BIOLOGICAL OVERVIEW

Enhancer of Polycomb [E(Pc)] is a gene with a dual identity, serving as a suppressor of position-effect variegation (PEV) (Sinclair, 1998a) and as a Polycomb Group (PcG) gene. E(Pc) and its relationship to PEV will be discussed first.

For many years, histologists have distinguished between two types of chromatin: the genetically active euchromatin and the genetically inactive heterochromatin. Position-effect variegation occurs when chromosomal rearrangements juxtapose a euchromatic gene next to a broken segment of heterochromatin. Expression of the transposed gene is repressed in some cells but not in others, producing a mosaic phenotype. Such heterochromatin-induced gene silencing occurs because there are differences in structure and in the regulation of gene expression between heterochromatin and euchromatin, and these changes can be transmitted to the translocated gene. Repression is probably caused by the spreading of heterochromatin into the euchromatic gene, causing inactivation, although models invoking sub-nuclear localization have been gaining support. Many modifiers of PEV have been identified (Sass, 1998). Suppressors of variegation [Su(var)s] are predicted to encode either structural components of heterochromatin, or proteins that regulate heterochromatin components. To date, several Su(var)s have been cloned. These include: Su(var) 3-7, a zinc finger protein; Su(var) 205, which encodes HP1, a protein purified from heterochromatin; modulo, a DNA-binding protein (Garzino, 1992); Su(var) 3-6, the protein phosphatase 1 catalytic subunit (Dombradi, 1992); Su(var) 3-9, a protein that contains domains found in other chromatin regulators (Tschiersch, 1994) and the gene encoding S-adenosylmethionine synthetase, which is required for spermine production (Larsson, 1996). The known Su(var)s havestructures consistent with a role in heterochromatin formation (Stankunas, 1998 and references).

In addition to its role in the enhancement of PEV, Enhancer of Polycomb is also a member of the Polycomb Group (PcG) gene family. Several groups have investigated whether PcG mutations modify PEV, or if Su(var)s have homeotic phenotypes, as a test of the hypothesis that the two groups have overlapping functions. The results suggest that there is little overlap between the groups, with the exception of the PcG genes Enhancer of Polycomb and Enhancer of zeste, which act as Su(var)s (Sinclair, 1998a). However, a trithorax group protein, Additional sex combs, can serve as an enhancer of PEV, supporting the notion that there is an overlap between regulators of homeotic genes and modifiers of PEV (Sinclair, 1998b). E(Pc) is unusual among PcG mutations because it does not, by itself, possess the capacity to generate a homeotic phenotype in embryos or adults (only Abd-B shows a very modest ectopic expression in embryos), once the maternal contribution of E(Pc) protein or mRNA is removed. However, mutations in E(Pc) enhance homeotic mutations in the PcG genes Polycomb, Polycomb like, polyhomeotic, Sex combs extra, Sex comb on midleg and super sex combs, suggesting that E(Pc) is important for PcG function. It may be that, like Su(z)2, E(Pc) is partially functionally redundant and, therefore, lacks homeotic effects in embryos and adults. One other gene, S-adenosylmethionine synthetase, has been identified that has no homeotic phenotype: it enhances phenotypes of PcG mutants and acts as a Su(var) (Larsson, 1996). Like S-adenosylmethionine synthetase, E(Pc) may be indirectly required for PcG function. It may be that E(Pc) regulates PcG or Su(var) expression, or has other indirect effects. E(Pc) has now been cloned from Drosophila and mouse. Both gene products contain a large novel domain that is also conserved in yeast and nematode proteins. The E(Pc) protein of Drosophila is ubiquitously expressed and binds to polytene chromosomes at about 100 sites, of which only about a third overlap with Pc-binding sites. Interestingly, E(Pc) is not detected at the heterochromatic chromocenter, supporting a model in which the E(Pc) has a functional rather than a structural role in heterochromatin, and supporting the conclusion that there is less overlap between mechanisms of heterochromatin formation and PcG repression than had previously been supposed (Stankunas, 1998).

A simple model to explain how E(Pc) functions as both a PcG protein (promoting gene silencing) and a Su(var) proposes that E(Pc) protein has a structural role in heterochromatin and in PcG complexes. This model is very unlikely because E(Pc) is not detected in heterochromatin of the chromocenter, as is HP1, another Su(var). The sequence analysis of E(Pc) does not reveal active sites of any known enzymes, making it unlikely that E(Pc) has an enzymatic function and argues against a hypothesis that E(Pc) modifies the chromodomain of HP1 and Pc (Kennison, 1995). The alternative suggestion (Kennison, 1995) that E(Pc) interacts with the chromodomains of HP1 and Pc has not been ruled out. These experiments do not address the possibility that E(Pc) directly affects nuclear compartmentalization of chromatin into active and inactive compartments, or that E(Pc) is needed for the establishment of a nuclear architecture required for establishment of repression. Nevertheless, the observation that E(Pc) is a chromatin protein of limited distribution makes it probable that E(Pc) regulates genes and thus may have indirect effects on nuclear architecture. The discrete binding of E(Pc) to polytene chromosomes makes it unlikely that E(Pc) plays a general role in chromosome or chromatin structure which is only indirectly affected by heterochromatin formation and repression by PcG proteins (such as S-adenosylmethione synthetase). Rather, it is suggested that E(Pc) has an indirect effect on the formation of heterochromatin, and thus on position-effect variegation; this probably occurs via the regulation of genes required for heterochromatin formation. Consistent with this idea, E(Pc) binds to the locations of some modifiers of PEV, but so far there is no evidence for direct regulation of modifiers of PEV by E(Pc) (Stankunas, 1998).

