRYAB suppressed NPC tumor proach and microsatellite typing, 11q13 and 11q22-23 formation in nude mice. Overexpression of CRYAB were identified as tumor-suppressive critical regions affected NPC progression-associated phenotypes such as (Cheng et al., 2000). Tumor-suppressor genes, THY1 loss of cell adhesion, invasion, interaction with the tumor and CADM1 (formerly called TSLC1) in 11q22-23, were microenvironment, invasive protrusion formation in three identified; their loss is associated with invasive and dimensional Matrigel culture, as well as expression metastatic properties characteristic of late- and ad- of epithelial–mesenchymal transition-associated markers. vanced-stage NPC (Lung et al., 2005, 2006, 2010). These CRYAB mediates this ability to suppress cancer progres- findings suggest that 11q22-23 harbors tumor-suppression by inhibition of E-cadherin cytoplasmic internaliza- sor genes whose loss is associated with NPC progres- tion and maintenance of b-catenin in the membrane that sion. Alpha B-crystallin (CRYAB), located in the tumor- subsequently reduces the levels of expression of critical suppressive 11q22-23 region, is a small heat-shock downstream targets such as cyclin-D1 and c-myc. Both protein and molecular chaperone. Downregulated ectopically expressed and recombinant CRYAB proteins CRYAB expression is associated with NPC progression were associated with endogenous E-cadherin and (Lung et al., 2008), as well as testicular (Takashi et al., b-catenin, and, thus, the cadherin/catenin adherens 1998) and breast (Lin et al., 2006) cancers, which junction. The CRYAB a-crystallin core domain is indicate that CRYAB is a potential tumor-suppressor responsible for the interaction of CRYAB with both gene in these cancers. E-cadherin and b-catenin. Taken together, these results Epithelial–mesenchymal transition (EMT) has an indicate that CRYAB functions to suppress NPC important role in cancer progression (Thiery et al., progression by associating with the cadherin/catenin 2009). EMT is regulated by multiple signals during adherens junction and modulating the b-catenin function. different stages in cancer progression. Imbalances in the Oncogene (2012) 31, 3709–3720; doi:10.1038/onc.2011.529; cell polarity networks and disruption of cell–cell published online 12 December 2011 junctions initiate the EMT process (Lee et al., 2006). The adherens junction mainly functions to maintain the Keywords: Alpha B-crystallin; nasopharyngealcarcinoma; physical association between cells. Internalization of E- E-cadherin; b-catenin cadherin caused by aberrant cellular signals disrupts the adherens junction (Palacios et al., 2005). The consequent Correspondence: Dr HL Lung or Professor ML Lung, Department of release of b-catenin from the membrane to the Clinical Oncology, University of Hong Kong, Room L2-23, 2/F, cytoplasm and nucleus activates its downstream tran- Laboratory Block, 21 Sassoon Road, Pokfulam, Hong Kong (SAR), scriptional program and promotes EMT. EMT and PR China. E-mail: [email protected] or [email protected] b-catenin signaling are important in NPC development Received 24 April 2011; revised 9 August 2011; accepted 6 October 2011; and progression (Chou et al., 2008). Loss of tumor- published online 12 December 2011 suppressor genes associated with NPC progression may Role of CRYAB in suppressing NPC progression Z Huang et al 3710

Figure 1 CRYAB suppresses NPC tumorigenicity in vivo.(a) Real-time quantitative reverse transcription–PCR analysis of CRYAB gene expression in NP460, BSD1 vector-alone and CRYAB stable transfectants. NP460 was used as the positive control of CRYAB expression. The relative CRYAB expression fold differences of each cell line were compared with the BSD1 vector-alone transfectant control. ±dox: Treatment with and without dox. (b) Western blot analysis of CRYAB protein expression levels in NP460, BSD1 vector-alone and CRYAB stable transfectants (±dox). a-Tubulin was used for protein normalization. (c) Tumor growth kinetics of BSD1 vector-alone and CRYAB stable transfectants (±dox) in nude mice. Each data point represents the average tumor volume of all six sites inoculated for each cell line.

facilitate EMT. To date, however, little is known about presence of dox ( þ dox), the tTA transcriptional their association. activator was inactivated, resulting in a reduction of In this study, we aim to investigate the functional role CRYAB mRNA and protein expression. of CRYAB in tumorigenesis and progression of NPC, as The BSD1 vector-alone and CRYAB stable transfec- well as the molecular mechanisms responsible for tants were subcutaneously injected into athymic nude CRYAB function. Re-expression of CRYAB in NPC mice. The control BSD1 was highly tumorigenic and cell lines reduced their tumorigenicity in vivo. NPC formed palpable tumors of 150 mm3 within 4 weeks in all progression-associated phenotypes, as well as EMT- six injection sites (Table 1). Overexpression of CRYAB associated markers, were suppressed upon CRYAB re- in the absence of dox significantly suppressed tumor- expression. The effects of CRYAB on the internalization igenicity (Figure 1c). No tumors arising from CRYAB42 of E-cadherin and b-catenin functions, and its associa- and only two tumors with CRYAB46 were observed at tion with the cadherin/catenin adherens junction, were week 4 after injection. The difference was statistically investigated in the present study. significant (Po0.05). Addition of dox reduced the levels of CRYAB and partially reversed the tumor-suppressive effect. However, there was still a clear growth-inhibitory Results effect compared with BSD1. This may be attributed to some leakiness in the system as reported previously CRYAB suppresses NPC tumor formation in vivo (Lung et al., 2010). Excised tumors (±dox) were CRYAB was previously shown to be downregulated in examined by western blots and no longer express NPC cell lines and tumors (Lung et al., 2008). We used a CRYAB (data not shown). These results are consistent tetracycline-regulated gene expression system (Protopo- with CRYAB being a potent suppressor of NPC tumor pov et al., 2002) to further study its tumor-suppressive formation in vivo. function in vivo. pETE-Bsd-CRYAB was transfected into the tetracycline transcriptional activator (tTA)- Overexpression of CRYAB affects NPC adhesion expressing HONE1-2 NPC cell line. Stable transfec- and invasion tants, CRYAB42 and CRYAB46, were selected. As Adhesion assays show that overexpression of CRYAB in control, pETE-Bsd was transfected into HONE1-2 to CRYAB42 and CRYAB46 (Àdox) significantly in- generate the stable transfectant, BSD1. Both CRYAB creases the cellular adhesive ability compared with mRNA and protein expression were induced in BSD1 (Po0.05) (Figure 2a). Reduction of CRYAB CRYAB42 and CRYAB46 in the absence of doxycy- expression in CRYAB42 and CRYAB46 ( þ dox) cline (Àdox) and verified by quantitative real-time PCR decreased the adhesive ability to similar levels observed (Figure 1a) and western blot analyses (Figure 1b). In the in BSD1. In the transwell invasion assay, when CRYAB

