The ISG15 isopeptidase UBP43 is regulated by proteolysis via the SCFSkp2 ubiquitin ligase.

The Skp2 oncoprotein belongs to the family of F-box proteins that function as substrate recognition factors for SCF (Skp1, cullin, F-box protein) E3 ubiquitin-ligase complexes. Binding of the substrate to the SCFSkp2 complex catalyzes the conjugation of ubiquitin molecules to the bound substrate, resulting in multi-ubiquitination and rapid degradation by the 26 S proteasome. Using Skp2 as bait in a yeast two-hybrid screen, we have identified UBP43 as a novel substrate for Skp2. UBP43 belongs to the family of ubiquitin isopeptidases and specifically cleaves ISG15, a ubiquitin-like molecule that is induced by cellular stresses, such as type 1 interferons (IFN), nephrotoxic damage, and bacterial infection. UBP43 was originally identified as an up-regulated gene in knock-in mice expressing an acute myelogenous leukemia fusion protein, AML1-ETO, as well as in melanoma cell lines treated with IFN-beta. The phenotype of UBP43 knockout mice includes shortened life span, hypersensitivity to IFN, and neuronal damage, suggesting that tight regulation of ISG15 conjugation is critical for normal cellular function. In this study, we demonstrate that UBP43 is ubiquitinated in vivo and accumulates in cells treated with proteasome inhibitors. We also show that Skp2 promotes UBP43 ubiquitination and degradation, resulting in higher levels of ISG15 conjugates. In Skp2-/- mouse cells, levels of UBP43 are consistently up-regulated, whereas levels of ISG15 conjugates are reduced. Our results demonstrate that the SCFSkp2 is involved in controlling UBP43 protein levels and may therefore play an important role in modulating type 1 IFN signaling.

Modification of proteins by ubiquitin and ubiquitin-like (Ubl) 1 molecules, including SUMO, Nedd8, and ISG15, has emerged as a critical regulatory process in eukaryotes, controlling pathways such as the cell cycle, cellular stress response, intracellular signaling, development, and the immune response (1)(2)(3)(4)(5). Deregulation of ubiquitin or Ubl modification can cause autoimmune and neurodegenerative diseases, developmental abnormalities, and cancer.
Conjugation of ubiquitin and Ubls involves a three-step mechanism initially demonstrated for ubiquitin as follows. A single E1 ubiquitin-activating enzyme activates the Ubl molecule via formation of a thioester bond. Activated Ubl is then transferred to one of a large family of E2 ubiquitin-conjugating enzymes. In most cases, E2 enzymes are targeted to appropriate substrates by a class of substrate receptor complexes termed E3 ubiquitin ligases. Together, the E2 and E3 enzymes catalyze the formation of isopeptide bonds between ubiquitin and lysine residues on the target proteins. In the case of multiubiquitination, additional ubiquitin molecules are added to form ubiquitin chains. The multiubiquitinated proteins are then recognized and rapidly degraded into short peptides by the 26 S proteasome (3,6). Modification by ubiquitin and Ubl molecules is a reversible process mediated by a large family of isopeptidases that exist to remove these molecules from their substrates. Although the function and regulation of these isopeptidases is poorly defined, the few that have been studied indicate an important role in growth and development (7,8).
Specificity of the ubiquitin and Ubl pathways is conferred by the nature and activity of the E3 complexes. One of the beststudied E3 ligase complexes is the SCF complex. SCF complexes are composed of four subunits: the three core components Skp1, a Cullin family member, the RING finger protein Rbx1/Roc1, and one variable component, the F-box protein, that acts as the substrate recognition factor. A large number of F-box proteins have been identified in organisms ranging from yeast to humans. The mammalian F-box protein Skp2 may play a pivotal role in oncogenesis and has been implicated in degradation of several key regulators of cell proliferation including the tumor suppressor proteins p27 Kip1 , p57 Kip2 , (9) and p130 (10) and the replication factor, hORC1 (11).
