Inefficient clearance of myelin debris by microglia impairs remyelinating processes

In the cuprizone model, CNS infiltration of BM-derived cells (BMDCs) has been reported by generating chimeric mice using total body irradiation coupled with BM transplantation (McMahon et al., 2002). However, the current understanding of the subject is that irradiation along with BM transplantation induces unwanted infiltration of cells into the CNS of untreated healthy animals and compelling evidence has demonstrated that this infiltration is not a physiological process (Ajami et al., 2007; Mildner et al., 2007). To address this issue, BM cells extracted from either GFP^+/− or CCR2^−/− donor mice were injected into the tail vein of mice pretreated with a busulfan/cyclophosphamide (BuCy) based chemotherapy regimen (Fig. 1 A). This regimen leads to a strong chimerism (>90%) in the circulation (Fig. 1 B) but does not induce artificial infiltration of BMDCs into the CNS under physiological conditions (Fig. 1 D) as expected (Lampron et al., 2012). 6 wk after transplantation, cuprizone supplemented chow was given to these mice that were sacrificed at various time points (Fig. 1 A). A significant infiltration of peripheral cells into the CNS was observed, with a peak of infiltration at weeks 4 and 5 of treatment (Fig. 1 C). Infiltrated cells were attracted to the sites of demyelinated white matter, most notably the corpus callosum (CC; Fig. 1 E) and striatum, but not the cerebellum (unpublished data). This infiltration was transient, as limited numbers of GFP⁺ cells were seen after 6 wk of treatment and even less when regular chow was reintroduced for 2 wk (Fig. 1 C, RM). The knockout of CCR2 in the periphery abrogated the infiltration of BMDCs into the CNS of cuprizone-fed mice by 96% (Fig. 2, A and C). 81% of infiltrated cells were also positive for microglial marker Iba1 (Fig. 2, A and B) and also expressed CD68, a marker of the myeloid lineage (Pulford et al., 1990; not depicted). A lower proportion of the few (<30) GFP⁺ cells detected in CCR2^−/− chimeric mice were Iba1 positive (38%, Figs. 1 F and 2 A). T cell marker CD3 was detected in few GFP+ cells found in CCR2^−/− and GFP chimeric mice (data not shown). F4/80-immunoreactivity (ir) signal was also detected in a low proportion of infiltrating cells (Fig. 2, D and E). Black Gold staining was used to measure the area of demyelination in the CC of these animals after 6 wk of cuprizone diet (Fig. 3, A and B) or the extent of remyelination when regular chow was reintroduced for 2 wk (Fig. 3 C). There was extensive demyelination of the CC in both GFP and CCR2^−/− chimeric mice (Fig. 3 A). Peripheral CCR2^−/− knockout and the blockade of BMDC infiltration into the CNS did not impact the subsequent remyelination either (Fig. 3 C). These results show that there is a regulated recruitment of monocytes to the sites of demyelination in the cuprizone model, but that these CCR2-expressing cells seem to have little impact on the demyelination or remyelination processes.

Although infiltration of CCR2-expressing cells did not have a significant impact on the demyelination induced by cuprizone feeding (Fig. 1–3), there is a robust accumulation of microglial cells at the sites of demyelination throughout the cuprizone protocol (Fig. 1, E–F) and their importance has been hinted at for many years (Gudi et al., 2014). CX3CR1 is a crucial receptor controlling microglial physiology and orchestrating the crosstalk between microglia and neurons that constitutively express its sole known ligand, fractalkine (CX3CL1) (Imai et al., 1997; Bellavance et al., 2014; Paolicelli et al., 2014). Remarkably, CX3CR1-deficient mice seemed almost immune to cuprizone as histochemical staining of myelin in these mice revealed a very mild level of demyelination in the CC when compared with WT littermates at 3 and 5 wk (Fig. 4, A and B). However, both genotypes exhibited similar behavioral defects as measured by the time spent in the open arm of an elevated plus maze (Fig. 4 C). Food intake and weight changes (Fig. S1, A and B) were monitored throughout the protocols and no difference could be found between WT and CX3CR1^−/− mice, showing that the difference between genotypes was not attributable to differential metabolic responses to a cuprizone-supplemented diet. Cuprizone treatment leads to a specific depletion of ODs in the CNS (Bénardais et al., 2013). Stereologic assessment of the number of mature ODs in the CC showed a similar depletion of ODs (Fig. 4 D). Furthermore, electron microscopy imaging similarly showed apoptotic ODs in CX3CR1^−/− mice (Fig. 4 E). Thus, preservation of myelin in CX3CR1 knockout mice does not result from increased survival of ODs in these mice, as these cells die by apoptosis at a similar extent to what’s observed in WT mice exposed to cuprizone.

