TnaA, an SP-RING Protein, Interacts with Osa, a Subunit of the Chromatin Remodeling Complex BRAHMA and with the SUMOylation Pathway in Drosophila melanogaster

Tonalli A (TnaA) is a Drosophila melanogaster protein with an XSPRING domain. The XSPRING domain harbors an SP-RING zinc-finger, which is characteristic of proteins with SUMO E3 ligase activity. TnaA is required for homeotic gene expression and is presumably involved in the SUMOylation pathway. Here we analyzed some aspects of the TnaA location in embryo and larval stages and its genetic and biochemical interaction with SUMOylation pathway proteins. We describe that there are at least two TnaA proteins (TnaA130 and TnaA123) differentially expressed throughout development. We show that TnaA is chromatin-associated at discrete sites on polytene salivary gland chromosomes of third instar larvae and that tna mutant individuals do not survive to adulthood, with most dying as third instar larvae or pupae. The tna mutants that ultimately die as third instar larvae have an extended life span of at least 4 to 15 days as other SUMOylation pathway mutants. We show that TnaA physically interacts with the SUMO E2 conjugating enzyme Ubc9, and with the BRM complex subunit Osa. Furthermore, we show that tna and osa interact genetically with SUMOylation pathway components and individuals carrying mutations for these genes show a phenotype that can be the consequence of misexpression of developmental-related genes.


Introduction
SUMOylation is a post-translational protein modification that can change the location, stability, activity or the interactions of the protein targets involved in many cellular processes, including cell death, cell cycle, signal transduction, and gene expression [1]. SUMOylation is the addition of SUMO (Small Ubiquitin-related MOdifier) to lysine residues of the target protein in the consensus amino acid sequence YKxE (Y represents a hydrophobic amino acid) [2]. Hundreds of proteins are SUMOylated in Drosophila [3]. The SUMOylation pathway starts with processing of an immature SUMO protein by the Ulp/SENP family of proteases. Next, the activating enzyme E1 (an Aos1/Uba2 heterodimer) generates a mature SUMO-adenylate intermediary which then forms a thioesther bond between the catalytic cysteine of Uba2 and SUMO. SUMO is next transferred to the E2 conjugating enzyme (Ubc9), which transfers SUMO to the target proteins. The SUMO E3 ligases function by stimulating the activity of Ubc9 or by facilitating the formation of an Ubc9-substrate complex. Finally, proteins of the Ulp/SENP family proteases make this whole process reversible [4].
The tna gene was identified in a genetic screen designed to find brahma (brm)-interacting genes [5]. brm encodes the SNF2 type-ATPase of the BRM chromatin remodeling complexes [6,7]. The osa gene encodes an exclusive subunit of one type of BRM complexes [6,8,9]. Besides interacting with brm, tna interacts even stronger with osa. All three genes (brm, osa, and tna) are required for proper expressions of the homeotic genes [5]. Homeotic genes determine the identity of body segments in Drosophila [10,11].
The role of various components of the SUMOylation pathway have been studied in Drosophila development [12,13]. tna is involved in homeotic gene expression but little is known about the proteins encoded by this locus. tna expresses a at least one putative isoform called TnaA [5]. This isoform has an XSPRING (eXtended SP-RING) domain that harbors a zinc finger of the SP-RING type {Siz/PIAS (Protein Inhibitors of Activated STAT [Signal Transducers and Activator of Transcription])-RING (Really Interesting New Gene)}. This zinc finger is present in one of the four major groups of proteins that have SUMO E3 ligase activity [1]. The only SP-RING finger proteins with putative SUMO E3 ligase activity that have been identified in the Drosophila proteome are Su(var)2-10 [14] and TnaA [5].
Here we show that TnaA physically interacts with both Ubc9 (the SUMO E2 conjugating enzyme) and with Osa (a putative in vivo target). We determined the dynamics of different TnaA species throughout development and showed that TnaA is an embryonic nuclear protein and is also present at discrete bands on polytene salivary gland chromosomes of third instar larvae. We also found that defects in tna cause larval lethality, abnormalities in the whole protein profile and an extension of the lifespan at this stage. Finally, we found genetic interactions between tna and osa and genes encoding the SUMOylation pathway components.

