CSF1R antagonism depleted microglia in adult and aged BALB/c mice
We and others report that microglia of the aged brain have a primed and immune-reactive profile [5, 6, 13, 24, 26, 27, 29, 32, 47, 56, 61, 63, 64, 68, 73]. Thus, the objective of this study was to determine if CSF1R antagonist-mediated depletion of microglia in the aged brain would result in repopulation with new and unprimed microglia. First, adult and aged BALB/c mice were administered vehicle or PLX5622 chow for 21 d and the number of microglia (CD11b+/CD45low) in the brain was assessed (Fig. 1a). Representative bivariate dot plots of CD45 and CD11b labeling of Percoll-enriched microglia are shown (Fig. 1b). As expected, there was a significant main effect of PLX5622 treatment on the number of microglia in the brain (F(1, 16) = 37.11, P < 0.0001; Fig. 1c). Moreover, post hoc analysis confirmed PLX5622 administration resulted in an 80–85% reduction in microglia isolated from the brain of adult and aged mice compared to controls (P < 0.01, for each). Additionally, RNA was collected from the hippocampus, a region exhibiting profound neuroinflammatory changes in aged rodents [3, 9, 49], and the expression of several microglia-related genes was determined by nanoString nCounter analysis. Principle component analysis (PCA) of hippocampal gene expression shows unsupervised clustering of samples based on age and PLX5622 treatment (Fig. 1d). As expected, expression of key microglial signature genes (e.g., Cd45, Cd68, Cd86, Csf1r, Cx3cr1, P2ry12, Trem2) was significantly reduced in mice administered PLX5622 (Fig. 1e). Moreover, this reduction in microglial gene expression with PLX5622 was independent of age. Thus, oral administration of PLX5622 robustly reduced the number of microglia in the brain of both adult and aged BALB/c mice.
Microglia repopulated independent of age after CSF1R antagonist-mediated depletion
Adult and aged BALB/c mice were administered vehicle or PLX5622 chow for 21 d to deplete microglia. After this, all mice were administered vehicle chow and the level of microglial repopulation was determined 0, 1, 3, 5, 7, or 21 days later (Fig. 2a). Quantification of Iba1+ microglia in the cortex showed a main effect of time (F(6, 49) = 42.32, P < 0.0001) on microglial repopulation (Fig. 2b, c). For example, after 0 days of repopulation, there was a significant reduction in Iba1+ microglia in PLX5622-treated adult and aged mice compared to controls (P < 0.0001 for both). In addition, there was a reduced number of microglia after 5 days of repopulation in both adult and aged mice compared to controls. By 7 days, however, microglia returned to control levels in both age groups. Both microglial depopulation and repopulation were independent of age. Overall, the microglia of both adult and aged BALB/c mice were capable of full repopulation after PLX5622-mediated depletion.
Microglial depletion and repopulation reversed CD68+ lysosome size and lipofuscin accumulation in the microglia of aged mice
Hallmarks of aging include increased CD68 expression and increased accumulation of lipid debris (i.e., lipofuscin) in microglia [60]. CD68 is a lysosomal protein associated with increased macrophage phagocytosis and lipid accumulation. In addition, lipofuscin accumulates in cells over time and is associated with increased cellular auto-fluorescence with age [10, 54]. Thus, we next determined the levels of CD68 and lipofuscin in the microglia of adult and aged mice after microglial depletion and repopulation as described above. A repopulation period of 21 d was selected to avoid confounds of transient microglial activation and cellular debris after PLX5622. The cortex was used for image analysis to minimize confounds of neuronal lipofuscin accumulation. CD68 expression in microglia was relatively abundant in the brain, and there was a significant main effect of age (F(1, 32) = 7.727, P < 0.01) and repopulation (F(1, 32) = 8.448, P < 0.01) on microglial lysosome size (Fig. 3a, b). Post hoc analysis revealed CD68+ lysosome size was increased in cortical microglia of aged mice compared to adults (P < 0.01). Moreover, microglial repopulation attenuated this age-associated lysosome enlargement (P < 0.01). These data indicate microglial depletion and repopulation normalizes the increased CD68 expression in aged microglia to adult levels.
Next, auto-fluorescence was assessed in microglia after depletion and repopulation. Representative histograms of auto-fluorescence detected in microglia (CD11b+/CD45low) enriched from the brains of adult and aged mice with or without forced turnover show different distributions of auto-fluorescence between groups (Fig. 3c). As expected, there was a significant main effect of age (F(1, 49) = 58.79, P < 0.0001) and repopulation (F(1, 49) = 20.56, P < 0.0001) on microglial auto-fluorescence (Fig. 3c). Furthermore, there was a significant interaction between age and repopulation (F(1, 49) = 8.14, P < 0.01). Post hoc analysis confirmed microglia from aged mice had higher auto-fluorescence compared to adult controls (P < 0.0001). Moreover, this age-associated increase was attenuated in aged mice subjected to microglial depletion and repopulation (P < 0.0001).
