If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
To whom correspondence should be addressed: Dept. of Experimental Oncology, European Institute of Oncology, Via Ripamonti 435, 20141 Milan, Italy. Tel.: 39-02-57489835; Fax: 39-02-57489851
* This work was supported by grants from the Associazione Italiana per la Ricerca sul Cancro (to S. C.) and from the Italian Ministry of Health. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. 1 Present address: Congenia S.r.l., Via Adamello 16, 20139 Milan, Italy.
SUMO-1 (small ubiquitin-related modifier-1) is a ubiquitin-like family member that is conjugated to its substrates through three discrete enzymatic steps, activation (involving the E1 enzyme (SAE1/SAE2)), conjugation (involving the E2 enzyme), and substrate modification (through the cooperation of the E2 and E3 protein ligases). The adenoviral protein Gam1 inactivates E1, both in vitro and in vivo, followed by SAE1/SAE2 degradation. We have shown here that Gam1 possesses a C-terminal SOCS domain that allows its interaction with two cellular cullin RING (really interesting new gene) ubiquitin ligases. We demonstrate that Gam1 is necessary for the recruitment of SAE1/SAE2 into Cul2/5-EloB/C-Roc1 ubiquitin ligase complexes and for subsequent SAE1 ubiquitylation and degradation. The degradation of SAE2 is not tightly related to Gam1 but is a consequent effect of SAE1 disappearance. These results reveal the mechanism by which a viral protein inactivates and subsequently degrades an essential cellular enzyme, arresting a key regulatory pathway.
Post-translational modifications of proteins are resourceful tools that cells use to control the function of proteins by regulating their activity, subcellular localization, and stability, as well as their interaction with other proteins. They are also important for promptly adjusting protein functions in response to changes in the state of a cell or its environment. The dynamic post-translational process that covalently transfers ubiquitin to itself or to other proteins is called ubiquitylation. Ubiquitin transfer requires distinct chemical steps catalyzed by the sequential activities of different enzymes (
). The ubiquitin E3 ligases perform the rate-limiting selectivity step of direct substrate recognition and are the most numerous and diverse in the ubiquitylation pathway. They can be subdivided into different general protein families, based on their structural features and their characteristic motifs. These include the cullin RING ligases (CRLs).
Through the assembly of their modular enzymatic core with different substrate recognition components, CRLs can promote the specific ubiquitylation of a wide array of targets (
The physiological importance of the CRL pathways is emphasized, considering that many receptors and adaptors of these ligases are exploited by viral and bacterial pathogens to subvert normal cellular processes (
). For example, the human immunodeficiency virus-encoded protein Vif interacts with EloB/C, Cul5, and Roc1 to form an E3 that eliminates the host anti-viral factor APOBEC3G (
). Another example is the complex formed from the adenoviral proteins E4orf6 and E1B55K and the host Cul5 complex that catalyzes the elimination of the tumor suppressor p53 to allow efficient viral replication (
). Sumoylation, the covalent addition of SUMO proteins to its substrates, in a similar manner to ubiquitylation, exploits different enzymatic reactions involving the E1-activating enzyme, the SAE1/SAE2 heterodimer, the E2-conjugating enzyme, and E3 SUMO ligases (
). In this report, we have shown that Gam1, recruiting Cul2/5-EloB/C-Roc1 through its C terminus-degenerate SOCS motif, targets the SUMO E1 heterodimer to these CRLs and promotes specifically SAE1 ubiquitylation. Subsequent proteasomal degradation of SAE2 depends on SAE1 disappearance, demonstrating that the two SUMO E1 subunits are mutually stabilized in vivo.
EXPERIMENTAL PROCEDURES
Plasmid Constructs—pSG9m Gam1 WT (Gam1 WT-Myc), pSG9m Gam1 L258/265A (Gam1 LL/AA-Myc), glutathione S-transferase (GST)-Gam1 WT, and GST-Gam1 L258/265A were previously described (
). pSG9m Gam1 S251A, L252A, Q253A, D254A, W255A, A256G, R257A, L258A, V260A, L272A, and P278A were generated by site-specific mutagenesis. pCDNA3 SAE1-SV5 and pCDNA3 SAE2-hemagglutinin were kindly provided by Ron Hay (
). pGEM3 elongin B and pGEM3 elongin C were kindly provided by Michele Pagano.
