The Interferon-inducible Ubiquitin-protein Isopeptide Ligase (E3) EFP Also Functions as an ISG15 E3 Ligase*

The expression of the ubiquitin-like protein ISG15 and protein modification by ISG15 (ISGylation) are strongly activated by interferons. Accordingly, ISG15 expression and protein ISGylation are strongly activated upon viral and bacterial infections and during other stress conditions, suggesting important roles for the ISG15 system in innate immune responses. Here, we report the identification of the ubiquitin-protein isopeptide ligase (E3) EFP (estrogen-responsive finger protein) as the ISG15 E3 ligase for 14-3-3σ protein. Like other known components of the protein ISGylation system (ISG15, UBE1L, UBP43, and UBC8), EFP is also an interferon-inducible protein. Expression of EFP small interfering RNA decreased the ISGylation of 14-3-3σ in the 293T cell ISGylation system as well as in MCF-7 cells upon interferon treatment. Furthermore, the ISGylation enzyme activity of EFP was RING domain-dependent. These findings indicate that EFP is an ISG15 E3 ligase for 14-3-3σ in vivo. The fact that both UBC8 and EFP are common components in the ubiquitin and ISG15 conjugation pathways suggests a mechanism whereby a limited set of enzymes accomplishes diverse post-translational modifications of their substrates in response to changes in environmental stimulations.

ISG15 is one of several known Ubl ubiquitin-like modifiers (1,2). Upon type I interferon (IFN) 3 treatment, ISG15 forms covalent conjugates with cellular proteins, a process similar to ubiquitin modification (ubiquitination or ubiquitylation) (1). Ubiquitylation is now understood to be the dominant mechanism whereby cellular proteins are marked for degradation (3,4). However, unlike ubiquitin modification, the role of ISG15 modification (ISGylation) has not been clearly defined, although ISG15 has been known since 1979 (5,6).
ISG15 is composed of two domains, each of which bears sequence identity to ubiquitin (33 and 32% for the N-and C-terminal domains, respectively) (7). The process of ISGylation is expected to be similar to that of protein ubiquitylation. Mass spectrometry analysis confirmed that an isopeptide bond can be formed between the C terminus of ISG15 and the lysine ⑀-amino group of the target protein (8). There are a series of distinct enzymes involved in the process of protein ubiquitylation and deubiquitylation, viz. the ubiquitin-activating enzyme (E1), ubiquitin carrier protein (E2), ubiquitin-protein isopeptide ligase (E3), and the Dub deubiquitining proteases (9,10). An E1 enzyme for ISGylation (UBE1L) has been cloned and shows high homology to the E1 enzyme UBE1 (11). UBE1L-deficient cells express only free ISG15 and do not show ISG15 conjugation, indicating that UBE1L is necessary for ISGylation (12,13). UBP43 (USP18) has been identified as a de-ISGylating isopeptidase and is highly specific in removing ISG15, but not Nedd8 and SUMO (small ubiquitin-like modifier) from their conjugates (14). UBP43 is a member of the UBP (USP) family of ubiquitin-specific proteases and contains conserved Cys and His domains, which are common in all members of this family (15).
