Loss of astrocytic domain organization in the epileptic brain

Abstract

Gliosis is a pathological hallmark of posttraumatic epileptic foci, but little is known about these reactive astrocytes beyond their high glial fibrillary acidic protein (GFAP) expression. Using diolistic labeling, we show that cortical astrocytes lost their nonoverlapping domain organization in three mouse models of epilepsy: posttraumatic injury, genetic susceptibility, and systemic kainate exposure. Neighboring astrocytes in epileptic mice showed a 10-fold increase in overlap of processes. Concurrently, spine density was increased on dendrites of excitatory neurons. Suppression of seizures by the common antiepileptic, valproate, reduced the overlap of astrocytic processes. Astrocytic domain organization was also preserved in APP transgenic mice expressing a mutant variant of human amyloid precursor protein despite a marked upregulation of GFAP. Our data suggest that loss of astrocytic domains was not universally associated with gliosis, but restricted to seizure pathologies. Reorganization of astrocytes may, in concert with dendritic sprouting and new synapse formation, form the structural basis for recurrent excitation in the epileptic brain.

Figure 1.

Cortical astrocytes are organized in nonoverlapping domains. A, Schematic of the procedure used for diolistic labeling of lightly fixed brain slices with DiI (green) and DiD (red). B, Representative labeling of cortical astrocytes with DiI and one with DiD. DiI and DiD distribute evenly in the plasma membrane of single astrocytes and outline their bushy structure in a control mouse. Scale bar (SB), 20 μm. C, Immunostaining against GFAP in cortex. Only the cell body and major processes of cortical astrocytes are GFAP positive. SB, 20 μm. D, Morphometric analysis of cortical astrocytes based on diolistic labeling and GFAP immunostaining (mean ± SEM). E, High-power image of box in B. SB, 10 μm. F, Area of overlap between two neighboring astrocytes is delineated in gray. The red line indicates the border of the domain of the DiD-labeled astrocyte (red), whereas the green line indicates the border of the domain of the DiI-labeled astrocyte (green). G, Yellow lines highlight the processes that extend into their neighbor's domain.

Figure 2.

Structure and organization of reactive astrocytes 1 week after injury. A, Cortical slice immunolabeled for GFAP 1 week after iron injection. A marked upregulation of GFAP (white) is evident around the lesion. The yellow asterisk indicates the lesion site. B, High-power image of the border of the injury site. Astrocytes directly adjacent to the lesion extend long processes oriented toward the lesion site (palisading astrocytes; 0–200 μm). Astrocytes located at a greater distance from the lesion (200–1000 μm) also exhibited reactive changes, but their processes were not oriented toward the injury site (hypertrophic astrocytes). White, GFAP; red, Map2; blue, Sytox. Scale bar (SB), 50 μm. Top right, Palisading astrocytes next to area of injury. SB, 10 μm. Bottom right, Hypertrophic astrocytes. SB, 10 μm. C, Morphometric analysis of diameter, number of main processes, and longest and thickest process based on GFAP (control, n = 65 cells, 6 mice; 1 week after injury, n = 65 cells, 7 mice; mean ± SEM; *p < 0.001). D, Representative EEG recording 2 d after injury.

Figure 3.

Loss of astrocytic domain organization 1 week after injury. A, Diolistic labeling of a cortical slice from an adult mouse expressing GFP under the hGFAP promoter 1 week after iron injection. Two adjacent GFP-positive astrocytes are labeled with DiI and DiD. Blue, DAPI; white, GFP; green, DiI; red, DiD. Scale bar (SB), 20 μm. B, Diolistic labeling of palisading reactive astrocytes demonstrating loss of the domain organization. Green, DiI; red, DiD; blue, DAPI. SB, 20 μm. C, Quantification of the volume of reactive astrocytes 1 week postinjury (PI) compared with astrocytes from control mice. The volume represents the total area of the astrocytic soma and processes labeled with DiI or DiD in serial sections (n = 20 cells, 7 mice; mean ± SEM; *p < 0.001). D, Quantification of the area of overlap between reactive astrocytes 1 week after injury compared with astrocytes in control mice over 10 μm stacks, and summation of the total length of processes that enter an adjacent cell's domain 1 week after injury compared with control over 10 μm stacks. The area of overlap and the total length of overlapping processes increased 15-fold in epileptic mice (control, n = 30 cells, 7 mice; 1 week after injury, n = 34 cells, 10 mice; mean ± SEM; *p < 0.001). E–H, High-power image of blue box in A of reactive astrocytes 1 week after injury. F, Area of overlap is delineated in gray, the red line is the border of the domain of the red cell, and the green line is the border of the domain of the green cell. G, H, Yellow lines indicate the processes of the cell that enter the domain of the adjacent astrocyte. I–L, Similar analysis in control mouse in which astrocytes are organized in essentially nonoverlapping domains.

