A Drosophila ortholog of the human cylindromatosis tumor suppressor gene regulates triglyceride content and antibacterial defense

Familial cylindromatosis is an autosomal-dominant predisposition to tumors of skin appendages called cylindromas(Bignell et al., 2000). This familial disease and the majority (57%) of sporadic cases of cylindromatosis are causally associated with defects in a single gene, named CYLD(Biggs et al., 1996; Biggs et al., 1995). Tumors develop in individuals who have both copies of the gene inactivated,indicating that CYLD is a tumor suppressor gene. CYLD codes for a 956 amino acid cytoplasmic protein (CYLD) with deubiquitylating activity, which has been associated genetically with its tumor-suppressing function (Borodovsky et al.,2002; Kovalenko et al.,2003; Trompouki et al.,2003). CYLD is also a negative regulator of NF-κB activation by members of the tumor necrosis factor receptor superfamily(Brummelkamp et al., 2003; Kovalenko et al., 2003; Regamey et al., 2003; Trompouki et al., 2003). The negative regulation of this pathway appears to be mediated by the ability of CYLD to induce deubiquitylation of TRAF2, TRAF6 and/or NEMO [also known as IKBKG and IKKγ-Human Gene Nomenclature Database (HUGO)]. Interestingly,mice with both Cyld alleles inactivated are more susceptible to chemically induced skin cancer than wild-type mice and their susceptibility is a consequence, at least in part, of increased Bcl3-dependent NF-κB activity (Ikeda and Dikic,2006; Massoumi et al.,2006). Recent reports have indicated a broader role for CYLD in mammalian pathophysiology, consistent with its broad tissue distribution. This protein is drastically downregulated in human inflammatory bowel disease and it is required for the proper development of T lymphocytes in mice(Costello et al., 2005; Reiley et al., 2006). CYLD is significantly conserved across evolution, because putative orthologs of CYLD can be readily identified in a variety of organisms, including Drosophila melanogaster and Caenorhabditis elegans(Bignell et al., 2000).

The broad tissue distribution of CYLD, its implication in two complex and highly regulated signaling pathways, and the fact that multiple tissues are affected when the gene is mutated highlight the complexity of the biological roles of this protein. To facilitate the functional characterization of CYLD,we sought to establish and characterize a simpler, yet appropriately representative, animal model system. Drosophila was chosen for this analysis, because it is well-characterized, offers enormous advantages for genetic analysis and contains a putative ortholog of CYLD. DrosophilaCYLD (FlyBase gene ID: CG5603) is 45% homologous to human CYLD and the two proteins have a similar domain organization(Bignell et al., 2000). Based on previous findings using human cells, we sought to validate the model by investigating the involvement of Drosophila CYLD in NF-κB/IKK signaling. In this context, we analyzed the potential involvement of Drosophila CYLD in IMD-mediated innate immunity.

In Drosophila, IMD (for immune deficiency) defines a signaling cascade broadly resembling the mammalian TNFR1 (also known as TNFRSF1A-HUGO)pathway (Kaneko and Silverman,2005; Naitza and Ligoxygakis,2004). Following microbial recognition via the Drosophilapeptidoglycan recognition proteins (PGRPs) PGRP-LC and PGRP-LE, the signal is transduced from the cell surface to IMD, a central player of the pathway and a homolog of the mammalian TNF receptor interacting protein 1 (RIP1)(Georgel et al., 2001). In the TNFR1 pathway, RIP1 is essential for NF-κB and mitogen activated kinase(MAPK) activation (Kelliher et al.,1998; Stanger et al.,1995). In Drosophila, IMD transduces a signal that involves similar components. Flies deficient in IMD, dFADD (also known as BG4-FlyBase) or dTAK1 (also known as TAK1-FlyBase) are extremely sensitive to Gram-negative bacterial infections and fail to activate the relevant antimicrobial peptides (AMPs) (Leulier et al., 2002; Naitza et al.,2002; Vidal et al.,2001).

The Drosophila Iκβ kinase (IKK) complex (see below)acts downstream of dTAK1 in the activation of NF-κB. This complex contains two subunits identified by the ird5 and kenny(key) mutants. The former codes for the IKKβ homolog, whereas the product of the latter shares significant homology to the regulatory subunit of the human NEMO protein (Lu et al., 2001; Rutschmann et al.,2000b; Silverman et al.,2000). Activation of the IKK complex depends on the fly homologs of the ubiquitin-conjugating enzymes UBC13 (also known as UBE2N-HUGO) and UEV1α (also known as UBE2V1-HUGO), similarly to the activation of mammalian IKK, which depends on the mammalian proteins(Deng et al., 2000; Wang et al., 2001; Zhou et al., 2005). The Drosophila IKK complex phosphorylates the precursor form of the Relish protein, an NF-κB family member that can be activated by proteolytic cleavage. DREDD, a caspase-type protease, is involved in the full activation of Relish along with the Drosophila IKK complex(Stoven et al., 2003). Active Relish regulates directly the transcription of antimicrobial peptide genes.