In the CNS, CX3CR1 is expressed exclusively by microglia and perivascular macrophages (Cardona et al., 2006). Because infiltrating CCR2 cells had limited effects on cuprizone-induced demyelination (Fig. 1–3) and CX3CR1 knockout did not impact metabolic processes or OD death (Fig. 4, D and E), we hypothesized a deficient microglial activity to be responsible for the protective effect on myelin content observed against cuprizone in CX3CR1-deficient mice. Although the CC of CX3CR1^−/− mice showed lower Iba1 immunoreactivity than WT mice (Fig. 5 A), a massive infiltration of the region by microglia is still observed when compared with littermates fed with a normal chow (Fig. 5 A). To characterize the inflammatory response in each genotype, we determined the expression level in the CC of TLR2 and C1q-α, markers of NF-κB–induced or complement-activated inflammatory processes, respectively. As expected, a robust and significant induction of both genes was noted in WT mice comparable to that observed in CX3CR1^−/− mice (Fig. 5, B and C), showing no significant difference in the inflammatory reaction between genotypes. However, the expression of the gene encoding for TREM2, a protein involved in phagocytosis (Takahashi et al., 2005), was much lower in CX3CR1^−/− than in WT mice (Fig. 6 A). As a protein implicated in antigen presentation, the induction of CD11c in microglia is also indicative of the phagocytic activity of microglia. Flow cytometry analysis of the expression of CD11c on live CD45⁺CD11b⁺ cells in the CNS (Fig. 6 B) revealed a similar pattern of expression, as its induction was lower in CX3CR1^−/− mice than in WT littermates (Fig. 6 C). Electron microscopy imaging of Iba1-stained microglia (Fig. 6, D and E) confirmed these observations as WT microglia showed extensive accumulation of nonstained endosomes (Fig. 6 D, red arrow), myelin products often partially digested and localized within the endosomes (Fig. 6 D, green arrow), and cholesterol crystals inside of their cytoplasm (Fig. 6 D, blue arrow). However, microglia in CX3CR1^−/− mice were almost devoid of endosomes, internalized myelin, and cholesterol crystals (Fig. 6 E), suggesting a reduced phagocytic activity in the CX3CR1^−/− mice (Tangirala et al., 1994; Klinkner et al., 1995; Lindberg et al., 2014). Collectively, these results show that although inflammatory responses are not affected by the knockout of CX3CR1, there is a clear deficit in microglial phagocytic processes.

