Giant mossy fiber bouton area and complexity are increased one and three months after status epilepticus
Giant bouton area and complexity were assessed in one and three month control mice, and mice collected one (1M SE) and three (3M SE) months after pilocarpine-induced status epilepticus. Analysis of the one month and three month control groups revealed that they were statistically equivalent for all parameters examined in the present study (t-test), so the two groups were pooled, and from here on are referred to as controls. A total of 222 giant boutons from fourteen control mice, 209 giant boutons from eight 1M SE mice and 135 boutons from seven 3M SE mice were examined. Mossy fiber bouton area was significantly increased in both groups of epileptic animals relative to controls (; P<0.001, ANOVA with Tukey’s post test). The frequency of giant boutons connected to “satellite” boutons by short axonal processes was also increased at both post-status time points relative to controls (; P<0.001, Kruskal-Wallis rank sum test with Dunn’s post test).
Figure 2 Dentate granule cell giant mossy fiber bouton (MFB) area is significantly increased both one and three months after status epilepticus (SE) relative to controls. The percentage of giant boutons with satellites was also increased at both time points after (more ...)
Thorny excrescence labeling in Thy1-GFP-expressing mice
GFP-labeling of CA3b pyramidal cell thorny excrescences in the Thy1-GFP line was robust, allowing for easy quantification of their numbers and complexity. A total of 50 cells were imaged from control animals, 33 cells from 1M SE animals and 26 cells from 3M SE animals. All CA3 pyramidal cells examined conformed to the morphology of “classical” CA3 pyramidal cells (Lorente de No, 1934
; Amaral, 1978
; Frotscher et al., 1988
; Ishizuka et al., 1990
; Li et al., 1994
; Seress and Ribak, 1995
; Buckmaster and Amaral, 2001
). Cell bodies were located in the pyramidal cell layer and possessed one to three prominent apical dendrites (although a single cell with four apical dendrites was observed in a control). Apical dendrites projected through stratum lucidum, stratum radiatum and on into stratum lacunosum-moleculare. Basal dendrites projected into stratum oriens. These findings suggest that pyramidal cells labeled in the Thy1-GFP line are representative of the entire population; a conclusion consistent with previous work demonstrating that labeled dentate granule cells in this line are indistinguishable from granule cells labeled with other approaches (Vuksic et al., 2008
The present study focused on segments of CA3 pyramidal cell apical dendrites contained within stratum lucidum, the projection field of granule cell mossy fiber axons and the principal localization of pyramidal cell thorny excrescences. Stratum lucidum is relatively thin − 50 μm or so – and correspondingly, only a small portion of a CA3 pyramidal cells’ apical dendritic tree is localized to this region. Pyramidal cell apical dendritic branches within this region were typically first, second or third order (). Fourth and fifth order branches were occasionally observed, but were not further analyzed due to their low incidence. Total dendritic length examined, broken down by branch order, was 758 μm (1st order), 1775 μm (2nd order) and 1800 μm (3rd order) for control neurons. For animals examined one month after status, 616 μm, 1441 μm and 1143 μm were scored for first, second and third order branches, respectively. Finally, 512 μm, 1276 μm and 884 μm were scored from animals collected three months after status.
Thorny excrescence density varies by branch order
Intriguingly, although the localization of first, second and third order branches to stratum lucidum renders them as potential targets for mossy fiber innervation, thorns were not distributed equally among segments. Specifically, comparisons of thorn density among branch orders for control animals revealed that density increased significantly with higher branch order (1st order, 0.370±0.237 thorns/μm; 2nd order, 1.000±0.394; 3rd order, 0.778±0.167; P=0.005 for 2nd and 3rd vs. 1st, Kruskal-Wallis rank sum test with Dunn’s post test). In light of these findings, comparisons between control and epileptic animals were made among equivalent branch orders.
Thorny excrescence density is transiently reduced following status epilepticus
One month after pilocarpine-induced status epilepticus, the density of thorns along 3rd order CA3 pyramidal cell dendritic segments was significantly reduced relative to pyramidal cells from control animals (; P=0.043, ANOVA with Tukey’s post test). Similar trends towards reduced thorn density were observed for 1st and 2nd order dendritic segments, although the effect did not reach significance (&). Notably, the reduction in density along 3rd order segments was transient, and three months after status epilepticus thorn densities were statistically indistinguishable from controls ().
Figure 3 The density of thorns was significantly decreased along 3rd order dendritic segments one month after status epilepticus (SE) relative to control animals and animals examined three months after status. Similar reductions were observed for first and second (more ...)
