Defective axonal transport of neurofilament proteins in neurons overexpressing peripherin

Authors

  • Stéphanie Millecamps,

    1. Research Centre of Centre Hospitalier Universitaire de Québec, Department of Anatomy and Physiology of Laval University, Quebec, Canada
    Search for more papers by this author
  • Janice Robertson,

    1. Centre for Research in Neurodegenerative Diseases, Toronto, Ontario, Canada
    Search for more papers by this author
  • Roxanne Lariviere,

    1. Research Centre of Centre Hospitalier Universitaire de Québec, Department of Anatomy and Physiology of Laval University, Quebec, Canada
    Search for more papers by this author
  • Jacques Mallet,

    1. Centre National de la Recherche Scientifique, Laboratoire de Genetique Moleculaire de la Neurotransmission et des Processus Neurodegeneratifs, Hôpital de la Pitié-Salpêtrière, Paris, France
    Search for more papers by this author
  • Jean-Pierre Julien

    1. Research Centre of Centre Hospitalier Universitaire de Québec, Department of Anatomy and Physiology of Laval University, Quebec, Canada
    Search for more papers by this author

Address correspondence and reprint requests to Jean-Pierre Julien, Research Centre of CHUQ, Department of Anatomy and Physiology, Laval University, 2705 Boulevard Laurier, Quebec, QC, Canada G1V 4G2. E-mail: jean-pierre.julien@crchul.ulaval.ca

Abstract

Peripherin is a type III neuronal intermediate filament detected in motor neuron inclusions of amyotrophic lateral sclerosis (ALS) patients. We previously reported that overexpression of peripherin provokes late-onset motor neuron dysfunction in transgenic mice. Here, we show that peripherin overexpression slows down axonal transport of neurofilament (NF) proteins, and that the transport defect precedes by several months the appearance of axonal spheroids in adult mice. Defective NF transport by peripherin up-regulation was further confirmed with dorsal root ganglia (DRG) neurons cultured from peripherin transgenic embryos. Immunofluorescence microscopy and western blotting revealed that excess peripherin provokes reduction in levels of hyperphosphorylated NF-H species in DRG neurites. Similarly the transport of a green fluorescent protein (GFP)-tagged NF-M, delivered by means of a lentiviral construct, was impaired in DRG neurites overexpressing peripherin. These results demonstrate that peripherin overexpression can cause defective transport of type IV NF proteins, a phenomenon that may account for the progressive formation of ALS-like spheroids in axons.

Abbreviations used
ALS

amyotrophic lateral sclerosis

CMV

cytomegalovirus

DIV

days in vitro

DMEM

Dulbecco's modified Eagle's medium

DRG

dorsal root ganglia

GFP

green fluorescent protein

IF

intermediate filament

NF

neurofilament

NGF

nerve growth factor

PBS

phosphate-buffered saline

SDS

sodium dodecyl sulfate

SEM

standard error of the mean

SC

slow component

WT

wild type

Intermediate filaments (IFs) comprise a large family of proteins that maintain the cytoarchitecture of cells. Neuronal IFs associate to produce a rope-like structure with a diameter of a size (10 nm) intermediate between actin microfilaments (7 nm) and microtubules (24 nm) in axons (Hirokawa et al. 1984). Neurofilament (NF) triplet proteins, termed NF-L (68 kDa), NF-M (150 kDa) and NF-H (200 kDa) for light, medium and high molecular weight NFs, respectively, as well as α-internexin (66 kDa) and peripherin (57 kDa), are the five members of the neuronal IFs (Julien and Mushynski 1998). These members are differentially expressed in adult brain. NFs are expressed in most differentiated neurons with large caliber axons. α-Internexin expression is restricted to specific populations of neurons with smaller axons, such as cerebellar granule cells and interneurons (Pachter and Liem 1985; Chien and Liem 1994; Fliegner et al. 1994). Peripherin expression is limited to the peripheral nervous system (PNS) neurons, to central nervous system (CNS) neurons with peripheral projections (sensory and motor neurons originating from brainstem or spinal cord) and to subsets of interneurons in the cortex and hippocampus (Portier et al. 1983; Leonard et al. 1988; Parysek and Goldman 1988; Brody et al. 1989; Escurat et al. 1990).

NFs and peripherin have opposite temporal gene expression in response to injury. Peripherin expression is increased in sensory neurons and in spinal motor neurons after axotomy of the sciatic nerve (Oblinger et al. 1989b; Troy et al. 1990; Wong and Oblinger 1990; Terao et al. 2000) whereas expression of NF subunits is reduced (Portier et al. 1982; Greenberg and Lasek 1988; Oblinger et al. 1989a; Muma et al. 1990). We have also reported that peripherin expression is induced after brain injuries in subsets of neurons where it is usually silent (Beaulieu et al. 2002). These data suggest that peripherin may play a role in neuronal regeneration.

