Aging triggers an upregulation of a multitude of cytokines in the male and especially the female rodent hippocampus but more discrete changes in other brain regions

C57BL/6J and BALB/cJ mice were originally obtained from Jax and then bred onsite at the University of South Carolina School of Medicine. Young mice ranged from 2 to 5 months old, middle-aged mice ranged from 9 to 12 months old, and old mice ranged from 18 to 24 months old. Litter effects are unlikely to account for effects described herein as mice were derived from multiple litters and were housed across multiple cages. Further, several sample sets analyzed by Western blot were obtained from multiple cohorts of mice born and raised at different times in our facility, with some cohorts including young, middle-aged and old mice, and other cohorts including only young and old mice. Young (4 months old) and old (19 months old) Brown Norway Rats were obtained directly from the NIA colony and allowed to acclimate to the animal facility at the University of South Carolina School of Medicine for 1 week prior to tissue harvest. Animals are housed on a 12:12 light:dark cycle and allowed ad lib access to food and water. Experiments were carried out in accordance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals (Pub 85-23, revised 1996) and were fully approved by the Institutional Animal Care and Use Committee of the University of South Carolina and the University of Maryland, Baltimore. See Table 1 for specific n’s in each experiment. Note that males and females were used throughout all experiments but not necessarily in sufficient number to power an analysis of age x sex. Mice are generally healthy at the time of tissue harvest. We do not conduct gross pathology; however, mice are routinely assessed by husbandry, veterinary, and laboratory staff and mice with palpable tumors, lethargy, altered gait, signs of malnutrition, or dehydration are removed from the study and euthanized. If mice were to demonstrate evidence of a striking anatomical abnormality of the brain upon dissection (e.g., a pituitary tumor), they would also not be included since it is our intent is to study the effects of healthy aging. Note that a previous study found ~35% of C57BL6 female mice raised in a pathogen-free facility exhibited pituitary tumors [44]; however, we came upon only a handful of these readily identifiable tumors (or tumors of any sort) in the brains of our C57BL/6J mice raised in a conventional animal facility.

Mice were euthanized during the light cycle via cervical dislocation, whilst rats were euthanized during the light cycle by CO₂ inhalation. Mouse brains were harvested fresh, dissected on wet ice, and stored at −80°C for further processing. Rat brains were harvested fresh, hemisected over wet ice, frozen in isopentane over dry ice, and half brains were then stored at −80°C. Later, the cerebellum and prefrontal cortex were dissected from frozen rat half brains on dry ice. All brain regions examined in the BALB/cJ mice were dissected from the same group of animals.

A Bio-Plex Pro Mouse Cytokine 23-Plex (Bio-Rad, Hercules CA) was used to probe ventral hippocampal homogenates for cytokines. Methods for the tissue preparation were based on previously published work in rat brain homogenates [45, 46]. Briefly, ventral hippocampal samples were homogenized in boiling lysis buffer (50-mM NaF/1% SDS [47];), and then, homogenates were diluted in sample diluent containing 1% FBS. Standards were reconstituted in lysis buffer containing 1% FBS and diluted via serial dilution such that all analytes reached a minimum concentration of 0.2 pg/mL. Bioplex data were collected over 3 runs using the same plate. In the first run, ½ of the wells were used to test all samples at a concentration of 0.275 μg/μL based on our previous work with rat tissue [45, 46]. Unfortunately, many targets were unable to be detected due to matrix effects or reached ceiling effects at that high of a concentration (See Table S1). In the 2nd run, we used approximately ¼ of the wells to retest ½ of these samples at a concentration of 0.1 μg/μL. We found 0.1 μg/μL allowed detection of all targets except IL-5. Therefore, in a 3rd run we used the remaining ¼ of the plate to retest the other ½ of the samples at a concentration of 0.05 μg/μL. 0.05 μg/μL enabled detection of all targets. The pattern of effects was consistent across the subsets of samples tested at 0.1 μg/μL and 0.05 μg/μL. As such, data from each concentration were expressed as a fold change of the young mean and combined in analyses. Bead preparation, handling, and plate processing were conducted according to manufacturer protocol. Plates were washed using a Bio-Plex Pro II Wash Station (Bio-Rad, Hercules, CA) and read using a Luminex SD system (Bio-Rad, Hercules CA) housed within the Instrument Resource Facility at the University of South Carolina School of Medicine.

