Regulation of Gonad Morphogenesis in Drosophila melanogaster by BTB Family Transcription Factors

During embryogenesis, primordial germ cells (PGCs) and somatic gonadal precursor cells (SGPs) migrate and coalesce to form the early gonad. A failure of the PGCs and SGPs to form a gonad with the proper architecture not only affects germ cell development, but can also lead to infertility. Therefore, it is critical to identify the molecular mechanisms that function within both the PGCs and SGPs to promote gonad morphogenesis. We have characterized the phenotypes of two genes, longitudinals lacking (lola) and ribbon (rib), that are required for the coalescence and compaction of the embryonic gonad in Drosophila melanogaster. rib and lola are expressed in the SGPs of the developing gonad, and genetic interaction analysis suggests these proteins cooperate to regulate gonad development. Both genes encode proteins with DNA binding motifs and a conserved protein-protein interaction domain, known as the Broad complex, Tramtrack, Bric-à-brac (BTB) domain. Through molecular modeling and yeast-two hybrid studies, we demonstrate that Rib and Lola homo- and heterodimerize via their BTB domains. In addition, analysis of the colocalization of Rib and Lola with marks of transcriptional activation and repression on polytene chromosomes reveals that Rib and Lola colocalize with both repressive and activating marks and with each other. While previous studies have identified Rib and Lola targets in other tissues, we find that Rib and Lola are likely to function via different downstream targets in the gonad. These results suggest that Rib and Lola act as dual-function transcription factors to cooperatively regulate embryonic gonad morphogenesis.


Introduction
in the embryonic central nervous system [19]. While Lola is hypothesized to function primarily as a transcriptional repressor, expression of copia appears reduced in the embryonic gonad of lola mutants [19]. These results suggest that Lola isoform function may be determined in part by its binding partners, or that different Lola isoforms regulate gene expression in different tissues. Gene expression profiling of lola null mutant embryos compared to controls has resulted in the identification of hundreds of genes that exhibit increased or decreased expression upon lola mutation, although the direct transcriptional regulation of these targets has not yet been demonstrated [33].
Unlike lola, which has at least 20 protein isoforms, the rib gene locus encodes three protein isoforms. All Rib isoforms have a DNA binding motif located C-terminal to the BTB domain that is referred to as the Pipsqueak (PSQ) motif ( Fig 1B) [15,16]. The PSQ motif is a 50-amino acid sequence that binds to DNA [34,35]. Recent work has resulted in the identification of multiple Rib transcriptional targets in the salivary gland, some of which are repressed while others are activated [36]. These results suggest that Rib transcriptional activity may also depend on its interaction with different binding partners.
Both rib and lola were previously identified in genetic screens for mutants that affect early gonad formation [37,38], and have been shown to function in other contexts to regulate organ morphogenesis. The rib gene was first identified in an ethyl methanesulfonate mutagenesis screen for larval cuticle abnormalities [39]. Subsequently, rib was shown to regulate cell migration and morphogenesis of the trachea and salivary glands [15,16,40]. When rib is mutated, the tracheal cell bodies and apical surface have severe migration defects and impaired morphogenesis, while the salivary glands fail to elongate [15,16]. Recent work has demonstrated that Rib is required to promote the changes in cell shape and volume needed for salivary gland morphogenesis [36]. Molecular studies of the downstream effectors of Rib are consistent with these cellular functions. rib mutants exhibit reduced levels of Crumbs (Crb), an apical membrane protein that functions in epithelial cell polarity and in apical membrane growth [41,42]. Rib also regulates the activity of Moesin (Moe), the only Ezrin-Radixin-Moesin (ERM) family protein in Drosophila, which plays a role in linking the plasma membrane to the actin cytoskeleton [42]. In rib mutants, levels of active, phosphorylated Moe are increased, suggesting that down-regulation of Moe activity is required for salivary gland and tracheal morphogenesis [42]. Finally, levels of the recycling endosomal protein Rab11 were reduced in rib mutants, consistent with an observed decrease in the number of apically-localized vesicles [42]. In the context of salivary gland and tracheal morphogenesis, Rib has also been demonstrated to genetically interact with another BTB protein, Lola like (Lolal) [42]. Given that Rib interacts with other BTB domain containing proteins such as Lolal, it is possible that Rib may interact with other BTB family proteins in a tissue-specific manner to regulate development of other tissues.  46.38/22.05 embryonic gonads exhibiting fusion (D) and compaction (E) defects. (E-F) rib 35.14/55.25 embryonic gonads exhibiting (F) fusion and (G) compaction defects. (H) Quantification of the frequency of gonad phenotypes observed in control, lola mutant and rib mutant embryos. Gonad morphology was assessed using the 68-77-lacZ enhancer trap. The following phenotypes were scored: fusion (red), compaction (blue) and wild-type (green). A chisquare test was performed to test the null hypothesis that the phenotypic ratios will be the same across all genotypes.
