Ablation of NF1 function in neurons induces abnormal development of cerebral cortex and reactive gliosis in the brain

Abstract

Neurofibromatosis type 1 (NF1) is a prevalent genetic disorder that affects growth properties of neural-crest-derived cell populations. In addition, approximately one-half of NF1 patients exhibit learning disabilities. To characterize NF1 function both in vitro and in vivo, we circumvent the embryonic lethality of NF1 null mouse embryos by generating a conditional mutation in the NF1 gene using Cre/loxP technology. Introduction of a Synapsin I promoter driven Cre transgenic mouse strain into the conditional NF1 background has ablated NF1 function in most differentiated neuronal populations. These mice have abnormal development of the cerebral cortex, which suggests that NF1 has an indispensable role in this aspect of CNS development. Furthermore, although they are tumor free, these mice display extensive astrogliosis in the absence of conspicuous neurodegeneration or microgliosis. These results indicate that NF1-deficient neurons are capable of inducing reactive astrogliosis via a non-cell autonomous mechanism.

Figure 1

Generation of NF1 flox mice. (A) Schematic of the wild-type (WT) NF1 allele, targeting vector, and the flox allele. A loxP site with the pGKneo cassette was inserted into intron 30, and the second loxP site was introduced into intron 32. In this manner, exons 31 and 32 of the NF1 gene are flanked by two loxP sites. A 5′ external probe (probe A) and a 3′ internal probe (probe B) for Southern analysis are shown. P, PstI; H, HindIII; R, EcoRI. (B) (Left panel) Southern analysis with 5′ probe (probe A) shows a mutant (left lane) and a WT (right lane) ES clones. The mutant allele produces a 10.2-kb fragment; the WT allele produces an 11.8-kb fragment. (Right panel) Genomic DNA from the same ES clones hybridized with 3′ probe (probe B), confirming that the mutant ES clone has undergone homologous recombination in the NF1 locus. The mutant allele produces a 1.2-kb fragment; the WT allele produces a 3.2-kb fragment. (C) Southern analysis with the probes (as described above) on the littermates of NF1flox/+ intercross. (D) Southern analysis with 3′ probe shows different alleles of the NF1 gene: WT allele (3.2 kb), knockout (KO) allele (2.6 kb), and flox allele (1.2 kb). These mice were subjected to survival study. (E) Survival curve of NF1flox/flox and flox/KO mice. It shows the normal survival profile of NF1 flox allele. (F) A representative E13.5 NF1KO/Δ and WT littermate.

Figure 1

Generation of NF1 flox mice. (A) Schematic of the wild-type (WT) NF1 allele, targeting vector, and the flox allele. A loxP site with the pGKneo cassette was inserted into intron 30, and the second loxP site was introduced into intron 32. In this manner, exons 31 and 32 of the NF1 gene are flanked by two loxP sites. A 5′ external probe (probe A) and a 3′ internal probe (probe B) for Southern analysis are shown. P, PstI; H, HindIII; R, EcoRI. (B) (Left panel) Southern analysis with 5′ probe (probe A) shows a mutant (left lane) and a WT (right lane) ES clones. The mutant allele produces a 10.2-kb fragment; the WT allele produces an 11.8-kb fragment. (Right panel) Genomic DNA from the same ES clones hybridized with 3′ probe (probe B), confirming that the mutant ES clone has undergone homologous recombination in the NF1 locus. The mutant allele produces a 1.2-kb fragment; the WT allele produces a 3.2-kb fragment. (C) Southern analysis with the probes (as described above) on the littermates of NF1flox/+ intercross. (D) Southern analysis with 3′ probe shows different alleles of the NF1 gene: WT allele (3.2 kb), knockout (KO) allele (2.6 kb), and flox allele (1.2 kb). These mice were subjected to survival study. (E) Survival curve of NF1flox/flox and flox/KO mice. It shows the normal survival profile of NF1 flox allele. (F) A representative E13.5 NF1KO/Δ and WT littermate.

