Identification of Siah-Interacting Protein As a Potential Regulator Of

Oncogene (2010) 29, 6357–6366 & 2010 Macmillan Publishers Limited All rights reserved 0950-9232/10 www.nature.com/onc ORIGINAL ARTICLE Identiﬁcation of Siah-interacting protein as a potential regulator of apoptosis and curcumin resistance

J Luo1, J Yang1, B-Y Yu1, W Liu2,MLi2 and S-M Zhuang1

1Key Laboratory of Gene Engineering of the Ministry of Education, State Key Laboratory of Biocontrol, School of Life Sciences, Sun Yat-sen University, Guangzhou, PR China and 2Proteomics Center, Zhongshan School of Medicine, Sun Yat-sen University, Guangzhou, PR China

The mechanism underlying curcumin (diferuloylmethane) molecules that regulate apoptosis is therefore essential resistance is still largely unknown. Here we employed for overcoming this global problem. proteomic approach to identify the Siah-interacting Curcumin (diferuloylmethane) is an active ingredient protein (SIP) as a candidate for detailed study, because isolated from turmeric. Growing evidences indicate that the spot intensity of SIP on a two-dimensional gel curcumin is able to induce apoptosis in various cancer displayed 70–90% reduction in curcumin-sensitive cells, cells (Karunagaran et al., 2005) and to suppress but remained unchanged in curcumin-resistant sublines, tumorigenesis, angiogenesis and metastasis in diverse after curcumin treatment. Both gain- and loss-of-function chemical-induced or xenograft tumor models (Kawamori studies revealed that SIP promoted curcumin-induced et al.,1999;Liet al., 2002; Aggarwal et al.,2005; apoptosis. Moreover, SIP underwent phosphorylation and Bhandarkar and Arbiser, 2007). Moreover, both phase I nuclear translocation in curcumin-sensitive but not and phase II clinical trials show that curcumin is effective resistant cells, upon curcumin exposure. The nuclear and well tolerated for cancer patients (Sharma et al., translocation of SIP was remarkably impaired when a 2004; Dhillon et al., 2008). To date, comprehensive putative nuclear localization sequence (NLS, amino acid studies have been conducted to characterize the anti- (aa) 143–159) was deleted or the serine 141 was mutated cancer effect of curcumin, but little is known about its into alanine, whereas truncation of the N-terminal region resistance. In a previous study, we showed that Chk1- (aa 1–43) obviously increased the nuclear import of SIP. mediated G2/M arrest may serve as a mechanism for the In accordance with their nuclear localization, N-terminal resistance of hepatoma cells to curcumin-induced apop- truncation significantly enhanced the proapoptotic effect tosis (Wang et al., 2008). Furthermore, our preliminary of SIP, whereas NLS deletion or Ser141Ala mutation results revealed that curcumin could induce apoptosis in attenuated the apoptosis-promoting activity of both wild- MOLT-4, a human lymphocytic leukemia cell line, and type- and N-terminal truncated-SIP. These data suggest its 6-thioguanine (TG)-resistant subline (MOLT-4/TG), that SIP plays a role in apoptosis and curcumin resistance, but not in its 9-b-D-arabinofuranosyl guanine (Ara-G)- and the function of SIP may be regulated by different resistant derivative (MOLT-4/AraG). To identify pro- motifs, such as the NLS, N-terminal region and serine 141. teins involved in curcumin resistance, a proteomic Our findings provide new insights into the biological approach was used to screen for differential proteins in significance of SIP and the mechanisms of drug resistance. MOLT-4 and its derivatives with and without curcumin Oncogene (2010) 29, 6357–6366; doi:10.1038/onc.2010.358; treatment. As a result, the Siah-interacting protein (SIP) published online 23 August 2010 was identified as a candidate for detailed study, because spot intensity of SIP on a two-dimensional gel (2D-gel) Keywords: Siah-interacting protein; SIP; curcumin; obviously reduced in MOLT-4 and MOLT-4/TG, but not apoptosis in MOLT-4/AraG after curcumin treatment. The SIP protein, also named as Calcyclin binding protein, was originally identified in mouse brain as a binding protein of S100A6 (Calcyclin) (Filipek and Introduction Wojda, 1996), whereas human homolog of SIP was discovered as a protein interacting with Siah (Matsuzawa Drug resistance is a major obstacle in cancer therapy and Reed, 2001). It has been shown that SIP regulates and is responsible for high cancer mortality (O’Connor, the ubiquitination and degradation of b-catenin (Mat- 2009). Dysfunction of apoptotic pathways has been suzawa and Reed, 2001; Santelli et al., 2005). Further- implicated in drug resistance, and identification of more, SIP was downregulated in renal cell carcinoma tissues, and overexpression of SIP inhibited the ability of Correspondence: Dr S-M Zhuang, Key Laboratory of Gene Engineer- tumor cells to form colony in soft agar and to develop ing of the Ministry of Education, School of Life Sciences, Sun Yat-sen tumor in nude mice (Ning et al., 2007; Sun et al., 2007). University, Guangzhou 510275, PR China. E-mail: [email protected] or [email protected] These data suggest that SIP deregulation may contribute Received 25 November 2009; revised 1 July 2010; accepted 14 July 2010; to cell proliferation and tumorigenesis. However, the role published online 23 August 2010 of SIP in apoptosis and in drug resistance is still unclear. Siah-interacting protein in apoptosis J Luo et al 6358 In this study, we reveal that SIP promotes curcumin- induced apoptosis. Curcumin treatment results in phosphorylation and nuclear translocation of SIP in curcumin-sensitive but not in resistant cells. The N- terminal region, a nuclear localization sequence (NLS) and a potential phosphorylation site Ser141 are crucial for the nuclear import of SIP. Furthermore, defect in SIP nuclear translocation is associated with cell resistance to curcumin- and also to other stimuli-induced apoptosis. These results imply SIP as a regulator of apoptosis and drug resistance, and as a potential target for cancer therapy.

