1. Immunology and Inflammation
  2. Neuroscience
Download icon

Differential accumulation of storage bodies with aging defines discrete subsets of microglia in the healthy brain

  1. Jeremy Carlos Burns
  2. Bunny Cotleur
  3. Dirk M Walther
  4. Bekim Bajrami
  5. Stephen J Rubino
  6. Ru Wei
  7. Nathalie Franchimont
  8. Susan L Cotman
  9. Richard M Ransohoff
  10. Michael Mingueneau  Is a corresponding author
  1. Multiple Sclerosis & Neurorepair Research Unit, Biogen, United States
  2. Department of Pharmacology & Experimental Therapeutics, Boston University School of Medicine, United States
  3. Emerging Neurosciences Research Unit, Biogen, United States
  4. Chemical Biology and Proteomics, United States
  5. Multiple Sclerosis Development Unit, Biogen, United States
  6. Center for Genomic Medicine, Department of Neurology, Massachusetts General Hospital, Harvard Medical School, United States
  7. Third Rock Ventures, United States
Tools and Resources
  • Cited 0
  • Views 2,353
  • Annotations
Cite this article as: eLife 2020;9:e57495 doi: 10.7554/eLife.57495

Abstract

To date, microglia subsets in the healthy CNS have not been identified. Utilizing autofluorescence (AF) as a discriminating parameter, we identified two novel microglia subsets in both mice and non-human primates, termed autofluorescence-positive (AF+) and negative (AF). While their proportion remained constant throughout most adult life, the AF signal linearly and specifically increased in AF+ microglia with age and correlated with a commensurate increase in size and complexity of lysosomal storage bodies, as detected by transmission electron microscopy and LAMP1 levels. Post-depletion repopulation kinetics revealed AF cells as likely precursors of AF+ microglia. At the molecular level, the proteome of AF+ microglia showed overrepresentation of endolysosomal, autophagic, catabolic, and mTOR-related proteins. Mimicking the effect of advanced aging, genetic disruption of lysosomal function accelerated the accumulation of storage bodies in AF+ cells and led to impaired microglia physiology and cell death, suggestive of a mechanistic convergence between aging and lysosomal storage disorders.

eLife digest

Microglia are a unique type of immune cell found in the brain and spinal cord. Their job is to support neurons, defend against invading microbes, clear debris and remove dying neurons by engulfing them. Despite these diverse roles, scientists have long believed that there is only a single type of microglial cell, which adapts to perform whatever task is required. But more recent evidence suggests that this is not the whole story.

Burns et al. now show that we can distinguish two subtypes of microglia based on a property called autofluorescence. This is the tendency of cells and tissues to emit light of one color after they have absorbed light of another. Burns et al. show that about 70% of microglia in healthy mouse and monkey brains display autofluorescence. However, about 30% of microglia show no autofluorescence at all. This suggests that there are two subtypes of microglia: autofluorescence-positive and autofluorescence-negative.

But does this difference have any implications for how the microglia behave? Autofluorescence occurs because specific substances inside the cells absorb light. In the case of microglia, electron microscopy revealed that autofluorescence was caused by structures within the cell called lysosomal storage bodies accumulating certain materials. The stored material included fat molecules, cholesterol crystals and other substances that are typical of disorders that affect these compartments. Burns et al. show that autofluorescent microglia contain larger amounts of proteins involved in storing and digesting waste materials than their non-autofluorescent counterparts. Moreover, as the brain ages, lysosomal storage material builds up inside autofluorescent microglia, which increase their autofluorescence as a result. Unfortunately, this accumulation of cellular debris also makes it harder for the microglia to perform their tasks.

Increasing evidence suggests that the accumulation of waste materials inside the brain contributes to diseases of aging. Future work should examine how autofluorescent microglia behave in animal models of neurodegenerative diseases. If these cells do help protect the brain from the effects of aging, targeting them could be a new strategy for treating aging-related diseases.

Introduction

Microglia are a unique population of tissue resident macrophages residing in the central nervous system (CNS) accounting for 10% to 15% of all cells within the CNS. While displaying some canonical macrophage activities such as the phagocytosis of debris and apoptotic bodies (Chan et al., 2001; Janda et al., 2018), microglia are also endowed with functions specific to the CNS microenvironment (Clayton et al., 2017; Li and Barres, 2018; Ransohoff, 2016; Ransohoff and Khoury, 2016), such as synaptic remodeling (Paolicelli et al., 2011; Stephan et al., 2012; Stevens et al., 2007; Weinhard et al., 2018), neuronal support (Parkhurst et al., 2013; Ueno et al., 2013), and oligodendrogenesis (Hagemeyer et al., 2017; Wlodarczyk et al., 2017).

However, despite this diversity of functions, no durable subsets have been identified in the healthy adult brain at steady-state. The disease-associated microglia (DAM) and the closely-related microglia subset expressing a neurodegenerative phenotype (MGnD) were reported to arise in response to the accumulation of β-amyloid plaques in Alzheimer’s Disease (AD) transgenic mouse models (Kamphuis et al., 2016; Keren-Shaul et al., 2017; Krasemann et al., 2017; Mrdjen et al., 2018), as well as in other models of neurodegeneration and aging (Chiu et al., 2013; Holtman et al., 2015; Spiller et al., 2018; Wlodarczyk et al., 2014). Mirroring the gene signature and function of DAM and MGnD subsets in disease conditions, the proliferative-region-associated microglia (PAM) subset was associated with the phagocytosis of newly formed and dying oligodendrocytes during normal post-natal development (Felsky et al., 2019; Hagemeyer et al., 2017; Li et al., 2019). Dark microglia, a subset identified by their condensed, electron-dense cytoplasm visible by transmission electron microscopy, are frequently observed during pathologic states and believed to actively associate with neuronal synapses (Bisht et al., 2016). Beyond conditions of neuronal or oligodendrocyte cell death, microglia heterogeneity was also observed within the adult steady-state CNS. In particular, regional variations in cellular density (Lawson et al., 1990), but also in microglia gene expression profiles were reported, pointing to region-specific microglia functions and possibly subsets in healthy brain (Ayata et al., 2018; Grabert et al., 2016).

While these emerging transcriptomic findings suggest that steady-state microglia subsets are likely present, they remain to be identified and characterized. By using cellular autofluorescence (AF) as a novel photophysical parameter to explore microglia heterogeneity in unperturbed conditions, we report that steady-state microglia exist in two discrete states at a regulated ratio throughout the entire lifespan of rodent and non-human primate species. Leveraging this physical property, we devised a novel probe-free method to isolate and characterize these subsets and established that AF+ and AF microglia differed in their ultrastructural features, homeostatic dynamics, proteomic content and physiological properties.

Results

Cellular autofluorescence identifies two discrete microglia subsets

Based on the observation that certain peripheral myeloid populations are highly autofluorescent, we initially included an empty fluorescence channel in our microglia flow cytometry analyses. Doing so revealed an unexpected bimodal distribution of autofluorescence (AF) intensity in CD45dimCD11B+ microglia isolated from 6-month-old naïve C57BL/6 mice (Figure 1A and Figure 1—figure supplement 1A). This bimodal distribution identified two subsets of microglia: an AF-positive (AF+) subset showing a strong AF signal, and an AF-negative (AF) subset which displayed no or minimal levels of AF. The two subsets appeared at a highly reproducible ratio of 1:2.5 (AF:AF+), with an average frequency of AF+ microglia of 71 ± 1.2% (n = 13; CV = 1.6%) (Figure 1—figure supplement 1B). To further characterize the spectral properties of the AF signal, we included additional empty fluorescence channels. The AF signal showed maximal intensity in the 660–735 nm emission range upon excitation with a 488 nm (Blue) laser but was also detected across multiple combinations of laser excitation wavelengths and emission filter ranges (Figure 1B). Extraction of single cell-level data from flow cytometry analyses revealed that the majority of microglia that were either positive or negative for AF in the Blue-710 nm AF channel were also positive or negative across most other AF channels tested, respectively (Figure 1C). In a few channels such as Red-780 nm for instance, 10% to 20% of the cells that were positive in Blue-710 nm appeared low or negative, which likely resulted from a differential sensitivity of the channels used to detect AF as shown in Figure 1B by the differential brightness of the AF+ population in those two channels. Both AF microglia subsets were detected across different regions of the brain and while there were minor differences in the frequency of AF+ microglia between regions, AF+ cells displayed similar levels of AF signal in cerebellum, cortex and hippocampus (Figure 1—figure supplement 1C,D). Both AF+ and AF subsets of microglia were positive for microglia homeostatic markers, including CX3CR1, P2RY12 and TMEM119 (Figure 1D and Figure 1—figure supplement 1E). Finally, there were no gender differences observed in the frequency of AF+ microglia nor the intensity of AF (Figure 1—figure supplement 1F,G). This observation was conserved across species as microglia isolated from the brains of 3- to 4-year-old Cynomolgus monkeys showed a very consistent bimodal pattern of AF in the Blue-710 nm channel with 80 ± 3.6% of microglia that were AF+ on average (Figure 1E,F). Altogether, these analyses identified cellular AF as a novel photophysical property for discriminating two discrete subsets of microglia present at steady-state and conserved between rodents and non-human primates.

Figure 1 with 1 supplement see all
Cellular autofluorescence identifies two discrete and novel microglia subsets.

(A) Representative flow cytometry scatterplot showing CD45 and CD11B surface levels in a brain single cell suspension and histogram of autofluorescence (AF) signal in DAPI CD45dim CD11B+ microglia detected with a blue laser and 710 nm detector. (B) Representative flow cytometry histograms of AF intensity in the entire microglia population for multiple excitation lasers and emission filters. (C) Heatmap of AF signal detected across multiple cytometer channels (columns) in 500 single microglial cells (rows) and hierarchically clustered by Manhattan distance. (D) Representative flow cytometry histograms of TMEM119, CX3CR1 and P2RY12 surface levels in AF+ and AF microglia subsets. (E) Representative flow cytometry scatterplots used to identify microglia in brain single cell suspensions from Cynomolgus macaques and representative histogram of AF levels detected in microglia. (F) Quantitation of results presented in (E). (G) Representative scatterplots of AF levels detected in CD45dimCD11B+ microglia and corresponding microscopy images of selected AF+ and AF cells (highlighted in orange and numbered in the scatterplots and images) analyzed using imaging flow cytometry. Scale bar = 7 μm. (A, B) Representative of n = 12 animals from at least four independent experiments. (D) n = 6 animals from at least two independent experiments. (E, F) n = 5 animals from two independent experiments. B = blue laser, R = red laser, YG = yellow green; AF = autofluorescence; BF = bright field; SSC = side scatter. See also Figure 1—figure supplement 1.

The autofluorescence signal in AF+ microglia originates from intracellular organelles

To define the subcellular origin of the AF signal in microglia, we utilized imaging cytometry. Microscopy images of AF+ cells identified by the fluorescence intensity detected in two empty channels (Figure 1G) revealed that the AF signal was not diffusely distributed throughout the cellular volume but was punctate and localized within intracellular organelles. The subcellular AF compartments were observed in all tested AF channels and systematically colocalized (Figure 1—figure supplement 1H and I). AF cells did not display detectable AF subcellular compartments in any of the tested channels, altogether establishing that these two subsets of microglia, identified solely by their AF profiles, differed by the presence of highly autofluorescent intracellular organelles that were restricted to the AF+ microglia subset.

AF signal intensity increases linearly with aging, but solely within AF+ microglia

In contrast to the punctate AF signal detected in 3-month-old animals, imaging flow cytometry applied to naïve mice from a range of ages revealed that AF subcellular structures became largely confluent in 10-month-old animals and occupied a larger fraction of the cytosol (from 11 μm2 to 17 μm2 on average in 3- and 10- month-old animals, respectively) (Figure 2A,B). Consistent with these results, flow cytometry analyses established a nearly linear increase of AF signal intensity in AF+ microglia with aging (Figure 2C,D), resulting in a cumulative 3-fold increase of AF signal between 3- and 12-month-old mice across multiple fluorescence channels (Figure 2D and Figure 2—figure supplement 1A). While a clear AF+ population was not detectable in mice aged 15 days post-natal or younger, a bimodal distribution appeared as early as 30 days post-natal and AF increased linearly with age from that point on (Figure 2—figure supplement 1B,C). In contrast, age-dependent increases in cellular AF were not detected in peripheral CD11B+ cells residing in the spleen (Figure 2—figure supplement 1D). Despite the large increase in the AF signal observed in AF+ cells, their frequency remained largely unchanged during aging, decreasing only slightly from 75% to 72% from 3 to 12 months of age (Figure 2E). Finally, neither the proportion of AF cells (Figure 2E) nor the area of intracellular AF signal detected in AF microglia (Figure 2A,B) were altered by aging, revealing a selective impact of aging on the AF+ microglia subset.

Figure 2 with 1 supplement see all
Natural aging increases microglia AF signal intensity solely within AF+ cells.

(A) Representative imaging flow cytometry images and (B) corresponding quantitation of average AF area per cell in AF+ and AF CD45dimCD11B+ microglia isolated from 3- and 10-month old mice. Significance established with 2-way ANOVA followed by Sidak’s post-hoc correction, n = 4000–8000 cells. (C) Representative flow cytometry histogram and (D) corresponding quantitation of AF intensity in AF+ CD45dim CD11B+ microglia isolated from mice at the indicated ages and presented as normalized values to AF levels observed in 3 month old animals. Significance established with 1-way ANOVA followed by Dunnett’s post-hoc. The line depicts the linear regression (R2 = 0.95). (E) Percent AF+ microglia at indicated ages. Significance established with 1-way ANOVA followed by Dunnett’s post-hoc. All data are presented as mean ± SD. For panels C-E, n ≥ 6 animals per age from at least two independent experiments. BF = bright field, AF = autofluorescence, SSC = side scatter, ***p<0.001. See also Figure 2—figure supplement 1.

AF+ microglia selectively accumulate intracellular storage bodies with age

We next isolated AF+ and AF microglia from 3- and 18-month-old mice using fluorescence-activated cell sorting (FACS) and performed transmission electron microscopy (TEM) analyses. At 3 months of age, the intracellular organization of AF microglia subsets differed in that AF+ cells almost systematically contained large storage bodies filled with osmophilic electron-dense deposits (Figure 3A). These electron-dense storage bodies were observed in 80% of AF+ microglia whereas only 46% of AF microglia contained electron-dense organelles (Figure 3A,C). When observed in AF cells, storage bodies were devoid of complex storage material and were more regularly shaped than those observed in AF+ cells (Figure 3A). In aged mice, the storage bodies within AF+ microglia changed dramatically in both size and complexity (Figure 3A). Between 3 and 18 months of age, the percentage of visible cytoplasm occupied by storage bodies increased from 9% to 23% on average and approximately 30% of AF+ cells from aged mice contained AF material that occupied more than a third of the cytoplasm (Figure 3D). The ultrastructural composition of the storage bodies in AF+ cells varied with frequent curvilinear (black arrowheads) and fingerprint-like profiles (red arrowheads) (Figure 3B). White, linear, rod-like material with fine-tipped ends (white arrowheads) were observed within membrane-bound lipoid bodies and closely resembled cholesterol crystal deposits (Figure 3B). Most prominent, however, was the proportion and volume of lipid droplets (asterisks) (Figure 3B). In contrast, both the frequency of cells containing storage bodies and the proportion of the cytosol occupied by storage bodies remained unchanged in the AF microglia subset (Figure 3A,C,D).

