High-throughput Immunoblotting UBIQUITIN-LIKE PROTEIN ISG15 MODIFIES KEY REGULATORS OF SIGNAL TRANSDUCTION*

ISG15 is a ubiquitin-like protein that conjugates to numerous proteins in cells treated with interferon or lipopolysaccharide. Dysregulation of protein ISG15 modification (ISGylation) in mice leads to decreased life expectancy, brain cell injury, and hypersensitivity to interferon. Although ISG15 was identified more than two decades ago, the exact biochemical and physiological functions of ISG15-modification remain unknown, and the proteins targeted by ISG15 have not been identified. The major purpose of this work was to identify ISG15 targets among well characterized proteins that could be used as models for biological studies. We purified ISGylated proteins from human thymus by immunoaffinity chromatography and analyzed ISG15 conjugates by a high-throughput Western blot screen (PowerBlot TM ). We found that three key regulators of signal transduction, phospholipase C (cid:1) 1, Jak1, and ERK1 are modified by ISG15. In addition to that, we demonstrate that transcription factor Stat1, an immediate substrate of Jak1 kinase, is also ISGylated. Using whole cell protein extracts and phospholipase C (cid:1) 1 as an example we demonstrate that ISG15 conjugates are not accumu-lated in cells treated with specific inhibitors of proteasomes. ISG15

ISG15 is one of the most strongly induced genes after interferon (IFN) 1 treatment (1,2) and is also significantly induced by viral infection (3) and lipopolysaccharide (LPS) (4). The sequence of ISG15 protein was noted to possess significant homology to a diubiquitin sequence, accounting for its cross-reactivity with some anti-ubiquitin antibodies (3). Several reports demonstrate that ISG15 is released by various cell types and can act as a cytokine leading to proliferation of NK cells (5). It has also been demonstrated that ISG15 is induced in the uterine endometrium during early pregnancy and was suggested to play a significant role in embryo implantation (6). ISG15 sequences are absent in yeast, nematode (Caenorhabditis), plant (Arabidopsis), and insect (Drosophila) indicating that the ISG15 conjugation system is restricted to higher animals with evolved IFN signaling.
Most remarkably, ISG15 was found to be conjugated to intracellular proteins via an isopeptide bond in a manner similar to ubiquitin (Ub) and other Ub-like proteins (Ubls) such as SUMO and Nedd8 (7). Conjugation of Ubls involves a threestep mechanism whereby specific enzymes (or enzyme complexes) activate and covalently link Ubls to their substrates (8,9). ISG15 conjugation occurs via a similar but distinct pathway compared with Ub conjugation (10), and an activating enzyme for ISG15 has recently been rediscovered as specific to ISG15 and not Ub (11). Similar to modification by other Ubls, the conjugation of ISG15 is reversible and is accomplished by a highly specific protease UBP43 (12). Based on sequence homology, UBP43 belongs to a family of Ub-specific proteases and was initially cloned in our laboratory from leukemogenic AML-ETO knock-in mice (13,14). UBP43 has since been cloned independently by other groups (15)(16)(17). UBP43 is induced by IFN, LPS, and viral infection and, similarly to the ISG15 gene, is regulated via p38 MAPK pathway and IFN regulatory factor 3 (18). A UBP43 Ϫ/Ϫ mouse model generated in our laboratory exhibits a massive accumulation of ISG15 conjugates in various tissues, which leads, either directly or indirectly, to decreased life expectancy, brain cell injury (19), and hypersensitivity to IFN stimulation (20). Although the role of Ub-, Nedd8-, and SUMO-modification has been assessed in numerous studies (8,21,22), the exact biochemical and physiological role of ISGylation has not been studied.
A major reason for the undefined biochemical function of ISG15 conjugation is the lack of known ISG15 targets that may serve as experimental models. Only one target, serpin 2a, has recently been reported, but its exact function is not known (23). In this study we used a large-scale Western blot-based screening process to identify new targets of ISG15 modification from a group of well characterized signal transduction proteins.