Results

The intensity change of SIP spot is different between curcumin-sensitive and -resistant cells upon curcumin treatment Our preliminary results showed that 25 mM curcumin was able to induce obvious apoptosis in MOLT-4 and MOLT- 4/TG upon treatment for 24 h, whereas MOLT-4/AraG was extremely resistant to curcumin-induced apoptosis even at an extended exposure time of 48 h and at a high dose of 50 mM (Figure 1a, Supplementary Figure S1 and data not shown). To investigate the molecular basis of curcumin resistance, we compared the proteomes of curcumin-treated MOLT-4 and its derivative sublines (MOLT-4/TG and MOLT-4/AraG) with that of DMSO (vehicle control)-treated cells. The proteomes of DMSO- treated MOLT-4, MOLT-4/TG and MOLT-4/AraG cells were also compared. As a result, 332 spots with significant intensity differences (ratio X1.5, Po0.05) between either of the compared groups were revealed, and further MALDI-TOF mass spectroscopy analysis successfully identified 45 proteins from these spots (data not shown). Among them, the spot intensity of SIP (Supplementary Table S1) on a 2D gel displayed 70–90% reduction in MOLT-4 and MOLT-4/TG, but remained almost the same in MOLT-4/AraG cells, after curcumin treatment (Figures 1b and c). Furthermore, the spot intensity of SIP was higher in MOLT-4/AraG compared with MOLT-4 and MOLT-4/TG (Figures 1b and c). These findings indicate that the intensity change of SIP spot may be related to the sensitivity of tumor cells to curcumin- induced apoptosis. Therefore, the SIP protein was chosen for further investigation.

SIP displays distinct phosphorylation pattern and cellular Figure 1 The intensity change of SIP spot is different between localization in curcumin-sensitive and -resistant tumor curcumin-sensitive and -resistant tumor cells upon curcumin treatment. (a) The sensitivity of MOLT-4 and its derivative lines cells to curcumin-induced apoptosis. Cells were treated with 25 mM To verify the above results, reverse transcription (RT)– curcumin for the indicated times, followed by apoptosis analysis PCR was initially employed and we found that using DAPI staining. (b) Representative 2D gel for MOLT-4 and curcumin exposure did not decrease the level of SIP its derivative lines. Cells were treated with DMSO or 25 mM curcumin (CUR) for 14 h, followed by 2D-gel analysis. Spot mRNA in sensitive MOLT-4 and MOLT-4/TG cells identiﬁed as SIP protein is indicated by arrows. (c) Semi- (Figure 2a), implying that SIP protein may be regulated quantitative measurement of the SIP protein spots from 2D gel. at a post-transcriptional level upon curcumin treatment. The intensity of each protein spot was determined, normalized to Due to the lack of commercial antibody, we raised a the sum of pixel intensities of all spots on the gel, and quantiﬁed as polyclonal antiserum against SIP (anti-SIP), and de- a percentage volume using ImageMaster 2D Platinum software 5.0. For (a) and (c), values represent the mean±s.d. of three monstrated that it was suitable for immunoblotting, independent experiments. *Po0.05; **Po0.01; ***Po0.001. Oncogene Siah-interacting protein in apoptosis J Luo et al 6359

Figure 2 SIP displays distinct phosphorylation pattern in curcumin-sensitive and -resistant tumor cells. (a, b) The mRNA and protein levels of SIP in MOLT-4 and its derivative cells. Cells were treated with DMSO or 25 mM curcumin (CUR) for 14 h, followed by reverse transcription–PCR (a) or western blotting (b) analysis. hPBGD and GAPDH were used as internal controls. (c) The phosphorylation status of SIP in MOLT-4 and MOLT-4/AraG cells. Cells were treated with DMSO, 25 mM curcumin or 200 ng/ml PMA for 14 h, followed by immunoprecipitation (IP) with anti-SIP (SIP) and then western blotting (WB) with anti-SIP or anti-phospho-serine (p- Ser). (d) The sensitivity of HepG2 cells to curcumin-induced apoptosis. Cells were treated with 25 or 50 mM curcumin for the indicated times, followed by apoptosis analysis using DAPI staining. Values are expressed as mean±s.d. of three independent experiments. (e) The protein level of SIP in HepG2 cells. Cells were treated with 25 or 50 mM curcumin for the indicated times and then analyzed by western blotting. b-actin, internal control. (f) The phosphorylation status of SIP in HepG2. Cells were treated with DMSO, 50 mM curcumin or 200 ng/ml PMA for 12 h, followed by IP/WB analysis. The value under each lane in (a), (b) and (e) indicates the relative expression level of SIP. The intensity for each band was densitometrically quantified and the intensity of SIP was then normalized by that of the internal control in each lane. The normalized intensity of SIP in the first lane was set as relative expression level 1. immunoprecipitation and immunostaining (Supplemen- to the altered phosphorylation pattern of SIP, the tary Figure S2). This anti-SIP was thereafter used for the whole cell extracts from MOLT-4 or MOLT-4/AraG detection of both exogenous and endogenous SIP treated with curcumin or DMSO were immunoprecipi- protein. Interestingly, although immunoblotting analy- tated with anti-SIP, followed by immunoblotting sis revealed a slight but reproducible reduction of SIP with anti-phosphoserine antibody (Figure 2c). Cells protein in both MOLT-4 and MOLT-4/TG after treated with PMA, a protein kinase C (PKC) activator curcumin treatment and a higher level of SIP in proved to induce SIP phosphorylation (Filipek et al., MOLT-4/AraG than in MOLT-4 and MOLT-4/TG 2002b), served as a positive control. Interestingly, cells (Figure 2b), the difference was much less evident similar to PMA-treated cells, a strong band indicating than that on 2D gel (Figures 1b and c). These data phosphorylation in the serine residues of SIP suggest that other modifications, besides the total was detected in curcumin-exposed MOLT-4 cells amount of SIP protein, also contribute to the intensity (Figure 2c, bottom left panel). In contrast, serine change of SIP spot on 2D gel. phosphorylation was found neither in curcumin- nor It has been demonstrated that SIP is phosphorylated in PMA-treated MOLT-4/AraG cells (Figure 2c, in unknown serine residues in neuroblastoma cells bottom right panel), suggesting that the phosphoryla- upon KCl or retinoic acid treatment (Filipek et al., tion pattern of SIP is distinct in curcumin-sensitive 2002b; Wu et al., 2003). Furthermore, phosphorylation (MOLT-4) and -resistant (MOLT-4/AraG) tumor may cause changes in both isoelectric point and cells, and the kinase(s) or other upstream factor(s), molecular weight of protein and therefore lead to the which are critical for SIP phosphorylation in response to relocation of protein spots on 2D gel. To verify whether either curcumin or PMA stimuli, may be deficient in the observed intensity change in SIP spot was due MOLT-4/AraG.