Samples for select C57BL/6J Westerns, BALB/cJ Westerns, and rat Westerns were homogenized using a sonic dismembrator (a.k.a. tissue sonicator), as previously described [47, 48], in boiling lysis buffer (50 mM NaF/1% SDS). The tissue for the remaining C57BL/6J Westerns was homogenized as previously described [47] in ice-cold lysis buffer (20 mM Tris-HCl, pH 7.5; 2 mM MgCl_2; Thermo Pierce Scientific phosphatase tablet #A32959 and protease inhibitor 3 #P0044) in preparation for subsequent biochemical fractionation of the samples. There were no differences in the pattern of Western results obtained with one or the other lysis buffer, and so data were combined. Total protein quantity was determined for all homogenized tissue using the DC Protein Assay kit (BioRad, Inc.; Hercules, CA), and western blots were carried out as previously outlined [47, 48]. For all blots, 36.3 μg of total protein or 22 μg of fractionated protein was loaded onto 4–12% Bis-Tris gels (Life Technologies) for electrophoresis. Following transfer to nitrocellulose membranes (#10600008, Amersham), all blots except those testing the biochemical fractionations were subjected to PonceauS (#6266-79-5, Fisher Scientific) to stain the total protein in each sample. Following image capture, The PonceauS was washed off with ultrapure water and TBST and the membranes were cut according to the molecular weight of the target of interest (Figure S1). Superblock (#37515, Thermofisher) was then used at room temperature to block non-specific binding sites on membranes, and membranes were subsequently probed overnight at 4°C with primary antibodies against IL-6 (early experiments with 1:200 of MAB406 from R&D Systems; later experiments with 1:2500 of ARX0962 from Life Technologies), IL-1β (ab106034 from Abcam, 1:5,000), IL-10 (ARG2419 from Arigo Biolaboratories, 1:500), and actin (A2066 from Sigma, 1:10,000) in Pierce Superblock (P137517)/0.1% Tween20 (Fisher BP337-500). The following day, blots were washed in TBST and incubated for 1 h at room temperature with a species-specific HRP-tagged secondary antibody (Jackson, 1:10,000). Blots probed with IL-6 and IL-1β were developed using WesternSure Premium Chemiluminescent Substrate (926-95000), whereas blots probed with actin were developed using Pierce SuperSignal West Pico CL Substrate (#34078). Blots were then apposed to film, scanned in at 1200 dpi, and quantified by densitometry using ImageJ (NIH) by a group-blinded experimenter. Data from the biochemical fractions were normalized to actin as a loading control as these data were obtained several years prior to the lab implementing the use of PonceauS. All other Western data were normalized to PonceauS staining intensity as a loading control. Unadjusted images of the full blots and membranes from which cropped images were pulled are shown in Figures S1-S8.

Biochemical data were collected by an experimenter blind to treatment, and Sigmaplot 11.2 was used to analyze data. As described above, Bioplex data were obtained from 2 separate runs, each using a different concentration of tissue. Therefore, Bioplex data from each run were normalized to the young group from that run. Similarly, Western data for a given experiment span multiple gels and so are normalized to the Young group on each gel in order to mitigate any non-specific differences between blots related to transfer efficiencies, film exposures, etc. (as in [47, 48]). For Westerns, each brain region and fraction were run on separate sets of blots and so resultant data are analyzed with individual statistical tests per region or fraction. Both males and females were included in each group, but not always in sufficient number to analyze for an effect sex. For experiments with fewer than n=4/sex/group, data were analyzed for age only. For experiments that included at least n=4/sex/group, data were analyzed for both age and sex. Parametric statistics were used (i.e., 2-factor ANOVA or Student t test) when datasets passed normality (Shapiro-Wilk test) and equal variance (Levene’s test). In cases where 2-factor ANOVAs failed assumptions of normality and/or equal variance, statistical tests for each factor were conducted separately. When statistical tests failed normality and/or equal variance, a nonparametric Whitney rank sum test or Kruskal-Wallis ANOVA on ranks was used. Correlations were conducted using Spearman rank order. Given all Bioplex data were derived from the simultaneous measurement of multiple endpoints, a false-rate discovery (FDR) correction was applied to all P values to mitigate the risk of type I error associated with multiple comparisons. Outliers >2 standard deviations from the mean were removed from analyses consistent with our previous publications (e.g., [47, 49]; outliers/total data points: Fig. 2A, 3/72; Fig. 3F, 1/72; Fig. 3G, 3/92).

