Astrocytic changes with aging and Alzheimer’s disease-type pathology in chimpanzees

Abstract

Astrocytes are the main homeostatic cell of the central nervous system. In addition, astrocytes mediate an inflammatory response when reactive to injury or disease known as astrogliosis. Astrogliosis is marked by an increased expression of glial fibrillary acidic protein (GFAP) and cellular hypertrophy. Some degree of astrogliosis is associated with normal aging and degenerative conditions such as Alzheimer’s disease (AD) and other dementing illnesses in humans. The recent observation of pathological markers of AD (amyloid plaques and neurofibrillary tangles) in aged chimpanzee brains provided an opportunity to examine the relationships among aging, AD-type pathology, and astrocyte activation in our closest living relatives. Stereologic methods were used to quantify GFAP-immunoreactive astrocyte density and soma volume in layers I, III, and V of the prefrontal and mid temporal cortex, as well as in hippocampal fields CA1 and CA3. We found that the patterns of astrocyte activation in the aged chimpanzee brain are distinct from humans. GFAP expression does not increase with age in chimpanzees, possibly indicative of lower oxidative stress loads. Similar to humans, chimpanzee layer I astrocytes in the prefrontal cortex are susceptible to AD-like changes. Both prefrontal cortex layer I and hippocampal astrocytes exhibit a high degree of astrogliosis that is positively correlated with accumulation of amyloid-beta and tau proteins. However, unlike humans, chimpanzees do not display astrogliosis in other cortical layers. These results demonstrate a unique pattern of cortical aging in chimpanzees and suggest that inflammatory processes may differ between humans and chimpanzees in response to pathology.

Graphical Abstract

Unlike humans, chimpanzee cortex does not exhibit age-related astrogliosis. Similar to humans, astrogliosis in chimpanzees does increase with Alzheimer’s-like pathology. Photomicrographs show the absence of astrogliosis in chimpanzee cortex that lacks pathology (a) and prevalence associated with pathology (b). Scale bars = 25 μm.

1. INTRODUCTION

Astrocytes play an important role in the maintenance of a homeostatic environment in the central nervous system (CNS). These homeostatic functions include maintaining ion and water balance, preserving the integrity of the blood-brain barrier, metabolic regulation, neurotransmitter regulation, and synaptic modulation (Avila-Muñoz & Arias, 2014; Cotrina & Nedergaard, 2005; Magistretti & Pellerin, 1999; Magistretti, Pellerin, Rothman, & Shulman, 1991; Magistretti, 2006; Magistretti and Allaman, 2015; Oberheim et al., 2009; Parpura & Zorec, 2010; Pellerin et al., 1998; Sibille, Pannasch, & Rouach, 2014; Wallraff et al., 2006; Verkhratsky & Nedergaard, 2017). Astrocytes have also been shown to influence learning and memory due to their ability to modulate energy metabolism, synaptic transmission, shuttling lactate, and enhancing long-term potentiation, (Fellin, Pascual, & Haydon, 2006; Han et al., 2013; Henneberger, Papouin, Oliet, & Rusakov, 2010; Panatier et al., 2006; Perea & Araque, 2005; Pellerin et al., 1998; Suzuki et al., 2011). In addition to their roles in the healthy CNS, astrocytes actively participate in the inflammatory responses associated with pathological conditions, such as amyotrophic lateral sclerosis, Down syndrome, Parkinson’s disease, Alzheimer’s disease (AD), traumatic brain injury, and epilepsy (Heneka, Rodríguez, & Verkhratsky, 2010; Markiewicz & Lukomska, 2006; Sofroniew, 2009; Sofroniew & Vinters, 2010). The response by astrocytes to pathology is referred to as astrogliosis, or astrocyte activation (Heneka et al., 2010). Astrogliosis is defined as a spectrum of molecular, cellular, and functional changes in astrocytes marked by upregulation of glial fibrillary acidic protein (GFAP), hypertrophy of the soma and cellular processes, and loss of domain organization (Pekny & Pekna, 2004; Sofroniew & Vinters, 2010; Wilhelmsson et al., 2006). Therefore, the response of astrocytes varies and can be represented as continuum of changes that corresponds to the severity and progression of the underlying condition (Sofroniew & Vinters, 2010).

Some degree of astrogliosis, as indicated by the upregulation of GFAP mRNA and protein expression, has also been associated with normal aging in humans and rodents (Beach, Walker, & McGear, 1989; Contrina & Nedergaard, 2002; Finch et al., 2002; Goss, Finch, & Morgan, 1991; Markiewicz & Lukomska, 2006; Morgan et al., 1997; Nicholas, Day, Laping, Johnson, & Finch, 1993). Two studies reported that the overall number of astrocytes does not change with age, but their analysis did not include GFAP-expressing astrocytes specifically (Pakkenberg, 2003; Pelvig, Pakkenberg, Stark, & Pakkenberg, 2008). It appears, however, that although there is no change in absolute numbers with age, there is a phenotypic shift that includes the up-regulation of GFAP and morphological changes such as an increase in dendritic length and altered soma size (Serrano-Pozo, Gómez-Isla, Growdon, Frosch, & Hyman, 2013; Sofroniew & Vinters, 2010).