Ethics Statement
All animal handling was approved by the Instituto de Biotecnología Bioethics Comittee, Permit Number 226 (2009/ 12/04), which follows NOM-062 animal welfare mexican law. All efforts were made to minimize animal suffering. Animals were sacrificed by CO 2 euthanasia.

Protein Extraction and Analyses
Soluble protein extracts for the developmental Western were obtained from 1 g of Ore-R individuals from each developmental stage with Trizol (Invitrogen). For cellular localization of the TnaA proteins, soluble nuclear (SNF) and cytoplasmic fractions were obtained from Ore-R embryo collections of 3-21 hour postfertilization [15]. The SNF was also used for the TnaA coimmunoprecipitation (Co-IP) assays. For Osa Co-IP assays, a total soluble protein fraction was obtained from Ore-R embryo collections of 3-21 hour postfertilization [16]. Protein extracts from salivary glands of third instar larvae were obtained by collecting the glands in PBS buffer plus Complete protease inhibitors [EDTA-free protease inhibitor tablet (ROCHE)], and boiling them for 5 minutes in sample loading buffer. The proteins were separated by SDS-PAGE and electrotransfered to nitrocellulose membranes for Western blot analyses. Immunoblots were done according to standard procedures and proteins of interest were detected with specific antibodies using different chemoluminiscence kits (Supersignal West Pico Chemiluminescent Substrate from Thermo scientific, ECL Plus Western Blotting Detection System or ECL Advanced Western Blotting Detection kit from Amersham, GE Healthcare, USA), according to manufacturers instructions.

Production and Affinity Purification of TnaA Antibodies
To generate antibodies against different TnaA regions, we used the TnaA cDNA that contains the TnaA translated exons from the ZAP1 clone [5] that represent the TnaA RD transcript [17]. The TnaA cDNA clone was digested with BamHI and two fragments were independently subcloned into the pGEX2T vector to generate glutathione S-transferase (GST) fusion proteins harboring the TnaA amino-termini (amino acids 159-432, GST-TnaA NH2-1 ) and the XSPRING domain (amino acids 433-856, GST-TnaA X-SPRING1 ). GST-fusion proteins were expressed and purified [18] to inject Winstar rats to raise polyclonal antibodies [19]. The antibodies from total sera were affinity-purified [20].

Pull-down and Immunoprecipitation Assays
All the clones used in this work were nucleotide-sequenced. The Drosophila Ubc9 cDNA (BDGP Gold collection of Drosophila Genomics Resource Center) was amplified with the Forward: 59-AGTTCGGAGAATTCTCCGGCATTGCTATTACACG-39 and Reverse: 59-CGGAATCCTCGAGGCG-CTTCTCGTACTCCAG-39 primers, and cloned in the EcoRI and XhoI sites of the pGEX-4T vector. Pull-down assays were done as described previously [21]. Immunoprecipitations were done on SNF or total protein extracts from 3-21 hour postfertilization Ore-R embryos [22]. In these assays we made two preclearings steps and we used the Buffer PD (20 mM HEPES, pH 7.9, 100 mM NaCl, 1 mM EDTA, 4 mM MgCl 2 , 1 mM DTT, 0.1% NP-40, 10% glycerol and 0.2 mM PMSF).

Fly Strains, Genetic Procedures, and Larval Staging
Unless otherwise noted, all mutations are described in Flybase [17]. Briefly, tna 1 , tna 5 , osa 1 and osa 2 are EMS-induced mutations. In tna 1 Gln 566 changed to a stop codon [5]. tna 5 was recovered after EMS mutagenesis in a genetic screen to identify brminteracting mutations (J. A. K., unpublished results). The lesion in the lwr 5 allele (Arg 104 to His) is located in a region that has been involved in the interaction between ubiquitin-conjugating enzymes with the HECT or RING ubiquitin E3 ligases [24]. The lwr 4-3 and lwr 13 were both derived from imprecise excision of P-elements inserted in the 59 regulatory zone [25,26]. smt3 04493 is a P-element insertion 10 bp upstream of the first exon of smt3 [27]. Fly cultures and crosses were performed according to standard procedures. Flies were raised on cornmeal-molasses media at 25uC unless otherwise noted. Media were supplemented with 0.05% of bromophenol blue to stage third instar larvae according to the gut dye clearance [28].