One candidate for this interaction is Lola. Previous studies have demonstrated a critical role for Lola in Drosophila nervous system development. lola is required for axon growth in the embryonic central nervous system (CNS), and mutation of lola results in axon pathfinding defects along the longitudinal tracts of the CNS [43]. In addition, Lola is required to prevent midline crossing of longitudinal axons [44]. In this context, Lola appears to up-regulate expression of the midline repellent protein, Slit, and its longitudinal axonal receptor Roundabout (Robo) [28,44]. Previous studies have demonstrated that Lola is also required for embryonic gonad morphogenesis and for gametogenesis in the adult [22,37,38,45].
Here, we have further characterized the role of Lola and Rib in embryonic gonad morphogenesis. Using genetic and physical interaction analysis, we have examined the relationship between Rib and Lola and our results suggest that they cooperate to promote gonad morphogenesis. Molecular modeling suggests that both Rib and Lola are capable of homo-and heterodimerization via their BTB domains. The colocalization of Rib and Lola with marks of transcriptional activation and repression and with each other at a subset of sites on polytene chromosomes suggests that Rib and Lola function as dual-function transcriptional regulators to regulate tissue morphogenesis.

Yeast-two hybrid interaction
The yeast strains Y2H-Gold and Y187 were transformed with pGADT7 and pGBKT7 vectors, respectively, by standard Lithium acetate-Tris-EDTA transformation and transformants were selected on SD-Leu or SD-Trp plates. The yeast two-hybrid was performed according to the Matchmaker 1 Gold Yeast Two-Hybrid System User Manual (Clontech) with a few modifications. For yeast mating, individual yeast colonies were inoculated into 5 ml of YPDA in 250 ml Erlenmeyer flasks in the morning. In the mid/late afternoon, yeast growth was measured by a spectrophotometer to ensure similar culture density. The desired mating combinations were mixed with equal amounts of media in a 500 ml Erlenmeyer flask and then the volume was brought up to~10 ml for incubation overnight at 30˚C. After 20 hours, cultures were examined for the presence of zygotes in the media and plated on SD-Trp-Leu media to select for successful mating. Following successful mating, colonies were streaked on SD-Leu-Trp-His-Ade plates supplemented with X-α-Gal and Aureobasidin A for yeast two-hybrid analysis.

Protein modeling
Evolutionarily related structures for the sequences of Lola and Rib were found by using Blast [56] and HHBlits [57] to search the SWISS-MODEL template library (SMTL version 2016-03-30, PDB release 2016-03-25) [58]. Homology models for the BTB domain dimers of Lola and Rib were constructed using SWISS-MODEL [59] using the crystal structure of the human Bach2 POZ domain (PDBid: 3OHV) [60] as a template. The initial dimer for Lola-Rib was modeled using one subunit from each of the homodimers. The three initial dimers were then refined using a combination of energy minimization and molecular dynamics. Each simulation box, containing one dimer, a TIP3 water box extending at least 10 Å beyond the protein in all directions and 0.15 M NaCl adjusted to neutralize the charge in the water box, was assembled using the molecular graphics program VMD [61]. The simulation box was then brought to equilibrium using the molecular dynamics program NAMD [62]. The equilibration procedure involved energy minimization with and without restraints on the protein coordinates (3000 steps each), slow heating from 10 to 310 K (30,000 steps), and then pressure and temperature equilibration using a Langevin piston (10,000 steps). Finally, unrestrained dynamics for 100,000 steps was done before data was acquired. The time step was 2 fs with every 150th step being saved in the trajectory for analysis. Periodic boundary conditions were used. The cutoffs for nonbonding (van der Waals and electrostatic) interactions were 12 Å. The switch distance was 10 Å, and 1.0 1-4 scaling factor was used. All calculations were done using CHARMM 27 parameters [63]. The van der Waals and electrostatic dimer interaction energy values presented in Table 1 were determined for the final 100 steps (30 ps) in each simulation and then averaged. All molecular graphics diagrams were generated using VMD [61].

