Disruption of the NF-H gene increases axonal microtubule content and velocity of neurofilament transport: relief of axonopathy resulting from the toxin beta,beta'-iminodipropionitrile

Abstract

To investigate the role of the neurofilament heavy (NF-H) subunit in neuronal function, we generated mice bearing a targeted disruption of the gene coding for the NF-H subunit. Surprisingly, the lack of NF-H subunits had little effect on axonal calibers and electron microscopy revealed no significant changes in the number and packing density of neurofilaments made up of only the neurofilament light (NF-L) and neurofilament medium (NF-M) subunits. However, our analysis of NF-H knockout mice revealed an approximately 2.4-fold increase of microtubule density in their large ventral root axons. This finding was further corroborated by a corresponding increase in the ratio of assembled tubulin to NF-L protein in insoluble cytoskeletal preparations from the sciatic nerve. Axonal transport studies carried out by the injection of [35S]methionine into spinal cord revealed an increased transport velocity of newly synthesized NF-L and NF-M proteins in motor axons of NF-H knockout mice. When treated with beta,beta'-iminodipropionitrile (IDPN), a neurotoxin that segregates microtubules and retards neurofilament transport, mice heterozygous or homozygous for the NF-H null mutation did not develop neurofilamentous swellings in motor neurons, unlike normal mouse littermates. These results indicate that the NF-H subunit is a key mediator of IDPN-induced axonopathy.

Figure 1

Targeted disruption of the NF-H gene. A shows the restriction map of the mouse genomic NF-H and targeting vector. The targeting vector was generated by inserting a 1.1-kb Neo cassette from pMC1NEOpolyA into the second XmaI site of a BamHI/BamHI NF-H fragment of ∼8 kb. The EcoRI/ BamHI fragment of the 5′ end was used as a probe for Southern blotting to detect a 2.8-kb EcoRI fragment indicative of homologous recombination. As shown in B, no changes in the levels of mRNA for NF-M and NF-L were detected in the brain of NF-H heterozygous (+/−) and homozygous (−/−) mutant mice. The levels of NF-H mRNA decreased by ∼50% in NF-H +/− mice as compared with +/+ littermate. As expected, no NF-H mRNA was detected in the NF-H −/− mice. In C, the Coomassie blue–stained gels of Triton X-100–insoluble extracts at 4°C from the spinal cord and sciatic nerve confirmed the lack of NF-H protein in NF-H −/− mice. The loss of NF-H did not provoke significant changes in the detected levels of NF-L (C) or of β-tubulin and actin as revealed by the immunoblots in D. However, there was a modest increase of ∼20% in levels of NF-M protein as detected by Coomassie blue staining (C). As shown in D, the SMI-31 antibody that recognizes primarily the hyperphosphorylated form of NF-H in normal mice yielded a dramatic increase of ∼10-fold in reactivity for the NF-M band in samples from NF-H null mice. This is likely the reflect of a compensatory increase in phosphorylation state of NF-M protein that contains, like NF-H, multiple Lys-Ser-Pro phosphorylation sites.

Figure 1

Targeted disruption of the NF-H gene. A shows the restriction map of the mouse genomic NF-H and targeting vector. The targeting vector was generated by inserting a 1.1-kb Neo cassette from pMC1NEOpolyA into the second XmaI site of a BamHI/BamHI NF-H fragment of ∼8 kb. The EcoRI/ BamHI fragment of the 5′ end was used as a probe for Southern blotting to detect a 2.8-kb EcoRI fragment indicative of homologous recombination. As shown in B, no changes in the levels of mRNA for NF-M and NF-L were detected in the brain of NF-H heterozygous (+/−) and homozygous (−/−) mutant mice. The levels of NF-H mRNA decreased by ∼50% in NF-H +/− mice as compared with +/+ littermate. As expected, no NF-H mRNA was detected in the NF-H −/− mice. In C, the Coomassie blue–stained gels of Triton X-100–insoluble extracts at 4°C from the spinal cord and sciatic nerve confirmed the lack of NF-H protein in NF-H −/− mice. The loss of NF-H did not provoke significant changes in the detected levels of NF-L (C) or of β-tubulin and actin as revealed by the immunoblots in D. However, there was a modest increase of ∼20% in levels of NF-M protein as detected by Coomassie blue staining (C). As shown in D, the SMI-31 antibody that recognizes primarily the hyperphosphorylated form of NF-H in normal mice yielded a dramatic increase of ∼10-fold in reactivity for the NF-M band in samples from NF-H null mice. This is likely the reflect of a compensatory increase in phosphorylation state of NF-M protein that contains, like NF-H, multiple Lys-Ser-Pro phosphorylation sites.

Figure 2

Light microscopy of peripheral myelinated axons. The lack of NF-H subunit in +/− and −/− mice had little effects on the caliber of axons from DRG neurons (A, C, E) or from L5 ventral roots (B, D, F). Bar: (A, C, E) 100 μm; (B, D, F) 20 μm.