Based on the information on UBE1L and UBP43, it is logical to hypothesize that ISG15 E2 and E3 are also members of the ubiquitin E2 and E3 families. Recently, Kim et al. (16) and Zhao et al. (17) reported the identification of the ubiquitin E2 enzyme UBC8 (UBCH8/UbcM8) as an ISG15-conjugating enzyme. Using the small interfering RNA (siRNA) approach, both groups indicated that UBC8 is a predominant ISG15 E2 enzyme for IFN-induced protein ISGylation in HeLa cells. Interestingly, UBC8 also functions as a ubiquitin E2 enzyme (18), indicating the possible cross-talk of protein ISGylation and ubiquitylation. In the ubiquitin system, the transfer of ubiquitin from E2 enzymes to target proteins is mediated by E3 enzymes, which interact with both E2 and its target proteins and function as a bridge between E2 and the target protein (19). UBC8 has been shown to interact with a number of ubiquitin E3 ligases to mediate protein ubiquitylation, such as E6AP, HHARI, Parkin, cIAP, Dorfin, EFP, RLIM, and Staring (18, 20 -26). The fact that UBC8 functions as a dual E2 enzyme for both ISG15 and ubiquitin raises the possibility that some UBC8-interacting ubiquitin E3 ligases can function as the ISG15 E3 ligase, mediating the interaction between UBC8 and the ISG15 target proteins and helping in the transfer of ISG15 from UBC8 to target proteins. Here, we report that a UBC8interacting protein, EFP (estrogen-responsive finger protein), can help in the ISG15 modification of 14-3-3 protein both in transfection experiments and in vivo. EFP depends on its RING finger domain to support the ISGylation of 14-3-3. These data indicate that EFP is an ISG15 E3 ligase for 14-3-3. This is the first study reporting the identification of ISG15 E3 in vivo, and we provide evidence that EFP protein is a common component in the ubiquitin and ISG15 conjugation pathways.
Northern Blotting-Total RNA from WM9 and MCF-7 cells was isolated using RNA-Bee reagent (Tel-Test) according to the manufacturer's instructions. Ten micrograms of total RNA from each time point was separated on a formaldehyde-agarose gel (0.22 M), blotted onto Hybond N ϩ membrane (Amersham Biosciences), and probed with 32 Plabeled cDNAs.
RNA Interference for EFP-The mammalian expression vector pSU-PER.retro.puro (OligoEngine, Seattle, WA) was used for constructing small hairpin RNA (shRNA) for EFP. The targeted sequences for siRNA are EFP cDNA regions 1158 -1176 (ggaaaagaaatccaagaaa; siRNA1) and 711-729 (ggtggagcagctacaacaa; siRNA2). The synthesized oligonucleotides for siRNA1 are 5Ј-Gatccccggaaaagaaatccaagaaattcaagagatttcttggatttcttttccttttta-3Ј and 5Ј-Agcttaaaaaggaaaagaaatccaagaaatctcttgaatttc-ttggatttcttttccggg-3Ј; the synthesized oligonucleotides for siRNA2 are 5Ј-Gatccccggtggagcagctacaacaattcaagagattgttgtagctgctccaccttttta-3Ј and 5Ј-Agcttaaaaaggtggagcagctacaacaatctcttgaattgttgtagctgctccaccggg-3Ј. The annealed oligonucleotides were digested with BglII and HindIII and inserted into the pSUPER.retro.puro vector. To test the efficiency of EFP shRNA in the cotransfection experiments, 293T cells were transfected with pSUPER-EFP shRNA and mammalian expression plasmids encoding FLAG-14-3-3 and His-ISG15. The expression of EFP and the ISGylation of 14-3-3 were analyzed 48 h post-transfection. To test the effect of EFP shRNA on IFN-induced 14-3-3 ISGylation in vivo, we packaged retrovirus by cotransfection of pSUPER-EFP shRNAs with the amphotropic retroviral packaging vector pCL-10A1 into 293T cells. MCF-7 cells seeded in 6-well plates were infected with control pSUPER.retro.puro vector retrovirus or pSUPER-EFP shRNA retrovirus. Twenty-four hours post-infection, cells were selected with 6 g/ml puromycin. Dead cells were washed out with PBS, and attached live cells were amplified, resulting in stable shRNA-expressing cell pools. MCF-7 cells and different cell pools were treated with 1000 units/ml human IFN-␣. After 48 h, the expression of EFP and the ISGylation of 14-3-3 were analyzed.

RESULTS
EFP Is Up-regulated upon IFN Treatment-Because the primary and tertiary structures of ISG15 are similar to those of ubiquitin (30), the mechanism of ISGylation is expected to be similar to that of protein ubiquitylation and is predicted to involve the activity of the E1, E2, and E3 enzymes. Notably, both ISG15 expression and protein ISGylation are highly induced upon IFN stimulation. Furthermore, all known components of the protein ISGylation system (i.e. ISG15, UBE1L, UBP43, and UBC8) are IFN-inducible proteins (6,31). We hypothesized that ISG15 E3 ligases are UBC8-interacting ubiquitin E3 ligases that may also be encoded by IFN-inducible genes. Based on published reports (24,28) and available information from Web sites, we identified one candidate protein (EFP) that fully matched the hypothesized criteria.
