Background
As humans age, cognitive decline can negatively affect everyday life even in relatively healthy individuals [
1,
2]. This cognitive decline is associated with atrophy and reduced plasticity in brain regions such as the hippocampus, cortex, striatum, and, eventually, the cerebellum (c.f [
3‐
5].). In the hippocampus, for example, aging is associated with decreased neurogenesis (e.g., [
6]), as well as decreased neuronal size, complexity, and connectivity [
1]. The molecular mechanisms underlying this age-related deterioration are not yet fully understood, but increased neuroinflammation is thought to play a role.
Many types of neuroinflammatory and neuroimmune pathways have been implicated in brain aging, including signals related to the Complement Cascade, Toll-like receptor signaling, antigen presentation, and IFϏb/NFϏb signaling as well as macrophage and microglia activation (e.g., [
7,
8]). Microglia are the brain’s innate immune cell. Microglia normally reside in a quiescent state until a foreign antigen activates them, at which time they produce pro-inflammatory cytokines (including interleukins, tumor necrosis factors, interferons, and chemokines). Once the threat is resolved, microglia then produce anti-inflammatory cytokines that initiate the return to homeostasis [
9]. Within the aging brain, an increase in macrophage infiltration, microglia priming, and/or microglia activation has been observed [
9‐
13], along with a failure to return to homeostasis [
9,
11,
14,
15]. Although the early stages of an immune response can be neuroprotective, chronic activation can be detrimental [
2,
13,
16,
17]. The chronic activation of microglia that is observed in the aging brain, particularly that within the hippocampus, is thought to cause persistent neuroinflammation that detrimentally affects cognitive function [
6,
13,
15,
16,
18‐
20].
Despite widespread acceptance that persistent neuroinflammation contributes to age-related cognitive decline [
13], studies comparing protein expression of cytokines in young versus old brains are surprisingly limited in terms of the number of cytokines and brain regions studied and the fact that most studies only examined effects in either males or females. Studies suggest that aging is associated with increased tumor necrosis factor (TNF) protein expression in the hippocampus and prefrontal cortex [
21,
22] but not the amygdala [
22]. Although many studies have reported age-related increases in basal interleukin 1β (IL-1β) protein levels in the hippocampus of multiple mouse strains and Wistar rats [
23‐
29], others find no such age-related increases in basal IL-1β protein expression in the hippocampus of F344xBN F1 rats specifically [
22,
30‐
32]. That said, these same reports do find F344xBN F1 rats exhibit age-related exacerbation of IL-1β induction following a challenge [
22,
30‐
32]. Outside of the hippocampus, reports are more limited with studies in SAMR1 mice reporting age-related increases in baseline IL-1β protein expression in the cortex, hypothalamus, and brain stem [
29], but studies in C57BL/6J mice and F344xBN F1 rats reporting no such age-related increases in baseline IL-1β protein expression in the cortex, hypothalamus, or amygdala [
22,
23,
30,
31]. Findings around IL-6 are even more discrepant. Although 2 studies reported age-related increases in IL-6 protein expression in the hippocampus (BALB/cJ mice, [
33]; strain not specified, [
21]), 2 other studies did not (C57BL/6J males, [
23]; F344xBN F1 Rats, [
22]). Similarly, 3 studies reported age-related increases in IL-6 protein expression in the cortex (BALB/cJ mice, [
33]; C57BL/6J females, [
34]; strain not specified, [
21]); however, a fourth did not (C57BL/6J males, [
23]).