We used the yeast two-hybrid technique to isolate potential Skp2-interacting proteins. We identified seven proteins, including UBP43, which will be described in this report. UBP43 belongs to the family of ubiquitin isopeptidases, one of the two classes of ubiquitin and Ubl deconjugating enzymes (12). UBP43 was identified as strongly expressed mRNA in hematopoietic tissue from an acute myeloid leukemia mouse model (13). Thereafter, UBP43 was identified as an up-regulated gene in melanoma cells treated with IFN-␤ and the protein kinase C activator MEZ. This treatment causes cells to irreversibly lose proliferation potential and begin progression toward terminal differentiation (14). UBP43 functions as an isopeptidase re-sponsible for cleaving the ubiquitin-like protein ISG15 from substrates (15). Upon treatment of cells with type I IFN or lipopolysaccharide (LPS), ISG15, UBP43, and the ISG15 E1 enzyme, UBE1L, are rapidly up-regulated (16). It seems that tight regulation of both the ISG15 conjugation and deconjugation pathways is required to ensure proper response to cellular stress. In UBP43 knockout mouse embryonic fibroblasts (MEFs), ISG15 cleavage from cellular substrates is strongly reduced, suggesting that UBP43 is the major ISG15 isopeptidase (17). UBP43Ϫ/Ϫ MEFs demonstrate prolonged STAT1 signaling and IFN hypersensitivity, which is in accordance with data showing that signal transducer and activator of transcription 1 is an ISG15-modified protein (18). These observations and the identification of UBP43 in the two tumor models suggest that UBP43 plays an important role in cellular proliferation and differentiation and that UBP43 levels need to be carefully controlled in cells. Indeed, UBP43 protein levels are regulated at the level of transcription (16) and post-translationally, as we show here, via the SCF Skp2 -mediated ubiquitin pathway.

MATERIALS AND METHODS
DNA Constructs-Skp1, Skp2, and UBP43 were PCR-amplified using a human lymphocyte cDNA library (HL4006AE; BD Biosciences Clontech). Skp1 and Skp2 were cloned into MatchMaker (BD Biosciences Clontech) yeast two-hybrid vectors pGADT7 and pGBKT, respectively. All other plasmids were created with the GATEWAY cloning system using vectors from Invitrogen following manufacturer's protocols. The C-terminally V5-tagged UBP43 adenovirus construct was made using the pADeasy system (Qbiochem). Amplified clones were verified by sequencing.
Antibodies-The Skp2 antibody used for immunoprecipitation was purchased from Zymed Laboratories Inc.. Skp2 (His-435) and Cdk2 (Glu-119) antibodies used in immunoblotting analysis were purchased from Santa Cruz. His, HA, and V5 antibodies were purchased from Amersham Biosciences, Babco, and Invitrogen, respectively. The affinity-purified polyclonal ISG15 antibody was kindly supplied by Dr. Arthur Haas and does not cross-react with ubiquitin (19).
Yeast Two-hybrid Screen-The MatchMaker (BD Biosciences Clontech) yeast two-hybrid system was used to screen for Skp2-interacting proteins. Yeast strain AH109 was transformed with plasmid pGBKT-Skp2 containing full-length Skp2 fused to the GAL4 DNA-binding domain. Transformed cells were streaked on plates lacking tryptophan to select for single clones. This strain was then transformed with a human lymphoma cDNA library (HL4006AE; Invitrogen) cloned into the vector pACT1 that fuses cDNAs to the GAL4 activating domain. 2 ϫ 10 6 clones were screened on plates lacking histidine (ϩ10 mM 3AT) or adenine and histidine (ϩ10 mM 3AT). Interaction of positive clones was verified by retransforming clones into the original Skp2 bait strain and assaying for interaction. Skp2 and UBP43 constructs were then cloned into yeast expression vectors compatible with the ProQuest two-hybrid system from Invitrogen. Constructs were transformed into AH109 and screened on plates lacking leucine, tryptophan, or histidine (ϩ30 mM 3AT).