Transfection, Western Blot Analysis, and Immunoprecipitation—HeLa or Phoenix cells were transfected with the calcium-phosphate method using the indicated plasmids. After 24 or 48 h, cellular extracts were obtained, lysing cells in E1A buffer (50 mm Hepes, pH 7.0, 250 mm NaCl, 0.1% Nonidet P-40, 5 mm EDTA, 1 mm dithiothreitol, 0.2 mm phenylmethylsulfonyl fluoride; 1 mg/ml leupeptin, and 1 mg/ml aprotinin). In the experiments depicted in Fig. 4, E and F, the cellular extracts were obtained using a SDS buffer (
). Immunoprecipitations (IPs) were done using protein extracts incubated with the indicated antibodies in E1A buffer. The following antibodies were used in these assays: anti-Myc epitope 9E10, anti-hemagglutinin epitope (12CA5), anti-hemagglutinin probe (Y-11), anti-vinculin, anti-EloB (FL-118), anti-EloC (R-20), and anti-Cul5 (H-300) from Santa Cruz Biotechnology, anti-SV5 epitope (SV5-Pk1 clone, Serotec), anti-SAE1 and anti-SAE2 (kindly provided by Dr. Ron Hay), anti-Roc1 (BIOSOURCE), anti-Cul1, anti-Cul2, and anti-Cul3 from Zymed Laboratories Inc., anti-Cul4A (Biodesign), and anti-histone deacetylase 2 (Abcam).
FIGURE 4Gam1 recruits SAE1/SAE2 in cullin-based ubiquitin ligase complexes and ubiquitinates SAE1.A, Gam1 binds directly the SUMO E1 subunits SAE1 and SAE2 in vitro. [35S]Methionine-labeled in vitro translated SAE1 or SAE2 were incubated with GST or GST-Gam1 WT as described under “Experimental Procedures.” The samples were loaded onto a 12% SDS-polyacrylamide gel. GST proteins were detected by staining with Coomassie Blue, whereas the in vitro interactions were detected by autoradiography. B, SUMO E1 co-elutes with elongin C and Gam1 in vivo. Phoenix cells were transfected with the indicated plasmids, lysed, and processed for a gel filtration analysis (described under “Experimental Procedures”). 1/5 V/V of the indicated fractions were loaded in a 17% SDS-polyacrylamide gel and immunoblotted. The dotted boxes underline the shift of the proteins that co-elute with Gam1. C and D, Gam1 recruits SAE1 and SAE2 in the cullin2/5-based complexes in vivo. Phoenix cells were transfected with the indicated plasmids, lysed in E1A buffer, and immunoprecipitated (IP) using the indicated antibodies. E, Gam1 induces SAE1 in vitro ubiquitylation. Phoenix cells were transfected with the indicated plasmids, lysed in E1A buffer, and immunoprecipitated using the indicated antibodies. The immunoprecipitated samples were incubated as a source of ubiquitin E3 ligase activity in an in vitro ubiquitylation reaction as described under “Experimental Procedures.” [35S]Methionine-labeled in vitro translated SAE1 or SAE2 was used as a substrate and detected by autoradiography. F, SUMO E1 subunits are mutually stabilized. HeLa cells were transfected with plasmids encoding the indicated shRNAs. 5 days later, cells were lysed and processed for Western blot analysis. G, Gam1 exploits both Cul2 and Cul5 to degrade SAE1. HeLa cells were interfered against the indicated cullins as described under “Experimental Procedures.” Thereafter, the interfered cells were transfected with empty vector or Myc-Gam1. After 48 h, the cells were lysed and processed for Western blot analysis.
Small Interfering RNA and Short Hairpin RNA (shRNA)—The small interfering RNA experiment (see Fig. 4F) was performed using the following SMARTpool reagents (Dharmacon) according to the experimental procedures manual: human Cul2 RNA interference (M-007277), human Cul5 RNA interference (M-019553), Control (CTRL) (D-001206-13). The transfections were performed using the Oligofectamine reagent (Invitrogen). The DNA for the shRNA experiment (see Fig. 4E) encoded a 21-nucleotide hairpin-specific sequence, with a loop sequence (-ttcaagaga-) separating two complementary domains. The specific 21-nucleotide sequences were cloned in pSUPER vectors shSAE1 (GTTCTTTACAGGAGATGTT), shSAE2 (AGTGGAACAGCTGGGTATC), and shControl (CGTACGCGGAATACTTCGA).
