The Polarity Protein Par-3 Directly Interacts with p75NTR to Regulate Myelination

See allHide authors and affiliations

Science  03 Nov 2006:
Vol. 314, Issue 5800, pp. 832-836
DOI: 10.1126/science.1134069


Cell polarity is critical in various cellular processes ranging from cell migration to asymmetric cell division and axon and dendrite specification. Similarly, myelination by Schwann cells is polarized, but the mechanisms involved remain unclear. Here, we show that the polarity protein Par-3 localizes asymmetrically in Schwann cells at the axon-glial junction and that disruption of Par-3 localization, by overexpression and knockdown, inhibits myelination. Additionally, we show that Par-3 directly associates and recruits the p75 neurotrophin receptor to the axon-glial junction, forming a complex necessary for myelination. Together, these results point to a critical role in the establishment of cell polarity for myelination.

The myelin sheath is a specialized membrane component in the vertebrate nervous system that is essential for the optimal transmission of neuronal action potentials. In the peripheral nervous system, Schwann cells (SC) are responsible for myelinating axons. Recently, environmental signals, particularly the neuregulins (1) and the neurotrophins (2, 3), have been shown to regulate SC myelination. Specifically, neurotrophin 3 (NT-3) promotes SC migration and inhibits myelination (24). In contrast, brain-derived neurotrophic factor (BDNF) inhibits SC migration and promotes myelination through the p75 neurotrophin receptor (NTR) (2, 3, 5). Much less is known, however, about the intrinsic mechanisms governing SC myelination. The formation of myelin by SCs is a highly polarized process, which consists of the unidirectional wrapping of multiple layers of membrane concentrically around an axon, initiated exclusively at the site of the axon-glial junction, raising the question of what regulates the asymmetric initiation of myelination.

To address this question, we examined the distribution of Par-3, a member of the Par family of adaptor proteins involved in the establishment of cell polarity in various cellular contexts (615) in SC/dorsal root ganglion (DRG) neuronal cocultures (Fig. 1A) and in sciatic nerves (Fig. 1B). Interestingly, we found that Par-3 is enriched asymmetrically at the membrane of premyelinating SCs (Fig. 1, A and B) and did not colocalize with markers for the Golgi or the endoplasmic reticulum (fig. S1). This asymmetric localization of Par-3 formed a bandlike pattern at the site of contact between the Schwann cell and the axon, as revealed most clearly when staining for neurofilament to reveal the axons, which did not express detectable amounts of Par-3 (Fig. 1, C and D). To better visualize the axon-glial junction, we stained for N-cadherin, an adhesion protein enriched at the SC/axon junction (16). In the SC/DRG cocultures, we found that Par-3 and N-cadherin colocalize (Fig. 1E). Similar to previous findings with N-cadherin, the asymmetric localization of Par-3 was also identified in SCs cultured without neurons, but only when SCs made contact with one another (Fig. 1F, arrow). In the absence of contact, Par-3 localization was diffuse and uniform (Fig. 1F, arrowhead). Thus, it appears that the redistribution of Par-3 is a direct result of cell-cell contact, which can be either SC-SC or SC-axon contact, consistent with previous studies showing that cell-cell contact is necessary for asymmetric localization of Par proteins (10, 11, 15, 17, 18). To visualize the ultrastructural localization of Par-3 at the axon-glial junction, immunogold electron microscopy (EM) was performed on premyelinating SC/DRG cocultures. The asymmetric localization of Par-3 was identified by the electron-dense gold particles deposited in the SC at the site of the SC/axon junction in cross section (Fig. 1G) and in longitudinal section (Fig. 1H).

Fig. 1.

Asymmetric localization of Par-3 in SCs at the axon/glial junction. (A and B) Immunostaining for Par-3 protein (green) in premyelinating SCs cultured with DRG neurons (A) and in teased fibers of embryonic rat sciatic nerve at 18 days gestation (B). Note the asymmetric localization of Par-3 (arrows). (C and D) Par-3 protein (green, arrows) and neurofilament (red, arrowheads) in SC/DRG neuron cocultures (C) and in sciatic nerve sections from embryonic rats at 18 days gestation (D). Par-3 and neurofilament are juxtaposed to one another and Par-3 is absent in the axons. (E) Immunostaining for Par-3 (green) and N-cadherin (red) in SC/DRG neuron cocultures. The arrows point to the colocalization (yellow). (F) SCs cultured in the absence of neurons display asymmetric localization of Par-3 (red, arrow) only upon contact with other SCs. In the absence of SC-SC contact, diffuse and uniform Par-3 staining is observed (arrowhead). Nuclei were stained using Hoechst (blue). Scale bars, 10 μm. (G) Immunogold EM on premyelinating SC/DRG cocultures in cross section. Par-3 is enriched in SCs at the site of axon contact, as identified by the electron-dense gold particles (arrowheads). (H) A longitudinal section of a premyelinating SC and axon. Par-3 localization is enriched in the SC membrane ensheathing the axon, as identified by the electrondense gold particles (arrowheads). Scale bars, 500 nm.

