Myelin phagocytosis by astrocytes after myelin damage promotes lesion pathology

Tissue samples

Human CNS tissue was obtained at autopsy according to Institutional Review Board-approved protocols. CNS tissue was obtained from nine subjects with multiple sclerosis, one subject with progressive multifocal leukoencephalopathy, one subject with metachromatic leukodystrophy and five subjects with subacute ischaemic infarct (Supplementary Table 1). Post-mortem intervals ranged between 3 and 32 h. Brain tissue was fixed with 10% formalin for 2 to 4 weeks and embedded in paraffin.

The multiple sclerosis patient cohort consisted of one acute case who was biopsied (Patient MS1), five cases of relapsing/remitting multiple sclerosis, two cases of secondary progressive multiple sclerosis and one case of asymptomatic multiple sclerosis that was incidentally discovered at autopsy (Patient MS9). Patient MS1 was a 62-year-old female with one month disease duration. Patient MS2 was a 50-year-old female with 13 years disease duration who died of cardiopulmonary arrest. Patient MS3 was a 31-year-old female with 8 years disease duration who died of respiratory failure. Patient MS4 was a 32-year-old male with 10 years disease duration who died of pneumonia. Patient MS5 was a 50-year-old female with 10 years disease duration who died of respiratory failure. Patient MS6 was a 39-year-old male with >15 years disease duration who died of haemorrhagic bronchopneumonia. Patient MS7 was a 65-year-old male with 30 years disease duration who died of chronic cystitis. Patient MS8 was a 58-year-old male with 27 years disease duration with unknown cause of death; and Patient MS9 was a 68-year-old male with unknown disease duration who died of cardiac arrhythmia. The expanded disability status scale (EDSS) scores of the patients at the time of their deaths are not known. We examined a total of 25 white matter lesions of which 20 lesions were early or chronic active and five lesions were chronic silent.

The patient with progressive multifocal leukoencephalopathy (Patient PML1) was a 55-year-old male with multiple sclerosis of 6 years disease duration who subsequently developed and died of progressive multifocal leukoencephalopathy after receiving treatment with natalizumab for 4 years. The patient with metachromatic leukodystrophy (Patient MLD1) was a 3-year-old male with metachromatic leukodystrophy who died in status epilepticus.

Infarct patient 1 (Patient CVA1) was 68-year-old male with a 20-day-old infarct of the entire right middle cerebral artery territory who died of brainstem compression. Infarct patient 2 (Patient CVA2) was a 79-year-old male with a 2-month-old infarct in the left middle cerebral artery territory involving the superior and middle temporal gyri, who died of sepsis. Infarct patient 3 (Patient CVA3) was 69-year-old male with multiple 2-week-old white matter watershed infarcts, who died of acute myocardial infarction. Infarct patient 4 (Patient CVA4) was 67-year-old female with an 18-day-old infarct of the entire left middle cerebral artery territory who died of metastatic pancreatic adenocarcinoma. Infarct patient 5 (Patient CVA5) was a 48-year-old male with a subacute infarct of the right frontal lobe (middle cerebral artery territory) who died of sepsis.

Immunohistochemistry and histochemistry

Formalin-fixed, paraffin-embedded (FFPE) sections were cut at 5 -μm thickness. After quenching with 0.03% hydrogen peroxide and blocking with normal serum, sections were incubated with primary antibodies overnight (Supplementary Table 2), processed with the appropriate biotinylated secondary antibody and avidin/biotin staining kit with diaminobenzidene and VIP as chromogens (Vector ABC Elite Kit, DAB Kit and VIP Kit, Vector Laboratories), and counterstained with haematoxylin (Pitt et al., 2003). For fluorescence, sections were incubated with fluorescent-labelled secondary antibodies and subsequently treated with 0.7% Sudan Black in ethanol and CuSO₄ to quench auto-fluorescence. Sections were counterstained with DAPI and mounted with VectaShield mounting medium (VectaShield Kit, Vector Laboratories). Adequate controls using isotype control antibodies were performed with each primary antibody.