Discrepancies in the above finding may be related to environmental factors that differ between labs, the brain region examined, the species/strain used in the study, and/or the sex of the subjects employed. For example, diet, immune status (e.g., rearing in a pathogen-free facility), and other environmental factors have been shown to preferentially upregulate cytokine expression in aged vs. young adult brains [
11,
12,
35‐
37]. Pathway analyses suggest that age-related changes in inflammation- and immune-related gene expression may be more pronounced in the hippocampus compared to other brain regions and more pronounced in females versus males [
7,
8]. Indeed, we noted that studies employing female subjects alone or in combination with male subjects routinely report higher basal IL-6 mRNA or protein expression in old vs. young brains [
8,
34,
38,
39]; whereas, only a subset of studies employing male subjects alone report this age-related increase in basal expression (increased mRNA: [
36]; increased protein: [
21,
33,
40‐
42]; no change mRNA: [
11,
12,
35,
43]; no change protein: [
22,
23,
37]). As such, we sought to clarify how pro-inflammatory (IL-1α, IL-1β, IL-2, IL-3, IL-5 IL-6—see discussion, IL-9, IL-12p40, IL-12p70, IL-17, eotaxin, interfeuron δ, KC, MIP-1a, MIP-1b, rantes, TNFα, MCP-1, or GM-CSF) and anti-inflammatory cytokine expression (IL-4, IL-10, IL-13, G-CSF) may change with age by studying multiple cytokines across multiple brain regions of both male and female subjects, using 2 strains of mice bred in-house as well as rats obtained from the NIA aging colony. Here, we find that in the ventral hippocampus, both pro- and anti-inflammatory cytokines increased with the age in males and even more so in females. Interestingly, while IL-6 was dramatically upregulated with age in all brain regions and subjects examined, IL1-β was only elevated in the ventral and dorsal hippocampus.
Discussion
Here we showed that age-related increases in cytokine expression are pervasive in the ventral hippocampus of C57BL/6 mice, with 20 out of the 23 cytokines showing significantly greater expression in aged vs. young adult mice. The 20 cytokines upregulated with age included the pro-inflammatory cytokines IL-1α, IL-1β, IL-2, IL-3, IL-6 (see further discussion below in paragraph 4), IL-9, IL-12p40, IL-12p70, IL-17, eotaxin, interfeuron δ, KC, MIP-1a, MIP-1b, rantes, and TNFα as well as the anti-inflammatory cytokines IL-4, IL-10, IL-13, and G-CSF (Fig.
1). Cytokine expression levels also became much more strongly correlated in old mice (80 significant correlations) vs young mice (9 significant correlations (Fig.
1X, Tables S
2 and S
3), with a unique module/network of 11 highly intercorrelated cytokines emerging in old mice (IL-1α, IL-1β, IL-3, IL-6, KL12p40, IL-13, IL17, KC, MCP-1, MIP-1b, and Rantes). It has been suggested that the identification of such disease-associated modules may be useful biomarkers for diagnosis or predicting patient outcomes and/or treatment responses (e.g., [
50,
54,
55]). Further, IL-10 and IL-1β exhibited a shift from glycosylated to unglycosylated isoforms in the ventral hippocampus, suggesting even higher specific activity [
56]. Interestingly, age-related increases in cytokine expression outside of the hippocampus appear to be more discreet, with the prefrontal cortex, striatum, and cerebellum demonstrating an age-related increase in IL-6 expression but not IL-1β. Importantly, we found this widespread age-related increase in IL-6 is conserved across species, occurring in C57BL/6J mice, BALB/cJ mice, and Brown Norway rats. In select brain regions, the age-related increases in IL-6 were more pronounced in females relative to males, and biochemical fractionation suggests the age-related increases in IL-6 disproportionately target pro- vs. anti-inflammatory signaling cascades. Together, our data are consistent with pathway analyses that suggest age-related changes in inflammation- and immune-related gene expression are more pronounced in the hippocampus compared to other brain regions and more pronounced in females versus males [
7,
8,
57].
