Cg promotes niche formation and represses precocious PGC differentiation

In a screen to identify novel regulators of ovarian development [13] we identified the transcription factor Combgap (Cg) as required for proper niche formation and PGC maintenance. To follow the differentiation status of PGCs, we used anti-Hts antibodies that label the fusome, an intracellular organelle within germ cells. In wild type LL3 ovaries, PGCs do not yet differentiate to form germline cysts, and each PGC carries a spherical fusome (Fig 1B, arrowhead) [6,24,25]. By contrast, ~50% of cg-mutant gonads from three different allelic combinations contained branched fusomes, suggesting PGCs differentiated precociously (Fig 1C and 1D, arrowheads, insets, for quantification see legend). Anti-Orb antibody, which robustly stains more developed 8- and 16 cell cysts [26], confirmed these observations. No Orb labeling was observed in WT ovaries (Fig 1E). However, Orb was expressed in cg-mutant gonads (Fig 1F and 1G). Thus, both fusome morphology and Orb expression demonstrate that the wild-type function of Cg is required to prevent precocious PGC differentiation.

In addition to PGC maintenance, Cg also affected niche formation. To follow TF cell accumulation and stack formation we used anti-Engrailed (En), which specifically stains TF cells [27]. At LL3, TF cells in WT or heterozygous ovaries were already stacked to form the adult number of niches (Fig 1H, Table 1) [8,12,28]. By contrast, TF formation was defective in cg mutants. This was particularly noted in the cg¹/cg^KG00882 combination (Fig 1I, Table 1). However, fewer TFs were also noted in other allelic combinations (Table 1). These ovaries also displayed shorter TFs (Fig 1J, arrows), an additional indication for a difficulty in TF formation. These phenotypes suggest that Cg is required for the proliferation/survival of somatic precursors, for the process of TF specification or for its correct timing [7].

In accordance with a possible role for Cg in precursor cell proliferation, cg-mutant gonads were significantly smaller than WT (Table 1). Vital dye staining (Propidium Iodide, PI) indicated that cell death did not increase in cg-mutant ovaries (of 18 cg¹/cg^KG00882 ovaries, 16 were not labeled by PI at all, and 2 had 1–3 dead cells, a similar level to WT ovaries)[8]. On the other hand, phospho-Histone H3 (pH3) staining, which marks mitotic cells, was significantly decreased in cg-mutants (0.15±0.03 pH3-positive cells per 1 μm³ in WT ovarian volume, n = 21, as compared to 0.08±0.02 per 1 μm³ in cg¹/cg^KG00882 ovaries, n = 14, t-test p-value 1.35e-6). Thus, Cg affects gonadal size by promoting cell proliferation but not cell survival.

Cg is required in the soma for ovarian development

Cg promotes somatic development and niche formation on the one hand, and represses precocious PGC differentiation on the other. To determine how might Cg affect both soma and germline, we first examined its expression using anti-Cg antibodies [4], and found that it was present in all nuclei of WT LL3 ovaries (Fig 2A). Similar results were obtained by examining the expression of a GFP protein trap inserted in the Cg locus (Fig 2B)[29]. The ubiquitous ovarian Cg expression is in accord with its ubiquitous expression in other organs [2,4].

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Cg is required in somatic cells for gonad morphogenesis.

(A, B) Both somatic and germline nuclei are stained by anti-Cg (A, green n = 30) or anti-GFP (B, green, n = 28). (C, D) Anti-GFP (green) marks WT cells. Anti-Hts (magenta) outlines somatic cells and marks fusomes in PGCs. cg² (C, n = 52) or cg^KG00882 (D, n = 21) mutant PGCs remain undifferentiated and harbor spherical fusomes (arrowheads). (E, F) Anti-Cg is in green and TFs are marked by anti-En (magenta). (E) In control LL3 ovaries, all nuclei express Cg and TFs are long and well formed (arrows). (F) Cg-GFP ovaries. Cg is reduced by tj-Gal4 driving GFP^RNAi in somatic cells. Cg is still expressed in germ cells (arrowheads) and a few anterior somatic nuclei. Ovaries are smaller and contain less TFs (quantification in Table 1). In addition, many TFs are shorter than WT TFs (compare arrows in 2E, 2F), indicating less developed TFs. Bars in A, B, E, F are 20 μm, and in C, D are 10 μm.

https://doi.org/10.1371/journal.pgen.1006330.g002

To determine the origin of the somatic and germline phenotypes of cg-mutants, we used mosaic analysis. Germline clones of either cg² or cg^KG00882 did not develop as cysts within larval ovaries, and contained spherical fusomes (Fig 2C and 2D). Similarly, depletion of Cg specifically from PGCs by expressing UAS-cg^RNAi with nos-Gal4 resulted in normal development of both somatic and germline lineages (S1 Fig). We conclude that Cg does not act within PGCs to maintain normal ovarian development.

Expression of UAS-cg^RNAi with tj-Gal4, which is specific to the ovarian soma, also failed to produce phenotypes, possibly due to a low efficiency of the RNAi (S1 Fig, Table 1). Indeed, expression of an efficient RNAi line against GFP in Cg-GFP ovaries resulted in low Cg levels in most somatic ovarian nuclei (Compare Fig 2E and 2F and S1 Fig). GFP^RNAi ovaries were significantly smaller and contained fewer TFs (Table 1). In addition, the formed TFs in the GFP^RNAi ovaries were less developed and incorporated fewer TF cells per TF stack (Compare TFs under arrows in 2E, 2F, S1 and S2 videos). Thus, Cg reduction in the ovarian soma can phenocopy the somatic ovarian phenotypes of cg mutants. By contrast, PGC development was normal in the GFP^RNAi ovaries (S1 Fig). One possibility is that PGC differentiation requires simultaneous depletion of Cg in both somatic and PGC nuclei. Alternatively, systemic effects could be causing PGC differentiation in cg mutant animals (see below).

