The Ets protein Pointed prevents both premature differentiation and dedifferentiation of Drosophila intermediate neural progenitors

The generation of brain complexity in higher order animals involves the production of intermediate neural progenitors (INPs) from neural stem cells (NSCs) (Lui et al., 2011; Pontious et al., 2008). INPs transiently proliferate to amplify the NSC output. However, INPs have very limited developmental potential and can generate only fate-restricted progeny. Defects in maintaining INP proliferation owing to precocious differentiation and cell cycle exit could lead to reduced brain complexity and brain malformation (Colasante et al., 2015; Quinn et al., 2007; Reillo et al., 2011), whereas the aberrant dedifferentiation of INPs may result in tumorigenic overgrowth (Liu et al., 2011; Walton et al., 2009). Therefore, INPs need to avoid not only differentiation but also dedifferentiation in order to produce a precise number of progeny with specific cell fates. It is crucial to decipher the mechanisms that prevent INP differentiation and dedifferentiation in order to understand the generation of brain complexity and the formation of brain tumors.

In Drosophila larval brains, INPs have similar roles to mammalian INPs in amplifying neuronal output from type II neuroblasts (NBs). Unlike ganglion mother cells (GMCs) generated from type I NBs, which divide only once and produce two neurons (Hartenstein et al., 2008), each INP generated from type II NBs produces ∼10 neurons by dividing in multiple rounds, thereby generating several GMCs (Bello et al., 2008; Boone and Doe, 2008; Bowman et al., 2008; Wang et al., 2014). INPs must maintain their self-renewal and avoid differentiation while dividing to produce neurons. However, newly generated INPs are immature and are prone to dedifferentiate into NBs if they fail to differentiate into mature INPs (Bowman et al., 2008; Eroglu et al., 2014; Koe et al., 2014). Thus, Drosophila INPs provide an excellent model for studying how INPs avoid dedifferentiation and premature differentiation.

In recent years, studies have begun to identify key genetic programs that prevent the dedifferentiation or premature differentiation of INPs. For example, preventing the dedifferentiation of immature INPs (imINPs) requires the cell fate determinants Brain tumor (Brat) and Numb, the Fez family transcription factor Earmuff (Erm), as well as the SWI/SNF and Histone deacetylase 3 (HDAC3) chromatin remodeling complexes (Bowman et al., 2008; Eroglu et al., 2014; Koe et al., 2014; Weng et al., 2010). The loss of any of these factors could result in imINP dedifferentiation and supernumerary type II NBs. By contrast, preventing the premature differentiation of INPs requires the suppression of the homeodomain protein Prospero (Pros) in INPs. Our recent studies show that the Sp8 family transcription factor Buttonhead (Btd) is required to suppress Pros in imINPs (Xie et al., 2014). The loss of Btd results in ectopic Pros expression in imINPs and the premature differentiation of INPs into GMCs. However, our understanding of the mechanisms that prevent the dedifferentiation and premature differentiation of INPs is still incomplete.

Our previous studies revealed that the Ets family transcription factor Pointed P1 (PntP1) is specifically expressed in type II NBs and imINPs (Zhu et al., 2011). Functional analyses show that inhibiting PntP1 activity by its antagonist Yan (Anterior open – FlyBase) transforms type II NBs into type I NBs and eliminates INPs, whereas PntP1 misexpression transforms type I NBs into type II NBs and promotes INP generation, suggesting that PntP1 is required for type II NB specification and INP generation (Zhu et al., 2011). However, it is still unknown exactly how PntP1 promotes of INP generation

In this study, we investigated PntP1 function in imINPs by taking advantage of our novel pntP1-specific mutant alleles and using RNAi knockdown. We show that PntP1 prevents both the dedifferentiation and premature differentiation of INPs by regulating the expression of distinct target genes at different stages of imINP development. Our work reveals the mechanistic details of the PntP1-mediated generation of INPs in type II NB lineages and provides novel insights into the mechanisms that maintain the balance between INP dedifferentiation and differentiation.