Oncogene Siah-interacting protein in apoptosis J Luo et al 6360

Figure 3 SIP displays distinct cellular localization in curcumin-sensitive and -resistant tumor cells. (a) Localization of endogenous SIP in MOLT-4 and its derivative lines. Cells were treated with DMSO or 25 mM curcumin (CUR) for 14 h, followed by immunostaining using anti-SIP. (b) Localization of endogenous SIP in HepG2. Cells were treated with 25 or 50 mM curcumin for indicated times, followed by immunostaining using anti-SIP. Nuclei were visualized by PI staining. Images were captured under a ﬂuorescence microscope. The results are representative of three independent experiments.

We further employed a human hepatoma cell line, Nuclear translocation of SIP is regulated by different HepG2, to verify the above observations. Curcumin motifs treatment resulted in increased apoptosis and decreased As reported (Filipek et al., 2002a; Santelli et al.,2005), SIP in HepG2 cells in a time- and dose-dependent the N-terminal amino acids (aa) 1–47 of SIP comprise a manner (Figures 2d and e). Significantly, enhanced self-dimerization domain, and the C-terminal contains phosphorylation in the serine residues of SIP was also binding motifs for SKP1 and S100 proteins (Figure 4a). detected in both curcumin- and PMA-exposed HepG2 To explore the mechanisms underlying nuclear transloca- cells (Figure 2f), which was similar to that in MOLT-4 tion of SIP, the protein sequence of SIP was analyzed cells (Figure 2c, left panel). computationally. We predicted a region with putative Collectively, these results suggest that curcumin NLS in aa 143–159 using PSORT II system (http:// treatment may result in increased phosphorylation of psort.hgc.jp/form2.html; Nakai and Horton, 1999), and a SIP in curcumin-sensitive but not resistant cells. cluster of three potential phosphorylation sites (Ser137, We next examined whether the cellular localization Ser141 and Ser142) near the putative NLS using Disphos of SIP changed upon curcumin treatment, considering (http://www.ist.temple.edu/disphos; Iakoucheva et al., that it underwent both nuclear translocation and 2004), NetPhos (http://www.cbs.dtu.dk/services/NetPhos; phosphorylation in KCl- or retinoic acid-treated cells Blom et al., 1999) and KinasePhos (http://kinasephos. (Filipek et al., 2002b; Wu et al., 2003). Using immuno- mbc.nctu.edu.tw; Huang et al., 2005) programs. Based fluorescence staining, we found that curcumin exposure on these pieces of evidence, a series of SIP mutants caused a significant translocation of endogenous SIP (Figures 4a and b) were constructed and their effects on from the cytoplasm to nucleus in both MOLT-4 and the cellular localization of SIP were examined. The MOLT-4/TG but not in MOLT-4/AraG cells HepG2 cell line was employed for the following studies (Figure 3a). Furthermore, the staining of SIP was as it displayed similar phenotypes as MOLT-4 upon obviously stronger in MOLT-4/AraG than in MOLT-4 curcumin treatment and it demonstrated much higher and MOLT-4/TG (Figure 3a), which was consistent transfection efficiency than MOLT-4. We found that with the results from immunoblotting (Figure 2b). The transiently expressed EGFP-fused wild-type SIP (SIP- translocation phenomenon was also observed in WT) was mainly present in the cytoplasm of HepG2 curcumin-treated HepG2 cells, in which SIP was initially cells, whereas curcumin exposure obviously increased detected in the cytoplasm and then translocated to the proportion of cells with nuclear accumulation of SIP the nucleus in a time- and dose-dependent manner (Figure 4c and Supplementary Figure S3), suggesting (Figure 3b). Interestingly, similar to what we observed that both endogenous (Figure 3) and exogenous SIP can in MOLT-4/AraG cells (Figure 3a), curcumin treatment translocate from the cytoplasm to nucleus in response to could not affect the location of SIP in another hepatoma curcumin treatment. The nuclear translocation effi- cell line Hep3B (data not shown), which was previously ciency of the exogenously expressed mutant SIP was found to be resistant to curcumin-induced apoptosis then assessed. Compared with SIP-WT, the Ser141Ala (Wang et al., 2008). mutant (WT-S141A) and the NLS-deleted SIP (WT- In summary, our results indicate that both phospho- ND) exhibited dramatic reduction in the nuclear rylation and nuclear accumulation of SIP are crucial for translocation, whereas the WT-S137A and WT-S142A curcumin-triggered apoptosis. mutants revealed a similar rate of nuclear accumulation

Oncogene Siah-interacting protein in apoptosis J Luo et al 6361

Figure 4 Nuclear translocation of SIP is regulated by different motifs. (a) Schematic diagram of the vectors expressing SIP-WT and its mutant forms. The residue numbers are indicated according to the protein sequence of wild-type SIP. (b) Detection of exogenous SIP and its mutant forms by immunoblotting. HepG2 cells were transfected with plasmids expressing EGFP or EGFP-fused wild type or mutant SIP, then analyzed by western blotting using anti-GFP antibody. (c) Nuclear localization of SIP and its mutant forms. HepG2 cells were transfected with plasmids expressing EGFP-fused wild type or mutant SIP for 24 h, then incubated with 25 or 50 mM curcumin (CUR) or DMSO for 12 h, followed by DAPI staining. Only EGFP-positive viable cells were scored. The cell with green ﬂuorescence in the nucleus was deﬁned as nuclear accumulation of SIP. At least 400 cells were counted in each sample. Values represent the mean±s.d. of three independent experiments. *Po0.05; ***Po0.001. (d) Phosphorylation status of SIP and its mutant forms. HepG2 cells were transfected with plasmids expressing His-tagged wild-type or mutant SIP for 24 h, then exposed to 50 mM curcumin for 12 h, followed by immunoprecipitation (IP) using anti-His (His) and then western blotting (WB) using anti-His or anti-phosphoserine (p-Ser).

