Primary microglia were isolated from young adult (6–8 weeks) mice to assess rates of engulfment of myelin debris and latex beads (Fig. 1a). After 4 h of exposure, there was more uptake of myelin compared to latex beads, and this progressively increased with time (Fig. 1b). As phagocytosis declines with age in macrophages [34, 39], we sought to corroborate this finding in microglia from aged (> 15 month) mice. We observed a significant reduction in the uptake of myelin by microglia isolated from aged adult mice compared to microglia isolated from neonate or young adult mice (Fig. 1c, d).
The results of preferential phagocytosis of myelin debris over inert beads suggest the involvement of specific receptor(s). To address this, we first determined the time lag between microglia activation and phagocytosis. Microglia were exposed to myelin debris and staining for intracellular iNOS demonstrate that 80% of cells were stimulated after 2 h (Fig. 1e, f). We thus chose the 2 h time point for RNA sequencing and assessed the expression of selected signature genes associated with pro-inflammatory (Il1α, Tnf) or anti-inflammatory (Arg1, Tgfb1) microglia activity. Myelin debris upregulated the pro-inflammatory Il1α and Tnf but not the anti-inflammatory genes (Fig. 1g). Overall, 1714 genes were differentially expressed with age and activation state in myelin-stimulated young and aged microglia (Fig. 1h).
To determine the receptor(s) mediating phagocytosis of myelin debris, we examined the RNA sequencing dataset for genes differentially expressed with age and activation state (DAA). When comparing a scavenger receptor gene set to the DAA genes, we observed an overlap of 39.28% (11 out of 28 scavenger receptor genes; Fig. 1h). At 2 h and another 12 h time point, we found significant alterations in the scavenger receptors Cd36, Stab1, Cd68, Cd14, Msr1, Cd44, Cxcl16, Clec7a, Scarb1, Olr1, and Cd163 between young and aged microglia (Fig. 1i, j). As CD36 was significantly downregulated in aged microglia and is a known receptor for the phagocytosis of oxidized phospholipids present in myelin debris , we focused on CD36. We found that a CD36 receptor antagonist (sulfo-N-succinimidyl oleate sodium, SSO) reduced phagocytosis by aged microglia in a concentration-dependent manner (Fig. 1k, l). Overexpression of CD36 through in vitro transcription resulted in a significant increase in phagocytosis in microglia isolated from young and aged mice together with microglia isolated from human subjects (Fig. 1m, n).
To assess whether these findings translated to an in vivo model, the demyelinating toxin lysolecithin was injected into the ventrolateral funiculus of young (2–3 months) and middle-aged (9–12 months) mice. This typically produces in young mice an initial complete demyelination followed by repopulation of OPCs by day 7 and remyelination by day 14–21 . Analysis of lesion epicenter areas between young and middle-aged mice with the myelin stain, eriochrome cyanine, showed no statistically significant differences on days 3, 7, or 21 (Supplementary Fig. 1, online resource). Staining for the pan-macrophage/microglia marker Iba1 showed an initial low level of immunoreactivity in lesions of middle-aged mice (17% decrease compared to young mice) that eventually reached levels of young mice on day 21 (Fig. 2a, b). This delay was due to lower number of monocyte-derived macrophages rather than microglia in lesions of middle-aged animals when CX3CR1CreER:Rosa26Tdt (Ai9) mice were used to discriminate Ai9+ microglia from Ai9− monocytes (24 h: 32% decrease; 72 h: 65% decrease)  (Fig. 2c–e). Phagocytosis in lesions of middle-aged mice was examined with Oil Red O (ORO) to detect neutral lipids [24, 41]; phagocytosed lipids concentrated in phagolysosomes have an intense, punctate appearance (Fig. 2f). Quantitation of punctate ORO showed significantly less ingested lipids on day 7 (42% decrease) in lesions of middle-age mice compared to lesions of young mice (Fig. 2g).
We previously screened a 1040 drug library for their capacity to enhance TNF-α production in lipopolysaccharide (LPS)-exposed adult human microglia and identified niacin (nicotinic acid, vitamin B3), meclocycline, bumetanide and orlistat with this effect . Here, we used macrophages since they are a major phagocytic population in many neurological lesions , and can be purified easily from the bone marrow of young (2–3 months) and middle-aged (9–12 months) mice. All four compounds by themselves did not elicit an elevation of TNF-α (data not shown), but they significantly enhanced TNF-α levels in LPS-stimulated bone marrow-derived macrophages (BMDM) from both age groups (Fig. 3a, b). In considering which stimulator would be the best candidate for potential clinical use, we noted that meclocycline has an unfavorable safety profile, oral orlistat is not absorbed from the intestinal tract, and bumetanide has poor CNS penetrance (www.drugs.com). Niacin, however, is widely used and tolerated at high doses for prolonged periods and displays high bioavailability when taken orally to treat dyslipidemia [8, 25]. Moreover, the concentration we used of 100 μM is achievable in plasma in humans  and niacin is detected in the human brain after administration . We assessed whether niacin is able to improve microglial phagocytosis of myelin debris and found a significant enhancement of myelin debris uptake by both young and aged mouse microglia, as well as human microglia (Fig. 3c, d).
Niacin exerts its anti-lipolytic effects through the niacin receptor, also known as GPR109A or the hydroxycarboxylic acid receptor 2 (Hcar2) [3, 25, 49]. To determine the involvement of this receptor, we isolated BMDM from wildtype and Hcar2−/− mice. Multiplex Luminex analyses supported the TNF-α ELISA results and niacin induced elevation of several cytokines and chemokines in LPS-exposed wildtype BMDM that was abrogated in Hcar2−/− cells (Supplementary Fig. 2a–h, online resource).
To determine whether niacin was binding to Hcar2 to mediate the increase in phagocytosis, we treated wildtype and Hcar2−/− cells with niacin and only observed enhancement in phagocytosis in wildtype cells (Fig. 3e). To address potential scavenger receptors mediating this increase in phagocytosis, we assessed transcripts from wildtype and Hcar2−/− BMDM treated with niacin. We found that Cd68 and Mertk were not affected by niacin treatment (Supplementary Fig. 2i, j, online resource). However, niacin elevated Cd36, which is downregulated in aged microglia and important for the phagocytosis of myelin debris (Fig. 3f).
Next, we tested whether the niacin-mediated enhancement in phagocytosis was through CD36. Microglia were exposed to myelin and were treated with niacin in the presence of the CD36 receptor antagonist SSO. When niacin was added alone to microglia from young and aged mice, as well as from humans, a significant increase in phagocytosis was observed compared to control (Fig. 3g, h). In contrast, SSO ameliorated niacin-enhanced myelin phagocytosis (Fig. 3g, h). These results identify niacin as a novel stimulator of macrophage/microglia phagocytosis in culture that acts through Hcar2 to upregulate CD36 culminating in myelin engulfment.
Since Hcar2 expression was elevated in demyelinating lesions (young: 4361% increase; middle-aged: 1896% increase; Fig. 4b), we asked whether Hcar2 is necessary for remyelination to proceed efficiently. We induced lysolecithin lesions in young wildtype and Hcar2−/− mice and found no difference in the rate of OPC recruitment or remyelination, suggesting that Hcar2 is not necessary for spontaneous remyelination (Supplementary Fig. 3a, b, online resource). We next addressed whether pharmacological stimulation of Hcar2 with the systemic administration of niacin would be sufficient to enhance remyelination in middle-aged mice (Fig. 4a). Wildtype or Cx3cr1GFP/+:Thy1YFP+ mice, demyelinated with lysolecithin, were treated with saline vehicle or niacin (100 mg/kg IP) once a day, with treatment starting 24 h after lysolecithin (Fig. 4a). From the literature, this dose results in an initial spike in plasma concentration, after which the concentration is maintained at approximately 240 µM from 2 to 6 h post-administration . From 6 to 9 h, the plasma concentration decreases to 120 µM, resulting in a concentration similar to the one achieved in humans taking 1 g of niacin . In our experiments, live imaging revealed that the macrophages/microglia in niacin-treated middle-aged mice had higher mean surface area (59% increase) and volume (82% increase), and lower mean sphericity (that is, less ameboid; 4% decrease) (Fig. 4c–f), indicating that niacin enhanced their process extension and elevated their surveillance within lesions.
