Ciliary Neurotrophic Factor (CNTF) Enhances Myelin Formation: A Novel Role for CNTF and CNTF-Related Molecules

In myelinating cocultures derived from 1900bp-MBP lacZ transgenic brains, assay of β-galactosidase enzymatic activity provides a reliable index of myelination

In cocultures derived from embryonic 1900bp-MBP lacZcerebral hemispheres, oligodendroglial differentiation and myelination occurred with the same timing as described previously for similar cultures derived from wild-type animals (Lubetzki et al., 1993;Demerens et al., 1996). After 11–12 DIV, the oligodendroglial population consisted of O4⁺preoligodendrocytes (Sommer and Schachner, 1981) and GalC-expressing immature oligodendrocytes labeled with the R-mAb (Ranscht et al., 1982). The first MBP-expressing cells were detected at 12–14 DIV. More mature MOG⁺ oligodendrocytes, characterized by their pauci-branched arborization, started to be observed at 15 DIV. The drastic change in morphology observed between 12 and 15 DIV corresponds to the transition between a mature non-myelin-forming cell (MOG⁻) (Fig.1A,C) and a myelinating phenotype (MOG⁺) (Fig.1B) (Solly et al., 1996). Combination of X-gal staining, to detect β-galactosidase enzymatic activity, and anti-MBP immunolabeling demonstrated that, in myelinating cocultures derived from 1900bp-MBP lacZ animals, transgene expression was restricted to mature pauci-branched oligodendrocytes, either myelinating (Fig. 1E) or pseudo-myelinating (Fig.1D). Pseudo-myelinating oligodendrocytes had not established contact with axons but are MOG⁺ cells, which have formed whorls of poorly compacted membranes at the tip of their processes (Fig.1B,D), as shown previously by electron microscopy (Lubetzki et al., 1993). In addition, all β-galactosidase-positive cells were also MOG⁺ (data not shown). In contrast, MOG⁻ multibranched non-myelin-forming cells (Fig. 1A,B) were never stained with the X-gal substrate, and thus did not express detectable levels of β-galactosidase (Fig.1A–C).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 1.

In cultures derived from 1900bp-MBP lacZ fetuses, β-galactosidase expression is restricted to myelinating oligodendrocytes. Dissociated cultures from E151900bp-MBP lacZ mouse brain cultured for 3 weeks were doubly stained with anti-GalC (A) and anti-MOG (B) antibodies or with anti-MBP antibodies and X-gal substrate to reveal β-galactosidase enzymatic activity (C, D, E). A, B, The same field was photographed with fluorescein (A) and rhodamine (B) optics. In A, two GalC⁺ cells are seen (green): the cell in the bottom left has the typical multibranched morphology of a mature non-myelin-forming oligodendrocyte. This cell is MOG⁻ (B). In contrast, in thetop right, another GalC⁺ cell is also MOG⁺ (red). This GalC⁺/MOG⁺ cell (arrowhead points to the cell body) is identified as a pseudo-myelinating oligodendrocyte based on its poorly branched morphology and the presence of whorls of myelin-like figures at the tips of its processes (arrows). C, A mature non-myelin-forming oligodendrocyte with the typical “sun-like” morphology is MBP⁺(brown) but X-gal-negative. D, A pseudo-myelinating oligodendrocyte with characteristic whorls of myelin-like figures (arrows) is MBP⁺(brown) and X-gal-positive (blue stainingof the cell body). E, A typical field of fibers myelinated by a single oligodendrocyte. This myelinating oligodendrocyte is MBP⁺/X-gal⁺. Note that the myelinated internodes are strongly MBP⁺, whereas the X-gal⁺ cell body appears MBP⁻ because in myelinating oligodendrocytes most of the MBP migrates out of the cell body and MBP immunoreactivity is mostly confined to the myelin sheath. Scale bars, 10 μm.

