Heterogenous response to aging of astrocytes in murine Substantia Nigra pars compacta and pars reticulata

Currently, little is known about the impact of aging on astrocytes in substantia nigra pars compacta (SNpc), where dopaminergic neurons degenerate both in physiological aging and in Parkinson's disease, an age-related neurodegenerative disorder. We performed a morphometric analysis of GFAP+ astrocytes in SNpc and, for comparison, in the pars reticulata (SNpr) of young (4-6 months), middle-aged (14-17 months) and old (20-24 months) C57BL/6J male mice. We demonstrated an age-dependent increase of structural complexity only in astrocytes localized in SNpc, and not in SNpr. Astrocytic structural remodelling was not accompanied by changes in GFAP expression, while GFAP increased in SNpr of old compared to young mice. In parallel, transcript levels of selected astrocyte-enriched genes were evaluated. With aging, decreased GLT1 expression occurred only in SNpc, while xCT transcript increased both in SNpc and SNpr, suggesting a potential loss of homeostatic control of extracellular glutamate only in the subregion where age-dependent neurodegeneration occurs. Altogether, our results support an heterogenous morphological and biomolecular response to aging of GFAP+ astrocytes in SNpc and SNpr.


Introduction
Among glial cells, astrocytes represent a highly heterogeneous population from a molecular, morphological, and functional point of view. Astrocytes are key homeostatic cells in the central nervous system. Among their pleiotropic functions, they do provide trophic support to neurons, participate in synaptic function and plasticity including uptake and recycling of neurotransmitters, regulate blood-brain barrier integrity and modulate adult neurogenesis ( Spampinato et al., 2019 ;Verkhratsky and Nedergaard, 2018 ). Beyond such homeostatic functions, in response to several endoge-nous and exogenous signals, including potential noxae , astrocytes undergo profound morphological and functional changes, a process referred to as astroglial plasticity ( Czéh et al., 2006 ;Rodríguez-Arellano et al., 2016 ).
Given the broad array of functions played by astrocytes, their age-related morphofunctional changes may deeply impact brain activity ( Verkhratsky et al., 2020 ). Interestingly, it has been reported that astrocytes change their morphological structure in response to aging in a regional and sub-regional specific manner Popov et al., 2021 ;Rodríguez-Arellano et al., 2016 ). Potentially, morpho-functional heterogeneity in astrocyte response to aging may also account for selective vulnerability or resilience to neurodegeneration in distinct brain regions ( Verkerke et al., 2021 ;. The substantia nigra (SN), a basal ganglia structure, is highly sensitive to aging ( Costa, 2014 ). The area can be anatomically and functionally divided into 2 subregions: (1) the pars compacta (SNpc), enriched in dopaminergic (DA) neurons; (2) the pars reticulata (SNpr), a major output nucleus of basal ganglia, enriched in GABAergic neurons which receive DA from dendrites of SNpc DA neurons ( Sonne et al., 2020 ). Progressive DAergic neuronal loss in SNpc is an inevitable physiological change that occurs with aging ( Costa, 2014 ). A loss of more than 60% SNpc DA neurons con-tributes to classical motor symptoms associated to Parkinson's Disease (PD) ( Collier et al., 2017 ). On the contrary, in SNpr neurons are relatively spared by aging and PD-associated neurodegeneration.
Very little is known about differences in astrocytic populations within SN subregions and if they are differentially affected by aging, a major risk factor for several neurodegenerative disorders, including PD ( Tanner and Goldman, 1996 ). Herein we explored structural modifications of astrocytes in SNpc and SNpr of murine brain during aging and correlated these changes with biomolecular alterations which may potentially contribute to functional alterations in their homeostatic functions and vulnerability to neurodegeneration of SNpc DA neurons.

Image acquisition and morphometric analysis
Image acquisition and morphometric analysis were performed by an operator blinded to animal group identity. Confocal 3D im-ages were acquired using the z-stack function of a LSM700 laserscanning confocal microscope (Carl Zeiss). Images were collected at 0.4 μm interval with a 20 × objective (1024 × 1024 pixel, 8bit depth, pixel size 0.313 μm). In each animal, 12 astrocytes per region with apparently fully intact GFAP-immunostained primary processes were selected for morphological analysis. 3D reconstructions of astrocytes were performed using FIJI software (version 1.52) with the Simple Neurite Tracer plugin ( Longair et al., 2011 ). Sholl analysis was performed with Sholl analysis plugin consisting in counting the number of intersections between cell processes and concentric spheres emanating from the center of cell soma at fixed distances (5 μm), as previously described Ferreira et al., 2014 ). Sholl Intersection Profile (SIP) allows an intuitive comparison of astrocytic morphology. Moreover, 3D reconstructions were exported as SWC files and analyzed with L-measure tool to evaluate additional morphometric features ( Scorcioni et al., 2008 ).

RT-qPCR
Messenger RNA concentration and purity were measured using a Nanodrop (Implen) and reverse transcribed by ImProm-II Reverse Trascription System (Promega) according to manufacturer's instructions. Real time qPCR was performed through the iTaq Universal SYBR Green Supermix (Biorad) according to manufacturer's instructions. PCR reaction was carried out with CFX Connect Real-Time PCR Detection System (Bio-Rad). Relative gene expression was obtained after normalization to housekeeping genes (S18) using the standard formula 2 − CT as previously described ( Rocchio et al., 2019 ). Primer sequences are detailed in Table S1 (Supplementary material).

Statistical analysis
All statistical analysis and data visualizations for morphometric results were performed in RStudio version 1.2.5 (RStudio Team, 2015) using R version 3.5.1 (R Core Team, 2016) and the packages ggplot2 ( Wickham, 2016 ), tidyverse ( Wickham et al., 2019 ), lme4 ( Bates et al., 2015 ) and multcomp ( Hothorn et al., 2008 ). For statistical analysis of morphological parameters, a linear mixed effects model was used to model the data of each parameter, with age as fixed effects and animal as a random effect (with lmer function). Using this approach, it is possible to overcome the dependency of the repeated observations within each animal . The presence of significant differences was tested using one-way ANOVA. SIPs were analyzed by mixed-effects nested ANOVA approach, with individual animal as random effect and radius nested within astrocytes nested within age. Statistical analysis of morphological differences between brain areas within each age-group was performed using a mixed-effect models with post hoc Tukey correction (using multcomp package). For RT-qPCR and western blot, statistical analysis and data visualizations were performed in GraphPad Prism 8. According to the number of groups, results were analyzed with Student's t-test or one-way ANOVA with Tukey's post-hoc test. For all analyses, significance was defined as p < 0.05. Data are presented as mean ± SEM.

Aging affects TH and GFAP expression in murine Substantia nigra
C57Bl/6J male mice of 3 different ages were used: young (4-6 month-old, Y), middle-aged (14-17 month-old, MA) and old (20-24 month-old, O). As it can be appreciated by the immune labeling for tyrosine hydroxylase (TH), the rate-limiting dopamine synthetic enzyme, an age-dependent reduction of TH + cells occurred with aging in the SNpc where DAergic neurons are located ( Fig. 1 A). In parallel studies, western blot analysis in total SN homogenates demonstrated an age-dependent reduction of TH protein levels which reached statistical significance in old compared to young mice ( Fig. 1 A; Y vs. O: p .value = 0.03). In the same experimental groups Glial Fibrillary Acidic Protein (GFAP), a classical astrocytic protein marker, was used to label astrocytes in both subregions of the SN by immunohistochemistry ( Fig. 1 B). GFAP + astrocytes are more abundant in SNpr than in SNpc, and no overt change in cell density appeared to occur as a function of age. When GFAP expression levels were evaluated in total SN homogenates of different age groups, western blot analysis demonstrated a statistically significant increase in MA and O groups compared to young mice group

Aging affects GFAP + astrocytic morphological complexity in SNpc, but not in SNpr
To dissect potential subregion-specific astroglial alterations in SNpc we performed a morphometric analysis on a 3-dimensional (3D) reconstruction of GFAP immune labeled astrocytes ( Fig. 2 A and B). As shown by SIP analysis, in the SNpc ( Fig. 2 C) we observed a significant increase in the number of intersections at 10-35 μm from the soma in MA as compared to Y mice. Moreover, there was a remarkable increase in the number of intersections at 5-45 μm radii in O mice as compared to both Y and MA mice ( Fig. 2 C). The age dependent increase in the complexity of GFAP + cells in SNpc was also confirmed by quantification of the total number, the average and the maximum number of intersections and the critical radius (the radius where maximum count of intersections occurred) ( Fig. 2 D, Table 1 ). To further confirm the increase in SNpc astrocyte complexity with aging, we also performed a quantitative analysis of a large panel of additional morphological features. The number of terminal points of astrocytes increased in an age-dependent manner ( Fig. 2 E, Table 1 ). Astrocytes from old group were characterized by a higher number of branches, primary branches, and ramification points compared with young mice group ( Fig. 2 E, Table 1 ). We also compared the length of astrocyte arborizations across ages. The total arborization length significantly increased with aging ( Fig. 2 E, Table 1 ). Similarly, the maximum process length increased with aging ( Fig. 2 E, Table 1 ). Conversely, aging did not impact on the average length of arborization branches ( Fig. 2 E, Table 1 ). To evaluate whether increased morphological complexity was paralleled by increased territory occupied by each astroglia cell, we used the 3D convex hull analysis, which measures the volume enclosed by a polygon that connects terminal points of the cellular processes. Only astrocytes belonging to the O mouse group showed a significant expansion in their volume compared to astrocytes from the other experimental groups ( Fig. 2 Table 1 ).
We then performed 3D reconstruction of GFAP + astrocytes in the SNpr ( Fig. 3 A Table 1 ) across different ages. More detailed morphological analysis also confirmed that branching complexity of astrocytes and their volume were not different in the 3 experimental groups, suggesting no effect of aging on astrocytes within SNpr ( Fig. 3 E, Table 1 ).

Morphometric differences in SNpc versus SNpr astrocytes across ages
Since only astrocytes in the SNpc undergo structural changes with aging, we further examined astrocytic morphology in the 2 SN subregions at each individual age. In young mice, SNpc astrocytes displayed significantly longer processes and reached maximum complexity at higher distance from soma compared to those in the SNpr ( Fig. 4 A, Table 1 ). Nevertheless, SNpc and SNpr astrocyte did not differ in terms of number of branches and total length, as demonstrated by SIP comparison ( Fig. 4 A, Table 1 ). In MA mice, SIP differences between cells from SNpc and SNpr increased remarkably ( Fig. 4 B, Table 1 ). At this age, astrocytes in the SNpc displayed higher number of processes, resulting in a general increase in arborization length ( Fig. 4 B, Table 1 ). In old mice, the morphological differences between astrocytes in the SNpc and SNpr further increased, as demonstrated by comparison of SIP and other morphometric parameters ( Fig. 4 C, Table 1 ).

Aging affects substantia nigra astrocyte enriched transcripts and GFAP protein expression in a subregion-specific manner
The age-dependent morphological alterations in GFAP + astrocytes of the SNpc could possibly underlie changes in astroglial functions. In order to identify biochemical/molecular correlates that may link structural remodeling with functional changes, we isolated SNpc and SNpr from young (n = 6) and old (n = 5) mice for additional analysis. Since an age dependent increase in GFAP levels was observed in SN whole lysates, first we quantified GFAP protein levels in SNpc and SNpr of young and old mice. In the SNpc no change in GFAP protein expression was observed by western blot ( Fig. 5 A). Conversely, in the SNpr GFAP protein levels significantly increased in old compared to young mice ( Fig. 5 B, p value = 0.04). Similar changes occurred for GFAP transcript levels, as measured by RT-qPCR. Indeed, no change in GFAP expression occurred    in SNpc ( Fig. 5 C), whereas GFAP levels were significantly increased in SNpr of old versus young mice ( Fig. 5 D, p value = 0.03). Next, we extended our analysis to other astrocyte enriched transcripts. The expression of the transcript encoding the glycolytic enzyme aldolase C (AldoC) was significantly upregulated in the SNpc of old mice compared to young ones ( Fig. 6 A, p value = .005) and a similar trend, although not reaching significance, was confirmed in the SNpr ( Fig. 6 B, p value = 0.051). No statistically significant difference was observed in Aquaporin 4 (AQP4) transcript expression in both subregions of old compared to young mice ( Fig 6 A and B). We then focused our attention on genes involved in glutamate homeostasis, a key function of astrocytes. Interestingly, a statistically significant decrement in gene expression of the glutamate transporter GLT1 was observed in old compared to young mice only in SNpc ( Fig. 6 A, p . value = 0.005), but not in SNpr ( Fig. 6 B). Conversely, expression levels of glutamate aspartate transporter (GLAST) transcripts were similar in both SN subregions of young and old mice. We also quantified transcripts for metabotropic glutamate receptor 3 (mGluR3), an astrocyte enriched glutamate receptor, and the other group II member, metabotropic glutamate receptor 2 (mGluR2): expression levels of both transcripts were not significantly different in old versus young mice in SNpc and SNpr subregions ( Fig. 6 A and  B). Finally, we evaluated expression levels of the glutamate/cystine antiporter subunit (xCT) transcript. xCT was significantly increased in both SNpc ( Fig. 6 A, p value = 0.0 0 03) and SNpr ( Fig. 6 B, p value = 0.0 0 04) of old mice, compared to young ones.