Oncogene Role of CRYAB in suppressing NPC progression Z Huang et al 3711 Table 1 Tumorigenicity assays of CRYAB transfectants Cell line Identiﬁcation Dox Tumor formation P*^ (no. tumors/no. sites) in week 4 after injection

BSD1 HONE1-2 Â pETE-Bsd À 6/6 þ 6/6 0.692* CRYAB42 HONE1-2 Â pETE-Bsd-CRYAB À 0/6 0.001^ þ 4/6 0.023*, 0.022^ CRYAB46 HONE1-2 Â pETE-Bsd-CRYAB À 2/6 0.003^ þ 4/6 0.507*, 0.018^

Abbreviations: CRYAB, Alpha B-crystallin; Dox, doxycycline. *P-value obtained by comparison between tumor sizes ±dox. ^P-value obtained by comparison with BSD1. was overexpressed (Àdox), the relative invasive ability CRYAB42 and CRYAB46. Significant reduction of was reduced to 62% and 74% (Po0.05), respectively, in viable cell numbers was observed by MTT assay for CRYAB42 and CRYAB46 (Figure 2b), whereas the CRYAB42 and CRYAB46 after 2 weeks in 3D Matrigel invasive ability in CRYAB42 and CRYAB46 ( þ dox) culture (Figure 3b), indicating that reduction of was restored. The migration potential of CRYAB42 and spheroid sizes in CRYAB42 and CRYAB46 is due to CRYAB46 was not changed significantly in the trans- decreased ability to proliferate in Matrigel culture. well migration assay under the same conditions as in the EMT is a central process occurring during cancer invasion assay (data not shown). Hence, the decreased progression (Thiery et al., 2009). Loss of epithelial cell number observed in the invasion assay was not due polarity and formation of disorganized, non-cohesive to change of migration ability. To further confirm and irregular invasive protrusions in 3D culture are inhibition of invasion by CRYAB, real-time quantitative hallmarks of the morphological phenotypes indicative of cell migration and invasion assays were performed. In EMT (Grunert et al., 2003). In 3D Matrigel culture, the invasion assay, 1:10 diluted Matrigel was coated in BSD1 (±dox) and CRYAB42 and CRYAB46 ( þ dox) the chambers. Matrigel is a gelatinous protein mixture, showed typical EMT phenotypes with disorganized, which mimics the complex extracellular environment of non-cohesive and irregular invasive protrusions. By the cancer cells. A rapid increase in cell index was contrast, CRYAB42 and CRYAB46 (Àdox) showed observed after 30 h in BSD1 cells compared with polarized and well-organized spheroids (Figure 3c). CRYAB42 and CRYAB46, indicating their reduced In addition to morphological changes, we examined invasion abilities (Figure 2c). At 50 h, the cell index of the expression levels of the EMT-associated markers BSD1 was 1.5942, whereas the cell indices of CRYAB42 E-cadherin and vimentin. Upregulation of E-cadherin and CRYAB46 were 0.3606 and 0.2736, respectively. and downregulation of vimentin were observed in In the migration assay, the cell index did not show as CRYAB42 and CRYAB46 (Àdox) at both the mRNA much difference between BSD1 vector-alone and (P ¼ 0.02 and 0.026 for E-cadherin; P ¼ 0.034 and 0.037 CRYAB stable transfectants. At 50 h, the cell index for vimentin) (Figure 3d) and protein (Figure 3e) levels, was 2.3856 for BSD1, and 1.8547 and 1.4072 for as compared with the parental cell line HONE1-2 and CRYAB42 and CRYAB46, respectively. Another Ep- the BSD1 vector-alone transfectant. The þ dox treat- stein–Barr virus-positive NPC cell line, C666, was also ment restored the altered expression of E-cadherin and tested for its invasive ability; pCR3.1 and pCR3.1- vimentin in CRYAB42 and CRYAB46 clones. CRYAB, CRYAB plasmids were transiently transfected into thus, has a potential role in suppressing NPC progres- C666, followed by transwell invasion assay. The western sion-associated properties. blot shows that CRYAB is expressed in the transiently transfected C666 cells, which showed significantly reduced invasiveness when CRYAB is expressed CRYAB affects E-cadherin localization (Figure 2d). These results suggest that overexpression Disruption of cell–cell junctions such as the cadherin/ of CRYAB potently affects both the adhesive and catenin adherens junction is associated with EMT invasive properties of the NPC cells. (Thomson et al., 2011). In BSD1 vector-alone transfectants, E-cadherin internalized from the cell membrane to the cytoplasm, indicating disruption of the cadherin/ CRYAB suppresses NPC progression-associated catenin adherens junction in these cells. By contrast, E- properties cadherin remained in the cell membrane in CRYAB42 To study CRYAB function in the tumor microenviron- and CRYAB46, indicating maintenance of the cadherin/ ment, three dimensional (3D) Matrigel culture was used. catenin adherens junction in the cell membrane When cultured in Matrigel for 14 days, BSD1 (±dox) (Figure 4a), which was also observed by immunohisto- formed large spheroids. Overexpression of CRYAB chemical staining in clinical samples (Supplementary reduced the relative spheroid sizes to 54% and 61%, Figure 1). This difference was statistically significant respectively, in CRYAB42 and CRYAB46 (Àdox), (Po0.05) and was confirmed after subcellular fractiona- compared with BSD1 (Figure 3a). Reduction of CRYAB tion and western blot analysis. In the membrane expression ( þ dox) restored the relative spheroid sizes in fractions, E-cadherin accumulated to higher levels in the

