The effect of age and sex on the expression of GABA signaling components in the human hippocampus and entorhinal cortex

Gamma-aminobutyric acid (GABA) is the primary inhibitory neurotransmitter in the nervous system. The GABA signaling system in the brain is comprised of GABA synthesizing enzymes, transporters, GABAA and GABAB receptors (GABAAR and GABABR). Alterations in the expression of these signaling components have been observed in several brain regions throughout aging and between sexes in various animal models. The hippocampus is the memory centre of the brain and is impaired in several age-related disorders. It is composed of two main regions: the Cornu Ammonis (CA1-4) and the Dentate Gyrus (DG), which are interconnected with the Entorhinal Cortex (ECx). The age- and sex-specific changes of GABA signaling components in these regions of the human brain have not been examined. This study is the first to determine the effect of age and sex on the expression of GABA signaling components-GABAAR α1,2,3,5, β1-3, γ2, GABABR R1 and R2 subunits and the GABA synthesizing enzymes GAD 65/67-in the ECx, and the CA1 and DG regions of the human hippocampus using Western blotting. No significant differences were found in GABAAR α1,2,3,5, β1-3, γ2, GABABR R1 and R2 subunit and GAD65/76 expression levels in the ECx, CA1 and DG regions between the younger and older age groups for both sexes. However, we observed a significant negative correlation between age and GABAAR α1subunit level in the CA1 region for females; significant negative correlation between age and GABAAR β1, β3 and γ2 subunit expression in the DG region for males. In females a significant positive correlation was found between age and GABAAR γ2 subunit expression in the ECx and GABABR R2 subunit expression in the CA1 region. The results indicate that age and sex do not affect the expression of GAD 65/67. In conclusion, our results show age- and sex-related GABAA/BR subunit alterations in the ECx and hippocampus that might significantly influence GABAergic neurotransmission and underlie disease susceptibility and progression.

www.nature.com/scientificreports/ relatively well-preserved during aging. No significant differences were found in protein levels in the ECx, CA1 and DG regions between the younger and older age groups for both sexes (Figs. 1,2,3,4,5,. However, we observed significant correlations between age and signal intensity for a few GABA signaling components. We detected a significant negative correlation between age and GABA A R α1 subunit expression level in the CA1 region of females ( Fig. 1h) (r = −0.695, P = 0.044). In the DG, there were significant negative correlations between age and expression of GABA A R β1 (Fig. 2l) (r = −0.682, P = 0.012), β3 ( Fig. 3l) (r = −0.559, P = 0.05) and γ2 subunits (Fig. 4l) (r = −0.616, P = 0.028) for males, respectively. The difference in GABA A R γ2 subunit expression level in the ECx between younger females and older females was not statistically significant ( Fig. 4d) but showed a trend towards increase in older females. Accordingly, in the ECx we found a significant strong positive correlation between age and expression level of the GABA A R γ2 subunit in females (Fig. 4g) (r = 0.793, P = 0.008). A significant positive correlation between age and expression level of GABA B R R2 subunit was observed in the CA1 region of females (Fig. 5 h) (r = 0.664, P = 0.041). There were no significant age-specific differences in the expression of the GAD 65 (Suppl Fig. 6d,e,f) and GAD 67 enzymes (Suppl Fig. 7d,e,f) between the younger and older age groups. Furthermore, no significant correlations were detected between age and expression level of GAD 65 (Suppl Fig. 6 g-l) or GAD 67 (Suppl Fig. 6 g-l) enzymes in the DG, CA1 region or ECx.