Discussion
Herein we describe, for the first time, a subregion-specific structural remodeling of SN astrocytes during mouse aging.
Analysis was performed in animals of 3 age ranges: 4-6, 14-17, 20-24 month old. In the 3 experimental groups we performed a morphometric analysis of astrocytes in SNpc and SNpr subregions. To this purpose, astrocyte 3D reconstructions were obtained using GFAP immune labeling. It is well established that GFAP is mainly expressed in the larger cell processes of mature astrocytes and not in fine processes such as peri-synaptic processes ( Reichenbach et al., 2010 ). Despite such limitations, this labelling approach has been commonly used for comparative analysis of key morphometric features of astrocytes by several groups Klein et al., 2020 ;Olabarria et al., 2010 ;SheikhBahaei et al., 2018 ;Yeh et al., 2011 ). Our morphometric analysis confirmed that GFAP + astrocytes undergo a significant increase in their morphological complexity only in the SNpc, and not in the SNpr, as demonstrated by SIP comparison in mice at different ages. In pars compacta we also observed an increase in parameters such as number of branches, ramification and terminal points, maximum length and overall arborization length of astrocytes. Importantly, such remarkable morphological changes occurred in SNpc in parallel with age-dependent reduction of TH immunoreactivity and protein expression levels, a correlate of DAergic neuron dysfunction. Interestingly, we also demonstrated that although morphological complexity of SNpc and SNpr astrocytes is similar in 4-6 month old mice their morphological profiles diverge with increasing age. This observation strengthens the idea that in substantia nigra GFAP + astrocytes undergo structural remodeling in a strictly subregionspecific manner in response to aging.
Of note, increased morphological complexity in SNpc astrocytes with aging was not paralleled by increased GFAP expression. Although GFAP protein levels in total SN extracts were significantly augmented as a function of age and in line with previous observations even in human brain ( Venkateshappa et al., 2012 ), the astrocytic marker increased only in SNpr, and not in SNpc, of old compared with young mice. Similar changes were also observed for GFAP transcript levels in old versus young mice. The lack of correlation between GFAP + astrocyte remodeling and changes in GFAP expression levels with aging is not surprising since it was previously observed in other brain areas ( Cerbai et al., 2012 ;Verkhratsky et al., 2020 ).
Recent studies have tried to understand astrocyte heterogeneous response to aging by applying state of the art omic approaches. For example, transcriptomic studies demonstrated that astrocytes from distinct brain regions of aged murine and human brain tissues have distinctive molecular signatures ( Clarke et al., 2018 ;Soreq et al., 2017 ). Of note, these studies proved that in aging the most dramatic changes in gene expression profiles occur not in neurons but in glia, especially in astrocytes and oligodendrocytes ( Soreq et al., 2017 ). Moreover, such changes mainly took place in brain areas such as human hippocampus and substantia nigra , incidentally 2 brain regions mostly affected by agedependent neurodegeneration ( Soreq et al., 2017 ). These findings provide further insights into the potential role of astrocytes, when dysfunctional, in region-specific neuronal vulnerability, a feature of age-related neurodegenerative diseases. No previous transcriptional studies are available to address molecular signatures of astrocytes  in SN pars compacta and pars reticulata . In our study, we attempted to correlate subregion-specific changes in astrocytic morphology with age-related biomolecular changes in the expression of selected astrocyte-enriched transcripts. It is well established that astrocytes are key actors for glutamate turnover/homeostasis in the brain. Neurons, indeed, are unable of de novo glutamate synthesis and they rely on astrocytes for continuous glutamine supply ( Hertz and Rothman, 2017 ). Moreover, astrocytes express excitatory amino acid transporters 1 and 2 (EAAT1/GLAST