Figure 4.

Structure and organization of reactive astrocytes 2 and 6 months after injury. A, Site of injury 6 months after injection of iron. The center of the lesion (yellow asterisk) is surrounded by palisading astrocytes and, at a greater distance, by hypertrophic astrocytes. White, GFAP; red, Map2; blue, Sytox. Scale bar (SB), 100 μm. B, Morphometric analysis of diameter, number of main processes, and longest and thickest process based on GFAP (2 months after iron injection, n = 50 cells, 3 mice; 6 months, n = 65 cells, 5 mice; mean ± SEM; *p < 0.001). The diameter, maximum length of process, and maximum thickness of GFAP-positive processes were significantly increased at both 2 and 6 months after injury compared with control. C, Diolistic labeling of cortical astrocytes near the area of injury, 6 months after injection. SB, 20 μm. D–F, High-power image of yellow box in C. E, Gray area delineates area of overlap between the adjacent cells. F, Yellow lines represent overlapping processes into the adjacent cell's domain. G, H, Quantification of overlapping area and summation of lengths of overlapping processes for 1 week and 2 and 6 months after injury. Area of overlap at 1 week and 2 and 6 months after injection was significantly increased compared with control (2 months, n = 18 cells, 3 mice; 6 months, n = 18 cells, 6 mice; mean ± SEM; *p < 0.001). I, Representative EEG trace recorded 6 months after iron injection. J, Quantification of the volume of reactive astrocytes 1 week and 2 and 6 months after iron injection compared with control (2 months, n = 7 cells, 3 mice; 6 months, n = 7 cells, 4 mice; mean ± SEM; *p < 0.05).

Figure 5.

Valproate reduces postinjury seizure activity and overlap of astrocytic processes. A, B, Comparison of the injury border in a mouse that did not respond to valproate (no reduction in seizure activity; A) and a responder (significant reduction in seizure activity; B) 1 week after iron injection. The yellow asterisk indicates the lesion site. White, GFAP; red, Map2; blue, DAPI. Scale bars (SBs), 100 μm. Astrocytes exhibited less reactive changes in mice that responded to valproate. C, Diolistic labeling of two hypertrophic astrocytes near the lesion of an animal that responded to valproate treatment. The astrocytes exhibited less overlap of their processes than in mice that did not receive or did not respond to valproate. The yellow box indicates the border between two adjacent cells. SB, 10 μm. D, Morphometric analysis based on GFAP for animals 1 week after injury with and without valproate treatment (responders). There is a significant decrease in the length of longest process, diameter, and diameter of thickest process in animals successfully treated with valproate compared with those without (n = 65 cells, 5 mice; mean ± SEM; *p < 0.01). E, The volume of reactive astrocytes was significantly reduced in animals that received and responded to valproate 1 week after injury (mean ± SEM responders, n = 18, 4 mice; *p < 0.005; nonresponders, n = 11, 4 mice; p = 0.56). F, Quantification of the domain organization in animals treated with valproate, both responders and those in which the drug had minimal effect. There was a significant decrease in both the area of overlap and summation of lengths of overlapping processes in animals 1 week after iron injection in which valproate treatment reduced seizure activity compared with those without treatment (responders, n = 54 cells, 5 mice; mean ± SEM; *p < 0.001). There was no significant difference in the domain organization in animals treated with valproate that continued to have seizures (nonresponders), compared with mice that did not receive antiepileptic treatment after iron injection (nonresponders, n = 38 cells, 4 mice; mean ± SEM; p = 0.8 for area of overlap and p = 0.9 for summation of overlapping processes). G, Representative EEG recordings from a mouse that responded to valproate and exhibited a reduced number of EEG abnormalities.

Figure 6.

