HERC5 is an IFN-induced HECT-type E3 proteinligase that mediates type I IFN-inducedISGylation of protein targetsJoyce Jing Yi Wong*, Yuh Fen Pung*, Newman Siu-Kwan Sze†, and Keh-Chuang Chin*‡§¶

*Immunology and Virology Laboratory and †Proteomics Laboratory, Genome Institute of Singapore, 60 Biopolis Street, #02-01 Genome, Singapore 138672;‡Department of Microbiology and Immunology, Yong Loo Lin School of Medicine, Block MD4, 5 Science Drive 2, Singapore 117597; and §ImmunologyProgram, National University of Singapore, Office of Life Sciences Satellite Laboratory, Defense Medical and Environmental Research Institute Building,#03-01, 27 Medical Drive, Singapore 117510

Edited by Peter Cresswell, Yale University School of Medicine, New Haven, CT, and approved May 26, 2006 (received for review January 16, 2006)

Type I IFNs induce the expression of IFN-stimulated gene 15 (ISG15)and its conjugation to cellular targets. ISGylation is a multistepprocess involving IFN-inducible Ube1L, UbcH8, and a yet-to-beidentified E3 ligase. Here we report the identification of an IFN-induced HECT-type E3 protein ligase, HERC5�Ceb1, which mediatesISGylation. We also defined a number of proteins modified byISG15 after IFN triggering or HERC5 overexpression. A reduction inendogenous HERC5 by small interfering RNA inhibition blocks theIFN-induced ISG15 conjugation. Conversely, HERC5 coexpressionwith Ube1L and UbcH8 induces the ISG15 conjugation in vivoindependent of IFN stimulation. A targeted substitution of Cys-994to Ala in the HECT domain of HERC5 completely abrogates its E3protein ligase activity. Therefore, this study demonstrates thatHERC5�Ceb1 is involved in the conjugation of ISG15 to cellularproteins.

ISG15 � Ceb1 � innate immunity � antiviral proteins

Type I IFNs (IFN-���) play an essential role in both innateantiviral and adaptive immune responses and are rapidly pro-

duced in response to microbial infection (1–3). They exert signalsthrough the activation of the Janus kinase–signal transducer andactivator of transcription pathway that mediates rapid induction ofIFN-stimulated genes (ISGs) (4, 5). ISG15 is one of the moststrongly induced genes after IFN treatment (6, 7) and is alsosignificantly induced by viral infection (8, 9) and LPS treatment (10,11). The ISG15 protein starts with two ubiquitin-related domainsthat have �27% sequence identity to ubiquitin and terminate in aconserved 152LRLRGG157 ubiquitin C-terminal motif. This studysuggests that ISG15 could act in a similar way to ubiquitin and otherubiquitin-like proteins such as SUMO by forming an isopeptidebond with cellular proteins (6, 12, 13). The crystal structure ofISG15 revealed that ISG15 consists of two domains with ubiquitin-like folds joined by a linker sequence (14).

Conjugation of ISG15 to cellular proteins occurs in a parallel butdistinct mechanism to that of ubiquitin (15–17). The E1 enzyme forISG15, Ube1L, is a single-subunit enzyme and is identified in vitroby its ability to catalyze the formation of a thioester bond betweenISG15 and Ube1L (17, 18). The Ube1L protein is highly similar tothe E1 enzyme for ubiquitin at the protein level. However, thisprotein does not form a conjugate with ubiquitin, indicating thatUbe1L is an E1 enzyme for the ISG15 conjugation system (ISGy-lation). Influenza B virus blocks protein ISGylation by inhibitingthe activation step through the interaction of the NS1B viral proteinwith ISG15 (18). This finding was the first suggestion that ISGy-lation might be important for protecting cells from viral infection.

Two groups recently found that a member of the ubiquitinE2-conjugating enzyme family, UbcH8, is involved in the ISGyla-tion (19, 20). Like ISG15 and Ube1L, the expression of UbcH8 isalso induced by IFN (21). The suppression of UbcH8 proteinexpression by RNA interference is shown to dramatically inhibit the

total level of IFN-induced conjugation of ISG15 to cellular proteins(19, 20).

In the ubiquitin and ubiquitin-like proteins (ublps) system, E3enzymes play a critical role in transferring ubiquitin or ublps suchas SUMO from the E2-conjugating enzyme to a specific substrate.The substrate protein interacts directly with a specific domain in theE3 enzymes or through an adaptor associated with the E3 enzyme(22). E3 enzymes are known to be divided into two groups: HECT(homologous to E6-AP C terminus) and RING (really interestingnew genes) proteins (22–25). Whereas RING-type proteins transferubiquitin directly from an E2 to a target, the HECT-type proteinforms a thioester bond with ubiquitin or ublps by its active cysteineresidue before transferring it to a substrate (25). The identity of theE3 enzyme(s) in the ISG15 conjugation system has not beenresolved.

Here we report the identification of target proteins conjugatedwith ISG15 and present data supporting the contention that thecovalent modification of cellular proteins by ISG15 is regulatedby an IFN-inducible HECT domain-containing E3 protein li-gase, HERC5, which itself is also a target for modification byISG15. HERC5 catalyzes ISGylation by means of a catalyticcysteine residue at position 994 in the HECT domain. Moreover,HERC5 is found to be sufficient for inducing ISGylation in theabsence of IFN. Knock-down of HERC5 by small interferingRNA (siRNA) results in the inhibition of IFN-induced ISG15conjugation. Thus, HERC5 is an IFN-induced E3 protein ligasethat mediates ISGylation.

