Nutrition-responsive glia control exit of neural stem cells from quiescence

Abstract

The systemic regulation of stem cells ensures that they meet the needs of the organism during growth and in response to injury. A key point of regulation is the decision between quiescence and proliferation. During development, Drosophila neural stem cells (neuroblasts) transit through a period of quiescence separating distinct embryonic and postembryonic phases of proliferation. It is known that neuroblasts exit quiescence via a hitherto unknown pathway in response to a nutrition-dependent signal from the fat body. We have identified a population of glial cells that produce insulin/IGF-like peptides in response to nutrition, and we show that the insulin/IGF receptor pathway is necessary for neuroblasts to exit quiescence. The forced expression of insulin/IGF-like peptides in glia, or activation of PI3K/Akt signaling in neuroblasts, can drive neuroblast growth and proliferation in the absence of dietary protein and thus uncouple neuroblasts from systemic control.

Graphical abstract

Figure 1

Nutritional Dependence of Neuroblast Reactivation (A–F) grh-GAL4 drives strong expression of UAS-mCD8-GFP in one-third of neuroblasts in the thoracic VNC (tVNC) (∼16/47 per thoracic segment; ∼48/141 total). Yellow arrows highlight examples of grh-GAL4-expressing neuroblasts. White arrows highlight examples of neuroblasts that do not express grh-GAL4. (A and A′) In just-hatched larvae (0–1 hours posthatching [hph]), the cell body diameter (ø) of a neuroblast is ∼3–4 μm. (B and B′) By 24 hph, most neuroblasts have increased in diameter but maintain their primary process (white arrowheads) prior to division. The dashed box in (B) shows a snapshot from a 3D reconstruction of a neuroblast (ventral, V; dorsal, D). (C and C′) By 48 hph, neuroblasts have fully enlarged and undergone several divisions. Note the small GFP-marked, Dpn-negative progeny (e.g., yellow arrowhead). (A′), (B′), and (C′) are snapshots from 3D reconstructions of the VNCs shown in (A), (B), and (C), respectively. (D–F) In larvae deprived of amino acids (sucrose-only diet), neuroblast growth and cell-cycle re-entry never occur (Britton and Edgar, 1998). Neuroblasts maintain their quiescent size and primary process. Compare (D), (E), and (F) with (A′), (B′), and (C′), respectively. Z projections of tVNCs at indicated time points. GFP, green; Deadpan (Dpn; neuroblast nuclei, red); Discs Large (Dlg; cell cortices, blue). Scale bars, 20 μm.

Figure 2

Glia Express dILP6 and dILP2 during Reactivation (A–B′) dilp6-GAL4 marks a subset of the outermost, perineurial (Stork et al., 2008) glia during first- and second-larval instars. dilp6-GAL4-driving UAS-mCD8-GFP, red; glial nuclei, blue (anti-Repo). Scale bars, 15 μm. (C) Anti-dILP2 (green) in the tVNC at 24 hph shows a punctate perinuclear enrichment in surface glial cells (see pink arrows in C′), consistent with secretory vesicle processing. Z projection of ventral surface glial layer. (D and E) dILP6-positive glia (dilp6-GAL4 > UAS-mCD8-GFP [green]) lie just above neuroblasts (Dpn, red) and below the basement membrane (dPerlecan, blue). Sequential sections from ventral surface of VNC (D and D″) and in cross-section (E). Scale bars: D and D″, 10 μm; E, 1μm. See also Figure S1.