Production of Recombinant Proteins in Bacteria and GST Purification-Escherichia coli BL21 (Invitrogen) cells were used to express the vector, pDEST15 (Invitrogen) containing GST or a GST-Skp2 fusion protein. One liter of each strain was induced with 100 mM isopropyl ␤-D-thiogalactoside for 2 h. Protein was isolated using a 3ϫ freeze-thaw protocol with a dry ice and ethanol bath. Extracts were purified over glutathione-Sepharose 4B matrix (Amersham Biosciences) according to manufacturer's instructions. Extracts from HEK 293 cells transfected with indicated plasmids were added to the matrix. Bound fractions were analyzed by Western blotting using the indicated antibodies.
Immunoprecipitation-80% confluent A549 cells on a 10-cm plate were infected with adenoviruses expressing Skp2 with an N-terminal FLAG tag, and UBP43 with a C-terminal V5 tag. Infection was confirmed with GFP. 36 h after infection, cells were harvested in mammalian lysis buffer. The lysate was subjected to a FLAG column (Sigma), and bound proteins were eluted with 1.5ϫ FLAG peptide (Sigma) and subjected to Western blot analysis.
Ubiquitination-Skp2 was subcloned into pDEST31 (Invitrogen) containing an N-terminal His 6 epitope. UBP43 was subcloned into pDEST40 (Invitrogen) containing C-terminal V5 and His 6 epitopes. HA-ubiquitin and His 6 -ubiquitin constructs were a gift from Dr. Rosalie Sears (Oregon Health and Sciences University, Portland, OR). A549 cells, transfected with indicated plasmids, were treated for 5 h with 10 M MG132 (Calbiochem). 35 S Pulse-chase-REF52 and A549 cells were infected with UBP43-V5 adenovirus for 20 h. Skp2 wild-type (ϩ/ϩ) and Skp2 knockout (Ϫ/Ϫ) primary MEFs were infected in the presence of Effectene reagent. Cells were then labeled for 30 min with 7 mCi of [ 35 S]methionine, chased with media containing 5 mM methionine and 3 mM cysteine, washed, and collected at 0, 15, 30, 60, 120, and 180 min. Cell extracts were denatured and then renatured for V5 immunoprecipitation with anti-V5 antibody from Invitrogen. Proteins were separated by SDS-PAGE and visualized by autoradiography. Proteasome inhibitors (5 M MG132 ϩ 5 M lactacystin) were added 5 h before the pulse and maintained throughout the pulse and chase phases.
IFN-␤ Treatment-Skp2 wild-type and SkpϪ/Ϫ MEFs were treated with 1000 units/ml mouse IFN-␤, and extracts were harvested 24 h later. For cell growth measurements, 3.6 ϫ 10 4 immortalized MEF cells were plated in six-well plates, treated with 500 units/ml of mouse IFN-␤, and harvested in duplicate. For each time point, total viable cell number was assessed by counting with a hemacytometer. Trypan blue staining was used to identify dead cells. A minimum of 150 cells was counted per sample.

UBP43
Interacts with Skp2-To identify Skp2-interacting proteins, we performed a yeast two-hybrid screen and obtained eight clones coding for putative Skp2 interacting proteins (see "Materials and Methods"). Sequencing of the cDNA inserts revealed two previously identified and five novel Skp2 interactors. In validation of our screen, we isolated Skp1 (two clones) and Cks1. Skp1 is the scaffold protein that anchors Skp2 to the SCF complex via the Skp2 F-box domain (18). Cks1 is a protein recently identified to interact with Skp2 and facilitate substrate recognition of p27 (20,21). Of the other five clones isolated, we focused our attention on a clone coding for a Cterminal portion of UBP43. Both the UBP43 C terminus (amino acids 121-373) that was isolated in the original screen and full-length UBP43, but not the N terminus (amino acids 1-121), were able to interact with Skp2, as assayed by production of ␤-galactosidase and growth on plates lacking histidine (Fig. 1A). To further isolate the Skp2-interacting region of UBP43, we made successive C-terminal truncations of the 121-373 fragment (Fig. 1B). We located a region between amino acids 183 and 352 of UBP43 that seems to be involved in its interaction with Skp2. The deletion construct containing the entire C-terminal region (121-373) as well as a construct lacking the last 21 amino acids, fragment 121-352, demonstrated robust interaction. However, a construct containing amino acids 121-285 showed markedly reduced interaction. When fragment 121-183 was expressed, interaction with Skp2 was completely abrogated. We therefore conclude that the region between amino acids 121 and 285 of UBP43 is required for the interaction between UBP43 and Skp2.
