Brain region‐specific amyloid plaque‐associated myelin lipid loss, APOE deposition and disruption of the myelin sheath in familial Alzheimer’s disease mice

There is emerging evidence that amyloid beta (Aβ) aggregates forming neuritic plaques lead to impairment of the lipid‐rich myelin sheath and glia. In this study, we examined focal myelin lipid alterations and the disruption of the myelin sheath associated with amyloid plaques in a widely used familial Alzheimer's disease (AD) mouse model; 5xFAD. This AD mouse model has Aβ42 peptide‐rich plaque deposition in the brain parenchyma. Matrix‐assisted laser desorption/ionization imaging mass spectrometry of coronal brain tissue sections revealed focal Aβ plaque‐associated depletion of multiple myelin‐associated lipid species including sulfatides, galactosylceramides, and specific plasmalogen phopshatidylethanolamines in the hippocampus, cortex, and on the edges of corpus callosum. Certain phosphatidylcholines abundant in myelin were also depleted in amyloid plaques on the edges of corpus callosum. Further, lysophosphatidylethanolamines and lysophosphatidylcholines, implicated in neuroinflammation, were found to accumulate in amyloid plaques. Double staining of the consecutive sections with fluoromyelin and amyloid‐specific antibody revealed amyloid plaque‐associated myelin sheath disruption on the edges of the corpus callosum which is specifically correlated with plaque‐associated myelin lipid loss only in this region. Further, apolipoprotein E, which is implicated in depletion of sulfatides in AD brain, is deposited in all the Aβ plaques which suggest apolipoprotein E might mediate sulfatide depletion as a consequence of an immune response to Aβ deposition. This high‐spatial resolution matrix‐assisted laser desorption/ionization imaging mass spectrometry study in combination with (immuno) fluorescence staining of 5xFAD mouse brain provides new understanding of morphological, molecular and immune signatures of Aβ plaque pathology‐associated myelin lipid loss and myelin degeneration in a brain region‐specific manner.