In gene expression profiling studies using oligonucleotide array, Leaman et al. (28) showed that EFP is one of the IFN-stimulated genes in human melanoma cell lines WM9 and WM35. To determine whether EFP mRNA is truly up-regulated by IFN treatment, Northern blot analysis was performed with total RNA prepared from WM9 human melanoma cells and MCF-7 human breast cancer cells. EFP has been reported to be an estrogen-inducible gene in MCF-7 cells (32). Northern blotting showed that EFP mRNA could be up-regulated by IFN treatment at 2 h, reaching a maximum at ϳ5 h. EFP mRNA levels at 8 and 24 h were still higher than the basal level (Fig. 1A). ISG15 mRNA levels were also analyzed by Northern blotting as a positive control (Fig. 1A).
To determine whether the induction of EFP is at the level of protein expression, Western blot analysis was performed with lysates prepared from WM9 human melanoma cells and MCF-7 human breast cancer cells. Immunodetection using EFP-specific antibody showed that IFN increased the protein level of EFP, reaching a maximum at ϳ8 h (Fig.  1B). As a positive control, ISG15 protein was also induced by IFN treatment.
The promoter regions of genes related to ISG15 modification, including ISG15, UBE1L, UBP43, and UBC8, contain the IFN-stimulated response element (ISRE) (6,16), which is responsible for promoter activation by type I IFN via its interaction with the IFN-stimulated gene factor-3 complex containing phosphorylated Stat1, Stat2, and IRF9 (33,34). As shown in Fig. 1C, the 5Ј-flanking region of the human EFP gene (32) also contains an ISRE sequence. The facts that the EFP promoter contains the ISRE sequence and that EFP protein is induced by IFN treatment support the involvement of EFP protein in IFN-stimulated ISGylation.
14-3-3 Can Be Modified by ISG15 in the 293T Transfection System-EFP protein is a RING finger-type ubiquitin E3 ligase that directs 14-3-3 ubiquitination and promotes 14-3-3 proteolysis in a proteasome-dependent manner (24). EFP can interact with UBC8 via its RING finger domain and mediates the transfer of ubiquitin from UBC8 to 14-3-3 (24). Recently, UBC8 has been identified as an ISG15 E2-conjugating enzyme (16,17). These facts support the possible role of EFP in the ISG15 modification of 14-3-3 protein. To determine whether 14-3-3 could be modified by ISG15, we cotransfected plasmid DNA expressing His-ISG15, UBE1L, UbcM8, and FLAG-14-3-3 into 293T cells. His-ISG15 and covalently linked proteins were enriched by Ni-NTA pull down, and antibody against the FLAG epitope was used in Western blot analysis to detect 14-3-3. As shown in Fig. 2, 14-3-3 associated with Ni-NTA resin independently of ISG15 modification. At the same time, a band that matched the molecular mass of a single ISG15-conjugated 14-3-3 protein was detected. This band was observed in cell extracts cotransfected with 14-3-3 and ISG15 (Fig. 2,  lane 2). Expression of UBE1L and UBC8 increased the intensity of this additional band (Fig. 2, compare lanes 4 and 6 with lane 2). Cotransfection of UBE1L and UBC8 expression plasmids strongly increased the ISG15-modified form of 14-3-3 (Fig. 2, lane 8). The total levels of 14-3-3 protein were determined by direct Western blotting in the same set of experiments. Although the total levels of 14-3-3 protein were quite equal, significantly higher levels of unmodified 14-3-3 were pulled down by Ni-NTA accompanied by higher levels of ISGylated 14-3-3 (Fig. 2, for example, lane 8). The result was very repeatable (also see Fig. 4). It is known that 14-3-3 proteins can form dimers and that 14-3-3 dimeric structure is very stable (35). It is likely that ISGylated 14-3-3 can still form dimers with unmodified 14-3-3, which was also pulled down by Ni-NTA resin through the interaction with His-ISG15/14-3-3.

