The SERTAD protein Taranis plays a role in Polycomb-mediated gene repression

The Polycomb group (PcG) proteins have been implicated in epigenetic transcriptional repression in development, stem cell maintenance and in cancer. The chromodomain protein Polycomb (Pc) is a key member of the PcG. Pc binds to the histone mark, trimethylated histone 3 lysine 27 (H3K27me3), to initiate transcriptional repression. How PcG proteins are recruited to target loci is not fully understood. Here we show that the Drosophila SERTA domain protein Taranis (Tara) is involved in transcriptional regulation of Pc target genes. Embryos lacking Tara exhibit a partial homeotic transformation of cuticular the segments, a phenotype associated with the loss of Pc function. Moreover, Drosophila embryos homozygous for a tara hypomorphic allele also misexpress engrailed, a Pc-regulated gene, and this phenotype is associated with the loss of Pc binding to the cis response element in the engrailed enhancer. In relation to that, Pc recruitment is reduced on the salivary gland polytene chromosomes and specifically at the engrailed locus. These results suggest that Tara might be required for positioning Pc to a subset of its target genes.


Introduction
Epigenetic mechanisms dictate the developmental fates of cells in an organism. Once differentiation is complete, transcriptional memory is normally retained in every cell across cell divisions. Two important classes of proteins, implicated in this dynamic and multifaceted process, are highly conserved members of the Polycomb (PcG) and Trithorax (TrxG) groups [1]. The PcG proteins are transcriptional repressors whereas the TrxG proteins counteract transcriptional silencing by antagonizing PcG function. The PcG proteins include members such as Enhancer of zeste and Polycomb (Pc). Enhancer of zeste is a SET domain methyl transferase that methylates histone 3 at lysine 27 to create trimethylated histone 3 lysine 27 (H3K27me3), a repressive chromatin mark. Pc is a chromodomain protein and key member of the PcG, which specifically recognizes and binds to the H3K27me3, leading to silencing of its target genes [2]. a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 Polycomb-mediated repression depends on cis DNA sequence known as the Polycomb Response Element (PRE) [3]. In Drosophila, PREs contain overlapping DNA binding sequences for Zeste, GAGA factor, Dsp1,Pleiohomeotic and ADF1 [4][5][6]. PcGs also exhibit long-range chromosomal interaction in the nuclear environment, where repressed loci contact each other. The PREs in the Bithorax complex (BX-C) demonstrate this type of three-dimensional chromosomal interactions [7][8][9]. During Drosophila development, the dynamic interplay between PcGs determines segmental identity along the anterio-posterior axis of the developing embryo [10]. Genetic manipulation of PcGs results in homeotic transformation in the abdominal segments of the embryo. This is manifested in the misexpression of genes located in the BX-C homeotic cluster namely Ultrabithorax (Ubx), abdominal-A (abd-A) and Abdominal-B (Abd-B) among many other targets [11,12]. Potentially, the PcGs control more than 300 genes as determined by genome-wide chromatin immunoprecipitation [13]. However, the complexity of PcG mediated repression is yet to be fully understood in the context of individual target genes.
Drosophila SERTA (SEI, RBTA and Tara) protein Taranis (Fig 1A) was identified as a modifier of PcG and trxG [14]. Tara protein shows two separate motifs-Plant Homeodomain (PHD)-bromodomain binding domain (PHBBD) followed by SERTA (SEI, RBTA and Tara) motif arranged from the N-terminal to the C-terminal [14] (Fig 1A). PHD domain proteins can recognize acetylated histone tails, therefore are important modulators of chromatin modifications [15]. The bromodomain is also critical for recognition followed by docking onto acetylated lysine in histone proteins [16]. Hence, The PHBBD in Tara indicates possible important functions in chromatin regulation. The SERTA domain proteins are conserved family of transcriptional factors. However, the function of SERTA proteins in epigenetic regulation is not fully understood.
In this study, we provide genetic and biochemical evidence consistent with a role for Tara in PcG-mediated gene silencing. We find that Tara contributes to Pc recruitment to a subset of loci and modulates PcG target gene expression during Drosophila development. Specifically, we show that Tara influences the engrailed PRE activity, thereby regulating engrailed expression in the developing Drosophila embryo.

