|Home | About | Journals | Submit | Contact Us | Français|
The Spalt-like family of zinc finger transcription factors is conserved throughout evolution and is involved in fundamental processes during development and during embryonic stem cell maintenance. Although human SALL1 is modified by SUMO-1 in vitro, it is not known whether this post-translational modification plays a role in regulating the activity of this family of transcription factors. Here, we show that the Drosophila Spalt transcription factors are modified by sumoylation. This modification influences their nuclear localization and capacity to induce vein formation through the regulation of target genes during wing development. Furthermore, spalt genes interact genetically with the sumoylation machinery to repress vein formation in intervein regions and to attain the wing final size. Our results suggest a new level of regulation of Sall activity in vivo during animal development through post-translational modification by sumoylation. The evolutionary conservation of this family of transcription factors suggests a functional role for sumoylation in vertebrate Sall members.
The Spalt-like (Sall) zinc finger transcription factors are involved in diverse biological processes such as organogenesis, maintenance of embryonic stem cells, and carcinogenesis (1, 2). In humans, mutations in SALL genes have been associated with inherited syndromes such as the Townes-Brocks syndrome, caused by mutations in SALL1, and the Okihiro syndrome, caused by mutations in SALL4 (3,–5). In Drosophila, the Sall family is formed by Spalt (Sal) and Spalt-related (Salr) (6, 7). During larval stages, Sall proteins are involved in the formation of the central part of the wing imaginal disc (8). Their expression domain encompasses the longitudinal LII provein until the LV provein. There, the levels of Sall proteins are not homogeneous, with a reduced expression domain that overlaps with provein LII (see Fig. 5A, inset) (9). The absence or ectopic expression of sall genes in the imaginal disc has important phenotypic consequences in the adult wing, affecting its final size and the formation of LII and LV (8, 10). This is due to the deregulation of Sall target genes, such as the knirps complex (kni). Kni is a zinc finger transcription factor, the expression of which in provein LII regulates the genes necessary for this vein differentiation (11).
In humans, SALL1 is modified in vitro by SUMO-1 (small ubiquitin-like modifier 1) (12); however the consequences of this modification are unknown. SUMO is structurally related to ubiquitin, although in contrast to ubiquitylation, sumoylation does not generally direct target proteins to degradation but influences properties such as subcellular localization, activity, and stability (13). Drosophila has a unique SUMO homolog, Smt3, closely related to human SUMO-3, with the rest of the sumoylation machinery being also conserved (14). The biological roles of sumoylation include wing morphogenesis through the modification of Vestigial (Vg) (15), a factor essential for cell proliferation and differentiation of the central region of the wing imaginal disc (16). Here, we report that Drosophila Sall proteins can be sumoylated and that this modification alters their nuclear localization. Furthermore, we show that the sumoylation status of these proteins influences the repression of vein formation in intervein regions as well as the regulation of a target gene in vivo.
Strain w; P[UAS-2xEGFP]AH2;PZ-lwr (P[PZ]lwr05486cn1/CyO;ry506), a recessive lethal insertion in the 5′-untranslated region of lwr; lwr4-3 (y1w;lwr4-3P[neoFRT]40A/CyO,y+) (17), and strain PZ-smt3 (P[PZ]smt304493cn1/CyO;ry506), a recessive lethal insertion 5′ of smt3, were obtained in from the Bloomington Drosophila Stock Center. Other strains used were Df-5 (Df(2L)32FP-5/CyO) (18), sal445/CyO (7), and SalEPv (SalEPv-Gal4) (10). Rearing and crosses were carried out at 25 °C in standard media. Transgenic lines UAS-sal, UAS-salr, UAS-sal-IKDP, and UAS-salr-IKEA-IKVA were generated using standard methods. Wing preparation was done as described previously (19). Areas were calculated using ImageJ on images taken at the same magnification. The numbers of wings analyzed were as follows: wild-type (WT)5 = 89, Df-5/+ = 15, PZ-smt3/+ = 20, Df-5/PZ-smt3 = 22, PZ-lwr/+ = 23, Df-5/PZ-lwr = 11, SalEPv = 48, SalEPv> sal = 24, SalEPv>sal-IKDP = 69, SalEPv>salr = 47, and SalEPv>salr-IKEA-IKVA = 99.