14-3-3 Can Be Modified by ISG15 upon Interferon Treatment in MCF-7 Cells-
To determine whether endogenous 14-3-3 is subjected to ISG15 modification, we performed immunoprecipitation with anti-14-3-3 antibody and anti-mouse IgG1 control antibody using MCF-7 cells extracts with or without stimulation by human IFN-␣. The immunoprecipitates were analyzed by Western blotting using antibody against human ISG15 or 14-3-3. As shown in Fig. 3A, a species with the predicted molecular mass of 14-3-3 modified with a single ISG15 moiety (28 ϩ 17 kDa) was detected in the IFN-treated sample, but not in the untreated one. Notably, 14-3-3/ISG15 conjugation was not observed in the immunoprecipitates of the anti-mouse IgG1 control antibody. The immunoprecipitates were also detected by anti-14-3-3 antibody (Fig. 3B). As shown in Fig. 3B, besides unmodified 14-3-3, an additional band was detected in the IFN-treated samples, but not in the untreated one. The size of this band corresponds to the ISG15 band detected in Fig. 3A. The molecular mass of the 14-3-3/ISG15 conjugate in Fig. 3 corresponds to that of the predominant 14-3-3 species detected in the 293T ISG15-conjugating system described in Fig. 2. We could also detect the ISGylation of 14-3-3 in IFN-treated HeLa cells and A549 cells (data not shown).
EFP Promotes ISGylation of 14-3-3-The observation that 14-3-3 is a target of ISG15 modification prompted us to investigate whether EFP can function as an ISG15 E3 ligase for 14-3-3. We first used the 293T system to examine the effect of EFP on 14-3-3 ISGylation. FLAG-14-3-3 was expressed in 293T cells along with the His-ISG15, UBE1L, and UBC8 expression constructs in the absence or presence of HA-EFP. The ISGylated proteins were enriched with Ni-NTA-agarose and detected by Western blotting. As described above, ISGylated 14-3-3 was detected when it was coexpressed with His-ISG15, UBE1L, and UBC8 without the addition of HA-EFP (Fig. 4A, lane 4); however, the amount of ISGylated 14-3-3 was dramatically increased by HA-EFP expression (lane 5). On film with longer exposure times, the ISGylated 14-3-3 signal was also detected when FLAG-14-3-3 was coexpressed with His-ISG15, and EFP expression increased this signal (Fig.  4A, lanes 2 and 3) (data not shown). As in Fig. 2, higher levels of unmodified 14-3-3 were pulled down by Ni-NTA accompanied by higher levels of ISGylated 14-3-3. The whole cell lysates were analyzed  directly by Western blotting for determination of total FLAG-14-3-3 levels and total ISG15 conjugation levels. Although EFP significantly increased the level of ISGylated 14-3-3, it did not enhance the total amount of protein ISGylation.