β-galactosidase enzymatic activity was assayed and compared with the stage of differentiation of the oligodendroglial population (Fig.2). As expected, no β-galactosidase enzymatic activity was detected before myelin deposition and/or MOG expression: β-galactosidase activity was first detected at 15 DIV and then increased threefold and eightfold at 19 and 25 DIV, respectively. This increase paralleled myelin formation in the cultures, as evaluated by the increase in the number of myelinated internodes (Fig.2A). During the same period, the ratio of GalC⁺ oligodendrocytes double-labeled with anti-MOG mAb increased from 15 to 39%, whereas the number of O4⁺ cells remained relatively stable. Because GalC⁺ and MOG⁺ cells still retain the expression of O4-recognized antigen (Bansal et al., 1989), the stability of the O4⁺ population suggested that after 11 DIV few if any, new oligodendrocyte precursor cells were generated under our culture conditions (Fig. 2B). These data demonstrated that in the 1900bp-MBP lacZ transgenic line, the increase in the level of β-galactosidase between 15 and 25 DIV in the absence of growth factor specifically reflects the increase in myelin formation and not the generation of new oligodendrocytes. Therefore, quantification of β-galactosidase enzymatic activity could be used as a reliable index of myelination in control situations.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 2.

Assay of β-galactosidase enzymatic activity in myelinating cultures derived from 1900bp-MBP lacZprovides an index of myelination. A, β-galactosidase activity (expressed in counts per minute per nanogram of protein) was assayed as a function of time in vitro. In cultures derived from 1900bp-MBP lacZ, β-galactosidase activity started to be detected at 15 DIV, whereas no activity was detectable in cultures derived from nontransgenic OF1 animals. The intensity of myelination in sister cultures was evaluated by counting the number of myelinated MBP positive internodes and is indicated by + (onset), ++ (moderate), and +++ (maximum). B, Sister cultures were also immunostained with O4, and either anti-GalC or anti-MOG mAbs. Note that the number of O4⁺ cells remained constant between 11 and 25 DIV, whereas the number of GalC⁺cells increased during the same period. MOG⁺ cells were not detected before 15 DIV, and the increase in their number paralleled the increase in β-galactosidase activity (A). Results are expressed as the number of positive cells per field (objective, 40×) (means ± SEM of cell count from 6 fields per culture, with 4–6 cultures per experiment).

Effect of different neurotrophic and growth factors on myelination

We used this assay to screen for chemokines or cytokines that could play a role in myelin formation. Based on the family of receptors they activate, the factors investigated were classified into four groups: (1) neurotrophins (NGF, BDNF, NT-3, and NT-4/5) acting through the Trk family of receptors; (2) GDNF and NTU, which are growth factors acting through Ret receptors; (3) PDGF-AA, FGF-2, or insulin, which mediate their action by binding to tyrosine kinase receptors; and finally (4) members of the CNTF family of ligands, which interact with gp130-containing receptors. The factors to be tested were added to the culture medium of myelinating cocultures, derived from embryonic day 15 (E15) 1900bp-MBP lacZ transgenic animals between 11 and 25 DIV, and myelin formation was quantified by assaying β-galactosidase enzymatic activity at 25 DIV. No promyelinating effect was observed when neurotrophins (10 ng/ml) were added to the cultures (Fig. 3A). Neurotrophins were also used at concentrations varying between 0.1 and 50 ng/ml, without any effect (data not shown). Similarly, GDNF, NTU (Fig. 3B), PDGF-AA, FGF-2 (each at 10 ng/ml), and insulin (100 μg/ml) (Fig. 3C) did not promote myelin formation. In contrast, CNTF dramatically increased myelin formation: the mean ± SEM β-galactosidase activity was 5.4 ± 0.9-fold higher in cultures treated with CNTF (10 ng/ml) than in untreated control cultures. The treatment of cultures with concentrations of CNTF varying between 0.01 and 50 ng/ml showed a clear dose response (Fig.3D). The CNTF-related factors LIF, CT-1, OsM, IL-6, and IL-11 were also tested. LIF, CT-1, and OsM had a promyelinating effect of the same amplitude as CNTF. When used at 10 ng/ml each, the increase in myelin formation was 5.9 ± 0.9-fold for CT-1, 4.6 ± 0.6-fold for LIF, and 4.8 ± 0.3-fold for OsM. This effect was clearly dose dependent (Fig. 3D). We did not observe any additive or synergistic action when these factors were added together (data not shown); addition of either neurotrophins or GDNF to CNTF-treated cultures did not potentiate the promyelinating effect of CNTF (data not shown). In contrast, IL-11 (2.3 ± 0.3-fold increase) or IL-6 (no increase) had little or no effect on myelination (Fig. 3D). The lack of effect of IL-6 was not attributable to species specificity, because we evaluated both human- and mouse-derived IL-6 with the same negative result.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 3.