AF+ microglia selectively accumulate intracellular storage bodies with age.

(A) Representative TEM images of sorted microglia subsets from 3- to 18-month-old mice. Red arrowheads point to storage bodies. Scale bars = 2 µm. Insets depict higher magnification images of selected AF+ microglia (dashed square lines). Inset scale bars = 500 nm. (B) Representative TEM images of storage bodies observed within aged AF+ microglia. Black arrowheads, curvilinear storage material; white arrowheads, crystal bodies; red arrows, fingerprint-like storage material; asterisks, lipid storage content. Scale bars = 500 nm. (C) Frequency of cells containing storage bodies in microglia AF subsets sorted and pooled from n = 10 mice at the indicated ages. Significance established with Fischer’s exact test. n = 48–59 cells/group. (D) Average percent of cytosolic area occupied by storage material in cells containing at least one storage body. Significance established with 2-way ANOVA followed by Tukey’s post-hoc. n = 11–21 cells/group. (E) Representative flow cytometry histograms and (F) corresponding quantitation of LAMP1 and CD68 staining in AF and AF+ subsets at indicated ages. Relative levels of LAMP1 and CD68 calculated as net geometric mean fluorescence intensity after subtraction of background AF signal and normalization to AF intensity detected in the AF subset at 3 months of age. Significance established with 2-way repeated-measures ANOVA followed by Tukey’s post-hoc. n ≥ 6 animals per age from at least two independent experiments. All data presented as mean ± SD. ns = not significant, **p<0.01 ***p<0.001.

In addition to these ultrastructural differences, AF+ cells expressed higher levels of LAMP1 and CD68 compared to AF cells (Figure 3E,F), indicating an enlargement of endolysosomal storage compartments in AF+ cells. Furthermore, a gradual age-dependent increase in LAMP1 and CD68 protein levels was observed in the AF+ subset (Figure 3E,F) whereas AF microglia did not show changes. Altogether these observations indicated that AF+ microglia differed from AF cells by their unique accumulation of endolysosomal storage compartments with aging.

Microglia AF subsets exhibit differential population dynamics upon depletion and replenishment of the microglia niche

To explore the cellular dynamics and ontogenic relationships between these two novel subsets of microglia, we depleted microglia in 14-month-old mice with the CSF1R-small molecule antagonist BLZ945 (Krauser et al., 2015; Pyonteck et al., 2013). Twenty-four hours following the treatment period, depletion was nearly complete, as assessed by the remaining frequency and absolute numbers of microglia (Figure 4A,B and data not shown). Consistent with previous reports (Huang et al., 2018; O’Neil et al., 2018), microglia rapidly repopulated the CNS at 7 and 14 days post-treatment, recovering on average to 25% (at 7 days) and 87% (at 14 days) of steady-state microglia numbers (Figure 4A and data not shown). However, repopulation by AF+ and AF subsets showed distinct kinetics. While AF microglia cell numbers reached 67% of steady-state levels by day 7, AF+ cells only repopulated to 4% of steady-state levels by that time (Figure 4B). At 14 days repopulating AF+ cell numbers only reached 14% of steady-state AF+ values in vehicle-treated mice while the repopulating AF subset surpassed steady-state levels by approximately 2-fold (Figure 4B) before normalizing to steady-state levels at day 70. Altogether these results established that the AF subset was the first subset to repopulate the depleted brain while the repopulation by AF+ cells was delayed, suggesting a possible conversion from AF to AF+ state during repopulation.

Figure 4 with 1 supplement see all
Microglia AF subsets exhibit differential population dynamics upon depletion and replenishment of the microglia niche.

(A) Representative flow cytometry scatterplots showing CD45 and CD11B surface levels in brain single cell suspensions from mice at indicated timepoints post treatment with vehicle or BLZ945 and (B) corresponding quantitation of absolute cell numbers of AF+ and AF microglia post-treatment. Significance established with 1-way ANOVA followed by Dunnett’s post-hoc test. (C) Representative flow cytometry histograms and (D) corresponding quantitation of AF intensity in the entire microglia population detected across multiple combinations of excitation lasers and emission filters and normalized to levels observed in vehicle-treated mice. In the scatter plot, the line represents the linear regression (R2 = 0.99). Significance established with 1-way ANOVA followed by Dunnett’s post-hoc test for each AF channel. (E) Representative flow cytometry scatterplots depicting KI-67 levels and AF intensity in microglia isolated at indicated time points post-treatment and (F) corresponding quantitation of the percent microglia positive for KI-67. Significance established with Welch’s ANOVA followed by Dunnett’s T3 post-hoc. n ≥ 8 animals per genotype group from at least two independent experiments. All data represented as mean ± SD. d = day; h = hour; B = blue laser; YG = yellow green laser; AF = autofluorescence; ns = not significant, ***p<0.001. See also Figure 4—figure supplement 1.

Supporting this hypothesis, the AF intensity of the repopulated microglia slowly increased from barely detectable levels at day seven to slightly increased levels at day 14, ultimately returning to a bimodal distribution by day 70 (Figure 4C,D). Even at the latter timepoint, only a small fraction of this newly-formed AF+ subset (19% on average) displayed AF intensity levels comparable to those seen in vehicle-treated mice (Figure 4C,D and Figure 4—figure supplement 1A,B) while most repopulating AF+ cells displayed 35% weaker AF signal intensity on average. This gradual accumulation of AF material over time in repopulating microglia supported the possibility that AF+ microglia were derived from the conversion of AF cells during replenishment. Further validating this conclusion, 56% of microglia displayed positivity for the proliferation-associated marker KI-67 at day seven before returning to steady-state values by day 14 (Figure 4E,F). Because AF+ microglia were virtually absent during this early repopulation phase (Figure 4B), these proliferation kinetics implied that the AF subset of microglia was the predominant subset responsible for the repopulation of the microglia compartment following depletion and that AF+ microglia were derived from the conversion over time of a subset of AF microglia.

Proteomic analysis of isolated AF+ and AF microglia subsets reveals molecular differences in endolysosomal, autophagic and metabolic pathways

To determine whether microglia AF subsets were distinct at the molecular level, FACS-isolated AF+ and AF microglia were analyzed by nano liquid chromatography mass spectrometry (nLC-MS). A total of 4231 proteins were detected by label-free LC-MS/MS and after filtering for proteins quantified in at least 50% of samples in either subset, 3492 remained for analysis. 351 proteins showed significant differences in abundance between AF+ and AF microglia (Benjamini-Hochberg adj. p value < 0.01, fold-change > |1.3|), with 254 and 97 upregulated and downregulated differentially expressed proteins (DEPs), respectively (Figure 5A and Figure 5—source data 1). When ranked by significance, 32 of the top 50 DEPs upregulated in AF+ microglia were associated with endolysosomal biology, among which was LAMP1, validating prior flow cytometry results (Figure 3E,F). Furthermore, a large number of lysosomal enzymes were upregulated in AF+ microglia including cathepsins (CTSA, CTSB, CTSD, CTSF, CTSL, CTSZ) as well as several enzymes involved in lysosomal degradation such as amidases (ASAH1, GBA, NAAA), thioesterases (PPT1, PSAP, NAGA), proteases (TPP1, LGMN) and glycosyl hydrolases (HEXA, HEXB, GLB1). Many other DEPs involved in the biology, trafficking, or fusion of endosomes with lysosomes and phagosomes (ATP6V0D1, SCARB2, GRN, ARL8B, TOM1, STX7, TMEM55B) were also upregulated in AF+ microglia. Underscoring the importance of these proteins towards maintaining proper CNS homeostasis, genetic perturbations in 15 of the top 50 DEPs were associated with severe CNS-related storage disorders, such as Neuronal Ceroid Lipofucinosis and Niemann-Pick. Lastly, DEPs that are typically associated with neurons (CBLN1, CBLN4, SNAP25), oligodendrocytes (MOBP, SYNE2) and astrocytes (GJA1) were detected at significantly higher levels in AF+ microglia, suggesting functional differences in either the phagocytic capacity of this subset or its ability to fully degrade ingested material.

Proteomic analysis of isolated AF+ and AF microglia subsets reveals molecular differences in endolysosomal, autophagic and metabolic pathways.

(A) Volcano plot comparing protein abundance in AF+ and AF microglia (x-axis: Log2 abundance difference, y-axis: negative log Benjamini-Hochberg adjusted p value from two-tailed Student’s t-test) with most-differentially expressed proteins annotated. Red lines: significance cutoffs (adjusted p value < 0.01, |Fold change| > 1.3). (B) Scatter plot displaying adjusted p value and GO term fold-enrichment results among differentially expressed proteins (DEPs) upregulated in AF+, with selected pathways highlighted. Dot color indicates number of proteins within GO pathway detected in the dataset (see legend). (C) Volcano plots comparing protein levels in AF+ and AF microglia with indicated GO term signatures overlaid in red. p value, χ2 statistical test for bias in distribution of GO term pathway members. (D) Top canonical pathways and (E) upstream regulators identified by IPA as differentially regulated between AF subsets. (D) Bars and dots respectively indicate p values and the percent of pathway proteins detected. (D, E) Predicted activation or inhibition of pathway or transcriptional regulator is indicated by positive and negative Z-scores, respectively, see legend. Analysis based on 351 DEPs identified using adjusted p value < 0.01 and fold change > |1.3|. See also Figure 5—source data 1, 2 and 3.

To systematically explore additional pathways distinguishing AF subsets at the molecular level, Gene Ontology (GO) term enrichment was performed with Panther (Figure 5B and Figure 5—source data 2). DEPs upregulated in AF+ microglia were enriched in pathways related to intracellular vesicle-mediated transport, lysosomal organization, protein transport, lipid catabolic processes and regulation of TOR signaling (Figure 5B,C). DEPs downregulated in AF+ microglia showed enrichment in RNA-related biological processes (transcription, RNA metabolism, splicing) and chromatin silencing. In agreement with GO term enrichment analysis, the top canonical pathways identified by Ingenuity Pathway Analysis (IPA) as enriched in AF+ microglia included phagosome maturation, autophagy, numerous catabolic pathways (amino acid and glutaryl-coA catabolism, ketogenesis and fatty acid β-oxidation), mitochondrial dysfunction, and pathways that were indicative of mTOR deregulation (mTOR, AMPK) (Figure 5D and Figure 5—source data 3). Finally, upstream regulators predicted to explain the distinct proteome displayed by AF+ cells included transcription factors involved in the regulation of cell cycle, senescence and apoptosis (TP53, MYC, CDKN2A), ER stress and unfolded protein response (XBP1), autophagy and lysosomal biogenesis (TFEB) and inflammatory responses (NFKBIA, PPARA) (Figure 5E and Figure 5—source data 3). Altogether, these results established that AF+ microglia expressed a unique proteome characterized by an increased representation of endolysosomal, autophagic, catabolic, and mTOR-related proteins.

Myelin, Fc receptor-mediated and TREM2-mediated phagocytosis are not dominant mechanisms contributing to microglia AF accumulation

Given the abundance of intracellular storage bodies and the increased detection of proteins involved in endocytosis and phagosome maturation observed in AF+ microglia, we investigated whether differential phagocytosis of myelin debris contributed to the subset-specific accumulation of AF storage material. Shiverer mice (Mbpshi/shi) harbor a spontaneous autosomal-recessive mutation in myelin basic protein (MBP) which results in defective myelin compaction and physical instability of the myelin sheath (Privat et al., 1979; Weil et al., 2016). Despite this myelin deficiency, there were only small differences in the frequency of AF+ cells, AF signal intensity, and LAMP1 levels in microglia from 70 day old Mbpshi/shi mice as compared to control animals (Figure 6A,B). Importantly, however, microglia from Mbpshi/shi mice upregulated markers known to be induced by the phagocytosis of apoptotic neuronal bodies such as CLEC7A/DECTIN1 (Krasemann et al., 2017; Figure 6—figure supplement 1A,B). Altogether, these observations validated that myelin instability was associated with increased microglia phagocytic activities in this model while also indicating that phagocytosis of unstable myelin sheaths or cellular debris was unlikely to be the primary driver of AF accumulation with aging.

Figure 6 with 2 supplements see all
The accumulation of AF storage bodies in microglia is genetically-controlled.

(A, C, E, G, H, J) Representative flow cytometry histograms and quantitation of AF subset frequency and/or AF signal intensity in AF+ microglia isolated from mice of indicated genotypes. Significance established with unpaired Student’s t-test (A, C, E, G), 2-way repeated measures ANOVA followed by Sidak’s post-hoc (H), and 1-way ANOVA followed by Tukey’s post-hoc (J). (B, D, F, I, K) Quantitation of LAMP1 levels in microglia isolated from mice of indicated genotypes, calculated as net geometric mean fluorescence intensity after subtraction of background AF signal and normalization to AF intensity detected in AF microglia from wild-type age-matched control mice. Significance established with unpaired Student’s t-test (B, D, F, I) and 1-way ANOVA followed by Tukey’s post-hoc (K). n ≥ 6 animals per genotype group from at least two independent experiments. All groups were age-matched and all data presented as mean ± SD. AF = autofluorescence; B = blue laser; YG = yellow green laser; ns = not significant, *p<0.05, **p<0.01, ***p<0.001. See also Figure 6—figure supplements 1 and 2.

To explore whether other phagocytic activities were contributors to AF accumulation, we examined Fc-receptor mediated and Triggering Receptor Expressed on Myeloid cells 2 (TREM2)-mediated phagocytosis using Fcer1g−/− mice, which lack the common Fc-gamma chain required for functional expression of activating Fc-receptors (Takai et al., 1994), and Trem2−/− mice, which were reported to have microglia defective in the engulfment of apoptotic neurons, myelin debris, and synapses (Cantoni et al., 2015; Hsieh et al., 2009; Wang et al., 2015). Microglia isolated from 9-month-old Fcer1g−/− and 6-month-old Trem2−/− mice displayed no differences in the frequency of AF+ microglia or AF intensity (Figure 6C,E), nor did they differ in LAMP1 levels when compared to controls (Figure 6D,F). Together, these results indicated that the generation of microglia AF was not dependent on either Fc-receptor- or TREM2-mediated phagocytosis.