Oncogene Role of CRYAB in suppressing NPC progression Z Huang et al 3712 CRYAB42 and CRYAB46 clones. The E-cadherin levels fractionation and western blot analysis of CRYAB were similar in the nuclear fractions and were weak in showed that the majority of CRYAB protein in the the cytoplasmic fractions (Figure 4b). Subcellular CRYAB42 and CRYAB46 clones was located in the

Oncogene Role of CRYAB in suppressing NPC progression Z Huang et al 3713 cytoplasm. CRYAB was moderately expressed in the cell catenin adherens junction. In co-immunoprecipitation membrane and was very weak in the nucleus (Figure 4c). assays using CRYAB (Figure 6a) and b-catenin (Figure 6b) antibodies, the association between CRYAB and endogenous b-catenin and E-cadherin was detected CRYAB affects b-catenin function in CRYAB42 and CRYAB46. Moreover, the level of b-Catenin, when released from the cadherin/catenin b-catenin associating with E-cadherin was dramatically adherens junction, translocates to the cytoplasm and increased in the CRYAB42 and CRYAB46 clones nucleus, and acts as a transcription factor to promote compared, indicating a strong association of b-catenin tumor progression and EMT (Heuberger and Birchmeier, and E-cadherin with CRYAB. 2010). As CRYAB suppressed the internalization of This association was further confirmed in glutathione E-cadherin, we further investigated whether CRYAB S-transferase (GST) pull-down assays. GST-tagged affected b-catenin function. Immunofluorescence staining CRYAB full-length and truncated proteins (Figure 6c), showed that b-catenin was located throughout the namely CRYAB-DN (deleted N-terminus), CRYAB- cytoplasm and nucleus in the control BSD1, whereas it DN core (deleted N-terminus and a-crystallin core localized mainly to the cell membrane in the CRYAB42 þ domain) and CRYAB-DC (deleted C-terminus), were and CRYAB46 transfectants (Figure 5a). This difference expressed, purified and incubated with total cell lysates was statistically significant (P 0.05). Membrane localiza- o of the vector-alone BSD1. Endogenous E-cadherin was tion of b-catenin was also observed by immunohistochem- pulled down by the CRYAB full-length, CRYAB-DN ical staining in clinical samples (Supplementary Figure 1). and CRYAB-DC, but not by the CRYAB-DN core This observation was further confirmed after subcellular þ proteins. Endogenous b-catenin was pulled down by fractionation and subsequent western blot analysis. In CRYAB full-length, moderately pulled down by both nuclear and cytoplasmic fractions, the b-catenin CRYAB-DN and CRYAB-DC, but not pulled-down levels were reduced in CRYAB42 and CRYAB46 as by the CRYAB-DN core proteins (Figure 6d). This compared with BSD1, whereas in the membrane fractions, þ indicates that the a-crystallin core domain is responsible there were increased levels of b-catenin in the CRYAB42 for the interaction of CRYAB with both E-cadherin and and CRYAB46 clones (Figure 5b). b-catenin. The N-terminus and C-terminus of CRYAB The NPC cell line HONE1-2 and the BSD1 vector- are partially responsible for the interaction with b- alone stable transfectant strongly express b-catenin and catenin. Taken together, these results indicate that its downstream targets c-myc and cyclin-D1. By CRYAB is likely to be associated with the cadherin/ contrast, in CRYAB42 and CRYAB46 (Àdox), the catenin adherens junction in NPC. protein expression levels of b-catenin, c-myc and cyclin- D1 were downregulated; reduction of CRYAB expression in CRYAB42 and CRYAB46 ( þ dox) restored these expression levels (Figure 5c). Downregulation of Discussion b-catenin, c-myc and cyclin-D1 was also observed in another NPC cell line, CNE2, when CRYAB was Our earlier study showed that CRYAB is a candidate transiently expressed (Figure 5d). These results indicate tumor-suppressor gene in NPC (Lung et al., 2008). In that overexpression of CRYAB maintains b-catenin in the current study, we show that overexpression of the cell membrane and downregulates target genes such CRYAB significantly suppresses NPC tumorigenicity in as c-myc and cyclin-D1. nude mice. However, there is still a clear growth inhibition in the CRYAB stable transfectants as The CRYAB protein is associated with the cadherin/ compared with the BSD1 vector-alone transfectant catenin adherens junction control, even when expression of CRYAB is repressed Previous studies showed that CRYAB interacts (Ghosh by dox. This may be attributed to some leakiness of et al., 2007b) and colocalizes (Maddala and Rao, 2005) CRYAB expression in vivo, which was also reported in with b-catenin, as well as interacts with the kidney- earlier studies using this dox-regulated system (Li et al., specific cadherin (Thedieck et al., 2008). CRYAB is, 2004; Lung et al., 2006). The expression of CRYAB is therefore, potentially associated with the cadherin/ lost from all excised tumors of both CRYAB42 and

Figure 2 CRYAB affects NPC adhesion and invasion abilities. (a) Adhesion ability in BSD1 vector-alone and CRYAB stable transfectants. Significant increase of cell adherent ability was observed in CRYAB stable transfectants (Àdox) as compared with BSD1 vector-alone and CRYAB stable transfectant ( þ dox) controls. Scale bar: 100 mm. The number of adherent cells in each cell line was counted. * and **Statistically significant difference from BSD1 vector-alone transfectants and CRYAB transfectants ( þ dox), respectively (Po0.05). (b) Transwell invasion assay of BSD1 vector-alone and CRYAB stable transfectants. Scale bar: 100 mm. Invading cells at the bottom surface of the transwell filter were stained and counted. Relative invasive ability in each cell line was calculated. * and **Statistically significant difference from BSD1 vector-alone transfectants and CRYAB transfectants ( þ dox) (Po0.05). (c) Quantitative assessment of migration and invasion abilities of BSD1 vector-alone and CRYAB stable transfectants. Real- time quantitative cell migration and invasion assays were performed using CIM-plate. Cell index represents the number of cells migrated or invaded through the chamber at different time points. (d) Transwell invasion assay of C666 with or without transient transfection of CRYAB. Scale bar 100 mm. *Statistically significant difference from C666 transient transfected with pCR3.1 (Po0.05). CRYAB protein expression in C666 with or without transient transfection of CRYAB was examined by western blotting.