(Figure 4c and Supplementary Figure S3). These obser- cell model with and without curcumin exposure (data vations were further conﬁrmed using wild-type and not shown), it is worthwhile to investigate whether mutant SIP with His-tag instead of EGFP (Supplemen- deletion of the aa 1–43 region could affect the nuclear tary Figure S4). Notably, without curcumin treatment, translocation of SIP. EGFP-fused SIP-S construct and a nuclear localization of endogenous SIP was only series of SIP-S mutant were generated and transfected observed in 2% HepG2 cells (Figure 3b, the DMSO- into HepG2 cells (Figures 4a and b). Surprisingly, SIP-S treated group), whereas nuclear import of exogenous underwent nuclear translocation spontaneously, inde- SIP-WTwas found in 24% cells (Figure 4c), implying pendent of curcumin exposure, suggesting that the N- that a large amount of overexpressed SIP might terminal region may be important for segregating SIP promote its nuclear import. protein in cytoplasm (Figure 4c, Supplementary Figures It is reported that there exists a transcription variant S3 and S4). Consistent with the ﬁndings from wild-type of hCacyBP/SIP (Iakoucheva et al., 2004), which results SIP, S-ND and S-S141A mutants exhibited decreased in an isoform of SIP protein (hereafter named as SIP-S) nuclear accumulation, whereas S-S137A and S-S142A that displayed truncation in the N-terminal aa 1–43 mutants showed a similar rate of nuclear localization, (GenBank accession no. NP_055227.1). Although we compared with SIP-S (Figure 4c and Supplementary failed to detect the expression of SIP-S protein in our Figure S3).

Oncogene Siah-interacting protein in apoptosis J Luo et al 6362 To further validate the relationship between phosphorylation and nuclear import of SIP, we analyzed the phosphorylation patterns of SIP-WT, WT-S141A and WT-S142A, under DMSO or curcumin treatment (Figure 4d). Compared with cells transfected with SIP-WT, WT-S141A but not WT-S142A transfectants displayed signiﬁcantly less serine phosphorylation (Figure 4d). Collectively, these data suggest that the N-terminal, the NLS region and the phosphorylation of Ser141 residue are crucial in regulating the cellular localization of SIP.

Nuclear accumulation of SIP is associated with cell sensitivity to apoptosis To unravel the biological significance of nuclear accumulation of SIP, HepG2 cells transfected with EGFP (vector control), EGFP-fused wild type or mutant SIP were treated with DMSO or curcumin for 24 h. Apoptosis analysis by morphological examination revealed that compared with EGFP-transfected cells, SIP-WT transfectants displayed higher apoptotic rates with and without curcumin exposure (Figure 5a), suggesting that SIP may have a proapoptotic function. Furthermore, the proapoptotic effect of SIP became more evident when the N-terminal aa 1–43 was deleted (SIP-S in Figure 5a), which was in accordance with the increased nuclear accumulation of SIP-S. Interestingly, the NLS deletion and Ser141Ala mutation, which disturbed the nuclear translocation of SIP, dramatically attenuated the apoptosis-promoting effect of both SIP-WT and SIP-S (Figure 5a). In contrast, Ser137Ala or Ser142Ala mutation, which had no impact on SIP translocation (Figure 4c), could not affect the proapoptotic activity of SIP (data not shown). These results were reproducible in HepG2 cells exposed to different concentrations of curcumin for 12 h (Supplementary Figure S5) and were confirmed by terminal deoxynu- cleotidyl transferase-mediated 20-deoxyuridine 50-tripho- sphate nick end labeling (TUNEL) analysis, an alternative technique for apoptosis detection (Supple- mentary Figure S6). Taken together, our data indicate Figure 5 The impact of SIP on curcumin-induced apoptosis. that defects in the nuclear translocation of SIP may (a) Apoptosis-promoting effect of SIP-WT and SIP-S was attenuated by NLS-deletion and Ser141Ala mutation. HepG2 cells contribute to curcumin resistance. were transfected with plasmids expressing EGFP-fused wild-type To further confirm the role of SIP in apoptosis, loss- or mutant SIP for 24 h, then incubated with DMSO or 25 mM of-function analyses were performed using HepG2 cells curcumin (CUR) for 24 h, followed by apoptosis analysis using that stably expressed the short-hairpin RNA (shRNA) DAPI staining. Only EGFP-positive cells were evaluated. At least 400 cells were analyzed in each sample. (b) SIP knockdown targeting coding sequence (G2-659) or 3-untranslated impaired curcumin-induced apoptosis. (c–d) SIP knockdown region of SIP (G2-1170), or the scrambled control suppressed curcumin-induced depolarization of mitochondrial shRNA (G2-NC). For each shRNA, two stable clones membrane. For (b–d), stable SIP-knockdown HepG2 lines (G2- were generated from independent retroviral infections 659 and G2-1170) and its control cells (G2-NC) were treated with (Supplementary Figure S7a). Both G2-659 and G2-1170 35 mM curcumin for 12 h, then applied to apoptosis analysis using DAPI staining (b), and to detection of mitochondrial membrane cells displayed significantly lower apoptotic rates than potential (Dcm) using MTRed and MTGreen staining followed by G2-NC upon curcumin treatment (Figure 5b and flow cytometry analysis (c, d). The intensity of MTRed staining Supplementary Figure S7b), indicating that loss of SIP depends on Dcm, whereas the intensity of MTGreen remains may inhibit curcumin-induced apoptosis in HepG2. the same regardless of Dcm and was thus used as an internal staining control. As for Dcm analysis, both bar chart (c) and repre- The mitochondrial membrane potential (Dcm) was sentative dot plots (d) are shown. For (a–c), values represent the then examined using double staining with MitoTracker mean±s.d. of three independent experiments. *Po0.05; Deep Red FM (MTRed) and MitoTracker Green FM **Po0.01; ***Po0.001. (MTGreen). In agreement with the apoptotic rates, curcumin-triggered loss of Dcm was abated in SIP- knockdown cells compared with the control cells

