Abstract
The Toll signaling pathway has a highly conserved function in innate immunity and is regulated by multiple factors that fine tune its activity. One such factor is β-arrestin Kurtz (Krz), which we previously implicated in the inhibition of developmental Toll signaling in the Drosophila melanogaster embryo. Another level of controlling Toll activity and immune system homeostasis is by protein sumoylation. In this study, we have uncovered a link between these two modes of regulation and show that Krz affects sumoylation via a conserved protein interaction with a SUMO protease, Ulp1. Loss of function of krz or Ulp1 in Drosophila larvae results in a similar inflammatory phenotype, which is manifested as increased lamellocyte production; melanotic mass formation; nuclear accumulation of Toll pathway transcriptional effectors, Dorsal and Dif; and expression of immunity genes, such as Drosomycin. Moreover, mutations in krz and Ulp1 show dosage-sensitive synergistic genetic interactions, suggesting that these two proteins are involved in the same pathway. Using Dorsal sumoylation as a readout, we found that altering Krz levels can affect the efficiency of SUMO deconjugation mediated by Ulp1. Our results demonstrate that β-arrestin controls Toll signaling and systemic inflammation at the level of sumoylation.
THE establishment of the immune system during development and subsequent maintenance of its homeostasis in the adult is under the control of several conserved signaling pathways (Evans et al. 2003; Lemaitre and Hoffmann 2007). A proper balance of activating and inhibitory mechanisms ensures a robust defense response to infection and at the same time prevents inappropriate activation of the immune pathways that can lead to inflammation, tissue damage, and cancer.
In both insects and mammals, Toll and related receptors are central for the function of the immune system (Lemaitre and Hoffmann 2007; Ganesan et al. 2011). The core components of the Toll pathway in Drosophila include the ligand Spätzle, the receptor Toll, the intracellular adaptors MyD88 and Tube, and the kinase Pelle. Toll signaling regulates the nuclear accumulation of the NF-κB homologs Dorsal (Dl) and Dif. Inappropriate activation of the Toll pathway often leads to a systemic inflammation phenotype, which is defined as activation of blood cells and elevated expression of innate immunity genes in the absence of infection (Paddibhatla et al. 2010; Ganesan et al. 2011).
An emerging control mechanism involved in the regulation of the core components of Toll/NF-κB signaling is sumoylation. Modification of proteins by SUMO (small ubiquitin-related modifier) has been shown to play a role in controlling multiple cellular functions, such as nucleocytoplasmic transport, transcriptional regulation, protein stability, signal transduction, and cell cycle progression (Geiss-Friedlander and Melchior 2007). In mammalian systems, the effects of sumoylation on NF-κB signaling are diverse and complex, including regulation of IκBα degradation, modulation of NEMO/IKKγ activity, and both activation and repression of transcription by NF-κB (Mabb and Miyamoto 2007). In the Drosophila genome, components of sumoylation machinery are mostly represented by single genes, making it an attractive system to dissect the role of sumoylation in Toll/NF-κB signaling (Talamillo et al. 2008a; Smith et al. 2012).
Genetic analyses as well as experiments in cultured cells showed the important but conflicting roles of sumoylation in regulating Toll signaling in Drosophila. Loss-of-function mutations in lesswright (lwr), which encodes the Drosophila E2 SUMO-conjugating enzyme Ubc9, were found to cause overproliferation of hematopoietic precursors, production of abnormally high levels of the differentiated blood cells called lamellocytes, and formation of lamellocyte-containing melanotic masses (Chiu et al. 2005; Huang et al. 2005; Paddibhatla et al. 2010; Kalamarz et al. 2012). Loss of lwr was also associated with increased expression of Drosomycin (Drs) as well as nuclear accumulation of the Toll transcriptional effector Dl in the hemocytes, suggesting that Ubc9 exerts an inhibitory effect on Toll signaling in larvae and protects the organism from abnormal inflammation (Chiu et al. 2005; Huang et al. 2005). However, in other reports, Ubc9/Lwr was shown to bind Dl directly and promote its sumoylation, which increased Dl nuclear retention and potentiated Dl-dependent transcriptional activation in cultured cells (Bhaskar et al. 2000, 2002). Moreover, components of the sumoylation machinery such as SUMO/Smt3 and Ubc9/Lwr were required for expression of antimicrobial peptide genes in Drosophila larvae and cultured cells after exposure to microbial elicitors (Bhaskar et al. 2002), suggesting a positive role for sumoylation in Toll activation. Recently, sumoylation was shown to be required for antimicrobial gene expression in the second major regulator of immunity in Drosophila, the immune deficiency (IMD) pathway (Fukuyama et al. 2013). Therefore, further investigation at the genetic and molecular levels is required to elucidate the complex role of sumoylation in Toll signaling, immunity, and inflammation.
Another level of control over the Toll/NF-κB pathway is exerted by β-arrestins. β-Arrestins were initially characterized as mediators of G protein coupled receptor (GPCR) desensitization and endocytosis (Pierce and Lefkowitz 2001). More recently, our knowledge of their signaling functions has dramatically broadened to involve a wide variety of signaling pathways and modes of regulation (Kovacs et al. 2009). The Drosophila genome encodes a single ortholog of β-arrestins, Kurtz (Krz) (Roman et al. 2000), which has been implicated in the regulation of GPCR signaling, as well as Notch, Hedgehog, receptor tyrosine kinase, and Toll pathways (Mukherjee et al. 2005; Ge et al. 2006; Tipping et al. 2010; Molnar et al. 2011; Li et al. 2012).
