Reduced cholesterol alters the biophysical properties of repaired myelin

Abstract

The myelin sheath is a lipid-rich membrane that ensheathes axons and is required for healthy and efficient signal transduction. Myelin is damaged in neurological diseases like multiple sclerosis, but remyelination can occur through the action of oligodendrocyte precursor cells (OPCs), which differentiate into mature oligodendrocytes that wrap axons to form repaired myelin. In this study, a genetic-based mouse model of demyelination was used, which features near-complete demyelination followed by robust remyelination in the brain. Lipid mass spectrometry on isolated myelin from the remyelinated brain revealed a decrease in the percent mole fraction of cholesterol when compared to healthy myelin. Biophysical studies on monomolecular lipid films formed using myelin lipid extracts from repaired myelin showed changes in the surface behavior of the lipid films, compared to the healthy myelin. Films formed using the remyelinated lipid extracts resulted in lower surface pressures and lower compressional moduli when compared to healthy controls, suggesting that repaired myelin membranes have lower lateral molecular packing within the lipid film. Synthetically prepared model membranes, based on the major lipid compositions of the healthy and diseased extracts, revealed that changes in cholesterol levels were the primary contributor to the changes in biophysical properties. Supplementation of the diseased lipid extracts with cholesterol led to a robust improvement in membrane surface pressures and compressibility. Together, these results suggest that high cholesterol levels are required for myelin membrane stability and that reduced cholesterol in repaired myelin may have a profound impact on the biophysical properties of the myelin membrane.

1. Introduction

The myelin sheath comprises a unique lipid-rich and cholesterol-rich multilamellar structure formed by the extension of the oligodendrocyte membrane, which wraps tightly around an axon. Damage to the myelin sheath, resulting in demyelination, is a hallmark of multiple sclerosis, but it also occurs in other neurological disease settings. Following damage, remyelination can occur in which oligodendrocyte precursor cells (OPCs) differentiate into oligodendrocytes that form new myelin [1]. A hallmark of remyelination is that the new myelin has fewer wraps and is thinner than the original developmental myelin [2].

The importance of lipids in maintaining healthy myelin is established by our knowledge of genetic leukodystrophies including Krabbe disease and X-linked adrenoleukodystrophy in which accumulation of certain lipids leads to myelin damage and dysfunction [3]. Lipids are a major component of the myelin membrane; myelin has a higher percentage of lipids (relative to protein) than a typical plasma membrane (~75% vs. ~40%) and a higher percentage of the lipids are cholesterol (~40% vs. ~20%) [4,5]. Cholesterol can have a large impact on the biophysical properties of a lipid bilayer, and it has been demonstrated in model lipid monolayers that a cholesterol percentage of 44% combined with other myelin lipids has optimum physicochemical properties including membrane fluidity and binding with myelin basic protein, which is a critical protein component regulating myelin compaction [6].

Although the lipid composition of myelin was originally determined 60 years ago [4,7], there have been only a handful of studies that have measured myelin lipid levels using modern mass spectrometry [8–10], and little is known about how the myelin lipid composition changes after demyelination. More importantly, to our knowledge, the lipid composition of the new myelin formed during remyelination has not been quantified from model systems or from human samples. While it is accepted that lipids play a role in the membrane structure, currently there is no information on how the structural integrity of the remyelinated myelin membrane differs from healthy controls. Monomolecular films have been previously used as a model of an individual leaflet of the myelin membrane to characterize the lateral organization and structural properties of myelin sheath [13,14]. However, the surface behavior of remyelinated myelin membranes is currently unknown.

In this study, we use lipid mass spectrometry to measure the remyelinated lipidome, using a mouse model that undergoes near complete demyelination and remyelination in the brain [11,12]. We demonstrate that the remyelinated lipidome is significantly altered with major shifts in the relative concentrations of cholesterol, phosphatidylcholine (PC), and phosphatidylethanolamine (PE). These shifts led us to hypothesize that the altered lipid composition would have profound effects on the biophysical properties of the membrane. To test this hypothesis, we characterize the surface behavior of monomolecular lipid films prepared from diseased and healthy lipid extracts from the mouse model [11]. We measured surface pressure changes when subjected to compressional forces and membrane compressibility using a Langmuir-Pockels trough. Our results show conclusive evidence of alterations in the lateral organization of the lipid molecules and film compressibility in the myelin lipid extracts from demyelinated mice when compared to healthy mice. We then prepared model membranes containing the high abundance lipids, with compositions based on our lipidomics data. We observed that reduced cholesterol, but not phospholipids, was a major contributor to the observed alteration in the surface properties of these model membranes and remyelinated lipid extracts. Taken together, our results indicate that the reduced cholesterol levels in the remyelinated myelin lipidome have a large effect on membrane packing and compressibility.

