1. Neuroscience
Download icon

Mature oligodendrocytes bordering lesions limit demyelination and favor myelin repair via heparan sulfate production

  1. Magali Macchi
  2. Karine Magalon
  3. Céline Zimmer
  4. Elitsa Peeva
  5. Bilal El Waly
  6. Béatrice Brousse
  7. Sarah Jaekel
  8. Kay Grobe
  9. Friedemann Kiefer
  10. Anna Williams
  11. Myriam Cayre
  12. Pascale Durbec  Is a corresponding author
  1. Aix Marseille Univ, CNRS, IBDM, France
  2. MRC Centre for Regenerative Medicine, Multiple Sclerosis Society Centre for Translational Research, University of Edinburgh, United Kingdom
  3. Institute of Physiological Chemistry and Pathobiochemistry and Cells-in-Motion Cluster of Excellence (EXC1003-CiM), University of Münster, Germany
  4. Max Planck Institute for Molecular Biomedicine, Germany
Research Article
  • Cited 1
  • Views 935
  • Annotations
Cite this article as: eLife 2020;9:e51735 doi: 10.7554/eLife.51735

Abstract

Myelin destruction is followed by resident glia activation and mobilization of endogenous progenitors (OPC) which participate in myelin repair. Here we show that in response to demyelination, mature oligodendrocytes (OLG) bordering the lesion express Ndst1, a key enzyme for heparan sulfates (HS) synthesis. Ndst1+ OLG form a belt that demarcates lesioned from intact white matter. Mice with selective inactivation of Ndst1 in the OLG lineage display increased lesion size, sustained microglia and OPC reactivity. HS production around the lesion allows Sonic hedgehog (Shh) binding and favors the local enrichment of this morphogen involved in myelin regeneration. In MS patients, Ndst1 is also found overexpressed in oligodendroglia and the number of Ndst1-expressing oligodendroglia is inversely correlated with lesion size and positively correlated with remyelination potential. Our study suggests that mature OLG surrounding demyelinated lesions are not passive witnesses but contribute to protection and regeneration by producing HS.

Introduction

In multiple sclerosis (MS), OLG are the target of inflammatory and immune attacks and their death results in multiple focal demyelinated lesions in the CNS. Myelin loss remains in defined areas, rather than expanding to involve all of the white matter, and the mechanism by which demyelinated lesions stop expanding is not understood. Some of these lesions may stop expanding as inflammation is controlled and lesions are remyelinated. Remyelination involves OPC recruitment to the lesion, differentiation into myelin forming cells and remyelination of denuded axons, and the success of this depends on environmental context, including secreted factors from neighboring cells (Franklin and Ffrench-Constant, 2008; Patrikios et al., 2006). This repair process, which occurs spontaneously in MS patients, is highly variable between patients and between lesions (Patrikios et al., 2006). Regenerative failure is mainly attributed to defects in OPC recruitment towards the demyelinated areas (Boyd et al., 2013) and/or to their incapacity to differentiate into myelinating OLG at the lesion site (Franklin and Ffrench-Constant, 2008; Wolswijk, 1998; Chang et al., 2002; Franklin, 2002).

Multiple factors are involved in this regenerative process including those produced by reactive astrocytes or microglia and macrophages (Miron et al., 2013). These contribute to myelin destruction, but also to myelin debris removal and beneficial effects by secreting factors that directly or indirectly support remyelination (El Waly et al., 2014; Emery, 2010; Aguirre et al., 2007; Williams et al., 2007; Courtès et al., 2011; Vernerey et al., 2013). Interestingly oligodendrocyte lineage cells also produce factors that modulate remyelination, such as the morphogen shh which is produced by OLG and OPC at the onset of demyelination in lysophosphatidyl choline (LPC)-induced lesions (Ferent et al., 2013). In this context, blocking Shh activity induces an increase in lesion size and a block in OPC proliferation and differentiation, and conversely Shh overexpression leads to the attenuation of the lesion extent and promotes oligodendrogenesis (Ferent et al., 2013). Identifying such actors involved in myelin damage and remyelination is needed for the design of future protective and regenerative therapies.

The presence of a multitude of signals regulating specific steps of remyelination raises the hypothesis that key factors may be necessary to integrate all these cues. One of these key factors may be HS proteoglycans (HSPG), as there is now compelling evidence that HSPGs play a critical role in regulating spatiotemporal coordination of signals in the extracellular microenvironment of many tissues during brain development and in adulthood (Sarrazin et al., 2011). HS chains consist of linear repeated disaccharide units of N-acetyl glucosamine and glucuronic acid which are synthesized on proteoglycan core proteins. Ndst enzymes perform the first step of these sugar modifications thus specifying the functional properties of HSPGs (Lindahl et al., 1998; Perrimon and Bernfield, 2000; Carlsson et al., 2008). Among the four known Ndst enzymes, Ndst1 appears as the key enzyme for addition of N-sulfated motifs to HS chains in brain during development, as shown by limited functional redundancy mediated by other Ndst enzymes (2-4) in Ndst1 KO mice (Grobe, 2005; Pallerla et al., 2007). During development, HS proteoglycans provide an important signaling scaffold allowing spatial concentration or trapping of numerous molecules such as morphogens and growth factors (Matsuo and Kimura-Yoshida, 2014) and the control of receptor activity (Matsuo and Kimura-Yoshida, 2014; Gallagher, 2001; Häcker et al., 2005; Parker and Kohler, 2010). Following CNS injury, HSPGs are known to play a pivotal role in post-lesional plasticity and regeneration (Iseki et al., 2002; Hagino et al., 2003). Some HS proteoglycans are over-expressed by reactive astrocytes in injured mouse brain and provide positive (Iseki et al., 2002) or negative (Hagino et al., 2003) environmental support for axon regenerative responses. In vitro, HS proteoglycans can prevent OLG differentiation, maintaining OPC in an immature proliferative phenotype by acting as a FGF-2 co-receptor (McKinnon et al., 1990; Bansal and Pfeiffer, 1997). Therefore, we hypothesized that HS proteoglycans play an organizing role in controlling myelin damage and repair.

Here we show that mature OLG bordering a demyelinated lesion limit lesion extension and influence OPC mobilization via HS production. Using a model of acute focal demyelination of the corpus callosum in mice, we show that Ndst1 expression is induced in OLG around the lesion throughout the phases of demyelination and remyelination. Ndst1 expression and subsequent HS accumulation mostly accumulate at the margin of the lesion, delimiting the lesion from the intact corpus callosum during demyelination. To evaluate the relevance of Ndst1 induction for lesion formation and repair, we exposed genetically modified mice with selective deletion of Ndst1 in oligodendroglia to focal demyelination of the corpus callosum. Lack of Ndst1 in OLG resulted in an increased lesion size, and a sustained OPC and microglia/macrophage activation at the early stage of remyelination. HS enrichment correlates with and is necessary for the binding around the lesion site of the morphogen Shh, suggesting that Ndst1 expression and HS secretion by OLG enhances Shh signaling after demyelination, thus favoring remyelination (Ferent et al., 2013; Zakaria et al., 2019). Furthermore, NDST1 expression in OLG was also increased in human postmortem tissues from multiple sclerosis patients. This increased density of NDST1+ OLG in lesions was inversely correlated with the size of the lesion and positively correlated with remyelination.

Results

Demyelination triggers Ndst1 up-regulation by OLG and creates a transient N-sulfated belt around the lesion

To identify candidates that could regulate interactions between progenitors and the injured environment, a microarray analysis was performed to compare gene expression in purified oligodendroglia from adult healthy and demyelinated animals (Cayre et al., 2013). One of the most robustly and significantly up-regulated genes after demyelination was Ndst1, a key enzyme of HS proteoglycan synthesis (fold increase of 48.9 and 14.0 in two different trials; p≤0.001; microarray data are available at GEO with accession number GSE47486). This up-regulation of Ndst1 was confirmed in vivo at 21 days in mice exposed to EAE by in situ hybridization combined with Olig2 labeling, a pan OLG marker. While Ndst1 was not detected in the corpus callosum of control brains (Figure 1—figure supplement 1A), it was highly expressed by the Olig2+ population after EAE in the corpus callosum (Figure 1—figure supplement 1B–C) in close proximity to lesion sites (Figure 1—figure supplement 1C).

To characterize the up-regulation of Ndst1 after demyelination, we used LPC to trigger focal demyelination lesions in the mouse corpus callosum (Figure 1A). In this model, demyelination is not T cell driven, and demyelination and remyelination proceed in a stereotypic sequence: demyelination occurs within few days, endogenous progenitor mobilization peaks at eight dpi and is followed by OPC differentiation (El Waly et al., 2014). Production of new myelin is then observed after 2 weeks. Demyelination of the corpus callosum is clearly visible after LPC injection in a reporter mouse where myelin fluoresces green (plp-GFP) (Spassky et al., 2001; Le Bras et al., 2005) by the total loss of GFP signal around the injection site (Figure 1—figure supplement 2). The lesion is also characterized by a strong increase in cell density (due to glia proliferation and to microglia/macrophage infiltration) observed by Hoechst staining (Figure 1—figure supplement 2) that strictly coincides with loss of GFP fluorescence.

Figure 1 with 2 supplements see all
Ndst1 up-regulation upon LPC-induced demyelination of the corpus callosum.

(A) Scheme showing the site of LPC injection (red point) in the adult corpus callosum and the location of picture shown in C (red rectangle). (B) Ndst1 expression levels (RT-qPCR) in the corpus callosum of healthy or demyelinated mice, contralateral (contra) and ipsilateral (ipsi) to the lesion site showing the Ndst1 up-regulation in the ipsilateral side. Tissues from five mice were pooled in each condition. Error bars represent S.E.M. *p<0.05, non-parametric ANOVA followed by Kruskal-Wallis test (independent two group comparisons). (C–H) Ndst1 in situ hybridization performed at 5 (C-F, n = 4), 8 (G, n = 4) and 14 (H, n = 4) dpi illustrating the Ndst1 expression pattern at different time points of demyelination (C–F) and remyelination (G–H). (D–E) Enlarged views of the CC in (C) corresponding to contralateral side (D) and positive cells at the margin of the demyelinated area at the site of LPC injection (E). CC, corpus callosum; Cx, cortex; SVZ, sub-ventricular zone; V, ventricle (structures are delineated by brown dotted lines, lesion with white dotted lines). Scale bars: 50 µm in F, G and H; 20 µm in D, F, H; 10 µm in D and E Asterisk in G indicates the site of injection since the demyelinated lesion is no longer visible at 14 dpi.

We first evaluated Ndst1 expression levels by performing RT-qPCR analysis using the corpus callosum of healthy or demyelinated mice on the ipsi- and contralateral sides to the LPC injection, 8 days post injection (dpi). We quantified a mean 41% increase in the Ndst1 expression level in the demyelinated corpus callosum compared to healthy corpus callosum (Figure 1B; p=0.05). Ndst1 transcripts were found up-regulated in demyelinated corpus callosum by in situ hybridization at five dpi during demyelination (Figure 1C–F), eight dpi (Figure 1G), and 14 dpi (Figure 1H). At days 5 and 8 dpi, Ndst1 expressing cells delimited a belt around the lesion site. Weak staining was observed distal to the lesion, in the contralateral side of the corpus callosum and at the core of the lesion (Figure 1C–D). Thus, the induction of demyelination in the corpus callosum triggers Ndst1 up-regulation, and this change is sustained throughout the phases of demyelination and remyelination.

Since Ndst1 catalyzes a mandatory step in the synthesis of HS chains, its expression is likely to reflect the distribution of HS. We thus examined the outcome of Ndst1 activity by analyzing the distribution of N-sulfated motifs after demyelination. To do so, we used the anti-HS antibody 10E4 (Grobe, 2005; Pan et al., 2014), which exclusively detects the N-sulfated motifs produced by Ndst1 activity (Pallerla et al., 2007). While no staining was detected in the contralateral corpus callosum (Figure 2A), N-sulfated HS positive puncta formed a belt around the demyelination lesion (Figure 2B–C), similar to the profile of Ndst1-induction. Furthermore, as with Ndst1 expression, no HS staining was detected within the lesion (Figure 2B–C). This staining is specific as it was absent after treatment with Heparinase, an enzyme that digests N-sulfated motifs on the HS chains (David et al., 1992; Figure 2D). This correlation between Ndst1 up-regulation and the presence of a highly N-sulfated microenvironment indicates the functional activity of Ndst1 after LPC-induced demyelination in the corpus callosum.