For staining of myelin proteins with Luxol Fast Blue (LFB), FFPE sections were incubated overnight with LFB at 56°C and subsequently washed in 95% ethanol and dH₂O. The colour was discriminated in Li₂CO₃ solution and differentiated in 70% ethanol. In formalin-fixed, OCT-embedded sections, i.e. in tissue in which lipids are not extracted through deparaffinization (Patient MS9), myelin lipids were detected with Sudan Red 5B (Oil Red O), which stains neutral lipids. Sections were incubated with Oil Red O for 10 min, washed with tap water, counterstained with haematoxylin and cover-slipped with aqueous mounting medium without dehydration (Mehlem et al., 2013). Both stains were combined with immunohistochemical labelling against GFAP and CD68.

Brightfield images were acquired with a Leica DM5000 B microscope using a Leica colour camera DFC310 Fx and the Leica Application Suite (version 4.2.0) imaging software. Fluorescent images were taken with an UltraVIEW VoX (Perkin Elmer) spinning disc confocal Nikon Ti-E Eclipse microscope. Image acquisition, visualization and quantification were performed using the Volocity 6.3 software (Improvision). Images were processed with the ImageJ software (Schneider et al., 2012).

Nuclear localization of NF-κB and chemokine expression levels of CCL3, CCL5, CCL20 and CXCL10 as well as inducible nitric oxide synthase (iNOS) were quantified by densitometric analysis of fluorescent immunoreactivity of various cell types. Acquisition and analysis were performed in a double-blinded manner. Statistical analysis was based on expression levels of at least 20 cells per cell type per case in four multiple sclerosis cases, two cerebral infarct cases and one case of progressive multifocal leukoencephalopathy.

Myelin purification

Myelin was purified as described by Norton and Poduslo (1973). In brief, adult male Sprague-Dawley rat spinal cords were homogenized in 0.32 M sucrose using a Dounce homogenizer. Myelin was recovered by three successive sucrose gradient centrifugation steps and washed extensively in Tris·Cl buffer, pH 7.4. The pellet was resuspended in 0.32 M sucrose, layered over 0.85 M sucrose and subjected to centrifugation at 75 000 g for 30 min. Myelin was recovered and suspended in phosphate-buffered saline (PBS) pH 7.4. Protein concentration of the preparation was analysed with a BCA-Kit (Pierce), according to the manufacturer’s instructions.

Cell culture

Cultures of glial cells were prepared according to the method of McCarthy and de Vellis (1980). Briefly, cerebral cortex from postnatal Day 1 Sprague-Dawley rat pups was isolated, minced, and digested for 30 min at 37°C in Hanks' balanced salt solution (HBSS) containing 0.25% trypsin (Life Technologies). Cells from each rat were plated in separate 75 cm² tissue culture flasks coated with 10 μg/ml of poly-l-lysine. Cultures were maintained in Dulbecco’s modified Eagle medium (DMEM) with 10% foetal bovine serum (FBS) and penicillin/streptomycin for 10 days. Flasks were agitated by rotation (180 rpm) for 30 min at 37°C to detach and collect microglia. Finally, astrocytes were collected after Accutase® treatment and were further purified with astrocyte-specific glutamate transporter anti-glutamate aspartate transporter (GLAST) antibodies linked to magnetic microbeads (Miltenyi Biotec) according to the manufacturer’s instructions. Astrocytes were plated in tissue culture dishes coated with 10 μg/ml of poly-l-lysine and cultured in DMEM with 10% FBS and penicillin/streptomycin. To assess purity of the primary cultures, immunostaining was performed using cell-type-specific antibodies targeting isolectin B4 (microglia) and GFAP (astrocytes).

Myelin uptake assays

Purified myelin was labelled with Oregon Green® 488 Dye (Life Technologies) according to the manufacturer's instructions. Myelin was incubated with cells for various durations ranging from 30 min to 72 h at 37°C. For quantification of myelin uptake by astrocytes, we washed astrocytes three times with PBS to release surface-associated myelin. We omitted pronase treatment of astrocytes suggested in previous protocols (Gaultier et al., 2009) as the washing step sufficiently removed myelin debris from astroglial cell surfaces, as demonstrated with confocal z-stack images of astrocytes at various time points of myelin exposure. Quantitative assessment of myelin uptake was performed with a fluorescence plate reader (Infinite M1000 Pro, Tecan). Myelin uptake was assessed qualitatively up to 24 h after incubation with fluorescence microscopy using a spinning disc confocal Nikon Ti-E Eclipse microscope. Images were processed using the ImageJ software (Schneider et al., 2012). In some studies, LysoTracker® or CellROX® (Life Technologies) was added 30 min prior to fixation.

