Aβ-Induced Alterations in Membrane Lipids Occur before Synaptic Loss Appears

Loss of active synapses and alterations in membrane lipids are crucial events in physiological aging as well as in neurodegenerative disorders. Both are related to the abnormal aggregation of amyloid-beta (Aβ) species, generally known as amyloidosis. There are two major known human Aβ species: Aβ(1–40) and Aβ(1–42). However, which of these species have more influence on active synapses and membrane lipids is still poorly understood. Additionally, the time-dependent effect of Aβ species on alterations in membrane lipids of hippocampal neurones and glial cells remains unknown. Therefore, our study contributes to a better understanding of the role of Aβ species in the loss of active synapses and the dysregulation of membrane lipids in vitro. We showed that Aβ(1–40) or Aβ(1–42) treatment influences membrane lipids before synaptic loss appears and that the loss of active synapses is not dependent on the Aβ species. Our lipidomic data analysis showed early changes in specific lipid classes such as sphingolipid and glycerophospholipid neurones. Our results underscore the potential role of lipids as a possible early diagnostic biomarker in amyloidosis-related disorders.


Introduction
Amyloidosis is used as an umbrella term for rare, serious diseases caused by the deposit of misfolded proteins. In brain tissue, it is characterised by the accumulation of amyloid-beta (Aβ), such as occurs in Alzheimer's disease (AD) [1]. Aβ species are products of a proteolytic cleavage, generated from amyloid precursor protein (APP) by αor β-secretase and γ-secretase activity, and their characteristics have been extensively reviewed [1][2][3]. However, the presumed neurotoxic effects of the major Aβ species, Aβ  and Aβ  , under pathological and physiological conditions remain unclear. So far, the molecular and cellular mechanism of Aβ species and their impact on the loss of synaptic sites, as well as on the changes of membrane lipids of brain cells, is poorly understood.
APP has been shown to play a pivotal role in synaptic and neural plasticity [4]. In vitro and in vivo studies have demonstrated that soluble Aβ species accumulate at the synaptic sites, resulting in disrupted synaptic plasticity and long-term potential [5][6][7][8]. However, Aβ  is thought to be more aggregation-prone compared to Aβ  . The aggregation status of Aβ species in senile plaques in AD is strongly regulated by time and their aggregation affinity [9,10]. Therefore, as has been demonstrated in vitro and in vivo, Aβ species have different influences on the pre-and postsynaptic densities dependent on the Aβ concentrations, the chemical structure of Aβ (α-helices or β-sheets), the time of Aβ treatment, and the Aβ aggregation status [2,[11][12][13][14][15][16][17][18][19][20][21]. Previously, it was also shown that glial cells are involved in Aβ-induced inflammatory responses and play an important role in Aβ clearance and degradation [22]. In addition, glial activation itself could play a protective role against Aβ-induced toxicity on neurones [22].
APP and the cleavage products Aβ  and Aβ  have different influences on lipid homeostasis [23]. Major molecular targets for Aβ in the cholesterol and sphingolipid metabolic pathways are 3-Hydroxy-3-methylglutaryl-coenzyme A (HMG-CoA) reductase and sphingomyelinases (nSMase) [23]. Aβ  activates nSMase, whereas Aβ  suppresses the activity of HMG-CoA reductase, resulting in decreased cholesterol levels. Sphingomyelins (SM) reduce γ-secretase activity, thus reducing Aβ levels; cholesterol, on the contrary, induces γ-secretase activity and is responsible for elevated Aβ levels [23,24]. Inhibition of HMG-CoA reductase is responsible for a reduction of intracellular as well as extracellular Aβ  and Aβ  peptides, resulting in elevated levels of cholesterol [25]. Sphingosine-1-phosphate (So1P) has been shown to be protective for neuronal cells, whereas ceramide promotes Aβ biogenesis by influencing the β-secretase of APP [26]. Ceramides also interfere in the control of many cellular processes, influencing, e.g., Aβ aggregation in physiological and pathological aging processes [27]. Phosphatidylcholines can alter the Aβ  mediated aggregation, depending on the thickness of the lipid membrane [28]. Charged phospholipid bilayers consisting of phosphatidylcholines and phosphoglycerol showed an increase in Aβ  fibril formation [29].
In this study, we investigated the direct effect and time-dependent influence of two major human Aβ species on primary hippocampal neurones and on glial cells. Hence, we examined whether the loss of active synapses in primary hippocampal neurones is Aβ-species-and time-dependent in vitro.
Our results show for the first time a detailed lipid profiling in two different types of brain cells, primary hippocampal neurones and glial cells, after 3 h (3 h) and 12 h (12 h) of Aβ  and Aβ  treatment. We established that both Aβ species cause changes in the membrane lipids of primary hippocampal neurones and glial cells after 3 h and 12 h of treatment. Interestingly, Aβ-induced alterations in membrane lipids occurred prior to the loss of active synapses.

