Exploiting Sphingo- and Glycerophospholipid Impairment to Select Effective Drugs and Biomarkers for CMT1A

Abstract

In Charcot-Marie-Tooth type 1A (CMT1A), Schwann cells exhibit a preponderant transcriptional deficiency of genes involved in lipid biosynthesis. This perturbed lipid metabolism affects the peripheral nerve physiology and the structure of peripheral myelin. Nevertheless, the identification and functional characterization of the lipid species mainly responsible for CMT1A myelin impairment currently lack. This is critical in the pathogenesis of the neuropathy since lipids are many and complex molecules which play essential roles in the cell, including the structural components of cellular membranes, cell signaling, and membrane trafficking. Moreover, lipids themselves are able to modify gene transcription, thereby affecting the genotype-phenotype correlation of well-defined inherited diseases, including CMT1A. Here we report for the first time a comprehensive lipid profiling in experimental and human CMT1A, demonstrating a previously unknown specific alteration of sphingolipid (SP) and glycerophospholipid (GP) metabolism. Notably, SP, and GP changes even emerge in biological fluids of CMT1A rat and human patients, implying a systemic metabolic dysfunction for these specific lipid classes. Actually, SP and GP are not merely reduced; their expression is instead aberrant, contributing to the ultrastructural abnormalities that we detailed by X-ray diffraction in rat and human internode myelin. The modulation of SP and GP pathways in myelinating dorsal root ganglia cultures clearly sustains this issue. In fact, just selected molecules interacting with these pathways are able to modify the altered geometric parameters of CMT1A myelinated fibers. Overall, we propose to exploit the present SP and GP metabolism impairment to select effective drugs and validate a set of reliable biomarkers, which remain a challenge in CMT1A neuropathy.

Keywords: CMT1A; Schwann cell; biomarker; demyelination; drug; lipid metabolism; myelin; peripheral neuropathy.

Figure 1

Multivariate data analysis (MVA) of rat sciatic nerve lipidome clearly discriminates CMT1A from controls displaying a specific impairment of sphingolipid and glycerophospholipid metabolism. (A) Score plot from principal component analysis (PCA) of untargeted lipidomics data. PCA was able to reliably discriminate CMT1A nerves (red, n = 11) from the wild type ones (black, n = 8). (B) Heatmap generated with the most significant features (with highest fold change and statistical significance) detected by MVA. One hundred twenty-one features were mainly responsible for the difference between the two groups. (C) Graph presenting the probability (y-axis) and the impact (x-axis) that a pathway is responsible for the difference shown in lipidomic profiles. Each circle represents a specific lipid pathway; the circle size represents the number of hits per pathway. Red–orange–yellow–white diminishing scale represents the degree of involvement in lipidomic profiles. (D) Table indicating only the significant pathways presented in the graph.

Figure 2

CMT1A myelin displayed perturbed sphingolipid (SP) and glycerophospholipid composition and ultrastructural abnormalities. The most abundant SP species were analyzed by an optimized protocol of targeted mass spectrometry. This method uses 25 standards to calculate the absolute concentration (nM) of sphingolipids of interest. (A–D) Ceramides, sphingomyelins, hexosylceramides, and sphingosine were all reduced in the CMT1A samples. (E,F) Phosphatidylinositides were the lipid species mainly altered in CMT1A myelin by shotgun untargeted analysis. The compound names are presented on the y-axis, while different acyl chains, sature, or insature, are presented on the x-axis. Data are presented as mean ± mean of standard error. Wild type (WT) (black), CMT1A (red) n = 4. Statistics was calculated with unpaired t-test, two-tailed. ****p < 0.0001, ***p < 0.001, **p < 0.01, and *p < 0.05. (G) X-ray diffraction performed on WT and CMT1A rat sciatic nerves from four different litters at 1.5 months (filled symbols) and at 3 months (open symbols) of age. The whole nerves were fixed in 2.5% glutaraldehyde in cacodylate buffer, pH 7.4, for 24 h. In the graph, the relative amount of myelin vs. myelin period is reported. The CMT1A rats showed a significantly reduced myelin content and an enlarged myelin periodicity compared to the WT littermates, which demonstrates the presence of myelin ultrastructural alterations in this neuropathy (WT black, n = 12; CMT1A red, n = 9, unpaired t-test, two-tailed; p < 0.0001 for both relative amount of myelin and myelin periodicity). The inset shows examples of the diffraction patterns, expressed as diffracted intensity vs. detector channel number. The patterns correspond to the data points marked by the asterisks. The small shift in the positions of the Bragg peaks indicates differences in periodicity, and the weaker peaks indicate less myelin. The middle region of each pattern, approximate channel numbers 950–1,050, is central scatter from the direct beam around the beam stop and, therefore, is excluded from the analysis. (H) X-ray diffraction performed on human sural nerve biopsies of patients affected by CMT1A compared to patients affected by other neurological diseases (OND). The nerves were fixed in buffered glutaraldehyde, processed for electron microscopy, and embedded in Epon, which accounts for the differences in periodicities with the results in (G). Notably, the CMT1A patients (red, n = 3) displayed enlarged myelin periodicity compared to the control patients (OND, black, n = 4, unpaired t-test, two-tailed; p < 0.01) as was found in the CMT1A rat. Data are presented as mean ± standard deviation.

Figure 3

Modulation of sphingolipid (SP) and glycerophospholipid (GP) pathways affects the CMT1A myelin physical structure. (A) Schematic illustration of treatment schedule adopted to demonstrate the specific involvement of SP and GP in CMT1A myelinopathy. The CMT1A and WT DRG cultures were chronically treated with several molecules selected for their proven efficacy on SP and GP pathway modulation in the presence of 15% newborn calf serum, ascorbic acid (100 μg/ml final concentration), and nerve growth factor at 5 ng/ml final concentration. In particular, we analyzed the effects on myelination of LPA, PA, CDP-choline, PIP3, VO-OHpic, 2OHOA, desipramine, SM, and L-serine. (B) Advanced neuropathology (see also the Video S1) performed on dorsal root ganglia (DRG) myelinated fibers allowed us to select PIP3, LPA, VO-OHpic, and L-serine as the most effective molecules. Interestingly, we found that these molecules improved the CMT1A myelinopathy—a reduced amount of myelinated fibers and shortening of the internode length—in a different way. In fact, while PIP3 and LPA significantly increased the amount of myelinated fibers without any effect on their structure [CMT1A Ctrl (n = 180) vs. CMT1A PIP3 (n = 215), mean ± SD: 0.53 ± 0.43 vs. 0.75 ± 0.50; CMT1A Ctrl (n = 405) vs. CMT1A LPA (n = 511), mean ± SD: 0.35 ± 0.27 vs. 0.58 ± 0.40), L-serine and VO-OHpic just increased the internode length (CMT1A Ctrl (n = 102) vs. CMT1A L-serine (n = 113), mean ± SD: 0.79 ± 0.12 vs. 0.85 ± 0.23; CMT1A Ctrl (n = 128) vs. CMT1A VO-OHpic (n = 155), mean ± SD: 0.67 ± 0.11 vs. 0.75 ± 0.11], suggesting the existence of at least two independent mechanisms essential for correct myelination. This hypothesis is further strengthened by the cumulative effect of PIP3 and L-serine simultaneous administration to CMT1A DRG cultures, which was able to improve both the quantity and the quality of pathological myelin [CMT1A Ctrl (n = 112) vs. CMT1A PIP3+L-serine (n = 145), myelinated area mean ± SD: 0.27 ± 0.20 vs. 0.46 ± 0.36; CMT1A Ctrl (n =112) vs. CMT1A PIP3+L-serine (n = 145), internode length mean ± SD: 0.62 ± 0.10 vs. 0.69 ± 0.12]. LPA, lysophosphatidic acid; PA, phosphatidic acid; DAG, diacylglycerol; CDP-choline, cytidine-5′-diphospho-choline; PI, phosphoinositide; PC, phosphatidylcholine; PIP, phosphatidylinositol phosphate; PTEN, phosphatase and tensin homolog; VO-OHpic, a PTEN inhibitor; PI3K, phosphatidylinositol-3-kinase; PIP3, phosphatidylinositol tris-3,4,5-phosphate; 2OHOA, 2-hydroxy oleic acid; SM, sphingomyelin; aSMase, acid sphingomyelinase; SMS, sphingomyelin synthase; Cer, ceramide. For myelinated area and internode length, we performed the D'Agostino–Pearson normality test to assess the type of data distribution. Therefore, statistical differences were determined using the non-parametric Mann–Whitney test (n represents the number of analyzed images in at least three biological replicates). For internode length frequency distributions, we performed non-parametric Kruskal–Wallis test followed by Dunn's multiple-comparisons test (n represents the total number of analyzed internodes). ns, not significant; **p < 0.01 and ****p < 0.0001.

Figure 4

Sphingolipid (SP) and glycerophospholipid (GP) pathways present a reliable source of CMT1A wet biomarkers. (A) Score plot from Orthogonal Projections to Latent Structures Discriminant Analysis (OPLS-DA) of untargeted lipidomics data. (A) OPLS-DA analysis was able to reliably discriminate CMT1A CSF (red, n = 9) from the wild type (WT) one (black, n = 8). (B) Corresponding OPLS-DA analysis of rat serum lipidome displayed a clear separation between the two phenotypes (WT, black, n = 5 and CMT1A, red, n = 5). (C) Untargeted lipidomics was also performed on the serum of 15 healthy donors and 28 CMT1A patients. Notably, OPLS-DA analysis demonstrated a clear clustering of the subjects into two groups according to the genotype. (D–F) Graphs presenting the probability (y-axis) and the impact (x-axis) that a pathway is responsible for the differences shown in lipidomic profiles. Each circle represents a specific lipid pathway; the circle size represents the number of hits per pathway. Red–orange–yellow–white diminishing scale represents the degree of involvement in lipidomic profiles. (G) Table indicating only significant pathways presented in the graphs. (H) A comparative pathway analysis (Venn diagram) shows that the dysregulation of SP and GP metabolism is mainly responsible for the difference between CMT1A and controls in both rat biofluids and the serum of human subjects. The diagram was designed by Biovenn online software (http://www.biovenn.nl/).