Cg is required for the correct expression of Broad-Complex

An additional phenotype of cg-mutant animals was their failure to pupariate at the end of larval development. cg-mutant larvae continued feeding and growing for several days after their WT siblings pupariated, eventually dying as giant larvae (S2 Fig). Such phenotypes suggest that Cg may somehow affect the ecdysone pathway, which induces molting and pupariation [30]. Since ecdysone signaling is required for ovarian development [8,13,31,32], we investigated whether Cg affected the ovarian ecdysone response, and in particular the Broad-complex (BR-C), its major ovarian target during larval development [8,13].

Of the four Br isoforms, Br-Z3 is not expressed in the ovary and Br-Z4 is expressed at very low levels at the larval-pupal transition, after gonad morphogenesis has largely occurred. We therefore concentrated our analysis on Br-Z2, which is expressed in the somatic ovary from early larval stages, and on Br-Z1, which is induced by ecdysone at mid-3^rd instar [13]. Isoform-specific antibody staining in cg¹/cg^KG00882 ovaries revealed a reduction in Br-Z1 and Br-Z2 protein levels (For Br-Z1, n = 13, t-test p-value 1.44e-7, Fig 3A and 3B. For Br-Z2, n = 19, t-test p-value 2.13e-10, Fig 3C and 3D). Since BR-C levels could be indirectly affected by the retarded development of cg-mutant animals, we also analyzed Br-Z1 and Br-Z2 in somatic mutant clones that were generated in a WT heterozygous background. Interestingly, cg-mutant somatic clones showed no reduction of Br-Z1 (S3 Fig), suggesting that gonadal Br-Z1 expression could be indirectly affected by the developmental retardation of cg mutants. By contrast, Br-Z2 levels were consistently reduced in cg somatic clones (Fig 3E–3F’). This demonstrates that Cg promotes Br-Z2 expression in a cell-autonomous manner, which is independent of changes in larval physiology or development.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. Reduced Broad expression in cg mutant ovaries.

(A, B) Ovaries were imaged at the same time with the same confocal settings. Germ cells are labeled by anti-Vasa (green). (A) At LL3, all WT somatic nuclei express Br-Z1 (magenta). (B) Reduced Br-Z1 expression in cg-mutant ovaries. (C, D) Ovaries were imaged at the same time with the same confocal settings. TF cells are labeled by anti-En (magenta). (C) At LL3, all wild type somatic nuclei express Br-Z2 (green). (D) Reduced Br-Z2 expression in cg-mutant ovaries. (E-F’) Mosaic analysis of Br-Z2 expression in cg-mutant cells. GFP (anti-GFP, green) marks wild type cells. cg-mutant clones are outlined. (E, E’) cg² mutant cells express lower Br-Z2 levels (magenta in E, grayscale in E’). (F, F’) cg^KG00882 mutant cells express lower Br-Z2 levels (magenta in F, grayscale in F’). Bars are 10 μm each.

https://doi.org/10.1371/journal.pgen.1006330.g003

To determine the extent to which reduced Br-Z2 expression contributes to gonadal phenotypes, we over-expressed Br-Z2 in cg¹/cg^KG00882 ovaries using a UAS promoter. Br-Z2 gonadal expression was rescued in these animals, and gonadal size and TF numbers increased, as compared to cg¹/cg^KG00882 mutants (Table 1, S4 Fig). However, gonads were not restored to WT state, suggesting that this pleiotropic gene affects ovarian development via multiple changes in gene expression.

Cg over-expression represses Broad expression and blocks ovarian development

To further probe how Cg might affect BR-C expression, we over-expressed it using a UAS- insertion line upstream of cg. Cg over-expressing LL3 ovaries were markedly smaller than WT ovaries and contained very few TFs (Fig 4A and 4B and Table 1). To further determine the developmental stage of Cg over-expressing ovaries, we stained them with anti-Tj. In young WT ovaries, Tj is expressed in all somatic gonadal nuclei, but during third instar, Tj expression is gradually restricted to ICs [5,33]. Indeed, only ICs were labeled by anti-Tj in WT LL3 ovaries (Fig 4C). By contrast, all somatic cells were labeled in the Cg over-expressing gonads, and none of these cells intermingled with germ cells (Fig 4D). Thus, Cg-over expressing ovaries seem arrested at a younger developmental stage. Another feature that characterized both Cg-over expressing ovaries and young ovaries, was the absence of a posterior group of somatic cells (swarm cells), which migrate to the posterior during third instar. In Cg-over expressing ovaries, swarm cells failed to migrate and remained at the medial side (Compare Fig 4E to 4F, white star). Combined, these features suggest that Cg-over expressing ovaries failed to mature.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. Over-expression of Cg disrupts ovarian morphogenesis.