Mammalian β-arrestin proteins can down-regulate NF-κB signaling by binding and stabilizing IκBα (Gao et al. 2004; Witherow et al. 2004) and by preventing autoubiquitination of TRAF6 (Wang et al. 2006). We have found that Krz limits the extent of Toll activation during embryogenesis (Tipping et al. 2010); however, the molecular details of this regulation are unknown. Here, we show for the first time that β-arrestin can control Toll signaling and systemic inflammation at the level of sumoylation. We demonstrate that loss of krz results in an up-regulation of Toll downstream effectors and that Krz exerts its functions by binding to Ulp1, a SUMO protease. Knockdown of Ulp1 leads to an increase in global sumoylation in vivo and causes Toll hyperactivity phenotypes that are similar to the effects of loss of krz. Using Dl as a target for Ulp1-mediated desumoylation, we show that Krz is required for the optimal SUMO protease activity of Ulp1, and that both proteins are necessary for limiting the Toll pathway activity and preventing an inappropriate inflammatory response.
Materials and Methods
Plasmid construction
Drosophila Ulp1, Dorsal, SUMO/Smt3, and human SENP1 (Origene) open reading frames were amplified by PCR using tag and/or restriction site-containing primers and cloned into pMT/V5-His series vectors (Invitrogen) to generate carboxy-terminally tagged Ulp1-V5, Ulp1-SBP, and Dorsal-V5, and amino-terminally tagged Flag-SUMOGG (active processed form of SUMO) and Flag-SENP1. For in vitro translation, HA-Krz and Ulp1-V5 full-length open reading frames and fragments were cloned into the pSP73 vector (Promega). Dl-K382R-V5 was generated using site-directed mutagenesis. HA-Krz, HA-ARRB1, and HA-ARRB2 constructs were described previously (Tipping et al. 2010). Ulp1-V5 was cloned into the pUAST-attB vector (Bischof et al. 2007) for transgenic expression in flies.
Fly stocks
The following lines were described previously: krzc01503 was isolated in a large transposition screen (Thibault et al. 2004), krz1 (Roman et al. 2000; Tipping et al. 2010), Dif1 (Rutschmann et al. 2000), da-GAL4 (Wodarz et al. 1995), Cg-GAL4 (Asha et al. 2003), ppl-GAL4 (Zinke et al. 1999), UAS-Dcr-2 (Dietzl et al. 2007), Histone-GFP (Tipping et al. 2010), Drs-GFP, Dpt-lacZ (DD1, provided by D. Ferrandon) (Manfruelli et al. 1999), and krz5.7 (genomic krz rescue construct) (Tipping et al. 2010). For analysis in homozygous animals, the krzc01503 chromosome was extensively recombined with the yw stock. UAS-driven krz and Ulp1 RNA interference (RNAi) lines were obtained from the Vienna Drosophila RNAi Center. krz1 mutant clones in the fat body were generated by crossing FRT82B krz1/TM6B males with HS-FLP122;;FRT82ubiGFP/TM6B females and heat shocking the progeny embryos for 1 hr at 37°. The UAS-Ulp1-V5 line was generated via phiC31-mediated integration into the attP2 landing site (Bischof et al. 2007; Venken and Bellen 2007).
Immunostaining, antibodies, and lamellocyte counting
Antibodies used for tissue immunostaining were as follows: mouse anti-Dorsal (1:70, Developmental Studies Hybridoma Bank) and rabbit anti-Dif (Rutschmann et al. 2000) (1:200, provided by D. Ferrandon). For Western blotting, the following antibodies were used: guinea pig anti-Krz (Tipping et al. 2010), rabbit anti-pan-arrestin (Affinity Bioreagents), mouse anti-HSP70 (Sigma), rabbit anti-SUMO (provided by A. Courey), rabbit anti-beta-Gal (Molecular Probes), rabbit anti-Flag (Sigma), mouse anti-V5 (Sigma), mouse anti-HA (Sigma), and rabbit anti-GFP (Molecular Probes). Secondary antibodies were from Jackson Immunoresearch (immunofluorescence) and GE Healthcare (Westerns). For immunoflourescence staining of fat bodies, larvae were dissected in 1× PBS and fixed in 4% formaldehyde for 30 min. Tissues were incubated in PBT (1× PBS with 0.1% Tween-20) with 1% BSA (Rockland) with primary antibody overnight at 4° and with secondary antibody for 2 hr at room temperature. Tissues were mounted with Prolong Gold antifade mounting reagent with DAPI (Invitrogen) and images were acquired with a Zeiss LSM510 confocal microscope.
For lamellocyte counting, 10 third instar larvae of each genotype were washed with 1× PBS. Hemolymph was collected into 700 µl of 1× PBS by making a small incision in the cuticle. Each sample was transferred into a well of a six-well plate with a coverslip treated with concanavalin A and incubated for 2 hr to allow cell attachment. Cells were fixed with 4% formaldehyde followed by incubation with Western blocking reagent (Roche) diluted with PBT and anti-lamellocyte primary antibody (1:10, provided by I. Ando) overnight at 4°. Coverslips were washed with PBT and incubated with fluorescent secondary antibody at room temperature. Coverslips were mounted in Prolong Gold antifade mounting reagent with DAPI (Invitrogen), and images acquired with an Olympus BX60 fluorescence microscope. A total of 20–30 fields of view were collected for each condition, and the assay was repeated in biological triplicates. The percentage of lamellocytes was determined by dividing the total number of lamellocytes observed by the total number of DAPI-positive cells.