Figure 3

Neuroblast Reactivation Requires Cell-Intrinsic dInR/PI3K Signaling grh-GAL4 driving: mCD8-GFP (A), mCD8-GFP + dominant-negative PI3K (Δp60) (B), mCD8-GFP + dPTEN (C), and mCD8-GFP + dominant-negative insulin receptor (dInR^K1409A) (D). (A) By 24 hph, all neuroblasts in the tVNC have begun to enlarge, and average cell body diameter has increased from ∼4 μm to ∼7 μm. (B–D) Expression of Δp60, dPTEN, or dInR^K1409A retards growth and cell-cycle re-entry (white arrowheads). Neuroblasts that do not express grh-GAL4 show normal cell growth (compare yellow arrows with white arrowheads). (A′–D′) are projections of VNCs shown in (A–D), respectively. White arrowheads in (A′–D′) point to the same neuroblasts as in (A–D), respectively. (E) A quantification (box and whisker plot) of the experiments represented in (A–D). GFP only (control), n = 52 (6 VNCs), mean = 7.45 μm, SD = 1.24. +PI3K (Δp60), n = 62 (5 VNCs), mean = 4.21 μm, SD = 0.87. +dPTEN, n = 114 (12 VNCs), mean = 4.22 μm, SD = 0.76. +dInR^K1409A, n = 109 (12 VNCs), mean = 4.54 μm, SD = 0.94. (n equals number of neuroblasts assayed). p values were generated using Student's t test. GFP, green; Dpn, red; Dlg, blue. Scale bars, 20μm. See also Figure S2 and Figure S3.

Figure 4

PI3K and Akt Signaling Are Sufficient for Neuroblast Reactivation (A–E) grh-GAL4 driving UAS-mCD8::GFP (green) and UAS-dp110^CAAX (a constitutively active form of the PI3K catalytic subunit) in larvae fed a sucrose-only (amino acid-free) diet. (A′) and (B′) are projections of the VNCs in (A) and (B), respectively. (A and A′) Neuroblasts in which PI3K signaling is activated by dp110^CAAX are quiescent at 0–1 hph. Scale bars, 20 μm. (B and B′) Neuroblasts can fully reactivate during the first-larval instar despite the absence of a nutritional stimulus. Arrows in (B) and (B′) point to an enlarged, reactivated neuroblast. The arrowhead in (B) points to one of the progeny of a reactivated neuroblast. Dpn, red; GFP, green; Dlg, in blue. Scale bars, 20 μm. (C and D′) The adaptor protein Miranda (red) is asymmetrically localized and partitioned to daughter cells of dp110^CAAX-reactivated neuroblasts (yellow arrowheads). Scale bar, 5 μm. (E) The cell-fate determinant Prospero (red) is also partitioned to dp110^CAAX-reactivated neuroblast progeny (see white arrowheads). Scale bar, 5 μm. (F–F″) Neuroblasts (Dpn, blue) in which PI3K signaling is upregulated by expression of dp110^CAAX show significantly increased levels of phosphorylated (active) Akt (pAkt, red) (blue arrowhead). Example control neuroblasts indicated by pink arrowheads. grh-GAL4 driving UAS-mCD8::GFP (green) and UAS-dp110^CAAX in third-instar larvae fed a normal diet (fresh yeast). Scale bars, 10 μm. (G–I′) grh-GAL4 driving UAS-mCD8::GFP (green) and UAS-myr-Akt (a constitutively active form of Akt) in larvae fed a sucrose-only (amino acid-free) diet. (G′–I′) are projections of VNCs in (G–I), respectively. Dpn, red; Dlg, blue; pH3-labeled mitotic cells, white. Scale bars, 20 μm. (G and G′) Neuroblasts in which Akt signaling is activated by myr-Akt are quiescent at 0–1 hph. (H and H′) These neuroblasts can fully reactivate during the first larval instar despite the absence of a nutritional stimulus. The yellow arrow points to a neuroblast not expressing grh-GAL4 that has failed to reactivate in the absence of the nutritional stimulus. (I and I′) Neuroblasts and their progeny are seen dividing at 48 hph (pH3, white). See also Figure S4.