siRNA for EFP Decreases the IFN-induced ISGylation of 14-3-3-After showing that EFP supports 14-3-3 ISGylation in 293T cell transfection assays, we further tested whether EFP depletion in vivo would cause a decrease in 14-3-3 ISGylation in IFN-treated cells. We reduced cellular EFP expression through the siRNA approach using a retroviral expression vector (pSUPER-EFP1 or pSUPER-EFP2) (36). First, we tested whether these siRNA constructs could decrease the expression of EFP and the ISGylation of 14-3-3 in 293T cells. Plasmids for His-ISG15 and FLAG-14-3-3 were transfected into 293T cells with increasing amounts of a mixture of pSUPER-EFP1 and pSUPER-EFP2. Cotransfection of EFP siRNA constructs reduced the levels of EFP protein and 14-3-3 ISGylation in a dose-dependent manner (Fig. 5A). To investigate the effects of EFP siRNA on IFN-induced ISGylation of endogenous 14-3-3, MCF-7 cells were infected with pSUPER-EFP1 and pSUPER-EFP2 retroviruses that also express the puromycin selection marker. After puromycin selection, the dead cells were washed out with PBS, and attached live cells were amplified, resulting in stable shRNA-expressing cell pools. The levels of EFP and 14-3-3 proteins were examined. As shown in Fig. 5B, the EFP protein level was strongly inhibited by pSUPER-EFP2 shRNA expression and weakly inhibited by pSUPER-EFP1 shRNA expression in untreated cells. After interferon treatment, the protein level of EFP was up-regulated, and the EFP expression was clearly inhibited by EFP siRNA2. The expression of both siRNAs had little effect on the total 14-3-3 protein level. However, the amount of 14-3-3 ISGylation was substantially decreased by EFP siRNA2 expression (Fig. 5B, lane 8). Western blotting confirmed that the amount of immunoprecipitated 14-3-3 was quite equal in the siRNA and control lanes. These results confirm that EFP is necessary in vivo for 14-3-3 ISGylation.

The RING Finger Domain Is Required for EFP to Function as an ISG15
E3 Ligase-Ubiquitin E3 had been classified into two major families, the HECT and RING finger families, which contain the conserved HECT and RING finger domains, respectively (3). RING fingers are "crossbrace" structures formed by conserved cysteine and histidine residues that coordinate two zinc ions (37). A previous study has shown that the RING finger domain is important for EFP in the physical interaction with UBC8 and in the ubiquitination of 14-3-3 (24). To determine whether the ISG15 E3 ligase activity of EFP is RING finger domain-dependent, we constructed a number of EFP mutants (Fig. 6A) and examined their E3 ligase activities for 14-3-3 ISGylation. 293T cells were transfected with plasmids expressing His-ISG15 and FLAG-14-3-3 together with different EFP expression constructs. RING finger-deleted EFP did not support the ISGylation of 14-3-3 (Fig. 6B, lane 4). Replacement of conserved cysteines in the EFP RING finger domain with serine also disrupted the function of EFP (Fig. 6B, lanes 6 and 7). At the same time, the RING finger domain alone could not support the ISGylation of 14-3-3. These results confirm that the RING finger domain is required for EFP to function as an ISG15 E3 ligase for 14-3-3.

DISCUSSION
The Ubiquitin-like protein ISG15 is highly induced upon IFN stimulation (1, 5, 38) and is detected as a free protein and a target protein- . The proteins were also immunoprecipitated with anti-mouse IgG1 antibody as a negative control. The positions of unmodified 14-3-3, ISGylated 14-3-3, and the immunoglobulin heavy (HC) and light (LC) chains are shown. 14-3-3 expression and protein ISGylation in cellular protein extracts were analyzed by Western blotting with antibodies specific for 14-3-3 and ISG15 (C). Conj., conjugate. conjugated form in cells (39). Although ISG15 has been known for 26 years (5), the process of ISGylation and the biological effects of ISGylation have not been well studied. Recently, research on ISGylation has been accelerated by the identification of the enzymes involved in the ISG15 conjugation process, including the ISG15 E1 enzyme UBE1L, the E2 enzyme UBC8, and the de-ISGylating enzyme UBP43 (12,14,16,17). A significant number of ISG15 target proteins came to light starting with serpin-2a (40) and Jak1 (Janus kinase-1), Stat1, Erk1 (extracellular signal-regulated kinase-1), and phospholipase C␥ (29), followed by Ͼ200 new proteins that function in diverse cellular pathways, including RNA splicing, chromatin remodeling, transcription, cytoskeletal organization, stress responses, and translation (41,42). Based on the similarity between the ubiquitin and ISG15 systems, we hypothesized that ISG15 E3 ligases are members of the ubiquitin E3 family, interact with the ISG15 E2 enzyme UBC8, and may be encoded by IFN-inducible genes. Recently, using purified Saccharomyces cerevisiae Rsp5p E3, Zhao et al. (17) showed that ISG15 conjugation can be accomplished by a ubiquitin E3 ligase in vitro. However, there are no in vivo data supporting this hypothesis. In this study, we have demonstrated that the RING finger ubiquitin E3 ligase EFP can promote ISG15 conjugation to 14-3-3 upon IFN treatment. We have also demonstrated that the RING domain, which is required for the ubiquitin E3 ligase activity of EFP (24), is required for promoting 14-3-3 ISGylation. These results suggest that EFP acts as an ISG15 E3 ligase for 14-3-3. To our knowl-edge, this is the first report on the identification of an ISG15 E3 ligase in vivo. These results indicate that the pathways of ubiquitin and ISG15 overlap in vivo.
