Subpopulations of neurons expressing parvalbumin in the human amygdala

Abstract

Amygdalar intrinsic inhibitory networks comprise several subpopulations of gamma-aminobutyric acidergic neurons, each characterized by distinct morphological features and clusters of functionally relevant neurochemical markers. In rodents, the calcium-binding proteins parvalbumin (PVB) and calbindin D28k (CB) are coexpressed in large subpopulations of amygdalar interneurons. PVB-immunoreactive (-IR) neurons have also been shown to be ensheathed by perineuronal nets (PNN), extracellular matrix envelopes believed to affect ionic homeostasis and synaptic plasticity. We tested the hypothesis that differential expression of these three markers may define distinct neuronal subpopulations within the human amygdala. Toward this end, triple-fluorescent labeling using antisera raised against PVB and CB as well as biotinylated Wisteria floribunda lectin for detection of PNN was combined with confocal microscopy. Among the 1,779 PVB-IR neurons counted, 18% also expressed CB, 31% were ensheathed in PNN, and 7% expressed both CB and PNN. Forty-four percent of PVB-IR neurons did not colocalize with either CB or PNN. The distribution of each of these neuronal subgroups showed substantial rostrocaudal gradients. Furthermore, distinct morphological features were found to characterize each neuronal subgroup. In particular, significant differences relative to the distribution and morphology were detected between PVB-IR neurons expressing CB and PVB-IR neurons wrapped in PNNs. These results indicate that amygdalar PVB-IR neurons can be subdivided into at least four different subgroups, each characterized by a specific neurochemical profile, morphological characteristics, and three-dimensional distribution. Such properties suggest that each of these neuronal subpopulations may play a specific role within the intrinsic circuitry of the amygdala.

Fig. 1

Diagrams representing coronal sections through the amygdala chosen as standards for quantification of PVB-IR neurons in the BLC-CO. In this study, neurons were counted in four representative coronal sections of the amygdala in each of five normal subjects. The diagrams shown here from rostral (section 1) to caudal (section 4) were used as standards for selecting such sections so that each rostrocaudal level could be closely matched across cases. These sections were chosen to maximize the number of PVB-IR detected. Nuclei included in the BLC-CO definition are filled in gray. AA, anterior amygdaloid area; AB, accessory basal nucleus; BN, basal nucleus; Ce, central nucleus; CO, cortical nucleus; Ent, entorhinal cortex; LN, lateral nucleus; Me, medial nucleus; pha, parahippocampalamygdaloid transitional area (modified from Mai et al., 1997).

Fig. 2

Photomicrographs showing examples of immunoreactive neurons in the LN. A–K show confocal microphotomicrographs; pseudocolors were red for PVB, green for PNN, and blue for CB. A–D show two PVB/PNN/– neurons. These neurons were found to express PVB (A) and to be surrounded by a PNN (B), but they were not IR for CB (C). D displays the overlay of the three markers. The pseudocolor yellow represents overlap between the red and the green. E–H illustrate a PVB/PNN/CB neuron. This neuron was demonstrated to express PVB (E) and CB (G) and to be wrapped by a PNN (F). E–G are shown superimposed in H to confirm the colocalization of the three markers (pseudocolor white). I–K show examples of neurons exclusively IR for PVB, PNN, and CB, respectively. Note that these microphotographs are overlays of the three channels, so that cell bodies and neuropil IR for each of the three markers are visible in the background. L shows a light microscopic photomicrograph of a PVB-IR neuron (black) surrounded by a PNN (brown). Scale bars = 100 μm in D (applies to A–D), H (applies to E–H), L (applies to I–L).

Fig. 3

Examples of four morphological types of PVB-IR neurons in the LN. Light microscopy photomicrographs of PVB-IR neurons belonging to each of the four morphological categories identified in this study. A and B show examples of large and small multipolar neurons, respectively. These neurons had irregularly shaped or rounded somata and several primary dendrites emerging in various directions. C and D show examples of large and small bipolar neurons, respectively. Note the fusiform shape of the somata and the two primary dendrites emerging at the opposite poles. Scale bar =30 μm in D (applies to A–D).

Fig. 4

Plots and photomicrographs of CB-IR neurons, PNN, and PVB-IR neurons within the LN, BN, AB, and CO. In order to show, and compare, the distribution of CB- and PVB-IR neurons and PNNs in the human amygdala, 40-μm-thick adjacent sections from one of the five subjects included in this study were immunostained for CB, PVB, and PNN by using a nickel-enhanced diaminobenzidine protocol. Plots were made via computer-assisted light microscopy (Bioquant Nova Prime v6.0; R&M Biometrics Inc.). Sections are displayed in rostrocaudal order (upper to lower). Each dot represents a neuron. CB-IR neurons were found to be scattered throughout the BLC-CO, with the AB showing by far the highest numbers. PVB-IR neurons and PNNs were found mostly in the lateral and ventrolateral subdivisions of the LN and, to a lesser extent, in the magnocellular subdivision of the BN. Note the large degree of overlap in the distribution of these two latter markers. Bi, basal nucleus intermediate; Bmc, basal nucleus magnocellular subdivision; Bpc, basal nucleus, parvicellular subdivision; LnL, lateral nucleus lateral subdivision; LnM, lateral nucleus medial subdivision; LnvL, lateral nucleus ventrolateral subdivision. Scale bar = 120μm (applies to all photomicrographs).

