Report

Control of Peripheral Nerve Myelination by the ß-Secretase BACE1

See allHide authors and affiliations

Science  27 Oct 2006:
Vol. 314, Issue 5799, pp. 664-666
DOI: 10.1126/science.1132341

Abstract

Although BACE1 (beta-site amyloid precursor protein–cleaving enzyme 1) is essential for the generation of amyloid-b peptide in Alzheimer's disease, its physiological function is unclear. We found that very high levels of BACE1 were expressed at time points when peripheral nerves become myelinated. Deficiency of BACE1 resulted in the accumulation of unprocessed neuregulin 1 (NRG1), an axonally expressed factor required for glial cell development and myelination. BACE1–/– mice displayed hypomyelination of peripheral nerves and aberrant axonal segregation of small-diameter afferent fibers, very similar to that seen in mice with mutations in type III NRG1 or Schwann cell–specific ErbB2 knockouts. Thus, BACE1 is required for myelination and correct bundling of axons by Schwann cells, probably through processing of type III NRG1.

The generation, aggregation, and deposition of amyloid-b peptide (Ab) in the brains of Alzheimer's disease (AD) patients is an invariant pathological feature of this devastating neurodegenerative disease (1). Ab is generated from the membrane-spanning b-amyloid precursor protein (APP) by endoproteolytic processing. Sequential cleavages, first by b-secretase and then by g-secretase, are required to liberate Ab into the extracellular space (2). β-Secretase activity is conferred by a type I transmembrane aspartyl protease, BACE1 (37). Other substrates for BACE1 include the sialyltransferase ST6Gal I (8, 9), the adhesion protein P-selectin glycoprotein ligand-1 (PSGL-1) (10), b subunits of voltage-gated sodium channels (11), APP-like proteins (APLPs) (12), and Ab itself (1315). BACE1 is the sole b-secretase because no Ab is synthesized in BACE1–/– mice (1619). However, only modest behavioral alterations are seen after targeted mutation of BACE1, and its physiological function remains unclear (19, 20).

Expression of BACE1 is confined mainly to neurons. Its activity could thus lead to release of substrates from neuronal membranes and thereby mediate paracrine signaling to neighboring cells, i.e., function in neuron-neuron or neuron-glia interactions. We investigated the expression of BACE1 in postnatal mice. The highest levels of BACE1 expression were observed in early postnatal stages, when myelination occurs (Fig. 1A). BACE1 expression declined in the second postnatal week, attaining low levels in adult animals (Fig. 1A). We hypothesized that the high expression of BACE1 in neurons around birth could be linked to the onset of myelination by Schwann cells, the ensheathing glia of peripheral nerves, which occurs at this time and which depends on signaling from the accompanying axons. In particular, the type III isoform of the epidermal growth factor (EGF)–like factor neuregulin 1 (NRG1), an axonal signal that activates heteromeric ErbB2 and ErbB3 receptors on Schwann cells, is important during Schwann cell development and myelination (21, 22). Type III NRG1 adopts a two-transmembrane structure with the active EGF domain in the lumenal portion, which may require endoproteolysis for its signaling capacity (23). Haploinsufficiency of type III NRG1 or conditional knockout of ErbB signaling in Schwann cells leads to aberrant axonal ensheathment and hypomyelination in peripheral nerves, whereas overexpression of type III NRG1 results in hypermyelination (2426). Owing to the tight temporal link between BACE1 expression and peripheral nerve myelination, we investigated whether BACE1 and type III NRG1 were coexpressed within sensory and motor neurons, whose axons project within peripheral nerves. In situ hybridization on spinal cord and dorsal root ganglia (DRG) from young postnatal mice revealed expression of BACE1 within the central nervous system (CNS), notably in ventral horn motor neurons, as well as in peripheral sensory neurons in the DRG (Fig. 1, B to D). No expression of BACE1 could be detected in satellite glia of DRGs or Schwann cells (Fig. 1, C and D, asterisks). The pattern of BACE1 expression in spinal cord and DRGs was highly similar to that of type III NRG1 (Fig. 1, E to G), particularly in sensory and motor neurons.

Fig. 1.