To determine if the ventral hippocampus exhibits a widespread upregulation of cytokines with age, we used a Bioplex to simultaneously measure 23 such endpoints in young versus old C57BL/6J mice. Relative to young mice, old mice expressed significantly higher levels of IL-1α, IL-1β, IL-2, IL-3, IL-4, IL-6, IL-9, IL-10, IL-12p40, IL-12p70, IL-13, IL-17, eotaxin, granulocyte colony-stimulating factor, interferon δ, KC, macrophage inflammatory protein 1a (MIP-1a), MIP-1b, rantes, and TNFα (Fig. 1; Table 1). The age-related increases in the above noted pro-inflammatory and anti-inflammatory cytokines occurred in equal proportion, with only the ratio of IL-1β/IL-13 differing significantly between old and young mice (Table 2). Effects noted in females tended to be stronger than those noted in males (Figure S9); however, the study was not powered for an analysis of sex effects (see Western data below). The age-related increases in IL-5 and monocyte chemoattractant protein 1 failed to reach the level of statistical significance and granulocyte-macrophage colony-stimulating factor showed no difference in expression between young and old mice. In addition to expression changes, aging was also associated with the emergence of novel correlations in expression that were not observed in young mice. Whereas young mice only showed 9 significant correlations (no more than 3 for a given cytokine), old mice showed a total of 80 significant correlations (Table S2; Figure 1X). Interestingly, these significant correlations in old mice were not randomly distributed. Rather, 65/80 significant correlations fell within a module (i.e., a cluster of cytokines whose signals strongly correlate with each other [50]) of 11 cytokines (IL-1α, IL-1β, IL-3, IL-6, KL12p40, IL-13, IL17, KC, MCP-1, MIP-1b, and Rantes; Table S3). The degree of correlation was not related to the effect size of the age-related increases, with cytokines showing no, small or large changes represented amongst both the highly and rarely correlated cytokines. The degree of correlation was also not related to the designation of anti- vs pro-inflammatory, with members of each family amongst the highly versus rarely correlated cytokines. Among the highly correlated cytokines, IL-1β, IL-3, IL-6, KL12p40, IL-13, IL17, KC, and MCP-1 formed a particularly tightly coupled core module, with each correlated to the other. Taken together, these data suggest the ventral hippocampus exhibits a dramatic change in the regulation of cytokines with age.

To confirm results obtained from the Bioplex and further investigate potential sex effects, IL-10, IL-6, and IL-1β expression in the ventral hippocampus were assessed using Western blots. In the ventral hippocampus of C57BL/6J mice, expression of presumed unglycosylated IL-10 (i.e., lower band on Western blot) appears to increase with age across males and females, whereas expression of presumed glycosylated IL-10 (i.e., upper band on Western blot) appears to remain stable (Fig. 2A; see Table 1 for statistics; post hoc: young vs. middle age P=0.035, young vs. old P=0.046). As such, the ratio of glycosylated/unglycosylated IL-10 decreased with age (Figure S10A). In contrast, both presumed unglycosylated and glycosylated forms of IL-1β increased in expression with age in both males and females (Fig. 2C; see Table 1 for statistics). Even though both isoforms of IL-1β increased with age, the ratio of glycosylated/unglycosylated IL-1 β still decreased with age as was seen with IL-10 (Figure S10B). IL-6 expression also increased across the lifespan of male and female C57BL/6J mice (Fig. 2B; see Table 1 for statistics; post hoc: young vs. middle-aged P=0.049, young vs. old P<0.001; middle-aged vs old P<0.001), but females showed a much larger age-related increase than males. Together, these data support a widespread upregulation of cytokines in the ventral hippocampus with age.

