Transient nuclear Prospero induces neural progenitor quiescence.

Stem cells can self-renew, differentiate, or enter quiescence. Understanding how stem cells switch between these states is highly relevant for stem cell-based therapeutics. Drosophila neural progenitors (neuroblasts) have been an excellent model for studying self-renewal and differentiation, but quiescence remains poorly understood. In this study, we show that when neuroblasts enter quiescence, the differentiation factor Prospero is transiently detected in the neuroblast nucleus, followed by the establishment of a unique molecular profile lacking most progenitor and differentiation markers. The pulse of low level nuclear Prospero precedes entry into neuroblast quiescence even when the timing of quiescence is advanced or delayed by changing temporal identity factors. Furthermore, loss of Prospero prevents entry into quiescence, whereas a pulse of low level nuclear Prospero can drive proliferating larval neuroblasts into quiescence. We propose that Prospero levels distinguish three progenitor fates: absent for self-renewal, low for quiescence, and high for differentiation.

Introduction

Quiescent stem cells preserve the stem cell pool for activation following disease or injury. Moreover, cancer stem cells can enter remission during treatment, possibly using mechanisms similar to stem cell quiescence, and re-initiate proliferation months to years later. Thus, studying stem cell quiescence may help design clinical approaches to prevent tumor dormancy and cancer recurrence, as well as to aid in activating stem cells for tissue repair.

Drosophila neuroblasts have served as a model system for identifying conserved signaling pathways that regulate stem cell proliferation and quiescence (Ebens et al., 1993; Datta, 1995; Britton and Edgar, 1998; Doe et al., 1998; Egger et al., 2008; Tsuji et al., 2008; Sousa-Nunes et al., 2010; Homem and Knoblich, 2012; Weng and Cohen, 2012). Drosophila neuroblasts delaminate from neuroectoderm during early embryogenesis, and then they go through multiple rounds of asymmetric cell division before exiting cell cycle at the end of embryogenesis; upon cell cycle exit, neuroblasts can undergo apoptosis (Karcavitch and Doe, 2005; Maurange and Gould, 2005; Ulvklo et al., 2012) or enter quiescence (Truman and Bate, 1988; Datta, 1995; Tsuji et al., 2008; Chell and Brand, 2010; Sousa-Nunes et al., 2011). During neuroblast asymmetric division, the scaffolding protein Miranda (Mira) is partitioned into the differentiating daughter cell, called a ganglion mother cell (GMC), and it carries at least three cargo proteins: the translational repressors Staufen and Brain tumor (Brat) and the transcriptional repressor Prospero (Hirata et al., 1995; Broadus and Doe, 1997; Ikeshima-Kataoka et al., 1997; Broadus et al., 1998; Schuldt et al., 1998; Shen et al., 1998; Bello et al., 2006; Betschinger et al., 2006; Lee et al., 2006). In the GMC, Mira appears to be degraded and its cargo is released into the cytoplasm (Staufen, Brat) or the nucleus (Prospero). Prospero is an atypical homeodomain protein that directly binds and represses progenitor and cell cycle genes to initiate GMC and neuronal differentiation (Li and Vaessin, 2000; Choksi et al., 2006); the mammalian ortholog Prox1 has a similar role in repressing cell cycle gene expression (Dyer, 2003; Foskolou et al., 2013). In addition, keeping high levels of Prospero out of the neuroblast nucleus is essential to prevent neuroblast differentiation (Choksi et al., 2006; Cabernard and Doe, 2009; Bayraktar et al., 2010).

In this study, we show that the Prospero differentiation factor is transiently detected at low levels in the neuroblast nucleus just prior to entry into quiescence. We find that loss of Prospero prevents entry into quiescence, whereas a pulse of low level nuclear Prospero drives proliferating larval neuroblasts into quiescence. We propose a model in which low levels of Prospero repress all cell cycle and progenitor genes except deadpan, whereas high levels of Prospero additionally repress deadpan to allow neural differentiation.

Results and discussion

Quiescent neuroblasts have a novel transcriptional profile lacking both progenitor and differentiation markers

To investigate the transition from a proliferating neuroblast to a quiescent neuroblast, we examined known proliferation/progenitor and differentiation marker expression during this transition. We hypothesized that loss of a progenitor marker or gain of a differentiation marker might induce neuroblast quiescence. As a model system we used the identified neuroblast 3–3 in the thoracic segments (NB3-3T) which reliably enters quiescence at embryonic stage 15 (Tsuji et al., 2008). Moreover, neuroblast 3–3 in abdominal segments (NB3-3A) remains proliferative until the end of embryogenesis and provides an excellent internal control (Tsuji et al., 2008). Previous work showed that the quiescent NB3-3T can be recognized by Deadpan, a basic helix-loop-helix transcription factor (Zhu et al., 2008; Chell and Brand, 2010). The coiled-coil Miranda protein is also reported to mark quiescent neuroblasts (Tsuji et al., 2008), which we confirm here (Figure 1), but note that mira transcription is down-regulated in quiescent neuroblasts based lack of mRNA in the stage 17 CNS in the BDGP in situ database (Tomancak et al., 2007). Interestingly, we find that all other tested progenitor markers—Worniu, Asense, and Cyclin E (CycE) (Brand et al., 1993; Caldwell and Datta, 1998; Ashraf et al., 2004; Lai et al., 2012)—are not expressed in the quiescent NB3-3T (Figure 1; data not shown), although they can be robustly detected in the proliferative NB3-3A (Figure 1). Similarly, larval quiescent neuroblasts lack expression of Worniu, Asense, CycE (and Miranda), while retaining Deadpan protein and active deadpan transcription (Figure 1—figure supplement 1). However, misexpression of Worniu (data not shown), Asense, or CycE (Figure 1—figure supplement 2) has no effect on the timing of NB3-3T quiescence.