ResultsIdentification and Purification of FLAG–ISG15-Associated and�or-Modified Proteins. To search for potential E3 enzyme(s) as well asproteins that were physically associated or covalently conjugatedwith ISG15, we generated A549 cell lines stably expressing ISG15with two copies of FLAG tag (Fig. 1A). The FLAG tags wereinserted at the N terminus of ISG15 because of the involvement ofthe conserved C terminus diglycine motif in conjugation withcellular proteins (Fig. 1A). The FLAG–ISG15 retains its conjuga-tion activity to cellular proteins after 48 h of IFN-� treatment ascompared with untreated control (Fig. 7A, which is published assupporting information on the PNAS web site). These data indicatethat FLAG tags inserted at the N terminus of ISG15 do not disruptthe conjugation to cellular proteins after IFN-� treatment.

We next carried out large-scale isolation of ISG15-associated�

Conflict of interest statement: No conflicts declared.

This paper was submitted directly (Track II) to the PNAS office.

Freely available online through the PNAS open access option.

Abbreviations: siRNA, small interfering RNA; shRNA, short hairpin RNA; ISG, IFN-stimulatedgene; RLD, RCC1-like domain; HA, hemagglutinin.

¶To whom correspondence should be addressed. E-mail: chinkc1@gis.a-star.edu.sg.

© 2006 by The National Academy of Sciences of the USA

www.pnas.org�cgi�doi�10.1073�pnas.0600397103 PNAS � July 11, 2006 � vol. 103 � no. 28 � 10735–10740

IMM

UN

OLO

GY

Dow

nloa

ded

by g

uest

on

Janu

ary

25, 2

021

conjugated proteins by anti-FLAG immunoaffinity purification(Fig. 7B). Control A549 fractions were also analyzed to identifynonspecific, copurified proteins and hence eliminate false positives.A total of 174 unique candidate proteins, not found in controlfractions, and represented by at least two peptides in the tandemMS analysis, were identified. The ISG15 target proteins thatfulfilled these stringent criteria are listed in Table 1, which ispublished as supporting information on the PNAS web site.

Twenty-six target proteins were selected for further validation oftheir conjugation or association with ISG15. Twenty-four of 26putative ISG15 target proteins were found to form conjugates withISG15 after IFN-� treatment (Table 1, bold). Two other ISG15-copurified proteins, Hsp70 and pICln, were not directly conjugatedto ISG15 but formed a specific, noncovalent interaction (data notshown). Of the proteins that were substrates for ISG15 conjugation,a number, such as destrin and cofilin, were found to conjugate witha single ISG15 as they migrated at the expected molecular massesfor destrin*FLAG–ISG15 (Mr � 37 kDa) (Fig. 1B, lane 2) andcofilin*FLAG–ISG15 (Mr � 37 kDa) (Fig. 1C, lane 2). On the otherhand, some proteins, such as enolase and GAPDH, existed asmultiply conjugated species that were possibly modified by one ormore ISG15s (Fig. 1 D and E, lane 2). At least two isoforms ofenolase-conjugated species were observed, migrating at the posi-tions expected for one and three ISG15s (Fig. 1D, lane 2). Similarto enolase, GAPDH also displayed a number of conjugated speciesthat migrated at the positions expect for one and two ISG15s inSDS�PAGE (Fig. 1E, lane 2). In addition to conjugation, we alsoobserved noncovalent association between ISG15 and cellularsubstrates (Fig. 1 B–E, shown by arrowhead). Thus, we haveidentified a large number of cellular proteins that form either singleor multiple conjugation and noncovalent interactions with ISG15.

HERC5�Ceb1 Is Copurified with ISG15, and Its Expression Is Regulatedby Type I IFN. Conjugation of ISG15 with cellular proteins iscatalyzed by the E1 activating enzyme Ube1L and the E2-conjugating enzyme UbcH8 (17). Thus far, no E3 enzyme(s) hasbeen identified. In ubiquitin system, E3 enzymes, with either aHECT domain or a RING domain, play a critical role in recruitingthe ubiquitin-loaded E2, recognizing specific substrates and facili-tating the transfer of the ubiquitin from the E2 to the lysine residuein the substrates (22). Among all ISG15-associated�conjugatedcandidate proteins, we noticed that a protein known as HERC5 (orCeb1) contains a HECT-type domain at its C terminus (Fig. 8,which is published as supporting information on the PNAS web site)(26). As shown in Table 2, which is published as supporting

information on the PNAS web site, HERC5 was represented by fiveindividual peptides from tandem MS sequencing and was not foundin the control sample, indicating that this protein is specificallycopurified with FLAG–ISG15.

The primary sequence of HERC5 revealed the presence ofRCC1-like (regulator of chromosome condensation-1) and HECTdomains that span residues 209–258 and 676-1024, respectively (Fig.8A) (26). The HECT domain was previously demonstrated tointeract with a ubiquitin-conjugating enzyme (23). Within theHECT domain of HERC5, a conserved cysteine residue residing inall known HECT-type protein ligases was also found and located atposition 994 (Fig. 8B) (26). Because HERC5�Ceb1 possesses all ofthe unique structural features of a mammalian E3 protein ligase, wehypothesized that HERC5 functions as an E3 protein ligase inISGylation.