A robust recruitment of microglia takes place in demyelinating regions of WT mice, but this phenomenon was attenuated in CX3CR1^−/− mice after both 3 (Fig. 7, A and B) and 5 (Fig. 7, C and D) weeks of cuprizone. It is interesting to note the presence of myelin debris in these microglia, especially in WT animals, suggesting a potent phagocytic response by these cells. Such a response by microglia is clearly weakened in CX3CR1^−/− (Video 1). In the CX3CR1^−/− mice, additional analysis of myelin patterns by electron microscopy revealed abundant myelin aberrations throughout sections of the CC (Fig. 7, G and J–M). Large, round, and empty myelin-enclosed structures reminiscent of spheroids, a hallmark of MS (Ferguson et al., 1997; Trapp et al., 1998) and found in other models of CNS pathology (Stirling et al., 2013), were also observed (Fig. 7 M). Whereas some aberrations were present in WT mice, spheroid-like structures were not observed in these mice. As such, it appears that disrupting myelin clearance by microglia leads to myelin disorganization and that deleting CX3CR1 is not a protective mechanism, as first suggested by the histochemical staining (Fig. 4 A). These results prompted the analysis of remyelinating processes in the case of inefficient clearance of aberrant myelin patterns, sacrificing mice 1 and 2 wk after cuprizone was removed from 6 wk diet. OPCs are normally recruited to the demyelinated CC from the beginning of week 4 to the end of week 5 (Blakemore and Irvine, 2008). As a marker of the earliest OD progenitors, PDGFRα expression was assessed by in situ hybridization, which revealed a significantly lowered level of induction of this gene in CX3CR1^−/− mice after 4 wk of cuprizone (Fig. 8, A and B) and by immunohistochemistry after 5 wk of cuprizone (Fig. 8 E). Furthermore, stereologic assessment of Olig2⁺ cells confirmed a deficient recruitment of OPCs in CX3CR1^−/− mice at 5 wk (Fig. 8, C and D). Proliferating OPCs were determined by the presence of Ki67 staining in Olig2⁺ cells in the brain of WT mice, a phenomenon that was marginal in the CC of CX3CR1^−/− mice (Fig. 8 E). This was also associated with a low PDGFRα-ir and IGF-1-ir signal, suggesting a defect in the mechanisms that trigger OPC proliferation in the CC of CX3CR1^−/− (Fig. 8 E). Surprisingly, Black Gold staining showed efficient remyelination of the CC in both WT and CX3CR1^−/− mice, although myelin-positive staining remained lower in WT mice at 1 wk after the cuprizone diet was replaced by normal chow (Fig. 9 A). Electron microscopy analysis of myelin integrity revealed another set of aberrant patterns, including myelin disorganization, separation of myelin layers, hypermyelination, vacuolization of axons, and a persistence of large spheroid-like structures (Fig. 9 B). Although 14.6% of axons displayed some of these patterns (but not spheroids) in WT mice, the frequency of aberrantly myelinated axons reached 67.9% in CX3CR1^−/− mice (P < 0.0001; Fig. 9 C). Collectively, these results show that impaired clearance of myelin by microglia through CX3CR1 depletion leads to deficits in myelin organization creating defective remyelination of axons in the cuprizone model.

Skewing the balance toward remyelination is a key goal for the long-term treatment of patients with demyelinating diseases such as MS. In this study, we used the toxin cuprizone to model a primary demyelinating insult and to study the impact of monocytes and microglia on demyelination and remyelination. We have shown that monocytes from the periphery can enter the CNS at demyelinating sites, but that these infiltrating cells do not significantly impact the processes of demyelination or remyelination. However, knocking out CX3CR1 severely impeded phagocytosis of myelin by microglia, leading to the persistence of myelin debris throughout the white matter of cuprizone-fed mice. This resulted in inefficient axonal remyelination characterized with aberrant myelin patterns, thus demonstrating the critical role of CX3CR1 in promoting microglial clearance of degenerate myelin and its importance for an efficient remyelination of demyelinating sites.

Infiltration of circulating myeloid cells to sites of demyelination has been reported in earlier studies (McMahon et al., 2002; Remington et al., 2007). These studies used irradiation-conditioned chimeric mice or flow cytometry to dissociate between macrophages and microglia. A study using head-protected, irradiation-conditioned, GFP chimeric mice questioned these results (Mildner et al., 2007). The authors of that study showed that peripheral macrophage infiltration does not occur when the head is protected during irradiation and concluded that cell recruitment to the CC during cuprizone-induced demyelination is an artifact resulting from irradiation-related side effects, namely BBB dysfunction and neuroinflammation. The cuprizone model induces a severe demyelination of specific sites, whereas the BBB remains intact as demonstrated by several studies (Bakker and Ludwin, 1987; Kondo et al., 1987; Matsushima and Morell, 2001). In our study, we showed that recruitment occurs when chimeric mice are generated using a chemotherapeutic conditioning, a procedure we have demonstrated to maintain BBB integrity and to be nonpermissive for the entry of cells into the CNS (Lampron et al., 2012; Michaud et al., 2013; Bellavance et al., 2014), in contrast to conditioning with whole-body irradiation (Priller et al., 2001; Simard and Rivest, 2004). Combining cuprizone and our chemotherapy regimen did not have a synergistic adverse effect on BBB integrity, as we could not detect any extravasation of large circulating proteins, such as albumin or IgG, into the CNS (unpublished data). Interestingly, infiltrated cells were short-lived cells as no GFP⁺ cells were observed in the CC of mice 3 wk after cuprizone removal. The transient nature of these cells suggests that they originate from monocytes rather than early myeloid progenitors introduced artificially in the bloodstream because of the BM transplantation procedure, making them actual monocyte-derived macrophages (MDMs). MDMs were shown to engraft the CNS transiently (Ajami et al., 2011), unlike uncommitted BM progenitors which are thought to engraft the brain sustainably (Davoust et al., 2008; Capotondo et al., 2012). The decline of infiltrating cells is probably caused by apoptotic processes, as is the case after experimental stroke (Li et al., 2013).