Figure 3

Cytoskeletal changes in L5 ventral root axons lacking NF-H. Comparison of axonal calibers (A), neurofilament content (B), and nearest neighbor distances (C), and density of microtubules in L5 ventral root axons from NF-H +/+, NF-H +/−, and NF-H −/− mice. The n (A) represents the number of animals and results were obtained from three sets of littermates age of 71 and 100 d. The results in B and C were derived from 71-d-old littermates. The most remarkable cytoskeletal change in the absence of NF-H was the dramatic increase (∼2.4-fold) in the density of microtubules. The data in D were derived from four NF-H knockout mice obtained from three independent ES cell lines called 37C2, 41B8, and 42B6. The age of the mice is shown at the bottom of the figures. The data show the mean ± SD. The t test was applied. *P < 0.0001.

Figure 4

Electron microscopy of ventral root axons. Transverse sections (A, C, E) and longitudinal sections (B, D, F) of myelinated axons >5 mm in diam in the internode from the L5 ventral roots of normal (A and B) and NF-H +/− (C and D) and knockout mice (E and F). The lack of NF-H did not significantly affect the structure of 10-nm neurofilaments. Note, however, a substantial increase in the number of microtubules in both NF-H +/− and −/− mice. Open arrows, microtubules; filled arrows, 10-nm neurofilaments. Bar, 0.4 μm.

Figure 5

Increase of assembled tubulin in mice lacking NF-H. Taxol-stabilized cytoskeletal preparations from the sciatic nerve of normal (+/+) and NF-H knockout (−/−) mice were fractionated on SDS-PAGE. In A, the Coomassie blue–stained gels show increased levels of assembled tubulin in the sciatic nerve of NF-H −/− mice as compared with normal mice. B shows immunoblots of taxol-stabilized cytoskeletal preparations using antibodies against NF-L and acetylated tubulin. In C, the densitometric analysis of the immunoblots of samples from five mice revealed an increase of ∼2.5-fold in levels of acetylated tubulin in taxol-stabilized cytoskeletal preparations from the sciatic nerve of the NF-H −/− mice. Note a decreased level of assembled actin in the NF-H homozygous −/− mice as compared with normal mice. The positions of molecular weight markers are indicated at the left.

Figure 6

Increased immunodetection of tau protein in cytoskeletal fraction of sciatic nerve from the NF-H knockout mice. After SDS-PAGE, the protein samples described in Fig. 5 were blotted on membrane and treated with alkaline phosphatase. The membrane was then incubated with the TAU-1 antibody (Boehringer Mannheim Corp., Indianapolis, IN) that recognizes hypophosphorylated tau. The sample from the NF-H −/− yielded an increased signal for a band corresponding to low molecular weight tau.

Figure 7

Increased velocity of NF-L and NF-M transport in axons of NF-H–deficient mice. (A–C) Fluorographs of slow transport profiles in motor axons of the sciatic nerve from 3-mo-old normal, NF-H +/− and NF-H −/− mice at 30 d after intraspinal injection of [³⁵S]methionine. For each panel, each successive lane represents a 3-mm nerve segment extending distally to the right. (D–F) show transport of NF-L and NF-M proteins quantified by densitometry scanning of fluorograph profiles. In the NF-H −/− mice, the leading edge of radiolabeled NF-L and NF-M proteins was detected more distally than in normal mice. No substantial changes of axonal transport occurred in the NF-H +/− mice. The asterisk in C points to an ∼180-kD band detected in the slow transport component of the sciatic nerve from NF-H −/− mice. The identity of this band remains unclear. The genotyping of mice and immunoblotting of sciatic nerve samples with anti–NF-H antibodies certified that this 180-kD band is not a NF-H species.

Figure 8

Relief of motor axonopathy resulting from IDPN intoxication in mice lacking NF-H protein. Light micrographs show the anterior horn of the L5 spinal cord in control (A–C), NF-H heterozygous +/− (D–F) and homozygous −/− mice (G–I) at 0 d (A, D, G), 7 d (B, E, H), and 15 d (C, F, I) after IDPN treatment. Note the ventral root axons within the spinal cord are enlarged at 7 d and reach massive proportions at 15 d after IDPN intoxication only in normal mice (B and C) but not in the NF-H +/− (E and F) and −/− mice (H and I). Open arrows, the ventral root axons within the spinal cord; VT, ventral root axons. Bar, 100 μm.

Figure 9

Electron microscopy of transverse sections from the L5 ventral root axons within the DRG after 7 d IDPN treatment in normal littermate +/+ (A) and NF-H homozygous −/− mice (B). The segregation of microtubules from neurofilament caused by IDPN intoxication is only evident in the control (A) as indicated by large filled arrows but not in NF-H homozygous mice (B). Small open arrows, non-segregated microtubules. Bar, 0.4 μm.