Based on these results, we hypothesized that the asymmetric distribution of Par-3 at the axonglial junction might be required to initiate myelin formation by SCs. If this were the case, disruption of Par-3 localization by either overexpression or knockdown in SCs should inhibit myelination. We used a well-established SC/DRG coculture system and a retroviral expression system to manipulate Par-3 expression and assess the ability of SCs to myelinate DRG axons (see supporting online material). Because retroviral vectors infect only actively dividing cells, this method ensures that SCs are infected whereas the DRG neurons remain uninfected. When we overexpressed Par-3 in SCs, we disrupted the polarized asymmetric localization of Par-3 (fig. S2), and SCs were unable to myelinate DRG axons, as revealed by staining for myelin basic protein (MBP) (Fig. 2A) and other myelin markers. Non-infected SCs on the same axon formed myelin normally (Fig. 2A). We performed multiple experiments and could not identify any Par-3–overexpressing SCs forming mature myelin. Interestingly, Par-3–overexpressing SCs aligned on axons normally, as observed by phase contrast microscopy, and formed what looked like internodes (Fig. 2A, arrows).

Fig. 2.

Disruption of Par-3 localization and function by overexpression or knockdown inhibits SC myelination. (A) Control-infected SCs (top; green, arrowhead) formed myelin normally as detected by immunostaining for MBP (red). Arrows indicate the length of the myelin internode. Par-3–overexpressing cells (bottom; green, arrowhead) fail to form myelin. (B) The efficiency of the shRNA for Par-3 knockdown was examined in purified rat SCs. Cells were either not transfected (lane 1) or were transfected with a retroviral vector expressing GFP only (lane 2), a nonblocking shRNA for Par-3 (lane 3), or the experimental shRNA for Par-3 (lane 4); Par-3 protein expression was detected by Western blot. (C) The Par-3 shRNA retroviral vector was also tested by infecting SCs in cocultures, and 3 days later Par-3 expression was detected by immunostaining (red). Infected cells (green, arrowheads) show a dramatic reduction of Par-3 expression compared with adjacent noninfected cells. (D) SCs cocultured with DRG neurons were infected with either control or Par-3 shRNA retroviral vectors. Infected cells express GFP (green, arrowheads), and myelin is detected by staining with MBP (red). (E) Quantification of the Par-3 knockdown experiment was achieved by counting the percentage of infected SCs actively myelinating. Error bars, SD; *P < 0.005, unpaired Student's t test. Scale bars, 10 μm.

We next asked whether Par-3 was required for myelin formation by SCs. We generated retroviral vectors expressing both green fluorescent protein (GFP) and a short-hairpin RNA (shRNA) sequence for Par-3. When tested in SCs, the Par-3 shRNA significantly reduced Par-3 expression (Fig. 2, B and C), whereas a control shRNA directed against another region of Par-3 did not alter expression and neither did a control retroviral vector encoding GFP only (Fig. 2B). We infected SCs with these retroviral vectors and tested the ability of SCs to myelinate DRG axons. Although SCs infected with the different control shRNA formed myelin normally, knocking down Par-3 expression in SCs significantly reduced their ability to myelinate (Fig. 2, D and E) without affecting proliferation (fig. S3). In some cases, SCs appeared to synthesize myelin proteins as seen in Fig. 2D (open arrow) but could not adequately myelinate the axons. Similar results were obtained with three additional nonoverlapping shRNA constructs previously reported to knock down Par-3 expression (19) (fig. S4). Importantly, disruption of Par-3 expression did not affect axon/glial junction formation, as revealed by N-cadherin staining (fig. S5). These results indicate that disrupting the asymmetric localization of Par-3 in premyelinating SCs by either overexpression or knockdown inhibits myelination, suggesting that asymmetric localization of Par-3 may be necessary for myelination.

How does a polarity protein such as Par-3 regulate myelination? Previous studies have shown that BDNF binds p75NTR to inhibit SC migration and promote myelination (3, 5). Consistent with these results, p75NTR knockout mice display a delay in myelination, thinner myelin sheaths, and an overall reduction in myelinated fibers in adulthood (3). Because Par-3 is a wellknown scaffolding protein, it raises the possibility that Par-3 and p75NTR might form a complex at the site of the axon-glial junction and play a part in the initiation of myelination. Consistently, we found that Par-3 and p75NTR are colocalized at the axon-glial junction (Fig. 3A). In addition, Par-3 coimmunoprecipitates with an antibody to p75NTR, and vice versa, in newborn mouse sciatic nerve extracts, which suggests an interaction in vivo (Fig. 3B). To investigate the timing of the Par-3/p75NTR association, we turned to the SC/DRG coculture system, where we can precisely control the onset of myelination. As expected, Par-3 and p75NTR were detected in the cocultures both before and after the induction of myelination, but 10 days after induction, when the large majority of myelin had already formed, Par-3 and p75NTR protein levels were reduced (fig. S6), suggesting that they may be required only in the early stages of myelination. To test this hypothesis, we coimmunoprecipitated Par-3 with an antibody to p75NTR, using protein extracts isolated at different time points from the SC/DRG cocultures (Fig. 3C). We found that Par-3 and p75NTR did not interact in SC/DRG cocultures before the induction of myelination, whereas after induction, Par-3 and p75NTR transiently associated, peaking at 2 days after induction (Fig. 3C). This finding was further confirmed by quantifying the percentage of SCs that displayed either asymmetric Par-3 or both asymmetric Par-3 and p75NTR by immunostaining. About 10% of the total number of SCs displayed asymmetric Par-3, but only 5% of the SCs displayed a colocalization of Par-3 and p75NTR (Fig. 3D). This colocalization was transient and peaked at the initiation of myelination (Fig. 3D).

Fig. 3.

Par-3 and p75NTR associate transiently at the initiation of myelination. (A) Premyelinating SCs cocultured with DRG neurons displayed asymmetric localization of Par-3 and p75NTR. Colocalization was observed (yellow) when the corresponding images were merged. A z-stack of images was acquired with structured illumination fluorescence microscopy and reconstructed in three dimensions to confirm colocalization. A projection of the z-stack is shown. Scale bar, 10 μm. (B) Coimmunoprecipitation for Par-3 and p75NTR was performed using sciatic nerves (SN) of newborn mouse pups. Par-3 coimmunoprecipitated with an antibody to p75NTR and was detected by immunoblotting for Par-3, whereas p75NTR coimmunoprecipitated with an antibody to Par-3 and was detected by immunoblotting for p75NTR. Normal rabbit immunoglobulin Gs were used as controls for the immunoprecipitation. (C) Purified DRG neurons, SCs, and cocultures at different days in culture (d) and days after induction of myelination (dI) were used for the coimmunoprecipitation of Par-3 with an antibody to p75NTR. Par-3 was detected by immunoblotting for Par-3 (top). The blot was then stripped and reprobed with an antibody to p75NTR (bottom). (D) Quantification SCs displaying only asymmetric Par-3 or both asymmetric Par-3 and p75NTR in SC/DRG cocultures. The cultures were fixed and stained at 2, 4, or 6 days after induction of myelination. Error bars, SD. (E) Coimmunoprecipitation with the p75 antibody from cocultures at 2 days after induction of myelination in the presence of exogenous BDNF, or the BDNF scavenger TrkB-Fc. The blot was probed for Par-3 (top), stripped, and reprobed for p75NTR (bottom).

Because BDNF is known to promote SC myelination by acting through p75NTR, and the removal of BDNF inhibits this process, we sought to determine the specific nature of the Par-3/p75NTR interaction in the myelination program, which would be expected to increase in conditions favorable for myelination. Consistent with this possibility, the addition of exogenous BDNF to the culture medium enhanced the Par-3/p75NTR association beyond that of control cultures, and removing endogenous BDNF, using the BDNF scavenger TrkB-Fc, significantly inhibited this association (Fig. 3E). These results were recapitulated by examining the colocalization of Par-3 and p75NTR by immunostaining in the presence or absence of BDNF. The percentage of cells that displayed both asymmetric Par-3 and p75NTR doubled with the addition of exogenous BDNF. These results demonstrate a ligand-dependent association of Par-3 and p75NTR and provide further evidence for a specific interaction and function that is critical for myelination.

The p75NTR has a PDZ binding motif at the C terminus (20), and Par-3 contains three PDZ domains (11), raising the possibility that Par-3 and p75NTR might associate via PDZ domains. To test this hypothesis, we constructed retroviral vectorsencodingbothGFPandeachofthethree PDZ domains of Par-3, or a 20 amino acid portion of the intracellular domain (ICD) of p75NTR, which contains the PDZ binding motif. Infection of SCs cocultured with DRG neurons with each of the constructs revealed that overexpression of p75NTR-ICD and the PDZ-1 and PDZ-2 domains of Par-3 significantly inhibited myelination, but overexpression of the PDZ-3 domain of Par-3 had no effect (Fig. 4, A and B). To determine whether Par-3 directly interacts with p75NTR, we used an in vitro binding assay with purified FLAG-tagged PDZ domains and a glutathione S-transferase (GST)–p75NTR-ICD. Remarkably, we found that only the PDZ-1 domain of Par-3 binds the ICD of p75NTR, whereas the other PDZ domains did not associate with p75NTR-ICD (Fig. 4C).