In this article, we show that, like its vertebrate counterpart, Drosophila CYLD is a deubiquitylating enzyme that is broadly expressed in embryos and adults. Flies with impaired CYLD expression are sensitive to bacterial infections, a phenotype that is reversed upon restoration of CYLD expression. In addition, we demonstrate that Drosophila CYLD is biochemically and functionally associated with the IMD pathway and that it regulates fat homeostasis.

Full-length D. melanogaster CYLD was amplified by PCR from D. melanogaster complementary (c)DNA using the primers: DMCYLDF1(5′-GCTCTAGAATGAATTCCAAATCCGATTATGAAGC-3′) and DMCYLDR4(5′-CTGACCTACGCCGGCGGCTCTAATGGTACATCATTATATCTG-3′). The resulting PCR product was digested with EcoRI and NotI, and was cloned into the corresponding sites of pCDNA3FLAG in frame with the FLAG epitope to generate pCDNA3FLAGDMCYLD (Trompouki et al., 2003). pCDNA3FLAGDMCYLD(C284S) was generated by three PCR reactions. The first PCR was performed with pCDNA3FLAGDMCYLD as the template and the primers DMCYLDF3 (5′-GGCTCGATACCACCATTACAAA-3′) and DMCYLDR5 (5′-ATCAAGGTAGCTGGAATTATGG-3′). The second PCR was performed with the same template and the primers DMCYLDF4(5′-CACCATAATTCCAGCTACCTTGAT-3′) and DMCYLDR6(5′-CCGATGAGTGAAAGCTTTGC-3′). The third PCR was performed with the products of the previous two PCRs as template and the primers DMCYLDF3 and DMCYLDR6 (see above). The resulting PCR product was digested with NdeI and HindIII and cloned into the NdeI and HindIII sites of pCDNA3FLADDMCYLD to generate pCDNA3FLAGDMCYLD(C284S). pUAST.1DMCYLD was generated by digesting pCDNA3FLAGDMCYLD with KpnI and XbaI and subcloning the fragment into the corresponding sites of the plasmid pUAST.1. pUAST.1DMCYLD(C284S) was generated in the same manner as pUAST.1DMCYLD. The full-length DMCYLD fragment produced from pCDNA3FLAGDMCYLD or pCDNA3FLAGDMCYLD(C284S) by digestion with EcoRI and NotI was cloned into pGEX5X-3 (Amersham-Pharmacia Biotech) digested with EcoRI and NotI to generate glutathione-S-transferase (GST)-DmCYLD or GST-DmCYLD(C284S), respectively. GST-DmCYLD(1-317) was generated by cloning into the EcoRI and NotI sites of pGEX5X-3 a PCR fragment generated using pCDNA3FLAGDMCYLD as a template and the primers DMCYLDF1 and DMCYLDR2 (5′-GACTGGATGCGGCCGCCTGTAATTTCGAATATCTTGTGGGC-3′)digested with EcoRI and NotI. To generate 6×HIS- and c-MYC-tagged Drosophila CYLD, the Drosophila CYLD-RC open reading frame (ORF) was cloned into the EcoRI-XhoI site of pAc5.1/c-MYC-HIS A using the following PCR primers: AcCYLD-F:5′-CCCGAATTCCAAAATGATCTTAAACAACAAAAGTAAAAC-3′ and AcCYLD-R:5′-CCCCTCGAGATGGTACATCATTATATCTGTGC-3′. Plasmid pAc5.1/c-MYC-HIS A was generated by replacing the V5 epitope of plasmid pAc5.1/V5-HIS A(Invitrogen) with a c-MYC epitope by ligating the following annealed phosphorylated oligos into BstBI-AgeI-digested pAc5.1/V5-HISA: cMyc-S, 5′-CGAAGAACAAAAACTTATTTCTGAAGAAGATCTGA-3′and cMyc-AS, 5′-CCGGTCAGATCTTCTTCAGAAATAAGTTTTTGTTCTT-3′. V5 and 6×HIS C-terminal-tagged kenny (key), the full-length key ORF, was PCR amplified from the cDNA clone LD18356(Drosophila Genomics Resource Center) using the following primers:AcKEY-F, 5′-CCCGAATTCCAAAATGTCGGACGAAGAGTCATTC-3′ and AcKEY-R,5′-CCCCTCGAGGTTTTTATCCAAACAATCGTTAACG-3′, and cloned into EcoRI-XhoI sites of the pAc5.1/V5-HIS A vector.