N-sulfate-enriched microenvironment forms a belt around the demyelinated lesion.

HS (10E4) labeling on the contra- (A) and ipsi- (B–C) lateral side to the lesion illustrates the generation of a N-sulfated microenvironment surrounding the lesion (delimited by white dashed lines) at five dpi (n = 3). No immunoreactivity was found after Heparinase I treatment (D) thus validating the 10E4 antibody specificity. Scale bars: 20 µm in A, B, D; 10 µm in C. CC, corpus callosum; V, ventricle.

The phenotype of Ndst1 expressing cells around the demyelinated lesion was examined at five dpi, using co-staining for several markers combined with Ndst1 in situ hybridization. We found that Ndst1 expressing cells are immunopositive for Olig2 (Figure 3A) and Plp+ using double in situ hybridization (Figure 3B). We quantified that 98.0 ± 1.5% Ndst1+ cells express Olig2 indicating that virtually all Nsdt1 cells belong to the oligodendroglia lineage (Figure 3—source data 1). We found that Ndst1 expressing cells are immunopositive for Olig2, PDGFRα and CC1 (Figure 3A,C,D) and Plp+ using double in situ hybridization (Figure 3B) At five dpi, 97.5 ± 1.7% of Ndst1 expressing cells were co-labeled with CC1, a mature OLG marker (Figure 3D), while only 0.8 ± 0.3% co-expressed the OPC marker PDGFRα (Figure 3C).

Ndst1 expressing cells around the lesion belong to the oligodendroglial lineage.

(A–B) Ndst1 in situ hybridization successively combined with Olig2 immunostaining (A) or Plp in situ hybridization (B) labeling, two OLG markers, illustrating Ndst1 up-regulation in oligodendroglia lineage cells surrounding the lesion site at five dpi (n = 3). (C–D) Representative images of Ndst1/PDGFRα (C) and Ndst1/CC1 (D) co-labeling illustrating that both OPC (C) and mature OLG (D) up-regulate Ndst1 after demyelination at five dpi (n = 4). Inserts in (A–D) illustrate boxed regions at high magnification. Scale bars: 20 µm.

Of note, in the belt delimitated by Ndst1 staining surrounding the lesion, only half of the Olig2+ cells expressed Ndst1 (45.8 ± 3.4%). This may indicate that not all stages of oligodendroglial maturation or not all oligodendrocytes respond equally to the lesion. We observed that, 8.5 ± 3.2% of PDGFRα+ cells and 48.9 ± 8.5% of CC1+ cells surrounding the lesion co-labeled with Ndst1. Our data indicate that the majority of Ndst1+ cells around the lesion are mature OLG which are more prone than OPC to activate Ndst1 expression in response to demyelination.

Deletion of Ndst1 in the Olig2+ population transiently worsens the extent of demyelination and modifies OPC reactivity after LPC injection

To test if Ndst1 activity in oligodendroglia controls demyelination and/or remyelination, we generated transgenic mice with a conditional deletion of Ndst1 in Olig2+ cells, by breeding Olig2-Cre+/- mice and Ndst1 Flox/Flox mice (Grobe, 2005; Dessaud et al., 2007). The efficiency of inactivation of Ndst1 expression in Olig2 cells was monitored by in situ hybridization in the context of LPC-induced demyelinating lesion, revealing a drastic decrease in Ndst1 expression in lesioned mutants compared to control (Figure 4A–F). As revealed by immunostaining using the anti-HS antibody, a significant reduction of 53% of the staining in N-sulfated HS positive puncta around the demyelination lesion was also observed in mutant compared to control (Figure 4G–I; p=0.05). In healthy mice, quantitative analysis of myelin content (Figure 4—figure supplement 1A–C; p=0.2), astrocyte density (Figure 4—figure supplement 1D–F; p=0.4) and oligodendroglial lineage cell density (Figure 4—figure supplement 1G–L; p=0.6, 0.2 and 0.2 for Olig2, CC1 and PDGFRα cell density respectively) revealed no difference between control (Olig2-Cre+/-) and mutant (Olig2-Cre+/-; Ndst1 Flox/Flox) adult mice, thus indicating that conditional deletion of Ndst1 in the Olig2+ cell population does not interfere with brain development and with subsequent myelin maturation.

Figure 4 with 1 supplement see all
Ndst1 inactivation in oligodendrocyte lineage cells in Olig2-Cre+/-; Ndst1 Flox/Flox mice.

(A–B) Representative images of the lesion site (delineated by white dashed lines) in the corpus callosum of control (A) (n = 2) and mutant (B) (n = 2) mice at eight dpi illustrating the enlargement of the lesion size in mutant mice compared to control mice. Olig2 (in red) is used to label oligodendrocyte lineage cells. In situ hybridization revealed a marked reduction in Ndst1 expression surrounding the lesion site in mice with conditional inactivation in the oligodendroglial lineage cells (B, D, F) compared to control mice (A, C, E). C and D are high magnifications of the squares in A and B respectively. E and F are high magnifications of the squares in C and D respectively. Representative images of 10E4 immunostaining at the lesion site (delineated by white dashed lines. 8dpi) in the corpus callosum of control (G) and mutant (H) mice showing a strong reduction of heparan sulfate labeling in absence of Ndst1 in oligodendrocytes. (I) Quantitative analysis of heparan sulfate labeling area fraction in control and Mutant conditions (n = 4 mice per condition). Error bars represent S.E.M. *p<0.05, non-parametric Mann-Whitney test (independent two group comparisons). CC, corpus callosum, V, ventricle, St, Striatum. Scale bars: 100 µm in A-; 20 µm in C-D. 30 µm in G-H. Source files of the quantitative analyses are available in the Figure 4—source data 1.

We first performed LPC-induced demyelination of the corpus callosum in control and mutant mice and measured the size of the lesion at 4, 8 and 14 dpi. As before, lesions were identified based on the high density of nuclei at the injection point (Figure 1—figure supplement 2). While no difference was detected at four dpi (0.226 ± 0.036 vs. 0.159±0.033 mm³ in control and mutant mice, respectively; p=0.23), a significant two-fold increase in lesion size was observed in mutant compared to control mice at eight dpi (0.199 ± 0.032 vs. 0.097±0.022 mm³, p=0.023) (Figure 5A–C). During the remyelination phase (between 8 and 14 dpi), the lesion area decreased in both groups reaching comparable sizes at 14 dpi (0.033 ± 0.02 vs. 0.028±0.012 mm³ in control and mutant mice, respectively; p=0.97) (Figure 5C).

Deletion of Ndst1 in Olig2+ cells affects lesion size and OPC mobilization after LPC-induced demyelination of the corpus callosum.

(A–B) Representative images of the lesion site (delineated by white dashed lines) in the corpus callosum of control (A) and mutant (B) mice at eight dpi illustrating the enlargement of the lesion size in mutant mice compared to control mice. (C) Quantitative analysis of the lesion size at 4, 8 and 14 dpi (n = 8,9,4 control and n = 9,12,6 mutant mice respectively). (D–E) Oligodendroglia labeled by Olig2 staining within the demyelinated area at eight dpi (E) compared to control mice (D). (F) Olig2 mean cell density in healthy (CTL) or demyelinated control and mutant mice at 4, 8, 14 dpi. (G–H) Mature OLG co-labeled by Olig2/CC1 within the demyelinated lesion at eight dpi in control (G) and mutant (H) mice. (I) Quantification of mean cell density of Olig2+/CC1+ cells within the demyelination lesion in healthy (CTL) or demyelinated control and mutant mice at 4, 8, 14 dpi. (J–K) Ki67+ immunolabeling shows the proliferation status of cells within the lesion 8dpi in control (J) and mutant (K) mouse. (L) Graph represents the cell proliferation (Ki67+ cells) in mutant relative to control mice at 4 and 8 dpi (n = 9,12 control and n = 8,16 mutant mice respectively). (M–N) Co-immunolabelling of Olig2 and Ki67 showing OPC proliferation in control (M) and mutant (N) mouse 8dpi. (I) Quantification of proliferating OPC (Ki67+/olig2+ cells) in lesion sites at 4 and 8 dpi (n = 6,11 control and n = 7,13 mutant mice respectively). Error bars represent S.E.M. *p<0.05, ***p<0.001, non-parametric Mann-Whitney test (independent two group comparisons). Scale bars: 50 µm in A, B, D, E and 10 µm in, G, H, J, K, M and N. Source files of quantitative analyses are available in the Figure 4—source data 1.

We examined how these changes in the local environment in these mutant mice affect OPC mobilization during remyelination by analyzing Olig2+ cells density (Figure 5D–F), maturation status (Figure 5G–I) and proliferation (Figure 5J–O). In accordance with demyelination, there was a marked decrease in Olig2+ cell density within the demyelinated area compared to healthy corpus callosum in both groups at four dpi (55.5 ± 3.3% and 57.3 ± 3.9% decrease in control and mutant mice respectively, p=0.03 and p=0.001) (Figure 5F), reflecting the loss of oligodendrocytes. We observe that in both conditions the density of Olig2+ cells returned to uninjected control values at 14 dpi. Quantification of mean cell densities of mature OLG (CC1+) (Figure 5G–I) within the lesion throughout the time course revealed no significant difference between the two groups indicating that cell differentiation is not affected by Ndst1 inactivation.

We observed that the density of Ki67+ proliferating cells within the lesion area was two-fold increased in Ndst1 mutant mice compared to control mice at eight dpi (217.8 ± 42.8% vs. 100±13.5% in mutant and control mice, respectively, p=0.037) (Figure 5J–K). Some of these proliferative cells are OPC since they co-express Olig2 and Ki67 (Figure 5M–O). At four dpi, the percentage of proliferating Olig2+ cells was lower (but not significantly different) in mutant mice compared to control (23 ± 3.6% vs. 17 ± 2.6% of Ki67+/Olig2+ cells in control and mutant mice, respectively, p=0.29) (Figure 5O). At eight dpi, the percentage of proliferating Olig2+ cells was significantly increased in mutant mice (7.6 ± 0.8% vs 3.2 ± 0.5% in mutant and control mice, respectively, p=0.0004) indicating a prolongation of OPC reactivity during the repair phase. These data show that OPC reactivity is altered in absence of Ndst1 at the onset of remyelination (8dpi). To note, no significant difference in cell death was detectable in the lesion between the two groups at four dpi (341.7 ± 4.5 and 357 ± 111.1 caspase3+ cells per mm² in control and mutant mice, respectively, p=0.28).

Together, these results suggest that Ndst1 expression in the Olig2+ population has no effect on initial demyelination (equivalent lesion size at four dpi) but protects the lesion from enlarging and participates in the control of OPC mobilization.

Deletion of Ndst1 in the Olig2+ population modulates microglia/macrophage activation

While the total number of proliferating cells within the lesion area was strongly increased in Ndst1 mutant mice compared to control mice at eight dpi (Figure 5J–L), the percentage of OPC among these cells represent only 7.6% in the mutant. These data suggest that Ndst1 loss in the Olig2 population indirectly modulates proliferation of surrounding cell types in the context of a demyelinating lesion. To address this, we evaluated the proliferation and activation states of the macrophage/microglia participating in demyelination-remyelination in this acute demyelination model. We found a robust increase in proliferation of CD68+ cells in mutant compared to control mice at eight dpi (166.4 ± 24.3 vs. 79.3 ± 20.6 CD68+/Ki67+ cells per mm² in mutant and control mice, respectively, p=0.026) (Figure 6A–C). Upon CNS insult, microglia/macrophages are quickly activated, changing their shape from ramified to rhomboid. Rhomboid versus ramified polarization of total or activated microglia/macrophages was examined using respectively Iba1 (Figure 6D–E) and CD68 (Figure 6F–H) immunostaining. We observed a switch of the microglia/macrophage polarization among the whole Iba1 and CD68 population in favor of the rhomboid phenotype in mutant mice compared to control at eight dpi. This effect was quantified for activated microglia (ratio of rhomboid/ramified CD68+ cells of 0.28 ± 0.05 in control vs. 0.66 ± 0.1 in mutant mice, p=0.038) (Figure 6H). While the activation phenotype tended to decrease between 4 and 8 dpi in control mice, it tended to increase in mutant animals. We then evaluated the expression level of Cox2, a marker of pro-inflammatory (M1) microglia/macrophage (Chhor et al., 2017) and observed a significant 77% increase in the number of Cox2+ microglial cells in mutant mice compared to control (p=0.01), indicating a delay in the pro-inflammatory (M1) to pro-regenerative (M2) switch in the absence of Ndst1 in oligodendroglia (Figure 6I–K). These results demonstrate that Ndst1 deletion in the Olig2 population is sufficient to enhance microglia/macrophage proliferation and activation at the lesion site at the onset of remyelination.

