FACS purification and transcriptome analysis of drosophila neural stem cells reveals a role for Klumpfuss in self-renewal

Abstract

Drosophila neuroblasts (NBs) have emerged as a model for stem cell biology that is ideal for genetic analysis but is limited by the lack of cell-type-specific gene expression data. Here, we describe a method for isolating large numbers of pure NBs and differentiating neurons that retain both cell-cycle and lineage characteristics. We determine transcriptional profiles by mRNA sequencing and identify 28 predicted NB-specific transcription factors that can be arranged in a network containing hubs for Notch signaling, growth control, and chromatin regulation. Overexpression and RNA interference for these factors identify Klumpfuss as a regulator of self-renewal. We show that loss of Klumpfuss function causes premature differentiation and that overexpression results in the formation of transplantable brain tumors. Our data represent a valuable resource for investigating Drosophila developmental neurobiology, and the described method can be applied to other invertebrate stem cell lineages as well.

Graphical abstract

Figure 1

Pure Populations of Larval NBs and Differentiated Cells Can Be Obtained by FACS (A) After dissection of larval CNS, tissue was disrupted and the cell suspension was subjected to quality control experiments (cell culture) or FACS sorting. Plotting GFP intensity (vertical axis) to cell size (horizontal axis) shows a small population of large cells with high GFP signal (NBs) and a population of smaller cells with a lower GFP signal (differentiated cells). FACS sorted cells were either subjected to quality control (cell culture or qRT-PCR analysis) or to paired-end Illumina Solexa mRNA sequencing. (B) Immunofluorescence staining for the NB marker aPKC, the differentiation marker ELAV and DAPI of an unsorted cell suspension (left column), FACS-sorted NBs (middle column), and neurons (right column). NBs are large aPKC-positive cells with a strong nuclear GFP signal (yellow arrowheads in left column), whereas their differentiated sibling cells are small, with a weaker GFP signal and stain positive for ELAV. The FACS-sorted NBs and neurons stain only for aPKC and ELAV, respectively (see single channels). Genotypes are ase-Gal4, UAS-stingerGFP. Scale bars: 20 μm.

Figure 2

FACS-Sorted NBs Express Correct Markers and Divide Asymmetrically (A) NB in a dissociated cell culture in telophase shows localization of aPKC to the apical and Miranda (Mira) to the basal cortex. (B) FACS-sorted NBs in metaphase (yellow arrowheads, left panel) and interphase (white arrowheads, left panel), or telophase (middle panel) asymmetrically localize aPKC (left panel) or Mira (middle panel) to the apical and basal cortex, respectively. The right panel shows FACS-sorted NBs arrested in mitosis with colchicine, which display correct localization of aPKC and Mira (n = 2). Metaphase DNA in (A) and (B) is marked with DAPI. (C) Stills from a movie (see also Movie S1) of an NB (yellow arrow) in an unsorted dissociated culture showing multiple rounds of asymmetric divisions. GMCs (yellow arrowhead) also divide to give rise to two neurons (white arrowhead). (D) Stills from a movie (see also Movie S2) of a FACS-sorted NB showing multiple rounds of asymmetric divisions. GMCs divide terminally to give rise to two neurons. (E) Quantification of cell cycle lengths from ten NBs show that the first and second divisions of sorted NBs, as well as the division of the first GMC, are only slightly affected, whereas later divisions are delayed compared with unsorted NBs. Three subsequent divisions, and the time point of the first GMC division, of ten NBs from three independent experiments each were measured; p = 0.39 (first NB), 0.02 (second NB), 0.0001 (third NB), and 0.92 (GMC), n.s. = not significant (Student’s t test). (F) FACS-sorted Dpn-positive NBs cultured for 5.5 hr show cortical localization of Pros (arrows) and multiple smaller Pros-positive, Dpn-negative GMCs. Genotypes are ase-Gal4, UAS-stingerGFP. Scale bars in (A-D): 12 μm; in (B) right panel and in (F): 20 μm.

Figure 3

Bioinformatic Analysis of Transcriptome Data (A) Expression levels of FACS-sorted NBs obtained by qRT-PCR and RNA-Seq data for known NB- and neuron-specific genes correlate (n denotes the number of experiments, error bars represent SD). (B) Numb transcript is alternatively spliced. The gene model from flybase (r5.44) is indicated, numb-RB is specifically expressed in neurons (upper track), numb-RA in NBs (lower track). (C) Hrb27c and brat are shown as examples of genes with 3′ UTR extensions. RNA-Seq tracks for neurons (upper track) and NBs (lower track) are shown. Note that for both genes the read coverage extends the 3′ UTR annotated in flybase (r5.44). (D) A network of genes involved in cell growth and NB self-renewal, resulting in loss or underproliferation of NBs knockdown phenotypes (Neumüller et al., 2011). Correlation with gene expression data shows that 71% of genes are significantly upregulated in NBs. “V” node shape denotes underproliferation. Small gray nodes are genes that are not expressed in NBs or neurons. Red nodes are genes expressed significantly higher in NBs (the strength of the color indicates fold change levels). White nodes are genes expressed at the same level in NBs and neurons. See also Figure S1 for an expanded network of asymmetric cell division.

Figure 4

Klumpfuss is an NB Identity Factor and Part of a Hypothetical Transcriptional Network for NB Self-Renewal (A) Hypothetical transcriptional network of 28 strong and differentially expressed TFs in NBs (FDR 0.01, log2FC > 3, FPKM value > 15) based on correlative Drosophila microarray expression data. Genes involved in growth control (mod, CG10565, bigmax, and TFAM), genes downstream of or regulated by Notch signaling (HLHmγ, dpn, wor, klu, and grh), and the chromatin remodeler Ssrp are marked in green, magenta, and yellow, respectively. Hubs are indicated with squares. (B) Larval brains overexpressing klu, dpn, HLHmγ, and control stained for Dpn, Pros, and Ase (Ase only shown for control and UAS-klu). Ectopic expression of these genes leads to overproliferation of Dpn-positive cells at the expense of differentiating cells (loss of Ase/Pros staining). White arrowheads indicate Ase-positive type I NBs in the klu overexpression brain. See also Figure S2 for Klu overexpression with the type II lineage-specific PntP1-Gal4, which also leads to tumor formation. Scale bars: 20 μm.

Figure 5

Immature INPs Reacquire NB Identity upon Ectopic Expression of Klumpfuss (A) In type II lineages (left column), Klu is expressed in primary NBs, reexpressed together with Dpn in Ase-positive mature INPs (see also single magenta/GFP channels, and schematic drawing) and turned off in GMCs/neurons. In type I NBs, Klu is expressed in primary NBs but, similarly to Dpn, not in GMCs or neurons (magenta/GFP in right column). (B) GFP-marked fragments from Klu overexpression brain tumors transplanted into the abdomens of host flies resulted in tumors in the adult hosts (filled green abdomen) 9 days after transplantation. Extracted tissue (right panels) stained with Mira and ELAV shows that the tumor tissue consists mostly of NBs. (C) As in wild-type type II lineages (left panels), mitotic (pH3 positive) NBs overexpressing Klu (right panels) show asymmetric localization of aPKC (apical marker, upper panels) or Mira (basal marker, lower panels). (D) Time-controlled induction of klu expression reveals blocked maturation of INPs and their dedifferentiation into Dpn-positive NB-like cells (white arrowheads, lower panels). In control type II lineages (upper panels), two to three Ase and Dpn negative immature INPs (white arrows) and three to four Ase-positive, Dpn-negative mature INPs (yellow arrowheads) are located next to the Dpn-positive, Ase-negative primary NB (asterisk). Lower panels show type II lineage after 19h of klu overexpression. Ase- and Dpn-negative immature INPs (white arrows) can be seen close to the primary NB (asterisk). Ectopic NB-like cells expressing Dpn, but not Ase, are found close to the primary NB (white arrowheads). See also schematic drawing for phenotype description. Smaller panels show blowups of the marked area in the left panels. Two subsequent focal planes are shown for control and klu overexpression (ten brains from two independent experiments were investigated). See also Figure S3 for Klu overexpression with the mature INP specific R9D11-Gal4, which does not lead to tumor formation. Scale bars: 20 μm.

Figure 6

Klumpfuss Is Required for NB Self-Renewal and Growth (A) Klumpfuss knockdown by RNAi causes loss of type II lineages. Larval brains are stained for Dpn and Pros, and type II lineages (left) and areas with loss of type II lineages (right panel) are marked. (B) Quantification of number of type II NBs upon klu knockdown by RNAi. On average, only one type II lineage can still be identified (n denotes the number of brains counted, error bars represents SD; p value 7.7 × 10⁻³¹, Student’s t test). (C) Quantification of the number of type I NBs upon klu knockdown by RNAi shows a reduction in total type I NBs per brain hemisphere (n denotes the number of brains counted, error bars represents SD; p value 2.7 × 10⁻⁵, Student’s t test). (D) Quantification of type I NBs cell size in the larval central brain shows that loss of klu function leads to a reduction in NB size (n denotes the number of NBs measured). (E) MARCM loss of function clones of kluR51 induced 30-48 hr AEL (right panel) show loss of primary NBs (right panel, type II lineage is shown) compared with wild-type MARCM clones (left panel). Larval brains were stained for Dpn and Pros, and the locations of NBs are indicated with asterisks. See also Figure S4 for the transheterozygous combination of mutant alleles klu^G410 and klu^R51, which recapitulates the klu clonal and shmiR loss of NB phenotypes. Scale bars: 15 μm.

Figure S1

Transcriptome Information Extending the Phenotype-Based Network of Asymmetric Cell Division, Cell Growth, and NB Self-Renewal, Related to Figure 3 The joined phenotypic network of asymmetric cell division, cell growth, and NB self-renewal as described by Neumüller et al. (2011) was expanded by adding strongly linked genes with available transcriptome data and five or more interaction partners in the phenotypic network. Interaction data were extracted from DroID (v5; http://www.droidb.org), STRING (v7.0 and v8.2), and BioGRID (v2.0.40), including experimental, genetic text-mining, and interolog interaction data. Genes are shown as nodes and the color of the nodes reflects the differential expression of genes, with red denoting higher expression in NBs, blue indicating higher expression in neurons, and white indicating no differential expression. For differentially expressed genes, the color intensity reflects the strength of the log-fold change. Approximately 12 genes from the phenotypic network were not expressed; they are shown in gray and small size. The shape of the nodes denotes phenotypes from the RNAi screen (Neumüller et al., 2011). The V shape is underproliferation, triangle is overproliferation, diamond is over- and underproliferation, hexagon is “other” (e.g. GFP aggregates), and ellipse is no phenotype. The white border and transparent colored nodes indicate genes from the original network, and black borders and solid node color denote genes with which the network was expanded. Thus, this network could form the basis for further studies and might help increase our understanding of neural stem cell biology.

Figure S2

Overexpression of klu with PntP1-Gal4 Results in Tumor Formation, Related to Figure 4 Expression of klu from PntP1-Gal4 resulted in tumors that showed an excess of ectopic NB-like cells at the expense of differentiating GMCs or neurons.

Figure S3

Overexpression of klu in Mature INPs Does Not Cause a Tumor, Related to Figure 5 Overexpression of klu using R9D11-Gal4, which starts expressing in mature Ase-positive, Dpn-negative INPs (Weng et al., 2010), does not result in an over-proliferation phenotype.

Figure S4

Transheterozygous Combination of Two Klu Mutant Alleles Causes Loss of Type I and Type II NBs, Related to Figure 6 The transheterozygous combination of the two mutant alleles klu^G410 and klu^R51 causes a loss of type II lineages (A) and a reduction of type I NBs by 15% (B). (n denotes number of brains counted, error bars represents SD, p = 0.01(Student’s t test)).