Because HERC5 is identified by the pulling down of ISGylatedproteins and ISG15-associated proteins, we first determinedwhether ISG15 binds HERC5 or forms covalent conjugated specieswith HERC5. HeLa cells were cotransfected with FLAG–HERC5and Myc–ISG15 (Fig. 2A, lanes 1–4), and cells were treated with(Fig. 2A, lanes 3 and 4) or without (Fig. 2A, lanes 1 and 2) IFN-�.We performed either anti-FLAG (Fig. 2A, lanes 2 and 4) or control(Fig. 2A, lanes 1 and 3) immunoprecipitation followed by anti-Mycor anti-FLAG Western blotting to determine whether higher-molecular-weight forms of HERC5 or free forms of ISG15 couldbe detected. As shown in Fig. 2A, we saw multiple species of highmolecular forms of HERC5 in IFN-�-treated cells expressing bothFLAG–HERC5 and Myc–ISG15 by anti-Myc and anti-FLAGWestern blotting, respectively. Moreover, we saw free forms ofISG15 copurified with HERC5 in both untreated and IFN-�-treated cells (Fig. 2A Left). These data indicate that HERC5physically associates with ISG15 and itself is a target for modifica-tion by ISG15.

We then sought to examine whether type I IFN induces HERC5mRNA expression in both HeLa and A549 cell lines because theexpression of Ube1L and UbcH8 has previously been shown to beinduced by IFN-� (21). Both HeLa and A549 cells were treated withIFN for 6, 12, 24, and 48 h. As shown in Fig. 2B, we detected basallevels of HERC5 mRNA in both untreated HeLa (Left) and A549(Right) cells. Upon IFN-� stimulation, HERC5 mRNA was rapidlyinduced in both cell lines after 6 h (Fig. 2B). At 12 h there were�30-fold increments in HERC5 mRNA expression levels comparedwith their respectively untreated controls for both HeLa and A549cells (Fig. 2B). The mRNA for HERC5 was continuously expressedat high levels at 24 and 48 h after IFN-� treatment as compared with

Fig. 1. Identification and purification of ISG15-associated and�or -modified proteins. (A) Schematic representation of the ISG15 with two copies of FLAG atthe N terminus. Mature ISG15 terminates in a conserved GG (bold), and the arrow indicates the cleavage site. ISG15 protein contains two ubiquitin-like domainsand is therefore predicted to be �15 kDa. Indeed, ISG15 protein runs as 15 kDa in SDS�PAGE. (B–E) Validation of ISG15 conjugation of destrin, cofilin, enolase,and GAPDH. HeLa cells were transfected with plasmids encoding FLAG–ISG15 and treated with IFN-� for 48 h. Total lysates were immunoprecipitated withanti-FLAG M2-conjugated agarose and immunoblotted with antibodies against destrin, cofilin, enolase, and GAPDH.

10736 � www.pnas.org�cgi�doi�10.1073�pnas.0600397103 Wong et al.

Dow

nloa

ded

by g

uest

on

Janu

ary

25, 2

021

the controls (Fig. 2B). We also measured the induction of ISG15conjugates in both HeLa (Fig. 2C Left) and A549 (Fig. 2C Right)cells upon IFN-� treatment. In contrast to the kinetics of HERC5mRNA expression, ISG15 protein conjugates were only detected inboth HeLa and A549 cells at 24 h after IFN-� treatment (Fig. 2C,lane 4) and were continually observed at 48 h (Fig. 2C, lane 5).Together, these data indicated that HERC5 is an IFN-inducibleprotein. Also, the strong induction of HERC5 mRNA expression byIFN-� before ISGylation is paralleled to the observations made forUbe1L and UbcH8. Thus, we hypothesized that HERC5 couldfunction as a protein ligase in concert with Ube1L and UbcH8 topromote ISGylation upon IFN-� treatment.

Depletion of Endogenous HERC5 Inhibits ISGylation upon Type I IFNTreatment. To investigate the role of HERC5 in ISGylation, we usedsiRNA technology to suppress the expression of endogenousHERC5 mRNA in A549 cells. Two small siRNAs, namely siRNA–HERC5-I and siRNA–HERC5-II, which target nucleotide se-quences 536–554 bp and 715–735 bp, respectively, in HERC5 wereused. A549 cells were transfected with either control siRNA or

siRNAs against HERC5. After 48 h of IFN-� treatment, totallysates were made and separated in SDS�PAGE. ISG15 conjugateswere detected by immunoblotting with an antibody against endog-enous ISG15. A dramatic reduction in the expression of ISG15conjugates was observed in A549 cells transfected with the twosiRNAs against HERC5 compared with control siRNA (Fig. 3A).As shown in Fig. 3B, both the siRNAs against HERC5 were effectivein reducing HERC5 mRNA expression by 80% after 24 h of IFN-�treatment (Fig. 3B), but not its closest homologue, HERC6 (Fig.9A, which is published as supporting information on the PNAS website). It is crucial to note that the reduction in HERC5 mRNAexpression had no effect on the expression of ISG15 (Fig. 3A),Ube1L, or UbcH8 (Fig. 3C) protein levels. Thus, these resultsindicate that HERC5 is a major E3 protein ligase whose inductionby IFN-� is necessary for catalyzing ISGylation in A549 cells.