UBC8 and its interacting E3 ligase EFP have the capacity to catalyze both ubiquitin and ISG15 conjugation. How are the specificities of these enzymes determined for their involvement in ubiquitin or ISG15 conjugation? Ubiquitin is a highly conserved housekeeping gene. It is ubiquitously and constitutively expressed and plays fundamental roles in many cellular processes (3,43). In contrast, ISG15 has been identified only in vertebrates, and its sequence conservation is much lower than that of ubiquitin (44), suggesting that ISG15 does not have essential housekeeping functions. ISG15 expression is almost undetectable under normal conditions and is strongly up-regulated during certain stress responses, especially type I IFN stimulation. Therefore, in the absence of ISG15, UBC8 and EFP can function as ubiquitylation enzymes. Upon the induction of ISG15 synthesis, they also serve as protein ISGylation enzymes. Many members of the ubiquitin E2 family are expressed in a given cell. These E2 enzymes can accept activated ubiquitin from UBE1. However, only a few of these members (UBC8 is the only currently identified member) can accept activated ISG15 from UBE1L. ISG15 competes with ubiquitin for its E2 when it is highly expressed. Furthermore, the spatial-temporal availability of UBC8-linked ubiquitin or ISG15 in vivo determines the specific functions of EFP and possibly other UBC8-interacting E3 ligases. The fact that both UBC8 and EFP are the common components in the ubiquitin and ISG15 conjugation pathways suggests a mechanism whereby a limited set of enzymes accomplishes diverse post-translational modifications of their substrates in  response to changes in environmental stimulations, resulting in distinct molecular consequences.
ISG15 expression and ISGylation are regulated by type I IFN. Expression of all known components of the ISGylation system, including ISG15, UBE1L, UBP43, and UBC8, is inducible by IFN (6,16,31). The promoters of these genes contain the ISRE, which is responsible for activation by type I IFN. In this study, we have also shown that EFP expression is increased upon IFN treatment and that the promoter region of EFP contains an ISRE (Fig. 1). EFP is also an estrogen-responsive gene (45). Although EFP protein is widely expressed, its level of expression is not uniformly high in various human tissues (46). It is possible that its expression in certain types of cells is heavily dependent on type I IFN signaling and that the major function of EFP in these cells is related to ISGylation of its substrates.
The function of ISG15 modification has been an enigma for many years. Although ISG15 has been implicated in a variety of biological activities based mostly on its similarities to ubiquitin and other ubiquitin-like proteins, the generation and analysis of ISG15-defficient mice (47) and ISG15 conjugation-deficient Ube1L knockout mice (48) did not reveal any significant developmental abnormalities, suggesting that ISG15 targeting is most likely a fine-tuning process that is induced by IFN. Very thorough studies must be performed to identify the exact consequences of ISGylation for some special target proteins. In this work, we have identified 14-3-3 as a new in vivo target of ISG15 modification. More important, we have demonstrated that EFP is an ISG15 E3 ligase for 14-3-3. Taking into account the large number and diversity of 14-3-3 ligands and the importance of 14-3-3 proteins in many cellular processes (49, 50), 14-3-3 is an attractive target for further study of the biological consequences of ISG15 modification. These findings will facilitate further analysis and accelerate our understanding of the biological effects of protein ISGylation.