Skp2 Interaction with UBP43 Requires the Skp2 Leucine-rich Repeat (LRR) Domain-Next, we analyzed the interaction of UBP43 and Skp2 in vivo. Fig. 1C demonstrates that full-length Skp2 can co-immunoprecipitate UBP43 ectopically expressed in the human lung cancer cell line A549. A construct lacking the LRR and C terminus showed no binding to UBP43. However, constructs lacking the N terminus or both the N terminus and F-box regions but retaining the LRR and C terminus did associate with UBP43. The LRR region of Skp2 has been implicated in substrate binding. Our data thus suggest that UBP43 is a substrate of SCF Skp2 .
Skp2 and UBP43 Interact in Vitro-To verify the interaction of UBP43 with Skp2, we performed in vitro binding experiments. Escherichia coli that expressed GST-Skp2 or GST alone were bound to glutathione beads, followed by incubation with extract from human embryonic kidney (HEK) 293 cells that expressed ectopic UBP43. Fig. 1D shows that GST-Skp2, but not GST, binds to UBP43 in vitro.
Levels of UBP43 Are Modulated by Skp2 and by Proteasome Inhibitors-Our finding that UBP43 interacts with the LRR domain of Skp2 suggested that UBP43 is a substrate of SCF Skp2 . Therefore, UBP43 protein levels could be controlled by protein degradation. To address this question, we expressed Skp2 and UBP43 in HEK293 cells. UBP43 levels were strongly reduced when Skp2 was expressed (Fig. 2A, lane 2) compared with when UBP43 was expressed alone (lane 1). Because Skp2 functions in an E3 ligase complex that targets substrates for degradation by the 26 S proteasome, we next assayed whether UBP43 accumulates in cells that have been treated with the proteasome inhibitor MG132. Indeed, under these conditions,  Hisϩ3-AT). Right, ␤-galactosidase activity. B, yeast strains expressing the indicated constructs were grown on ϪHϩ3-AT plates to assess protein interaction as above. C, mammalian co-immunoprecipitation. Human lung carcinoma A549 cells were infected with adenovirus expressing UBP43-V5 and various FLAG-Skp2 constructs. Cell extracts were immunoprecipitated with anti-FLAG beads, separated by SDS-PAGE, and analyzed by immunoblotting with the indicated antibodies. The asterisk indicates a background band. Endogenous Skp2 is denoted by a white arrowhead, ectopic Skp2 by black arrowheads. The proteasome inhibitor lactacystin was added to a final concentration of 10 M 5 h before harvest. IB, immunoblot; IP, immunoprecipitation. D, GST-Skp2 binds UBP43 in vitro. GST-Skp2 (lanes 1 and 2) or GST (lanes 3 and 4) were expressed in E. coli and purified over glutathione beads. Beads were incubated with HEK293 cell extracts expressing UBP43 (lanes 1 and 3) or empty vector (lanes 2 and 4).
robust UBP43 accumulation can be observed ( Fig. 2A, lane 4). Together, these results suggest that UBP43 is degraded in the proteasome most likely by SCF Skp2 -mediated ubiquitination.