Figure 1.

Quiescent neuroblasts have a novel transcriptional profile lacking both progenitor and differentiation markers.

(A–F) Proliferating NB3-3A contains nuclear Deadpan (A), cytoplasmic Mira (B), nuclear Asense (C), nuclear Worniu (D), and nuclear CycE (E) at embryonic stages 14 and 16; quantification is shown in (F). Neuroblast lineages are marked by Eg-GFP (Eg-gal4 UAS-mCD8:GFP), and the neuroblast is identified by Dpn (shown in insets) and indicated by arrows. Anterior side is up, and lateral is at right. Scale bar: 5 μm. (G–L) Proliferating NB3-3T expresses all above-mentioned neuroblast markers at stage 14 (S14). At stage 16 (S16), NB3-3T is quiescent and contains Deadpan (G) and Mira (H) but lacks Asense (I), Worniu (J) and CycE (K). Quantified in (L). Scale bar: 5 μm. (F and L) Quantification; number of neuroblasts scored shown in bar. (M) Schematic summary of neuroblast marker profiles of proliferating neuroblast, quiescent neuroblast and neuron.

DOI: http://dx.doi.org/10.7554/eLife.03363.003

(A) The newly hatched larval (NHL) central brain contains five persistently proliferative neuroblasts (the four mushroom body neuroblasts and the lateral antennal lobe neuroblast) and ∼100 quiescent central brain neuroblasts in each lobe. (B–E) Proliferative neuroblasts in the NHL contain progenitor markers Deadpan (Dpn), Asense (Ase), and Worniu (Wor) but lack the differentiation markers Prospero (Pros), Elav, and Repo and incorporate EdU (top). Quiescent neuroblasts maintain Dpn, but lack the progenitor markers and the differentiation markers, and fail to incorporate EdU (bottom).

DOI: http://dx.doi.org/10.7554/eLife.03363.004

No significant change in the number of proliferating NB3-3T in stage 15 embryos with overexpressed Asense (AseOXN) or Cyclin E (CycEOXN). Error bars, standard deviation; n.s., not significant. Fly genotype: AseOXN, worniu-gal4 UAS-asense; CycEOXN, worniu-gal4 UAS-cyclin E.

DOI: http://dx.doi.org/10.7554/eLife.03363.005

Figure 1—figure supplement 1.

Larval quiescent neuroblasts are Deadpan-positive but lack the progenitor markers Miranda, Asense, Cyclin E, Worniu and the differentiation markers Prospero, Elav, and Repo.

(A) The newly hatched larval (NHL) central brain contains five persistently proliferative neuroblasts (the four mushroom body neuroblasts and the lateral antennal lobe neuroblast) and ∼100 quiescent central brain neuroblasts in each lobe. (B–E) Proliferative neuroblasts in the NHL contain progenitor markers Deadpan (Dpn), Asense (Ase), and Worniu (Wor) but lack the differentiation markers Prospero (Pros), Elav, and Repo and incorporate EdU (top). Quiescent neuroblasts maintain Dpn, but lack the progenitor markers and the differentiation markers, and fail to incorporate EdU (bottom).

DOI: http://dx.doi.org/10.7554/eLife.03363.004

Figure 1—figure supplement 2.

Overexpressing Asense and Cyclin E does not change the timing of neuroblast quiescence.

No significant change in the number of proliferating NB3-3T in stage 15 embryos with overexpressed Asense (AseOXN) or Cyclin E (CycEOXN). Error bars, standard deviation; n.s., not significant. Fly genotype: AseOXN, worniu-gal4 UAS-asense; CycEOXN, worniu-gal4 UAS-cyclin E.

DOI: http://dx.doi.org/10.7554/eLife.03363.005

We next turned to examining differentiation markers. We assayed nuclear Prospero, which marks differentiating GMCs and young neurons in the larval CNS (Carney et al., 2013), Embryonic lethal and abnormal vision (Elav), which marks all neurons (Robinow and White, 1988); and Reversed polarity (Repo), which marks all non-midline glia (Xiong et al., 1994). We found that the mature quiescent NB3-3T did not express nuclear Prospero, Elav, or Repo (data not shown), nor were these differentiation markers observed in quiescent larval brain neuroblasts (Figure 1—figure supplement 1). Interestingly, we did observe transient low level nuclear Prospero in NB3-3T just prior to its entry into quiescence (see next section). We conclude that quiescent neuroblasts have a novel transcriptional profile that lacks both progenitor and differentiation markers.

A pulse of low level nuclear Prospero marks neuroblast entry into quiescence

Here, we characterize further our observation that NB3-3T showed low level nuclear Prospero at stage 15, concurrent with the entry into quiescence. Whereas NB3-3T showed a pulse of nuclear Prospero at stage 15 concurrent with entry into quiescence (Figure 2A,B), NB3-3A did not show nuclear Prospero and did not enter quiescence (Figure 2C,D). The levels of Prospero in the NB3-3T neuroblast are clearly detectable by immunofluorescent staining, but at significantly lower levels than in differentiating GMCs (Figure 2B, arrow; Figure 2—figure supplement 1). To probe the correlation of transient nuclear Prospero and neuroblast quiescence in more detail, we assayed temporal identity mutations that are known to shift the timing of NB3-3T quiescence: nubbin/pdm2 (pdm) mutants cause precocious NB3-3T quiescence, whereas castor (cas) mutants cause a delay in NB3-3T quiescence (Tsuji et al., 2008). Strikingly, the pulse of nuclear Prospero precisely matched the timing of neuroblast quiescence: it occurred earlier in pdm mutants, and later in cas mutants (Figure 2E,F). We conclude that a pulse of low level nuclear Prospero marks neuroblast entry into quiescence, suggesting a functional relationship.