In a related study, lipofuscin volume was determined in cortical microglia after depletion and repopulation. Notably, there was robust lipofuscin auto-fluorescence in the cortex of aged mice, but minimal auto-fluorescence in the cortex of adult mice (Fig. 3e, f). There was a significant main effect of age (F(1, 32) = 47.41, P < 0.0001) and microglial repopulation (F(1, 32) = 8.35, P < 0.01) on microglial lipofuscin volume. Furthermore, there was a significant interaction between age and repopulation (F(1, 32) = 7.403, P < 0.05). Post hoc analysis confirmed microglial lipofuscin content was higher in aged mice compared to adult controls (P < 0.0001), and this age-associated lipofuscin accumulation in microglia was reduced by microglial depletion and repopulation (P < 0.01). There was also an age-associated increase in non-microglial lipofuscin (Fig. 3e). NeuN immunolabeling showed neuronal accumulation of lipofuscin with age (F(1, 32) = 41.48, P < 0.0001), but this was unaffected by microglial depletion and repopulation (Fig. 3g). Taken together, microglial depletion and repopulation reduced the level of lipofuscin in aged microglia, but not in neurons.
Depletion and repopulation of microglia partially reversed the microglial aging transcriptional signature
Next, we sought to determine the mRNA signature of microglia in adult and aged mice after depletion and repopulation. Therefore, adult and aged mice were administered vehicle or PLX5622 chow for 21 d to deplete microglia. After 21 d, all mice were administered vehicle chow for an additional 21 d to allow for microglial repopulation. CD11b+/CD45low microglia were then Percoll-enriched, purified using fluorescence-activated cell sorting (FACS), and RNA was sequenced (Fig. 4a). PCA on the 500 most variable genes between the experimental groups shows clustering of samples by age, independent of microglial repopulation (Fig. 4b).
Differential expression analysis between Aged Control and Adult Control groups shows 511 genes significantly regulated by age (Padj < 0.05, absolute fold change > 1.5). Of these, expression of 455 genes was increased (Fig. 4c) while 56 were decreased (Fig. 4d). Following microglial repopulation, age-associated changes in 127 genes were reversed (117 increased, 10 decreased), such that expression in Aged Repopulation was no longer different from Adult Control. For instance, repopulation attenuated age-associated increases in A2m, Apoe, Bmp6, Olr1, Sorl1, and Tgfb2 expression. Microglial repopulation also reversed age-associated decreases in expression of Cdkn1a, Dennd2c, and Socs3. Next, gene expression between Aged Repopulation and Aged Control groups were compared to determine genes that were either partially reversed or exacerbated by repopulation (P < 0.05). Partially reversed genes showed expression closer to Adult Control, but were not fully restored. In contrast, exacerbated genes were further increased or decreased by repopulation. Age-associated differential expression of 58 genes (53 increased, 5 decreased), was partially reversed by microglial repopulation (Fig. 4c, d). The aged associated increase in Lyz2 and Tgfbr3, and decrease in Ccr1 and Tlr5, were partially reversed by microglial depletion and repopulation. Notably, expression of 19 genes (14 increased, 5 decreased) were exacerbated by microglial repopulation. For instance, age-associated increases in expression of Axl, Oas2, and Tnfsf8 were further increased by repopulation (Fig. 4c). Collectively, of the 511 genes differentially regulated by age, 127 were reversed by microglial depletion and repopulation, 53 were partially reversed, and 307 genes (271 increased and 36 decreased) remained unaffected. Genes unaffected by microglial depletion and repopulation included several inflammation-related genes (e.g., C3, Clec7a, Ifi44l, Il1b, Il1rn, Mrc1, Tlr8), indicating the age-associated inflammatory profile of microglia was unaffected by forced turnover. Overall, microglia exhibited a robust age-associated mRNA signature that was only partially influenced by forced turnover of microglia.
Differentially regulated pathways and gene ontologies in microglia were not affected by microglial depletion and repopulation
The genes differentially expressed by microglia in the Aged Control and Aged Repopulation groups compared to Adult Control (Fig. 4) were analyzed using IPA and PANTHER gene annotation. Canonical pathways, diseases and functions, and upstream regulators that were enriched in each differential expression comparison were compared (Fig. 5a). Overall, there was an age-associated increase in numerous inflammatory pathways, including NF-κb signaling, neuroinflammatory signaling, and production of NO and ROS by macrophages. Furthermore, microglial gene expression in aged mice was consistent with increased signaling of interferon (IFN)-γ, tumor necrosis factor (TNF), interleukin (IL)-1β, IFN-α, and IL-4. Overall, none of these inflammatory pathways were significantly reversed by microglial depletion and repopulation.