Details for the construction and oligonucleotides are provided under supplemental “Experimental Procedures.” Mutants sal-IKDP, sal-IKHA, salr-IKVA, and salr-IKEA-IKVA were made using the QuikChange® II site-directed mutagenesis kit (Stratagene). For the N-terminal constitutively sumoylated fusion proteins, degenerated smt3 (GenBankTM accession number FN539078) was inserted into the SacI-AclII sites of sal or into the BsrGI-PvuI sites of salr to generate smt3-sal or smt3-salr, respectively.
HEK-293FT cells were transfected using the calcium phosphate method. Immunofluorescence was done as described previously (19). Imaginal disc immunohistochemistry was performed in third-instar wandering larvae following standard procedures. The antibodies used were guinea pig anti-Kni (20); rat anti-Sal (21); mouse monoclonal anti-B23 (Santa Cruz Biotechnology); and anti-mouse, anti-guinea pig, and anti-rat Alexa 568 (Molecular Probes). Nuclei were labeled with DAPI (Sigma) or TO-PRO-3® (Invitrogen). Confocal images were taken with a Leica DM IRE2 microscope using the ×40 or ×63 objectives, a resolution of 1024 × 1024 dpi, and a zoom of 1 or 4. Measurements of the Kni stripe were taken independently three times for each disc at the proximal part of the ventral compartment on images taken at the same magnification and settings using Adobe Photoshop. The length of the wing blade was taken at the anterior-posterior compartment border. The numbers of discs analyzed were as follows: WT = 9, SalEPv>GFP = 10, SalEPv>sal = 14, SalEPv>sal-IKDP = 6, SalEPv>salr = 5, and SalEPv>salr-IKEA-IKVA = 5.
Sumoylation motifs were identified using Sumoplot software. In vitro translation was carried out using TnT® T7 (wheat germ extract; Promega), adding [35S]methionine (Amersham Biosciences and Pierce). Translated proteins were incubated with an ATP-regenerating system, SUMO-1, Ubc9, and purified activating enzyme E1 (BIOMOL). Reactions were incubated at 30 °C for 2 h. Products were resolved by SDS-PAGE and exposed.
The sumoylation consensus motif ΨKXE (where Ψ is a hydrophobic amino acid, and X is any amino acid) (22, 23) is present in the Sal motifs IKEE (Lys641) and IKHE (Lys779), as well as in the Salr motifs IKED (Lys962) and IKVE (Lys1229) (Fig. 1A). These motifs are conserved in other Drosophila species analyzed (Fig. 1A). Sal motifs are located in the central part of the protein, between zinc finger domains 2 and 3, whereas Salr motifs are localized between zinc finger domains X and 5 (IKED) and in the C-terminal part of the protein (IKVE) (Fig. 1A). The presence of these conserved motifs suggested that the Drosophila Sall proteins could be modified by sumoylation. Accordingly, both Sal and Salr were modified in the presence of human SUMO-1 in vitro (Fig. 1, B and C). In the case of Sal, fragment 403–933, which contains the IKEE and IKHE motifs, was sumoylated (Fig. 1B). Whereas the mutation of IKHE to IKHA did not disrupt sumoylation, the additional mutation of IKEE to IKDP did. Sal fragment 403–516, which lacks any consensus motif, was not sumoylated (data not shown). In the context of the full-length protein, mutation of IKEE to IKDP reduced the modification drastically (Fig. 1B), suggesting that Sal sumoylation is mediated mainly through IKEE.