The rabbit anti-CYLD antibody was generated by Santa Cruz Biotechnology using a GST fusion of Drosophila CYLD amino acids 1-317 as an antigen.

Embryos from each developmental stage were dechorionated, fixed in 180 mM HEPES pH 6.9, 4 mM MgSO₄ and 2 mM EGTA in 4% formaldehyde and 58.8%heptane, and rinsed with methanol. After rehydration with BBT[1×phosphate-buffered saline (PBS), 0.1% Tween-20, 1% bovine serum albumin (BSA)], embryos were blocked with BBT-250 (BBT, 250 mM NaCl)containing 2% normal goat serum (NGS) for 2 hours, stained with the appropriate primary antibody (400 ng/ml) overnight at 4°C, rinsed the following day with BBT-250 and stained with secondary antibody (1:2000 dilution) for 2 hours at room temperature. Finally, the embryos were rinsed with PBT (1×PBS, 0.1% Tween-20), stained with ABC reagent (Vectastain Elite HRP kit), developed with DAB solution (1 ml PBT, 1 mg DAB, 0.1%H₂O₂) and mounted in liquid Canada Balsam (Sigma). All antibodies were diluted in BBT-250 containing 2% NGS.

Sagittal paraffin sections (8 μm) of entire adult flies were obtained and processed for immunohistochemistry as described previously(Philip et al., 2001; Skoulakis and Davis, 1996). The primary antibody was used at 5 μg/ml and the secondary antibody at 1:2000. After washing, the sections were mounted with DAKO mounting medium(DAKO Corporation). All antibodies were diluted in PBHT (20 mM NaH₂PO₄, 500 mM NaCl, 0.2% Triton X-100, pH 7.4)containing 5% NGS.

Female flies up to 5 days old were fixed for 5 hours at 4°C in PBS containing 4% paraformaldehyde. Subsequently, they were rinsed in PBS for 10 minutes at room temperature and were incubated overnight at 4°C in PBS containing 25% sucrose. The following day, the flies were placed in OCT(Sakura Finetek Europe) and were cryosectioned in 8 μm sections. The sections were placed on gelatin-covered glass slides and rinsed for 10 minutes at room temperature in PBS. Subsequently, they were placed in Nile red solution (0.5 mg/ml Nile red in acetone) for 5 minutes at room temperature,and then they were rinsed three times with PBS and mounted with DAKO mounting medium (DAKO Corporation) for microscopy.

MC1061 bacteria (Escherichia coli) expressing a ubiquitin-Met-β-galactosidase fusion were transformed with plasmids expressing GST, GST-CYLD(541-956), GST-CYLD or GST-CYLD(C284S). Overnight cultures of the transformants were diluted tenfold, grown for 1 hour and then incubated for 4 hours in the presence of 0.1 mM IPTG to induce the expression of GST fusion proteins. Cells (1.5 ml) were lysed in 150 μl of cracking buffer (10 mM phosphate pH 7.4, 8 M urea, 1% SDS, 1% bmercaptoethanol) and boiled for 5 minutes. Extracts were subjected to SDS-PAGE (6% gel) and western blotting with a rabbit anti-β-gal antibody. Expression of GST fusion proteins was detected with SDS-PAGE (8.5% gel) and western blotting using anti-GST antibody.

RNA was extracted from either eight male or five female flies or from the corresponding body parts using the RNA-Wiz kit (Ambion), according to manufacturer's instructions. cDNA was prepared using the Revert Aid cDNA synthesis kit (Fermentas), according to manufacturer's instructions. PCR analysis was performed using the primers DMCYLDF7(5′-TAGAGCCGGAGGAACCTTTAC-3′) and DMCYLDR9(5′-GCATCTGTTGGCTGGTAC-3′); DMCYLDF3 and DMCYLDR6; and rp49F1(5′-GATCGTGAAGAAGCGCAC-3′) and rp49R2(5′-CGCTCGACAATCTCCTTG-3′) for loading control.

For triglyceride (TAG)-content analysis, three groups of eight male flies per genotype, approximately 36 hours old, were homogenized in 100 μl 0.05%Tween-20, incubated at 70°C for 5 minutes and cleared twice by centrifugation at 5000 rpm (6150 g) for 1 minute and 13,000 rpm (16,000 g) for 3 minutes. Supernatants were analyzed for TAG content using a commercial kit (Biosis) according to manufacturer's instructions. Total protein was measured using the Bio-Rad D_CProtein Assay kit (Bio-Rad) according to manufacturer's instructions. For the starvation assay, three groups of 25 male flies per genotype, approximately 36 hours old, were starved in food-free vials with unlimited water supply at 25°C. The death rate was estimated by counting the number of flies unable to exhibit a sit-up response in 12-hour intervals(Gronke et al., 2003).