The involvement of glial cells, including astrocytes, in AD was in fact first reported by Alzheimer who noted glial cells surrounding neuritic plaques and in association with damaged neurons (Alzheimer, 1911). A more recent model describing the phenotypic response of astrocytes to AD indicates that both degeneration and astrogliosis cooccur (Verkhratsky, Olabarría, Noristani, Yeh, & Rodíguez, 2010). It is possible that the initial deficits observed in AD, such as impaired synaptic function and loss of synapses, are the result of a generalized atrophy of astrocytes that leads to their inability to modulate synaptic transmission (Kettenmann & Verkhratsky, 2008; Verkhratsky et al., 2010; Verkhratsky & Parpura et al., 2016). Therefore, astrocytic atrophy and consequent deficit in homeostatic regulation could contribute to the early cognitive changes observed in AD. The initial astrocyte atrophy is then followed by marked activation of an additional adjacent populations of astrocytes. This cascade of astrocyte activation is triggered by the accumulation of amyloid beta protein (Aβ) plaques and hyperphosphorylated tau neurofibrillary tangles (NFT; Armstrong, 2009; Montine et al., 2010; Olabarría, Noristani, Verkhratsky, & Rodríguez, 2010; Rodríguez & Verkhratsky, 2011; Verkhratsky et al., 2010). Astrogliosis is observed in regions surrounding amyloid deposition and the dendritic processes of activated astrocytes participate in the formation of neuritic plaques (Beach & McGeer, 1988; Beach et al., 1989; Duffy, Rapport, & Graf, 1980; Mancardi, Liwinicz, & Mandybur, 1983; Mandybur, 1989; Nagele, Wegiel, Venkataraman, Imaki, & Wang, 2004; Rodríguez, Olabarria, Chvatal, & Verkhratsky, 2009; Schechter, Yen, & Terry, 1981). Reactive astrocytes are also found in proximity to NFT (Serrano-Pozo et al., 2011).

Astrocyte morphology and function appears to have evolved in humans and in tandem with other cortical microstructural features may work to render humans more susceptible to neuropathologies such as AD (Colombo, Yáñez, Puissant, & Lipina, 1995; Colombo, Quinn, & Puissant, 2002; Falcone et al., 2018; Magistretti, 2006; Oberheim et al., 2006; Oberheim et al., 2009; Oberheim, Wang, Goldman, & Nedergaard, 2006). Studies in aged great apes have reported the presence of senile plaques and tau pathology (Edler et al., 2017; Gearing, Rebeck, Hyman, Tigges, & Mirra, 1994; Gearing, Tigges, Mori, & Mirra, 1996; Gearing, Tigges, Mori, Mirra, 1997; Kimura et al., 2001; Perez et al., 2013; Rosen et al., 2011; Selkoe, Bell, Podlisny, Price, & Cork 1987). Therefore, to gain a better understanding of the role that astrocytes play in aging as well as in the pathogenesis of AD and other degenerative brain disorders, we examined the response of astrocytes in the cerebral cortex of aged chimpanzees.

2. MATERIALS AND METHODS

2.1. Specimens and cortical regions

The study sample included the brains of 27 chimpanzees (Pan troglodytes; 16 females and 11 males) ranging from 12 to 62 years of age (Table 1). A subset of these brains (chimpanzees aged 37 to 62 years) was included in a previously published study that examined AD-type pathology in chimpanzees (Edler et al., 2017). All postmortem brains samples were obtained from AAALAC or AZA-accredited zoological or biomedical research facilities. All brain specimens were provided by the National Chimpanzee Brain Resource (C.C.S and W.D.H.).

TABLE 1.

Specimens used in the present study. Sex and age (years) for all chimpanzees included in the study. Any regions omitted from statistical analyses are listed in the last column.

Sex	Age	Omitted Regions/Parameters
F	12
F	16	HC Density, HC Volume
M	17
M	19
M	20.5
F	20.9
M	24
F	29	MTG Density
F	35.4	HC Density, HC Volume
F	37
F	39
M	39	CA3 Volume
F	40
M	40
F	40
M	40.5
M	40.7
F	41	CA3 Volume
F	43
F	45	CA1 Volume, CA3 Volume
M	46.3
F	49
F	51	CA3 Volume
M	57	CA1 Volume, CA3 Volume
F	58	CA1 Volume
F	58
M	62

The brain regions analyzed included the hippocampus (HC), mid temporal gyrus (MTG; Brodmann’s area 21), and prefrontal cortex (PFC; Brodmann’s areas 9 or 10). These regions were chosen based on guidelines set forth for the neuropathologic assessment of AD, staging for NFT, and Aβ accumulation (Braak & Braak, 1991; Montine et al., 2012; Thal, Rub, Orantes, & Braak, 2002). The pyramidal layer of HC fields CA1 and CA3 and layers I, III, and V of PFC and MTG were included in the analyses. Layer I interlaminar astrocytes have been shown to be uniquely susceptible to disruption in AD and layers III and V were chosen as they contain higher densities of amyloid plaques and NFT than other layers in AD (Arnold, Hyman, Flory, Damasio, & Van Hosen, 1991; Colombo et al., 2002; Davies, Mann, Sumpter, & Yates, 1987; Lewis, Campbell, Terry, & Morrison, 1987; Pearson, Esiri, Hiorns, Wilcock, & Powell, 1985). Furthermore, a previous study examined astrogliosis patterns in human AD tissue, confirming that activated astrocytes are present in layers II, III, and V (Beach et al., 1989). Right or left hemispheres were used based on specimen availability. Aβ and NFT have been shown to be deposited symmetrically in humans, suggesting that examining AD pathology and astrogliosis in one hemisphere should not skew results (Giannakopoulos, Kövari, Herrmann, Hof, & Bouras, 2009).

2.2. Tissue processing

Chimpanzee brains were collected with a postmortem interval of less than 20 hours following natural death or health-related euthanasia. Brains were immersion-fixed in 10% buffered formalin solution for 7 to 10 days and then transferred to a 0.1 M phosphate buffered saline solution (PBS) containing 0.1% sodium azide for storage at 4° C.

Prior to sectioning, samples were cryoprotected in a graded series of sucrose solutions (10, 20, and 30%). Brains were sectioned using a freezing-sliding microtome (SM200R, Leica, Chicago, IL). Sections were individually placed in sequentially numbered centrifuge tubes containing freezer storage solution (30% each distilled water, ethylene glycol, and glycerol and 10% 0.244 M PBS) and stored at −20° C until further histological or immunohistochemical processing. Every 10^th section was stained for Nissl substance using a 0.5% cresyl violet solution to visualize the cytoarchitecture. Nissl-stained sections were used to delineate HC subfields, neocortical regions, and layers (Figs. 1, 2).

FIGURE 1.

GFAP-immunostained section (a) and Nissl-stained section (b) with cortical layers labeled in the PFC of a 62 year-old chimpanzee. Nissl-stained sections were used to locate the appropriate cortical layers for quantification of the immunohistochemically processed sections. Scale bars = 250 μm.

FIGURE 2.

GFAP-immunostained section (a) and Nissl-stained section (b) in a 62 year-old chimpanzee HC. All hippocampal subfields are labeled but only CA1 and CA3 were quantified for GFAP immunoreactivity density and soma volume. Nissl-stained sections were used to locate the appropriate subfield for quantification. Scale bars = 250 μm.

2.3. Antibody characterization

Table 2 describes the antibodies used in the present study. GFAP is a rabbit polyclonal antibody raised against glial fibrillary acidic protein (1:12,500 dilution, Millipore, Burlington, MA, Cat# AB5804, RRID: AB_2109645). 6E10 is a mouse monoclonal IgG1 antibody raised against amino acid residue 1–16 (DAEFRHDSGYEVHHQK) of Aβ (1:7,500 dilution, Covance Research Products, San Diego, CA, Cat# SIG-39320, RRID: AB_2313952). This antibody also has an epitope of residues 3–8 which recognizes Aβ isoforms and its precursor APP (Kim et al., 1988). AT8 is a mouse monoclonal IgG1 antibody raised against partially purified human paired helical filament (PHF) tau with an epitope at residues phosphoserine202/phosphothreonine205 which recognizes phosphorylated tau isoforms (1:2,500 dilution, Thermo Fisher Scientific, Waltham, MN, Cat# MN1020, RRID: AB_223647; Goedert et al., 1995a; Rosen et al., 2008). CP13 is a mouse monoclonal IgG1 antibody with an epitope at residue phosphoserine202 which recognizes phosphorylated tau (1:10,000 dilution, gift from Dr. P. Davies, RRID: AB_2314223; Jicha, Bowser, Kazam, & Davies, 1997). Specificity of antibodies AT8 and CP13 were determined by Western blot (Goedert et al., 1995a; Jicha et al., 1997). Antibody 6E10 specificity was verified by Kim et al. (1988). The GFAP antibody used in immunohistochemistry protocols positively stained astrocytes congruent with previously published research (Beach et al., 1989; Hansen, Armstrong, & Terry, 1987; Rodríguez et al., 2014; Sofroniew & Vinters, 2010).

TABLE 2.

Primary antibodies used in this study.

Antigen	Antibody	Host/Type	Dilution	Suppler/Catalog Number	RRID
GFAP	Glial fibrillary acidic protein	Rabbit, polyclonal	1:12,500	Millipore, AB5804	AB_2109645
APP/Aβ (6E10)	Aβ/APP residues 1-16 (DAEFRHDSGYEVHHQK)	Mouse, monoclonal	1:7,500	Covance, SIG-39320	AB_2313952
Tau (AT8)	Paired helical filaments – tau phosphoSer202 and phosphoThr205	Mouse, monoclonal	1:2,500	ThermoFisher, MN1020	AB_223647
Tau (CP13)	Phosphorylated tau – phosphoSer202	Mouse, monoclonal	1:10,00	Gift from Dr. Peter Davies	AB_2314223

2.4. Immunohistochemistry

A series of brain sections from each region were immunostained using the avidin-biotin-peroxidase method for either amyloid (6E10) or tau (AT8/CP13) as well as GFAP to visualize both AD pathology and astrocytes within the tissue. Tissue sections were first pretreated for antigen retrieval for 30 minutes in a 0.05% solution of citraconic acid (pH 7.4) at 85° C. Endogenous peroxidase was quenched for 20 minutes using a solution of 75% methanol, 2.5% hydrogen peroxide (30%), and 22.5% distilled water. Sections were then pre-blocked in a solution containing 90.4% 0.1 M PBS (pH 7.4), 0.6% Triton X-100, 4% normal serum, and 5% bovine serum albumin. Tissue was incubated for 48 hours in a primary antibody. One of three primary antibodies were used 6E10 at a dilution of 1:7,500 (mouse monoclonal antibody, Covance, Cat# SIG-39320, RRID: AB_2313952), AT8 at a dilution of 1:2,500 (mouse monoclonal antibody, Thermo Fischer Scientific, Cat# MN1020, RRID: AB_223647) or CP13 at a dilution of 1:10,000 in PBS (mouse monoclonal antibody, gift from Dr. P. Davies, RRID: AB_2314223). After incubation, sections were placed in a solution of biotinylated secondary antibody (1:200) and 2% normal serum. Sections were then placed in avidin-peroxidase complex (PK-6100, Vector Laboratories, Burlingame, CA, USA) followed by 3, 3’-diaminobenzidine-peroxidase substrate with nickel enhancement (SK-4100, Vector Laboratories). Next, tissue sections were incubated in a blocking solution containing 5% normal serum and 90.5% 0.1 M PBS. The tissue was then incubated for 48 hours in anti-GFAP antibody at a dilution of 1:12,500 in PBS (rabbit polyclonal antibody, Millipore, Cat# AB5804, RRID: AB2109645). Following incubation, sections were placed in a solution of biotinylated secondary antibody (1:200) and 2% normal serum. Lastly, sections were placed in avidin-peroxidase complex followed by NovaRED (SK4800, Vector Laboratories).

2.5. AD-type pathology

All chimpanzees aged 37 years and older were also included in a recent study of AD pathology in chimpanzees (Edler et al., 2017). To evaluate the response of astrocytes to the presence of neuropathological changes, the tau, amyloid, and pathology scores were calculated as previously described by Edler et al (2017). Briefly, the amyloid score was determined using three methods to assess Aβ plaque frequency, progression of amyloid pathology, and staging for cerebral amyloid angiopathy (CAA; Attems, Quass, Jellinger, & Lintner, 2007; Mirra et al., 1991; Montine et al., 2012; Thal et al., 2002; Vonsattel et al., 1991). For each individual, all four regions (PFC, MTG, CA1, and CA3) were examined and given a score from 0 to 3 for each of the three scoring methods. The regional scores were then totaled to provide a score ranging from 0 to 12 for each chimpanzee. This was repeated for each scoring method, resulting in an overall score of 0 to 36, representing the cumulative amyloid load within the brain. For tau score, two scoring methods measuring the frequency of tau neuritic plaques and NFT were utilized (Braak & Braak, 1991; Braak, Alafuzoff, Arzberger, Kretzschmar, & Del Tredici, 2006; Mirra et al., 1991; Montine et al., 2012). As before, each of the four brain regions was giving a score ranging from 0 to 3 for one scoring method and these scores were then summed in each individual to provide a score from 0 to 12. This was completed for both scoring methods. Therefore, each chimpanzee generates a tau score ranging from 0 to 24 that represents the overall amount of tau neuritic plaques and NFT in the brain. The pathology score for each chimpanzee is the sum of the amyloid and tau score (0 to 60) and represents the overall disease state of the brain.

Several quantitative measures of pathological markers collected by Edler et al. (2017) were also used as a means to examine the association of activated astrocytes to AD-type pathology. The pathological markers specific for amyloid pathology include amyloid precursor protein (APP)/Aβ plaque volume, APP/Aβ vessel volume, and Aβ42 vessel volume. All immunohistochemical procedures for APP/Aβ and tau (using antibody 6E10 for amyloid and antibody AT8 or CP13 for tau) were described in Edler et al. (2017) and the same materials were used in the present study.

2.6. Quantification and statistics

GFAP-immunoreactive (ir) astrocyte density (mm³) and soma volume (pm³) were quantified using an Olympus BX-51 photomicroscope equipped with Stereoinvestigator software (MBF Bioscience, Williston, VT, USA, version 11). Nissl-stained sections adjacent to the immunostained sections were used to guide the delineation of cortical layers and subfields at low magnification (4×, 0.13 N.A.; see Figs. 1 and 2). Astrocyte density was quantified using the optical fractionator probe at high magnification (40×, 0.75 N.A.). The counting frame was set at 100 × 100 μm with an optical disector height of 7 μm and a guard zone of 2 μm. Section thickness was measured at every fifth sampling site. All GFAP-ir astrocytes within the counting frame were counted. GFAP-ir soma volume was collected simultaneously using the nucleator probe in Stereoinvestigator. Soma volume was measured for every second marked astrocyte and 6 rays were used to define the soma edge. GFAP-ir astrocyte density was calculated as the estimated population divided by planimetric volume of the region of interest sampled. All statistical analyses were completed using IBM SPSS Statistics Data Editor 24.0 software. Significance was set at 0.05 except for tests examining sex differences, and Bonferroni correction post hoc analyses of cortical layer and region differences was used as appropriate for multiple t-test or Mann-Whitney U comparisons (α = .006 for sex and α = .017 for cortical layers and regions).

Mean astrocyte density and soma volume in each region by cortical layer were examined for sex differences using independent samples t-test. Mean differences were also analyzed between HC subfields CA1 and CA3 using the same statistical test. GFAP immunoreactivity density and soma volume, analyzed by cortical layer and subfield, were regressed on age, three scores (tau, amyloid, and pathology), and several markers of AD pathology obtained from our previous study, (see Edler et al., 2014 for full methodological details). For regressions using pathology scores and pathological markers, only chimpanzees 37 years of age and older were included in the analyses. Chimpanzees younger than 37 years (n = 9) did not show any signs of AD pathology as assessed with immunohistochemistry (i.e., 6E10 for amyloid and CP13 or AT8 for tau).

To analyze the differences among the three cortical layers (I, III, V) of the PFC and MTG, a one-way analysis of variance (ANOVA) was used. To compare GFAP-ir astrocyte density and soma volume by region, an overall average for each individual for each area was calculated. Average astrocyte density and soma volume for the PFC and MTG were calculated by adding the totals for each cortical layer together and then dividing by 3. For the HC, CA1 and CA3 densities or volumes were added together then divided by 2. These averages were compared using a one-way ANOVA. For the ANOVA of astrocyte density, the test of the homogeneity of variance was rejected. To correct for this, the nonparametric Kruskal-Wallis test was performed instead. All post hoc testing was conducted using independent samples t-tests or a Mann-Whitney U test.

3. RESULTS

3.1. Qualitative assessment of GFAP-ir astrocytes

Immunostaining for GFAP was robust (Fig. 3) and GFAP-ir astrocytes were visualized in all examined regions (Fig. 3a). GFAP-ir astrocytes were present in higher quantities in the outer neocortical layers (layers I-III) with very few reactive astrocytes present in the deeper cortical layers (layers V and VI). Astrocytes in the superficial cortical layers often break their domains, creating regions or clusters of activated astrocytes (Fig. 3b, c).

FIGURE 3.

Photomicrographs of GFAP-ir astrocytes in the PFC of a 37 (a) 58 (b, c), 50 (d), and 57 (e) year-old chimpanzees demonstrating robust GFAP immunoreactivity and the presence of astrogliosis. GFAP immunolabeling can be seen throughout the cortical layers with astrogliosis occurring predominantly in the outer cortical layers (I-III; a). Activated astrocytes in the superficial cortical layers formed clusters (white arrows) with overlapping processes signifying a loss of domain organization (b, c). Panels d and e show double immunolabeling for GFAP and tau (white arrows). These astrocytes are morphologically consistent with what has been termed granular/fuzzy tau-ir astrocytes (white arrows; b, c). Scale bars = 250 μm (a) and 25 μm (b-e).