Next we sought to test whether the inducibility of endogenousISG15 conjugates by IFN-� in HeLa cells is affected by a shorthairpin RNA (shRNA)-mediated reduction in HERC5. A shRNAexpression construct, designated as shRNA–HERC5–1-4, was gen-erated. This shRNA was chosen to target a separate region inHERC5, 1,606–1,639 bp. HeLa cells were transfected with shRNA–HERC5–1-4, and cells stably expressing shRNA against HERC5were then selected by puromycin treatment. The expression ofHERC5 mRNA after IFN-� treatment was examined by real-timePCR using primers specific for HERC5. HERC5 mRNA was foundto be significantly reduced in HeLa cells constitutively expressingshRNA against HERC5 compared with control cells after IFN-�treatment (Fig. 3D), but not its closest homologue, HERC6 (Fig.9B). The reduction of HERC5 expression in HeLa cells alsosignificantly impaired the induction of endogenous ISG15 conju-gates after 48 h of IFN-� treatment (Fig. 3E). Taken together, theseresults indicate that HERC5 plays an essential role in catalyzingIFN-�-induced ISGylation in both HeLa and A549 cells.

Coexpression of HERC5 with Ube1L and UbcH8 Restores ISG15 Conju-gates in Vivo Without Type I IFN for Induction. Because HERC5expression is induced by IFN, we tested whether overexpression ofHERC5 in combination with Ube1L and UbcH8 would lead toISGylation in the absence of IFN-�. HeLa cells transfected witheither Ube1L (Fig. 4A, lane 2) or UbcH8 (Fig. 4A, lane 3) expressedno ISG15 conjugates. A partial expression of ISG15 conjugates wasobserved in HeLa cells transfected with Ube1L and UbcH8 (Fig.4A, lane 4), which may have been because of a low level of HERC5expression in HeLa cells (refer to RNA work in Fig. 2B Left). WhenHERC5 is overexpressed, the production of ISG15 conjugates byHeLa cells is dramatically enhanced (Fig. 4A, lanes 5). These datastrongly implicated HERC5 as an E3 protein ligase whose overex-pression is sufficient to drive ISGylation.

Next we examined a number of proteins known to be ISGy-lated after IFN-� treatment, for conjugation with ISG15 in HeLacells overexpressing HERC5. HeLa cells were transfected withFLAG–ISG15 alone (Fig. 4 B–I, lane 1), FLAG–ISG15 plusvectors expressing Ube1L and UbcH8 (Fig. 4 B–I, lane 2), orHERC5, Ube1L, and UbcH8 (Fig. 4 B–I, lane 3). Cells wereharvested after 48 h. FLAG–ISG15 conjugates were immuno-precipitated with anti-FLAG antibody. The protein complexeswere separated in SDS�PAGE and detected with specific anti-bodies. All of the target proteins examined were found to becovalently conjugated with ISG15 when HERC5 is overex-pressed (Fig. 4 B–I, lanes 3). Three target proteins, destrin,enolase, and GAPDH, were observed to have low levels ofconjugation in HeLa cells transfected with Ube1L and UbcH8alone (Fig. 4 B, D, and E, lane 2). However, their levels ofconjugation were dramatically enhanced in cells transfected withHERC5, Ube1L, and UbcH8 (Fig. 4 B, D, and E, lane 3).Interestingly, the presence of multiple conjugated species forboth enolase and GAPDH were also observed after HERC5overexpression, indicating that HERC5 is essential for both

Fig. 2. HERC5 is an IFN-inducible protein and is covalently modified withISG15. (A) HERC5 interacts with ISG15 and forms conjugated species withISG15. HeLa cells were cotransfected with FLAG–HERC5 and Myc–ISG15, andcells were treated (lanes 3 and 4) or not treated (lanes 1 and 2) with IFN-�. Totallysates were prepared and immunoprecipitated with either control (lanes 1and 3) or FLAG (lanes 2 and 4) antibody and then separated on SDS�PAGE. Theprotein complexes were detected with either anti-Myc (Left) or anti-FLAG(Right) antibody. (B) Kinetics of IFN-induced HERC5 mRNA expression in HeLa(Left) and A549 (Right) cells. HeLa and A549 cells were either untreated ortreated with IFN-� for 6, 12, 24, and 48 h. Total RNAs were isolated, andreal-time PCR was performed with HERC5-specific primers. (C) Induction ofISG15 conjugation in HeLa (Left) and A549 (Right) cells after IFN-� treatment.The unstimulated and IFN-stimulated HeLa and A549 cells were harvested at6, 12, 24, and 48 h. Total lysates were separated in SDS�PAGE and immuno-blotted with anti-ISG15 antibody.

Wong et al. PNAS � July 11, 2006 � vol. 103 � no. 28 � 10737

IMM

UN

OLO

GY

Dow

nloa

ded

by g

uest

on

Janu

ary

25, 2

021

single and multiple ISGylation in target proteins (Fig. 4 D andE, lane 3). In contrast to enolase and GAPDH, the other fivetarget proteins tested (cofilin, Hsp27, Hsc70, peroxiredoxin I,and peroxiredoxin VI) displayed no detectable ISGylation inHeLa cells transfected with only Ube1L and UbcH8 (Fig. 4 Cand F–I, lane 2). These proteins all exhibited significant levels ofISG15 conjugation in the HeLa cells overexpressed with HERC5(Fig. 4 C and F–I, lane 3). Together, these data indicate thatHERC5 is an E3 protein ligase involved in ISGylation. Bothsingle and multiple conjugations in target proteins require theexpression of HERC5. Once expressed, the enzymatic activity ofHERC5 is not dependent on type I IFNs.