Here we report the first systematic attempt to establish the identity of ISGylated proteins. We show that several key regulators of signal transduction (namely, PLC␥1, Jak1, Stat1, and ERK1) are targeted by ISG15. Our data describes valid methodology for identification of new ISG15 targets, offers attractive models to study the function of ISGylation, and suggests a biological role for ISG15 modification in regulation of multiple signal transduction pathways.

Immobilization of Antibodies and Purification of ISGylated
Proteins-Hybridoma (clone 2.1) (24) that produces monoclonal mouse anti-human ISG15 antibodies was cultured in Hybridoma-serum-free medium (Invitrogen, Carlsbad, CA). Cells were removed by centrifugation at 2000 ϫ g for 5 min, and the supernatants were filtered through a 0.22-m filter to remove debris. IgGs were purified on a protein G column (Amersham Biosciences) according to manufacturer's instructions. Eluted IgGs were dialyzed against coupling buffer (0.2 M Na 2 HPO 4 , 0.2 M NaCl, pH 8.5). Six mg of purified IgGs were mixed with 1 ml of glyoxal-activated agarose (Active Motif, Carlsbad, CA) and NaBH 3 CN (Sigma) to a final concentration of 50 mM. Coupling proceeded overnight at 24°C with constant rocking. Coupled resin was removed, and the supernatant was allowed to couple with a fresh batch of glyoxal-activated agarose. Success of immobilization was determined by measuring protein concentration in the buffer before and after coupling and was estimated to be 2.6 and 1.6 mg of IgG per ml of resin after first and second coupling, respectively. Unreacted sites on coupled resin were blocked with 10 mM ethanolamine, pH 8.0 (2 h at 24°C). IgG-resin was washed sequentially with 10 volumes of each of the following: PBS, 1% Triton X-100 in PBS, PBS, 2 M NaCl in PBS, and PBS and stored in PBS containing 0.01% thimerosal (Sigma) at 4°C. Immediately before use in immunoaffinity purification of ISG15 conjugates the IgG-resin was washed with 10 volumes of 0.1 M glycine, pH 2.5.
Thymus samples were obtained (with ethical approval) from children (aged 1-10 years) during routine cardiac surgery. Nine grams of tissue were macerated with a razor blade and homogenized using a tissue homogenizer in 25 ml of RIPA buffer (50 mM Tris, pH 7.5, 5 mM EDTA, 150 mM NaCl, 0.1% w/v SDS, 1% v/v Triton X-100, 0.5% w/v sodium deoxycholate, 2 mM phenylmethylsulfonyl fluoride, and protease inhibitor mixture (Sigma, product number P2714)). The slurry was sonicated (four 30-s pulses), and insoluble material was removed by 10-min centrifugation at 18,000 ϫ g. This extraction procedure yielded 500 mg of total protein. Supernatant was stored at Ϫ80°C and passed through a 0.4-m filter immediately before use in immunoaffinity purification. In a series of pilot experiments we established that certain ISG15 conjugates would bind to IgG-resin inefficiently at pH 8.0 but bound efficiently at pH 7.0. The addition of ethylene glycol improved conditions for binding of an overlapping but distinct set of ISG15 conjugates. Combination of pH 7.0 with ethylene glycol, on the other hand, resulted in massive binding of certain nonspecific proteins. We therefore adopted the following two-step protocol. IgG-resin (0.5 ml) was mixed with thymic protein extract (250 mg of total protein, at a final concentration of 3 mg/ml) in RIPA buffer, pH 7.0, final volume 83 ml. Binding of ISG15 conjugates was carried out overnight at 4°C. The resin was then washed three times with 20 volumes of RIPA buffer, pH 7.0, and stored at Ϫ20°C. The protein extract was adjusted to pH 8.0, and ethylene glycol was added to a final concentration of 25% v/v. Fresh IgG-resin was added, and binding was carried out as above. The resin was washed three times with 20 volumes of RIPA buffer, pH 8.0, and 25% ethylene glycol and stored at Ϫ20°C. Bound proteins resulting from both steps were eluted from IgG-resins by the addition of 2 volumes of SDS-PAGE loading buffer and a 5-min incubation at 90°C. The ISG15 conjugates from both binding steps (pH 8.0 and pH 7.0, 25% ethylene glycol) were pooled and analyzed by highthroughput Western blot.
