Local insulin-like growth factor I expression is essential for Purkinje neuron survival at birth

Abstract

IGF1, an anabolic and neuroprotective factor, promotes neuronal survival by blocking apoptosis. It is released into the bloodstream by the liver, or synthesized locally by muscles and neural cells, acting in an autocrine or paracrine fashion. Intriguingly, genetic studies conducted in invertebrate and murine models also suggest that an excess of IGF1 signaling may trigger neurodegeneration. This emphasizes the importance of gaining a better understanding of the mechanisms controlling IGF1 regulation and gene transcription. In the cerebellum, Igf1 expression is activated just before birth in a subset of Purkinje cells (PCs). Mice carrying a null mutation for HLH transcription factor EBF2 feature PC apoptosis at birth. We show that Igf1 is sharply downregulated in Ebf2 null PCs starting before the onset of PC death. In vitro, EBF2 binds a conserved distal Igf1 promoter region. The pro-survival PI3K signaling pathway is strongly inhibited in mutant cerebella. Finally, Ebf2 null organotypic cultures respond to IGF1 treatment by inhibiting PC apoptosis. Consistently, wild type slices treated with an IGF1 competitor feature a sharp increase in PC death. Our findings reveal that IGF1 is required for PC survival in the neonatal cerebellum, and identify a new mechanism regulating its local production in the CNS.

Figure 1

Ebf2 null PCs die by apoptosis shortly after birth. (a–d) Double immunofluorescence on P0.5 sagittal sections, as labeled. In wild-type (wt) mice, cell death, as assessed by activated caspase 3 (AC3) immunoreactivity, is only detectable in the prospective IGL/white matter (pwm), not in the PC layer (arrowheads in a). In the mutant cerebellum, most apoptotic cells are retinoic acid-related orphan receptor α (RORα)-positive PCs (b, arrows in c). However, dying PCs, positive for calbindin (CaBP), are also found in the mutant pwm (arrows in d). (e–h), TUNEL assay on frontal sections from wt (e,g) and null mutant (f,h) cerebella. DNA fragmentation (red fluorescent dots) is evident only in the Ebf2 null PC layer (pcl) (arrows in f,h). Note the comparable levels of apoptosis in the external granular layer (egl) of Ebf2 null versus wt control sections. (i–k) In the mutant cerebellum only few TUNEL+ cells located in the pwm are positive for Pax2, Pax6 or Olig2 (arrows). (l–o) Electron microscopic imaging of wt (l,n) and Ebf2 null (m,o) PCs. Wt PCs feature large nuclei with round nucleoli (asterisk). Their cytoplasm is rich in polysomes (circled area) and contains intact mitochondria (arrowheads). Conversely, in the mutant cerebellum (m,o), numerous apoptotic bodies are found, characterized by a pycnotic nucleus (nc) and condensed cytoplasm (arrow). Scale bar: a–c, 200 μm; d, 50 μm; e–h, 100 μm; i–k, 20 μm; l,n,o, 1 μm; m, 2 μm

Figure 2

Ebf2 and Igf1 are co-expressed in PCs shortly after birth. (a) β-gal, a reporter of Ebf2 expression in Ebf2^+/LacZ heterozygotes, is preferentially expressed in PCs located in the anterior lobe (a, higher magnification in b). (c) In situ hybridization on sagittal cerebellar sections hybridized with an Ebf2 or with an Igf1 cRNA probe. PCs that express Igf1 are also positive for Ebf2 (c). (d) Double immunostaining for CaBP and IGF1 on sagittal cerebellar sections; (e) higher magnification in (d). At both the transcript and protein levels, Ebf2 and IGF1 are found expressed in PCs. pcl: Purkinje cell layer; egl: external granular layer. Scale bar: a,d, 500 μm; b, 40 μm; c, 200 μm; e, 20 μm

Figure 3

Igf1 expression is downregulated in Ebf2 null PCs. (a–d) In situ hybridization on E19.5 and P0.5 frontal cerebellar sections (sectioning plane sketched at bottom left). In the wt, Igf1 is detectable in a subset of PCs as revealed by comparing its distribution to that of RORα in adjacent sections. The Igf1 transcript is almost absent in Ebf2 null PCs (arrows). (e) Western blot analysis of IGF1 protein shows a significant reduction in Ebf2 null lysates compared to the wt (protein levels plotted in f). The experiment was repeated on protein extracts from three wt and mutant cerebella with the S.D. in each case. pcl, Purkinje cell layer; egl, external granular layer. Scale bar: 200 μm. ^*P<0.05