Immunocytochemistry

After treatment, astrocyte cultures were washed with PBS, fixed in 4% paraformaldehyde/PBS for 20 min, washed, blocked and permeabilized for 1 h in 10% goat serum, 1% bovine serum albumin, and 0.01% Triton™ X-100 at room temperature. Cultures were incubated in blocking buffer with primary antibody (2 µg/ml) overnight at 4°C and subsequently with fluorescent-labelled secondary antibody (1 µg/ml, Life Technologies) for 2 h at room temperature. Primary antibodies used to identify astrocytes and cell organelles are listed in Supplementary Table 2. Cell nuclei were visualized with Hoechst® dye 33 342 (1:100 000 in PBS for 5 min). Cultures were washed three times and mounted in ProLong® Gold antifade reagent (Life Technologies).

Flow cytometry, ELISPOT and immunoblotting

Astrocytes were incubated with purified myelin ranging from 15 min to 2 h and dissociated with Accutase® (Life Technologies) for flow cytometry analysis following published procedures (Housley et al., 2011). Briefly, cells were stained for 15 min at room temperature with fluorescent-labelled primary antibodies, washed, and analysed on either a FACSCalibur™ or a LSRII flow cytometer (BD Biosciences). Cytokine release was quantified using an enzyme-linked immunospot (ELISPOT) array (RAT Cytokine Proteome Profiler, R&D Systems). Supernatants from astrocyte cultures exposed previously to purified myelin for 24–48 h were collected and centrifuged (3000 rpm for 10 min) before analysis. All assays were performed according to the manufacturer’s instructions. Immunoblotting was performed by published procedures (Ponath et al., 2007). Briefly, following stimulation with purified myelin, cultured astrocytes were lysed for whole protein fractions in the presence of a proteinase inhibitor cocktail. Sodium dodecyl sulphate-polyacrylamide gel electrophoresis (SDS-PAGE) was performed using NuPAGE® 10% Bis–Tris gels (Life Technologies) with MOPS running buffer and subsequent transfer of blots on nitrocellulose membranes. Primary antibodies used for immunodetection are listed in Supplementary Table 2. After incubation with secondary antibodies, proteins were visualized with an enhanced chemiluminescence using an ImageQuant™ LAS 4000 camera (GE Healthcare).

Extraction of total RNA and polymerase chain reaction array

For expression profiling of 84 key genes involved in autoimmune and inflammatory immune responses, we used the Rat Inflammatory Response & Autoimmunity RT² Profiler PCR Array (Qiagen), according to the manufacturer’s instructions. Astrocytes were incubated with 4 ng/ml purified myelin for 4 h before total RNAs were extracted (miRNeasy® Mini Kit; Qiagen). The PCR array was run on a StepOne™ Real-time PCR system (Life Technologies). Data were analysed using web-based software (http://pcrdataanalysis.sabiosciences.com /pcr/arrayanalysis.php), based on the ΔΔC_T method with normalization of the raw data to five housekeeping genes.

Statistical analysis

Results were expressed as standard deviation (SD) of the mean. Data were analysed by one-way ANOVA followed by the Tukey-Kramer multiple comparison test.

Study approval

Human CNS tissue was obtained according to Institutional Review Board-approved protocols.

Yale’s Institutional Animal Care and Use Committee (IACUC) granted approval for the use of animals in this project (#2013-11543). Yale University has an approved Animal Welfare Assurance (#A3230-01) on file with the NIH Office for Protection from Research Risks. The Assurance was approved May 12, 2011.