Effect of Ndst1 deletion on microglia/macrophage activation.

(A–B) CD68+/Ki67+ co-immunolabeling shows the proliferation status of activated microglia/macrophages. (C) Quantification of proliferating microglia/macrophages (Ki67+/CD68+ cells) in lesion sites at 4 and 8 dpi (n = 3,7 control and n = 3,7 mutant mice respectively). Iba1 (D–E) and CD68 immunolabeling (F–G) shows the increase in rhomboid-polarized microglia/macrophages in the demyelinated area of mutant mice at eight dpi. (H) Quantification of the ratio of rhomboid/branched CD68+ cells in lesion sites at 4 and 8 dpi (n = 3,4 control and n = 3,6 mutant mice respectively) showing a switch of the microglia/macrophage polarization in favor of the rhomboid phenotype in mutant mice at eight dpi. (I–J) Cox2 immunolabeling shows an increase in this M1 phenotype marker at 8dpi in mutant mice. (K) Quantification of Cox2+ cells in lesion sites at 8dpi (n = 4 control and n = 5 mutant mice). Error bars represent S.E.M. *p≤0.05, non-parametric Mann-Whitney test (independent two group comparisons). Scale bars: 50 µm in I-J and 10 µm in A-B, D-E, F-G. Source file of quantitative analyses are available in the Figure 5—source data 1.

Shh binds to HS around the focal LPC-induced demyelinated lesion in the corpus callosum

As previously mentioned, HSPGs form a scaffold that shapes the distribution and activity of numerous growth factors and morphogens during development and provide environmental support for regenerative responses following CNS injury. Among several known HSPG-binding morphogens, Shh was previously identified as a positive regulator for myelin repair (Ferent et al., 2013; Zakaria et al., 2019). In order to determine whether lesion-induced HS enrichment around the lesion site could influence Shh distribution (hence signaling), we used an Alkaline Phosphatase (AP) tagged version of Shh (AP-SHH) to directly assay its binding capacity in demyelinating context. Knowing that the CW sequence serves as a major HS-binding site for Shh (Carrasco et al., 2005; Rubin et al., 2002), we also used AP-SHH recombinant proteins deleted for the CW sequence (AP-SHH-CWdeleted), as a control. Probes were incubated on fresh brain sections obtained 4 days after LPC injections. AP-Shh binding was observed in the cortex in healthy conditions (data not shown) and after lesion (Figure 7B,B’). While no AP-Shh binding was observed in healthy or uninjured contralateral corpus callosum, AP-Shh binding delimited a clear belt surrounding the lesion site in the corpus callosum after LPC injection (Figure 7B–B’). In contrast, AP-Shh-CWdeleted did not bind around the same lesion on adjacent sections (Figure 7C–D). As AP-Shh binding depends on the integrity of its HS-binding motif, this indicates that endogenous Shh localization and concentration may be controlled by HS production by peri-lesional OLG.

AP-tagged Shh protein binds to HS concentrated around LPC-induced lesions in the corpus callosum.

Representative images of adjacent serial coronal sections derived from control mice 4 days after LPC injection and incubated with the fusion proteins AP-Shh-WT (A–B’) or AP-Shh-CW in which the CW sequence responsible for HS binding is absent (C–D) (n = 4). The lesion site is delineated by dashed lines. Staining using B-gal is clearly visible around the lesion after AP-Shh incubation (B–B’), while no staining is observed when the AP-Shh-CW deleted protein is used (D). These data show that Shh is concentrated around the lesion and that this distribution depends on the integrity of the HS binding motif. (E) Quantification of Ptch1 expression at 8dpi in control and mutant mice reported in number of dots per cell (n = 4 control and n = 5 mutant mice, p=0.07). (F–G) Illustration of Ptch1 expression in peri-lesional areas in control (F) and mutant (G) mice after labeling as detected by RNAscope technology. CC, corpus callosum; Cx, cortex. Scale bars: 100 µm in A-D. 10 µm in F and G. Source file of quantitative analysis is available in the Figure 6—source data 1.

In order to assess whether SHH signaling was indeed reduced in Ndst1 mutant mice compared to control mice, we quantified Ptch1 (the main SHH receptor) expression around LPC-induced demyelination lesions at eight dpi using RNAscope (Figure 7E–G). Ndst1 mutant mice exhibited 38% decrease in Ptch1 expression compared to control mice (8.6 ± 0.7 vs 13.8 ± 3.0 dots/cells in mutant and control mice respectively, n = 5 mice/group) although it did not quite reach significance (p=0.07) (Figure 7E). Altogether these results suggest that lack of NDST1 in OLG lineage attenuates SHH signaling following demyelination insult.

NDST1 is expressed by oligodendroglia in multiple sclerosis lesions and correlates with lesion size and remyelination

To examine the relevance of our findings for multiple sclerosis (MS) physiopathology, we examined Ndst1 expression in MS tissue. We first probed the snRNAseq data provided by Jakel et al work (Jäkel et al., 2019), and observed that few cells express Ndst1 in both control and MS tissues but when the oligodendrocytes that express Ndst1 are compared, there is a trend to increased expression in MS tissue (Figure 8—figure supplement 1). Because such approach identifies around 15% only of nuclear RNA, we then directly examined the expression pattern of NDST1 protein in MS patient brain sections. Normal appearing white matter (WM), remyelinating, active, chronic active or chronic inactive lesions were analyzed. While NDST1 staining was very weak in control WM (without MS), we observed a significant increase of NDST1 labeling in MS patients WM (Figure 8A–B). Comparison of healthy control, MS normal appearing WM and MS lesions showed that there is a significant increase of NDST1 staining in multiple sclerosis lesions vs. control (p=0.0014) (Figure 8C). Comparison of each MS lesion with its surrounding normal appearing WM using a paired t test, revealed that there is significantly more NDST1 labeling in MS lesions compared to their surrounding normal appearing WM (paired two-tailed t test, t9 = 3.39, p=0.0095). However, NDST1+ cells were distributed evenly throughout the lesion, rather than forming a delimiting band (Figure 8—figure supplement 2D).

Figure 8 with 2 supplements see all
NDST1 is highly expressed in MS tissue and NDST1+OLIG2+ cell density negatively correlates with lesion size.

(A–B) Representative images of NDST1 staining in control (A) and MS (B) WM. (C) Quantification of NDST1 labeling shows a significant over-expression of NDST1 in MS lesions (n = 9) compared to control tissue (n = 4) (Kruskal-Wallis test, H = 13.09, n = 4,9,9, p<0.01, means plus standard deviation). The colors represent paired samples from the same patients. (D–G) Representative images of immunostaining against NDST1 successively co-labeled with OLIG2+ for oligodendroglia (D), GFAP+ for astrocytes (E), NEUN+ for neurons (F), and IBA1+ for microglia/macrophages (G). (H) Quantification of the proportions of different NDST1+ cell types in normal appearing WM and various MS lesions shows that NDST1 expressing cells are mainly oligodendroglia. (I) The proportion of OLIG2+ cells which is NDST1+ is significantly increased in active lesions compared to control (Kruskal-Wallis test, H = 13.92, n = 7,21,4,14,14 p<0.05). Overall, the majority of OLIG2+ cells are NDST1+ in MS lesions and NAWM while this is not true in control brain tissue. (J) The number of oligodendroglia expressing NDST1 is inversely correlated to lesion size. (K) NDST1+ cell numbers positively correlate with the remyelination score assigned to each patient, summing all lesions within blocks from the same MS patients (see Materials and methods). NAWM, normal appearing white matter; RM, remyelinated lesion; A, active lesion; CA, chronic active lesion; CI, chronic inactive lesion. Scale bars represent 50 µm (A–B) or 10 µm (D–G). Source files of quantitative analyses are available in the Figure 7—source data 1.

We then performed double-labeling immunohistochemistry to characterize NDST1-positive cells in various types of lesions and normal appearing WM, using OLIG2 for oligodendroglia (Figure 8D), GFAP for astrocytes (Figure 8E), NEUN for neurons (Figure 8F) and IBA1 for microglia/macrophages (Figure 8G). Quantitative analysis showed that the majority of NDST1 cells were oligodendroglia in all types of lesions (remyelinated, active, chronic active, chronic inactive) and in normal appearing WM (Figure 8H).

The vast majority of OLIG2+ cells in the lesions expressed NDST1, with a gradual reduction in the proportion of OLIG2+NDST1+ cells as lesions become more chronic, and with significantly fewer in control tissue (Figure 8I; p=0.016). The number of NDST1+ oligodendroglia in each lesion was inversely correlated with the lesion’s size (Figure 8J). As blocks of MS tissue contained multiple lesions and sometimes we had multiple blocks from the same patient (see Table 1), we gave each patient an overall remyelination ability score corresponding to how many lesions in the blocks from that patient were remyelinated, or likely to remyelinate if the patient had survived. Here, we were aiming to see whether patients considered being ‘good remyelinators’ using this score express more NDST1. A lesion was given a score of 3 points (complete remyelination), two points (active - likely to remyelinate), one point (chronic active less likely to remyelinate) and 0 points (chronic inactive -unlikely to remyelinate). The total score of all the lesions per patient was then divided by the number of lesions per patient, to allow comparisons. We showed that NDST1+ cell density positively correlated with patient’s score of remyelination ability (Figure 8K). These data reveal that MS tissues with a higher repair potential (containing most active and remyelinated lesions) display a high number of NDST1+ cells therefore suggesting that higher numbers of NDST1+ cells in a lesion may provide a positive environmental support for myelin repair.

Table 1
Classification and characteristic of human post-mortem samples.
PatientSexAge (years)MS typeDisease duration (years)Time to post mortem (h)Number of lesionsActiveChronic activeChronic inactiveRemyeli-nating
MSMS100M46SP8760042
MS121F49SP142421010
MS122M44SP101621100
MS136M40SP91091035
MS154F34SP211242011
MS176M37PP271270025
MS187F57SP271340004
MS207F46SP251080332
MS230F42SP193142002
ControlCO14M64--26-----
CO25M35--22-----
CO28F60--13-----
CO39M82--21-----
Total46741421

Discussion

In this study, we investigated the role of Ndst1 and HS after demyelination. Using LPC-induced demyelination of the corpus callosum in mouse, we showed for the first time that Ndst1-dependent N-sulfate sugar modifications occur at the onset of demyelination and during remyelination. These modifications limit the extension of demyelination and create a permissive substrate enhancing remyelination. First, we show that Ndst1 is almost exclusively expressed by oligodendroglia present at the margin of the lesion delimiting the lesion from the intact corpus callosum. Second, we found that the conditional deletion of Ndst1 in the Olig2 population concomitantly triggers an enlargement of the LPC-induced demyelinated area, alters OPC mobilization and favors the pro-inflammatory (M1) phenotype in microglia. We propose that these effects could be mediated through HS dependent binding of the morphogen Shh which has been shown to be a positive regulator of myelin repair through increased oligodendrogenesis and microglial regulation (Ferent et al., 2013; Zakaria et al., 2019). Finally, using MS brain tissue samples, we show that NDST1 is up-regulated, especially within lesions and that the density of NDST1+ cells in these lesions is negatively correlated to lesion size, and positively correlated to the patient’s potential remyelination ability. Our data suggest that Ndst1/HS expressed by oligodendrocytes around the lesion create a protective and permissive environment playing a positive role in myelin repair.