Drosophila Reptin and other TIP60 complex components promote generation of silent chromatin

Histone acetyltransferase (HAT) complexes have been linked to activation of transcription. Reptin is a subunit of different chromatin-remodeling complexes, including the TIP60 HAT complex (see Tip60). In Drosophila, Reptin also copurifies with the Polycomb group (PcG) complex PRC1, which maintains genes in a transcriptionally silent state. Genetic interactions have been demonstrated between reptin mutant flies and PcG mutants, resulting in misexpression of the homeotic gene Scr. Genetic interactions are not restricted to PRC1 components, but are also observed with another PcG gene. In reptin homozygous mutant cells, a Polycomb response-element-linked reporter gene is derepressed, whereas endogenous homeotic gene expression is not. Furthermore, reptin mutants suppress position-effect variegation (PEV), a phenomenon resulting from spreading of heterochromatin. These features are shared with three other components of TIP60 complexes, namely Enhancer of Polycomb, Domino, and dMRG15. It is concluded that Drosophila Reptin participates in epigenetic processes leading to a repressive chromatin state as part of the fly TIP60 HAT complex rather than through the PRC1 complex. This shows that the TIP60 complex can promote the generation of silent chromatin (Qi, 2006).

It is proposed that Reptin acts as a subunit of the TIP60 HAT complex to generate a repressive chromatin state. This is a novel activity of a HAT complex previously shown to promote transcription. This study shows that Reptin copurifes with the Polycomb complex PRC1. This prompted an investigation of whether the biochemical interaction with PRC1 was accompanied by a genetic interaction. It was shown that Reptin and PRC1 components genetically interact to regulate expression of the Hox gene Scr. However, Reptin also interacts with a PcG gene product not associated with the PRC1 complex, Pcl. Although no interactions were detected between reptin heterozygous mutants and several PREs tested, a PRE from the Ubx gene is derepressed in reptin homozygous mutant cells. This shows that Reptin contributes an essential function to the activity of this PRE. However, unlike most PcG genes, reptin homozygous mutants do not derepress endogenous Hox gene expression. It appears that repression of endogenous Hox genes is more complex and not as sensitive to the loss of Reptin as the Ubx PRE. In contrast to most PcG genes, reptin mutants suppress PEV. Interestingly, derepression of the Ubx PRE also occurs in embryos mutant for other suppressors of PEV, indicating that this PRE may be highly sensitive to the chromatin environment in its vicinity. Since reptin mutants suppress PEV and fail to derepress endogenous Hox gene expression, reptin is not considered a bona fide PcG gene, and it is found unlikely that Reptin protein contributes an essential function to the PRC1 complex. In fact, the biochemical activities ascribed to PRC1 can be reconstituted either with recombinant dRing1/Sce or with four core components whose activity can be further enhanced by the DNA-binding proteins Zeste and GAGA (Qi, 2006).

Given that Reptin is present in TIP60 complexes in mammals and recently was shown to be a component of a Drosophila TIP60 complex, the possibility is considered that the genetic interactions observed with PcG genes are due to the presence of Reptin in the fly TIP60 complex. The products of two previously characterized Drosophila genes, E(Pc) and domino, are also present in the TIP60 complex. Strikingly, E(Pc) and domino mutants share with reptin the ability to genetically interact with PcG genes and suppress PEV. E(Pc) is an unusual PcG gene that has very minor effects on Hox gene expression, and unlike most PcG genes, modifies PEV. In both yeast and humans, E(Pc) homologs form a core complex with Esa1 (TIP60) and Yng2 (ING3) that is sufficient for the nucleosomal acetylation of histones H4 and H2A by the NuA4 complex. That such an integral NuA4/TIP60 complex component displays phenotypes similar to reptin mutants suggests that Reptin functions through the fly TIP60 complex (Qi, 2006).

Domino protein is similar to p400 and to SRCAP in mammals and to Swr1 in yeast. Swr1 has recently been shown to exchange the variant histone H2A.Z (Htz1 in yeast) for H2A in nucleosomes. Intriguingly, an involvement of Htz1 (H2A.Z) in controlling the spreading of silenced chromatin has recently been demonstrated in yeast. Exchange of variant histones may be a conserved feature of chromatin regulation since a recent report demonstrates that Drosophila H2Av behaves genetically as a PcG gene and suppresses PEV. Domino exchanges phosphorylated and acetylated H2Av for unmodified H2Av after DNA damage. However, no change was found in binding of H2Av to polytene chromosomes prepared from domino mutant larvae (Qi, 2006).