Loss of domain organization among hypertrophic astrocytes in other epilepsy models. A, Low- and high-power images of sensory-motor cortex 6 months after exposure to kainic acid. A clear upregulation of GFAP throughout the cortical layers is evident. Scale bars (SBs): left, 100 μm; right, 10 μm. B, Low- and high-power images of the sensory-motor cortex of the SWXL-4 mouse that also exhibited an increase in GFAP expression. SBs: left, 100 μm; right, 10 μm. C, D, Morphometric analysis in mice exposed to kainate (KA) and the SWXL-4 mice based on GFAP (KA, n = 65 cells, 3 mice; SWXL-4, n = 85 cells, 6 mice; mean ± SEM; *p < 0.001). E, Representative EEG recording of a mouse exposed to kainate 6 months earlier. F, EEG recording of SWXL-4 mice 10 months of age. G, Diolistic labeling revealed extensive overlap of processes between two reactive astrocytes in the cortex of a kainate-treated mouse. SB, 10 μm. Inset, Overlapping processes of two astrocytes in CA3. H, I, High-power images of blue box in G demonstrating overlapping area and processes of kainate-treated cortical astrocytes. J, Two adjacent astrocytes in SWXL-4 mice with overlapping domains. SB, 10 μm. K, L, High-power images of blue box in J demonstrating overlapping area and processes in SWXL-4 cortical astrocytes. M, Quantification of the change in volume in kainate-treated and SWXL-4 mice compared with control. There is a significant increase in volume of reactive astrocytes in both the kainate-treated and SWXL-4 mice compared with control (KA, n = 10, 3 mice; SWXL-4, n = 16, 6 mice; mean ± SEM; *p < 0.05). N, O, Quantification of loss of domain organization in the kainate-treated and SWXL-4 cortical astrocytes. There is a significant increase in the amount of overlap between astrocyte domains in both kainate-treated and SWXL-4 mice compared with control (KA, n = 16 cells, 4 mice; SWXL-4, n = 44 cells, 7 mice; mean ± SEM; *p < 0.001).

Figure 7.

Cortical astrocytes in seizure-free Tg2576 transgenic mice are reactive but do not lose their domain organization. A, Control cortex with low GFAP expression throughout cortex. Scale bar (SB), 100 μm. B, Cortex of a Tg2576 mouse with upregulation of GFAP throughout all cortical layers. SB, 100 μm. C, Higher-power image of cortical astrocytes in Tg2576 mouse. SB, 10 μm. D, Morphometric analysis based on GFAP. Astrocytes in the Tg2576 mice exhibited only a significant increase in the maximum thickness of GFAP-positive processes (Tg2576, n = 65, 3 mice; mean ± SEM; *p < 0.001). E, Representative EEG recording of Tg2576 mice. F, Diolistic labeling of cortical astrocytes in Tg2576 mouse. SB, 10 μm. G–I, High-power image of blue box in F showing limited overlap between processes of two adjacent astrocytes. J–L, Adjacent control astrocytes showing limited overlap and extension of processes into the adjacent cell's domains. M, Quantification of the volume of reactive astrocytes in Tg2576 compared with control. There is no significant increase in the volume of reactive astrocytes in the Tg2576 model (n = 12, 3 mice; mean ± SEM; p = 0.45). N, Quantification of amount of overlap between adjacent cells in Tg2576 and control mice (Tg2576, n = 30, 3 mice; mean ± SEM; there is no significant increase in amount of overlap compared with control; area, p = 0.5; overlapping process, p = 0.8).

Figure 8.

Changes in dendritic structure parallel reactive changes in astrocytes. A, Low-power image of a Thy1-YFP mouse 1 week after iron injection in the hemisphere contralateral to the injury site. White, GFAP; green, YFP. Scale bar (SB), 100 μm. B, Low-power image of a Thy1-YFP mouse 1 week after iron injection near the injury site; yellow lines delineate area of palisading astrocytes. SB, 100 μm. C, High-power image of normal-appearing apical dendrites of pyramidal cells in the hemisphere contralateral to the injection site. SB, 20 μm. Inset, High-power image of dendrite and spines. SB, 2 μm. D, High-power image of zone with palisading astrocytes and thin dendrites lacking spines. SB, 20 μm. Inset, High-power image of dendrite and spines within 200 μm of the injection site. SB, 2 μm. E, High-power image of area more distant to the lesion with hypertrophic astrocytes and thick dendrites exhibiting a significant increase in number of spines. SB, 20 μm. Inset, High-power image of dendrite and spines >200 μm from the injection site. SB, 2 μm. Yellow arrowheads (C–E) indicate dendrites seen in their respective high-power insets. F, Morphometric analysis of apical dendrites and controls (control, n = 40 dendrites, 3 mice; iron injection, n = 40 dendrites, 5 mice; mean ± SEM; *p < 0.001). G, Quantification of spine density compared with control (control, n = 24 dendrites; <200 μm, n = 28 dendrites; >200 μm, n = 31 dendrites; contralateral, n = 30 dendrites, 5 mice; mean ± SEM; *p < 0.005). H, Representative EEG recording from a Thy1-YFP mouse during the first week after iron injection. I, Apical dendrite in control, seizure-free Thy1-YFP cortex. SB, 2 μm. J, From within 200 μm of the iron injection, demonstrating thinning of the dendrite and loss of spines. SB, 2 μm. K, More than 200 μm from the iron injection, showing hypertrophy and increase in spine density. SB, 2 μm. L, From the contralateral hemisphere. SB, 2 μm.