Amyloid-Beta, Tau, and Microglial Activation in Aged Felid Brains

Abstract

Alzheimer’s disease (AD) and its associated pathology have been primarily identified in humans, who have relatively large brains and long lifespans. To expand what is known about aging and neurodegeneration across mammalian species, we characterized amyloid-beta (Aβ) and tau lesions in five species of aged felids (n = 9; cheetah, clouded leopard, African lion, serval, Siberian tiger). We performed immunohistochemistry to detect Aβ40 and Aβ42 in plaques and vessels and hyperphosphorylated tau in the temporal lobe gyrus sylvius and in the CA1 and CA3 subfields of the hippocampus. We also quantified the densities and morphological types of microglia expressing IBA1. We found that diffuse Aβ42 plaques, but not dense-core plaques, were present more often in the temporal cortex and tended to be more common than Aβ40 plaques across species. Conversely, vascular Aβ was labeled more consistently with Aβ40 for each species on average. Although all individuals showed some degree of Aβ40 and/or Aβ42 immunoreactivity, only the cheetahs and clouded leopards exhibited intraneuronal hyperphosphorylated tau (i.e., pretangles), which was more common in the hippocampus. Reactive, intermediate microglia were significantly associated with total Aβ40 vessel area and pretangle load in the hippocampus. This study demonstrates the co-occurrence of Aβ and tau pathology in two felid species, cheetahs and clouded leopards. Overall, these results provide an initial view of the manifestation of Aβ and tau pathology in aged, large-brained felids, which can be compared with markers of neurodegeneration across different taxa, including domestic cats, nonhuman primates, and humans.

Graphical Abstract

To expand what is known about aging and neurodegeneration across mammalian species, we characterized amyloid-beta (Aβ) and tau pathology in five species of aged felids. We also quantified microglia expressing IBA1. Aβ42 deposition tended to appear in diffuse plaques, whereas Aβ40 was more often vascular. Although all individuals showed Aβ immunoreactivity, only cheetahs and clouded leopards also exhibited tau pretangles.

Introduction

Alzheimer’s disease (AD) is a progressive, irreversible neurodegenerative condition characterized by cognitive symptoms of dementia and two distinct and co-occurring pathological hallmarks: aggregations of amyloid-beta (Aβ) forming extracellular plaques and hyperphosphorylated tau protein within neurons resulting in neurofibrillary tangles (NFT) (Alzheimer et al., 1995; Asik et al., 2021; Bancher et al., 1989; Bussière et al., 2003; Glenner and Wong, 1984; Goedert, 1993; Montine et al., 2012). Additionally, microglia—the brain’s primary immune cells—are often found responding to Aβ plaques and NFT (see Edler et al., 2021 for review). Although humans are suggested to be unique in their predisposition to AD, recent studies have shown that aged chimpanzees can simultaneously possess both types of AD pathology (Edler et al., 2017). Age-related pathology has also been reported in other nonhuman primates, pinnipeds, canines, elephants, cetaceans, cattle, horses, sheep, and bears (see Gołaszewska et al., 2019; Youssef et al., 2016 for summary; Head et al., 2001; Takaichi et al., 2021). The current study aimed to investigate Aβ and tau pathology in the brains of felids. There is an important distinction to be made between brain aging pathology that may be nearly universal among mammals, as opposed to AD. AD has unique features, both on the cellular and behavioral level, that have not been reported in nonhuman species to the same severity, and the handful of other species that display AD-like pathology do not appear to undergo the extensive brain volume loss, neuronal death, or extreme dementia that characterizes human AD (e.g., Edler et al., 2020; Herndon et al., 1999; Heuer et al., 2012; Hof et al., 2002; Latimer et al., 2019; Picq et al., 2010).

Humans are thought to be particularly susceptible to AD because of our extended lifespan and very large and energetically costly brains (Bordone et al., 2019; Bruner & Jacobs, 2013; Bufill et al., 2013). One common explanation for the widespread prevalence of AD is that the disease is a nearly inevitable “metabolism reduction program” that comes naturally with increased age (Reser, 2009). When researchers study AD, though, models typically include animals with smaller brains and much shorter lifespans, such as laboratory mice and rats (e.g., Cohen et al., 2013; Götz et al., 2018; Pang et al., 2022; Rapoport & Nelson, 2011; Yokoyama et al., 2022; Youssef et al., 2016). Given that the underpinnings of AD may rely on the combination of a long lifespan and an energetically expensive brain, rodents likely do not possess the metabolic requirements or life history characteristics to accurately determine if the progression of age-related pathology to AD is unique to humans. Rodents also do not naturally develop Aβ plaques or NFT (Yokoyama et al., 2022). Even transgenic mouse models with human gene insertions tend to only develop amyloid pathologies (Poon et al., 2020), whereas tauopathy appears more difficult to replicate (Wenger et al., 2023), especially in the same model organism. In contrast, normal aging in nonhuman primates, who have relatively larger brains and longer lifespans compared to rodents, features the presentation of neurodegenerative pathology, dendritic attrition, synaptic loss, and cognitive decline (Hof et al., 2002; Hof & Morrison, 2004). Although severity and degree of co-occurrence varies, many primate species naturally develop the general characteristics of age-related pathology in their brains. Aβ plaques (often diffuse, not compact), vascular Aβ, and tauopathy have all been found to occur in aged chimpanzees, orangutans, and gorillas, as well as some non-apes, such as macaques, baboons, vervets, and marmosets (see Freire-Cobo et al., 2021 for review; Heuer et al., 2012).

Felid brain sizes and lifespans are more similar to primates than rodents, with maximum longevity ranging from 13 to 30 years of age (de Magalhães & Costa, 2009; de Magalhães et al., 2024; Radinsky, 1975). As such, this taxon offers a valuable opportunity to study typically age-related pathology in other large mammals and investigate whether the brain aging effects seen in apes and large monkeys extend to a clade that is separated from primates by approximately 100 million years of independent evolution. Aβ and tau pathology have previously been described in select felid species. In domestic cats, both Aβ and tau pathology have been documented (Chambers et al., 2015; Fiock et al., 2020; Gunn-Moore et al., 2006; Head et al., 2005; Vidal-Palencia et al., 2023), though the latter is rare and often limited to the entorhinal cortex. Importantly, Aβ plaques were diffuse, similar to that seen in normal human aging without disease, rather than the dense-core compact pattern that typically characterizes human AD (Malek-Ahmadi et al., 2016). Although primary tauopathies appear to be rare in felines, pretangles have been identified in domestic cats, predominantly in the cortex (Sordo et al., 2021; Vidal-Palencia et al., 2023). In addition, aged cheetahs have co-occurrence of Aβ pathology and NFT in the parahippocampal cortex and CA1 subfield of the hippocampus (Serizawa et al., 2012). Taken together, these few studies on domestic cats and large felids demonstrate that Aβ and tau pathology can occur and co-exist in a variety of species, although the specific presentation and progression differs.