Effect on myelination of different neurotrophic and growth factors. Myelinating cultures derived from 1900bp-MBP lacZ embryos were treated between 11 and 25 DIV; β-galactosidase activity was assayed at 25 DIV. A–C, Members of the neurotrophin (A) and GDNF (B) families and growth factors (C) were used at a concentration of 10 ng/ml, except for insulin, which was used at 100 μg/ml. D, Members of the CNTF family were tested at concentrations varying between 0.01 and 50 ng/ml. For each culture well, myelin formation was assessed by the quantification of the β-galactosidase level normalized to protein level (expressed as counts per minute per nanogram of protein). Results are expressed as the percentage of mean control values. A–C, Results are means ± SEM of three independent experiments with four to six cultures per experiment.D, Results are the means ± SEM of six cultures from one representative experiment.

Promyelinating effect of CNTF is mediated by favoring oligodendrocyte maturation

To determine whether the promyelinating effect of CNTF was related to an effect on oligodendrocyte maturation, we evaluated the proportion of GalC⁺ oligodendrocytes coexpressing the maturation marker MOG. Compared with controls, treatment with CNTF resulted in a 1.9-fold increase in the percentage of GalC⁺/MOG⁺mature oligodendrocytes (Fig.4A). In addition, in separate experiments, the percentage of MOG⁺ cells among total O4⁺ oligodendrocytes was increased 1.8-fold in CNTF-treated cultures (36.6 ± 4.8% vs 66.9 ± 0.8% in control and treated cultures, respectively). The promyelinating effect observed with CNTF was not attributable to a selective activation of the 1900bp-MBP promoter, because a 3.7 ± 0.3-fold increase in β-galactosidase activity was observed after the addition of CNTF to cultures derived fromplp-lacZ mice, another transgenic line, in which thelacZ transgene is under the control of the proteolipid protein (PLP) promoter (Spassky et al., 1998) (Fig.4B). Finally, the promyelinating effect of CNTF detected by β-galactosidase assay was then confirmed by direct quantification of the extent of axon wrapping. In these latter experiments, an increase in the number of myelinated internodes was indeed observed for concentrations of CNTF of 0.1 and 1 ng/ml (Fig.4C). Interestingly, the effect was less pronounced for higher concentrations of CNTF, suggesting that a strong upregulation of myelin gene transcription could compromise myelin formation or maintenance.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 4.

The promyelinating effect of CNTF is mediated by enhancing oligodendrocyte maturation. Myelinating cultures derived from1900bp-MBP lacZ (A–C) orplp-lacZ (PLP-lacZ) embryos (B) were treated with CNTF and analyzed at 25 DIV for MOG expression (A), β-galactosidase activity (B), or axon wrapping (C). A, Dissociated cultures were doubly stained with anti-GalC and MOG mAb, and the percentage of GalC⁺ cells expressing MOG was quantified in CNTF-treated (10 ng/ml, 11–25 DIV) and control cultures. Results are expressed as the means ± SEM of one representative experiment (determination from 6 different culture wells per condition).BS, Bottenstein and Sato medium. B, β-galactosidase activity, expressed as a percentage of control untreated cultures, was assayed in CNTF-treated (11–25 DIV) cultures derived from either 1900bp-MBP lacZ orplp-lacZ transgenic embryos. Results are the means ± SEM of three independent experiments representing four to six cultures per experiment. ***p < 0.001; Student'st test. C, The number of myelinated internodes was counted in control and CNTF-treated cultures. Results are expressed as means ± SEM (determination of 5 different culture wells per condition).