The GABA signaling components examined showed similar expression between females and males and younger and older age groups (Figs. 1, 2, 3, 4, 5, Suppl Figs. 1-7) applying a linear mixed model (Suppl Tables 3-4). This analysis, that takes into consideration the interactions between the brain regions, identified GABA A R α1 subunit expression in the CA1 region was significantly different from the DG region (P = 0.008); GABA A R α3 subunit expression in the ECx region was significantly different from the CA1 region (P < 0.001) and the DG (P = 0.029); GABA A R α5 subunit expression in the ECx region was significantly different from the CA1 region (P < 0.001) and the DG (P = 0.039), furthermore the GABA A R α5 subunit expression in the DG region was significantly different from the CA1 region (P = 0.019); GABA A R β3 subunit expression in the CA1 region was significantly different from the ECx region (P < 0.001) and the DG (P < 0.001); GABA A R γ2 subunit expression in the CA1 region was significantly different from the ECx region (P = 0.039) and the DG (P = 0.009), and the GABA A R γ2 subunit expression in the DG region was significantly different from the ECx region (P < 0.001); GABA B R R1 subunit expression in the CA1 region was significantly different from the DG region (P < 0.001) (Suppl Tables 3-4).

Discussion
In this study, we have investigated the effect of age and sex on the expression of GABA A R subunits, GABA B R subunits, and GABA synthesizing enzymes in the human hippocampus and ECx. We report a significant negative correlation between age and GABA A R α1subunit level in the CA1 region for females; a significant negative correlation between age and GABA A R β1, β3, and γ2 subunit expression in the DG region for males. In females a significant positive correlation was found between age and GABA A R γ2 subunit expression in the ECx and GABA B R R2 subunit expression in the CA1 region. However, the results indicate that age and sex do not affect the expression of GAD 65/67. Given the lack of human studies in the literature, the current study provides an important first glimpse into potential age-and sex-related GABA receptor subunit expression changes occurring in the human hippocampus.
In the ECx we found a significant strong positive correlation between age and expression level of the GABA A R γ2 subunit in females. The sex-related GABA A R γ2 subunit expression increase with age in females might be the result of underlying hormonal differences between sexes. Weiland and Orchinik 27 showed an increase in GABA A R γ2 subunit mRNA levels in the CA1, CA2, and CA3 regions in adult female rats after administration of progesterone via injection. However, in the context of the current study, evidence has shown that there is a decrease in progesterone level with age, with a 30-50% decrease in post-menopausal women and a 35% decrease in estrogen levels 38 . Given that the mean age of the older female group (77.2 ± 3.96 years), the females in the group would be expected to have a reduction of progesterone and estrogen 39 . However, there is no definite evidence that hormonal differences could lead to the sex-specific differences in GABA A R γ2 subunit expression observed in this study. A single progesterone treatment might affect the expression of the GABA A R γ2 subunit on a different way compared to a long-term reduction in hormonal levels. Additionally, in the in vivo human multiple hormones are acting at any one time.
We also observed a significant strong negative correlation between age and GABA A R γ2 subunit expression in the DG of males. In females a significant positive correlation was found between age and GABA A R γ2 subunit expression in the ECx, suggesting that these age-related correlations are sex-specific. We found no significant agerelated GABA A R γ2 subunit level changes in the hippocampus or ECx between the younger and older groups and our recent mouse study shows no age-related alteration when comparing hippocampal (CA1, CA2/3, and DG) γ2 subunit levels in 6 months old (young) and a 21 months (old) male mice 11 . One study found an increase in γ2 subunit expression with age in the rat hippocampus 13 , while others showed an age-related decline in γ2 subunit expression in the mouse hippocampus 14 . However, these studies did not differentiate between hippocampal subregions, sex of the animals, and the rat study examined two age groups only, making it difficult to compare the findings with our results and draw any conclusions about the effect of sex.
Given the role of GABA as the primary inhibitory neurotransmitter in the brain, GABAergic dysfunction has been implicated as one of the factors in epileptic seizure incidences 34,40,41 . GABAergic signaling components are also affected in the epileptic brain, particularly GABA A Rs 34,42 . In rat models, treatment with GABA A R antagonists such as bicuculline and picrotoxin caused severe motor seizures (Fisher, 1989). Previous studies show the γ2 subunit is widely expressed in the hippocampus and cortical brain areas and is required for the majority of GABA A Rs assembly 11,[43][44][45][46] . Knockout of the GABA A R γ2 subunit in mice have shown a reduction in receptor clustering and decreased synaptic function, while γ2 subunit gene disruption showed a near-complete abolishment www.nature.com/scientificreports/ of all GABA A Rs to benzodiazepine site ligands 47 . The higher γ2 subunit levels observed in older females in this study indicate that GABAergic inhibition and effects of benzodiazepines might be more effective in older females in the ECx compared with younger females and both younger and older males. This may be particularly important in seizure therapy, as it has been shown that the ECx is involved in temporal lobe epilepsy (TLE), with (d, e, f) Signal intensity graphs for each group comparing GABA α1 Western Blot band was measured and normalised to their corresponding β-actin signal for each age group. The data is graphed as mean ± SEM (N = 4-7). (g, h, i, j, k, l) Correlation graphs for males and females plotting the relationship between age and signal intensity of GABA α1 bands.  48,49 . The reduced expression of γ2 subunit given the reduced volume and lower expression levels might have more severe implications in males and younger females, however, more research is required in this regard. (d, e, f) Signal intensity graphs for each group comparing GABA β1 Western Blot band was measured and normalised to their corresponding β-actin signal for each age group. The data is graphed as mean ± SEM (N = 4-7). (g, h, i, j, k, l) Correlation graphs for males and females plotting the relationship between age and signal intensity of GABA β1 bands (p * ≤ 0.05). www.nature.com/scientificreports/ We detected a significant negative correlation between age and GABA A R α1 subunit level in the CA1 region for females. Hippocampal GABA A R α1 subunit expression is high in the CA1 region 11,45,50 , and few rodent studies have shown that this expression stays at the same level 11,14 or increases 12,13 with age. However, the expression was not evaluated in females and most studies examined two age groups only. Thus, the results are not comparable (d, e, f) Signal intensity graphs for each group comparing GABA A R γ2 Western Blot band was measured and normalised to their corresponding β-actin signal for each age group. The data is graphed as mean ± SEM (N = 4-7). (g, h, i, j, k, l) Correlation graphs for males and females plotting the relationship between age and signal intensity of GABA A R γ2 bands (p * ≤ 0.001). www.nature.com/scientificreports/ to the findings in our study. These findings were also contradicted by studies that found an age-dependent decrease in the α1 subunit expression in the CA1 region of the rhesus monkey and human hippocampus using immunohistochemistry combined with densitometric analysis 51,52 . Previous studies have shown there is a significant loss of GABA A R α1 subunit expression in the CA1 and CA2 regions of human hippocampal sections of patients with temporal lobe epilepsy 53,54 . The negative correlation of age and expression of the α1 subunit in the CA1 subregion in females indicates that anti-epileptic drugs might require different dosing for older females, to achieve the best outcomes. Further significant strong negative correlations were also observed between age and GABA A R β1 and β3 subunit levels in the DG region of males. Reduction or elimination of activity at β1 subunit-containing GABA A Rs has been shown to increase the efficacy of anxiolytic benzodiazepines in rats 55 . Therefore, age-related decreases in β1 subunit expression in older males might be beneficial for the efficacy and reducing potential side-effects of particular benzodiazepine treatments for anxiety and depression. Previous rat, mouse, and nonhuman primate studies found no significant age-related changes in GABA A R β3 subunit expression in the hippocampus [11][12][13]52 . The α5β3γ2 receptor configuration is particularly high in extra-synaptic sites in the DG 56 with an important role in cognition and memory, and as a target for benzodiazepine agonists 56,57 . Reduced expression of GABA A R β3 subunit in the DG with age in males, as seen in the current study, can affect the receptor configuration, thus reducing GABAergic inhibition and benzodiazepine effectiveness. However, this is not only applicable to benzodiazepines as other drugs also target GABA A Rs containing the β3 subunit. An example is loreclezole, an anticonvulsant drug that is a positive allosteric modulator of the GABA A R, that is highly selective for receptors containing the GABA A R β2 or β3 subunits, unlike classical benzodiazepines which bind to the benzodiazepine binding site between the α and γ subunits of the GABA A R 58 . Reduced expression of the GABA A R β3 subunit with age can, therefore, affect the therapeutic ability of this and other similar drugs.