There have been many inconsistencies in the literature with regard to reports of age-related changes—or lack thereof—in cytokine protein expression. Our results are consistent with several past studies showing age-related increases in basal IL-1β protein in the hippocampus of a variety of naturally aging mice and rats [
23‐
28], but not in brain regions outside of the hippocampus [
22,
23,
30‐
32,
37]. That said, our results differ from several studies specifically using the F344xBN rats from the NIA colony that found no effect of age on basal IL-1β protein expression; although they did report age-related exacerbation of IL-1β protein induction caused by a high-fat diet,
Escherichia coli infection or surgery [
22,
30‐
32]. As noted above, findings around IL-6 are particularly diverse, with many studies reporting age-related increases in basal IL-6 expression in the brains of naturally aging rodents [
21,
33,
40‐
42] and several others finding no age-related changes [
22,
23,
37]. Inconsistencies in the literature around IL-6 cannot be so easily explained by differences in the strains used or the specific brain region examined. For example, one study employing C57BL/6J mice found age-related increases in IL-6 protein expression in the cortex as did we [
34]; however, the effect in another study failed to reach statistical significance [
23]. Such inconsistencies in the literature may be related to differential sensitivities of antibodies. Indeed, we found Femto chemiluminescence substrate was needed to reliably detect cytokine expression in our whole-tissue homogenates. Differing environmental factors across labs may also contribute to reported inconsistencies. For example, diet, and other environmental factors have been shown to preferentially upregulate cytokine expression in aged vs. young adult brains [
11,
12,
35‐
37]. Differences in ambient temperatures from facility to facility may even be to blame since higher temperatures can increase the expression of cytokines [
58] or increase the transport of cytokines from the periphery into the brain [
59]. The age-related increases in IL-6 we have measured are the most robust biochemical finding ever observed in our lab. They have been detected using 2 different IL-6 antibodies, using tissue from multiple brain regions from multiple cohorts of mouse strains raised at different times in our facility and even tissue from rats raised in a different animal facility. That said, we cannot rule out the possibility that our ability to detect such reliable age-related increases in IL-6 expression may be directly related to environmental factors present in the animal facility at the University of South Carolina at the time of tissue harvesting, since the rats obtained from the NIA colony did have to habituate for 1 week prior to experimentation.
It will be of interest to future studies to determine the cell type and mechanism driving the region-specific changes in cytokine expression described herein. Microglia release a myriad of cytokines [
60] but cytokines can also be released by brain endothelial cells, astrocytes, and neurons as well [
42,
60,
61]. Neurons can further regulate cytokine levels by releasing signals that either activate (i.e., “on signals”) or deactivate/inhibit microglia (i.e., “off signals”) [
62]. Our findings suggest that age-related increases in the hippocampus may be more extensive than those in the prefrontal cortex, striatum, or cerebellum. This increased sensitivity of the hippocampus is consistent with the fact that the hippocampus is one of the brain regions most populated with microglia [
13].
Upon biochemical fractionation of the hippocampus, we found that IL-6 expression was increased with age in all fractions but that the magnitude of the increase was far greater in the cytosolic versus nuclear or membrane fraction. We cannot completely exclude the possibility that expression observed in the nuclear fraction reflects contamination by unsheared cells; however, the fact that we do not find substantial expression of the cytosolic marker pAKT or the membrane marker synaptophysin in the nuclear fractions suggests such contamination of our nuclear fraction is minimal. Even if the IL-6 expression found in the nuclear fraction actually reflects contamination from the membrane of unsheared cells, it would not change the conclusion drawn from the experiment—that is, that the magnitude of age-related increases is larger in the soluble fraction. This pattern has important functional implications. IL-6 can signal via membrane-bound IL-6 receptors in the “classic” pathway to elicit anti-inflammatory responses, or IL-6 can signal via soluble IL-6 receptors in the “trans-signaling” pathway to elicit pro-inflammatory responses [
51,
52]. The fact that age-related increases in hippocampal IL-6 are far greater in the cytosolic versus membrane fraction points to a proinflammatory response. Indeed, increased activation of IL-6 trans-signaling in the brain has been implicated in several inflammatory age-related diseases of the nervous system, including Alzheimer’s and Parkinson’s Disease (c.f., [
52]). The fact that we observe age-related increases in the nuclear fraction suggests IL-6 may also participate in a non-canonical cytokine signaling pathway whereby cytokine-bound receptors are internalized to the cytosol for transport to the nucleus (e.g., [
63]). It will be of interest to future studies to understand if targeting soluble vs membrane IL-6 receptors may prove therapeutic for the aging brain [
64,
65].
Age-related increases in pro-inflammatory cytokines are thought to be detrimental since higher expression of pro-inflammatory cytokines in humans and rodents correlates with deficits in cognitive function, synaptic plasticity, neurogenesis, and neurotrophic factor expression [
9,
14,
25,
27,
30,
32,
66‐
73]. In contrast, an age-related increase in anti-inflammatory cytokines is thought to be protective, while their loss impairs plasticity and cognition [
26,
74‐
77]. Here, we found that pro- and anti-inflammatory cytokines were upregulated largely in parallel with each other (Table
2). Further, the module of highly intercorrelated cytokines that emerged with age (i.e., IL-1α, IL-1β, IL-3, IL-6, KL12p40, IL-13, IL17, KC, MCP-1, MIP-1b, and Rantes) included both pro- and anti-inflammatory pathway members. Together, this suggests an attempt of anti-inflammatory pathways to compensate for the dysregulation of pro-inflammatory pathways. It is likely that the widespread age-related increase in IL-6 noted herein is detrimental since the majority of studies suggest elevated expression impairs cognitive function ([
66,
67,
78,
79], but see [
42]). As such, therapeutics that restore cytokine signaling may prove beneficial in the treatment of age-related disorders.