Oncogene Role of CRYAB in suppressing NPC progression Z Huang et al 3714 CRYAB46 (±dox), indicating CRYAB is clearly in vivo. The nude mouse assay provides key functional inactivated in the tumorigenic revertants, and further evidence that CRYAB suppresses NPC growth in vivo supports the tumor-suppressive function of CRYAB and is a potent tumor-suppressor gene in NPC.

Oncogene Role of CRYAB in suppressing NPC progression Z Huang et al 3715 CRYAB is correlated with lymph node metastasis in 2003). In this current study, overexpression of CRYAB NPC (Lung et al., 2008); reduced in peri-neural invasion functionally suppresses tumor progression-associated of head and neck cutaneous squamous cell carcinoma properties such as loss of cell adhesion and invasion, (Solares et al., 2010); and is signiﬁcantly associated with as well as viability and appearance of irregular invasive adverse ovarian cancer patient survival (Stronach et al., protrusion formation in 3D Matrigel culture in NPC,

Figure 4 CRYAB function associated with inhibition of cytoplasmic internalization of E-cadherin. (a) Immunofluorescence staining analysis of E-cadherin. CRYAB inhibits the internalization of E-cadherin. The arrows indicate membrane-bound E-cadherin. Scale bar: 1 mm. E-cadherin localization from different fields in a minimum of 150 cells was counted and the percentage of cells showing E-cadherin localization at different sites was calculated. *Statistically significant differences (Po0.05). (b) Subcellular localization of E-cadherin in BSD1 vector-alone and CRYAB stable transfectants. Subcellular fractionation followed by immunoblotting was performed with BSD1 vector-alone and CRYAB stable transfectants. a-Tubulin was used as a marker for normalization of the cytoplasmic fraction. Histone3 (H3) was used as a marker for normalization of the nuclear fraction. The membrane fraction was normalized to the Coomassie Blue staining of the proteins in that fraction. (c) Subcellular localization of CRYAB in BSD1 vector- alone and CRYAB stable transfectants. Subcellular fractionation followed by immunoblotting was performed with BSD1 vector-alone and CRYAB stable transfectants. a-Tubulin was used as a marker for normalization of the cytoplasmic fraction. Histone3 (H3) was used as a marker for normalization of the nuclear fraction. The membrane fraction was normalized to the Coomassie Blue staining of the proteins in that fraction.

Figure 3 CRYAB suppresses NPC progression-associated properties. (a) 3D Matrigel culture assay of BSD1 vector-alone and CRYAB stable transfectants. Scale bar: 100 mm. The relative spheroid sizes of each cell line compared with BSD1 vector-alone were counted and calculated 2 weeks after inoculation. * and **Statistically significant difference from BSD1 vector-alone transfectants and CRYAB transfectants ( þ dox) (Po0.05). (b) Cell viability in 3D Matrigel culture was measured by MTT assay. *Statistically significant difference from BSD1 vector-alone transfectants and CRYAB transfectants ( þ dox) (Po0.05). *Statistically significant difference from BSD1 vector-alone transfectants and CRYAB transfectants ( þ dox) (Po0.05). (c) Invasive protrusion structure formation in 3D Matrigel culture of BSD1 vector-alone and CRYAB stable transfectants. The BSD1 vector-alone transfectant and CRYAB transfectants ( þ dox) showed a disorganized morphology, with invasive protrusion structures in 3D Matrigel culture 4 weeks after inoculation, whereas CRYAB transfectants (Àdox) formed polarized cell spheroids. Higher magnification ( Â 10) images of invasive protrusion structures and polarized cell spheroids are shown. Phalloidin was used to stain the actin filaments. Scale bar: 50 mm. (d) Real-time quantitative PCR analysis of E-cadherin and Vimentin in BSD1 vector-alone and CRYAB stable transfectants (±dox). The relative fold differences of E-cadherin and Vimentin expression of each cell line were compared with the BSD1 vector-alone transfectant control. *Statistically significant differences (Po0.05). (e) Western blot analysis of vimentin and E-cadherin in the parental NPC cell line HONE1-2, BSD1 vector-alone and CRYAB stable transfectants (±dox). a-Tubulin was used for normalization.

Oncogene Role of CRYAB in suppressing NPC progression Z Huang et al 3716

Figure 5 CRYAB affects b-catenin function. (a) Immunostaining of b-catenin in BSD1 vector-alone and CRYAB stable transfectants are shown. The arrow indicates typical membrane staining. Scale bar 1 mm. b-Catenin localization from different ﬁelds in a minimum of 500 cells was counted and the percentage of cells showing b-catenin localization at different sites was calculated. *Statistically signiﬁcant differences (Po0.05). (b) Subcellular localization of b-catenin in BSD1 vector-alone and CRYAB stable transfectants. Subcellular fractionation followed by immunoblotting was performed with BSD1 vector-alone and CRYAB stable transfectants. a-Tubulin was used as a marker for normalization of the cytoplasmic fraction. Histone3 (H3) was used as a marker for normalization of the nuclear fraction. The membrane fraction was normalized to the Coomassie Blue staining of the proteins in that fraction. (c) Western blot analysis of b-catenin and its downstream targets c-myc and cyclin-D1 in the parental NPC cell line HONE1-2, BSD1 vector-alone and CRYAB stable transfectants. a-Tubulin was used for normalization. (d) Western blot analysis of b-catenin and its downstream targets c-myc and cyclin-D1 in the NPC cell line CNE2, with or without transient transfection of CRYAB. a-Tubulin was used for normalization.