High-throughput Western Blot Screening and Data Analysis-Primary screening (pay per service) was performed at BD Biosciences Transduction Laboratories (Lexington, KY) using the PowerBlot TM assay. Briefly, the purified ISG15 conjugates were separated on six highresolution gradient gels and transferred onto nitrocellulose membranes. Each membrane was divided into 40 lanes by applying a chamber-forming grid. To each chamber a mixture of mAbs was added (1-8 mAbs per mixture; identities of mAbs and location with respect to lanes and templates are available from BD Biosciences Transduction Laboratories). After 1 h of incubation at room temperature the chambers were rinsed and incubated with secondary antibodies under the same conditions (Alexa 680-labeled goat anti-mouse IgGs; Molecular Probes, Eugene, OR). The images were acquired with an infrared scanner (Li-Cor, Lincoln, NE). The bands were identified, and molecular masses were assessed using specialized software by the service. Each observed band, where possible, was manually matched against the expected molecular mass of a protein recognized by an individual antibody in the mixture. The computer-generated data were scrutinized in our laboratory by careful manual analysis and were found to be more than 90% accurate.
Immunoprecipitations and Western Blots-To verify the results of the primary screen we performed immunoprecipitations (IPs) with additional controls as well as reciprocal IPs. Anti-Stat1 (rabbit polyclonal, antipeptide) antibodies were from Santa Cruz Biotechnology (Santa Cruz, CA). Anti-ERK1, anti-PLC␥1, and anti-Jak1 mAbs were purchased from BD Biosciences Transduction Laboratories. Rabbit antihuman ISG15 polyclonal antibodies were previously described (24). Rabbit anti-mouse ISG15 polyclonal antibodies were generated as follows. Murine ISG17 (pro-ISG15) was PCR-amplified (primers TGGAA-TTCTTAGGCACACTGGTCCCC and AATTCGATTCTGGATCCGCCT-GGGACC) from a cDNA library (a kind gift of Drs. M. Robek and S. Uprichard, The Scripps Research Institute) prepared from hepatitis B virus -infected hepatocytes and cloned into pGEM-T Easy vector (Promega, Madison, WI). The correct sequence was verified by sequencing, the insert excised with BamHI and SalI restriction endonucleases and recloned into respective sites of pQE-30 vector (Qiagen, Valencia, CA) to yield His 6 -tagged ISG17. His 6 -ISG17 was then expressed in Escherichia coli and purified on nickel-nitrilotriacetic acid resin (Qiagen) as recommended by the manufacturer. Rabbit sera were generated by Covance (Denver, PA). Anti-ISG15-specific IgGs were immunoaffinity-purified on immobilized His 6 -ISG15 and depleted against immobilized Ub (Sigma) to remove any cross-reacting IgGs. His 6 -ISG15 and Ub were immobilized on glyoxal-activated agarose as described above for mAbs.
With the exception of anti-ERK1 (lysate was prepared in denaturing conditions as recommended by the manufacturer of mAbs), all lysates were prepared and IPs were performed in RIPA buffer. Protein A-Sepharose was used with polyclonal antibodies, and protein G-Sepharose was used with mAbs. Western blotting was performed as previously described (12).
Cell Culture-Wild type and UBP43 Ϫ/Ϫ MEFs (murine embryonic fibroblasts) were generated from a litter of embryos on embryonic day 12.5. Briefly, embryonic torsos were minced and trypsinized for 30 min at 37°C. Cells were harvested, resuspended in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum, and plated on 10-cm dishes. MEFs (9 ϫ 10 5 ) were plated on 60-mm plates and replated every 3 days for over 20 passages until they were immortalized. Where indicated MEFs were treated with 200 units/ml of IFN␤, 5 M lactacystin, and 10 M MG132 (Calbiochem, La Jolla, CA). Thymocytes and bone marrow cells were isolated from 4 -6 week-old mice and were cultured in RPMI 1640 supplemented with 10% fetal bovine serum and 100 units/ml penicillin and 100 g/ml streptomycin (Invitrogen). Bone marrow cultures were also supplemented with the following growth factors: 10 ng/ml interleukin-3, 10 ng/ml interleukin-6, and 100 ng/ml stem cell factor (PeproTech, Rocky Hill, NJ).