Fig. 5

Pie charts representing the percentages of each neuronal subpopulation identified within the BLC-CO. Among PVB-IR neurons, the two largest subgroups were represented by PVB/PNN/– and PVB/–/–neurons. Interestingly, PVB/PNN/CB neurons represented the smallest population, whether considered as percentages of PVB (A), CB (B), or PNN (C). It is also worth noting that –/PNN/CB neurons were not detected.

Fig. 6

Percentage distributions and neuron numbers of subgroups of PVB-IR neurons vary across a rostrocaudal gradient. In A and B, subgroups of neurons are shown as percentages of the entire population of PVB-IR neurons counted in the BLC-CO (four sections/ case; five cases). Data from the rostral (sections 1, 2; A) and the caudal (sections 3, 4; B) portions of the BLC-CO are represented separately to show changes within the rostrocaudal gradient. These changes aredue in great part to an increase of the percentage of PVB/PNN/–neurons in caudal portion at the expense of PVB/–/CB neurons. In C, bar graphs show comparisons between total numbers of cells (logarithmic transformation) for each neuronal subgroup detected in the rostral and caudal portion of the BLC-CO. Numbers of PVB/–/–, PVB/PNN/–, –/–/CB, and –/PNN/– neurons were significantly higher in the caudal portion.

Fig. 7

Distribution of neuronal subgroups along the rostrocaudal axis of the BLC-CO. Numbers of cells for each neuronal subgroup examined were plotted for each of the four sections included in the analysis to show their rostrocaudal distribution. Numbers of neurons were averaged across subjects (error bars represent SEM). In the BLC-CO, numbers of PVB/PNN/– and PVB/–/– neurons progressively increase in more caudal sections, whereas numbers of PVB/–/CB show the opposite tendency. [Color figure can be viewed in the online issue, which is available at www.interscience.wiley.com.]

Fig. 8

Comparisons between distribution slopes of subgroups of IR neurons. Numbers of neurons counted in each of the four sections analyzed were averaged across cases. Paired t-tests, corrected for multiple comparisons, were used to test whether the rostrocaudal distribution slopes of pairs of neuron subgroups were significantly different. Numbers of PVB/PNN/– neurons sharply increased in the caudal portion of the BLC-CO (sections 3, 4), whereas PVB/–/CB neurons showed a decrease. The distribution slopes of these two neuron subgroups were found to be significantly different (P=0.04; A). The distributions of the other pairs considered did not show significant differences (B–D).

Fig. 9

Distribution of neuronal subgroups along the rostrocaudal axis of the LN and BN. In one representative amygdala, numbers of cells for each neuronal subgroup were plotted for each of the four sections included in the analysis to show their rostrocaudal distribution. In the LN, numbers of PVB/PNN/– and PVB/–/– neurons progressively increased in more caudal sections, whereas numbers of PVB/–/CB showed the opposite tendency. These distribution patterns were not detected in the BN, suggesting that rostrocaudal gradients detected in the BLC-CO may be driven mainly by the LN. [Color figurecan be viewed in the online issue, which is available at www.interscience.wiley.com.]

Fig. 10

Observed distribution of four subgroups of PVB-IR neurons within two coronal planes of the human amygdala. Neurons were plotted from one of the five amygdalas included in the study. In the rostral plane (compare with section 2 in Fig. 1), PVB/PNN/– and PVB/PNN/CB neurons were found mostly in the ventrolateral portion of LN, whereas most PVB/–/CB and PVB/–/– were detected in the lateral LN. The magnocellular portion of the BN contained mainly PVB/–/– and PVB/PNN/– neurons. In more caudal planes (compare with section 4 in Fig. 1), PVB/PNN/–, PVB/PNN/CB, and PVB/–/–neurons were strictly confined to the ventrolateral subdivision of the LN, and fewer PVB/–/CB were found in its lateral subdivision. [Color figure can be viewed in the online issue, which is available at www.interscience.wiley.com.]

Fig. 11

Contingency table representing the morphological subtypes for each amygdalar neuronal subpopulation included in the study. Specific neurochemical subgroups were strongly associated with distinct morphological phenotypes (Chi-squre P=0.0001). Strikingly, neurons possessing PNNs showed a strong tendency to be large and multipolar, with the conspicuous exception of PVB/PNN/CB neurons, which were all small bipolar cells. Consistently, PVB/–/CB neurons also were mostly small fusiform. LB, large bipolar; LM, large multipolar; MP, multipolar; SB, small bipolar; SM, small multipolar. [Color figure can be viewed in the online issue, which is available at www.interscience.wiley.com.]