Address correspondence and reprint requests to Ashok K. Shetty PhD, Associate Research Professor, Division of Neurosurgery, DUMC Box 3807, Duke University Medical Center, Durham, NC 27710, USA.E-mail: firstname.lastname@example.org
Increased excitability of principal excitatory neurons is one of the hallmarks of aging in the hippocampus, signifying a diminution in the number and/or function of inhibitory interneurons with aging. To elucidate this, we performed comprehensive GABA-ergic interneuron cell counts in all layers of the dentate gyrus and the CA1 and CA3 subfields, using serial sections from adult, middle-aged and aged Fischer 344 rats. Sections were immunostained for glutamate decarboxylase-67 (GAD-67, a synthesizing enzyme of GABA) and GAD-67 immunopositive interneurons were counted using an unbiased cell counting method, the optical fractionator. Substantial declines in the absolute number of GAD-67 immunopositive interneurons were found in all hippocampal layers/subfields of middle-aged and aged animals, in comparison with the adult animals. However, the counts were comparable between the middle-aged and aged groups for all regions. Interestingly, determination of the absolute number of interneurons using neuron-specific nuclear antigen (NeuN) expression in the strata oriens and radiatum of CA1 and CA3 subfields revealed an analogous number of interneurons across the three age groups. Furthermore, the ratio of GAD-67 immunopositive and NeuN positive interneurons decreased from adult age to middle age but remained relatively static between middle age and old age. Collectively, the results underscore that aging in the hippocampus is associated with wide-ranging decreases in the number of GAD-67 immunopositive interneurons and most of the age-related changes in GAD-67 immunopositive interneuron numbers transpire by middle age. Additionally, this study provides novel evidence that age-related reductions in hippocampal GAD-67 immunopositive interneuron numbers are due to loss of GAD-67 expression in interneurons rather than interneuron degeneration.
Multiple studies have demonstrated that aging is associated with increased excitability of principal hippocampal neurons (Landfield et al. 1986; Barnes et al. 1987; Kerr et al. 1991; Bekenstein and Lothman 1993; Barnes 1994; Papatheodoropoulos and Kostopoulos 1996). This was also evidenced by the absence of silent cells in the aged hippocampus in any environment (Tanila et al. 1997), suggesting that aging considerably impairs inhibitory processing in hippocampal circuits. Therefore, it is imperative to determine the extent of alterations in number and function of inhibitory gamma-amino-butyric-acid positive (GABA-ergic) interneurons in different regions of the hippocampus as a function of age. Previous studies using representative sections and less stringent counting methods have reported that hippocampal interneuron populations undergo various alterations during the course of aging. A previous study from our laboratory has shown that the density of interneurons positive for GABA synthesizing enzyme glutamate decarboxylase-67 (GAD-67) or distinct calcium binding proteins declines as a function of aging in the septal hippocampus of Fischer 344 rats (Shetty and Turner 1998). Recent studies, in addition, demonstrate that the density of hippocampal interneurons positive for somatostatin and neuropeptide Y (NPY) decrease with aging (Cadiacio et al. 2003; Vela et al. 2003). Nevertheless, it is uncertain whether these decreases in density reflect diminution in the absolute number of hippocampal interneurons or hypofunctionality of interneurons with altered phenotype, as age-related alterations in hippocampal GABA-ergic interneuron numbers have not been measured rigorously using unbiased stereological methods. Additionally, the potential link between the age-related decreases in the hippocampal GABA-ergic interneuron numbers and age-related interneuron degeneration has not been elucidated.
Inhibitory input from GABA-ergic hippocampal interneurons serves to thwart principal excitatory hippocampal neurons from becoming hyperexcitable (Freund and Buzsaki 1996). Generally, the GABA-ergic interneurons restrain principal excitatory neurons in the hippocampus through two different ways (Dvorak-Carbone and Schuman 1999; Yan et al. 2003). One way is through inhibition of input signals by projecting to the presynaptic axons, which serves to control the input information flow from one region to the other. The other means is through direct innervation of principal cell soma and dendrites, which controls the excitability of principal excitatory cells (Karnup and Stelzer 1999). Thus, inhibition mediated by interneurons sets a threshold for the excitation of pyramidal cells, and the strength of interneuron input controls the level of discharges of principal hippocampal cells. This is evidenced by epileptic-like conditions in animal models following disinhibition of the principal excitatory neurons (Prince 1978; Dingledine and Gjerstad 1980). Likewise, compromised inhibitory input from the GABA-ergic interneurons appears to be a major factor that underlies hyperexcitability in both temporal lobe epilepsy (TLE) and animal models of TLE (Franck et al. 1988; Cornish and Wheal 1989; Williams et al. 1993; Perez et al. 1996). In this context, it is possible that increased excitability of principal hippocampal neurons during aging is a result of age-related alterations in hippocampal GABA-ergic interneurons.
We hypothesize that aging in the hippocampus is associated with wide-ranging decreases in the absolute number of interneurons that synthesize GABA but not interneuron cell loss. To address this, we performed comprehensive GABA-ergic interneuron cell counts in all layers and subfields of the entire hippocampus using adult, middle-aged and aged Fischer 344 rats. The optical fractionator method was utilized for cell counting, as it provides an unbiased method for counting neurons (Dorph-Petersen et al. 2001). To visualize GABA-ergic interneurons, we used an antibody against GAD-67, which is an excellent immunohistochemical marker for identifying GABA-ergic interneurons (Esclapez et al. 1994; Dupey and Houser 1996; Shetty and Turner 2000, 2001). As significant reductions in GAD-67 immunopositive interneuron numbers were found with aging, we further determined whether the reduced GABA-ergic interneuron number was due to pervasive interneuron degeneration or a loss of GAD-67 expression. This was accomplished by quantifying neurons positive for neuron-specific nuclear antigen (NeuN) in the strata oriens and radiatum of CA1 and CA3 subfields. The NeuN is a marker of mature neurons (Mullen et al. 1992) and, in the hippocampus, NeuN is expressed in all neurons including GABA-ergic interneurons. As neurons in the strata oriens and radiatum of CA1 and CA3 subfields are exclusively interneurons, immunocytochemical staining using NeuN antibody visualizes all interneurons in these layers.
Materials and methods
Tissue processing and selection of sections
Adult (7 months old), middle-aged (15 months), and aged male Fischer 344 rats (23 months) were acquired from the National Institute of Aging colony located at Harlan Sprague-Dawley (Indianapolis, IN, USA). The animals were individually housed in an environmentally controlled room (∼23°C) with a 12 : 12-h light-dark cycle, and were given food (commercial rat chow) and water ad libitum. We used Fischer 344 rats in this study as this rat strain has several advantages. The genetic background is known, and the normal life span and development of the Fischer 344 rats are well defined; 1.5–2-month-old rats are juvenile, 3–6-month-old rats are young adults, 7–11-month-old rats are adults, 12–21-month-old rats are middle aged, and 22-month-old or older rats are considered aged. Their life expectancy is approximately 29 months, with a maximal survival time of approximately 36 months (Coleman et al. 1977). One week after their arrival, rats were deeply anesthetized with halothane and subsequently perfused through the heart with 200 mL of physiological saline containing 0.1% heparin for 3 min followed by 500 mL of 4% paraformaldehyde in 0.1 m phosphate buffer (PB, pH 7.4) for 30 min. All animals in this study were perfused as above during the afternoon (between 14.00 and 18.00 h). Additionally, all experiments were performed as per the animal protocol approved by the animal studies subcommittee of the Durham Veterans Affairs Medical Center and the Institutional Animal Care and Use Committee of the Duke University Medical Center. The brains were removed, post fixed in the same 4% paraformaldehyde solution for 18 h at 4°C, and cryoprotected using sucrose solution prepared in PB. Thirty-micrometer-thick cryostat sections were cut coronally through the entire hippocampus and collected in PB. For GAD-67 and NeuN immunostaining, serial sections (every 10th section for GAD-67 and every 20th section for NeuN) through the entire hippocampus were selected in each animal and processed for quantitative immunocytochemical analysis. The above tissue selection criteria ensured that the sections selected for immunohistochemical analyses were entirely independent of one another, and prevented counting interneurons from contiguous sections.
The GAD-67 immunostaining was performed using indirect immunoperoxidase method, as described in our earlier studies (Shetty and Turner 1998, 2000, 2001). In brief, sections were rinsed in 0.1 m phosphate-buffered saline (PBS) and incubated in 1% sodium borohydride solution for 15 min. Sodium borohydride reduces double bonds and free aldehyde groups, which results in enhanced immunoreactivity of protein antigens including GAD-67 (Toth and Freund 1992; Shetty and Turner 1998, 2001). Sections were washed in PBS, treated with 10% normal goat serum in PBS for 30 min, incubated overnight in the GAD-67 antibody (1 : 2000 dilution, Chemicon, Temecula, CA, USA), which is raised in the rabbit and has been shown to specifically recognize GAD-67 with western blot. Following this, sections were rinsed in PBS, treated with goat anti-rabbit peroxidase (1 : 500 dilution) for 2 h, washed in PBS, and the peroxidase reaction was developed using 3,3-diaminobenzidine (DAB) and nickel chloride as chromogens (Vector, Burlingame, CA, USA). For NeuN immunostaining, sections were treated with PBS solution containing 20% methanol and 3% hydrogen peroxide for 20 min, rinsed in PBS, and incubated for 15 min in 1% sodium borohydride in distilled water. Following this, sections were washed in PBS, blocked with 10% normal horse serum in PBS containing 0.1% Triton-X 100, and incubated overnight in the mouse NeuN antibody (1 : 1000 dilution, Chemicon). The NeuN immunostaining was visualized using the avidin-biotin complex method (elite mouse ABC kit, Vector). The peroxidase reaction was developed using DAB as the chromogen (Vector). For both GAD-67 and NeuN immunostaining, the chromogen reaction was initially standardized under a microscope to find the length of incubation that would result in optimal cell staining with minimal background. This same incubation length was subsequently used for staining across the three age groups. The sections were mounted on gelatinized slides, dehydrated, cleared, and coverslipped with vectamount (Vector). To exclude any possible effects of staining protocol on the number of GAD-67 or NeuN positive neurons, sections from the three age groups were processed in parallel with identical concentrations of primary and secondary antibody solutions. Furthermore, the concentration of DAB and hydrogen peroxide for the chromogen reaction was kept constant.
Quantification of GAD-67 and NeuN positive neurons
For quantification, the number of GAD-67 immunopositive neurons in each animal belonging to the three age groups (n = 5/age group) was measured separately in two regions of the dentate gyrus (dentate hilus plus granule cell layer and the dentate molecular layer) and three distinct layers (strata oriens, pyramidale and radiatum) of CA1 and CA3 subfields. Thus, all hippocampal layers except for the stratum lacunosum of CA1 were selected for counting of GAD-67 immunopositive cells. For quantification of NeuN positive neurons, the number of NeuN positive neurons in each animal was measured in the two layers (strata oriens and radiatum) of CA1 and CA3 subfields. The boundaries were determined for each layer and subfield according to Paxinos and Watson (1986). The layers of smaller CA2 subfield were included with the layers of CA1 subfield. Cells were counted (in every 10th section for GAD-67 and every 20th section for NeuN) through the entire antero-posterior extent of the hippocampus using the Stereo Investigator system (Microbrightfield Inc., Williston, VT, USA). The Stereo Investigator system consisted of a color digital video camera (Optronics Inc., Muskogee, OK) interfaced with a Nikon E600 microscope.
Counting frame and sampling scheme
In every animal, for each hippocampal layer listed above, GAD-67/NeuN positive cells were counted from 50 to 400 frames (each measuring 50 × 50 μm, 0.0025 mm2 area) in each of the selected sections using the 100 × oil immersion objective lens. The frames were selected using the systematic random sampling scheme. The systematic random sampling procedure provides an unbiased and efficient sampling technique. In this scheme, once the region of interest is marked with a contour, sampling sites are evenly distributed throughout the region of interest. The distribution pattern of the sampling sites in each tissue section is systematic, as the distance from each sampling site to the next is the same. The placement of this pattern of sampling sites is the random component. The number and density of frames were selected using the optical fractionator component of the Stereo Investigator system. For each layer, the number of frames per section counted varied because, the overall area for each hippocampal layer changed from section to section. For instance, areas of many layers in anterior-most and posterior-most sections through the hippocampus were consistently much smaller than areas in sections through the intermediate region of the hippocampus; as a result, cells were counted from a larger number of frames in sections through the middle of the hippocampus. The above sampling scheme is consistent with the principle of the optical dissector counting method that the density of sampling units must remain constant for each section. Thus, counting of cells from 50 to 400 randomly and methodically chosen frames in every 10th (for GAD-67) or 20th (for NeuN) section through different layers of the hippocampus guaranteed that effectively every GAD-67/NeuN positive cell in these layers had an equal odds of being counted. This is especially imperative because the distribution of interneurons in all layers of the hippocampus was characteristically heterogeneous.
Initial section thickness and shrinkage of tissue following GAD-67/NeuN immunostaining along the Z-axis
For these studies, we cut 30-μm-thick sections through the hippocampus using a cryostat that has been calibrated. Measurement of the thickness of sections immediately following sectioning has revealed that the variability between sections is minimal (i.e. ± 1 μm). With GAD-67/NeuN immunostaining, sections showed significant shrinkage along the Z-axis. However, the extent of shrinkage between sections from different animals in the three age groups was similar. Measurement of the thickness of sections in different regions of the hippocampus using a microcater (resolution 0.5 μm) revealed that the average thickness of sections was reduced to 53% of the initial section thickness and, hence, at the time of data collection, the thickness of sections was 16 μm. Thus, there was homogenous uniform deformation of the sections in regions of the hippocampus along the Z-axis.
Counting procedure and data analysis
In every section, the contour of every hippocampal layer selected was first delineated for counting using the tracing function of the Stereo Investigator. Following this, the optical fractionator component was activated, and the number and location of counting frames and the counting depth for that section was determined by entering parameters such as the grid size (150 × 150 μm), the thickness of guard zone (4 μm) and the optical dissector height (i.e. 8 μm). A computer-driven motorized stage then allowed the section to be analyzed at each of the counting frame locations. In every counting frame location, the top of the section was set, after which the plane of the focus was moved 4 μm deeper through the section (guard zone) to prevent counting inaccuracies due to uneven section surface. This plane served as the first point of the counting process. Any GAD-67/NeuN positive cells that came into focus in the next 8 μm section thickness were counted if they were entirely within the counting frame or touching the upper or right side of the counting frame. Thus, all GAD-67/NeuN positive cells that were present in the middle 8-μm section depths were counted in every section. Based on the above parameters and cell counts, the Stereo Investigator program calculated the total number of GAD-67/NeuN positive cells per selected hippocampal layer by utilizing the optical fractionator formula, N = 1/ssf.1/asf.1/hsf.EQ– (Dorph-Petersen et al. 2001). The abbreviation ssf represents the section sampling fraction, which is 10 for GAD-67 counting and 20 for NeuN counting; asf symbolizes the area sampling fraction, which is calculated by dividing the area sampled with the total area of the layer; hsf stands for the height sampling fraction, which is calculated by dividing the height sampled (i.e. 8 μm in this study) with the section thickness at the time of analysis (i.e. 16 μm); EQ– denotes the total count of particles sampled for the entire layer.
An average of 17–18 serial sections were measured in each animal for counting of GAD-67 immunopositive interneurons and an average of eight sections were measured for counting of NeuN positive neurons. An option in the Stereo Investigator program allowed the experimenter to remain unaware of the running cell count totals until all sections for each animal were completed. The value for different layers and subfields were calculated separately for every animal before calculating the mean and standard errors for the three age groups. The data between adult (n = 5), middle-aged (n = 5), and aged (n = 5) animals were compared using a one-way analysis of variance (anova) with a student-Newman-Keuls multiple comparisons post hoc test. All data are presented as means ± standard errors (SEM).
Calculation of ratios of GAD-67-positive interneurons to NeuN-positive interneurons in the anterior and posterior regions of the hippocampus
To ascertain changes in the ratio of GAD-67 immunopositive interneurons to NeuN immunopositive interneurons in the strata oriens and radiatum of CA1 and CA3 subfields along the antero-posterior axis of the hippocampus, we calculated GAD-67: NeuN ratios for these layers in both anterior and posterior halves of the hippocampus. The ratio for the anterior half of each layer was calculated in every animal by determining the absolute number of GAD-67 and NeuN immunopositive cells present in the anterior half of the respective layer. The anterior half of the hippocampus in this study comprised the nine-most anterior GAD-67 immunostained sections (every 10th) and the four-most anterior NeuN immunostained sections (every 20th). The ratio for the posterior half of each layer was calculated in every animal by determining the absolute number of GAD-67 and NeuN immunopositive cells present in the posterior half of the respective layer. The posterior half of the hippocampus in this study comprised 8–9-most posterior GAD-67 immunostained sections (every 10th) and the four-most posterior NeuN immunostained sections (every 20th).
Morphology and distribution of GAD-67 immunoreactive elements in the adult, middle-aged and the aged hippocampus
In all age groups, immunostaining of hippocampal sections with GAD-67 antibody revealed clear interneurons in all layers and subfields of the hippocampus (Fig. 1). The soma of pyramidal cells in the stratum pyramidale of CA1 and CA3 subfields and granule cells in the granule cell layer of the dentate gyrus clearly lacked GAD-67 expression, but GAD-67 immunoreactive axon terminals were clearly visible in these cell layers. In addition, GAD-67 immunoreactivity was seen in the axons of dentate granule cells (mossy fibers) throughout the stratum lucidum of CA3 region in all age groups (Fig. 1), except for a few from the aged group. Thus, the overall pattern of GAD-67 immunostaining was comparable across the three age groups.
However, a closer examination of the distribution of GAD-67 immunopositive interneurons in different subfields clearly revealed decreased density of GAD-67 immunopositive interneurons as a function of age (Figs 2, 3, and 4). In the middle-aged and the aged dentate gyrus, the reduction in interneuron density was particularly striking at the junction of the dentate hilus and the granule cell layer (i.e. the region of basket cells; Fig. 2). In CA1 and CA3 regions of the middle-aged and aged hippocampus, reductions were particularly obvious in the stratum radiatum (Figs 3 and 4). The intensity of GAD-67 immunostaining within a fraction of interneurons also appeared diminished in the aged group. Furthermore, the immunoreactivity for GAD-67 was prominently seen in both soma and dendrites of interneurons in the adult group. In contrast, in the middle-aged and aged groups, a fraction of interneurons expressed GAD-67 in only the soma (Figs 2 and 4).
The absolute number of GAD-67 immunopositive interneurons was quantified in the dentate gyrus and CA1 and CA3 subfields using serial sections and an unbiased cell counting method, the optical fractionator. In the dentate gyrus, absolute numbers were calculated separately for the dentate hilus plus granule cell layer and the molecular layer. In CA1 and CA3 subfields, absolute numbers of interneurons were measured separately for three distinct layers: the strata oriens, pyramidale and radiatum. In each subfield, cumulative interneuron counts from different layers were expressed as the number per subfield, and cumulative interneuron counts from the three subfields (dentate gyrus and CA1 and CA3 subfields) were expressed as the absolute number for the entire hippocampus. To determine the layers and regions where GAD-67-positive interneuron numbers in the hippocampus decrease significantly as a function of age, the absolute numbers of interneurons for different layers, subfields, and the entire hippocampus were compared across the adult (n = 5), middle-aged (n = 5), and aged rats (n = 5) using anova with Student-Newman-Keuls multiple comparison post-hoc test.
Age-related changes in the number of GAD-67-positive interneurons in different layers of the hippocampus
In the dentate gyrus, in comparison with the adult hippocampus, the number of GAD-67 immunopositive interneurons within the DH and the GCL exhibited 41% decrease (p < 0.001) in the middle-aged hippocampus and 50% reduction (p < 0.001) in the aged hippocampus (Fig. 5a). The interneurons in the molecular layer exhibited 46% reduction (p < 0.001) in the middle-aged group and this level of reduction persisted in the aged group (Fig. 5a). In the CA1 subfield, in comparison with the adult hippocampus, the age-related reductions were significant in all three strata of middle-aged and aged animals (Fig. 5b). The reductions varied from 34 to 37% (p < 0.01) in the stratum oriens,35–48% (p < 0.01) in the stratum pyramidale, and 28–35% (p < 0.01) in the stratum radiatum (Fig. 5b). Furthermore, in all of the above three layers, no significant differences in cell number existed between the middle-aged and aged groups (p > 0.05; Fig. 5b). An age-related decrease in the number of GAD-67 immunopositive interneurons was also observed in each of the three layers of the CA3 subfield (Fig. 5c). The stratum oriens exhibited 49% decrease (p < 0.01) in the middle-aged group and 52% decrease (p < 0.01) in the aged group (Fig. 5c). The decrease was 42–45% (p < 0.01) in the stratum pyramidale, and 42–54% (p < 0.01) in the stratum radiatum. As observed for both dentate gyrus and the CA1 subfield earlier, cell counts in all three layers of the CA3 subfield were comparable between middle-aged and aged groups (Fig. 5c). Thus, most of the decrease in the number of GAD-67 immunopositive interneurons in different hippocampal layers occurs by middle-age and persists close to this level at old age, implying that the number of GAD-67 immunopositive interneurons remains mostly unchanged between middle-age and old age.
Age-related alterations in the absolute number of GAD-67 immunopositive interneurons within the entire dentate gyrus, CA1 and CA3 subfields, and the whole hippocampus
In order to measure the extent of overall age-related reductions in the number of GAD-67 immunopositive interneurons in the three subfields of the hippocampus, in every animal the GAD-67 cell counts from different layers of each subfield were combined to produce GAD-67 interneuron cell number for the entire dentate gyrus, and the CA1 and CA3 subfields (Fig. 6a). The dentate gyrus in the adult hippocampus comprised an average of 55 699 (Mean ± SEM = 55 699 ± 3183) GAD-67 immunopositive interneurons. This number decreased to 31 752 ± 1996 by middle age and 28 829 ± 1245 by old age (43–48% reduction, p < 0.001, F ratio = 41.6; Fig. 6a). The adult CA1 region contained 73 915 ± 5792 GAD-67 immunopositive interneurons, which decreased to 50 119 ± 4143 interneurons by middle age and 45 122 ± 4017 interneurons by old age (32–39% reduction, p < 0.01, F ratio = 10.6; Fig. 6a). In the adult CA3 region, 43 546 ± 3358 GAD-67 immunopositive interneurons were found. This number decreased to 24 516 ± 731 in the middle-aged group and 21 456 ± 970 in the aged group (44–51% decrease, p < 0.01, F ratio = 33.7; Fig. 6a). However, within all subfields of the hippocampus, differences in GAD-67 immunopositive interneuron numbers between middle-aged and aged groups were insignificant, consistent with the results described earlier for different layers of the hippocampus. To ascertain the total number of GAD-67 immunopositive interneurons for the entire hippocampus, in each animal, the GAD-67 cell counts from the dentate gyrus and the CA1 and CA3 subfields were combined (Fig. 6b). This measurement revealed that the entire hippocampus contains an average of 173 160 GAD-67 immunopositive interneurons during adult age, 106 387 GAD-67 immunopositive interneurons during middle age (i.e. 39% reduction, p < 0.001) and 95 407 GAD-67 immunopositive interneurons during old age (i.e. 45% reduction compared with the adult hippocampus, p < 0.001). This suggests that the overall number of GAD-67 immunopositive interneurons in the hippocampus decreases considerably between adulthood and middle age but exhibits no significant decrease thereafter.
Age-related changes in the number of NeuN-positive interneurons in the strata oriens and radiatum of CA1 and CA3 subfields
Immunostaining with NeuN antibody clearly revealed sparsely distributed interneurons in strata oriens and radiatum of CA1 and CA3 subfields, in addition to principal hippocampal cell layers (strata pyramidale of CA1 and CA3), the dentate granule cell layer and neurons in the dentate hilus (Fig. 7). The pattern of NeuN staining in different layers of the hippocampus appeared mostly comparable across the three age groups (Fig. 8). To ascertain whether the widespread reductions observed in the number of GAD-67 immunopositive neurons reflects actual cell death of a major fraction of interneurons or just a down-regulation of GAD-67 protein expression, we performed parallel counts of interneurons in strata oriens and radiatum of CA1 and CA3 subfields using NeuN immunostained serial sections and the optical fractionator cell counting method. Estimation of the NeuN immunostained interneurons was not performed in the dentate hilus, dentate granule cell layer, and stratum pyramidale of CA1 and CA3, as interneurons cannot be differentiated from other (mostly excitatory) neurons in these layers using NeuN immunostained sections. This analysis demonstrated that the absolute number of interneurons in strata oriens and radiatum of CA1 and CA3 subfields in the middle-aged and aged hippocampus is highly comparable with the adult hippocampus (p > 0.05; Fig. 9). Thus, widespread reductions observed in GAD-67 immunopositive interneurons as a function of aging largely reflect severe down-regulation of GAD-67 protein expression in a major fraction of interneurons in the middle-aged and the aged hippocampus.
Age-related changes in the ratio of GAD-67 immunopositive and NeuN-positive interneurons
We determined the ratio of GAD-67 immunopositive and NeuN positive interneurons in anterior and posterior segments of the hippocampus for strata oriens and radiatum of CA1 and CA3 subfields (Table 1). This analysis revealed that, in the anterior segment of the hippocampus, the ratio of GAD-67 immunopositive and NeuN positive interneurons in the adult hippocampus is 0.5. In the middle-aged and aged hippocampus, this ratio declines to 0.3. In the posterior segment of the hippocampus, the ratio is 0.31 in the adult hippocampus, 0.21 in the middle-aged hippocampus and 0.16 in the aged hippocampus. Thus, the expression of GAD-67 in NeuN-positive interneurons diminishes considerably in both anterior and posterior regions of the hippocampus from adult to middle age but undergoes no significant change thereafter.
Table 1. Ratio of glutamate decarboxylase-67 (GAD-67) positive and neuron-specific nuclear antigen (NeuN)-positive interneurons in anterior and posterior regions of adult, middle-aged and aged hippocampus *
Anterior region (mean ± SEM)
Posterior region (mean ± SEM)
Values for each group represent averages of values from five different animals. In comparison with the adult animals, the GAD-67 and NeuN ratio in the anterior region of the hippocampus exhibited 42% reduction in the middle-aged animals ( p < 0.01) and 37% reduction in the aged animals ( p < 0.01). In the posterior region of the hippocampus, the ratio demonstrated 32% decline in the middle-aged animals ( p < 0.05) and 48% reduction in the aged animals ( p < 0.01).
0.49 ± 0.04
0.31 ± 0.04
0.28 ± 0.02
0.21 ± 0.02
0.31 ± 0.05
0.16 ± 0.01
This quantitative study on hippocampal interneurons provides three novel findings. First, aging in the rat hippocampus is clearly coupled with widespread reductions in the absolute number of GAD-67 immunopositive interneurons in all layers and subfields. This finding is consistent with the decreased density of GAD-67 immunopositive interneurons observed earlier in the aged septal hippocampus (Shetty and Turner 1998). Secondly, most of the age-related declines in GAD-67 immunopositive interneuron numbers in the hippocampus emerge by middle age. Thirdly, age-related reductions in hippocampal GAD-67 immunopositive interneuron numbers are due to a loss or diminution of GAD-67 protein expression in a sizable fraction of interneurons. This is because age-related declines in the number of GAD-67 immunopositive interneurons did not appear to be due to interneuron degeneration, at least in the strata oriens and radiatum of CA1 and CA3 subfields, the layers in which potential GAD-67 immunopositive interneuron degeneration was quantitatively ascertained using parallel NeuN cell counts.
Magnitude and basis of age-related reductions in the number of GAD-67 immunopositive interneurons
This study represents the first report of complete GAD-67 immunopositive (GABA-ergic) interneuron counts in the hippocampus as a function of aging. The results underscore that the absolute number of interneurons expressing GAD-67 protein substantially decreases in all layers and subfields of the hippocampus with aging. The extent of reductions within different layers of the middle-aged and aged hippocampus varied from 41 to 50% in the dentate gyrus, 28–48% in the CA1 subfield, and 42–54% in the CA3 subfield. These results are consistent with the recent observation that aging leads to a decreased expression of GAD-67 mRNA in the hippocampus (Vela et al. 2003). The age-related reductions in the absolute number of GAD-67 immunopositive interneurons could signify either degeneration of a considerable number of interneurons throughout the hippocampus or a prevalent down-regulation of the GAD-67 protein expression with conservation of interneurons or both of these mechanisms. We ascertained these possibilities by measuring the absolute number of interneurons positive for NeuN within the strata oriens and radiatum of CA1 and CA3 subfields, which revealed no changes in the absolute number of interneurons as a function of age in these layers. Collectively, these correlative observations imply that reductions in the number of GAD-67 immunopositive hippocampal interneurons observed with aging largely reflect a drastic down-regulation or loss of GAD-67 protein expression within a substantial fraction of interneurons in the middle-aged and aged hippocampus. A significant decline observed in the ratio of GAD-67-immunopositive and NeuN-positive interneurons in both anterior and posterior regions of strata oriens and radiatum of CA1 and CA3 subfields as a function of age also corroborates the above conclusion. Down-regulation of GAD-67 protein expression without affecting interneuron survival has been observed under certain conditions in many areas of the brain, including the hippocampus. For instance, removal of afferents to the mouse somatosensory cortex leads to transient down-regulation of GAD-67 protein in cortical interneurons (Gierdalski et al. 1999). Additionally, in animal models of temporal lobe epilepsy, a substantial number of interneurons exhibit permanent down-regulation of GAD-67 protein expression leading to a significantly less postsynaptic inhibition and hyperexcitability in the lesioned hippocampus (Obenaus et al. 1993; Esclapez et al. 1997; Shetty and Turner 2000, 2001). Thus, aging in the hippocampus does not appear to be associated with degeneration of interneurons, but is linked with a reduced number of interneurons that synthesize GABA. In this context, it is plausible that increased excitability of principal excitatory neurons in the aged hippocampus is a result of decline in the amount of the inhibitor neurotransmitter GABA.
Potential mechanisms of loss of GAD-67 protein in hippocampal interneurons with aging
The mechanisms of the loss of GAD-67 protein expression in a considerable fraction of interneurons in the middle-aged and aged hippocampus are unclear. However, this could be due to age-related reductions in the afferent trophic support and synaptic input from hippocampal pyramidal neurons, as hippocampal pyramidal neurons provide both trophic support and direct afferent innervation to multiple populations of interneurons in the adult hippocampus (Buzsaki 1984; Scharfman 1994; Freund and Buzsaki 1996). Furthermore, GABA-ergic interneurons in the hippocampus are dependent on brain-derived neurotrophic factor (BDNF) for proper functioning (Jones et al. 1994; Marty et al. 1996) but are incapable of synthesizing BDNF, and hence rely on principal hippocampal neurons for a steady supply (Rocomora et al. 1996; Conner et al. 1997). Additionally, the above contention is supported by the following findings in earlier studies. First, aging in the hippocampus is associated with substantial declines in the synthesis of neurotrophic factors, particularly the BDNF (Smith 1996; Smith and Cizza 1996; A. K. Shetty, unpublished data). Secondly, although aging in the hippocampus is not characterized by significant pyramidal neuron loss, a substantial reduction in the number of synapses is one of the hallmarks of aging (Cotman and Anderson 1988; Rapp and Gallagher 1996; Rasmussen et al. 1996; Poe et al. 2000; Rosenzweig and Barnes 2003). Thus, both decreased BDNF support and reduced direct feed-forward afferent input (due to loss of synapses) from hippocampal pyramidal neurons to interneurons during the course of aging are likely among the factors causing the loss of GAD-67 expression in a significant fraction of hippocampal interneurons with aging.
Fate of interneurons that lack GAD-67 in the aging hippocampus
Comparison of the numbers of NeuN immunopositive interneurons in the strata oriens and radiatum of CA1 and CA3 subfields between the adult hippocampus and the middle-aged and the aged hippocampus suggest that interneurons that lose GAD-67 during the course of aging continue to survive. However, it is unclear whether these GAD-67-deficient interneurons retain some of their major function (i.e. inhibition of the principal hippocampal neurons). Additionally, interesting issues for future studies are to determine whether or not these GAD-67-deficient interneurons in the aging hippocampus: (i) express other neurotransmitters or neuromodulators including calcium binding proteins and neuropeptides, (ii) lose all or some roles they might play on hippocampal circuitry, and (iii) exhibit dendritic regression and axonal retraction. In view of the previous studies showing that hippocampal interneurons positive for calcium binding proteins calbindin and parvalbumin also exhibit reductions in number with aging (Potier et al. 1994; Shetty and Turner 1998), it is likely that a greater fraction of GAD-67 deficient interneurons in the aging hippocampus lack other key proteins as well. Combined physiological and anatomical analyses of GAD-67 deficient interneurons in the aging hippocampus in future studies could address the above interesting issues.
Implications of widespread loss of GAD-67 protein expression in GABA-ergic interneurons
From the results of this study, the reduced functional inhibition observed earlier in the aged hippocampus appears to be related to a decline in the absolute number of functional GABA-ergic interneurons. As GABA-ergic interneurons function by stabilizing excitatory influences and synchronizing principal excitatory neuron populations, reduced functional inhibition may increase the vulnerability of the aged hippocampus to hyperexcitability under certain conditions (Turner 1990; Potier et al. 1993; Barnes 1994; Cobb et al. 1995; Freund and Buzsaki 1996). Indeed, studies suggest that pyramidal neurons in the aging hippocampus exhibit higher frequency bursting (comprising sharp-wave/ripple complexes) during large irregular activity (Smith et al. 2000), and memory-impaired aged rats fail to demonstrate silent cells in the hippocampus in any environment (Tanila et al. 1997). Physiological reports on age-dependent changes in neuronal excitability are contentious owing to disparities in recording procedures, animal ages, strains, and behavioral status (Barnes 1994). Assessment of CA1 pyramidal neurons demonstrated a decrease in action potential threshold in unimpaired or untested animals (Landfield et al. 1986; Barnes et al. 1987; Kerr et al. 1991), an increase in neuronal excitability from middle age to senescence (Papatheodoropoulos and Kostopoulos 1996), and an increase in action potential threshold in impaired aged animals (Turner and Deupree 1991; Moyer et al. 1992; Potier et al. 1992, 1993). In addition, both deficits in the induction and maintenance of long-term potentiation (Rosenzweig and Barnes 2003) and reductions in GABA-ergic interneuron-mediated inhibitory postsynaptic potential (Billard et al. 1995) have been observed in the aged hippocampus.
Thus, the overall changes in hippocampal neuronal excitability in aged animals likely represent multiple age-related physiological alterations in different conditions (Rosenzweig and Barnes 2003). Nonetheless, the increased excitability of pyramidal neurons in the aged hippocampus under certain conditions is likely linked to the substantial age-related reductions in the number of GABA-synthesizing interneurons, as observed in this study. Absence of reductions in the postsynaptic GABAA and GABAB receptors in the aging hippocampus also supports the above conclusion (Potier et al. 1992, 1994; Billard et al. 1995; Gutierrez et al. 1996). Additionally, the current study underscores that the age-related decreases in the number of functional GABA-ergic interneurons emerge as early as the middle age. Given the potential importance of disinhibition in epileptogenesis, it is possible that increased vulnerability of the aging hippocampus to excitotoxins and to epileptic seizures after injury (Dawson and Wallace 1992; Eisenschenk and Gilmore 1999; Shetty and Turner 1999a,b; Kerr et al. 2002) is linked to reductions in the number of functional GABA-ergic interneurons. Furthermore, the age-related decline in the functional GABA-ergic interneuron numbers in the aging hippocampus could contribute significantly to impairments in both memory acquisition and spatial tuning of hippocampal place cells observed in aged animals (Tanila et al. 1997; Yan et al. 2003), as firing properties of place cells are known to be associated with the activity of hippocampal interneurons (Mizumori et al. 1989; Wilson and McNaughton 1993).
Prospective strategies for recovering the functional GABA-ergic interneuron numbers in the aging hippocampus
The finding that age-related loss of GAD-67 protein expression within a sizable fraction of hippocampal interneurons does not appear to be associated with interneuron degeneration supports the hypothesis that the ‘skeleton’ or the structural basis of the local hippocampal inhibitory network largely remains intact during aging, particularly interneuron soma and their major efferent projections onto principal cells. In this context, the dormant basket cell hypothesis proposed by Sloviter for epilepsy models is also relevant for the aging hippocampus (Sloviter 1991; Bernard et al. 1998). This hypothesis suggests that loss of critical excitatory afferent inputs from pyramidal neurons to interneurons makes interneurons hypofunctional. The hypofunctional status of interneurons in the aging hippocampus is reflected in this study by the loss of GAD-67 in a substantial fraction of interneurons. Nevertheless, the decline observed in the number of functional GABA-ergic interneurons during the course of aging in the hippocampus appears amenable to treatment strategies that have the potential to induce GABA-expression in GABA-deficient interneurons. For example, in a kainate model of temporal lobe epilepsy, transplantation of fetal hippocampal CA3 pyramidal neurons leads to normalization in the number of functional GABA-ergic interneurons within both dentate gyrus and the CA1 subfield, presumably through re-innervation of GABA deficient interneurons by axons of grafted CA3 pyramidal neurons (Shetty and Turner 2000, 2001). In the same vein, grafting of fresh embryonic hippocampal CA1 and CA3 pyramidal neurons into the aging hippocampus might normalize the number of functional GABA-ergic interneurons in the aging hippocampus by providing both trophic support and additional feed-forward afferent input to GABA-deficient interneurons.
We provide novel evidence that the GAD-67 immunopositive interneurons in all layers and subfields of the hippocampus undergo reduction in total number by middle age, but no significant change thereafter. Fascinatingly, these changes in GAD-67 immunopositive interneuron numbers do not appear to be due to interneuron degeneration. This was evidenced by the analogous number of NeuN positive interneurons in the strata oriens and radiatum of CA1 and CA3 subfields across the three age groups. Thus, age-related diminutions in the number of GAD-67 immunopositive interneurons in the hippocampus largely reflect loss of GAD-67 protein expression in interneurons. Age-related decline in the functional GABA-ergic interneuron numbers may underlie some of the hippocampus-related behavioral alterations observed in aged animals, particularly higher frequency bursting during large irregular activity, increased vulnerability to epileptic seizures after injury or exposure to excitotoxins, and impairments in both memory acquisition and spatial tuning of hippocampal place cells.
This research was supported by grants from the Department of Veterans Affairs (VA Merit Review Award to AKS) and National Institutes of Health (National Institute of Aging grant RO1AG20924 to AKS). We thank Dr Vandana Zaman for her excellent contribution to tissue processing and immunostaining in this study.