2. Materials and Methods

2.1. Materials

HPLC-grade organic solvents used for equipment cleaning and sample preparation (chloroform, acetone, and isopropanol), reagent-grade HEPES and NaCl were purchased from Fischer Scientific (Hampton, NH). Reagent grade ETDA (Ethylenediamine-tetraacetic acid), NaOH, and Cholesterol were purchased from Millipore Sigma (St. Louis, MO). Purified water was obtained from a Milli-Q system (EMD Millipore - Billerica, MA), with 18.2 MΩ.cm of resistivity at 25 °C. HEPES buffer (5 mM, pH 7.3) was prepared and filtered using 0.2 μm Nylon membranes (Merck Millipore Ltd., Tullagreen, IRL). Porcine brain PC (L-α-phosphatidylcholine), porcine brain PE (L-α-phosphatidylethanolamine), and porcine brain PS (L-α-phosphatidylserine) were purchased from Avanti Polar Lipids (Alabaster, AL). Stock solutions of brain PC and PE were received at 25 mg/mL in chloroform, and brain PS was received as a sodium salt. Phospholipid solutions were prepared by dilution or solubilization in chloroform or chloroform:methanol mixtures to obtain a final stock concentration of 1 mg/mL. Samples were sealed with Teflon tape, protected from light, and stored at −20 °C when not in use.

2.2. Animal Husbandry

The demyelination mouse model utilizes Plp1-iCKO-Myrf mice (Myrf fl/fl; Plp1-CreERT) which have a C57BL6/J background [12]. The Plp1-iCKO-Myrf mice were bred at the University of Kansas by crossing Myrf(fl/fl) mice [15] with Plp1-CreERT mice [16]. Mice were housed in a climate-controlled room (24 ± 1 °C) with a 12 h light / dark cycle (12h on, 12h off) with food and water ad libitum. Myrf loss and demyelination was induced at 8 weeks of age with intraperitoneal injections containing 2 mg of tamoxifen (100 μl of 20 mg/ml tamoxifen in corn oil) daily for 5 consecutive days. Cre positive mice, which experience Myrf loss and demyelination, represented the demyelinating “diseased” group. Cre negative mice, which do not lose Myrf and do not undergo demyelination, were used as controls in the experiment and also received tamoxifen injections. Both male and female mice were used, and the data was combined for analysis. The number of mice used in each experiment is indicated in the figure legend. Euthanasia was performed at 24 weeks after tamoxifen by carbon dioxide inhalation and cervical dislocation. Brain tissue was collected immediately, frozen on dry ice, and stored at −80 °C until processing. All experiments were approved by the Institutional Animal Care and Use Committee at the University of Kansas.

2.3. Myelin isolation

Sucrose gradient centrifugation, a previously validated method, was used for myelin isolation [17]. In brief, minced half brains were placed in a cold 7mL Dounce homogenizer (DWK Life Sciences) containing 0.3 M sucrose solution (0.3 M sucrose, 20 mM Tris Cl, 2 mM Na₂EDTA, 1 mM dithiothreitol, pH 7.45). Tissue was homogenized using five strokes of the loose pestle and seven strokes of the tight pestle. Homogenate was carefully layered over a 0.83 M sucrose solution (0.83 M sucrose, 20 mM Tris Cl, 2 mM Na₂EDTA, 1 mM dithiothreitol, pH 7.45). and ultracentrifuged for 30 minutes at 75,000 × g at 4°C. The layer of myelin formed at the interface was carefully collected and resuspended in Tris Cl buffer solution (20 mM Tris Cl, 2 mM Na2EDTA, 1 mM dithiothreitol, pH 7.45). The resuspended myelin solution was ultracentrifuged for 15 minutes at 75,000 × g at 4°C and the myelin pellet was collected. The collected myelin pellet was subjected to an osmotic shock by resuspending in Tris-Cl buffer and centrifuging for 15 minutes at 12,000 × g at 4°C. A second osmotic shock was provided by resuspending the pellet and centrifuging for 10 minutes at 12,000 × g at 4°C. The myelin pellet was again resuspended in 0.3 M sucrose solution and layered over a 0.83 M sucrose solution, and the subsequent steps were repeated to obtain an enriched myelin fraction from the final pellet. Finally, the enriched myelin was resuspended in 6 ml of Tris Cl buffer and an aliquot was removed for protein quantification. The remaining myelin was lyophilized to obtain a crude myelin fraction. Protein levels were quantified in duplicate for each sample using a BCA assay.

2.4. Lipid mass spectrometry

For the lipid mass spectrometry experiments, myelin lipids were extracted by dissolving the crude myelin fraction in 1 ml of chloroform: methanol: water (35:65:3.5), followed by vortexing (Multitube vortexer, Fisher brand) for 2 hours in 10-minute intervals. Samples were then centrifuged at 1300 rpm for 10 minutes, and the lower layer was collected and vacuum-dried.