Figure 2.

Transient low level nuclear Prospero is tightly correlated with neuroblast entry into quiescence.

(A–B) NB3-3T shows transient nuclear Prospero (Pros) during entry into quiescence at stage 15 (S15) (B). Differentiated daughter cell shows strong nuclear Prospero and is indicated by the arrow. Neuroblast lineages are identified by Eg-GFP (Eg-gal4 UAS-mCD8:GFP) (A), and the neuroblast is identified by Dpn and outlined by dashed lines. Summarized in schematic below (Prospero, green); quantified in (E). Anterior side is up, and lateral is at right. Scale bar: 5 μm. (C–D) Proliferating NB3-3A does not contain nuclear Prospero. Summarized in schematic below; quantified in (E). Scale bar: 5 μm. (E) Temporal identity factors schedule the timing of neuroblast entry into quiescence and the expression of nuclear Prospero. In cas mutant (cas), neuroblast quiescence is delayed, as is the timing of nuclear Prospero. In the deficiency allele Df(2L)ED773 which removes both nubbin/pdm2 (pdm mutant), NB3-3T precociously enters quiescence, and the timing of nuclear Prospero is advanced. (F) Schematic of the timing of neuroblast entry into quiescence and the expression of nuclear Prospero.

DOI: http://dx.doi.org/10.7554/eLife.03363.006

(A) Confocal images of a wild type NB3-3T (blue circle, identified by Eg-GFP and Dpn) and its adjacent GMC (yellow circle, identified by Eg-GFP and its smaller size) in stage 15 embryos. Scale bar: 5 μm. (B) Quantification of nuclear Prospero intensity in unit area in a neuroblast vs an adjacent GMC in stage 15 embryos. Prospero intensity in each confocal stack was measured with the open source software FIJI to obtain gray value and the nuclear area. The total Prospero intensity was summed from the confocal stacks, and then divided by nuclear volume to obtain the Prospero intensity per unit volume. Small triangles represent the intensity quantified from each individual neuroblast, and the large triangle is the average. Small circles represent the intensity quantified from each individual GMC, and the large circle is the average.

DOI: http://dx.doi.org/10.7554/eLife.03363.007

Figure 2—figure supplement 1.

Nuclear Prospero levels are lower in stage 15 NB3-3T than in differentiating GMCs.

(A) Confocal images of a wild type NB3-3T (blue circle, identified by Eg-GFP and Dpn) and its adjacent GMC (yellow circle, identified by Eg-GFP and its smaller size) in stage 15 embryos. Scale bar: 5 μm. (B) Quantification of nuclear Prospero intensity in unit area in a neuroblast vs an adjacent GMC in stage 15 embryos. Prospero intensity in each confocal stack was measured with the open source software FIJI to obtain gray value and the nuclear area. The total Prospero intensity was summed from the confocal stacks, and then divided by nuclear volume to obtain the Prospero intensity per unit volume. Small triangles represent the intensity quantified from each individual neuroblast, and the large triangle is the average. Small circles represent the intensity quantified from each individual GMC, and the large circle is the average.

DOI: http://dx.doi.org/10.7554/eLife.03363.007

Prospero is required for neuroblast quiescence

Having shown a strong correlation between the timing of nuclear Prospero and neuroblast entry into quiescence, we next asked if Prospero is required for neuroblast entry into quiescence and loss of nuclear Prospero can delay the timing of neuroblast quiescence. In theory, this can be done by quantifying the number of EdU+ neuroblasts in prospero null mutant embryos. The problem is that prospero mutants show de-differentiation of GMCs into proliferating neuroblasts (Choksi et al., 2006; Lee et al., 2006), making it necessary to distinguish the parental neuroblasts from the de-differentiated GMCs. To resolve the issue, we used the Notch signaling reporter, mγ-LacZ (Wech et al., 1999; Cooper et al., 2000), to identify the parental neuroblasts. During neuroblast asymmetric cell division, Miranda-Prospero and Partner of numb (Pon)-Numb protein complexes are independently segregated into the GMC to promote differentiation (Doe, 2008). In prospero mutants, Numb is still properly segregated to GMCs where it represses Notch activity (Spana and Doe, 1995) and consequently the Notch reporter mγ-LacZ is restricted to parental neuroblasts. mγ-LacZ is transcribed only in neuroblasts; in wild type the protein perdures into neuroblast progeny, whereas in prospero mutants the protein is restricted to neuroblasts because the progeny proliferate without expressing mγ-LacZ and thus dilute out the protein (Choksi et al., 2006; Figure 3A,B). Thus, we scored for EdU incorporation in the mγ-LacZ+ parental neuroblasts. Whereas in wild type nearly all neuroblasts ceased EdU incorporation by stage 16 (Figure 3A), many mγ-LacZ+ neuroblasts continued to incorporate EdU in prospero mutants at the same stage (Figure 3B). This strongly supports a model in which Prospero is required for neuroblast entry into quiescence. In addition, Prospero is unique among basal cell fate determinants in regulating neuroblast quiescence: loss of function mutations in numb or brat, or the Prospero target gene string, showed no effect on the timing of neuroblast quiescence (data not shown).

Figure 3.

Prospero is required for neuroblast quiescence.

(A–B) Wild type and prospero mutant (pros) stage 16 embryos. Parental neuroblasts (NB) stain for Deadpan and mγ-LacZ, whereas ectopic ‘de-differentiated’ neuroblasts stain for Deadpan but not mγ-LacZ. Proliferating neuroblasts are marked by EdU incorporation. (A) In wild type, most thoracic neuroblasts are in quiescence and do not incorporate EdU. (B) In prospero mutants, most parental neuroblasts (mγ-LacZ+) have not entered quiescence and still incorporate EdU. Ventral side is up and anterior is at left. Dashed line, boundary of neuroblast and neuron (N) layers. Scale bars: 20 μm. (C–E) NB3-3T remains proliferative in prospero mutant stage 16 embryos. (C) In wild type, NB3-3T is Mira+ mγ-LacZ+ EdU−; (D) in prospero mutants, NB3-3T is Mira+ mγ-LacZ+ EdU+. The NB3-3T lineage was identified by Eg-GFP (Eg-gal4 UAS-mCD8:GFP) and the neuroblast is indicated by arrowheads. (E) Quantification. n.s., not significant; ****p < 0.00001.

DOI: http://dx.doi.org/10.7554/eLife.03363.008

To test further whether Prospero is required for neuroblast quiescence, we returned to the NB3-3T model system. We used mγ-LacZ, Miranda, and EdU incorporation to assay the timing of NB3-3T quiescence in wild type and prospero mutant embryos. In both wild type and prospero haplo-insufficiency embryos, the mγ-LacZ+ NB3-3T was proliferative at stage 14 and quiescent by stage 16 (Figure 3C, quantified in Figure 3E). In prospero mutant embryos, the mγ-LacZ+ NB3-3T was proliferative at both stage 14 and 16 (Figure 3D, quantified in Figure 3E). The results suggest that the level of Prospero required for neuroblast quiescence is lower than that in prospero haplo-insufficiency and that a low level of Prospero is sufficient to promote neuroblast quiescence (see next section). We conclude that Prospero is required for neuroblast quiescence.

Prospero is sufficient to induce neuroblast quiescence

The central brain neuroblasts are continuously proliferating during the third larval instar (Ito and Hotta, 1992), and so we used this stage to determine whether transient Prospero expression could induce neuroblast quiescence. We used the TARGET method (McGuire et al., 2003) to transiently induce Prospero expression for 2 hr, and then assayed for neuroblast quiescence 12 hr later (Figure 4). Because high levels of Prospero can induce neuroblast differentiation (Figure 4—figure supplement 1) (Choksi et al., 2006; Cabernard and Doe, 2009; Bayraktar et al., 2010), we used Prospero levels low enough to minimize neuroblast differentiation, as shown by the persistence of most Deadpan+ neuroblasts after misexpression (Figure 4D). We identified quiescent neuroblasts by their failure to incorporate EdU, lack of progenitor marker Worniu, and lack of the differentiation marker Elav—together a robust signature for quiescent neuroblasts (see Figure 1). In wild-type larval brains, we found few or no quiescent neuroblasts, as expected (Figure 4E,F). In contrast, a pulse of Prospero resulted in the appearance of numerous quiescent neuroblasts (Figure 4E,F; 12 hr timepoint). To rule out the possibility that these ‘quiescent neuroblasts’ were early in the differentiation pathway, we reasoned that if they were quiescent they could reactivate proliferation in a nourishing environment (Chell and Brand, 2010; Sousa-Nunes et al., 2010) whereas if they were differentiating they would never re-enter the cell cycle. Thus, we exposed larval neuroblasts to the same 2 hr pulse of Prospero, but waited 24 hr to assay neuroblasts. We found that nearly all of the neuroblasts that were quiescent at 12 hr after induction had reactivated proliferation and re-expressed progenitor markers (Figure 4E,F; 24 hr timepoint). In addition, precocious low level nuclear Prospero could advance the timing of NB3-3T quiescence (Figure 4—figure supplement 2). We conclude that transient, low-level nuclear Prospero is sufficient to induce neuroblast quiescence. Interestingly, Prospero can also transiently arrest the cell cycle without inducing differentiation in embryonic longitudinal glial progenitors (Griffiths and Hidalgo, 2004), although the relative levels of Prospero have not been explored in this system.

Figure 4.

Prospero is sufficient to induce neuroblast quiescence.

(A–C) Transient overexpression of nuclear Prospero in neuroblasts (worniu-gal4 UAS-HA:prospero tub-gal80) results in neuroblast quiescence (A) or differentiation (B). Third-instar larvae were shifted from 22°C to 30°C to inactivate Gal80, which resulted in nuclear Prospero in neuroblasts. Low levels of nuclear Prospero result in neuroblast quiescence (Dpn+ Elav−) (A) whereas high levels of nuclear Prospero suppress Deadpan and activate Elav to induce differentiation (Dpn− Elav+) (B). Prospero levels were quantified in (C). Scale bars: 5 μm. (C) Quantification of Prospero levels in quiescent neuroblasts (Dpn+ Elav−) or differentiated neuroblasts (Dpn− Elav+). Differentiated neuroblasts were identified by their nuclear size larger than 7 μm in diameter. Prospero intensity was determined by totaling the gray value in nucleus of confocal stacks, followed by normalization to the total gray value of DNA marker 4ʹ,6-diamidino-2-phenylindole (DAPI). Number of neuroblasts quantified shown in bar. Error bars: standard deviation. (D–F) Larval neuroblast phenotype upon transient overexpression of Prospero. (D) Neuroblasts stay Deadpan+ (Dpn+) and thus do not differentiate. (E) There is an increase of quiescent neuroblasts 12 hr after Prospero induction, as determined by the elevated percentage of EdU− (D) and Wor− (E) neuroblasts. The quiescent neuroblasts re-enter the cell cycle and express the progenitor marker Wor 24 hr after induction. Error bars: standard deviation. n.s., not significant; *p < 0.05; **p < 0.01; ***p < 0.0001. (G) Model. Prospero levels distinguish three progenitor fates: absent for self-renewal, low for quiescence, and high for differentiation.

DOI: http://dx.doi.org/10.7554/eLife.03363.009

(A) Two levels of Prospero were generated by increasing time of misexpression (4 or 12 hr) with one copy of UAS-HA:prospero transgene. Whereas 2 hr can induce neuroblast quiescence (Figure 4), 4 hr and 12 hr can robustly induce neuroblast differentiation. (B) Two levels of Prospero were generated by misexpressing one vs two copies of the UAS-HA:prospero transgene for the same time (2 hr). A significant increase of the number of neuroblasts differentiated with two copies of UAS-HA:prospero transgene.

DOI: http://dx.doi.org/10.7554/eLife.03363.010

(A–G) Transient overexpression of nuclear Prospero in neuroblasts (worniu-gal4 UAS-HA:prospero tub-gal80) induces neuroblast cell cycle arrest. Embryos were shifted from 22°C to 30°C for 1 hr to inactivate Gal80, which resulted in transient nuclear Prospero in NB3-3T at stage 14 (A, wild type; B, transient nuclear Prospero). Scale bars: 2 μm. (C–G) Quantification of the Eg+ NB3-3T phenotype upon transient overexpression of Prospero. (C) Some neuroblasts have detectable nuclear Prospero. (D) Most neuroblasts fail to incorporate EdU. (E–G) There is a slight loss of Worniu+ (E), Asense+ (F) and CycE+ (G) neuroblasts.

DOI: http://dx.doi.org/10.7554/eLife.03363.011

Figure 4—figure supplement 1.

Low nuclear Prospero induces neuroblast quiescence, whereas high Prospero induces neuroblast differentiation.

(A) Two levels of Prospero were generated by increasing time of misexpression (4 or 12 hr) with one copy of UAS-HA:prospero transgene. Whereas 2 hr can induce neuroblast quiescence (Figure 4), 4 hr and 12 hr can robustly induce neuroblast differentiation. (B) Two levels of Prospero were generated by misexpressing one vs two copies of the UAS-HA:prospero transgene for the same time (2 hr). A significant increase of the number of neuroblasts differentiated with two copies of UAS-HA:prospero transgene.

DOI: http://dx.doi.org/10.7554/eLife.03363.010

Figure 4—figure supplement 2.

Prospero is sufficient to induce embryonic neuroblast quiescence.

(A–G) Transient overexpression of nuclear Prospero in neuroblasts (worniu-gal4 UAS-HA:prospero tub-gal80) induces neuroblast cell cycle arrest. Embryos were shifted from 22°C to 30°C for 1 hr to inactivate Gal80, which resulted in transient nuclear Prospero in NB3-3T at stage 14 (A, wild type; B, transient nuclear Prospero). Scale bars: 2 μm. (C–G) Quantification of the Eg+ NB3-3T phenotype upon transient overexpression of Prospero. (C) Some neuroblasts have detectable nuclear Prospero. (D) Most neuroblasts fail to incorporate EdU. (E–G) There is a slight loss of Worniu+ (E), Asense+ (F) and CycE+ (G) neuroblasts.

DOI: http://dx.doi.org/10.7554/eLife.03363.011

Our results confirm that high levels of Prospero can trigger differentiation and reveal that low levels of Prospero induce quiescence in the Drosophila neuroblasts. High levels of the vertebrate ortholog Prox1 can also trigger differentiation (Dyer, 2003; Takahashi et al., 2006; Foskolou et al., 2013); thus it is interesting that low levels of nuclear Prospero promote neuroblast quiescence without triggering differentiation (Figure 4—figure supplement 1). Prox1 activates some target genes at high levels but represses some target genes at low levels (Chen et al., 2008a, 2008b). We suggest a speculative model for Prospero function: low levels of Prospero are sufficient to repress most progenitor-specific genes in neuroblasts (worniu, asense, miranda, and cycE) but do not repress the expression of deadpan (perhaps due to lack of high affinity Prospero binding sites). In contrast, high levels of Prospero are sufficient to repress deadpan (perhaps via low affinity binding sites) and may even activate neuronal differentiation genes (Figure 4G).

In embryonic neuroblasts, the temporal transcription factors Pdm and Cas schedule the timing of neuroblast quiescence, but how they regulate Prospero nuclear import is unknown. In larval neuroblasts, Grh prevents accumulation of nuclear Prospero which would induce neuroblast cell cycle exit and differentiation (Maurange et al., 2008; Chai et al., 2013), again by an unknown mechanism. The protein Caliban is known to promote Prospero nuclear export in S2 cells (Bi et al., 2005), but it is unknown whether Caliban has a similar role in neuroblasts or if down-regulation of Caliban levels or activity leads to increased nuclear Prospero in neuroblasts entering quiescence.

Materials and methods

Fly genetics

The following flies were used in this study: (1) UAS-mCD8:Gfp;eagle-gal4; (2) cas/TM3 Sb Ubx-lacZ; (3) Df(2L)ED773/CyO wg-lacZ; (4) mγ-lacZ;UAS-mCD8:Gfp; eg-gal4 pros/TM3 Sb Ubx-lacZ; (5) UAS-HA:prospero (attP40); (5) UAS-HA:prospero (attP2); (6) wor-gal4;tubulin-gal80.

Induction of Prospero, EdU incorporation, antibody staining and imaging

Embryos with genotype UAS-HA:prospero/wor-gal4;tubulin-gal80/+ were collected for 4 hr at 22°C, and then cultured at 22°C for 5 days. The larvae were transferred to 30°C to inactivate Gal80, which resulted in the activation of Gal4 to express HA:Prospero. After heat shock, the larvae were immediately dissected or moved back to 22°C to incubate for 12 or 24 hr for later dissection, followed by 2-hr EdU incorporation, fixation, and antibody staining. EdU incorporation was performed in PBS at the concentration of 200 μg/ml. Antibody staining and EdU detection were previously described (Lai et al., 2012). Antibodies used in this study included: rabbit anti-Asense (1:1000; Cheng-Yu Lee, University of Michigan, Ann Arbor, MI), mouse anti-β-galactosidase (1:1000; Promega, Madison, WA), mouse anti-Cyclin E (1:50; Developmental Studies Hybridoma Bank (DHSB) developed under NICHD and maintained by University of Iowa), rabbit anti-Cyclin E (1:300; Santa Cruz Biotechnology, Dallas, TX); rat anti-Deadpan monoclonal (1:50; Doe lab), guinea pig anti-Deadpan (1:1000; Jim Skeath, Washington Univ., St. Louis, MO), mouse anti-Eagle (1:100; Doe lab); rabbit anti-Eagle (1:500) (Higashijima et al., 1996; Freeman and Doe, 2001); rat anti-Elav (1:50, DSHB); chicken anti-GFP (1:500; Aves Labs, Tigard, OR), mouse anti-HA (1:1000; Covance, Princeton, NJ), chicken anti-HA (1:1000, Bethyl Laboratories, Montgomery, TX); guinea pig anti-Miranda (1:2000; Doe lab), mouse anti-Prospero monoclonal purified IgG (1:1000; Doe lab), and rat anti-Worniu (1:2; Doe lab). Microscopy was done using a Zeiss LSM700 or LSM710; image processing and quantification were performed with the open source software FIJI.

eLife posts the editorial decision letter and author response on a selection of the published articles (subject to the approval of the authors). An edited version of the letter sent to the authors after peer review is shown, indicating the substantive concerns or comments; minor concerns are not usually shown. Reviewers have the opportunity to discuss the decision before the letter is sent (see review process). Similarly, the author response typically shows only responses to the major concerns raised by the reviewers.

Thank you for sending your work entitled “Transient nuclear Prospero induces neural progenitor quiescence” for consideration at eLife. Your article has been favorably evaluated by K (Vijay) VijayRaghavan (Senior editor), a Reviewing editor, and 3 reviewers.

The Reviewing editor and the other reviewers discussed their comments before we reached this decision, and the Reviewing editor has assembled the following comments to help you prepare a revised submission.

The current study addresses the role of the Prospero transcription factor in inducing neuroblast quiescence in the embryonic and larval Drosophila CNS. They find that the NB3-3T lineage displays a brief pulse of weak Pros nuclear expression just prior to entering quiescence. Mutant and mis-expression analysis lends support for the notion that this brief pulse in fact induces quiescence. They also address this notion in the postembryonic neuroblasts, and find support for a similar role for Pros in quiescence. Since we know very little about the fascinating, and even potentially medically important phenomena of stem cell quiescence, this study is of importance, and would likely be interesting for a broader audience. However, the manuscript is not without problems, and several issues should be addressed prior to publication.

Substantive concerns:

1) If a weak and brief pulse of Pros protein in NB nuclei is indeed important for NB quiescence, it is possible that pros may show haplo-insufficiency. To this end, they could analyze NB3-3T quiescence in pros heterozygotes. In addition, given the role of grh in quiescence, any heterozygotic effect of pros may be enhanced in transhet with grh i.e., pros+/-; grh+/-. Any such genetic dosage effect would greatly strengthen the study.

2) In the Introduction section and Figure 1–figure supplement 1: They state that “However, loss or misexpression of Worniu, Asense, or CycE has no effect on the timing of NB3-3T quiescence (Figure 1–figure supplement 1; data not shown)”. I cannot find any data regarding that misexpression of these factors cannot affect the timing of NB3-3T quiescence. This is an important point in the paper, and some data should be provided.

3) In pros mutants GMCs are de-differentiated into NBs, and hence express NB markers, e.g. Dpn. Hence, they use mgamma-LacZ to identify NBs in pros mutants. This is a reasonable approach, given that the asymmetric localization of Numb into GMCs (which is not affected in pros) results in expression of mgamma-lacZ only in the NBs. However, given that b-gal has a half-life in embryos of 8-12 hrs, it is not clear to me how the b-gal can label only NBs. Yes, the transgene is only transcriptionally active in NBs, but surely the b-gal protein, being quite stable and not subjected to asymmetric distribution, must label daughter cells as well. We need to be convinced, in this context, that all or in fact any of the ectopically EdU labelled cells in Figure 3B are actually NBs. Similarly, in Figure 3C-D, in pros mutants there is b-gal expression (as well as Mira expression) in several cells in the NB3-3T lineage; how can the authors be sure that the EdU positive cell is the NB?

4) Figure 4–figure supplement 1 and 2: They mis-express pros in NB3-3T, and find that it can induce NB quiescence. This is not surprising, given the role of pros in terminating proliferation.

5) Regarding postembryonic NBs, there are at least two previous publications that have addressed the function of pros in postembryonic NBs. In the paper from the Gould lab (Maraunge et al, 2008, Supplement Figure 3C) they show ectopic dividing cells in pros MARCM clones. However, since the clones are quite large, and they use few markers, ectopically dividing cells may just as well be GMCs and not NBs. Similarly, in the recent paper from the Doe lab itself (Bayraktar et al., Neural Dev, 2010,) they only study GOF effects of Pros, and find no pros LOF effects in Type I NBs. In the current study, they again show that GOF of pros can trigger NB quiescence in postembryonic NBs. This is not surprising given the two previous publications. However, they provide no new LOF data supporting the idea that Pros is necessary for NB quiescence.

6) The targets of Pros that induce/repress neuroblast quiescence are unknown. Pros has been shown to repress neuroblast cell-fate genes and cell-cycle genes in GMCs (Choksi SP, et al., Dev Cell 2006). The authors mentioned that loss or misexpression of Wor, Ase or CycE has no effect on the timing of NB3-3T quiescence. Have they tested these in 3rd instar central larval brain neuroblasts, in which the transient Pros expression was shown to induce neuroblast quiescence? Also, have they tested the role of String, an important cell cycle regulator that was repressed by Pros (Choksi SP, et al., Dev Cell 2006)?

7) The authors could also discuss the potential role of other cell fate determinants (Numb and Brat) in neuroblast quiescence. It will be valuable to know if the role of Pros represents a general function of cell fate determinants in regulating neuroblast quiescence or is Pros unique among the cell fate determinants in this process.

1) If a weak and brief pulse of Pros protein in NB nuclei is indeed important for NB quiescence, it is possible that pros may show haplo-insufficiency. To this end, they could analyze NB3-3T quiescence in pros heterozygotes. In addition, given the role of grh in quiescence, any heterozygotic effect of pros may be enhanced in transhet with grh i.e., pros+/-; grh+/-. Any such genetic dosage effect would greatly strengthen the study.

We appreciate reviewers raising the point and help us to further strengthen our conclusion: “low levels of Prospero promote neuroblast quiescence.” We first performed EdU incorporation in the pros heterozygous mutants and found that the neuroblast quiescence timing was not delayed. The results suggest that the levels of Prospero required for neuroblast quiescence is lower than that in pros haplo-insufficiency, which is consistent with our conclusions from the overexpression studies. We have added the result in Figure 3E and revised the Results section: “In both wild type and prospero haplo-insufficiency embryos, the mγ-LacZ+ NB3-3T was proliferative at stage 14 and quiescent by stage 16 (Figure 3C, quantified in 3E). In prospero mutant embryos, the mγ-LacZ+ NB3-3T was proliferative at both stage 14 and 16 (Figure 3D, quantified in 3E). The results suggest that the level of Prospero required for neuroblast quiescence is lower than that in prospero haplo-insufficiency and that a low level of Prospero is sufficient to promote neuroblast quiescence.”

Although grh plays an important role in regulating Pros activity in larval neuroblasts (Maurange et al., 2008; Chai et al., 2013), it is unknown whether grh plays a similar role in embryonic neuroblasts. If grh regulates the timing of neuroblast quiescence by preventing Pronuclear accumulation, trans-heterozygous pros+/- grh+/- will show no phenotype as Grh and Pros counteract each other in regulating neuroblast quiescence. As expected, we didn’t find any phenotype regarding the timing of neuroblast quiescence in the pros+/- grh+/-trans-heterozygous mutants (data not shown). Nevertheless, the concern led us to put back the discussion we removed in the previous submission due to the limitation of space: “In embryonic neuroblasts, the temporal transcription factors Pdm and Cas schedule the timing of neuroblast quiescence, but how they regulate Prospero nuclear import is unknown. In larval neuroblasts, Grh prevents accumulation of nuclear Prospero which would induce neuroblast cell cycle exit and differentiation (Chai et al., 2013; Maurange et al., 2008), again by an unknown mechanism.”

2) In the Introduction section and : They state that “However, loss or misexpression of Worniu, Asense, or CycE has no effect on the timing of NB3-3T quiescence (; data not shown)”. I cannot find any data regarding that misexpression of these factors cannot affect the timing of NB3-3T quiescence. This is an important point in the paper, and some data should be provided.

We agree, and we have added a new figure with the requested data (Figure 1–figure supplement 2). In summary, when we overexpressed Ase or CycE, we didn’t observe significant change in the number of proliferating neuroblasts.

3) In pros mutants GMCs are de-differentiated into NBs, and hence express NB markers, e.g. Dpn. Hence, they use mgamma-LacZ to identify NBs in pros mutants. This is a reasonable approach, given that the asymmetric localization of Numb into GMCs (which is not affected in pros) results in expression of mgamma-lacZ only in the NBs. However, given that b-gal has a half-life in embryos of 8-12 hrs, it is not clear to me how the b-gal can label only NBs. Yes, the transgene is only transcriptionally active in NBs, but surely the b-gal protein, being quite stable and not subjected to asymmetric distribution, must label daughter cells as well. We need to be convinced, in this context, that all or in fact any of the ectopically EdU labelled cells in are actually NBs. Similarly, in , in pros mutants there is b-gal expression (as well as Mira expression) in several cells in the NB3-3T lineage; how can the authors be sure that the EdU positive cell is the NB?