Next, PANTHER was used to determine the gene ontology (GO) designations for microglial genes significantly regulated by age with or without repopulation. Genes related to several biological processes (Fig. 5b) and molecular functions (Fig. 5c) were significantly altered by age. The most prevalent biological processes regulated by age in microglia were categorized as cellular process (GO:0009987), metabolic process (GO:0008152), response to stimulus (GO:0050896), biological adhesion (GO:0022610), and biological regulation (GO:0065007). Molecular functions of genes regulated by age included catalytic activity (GO:0003824), binding (GO:0005488), transporter activity (GO:0005215), and receptor activity (GO:0004872). The biological processes and molecular functions of genes regulated by age were unaffected by microglial depletion and repopulation. Taken together, these data indicate that microglial repopulation did not significantly alter the overall pathway-level or cellular systems-level effects of aging on microglial gene expression.
LPS-induced sickness behavior was prolonged in aged mice and unaffected by microglial depletion and repopulation
We and others have reported that primed microglia in models of aging, traumatic brain injury, and stress, exhibit an exaggerated immune-reactive profile after secondary immune challenge [23, 43, 72]. This amplified neuroinflammation corresponds with increased cognitive impairment and prolonged sickness behavior. Therefore, we sought to determine if the intermediate restoration of the microglial mRNA profile corresponded with an attenuated response to peripheral LPS challenge. Thus, adult and aged BALB/c mice were administered vehicle or PLX5622 chow for 21 d to deplete microglia. After 21 d, all mice were administered vehicle chow for an additional 21 d to allow for microglial repopulation, after which mice were injected with i.p. saline or LPS (Fig. 6a). The social exploratory behavior of each mouse was evaluated 4 and 24 h after saline or LPS injection. At both 4 and 24 h post-injection, there was no significant effect of age or repopulation on social exploratory behavior in saline-treated mice; therefore, all saline-treated mice were combined into a single group for subsequent analysis. At 4 h post-injection, there was a significant main effect of LPS (F(1, 46) = 82.16, P < 0.001) and age F(1, 20) = 7.56, P < 0.05) on social exploratory behavior (Fig. 6b). Post hoc analysis revealed that all LPS-injected mice had decreased social exploratory behavior compared to saline-treated mice (P < 0.0001 for all). At 24 h post-injection, there was a main effect of age (F(1, 71) = 37.26, P < 0.0001) and LPS (F(1, 71) = 8.39, P < 0.01) on social exploratory behavior. Moreover, there was a significant interaction between age and LPS (F(1, 71) = 9.53, P < 0.01). Post hoc analysis revealed that only aged LPS-injected mice displayed decreased social exploratory behavior at 24 h compared to saline-treated mice (P < 0.0001 for both Control and Repopulation). Within LPS-treated mice, there was a significant main effect of age (F(1, 33) = 47.19, P < 0.0001) on social exploration, but no effect of microglial repopulation. Moreover, aged mice spent approximately 70% less time interacting with the novel juvenile compared to adult mice (P < 0.001). It is important to note, however, that there was no effect of microglial repopulation on social exploration in either adult or aged mice at either time point. Taken together, aged mice had an exaggerated and prolonged sickness behavior that was not ameliorated by microglial turnover.
To characterize the neuroinflammatory transcriptional response to peripheral immune challenge, hippocampal mRNA was collected 24 h after LPS. RNA was analyzed using the Mouse Inflammation v2 nanoString gene array. PCA showed unsupervised clustering by LPS and age, but not by microglial repopulation (Fig. 6c). Next, LPS-induced changes in inflammatory gene expression were determined in adult and aged mice with or without microglial repopulation (Fig. 6d). Genes were classified as either exacerbated by age (increased by LPS in adults and further increased with age) or uniquely regulated by LPS in aged mice (increased by LPS in aged, but not adult, mice). LPS increased expression of 11 genes (e.g., C1qa, C1qb, C3ar1, Cfb, Il1b, Tyrobp) in the hippocampus of adult mice that were further exacerbated in aged mice. Of these, only expression of Ccl8 and Ifit2 were reversed by microglial repopulation. LPS increased expression of 32 genes (e.g., C3, Cd163, Mrc1, Myd88) in the hippocampus of aged mice that were not increased in adults. Three of these uniquely regulated genes were prevented by repopulation (Creb1, Daxx, Il12a). Taken together, aged mice had an exaggerated and more comprehensive response to innate immune challenge in the hippocampus compared to adult mice and this response was not reversed by microglial depletion and repopulation.