NFs are made by the heteropolymerization of NF-L with NF-M or NF-H (Lee et al. 1993). In contrast to NFs, peripherin can self-assemble into homopolymers (Cui et al. 1995; Ho et al. 1995; Beaulieu et al. 1999b) or interact with each of the three NF subunits in vitro (Ching and Liem 1993; Athlan and Mushynski 1997; Beaulieu et al. 1999b) or in vivo (Parysek et al. 1991). Using SW13 cells with no endogenous cytoplasmic IF proteins, we showed that in the absence of NF-L, peripherin is unable to assemble properly with heavy NF subunits, and such interactions form punctuate protein structures (Beaulieu et al. 1999b).

Cytoskeletal proteins are dynamic components undergoing a continuous turnover. The newly synthesized subunits are actively transported from the cell body to the neuronal processes and nerve endings, in a so-called anterograde manner, along the microtubule track. The slow axonal transport conveys cytoskeletal proteins and is divided into two rate components. The slow component a (SCa) carries NFs and microtubules at a rate of 0.1–1 mm/day. α-Internexin and peripherin are co-transported with NF triplets in axons (Filliatreau et al. 1988; Kaplan et al. 1990; Chadan et al. 1994). The slow component b (SCb) is slightly faster (2–8 mm/day) and carries actin, spectrin and cytoplasmic proteins (Hoffman and Lasek 1975; Nixon and Logvinenko 1986).

Recent studies based on the live visualization of fluorescently-labeled NFs, expressed in cultured sympathetic neurons, provided new insight into the molecular basis of NF movement (Roy et al. 2000; Wang et al. 2000). The transport of the fluorescent NF polymers was mainly anterograde. It consisted of intermittent rapid movements (few µm/s), propelled by a powered moving motor protein, interspersed with long pauses that could be the consequence of dissociation of NFs with the motor protein (Roy et al. 2000; Wang et al. 2000). The SCa slow transport wave observed in vivo by traditional pulsed radiolabeled techniques probably represents the sum of individual NF movements spending most of their time pausing (Brown 2000).

Defects in NF metabolism leading to the disorganization of the NF network is a proposed mechanism for neurodegenerative disorders such as motor neuron diseases, including amyotrophic lateral sclerosis (ALS) (Carpenter 1968; Hirano et al. 1984; Rouleau et al. 1996; Wong et al. 2000) and infantile spinal muscular atrophy (SMA) (Lippa and Smith 1988; Lee et al. 1989; Cifuentes-Diaz et al. 2002), as well as sensorimotor neuropathies, including the axonal form of Charcot-Marie-Tooth disease (CMT2) (Brownlees et al. 2002; Perez-Olle et al. 2002) and giant axonal neuropathy (GAN) (Asbury et al. 1972; Bomont et al. 2000). Concomitant NF and peripherin accumulations in perikaryal and axonal aggregates are a hallmark of degenerating spinal motor neurons in ALS, a devastating motor neuron disease characterized by a progressive loss of upper and lower motor neurons resulting in paralysis and death (Carpenter 1968; Hirano et al. 1984; Corbo and Hays 1992; Migheli et al. 1993; Rouleau et al. 1996; Wong et al. 2000; He and Hays 2004). Involvement of peripherin in ALS pathogenesis is further supported by the recent discovery of a peripherin gene frameshift mutation in one sporadic ALS case (Gros-Louis et al. 2004). We also reported increased peripherin expression in ALS tissue (Robertson et al. 2003). Moreover, peripherin is toxic for motor neurons when overexpressed in primary culture (Robertson et al. 2001), or in transgenic mice (Beaulieu et al. 1999a). Indeed, transgenic mice overexpressing the mouse peripherin gene (Per mice) develop a late onset motor neuron disease reminiscent of some aspects of human ALS (Beaulieu et al. 1999a). Unlike transgenic mice overexpressing wild-type NFs (Cote et al. 1993; Xu et al. 1993; Lee et al. 1994; Beaulieu et al. 1999a), these Per mice exhibited massive motor neuron death that was accelerated and worsened by NF-L deficiency (Beaulieu et al. 1999a, 2000). Such a decrease in expression of NF-L has been reported in motor neurons of ALS patients (Bergeron et al. 1994; Wong et al. 2000). The noxious effect of peripherin overexpression in Per mice may be due to the accumulation of the protein in the proximal axons of motor neurons that might block axonal transport (Beaulieu et al. 1999a). In order to clarify the mechanism(s) leading to motor neuron degeneration in these mice, we carried out axonal transport studies using the [35S]-methionine pulse chase approach. We also studied the distribution of phosphorylated NF-H in dorsal toot ganglia (DRG) cultures prepared from peripherin transgenic mice. Finally, we used a lentiviral vector encoding green fluorescent protein (GFP)-tagged NF-M to study the distribution of newly transported NFs in a Per DRG culture. Our results demonstrate that intracellular transport of NF proteins is defective in peripherin transgenic mice, and that these defects may account for the progressive accumulation of IF aggregates in motor and sensory axons.