The term aging-related tau astrogliopathy (ARTAG) was recently applied to describe astroglial pathology detected using tau immunohistochemical markers (Kovacs et al., 2016; Kovacs et al., 2018). Although not quantified here, we did observe tau-positive astrocytes in the chimpanzee cortex, with morphological types consistent with what has been termed granular/fuzzy tau-ir astrocytes (Fig. 3d, e). Future studies are needed to determine if the chimpanzee pattern follows that of what has been reported in human aging (Kovacs et al., 2018).

Astrocytes were also more numerous in the neocortical regions compared to the HC. In addition, GFAP-ir astrocytes were frequently observed in association with blood vessels (Fig. 4), The processes of these astrocytes extended toward the vasculature in a web-like fashion (Fig. 4a). Many of these blood vessels were immunoreactive for AD pathology (Fig. 4b, c; Edler et al., 2017). GFAP-ir activated astrocytes were also found in association with pretangles and chimpanzee specific neuritic clusters (Fig. 5a, b; Edler et al., 2017).

FIGURE 4.

Astrogliosis was frequently found surrounding the vasculature in the PFC of a 57 (a) and 58 (b, c) year-old chimpanzees. The first panel shows astrocytes exhibiting tau accumulations (stained black; white arrows) lining blood vessels (a). Tau pathology was labeled using an antibody specific for phosphorylated tau (CP13). These astrocytes have many GFAP-ir processes (black arrows) that surround blood vessels in a web-like fashion (a). Panels b and c show activated astrocytes (black arrows) near blood vessels with amyloid pathology (stained black; white arrows). Amyloid pathology was labeled using antibodies specific for Aβ and amyloid precursor protein (6E10). Scale bars = 25 μm.

FIGURE 5.

Astrogliosis (black arrows) occurs near chimpanzee specific neuritic clusters (white arrows) in the frontal cortex of a 57 year-old male chimpanzee. Neuritic clusters are visualized with immunohistochemical staining for phosphorylated tau. Scale bars = 250 μm (a) and 25 μm (b).

3.2. Regional and laminar differences

There were no sex differences (Student’s t-test, all p’s > .006) for GFAP-ir astrocyte density and soma volume for all layers and regions examined. The Kruskal-Wallis analysis of GFAP-ir astrocyte density revealed a significant difference among the PFC, MTG, and HC (H = 41.88, p < .05). Post hoc testing showed that astrocyte density is significantly lower in the HC compared to the PFC (U = 23) and MTG (U = 51, both p’s < .017; see Fig. 6a). The ANOVA comparing astrocyte soma volume among regions was also significant (F_2,74 = 3.92, p < .05). Post hoc analysis revealed that soma volume was larger in the PFC compared to the HC (t₄₈ = 2.51, p < .017; Fig. 6b).

FIGURE 6.

GFAP immunoreactivity is stronger in neocortical regions. Astrocyte density was significantly higher in PFC and MTG compared to HC (a; p’s < .017). Astrocyte soma volume was also significantly higher in PFC compared to HC (b; p < .017).

ANOVA analyses were used to compare astrocyte density in the PFC and MTG among layers I, III, and V. Significant differences were found among the three layers in each region (F_2,78 = 17.29 for PFC and F_2,75 = 4.97 for MTG, both p’s < .05; Fig. 7). Post hoc tests revealed that GFAP-ir density was higher in layer I relative to layers III and V for the PFC (t_39.86 = 5.19 for layer III and t₅₂ = 4.4 for layer V, both p’s < .017; Fig. 7a). Similarly, layer I astrocyte density was higher compared to layer III for MTG (t₅₀ = 3.35, p > .017; Fig. 7b). The ANOVA analyses comparing astrocyte soma volume in the PFC and MTG by cortical layers was significant for both regions (F_2,78 = 3.0 for PFC and F_2,78 = 7.7 for MTF, both p’s < .05). Further post hoc tests showed that soma volume did not differ among individual cortical layers in the PFC (all p’s > .017). However, astrocyte soma volume was greater in layer I compared to layer V of the MTG (t₅₂ = 4.0, p < .017; Fig. 7c).

FIGURE 7.

GFAP immunoreactivity was higher in layer I in both PFC and MTG. In the PFC there was a significantly greater astrocyte density in layer I compared to layers III and V (a; p’s < .017). In the MTG, astrocyte density was significantly higher in layer I compared to layer III (b; p < .017). Astrocyte soma volume in the MTG was significantly greater in layer I compared to layer V (c; p < .017).

Student’s t-tests demonstrated that astrocyte density and soma volume significantly differed between hippocampal subfields. The results indicated that there is a greater density of GFAP-ir astrocytes in the CA3 subfield compared to CA1 (t_32.627 = −2.099, p < .05; Fig. 8a). However, CA1 astrocytes have a larger soma volume compared to the CA3 subfield (t₄₀ = 2.028, p < .05; Fig. 8b).

FIGURE 8.

GFAP-ir astrocyte density was significantly higher in the CA3 subfield compared to the CA1 indicating that more astrocyte within the CA3 upregulate GFAP expression in aging (a; p < .05). Astrocyte soma volume is significantly greater in the CA1 indicating astrocytic hypertrophy (b; p < .05). This indicates that CA3 astrocytes are in the earlier stages astrogliosis compared to the CA1.

3.3. Aging and pathological markers

Age did not correlate with GFAP-ir astrocyte density or soma volume (p’s > .05) for the cortical layers and subfields examined with one exception (Figs. 9–11). The astrocytic soma volume in MTG layer V decreased with age (F_1,25 = 5.87, p < .05; Fig. 10f). However, we believed this result was due to an outlier in the data (i.e. one value being more than 3 standard deviations from the mean) and the data were reanalyzed with this individual removed. The second test found that GFAP-ir astrocyte density does not decrease with age (F_1,24 = 3.61, p > .05).

