Biochemical Journal

Research article

Drosophila Sal and Salr are transcriptional repressors

Jonatan Sánchez, Ana Talamillo, Monika González, Luis Sánchez-Pulido, Silvia Jiménez, Lucia Pirone, James D. Sutherland, Rosa Barrio

Abstract

The SALL (Spalt-like) family of zinc-finger transcription factors is conserved in metazoans. In Drosophila Sal (Spalt) and Salr (Spalt-related) control the expression of genes involved in wing and central nervous system development, including cell adhesion and cytoskeletal proteins. In humans, SALL mutations associate with congenital disorders such as the Townes–Brocks and Okihiro syndromes. Human and Drosophila SALL proteins are modified by SUMO (small ubiquitin-related modifier), which influences their subnuclear localization. In the present study, we have analysed the transcriptional activity of Drosophila Sall proteins in cultured cells. We show that both Sal and Salr act as transcriptional repressors in Drosophila cells where they repress transcription through an AT-rich sequence. Furthermore, using the UAS/Gal4 heterologous system, Drosophila Sal and Salr repress transcription in human cells. Under our experimental conditions, only in the case of Salr is the repression activity dependent on the HDAC (histone deacetylase) complex. This complex might interact with the C-terminal zinc fingers of Salr. We describe the differential subcellular localizations of Sal and Salr fragments and identify their repression domains. Surprisingly, both repressors also contain transcription activation domains. In addition, under our experimental conditions SUMOylation has differential effects on Sal and Salr repressor activity. Phylogenetic comparison between nematodes, insects and vertebrates identifies conserved peptide sequences that are presumably critical for SALL protein function.

  • Drosophila
  • histone deacetylase (HDAC)
  • small ubiquitin-related modifier (SUMO)
  • Spalt-like (SALL)
  • transcriptional repression

INTRODUCTION

SALL (Spalt-like) proteins are zinc-finger transcription factors that are conserved from Caenorhabditis elegans to mammals. These proteins are involved in different biological processes such as organogenesis, carcinogenesis and the maintenance of pluripotency in embryonic stem cells [1,2]. The first SALL genes to be described were sal (spalt) and salr (spalt-related) in Drosophila [3,4]. These two genes participate in numerous developmental processes, including wing growth and vein formation under the control of the TGFβ (transforming growth factor β) pathway via the target genes iro-C and kni-C in the imaginal discs [57]. Sal and Salr are necessary for the determination of neuronal fate in the peripheral nervous system [8,9], and regulate the expression of cell adhesion and cytoskeletal proteins in the central nervous system and in imaginal discs [10,11]. However, the direct interaction between Sall proteins and the target DNA sequences has been reported only for Salr, which binds to an AT-rich region (AATTATGAAATGCCA) in the promoter of the chorion gene s15 [4,12].

In vertebrates, there are four homologous genes (SALL14). The importance of this family of proteins to human health is linked to mutations that are associated with different inherited diseases, such as TBS (Townes–Brocks syndrome) or OS (Okihiro syndrome) caused by mutations in SALL1 and SALL4 respectively. TBS is an autosomal dominant syndrome characterized by imperforate anus, limb malformations, dysplastic ears and sensorineural hearing loss [13,14]. OS (also known as Duane-radial ray syndrome) is an autosomal dominant disorder also characterized by limb malformations, together with ocular and renal anomalies [15,16]. The limb and neuronal abnormalities in those patients are reminiscent of the limb and nervous system defects in Drosophila sall mutants, suggesting that SALL proteins might be involved in the regulation of conserved developmental processes in metazoans [17].

In mammals, SALL1 proteins have been defined as transcriptional repressors. In humans and mice, two different domains mediate SALL1 repression ability: a poly(Gln) (polyglutamine) region located in the N-terminal part of the protein, and the central region where the ZF (zinc finger) pairs 2 and 3 are located [1823]. In mice, the HDAC (histone deacetylase) complex mediates repression by the SALL1 N-terminal region via interaction with a 12 amino acid sequence [20]. The human and murine regions involved in repression localize to subnuclear domains that coincide with heterochromatin, binding there to an AT-rich sequence [1820,24,25]. This heterochromatin interaction might represent an additional mechanism of repression beyond the HDAC interaction. However, not all of the SALL proteins have been identified as transcriptional repressors. In humans, SALL2 activates the transcription of the cyclin-CDK (cyclin-dependent kinase) inhibitor p21, important for G1 checkpoint control [26], and SALL4 activates Bmi-1 through a specific enhancer [27]. In addition, the maintenance of pluripotency in embryonic mouse stem cells is dependent on Sall4 activation of Nanog, Pou5f1 and Sall4 itself [28,29].