The only currently available pntP1-specific allele is a null allele, pnt^Δ33, in which the pntP1-specific exon 1, including part of the coding region, is deleted (O'Neill et al., 1994). However, pnt^Δ33 is not ideal for loss-of-function phenotypic analyses in Drosophila larval type II NB lineages for two reasons. First, PntP1 proteins perdure in pnt^Δ33 mutant type II NB clones even at late third instar larval stages (Zhu et al., 2011). Second, the embryonic lethality of pnt^Δ33 makes it impossible to examine type II NB lineage development in pnt^Δ33 homozygous mutant larvae. Therefore, we decided to generate novel pntP1-specific mutant alleles using CRISPR/Cas9 technology (Jinek et al., 2012; Port et al., 2014; Sebo et al., 2014) to investigate how PntP1 regulates INP generation. We chose two 20 nt sequences that are 249 and 536 bp downstream of the translation start codon as gRNA targets (Fig. 1Aa,b). We generated two mutant lines, named pntP1⁸² and pntP1⁹⁰, which carry small indels. These indels cause frame shifts and premature stop codons, resulting in deletions of 535 and 445 amino acids at the C-terminus, including the Ets DNA-binding domain, in the pntP1⁸² and pntP1⁹⁰ mutants, respectively (Fig. 1Ab,c). Both pntP1⁹⁰ and pntP1⁸² homozygotes are late third instar larval lethal, indicating that they are likely to be hypomorphic alleles.

We then examined type II NB lineages in pntP1⁹⁰ and pntP1⁸² homozygous mutant larvae in order to investigate how the loss of PntP1 would affect type II NB lineage development. In wild-type larval brains, there are only eight type II NBs per lobe, which can be distinguished from type I NBs by the absence of the proneural protein Asense (Ase). Each type II NB lineage contains ∼20-25 mature INPs, which express both Ase and the bHLH protein Deadpan (Dpn) (Fig. 1B,B′,I-K,P) (Bello et al., 2008; Boone and Doe, 2008; Bowman et al., 2008). We found that in pntP1⁹⁰ mutant brains, the number of Ase⁻ type II NBs is doubled. However, the number of INPs was reduced by 50% and ∼25% of Ase⁻ type II NBs did not produce mature INPs. Instead, only a few Ase⁺ Dpn⁻ cells that resembled GMCs were generated (Fig. 1C,C′,I-K,P). In pntP1⁸² mutants, the loss of INPs was even more severe. There were only two or three INPs per Ase⁻ NB on average and ∼75% of the Ase⁻ type II NBs did not produce INPs (Fig. 1D,D′,I-K,P). However, in contrast to the increase in the number of Ase⁻ type II NBs in pntP1⁹⁰ mutants, the number of Ase⁻ type II NBs was reduced to about six per lobe in the pntP1⁸² mutants. Because PntP1 is required to suppress Ase expression in type II NBs (Zhu et al., 2011), it is possible that some type II NBs might have become Ase⁺ in the pntP1⁸² mutants (and possibly in pntP1⁹⁰ mutants as well), as observed in Pnt knockdown type II NBs (see below).

To verify whether the phenotypes observed in the pntP1⁹⁰ and pntP1⁸² mutants are indeed caused by mutations of pntP1, we examined phenotypes in pntP1⁹⁰/pnt^Δ88 or pntP1⁸²/pnt^Δ88 transheterozygotes and compared them with phenotypes in Pnt knockdown type II NB lineages. pnt^Δ88 carries a deletion that covers the entire pntP1 transcript, including the 3′ exons shared by other pnt transcripts (Brunner et al., 1994). We found that pntP1⁹⁰/pnt^Δ88 mutants had similar, but more severe, phenotypes compared with pntP1⁹⁰ mutants (Fig. 1E,E′,I-K,P). In pntP1⁹⁰/pnt^Δ88 mutants, the number of Ase⁻ type II NBs was further increased to ∼25 per lobe and the number of INPs was reduced by nearly 70%. In the pntP1⁸²/pnt^Δ88 mutants, we observed a similar loss of INPs as in the pntP1⁸² mutants, but a slight increase in the number of Ase⁻ type II NBs (Fig. 1F,F′,I-K,P). Consistently, we observed a 50-100% increase in the total number of type II NBs and the complete elimination of INPs in over 90% of type II NB lineages when Pnt was knocked down by two independent UAS-pnt RNAi lines driven by type II NB lineage-specific pntP1-GAL4. Furthermore, Ase was ectopically expressed in 60-80% of Pnt knockdown type II NBs (Fig. 1G-H′,L-P), which is in accord with our previous results showing that the Yan-mediated inhibition of Pnt activity leads to ectopic ase activation in type II NBs. These data show that the loss of PntP1 leads to not only the transformation of type II NBs into type I NBs as we previously reported (Zhu et al., 2011) but also the elimination of INPs and the generation of extra type II NBs (Fig. 1P).

We then examined the expression of Pros in imINPs and the identity of the Ase⁺ progeny in pnt mutant or Pnt knockdown type II NB lineages without INPs to investigate why the loss of PntP1 led to the elimination of mature INPs. During INP maturation, newly generated imINPs first differentiate from an Ase⁻ state to an Ase⁺ state before they fully mature. Pros is not normally expressed in type II NBs or imINPs (Fig. 2A-A″) (Bowman et al., 2008). Our recent studies show that in the absence of Btd, Ase⁻ imINPs ectopically express nuclear Pros, which promotes the premature differentiation of INPs into GMCs and eliminates mature INPs (Xie et al., 2014). Therefore, we wondered whether the elimination of INPs resulting from the loss of PntP1 was also due to the Pros-mediated premature differentiation of INPs into GMCs, which express both Ase and nuclear Pros. We examined Pros expression in the Pnt knockdown and pntP1⁸² mutant type II NB lineages that showed a more consistent loss of INPs. Indeed, unlike in normal type II NB lineages, nuclear Pros was consistently expressed in newly generated Ase⁻ imINPs, which can be identified as Ase⁻ cells adjacent to the NBs, in the Pnt knockdown and pntP1⁸² mutant type II NB lineages. Furthermore, the Ase⁺ progeny that were generated in lineages without mature INPs also expressed nuclear Pros, indicating that they were GMCs (Fig. 2B-D″). These results suggest that PntP1 is required to suppress Pros expression in newly generated imINPs and to prevent imINPs from prematurely differentiating into GMCs.

Given that the main function of Pros is to promote cell cycle exit and differentiation (Choksi et al., 2006; Maurange et al., 2008), it is very likely that the loss of INPs resulting from the loss of PntP1 is due to the ectopic expression of nuclear Pros in imINPs. To test this, we examined whether reducing Pros expression could rescue the loss of INPs in Pnt knockdown type II NB lineages. Therefore, we knocked down Pnt in a pros¹⁷ heterozygous background. In the pros¹⁷ heterozygous mutants, type II NB lineages produced similar numbers of INPs as in the wild type (Fig. 3A-B″,E,F). However, unlike the Pnt knockdown in the wild-type background, which eliminated INPs in nearly all type II NB lineages, Pnt knockdown in pros¹⁷ heterozygotes only eliminated INPs in ∼50% of lineages and there were still an average of about six INPs per lineage (Fig. 3C-F), suggesting that reducing Pros expression partially rescued the loss of INPs in the Pnt knockdown type II NB lineages. These results indicate that the ectopic nuclear Pros expression in imINPs is responsible for the loss of INPs that results from PntP1 loss.

In addition to the rescue of INPs, we found that reducing Pros expression partially restored the suppression of Ase in Pnt knockdown type II NBs. When Pnt was knocked down in the wild-type background, only 25% of type II NBs remained Ase⁻. However, when Pnt was knocked down in the pros¹⁷ heterozygous background, which still increased the total number of type II NBs to 15.8±3.1 per lobe (mean±s.d., n=13), ∼70% of type II NBs were Ase⁻ (Fig. 3C-D″,G). Because nuclear Pros is not detected in the Pnt knockdown type II NBs, the restoration of Ase suppression in Pnt knockdown type II NBs is likely to be an indirect effect. We previously reported that maintaining PntP1 expression in type II NBs may require a feedback signal from INPs, and the loss INPs could lead to a loss/reduction of PntP1 expression in type II NBs (Xie et al., 2014). Therefore, one possibility is that the rescue of INPs by the reduction of Pros might help restore the feedback signal, which might in turn partially restore the expression of PntP1 and the suppression of Ase in the NBs, as occurs in btd mutant clones (Xie et al., 2014). However, we could not detect PntP1 proteins after knocking down PntP1 in either the wild-type or pros¹⁷ heterozygous background, suggesting that the partially restored PntP1 expression could still be below the detection limit (data not shown)

Because the loss of PntP1 or Btd leads to similar Pros-mediated premature differentiation of INPs, we next investigated whether PntP1 and Btd function in the same pathway by performing genetic interaction tests. We examined whether reducing the expression of Btd would enhance the loss of INPs in pntP1⁹⁰ homozygous mutants, which show a less severe loss of INPs and might be more sensitive to the reduction in Btd expression. We used btd^XG81 or btd-GAL4 heterozygous mutants to reduce Btd expression. btd^XG81 is a missense loss-of-function allele, and btd-GAL4 is a lethal p{GAL4} insertion allele, in which GAL4 is integrated into the btd promoter (Estella and Mann, 2010; Wimmer et al., 1993). We labeled type II NB lineages with mCD8-GFP driven by btd-GAL4, which is only expressed in type II NB lineages on the dorsal side of larval brains (Xie et al., 2014), to quantify the number of INPs in btd-GAL4/+; pntP1⁹⁰ mutants. For the pntP1⁹⁰ or btd^XG81/+; pntP1⁹⁰ mutants, we focused on Ase⁻ type II NBs for the quantification because we could not identify type II NBs if they became Ase⁺. In the pntP1⁹⁰ homozygous mutants the average number of INPs per Ase⁻ NB was reduced by ∼50%, and ∼25% of the Ase⁻ NBs did not have associated INPs (Fig. 4A,A′,C,C′,G,H), whereas btd^XG81 or btd-GAL4 heterozygous mutants did not exhibit an obvious loss of INPs (Fig. 4B,B′,E,E′,G,H). However, the loss of INPs in the pntP1⁹⁰ mutant type II NB lineages was dramatically enhanced in the btd^XG81 or btd-GAL4 heterozygous mutant background. In the btd-GAL4/+; pntP1⁹⁰ or btd^XG81/+; pntP1⁹⁰ mutants, the number of INPs was reduced by more than 80%, and ∼90% of NBs did not have associated INPs (Fig. 4D,D′,F,F′,G,H). The enhancement of INP loss in pntP1⁹⁰ mutants by the reduction of Btd suggests that Btd and PntP1 genetically interact to prevent the premature differentiation of INPs and that Btd and PntP1 are likely to function in the same pathway.

Interestingly, in addition to enhancing the loss of INPs, reducing Btd expression also promoted ectopic activation of ase in pntP1⁹⁰ mutant type II NBs. The pntP1⁹⁰ mutants usually had ∼13 Ase⁻ NBs per lobe. However, the number of Ase⁻ type II NBs was reduced to four to six per lobe in the btd^XG81/+; pntP1⁹⁰ or btd-GAL4/+; pntP1⁹⁰ mutants, although the btd^XG81 and btd-GAL4 heterozygotes had the same number of Ase⁻ type II NBs as the wild type (Fig. 4C-F′,I,J). By labeling type II NB lineages with mCD8-GFP driven by btd-GAL4, we found that the total number of type II NBs in the btd-GAL4/+; pntP1⁹⁰ mutants was still significantly increased compared with the wild type, similar to the pntP1⁹⁰ mutants, but over 60% of the type II NBs were Ase⁺ (Fig. 4F,F′,I,J), indicating that the reduction of Ase⁻ type II NBs in the btd-GAL4/+; pntP1⁹⁰ mutants (and probably in the btd^XG81/+; pntP1⁹⁰ mutants as well) was due to the ectopic activation of ase in the NBs. This ectopic ase activation is likely to have occurred throughout type II NB lineage development because the ectopic Ase expression could already be observed at 1 day after larval hatching, and the number of Ase⁺ type II NBs continued to increase as the total number of type II NBs increased from early to late larval stages (Fig. S1). The ectopic ase activation in pntP1⁹⁰ mutant type II NBs resulting from the reduction of Btd expression suggests that PntP1 and Btd function cooperatively to suppress Ase expression in type II NBs.