Cell culture, immunoprecipitation, in vitro translation, and Western blots
Drosophila 529SU cells used for sumoylation assay (Bhaskar et al. 2002) were provided by A. Courey. Schneider 2 (S2) and 529SU cells were cultured at 25° in Schneider’s Drosophila medium (Invitrogen) supplemented with 10% heat-inactivated fetal bovine serum (Invitrogen) and 5% Pen/Strep antibiotics (Invitrogen). For DNA transfections, Effectene transfection reagent (Qiagen) was used according to manufacturer’s instructions. Empty vector was added to maintain the constant amount of DNA in each transfection sample. The following day, cells were induced with 0.35 mM CuSO4 overnight. Cells were then lysed in default lysis buffer [50 mM Tris pH 7.5, 125 mM NaCl, 5% glycerol, 0.2% IGEPAL, 1.5 mM MgCl2, 1 mM DTT, 25 mM NaF, 1 mM Na3VO4, 1 mM EDTA and 2× Complete protease inhibitor (Roche)]. Cell lysates were incubated with anti-HA, anti-Flag, or anti-V5 beads (Sigma) at 4° for 2.5 hr; beads were washed; and proteins were eluted with SDS sample buffer. For Western blotting, proteins were separated on 8% SDS–PAGE and transferred onto PVDF membrane (Millipore). Membranes were blocked for 1 hr with blocking solution (5% nonfat milk, 2% BSA in TTBS: 50 mM Tris-HCl pH 7.5, 0.15 M NaCl, 0.1% Tween-20), incubated with primary antibodies overnight at 4°, and then with HRP-conjugated secondary antibodies (GE Healthcare) for 1 hr at room temperature. Signals were developed with SuperSignal West Dura Extended Duration Substrate (Thermo Scientific) and images were acquired using the Kodak 4000R Image Station.
For in vitro translation, proteins were produced from pSP73 vector-based constructs using TNT high-yield wheat germ in vitro system (Promega) according to manufacturer’s instructions. A total of 200 µl of 1× PBS was added to each reaction and samples were incubated with 20 µl anti-V5 beads at 4° for 2.5 hr. Samples were washed with 1× PBS and proteins were eluted with SDS sample buffer and analyzed by Western blotting.
Sumoylation analysis
Third instar larvae were collected in 1.5 ml Eppendorf tubes and washed with 1× PBS. Larvae were homogenized in default lysis buffer (see above) supplemented with 10 mM N-ethylmaleimide (NEM). Extracts were cleared by centrifugation at 15,000 rpm at 4° for 15 min, mixed with SDS sample buffer, and analyzed by Western blotting. For Dl sumoylation assay in cultured cells, appropriate constructs were transfected into 529SU cells as described above, cells were pelleted in 1.5-ml Eppendorf tubes at 500 g for 1 min, and washed once with 1× PBS. Cells were resuspended directly in 50 µl SDS sample buffer, and samples were heated at 90° for 5 min, followed by Western blotting (see above).
Quantitative real-time PCR
RNA was extracted from larval tissues using Absolutely RNA miniprep kit (Agilent). cDNA was prepared with Protoscript First Strand cDNA synthesis kit (New England BioLabs). RT–PCR was performed using DyNAmo SYBR Green qPCR kit (Thermo Scientific) on an Mx3005P qPCR machine (Agilent). RpL32 signal was used for normalization. Primers used for RT–PCR were as follows: Drs, ACCAAGCTCCGTGAGAACCTT/TTGTATCTTCCGGACAGGCAG and RpL32, GCTAAGCTGTCGCACAAATG/GTTCGATCCGTAACCGATGT.
Identification of Ulp1 peptides by mass spectrometry
Krz-interacting proteins were purified from cultured cells and embryos as described in Kyriakakis et al. (2008). Final samples obtained after affinity purifications were analyzed by LC-MS/MS as described previously (Bouwmeester et al. 2004). Ulp1 peptides were detected in samples from both cultured cells and embryos.
Production of double-stranded RNA and RNAi in cell culture
DNA fragments of 500–700 bp, complementary to mRNA sequences for the targeted proteins, were amplified by using PCR. A T7 RNA polymerase promoter sequence (TAATACGACTCACTATAGGGAGA) was included at the 5′ end of each primer. Conditions of double-stranded RNA (dsRNA) production and RNAi in Drosophila cell culture have been decribed previously (Clemens et al. 2000). PCR products were purified with QIAquick PCR purification kit (Qiagen), transcribed with RiboMAX Large Scale RNA Production system (Promega), and 25 µg of dsRNA was added to each well of a six-well plate containing 3 × 106 cells per well. The following primers were used for making dsRNA templates: Bla (beta-lactamase, for control treatments), CTAATACGACTCACTATAGGGAGACCACGAGTGGGTTACATCGAACTGG/CTAATACGACTCACTATAGGGAGACCACGTTCATCCATAGTTGCCTGAC; krz, TAATACGACTCACTATAGGGAGACCACAGTCTCGGTATTATGGTCCAC/TAATACGACTCACTATAGGGAGACCACATTCGACACTTTCAGCTTAAACC; and Ulp1, TAATACGACTCACTATAGGGAGACCACCTCAAGTCACAGTGAAAAGTG/TAATACGACTCACTATAGGGAGACCACGGTGTACTACTTCCTAAGGG. DNA template for making Ulp1 dsRNA was generated in an overlap PCR reaction from two segments of the 5′- and 3′-untranslated regions, with the following overlapping primers: GTAGCGAACACACGCGCGATGCGAATAAGG/CCTTATTCGCATCGCGCGTGTGTTCGCTAC. The sequence used for making Ulp1 dsRNA did not overlap with the coding region and therefore did not interfere with expression of Ulp1-SBP protein.