Immunostaining of Ring Glands, Salivary Glands, and Polytene Chromosomes of Third Instar Larvae
Immunostaining of ring and salivary glands were done as described by [29], and the immunostaining of polytene salivary gland chromosomes was done as reported by [30]. For immunostaining of polytene salivary gland chromosomes, the TnaA X-SPRING antibodies were preabsorbed with fixed 0-3 hour embryos [31]. Polytene chromosomes and salivary and ring glands images were captured on a Leitz DMIRB inverted photoscope equipped with a Leica TCS Nt laser confocal imaging system, a Zeiss Inverted Axiovert fluorescent microscope, a Leica Aristaplan fluorescent microscope or an Olympus Inverted confocal FV1000 microscope. Images were processed using Image J.

TnaA 130 and TnaA 123 in Space and Time throughout Development
The tna gene produces several large transcripts that are differentially expressed from embryo through adult stages [5,17]. The main large transcript is 6.1 kb and it peaks at the pupal stage [5]. Translation of this transcript predicts a protein product of 127 kDa that we named TnaA [5]. To study TnaA, we prepared two affinity-purified antibodies: TnaA NH2 that was raised against the amino-terminal region and TnaA XSPRING that was raised against the XSPRING domain ( Fig. 1 and Material and Methods). Both antibodies recognize the same proteins on adult male soluble extracts and they were used indistinctly along this work ( Fig. 2A).
Two main TnaA protein products, one of 130 kDa (TnaA 130 ) and another one of 123 kDa (TnaA 123 ) are present in varying abundance throughout development (Fig. 2B). The abundance does not correspond to the tna mRNA expression pattern [5] suggesting postranscriptional regulation. We sometimes observe another product heavier than TnaA 130 in embryos of 3-21 h (Fig. 2B). These three Tna species we found, are consistent with the three Tna polypeptides described in Flybase [17]. Nevertheless we cannot discard the possibility that TnaA could be postranslationally modified. For example, we determined using the SUMOsp 2.0 program [32] that TnaA has two putative SUMOylation sites and one putative SUMO Interacting Motif (SIM) [33] (data not shown). In extracts isolated from 0-3 hour embryos, we detected very low levels of TnaA 130 , while TnaA 123 was not detected. In extracts isolated from 3-21 hour embryos, we detected a TnaA form larger than TnaA 130 , and the levels of both TnaA 130 and TnaA 123 increased, reaching maximums in the first larval instar. Decreases in the abundances of both proteins were observed in second and third instar larvae, with the levels of TnaA 123 higher than those of TnaA 130 . Both forms abundance decreased substantially in pupae and TnaA 130 was observed again at the pharate stage meanwhile TnaA 123 is not detected. In adult flies of both sexes, TnaA 130 and TnaA 123 were both highly abundant at about equal levels. The appearance of TnaA 123 was always preceded by the presence of TnaA 130 .
Next, we investigated the subcellular location of the TnaA proteins in nuclear and cytoplasmic fractions from 3-21 hour embryos (Fig. 2C, upper panel). The largest subunit of RNA polymerase II and b-tubulin were used to test the purity of the fractions (Fig. 2C, middle and lower panels). We found that TnaA 123 was enriched in the nuclear fraction whereas TnaA 130 was enriched in the cytoplasmic fraction (Fig. 2C). It has been shown that SUMO is present in prothoracic gland nuclei [29] in third instar larvae. tna mutant individuals arrest development at the larval-pupal transition which is where less TnaA protein is expressed (see ahead). This suggests that TnaA may be expressed in specific tissues relevant for metamorphosis. We immunostained salivary (Fig. 2D, upper panel) and ring glands (Fig. 2D, lower panel) from third instar larvae with the TnaA XSPRING antibody and we found that TnaA was present most highly within the nucleus of the secretory cells of salivary glands and in prothoracic gland cells.