Figure 5

dILPs Are Required for Neuroblast Reactivation, and Their Glial Expression is Nutrition Dependent (A–C) dilp 2,3,5,6 mutants display impaired neuroblast reactivation (compare B and C with heterozygous control A). Dpn, red; Dlg, blue. Scale bars, 20 μm. (D and E) VNCs from Oregon R larvae at 24 hph. (D) dILP2 protein expression in the surface glia of larvae fed a normal diet. (E) In larvae reared on a sucrose-only diet, dILP2 expression is greatly reduced (DILP2, green; repo, red). VNCs were dissected, stained, and imaged together. Identical reagents and microscope settings were employed. Scale bars, 25 μm. (F) Q-PCR analysis of dilp6 in the VNC. dILP6 transcript levels at 12 hr and 24 hr posthatching in VNCs of larvae fed normal or sucrose-only diets, compared to dilp6 transcript levels at 0 hr (just hatched). dilp6 levels normally increase 8-fold during the first instar (0–24 h) but are abolished when larvae are reared on a sucrose-only diet. ^∗∗∗p < 0.02; Student's t test. Error bars represent standard deviations. Larvae fed a normal diet showed a mean fold change in dilp6 mRNA level of 1.7 and 7.9 at 12 and 24 hr, respectively, with SD of 0.01 and 1.55, respectively. Larvae fed a sucrose-only diet showed a mean fold change in dilp6 mRNA level of 1.1 and 1.2 at 12 and 24 hr, respectively, with SD of 0.11 and 0.15, respectively.

Figure 6

Glial dILP Expression Is Sufficient for Neuroblast Reactivation (A–D) Repo-GAL4 driving UAS-dilp6 and UAS-Histone H2B-mRFP (H-RFP, white) in larvae reared on a sucrose-only (amino acid-free) diet. Dpn, red; Dlg, blue; pH3, green. Scale bars, 20 μm. (A) At 0–1 hph, neuroblasts are quiescent, showing no sign of growth or division. (B) Forced expression of DILP6 in glia drives the reactivation of neuroblasts in the absence of the nutritional stimulus at 27 hr. Yellow arrowheads indicate mitotic neuroblasts. (C and D) Neuroblasts continue to divide at 48 and 72 hph, respectively. Yellow arrowheads indicate mitotic neuroblasts. (E and F) Control VNCs from larvae with UAS-dilp6 and UAS-H-RFP, but no GAL4 driver, reared on a sucrose-only (amino acid-free) diet. Neuroblasts never enlarge or divide. White arrowheads indicate neuroblasts. Scale bars, 20 μm. See also Figure S5.

Figure 7

Glia Are a Key Relay between Nutrition and Neuroblast Reactivation (A and B) Repo-GAL4 driving UAS-shi^ts and UAS-Histone H2B-mRFP (H-RFP) and control (no GAL4), reared at 33°C after larval hatching. Dpn, green; Dlg, red; pH3, blue; H-RFP, white. Scale bars, 20 μm. (A) At 72 hr (midthird instar), neuroblasts in the control are fully enlarged and proliferating. White arrowheads indicate mitotic neuroblasts. (B) At 72 hr, neuroblasts from animals in which glial dynamin function has been blocked remain quiescent. Yellow arrowheads indicate neuroblasts. (C and D) Quantification of neuroblast enlargement and proliferation, respectively. ^∗∗∗p < 0.005; Student's t test. The higher variation seen at 72 hr posthatching is due to a subset of larvae eventually showing neuroblast reactivation after a prolonged delay (40%; n = 10). (C) Box and whisker plot showing neuroblast growth is blocked by glial expression of shi^ts. At 29, 48, and 72 hr, control neuroblasts have mean diameters of 8.13, 10.27, and 11.46 μm, respectively, with SD of 1.49, 1.71, and 2.06, respectively. At 29, 48, and 72 hr, in larvae in which dynamin function has been blocked in glia, neuroblasts have mean diameters of 4.17, 4.87, and 5.58 μm, respectively, with SD of 0.47, 0.79, and 1.75, respectively. (D) Bar chart showing neuroblast proliferation is also suppressed by blocking dynamin function in glia. M phase neuroblasts were identified by the presence of pH3. Error bars represent standard deviations. At 29, 48, and 72 hr, control tVNCs have a mean number of M phase neuroblasts of 14.5, 14.6, and 13.14, respectively, with SD of 1.91, 0.71, and 3.02, respectively. At 29, 48, adn 72 hr, in larvae in which dynamin function has been blocked in glia, tVNCs have a mean number of M phase neuroblasts of 0, 0.5, and 4.1, respectively, with SD of 0, 0.58, and 5.55, respectively. (E) A model for the nutritional control of neuroblast reactivation. Previous work (Britton and Edgar, 1998) suggested that dietary amino acids are sensed by the fat body, triggering FBDM secretion into the hemolymph. The FBDM might then stimulate surface glia, which we show express and secrete dILPs in response to amino acids. These dILPs act on neuroblasts in a paracrine manner to activate the dInR/PI3K/Akt pathway, leading to cell growth and cell-cycle re-entry. dILPs, purple; active PI3K/Akt, green; asymmetrically localized cell fate determinants, red.