RNA extraction and real-time qPCR
Total RNA from Drosophila embryo was extracted using Trizol and cDNA was synthesized with Superscripts III. Primers . The consensus SERTA sequence and its similarity with mouse and human proteins is shown below (B) Transgenic PRE mediated pairing sensitive silencing of scalloped (sd) is suppressed in tara loss of function. Simplified representation of the Fab-X locus is shown (not to scale). Fab-X flies carrying transgenic Fab-7 PRE on the X chromosome show the malformed wing blade phenotype, when raised at 29˚C. tara 1 allele suppresses this phenotype as shown by the suppression in the sd phenotype for both wings. (C) tara does not genetically interact with scalloped. Hypomorphic sd 1 allele shows jagged wing margins in males, which is not modified by tara.
Polytene chromosome squash and fluorescent in situ hybridization followed by immunostaining Salivary glands were dissected from two wandering 3rd instar larvae and fixed for 1 min in a drop of the solution 1 (50ul Triton X-100, 400 μl PBS and 50 μl 37% p-formaldehyde) on a coverslip. Glands were then move to about 40ul of the solution 2 (50ul 37% p-Formaldehyde, 200 μl De-ionized water and 250 μl glacial acetic acid) on another coverslip. After 10mins, the coverslip was taken up on a poly-L-Lysine coated glass slide. The coverslip was tapped with the erasure end of a pencil to break up the salivary glands and spread the chromosomes. The slides were placed in a -80˚C freezer for at least 10mins. The coverslip was then removed with a sharp blade and the slides were washed twice in PBS followed by blocking in PBS-0.3% Triton-X 100 plus 2% BSA (Bovine Serum Albumin Fraction V). For antibody staining, 20 μl of diluted primary antibody was added onto slide area containing the squashed chromosomes (marked with a glass marker) and covered with a 22mmx22mm piece of parafilm followed by overnight incubation at 4˚C. The following day, slides were washed twice (15mins each) in PBS-T and incubated at RT for 2 hrs with the secondary antibody. The slides were mounted with Vectashield DAPI.
For FISH, the protocol from Epigenomic NoE (http://epigenesys.eu) was adapted. After polytene squash, the slides were placed in 2xSSC (diluted from 20x stock: 175.3g NaCl, 88.2g Sodium citrate in 1 L deionized water, pH 7.0) for 45mins at 72˚C and dehydrated in 70% and 95% ethanol (2x, 5mins each). DNA was denatured with 0.1N NaOH and the slides were washed with 2xSSC for 5mins. The slides were again dehydrated with ethanol and 14 μl probes in hybridization buffer (50% deionized formamide, 0.3M NaCl, 20mM Tris, 5mM EDTA, 1x Denhardt's solution, 10% dextran sulfate, 10mM DTT, 500ug/ml yeast tRNA) was added onto each slide, covered with a glass slide and sealed with rubber-cement. Hybridization was carried out at 37˚C for overnight in a moist chamber. After hybridization, slides were washed in 2xSSC and transferred to PBS and immunostaining to detect DIG. FISH was performed with sheep anti-DIG (2hrs RT) and Donkey anti-sheep Cy3 (2hrs RT). Polycomb or anti-flag staining was performed as mentioned before.
The engrailed genomic region specific probe was made using the Roche DIG labeling Kit according to manufacturer recommendation. An 8kb Sph1 fragment from engrailed locus (P(EN1)) plasmid described in [18]) was used to generate DIG labeled probes. This 8kb engrailed genomic DNA also contains the PRE necessary for engrailed regulation. We used an 8kb fragment form the engrailed promoter region [18]. This fragment was generated by SPHI restriction digest of P(EN1) plasmid mentioned in 19.

