An RNF11: Smurf2 complex mediates ubiquitination of the ... - Nature

Oncogene (2004) 23, 1801–1808

& 2004 Nature Publishing Group All rights reserved 0950-9232/04 $25.00 www.nature.com/onc

ORIGINAL PAPERS

An RNF11: Smurf2 complex mediates ubiquitination of the AMSH protein Haoxia Li1 and Arun Seth1,2,3,* 1

Molecular and Cellular Biology Research; 2Laboratory of Molecular Pathology, Sunnybrook and Women’s College Health Sciences Centre; 3Department of Laboratory Medicine and Pathobiology, University of Toronto, Toronto, Ontario, Canada

RING-finger proteins play crucial roles in ubiquitination events involved in diverse cellular processes including signal transduction, differentiation and apoptosis. Most of the RING-finger proteins have E3-ubiquitin ligase activity. RNF11 is a small RING-finger protein and harbors a RING-H2 domain and a PY motif that could facilitate protein:protein interaction(s) involved in oncogenesis. To isolate RNF11 protein partners and determine its role in normal and cancer cells, we performed yeast two-hybrid screening. Among 18 in-frame positive clones, three were found to be ZBRK1, Eps15 and AMSH (associated molecule with the SH3 domain of STAM). ZBRK1 is a KRAB domain containing Zinc-finger protein and is known to repress target gene transcription in a BRCA1dependent manner. Eps15 is monoubiquitinated and is part of an essential complex involved in the endocytosis of plasma membrane receptors via the clathrin-mediated internalization pathway. Recent studies have shown that AMSH protein is involved in BMP/TGF-b signaling pathway by binding to Smad6 and Smad7. The association of RNF11 with these binding partners suggests that it would be involved in biological processes such as gene transcription, BMP/TGF-b signaling and ubiquitinationassociated events. Previously, we have shown that RNF11 interacts with the HECT-type E3 ligases AIP4 and Smurf2. Here, we show that RNF11 binds to AMSH in mammalian cells and that this interaction is independent of the RNF11 RING-finger domain and the PY motif. Our results also demonstrate that AMSH is ubiquitinated by Smurf2 E3 ligase in the presence of RNF11 and that a consequent reduction in its steady-state level requires both RNF11 and Smurf2. RNF11 therefore recruits AMSH to Smurf2 for ubiquitination, leading to its degradation by the 26S proteasome. The potential functions of RNF11mediated degradation of AMSH in breast cancer are discussed. Oncogene (2004) 23, 1801–1808. doi:10.1038/sj.onc.1207319 Published online 2 February 2004 Keywords: RNF11; AMSH; Smurf2; TGF-b; breast cancer; RING-H2-finger domain; PY motif

*Correspondence: A Seth, Laboratory of Molecular Pathology, Rm E-423b, Sunnybrook and Women’s College Health Sciences Centre, 2075 Bayview Avenue, Toronto, Ontario, Canada M4N 3M5; E-mail: [email protected] Received 16 July 2003; revised 19 August 2003; accepted 24 October 2003

Introduction There is considerable epidemiologic, pathologic and molecular evidence to suggest that breast cancer progresses in a stepwise manner through a series of precursor stages to the development of invasive carcinoma and metastases to distant sites. In order to define genes that are essential to the initiation and development of the sporadic form of breast cancer, we isolated 41000 genes that may be activated or inactivated during progression of breast cancer. This was accomplished by subtractive hybridization and differential display cloning methods using matched breast tumor and normal breast cell-line RNAs (Burger et al., 1998). One of these genes encodes the RING-finger protein RNF11 (Kitching et al., 2003). Tissue microarray technology allowed for a rapid qand simultaneous immunohistochemical analysis of RNF11 protein expression in 125 primary tumors, wherein RNF11 was found to be markedly overexpressed in breast and to a somewhat lesser extent in pancreatic and colon cancers and only weakly expressed in prostate and renal cell carcinomas (Subramaniam et al., 2003). The RNF11 gene encodes an ORF of 154 amino acids, which contains modular domains and motifs known to interact with other proteins involved in cell signaling (Kitching et al., 2003). RNF11 harbors a RING-H2-finger domain that could facilitate protein: protein interaction(s) involved in ubiquitin-mediated proteolysis and a PY motif that could bind to WWdomain proteins, several of which are known to be HECT-type E3 ligases (Kitching et al., 2003; Subramaniam et al., 2003). HECT-type E3 ligases are a family of proteins that includes E6-AP, NEDD4, Smurf1 and Smurf2 (Weissman, 2001). Each is characterized by the HECT domain, a 350 residue region located near the carboxy-terminus containing an essential cysteine residue capable of forming a thiol-ester intermediate with ubiquitin (Nuber and Scheffner, 1999). Although E6AP lacks WW domains, they are present in NEDD4, AIP4, Smurf1 and Smurf2 E3 ligases. Within the WW domain, conserved tryptophan residues form a hydrophobic pocket capable of binding to the proline-rich sequences known as the PY motif (Kay et al., 2000; Rotin et al., 2000). Smad proteins (Smad2, -3 and -7) also contain PY motifs and act as substrates as well as adaptors for Smurf2 E3-ubiquitin ligase and mediate the targeted