We also found that this recruitment of myeloid cells strongly depends on CCR2 signaling even though other chemokines are highly expressed during demyelination (unpublished data). Remington et al. (2007) showed that CCR2 deficiency results in lower levels of macrophages in the CC without affecting the microglial response to demyelination. Using CCR2^−/− chimeric mice, we showed that MDMs are not actively implicated in demyelination, supporting previous observations (Mildner et al., 2007). A recent study using the lysolecithin model of demyelination suggested that monocytes are implicated in the process of remyelination as exposing old mice to a youthful hematopoietic system restored remyelination to a level similar to that observed in younger animals, and the effect was dependent on CCR2 expression (Ruckh et al., 2012). Our data suggest that monocytes do not influence remyelination after cuprizone treatment as no differences were observed in mice between 1 and 2 wk after withdrawal of cuprizone from the diet. However, peripheral myeloid cells might be more involved in de- and remyelination during lysolecithin-induced demyelination compared with cuprizone because of BBB alterations resulting from the stereotactic injection of lysolecithin, permissive for the infiltration of peripheral cells in untreated controls. Whereas the role played by these cells in the cuprizone model will need to be determined, our study is, to our knowledge, the first to demonstrate controlled infiltration of MDM into the CNS in a mouse model devoid of massive neuronal death or BBB disruptions. Additional studies will be needed to firmly establish the exact role played by infiltrating macrophages in de- and remyelination processes in the brain of mice treated with cuprizone.

Microglia are actively implicated in myelin removal during demyelination (Neumann et al., 2009). The CX3CL1–CX3CR1 axis is crucial in the maintenance of microglia under a resting state because of the constant crosstalk with neurons (Imai et al., 1997). Its activity modulates microglial behavior both under normal and pathological conditions (Bellavance et al., 2014; Paolicelli et al., 2014). At the time points where WT mice presented severe demyelination (after 4, 5, and 6 wk of cuprizone), CX3CR1^−/− mice showed very weak demyelination in histochemical stainings. Our data suggest that the resistance of CX3CR1^−/− mice to demyelination does not result from a resistance to cuprizone treatment, but rather to a defect in myelin removal by CX3CR1^−/− mice. These mice displayed both a defect in microglial migration to the CC and a defect in phagocytosis of myelin debris, whereas inflammatory responses were similar. Electron microscopy as well as myelin and Iba-1 staining revealed an almost complete absence of phagocytic inclusions in CX3CR1^−/− microglia, unlike WT microglia that were gorged with such inclusions. This suggested that poor phagocytosis of microglial cells is the primary defect responsible for impaired clearance of myelin debris in CX3CR1^−/− mice.