A cDNA corresponding to the full-length ORF of Drosophila CYLD was cloned into the LexA DNA-binding domain vector pHybLex/Zeo (Invitrogen). The resulting plasmid, pHyb-CYLD, was transformed into the Saccharomyces cerevisiae strain L40 [MATa his3Δ200 trp1-901 leu2-3112 ade2 LYS::(4lexAop-HIS3) URA3::(8)lexAop-lacZ) GAL4] (Invitrogen) and used as bait in a mating-based two-hybrid screen against other members of the IMD pathway (a full list of genes and PCR primer sequences is available upon request). Each pathway member was prepared by cloning their full-length cDNAs into the B42 activation-domain vector pYESTrp2 (Invitrogen) and transformed into the S. cerevisiae strain EGY48 (MATα ura3 trp1 his3 6lexAop-LEU2; Invitrogen). An interaction between JUN (bait;also known as JRA-FlyBase) and FOS (also known as KAY-FlyBase) using the abovementioned plasmids and yeast strains was employed as a positive control. Mating was performed by overlaying 1 μl of an overnight culture of L40 flies with 1 μl of EGY48 on a YPAD plate. Plates were incubated for 1-2 days at 30°C and diploids were selected. Activation of the lacZreporter gene was tested according to the protocols outlined in the Hybrid Hunter handbook (Invitrogen).

Drosophila S2 cells (Invitrogen) were maintained at 25°C in Schneider's Drosophila medium supplemented with 10% heat-inactivated FBS, 50 units ml^-1 penicillin G, and 50 μg/ml streptomycin sulfate (Invitrogen). Transfection was performed using the Effectene transfection reagent (Qiagen) according to the manufacturer's protocol. Briefly, 7.5×10⁵ cells ml^-1 were plated in 60 mm plates 24 hours before transfection. DNA (2 μg) was used for each transfection using a 1:10 Effectene ratio. Cells were washed once in 1×PBS at 8 hours post-transfection and incubated at 25°C for 48 hours.

Cell lysates were prepared from 5 ml cultures of transiently transfected cells with 2 μg each of pAcCYLD/c-Myc and pAcKEY/V5 or 2 μg of pAcKEY/V5 alone, in 100 μl RIPA buffer (150 mM NaCl, 1.0% Igepal CA-0630, 0.5% sodium deoxycholate, 0.1% SDS, 50 mM Tris, pH 8.0) (Sigma-Aldrich), supplemented with Complete Mini Protease Inhibitor Cocktail tablets (Roche Applied Science). Lysates were incubated with 100 μl of antic-MYC agarose affinity gel(Sigma-Aldrich) for 1.5 hours with agitation rocking at room temperature. Immunoprecipitates were washed with CoIP buffer (900 mM NaCl, 1.0% Triton X-100, 0.5% sodium deoxycholate, 0.1% SDS, 50 mM Tris, pH 8.0) six times, each for 10 minutes, followed by a final wash with RIPA buffer alone. Immunoprecipitates were eluted by boiling in 2×SDS sample buffer,separated by 13% SDS PAGE, and transferred to PVDF membranes. Blots were probed with anti-V5 antibody (1:5000 dilution; Invitrogen).

The following strains were made and used in this study: UAS-CYLD(wild type), UAS-CYLD/+; CYLD^f00814;hs-Gal4/MKRS(CYLD rescue) and four insertions from the insertional mutagenesis stocks generated by Exelixis and made available to the fly community via the Bloomington Stock Centre. These were f00814(CYLD^f00814), f02494, d10472 and f00135. Df(2L)j27/CyO flies were obtained from the Bloomington Stock Center. We used a white isogenic strain as a wild-type control in all experiments. A white isogenic strain was used as a background by Exelixis in generating a series of different P element and piggyBac insertions(Thibault et al., 2004). Additional strains used were relish(Hedengren et al., 1999) and dif (Rutschmann et al.,2000a) as positive controls for Gram-negative and Gram-positive bacterial infections, respectively, and the yolkGAL4-drivercontaining strain (Georgel et al., 2001). To rescue the CYLD^f00814 mutant phenotypes, strain UAS-CYLD/+; CYLD^f00814;hs-Gal4/MKRS was incubated at 37°C for 1 hour followed by recovery at 29°C for 1 hour. All strains were analyzed by PCR for the presence of the Gram-negative bacteria Wolbachia and were found not to be infected. Bacterial immune challenge and northern blotting was performed as previously described(Rutschmann et al.,2000a).