In addition to Aβ and tau, microglia are involved in brain aging and AD. Microglia are the primary immune cells of the central nervous system, and while they are crucial for several key functions throughout life, their function changes with aging. In the healthy brain, microglia use ramified processes to survey the cellular environment (Nimmerjahn et al., 2005). When infection, trauma, or neurodegeneration occur, microglia undergo rapid changes in cell morphology, gene expression, and function in a process known as microglial activation (Gehrmann et al., 1992; Jørgensen, 1993; Kofler et al., 2012; Lehrmann et al., 1997; Morioka et al., 1992, 1993). Phenotypically, microglial activation results in decreased arborization, an enlarged cell soma, and shortened or total loss of cellular processes. In AD, microglia are implicated in the inflammatory response to the formation of Aβ plaques and NFT (Akiyama et al., 2000; Cunningham, 2013; Gandy & Heppner, 2013; Heneka et al., 2014; Hickman & El Khoury, 2014; Krstic & Knuesel, 2013; Perry & Holmes, 2014; Prokop et al., 2013; Sudduth et al., 2013). Microglia with intermediate or amoeboid morphologies are specifically associated with dense-core Aβ plaques and NFT in human AD and transgenic mouse models of AD (Frautschy et al., 1998; Hendrickx et al., 2017; Itagaki et al., 1989; Mattiace et al., 1990; Perlmutter et al., 1992; Stalder et al., 1999; Yin et al., 2017). It has been proposed that reactive microglia stimulate neurons to overproduce Aβ, which results in the formation of extracellular plaques and hyperphosphorylation of tau, thus creating a positive feedback loop involving microglial activation and AD pathology development (Meraz-Rios et al., 2013; Wyss-Coray & Rogers, 2012).

From an evolutionary perspective, studying the neurobiology of aging in felids is critical to further understanding what is conserved or unique across species in terms of neurodegeneration. Since it is only recently that nonhuman primates like chimpanzees were reported to develop both plaques and tangles reminiscent of AD, it would be valuable to learn whether more evolutionarily distant mammals also have the potential to spontaneously develop this pathology as they age. Obtaining brain samples from large felids is difficult and opportunistic because it relies on donations from zoos. Accordingly, this study examined nine aged individuals of five different felid species (cheetah, clouded leopard, African lion, serval, Siberian tiger; Table 1). Three brain regions were evaluated and chosen because they are particularly affected in AD (Braak & Braak, 1991; Edler et al., 2017; Gearing et al., 1996, 1997; Montine et al., 2012; Rosen et al., 2008): the gyrus sylvius (analogous to primate middle temporal gyrus) and the CA1 and CA3 subfields of the hippocampus. For these regions of interest, we examined Aβ plaques, vascular Aβ, and tau lesions (pretangles and NFT). Furthermore, we characterized microglia density and morphology within these regions. We show that different species of felids exhibit either Aβ pathology alone, or both Aβ and tau, and a significant relationship between pretangle pathology load and microglial activation appears to occur in the hippocampus.

Table 1.

Sample of felids and presence/absence of pathologies.

Common name	Species name	Age (years)	Sex	Brain weight (g)	Aβ42 plaque	Aβ42 vessel	Aβ40 plaque	Aβ40 vessel	Pretangle	NFT
Cheetah	Acinonyx jubatus	15	M	140 g	+	+	+	+	+	+
Cheetah	Acinonyx jubatus	17	F	113 g	+	+	−	+	+	−
Clouded leopard	Neofelis nebulosa	18	F	82 g	+	+	+	+	+	+
Clouded leopard	Neofelis nebulosa	24	F	73 g	+	+	+	+	+	−
African lion	Panthera leo	18	F	186 g	+	+	+	+	−	−
African lion	Panthera leo	19	F	249 g	+	+	+	+	−	−
Serval	Leptailurus serval	18	F	56 g	+	+	+	−	−	−
Serval	Leptailurus serval	20	F	46 g	+	+	+	+	−	−
Siberian tiger	Panthera tigris altaica	18	M	282 g	+	+	+	+	−	−

See Figure 1 for phylogenetic relationships. Abbreviations: +, present; −, absent; M, male; F, female; NFT, neurofibrillary tangle.

Materials and Methods

Specimens and Sample Processing

Formalin-fixed postmortem brain specimens from nine aged felids (see Table 1) were used. All nine individuals originated from Association of Zoos and Aquariums-accredited zoos. The sample was selected to include individuals that died within approximately the last quarter of their species’ estimated maximum potential lifespan (based on data in AnAge: Animal Ageing and Longevity Database; https://genomics.senescence.info/species). Causes of death were reported by the veterinary facilities these felids came from, and included euthanasia due to symptoms from cancers, kidney disease, spinal spondylosis, and musculoskeletal lesions. Five different species were included (cheetah, clouded leopard, African lion, serval, Siberian tiger). Behavioral and cognitive data were not available for the individuals in this sample. After initial fixation for at least two weeks, brains were stored in a 0.1 M phosphate-buffered saline (PBS) containing 0.1% sodium azide at 4° C. A slab through the middle region of the brain hemisphere containing hippocampus, temporal cortex, and other associated structures was dissected and prepared for histological sectioning. Samples were cryoprotected in a graded series of 10, 20, and 30% sucrose solutions, and serially cut into 40 μm-thick sections using a Leica SM2000R freezing microtome (Buffalo Grove, IL). Sections were placed into individual centrifuge tubes containing cryoprotection solution (30% distilled water, 30% ethylene glycol, 30% glycerol, and 10% 0.244 M PBS), numbered sequentially, and stored at −20°C until histological or immunohistochemical processing. Every tenth section was stained for Nissl substance with a 0.5% cresyl violet solution to reveal cell bodies and define cytoarchitectural boundaries.

Identification and Regional Sampling

All regions were identified using Nissl-stained sections. Brain areas for investigation were selected based on prior research in nonhuman primates (Edler et al., 2017) and were regions that have been identified to be most impacted by neurodegeneration according to Braak staging of NFT and Thal phases of Aβ deposition in humans with AD (Braak & Braak, 1991; Thal et al., 2002). Regions sampled for quantification included layer III in the gyrus sylvius (GS) and the stratum pyramidale of the CA1 and CA3 subfields of the hippocampus (Figure 2) (Pakozdy et al., 2014). Layer III was a focus because prior research has shown that pyramidal neurons in neocortical layers III and V, as well as the hippocampal stratum pyramidale, display extensive neuronal and synaptic loss during AD in humans (Hof et al., 1990; Scheff & Price, 2006). Furthermore, Aβ plaques and NFT are typically more robust in those cortical layers (Akram et al., 2008; Bussière et al., 2003). Other brain regions were carefully examined and are described qualitatively.

Figure 2. Nissl-stained section showing regions of interest.

Coronal hemisphere of clouded leopard (section thickness 40 μm). Abbreviations: GS, gyrus sylvius. Scale bar 5 mm.