Levels of UBP43 and ISG15 Are Altered in Skp2Ϫ/Ϫ Cells-To test whether levels of UBP43 are increased in cells that are devoid of Skp2, we performed immunoblotting of primary, low passage MEF extracts derived from wild-type or Skp2Ϫ/Ϫ mice. As shown in Fig. 2B, UBP43 levels are significantly increased in Skp2Ϫ/Ϫ extracts (lane 2) compared with wild type (lane 1). Note that UBP43 levels are very low in unstimulated wild-type cells and are hardly detectable with the available antibody. We also observed a significant increase in steady-state UBP43 levels in Skp2Ϫ/Ϫ cells stimulated with LPS (data not shown). In addition to the increase in UBP43 levels in Skp2Ϫ/Ϫ cells, an increase in free ISG15 levels was observed (lane 2, arrowhead). This suggests that in the absence of Skp2, UBP43 levels are up-regulated, resulting in increased cleavage of ISG15 from substrates (see Fig. 5B) and a concomitant increase in free ISG15.
Skp2 Enhances UBP43 Ubiquitination in Vivo-Proteasomal degradation of proteins is triggered by multi-ubiquitination of targeted polypeptides. To determine whether UBP43 is ubiquitinated in vivo, we transfected HEK293 cells with plasmids encoding Skp2, V5-tagged UBP43, and HA-tagged ubiquitin. UBP43, in the presence of overexpressed Skp2, appeared in higher molecular mass forms consistent with ubiquitination (Fig. 3A). These high molecular mass bands were intensified upon addition of the proteasome inhibitor lactacystin. Next, we performed immunoprecipitation with anti-V5 antibodies to capture UBP43 protein. High molecular mass bands were observed that were immunoreactive against anti-HA antibodies, indicating that these bands represented ubiquitinated UBP43 (Fig. 3B, lanes 2 and 3). Ubiquitination was enhanced by ectopic Skp2 (lane 3) and was absent if either V5-UBP43 or HA-ubiquitin was omitted (lanes 1 and 4). If these high molecular mass bands correspond to ubiquitinated UBP43 species, then inhibition of the proteasome should result in an increase and possibly a shift to even higher molecular mass bands. This is indeed the case, as shown in Fig. 3C, lane 3. In this experiment, His-V5-tagged UBP43 was captured on nickel-nitrilotriacetic acid beads under denaturing conditions, followed by immunoblotting against the HA tag on ubiquitin. Because in these experiments, the amount of transfected UBP43 was high compared with Skp2, we did not observe a significant reduction in UBP43 steady-state levels upon Skp2 co-transfection. We conclude that UBP43 is ubiquitinated in vivo. To test whether Skp2 can increase the amount of ubiquitinated UBP43, we infected normal rat embryo fibroblast (REF52) cells with V5tagged UBP43 expressed from an adenovirus together with increasing levels of adeno-Skp2 virus. Fig. 3D shows that under proteasome inhibition conditions (MG132, lanes 3, 5, and 7), UBP43 ubiquitination is enhanced by increasing levels of Skp2. We kept the level of Skp2 at or near endogenous levels (data not shown), whereas UBP43 levels were higher to allow detection of ubiquitinated species. In conclusion, these experiments demonstrate that UBP43 is ubiquitinated in vivo by Skp2 and is then degraded in a proteasome-dependent manner.
Skp2 Promotes Degradation of UBP43 in Vivo-Our results suggested that UBP43 levels were controlled by SCF Skp2 -mediated degradation via the ubiquitin-proteasome pathway. To measure UBP43 degradation rates in vivo, we performed pulsechase experiments in adenovirus-UBP43 infected rat fibroblast (REF52) cells. UBP43 has a half-life of ϳ60 min under unstressed conditions (Fig. 4A, lanes 2-5). Again, UBP43 is very hard to detect under these conditions; however, the GFP control lane (lane 1) clearly demonstrates the specificity of the UBP43 band. When cells were treated with a proteasome inhibitor (Fig.  4A, lanes 6 -9), UBP43 was stabilized significantly. We obtained similar results in human A549 cells (data not shown). In addition, cycloheximide treatment of cultures followed by analysis of UBP43 protein levels confirmed the pulse-chase results (data not shown). Ectopic Skp2 expression resulted in accelerated degradation of UBP43 (data not shown). We repeated the pulse-chase analysis in low passage Skp2ϩ/ϩ and SkpϪ/Ϫ primary MEFs. The half-life of UBP43 was ϳ50 min in Skp2 wild-type cells compared with 120 min in knockout cells (Fig. 4, B and C). When Skp2 was re-expressed in the Skp2Ϫ/Ϫ cells, levels of UBP43 dropped dramatically at the 15-min time point and then remained at that low level for the rest of the chase period. We conclude that Skp2 can initiate rapid degradation of UBP43 via the ubiquitin-proteasome pathway.