Another significant strong positive correlation between age and expression of GABA B R2 subunit was found in the CA1 region of females in our study. The current literature is limited on age-and sex-related changes in the expression of the GABA B Rs in the hippocampus. McQuail et al. 19 found no significant differences in the rat hippocampal GABA B R R2 expression and GTP binding. Interestingly, Banuelos et al. (2014) also demonstrated a significant negative correlation between GABA B R R2 subunit expression and working memory test scores, while McQuail et al. 19 found no relationship. The increased expression in GABA B R R2 subunits may provide a foundation to understanding the age-related deficits in spatial and working memory, given the findings from studies above indicating increased GABA B R R2 subunit expression resulting in a reduced working memory performance 19 . The findings from our current study are in line with the results of Liao et al. 23 which showed an increase in GABA B R R2 subunit expression in the visual cortex of aged rhesus macaque monkeys. However, Pandya et al. 6 showed no age-related changes in several human cortical areas and cerebellum, and McQuail et al. 19 and Bañuelos et al. 21 showed an age-related decrease in GABA B R R2 expression in the rat PFC. These studies demonstrate that age-related GABA B R R2 subunit expression changes are brain region-specific.
It is known that the hippocampus is involved in spatial learning, memory consolidation, and memory transfer, however, the hippocampus and the GABAergic system are also implicated in neurological disease such as epilepsy, anxiety, depression, and AD 16,34,59,60 . Women are more likely to develop anxiety, depression and dementias such as AD than men 36,60-64 . Importantly, the observed sex-specific expression of GABA A R subunits related to anxiety (β2, β3 and γ2) and memory (α1 and α5) might be linked to increased vulnerability to these and other neurological diseases and neurodegenerative conditions. GABA A R subunit expression changes have been implicated in AD 16,45,65 . For a long time, the literature has indicated that the GABAergic system remains unperturbed in AD, however, evidence indicates that there are reductions of GABA currents in human cortical cells from AD brains 66 , as well as GABAergic nerve terminal damage and reduction in GABA uptake in the AD brain 67 . Given that the characteristic symptom of AD patients is memory loss, evidence shows the hippocampus is affected severely in AD and also shows age-related molecular and cellular changes [8][9][10] . Studies have shown that the subunit density of γ2 subunit expression was preserved 68 and α1 subunit immunolabeling was increased in human AD hippocampal tissue 69 . We reported significant increases in the GABA A R α1, α2, α5, β2, and γ2 subunit expression in the different layers of CA1-3 and DG regions in the AD human hippocampus 45 . This is an overall trend except for the α1 and α2 subunits that show decreased expression of the CA1 subregion in the stratum radiatum and pyramidale, respectively 45,60 . In the current study, the expression of GABA A R subunit α1 in the CA1 and β1, β3 in the DG in females and males respectively, all show significant negative trends of expression with age. This indicates an opposite effect of normal aging on GABA A R subunit expression compared to AD. There is increasing evidence of remodeling of the GABAergic, cholinergic, and glutamatergic neurotransmitter systems in AD leading to disruption of the excitatory/inhibitory balance 16,[69][70][71][72] . Alterations in glutamate receptor and transporter expression in the hippocampus have also been observed, contributing to glutamate-mediated excitotoxicity 65,71,[72][73][74][75][76]87 . The imbalance is further driven by neuronal cell death observed in AD 10,65,77 . Given the findings of these studies, there is some credence to the idea that surviving neurons in the AD hippocampus increase GABA A R subunit synthesis to help maintain the inhibitory circuit in the hippocampus, whereas in normal aging, there is a decrease in subunit synthesis.
The results from this study observed an age-related trend toward a decrease in α5 subunit expression with age in the ECx for females that did not reach statistical significance. Findings from a rat study showed a moderate α5 subunit decrease in the hippocampus during aging 14 but other studies did not detect age-related changes in the mouse and rat hippocampus 11,13 . In AD, increased expression of the GABA A R α5 subunit is found in the hippocampal CA1 subregion and subiculum of AD cases compared to healthy controls 45,50,60 . This is in line with findings showing mild memory and cognitive impairment in normal aging with increased GABA A R α5 subunit expression 78 . Therefore, it is plausible that the increased GABA A R α5 subunit expression is linked to negative cognitive and memory impairments in age-related disease conditions 79  www.nature.com/scientificreports/ lesion and AD models [78][79][80][81][82] . Therefore, an age-related decrease in α5 subunit expression might be compensatory, as it may contribute to maintaining a normal excitatory/inhibitory balance, long-term potentiation, and cognition. There was no significant age-or sex-specific differences in the expression of GABA A R α2 and α3 subunits in the current study. Rodent studies have shown that both the GABA A R α2 and α3 subunits decrease with age in the somatosensory and visual cortices 14,22,83,84 . However, a comprehensive mouse study by Palpagama et al. 11 did not find any age-related expression changes of the aforementioned subunits, which is in agreement with the findings of the current human study. In the human STG GABA A R α3 subunit expression is significantly lower in older males 6 but the subunit levels in most other cortical regions are not affected by age. Given that age-related decreases of the α2 and α3 subunits was only observed in sensory cortices, this could indicate that age-related visual and auditory impairment in the periphery might drive these changes.
Western blotting provides a robust quantitative analysis, but the limitation of this technique is that we were not able to examine the variations in expression across different cell types of neural circuits within hippocampal subregions and ECx. Further studies are required to examine such differences within individual cell types as these could have a significant functional consequence in terms of network activity in these brain regions. These studies will also help to validate our findings. This is important as there is a possibility of false positive and negative errors. While we applied strict case selection criteria and tissue processing and experiments were performed at the highest possible standards the variability of data is relatively high. The study would certainly benefit from more samples and samples that are more evenly spaced by age. However, the availability of human tissue is very limited, therefore minimizing the variance is challenging. In addition, further studies will be required to understand the functional implications of these age-related and sex-specific GABA A R subunit expression changes.

Conclusions
The current knowledge of GABAergic age-related and sex-specific differences across the hippocampus and surrounding cortices is limited. Our study shows that the GABAergic system in the hippocampus and ECx is relatively robust to age-related changes. However, the observed sex-specific negative correlation of GABA A α1, β1, β3, and γ2 subunit expression with age, particularly in the DG and CA1 subregion, and the positive correlation of the γ2 and GABA B R2 subunit expression with age in the ECx and CA1 subregion, respectively, might significantly influence the function of the receptors and affect GABAergic inhibition within the human hippocampus. As discussed above the GABAergic system is implicated in the development and progression of several neurological disorders, therefore important to consider sex-and age-specific differences in the expression of GABA signaling components when designing new therapeutics or improving current treatments. Age-related functional changes such as cognitive decline, depression, and increased risk of neurodegenerative disease are becoming more of a pressing issue given the growing elderly population. Understanding the mechanisms involved in aging of a critically important neurotransmitter system, the GABAergic system, will help provide better understanding of pathological changes that might be accelerated by age. Further research will be important to shed light on the implications of age-and sex-related GABAergic alterations in disease conditions, to design improved and more effective personalized treatments.

Methods
Human brain tissue preparation and neuropathological analysis. The study was conducted at the University of Auckland, Centre for Brain Research. The tissue was acquired through a donation program to the Neurological Foundation Human Brain Bank. The procedures were approved by the University of Auckland Human Participants' Ethics Committee. All experiments were performed in accordance with relevant guidelines and regulations. Processing of tissue followed the procedure described previously 85 . The brain was dissected to separate the hemispheres, with the left hemisphere cut into anatomical blocks and freshly frozen and stored at −80 °C. Standard sections, including the middle frontal-, middle temporal-, cingulate gyrus, hippocampus, caudate nucleus, substantia nigra, locus coeruleus, cerebellum from all cases were examined by a neuropathologist. All cases included in this study had no history of any primary neurodegenerative, psychiatric disorder, neurological disease abnormalities, or excessive alcohol consumption (Suppl Table 1). None of our records show that that any of the female cases were on hormone replacement therapy. The cases were sorted into four groupsyounger female (YF-53.2 years ± 11.9; N = 6), older female (OF 77.2 years ± 3.96; N = 6), younger male (YM-49.5 years ± 6.5; N = 6), older male (OM-78.14 years ± 6.47; N = 7). All effort was made to have the largest possible age gap between the younger and older age groups.
Western blotting. The human tissue blocks were cut using a cryostat (CM3050, Leica Microsystems, Germany) at 60-µm thickness and collected on glass slides. Hippocampal areas of interest-DG, CA1 and ECx-were collected with a blade into sterile 1.7 ml tubes for each area of interest. The tissue was homogenised in a buffer containing 0.5 M Tris, 100 mM EDTA, 4% SDS, pH 6.8, supplemented with 100 mM phenylmethanesulfonyl fluoride (Sigma, St. Louis, MO, USA) and 0.5 mm glass beads (Mo-Bio Laboratories, Solano Beach, CA, USA) in a Mini Bullet Blender Tissue Homogenizer (Next Advance, Inc., NY, USA) at speed 8 for 8 min. The homogenates were incubated for 1 h on ice, centrifuged at 10621 g for 10 min. The resulting supernatant collected and stored at −20 °C. The protein concentration of each sample was determined using detergent-compatible protein assay (DC Protein Assay, 500-0116, Bio-Rad, Hercules, CA, USA) as per manufacturer instructions.
Protein samples for each case were numbered from 1 to 25 and randomized. The experimenter was blinded to avoid any potential bias during the experiment, image acquisition and analysis.
Twenty μg of each protein extract was run on a gradient polyacrylamide electrophoresis gel (NU PAGE 4-12% BT 1.5, NP0336BOX; Life Technologies, Carlsbad, CA, USA) and then blotted using the Thermofisher XCell Blot Module (Thermofisher, CA, USA), which was then transferred onto nitrocellulose membranes www.nature.com/scientificreports/ (Amersham-Protran, GE Healthcare, Germany) for immunolabeling. The gels were also loaded with three molecular weight ladders to verify labeled band sizes: MagicMark, SeeBlue and Molecular Weight (Invitrogen, CA, USA). Membranes were blocked with Odyssey blocking buffer (LI-COR Biosciences, NE, USA) at room temperature for 30 min, followed by incubation with primary antibodies (Suppl Table 2) in 5% BSA-Tris-Buffered saline (TBS) pH 7.6, 0.1% Tween (TBST) at 4 °C for 24 h. The membranes were then washed 3 × 10 min in TBST and incubated with the appropriate IRDye (1:10,000, goat anti-rabbit IRDye 680RD, 926-68,071, RRID:  AB_10956166: goat anti-mouse IRDye 800CW, 926-32,214, RRID: AB_621846: LI-COR Biosciences, NE, USA) secondary antibody for 1 h at room temperature. Membranes were washed and scanned on an Odyssey Infrared Imaging System (LI-COR Biosciences, NE, USA). Detection of the immunofluorescence signal was carried out at the 680 nm and 800 nm spectrum. No data points were excluded but few bands were not included in analysis due to technical problems, such as problems related to loading, running, background and damage of the gels or membranes. Representative bands from blots were cropped from different parts of the same gel (labeled with asterisk), for full length blots see Supplementary Information.

Analysis.
To measure the signal intensities of each sample band, the analysis was conducted using ImageJ software (National Institutes of Health, USA). Signal intensity of each sample was normalised to β-actin. The logarithm of the protein expression was analysed per protein using a linear mixed model with all interactions between Region, Gender and AgeClass as fixed terms and person-id as a random term. The analysis was performed in R, version 4.0.3. (https:// www.r-proje ct. org), using package lme4 86 . Subsequent residual analysis was done using the package DHARMa (https:// CRAN.R-proje ct. org/ packa ge= DHARMa. Correlation analysis was performed using a Spearman's test in Prism (version 8; GraphPad Software) with a value of P ≤ 0.05 considered significant. Data in all figures was expressed as mean ± SEM. Adobe Photoshop CC 2021 (Adobe Systems Software) was used to prepare the figures.

Ethics approval. All procedures were approved by the University of Auckland Human Participants' Ethics
Committee (Approval number: 001654).

Data availability
The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.