Drosophila developmentDrosophila contains a putative ortholog of the mammalian tumor suppressor gene CYLD. Five transcripts of Drosophila CYLD have been identified by genomic approaches (Fig. 1A, www.ensembl.org). The expression pattern of Drosophila CYLD was investigated by reverse transcriptase (RT)-PCR, and CYLD-specific transcripts were found in all developmental stages (embryo, larva and pupa) and in all three major adult body segments (Fig. 1B). The protein distribution pattern was then investigated using an anti-CYLD antibody that we developed (see Materials and methods). The specificity of this antibody to Drosophila CYLD was confirmed because embryos homozygous for the Df(2L)J21 deficiency, which deletes the 31C7-D1 chromosomal area in which CYLD resides, did not show any reactivity to the antibody(Fig. 2D). DrosophilaCYLD was present throughout stage 1-5 embryos (data not shown) but, by stage 7, the protein was detected mainly in mesoderm and in developing segments of the anterior and posterior midgut (Fig. 2A). During embryonic stage 12, CYLD continued to accumulate in the developing midgut and was relatively abundant in the brain and ventral nerve cord (Fig. 2B). By embryonic stage 14, accumulation of CYLD in the brain and ventral nerve cord was notably stronger than that of earlier stages, while the protein remained abundant in the midgut and posterior intestine(Fig. 2C). Together, these data suggest a role for CYLD in the development or function of the midgut and mesodermal derivatives, as well as in the nervous system.

In adult flies, CYLD was detected in several tissues. Strong expression was detected in the cytoplasm of nurse and follicle cells in the developing oocyte(Fig. 2F), and in the testes(Fig. 2G). Weaker protein accumulation was detected in intestinal epithelial cells (data not shown) and in the fat body (Fig. 2H). In addition, CYLD accumulated strongly in the proboscis, antenna(Fig. 2I) and in the eye(Fig. 2J), whereas the protein was weakly detectable in various areas of the brain, such as the optic lobe(data not shown). The reactivity of adult tissues to the anti-CYLD antibody could be almost completely eliminated with the addition of excess antigen that was used to raise the antibody (data not shown).

The high degree of similarity between the ubiquitin-specific protease domain of human CYLD and the corresponding C-terminal region of Drosophila CYLD prompted an investigation into the ability of Drosophila CYLD to act as a deubiquitylating enzyme. For this purpose, Drosophila CYLD was introduced into bacteria expressing a ubiquitin-β-galactosidase (Ub-βgal) fusion, and its ability to cleave the ubiquitin moiety from the Ub-βgal fusion protein was tested. Wild-type Drosophila CYLD readily cleaved ubiquitin from the Ub-βgal fusion (Fig. 3,compare lanes 1 and 3). By contrast, a mutated form of DrosophilaCYLD, containing a serine instead of a conserved cysteine in the putative catalytic domain, was unable to cleave ubiquitin from the Ub-βgal fusion(Fig. 3, compare lanes 1 and 4). As expected, human CYLD was also capable of cleaving the Ub-βgal fusion (Fig. 3, lane 2). Therefore, like its vertebrate ortholog, Drosophila CYLD has autonomous deubiquitylating activity that depends on the integrity of its conserved C-terminal catalytic domain.

In order to study the biological role of CYLD in Drosophila, four strains containing P-element insertions in the CYLD locus were obtained from the Exelixis collection. The genomic locations of all transposons were verified by PCR (data not shown). Strain f00814,which contains a P-element insertion in the first intron(Fig. 4A), was chosen for further analysis because this allele can be considered a strong CYLDhypomorph lacking detectable expression of all possible transcripts from the gene (Fig. 4B). By contrast,the insertions in intron 2 of Drosophila CYLD either did not appear to disrupt or disrupted incompletely the expression of CYLD-associated mRNAs in the f00135, f02494 and d10472 strains. Consistent with the RT-PCR analysis, staining of f00814 homozygous-mutant embryos with anti-CYLD antibody failed to detect CYLD protein (data not shown). Therefore, this strain represents a bona fide mutant allele of the gene, henceforth called CYLD^f00814. Given the protein distribution in developing embryos, CYLD^f00814 homozygotes were surprisingly viable and did not exhibit obvious external morphological anomalies. This indicates that Drosophila CYLD function is either dispensable for proper embryonic development, or it is functionally redundant with another gene in the genome. However, histological analysis revealed that the majority of CYLD^f00814 adult homozygotes exhibited fat body cells of an abnormal size and morphology (Fig. 5B, Table 1). This phenotype was reversed upon inducible expression of a CYLD transgene under the control of the heat-shock promoter in the CYLD^f00814 mutant background(UAS-CYLD/+;CYLD^f00814;hs-Gal4/MKRS, referred to as the CYLD-rescue strain from this point on)(Fig. 5C, Table 1). The absence of CYLD did not precipitate apparent morphological defects in other tissues and organs[compare the oocyte from wild-type (Fig. 5D), f00814 (Fig. 5E) and CYLD-rescue (Fig. 5F) flies]. Nile red staining showed that fat body cells in CYLD^f00814 adult homozygotes were able to accumulate lipids. This in turn indicated that they retained a degree of functionality despite their morphological alterations(Fig. 5G,H). A measurement of lipid droplets revealed that CYLD^f00814 adult homozygotes had significantly more but smaller lipid droplets (5.77±0.26 lipid droplets cm^-2) compared to the wildtype strain w1118(3.08±0.20 lipid droplets cm^-2).