The chromatin remodeling factor Bap55 functions through the TIP60 complex to regulate olfactory projection neuron dendrite targeting

The Drosophila olfactory system exhibits very precise and stereotyped wiring that is specified predominantly by genetic programming. Dendrites of olfactory projection neurons (PNs) pattern the developing antennal lobe before olfactory receptor neuron axon arrival, indicating an intrinsic wiring mechanism for PN dendrites. These wiring decisions are likely determined through a transcriptional program. This study found that loss of Brahma associated protein 55 kD (Bap55) results in a highly specific PN mistargeting phenotype. In Bap55 mutants, PNs that normally target to the DL1 glomerulus mistarget to the DA4l glomerulus with 100% penetrance. Loss of Bap55 also causes derepression of a GAL4 whose expression is normally restricted to a small subset of PNs. Bap55 is a member of both the Brahma (BRM) and the Tat interactive protein 60 kD (TIP60) ATP-dependent chromatin remodeling complexes. The Bap55 mutant phenotype is partially recapitulated by Domino and Enhancer of Polycomb mutants, members of the TIP60 complex. However, distinct phenotypes are seen in Brahma and Snf5-related 1 mutants, members of the BRM complex. The Bap55 mutant phenotype can be rescued by postmitotic expression of Bap55, or its human homologs BAF53a and BAF53b. These results suggest that Bap55 functions through the TIP60 chromatin remodeling complex to regulate dendrite wiring specificity in PNs. The specificity of the mutant phenotypes suggests a position for the TIP60 complex at the top of a regulatory hierarchy that orchestrates dendrite targeting decisions (Tea, 2011).

The stereotyped organization of the Drosophila olfactory system makes it an attractive model to study wiring specificity. The first olfactory processing center is the antennal lobe, a bilaterally symmetric structure at the anterior of the Drosophila brain. It is composed of approximately 50 glomeruli in a three-dimensional organization. Each olfactory projection neuron (PN) targets its dendrites to one of those glomeruli to make synaptic connections with a specific class of olfactory receptor neurons. Each PN sends its axon stereotypically to higher brain centers (Tea, 2011).

During development, the dendrites of PNs pattern the antennal lobe prior to axons of olfactory receptor neurons. The specificity of PN dendrite targeting is largely genetically pre-determined by the cell-autonomous action of transcription factors, several of which have been previously described. Furthermore, chromatin remodeling factors have been shown to play an important role in PN wiring (Tea, 2010), although very little is currently known about their specific functions. This study reports a genetic screen for additional factors that regulate PN dendrite wiring specificity; Brahma associated protein 55 kD (Bap55) was identified as a regulator of PN dendrite wiring specificity as part of the TIP60 chromatin remodeling complex (Tea, 2011).

Bap55 is an actin-related protein, the majority of which physically associates with the Brahma (BRM) chromatin remodeling complex in Drosophila embryo extracts. There are two distinct BRM complexes: BAP (Brahma associated proteins; homologous to yeast SWI/SNF) and PBAP (Polybromo-associated BAP; homologous to yeast RSC), both of which contain Brahma, Bap55, and Snf5-Related 1 (Snr1). The human homologs of the BAP and PBAP complexes are called the BAF (Brg1 associated factors) and PBAF (Polybromo-associated BAF) complexes, respectively. The BRM/BAF complexes are members of the SWI/SNF family of ATP-dependent chromatin-remodeling complexes, and have been shown to both activate and repress gene transcription, in some cases, of the same gene (Tea, 2011).

In experiments purifying proteins in complex with tagged Drosophila Pontin in S2 cells, Bap55 was also co-purified as a part of the TIP60 complex, as determined by mass spectrometry. The TIP60 histone acetyltransferase complex has been shown to be involved in many processes, including both transcriptional activation and repression. The complex contains many components, including Bap55, Domino (Dom), and Enhancer of Polycomb (E(Pc)). Dom, homologous to human p400, is the catalytic DNA-dependent ATPase; its ATPase domain is highly similar to Drosophila Brahma and human BRG1 ATPase domains. E(Pc) is homologous to human EPC1 and EPC2 and is an unusual member of the Polycomb group; it does not exhibit homeotic transformations on its own, but rather enhances mutations in other Polycomb group genes (Tea, 2011).

This study provides evidence that Bap55 functions as a part of the TIP60 complex rather than the BRM complex in postmitotic PNs to control their dendrite wiring specificity (Tea, 2011).

To further understanding of dendrite wiring specificity in Drosophila olfactory PNs, a MARCM-based forward genetic screen was performed using piggyBac insertional mutants. MARCM allows visualization and genetic manipulation of single cell or neuroblast clones in an otherwise heterozygous background, permitting the study of essential genes in mosaic animals. In this screen, GH146-GAL4 was used to label a single PN born soon after larval hatching, which in wild-type (WT) animals always projects its dendrites to the dorsolateral glomerulus DL1 in the posterior of the antennal lobe. The DL1 PN also exhibits a stereotyped axon projection, forming an L-shaped pattern in the lateral horn, with additional branches in the mushroom body calyx. A mutant, called LL05955, was identified in which DL1 PNs mistargeted to the dorsolateral glomerulus DA4l in the anterior of the antennal lobe. This phenotype is strikingly specific, with 100% penetrance. Arborization of mutant axons, however, was not obviously altered. The piggyBac insertion site was identified using inverse PCR and Splinkerette PCR. LL05955 is inserted into the coding sequence of Bap55, encoding a homolog of human BAF53a and BAF53b. Precise excision of the piggyBac insertion reverted the dendrite mistargeting phenotype, confirming that disruption of the Bap55 gene causes the dendrite mistargeting (Tea, 2011).