Cys-994 Within the HECT Domain of HERC5 Is Essential for Its ProteinLigase Activity in ISGylation. All HECT-type ubiquitin proteinligases contain a conserved active cysteine residue that forms an

intermediate bond with ubiquitin before transferring it to thesubstrates (Fig. 8B). To further demonstrate that HERC5 is anE3 protein ligase for ISGylation, the putative active cysteineresidue (C994), based on conservation with other HECT do-main-containing E3s, was mutated to an alanine residue bysite-specific mutagenesis. HeLa cells were cotransfected withMyc–ISG15 and either wild-type FLAG–HERC5 (Fig. 5, lane 4)or mutant FLAG–HERC5 (Fig. 5, lane 5). This point mutationin the cysteine residue (C994) completely abolishes its E3 activityin ISGylation (Fig. 5 Upper, compare lanes 4 and 5). Theexpression levels of wild-type and mutant HERC5 proteins werefound to be comparable (Fig. 5 Lower). These data demonstratethat HERC5 acts as an E3 protein ligase in ISGylation by meansof a conserved cysteine residue, C994, found in the HECTdomain.

Fig. 3. Depletion of HERC5 inhibits the induction of ISG15 conjugation by IFN-�. (A) HERC5 depletion by siRNAs inhibits ISG15 conjugation induced by IFN-�in A549 cells. The A549 cells were transfected with control siRNA (lane 1) or siRNAs against HERC5 (lanes 2 and 3). Twenty-four hours after transfection, A549cells were treated with IFN-� for another 48 h. Total lysates were harvested and separated in SDS�PAGE. ISG15 conjugates were detected by anti-ISG15 antibody.(B) Suppression of HERC5 mRNA expression in A549 cells by siRNAs. Efficiency of siRNA–HERC5-I and siRNA–HERC5-II in inhibition of HERC5 RNA expression wasdetermined in A549 cells transfected with siRNA control and siRNAs for HERC5 24 h after IFN-� treatment. Expression of HERC5 mRNA was analyzed by real-timePCR with specific primers for HERC5. (C) HERC5 elimination has no effect on the induction of Ube1L and UbcH8 by type I IFN. Total lysates were prepared fromA549 cells transfected with control siRNA (lane 1) or siRNAs against HERC5 (lanes 2 and 3) at 24 h after IFN-� treatment. The levels of protein expression of Ube1Land UbcH8 were detected by antibodies against Ube1L (C Upper) and UbcH8 (C Lower), respectively. (D) Suppression of IFN-�-induced HERC5 mRNA expressionin HeLa cells constitutively expressing shRNA–HERC5–1-4. Efficiency of shRNA–HERC5–1-4 in inhibition of endogenous HERC5 RNA expression was determinedby using real-time PCR with HERC5-specific primers. (E) Inhibition of IFN-�-induced ISG15 conjugation in HeLa cells stably expressing shRNA against HERC5. HeLacells were stably transfected with vectors expressing shRNA–HERC5–1-4 and selected with 1 �g�ml puromycin. Cells were treated with IFN-� for 48 h, and ISG15conjugates were analyzed by immunoblotting with anti-ISG15 antibody.

Fig. 4. HERC5, together with Ube1L and UbcH8, induces the FLAG–ISG15 conjugation in HeLa cells without type I IFN treatment. (A) HeLa cells were transfectedwith expression vector encoding for FLAG–ISG15 plus vectors encoding Ube1L (lane 2), UbcH8 (lane 3), Ube1L and UbcH8 (lane 4), or Ube1L, UbcH8, and HERC5(lanes 5). Twenty-four hours after transfection, total lysates were prepared, and FLAG–ISG15 conjugates were detected with anti-FLAG antibody. (B–I) HeLa cellswere transfected with expression vector encoding for FLAG–ISG15 (lane 1), FLAG–ISG15 plus vector encoding Ube1L and UbcH8 (lane 2), or FLAG–ISG15 constructtogether with vectors encoding Ube1L, UbcH8, and HERC5 (lanes 3). Total lysates were harvested, FLAG–ISG15 conjugates were immunoprecipitated withanti-FLAG M2-conjugated agarose, and protein complexes were separated in SDS�PAGE and then immunoblotted with antibodies against destrin (B), cofilin (C),enolase (D), GAPDH (E), Hsp27 (F), Hsc70 (G), peroxiredoxin I (H), and peroxiredoxin VI (I).

10738 � www.pnas.org�cgi�doi�10.1073�pnas.0600397103 Wong et al.

Dow

nloa

ded

by g

uest

on

Janu

ary

25, 2

021

HERC5 Interacts with ISGylated Substrates, HSC70, and ThioredoxinReductase. To further support that HERC5 is a protein ligase forISG15, we next tested whether HERC5 interacts with ISGylatedsubstrates, Hsc70, and thioredoxin reductase. HeLa cells weretransfected with FLAG–HERC5 (Fig. 6, lane 3), ISGylated sub-strate (Fig. 6, lane 2), FLAG–HERC5 plus HA-Hsc70 (Fig. 6A,lane 4) or HA–thioredoxin reductase (Fig. 6B, lane 4). Cell extractswere immunoprecipitated with an anti-FLAG monoclonal anti-body. The immunoprecipitates were then subjected to SDS�PAGEand immunoblotted with antibodies against FLAG or hemaggluti-nin (HA). As shown in Fig. 6, the antibody against FLAG precip-itated FLAG–HERC5 in cells expressing FLAG–HERC5 (Fig. 6Middle, lanes 3 and 4), but not in the control cells (Fig. 6 Middle,lanes 1 and 2). The FLAG antibody also precipitated Hsc70 (Fig.6A Top, lane 4) or thioredoxin reductase (Fig. 6B Top, lane 4). Theexpression of cellular substrates was the same by immunoblottingcell lysates with an HA antibody (Fig. 6 Bottom, lanes 2 and 4).These results suggest that HERC5 acts as an E3 protein ligase for

ISGylation by recruiting cellular substrates for modification byISG15, which is also found to be associated with HERC5 (Fig. 2A).