| INTRODUC TI ON
According to the amyloid hypothesis, accumulation of amyloid beta (Aβ) plays a central and causative part in Alzheimer's disease (AD) pathogenesis and the rest of the disease process (Hardy & Higgins, 1992). Aβ aggregates into oligomers that form diffuse and neuritic plaques in the brain parenchyma, which have been suggested to cause dysfunction and loss of neurons and synapses (Lacor et al., 2004;Palop & Mucke, 2010).
However, there is also emerging evidence that this process may lead to impairment of the lipid-rich myelin sheath and the oligodendrocyte degeneration present in the brains of AD patients and transgenic AD mice models (Behrendt et al., 2013;Desai, Guercio, Narrow, & Bowers, 2011;Nasrabady, Rizvi, Goldman, & Brickman, 2018). The presence of intracellular granular lipid deposits in multiple cell types and myelin deficits in the brain parenchyma in AD was initially described by Aloysius Alzheimer in 1911 (Foley, 2010;Möller & Graeber, 1998). Further, the E4 allelic variant of the apolipoprotein (APOE) gene which encodes a protein involved in the sterol and sphingolipid metabolism such as the modulation of sulfatide content, and lipid transport (Bu, 2009;Han, Cheng, Fryer, Fagan, & Holtzman, 2003;Kim, Basak, & Holtzman, 2009) was suggested to be a prominent risk factor for sporadic AD and late-onset familial AD (FAD) (Corder et al., 1993;Saunders et al., 1993;. APOE linked the aberrant lipid biochemistry to AD pathogenesis in postmortem human AD and transgenic AD mice brains (Bandaru et al., 2009;Han, M. Holtzman, W. McKeel, Kelley, & Morris, 2002). Therefore, the role of intracellular lipid deposits, lipid biochemistry and myelin disruption in association with amyloid pathology in AD pathogenesis has been widely scrutinized (Bartzokis, 2004;Collins-Praino et al., 2014;Di Paolo & Kim, 2011;Mitew et al., 2010;Palavicini et al., 2017;Yang et al., 2014). Morphological data from human presenilin-1 familial, sporadic and preclinical AD cases, as well as in aged transgenic mice models, tg2576 (APP Swe 670/671 ) and APP/ PS1 (APP Swe ˟ PS1 M146L ) has revealed a clear association between the fibrillary Aβ pathology and focal disruption of the myelin sheath in the gray matter of the neocortex compared with the age-matched control tissue (Mitew et al., 2010). Moreover, there was a focal loss of oligodendrocytes in sporadic and preclinical AD cases (Mitew et al., 2010) which is in line with the finding that amyloid-β peptides are cytotoxic to oligodendrocytes (Xu et al., 2001). However, molecular aspects of these observations still remain unknown.
Early lipidomics studies using brain tissue extracts revealed that sulfatides and plasmalogens substantially and specifically deplete, while ceramides elevate at the very early stages of AD (Han, Holtzman, & McKeel, 2001a;Han et al., 2002). Further, a body of evidence deduced from shotgun lipidomic analysis of brain tissue extracts suggested that APOE mediates sulfatide depletion in APP transgenic mice models (Cheng, Zhou, Holtzman, & Han, 2010). However, there was no direct correlation between sulfatide depletion and amyloid plaque-associated pathology. While recent data has revealed focal depletion of several sulfatide species in the amyloid plaques in the hippocampus and cerebral cortex of the brain in transgenic AD mice models, tgArcSwe (APP E693G(Arctic)/ KM670/671Nl(Swedish) ) (Kaya, Brinet, Michno, Başkurt, et al., 2017;Kaya, Brinet, Michno, Syvänen, et al., 2017) and tgSwe (APP KM670/671Nl ) (Michno et al., 2018), it still remains unknown whether these observations are associated with myelin disruption and/or other factors such as APOE in association with amyloid plaques in specific brain regions.
In this study, high-spatial resolution MALDI-IMS of coronal brain tissue sections of 12-month-old 5xFAD mice revealed focal amyloid plaque-associated depletion of several myelin-associated lipid species. Double staining of the consecutive sections with fluoromyelin and amyloid-specific antibodies revealed amyloid plaque-associated myelin lipid disruption which was specific to the edges of the corpus callosum in the white matter while there was no disruption of the myelin sheath observed in amyloid plaque-rich areas in several other brain regions in the gray matter. APOE accumulation in amyloid plaques was found to be in line with sulfatide depletion in several brain regions, suggesting that sulfatide depletion might be APOE deposition-associated in all the plaques in the white and gray matter regions in 5xFAD mouse brain. Our data suggests amyloid plaque pathology-associated disruption of the myelin sheath through the loss

K E Y W O R D S
Alzheimer's disease, amyloid plaques, apolipoprotein E (APOE), MALDI Imaging Mass Spectrometry, myelin lipids, sulfatides of myelin lipids which leverages the understanding of gray matter and white matter pathologies in AD.

| Chemicals and reagents
All chemicals for matrix and solvent preparation were pro-analysis grade and obtained from Sigma-Aldrich (Stockholm) unless otherwise specified. Anti-beta amyloid antibody (rabbit polyclonal to beta amyloid), anti-APOE antibodies (rabbit monoclonal to apolipoprotein E) and thioflavin S, DAPI fluorescent stains were purchased from Abcam.
FluoroMyelin™ green fluorescent myelin stain was purchased from ThermoFisher Scientific (Stockholm). TissueTek optimal cutting temperature (OCT) compound was purchased from Sakura Finetek (AJ Alphen aan den Rijn). The deionized H 2 O was obtained from a Milli-Q purification system (Merck Millipore).

| Animals, tissue sampling and sectioning
The 5xFAD transgenic mouse model was purchased from Jackson

| Sample preparation and matrix application for MALDI-IMS
For MALDI-IMS in dual polarity, 1,5-diaminonapthalene matrix was deposited onto the tissue by sublimation-based matrix coating (Thomas, Charbonneau, Fournaise, & Chaurand, 2012). Matrix deposition for lipid analysis was carried out using a vacuum sublimation apparatus (Sigma-Aldrich) as previously described in detail elsewhere (Kaya, Jennische, Lange, & Malmberg, 2018a). 1,5-Diaminonapthalene matrix was deposited over the brain tissue sections with the optimized sublimation parameters: 20 min at a temperature of 130°C under a stable vacuum of ~0.8 mbar according to the previously published reports (Kaya, Michno, et al., 2017).

IMS analysis of tissue sections was performed on a MALDI-TOF/TOF
UltrafleXtreme mass spectrometer equipped with a SmartBeam II Nd:YAG/355 nm laser operating at 1 kHz (Bruker Daltonics). MALDI-IMS data for lipids over a mass range of 300-3000 Da. was collected in reflective ion mode using 5-10 laser shots per pixel in negative polarity and 20-25 laser shots in positive polarity as previously described in detail elsewhere (Kaya, Michno, et al., 2017).  Figure S1).
The localization of Aβ 1-40 and Aβ 1-42 in 5xFAD mouse brain has previously been shown on these samples (Kaya et al., 2019), with MALDI-MS and MALDI-IMS which are able to distinguish between the Aβ 1-40 and Aβ 1-42 in AD brain tissues (Carlred et al., 2016;Stoeckli, Staab, Staufenbiel, Wiederhold, & Signor, 2002). Here we focus solely on the lipid changes associated with the plaques and instead correlate the MALDI-IMS with immunohistochemical staining of the plaques.
Quantitative analysis of MALDI-MSI data is notoriously difficult.
So as not to imply quantitation with the inclusion of numerical statistics the robustness and reproducibility of the results are demonstrated by the inclusion of multiple images from multiple 5xFAD and littermate control mice ( Figures S6, S7).

| Immunohistochemistry, fluorescence staining, and hematoxylin staining
In order to visualize amyloid aggregates, the tissue sections were fixed in methanol, blocked with 2.5% horse serum and incubated with a rabbit polyclonal antibody against β-amyloid (ab2539; Abcam) diluted 1:2000 in phosphate-buffered saline (PBS), overnight. Impress reagent, anti-rabbit Ig (Vector Laboratories) was used as secondary reagent and the reaction was visualized using Liquid DAB + sub-

| Myelin lipids are depleted in amyloid plaques in the hippocampus, cortex and on the edges of corpus callosum
Initially, chromogenic staining of amyloid plaques with anti-beta amyloid antibody and hematoxylin staining was performed on a coronal tissue section to reveal the plaque distributions over the brain regions in 12-month-old 5xFAD mice brains. Amyloid plaques were visualized abundantly in the cornu ammonis (CA), subiculum and deep cortical layers ( Figure S2). For the inspection of the amyloid plaque-associated alterations of myelin lipids, regions containing both white matter and gray matter areas including the cortex, corpus callosum, and CA region in the hippocampus in coronal sections of 12-month-old 5xFAD mice brains were imaged using MALDI-IMS at 10 μm spatial resolution in negative and positive polarities (Figures 1 and 2). Further, consecutive tissue sections were stained with an anti-Aβ antibody and hematoxylin to distinguish amyloid plaques and brain regions, respectively in the gray matter. This suggests amyloid plaque-associated disruption of the lipid reach myelin sheath which links amyloid pathology to white matter lipid degeneration. Results from three more 12-monthold 5xFAD mice brains are presented in Figure S8.

| APOE is present in amyloid plaques in the gray matter and on the edges of corpus callosum
APOE is synthesized and secreted primarily by astrocytes and microglia in the CNS (Boyles, Pitas, Wilson, Mahley, & Taylor, 1985;Nakai, Kawamata, Taniguchi, Maeda, & Tanaka, 1996). As mentioned above, it has been postulated that APOE mediates sulfatide depletion in APP transgenic mice models using shot gun lipidomic analysis of brain tissue extracts (Cheng et al., 2010). It has also been reported that the densities of astrocytes (glial fibrillary acidic protein) and microglia (F4/80) implicated in neuroinflammation are increased in/around Aβ 42 -rich amyloid plaques in aged 5xFAD mouse brain (Oakley et al., 2006) which suggests a possible route for accumulation of APOE in/ around amyloid plaques. Therefore, to test if APOE is accumulated and can be correlated with severe sulfatide depletion in amyloid plaques in  the gray matter and on the edges of the corpus callosum in 12-monthold 5xFAD mouse brain (Figure 1), coronal brain tissue sections were double stained with thioflavin S and monoclonal anti-APOE antibody ( Figure 6). The results showed accumulation of APOE protein in all the amyloid plaques in the gray matter and on the edge of the corpus callosum supporting the hypothesis of an association of APOE with sulfatide depletion in the amyloid plaques in 12-month-old 5xFAD mouse brain. Results from three other 12-month-old 5xFAD mice brains are presented in Figure S9.