Chromatin immunoprecipitation from Drosophila larvae
Larvae were grown on apple agar plates at 29˚C with excess yeast paste. Approximately 1mg of 2nd instar larvae of the indicated genotypes were crushed in PBS with a pestle. The larval cells were cross-linked at room temperature for 15 minutes with 1% final concentration of formaldehyde, then 0.125mM final concentration of glycine was added and the samples were incubated for another 5 minutes at room temperature. The cells were collected by centrifugation and washed 3 times with PBS-T (1X PBS, pH 7.6 with 0.3% Triton-X). After that, cell lysis buffer was added (50mM HEPES-KOH, 140mM NaCl, 1mM EDTA, pH 8.0, 1% Triton-X, 0.1% Sodium Deoxycholate, 5mM PMSF, 1X PIC) and the cells were sonicated 8 times with 15 second pulses with a Branson S-450 Sonicator set at 40% and output 5. The sample was centrifuged at 13,000xg for 2 minutes at 4˚C to remove the cell debris. A portion of chromatin lysate was heated in Elution Buffer (1% SDS, 100mM NaHCO3) at 65˚C overnight and input chromatin was prepared using the QIAquick PCR Purification Kit (Qiagen Cat. 28106). The remaining chromatin lysate was incubated overnight in 4˚C with 4 μg of anti-Polycomb (Santacruz Biotech, Sc-25762) or an equivalent volume of water for mock treatment. Agarose-G beads were pre-blocked overnight at 4˚C in 1.5 μg salmon sperm DNA per 20 μl beads, then 20 μl was added to the chromatin lysate and the samples were kept on a rocking platform overnight at 4˚C. The beads were washed three times for 5 minutes each wash in 4˚C in Wash Solution (0.1% SDS, 1% Triton-X, 2mM EDTA, 150mM NaCl, 20mM Tris-Cl pH 8.0), then 2 hours at 4˚C in Final Wash Solution (0.1% SDS, 1% Triton-X, 2mM EDTA, 500mM NaCl, 20mM Tris-Cl, pH 8.0). The beads were incubated in room temperature for 20 minutes in Elution Buffer and de-cross linked overnight at 65˚C. Chromatin was isolated using the QIAquick PCR Purification Kit and 1ul was used for each Sybr Green qPCR reaction. QPCR was performed using standard settings in a Biorad iCycler IQ5 instrument. The following sets of primer sets were used from [19].

Western blotting
Total lysate was prepared by homogenizing 10 pairs of salivary gland of the indicated genotypes in 20 μl of cell Lysis Buffer and separated on a 10% gel. Pc antibody (Santacruz Biotech, Sc-25762) was used with 1:500 dilutions and anti-rabbit HRP was used as the secondary.

Tara modulates a transgenic PRE-mediated phenotype
Tara has been previously implicated in PcG/TrxG function [14,20]. We have formerly identified tara in a genetic screen as a potential new component of the RNAi pathway [21]. Since transcriptional silencing mediated by PcGs in certain scenarios requires RNAi components like Argonaute and Dicer [22], we wanted to examine whether Tara could potentially connect PcGs to the RNAi machinery. To this end, we used a transgenic assay system that contains a boundary element for the iab-7 Polycomb response element (PRE) from the BX-C locus called Frontoabdominal 7 (Fab-7) [8,23,24] It regulates selector gene Abdominal-B in the wild type context. We used the Fab-X line, a specific transgenic insertion on X of a 3.6kb fragment from Fab-7 locus containing both the boundary and the respective PRE. Fab-x flies harbor two copies in tandem [8,25]. Interestingly, Fab-X silences the nearby scalloped (sd) gene along with the mini-white reporter, resulting in a stumpy wing phenotype (referred to as sd phenotype) [8] ( Fig 1B). The silencing of sd occurs only in homozygotes (females in this case) when raised at 29˚C, and is dependent on the Polycomb group of proteins and RNAi components like Dicer and Argonaute, suggesting that the interaction between PcG and RNAi is required for the silencing process [8,22]. We tested whether tara can modulate the Fab-X induced sd wing phenotype. We observed that, the heterozygous tara mutant alleles could partially suppress the sd phenotype (Fig 1B). The tara 1 and tara 03881 are two independent alleles both caused by a P-element insertion at the 5' region of the tara locus. Both alleles are associated with embryonic lethality in homozygotes, and they are phenotypically indistinguishable. Expression of tara from a transgene under an ubiquitin promoter rescues the lethality, suggesting the lethality is due to loss of tara [14,17,20]. In the absence of tara mutations, more than 90% of the Fab-X flies show sd phenotype affecting both wings. In tara loss-of-function heterozygous background, we found that this phenotype is reduced to 25%. On the other hand, tara heterozygosity does not suppress wing phenotypes, which is caused by a hypomorphic scalloped allele sd 1 (Fig 1C) [26], suggesting that the observed suppression was unlikely due to genetic interaction between scalloped and tara. Thus, Tara may play a role in Fab-X induced silencing of the nearby sd gene.