HS and chondroitin sulfates are the two main classes of sulfated proteoglycans constituting the extracellular space. Chondroitin sulfates are strongly expressed by astrocytes and microglia providing a hostile environment impeding regeneration and remyelination following brain injury (Siebert and Osterhout, 2011; Pendleton et al., 2013; Deng et al., 2015; Lau et al., 2012). Enzymatic degradation of chondroitin sulfate proteoglycans using chondroïtinase ABC treatment in vivo promotes OPC mobilization and remyelination (Lau et al., 2012). In vitro, chondroitin sulfates reduce OPC maturation (Karus et al., 2016) and in vivo the use of a CSPG synthesis inhibitor promotes OPC maturation and accelerates remyelination following focal demyelination in mice (Keough et al., 2016). A recent study has shown that Surfen, a proteoglycan binding agent, reduces inflammation and delays remyelination. (Warford et al., 2018). Our study provides evidence for a beneficial effect of HS proteoglycans (and upstream enzyme Ndst1) on limiting demyelinated lesion size and promoting remyelination. Here we propose that HS on mature OLG changes the amount/stability of soluble factors, and thus the recruitment and proliferation of surrounding cells (such as OPC or microglia). Thus, OLG modify their environment which in turn regulates the behavior of cells in the lesioned CC including microglia and OPC. Therefore, two classes of sulfated proteoglycans have opposite effects on myelin repair; one being detrimental (chondroitin sulfates), and the other (HS) having a beneficial effect. Interestingly, both are found at the margin of the lesion, but while the chondroitin sulfates are secreted by astrocytes and microglia, HS are expressed by oligodendroglia. Our data show that the majority of Ndst1+ cells at the margin of the LPC-induced lesion are mature CC1+ OLG revealing that mature OLG around a demyelination lesion, respond to post lesional cues. Interesting, recent data have shown that pre-existing mature OLG can also participate to myelin regeneration in human and rodent by forming new myelin sheaths further indicating that mature OLG have an active role in myelin repair (Yeung et al., 2019; Duncan et al., 2018).

This OLG response to nearby demyelination appears propitious to regeneration by restricting the lesion spread, favoring OPC mobilization and modulating microglia response.

Macrophages/microglia participate to myelin repair through myelin debris clearance and the secretion of regenerative factors that altogether promote the recruitment of OPC, their proliferation and differentiation into mature myelin-forming cells. Activated microglia also produces various pro-inflammatory mediators (cytokines, chemokines…) that may affect OPC since they express a battery of receptors (Peferoen et al., 2014). Recent studies have shown that efficient remyelination required the dynamic regulation of functional microglia phenotype (Miron et al., 2013; Lloyd et al., 2019). Microglia respond to demyelination with initial pro-inflammatory phenotype (M1) followed later by pro-regenerative phenotype (M2) which actively contributes to myelin repair (Miron et al., 2013; Olah et al., 2012). Interestingly, this transition from pro-inflammatory to pro-regenerative phenotype is a rate-limiting step in the repair process since intra-lesional depletion of pro-regenerative microglia blocks oligodendrocyte differentiation and delays the regenerative process (Miron et al., 2013). In our study, lack of Ndst1/HS from Olig2+ cells in transgenic mice leads to increased microglial proliferation, a strongly activated rhomboid polarization of CD68-expressing cells and an increased density of pro-inflammatory (M1) microglia at an early stage of the remyelination phase (eight dpi). Altogether these effects may contribute to enlargement of the demyelinated lesion and delayed myelin repair. This non-cell autonomous effect observed on microglia activation could in turn disturb OPC mobilization and thus modify the repair process (Miron et al., 2013).

The beneficial action of HS may be related to their ability to bind numerous growth factors and morphogens, as observed during development (Häcker et al., 2005). HS can act as co-receptors for these ligands or are involved in the stabilization and/or local concentration of ligands in the extracellular space, which modulates cell signaling. Ndst1 global knock out mice display developmental defects that mainly resemble those found in embryos deficient for Shh or FGFs (Pallerla et al., 2008). FGF and Shh implication in myelin regeneration and glial reactivation have been extensively examined in mouse models of demyelination (for review, see El Waly et al., 2014). A role of both factors in remyelination was first inferred by correlating their spatial and temporal upregulation after demyelination particularly in LPC-induced demyelination (Ferent et al., 2013; Gudi et al., 2011; Hinks and Franklin, 1999; Tourbah et al., 1992).

Concerning FGF, conflicting results have been published on its activity (Ruffini et al., 2001; Dehghan et al., 2012; Kumar et al., 2007; Armstrong et al., 2002; Zhou et al., 2012; Kang et al., 2019). A recent report has addressed this issue using the simultaneous ablation of both FGFR1 and FGFR2 specifically in cells from the oligodendrocyte lineage (Furusho et al., 2015). This study revealed that FGF signaling is not required for myelin regeneration in acute models of demyelination including LPC-induced demyelination of the spinal cord and cuprizone intoxication (Furusho et al., 2015). Overall, in all these analyses the phenotypes observed after LPC-induced demyelination never recapitulate (even partially) the phenotype observed in the present report using Olig2-Cre; Ndst1 Flox/Flox mice. This suggests that FGF is probably not the main ligand regulated by HS activity during myelin repair in this model.

By contrast, blocking Shh activity leads to an increase in demyelinated lesion size and an altered OPC mobilization after LPC-induced demyelination (Ferent et al., 2013) similar to what we observe in Olig2-Cre; Ndst1 Flox/Flox mice. However, these effects persist at later time points at the end of remyelination, and OPC maturation is also inhibited after Shh inactivation (Ferent et al., 2013). Of note, recent in vitro findings show that the proteolytic processing of Shh required for signaling pathway activation (Ohlig et al., 2012) is finely regulated by HS chains (Ortmann et al., 2015), perhaps influencing Shh concentration and/or spreading. Here we show that Shh binds to HS around demyelinated lesions in mouse. Thus, we propose that HS removal may delay the local accumulation around the demyelinated lesion of Shh produced by OLG (Ferent et al., 2013) and/or delay Shh activation and spreading, leading to a reduction in signaling as suggested by reduced Ptch1 expression at early time-points. Later, the sustained production of Shh may overtake the absence of HS, leading to efficient recovery. Interestingly, Ferent et al., 2013 have shown that the main Shh responding cells (cells expressing Gli1 and/or Smo) after LPC-induced demyelination of the corpus callosum are OLG and microglia. These observations further support the idea that sustained microglia and OPC activation in the absence of HS are at least in part due to altered Shh signaling.

Finally, we can confirm that NDST1 is also over-expressed in human MS brain samples, mostly in Olig2-positive oligodendroglia. High levels of HS proteoglycans and chondroitin sulfate proteoglycans have been previously associated with inflammatory CNS diseases such as MS (Briani et al., 2002; Berezin et al., 2014; van Horssen et al., 2006; Satoh et al., 2009). Here, we show that NDST1 is upregulated within and surrounding MS lesions in post mortem tissue compared to control. NDST1 expression is significantly higher in MS lesions compared to surrounding normal appearing white matter (NAWM), irrespective of lesion type (remyelinated, active, chronic active or chronic inactive). The distribution of the NDST+ cells was different from the mouse model, in that we saw no surrounding band of positive cells, but instead the entire lesions contained positive cells, though the majority of these cells were OLIG2+ oligodendroglia. This difference may be related to the timing of examination of the tissue after the lesion onset, which is later in the human tissue, and secondary to poorer repair in humans. In MS lesions, a large proportion of these OLIG2+ cells express NDST1, suggesting that at least some of these are mature, but we were unable to distinguish mature and immature oligodendroglia in this tissue (due to limitations in double-labelling using effective antibodies in human tissues). However, these observations still concur with our mouse data indicating that oligodendroglia respond to neighboring demyelination. Also consistent with our results showing increased lesion size in Olig2-Cre+/-; Ndst1 Flox/Flox mice, in human samples we observed an inverse correlation between lesion size and density of NDST1-expressing Olig2+ cells. NDST1 cell density also positively correlates with a pathological score of potential remyelination ability in patients.

Overall, our results in mouse and human tissues suggest that NDST1/HS levels are an indicator of oligodendroglial reactivity after demyelination, and are involved in both limiting the size of the lesion and creating a permissive environment for myelin regeneration. Furthermore, this study shows for the first time that mature oligodendrocytes around lesion are active players during demyelination/remyelination by producing HS and thus modifying the local environment. This study will help improve understanding the neuropathology of MS in both limitation of damage and promotion of remyelination, which may in the future help target pharmacological approaches to potentiate myelin repair.

Materials and methods

Key resources table
Reagent type
(species) or resource
DesignationSource or referenceIdentifiersAdditional
information
Genetic reagent (M. musculus)Olig2CrePMID:18046410B6D2F1J/Rj genetic background
Genetic reagent (M. musculus)Ndst1flox/floxPMID:16020517Dr. Kay Grobe (University of Münster, Münster, Germany)
Genetic reagent (M. musculus)PlpgfpPMID:15906234
PMID:11756747
Dr. Bernard Zalc (University of Sorbonne, Paris, France)
Biological sample (H. sapiens)Brain tissue from 9 MS patientsUK Multiple Sclerosis Tissue Bank
(MREC/02/2/39)
Postmortem unfixed frozen
Biological sample
(H. sapiens)
Brain tissue from 4 Control patientsUK Multiple Sclerosis Tissue Bank
(MREC/02/2/39)
Postmortem unfixed frozen
Cell line (H. sapiens)293T HEKATCCCRL3216
Transfected construct (M. musculus)pWiz-AP-SHHPMID:16020517Production of AP-tagged SHH recombinant protein
Transfected construct (M. musculus)pWiz-AP-SHH-CWdeletedPMID:11959830Production of AP-tagged deleted SHH recombinant protein
AntibodyRabbit polyclonal anti-OLIG2MilliporeAB9610IF (1/1000)
AntibodyRabbit polyclonal anti-OLIG2Sigma-AldrichHPA003254IF (1/100)
AntibodyMouse monoclonal anti-APC (clone CC1)CalbiochemOP-80IF (1/400)
AntibodyRat monoclonal anti-PDGFRa (clone APA5)MilliporeCBL1366IF (1/250)
AntibodyMouse monoclonal anti-MBPMilliporeMAB384IF (1/500)
AntibodyMouse monoclonal anti-Ki67BD Pharmingen556003IF (1/500)
AntibodyRabbit polyclonal anti-Caspase 3Cell Signalling9661IF (1/200)
AntibodyRabbit polyclonal anti-GFAPDakoZ0334IF (1/400)
AntibodyGoat polyclonal anti-IBA1AbcamAb5076IF (1/500)
AntibodyRabbit polyclonal anti-IBA1Wako Chemicals019–19741IF (1/500)
AntibodyRat monoclonal
Anti-CD68
AbcamAb53444IF (1/400)
AntibodyRabbit polyclonal anti-COX2AbcamAb15191IF (1/400)
AntibodyMouse monoclonal IgM anti-N-sulfated motifs on HS chains (clone10E4)Amsbio370255–1IF (1/500)
AntibodyMouse monoclonal anti-NDST1Abcamab55296IF (1/50)
AntibodyRabbit polyclonal anti-NeuNAbcamAb104225IF (1/500)
Sequence-based reagent Ndst1_FEurofins GenomicsRT-qPCR primersgctggacaagatcatcaatgg
Sequence-based reagent Ndst1_REurofins GenomicsRT-qPCR primersacacagtacttctacgactatcc
Sequence-based reagentGapdh_FEurofins GenomicsRT-qPCR primersgggttcctataaatacggactgc
Sequence-based reagentGapdh_REurofins GenomicsRT-qPCR primersctggcactgcacaagaagat
Sequence-based reagentplp/dm20PMID:9373029Probe for ISH
Sequence-based reagentNdst1PMID:16020517Probe for ISH
Sequence-based reagentPtch1Advanced Cell Diagnostics402811-C2Probe for RNAScope
Peptide, Recombinant ProteinHuman NDST1Abcamab116875
Commercial assay or kitRNAscope Multiplex Fluorescent kitAdvanced Cell Diagnostics323133
Commercial assay or kitDAB Peroxidase (HRP) Substrate Kit (with Nickel)Vector LaboratoriesSK-4100
Commercial assay or kitVECTOR Blue AP Substrate KitVector LaboratoriesSK-5300
Commercial assay or kitImmPRESS-AP Anti-Rabbit IgG Polymer Detection KitVector LaboratoriesMP-5401
Commercial assay or kitImmPRESS HRP Anti-Mouse IgG Polymer Detection KitVector LaboratoriesMP-7402
Chemical compound, drugLysolecithinSigma-Aldrich-MerckL1381
Chemical compound, drugHeparinaseAmsbio100700
Chemical compound, drugLipofectamine
2000
Invitrogen11668–030
Chemical compound, drugVector BloxallVector LaboratoriesSP-6000
Software, algorithmImageJhttps://imagej.nih.gov/ij/
Software, algorithmZen two liteZeiss
Software, algorithmGraphPad Prismhttps://graphpad.com