In Vitro Binding—Proteins were in vitro translated in a rabbit reticulocyte lysate system kit (Promega) and [35S]methionine-labeled (Amersham Biosciences) according to the manufacturer's instructions. The in vitro translations were incubated in E1A buffer with 5 μg of GST or GST-fused proteins for 1 h at room temperature on rotation. After three washes in E1A buffer, the samples were loaded in SDS-polyacrylamide gels, stained with Coomassie staining, dried, and exposed for autoradiography (see Figs. 3B and 4A) or loaded on SDS-polyacrylamide gels and immunoblotted with the indicated antibodies (see Fig. 3, C and B).
FIGURE 3Identification of critical residues in the Gam1 SOCS motif required for association with elongins/cullins complexes.A, Phoenix cells were transfected with the indicated plasmids, lysed in E1A buffer, immunoprecipitated (IP) using an anti-Myc antibody, and immunoblotted as indicated. B, [35S]methionine-labeled in vitro translated EloB and/or EloC were incubated with GST, GST-Gam1 WT, or GST-Gam1 LL/AA as described under “Experimental Procedures.” The samples were loaded onto a 17% SDS-polyacrylamide gel. GST proteins were detected by staining with Coomassie Blue, whereas the in vitro interactions were detected by autoradiography. In vitro translated Cul2-Myc-tagged (C) or in vitro translated Cul5-Myc-tagged (D) were incubated with GST, GST-Gam1 WT, or GST-Gam1 LL/AA as described under “Experimental Procedures.” The samples were loaded onto a 12% SDS-polyacrylamide gel and immunoblotted with the indicated antibodies.
In Vitro Ubiquitylation Assay—Phoenix cells were transfected using the indicated plasmids. 48 h after transfection, the cells were lysed in E1A buffer and immunoprecipitated using an anti-Myc antibody. An aliquot of each IP was checked by Western blot analysis to verify the equal amount of immunoprecipitated proteins. The immunoprecipitated samples were incubated thereafter with the indicated combinations of 200 mm ubiquitin-E1, 500 mm UbcH5a, ubiquitin (Biomol) and SAE1, or [35S]methionine-labeled SAE2 in vitro translation. The reactions were incubated 2 h at 30 °CinATP buffer at 10× (500 mm Tris, pH 7.4, 50 mm MgCl2, 20 mm ATP). The samples loaded on SDS-polyacrylamide gels were then stained with Coomassie staining, dried, and exposed for autoradiography.
Gel Filtration—24 h after transfection, 4 mg of each of the total cellular lysates were loaded into a Superose 6 HR 10/30 column (Amersham Biosciences). The running conditions were designed as described in the manufacturer's procedures manual. The column was equilibrated with E1A buffer without Nonidet P-40. 40 fractions were collected. 1/5 V/V of the indicated fractions were loaded on SDS-polyacrylamide gels and immunoblotted.
RESULTS
Gam1 Recruits Cullin-based Ligase Complexes (CRLs)—We have demonstrated that the reduction of SAE1 and SAE2 proteins induced by Gam1 expression is not due to transcription regulation and is prevented by the addition of the proteasome inhibitor MG132 (
). Furthermore, the disappearance of SUMO E1 subunits was due to a substantial decrease in their half-lives induced by the viral protein, thus implicating the ubiquitin degradation pathway in Gam1 function (
). We therefore started a bioinformatics analysis to identify any conserved domain in the Gam1 amino acid sequence that might be traced back to known ubiquitin E3 complexes. Using SMART (simple modular architecture research tool) software, we found a putative SOCS motif (
). Examining the multiple alignment between the amino acid sequence of Gam1 and the established SOCS motifs of known proteins, we observed that Gam1 has a conserved pattern of amino acids mainly in the BC-box (Fig. 1).
FIGURE 1Gam1 is a putative SOCS-box-containing protein. Shown is the sequence alignment of SOCS-box motifs from cellular proteins that bind EloC. Conserved residues are shaded. Proteins that bind cullin2 or cullin5 differently are classified in two distinct groups. The alignment and the consensus motif were generated using ClustalW software (PAM matrix). p, polar amino acids; l, aliphatic amino acids; b, big amino acids; s, small amino acids.