Loss of tara function leads to homeotic transformation
To understand whether Tara plays a role in Pc function, we investigated the biological function of tara during Drosophila development. We examined Drosophila embryos homozygous for the tara mutations. Since both tara loss-of-function alleles show similar phenotype in relation to homeotic transformation in the embryos, we focused on tara 03881 to investigate Polycomb related mechanisms. Previous RNA in situ hybridization indicates that the maternal component for tara is minimal, as the general pattern of its expression is mostly ubiquitous. However, it is not significantly expressed before the developmental stage 5 [14]. Thus, we used zygotic mutant embryos lacking tara for this study.
During development, Drosophila embryos generate unique patterns of cuticular bristles on the ventral side. These "denticle bands" are arranged in three thoracic (T1-T3) and eight abdominal segments (A1-A8) from anterior to posterior region of the embryo [27] (Fig 2A). Each denticle band has a distinct length and shape and it is thought to be representative of the segment's identity in the embryo (Fig 2B). In wild-type embryos, the A2-A7 bands in the embryonic cuticle appear trapezoidal, whereas the A8 is rectangular. Cuticles of tara 03881 homozygous embryos show homeotic transformation mostly in the posterior segment, such that abdominal segments following the 5th are transformed to resemble the last abdominal segment (A8) (Fig 2B). The A6 band of tara 03881 homozygous embryos is shortened and is rectangular similar to A8, instead of trapezoidal as seen in wildtype embryos, suggesting a homeotic transformation towards a posterior segmental identity. This transformation is reminiscent of the well-characterized Polycomb phenotype [28], albeit less pronounced. For comparison, a null allele of Polycomb (Pc 3 ) is shown [29]. Pc 3 embryos show complete homeotic transformation across all segments in the embryo. In these embryos, all other abdominal segments transform into segment A8 (Fig 2A). Double homozygous mutant (Pc 3 , tara 03881 ) embryos lacking both tara and Pc do not show further enhancement regarding homeotic transformation, suggesting that the effect of tara might be mediated by Pc. Since tara predominantly causes transformation in the posterior region of the embryo, we quantified the ratio of the length of A4 over A6 bands (Fig 2C). The length here is measured by the distance between two furthest points for each denticle band. Hence, it is the length of the base of trapezoidal or rectangular shape of the specific band measured. In wild-type embryos, the ratio is~1. 16. In Pc 3 homozygous embryos, the ratio is~1. Similar to Pc mutants, the observed ratio in the tara 03881 homozygotes is significantly smaller (~1.03) than that of wildtype embryos. Expectedly, no statistically significant difference was observed between Pc 3 and Pc 3 ,tara 03881 double mutant embryos (Fig 2C). The partial homeotic transformation indicates that Tara could be involved in patterning process that is regulated by PcGs.

Tara genetically interacts with of the BX-C gene Ultrabithorax
Since homeotic transformation is attributable to misexpression of Pc-regulated homeotic genes like Ultrabithorax (Ubx), we next asked whether tara could modulate Ubx dependent phenotype. Ubx is a part of the Drosophila BX-C cluster of genes, which act in concert to regulate posterior segmental identity in the embryo [30].