Animals
For all experiments, timed-pregnant and postnatal mice were obtained from the central animal facility of the Carl von Ossietzky Universität Oldenburg. The primary hippocampal neurones were derived from C57 BL/6 mouse embryos at embryonic stage 18 (E18). These experiments were carried out in accordance with the institutional guidelines for animal welfare and approved by the "Niedersächsisches Landesamt für Verbraucherschutz und Lebensmittelsicherheit" (33.19-45502-04-18/2766). The glial cells were obtained from postnatal day 2 (P2) mouse pups in accordance with the institutional guidelines of German animal welfare ( §4) for the use of laboratory animals at the Carl von Ossietzky Universität Oldenburg.

Primary Glial Cell Cultures
Glial cell cultures were prepared from cortical tissue of P2 mouse pups. The brains were dissected and washed twice in HBSS (1X, Phenol red, Thermo Fisher Scientific). After removing the meninges from the brains, cortical tissue was isolated, washed twice in HBSS and incubated with 0.05% trypsin-EDTA (Gibco, Thermo Fisher Scientific) for 12 min at 37 • C. The supernatant was then carefully removed and replaced by plating media, consisting of DMEM (high glucose with L-glutamine, 4500 mg/L D-glucose, 110 mg/L sodium pyruvate, Gibco, Thermo Fisher Scientific) supplemented with 10% FBS (Pan-Biotech), 10% HS (Pan-Biotech) and 100 U/mL penicillin with 100 µg/mL streptomycin (Pan-Biotech). Tissue was carefully triturated and centrifuged for 1 min at 20× g. Supernatant was cautiously transferred to a new falcon tube and centrifuged for 5 min at 300× g. The supernatant was discarded and the pellet resuspended in plating media. Before plating, the T75 flask (Sarstedt, Label A, The Netherlands), coated overnight with 0.1 mg/mL PLL (P2636-100MG, Sigma-Aldrich) in ortho-boric acid (VWR chemicals) buffer (pH 8.5), was rinsed three times in MilliQ water. The dissociated glial cells were filtered through a 70 µm cell strainer (Corning, VWR, Darmstadt, Germany) and plated into the T75 flasks. After 2-3 days, the culture medium was changed and further maintained in 37 • C at 5% CO 2 . After 10-12 DIV, microglia were separated from astrocytes by agitation (230 rpm) at 37 • C for 3-4 h for obtaining the microglia. Adherent astrocytes were harvested using 0.25% trypsin-EDTA for 5 min in an incubator. Astrocytes were collected and centrifuged at 473× g for 10 min. The cell pellet was resuspended in plating media consisting of 45.5% DMEM (high glucose with L-glutamine, 4500 mg/L D-glucose, 110 mg/L sodium pyruvate, Gibco, Thermo Fisher Scientific), 45.5% HAMS-F12 nutrient mix (Gibco, Thermo Fisher Scientific), 8% FBS (Pan-Biotech), 1% Pen/Strep (Pan-Biotech), and seeded into 60 mm-diameter Petri dishes (Nunclon TM Delta surface, Thermo Fisher Scientific).