DiscussionHERC5 Is an IFN-Inducible Protein Ligase for ISG15 Conjugation Sys-tem. We have described the properties of HERC5 identifiedthrough its interaction with ISG15. This interaction was detected byusing an affinity-based purification scheme involving an antibodyagainst FLAG–ISG15 in A549 cells treated with IFN-�. HERC5 isa member of a group of related proteins known as the HERC family(27, 28). The human HERC family currently consists of six mem-bers, HERC1 to HERC6, and is characterized by the presence ofa HECT domain and one or more RCC1-like domains (RLDs).Proteins with a HECT domain are now thought to act as E3 proteinligase for ubiquitin and ublps, such as E6-AP and Nedd4 (29). Herewe have provided evidence suggesting that HERC5 is an IFN-inducible E3 protein ligase that is required and sufficient for theISG15 conjugation system in vivo. First, a knock-down of HERC5expression by RNA interference technology completely inhibitedthe induction of ISG15 conjugation induced by IFN. Second,HERC5 coexpression with the Ube1L and UbcH8 suffices tomediate ISG15 conjugation in vivo even if in the absence of IFNtreatment. Third, a mutation of a conserved cysteine residue in theHECT domain of HERC5 to an alanine residue abolished itsprotein ligase activity in vivo. This result is consistent with thehypothesis that C994 is important for transferring ISG15 fromUbcH8 to specific target proteins. Fourth, HERC5 interacts withISG15 and ISGylated substrates such as Hsc70 and thioredoxinreductase. The ability of HERC5 to interact with both ISG15 andcellular substrate is consistent with the hypothesis that HERC5could act as an E3 protein ligase for ISGylation. In addition, we findthat knock-down of Ube1, E1 enzyme for the ubiquitination system,has no affect on the ISGylation, and this result rules out thepossibility that HERC5 acts an E3 enzyme for ubiquitination byactivating an uncharacterized E3 enzyme for ISGylation and in-ducing ISGylation (Fig. 10, which is published as supporting infor-mation on the PNAS web site). The finding that HERC5 is involvedin ISGylation in this study was also demonstrated by an independentgroup while this article was under review (30). One of the chal-lenging tasks is to express HERC5 for promotion of ISGylation invitro; however, we and Dastur et al. (30) had some difficultyexpressing either the full-length or the HECT domain of HERC5in the bacterial system, and this issue remains to be resolved.

The N-terminal region of HERC5 possesses at least one RLD.RLDs have been shown to interact with several proteins, and it ispossible that the RLDs serve as the regions for recruiting specificprotein substrates. In addition, the middle region, between RLDsand the HECT domain of HERC5, displays no obvious homologyto any known proteins. This region may also serve a role inrecognizing target proteins. Further structure–function studies inHERC5 will be needed to provide insight into the specificity ofHERC5 in recognizing its target proteins.

Target Proteins Modified by HERC5. By using a combination ofaffinity purification and MS, we identified 174 candidate proteinsthat were covalently conjugated or interacted with ISG15 upon IFNtreatment. Of 27 target proteins examined, 24 were identified toconjugate with ISG15, and the other three proteins were found tointeract with ISG15 (Table 1 and data not shown). Given that mostof the candidates we chose to pursue were modified by the ISG15conjugation pathway, we are highly confident that our large data setrepresents bona fide substrates for ISG15 in vivo.

Eight putative IFN-inducible ISG15 targets (MxA, IFIT1, signaltransducer and activator of transcription 1, Ube1L, tryptophanyltRNA synthetase, TRIM21, gelsolin, and HERC5) are identified(4, 5). We chose four of the IFN-induced proteins (MxA, IFIT1,signal transducer and activator of transcription 1, and tryptophanyltRNA synthetase) and verified that they are ISG15-conjugated in

Fig. 5. Conserved Cys-994 within the HECT domain of HERC5 is essential forits ligase activity in ISGylation. Cys-994 within the HECT domain of HERC5 wasmutated to alanine residue. HeLa cells were transfected with vectors express-ing 3Myc–ISG15 (lane 2); 3Myc–ISG15, Ube1L, and UbcH8 (lane 3); 3Myc–ISG15,Ube1L, UbcH8, and wild-type FLAG–HERC5 (lane 4); and 3Myc–ISG15, Ube1L,UbcH8, and FLAG–HERC5–C994A (lane 5). Forty-eight hours after transfection,cell lysates were collected, and Myc–ISG15 conjugates were detected byimmunoprecipitation followed by immunoblotting with anti-Myc antibody(Upper). The expression of wild-type and mutant HERC5 was detected withanti-FLAG antibody (Lower).

Fig. 6. HERC5 interacts with ISGylated substrates, Hsc70, and thioredoxinreductase. (A) HeLa cells were transfected with HA-Hsc70 (lane 2), FLAG–HERC5 (lane 3), or FLAG–HERC5 plus HA-Hsc70 (lane 4). Total lysates weremade and immunoprecipitated with anti-FLAG agarose. The immunocom-plexes were separated and immunoblotted with anti-HA (Top) for detectionof Hsc70 or anti-FLAG (Middle) for HERC5. The total amount of Hsc70 wasconfirmed by immunoblotting of cell lysate with anti-HA antibody (Bottom).(B) Same as above, except that the HA-tagged form of thioredoxin reductase(HA-THR) was used in the transfection.