| D ISCUSS I ON
The myelin sheath is a lipid rich, multilamellar extension of the plasma membrane of oligodendrocytes. In contrast to most biological membranes, myelin contains remarkably high lipid levels which account for 70%-85% of its dry weight (Wang, Palavicini, & Han, 2018). Lipidomics studies have revealed that galactosylceramide and their sulfated form, sulfatides, are the most typical lipids of myelin, where they are highly enriched and their abundance was found to be proportional to the amount of myelin present in the rat brain (Norton & Poduslo, 1973). In addition to galactolipids, cholesterol, ethanolamine-containing plasmalogens, phosphatidylethanolamines, phosphatidycholines and sphingomyelin are also major lipid constituents of the myelin sheath (Norton & Poduslo, 1973) (Wang et al., 2018).
This study reports the loss of myelin-associated lipids, disruption of myelin sheath, and APOE deposition localized to amyloid plaques in aged (12-month-old) 5xFAD mouse brain tissue. The goal was to shed further light on the molecular, morphological, and immune signatures of myelin lipid loss in association with amyloid pathology in AD in a spatially specific manner. Amyloid plaques are exceedingly rare in the white matter in AD pathology (Nasrabady et al., 2018). This is apparent here with 12 month-old 5xFAD mouse brain in which we observed amyloid plaque accumulations abundantly in the gray matter areas including subiculum, cortex, CA and DG, while there were a few plaques visible on the edges of corpus callosum in the white matter ( Figure S2b). The presence of the latter results in a direct exposure of myelin-rich corpus callosum in the white matter to amyloid plaques rich in Aβ 1-42 and Aβ 1-40 (Oakley et al., 2006).
It was previously demonstrated that Aβ peptides, particularly Aβ 1-42, are toxic to oligodendrocytes and can lead to their increased apoptotic cell death (Desai et al., 2010;Xu et al., 2001). The abundant lipid structural components of the myelin sheath, sulfatides, are mainly found in oligodendrocytes (the myelinating cells) in the CNS (Hirahara et al., 2017), whereas low amounts have been detected in neurons and astrocytes (Isaac et al., 2006). Therefore, it can be inferred that Aβinduced degeneration of oligodendrocytes might lead to severe loss of myelin lipids, particularly sulfatides, which can cause degeneration of the myelin sheath in amyloid plaques on the edges of the corpus callosum as observed here. Shotgun lipidomic analysis of brain tissue extracts has previously revealed that decreases in sulfatide levels coincide with the elevation of ceramides in early stages of AD (Han et al., 2002). Here, MALDI-IMS reveals co-localization of the accumulation of ceramides and the depletion of sulfatides in amyloid plaques in several brain regions including cortex, subiculum and CA and on the edges of corpus callosum in the 5xFAD mouse brain (Figure 3). While ceramides may arise as possible precursor/degradation products of sulfatides, a body of evidence suggests that the ceramide elevation results from Aβ-mediated activation of sphingomyelinases (SMases) that catalyze the breakdown of sphingomyelins (SMs) to ceramides (Jana & Pahan, 2004;Lee et al., 2004). Grimm et al. reported increased activity of SMases in presenilin-FAD mutations (that also exist in 5xFAD), where Aβ1-42 directly activates neutral SMase (Grimm et al., 2005). In our imaging analysis, no clear changes in sphingomyelin levels in amyloid plaques were observed in 12 month old 5xFAD mouse brain (data not shown).