In the case of Salr, sumoylation assays of Salr fragments 803–1126 and 1096–1251 showed that mutations of IKED to IKEA and of IKVE to IKVA blocked sumoylation (Fig. 1C). Salr fragments 106–637 and 633–921, which did not contain consensus motifs, were not sumoylated (data not shown). In the context of the full-length protein, the mutation of IKVE to IKVA did not abolish sumoylation, whereas the additional mutation of IKED to IKEA drastically reduced the modification by SUMO-1 (Fig. 1C). In summary, our results suggest that Sal sumoylation occurs mainly through the IKEE motif and that Salr sumoylation occurs in both IKVE and IKED motifs.
We next analyzed whether sumoylation influences the subcellular localization of Sall proteins in vivo. In WT wing imaginal discs, endogenous Sal detected with specific antibodies (21) localized in the cell nuclei, exhibiting a diffuse distribution (Fig. 2A). However, in lwr4-3/+ discs, a hypomorph mutation for the Ubc9 homolog lesswright (lwr), Sal aggregated in discrete domains (Fig. 2B), indicating that sumoylation influences Sal localization.
To further analyze the effect of sumoylation on Sall localization, we compared the behavior of WT and mutant proteins in cultured cells in either the presence or absence of exogenous Smt3. Proteins mutated in the sumoylation sites were created. In all cases, aspartic or glutamic acids in the consensus sites were substituted, as they are the amino acids that are involved in the interaction with the Ubc9 conjugating enzyme and also so as not to interfere with other possible modifications of the lysines such as ubiquitylation. We also analyzed the localization of Smt3 N-terminal fusions to Sal and Salr, which represent a model for constitutive sumoylation. A synthetic sequence for smt3 was designed that substituted glycines 87 and 88 with alanines to block the processing of Smt3 by the Ulp1 protease, thus avoiding proteolysis or transference of the Smt3 moiety to other proteins.
Our results show that Sal localized in discrete nuclear domains (Fig. 2C). Interestingly, the number and size of Sal nuclear domains decreased in the presence of overexpressed Smt3 (Fig. 2E), similar to the fusion protein Smt3-Sal (Fig. 2F), suggesting that the observed change in Sal localization depends on the direct modification of Sal by Smt3. This effect probably does not reflect changes in the stability of Sal, as the half-life of the sumoylated form did not show significant differences with respect to the WT protein (see Fig. 4, A and B). Furthermore, the mutant Sal-IKDP, which localized in subnuclear domains similarly to WT Sal (Fig. 2G), did not change upon cotransfection with Smt3 (Fig. 2H). Altogether, these results indicate that the IKDP mutant is insensitive to Smt3 and that the observed change in localization of WT Sal depends on its direct modification through this motif. We observed a similar effect when human SALL1 fused to green fluorescent protein (GFP), which normally localized in nuclear domains (Fig. 2I) (24), was coexpressed with Smt3 (Fig. 2J), implying a conserved mechanism for Sall protein localization dependent on sumoylation.
Salr localized in the same subnuclear domains as Sal (Fig. 3A), suggesting a possible interaction between the two proteins. However, fluorescence resonance energy transfer and yeast two-hybrid assays failed to demonstrate their direct interaction (data not shown). In contrast to Sal, Salr formed large nuclear aggregates when coexpressed with Smt3 (Fig. 3B). The fusion protein Smt3-Salr mimicked this accumulation (Fig. 3C), suggesting that the change in localization was due to the direct modification of Salr by Smt3. The aggregates formed did not correspond to the nucleolus, as they did not colocalized with the nucleolar marker B23 (Fig. 3D). We cannot discard that constitutively sumoylated Salr might be subject to an increased turnover rate based on measurements of half-life and the appearance of possible cleavage products (Fig. 4, C and D). Similar to WT Salr, the mutants Salr-IKVA and Salr-IKEA-IKVA localized in nuclear domains (Fig. 3, E and G), indicating that these mutations do not affect the localization of the protein. Also similar to the WT protein, the single mutant Salr-IKVA formed large aggregates when coexpressed with Smt3 (Fig. 3F). However, the double mutant Salr-IKEA-IKVA was insensitive to the presence of Smt3 (Fig. 3H). Taken together, these results indicate that the localization of Sall proteins is influenced by their sumoylation status. These results are in agreement with our in vitro data, indicating that both sites can mediate the sumoylation of Salr. In addition, changes in nuclear sizes were seen when Sall proteins and Smt3 were overexpressed.
