Hypoxic preconditioning differentially affects GABAergic and glutamatergic neuronal cells in the injured cerebellum of the neonatal rat

Abstract

In this study we examined cerebellar alterations in a neonatal rat model of hypoxic-ischemic brain injury with or without hypoxic preconditioning (Pc). Between postnatal days 7 and 15, the cerebellum is still undergoing intense cellular proliferation, differentiation and migration, dendritogenesis and synaptogenesis. The expression of glutamate decarboxylase 1 (GAD67) and the differentiation factor NeuroD1 were examined as markers of Purkinje and granule cells, respectively. We applied quantitative immunohistochemistry to sagittal cerebellar slices, and Western blot analysis of whole cerebella obtained from control (C) rats and rats submitted to Pc, hypoxia-ischemia (L) and a combination of both treatments (PcL). We found that either hypoxia-ischemia or Pc perturbed the granule cells in the posterior lobes, affecting their migration and final placement in the internal granular layer. These effects were partially attenuated when the Pc was delivered prior to the hypoxia-ischemia. Interestingly, whole nuclear NeuroD1 levels in Pc animals were comparable to those in the C rats. However, a subset of Purkinje cells that were severely affected by the hypoxic-ischemic insult--showing signs of neuronal distress at the levels of the nucleus, cytoplasm and dendritic arborization--were not protected by Pc. A monoclonal antibody specific for GAD67 revealed a three-band pattern in cytoplasmic extracts from whole P15 cerebella. A ∼110 kDa band, interpreted as a potential homodimer of a truncated form of GAD67, was reduced in Pc and L groups while its levels were close to the control animals in PcL rats. Additionally we demonstrated differential glial responses depending on the treatment, including astrogliosis in hypoxiated cerebella and a selective effect of hypoxia-ischemia on the vimentin-immunolabeled intermediate filaments of the Bergmann glia. Thus, while both glutamatergic and GABAergic cerebellar neurons are compromised by the hypoxic-ischemic insult, the former are protected by a preconditioning hypoxia while the latter are not.

Figure 1. Effect of treatment conditions on NeuroD1 and vimentin-stained cells in the posterior cerebellum.

Double immunolabeling for the pro-neuronal marker NeuroD1 (brown, HRP-DAB) and the glial intermediate filament protein vimentin (grey-black, HRP-DAB-nickel chloride) in the ML (A–D) and IGL (E–H) of control (C) and treated animals: hypoxic preconditioning (Pc), hypoxic-ischemic lesion (L) and injury with prior preconditioning (PcL). In A–D, migrating spindle-like NeuroD1-positive cells (black arrowheads) lying on the vimentin-positive Bergmann glial scaffolds (narrow black arrows) are shown. In panel C, fewer migrating cells are present in the ML of L animals, some of the NeuroD1-positive nuclei show a round profile (thick black arrow), while the vimentin labeling of the Bergmann glia projections is attenuated compared to the other three conditions. In D, an apparent recovery in the number of cells migrating on the vimentin-labeled Bergmann glia scaffold is observed in the PcL cerebellum. E-H: Images of the IGL from the same animals with intense NeuroD1 labeling of the granule cell nuclei (white arrows). Note the high number of granule cells in the C group (E) compared to the more scattered pattern in the Pc and L animals (F–G). In contrast, the PcL cerebellum shows an increased density of granule cells (H) compared to the individual treatments. Pc induces vascularization (white arrowhead, F). A–H: 60x, scale bar: 20 µm. I: Assembly of 4× images of the whole cerebellum immunolabeled with the anti-NeuroD1 antibody. The rectangle illustrates the posterior lobes where the quantifications of NeuroD1-positive cells in the ML (J) and in the IGL (K) were performed. Data are expressed as mean ± SEM in an area of 1×10⁻² mm². EGL: External granular layer. IGL: Internal granular layer. ML: Molecular layer. ND1: NeuroD1. VIM: Vimentin. Statistics: One-way ANOVA followed by Bonferroni's post-hoc tests; p<0.05; 0.01 and 0.001 (*, **, ***, respectively) were considered significant.

Figure 2. Cerebellar NeuroD1 levels under treatment conditions.