The mechanism of sulfatide depletion in amyloid plaques still remains elusive. The content of sulfatides in the CNS is modulated by APOE protein in an isoform-dependent manner through APOEcontaining CNS lipoproteins (Han et al., 2003). APOE, the lipid transport protein expressed by microglia (Nakai et al., 1996) and astrocytes (Boyles et al., 1985) in the brain was also evidenced to be essential for Aβ deposition in APP V717F AD mouse model (Bales et al., 1999). To identify the mechanisms of sulfatide depletion in AD, Han et al. (Han, 2007) proposed a working mechanism which suggests APOE-associated lipoprotein particles released from astrocytes can acquire sulfatides from the myelin sheath through a ''kiss-and-run'' mechanism. This is then followed by metabolism and degradation of the resulting sulfatide-containing APOE-associated lipoprotein particles through endocytic pathways. This hypothesis was tested by Cheng et al. (Cheng et al., 2010) using two human APP expressing tg mice, PD (APP V71F ) and tg2576 (APP Swe 670/671 ), both of which display extensive amyloid plaques. Analysis by shotgun lipidomics was used to demonstrate that the sulfatide levels were reduced in both animals in an Aβ pathology-dependent and age-dependent manner and sulfatide depletion did not occur in APP mutant, APOE null (APOE −/− ) animals relative to the APOE −/− controls. This provides evidence for the association of APOE with sulfatide depletion in AD pathogenesis (Cheng et al., 2010). Further, the brains of APP tg mice, tg2576 (APP Swe 670/671 ), and indeed 5xFAD (Hong et al., 2013) have higher APOE levels relative to age-matched controls. Accumulation of astrocytes (glial fibrillary acidic protein) and microglia (F4/80) as a response to amyloid pathology in 5xFAD mouse brain (Oakley et al., 2006) suggests a possible route for accumulation of APOE in/around amyloid plaques. In our study, we imaged the APOE distribution and observed clear deposition of APOE in/around the amyloid plaques in the gray and white matter in 12-month-old 5xFAD mouse brain ( Figure 6). This is in line with the severe depletion of several sulfatide species in all the amyloid plaques in the grey matter and white matter ( Figure 1). Thus, it can be hypothesized that the elevated levels of APOE, a result of the immune response to amyloid deposition, irrespective of the brain region, can facilitate the depletion of sulfatides through APOE-associated lipoprotein metabolism in/around Aβ plaques.
Significant depletion of ethanolamine plasmalogen phospholipids has been reported in human AD subjects and transgenic AD mice models (Ginsberg, Rafique, Xuereb, Rapoport, & Gershfeld, 1995;Han et al., 2001b;Igarashi et al., 2011) and this depletion is AD specific compared to the primary site of neurodegeneration in either Huntington's disease (caudate nucleus) or Parkinson's disease (substantia nigra) (Ginsberg et al., 1995). PC species were also reported to be diminished in AD subjects (Whiley et al., 2014).
Oxidative stress (Farooqui, Rapoport, & Horrocks, 1997), inflammation (Katafuchi et al., 2012), and peroxisome dysfunction (Kou et al., 2011) are possible amyloid plaque-associated mechanisms that have been proposed for plasmalogen depletion in AD. Indeed, Aβ peptide deposits, damaged neurons, and glial cells are obvious stimuli for focal inflammation in AD due to their discrete, localized accumulation (Akiyama et al., 2000;Heneka et al., 2015). Amyloid peptides and inflammatory mediators accumulated in amyloid plaques can upregulate the activity of PLA 2 enzymes which can result in the depletion of plasmalogen PEs and phosphatidylcholines in amyloid plaques. Indeed, partial accumulation of PLA 2 around Aβ plaques has been shown by immunostaining in 5xFAD mouse brain, which suggests a stimulated activity of this enzyme and underlies focal amyloid-associated inflammatory processes (Hong et al., 2016). Here, accumulations of LPE and LPC species in amyloid plaques ( Figure 3) suggest a mechanism involving inflammatory activation of PLA 2 in association with amyloid plaques. On the other hand, plaque-associated oxidative stress can result in degradation of plasmalogens by reactive oxygen species (ROS). Plasmalogens are particularly susceptible to oxidative stress due to a vinyl-ether group in their molecular structure and may act as scavengers to protect lipids and lipoproteins from oxidative damage (Su, Wang, & Sinclair, 2019). Therefore, depletion of plasmalogens can exacerbate the focal oxidative damage to the lipids and proteins in amyloid plaques in 5xFAD mice brains. Additionally, increased levels of Aβ-induced ROS can also induce peroxisomal dysfunction, thus reducing alkyl-dihdroxyacetone phosphate synthase (AGPS) protein stability, which in turn decreases PE-PL synthesis, due to the dysfunction in peroxisomes where plasmalogens are biosynthesized (Grimm et al., 2011).