Interestingly, we observed no modulation of Cx3cl1 mRNA expression during the cuprizone treatment and no difference in its expression between WT and CX3CR1^−/− mice (unpublished data). A link between CX3CR1 signaling and phagocytosis of debris by microglial cells has been reported in mouse models of Alzheimer’s disease (AD) and excitotoxicity. In AD, it was reported that CX3CR1 deficiency enhances phagocytosis of the toxic peptide Aβ (Lee et al., 2010). As we observed during demyelination, excitotoxicity does not influence the expression of the gene encoding for CX3CL1, but promotes its cleavage from neuronal membrane (Chapman et al., 2000), resulting in a soluble isoform that promotes clearance of damaged neurons by microglia (Noda et al., 2011). The discrepancy between the effect of CX3CR1 depletion on the phagocytic capacity of microglial cells in AD and during excitotoxicity might result from the different types of phagocytosis involved (Neumann et al., 2009). Apoptotic neurons are phagocytized without inflammation via receptors like TREM-2 (Takahashi et al., 2005), which is also implicated in myelin removal during EAE (Piccio et al., 2007; Takahashi et al., 2007). Interestingly, CX3CR1^−/− mice show a lower level of Trem2 mRNA in the CC during period of severe demyelination. However, phagocytosis of Aβ seems to depend on inflammation mediated by innate immune receptors like TLRs (Liu et al., 2005). Therefore; it is possible that CX3CR1 promotes phagocytosis of apoptotic neurons and myelin debris, but inhibits phagocytosis of Aβ. This line of thought echoes recently published results by Yamasaki et al. (Yamasaki et al., 2014) demonstrating in an EAE model that microglia actively cleared myelin debris while MDMs had other functions. However, we have shown before that MDMs were more efficient in the clearance of Aβ than resident microglia (Simard et al., 2006). It becomes tempting to suggest that microglia have evolved to become experts in dealing with products normally found in the CNS such as myelin but lack in efficiency when it comes to abnormal oligomeric aggregates like Aβ.

It will be interesting to evaluate the state of activation of CX3CR1-deficient microglial cells compared with WT as the microglial phenotype seems crucial for the promotion of OPC recruitment and tissue regeneration (Olah et al., 2012). It has been shown that microglia/macrophages exhibit a more M1-like phenotype during OPC recruitment and that there is a switch toward a M2 phenotype to allow OPC maturation (Boven et al., 2006; Miron et al., 2013). Kotter et al. showed in a series of elegant experiments that macrophages are essential for myelin removal during demyelination (Kotter et al., 2001, 2005) and that myelin debris impair CNS remyelination by inhibiting OPC differentiation (Kotter et al., 2006). Accordingly, we expected to observe a delayed myelin clearance in CX3CR1^−/− mice and a lower level of remyelination after removal of cuprizone from the diet when compared with WT mice, as Skripuletz et al. observed in mice devoid of astrocytes (Skripuletz et al., 2013). Surprisingly, histochemical staining revealed that, in CX3CR1^−/− mice, remyelination occurred at a similar extent to WT mice despite lower levels of OPC in these mice after 5 wk of cuprizone. It has previously been shown that OPC density is not a rate-limiting factor for efficient remyelination (Woodruff et al., 2004). It rather seems that remyelination occurs, but in an aberrant way. Indeed, electron microscopy observations demonstrated that although CX3CR1^−/− mice had similar numbers of myelinated axons as WT mice, a higher proportion of axons exhibited myelin abnormalities such as hypermyelination, sheath breakdown, and vacuolated formations. Studying the long-term consequences of this aberrant remyelination on axonal damages and neurological deficits may be informative.

Immunomodulators have been very efficient in the control of acute demyelinating events in relapsing–remitting courses of MS. However, these molecules have largely been ineffective in alleviating chronic demyelination in progressive forms of the disease, suggesting a primary neurodegenerative course (Stys et al., 2012). As autoimmune defects do not seem to play a significant role in progressive forms of MS, a balance between limiting demyelination and boosting remyelination of affected sites must be reached for long-term therapeutic support of these patients (Fox et al., 2012). However, the results reported in this study propose important concepts to take into consideration: there can be no efficient remyelination if microglia are unable to clear degenerate myelin from affected axons. As such, an optimal treatment strategy could consist in boosting microglial phagocytic processes whereas limiting inflammatory responses, combined with agents targeting OD physiology. Monocytes and microglia are easily targetable, even with compounds that do not cross the BBB, making them prime candidates for novel therapeutic options in progressive forms of demyelinating diseases.