To investigate the role of tara in Ubx regulation, we utilized a gain-of-function allele of Ubx called Ubx Cbx-1 [31]. The Ubx Cbx-1 allele arose due to a transposition event causing a 17-kb fragment from Ubx upstream region integrating into a downstream intronic region of the gene, causing upregulation of Ubx in parasegment 5 [32,33]. Interestingly, even with a Ubx loss-of-function in cis (Ubx 1 allele), Ubx Cbx-1 was able to activate the wildtype counterpart in trans when present in heterozygous condition (Ubx cbx-1 , Ubx 1 /+), causing ectopic expression of Ubx in wing disc, a phenomenon known as transvection [34]. The Ubx Cbx-1 , Ubx 1 /+ heterozygous flies exhibit partial wing-to-haltere transformation, which is typically enhanced in the PcG mutant background (Fig 3). This phenotype is fully penetrant but varies in expressivity. We categorized the phenotype into four different classes-I to IV-depending on the severity of the transformation (Fig 3). The Class I category wings has a slight curved region on the posterior boundary of the wing, Class II wings show partial loss of the fifth wing-vein and the area between the 4th and 5th vein appear crumpled. The wing morphology is severely affected in Class III, with the presence of bubbles and crumpled posterior margin. The Class IV wings are the most affected as the wing-to-haltere transformation is prominent in the stumpy wing morphology. The severity of wing phenotype is positively correlated with the expression of Ubx in the larval wing disc tissue [32]. At 22˚C, about 75% of Ubx Cbx-1 , Ubx 1 heterozygous flies show mild transformation (Class I), and only 9% shows Class II and 15% shows Class III phenotypes, respectively. In the presence of one copy of tara 03881 , the majority of the wings fall into the Class II (46%) and Class III (53%) categories, suggesting that loss of tara enhances the wing-to-haltere transformation of Ubx Cbx-1 , Ubx 1 heterozygous flies. The wing phenotype, however, is not as severe as that of Ubx Cbx-1 , Ubx 1 /Pc 3 flies, in which 86% of the wings are in Class IV category. These results suggest that Tara normally negatively regulates Ubx expression, although Pc appears more important in repressing Ubx.

Loss of tara function reduces Pc recruitment to PREs
The genetic experiments described above indicate that Tara might be involved in Pc function. The suppression of the Fab-X and Ubx Cbx-1 phenotypes by tara mutations could have resulted from a failure of Pc recruitment to the PRE in Fab-X or Ubx regulatory regions. Two of the important steps in the canonical PcG functions are trimethylation of histone 3 lysine 27 (H3K27me3) followed by Pc binding to this epigenetic mark via its chromodomain  [35]. Tara could conceivably be involved in either of these processes. We next set out to investigate these possibilities. To this end, we first created mitotic clones using the FLP-FRT technique and examined Pc levels in hypomorphic homozygous tara 03881 cells of the 3rd instar larval salivary glands (see Methods). We observed a significant reduction in Pc fluorescence in tara 03881 / tara 03881 cells (Fig 4A, arrows).
Drosophila polytene chromosome shows distinctive pattern of Pc binding as multiple high and low intensity bands. This provides us with the necessary setup to examine Pc binding to chromosomes at higher resolution compared to whole-mount immunostaining in salivary gland. We knocked down tara in the salivary gland by expressing dsRNA against tara using the UAS-Gal4 system followed by immunostaining with anti-Pc and anti-H3K27me3 (see Methods; S1 Fig). The salivary gland-specific Sgs3-gal4 driver was used for this purpose to drive RNAi specifically in the salivary gland. We observed a significant reduction in the number of Pc bands in tara knockdown (Fig 4B bottom panel) compared with control (Fig 4B top  panel), corroborating the observation in tara loss-of-function clones. Pc bands were decreased to an average of 44 per nuclei. In comparison, Sgs3 > w1118 (control) shows 63 Pc bands ( Fig  4B, graph). Magnified view of the third chromosome is shown in the panel next to the whole chromosome for better comparison. However, tara RNAi appears to decrease the level of Pc in most of the low intensity bands, whereas the number of high intensity bands remains unchanged. Our result indicates that Tara might be important for Pc recruitment at specific target loci but may not be necessary for all the genes regulated by Pc.
We did not observe any significant changes in the levels and distribution of H3K27me3 in polytene chromosomes after knocking down tara (S2 Fig). When compared to the Tubulin80--Gal4 driver only control, there was no significant change in Pc protein levels for embryos with tara knockdown (S3 Fig). Our data indicate that Tara might be involved in Pc recruitment rather than affecting histone methylation of H3K27 or the production of Pc per se.