(A, B) LL3 ovaries, Anti-Cg is in green and anti-En marks TFs (magenta). Images were taken at the same day and in the same confocal settings. (A) In wild-type ovaries, TF stacks are completely organized and all somatic nuclei express uniform Cg levels. (B) Over-expression of Cg (outlined) results in extremely small ovaries containing few TFs. (C-F) PGCs are marked by anti-Vasa (green), anti-Hts (magenta) outlines somatic cells and fusomes within PGCs. (C, D) LL3 ovaries, anti-Tj marks ICs in Cyan. (C) In the wild type, ICs intermingle with PGCs. (D) In Cg-over-expressing ovaries, no ICs are present between PGCs. Tj is expressed in all somatic cells. (E) PGCs occupy the middle of the ovary and are interspaced with ICs. Swarm cells are at the posterior (white star, bottom) of the ovary. (F) Cg-over expressing ovary. PGCs are grouped together without ICs to separate them. Swarm cells, which failed to migrate to the posterior, are located medially (white star at the side). Bars are 10 μm each.

https://doi.org/10.1371/journal.pgen.1006330.g004

The gross morphological abnormalities of Cg over-expressing ovaries resembled the developmental defects in ovaries over-expressing the dominant negative EcR [13]. Indeed, examination of Broad expression using an antibody directed against the core region of BR-C, which labels all Broad isoforms, revealed a strong Broad suppression in Cg over-expressing cells. At ML3, all somatic WT nuclei express BR-C (Fig 5A–5A’’). tj-Gal4, which is expressed in all somatic ovarian nuclei at this stage [5], led to Cg over-expression in most somatic nuclei (Fig 5B and 5B’’). Consequently, hardly any BR-C expression was observed in these nuclei (Fig 5B’ and 5B’’).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 5. Over-expression of Cg abolishes Broad expression.

In all panels, Anti-Cg is in green, Anti-BR-C is in magenta. (A, B) ML3 ovaries. In the wild type (A-A”) Cg and BR-C are both present in somatic nuclei in uniform levels. (B-B’) In Cg-over expressing ML3 ovaries, almost all somatic nuclei express high Cg levels and BR-C labeling is lost. (C-C”) Wild type LL3 ovaries, all somatic nuclei co-stain with BR-C and Cg. (D-D”) Regions of cells that over-express Cg (outlined) display reduced or no BR-C staining. Lack of BR-C expression is particularly apparent in the middle of the over-expressing region, where cells do not face wild type neighbors. Bars are 10 μm each.

https://doi.org/10.1371/journal.pgen.1006330.g005

At LL3, due to the morphological defects caused by Cg over-expression, cells expressing high Cg levels were located at lateral ovarian regions (Fig 5D). In these cells, BR-C levels were very low or non-existent (Fig 5D’ and 5D’’). Interestingly, cells at the border of the region over-expressing Cg retained some BR-C labeling, suggesting a non-autonomous contribution to BR-C expression (Fig 5D’, arrowheads). Together, these results show that correct Broad expression requires wild-type levels of Cg, and any change in Cg amounts or function disrupts expression from this important locus.

EcR binding to polytene chromosomes depends on correct Cg levels

To determine how direct the control of Cg on BR-C expression is, we asked whether Cg is localized to the BR-C locus. At LL3, increased transcription from BR-C can be directly observed as puffing of the locus in polytene chromosomes of salivary glands [34]. Staining with anti-Cg revealed that Cg was localized to the BR-C puff in 84.5% of all chromosomes tested (n = 58, Fig 6A’ and 6A’’’, green arrow). Moreover, co-labeling with anti-EcR revealed that in all these cases, Cg co-localized with EcR (Fig 6A” and 6A’’’, white arrow). This suggests that transcription from BR-C may depend directly on Cg. EcR and Cg co-localized in some loci but not in others (Fig 6A’–6A’’’, green, magenta and white arrowheads). This suggests that Cg may not be an obligatory component of the EcR transcriptional complex. Alternatively, Cg may be present in some EcR-containing polytene chromosome bands at levels that are undetected by antibody staining, or associate with EcR complexes in a temporally-restricted manner.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 6. Cg affects EcR binding to polytene chromosomes.

Salivary glands polytene chromosomes spreads. For all panels, imaging was performed with the same confocal settings. DAPI (white) stains DNA, anti-EcR is in magenta, and anti-Cg is in green. (A-A”) A portion of the X chromosome is shown. A green arrow in (A’) shows where Cg is bound to the BR-C puff. EcR is bound at the same puff (magenta arrow in A”). In the composite (A”‘), co-localization of the two antibodies appears in white. Two such bands are marked (white arrows). In addition numerous bands demonstrate unique EcR or Cg binding (magenta and green arrows, respectively). (B-B”‘) Chromosomes spread from cg¹/cg^KG00882 mutant larva. Chromatin bands are not as tight, very little Cg is observed and EcR binding to chromatin is reduced. (C-C”‘) Cg was over-expressed in salivary glands using sgs3-Gal4. A large amount of Cg, which spreads throughout the chromatin is observed. Little EcR can be detected on such chromosomes.

https://doi.org/10.1371/journal.pgen.1006330.g006

As EcR and Cg affect BR-C expression, and both co-localized at the BR-C locus, we investigated whether EcR localization may depend on Cg. In the strong cg¹/cg^KG00882 allelic combination, lower Cg levels were detected on polytene chromosomes (100% of spreads, n = 18, Fig 6B’ and 6B’’’). On such chromosomes, EcR was still present, but at reduced levels (89% of spreads, n = 18, Fig 6B’’ and 6B’’’). In addition, the chromatin-banding pattern was loose, suggesting a global effect on chromatin by Cg (Fig 6B). Other allelic combinations displayed similar, albeit weaker phenotypes: The reduction in Cg levels in the weaker allelic combinations was not as severe, resulting in a smaller reduction in EcR binding to chromatin (S5 Fig).