The altered morphology of the fat body prompted an investigation into the fat content of flies with defective CYLD expression. The fat body is the major site of fat storage in Drosophila. Fat is stored primarily in the form of triglycerides (TAGs) in characteristic structures called lipid droplets. These structures consist of a single phospholipid layer that surrounds the TAGs. They are opaque and usually occupy a large portion of the cell, thus restricting the cytoplasm and nucleus to one side, and therefore contributing to the characteristic morphology of the fat body(Gronke et al., 2005). Total TAG content was found to be inversely proportional to the level of Drosophila CYLD expression. More specifically, total TAGs were increased by 79.8% in CYLD^f00814 adult homozygotes compared with control animals (Fig. 6A). However, CYLD^f00814 homozygotes that expressed transgenic Drosophila CYLD contained almost wild-type levels of TAGs (CYLD rescue, Fig. 6A). Significantly, in the absence of the heat-shock required for the expression of Drosophila CYLD transgenes, UAS-CYLD/Y;CYLD^f00814;hs-Gal4/MKRS flies possessed a TAG content that was not significantly different from that of CYLD^f00814 homozygotes[Fig. 6A, CYLD rescue (NO HS)]. Therefore, our data suggest that Drosophila CYLD is required for TAG homeostasis, a function that is most likely related to the altered morphology of the fat body in CYLD^f00814 homozygotes.

The increased levels of TAGs in flies with impaired CYLDexpression suggested that CYLD^f00814 homozygotes might be able to survive for longer periods of time under starvation conditions. For this purpose, newly emerged adults were allowed to feed normally for up to 36 hours and were then placed in vials without food but with free access to water. CYLD^f00814 homozygotes survived significantly longer without a source of nutrients in comparison with controls(Fig. 6B). Control animals and CYLD-rescue flies were dead after 4 days, whereas more than 40% of CYLD^f00814 homozygotes remained alive at the end of this time period. Heat-shock did not significantly alter the resistance to starvation of control animals (data not shown). Collectively, the data suggest that reduced levels of CYLD result in an elevation in the levels of TAGs,which is the most likely cause of the observed resistance to starvation of CYLD^f00814 homozygotes.

Because of the deficits in fat body structure and function described above,and the known principal role of this organ in Drosophila host defence, we wanted to determine whether loss of Drosophila CYLDaltered susceptibility to infections. Furthermore, this would be consistent with the given connection of mammalian CYLD with NF-κB/IKK signaling and with the function of the immune system(Brummelkamp et al., 2003; Kovalenko et al., 2003; Reiley et al., 2006; Trompouki et al., 2003).

The hallmark of the Drosophila systemic immune response is the synthesis and secretion by the fat body of potent AMPs. Although these responses have broad specificity, their pattern of expression depends largely on the type of invading pathogen (Naitza and Ligoxygakis, 2004). Whereas anti-fungal and anti-Gram-positive bacterial responses are controlled by the Toll signaling cascade,Gram-negative sepsis depends on the IMD pathway, the Drosophilahomolog of mammalian TNFR signaling(Kaneko and Silverman, 2005). In human cells, CYLD interacts with NEMO and regulates the activation of NF-κB by deubiquitylating signaling molecules of TNFR-associated pathways (Brummelkamp et al.,2003; Kovalenko et al.,2003; Trompouki et al.,2003). We used a quantitative yeast two-hybrid assay to investigate the ability of Drosophila CYLD to interact with known members of the IMD pathway in Drosophila (PGRP-LC, PGRP-LE, IMD,dFADD, DREDD, dTAK1, IRD5 and KENNY). We found that Drosophila CYLD was able to interact with KENNY, which is the Drosophila counterpart of the mammalian NEMO (IKKγ) protein, but not with other proteins implicated in the IMD pathway (Fig. 7A). The interaction between CYLD and KENNY was confirmed via coimmunoprecipitation in Schneider-2 (S2) cells(Fig. 7B). Next, we investigated the effect of impaired Drosophila CYLD expression on Toll and IMD signaling with and without infection, by measuring the expression of the AMP genes Diptericin (Dipt; DptB) and Drosomycin (Drs) as read-outs for IMD and Toll signaling,respectively. In uninfected flies, Dipt expression in mutant homozygotes was significantly higher than in wild-type flies, whereas no significant difference was observed between CYLD^f00814homozygotes and wild-type flies upon infection with Escherichia coli(Fig. 8A). No significant difference in Drs expression was detected between mutant and wild-type flies without infection or following Gram-positive bacterial challenge (data not shown), suggesting that Drosophila CYLD does not regulate the Toll pathway.