providing functional evidence of the impact of CRYAB stable transfectants further suggest association of in NPC progression. CRYAB with EMT. Loss of cell polarity and disruption EMT has a crucial role in cancer progression. In of cell–cell junctions such as the adherens junction is NPC, EMT was shown to affect cancer invasion in both involved in the initiation of EMT (Lee et al., 2006). Epstein–Barr virus-dependent and independent manners E-cadherin is one of the major components in the (Lin et al., 2009; Kong et al., 2010). E-cadherin and adherens junctions and it is internalized from the cell vimentin are two well-studied markers associated with membrane to the cytoplasm during disruption of the EMT that are aberrantly expressed in NPC (Zheng adherens junction (Lu et al., 2003). Inhibition of et al., 1999; Lo et al., 2003). The observed expression cytoplasmic internalization of E-cadherin in the changes of E-cadherin and vimentin in the CRYAB CRYAB stable transfectants indicates that CRYAB

Oncogene Role of CRYAB in suppressing NPC progression Z Huang et al 3717

Figure 6 CRYAB protein is associated with the cadherin/catenin complex. Ectopically expressed CRYAB protein is associated with endogenous cadherin/catenin complex in vivo as shown by co-immunoprecipitation assay using CRYAB or b-catenin antibodies. Cell lysates prepared from BSD1 vector-alone and CRYAB stable transfectants were precipitated with antibodies against CRYAB (a) and b-catenin (b), or purified IgG controls, followed by immunoblotting with the indicated antibodies. Input: Cell lysate without immunoprecipitation. (c) A schematic diagram showing the CRYAB full-length and truncated variants. (d) The purified recombinant CRYAB protein interacts with endogenous E-cadherin and b-catenin. The total cell lysate of the BSD1 vector-alone transfectant was incubated with a GST control, GST-CRYAB full-length or GST-CRYAB truncated variants. Immunoblotting with anti-GST antibodies served as loading controls. may be involved in the disruption of the cadherin/ actin cytoskeleton and mediates its mechanical stability catenin adherens junction and EMT initiation. Recently, (Green et al., 2010). In our study, we demonstrated the nuclear localization of E-cadherin was reported in interaction among CRYAB, E-cadherin and b-catenin Merkel cell carcinomas, solid pseudopapillary tumors in vitro and in vivo, which suggests the involvement of of the pancreas and renal cell carcinomas (Chetty and CRYAB in the cadherin/catenin adherens junction. Serra, 2008). The detailed mechanisms of E-cadherin Previous studies found that the CRYAB protein nuclear localization are still poorly understood. p120, interacts with b-catenin at the N-terminus, the which regulates the trafficking of E-cadherin, reportedly a-crystallin core domain and the C-terminus (Ghosh may have a role (Reynolds and Roczniak-Ferguson, et al., 2007b), and interacts with Ksp–cadherin, another 2004). The nuclear localization of E-cadherin in BSD1 member of the cadherin family, in the N-terminus and CRYAB stable transfectants remains to be inves- (Thedieck et al., 2008). Our GST pull-down assay shows tigated further. that the a-crystallin core domain is mainly responsible Disruption of the cadherin/catenin adherens junction for the interaction with b-catenin. The moderate effects releases the membrane-bound b-catenin into the cyto- seen with the N-terminal and C-terminal fragments are plasm. The b-catenin then further accumulates and consistent with the previous finding of Ghosh et al. interacts with the T-cell factor/lymphoid enhancer (2007b). The a-crystallin core domain, but not the factor (TCF/LEF) complex to enhance the transcription N-terminus or the C-terminus, is required for the of genes promoting tumor progression (Nelson and interaction with E-cadherin. The a-crystallin core Nusse, 2004). In the current study, overexpression of domain is the major interaction site of CRYAB with CRYAB in NPC suppresses the b-catenin oncogenic cytoskeleton proteins and also facilitates actin polymer- function by maintaining the membrane-bound b-catenin. ization (Ghosh et al., 2007a; Ohto-Fujita et al., 2007). c-myc and cyclin-D1 are well-studied b-catenin/TCF/ The interaction of CRYAB with E-cadherin and LEF target genes (Heuberger and Birchmeier, 2010). b-catenin in the a-crystallin core domain may facilitate Downregulation of c-myc and cyclin-D1 in the CRYAB the interaction between the adherens junction and the stable and transient transfectants further indicates that actin cytoskeleton, which is required for adhesion suppression of b-catenin is associated with overexpres- strengthening, and junctional plaque assembly and sion of CRYAB. maintenance (Green et al., 2010). In the cadherin/catenin adherens junctions, E-cadher- Growing evidence suggests the impact of chromosome in interacts with p120–catenin, g-catenin and b-catenin. 11q tumor-suppressor genes in NPC progression (Lung b-Catenin interacts with cortical actin-associated et al., 2006, 2010). Our current study now provides clear a-catenin, which links the adherens junctions to the evidence that CRYAB suppresses tumorigenesis, as well

Oncogene Role of CRYAB in suppressing NPC progression Z Huang et al 3718 as progression-associated phenotypes and EMT-asso- Adhesion assay ciated markers in NPC. CRYAB inhibits E-cadherin A total of 1 Â 104 cells was seeded in a 96-well plate and cytoplasmic internalization and also affects the onco- incubated in culture medium for 8 h, followed by washing in genic function of b-catenin by maintaining membrane- phosphate-buffered saline twice and ﬁxing with 70% ethanol bound b-catenin. In addition, we demonstrate that the for 15 min at room temperature. The remaining adherent cells were washed by phosphate-buffered saline and stained with CRYAB protein is also associated with the cadherin/ 0.5% crystal violet for 10 min. Adherent cells in each well were catenin adherens complex through the a-crystallin core counted. domain. Thus, CRYAB is a potent tumor-suppressor gene, which is associated with tumor progression in Invasion assay NPC. Additional studies are needed to unravel the A total of 2 Â 105 cells was seeded into the chamber of a 24- detailed mechanisms of CRYAB function in the well Matrigel-coated membrane ﬁlter (Becton Dickinson cadherin/catenin adherens complex. Labware, Franklin Lakes, NJ, USA). The cell invasion assay was performed as described previously (Lung et al., 2010).