To determine whether Gam1 could interact with the elongins, we performed an in vitro binding assay using a bacteria-purified GST fusion version of the Gam1 protein incubated with [35S]methionine-labeled in vitro translated EloB and/or EloC (Fig. 2A). We demonstrated that Gam1 was able to interact directly with EloC but not with EloB, that is recruited to Gam1 only in the presence of EloC (Fig. 2A). To establish whether Gam1 could physically interact in vivo with the endogenous CRLs, and in particular with the elongins, cells were transfected with a Myc-tagged Gam1 expression vector followed by immunoprecipitation. In agreement with the architecture of these complexes, Western blot analysis showed that, in vivo, Gam1 was able to immunoprecipitate strongly EloB, EloC, and the RING protein Roc1, generating nonphysiological aggregates (Fig. 2B).
FIGURE 2Gam1 interacts with cullin2/5-based aggregates.A, Gam1 recruits directly EloC and indirectly EloB in vitro. [35S]Methionine-labeled in vitro translated EloB and/or EloC were incubated with GST or GST-Gam1 as described under “Experimental Procedures.” The samples were loaded into a 17% SDS-polyacrylamide gel. GST proteins were detected by staining with Coomassie Blue, whereas the in vitro interactions were detected by autoradiography. B, Gam1 forms a complex with cullin2/5-based complexes in vivo. Phoenix cells were transfected with the indicated constructs and lysed in E1A buffer. The samples were then immunoprecipitated (IP), electrophoresed on SDS-polyacrylamide gels, and immunoblotted with the indicated antibodies. Input, whole cellular extract (WCE). C, Gam1 binds Cul2 and Cul5 in distinct complexes. Phoenix cells were transfected with the indicated plasmids two days later, lysed in E1A buffer, immunoprecipitated, and immunoblotted, as indicated. D, Gam1 co-elutes with Cul2/Cul5-based complex subunits in vivo. Phoenix cells were transfected with the indicated plasmids, lysed, and processed for a gel filtration analysis (described under “Experimental Procedures”). 1/5 V/V of the indicated fractions were loaded in 17% SDS-polyacrylamide gels and immunoblotted. The dotted boxes underline the shift of elongins and cullins complexes induced by the expression of Gam1.
As expected, Gam1 does not interact with Cul1, Cul3, or Cul4A but is able to bind both Cul2 and Cul5 (Fig. 2B). This result is quite peculiar, because the cellular and viral proteins that interact with EloB and EloC, usually form a complex in an exclusive manner with either Cul2 or Cul5. This specificity of cullin seems to be due to the C-terminal P/L-rich region of the SOCS motif (
), which is absent from the Gam1 amino acid sequence. To investigate whether Gam1 recruits Cul2 and Cul5 in a common protein complex, cells were transfected with combinations of Gam1 and differentially tagged Cul2 and Cul5 expression vectors. As shown in Fig. 2C, the immunoprecipitated Cul2 or Cul5 does not aggregate with the other overexpressed cullin, even if bound with Gam1. These data confirm the individual interactions between Gam1 and Cul2 or Cul5 and show that the two cullins generate distinct and new complexes with the viral protein.
Subsequently, to validate the in vivo interaction between Gam1 and the Cul2/Cul5-based CRLs, we followed the distribution of Gam1 protein using a gel filtration technique and observed complexes with a representative range of molecular weights. As shown in Fig. 2D, normally EloB and EloC co-elute in the same fractions and are mainly present in aggregates between 44 and 158 kDa (Fig. 2D, upper panel). Significantly, the expression of Gam1 both induces a clear increase in the molecular weight of the elongin complexes and co-elutes with them. The presence of Cul2, Cul5, and Roc1 in the same elution fractions suggests that Gam1 recruiting EloB/C into new complexes (Fig. 2D, lower panel) could reconvert the functionality of these endogenous CRLs.