Characterization of the role of Rib and Lola in gonad morphogenesis
Previously, lola and rib were identified in screens for genes that are required for embryonic gonad morphogenesis [37,38]. In order to further characterize the gonad phenotypes of rib and lola mutants, the 68-77-lacZ enhancer trap, which is expressed in the cytoplasm of the SGPs, was used to mark the cytoplasm of SGPs and examine gonad morphology ( Fig 1C) [4,5,46]. During normal gonad development, SGP clusters fuse into an elongated, coalesced gonad by the end of stage 13, and PGCs and SGPs compact into a spherical gonad with SGPs ensheathing the PGCs by the end of stage 15 (Fig 1A and 1C). Therefore, we scored rib and lola embryonic gonads for a failure of SGP clusters to coalesce, referred to as fusion defects, and a failure of gonads to form a round spherical gonad, referred to as compaction defects.
Immunostaining of SGPs of the gonad in both rib and lola mutants revealed defects in SGP cluster fusion and gonad compaction (Fig 1D-1H; S1 Fig). Quantification of these phenotypes demonstrates that the control embryos, which carry the 68-77-lacZ enhancer trap, exhibit a low frequency of fusion and compaction defects of~7% ( Fig 1H). Less than 17% of the gonads in embryos heterozygous for rib and lola mutations exhibited fusion and compaction defects ( Fig 1H). For analysis of rib and lola homozygous mutants, heteroallelic combinations were used to minimize the potential contribution of second site mutations to the observed phenotype, as these mutant alleles were obtained from a mutagenesis screen [37]. Four lola alleles were used in these studies: mutations in lola ORE76 and lola 22.05 (Q97STOP) are predicted to affect all Lola protein isoforms [28,38], while lola 46.38 encodes a H844L mutation in the second the zinc finger of Lola isoform T (Flybase designation PR/PG) and lola ORC4 encodes a premature stop codon (P771STOP) in Lola isoform K (Flybase designation PI) ( Fig 1B). We observed defects when each isoform-specific allele was combined with a null or hypomorphic allele affecting all isoforms ( Fig 1D, 1E and 1H). In the case of the lola 46.38/22.05 and lola 22.05/ORE76 mutants, fusion and compaction defects were observed in just under 40% of gonads ( Fig 1H). A small increase in the frequency of gonad defects to 44% was observed in lola 46.38/ORE76 mutants ( Fig 1H), suggesting that despite an early stop codon in lola 22.05 it may not be as strong of a loss of function mutant as lola ORE76 . lola 22.05/ORC4 mutants carrying one copy of the Lola isoform K-specific mutation in combination with a mutation affecting the Lola common region exhibited gonad defects 34% of the time ( Fig 1H). Thus, all heteroallelic lola mutants exhibited an increase in fusion and compaction defects when compared to their heterozygous counterparts. These results also suggest that multiple Lola isoforms, namely Lola-T and Lola-K, function in gonad morphogenesis.
In the case of the rib alleles, only the rib 1 allele encodes a protein with a premature stop codon (R283STOP) [15], while rib 35.14 encodes a protein with a D406V missense mutation, which is localized to the PSQ DNA binding motif and is likely to function as a hypomorph ( Fig 1B). While the precise mutation in the rib 55.25 allele is unknown, sequencing has revealed that it is not within the coding sequence. Examination of rib 35.14/55.25 mutants reveals the presence of fusion and compaction defects in approximately 52% of embryonic gonads (Fig 1F-1H). rib 55.25/1 heteroallelic mutants exhibited the highest frequency of defects with a total of 61% ( Fig 1H). Thus, the rib 55.25 mutant allele is likely a stronger allele than the rib 35.14 allele. Overall, these results suggest that Rib and Lola are critical regulators of gonad coalescence and compaction.
Previous studies also demonstrated a requirement for Lola in germ cell migration, as germ cells were observed outside of the coalesced gonad [38]. Therefore, we scored the number of extragonadal germ cells in rib and lola mutant embryos. While controls had an average of less than one extragonadal germ cell per embryo, rib heterozygotes had a slight increase to an average of 1.