Figure 2

NF1−/− neurons generated by Cre-mediated recombination survive in the absence of neurotrophins. (A) Schematic drawing of the NF1flox allele and the recombined allele resulted from Cre-mediated recombination. Southern analysis with 3′ probe (probe B) produces an 11.6-kb fragment from the flox allele and knockout (KO) allele, and a 7.2-kb fragment from the recombined (Δ) allele. Thus, the flox allele and recombined flox allele (Δ) can be distinguished by Southern analysis. Positions of primers (P1, P2, P3, and P4) are illustrated. The combination of P1, P3, and P4 primers is used to genotype NF1flox mice; the pair of P1 and P2 primers (R-Cre PCR) is used to detect Cre-mediated recombination, P1 and P4 primers are used to detect the presence of the flox allele (flox-PCR; see Materials and Methods). H, HindIII. (B) Southern analysis of NF1flox/KO dorsal root ganglion (DRG) after exposure to Cre-adenovirus. Genomic DNA from NF1flox/KO DRG cultures infected with Cre-adenovirus were hybridized with probe B. The multiplicity of infection (MOI) of each culture is illustrated. Of note, in this restriction analysis (HindIII), the KO allele produces a similar fragment (11.6 kb) as the flox allele. DRG cultures with MOI > 20 underwent complete deletion: The 11.6 kb band represents the KO allele; the 7.2-kb band represents the deleted (Δ) allele after Cre-mediated recombination. (C) Neurotrophin survival assay on E13.5 DRG cultures from NF1 flox/KO or flox/flox E13.5 embryos. Cultures were either mock infected or subjected to LacZ-, and Cre-adenovirus for 3 d in the presence of NGF (10 ng/mL). NGF was then removed by dilution of medium and addition of anti-NGF antibody. Cell counts were performed every 24 h for 5 d after plating. The results represent the mean of three independent experiments.

Figure 3

Characterization of Synapsin I–Cre transgene (Tg) expression. E12.5 embryos (A) Cre Tg, (B) LacZ Tg, (C) and (D) Cre/LacZ double Tg were subjected to LacZ staining. Only the compound transgenic mice exhibit Cre-mediated activation of the β-galactosidase gene. Double-labeling immunoanalysis of Cre/LacZ double Tg with anti-NeuN and anti-LacZ antibodies for the cortex (E–J); spinal cord (K–M); hippocampus (N–P); and cerebellum (Q–S). Colocalization of LacZ and NeuN signals indicates that Cre-mediated recombination specifically occurs in differentiated neurons. Of note, the neuronal marker, NeuN is not expressed in the Purkinje cells. Ctx, cortex; SC, spinal cord; Hp, hippocampus; and Cb, cerebellum. Objective magnification: E–G and N–P, 20×; H–M and Q–S, 40×.

Figure 4

NF1^SynIKO mice lose neurofibromin immunoreactivity in most neuronal populations. (A–D) Anti-NF1 immunoreactivity in the cortex (Ctx), cerebellum (Cbx), and spinal cord (SC) of control mice showing dark staining of neurons. (E–H) Loss of NF1 staining in the majority of neuronal populations in the mutant mice. An exception, however, is a subset of the Purkinje neurons in the Cbx (G), which retain NF1 immunoreactivity. A–C are from the same section, as are E–G. Scale bar, 50 μm.

Figure 5

NF1^SynIKO astrocytes do not express Cre recombinase. Double immunoanalysis with anti-LacZ/GFAP (A,C) and anti-Cre/GFAP (B,D). In the CA2/3 regions (A) and dentate gyrus (DG) (C) of the hippocampus, most of LacZ positive cells are localized in the neuronal layer with big nuclei. Of note, the inset shows the relative size of neuronal and small astrocytic nuclei and examples of colocalization of LacZ and glial fibrillary acidic protein (GFAP) signals. The arrow in A points out the apparent colocalization of LacZ and GFAP signals, which is actually the result of independent cells on two planes of the 50 μm section. (C) Some red nuclei are visible outside the DG. These are large nuclei that do not overlap with GFAP and likely represent interneuron nuclei. (D) High magnification of B, showing that Cre-positive cells are not GFAP positive. Scale bar, 50 μm.