Our data show that in the somatic cells of the ovary, a reduction in Cg protein results in lower Br-C induction, while over-expressing Cg results in an almost complete block of Br-C expression (Figs 3 and 5). We therefore investigated the status of EcR binding to polytene chromosomes of salivary glands that over-expressed Cg. In 45% of chromosome spreads, Cg was indeed over-expressed, coating the entire chromatin (n = 58, Fig 6C’ and 6C’’’). In such spreads, EcR binding was almost completely undetected (92% of Cg over-expressing spreads, Fig 6C’’ and 6C’’’). The effects of Cg levels on EcR binding to chromatin (reduction in the mutant background and severe reduction upon Cg over-expression), match the manner by which changes in Cg levels affect BR-C expression. Together, these results confirm that normal EcR activity depends on tight regulation over Cg levels and function.

Cg promotes long-range chromatin contacts at the BR-C locus

The changes in chromatin structure observed in polytene chromosome of cg¹/cg^KG00882 animals, suggested that Cg may affect the chromatin structure at the BR-C locus. To test this possibility we used circularized chromosome conformation capture with high-throughput sequencing (4C). To compare chromatin conformation prior to- and following exposure to the ligand, and to ask how may Cg affect this response, we used Kc167 cells, which respond well to the ligand (see below). Since our observations in polytene chromosomes suggested a connection between EcR and Cg, we selected as viewpoints into the BR-C locus two regions that are associated with EcR. The first viewpoint, B1 (at 1609904), was located between the distal and proximal promoters, 2 Kilo-Bases (KB) away from EcR binding sites. The second viewpoint (B2, at 1619647) was close to the proximal promoter, in a region within the first intron, which binds EcR in-vivo (S6 Fig) [35,36,37].

We observed a domain-like contact landscape around the B1 viewpoint, with asymmetric contacts extending some ~250KB downstream (Fig 7A, color coded domainogram), and ~125KB upstream. Analysis of B2 indicated several long-range contacts that were enriched over the typical domain architecture (Fig 7A, right). We selected the three loci with the strongest evidence for long-range contacts for further quantitative analysis (X1-X3). Significantly, these three contact points constituted the three sites with the highest EcR-binding content within this ~370KB region (marked as red bars underneath X1-X3, Fig 7A, right, S6 Fig) [36,37]. Thus, the EcR binding region within BR-C is engaged in long-range contacts with other EcR-binding regions.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 7. 4C-seq profiles for BR-C locus-specific contacts in Kc167 cells and in wing imaginal discs.

Regions within X chromosome are according to Ensembl_6 of GBrowse. (A) 4C domainogram for two different viewpoints (chromosome X, bait B1 1609904 on left, and bait B2 1619647 on right) and five experimental conditions are shown. The average contact intensity (smoothed number of captured ligations) is depicted as a trend with error band representing two standard deviation of the estimated mean. Raw data points are shown as small black dots. A color-coded depiction of contact intensity in increasing scales (10 to 300 restriction fragments) is shown below each trend. X1-3 indicates the position of loci we select for quantitative analysis below. N, represent the number of inferred ligation evens used to generate each profile, typically based on 1-3M sequenced reads. (B) Comparison of 4C profiles in control cells treated with ecdysone (red) to untreated cells (black, top) or treated cells in which Cg was knocked down (blue, bottom). Each comparison shows overlaid contact intensity trends on top. The domainogram is showing the differential contact intensity in increasing genomic window sizes. Red bands on the dominogram indicates contacts in the region are stronger following ecdysone treatment. (C) Bar-plots represent the mean contacts for windows X1, X2, 200 fragments per each and 30 fragments for X3. The number of contacts per window is color-coded. (D) Bar-plots representing the mean contacts for X1 with viewpoints B1 and B2 in wing imaginal discs derived from wild type and cg¹/cg^KG00882 animals, 40 fragments. **P<0.001, *P<0.05 (Chi-Square pair-wise test).

https://doi.org/10.1371/journal.pgen.1006330.g007

We compared systematically the 4C contact profiles around the B1 and B2 viewpoints between conditions, using differential contact analysis at multiple scales (Fig 7B). Quantitative follow up of the contact intensity at the X1, X2 and X3 loci (Fig 7C) showed that the overall contact architecture was stable. However, ligand exposure and Cg presence modified some contacts in significant ways: The closest contact (X1) was located ~30KBs upstream, next to the distal BR-C promoter. Our data indicated about 27%-22% increase in the number of contacts between X1 and either B1 or B2, respectively, following ecdysone treatment (Fig 7C). Notably, a significantly weaker increase was detected when cells were pre-treated with cg ds-RNA. Thus, exposure to ecdysone increases BR-C transcription and contact stability. Both these phenomena are Cg-dependent. Reciprocal 4C confirmed the increased number of contacts between B1 and B2 to X1 upon ecdysone treatment and strengthened the dependency of this interaction on Cg (S7 Fig).

To further substantiate the effects of Cg-depletion on the BR-C locus, we studied its conformation in wing imaginal discs derived from wild type and from cg¹/cg^KG00882 late third instar animals. At this developmental stage, ecdyosne titers are high, and BR-C transcription peaks. Contacts between X1 and B1 or B2 were significantly lower in cg-mutant animals as compared to wild type (Fig 7D). This conforms to the reduced BR-C expression and reduced EcR- binding to the locus that is observed in mutant animals. Thus, in-vivo studies confirm the dependency of contacts at the BR-C locus on normal Cg levels.