TnaA is Critical for Larval Development
While we can detect TnaA 130 and TnaA 123 in Ore-R and in tna 1 /+ or tna 5 /+ individuals, TnaA 130 is barely detectable and TnaA 123 decreases dramatically in tna 1 /tna 5 larvae (Fig. 3A, left panel). The tna 1 mutation changes Gln 566 to a stop codon, is recessive lethal [5] and behaves as a dominant negative. tna 1 is a much stronger dominant enchancer of osa 1 than is a deficiency of the tna region (Table 1). tna 1 would produce a truncated protein of 62 kDa that we have been able to observe in heterozygous tna 1 /+ salivary glands soluble extracts (Fig. 3A, right panel). The molecular lesion of tna 5 has not been determined, but it behaves genetically as a hypomorphic allele and its product can be detected in tna 1 /tna 5 third instar larvae extracts (Fig. 3A, left panel).
To better understand tna function we studied the lethality of tna 1 /tna 5 animals. The tna 1 /tna 5 larvae (Fig. 3B) did not have melanotic tumors as observed in lwr or aos1 mutant individuals [34,35,36], nor are they a larger size as observed for smt3 knockdowned larvae [29]. We found that 65% of tna 1 /tna 5 individuals reach the third instar larval stage (Fig. 3C), but only 41% pupated and only 8% of the expected individuals reached the pharate stage. No tna 1 /tna 5 individuals eclosed as adults (Fig. 3C). We also noticed that the tna 1 /tna 5 third instar larvae that did not pupate often survived long after their heterozygous tna 1 /+ or tna 5 / + siblings larvae pupated. Some of these tna 1 /tna 5 larvae have an extended lifespan of at least two weeks (Fig. 3C). A similar extension of larval lifespan was previously observed in animals with reduced levels of SUMO [29], Aos1 (one of the E1 subunits) [34] or Ubc9 (E2) [35,36].
Given the abnormal behavior of tna 1 /tna 5 larvae and knowing that the TnaA profile is altered (Fig. 3A), we characterized the protein profile of their salivary glands (Fig. 3D). We staged the larvae by feeding them with bromophenol blue [28] and divided them in early (blue) and late (white) larvae. All tna 1 /tna 5 larvae remained as early larvae (blue). They were collected 24 hours after they crawled from the food to obtain their salivary glands and we determined their protein profile (Fig. 3D). Although tna 1 /tna 5 larvae remained blue, the protein profile differed from both the early and late wild-type Ore-R salivary glands obtained under the same conditions. Differences in the quantity and quality of proteins present in tna 1 /tna 5 salivary glands fall mostly in the range over 72 kDa (Fig. 3D).