We analyzed the possible genetic interaction between the sumoylation machinery components and the Drosophila sall genes. Df-5 mutants carry a small genomic deficiency that removes sal and salr genes and produces embryonic lethality (18). The Df-5/+ heterozygous fly wings show a small delta in the distal part of vein LIV and a modest size reduction but no major disturbances in veins LII and LIII or in the LII-LIII intervein region (Fig. 5, C and E). Loss-of-function mutant lines PZ-smt3 and PZ-lwr did not exhibit a phenotype in heterozygosis (Fig. 5E and data not shown). However, the double heterozygous wings Df-5/PZ-smt3 showed ectopic vein LIII in 86% of the cases and a >10% reduction in wing size, similar to trans-heterozygous Df-5/PZ-lwr (Fig. 5, D and E; and data not shown). The repression of vein LIII in the LII-LIII intervein region might depend on both sal and salr or, alternatively, only on salr, as the double heterozygous for the single mutant sal445, an amorph allele of sal, and PZ-lwr did not show ectopic vein tissue (data not shown).
To gain more insight into this functional relationship, we used the SalEPv-Gal4 line to overexpress the Sall proteins in the central part of the wing from the LII provein to the LV provein (Fig. 5A) (10). Overexpression of Sal caused the formation of ectopic veins LII and LIII in the LII-LIII intervein region and a general wing size reduction (Fig. 6, A, E, and F), whereas overexpression of Salr caused a mild phenotype (Fig. 6, C, E, and F). These were similar to the phenotypes observed using the previously reported lines UAS-sal and UAS-salr (8), suggesting that the fluorescent tags did not dramatically affect the function of the proteins (data not shown). However, the mutations in the sumoylation sites produced dramatic changes in the activities of the respective Sall proteins. The mutation IKDP significantly inhibited the Sal overexpression phenotype (Fig. 6, B, E, and F). Conversely, the mutation IKEA-IKVA significantly augmented the Salr overexpression phenotype (Fig. 6, D–F). These results suggest that the sumoylation status of Sall proteins influences their capacity to regulate the formation of ectopic veins.
The formation of the LII extra vein might indicate a misregulation of kni (11). To test this possibility, we measured the width of the kni ventral expression domain in the LII provein in imaginal discs. The results showed that the kni expression domain was broader when Sal-IKDP or especially WT Sal was overexpressed compared with overexpression of GFP (Fig. 7, compare C and D with A). Furthermore, kni expression expanded when the mutant Salr-IKEA-IKVA, but not WT Salr, was overexpressed (Fig. 7, compare F with B and E). These results suggest that the sumoylation status of Sall proteins influences their capacity to regulate the expression of downstream genes.
The activity and the location of some transcription factors can be influenced by their post-translational modifications. For instance, murine Sall1 phosphorylation by protein kinase C disrupts its interaction with the histone deacetylase complex and its transcriptional ability (25). Here, we investigated how sumoylation might affect the activity of Drosophila Sall proteins in vivo and their localization in cultured cells. Our results show that both Sal and Salr are sumoylated and that their sumoylation status might influence their subnuclear localization and their activity in vivo.
Our results indicate that localization of Sall proteins depends on their sumoylation status. On one hand, we showed that the subnuclear localization of Sal changes in wing imaginal discs when sumoylation is compromised. On the other hand, Sal localization changes when Smt3 is overexpressed in cultured cells. Although HEK-293FT cells express SUMO (26), only a proportion of the overexpressed Sall factors might be sumoylated. This proportion might be augmented when Smt3 is overexpressed, making the change in localization visible. Therefore, our results in cultured cells are in accordance with the changes in Sal localization seen in imaginal discs: in WT discs, where Smt3 is expressed throughout the wing blade (27), Sal shows a more diffuse localization than in discs deficient for sumoylation (Fig. 2, A and B). Interestingly, our analysis in imaginal discs also showed that Sal expression is not abolished when sumoylation is impeded. In addition, Sal is still expressed when smt3 is down-regulated by RNA interference in imaginal discs (data not shown), despite that the function of signaling of Vg and Decapentaplegic, two known positive regulators of sal in the wing, could be affected by the lack of sumoylation (10, 15, 28).