Upregulation of LPC species in amyloid plaques can be also implicated in focal demyelination (Mitew et al., 2010). LPC induces focal demyelination (Allt, Ghabriel, & Sikri, 1988;Hall, 1972) and disrupts myelin lipids nonspecifically (Plemel et al., 2018). This is in line with the myelin lipid loss in amyloid plaques, particularly in the myelin lipid-rich corpus callosum, in aged 5xFAD mouse brain (Figure 3), which might suggest a role of LPC for the disruption of the lipid-rich myelin sheath in this region.
Additionally, depletion of myelin lipids in amyloid plaques can be involved in the discussion of well-studied lipid metabolism-related amyloid plaque immunity in 5xFAD mouse brain such as the studies on the microglial receptor, triggering receptor expressed on myeloid cells 2 (TREM2). TREM2 lipid sensing was shown to sustain microglial response and metabolic fitness around the amyloid plaques and limits the diffusion and toxicity of amyloid plaques (Wang et al., 2016) in 5xFAD mice brain. Human TREM2 reporter cells were found to be stimulated by various phospholipids and non-phosphate anionic and zwitterionic lipids including sulfatides (Ulland et al., 2017;Wang et al., 2015). Further, elevated TREM2 expression in microglia by elevated TREM2 gene dosage in 5xFAD mice reshapes microglial response which ameliorates AD neuropathology (Lee et al., 2018). Therefore, it might be hypothesized that significant focal loss of anionic sulfatides and plasmalogen PEs might impair TREM2 lipid sensing, particularly in the myelin lipid-rich corpus callosum, and limit the microglial activation for degradation of Aβ in 5xFAD mice brain.
Taken together, high-spatial resolution MALDI-IMS of coronal brain tissue sections of 12-month-old 5xFAD mouse revealed focal amyloid plaque-associated depletion of several myelin-associated lipid species.
Double staining of the consecutive sections with fluoromyelin and amyloid-specific antibody revealed amyloid plaque-associated myelin lipid disruption which occurs specifically on the edges of the corpus callosum in the white matter while there was no disruption of myelin sheath observed in amyloid plaque areas in several other brain regions in the grey matter. APOE accumulation in amyloid plaques is in line with sulfatide depletion in several brain regions suggesting that sulfatide depletion might be APOE deposition associated as an immune response to amyloid aggregation in all the plaques in the white and grey matters in 5xFAD mouse brain. Our data leverage the understanding of morphological, molecular, and immune signatures of amyloid plaque-associated myelin lipid loss in a brain region-specific manner. It is particularly F I G U R E 3 Matrix-assisted laser desorption/ionization imaging mass spectrometry (MALDI-IMS) reveals colocalized depletions of sulfatides and elevations of long-chain ceramides in amyloid plaques in the cortex, and hippocampus in 12-month old 5xFAD mouse brain (n = 4 mice, n = 3 sections each). (a) Spatial distribution of amyloid aggregates (brown aggregates) in the hippocampus and adjacent cortical regions in 12-month-old 5xFAD mouse brain visualized with anti-beta amyloid antibody and counterstaining with hematoxylin. F I G U R E 4 Double staining of coronal brain tissue sections with FluoroMyelin (in green), and anti-amyloid beta (Aβ) antibody (in red) reveals focal amyloid plaque-associated disruption of myelin on the edge of corpus callosum in the white matter of 12-month-old 5xFAD mice brain (n = 4 mice, n = 3 sections each). Coronal tissue sections were stained with (