In addition, we also studied the effects of a gain-of-function condition for Drosophila CYLD, whereby UAS-CYLD flies were crossed to the fat body-specific yolk-GAL4 driver and Dipt gene expression was measured (Fig. 8B). Dipt induction was significantly below that of wild type in infected flies.

Interestingly, following infection with Gram-positive or Gram-negative bacteria, CYLD^f00814 homozygotes exhibited a reduced rate of survival compared with wild type (Fig. 9). Flies from this strain showed reduced survival after infection by the Gram-negative bacterium E. coli(Fig. 9A) or by the Gram-positive bacterium Enterococcus faecalis(Fig. 9B), compared with wild-type or CYLD-rescued flies. Parallel experiments with the CYLD^f00814 allele over a deficiency uncovering the Drosophila CYLD locus produced comparable results (data not shown). This was not the case when heat-killed bacteria were used (data not shown). Notably, restoration of Drosophila CYLD expression in CYLD-rescue flies also restored the ability of these flies to respond to both kinds of challenges (Fig. 9, CYLD-rescue curves). Taken together, these results identify a crucial role of CYLD in anti-bacterial responses in Drosophila.

Drosophila melanogaster is emerging as one of the most effective tools for analyzing the function of human disease genes, including those responsible for developmental and neurological disorders, cancer,cardiovascular disease, metabolic and storage diseases, and genes required for the function of the visual, auditory and immune systems. Flies have several experimental advantages, including their rapid life cycle and the large numbers of individuals that can be generated, which make them ideal for sophisticated genetic screens (Bier,2005).

In the present study, we characterized the expression and function of the Drosophila ortholog of the human tumor suppressor gene CYLD. Flies with impaired Drosophila CYLD expression showed severe susceptibility to bacterial infection. By inference from its human counterpart, which is a negative regulator of TNF/NF-κB signaling, it can be hypothesized that IMD-mediated AMP gene expression would be upregulated in CYLD loss-of-function mutant flies, whereas a gain-of-function condition should have the opposite effect. As expected, overexpressing Drosophila CYLD suppressed IMD signaling in uninfected and infected flies. In uninfected null mutants, Dipt expression was twice as high as in wild-type flies. Interestingly, Dipt levels in infected CYLD mutants were not significantly different from wild-type flies,whereas basal levels were elevated in comparison with wild-type controls. The most likely reason for this is that, when septic injury is used as a means of infection, the system is overloaded and Dipt expression reaches its maximum, such that even when an inhibitor of IMD signaling is eliminated, any further increase is not possible.

To our knowledge, this article is the first to describe a deubiquitylating enzyme implicated in the regulation of NF-κB signaling in Drosophila. Furthermore, the discovery that the loss of Drosophila CYLD leads to a concomitant increase in Diptexpression suggests that the IMD pathway, like its mammalian counterpart (the TNFR pathway), is activated by ubiquitylation. Our results suggest that Drosophila IKKγ is a likely target for this activation. Its interaction with Drosophila CYLD could serve as a switch to de-activate the pathway, as has been proposed for the NEMO-CYLD interaction in mammals (Brummelkamp et al.,2003; Kovalenko et al.,2003; Trompouki et al.,2003). An additional IKK-regulating deubiquitylating enzyme, A20(also known as Tnfaip3-Mouse Genome Informatics), has been described in mice. The knock-out effect of its gene has revealed a crucial function in limiting inflammation by terminating TNFα-induced NF-κB responses(Lee et al., 2000). More-recent experiments have revealed that it is, at the same time, a deubiquitylase of K63-linked polyubiquitin chains, and it promotes K48-linked polyubiquitylation of RIP1 (Boone et al.,2004; Evans et al.,2004; Wertz et al.,2004). Drosophila contains an ortholog of A20,and P-element insertions in the 5′ untranslated region (UTR) of this gene have an immune phenotype (A.E.B. and P.L., unpublished). Systematic screening for deubiquitylating enzymes in the fly should strengthen the concept of the non-degradative function of polyubiquitin chains in NF-κB signaling.