Ubiquitination of the AMSH protein H Li and A Seth

1802

destruction of cellular proteins. (Zhu et al., 1999; Bonni et al., 2001; Zhang et al., 2001). Recently, we found that WW domains of Smurf2 and AIP4 interact with the PY motif of RNF11 in mammalian cells (Kitching et al., 2003; Subramaniam et al., 2003). There is a possibility that the RNF11 protein could recruit its binding partners to these E3 ligases for ubiquitination and subsequent degradation. In order to discover such partner proteins of RNF11, we performed yeast two-hybrid screening using human ovary cDNA library. Among 18 positive clones, we isolated a gene that encodes AMSH (associated molecule with the SH3 domain of STAM), a protein involved in BMP/TGF-b signaling through interactions with Smad6 and Smad7 (Tanaka et al., 1999; Ishii et al., 2001; Itoh et al., 2001). AMSH was first cloned as a novel STAM-associated molecule involved in intracellular signal transduction mediated by cytokines (Tanaka et al., 1999). AMSH inhibits the antagonistic effects of Smad6 on the BMP–Smad pathway and its expression level plays a critical role in determining the potency of BMP-induced responses (Itoh et al., 2001). Furthermore, AMSH can interact with Smad7 and also affect the TGF-b signaling pathway. In the present study, we have investigated interactions between Smurf2, RNF11 and AMSH in mammalian cells using mutants of the RNF11 PY motif and its RING-finger domain. In addition, the effects of these interactions on ubiquitination and TGF-b-induced transcription activation were also examined.

Results Isolation of RNF11 partner proteins To gain an insight into the mechanisms of RNF11 action in normal and cancer cells, we used a yeast twohybrid system to identify interacting proteins. The wildtype (wt) RNF11 could not be used as a bait because it yielded high background reporter activity when fused to the Gal4 DNA-binding domain. To circumvent this problem, we constructed three RNF11 mutants: (1) RNF11 mtPY, in which tyrosine 40 in the PY motif was mutated to alanine, (2) RNF11 mtRing, in which cysteines 89 and 93 in the RING-finger domain were mutated to serine residues and (3) RNF11 mtDouble, which contained both the PY and the RING-finger domain mutations. The RING and the double mutant did not show background activity, thus we chose the mtRing construct as the bait in our yeast two-hybrid

screen of a human ovary cDNA library. In all, 98 positive clones were sequenced revealing 18 clones to be in-frame with the Gal4 DNA-binding domain. Among the positive clones, three binding partners were detected more than once, thus representing 15 unique genes. Clones bioinformatically identified as known proteins included ubiquitination pathway-related proteins such as UbC, RPS27 and EPS15, which is monoubiquitinated and involved in protein trafficking, and the transcription repressor protein ZBRK1 (Table 1). AMSH, a protein involved in BMP/TGFb signaling, was also isolated as an RNF11 partner. AMSH and all the other 14 binding partners showed specific interactions with RNF11 by genetic selection for yeast bearing both plasmids on low and high stringency plates (Figure 1). RNF11 and its mutants bind AMSH To verify that AMSH binds wt RNF11 in mammalian cells and to investigate whether the PY motif is responsible for such an association, HEK293T cells were transiently cotransfected with glutathione S-transferase (GST)-AMSH and Flag-RNF11 expression vectors. Immunoprecipitation of cell lysates with Flag antibody and subsequent Western blotting using GST antibody revealed that GST-AMSH coimmunoprecipitated with Flag-RNF11 but not with control GST product, suggesting that AMSH binds to RNF11 protein (Figure 2a, upper panel). We also investigated the binding of AMSH to the RNF11mtRing and RNF11mtDouble proteins (Figure 2b). HEK293T cells were also cotransfected with HA-AMSH in combination with the GST-RNF11mtRing or the GSTRNF11mtDouble or control GST expression vectors. GST pull-down and Western blot analysis revealed that HA-AMSH interacts with both mutants of RNF11, but not with the GST protein alone (Figure 2b). RNF11 recruits AMSH to Smurf2 E3 ligase Both wt RNF11 and the RING domain mutant of RNF11 interact with AMSH (Figure 2), suggesting that the PY motif of RNF11 is not responsible for binding to AMSH. On the other hand, RNF11 binds Smurf2 through this PY motif raising the possibility that RNF11 acts as an adapter protein bringing AMSH and Smurf2 together (Subramaniam et al., 2003). To test this hypothesis, lysates from HEK293T cells cotransfected with GST-RNF11, GST-AMSH and FlagSmurf2 were analysed by immunoprecipitation and immunoblotting (Figure 3). Anti-GST immunoblotting