and EAAT2/GLT1) which clear approximately 80% of glutamate released as neurotransmitter in synaptic clefts ( Levy et al., 1995 ;Magi et al., 2019 ), thereby preventing the occurrence of excitotoxic events. Recent stereological studies in adult rats have attempted to gain informa-tion about the organization of astrocytes and excitatory synapses in substantia nigra ( Kessler et al., 2020 ). These morphometric studies demonstrated a very high ratio of GLT1 expressing glial membranes to glutamatergic synapses both in SNpc and SNpr, suggesting an efficient control of extracellular glutamate concentrations in the 2 subregions. Interestingly, in our current study, we observed a decrease of GLT1 gene expression occurring only in SNpc, and not in SNpr, in old compared to young mice. In principle, subregion-specific GLT1 dysfunction could potentially lead to extracellular glutamate accumulation that might result, in turn, in increased neuronal degeneration. In our study, we also quantified cysteine-glutamate exchanger (xCT) gene expression in both SN subregions in old compared to young mice. The xCT protein is a subunit of the x c (-) system, an antiporter that exchanges intracellular glutamate for extracellular cysteine, thereby promoting glutathione (GSH) synthesis as well as non-vesicular glutamate release. xCT is predominantly expressed in glial cells, especially microglia and astrocytes ( Ottestad-Hansen et al., 2018 ). Malfunctioning of the xCT system has been suggested to cause oxidative stress and excitotoxicity, both important phenomena in the pathogenesis of several neurological and neurodegenerative disorders, including PD ( Lewerenz et al., 2013 ). Of note, mice lacking xCT were less susceptible to 6-hydroxydopamine (6-OHDA)-induced neurodegeneration in the SNpc compared to age-matched wild-type littermates ( Massie et al., 2011 ). Reduced sensitivity to 6-OHDA might be related to marked reduction in striatal extracellular glutamate levels which was reported in mice lacking xCT ( Massie et al., 2011 ). To our knowledge, although preliminary, our data are the first demonstration that xCT expression may be increased in SN with aging. Interestingly, we did not observe changes in expression levels of transcripts encoding other key proteins involved in the modulation of glutamate neurotransmission, such as mGluR2 and mGluR3 receptors, in both SN subregions of old mice compared to the young counterpart. We also detected a significant increase in Aldolase C (AldoC) transcripts in the SNpc of old compared to young mice, with a similar trend in the SNpr. AldoC is a brain-specific protein, expressed mainly in astrocytes ( Fujita et al., 2014 ), which is involved in glycolysis but also in non-canonical functions, including cytoskeletal rearrangement ( Merkulova et al., 2011 ;Volker and Knull, 1997 ) which is essential for structural remodeling. At the current state of knowledge, functional implications of AldoC transcript changes during aging and in SN are not understood.
Age-dependent increase in astrocyte complexity has been commonly regarded as a hallmark of astroglia reactivity ( Schiweck et al., 2018 ). Recently the prevalent and "classical" concept of an age-dependent increase in reactive astrogliosis has been revised. At present a vast array of data support indeed the idea that astrocytic asthenia or atrophy more often contribute to aging of the brain ( Popov et al., 2021 ;Verkhratsky et al., 2021Verkhratsky et al., , 2020. Altogether, it is currently envisioned that during aging astrocytic changes may also contribute to brain dysfunction through initiation and propagation of neuroinflammation ( Clarke et al., 2018 ), oxidative stress and mitochondrial dysfunction ( Acioglu et al., 2021 ), but mainly through reduction and/or loss in their homeostatic functions including neurotransmitter uptake and gliotransmitter release ( Piacentini et al., 2017 ;Popov et al., 2021 ;Verkhratsky et al., 2021Verkhratsky et al., , 2020. GLT1 expression, which we found to be decreased only in SNpc and not in SNpr, has been shown to be downregulated in reactive astrocytes ( Sheldon and Robinson, 2007 ). The