Animals and treatments

Request a detailed protocol

All experimental and surgical protocols were performed following the guidelines established by the French Ministry of Agriculture (Animal Rights Division). The architecture and functioning rules of our animal house, as well as our experimental procedures have been approved by the ‘‘Direction Départementale des Services Vétérinaires’’ and the ethic committee (ID numbers F1305521 and 2016071112151400 for animal house and research project, respectively). Surgery and perfusions were performed under ketamine (100 mg/kg, MERIAL, Lyon, France))/xylazine (10 mg/kg, BAYER, Puteaux, France) anesthesia. C57BL/6 wild-type and transgenic mice were successively used to characterize post-lesional expression of Ndst1 and HS after demyelination and to investigate the impact of conditional deletion of Ndst1 in the Olig2-positive cell population. Heterozygous Olig2-Cre+/- (from B6D2F1J/Rj genetic background) (Dessaud et al., 2007) and double transgenic Olig2-Cre+/-; Ndst1 Flox/Flox mice (Grobe, 2005; Dessaud et al., 2007) will be referred below as control and mutant mice, respectively. Mice expressing GFP under the control of the proteolipid protein (plp, a protein largely present in myelin) promoter were used in some experiments to better observe demyelination lesions (called thereafter plpGFP mice). Animals were housed under standard conditions with enrichment and access to water and food ad libitum on a normal 12 hr light/dark cycle.

Human postmortem samples

Request a detailed protocol

Postmortem unfixed frozen tissues were obtained from the UK Multiple Sclerosis Tissue Bank via a UK prospective donor scheme with full ethical approval (MREC/02/2/39). Luxol fast blue (LFB) (staining myelin; Figure 8—figure supplement 2C) and Oil Red O (staining lipids phagocytosed by macrophages) were performed to characterize and classify the lesion types (Boyd et al., 2013). Active lesions have indistinct borders on LFB and lipid-laden macrophages/microglia. Chronic active lesions have a ring of lipid-laden macrophages/microglia and a core with few immune cells. Chronic inactive lesions have a distinct border on LFB and few immune cells. Finally, shadow plaques, thought to represent remyelination, have less intense staining on LFB. This classification was done by two independent researchers for a previous publication (Boyd et al., 2013). In this study, we used active (n = 7), chronic active (n = 4), chronic inactive (n = 14) and remyelinated (shadow) MS plaques (n = 21) from 14 blocks of brain tissue from 9 MS patients and 4 blocks of brain tissue from four controls with no neurological disease (Table 1).

Focal demyelination in the corpus callosum and tissue processing

Request a detailed protocol

Focal demyelination was performed by stereotactic injection of Lysolecithin (LPC) (SIGMA-ALDRICH, St Louis, USA) as described previously (Cayre et al., 2013; Magalon et al., 2007). The corpus callosum from healthy or demyelinated mice, from the ipsilateral and contralateral side to the LPC-induced lesion, were dissected 7 days post injection (dpi) from 1 mm thick coronal slices in cold Hank's Balanced Salt Solution (GIBCO by life technologie, Paisley, UK) and processed for RT-qPCR analysis (Ndst1 primers : exon 6 (forward, 5′-gctggacaagatcatcaatgg-3′) and exon 7 (reverse, 5′-acacagtacttctacgactatcc-3′); for Gapdh: exon1 (forward, 5’-gggttcctataaatacggactgc-3’) and exon2 (reverse 5’-ctggcactgcacaagaagat-3’). Primers from EUROFINS GENOMICS, Ebersberg, GERMANY). For histological analysis, mice were anesthetized and perfused with ice-cold 4% paraformaldehyde (FISHER SCIENTIFIC, Loughborough Leics, UK) in PBS (GIBCO by life technologie, Paisley, UK). Brains were post-fixed overnight in 4% paraformaldehyde in PBS and cut on a vibratome (Leica) in 4 series of coronal sections (50 µm thick) for immunofluorescence, or cryopreserved and cut with cryostat (20 µm thick) for in situ hybridization.

In situ hybridization and immunohistochemistry

Request a detailed protocol

In situ hybridization was performed using plp/dm20 (Peyron et al., 1997) and Ndst1 probes (Grobe, 2005) as described in Virard et al., 2006. RNAscope Multiplex Fluorescent kit (323133; Advanced Cell Diagnostics) was used to detect Ptch1 mouse mRNA. Briefly cryosections were baked for 30 min at 60°C and dehydrated, incubated for 30 min at 40°C with protease III before incubation with RNAScope probe Ptch1 (402811-C2) for 2 hr at 40°C. Immunohistochemistry was performed as described in Cayre et al., 2013. The following primary antibodies were used: rabbit anti-Olig2 (AB9610; 1/1000, Millipore, USA), mouse anti-APC (CC1) (1/400; Calbiochem, USA), and rat anti-PDGFRα (CBL1366; 1/250; Millipore, USA) for oligodendroglial lineage cells; anti-MBP (mouse, 1/500, Chemicon, Millipore S.A.) for myelin sheaths; mouse anti-Ki67 (556003; 1/500; BD Pharmingen) for proliferating cells; rabbit anti-caspase 3 (9661; 1/200; Cell Signaling) for apoptotic cells, rabbit anti-GFAP (1/400) for astrocytes; goat anti-Iba1 (1/500, Abcam) for microglia and macrophages; rat anti-CD68 (1/400, Abcam) for activated microglia and macrophages; rabbit anti-Cox2 (1/400, Abcam) for proinflammatory M1 microglia/macrophage; mouse IgM anti-N-sulfated motifs on HS chains (10E4 antibody, 1/500; Seikagaku, Japan). For Olig2 and Ki67 immunofluorescence, antigen unmasking was performed by 20 min incubation in boiling citrate buffer (10mM pH6). For N-sulfated motifs labeling, floating sections from PFA perfused-brain were incubated for 2 hr 30 at 37°C in buffer (100 mM Sodium Chloride, 1 mM Calcium Chloride, 50 mM Hepes 5 µg, BSA pH 7) with or without Heparinase (3.3 mU from Flavobacterium heparinum, Seikagaku Kogyo Co. # 100700, Japan) (David et al., 1992) before permeabilization. Secondary antibodies coupled to alexa 488, 555 and 647 (1/500, Invitrogen Molecular Probes) were applied for 2 hr 30 at RT in a humid chamber. Sections were counterstained with Hoechst 33342 (1/500, Sigma).

AP-Shh recombinant protein binding test in mice

Request a detailed protocol

Plasmids containing sequences for AP-tagged N-terminal WT or deleted Shh were produced by PCR and ligated into pWIZ vector as described in Grobe, 2005; Rubin et al., 2002. Briefly, plasmids were transiently transfected into HEK cells using lipofectamine 2000 (Invitrogen). Transfection proceeded for 3 hr. Culture supernatants were collected after 60 hr and filtered through 0.45 µm filters (Corning Incorporated, Durham, USA). Hepes 10mM pH7 was added to increase stability. Shh concentration was then evaluated measuring AP activity in culture supernatants. Preparations from mock-transfected HEK cells were generated and used as vehicle controls. The AP-Shh binding test was performed as described in Rubin et al., 2002.

Fresh frozen brain sections were post-fixed with ice-cooled methanol for 8 min. After rinsing with phosphate-buffered saline containing 4 mM MgCl2 and blocking with 1% Bovine Serum Albumin (SIGMA-ALDRICH, St Louis, MO, USA) 1 hr at RT, frozen adjacent sections from healthy or demyelinated C57BL/6 mice were incubated with 5 nM of two AP tagged versions of Shh: AP-Shh recombinant proteins carrying the N-terminal CW sequence (AP-SHH), the main HS-binding site for Shh (Rubin et al., 2002; Carrasco et al., 2005), or lacking this motif (AP-SHH-CWdeleted). Sections were then washed with PBS to dissociate any low affinity interaction and endogenous phosphatases were inactivated by heating at 65°C for 2 hr. AP was revealed by incubating overnight in NBT (100 mg/ml)/BCIP (50 mg/ml) in 100 mM Tris pH 9,5 with 100 mM NaCl and 50 mM MgCl2.

Immunohistochemistry on human post-mortem tissue

Request a detailed protocol

Tissue was fixed in 4% paraformaldehyde in PBS for 30 min. Endogenous peroxidase and AP activity was blocked by 10 min incubation with Vector Bloxall (Vector, SP-6000 VECTOR LABORATORIES, Burlingame, USA). Slides were blocked with ready-to-use 2.5% normal horse serum from Vector secondary antibody kits for at least 20 min. Primary antibodies were incubated overnight in antibody diluent (Spring Bioscience, ADS-125) at 4°C. Primary antibodies used: mouse anti-NDST1 (1/50; Abcam, ab55296), rabbit anti-NeuN (1/500; Abcam, ab104225), rabbit anti-IBA1 (1/500; Wako chemicals, 019–19741), rabbit anti-Olig2 (1/100; Sigma, HPA003254). HS staining with the mouse IgM anti-N-sulfated motifs on HS chains (10E4 antibody, Seikagaku, Japan) did not give any signal on human tissue. NDST1 intensity was evaluated after a short exposition (exactly 2 min). All other stainings were fully developed. To ensure antibody specificity, the NDST1 antibody was pre-absorbed with human NDST1 recombinant protein (Abcam, ab116875), and added to tissue sections, with no staining seen (Figure 8—figure supplement 2A–B).

Secondary antibodies were incubated at RT for 1 hr. Staining was developed with a DAB Peroxidase (HRP) Substrate Kit (with Nickel), 3,3’-diaminobenzidine (Vector, SK-4100) and a VECTOR Blue AP Substrate Kit (Vector, SK-5300) as per manufacturer’s guidelines. Secondary antibodies used: ImmPRESS-AP Anti-Rabbit IgG Polymer Detection Kit (Vector, MP-5401) and ImmPRESS HRP Anti-Mouse IgG Polymer Detection Kit, made in Horse (Vector, MP-7402). PBS washes were performed between each treatment.

Microscopy and quantification

Request a detailed protocol

For mouse tissue analysis, imaging was performed with the Apotome system (Zeiss). The demyelinated area and cell counts were evaluated using Zen software (Zeiss). Immunofluorescent or in situ hybridization positive cells were counted in every fourth section through the whole demyelinated lesion per mouse and averaged for each mouse. Cell counts are presented as the mean of at least three mice. For RNAscope ISH, each punctate dot signal was counted around lesion (by using ROI and analyze particule Fiji Plugins) and reported to total nuclei number. For Heparan sulfate labeling quantification, 10E4-mmunopositive area was analyses on five sections per mouse. A constant exposure time was applied to all sections and for image acquisition, and the area occupied by 10E4+ labeling was quantified using ImageJ software. Lesion size was quantified by measuring the area of high density of nuclei in every fourth section through the whole demyelinated lesion per mouse. In this analysis, high density of nuclei was correlated with myelin loss visualized by MBP staining or by loss of fluorescence in plpGFP mice (Figure 1—figure supplement 2).

For the human post-mortem tissue analysis, slides were imaged using a ZEISS Axio Scan.Z1 slide scanner. One researcher marked out the lesion sites and normal appearing white matter (WM) as areas of interest, while another counted single positive (NDST1+ cells) and double positive cells (NDST+ cells and other brain cell markers combined as above) in these areas of interest (ensuring blinding of counting).