Misregulation of engrailed in tara loss-of-function embryo
As mentioned previously, Pc function in Drosophila is mediated by cis DNA element called PRE. We have found tara to be important in the transgene Fab-X mediated silencing and possible repression of Ubx, both of which require PRE activity (Figs 1B and 3). We wanted to test whether tara can affect the PREs of other known Pc targets. Drosophila segment polarity gene engrailed (en) is a Pc target with two well-characterized PREs in a 2.4kb region upstream of its transcription start site. The proximal PRE is located 181bp upstream of the transcription start site and is characterized by Pc binding [36]. We first looked at en expression pattern in tara 03881 homozygous embryos. en is expressed in 14 distinct stripes during embryonic development at germ-band extension [37,38]. In each stripe, en expression is limited to 2-3 rows of cells in wildtype embryos (Fig 5A). Pc 3 homozygous embryos show a misexpression of en, as the total number of cells expressing en in a stripe increases. We observed a very similar phenotype in the tara embryos homozygous for tara 03881 (Fig 5A). Interestingly, unlike the homeotic transformation phenotype (Fig 2A), the tara 03881 and Pc 3 double mutant embryos show a further increase in the number of cells that express en than that of single mutant embryos ( Fig  5A). The effect is most discernible in the 13 th parasegment. Similar synergistic relationship is known to exist between PcG proteins in the context of engrailed regulation [39]. To quantify en expression, we counted the number of en-expressing cells in the 13th parasegment of each embryo. In wildtype embryos, the en stripe in the 13th parasegment contains average 17 cells (Fig 5A, white box and right), whereas in tara 03881 homozygous embryo, this number is increased to 22. Pc 3 homozygous mutation results in expression of en in average 29 cells, and the tara 03881 Pc 3 double mutant shows enhancement as the number of en expressing cells reaches 33 (Fig 5A, white box and right).
Tara is required for efficient Pc binding to the engrailed PRE Next, we asked whether misexpression of en could be attributable to loss of Pc binding to the en locus and PRE. We employed chromatin immunoprecipitation (ChIP) to investigate Pc binding to the en locus. For FISH-I we used the Sgs3-gal4 driver to knockdown tara in the salivary gland and examined polytene chromosomes from salivary glands of wandering third instar larvae. Squashed polytene chromosome from control (Sgs3 >w1118 and Sgs3 > taraRNAi were subjected to FISH-I (see Methods). In the driver-only control, a distinct Pc band (red) is coincident with the FISH signal (green) from the engrailed locus (Fig 5B, top panel). The intensity of this band is lower than a nearby high intensity Pc band. Knocking down tara resulted in a much-reduced Pc signal at the en locus (Fig 5B, bottom  panel). The nearby high intensity Pc band, however, does not diminish, consistent with our observation that not all Pc bands are affected by knocking down tara (Fig 4B). This suggests that Tara might be required for Pc binding to a subset of Pc regulated loci including en.
We also examined whether Tara is required for Pc binding to en PRE with Chromatin IP (ChIP). We used the ubiquitous Tubulin80-gal4 driver to globally knockdown tara. Tubulin80 > taraRNAi and control 2nd instar larvae were collected for chromatin extraction. ChIP was performed with primers described in [19]. One primer set was used to amplify the right panel, N = 6 (wt, tara) and 8(pc, pc-tara) (B) Salivary glands from Sgs3 > taraRNAi or control wandering 3rd instar larvae were squashed and subjected to FISH for the en locus and followed by immunostaining with antibody against Pc. Top: Sgs3 >w1118; Bottom: Sgs3 > taraRNAi at 29˚C. To maximize knockdown, larvae were raised at 29˚C. (C) Tubulin80-gal4>UAS-taraRNAi and control 2nd instar larvae were collected for chromatin extraction. Chromatin IP was performed with primers to amplify an en PRE fragment located at 0.2 Kb upstream from the transcription start site and an intergenic fragment located 1.3 Kb downstream. The diagram shows approximate location of the PRE and intergenic region as a negative control (not to scale).
https://doi.org/10.1371/journal.pone.0180026.g005 en PRE, located 0.2kb upstream from the transcription start site. The other set of primers were for a fragment located 1.3 kb downstream site in the en gene region (Fig 5C), which does not contain any PRE or Pc binding sequence, hence can be considered as an internal negative control. In Tubulin80 >w1118 larvae, Pc binding to en PRE was observed but not in the control en gene region (Fig 5C). In contrast, Tubulin80 > taraRNAi larvae show reduced Pc binding at en PRE (p = 0.04). As expected, the internal gene control did not show any Pc enrichment. These results suggest that Tara is required for the recruitment of Pc to the endogenous en PRE to modulate its transcription.