The second contact region (X2) was ~228KB downstream of B2, within CG42666, and showed an ecdysone -dependent decrease in contacts with B1 and B2 (Fig 7C). Finally, The most distant contact (X3) was ~327KB downstream of B2, within the first intron of DHR4. X3 showed a significant increase in contacts with B2 upon ecdysone treatment. Contacts between B1/B2 and X2 or X3 were not affected by cg knockdown (Fig 7C).

In summary, our 4C analysis showed a complex chromatin structure at the BR-C locus, featuring long-range contacts between EcR-binding regions across ~370KB. These connections are differentially modulated by exposure to ligand, but only the contacts within the BR-C locus are Cg-dependent. 4C profiles around BR-C serve as a preliminary indication to the diversity of changes both ligand and Cg confer upon the BR-C chromatin landscape.

Cg is required for proper EcR response in Kc167 cells

Our results demonstrate that EcR binding regions in BR-C are associated with two other genes containing EcR-binding regions, and that EcR binding to chromatin depends on Cg. This prompted us to ask whether Cg and EcR might physically interact, and whether Cg may also affect the transcription of DHR4 and CG24666. Our attempts to co-immunoprecipitate EcR and Cg from either adult ovaries or Kc167 cells were unsuccessful. Thus, either the two proteins do not physically interact, or they are part of a temporally restricted or labile complex. To examine this last option, we measured EcR-Cg interactions using a proximity ligation assay (PLA), which can detect physically associated proteins without cell disruption.

Control, β-Gal ds-RNA, cells contained an average of 5.09 PLA spots located at their nuclei (n = 461 cells, Fig 8A and 8B). Treatment of control cells with 1μM ecdysone for 2 hours increased the average number of spots to 7.26 (n = 490 cells, Fig 8A and 8B). A Mann-Whitney U test determined that this change is highly significant (p vaue 3.54E-13). cg dsRNA cells treated with ecdysone contained only 3.27 spots on average (n = 320, Fig 8B, p value 1.06E-35, compared to ecdysone treated control cells), attesting to the specificity of the assay. Together, the PLA and polytene chromosome studies suggests that only a fraction of Cg and EcR may reside within the same complex. Some of these complexes may be preformed, while others either form de-novo, or are stabilized upon exposure to ecdysone.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 8. Cg is required for EcR-mediated transcription.

(A) Representative data from a PLA experiment. A compression of a few Z sections is shown. PLA signal is in magenta and DNA (DAPI) is in blue. Complexes appear in control, untreated β-Gal dsRNA cells (naïve), and their numbers increase following treatment with ecdysone. (B) Summary of PLA data for control β-Gal dsRNA cells (teal), β-Gal dsRNA cells following ecdysone treatment (orange) and cg dsRNA following ecdysone treatment (Blue). The plot shows the number of cells (Y axis) containing a particular number of PLA spots (from 0 to 24, X axis). (C) qPCR quantification of fold-increase in BR-Z1, BR-Z2, CG42666 and DHR4, following ecdysone treatment, as compared to transcripts levels in cells that were not treated by ecdysone. Control, β-Gal-dsRNA is in blue and cg-dsRNA in red. Paired t-test p values are indicated.

https://doi.org/10.1371/journal.pgen.1006330.g008

We next tested how may Cg affect transcription of genes in contact with BR-C. We first measured isoform-specific induction of BR-C mRNA levels upon ecdysone treatment. Control Kc167 cells treated with dsRNA against β-Gal, responded to ecdysone exposure by up-regulating both BR-Z1 and BR-Z2 transcripts, with BR-Z2 induction consistently higher than BR-Z1 (Fig 8C). When cg levels were reduced by cg-dsRNA (S8 Fig), up-regulation of either transcript was attenuated by about half (Fig 8C). The dependence of BR-C expression on Cg in both gonadal and Kc167 cells, which originate from embryonic blood cells, suggests that control of this locus by Cg is a ubiquitous phenomenon.

The first associated gene, CG42666, lying next to X2, was not induced by ecdysone (Fig 8C). Since contacts between BR-C and CG42666 were reduced by ligand treatment (Fig 7C), this suggests that following exposure to ligand, some long-range contacts with inactive EcR targets may be diminishing.

On the other hand, transcription of DHR4 (which encompasses X3) was greatly increased by ecdysone treatment. Induction of this gene was an order of magnitude higher than that of BR-C (Fig 8C). Contacts between BR-C and DHR4 increased upon ligand exposure (Fig 7). Thus, the contacts with the un-induced gene diminish, while those with the induced gene increase. Significantly, DHR4 transcriptional induction by ecdysone was reduced in cg-RNAi cells. Thus, Cg can affect other ecdysone-induced genes besides BR-C.

Cg and EcR-mediated BR-C expression

Cg is associated with the BR-C locus and its RNA and protein products in three different cellular systems: somatic ovarian cells, Kc167 cells and in salivary glands. One major inducer of transcription from this locus is EcR [15,17,34]. Several lines of evidence suggest that Cg effects BR-C via EcR. First, EcR and Cg co-localize at the BR-C locus. Second, we observe reduced EcR binding to polytene chromosomes in cg mutants. Third, EcR binding to chromatin is almost completely abolished when Cg is over-expressed in salivary glands. These changes in EcR binding to chromatin fit very well with how Cg levels affect BR-C in somatic ovarian cells. We observe a reduction in Br-Z2 levels in cg-mutant clones, and a complete block of all BR-C isoforms in cells over-expressing Cg. Cg levels affect EcR in a global manner, suggesting other EcR targets are likely affected. Indeed, the transcription of another EcR target, DHR4, is also reduced in cg-RNAi Kc167 cells.