Astroglia containing myelin debris are present in various demyelinating CNS diseases

To investigate the prevalence of astrocytes with myelin debris in diseases with prominent myelin injury, we examined autopsy tissue from patients with multiple sclerosis, progressive multifocal leukoencephalopathy, metachromatic leukodystrophy and subacute ischaemic infarct. The majority of astrocytes in acute multiple sclerosis lesions (biopsy, Fig. 1A–D) and at the edges of ongoing demyelination in chronic active lesions (Fig. 1E–N), had a hypertrophic morphology characterized by a ballooned appearance with short, thickened processes and enlarged, occasionally multiple nuclei. These astrocytes contained LFB-positive inclusions and Oil Red O-positive lipid droplets, suggesting presence of myelin debris (Fig. 1C, G and H). Astrocytes in gliotic, inactive lesion centres showed small nuclei and small amounts of delicate fibrillary cytoplasm (Fig. 1B, C, E, F, I and L) and were uniformly LFB/Oil Red O-negative (Fig. 1F and L). Confocal microscopy confirmed that hypertrophic astrocytes contained intracellular MBP- and PLP1-positive puncta (Fig. 1D, J and K), whereas fibrous astrocytes in normal-appearing white matter or gliotic lesions were immunonegative for myelin proteins. In addition, PLP-positive puncta within astrocytes were located within lysosomal ring structures visualized with the lysosomal marker, LAMP1 (Fig. 1D and K). This finding further establishes the intracellular localization of myelin debris and its degradation through the lysosomal pathway. Moreover, myelin debris co-localized with the cell surface scavenger receptor low density lipoprotein receptor-related protein 1 (LRP1; Fig. 1M), indicating that myelin uptake is mediated by LRP1. The number of myelin-containing, hypertrophic astrocytes was approximately one-third of myelin-laden macrophages; however, the amount of myelin debris within astrocytes was only 5% of that in macrophages (Supplementary Fig. 1), suggesting that the bulk of myelin is removed by macrophages. In addition, we found that myelin-containing hypertrophic astrocytes were present to a lesser degree in the perimeter of active lesions, i.e. in normal-appearing white matter, which contains scattered activated microglia but not myelin-laden phagocytes (Supplementary Fig. 2).

We also studied progressive multifocal encephalopathy, a demyelinating disease caused by JC virus (Boster et al., 2013). Acute progressive multifocal leukoencephalopathy lesions and occasionally adjacent brain tissue displayed numerous hypertrophic astrocytes with enlarged, bizarrely shaped nuclei, a characteristic feature of progressive multifocal leukoencephalopathy (Fig. 2A–C). The majority of hypertrophic astrocytes within the demyelinating lesion but not in adjacent white matter contained small LFB-positive inclusions (Fig. 2D and E). Confocal microscopy on fluorescent-labelled sections confirmed the presence of MBP in lesional hypertrophic astrocytes but not in astrocytes in normal-appearing white matter (Fig. 2F and G). In progressive multifocal leukoencephalopathy, we observed a population of myelin-negative hypertrophic astrocytes that was not present in multiple sclerosis or cerebral infarcts. This is unsurprising as astroglial hypertrophy is a hallmark of progressive multifocal leukoencephalopathy presumably caused by JC virus proliferating within astrocytes.

Furthermore, we examined CNS tissue from a case of metachromatic leukodystrophy. Metachromatic leukodystrophy is a U-fibre sparing leukodystrophy in which sulphatide and lysosulphatide accumulate due to arylsulphatase A deficiency in lysosomes (Gieselmann and Krageloh-Mann, 2010). Myelin loss along with axonal destruction, the pathological hallmark of metachromatic leukodystrophy is thought to be caused by critical sulphatide levels in oligodendrocytes and Schwann cells. Immunolabelling with MBP and CD68 showed a gradient of demyelination from near-intact U-fibres to myelin-depleted white matter with diffuse macrophage infiltration (Fig. 2H–K). As with multiple sclerosis and progressive multifocal leukoencephalopathy, hypertrophic astrocytes were present in zones of ongoing myelin destruction and contained PLP-positive myelin debris (Fig. 2M and P). In contrast, myelin-depleted white matter in metachromatic leukodystrophy contained a dense gliotic mesh with infrequent and small astrocytic cell bodies from which myelin debris was absent (Fig. 2N and Q). Reactive astrocytes in subcortical areas with preserved myelin were also PLP-negative (Fig. 2L and O).