Oncogene Siah-interacting protein in apoptosis J Luo et al 6363

Figure 6 The Ser141Ala mutation attenuates the nuclear accumulation and the apoptosis-promoting effect of SIP in different cells treated with diverse apoptosis stimuli. The indicated cells were transfected with plasmids expressing EGFP-fused wild-type or mutant SIP for 24 h, incubated for 12 h with 25 mg/ml etoposide (ETO), 500 ng/ml doxorubicin (DOXO), 25 mM nitro mustard (NM), 100 J/m2 ultraviolet irradiation (UV) or 35 mM curcumin (CUR), then applied to analyses for SIP localization and cell apoptosis. For details, see the legends to Figures 4c and 5a. At least 400 cells were counted in each sample. Values represent the mean±s.d. of three independent experiments. *Po0.05; **Po0.01; ***Po0.001.

(Figures 5c and d, Supplementary Figures S7c and d), nuclear translocation are crucial for the function of SIP. implying that SIP may be involved in the mitochondrial- We further demonstrate that phosphorylation at the mediated apoptosis. Ser141 residue, the N-terminal region (aa 1–43) and the To investigate whether the above observations were putative NLS region (aa 143–159) are crucial for nuclear general phenomena in different cells with different translocation of SIP. These data are consistent with the stimuli-triggered apoptosis, the effect of Ser141Ala contention that the nuclear import of most signaling mutation was subsequently examined in HepG2 cells proteins is NLS-dependent and is regulated by protein exposed to other apoptosis stimuli, including etoposide, modifications, such as phosphorylation (Jans and doxorubicin, nitro-mustard, ultraviolet irradiation Hubner, 1996; Lange et al.,2007).Interestingly,although (Figure 6, left panels) and in another two human curcumin-resistant subline MOLT-4/AraG shows hepatoma cell lines (QGY-7703 and BEL-7404) treated defect in both SIP phosphorylation and nuclear trans- with curcumin (Figure 6, right panels). Consistently, location, it contains an SIP-WT sequence identical to SIP-S displayed more pronounced nuclear accumulation that in MOLT-4, according to our direct sequencing data and proapoptotic activity than SIP-WT, whereas (data not shown). These findings implicate that the Ser141Ala mutation significantly attenuated the effect kinases/upstream factors that regulate SIP phospho- of both SIP-WT and SIP-S (Figure 6). These results rylation may be deficient in MOLT-4/AraG. implicate that nuclear translocation of SIP, which is It has been revealed that SIP is phosphorylated by regulated at least partly by Ser141, may be involved in a PKC in vitro (Filipek et al., 2002b) and curcumin common apoptotic signaling pathway. activates PKC in a membrane-dependent and calcium- regulated manner (Mahmmoud, 2007). In our study, both curcumin and PKC activator (PMA) cause SIP Discussion phosphorylation in MOLT-4 but not MOLT-4/AraG cells. Furthermore, NetPhosK (Blom et al., 2004) and Results from others (Filipek et al., 2002b; Wu et al., KinasePhos (Huang et al., 2005) programs predict the 2003) and our studies suggest that phosphorylation and Ser141 residue of SIP as the phosphorylation site of