Materials and methods Real-time quantitative cell migration and invasion assays The assays were performed using the xCELLigence System Cell lines and culture conditions (Roche, Penzberg, Germany) for real-time cell analysis. In The NPC cell lines, CNE2 (Sizhong et al., 1983) and C666 brief, cells are seeded in a 16-well CIM-plate containing (Cheung et al., 1999), the immortalized nasopharyngeal electronic sensors for direct study of cell migration and epithelial cell line, NP460 (Li et al., 2006), and the engineered invasion on the xCELLigence RTCA DP Instrument (Roche). NPC HONE1-2 cells (Protopopov et al., 2002) were cultured For the migration assay, 40 000 cells were seeded directly on as described previously (Lung et al., 2008). Stable transfectants the upper chamber of the CIM-plate. For the invasion assay, with the CRYAB transgene or the pETE-Bsd vector alone were 40 000 cells were seeded on the upper chamber coated with 1:10 maintained in culture medium containing 5 mg/ml blasticidin. diluted Matrigel (BD Biosciences). The bottom chambers were ﬁlled with Dulbecco’s modiﬁed Eagle’s medium plus 5% serum Quantitative reverse transcription–PCR analysis as chemoattractants. Migration and invasion were monitored Quantitative reverse transcription–PCR analysis was for 70 h. Cell index represents the number of cells inside the performed as described previously (Cheung et al., 2009). wells based on the measured electrical impedance. CRYAB and GAPDH (glyceraldehyde-3-phosphate dehydro- genase) Taqman probes and the SYBR Green PCR master mix Matrigel culture spheroid and invasive protrusion formation were purchased from Applied Biosystems (ABI, Carlsbad, CA, assays USA). The sequences of primers are shown in Supplementary A Matrigel basement membrane matrix (BD Biosciences) was Table 1. coated onto 24-well plates and then 2000 cells were seeded on top. For the spheroid formation assay, the Matrigel culture 1 Western blot analysis was incubated at 37 C for 14 days and spheroids formed in the Preparation of cell lysates, sodium dodecyl sulfate–PAGE and culture were counted and spheroid sizes were measured under transfer were performed as described previously (Lung et al., an inverted light microscope (Nikon Instruments, Melville, 2005). CRYAB (SPA-222; Stressgen, Victoria, BC, Canada), NY, USA). For protrusion formation, the Matrigel culture 1 E-cadherin (36/E-cadherin; BD Biosciences Labware, MA, was incubated at 37 C for 30 days. Fixing and staining were USA), vimentin (RV202; BD Biosciences), b-catenin (L54E2; performed as described previously (Wong et al., 2012). Cell Signaling Technology, Danvers, MA, USA), cyclin-D1 (sc-753; Santa Cruz Biotechnology, Santa Cruz, CA, USA), MTT assay c-myc (sc-764; Santa Cruz Biotechnology), a-tubulin (Ab-1; A Matrigel basement membrane matrix (BD Biosciences) was Calbiochem, Darmstadt, Germany) and histone H3 (sc-10809; coated onto 24-well plates and then 2000 cells were seeded on Santa Cruz Biotechnology) antibodies were used for western top. The MTT assay was performed as described previously blots. (Lung et al., 2006).

Transient and stable transfection of CRYAB Immunofluorescent staining The 528 bp CRYAB cDNA fragment was subcloned from Immunofluorescence staining was performed as described pcDNA3-CRYAB (Simon et al., 2007) into the pCR3.1 vector previously (Leung et al., 2008) using the following primary (Invitrogen, Carlsbad, CA, USA) at the HindIII and EcoRI antibodies as mentioned previously: b-catenin (1:200) and sites, and into the pETE-Bsd vector (Protopopov et al., 2002) E-cadherin (1:100). Cells with different b-catenin and at the BglII and NotI sites. Transient and stable transfection of E-cadherin localization were counted using the ImageJ soft- CRYAB was performed with the Lipofectamine 2000 reagent ware (NIH, Bethesda, MD, USA). The percentage of cells with (Invitrogen), as described previously (Lung et al., 2006). membrane or cytoplasmic and nuclear localization of b-catenin or E-cadherin was calculated. In vivo tumorigenicity assay The tumorigenicity of cell lines was examined by subcutaneous Subcellular fractionation injection as described previously (Lung et al., 2006). In brief, Subcellular fractionation was performed following the proto- 1 Â 107 cells were injected into six sites on three 6 to 8-week-old col by Abcam (http://www.abcam.com/ps/pdf/protocols/sub female immunodeficient athymic BALB/C nu/nu nude mice cellular_fractionation.pdf). In brief, cells were lysed with a and tumor growth was measured weekly. To inhibit tetra- subcellular fractionation buffer and cell lysates were centri- cycline-regulated expression of CRYAB, 200 mg/ml dox was fuged. The pellet was washed in fractionation buffer and added to the drinking water (Lung et al., 2010). resuspended in nuclear buffer; this is the nuclear fraction. The