Immunohistochemistry

For each individual, every tenth section through the region of interest was immunohistochemically processed for tau phosphorylated at Ser202 and Thr205, Aβ40, or Aβ42 (Table 2 for antibody information) following previously established protocols (Edler et al., 2017; Raghanti et al., 2008, Raghanti et al., 2009). Neocortical tissue from the brain of a chimpanzee that was found to have both NFT and dense-core Aβ plaques was used as a positive control to ensure antibodies were functioning as expected. Free-floating sections were pretreated for antigen retrieval by incubation in citric acid for 30 min at 37° C and 20 min at room temperature (RT). Endogenous peroxidase was quenched in a hydrogen peroxide (2.5%) and methanol (75%) solution for 20 min at RT, and sections were pre-blocked for 1h in a solution of 0.1 M PBS (pH 7.4), 0.6% Triton X, 4% normal goat serum, and 5% bovine serum albumin at RT. Sections were then placed in a primary antibody diluted in PBS for 24 h at RT. Next, sections were incubated in a biotinylated secondary antibody (1:200 dilution) in a solution of PBS and 2% normal goat serum (1 h, RT), followed by an avidin-biotin-peroxidase complex (1 h, RT; Vector Laboratories, Burlingame, CA) and 3,3′-diaminobenzidine (Vector Laboratories). After a final rinse, sections were mounted on slides, dehydrated (10 min each 50, 70, 95, 100% EtOH and 20 min Citra Solv), and coverslipped using Permount mounting medium (Fisher Scientific, Hampton, NH). Sections were selected and processed the same way for microglia expressing IBA1 (Table 2 for antibody information).

Table 2.

Primary antibodies.

Antigen	Host species	Antibody	Pathology	Dilution	Company	Catalog #	RRID
Aβ40	Rabbit	Rabbit polyclonal to a seven aa peptide sequence from the C terminus of human beta-amyloid 1–40	Aβ40 plaques and vessels	1:15,000	Millipore Sigma	AB5074P	AB_91672
Aβ42	Rabbit	Rabbit IgG monoclonal raised against C terminus of human Aβ A4 protein	Aβ42 plaques and vessels	1:25,000	ThermoFisher Scientific	H31L21	AB_2532306
Tau	Mouse	Mouse IgG1 monoclonal (AT8) to a partially purified human PHF-tau with epitope at residues pSer202/pThr205	Pretangles, NFT, tau neuritic clusters	1:2,500	ThermoFisher Scientific	MN1020	AB_223647
IBA1	Rabbit	Rabbit IgG polyclonal raised against the C terminus of amino acids 81–93 of human AIF-1	Ramified, intermediate, and amoeboid microglia	1:10,000	Wako	019–19741	AB_839504

Data Acquisition

Quantitative analysis was performed using computer-assisted stereology with a Zeiss Axioplan 2 photomicroscope equipped with a digital camera and Stereo Investigator software (MBF Bioscience, Williston, VT, RRID:SCR_002526) by a single observer. Appropriate sampling parameters for each probe were determined based on prior studies (Edler et al., 2017, 2018).

Densities for neurons and glia using Nissl-stained sections were obtained using the optical fractionator probe with a 40x objective lens (N.A. 0.75) under Köhler illumination. Grid size was 200 × 200 μm for CA1 and CA3, and 300 × 300 μm for GS. Disector height was 6 μm, with an upper guard zone of 1 μm. Beginning at a random starting point, three equidistant sections (every 10th section) per region of interest and individual were selected for analysis. For consistency, the sections from different individuals were all selected to be at similar coronal levels of hippocampal anatomy. Mounted section thickness was measured every eighth sampling site. A different marker was used for neurons and glia, and counting frame size was 65 × 65 μm for neurons and 30 × 30 μm for glia.

Densities for microglia expressing IBA1—including ramified, intermediate, and amoeboid morphology—were quantified separately using the optical fractionator probe with the same parameters as above and a counting frame of 65 × 65 μm. The nucleator probe was used to measure microglia soma volumes for every fourth marker placed. A different marker for each morphology type was placed when encountered within the optical disector frame.

Densities of pretangles and NFT containing tau phosphorylated at Ser202 and Thr205 were similarly obtained using the optical fractionator probe. A different marker was placed for each pathology type (i.e., pretangle, NFT) when encountered within the optical disector frame. Densities (per mm³) for each region were calculated as the number of markers counted divided by the total volume of the sum of disectors investigated. To correct for tissue shrinkage in the z-axis, the height of the disector was multiplied by the ratio of section thickness to the actual weighted mean thickness after mounting and dehydration (Dorph-Petersen et al., 2001). For certain analyses, densities for CA1 and CA3 were averaged to calculate overall average hippocampal (HC) density. Densities for all three regions (CA1, CA3, GS) were averaged for total tau pathology density.

Percent (%) area occupied by Aβ40 or Aβ42 in plaques and vessels was measured in all regions of interest using the area fraction fractionator (AFF) probe, which is founded on a Cavalieri point-counting system. Using a 10x objective lens (N.A. 0.25), markers were placed on a grid of points (300 × 300 μm for HC, 500 × 500 μm for GS) overlaying the sampling area. Every point on the grid received one of three potential markers: non-Aβ, Aβ plaque, or Aβ vessel. Area fractions were calculated by the AFF probe and reported as a percentage for each region per individual. Area fractions occupied by Aβ40 or Aβ42 plaques for CA1 and CA3 were summed to calculate hippocampal percentage, when needed. Fractions from all three regions (CA1, CA3, GS) were summed for total % area occupied by Aβ40 or Aβ42 plaques. The same procedure was conducted for Aβ40 and Aβ42 vessels.

Pathology Identification and Statistical Analysis

Aβ and tau pathology were defined as in Edler et al. (2017), who based their protocol on Serrano-Pozo et al. (2011). Aβ plaques were defined as extracellular deposits of insoluble Aβ. Pretangles were defined as containing tau phosphorylated at Ser202 and Thr205, but in healthy-looking neurons with the presence of diffuse, punctate tau staining in the cytoplasm, well-preserved dendrites, and a centered nucleus. NFT contained intraneuronal aggregates of hyperphosphorylated tau, with a nucleus that was either displaced toward the periphery of the soma or was absent. Dendrites and axons were distorted, shortened, or absent in NFT. No tau dystrophic neurites were observed in this sample of felids. In addition to the three regions of interest (CA1, CA3, GS), each coronal section was carefully examined for the occurrence of pathology markers elsewhere. This resulted in quantification of pathology markers in the entorhinal cortex of both clouded leopards and both cheetahs.