In addition to causing DL1 mistargeting, Bap55 mutants also display neuroblast clone phenotypes. In WT, GH146-GAL4 can label three distinct types of PN neuroblast clones generated in newly hatched larvae. Two of these clones, the anterodorsal neuroblast clone and the lateral neuroblast clone, possess cell bodies that lie dorsal or lateral to the antennal lobe, respectively. PNs from these two lineages project their dendrites to stereotyped and nonoverlapping subsets of glomeruli in the antennal lobe. The third type of clone, the ventral neuroblast clone, has cell bodies that lie ventral to the antennal lobe and dendrites that target throughout the antennal lobe due to the inclusion of at least one PN that elaborates its dendrites to all glomeruli (Tea, 2011).

In Bap55^-/- PNs, anterodorsal neuroblast clones display a mild reduction in cell number, and their dendrites are abnormally clustered in the anterior dorsal region of the antennal lobe, including the DA4l glomerulus. Lateral neuroblast clones display a severe reduction in cell number, and the remaining dendrites are unable to target to many glomeruli throughout the antennal lobe. Ventral neuroblast clones display a mild reduction in cell number and a reduced dendrite mass throughout the antennal lobe. During development, the lateral neuroblast first gives rise to local interneurons before switching to produce PNs; in mutants affecting cell proliferation, this property of the lateral neuroblast displays as a severe reduction in GH146-labeled PNs. The severely reduced cell number in Bap55 mutants suggests that Bap55 is essential for neuroblast proliferation or neuronal survival. In the anterodorsal and ventral neuroblasts, PN numbers are not drastically changed; thus, the phenotypes indicate that Bap55 is important for dendrite targeting in multiple classes of PNs (Tea, 2011).

In WT, Mz19-GAL4 labels a subset of the GH146-GAL4 labeling pattern. It labels a small number of PNs derived from two neuroblasts, which can be clearly identified in WT clones generated in newly hatched larvae. Anterodorsal neuroblast clones target their dendrites to the VA1d glomerulus, as well as the DC3 glomerulus residing immediately posterior to VA1d (not easily visible in confocal stacks). Lateral neuroblast clones target all dendrites to the DA1 glomerulus. Unlike GH146-GAL4, WT Mz19-GAL4 never labels ventral neuroblast clones because it is not normally expressed in those cells (Tea, 2011).

In Bap55 mutant PN clones, however, Mz19-GAL4 labels additional PNs in anterodorsal, lateral, and ventral clones compared to their WT counterparts. This phenotype suggests that some Mz19-negative PNs now express Mz19-GAL4. In anterodorsal clones, Mz19-GAL4 labels additional cells, although not as many as are labeled by GH146-GAL4. The PNs also mistarget their dendrites to the anterior antennal lobe, similar to mutant GH146-GAL4 anterodorsal neuroblast clones. WT lateral neuroblast clones normally contain GH146-positive PNs and GH146-negative local interneurons. In Bap55^-/- lateral neuroblast clones, Mz19-GAL4 predominantly labels local interneurons that send their processes to many glomeruli throughout the antennal lobe and do not send axon projections to higher brain centers. Lateral clones also show ectopic PN labeling with a lower frequency. The Bap55 mutant appears to cause derepression of Mz19-GAL4, resulting in labeled local interneurons. Ventral neuroblast clones are never labeled in WT Mz19-GAL4, yet are labeled in Bap55 mutants. This further indicates a derepression of the Mz19-GAL4 labeling pattern (Tea, 2011).

Unlike GH146-GAL4, WT Mz19-GAL4 never labels single cell clones when clone induction is performed in newly hatched larvae. This is because Mz19-GAL4 is not expressed in the DL1 PN, the only GH146-positive cell generated during this heat shock time of clone generation. However, in Bap55 mutant PN clones, Mz19-GAL4 ectopically labels single cell anterodorsal PN clones targeting to the DA4l glomerulus, which show an L-shaped pattern in the lateral horn with branches in the mushroom body calyx, similar to GH146-GAL4 labeling. The simplest interpretation is that this compound phenotype reflects first a derepression of Mz19-GAL4 in the DL1 PN, and second a DL1 to DA4l mistargeting phenotype in Bap55 mutants (Tea, 2011).

To test whether Bap55 functions in the neuroblast or postmitotically in PNs, GH146-GAL4, which expresses only in postmitotic PNs, was used to express UAS-Bap55 in a Bap55^-/- single cell clone. The dendrite mistargeting phenotype was shown to be rescued to the WT DL1 glomerulus and it is concluded that Bap55 functions postmitotically to regulate PN dendrite targeting. The axon phenotype remains the stereotypical L-shaped pattern (Tea, 2011).

The Drosophila Bap55 protein is 70% similar and 54% identical to human BAF53a and 71% similar and 55% identical to human BAF53b. BAF53a and b are 91% similar and 84% identical to each other. Using GH146-GAL4 to express human BAF53a or b in a Bap55^-/- single cell clone, it was found that the human homologs can effectively rescue the Bap55^-/- dendrite mistargeting phenotype. Interestingly, both also cause the de novo DM6 dendrite and ventral axon mistargeting phenotypes in 6 out of 19 cases for BAF53a and 2 out of 32 cases for BAF53b. Thus, human BAF53a and b can largely replace the function of Drosophila Bap55 in PNs (Tea, 2011).