Disruption of autophagy and lysosomal processing alter AF accumulation within AF+ microglia

Given that the proteomic analysis identified upregulation of autophagic proteins in AF+ microglia, we reasoned that the recycling of cellular components and their subsequent lysosomal degradation might contribute to the accumulation of AF material in microglia. To disrupt the formation of autophagic vacuoles in microglia, Atg5flox/flox mice were crossed to the Cx3cr1CRE-ERT2 mouse line and treated with tamoxifen at 6 weeks of age to induce Atg5 conditional deletion in CX3CR1+ cells (Figure 6—figure supplement 2A), hereafter referred to as Atg5−/−. In 12-month-old Atg5−/− mice, there was a significant decrease both in the frequency and the relative AF intensity of the AF+ subset (Figure 6G). Interestingly, in Atg5−/− mice the distribution of AF intensity in AF+ cells broadened significantly compared to controls. To better characterize this change in distribution, we divided the AF+ subset into AFdim and AFhi populations (Figure 6G). In Atg5−/− mice, the proportion of AFhi microglia was decreased compared to Atg5+/+(30% versus 36%, p=0.012, Figure 6H). While no difference was observed in LAMP1 (Figure 6I), CD68 was significantly lower in AF+ microglia from Atg5−/− mice (Figure 6—figure supplement 2B).

We next examined the contribution of lysosomal pathways in the regulation of AF accumulation. To this end, we isolated microglia from Cln3Δex7/8 homozygous knock-in mice (hereafter referred as Cln3KI/KI), which express the most common loss-of-function allelic variant of the lysosomal gene Cln3 found in patients with juvenile-forms of neuronal ceroid lipofuscinosis (NCL) (Cotman et al., 2002). While the proportion of AF+ microglia was modestly decreased in 5-month-old Cln3KI/KI mice, AF+ microglia isolated from Cln3KI/KI animals showed a striking increase in AF signal intensity compared to microglia from control mice (Figure 6J). Furthermore, the AF signal intensity was not equivalently increased across the entire light emission spectrum, with the highest difference (2.4-fold on average) observed in the short emission wavelengths (525 nm) whereas differences were more limited at 585 nm and no longer detected at 710 nm (Figure 6J), likely reflecting differences in both the magnitude and composition of AF accumulation in Cln3KI/KI animals. Consistent with a lysosomal origin of AF signal in wild-type mice (Figure 3E,F), LAMP1 protein was increased by more than 2-fold in AF+ microglia from Cln3KI/KI mice (Figure 6K). In contrast, LAMP1 levels in AF microglia remained unchanged across all genotypes (Figure 6K), establishing that lysosomal dysfunction selectively affected the subset of microglia accumulating AF material. In summary, autophagy and lysosomal dysfunction respectively attenuated and increased the accumulation of AF material in AF+ microglia, thereby establishing autophagy and lysosomal biology as important mechanisms contributing to the formation of the AF+ microglia subset.

Advanced aging and lysosomal dysfunction lead to the preferential cellular loss of the AFhi microglia population

Given that the relative proportion of AF+ microglia remained steady during the first year of life (Figure 2E), we next analyzed AF microglia subsets during advanced aging, defined in mice as 18 to 24 months of age. Similar to what we observed between 3 and 12 months of age (Figure 2C), the maximal intensity of AF in AF+ cells continued to increase between 12 and 24 months of age (Figure 7A). In contrast to the largely bimodal distribution of AF seen during the first 12 months of age, the distribution observed at 24 months of age was more continuous, with a noticeable downward shift in AF intensity detected on a per cell basis (Figure 7A). To quantify this change in distribution, we divided the AF+ subset into AFdim and AFhi populations as done previously (Figure 6G). While at 12 months of age 50% of microglia were AFhi, this population decreased to 32% on average by 24 months of age, with a concomitant increase in AFdim and AF cells (Figure 7B). This apparent relative loss of the AFhi microglia subset was accompanied by a 31% decrease in the overall number of microglia recovered from 24-month-old mice as compared to 12-month-old animals. This decreased recovery did not affect all AF microglia subsets equally but was uniquely attributed to a 58% decrease in the absolute number of AFhi cells (Figure 7C).

Figure 7 with 1 supplement see all
Advanced aging and lysosomal dysfunction lead to the preferential cellular loss of the AFhi microglia population.

(A) Representative flow cytometry histograms and (B, C) corresponding quantitation of AF subset frequencies (B) and absolute cell numbers (C) in 12- and 24-month-old naïve mice. Significance established with 2-way repeated measures ANOVA followed by Sidak’s post-hoc. (D–G) Representative scatterplots depicting DAPI and AnnexinV staining (D) and KI67 and CX3CR1 staining (F) in microglia subsets from 12- and 24-month-old mice and corresponding quantitation of the frequency of AnnexinV+ (E) and KI67+ (G) cells in each microglia subset. Significance established with 2-way repeated measures ANOVA followed by Sidak’s post-hoc. (H) Representative flow cytometry histograms of AF intensity in microglia from mice of indicated genotypes and ages and detected using two distinct cytometer channels and (I) corresponding quantitation of absolute cell numbers of microglia subsets in 18-month-old animals. Significance established with 2-way ANOVA followed by Sidak’s post-hoc. (J) Quantitation of percent microglia positive for AnnexinV in indicated AF subsets from 18-month-old animals of indicated genotypes. (K–L) Quantitation of CellROX DeepRed ROS-indicator staining in microglia subsets from mice at indicated ages and after subtraction of background AF signal at (K) baseline and (L) after treatment with ROS inducer tBHP. (M) Quantitation of Mitotracker dye staining as an indicator of mitochondrial content in microglia subsets from mice at indicated ages. Significance established with 2-way ANOVA followed by Sidak’s post-hoc. n ≥ 7 animals per genotype group from at least two independent experiments. All data presented as mean ± SD. AF = autofluorescence; B = blue laser; ROS = reactive oxygen species; tBHP = tert Butyl hydroperoxide; ns = not significant *p<0.05 **p<0.01 ***p<0.001. See also Figure 7—figure supplement 1.

To explore the possible factors involved in the preferential loss of AFhi microglia with age, we examined rates of proliferation and cell death in adult and aged animals. The frequency of microglia positive for AnnexinV increased in aged mice (4.0 ± 0.9%, n = 8) as compared to adult mice (1.9 ± 0.2%, n = 7) and was mostly attributed to AFhi and AFdim cells, which showed a 2-fold increase in the cell death rate at 24 months of age as compared to 12 months of age (Figure 7D,E). By contrast, the cell death rate seen in AF cells did not significantly increase between 12 and 24 months of age. When quantifying specifically early-apoptotic (AnnexinV+ DAPI) cells, AFhi and AFdim microglia exhibited similar relative increases in early-apoptotic rates between 12 and 24 months of age (Figure 7—figure supplement 1A), altogether suggesting that increased rates in apoptotic and necrotic death were both contributing to the overall 2-fold higher cell death rate seen in AFhi and AFdim microglia with aging. Unlike cell death rates, homeostatic proliferation rates at 12 months of age were generally higher in AF microglia than in AFdim (by 2-fold on average) and AFhi cells (by 7-fold on average) and did not change with aging, with the exception of AFhi microglia which showed a 2.6-fold increase in proliferative rate between 12- and 24 months of age (Figure 7F,G).

Because microglia from Cln3KI/KI mice accumulated higher levels of AF in the AF+ population (Figure 6J), we reasoned that accelerated accumulation of AF material may exacerbate the differences in AF subset survival. Accordingly, as early as at 9 months of age and more prominently at 18 months of age, we observed a marked decrease in the frequency of the AFhi subset, using channels that either showed increased AF signal at 5 months of age (e.g., Blue-525 nm) or channels that showed no increase in AF signal at this age (e.g., Blue-710 nm) (Figure 7H and Figure 7—figure supplement 1B,C). This decreased proportion also correlated with AF dose-dependent reductions in microglia cell numbers (by 30% and 85%, respectively, for AFdim and AFhi microglia, Figure 7I) and increases in overall cell death and early-apoptotic rates in AFhi and AFdim microglia, but not AF cells from Cln3KI/KI mice as compared to control mice (by 3.0- and 2.0-fold, respectively, Figure 7J and Figure 7—figure supplement 1D). Altogether, these observations supported a model where the age-associated progressive accumulation of AF material directly caused increased cell death rates, thereby leading to the selective decline of the AF+ subset observed during natural aging and the premature collapse of this subset upon accelerated accumulation of AF caused by genetic manipulation of the Cln3 pathway.

The AF+ subset is associated with an increased production of reactive oxygen species

Based on proteomic analyses, we hypothesized that the higher apoptotic rates observed in AF+ cells could result from mitochondrial dysfunction, the highly catabolic metabolism of AF+ cells or their reliance on fatty acid β-oxidation as a source of energy, all metabolic processes which are known to generate high levels of ROS. As measured by a ROS-sensitive cellular dye, AF+ microglia from 3- and 24-month-old mice displayed baseline ROS levels that were on average 2-fold and 2.4-fold higher than those observed in AF microglia, respectively (Figure 7K). After treatment with tert-Butyl hydroperoxide (tBHP), a potent inducer of cellular ROS, AF+ microglia generated ROS levels that were 1.5-fold and 2-fold higher than those observed in AF microglia at 3 and 24 months of age, respectively (Figure 7L). tBHP-induced ROS generation increased with aging and this increase was selectively observed in AF+ microglia that generated 1.6-fold more ROS at 24 months than at 3 months of age (Figure 7L).

To determine whether differences in mitochondrial content were contributing to the increased levels of ROS observed in AF+ cells, we assayed microglia cellular content using the Mitotracker probe. AF+ microglia exhibited, 1.5-fold and 1.6-fold higher levels of Mitotracker fluorescence on average than AF cells at 3 and 24 months of age, respectively (Figure 7M). There was no age-dependent increase in Mitotracker signal detected for AF+ microglia. Together, these data established that AF accumulation in AF+ microglia was associated with increased generation of ROS, which may contribute to the changes selectively observed in AF+ physiology with aging and lysosomal dysfunction.

Discussion

Generally, a cellular subset should exhibit shared properties with other cells within the cell type while exhibiting unique features and selective physiological functions that are independent of their microenvironment or external stimuli. It has therefore been suggested that true microglial subsets should be defined in unchallenged conditions by their intrinsic properties which translate into unique physiological functions during development or aging (Stratoulias et al., 2019). This report describes novel microglia subsets that precisely satisfy each criteria of this definition: (1) Cellular autofluorescence constitutes a cell-intrinsic photophysical parameter that unexpectedly distributed in a bimodal fashion in microglia; (2) The distinct subsets of microglia identified by this parameter were found in healthy, non-challenged conditions in both mice and non-human primates, irrespective of brain region or local microenvironment (e.g., various levels of myelin content such as in Shiverer mice) and independently of any specific stimulus or age (e.g., these subsets were present in young healthy brain as well as in aged animals); (3) AF+ cells selectively accumulated lysosomal storage bodies, therefore identifying the subset of microglial cells endowed with clearance functions in the brain and suggesting that this intrinsic AF property reflected unique physiological functions; (4) These subsets were detected longitudinally at a fixed ratio throughout most of adult life, implying the existence of homeostatic mechanisms maintaining a strict abundance ratio between them. Microglia that were negative for autofluorescence were detected even in advanced aging conditions, excluding the possibility that these subsets could be simple cellular phenotypes. While DAM, MgnD, LDAM and PAM microglia states (Hagemeyer et al., 2017; Kamphuis et al., 2016; Keren-Shaul et al., 2017; Krasemann et al., 2017; Li et al., 2019; Marschallinger et al., 2020) were previously described in disease settings, aging and during development, functionally-defined populations of microglia have not been identified or characterized in steady-state settings ‒ to the best of our knowledge ‒, despite the heterogeneity of microglia observed by mass cytometry and single cell transcriptomic methods (Hammond et al., 2019; Li et al., 2019; Masuda et al., 2019; Matcovitch-Natan et al., 2016; Mrdjen et al., 2018). The novel subsets described here, which we propose to name AF+ and AF, are predicted to have distinct functions and potentially represent the first defined subsets of microglia in the unchallenged healthy brain.

Lines of evidence point to a model where AF+ microglia derive from the gradual and synchronous conversion of a subset of AF cells: (1) AF was found to linearly increase from 3 months to 12 months of age, indicative of a progressive and cumulative biological process during aging; (2) Microglia proliferation following depletion was largely restricted to the AF subset which fully replenished within 14 days; (3) The post-depletion re-emergence of the AF+ subset was delayed until after proliferation had subsided. One limitation of these repopulation experiments is that we were unable to exclude the possibility that a small number of surviving AF+ cells were responsible for repopulating the AF+ niche over a longer period of time. Only future lineage tracing experiments or the adoptive-transfer of congenically-labeled AF+ and AF microglia into neonatal hosts as recently described (Hasselmann et al., 2019) will definitely establish the ontogenic relationship between these microglia subsets. Our repopulation results are in agreement with the reported decrease in both lipofuscin content and CD68+ lysosomal volume described by O’Neil and colleagues at 21 days following PLX5622-induced microglia depletion and repopulation (O’Neil et al., 2018). While the authors concluded that enforced microglia turnover reversed age-related lipofuscin accumulation, our data indicate that a discrete AF+ microglia population re-emerges post depletion and that repopulating AF+ microglia progressively accumulate AF with time, further establishing that AF accumulation is a progressive physiological process directly linked to aging. However, several important questions remain to be addressed in future studies, among which are: (1) the formal demonstration of the irreversibility of AF accumulation with aging; (2) the identification of the molecular cues responsible for the conversion from AF to AF+; and (3) the elucidation of the mechanisms maintaining AF subset ratios.

Historically, the gradual increase of AF in the brain with aging has been attributed to the accumulation of lipofuscin within postmitotic neurons and glial cells (Nakanishi and Wu, 2009; Seehafer and Pearce, 2006; Xu et al., 2008). Lipofuscin is characterized by a distinctive cytoplasmic accumulation of non-degradable fluorescent storage material comprised of highly crosslinked, polymeric and oxidized macromolecules as well as metal cations (Brunk and Terman, 2002). While lipofuscin typically displays spectral excitation and emission maxima in the range of 320–480 nm and 460–630 nm, respectively (Jung et al., 2010; Moreno-García et al., 2018; Warburton et al., 2007), the exact composition is known to vary between brain regions and cell types, likely impacting its precise spectral properties (Gilissen and Staneva-Dobrovski, 2013). In contrast, several exciting laser wavelengths (405, 488, 561, 640 nm) led to the emission of AF in microglia while the AF emission spectrum was also characterized by an equally wide range of wavelengths (450–780 nm) observed on a per cell basis. The latter property suggests that the AF signal detected within each AF+ microglia originates from complex subcellular content that may only partially overlap with traditional lipofuscin detected in other cell types. For example, while ceroid and lipofuscin exhibit some overlapping characteristics, these storage materials have unique compositions that in turn display subtle spectral differences (Seehafer and Pearce, 2006). The possibility that microglia AF observed by flow cytometry is detecting a wide range of storage material is further supported by the genotype-dependent spectral differences observed in Cln3KI/KI microglia, which displayed increased fluorescence intensity in select channel wavelengths, likely reflecting the specific accumulation of ceroid and not all possible subtypes of autofluorescence-emitting storage materials that may accumulate in AF+ microglia with normal aging. One limitation of this study is that the identification of AF subsets currently requires analysis of a single cell suspension by flow cytometry. Efforts leveraging proteomics-generated DEPs are currently ongoing to understand the spatial distribution in situ of AF+ and AF microglia in the context of different CNS regions and in relation to other cell types or subcellular structures such as synapses or apoptotic bodies. Understanding how aging and lysosomal dysfunction modify the spatial distribution of these two microglia subsets within the CNS should ultimately improve our understanding of the differential origin and composition of storage bodies accumulating in AF+ microglia. When imaged by TEM, several types of storage compartments were detected in AF+ microglia, among which were: lipid-based storage material which was reminiscent of the lipid droplets recently described in lipid droplet-accumulating microglia (LDAM) (Marschallinger et al., 2020; crystalline deposits which resembled cholesterol crystals previously described in foamy macrophages, lipid laden microglia and endothelial cells (Baumer et al., 2017; Cantuti-Castelvetri et al., 2018; Klinkner et al., 1995; Tangirala et al., 1994); and finally, curvilinear and fingerprint-like pattern depositions, which are hallmark features of storage disorders such as NCL and progranulin haploinsufficiency (Anderson et al., 2006; Cotman et al., 2002; Ward et al., 2017) and suggested a mechanistic convergence between lipofuscinosis/ceroid pathologies and microglia dysfunction associated with aging (Spittau, 2017).