Synaptic Study Immunofluorescence
After 3 h and 12 h DMSO and Aβ treatment, the conditioned medium was removed and the cultured hippocampal neurones were washed once with warm phosphate buffer saline (PBS (1X), Gibco, Thermo Fisher Scientific), followed by fixation with 4% paraformaldehyde (PFA, Merck Millipore, Darmstadt, Germany) in PBS (1X) containing 15% D-(+)-saccharose (≥99.7%, Carl Roth) at 4 • C for 10 min. After fixation, the cells were washed three times with PBS (1X) before being permeabilised with 0.1% triton X-100 (Carl Roth) for three minutes at room temperature (RT). The neurones were then washed three times with PBS (1X) for 10 min while gently shaking on a shaker at RT. The neurones were blocked in PBS (1X) containing 10% foetal bovine serum (FBS, Pan Biotech) and 1% normal goat serum (NGS, Vector laboratories, Biozol, Germany) for 1 h on a shaker at RT. Afterward, they were incubated with two primary antibodies: guinea-pig anti-vesicular glutamate transporter-1 (VGlut-1 (1:200), Synaptic Systems, 135304, Göttingen, Germany) and mouse anti-postsynaptic density protein-95 (PSD-95 (1:1000), clone (7E3-1B8), Thermo Fisher Scientific, MA1-046) overnight on a shaker at 5 • C. After washing the neurones three times with PBS they were incubated with two secondary antibodies: goat anti-mouse Alexa Fluor 488 (1:1500, Molecular probes) and goat anti-guinea-pig cyanine fluorescent dye (Cy3) (1:1000, Jackson Immuno Research, Cambridgeshire, UK) on a shaker for 1.5 h at RT. Nuclear staining was performed using DAPI (1:2000). All primary and secondary antibodies, as well as DAPI (Carl Roth), were diluted in PBS (1X) containing 5% FBS and 1% NGS. The coverslips with neurones were mounted on glass slides (Duran 76 mm × 26 mm, Carl Roth) with Immu-mount ® (Shandon, Thermo Fisher). Immunofluorescent images for quantification of synaptic profiling between pre-and postsynaptic markers were captured using an Olympus IX83 invert microscope with DP80 camera (Olympus, Shinjuku, Japan) using the UPlanSApo 100×/1.4 oil objective with the following filter modules: U-F39002 AT-FITC for Alexa Fluor 488, U-F39004 AT-CY3 for Cy3, and U-FF for DAPI. Background correction, adjustment of brightness and contrast, and selection of regions of interest (ROI) were performed by CellSense software (Olympus, Shinjuku, Japan). Fluorescent images ( Figure 1A) were captured with an inverted laser-scanning confocal microscope (SP8, Leica, Wetzlar, Germany) with a 63×/1.4 oil objective (zoom 3) using the Leica software. Images were taken using 488 nm and 564 nm lasers. Background correction, brightness, and contrast were adjusted by ImageJ software (NIH, Bethesda, MD, USA). Further processing of the images was carried out using Adobe Illustrator CC 2020.
2.6. Lipidomic Study 2.6.1. Samples Collection for Lipid Analysis Hippocampal primary neurones (12 DIV) and glial cells were harvested and collected in Eppendorf ® LoBind micro-centrifuge tubes (Eppendorf, Omnilab) after 3 h and 12 h with and without Aβ  and Aβ  treatment. The supernatant was discarded. Dishes were washed once with cold PBS (1X), neurones and glial cells collected in LoBind tubes after 2 min' centrifugation (9391× g) at 4 • C, the supernatant was discarded, and the pellets were snap-frozen and stored at −80 • C.  The synaptic study showed the colocalisation of presynaptic (Vglut-1) and postsynaptic (PSD-95) markers, which was defined as active synapses, in hippocampal neurones at DIV12 for both control groups (CTRL and DMSO) and Aβ-treated groups ((1 µM) Aβ  and (1 µM) Aβ  ) at different time points (3 h and 12 h) of treatment (all groups from four independent cultures (n = 4)). In the 3 h treatment, the following average number of neurones was counted in each group: CTRL (n = 12), DMSO (n = 9), Aβ (1-40) (n = 11) and Aβ (1-42) (n = 9). In the 12 h treatment, the following average number of neurones was counted in each group: CTRL (n = 14), DMSO (n = 13), Aβ  (n = 13), and Aβ (1-42) (n = 17). For both 3 h and 12 h treatments, 3 ROI from each neurone was counted. (A) Representative confocal images of control groups and Aβ-treated groups of hippocampal neurones stained for both presynaptic (Vglut-1) and postsynaptic markers (PSD-95) are shown. There was a reduction in active synapses after 12 h of Aβ (1-40)-and Aβ  -treated neurones compared to control groups. White boxes show enlargements of an example of defined active synapses for both control groups and Aβ-treated groups (3 h and 12 h). (B,C) Quantification of active synapses from controland Aβtreated groups of hippocampal neurones. (B) No significant effect on active synapses was seen in Aβ-treated groups after 3 h (3 ROI of average counted cells; n = 10, from four independent cultures (n = 4)). (C) A significant reduction of active synapses was observed after 12 h of Aβ  and Aβ  -treated neurones compared to control groups. No significance (ns) was seen between CTRL and DMSO, nor between Aβ (1-40) -and Aβ  -treated neurones (3 ROI of average counted cells; n = 14, from four independent cultures (n = 4)). *** p < 0.0001; scale bars represent 5 µm. The size of the ROI was 47 µm 2 .