FIGURE 9.

Linear regression results comparing GFAP-ir astrocyte density and soma volume by age in the layers of the PFC. All analyses reveal that astrocyte activation does not significantly correlate with age (p’s > .05). The R² value of each regression is indicated in each panel. Open symbols, females; black symbols, males.

Figure 11.

Linear regression comparing age to GFAP-ir astrocyte density and soma volume in each hippocampal subfield. The results show that GFAP-ir astrocyte density and soma volume were not significantly correlated with age (p’s > .05). The R² value of each regression is indicated in each panel. Open symbols, females; black symbols, males.

FIGURE 10.

Linear regression analyses examining GFAP-ir astrocyte density and soma volume by age in the MTG by cortical layers. The results indicate that astrocyte activation does not significantly correlate with age in the MTG except for GFAP-ir astrocyte soma volume in layer V (p’s > .05 for a-e; p < .05 for f). In layer V, one sample’s soma volume was greater than 3 standard deviations away from the mean (black arrow) and thus was removed from the data set and the statistical analysis was rerun (f). Upon rerunning the analysis, GFAP-ir soma volume no longer significantly increased with age (f; p > .05). The R² value of each regression is indicated in each panel. Open symbols, females; black symbols, males.

Regression analyses of tau, amyloid, and pathology scores revealed no significant relationship with astrocyte density or soma volume in the layers and subfield examined (all p’s > .05) except layer I of the PFC. GFAP-ir astrocyte density increased in PFC layer I as all three scores increased (F_1,16 = 5.09 for tau score, F_1,16 = 10.74 for amyloid score, and F_1,16 = 13.06 for pathology score, all p’s < .05; Fig. 12a-c).

FIGURE 12.

Linear regression analyses in PFC layer I comparing GFAP-ir astrocyte density to the three individual AD scores (a-c). Astrogliosis was significantly increased in relation to tau score (a), amyloid score (b) and pathology score (c; all p’s < 0.5). Linear regressions comparing GFAP immunolabeling density in PFC layer I to AD-type pathology (c-d). Each regression analyses showed a significant correlation to astrocyte GFAP immunoreactivity density in layer I in response to neuritic cluster density (d), APP/Aβ vessel volume (e), Aβ42 vessel volume (f; all p’s < .05). AD-type pathology appears to trigger an increase in the density of reactive astrocytes in layer I indicating the possible vulnerability of these astrocytes. The R² value of each regression is indicated in each panel. Open symbols, females; black symbols, males.

Examination of the individual AD markers collected by Edler and colleagues (2017) showed that astrocyte density increased with neuritic cluster density (F_1,16 = 20.9; Fig. 12d), APP/Aβ vessel volume (F_1,16 = 8.18; Fig. 12e), and Aβ42 vessel volume (F_1,16 = 9.96, all p’s < .05; Fig. 12f) in the PFC layer I. However, astrocyte density was not associated with APP/Aβ plaque volume or pretangle density (all p’s > .05). Astrocyte soma volumes in the PFC were not correlated with any specific pathological marker of AD (all p’s > .05). GFAP-ir astrocyte density and soma volume in the MTG were not correlated with any pathological marker of AD examined in this study.

In the CA1, astrocyte soma volume increased with neuritic cluster density (F_1,12 = 6.36, p < .05; Fig. 13a), whereas astrocyte density significantly increased with APP/Aβ plaque volume (F_1,16 = 29.85; Fig. 13b) and Aβ42 vessel volume (F_1,15 = 4.82, both p’s < .05; Fig. 13c). For hippocampal subfield CA3, astrocyte density was shown to increase with APP/Aβ vessel volume (F_1,16 = 4.58; Fig. 14a) and Aβ42 vessel volume (F_1,15 = 9.2, both p’s < .05; Fig. 14b). Astrocyte soma volume in the CA3 subfield was not associated with any of the pathological markers of AD (all p’s > .05).

FIGURE 13.

Linear regression analyses show that GFAP-ir soma volume significantly increases with tau neuritic cluster density in the CA1 (a; p < .05). The results of the linear regressions of GFAP-ir astrocyte density in the CA1 significantly increases with an increase APP/Aβ plaque volume (b) and Aβ42 vessel volume (c; both p’s < .05). Each regression indicates that activated astrocytes measured by overall density and soma volume increases in relation to AD-type pathology in the chimpanzee CA1. The R² value of each regression is indicated in each panel. Open symbols, females; black symbols, males.

FIGURE 14.

Linear regression analyses in the CA3 show a significant increase in GFAP-ir astrocyte density with an increase in APP/Aβ vessel volume (a) and Aβ42 vessel volume (b; both p’s < .05). These results indicate that there is a strong response by astrocytes to amyloid accumulation in the vasculature. The R² value of each regression is indicated in each panel. Open symbols, females; black symbols, males.

4. DISCUSSION

The current sample included chimpanzees that ranged in age from 12 to 62 years. The lifespan of chimpanzees, even in captivity, does not usually exceed 50 years, therefore, the present sample is representative of an aged population from adolescence to maximum lifespan (Rowe & Meyers, 2012). Our results revealed that astrogliosis, as measured by GFAP-ir astrocyte density and soma volume, does not correlate with age in chimpanzees. These results differ from previous studies in humans and rodents that reported an increase in GFAP mRNA, GFAP protein expression, and hypertrophy of astrocyte soma volume with age (Table 2; Beach et al., 1989; David et al., 1997, Finch et al., 2002; Goss et al., 1991; Janota, Brites, Lemere, & Brito, 2015; Landfield, Rose, Sandles, Wohstaddter, & Lynch, 1977; Morgan et al., 1997; Nichols et al., 1993, O’Callaghan & Miller, 1991). However, these results are consistent with reports of no association of GFAP-ir astrocytes with age in rhesus macaques (Haley, Kohama, Urbanski, & Raber, 2010; Robillard, Lee, Chiu, & MacLean, 2016).

