Ectopic expression of ceramide synthase 2 in neurons suppresses neurodegeneration induced by ceramide synthase 1 deficiency

Abstract

Sphingolipids exhibit extreme functional and chemical diversity that is in part determined by their hydrophobic moiety, ceramide. In mammals, the fatty acyl chain length variation of ceramides is determined by six (dihydro)ceramide synthase (CerS) isoforms. Previously, we and others showed that mutations in the major neuron-specific CerS1, which synthesizes 18-carbon fatty acyl (C18) ceramide, cause elevation of long-chain base (LCB) substrates and decrease in C18 ceramide and derivatives in the brain, leading to neurodegeneration in mice and myoclonus epilepsy with dementia in humans. Whether LCB elevation or C18 ceramide reduction leads to neurodegeneration is unclear. Here, we ectopically expressed CerS2, a nonneuronal CerS producing C22-C24 ceramides, in neurons of Cers1-deficient mice. Surprisingly, the Cers1 mutant pathology was almost completely suppressed. Because CerS2 cannot replenish C18 ceramide, the rescue is likely a result of LCB reduction. Consistent with this hypothesis, we found that only LCBs, the substrates common for all of the CerS isoforms, but not ceramides and complex sphingolipids, were restored to the wild-type levels in the Cers2-rescued Cers1 mutant mouse brains. Furthermore, LCBs induced neurite fragmentation in cultured neurons at concentrations corresponding to the elevated levels in the CerS1-deficient brain. The strong association of LCB levels with neuronal survival both in vivo and in vitro suggests high-level accumulation of LCBs is a possible underlying cause of the CerS1 deficiency-induced neuronal death.

Keywords: ceramide; ceramide synthase; long-chain base; neurodegeneration; sphingolipid.

Fig. 1.

Cerebellar neuronal expression of CerS2 from a transgene. (A and B) Real-time PCR results showing changes of the expression of Cers2 (A) and Cers1 (B) transcripts in brains of 4-wk-old Cers1 wild-type (+/+), Cers1 wild-type with the neuron-specific Cers2 transgene (+/+ Tg), homozygous Cers1 toppler mutant (to/to), and Cers1 toppler mutant with the Cers2 transgene (to/to Tg). Values are mean ± SD (three mouse samples, four technical replicates for each sample). **P ≤ 0.01 (one-way ANOVA, multiple comparisons). (C) A representative Western blot analysis showing the CerS2 protein level in the brain. Genotypes of mice are as described in A and B. The blot was redeveloped with an antibody against α-tubulin as a loading control. (D–l) Immunohistochemistry of brain sections from 4-wk-old wild-type (D–I, +/+) and Tg-CerS2 (J–L, +/+ Tg) mice, using antibodies against CerS2 (D, G, and J), APC (E), and IP3R (H and K). Merged images are also shown (F, I, and L). Purkinje cells and oligodendrocytes are marked with large and small arrowheads, respectively. Cerebellar cell layers are labeled as GCL (granule cell layer), WM (white matter), ML (molecular layer), or PCL (Purkinje cell layer). (Scale bar, 50 µm.)

Fig. S1.

Cerebral neuronal expression of CerS2 from a transgene. (A–F) CerS2 expression in the hippocampa (areas enclosed by dashed lines) from the wild-type (A–C, +/+) and Tg-CerS2 (D–F, +/+ Tg) mice, using antibodies against CerS2 (A and D) and NeuN (B and E). Merged images were shown (C and F). (G–L) CerS2 expression in the cerebral cortex from the wild-type (G–I, +/+) and Tg-CerS2 (J–L, +/+ Tg) mice using antibodies against CerS2 (G and J) and NeuN (H and K). Cortical layers (I–IV) are delineated by dashed lines in merged images (I and L). (Scale bar, 100 µm.)

Fig. 2.

Neuronal expression of Cers2 suppresses Purkinje cell death caused by Cers1-deficiency. (A–F) Calbindin D-28K immunohistochemistry of cerebella from 10-mo-old wild-type (A), Cers1^to/to (B), and Cers1^to/to;Tg-Cers2 (C) mice. Lobule X of each cerebellum was shown in detail (D–F). Sections are close to midline. [Scale bars, 1 mm (A–C) or 100 µm (D–F).] (G–I) Cresyl Violet staining of cerebella from 10-mo-old wild-type (G), Cers1^to/to (H), and Cers1^to/to;Tg-Cers2 (I) mice. Detail in Lobule X was shown. Arrows point to Purkinje cells. (Scale bar, 100 µm.)

Fig. 3.

Neuronal expression of Cers2 suppresses accumulation of ubiquitin-positive deposits caused by Cers1 deficiency. Immunohistochemistry showing ubiquitin (A, D, and G), NeuN (B, E, and H) staining, or merge images (C, F, and I) of hippocampal CA3 region of 15-mo-old wild-type (A–C, +/+), Cers1^to/to (D–F, to/to), and Cers1^to/to;Tg-Cers2 (G–I, to/to Tg) mice. Two cells still harboring ubiquitin-positive deposits in the to/to Tg-Cers2 hippocampus are marked with arrowheads (G and I). Note the reduction of NeuN staining in the to/to hippocampus. (Scale bars, 50 µm.)

Fig. 4.

Ceremide profiles in the brain with neuronal expression of Cers2. (A) Simplified schematic diagram of the sphingolipid biosynthesis pathway in mammal, showing key metabolites such as dihydrosphingosine (dhSph), dihydroceramide (dhCer), ceramide (Cer), sphingosine (Sph), and sphingosine-1-phosphate (S1P), as well as a few enzymes involved including ceramide synthase 1–6 (CerS1–CerS6) and ceramidases (CDases). (B and C) Brain ceramide levels of 40-d-old Cers1 wild-type (+/+, unfilled bars), homozygous Cers1^to/to mutant (to/to, filled bars), and Cers1^to/to;Tg-Cers2 (to/to Tg, hatched bars) mice. Profiles of total amounts of ceramides (B) and individual ceramides species (C) are shown. Three mice (n = 3) of each genotype were used. Values are mean ± SD. *P ≤ 0.05; **P ≤ 0.01 (one-way ANOVA, multiple comparisons).

Fig. S2.

Sphingomyelin and simple glycosphingolipid profiles in the brain with neuronal expression of Cers2. (A) Sphingomyelin species and total amount of sphingomyelin (Inset). (B) Hexosylceramide species and total amount of hexosylceramide (Inset). (C) Lactosylceramide species of 40-d-old Cers1 wild-type (+/+, unfilled bars), homozygous Cers1^to/to mutant (filled bars), and Cers1^to/to;Tg-CerS2 (hatched bars) mice. Three mice (n = 3) of each genotype were used. Values are mean ± SD. *P ≤ 0.05; **P ≤ 0.01 (one-way ANOVA, multiple comparisons).

Fig. S3.

Gangliosides profiles in the brain with neuronal expression of Cers2. (A) TLC of gangliosides purified from the brains of 40-d-old wild-type (+/+), homozygous Cers1^to/to mutant (to/to), and Cers1^to/to;Tg-CerS2 (to/to Tg) mice. (B–D) Quantifications of GM1 (B), GD1a (C), and GT1b (D) gangliosides on the TLC. Three mice (n = 3) of each genotype were used. Values are mean ± SD. *P ≤ 0.05 (one-way ANOVA, multiple comparisons).

Fig. 5.

Sphingolipid metabolite profiles in the brain with neuronal expression of Cers2. (A–E) Free LCB levels in the brain, including dihydrosphingosine (dhSph; A), sphingosine (Sph, B), dihydrosphingosine-1-phosphate (dhS1P; C), sphingosine-1-phosphate (S1P; D), 1-deoxy-sphinganine (DeoSa; E), and 1-deoxy-sphingosine (DeoSph; F). (G) Total LCB levels, including free LCBs and LCB moieties in brain sphingolipids. (H) SPT activity of brain microsomes using 20 µM C16-CoA substrate. Four mice (n = 4) of wild type (+/+) and three mice (n = 3) of CerS1^to/to or CerS1^to/to;Tg-Cers2 were used. Values are mean ± SD. *P ≤ 0.05; **P ≤ 0.01 (one-way ANOVA, multiple comparisons).

Fig. 6.

Toxicity of LCBs to cultured neurons. (A–L). Dihydrosphingosine (dhSph) treatment of cortical neurons. Cortical neurons from E18 embryos were untreated (A–D), or treated with 1 µM (E–H) or 15 µM (I–L) dhSph. (M–P) Deoxysphinganine (DeoSa) treatment of cortical neurons. Neurons were treated with 0.15 µM DeoSa. An area is enlarged in the Inset in P to show axon fragmentation at its early stage (arrowheads). The rationale for using these dhSph and deoSa concentrations was explained in detail in the Results. Neurons, after fixation, were incubated with an antibody against neuron-specific class III β-tubulin (A, E, I, and M; Tuj1), and then subjected to TUNEL assay (B, F, J, and N) and counterstained with DAPI (C, G, K, and O). Merged images were shown (D, H, L, and P). (Scale bar, 20 µm.)