Myelin content was evaluated by double blind scoring of images taken from Plp immunostaining on brain sections (three photos per section and three sections per brain). Score of 4 was attributed to maximum myelination down to 0 for absence of myelin. The mean score for the control group was considered as 100%.

Gene expression profile of demyelinated versus healthy mouse progenitors

Request a detailed protocol

This protocol is fully described in Cayre et al., 2013. Briefly, OPC from eight mice induced for experimental autoimmune encephalomyelitis (EAE mice) at the peak of paralytic symptoms and from eight adult healthy mice as controls were purified using magnetic cell sorting (Miltenyi Biotec). This experiment was replicated in an independent similar experiment. cDNAs were prepared and used (250 ng) as template for Cy3 and Cy5, combined and hybridized to Agilent Whole Mouse Genome Oligo Microarrays 4 Å~44K. Agilent Feature Extraction Software (FES) determines feature intensities and ratios (including background subtraction and normalization), rejects outliers and calculates statistical confidences (P-values). We obtained a gene list with all normalized Cy5/Cy3 log10 ratios, Cy5/Cy3 fold changes, sequence description and P-values. Microarray data are available at GEO with accession number GSE47486.

Statistical analysis

Request a detailed protocol

All the presented values in mice are means ± S.E.M unless otherwise stated. Data were statistically processed with the non-parametric Mann-Whitney test (independent two group comparisons). p<0.05 was considered significant and p<0.01 highly significant. All measurements and subsequent evaluations were performed blind to the experimental group to which the animals belonged. For the human post-mortem tissue analysis, a d'Agostino and Pearson omnibus normality test was used to test whether the data fit a normal distribution and a parametric test were done only if all compared data sets passed the normality test. The NDST1+ cells in control versus multiple sclerosis WM was compared using a two-tailed Mann Whitney U test. Multiple sclerosis lesions and their surrounding WM were compared using a paired two-tailed t test. The absolute numbers of NDST1+ cells and double positive NDST1+ OLIG2+ cells in individual lesions/normal appearing WM were compared by Kruskal-Wallis test. As MS tissue blocks contained more than one lesion, and we had several blocks from the same patients, we gave each patient an overall remyelination ability score corresponding to how many lesions in the blocks from that patient were remyelinated, or likely to remyelinate if the patient had survived. Remyelinated lesions received an arbitrary three points, active lesions two points, chronic active lesions one point and chronic inactive lesions 0 points. This was divided by the number of lesions counted for each patient, to allow comparisons.

References

  1. 1
  2. 2
  3. 3
  4. 4
  5. 5
  6. 6
  7. 7
  8. 8
  9. 9
  10. 10
  11. 11
  12. 12
  13. 13
  14. 14
  15. 15
  16. 16
  17. 17
  18. 18
  19. 19
  20. 20
  21. 21
  22. 22
  23. 23
  24. 24
  25. 25
  26. 26
  27. 27
  28. 28
  29. 29
  30. 30
  31. 31
  32. 32
  33. 33
  34. 34
  35. 35
  36. 36
  37. 37
  38. 38
  39. 39
  40. 40
  41. 41
  42. 42
  43. 43
  44. 44
  45. 45
  46. 46
  47. 47
  48. 48
  49. 49
  50. 50
  51. 51
  52. 52
  53. 53
  54. 54
  55. 55
  56. 56
    Cerebellar proteoglycans regulate sonic hedgehog responses during development
    1. JB Rubin
    2. Y Choi
    3. RA Segal
    (2002)
    Development 129:2223–2232.
  57. 57
  58. 58
  59. 59
  60. 60
  61. 61
  62. 62
  63. 63
  64. 64
  65. 65
  66. 66
  67. 67
  68. 68
  69. 69
  70. 70
  71. 71

Decision letter

  1. Klaus-Armin Nave
    Reviewing Editor; Max Planck Institute of Experimental Medicine, Germany
  2. Gary L Westbrook
    Senior Editor; Oregon Health and Science University, United States
  3. James W Fawcett
    Reviewer; University of Cambridge, United Kingdom

In the interests of transparency, eLife publishes the most substantive revision requests and the accompanying author responses.

Acceptance summary:

This manuscript explored the role of mature oligodendrocytes in regions surrounding demyelinating lesions in the production of heparin sulfates. These results have implications not only for limiting demyelination, but also in remyelination in MS and other demyelinating disorders.

Decision letter after peer review:

[Editors’ note: the authors submitted for reconsideration following the decision after peer review. What follows is the decision letter after the first round of review.]

Thank you for submitting your work entitled "Mature oligodendrocytes bordering lesions limit demyelination and favor myelin repair via heparan sulphate production" for consideration by eLife. Your article has been reviewed by three peer reviewers, one of whom is a member of our Board of Reviewing Editors, and the evaluation has been overseen by a Senior Editor. The following individual involved in review of your submission has agreed to reveal their identity: James Fawcett (Reviewer #3).

Our decision has been reached after consultation between the reviewers. Based on these discussions and the individual reviews below, we regret to inform you that your work will not be considered further for publication in eLife, at least not in its present scope. As you will see from the individual reports there was certainly interest in your findings of an oligodendroglial response to a demyelinating lesion. However, the reviewers felt that at present the experiments aimed at mechanistic insight resulted mostly in correlative data. Perhaps, using genetic mouse models would be a strategy to reach stronger conclusions on causation. These would be time-consuming additional experiments and we do not invite a revised manuscript at this point. Any new submission would be independently assessed, possibly with new reviewers.

Reviewer #1:

This paper is interesting because it provides to my knowledge the first report of a molecular response of intact oligodendrocytes close to a demyelinating lesion, the induction of an enzyme (Ndst1) that increases the expression of heparan sulfate and thus HS-modifed proteins. Based on the phenotype of OPC/OL-specific Ndst1 cKO mice the authors suggest that this oligodendroglial response limits the extent of demyelination and promotes remyelination, possibly by locally increasing sonic hedgehog signaling. The authors also show a similar upregulation of NDST1 within (not around) human MS lesions, where the expression level inversely correlates with lesion size.

Major points:

1) Since primary development and myelination are not perturbed in the brains of Ndst1 cKO mice, another possible explanation of the data is that enhanced activation of microglia (or their altered phenotype) affects remyelination in a non-cell autonomous way.

2) Thus, the Shh data are plausible but only correlations, and one should not leave the impression that reduced Shh binding was shown to cause the delay of remyelination.

3) It is unclear how "remyelination" in the human MS autopsy samples was assessed (pale blue staining could also be incomplete demyelination). In fact, this may explain the different pattern of Ndst1 expression in mouse and human demyelinating lesions.

4) Figure 3. It is unclear what the anti-PLP staining is actually labeling outside the lesion area (in Figure 3B), as PLP is still a myelin marker

5) Figure 1C-H: In-situ hybrization of Ndst1 mRNA in demyelinaing lesions: these observations are critical. but lack quantification (required standard). The alignment of the "lesion" with the "belt" is not obvious when using only dashed lines. Why is Ndst1 expression not shown also as an overview type image (similar to Figure 1—figure supplement 2)? Same is true for Figure 2, which also lacks quantitation. While the "estimated" level of upregulation is stronger than in Ndst1 in Figure 1B, the IHC images are (at present) only case reports.

Reviewer #2:

This paper examines the role of heparan sulphate in remyelination by examining the expression of a key enzyme NDST1 and the consequences of an NDST1 conditional knockout in a mouse model of remyelination. In addition, the authors examine tissue from patients with MS. They conclude that HS containing proteoglycans are involved in remyelination as the enzyme is expressed around demyelinating lesions and its loss delays remyelination as evidenced by an increase in lesion size. Moreover there is a correlation between the degree of expression and the degree of remyelination in MS tissue.

While I think that the role of proteoglycans in remyelination is interesting, this paper is predominantly descriptive and correlative and doesn't provide sufficient mechanistic detail. The evidence for SHH binding in the region of putative HS expression is noteworthy but it is purely correlative. Equally the increase in microglial proliferation is interesting but not explored further.

My view would be that experiments are required to define one or more mechanisms by which HS is involved in remyelination, with gain and/or loss of function experiments clearly proving that the proposed mechanism is both linked to HS expression and regulates oligodendrocyte behaviour during remyelination in vivo.

Additionally I think there are three other areas that require attention.

First, the authors argue that the distribution around the experimental lesions is significant and that they have uncovered a novel function for surviving oligodendrocytes around lesions. It's notable however that they do not see this distribution in the MS tissue and I wonder whether their results simply reflect the lack of any surviving oligodendrocytes following LPC injection. In other words the distribution they see is simply a consequence of the model rather than having any great biological significance

Second, the characterization of the OPC response in the presence or absence of HS is very poor and studies of proliferation and differentiation, including the quantification of newly formed oligodendrocytes, is required

Third, a better characterization of the microglial phenotype (e.g. M1/M2 or similar) is required if the significance of this response is to be understood

Reviewer #3:

This is a thorough and interesting account of the role of HSPGs in remyelination. Many mechanisms of remyelination are known, but the idea that a particular sulfation change in HSPGs could be important is new. There is also data from human MS pathology in the paper, which gives clinical relevance.

1) The paper focuses on NDST which gives a particular form of sulfation to HS. There needs to be some information about the sulfation motif that is produced and any known information on relevant molecules that might bind to this form of HS. Shh is binding is shown, but it is unlikely to be the only effector of the local HS change.

2) If there is an increase in NDST, do the authors have information about any HS core proteins that might also be upregulated?

3) Why would a sulfation change in mature oligos increase myelination? Surely OPCs would be more relevant?

4) The authors show that there is some increase in Shh binding around lesions, but not that it does anything. The statement in the Discussion needs to be modified. Also they do not show local upregulation. Any published information that might link Shh or Shh inhibitor to remyelination needs to be stated. This part of the paper needs attention.

5) Do the authors believe that the changes they see in NDST etc. are specific to demyelination, or are similar changes seen with traumatic lesions and inflammation?

6) Figure 4: Is the increase in cell proliferation in the knockouts entirely accounted for by increased division of microglia?

7) Figure 6 shows some increase in Shh binding around the lesion, but also widespread binding elsewhere. Some comment is warranted.

[Editors’ note: further revisions were suggested prior to acceptance, as described below.]

Thank you for submitting your article "Mature oligodendrocytes bordering lesions limit demyelination and favor myelin repair via heparan sulphate production" for consideration by eLife. Your article has been reviewed by three peer reviewers, one of whom is a member of our Board of Reviewing Editors, and the evaluation has been overseen by Gary Westbrook as the Senior Editor. The following individual involved in review of your submission has agreed to reveal their identity: James W Fawcett (Reviewer #2).

The Reviewing Editor has drafted this decision to help you prepare a revised submission.

Summary

The authors have submitted a comprehensive revision of the manuscript, with several pieces of new data. These changes address the main issues raised by the reviewers, and the paper is considerably improved. However, the authors have not addressed the mechanistic underpinning of the proposed mechanism – i.e. the belt of HS synthesis and deposition in response to Ndst1 upregulation in oligodendrocytes around the lesion, and the proposal that binding of Shh promotes both, OPC proliferation and microglia polarization. Three points require further revisions.

1) We think a fairly simple IHC experiment with existing materials should provide direct mechanistic insight:

a) by showing that HS production around the LPC lesions is lost or reduced in the Ndst1 KO

b) by showing that after LPC lesion the Shh downstream targets in OPCs (and in other cells) are less activated in Ndst1 KO mice

c) by showing that M2 microglia are indeed increased in number.

2) The authors are also encouraged to check for Ndst1 upregulation in the available databases on single nuclei gene expression in MS tissue.

3) Finally, the role of microglia has been experimentally explored by additional experiments, but the possibility of non-cell autonomous effects on remyelination (by microglia) has not been adequately discussed, despite some reassurance in the rebuttal letter.

https://doi.org/10.7554/eLife.51735.sa1

Author response

[Editors’ note: the authors resubmitted a revised version of the paper for consideration. What follows is the authors’ response to the first round of review.]