Wong et al. PNAS � July 11, 2006 � vol. 103 � no. 28 � 10739

IMM

UN

OLO

GY

Dow

nloa

ded

by g

uest

on

Janu

ary

25, 2

021

IFN-�-treated cells. The biological consequences of ISG15 modi-fication or its association with these proteins is unclear. Previousimmunofluorescence and biochemical studies have shown thatISG15 and its conjugates can be located on the cytoskeleton(31–33). One possible mechanism could be to move IFN-inducedproteins from their normal location to the cytoskeleton, where theyexert their antiviral activity by inhibiting either the trafficking ofviral proteins or the exocytosis of viral particles.

While we were studying the role of HERC5 in ISG15 conjugationsystem, two groups independently reported using a proteomicapproach to identify ISG15-modified proteins from HeLa cells andU937- or UBP43-deficient mouse embryonic fibroblasts (34, 35). Of171 candidates identified in our study, only 61 proteins were foundin the lists reported by the other two groups, and thus the rest ofcandidates represent novel substrates for ISG15 modification. Sev-eral proteins identified in our screening but not in the other twogroups were examined, and they were found to be conjugated withISG15. These proteins are 14-3-3�, destrin, peroxiredoxin VI, andEBNA2 coactivator.

In conclusion, we have identified a novel factor involved inISGylation, HERC5, which acts as an ISG15 E3 protein ligase. Thisactivity is strictly dependent on a conserved Cys residue at position994 within the HECT domain. We also report the identification ofboth constitutively and IFN-induced cellular proteins that aresubstrates for modification by ISG15. The ISGylation of the iden-tified protein targets was observed after type I IFN treatment or inthe presence of HERC5 overexpression.

Materials and MethodsGeneration of FLAG–ISG15, Myc–ISG15, UbcH8, Ube1L, and HERC5Expression Constructs. cDNA encoding for ISG15 was amplifiedfrom a human splenic cDNA library purchased from CLONTECH.Either two copies of FLAG tag or three copies of Myc tag wereinserted at the N terminus of ISG15 and subcloned into pQXIXexpression vector (CLONTECH). UbcH8 was also cloned by PCRfrom the human splenic cDNA library and was further subclonedinto pLNCX2 expression vector (CLONTECH). Ube1L cDNA wasa gift from E. Dmitrovsky (Dartmouth Medical School, Hanover,NH). Its cDNA was amplified by PCR and subcloned into pQXIXexpression vector (CLONTECH). cDNA for HERC5 was kindlyprovided by Motoaki Ohtsubo (Kurume University, Fukuoka-ken,Japan), and an insertion of a nucleotide A at position 437 inHERC5 was deleted by site-directed mutagenesis.

Generation of A549 Cells Stably Expressing FLAG–ISG15. A549 cellswere transfected with expression construct encoding FLAG–ISG15by Lipofectamine 2000 (Invitrogen). Cells were selected by using500 �g�ml G418 and were further screened for FLAG–ISG15expression by immunoblotting with anti-FLAG antibody.

Purification of ISG15 Associated and�or Modified Proteins. A549 cellsstably expressing FLAG–ISG15 (5 � 109) were treated with 500units�ml IFN-�. Cells were lysed in 200 ml of 1% Triton X-100 in10 mM Tris and 150 mM NaCl (pH 7.4) (TBST) on ice for 1 h. Afterremoving nuclei and cell debris by centrifugation, crude lysate wasfirst subjected to a mouse IgG-conjugated agarose column (Sigma)followed by an anti-FLAG M2-conjugated agarose column (Sigma)at 4°C, respectively. ISG15-associated and�or -modified proteinswere eluted with 3.5 M MgCl2 after washing extensively with 1%TBST buffer. Portions of the fractions collected were separated bySDS�PAGE and were further detected by using an anti-FLAGantibody. Fractions containing FLAG–ISG15 were pooled andconcentrated by using Centricon (Amicon).

MS Analysis. MS analysis was performed in-house at the proteomicfacility of the Genome Institute of Singapore. In brief, the sampleswere reduced and alkylated with iodoacetamide, digested withtrypsin, and subjected to LC-MS�MS analysis on an LCQ Deca PlusIon-trap mass spectrometer. Peptide masses were queried againstentries in the International Protein Index human database by usingMASCOT (Matrix Science). For a protein to be considered as a ‘‘hit,’’a minimum of two matching peptides was required.

Verification of ISG15 Target Proteins. HeLa cells were plated on100-mm dishes (2.2 � 106) and transfected by using Lipo-fectamine 2000. After 4 h, dishes were washed and replaced withcomplete DMEM-H media either with or without IFN-� (500units�ml). The cells were harvested after 48 h. Total lysates wereextracted for Western blotting or immunoprecipitation withanti-FLAG antibody. The sequences of siRNAs and antibodiesused can be found in Fig. 10. For more information, seeSupporting Materials and Methods, which is published as sup-porting information on the PNAS web site.

We thank Sukjo Kang, P. A. Macary, and Stephen Ogg for comments onthe manuscript.

1. Decker, T., Muller, M. & Stockinger, S. (2005) Nat. Rev. Immunol. 5, 675–687.2. Perry, A. K., Chen, G., Zheng, D., Tang, H. & Cheng, G. (2005) Cell Res. 15, 407–422.3. Smith, P. L., Lombardi, G. & Foster, G. R. (2005) Mol. Immunol. 42, 869–877.4. Levy, D. E. & Darnell, J. E., Jr. (2002) Nat. Rev. Mol. Cell Biol. 3, 651–662.5. Stark, G. R., Kerr, I. M., Williams, B. R., Silverman, R. H. & Schreiber, R. D. (1998) Annu.