Oncogene Role of CRYAB in suppressing NPC progression Z Huang et al 3719 supernatant was re-centrifuged at a higher speed (40 000 Statistical analysis r.p.m.); the supernatant, after ultra-centrifugation, is the Student’s t-test was performed for adhesion, invasion and 3D cytoplasmic fraction. The pellet was washed with fractionation Matrigel culture assays; quantitative real-time reverse tran- buffer and resuspended in nuclear buffer; this is the membrane scription–PCR; and immunofluorescence staining for statisti- fraction. cal comparison between the BSD1 vector-alone and CRYAB stable transfectants. A P-value of o0.05 was considered statistically significant. Co-immunoprecipitation The co-immunoprecipitation assay was performed as described previously (Zhang et al., 2010). In brief, cells were lysed and Conflict of interest then pre-cleared with rProtein-G-agarose (Invitrogen) and incubated with CRYAB (Stressgen) (1:250), b-catenin (Cell The authors declare no conflict of interest. Signaling Technology) (1:250) or a control purified rabbit IgG (Invitrogen), at 4 1C overnight. Lysates were then immuno- precipitated with rProtein-G-agarose, washed, and then Acknowledgements subjected to sodium dodecyl sulfat-PAGE and western blot analysis. We thank Patrick Vicart for the pcDNA3-CRYAB construct. We acknowledge the funding support from the Research Grants Council grants, and the University Grants Council of GST recombinant protein expression and purification Hong Kong Special Administrative Region, People’s Republic GST recombinant protein expression and purification were of China, for AoE/M-06/08 to MLL; The University of Hong performed as described previously (Wong et al., 2011). The Kong Seed Funding Programme for Basic Research to HLL; and cloning and purification of the truncated CRYAB proteins are the Swedish Cancer Society, the Swedish Research Council, the described in the Supplementary information and Supplemen- Swedish Institute and Karolinska Institute to ERZ. We acknowl- tary Table 2. edge the Area of Excellence Hong Kong NPC Research Tissue Bank for the NPC specimens and tissue blocks.

References

Cheng Y, Stanbridge EJ, Kong H, Bengtsson U, Lerman MI, Lung ML. Kong QL, Hu LJ, Cao JY, Huang YJ, Xu LH, Liang Y et al. (2010). (2000). A functional investigation of tumor suppressor gene activities in Epstein–Barr virus-encoded LMP2A induces an epithelial–mesench- a nasopharyngeal carcinoma cell line HONE1 using a monochromo- ymal transition and increases the number of side population stem- some transfer approach. Genes Chromosomes Cancer 28: 82–91. like cancer cells in nasopharyngeal carcinoma. PLoS Pathog 6: Chetty R, Serra S. (2008). Nuclear E-cadherin immunoexpression: e1000940. from biology to potential applications in diagnostic pathology. Adv Lee JM, Dedhar S, Kalluri R, Thompson EW. (2006). The epithelial– Anat Pathol 15: 234–240. mesenchymal transition: new insights in signaling, development, and Cheung AK, Lung HL, Ko JM, Cheng Y, Stanbridge EJ, Zabarovsky disease. J Cell Biol 172: 973–981. ER et al. (2009). Chromosome 14 transfer and functional studies Leung AC, Wong VC, Yang LC, Chan PL, Daigo Y, Nakamura Y identify a candidate tumor suppressor gene, mirror image poly- et al. (2008). Frequent decreased expression of candidate tumor dactyly 1, in nasopharyngeal carcinoma. Proc Natl Acad Sci USA suppressor gene, DEC1, and its anchorage-independent growth 106: 14478–14483. properties and impact on global gene expression in esophageal Cheung ST, Huang DP, Hui AB, Lo KW, Ko CW, Tsang YS et al. carcinoma. Int J Cancer 122: 587–594. (1999). Nasopharyngeal carcinoma cell line (C666-1) consistently Li HM, Man C, Jin Y, Deng W, Yip YL, Feng HC et al. (2006). harbouring Epstein–Barr virus. Int J Cancer 83: 121–126. Molecular and cytogenetic changes involved in the immortalization Chou J, Lin YC, Kim J, You L, Xu Z, He B et al. (2008). of nasopharyngeal epithelial cells by telomerase. Int J Cancer 119: Nasopharyngeal carcinoma—review of the molecular mechanisms 1567–1576. of tumorigenesis. Head Neck 30: 946–963. Li J, Wang F, Protopopov A, Malyukova A, Kashuba V, Minna JD Ghosh JG, Houck SA, Clark JI. (2007a). Interactive sequences in the et al. (2004). Inactivation of RASSF1C during in vivo tumor stress protein and molecular chaperone human alphaB crystallin growth identiﬁes it as a tumor suppressor gene. Oncogene 23: 5941– recognize and modulate the assembly of ﬁlaments. Int J Biochem 5949. Cell Biol 39: 1804–1815. Lin DI, Barbash O, Kumar KG, Weber JD, Harper JW, Klein-Szanto Ghosh JG, Shenoy Jr AK, Clark JI. (2007b). Interactions between AJ et al. (2006). Phosphorylation-dependent ubiquitination of cyclin important regulatory proteins and human alphaB crystallin. D1 by the SCF (FBX4–alphaB crystallin) complex. Mol Cell 24: Biochemistry 46: 6308–6317. 355–366. Green KJ, Getsios S, Troyanovsky S, Godsel LM. (2010). Intercellular Lin JC, Liao SK, Lee EH, Hung MS, Sayion Y, Chen HC et al. (2009). junction assembly, dynamics, and homeostasis. Cold Spring Harb Molecular events associated with epithelial to mesenchymal transi- Perspect Biol 2: a000125. tion of nasopharyngeal carcinoma cells in the absence of Epstein– Grunert S, Jechlinger M, Beug H. (2003). Diverse cellular and Barr virus genome. J Biomed Sci 16: 105. molecular mechanisms contribute to epithelial plasticity and Lo AK, Liu Y, Wang XH, Huang DP, Yuen PW, Wong YC metastasis. Nat Rev Mol Cell Biol 4: 657–665. et al. (2003). Alterations of biologic properties and gene Heuberger J, Birchmeier W. (2010). Interplay of cadherin-mediated expression in nasopharyngeal epithelial cells by the Epstein–Barr cell adhesion and canonical Wnt signaling. Cold Spring Harb virus-encoded latent membrane protein 1. Lab Invest 83: Perspect Biol 2: a002915. 697–709. Hui AB, Lo KW, Leung SF, Choi PH, Fong Y, Lee JC et al. (1996). Lo KW, Teo PM, Hui AB, To KF, Tsang YS, Chan SY et al. (2000). Loss of heterozygosity on the long arm of chromosome 11 in High resolution allelotype of microdissected primary nasopharyn- nasopharyngeal carcinoma. Cancer Res 56: 3225–3229. geal carcinoma. Cancer Res 60: 3348–3353.