Figure 4

EBF2 activates Igf1 expression in P19 cells. (a) To assess the efficiency of the Ebf2-specific shRNA in downregulating EBF2 protein levels, we overexpressed a flag-tagged Ebf2 in HEK-293 cells that do not exhibit detectable levels of endogenous Ebf2. Western blotting results show that flag-Ebf2-overexpressing HEK-293 cells cotransfected with an Ebf2 shRNA (Ebf2-sh) exhibit a complete downregulation of flag-EBF2 protein levels, compared to cells cotransfected with a mock sh-RNA (mock-sh). (b) Quantitative analysis of the experiment described in (a). Band intensities in sh-treated samples, relative to mock treatment, were measured using the ImageQuant software. All protein levels were also normalized to β-actin. (c) Time course of endogenous Ebf2 gene expression in neuralized P19 cells after EBF2 knock-down with the Ebf2-sh vector (black columns), relative to the corresponding transcript level in cells treated with mock shRNA (empty column), as measured by RT-qPCR. (d) Time course of endogenous Igf1 gene expression in cells treated as detailed in c (RT-qPCR). (e) Induction of endogenous Igf1 gene expression in neuralized P19 cells transfected with an Ebf2 expression vector after 24 h (RT-qPCR). (c–e) S.D. are indicated in each histogram. All the experiments shown were repeated three times. ^*P<0.05; ^** P<0.01

Figure 5

EBF2 binds two sequences of Igf1 promoter 1 and enhances promoter activity. (a) Results of luciferase assays in COS-7 cells transiently transfected with an expression plasmid for Ebf2 (or an empty vector) and with the Igf1 promoter–reporter plasmids, as sketched. The graph summarizes results of four independent experiments (mean±S.E.), each performed in duplicate and normalized for total protein content with the S.D. in each case. The asterisks indicate significant difference in EBF2-induced promoter activity (P<0.03) in comparison to the –1711 promoter–reporter plasmid that spans putative-binding sites IGF EBF-A and –B. Basal luciferase values ranged from 10 000 to 135 000 relative light units/10 s. (b) Results of gel-mobility shift assays using IR-labeled double-stranded oligonucleotides for either IGF EBF-A or -B of rat Igf1 promoter 1. The DS oligos were incubated with nuclear protein extracts from COS-7 cells transfected with expression plasmids encoding FLAG-tagged EBF2 or the unrelated STAT5b. Protein–DNA complexes are observed only with nuclear extracts from EBF2-expressing cells. 50-fold excess of unlabeled homologous oligonucleotide and a previously described EBF-binding site of the OcNC gene effectively compete off binding, whereas an Oct-1 recognition site does not. A supershift of each protein–DNA complex (^*) is seen with an antibody to the FLAG-epitope, but not with an irrelevant antibody directed against 11 amino acids from the T7 phage gene 10 leader peptide. Unbound probe is observed at the bottom of each gel

Figure 6

Reduced AKT1 phosphorylation in the Ebf2 null cerebellum. (a) P0.5 sagittal sections from an Ebf2^+/LacZ cerebellum immunostained for CaBP and the IGF1 effector AKT, phosphorylated in Ser⁴⁷³. Phospho-AKT staining in PCs reveal the activation of the IGF1R/PI3K-signaling pathway in these cells, specifically in dendrites (arrows in b). (c) Cerebellar protein extracts from wt and Ebf2 null mice were analyzed by western blotting at the same stage (P0.5). The levels of AKT phosphorylation at both Ser473 and Thr308, and GSK3β phosphorylation at Ser9 are dramatically reduced in Ebf2 –/– lysates, whereas total AKT and IGF1R levels are unchanged. (d) The intensity of immunoreactivity was quantified using the ImageJ software. Protein levels were normalized to β-actin. Each immunoblotting experiment was repeated on protein extracts from four independent wt and null mutant mice. S.D. is shown in each case. Size bars: (a) 500 μm; (b) 20 μm. ^*P<0.05; ^**P<0.01

Figure 7

IGF1 is essential for PC survival in organotypic cultures of the perinatal cerebellum. (a,b) E19.5 wt cerebella were divided in two parts and each hemicerebellum was treated with an IGF1 competitor (H1356 Bachem) (a) or vehicle (1 × PBS) (b), and maintained in culture for 2 days. The number of apoptotic cells is revealed by immunostaining for CaBP and active caspase 3 (AC3). Cell death is sharply increased in competitor-treated slices (plotted in c). (d–g) Cerebella from Ebf2 null mutants and wt controls were divided in two parts and each hemicerebellum was treated with vehicle (1 × PBS in d,f) or murine purified IGF1 (e,g), and maintained in culture for 2 days. The number of apoptotic PCs was dramatically reduced in IGF1-treated Ebf2 null slices (e) in comparison to vehicle-treated null slices (d). Conversely, IGF1 treatment had no effect on cell death in control slices when compared to vehicle treatment (f and g, respectively). (c,h) Statistical significance of both experiments was estimated computing the ratio of AC3-positive cells located in the PC layer to the total number of CaBP-positive PCs, relative to IGF1 treatment. All treatments were repeated three times for each genotype. The S.D. is indicated in all cases examined. ^*P<0.05, **P<0.01. Size bar: 50 μm