Figure 6

NF1^SynIKO mice have reduced cerebral cortex thickness and increased cell density. Serial coronal sections of brains from control and mutant littermates were cut at 50 μm and analyzed by hematoxylin and eosin (H&E) staining, NeuN, MAP2, and Golgi staining. (A,E) Control cortex, H&E; (B,F) mutant cortex, H&E; (C,G) control cortex, NeuN; (D,H) mutant cortex, NeuN; the reduced cortical thickness is readily apparent in the mutant cortex. (I) Control cortex, MAP2; (J) mutant cortex, MAP2, arrows point out the presence of MAP2-positive pyramidal neurons in the mutant cortex; (K) control and (L) mutant cortex, Golgi stain. The morphology of the stained neurons appeared normal in the mutant cortex. Scale bar, 50 μm.

Figure 7

Increased cell density in the cerebral cortex of NF1^SynIKO mice. Anterior-to-posterior morphometric analysis of the secondary somatosensory cortex (S2 Ctx) revealed a significant reduction in cortical thickness in NF1^SynIKO mice compared to control (CTR) animals. The reduction was significant particularly at caudal levels; −0.94 (P < 0.05), −1.46, and −1.94 (P < 0.01) mm from Bregma (A). A comparison of the average cortical thickness in these mice showed an ∼20% decrease (P < 0.05) in the NF1^SynIKO mice (B). Despite the reduced cortical thickness in mutant mice, the rest of the brain size is comparable to that in CTR mice. Measurement of the horizontal brain length, excluding the cortex (H-Ctx) in mutant mice showed no significant differences throughout the anterior-to-posterior axis when compared to CTR mice (C). Quantitative analysis of the total number of cells in the cortex failed to demonstrate any differences because of genotype (D). *, P < 0.05; **, P < 0.01.

Figure 8

NF1^SynIKO mice display extensive astrogliosis. Serial coronal sections were cut at 50 μm thickness (cryostat) or 5 μm thickness (paraffin), and subjected to GFAP immunoanalysis. (A–D) Four sections from rostral to caudal, with the left hemispheres in each panel from control mice and the right hemispheres from mutant mice. Note the overall darker brown staining in the right hemispheres, indicating extensive astrogliosis in the mutant brains. (E–H) High magnification of the control cortex (the boxed area of the left hemisphere in A–D). (I–L) High magnification of the mutant cortex (the boxed area of the right hemisphere in A–D). (M–O) Control and (R–T) mutant hippocampus. Note the hypertrophic nature of the reactive astrocytes (compare T to O). Control (P,Q) and mutant (U,V) brainstem. Scale bar: A–D, 1 mm; E–V, 50 μm.

Figure 9

The number of glial fibrillary acidic protein (GFAP)-positive cells is significantly increased in the mutant brains. The quantification of GFAP-positive cells in the cortex (A), hippocampus (B), and brainstem (C). (D) Mean value ± SEM (number of brains examined).

Figure 10

Activation of MAP kinase in NF1^SynIKO neurons. (A) (Top panel) Western blot analysis of tissues from different parts of control and mutant brains using anti-phosphorylated erk (P-erk) antibody. (Bottom panel) The same blot was stripped and reprobed with anti-P-erk antibody as a loading control. (B) P-erk and erk signals were quantified by densitrometry (see Materials and Methods). The level of P-erk in each sample is normalized to level of erk protein signal. Each number represents mean value ± standard error of the mean (SEM) of ratio of P-erk levels of three mutants to three controls. Sections of cerebral cortex from control (C) and mutant mice (D) were analyzed with anti-P-erk antibody. The mutant cortex exhibits greater number of phospho-erk positive cells corresponding to neurons and not to reactive astrocytes. E and G are high magnification of the top and bottom boxed area of C; F and H are high magnification of the top and bottom boxed area of D. Ctx, cerebral cortex; Hp, hippocampus; Bs, brainstem; Cb, cerebellum; CC, corpus callosum. Scale bar, 100 μm.