In addition to differences in subcellular content and population dynamics, AF+ microglia showed distinct differences at the proteome level, characterized by increased representation of endolysosomal and autophagic pathways as well as proteins involved in oxidative phosphorylation, catabolic metabolism and fatty acid β-oxidation. These proteomic differences are reminiscent of previously reported effects of aging on the microglia proteome, which consisted of a bioenergetic shift towards fatty-acid utilization, perturbations in the regulation of gene expression and an upregulation of proteins suggestive of mitochondrial dysfunction (Flowers et al., 2017). However, while those changes were observed in microglia from 24-month-old mice, we detected these differences in AF+ microglia from mice as young as 12-month-old, suggesting that this aging signature may be primarily driven by the AF+ subset. GO term enrichment and IPA analyses also identified dysregulation of mTOR-related proteins and pathways in AF+ microglia. mTORC1, which integrates upstream pathways sensing nutrient availability, cellular energy and growth factors, regulates cell growth through its control of metabolism, autophagy, protein translation and organelle biogenesis (Liu and Sabatini, 2020). The activity of mTORC1 is tightly controlled by its recruitment to the lysosomal surface by the Rag-Ragulator complex, where is it brought into close proximity to the activating GTPase Rheb. The elevated levels of Ragulator components (LAMTOR1, 2, 4, 5) and Rag members RRAGA/B and RRAGC in AF+ cells are likely reflective of the increased lysosomal content present in these cells. The oxidative and catabolic metabolic profile of AF+ microglia along with their overrepresentation of autophagic proteins points to an impaired activation of mTOR in AF+ microglia and suggests that pharmacological modulation of mTOR may restore AF+ microglia physiology, a hypothesis that remains to be tested. Intriguingly, recent work exploring the effects of elevated mTORC1 signaling in Tsc1-deficient microglia described increased lysosomal biogenesis and phagocytosis that was coincident with decreased synaptic density and a more reactive microglial morphology (Zhao et al., 2018).

In prior studies which analyzed lipofuscin in microglia during aging (Mendes-Jorge et al., 2009; Moreno-García et al., 2018; Nakanishi and Wu, 2009), the assumption was that the accumulation occurred equivalently across all microglial cells. Instead, we have established that the progressive increase of AF material and associated accumulation of LAMP1+ storage bodies are occurring exclusively within AF+ microglia, highly suggestive of functional differences between subsets. In contrast, AF microglia remained devoid of AF signal during natural aging and even when lysosomal function was impaired by the Cln3Δex7-8 mutation. The persistence of the AF subset in those perturbed conditions suggests specific mechanisms or regional cues governing the distinct functions unique to each AF subset that will be the object of future investigation. Interestingly, aging-related myelin degradation, shedding and subsequent phagocytosis by microglia were proposed to be the primary contributors to the accumulation of lipofuscin (Safaiyan et al., 2016). However, our observations in the MBP-mutant Shiverer mice (Molineaux et al., 1986), which exhibit CNS hypomyelination and myelin sheath instability (Bird et al., 1978; Kirschner and Ganser, 1980; Privat et al., 1979; Safaiyan et al., 2016; Weil et al., 2016), do not fully support this model as the bimodal AF distribution was largely unaltered in Mbpshi/shi microglia despite their upregulation of CLEC7A/DECTIN1, which likely reflected their active phagocytosis of apoptotic bodies and cellular debris caused by myelin instability. Therefore, myelin phagocytosis and degradation are unlikely to be the primary contributor to the microglia AF signal. This conclusion is compatible with observations made in the subretinal layers of the eye where the increased lipofuscin signal observed in microglia with aging or age-related macular degeneration was not linked to myelin phagocytosis, but to the phagocytosis of rod outer segments and apoptotic retinal pigment epithelial cells (Santos et al., 2010; Xu et al., 2008). While the lack of alterations in AF with the loss of Fc-receptor- or TREM2-mediated phagocytosis indicates that those particular pathways are not key contributors, AF accumulation may still be impacted by alternative phagocytic pathways active in microglia that were not tested in this study, such as TAM (Tyro3, Axl, Mer) (Fourgeaud et al., 2016; Tufail et al., 2017), SIRPα-CD47 (Hutter et al., 2019) and C3-CD11b (Fu et al., 2012; Schafer et al., 2012). In contrast, we established that lysosomal and autophagosomal degradation pathways are primary mechanisms of accumulation of AF material, as evidenced by the accelerated and decreased AF accumulation respectively observed in AF+ microglia from Cln3Δex7-8 mice and Atg5-deficient microglia. Of all cell types in the brain, microglia express the highest transcript levels of Cln3, which is involved in the regulation of lysosomal pH, cathepsin activity, and endocytic trafficking (Cárcel-Trullols et al., 2017; Cotman and Staropoli, 2012; Golabek et al., 2000; Schmidtke et al., 2019). Although the disruption of Cln3 expression was present in both AF+ and AF cells in Cln3Δex7-8 mice, the latter subset was unaffected by the perturbation of the lysosomal pathway, suggesting that CLN3-dependent lysosomal degradation is dispensable in AF microglia, which further highlights the molecular differences between AF+ and AF microglia.

While lipofuscin accumulation has been linked to age-related neuronal and microglial functional decline (von Bernhardi et al., 2015; Brunk and Terman, 2002; Höhn and Grune, 2013; Jung et al., 2007; Kurz et al., 2008; Safaiyan et al., 2016; Sierra et al., 2007), the ability to precisely characterize its functional impact was impeded by the inability to isolate microglia which accumulated AF storage material from those which did not. Here, we overcame this limitation by developing a method permitting the isolation of these microglia subsets and established that the age-dependent accumulation of AF storage material was associated with increased mitochondrial mass and ROS production, decreased homeostatic proliferation and increased cell death rates in AF+ cells. Increased ROS production by aging AF+ microglia could be a direct consequence of the accumulation of AF material as lipofuscin can incorporate transition metals such as copper and iron, which form a redox-active surface catalyzing the Fenton reaction and promote mitochondria-independent cytotoxic effects (Höhn et al., 2010). Alternatively, the increased generation of ROS in AF+ could result from mitochondrial dysfunction, a pathway that was enriched in the AF+ proteome or the highly catabolic metabolism of AF+ cells and their reliance on fatty acid β-oxidation which is another major a source of ROS. In addition to ROS, the accumulation of undegraded macromolecules within lysosomes, which is known to inhibit key catabolic enzymes and permeases (Lamanna et al., 2011; Prinetti et al., 2011; Walkley and Vanier, 2009), could further propel the lysosomal system into complete dysfunction. The subsequent overaccumulation of storage material may also physically damage the lysosomal membranes, causing dispersal of storage contents into the cytosol or the extracellular space and subsequent inflammasome activation and neuronal cell death, as it was shown recently in a genetic model for acid sphingomyelinase deficiency and in a demyelination model (Cantuti-Castelvetri et al., 2018; Gabandé Rodríguez et al., 2018). Foam cell macrophages, which are generated by the overaccumulation of oxidized low-density lipid protein and cholesterol in atherosclerotic plaques, have been shown to contribute to vascular pathology via these precise mechanisms (Childs et al., 2016; Duewell et al., 2010; Gibson et al., 2018; Hakala et al., 2003). Interestingly, foam cell macrophages exhibit many of the ultrastructural features observed in aged AF+ microglia and thus may share common pathways precipitating cellular dysfunction and tissue pathology. Lastly, another possible mechanism by which AF material could trigger microglia dysfunction and cytotoxic effects is via its binding and inhibition of the proteasome and the associated perturbation of cellular proteolytic functions, as it was previously reported for lipofuscin (Höhn et al., 2011; Sitte et al., 2000; Szweda et al., 2003). Despite the striking stability of the AF subset ratio throughout most adult mice, cumulative and age-dependent AF accumulation ultimately resulted in the collapse of the ratio of AF+ to AF microglia in advanced aging, likely due to one of the aforementioned mechanisms. The accelerated accumulation of AF material in AF+ microglia from Cln3Δex7-8 mice phenocopied the effect of advanced aging on microglia by selectively increasing apoptotic rates of AF+ cells and resulting in an early and precipitous decline of the AF+ subset. Altogether, these results suggest a convergence of mechanisms operating during aging and in lysosomal storage disorders whereby the accumulation of storage bodies in AF+ cells drives microglia dysfunction, possibly contributing to neurodegeneration known to be associated with both conditions.

Together, these results provide novel insight into microglia physiology by identifying two previously unknown subsets which exist at steady-state in the murine and non-human primate CNS. The ability to identify and isolate distinct microglia subsets via a cell-intrinsic, label-free method presents future opportunities to further characterize any additional functions that may shed light on their subset-specific roles in maintaining the delicate balance between CNS homeostasis and neurodegeneration. The significance of these subsets is clearly highlighted by a model of juvenile neuronal ceroid lipofuscinosis in which lysosomal dysfunction was born solely by AF+ microglia, whereas AF microglia appeared unaffected. Lastly, the finding that the progressive depletion of AF+ microglia occurred as function of natural brain aging raises the possibility that this subset may be functionally protective and warrants further investigation into the biological factors regulating their survival.

Materials and methods

Key resources table
Reagent type
(species) or resource
DesignationSource or referenceIdentifiersAdditional information
Chemical compound, drugBLZ945;
CSF1R inhibitor
PMID:24056773CAS No. 953769-46-5
Chemical compound, drugTamoxifenMillipore-SigmaCat# T5648
Biological sample (C. macaque)NHP brain samplesCharles River Labsfreshly isolated same day
OtherDAPIThermoFisherCat# 62248Cell impermeant blue-fluorescent DNA stain for viability used at a concentration of 0.1 µg/mL
Antibodyanti-mouse CD11b BV510 (rat monoclonal)BiolegendRRID:AB_2561390FlowCyt (1:200)
Antibodyanti-mouse CD45 BV785 (rat monoclonal)BiolegendRRID:AB_2564590FlowCyt (1:200)
Antibodyanti-mouse CX3CR1 BV421 (rat monoclonal)BiolegendRRID:AB_2565706FlowCyt (1:200)
Antibodyanti-mouse P2RY12 PE (rat monoclonal)BiolegendRRID:AB_2721644FlowCyt (1:200)
Antibodyanti-mouse TMEM119 (rabbit monoclonal)AbCamRRID:AB_2744673FlowCyt (1:100)
Antibodyanti-rabbit AlexaFluor488 (rat monoclonal)BiolegendRRID:AB_2563203FlowCyt (1:400)
OtherAnnexinV AlexaFluor647ThermoFisherCat# A23204Apoptotic cell labeling reagent
Antibodyanti-mouse CD68 FITC (rat monoclonal)BiolegendRRID:AB_10575475FlowCyt (1:300)
Antibodyanti-mouse LAMP1 AlexaFluor647 (rat monoclonal)BiolegendRRID:AB_571990FlowCyt (1:200)
Antibodyanti-mouse KI-67 BV421 (rat monoclonal)BiolegendRRID:AB_2629748FlowCyt (1:100)
Antibodyanti-mouse Ly6G PE (rat monoclonal)BD BiosciencesRRID:AB_394208FlowCyt (1:100)
Antibodyanti-mouse Ly6C PE (rat monoclonal)BiolegendRRID:AB_1186133FlowCyt (1:100)
Antibodyanti-mouse CD3 PE (rat monoclonal)BD BiosciencesRRID:AB_395699FlowCyt (1:100)
Antibodyanti-mouse CD19 PE (rat monoclonal)BD BiosciencesRRID:AB_395050FlowCyt (1:100)
Antibodyanti-mouse NK1.1 PE (rat monoclonal)BD BiosciencesRRID:AB_396674FlowCyt (1:100)
Antibodyanti-mouse CD45 AF488 (rat monoclonal)BiolegendRRID:AB_493531FlowCyt (1:100)
Antibodyanti-mouse CX3CR1 BV785 (rat monoclonal)BiolegendRRID:AB_2565938FlowCyt (1:200)
Antibodyanti-mouse CD11b APC (rat monoclonal)BiolegendRRID:AB_312794FlowCyt (1:200)
Antibodyanti-NHP CD45 BV786 (mouse monoclonal)BD BiosciencesRRID:AB_2738454FlowCyt (5 µL/test)
Antibodyanti-NHP CD11b BV510 (mouse monoclonal)BD BiosciencesRRID:AB_2737996FlowCyt (5 µL/test)
Genetic reagent (M. musculus)Cx3cr1tm2.1(cre/ERT2)Jung (Cx3cr1–CreERT2)PMID:23273845RRID:IMSR_JAX:020940Dr. Steffen Jung (Weizmann Institute of Science, Israel)
Genetic reagent (M. musculus)Atg5floxPMID:16625204Dr. Noboru Mizushima, (University of Tokyo)
Genetic reagent (M. musculus)Fcer1g-/-Taconic BiosciencesModel 583PMID:8313472
Genetic reagent (M. musculus)Trem2-/-KOMP repositoryRRID:MMRRC_050209-UCD
Genetic reagent (M. musculus)MbpshiJackson LaboratoryRRID:IMSR_JAX:001428PMID:6168677
Genetic reagent (M. musculus)Cln3e × 7/ex8 (Cln3 KI)PMID:12374761Dr. Susan Cotman (Harvard Medical School)
Software, algorithmFlowJo V10TreeStarRRID:SCR_008520
Software,
algorithm
Prism 8.0GraphpadRRID:SCR_002798
Software, algorithmFIJIFIJIRRID:SCR_002285
Software, algorithmMaxQuant Version 1.6.0.16Max PlanckRRID:SCR_014485
Software, algorithmPerseusMax PlanckRRID:SCR_015753
Software, algorithmIngenuity Pathway AnalysisQiagenRRID:SCR_008653
OtherPercollGE HealthcareCat# 17-0891-01Medium for density
gradient centrifugation
OtherMitotracker DeepRedFMThermoFisherCat# M22426Mitochondrial fluorescent dye, used at 50 nM
OtherCellROX DeepRedThermoFisherCat# C10491ROS fluorescent detection agent, used at 500 nM
Sequence based reagentAtg5 Taqman primersThermoFisherAssayID: Mm01187301_m1, Cat# 4351372Spanning exons 3–4
Commercial assay or kitRNEasy Plus MicroQiagenCat# 74034
Commercial assay or kitSuperScript IV VILO kitThermoFisherCat# 11754050
Sequence based reagentGapdh endogenous control primers, M. musculusThermoFisherCat# 4352339E
Commercial assay or kitQuantiTect Multiplex PCRQiagenCat# 204541
Software, algorithmPANTHERPANTHERRRID:SCR_004869
Peptide, recombinant proteinLysyl Endopeptidase, Mass Spectrometry Grade (Lys-C)WakoCat# 121–05063
Peptide, recombinant proteinSequencing Grade Modified TrypsinPromegaCat# V5113