Table 1 Positive clones obtained from yeast two-hybrid screen of RNF11 Name AMSH – associated molecule with the SH3 domain of STAM EPS15 – epidermal growth factor receptor pathway substrate 15 RPS27A – ribosomal protein S27a UbC – ubiquitin C ZBRK1 – Zinc-finger protein ZBRK1 Oncogene

Unigene

Function

Hs.12479 Hs.79095 Hs.311640 Hs.183704 Hs.130965

TGF-b family signal transduction Monoubiquitinated during receptor-mediated endocytosis Carboxyl extension processed to ubiquitin and RPS27 Proteasomal degradation Nuclear complex with BRCA1


1803

Figure 1 Activation of the ADE2 and HIS3 reporter genes by the interaction of RNF11mtRing with AMSH. Separate quarters of growth plates were streaked with yeast strains derived from AH109 cells transformed with the following pairs of plasmids: pGBKT7RNF11mtRing with pACT2-AMSH; pGBKT7-RNF11mtRing with the prey vector pACT2; the bait vector pGBKT7 with pACT2AMSH; pGBKT7 with pACT2. -Trp-Leu: medium without Trp and Leu. -Trp -Leu, -His: low stringency plate lacking -Trp, -Leu and -His. -Trp -Leu, -His -Ade: high stringency plate lacking Trp, Leu, His and Ade

of lysates from cells transfected with AMSH and the RNF11 GST-fusion protein expression vectors indicated that GST-AMSH and GST-RNF11 were present where expected (Figure 3a). Anti-Flag immunoblotting showed that the Flag-Smurf2 protein was in lysates from cells transfected with the Smurf2 expression vector (Figure 3b, lower panel). As also reported elsewhere, we found that GST-RNF11 coimmunoprecipitates with Flag-Smurf2 from lysates of cells transfected with expression vectors for these two proteins (Figure 3b, upper panel, lane 4) (Subramaniam et al., 2003). In addition, Flag-Smurf2 is also able to coimmunoprecipitate GST-AMSH in the presence of GST-RNF11 (Figure 3b, upper panel, lane 1). A faint band that comigrates with GST-AMSH in lane 3 could either be a result of weak interaction with Smurf2 protein or nonspecific interactions with the GST beads. Both RNF11 and Smurf2 regulate AMSH steady-state levels

Figure 2 (a) RNF11 interacts with AMSH in mammalian cells. Proteins immunoprecipitated by anti-Flag antibody from lysates of HEK293T cells cotransfected with both Flag-RNF11 and GSTAMSH or Flag-RNF11 alone were detected by anti-GST immunoblotting. Also shown is an anti-Flag immunoblot of the lysates used for anti-Flag immunoprecipitation. (b) AMSH interacts with mutants of RNF11. Proteins isolated by GSTpulldown from lysates of HEK293T cells cotransfected with HA-AMSH and empty GST vector or GST-RNF11mtRing or GST-RNF11mtDouble were detected by anti-HA immunoblot

AMSH protein has been reported to be involved in the IL-2 and TGF-b signaling pathway (Tanaka et al., 1999; Ishii et al., 2001; Itoh et al., 2001). However, mechanisms for AMSH protein regulation are not yet understood. Smurf2 is an E3-ubiquitin ligase that catalyses the ubiquitination of target protein substrates destined for ubiquitin-mediated proteolysis. Since RNF11 binds both Smurf2 and AMSH proteins, we reasoned that RNF11 could recruit AMSH to Smurf2 for ubiquitination and subsequent degradation by the 26S proteasome. To test this hypothesis, AMSH was overexpressed with RNF11 or with Smurf2 or with both RNF11 and Oncogene


1804

Figure 4 RNF11 and Smurf2 alter the steady-state levels of AMSH. HEK293T cells were cotransfected with various combinations of GST-AMSH, Flag-Smurf2 and Flag-RNF11 vectors (Material and Methods). Steady-state levels of AMSH in the presence or absence of RNF11 and Smurf2 were examined by immunoblotting using indicated antibodies