De-regulation of the IMD pathway, as exemplified by the observed alteration in Dipt regulation, cannot account, however, for the distinct sensitivity of the Drosophila CYLD mutants to infection. Although measuring AMP gene expression is a good indication of the humoral aspect of the response, overall survival to overt infection is a more general phenomenon involving several pathways and cellular activities(Boutros et al., 2002). An intriguing explanation for the death of CYLD^f00814homozygotes following infection could be offered by the dramatic change in fat body morphology and physiology. The fat body is the major immuno-responsive organ during Drosophila systemic defence and significant changes would be expected to influence survival to immune challenges. Interestingly, a recent report has identified a crucial role for murine CYLD in the development of T lymphocytes (Reiley et al.,2006). Whether higher fat levels could potentially be detrimental as well, remains to be established. A possible hypothesis could be that fly hemocytes, which are macrophage-like cells(Meister and Lagueux, 2003),are not able to engage in phagocytosis because they are overloaded with the function of internalizing excess lipids (that could presumably accumulate in the blood). In the absence of phagocytic activity, bacteria could proliferate freely and kill the host. Following the same line of thinking, mutants with high blood TAGs would also be sensitive to infection. Flies mutant for brummer (bmm), a TAG lipase homologous to human adipocyte TAG lipase, have 100% more TAGs in their blood than their wild-type counterparts (Gronke et al.,2005). Nevertheless, bmm flies showed a survival typical of wild-type flies when challenged with Gram-positive or Gram-negative bacteria (data not shown). This finding indicates that high levels of TAGs are not sufficient to cause immune deficiency and strengthens our hypothesis that fat body structure and function might be the principal determinant of immune deficiency in CYLD^f00814 homozygotes. Collectively,findings by Reiley et al. (Reiley et al.,2006) together with the results of this study establish a clear functional relationship between mammalian and Drosophila CYLD proteins in host defence.

Even though the analysis of CYLD activity in cell lines suggests that it functions as a general negative regulator of IKK and JNK activation by cleaving Ub^Lys63 chains, cylindromatosis patients display a phenotype that is restricted to epidermal body regions. The TNFR family member ectodysplasin A receptor controls the development of epidermal appendices by activating NF-κB (Schmidt-Ullrich et al., 2001). Therefore, under physiological conditions, the role of CYLD may be restricted to specific cell types and stimulatory conditions,which points to a functional redundancy with other ubiquitin hydrolases. Similarly, our analysis showed a broad pattern of Drosophila CYLDexpression in both embryonic and adult tissues. However, a compromise of CYLD expression did not appear to affect embryonic development, but rather the major effect we observed was restricted to the fat body and the physiological properties attached to it (fat storage, host defence). Again,redundancy with proteins performing similar functions in other tissues could be envisaged (Chen and Fischer,2000). Nevertheless, possible functional deficits in the nervous system, where Drosophila CYLD accumulates, were not investigated in CYLD^f00814 animals, but will be the focus of future experiments. We demonstrated that impaired Drosophila CYLD expression is associated with increased TAG content and resistance to nutrient starvation. This result suggests that CYLD is involved in fat storage and/or metabolism in Drosophila, and possibly in mammals. Increased TAG content has been observed in flies with mutated insulin receptor (INR) or Chico, an INR substrate (Bohni et al.,1999; Clancy et al.,2001; Tatar et al.,2001). Therefore, it is possible that Drosophila CYLD is required for proper signal transduction by INR. INR is activated by neuropeptides [insulin-related peptides (ILPs)] homologous to human insulin(Brogiolo et al., 2001). We have found that the expression of Drosophila ILP2, the major ILP, is not affected by the decrease in CYLD levels (data not shown). This finding suggests that Drosophila CYLD might regulate INR signaling downstream of receptor activation. One such possible target is the transcription factor FOXO, which is activated by INR signaling(Goberdhan and Wilson, 2003). It has been shown that FOXO is activated by the JNK pathway, which itself was recently shown to be negatively regulated by CYLD in mammalian cells(Reiley et al., 2004; Wang et al., 2005). In Drosophila, activation of the JNK pathway contributes to oxidative stress resistance (Wang et al.,2003), and this, in conjunction with increased TAG levels observed in CYLD-deficient flies, might explain the resistance of CYLD^f00814 homozygotes to nutrient starvation. Further analysis is required to determine whether Drosophila CYLD, like its human counterpart, is also involved in the regulation of the JNK pathway,establishing further the valuable cross-talk between flies and humans in the context of understanding disease.

The authors would like to thank I. Livadaras and C. Savakis (Institute of Molecular Biology and Biotechnology, Foundation for Research and Technology,Heraklion, Greece) for advice and technical assistance with the generation of transgenic flies; the curators of the Exelixis collection for providing the mutant strains used in this study; D. Sherratt (University of Oxford) for the use of the Fuji-film FLA 3000 phosphorimager; and Steve Tronick (Santa Cruz Biotechnology) for assistance with the generation of the anti-CYLD antibody. This work was supported by an international scholarship from the Howard Hughes Medical Institute and by funding under the Sixth Research Framework Programme of the European Union, Project INCA (LSHCCT-2005-018704) (to G.M.) and by a Career Establishment Grant from the Medical Research Council UK (to P.L.). G.M. is a Scholar of the Leukemia and Lymphoma Society of America.