RESULTS
Primary Screening by High-throughput Western Blot-ISGylated proteins purified from human thymus by immunoaffinity chromatography were submitted for analysis by BD Transduction Laboratories. The proteins in this assay separated on SDS-PAGE and transferred onto nitrocellulose membrane, and the membrane is divided into multiple lanes. Each individual lane is incubated with a blend of several specifically selected monoclonal antibodies that recognize proteins of nonoverlapping molecular weights (see "Materials and Methods" for further details). The assay uses 860 individual mAbs of which 710 cross-react with human proteins. Some of the mAbs were different clones recognizing the same protein or phosphorylated versions of the same protein. Therefore, 645 individual proteins can be detected by this technology in various human tissues and cells. Computer-assisted and subsequent manual examination of detected signals revealed 149 bands. Based on the confidence with which the identity of the protein could be deduced we separated the bands into three groups. Group one (72 bands) included proteins that (a) matched very well the expected molecular mass of unmodified proteins and (b) were unlikely to be the result of a smaller protein that has been modified by ISG15. Proteins of this group most likely were nonspecifically bound to IgG-resin (either agarose itself or IgGs) but also might have been co-purified with ISGylated protein(s). Group one was treated as irrelevant and is not discussed in this study. Group two (73 bands) incorporated the bands of ambiguous identity. Apparent molecular masses of group two bands corresponded to an expected molecular mass of an unmodified protein but also may correspond to a molecular mass of another protein modified by one or two molecules of ISG15. The proteins that belong to this group are currently being validated in our laboratory.
Molecular masses of group 3 proteins (three bands or groups of bands) did not correspond to any protein and were considered as the candidates likely to be ISG15-modified. Fig. 1 shows three fragments of respective Western blot panels. PLC␥1 was represented by four bands migrating in the range of 145-185 kDa (fragment A), whereas ERK1 and Jak1 were represented by single bands of 57 and 145 kDa, respectively. The molecular masses of these bands (with the exception of the 145-kDa band recognized by PLC␥1, which may be a product of proteolysis; see Figs. 2, 3 and "Discussion") were in good correlation with expected shifts in mobility upon addition of one or more molecules of ISG15.
Verification with Individual Antibodies-To validate the results of high-throughput Western blot analysis we performed IPs with anti-ISG15 mAb and analyzed the immunocomplexes by Western blot with anti-PLC␥1, anti-Jak1, and anti-ERK1 mAbs. Bands identical to those observed on high throughput immunoblotting images were detected only when specific anti-ISG15 mAb but not when an isotype control IgG were used for IP ( Fig. 2A). We consequently performed reciprocal IPs with individual mAbs against PLC␥1, Jak1, and ERK1 (Fig. 2B). Immunocomplexes were analyzed by Western blot with polyclonal rabbit anti-human ISG15 antibodies, and again, the signals corresponding to ISGylated proteins were detected. These results strongly suggest that PLC␥1, Jak1, and ERK1 are ISGylated proteins. The anti-ERK1 mAbs (BD Biosciences Transduction Laboratories, catalog number 610030) recognize both the 44-and 42-kDa bands that may represent ERK1 and ERK2, respectively. It is therefore not excluded that ISG15modified protein identified in our work may be ERK2.