Results
Loss of krz leads to an increased production of lamellocytes and an up-regulation of Toll signaling
To study the effects of krz loss of function on Toll signaling, we analyzed a homozygous lethal P-element insertion allele, krzc01503 (Thibault et al. 2004), as well as various conditions of dsRNA-mediated krz knockdown. Full-length Krz protein was not detectable in krzc01503 homozygous larvae, but its expression, as well as adult viability, was restored in the presence of a genomic rescue construct, krz5.7 (Supporting Information, Figure S1A). A knockdown of krz in whole larvae using the da-GAL4 driver (Wodarz et al. 1995), or in the fat body and the lymph gland using the Cg-GAL4 driver (Asha et al. 2003), in the presence of coexpressed Dicer-2 (Dcr-2) (Dietzl et al. 2007), also almost completely eliminated the Krz protein (Figure S1, A and B). We observed melanotic masses in homozygous krzc01503 third instar larvae (Figure 1B), as well as in larvae in which krz was knocked down using the Cg-GAL4 driver (Figure 1C), consistent with a previous report showing melanotic mass formation in krz1 homozygotes (Roman et al. 2000). Melanotic masses were detected not only next to the fat body cells, but also in and near the lymph gland, nervous system, dorsal vessel, brain stem, and in free circulation. Melanotic mass formation is often associated with increased numbers of lamellocytes, which normally constitute only a small proportion of all blood cells (Minakhina and Steward 2006). Using a lamellocyte-specific antibody (gift of I. Ando), we observed a ninefold increase in the percentage of circulating lamellocytes in krzc01503 homozygous third instar larvae and a threefold increase in Cg-GAL4-driven RNAi knockdown animals, compared to controls (Figure 1D). Lamellocyte counts returned to normal levels when a genomic krz rescue construct was combined with krzc01503, confirming that this defect was specifically attributable to loss of krz (Figure 1D).
Loss of krz leads to an increased production of lamellocytes and activation of Toll signaling. (A–C) Melanotic mass formation in control and krz loss-of-function larvae (arrows). (D and E) Quantification of circulating lamellocytes in the hemolymph of third instar larvae as percentage of all hemocytes. (D) Proportion of lamellocytes increased approximately ninefold in krzc01503 homozygous larvae and threefold in RNAi knockdown animals, relative to yw controls. (E) Proportion of lamellocytes was significantly lower in Dif1; krzc01503 animals, compared to krzc01503 homozygotes. (F) Live fluorescence image of third instar larvae carrying Drosomycin-GFP reporter as part of the DD1 chromosome. Drs-GFP was highly expressed in krzc01503 homozygotes. Arrows indicate melanotic masses. (G) Western blot analysis of expression from Drs-GFP and Dpt-lacZ (beta-gal) reporters in whole third instar larvae. Drs-GFP expression was not detected in the DD1 line but was highly elevated in DD1; krzc01503 animals, whereas Dpt-LacZ levels were not affected. (H) Quantitative RT–PCR of endogenous Drosomycin gene expression in whole second instar larvae. Drs levels were higher in krzc01503 homozygous larvae but were reduced when krzc01503 was combined with either loss of Dif or genomic krz rescue (krz5.7). **P < 0.01; *P < 0.05. Error bars represent standard deviation. a.u., arbitrary units.
Formation of melanotic masses and an increased differentiation of lamellocytes were reported for several mutations that cause hyperactivation of Toll signaling (Gerttula et al. 1988; Roth et al. 1991; Lemaitre et al. 1995), suggesting that Krz may be required for limiting the activity of the Toll pathway. To determine whether Toll pathway activity is responsible for the melanotic mass phenotype in krz mutants, we combined the krzc01503 allele with loss of Dif, which is a Toll effector protein involved in lamellocyte differentiation (Huang et al. 2005). Lamellocyte counts were significantly lower in Dif1; krzc01503 larvae, compared to krzc01503 mutants (Figure 1E), suggesting that this phenotype in krz mutant animals is dependent on downstream Toll pathway activation.
We next studied the effects of krz loss on Toll target gene expression. We used a fly line (DD1) that carries two immunity-related reporters: Drosomycin (Drs) promoter-GFP, which responds to Toll activation, and Diptericin (Dpt) promoter-LacZ, which is downstream of IMD signaling (Manfruelli et al. 1999). Live imaging of DD1; krzc01503 larvae revealed a strong up-regulation of Drs-GFP expression (Figure 1F), which was also confirmed by Western blotting (Figure 1G). In contrast, Dpt-LacZ expression was not affected (Figure 1G). These results indicate that loss of krz function specifically affects Toll signaling, without altering the IMD pathway.
For a more direct measurement of endogenous Drs expression, we performed quantitative RT–PCR. Analysis of expression in second instar larvae showed that Drs levels were modestly but significantly higher in krzc01503 homozygotes, compared to controls, and that this effect could be reversed when a genomic rescue construct was present (Figure 1H). Importantly, Drs levels were similar to controls in Dif1; krzc01503 double mutant second instar larvae, suggesting that Dif is required for the observed overexpression of Drs in krz mutants. In krzc01503 third instar larvae, however, we did not detect significant changes in Drs expression (data not shown). It is possible that krz mutants accumulate multiple developmental defects by the third instar stage, which may have some adverse effects on gene expression. Consistent with such a possibility, overall level of Dl was lower in krzc01503 third instars (Figure S1C), and clonal analysis showed a significantly lower level of Dl and Dif in krz1 mutant fat body cells (Figure S1, D–E′′′).