Finally, we examined five cases of subacute cerebral infarcts ranging from 2 to 9 weeks in age. In all cases, the necrotic core of the infarct was surrounded by a myelin-depleted, reactive zone with hypertrophic (gemistocytic) astrocytes (Fig. 3A–E and G). In areas of acute myelin breakdown within the reactive zone, hypertrophic astrocytes but not activated astrocytes in the bordering myelinated white matter contained PLP1-positive myelin debris (Fig. 3F, G, I and J). In the necrotic core, no astrocytes were present and myelin debris was found in infiltrating macrophages (Fig. 3H and K). As with multiple sclerosis, the ratio of myelin-containing astrocytes to macrophages indicates that the majority of damaged myelin is cleared by macrophages (Supplementary Fig. 1).

In summary, in all CNS diseases that we examined, hypertrophic astrocytes were the predominant astroglial phenotype in areas of acute myelin breakdown of which the vast majority contained myelin debris. Reactive astrocytes in normal appearing white matter and fibrotic astrocytes within gliotic scars were uniformly myelin-negative. Moreover, intracellular myelin debris was associated with LRP1 and with LAMP1 suggesting a receptor-based mechanism for astroglial myelin uptake and lysosomal degradation.

Astrocytes take up myelin debris in a concentration-dependent manner

We modelled the functional impact of myelin uptake on astrocytes by exposing cultured primary cortical rat astrocytes to purified myelin fragments. Differentiated astrocytes (in vitro Day 14) were purified with GLAST antibodies linked to magnetic microbeads (>98% purity, Supplementary Fig. 3) and cultured as a confluent monolayer and incubated with various concentrations of fluorescent-labelled myelin purified from rat spinal cord. Within 24 h, all myelin was cleared from the supernatant and was present in the majority of astrocytes (Fig. 4A). Flow cytometry confirmed that >98% of astrocytes contained fluorescent myelin debris as early as 45 min after myelin exposure. Intracellular localization of myelin was visualized by orthogonal views of z-stack images acquired by confocal microscopy (Fig. 4B).

Myelin uptake in isolated astrocyte cultures was concentration-dependent and linear within the first hour at a concentration range of 0.125 to 2 µg/ml (Fig. 4C). Myelin uptake by astrocytes was also observed when co-cultured with microglia, albeit at a lower level, since microglia competed for myelin debris and phagocytosed with higher efficiency. In addition, myelin uptake was independent of astroglial density and maturation stage (in vitro Days 14–40). Astroglial activation with TNFα and IFNγ increased myelin uptake by 25% [one-way ANOVA, F(3,8) = 9.51, P = 0.005; Tukey–Kramer test, P < 0.05], while incubation with lipopolysaccharide (LPS; all 10 ng/ml for 24 h) did not change myelin uptake compared to resting astrocytes.

Myelin is taken up by LRP1-mediated endocytosis and is degraded in lysosomes

To examine uptake mechanisms and intracellular trafficking of myelin debris, we immunolabelled cultured astrocytes for the cell surface receptor LRP1 and with markers of the endosomal lysosomal pathway. Fluorescent-labelled myelin fragments within astrocytes co-localized with LRP1 after 30 min, with the endosomal marker early endosome antigen-1 (EEA1) after 60 min and with the lysosomal markers, Lysotracker Red^TM and CellROX^TM at 120 min (Fig. 4E–H). Time-lapse images demonstrated lysosomal degradation of myelin debris over time indicated by loss of green fluorescence within lysosomes (Fig. 4H). To confirm that myelin uptake by astroglia was a receptor-mediated, endocytotic process, we quantified myelin phagocytosis in astrocyte cultures preincubated with the competitive LRP1 antagonist, receptor associated protein (RAP; 5 µg/ml) and with Dynasore (0.1 µM, both 30 min), an inhibitor of dynamin-dependent endocytosis. Incubation of both RAP and Dynasore resulted in significantly reduced myelin endocytosis by astrocytes. Similarly, preincubation with LPS for 2 h resulted in reduced myelin uptake, consistent with LRP1-depleting effect of LPS from the cell surface (Fig. 4I) (Costales et al., 2013). These results were confirmed by flow cytometry where pretreatment with Dynasore blocked myelin ingestion by >90% (Fig. 4J).

