In order to investigate microglial maturation and function in the R6/2 mouse model of HD, we performed IBA1 immunostaining followed by densitometric and morphometric analysis in the striatum of 3-week, 10-week, and 13-week old animals (Fig. 1A-F). Densitometric studies included measurement of density (cells/mm2) as well as the nearest neighbor distance (NND, nearest cell to every other cell), and spacing index (compiling the NND and density) between microglia. R6/2 animals displayed an age-dependent decrease in microglial density but had higher microglial density compared to control animals at all ages investigated (Fig. 1G). The NND of both R6/2 and control animals increased with age, and WT animals had significantly larger NND compared with R6/2 animals at both 10 and 13 weeks of age (Fig. 1H). However, the spacing index did not differ significantly between groups at any age investigated (Fig. 1I).
In addition to these changes in densitometric maturation, microglia in both genotypes exhibited an age-dependent morphological maturation. Microglial cell body area was stable across ages in WT animals, while cell body area decreased with age in R6/2 animals (Fig. 1J). Further morphological investigation found microglia in control animals were characterized by an increase in their arborization area (Fig. 1K) while R6/2 animals’ microglial arborization area was already elevated at 3 weeks of age. Both genotypes displayed an age-dependent decrease in their morphological index (the ratio of cell body area divided by the arborization area) (Fig. 1L). In fact, R6/2 mouse microglia already displayed a significantly reduced morphological index compared with WT microglia at 3 weeks of age (Fig. 1L). Together, these data indicate that microglial number remained elevated in the striatum of R6/2 mice compared with WT controls and their morphology was significantly changed as early as 3 weeks of age. This age corresponds to a time point preceding any known inflammatory signaling or neurodegeneration in this model [45].
Microglia are known to play a major role in synaptic removal and plasticity, notably via phagocytosis in healthy and disease states [20, 46]. To determine the functional implications of the decreased morphological indices we observed in R6/2 mice, we investigated the phagocytic index in microglia by measuring levels of CD68, a transmembrane protein highly expressed by microglia and macrophages that is enriched in their phagolysosomal compartments [47]. IBA1 + microglia in both R6/2 and WT mice displayed abundant CD68 + puncta (Fig. 2A-F). We quantified the number of CD68 + puncta per microglial cell body (phagocytic index) in the striatum of 3, 10, and 13 week old animals. WT microglia decreased their phagocytic index as the animals matured (Fig. 2G). However, microglia in R6/2 mice had elevated levels of CD68 + puncta at all ages investigated, and did not display a decrease in phagocytosis over time with maturation (Fig. 2G). Microglia in R6/2 mice may be performing aberrant excess phagocytosis as often seen in neurodegenerative disease conditions [20], or their phagolysosomal system may be overwhelmed and not processing phagocytosed material properly, causing the cells to become overloaded with phagocytic debris.
In order to further investigate the types of phagocytic cargo associated with the CD68 + puncta of both WT and R6/2 animals, we performed immunoEM in the dorsomedial striatum of 3-week versus 10-week old WT and R6/2 animals. We focused on the dorsomedial region of the striatum, which is one of the earliest regions affected by HD pathology [12]. Microglial cell bodies in both WT and R6/2 model mice displayed characteristic ultrastructural features, including a cheetah-like heterochromatin pattern in their ovoid nuclei surrounded by a narrow band of IBA1 + cytoplasm (Fig. 3A-D). Microglial cell bodies were often found directly juxtaposed with neuronal elements such as cell bodies and dendrites, as well as synaptic elements, including axon terminals and dendritic spines (Fig. 3A-D). While microglia in both 3-week and 10-week old WT mice rarely showed processes contiguous with their cell body in ultrathin sections, microglia in 10-week old R6/2 mice often had long, ramified processes connected to their soma in ultrathin sections (Fig. 3D). Microglial cell bodies across all experimental groups also displayed characteristic long stretches of ER and occasional lipidic inclusions, lipofuscin granules as well as lysosomes, all common in microglial cell bodies. Microglia in R6/2 animals were nevertheless more likely to contain phagosomes, and had more phagosomes per cell body than WT animals at 3 weeks of age (Fig. 3E, F) consistent with our light microscopy observations. Furthermore, these phagosomes often held partially digested material, and microglia in R6/2 animals contained more partially digested inclusions than WT animals at 10 weeks of age (Fig. 3G), indicative of a possible impairment in phagolysosomal maturation.
Additionally, microglia in R6/2 striatum displayed increased frequency of dilated ER, as a much higher proportion of R6/2 cell bodies contained dilated ER than WT cell bodies (Fig. 3H). Dilated ER is a well-known marker of cellular stress that has been described using EM in numerous contexts of neurodegeneration including amyotrophic lateral sclerosis and Alzheimer’s disease pathology [43, 48, 49]. We also identified two microglia in 3-week old R6/2 mice with reduced IBA1 immunoreactivity and a condensed cytoplasm as well as nucleoplasm (Supplemental Fig. 1). These cells, which further displayed dilated ER, are reminiscent of the dark microglia seen in ageing and other neurodegenerative disease models [43, 48]. Their long processes formed acute angles and interacted with synaptic structures as well as the vasculature, all characteristic of dark microglia. However, we did not identify cells with the hallmark loss of nuclear chromatin pattern typically associated with dark microglia [48].