To confirm the paired helical filament (or fibrillar) structure of presumed NFT, Thioflavin S staining was conducted as follows for relevant sections. Sections were rinsed in a 0.1 M PBS, mounted on slides, air dried, placed in Citri-Solv for 10 minutes, and hydrated in a series of ethanol solutions (100%, 95%, 70%, and 50%) followed by distilled water (Bussière et al., 2004; Edler et al., 2017; Guntern et al., 1992; Rajamohamedsait & Sigurdsson, 2012). Then, sections were incubated in a filtered 1% aqueous thioflavin S solution (Sigma, T1892). After, the tissue was dehydrated in a series of ethanol solutions. Next, sections were placed in a filtered 1% Sudan Black B solution (Sigma, 19966–4), dissolved in 70% ethanol, to eliminate autofluorescence and placed in Citri-Solv. After coverslipping, slides were stored at 4° C for 48–72 hours before images were taken on the Zeiss Axioplan 2 photomicroscope (MBF Bioscience, Williston, VT). A FITC filter (blue-green) for fluorescence was used to match the optimal excitation (~440–480 nm) and emission wavelengths for thioflavin S (~515–550 nm) (Sun et al., 2002).

Due to the opportunistic nature of the sample, all nine felids were treated as one group for data analysis purposes. To try to alleviate the inherent differences that come with five different species in a single sample, repeated measures ANOVA or its non-parametric counterpart, the Friedman test, were used. Using this approach, the goal was to test whether patterns were consistent across individual brains, rather than generalizing across all nine individuals. Regression analyses were performed to determine relationships between tau densities and Aβ variables. Statistical analyses were conducted and figures were generated using R-4.2.3 for Windows (R Core Team, 2023), and level of significance (α) was set at 0.05.

Results

Aβ42 Plaques and Vasculature

All Aβ42 plaque deposits were diffuse, as determined by Thioflavin S staining, with no dense-core morphology. See Table 3 for a summary of Aβ42 area fractions across individuals. The majority of Aβ42 plaques were found in the GS of the older cheetah (0.023% plaque area; Figure 3A), the older clouded leopard (0.006% plaque area; Figure 3B), and the younger serval (0.014% plaque area; Figure 3C). The Siberian tiger (Figure 3D) had much stronger immunoreactivity in the GS compared to the hippocampus for both Aβ42 plaques and vessels (Figure 5), whereas the lions (Figure 3E) had weak Aβ42 immunoreactivity in both the hippocampus and GS. The older cheetah had the highest % area for Aβ42 cortical plaques (0.023% plaque area, Table 3), combined with an Aβ42 vessel load in its hippocampus (0.008% area), whereas the younger cheetah tended to have a relatively high amount of Aβ42 plaques in hippocampal regions (% area = 0.01%, Table 3). The younger clouded leopard (Figure 3F) consistently had weaker Aβ42 immunoreactivity than its older conspecific. The older serval was another individual that possessed a high % area of Aβ42 vessels in the hippocampus (% area = 0.01%, Table 3). Aβ42 plaques cumulatively occupied more % area than vessels across the three regions of interest (CA1, CA3, GS) in all individuals except the older clouded leopard and older serval (Table 3, Figure 5).

Table 3.

Overview of Aβ plaque and vessel area fractions and tau pathology densities across felids.

		Cheetah		Clouded leopard		Lion		Serval		Siberian tiger
		15-year-old	17-year-old	18-year-old	24-year-old	18-year-old	19-year-old	18-year-old	20-year-old	18-year-old
Aβ42 plaque area fraction (%)	CA1	0.0036	0	0	0.0007	0.0046	0	0.0013	0.001	0.0011
	CA3	0.0061	0.002	0.0013	0.0021	0.0008	0	0.0016	0.0028	0.0005
	Hippocampus (CA1 + CA3)	0.0097	0.002	0.0013	0.0028	0.0054	0	0.0029	0.0038	0.0016
	Gyrus sylvius (GS)	0.0029	0.023	0.0019	0.0064	0.003	0.0038	0.014	0.0073	0.019
	Entorhinal cortex (EC)	0.0077	0.005	0.0066	0.098	N/A	N/A	N/A	N/A	N/A
Aβ42 vessel area fraction (%)	CA1	0.0012	0.0038	0	0	0.0005	0.0013	0.0007	0.0029	0
	CA3	0	0.0041	0	0.0014	0	0.0007	0	0.0073	0
	HC	0.0012	0.0079	0	0.0014	0.0005	0.002	0.0007	0.0102	0
	GS	0	0.0009	0.0028	0.022	0.005	0	0.0009	0.0038	0.012
	EC	0.0006	0.0002	0	0.056	N/A	N/A	N/A	N/A	N/A
Aβ40 plaque area fraction (%)	CA1	0	0	0.0004	0.0006	0	0.0025	0.002	0	0
	CA3	0	0	0	0	0	0	0	0	0
	HC	0	0	0.0004	0.006	0	0.0025	0.002	0	0
	GS	0.0079	0	0.0013	0	0.0013	0	0	0.0006	0.0044
	EC	0	0	0	0	N/A	N/A	N/A	N/A	N/A
Aβ40 vessel area fraction (%)	CA1	0.0026	0.0028	0.0027	0.0024	0.0056	0	0	0.0009	0.0011
	CA3	0	0.0013	0.0014	0.0006	0.0089	0.0005	0	0.0029	0.0023
	HC	0.0026	0.0041	0.0041	0.003	0.015	0.0005	0	0.0038	0.0034
	GS	0.0079	0.0018	0.0025	0.0048	0	0.0007	0	0.0018	0.0044
	EC	0.0014	0.0002	0.0005	0.0002	N/A	N/A	N/A	N/A	N/A
Pre-tangle density (mm 3 )	CA1	909.6	0	241.9	409.9	0	0	0	0	0
	CA3	202.4	0	0	125.6	0	0	0	0	0
	HC	1112	0	241.9	535.5	0	0	0	0	0
	GS	916.9	1084.8	0	0	0	0	0	0	0
	EC	1715.4	1139.1	1616	0	N/A	N/A	N/A	N/A	N/A
NFT density (mm 3 )	CA1	82.7	0	0	0	0	0	0	0	0
	CA3	0	0	0	0	0	0	0	0	0
	HC	82.7	0	0	0	0	0	0	0	0
	GS	0	0	0	0	0	0	0	0	0
	EC	0	0	346.3	0	N/A	N/A	N/A	N/A	N/A

Hippocampus (HC) ROI: CA1 and CA3. Cortical ROI: gyrus sylvius (GS). Additional ROI: entorhinal cortex (EC), for cheetahs and clouded leopards only. N/A to signify that quantification was not done in the EC of the other species. All values given as % area fraction or density per mm³.

Figure 3. Examples of Aβ42 diffuse plaques and vessels across nine felids.

A. Cheetah GS, diffuse plaques. B. Clouded leopard GS, vessels and diffuse plaques. C. Serval GS, diffuse plaques. D. Siberian tiger GS, diffuse plaque. E. Lion CA1, vessel. F. Clouded leopard (different individual than B.) GS, diffuse plaques. Scale bar 250 μm for A, B, F; 25 μm for C-E. Abbreviations: GS, gyrus sylvius.

Figure 5. Aβ pathology area fractions by region.

Average total Aβ plaque and vessel area fraction (%) across three regions of interest (CA1, CA3, GS) by type of pathology (vessel, plaque) and form of Aβ (40, 42).