BACE1 is highly expressed during the period of myelin formation and is coexpressed with type III NRG1 in neurons projecting to the periphery. (A) Immunoblot analysis of membrane preparations from CNS lysates of mice 1, 5, 8, 12, and 17 days old or of tissue from adults. BACE1 is highly expressed in the first postnatal week and is subsequently down-regulated. Levels of the BACE1 substrate, amyloid precursor protein (APP), remain relatively constant in postnatal animals of different ages. (B to G) Analysis of BACE1 expression (B to D) or type III NRG1 (E to G) at postnatal day 5 with digoxigenin-labeled riboprobes (B, C, E, and F). Both BACE1 and type III NRG1 are coexpressed in motor neurons (B, E, and higher-magnification insets B′ and E′; motor neurons are indicated by arrowheads) and peripheral sensory neurons in the DRG (C and F). Counterstaining with a nuclear marker [D and G; false-color DAPI (4′,6-diamidino-2-phenylindole) counterstain of the sections shown in C and F] shows an absence of BACE1 and type III NRG1 expression in Schwann cells and satellite glia (asterisks highlight glia in the nerve bundle within the DRG). Bars: (B and E), 200 mm; (B′ and E′), 40 mm; (C, D, F, and G), 100 mm.

Because defects in the extent of myelination are seen in mice with reduced signaling of type III NRG-1 (25, 26), we studied the progression of myelination in the sciatic nerve of BACE1–/– mice using electron microscopy. Myelin sheath thickness, expressed as the G ratio [internal/external fiber diameter (27)], was determined in several hundred axons from pairs of mutants and controls at day 8, 12, and 17 of postnatal development and in adult animals. Axons of BACE1–/– mice were hypomyelinated at all stages (Fig. 2, A to F, and figs. S1 and S2; P < 0.001), whereas myelin ultrastructure was unchanged (Fig. 2, A and B, insets). Furthermore, individual large-diameter axons surrounded by single Schwann cells, which had failed to initiate myelination, were numerous in BACE1–/– mutants at postnatal day 8 (Fig. 2B, arrow). Hypomyelination was also observed in an independently generated BACE1 knockout (17) (fig. S3). We also examined the morphology of small-diameter axons, which are ensheathed and separated from each other by cytoplasmic processes of nonmyelinating Schwann cells to form Remak bundles. The bundling of such axons by nonmyelinating Schwann cells was significantly altered in BACE1–/– mice (Fig. 2, G to I) (P < 0.001), such that Remak bundles contained aberrantly large numbers of unseparated or poorly segregated axons.

Fig. 2.

BACE1–/– mice display hypomyelination of peripheral nerves and axonal-bundling abnormalities. Electron microscopy analysis of sciatic nerves of control (A, D, G) and BACE1 mutant mice (B, E, H) in 8-day-old (A and B) and adult mice (D to H). G-ratio determinations at postnatal day 8(C) and adult stages (F) are shown as the percentage of total myelinated axons. Insets (A and B) show normal ultrastructure of myelin lamellae in mutant mice. Amyelinated axons are abundant in BACE1–/– mice at postnatal day 8 (B, arrow). (G to I) Sorting of small-diameter axons by Schwann cell cytoplasmic processes within Remak bundles is defective in BACE1 mutants. Arrowheads in (G) and (H) indicate cytoplasm between different axons, which isolates each individual axon within the Remak bundles of controls. BACE1 mutants contain unusually large Remak bundles; most of the mutant axons are not separated by the Schwann cell and remain directly apposed to each other. (G and H) are magnifications of Remak bundles seen in (D) and (E), respectively; bars: (A, B, D, and E), 2 mm; (G and H), 1 mm.

β-Secretase activity in vitro can be demonstrated for both BACE1 and to a lesser extent for the homologous enzyme BACE2 (2831). We thus analyzed the peripheral nerves of BACE2 mutant mice, as well as BACE1/BACE2 compound mutants, using the independently generated BACE1 mutant for comparison (17, 32) (fig. S3). Peripheral nerve myelination in BACE2 mutant mice was unchanged, whereas BACE1/BACE2 compound homozygotes displayed hypomyelination, very similar to that seen in BACE1 homozygotes. Thus, regulation of nerve myelination by b-secretase activity is restricted to BACE1.