6–8-wk-old male C57BL/6j mice from The Jackson Laboratory were used in this study. GFP mice were crossed with CCR2^−/− mice to generate the GFP-CCR2^−/− mice. GFP and GFP-CCR2^−/− mice have all C57BL/6j background and were generated from our colonies. The backcross for GFP, WT, and GFP-CCR2^−/− were F8, F5, and F5, respectively. All animals were acclimated to standard laboratory conditions with ad libitum access to mouse chow and water. Mice were housed up to three per cage. All protocols were performed in accordance to the Canadian Council on Animal Care guidelines, as administered by the Laval University Animal Welfare Committee.

0.2% wt/wt cuprizone (Sigma-Aldrich) was mixed with regular ground chow and fed to experimental animals for 2, 3, 4, 5, and 6 wk. The feed was changed every 2 d and food intake was monitored throughout the protocols. Control animals were fed regular ground chow and manipulated as often as cuprizone-fed mice. For remyelination protocols, cuprizone-supplemented chow was replaced by regular chow after 6 wk and the mice were killed 1 and 2 wk later. Mice were 6–8 wk of age at the beginning of the diet. Each experiment was repeated twice.

Experimental animals received a total of 80 mg/kg of Busulfan administered i.p. every 12 h for 4 d, followed by 2 d of single i.p. injections of 100 mg/kg cyclophosphamide. After a 24-h rest, 1.5–2 × 10⁷ BM cells isolated from the tibia and femur of donor mice were injected into the tail vein of target animals. GFP and GFP-CCR2^−/− mice were used as donors. For details on this procedure, please refer to previous studies (Lampron et al., 2012).

Blood samples were collected from the submandibular vein and kept from light on a rotator for <1 h. 65 µl of each blood samples was diluted with 35 µl of DPBS without Ca²⁺ or Mg²⁺ and incubated 15 min on ice with purified rat anti–mouse CD16/CD32 antibody (Mouse BD Fc Block; BD). Cells were then labeled at 4°C during 40 min with the following rat anti–mouse antibodies: V500-CD45 (clone 30-F11; BD), Alexa Fluor 700-Cd11b (clone M1/70; eBioscience), and APC-CD115 (clone AFS98; eBioscience). Next, red blood cells were lysed with BD Pharm Lyse during 20 min at room temperature, and the remaining leukocytes were washed and resuspended with DPBS w/o Ca²⁺ and Mg²⁺. FACS and data acquisition were performed using a SORP LSR II and FACSDiva software (both from BD), respectively. Results were analyzed with the FlowJo software (Tree Star). Monocytes were identified as CD45⁺CD11b⁺ CD115⁺ cells.

Mice were deeply anesthetized via an i.p. injection of a mixture of ketamine hydrochloride and xylazine and then perfused intracardially with ice-cold DPBS. Brains were extracted and immediately homogenized in DMEM supplemented with 10 mM Hepes and 2% FBS and filtered through a 70-µm filter. The cell suspension was then centrifugated at 300 g for 10 min at room temperature. After centrifugation, the supernatant was aspirated and cells were gently resuspended in 1 ml of 37% Percoll. The cell suspension was then underlaid with 70% Percoll and centrifugated at 600 g for 40 min with minimal acceleration and deceleration. The cell ring at the interphase was then collected and mixed thoroughly with DPBS and 2% FBS. The solution was then centrifuged at 300 g for 10 min and washed with DPBS + 2% FBS twice. Cells were first incubated on ice for 15 min with purified rat anti–mouse CD16/CD32 (Mouse BD Fc Block; BD). The mix was then incubated on ice with CD45-PE, CD11b-APC, and CD11c-PE (All from BD) along with the blue Live/Dead marker of cellular death (Molecular Probes) for 35 min and then washed with DPBS + 2% FBS. The cells were analyzed using a two-laser, six-color FACS Canto II flow cytometer and data acquisition was done with BD FACSDiva software (Version 6.1.2; BD). Cells were then acquired according to the different fluorescent antibodies. Results were analyzed using FlowJo software (Tree Star).