Lola and Rib function in the SGPs to promote gonad development
The fusion and compaction defects observed in the gonad suggest that these proteins function in the SGPs during gonad morphogenesis, while germ cell migration defects suggest they may also be required in the PGCs. Therefore, immunohistochemistry for Rib and Lola proteins was performed to determine where Rib and Lola are expressed in the developing gonad. Antibody staining for Lola with an antibody that recognizes all Lola isoforms demonstrates that Lola is expressed in the SGPs and PGCs and the surrounding mesoderm in stage 13 embryos and and primarily in the SGPs and surrounding mesodermal cells in 15 embryos (Fig 2A and 2B). As our genetic analysis suggested Lola isoforms T and K may function in gonad morphogenesis, we also examined expression of these isoforms using Lola-T and Lola-K GFP fusion proteins. Antibody staining for Rib revealed expression of Rib in the SGPs and no staining in the PGCs in stage 13 and 15 embryos (Fig 2C and 2D). Given the similar phenotypes observed in rib and lola mutants and similar expression patterns in the SGPs of the embryonic gonad, the colocalization of both of these proteins was examined. Immunohistochemistry analysis reveals the colocalization of Lola and Rib in SGPs in stage 13 and 15 embryos (Fig 2E and 2F), suggesting that Lola and Rib are both required in the SGPs to promote gonad morphogenesis. Rescue experiments previously demonstrated that overexpression of Lola in the mesoderm could rescue lola mutant gonad phenotypes [38]. Therefore, the ability of Rib overexpression in the mesoderm to rescue the rib mutant phenotype was tested. Results demonstrate that Rib overexpression in the mesoderm rescues the rib mutant phenotype (S4 Fig).
Because lola and rib expression is also observed in the surrounding mesoderm, it was necessary to eliminate the possibility that the gonad defects were due to a more generalized defect in mesoderm development. In order to look at mesoderm development, stage 12 embryos were immunostained with anti-Fasciclin 3 to examine visceral mesoderm development (Fig 3). Visceral mesoderm development was indistinguishable from controls for both lola and rib mutant embryos (Fig 3), demonstrating that mesoderm development was not globally disrupted in rib and lola mutants. Thus, expression data suggests Rib and Lola function specifically in the SGPs to promote gonad morphogenesis.

rib and lola genetically interact
With the observations that both rib and lola exhibit similar mutant defects and are both expressed in the SGPs, we explored the possibility these proteins may cooperate to regulate gonad morphogenesis through genetic interaction studies. Gonad development was examined in embryos heterozygous for both rib and lola alleles. Stage 15 embryonic gonads were scored as wild-type, or as having fusion or compaction defects. The frequency of defects in double heterozygotes was compared to control embryos, as well as to embryos heterozygous for either rib or lola (lola 46.38/+ , rib 35.14/+ , lola 22.05/+ , and rib 55.25/+ ) (Fig 4). The percentage of defective gonads ranged from 8-17% in the controls (Fig 4). In the case of embryos heterozygous for both rib and lola (lola 46.38/+ , rib 55.25/+ and lola 22.05/+ , rib 35.14/+ ), the frequency of gonad defects increased significantly to between 65-85% (Fig 4), demonstrating a synergistic effect and suggesting that rib and lola cooperate to regulate gonad morphogenesis. In addition, heteroallelic mutants for rib or lola were examined that were also heterozygous for lola or rib, respectively (lola 22.05/+ , rib 35.14/55.25 and lola 46.38/22.05 , rib 55.25/+ ). These mutants showed a dramatic increase in gonad defects relative to rib and lola mutants alone with more than 85% of the gonads exhibiting fusion and compaction defects, compared to~50% in rib and~40% in lola heteroallelic combinations (Fig 4). These results suggest that Rib and Lola function cooperatively in the same pathway or in parallel pathways to regulate embryonic gonad development.

Rib and Lola do not regulate expression of each other
The genetic interaction of rib and lola led us to hypothesize that Rib and Lola could function (1) in a stepwise fashion in a single pathway; (2) in parallel pathways to impinge on a common set of downstream targets; or (3) as part of a complex to regulate a common set of target genes to promote gonad morphogenesis. Therefore, we first explored the possibility that Rib and Lola regulate expression of one another in a stepwise fashion by examining expression of Lola in rib mutants and expression of Rib in lola mutants. Expression of Lola in rib heterozygotes versus rib heteroallelic mutants revealed no difference in Lola expression, suggesting that Rib does not regulate Lola expression levels (Fig 5A and 5B). Similarly, Rib expression was compared in lola heterozygotes and lola heteroallelic mutants, and Rib expression was unchanged (Fig 5C and 5D). These results suggest that Rib and Lola are not regulating expression of each other, but rather function in parallel pathways or as a complex to regulate gene expression during embryonic gonad development.

