Regional and age-related differences in GAD67 expression of parvalbumin- and calbindin-expressing neurons in the rhesus macaque auditory midbrain and brainstem

Histological processing and antibody characterization

Animals were euthanized with an overdose of sodium pentobarbital (60 mg/kg, i.v.) and transcardially perfused with a solution of 4% paraformaldehyde in 0.01% phosphate-buffered saline (PBS; pH 7.4), followed by a mixture of 4% paraformaldehyde and 10% sucrose. After perfusion, the brains were extracted and placed in a solution of 4% paraformaldehyde and 30% sucrose for cryoprotection. All brains were cut transversely at either 25 or 30 μm. Sections were stained for PV or CB immunohistochemistry along with GAD67 immunohistocemistry (see below). In total, at least eight and up to 20 individual sections showing colocalization of both PV and CB with GAD67 were analyzed per animal, and this range depended on the size of the region studied. Adjacent sections were always processed for different combinations of CaBP and GAD67 or with other markers not described in the current article, which allowed for a minimum spacing of 100–120 μm between sections processed for a given combination. This spacing was sufficient in sampling neurons from the majority of the rostrocaudal axis of both the IC and the SOC in all animals.

Brain sections were blocked for 24 hours in 3% normal horse serum (Vector, Burlingame, CA; S-2000) and 0.25% Triton X-100 (Sigma-Aldrich, St. Louis, MO), followed by a 24-hour incubation in a solution containing a mixture of either the primary PV antibody (anti-PV in mouse, 1:2,000; Sigma-Aldrich; p3088; RRID: AB_477329) or primary CB antibody (anti-CB in mouse, 1:2,000; Sigma-Aldrich; c9848; RRID: AB_476894) along with the primary GAD67 antibody (anti-GAD1 in rabbit; 1:1,000; Sigma-Aldrich; SAB4300642; RRID: AB_11129033). After this incubation, the sections underwent three 15-minute washes in 0.01% PBS. The sections were then incubated for 48 hours in a solution containing two secondary fluorescent antibodies of different emission wavelengths, one specific to mouse IgG (Alexa Fluor 594; 1:250; Life Technologies, Grand Island, NY; A21203; RRID: AB_10563558) and the other specific to rabbit IgG (Alexa Fluor 488; 1:250; Life Technologies; A21206; RRID: AB_10049650). After three more 15-minute rinses in PBS, the sections were mounted with a Vectashield fluorescent mounting medium containing 4',6-diamidino-2-phenylindole (DAPI; Vector; H1200) and coverslipped. The information for all antibodies and chemicals used in these procedures is summarized in Table 2.

Several tests were used to confirm the isotype specificity of all antibodies used in the study. The Sigma ImmunoType Kit (product code ISO-1) and a double diffusion immunoassay using mouse monoclonal antibody isotyping reagents (product code ISO-2) confirmed isotype specificity of the anti-PV antibody as frog leg PV and the anti-CB antibody as purified bovine kidney CB-D-28k. Immunoblotting of the anti-GAD antibody against extracts from rat and mouse brain yielded a single band, confirming specificity to the GAD67 protein. Furthermore, exclusion of the primary antibody yielded only weak background signal in all animals and no punctuate cellular labeling, indicating that positive signals resulted from the antibodies used and not extraneous label.

Identification of the IC and SOC

The different subregions of the IC (central and external nucleus) and the SOC (medial, lateral, and trapezoid regions) were defined by using the DAPI counterstain on all CaBP-stained sections. In accordance with previous investigations, the IC was defined as being dorsal to the cuneiform nucleus and the nucleus of the lateral lemniscus, lateral to the periaquaductual gray, ventral and caudal to the superior colliculus, and caudal to the central tegmental field (Oliver, 2005). We differentiated the central nucleus (ICc) from the peripheral cortex (ICx) based on their relative immunoreactivity to PV and CB, because the ICc and ICx of primates have previously been segregated based on the immunoreactivity to these molecules, with the ICc being PV rich and the ICx being CB rich (Jones, 2003; Parviz and Damasio, 2003; Engle et al., 2014; Fig. 1A,C). The three principle subdivisions of the SOC, the medial nucleus of the trapezoid body (MNTB), medial superior olive (MSO), and lateral superior olive (LSO), were segregated based on the conventions of Strominger and Hurwitz (1976) and Irving and Harrison (1967). Briefly, the MSO was characterized as a vertically oriented cell-rich region in the caudal pontine tegmentum surrounded by heavy staining of the neuropil. The LSO was characterized by a globular cell-rich region located laterally and slightly dorsally to the MSO, and the MNTB was identified as a cell rich region with large perikarya lying medially and ventrally to the MSO (Fig. 1B; see also Gray et al., 2014a).

Data analysis

Fluorescence microscopy was used to coregister CaBP immunoreactivity (PV or CB) with GAD67 immunoreactivity. Individual neurons were sampled by taking ×20 or ×40 images of the different subdivisions of the IC and SOC using an appropriate set of fluorescent filters. All images were line averaged to reduce background noise and imported into Adobe Photoshop CS5, in which they were coregistered. Images of the counting fields were taken away from the anatomical borders of the regions of interest to ensure that all sampled neurons belonged to the appropriate subdivisions.

To assess the proportion of neurons coexpressing CaBPs and GAD67, the total numbers of neurons staining positively for PV or CB and GAD67 were counted by two observers blind to the identity of the animals. A CaBP-immunopositive neuron was defined as containing label and having a punctate, cell-like morphology. Colabeled neurons were identified in the overlaid images. More specifically, the PV and CB label filled entire cells, and the GAD67 label was found intracellularly in much smaller quantities within either the cytoplasm or the nucleolus, which is in agreement with previous reports for this level of the auditory system (Ito et al., 2010). Neurons with minimal label in any filter were not considered coexpressing to minimize the effects of the weak background noise of these labels on the estimations of the proportion of coexpressing neurons (see below). Comparisons between the two observers showed no significant differences in their counts of all cell types (paired t-tests; all P > 0.2), so their counts were combined for analysis. Proportions of colocalizing neurons (P_C) were calculated as

with C being the coregistered cells and N being non-coregistered cells. These standards were used for all subdivisions of the SOC and IC. Furthermore, to ensure that our counting criteria and quantification metrics did not skew the data in any way, sections from the striatum underwent an identical analysis, because PV- and CB-expressing neurons in these areas have been shown to be primarily GABAergic, and the results of this analysis confirm the appropriate sensitivity of this analysis (see Results).

Statistical analysis

Two statistical tests were performed, first at the level of individual neurons and second at the level of individual animals. In the first analysis, neurons from all animals were grouped and categorized as olivary/collicular for the between-region analysis and young/old for the aging analysis. A Mann-Whitney U test for nonparametric populations was used to analyze whether the proportions of neurons that coexpressed GAD67 and PV/CB were significantly different between the IC and the SOC or between age groups. The age cutoff was 21.66 years (roughly 65 human years). A relatively stringent significance criterion of P < 0.01 was used in these analyses. The total number of neurons sampled in each comparison ranged between 272 and 498 neurons per group for the regional analysis and between 51 to 202 neurons per group for the aging analysis. This grouping increased the sample size because each neuron was considered a data point that either did or did not express GAD67. This analysis provided more power to detect changes (given the relatively small number of animals) than the multiple-way ANOVA described below, which uses each individual animal as a data point.

In the second analysis, we sought to understand the data at the level of individual animals by using a two-way ANOVA with region (IC vs. SOC) and age as factors. A significance criterion of P < 0.05 was used in this analysis because protection against the effects of multiple comparisons is built into the ANOVA model. Overall, this is a more stringent analysis than the Mann-Whitney U test, especially given the much smaller sample size (n = 8 animals). Both of these analyses together are useful for understanding the degree of intersubject variability (reflected strongly in the ANOVA) for both region- and age-related differences (detected through the less stringent Mann-Whitney U test).

Finally, to determine age-related density changes in PV⁺ and CB⁺ neurons (as opposed to the proportions described above), the density values from young and old animals underwent an unpaired t-test with a significance criterion of P < 0.01.

Characterization of the IC and SOC

We examined the IC and SOC in transverse histological sections from eight rhesus macaques ranging in age from 12 through 28 years, which roughly corresponds to 36 through 84 human years (Davis and Leathers, 1985). Neurons from these sections were labeled with GAD67 antibodies along with either PV or CB antibodies (see Materials and Methods). All animals stained positively for all antibodies used, so both combinations of PV-GAD67 and CB-GAD67 were observed and analyzed. Borders of the IC and SOC (Fig. 1A,B) were determined with the help of a DAPI counterstain (see Materials and Methods).

The proportion of PV⁺ and CB⁺ neurons expressing GAD67 differed in the IC and SOC

In rodents, most PV⁺ and CB⁺ neurons in the IC express GAD67, whereas for the SOC this is not the case (Fredrich and Reisch, 2009). Because these observations have not been made in the macaque, at the first level of analysis we compared the proportion of sampled PV⁺ and CB⁺ neurons that expressed GAD67 in the IC and SOC (Fig. 2A,B). We observed PV⁺ and CB⁺ neurons expressing GAD67 in both the IC and the SOC, although in different proportions. In the IC, 57% of PV⁺ neurons sampled were also GAD67⁺, whereas only 42% expressed GAD67 in the SOC (Mann-Whitney U, P < 0.01; Fig. 2D). Conversely, CB⁺ neurons in the IC and SOC did not differ in the proportion of the neurons expressing GAD67 (61% and 62%, respectively; Mann-Whitney U, P > 0.05; Fig. 2D). These data suggest that at the level of individual neurons there are regional differences in the expression of GAD67 for populations of PV neurons but not CB neurons. These observations held true at the level of individual animals (two-way ANOVA; PV: P = 0.03; CB: P > 0.05), suggesting that there is little variability in these observations between monkeys.

As a control for the GAD67 specificity observed in the IC and SOC, we also sampled neurons from the striatum (Fig. 2C), because the molecular profiles of PV⁺ and CB⁺ neurons in this region are well known (Carder et al., 1996; Parent et al., 1996; Prensa et al., 1998; Cicchetti et al., 2000; Rudy et al., 2011) and make up separate chemically defined classes of GABAergic interneurons. Our results are consistent with these previous findings in that PV⁺ and CB⁺ neurons in the striatum expressed GAD67 in most neurons sampled. Quantification of these observation revealed that PV⁺ neurons expressed GAD67 in 83% of neurons sampled, and CB⁺ neurons expressed GAD67 in 92% of the neurons sampled (Fig. 2, light gray bars). These proportions are significantly different from the proportion of PV⁺ and CB⁺ neurons expressing GAD67⁺ in the IC and SOC (Mann-Whitney U, P < 0.01; Fig. 2D, asterisks). Again, a two-way ANOVA yielded identical results (both PV and CB P < 0.05), indicating that these differences are also seen at the level of individual animals. Thus, our methodology yielded results consistent with those found previously for the striatum, and the percentage of PV⁺ and CB⁺ neurons expressing GAD67 in the macaque SOC and IC is significantly different from that in the striatum. Together these data suggest that there exist subclasses of CaBP-expressing neurons in the auditory brainstem and midbrain that differ chemically and physiologically from cerebral neurons expressing the same CaBPs and that these observations are consistent across animals.

Age-related changes in PV⁺ and CB⁺ neurons expression

To determine whether there is an age-related change in the density of neurons expressing these CaBPs, as reported previously (Gray et al., 2013, 2014a,b; Engle et al., 2014), we quantified the age-related density changes of PV⁺ and CB⁺ neurons by counting the number of PV⁺ or CB⁺ neurons per section analyzed and compared these values across age groups. The results of the current study were consistent with prior studies in that PV expression was significantly increased in the MSO, LSO, and ICc with age (unpaired t-test, P < 0.01; Fig. 3A), but not in the MNTB or ICx. Age-related increases in CB expression were noted only in the MSO (unpaired t-test, P < 0.01; Fig. 3B), and to our knowledge this is the first report of age-related changes in the density of CaBP-expressing neurons in the macaque auditory system.