Figure 3 RNF11, AMSH and Smurf2 form a trimeric complex. (a) GST-fusion proteins in lysates of HEK293T cells cotransfected with combinations of expression vectors for GST-RNF11, GSTAMSH, Flag-Smurf2 and the empty vectors were detected by anti-GST immunoblotting. (b) Proteins immunoprecipitated by anti-Flag antibody from equal amounts of lysates of the same HEK293T cells cotransfected with combinations of expression vectors for GST-RNF11, GST-AMSH, Flag-Smurf2 and the empty vectors were detected by anti-GST immunoblots. Also shown is an anti-Flag immunoblot of the lysates from the same cells transfected in panel A

Smurf2 (Figure 4). Anti-GST immunoblots of total cell lysates revealed a significant reduction of AMSH steadystate levels in the presence of RNF11 and Smurf2 in comparison to the lanes in which AMSH was cotransfected with either RNF11 or Smurf2 alone. Smurf2 promotes ubiquitination of AMSH in the presence of wt RNF11 Interaction of the RNF11 with Smurf2 and AMSH proteins and the reduction of AMSH protein levels in the presence of Smurf2 and RNF11 suggested that Oncogene

AMSH could be ubiquitinated by the Smurf2:RNF11 complex (Figures 3 and 4). To investigate this notion, HEK293T cells were cotransfected with GST-AMSH, GST-RNF11 or its mutants and Flag-Smurf2 or mutant Flag-Smurf2 together with HA-tagged ubiquitin. Whole-cell lysates were subjected to HA and Flag immunoprecipitation and GST immunoblotting to detect ubiquitination of AMSH protein (Figure 5). Flag immunoprecipitation followed by GST immunoblotting resulted in coimmunoprecipitation of proteins corresponding to ubiquitinated and unubiquitinated GSTRNF11 and AMSH from lysates of transfected cells (Figure 5c). Similar to the negative control lacking RNF11 (Figure 5c lane4), the RNF11mtPY mutant (Figure 5c lane2) resulted in very weak immunoprecipitation of GST-AMSH protein by Smurf2. Polyubiquitinated GST-AMSH bands were revealed by GST immunoblots of proteins immunoprecipitated with anti-HA antibodies (Figure 5a, lanes 1 and 5f, lane 5), but not when mutant Smurf2 was used (Figure 5f, lane 2). However, a band representing presumably monoubiquitinated GST-AMSH protein can be seen with and without exogenously expressed Smurf2 and RNF11 in Figure 5a (lanes 1–5 and 5F, lanes 1–5). This ubiquitinated band could arise by the interaction of GST-AMSH with endogenous Smurf2 and RNF11 proteins or an alternative pathway for ubiquitination of AMSH may exist. However, strong polyubiquitinated bands are seen only when both wt RNF11 and wt Smurf2 are present (Figure 5a, lane 1 and Figure 5f, lane 5). Similar polyubiquitinated AMSH bands are not


1805

depends upon additional interaction between the wt RING domain and UBCH5c E2 (Subramaniam et al., 2003). The consequences of this for ubiquitination of AMSH are seen in Figure 5c (lane 3), where although a large amount of GST-AMSH was immunoprecipitated with Smurf2, very little polyubiquitination of GSTAMSH is seen (Figure 5a, lane 3). In addition, strong Smurf2 bands in lanes 1 and 6 of Figure 5b indicate that wt RNF11 mediates the activation of the Smurf2 ligase activity because ubiquitination of Smurf2 is observed only when both of these proteins are present (Figure 5). The GST immunoblot of whole-cell lysates revealed that cotransfection with the Flag-Smurf2 construct reduced the amount of GST-AMSH available for immunoprecipitation (Figure 5e, lanes 1–4). The strong GSTAMSH band from cell lysates lacking the Flag-Smurf2 protein could thus be a result of reduced degradation in the absence of Smurf2 (Figure 5e, lane 5). Taken together, our data suggest that RNF11 interaction with Smurf2 is a major factor in the polyubiquitination of AMSH protein. RNF11 interferes with the AMSH-induced transcription from a TGF-b-responsive promoter The effect of interactions between AMSH, RNF11 and Smurf2 on TGF-b-dependent transcription activation was investigated by transient cotransfection with 3TPLux promoter–reporter plasmids (Figure 6). The 3TPLux reporter vector contains a TGF-b-responsive portion of the PAI promoter linked to the coding region of the firefly luciferase gene. The PAI promoter is highly active in TGF-b-treated cells and inactive in the absence of TGFb (Figure 6). Interestingly, reporter gene activity is increased when AMSH is included in the cotransfection. This effect is reduced when both AMSH and RNF11 constructs are present possibly due to the degradation of AMSH induced by RNF11. Inclusion of Smurf2 with AMSH and RNF11 reduced the activation