Aβ40 Plaques and Vasculature

In general, there was weaker immunoreactivity to the Aβ40 antibody compared to Aβ42, and when present, it was mostly in the form of vascular Aβ. See Table 3 for a summary of Aβ40 area fractions across individuals. The younger lion (Figure 4A) and the younger cheetah (Figure 4E) had the most widespread Aβ40 immunoreactivity in the GS and hippocampus, respectively. Both clouded leopards (Figures 4B, F) and the Siberian tiger (Figure 4D) also had quite prevalent vascular Aβ40 across the regions of interest (Table 3). The only exception to the majority of Aβ40 presenting as vessels was the younger serval, who only had diffuse Aβ40 plaques in the CA1 (0.002% area; Figure 4C). The older serval, though, had Aβ40 vessels in both hippocampal (0.004% area) and cortical (0.002% area) regions.

Figure 4. Examples of Aβ40 diffuse plaques and vessels across nine felids.

A. Lion CA1, vessel. B. Clouded leopard GS, vessels. C. Serval CA1, diffuse plaques. D. Siberian tiger GS, vessels. E. Cheetah GS, diffuse plaques. F. Clouded leopard (different individual than B) GS, vessels. Scale bar 25 μm for A-C, E-F; 250 μm for D. Abbreviations: GS, gyrus sylvius.

Aβ Area Fractions

Each type of pathology occupied less than 0.02% area of the three regions of interest in nearly all cases, the exceptions being 0.023% in the older clouded leopard (driven by 0.022% cortical Aβ42 vessel area) and 0.025% in the older cheetah (led by 0.023% cortical Aβ42 plaque area). Average area fraction occupied by both Aβ forms of plaques and vessels was not significantly different using Friedman test (non-parametric repeated measures ANOVA) (Figure 5). When assessing total Aβ load across all three regions of interest, a non-significant trend was noted toward greater Aβ42 plaques than Aβ40 and more Aβ40 vessels than Aβ42. Overall, the predominant forms of pathological lesions were Aβ42 plaques and Aβ40 vessels. This was also the case in the hippocampus, while in the GS it was Aβ42 plaques and vessels. On average, each region possessed a greater % area occupied by Aβ42 plaques compared to Aβ42 vessels, and a greater % area occupied by Aβ40 vessels compared to Aβ40 plaques (Figure 5).

There was no significant relationship between % area occupied by Aβ40 plaques and vessels or Aβ42 plaques and vessels (Figure 6). Although there was no significant difference between total % Aβ42 plaque and vessel area, or total % Aβ40 plaque and vessel area (Friedman test p-value > 0.05), there does appear to be a slightly positive trend by isoform, particularly for Aβ40 (rs = 0.31, Figure 6). Aβ40 and Aβ42 vessel % area very slightly trended toward a positive relationship but was not statistically significant (rs = 0.13, p > 0.05; Figure 7). Aβ40 and Aβ42 plaque % area had no true relationship, and Spearman’s rho was negative (rs = −0.067). None of the total Aβ-related pathology loads were significantly correlated with one another in any combination, using Spearman’s rank correlation.

Figure 6. Total Aβ vessel vs plaque area fraction.

Total % plaque area vs total % vessel area, by Aβ type.

Figure 7.

A. Total Aβ40 vs Aβ42 plaque area percent. Total % Aβ40 plaque area vs total % Aβ42 plaque area across three regions of interest (CA1, CA3, GS). B. Total Aβ40 vs Aβ42 vessel area percent. Total % Aβ40 vessel area vs total % Aβ42 vessel area across three regions of interest (CA1, CA3, GS).

Every type of Aβ pathology was analyzed for possible correlation with absolute age and brain weight. A significant relationship between absolute age and total Aβ42 vessel load was found (rs = 0.67; p = 0.047; Figure 8). No other significant relationships were observed between absolute age and pathology burdens. Also of note, the 15-year-old cheetah had the highest Aβ40 plaque load (0.008% area, all in the GS) despite being the youngest of the nine felids. There was indeed a negative trend between absolute age and Aβ40 plaque area fraction (data not shown). These results must be considered cautiously since these are absolute ages, which might represent somewhat different stages of the neurodegenerative process in each species. There was no statistically significant correlation between brain weight and Aβ pathology load of any kind, although both types of total Aβ40 pathology burden had the largest Spearman’s rho (Aβ40 plaque rs = 0.54, Aβ40 vessel rs = 0.42).

Figure 8. Absolute age vs total Aβ42 vessel area fraction.

Absolute age (in years) vs total % Aβ42 vessel area across three regions of interest (GS, CA1, CA3).

Pretangle and NFT Densities

Only the clouded leopards (Figures 9A, D) and cheetahs (Figure 9B–C) displayed tau pathology in the regions of interest. See Table 3 for a summary of pretangle and NFT densities across individuals. One of the cheetahs is the youngest sample in absolute age (15 years old; Figure 9B), but the clouded leopard in Figure 9A is the oldest at 24 years old. The younger cheetah possessed significant pretangles in all three regions of interest (CA1 = 910/mm³, CA3 = 202/mm³, GS = 917/mm³, Table 3), whereas the older cheetah only had pretangles in the GS (1085/mm³, Table 3). Both clouded leopards had pretangles solely in the hippocampal regions, with the older individual possessing a higher density (536/mm³, Table 3).

Figure 9. Examples of tau pretangles and NFT in the CA1, GS, and entorhinal cortex of three felids.

A. Clouded leopard CA1, pretangles. B. Cheetah CA1, pretangles. C. Cheetah (different individual than B) EC, pretangles. D. Clouded leopard (different individual than A) CA1, pretangles. E. Clouded leopard (same individual as D) coronal hemisphere section (40 μm thick), with visibly darker staining in EC. F. Cheetah EC, pretangles (same individual as B). G-I. Clouded leopard (same individual as D-E) EC, NFT. Scale bar 25 μm for A-C, G-I; 250 μm for D, F; 5 mm for E. Abbreviations: EC, entorhinal cortex. GS, gyrus sylvius.

One of the clouded leopards (Figures 9D–E, G-I) and both cheetahs (Figures 9B–C, F) showed strong hyperphosphorylated tau staining in the entorhinal cortex (EC). As this immunoreactivity was even visible without magnification in some individuals (Figure 9E), we elected to also quantify Aβ and tau pathology in the EC of both cheetahs and clouded leopards. The cheetah EC displayed pretangles (younger individual’s pretangle density = 1715/mm³, older individual’s pretangle density = 1139/mm³), but the younger clouded leopard, that had a lower pretangle density (241.9/mm³) than its older conspecific in the CA1 and CA3, showed both pretangles (1616/mm³) and NFT (346/mm³) in the EC. Despite this interesting finding, no correlation was identified between tau pathology and Aβ pathology in the EC of the same individuals. Figure 10 gives an example of Aβ42 plaque % area and pretangle density in the EC (rs = 0.63, p = 0.069). All areas of the cortex present in the sampled coronal sections were carefully examined for the presence of pathology markers as well, but no other striking occurrences were found.

