AF4 is a critical regulator of the IGF-1 signaling pathway during Purkinje cell development

Abstract

Deregulation of the insulin-like growth factor 1 (IGF-1) signaling pathway is a recurrent finding in mouse models and human patients with cerebellar ataxia and thus represents a common pathological cascade in neuronal cell death that may be targeted for therapy. We have previously identified a point mutation in AF4, a transcription cofactor of RNA polymerase II elongation and chromatin remodeling, that causes progressive and highly specific Purkinje cell (PC) death in the ataxic mouse mutant robotic, leading to the accumulation of AF4 in PCs. Here we confirm that the spatiotemporal pattern of PC degeneration in the robotic cerebellum correlates with the specific profile of AF4 upregulation. To identify the underlying molecular pathways, we performed microarray gene expression analysis of PCs obtained by laser capture microdissection (LCM) at the onset of degeneration. Igf-1 was significantly downregulated in robotic PCs compared with wild-type controls before and throughout the degenerative process. Consistently, we observed a decrease in the activation of downstream signaling molecules including type 1 IGF receptor (IGF-1R) and the extracellular signal-regulated kinase (ERK) 1 and ERK2. Chromatin immunoprecipitation confirmed that Igf-1 is a direct and the first validated target of the AF4 transcriptional regulatory complex, and treatment of presymptomatic robotic mice with IGF-1 indeed markedly delayed the progression of PC death. This study demonstrates that small changes in the levels of a single transcriptional cofactor can deleteriously affect normal cerebellum function and opens new avenues of research for the manipulation of the IGF-1 pathway in the treatment of cerebellar ataxia in humans.

Figure 1.

Spatiotemporal expression of the Af4 gene in PCs during postnatal cerebellum development. A, B, Af4 transcript levels were determined by qRT-PCR from PCs of wild-type mice obtained by LCM (n = 3). Analysis was performed from all 10 lobes (A), or from individual cerebellar regions (B). Results are shown ± SD. The asterisks indicate significant differences between time point expression values (**p < 0.01, ***p < 0.001).

Figure 2.

Af4 gene expression is not altered in robotic PCs. Af4 transcript levels were determined by qRT-PCR from PCs of wild-type (WT) and robotic (ROB) mice obtained by LCM (n = 3). Results are shown ± SD.

Figure 3.

AF4 protein accumulates in robotic PCs. Immunostaining of AF4 in parasagittal cerebellum sections from wild-type (WT) and robotic (ROB) mice. To enable a direct comparison, pictures were taken using the same exposure time. The localization of AF4 to PCs was confirmed by double staining for calbindin (data not shown). Images show lobe III only and are representative of the results obtained for WT and ROB littermates from three independent litters. Scale bar, 100 μm.

Figure 4.

Igf-1 gene expression is decreased in robotic PCs. Igf-1 transcript levels were determined by qRT-PCR from PCs of wild-type (WT) and robotic (ROB) mice obtained by LCM (n = 3). Results are shown ± SD. The asterisks indicate significant differences between WT and ROB expression values (*p < 0.05, ***p < 0.001).

Figure 5.

The AF4 transcriptional regulatory complex binds to the Igf-1 locus in the cerebellum. Cerebellum from wild-type and robotic mice at the indicated age (n = 3 in each group) were subjected to ChIP analysis using the AF4 antiserum. The isolated chromatin was used as a template for qPCR with 4 primer sets designed within the 5′ untranslated (1), coding (2 and 3), and 3′ untranslated (4) regions of the Igf-1 genomic sequence. Nonspecific binding was determined through the use of negative ChIP controls and was subtracted, as described in Materials and Methods. Results are expressed as a percentage of input, in fold change of wild-type in robotic.

Figure 6.

Activation of the IGF-1 signaling pathway is reduced in robotic PCs. Immunostaining of parasagittal cerebellum sections of P21 wild-type (WT) and robotic (ROB) mice for p-IGF-1R and p-ERK1/2. To enable a direct comparison between WT and ROB, pictures were taken using the same exposure time. Pictures show lobe III only and are representative of the results obtained for WT and ROB littermates from three independent litters. Scale bar, 100 μm.

Figure 7.

Igbp-3 gene expression is decreased in robotic PCs. Igfbp-3 transcript levels were determined by qRT-PCR from PCs of wild-type (WT) and robotic (ROB) mice obtained by LCM (n = 3). Results are shown ± SD. The asterisks indicate significant differences between WT and ROB expression values (**p < 0.01, ***p < 0.001).

Figure 8.

Igf-1 gene expression is altered in neuronal and non-neuronal affected tissues of the robotic and Af4 KO mice. Igf-1 transcript levels were determined by qRT-PCR from tissues of robotic (ROB), and Af4 heterozygous (Af4^−/+) and homozygous (Af4^−/−) KO mice (n = 3). Analysis was performed in the cerebellum, thymus, and bone marrow at P21, and in the lens at P70. Results are shown in fold change of wild-type in robotic or KO mutants ± SD.

Figure 9.

Treatment of robotic mice with IGF-1 slows down PC death. P7 robotic mice were injected daily for 4 weeks with IGF-1 at 1 mg/kg or the vehicle (n ≥ 7 in each group). A, Analysis of PC degeneration in the cerebellum following IGF-1 treatment at P70. Parasagittal sections were immunostained for calbindin. Typical signs of PC degeneration including swelling of PC dendrites (arrow) and axonal torpedoes (arrowhead) can be observed. Scale bars, 1 mm (lobes I–III); 100 μm (lobe IV). B, Analysis of IGF-1 pathway activation in the cerebellum following IGF-1 treatment at P70. Parasagittal sections were immunostained for IGF-1, p-IGF-1R, and p-ERK1/2. To enable a direct comparison, pictures were taken using the same exposure time. Scale bar, 100 μm. In A and B, images are representative of the results obtained for three animals from each group.