Analysis of ISG15 Conjugates in UBP43 Knock-out Mice-Our laboratory has developed a knock-out mouse model in which a gene coding for the ISG15-specific protease UBP43 is deleted (19). Most tissues of UBP43 Ϫ/Ϫ mice have a massive increase of ISG15 conjugates, relative to wild type. This difference can be further increased following dosing with LPS or poly(I-C). We used MEFs derived from UBP43 Ϫ/Ϫ and UBP43 ϩ/ϩ mice to assess the relevance of our findings in human tissue to the murine model. MEFs of both genotypes were incubated with IFN␤ for 24 h. ISGylated proteins were immunoprecipitated with purified rabbit anti-mouse ISG15 antibodies, and immunocomplexes were analyzed by Western blot. The results presented in Fig. 3 demonstrate that murine PLC␥1 and ERK1 are also modified by ISG15. The modification, however, was only obvious after IFN␤ treatment and, as expected, was stronger in the UBP43 Ϫ/Ϫ cells. Interestingly, the band of 145 kDa that was detectable with anti-PLC␥1 in immunoprecipitates from human thymus was also present in MEFs and was likely to be a product of specific proteolysis of ISGylated PLC␥1. We were not able to detect ISGylation of Jak1 in MEFs; however, the modification was detectable in murine thymocytes (not shown) suggesting that the set of ISGylated proteins may be cell-specific. In a separate paper we report that Stat1, a transcription factor and substrate of Jak1, remains activated for a longer period of time in UBP43 Ϫ/Ϫ cells (20). In many Western blot experiments, where UBP43 Ϫ/Ϫ animals or isolated cells were challenged with Jak-Stat activators we observed the appearance of higher molecular mass bands recognized by Stat1 antibodies (Fig. 4A). We hypothesized that these high molecular mass bands were Stat1 molecules conjugated by ISG15. Reciprocal immunoprecipitations and Western blotting with rabbit polyclonal antibodies against Stat1 and ISG15 revealed the presence of up to five specific bands that are likely to be Stat1-ISG15 conjugates (Fig. 4B). The two bands (Stat1␣/␤, 91 and 84 kDa) corresponding to unmodified Stat1 are non-specifically recognized by ISG15 antibodies due to the overload of Stat1 from immunoprecipitation using Stat1 antibodies. These two bands also appeared on the membrane with Ponceau S staining (Fig. 4B, left panels). Noticeably, the same bands corresponding to unmodified Stat1␣/␤ are detectable with anti-Stat1 in the proteins immunoprecipitated with anti-ISG15 (Fig. 4B, right panels). Stat1 is known to form homodimers (25), and the copurification of unmodified molecules may be caused by protein-protein interaction between ISGylated and unmodified Stat1. In addition, other proteins that interact with Stat1 to form ISGF3 and other complexes may be ISGylated and cause the copurification of unmodified Stat1. To answer the question whether Stat1 is ISGylated in human thymus we performed Western blot with anti-Stat1 on purified ISG15 conjugates, the same preparation that was used for high throughput immunoblotting (Fig. 4C). A banding pattern similar to that observed in murine T-cells (Fig. 4B) was detected.
Proteasome Inhibitors Do Not Increase Amount of ISG15 Conjugates-Attachment of monoubiquitin to its target protein serves to modulate activity or location (26). However, the major function of ubiquitin is the formation of polyubiquitin chains and localization of targeted proteins to proteasomes for subsequent degradation (27,28). Loeb and Haas (7) point out that pulse studies by Knight and Cordova (29) indicated rapid turnover of both free and conjugated ISG15. It is not known whether conjugated ISG15 may act as Ub and direct the conjugated proteins for proteasomal degradation. Although reports point to the evidence (or absence of such) against it, no experimental data have been presented (7,23). UBP43 Ϫ/Ϫ cells, due to the absence of ISG15 deconjugation, appear to be an excellent model to analyze the relation between Ub, ISG15, and the proteasome. We treated UBP43 ϩ/ϩ and UBP43 Ϫ/Ϫ cells, uninduced or IFN␤-induced, with a highly specific pro- Thymi were excised from mice injected with poly(I-C) 1 or 24 h prior to sacrifice. B, reciprocal immunoprecipitations from IFN␤-treated UBP43 ϩ/ϩ and UBP43 Ϫ/Ϫ thymocytes were performed using rabbit anti-Stat1 or anti-ISG15 antibodies and then probed with anti-ISG15 or anti-Stat1, respectively. Proteins were precipitated with nonspecific rabbit