Expression of downstream targets is induced by Dl and Dif that accumulate in the nuclei in response to activated Toll signaling (Ip et al. 1993; Lemaitre et al. 1995, 1996). We asked whether localization of Dl and Dif was affected by loss of krz function. In the fat bodies of control yw third instar larvae, both Dl and Dif were evenly distributed throughout the cell (Figure 2, A–A′′ and C–C′′). In contrast, both proteins showed predominantly nuclear localization in third instar larval fat bodies expressing krz dsRNA under the control of the Cg-GAL4 driver (Figure 2, B–B′′ and D–D′′). Consistently, staining of fat bodies from krzc01503 second instar larvae showed that Dl was nuclear in most cells (Figure S1, F–G′′′), and in krz1 mutant clones residual Dl and Dif proteins accumulated in the nuclei, though the latter effect could also be due to a more rapid degradation of Dl and Dif in the cytoplasm (Figure S1, D–E′′′). Taken together, formation of melanotic masses, increased levels of lamellocytes, nuclear accumulation of transcriptional effectors, and hyperactivation of a downstream target gene in krz mutants suggest that Krz is necessary to prevent inappropriate activation of Toll signaling in Drosophila larvae. Loss of krz function results in Toll pathway hyperactivity, which can contribute to systemic inflammation phenotypes observed in krz mutants, even in the absence of infection by pathogens.
Loss of krz results in nuclear accumulation of Dorsal and Dif. (A–D′′) Antibody staining of Dorsal and Dif proteins (red) in the fat bodies of third instar larvae. DAPI (blue) marks the nuclei. Dl (A–A′′) and Dif (C–C′′) staining of the fat bodies from control yw larvae showed diffuse subcellular distribution. In the fat bodies of larvae coexpressing krz dsRNA and Dcr-2 with the Cg-GAL4 driver, both Dl (B–B′′) and Dif (D–D′′) were preferentially localized in the nuclei.
Krz binds to the Ulp1 protein
To gain insight into possible molecular mechanisms of Krz involvement in controlling Toll signaling and systemic inflammation, we carried out a proteomic analysis of Krz interacting proteins in embryos and cultured cells (Kyriakakis et al. 2008). Among the interactors identified in both cells and embryos was Ulp1, a SUMO protease (Figure S2A) (Bhaskar et al. 2002; Smith et al. 2004). To validate the binding between Krz and Ulp1, we expressed the HA-Krz and Ulp1-V5 proteins in Drosophila S2 cells and were able to co-immunoprecipitate (co-IP) them in both directions (Figure 3, A and B), confirming that Krz and Ulp1 form a complex. The binding between these proteins appears to be direct, as Ulp1-V5 could also be efficiently immunoprecipitated by HA-Krz when both proteins were produced using in vitro translation (Figure S2C). Mapping of a possible Ulp1-Krz interaction region identified a carboxy terminal domain (part C) of Ulp1 as the most likely interaction surface, with the middle portion of the Ulp1 protein (part B) contributing to binding (Figure S2, B and D). Part C in this study corresponds to the conserved catalytic domain in Ulp1, which is responsible for its desumoylation activity.
Ulp1 is a Krz-interacting protein, and its knockdown results in Toll pathway activation. (A and B) Co-immunoprecipitation of Krz and Ulp1 from Drosophila S2 cells. HA-Krz and Ulp1-V5 were transfected in the indicated combinations, immunoprecipitated with V5 or HA affinity resin, and immunoblotted with the corresponding antibodies. Binding between HA-Krz and Ulp1-V5 was observed in both directions of co-IP. (C) Co-immunoprecipitation of mammalian β-arrestins and SENP1 from Drosophila S2 cells. HA-ARRB1, HA-ARRB2, and Flag-SENP1 were transfected in the indicated combinations, immunoprecipitated with HA affinity resin, and immunoblotted. Flag-SENP1 formed a complex with β-arrestin 2 but not β-arrestin 1. (D) Melanotic mass formation in a third instar larva in which Ulp1 dsRNA and Dcr-2 were expressed using the da-GAL4 driver. Melanotic masses are visible throughout the body. (E) Quantification of circulating lamellocytes in the hemolymph of third instar larvae as percentage of all hemocytes. Proportion of lamellocytes increased ∼15-fold in Ulp1 knockdown animals. (F) Quantitative RT–PCR of endogenous Drosomycin gene expression in whole third instar larvae. Drs levels were 60-fold higher in Ulp1 knockdown animals, compared to controls. (G–H′′) Immunostaining of Dorsal protein (red) in the fat bodies of third instar larvae. DAPI (blue) marks the nuclei. Compared to the Hist-GFP controls (G–G′′), Dl nuclear levels were higher in the Ulp1 RNAi larvae (H–H′′). **P < 0.01. Error bars represent standard deviation. a.u., arbitrary units; IB, immunoblot; IP, immunoprecipitation.
To determine whether the interaction between Krz and Ulp1 that we discovered is evolutionarily conserved, we tested whether human homologs of Krz (β-arrestins 1 and -2) and Ulp1 (SENP1) can bind each other. HA-tagged β-arrestin 2 (HA-ARRB2), but not β-arrestin 1 (HA-ARRB1), was immunoprecipitated with Flag-SENP1 when expressed in Drosophila S2 cells (Figure 3C). These experiments have therefore identified a novel, conserved interaction between Krz/β-arrestin and Ulp1/SENP1.