Figure 5 RNF11 recruits AMSH to Smurf2 for ubiquitination. HEK293T cells were cotransfected with combinations of GSTAMSH, GST-RNF11, GST-RNF11 mutants and Flag-Smurf2 or Flag-mtSmurf2 expression vectors along with HA-Ub plasmid. Aliquots of cell lysates were immunoprecipitated by anti-HA or anti-Flag antibodies, followed by immunoblotting with indicated antibodies to detect respective proteins

detected when mutant Smurf2 (Figure 5f, lane 2) or mutant RNF11 (Figure 5a, lane 2 and 3) are used. Although the RING domain is not required for interaction with Smurf2, ubiquitination of RNF11

Figure 6 AMSH and RNF11 affect TGF-b-responsive promoter expression. HepG2 cells grown with and without TGF-b were transiently transfected with p3TP-Lux, RNF11, Smurf2 and/or AMSH as indicated. Luciferase activity was normalized to b-gal activity and is expressed as the mean 7s.d. of triplicates from a representative experiment Oncogene


1806

of the reporter gene by 50%. In the absence of RNF11 and AMSH, Smurf2 inhibits the activation of the TGFb-responsive PAI. This inhibitory effect of Smurf2 on TGF-b responsiveness is consistently seen by us and others (Kavsak et al., 2000; Subramaniam et al., 2003). The partial relief of Smurf2 inhibition in the presence of RNF11 may be due to an AMSH-independent function of RNF11 reported elsewhere (Subramaniam et al., 2003). Discussion The regulation of protein degradation via the ubiquitin– proteasome pathway is a crucial mechanism by which cells modulate the expression levels of specific proteins involved in various cellular processes including proliferation, differentiation and apoptosis. Ubiquitinmediated proteolysis has critical roles in tumorigenesis through the regulation of cancer-associated proteins. Ubiquitination involves a highly specific enzymatic cascade in which ubiquitin moiety is initially activated by an E1 enzyme and subsequently coupled to substrates by either an E2 or E3 enzyme. E3-ubiquitin ligases play an essential role in recognizing the substrates for ubiquitination. There are two major classes of E3ubiquitin ligases, the RING type and the HECT type. The RING E3 ligases such as CBL and MDM2 proteins contain a RING-finger domain that interacts with E2 enzymes and a domain that binds to the substrate to be ubiquitinated. The HECT-type E3 ligases such as Smurf2, AIP4 and NEDD4 possess WW domains capable of binding PY motifs of substrates or adaptor proteins that recruit substrates to be ubiquitinated. The RNF11 protein is unique in that it contains both a RING-H2-finger domain and a PY motif that can potentially interact with both E2s and E3s. Indeed, we have reported that RNF11 interacts with UBCH5 E2s through its RING-H2-finger domain and the HECTtype E3 ligases such as NEDD4, AIP4 and Smurf2 via its PY motif (Kitching et al., 2003; Subramaniam et al., 2003). Previously, we have shown that RNF11 is overexpressed in breast and pancreatic cancers; in the present study in order to investigate its biological functions, we performed yeast two-hybrid screening to isolate and characterize its binding partners (Subramaniam et al., 2003). We obtained 18 positive clones including genes for proteins relevant to breast tumorigenesis, ubiquitination-related events and TGF-b/BMP signal transduction (Table 1). Among the positive clones, we isolated a gene encoding the AMSH openreading frame, recently recognized as a protein that binds directly to the TGF-b family of signaling molecules Smad6 and Smad7, thereby inhibiting the binding of I-Smads to activated type I receptors or activated R-Smads (Itoh et al., 2001). We demonstrated that AMSH binds to RNF11 with high affinity when expressed in HEK293T cells. We also showed that AMSH interacts with RNF11 mutants including RNF11mtRing and double mutant, indicating that it Oncogene