To collect the brain tissues, mice were deeply anesthetized via an i.p. injection of a mixture of ketamine hydrochloride and xylazine, and then rapidly perfused transcardially with 0.9% saline, followed by 4% paraformaldehyde in sodium phosphate buffer (pH 7.4 at 4°C). Brains were rapidly removed from the skulls, postfixed overnight, and then placed in a solution containing 10% sucrose diluted in 4% paraformaldehyde (pH 7.4) for at least 48 h at 4°C. The frozen brains were mounted on a microtome (Reichert-Jung; Cambridge Instruments Company), frozen with dry ice, and cut into 25-µm coronal sections from the olfactory bulb to the end of the medulla. The slices were collected in a cold cryoprotectant solution (0.05 M sodium phosphate buffer, pH 7.3, 30% ethylene glycol, 20% glycerol) and stored at −20°C. Free-floating sections were incubated for 30 min in KPBS containing 4% goat serum, 1% BSA, and 0.4% Triton X-100. Using the same blocking solution, the sections were then incubated overnight in primary Ab at 4°C. The sections were then rinsed four times for 5 min in KPBS, followed by a 90 min incubation in fluorochrome- or biotin-conjugated goat secondary anitbody at room temperature. For multiple stainings, the previous wash and incubation steps were repeated with the other antibodies. For nonfluorescent staining, biotin-conjugated secondary Ab were detected using the peroxidase-based Vectastain ABC kit (Vector Labs) following the manufacturer’s instructions. Sections were then rinsed four times for 5 min in KPBS, revealed with 3,3′-Diaminobenzidine tetrahydrochloride (Sigma-Aldrich) and mounted onto SuperFrost slides (Thermo Fisher Scientific), stained with DAPI if needed (0.0002% dilution for 10 min; Molecular Probes), and coverslipped with antifade medium composed of 96 mM Tris-Hcl, pH 8.0, 24% glycerol, 9.6% polyvinylalcohol, and 2.5% diazabicyclooctane (Sigma-Aldrich) or DPX mountant (Electron Microscopy Science). Confocal laser scanning microscopy was performed with a BX-61 microscope equipped with the Fluoview SV500 imaging software 4.3 (Olympus America Inc.). Confocal images were acquired by sequential scanning using a two-frame Kalman filter and a z-separation of 1 µm. Image mosaics were acquired using Micro-Manager software (version 1.3.36 β) with an Olympus IX81 microscope, equipped with a 4X lens, a GFP filter, a Prior ProScan motorized stage and a Qimaging Retiga EXi digital camera. Acquired images were thereafter stitched together using XuvStitch x64 software (version 1.8.0; Emmenlauer et al., 2009) and exported as hdf5 files, then converted to tiff with ImageJ x64 software (version 1.46m; Schneider et al., 2012). Stereological analysis was performed as previously described (Simard et al., 2006). Contours of regions of interest were traced as virtual overlay on the streamed images. Individual cells were counted and reported as a density of events in relation to the surface of the area containing them.

Free floating tissue sections were incubated 30 min in 0.5 mM 1,1′-Dioctadecyl-3,3,3′,3′-tetramethylindocarbocyanine perchlorate or DiI (Sigma-Aldrich) in the blocking solution, which was followed by 3 washing steps in KPBS (see previous section).

Brain sections collected as described above were washed three times for 10 min in cold KPBS. 3–4 slices of interest were then put into a 1.5 mL sample tube (Eppendorf) containing 200 µL of Black Gold (EMD Millipore) diluted 0.3% into 0.9% NaCl. These were incubated exactly 10 min at 62°C, transferred to KPBS, mounted onto Superfrost slides, and coverslipped, as described above. 8-bit grayscale TIFF images of the regions of interest were taken in a single sitting for whole protocols with a Qimaging camera, with the same gain/exposure settings for every image. To quantify myelination, these images were imported into ImageJ and myelination of a given area was measured as the surface proportion of staining intensity above a determined threshold. This threshold was set to detect at least 95% of myelination in untreated WT controls and average 30% in WT mice treated for 3, 5, 6, and 7 wk.