Figure 10. Aβ42 plaque area fraction vs pretangle density in EC.

Aβ42 plaque % area vs tau pretangle density (per mm³) in the entorhinal cortex (EC). Lion, serval, and Siberian tiger all at (0, 0).

Brain weight and absolute age did not correlate with pretangle pathology load (data not shown). Additionally, there was no significant relationship between cortical and hippocampal pretangle densities.

Co-occurrence of Aβ and Tau Pathologies

Using Spearman’s rank analysis, total pretangle density was not associated with total Aβ pathology of any kind (data not shown), nor was cortical or hippocampal pretangle density with cortical or hippocampal Aβ pathology of any kind (example in Figure 11). No significant correlation (p > 0.05) was found between pretangle density and Aβ40 vessel % area across regions. Instead, we observed that for a similar vessel area fraction (0.002–0.004%), pretangle density can be quite varied in different felids, including between members of a single species, such as clouded leopard. Although one clouded leopard had a vessel fraction of 0.003% in the hippocampus and the other had 0.004%, the former had a higher pretangle density of 536/mm³, whereas the latter had 242/mm³. Conversely, despite a similar pretangle density in the GS of both cheetahs (917/mm³ and 1085/mm³), there is a strong discrepancy in Aβ40 vessel area fractions (0.008% and 0.002%, respectively), wherein the younger male cheetah had more Aβ40 pathology.

Figure 11. Pretangle density vs Aβ40 vessel area fraction by region.

Pretangle density (per mm³) in hippocampus (HC; CA1 and CA3) and cortex (GS) vs Aβ40 vessel area fraction in hippocampus and GS.

Regional Neuron and Glia Densities

Neuron and glia densities were quantified in the three regions of interest (CA1, CA3, GS) (Figure 12). Brain weight did not correlate with neuron density in any brain region (p > 0.05; Figure 13). The cheetah with the youngest absolute age (15 years old) consistently was above the regression line for all three regions, indicating relatively higher neuron density. One clouded leopard, the oldest individual at 24 years, is only below the regression line for CA1 neuron density. CA1 and GS consistently trended toward higher neuron densities than CA3 (repeated measures ANOVA, p = 0.076, F = 3.043; Figures 13, 14A). Although the ANOVA was not statistically significant, post-hoc analysis revealed differences between CA3 and GS neuron densities in particular (non-adjusted p-value = 0.02; Bonferroni adjusted p-value = 0.058). Glia density was significantly greater in CA3 compared to CA1 and GS, according to repeated measures ANOVA (Figure 14B; p = 7.86e-05, F = 18.073).

Figure 12. Clouded leopard Nissl stain of neurons and glia in regions of interest.

A. CA3 stratum pyramidale. B. CA1 stratum pyramidale. C. Gyrus sylvius, layer III. Cyan arrow, neuron; red arrow, glia. Scale bar 25 μm.

Figure 13. Log brain weight vs log neuron density.

Total log neuron density (per mm³) across three regions of interest (CA1, CA3, GS) versus log brain weight (in grams).

Figure 14.

A. Neuron density by region. B. Glia density by region. C. Glia:neuron ratio by region. Note the different y-axis scale for each subset.

When comparing glia-neuron ratios across the three regions of interest, repeated measures ANOVA showed significant variation (p = 0.001, F = 10.74) and post-hoc pairwise comparison showed that the CA3 glia-neuron ratio was higher than both CA1 and GS ratios (Figure 14C; p = 0.004, Bonferroni corrected p = 0.012), likely driven by a combination of relatively higher glia densities and relatively lower neuron densities. Aβ and tau pathologies in felids were more prevalent in the CA1 and GS, regions that appear to have high neuron densities and low glia densities, whereas the opposite pattern is present in the CA3, which generally had a lower pathology burden.

No significant correlation was noted between glia and intermediate microglia densities or total microglia densities across brain regions (data not shown).

Regional Densities of Microglia

Microglia were classified as ramified, intermediate, or amoeboid morphology based on cell soma size and number of processes (examples in Figure 15). No significant difference among regions was detected for any of the microglia morphology densities, according to repeated measures ANOVA (Figure 16A–C; all p-values > 0.10, all F-statistics ≤ 2.4). In the case of amoeboid microglia, a non-parametric Friedman test was used because of an extreme outlier in the CA3 of the older clouded leopard.

Figure 15. Examples of different microglial morphologies in the cheetah.

Ramified (A), intermediate (B), and amoeboid (C) microglia. Scale bar 25 μm.

Figure 16.

Density of (A) ramified, (B) intermediate, and (C) amoeboid microglia by region. Note the different y-axis scale for each morphology type. P-value > 0.05 for all three activation states. D. Absolute microglia morphology densities by region.

In all three regions of interest, the predominant microglia morphology was intermediate followed by ramified (Figure 16D). Very few amoeboid microglia were observed among the felids, although their density was slightly higher on average in the GS compared to hippocampal regions. Average microglia soma volume across individuals was highest for amoeboid morphology at 5163 mm³, with intermediate morphology having an average soma volume of 504 mm³, and ramified morphology soma volume of 72 mm³. As expected, there is a clear trend for amoeboid microglia soma volume to be larger than both intermediate and ramified states, with intermediate also having larger soma volumes than ramified on average (Torres-Platas et al., 2014).

Intermediate microglia were significantly associated with total Aβ40 vessel area fraction (Figure 17A; rs = 0.89, p = 0.001), as well as pretangle load in the hippocampus (Figure 17B; rs = 0.82, p = 0.007). Neither amoeboid nor total microglia densities were significantly correlated with any type of Aβ or tau pathology.

Figure 17.

A. Intermediate microglia and total Aβ40 vessel area fraction. Intermediate microglia densities (per mm³) vs total % Aβ40 vessel area across three regions of interest (GS, CA1, CA3). B. Intermediate microglia and pretangle densities in HC. Intermediate microglia densities (per mm³) in hippocampus (HC; CA1 and CA3) vs pretangle density (per mm³) in HC.

Discussion

This is the first characterization of brain aging pathology in clouded leopard, serval, lion, and Siberian tiger, and provides additional data on cheetah which has previously been studied (Serizawa et al., 2012). These findings establish that Aβ40 and Aβ42 vascular and diffuse plaque deposits can be found in these species, either independently or co-occurring with hyperphosphorylated tau in some cases. Furthermore, this study supports the work of Serizawa et al. (2012), who found Aβ and intracellular tau in a sample of aged captive cheetahs. The incidence of AD-like pathology is still rare within felids, with vascular and plaque forms of Aβ40 and Aβ42 each only occupying less than 0.025% of total area fraction of the CA1, CA3, and GS on average. This is a lower percentage than chimpanzees, which in one study were calculated to have Aβ40 area fractions of 0.28% in the neocortex and 0.12% in the hippocampus, and Aβ42 area fractions of 0.24% in the neocortex and 0.08% in the hippocampus (Edler et al., 2017). While tau pathology was comparatively rare in felids, only appearing in half the sample, the immunoreactivity was generally stronger than the diffuse nature of Aβ40 and Aβ42. Pretangles and NFT containing tau phosphorylated at Ser202 and Thr205 were present in the cheetahs and clouded leopards, although the individuals each displayed this pathology to vastly different extents.