Rev. Biochem. 67, 227–264.6. Haas, A. L., Ahrens, P., Bright, P. M. & Ankel, H. (1987) J. Biol. Chem. 262, 11315–11323.7. Korant, B. D., Blomstrom, D. C., Jonak, G. J. & Knight, E., Jr. (1984) J. Biol. Chem. 259,

14835–14839.8. Nicholl, M. J., Robinson, L. H. & Preston, C. M. (2000) J. Gen. Virol. 81, 2215–2218.9. Kim, M. J., Latham, A. G. & Krug, R. M. (2002) Proc. Natl. Acad. Sci. USA 99, 10096–10101.

10. Ritchie, K. J., Malakhov, M. P., Hetherington, C. J., Zhou, L., Little, M. T., Malakhova,O. A., Sipe, J. C., Orkin, S. H. & Zhang, D. E. (2002) Genes Dev. 16, 2207–2212.

11. Doyle, S., Vaidya, S., O’Connell, R., Dadgostar, H., Dempsey, P., Wu, T., Rao, G., Sun, R.,Haberland, M., Modlin, R. & Cheng, G. (2002) Immunity 17, 251–263.

12. Ritchie, K. J. & Zhang, D. E. (2004) Semin. Cell Dev. Biol. 15, 237–246.13. Blomstrom, D. C., Fahey, D., Kutny, R., Korant, B. D. & Knight, E., Jr. (1986) J. Biol. Chem.

261, 8811–8816.14. Narasimhan, J., Wang, M., Fu, Z., Klein, J. M., Haas, A. L. & Kim, J. J. (2005) J. Biol. Chem.

280, 27356–27365.15. Huang, D. T., Walden, H., Duda, D. & Schulman, B. A. (2004) Oncogene 23, 1958–1971.16. Loeb, K. R. & Haas, A. L. (1992) J. Biol. Chem. 267, 7806–7813.17. Dao, C. T. & Zhang, D. E. (2005) Front. Biosci. 10, 2701–2722.18. Yuan, W. & Krug, R. M. (2001) EMBO J. 20, 362–371.19. Zhao, C., Beaudenon, S. L., Kelley, M. L., Waddell, M. B., Yuan, W., Schulman, B. A.,

Huibregtse, J. M. & Krug, R. M. (2004) Proc. Natl. Acad. Sci. USA 101, 7578–7582.20. Kim, K. I., Giannakopoulos, N. V., Virgin, H. W. & Zhang, D. E. (2004) Mol. Cell. Biol. 24,

9592–9600.

21. Nyman, T. A., Matikainen, S., Sareneva, T., Julkunen, I. & Kalkkinen, N. (2000) Eur.J. Biochem. 267, 4011–4019.

22. Pickart, C. M. & Eddins, M. J. (2004) Biochim. Biophys. Acta 1695, 55–72.23. Scheffner, M., Huibregtse, J. M., Vierstra, R. D. & Howley, P. M. (1993) Cell 75,

495–505.24. Huibregtse, J. M., Scheffner, M., Beaudenon, S. & Howley, P. M. (1995) Proc. Natl. Acad.

Sci. USA 92, 5249.25. Joazeiro, C. A. & Weissman, A. M. (2000) Cell 102, 549–552.26. Mitsui, K., Nakanishi, M., Ohtsuka, S., Norwood, T. H., Okabayashi, K., Miyamoto, C.,

Tanaka, K., Yoshimura, A. & Ohtsubo, M. (1999) Biochem. Biophys. Res. Commun. 266,115–122.

27. Garcia-Gonzalo, F. R. & Rosa, J. L. (2005) Cell. Mol. Life Sci. 62, 1826–1838.28. Hochrainer, K., Mayer, H., Baranyi, U., Binder, B., Lipp, J. & Kroismayr, R. (2005)

Genomics 85, 153–164.29. Mueller, D. L. (2004) Nat. Immunol. 5, 883–890.30. Dastur, A., Beaudenon, S., Kelley, M., Krug, R. M. & Huibregtse, J. M. (2006) J. Biol. Chem.

281, 4334–4338.31. Austin, K. J., Carr, A. L., Pru, J. K., Hearne, C. E., George, E. L., Belden, E. L. & Hansen,

T. R. (2004) Endocrinology 145, 967–975.32. Loeb, K. R. & Haas, A. L. (1994) Mol. Cell. Biol. 14, 8408–8419.33. Bebington, C., Bell, S. C., Doherty, F. J., Fazleabas, A. T. & Fleming, S. D. (1999) Biol.

Reprod. 60, 920–928.34. Zhao, C., Denison, C., Huibregtse, J. M., Gygi, S. & Krug, R. M. (2005) Proc. Natl. Acad.

Sci. USA 102, 10200–10205.35. Giannakopoulos, N. V., Luo, J. K., Papov, V., Zou, W., Lenschow, D. J., Jacobs, B. S.,

Borden, E. C., Li, J., Virgin, H. W. & Zhang, D. E. (2005) Biochem. Biophys. Res. Commun.336, 496–506.

10740 � www.pnas.org�cgi�doi�10.1073�pnas.0600397103 Wong et al.

Dow

nloa

ded

by g

uest

on

Janu

ary

25, 2

021