TnaA is Chromatin-associated at Discrete Sites on Polytene Salivary Gland Chromosomes
We have shown that TnaA 123 is nuclear in Drosophila embryos (Fig. 2C) and that TnaA (probably TnaA 123 ) is mainly nuclear in salivary and ring glands from third instar larvae (Fig. 2D). We immunostained polytene salivary gland chromosomes of third instar larvae and found that TnaA is associated with discrete sites (Fig. 4A). The number of TnaA sites suggests that TnaA might be required for the transcription of more than just the homeotic genes. Interestingly, most of the TnaA signals detected on polytene salivary gland chromosomes are located in interbands which are thought to have decondensed chromatin where transcription can occur (Fig. 4B). Because of the strong genetic interactions between tna and osa [5], we coimmunostained for TnaA and Osa on polytene salivary gland chromosomes. TnaA colocalizes with Osa at some sites, but not at others (Fig. 4C, upper and bottom panels). We do not know whether this is because TnaA is not required at all genes regulated by Osa, or whether it is due to an interaction between TnaA and Osa that is more transient than Osa localization.
TnaA Physically Interacts with Ubc9 and with Osa SUMO E3 ligases function for selection of SUMOylation targets and/or for enhancement of the SUMO conjugation process. TnaA has an SP-RING zinc finger that is also present in a subclass of SUMO E3 ligases that includes the PIAS proteins in mammals [37] and Su(var)2-10 in Drosophila [14]. Since the SP-RING in the PIAS proteins physically interacts with Ubc9 [38,39], we explored whether TnaA physically interacts with Drosophila Ubc9, using yeast two-hybrid assays and pull-down assays.
For the yeast two-hybrid assays we first used the full-length TnaA protein (Fig. 1) fused to the yeast GAL4-DNA binding domain as ''bait'', and the full-length Drosophila Ubc9 protein (Fig. 5A) fused to the GAL4-activation domain as ''prey''. We found that the full-length TnaA protein was able to activate the transcription of at least two reporter genes in the absence of a ''prey'' (Fig. 5B), and as a consequence the full-length TnaA protein could not be used to test for the Ubc9 interaction in this assay. We then split the TnaA protein into five fragments that cover the whole TnaA protein (Fig. 1). Two out of the five fragments contain the SP-RING zinc finger (TnaA XSPRING2 and TnaA SP-COO Qless ). The other fragments have different TnaA regions that include the two glutamine-rich domains (TnaA NH2-2 ), the bipartite nuclear location signal (TnaA QLess ) and the carboxyending (TnaA COO ). We found that the TnaA XSPRING2 fragment interacted with Ubc9 in the yeast two-hybrid assay while the other fragments, including TnaA SP-COO QLess , did not interact (Fig. 5B). These results show that the TnaA SP-RING zinc finger is necessary but not sufficient for the TnaA interaction with Ubc9 in this assay.
Osa is a subunit of some BRM complexes, and the osa gene strongly interacts with tna [5]. Since it was found that Osa is modified by SUMO in Drosophila embryos [3], we thought that TnaA might be involved in Osa SUMOylation. We searched for SUMOylation consensus sites (yKxE) in the Osa protein sequence (2713 aa) using the SUMOsp 2.0 program [32] and found eight putative SUMOylation sites (Fig. 5A), six of them located within a segment located from amino acids 1951 to 2600 surrounding the C2 domain [40]. We will refer to the fragment with the six putative SUMOylation sites as Osa C2 in this work. We synthesized the Osa C2 cDNA from polyA + RNA of 3-21 hour embryos and fused it to the GAL4-activation domain to use as ''prey'' in the yeast two-hybrid assay. We tested the six TnaA baits already described (including full-length TnaA), and found that baits harbouring the SP-RING (TnaA XSPRING2 or TnaA SP-COO QLess ) did not interact with the Osa C2 prey. Although TnaA NH2-2 (and to a lesser extent, full-length TnaA) interacted with Osa C2 , these baits also interacted with pGADT7 or pGADT7-SV40 negative control samples, preventing us from concluding whether the interactions with Osa C2 are bona fide. In contrast, we found that the TnaA Qless bait cleanly interacts physically with Osa C2 (Fig. 5B).