Mice

This study was performed in accordance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals. Research animals at Biogen were housed under specific pathogen free conditions in an AAALAC accredited facility with a 12 hr - 12 hr light - dark cycle and environmental conditions controlled at 21°C and 40–60% humidity. Animals handled according to an approved institutional animal care and use committee (IACUC) protocol (#756). This study was reviewed and approved by the Massachusetts General Hospital (MGH) Subcommittee of Research Animal Care (SRAC), which serves as the Institutional Animal Care and Use Committee (IACUC) for MGH (Protocol #2008N000013). C57BL/6J, B6.129P2-Apoetm1Unc/J, C3Fe.SWV-Mbpshi/J (Chernoff, 1981), B6J.B6N(Cg)-Cx3cr1tm1.1(cre)Jung/J (Cx3cr1CreERT2) (Yona et al., 2013) were purchased from Jackson Labs. B6.129S-Atg5<tm1Myok> (Atg5flox) (Hara et al., 2006) were acquired from Dr. Noboru Mizushima (University of Tokyo). Cln3Δex7/ex8 knock-in mice (Cotman et al., 2002) were generously provided by Dr. Susan Cotman (Harvard Medical School) and carry a ~ 1 kb genomic deletion in the lysosomal gene Cln3 that is orthologous to the commonly observed mutation in Juvenile Neuronal Ceroid Lipofuscinosis patients. This exonic deletion results in the production of an alternatively spliced mRNA species encoding a detectable non-truncated CLN3 mutant protein. B6.129P2-Fcer1gtm1Rav N12 (Fcer1g−/−) (Takai et al., 1994) were purchased from Taconic. Trem2tm1(KOMP)Vlcg (Trem2−/−) were acquired from the KOMP repository. Atg5−/− mice were generated by crossing the Atg5flox and Cx3cr1CreERT2 lines. Age-matched cohorts were used in all experiments. All mice used in studies were female unless otherwise noted. Mice were euthanized by CO2 inhalation.

Tamoxifen treatment

Request a detailed protocol

One gram of tamoxifen (Cat# T5648, Millipore-Sigma) was dissolved in 5 mL ethanol and then mixed with 45 mL of corn oil USP (Cat# CO136, Spectrum Chemical) warmed at 37°C for a final concentration of 20 mg/mL. 6 week old mice were given two 200 µL subcutaneous injections of tamoxifen-corn oil, spaced 48 hr apart to induce genetic recombination.

Isolation of murine microglia for flow cytometry

Request a detailed protocol

All solutions were used ice-cold unless otherwise noted. Immediately following CO2 euthanasia, mice were transcardially perfused with 20 mL of phosphate buffered saline (PBS) containing 3 mM EDTA. Brains were dissected and transferred into a 15 mL conical tube filled with Hanks’ Balanced Salt Solution (HBSS) on ice. For tissue homogenization, brains were transferred to a Petri dish and minced over ice with a scalpel. Minced tissue was transferred to a 7 mL glass dounce homogenizer (Cat# 57542, Wheaton) containing 5 mL of HBSS + 25 mM HEPES and homogenized with a pestle for approximately 10 strokes each. The single cell suspension was transferred to a 15 mL conical tube and centrifuged at 300xg for 5 min at 4°C. The supernatant was aspirated, and the cell pellet was gently resuspended in 1 mL of fetal bovine serum (FBS). A 33% isotonic Percoll solution was prepared by combining 9 mL of Percoll (Cat# 17-0891-01, GE Healthcare) with 1 mL of 10X HBSS and 20 mL of 1X HBSS + 25 mM HEPES. 9 mL of the 33% isotonic Percoll solution was then added and the cell suspension mixed by inversion. 1 mL of a 10% FBS/HBSS solution was overlaid on the Percoll suspension and the cells were centrifuged at 800 g for 15 min at 4°C with no brake. The resulting cell pellet was washed once in 10 mL HBSS + 25 mM HEPES, centrifuged at 300xg for 5 min at 4°C and resuspended in a final volume of 1 mL HBSS + 25 mM HEPES. Antibody staining was performed in a 96 well v-bottom plate (Cat# 3897, Corning) with the following antibodies after Fc-receptor blocking for 5 min (anti-CD16/32, clone 93, Biolegend): anti-CD45 BV785 (clone 30-F11, BioLegend), anti-CD11B BV510 (clone M1/70, BioLegend), anti-CX3CR1 BV421 (clone SA011F11, BioLegend), and anti-P2RY12 PE (clone S16007D, BioLegend). For unconjugated anti-TMEM119 (clone 101–6, Abcam) an anti-rabbit AlexaFluor488 secondary antibody (clone Poly4064, BioLegend) was used. DAPI (Cat# 62248, ThermoFisher Scientific) was used to assess viability. For apoptotic cell analysis, AnnexinV conjugated to AlexaFluor647 (Cat# A23204 ThermoFisher Scientific) was used according to the supplier protocol after surface staining. For intracellular antigens, cell suspensions were fixed and permeabilized using the Transcription Factor Staining kit (Cat# 00-5523-00, ThermoFisher Scientific) after surface staining with ZombieUV Fixable Viability dye (Cat# 423107 BioLegend). Fixation was followed by staining with CD68 FITC (clone FA11, BioLegend) LAMP1 AlexaFluor647 (clone 1D4B, BioLegend) and KI-67 BV421 (clone 11F6, BioLegend). Absolute cell counts were determined using 123count eBeads (Cat# 01-1234-42, ThermoFisher Scientific) Flow cytometry data were acquired on a five laser-equipped LSRFortessa X-20 (BD Biosciences) and analyzed in FlowJoV10 (Treestar). Non-autofluorescent peripheral immune cells were used as negative controls to determine the initial bifurcating gate to identify AF and AF+ microglia subsets. To subdivide the AF+ subset, a gate encompassing AFhi microglia was drawn to capture the AF+ gaussian peak. The AFdim gate was drawn between the edge of the AF population and the base of the AF+ gaussian peak (lower boundary of AFhi gate). Imaging flow cytometry data were acquired with an AMNIS Imagestream MKII (Millipore-Sigma) and analyzed with AMNIS IDEAS software.

Isolation of non-human primate microglia for flow cytometry

Request a detailed protocol

Immediately following euthanasia, Cynomolgus macaques were transcardially perfused with an ice-cold Lactated Ringers Solution (LRS). Brains were removed and placed into ice-cold LRS and kept chilled until processing. Approximately 500 mg of tissue from the frontal cortex was excised, minced and gently homogenized on ice in a 7 mL glass dounce homogenizer (Cat# 57542, Wheaton) containing 5 mL of HBSS + 25 mM HEPES. Microglia were then isolated using the same procedure as described above for murine microglia. Antibody staining was performed in a 96 well v-bottom plate (Cat# 3897, Corning) with the following antibodies after Fc-receptor blocking (Human TruStain FcX, Cat#422301, Biolegend): anti-CD45 BV786 (clone D058-1283, BD Biosciences), anti-CD11B BV510 (clone ICRF44, BD Biosciences).

Isolation of murine microglia for fluorescence activated cell sorting (FACS)

Request a detailed protocol

Single cell suspensions were prepared as previously described for flow cytometry microglia isolations through the initial centrifugation. The resulting cell pellet was resuspended in 5 mL of a room-temperature (RT) 70% isotonic Percoll solution, which was carefully overlaid with 5 mL of a 37% RT isotonic Percoll solution. The Percoll cell suspension was then centrifuged at 800 g for 25 min at 22°C with acceleration set to 50% and brake set to 10%. Enriched microglia were then carefully collected from the 37%/70% Percoll interphase to a new 15 mL conical tube previously blocked with 2% bovine serum albumin (BSA) in PBS at RT for 3 hr. Microglia were washed once with 10 mL of ice-cold HBSS + 25 mM HEPES and centrifuged at 300 g for 5 min at 4°C. The resulting microglia cell pellet was resuspended in 100 μL of HBSS + 2% BSA. Cell suspensions were Fc-blocked and then stained with anti-CD45 AF488 (clone 30-F11, BioLegend), anti-CD11B APC (clone M1/70, BioLegend), anti-CX3CR1 BV785 (clone SA011F11, BioLegend). DAPI was used for viability and a dump gate in the PE channel was created with the following antibodies: anti-Ly6G PE (clone 1A8, BD Biosciences), anti-Ly6C PE (clone HK1.4, BioLegend), anti-CD3 PE (clone 17A2, BD Biosciences), anti-CD19 PE (clone 1D3, BD Biosciences), anti-NK1.1 PE (clone PK136, BD Biosciences). Cell sorting was performed on a BD FACSAria Fusion instrument equipped with an 85 μm nozzle.

Proteomics sample preparation and data acquisition

Request a detailed protocol

75,000 microglia were FACS-isolated into tubes containing the dried residual of 150 μL 8M urea, 5 mM EDTA, 0.1M Tris/HCl pH 8.5, resulting in final volumes of 300 μL. The suspensions were mixed and stored frozen until used. Lysates were thawed and treated with a E220 focused beam ultrasonicator (Covaris) 20 times at 2 s each with 150W output. Lysates were clarified by centrifugation, reduced with 10 mM final concentration of DTT for 20 min and subsequently alkylated with 30 mM 2-iodoacetamide for 30 min at room temperature. After a two-fold dilution in water, samples were digested overnight with 0.05 μg LysC (Wako) and 0.2 μg modified trypsin (Promega). Peptides were acidified to a final concentration 0.5% (v/v) trifluoroacetic acid and desalted using C18 StageTips (Rappsilber et al., 2007). 50% of resulting peptides were separated on a 50 cm, 75 μm inner diameter EasySpray column packed with 2 μm PepMap C18 RSLC material (ThermoFisher) over 120 min using an EASY-nLC 1200 system (ThermoFisher). Peptides were analyzed online with an Orbitrap Fusion-Lumos mass spectrometer (ThermoFisher) in data-dependent acquisition mode. Full scans were acquired at a resolution of 240,000 in the Orbitrap analyzer and the most abundant precursors were selected in a 1 s scan cycle for higher-energy dissociation (HCD) with a 0.7 Th isolation window, a target of 10,000 ions and a maximum injection time of 25 ms, followed by detection fragment ion in the ion trap.

Proteomics data analysis

Request a detailed protocol

Raw data were processed in MaxQuant Version 1.6.0.16 (Cox and Mann, 2008) and searched with Andromeda (Cox et al., 2011) against a comprehensive SwissProt release for mouse with a false discovery rate of 1% at the peptide and protein level. Identifications of MS1 features were transferred between samples using the Match Between Runs option with a 0.2 min matching tolerance. Label-free protein quantification was performed using the MaxLFQ algorithm (Cox et al., 2014). Data analysis was performed in Perseus (Tyanova et al., 2016). After log transformation and applying a filter for proteins quantified in at least 50% of samples in either group, missing values were imputed with a normal distribution with a mean shifted down by 1.8-fold and a width of 0.3-fold in relation to the standard distribution of measured values of the respective sample. p values were calculated using Student’s t-test and the false discovery rate was controlled with the Benjamini-Hochberg method. Proteins were determined to be differentially expressed if they met an adjusted p value threshold of <0.01 and an absolute fold change difference threshold of >1.3. Human orthologs of identified mouse proteins were assigned using Panther and GO term enrichment of differentially expressed proteins was calculated with the Panther Overpresentation Test (Mi et al., 2019). Canonical pathway analysis and identification of upstream regulators was performed with Ingenuity Pathway Analysis (Qiagen).

Transmission electron microscopy

Request a detailed protocol

Microglia from pooled animals (n = 10 per age) were FACS-isolated directly into tubes containing 4% glutaraldehyde/0.1M sodium-cacodylate solution, pH 7.4 (Electron Microscopy Sciences, Hatfield, PA). Fixed cells were pelleted and post-fixed in 1.0% osmium tetroxide in cacodylate buffer for 1 hr at RT, rinsed in cacodylate buffer, and stabilized using 2% agarose in PBS. Agarose-embedded pellets were dehydrated through a graded series of ethanols to 100%, then dehydrated briefly in 100% propylene oxide. Specimens were infiltrated in a mix of 2:1 propylene oxide/Eponate resin (Ted Pella, Redding, CA) 2 hr at room temperature on a gentle rotator, then switched into a 1:1 mix of propylene oxide/Eponate resin and allowed to infiltrate overnight on a gentle rotator. 24 hr later, specimens were placed into fresh 100% Eponate resin and allowed to infiltrate for several hours, then embedded in flat molds with fresh 100% Eponate. Polymerization occurred within 24–48 hr at 60°C. Thin (70 nm) sections were cut using a Leica EM UC7 ultramicrotome, collected onto formvar-coated grids, contrast-stained with uranyl acetate and Reynold's lead citrate and examined in a JEOL JEM 1011 transmission electron microscope at 80 kV. Images were collected using an AMT digital imaging system with proprietary image capture software (Advanced Microscopy Techniques, Danvers, MA). TEM image analysis and area determinations for subcellular features were calculated in FIJI (Schindelin et al., 2012). Cytosolic area values were calculated by determining the full cellular area and subtracting the nuclear region.

Mitochondrial content

Request a detailed protocol

Mitotracker DeepRedFM (Cat# M22426, ThermoFisher Scientific) reagent was used according to the supplier protocol. Briefly, a single cell suspension of isolated microglia was surface stained with CD45 BV786 and CD11B BV510. Cells were pelleted and then resuspended gently in 500 μL prewarmed 37°C staining solution containing 50 nM of Mitotracker reagent. Cells were incubated for 30 min at 37°C in a humidified 5% CO2 incubator then pelleted and resuspended in 200 μL of HBSS+2% BSA for analysis by flow cytometry.