It has been proposed that the increase in GFAP expression in astrocytes may represent a response to increased inflammation and oxidative stress loads of the aging brain (Cotrina & Nedergaard, 2002; Morgan et al., 1997; Sohal & Weindruch, 1996). The lack of an increased GFAP expression with aging in chimpanzees suggests either a lower oxidative stress load with age or the existence of protective mechanisms that renders chimpanzee astrocytes less sensitive to disturbances in their environment. One study found that gene and protein expression differed between chimpanzees and humans specifically in relation to aerobic metabolism (Bauernfeind et al., 2015). This is intriguing as aerobic metabolism is one of the main producers of oxygen free radicals which can cause oxidative stress (Chance, Sies, & Boverls, 1979; Fridovich, 1978; Halliwell, 1994; Uttara, Singh, Zamboni, & Mahajan, 2009). Human astrocytes are also more integrated into the surrounding neuronal networks and have increased abilities to modulate neuronal transmission (Cotrina & Nedergaard, 2005; Oberheim et al., 2009; Parpura & Zorec, 2010; Pellerin et al., 1998; Sibille et al., 2014; Wallraff et al., 2006). It is possible that these changes make human astrocytes more sensitive to their local environments and thus more easily express a reactive phenotype.

The present study as well work in rhesus macaques examined astrogliosis using immunohistochemical methods to label GFAP expressing cells (Haley, Kohama et al., 2010; Robillard et al, 2016). However, the majority of work examining GFAP expression with age used methods that allowed for the quantification of the overall amount of protein or mRNA upregulation in the brain (Contrina & Nedergaard, 2002; Finch et al., 2002; Goss et al, 1991; Morgan et al., 1997; Nicholas et al, 1993). Unfortunately, these approaches cannot be reliably applied in most primate brain samples due to the method of postmortem tissue fixation. Therefore, even though we did not find an increase in the number of astrocytes labeled for GFAP, we cannot definitively state that the overall amount of this protein does not change with age. However, although our methods are limited, the results of this study are consistent with earlier reports that found differences between humans and chimpanzees in brain aging (Autrey et al., 2014; Chen et al., 2013; Sherwood et al., 2011).

A major risk factor of AD in humans is age (Akinyemi, Mukaetova-Ladinska, Attems, Ihara, & Kalaria, 2013; National Institute of Aging, 2012). The presence of AD pathological markers in aged chimpanzees provides an opportunity to examine the downstream consequences of those naturally-occurring pathologies in a nonhuman species. In humans, AD pathology (plaques and tangles) and activated astrocytes are common in neocortical layers II, III, and V (Arnold et al., 1991; Davies et al., 1987; Pearson et al., 1985; Lewis et al., 1987; Beach et al., 1989). However, in the present chimpanzee sample, little astrogliosis was noted outside the supragranular layers (I-III) or in the HC, and the densest astrocytic response appeared in layer I of the neocortex. This pattern of astrogliosis appears to differ from the pattern seen in a series of AD human brains which found that astrogliosis occurs throughout the cortex (Table 3; Beach et al., 1989). However, a second study examining GFAP expression in an AD human cohort found dense astrogliosis in layer I (Simpson et al., 2010). Therefore, high GFAP expression in the upper cortical layers may be a shared response to pathology for both humans and chimpanzees.

TABLE 3.

Pattern of astrogliosis due to aging and AD pathology in chimpanzees and humans.

Pattern of Astrogliosis in Alzheimer’s Disease
Chimpanzees	Humans	Both
Lack of astrogliosis in neocortical layers where AD pathology is present.	Astrogliosis occurs heavily in neocortical layers II, III, and V.1	Disruption of layer I interlaminar astrocytes in the neocortex.2,3
		Astrogliosis is triggered by amyloid pathology including CAA.4–7
		Astrogliosis increases in response to tau pathology (tau neurtic plaques and NFT’s).8
		Increase in astrogliosis with the progression AD pathological burden (i.e. pathology score).9
Pattern of Astrogliosis in Nonpathological Aging
Chimpanzees	Humans	Both
GFAP-ir astrocyte density does not increase with age.	GFAP mRNA and protein increases with age.1,10–12	Qualitative examinations reveal astrogliosis occurs in upper cortical layers in response to normal aging.1

^1.Beach et al., 1989;

^2.Colombo et al., 2002;

^3.Colombo et al., 2004;

^4.Duffy et al., 1980;

^5.Kimbrough et al., 2015;

^6.Mancardi et al., 1983;

^7.Nagele et al., 2004;

^8.Serrano-Pozo et al., 2011;

^9.Serrano-Pozo et al., 2013;

^10.David et al., 1997;

^11.Finch et al., 2002;

^12.Nichols et al., 1993

We also examined astrogliosis in response to AD pathology in the chimpanzee brains. In the HC, we found a different regional pattern of GFAP-ir astrocyte density and soma volume between the hippocampal subfields. CA1 GFAP-ir astrocyte density was lower than that of CA3. However, CA1 astrocytes had a greater soma volume. Astrogliosis was associated with neuritic cluster density, APP/Aβ plaque volume, and Aβ42 vessel volume within the CA1 subfield, whereas astrogliosis within the CA3 was only associated with amyloid accumulation within the vasculature. These results are interesting in light of the fact that in humans, the CA1 is more vulnerable to AD pathology relative to CA3, further highlighting the differences between these two species in their response to pathology (Braak & Braak, 1991; Mueller et al., 2010; Rössler, Zarski, Bohl, & Ohm 2002). Further support for these differences were found by Edler et al (2018), who reported a higher density of activated microglia in the CA3 compared to the CA1 in aged chimpanzees which included samples with both Aβ and tau pathology.