Oncogene Siah-interacting protein in apoptosis J Luo et al 6364 PKC. The role of PKC in the apoptosis signaling of SIP signaling may activate an SIP-independent pathway is contradictory and depends on the isozyme apoptotic pathway. and cell context (Gutcher et al., 2003). For example, Although little is known about the mechanism of SIP- PKCe-deficient thyroid cells display reduced sensitivity mediated apoptotic signaling, some insights may be to apoptosis induced by DNA-damaging agents, obtained from previous studies on proteins interacting including actinomycin D, ultraviolet irradiation and with SIP. It has been shown that S100 family members doxorubicin (Knauf et al., 1999), whereas overexpres- interact with the C-terminal region of SIP (Filipek et al., sion of PKCe in MCF-7 cells downregulates p53 and 2002a; Bhattacharya et al., 2005) and some of the inhibits TRAIL-induced apoptosis (Shankar et al., members, such as S100A1, S100A2, S100A6 and 2008). In addition to PKC, KinasePhos also predicts S100B, are involved in apoptosis regulation, and many Cdk1 as the potential kinase for Ser141 phosphoryla- S100 proteins display cellular translocation in response tion. Interestingly, our previous study shows that to various stimuli (Hsieh et al., 2002; Orre et al., curcumin treatment results in Cdk1 inactivation in 2007; Joo et al., 2008). Besides, SIP has been found to curcumin-resistant but not -sensitive cells (Wang et al., interact with ERK1/2 and downregulate ERK activity 2008). Therefore, the effect of PKC and Cdk1 in (Kilanczyk et al., 2009). Furthermore, suppression of SIP-mediated apoptosis is worth to be explored in the ERK-involved pathway sensitizes cancer cells to the future. apoptosis (Tran et al., 2001; Wu et al., 2004) and a series Notably, both theoretical estimation and immuno- of inhibitors targeting ERK pathway have been blotting analysis reveal that the molecular weight of SIP identified as potential anti-cancer drugs (Daouti et al., is around 30 kDa. As we know, small proteins (MW 2009). It has also been shown that SIP serves as a o40 kDa) may, in principle, diffuse passively through molecular bridge between Siah-1 and Skp1 proteins, and the nuclear pores (Lange et al., 2007). However, it has forms a ubiquitin ligase complex together with Ebi been shown that SIP can self-dimerize through the and in turn promotes b-catenin degradation. b-Catenin N-terminal aa 1–47 (Santelli et al., 2005). In that case, is a widely studied oncogene and is able to impel the resulting protein dimer may become too large to the transcription of regulators of cell proliferation diffuse passively into the nucleus. On the other hand, the and apoptosis, such as cyclin D1, c-Myc and survivin nuclear import of the exogenously expressed SIP-S (Jin et al., 2008; Takigawa and Brown, 2008). Taken protein, which has the truncation of N-terminal aa 1–43 together, we speculate that SIP may exert its function on within the self-dimerization domain, is significantly apoptotic signaling by interacting with other proteins. attenuated when its NLS is deleted or Ser141 is Further efforts to identify these proteins and their substituted with alanine. These data indicate that the function in SIP-mediated apoptotic signaling are surely nuclear transport of SIP is applied to precise regulation warranted. rather than passive diffusion. The above results also In summary, we investigate the role of SIP in emphasize that the NLS domain and the Ser141 residue, apoptotic signaling and its underlying mechanisms, as well as the N-terminal region, have important roles in and reveal that defects in SIP phosphorylation and cellular translocation of SIP and may have further nuclear translocation may contribute to the reduced cell impact on SIP-mediated apoptotic signaling. sensitivity to apoptosis stimuli. Our data provide new The significant correlation between nuclear accumu- insights into the biological function of SIP protein and lation of SIP and its proapoptotic activity further the potential mechanisms of curcumin resistance. implicate that the nuclear import of SIP is an important step rather than an accompanying phenomenon during the apoptosis process, and impaired translocation of SIP Materials and methods may render cells resistant to apoptotic stimuli. Interest- ingly, we also notice that the apoptotic rates are similar Detailed description on the reagents and experimental between SIP-WT- and WT-S141A-transfected HepG2 procedures can be found in the Supplementary Methods. cells treated with etoposide or in BEL-7404 cells exposed to curcumin (Figure 6), although SIP-WT shows much Cell lines higher nuclear translocation efficiency than WT-S141A. MOLT-4/TG and MOLT-4/AraG (generous gifts from These discrepancies are likely due to the difference in Dr Curt Peterson, Linko¨ping University, Linko¨ping, Sweden) cell context and the types of stimuli. were derived from human acute lymphocytic leukemia cell Yang et al. (2006) report that antisense oligonucleo- lines MOLT-4 by long-time exposure to low doses of TG and tides against SIP significantly inhibit apoptosis in Ara-G, respectively (Lotfi et al., 2002; Mansson et al., 2002). endometrial stromal cells 12 h after ultraviolet irradia- The IC50 values of TG and Ara-G for these cell lines are shown tion. Consistently, SIP knockdown by RNAi attenuates in Supplementary Table S2. apoptosis triggered by curcumin at the 12-h time point. Stable SIP-knockdown HepG2 cell lines (G2-659 and G2- Intriguingly, the apoptotic rates of SIP-knockdown cells 1170) and their control line (G2-NC) were generated using retroviral vectors pSR-puro-sh659, pSR-puro-sh1170 and increase to a similar level as those of control cells at 24 h pSR-puro-NC, respectively. and become even higher than control cells at 48 h after MOLT-4 and its derivatives were grown in RPMI 1640 curcumin exposure (data not shown). These data may (Gibco Invitrogen, Carlsbad, CA, USA), and human hepato- suggest that SIP has an important role in the initiation ma cell lines (HepG2, Hep3B, QGY-7703 and BEL-7404) were step of curcumin-induced apoptosis, whereas inhibition grown in DMEM (Gibco Invitrogen), supplemented with 10%

Oncogene Siah-interacting protein in apoptosis J Luo et al 6365 FBS (PAA, Haidmannweg, Austria) and 2 mML-glutamine searching in the publicly available database NCBInr or (Sigma, St Louis, MO, USA). SWISS-PROT using Profound search engine (Amersham).

Plasmids, siRNA duplexes and cell transfection Western blotting and immunoprecipitation All primer sequences used for plasmid construction and For western blotting, proteins were separated by sodium dodecyl siRNA sequences are shown in Supplementary Table S3. sulfate–polyacrylamide gel electrophoresis (8–15%), electrophor- The plasmids used in this study included the following. The etically transferred to 0.22 mm Immobilon-P PVDF membranes prokaryotic expression vector pET-28b-SIP that expressed a (Millipore, Billerica, MA, USA), and detected by speciﬁc His-tagged SIP was used to immunize rabbits and produce antibodies and an ECL kit (Pierce, Rockford, IL, USA). anti-SIP; p3FH-SIP that expressed Flag- and HA-tagged SIP For immunoprecipitation assay, cells were lysed in protein was used to verify the effectiveness of anti-SIP in immuno- extraction buffer (20 mM Tris–HCl (pH 7.5), 150 mM NaCl, 8 mM blotting and immunoprecipitation; the eukaryotic expression MgCl2 and 1% NP-40) containing Complete Mini EDTA-free vectors of His or EGFP-tagged SIP and its mutant forms, protease inhibitor cocktail and PhosSTOP phosphatase inhibitor constructed based on pcDNA3.0 (Invitrogen, Seattle, WA, cocktail (Roche, Mannheim, Germany). Extracts were incubated USA), were used to analyze the SIP cellular localization and with anti-HA or anti-SIP and protein A/G plus beads at 4 1Cfor their effect on apoptosis; the pSR-puro-sh659 and pSR-puro- 2 h. After three washes with buffer containing 20 mM Tris–HCl sh1170 that expressed shRNA were used to target the coding (pH 7.5), 150 mM NaCl and 0.1% NP-40, the immune complexes sequence (659–677 nt, NM_014412.2) and the 3-untranslated were applied to immunoblotting. region (1170–1188 nt) of SIP, respectively; the pSR-puro-NC vector, a generous gift from Dr Zeng MS (Cancer Center, Sun RNA extraction and semiquantitative RT–PCR Yat-sen University, Guangzhou, PR China), expressed the Total RNA from cells was extracted using Trizol (Invitrogen). scrambled shRNA and was used as a negative control. To obtain cDNA, 2 mg of total RNAs was reversely Lipofectamine and PLUS reagent (Invitrogen) were used for transcribed using Moloney murine leukemia virus reverse plasmid transfection and Lipofectamine RNAiMAX (Invitro- transcriptase (Promega). Speciﬁc primers used for RT–PCR gen) for the transfection of siRNA duplex. are listed in Supplementary Table S3.