Nuclear extract proteins from whole cerebella were analyzed for NeuroD1 and histone H3 levels via Western blot. A: Representative blot of three independent experiments using different litters from the four experimental groups showing a specific band for NeuroD1 of about 50 kDa (black arrow). The bands for the nuclear marker histone H3 for the same membrane are shown at the bottom. Quantifications expressed as the mean of the optical density (OD) of NeuroD1 relative to histone H3 ± SEM are graphed in B. H3: Histone H3. ND1: NeuroD1. MW: Molecular weight. Statistical analysis: One-way ANOVA followed by Bonferroni's post-hoc tests; p<0.05 and p<0.01 (* and **, respectively) were considered significant.

Figure 3. Treatment effects on cerebellar Purkinje cells revealed by GAD67 immunostaining.

Posterior lobes of the four different groups immunolabeled for the GABAergic enzyme GAD67, and developed with HRP-DAB (A–D) and Alexa Fluor 488 (E–H) are shown. Both methods preferentially stain a string of large and evenly distributed cells in all treatment groups. Under control conditions (A and E), cell density and regularity in the PkL were highest. In the experimental animals that suffered hypoxia alone (Pc) or hypoxia and ischemia (L and PcL), the continuity of the string appears disrupted and the somas of some of these neurons show morphological modifications (B–D and F–H). In the three treated groups, at least two subpopulations of Purkinje cells were clearly distinguished that we categorized as normal (white arrowhead) or abnormal (white arrow) (I and I’ for the L group). In these enlarged images counterstained with propidium iodide (PI) to reveal nuclear compaction, the intensely stained nuclei of the abnormal cells and the mis-orientation of one of them with its long axis parallel to the ML and IGL are clearly seen. The quantification of the number of GAD67-positive Purkinje cells in the whole cerebella of the different groups of animals is shown in (J). The percentage of abnormal cells was also calculated and graphed (K). The data are expressed as the mean ± SEM in an area of 4×10⁻² mm². A-D: Optical microscopy, 10x, scale bar: 150 µm. E–H: Confocal microscopy, 20x, scale bar: 100 µm. I-I’: Epifluorescence microscopy, 2× zoom from 40x, scale bar: 15 µm. GAD67: Glutamate decarboxylase 67. PkL: Purkinje cell layer. Statistical analysis: One-way ANOVA with Bonferroni's post-hoc tests; (***) represents p<0.001 between the groups indicated by the horizontal bars on top.

Figure 4. Total GAD67 protein expression in the rat cerebellum under treatment conditions.

Levels of GAD67 in cytoplasmic extracts from whole cerebella were determined by WB and normalized for total actin in each sample. A: Western blot of one of three independent experiments using different litters from the four experimental groups. The black arrow points to a band of about 110 kDa, the white arrow points to a 67 kDa band, and the black arrowhead to a band of around 55 kDa (top). Actin was used as a control for protein loading (bottom). B–D: Quantification of the ∼110 kDa, 67 KDa and ∼55 kDa bands from the different experimental groups. Data were expressed as mean ± SEM. Statistical analysis: One-way ANOVA and Bonferroni's post-hoc tests. Comparisons between groups are shown by horizontal bars at the top; p<0.001 (***) was considered significant.

Figure 5. β–tubulin III (Tuj1) expression in cerebella under treatment conditions.

A–D: Combined Tuj1 immunolabeling (green, Alexa Fluor 488) and PI staining of cell nuclei (red). Purkinje cells with normal and abnormal morphologies are indicated with white arrowheads and white arrows, respectively. E–H: The same fields viewed with the Tuj1 channel alone. Prominent dendritic trunks oriented towards the ML are observed in control animals (A and E, black arrow with white borders), but not under the treatment conditions. E’-H’: Enlargements of the insets shown in E–H. The IGL from control animals shows high cellular density with granule cell nuclei surrounded by narrow bands of cytoplasm and abundant highly immunoreactive spots (E’: black arrowheads with white borders). In the IGL from the Pc and L groups granule cells show a wider immunolabeled cytoplasm (F’-G’). In the PcL cerebellum the IGL has an intermediate appearance, between that of the untreated animals and the same layer in L and Pc rats (H’). n: Granule cell nuclei. *: Wider Tuj-1-positive cytoplasms. A-H: Confocal microscopy, 60x, scale bar: 20 µm. E’-H’: 4.5× digital zoom, scale bar: 10 µm.