ROS detection

Request a detailed protocol

CellROX Deep Red (Cat# C10491, ThermoFisher Scientific) was used according to supplier protocol. Briefly, a single cell suspension of isolated microglia was surface stained with CD45 BV786 and CD11B BV510. Cells were pelleted and then resuspended gently in prewarmed RPMI1640 (Cat# 11875085, ThermoFisher Scientific) containing 10% FBS. Cells were treated with a 100 μM solution containing tert-butyl hydroperoxide for 30 min at 37°C in a humidified 5% CO2 incubator to induce ROS. CellROX reagent was then added to a final concentration of 500 nM and the cells were incubated for 45 min at 37°C. Cells were pelleted, resuspended in 200 μL of prewarmed RPMI1640/10%FBS containing DAPI (0.1 μg/mL final concentration) and then promptly analyzed on a flow cytometer.

Microglia depletion

Request a detailed protocol

To deplete microglia, 14-month-old mice were orally treated q.d. for 7 days with BLZ945 at a dose of 200 mg/kg body-weight (Krauser et al., 2015) in a 20% captisol solution. Vehicle-treated mice received only 20% captisol solution. Tissues were collected at the indicated time after last dose.

RNA isolation and quantitative real-time PCR

Request a detailed protocol

50,000–100,000 microglia were FACS-isolated directly into chilled RNA/DNA free microfuge tubes containing 500 μL RLT-plus lysis buffer supplemented with 1% 2-mercaptoethanol. RNA was isolated with the RNEasy Plus Micro Kit (Cat# 74034, Qiagen) according to the kit protocol. RNA quality was assessed using the Agilent RNA6000 Pico kit (Cat# 5067–1513, Agilent) on an Agilent Bioanalyzer system. RNA quantity was determined with the Quant-iT RiboGreen Fluorescence assay (Cat# R11490, ThermoFisher Scientific). Isolated RNA was then converted to cDNA using the SuperScript IV VILO kit (Cat# 11754050, ThermoFisher Scientific) per manufacturer protocol. Microglia cDNA was combined with a QuantiTect Multiplex PCR (Cat# 204541 Qiagen) master mix containing Taqman primers spanning Atg5 exons 3–4 (AssayID Mm01187301_m1, Cat# 4351372, ThermoFisher Scientific) and Gapdh endogenous control primers (Cat# 4352339E, ThermoFisher Scientific). qRT-PCR reaction was run for 40 cycles on a QuantStudio12K Flex instrument (ThermoFisher Scientific) and relative quantity (RQ) of Atg5 transcript was calculated using the 2ΔΔCt method.

Statistical analysis

Request a detailed protocol

Statistical analysis was performed using Prism software (GraphPad, San Diego, CA). To determine significance values when comparing AF subsets, a paired, two-tailed t-test or 2-way repeated measures ANOVA with Tukey’s or Sidak’s post-hoc multiple comparison’s test were utilized. For comparisons between genotypes, significance testing was performed using two-tailed independent t-test or 1-way ANOVA followed by Tukey’s post-hoc for multiple comparisons. For BLZ945 experiments, a 1-way ANOVA followed by Dunnett’s post-hoc test for multiple comparisons was used. A value of p<0.05 was considered statistically significant. All data were presented as the mean with standard deviation.

References

  1. 1
  2. 2
  3. 3
  4. 4
  5. 5
  6. 6
  7. 7
  8. 8
  9. 9
  10. 10
  11. 11
  12. 12
  13. 13
  14. 14
  15. 15
  16. 16
  17. 17
  18. 18
  19. 19
  20. 20
  21. 21
  22. 22
  23. 23
  24. 24
  25. 25
  26. 26
  27. 27
  28. 28
  29. 29
  30. 30
  31. 31
  32. 32
  33. 33
  34. 34
  35. 35
  36. 36
  37. 37
  38. 38
  39. 39
  40. 40
  41. 41
  42. 42
  43. 43
  44. 44
  45. 45
  46. 46
  47. 47
  48. 48
  49. 49
  50. 50
  51. 51
  52. 52
  53. 53
  54. 54
  55. 55
  56. 56
  57. 57
  58. 58
  59. 59
  60. 60
  61. 61
  62. 62
  63. 63
  64. 64
  65. 65
  66. 66
  67. 67
  68. 68
  69. 69
  70. 70
  71. 71
  72. 72
  73. 73
  74. 74
  75. 75
  76. 76
  77. 77
  78. 78
  79. 79
  80. 80
  81. 81
  82. 82
  83. 83
  84. 84
  85. 85
  86. 86
  87. 87
  88. 88
  89. 89
  90. 90
    Formation of cholesterol monohydrate crystals in macrophage-derived foam cells
    1. RK Tangirala
    2. WG Jerome
    3. NL Jones
    4. DM Small
    5. WJ Johnson
    6. JM Glick
    7. FH Mahlberg
    8. GH Rothblat
    (1994)
    Journal of Lipid Research 35:93–104.
  91. 91
  92. 92
  93. 93
  94. 94
  95. 95
    Secondary lipid accumulation in lysosomal disease
    1. SU Walkley
    2. MT Vanier
    (2009)
    Biochimica Et Biophysica Acta (BBA) - Molecular Cell Research 1793:726–736.
    https://doi.org/10.1016/j.bbamcr.2008.11.014
  96. 96
  97. 97
    Proteomic and phototoxic characterization of melanolipofuscin: correlation to disease and model for its origin
    1. S Warburton
    2. WE Davis
    3. K Southwick
    4. H Xin
    5. AT Woolley
    6. GF Burton
    7. CD Thulin
    (2007)
    Molecular Vision 13:318–329.
  98. 98
  99. 99
  100. 100
  101. 101
  102. 102
  103. 103
  104. 104
  105. 105

Decision letter

  1. Suzanne R Pfeffer
    Senior and Reviewing Editor; Stanford University School of Medicine, United States

In the interests of transparency, eLife publishes the most substantive revision requests and the accompanying author responses.

Acceptance summary:

This study provides a novel description of microglial heterogeneity during the process of aging, with careful characterization of a helpful and easily accessible biomarker-the presence or absence of autofluorescence. Increased autofluorescence in AF+ microglia occurs during normal aging, can be regulated by lysosomal pathways and autophagy, and leads to increased microglial apoptosis and ROS production.

Decision letter after peer review:

Thank you for submitting your article "Storage body accumulating microglia, a novel subset in healthy brain selectively impacted by aging" for consideration by eLife. Your article has been reviewed by Suzanne Pfeffer as the Senior Editor and Reviewing Editor and three reviewers. The reviewers have opted to remain anonymous.

The reviewers have discussed the reviews with one another and the Reviewing Editor has drafted this decision to help you prepare a revised submission. While we normally highlight the specific actions needed to revise the story, I instead include the detailed reviewer comments as many of the comments relate to the presentation of the data rather than serving as a list of needed experiments. Because there is no functional characterization, this would either need to be completed, or comments pertaining to function would need to be amended and a section outlining these caveats and future steps included. Also, data for other brain regions which was alluded to should be included; integration with the body of literature on lipofuscin would also be helpful.

We hope you will find the comments constructive in preparing a revised manuscript that we believe should be considered in the category of RESOURCE article.

We would like to draw your attention to changes in our revision policy that we have made in response to COVID-19 (https://elifesciences.org/articles/57162). Specifically, when editors judge that a submitted work as a whole belongs in eLife but that some conclusions require a modest amount of additional new data, as they may with your paper, we are asking that the manuscript be revised to either limit claims to those supported by data in hand, or to explicitly state that the relevant conclusions require additional supporting data.

Our expectation is that the authors will eventually carry out the additional experiments and report on how they affect the relevant conclusions either in a preprint on bioRxiv or medRxiv, or if appropriate, as a Research Advance in eLife, either of which would be linked to the original paper.Reviewer #1:

In "Storage body accumulating microglia, a novel subset in healthy brain selectively impacted by aging", the authors identify two subsets of microglia based on the presence or absence of autofluorescence (termed AF+ and AF-). The authors used young and aged mice to characterize changes in the AF+ microglial population during normal aging, including in knockout mice for phagocytic and lysosomal pathway genes. The authors conclude that (1) increased intensity of autofluorescence in AF+ microglia occurs during normal aging and is associated with increased storage body accumulation and proteomic changes, (2) AF intensity can be regulated by lysosomal pathways and autophagy, and (3) in advanced aging the accumulation of AF leads to increased microglial apoptosis and ROS production.

This study provides a novel description of microglial heterogeneity during the process of aging, with careful characterization of a helpful and easily accessible biomarker. It is an appealing topic to neuroimmunology and aging fields. Although the manuscript does not delineate the mechanisms regulating autofluorescence, nor its consequences on physiology, these topics represent exciting future directions. There are several points that, if addressed, would broaden appeal, increase interpretability, and/or strengthen conclusions.

1) All presented data examines single cell suspensions of microglia. Identification of AF+ microglia requires flow cytometry, the specific mechanisms regulating microglial autofluorescence are not known, and AF+ microglia have a range of different cellular inclusions. It is therefore challenging to know how to identify AF+ versus AF- microglia in tissue sections, but this would greatly enhance our ability to gauge their relevance. If the authors have any data or thoughts about how to identify AF+ microglia in situ, they would be highly useful to the field. If not, it could be helpful to mention this understandable limitation.

2) Subsection “Cellular autofluorescence identifies two discrete microglia subsets”: "AF microglia subsets were detected across different regions of the brain when isolated from cerebellum, cortex and hippocampus (data not shown)." Given ongoing debates around regional microglial heterogeneity, this is an interesting and important statement. If data are already in hand pertaining to regional differences in the frequency of AF+ and AF- microglia (or lack thereof), this would be of great appeal. If not, this sentence could be removed, as it is currently provocative but difficult to interpret.

3) Subsection “Microglia AF subsets exhibit differential population dynamics upon depletion and replenishment of the microglia niche”: "Because AF+ microglia were virtually absent during this early repopulation phase (Figure 4B), these proliferation kinetics implied that the AF- subset of microglia was the predominant subset responsible for the repopulation of the microglia compartment following depletion and that AF+ microglia were derived from the conversion over time of a subset of AF- microglia." The authors lack direct evidence that AF- cells progress into AF+ cells, and it is possible that a small number of remaining AF+ microglia can directly repopulate the AF+ niche, just with slower proliferation kinetics. New experimental data is not needed, but alternative hypotheses should be discussed.

4) Regarding Figure 7: To identify true, early apoptotic cells DAPI-AnnexinV+ cells should be the only ones counted. Cells stained for both AnnexinV and DAPI are not necessarily apoptotic, since membrane permeability permits Annexin V binding to the inner membrane leaflet. Cells may have a permeable membrane due to other pathways of cell death i.e. necroptosis, necrosis, pyroptosis. Presenting data on only DAPI-AnnexinV+ cells would increase interpretability of these results and strengthen the conclusion that AFhigh microglia are more prone to apoptotic cell death. Alternatively, the authors could change their language to include the possibility of other types of cell death occurring in AF+ microglia, or note these limitations in the Discussion section.Reviewer #2:

This manuscript describes a new classification system for microglia in the rodent and non-human primate brain that is based on autofluorescence. These AF+ and AF- microglia are equally distributed across brain regions (though this is not shown and would be a nice addition as the data is stated as being generated), does not differ according to sex, and maintains similar percentages following repopulation after depletion of microglia. Unlike other classifications of microglia in the disease state (e.g. disease-associated microglia, DAMs), AF+/AF- cells are present across species, and the proteomic analysis that suggests that have lysosomal instead of phagocytic differences is very interesting (this was validated using Trem2-/- and Fcer1g-/- mice). Overall, this is an interesting hypothesis, and provides a new avenue for characterizing microglia for future functional investigation.

My concerns are based largely on this function characterization – as while the authors state they have provided an improvement on other recently published transcriptomic studies by assessing function of these AF+/AF- microglia, they have not. What they have provided is a proteomic analysis of these cells – as no functional assays are provided in the manuscript. This does not detract from the overall hypothesis of the manuscript, but it would benefit from some taming of the language.

As there has been much controversy in the microglia field about binary nomenclature, perhaps some caution here is prudent. If the cells can be shown to have a functional difference, this would go a long way to dispel concerns in the field (otherwise more appropriate descriptions and removal of statements that suggest functional testing has been completed here would suffice).

The authors state that they have measured similar AF+/AF- cells in the cerebellum, cortex, and hippocampus. Given this statement is provided as proof of continuity of the subsets at several places throughout the manuscript – these data should be shown.

The data presented in Figure 1C is great, and the percentages seem more or less correct for the B-710 group of cells, but appears more 50/50 for R-780 cells. These are presumably lowly abundant? Can some quantification of these subtypes be extracted from these data?

The authors spend considerable time describing morphological changes in Figure 3 (panels A) – suggesting that AF- cells are more regularly shaped that AF+ cells, however the representative images in Figure 3A appears to show a different story – with AF- microglia having many 'ruffled' edges, while AF+ microglia are smooth. This is particularly noticeable in the 18 month cells. This highlights a real problem in the field of describing cell subsets using morphology – it can be misleading and biased. Considering the clear AF phenotypes, the morphology of the cells is not a strong addition here. Differences in cellular inclusions remain clear, though this might only be true at 3 months. This remains a weak section of the manuscript (which on the whole is strong and exciting). Similarly, in panels B the only described component here that is clearly visible is the crystal inclusions (white arrows). The curvilinear and fingerprint like profiles are not visible. This may be a resolution problem in review images, but this should be carefully checked and amended if necessary. This is also highlighted at the bottom of Page 16 where it is suggested that these 'crystal deposits resemble cholesterol crystals found in atherosclerotic plaues'. This is highly interesting, but this statement should be qualified or measured/validated in some way. Perhaps some careful review and rewriting could address this.

The AF+ dim/lo section of the manuscript is the weakest section of the manuscript and is not well-defined or incorporated into the remainder of the very compelling story. The dim/hi populations are not defined, and it feels like a move back to CD45hi/low populations that have caused problems in the field before. Indeed, the split of these cells in Figure 6, panel G, is arbitrary and very unclear how this was chosen. In Figure 7 dim and hi populations only appear visible on FACS plots in the 24 month animals, but the split in younger animals is still unclear, and seems arbitrary to gate them as such at earlier ages as the AF is a smooth transition from lower to higher levelsReviewer #3:

This manuscript unveils the presence of two distinct microglial populations defined by the presence or absence of autofluorescent (AF) material. Surprisingly, the proportion of AF+ and AF- cells is constant across animals, species (mouse and a non-human primate), and ages, while AF intensity grows linearly with aging in AF+ cells. In AF+ cells, the autofluorescence localizes in intracellular organelles. Ultrastructurally, AF+ cells exhibited large and complex storage bodies whose size markedly increases with aging, along with the signals of LAMP1 and CD68 (lysosomal membrane proteins) in AF+ (but not AF-) cells. Proteomics experiments revealed upregulation of proteins of the autophagy-lysosome pathway in AF+ cells.

The authors also show that AF+ and AF- cells exhibit dramatically different dynamics in post-microglia-depletion brain repopulation, with AF- cells repopulating the niche first, and then partially converting to AF+ cells.

