Characterization of Vimentin-Immunoreactive Astrocytes in the Human Brain

Abstract

Astrocytes are commonly identified by their expression of the intermediate filament protein glial fibrillary acidic protein (GFAP). GFAP-immunoreactive (GFAP-IR) astrocytes exhibit regional heterogeneity in density and morphology in the mouse brain as well as morphological diversity in the human cortex. However, regional variations in astrocyte distribution and morphology remain to be assessed comprehensively. This was the overarching objective of this postmortem study, which mainly exploited the immunolabeling of vimentin (VIM), an intermediate filament protein expressed by astrocytes and endothelial cells which presents the advantage of more extensively labeling cell structures. We compared the densities of vimentin-immunoreactive (VIM-IR) and GFAP-IR astrocytes in various brain regions (prefrontal and primary visual cortex, caudate nucleus, mediodorsal thalamus) from male individuals having died suddenly in the absence of neurological or psychiatric conditions. The morphometric properties of VIM-IR in these brain regions were also assessed. We found that VIM-IR astrocytes generally express the canonical astrocytic markers Aldh1L1 and GFAP but that VIM-IR astrocytes are less abundant than GFAP-IR astrocytes in all human brain regions, particularly in the thalamus, where VIM-IR cells were nearly absent. About 20% of all VIM-IR astrocytes presented a twin cell morphology, a phenomenon rarely observed for GFAP-IR astrocytes. Furthermore VIM-IR astrocytes in the striatum were often seen to extend numerous parallel processes which seemed to give rise to large VIM-IR fiber bundles projecting over long distances. Moreover, morphometric analyses revealed that VIM-IR astrocytes were more complex than their mouse counterparts in functionally homologous brain regions, as has been previously reported for GFAP-IR astrocytes. Lastly, the density of GFAP-IR astrocytes in gray and white matter were inversely correlated with vascular density, but for VIM-IR astrocytes this was only the case in gray matter, suggesting that gliovascular interactions may especially influence the regional heterogeneity of GFAP-IR astrocytes. Taken together, these findings reveal special features displayed uniquely by human VIM-IR astrocytes and illustrate that astrocytes display important region- and marker-specific differences in the healthy human brain.

Keywords: GFAP; astrocyte; human; morphology; mouse; stereology; vimentin.

FIGURE 1

Vimentin immunohistochemistry strongly labels blood vessels and astrocytes in healthy adult human brain. (A) Example of a VIM-IR astrocyte in the caudate nucleus juxtaposed to and contacting (arrowheads) VIM-IR blood vessels. (B) VIM-IR astrocytes in the prefrontal cortex gray matter outnumbered those in white matter and are clustered along the gray/white matter boundary (dashed line). (C) Although distributed in a similar pattern along the gray/white matter boundary (dashed line), fewer VIM-IR astrocytes were observed in the primary visual cortex compared to the prefrontal cortex. (D) VIM-IR astrocytes were most abundant in the caudate nucleus. (E) VIM-IR astrocytes were almost completely absent from the mediodorsal thalamus. Scale bars = 50 μm.

FIGURE 2

VIM-IR cells express other astrocytic markers in adult human and mouse brain. (A) Representative example in the human prefrontal cortex of a VIM-IR astrocyte revealed using immunofluorescence (white), containing Aldh1L1 mRNA (green) and GFAP mRNA (red), revealed using fluorescent in situ hybridization (FISH), within the nucleus (DAPI-stained, blue). (B) By combining FISH with immunofluorescence, Aldh1L1 mRNA was observed in the majority of DAPI-stained nuclei of VIM-IR astrocytes in the PFC and the caudate nucleus (n = 2). (C) By combining FISH with immunofluorescence, GFAP mRNA was observed in the majority of DAPI-stained nuclei of VIM-IR astrocytes in the PFC and the caudate nucleus (n = 2). (D) Using coimmunofluorescence, the majority of astrocytes in the PFC and the caudate nucleus coexpressed VIM protein and GFAP protein (n = 2). (E) In transgenic adult mice, the fluorescence of cre reporter proteins in astrocytes expressing Aldh1L1 was observed in many VIM-IR astrocytes in the caudate-putamen (CPu) (n = 3). (F) VIM protein expression (red) was located within thick processes of transgenically Aldh1L1-labeled (green) astrocytes in the mouse CPu. Scale bars = 25 μm.