Myelin uptake induces astroglial activation

To elucidate the effect of myelin uptake on astrocyte function, we examined astroglial proliferation and activation markers. Astroglial proliferation following myelin endocytosis was measured through bromodeoxyuridine (BrdU)-incorporation and showed a 3-fold increase over untreated controls, compared to a 5-fold increase with TGF-β (10 ng/ml), a potent inducer of astroglial proliferation (Fig. 5A) (Yu et al., 2014). In addition, GFAP was upregulated in astrocytes after 24 h of myelin exposure and continued to increase for at least 9 days as shown by western blot (Fig. 5B and C). Furthermore, myelin phagocytosis resulted in elevated levels of intracellular reactive oxidative species after 24 h as shown by live cell imaging with a fluorescent probe for oxidative stress (Fig. 5D). Quantification with densitometric analysis of fluorescence intensity showed a 3-fold increase in reactive oxidative species generation [mean_untreated = 1753 ± 416, mean_{myelin-treated} = 5309 ± 1409; one-way ANOVA, F(1,16) = 29.8, P < 0.0001].

We further examined intracellular astroglial signalling pathways by determining activation of NF-κB and mitogen-activated protein (MAP) kinase p38, a downstream target of LRP1. Myelin uptake by astroglia resulted in nuclear translocation of p65 NF-κB, transient phosphorylation of astroglial p65 NF-κB and degradation of IκBα, suggesting induction of NF-κB signalling (Fig. 6A–C). Astroglial nuclear translocation of p65 NF-κB after incubation with myelin debris was also observed in astrocyte-microglia co-cultures (data not shown). Similarly, p38 MAP kinase signalling was increased in astrocytes after myelin uptake, as indicated by p38 MAPK phosphorylation (Fig. 6D and E). Inhibition of myelin endocytosis through pretreatment with the dynamin inhibitor, Dynasore (0.1 µM, 30 min), reduced IκBα degradation and p38 MAPK phosphorylation, suggesting that myelin uptake rather than LRP1 binding induced both signalling pathways (Fig. 6C and D).

Myelin uptake induces astroglial chemokine production

To determine the effect of myelin phagocytosis on possible immune functions of astrocytes, we analysed mRNA expression and secretion of chemokines, as well as recruitment of T cells and microglia. Expression of 84 immune-relevant genes in untreated and myelin-treated astrocytes were quantified with a PCR array (Supplementary Fig. 4). The genes that were expressed 3.5-fold or higher from baseline are listed in Supplementary Fig. 5A, and are visualized in a scatter plot (Fig. 7A). Of the 17 highly expressed genes, 65% (11) encoded for chemokines, highlighted in red in the scatter plot. Other highly upregulated transcripts included Nos2, encoding for inducible nitric oxide synthase (iNOS), and the interleukins Il1a (IL-1α) and Il1b (IL-1β) (Fig. 7A).

To confirm that transcriptional upregulation resulted in increased production of chemokines, we examined supernatant of cultured astrocytes 24 h after myelin treatment using a cytokine profiler ELISPOT array (Fig. 7B). Myelin exposure induced de novo secretion of chemokines (CCL3, CCL5 and CXCL9) and significantly upregulated others (CXCL2, CXCL3, CXCL5, CXCL10 and CCL20). Importantly, interleukins were not detected suggesting that myelin-uptake induced a specific astroglial activation pattern that predominantly supports recruitment of immune cells (Supplementary Fig. 5B).

Finally, to confirm that astroglial chemokine secretion was effective in recruiting immune cells, we analysed migration of polarized T_h1 CD4⁺ T cells and microglia through a transwell-chamber with conditioned medium from astrocytes in the lower chamber. Conditioned medium from astrocytes treated with myelin increased T_h1 CD4⁺ cell and microglial migration by 3-fold and 1.6-fold, respectively (Fig. 7C). These data underscore the ability of myelin-stimulated astrocytes to recruit inflammatory cells. Furthermore, we observed that supernatant from myelin-treated astrocytes led to microglial activation in vitro, as indicated by their morphological change from an elongated and bipolar to a rounded shape with short processes (Supplementary Fig. 6).