Age-related changes in the proportion of GAD67-expressing PV⁺ and CB⁺ neurons in the IC and SOC

Given that we observed age-related changes in the density of PV⁺ and CB⁺ neurons, the next level of analysis was to determine whether there was also a difference in the percentage of these neurons that also expressed GAD67. The quantification of the proportion of these cells within the IC and SOC revealed changes in some, but not all, subdivisions of these structures. PV and GAD67 showed no significant changes in coexpression in the ICc as a function of age (Mann-Whitney U; P > 0.05) but an increase with age in the ICx (Mann-Whitney U; P < 0.01; Fig. 4A). The proportion of CB⁺ neurons in both the ICx and the ICc expressed GAD67 equally with age (Mann-Whitney U; P > 0.05; Fig. 4B). Changes were also noted in the SOC, although they differed from the changes seen in the IC. The proportion of PV⁺ neurons expressing GAD67 significantly decreased with age in the LSO (Mann-Whitney U; P < 0.01; Fig. 4A) and showed a trend to decrease in the MSO (Mann-Whitney U; P = 0.06). The changes in CB and GAD67 coexpression with age were confined to the MSO, where significant increases were noted (Mann-Whitney U; P < 0.01; Fig. 4B).

Because one or more subdivisions showed either significant differences or trends toward increases or decreases in the coexpression of GAD67 and PV⁺/CB⁺ neurons as a function of age, we combined the values of all IC and SOC subdivisions. This analysis revealed that as a population there was no change in the coexpression of GAD67 with either PV⁺ or CB⁺ neurons in the IC with age; however, in the SOC, PV⁺ neurons express GAD67 less often and CB⁺ neurons express GAD67 more often with age (Mann-Whitney U, P < 0.01; Fig. 4C). Together these data suggest that, at the level of individual neurons, age-related changes in the inhibitory molecular profile in populations of PV⁺ and CB⁺ neurons exist in the SOC, but not the IC, and are different among CaBPs. In general, the populations of PV⁺ neurons became less inhibitory with age, whereas populations of CB⁺ neurons become more inhibitory with age, but these observations are not absolute (PV⁺ neurons in the ICx) nor present in all auditory subdivisions. When these same data underwent a two-way ANOVA, no statistically significant changes were observed, indicating that there is considerable inter-animal variation in these age effects that, given the small sample size, masks any age-related effects at the level of individual animals (Fig. 4D). Together these data suggest that populations of PV- and CB-expressing neurons in the auditory brainstem change their expression of GAD67, but these changes do not necessarily occur in all aged animals.

Age-related changes in PV and CB neuron density

Neurons of the central auditory system containing CaBPs have repeatedly been shown to change in density during the natural aging process of both rodents (O'Neill et al., 1997; Zettel et al., 1997; Idrizbegovic 2003, 2004, 2006; Ouda et al., 2008, 2012) and nonhuman primates (Gray et al., 2013, 2014a, b; Engle et al., 2014; Fig. 3). The present study is the second report of age-related density changes in PV-expressing neurons within the auditory system of the macaque and the first report of age-related changes in CB-expressing neurons. Although the density changes of CB-expressing neurons were confined to the MSO, unlike the more widespread density changes of PV-expressing neurons, it is clear that the number of neurons expressing both of these CaBPs can change in aged auditory nuclei. Notably, all reported age-related changes in the density of CaBP-expressing neurons have resulted in increases, whereas both increases and decreases have been shown in rodents. These observations suggest that there are cross-species differences in the expression of these neurons and in their density changes with age. The root of these inconsistencies may arise from cross-species differences in sound processing or from differences in the distribution of inhibitory neurons throughout the auditory pathway. For example, the auditory thalamus of rodents does not contain inhibitory neurons, but the same structure in primates and felines does (Winer and LaRue, 1996; Bartlett and Smith, 1999). Also, the specifics of binaural sound processing can differ significantly between species with different hearing ranges (for reviews see Moore, 1991; Pollack et al., 2002; Grothe and Koch, 2011). Regardless of the exact cross-species differences, the results from rodent models cannot be completely generalized to primates, and vice versa.

What remains unclear, however, is the role of these changes in the aging process. This gap in understanding arises partially from our minimal knowledge of the physiological response properties of CaBP-expressing neurons of subcortical structures. Most studies investigating the physiology of these neurons have used the cerebral cortex, striatum, or hippocampal formation, where most are GABAergic inhibitory neurons; therefore, the first goal of the present study was to uncover the GABAergic profiles of populations of PV⁺ and CB⁺ neurons in the auditory midbrain and brainstem.