Knockdown of Ulp1 results in Toll pathway hyperactivity
Based on our observation of excessive Toll pathway activity in krz mutants, we asked whether loss of Ulp1 function would also affect Toll signaling. A single P-element insertion that disrupts the Ulp1 gene, P{lacW}Ulp1G0026 (Peter et al. 2002), has been lost (U. Schäfer, personal communication), and the next nearest P-element insertion is located outside of the Ulp1 transcript and is homozygous viable without an apparent phenotype (A. Veraksa, unpublished observation). As a genetic lesion in the Ulp1 gene is not available, we analyzed larvae in which the function of Ulp1 was reduced by expression of Ulp1 dsRNA and Dcr-2 using the ubiquitous da-GAL4 (Wodarz et al. 1995) driver. Ulp1 RNAi knockdown larvae developed extensive melanotic masses located throughout the body and did not survive beyond third instar stage (Figure 3D). The proportion of circulating lamellocytes was increased ∼15-fold in Ulp1 RNAi third instars, compared to yw controls (Figure 3E). Quantitative RT–PCR showed that the endogenous Drs gene was up-regulated by ∼60-fold in the Ulp1 knockdown animals (Figure 3F). The Toll pathway transcriptional effector Dl was predominantly nuclear in many fat body cells of Ulp1 RNAi third instar larvae (Figure 3, G–H′′). These phenotypes together suggest that, like loss of krz, reduction in Ulp1 function leads to hyperactivation of Toll signaling and an inflammatory syndrome in the absence of infection, manifested in the differentiation of lamellocytes, formation of melanotic masses, and increased expression of antimicrobial peptide genes.
Ulp1 desumoylates Dorsal and other proteins
Because Ulp1 functions as a SUMO protease (Smith et al. 2004), we asked whether reduced levels of Ulp1 would affect sumoylation in larvae. Knockdown of Ulp1 using da-GAL4 and Dcr-2 resulted in a strong increase of global sumoylation in third instars (Figure 4A), consistent with previous Ulp1 RNAi experiments performed in cultured cells (Smith et al. 2004, 2011). Knockdown of krz using the same driver did not change the pattern of overall sumoylation in third instar larvae (Figure 4A).
Desumoylating activity of Ulp1. (A) Western blot analysis of overall sumoylation levels in whole third instar larvae. Knockdown of Ulp1 significantly increased global sumoylation in vivo, whereas knockdown of krz did not. (B) Sumoylation of Dorsal in 529SU cells. Cells were treated with control (Bla, beta-lactamase), krz, or Ulp1 dsRNA and transfected with the indicated Dl constructs together with Flag-SUMOGG. Total cell extracts were immunoblotted with the indicated antibodies. Knockdown of Ulp1 resulted in an increased sumoylation of Dl. Transfection of high levels of Ulp1-SBP completely eliminated Dl sumoylation. A mutated form of Dl (K382R) was not sumoylated under any condition. Arrow indicates the nonsumoylated form of Dl; arrowhead marks the position of the sumoylated form. (C) Quantification of Dl sumoylation in three independent experiments. A ratio of sumoylated to a nonsumoylated form is shown. *P < 0.05. Error bars represent standard deviation.
Dl is sumoylated on lysine 382 and desumoylated by Ulp1 in cultured cells (Bhaskar et al. 2002). Our attempts to detect Dl sumoylation in vivo were not successful, so we studied the effects of Krz and Ulp1 on Dl sumoylation using a Drosophila 529SU cell line (gift of A. Courey), in which sumoylation is enhanced by coexpression of Smt3 (Drosophila SUMO) and Ubc9/Lwr (an E2 SUMO conjugation enzyme) (Bhaskar et al. 2002). As would be expected from changes in global sumoylation observed in vivo (Figure 4A), knockdown of krz did not appreciably alter the level of Dl sumoylation in 529SU cells, whereas Ulp1 RNAi resulted in a significant increase in Dl sumoylation (Figure 4B, quantified in Figure 4C). Overexpression of Ulp1 tagged with a streptavidin binding peptide (Ulp1-SBP) completely eliminated Dl sumoylation in both krz and Ulp1 RNAi conditions, as well as reduced global sumoylation. A nonsumoylatable form of Dl, Dl-K382R (Bhaskar et al. 2002), was not sumoylated in this assay, confirming the identity of the Dl-SUMO band (Figure 4B). Taken together, these results show that reduction of Ulp1 function resulted in an increase in global sumoylation in vivo and in cultured cells, whereas a knockdown of krz alone did not have a detectable effect. Sumoylation of Dl, which is one of the Ulp1 targets, followed this overall pattern.
Synergistic effects of Krz and Ulp1 on sumoylation and systemic inflammation
We then asked whether krz and Ulp1 interact genetically. When either krz or Ulp1 was singly knocked down using Cg-GAL4 (without Dcr-2, for a weaker knockdown effect), no melanotic masses were observed in third instar larvae (Figure 5, A and B). Remarkably, when krz and Ulp1 dsRNAs were coexpressed with Cg-GAL4, multiple melanotic masses formed in third instar larvae (Figure 5C). Correspondingly, the levels of circulating lamellocytes were similar to those in controls in single krz or Ulp1 knockdown conditions, but were 10-fold higher when krz and Ulp1 were knocked down together (Figure 5D). Simultaneous knockdown of krz and Ulp1 using Cg-GAL4 also led to a 25-fold up-regulation of Drs (Figure S5A). Moreover, when the da-GAL4 driver was used for single knockdowns without Dcr-2, krz and Ulp1 RNAi animals survived to adulthood, but a simultaneous knockdown of both krz and Ulp1 with da-GAL4 resulted in 100% lethality at the larval stages. The phenotype of a combined weak knockdown of both krz and Ulp1 was therefore similar to a strong knockdown of either krz or Ulp1 alone (compare Figures 1, 3, and 5), suggesting that these genes are involved in the same developmental pathway whose disruption results in an increased production of lamellocytes and formation of melanotic masses.