In addition to microglial cell body ultrastructure, we utilized immunoEM to glean information into microglial processes activities in the striatum. Microglial processes are IBA1+, allowing them to be investigated at ultra-high resolution in EM (Fig. 4A-D). Microglial processes in ultrathin sections are not usually contiguous with their cell body, and form a variety of shapes and sizes as they move throughout the neuropil and survey their environment. Similarly to cell bodies, processes often contained phagocytosed material in our ultrastructural analyses (Fig. 4A, B) and made frequent direct contacts with extracellular degraded elements or debris (referred to as “extracellular degradation,” Fig. 4C, D). Microglial processes observed in WT and R6/2 mice similarly displayed age-dependent decreases in their perimeter (Fig. 4E), suggesting that microglia are taking on a more surveillant morphology as indicated by our light microscopy studies. Interestingly, R6/2 microglial processes had larger perimeters than WT processes at 3 weeks of age, but their process perimeter was significantly reduced with age and became smaller than WT processes at 10 weeks of age (Fig. 4E). Microglial processes in both WT and R6/2 striatum reduced their areas with age, but again, microglial processes in R6/2 mice became significantly smaller in area than those of WT mice by 10 weeks of age (Fig. 4F). Microglial processes in R6/2 mice were also more likely to perform extracellular degradation than processes in WT animals, although this phenomenon also decreased with age (Fig. 4G). Microglial processes in both WT and R6/2 further displayed an age-dependent decrease in phagocytosed material (Fig. 4H). Interestingly, this is in contrast with our findings for cell bodies indicating there may be a shift in phagocytic cargo trafficking between processes and cell bodies with maturation.
One of the most striking changes in microglial process ultrastructure in R6/2 animals were at the level of their interactions with synaptic structures. We found many incidences of microglial processes interacting with synaptic structures, including axonal terminals, dendritic spines, and direct contact with excitatory synaptic clefts, defined as the site of contact between a presynaptic axon terminal and a postsynaptic dendritic spine (Fig. 5A-D). Microglia in WT dorsomedial striatum consistently interacted with the same number of excitatory synapses onto dendritic spines as the animals matured (Fig. 5E, F). Microglia in R6/2 dorsomedial striatum were significantly more likely to interact with synaptic clefts than microglia in WT striatum at 3 weeks of age, but significantly less likely to interact with synaptic clefts at 10 weeks of age, and displayed an age-related decrease in synaptic interactions (Fig. 5E, F). Microglial processes in R6/2 mice were also significantly less likely to interact with either presynaptic axon terminals or postsynaptic dendritic spines at 10 weeks of age (Fig. 5G, H), displaying an age-dependent decrease in their synaptic interactions.
Microglia-synapse interactions may have an impact on synaptic density and could be impacted by a large number of factors including synaptic number. In order to investigate these changes, we shifted our focus to investigate dendrites in the striatum of 3-week old and 10-week old animals. We performed FIB-SEM experiments to image 150–250µm3 regions outside of striosomes and devoid of blood vessels, myelinated axons and cell bodies within the dorsomedial striatum of WT and R6/2 mice (Fig. 6A). Afterwards, we segmented randomly selected dendrites of lengths varying between 4.5 and 10 µm, dependent upon their orientation through the imaged volume (Fig. 6B-E). We correlated the segmented dendrite with the original images to count the number of excitatory synapses and determine the number of synapses per micron of dendrite (Fig. 6F). This analysis revealed that synaptic density was not affected by genotype in 3-week old animals. However, synaptic density increased between 3 and 10 weeks in WT animals, without concomitant increase of synaptic density in R6/2 animals (Fig. 6F). This data supports other studies finding impairment of corticostriatal communication in 10–12 weeks old R6/2 mice [50], and could be related either to synaptic loss or a defect of synapse formation or maturation.
Because we segmented dendrites and counted synaptic density at nanoscale resolution, we were also able to discriminate between non-synaptic (spines which were not juxtaposed with an axon terminal containing synaptic vesicles) and synaptic spines (spines directly juxtaposed with an axon terminal containing synaptic vesicles). We were also able to count en face synapses formed directly onto the dendrite trunk (Fig. 6A, inset). Although there was no change in synaptic number at 3 weeks of age (Fig. 6F), synapses in 3-week old R6/2 animals were significantly less likely to contact spines, and did not increase synapses onto dendritic spines with age (Fig. 6G). Synapses in R6/2 animals were more likely to directly target the dendritic trunk itself compared to WT animals at 3 weeks of age (Fig. 6H). These data indicate that, while total synaptic density may not be affected in 3-week old R6/2 animals, there is already a difference in the type of synaptic input made onto the medium-sized spiny neurons in the dorsomedial striatum. This shift in synaptic location (higher proportion of en face synapses to spine synapses) persisted in 10-week old R6/2 animals (Fig. 6I).