Subcortical PV⁺ and CB⁺ cells differ from cerebral PV⁺ and CB⁺ neurons

The results of the current study reveal two important findings. First, as a population, PV⁺ neurons of the IC express GAD67 more often than in the SOC, a difference that is not seen with populations of CB⁺ neurons. Therefore, populations of CaBP-expressing neurons can differ in their net GAD67 expression between the midbrain (IC) and the brainstem (SOC), but this is not the case with all classes of CaBP-expressing neurons. Second, as a population, subcortical PV⁺ and CB⁺ neurons expressed GAD67 in significantly fewer cells than in populations of chemically analogous (with respect to CaBPs) neurons in the cerebrum (Fig. 2). This observation suggests that the presence of PV and CB in these subcortical structures does not relate to a single specific physiological profile of neurons as it does in the cerebrum but rather PV and CB are likely expressed in neurons performing different cellular processes, some of which are likely affected during the aging process.

Together these data suggest that CaBP-expressing neurons of the subcortical auditory system cannot, as a population, be considered inhibitory through GABAergic neurotransmission. Caution should be taken, however, when interpreting these results. It is possible that GAD67 expression is simply lower in SOC neurons compared with IC neurons, bringing them below detection levels. Also, we cannot definitively rule out that these non-GAD67-expressing neurons are not inhibitory in these areas of the brain through a different neurotransmitter system. As an example, many subcortical auditory neurons use glycine as an inhibitory neurotransmitter instead of GABA. To investigate these possibilities, it would be informative for future studies to colabel these CaBP-expressing neurons with other markers of GABA neurotransmission, such as the GABA vesicular transporter, GAT, or markers for other neurotransmitters. Alternatively, some neurons may express a different isoform of GAD, such as GAD65, which would elude the present analysis.

Despite the possibilities described above, substantial proportions of both PV⁺ and CB⁺ neurons do express GAD67 in both the SOC and the IC. These large values suggest that subclasses of CaBP-expressing neurons contain GAD67 and can be differentiated from one another based on independent morphological or chemical characteristics. The distribution of these subclasses of neurons may differ between regions of the auditory system, which would explain the IC–SOC differences in PV⁺ inhibitory cells. It is unknown what morphological cell types comprise the chemically defined neurons described in the current study. A better understanding of the morphological and chemical properties characterizing these CaBP-containing, GAD67-expressing neurons would be useful in elucidating their different expression patterns observed in the IC and SOC.

Functional implications

Aged neurons of the auditory system of both rodents and nonhuman primates have increased spontaneous and driven firing rates in both subcortical and cortical structures (Caspary et al., 2005, 2006; Juarez-Salinas et al., 2010; Engle and Recanzone, 2012). These increases result in broader spectral and spatial tuning of auditory neurons, both deficits that correlate with decreases in psychophysical metrics of auditory perception (Engle and Recanzone, 2012). Aside from disrupting auditory perception, these age-related increases in neural activity put auditory neurons at risk for excitotoxicity. This toxicity occurs when relatively high levels of glutamate cause greater depolarizations and more frequent action potentials, increasing the influx of calcium, which activates intracellular enzymes that in turn damage vital cellular components (Manev et al., 1989). Given the increased physiological activity in aged auditory neurons, mechanisms to prevent such excitotoxicity are likely necessary for normal cell maintenance. An effective way to compensate for the increased intracellular calcium would be to increase the expression of CaBPs, which act as calcium buffers.

The current data suggest that in the macaque, as in rodents (Fredrich and Reisch, 2009), the presence of CaBP within subcortical auditory neurons does not necessarily relate to the neurotransmitters used by these neurons. Rather, the CaBPs likely play different, nonelectrophysiological roles in the aging process of the auditory system. One hypothesis explaining the age-related increases in these neurons is that they function as a homeostatic response combatting the toxic effects of excitotoxicity. This idea is applicable to both rodent and nonhuman primate models because both experience age-related hyperexcitability in the auditory system (Caspary et al., 2006; Juarez-Salinas et al., 2010), both possess subcortical PV- and CB-expressing neurons that do not relate to a single class of neurotransmitter, and the basic intracellular physiological mechanisms (metabolic homeostasis, control of Ca²⁺ levels, etc.) are very similar. Whether this is true remains to be seen; however, these observations give credibility to such a preventive function for these cells.