Krz and Ulp1 synergistically control sumoylation and Toll pathway activity. (A–C) Formation of melanotic masses in third instar larvae. When expressed without Dcr-2 using the Cg-GAL4 driver, knockdown of neither krz nor Ulp1 alone resulted in melanotic mass formation (A and B). However, extensive melanotic masses were observed when krz and Ulp1 dsRNAs were coexpressed (C, arrows). (D) Quantification of circulating lamellocytes in the hemolymph of third instar larvae as percentage of all hemocytes. Proportion of lamellocytes increased ∼10-fold in the double knockdown animals, compared to yw controls. (E) Quantitative RT–PCR of endogenous Drosomycin gene expression in whole third instar larvae grown at 29°, using the ppl-GAL4 driver. Drs levels were 3.5-fold higher in the double knockdown animals, compared to controls. (F and G) Synergistic effects of Krz and Ulp1 on Dl sumoylation in 529SU cells. (F) Cells were treated with control (Bla, beta-lactamase), krz, and/or Ulp1 dsRNA and transfected with the Dl-V5 construct together with Flag-SUMOGG and increasing (but small) amounts of Ulp1-SBP. Total cell extracts were immunoblotted with the indicated antibodies. Arrow indicates the nonsumoylated form of Dl; arrowhead marks the position of the sumoylated form. (G) Quantification of Dl sumoylation in three independent experiments. Knockdown of both krz and Ulp1 resulted in a significantly higher level of Dl sumoylation, compared to a knockdown of Ulp1 alone. (H) Model summarizing the activity of Krz and Ulp1 in controlling sumoylation and Toll pathway activity. Krz promotes desumoylating activity of Ulp1 toward Dorsal and other targets. A proper balance of sumoylation is important for limiting Toll pathway activity and preventing an inappropriate inflammatory response in the absence of pathogens. **P < 0.01; *P < 0.05. Error bars represent standard deviation.
The Cg-GAL4 driver is expressed both in the fat body and in the lymph gland (Asha et al. 2003), and it is possible that the phenotypes in double knockdown animals involved some communication between these organs. To test for the specificity of the observed effects, we used the ppl-GAL4 driver, which is expressed exclusively in the fat body (Zinke et al. 1999). Double knockdown of both krz and Ulp1 at 29° resulted in a significant up-regulation of endogenous Drs expression, suggesting that the loss of these two genes specifically in the fat body is sufficient to generate a mutant phenotype (Figure 5E).
To gain insight into a possible functional significance of Krz and Ulp1 interaction, we studied Dl sumoylation in 529SU cells under the conditions of single or joint knockdown of krz and Ulp1. As expected from the results described above (see Figure 4B), knockdown of krz alone did not increase Dl sumoylation, whereas knockdown of Ulp1 did (Figure 5F, lanes 2 and 3). However, when cells were treated with dsRNA for both krz and Ulp1, Dl sumoylation was further increased to levels that were significantly higher than those observed for Ulp1 RNAi alone (Figure 5F, lane 7). These results were confirmed in three independent experiments (Figure 5G). Furthermore, we could titrate out Dl sumoylation by expressing small increasing amounts of Ulp1-SBP in the background of Ulp1 knockdown (Figure 5F, lanes 4–6). When both krz and Ulp1 were knocked down, Dl sumoylation was more persistent and required higher levels of Ulp1-SBP to achieve the same degree of reduction (Figure 5F, lanes 8–10). A similar pattern could be seen for overall protein sumoylation levels in this experiment (Figure 5F, SUMO blot). Taken together, these results suggest that Krz is a cofactor for the Ulp1 desumoylation activity, and that this role of Krz becomes apparent when Ulp1 is present in limiting amounts.
Discussion
In this study, we present evidence that reveals a novel function of the β-arrestin Krz in controlling sumoylation and preventing systemic inflammation in Drosophila larvae. Our results can be summarized in the model presented in Figure 5H: at least part of Krz protein function is carried out via its direct association with Ulp1, a SUMO protease. Krz and Ulp1 synergistically regulate the level of sumoylation of multiple proteins, including the Toll transcriptional effector Dorsal. Sumoylation can regulate nuclear localization, transcriptional activity, and other protein functions that converge to control the Toll pathway and possibly other regulators of the larval immune system. The effects of Ulp1 and Krz counterbalance the function of Ubc9/Lwr, which promotes sumoylation. A proper balance of sumoylation is important for preventing inappropriate inflammatory responses in the absence of pathogens.
Modulation of Ulp1 activity by Krz was most easily revealed in sensitized backgrounds of partial Ulp1 loss of function. While loss of krz by itself did not have a detectable effect on sumoylation, it showed a clear synergism with loss of Ulp1, which was manifested in greater sumoylation of Dl and other proteins and stronger phenotypes in joint krz and Ulp1 knockdown conditions (Figure 5). Our data suggest that normally the Krz protein facilitates the SUMO deconjugation function of Ulp1, but the molecular details of this interaction are currently unknown. One possibility is that Krz affects the binding of Ulp1 to its substrates. In transfected cells, the binding between Ulp1 and Dl was not affected by a knockdown of krz (Figure S3), making this a less favorable mechanism. It is also possible that Krz alters Ulp1 enzyme activity. We speculate that this is the more likely scenario given that Krz binds to Ulp1 in or near its catalytic domain (Figure S2).