The changes observed in peripheral nerves of BACE1–/– mice phenocopied those previously observed in mice with haploinsufficiency of NRG1 or in mice that lack ErbB signaling in Schwann cells, i.e., amyelination of a subset of large-diameter axons at early stages, hypomyelination into adulthood, and altered sorting of axons within Remak bundles by nonmyelinating Schwann cells (2426). Because axonally expressed type III NRG1 determines the ensheathment fate of axons (25, 26), we reasoned that the myelination phenotype in BACE1–/– mice might be due to an alteration in the BACE1-dependent presentation or availability of neuronal type III NRG1 to associated glia. Indeed, biochemical analysis of CNS membrane preparations revealed a robust accumulation of a 130-kD isoform of NRG1 in BACE1–/– mice (Fig. 3A), corresponding to the uncleaved type III precursor (23). The expression of NRG1 mRNA, in contrast, was unchanged in the brains of BACE1–/– mice (fig. S4). This is consistent with the accumulation of uncleaved NRG1 protein being due to loss of BACE1-dependent processing. The findings with BACE1–/– mice suggested that axonally bound type III NRG1 could be a physiological substrate for BACE1 in vivo. NRG1 isoforms with an EGF-b domain (NRG1-β) have a higher affinity for ErbB homo- and heterodimeric receptors and are thereby more potent in signaling than those with an EGF-a domain (NRG1-α) (33). The a-type NRG1 isoforms are dispensable, whereas the b-type isoforms are required for normal nervous system development (34). Consistent with this, the predominant NRG1 isoforms expressed in the nervous system were transmembrane isoforms with a b-type EGF domain (NRG1-β1) (35), which was confirmed for postnatal mouse brain by reverse transcription–polymerase chain reaction (fig. S4). To provide further evidence for BACE1-mediated processing of NRG1, we studied the release of secreted alkaline phosphatase (SEAP) proteins fused to NRG1-β1 sequences (23, 36). Coexpression of SEAP–NRG1-β1 with BACE1 resulted in a strong increase (P < 0.001) in shedding of SEAP–NRG1-β1, which was inhibited in a dose-dependent manner by the BACE1-specific inhibitor C3 (37) (Fig. 3B). This result supports the hypothesis that BACE1 mediates the cleavage of NRG1-β1.

Fig. 3.

Type III NRG1 is a physiological substrate for BACE1 in vivo. (A) Membrane preparations from brain lysates of 5-day-old wild-type (WT) and BACE1–/– mice were analyzed by immunoblotting with antibodies directed against the N-terminal domain of BACE1, a C-terminal domain of NRG1, or amyloid precursor protein (APP), as indicated. The full-length NRG1 precursor (130 kD) accumulates in BACE1–/– mice. (B) BACE1-dependent cleavage of SEAP–NRG1-β1 (36) was evaluated by SEAP assays with supernatants of control human embryonic kidney 293 (HEK293) cells (SEAP–NRG1-β1) or HEK293 cells coexpressing BACE1 (SEAP–NRG1-β1 + BACE1). A fourfold increase in shedding of the SEAP–NRG1-β1 fusion protein was measured upon BACE1 coexpression (P < 0.001). Shedding in coexpressing cells could be suppressed by applying the BACE1-specific inhibitor C3 (37). (*) Denotes significance compared to all other groups (P < 0.001; n = 6 independent measurements per group).

Our findings define a physiological function of BACE1. BACE1 is required for peripheral nerve myelination and axonal bundling by Schwann cells probably via processing of type III NRG1. However, other proteases and additional substrates can so far not be excluded. The tumor necrosis factor–a (TNF-a) converting enzyme (TACE), for example, is known to be the responsible shedding protease for some NRG1 splice variants (26, 36, 38, 39) and other EGF receptor ligands (40). Previous evidence suggested that specifically cell surface–associated type III NRG1 is involved in neuron-glial signaling and peripheral nerve myelination (26, 41). Our in vivo data suggest that neurons specifically trigger myelination and axonal sorting within Remak bundles by the proteolytic activity of BACE1. The data obtained with SEAP–NRG-b1 fusion proteins indicate that one recognition site for BACE1 resides in the stalk region of NRG1-β1 isoforms (present in types I and III). We cannot yet exclude the possibility that BACE1 activity results further in the complete release of type III NRG1 from the membrane via cleavage on the N-terminal side of the EGF domain. Whether BACE1 also plays important roles during myelination of the CNS remains to be shown, although accumulation of unprocessed NRG1 in the brain of BACE1–/– mice (Fig. 3A) may be indicative of a CNS function for BACE1.

The inhibition of β- and γ-secretase is currently one of the most hopeful approaches for AD therapy besides amyloid b-peptide vaccination (2, 4244). However, likely side effects due to the inhibition of physiological function of secretases must be carefully considered. Indeed, blocking g-secretase function resulted in rather severe side effects due to the inhibition of Notch signaling (2). Our findings define a physiological function for BACE1 in myelination, which may allow monitoring of the effects for b-secretase inhibition in vivo.

Supporting Online Material

www.sciencemag.org/cgi/content/full/1132341/DC1

Materials and Methods

Figs. S1 to S4

References

References and Notes

View Abstract

Navigate This Article