Aβ Pathology in Felids

The presentation of Aβ in plaques was diffuse across the sample, as has been observed in non-cognitively impaired older humans and aged individuals from other species, rather than the dense-core manifestation that truly characterizes AD. Elderly domestic cats with Aβ deposits also only have diffuse plaques (Sordo et al., 2021). Although the trends were not significant, there were in total more Aβ42 plaques than Aβ40 and more Aβ40 vessels than Aβ42, on average. The presence of more Aβ42 plaques aligns with the literature on AD in humans, where diffuse plaques are seen to be exclusively positive for Aβ42, but there is a strong correlation between Aβ40 positivity and mature plaques (Iwatsubo et al., 1994). Since all plaques were diffuse, it follows that the individuals’ Aβ42 immunoreactivity was stronger. In terms of vascular Aβ, humans exhibit Aβ vessels with stronger Aβ40 than Aβ42 immunoreactivity (Gravina et al., 1995), and this appears to be the case in this group of felids, as well. Alternatively, some researchers argue that vascular Aβ deposition is not directly associated with plaque formation in humans (e.g., Lippa et al., 1993).

Although there were no statistically significant differences, most often Aβ42 pathology—both vessels and plaques—had higher density in the cortex (GS) compared to hippocampal regions. GS Aβ42 % vessel area fraction is correlated with total % vessels (p < 0.05), suggesting the majority of total % area fraction is from the cortex. Elderly cats also mainly had Aβ deposits in the cortex, with only occasional hippocampal aggregates (Sordo et al., 2021), suggesting that felid Aβ aggregation only progresses to the hippocampus at later stages of pathology formation. This is reiterated in Fiock et al.’s (2020) findings that domestic cats had more Aβ accumulation in neocortical regions, Edler et al.’s (2017) findings that Aβ40 and Aβ42 plaques were more prevalent in neocortical areas of chimpanzees, and the Thal phases of Aβ deposition in humans (Thal et al., 2002). Perhaps hippocampal Aβ deposits occur later in pathology expansion for felids and felines, as they do in humans and nonhuman primates.

Tau Pathology in Felids

The progression of tau pathology in this study’s sample of felids may be more like primates, including humans, than domestic cats. In humans, NFT are reported to occur most frequently in the entorhinal and perirhinal cortices and the CA1 and subiculum fields of the hippocampus (Arriagada et al., 1992; Braak & Braak, 1991; Montine et al., 2012). Similarly, NFT are more widespread in the hippocampus, as opposed to the cortex, of chimpanzees (Edler et al., 2017). This aligns with the observation of NFT in the EC of one clouded leopard and in the CA1 of one cheetah. No NFT were seen in neocortical areas, suggesting that the early stages of tau pathology progression may be similar in felids, humans, and nonhuman primates. Pretangles in felids were also more common in the hippocampus, particularly CA1. Hippocampal (CA1, CA3) pretangle density was positively correlated with total pretangle density (p < 0.05), suggesting that the hippocampus is likely one of the earliest regions affected by tau pathology. This is unlike aged domestic cats and chimpanzees, who had more tau pretangles in the cortex compared to the hippocampus (Edler et al., 2017; Sordo et al., 2021). Felids tend to have a higher ratio of pretangles to NFT compared to that seen in human AD and lack dystrophic neurites, similar to domestic cats (Fiock et al., 2020), although this type of pathology was seen in aged chimpanzees (Edler et al., 2017) and pinnipeds (Takaichi et al., 2021).

Aβ and Tau Pathology Co-occurrence

Only cheetahs and clouded leopards were species that simultaneously possessed both Aβ and tau pathology. There were no significant correlations between Aβ and tau pathology of any kind, as was the case in studies of AD-like pathology in aged chimpanzees, gorillas, and crabeating macaque (Edler et al., 2017; Oikawa et al., 2010; Perez et al., 2013). The co-occurrence of extracellular Aβ accumulation and intraneuronal NFT has previously been observed in aged domestic cats (Sordo et al., 2021) and Tsushima leopard cats (Prionailurus bengalensis euptilurus). The latter were reported to have granular Aβ42 deposits throughout the cerebral cortex and in the pyramidal cell layer of the hippocampus but lacked plaques, and most of the cats with Aβ42 pathology also developed NFT (Chambers et al., 2012).

With a sample size of nine felids from five different species, one can expect a high degree of variation in how Aβ and tau pathology manifests. The two clouded leopards, aged 18 and 24 years, presented with different pathological profiles, with the older clouded leopard possessing more Aβ pathology, while the younger had more tau pathology (when the EC is included). Although the relative quantities of these pathologies differ within a single species, this study presents evidence for their consistent co-occurrence. Interestingly, it was the youngest individual (15-year-old cheetah) that had the highest density of pretangles and one of the greater % areas occupied by Aβ42 and Aβ40 plaques. This is different from chimpanzees, where there was a positive association between pretangle density and age (Edler et al., 2017). This could be due to small sample size in the present study or possible felid interspecies differences in propensity for neurodegenerative pathology. What is consistent between the cheetahs and chimpanzees, though, is that more Aβ pathology tends to relate to more severe tau pathology (although this is not the case in other felids, like the clouded leopards).

The presence of some type of Aβ pathology in each of the nine individuals suggests that moderate Aβ aggregation may be a component of the typical aging process in felids, especially since the plaques were diffuse rather than dense. Generally, the cheetahs and clouded leopards had some of the highest pathology loads (with the exception of Aβ42 plaques, where the servals had more than the clouded leopards), but these individuals differed in what pathology was most prevalent in which region. For example, the younger clouded leopard had NFT and pretangles in the EC, but this was not related to more Aβ pathology in that same region. In contrast, the older clouded leopard had a higher Aβ load in the hippocampus but few pretangles. This indicates that having a high Aβ burden is not a prerequisite for tau pathology development, or vice versa. The lion and Siberian tiger share the most recent common ancestor out of the group of nine felids (Figure 1) and shared the tendency to have relatively low average pathology loads. Such a phylogenetic connection did not hold up for higher pathology burdens, because the cheetah and clouded leopard are not each other’s closest relatives (Kumar et al., 2022).

Figure 1. Phylogenetic tree of species in this study.

The last common ancestor of all species is estimated to have lived 12 million years ago. All images are Creative Commons license. Lifespan estimates from AnAge: Animal Ageing and Longevity Database (de Magalhães et al., 2024; https://genomics.senescence.info/species). Date estimates from TimeTree (Kumar et al., 2002; http://www.timetree.org).