Although the TnaA XSPRING2 region interacted physically with Ubc9 in the yeast two-hybrid assays, we wanted to test for TnaA/ Ubc9 physical interactions in Drosophila embryos. We performed pull-down assays using as bait a purified GST-Ubc9 fusion protein incubated with a nuclear protein extract from 3-21 hour embryos where we know TnaA is present (Fig. 2B). After extensive stringent washing, the presence of TnaA amongst the GST-Ubc9-interacting proteins was assessed by Western analyses with the TnaA X-SPRING antibody (Fig. 5C). As expected, we found that full-length TnaA from nuclei of Drosophila embryos interacts with full length GST-Ubc9, confirming the results that we obtained with the yeast two-hybrid assays using TnaA fragments and further suggesting that these proteins interact in vivo.
In all reported cases it is known that only a fraction of the whole pool of a SUMOylatable protein in a cell is SUMOylated, either because of spatial restrictions (the target protein should be located where the SUMO and the SUMOylation enzymes are) or because fine regulation constricts the amount of the SUMOylated protein [4]. We showed that Osa C2 interacts with a fragment of TnaA (TnaA QLess ) in a yeast two-hybrid assay (Fig. 5B). To test whether this interaction can be observed with the full-length proteins in Drosophila embryos, we performed TnaA or Osa coimmunoprecipitation assays from total or nuclear protein extracts from 3-21 hour embryos. For this purpose, we first showed that the TnaA XSPRING and Osa antibodies are able to immunoprecipitate TnaA and Osa, respectively (Fig. S1), and that the control proteins Hsp70 and Cdk7 do not coimmunoprecipitate with TnaA or with Osa, respectively (Fig. S2). Interestingly, we found that TnaA coimmunoprecipitates with a fraction of Osa found in nuclear protein extracts from 3-21 hour embryos (Fig. 5D), and that reciprocally, Osa coimmunoprecipitates with TnaA from a total protein extract of 3-21 hour embryos (Fig. 5E). Since we found that TnaA interacts physically with Osa and with Ubc9 (Fig. 5) we tried to test whether TnaA has SUMO E3 ligase activity on the Osa C2 fragment using a mammalian in vitro assay (Active Motif kit). Although Osa C2 is SUMOylated in this assay, we were not able to get convincing evidence that TnaA has SUMO E3 ligase activity under these conditions (data not shown).
tna and osa Genetically Interact with Components of the SUMOylation Pathway tna genetically interacts with brm and osa [5]. Transheterozygous adult flies carrying mutations in combinations of any of these three genes have a strong held-out wing phenotype [5,8] (Fig. 6). This phenotype appears to result from reduced expression from the P2 promoter of the homeotic gene Antp [8]. The interactions with tna might be a consequence of reduced SUMOylation of Osa (and/or Brm) proteins. If so, mutations in other components of the SUMOylation pathway might also show genetic interactions. We generated transheterozygous flies carrying mutant alleles of either the SUMO E2 conjugating enzyme Ubc9 (lwr 5 , lwr 4-3 , and lwr 13 ) [25,26] or SUMO (smt3 04493 ) [27] in combination with mutant alleles of tna (tna 1 and tna 5 ) or osa (osa 1 and osa 2 ). All of these individuals have at least one wild type copy of each gene to allow survival to the adult stage.
Individuals carrying tna alleles other than tna 1 (tna 3 or tna 5 ), or deficiencies uncovering the tna region [Df(3L)vin2 or Df(3L)lxd6] do not show the held-out wing phenotype. This phenotype is also not shown by transheterozygous individuals carrying tna alleles other than tna 1 or the tna deficiencies in combination with smt3 or lwr alleles. In contrast, we found that smt3 04493 , lwr 4-3 and lwr 13 enhance both the penetrance and the expressivity of the held-out wing phenotype of tna 1 individuals (from 19% with weak expressivity to 56%, 46% and 52% with stronger expressivity, respectively, Fig. 6 and Table 1). Thus, we confirmed genetically that tna interacts with the SUMOylation pathway genes smt3 and lwr. Interestingly, we found a maternal effect in the enhancement of the tna 1 held-out wings phenotype in double heterozygous smt3 04493 /+; tna 1 /+ individuals. The enhancement is only observed when there is not maternal contribution of tna (Table 1). We did not observe this maternal effect with the other genes tested. Since we observed strong genetic interactions between tna and osa (Fig. 6) [5], therefore we tested whether the SUMOylation genes lwr and smt3 also interact with osa (osa 1 and osa 2 in Table 1 and Fig. 6). We found that smt3 04493 augmented the penetrance and the expressivity of the weak held-out wing phenotype of osa 1 (Fig. 6, Table 1) but that it did not interact with the weaker osa 2 allele. None of the lwr alleles tested interacted with osa 1 or osa 2 , suggesting that Ubc9 activity in these heterozygous individuals is sufficient to reach appropriate SUMOylation levels.