Preparation of antisera against human SIP protein Analysis of subcellular localization of SIP To generate rabbit polyclonal antisera against SIP, the His- To detect endogenous SIP or exogenous His-tagged SIP, tagged SIP protein was expressed in Escherichia coli strain immunostaining was performed as reported (Wang et al., BL21 (Novagen, Madison, WI, USA) using pET-28b-SIP, 2008), using anti-SIP or anti-His as primary antibody and then purified with HisBind Purification Kits (Novagen) and goat-anti-rabbit or mouse Alexa-488 conjugated secondary applied to immunize rabbits. The male white New Zealand antibody. The nucleus was counterstained with DAPI or PI. rabbit was injected with 1 mg His-tagged SIP protein All images were captured under a fluorescence microscope. emulsified with complete Freund’s adjuvant (Sigma). Three Cells with green fluorescence in nucleus were defined as nuclear booster injections with 500, 250 and 125 mg protein emulsified localization of SIP. with incomplete Freund’s adjuvant (Sigma) were then conducted at 2-week intervals. Antisera were collected 1 week after Measurement of mitochondrial membrane potential the last immunization. Cells were stained with both MitoTracker Deep Red FM and MitoTracker Green FM (Invitrogen) at 37 1C for 20 min in the Detection of apoptosis dark and washed in Ca2 þ -free phosphate buffered saline, Apoptosis was evaluated by morphological examination using followed by flow cytometry analysis. DAPI or PI staining, by flow cytometry analysis using Annexin V-PI staining, and by TUNEL assay using DeadEnd Statistical analysis Fluorometric TUNEL System (Promega, Madison, WI, USA). Values represent the mean±s.d. from at least three independent experiments. Differences between the experimental groups Two-dimensional gel electrophoresis-based comparative were analyzed using Student’s t-test. A value of Po0.05 was proteomics considered statistically significant. For comparative proteomic analysis, cells were treated with DMSO or 25 mM curcumin for 14 h. Proteins (600 mg) were applied to 24 cm Immobiline Drystrip (pH 3–10, linear Conflict of interest gradient; Amersham, GE Healthcare, Uppsala, Sweden) for the first dimensional isoelectric focusing, and then separated The authors declare no conflict of interest. by sodium dodecyl sulfate-polyacrylamide gel electrophoresis using EttanTM DALTsix system (Amersham). Gels from three independent experiments were analyzed using ImageMaster Acknowledgements 2D Platinum software 5.0 (Amersham). Only spots that showed significant and reproducible changes in the intensity This study was supported by grants from the National Natural (ratio X1.5, Po0.05) were selected for protein identification. Science Foundation of China (30470664), Ministry of Science Peptide mass fingerprint of protein was obtained using Ettan and Technology of China (2010CB912803, 2005CB724600) MALDI-TOF Pro (Amersham). Proteins were identified by and Ministry of Education of China (105136).

References

Aggarwal BB, Shishodia S, Takada Y, Banerjee S, Newman RA, the paclitaxel-induced nuclear factor-kappaB pathway in Bueso-Ramos CE et al. (2005). Curcumin suppresses breast cancer cells and inhibits lung metastasis of