Lastly, the PLA approach suggests that at least some nuclear complexes contain both EcR and Cg. It is likely that EcR and Cg do not directly bind each other, as co-IP approaches failed to establish a direct connection. In addition, Cg is not dedicated exclusively to the ecdysone pathway, and also affects Wingless and Hedgehog target genes [2,3,4]. Indeed, the amount of PLA spots and co-stained bands in salivary glands suggest that only a minor fraction of Cg is engaged in such stable complexes. We hypothesize Cg may have pleiotropic roles, and engage in different or more general complexes. Indeed, a recent publication identified Cg as a PRE-binding protein, and as co-localizing with pleiohomeotic and polyhomeotic (Ph) [38]. It is unclear if this interaction overlaps with how Cg affects EcR and the BR-C locus, or if it represents an entirely independent function. The exact biochemical function of Cg as a more general mediator of protein binding onto chromatin requires further investigation.

Association between EcR-enriched loci

4C analysis shows that EcR-enriched regions within the first intron of BR-C contact EcR-enriched binding site close to the distal BR-C promoter, ~30KBs upstream. The interactions between these EcR-enriched regions existed in Kc167 cells prior to exposure to ecdysone, but increased following ligand binding. Our results support a model where the chromatin is pre-programmed for specific transcriptional responses, and ligand exposure can modulate the intensity of these pre-formed connections. The ecdysone response is rapid, and would benefit from a pre-arranged chromatin structure that enables a coordinated transcriptional response for the hormone.

The contacts of EcR-enriched regions between BR-C, CG42666 and DHR4 suggest that BR-C is transcribed in association with other genes, or in a ‘multigene’ complex. Such complexes are prevalent in mammalian genomes [39]. Interestingly, contact intensity increases upon ecdysone treatment for BR-C and DHR4, the two genes displaying ecdysone-induced transcription. Contacts intensity is reduced for CG42666, which does not respond to ecdysone treatment. At this stage we cannot determine whether the changes in chromosome conformation participate in regulating ecdysone-induced transcription.

Within the Br-C locus, increased contact intensity is Cg-dependent. The decreased intensity following cg-reduction is observed in both Kc167 cells, and to a larger extent, in cg-mutant animals. The consistency of the effect in cells of different origin suggests that Cg affects the conformation of DNA at the BR-C locus in a global manner. The larger effect in the mutant animals is likely due to the inefficiency of the dsRNA technique in tissue culture.

Unlike the contacts between the distal BR-C promoter and the BR-C first intron, the contacts with DHR4 are Cg-independent. As Cg affects the transcription of both BR-C and DHR4, this adds to the complexity of the regulatory interactions in this region. In a similar vein, qPCR measurements show an order of magnitude difference in ecdysone induction levels between BR-C and DHR4. Thus, induction strength is not determined in this case by the association per-se, but by individual promoters and/or the existence of other elements such as insulators, which participate in ecdysone-induced chromatin looping [35,40]. Whether Cg may participate in these complexes or even mediate some or their differences await biochemical analysis.

Cg, BR-C transcription, and gonad morphogenesis

While qPCR measurements in Kc167 cells show cg-dsRNA reduces transcription of both BR-Z2 and BR-Z1, protein measurements in somatic ovarian clones show a marked reduction only for BR-Z2 protein levels. One option to explain this difference involves the smaller reduction of BR-Z1 transcripts in cg-dsRNA as compared to BR-Z2. This smaller difference may still be detected by qPCR, but not by antibodies. A second, not mutually exclusive explanation may be connected to post-transcriptional control. The reduced BR-Z1 levels may trigger some compensation at the level of splicing or protein translation. Alternatively, the manner by which Cg controls transcription from the BR-C locus in ovarian and Kc167 cells may be somewhat different.

BR-Z1 protein levels are reduced in whole-body mutants but not in clones, suggesting a systemic component for cg-depletion on ovarian BR-C transcription. During late larval stages, BR-C acts in a feed-forward loop, augmenting ecdysone production [41]. Cg may therefore affect ovarian BR-C expression both autonomously and systemically. An antagonistic effect may be caused by DHR4, which is a negative regulator of ecdysone production [42] and is also reduced by cg-depletion. How these two opposing effects balance out in the prothoracic gland, which synthesizes ecdysone, remains to be investigated. However, cg-mutant larvae, which do not pupariate, suggest that Cg depletion results in insufficient ecdysone production.

Finally, our previous work showed that ecdysone signaling promotes three processes during gonadal development: gonadal growth, niche differentiation, and PGC differentiation. ecdysone signaling in the larval ovary promotes BR-Z1 expression exactly at the time when niches and PGCs differentiate. This suggests that BR-Z1 is necessary for stem cell unit differentiation. Indeed, expression of a dominant negative EcR abolishes both BR-Z1 expression and stem cell unit differentiation [13]. BR-Z2 is expressed even prior to stem cell unit differentiation, and therefore cannot be sufficient to induce differentiation. However, BR-Z2 may still be necessary for the correct differentiation of niches and PGCs, together with BR-Z1. Indeed, BR-Z2 could partially rescue both ovarian growth and niche differentiation.