Analysis of mouse lines with defective genes for myelination and phagocytosis indicated that these two pathways do not contribute to the generation of AF+ microglia. Conversely, KO of an autophagy gene decreased, whereas KO of a lysosomal gene increased, autofluorescence in AF+ microglia, revealing intersections with these pathways. Advanced aging was accompanied by a striking and selective loss of microglia with highest AF levels, which was accelerated in a genetic model of lysosomal dysfunction and was paralleled by an increase in ROS production and mitochondrial content.

The story is novel, well developed, and significant. The experiments are well controlled and largely supports the main conclusions of the work. The identification of a subpopulation of microglial cells that is likely responsible for managing catabolic challenges in aging and neurodegenerative disease is significant because it defines potential targets of treatments. More generally, the finding that microglial cells are divided in two distinct populations of fairly constant proportions is intriguing and raises several questions, including (1) How the conversion from AF- to AF+ cells is trigged at the molecular level? (2) Is this conversion irreversible? (3) What keeps the proportion of the two populations constant? These questions, however, are better addressed in follow-up studies, and I am mentioning them here only to underline how this work raises innovative questions in the field.

[Editors' note: further revisions were suggested prior to acceptance, as described below.]

Thank you for resubmitting your work entitled "Storage body accumulating microglia, a novel subset detected in the healthy brain and selectively impacted by aging" for further consideration by eLife. Your revised article has been evaluated by Suzanne Pfeffer (Senior Editor) and a Reviewing Editor. We would be pleased to publish this article after 2 minor issues are addressed.

Please address:

1) Figure 6—figure supplement 2, panel A lacks annotation of statistical significance testing, which should be added.

2) The title was (and remains) difficult to interpret. Please try to improve?

Reviewer #2 comments included for the authors' benefit (no changes required):

I particularly appreciate the addition of AF+ microglia data from additional brain regions. I think this adds a great deal to the manuscript to show these cells as a population that exists throughout the CNS. I look forward to the efforts described in response to reviewer #1 to visualize these cells using in situ. The high mag/high resolution EM image is really beautiful – and I appreciate this addition. I also appreciate the difficulties in isolation and culture of microglia to run functional tests. Having said this, the assays in the paper remain as descriptive and correlative, which is not a problem at all, but they remain not being functional assays. I do however appreciate that the LAMP1 staining is nice, the presence of oligo and neuron material inside the cells also suggests phagocytosis. Repopulation following depletion does provide information about a possible transition from AF- to AF+ with age, but as discussion by reviewer #1 there remains a possibility that AF+ cells repopulate from a very small population. I think however this has been addressed by changing wording – e.g. subsection “Proteomic analysis of isolated AF+ and AF microglia subsets reveals molecular differences in endolysosomal, autophagic and metabolic pathways” – and this is fine. As the first paper describing this subset/phenomenon, I am sure future studies will cover these specifics. The clarification on B-170 R-780 cell percentages is interesting, and the description of the laser integrity in pulling out individual populations is also interesting, but as the authors suggest likely not of biological importance.

https://doi.org/10.7554/eLife.57495.sa1

Author response

Reviewer #1:

In "Storage body accumulating microglia, a novel subset in healthy brain selectively impacted by aging", the authors identify two subsets of microglia based on the presence or absence of autofluorescence (termed AF+ and AF-). The authors used young and aged mice to characterize changes in the AF+ microglial population during normal aging, including in knockout mice for phagocytic and lysosomal pathway genes. The authors conclude that (1) increased intensity of autofluorescence in AF+ microglia occurs during normal aging and is associated with increased storage body accumulation and proteomic changes, (2) AF intensity can be regulated by lysosomal pathways and autophagy, and (3) in advanced aging the accumulation of AF leads to increased microglial apoptosis and ROS production.

This study provides a novel description of microglial heterogeneity during the process of aging, with careful characterization of a helpful and easily accessible biomarker. It is an appealing topic to neuroimmunology and aging fields. Although the manuscript does not delineate the mechanisms regulating autofluorescence, nor its consequences on physiology, these topics represent exciting future directions.

The authors would like to thank reviewer #1 for this in-depth review and appreciation of our findings as well as for highlighting the novelty and utility of microglia autofluorescence as a biomarker of microglia heterogeneity during aging. While we agree that we did not identify the exact biochemical nature of the storage bodies associated with the AF signal nor the functional consequences of AF accumulation on microglia physiology at the organismal level, we would like to emphasize two findings which partially answer these limitations:

a) Genetic analyses shown in Figure 6 identified autophagy and lysosomal functions as key mechanisms regulating autofluorescence accumulation.

b) Characterization of the effect of advanced aging and lysosomal dysfunction on the physiology of AF- and AF+ microglia (Figure 7) allowed us to define functional consequences of early and accelerated AF accumulation at the cellular level, including increased apoptosis, higher ROS production and selective loss of the AF+ cells.

There are several points that, if addressed, would broaden appeal, increase interpretability, and/or strengthen conclusions.

1) All presented data examines single cell suspensions of microglia. Identification of AF+ microglia requires flow cytometry, the specific mechanisms regulating microglial autofluorescence are not known, and AF+ microglia have a range of different cellular inclusions. It is therefore challenging to know how to identify AF+ versus AF- microglia in tissue sections, but this would greatly enhance our ability to gauge their relevance. If the authors have any data or thoughts about how to identify AF+ microglia in situ, they would be highly useful to the field. If not, it could be helpful to mention this understandable limitation.

The authors fully agree with reviewer #1 that developing a microscopy method to identify AF- and AF+ microglia subsets in situ would further our understanding of these microglia subsets in the context of CNS tissue regions and their interactions with other cell types or subcellular structures (e.g., synapse or apoptotic bodies). Efforts for developing IHC/IF methods to localize AF+ and AF- microglia in situ are currently ongoing and leveraging our proteomic dataset to define suitable markers. We have now updated the Discussion section to highlight this limitation of our findings and ongoing efforts to develop methods to identify these subsets in situ.

2) Subsection “Cellular autofluorescence identifies two discrete microglia subsets”: "AF microglia subsets were detected across different regions of the brain when isolated from cerebellum, cortex and hippocampus (data not shown)." Given ongoing debates around regional microglial heterogeneity, this is an interesting and important statement. If data are already in hand pertaining to regional differences in the frequency of AF+ and AF- microglia (or lack thereof), this would be of great appeal. If not, this sentence could be removed, as it is currently provocative but difficult to interpret.

We agree with reviewer #1 that the detection of AF subsets across different anatomical regions of the brain is an important finding. We have now included the corresponding data (Figure 1—figure supplement 1C,D) which establishes the presence of both AF- and AF+ microglia in cerebellum, cortex and hippocampus. This figure panel is now described in the Results section.

3) Subsection “Microglia AF subsets exhibit differential population dynamics upon depletion and replenishment of the microglia niche”: "Because AF+ microglia were virtually absent during this early repopulation phase (Figure 4B), these proliferation kinetics implied that the AF- subset of microglia was the predominant subset responsible for the repopulation of the microglia compartment following depletion and that AF+ microglia were derived from the conversion over time of a subset of AF- microglia." The authors lack direct evidence that AF- cells progress into AF+ cells, and it is possible that a small number of remaining AF+ microglia can directly repopulate the AF+ niche, just with slower proliferation kinetics. New experimental data is not needed, but alternative hypotheses should be discussed.

The authors agree that without additional lineage tracing experiments, it is technically possible that a small number of surviving AF+ cells were responsible for repopulating the AF+ niche over a longer period of time. We have now highlighted this alternative hypothesis in the revised version of the manuscript (Discussion section). While we can’t directly exclude this alternative hypothesis, we had included several lines of evidence that argued against that possibility (Discussion section). In the absence of lineage tracing capabilities available to us at this time to label and trace AF+ and AF- microglia, we are currently exploring alternative strategies to tackle this question, such as the transplantation of FACS-sorted microglia AF subsets from congenic donors into neonate recipients (PMID: 31375314). This future avenue of investigation is now highlighted in the Discussion section.

4) Regarding Figure 7: To identify true, early apoptotic cells DAPI-AnnexinV+ cells should be the only ones counted. Cells stained for both AnnexinV and DAPI are not necessarily apoptotic, since membrane permeability permits Annexin V binding to the inner membrane leaflet. Cells may have a permeable membrane due to other pathways of cell death i.e. necroptosis, necrosis, pyroptosis. Presenting data on only DAPI-AnnexinV+ cells would increase interpretability of these results and strengthen the conclusion that AFhigh microglia are more prone to apoptotic cell death. Alternatively, the authors could change their language to include the possibility of other types of cell death occurring in AF+ microglia, or note these limitations in the discussion section.

We would like to thank reviewer #1 for highlighting this lack of precision and accuracy in how we presented cell death data in Figure 7. The initial data presented were indeed inclusive of multiple modes of cell death and did not strictly reflect apoptotic rates, but overall cell death rates. The text has been modified accordingly in the Results section. We also reanalyzed the data to separately quantify apoptotic (AnnexinV+, DAPI-) and necrotic cells (AnnexinV+, DAPI+). We found that AFhi and AFdim microglia exhibited a 2-fold increase in both apoptotic and necrotic death rates between 12 and 24 months of age (Figure 7—figure supplement 1A), altogether suggesting that both increased apoptotic and necrotic death are contributing to the higher overall cell death rates of these subsets relatively to AF- microglia. In addition to Figure 7—figure supplement 1A, we performed a similar analysis in Cln3KI/KI mice and these results are now presented in Figure 7—figure supplement 1D.

Reviewer #2:

This manuscript describes a new classification system for microglia in the rodent and non-human primate brain that is based on autofluorescence. These AF+ and AF- microglia are equally distributed across brain regions (though this is not shown and would be a nice addition as the data is stated as being generated), does not differ according to sex, and maintains similar percentages following repopulation after depletion of microglia. Unlike other classifications of microglia in the disease state (e.g. disease-associated microglia, DAMs), AF+/AF- cells are present across species, and the proteomic analysis that suggests that have lysosomal instead of phagocytic differences is very interesting (this was validated using Trem2-/- and Fcer1g-/- mice). Overall, this is an interesting hypothesis, and provides a new avenue for characterizing microglia for future functional investigation.

The authors would like to thank reviewer #2 for their positive feedback on our work and appreciation that microglia AF subsets are present across species, something which is not true (or at least being debated) for other subsets of microglia described previously. As per reviewer #2 suggestion, we have now included in the revised manuscript the characterization of AF subsets across different anatomical regions of the brain (Figure 1—figure supplement 1C,D). This figure establishes the presence of AF- and AF+ microglia in cerebellum, cortex and hippocampus. Description of these observations is included in subsection” Cellular autofluorescence identifies two discrete microglia subsets”.

My concerns are based largely on this function characterization – as while the authors state they have provided an improvement on other recently published transcriptomic studies by assessing function of these AF+/AF- microglia, they have not. What they have provided is a proteomic analysis of these cells – as no functional assays are provided in the manuscript. This does not detract from the overall hypothesis of the manuscript, but it would benefit from some taming of the language.

As there has been much controversy in the microglia field about binary nomenclature, perhaps some caution here is prudent. If the cells can be shown to have a functional difference, this would go a long way to dispel concerns in the field (otherwise more appropriate descriptions and removal of statements that suggest functional testing has been completed here would suffice).

We are in agreement with reviewer#2 that we did not perform assays to directly assess functional differences between microglia subsets and moderated the functional claims throughout the manuscript. The principle reasons for the decision to eschew in vitro functional assays is the well-documented, rapid loss of homeostatic microglia identity that occurs upon culture on a plastic dish (PMID:28521131, PMID:30108586) in addition to the limited physiological relevance of the corresponding in vitro assays. Instead, we chose to provide several lines of indirect in vivo and ex vivo evidence in support of functional differences between AF subsets. These observations include:

- The steady, age-dependent increases in AF signal, LAMP1 levels and storage bodies observed selectively in AF+ microglia (Figure 1, Figure 2 and Figure 3) which are wholly reflective of a difference in the functional abilities of these microglia subsets in either the ingestion or the degradation of extracellular material. Of note, none of these readouts increased with aging in AF- microglia, thus pointing to a functional divergence in the lysosomal / degradative activity of the two subsets with aging.

- Further supporting these observations, proteins that are typically associated with oligodendrocytes (MOBP, SYNE2) or neurons (SNAP25, CBLN1, CBLN4) were detected in our proteomics dataset at significantly higher levels within AF+ microglia than AF-, providing further indirect evidence of increased phagocytic capacity of AF+ microglia, differences in the material they ultimately ingested, or differences in the degradative capacity of these subsets. We have now included these observations in subsection “ Myelin, Fc receptor-mediated and TREM2-mediated phagocytosis are not dominant mechanisms contributing to microglia AF accumulation” to highlight the differential detection of these neuronal/oligodendrocyte proteins.

- The differential ability of AF+ and AF- microglia to repopulate the brain upon depletion (Figure 4), pointing to another functional difference between those subsets that is also reflected in the differential levels of homeostatic proliferation, ROS production and cell death seen in AF+ and AF- cells (Figure 7).

-The differential genetic regulation of those subsets was highlighted by the lack of impact of Cln3 mutation on AF- microglia whereas the same mutation led to early cell death and loss of AF+ cells, indicating that AF+, but not AF- microglia, rely on lysosomal-related functions and indirectly demonstrating that these subsets are endowed with distinct functions.

The authors state that they have measured similar AF+/AF- cells in the cerebellum, cortex, and hippocampus. Given this statement is provided as proof of continuity of the subsets at several places throughout the manuscript – these data should be shown.

As mentioned in our response to reviewer #1, we agree that the detection of AF subsets across different anatomical regions of the brain is an important finding. As such, we have now included the data as part of Figure 1—figure supplement 1C,D.

The data presented in Figure 1C is great, and the percentages seem more or less correct for the B-710 group of cells, but appears more 50/50 for R-780 cells. These are presumably lowly abundant? Can some quantification of these subtypes be extracted from these data?

Reviewer#2’s observation is correct in that in Figure 1C where each row represents a single analyzed cell, the proportion of single cells positive for AF in the B-710 channel is roughly 70%. In contrast, this proportion is different in R-780 channel where it is roughly 50%. We believe that this difference is only apparent and is not necessarily reflecting a different subtype of AF microglia. Instead, we favor the hypothesis that these differences are more likely caused by the higher sensitivity of the B-710 channel compared to R-780. This distinct sensitivity in AF detection is very clear in Figure 1B when considering the brightness of the AF+ population in those two channels. We have included scatterplots below to better illustrate the differences in sensitivity between B-710 and R-780 in detecting AF+ microglia at different ages. Using alternate high-sensitivity channels to detect AF, such as YG-610 or YG-670, one can identify a similar proportion of AF+ microglia (~70%) whereas weaker channels such as R-730 and R-670 yield lower proportions similar to the R-780 channel. These important considerations have now been added in subsection “Cellular autofluorescence identifies two discrete microglia subset”.