FIGURE 3

VIM immunohistochemistry labels astrocytic subtypes previously described for GFAP-IR astrocytes in human neocortex. (A) Fine unbranched fibers (black arrowheads) emerging from darkly labeled VIM-IR interlaminar astrocytes (white arrowheads) and projecting to layer III. (B) VIM-IR astrocytes in the prefrontal cortex gray matter have a non-overlapping domain organization. (C) An example of a VIM-IR astrocyte in deep cortical gray matter (prefrontal cortex) with the attributes of a varicose projection astrocyte. The inset shows a magnified view of regularly spaced varicosities (arrows) on VIM-IR varicose projection astrocyte processes. (D) In cortical white matter and thalamus, VIM-IR astrocytes (white arrowhead) extended rather straight and mostly unbranched processes, as is typical for GFAP-IR fibrous astrocytes. VIM-IR microglia (black arrowheads) were also observed in cortical white matter. In general, these cells were weekly stained and easily distinguishable from VIM-IR astrocytes. Scale bars = 50 μm.

FIGURE 4

VIM-IR astrocytes display diverse morphological features. (A) Two adjacent VIM-IR astrocyte cell somas (arrowheads) displaying a twin cell morphology were commonly observed in all human brain regions examined. (B) In the caudate nucleus, we observed long bundles of parallel VIM-IR fibers (arrowheads) of no immediately discernable cell origin or target, but which often received contacts from neighboring VIM-IR astrocytes. (C) Some VIM-IR astrocytes extended parallel projections (arrowheads) which may be the origin of the fiber bundles illustrated in (C) and which appear (black outline) from outside of the z-plane of the section, suggesting that VIM-IR astrocytes contact distal targets. Scale bars = 50 μm.

FIGURE 5

Regional stereological estimates of VIM-IR and GFAP-IR astrocyte density in the adult human brain. (A) There were significantly more VIM-IR astrocytes in the caudate nucleus than in all other brain regions examined, including the thalamus where VIM-IR astrocytes were mostly absent. (B) All human brain regions included in this study were more densely populated by GFAP-IR astrocytes than by VIM-IR astrocytes. There were also significantly more GFAP-IR astrocytes in either subcortical region than in neocortical areas. **p ≤ 0.01 (n = 8; Matched One-Way ANOVA).

FIGURE 6

Regional heterogeneity in human VIM-IR astrocyte morphometry. (A) Representative VIM-IR astrocytes from prefrontal cortex gray matter (left) and white matter (right). (B–H) Branched Structure Analysis (BSA) of VIM-IR astrocytes revealed morphometric differences across markers and regions. BSA measurements for VIM-IR astrocytes were generally similar for cortical gray matter and caudate nucleus, and for cortical white matter and mediodorsal thalamus. Scale bars = 50 μm. **p ≤ 0.01, ****p ≤ 0.0001 (n = 5; Friedman Test).

FIGURE 7

Lack of regional heterogeneity in mouse VIM-IR astrocyte morphometry. (A) Representative VIM-IR astrocyte from the mouse frontal association cortex gray matter. (B–H) In the mouse brain, comparing VIM-IR astrocytes between regions revealed no significant difference (p > 0.05) for any of the BSA. Scale bars = 10 μm (n = 5; Friedman Test).

FIGURE 8

VIM-IR astrocyte morphometry reveals considerably more complex cerebral astrocytes in humans than in mouse. (A) Representative VIM-IR astrocytes from the human (left) and the mouse (right) primary visual cortex. (B–H) Human VIM-IR astrocytes were distinguishable from mouse VIM-IR astrocytes for all regions and BSA measures, except for process number in the caudate nucleus. Scale bars = 10 μm. n.s. p > 0.05, ***p ≤ 0.001 (n = 5; Mann–Whitney U test).

FIGURE 9

Regional differences in vascularization correlate with regional differences in astrocyte density. (A,B) The area occupied by CD31-IR and VIM-IR blood vessels was significantly greater in cortical than in subcortical regions. (C) There were no significant differences in the area occupied by CD31-IR blood vessels between mouse brain regions, but this coverage was substantially lower that that measured in human samples (A,B). (D) Human VIM-IR and GFAP-IR astrocyte density negatively correlated with CD31-IR vascularization across regions. One value from the caudate nucleus was excluded as an outlier for both regressions, and both the thalamus and cortical white matter values were excluded from the VIM-IR density regression as VIM-IR cells were mostly absent from these regions. Scale bars = 50 μm. n.s. p > 0.05, ****p ≤ 0.0001 (Human: n = 8; Mouse: n = 5; Matched One-Way ANOVAs).