The role of Sal and Salr in the formation of vein LII is not fully understood. Whereas the absence of both genes inhibits vein formation, the absence of only Sal promotes the formation of ectopic vein LII, suggesting a possible role for Salr as an vein activator (8). However, the overexpression of either protein using Gal4 lines that drive expression throughout the wing blade, such as 638-Gal4 and Nubbin-Gal4, inhibits vein LII formation through the inhibition of kni (8,–10). Using SalEPv-Gal4, we increased the expression of sall constructs in the central part of the wing, with this effect being more acute in the presumptive LII region (Fig. 7A). We could hypothesize that by overexpressing either the WT or sumoylation mutant forms of the proteins, we might alter the proportion of sumoylated versus non-sumoylated forms in vivo, with this affecting the regulation of kni. In the case of vein LIII, our analysis of trans-heterozygous flies suggests that Sall proteins and Smt3 collaborate in the repression of vein LIII in the LII-LIII intervein region. In addition, the overexpression of the sumoylation mutant Salr-IKEA-IKVA promoted the formation of ectopic vein LIII, suggesting that sumoylated Salr is necessary for LIII repression in the LII-LIII intervein region.
One conclusion that emerges from our results is that sumoylation seems to affect the localization and vein repressor capacities of Sal and Salr in opposite ways: whereas sumoylation modifies the subnuclear localization of Sal toward a more diffuse pattern in the nucleus, Salr sumoylation promotes the formation of large subnuclear aggregates. Sal and Salr exhibit transcriptional repressor capacity in cultured cells,6 and it could be possible that these dramatic changes in the localization of Sall proteins could alter their transcriptional repressor capacity in vivo. In the adult wing, overexpression of the sumoylation mutant of Sal produces a milder phenotype than the overexpression of WT Sal, whereas the sumoylation mutant of Salr increases the formation of ectopic veins in comparison with WT Salr. In addition, our results on the capacity to regulate the expression of kni fit well with the phenotypic consequences observed in adult wings: the overexpression of the mutant form of Sal affects kni expression in a milder way than the overexpression of the WT form. In contrast, the overexpression of mutant Salr increases the kni expression territory, suggesting that the sumoylated form of Salr regulates the repression of kni in the intervein region more effectively compared with the non-sumoylated form. Thus, sumoylation seems to affect Sal and Salr in opposite ways. As both proteins are in many cases coexpressed and might participate in the same biological processes, their regulation by sumoylation opens an interesting venue in which the same modification produces opposite effects, contributing to the fine-tuning of the role of Sall proteins in gene regulation.
In summary, the sumoylation of Sall proteins affects their nuclear localization. Sumoylation is necessary for repressing vein LIII by Sall in the LII-LIII intervein region and for attaining the wing final size, suggesting that these two functions are sensitive to the doses of sumoylated Sall proteins.
We thank Dr. J. Kohlhase for GFP-SALL1, Dr. J. Jaeger for anti-Kni antibody, Dr. R. T. Hay for advice on sumoylation, Dr. R. Schuh for constructs, the Bloomington Drosophila Stock Center, and Dr. J. F. de Celis for critical reading of the manuscript.
*This work was supported by Spanish Ministerio de Ciencia e Innovación Grants BFU2008-01884 and CSD2007-008-25120, Departments of Education and Industry of the Basque Government Grant PI2009-16 and Etortek Research Programs 2008/2009, and the Bizkaia County.
The on-line version of this article (available at http://www.jbc.org) contains supplemental “Experimental Procedures” and additional references.
6J. Sánchez, A. Talamillo, C. Pérez, and R. Barrio, unpublished data.
5The abbreviations used are: