Opposing intrinsic temporal gradients guide neural stem cell production of varied neuronal fates

See allHide authors and affiliations

Science  16 Oct 2015:
Vol. 350, Issue 6258, pp. 317-320
DOI: 10.1126/science.aad1886

Changes over time build neuronal diversity

Although neural progenitors can keep generating new neurons, they can generate different neurons as the organism develops. Two different sections of the Drosophila brain, the mushroom bodies and the antennal lobes, show this characteristic, although the antennal lobes produce more different types of neurons over development than do the mushroom bodies. Liu et al. identified two RNA-binding proteins that manage this change over development in both settings.

Science, this issue p. 317


Neural stem cells show age-dependent developmental potentials, as evidenced by their production of distinct neuron types at different developmental times. Drosophila neuroblasts produce long, stereotyped lineages of neurons. We searched for factors that could regulate neural temporal fate by RNA-sequencing lineage-specific neuroblasts at various developmental times. We found that two RNA-binding proteins, IGF-II mRNA-binding protein (Imp) and Syncrip (Syp), display opposing high-to-low and low-to-high temporal gradients with lineage-specific temporal dynamics. Imp and Syp promote early and late fates, respectively, in both a slowly progressing and a rapidly changing lineage. Imp and Syp control neuronal fates in the mushroom body lineages by regulating the temporal transcription factor Chinmo translation. Together, the opposing Imp/Syp gradients encode stem cell age, specifying multiple cell fates within a lineage.

Diverse neural stem cells produce distinct sets of specialized neurons. The suite of daughter neurons generated by common neural stem cells can further change as development progresses, which suggests that the neural stem cells themselves change over time (14). In Drosophila, the neuroblast, a type of neural stem cell, can bud off about 100 ganglion mother cells that each divide once to produce two, often different, daughter neurons (5, 6). Mapping the serially derived neurons based on birth order/time has revealed that each individual neuroblast makes an invariant series of morphologically distinct neuronal types (79).

Drosophila central brain neuroblasts differ greatly in the number of neuronal types produced and the tempo at which changes occur. The four mushroom body neuroblasts divide continuously throughout larval and pupal development, but each produces only three classes of neurons (7). By contrast, the antennal lobe anterodorsal 1 (ALad1) neuroblast generates 22 neuronal types during larval development (9). Although sequential neuroblast expression of temporal transcription factors specifies neuronal cell fates in lineages of the Drosophila ventral nerve cord and optic lobes (10, 11), this mechanism is not easily applied to central brain. Here, we analyze fate determinants that direct neuronal diversification based on the age of the neuroblast.

Genes with age-dependent changes in expression levels throughout the life of a neuroblast could confer different temporal fates upon the neuronal daughter cells born at different times. Thus, we aimed to find genes dynamically expressed in mushroom body and antennal lobe neuroblasts by sequencing transcriptomes over the course of larval and pupal neurogenesis. We marked mushroom body or antennal lobe neuroblasts persistently and exclusively with green fluorescent protein (GFP) by genetic intersection and immortalization tactics (12, 13) (fig. S1, A to F). We isolated approximately 100 GFP+ neuroblasts for each RNA/cDNA preparation. Quantitative polymerase chain reaction (qPCR) showed that known neuroblast genes, including deadpan (dpn) and asense, are enriched in mushroom body and antennal lobe neuroblasts compared with total larval brains (fig. S1, G and H). Samples passing this qPCR quality check were sequenced. We obtained the transcriptomes of mushroom body neuroblasts at 24, 50, and 84 hours after larval hatching (ALH), as well as 36 hours after puparium formation (APF), and the transcriptomes of antennal lobe neuroblasts at 24, 36, 50, and 84 hours ALH (table S1). We did not sequence the antennal lobe neuroblasts at 36 hours APF because they stop producing neurons around puparium formation. Also, we did not analyze mushroom body neuroblasts at 36 hours ALH because the mushroom body lineages do not undergo detectable fate or molecular changes between 24 and 50 hours ALH.

We defined strongly dynamic genes as those with a greater than fivefold change in expression level across different time points and a maximum average abundance higher than 50 transcripts per million (see the supplementary materials). We found 83 strongly dynamic genes in the mushroom body and 63 in the antennal lobe (fig. S2). The two sets shared 16 genes in common (Fig. 1A). Among these 16 common genes, pncr002:3R, a putative noncoding RNA, ranks highest in absolute abundance, the importance of which remains unclear.

Fig. 1 RNA sequencing identified temporally dynamic genes in neuroblasts.

(A) Sixteen common dynamic genes in mushroom body (MB) or antennal lobe (AL) neuroblasts were shown. The entire lists of 83 dynamic genes in mushroom body and 63 in antennal lobe neuroblasts are in fig. S2. On the left, normalized expression levels (maximum set to 1) are shown in blue (low) to red (high). On the right, logarithmic scales of the maximum expression level [transcripts per million (TPM)] are shown in white (low) to orange (high). Imp and Syp are indicated with arrowheads. (B) Dynamic changes of Imp and Syp expression (mean ± SD, n = 3 replicates. Each replicate contains 100 GFP+ neuroblasts.) are shown. (C) Immunostaining shows that Imp protein decreases and Syp protein increases over time in mushroom body neuroblasts and newly born neurons (circled regions). Reciprocal protein levels of Imp and Syp are also present in older neurons (arrows and arrow heads). Representative images of n = 15 brain tissue samples. Scale bars, 20 μm.

IGF-II mRNA-binding protein (Imp) and Syncrip (Syp), which code for two evolutionally conserved RNA-binding proteins (14, 15), rank second and third in absolute abundance (Fig. 1A). Imp is expressed abundantly at 24 hours ALH and declines to a minimum at 84 hours ALH in antennal lobe and 36 hours APF in mushroom body neuroblasts, whereas Syp increases from minimal expression at 24 hours ALH to become one of the most abundant genes at late larval stages (Fig. 1B). Imp/Syp gradients with larger amplitudes and steeper slopes characterize antennal lobe neuroblasts (Fig. 1B), which yield more diverse neuron types at faster tempos than the mushroom body neuroblasts. Antibodies to Imp and Syp (14, 16) showed similar patterns in shifts of protein abundance in both mushroom body neuroblasts and neuronal daughter cells (Fig. 1C). We therefore investigated the roles of Imp and Syp in neuronal temporal fate specification in both mushroom body and antennal lobe lineages.

The post-embryonic mushroom body neuroblasts sequentially produce γ, α′/β′, and α/β neurons, which can be distinguished by a variety of markers (7, 17, 18). Fasciclin II (FasII) is expressed in the perimeter of the α/β lobes, is weakly expressed in the γ lobe, and is not expressed in the α′/β′ lobes. Trio, a Dbl family protein, is expressed in the γ and α′/β′ lobes. The γ neurons can also be identified in wandering larvae by expression of ecdysone receptor B1 isoform (EcR-B1) (19). Moreover, one can predict the fate of newborn mushroom body neurons based on the protein levels of Chinmo, a known temporal transcription factor, as abundant Chinmo specifies the γ fate, weak Chinmo expression confers the α′/β′ fate, and absence of Chinmo permits the α/β fates (20).

RNA interference (RNAi) aimed to reduce Imp expression resulted in up-regulated Syp, whereas knocking down Syp expression caused increased Imp expression (Fig. 2, A to D). This reciprocal derepression was evident in protein and transcript content (Fig. 2, A to E) as well as phenotype. Silencing Imp triggered precocious production of α/β neurons throughout larval development; these neurons lacked EcR-B1 at the wandering larval stage (Fig. 2, F and G) and showed no Chinmo at 24 hours ALH (Fig. 2, H and I). Imp-depleted neuroblasts ended neurogenesis prematurely: By 28 hours APF, no neuroblasts remained in the mushroom body (n = 10 brain tissue samples). This resulted in small adult mushroom bodies with only α/β lobes (Fig. 3, A and B, and fig. S4, A and B) (21). By contrast, silencing Syp extended the production of Chinmo-positive γ neurons through pupal development (Fig. 2, J and K, and fig. S3). The Syp-depleted adult mushroom bodies consisted of a single prominent γ lobe (Fig. 3C and fig. S4C). The reciprocal temporal fate transformations were also seen in the mushroom body neuroblast clones homozygous for various Imp or Syp loss-of-function mutations (fig. S4, D to F). Thus, Imp specifies early γ neurons and Syp specifies late α/β neurons.

Fig. 2 Knocking down Imp or Syp derepressed the other and caused opposite phenotypes.

(A to E) Depleting Imp or Syp by targeted RNAi (driven by GAL4-OK107) derepressed the other in MB neuroblasts and newly born neurons (circled regions), as evidenced by immunostaining [(A) to (D)] and qPCR (mean ± SD, n = 3 replicates. Each replicate contains 100 GFP+ neuroblasts.) (E). Student’s t test was performed. *P < 0.05, **P < 0.01, ****P < 0.0001. Gray, control (CTRL) group at 24 hours ALH; black, Imp-RNAi at 24 hours ALH; red, CTRL at 36 hours APF; blue, Syp-RNAi at 36 hours APF. (F and G) EcR-B1 was absent in Imp-depleted wandering-larval MB neurons (circled regions). (H to K) Chinmo protein precociously became undetectable in Imp-depleted MB early larval-born neurons [(H) and (I)] but ectopically existed in Syp-depleted MB pupal-born neurons [(J) and (K)]. Circled regions contain one MB neuroblast (Dpn+) and its newborn daughter cells. All images are representative of n = 15 brain tissue samples. Scale bars, 20 μm.

Fig. 3 Imp and Syp govern MB neuronal temporal fates via Chinmo regulation.

(A to G) Representative confocal projections of adult MBs immunostained by GFP and FasII antibodies. Cartoons on the right illustrate the lobe phenotypes and were constructed by examining individual Z planes. Consistent phenotypes were seen in all examined samples (n = 20 brain tissue samples per genotype). Notably, transgenic Chinmo partially rescued early γ neurons in the Imp-depleted MB [compare (B) with (F)], and depleting Chinmo converted the supernumerary early γ neurons present in the Syp-depleted MB into late-born α/β neurons [compare (C) and (G)]. Scale bars, 20 μm.

We next asked whether prolonged coexpression of Imp and Syp can increase the number of α′/β′ neurons, which are typically born at a time when Imp and Syp expression levels are similar. Ectopic induction of Imp or Syp transgenes enhanced Imp or Syp protein levels in mushroom body newborn neuronal daughter cells but not in the neuroblasts (fig. S5). In the case of Syp overexpression, the early larval neuroblasts still contained abundant Imp, as did their newborn neurons expressing ectopic Syp (fig. S5C). Analogously, overexpressing Imp rendered the pupal-born neurons strongly positive for Syp as well as ectopic Imp (fig. S5E). Therefore, after Imp or Syp overexpression, many more newborn neurons simultaneously expressed Imp and Syp, and the normally modest α′/β′ neuronal lobes were enlarged in adult brains (Fig. 3, D and E, and fig. S4, G and H). Cell death was unlikely to distort the developmental outcomes, as rare sporadic cell death was only detected in mushroom body neurons expressing ectopic Syp at 50 hours ALH (fig. S6). Taken together, our data demonstrate that relative levels of Imp and Syp dictate mushroom body neuronal temporal fates.

The altered Chinmo protein levels upon Imp or Syp depletion (Fig. 2, H to K, and fig. S3) prompted us to ask whether Imp and Syp regulate chinmo expression. The abundance of chinmo transcripts normally decreases as neuroblasts age (fig. S7, A and B). This pattern was unperturbed in Imp or Syp-depleted neuroblasts (fig. S7, C and D). Together, these data suggest that Imp and Syp regulate chinmo expression at a posttranscriptional level.

We further explored by epistasis whether Imp and Syp act to regulate mushroom body neuronal temporal fates through Chinmo. Overexpressing a chinmo transgene partially restored the production of γ neurons by the short-lived, Imp-depleted neuroblasts (Fig. 3F). Moreover, silencing chinmo transformed the supernumerary γ neurons made by the Syp-depleted neuroblasts into α/β neurons (Fig. 3G). Together, these observations place Chinmo downstream of Imp and/or Syp in the temporal fate specification of mushroom body neurons.

To ascertain whether Imp/Syp gradients serve as a general temporal-fating mechanism, we examined the roles of Imp and Syp in the rapidly changing antennal lobe anterodorsal 1 (ALad1) lineage that yields ~60 larval-born neurons of 22 types. Although all 22 types express acj6-GAL4, only the first 12 types generated express GAL4-GH146 (9). Imp depletion reduced the ALad1 daughter cell number (acj6+) from 64.0 ± 2.0 to 50.7 ± 4.9 (P < 0.05) but increased the ratio of late-type to early-type neurons (early-type GH146+ neurons decreased from 40.3 ± 2.1 to 13.3 ± 2.6, P < 0.0001) (Fig. 4 and fig. S8). By contrast, Syp depletion increased total ALad1 daughter cells from 64.0 ± 2.0 to 100.0 ± 4.1 (P < 0.0001) and increased the percentage of the early-type (GH146+) neurons from 63.0 ± 3.2% to 93.5 ± 5.1% (P < 0.001) (Fig. 4 and fig. S8). Despite precocious production of late-type neurons or prolonged generation of early-type neurons, 21 of the 22 neuron types were preserved (fig. S8). In summary, the opposing Imp/Syp gradients govern temporal fates in at least two different neuroblast lineages that produce mushroom body and antennal lobe neurons functioning in memory and olfaction, respectively.

Fig. 4 Depleting Imp or Syp elicited temporal fate transformation of antennal lobe neurons.

Composite confocal images of adult fly brain antennal lobe regions counterstained with nc82 monoclonal antibody (blue). ALad1 neuroblast clones of various genotypes were labeled with GFP (magenta) driven by GAL4-GH146 (A to C) or acj6-GAL4 (D to F), respectively; cell counts (mean ± SD) are indicated at the bottom left of each panel. Compared to control (CTRL) (D), total cell (acj6+) numbers of the ALad1 lineage were significantly decreased (P < 0.05) in Imp-RNAi (E) but increased (P < 0.0001) in Syp-RNAi clones (F). All 22 neuronal types were maintained, with the exception of VA3 (arrows) in Imp-RNAi clones. However, the cell number ratio between early- and late-born neuronal types were significantly decreased (P < 0.001) in Imp-RNAi but increased (P < 0.001) in Syp-RNAi clones. Consistent phenotypes were seen in all examined neuroblast clones (n = 6 brain tissue samples per genotype). Scale bars, 20 μm. Detailed analysis is presented in fig. S8.

The opposite temporal gradients of Imp and Syp in neuroblasts confer the serially derived daughter cells with graded levels of Imp/Syp. The acquisition of distinct daughter cell fates based on the Imp/Syp morphogens is reminiscent of early embryonic patterning by the opposite spatial gradients of maternally inherited Bicoid and Nanos. Different levels of Bicoid and Nanos are incorporated into cells along the anterior-posterior axis after cellularization of the blastoderm. Bicoid and Nanos function as RNA-binding proteins to initiate anterior-posterior spatial patterning via translational control of maternal transcripts that encode transcription factors (22). Analogously, temporal fate patterning of newborn neurons is orchestrated by post-transcriptional control of chinmo and potentially other genes by Imp and Syp. Because Imp and Syp may share common targets (14, 15) but show affinity for different RNA motifs (15, 23), it is possible that Imp and Syp can both bind chinmo transcripts, but they may differentially target chinmo transcripts for translation versus sequestration. Descending Imp temporal gradients governs aging of Drosophila testis stem cell niche (24). Imp-1, the mammalian ortholog of Imp, is also needed to maintain mouse neural stem cells (25). We propose that graded Imp/Syp expression constitutes an evolutionally conserved mechanism for governing time-dependent stem cell fates, including temporal fate progression in neural stem cells and their derived neuronal lineages.

Supplementary Materials

Materials and Methods

Figs. S1 to S8

Table S1

References (2629)

References and Notes

  1. Acknowledgments: We thank L. Jones, T. S. Hays, P. Macdonald, and I. Davis for sharing reagents. We thank the Transgenic RNAi Project at Harvard Medical School (NIH/NIGMS R01-GM084947) and the Vienna Drosophila Resource Center for providing transgenic RNAi fly stocks used in this study. We thank R. Miyares, G. Rubin, J. Truman, C. Wu, and I. Davis and his group for comments. This work was supported by Howard Hughes Medical Institute. RNA-sequencing data are available in the National Center for Biotechnology Information Gene Expression Omnibus, accession no. GSE71103. The supplementary materials contain additional data.
View Abstract

Stay Connected to Science

Navigate This Article