binds to AMSH neither by RING-finger domain nor by PY motif but a separate region. We found that the steady-state level of AMSH was decreased in the presence of both RNF11 and Smurf2, but not in the presence of RNF11 or Smurf2 alone (Figure 4). However, in Figure 5e (lane 4) Flag-Smurf2 appears to reduce the level of GST-AMSH even in the absence of GST-RNF11, perhaps due to endogenous RNF11 or other regulators of AMSH stability. This was not apparent in Figure 4, where we used a different expression vector. Furthermore, polyubiquitinated AMSH protein bands were observed in the presence of both RNF11 and Smurf2 compared to when either the RNF11 or Smurf2 was absent. No polyubiquitinated AMSH bands were detected in the presence of mutant Smurf2, which lacks E3 ligase activity. Moreover, the amount of polyubiquitinated GST-AMSH protein depends upon which mutants of RNF11 were cotransfected with the AMSH along with Smurf2 expression vectors. Thus, our data indicate that RNF11 recruits AMSH to Smurf2 for ubiquitination as modeled in Figure 7. AMSH protein has been shown to upregulate BMPinduced transcription and it is possible that the ubiquitination and degradation mediated by RNF11 may negatively affect BMP and/or TGF-b signaling (Itoh et al., 2001). AMSH interacts with Smad7 in cultured cells upon TGF-b receptor activation and augments TGF-b-induced luciferase reporter activity, suggesting that AMSH may regulate the TGF-b–Smad pathway in a similar way as the BMP–Smad pathway. Thus, we explored the effects of overexpressing both RNF11 and AMSH on TGF-b-responsive gene activity. AMSH was found to have a positive effect on the PAI promoter that is reduced by the overexpression of RNF11 and/or Smurf2 (Figure 6). This is consistent with our model, wherein a Smurf2/RNF11 complex leads to the ubiquitination and degradation of AMSH (Figures 5 and 7). We have speculated that protein partners of RNF11 would be targets of ubiquitination by the RNF11/ Smurf2 complex and could have roles in the development of breast and other cancers (Subramaniam et al., 2003). The interaction of RNF11 and AMSH may be involved in breast cancers by negating the positive effects of AMSH on TGF-b signaling leading to cell

Figure 7 RNF11 mediates the ubiquitination of AMSH protein


1807

proliferation and malignant progression. Indeed, we have found that AMSH mRNA is differentially expressed in breast cancer cell lines (data not shown) and overexpression of RNF11 protein in invasive breast cancers and pancreatic tumors (Subramaniam et al., 2003). In addition to AMSH, we found positive clones encoding ZBRK1, UbC, RPS27A and Eps15 open reading frames. ZBRK1 is a KRAB domain containing Zinc-finger protein that binds to BRCA1 protein, the quintessential hereditary breast and ovarian cancer susceptibility gene and is involved in coordinating the cellular DNA damage response (Welcsh et al., 2000; Zheng et al., 2000). One of the RNF11-binding proteins is UbC. In eukaryotes, a family of three genes (i.e. UbA, UbB and UbC) encodes protein products that are processed to yield free ubiquitin. Ubiquitin is a highly conserved 76 amino-acid polypeptide that covalently modifies various regulatory proteins such that substrates marked with a polyubiquitin chain are selectively targeted to the 26 proteasome for degradation. Ubiquitin is synthesized as precursor proteins that consist either of polyubiquitin chains or single ubiquitin moieties fused to unrelated carboxyl extension proteins (UbA type) such as UbA80 (ribosomal protein S27a, RPS27a), and we also found one such gene, RPS27, among the yeast two-hybrid clones that interact with RNF11. Eps15 is another gene that we isolated by RNF11 yeast two-hybrid cloning. Eps15, which was originally identified as a substrate of the EGF receptor tyrosine kinase, can also be phosphorylated following the activation of platelet-derived growth factor receptor and is involved in clathrin-mediated internalization of plasma membrane receptors (Carbone et al., 1997; Salcini et al., 1999; Sorkina et al., 1999; Bean et al., 2000). Interestingly, Nedd4 E3 and UbcH5 E2 have been shown to monoubiquitinate Eps15 protein. Eps15 forms a multivalent complex with Hrs and STAM proteins that sorts ubiquitinated proteins into the multivesicular body pathway (Bache et al., 2003). RNF11 also interacts with AMSH, a partner protein of STAM and mediates its ubiquitination. (Table 1, Figures 2–5) (Itoh et al., 2001). We have also reported that RNF11 interacts with both UbcH5c (an E2) and at least two different HECT-type E3 ligases leading to ubiquitination of substrates (Kitching et al., 2003; Subramaniam et al., 2003). Monoubiquitination of Eps15 and Hrs is necessary for the formation of the multivalent complex comprising Eps15, Hrs and STAM that regulates receptor signaling and trafficking (Bache et al., 2003). Thus, the complex formed by RNF11-, UbcH5 E2- and HECT-type E3 ligase could also contribute to the ubiquitination of key proteins regulating the endocytic trafficking of cell-surface proteins involved in breast cancer such as the EGF family of receptors (Waterman and Yarden, 2001). In this report, we have described the association of RNF11 with a number of proteins that potentially constitute a variety of multiprotein complexes. RNF11 may therefore function as an adaptor molecule for the

recruitment of specific substrates to E3 ligase complexes for ubiquitination and subsequent degradation by 26S proteasome. Thus, RNF11 and associated proteins are likely to be involved in multiple cancer-related cellular processes such as gene transcription, protein ubiquitination and ubiquitination-associated events like receptor endocytosis and protein sorting.