To address whether Bap55 functions as a part of the BRM complex in PN dendrite targeting, two additional BRM complex mutants were tested for their PN dendrite phenotypes. First, Brahma (brm), the catalytic ATPase subunit of the BRM complex, which is required for the activation of many homeotic genes in Drosophila, was tested. Null mutations have been shown to decrease cell viability and cause peripheral nervous system defects. RNA interference knockdown of brm in embryonic class I da neurons revealed dendrite misrouting phenotypes, although not identical to the Bap55 phenotype. The human homologs of brm, BRM and BRG1 (Brahma-related gene-1), both have DNA-dependent ATPase activity. Inactivation of BRM in mice does not yield obvious neural phenotypes, but inactivation of BRG1 in neural progenitors results in reduced proliferation. BRG1 is likely to be required for various aspects of neural development, including proper neural tube development (Tea, 2011).

In PNs, brm mutants displayed anterodorsal single cell clone mistargeting to non-stereotyped glomeruli throughout the antennal lobe, with each clone differing from the next. This is in contrast to the highly stereotyped DA4l mistargeting of Bap55 mutants. brm^-/- neuroblast clones also displayed phenotypes where dendrites make small, meandering projections throughout the antennal lobe, which does not resemble the Bap55^-/- phenotype. They additionally exhibit a perturbed cell morphology phenotype, which is markedly more severe than the Bap55 mutant phenotype (Tea, 2011).

Next, Snr1, a highly conserved subunit of the BRM complex, was tested. It is required to restrict BRM complex activity during the development of wing vein and intervein cells and functions as a growth regulator. Its human homolog, SNF5, is strongly correlated with many cancers, yet little is known about its specific role in neurons (Tea, 2011).

In PNs, Snr1 mutants displayed similar phenotypes to brm mutants. The single cell clones displayed mistargeting to non-stereotyped glomeruli throughout the antennal lobe, with each clone demonstrating a unique phenotype. The neuroblast clones exhibited small meandering dendrites throughout the antennal lobe, which also showed extremely perturbed cell morphology. These results, especially the non-sterotyped single cell clone phenotypes, indicate that the BRM complex functions differently from Bap55 in controlling PN dendrite targeting (Tea, 2011).

brm and Snr1 mutants were further analyzed with Mz19-GAL4 to determine whether their phenotypes resembled the Bap55 mutant phenotype of derepression. It was not possible to generate any labeled clones, indicating one of three possibilities: increased cell death, severe cell proliferation defects, or repression of the Mz19-GAL4 labeling pattern. In any of the three cases, the phenotype does not resemble the Bap55^-/- mutant phenotype of abnormal activation of Mz19-GAL4 in single cell or neuroblast clones, indicating that the BRM complex functions differently from Bap55 in PNs (Tea, 2011).

In the same screen in which the Bap55 mutation was identified, LL05537, a mutation in dom that resulted in a qualitatively similar phenotype to Bap55 mutants was identified. dom^-/- DL1 PNs split their dendrites between the posterior glomerulus DL1 and the anterior glomerulus DA4l. Their axons exhibit a WT L-shaped pattern in the lateral horn (Tea, 2011).

The LL05537 allele contains a piggyBac insertion in an intron just upstream of the translation start of dom. Because the piggyBac insertion contains splice acceptor sites and stop codons in all three coding frames, this allele likely disrupts all isoforms of dom. Similarly to Bap55, the piggyBac insertion site was identified using inverse PCR and Splinkerette PCR. Precise excision of the piggyBac insertion reverted the dendrite targeting phenotype, confirming that disruption of the dom gene causes the dendrite mistargeting. In addition, a BAC transgene that contains the entire dom transcriptional unit rescued the dom^-/- mutant phenotypes (Tea, 2011).

Dom is the catalytic DNA-dependent ATPase of the TIP60 complex and has been shown to contribute to a repressive chromatin structure and silencing of homeotic genes. Dom is a member of the SWI/SNF family and its ATPase domain is highly similar to the Drosophila Brahma and human BRG1 ATPase domains. The human homolog of Dom is p400, which is important for regulating nucleosome stability during repair of double-stranded DNA breaks and an important regulator of embryonic stem cell identity (Tea, 2011).

To determine whether Bap55 and Dom genetically interact, UAS-Bap55 was expressed in a dom^-/- PN. This manipulation did not significantly alter the dendrite phenotype. The axon branching pattern also was not altered (Tea, 2011).

Another component of the TIP60 complex, E(Pc), was also examined. In Drosophila, E(Pc) is a suppressor of position-effect variegation and heterozygous mutations in E(Pc) result in an increase in homologous recombination over nonhomologous end joining at double-stranded DNA breaks. Following ionizing radiation, heterozygous animals also exhibit higher genome stability and lower incidence of apoptosis. Yet little is known about its role in neurons (Tea, 2011).