Rib and Lola physically interact via their BTB domains
Given the observations that rib and lola show a strong genetic interaction, co-localize in the SGPs, and both contain a BTB domain, we next tested the possibility that these proteins may physically interact to regulate embryonic gonad development. Homology modeling and molecular dynamics simulations were used to predict the potential interaction between Lola and Rib BTB domains (Fig 6). Interaction energy calculations (Table 1) suggest that Rib and Lola have the ability to heterodimerize as well as homodimerize.
Therefore, we tested the ability of the Lola and Rib BTB domains to interact physically by performing a yeast two-hybrid assay. The Rib and Lola BTB domains were each fused to the yeast GAL4 DNA binding domain (BD) and the GAL4 activation domain (AD). Positive and negative controls, as well as strains expressing BD-RIB and AD-Lola, AD-Rib and BD-Lola, AD-Rib and BD-Rib, and AD-Lola and BD-Lola were successfully mated and found to be viable when carrying all fusion protein combinations (Fig 7A and 7A'). The ability of AD and BD fusion proteins to interact were tested using four reporters under the control of the GAL4 upstream activating sequence: His3, Ade2, LacZ, and the AUR1-C gene, which confers resistance to Aureobasidin A (Fig 7B and 7B'). The positive control (Fig 7B #1) exhibited activation of all reporters, while the negative control (Fig 7B #2) failed to grow on the selection plates.

Regulators of Gonad Morphogenesis
Lola-BD and Rib-BD fusions were tested with the GAL4-AD alone and failed to activate reporters (Fig 7B' #7-8), demonstrating that Rib and Lola BTB domains do not autoactivate reporters. Additionally, Lola-AD and Rib-AD fusions were also tested in combination with the GAL4-BD alone and failed to activate reporters (Fig 7B' #9-10), demonstrating that the Lola and Rib BTB domains do not bind to the reporters or the GAL4-BD nonspecifically. Mating of BD-Lola with AD-Rib and BD-Rib with AD-Lola resulted in activation of all reporters (Fig 7B'  #5-6). In addition, we tested the ability of the Rib and Lola BTB domains to form homodimers, and observed homodimerization of both Rib and Lola BTB domains, with a more robust Regulators of Gonad Morphogenesis interaction for Lola than for Rib (Fig 7B #3-4). These results are consistent with molecular modeling, which predicted stronger interactions for Lola-Lola homodimers than for Rib-Rib homodimers (Table 1). These results demonstrate that Rib and Lola BTB domains are capable of homo-and heterodimerization, and suggest that Rib and Lola form a complex to cooperatively regulate gene expression during gonad morphogenesis rather than simply functioning in parallel.

Rib and Lola are transcriptional activators and repressors
While earlier studies suggested that Rib and Lola function as transcriptional repressors, more recently it has been suggested that Rib and Lola also function in transcriptional activation [19,36,42]. In order to further examine the roles of Rib and Lola as transcriptional activators and/ or repressors, we examined the localization of the Rib and Lola protein with marks of transcriptional activation and repression on polytene chromosomes. First, Rib and Lola expression were examined in combination with immunostaining for H3K27me3, a mark of transcriptional repression. Results demonstrated that Lola colocalized with H3K27me3 on numerous sites on the chromosome (Fig 8A). As the Rib antibody did not exhibit strong staining on polytene chromosomes, we expressed a 3xHA-tagged Rib in salivary glands using forkhead-Gal4. We observed colocalization of 3xHA-Rib and H3K27me3 at numerous sites on polytene chromosomes (Fig 8C). While Lola and Rib colocalize with H3K27me3 at a number of loci on polytene chromosomes, the presence of Lola and Rib at other distinct sites suggests that Rib and Lola may also localize at sites of transcriptional activation. In order to determine if Lola and Rib may function in transcriptional activation, we examined Rib and Lola colocalization with sites of RNA Polymerase II phosphoserine 5 (PolIIser5). Lola and Rib also colocalized with PolIIser5 at many sites on polytene chromsomes (Fig 8B and 8D). These results suggest that Rib and Lola have dual functions as transcriptional activators and repressors. As our earlier results suggest that Rib and Lola may coregulate transcription of target genes, we examined colocalization of 3xHA-Rib and Lola on polytene chromosomes. While Lola and Rib colocalize at some loci, Lola staining on polytene chromosomes is significantly reduced upon overexpression of 3xHA-Rib, and there are a number of loci to which only Rib localizes (Fig 8E). These results suggest that Rib does not require Lola to localize to polytene chromosomes. In addition, as Lola staining was present at more sites on the polytene chromosomes when endogenous Rib was not observed, it does not appear that Rib is required to recruit Lola. The reduction of Lola staining on chromosomes upon overexpression of 3xHA-Rib suggests that the 3xHA tag may affect the ability of Rib to dimerize or that overexpressed Rib may result in an increased population of Rib homodimers at the expense of Rib-Lola heterodimers. The continued colocalization of Rib and Lola at some sites suggests that Rib and Lola cooperate to regulate transcription of a subset of target genes, while functioning independently of each other to regulate other targets.