Reviewer #1:

This paper is interesting because it provides to my knowledge the first report of a molecular response of intact oligodendrocytes close to a demyelinating lesion, the induction of an enzyme (Ndst1) that increases the expression of heparan sulfate and thus HS-modifed proteins. Based on the phenotype of OPC/OL-specific Ndst1 cKO mice the authors suggest that this oligodendroglial response limits the extent of demyelination and promotes remyelination, possibly by locally increasing sonic hedgehog signaling. The authors also show a similar upregulation of NDST1 within (not around) human MS lesions, where the expression level inversely correlates with lesion size.

Major points:

1) Since primary development and myelination are not perturbed in the brains of Ndst1 cKO mice, another possible explanation of the data is that enhanced activation of microglia (or their altered phenotype) affects remyelination in a non-cell autonomous way.

We fully agree with this interpretation of our results. As requested by reviewers, we have further clarified this point and better described microglia activation. We have performed new experiments to further characterize the proliferation and activation states of the macrophage/microglia participating in demyelination-remyelination in the acute demyelination model (See new version of Figure 5). We have shown:

1) An increase in proliferation of CD68+ cells in mutant compared to control mice at 8 dpi (Figure 5A-C).

2) A switch of the microglia/macrophage polarization among the whole Iba1 and CD68 population in favor of the rhomboid phenotype in mutant mice compared to control at 8 dpi (Figure 5H).

3) A significant increase in Cox2 expression (a marker of pro-inflammatory (M1) microglia/macrophage) in mutant mice compared to control, indicating a delay in the pro-inflammatory (M1) to pro-regenerative (M2) switch in the absence of Ndst1 in oligodendroglia.

These results demonstrate that Ndst1 deletion in the Olig2 population is sufficient to enhance microglia/macrophage proliferation and activation at the lesion site at the onset of remyelination. The implication of Microglia (in a non cell autonomous effect) is presented in the Discussion.

2) Thus, the Shh data are plausible but only correlations, and one should not leave the impression that reduced Shh binding was shown to cause the delay of remyelination.

Previous reports from our lab and others have shown that Shh is a strong candidate for myelin repair acting both on oligodendrogenesis and regulation of microglia activation (Ferent et al., 2013; Zakaria et al., 2019). Indeed, Shh is produced by OLG and OPC at the onset of demyelination in lysophosphatidyl choline (LPC)-induced lesions. In this context, blocking Shh activity induces an increase in lesion size and a block in OPC proliferation and differentiation, and conversely Shh overexpression leads to the attenuation of the lesion extent and promotes oligodendrogenesis. Microglia express Shh receptors (Ferent et al., 2013) and Shh treatment reduces the number of activated microglia in LPC model. Thus, Shh production by OLG lineage cells could act both directly on OPC (cell autonomous) and on microglia activation (non cell autonomous). Since Shh is bound and enriched at the border of the lesion in a HS dependent manner, it is likely that this morphogen contributes to the role of Ndst1/HS in the regeneration process.

We made the mistake of not presenting these results (Ferent et al….) in the Introduction of the paper but only in the Discussion. To better underline these key findings concerning Shh for the understanding of our work, we described these results and references in the Introduction of new version of the manuscript.

A second possible candidate could be FGF since HS can also modulate its signaling and since Ndst1 global knock out mice display developmental defects that mainly resemble those found in embryos deficient for Shh or FGFs. Concerning FGF recent work using the simultaneous ablation of both FGFR1 and FGFR2 specifically in OLG revealed that FGF signaling is not required for myelin regeneration. Overall, in all the analyses modulating FGF signaling in acute model of demyelination the phenotypes observed never recapitulate the phenotype observed in the present report using Olig2-Cre; Ndst1 Flox/Flox mice. This suggests that FGF is probably not the main ligand regulated by HS activity during myelin repair in this model. These data from the literature were absent from the first version of the manuscript and are now added in the Discussion.

3) It is unclear how "remyelination" in the human MS autopsy samples was assessed (pale blue staining could also be incomplete demyelination). In fact, this may explain the different pattern of Ndst1 expression in mouse and human demyelinating lesions.

“Shadow” plaques where there are patches of pale blue staining on luxol fast blue, suggesting thinner myelin sheaths, have long been suggested to be remyelinated lesions. This assumption is based on their similarity to remyelinated lesions seen in animal models. These lesions are not surrounded by many myeloid cells, and these do not contain myelin debris, again suggesting that these are at least not recently demyelinated. Thus, we believe these are repaired lesions but will call these lesions shadow plaques, to cover all possibilities including remyelination or partial demyelination.

4) Figure 3. It is unclear what the anti-PLP staining is actually labeling outside the lesion area (in Figure 3B), as PLP is still a myelin marker

There is a misunderstanding. In Figure 3 for plp labeling a double in situ hybridization staining was performed. We make it clearer in the text by writing “We found that Ndst1 expressing cells are immuno-positive for Olig2 (Figure 3A) and Plp+ using double in situ hybridization (Figure 3B)”.

5) Figure 1 C-H: In-situ hybrization of Ndst1 mRNA in demyelinaing lesions: these observations are critical. but lack quantification (required standard). The alignment of the "lesion" with the "belt" is not obvious when using only dashed lines. Why is Ndst1 expression not shown also as an overview type image (similar to Figure 1—figure supplement 2)? Same is true for Figure 2, which also lacks quantitation. While the "estimated" level of upregulation is stronger than in Ndst1 in Figure 1B, the IHC images are (at present) only case reports.

Concerning quantitative analysis. At this stage of the work (Figure 1 and 2) we describe the expression pattern of ndst1 and HS in the lesioned brain. Both are virtually absent in the healthy corpus callosum as well as in the contralateral part after lesion using ISH and IF. We thus decided to perform quantitative analysis using RT-qPCR, to compare ndst1 expression in the corpus callosum of healthy or demyelinated mice using 5 mice in each condition. These quantitative data are presented in Figure 1. In a second step, we characterized the phenotype of cells expressing ndst1 and performed quantitative analysis showing that 100% of ndst1 positive cells are Olig2 positive.

As mentioned in the manuscript Ndst1 expression is monitored by in situ hybridization. This technic is not compatible with good quality myelin immunostaining. Thus, in order to delimit precisely lesions in the LPC model, we use Hoechst staining since the lesion is also characterized by an increase in cell density due to glia and microglia proliferation. In order to validate this approach we have performed injection of LPC in a reporter mouse (plp-GFP) and show that the limit of the lesion visualized by loss of GFP fluorescence strictly corresponds with the area exhibiting strong increase in cell density. These data are presented in Figure 1—figure supplement 2. Thus the alignment of the lesion was based on cell density after in situ hybridization.

Reviewer #2:

This paper examines the role of heparan sulphate in remyelination by examining the expression of a key enzyme NDST1 and the consequences of an NDST1 conditional knockout in a mouse model of remyelination. In addition, the authors examine tissue from patients with MS. They conclude that HS containing proteoglycans are involved in remyelination as the enzyme is expressed around demyelinating lesions and its loss delays remyelination as evidenced by an increase in lesion size. Moreover there is a correlation between the degree of expression and the degree of remyelination in MS tissue.

While I think that the role of proteoglycans in remyelination is interesting, this paper is predominantly descriptive and correlative and doesn't provide sufficient mechanistic detail. The evidence for SHH binding in the region of putative HS expression is noteworthy but it is purely correlative. Equally the increase in microglial proliferation is interesting but not explored further.

My view would be that experiments are required to define one or more mechanisms by which HS is involved in remyelination, with gain and/or loss of function experiments clearly proving that the proposed mechanism is both linked to HS expression and regulates oligodendrocyte behaviour during remyelination in vivo.

Additionally I think there are three other areas that require attention.

First, the authors argue that the distribution around the experimental lesions is significant and that they have uncovered a novel function for surviving oligodendrocytes around lesions. It's notable however that they do not see this distribution in the MS tissue and I wonder whether their results simply reflect the lack of any surviving oligodendrocytes following LPC injection. In other words the distribution they see is simply a consequence of the model rather than having any great biological significance

First, as shown in Figure 1—figure supplement 1, we have observed the same pattern of Ndst1 expression in the EAE model were Ndst1 is up-regulated by the Olig2+ cell population in close proximity to inflammation sites in corpus callosum. Since we have observed this expression in both mouse models (LPC and EAE), it is probably relevant.

Concerning the distribution of Ndst1 expression in the LPC model: Even though OLG are massively killed within the lesion, remaining OLG in close proximity to the lesion do express ndst1 while OLG remote from the lesion in the CC do not. This shape “as a belt” is may be linked to the model but the presence of reactive mature OLG ndst1+ at the lesion site observed in all models including human is biologically relevant.

This point has been discussed in the new version of the manuscript.

Second, the characterization of the OPC response in the presence or absence of HS is very poor and studies of proliferation and differentiation, including the quantification of newly formed oligodendrocytes, is required

We have performed new experiments presented now in Figure 4 in the new version of the manuscript. We performed a quantification of proliferating of OPC (Ki67+/olig2+ cells) in lesion sites at 4 and 8 dpi. We show that the dynamic of OPC proliferation is altered in mutant mice. Overall, Our data show that Ndst1 expression in the Olig2+ population has no effect on initial demyelination (equivalent lesion size at 4 dpi) but protects the lesion from enlarging and participates in the control of OPC mobilization. We have included these data in the new version of the manuscript.

Third, a better characterization of the microglial phenotype (e.g. M1/M2 or similar) is required if the significance of this response is to be understood

See response to reviewer 1. We have performed new data showing that Ndst1 deletion in the Olig2 population is sufficient to enhance microglia/macrophage proliferation and activation at the lesion site at the onset of remyelination. We demonstrate accentuated M1 phenotype in mutant mice compared to wild-type mice.

Reviewer #3:

This is a thorough and interesting account of the role of HSPGs in remyelination. Many mechanisms of remyelination are known, but the idea that a particular sulfation change in HSPGs could be important is new. There is also data from human MS pathology in the paper, which gives clinical relevance.

1) The paper focuses on NDST which gives a particular form of sulfation to HS. There needs to be some information about the sulfation motif that is produced and any known information on relevant molecules that might bind to this form of HS. Shh is binding is shown, but it is unlikely to be the only effector of the local HS change.

We have now included in the new version of the manuscript a full discussion concerning which factors could bind to HS in this context.

Concerning Shh binding, we have used an AP tagged version of Shh (AP-SHH) to directly assay its binding capacity in demyelinating context. We believe that an important piece of information is the data presented using the AP-SHH recombinant proteins deleted for the CW sequence (AP-SHH-CW deleted) since CW sequence serves as a major HS-binding site for Shh (Carrasco et al., 2005; Rubin et al., 2002). This sequence is necessary for Shh binding to HS.

Concerning other factors and in particular FGF signalling, please see response 2 to reviewer 1.

2) If there is an increase in NDST, do the authors have information about any HS core proteins that might also be upregulated?

We have preliminary data concerning this issue. We compared by qRT-PCR the expression level of transcripts of the main proteoglycan HS transcripts in mouse CC seven days after injection of the LPC and showed that only glypicans 4 and 5 are over-expressed after injury. Nevertheless, we would rather not include these data in the work since there are several carriers, Glypicans, Syndecans and perlecans, and since their activity is also depending on the fact that some are membrane bound others are soluble. Here focusing on Ndst1 we work on HS “as a whole” and would rather keep this level of complexity. However, these data can be provided and included in the manuscript as a supplementary figure if requested by the reviewer.

3) Why would a sulfation change in mature oligos increase myelination? Surely OPCs would be more relevant?

As discussed in the manuscript we believe that HS on OLG changes the amount/stability of soluble factors, and thus the recruitment and proliferation of surrounding cells (such as OPC or microglia). Thus, OLG trigger non cell autonomous effects by generating a matrix (environment) that in turn regulates the behavior of cells in the lesioned CC. This point is presented in the Discussion.

4) The authors show that there is some increase in Shh binding around lesions, but not that it does anything. The statement in the Discussion needs to be modified. Also they do not show local upregulation. Any published information that might link Shh or Shh inhibitor to remyelination needs to be stated. This part of the paper needs attention.