Cytokines activating the gp130/Janus kinase pathway stimulate CNS myelination

CNTF, LIF, OsM, and CT-1 use receptors containing LIF receptor (LIFR)/gp130 heterodimers, whereas IL-6 and IL-11 have been shown to act through gp130 homodimer-containing receptors. To demonstrate unambiguously that a promyelinating action could be obtained independently of the LIFR unit, we conducted additional experiments in which IL-6 and its soluble α-subunit receptor were added together. Whereas neither IL-6 nor the soluble α subunit alone induced any effect on myelin formation, the simultaneous addition of IL-6 and soluble α-subunit receptor caused a fivefold increase in β-galactosidase activity, but only when IL-6 was used at a concentration of 50 ng/ml, and this activity was no longer detected at 10 ng/ml (Fig. 5A).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 5.

The promyelinating effect of CNTF is signaled by gp130-containing receptors. A, Cultures were treated with IL-6, the IL-6-soluble receptor α subunit (sR-IL-6), or IL-6 +sR-IL-6 at either 10 or 50 ng/ml each. Results are expressed as a percentage of control β-galactosidase levels (means ± SEM of 3 independent experiments with 4–6 different cultures per experiment). B, CNTF (10 ng/ml) and JAK inhibitor AG490 (AG) were added daily into the culture medium between 14 and 18 DIV, and β-galactosidase was assayed at 19 DIV. β-galactosidase levels are expressed as means ± SEM (6 different culture wells for each condition) of one representative experiment. ***p < 0.001; Student'st test.

To determine whether the promyelinating effect of CNTF was mediated through the JAK, we attempted to block this transduction pathway using the JAK inhibitor AG490 (Meydan et al., 1996). Simultaneous addition of increasing concentrations of AG490 to CNTF (10 ng/ml) resulted in a dose-dependent inhibition of the promyelinating effect of CNTF, whereas AG490 alone had no influence on β-galactosidase activity (Fig. 5B).

In the 1900bp-MBP lacZ transgenic, assay of β-galactosidase enzymatic activity is a reliable and simple alternative to quantify myelin formation

In neuron–oligodendrocyte coculture, myelination was previously quantified by counting the number of either myelinated internodes or myelinating and pseudo-myelinating cells (Lubetzki et al., 1993;Demerens et al., 1996, 1999; Charles et al., 2000). However, this system of quantification is not suitable to screen for a large number of molecules influencing myelination. In the 1900bp-MBP lacZtransgenic mouse, we have shown previously that expression of the reporter transgene is turned on after MBP expression and only when oligodendrocytes start to myelinate (Stankoff et al., 1996). Here we show that in myelinating cultures derived from 1900bp-MBP lacZ embryos, expression of the transgene is restricted to myelinating and pseudo-myelinating oligodendrocytes (Fig. 1). Pseudo-myelinating oligodendrocytes are mature oligodendrocytes that have not established an axonal contact in vitro. Our observation that the reporter gene is expressed by both myelinating and pseudo-myelinating cells makes it a powerful tool to screen for molecules that favor the final maturation of oligodendrocytes. Because the assay of β-galactosidase is easy and very sensitive, quantification of this enzymatic activity in myelinating cocultures derived from 1900bp-MBP lacZ transgenic mice provides a simple and reliable index of myelin synthesis. Using this index, we provide evidence that only CNTF-related factors induce a promyelinating effect.