Expression analysis of Rib and Lola transcriptional targets
Previous studies have resulted in the identification of a number of Rib and Lola downstream targets in other tissues. Expression profiling and genetic analysis has demonstrated that Lola Regulators of Gonad Morphogenesis promotes expression of Slit and Robo in the developing central nervous system and inhibits expression of the cytoskeletal regulator Spire (Spir). Both Slit and Robo have been demonstrated to function in gonad morphogenesis; however, only Robo has been localized to the gonad by immunostaining [37]. Therefore, we examined expression of Robo and Spir in lola null homozygotes, and compared expression to lola heterozygotes. Immunostaining for Robo revealed that mutation of lola does not alter Robo expression in the gonad (S5A and S5B Fig).
These results are consistent with previous observations that Lola does not regulate Robo expression in the gonad [38]. Next we examined expression of Spir in the gonad. Antibody staining reveals little enrichment of Spir in the developing gonad, and its expression does not appear to be altered in Lola homozygotes, as compared to heterozygous controls (S5C and S5D Fig). These results suggest that Lola regulates different targets in a tissue-specific manner.
In the case of Rib, chromatin immunoprecipitation and gene expression analysis have resulted in the identification of more than a dozen Rib transcriptional targets in the salivary gland, including crb, which encodes an apical membrane protein involved in epithelial polarity. In addition, Moesin (Moe), the sole Ezrin-Radixin-Moesin protein in Drosophila, was also found to exhibit increased phosphorylation upon rib mutation. Therefore, expression of crb and activation phospho-Moesin (pMoe) were examined in rib mutant embryos. Analysis of crb expression in the gonad reveals that crb is not highly expressed in the developing gonad, and its expression levels do not change in rib homozygotes relative to heterozygous controls (S6A and S6B Fig). While pMoe is present in the gonad at the PGC membrane, it does not appear in the SGPs and its expression also remains unchanged upon rib mutation (S6C and S6D Fig). These results suggest that Rib functions through different downstream targets in the gonad as compared to the salivary gland.

Discussion
In this study we demonstrated the requirement for two BTB family transcription factors, Rib and Lola, for embryonic gonad morphogenesis. Both rib and lola are expressed in the SGPs of the gonad, and rib and lola mutants exhibit defects in SGP cluster fusion and gonad compaction. Genetic analysis revealed that these genes interact with each other in the context of gonad development, while molecular modeling and yeast two-hybrid studies demonstrate the ability of these proteins to interact physically via their BTB domains. We show that Rib and Lola colocalize with marks of transcriptional activation and repression, suggesting that these proteins act as dual-function transcriptional regulators to promote tissue morphogenesis. Finally, we find that Rib and Lola do not regulate the same downstream targets in the gonad as those previously identified in other developmental contexts.
Rib and Lola were identified in screens for mutations that affect embryonic gonad morphogenesis [37,38]. The defects in SGP cluster fusion and gonad compaction that we and others have observed, as well as previous work demonstrating SGP cluster fusion and compaction do not require PGCs, suggest that Rib and Lola function specifically in the SGPs to regulate gonad development [1,37,38,64]. Consistent with the mutant phenotypes, both proteins are strongly expressed in the SGPs; however, they are also expressed strongly in the surrounding mesoderm and in the case of Lola, weakly in the PGCs (Fig 2). The proper specification of other mesodermally derived tissues, like the visceral mesoderm (Fig 4), suggests that the gonad defects observed in rib and lola mutants are due to a specific requirement for these proteins in the SGPs. Supporting this hypothesis, previous work demonstrated that overexpression of lola in the mesoderm can rescue the lola mutant phenotype [38], and we find that overexpression of Rib in the mesoderm can rescue the rib mutant phenotype (S4 Fig). Interestingly, extragonadal PGCs were also observed in many rib and lola mutant embryos (S2 Fig), suggesting that Rib and Lola may regulate the migration of PGCs and/or their ability to interact with SGPs cell autonomously or non-autonomously. As Rib expression is not observed in PGCs (Fig 2C and  2D), and Lola rescue experiments indicated that Lola mesodermal expression is sufficient to rescue the lola mutant phenotype [38], we favor the hypothesis that regulation of PGC migration by Rib and Lola must be cell non-autonomous.