Figure 4 Neuronal reconstructions showing apical dendritic segments from control animals and animals examined one (1M SE) and three (3M SE) months after status epilepticus (SE). Control animals, and animals examined three months after status occasionally exhibited (more ...)
Analysis of pyramidal cell variability in epileptic animals revealed that the reduction in thorn density evident one month after status epilepticus was due largely to the disappearance of pyramidal cells with high densities. While 23.7% (9 of 38) of pyramidal cells from control animals exhibited densities greater than 1 thorn/μm on 3rd order branches, none of the cells examined one month after status exhibited such high densities on 3rd order branches (0 of 20). Three months after status, cells with dense accumulations of thorns reappeared, making up 27.7% (5 of 18) of the cell population examined.
Altered thorn distribution three months after status epilepticus
Intriguingly, although thorn density was similar between control pyramidal cells and cells examined three months after status, casual observation suggested that the distribution of thorns was altered after status – particularly for cells with higher thorn densities. Specifically, while thorns along control dendrites tended to be highly clustered, thorns in animals examined three months after status were more evenly distributed (). Simple density measurements, however, failed to capture this phenomenon (e.g. ten clustered thorns produces the same overall density as ten evenly distributed thorns for a given length of dendrite), so two alternate strategies were developed to analyze the data. Analyses focused on pyramidal cells with thorn densities greater than 1.0/micron – seven from control animals and seven from 3M SE animals.
Figure 5 A: Neurolucida reconstructions showing thorn distributions along pyramidal cells from control animals and animals examined three months after status epilepticus (3M SE). Note the tendency for individual thorns to form clusters along the dendrites of cells (more ...)
The first approach used to examine thorn distributions patterns involved calculating the mean distance between thorns along individual dendrites. The mean value for the distance between adjacent spines (nearest neighbors) was 0.806μm (95%CI 0.793 – 0.819) for cells from control animals and 0.880 μm (95%CI 0.860 – 0.900) for cells from animals examined three months after status epilepticus. The difference was significant (P<0.001), confirming qualitative impressions that thorn distributions differed.
To explore whether this difference might have biological significance, a second analytical approach was developed to predict that number of giant mossy fiber boutons that would be required to innervate the thorns in each data set. Briefly, the three-dimensional coordinates for each thorn were used to determine the minimum number of spheres with a radius r required to cover every thorn along a dendritic tree one time. This value is defined as the minimum cover set, and was developed as a new means for quantifying thorn distribution to better assess the unique innervation patterns of CA3 pyramidal cells by granule cell giant boutons. Basically, the minimum cover set can be viewed as a measure of the smallest number of hypothetical giant mossy fiber boutons (hMFB) of a given radius r that would be required to cover every thorn along a length of dendrite without overlap. Here, the cover set was normalized for each dendrite by dividing it by the total number of thorns to give the cover fraction; the number of hMFB’s required to cover one thorn. This analysis was also used to determine the mean number of thorns per hMFB of a given radius. Analyses were run with radii set to 1.0, 1.4, 1.6, 2.0 and 2.2. Middle values (1.4, 1.6 and 2.0) were selected to represent the range of giant bouton cross sectional areas observed in the present study (See for hMFB values; measured MFB values ranged from 4.47 μm2 to 13.56 μm2 for the present study). Low and high end r values correspond to giant bouton areas that would be outside the biological range observed here (too small or excessively large, respectively).
Minimum cover set analysis revealed a significant difference between control and 3M SE animals when r was set to biologically relevant values (; r=1.4, 1.6 or 2.0). Cover fraction was significantly higher in 3M SE animals relative to control animals, indicating that thorns were more dispersed, and suggesting that a greater number of giant boutons would be required to cover all thorns present. These differences were present even though overall thorn density was statistically identical between the two groups (control, 2.59±0.34; 3M SE, 2.50±0.41; P=0.867, t-test), leading to a second finding: The mean number of thorns per hMFB significantly decreased in 3M SE animals for r=1.4, 1.6 or 2.0 (if cover fraction rises while thorn density remains the same, thorns/hMFB must drop; ). By contrast, significant differences vanished when r was set at values that would reflect giant boutons either too large or too small to be biologically relevant (), indicating that the findings are not arbitrary in nature.