Discussion
The presence of the SP-RING, the physical interaction of the SUMO E2 conjugating enzyme Ubc9 with TnaA, and the genetic interaction of tna with genes encoding SUMOylation pathway proteins suggest that TnaA may be involved in the SUMOylation pathway to activate transcription. TnaA may also have other functions not directly related to SUMOylation. These other functions may or may not act together with SUMOylation to positively regulate gene expression.

TnaA Function in Gene Expression Involving the SUMOylation Pathway
Gene expression involves the integration of many regulatory mechanisms. Recently, many examples of SUMOylation and/or ubiquitylation during transcriptional regulation have been described [4]. These examples include the clearance of activators to favor transcription cycles in inducible genes [41] and the assembly of different proteins into a complex [42,43]. Most of the tna interacting genes (osa, brahma, moira, kohtalo, skuld, and kismet) [5] encode subunits of complexes involved in chromatin remodeling and transcription by RNA polymerase II, suggesting that SUMOylation may be important at multiple aspects of gene regulation in Drosophila. Typically, SUMO-tagged proteins are recognized by a binding partner that contains a SIM (SUMO Interacting Motif) [33]. All of the proteins encoded by the tna interacting genes listed above have more than one SIM and SUMOylation sites (data not shown) and could be either SUMOylation targets, readers of the SUMO mark, or proteins that help TnaA exert its function(s). SUMO E3 ligases are required for the enhancement and/or for the specifity of the SUMOylation tagging on targets. In this work we utilized different approaches to show that TnaA is involved in the SUMOylation pathway possibly as a SUMO E3 ligase. We showed a TnaA physical interaction with Ubc9 and genetic interactions between tna and osa with SUMOylation pathway genes. SUMOylated Osa is found in early embryos (0-3 hour) [3] and embryonic TnaA and Osa coimmunoprecipitate reciprocally (this work). We also showed that a GST-Ubc9 fusion physically interacts with native nuclear TnaA from Drosophila embryos. Hence, we suggest that Osa is a good candidate to be a TnaA-SUMOylation target in vivo. Our data suggest that TnaAdependent SUMOylation of Osa and/or of other target(s), particularly proteins associated with Osa (e.g. other BRM complex subunits, histones, or others, see ahead), may be required for correct gene expression including homeotic genes. Osa is a large protein of around 280 kDa with an ARID domain which binds AT-rich sequences, LXXLL domains [8] that could help it to interact with nuclear receptors and has eight putative SUMOylation target sequences, six of them in the Osa C2 fragment (Fig. 5A). In humans there are three proteins related to Osa, BAF250a, BAF250b and BAF200/ARID2 [44] and it was reported that For expressivity of held-out wing phenotype (HWO) see Fig. 6.
Originally tna was identified in a screen to find Brm-interacting proteins [5]. Although we did not study here whether Brm can be SUMOylated, it has been reported that mammalian SUMO-2 can be acetylated at K33 to inhibit some SUMO-SIM interactions [46]. Interestingly, these authors also show that the bromodomain of p300, besides recognizing acetylated histones [47], can bind the SUMO acetylated form, opening the question of whether other bromodomains, such as the one present in the Brahma protein, would be able to recognize a putative Drosophila acetylated SUMO when present in any of its interactor proteins.
TnaA may also be promoting homeotic gene expression by inactivation through SUMOylation of a PcG protein. Indeed, SUMOylation of the PcG protein Scm (encoded by the Sex comb on midleg gene) decreases its levels at the PRE (Polycomb Response Element) located upstream the Ubx homeotic gene. SUMO compromised animals show a reduction of Ubx expression and it has been suggested that TnaA may be involved in Scm SUMOylation to promote homeotic gene expression [48].

Other TnaA Interactors and SUMO-independent Functions of TnaA
We found that TnaA 130 is mainly cytoplasmic and TnaA 123 is mainly nuclear. Although most studied SUMO enzymes and targets are in the nucleus, there are some examples of SUMOylation of proteins in the cytoplasm [49]. As TnaA 130 always precedes  In the Osa protein, the ARID, the C1 and C2 domains (grey boxes), the SUMO interacting motif (SIM) and the Osa C2 fragment (dark line) are indicated. Forward (black circles) and inverted (gray circles) putative SUMOylation consensus sites in these proteins are indicated. For TnaA baits see Fig. 1. (B) TnaA interaction with Ubc9 and Osa C2 in yeast two-hybrid assays. Yeast colony complementation of growth controls in SD-Trp/2Leu media due to the presence of pGBKT7 (Trp + ) and pGADT7, (Leu + ) plasmids (left) in the same yeast cells. Interaction assay in QDO +3-AT (SD-Trp/2Leu/2Ade/2His +3-AT) media (right). Growth is observed when baits and preys interact, allowing GAL4 reconstitution with the consequent ADE2 and HIS3 reporter genes transcription. Baits were TnaA fragments (Fig. 1) Fig. 2B), we think that TnaA may be processed to enter the nucleus to SUMOylate its targets. Notably, SUMOylation pathway proteins with well known nuclear activities also SUMOylate targets in the cytoplasm [50]. Thus, with what we know at present, we cannot discard the possibility that TnaA 130 can also function in the cytoplasm. We also found that tna interacts with the cTub23C gene that encodes an isoform of c-tubulin [51] and with taranis (tara) [5]. The significance of the interaction of tna with tara and cTub23C is currently unknown.
It is possible that TnaA could be necessary for BRM complex(es) function(s) regardless of SUMOylation, and that independently, SUMOylation could be required for function of other BRM complex(es) components. We cannot neither rule out the possibility that TnaA may have other functions independent of its possible role in the SUMOylation pathway, as has been (1 mg), and 3-21 hour embryos soluble nuclear fraction (3.7 mg) were used. The Western was revealed with the TnaA NH2 antibody (1:120). Lanes are labeled as above. The equivalent amount of an irrelevant antibody was used as mock (M). Molecular weight markers are indicated (left). doi:10.1371/journal.pone.0062251.g005 Figure 6. SUMOylation pathway mutations enhance held-out wing phenotype of tna and osa flies. Flies with different held-out wing phenotype expressivity. Fly genotype is indicated in each picture. Penetrance of the held-out wing phenotype in each genotype is in Table 1  reported for the PIAS proteins, known SP-RING SUMO E3 ligases [52,53,54]. The SP-RING plays a key role in this PIAS activity. The TnaA SP-RING is immersed in a 300-aminoacid region that we called the XSPRING domain that is shared with the vertebrate proteins Zimp7 and Zimp10 [KIAA1886 and KIAA1224 respectively, 5]. Although TnaA is related to the PIAS proteins because it has an SP-RING, it does not have the SAP (Scaffold attachment factor-A/B, Acinus and PIAS domain) nor the PINIT motifs that are PIAS signature domains.
The SAP and PINIT motifs in the PIAS proteins confer functions related to structural anchoring and transcriptional regulation. In mammals it has been shown that PIAS1 promotes the transcriptional repressive activity of Msx1 through regulating its location in a SUMO-independent way [53], it controls the stability of Msx1 by preventing its ubiquitination [55] and it regulates the transcriptional activity of GATA4 [56]. Similarly, in Xenopus, XPIASy down-regulate XSmad2 transcriptional activity independently from XPIASy SUMO E3 ligase activity [57].
Although human Zimp7, human Zimp10, and Drosophila TnaA do not have these other PIAS signature motifs they have transcriptional activation domains [58,59] (Fig. 5B). The presence of a transcriptional activation domain could explain why we could not use the TnaA-Gal4 DNA-binding domain fusion in the yeast two-hybrid system (Fig. 5B). This suggests that TnaA, besides its possible role in the SUMOylation pathway, has other functions in Drosophila transcriptional activation.