Materials and methods Yeast two-hybrid screen All vectors, yeast strains and methods were derived from the MATCHMAKER Two-Hybrid System 3 (Clontech), which is a Gal4-based system. AH109 and Y187 strains were used as hosts in the yeast two-hybrid assay (Clontech). AH109 contains two nutritional reporter genes for adenine and histidine. Both AH109 and Y187 contain the LacZ reporter gene. We found that the wt RNF11 activates reporter transcription in the absence of other transfected protein partners and decided to use mutant RNF11mtRing, where the first two cysteines of the RING-finger domain were changed to serines. This was generated using the Quickchange site-directed Mutagenesis Kit from Stratagene. The RNF11mtRing was then cloned in-frame with the Gal4 DNA-binding domain and the resulting plasmid was designated pGBKT7-RNF11mtRing. The expression of Gal4fusion protein RNF11mtRing was verified by Western blot analysis with anti-myc monoclonal antibody (Clontech) prior to library screening. The haploid AH109 yeast strain (MATa) was sequentially transformed with pGBKT7-RNF11mtRing. The pretransformed human ovary cDNA library in the haploid yeast strain (MATa) Y187 was bought from Clontech. Each cDNA in the library has been fused with Gal4 activation domain in the prey plasmid pACT2. The two strains were subsequently mated with eachother as haploid cells resulting in diploid cells containing both the bait and prey plasmids. Positive clones were selected and identified by growth on low stringency plates (Leu, Trp and His) or high stringency plates (Leu, Trp, His and Ade) and by b-galactosidase (b-gal) assays. Plasmid DNA was prepared from positive clones by transformation and growth in Escherichia coli DH5a and then cotransformed into AH109 with the original bait to verify the specific activation of reporter genes. The positive clones were confirmed in-frame with Gal4 activation domain by sequencing with T7 sequencing forward primer. DNA constructs for expression in mammalian cells Flag-Smurf2, Flag-mtSmurf2 and HA-tagged ubiquitin were generously provided by Drs Jeff Wrana and Liliana Attisano. HA-AMSH was subcloned by inserting AMSH fragment released from the prey vector pACT2 with BglII digestion into the pcDNA3.1 vector. GST-AMSH was created by inserting the BamHI/XhoI fragment from AMSH from pACT2 into the pEBG-GST vector. Flag-RNF11 was constructed in the pSG5 vector from pCMV-tag2C-RNF11. GST-RNF11 was cloned from Flag-RNF11 by inserting the BamHI/XhoI fragment into the pEBG-GST vector. Mutations of the RNF11 PY motif and RING domain were created as described (Subramaniam et al., 2003). Antibodies Flag (M2), b-actin and GST monoclonal antibodies were purchased from Sigma. HA monoclonal antibody (12CA5) Oncogene


1808 was from Roche and polyclonal antibody from Sigma. Protein sepharose beads were from Sigma and gluthathione sepharose beads were from Amersham Pharmacia. Cell culture and transfection The HEK293T cell line was maintained in high-glucose DME medium (GIBCO) containing 10% fetal bovine serum without antibiotics in 5% CO2 incubator. Cells were plated in 100 mm plates at a density of 2 106 cells/ml the day before the transfection and medium was changed 2–3 h prior to transfection. At 48 h after transfection, cells were lysed in 0.1% NP-40 lysis buffer (1 M Tris, pH 7.4, 5 M NaCl) with protease inhibitors. Immunoprecipitation and immunoblotting Cell lysates (250–500 mg) were incubated with antibodies on a shaker at 41C for 1 h. Preswollen protein sepharose beads (Sigma) were added to the above lysates and incubated for 30 min. Beads were then washed three times with the lysis buffer. SDS-loading buffer containing 100 mM DTT was added to the beads and were then boiled for 5 min. After spinning, the sample was applied to 12% SDS–polyacrylamide gel for electrophoretic separation. The proteins were transferred onto a nylon membrane overnight at room temperature at constant current. All procedures after blotting were carried out at room temperature. The membrane was blocked for