Pyramidal cells from epileptic animals are innervated by greater numbers of giant boutons
To test the prediction that altered thorn distribution in animals exposed to status reflects input by larger numbers of giant mossy fiber boutons, a method for simultaneous labeling of pyramidal cell thorny excrescences and apposed granule cell giant boutons was needed. Unfortunately, the Timm stain – a reliable histochemical method used to label mossy fiber axons based on their high zinc content – is not readily adaptable to dual labeling approaches, and antibodies for brain-derived neurotrophic factor (BDNF) and neuropeptide Y, albeit promising (Scharfman et al., 2002
), only labeled a subset of GFP-expressing giant boutons (Danzer et al., 2004
; S.C. Danzer, unpublished observations). A novel approach for revealing giant boutons was therefore developed using antibodies targeted against the zinc transporter, ZnT-3. Grossly, Znt-3 immunostaining produced a pattern of labeling virtually identical to the Timm stain (). When the approach was tested by examining GFP-labeled giant boutons and ZnT-3 immunoreactive puncta, and almost perfect correspondence was observed (). Specifically, of 147 GFP-expressing giant boutons examined (63 control, 84 3M SE), 145 were ZnT-3 positive (98.6%). Moreover, ZnT-3 labeling roughly corresponded to the borders of the giant boutons, as revealed by the GFP label. These data confirmed that Znt-3 immunostaining can be used as a reliable marker of granule cells giant mossy fiber boutons.
Figure 6 A: Confocal maximum projection showing GFP-expressing hippocampal granule cells and CA3 pyramidal cells (green) and ZnT-3 immunolabeling (blue). B: Confocal images of GFP-expressing giant mossy fiber boutons from control animals. Maximum projections throughout (more ...)
Casual analysis of ZnT-3 immunolabeling in GFP expressing brain sections yielded several intriguing findings. Firstly, a tight correspondence between GFP-labeled thorns and ZnT-3 immunoreactive puncta was readily apparent (&). This was particularly true for the elaborate thorny excrescences, which were invariably associated with ZnT-3 immunoreactive puncta. By contrast, although most isolated thorns were also associated with immunoreactive puncta (, middle row), this was not always the case (not shown), and 16.0±5.3% of thorns from controls, and 11.3±2.1% of thorns from 3M SE animals (P=0.368, t-test) were not apposed to immunoreactive puncta. Whether these thorns reflect silent synapses, input from cell populations other than granule cells, or other factors is not clear; however, given the nearly perfect correlation between ZnT-3 labeling and mossy fiber terminals, it seems unlikely that these thorns receive mossy fiber input.
Figure 7 Pseudocolored maximum projections of GFP-expressing CA3 pyramidal cells (red) and passing granule cell mossy fiber axons (green) are shown in the left column. Middle and right columns show ZnT-3 immunolabeling and merged ZnT-3+GFP labeling. The z-depth (more ...)
To estimate the number of giant boutons innervating segments of CA3 pyramidal cell dendrites with thorn densities greater than 1.0/micron, 12 GFP-expressing cells from control animals and 16 from 3M SE animals were imaged from brain sections stained with ZnT-3 antibodies. GFP-labeled thorns were identified, and the number of thorns per ZnT-3 immunoreactive puncta, and the number of puncta per dendrite was determined. While thorn density did not differ between groups (control, 2.45±0.21; 3M SE, 2.90±0.38; P=0.350, t-test), the number of ZnT-3 puncta/10 μm of dendrite was significantly increased three months after status (control, 2.64±0.38; 3M SE, 4.92±0.33; P<0.001, t-test). Correspondingly, with more puncta, but relatively similar numbers of thorns, the number of thorns per puncta was significantly reduced (control, 10.74±1.95; 3M SE, 5.09±0.38; P=0.039, Mann-Whitney RST). Finally, the cover fraction (# puncta to cover 100 thorns) increased significantly from 14.00±2.71 (control) to 21.28±1.52 (3M SE; P=0.02, t-test). Together, these data provide independent support for the conclusion that in epileptic animals, although overall thorn density is preserved, the pattern of innervation shifts towards larger numbers of giant boutons contacting fewer thorns.
