Loss of Corto extends the early competence window
Neuroblasts undergo two temporally-separated, distinct types of hb gene repression. In the first, hb is transcriptionally repressed by Svp after one-to-two divisions. In the second, several hours and divisions later, the already transcriptionally repressed hb gene physically relocates to the neuroblast nuclear lamina. This relocation causes heritable gene silencing and renders the hb gene refractory to activation in the descendent neurons, thereby closing the early competence window (Fig. 1A) [11, 23]. Thus, while neuroblasts typically generate one to two early-born neurons, they are capable of producing more until nuclear architecture reorganization terminates competence. We recently identified the PcG factors as critical players in the second step, in terminating neuroblast early competence. PcG proteins are chromatin factors best known for their function to regulate gene repression through the cooperative activity of two multimeric complexes, PRC1 and PRC2 [26, 28, 34,35,36,37,38,39]. We found that Psc is required for timely hb gene relocation to the neuroblast nuclear lamina and closing the early competence window [23].
Another class of proteins, the ETP proteins, has been shown to interact with the Trithorax and PcG complexes and regulate target genes. Corto, one such ETP protein, has been shown to genetically interact with PcG factors in Hox gene repression [29, 40, 41] and colocalizes with the majority of Psc-bound polytene chromosome sites [42]. Thus, we hypothesized that Corto may also function in regulating neuroblast competence. To test this, we focused on the NB7-1, one of the thirty neuroblast lineages of the embryonic VNC, for which we have markers to identify the progeny as well as have detailed knowledge of the temporal progression of its lineage and competence [9, 11, 12, 16, 23]. Briefly, we took advantage of the GAL4/UAS system [43] and used the engrailed GAL4 (en-GAL4) driver to drive strong, continuous Hb expression in NB7-1 throughout its entire lineage (en > hb) and determined the number of early-born neurons generated, a quantitative measure of the length of the early competence window. To distinguish between the endogenously activated Hb protein in response to early-born identity specification and the Hb protein misexpressed from the UAS-hb construct, we implemented an endogenously encoded bacterial artificial chromosome (BAC) transgene that includes the hb genomic locus and all of its regulatory enhancers. This BAC has been modified to include a hemagglutinin epitope tag (HA) fused to Hb, and our previous work has shown that this BAC insertion is sufficient to rescue a hb null animal and shows the same expression pattern as the native hb gene [11]. Thus, we are able to use HA as a proxy for endogenous Hb expression. We co-stained Hb and Even skipped (Eve), a marker for the U motoneuron progeny of the NB7-1 lineage, and quantified the number of early-born neurons (Eve+HA+) produced in response to continuous Hb misexpression. We note that in the competence assay, the degree of neuroblast response to Hb misexpression is variable even among homologous neuroblast lineages within an individual embryo. Thus, the length of the early competence window associated with a particular genotype is represented as an average of all lineages quantified. Compared to 5.1 ± 1.0 Eve+HA+ neurons in wild type NB7-1, we found 8.0 ± 2.3 early-born neurons in cortoL1 mutants (p < 0.0001). To control for any possible background mutations, we also examined transheterozygous animals of two independent loss of function alleles, corto L1 and corto420 [44] and observed consistent results, with 7.6 ± 2.7 early-born neurons generated (p < 0.0001) (Fig. 1B-D), indicating a prolonged early competence window. Thus, Corto plays an essential role in terminating early competence at mid-embryogenesis.
hb relocation to neuroblast nuclear lamina is delayed in corto mutant neuroblasts
We previously showed that the early competence window closes when the hb gene relocates to the neuroblast nuclear lamina at mid-embryogenesis [11, 23]. Thus, given the extension in early competence in corto mutants, we next examined hb gene positioning in corto mutant neuroblasts using in vivo immuno-DNA Fluorescence in situ Hybridization (FISH). Embryos were hybridized with a fluorescent DNA probe generated against approximately 10 kb at the hb gene locus and subsequently immunostained with Lamin Dm0, a B-type lamin intermediary filament that labels the nuclear envelope, and Worniu, a neuroblast-specific transcription factor. In corto mutant neuroblasts, we found that the proportion of hb gene loci at the nuclear lamina, which we defined as FISH signals that pixel-overlapped with lamin signals, was significantly reduced at mid-embryogenesis (stage 12: 56.9 ± 10.1% at lamina in wild type versus 33.7 ± 6.7% in corto−/−, P = 0.0004; stage 13/14: 56.4 + 6.6% at lamina in wild type versus 43.3 ± 2.8% in corto−/−, P = 0.0006) (Fig. 2A,B). We observed a reduction in hb localization to the nuclear lamina in corto mutant neuroblasts even when we included FISH signals that were near, but not touching the lamina (≤ 0.4µm from lamin) (Fig. 2B-ii). Interestingly, the difference between wild type and corto mutants diminished over time, and by late stage 15, there was no discernible difference in hb gene-lamina association between the genotypes (stage 15: 57.3 ± 3.8% at lamin in wild type versus 57.0 ± 3.3% in corto −/− mutants, P = not significant) (Fig. 2B). Thus, the data show that in corto mutants hb gene relocation to the neuroblast nuclear lamina is delayed. It is worth noting that we do not observe an all-or-nothing association of the hb gene with the nuclear lamina, and even at stage 11, during the early competence window, ~ 25% of the hb gene loci are localized at the lamina. In fact, reports from others show that even genomic regions outside of lamina-associated domains are located at the periphery ~ 30% of the time [45,46,47], underscoring the highly dynamic and non-static nature of the genome [48]. Importantly, we observe a two-fold increase in hb gene association with the neuroblast nuclear lamina at mid-embryogenesis in wild type embryos that has functional consequences on neuroblast competence, and this relocation is delayed in corto mutants.
To confirm that the delay in hb gene relocation in corto mutants is not due to a more general delay in neuroblast divisions, we co-stained embryos with Deadpan (Dpn), a pan-neuroblast marker, and phospho-histone 3 (PH3), a marker for dividing cells. There was no difference in numbers of dividing neuroblasts at either stage 12 or 14. Embryonic neuroblasts gradually divide more slowly as neurogenesis progresses, and this dynamic was unchanged in corto mutants, suggesting that neuroblast proliferation is progressing normally (Figure S1).
Normal Hb temporal dynamics in corto mutant neuroblasts
While genes localized at the nuclear lamina are typically in a repressed or silenced state, actively transcribed genes are often localized interiorly [49]. If the delay in hb gene relocation to the neuroblast nuclear lamina in Corto mutant embryos is due to prolonged hb transcription, we would expect a concomitant increase in the number of early-born neurons produced. By immunostaining for Eve, we did observe on occasion ectopic early-born neurons, but the occurrence was rare (2 Eve+Hb+ neurons in wild type compared to an average 2.1 in cortoL1/420 mutants) (Figure S2). Drosophila embryo development is highly stereotyped, allowing us to use morphological characteristics to compare stage-matched animals of different genotypes, and we found no qualitative differences in the timing of Hb repression in neuroblasts between wild type and corto mutants (Fig. 3A). By late stage 12, Hb is expressed by only a few neuroblasts in the thoracic segments (Fig. 3A, arrowheads) and is not detectable in any neuroblasts at stage 13, but rather is expressed in the neuronal layer, deeper in the embryo (Fig. 3B). Upon quantifying the proportion of Hb+ neuroblasts between stage 11 and 12, when Hb is rapidly being repressed across the neuroblast population, we measured no difference between wild type and corto mutants, indicating Hb was downregulated in a timely manner (Fig. 3C). Further arguing against a strong role for Corto as a hb transcriptional repressor, we found that Corto overexpression in neuroblasts had no effect on the production of early-born neurons (Figure S3), in contrast to hb’s known transcriptional repressor, Svp, which reduces the number of early-born neurons upon overexpression [21]. Thus, the impaired hb gene-lamina relocation phenotype in corto mutant neuroblasts we observed at stages 12–14 (7-11 h after egg lay, AEL), is well after hb is already transcriptionally downregulated, and the relocation phenotype is not due to prolonged hb transcription (prior to 7 h AEL) that keeps it in the nuclear interior.