To determine the extent to which cytokines are upregulated with age in the brain, Western blots were used to analyze IL-6 and IL-1β in the dorsal hippocampus, prefrontal cortex (PFC), striatum, and cerebellum of C57BL/6J mice. IL-6 expression increased with age in the dorsal hippocampus (Fig. 3A, see Table 1 for statistics; post hoc: young vs middle P=0.005, young vs. old P<0.001, middle vs. old P<0.001), prefrontal cortex (Fig. 3C; Table 1), striatum (Fig. 3E; Table 1), and cerebellum of mice (Fig. 3G; Table 1; post hoc: middle age or old vs. young, P<0.001). In contrast, IL-1β expression only increased with age in the dorsal hippocampus (Fig. 3B; Table 1). Unlike the ventral hippocampus, there was no age-related decrease in glycosylated/unglycosylated IL-1β in the dorsal hippocampus (Figure S10C) or prefrontal cortex (Figure S10D), and females actually showed an increase in glycosylated/unglycosylated IL-1β in the striatum (Figure S10E). Also unlike the ventral hippocampus, males and females showed equivalent age-related increases in IL-6 expression in the dorsal hippocampus, prefrontal cortex, striatum, and cerebellum suggesting the heightened age-related increases in IL-6 observed above in females are restricted to the ventral hippocampus in this mouse strain. Age-related increases in IL-6 expression outside of the hippocampus were confirmed in the second cohort of C57BL/6J mice with middle-aged mice (n=5M/6F) showing higher expression relative to young mice (n=5/sex) in the prefrontal cortex (young, 0.228 ±0.01 A.U.; middle, 1.00 ±0.07 A.U.; rank sum test: T(10,11)=55.00, P<0.001) and striatum (young, 0.275 ±0.03 A.U.; middle, 1.00 ±0.07 A.U.; rank sum test: T(10,11)=45.00, P<0.001). These data together with the Bioplex data suggest that while the hippocampus exhibits a widespread upregulation of cytokines with age, other brain regions exhibit a more restricted upregulation.

To determine if the observed age-related increases in IL-6 are conserved across mouse strains and rats, we measured IL-6 expression in the brains of young versus old BALB/cJ mice and Brown Norway Rats. Relative to young BALB/cJ mice, old BALB/cJ mice express significantly higher levels of IL-6 protein in the ventral hippocampus, dorsal hippocampus, prefrontal cortex, striatum, and cerebellum (Fig. 4A–E). Visual inspection of the data suggests the age-related increases in IL-6 may be more prominent in the female vs male BALB/cJ mice in the ventral hippocampus, PFC, and striatum, but the study was not sufficiently powered for a formal analysis of sex as a factor. Old Brown Norway Rats also showed higher expression of IL-6 protein in the PFC and cerebellum relative to young rats (Fig. 5F–G), with females showing higher IL-6 expression than males in the cerebellum (Table 1). Together, these data suggest that widespread age-related increases in IL-6 expression are conserved across species, are generally more pronounced in females, and are not tied to a single rearing environment.

IL-6 can signal through multiple pathways, each of which has different functional consequences. Specifically, IL-6 signals via membrane-bound IL-6 receptors in the “classic” pathway to elicit anti-inflammatory responses and soluble IL-6 receptors in the “trans-signaling” pathway to elicit pro-inflammatory responses [51‐53]. As such, we conducted biochemical fractionation on the ventral and dorsal hippocampus of young versus old C57BL/6J mice. In both ventral and dorsal hippocampus, we found age-related IL-6 increases occur in all fractions but that the magnitude of the age-related increase in cytosolic IL-6 was twice that observed in the membrane or nuclear fractions (Fig. 5; effect of a fraction within old: F(2,14)=13.35, P<0.001; post hoc: cytosolic vs. nuclear and membrane, P<0.001). Parsimoniously, this points to a predominantly pro-inflammatory consequence of these age-related increases.