Figure S1

dILP6 and dILP2 Are Expressed in the Same Cells, Related to Figure 2 24h VNC in which dilp6-GAL4 is driving the nuclear marker histone-mRFP (in red) (under UAS control). Glial cells expressing dILP2 protein (in green) are the same cells that express dilp6 GAL4 (red nuclei). Compare yellow arrows between the separate channels and overlay.

Figure S2

PI3K Is Active during, and Required for, Neuroblast Reactivation, Related to Figure 3 (A–B′) PH-GFP distribution during reactivation. During neuroblast reactivation there is a strong accumulation of PH-GFP at the cell membrane (compare arrowheads in A, A’ (quiescent neuroblasts) to those in B,B’ (reactivating neuroblasts)), indicating increased PIP3 levels and therefore increased PI3K activity. (GFP in green, Deadpan in red, scale bars represent 20μm). (C and E) The CNS is significantly reduced in size in PI3K (dp110) loss of function mutants. CNS from wandering third instar larvae stained with Discs large and false colored (CNS in red, and eye discs in green). Thoracic VNC (tVNC) highlighted with dashed white box. CB marks the central brain. (Scale bars represent 150μm). (D and F) Neuroblasts remain quiescent in PI3K (dp110) mutants. z-projections of the tVNC. In PI3K mutants, neuroblast cell growth is significantly retarded, with many neuroblasts remaining quiescent. Compare yellow arrows in (F), with white arrowheads in (D). (Dpn in red, scale bars represent 20μm).

Figure S3

S6K Is Enriched in Reactivating Neuroblasts, Related to Figure 3 Late first instar larval VNCs (genotype exhibits slightly delayed development) that are homozygous for the S6K-GFP protein trap (in which GFP has been inserted into the second intron of endogenous S6K) (Buszczak et al., 2007; Kelso et al., 2004). (A–A″) During reactivation S6K is enriched in a population of CNS cells which are negative for the neuron-specific transcription factor ElaV (in red) (see white arrows). (B–B″) The ElaV-negative cells with high S6K GFP levels are the neuroblasts, as evidenced by their Deadpan-positive nuclei (in red) (see white arrows). (Scale bars in all panels represent 20μm).

Figure S4

Akt Signaling Is Sufficient for Neuroblast Growth and Cell-Cycle Re-Entry, Related to Figure 4 24h VNC from a larva reared on a sucrose-only diet in which grh-GAL4 is driving UAS myr-Akt and UAS mCD8-GFP expression. myr-Akt expression is sufficient for neuroblast reactivation in the absence of the amino-acid stimulus. Confocal channels have been split to allow better visualization of representative mitotic figures. All grh-Gal4-positive neuroblasts have enlarged, and many have begun to divide as evidenced by smaller, GFP-retaining, daughter cells adjacent to the neuroblasts, and pH3 staining. Mitotic pH3 nuclei can be seen in both neuroblasts (white arrows), and neuroblast progeny (yellow arrow). (Scale bars represent 20μm).

Figure S5

dILP2 Can Also Reactivate Neuroblasts, Related to Figure 6 27h VNCs from larvae reared on a sucrose-only diet in which repo-GAL4 is driving the expression of either dILP6, dILP2, or dILP2 and dILP6, in the presence of tubGAL80^ts. Larvae were shifted from 18°C to 29°C at larval hatching to block the repressor activity of GAL80, and allow dilp expression. Both dILP6 and dILP2 were sufficient for neuroblast reactivation under these conditions (white arrowheads point to enlarging neuroblasts). Co-expression of dILP2 and dILP6 had no additive or synergistic effect on neuroblast reactivation under these conditions. (Scale bars represent 20μm).