CNTF-induced increase in myelination is not mediated by an increase in proliferation or survival of oligodendrocyte precursor cells

Because CNTF has been shown to increase the proliferation (Barres et al., 1996) and survival (Barres et al., 1993; Louis et al., 1993;Mayer et al., 1994) of oligodendrocyte precursors, the observed effect on myelination could be the consequence of a higher number of oligodendrocytes in the cultures. This is very unlikely for several reasons. First, we have evaluated bromodeoxyuridine incorporation in glial cultures derived from newborn animals and found no variation in oligodendrocyte precursor proliferation in the presence or absence of CNTF (B. Stankoff, unpublished results). Second, other growth factors with well documented mitogenic properties on oligodendrocyte precursors, such as PDGF-AA (Richardson et al., 1988), FGF-2 (Bogler et al., 1990), and NT-3 (Barres et al., 1994; Cohen et al., 1996), had no effect on the β-galactosidase level.

It has also been reported that CNTF promotes the survival of newly differentiated oligodendrocytes in culture. In these studies, oligodendrocyte survival has been evaluated either in low-density cultures (Barres et al., 1993) or in cultures grown in a poor minimal medium (Mayer et al., 1994) or against tumor necrosis factor α-induced toxicity (Louis et al., 1993), whereas in our experiments, cultures were at high density and grown in B-S medium containing survival factors such as PDGF-AA (during the first week in vitro) and high concentrations of insulin. The stability of the O4⁺ population of cells between 11 and 25 DIV indicates that a significant number of deaths of either oligodendrocytes or oligodendrocyte precursors did not occur during this period. Moreover, insulin alone, which is also a survival factor for oligodendrocytes in low-density cultures (Barres et al., 1993), had no effect on β-galactosidase levels, and addition of insulin to CNTF-treated cultures did not increase myelination (data not shown).

Together, these data suggest that the increase in myelin formation induced by CNTF and CNTF-related factors is not the result of an increase in oligodendrocyte proliferation and survival, but relates to a specific effect on myelin synthesis. Furthermore, our observation that CNTF enhances the final maturation of oligodendrocytes by increasing the proportion of MOG⁺myelinating oligodendrocytes as well as the increase in β-galactosidase activity measured in cultures derived fromplp-lacZ transgenic mice are additional arguments in support of an effect of CNTF and CNTF-related molecules on myelin formation. In this respect, because the increase in MOG⁺cells is smaller than the increase in β-galactosidase expression, it is likely that in addition to an effect on oligodendrocyte maturation, CNTF could also increase myelin synthesis per oligodendrocyte. A similar increase was also observed in the number of myelinated internodes, confirming the effect on myelin formation per se. However, the lower magnitude of this effect could be attributable to the limited number of qualified axons permissive to myelination (Charles et al., 2000). Together, our results provide evidence of a new role for CNTF and CNTF-related molecules. It is possible that CNTF acts directly on oligodendrocytes. Alternatively, because our cultures contain astrocytes in addition to neurons and oligodendrocytes, we cannot exclude the possibility that the reported effect on myelin formation is indirect, via astrocytes or neurons.

Promyelinating effect of CNTF is mediated by the LIFRβ/gp130 complex

The promyelinating effect induced by CNTF and CNTF-related factors, such as LIF, CT-1, OsM, was of the same order of magnitude with no additive or synergistic action. This functional overlap between related cytokines suggests that their mechanism of action on myelination might involve a similar transducing pathway. These cytokines use receptors containing the same signal transducer gp130, which forms a heterodimer with the other related partner, LIFRβ (Kishimoto, 1994; Stahl and Yancopoulos, 1994; Turnley and Bartlett, 2000). LIF binds with high affinity to the LIFRβ/gp130 complex, whereas for CNTF, CT-1, and OsM, ligand specificity is conferred by a third α subunit, which forms a complex with the LIFRβ/gp130 heterodimer. In contrast, neither IL-6 nor IL-11 acts through the formation of an LIFRβ/gp130 heterodimer. Signaling of IL-6 requires binding to the IL-6 α-subunit receptor and homodimerization of gp130. For IL-11, two controversial modes of action have been proposed: formation of either a gp130 homodimer or a heterodimer of gp130 with an as yet unidentified IL-11-specific subunit (Yin et al., 1993;Neddermann et al., 1996). Because IL-6 is inducing a promyelinating effect when added with its soluble α-subunit receptor, this suggests that dimerization of gp130 is sufficient to mediate this effect. However, to enhance myelin formation, IL-6 had to be used at a concentration at least fivefold higher than for CNTF, LIF, CT-1, or OsM. Therefore, it is likely that under physiological conditions, the promyelinating properties of CNTF-related factors use the formation of LIFRβ/gp130 heterodimer.

