Reorganization of Synaptic Connections and Perineuronal Nets in the Deep Cerebellar Nuclei of Purkinje Cell Degeneration Mutant Mice

Abstract

The perineuronal net (PN) is a subtype of extracellular matrix appearing as a net-like structure around distinct neurons throughout the whole CNS. PNs surround the soma, proximal dendrites, and the axonal initial segment embedding synaptic terminals on the neuronal surface. Different functions of the PNs are suggested which include support of synaptic stabilization, inhibition of axonal sprouting, and control of neuronal plasticity. A number of studies provide evidence that removing PNs or PN-components results in renewed neurite growth and synaptogenesis. In a mouse model for Purkinje cell degeneration, we examined the effect of deafferentation on synaptic remodeling and modulation of PNs in the deep cerebellar nuclei. We found reduced GABAergic, enhanced glutamatergic innervations at PN-associated neurons, and altered expression of the PN-components brevican and hapln4. These data refer to a direct interaction between ECM and synapses. The altered brevican expression induced by activated astrocytes could be required for an adequate regeneration by promoting neurite growth and synaptogenesis.

Figure 1

Labelling of calbindin expressing PC neurons in wt and pcd mice. Purkinje cells and their axons in the cerebellum show a strong immunoreactivity for calbindin. In one month old wt anti-calbindin antibodies detect the Purkinje cells and their axon in the cerebellum. The neurons of the cerebellum in one-month-old pcd mice reveal less calbindin immunoreactivity. Scale bar: 100 μm (a) and 50 μm (b).

Figure 2

Detection of glutamatergic and GABAergic terminals in DCN. The large DCN neurons are enwrapped by aggrecan-based ECM (red). (a) DCN neurons are innervated by GABAergic boutons, labeled by GAD 65/67. GABAergic terminals seem to be reduced in pcd. (b) The glutamatergic boutons at DCN neurons are discovered by moderate vGlut 1 and vGlut 2 staining. The staining in pcd appears slightly enhanced. Scale bar: 20 μm.

Figure 3

Quantification of GAD65/67 and vGlut in DCN. (a) Quantification shows the distribution of GABAergic terminals in different distances from the PN-bearing neurons. The total number of boutons are reduced in pcd compared to wt, regardless of the distance. (b) Somatic glutamatergic terminals at DCN neurons appear to be enhanced in pcd mice. The peripheral synapses remain unaffected by the insult. (c) Western blot analyses of GAD65/67 and vGlut1 with protein extracts of DCN sections. Typical specific bands are visible in both genotypes. Quantification of these bands reveals slight but no significant differences between wt and pcd (GAD65/67 p = 0.419; vGlut1 p = 0.087). Data are given as mean ± SEM.

Figure 4

Detection of different brevican fragments. (a) Immunoreaction with pan-specific brevican antibodies (BD Bioscience, FL) clearly surrounds DCN neurons in wt but not in pcd mice. As internal control nontarget region of PC axons, the cochlear nucleus shows no alterations in immunoreactions with FL. (b) The 50 kDa isoform of brevican seems to be nearly absent around DCN neurons of pcd mice, whereas the not affected region (CN) revealed brevican-bearing neurons. (c) The B756 antibodies detect mainly the 80/90 kDa and full length isoforms of brevican. These cleavage products aggregate around the neurons in DCN of wt mice and seem to be integrated in PNs. In pcd, PNs appear with lower intensity, but with potential higher parenchymatic reaction. PN-detection with all three antibodies in the internal nontarget control region (CN) is unchanged. Scale bar: 20 μm. (d) Biochemical detection of brevican with SDS-PAGE with pan-specific antibodies revealed most known isoforms at 50, 80, 90, and 145 kDa. Quantification of the 50 and 145 kDa brevican isoform showed a significant increased protein expression in pcd (p < 0.001), respectively, for the different isoforms. Therefore, the diagram is supposed to display the optical density (OD) values of pan-brevican chemiluminescent signal summed values (OD 50 kDa + 145 kDa brevican/actin). Data are given as mean ± SEM.

Figure 5

Comparison of link protein expression in DCN. DCN neurons are visualized by aggrecan immunoreaction (red). (a) Hapln1 labeling (green) surrounds the DCN neurons in both genotypes, matching the aggrecan immunoreactivity; additionally in pcd hapln1 immunoreaction is distributed throughout the whole parenchyma. (b) Hapln4 (green) encloses the DCN neurons in wt mice. In contrast, hapln4 in pcd exhibits virtually no immunoreaction. Scale bar: 20 μm. (c) Western blot reveals protein bands at approximately 40 kDa for link proteins. Quantification of the link proteins yielded an elevated protein level of both components in pcd (hapln1 p < 0.01; hapln4 p < 0.05). Data are given as mean ± SEM.

Figure 6

Distribution of hyaluronan and tenascin-R in DCN. The labeling shows important constituents of PNs. The neurons in DCN are surrounded by strong aggrecan immunoreaction (red). (a) Hyaluronan is ubiquitously distributed and concentrated around the neurons. (b) Tenascin-R is equally present in DCN of wt and pcd mice and mainly encloses the neurons. Scale bar: 20 μm.

Figure 7

Reactive astrogliosis in the DCN of pcd mouse brain. (a) DCN of wt mouse brain is characterized by the virtual absence of reactive astrocytes. In pcd, the degeneration process is accompanied by a strong astrocytic activation; the DCN seem to be filled with astrocytes. Scale bar: overview 100 μm, detail 20 μm. (b) Western blot analyses confirm the immunocytochemical data. In pcd tissue, the GFAP protein level is more than 2-fold increased (p < 0.01). Data are given as mean ± SEM.