Bace1 processing of NRG1 type III produces a myelin-inducing signal but is not essential for the stimulation of myelination

Abstract

Myelin sheath thickness is precisely adjusted to axon caliber, and in the peripheral nervous system, neuregulin 1 (NRG1) type III is a key regulator of this process. It has been proposed that the protease BACE1 activates NRG1 dependent myelination. Here, we characterize the predicted product of BACE1-mediated NRG1 type III processing in transgenic mice. Neuronal overexpression of a NRG1 type III-variant, designed to mimic prior cleavage in the juxtamembrane stalk region, induces hypermyelination in vivo and is sufficient to restore myelination of NRG1 type III-deficient neurons. This observation implies that the NRG1 cytoplasmic domain is dispensable and that processed NRG1 type III is sufficient for all steps of myelination. Surprisingly, transgenic neuronal overexpression of full-length NRG1 type III promotes hypermyelination also in BACE1 null mutant mice. Moreover, NRG1 processing is impaired but not abolished in BACE1 null mutants. Thus, BACE1 is not essential for the activation of NRG1 type III to promote myelination. Taken together, these findings suggest that multiple neuronal proteases collectively regulate NRG1 processing.

Fig. 1

Proteases serve as positive and negative regulators of myelination. (A) Pharmacological BACE1 inhibition blocks myelination in DRG dissociated explant cultures, whereas inhibition of ADAMs promotes myelination. Fluorescent imunostaining for MBP (green). Numbers of MBP-positive segments (mean ± SEM) after 14 days in vitro (DMSO, ADAMs inh n = 3 cultures; BACE1 inh n = 2 cultures; 2 coverslips/treatment; *P < 0.05, **P < 0.01). Scale bar, 200 μm. (B) Axonal association of Schwann cells is not impaired upon BACE1 or ADAMs inhibition in DRG neuron dissociated explant cultures. Immunostaining for axons (Tuj1; red) and counterstaining for nuclei (DAPI; blue) after 9 d of protease inhibitor treatment. Representative axon bundles used for quantification are delineated. Numbers of axon-associated Schwann cells (per bundle length in μm) are displayed as percentage ± SEM relative to DMSO treated controls (n = 2 cultures; 3 coverslips/condition; 3 bundles/coverslip; P = 0.38, **P < 0.01). Scale bar, 20 μm. (C) Cleavage of full-length (HA-NRG1^FL), but not “processed” NRG1 type III (HA-NRG1^GIEF) by BACE1. Western blot of transfected HEK293T cell lysates probed with an anti-HA antibody. Expression of BACE1 (+B) strongly reduced the level of full-length HA-NRG1^FL (black arrowheads) and resulted in the accumulation of a ∼75 kD N-terminal processing product (white arrowhead), which corresponds in size to HA-NRG1^GIEF. HA-NRG1^FL processing by BACE1 was blocked by BACE1 inhibitor treatment (+B/Bi). Expression of BACE1 had no visible impact on HA-NRG1^GIEF levels. Note that “base line” processing of HA-NRG1^FL (+Bi) was not affected by BACE1 treatment. (D) BACE1 cleaves HA-NRG1^FL in the juxtamembrane “stalk” region. Western blot of transfected HEK293T cell lysates probed with an antibody against the C-terminal peptide sequence “GIEF”. Note that the C-terminal in ‘GIEF’ epitope only accumulates in HA-NRG1^FL lysates when BACE1 was coexpressed (+B). HA-NRG1^GIEF lysates served as a positive control. The membrane was reprobed for tubulin as a loading control. [Color figure can be viewed in the online issue, which is available at wileyonlinelibrary.com.]

Fig. 2

BACE1 cleavage in the “stalk” region produces a myelination-inducing NRG1 type III variant. (A) Domain structure of HA-NRG1^FL and HA-NRG1^GIEF. BACE1 cleavage of HA-NRG1^FL is indicated by scissors. Protein domains: EGF-like domain (green), cystein-rich domain (purple/black), transmembrane domain (brown), HA epitope tags (red). EC: extracellular; IC: intracellular. (B) HA-NRG1^FL and HA-NRG1^GIEF induce heterodimerization of ErbB2 and ErbB3 receptors in a two-cell split-TEV assay as measured by a ∼5 fold increase in luciferase reporter activity compared with controls. Results are displayed as relative luminescence units (RLUs) ±SD (n = 6; ***P < 0.001). (C) HA-NRG1^FL and HA-NRG1^GIEF stimulate myelination in DRG neuron-Schwann cell co-cultures. Immunostaining for MBP (red) demonstrates a dramatic increase in the number of myelin segments upon lentiviral expression of HA-NRG1^FL or HA-NRG1^GIEF in sensory neurons compared with noninfected control cultures (ctrl) 7 days after myelination induction. Scale bar, 50 μm. (D) Sensory neurons prepared from NRG1 type III null mutants (CRD-NRG1−/−) are not myelinated 14 days after induction of myelination. Myelination is restored in NRG1 type III-deficient axons after lentiviral overexpression of HA-NRG1^FL or HA-NRG1^GIEF as demonstrated by immunostaining for MBP (red). Neurofilament (NF) staining (green) reveals the normal maintenance of sensory axons in mutant cultures. Scale bar, 50 μm. [Color figure can be viewed in the online issue, which is available at wileyonlinelibrary.com.]