Skp2 Activity Modulates the ISG15 Conjugation Pathway-ISG15 conjugation to substrates is induced upon IFN-␤ stimulation. To determine whether Skp2 has an effect on ISG15 conjugation by way of its regulation of UBP43, we overexpressed Skp2 in the A549 human lung carcinoma cell line. After treatment with LPS, cells overexpressing Skp2 showed a marked increase in ISG15 conjugation compared with cells transfected with empty vector (Fig. 5A). On the other hand, Skp2Ϫ/Ϫ MEFs induced with IFN-␤ displayed a reduction in ISG15 conjugation (Fig. 5B) and an increase in free ISG15 (Fig.  2B) compared with wild-type MEFs. We conclude that Skp2 can modulate the level of ISG15 conjugates, most likely via degradation of UBP43.
Cells from UBP43Ϫ/Ϫ mice exhibit increased levels of ISG15 conjugates; these animals are hypersensitive to induction of the type I IFN pathway (17). Therefore, we would expect that Skp2Ϫ/Ϫ cells, which have reduced levels of ISG15 conjugates, might be more resistant to the growth-inhibiting effects of IFN-␤ treatment. To test this, we analyzed the sensitivity of Skp2Ϫ/Ϫ cells toward IFN-␤. Skp2 wild-type cells showed a marked reduction in cell growth 24 h after IFN-␤ reduction (Fig. 5C). In contrast, the growth rate of Skp2Ϫ/Ϫ cells was unaffected by IFN-␤ induction at 24 h. These data suggest that absence of Skp2 decreases or delays the response to IFN-␤, most likely because of the higher levels of UBP43 observed (Fig. 2B). DISCUSSION The biological function of ISG15 modification is not well understood. It is clear, however, that carefully controlled  3. Skp2 increases UBP43 ubiquitination in vivo. A, extracts from HEK293 cells were transfected with UBP43-V5 (lanes 2-4), Skp2 (lanes  1, 3, 4), and HA-ubiquitin (lanes 2-4), separated by SDS-PAGE, and analyzed with the indicated antibodies. Lactacystin (10 M) was added 5 h before harvest. 〉, extracts from HEK293 cells transfected with UBP43-V5 (lanes 1-3), Skp2 (lanes 3 and 4), and HA-ubiquitin (lanes 2-4) were immunoprecipitated with anti-V5 antibody, separated by SDS-PAGE, and analyzed with the indicated antibodies. Asterisks mark heavy and light IgG bands. C, A549 cells were transfected with UBP43 V5⅐6HIS (lanes 2-4) and HA-ubiquitin (lanes 1, 3, and 4). Proteasome inhibitor, 5 M MG132, and/or 5 M lactacystin (lanes 1-3) was added 5 h before harvest. Extracts were purified using nickel-nitrilotriacetic acid (Ni-NTA) beads under denaturing conditions (8 M urea) to capture UBP43 V5⅐6HIS . Eluates, separated by SDS-PAGE, were analyzed by Western blotting with the indicated antibodies. D, REF52 cells were infected with constant amounts of adenovirus expressing UBP43-V5 or GFP and increasing amounts of adenovirus expressing Skp2 (Ad-Skp2) in the presence or absence of proteasome inhibitor (5 M MG132, 5 M lactacystin). Lysates were subjected to Western blot analysis. ISG15 conjugation and deconjugation to substrates is crucial for the health of a cell and of an organism. This suggests that the level and activity of enzymes that control ISG15 modification, including UBE1L and UBP43, need to be tightly regulated. Indeed, mice lacking UBP43 are short-lived, develop neuronal injury, exhibit hypersensitivity to IFN, and demonstrate increased apoptosis in hematopoietic tissues (17). On the other hand, ectopic expression of UBP43 blocks monocyte differentiation in cell culture (13). In addition, the E1 enzyme for ISG15, UBE1L, is absent in all 14 lung cancers examined for UBE1L expression, suggesting that the lack of ISG15 conjugation contributes to malignant transformation. Cellular levels of UBP43 are controlled at the level of transcription by LPS and IFN type 1 induction (14,16). We demonstrate here that the SCF Skp2 ubiquitin ligase controls the UBP43 protein level by ubiquitin-mediated degradation via the proteasomal pathway.
Our data demonstrate that Skp2 binds to UBP43 and initiates its multi-ubiquitination, resulting in UBP43 degradation via the proteasome. In MEFs lacking Skp2, levels of free ISG15 are high, and ISG15 conjugates are low, consistent with increased UBP43 levels. On the other hand, high levels of Skp2 result in an increase in ISG15-conjugated proteins. It was interesting that upon LPS treatment, UBP43 protein was stabilized, an effect that was countered by high Skp2 levels (data not shown). The coordinated induction of both ISG15 conjugating and deconjugating pathways suggests that ISG15 modification is a dynamic process that needs to be carefully controlled for normal cellular function and viability. Indeed, ectopic expression of ISG15 in various cell types initiates apoptosis. 2 In this context, Skp2-mediated degradation might play a fine-tuning role to adjust the levels of UBP43 according to the growth and stress conditions of a cell. Skp2 itself is regulated at the level of transcription and protein degradation (22,23). Skp2 protein is absent in G 0 and early G 1 cells, rises as cells enter S phase, and declines in mitosis. We have observed an inverse correlation between Skp2 and ectopic UBP43 levels in synchronized A549 cells (data not shown), consistent with the role of Skp2 in 2 S. Lanker, unpublished observations. degrading UBP43. We have not yet been able to follow endogenous UBP43 levels through the cell cycle, mainly because UBP43 levels are very low in cells not treated with IFN or LPS, and the available antibody does not detect endogenous UBP43. Our data do not exclude the possibility that more than one F-box protein participates in the degradation of UBP43; indeed, in Skp2Ϫ/Ϫ cells, UBP43 is still fairly unstable, with a half-life of about 2 h (Fig. 4C). However, UBP43 steady-state levels are greatly increased in Skp2Ϫ/Ϫ cells, and Skp2 reexpression reduces the levels back to normal, arguing that Skp2 does have a major effect on UBP43 protein levels.
The recognition of substrates by SCF complexes is often catalyzed by substrate phosphorylation at particular residues (24 -31). This is also true for SCF Skp2 (9,10,22,32,33). In addition, the adapter protein Cks1 was shown to be required for efficient targeting of p27 and p130 (20,21). We have preliminary evidence that UBP43 is phosphorylated, but whether phosphorylation is important for binding and whether Cks1 is needed for efficient interaction with Skp2 are under investigation.
ISG15 conjugation plays an important role in stress response, and is also implicated in controlling cell proliferation and differentiation. It is noteworthy that influenza virus proteins inhibit ISG15 expression and conjugation, suggesting that ISG15 mediates antiviral activity (34). It is intriguing that Skp2 controls the levels of cell cycle regulators and, as we show here, a factor in stress response. A connection between cell cycle control and stress response at the level of SCF-mediated degradation has been documented. For example, the yeast SCF Cdc4 complex controls the CDK inhibitors Sic1 and Far1 as well as the transcription factor Gcn4 involved in the response to amino acid starvation. SCF Grr1 degrades G 1 cyclins and has an important role in the cellular response to glucose starvation. It will be interesting to understand the molecular mechanism that ties Skp2 to cellular stress and the role of ISG15 conjugation in this process. Dissecting how Skp2 connects cell cycle control and cellular stress signaling will serve as a paradigm for similar pathways that integrate cell division and external signals.