The data concerning pattern of expression and role of shh during demyelination and myelin repair are already published and were mentioned in the Discussion of the first version of the manuscript. To be more accurate and better present these data we have now included a sentence in the Introduction “Interestingly oligodendrocyte lineage cells also produce factors that can modulate remyelination like the morphogen Shh which is produced by OLG and OPC at the onset of demyelination in lysophosphatidyl choline (LPC)-induced lesions (Ferent et al., 2013). In this context, blocking Shh activity induces an increase in lesion size and a block in OPC proliferation and differentiation, and conversely Shh overexpression leads to the attenuation of the lesion extent and promotes oligodendrogenesis” These data are also included in the Discussion. See also response 2 to reviewer 1.

5) Do the authors believe that the changes they see in NDST etc. are specific to demyelination, or are similar changes seen with traumatic lesions and inflammation?

As mentioned below and shown in Figure 1—figure supplement 1, we have observed the same pattern of Ndst1 expression in the EAE model. It would indeed be interesting to explore non-demyelinating models in the context of another study.

6) Figure 4: Is the increase in cell proliferation in the knockouts entirely accounted for by increased division of microglia?

We have performed co-immunolabelling of Olig2 and Ki67 showing a significant increase of OPC proliferation mutant mouse compared to control 8dpi. These results are now included in the new version of the manuscript.

7) Figure 6 shows some increase in Shh binding around the lesion, but also widespread binding elsewhere. Some comment is warranted.

Indeed as described in the literature Shh receptors are expressed in various cortical populations in the brain (Harwell et al., Neuron 2012 Mar 22;73(6); Shikawa et al., Dev Biol. 2011 Jan 15;349(2):147-59). As shown in Figure 6B of the presented manuscript AP-SHH probe binds on both size of the brain in the Cortex independently of lesion side.

[Editors’ note: what follows is the authors’ response to the second round of review.]

The authors have submitted a comprehensive revision of the manuscript, with several pieces of new data. These changes address the main issues raised by the reviewers, and the paper is considerably improved. However, the authors have not addressed the mechanistic underpinning of the proposed mechanism – i.e. the belt of HS synthesis and deposition in response to Ndst1 upregulation in oligodendrocytes around the lesion, and the proposal that binding of Shh promotes both, OPC proliferation and microglia polarization. Three points require further revisions.

1) We think a fairly simple IHC experiment with existing materials should provide direct mechanistic insight:

a) by showing that HS production around the LPC lesions is lost or reduced in the Ndst1 KO

All the requested experiment could not be performed on existing materials since tissue preparation is different depending on the antibodies used; for example, heparan sulfate immunostaining had to be performed on fresh frozen samples while microglia staining requested strong fixative procedure. We had thus to produce new cohorts of mutant mice and perform LPC-induced demyelination to satisfy the reviewer’s requests, explaining the delay in our response.

b) by showing that after LPC lesion the Shh downstream targets in OPCs (and in other cells) are less activated in Ndst1 KO mice

Despite numerous tests, we have reached the conclusion that shh signaling pathway activation cannot be monitored by immunostaining since there are no good antibodies against shh downstream targets. We contacted various experts including Dr Traifford, Dr Zeller and Dr Dahmane who confirmed these points.

We thus used RNAscope technic to perform comparative analysis of patched1 expression in wild-type and mutant animals. We could show that Ptch1 expression is decreased around LPC-induced demyelination lesions at 8 dpi in mutant animals compared to controls.

These data are now included in a new version of Figure 6 and a paragraph has been added in the Results section.

c) by showing that M2 microglia are indeed increased in number.

We believe that the request is concerning M1 Microglia which is the population we have observed to be affected in mutant condition. We gained information on microglia phenotype using Cox2 immunostaining, Cox2 being a pro-inflammatory enzyme thus labeling M1 microglia. We show a significant increase of Cox2+ cells /mm2 in Ndst1 KO mice 8 days post lesion indicating that the number of M1 microglia is increased in mutant mice after demyelination.

2) The authors are also encouraged to check for Ndst1 upregulation in the available databases on single nuclei gene expression in MS tissue.

Interrogation of our Jaekel et al., 2019 single nuclei RNAseq dataset shows few nuclei that express NDST1 RNAs generally, which is likely due to the dropseq platform only identifying around 15% of the nuclear RNAs, and so favors RNAs with high expression. This is also very similar on interrogation of the Rowitch dataset from human brain.

Therefore, there are too few nuclei here expressing NDST1 to be able to be confident. However, with this caveat, if the cells that express NDST1 are compared, there is a trend to increased expression in MS tissue, which also holds true if only oligodendroglia expressing NDST1 are compared, supporting our biological conclusions. We added part of these data to the manuscript as Figure 8—figure supplement 1.

3) Finally, the role of microglia has been experimentally explored by additional experiments, but the possibility of non-cell autonomous effects on remyelination (by microglia) has not been adequately discussed, despite some reassurance in the rebuttal letter.

A paragraph has been added in the Discussion.

https://doi.org/10.7554/eLife.51735.sa2

Article and author information

Author details

  1. Magali Macchi

    Aix Marseille Univ, CNRS, IBDM, Marseille, France
    Contribution
    Conceptualization, Formal analysis, Investigation, Methodology, Writing - original draft
    Contributed equally with
    Karine Magalon
    Competing interests
    No competing interests declared
  2. Karine Magalon

    Aix Marseille Univ, CNRS, IBDM, Marseille, France
    Contribution
    Conceptualization, Formal analysis, Investigation, Methodology, Writing - review and editing
    Contributed equally with
    Magali Macchi
    Competing interests
    No competing interests declared
  3. Céline Zimmer

    Aix Marseille Univ, CNRS, IBDM, Marseille, France
    Contribution
    Conceptualization, Formal analysis, Writing - original draft
    Competing interests
    No competing interests declared
  4. Elitsa Peeva

    MRC Centre for Regenerative Medicine, Multiple Sclerosis Society Centre for Translational Research, University of Edinburgh, Edinburgh, United Kingdom
    Contribution
    Formal analysis
    Competing interests
    No competing interests declared
  5. Bilal El Waly

    Aix Marseille Univ, CNRS, IBDM, Marseille, France
    Contribution
    Formal analysis
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0003-2991-3754
  6. Béatrice Brousse

    Aix Marseille Univ, CNRS, IBDM, Marseille, France
    Contribution
    Formal analysis
    Competing interests
    No competing interests declared
  7. Sarah Jaekel

    MRC Centre for Regenerative Medicine, Multiple Sclerosis Society Centre for Translational Research, University of Edinburgh, Edinburgh, United Kingdom
    Contribution
    Formal analysis
    Competing interests
    No competing interests declared
  8. Kay Grobe

    Institute of Physiological Chemistry and Pathobiochemistry and Cells-in-Motion Cluster of Excellence (EXC1003-CiM), University of Münster, Münster, Germany
    Contribution
    Formal analysis, Validation, Writing - original draft
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-8385-5877
  9. Friedemann Kiefer

    Max Planck Institute for Molecular Biomedicine, Münster, Germany
    Contribution
    Resources
    Competing interests
    No competing interests declared
  10. Anna Williams

    MRC Centre for Regenerative Medicine, Multiple Sclerosis Society Centre for Translational Research, University of Edinburgh, Edinburgh, United Kingdom
    Contribution
    Formal analysis, Supervision, Investigation, Methodology, Writing - original draft, Writing - review and editing
    Competing interests
    No competing interests declared
  11. Myriam Cayre

    Aix Marseille Univ, CNRS, IBDM, Marseille, France
    Contribution
    Conceptualization, Formal analysis, Supervision, Validation, Investigation, Methodology, Writing - original draft, Writing - review and editing
    Competing interests
    No competing interests declared
  12. Pascale Durbec

    Aix Marseille Univ, CNRS, IBDM, Marseille, France
    Contribution
    Conceptualization, Supervision, Funding acquisition, Validation, Methodology, Project administration
    For correspondence
    pascale.durbec@univ-amu.fr
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-9660-1809

Funding

Centre National de la Recherche Scientifique

  • Pascale Durbec

Aix-Marseille Université (Graduate student fellowship)

  • Pascale Durbec

Fondation pour la Recherche Médicale (DEQ20140329501)

  • Pascale Durbec

Agence Nationale de la Recherche (France-bioimaging/PICSL infrastructure ANR-10-INSB-04-01)

  • Pascale Durbec

Agence Nationale de la Recherche (ANR-15-CE16-0014-01)

  • Pascale Durbec

AM*DEX NeuroMarseille Institute (AMX-19-IET-004)

  • Pascale Durbec

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Acknowledgements

We are grateful to E Traiffort, F Helmbacher and C Bertet for critical reading of the manuscript.

Ethics

Human subjects: Human postmortem unfixed frozen tissues were obtained from the UK Multiple Sclerosis Tissue Bank via a UK prospective donor scheme with full ethical approval (MREC/02/2/39).

Animal experimentation: All experimental and surgical protocols were performed following the guidelines established by the French Ministry of Agriculture (Animal Rights Division). The architecture and functioning rules of our animal house, as well as our experimental procedures have been approved by the 'Direction Départementale des Services Vétérinaires' and the ethic committee (ID numbers F1305521 and 2016071112151400 for animal house and research project.

Senior Editor

  1. Gary L Westbrook, Oregon Health and Science University, United States

Reviewing Editor

  1. Klaus-Armin Nave, Max Planck Institute of Experimental Medicine, Germany

Reviewer

  1. James W Fawcett, University of Cambridge, United Kingdom

Publication history

  1. Received: September 17, 2019
  2. Accepted: June 9, 2020
  3. Accepted Manuscript published: June 9, 2020 (version 1)
  4. Version of Record published: June 22, 2020 (version 2)

Copyright

© 2020, Macchi et al.

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.

Metrics

  • 935
    Page views
  • 148
    Downloads
  • 1
    Citations

Article citation count generated by polling the highest count across the following sources: Crossref, PubMed Central, Scopus.

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Download citations (links to download the citations from this article in formats compatible with various reference manager tools)

Open citations (links to open the citations from this article in various online reference manager services)

Further reading

    1. Neuroscience
    Igor Gridchyn et al.
    Research Article Updated

    In vitro work revealed that excitatory synaptic inputs to hippocampal inhibitory interneurons could undergo Hebbian, associative, or non-associative plasticity. Both behavioral and learning-dependent reorganization of these connections has also been demonstrated by measuring spike transmission probabilities in pyramidal cell-interneuron spike cross-correlations that indicate monosynaptic connections. Here we investigated the activity-dependent modification of these connections during exploratory behavior in rats by optogenetically inhibiting pyramidal cell and interneuron subpopulations. Light application and associated firing alteration of pyramidal and interneuron populations led to lasting changes in pyramidal-interneuron connection weights as indicated by spike transmission changes. Spike transmission alterations were predicted by the light-mediated changes in the number of pre- and postsynaptic spike pairing events and by firing rate changes of interneurons but not pyramidal cells. This work demonstrates the presence of activity-dependent associative and non-associative reorganization of pyramidal-interneuron connections triggered by the optogenetic modification of the firing rate and spike synchrony of cells.

    1. Chromosomes and Gene Expression
    2. Neuroscience
    Sandra Fendl et al.
    Tools and Resources

    Neurotransmitter receptors and ion channels shape the biophysical properties of neurons, from the sign of the response mediated by neurotransmitter receptors to the dynamics shaped by voltage-gated ion channels. Therefore, knowing the localizations and types of receptors and channels present in neurons is fundamental to our understanding of neural computation. Here, we developed two approaches to visualize the subcellular localization of specific proteins in Drosophila: The flippase-dependent expression of GFP-tagged receptor subunits in single neurons and 'FlpTag', a versatile new tool for the conditional labelling of endogenous proteins. Using these methods, we investigated the subcellular distribution of the receptors GluClα, Rdl, and Dα7 and the ion channels para and Ih in motion-sensing T4/T5 neurons of the Drosophila visual system. We discovered a strictly segregated subcellular distribution of these proteins and a sequential spatial arrangement of glutamate, acetylcholine, and GABA receptors along the dendrite that matched the previously reported EM-reconstructed synapse distributions.