SALL proteins are modified by coupling to SUMO (small ubiquitin-related modifier) 1, a protein modification linked to processes including subnuclear localization and transcriptional activity [30]. In Drosophila, Sal and Salr are SUMOylated in vitro and interact genetically with the SUMOylation pathway during wing development. SUMOylation leads to subnuclear localization changes in Sall proteins, with phenotypic consequences for wing growth and vein formation [7]. In the present study, we analyse the functional domains of the Drosophila Sall proteins, their localization and their transcriptional activity in relation to HDACs and to SUMOylation. We show that both proteins are transcriptional repressors and that Salr (but not Sal) repression is mediated by HDACs. In addition, SUMOylation differentially affects Salr and Sal transcriptional activity. Furthermore, we report that both these repressor proteins also contain transcriptional activation domains. Finally, the evolutionary conservation of these domains among vertebrates and invertebrates is discussed.

EXPERIMENTAL

DNA constructs

NheI-ApaI or NotI-ApaI fragments containing FL (full-length) sal or salr were cloned in the XbaI or NotI-ApaI sites of pAC5.1-V5-His-A (Invitrogen) to generate pAc-Sal or pAc-Salr respectively. pAC-RLuc contains the RLuc (Renilla luciferase) gene cloned between the EcoRV and HpaI sites of pAC5.1-V5-His-A. pAC-SalrBE-RLuc contains five copies of the sequence AATTATGAAATGCCA [4] in the BlpI site of pAC-RLuc. pAC-SalrBE*-RLuc contains five copies of the mutated sequence AAccgaattccGCCA cloned in the same site (lower case letters indicate the mutated bases). The pAC5.1-CFLuc-V5His vector was purchased from Addgene [31].

The pECFP-His-Myc-sal, pEYFP-His-Fluo-salr, pGal4-DBD (DNA-binding domain), pGal4-sal, pGal4-salr, pB-CFP-sal-IKDP and pB-YFP-salr-IKEA-IKVA constructs have been described previously [7]. For the constructs pGal4-S1 to -S8, pGal4-S11, pGal4-R1 to -R8, and pGal4-R9 to -R12, sal or salr fragments were amplified by PCR using specific oligonucleotides that contained NheI and ApaI restriction sites, cloned into the pST-BlueI vector (Stratagene) and sequenced. The fragments digested with NheI-ApaI were cloned into the same sites of pGal4-DBD or pGal4-salr. Fragments S9, S10 and S12 were digested from pGal4-sal and cloned into the same sites of pGal4-DBD using KpnI-EcoRV, KpnI-EcoRI or KpnI-NotI restriction enzymes respectively.

pGal4-sal-IKDP was generated by cloning the EcoRI fragment from pB-CFP-sal-IKDP into pGal4-sal. pGal4-salr-IKEA-IKVA was generated by cloning the fragment XbaI-ApaI from pB-YFP-salr-IKEA-IKVA into the same sites of pGal4-salr.

The luciferase reporter pGal4tkLUC was a gift from Dr J. Ericsson (Ludwig Institute for Cancer Research, Biomedical Center, Uppsala, Sweden) [32]. pRL-SV40 (Promega) was used to normalize the activity in the transcriptional activity assays. Oligonucleotide sequences are given in the Supplementary Table S1 (at http://www.BiochemJ.org/bj/438/bj4380437add.htm).

Cell culture and immunocytochemistry

Drosophila S2R+ [33] and Kc167 cells [34] were obtained from the Drosophila Genomics Resource Center. Cells were cultured at 25 °C in Drosophila Schneider's medium (Invitrogen) supplemented with 10% FBS (fetal bovine serum; Gibco) and 1% penicillin/streptomycin (Gibco). Transfections were performed using Effectene (Qiagen) in 96-well plates.

HEK (human embryonic kidney) 293FT cells (Invitrogen) were maintained at 37 °C in DMEM (Dulbecco's modified Eagle's medium; Gibco) supplemented with 10% FBS, 1% penicillin/streptomycin and 0.2% primocin (InvivoGen). For transfections, the calcium phosphate method was used in 24-well plates.

For immunofluorescence, cells were fixed 24 h after transfection and treated as described previously [7] using an anti-Gal4 antibody (1:50 dilution; Santa Cruz Biotechnology) or DAPI (4′,6-diamidino-2-phenylindole, 1:2000 dilution; Sigma). Confocal images were taken using a Leica TCS-SP2 DM-IRE2 microscope using the ×63 objective at a resolution of 1024×1024 d.p.i. and different zoom values.