Gam1 Binds Elongins through Its SOCS-Box Motif—To prove the functionality of the Gam1 SOCS-box and its implication in EloC recruitment, few amino acid residues involved in this putative domain were mutated. The resulting Gam1 mutants were expressed and tested for elongin and cullin binding by immunoprecipitation (Fig. 3A). The alignment between Gam1 and the consensus SOCS motif (Fig. 1) reveals that replacement of the highly conserved cysteine with alanine (Ala-256) in Gam1 generates a domain most similar to SOCS domains of other viral proteins (
). Nevertheless, the mutation of this residue does not cause any effect on its binding property (Fig. 3A). Instead, the Gam1 point mutants L252A, W255A, L258A, and L265A exhibited a drastically lower binding with EloB and EloC, consolidating the basic role of the hydrophobic surfaces of the SOCS motif in the functional recruitment of E3 adaptor subunits (Fig. 3A). To exclude the possibility that these mutants have any other structural deficiencies, we tested binding with a different and uncorrelated Gam1 interactor, histone deacetylase 2 (
). All Gam1 mutants bind histone deacetylase 2 in vivo in the same manner (Fig. 3A), demonstrating that these point mutations impair only the hydrophobic pocket necessary for EloC interaction.
The Gam1 L252A, W255A, L258A, and L265A mutants show a virtually abolished binding with elongins but a residual interaction with Cul2 and Cul5, whereas the Gam1 double mutant L258A/L265A (
) was totally unable to immunoprecipitate all of the subunits of the E3 complexes (Fig. 3A). Supposing that Gam1 could interact also directly with Cul2 and Cul5, we evaluated the binding properties of the Gam1 mutant L258A/L265A (Gam1 LL/AA) in vitro. As shown in Fig. 3B, incubating the in vitro translated EloB and EloC with bacteria-purified GST-Gam1 WT or GST-Gam1 LL/AA, we demonstrated that the Gam1 mutant does not bind the EloB/C heterodimer. Subsequently, we investigated the interaction between bacteria-purified GST-Gam1 WT and its mutant incubated with either in vitro translated Cul2 (Fig. 3C) or Cul5 (Fig. 3D). We demonstrated that Gam1 binds either Cul2 or Cul5, and these direct interactions were not affected by the presence of a functional Gam1 SOCS motif in vitro.
Gam1 Recruits SUMO E1 in the CRL Complexes—We have shown that the Gam1 LL/AA can no longer bind Cul2/5-EloB/C proteins and cannot induce SUMO E1 disappearance in vivo (
). We therefore reasoned that, by recruiting the CRL complexes through its SOCS-box, Gam1 could work as a substrate receptor, allowing SAE1 and SAE2 ubiquitylation and their subsequent proteasomal degradation. Because Gam1 binds directly SAE1 and SAE2 (Fig. 4A) (
), we decided to establish whether they could be included into the Gam1-EloB/ C-Cul2/5 protein aggregates following their distribution in a gel filtration assay. As shown in Fig. 4B, upper panel, normally EloC does not co-elute with the heterodimer SAE1/SAE2. Instead the presence of Gam1 causes the co-elution of SUMO E1 components and EloC in the same fractions (Fig. 4B, lower panel, dotted box). These in vivo results suggest that Gam1 may recruit SAE1/SAE2 to the Cul2/5-EloB/C complexes, thus promoting their ubiquitylation.
To verify whether the SUMO E1 subunits are actually present in the same protein aggregates together with Gam1, EloB/C, and Cul2/Cul5, we immunoprecipitated the overexpressed tagged version of SAE1 and SAE2 with or without Gam1 WT or Gam1 LL/AA (Fig. 4, C and D). We demonstrated that SAE1 and SAE2 do not bind EloB/C and Cul2/5 normally but are recruited in the CRL complexes only through binding to the Gam1 WT protein (Fig. 4, C and D). As expected, Gam1 LL/AA, defective for SUMO E1 protein degradation, cannot operate as a substrate receptor and fails to join SAE1/SAE2 into Cu2/5-based complexes (Fig. 4D).
Gam1 Permits SAE1 in Vitro Ubiquitylation—To tightly correlate the degradation of SUMO E1 and the assembly of the functional ubiquitin E3s induced by Gam1, we followed the enzymatic activity of these nonphysiological protein aggregates. We transfected cells with empty vector, Gam1 WT, or Gam1 LL/AA Myc-tagged expression vectors and subsequently immunoprecipitated the Myc-tagged proteins. Equal amounts of the immunoprecipitated materials were used as a source of E3 activities in the in vitro ubiquitylation reactions of [35S]methionine-labeled in vitro translated SAE1 or SAE2 (Fig. 4E). Surprisingly, we showed that immunoprecipitated Gam1 could in vitro conjugate ubiquitin only on the SAE1 protein (Fig. 4E).
SAE1 and SAE2 Are Mutually Stabilized in Vivo—To explain the paradox in which the degradation of SAE1 and SAE2 in vivo mediated by Gam1 contrasts with its inability to post-translationally modify SAE2 in vitro, we investigated whether SUMO E1 subunits are mutually stabilized in vivo. To deplete the protein level of SUMO E1 subunits, we set an RNA interference approach (shRNAs) able to reduce successfully SAE1 or SAE2. As shown in Fig. 4F, the disappearance of any SUMO E1 subunit caused a strong reduction of the other protein constituent of the enzyme. Therefore, in the presence of Gam1, the disappearance of SAE2 seems to be directly related to the reduction in SAE1 protein level and is not tightly dependent on the effect of Gam1 on endogenous CRLs. Furthermore, we could conclude that the stability of the SUMO E1 heterodimer is tightly related to the presence of its two subunits, SAE1 and SAE2.
Gam1 Exploits Both Cul2 and Cul5 to Degrade SAE1—The original ability of Gam1 to interact with both Cul2 and Cul5 and the absence of any identifiable cullin selection motif in its amino acid sequence prompted us to elucidate the contribution of each cullin in SAE1 degradation. To understand whether they are both necessary for the function of Gam1, we employed an RNA interference approach to delete specifically Cul2 and/or Cul5. As shown in Fig. 4G, the expression of Gam1 in cells in which Cul2 and Cul5 are simultaneously depleted does not induce SAE1 degradation. The single cullin depletion does not restore the disappearance of SAE1 caused by Gam1, suggesting that Cul2 and Cul5 are redundant for Gam1 function (Fig. 4G).
DISCUSSION
A plethora of data has now implicated SUMO in fundamental biological activities, making SUMO as important in regulating cell activity as ubiquitin. We have described a novel mechanism of action adopted by the CELO adenoviral protein Gam1 that induces a total reduction of cellularly, sumoylated proteins by blocking the formation of an E1-SUMO thioester complex (
). We now reveal the mechanism underlying Gam1 function by showing that it exploits the endogenous ubiquitin pathway to convert the specificity of cullin2/5-based E3 complexes, triggering a nonphysiological ubiquitylation and degradation of SAE1.
Gam1 Is the Substrate Receptor of Ubiquitin-E3 Complexes—Having previously demonstrated that the function of Gam1 on sumoylation is related to the proteasome activity (
), we then investigated the potential role of the ubiquitin system in the Gam1 phenotype. In a similar manner to sumoylation, protein ubiquitylation catalyzes the formation of polyubiquitin chains onto substrate proteins via isopeptide bonds through a cascade of enzymes involving activating (E1), conjugating (E2), and ligating (E3) activities (
). Polyubiquitylated substrates are then rapidly delivered to and degraded by the 26 S proteasome. The substrate specificity of ubiquitin-dependent proteolysis is tightly mediated by hundreds of E3 ubiquitin ligases. Many viral and bacterial pathogens have evolved different proteins that convert the specificity of host multisubunit ubiquitin ligase complexes (E3), inducing a nonphysiological degradation of specific cellular targets (
). These considerations led us to identify a putative SOCS domain in the C-terminal region of the Gam1 protein (Fig. 1), implicating an involvement in Cul2/5-based E3 ligase complexes.
Although the amino acid sequences are not fully conserved among the SOCS proteins, they maintain the same helix structure, steric surface, and pack similar to the hydrophobic pocket of EloC. Interestingly, the Gam1 amino acids included between 251–265 are predicted to form an α-helix, and the Leu-252, Trp-255, Leu-258, and Leu-265 residues could be located on the same face of this generated helix. This probable hydrophobic cluster could support a successful interaction with the hydrophobic pocket of EloC. In agreement with these structural considerations, we demonstrated that Gam1 interacts directly with EloC (Fig. 2, A and B), and this binding is stabilized by the hydrophobic residues Leu-252, Trp-255, Leu-258, and Leu-265 of Gam1 (Fig. 3A). In fact, the double substitutions L258A and L265A (LL/AA) impair totally the Gam1 binding capability both in vitro and in vivo (Fig. 3, A and B).
Generally, within the SOCS motif, the presence of a C-terminal proline/leucine-rich region (P/L-rich) (
), allows us to distinguish Cul2 or Cul5-associated proteins (Fig. 1). Even if Gam1 could not be clustered into any of these groups, we demonstrated that this viral protein binds independently both Cul2 and Cul5 (Fig. 2, B and C). We could assume that other less well conserved residues in the Gam1 amino acid sequence are important for the cullin selection or that Gam1, through its SOCS-box, binds specifically elongins but nonspecifically either Cul2 or Cul5. Moreover, despite the direct binding between Gam1 and Cul2 or Cul5 (Fig. 3, C and D), we could assume that, in vivo, the correct recruitment of the EloB/C heterodimer is necessary to stabilize the entire Gam1 complex and that its dual interaction with Cul2 and Cul5 could theoretically provide a larger and stronger platform of substrates for Gam1 to ubiquitylate. We showed that the presence of Gam1 induces, as expected, a strong increase in the molecular weight of elongin complexes (Fig. 2D), supporting the hypothesis that Gam1 is able to reconvert the functionality of these endogenous protein aggregates.
Gam1 Targets SAE1 into Cul2/5-EloBC-Roc1 Complexes—We have shown that the Gam1 LL/AA mutant can no longer bind Cul2/5-EloB/C proteins (Fig. 3, A and B). Interestingly, this double point mutant was initially identified as the inactive version of Gam1 protein (
). Consequently, we speculated that Gam1, by recruiting the cullin E3 ligase complexes through its SOCS motif, allowed SAE1 and SAE2 ubiquitylation and their proteasomal degradation. Following this hypothesis, we demonstrated that SAE1 and SAE2 were stably associated, through the direct link of Gam1 WT, into the Cul2/5-EloB/C-Roc1 complexes (Fig. 3, B and C). In agreement with the initial assumption, Gam1 LL/AA fails to connect the SUMO E1 subunits to the endogenous CRL complexes (Fig. 3B). Therefore, we could assume that the inability of Gam1 LL/AA to degrade SUMO E1 in vivo is a direct effect of its failure to join, in a stable manner, the CRLs and SAE1/ SAE2. Consequently, we followed the ubiquitin-conjugating activity of the Gam1-based CRLs in vitro and showed that Gam1 could ubiquitin-modify only the SAE1 protein, suggesting that the in vivo phenotype could be due to overlapping but distinct events (Fig. 4E). Using a RNA interference approach, we demonstrated that the stability of the SUMO E1 heterodimer is tightly related to the presence of its two subunits, SAE1 and SAE2, and that the effect of Gam1 on SAE2 seems to be a consequence of the induced ubiquitylation and degradation of SAE1 (Fig. 4F).
We have shown that the viral protein Gam1, through a novel dual binding with Cul2- and Cul5-based aggregates (Fig. 2) could theoretically provide two nonphysiologically larger ubiquitin platforms with overlapping functions. In fact, Cul2 and Cul5 seem to have a redundant role during the degradation of SAE1 induced by Gam1 (Fig. 4G).
The atypical role of Gam1 could also be a viral evolutionarily conserved mechanism to assure specific substrate modification and degradation of essential cellular proteins. The flexibility of action of Gam1 supports its pivotal role in viral replication and reveals its potential for the study of cellular pathways.
Acknowledgments
We are very grateful to Andrea Musacchio and Colin Goding for helpful suggestions and Gioacchino Natoli for critical reading of the manuscript. We thank Tingting Yao, Joan Conaway, Michele Pagano, Xiaofang Yu, and Nakayama Keiichi for plasmids, Ron Hay for antibodies, all of the laboratory members for data discussions, and Paula Babarovic for generating shRNA plasmids. We thank all European Institute of Oncology (IEO) and IEO-FIRC Institute of Molecular Oncology (IFOM) campus facilities.
87411-0&id=gr4.jpg){kind=link}
87411-0&id=gr3.jpg){kind=link}
87411-0&id=gr1.jpg){kind=link}
87411-0&id=gr2.jpg){kind=link}