IgGs (N) or with specific antibodies (S). Ten micrograms of whole cell lysate were loaded (L) to locate positions of unmodified proteins. The Ponceau S staining shows the large amount of Stat1 in the immunoprecipitates (marked by asterisks). C, ISG15 conjugates immunoaffinity-purified on antihuman ISG15 mAb resin from human thymus (same as in Fig. 1) and probed with polyclonal anti-ISG15 or anti-Stat1. teasome inhibitor lactacystin. Western blot revealed no difference in the amount of ISG15 conjugates between lactacystintreated and untreated samples, while as expected, an increase in the amount of ubiquitinated proteins was observed (Fig. 5A). Noticeably, the level of ISGylation had no detectable effect on the appearance of Ub conjugates in either UBP43 Ϫ/Ϫ or UBP43 ϩ/ϩ cells treated with IFN␤. To confirm this observation we immunoprecipitated PLC␥1 from MEFs treated with a proteasome inhibitor and probed immunocomplexes with anti-ISG15 antibodies (Fig. 5B). Consistent with the data of total protein analysis we did not observe any increase in the amount of ISGylated PLC␥1 upon inhibition of proteasomes. These results demonstrate that ISGylated proteins are not degraded by proteasomes and suggest that ISGylation does not prevent conjugation of polyubiquitin. DISCUSSION A protein modified by ISG15 acquires two new characteristics: gains affinity to anti-ISG15 antibodies and migrates more slowly on SDS-PAGE. We used these features to our advantage and identified candidate proteins modified by ISG15. High throughput immunoblotting service was originally developed for comparative analysis of two or more samples (e.g. protein extracts from normal and diseased tissues) with the intent to identify proteins whose levels are altered. Our work proves that this method is a useful tool for identification of posttranslationally modified proteins in samples enriched for the modified species. Nevertheless, we have not utilized this technology to full potential, and there are several ways to improve the experimental scheme. First, although postsurgical thymus appeared to be a good source of ISGylated proteins, the amount of ISG15 conjugates in this tissue was approximately three times lower than in IFN-treated A549 or U937 cells. 2 Second, a significant amount of free ISG15 that is present in cells and tissues strongly binds to immobilized anti-ISG15 antibodies, occupies valuable sites, and decreases efficiency of purification of ISG15 conjugates. Free ISG15, however, can be efficiently removed by gel filtration chromatography on Sephadex G-50. 2 Third, false-positives detected in the primary screening step, which resulted from nonspecific binding to IgG-resin, can be eliminated if simultaneous immunoaffinity purification is performed on immobilized anti-ISG15 and isotype control IgG. Proteins absorbed to both resins can be then analyzed on parallel blots, and nonspecific bands can be identified. Fourth, the sensitivity of the method could be increased more than 10-fold since in our conditions only 0.5 g of total protein was loaded per lane, while, the technology permits loading of up to 7.5 g of total protein per lane.
All four proteins reported in this study are known to be major players in signal transduction and are the subjects of intensive investigation by numerous laboratories. The PLC isozymes hydrolyze phosphatidylinositol bisphosphate thus modulating both calcium and protein kinase C-regulated pathways (30). PLC␥1 is essential for growth factor-induced cell motility, mitogenesis (31), normal growth and development, carcinogenesis, and cellular transformation (32).
The Jak family of receptor-associated protein kinases is directly involved in response to IFN and other cytokines (33). Jak1 is rapidly phosphorylated in response to IFN and is required for the phosphorylation of the transcription factor Stat1 (25). We demonstrate that Stat1 is also modified by ISG15 in poly(I-C) or IFN␤-treated murine cells and found that Stat1 is also modified in human thymus (Fig. 4C). High-throughput immunoblotting failed to detect this, possibly because of lower sensitivity of the antibodies used by the service. Jak1-deficient mice exhibit perinatal lethality (34), whereas Stat1 knock-out mice, although viable, lack all the physiological functions associated with the IFNs and display a remarkable sensitivity to viral infections and other pathological agents (35,36).