In this study, it was found that E(Pc)^-/- DL1 PN dendrites also mistarget to the anterior glomerulus DA4l and exhibit the stereotyped L-shaped axon pattern in the lateral horn. A BAC transgene that contains the entire E(Pc) transcription unit rescued the E(Pc) mutant phenotypes. To determine whether Bap55 and E(Pc) genetically interact, UAS-Bap55 was expressed in an E(Pc)^-/- DL1 PN. This manipulation caused the dendrites to split between the DA4l and DM6 glomeruli, and resulted in axons targeting ventrally to the lateral horn (Tea, 2011).

Neuroblast clones mutant for dom also exhibit dendrite mistargeting phenotypes to inappropriate glomeruli throughout the antennal lobe. Anterodorsal and lateral neuroblast clones show a very mild reduction in cell number and their dendrites do not target to the full set of proper glomeruli. Ventral neuroblast clones, when compared to WT, exhibit incomplete targeting throughout the antennal lobe (Tea, 2011).

Further analysis of dom mutants by labeling with Mz19-GAL4 revealed the same derepression as in Bap55 mutants. dom mutant Mz19-GAL4 PN clones also label anterodorsal, lateral, and ventral neuroblast clones with phenotypes similar to GH146-GAL4 labeled neuroblast clones. In anterodorsal and lateral neuroblast clones, Mz19-GAL4 labels a large number of PNs that target to many glomeruli throughout the antennal lobe, although the cell number is smaller than GH146-GAL4 labeling. Ventral neuroblast clones are never labeled in WT Mz19-GAL4, yet are labeled in dom mutants. Mz19-GAL4 also labels single cell clones that split their dendrites between the DA4l and DL1 glomeruli and form the stereotypical L-shaped axon pattern in the lateral horn. As in Bap55 mutants, this compound phenotype likely results from ectopic labeling of a DL1 PN, which further mistargets to DA4l (Tea, 2011).

The E(Pc) phenotypes in GH146 and Mz19-GAL4 labeled neuroblast clones, as well as Mz19-GAL4 labeled single cell clones, displayed similar phenotypes to dom as described above. The phenotypic similarities in single cell clone dendrite mistargeting and derepression of a PN-GAL4 in mutations that disrupt Bap55, dom and E(Pc) strongly suggest that these three proteins act together in the TIP60 complex to regulate PN development (Tea, 2011).

This study has demonstrated a similar role for three members of the TIP60 complex in olfactory PN wiring. The TIP60 complex plays a very specific role in controlling dendrite wiring specificity, with a precise mistargeting of the dendrite mass in Bap55, dom, and E(Pc) mutants. This specific DL1 to DA4l mistargeting phenotype has only been seen in these three mutants, out of approximately 4,000 other insertional and EMS mutants screened, supporting the conclusion that the TIP60 complex has a specific function in controlling PN dendrite targeting. TIP60 complex mutants show discrete glomerular mistargeting, rather than randomly distributed dendrite spillover to different glomeruli. In contrast, perturbation of individual cell surface receptors often leads to variable mistargeted dendrites that do not necessarily obey glomerular borders, possibly reflecting the combinatorial use of many cell surface effector molecules. Even transcription factor mutants yield variable phenotypes. Interestingly, BRM complex mutants yield non-stereotyped phenotypes in PNs. No stereotyped glomerular targeting was seen for brm or Snr1 mutant dendrites; each PN spreads its dendrites across different glomeruli. These data suggest that different chromatin remodeling complexes play distinct roles in regulating neuronal differentiation. The uni- or bi-glomerular targeting to specific glomeruli implies that the TIP60 complex sits at the top of a regulatory hierarchy to orchestrate an entire transcriptional program of regulation (Tea, 2011).

This study suggests a function for Bap55 in Drosophila olfactory PN development as a part of the TIP60 complex rather than the BRM complex. Another possibility could be that Bap55 also serves as the interface between the BRM and TIP60 complexes. While loss of core BRM complex components results in a more general defect, loss of Bap55 could specifically disrupt interactions with the TIP60 complex but maintain other BRM complex functions, causing a more specific targeting phenotype mimicking loss of TIP60 complex components (Tea, 2011).

Interestingly, both human BAF53a and b can significantly rescue the Bap55^-/- phenotype. Though in mammals BAF53a is expressed in neural progenitors and BAF53b is expressed in postmitotic neurons, they can perform the same postmitotic function in Drosophila PNs. Further, both can function with the TIP60 complex in PNs to regulate wiring specificity. These data suggest that the functions for BAF53a and b (in neural precursors and postmitotic neurons, respectively) diverge after the evolutionary split between vertebrates and insects (Tea, 2011).

The discrete glomerular states of the mistargeting phenotypes may suggest a role for the TIP60 complex upstream of a regulatory hierarchy determining PN targeting decisions. It is possible that disrupting various components changes the composition of the complex. Additionally, overexpression of Bap55 in various mutant backgrounds might alter the sensitive stoichiometry of the TIP60 complex, resulting in targeting to different but still distinct glomeruli (Tea, 2011).