Oncogene Role of CRYAB in suppressing NPC progression Z Huang et al 3720 Lu Z, Ghosh S, Wang Z, Hunter T. (2003). Downregulation of Sizhong Z, Xiukung G, Yi Z. (1983). Cytogenetic studies on an caveolin-1 function by EGF leads to the loss of E-cadherin, epithelial cell line derived from poorly differentiated nasopharyn- increased transcriptional activity of beta-catenin, and enhanced geal carcinoma. Int J Cancer 31: 587–590. tumor cell invasion. Cancer Cell 4: 499–515. Solares CA, Boyle GM, Brown I, Parsons PG, Panizza B. (2010). Lung HL, Bangarusamy DK, Xie D, Cheung AK, Cheng Y, Kumaran Reduced alphaB-crystallin staining in perineural invasion of head MK et al. (2005). THY1 is a candidate tumour suppressor gene with and neck cutaneous squamous cell carcinoma. Otolaryngol Head decreased expression in metastatic nasopharyngeal carcinoma. Neck Surg 142: S15–S19. Oncogene 24: 6525–6532. Stronach EA, Sellar GC, Blenkiron C, Rabiasz GJ, Taylor KJ, Miller Lung HL, Cheung AK, Cheng Y, Kwong FM, Lo PH, Law EW et al. EP et al. (2003). Identification of clinically relevant genes on (2010). Functional characterization of THY1 as a tumor suppressor chromosome 11 in a functional model of ovarian cancer tumor gene with antiinvasive activity in nasopharyngeal carcinoma. Int J suppression. Cancer Res 63: 8648–8655. Cancer 127: 304–312. Takashi M, Katsuno S, Sakata T, Ohshima S, Kato K. (1998). Lung HL, Cheung AK, Xie D, Cheng Y, Kwong FM, Murakami Y Different concentrations of two small stress proteins, alphaB et al. (2006). TSLC1 is a tumor suppressor gene associated with crystallin and HSP27 in human urological tumor tissues. Urol Res metastasis in nasopharyngeal carcinoma. Cancer Res 66: 9385–9392. 26: 395–399. Lung HL, Lo CC, Wong CC, Cheung AK, Cheong KF, Wong N et al. Thedieck C, Kalbacher H, Kratzer U, Lammers R, Stevanovic S, Klein (2008). Identification of tumor suppressive activity by irradiation G. (2008). alpha B-Crystallin is a cytoplasmic interaction partner of microcell-mediated chromosome transfer and involvement of the kidney-specific cadherin-16. J Mol Biol 378: 145–153. alpha B-crystallin in nasopharyngeal carcinoma. Int J Cancer 122: Thiery JP, Acloque H, Huang RY, Nieto MA. (2009). Epithelial– 1288–1296. mesenchymal transitions in development and disease. Cell 139: Maddala R, Rao VP. (2005). alpha-Crystallin localizes to the leading 871–890. edges of migrating lens epithelial cells. Exp Cell Res 306: 203–215. Thomson S, Petti F, Sujka-Kwok I, Mercado P, Bean J, Monaghan M Nelson WJ, Nusse R. (2004). Convergence of Wnt, beta-catenin, and et al. (2011). A systems view of epithelial–mesenchymal transition cadherin pathways. Science 303: 1483–1487. signaling states. Clin Exp Metastasis 28: 137–155. Ohto-Fujita E, Fujita Y, Atomi Y. (2007). Analysis of the alphaB- Wong VC, Chen H, Ko JM, Chan KW, Chan YP, Law S et al. (2012). crystallin domain responsible for inhibiting tubulin aggregation. Tumor suppressor Dual-specificity Phosphatase 6 (DUSP6) impairs Cell Stress Chaperones 12: 163–171. cell invasion and epithelial–mesenchymal transition (EMT)-asso- Palacios F, Tushir JS, Fujita Y, D’Souza-Schorey C. (2005). ciated phenotype. Int J Cancer 130: 83–95. Lysosomal targeting of E-cadherin: a unique mechanism for the Wong VC, Ko JM, Qi RZ, Li PJ, Wang LD, Li JL et al. (2011). downregulation of cell–cell adhesion during epithelial to mesench- Abrogated expression of DEC1 during oesophageal squamous cell ymal transitions. Mol Cell Biol 25: 389–402. carcinoma progression is age- and family history-related and Protopopov AI, Li J, Winberg G, Gizatullin RZ, Kashuba VI, Klein G significantly associated with lymph node metastasis. Br J Cancer et al. (2002). Human cell lines engineered for tetracycline-regulated 104: 841–849. expression of tumor suppressor candidate genes from a frequently Zhang J, Yu L, Wu X, Zou L, Sou KK, Wei Z et al. (2010). The affected chromosomal region, 3p21. J Gene Med 4: 397–406. interacting domains of hCdt1 and hMcm6 involved in the chromatin Reynolds AB, Roczniak-Ferguson A. (2004). Emerging roles for p120– loading of the MCM complex in human cells. Cell Cycle 9: 4848–4857. catenin in cell adhesion and cancer. Oncogene 23: 7947–7956. Zheng Z, Pan J, Chu B, Wong YC, Cheung AL, Tsao SW. (1999). Simon S, Fontaine JM, Martin JL, Sun X, Hoppe AD, Welsh MJ et al. Downregulation and abnormal expression of E-cadherin and beta- (2007). Myopathy-associated alphaB-crystallin mutants: abnormal catenin in nasopharyngeal carcinoma: close association with phosphorylation, intracellular location, and interactions with other advanced disease stage and lymph node metastasis. Hum Pathol small heat shock proteins. J Biol Chem 282: 34276–34287. 30: 458–466.

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

Oncogene