Evidence for cell intrinsic rather than regional regulation of CA3 pyramidal cells thorn density
Finally, we sought to explore whether neuronal plasticity among cells with high thorn densities might be regulated by cell intrinsic or regional factors. To begin to address this issue, we first compared thorn density between different branches of the same cell. CA3 pyramidal cells frequently project several dendritic branches through stratum lucidum, and these branches are often separated by tens of microns. Apical dendritic branches from the same cell, therefore, may encounter different local environments. If local factors predominate in regulating the density of thorns, different branches belonging to the same neuron may vary considerably. Alternatively, if thorn density for a given neuron is regulated by intrinsic factors, different branches would be predicted to be similar despite physical separation. To explore these different scenarios, a subset of CA3 pyramidal cells for which multiple dendritic trees were present in stratum lucidum was analyzed. Despite striking variability among pyramidal cells present in the same tissue sections (), thorn densities on distinct dendritic trees belonging to the same cells were highly correlated (). Significant correlations were evident both for neurons from control animals (R=0.982, P<0.0001, Pearson Product Moment Correlation, N=30) and epileptic animals (1M SE, N=22, R=0.932, P<0.0001; 3M SE, N=9, R=0.988, P<0.0001). By contrast, regional factors were not predictive of thorn density. Density was not significantly correlated with a cell’s bregma coordinates (Paxinos and Franklin, 2001
), medial-lateral position within CA3b (measured relative to the dentate granule cell layer) or soma depth within the pyramidal cell layer for either control (Supplemental Table 1
) or epileptic groups (not shown). As a last note, no obvious associations between dendritic structure and thorn density were found (Supplemental Table 1
). Taken together, these finding suggest that pyramidal cell thorn density is regulated on a cell-by-cell, rather than a regional basis. That said, the present study was deliberately designed to select cells from relatively restricted anterior-posterior positions and pyramidal cell layer subregions (to reduce variability and increase the likelihood of detecting differences between control and epileptic animals). Whether comparisons between more disparate regions (e.g. CA3a vs. CA3c) would produce different results is not known.
Figure 9 Adjacent CA3b pyramidal cells can exhibit dramatically different thorn densities. A: Pseudocolored confocal maximum projection showing GFP-labeling in the hippocampus of a control Thy1-GFP-expressing mouse. Dentate granule cells (dg) and CA1 pyramidal (more ...)
Figure 10 The density of thorns along different branches of the same dendritic tree is highly correlated. Images show neuronal reconstructions of two different pyramidal cells, one with a high density of thorns (top, control) and one with low density (bottom, 3M (more ...)
Status epilepticus does not alter the distribution of GFP expressing cells
In the present study, CA3 pyramidal cells were labeled using the Thy1-GFP mouse line. Previous studies of hippocampal granule cells suggest that seizures do not alter the pattern of GFP labeling in these animals (Danzer and McNamara, 2004
; Walter et al., 2007
; Danzer et al., 2009
). To insure that this was also true for CA3 pyramidal cells, we examined the number and distribution of labeled cells in these animals. Counts of GFP-expressing CA3b pyramidal cells in dorsal hippocampus revealed a non-significant decrease in pilocarpine-treated animals relative to controls (control, 0.42±0.08 GFP-expressing pyramids/hippocampus; 1M SE, 0.34±0.13; 3M SE, 0.33±13; P=0.350, ANOVA on ranks), as would be expected given the well-established sensitivity of pyramidal cells to seizure-induced death (Shibley and Smith, 2002
; Borges et al., 2003
). The data suggests the perhaps as many as 20% of GFP expressing CA3 pyramidal cells are lost following pilocarpine treatment (note that due to the low density of GFP expressing pyramids – perhaps 1% – detecting such a small change would require a prohibitively large number of animals. Thus, the negative finding here should be interpreted cautiously). Most importantly for the present study, the predicted direction and small size of the change suggests that seizure activity is not dramatically altering GFP expression (e.g. a two-fold increase would be readily detectable with >80% statistical power).
As an additional measure to insure against the possibility of bias introduced by changes in GFP expression, the regional distribution of pyramidal cells used for thorn density counts was examined. Pyramidal cells were found at equivalent bregma coordinates (control, −2.39±0.04; 1M SE, −2.28±0.08; 3M SE, −2.34±0.07; P=0.43, ANOVA), equivalent distances from the granule cell layer (control, 544±31 μm; 1M SE, 506±38 μm; 3M SE, 549±47 μm; P=0.695, ANOVA) and equivalent depths with the pyramidal cell layer (control, 45.0±4.6%; 1M SE, 42.2±5.0%; 3M SE, 43.5±6.0%; P=0.898, Kruskal-Wallis rank sum test; values were normalized by dividing the distance of the soma from stratum lucidum by the total thickness of the pyramidal cell layer, and are expressed as percentages). In summary, although selective cell loss could contribute to the current findings; similar number and distribution of GFP labeled cells in control and treated animals supports the conclusion that Thy1 driven GFP expression is a reliable and consistent marker of CA3 pyramidal cells.