increase in GFAP protein levels has been commonly related to astroglial reactivity ( Eng et al.,20 0 0 ) and associated to aging brain ( Verkerke et al., 2021 ). Conversely, in our study increased morphological complexity of astrocytes in SNpc did not correlate with increased GFAP protein expression, which rather occurred in SNpr where astrocytic morphology was not affected by aging. Once again, these data support the idea that a rigid categorization of morphofunctional and molecular changes occurring in astrocytes is difficult and more complex than we can envision. Moreover, during aging structural and transcriptional changes occurring in astrocytes differ among brain regions ( Boisvert et al., 2018 ;O'Callaghan and Miller, 1991 ;Rodríguez et al., 2014 ). For example, in striatum and hippocampus astrocytes increase their expression of GFAP and become more hypertrophic with aging. In contrast, astrocytes in the entorhinal cortex are atrophic in the aged mice Rodríguez et al., 2014 ). Moreover, hippocampal but not cortical astrocytes upregulate genes which are abundant in reactive astrocytes ( Clarke et al., 2018 ). How such differences in astrocytes across different brain regions contribute to circuit function and dysfunction during aging is still largely unknown ( Lawal et al., 2022 ). Our current data now point to sub region-specific differences in the response to aging of astrocytes in s ubstantia nigra as well.
At present the full relevance of our morphometric and biomolecular data for astrocytes in SNpc is limited by their correlative nature. Despite that, we can hypothesize that age-related changes in SNpc both at the morphological and molecular levels may be associated with functional modifications of GFAP + astrocytes. Loss of homeostatic, protective functions like extracellular glutamate control could be hypothesized based on selective changes in GLT1 expression in the sub region. On the other hand, we cannot rule out the possibility that acquisition of detrimental functions may take place as a consequence of remodeling of GFAP + astrocytes only in the SNpc, where age-dependent neurodegeneration occurs. Indeed, a vast array of studies demonstrated that astrocytes not only exert a supportive role in neuronal circuitries, but they may also play an active role in age-related neurodegeneration ( Acioglu et al., 2021 ;Colombo and Farina, 2016 ;Reid and Kuipers, 2021 ;Verkerke et al., 2021 ;Verkhratsky et al., 2020 ). The fact that these changes occur only in SNpc and not in pars reticulata certainly deserves further investigation, in particular to understand the impact of these changes on neuronal functions.
In conclusion, our current data support, once again, the remarkable degree of heterogeneity at any level, morphological and molecular, within astrocytic populations. Moreover they contribute to increase the current knowledge and interest on the impact of physiological aging on astrocytes in a brain region which is closely linked to age-dependent neurodegeneration ( Jinno, 2011 ;Mosher and Schaffer, 2018 ).

Data availability
All data are available upon request.

Authors contribution
H.B. contributed to design research plan, performed IHC and morphometric analysis, acquired and processed images and figures, contributed to discussion, and wrote the manuscript; F.C. contributed to Western Blot and qPCR analysis, contributed to discussion and wrote the manuscript. I.M. performed acquisition of IHC images; V.B. contributed to discussion; P.L.C. provided expertise and resources and contributed to discussion; M.G. conceived, designed, and coordinated the research plan, wrote and edited the manuscript.

Submission declaration and verification
The authors declare that this paper has not been submitted elsewhere and will not be submitted elsewhere while under consideration at Neurobiology of Aging. If accepted, it will not be published elsewhere in the same form, in English or in any other language, including electronically without the written consent of the copyright-holder.
Publication of this work has been approved by all authors.

Disclosure statement
The authors report no conflicts of interest.