nonspecific binding with 5% milk powder in Tris-buffered saline (TBS), pH 7.4 for 1 h. First, the antibody was diluted in the blocking buffer and incubated for 1 h on a shaker. The membrane was then washed twice with TBS containing 0.1% Tween-20. Afterwards, the membrane was incubated with species-specific secondary antibody conjugated to horse-radish peroxidase enzyme for 1 h. The membrane was washed three times with wash buffer (TBS with 0.1% Tween-20) before exposure to X-ray film after treatment with the enzymeenhanced chemiluminescence kit from Pierce. Transcriptional response assay For TGF-b-inducible luciferase reporter assays, HepG2 cells were transiently transfected with the reporter plasmid (3TPLux), pCMV b-gal and the indicated constructs or with the empty vector alone. To induce the luciferase reporter, cells were treated overnight with 100 pM of TGF-b as described (Hayashi et al., 1997). Luciferase activity in cell lysates was measured using the luciferase assay system in a Berthold Lumat LB 9501 luminometer (Promega). Acknowledgements We thank Jeff Wrana and Liliana Attisano for helpful discussions and Richard for editorial assistance. This work was funded by grants from the Canadian Breast Cancer Research Alliance and Oncostasis, USA to A Seth.

References Bache KG, Raiborg C, Mehlum A and Stenmark H. (2003). J. Biol. Chem., 278, 12513–12521. Bean AJ, Davanger S, Chou MF, Gerhardt B, Tsujimoto S and Chang Y. (2000). J. Biol. Chem., 275, 15271–15278. Bonni S, Wang HR, Causing CG, Kavsak P, Stroschein SL, Luo K and Wrana JL. (2001). Nat. Cell Biol., 3, 587–595. Burger A, Li H, Zhang XK, Pienkowska M, Venanzoni M, Vournakis J, Papas T and Seth A. (1998). Oncogene, 16, 327–333. Carbone R, Fre S, Iannolo G, Belleudi F, Mancini P, Pelicci PG, Torrisi MR and Di Fiore PP. (1997). Cancer Res., 57, 5498–5504. Hayashi H, Abdollah S, Qiu Y, Cai J, Xu YY, Grinnell BW, Richardson MA, Topper JN, Gimbrone Jr MA, Wrana JL and Falb D. (1997). Cell, 89, 1165–1173. Ishii N, Owada Y, Yamada M, Miura S, Murata K, Asao H, Kondo H and Sugamura K. (2001). Mol. Cell. Biol., 21, 8626–8637. Itoh F, Asao H, Sugamura K, Heldin CH, ten Dijke P and Itoh S. (2001). EMBO J., 20, 4132–4142. Kavsak P, Rasmussen RK, Causing CG, Bonni S, Zhu H, Thomsen GH and Wrana JL. (2000). Mol. Cell, 6, 1365–1375. Kay BK, Williamson MP and Sudol M. (2000). FASEB J., 14, 231–241.

Oncogene

Kitching R, Gish G, Burger A, Landberg G and Seth A. (2003). Biochim. Biophys. Acta, submitted. Nuber U and Scheffner M. (1999). J. Biol. Chem., 274, 7576–7582. Rotin D, Staub O and Haguenauer-Tsapis R. (2000). J. Membr. Biol., 176, 1–17. Salcini AE, Chen H, Iannolo G, De Camilli P and Di Fiore PP. (1999). Int. J. Biochem. Cell Biol., 31, 805–809. Sorkina T, Bild A, Tebar F and Sorkin A. (1999). J. Cell Sci., 112 (Part 3), 317–327. Subramaniam V, Li HX, Wong M, Kitching R, Attisano L, Wrana J, Zubovits J, Burger A and Seth A. (2003). Br. J. Cancer, 89, 1538–1544. Tanaka N, Kaneko K, Asao H, Kasai H, Endo Y, Fujita T, Takeshita T and Sugamura K. (1999). J. Biol. Chem., 274, 19129–19135. Waterman H and Yarden Y. (2001). FEBS Lett., 490, 142–152. Weissman AM. (2001). Nat. Rev. Mol. Cell. Biol., 2, 169–178. Welcsh PL, Owens KN and King MC. (2000). Trends Genet., 16, 69–74. Zhang Y, Chang C, Gehling DJ, Hemmati-Brivanlou A and Derynck R. (2001). Proc. Natl. Acad. Sci., USA, 98, 974–979. Zheng L, Pan H, Li S, Flesken-Nikitin A, Chen PL, Boyer TG and Lee WH. (2000). Mol. Cell, 6, 757–768. Zhu H, Kavsak P, Abdollah S, Wrana JL and Thomsen GH. (1999). Nature, 400, 687–693.