Several mutants have been identified that cause DL1 PNs to mistarget to areas near the DM6 glomerulus (Tea, 2010). Interestingly, WT DM6 PNs have the most similar lateral horn axon arborization pattern to DL1 PNs. It is hypothesized that the transcriptional code for DM6 is similar to that of DL1, which is at least partially regulated by the TIP60 complex. The genes described in this manuscript are the only mutants that have yielded specific DA4l mistargeting to date. It is possible that the targeting 'code' for DA4l, DL1, and DM6 may be most similar, such that perturbation of the TIP60 complex might result in reprogramming of dendrite targeting. PNs have previously been shown to be pre-specified by birth order. Yet DA4l is born in early embryogenesis, DL1 is born in early larva, and DM6 is born in late larva. This implies that the TIP60 transcriptional code does not correlate with PN birth order. The mechanisms by which the TIP60 complex specifies PN dendrite targeting remain to be determined (Tea, 2011).

This study has characterize PN phenotypes of mutants in the BRM and TIP60 complexes, with a focus on Bap55, which is shared by the two complexes. The TIP60 complex was found to play a very specific role in regulating PN dendrite targeting; mutants mistarget from the DL1 to the DA4l glomerulus. This specific mistargeting phenotype suggests that TIP60 controls a transcriptional program important for making dendrite targeting decisions (Tea, 2011).

GENE STRUCTURE

The E(Pc) gene is located upstream of invected and is transcribed in the same direction (Stankunas, 1998).

Bases in 5' UTR - 1.1 kb

Exons - 7

Bases in 3' UTR - 1.4 kb

PROTEIN STRUCTURE

Amino Acids - 2023

Structural Domains

The protein has an estimated size of 220 kDa, making it the largest member of the PcG characterized so far. It contains many charged residues and has an estimated pI of 5.79. Two of the basic amino acid-rich sequences, amino acids 325-KKRKHK-330 and 675- KRRRLRRKK-683 are probable nuclear localization signals. E(Pc) is similar to a number of cloned PcG proteins because of the presence of multiple regions enriched in specific amino acids. E(Pc) contains 7 glutamine-rich regions, some of which are perfect repeats, the longest being 12 consecutive glutamines. Glutamine repeats are also found in Polyhomeotic (ph) and Additional sex combs (Asx) (Sinclair, 1998b). Glutamine repeats have been implicated in protein-protein interactions and in transcriptional activation. The 18 amino acid sequence from aa 835-852 contains 15 alanines. Alanine-rich motifs are found in Asx (Sinclair, 1998b), Sex comb on midleg and on a new member of the PcG, Cramped. Alanine-rich regions have been implicated in repression by transcription factors, but their function in PcG proteins is unknown. E(Pc) contains two arginine-rich sequences at aa 780-793 and 1976-1982, a feature also found in the carboxy terminus of Su(z)2. E(Pc) contains a putative leucine zipper at amino acids 644-672. Leucine zippers form coiled coils. Analysis of E(Pc) predicts a coiled coil between amino acids 651 and 690, consistent with the leucine zipper being functional. There are two additional coiled coils predicted to occur between amino acids 208 and 258, and 1633 and 1630 (Stankunas, 1998).

EVOLUTIONARY HOMOLOGS

Drosophila Enhancer of Polycomb, E(Pc), is a suppressor of position-effect variegation and an enhancer of both Polycomb and trithorax mutations. A homologous yeast protein, Epl1, is a subunit of the NuA4 histone acetyltransferase complex. Epl1 depletion causes cells to accumulate in G2/M and global loss of acetylated histones H4 and H2A. In relation to the Drosophila protein, mutation of Epl1 suppresses gene silencing by telomere position effect. Epl1 protein is found in the NuA4 complex and a novel highly active smaller complex named Piccolo NuA4 (picNuA4). The picNuA4 complex contains Esa1, Epl1, and Yng2 as subunits and strongly prefers chromatin over free histones as substrate. Epl1 conserved N-terminal domain bridges Esa1 and Yng2 together, stimulating Esa1 catalytic activity and enabling acetylation of chromatin substrates. A recombinant picNuA4 complex shows characteristics similar to the native complex, including strong chromatin preference. Cells expressing only the N-terminal half of Epl1 lack NuA4 HAT activity, but possess picNuA4 complex and activity. These results indicate that the essential aspect of Esa1 and Epl1 resides in picNuA4 function. It is proposed that picNuA4 represents a nontargeted histone H4/H2A acetyltransferase activity responsible for global acetylation, whereas the NuA4 complex is recruited to specific genomic loci to perturb locally the dynamic acetylation/deacetylation equilibrium (Boudreault, 2003).

There appear to be two E(Pc) paralogs in mammals, which have been named EPC1 and EPC2 in humans, and Epc1 and Epc2 in mice. A mouse EST clone containing sequences homologous to E(Pc) was used to screen an embryonic cDNA library: a 3.9 kb cDNA was recovered and termed Epc1-L. It contains 5' and 3' untranslated sequences, including a poly(A) tail, and a 2.3 kb ORF. Epc1-L encodes a protein of only 764 amino acids, about a third of the length of the Drosophila homolog. Another clone is a 1.65 kb cDNA, which contains 825 bp identical to Epc1-L, plus 109 bp not found in Epc1-L, and a downstream sequence that is identical to Epc1-L for the remainder of the clone. These divergent sequences probably correspond to alternatively spliced exons, although the genomic structure has not been determined to confirm this. The shorter splice variant has been termed Epc1-S. At the location corresponding to the insertion of the 109 bp in Epc1-S, Epc1-L contains 769 bp not found in Epc1-S. Both the 769 and 109 bp sequences are open through their entire length, but Epc1-L and Epc1-S use different reading frames downstream, even though their nucleotide sequences are identical. The result is that Epc-S terminates much earlier than Epc-L, to yield a protein of 344 amino acids. Another clone has also been sequenced, a short clone from Epc2 (Stankunas, 1998).