Lipid Extraction
Hippocampal primary neurone (12 DIV) and glial-cell pellets were transferred by glass pipettes into transparent glass centrifuge tubes (Corning) containing chloroform (CHCl 3 , SupraSolv ® , Merck, Darmstadt, Germany), methanol (LiChrosolv ® HPLC gradient grade, Merck), and fuming hydrochloric acid 37% (Rotipuran ® , p.a., ACS, ISO, Carl Roth) solutions supplemented with 1% butylhydroxytoluol (BHT, ≥99%, Carl Roth) and fractioned by pipetting up and down while the sample was on ice. Then, 1 µL TopFluor lysophosphatidic acid (LPA) (702.58 g/mol, Avanti 810280P-1MG), was added per 1 million cells. The TopFluor LPA was dissolved in 1 mL CHCl 3 (1 mg/mL). Then CHCl 3 and water (H 2 O, Rotisolv ® HPLC gradient grade, Carl Roth) were added, followed by vortexing and 30 min incubation in the dark at RT. Samples were centrifuged at 1260× g for 10 min at RT. Finally, the lipid phase was collected using a glass Pasteur pipette into a new glass vial (Thermo Fisher Scientific) and placed in a nitrogen chamber with 6% O 2 concentration, overnight in the dark. Until measurement, the samples were stored at −20 • C.

Data Analysis
For quantification of our synaptic study, we counted active synapses on hippocampal neurones from 12 DIV. Synapses in hippocampal neurones were stained with antibodies specific for synaptic proteins (PSD-95, postsynaptic in green; Vglut-1, presynaptic in red). Only puncta with obvious PSD-95/Vglut-1 overlap (in yellow) were counted as active synapses [32]. We analysed three regions of interest (ROI = 47 µm 2 ) on each neurone for an average of ≥12 neurones from four different cultures (n = 4) after both 3 h and 12 h of Aβ treatment. In the 3 h treatment, the following average amount of neurones were counted in each group: CTRL (n = 12), DMSO (n = 9), Aβ (1-40) (n = 11) and Aβ (1-42) (n = 9). For the 12 h treatment, the following average amount of neurones were counted in each group: CTRL (n = 14), DMSO (n = 13), Aβ  (n = 13) and Aβ  (n = 17). Statistical analysis was performed using GraphPad Prism 7 (GraphPad Software, San Diego, CA, USA). Values were analysed for normal distribution using the Shapiro-Wilk test. As the data were not normally distributed, they were analysed by performing a Kruskal-Wallis test followed by a post hoc Dunn's multiple comparison test. All data are presented as mean + standard deviation (SD) and considered to be significant if p < 0.05 (*** p < 0.0001). Tandem mass spectrometry data of three independent experiments (n = 3) from neurones and glial cells were further processed by GraphPad Prism 7, performing logarithmic 10 transformation, and plotted to our negative control (DMSO). This was because the raw data showed that a low concentration of DMSO (final concentration 0.03%) showed an effect on both the lipid composition of glial cells and hippocampal primary neurones, as previously demonstrated [33]. Z-score transformation was performed, and data were plotted as heat maps for each lipid class.  and Aβ  Treatment

Synaptic Loss after Aβ
To examine whether Aβ species have a preferential early or late effect on the loss of active synapses, primary hippocampal neurones at 12 DIV were treated with (1 µM) Aβ  or (1 µM) Aβ  for 3 h and 12 h. Active synapses were analysed by colocalisation of fluorescent markers recognizing vesicular glutamate transport 1 (Vglut-1) and postsynaptic density protein-95 (PSD-95) ( Figure 1A-C). We quantified three ROI (47 µm 2 ) on the dendritic tree of the hippocampal neurones by examining the colocalisation dots ( Figure 1A, white boxes) of a pre-and postsynaptic marker. No significant effect on active synapses was observed after 3 h treatment between our control groups (CTRL and DMSO) compared to Aβ-treated groups (Aβ  and Aβ  ). After 12 h, however, Aβ-treated groups showed a significant reduction in active synapses as compared to the control groups ( Figure 1A and Table 1). Both 3 h and 12 h DMSO treatments showed a small decrease in active synapses compared to the CTRL, but this was not significant ( Figure 1B,C and Table 1). We observed no significant difference after 3 h and 12 h treatment between Aβ  and Aβ  treatments (Figure 1B,C and Table 1). However, 3 h Aβ  and Aβ  treatment showed a slight decrease in active synapse numbers as compared to our control groups, but this was not significant ( Figure 1B,C and Table 1). Table 1. List of the statistically adjusted p-value (from Figure 1B,C) between separate groups (CTRL, DMSO, Aβ  and Aβ  ) after 3 h and 12 h of treatment conditions. A Kruskal-Wallis test was performed, followed by a Dunn's multiple comparison test.   0.3017 <0.0001 CTRL vs. Aβ  0.0867 <0.0001 DMSO vs. Aβ  >0.9999 <0.0001 DMSO vs. Aβ  >0.9999 <0.0001 Aβ  vs. Aβ  >0.9999 >0.9999

Disruption in Cellular Lipids of Hippocampal Neurones and Glial Cells after Aβ
Hippocampal neurones at 12 DIV and glial cells were treated with Aβ  and Aβ (1-42) for 3 h and 12 h to test whether Aβ species influence cellular lipids, whether that has a preference for a specific class of lipids, and to examine whether Aβ  and Aβ  have early effects on the lipid changes in these cells. Two different sets of analyses were performed: HPTLC and tandem MS. Lipids of hippocampal neurones and glial cells were extracted equally from our control groups (CTRL and DMSO) and Aβ-treated groups (Aβ  and Aβ  ) after 3 h and 12 h of treatment. TopFluor lysophosphatidic acid (LPA) was used as the internal standard, to see whether the same amounts of lipids were extracted from each sample of control-and Aβ-treated groups ( Figure S1A-D). To analyse which different lipid classes (e.g., sphingolipids and glycerophospholipids) were dysregulated, we showed HPTLC-derivatised copper (II)-sulphate images and their corresponding scanning profiles for hippocampal neurones and glial cells treated with Aβ  and Aβ  for 3 h and 12 h ( Figure S2A-D). These images and profiles revealed changes in different lipid classes (e.g., sphingolipids and glycerophospholipids), which were in accordance with the external standards used ( Figure S2, purple arrows and boxes). Each numbered purple arrow and box refers to external standards that could be detected from hippocampal neurones and glial cells after 3 h and 12 h of Aβ  and Aβ  treatment. The copper (II)-sulphate scanning profiles are shown to facilitate examination of the changes between control groups and Aβ-treated groups. The represented retention factor (Rf , Table S1) and arbitrary unit (AU , Table S1) are both based on the peak's end in the copper (II)sulphate scanning profiles. These show the changes in band intensity of the HPTLC copper (II)-sulphate images and correspond to the numbered labelled purple arrows and boxes ( Figure S2). Our data showed small changes in sphingolipids and glycerophospholipids after 3 h Aβ  and Aβ  treatment for both neurones ( Figure S2A) and glial cells ( Figure S2C), and large changes after 12 h of Aβ  and Aβ  treatment ( Figure S2B,D).  and Aβ

Treatment from Hippocampal Neurones and Glial Cells Using Tandem Mass Spectrometry Analysis
To corroborate whether Aβ  and Aβ  preferentially influence specific sphingolipid or glycerophospholipids isoforms, we treated hippocampal neurones at DIV 12 and glial cells for 3 h and 12 h with Aβ  and Aβ  and subsequently analysed them by tandem MS. Here, the characterisation of altered lipid isoforms for hippocampal neurones and glial cells after 3 h and 12 h Aβ  and Aβ  treatment are shown, focussing on sphingolipids such as ceramides (Cer), dihydroceramides (DHCer), lactosylceramides (LacCer), monohexosylceramides (HexCer), sphingosine (So), sphinganine (Sa), sphingosine-1-phosphate (Sa1P), sphinganine-1-phosphate (So1P), sphingosylphosphorylcholine (SPC), sphingomylines (SM), dihydrosphingomylines (DHSM), and glycerophospholipids such as phosphatidylcholine (PC), lyso-phosphatidylcholine (LPC), lysophosphatidylethanolamine (LPE), lyso-phosphatidylglycerol (LPG), lyso-phosphatidylserine (LPS), Lyso-platelet-activating factor (Lyso-PAF) (Figures 2-4). All results are presented as a log 10 transformation and were performed in triplicate. Furthermore, ratios between the Aβ-treated groups and our negative control (DMSO) were calculated, setting the negative control (DMSO) to 0. After taking the ratios, the Z-score was taken for the total data set. Changes in lipid isoforms are shown as tendencies. This means that the red colours refer to an increase of lipid quantities and the green colour indicates a reduction compared to our negative control (DMSO) which was set to 0 as baseline parameter (Figures 2-4).

Ceramides (Cer)
Both hippocampal neurones and glial cells showed alterations in Cer isoforms after 3 h and 12 h Aβ  and Aβ   treatment (Figure 2). Neurones showed a reduction in C14 Cer after 3 h Aβ  treatment. An increase of C14 Cer was observed after 3 h and 12 h of Aβ  and Aβ  treatment. Other ceramide isoforms (C16 Cer, C18:0 Cer, C20 Cer, C22 Cer, C24:1 Cer) showed an increase in 3 h Aβ  -and 12 h Aβ  -treated neurons. Aβ-treated glial cells showed a strong increase after 3 h Aβ  and Aβ  treatment for C14 Cer. After 12 h, Aβ  -treated glial cells showed a strong reduction compared to Aβ  for C14 Cer. A weak reduction in C18:1 Cer for 3 h Aβ  and 12 h Aβ   (Figure 2) was also observed. Other ceramide isoforms (C16 Cer, C18:0 Cer, C20 Cer, C22 Cer, C24:1 Cer) showed a weak increase after 12 h Aβ  -treatment on glial cells. An increase was observed in C24 Cer after 3 h Aβ  as well as after 3 h and 12 h Aβ  in treated neurones. No changes were observed in C24 Cer of Aβ  -and Aβ  -treated glial cells (Figure 2).

Dihydroceramides (DHCer)
In Aβ  -treated glial cells, C14 DHCer, C16 DHCer and C18:0 DHCer decreased after 3 h and increased after 12 h treatment. However, in glial cells C14 DHCer and C18:0 DHCer was increased after 3 h of Aβ  treatment. After 3 h Aβ  -and Aβ  treatment on glial cells, a reduction in C20 DHCer and an increase in C22 DHCer was observed. After 12 h there was a decrease in Aβ  -and an slight increase in Aβ  treated glial cells. C24:1 DHCer showed a weak reduction after 3 h Aβ  and Aβ  treatment as well as an increase after 12 h Aβ   (Figure 2). Aβ  -and Aβ  -treated hippocampal neurones showed a specific decrease in C20 DHCer after 3 h, and no changes after 12 h Aβ  and Aβ      and (1 µM) Aβ  of hippocampal neurons at DIV 12 and glial cells (all groups, from three independent experiments (n = 3)). Changes of these lipid classes are shown as logarithmic (log 10 ) relative intensity (arbitrary unit); the green colour refers to a reduction, and the red colour refers to an increase of lipid levels compared to our negative control (DMSO), which was set to 0 as baseline. Both hippocampal neurones and glial cells showed an increased and reduced intensity of Cer, DHCer, LacCer, and HexCer isoforms after both 3 h and 12 h Aβ  and Aβ  treatment. Different changes in the lipid classes are shown between treated hippocampal neurones and glial cells. Aβ-treated hippocampal neurones showed prominent increases and decreases in Cer isoforms and LacCer and HexCer isoforms after 3 h and 12 h of Aβ  and Aβ  treatment. There was both a prominent intensity increase and a decrease of DHCer isoforms after 3 h and 12 h of Aβ  and Aβ  treatment. Glial cells showed prominent lipid changes of Cer, DHCer, LacCer, and HexCer isoforms after 3 h and 12 h of Aβ  or Aβ  treatment.

Lactosylceramides (LacCer)
Glial cells treated with Aβ  and Aβ  showed a reduction in all LacCer isoforms after 12 h. In 3 h Aβ  -and 12 h Aβ  -treated glial cells, decreases in C18:0 LacCer and C22 LacCer and an increase after 12 h Aβ  treatment were seen. In C16 LacCer and C24:1 LacCer an increase after 12 h Aβ  and Aβ  treatment on glial cells was seen. A reduction was observed in C16 LacCer after 3 h Aβ  and a decrease after 3 h Aβ  treatment on glial cells. After 3 h, Aβ  -treated hippocampal neurones showed an increase in C18:0 LacCer. A decrease in C16 LacCer and C18 LacCer was seen in 3 h Aβ  -treated hippocampal neurons (Figure 2). A slight decrease was observed in C18:0 LacCer after 12 h Aβ  and Aβ  -treatment on neurones as well as in C22 LacCer after 3 h Aβ  treatment.

Sphingosine (So) and Sphingosine-1-Phosphate (So1P)
Hippocampal neurones and glial cells showed weak changes of d18:1 So after both 3 h and 12 h Aβ  and Aβ   treatments (Figure 3). There was a decrease in d18:1 So1P after 3 h Aβ  treatment in hippocampal neurones and an increase after 12 h Aβ  treatment in glial cells. A reduction in d18:1 So1P was also observed after 3 h and 12 h Aβ  treatment of hippocampal neurons (Figure 3). An increase in d18:1 So1P for both 3 h Aβ  -treated glial cells and 12 h Aβ  -treated hippocampal neurones was observed (Figure 3). Here, we examined different sphingophospholipid isoforms; sphingosine (d18:1 So), sphinganine (d18:0 Sa), sphingosine-1phosphate (d18:1 So1P), sphinganine-1-phosphate (d18:1 Sa1P), sphingomyelines (SM) and dihydrosphingomyelines (DHSM) after 3 h and 12 h treatment with (1 µM) Aβ  and (1 µM) Aβ  hippocampal neurones at DIV 12 and glial cells (all groups, from three independent experiments (n = 3)). Changes of these lipid classes are shown as logarithmic (log 10 ) relative intensity (arbitrary unit); the green colour refers to a reduction and the red colour refers to an increase in lipid levels compared to our negative control (DMSO), which was set to 0 as baseline. Major intensity changes in lipid classes of d18:1 So, d18:0 Sa, d18:1 So1P, d18:1 Sa1P after 3 h and 12 h Aβ  and Aβ  treatment of hippocampal neurones and glial cells were observed. Specific lipid intensity changes after Aβ  and Aβ  treatment of hippocampal neurones of very long fatty acid SM compared to Aβ-treated glial cells were detected. Sphinganine (Sa) and Sphinganine-1-Phosphate (Sa1P) A prominent reduction was seen in d18:0 Sa for glial cells after 3 h and 12 h Aβ  and Aβ  treatment ( Figure 3). In hippocampal neurones d18:0 Sa was slightly increased after 3 h Aβ  treatment and slightly reduced after 12 h Aβ  treatment. A reduction was also observed in d18:0 Sa1 P after 3 h Aβ  -and Aβ  -treated hippocampal neurones, as well as 3 h and 12 h Aβ  -treated glial cells. No strong reduction was observed in d18:0 Sa1P after 12 h Aβ  and Aβ  treatment of hippocampal neurones as well as in 3 h Aβ  -treated glial cells ( Figure 3).

Sphingosylphosphorylcholine (SPC)
Both, hippocampal neurones and glial cells showed alterations in SPC 16:0 after 3 h Aβ  and Aβ  treatment. The most prominent decrease in SPC 16:0 was shown after 3 h of Aβ  and Aβ  treatment for both hippocampal neurones and glial cells. A decrease was also observed after 12 h of Aβ  treatment for both hippocampal neurones and glial cells (Figure 3).

Phosphatidylcholines (PC) and Lyso-Phosphatidylcholines (LPC)
A prominent decrease in PC (38:0) in hippocampal neurones after 3 h Aβ  as well as 12 h Aβ  and Aβ  was observed, as well as an strong increase after 3 h Aβ  treatment. All other PC isoforms (PC ( Glial cells also showed a prominent increase in PC (36:0) after 3 h Aβ  and Aβ  treatment. No changes were observed in PC (38:0) after 12 h Aβ  treatment of glial cells. LPC (20:0) showed a strong increase after 3 h Aβ  treatment of hippocampal neurones and in 3 h Aβ  -and 12 h Aβ  -treated glial cells (Figure 4). A prominent decrease was observed in LPC (20:0) in 3 h Aβ  and 12 h Aβ  treated glial cells and a weak decrease in 12 h Aβ  -treated hippocampal neurones (Figure 4). Figure 4. Heat map-based mass spectrometry analysis of glycerophospholipid classes from hippocampal neurones and glial cells after 3 h and 12 h Aβ treatment. Here, we examined different glycerophospholipids isoforms: phosphatidylcholine (PC), lyso-phosphatidylcholine (LPC), lysophosphatidylethanolamine (LPE), lyso-phosphatidylglycerol (LPG), lyso-phosphatidylserine (LPS), and lyso-platelet-activating factor (Lyso-PAF) after 3 h and 12 h treatment with (1 µM) Aβ  and (1 µM) Aβ  of hippocampal neurones at DIV 12 and glial cells (all groups, from three independent experiments (n = 3)). Changes in these lipid classes are shown as logarithmic (log 10 ) relative intensity (arbitrary unit); the green colour refers to a reduction and the red colour refers to an increase in lipid levels compared to our negative control (DMSO), which was set to 0 as baseline. Hippocampal neurones specifically showed PC (38:0) intensity changes after 3 h and 12 h Aβ  and Aβ  treatment compared to an overall reduction in PC for Aβ  and Aβ  -treated glial cells. LPC, LPG, LPS, and Lyso-PAF showed specific changes after 3 h and 12 h Aβ  and Aβ  treatment in both hippocampal neurones and glial cells.

Discussion
In this study, using different quantitative cellular techniques in different Aβ-treated primary brain cells, we showed for the first time that lipid profiles change after treatment and before synaptic loss was observed. We showed that human Aβ species exogenously applied to primary neurones and glial cells influence the numbers of active synapses and led to lipid alterations in a time-dependent manner. We also demonstrated that Aβ  and Aβ  treatment have a specific-species dependent influence on the integrity of cellular lipids in hippocampal neurones and glial cells.
Aβ  and Aβ  induces synaptic loss only after 12 h of treatment. Synaptic dysfunctions due to Aβ accumulation are strongly associated with the cognitive disturbances of AD. The Aβ species Aβ  and Aβ  are the major forms of amyloid β peptides in the brain , whereby Aβ  seems to be more toxic than Aβ   [30,34,35]. We found no selective influences of Aβ  or Aβ  on the loss of active synapses in cultured hippocampal neurones. Moreover, both Aβ species induced synaptic loss after a 12 h treatment, whereas a 3 h treatment did not show effects on synapses either with Aβ  or Aβ  . Differences between these Aβ species on the synapse alterations are dependent on the concentration and aggregated forms of Aβ [30]. Moreover, Fu et al. (2017) [35] showed that the Aβ  oligomers have a moderate but significantly higher level of neurotoxicity, whereas monomers have the weakest neurotoxic effect (~20%) on neuronal cells, without differences between Aβ  and Aβ  . These results correspond to our findings on primary hippocampal neurones.
Aβ  and Aβ  differently influence cellular lipids before synaptic loss appears. Previous studies have shown changes in different lipid classes and a correlation to AD pathology. Although we do not fully understand the connection between AD and lipid metabolism, there is more and more evidence that lipids could be a useful blood biomarker in the diagnostics of AD, in addition to other risk factors [36,37]. Nevertheless, it is poorly understood which lipid isoforms are dysregulated on a cellular level. We showed here that changes in lipid composition in hippocampal neurones and glial cells occur early, even before synaptic loss becomes evident, and independently of the Aβ species.
Our data indicate an increase of ceramides (especially C14 Cer) in glial cells and a reduction of these in hippocampal neurones. Both types of brain cells indicate an early reduction of specific dihydroceramides (C20 DHCer and C22 DHCer). These different amounts of lipids caused by Aβ could be a result of activated apoptotic-survival pathways and inflammatory activity due to their response to oxidative stress [38].
In conclusion, we found that after Aβ  or Aβ  treatment, levels of specific sphingolipid and glycerophospholipid isoforms change before there is any detectable loss of active synapses. No significant differences were seen between the effects of Aβ  and Aβ  species in the loss of active synapses. Our data contributes to a better understanding of the cellular influences of Aβ species. It remains to be elucidated whether these influences of Aβ species on changes in membrane lipid profiles could be useful as an early diagnostic biomarker in amyloidosis-related disorders.