Fig. 4.

Par-3 directly associates with p75NTR via its PDZ1 domain. (A) Retroviral vectors expressing GFP, and each of the three PDZ domains of Par-3, or the ICD of p75NTR, were generated and used to infect SCs cocultured with DRG neurons. Infected cells overexpressed the individual domains of Par-3 and p75NTR and GFP (green, arrowheads). Myelin was identified by immunostaining for MBP (red). Scale bar, 10 μm. (B) Quantification of the overexpression experiment was achieved by counting the proportion of infected SCs actively myelinating. Error bars, SD; *P < 0.001, unpaired Student's t test. (C) The FLAG-tagged version of the PDZ domains of Par-3 and the bacterially produced GST-p75-ICD proteins were purified, and binding was assayed in a cell-free extract. The complex was pulled down by immobilizing the FLAG-PDZ domains, and detection of binding was determined by immunoblotting with an antibody to GST. (D) Schematic representation of a proposed model for the role of Par-3 in SC myelination. An illustration of the cross section of a myelin internode at different time points throughout the myelination process.

Is Par-3 required for the asymmetric localization of p75NTR, or vice versa? To address this question, we counted the proportion of SCs that displayed asymmetric distribution of p75NTR after Par-3 knockdown, compared with control-infected SCs. The number of SCs that displayed asymmetric p75NTR localization after Par-3 knockdown (19.2 ± 3.9% with control shRNA versus 8.1 ± 0.8% with Par-3 shRNA) was reduced by half, indicating that the asymmetric localization of p75NTR requires Par-3. Although some SCs still display asymmetric p75NTR after Par-3 knockdown, it is likely that this is due to residual Par-3 expression after shRNA knockdown. To determine whether p75NTR could be required for the asymmetric localization of Par-3 at the axon-glia junction, we analyzed cocultures of SCs and DRG neurons prepared from p75NTR knockout mice (21). We found that p75NTR knockout SCs displayed the same asymmetric Par-3 localization as wild-type SCs (fig. S7), indicating that asymmetric localization of Par-3 does not require p75NTR.

The results presented here identify a novel protein-protein interaction via the PDZ binding motif of p75NTR, to our knowledge the first demonstration of Par-3 directly associating with a membrane receptor. Our data also show an essential requirement for the establishment of SC polarity in myelination, because disruption of Par-3 localization by overexpression and knockdown, which in turn disrupts the Par-3/p75NTR complex, inhibits SC myelination. Because overexpression of Par-6, another polarity protein known to form a complex with Par-3, can also inhibit myelination (fig. S8), we propose that the Par polarity complex has a major role in SC myelination. Based on our results, we suggest the following model (Fig. 4D). SCs become polarized upon contact with axons, as reflected by the asymmetric enrichment of Par-3 at the site of the axon-glial junction (Fig. 4D, left). As SCs prepare to initiate myelination, BDNF is synthesized by DRG neurons, secreted and released along axons to promote SC myelination. BDNF binds to p75NTR on the SCs, which somehow accommodates the association between the PDZ-1 domain of Par-3 and the PDZ binding motif of p75NTR (Fig. 4D, left). Concomitantly, Par-3 recruits p75NTR to the axon-glial junction, ensuring that BDNF is continuously and efficiently received by the SC and translated into signaling events that lead to the formation of myelin (Fig. 4D, left and middle). Once myelination is complete, Par-3 and p75NTR are down-regulated, and the complex dissociates (Fig. 4D, right).

Our data raise the possibility that the asymmetric localization of Par-3 at the axon-glial junction causes a convergence of various receptors and adhesion molecules that are essential for the initiation of myelin formation. This is supported by our observation that overexpression of the PDZ-2 domain, which does not bind p75NTR-ICD, also efficiently abolishes myelination by SCs. It will be interesting to characterize the molecules that are recruited to this site and to determine their role in the process of myelination. Additionally, it will be important to determine whether cell polarity mechanisms are involved in nerve regeneration and in remyelination paradigms, because this will largely impact the design of therapies for various peripheral neuropathies and nerve injury.

Supporting Online Material

Materials and Methods

Figs. S1 to S8


References and Notes

Stay Connected to Science

Navigate This Article