TnaA in Drosophila Development
We described a genetic interaction between tna, osa, and SUMOylation pathway genes. TnaA interacts physically with Ubc9 through the SP-RING supporting the genetic interaction data. Animals derived from osa and tna mutant germline clones die at different stages of development. While the osa ones do not survive embryogenesis [8] the tna ones die mostly as third instar larvae [5]. A pool of Osa is found SUMOylated in embryos of 0-3 hour of development when zygotic expression has not started [3] and TnaA is barely detectable (overexposure of Fig. 2B, data not shown). Moreover, when we studied the tna and smt3 interaction, we found a tna maternal effect. The held-out wings phenotype in smt3/+; tna/+ adults is observed when the mother is tna defective, but we do not observe this when the mothers have low dosages of SUMO (Table 1). We think it is probable that SUMOylated Osa plays a role at early stages of development. SUMOylation of embryonic Osa can happen in the maternal germline or in the embryo with the help of the maternally-inherited SUMOylation pathway machinery. This machinery may include TnaA if TnaA is involved in SUMOylation or another protein with a SUMOrelated function. It is also possible that smt3/+; tna/+ embryos derived from smt3 mothers do not present the held-out wings phenotype because the SUMOylation pathway can compensate even with low dosages of SUMO. On the other hand, if TnaA is related to SUMOylation, embryos derived from tna mothers would lack correct SUMOylation of specific targets (such as Osa) causing later the appearance of the held-out wings phenotype.
Why do tna mutant animals die at later stages of development? One possibility is that proteins other than TnaA can exert its function on particular targets, such as Osa, or that they could only impact the TnaA targets in earlier stages of development, but not in later stages. SUMO is required for metamorphosis [29]. As the majority of tna mutant animals die as larvae or pupae and cannot proceed to metamorphosis (Fig. 3C) [5], and as TnaA is in prothoracic gland nuclei of third instar larvae (Fig. 2D) obvious candidates for regulation by tna would be the ecdysone-pathway, ecdysone-regulated or patterning genes.
The relevance of SUMOylation (and of genes like tna) in different developmental processes is just starting to emerge. The requirement of SUMOylation and of tna to maintain gene expression makes that the next challenges will be to find the SUMOylation and tna targets in vivo and to understand the consequences of this modification in proteins involved in chromatin dynamics and in gene expression. Figure S1 The TnaA XSPRING and Osa antibodies immunoprecipitate TnaA and Osa proteins, respectively. (A) TnaA was immunoprecipitated from 3-21 hour embryo-soluble nuclear fraction (500 mg) using TnaA XSPRING antibody (1 mg). The Western was revealed with TnaA XSPRING (1:100). The three panels correspond to films with increasing exposure times. Input