One important difference between the effects of BR-C and Cg on gonad morphogenesis is that while both factors promote ovarian growth and niche formation, Cg retards PGC differentiation while BR-C promotes it. This suggests that Cg affects other targets important for gonadal growth. Previous work uncovered many signaling pathways that are important for niche formation and GSC establishment [7]. Understanding how these pathways converge to form the organized organ is crucial for basic understanding of organogenesis and for applications in regeneration. Uncovering how pleiotropic proteins such as Cg contribute to coordinated transcription and hence to organogenesis will be part of this effort.

Fly stocks

cg^KG00882, cg¹, FRT42D, FRT42D,ubi-GFP, Cg-GFP (cg^CC01469), cg-RNAi (cg^HMS01145), UAS-Br-Z1, UAS-Br-Z3, UAS-Br-Z4 were from the Bloomington Stock collection. cg² was from Dr. William J. Brook (University of Calgary, Canada), tj-Gal4, nos-Gal4 from Dr. Ruth Lehmann (NYU, USA), c587-Gal4 was from Dr. Ting Xie (Stwoers Institute, USA), UAS-Cg (P{y+, Mae-UAS6.11}Cg was from Dr. John Merriam (UCLA, USA). UAS-Br-Z2 was prepared as follows: br-Z2 cDNA in pBS(SK-)was obtained from Dr. Lauren von Kalm (UCF, USA). Br-Z2 was PCR-amplified using the forward primer AAAAGAATTCAGCCAGAACCAGACACCCATCGAGATG and the reverse primer TTTTAGATCTCATGGTCGTGCTGTCTTTCATCGCTG. The PCR product was suBR-Cloned into pUAStattB via the BglII and EcoRI sites. Injection of UAS-Br-Z2 into the attP2 site on the third chromosome were performed by Genetic Services (Sudbury, MA, USA).

Larval dissections and staining

Larval developmental timing was determined as described previously [13]. In short, crosses were performed in bottles, to ensure a sparse larval population. In these strict under-crowded conditions, larval development is uniform. Laying was performed for two hours, and eggs were allowed to mature at 25°C until late third instar (120 h after laying). cg mutants, which do not pupariate, were dissected 5–7 days after laying. Females were dissected and stained as previously described [13,43].

Antibodies

Generation of BR-Z2-specific antibodies was according to [44]: The Br-Z2 isoform-specific region was PCR amplified (forward primer: GGATCCAAGGAAACTCGCCCAAGAAACTC, reverse primer: GAATTCTGCTGTGGCTGTTGGCTTTGC). The amplified region was cloned using BamHI and EcoRI, into pGEX-3X bacterial expression vector in E. coli BL21 strain. Bacterial cultures were grown until OD_600nm 0.5 and induced with 2mM IPTG for 4h. GST fusion proteins were purified on glutathione agarose (Sigma). The purified GST-BrZ2 was used to immunize New Zealand female rabbits. The first injection was in Freund's complete adjuvant and the boost, spaced for at least 3 weeks, was in Freund's incomplete adjuvant. A total of 3 boosts were performed and sera taken 2 weeks after the last boost was tested for immunoreactivity against the injected proteins. The resulting Ab reacts specifically with Br-Z2 (S9 and S10 Figs).

Anti-Cg (1:1000) was a kind gift from Dr. William J. Brook (University of Calgary, Canada), Rabbit-anti-GFP (1:1000) was from Invitrogen, Rabbit-anti-Vasa (1:5000) was a gift from Dr. Ruth Lehmann (NYU, USA), Guinea Pig anti-Tj (1:7000) was a gift from Dr. Dorothea Godt (Univeristy of Toronto, Canada). The following monoclonal antibodies were obtained from the Developmental Studies Hybridoma Bank, developed under the auspices of the NICHD and maintained by the University of Iowa, Department of Biology: anti-En (4D9, 1:20), anti-Hts (1B1, 1:20), anti-BR-C (25E9.D7, 1:10), anti-Br-Z1 (Z1.3C11.OA1, 1:10), anti-Orb (6H4, 1:20), anti-EcR.C (AG10.2, 1:20). Larval ovaries staging and staining was as previously published [13].

qPCR

0.4x10⁶ cells were seeded together with the dsRNA and RNA isolated 3 days later using PerfectPure RNA cultured cell kit (5Prime) according to manufacturer’s instructions. Reverse transcription was performed with the High Capacity cDNA Reverse Transcription Kit (Applied Biosystems). qPCR employed SYBR Green (Invitrogen) with the following primers (forward and Reverse) RpS17: CAAGATTGCCGGCTATGTCA and CCTGCAACTTGATGGAGATACCA. Combgap: TCCCCGAAGACCGAACTACA and GTACGGGCGTTCCTTCTTGA, CG42666: GGGCTAGGGACGACAGTTT and GTTTGGAAGCTCGCTACTGG, DHR4: CGCTCCTACCTGCAAAACTC and CACGAAGGGCACATAGAACA, or the Taqman assay: RpL32 (Dm02151827), Br-Z1 (Dm01837161_m1), Br-Z2 (Dm01821011_m1). Q-PCR was performed in Applied Biosystems’ StepOne TM, analyzed by DDCT and normalized to RpS17 or RPL32. For statistical analyses, two-tailed student’s T-test were performed. P values are indicated.