Age-associated changes in whole-brain transcription were unaffected by microglial repopulation
Collectively, we show that microglial repopulation partially reversed age-induced transcriptional changes and lipofuscin accumulation. This, however, was insufficient to reverse exaggerated behavioral and inflammatory responses to innate immune challenge with LPS. Thus, age-associated changes in the brain microenvironment may promote microglial priming independent of repopulation. To address this, we determined the mRNA signature of a coronal brain section, a representative sampling of the whole-brain transcriptome including the cortex, hippocampus, and hypothalamus, in adult and aged mice after microglial depletion and repopulation. Adult and aged mice were administered vehicle or PLX5622 chow for 21 d to deplete microglia. After 21 d, all mice were administered vehicle chow for an additional 21 d to allow for microglial repopulation, after which a coronal brain section was collected, and RNA was extracted and sequenced (Fig. 7a).
Differential expression analysis of whole-brain gene expression between Aged Control and Adult Control groups shows 409 genes significantly affected by age, with increased expression of 207 genes (Fig. 7b) and decreased expression of 202 genes (Fig. 7c). Moreover, these age-associated transcriptional changes persisted despite forced microglial turnover. Microglial repopulation reversed expression of 18 genes (5 increased, 13 decreased; e.g., Flnb, Ctcf, Gabra2, Gabrb1, Parp6, Snx). Only 3 genes were partially reversed, including Fnip2 (increased by age), Chl1, and Met (decreased by age). Age-associated decreases in Camk2n1, Camk4, and Camkk2 expression, and increase in H2-D1, were exacerbated by repopulation. Nonetheless, the majority of age-associated transcriptional changes in the brain were unaffected by microglial repopulation (382 total: 200 increased, 182 decreased). For example, age-associated increases in whole-brain mRNA expression of C1qa, Clec7a, Hif3a, Il33, and Smad9 persisted despite microglial repopulation (Fig. 7b). Similarly, age-associated decreases in expression of Grin3a, Negr1, Nrep, and Ntrk3 were unaffected by microglial repopulation (Fig. 7c). Notably, age-associated augmentation of astrocyte-related genes, including Gfap, S100b, and Vim, was unaffected by repopulation. Collectively, these data indicate that renewal of microglia following depletion and repopulation did not dramatically influence the whole-brain transcriptional responses to aging in mice.
Age-associated reactive astrogliosis was microglia-independent
Several reports indicate that astrocytes become more inflammatory with age [27, 48]. Therefore, we sought to determine the level of reactive astrogliosis in adult and aged mice after microglial depletion and repopulation. Adult and aged BALB/c mice were administered vehicle or PLX5622 chow for 21 d to deplete microglia. After 21 d, all mice were administered vehicle chow for an additional 21 days to allow for microglial repopulation. As expected, GFAP+ astrocyte density was increased in the aged hippocampus compared to adults (Fig. 8a, b). There was a significant main effect of age (F(1, 41) = 59.60, P < 0.0001) on GFAP+ astrocyte density, but not of microglial depletion or repopulation. These findings indicate that the age-associated increase in reactive astrogliosis was independent of microglia.
In a similar study, adult and aged mice were subjected to microglial depletion and repopulation as above. RNA was isolated from a coronal brain section and the expression of genes indicative of reactive astrogliosis was determined (Fig. 8c-e). As expected, there was a significant increase in Gfap (F(1, 7) = 287.5, P < 0.0001), S100b (F(1, 7) = 39.68, P < 0.001), and Vim (F(1, 7) = 44.65, P < 0.001) expression in aged mice compared to adults. Moreover, this age-associated increase in mRNA expression was independent of microglial depletion and repopulation. Taken together, these data show that reactive astrogliosis persisted in the aged brain after microglial repopulation.
Aged brain-conditioned media induces a hyper-inflammatory LPS response in neonatal microglia ex vivo
In order to assess the effect of the aged brain microenvironment on the inflammatory signature of microglia, culture media were conditioned with coronal brain sections from adult (8–10 weeks old) or aged (20 months old) BALB/c mice. Again, coronal brain sections were used to represent the global CNS environment. After 24 h, CM was collected and diluted with fresh media. Primary neonatal microglia were then incubated with adult or aged CM for 24 h and stimulated with LPS or vehicle. Microglial RNA was isolated after 4 h and expression of inflammatory cytokines determined (Fig. 9a). It is important to note incubation with CM did not affect microglial viability over this 24-h period (Fig. 9b). As expected, there was a significant main effect of LPS on expression of Il1b (F(1, 28) = 81.6, P < 0.0001), Il6 (F(1, 28) = 57.7, P < 0.0001), and Tnf (F(1, 28) = 176.5, P < 0.0001) mRNA expression (Fig. 9c-e). Moreover, microglia incubated with aged CM exhibited increased expression of Il1b, Il6, and Tnf mRNA in response to LPS compared to adult CM (P < 0.05 for all). Taken together, these data show soluble signals from the aged brain cause microglia to exhibit exaggerated inflammatory responses to LPS stimulation.