Identification of the downstream targets through which Rib and Lola function to regulate cell-cell interactions is critical for understanding how these proteins regulate tissue morphogenesis. Both Rib and Lola contain DNA binding domains, with Rib containing a PSQ motif, and many Lola isoforms containing at least one zinc finger domain. Earlier studies predicted Lola and Rib to function as transcriptional regulators [15,16,28], which has subsequently been confirmed [19,36]. In the case of Lola, numerous putative targets of Lola have been identified in the central nervous system, including the cytoskeletal regulator Spir, as well as members of the Slit/Robo signaling pathway [33,43]. Consistent with a previous study, we observe that robo expression does not appear to be regulated by Lola in the gonad [37,38] (S5A and S5B Fig). In addition, our results also demonstrate that Lola does not regulate spir expression to promote gonad morphogenesis (S5C and S5D Fig). These results suggest that Lola is likely to regulate different target genes in different tissues. In the case of the only confirmed direct transcriptional target of Lola, the copia retrotransposon, Lola appears to repress copia expression in the central nervous system, while activating its expression copia in the gonad [19]. These results suggesting that a single Lola isoform may have dual functions or that different Lola isoforms regulate copia expression in different tissues. The ability of Lola isoforms to function as transcriptional activators and repressors is supported by the localization of Lola with both active and repressive marks of transcription on polytene chromosomes (Fig 8A and 8B).
The existence of at least 20 different Lola protein isoforms, 17 of which contain one or two zinc fingers that lack sequence conservation and exhibit different binding specificities, suggests that the different Lola isoforms regulate distinct cohorts of genes and may function in a tissuespecific manner [22,[27][28][29][30]. Consistent with this prediction, we and others have found that different isoforms are expressed and function in different tissues [22,28,38]. In the embryo we find that Lola-T, which is specifically mutated by the lola 46.38 allele, is required for gonad morphogenesis, consistent with a previous study (Fig 1) [38]. However, we also observe defects in gonad morphogenesis with a Lola-K isoform-specific mutant allele, lola ORC4 (Fig 1). Comparison of the T and K isoforms reveals significant sequence similarity in their zinc finger regions [28], as well as similarity in their predicted DNA consensus binding sequences [30], suggesting that these isoforms may regulate transcription of a common set of target genes. Expression analysis of Lola-K and Lola-T GFP fusion proteins in the gonad revealed that Lola-T is expressed in SGPs in stage 13 gonads, while Lola-K-GFP was strongly expressed in the SGPs and PGCs of both stage 13 and 15 gonads (S3 Fig). These results suggest that both Lola-T and Lola-K function in the SGPs to regulate gonad morphogenesis. Expression of Lola-K in PGCs suggests that this isoform may be specifically required in the PGCs. Interestingly, it was previously reported that overexpression of lola-T rescues the lola mutant gonad phenotype in mutants predicted to lack all Lola isoforms [38]. This data suggests that different isoforms may be able to compensate for each other if they share DNA binding similarities, as is the case for Lola-K and Lola-T. It is also feasible that different Lola isoforms dimerize to cooperatively regulate gonad development, which is supported by our molecular modeling and yeast-two hybrid analysis demonstrating the ability of Drosophila BTB domains to homodimerize (Figs 6 and 7) [31,32].
The transcriptional activity of Lola could also depend on its interaction with other BTB and/or non-BTB domain-containing proteins, including Rib. Drosophila BTB domain-containing proteins have been demonstrated to interact with each other in yeast-two hybrid analysis [18]. In addition, BTB containing proteins, including Pipsqueak, Trithorax-like, Batman and Bric-à-brac, function together to limit ovariole number in the Drosophila ovary [65]. In the context of the salivary gland and the trachea, Rib interacts with another BTB family protein, Lolal, to regulate development [42]. Molecular modeling of Lola-Rib BTB domain interactions (Fig 6), suggested that Lola and Rib BTB domains are capable of forming heterodimers, as well as homodimers. This hypothesis was supported by genetic interaction analysis of rib and lola mutants and the physical interaction of their BTB domains (Figs 4 and 7). Colocalization of Rib and Lola on polytene chromosomes reveals that Rib and Lola colocalize at some, but not all sites on the chromosomes (Fig 8E). This data combined with molecular modeling results suggests that Rib and Lola homodimers and heterodimers are likely to regulate transcription of different cohorts of genes to promote gonad morphogenesis.