Since myelin debris is non-permissive for remyelination and needs to be removed for remyelination to occur [22, 37], we investigated the effect of niacin on myelin phagocytosis in vivo. Extending our tissue culture results (Fig. 3f), we found that niacin significantly increased the mean fluorescence intensity of CD36 expression by monocyte-derived macrophages and microglia within the lesion (Fig. 4g–j). To determine whether the elevated CD36 corresponded to lipid engulfment, we performed ex vivo live imaging experiments in which addition of the lipophilic dye, Nile Red, labels lipids that have been engulfed by Cx3cr1GFP/+ cells [39, 48]. We found that systemic niacin treatment significantly enhanced the amount of engulfed lipids within CX3CR1GFP/+ macrophages/microglia on day 3 in middle-aged mice (512% increase) (Fig. 4k, l). Niacin thus promotes the clearance of inhibitory myelin debris from lesions in middle-aged animals.
Other effects of niacin were examined. Ex vivo multiphoton live imaging of CX3CR1GFP/+ mice at day 3 showed that niacin did not influence the motility of macrophages/microglia within lesions in middle-aged animals (Supplementary Fig. 4, online resource). Niacin did not affect the inflammatory profile of circulating blood monocytes after 3 or 7 days of daily treatment (Supplementary Fig. 5, online resource). In lesions on both day 3 and day 7 (Supplementary Fig. 4b and 6a,b, online resource), niacin did not alter the density of CX3CR1GFP/+ or Iba1-stained macrophages/microglia, or changed the proportion of cells within the lesion that were microglia or infiltrating macrophages (Supplementary Fig. 6c,d, online resource). We conclude that niacin is not affecting circulating monocytes or the recruitment of macrophages/microglia into lesions in middle-aged animals, but that it promotes the intralesional phagocytic clearance of inhibitory myelin debris.
With the removal of inhibitory myelin debris, daily niacin treatment increased the density of Olig2+ oligodendrocyte lineage cells (41% increase), including Olig2+PDGFRα+ OPCs (76% increase) (Fig. 5a–c) in demyelinated lesions in middle-aged mice. As niacin does not have a direct effect on OPCs in vitro (Supplementary Fig. 7, online resource), it seems likely that niacin is acting through the phagocytic removal of inhibitory myelin debris to enhance OPC recruitment within the lesion.
We examined whether systemic niacin treatment for the first 7 days after lysolecithin injection, when myelin breakdown and removal is largely occurring, was sufficient to promote remyelination on day 21 in middle-aged mice. This was first addressed by quantifying the percentage of MBP-positive rings within lesions, presumed to be newly formed myelin around axons. Niacin significantly increased the percentage of MBP-positive area within the lesion (138% increase) (Fig. 5d, e). To confirm that remyelination had occurred, we examined lesions by transmission electron microscopy (EM) (Fig. 5f). Lesions from mice treated with vehicle or niacin had the same density of axons (Supplementary Fig. 8, online resource), suggesting that niacin was not neurotoxic despite its potential for elevating cytokines in culture (Fig. 3); the safety was corroborated by a similar expression of the pro-inflammatory IL-1β in lesions between groups (Supplementary Fig. 9, online resource).
The EM images with quantitation throughout the lesion core showed that the percentage of remyelinated axons was elevated by niacin (62% increase) (Fig. 5g). g ratio measurements for myelin thickness (1 is complete demyelination) across all axonal calibers (Fig. 5h, i) provided an initial impression of lower g ratios (i.e., remyelinated) on larger axons in the niacin group (Fig. 5h), so we delineated small and large axons based on a 2.5 μm axonal caliber cutoff. The results show remyelination occurring similarly on small axons in both niacin and vehicle groups (Fig. 5j), but large axons were minimally remyelinated in middle-aged mice unless treated with niacin (Fig. 5k). These results demonstrate that systemic treatment with niacin improves remyelination in the aging CNS, specifically in larger caliber axons that otherwise would not be remyelinated in aging after lysolecithin-induced demyelination.
Recently, it was shown that the ability for macrophages/microglia to export phagocytosed cholesterol is important for preventing the formation of cholesterol crystals and lysosomal rupture within macrophages/microglia in demyelinated lesions, and that this is reduced in aging . The cholesterol efflux is regulated by reverse cholesterol transporters such as adenosine triphosphate-binding cassette (ABC) A1 (ABCA1) and G1 (ABCG1) on the plasma membrane , which are in turn controlled by liver × receptor alpha  that was previously shown to enhance remyelination in the aging CNS . We did not find that niacin elevated Abca1 or Abcg1 in our study (Supplementary Fig. 10, online resource), thereby suggesting for the future that the combination of niacin promoting myelin engulfment together with a drug to stimulate cholesterol efflux could be even more effective for remyelination in aging animals.