Transcriptional activity assays

Drosophila S2R+ cells were seeded in 96-well plates and transfected with 25 ng of pAC-SalrBE-RLuc or pAC-SalrBE*-RLuc, either alone or co-transfected with 12.5 ng of pAc-Sal or pAc-Salr, and 12.5 ng of pAC5.1-CFLuc-V5His was used for normalization. HEK-293FT cells were seeded in 24-well plates and transfected with 500 ng of the pGal4-DBD constructs or 500 ng of pGal4tkLUC. For normalization, 1 ng of pRL-SV40 was added to each well. To test the effect of HDACs, at 24 h after transfection cells were treated with either TSA (trichostatin A, 90 ng/ml; Calbiochem), EX527 (10 mM; Cayman Chemical), NAM (nicotinamide, 10 mM; Sigma) or DMSO (Sigma). Transcriptional activity was measured 24 or 48 h after transfection using the Dual-Luciferase Reporter Assay System (Promega) for mammalian cells, or the Dual-Glo Luciferase Assay System (Promega) for Drosophila cells, following the manufacturer's instructions. Luminescence was measured in a microplate luminometer (Veritas). Results are given as means+S.D., unless otherwise stated. Differences between groups were calculated using Student's t test.

Western blot analysis

HEK-293FT cells were transfected in 24-well plates as described above. At 24 h after transfection, cells were collected in 100 ml of Laemmli buffer, the protein content was measured, and equivalent amounts were loaded on to 4–15% gradient polyacrylamide gels (Bio-Rad). Transfer on to PVDF membranes was done for 3 or 4 min using the iBlot system (Invitrogen). Anti-Gal4 (Santa Cruz Biotechnology) antibodies were used at a dilution of 1:1000. The expected molecular mass and number of amino acids for each fragment are given in Supplementary Table S2 at http://www.BiochemJ.org/bj/438/bj4380437add.htm.

Sequences comparison and phylogenetic analysis

Sequences search and comparison was performed using the BLAST tool (http://blast.ncbi.nlm.nih.gov/Blast.cgi). Alignments and unrooted N-J phylogeny trees were performed using T-Coffee (http://www.tcoffee.org). Database codes for the sequences used in the analysis are provided in the respective Figure legends.

RESULTS

Localization and transcriptional activity of Sall proteins in Drosophila cells

We analysed the localization and transcriptional capacity of Sall proteins in Drosophila cell culture. Sall proteins labelled with fluorescent tags were expressed in Drosophila Kc167 cells. As in mammalian cells [7], both Sall proteins localized to subnuclear domains (Figures 1A and 1B), with the Sal domains being larger in size and fewer in number than the Salr domains.

Figure 1 Sal and Salr are transcriptional repressors in Drosophila cells

(A and B) Confocal micrographs showing in green the localization of CFP–Sal (A) or YFP–Salr (B) in subnuclear domains of Drosophila Kc167 cells. The DAPI nuclear marker is shown in magenta. Grey pictures show single green channels (A' and B'). Scale bars represent 4 μm. (C and D) Graphic representation of RLuc activity of pAC-SalrBE-RLuc (C) or pAC-SalrBE*-RLuc (D) in the absence (−) or presence (+) of vectors expressing Sal or Salr. Values are the average of three independent experiments performed in quadruplicates. Values are means+S.D. Asterisks indicate statistically significant differences (P≤0.005).

To test the transcriptional capacity of Sall proteins in Drosophila cells, we used a Salr downstream region [4]. Salr binds to an AT-rich sequence, AATTATGAAATGCCA, which we refer to in the present study as SalrBE (Salr-binding element). The mutated SalrBE* sequence (AAccgaattccGCCA) blocks Salr binding [4]. Each of these sequences was placed between the strong actin 5C promoter and the TATA box, upstream of the RLuc reporter gene. Plasmids were transfected and luciferase assays were performed in S2R+ Drosophila cells. The pAC-SalrBE*-RLuc construct gave lower RLuc activity compared with the pAC-RLuc or pAC-SalrBE-RLuc constructs (results not shown). We found that the RLuc activity from the pAC-RLuc construct did not change in the presence of Sal or Salr (results not shown). In contrast, both pAC-Sal and pAC-Salr repressed RLuc activity from pAC-SalrBE-RLuc by more than 80% (Figure 1C). However, the activity of the pAC-SalrBE*-RLuc construct, carrying the mutated binding element, was not significantly reduced in the presence of either Sal or Salr (Figure 1D). These results confirm that the Sal and Salr repressors both act through the AT-rich SalrBE sequence.