Ligand binding followed by receptor complexing activates JAK (Stahl and Yancopoulos, 1994). The inhibition observed with the JAK inhibitor AG490 suggests that the CNTF promyelinating effect is signaled through the JAK pathway. Activation of JAK leads to docking of Src homology 2 domains of a variety of proteins, such as signal transducer and activator of transcription (STAT) proteins (Stahl and Yancopoulos, 1994; Segal and Greenberg, 1996). Activated STAT molecules dimerize and translocate to the nucleus, where they induce the expression of a variety of genes. Several JAK (JAK1, JAK2, Tyk2) and at least six different STAT proteins have been described. In addition to the JAK/STAT pathway, binding of CNTF-related factors can also activate the Ras mitogen-activated protein (MAP) kinase (Stahl et al., 1995;Giordano et al., 1997); in some cell types, signaling through phosphatidylinositol 3 (PI3) kinase has also been described previously (Oh et al., 1998). The different combination of JAK/STAT proteins and the alternative activation of MAP or PI3 kinases might explain the different types of responses induced in cells of the oligodendroglial lineage, depending on their developmental stages. For instance, because PDGF-AA and FGF-2 have been described as potent mitogens for oligodendrocyte precursors, and because signaling of these growth factors is mediated through the MAP kinase pathway, it can be postulated that CNTF-induced proliferation of oligodendrocyte precursors is also signaled by MAP kinase. PDGF-AA and CNTF also enhance oligodendrocyte precursor survival, and it has been shown that this survival effect is mediated by rapid tyrosine phosphorylation of JAK1, JAK2, STAT1α/β, and STAT3 (Dell'Albani et al., 1998). Therefore, it is tempting to speculate that, later during development, binding of CNTF to mature oligodendrocytes activates a different combination of JAK and STAT proteins, which in turn induce, or enhance, the coordinate expression of the subset of myelin-specific genes necessary to synthesize a quantity of membrane sufficient to form the myelin sheath.

Does CNTF favor myelination in vivo?

In addition to their effect on neurons, CNTF-related factors have been shown to enhance proliferation, survival, and differentiation of oligodendrocyte precursor cells (Barres et al., 1993; Louis et al., 1993; Mayer et al., 1994; Barres et al., 1996; Vos et al., 1996; Marmur et al., 1998). However, to our knowledge, this is the first time that CNTF-related factors have been shown to enhance myelination by acting on the last stages of oligodendrocyte maturation. Interestingly, in the rat optic nerve, it has been shown that astrocytes start to synthesize CNTF at the end of the first postnatal week (Stockli et al., 1991;Dobrea et al., 1992), which corresponds to the onset of myelination (Colello et al., 1995). This temporal concordance supports a physiological role of CNTF on myelin formation. Nevertheless, in CNTF-deficient mice both oligodendrocyte number and myelination attain wild-type values (Barres et al., 1993), presumably because of compensation by either the other gp130-stimulating cytokines or by the cytokine-like factor-1/cardiotrophin-like cytokine complex, which is the second ligand for the CNTF receptor (Elson et al., 2000). However, mice deficient for one of the main signaling elements involved in CNTF-related cytokine function, such as CNTF-α receptor, gp130, LIFRβ, or JAKs, die during development or perinatally (DeChiara et al., 1995; Li et al., 1995; Ware et al., 1995; Parganas et al., 1998;Rodig et al., 1998). Therefore, these mutants have failed to provide any models to investigate the role of the CNTF signaling pathway in myelinogenesis. However, the promyelinating effect of CNTF-related factors and the observation that they could also promote the differentiation of postnatal brain precursor cells in oligodendrocytes (Marmur et al., 1998) could be of therapeutic value in MS.