Comparison of E(Pc) with the Epc1-S sequence reveals three regions of sequence similarity termed EPcA-C, respectively. EPcA is an amino terminal region of 266 amino acids, which is 51% identical and 71% similar to the equivalent mouse domain. Within EPcA is a stretch of 54 amino acids in which 46 amino acids are identical to the equivalent mouse sequence and 50/54 amino acids are similar. Interestingly, this highly conserved sequence contains a consensus tyrosine phosphorylation sequence 214-RKNDEASY-221. The EPcA domain is hydrophilic and contains 34% charged residues, as opposed to 23% in the protein as a whole. The domain is also strikingly rich in methionine and tyrosine (6% and 4.15%, respectively), when compared to methionine and tyrosine in the protein as a whole (1.8% and 1.9% respectively). Structural analysis predicts that the EPcA domain in flies and mice contains three alpha helices in the regions of amino acids 5-55, 90-155 and 200-260. EPcB is an interior domain of 102 amino acids (52% identical and 64% similar in flies and mice) that contains 15% arginine and 10% serine residues. EPcC contains just 24 amino acids (54% identical and 75% similar) and is unremarkable except for its conservation over a long evolutionary distance. Epc-L contains a short glutamine repeat, but it does not contain the alanine or the arginine repeats found in E(Pc). Interestingly, Epc1-S contains an alanine repeat but does not contain EPcB or EPcC. Perhaps Epc1-L and Epc1-S have acquired different functions. Epc2 initiates at the same methionine used in E(Pc), unlike Epc1 which initiates at a methionine internal to that of E(Pc). Neither Epc1-L nor Epc1-S contains a putative leucine zipper (Stankunas, 1998).

To determine if EPC1 or EPC2 co-map with known human mutations that affect growth or differentiation, the cytological location of EPC1 and EPC2 was determined using fluorescence in situ hybridization (FISH). EPC1 maps to 10p11-12 and EPC2 to 22q13.3. EPC1 also localizes to region 10p11.2-12 on the human transcript map. No transcripts match the EPC2 sequence in the same database. These data show that EPC1 and EPC2 are distinct genes. However, no mutations that map to these regions have phenotypes affecting cell growth or determination (Stankunas, 1998).

E(Pc) homologs were sought in yeast: Caenorhabditis elegans databases and matches were found in both organisms. The EPcA domain is conserved [26% identity, 46% similarity when compared to E(Pc)] in YFL024C, a hypothetical yeast protein proposed to encode a 96.7 kDa protein of no known function, that has been rename EPL1, for Enhancer of Polycomb-like. EPL1, like E(Pc) contains glutamine repeats and an alanine repeat. Some of the EPcA domain is conserved (42% identity, 52% similarity over a 156 amino acid sequence) in a C. elegans EST, referred to as cEPc. The cEPc protein also contains the EPcB domain. When the EPcA domain was used in a BLAST search to rescreen the databases, another human protein, called BR140 (and also termed peregrin) was recovered (Thompson, 1994). BR140 contains only some of the EPcA domain. The conserved region shows 33% identity over a 64 amino acid region. Interestingly, BR140 contains a bromodomain, found in the trithorax Group gene brahma, as well as in other highly conserved proteins required for transcriptional activation, and a PHD domain, a cysteine cluster found in many proteins including trithorax and Polycomblike. Therefore, BR140 contains three separate domains shared with different Drosophila proteins implicated in gene regulation and chromatin. The structure of BR140, which shows significant conservation of a short sequence within EPcA suggests that EPcA is modular, and that different parts of the sequence have different functions. A modular organization for EPcA is also supported by the different spacing between conserved regions of EPL1 and those of higher eukaryotes (Stankunas, 1998).

The full-length murine Epc1 cDNA, which should hybridize to both splice variants, was used to probe Northern blots prepared from adult mouse tissues and from embryonic stages. A complex pattern of hybridization was seen. In adult tissues,there appear to be two mRNAs of 4.0 and 2.6 kb, plus one other smaller mRNA that varies in size from 1.3-2.0 kb, depending on the tissue. The 4.0 kb transcript is the major transcript in most tissues, but the 4.0 and 2.6 kb transcripts appear to be regulated independently, as can be seen by comparing amounts of these transcripts in liver or skeletal muscle with kidney. Epc1 is expressed in all tissues tested except spleen, and is present throughout embryonic development, although there are additional mRNAs of higher molecular weight expressed in embryogenesis. While the 4.0 and 2.6 kb transcripts likely represent Epc1-L and Epc1-S, respectively, the possibility cannot be ruled out that Epc2 transcripts or transcripts from uncharacterized loci are also detected (Stankunas, 1998).

Home page: The Interactive Fly © 1995, 1996 Thomas B. Brody, Ph.D.

The Interactive Fly resides on the
Society for Developmental Biology's Web server.