Localization and transcriptional activity of Sall proteins in human cells

Drosophila S2R+ and Kc167 cells are poorly adherent, which makes handling them more difficult. In addition, the small size of these cells is not ideal for determining subcellular localizations. Therefore localization and functional assays were made in the HEK-293FT kidney cell line. SALL1, SALL3 and SALL4 genes are expressed in the HEK-293FT cell line [35] (http://www.proteinatlas.org), SALL proteins having an important role in kidney development [36]. To analyse the transcriptional activity and localization of the different domains of the Drosophila Sall proteins, we used a bipartite UAS/Gal4 heterologous transcriptional reporter assay. We tested Gal4-DBD constructs fused to Sal or Salr, and co-transfected these into HEK-293FT cells together with a UAS-luciferase reporter. Using an anti-Gal4 antibody, we detected the localization of Sall proteins in subnuclear domains (Figures 2A and 2B), as was previously reported for CFP (cyan fluorescent protein)–Sal and YFP (yellow fluorescent protein)–Salr fusion proteins [7], indicating that the Gal4 fusion did not alter the localization of the proteins.

Figure 2 Drosophila Sal and Salr are transcriptional repressors in human cells

(A and B) Confocal micrographs showing in green the localization of Gal4–Sal (A) or Gal4–Salr (B) in subnuclear domains of HEK-293FT cells. The DAPI nuclear marker is shown in magenta. Grey pictures show single green channels (A' and B'). Scale bars represent 4 μm. (C) Graphic representation of firefly luciferase activity of pGal4tkLUC alone (Gal4), Gal4-DBD fused to FL Sal (Gal4-Sal) or FL Salr (Gal4-Salr), either in the presence of DMSO (black bars), TSA (dark grey bars), NAM (white bars) or EX527 (light grey bars). Values are the average of three independent transfections performed in triplicates. Values are means+S.D. Asterisks indicate statistically significant differences between Gal4-DBD and Gal4-Sal or Gal4-Salr (black asterisks, P≤10−6) or between Gal4-Salr with or without inhibitor treatments (white asterisks, P≤0.005).

Luciferase assays showed that both Sall proteins gave strong repression, the activity of Gal4–Sal and Gal4–Salr being approximately 20% the activity of Gal4-DBD alone (Figure 2C). These results indicated that Drosophila Sall proteins repress transcription in mammalian cells. To test whether the transcriptional activity of Sall proteins was mediated by HDACs, cells transfected with pGal4-sal or pGal4-salr were treated with the HDAC inhibitor TSA. Only Salr repression activity seems to be mediated by HDAC, the cells treated with TSA showing statistically significant differences from DMSO-treated cells. Cells transfected with Sal did not show statistically significant differences between control and TSA-treated cells, suggesting that Sal repression might be mediated by a different mechanism. TSA inhibits Class I and Class II, the so-called ‘classical’ HDACs. However, the Class III HDACs, composed of NAD+-dependent enzymes or sirtuins, are not affected by TSA [37,38]. In order to test whether the repression exerted by Sall proteins is mediated by sirtuins, we used two specific inhibitors: NAM, a general sirtuin inhibitor, and EX527, a specific inhibitor of Sirt1. NAM and EX527 released the repression exerted by Salr in a similar way to TSA. The activities shown by Salr in cells treated with TSA, NAM or EX527 are similar, but show statistically significant differences to non-treated cells (P< 0.005). Double treatments, with TSA and NAM, or with TSA and EX527, did not show an additive effect (results not shown). In contrast with Salr, these drug treatments gave no significant differences in Sal repression compared with untreated cells. Therefore these results suggest that in these experimental settings Sal and Salr might repress transcription through different mechanisms.

Differential distribution of ZF pairs in Sall proteins across evolution

The results described above show that, surprisingly, Sal and Salr might mediate transcription through different mechanisms, either HDAC-dependent or -independent. This might depend on the differential distribution of the ZFs present in Sall proteins (Figure 3). Specific ZFs at a given position are more similar to each other in diverse species than to ZFs at other positions. For example, ZF2 from human SALL1 is more related to ZF2 from Drosophila Sal than to ZF3, 4 or 5 from human SALL1, indicating that the distribution of ZF pairs typical of the SALL proteins occurred before the divergence of insects and mammals. ZF pairs ZF2 and 3 are present in all of the Sall genes from the insect and mammalian species analysed, the presence of ZF4 or 5 being variable (Figure 2, and Supplementary Figures S1–S5 at http://www.BiochemJ.org/bj/438/bj4380437add.htm). Human and murine SALL1 and SALL3 contain ZF2, 3, 4 and 5, whereas SALL4 does not contain ZF4. In the non-Drosophilid insects analysed we found a unique sall gene containing ZF2–5, the exception being the mosquito species where a unique sall gene lacking ZF5 was found. By contrast, all of the Drosophilid species analysed contain two sall genes in their genomes, one of them encoding a protein equivalent to Sal (ZF2, 3 and 4) and the other to Salr (ZF2, 3, X and 5), suggesting that sal/salr duplication occurred after the divergence of Drosophilidae from other dipterans. The presence of ZF4 in Drosophila Sal and ZF5 in Salr might be related to the differential repression mechanisms of these two Sall proteins.

Figure 3 ZF distribution in Sall proteins is conserved throughout evolution

Upper panel: schematic representation of the ZFs and poly(Gln) domains in SALL proteins of the indicated species belonging to the phyla specified on the left-hand side. Bottom panel: unrooted N-J tree analysis of the ZF domains of vertebrate and invertebrate SALL proteins, shown in Supplementary Figure S1 (at http://www.BiochemJ.org/bj/438/bj4380437add.htm). Species analysed and accession numbers of the sequences were: Bruma, Brugia malayi, A8P9M2; Caeel, Caenorhabditis elegans, Q17396; Drome, Drosophila melanogaster, P39770 (Sal) and P91639 (Salr); Hs, Homo sapiens, Q9NSC2 (SALL1), Q9Y467 (SALL2), Q9BXA9 (SALL3) and Q9UJQ4 (SALL4); Pripa, Pristionchus pacificus, Q27ZK0; Sako, Saccoglossus kowalevskii, XP_002738555; Stpur, Strongylocentrotus purpuratus, XP_781376.

Analysis of Sal transcriptional domains in human cells

Prior to protein domain analysis, we compared the D. melanogaster Sal and Salr sequences with that of other insect species to identify conserved peptide sequences (Supplementary Figures S2–S5). This analysis revealed conserved blocks that were used as a guide to generate different fragments of the proteins (Figures 4A and 5A). These fragments were cloned in-frame with Gal4-DBD and a series of transient transfections were performed to analyse their localization and transcriptional capacity. All of the constructs were expressed in HEK-293FT cells and protein fragments showed the expected molecular masses (Figures 4C and 5C).

Figure 4 Analysis of Sal repression and activation domains in human cells

(A) Schematic representation of the S1–S12 fragments used in the analysis. Ovals represent ZFs and gridded squares represent the poly(Gln) domain. (B) Graphic representation of firefly luciferase activity of pGal4tkLUC alone (Gal4), Gal4 fused to FL Sal (FL-Sal) or Gal4-fused Sal fragments (S3–S6, S8, S11 and S12). Values are the average of three independent transfections performed in triplicates. Values are means+S.D. Asterisks indicate statistically significant differences with respect to FL-Sal (P≤10−5). (C) Western blot analysis using anti-Gal4 antibodies showing the expression of the constructs used in the luciferase assays. Molecular mass markers in kDa are indicated to the right-hand side of each panel. The expected size for each construct is indicated in Supplementary Table S2 (at http://www.BiochemJ.org/bj/438/bj4380437add.htm). (DO) Confocal micrographs showing in green the localization of the indicated Sal fragments in HEK-293FT cells. The DAPI nuclear marker is shown in magenta. (D'–O') Black and white panels show single green channels. The scale bar represents 4 μm. All pictures were taken at the same magnification.

Figure 5 Analysis of Salr repression and activation domains in human cells

(A) Schematic representation of the R1–R12 fragments used in the analysis. Ovals represent ZFs and gridded squares represent the poly(Gln) domain. (B) Graphic representation of firefly luciferase activity of pGal4tkLUC alone (Gal4), Gal4 fused to FL Salr (FL-Salr) or Gal4-fused Salr fragments (R3, R5–R7, R11 and R12), either in the presence of DMSO (black bars) or TSA (grey bars). Values are the average of three independent transfections performed in triplicate. Values are means+S.D. Black asterisks indicate statistically significant differences respect to FL-Salr (P≤10−4). White asterisks indicate statistically significant differences of each fragment in the presence or absence of TSA (P≤10−3). (C) Western blot analysis using anti-Gal4 antibodies showing the expression of the constructs used in the luciferase assays. Molecular mass markers in kDa are indicated to the right-hand side of each panel. The expected size for each construct is indicated in Supplementary Table S2 (at http://www.BiochemJ.org/bj/438/bj4380437add.htm). (DO) Confocal micrographs showing in green the localization of the indicated Salr fragments in HEK-293FT cells. The DAPI nuclear marker is shown in magenta. (D'–O') Black and white panels show single green channels. The scale bar represents 4 μm. All pictures were taken at the same magnification.

The fragments showed different patterns of localization, either cytoplasmic, nuclear or both. In the case of Sal (Figure 4), S1 was mainly localized in the cytoplasm (Figure 4D), whereas S2, S7, S9 and S10 were localized in cytoplasm and nucleus (Figures 4E, 4J, 4L and 4M). Fragment S2, which contains the poly(Gln) sequence, and fragment S7 showed a punctate pattern in the cytoplasm. All of the fragments containing ZFs, in addition to fragments S3 and S5, were localized in the nucleus (Figures 4F–4I, 4K, 4N and 4O). The nuclear fragments S3, S4, S5, S6, S8 and S11 showed a different distribution to the FL protein. Fragment S12 showed co-localization with the FL CFP–Sal fusion protein, suggesting that ZF4 is not necessary for Sal subnuclear localization (Figure 6A).

Figure 6 S12 and R12 fragments co-localize with the FL proteins

(A and B) Confocal micrographs showing co-localization of CFP–Sal (A) or YFP–Salr (B) shown in green, with the S12 or R12 fragments shown in magenta respectively. Black and white panels show single green (A' and B') or magenta pictures (A” and B”). The scale bar represents 4 μm. All pictures were taken at the same magnification.

We tested the transcriptional activity of the nuclear fragments of Sal (Figure 4B). Surprisingly, fragments S3, S4, S5 and S11 showed strong transcriptional activation capacity. Although the S5 fragment was expressed at lower levels than others (Figure 4C), this fragment showed the highest activation capacity, having 5-fold more activity than Gal4-DBD alone. Interestingly, cells overexpressing S5 were smaller than cells overexpressing other constructs, indicating an interaction of S5 with cell size. Fragments S6, S8 and S12 showed a repression capacity similar to the FL protein. This suggests that Sal repression domains might be related to ZF3 and ZF4, and Sal activation is located between the poly(Gln) and ZF3 domains.

Analysis of Salr transcriptional domains in human cells

In the case of Salr, only the regions that contained ZFs were localized in the nucleus (Figures 5F, 5H–5J, 5N and 5O), R3 showing some cytoplasmic location (Figure 5F), whereas fragments R4, R8, R9 and R10 were localized mainly in the cytoplasm (Figures 5G, 5K–5M). Fragments R1 and R2 were localized in the nucleus and cytoplasm (Figures 5D and 5E). Nuclear fragments exhibited different patterns. Fragments R5, R6 and R7 showed a diffused expression pattern in the nucleus (Figures 5H–5J), R11 was localized in the nuclear periphery (Figure 5N), and R12 was localized in a similar way to FL Salr (Figure 5O). Indeed, R12 co-localized with the fusion protein YFP–Salr (Figure 6B), suggesting that ZF5 is not necessary for Salr localization in subnuclear domains.

We analysed the transcriptional activity of the nuclear fragments R3, R5, R6, R7, R11 and R12 (Figure 5B). The ZF2-containing fragment R3 of Salr showed significant transcriptional activation capacity (Figure 5B), when compared with Gal4–Salr or Gal4-DBD alone. Fragments R5 and R11 showed some repression ability, although less than FL Gal4–Salr, whereas R6, R7 and R12 did not show statistically significant differences with respect to the FL protein (Figure 5B). These results suggest that part of the Salr repression capacity resides in fragments R6 and R7, which contain ZFX and ZF5 respectively. In addition, R12, which lacks ZF5, showed a similar repression activity to FL Salr, suggesting that, although ZF5 shows repression activity by itself, it is not necessary for the total repression mediated by Salr.

We have shown above that Salr might act through HDACs. To test which domain could be responsible for this repression, we analysed the transcriptional activity of the protein fragments in the presence of TSA. Our analysis showed that R6, R7 and R12 presented statistically significant differences in the presence of TSA (Figure 5B), suggesting that the HDAC-dependent repression observed for Salr might be mediated through these fragments. Interestingly, fragments R6 and R7 contain ZFX and ZF5, the two ZFs characteristic of Salr. This suggests that the Salr interaction with HDACs might depend on fingers ZFX and ZF5, and might explain the different mechanisms observed between Sal and Salr. On the other hand, our results suggest that the lack of ZF5 does not affect the transcriptional repression by Salr or its dependence on HDACs, as there are no statistically significant differences between R12 and Salr (Figure 5B).

Effect of SUMOylation on the transcriptional activity of Sall proteins

SUMOylation might modify the transcriptional activity of Sall proteins. We analysed the transcriptional capacity of the SUMOylation-insensitive mutant forms of Sal and Salr [7]. These forms localize in a similar way to the WT (wild-type) proteins, but they are not modified by SUMOylation in vitro and they behave differently when overexpressed in vivo. The repression activity of Sal-IKDP was 2-fold higher than that of the WT protein (Figure 7). Surprisingly, in the case of Salr, the repression capacity of Salr-IKEA-IKVA was lower than that of the WT protein (Figure 7). These results suggest that, in our experimental settings, SUMOylation modifies the repression capacity of Sal and Salr in opposite ways.

Figure 7 SUMOylation influences the transcriptional activity of Drosophila Sall proteins in human cells

Graphical representation of firefly luciferase activity of pGal4tkLUC alone (Gal4), Gal4 fused to FL Sal (Gal4-Sal), SUMOylation-insensitive Sal mutant (Gal4-Sal-IKDP), FL Salr (Gal4-Salr) or SUMOylation-insensitive Salr mutant (Gal4-Salr-IKEA-IKVA). Values are the average of triplicate transfections. Values are means+S.D. Asterisks indicate statistically significant differences (P< 0.005).

DISCUSSION

In the present study we have shown that Sal and Salr act as transcriptional repressors in two different experimental settings. In Drosophila cells, Sal and Salr can repress transcription through an AT-rich sequence. In mammalian cells they act as repressors in a UAS/Gal4 heterologous system. Furthermore, we have shown that both proteins contain activation and repression domains, and that in our experimental settings they behave differently in relation to HDACs or SUMOylation. Despite the limitations inherent in the use of a heterologous system, our results have provided valuable information on Sal and Salr transcriptional activity, which is discussed below.

Comparison of insects and vertebrate Drosophila proteins

The Drosophila Sal and Salr proteins show transcriptional repressor activity in both Drosophila and mammalian cells. In Drosophila cells, both Sall proteins act through an AT-rich sequence present in the upstream regulatory region of the chorion gene s15. This is similar to mouse Sall1, which also represses transcription through an AT-rich sequence [25]. In vertebrates, there are two distinct domains of repression: one on the N-terminal part of the proteins, which is not conserved in Drosophila, and one formed by the central region ZF3 and ZF2. Sal and Salr ZF3 regions show repressive capacity in a manner similar to that observed in vertebrates. Interestingly, the ZF3 region of Salr that binds to the AT-rich sequence is by itself capable of causing an overexpression phenotype similar to that of the FL protein [6].

In contrast with Sall1, mouse Sall4 activates transcription through CT- or CG-rich sequences [28]. Surprisingly, Sal and Salr are also able to activate transcription through a region located between poly(Gln) and ZF3. Repression and activation capacities could reside in different SALL proteins in mammals, whereas both capacities could exist in the same molecule in Drosophila. With all SALL proteins, binding to specific sequences and cell-specific protein partners could result in different SALL protein conformations, thus exposing either repression or activation domains. Although no direct transcriptional activation of any target gene by Drosophila Sall proteins has been reported, this could exist in certain cell types or developmental processes.

Regions S3 in Sal and R3 in Salr possess modest, but reproducible, activation capacity. These regions contain a box rich in non-polar amino acids conserved among the insect sequences, suggesting a conserved function (Supplementary Figure S3, boxed in orange). Aside from the ZFs and the poly(Gln) region, two other conserved boxes among insects and mammals were identified. One of them is included in regions S6 and R5, both of them exhibiting transcriptional repression activity (Supplementary Figures S1 and S4, boxed in red). The other box is located at the C-terminal end of ZF5, it is rich in charged amino acids (Supplementary Figures S1 and S5, boxed in blue) and is contained in the R7 fragment. The R7 fragment possesses repression capacity, although it has been shown to be dispensable for the general repression of Salr and also for its subnuclear localization. Based on their high degree of conservation, these two novel motifs could be potentially relevant for the function of SALL proteins among species.

SUMOylation and the transcriptional activity of Sall proteins

In Drosophila, SUMOylation increases the transcriptional activity of Vg (Vestigial), Dorsal and p53 [3942], while reducing the transcriptional activity of Stat92E, GCMa and SoxN [43]. One conclusion that emerges from our results is that SUMOylation seems to affect the transcriptional capacity of Sal and Salr in opposite ways in our experimental settings. In the case of Sal, a mutation that prevents SUMOylation enhanced its capacity to repress transcription. In contrast, SUMOylation-site mutations reduce the Salr capacity to mediate transcriptional repression. This has been described as well for some mammalian transcription factors such as Elk-1 or SoxN, where SUMOylation increases their repression activity [4346].

The same ‘opposite’ effect of SUMOylation on the transcriptional activity of Sall proteins is in agreement with in vivo studies concerning Drosophila wing development [7]. We reported that the induction of ectopic wing vein formation by Sal overexpression is partially suppressed in Sal-IKDP mutants, whereas the capacity of Salr to induce veins increases when the SUMOylation mutant form is overexpressed. According to the genetic analysis, Sall proteins and SUMO collaborate in the repression of veins in intervein regions. There the lack of SUMOylation of Salr could diminish its repression capacity and cause the generation of ectopic veins. Thus the overexpression of the SUMOylation mutant Salr-IKEA-IKVA promoted the formation of ectopic vein LIII. We hypothesize that SUMOylated Salr is responsible for LIII repression in the intervein region based on its transcriptional repression capacity.

It has been proposed that SUMOylation has a general role in transcriptional repression, as it is necessary for the interaction of transcription factors with the HDAC components. Such is the case for Elk-1, where mutations in the SUMOylation domain block its interaction with HDACs [4446]. The Salr SUMOylation sequence IKED is located in the ZF5-containing fragment R7, which we have demonstrated to contain TSA-responsive repressor activity. Therefore we hypothesize that mutation in the IKED sequence would interfere with its putative interaction with HDAC components.

In summary, the results of the present study provide insight into the role of Drosophila Sall proteins. We have shown that both proteins act as transcriptional repressors in Drosophila and in mammalian cells, and that both contain repression and activation domains. In addition, we have shown that Sal and Salr behave differently in relation to HDACs and that SUMOylation affects their transcriptional capacity in opposite ways. Further work will be necessary to understand the role of the conserved boxes among insects and vertebrates and the relationship between the SUMOylation status of these proteins and their transcriptional capacity.

AUTHOR CONTRIBUTION

Jonatan Sánchez, Ana Talamillo, James Sutherland and Rosa Barrio conceived and planned the experiments; Jonatan Sánchez, Ana Talamillo, Monika González, Silvia Jiménez, Lucia Pirone and Rosa Barrio performed the experiments; Luis Sánchez-Pulido and Rosa Barrio performed the sequence analysis; Jonatan Sánchez, Monika González, James Sutherland and Rosa Barrio analysed the results; and Jonatan Sánchez and Rosa Barrio wrote the paper.

FUNDING

This work was supported by the Spanish Ministerio de Ciencia e Innovación [grant numbers BFU2008-01884, RyC-05002168, CSD2007-008-25120]; the Health Institute Carlos III [grant number PI070094/]; the Department of Education of the Basque Government [grant number PI2009-16]; the Department of Industry of the Basque Government [Etortek Research Programs 2008-2010]; and the Bizkaia County.

Acknowledgments

We thank R. Hjerpe (The Scottish Institute for Cell Signalling, University of Dundee, Dundee, Scotland, U.K.), J. Ericsson (Ludwig Institute for Cancer Research, Biomedical Center, Uppsala, Sweden), the Drosophila Genome Resource Center and Addgene for reagents. We thank C. Pérez for technical assistance and A. Carracedo for advise in the sirtuin inhibitory assays. We thank D. Gubb for critical reading of the manuscript prior to submission.

Abbreviations: CFP, cyan fluorescent protein; DAPI, 4′,6-diamidino-2-phenylindole; DBD, DNA-binding domain; FBS, fetal bovine serum; FL, full-length; HDAC, histone deacetylase; HEK, human embryonic kidney; NAM, nicotinamide; Sal, Spalt; SALL, Spalt-like; Salr, Spalt-related; SalrBE, Salr-binding element; SUMO, small ubiquitin-related modifier; OS, Okihiro syndrome; poly(Gln), polyglutamine; RLuc, Renilla luciferase; TBS, Townes–Brocks syndrome; TSA, trichostatin A; WT, wild-type; YFP, yellow fluorescent protein; ZF, zinc finger

References

View Abstract