The Toll pathway is one of the signaling systems controlled at the level of sumoylation. Our findings extend previous reports on the role of sumoylation in regulating Toll signaling and systemic inflammation in Drosophila and underscore the complexity of this regulation. It was previously shown that a reduction in global sumoylation in lwr (Ubc9) mutants resulted in hyperactivation of Toll signaling and induced melanotic mass formation (Chiu et al. 2005; Huang et al. 2005; Paddibhatla et al. 2010; Kalamarz et al. 2012). In this study, knockdown of Ulp1, which causes an increase in overall sumoylation, was surprisingly found to result in a similar systemic inflammation phenotype. Moreover, we found that overexpression of Ulp1-V5, which is predicted to decrease global sumoylation, also led to an abnormally high Drs expression and accumulation of Dl in the nucleus (Figure S4). Previous studies using assays in cultured cells showed that sumoylation generally increases the activity of Dl as a transcription factor and also increases its nuclear retention (Bhaskar et al. 2000, 2002), and it is likely that these factors contribute to the hyperactivation of Toll signaling in Ulp1 knockdown animals. However, since mutations in general sumoylation components such as Ubc9/Lwr and Ulp1 affect sumoylation of many targets in addition to Dl, it seems that the resulting phenotypes cannot be attributed solely to changes in the activity of Dl, but likely involve additional regulators.
When analyzing the effects of loss of krz and Ulp1 on Toll signaling, we noted that certain aspects of the mutant phenotypes were stronger than others. Thus, lamellocyte differentiation and Dl and Dif nuclear localization were altered more readily compared to the induction of antimicrobial peptides, particularly when the latter was compared to the condition of overexpression of Toll10b, an extreme gain-of-function Toll allele (Erdelyi and Szabad 1989) (Figure S5A). Correspondingly, a simultaneous knockdown of krz and Ulp1 had a modest effect on the ability of larval hemolymph to reduce the titer of pathogenic bacteria (Figure S5B). We speculate that this may be explained by sumoylation affecting only a subset of Toll functions, whereas the activated Toll receptor is capable of inducing all possible downstream effects, achieving maximum activation of its target genes.
At present, it is not known whether all of the observed effects in krz and Ulp1 mutant conditions are cell autonomous, though our clonal analysis suggests that most of the defects such as the accumulation of residual Dl in the nucleus were observed only in the cells mutant for krz (Figure S1), and a simultaneous knockdown of krz and Ulp1 specifically in the fat body resulted in Drs induction (Figure 5E). It is still possible that a communication between the fat body and other tissues contributes to systemic inflammation phenotypes observed in the krz and Ulp1 mutants, as was suggested for other mutants affecting sumoylation (Paddibhatla et al. 2010). Notwithstanding, the requirement for Toll pathway activation in this process is clear, as demonstrated by significant decreases in the severity of the phenotypes observed in the background of Dif1 mutation (Figure 1).
In addition to Toll, other signaling pathways may also be affected by global changes in sumoylation and thus contribute to the overall Ulp1 (and by extension, lwr) loss of function phenotypes. For example, sumoylation can control the activity of c-Jun N-terminal kinase (JNK) and Janus kinase/Signal transducer and activator of transcription (JAK/STAT) pathways in Drosophila, both of which are known to be involved in the regulation of immune system homeostasis (Gronholm et al. 2010; Huang et al. 2011). Other pathways controlled by sumoylation, including Ras (Nie et al. 2009), ecdysteroid signaling (Talamillo et al. 2008b), and the IMD pathway (Fukuyama et al. 2013), may also contribute to the observed effects such as systemic inflammation.
Our data have revealed a new regulatory mechanism for β-arrestin-mediated control of cellular signaling through modulation of sumoylation. Because the interaction between Krz/β-arrestin and Ulp1/SENP1 is conserved, it is likely that this mode of regulation is general and exists in other organisms. It will be of interest to determine whether mammalian β-arrestins can regulate Toll/NF-κB and other signaling pathways at the level of sumoylation.
Acknowledgments
We thank A. Courey for providing anti-SUMO antibody and the 529SU cell line, D. Ferrandon for the DD1 line and anti-Dif antibody, I. Ando for the lamellocyte-specific antibody, and S. Shvartsman for the Hist-GFP fly line. We also thank Matt Garrity for helpful comments on the manuscript. Several fly lines were obtained from the Bloomington Drosophila Stock Center at Indiana University and from the Vienna Drosophila RNAi Center, Austria. Anti-Dorsal monoclonal antibody was obtained from the Developmental Studies Hybridoma Bank. This work was supported by National Institutes of Health (NIH) grants GM097727 and CA156734 and National Science Foundation grant 0640700 (to A.V.). Y.T.I. is supported by NIH grant DK83450 and is a member of the UMass Diabetes and Endocrinology Research Center (DERC) (DK32520), the UMass Center for Clinical and Translational Science (UL1TR000161), and the Guangdong Innovative Research Team Program (no. 201001Y0104789252).
Footnotes
Communicating editor: N. Perrimon
- Received December 18, 2012.
- Accepted September 24, 2013.
- Copyright © 2013 by the Genetics Society of America