The family of kinases known as ERKs or MAPKs are activated after cell stimulation by a variety of hormones and growth factors. Numerous proteins represent the downstream effectors for the active ERK and implicate it in the control of cell proliferation, differentiation, as well as carcinogenesis and inflammation (37, 38). h. An equal volume of Me 2 SO (vehicle) was added to control samples. Cell lysates (10 g of total protein) were resolved on 8 -18% minigel, and Western blot was performed with rabbit anti-mouse ISG15. Membrane was stripped and reprobed with anti-Ub serum. B, UBP43 ϩ/ϩ (not treated with IFN) and UBP43 Ϫ/Ϫ (incubated with IFN␤ for 18 h) MEFs were treated with MG132 at a final concentration of 10 M for 3 h. An equal volume of Me 2 SO (vehicle) was added to control samples. PLC␥1 was immunoprecipitated, and immunocomplexes were resolved on 7% minigel. The membrane was probed with anti-ISG15 and, after stripping, was reprobed with anti-PLC␥1. Positions of unmodified PLC␥1 are shown with arrows and positions of conjugates are shown with asterisks. Cell lysates (10 g of total protein) were also resolved on an identical minigel to assess amount of ISG15 and Ub conjugates. Western blot was performed with rabbit anti-mouse ISG15, and after stripping, the membrane was reprobed with anti-Ub serum.
Several proteins in our report (PLC␥1, Stat1) consistently show more than one modified band (Figs. 1-4 (3). Therefore, the possibility that short polyISG15 chains (two to five ISG15 molecules) are formed remains open. Alternatively, modification of more than one lysine on the same protein may take place. Formation of polyUb on substrate proteins, however, does not stop after addition of a few Ub molecules but results in formation of long chains. This does not seem to be the case with ISGylated proteins, suggesting that conjugation of single ISG15 molecules to several lysines on the same protein is more likely. Although, the biochemical function of ISG15modification remains unknown the results of our experiments with proteasome inhibitors indicate that ISG15 conjugates are not degraded by proteasomes and ISG15 does not protect targeted proteins from ubiquitination and subsequent degradation. The proteins thus far known to be modified by ISG15 (Serpin 2a, ERK1, PLC␥1, Jak1, and Stat1) have diverse biochemical functions. It is reasonable to assume that the major role of ISG15 conjugation is IFN or other signalinduced modulation of such characteristics of the protein as solubility, stability, localization, etc. In this respect the role of ISG15 may be similar to polyubiquitin, which conjugates to a vast number of diverse proteins, yet the consequences of binding are the same: acquired affinity to a proteasomal subunit and degradation. Loeb and Haas (7) noticed that the data of Knight and Cordova (29) indicated rapid turnover of both free and conjugated ISG15. In most of our experiments we were able to observe fragments of PLC␥1, Jak1, and ERK1 (Figs. 1 and 2B) as well as Stat1 (data not shown). It is tempting to speculate that ISG15 conjugation promotes degradation of targeted proteins via a pathway alternative to proteasomal. However, it is also possible that protein ISGylation plays a specific role in regulating enzymatic or DNA binding activity of target proteins similar to other ubiquitin-like modifiers.
Coordinated induction of ISG15, UBP43, and UBE1L suggests that ISG15 conjugation is a dynamic and highly controlled process. ISG15-activating enzyme UBE1L was found to be absent in 14 different lung cancer cell lines suggesting that decrease of ISG15 conjugation may contribute to carcinogenesis (39). Certain viruses can specifically block conjugation or synthesis of ISG15 (11) possibly in an attempt to suppress host cell suicide and inflammatory response. Dysregulation of ISGylation due to deletion of UBP43 leads to decreased life expectancy, brain cell injury, hypersensitivity to interferon, and apoptosis in hematopoietic tissues (19,20). These reports combined with the fact that IFN suppresses cellular proliferation indicate that the balance of ISG15 conjugation is important for the control of cell differentiation and growth suppression. Our current report indicates that ISGylation may be directly involved in the regulation of several signal transduction pathways in organisms challenged with IFN elicitors. It is possible that upon IFN stimulation the cell fine tunes the degree of response via ISGylation of Jak1 and Stat1, crucial players in the IFN pathway. Moreover, other signaling pathways may also be directly affected by means of ISGylation of respective critical regulators (e.g. PLC␥1, ERK1) to orchestrate overall growth, differentiation, and survival.