Author response image 1

The authors spend considerable time describing morphological changes in Figure 3 (panels A) – suggesting that AF- cells are more regularly shaped that AF+ cells, however the representative images in Figure 3A appears to show a different story – with AF- microglia having many 'ruffled' edges, while AF+ microglia are smooth. This is particularly noticeable in the 18 month cells. This highlights a real problem in the field of describing cell subsets using morphology – it can be misleading and biased. Considering the clear AF phenotypes, the morphology of the cells is not a strong addition here. Differences in cellular inclusions remain clear, though this might only be true at 3 months. This remains a weak section of the manuscript (which on the whole is strong and exciting). Similarly, in panels B the only described component here that is clearly visible is the crystal inclusions (white arrows). The curvilinear and fingerprint like profiles are not visible. This may be a resolution problem in review images, but this should be carefully checked and amended if necessary. This is also highlighted at the bottom of Page 16 where it is suggested that these 'crystal deposits resemble cholesterol crystals found in atherosclerotic plaues'. This is highly interesting, but this statement should be qualified or measured/validated in some way. Perhaps some careful review and rewriting could address this.

The morphological characterization in subsection “AF+ microglia selectively accumulate intracellular storage bodies with age” was referring to intracellular storage bodies, not to the entire cell. This statement has been modified to minimize ambiguity. We apologize that the curvilinear fingerprint inclusions were not visible in the initial version of the manuscript that we submitted and will verify that the resolution is sufficient upon final submission. In the meantime, we have included below a high-resolution version of the image that was included in the initial submission.

Author response image 2

Cholesterol crystallization is a well-characterized phenomenon in lipid-laden macrophages. The TEM image we included in the manuscript matches well with previous reports of intracellular cholesterol crystals. For example, Klinkner et al., imaged macrophages incubated in vitro with oxidized-LDL by TEM (Klinkner et al., 1995) and described extensive lysosomal lipoid bodies, many in the process of conversion from the fluid to crystalline state. Fully developed crystal profiles were often enveloped by phospholipid lamellae that outlined the crystal contours and were observed in one of two distinct shapes: a short bar shape or a long thin arc. We have included a clearer description of the TEM cholesterol crystal observation in subsection “AF+ microglia selectively accumulate intracellular storage bodies with age” as well as additional references where this finding has been previously characterized by TEM are now included in the Discussion section.

The AF+ dim/lo section of the manuscript is the weakest section of the manuscript and is not well-defined or incorporated into the remainder of the very compelling story. The dim/hi populations are not defined, and it feels like a move back to CD45hi/low populations that have caused problems in the field before. Indeed, the split of these cells in Figure 6, panel G, is arbitrary and very unclear how this was chosen. In Figure 7 dim and hi populations only appear visible on FACS plots in the 24 month animals, but the split in younger animals is still unclear, and seems arbitrary to gate them as such at earlier ages as the AF is a smooth transition from lower to higher levels

We would like to thank reviewer #2 for their overall description of this manuscript as a compelling story, but also for highlighting this weaker section of the manuscript. We do agree that the introduction of AFdim and AFhi subsets may be confusing, especially in Figure 6. We believe however that subdividing the AF+ subset into AFdim and AFhi subpopulations is important in quantifying the change in AF distribution within the AF+ subset observed with advanced aging and in conditions of lysosomal or autophagy dysfunction. We have therefore made a number of changes to address reviewer#2’s comment:

- In Figure 7A, the histogram distribution of AF intensity in microglia from 24-month old mice (i.e., in conditions of advanced aging) shows a very different AF distribution compared to microglia from 12-month animals, with the clear alteration of the bimodal distribution and emergence of cells with intermediate levels of AF. In order to account for this significant change in distribution, we subdivided the AF+ population into 2 groups, AFhi and AFdim in the same manner performed in Figure 6G. Initially, non-autofluorescent peripheral immune cells were used as negative controls to determine the initial bifurcating gate to identify AF- and AF+ microglia subsets. To subdivide the AF+ subset, a gate encompassing AFhi microglia was drawn to capture the AF+ gaussian peak. The AFdim gate was drawn between the edge of the AF- population and the base of the AF+ gaussian peak (lower boundary of AFhi gate). To confirm the validity of this additional gating scheme, we leveraged the Cln3 data. While the AFhi cells normally formed a sharp gaussian peak in the B-710 channel, the peak completely dissipated at 18 months in Cln3KI/KI animals (Figure 7H, I), with only AF- cells and the intermediate AFdim cells remaining. The justification for this gating strategy is now included in the subsection “Mice”.

- We also amended Figure 6G to remain consistent with the neighboring panels while also introducing the subdivision of the AF+ population with gating used outlined in Figure 6G. We have also included a panel with the accompanying quantification of the AFhi and AFdim populations (Figure 6H).

Reviewer #3:

This manuscript unveils the presence of two distinct microglial populations defined by the presence or absence of autofluorescent (AF) material. Surprisingly, the proportion of AF+ and AF- cells is constant across animals, species (mouse and a non-human primate), and ages, while AF intensity grows linearly with aging in AF+ cells. In AF+ cells, the autofluorescence localizes in intracellular organelles. Ultrastructurally, AF+ cells exhibited large and complex storage bodies whose size markedly increases with aging, along with the signals of LAMP1 and CD68 (lysosomal membrane proteins) in AF+ (but not AF-) cells. Proteomics experiments revealed upregulation of proteins of the autophagy-lysosome pathway in AF+ cells.

The authors also show that AF+ and AF- cells exhibit dramatically different dynamics in post-microglia-depletion brain repopulation, with AF- cells repopulating the niche first, and then partially converting to AF+ cells.

Analysis of mouse lines with defective genes for myelination and phagocytosis indicated that these two pathways do not contribute to the generation of AF+ microglia. Conversely, KO of an autophagy gene decreased, whereas KO of a lysosomal gene increased, autofluorescence in AF+ microglia, revealing intersections with these pathways. Advanced aging was accompanied by a striking and selective loss of microglia with highest AF levels, which was accelerated in a genetic model of lysosomal dysfunction and was paralleled by an increase in ROS production and mitochondrial content.

The story is novel, well developed, and significant. The experiments are well controlled and largely supports the main conclusions of the work. The identification of a subpopulation of microglial cells that is likely responsible for managing catabolic challenges in aging and neurodegenerative disease is significant because it defines potential targets of treatments. More generally, the finding that microglial cells are divided in two distinct populations of fairly constant proportions is intriguing and raises several questions, including (1) How the conversion from AF- to AF+ cells is trigged at the molecular level? (2) Is this conversion irreversible? (3) What keeps the proportion of the two populations constant? These questions, however, are better addressed in follow-up studies, and I am mentioning them here only to underline how this work raises innovative questions in the field.

The authors would like to thank reviewer #3 for this very positive feedback on our submission and highlighting the novelty and potential of our findings to develop future therapeutics in the context of aging and neurodegeneration. We have now revised the Discussion section to include the important questions raised by reviewer #3 which we agree are of high importance for follow-up studies.

[Editors' note: further revisions were suggested prior to acceptance, as described below.]

Please address:

1) Figure 6—figure supplement 2, panel A lacks annotation of statistical significance testing, which should be added.

2) The title was (and remains) difficult to interpret. Please try to improve?

Reviewer #2 comments included for the authors' benefit (no changes required):

I particularly appreciate the addition of AF+ microglia data from additional brain regions. I think this adds a great deal to the manuscript to show these cells as a population that exists throughout the CNS. I look forward to the efforts described in response to reviewer #1 to visualize these cells using in situ. The high mag/high resolution EM image is really beautiful – and I appreciate this addition. I also appreciate the difficulties in isolation and culture of microglia to run functional tests. Having said this, the assays in the paper remain as descriptive and correlative, which is not a problem at all, but they remain not being functional assays. I do however appreciate that the LAMP1 staining is nice, the presence of oligo and neuron material inside the cells also suggests phagocytosis. Repopulation following depletion does provide information about a possible transition from AF- to AF+ with age, but as discussion by reviewer #1 there remains a possibility that AF+ cells repopulate from a very small population. I think however this has been addressed by changing wording – e.g. subsection “Proteomic analysis of isolated AF+ and AF- microglia subsets reveals molecular differences in endolysosomal, autophagic and metabolic pathways” – and this is fine. As the first paper describing this subset/phenomenon, I am sure future studies will cover these specifics. The clarification on B-170 R-780 cell percentages is interesting, and the description of the laser integrity in pulling out individual populations is also interesting, but as the authors suggest likely not of biological importance.

We thank the editorial staff for the quick decision on our manuscript revisions. We have amended Figure 6—figure supplement 2A and accompanying legend to include the statistics for that experiment. At the suggestion of the reviewing editor, we have also changed the manuscript title to "Differential accumulation of storage bodies with aging defines discrete subsets of microglia in the healthy brain" which we believe to be simpler and more precise than the previous title.

https://doi.org/10.7554/eLife.57495.sa2

Article and author information

Author details

  1. Jeremy Carlos Burns

    1. Multiple Sclerosis & Neurorepair Research Unit, Biogen, Cambridge, United States
    2. Department of Pharmacology & Experimental Therapeutics, Boston University School of Medicine, Boston, United States
    Contribution
    Conceptualization, Formal analysis, Investigation, Visualization, Methodology, Writing - original draft
    Competing interests
    currently a full-time employee of Biogen
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0003-3853-0237
  2. Bunny Cotleur

    Emerging Neurosciences Research Unit, Biogen, Cambridge, United States
    Contribution
    Investigation
    Competing interests
    currently a full-time employee of Biogen
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0003-4571-6435
  3. Dirk M Walther

    Chemical Biology and Proteomics, Cambridge, United States
    Contribution
    Data curation, Formal analysis, Investigation, Methodology
    Competing interests
    currently a full-time employee of Biogen
  4. Bekim Bajrami

    Chemical Biology and Proteomics, Cambridge, United States
    Contribution
    Formal analysis, Investigation, Methodology
    Competing interests
    currently a full-time employee of Biogen
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-4129-9390
  5. Stephen J Rubino

    Multiple Sclerosis & Neurorepair Research Unit, Biogen, Cambridge, United States
    Contribution
    Investigation, Writing - review and editing
    Competing interests
    currently a full-time employee of Biogen
  6. Ru Wei

    Chemical Biology and Proteomics, Cambridge, United States
    Contribution
    Supervision, Methodology, Writing - review and editing
    Competing interests
    currently a full-time employee of Biogen
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-7646-9046
  7. Nathalie Franchimont

    Multiple Sclerosis Development Unit, Biogen, Cambridge, United States
    Contribution
    Supervision, Writing - review and editing
    Competing interests
    currently a full-time employee of Biogen
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-2265-8378
  8. Susan L Cotman

    Center for Genomic Medicine, Department of Neurology, Massachusetts General Hospital, Harvard Medical School, Boston, United States
    Contribution
    Conceptualization, Resources, Supervision
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0003-3114-0543
  9. Richard M Ransohoff

    Third Rock Ventures, Boston, United States
    Contribution
    Conceptualization, Supervision, Writing - review and editing
    Competing interests
    currently a full-time employee of Third Rock Ventures
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0003-0175-6910
  10. Michael Mingueneau

    Multiple Sclerosis & Neurorepair Research Unit, Biogen, Cambridge, United States
    Contribution
    Conceptualization, Supervision, Writing - review and editing
    For correspondence
    michael.mingueneau@biogen.com
    Competing interests
    currently a full-time employee of Biogen
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-3873-7329

Funding

Batten Disease Research

  • Susan L Cotman

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Acknowledgements

The authors thank M Klein, A Nowell and Dr. E Butz for assistance in collection of Cln3Δex7/8 mice and the Batten Disease Research Foundation for funding and support of Dr. S Cotman’s laboratory. The authors acknowledge the support of the Boston University Biomolecular Pharmacology training program. The authors also acknowledge D Capen and the Microscopy Core of the Center for Systems Biology/Program in Membrane Biology (Mass General Hospital), which is partially supported by an Inflammatory Bowel Disease Grant DK043351 and a Boston Area Diabetes and Endocrinology Research Center (BADERC) Award DK05752.

Ethics

Animal experimentation: This study was performed in accordance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals. Research animals at Biogen were housed in an AAALAC accredited facility and handled according to an approved institutional animal care and use committee (IACUC) protocol (#756). This study was reviewed and approved by the Massachusetts General Hospital (MGH) Subcommittee of Research Animal Care (SRAC), which serves as the Institutional Animal Care and Use Committee (IACUC) for MGH (Protocol #2008N000013).

Senior and Reviewing Editor

  1. Suzanne R Pfeffer, Stanford University School of Medicine, United States

Publication history

  1. Received: April 21, 2020
  2. Accepted: June 21, 2020
  3. Accepted Manuscript published: June 24, 2020 (version 1)
  4. Version of Record published: July 17, 2020 (version 2)

Copyright

© 2020, Burns et al.

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.

Metrics

  • 2,353
    Page views
  • 403
    Downloads
  • 0
    Citations

Article citation count generated by polling the highest count across the following sources: Crossref, PubMed Central, Scopus.

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Download citations (links to download the citations from this article in formats compatible with various reference manager tools)

Open citations (links to open the citations from this article in various online reference manager services)

Further reading

    1. Immunology and Inflammation
    Matthew D Lauver et al.
    Research Article

    JCPyV polyomavirus, a member of the human virome, causes Progressive Multifocal Leukoencephalopathy (PML), an oft-fatal demyelinating brain disease in individuals receiving immunomodulatory therapies. Mutations in the major viral capsid protein, VP1, are common in JCPyV from PML patients (JCPyV-PML) but whether they confer neurovirulence or escape from virus-neutralizing antibody (nAb) in vivo is unknown. A mouse polyomavirus (MuPyV) with a sequence-equivalent JCPyV-PML VP1 mutation replicated poorly in the kidney, a major reservoir for JCPyV persistence, but retained the CNS infectivity, cell tropism, and neuropathology of the parental virus. This mutation rendered MuPyV resistant to a monoclonal Ab (mAb), whose specificity overlapped the endogenous anti-VP1 response. Using cryo EM and a custom sub-particle refinement approach, we resolved an MuPyV:Fab complex map to 3.2 Å resolution. The structure revealed the mechanism of mAb evasion. Our findings demonstrate convergence between nAb evasion and CNS neurovirulence in vivo by a frequent JCPyV-PML VP1 mutation.

    1. Immunology and Inflammation
    2. Neuroscience
    Nicola S Orefice et al.
    Research Article

    Cortical lesions represent a hallmark of multiple sclerosis and are proposed as a predictor of disease severity. microRNAs are suggested to be important players in the disease pathogenesis and the experimental autoimmune encephalomyelitis animal model. We implemented a mouse model recapitulating more closely the human pathology as it is characterized by both an autoimmune heterogeneity and the presence of cortical lesions, two parameters missing in experimental autoimmune encephalomyelitis. In our model, mice clustered in two groups displaying high or low clinical scores. Upon cortical cytokine injection, lesions appeared with a specific topography while cortical miRNA profiles were altered. These two features differed according to disease severity. We evidenced changes in miRNA regulators and targets suggesting that miRNA alteration had functional repercussions that could explain the differences in cortical lesions. This model represents a crucial tool for the study of both miRNA involvement and cortical lesion formation in disease pathogenesis.