Fig. 3

Neuronal overexpression of epitope-tagged NRG1 type III-variants in transgenic mice. (A) Schematic drawing of the Thy1.2 cassette used for neuronal expression of HA-NRG1^FL and HA-NRG1^GIEF cDNAs (tg). XhoI, restriction site for subcloning; NotI/PvuI, restriction sites for cassette release; exons Ia, Ib, II, IV, and locations of RT-PCR primers used in (B) are indicated. (B) HA-NRG1^FL and HA-NRG1^GIEF transgenes are expressed in spinal cord at the onset of peripheral myelination. Transgene-specific RT-PCR of spinal cord cDNAs from transgenic and wildtype (wt) mice at P1. Amplification of β-actin was used as a positive control. (C) HA-NRG1^FL and HA-NRG1^GIEF proteins are expressed in the spinal cord of transgenic mice. Western blot of spinal cord protein lysates (40 μg) at 2 months of age probed with an anti-HA antibody. Arrowheads mark full-length (∼ 140 kD) and processed HA-NRG1 type III (∼ 70 kD) as well as a lower molecular weight protein band (∼ 45 kD), which most likely results from further NRG1 type III processing. (D) Spinal cord motor neurons and sensory neurons express HA-NRG1^FL and HA-NRG1^GIEF proteins at their plasma membrane. Confocal images of spinal cord and DRG cross sections after immunostaining for the HA-epitope, neurofilament 200 (NF), and peripherin (Prph). White line marks the border between gray and white matter. Scale bars, 50 μm. [Color figure can be viewed in the online issue, which is available at wileyonlinelibrary.com.]

Fig. 4

Expression of the epitope-tagged NRG1 type III-variants in the brains of transgenic mice. (A) HA-NRG1^GIEF expression in the cortex exclusively localizes to neurons. Confocal images of cortical layer V immunostained for the HA-epitope and cell type-specific markers. HA-NRG1^GIEF accumulates in a subset of NeuN-positive cortical neurons but is absent from parvalbumin (Parv)-expressing interneurons and Olig2-positive oligodendrocytes. Scale bars, 25 μm. (B) Confocal image of a HA epitope-positive neocortical projection neuron from a HA-NRG1^GIEF transgenic mouse at 4 months of age. HA-NRG1^GIEF is expressed in the soma, axon (arrowheads), and proximal dendrites (left panel), as well as dendritic spines (right panel; arrowheads). Scale bars, 10 μm. (C) Electron micrograph of a cortical synapse (postsynapse is to the lower right) after immunogold labeling using anti HA- (black arrowhead, 10 nm gold) and anti-PSD95 antibodies (white arrowhead, 15 nm gold). The HA-NRG1^FL transgene preferentially localizes to the PSD95-positive postsynaptic compartment. Scale bar, 100 nm. [Color figure can be viewed in the online issue, which is available at wileyonlinelibrary.com.]

Fig. 5

Limited axonal transport of NRG1 type III. (A) (left panel) Endogenous full-length NRG1 (140 kD) and a C-terminal processing product (∼ 60 kD; arrowheads) are detected with an antibody against the C-terminus of NRG1 (αNRG1 C-term) in wt spinal cord protein lysates at P12, whereas corresponding sciatic nerve expression is virtually absent. (right panel) Endogenous NRG1 accumulates on the plasma membrane of spinal cord motor neurons after immunostaining with the αNRG1 C-term antibody. Nuclei are counterstained with DAPI (blue). Corresponding NRG1 immunoreactivity is absent from the sciatic nerve. Sciatic nerve axons are marked by tubulin (Tuj1; red). (B,C) (left panels) When probed with an anti HA-antibody (αHA), HA-NRG1 type III (140 kD) and two N-terminal processing products (∼ 70 kD and ∼ 45 kD; arrowheads) are present in spinal cord protein lysates from HA-NRG1^FL transgenic mice at P12, but barely detectable in sciatic nerve. Similarly, HA-NRG1^GIEF (∼ 70 kD) and a smaller protein species (∼ 45 kD; arrowheads) are prominently expressed in spinal cord lysates from HA-NRG1^GIEF mice, whereas sciatic nerve expression is low. (right panels) Immunostaining for axons (neurofilament 200, NF; red) and the HA-epitope (green) reveals localization of HA-NRG1^FL and HA-NRG1^GIEF at the surface of individual sciatic nerve axons (insets). Scale bars, 10 μm. (D) Confocal images of an individual sciatic nerve axon in cross section from an adult HA-NRG1^GIEF transgenic mouse. Immunostaining for neurofilament 200 (NF; red) and the HA tag (green) reveals expression of HA-NRG1^GIEF in vesicle-like structures (white arrowheads) and on the axonal surface (empty arrowheads). Scale bar, 2.5μm. [Color figure can be viewed in the online issue, which is available at wileyonlinelibrary.com.]