Lipidomics was performed at the Kansas Lipidomic Research Center at Kansas State University using direct infusion triple quadrupole mass spectrometry on a Sciex 4000 QTrap. The dried lipid samples were mixed with internal standards and solvents, and phospholipid analysis was performed following published protocols [18]. Internal standards and parameters related to acquisition and data processing are indicated in the Supplementary Information. Neutral loss scans were used to measure LPE, PE, PA, PI, and PS, while precursor ion scans were used to measure LPC and PC. Internal standards from each type of lipid were added to the samples. Corrections were applied for the differences in signal between the internal standards and SPLASH Lipidomix (Avanti Polar Lipids). Internal standards for PC were used for both SM and PC, and a correction was applied to the signal of SPLASH SM versus the PC internal standard. The signal responses were normalized and reported as normalized mass spectral intensity. The intensity of 1 nmol of internal standard gives a signal response value of 1.

Cholesterol was measured using gas chromatography-mass (GC-MS) spectrometry [19]. The sample was prepared by dissolving the dried lipid extract in 1 mL of chloroform, adding 10 nmol of coprosterol as an internal standard, and drying under nitrogen. The spiked sample was reconstituted with 1 mL of 1:9 3 N potassium hydroxide:methanol and heated at 80°C for 1 h. After cooling to room temperature, 2 mL of optima-grade water and 0.25 mL of saturated sodium chloride were added. The solution was extracted 3 times with 2 mL of hexane. The hexane layers were combined, dried under nitrogen, and dissolved in 50 μL of pyridine. Derivatization was performed by adding 25 μL of N-trimethylsilyl-N-methyltrifluoroacetamide with trimethylchlorosilane and incubating at 50°C for 60 min.

GC-MS analysis was performed on an Agilent 6890N GC coupled to an Agilent 5975N quadrupole mass selective detector (with EI). The GC was fitted with a VF-5MS capillary column (inert 5% phenylmethyl column, length: 30 mm, internal diameter: 250 μm, film thickness: 0.25 μm). Helium was used as the carrier gas at a column flow rate of 1 mL/min. The MS source temperature was at 230°C, the front inlet was operating at 250°C, and the quadrupole temperature was at 150°C. The Agilent 7683 autosampler was used to inject 1 μL of the derivatized sample in the splitless mode. The GC temperature program had an initial temperature of 150 °C, held for 1 min, ramped 30 °C/min to 300°C, and then ramped 3°C/min to a final temperature of 315°C, and held for 4 min. The total run time was 15 min. The mass spectrometer was operated in the electron impact mode at 70 eV ionization energy. The data acquisition was in scan mode (m/z 50 to 650). The data were processed with Agilent Chemstation software.

Statistical analysis was performed by comparing the different lipid classes in healthy samples and diseased samples using multiple t-tests with the Holm-Šídák correction for multiple comparisons.

2.5. Myelin lipid extraction

For the biophysical studies, myelin lipids were extracted using a modified Bligh–Dyer protocol [20]. The lyophilized myelin pellet was dissolved in 1 mL of water and combined with a mixture of chloroform (containing 0.01% butylated hydroxytoluene, BHT):methanol:water (3:2:1) in glass tubes. After shaking and vortexing thoroughly, the mixture was centrifuged (Sorvall ST 40R Centrifuge, Thermo Fisher Scientific) at 1300 rpm for 10 min. The lower layer was carefully removed and saved in a glass tube. The remaining top layer was further extracted twice with 1.25 mL of chloroform with 0.01% BHT; the lower layers were carefully removed and combined. The combined lower layer was then washed with 300 μL of 1 M KCl followed by 300 μL of water and vacuum-dried completely (Savant SpeedVac SPD130DLX vacuum concentrator, Thermo Fisher Scientific, USA) to obtain the dried lipid extract from myelin. Before performing the biophysical studies with the lipid extracts, they were resuspended in chloroform to obtain a final concentration of 1 mg/ml.

2.6. Biophysical studies

Surface pressure measurements were conducted at room temperature (23 ± 2 °C) using a Teflon Langmuir-Pockels trough with a Wilhelmy paper plate (Biolin Scientific Inc. - Gothenburg, Sweden). The trough apparatus was equipped with movable Delrin barriers to compress the lipid film by changing the surface area from 7750 to 1650 mm². A typical experiment started with a cleaning procedure consisting of wiping the trough and barriers with isopropanol and acetone, followed by pouring water and carefully aspirating the material from the air-liquid interface repeated times. After cleaning, the trough surface pressure sensor was calibrated using MilliQ water to read a value of 72.8 mN/m in air and 0 mN/m in water. With the barriers at the outmost position, the trough was filled with 37 mL of HEPES buffer, and appropriate amounts of working lipid samples were carefully spread at the air-liquid interface using a Hamilton Syringe (Reno, NE). Twenty minutes were allowed for solvent evaporation after sample spreading. After solvent evaporation, the lipid film was compressed by approximating the barriers at a 5 mm/min rate, while surface pressure measurements were recorded every second. For each animal group, healthy or diseased, samples were obtained from three independent mice. Additionally, for each tested sample, animal-derived or model lipid membranes, three independent replicates were collected. The surface pressure measurements reported are average values with the experimental standard deviations (shaded area), and t-tests with unequal variances and a 95% confidence interval were applied to determine statistical significance between samples.

The two-dimensional compressional modulus (C_s⁻¹) was then calculated from the isothermal surface pressure change during compression. C_s⁻¹ represents the film’s ability to store energy under compressional force, and is defined as C_s⁻¹=-A(dπ/dA)_T, where π is the surface pressure and A is the area of the trough. Numerically, high C_s⁻¹ moduli are observed in stiff films, which present high resistance to lateral compression. On the contrary, fluid interfacial films typically present low C_s⁻¹ moduli. The compressional moduli values presented here were obtained by numerical differentiation of the isotherms using the software Origin 2024b. To avoid amplifying noise, the surface pressure raw data was smoothened using a Fast Fourier Transform (FFT) filter with a 10-point window.

3. Results

3.1. Lipidomic analysis of normal and diseased myelin

To determine how the myelin lipidome changes after demyelination and remyelination, lipidomic analysis was performed on brain tissue from a mouse model of demyelination (Plp1-iCKO-Myrf mice). The strain is based on the inducible, conditional knockout (iCKO) of Myrf (myelin regulatory factor). Myrf encodes a transcription factor required for maintenance of healthy myelin and conditional deletion of Myrf in oligodendrocytes leads to oligodendrocyte cell death and substantial myelin loss [11,12,15]. Use of the Plp1-CreERT strain in combination with Myrf(fl/fl) strain restricts Cre recombinase expression and resulting in Myrf ablation to mature oligodendrocytes upon injection with tamoxifen. Loss of Myrf results in oligodendrocyte cell death and gradual demyelination that peaks at ~8–10 weeks post-tamoxifen. Oligodendrocyte progenitor cells (OPCs) still retain Myrf and are available to proliferate, differentiate, and remyelinate, which occurs between 10 and 24 weeks post-tamoxifen.

Demyelination in Plp1-iCKO-Myrf mice was induced at 8 weeks of age with 5 days of tamoxifen injections. Brain tissues were harvested at 24 weeks post-tamoxifen after both demyelination and subsequent remyelination had occurred. We performed analysis only on brain tissue as it was recently demonstrated that spinal cords in Plp1-iCKO-Myrf mice show limited remyelination,[21] and thus, spinal cords would be less suitable tissue for defining the remyelinated lipidome. As controls, healthy brains were harvested from Cre negative mice, which did not undergo Myrf deletion and subsequent demyelination. The myelin fraction was isolated from all brains (remyelinated and healthy) using sucrose gradient centrifugation [17]. The myelin lipids were further isolated using an organic extraction and were then analyzed by mass spectrometry.

Mass spectrometry analysis indicated that the absolute levels of all lipid classes were reduced in the remyelinated samples relative to healthy, normal myelin, except for cholesterol esters, which remain unchanged. (Figure 1A–C and Table S1). The levels were normalized by protein content, so the reductions observed suggest a shift in the lipid to protein ratio of remyelinated myelin. Both male and female mice were used in the experiment, and statistical analysis suggested that there were no sex-related differences in the lipid levels in the remyelinated myelin (Figure S1 and Table S2).

Figure 1. Lipid levels were reduced across all major classes in diseased myelin lipids.

Total lipid levels are shown for 13 major classes of lipids (split into three groups, (A) High Abundance Lipids, (B) Medium Abundance Lipids, (C) Low Abundance Lipids) in healthy myelin lipid extracts (n = 4–5) and diseased myelin lipid extracts (n = 8), where each sample represents a different mouse. (D) The percent mole fraction for each lipid class was calculated for the healthy and diseased myelin lipid extracts and plotted as pie charts.

For all lipids except cholesterol, the levels for each fatty acid length variant were measured using MS/MS and summed to determine the total level for each class. Cholesterol was measured using GC-MS. Statistical significance was determined across all classes using t-tests corrected for multiple comparisons with the Holm-Sidak method (****P < 0.0001, ***P < 0.001, **P < 0.01).

We noticed that the absolute levels of cholesterol showed a 6-fold decrease as compared to the 2-fold decrease in the absolute levels of phosphatidylcholine (PC) and phosphatidylethanolamine (PE). To examine how the shifting lipid levels might affect the relative lipid composition of the myelin membrane, we analyzed the same data by plotting the percent mole fraction of each lipid class as a pie chart (Figure 1D). The percent mole fraction was calculated by summing all lipids measured in the lipidomics experiment and setting that as the total moles of lipids; then each class of lipids was determined as the percent fraction of that total. This revealed a major shift in the overall lipid composition of the myelin membrane. Most notably, the combined percentage of PC and PE increased from 38% to 60%, while the percentage of cholesterol dropped from 46% to 24%.

3.2. Characterizing the surface properties of normal and diseased myelin membranes

To gain insight into how the altered lipidome in the remyelinated myelin changes the lateral molecular packing, which in turn can have functional consequences, we studied the mechanical properties of monomolecular lipid films formed using the myelin lipid extracts from the Plp1-iCKO-Myrf (diseased) mice and compared these measurements to healthy controls. The monomolecular films represent a single leaflet of the myelin sheath and can be subjected to compression, by using a barrier that modifies the total area available for lipid packing. Figure 2A shows the surface pressure vs. area per lipid isotherms of the monomolecular films of myelin lipids obtained from 3 mice in each category (healthy and diseased), during the compression cycle. Each measurement was repeated three times. The isotherms shown in Figure 2A represent an average of the three experiments for each lipid extract obtained from the different individual mice, and the error bars represent the error in measurement for each lipid extract. Figure 2A shows a complex series of responses that progressively change with compression of the lipid film (i.e. decreasing area per lipid). For the healthy myelin extract, initially, at large area/lipid molecule (3.25 cm²/μg), the surface pressure value is low (healthy mouse 3) or negligible (healthy mice 1 and 2), corresponding to a gas phase (zero surface pressure) or liquid-expanded phase [22]. As the film is compressed, a sharp increase in surface pressure is recorded with decreasing area, suggesting an increase in lateral interactions between the lipid molecules [22]. In contrast, for the remyelinated myelin lipid extracts (diseased model), our results show negligible surface pressure at area per lipid values where non-zero surface pressure was recorded for the healthy system (3.25 cm²/μg). In fact, the first non-zero surface pressure (“lift-off”) was recorded only when the packing density was significantly higher (between 0.4–0.75 cm²/μg), an over four-fold increase in the packing density.

Figure 2.

(A) Surface pressure vs. area per lipid isotherms for myelin lipid extracts obtained from healthy and diseased mice. The isotherms show that healthy lipid mixtures form monomolecular films that are more thermodynamically stable and demonstrate lateral interactions between the molecules forming the film. On the other hand, the diseased models show more compaction in the film and are not stable upon film compression. The continuous lines are averages of n=3 measurements, and the shaded areas are the standard deviations. (B) Compressional Modulus vs. area per lipid curves show that healthy lipids have more in-plane rigidity and higher resistance to deformation.

As we continue to compress the film, the progressive increase in surface pressure with decreasing area per lipid reaches a maximum value for the healthy controls, beyond which no further increase in surface pressure is seen, and a plateau appears at ~45 mN/m. For lipids extracted from the diseased mice, no such collapse pressure is attained.

Our results also show that while there is some variability between the isotherms from the three different mice in each group, overall, the behavior is reproducible. This variability between different animals within the same group is expected in animal studies. Further, the very tight error bars show the reproducibility of the data obtained with this measurement technique.

The surface pressure vs. area isotherms can also be used to calculate the area compressional moduli (C_s⁻¹) of a monomolecular film, which is a measure of the “in-plane elasticity”, and therefore the lateral packing state in the monolayer [23]. Figure 2B presents a comparison of the C_s⁻¹ values for the healthy and remyelinated myelin lipid extracts as a function of the surface area per lipid. For the healthy film, a sharp increase is recorded initially, followed by a more gradual increase until reaching area per lipid values that correspond to the monolayer collapse (below 1.5 cm²/μg). Beyond monolayer collapse, C_s⁻¹ drops sharply to near-zero values. This behavior is expected and corresponds to a phase change, as the film transitions from a well-packed 2-D lipid film to a 3-D structure.

For the remyelinated lipid extracts, the C_s⁻¹ curves show two distinct features, when compared with the healthy controls. First, we see a sharp increase in the C_s⁻¹, corresponding to a transition from the gas phase to the liquid-expanded phase (0.75 – 0.25 cm²/μg). Second, we do not observe the drop in compressibility modulus, corresponding to film collapse. Finally, and most importantly, the C_s⁻¹ values are found to be lower in magnitude for the diseased model, compared to the healthy lipid extracts, at all area per lipid values studied here.

3.3. Using model membranes designed to mimic healthy and diseased myelin to determine the contribution of different lipid classes on surface properties of myelin membranes

The experimental observations described above underscore the impact of the altered lipidome on the membrane mechanical properties, but do not provide information about the contribution of the different lipid classes. Therefore, to better understand the role of individual lipid classes, we designed model lipid mixtures containing the high abundance lipid classes (PC, PE, PS, and cholesterol) as these comprise >90% of the lipids measured in the lipidomic analysis in Figure 1.

We present the surface pressure vs. area isotherms for four different lipid mixtures, in Figure 3. Here the “healthy phospholipids” and “diseased phospholipids” represent the compositions of the phospholipid found in the healthy and diseased mice myelin extracts in our lipidomic studies, while 46% and 22% cholesterol represent the cholesterol content in the healthy and diseased mouse models. All phospholipids used here were obtained as total brain extracts from Avanti Polar lipids as these contain a mixture of lipids with differences in the acyl chain length and tail unsaturation. For example, brain-PC contained the most saturated lipids (16:0), while brain-PE has the largest content of unsaturated lipids, with a significant amount of polyunsaturated lipids (C22:4).

Figure 3:

(A) Surface pressure vs. area per lipid isotherms for model lipids films formed using the high abundance lipid classes found in the myelin extract. The isotherms show that the addition of 46%, representing the native cholesterol content, shows the highest surface pressure reached and the formation of the most stable films. These results also show that the models mimicking the phospholipid content in healthy and diseased animals do not show significant differences in their thermodynamic properties. The continuous lines are averages of n=3 measurements, and the shaded areas are the standard deviations. (B) Compressional Modulus vs. area per lipid curves show that the phospholipid content does not contribute significantly to the rigidity of the monolayer. The compressional modulus (ability to resist deformation) is the highest in films containing the native cholesterol amount.

As seen in Figure 3A, when comparing lipid mixtures containing only the three phospholipid classes (labeled “healthy brain” and “diseased brain” phospholipids), at high area per lipid values, both isotherms overlap. With decreasing surface area, the isotherms show a gradual increase in the surface pressure. Further, the two isotherms start to show a divergence at area per lipid values of 4 cm²/μg and lower, with the “healthy” mixture showing slightly larger surface pressure values, compared to the “diseased” set. This suggests formation of a more condensed monolayer in the “diseased” phospholipid mixture. Further, while both curves show monolayer collapse, the average collapse pressure for the model “healthy” mixture was found to be 34.9 $\pm$ 0.2 mN/m, compared to the disease model, with average collapse surfaces pressures of 32.5 $\pm$ 0.1 mN/m. A two-tailed t-test with unequal variances (Welch’s test) and 95% CI showed that this is a statistically significant difference (p value = 0.001). The details of this analysis is shown in Table S3.1–3.3.

Next, we consider isotherms that contain cholesterol, in addition to the phospholipids. The purple curve, containing 46% cholesterol and the green curve, containing 22% cholesterol, show that initially, the surface pressure is lower than the phospholipid-only mixtures. However, with compression, the film containing 46% cholesterol shows a more rapid increase in surface pressure, compared to all the other mixtures shown here. Further, the addition of 46% cholesterol leads to a very large increase in the collapse pressure, and the collapse pressure value is slightly higher than that observed for the healthy lipids (48 mN/m). Figure S2 in the Supporting Information presents our data where different cholesterol compositions were titrated into the phospholipid mixtures. As seen in this figure, the addition of 46% cholesterol showed the most significant increase in the maximum surface pressure reached before the films underwent monolayer collapse.

To quantify the response to applied compression, next we present the C_s⁻¹ for these four lipid mixtures, shown in Figure 3B. The phospholipid-only films representing the “healthy” and “diseased” models both show a sharp increase in the C_s⁻¹ values followed by a plateau value of 30 mN/m for both mixtures. This value is lower than that observed for the healthy animal lipid extract. In contrast, the addition of cholesterol leads to a lower C_s⁻¹ at the lowest packing densities, followed by a more gradual increase. While the film containing 22% cholesterol could only reach the same compressional modulus as the phospholipid films, the film containing 46% cholesterol demonstrated a sharp increase in C_s⁻¹ with increasing compression, reaching values of 80 mN/m at high packing densities. This maximum C_s⁻¹ value is even higher than that recorded for the healthy lipid extracts, shown in Figure 2.

Finally, given that the model mixtures containing 46% cholesterol showed the largest increase in the collapse surface pressure and maximum C_s⁻¹ value, we added cholesterol to the remyelinated lipid extract obtained from one of the genetically modified mice, to restore the cholesterol levels to that of the native system. The surface pressure vs. area isotherms and compressional moduli for this system are shown in Figure 4 and compared with a healthy model and a diseased model. Figure 4A shows that when additional cholesterol is added, the area per lipid value where “lift-off” occurs now shifts more towards the healthy model. Additionally, this film is able to resist collapse and reaches surface pressures higher than 40 mN/m. Further, the monolayer collapse occurs at area per lipid values similar to the healthy film (~ 1.2 cm²/μg). The collapse pressure is also similar in value to the healthy film, while the corresponding diseased film does not reach such high surface pressures or undergo monolayer collapse. When considering the area compressional moduli shown in Figure 4B, we find that initially this film shows no area compressibility modulus, followed by a sharp increase at the area per lipid values corresponding to lift-off. A kink in the C_s⁻¹ values is noted with compression, suggesting a phase transition that was previously not observed in either film. Finally, at a packing density corresponding to the monolayer collapse, the C_s⁻¹ values reached ~ 116 mN/m, which is much higher than even the values obtained in the film formed by the healthy myelin lipid extract.

Figure 4:

(A) Surface pressure vs. area per lipid isotherms for myelin extracts where cholesterol was added to the lipid mixture from the remyelinated myelin extract (green curve) compared to healthy and diseased models. The isotherms show that restoring the native cholesterol content in the remyelinated mixture shows a shift towards the healthy film, suggesting the role of cholesterol-lipid interactions in maintaining the in-plane thermodynamic stability. The continuous lines are averages of n=3 measurements, and the shaded areas are the standard deviations. (B) Compressional Modulus vs. area per lipid curves show that addition of cholesterol leads to a significant increase in the in-plane rigidity of the films, which in turn suggests formation of more stable films that are resistant to in-plane deformation.

4. Discussion

4.1. Lipidomic analysis reveals reduced cholesterol in diseased myelin

The lipidomic study presented here shows clear evidence of alteration in the lipid composition of the myelin lipid fraction extracted and enriched from the brain tissue of healthy vs. diseased mice. This study was enabled by the unique Plp1-iCKO-Myrf model that features near complete demyelination in the brain followed by remyelination [11,12]. Thus, the vast majority (>90%) of the isolated myelin was myelin that had been newly formed after demyelination. Specifically, our results show a lowering of all lipid classes (high, medium, and low abundance lipids) in the remyelinated myelin, suggesting that there are fewer myelin lipids, and therefore less myelin overall, after demyelination and remyelination. From our lipidomic studies, we conclude that after remyelination there is a shift in the relative abundance of cholesterol versus other phospholipids.

4.2. Remyelinated lipids from diseased models are less densely packed

Although recent studies have examined the myelin lipidome using mass spectrometry [8–10], we believe this is the first time that the “remyelinated” myelin lipidome has been measured. With this information in hand, we were able to determine how alterations in the remyelinated myelin lipidome lead to changes in the lateral organization of lipids and the mechanical properties of myelin. The surface pressure vs. area isotherms and the compressibility moduli data shown in Figure 2, taken together show significant differences between the packing of the healthy myelin lipids and the remyelinated lipids extracted from the diseased mouse model. The compression curves for the healthy brain lipids shown here are similar to surface pressure curves reported for healthy spinal cord myelin lipids [13], suggesting that the phase behavior of the brain and spinal cord is similar. Further, the collapse pressure seen in the myelin lipid extract discussed here is similar to the collapse pressure reported for lipids extracted from the white matter of marmoset control and EAE mice [24]. Monolayer collapse is common in insoluble monomolecular films and corresponds to the point where the monolayer can no longer resist applied compression, and “buckles” under the applied stress, as the molecular density increases [25]. At this point, the film transitions from a 2-D system to a 3-D system.

For the remyelinated lipids from the diseased mice, the observed shift in the lift-off to lower area per lipid values indicates a decrease in the lateral repulsion between the lipid molecules in the remyelinated lipid extract, or the formation of a more condensed film [23]. The lower surface pressures in the lipid films formed from the disease model further underscore the lack of long-range lateral interactions between the lipid molecules. Together, these observations can be attributed to the altered lipidome recorded for the remyelinated lipid extracts.

Our C_s⁻¹ data taken together suggest that the myelin films remain in the liquid-expanded phase throughout the compression cycle and the remyelinated membranes are more fluid than the healthy controls [23,26]. A significant shift in the area per lipid where the compressional moduli show a sharp increase also indicates that intermolecular interactions in the remyelinated lipid film are much lower than in the healthy control. We hypothesize that the decrease in the intermolecular interactions impacts the mechanical properties of the lipid films, which manifests as a lowering of the compressional moduli recorded for the remyelinated lipid films when compared to the healthy controls. In turn, we attribute this decrease in the film’s mechanical properties to the altered lipidome recorded for the remyelinated lipid extracts.

Biophysical studies reported to date have been performed on lipids extracted from the EAE mouse model, which has both demyelination and axonal degeneration, which limits the extent of remyelination [13,14,24,27,28]. Using the EAE model, Gasecka et al. have previously reported a lowering of the molecular ordering in the myelin sheath of the mouse spinal cord during demyelination [27]. On the other hand, using the white matter from the EAE model, only small differences in the surface pressure vs. area isotherms were reported. Our results using myelin extracts from mice that have undergone genetically induced demyelination, followed by robust remyelination, show not only an altered lipidome but that the myelin sheath is also structurally less stable (more fluid), due to a lack of molecular packing. This alteration in the lateral packing of the remyelinated sheath can contribute to disease progression in multiple sclerosis.

4.3. Cholesterol content plays a dominant role in the mechanical properties of myelin membranes

Previous studies using a EAE model have attributed observed changes in the surface pressure vs. area isotherms to changes in the acyl chain saturation, but not headgroup classes [24]. Our data using model mixtures containing different ratios of the brain PC, PE and PS headgroups with different composition of the acyl tails, allowed us to also assess if the differences in the acyl chains contributed to the significant difference seen in the lipid extracts from the animal models reported in Figure 2. Based on our results with the phospholipid only systems shown in Figure 3, we argue that the change in the content of the phospholipid headgroups or even the differences in the acyl chain length do not explain the difference observed in the animal models. On the other hand, addition of cholesterol did show a significant shift in the “lift-off” area to a higher packing density, particularly when 46% cholesterol was added. This suggests that the addition of cholesterol causes compaction in both films, with higher cholesterol content leading to more compaction. This cholesterol concentration also corresponds to the cholesterol content found in the healthy myelin extract in our lipidomic studies. Together, these isotherms suggest that the addition of cholesterol has a more significant impact on monolayer packing, leading to more compact films that resist deformation, while the films formed with phospholipid alone lead to more expanded, fluid films. In a recent study using lipid monolayers formed using brain lipid extracts and the myelin basic protein, Träger et al. showed that lipid monolayers composed of 44% cholesterol formed the most stable films in the presence of the myelin basic protein, suggesting that this cholesterol content also enables the most desired lipid-protein interaction [6]. While studying the lipid-protein interaction is beyond the scope of this study, we note that the cholesterol content reported in reference [6] was found to be similar to that seen here. The compressional moduli data also confirms that the addition of cholesterol leads to compaction in the film at lower packing densities, as well as being able to form a stable film at higher packing densities.

Finally, we were able to demonstrate that supplementing cholesterol in the diseased lipid extract could significantly restore the mechanical properties (Figure 4). Specifically, addition of cholesterol to the lipid extract from the diseased mouse model shows a shift in the “lift-off” to lower packing densities, as well as reappearance of the collapse plateau, similar to the healthy controls, even though the slope of the isotherm is not the same. Similarly, the addition of cholesterol to the lipid extract from the diseased model led to a compressional modulus that is even higher than the healthy controls. Therefore, we conclude that the observed differences in the biophysical properties of myelin extracts from the healthy and diseased animal models may be explained in part by differences in the cholesterol content. To fully reproduce the healthy control curve in future research, we will need to consider the potential role of other minor lipid classes, particularly sphingolipids and plasmalogens, and the exact fatty acyl composition of myelin lipids, that are beyond the scope of this study.

In this study, we discovered that remyelinated myelin in a mouse model has reduced cholesterol relative to other lipid species and we used biophysical studies to demonstrate that the lack of cholesterol profoundly affects lipid packing and membrane stability. Local cholesterol synthesis in the brain is required for successful remyelination in mouse models [29,30]. Our findings provide a possible biophysical mechanism for why remyelination is impaired when cholesterol availability is limited.[31] Reduced cholesterol content could decrease the stability and compressibility of the myelin membrane, and thus the myelin may produce fewer concentric wraps making the myelin “thinner” [29,30]. An intriguing implication is that availability of cholesterol could drive myelin thickness, suggesting that increasing cholesterol levels as a therapeutic strategy may result in more stable myelin membranes with more wraps and thicker remyelination. This mechanism is supported by previous work showing that dietary supplementation with cholesterol can increase myelin thickness in the lysolecithin mouse model of demyelination [32].

5. Conclusion

Our lipidomic data show that the composition of the remyelinated lipid is significantly different than the healthy controls with a large relative reduction in cholesterol content. Further, our biophysical studies show that this difference in the lipidome also leads to differences in the membrane mechanical properties, due to a reduction in the lateral organization of the lipids in the two-dimensional film. Further, these results also demonstrate that the decrease in the cholesterol content in the remyelinated myelin extract plays a dominant role in controlling the mechanical properties and thermodynamic stability of the lipid film, by controlling the in-plane rigidity of the monomolecular films.

Data availability

All data will be made available on request.