Tissue culture

Kc167 cells were maintained at 25°C in Shields and Sang M3 insect medium (Sigma) containing 10% fetal bovine serum (Gibco) and 1% penicillin–streptomycin (Gibco). For RNAi, dsRNA synthesis was according to DRSC protocol using Readymix (Sigma), MEGAscript T7 kit (Ambion) and RNAeasy (Qiagen). Amplicons from the DRSC database DRSC32301 for combgap and DRSC24562 for β-Gal were used. 2 μg of dsRNA against either combgap or LacZ together with 2 μl DharmaFECT4 (Dharmacon) were applied to 0.4x10⁶ cells in 12-well plates. Cells were exposed to dsRNA for 72 hours and then 1μM ecdysone or control 95% Ethanol were added to the medium for 2 hours. Following ecdysone exposure, cells were harvested for qPCR.

Proximity ligation assay (PLA)

PLA was performed according to manufacturer’s specifications using the Duolink kit (Sigma-Aldrich) with the PLA probe anti-Rabbit PLUS (DUO92002), the anti mouse Minus (DUO92004) and the detection reagent red (DUO92008). First antibodies used were Rabbit-anti-Cg (1:1000) and mouse-anti-EcR (1:20).

4C-Seq

The preparation of 4C template (five biological repeats) was performed as described previously [45], with some modifications. 5X10⁶ Kc167 cells were fixed and following cross-linking, the DNA was digested by the 4 cutter restriction enzyme DpnII (NEB), ligated by high concentrated T4 ligase (NEB). The ligation products were sonicated in to fragments at the average length of 500 bp, end repaired by fill-in and exonuclease reaction and then ligated to an adaptor sequence for the Illumina Trueseq process. 150–200 ng of the resulting 4C template were used for 2 sequential PCR reactions. The first PCR was performed by primers designed to target the region of our interest and illumina Truseq enrichment PCR primer 2. In the second PCR, we used primers that include the sequence illumina Truseq universal adapter AATGATACGGCGACCACCGAGATCTACACTCTTTCCCTACACGACGCTCTTCCGATCT followed by specific internal sequence together with illumina Truseq enrichment PCR primer 2 (see below). The primers for the first PCR were designed to target region that is far by 80 to 98 bp from the DpnII site. The primers (internal primes) for the second PCR were designed to target the region that is far by around 3–80 bp from the same DpnII site.

The primers that were used for two BR 4C viewpoints (B1, B2) and for the primers that were used for three contact regions (X1, X2, X3) are listed in the following table 2.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Primers used for 4C analysis.

https://doi.org/10.1371/journal.pgen.1006330.t002

Sequenced products were mapped to the genome using an in house pipeline (available upon request), and the precise sonication sites were used to eliminate PCR duplicates. 4C domainograms [46] show a smoothed contact intensity profile with gray error band representing 2 SD on the estimated mean contact intensity in a running window. It also shows a color-coded visualization of contact intensities in a series of increasing scales (from 10 restriction fragments to 300 restriction fragments). As the modified protocol provide precise control over PCR duplicates, linear means of the number of ligation evens per restriction fragment within windows can be used. Similarly, statistical tests comparing the total number of ligations in a given window between two conditions (following normalization for total sequence coverage) are robust. This was reconfirmed for the current experiments by quantifying technical variance on triplicate experiments.

Imaginal Discs harvesting for 4C analysis

Inverted head segments of 400 OR (wild type) or 200 cg^KG000882/cg¹ animals were fixed in 2% formaldehyde in PBS (20 min., Room Temperature). Fixed tissue was then washed twice with PBS/125mM Glycine/0.01% Triton-X-100, and a third time with PBS. Samples were washed again with PBS/1% protease inhibitor cocktail (P8340, Sigma). Wing Imaginal discs were then dissected out and placed in an eppendorf tube. Excess fluid was removed and tissues were snap-freezed in Liquid Nitrogen and stored at -80°C. To prepare DNA, discs were thawed on ice and centrifuged at 13,000 rpm for 20 sec. Excess fluid was discarded. Discs were resuspended in 50 ul lysis buffer (10 mM Tris-HCl pH 8.0, 10 mM NaCl, 0.2% Igepal CA360 (Sigma I8896)), and 10 ul/ml of protease inhibitors (Sigma P8340). Discs were homogenized with a plastic motorized pestle (3X2min). Following a brief centrifugation, 500 ul of lysis buffer and 50 ul of protease inhibitors were added and the suspension was centrifuges at 5000 rpm for 5 min, at RT. Lysate was washed twice with ice-cold 1.2X NEBuffer 3 at 5000 rpm, 5 min, at RT, and resuspended in 500 ul NEBuffer 3, supplemented with 7.5 ul 20% SDS. Mixture was incubated at 37°C, rotating at 900 rpm for 1h in a Thermomixer, followed by an addition of 50 ul of 20% T-X-100 and an additional incubation at 37°C, rotating at 900 rpm for 1h in a Thermomixer. Lysate was digested with 400 U of DpnII at 37°C, 900 rpm overnight in a Thermomixer. Enzyme inactivated at 65°C, 20 min. Ligation and continuation of the 4C protocol was as performed for KC cells.

Polytene chromosome squash

Larvae were grown at 18°C in a non-crowded culture. Wandering 3^rd instar larvae were washed and dissected in Ringer’s buffer. Extracted salivary glands were spread according to published protocol [47]. Rabbit-anti-Cg (1:1000) and mouse-anti-EcR (1:20) were used to localize EcR and Cg to the broad locus, and DAPI staining was used to highlight the banding pattern. The br locus was identified by the banding pattern that is specific to the tip of the X chromosome and by the puffing pattern specific to the ecdysone response at late third instar.