GFAP-ir astrocyte density was higher in the PFC and MTG relative to the HC. Soma volume of the PFC GFAP-ir astrocytes was also significantly larger than in the HC. These results correspond to the pattern of AD-type pathology that showed a higher total percent area occupied by amyloid plaques as well as greater vascular amyloid deposition in the neocortex compared to the HC (Edler et al., 2017). Amyloid deposition in the neocortex likely triggered the higher astrogliosis observed in these regions, as activated astrocytes typically surround amyloid plaques (Beach & McGeer, 1988; Beach et al., 1989; DeWitt, Perry, Cohen, Doller, & Silver, 1998; Duffy et al., 1980; Kimbrough, Robel, Roberson, & Sontheimer, 2015; Mancardi et al. 1983; Mandybur, 1989; Nagele et al., 2004; Schechter et al., 1981; Rodríguez et al., 2009). Similarly, much of the astrogliosis we observed occurred in relation to vasculature pathology. A few chimpanzees in this study had pathology comparable to moderate or severe CAA (Edler et al., 2017). CAA occurs in a large proportion of AD brains and is caused by the accumulation of Aβ in small arteries and leptomeninges and plays a role in the development of microvascular lesions in the aging brain (Joachim, Morris, & Selkoe, 1988; Kövari, Herrmann, Hof, Bouras, 2013; Thal, Griffin, de Vos, & Ghebremedhin, 2008; Vinters, 1987; Vinters, Wang, & Secor, 1996). Therefore, some of the astrogliosis observed in the neocortex and HC is likely in response to an increase in vascular amyloid deposition.

Layer I of the PFC showed a significant increase in GFAP-ir astrocyte density with increased APP/Aβ vessel volume, Aβ42 vessel volume, and neuritic cluster density. This layer also showed a significant correlation between GFAP-ir astrocyte density and amyloid, tau, and pathology scores (Table 3). This is comparable to human AD brains, which demonstrate prominent astrogliosis with the progression of the disease, and mostly in association with amyloid plaques (Beach & McGeer, 1988; Beach et al., 1989; Cullen, 1997; Duffy et al., 1980; Mancardi et al., 1983; Mandybur, 1989; Nagele et al., 2004; Rodríguez et al., 2009; Schechter et al., 1981; Serrano-Pozo, Gómez-Isla, Growdon, Frosch, & Hyman 2013). Also, Serrano-Pozo et al. (2011) reported that activated astrocytes increase linearly throughout the progression of AD in humans, which also correlates with an increase in NFT. Similar to humans, chimpanzee astrogliosis was not associated with pretangle density (Serrano-Pozo et al., 2011). However, unlike humans, astrogliosis appeared restricted to layer I of the PFC and a global increase in GFAP expression was not seen in relation to AD pathology in the neocortex.

Layer I astrocytes exhibited a unique vulnerability to AD-like pathology in chimpanzees. Layer I of the primate cortex contains interlaminar astrocytes, a distinctive class of astrocytes which show greater complexity in primates compared to other mammals (Colombo, 1996; Colombo et al., 1997; Colombo et al., 1998; Colombo et al., 2001; Colombo & Reisin, 2004; Falcone et al., 2018; Oberheim et al., 2009). The somata of these interlaminar astrocytes in layer I possess long processes that extend deep into the cortical layers, terminating in layers III and IV. These long processes are often described having a columnar or palisade organization (Colombo et al., 1995; Colombo, Sherwood, & Hof, 2004; Colombo & Reisin, 2004; Oberheim et al., 2009). The superficial cortical layers where these astrocytes are found exhibit a high density of neuritic plaques in AD (Delaère, Duyckaerts, He, Piette, & Hauw, 1991; Geddes et al., 1997). Though not quantified here, previous reports have shown that AD pathology can disrupt the columnar organization of these astrocytes, with a loss of palisade structure and prominent GFAP immunoreactivity (Colombo et al., 2002). Another study examining interlaminar astrocytes in the great apes (orangutans, gorillas, and chimpanzees) found that a small number of chimpanzees had disruptions to their interlaminar astrocytes, with two chimpanzees lacking the typical columnar organization as well as showing signs of astrogliosis (ages 5 and 37; Colombo et al., 2004).

Overall, we found that much of the astrogliosis associated with AD pathology in chimpanzees was specific to the PFC, and most pronounced in layer I. This region plays a role in both episodic and working memory, which are both impaired early in the pathogenesis of AD and also in mild cognitive impairment (MCI) of humans (Bauer & Fuster, 1976; Fuster & Alexander, 1971; Kane & Engle, 2002; Kubota & Niki, 1971; Miller & Orbach, 1972). MCI is defined as progressive memory loss distinct from normal aging and has been suggested as a transitional stage between healthy aging and AD (Almkvist et al., 1998; Baddeley, Bressi, Della Sala, Logie, & Spinnler, 1991; Belleville, Peretz, & Malenfant, 1991; Bowen et al., 1997; Hänninen et al., 1995; Morris et al., 2001; Peterson et al., 1999). Importantly, cognitive data are not available for these chimpanzees, and as such, we are unable to assess potential memory impairment due to astrogliosis in the PFC of chimpanzees. However, the astrogliosis present in layer I suggests that this might be a shared response to AD-type pathology in both chimpanzees and humans (Table 3).

In conclusion, the present data show that PFC layer I astrocytes demonstrate a unique and early vulnerability to pathology in chimpanzees. While we did not directly evaluate human tissue in this sample, this finding in chimpanzees seems to be consistent with what has been reported in human AD. In contrast, astrogliosis in the aged and pathological chimpanzee brain, as measured by GFAP-ir astrocyte density and soma volume, differed from the patterns reported in humans. However, brain aging in chimpanzees appears to be consistent with that of other nonhuman primates. While speculative, these differences could indicate specific vulnerabilities of the human brain to neurodegenerative diseases such as AD that could be mediated by divergent inflammatory cascades.