Oncogene Siah-interacting protein in apoptosis J Luo et al 6366 human breast cancer in nude mice. Clin Cancer Res 11: Lange A, Mills RE, Lange CJ, Stewart M, Devine SE, Corbett AH. 7490–7498. (2007). Classical nuclear localization signals: definition, function, Bhandarkar SS, Arbiser JL. (2007). Curcumin as an inhibitor of and interaction with importin alpha. J Biol Chem 282: 5101–5105. angiogenesis. Adv Exp Med Biol 595: 185–195. Li N, Chen X, Liao J, Yang G, Wang S, Josephson Y et al. (2002). Bhattacharya S, Lee YT, Michowski W, Jastrzebska B, Filipek A, Inhibition of 7, 12-dimethylbenz[a]anthracene (DMBA)-induced Kuznicki J et al. (2005). The modular structure of SIP facilitates its role oral carcinogenesis in hamsters by tea and curcumin. Carcinogenesis in stabilizing multiprotein assemblies. Biochemistry 44: 9462–9471. 23: 1307–1313. Blom N, Gammeltoft S, Brunak S. (1999). Sequence and structure- Lotfi K, Mansson E, Peterson C, Eriksson S, Albertioni F. (2002). Low based prediction of eukaryotic protein phosphorylation sites. JMol level of mitochondrial deoxyguanosine kinase is the dominant factor Biol 294: 1351–1362. in acquired resistance to 9-beta-D-arabinofuranosylguanine cyto- Blom N, Sicheritz-Ponteń T, Gupta R, Gammeltoft S, Brunak S. toxicity. Biochem Biophys Res Commun 293: 1489–1496. (2004). Prediction of post-translational glycosylation and phosphor- Mahmmoud YA. (2007). Modulation of protein kinase C by curcumin; ylation of proteins from the amino acid sequence. Proteomics 4: inhibition and activation switched by calcium ions. Br J Pharmacol 1633–1649. 150: 200–208. Daouti S, Wang H, Li WH, Higgins B, Kolinsky K, Packman K et al. Mansson E, Stridh H, Albertioni F. (2002). Resistance to mitochon- (2009). Characterization of a novel mitogen-activated protein kinase drial- and Fas-mediated apoptosis in human leukemic cells with kinase 1/2 inhibitor with a unique mechanism of action for cancer acquired resistance to 9-beta-D-arabinofuranosylguanosine. Bio- therapy. Cancer Res 69: 1924–1932. chem Biophys Res Commun 298: 338–344. Dhillon N, Aggarwal BB, Newman RA, Wolff RA, Kunnumakkara AB, Matsuzawa SI, Reed JC. (2001). Siah-1, SIP, and Ebi collaborate in a Abbruzzese JL et al. (2008). Phase II trial of curcumin in patients with novel pathway for beta-catenin degradation linked to p53 responses. advanced pancreatic cancer. Clin Cancer Res 14: 4491–4499. Mol Cell 7: 915–926. Filipek A, Jastrzebska B, Nowotny M, Kuznicki J. (2002a). CacyBP/ Nakai K, Horton P. (1999). PSORT: a program for detecting sorting SIP, a calcyclin and Siah-1-interacting protein, binds EF-hand signals in proteins and predicting their subcellular localization. proteins of the S100 family. J Biol Chem 277: 28848–28852. Trends Biochem Sci 24: 34–36. Filipek A, Jastrzebska B, Nowotny M, Kwiatkowska K, Hetman M, Ning X, Sun S, Hong L, Liang J, Liu L, Han S et al. (2007). Calcyclin- Surmacz L et al. (2002b). Ca2+-dependent translocation of the binding protein inhibits proliferation, tumorigenicity, and invasion calcyclin-binding protein in neurons and neuroblastoma NB-2a of gastric cancer. Mol Cancer Res 5: 1254–1262. cells. J Biol Chem 277: 21103–21109. O’Connor R. (2009). A review of mechanisms of circumvention and Filipek A, Wojda U. (1996). p30, a novel protein target of mouse modulation of chemotherapeutic drug resistance. Curr Cancer Drug calcyclin (S100A6). Biochem J 320: 585–587. Targets 9: 273–280. Gutcher I, Webb PR, Anderson NG. (2003). The isoform-specific Orre LM, Pernemalm M, Lengqvist J, Lewensohn R, Lehtio J. (2007). regulation of apoptosis by protein kinase C. Cell Mol Life Sci 60: Up-regulation, modification, and translocation of S100A6 induced 1061–1070. by exposure to ionizing radiation revealed by proteomics profiling. Hsieh HL, Schafer BW, Cox JA, Heizmann CW. (2002). S100A13 and Mol Cell Proteomics 6: 2122–2131. S100A6 exhibit distinct translocation pathways in endothelial cells. Santelli E, Leone M, Li C, Fukushima T, Preece NE, Olson AJ et al. J Cell Sci 115: 3149–3158. (2005). Structural analysis of Siah1-Siah-interacting protein inter- Huang HD, Lee TY, Tzeng SW, Horng JT. (2005). KinasePhos: a web actions and insights into the assembly of an E3 ligase multiprotein tool for identifying protein kinase-specific phosphorylation sites. complex. J Biol Chem 280: 34278–34287. Nucleic Acids Res 33: W226–229. Shankar E, Sivaprasad U, Basu A. (2008). Protein kinase C epsilon confers Iakoucheva LM, Radivojac P, Brown CJ, O’Connor TR, Sikes JG, resistance of MCF-7 cells to TRAIL by Akt-dependent activation of Obradovic Z et al. (2004). The importance of intrinsic disorder for Hdm2 and downregulation of p53. Oncogene 27: 3957–3966. protein phosphorylation. Nucleic Acids Res 32: 1037–1049. Sharma RA, Euden SA, Platton SL, Cooke DN, Shafayat A, Hewitt Jans DA, Hubner S. (1996). Regulation of protein transport to the HR et al. (2004). Phase I clinical trial of oral curcumin: biomarkers nucleus: central role of phosphorylation. Physiol Rev 76: 651–685. of systemic activity and compliance. Clin Cancer Res 10: 6847–6854. Jin T, George Fantus I, Sun J. (2008). Wnt and beyond Wnt: multiple Sun S, Ning X, Liu J, Liu L, Chen Y, Han S et al. (2007). mechanisms control the transcriptional property of beta-catenin. Overexpressed CacyBP/SIP leads to the suppression of growth in Cell Signal 20: 1697–1704. renal cell carcinoma. Biochem Biophys Res Commun 356: 864–871. Joo JH, Yoon SY, Kim JH, Paik SG, Min SR, Lim JS et al. (2008). Takigawa Y, Brown AM. (2008). Wnt signaling in liver cancer. Curr S100A6 (calcyclin) enhances the sensitivity to apoptosis via the Drug Targets 9: 1013–1024. upregulation of caspase-3 activity in Hep3B cells. J Cell Biochem Tran SE, Holmstrom TH, Ahonen M, Kahari VM, Eriksson JE. 103: 1183–1197. (2001). MAPK/ERK overrides the apoptotic signaling from Fas, Karunagaran D, Rashmi R, Kumar TR. (2005). Induction of TNF, and TRAIL receptors. J Biol Chem 276: 16484–16490. apoptosis by curcumin and its implications for cancer therapy. Wang WZ, Cheng J, Luo J, Zhuang SM. (2008). Abrogation of G2/M Curr Cancer Drug Targets 5: 117–129. arrest sensitizes curcumin-resistant hepatoma cells to apoptosis. Kawamori T, Lubet R, Steele VE, Kelloff GJ, Kaskey RB, Rao CV FEBS Lett 582: 2689–2695. et al. (1999). Chemopreventive effect of curcumin, a naturally Wu J, Tan X, Peng X, Yuan J, Qiang B. (2003). Translocation and occurring anti-inflammatory agent, during the promotion/progres- phosphorylation of calcyclin binding protein during retinoic acid- sion stages of colon cancer. Cancer Res 59: 597–601. induced neuronal differentiation of neuroblastoma SH-SY5Y cells. Kilanczyk E, Filipek S, Jastrzebska B, Filipek A. (2009). CacyBP/SIP J Biochem Mol Biol 36: 354–358. binds ERK1/2 and affects transcriptional activity of Elk-1. Biochem Wu J, Wong WW, Khosravi F, Minden MD, Penn LZ. (2004). Blocking Biophys Res Commun 380: 54–59. the Raf/MEK/ERK pathway sensitizes acute myelogenous leukemia Knauf JA, Elisei R, Mochly-Rosen D, Liron T, Chen XN, Gonsky R cells to lovastatin-induced apoptosis. Cancer Res 64: 6461–6468. et al. (1999). Involvement of protein kinase cepsilon (PKCepsilon) Yang YJ, Liu WM, Zhou JX, Cao YJ, Li J, Peng S et al. (2006). in thyroid cell death. A truncated chimeric PKCepsilon cloned from Expression and hormonal regulation of calcyclin-binding protein a thyroid cancer cell line protects thyroid cells from apoptosis. J Biol (CacyBP) in the mouse uterus during early pregnancy. Life Sci 78: Chem 274: 23414–23425. 753–760.

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

Oncogene