Discussion
We have investigated the role of Tara in Pc-mediated gene repression in Drosophila and have found that Tara modulates the epigenetic silencing of the developmental gene, engrailed, that is transcriptionally regulated by the Polycomb group proteins. The segment polarity gene engrailed is a target of the Polycomb-mediated transcriptional repression. We have further shown that Tara is required for Pc binding at of the engrailed Polycomb response elements upstream of the transcription start site, although the detailed mechanism awaits further investigation. We have found that Tara genetically interacts with the Hox gene Ubx. During development, the Ubx expression pattern at stage 16 ventral nerve cord seems to be uniform in tara loss-of-function embryos, instead of a gradient pattern as found in wildtype (not shown). However, this observation does not satisfactorily account for the embryonic homeotic transformation in tara loss-of-function. In the developing wildtype embryo, Abd-B is expressed in represses PS13 and PS14, whereas Ubx is absent in those segments. The mis-expression of Ubx in Parasegment 13 or 14 was not clearly visible in the tara-/-embryos. Alternatively, tara lossof-function embryos did not show any significant change in the Abd-B expression pattern either (S4 Fig for Abd-B). In PS10-12, Abd-B acts in concert with abd-A to determine parasegment identity by repressing Ubx [11]. However, we have not examined abd-A expression in our study to test the possibility of its deregulation in tara-/-embryos. Since the interplay between these Hox genes determines parasegment identities in the wildtype embryo, we are unable to determine the mechanistic roles tara might play in this respect. It is possible that, Tara might be indirectly involved in repressing Hox genes in cooperation with Pc. Alternatively, Tara could be regulating interacting partners of the Hox transcription factors, which might be required for their activity. Future experimentation will be necessary to tease out the role for tara in Hox gene regulation. It will be interesting to find how tara expression correlates with Abd-B and Ubx. It is interesting to note that tara is expressed at all stages during embryonic development as determined by in situ hybridization [14]. After gastrulation, tara expression covers the entire germband of the developing embryo. The expression at stage 11-12, is very strong and broadly covers the embryo.
The regulation of en expression is under the control of complex inter-regulation in Drosophila embryo and imaginal disc involving PcG and Trx proteins [39][40][41][42]. It has been shown that Pc is required for maintaining en expression in the posterior par segments in the developing embryo, and in the posterior compartments of imaginal discs [39]. In tara loss-of-function mutant embryos, we found that the number of cells expressing en increases (Fig 4A), which suggests that more cells adopt the posterior identity. Despite this, the anterio-posterior compartment boundary is still maintained. The En protein is repressed in the anterior region of the wing disc by PcG proteins [39]. Surprisingly, the tara (tara 03881 ) clones in the wing disc do not show any significant alteration in the engrailed expression as revealed by antibody staining (not shown; also see [20]. This might be due to the redundant functions of members of the PcG in repressing en [43,44], or that Tara is only required in tissue regeneration after damage [20,45]. In our study, we did not observe any physical interaction between Tara and Pc by co-immunoprecipitation. We also found that Tara localization on polytene chromosomes showed only partial colocalization with Pc band (not shown). We were unable to observe Tara recruitment on the en locus, as Tara signal on the polytene chromosome decreases significantly following immune-FISH procedure (not shown). It is likely that Tara might regulate Pc recruitment via indirect mechanisms, which remains to be investigated.
Tara was initially characterized as an enhancer of trithorax based on genetic interactions [14] and was recently shown to mediate the function of the PcG protein polyhomeotic in repressing engrailed expression [20]. Our work provides additional support for the biological functions of Tara as a mediator of PcG functions. It is noted that similar dual functions have been observed with other member of the Polycomb and Trithorax family [46], which include the histone methyltransferase Enhancer of zeste (E(z)), a trithorax mimic of E(z), called E(z) trm [47,48], Additional sex combs (Asx) [49], Corto [50], and batman [51]. Thus, PcG/trx dependent transcriptional regulation is complex and is context-dependent. Tara could be part of this multifaceted network and might behave as a PcG member for repression of certain targets while as a Trithorax member for activation of certain other genes. However, genes that are positively regulated by Tara have not yet been identified. It will be interesting to explore interaction between effectors of context-dependent transcriptional regulators and Tara in the future.