Elevated plus maze was used to quantify behavioral deficits in cuprizone-treated mice as previously described (Walf and Frye, 2007). The elevated plus-maze was made of beige Plexiglas illuminated under constant, homogenous lighting (350 lux). The apparatus consisted of four arms (30 × 5 cm) elevated 40 cm above floor level. Two of the arms contained 15-cm-high walls (enclosed arms) and the other two contained none (open arms). Each mouse was placed in the middle section facing an open arm and left to explore the maze for a single 5-min session with the experimenter out of view. After each trial, the floor was wiped clean with a damp cloth and dried. The following scores were taken: open arm entries and duration (inversely related to anxiety) and distance traveled (reflecting general ambulation), measured by a video camera and analyzed with a video tracking system (ANY-maze; Stoelting Co.), which considers an animal inside a zone whenever its center point is within it.

In situ hybridization was performed on every twelfth section of the brain, starting from the end of the olfactory bulb to the end of the cortex, using 35S-labeled cRNA probes as described previously (Laflamme and Rivest, 2001).

Neuronal death was labeled with the Fluoro-Jade B (FJB) method. In brief, every sixth section of the whole rostrocaudal extent of each brain was mounted onto Superfrost slides (Thermo Fisher Scientific), dried under vacuum for 2 h, dehydrated through graded concentrations of alcohol (50, 70, and 100% for 1 min each), and rehydrated through graded concentrations of alcohol (100, 70, and 50% for 1 min each) and 1 min in distillated water. They were then dipped into and shaken in potassium permanganate (0.06%) for 10 min, rinsed for 1 min in distillated water, and dipped into and shaken in a solution containing Fluoro–Jade B (Fluoro–Jade B 0.0004% [Histochem] + acetic acid 0.1% [Sigma-Aldrich] + DAPI 0.0002% [Molecular Probes]) for 20 min. The slides were thereafter rinsed three times in distillated water (1 min each), dried, dipped in xylene three times (2 min each), and coverslipped with DPX.

Mice were deeply anesthetized with sodium pentobarbital (80 mg/kg, i.p.) and perfused through the aortic arch with 3.5% acrolein (in 100 mM phosphate buffer [PB], pH 7.4) followed by 4% paraformaldehyde (PFA; in 100 mM PB, pH 7.4). Longitudinal sections of the brain (50 µm thick) were cut with a vibratome in ice-cooled PBS (0.9% NaCl in 50 mM phosphate buffer, pH 7.4) and immunostained with a rabbit anti-ionized calcium binding adaptor molecule 1 (IBA1) antibody (1:1,000 in blocking solution; Wako Pure Chemical Industries) as previously described (Tremblay et al., 2010). In brief, labeling was revealed with 3,3′-Diaminobenzidine tetrahydrochloride (DAB; 0.05 mg/ml) and hydrogen peroxide (0.03% in buffer solution; DAB Peroxidase Substrate kit; Vector Laboratories). Immunostained sections were postfixed flat in 1% osmium tetroxide, dehydrated in ascending concentrations of ethanol, impregnated in Durcupan (Electron Microscopy Sciences) overnight at room temperature, mounted between ACLAR-embedding films (Electron Microscopy Sciences), and cured at 55°C for 48 h. Areas of the corpus callosum were excised from the embedded sections, reembedded at the tip of resin blocks, cut with an ultra-microtome (Leica Ultracut S) at 65–80 nm thickness, collected on bare square-mesh grids, and examined with an FEI Tecnai G2 Spirit BioTwin electron microscope.

Data are presented as mean ± SEM. Statistical tests performed include two-tailed unpaired Student’s t test using Welch’s correction of unequal variances and 1-way ANOVA followed by Tukey or Bonferroni post-hoc test for multiple comparisons. All analyses were carried with the Prism software (version 6; GraphPad) and the α level was set at 0.05.

Fig. S1 shows food intake and body weight throughout a cuprizone protocol and Video 1 shows sequential images from deep corpus callosum by confocal microscopy.

The authors would like to thank Dr. Denis Soulet for his technical help in preparing the stitched image mosaic.

This work was supported by grants from the Canadian Institutes for Health Research and the Multiple Sclerosis Scientific Research Foundation of Canada.

The authors have no conflicting financial interests.