In addition to the elimination of INPs, the loss of PntP1 also increased the number of type II NBs. We next investigated the cellular origin of the extra type II NBs. Several previous studies have shown that imINPs are not fully committed to their cell fate and can revert back to the NB fate in the absence of tumor suppressors such as Brat, Numb or Erm (Bowman et al., 2008; Weng et al., 2010). Because PntP1 is highly expressed in imINPs (Zhu et al., 2011), we wondered whether the generation of ectopic type II NBs resulting from the loss of PntP1 was also due to imINP dedifferentiation. To test this, we tried to knockdown Pnt in imINPs using erm-GAL4(III) or erm-GAL4(II) (Xiao et al., 2012). erm-GAL4(III) is mainly expressed in Ase⁺ imINPs, whereas erm-GAL4(II) is expressed in both Ase⁻ and Ase⁺ imINPs, except for the newly generated Ase⁻ imINPs. Our results showed that Pnt knockdown by erm-GAL4(III) did not produce any extra type II NBs in third instar larval brains (Fig. 5A-B″,E), whereas Pnt knockdown by erm-GAL4(II) significantly increased the number of type II NBs to ∼16 per brain lobe (Fig. 5C-E). These results suggest that the generation of extra type II NBs resulting from the loss of PntP1 is likely to be due to dedifferentiation of Ase⁻ imINPs into type II NBs. The dedifferentiation of imINPs is likely to occur at a low frequency because the total number of type II NBs was only increased by 100-200% at the third instar larval stages when pntP1 was mutated or knocked down, and individual ectopic type II NB lineages were often well separated (e.g. Fig. 1G-H′, Fig. 4F). However, the dedifferentiation could occur at any stage during development, as indicated by the existence of lineages with multiple NBs at early and late larval stages (e.g. the second GFP-labeled lineage from the left in Fig. 4F and the second GFP-labeled lineage from the right in Fig. S1A).

Why does the loss of PntP1 lead to the dedifferentiation of imINPs into type II NBs? Our previous studies show that the misexpression of PntP1 in type I NB lineages is sufficient to induce the expression of Erm in INP-like cells, but that inhibiting PntP1 activity with Yan abolishes the expression of Erm in type II NB lineages (Zhu et al., 2011). Furthermore, Erm and PntP1 are co-expressed in imINPs (Janssens et al., 2014; Zhu et al., 2011), and loss of Erm similarly leads to an increase in the number of type II NBs. Therefore, Erm could be a PntP1 target, and the dedifferentiation of imINPs resulting from the loss of PntP1 could be due to the loss of Erm in imINPs. To test this, we examined Erm expression in Pnt knockdown type II NB lineages by immunostaining. In normal type II NB lineages, Erm is expressed in Ase⁻ and Ase⁺ imINPs (Fig. 6A-A″) (Janssens et al., 2014). However, we did not observe any obvious Erm staining in Pnt knockdown type II NB lineages, even when Ase was not ectopically expressed in the NBs (Fig. 6B-D), indicating that the expression of Erm was largely abolished by Pnt knockdown and the loss of Erm expression was not due to the transformation of type II NBs into type I-like NBs. The loss of Erm expression in Pnt knockdown type II NB lineages provides evidence to support the hypothesis that Erm is a PntP1 target.

To determine whether the loss of Erm expression was indeed responsible for the dedifferentiation of imINPs resulting from the loss of PntP1, we tested genetic interactions between erm and pntP1 by examining whether reducing Erm expression would further increase the generation of extra type II NBs resulting from the partial loss of PntP1 and if restoring Erm expression would suppress the generation of extra type II NBs. If the dedifferentiation of imINPs resulting from the loss of PntP1 is indeed due to the loss or reduction of Erm expression, then removing one wild-type copy of the erm gene would further increase the number of type II NBs. By contrast, restoring Erm expression in imINPs should suppress the dedifferentiation of imINPs and thus reduce the number of type II NBs.

Indeed, we found that although erm² heterozygous mutants did not have any extra type II NBs (Fig. 7A,B,K), knockdown of Pnt in erm² heterozygous mutants led to a significant increase in the total number of type II NBs (including both Ase⁻ and Ase⁺ type II NBs) compared with Pnt knockdown in the wild-type background. Similar increases were observed when two independent UAS-pnt RNAi lines were used (Fig. 7C-F,K). Consistently, the removal of one wild-type copy of erm doubled the number of Ase⁻ type II NBs in pntP1⁹⁰ mutant larvae (Fig. 7G,H,L). By contrast, when erm-GAL4(II) drove the expression of UAS-erm in imINPs, the generation of extra type II NBs in the pntP1⁹⁰ mutants was significantly suppressed (Fig. 7I,J,L). The total number of Ase⁻ type II NBs in the pntP1⁹⁰ mutants was reduced to ten per lobe when UAS-erm was expressed in imINPs, suggesting that restoring Erm expression largely prevents the dedifferentiation of imINPs.

Taken together, the enhancement and the suppression of the generation of ectopic type II NBs by reducing or maintaining Erm expression, respectively, indicate that the loss of Erm expression accounts for the dedifferentiation of imINPs into type II NBs that results from the loss of PntP1.

We have previously shown that PntP1 is required for the generation of INPs in type II NB lineages (Zhu et al., 2011), but the underlying mechanisms have not been elucidated. We show in this study that PntP1 has at least two distinct roles in INPs: preventing the premature differentiation of INPs into GMCs by acting together with Btd to suppress Pros expression in newly generated imINPs; and preventing the dedifferentiation of INPs into type II NBs by activating Erm expression late in imINP development. Therefore, PntP1 is able to maintain the balance between INP differentiation and dedifferentiation by functioning through two distinct pathways (Fig. 7M).

Our results show that a reduction in PntP1 function leads to both a loss of INPs and an increase in the number of type II NBs without completely transforming all type II NBs into type I-like NBs. The loss of INPs is particularly obvious in the pntP1⁸² mutant and Pnt knockdown type II NB lineages, possibly because there is a more severe reduction of PntP1 function in the pntP1⁸² mutants or after Pnt knockdown. We provide several lines of evidence to demonstrate that the loss of INPs is due to the Pros-mediated premature differentiation of INPs. First, the loss of INPs is not due to the transformation of type II NBs into type I NBs because the loss of INPs occurs independently of the ectopic Ase expression in the NBs. Second, when INPs are lost, the type II NBs generate GMCs instead. Third, nuclear Pros is ectopically activated in the newly generated imINPs when Pnt is lost. Fourth, the loss of INPs can be almost fully rescued by reducing Pros expression. Therefore, PntP1 normally inhibits Pros expression in the newly generated imINPs so that the imINPs can differentiate into mature INPs and undergo self-renewing divisions instead of becoming GMCs and exiting the cell cycle.

Although it remains to be investigated exactly how PntP1 suppresses Pros expression, our work suggests that it does so by functioning together with Btd, which is also required to suppress Pros in imINPs (Xie et al., 2014). We show that reducing Btd expression enhances the loss of INPs in pntP1⁹⁰ mutants. Although reducing Btd expression also enhances the ectopic activation of Ase in type II NBs, enhancement of INP loss also occurs in lineages without ectopic Ase expression. The genetic interaction between Btd and PntP1 suggests that they function in the same pathway to suppress Pros expression. However, it is unlikely that there is a direct regulatory relationship between Btd and PntP1 for several reasons. First, although Btd and PntP1 show similar expression patterns in type II NB lineages, Btd, but not PntP1, is also expressed in a subset of type I NB lineages. Second, we did not observe obvious changes in the expression of mCD8-GFP driven by btd-GAL4 in the pntP1⁹⁰ mutant or Pnt knockdown type II NB lineages (data not shown). Third, PntP1 expression is maintained in the majority of btd mutant type II NB clones (Xie et al., 2014). Fourth, PntP1 promotes the generation of INPs from type I NBs only when Btd is coexpressed, but not when PntP1 is expressed alone (Xie et al., 2014). It is well documented that Ets family proteins are able to bind to other transcription factors and that this partnership can enhance binding to the promoters of target genes and contribute to the functional specificity of Ets proteins (Hollenhorst et al., 2011). Therefore, one interesting possibility is that PntP1 and Btd physically interact and bind cooperatively to the promoters of their target genes.

In addition to preventing premature differentiation of INPs, our phenotypic analyses in the pntP1 mutants and Pnt knockdown type II NB lineages suggest that PntP1 also prevents imINP dedifferentiation. We show that there is a significant increase in the number of type II NBs in pntP1⁹⁰ homozygous or pntP1⁹⁰/pnt^Δ88 transheterozygous mutants or after knocking down Pnt. The extra type II NBs are likely to be derived from the dedifferentiation of Ase⁻ imINPs, as knocking down Pnt by erm-GAL4(II) but not erm-GAL4(III) leads to an increase in the number of type II NBs. We also provide several lines of evidence to show that the dedifferentiation of imINPs is at least in part due to the loss of Erm in imINPs. First, Erm expression in imINPs is lost after Pnt knockdown and the loss of Erm occurs independently of the transformation of type II NBs into type I NBs. Second, reducing the expression of Erm significantly enhances the generation of extra type II NBs resulting from the loss of PntP1, whereas maintaining Erm expression in Ase⁻ imINPs significantly suppresses the generation of extra type II NBs in the pntP1⁹⁰ mutants. Our results are consistent with a recent report showing that knockdown of Pnt by the type II NB driver wor-GAL4 ase-GAL80 increases the number of type II NBs, and reducing Pnt expression enhances the generation of ectopic type II NBs in erm mutants (Komori et al., 2014). However, our work identifies additional cellular and molecular mechanisms by which extra type II NBs are generated following the loss of PntP1.

One of the defining features of type II NBs is the lack of Ase expression. We previously reported that PntP1 was responsible for the suppression of Ase in type II NBs (Zhu et al., 2011). However, a recent study argues that PntP1 is only required for INP specification, but not type II NB specification, by showing that knockdown of Pnt by wor-GAL4 ase-GAL80 only increases the number of type II NBs but does not ectopically activate Ase in type II NBs (Komori et al., 2014). Here, we show that Pnt knockdown by pntP1-GAL4 is sufficient to ectopically activate Ase in ∼80% of type II NBs, confirming that PntP1 indeed specifies type II NB identity. The discrepancy could be due to differences in the efficiency of Pnt knockdown by RNAi involving different GAL4 drivers. However, our results show that reducing Pros expression not only rescues the loss of INPs but also restores the suppression of Ase expression in Pnt knockdown type II NBs, raising the question of whether PntP1 specifies type II NBs by directly acting in the NBs or by indirectly promoting INP generation. Our previous results show that PntP1 misexpression is able to suppress Ase expression in type I NBs, even if INPs are not generated, suggesting that PntP1 is likely to act in the NBs to suppress Ase expression. However, INPs are also likely to be involved in maintaining type II NB identity by providing a feedback signal to maintain PntP1 expression in the NB, as we proposed in a previous study (Xie et al., 2014). One candidate feedback signal could be the Notch ligand. Our recent studies suggest that Notch signaling is required to maintain PntP1 expression and type II NB identity (Li et al., 2016; Zhu et al., 2012). INPs could provide the ligand to activate Notch in the NBs, as suggested in a recent study (Song and Lu, 2011). This Notch-mediated feedback mechanism for maintaining neural progenitor cells is also conserved in mammals (Campos et al., 2001; Lui et al., 2011; Yoon et al., 2008).

However, it is surprising that Ase remains suppressed at least in a subset of pntP1⁹⁰ or pntP1⁸² mutant type II NBs. Given that PntP1 is a transcription factor, one might expect that the deletion of the Ets DNA-binding domain in the pntP1⁹⁰ and pntP1⁸² mutants would lead to a complete loss of PntP1 function and to ectopic activation of Ase in all type II NBs. However, our genetic data suggest they are likely to be hypomorphic alleles because the phenotypes in the pntP1⁹⁰ and pntP1⁸² mutants are weaker than in the pntP1⁹⁰/pnt^Δ88 transheterozygotes. Furthermore, the pntP1⁹⁰ and pntP1⁸² mutants show different degrees of severity in phenotype, which also indicates that deletion of the Ets DNA-binding domain does not completely abolish PntP1 function. Therefore, the truncated PntP1 protein generated in the pntP1⁹⁰ or pntP1⁸² mutants could still be partially functional and the DNA-binding domain of PntP1 might be dispensable for function. Our results show that reducing Btd expression significantly enhances the ectopic activation of Ase in pntP1⁹⁰ mutant type II NBs. Although we do not rule out the possibility that the enhanced ectopic Ase activation could be secondary to the enhanced loss of INPs and their feedback signal(s), Btd and PntP1 might actually function together to suppress Ase. In support of this notion, we previously showed that Btd overexpression was able to partially suppress Ase expression in a subset of type I NBs in larval brains (Xie et al., 2014). The truncated PntP1 proteins generated from the pntP1⁹⁰ or pntP1⁸² alleles might still be able to interact with Btd to regulate target gene expression.

In summary, our studies shed new light on the mechanistic details of the PntP1-mediated generation of INPs as well as type II NB specification. However, it remains to be investigated how PntP1 and Btd act together to specify type II NBs and inhibit Pros expression in INPs. Understanding this might rely on identification of the direct targets of PntP1 and Btd.

UAS-pnt RNAi lines (#31936 and #35038, Bloomington Drosophila Stock Center) were used for Pnt knockdown. Type II NB lineage-specific pntP1-GAL4 (named GAL4^14-94 previously) (Zhu et al., 2011) and erm-GAL4(II) or (III) (Pfeiffer et al., 2008; Xiao et al., 2012) were used to drive the expression of UAS-transgenes in type II NB lineages or in imINPs, respectively. Other fly lines include: pnt^Δ88/Tm6Tb (Brunner et al., 1994), which was used for pntP1 loss-of-function phenotypic analyses; pros¹⁷/TM6Tb (Doe et al., 1991) for reducing Pros expression; btd^XG81, FRT19A/FM7c, Kr-GFP and btd-GAL4, FRT19A/FM7c, Kr-GFP (Estella and Mann, 2010; Wimmer et al., 1993) for reducing Btd expression; and erm²/Cyo, act-GFP (Weng et al., 2010) for reducing Erm expression.

gRNA targets were selected using an online design tool (http://tools.flycrispr.molbio.wisc.edu/targetFinder/). The gRNA-expressing vectors were generated by a series of PCR reactions (see the supplementary Materials and Methods) (Sebo et al., 2014) and injected into y¹, P[vas-Cas9.S]ZH-2A, w¹¹¹⁸ (BDSC #52669) at 1 µg/ml (Rainbow Transgenic Flies, Camarillo, CA, USA) for generating mutant lines. Mutations were detected by PCR amplification of genomic DNAs followed by sequencing using pntP1-specific primers.

For RNAi knockdown or misexpression of transgenes, larvae were raised at 30°C after hatching. UAS-dcr2 was co-expressed with UAS-RNAi transgenes to enhance the efficiency of RNAi knockdown. Phenotypes were examined at third instar larval stages. To examine whether reducing Pros expression could rescue the loss of INPs resulting from Pnt RNAi knockdown, UAS-pnt RNAi was first recombined with pntP1-GAL4. Then, the UAS-pnt RNAi pntP1-GAL4 recombinant flies were crossed with either wild type or pros¹⁷/Tm6,Tb and their progenies raised at 25°C.

Dissection, fixation and immunostaining of larval brains were performed as described (Lee and Luo, 1999). Primary antibodies used for immunostaining include: guinea pig anti-Ase (1:5000; Brand et al., 1993) and rabbit anti-Dpn (1:500; Bier et al., 1992) (both gifts from Y. N. Jan), rat anti-mCD8 (Life Technologies, clone #5H10; 1:100), mouse anti-Pros (Developmental Studies Hybridoma Bank; 1:20) and rabbit anti-Erm (1:50; Janssens et al., 2014; a gift from H.Y. Wang). Secondary antibodies conjugated to Cy2, Cy3, Cy5 or DyLight 647 (Jackson ImmunoResearch) were used at 1:100 (Cy2) or 1:500 (Cy3, Cy5 or DyLight 647). A Zeiss LSM510 confocal microscope was used to acquire images, which were processed with Adobe Photoshop. Two-tailed Student's t-tests were used for statistical analyses.

We thank Drs Y. N. Jan, H. Y. Wang, C. Y. Lee, the Bloomington Drosophila Stock Center, and the Developmental Studies Hybridoma Bank for antibodies and fly stocks; M. Pan and X. Jackson for technical support; Dr F. Pignoni and S.Z. lab members for thoughtful discussion and comments.