Fig. 6

HA-NRG1^GIEF promotes hypermyelination in the PNS and CNS of transgenic mice. (A) Sciatic nerve hypermyelination in HA-NRG1^FL and HA-NRG1^GIEF transgenic mice. Methylenblue-Azure II staining for myelin on semithin sciatic nerve cross sections at 2 months of age. Scale bar, 10 μm. (B) Quantification of myelin sheath thickness by g-ratio analysis. HA-NRG1^FL and HA-NRG1^GIEF transgenic mice exhibit significantly thicker myelin compared with wt (*P < 0.05). Bars represent mean g-ratios ± SEM (n = 3; >100 axons/mouse). (C) The number of myelinated and nonmyelinated Schwann cells is not changed in HA-NRG1^GIEF transgenic mice compared with wt at 2 months of age (n = 3; mSchwann cells P = 0.08; nmSchwann cells P = 0.18). Error bars, ±SEM. (D) The total number of myelinated axons in sciatic nerves from HA-NRG1^GIEF transgenic mice at 2 months of age is unaltered (n = 3 each; P = 0.29). Error bars, ±SEM. (E) Electron micrographs of the corpus callosum at 2 months of age. Scatter plot (upper panel) displays g-ratios as a function of axon diameter. The average g-ratio (lower panel) in HA-NRG1^GIEF transgenic mice is significantly reduced when compared with wt (wt, n = 3; HA-NRG1^GIEFn = 4; *P < 0.05). Error bars, ±SEM. Scale bar, 0.25 μm. [Color figure can be viewed in the online issue, which is available at wileyonlinelibrary.com.]

Fig. 7

The myelination-promoting competence of NRG1 type III is not compromised in the absence of BACE1. (A) Semithin sciatic nerve cross sections stained for myelin at 2 months of age. Sciatic nerves from BACE1−/− mutants are severely hypomyelinated and HA-NRG1^FL transgenic nerves are hypermyelinated compared with wt. Expression of the HA-NRG1^FL transgene in a BACE1 null background restores myelination. Scale bar, 10 μm. (B) Quantification of myelin sheath thickness by g-ratio analysis. HA-NRG1^FL overexpression in BACE1−/− mutants significantly increases myelin sheath thickness compared with BACE1−/− mutants (**P < 0.01) and even wt (**P < 0.01). Myelin sheath thickness in HA-NRG1^FL*BACE1−/− mutants is not significantly reduced compared with HA-NRG1^FL transgenic mice (P = 0.35). (n = 3 per genotype, >100 axons/mouse). The scatter plot (lower panel) illustrates that HA-NRG1^FL expression in BACE1−/− mice rescues the myelination deficit of BACE1 mutants across all axon diameters. (C) Myelinating Schwann cells (left panel) are increased and nonmyelinating Schwann cells (right panel) decreased in adult BACE1−/− mutants (n = 4) compared with wt (n = 3) at 2 months of age (mean±SEM; *P < 0.05; **P < 0.01) (D) “Stalk” cleavage of NRG1 type III is reduced but not abolished in the absence of BACE1. Western blot of spinal cord protein lysates from 2 months old mice probed with an anti-HA antibody. Processed NRG1 type III (∼70 kD) is reduced, whereas full-length NRG1 type III (∼140 kD) accumulates in HA-NRG1^FL*BACE1−/− compound mutants compared with HA-NRG1^FL transgenic mice. Membrane was reprobed for tubulin as a loading control. Densitometric quantification demonstrates a significant increase in the relative signal ratio (∼140 kD/∼70 kD protein bands relative to tubulin) in HA-NRG1^FL*BACE1−/− mice ± SEM; *P < 0.05). [Color figure can be viewed in the online issue, which is available at wileyonlinelibrary.com.]