Psc and Corto genetically interact to restrict the early competence window
Corto has been reported to interact with PcG chromatin factors to maintain gene repression [29, 40]. In particular, Corto shares many binding sites on polytene chromosomes with Psc [40, 42], which plays a central role in PRC1 silencing activity [27]. We asked whether Corto and Psc genetically interact to terminate competence by testing the early competence window in transheterozygous mutant embryos. We examined embryos stages 14–16, the tail end of the NB7-1 lineage, and found that while neither Psc heterozygotes or corto heterozygotes alone have a competence phenotype (P value not significant), Psch27/+;cortoL1/+ transheterozygous animals have prolonged early competence (Fig. 4A) (Psch27/+;cortoL1/+ 6.4 ± 1.5 vs. wild type 5.4 + 1.1, P = 0.0004). Given the variability in response to Hb misexpression even among homologous neuroblasts within the same embryo, we additionally displayed the data as a scatterplot, showing the number of early-born neurons relative to Eve+ neurons, representing the total cells in the lineage upon Hb misexpression (Fig. 4A-ii). Psch27/+;cortoL1/+ transheterozygous neuroblasts generated, overall, a higher fraction of early-born neurons than wild type neuroblasts with similar numbers of Eve+ cells, indicating increased competence among neuroblasts with comparable levels of Hb misexpression. We note that for the stages analyzed, the number of Eve+ cells is higher in older embryos, as expected (stage 14: 18.7 ± 2.9 Eve+ cells at stages 14, n = 68 lineages from 5 embryos, vs stage15/16: 23.0 ± 3.6, n = 136 lineages from 8 embryos, p < 0.0001), whereas the number of Eve+HA+ (early-born) neurons did not correlate with stage (stage 14: 5.9 ± 1.5 Eve+HA+ neurons vs stage 15/16: 5.6 ± 1.4, not significant). This confirms that all embryos sampled had completed production of early-born neurons. Together, we conclude that Psc and Corto genetically interact to close the competence window to specify early-born neurons.
In contrast, overexpression of Corto did not have any effect on competence. While wild type animals had an average of 5.5 ± 0.3 Eve+HA+ neurons upon Hb misexpression (68 hemisegments quantified from n = 4 embryos), co-overexpression of Corto did not yield a statistically different result, 6.1 ± 0.3 Eve+HA+ neurons (86 hemisegments quantified from n = 5 embryos) (Fig. 4C). If Corto acts in neuroblasts primarily through its interactions with the PcG complex, perhaps the stoichiometry or activity level of one or more PcG complex subunits is rate-limiting.
Corto does not act to maintain repression of Hb or the Hox gene, Abdominal B, in neuroblasts
PcG factors have been well studied for their roles in maintenance of target gene repression. As an enhancer of PcG, Corto has been reported to play a similar role in the repression of Abd-B, a known PcG Hox target gene, and loss of Corto results in its derepression [41]. Our recent study [23] showed that the hb intron region is a strong PcG target site, and loss of PRC1 impairs hb gene-lamina relocation. Interestingly, however, loss of PcG does not result in Hb derepression in neuroblasts even in late stage embryos. Moreover, even the most severe loss of function mutants for PRC1 (Psc-Su(z)2P3C) and PRC2 (maternal and zygotic mutants of extra sex combs, escmat/zyg), did not show any neuroblast derepression of Hb in late-stage embryos [23], raising the possibility that PcG factors may not function in transcriptional repression in this cell type. We thus examined Hb expression in late-stage corto mutant embryos to determine whether Hb becomes derepressed. Similar to the PRC1 mutants, we also did not find Hb derepressed in neuroblasts of late stage corto mutant embryos (Fig. 5A). While we did observe Abd-B derepression in corto mutants, consistent with previous observations [41], surprisingly, the derepression was limited to the epithelia, and we did not find Abd-B derepressed in neuroblasts (Fig. 5B-C). In summary (Fig. 6), our results show that Corto is required for the timely relocation of the hb gene to the neuroblast nuclear lamina and genetically interacts with PRC1 to terminate early competence. However, it does not play a role in maintaining hb gene repression, and in a departure from its known function, it does not play a role in repression of its known target Abd-B, within neuroblasts.