We thank the reviewers for highlighting this issue, which obviously needed a better explanation! The short answer is that pros mutant GMCs do not transcribe mγ-lacZ but continue to divide and thus dilute out mγ−lacZ, resulting in high level mγ−lacZ specifically in the parental neuroblasts. We have clarified this in the Results section: “mγ-LacZ is transcribed only in neuroblasts; in wild type the protein perdures into neuroblast progeny, whereas in prospero mutants the protein is restricted to neuroblasts because the progeny proliferate without expressing mγ-LacZ and thus dilute out the protein (Choksi et al.; Figure 3A,B).”

We thank the reviewers for helping us clarify this important point, which further supports our identification of NB3-3T in the prospero mutants, and our conclusion that Prospero is required for neuroblast quiescence.

4) : They mis-express pros in NB3-3T, and find that it can induce NB quiescence. This is not surprising, given the role of pros in terminating proliferation.

Yes, we agree that a major role of Prospero in neural progenitor cells is to terminate proliferation as previously reported (Li and Vaessin, 2000; Choksi et al., 2006). However, we want to emphasize that our new discovery is the dosage effect of Prospero in the progenitor cells: high levels of Pros will terminate proliferation and induce differentiation, while low levels of Pros transiently arrest cell cycle but are not sufficient to change the neuroblast identity, which leads to quiescence. We have added the following sentence in the Result section to further emphasize our main conclusion in the current study: “Our results confirm that high levels of Prospero can trigger differentiation, and reveal that low levels of Prospero induce quiescence in the Drosophila neuroblasts.”

5) Regarding postembryonic NBs, there are at least two previous publications that have addressed the function of pros in postembryonic NBs. In the paper from the Gould lab (Maraunge et al, 2008, Supplement ) they show ectopic dividing cells in pros MARCM clones. However, since the clones are quite large, and they use few markers, ectopically dividing cells may just as well be GMCs and not NBs. Similarly, in the recent paper from the Doe lab itself (Bayraktar et al., Neural Dev, 2010,) they only study GOF effects of Pros, and find no pros LOF effects in Type I NBs. In the current study, they again show that GOF of pros can trigger NB quiescence in postembryonic NBs. This is not surprising given the two previous publications. However, they provide no new LOF data supporting the idea that Pros is necessary for NB quiescence.

In our Results section, we have shown that loss of Prospero can delay the timing of neuroblast quiescence (Figure 3). We have also shown that transient overexpression of low levels of Prospero can induce neuroblast quiescence (Figure 4). To help readers easily understand our studies, we have revised the first sentence in the section of “Prospero is required for neuroblast quiescence” as follows: “Having shown a strong correlation between the timing of nuclear Prospero and neuroblast entry into quiescence, we next asked if Prospero is required for neuroblast entry into quiescence and loss of nuclear Prospero can delay the timing of neuroblast quiescence.”

6) The targets of Pros that induce/repress neuroblast quiescence are unknown. Pros has been shown to repress neuroblast cell-fate genes and cell-cycle genes in GMCs (Choksi SP, et al., Dev Cell 2006). The authors mentioned that loss or misexpression of Wor, Ase or CycE has no effect on the timing of NB3-3T quiescence. Have they tested these in 3rd instar central larval brain neuroblasts, in which the transient Pros expression was shown to induce neuroblast quiescence? Also, have they tested the role of String, an important cell cycle regulator that was repressed by Pros (Choksi SP, et al., Dev Cell 2006)?

In our earlier paper (Lai et al., 2012, Dev Cell), we showed that wor mutant larval neuroblasts prematurely differentiate or die, while overexpression of Wor results in the cell cycle arrest but not quiescence. We have also used cycE-RNAi (FBti0140127) and ase-RNAi (FBti0158728) to knock down CycE and Ase, respectively, but could not induce larval neuroblast quiescence (data not shown). We thus conclude that loss of CycE, Ase or Wor do not induce larval neuroblast quiescence.

As requested, we tested the role of String in neuroblast quiescence. We examined stg mutants for neuroblast EdU incorporation and Ase, Wor and Mira markers all are absent in quiescent neuroblasts. Interestingly, string mutant neuroblasts are EdU- as expected, but did not downregulate Ase, Wor and Mira, showing that they did not match the gene expression profile of quiescent neuroblasts. We conclude that although string is repressed by Pros, string downregulation is not sufficient to induce neuroblast quiescence. We add: “In addition, Prospero is unique among basal cell fate determinants in regulating neuroblast quiescence: loss of function mutations in numb or brat, or the Prospero target gene string, showed no effect on the timing of neuroblast quiescence (data not shown)”.

7) The authors could also discuss the potential role of other cell fate determinants (Numb and Brat) in neuroblast quiescence. It will be valuable to know if the role of Pros represents a general function of cell fate determinants in regulating neuroblast quiescence or is Pros unique among the cell fate determinants in this process.

We thank the reviewers for this comment, which allowed us to add new experiments that further highlight the unique role of Pros in promoting neuroblast quiescence. We found that numb or brat null mutant embryos did not exhibit a delay or advance in neuroblast quiescence (data not shown). We add: “In addition, Prospero is unique among basal cell fate determinants in regulating neuroblast quiescence: loss of function mutations in numb or brat, or the Prospero target gene string, showed no effect on the timing of neuroblast quiescence (data not shown)”.