Our understanding of how Rib functions to regulate tissue morphogenesis comes from work examining its role in salivary gland and tracheal development. Previous studies revealed that rib mutants exhibit defects in tube elongation that arise from decreased apical membrane length and increased apical stiffness [15,16,41]. Consistent with these phenotypes, rib mutants exhibit changes in apically-localized proteins, including decreased expression of crb, increased levels of active of Moe, and decreased levels of the recycling endosomal protein Rab11 [42]. Further examination of the basis of the rib mutant phenotype showed that these defects arise from a failure of rib mutants to execute the cell shape and volume changes needed for salivary gland morphogenesis [36]. Recent Rib ChIP-seq analysis and expression profiling of rib mutants resulted in the identification of 20 potential targets for transcriptional activation and 40 potential targets for transcriptional repression in the salivary gland [36], including genes involved in cell migration, cell adhesion, and regulation of the cytoskeleton [36]. Similar to the salivary gland and trachea, gonad morphogenesis depends on changes in SGP cell shape [64], suggesting that Rib may function through some of the same downstream targets to promote proper gonad formation. We have examined the ability of Rib to regulate expression of crb and phosphorylation of Moe. In the case of Crb, low levels of Crb protein are observed in the gonad, unlike the strong membrane expression observed in the salivary gland (S6A Fig), which is not surprising giving that neither SGPs or PGCs have been observed to exhibit polarity. This expression is unchanged in rib homozygotes, demonstrating that crb does not appear to be a target of Rib in the gonad (S6A and S6B Fig). While pMoe is strongly enriched at the PGC membrane, levels of pMoe remain unchanged in rib homozygotes (Fig 6C and 6D). Therefore, it appears that Rib regulates different downstream targets in a tissue-specific manner. This activity may depend on its interaction partners, as Rib interacts with different binding partners in different tissues, namely Lolal in the salivary gland and trachea and with Lola in the gonad [42].
In addition to identifying the downstream targets of Rib and Lola in the developing gonad, it is also critical to characterize the molecular mechanisms regulating Rib and Lola expression and activity. Rib has been suggested to function downstream of the Mitogen Activated Protein Kinase (MAPK) signaling pathway based on the similarity of mutant phenotypes of rib and members of the Fibroblast Growth Factor-MAPK signaling pathway, as well as the presence of seven MAPK consensus phosphorylation sites in Rib [16]. In contrast, the mechanisms regulating Lola's functions are unknown. Identification of signaling pathways with which rib and lola genetically interact would provide insight into the mechanisms regulating expression and function of these proteins and allow us to better understand how they are functioning within a network to regulate tissue development.
Although there are no human orthologs of lola and rib, there are similar proteins to Lola in vertebrates, which include Zfp131, Miz-1, and Leukemia-Related Factor (LRF) [66]. Zfp131 exhibits the most similar expression pattern to Lola, as it is expressed in the testes, adult brain and the developing nervous system [66,67]. In contrast, Rib does not have any similar vertebrate proteins, as the BTB domain has not been observed in combination with the PSQ motif in vertebrates. However, other BTB family proteins may functionally substitute for Rib in the context of vertebrate gonad development. The characterization of the roles of Rib and Lola in embryonic gonad development, their genetic and physical interaction, and their colocalization with regions of active and inactive transcription on polytenes chromosomes suggests that cooperative regulation of gene expression by BTB family proteins may be used in a variety of developmental and disease contexts. The implication of Drosophila and mammalian BTB proteins in lymphocyte, skeletal, gonad and neurological development, as well as in cancer, illustrates the importance of understanding the mechanisms by which these proteins cooperate to regulate gene expression [21]. Identification of the direct downstream transcriptional targets of these genes and the molecular pathways in which they function is critical for understanding how these genes regulate development and contribute to disease. The following gonad phenotypes were scored: fusion (red), compaction (blue) and wild-type (green). Gonads were scored by staining somatic gonadal precursor cells for the 68-77-lacZ enhancer trap. A Chi-square test was performed to test the null hypothesis that the phenotype ratios will be the same for all genotypes. Results allow us to reject the null hypothesis: