Astrocyte-derived clusterin disrupts glial physiology to obstruct remyelination in mouse models of demyelinating diseases

Abstract

Multiple sclerosis (MS) is a debilitating demyelinating disease characterized by remyelination failure attributed to inadequate oligodendrocyte precursor cells (OPCs) differentiation and aberrant astrogliosis. A comprehensive cell atlas reanalysis of clinical specimens brings to light heightened clusterin (CLU) expression in a specific astrocyte subtype links to active lesions in MS patients. Our investigation reveals elevated astrocytic CLU levels in both active lesions of patient tissues and female murine MS models. CLU administration stimulates primary astrocyte proliferation while concurrently impeding astrocyte-mediated clearance of myelin debris. Intriguingly, CLU overload directly impedes OPC differentiation and induces OPCs and OLs apoptosis. Mechanistically, CLU suppresses PI3K-AKT signaling in primary OPCs via very low-density lipoprotein receptor. Pharmacological activation of AKT rescues the damage inflicted by excess CLU on OPCs and ameliorates demyelination in the corpus callosum. Furthermore, conditional knockout of CLU emerges as a promising intervention, showcasing improved remyelination processes and reduced severity in murine MS models.

Fig. 1. CLU is upregulated by astrocytes in MS as well as murine demyelinating lesions.

A, B Representative micrographs depicting immunohistochemical staining of CLU and GFAP within the white matter lesions of MS and NMS patients (n = 1 participant/group, p < 0.0001). C, D Immunohistochemical staining revealing elevated CLU levels in the astrocyte within the spinal cord of EAE (n = 3 mice/group, p = 0.0001). E, F Western blot analysis demonstrates a progressive increase in CLU levels correlated with disease severity in the spinal cord of EAE mice (n = 3 mice/group, E1 vs. Ctrl p = 0.4528; E2 vs. Ctrl p < 0.0001; E3 vs. Ctrl p < 0.0001; E4 vs. Ctrl p = 0.0001). G, H Increased CLU levels are observed in LPC-induced demyelinated foci (n > 5 mice/group, Dpi 5 vs. dapi 0 p < 0.0001, Dpi 10 vs. dapi 0 p < 0.0001, Dpi 20 vs. dapi 0 p < 0.0001). I–L In cuprizone-induced demyelinated foci, CLU levels remain unchanged, as evidenced by immunohistochemical staining (n = 3 mice/group, p = 0.6358) and western blot (n = 4 mice/group, p = 0.9852). For immunofluorescence pictures: Scale bar: 50 μm. Statistical analysis was performed using unpaired two-tailed Student’s t-tests, mean ± SD, *p < 0.05, **p < 0.01, ***p < 0.001. ns. = p not significant. NMS non-multiple sclerosis, MS multiple sclerosis, Ctrl control, EAE experimental autoimmune encephalomyelitis, E clinical score for EAE, LPC lysolecithin.

Fig. 2. Impact of CLU exposure on the transcriptome of OPCs.

A Schematic diagram outlining the protocols for RNA sequencing of OPCs treated with CLU. B GSEA revealed the toxic effects of CLU on OPCs after 3 h of treatment (n = 3 mice/group). C Volcano plot displaying DEGs in OPCs treated with CLU for 24 h (n = 3 mice/group). D–F Gene Ontology annotation using downregulated genes highlighted the enrichment of biological processes (BPs) relevant to OL differentiation, myelination, and axon ensheathment after 24 h of treatment with CLU at 200 ng/mL (n = 3 mice/group). G Gene Ontology annotation using upregulated DEGs revealed several GOBPs associated with DNA damage (n = 3 mice/group).

Fig. 3. Impact of CLU on OPC differentiation and apoptosis in vitro.

A Schematic diagram illustrating OPC differentiation, proliferation, and apoptosis assays treated with CLU. B, C CLU exhibited a dose-dependent decrease in the number of MBP⁺ cells in the OPC differentiation assay (n = 3 technical replicates/group, 50 ng/mL vs. 0 ng/mL p = 0.0141; 100 ng/mL vs. 0 ng/mL p = 0.0032; 200 ng/mL vs. 0 ng/mL p = 0.0013; 400 ng/mL vs. 0 ng/mL p = 0.0001). D, E In the OPC proliferation assay, CLU dose-dependently reduced the number of EdU⁺Olig2⁺ cells (n = 3 technical replicates/group, 50 ng/mL vs. 0 ng/mL p = 0.0035; 100 ng/mL vs. 0 ng/mL p = 0.0025; 200 ng/mL vs. 0 ng/mL p = 0.0008; 400 ng/mL vs. 0 ng/mL p = 0.0005). F, G CLU dose-dependently increased OPC apoptosis, as detected by cleaved caspase-3 immunohistochemical staining (n = 3 technical replicates/group, 50 ng/mL vs. 0 ng/mL p = 0.0005; 100 ng/mL vs. 0 ng/mL p = 0.0005; 200 ng/mL vs. 0 ng/mL p < 0.0001; 400 ng/mL vs. 0 ng/mL p < 0.0001). H, I CLU dose-dependently increased OPC apoptosis, confirmed by CC3 expression via western blot (n = 3 technical replicates/group, 1 h vs. 0 h p = 0.9447; 3 h vs. 0 h p = 0.4328; 6 h vs. 0 h p = 0.2383; 12 h vs. 0 h p = 0.0017; 24 h vs. 0 h p = 0.0014). J, K CLU inhibited the ability of OPCs to form a myelin sheath in cerebellar organic slice culture (n ≥ 3 technical replicates/group, p = 0.0033). L–N Numbers of Olig2⁺ OPCs and CC1⁺ OLs in the cerebellar slices were reduced after CLU treatment (n = 5 technical replicates/group, p < 0.0001; p = 0.0003). For immunofluorescence pictures: Scale bar: 50 μm. Statistical analysis was performed using unpaired two-tailed Student’s t-tests, mean ± SD, *p < 0.05, **p < 0.01, ***p < 0.001. ns. = p not significant. CC3 cleaved capase-3.

Fig. 4. Impact of elevated CLU on OPC and myelin loss in the corpus callosum.

A Schematic representation of NG2-CreERT2:: tau-mGFP reporter mice construction and experimental arrangement. B–D CLU administration into the corpus callosum for 7 days via implantable osmotic pumps significantly reduced the length of new myelin and the newly myelinated area, compared to DPBS vehicle control mice (n = 4 mice/group, p = 0.0107; p > 0.0001. Three independent experiments were conducted). E, F TEM revealed thinner myelin sheaths in CLU-administered mice compared to DPBS-administered mice. G No significant difference in axon diameter was observed between CLU-administered mice and DPBS-administered mice (n = 3 mice/group, p = 0.5586). H Schematic of viral constructs. I–K CLU upregulation by AAV significantly reduced the number of NG2⁺ OPCs and increased CC3⁺ OPCs (n = 4 mice/group, p = 0.0007; p < 0.0001). L–N CLU upregulation by AAV significantly reduced the number of CC1+ OLs and increased CC3+ OLs (n = 4 mice/group, p = 0.0001; p < 0.0001). O, P TEM uncovered a reduction in myelin sheath thickness in Dio-CLU mice relative to the Dio-NC group (n = 3, three independent experiments were conducted). Q, R The percentage of unmyelinated axons was elevated in Dio-CLU mice compared to the Dio-NC group (n = 3 mice/group, p = 0.0241), while the axon diameter was unchanged (n = 3 mice/group, p = 0.0789). For immunofluorescence pictures: Scale bar: 50 μm. For TEM: Scale bar: 1 μm. Statistical analysis was performed using unpaired two-tailed Student’s t-tests, mean ± SD, *p < 0.05, **p < 0.01, ***p < 0.001. ns. = p not significant. Tam tamoxifen.

Fig. 5. CLU-induced OPC injuries were attenuated by VLDLR knockdown in vivo.

A, B NG2-shRNA-mediated VLDLR knockdown efficiency was determined by Western blot (n = 3 technical replicates/group, p = 0.0005). C Schematic of viral constructs and representation of the experimental setup. D–F The reductions of NG2+ OPCs and Edu+ OPCs in the CLU-administered group were reversed by VLDLR’s specific knockdown of OPCs (n = 3 mice/group, p = 0.0035; p = 0.0083). G, H The increase of CC3+ OPCs in the CLU-administered group was reversed by VLDLR’s specific knockdown of OPCs (n = 3 mice/group, p = 0.0070). For immunofluorescence pictures: Scale bar: 50 μm. Statistical analysis was performed using unpaired two-tailed Student’s t-tests, mean ± SD, *p < 0.05, **p < 0.01, ***p < 0.001. ns. = p not significant. VLDLR very low-density lipoprotein receptor.

Fig. 6. CLU-induced OPCs and OL injury through inhibition of the PI3K-AKT pathway.

A–D Western blot confirmed a time-dependent reduction in p-PI3K, p-AKT, and p-mTOR after CLU stimulation (n = 3 technical replicates/group) (For p-mTOR/TOR, 1 h vs. 0 h p = 0.2193; 3 h vs. 0 h p = 0.0001; 6 h vs. 0 h p = 0.0335; 12 h vs. 0 h p = 0.0017; 24 h vs. 0 h p = 0.0001) (For p-ATK/AKT, 1 h vs. 0 h p = 0.3765; 3 h vs. 0 h p = 0.1101; 6 h vs. 0 h p = 0.0011; 12 h vs. 0 h p < 0.0001; 24 h vs. 0 h p < 0.0001) (For p-PI3K/PI3K, 1 h vs. 0 h p = 0.0155; 3 h vs. 0 h p = 0.0061; 6 h vs. 0 h p = 0.0008; 12 h vs. 0 h p = 0016; 24 h vs. 0 h p = 0007). E–I SC79 (an AKT agonist) could mitigate CLU damage to the OPC lineage, while MK2206 (an AKT antagonist) could exacerbate CLU damage to the OPC lineage in vivo (n = 4 mice/group) (For (F), DPBS vs. DPBS+MK2206 p = 0.0117, CLU vs. CLU+SC79 p = 0.0015, CLU+SC79 vs. CLU+MK2206 p = 0.0002) (For (G), DPBS vs. DPBS+MK2206 p = 0.0003, CLU vs. CLU+SC79 p < 0.0001, CLU vs. CLU+MK2206 p = 0.0005, CLU+SC79 vs. CLU+MK2206 p < 0.0001) (For (H), DPBS vs. CLU p < 0.0001, CLU vs. CLU+SC79 p = 0.0003, CLU+SC79 vs. CLU+MK2206 p < 0.0001) (For (I), DPBS vs. CLU p < 0.0001, CLU vs. CLU+SC79 p = 0.0006, CLU+SC79 vs. CLU+MK2206 p < 0.0001). For immunofluorescence pictures: Scale bar: 50 μm. Statistical analysis was performed using unpaired two-tailed Student’s t-tests, mean ± SD, *p < 0.05, **p < 0.01, ***p < 0.001. ns. = p not significant.

Fig. 7. Enhanced remyelination in the corpus callosum with CLU knockdown in the LPC model.

A Schematic of viral constructs. B Schematic representation of the experimental arrangement. C–E Increased proliferation ability of OPCs following CLU knockdown (n = 3 mice/group, p = 0.0001; p = 0.0013). F–H Increased numbers of NG2+ OPCs and CC1+ OLs after CLU knockdown (n = 3 mice/group, p < 0.0001; p = 0.0005). I, J Enhanced OPC differentiation ability after CLU knockdown, detected by EdU differentiation assay (n = 3 mice/group, p = 0.0011). K, L TEM analysis revealed increased myelin sheath thickness after CLU knockdown (n = 4 mice/group, three independent experiments were conducted). M, N The percentage of unmyelinated axons was decreased in A-shCLU mice compared to the A-shNC group (n = 4 mice/group, p = 0.0037), while the axon diameter was unchanged (n = 4 mice/group, p = 0.6686). For immunofluorescence pictures: Scale bar: 50 μm. For TEM Scale bar: 1 μm. Statistical analysis was performed using unpaired two-tailed Student’s t-tests, mean ± SD, *p < 0.05, **p < 0.01, ***p < 0.001. ns. = p not significant. LPC lysolecithin, EdU 5-ethynyl-2’-deoxyuridine, Brdu 5-bromo-2’-deoxyuridine.

Fig. 8. Improved myelin renewal with astrocyte-specific CLU conditional knockout in LPC and EAE models.

A Schematic representation of the ALDH1L1-CreERT:: CLU^fl/fl mice construction. B–D Increased numbers of Olig2⁺ OPCs and CC1⁺ OLs in the corpus callosum of CLU cKO^AST mice in the LPC model (n = 3 mice/group, p = 0.0013; p = 0.0004). E–H TEM analyses revealed thicker myelin sheaths in CLU cKO^AST mice compared to CLU^fl/fl mice. The percentage of unmyelinated axons decreased in CLU cKO^AST mice compared to the CLU^fl/fl group (n = 3 mice/group, p = 0.0052), while the axon diameter remained consistent (n = 3 mice/group, p = 0.8429). I Astrocyte-specific CLU conditional knockout decreased the clinical severity of EAE animals (n = 6 mice/group). J–L Astrocyte-specific CLU conditional knockout increased the number of NG2⁺ OPCs at the peak of the disease onset, while the number of Ki67+ OPCs was unchanged (n = 5 mice/group, p = 0.0005; p = 0.9679). M, N Astrocyte-specific CLU conditional knockout increased the number of CC3⁺ OPCs at the peak of the disease onset (n = 5 mice/group, p < 0.0001). For immunofluorescence pictures: Scale bar: 50 μm. For TEM Scale bar: 1 μm. Statistical analysis was performed using unpaired two-tailed Student’s t-tests, mean ± SD, *p < 0.05, **p < 0.01, ***p < 0.001. ns. = p not significant. LPC lysolecithin.