It is noteworthy that robust and consistent sex differences were observed in the present study [
80]. Whereas the female-aged brain was biased towards exaggerated inflammation relative to the male brain, no differences between males and females were evident in young or middle-aged mice. Although the majority of these studies were not powered to allow for statistical sex-based conclusions, the exacerbated neuroinflammation observed within the aging female brain was reliable and robust. For example, 14/17 old females exhibited VHIPP IL-6 levels that exceeded the mean value measured in old males. These data are in line with other studies indicating that neuroinflammation accumulates in the aged female hippocampus to a greater extent than males [
57] and highlights a potential mechanism whereby females display faster age-related cognitive decline than men [
81] and higher rates of Alzheimer’s disease [
82]. It will be of interest to future studies to understand the basis for these augmented female responses. Both female and male hormones fluctuate over time and change with age in rodents, with age-related decreases in pulsatile GnRH observed in male rodents and a multiplicity of ovarian states observed in aging female rodents (i.e., estropause in a persistent estrus phase, estropause in a persistent diestrus phase, or irregular cycling) [
83‐
87]. Such changes in ovarian aging status have been shown to influence inflammatory and metabolic gene expression in the rat hippocampus, albeit not differentially in the ventral versus dorsal hippocampus [
88]. Further, a number of peripheral inflammatory diseases increase with time—some of which occur differentially in males versus females and some of which elicit different cytokine profiles in males versus females (e.g., [
89,
90]). The fact that we did not track hormonal status or peripheral pathology may be considered a weakness of the present study. That said, it is not intuitive how global changes in blood cytokine levels triggered by hormones or tumors would elicit such brain region-specific effects (i.e., ventral but not dorsal hippocampus) or cytokine-specific sex effects (i.e., IL-6 but not IL-1 or IL-10). Whatever their basis, efforts to therapeutically target cytokine signaling should give serious consideration to these sex differences, particularly given that anti-inflammatory therapeutic responses [
91] and the half-life of cytokine antagonists [
92] differ in males versus females.
A number of therapeutic avenues are currently being pursued to target age-related increases in neuroinflammation. Natural products containing resveratrol have been shown to attenuate serum IL-6 and TNFα levels in healthy older adults along with memory retention and hippocampal functional connectivity [
93,
94]. Other antioxidants have reduced TNFα and IL-1β mRNA and protein expression in the brains of senescence-accelerated mouse models [
95]. Reported effects of melatonin are mixed with 1 study reporting reduced TNFα, IL-1β, and IL-6 protein expression in the hippocampus [
21] and the other reporting no effect on age-related increases in TNFα and worsening of age-related increases in brain IL-1α [
96]. Probiotics not only lowered TNFα and MCP1 protein expression in the serum while increasing IL-10 protein expression, but also improved cognition in a senescence-accelerated mouse model (males and females, [
97]). Biological approaches have also been taken to overcome the damaging effects of pro-inflammatory cytokines, including infusion of receptor antagonists or anti-inflammatory cytokines [
26,
28,
32,
69,
74,
76,
77,
98]. Of particular note—given our biochemical fractionation data—biologics that inhibit IL-6 trans-signaling specifically are being developed in the context of a number of inflammatory diseases (c.f., [
52]). Finally, behavioral therapeutic approaches, such as mindfulness training or exercise, have also shown promise in attenuating age-related increases in IL-1β and IL-6, at least in males [
99‐
102] and cognitive behavioral therapy has been shown to boost immune function across sexes by reducing proinflammatory molecules and improving immune cell counts [
103]. Importantly, peripheral markers of inflammation and immune activation may prove viable patient-selection biomarkers for such clinical trials given that several studies demonstrate parallel changes in the brain and blood or saliva [
7,
8,
33] as well as correlations between elevated cytokine expression in the serum and reduced cognitive function [
104].
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.