* Based on data from 2005. The 24-year-old clouded leopard in the present sample is now the oldest recorded individual for this species.

** Based on data for Panthera tigris.

Image credits: “Cheetah” by Leszek.Leszczynski is licensed under CC BY 2.0. via Flicker. “Leptailurus serval serval” by Wynand Uys is licensed under CC BY 4.0. via iNaturalist. “Clouded Leopard (Neofelis Nebulosa), Santago” by spencer77 is licensed under CC BY 2.0. via Flickr. “Amur or Siberian Tiger (Panthera tigris altaica)” by Smudge 9000 is licensed under CC BY-SA 2.0. via Flickr. “Panthera leo (male) Colchester Zoo” by William Warby is licensed under CC BY 2.0. via Wikimedia.

Microglia and Pathology

In human brains with AD, both tau and Aβ pathology are associated with increased microglial activation (Heneka et al., 2014; Marlatt et al., 2014; Meraz-Ríos et al., 2013; Villegas-Llerena et al., 2016). Microglia play an important role in the removal of Aβ peptides through phagocytosis (Hansen et al., 2018; Hickman et al., 2008). In this study, microglia were found predominantly in the reactive intermediate state, followed by ramified and then reactive amoeboid, like data from aged chimpanzees (Edler et al., 2018) and the human cerebral cortex (Torres-Platas et al., 2014). There were no significant differences between microglia densities across the three regions of interest in a single brain, although glia density was significantly greater in CA3 compared to CA1 and GS across individuals. This mirrors findings in humans and nonhuman primates, where glia density tends to be higher in the CA3 than CA1 (Edler et al., 2020).

The only significant correlation between microglia with intermediate morphology and pathology was related to Aβ40 vessel area fraction across the three regions of interest and pretangle density in the hippocampus. While microglia appear to be effective in removing Aβ (Lee & Landreth, 2010; Richter et al., 2020), their neuroinflammatory properties may actually also exacerbate Aβ and tau pathology, particularly through the release of proinflammatory cytokines (d’Errico et al., 2022; Hansen et al., 2018; Hickman et al., 2008; Leyns & Holtzman, 2017; Meraz-Ríos et al., 2013; Wang et al., 2015). Microglial activation has been implicated in tau hyperphosphorylation and neurodegeneration in human tauopathies (Bellucci et al., 2011; Gebicke-Haerter, 2001; Ishizawa & Dickson, 2001). Maphis et al. (2015) argue that reactive microglia drive tau pathology in humans and mouse models of AD. Despite these potential negative effects, microglia have also been shown to mitigate tau spread (e.g., Odfalk et al., 2022). An association between microglial activation and Aβ and tau pathology has been found in aged chimpanzees. Aβ pathology correlated with microglial activation (Edler et al., 2017), and later research also supported a decrease in ramified microglia in chimpanzees with NFT (Edler et al., 2023). More research is needed to reveal whether the microglia in aged, pretangle-affected felid brains are exercising either or both positive and negative effects.

In terms of amyloid, previous work has shown that only dense-core, not diffuse, Aβ deposits had reactive microglia associated with them, suggesting that Aβ accumulation likely precedes the activation of microglia (Itagaki et al., 1989). This may be the case for these felids; there is not yet a significant rise in microglial activation because the Aβ plaques are still diffuse. Alternatively, it may be possible that felids are simply resistant to developing dense-core plaques, similar to most nonhuman primates that only present with diffuse Aβ plaque morphology (e.g., Gearing et al., 1997; Geula et al., 2002; Mufson et al., 1994; Perez et al., 2013). Due to the lack of dense Aβ aggregates in these felids, the primary role of the microglia likely remains maintaining the brain’s homeostasis during routine, non-pathological age-related disturbances.

Limitations and Future Directions

A major limitation of this study includes its small and irregular sample of felids. As a result, correlations are based on only a few individuals, and those with the most severe pathology may be driving any statistically significant results. Similarly, potentially significant trends of Aβ and tau pathology in felids could be concealed because of small sample size using different species. Importantly, age is often considered the greatest risk factor for developing neurodegenerative diseases (Reser, 2009), yet the composition of this sample prevented us from analyzing relative age because maximum lifespan estimates for each species are uncertain and based on limited data availability. Due to the opportunistic nature of sample availability, the present study was also not able to include young individuals of each species for pathological comparison. Another important consideration is the fact that most of the sample was composed of female felids, apart from the younger cheetah and the Siberian tiger. Given the paucity of male samples, sex differences were not analyzed. This would be a particularly interesting future direction, as it is well-established that AD in humans is more common in females (e.g., Beam et al., 2018; Ferretti et al., 2018; Zhu et al., 2021), and minor sex differences in Aβ burden have been described in the brains of nonhuman primates (Edler et al., 2017; Perez et al., 2013).

Unfortunately, there is no formal behavioral or cognitive data for the felids in this sample. As all felid samples originated from zoos, one must consider the fact that life in captivity is very different from a wild environment. Carnivores in captivity are known to live longer (Tidière et al., 2016), and the everyday cognitive stimuli they experience greatly differ from their natural habitats, all of which would presumably affect the course of brain aging. In their study of age-related pathology in captive cheetahs, though, Serizawa et al. (2012) did find that the animals with the most severe tau pathology showed clinical cognitive dysfunction. Similarly, domestic cats with Aβ and tau deposits have been shown to have cognitive dysfunction syndrome, which manifests as various behavioral abnormalities and dementia-like symptoms (Sordo & Gunn-Moore, 2021; Vidal-Palencia et al., 2023). These are both promising examples that age-related pathology in felids may also be linked to cognitive symptoms, but more research is needed.

Despite these limitations, this study characterized Aβ and tau pathology in several new felid species, and it establishes the building blocks for further exploration of neurodegeneration in non-primate mammals with relatively long lifespans and large brains. Aged felids from five different species were found to exhibit some degree of Aβ and tau pathology, most for the first time. Although all individuals showed some level of Aβ40 and/or Aβ42 immunoreactivity, only cheetahs and clouded leopards possessed hyperphosphorylated tau pathology in the regions of interest (CA1, CA3, gyrus sylvius). As Edler et al. (2017) established for chimpanzees, and Serizawa et al. (2012) for cheetahs, this study also found co-occurrence of Aβ and tau pathology in certain felids. There was a trend toward greater Aβ42 deposition in diffuse plaques (versus Aβ40), but more vascular Aβ40 (versus Aβ42). There was also a trend toward more cortical Aβ pathology, but more hippocampal and hippocampus-adjacent tau pathology, primarily in the form of pretangles. We also observed that intermediate microglia density was significantly associated with total Aβ40 vessel area fraction and hippocampal pretangle load, which leads to further questions about microglia’s positive and negative neuroinflammatory properties. Overall, there was variability in how brain aging pathology manifested within this group of felids—even within a single species—but it was shown that such pathology can appear in these species and its presentation has both similarities and differences to Aβ and tau pathology progression in domestic cats, nonhuman primates, and humans.