Materials and methods

Transgenic animals

Heterozygous transgenic mice overexpressing the mouse peripherin gene under the control of peripherin promoter have been described previously (Beaulieu et al. 1999a). Control mice are the wild-type (WT) age-matched littermates. The use of animals and surgical procedures described in this study meet The Guide to the Care and Use of Experimental Animals of the Canadian Council on Animal Care.

Injection of [35S]-methionine into spinal cord

The procedure for injecting [35S]-methionine into spinal cord is detailed elsewhere (Millecamps and Julien 2004). Briefly, l-[35S]-methionine (1175 Ci/mmol; Perkin Elmer, Life Sciences, Boston, MA, USA) was dried by lyophilization and resuspended in a phosphate-buffered saline (PBS) solution to obtain a concentration of 250 µCi/µL. Mice (n = 6 per group) were deeply anaesthetized by intraperitoneal injection of 75 mg/kg xylazine (CDMV, St-Hyacinthe, QC, Canada) and 10 mg/kg ketamine (CDMV) diluted in PBS. Back muscles were carefully incised, using surgical scissors and microdissecting tweezers, to expose the laminae of vertebra. Under an operating microscope, a 1–2 mm2 window was drilled in the right lamina, at the level of L5 spinal cord, with a small electric drill (Dremel MultiPro, Rotary Tool 395T6 model, Racine, WI, USA) to expose the right spinal cord without damage. [35S]-methionine solution (1 µL) was injected into the anterior horn area, 1 mm deep from the dorsal surface and 1 mm from the middle groove, at a rate of 0.5 µL/min delivered by a syringe pump (KD Scientific, New Hope, PA, USA). After injection the muscles were closed in layers with silk sutures and the skin incision was closed by wound suture clips. One hour after waking, the mice were administered a subcutaneous injection of 0.1 mg/kg buprenorphine (Reckitt & Colman Pharmaceuticals, Inc., Richmond, VA, USA).

Four weeks after injection, the mice were killed by an overdose (1 g/kg) of chloral hydrate (Fisher Scientific, Nepean, ON, Canada). The right sciatic nerve and L5 ventral and dorsal roots were dissected out. The L5 roots and DRG were pooled into one fraction corresponding to an axonal length of 12 mm. The sciatic nerve was cut into 3 mm consecutive segments starting from the L5 dorsal root ganglion to the muscular extremity.

Protein extraction and fluorographs

Each 3 mm segment was homogeneized, using a Polytron PT1200C (Kinematica AG, Littau-Lucerne Switzerland), in 1% Triton X-100 extraction buffer consisting of 10 mm Tris-HCl pH 7.5, 150 mm NaCl, 1 mm EDTA and 10 µg/mL of each protease inhibitor (aprotinin, leupeptin, pepstatin). The mixture was centrifuged at 13 000 gfor 10 min. The pellet was then dissociated in 0.5% sodium dodecyl sulfate (SDS) buffer containing 8 m urea and 2%β-mercaptoethanol with protease inhibitors. Extracts were centrifuged at 13 000 gfor 10 min and supernatant fluids (Triton X-100-insoluble fraction) were collected. Protein concentrations were evaluated by the standard Bradford assay (Bio-Rad, Mississauga, ON, Canada).

Equal amounts (20 µg) of proteins prepared from each nerve segment were loaded into the consecutive wells of 8% SDS-polyacrylamide gels. Electrophoresis was carried out until the bromophenol blue dye had run to the bottom of the gel. The polyacrylamide gel was stained with Coomassie brilliant blue solution (1% Coomassie blue, 40% methanol, 10% glacial acetic acid) for 30 min and de-stained for 1 h in a fixative solution (30% methanol, 10% glacial acetic acid). For fluorographic enhancement of signal, the gel was soaked in Amplify reagent (Amersham Biosciences, Baie d'Urfe, QC, Canada) for 30 min and dried under vacuum to be exposed to Kodak Biomax film (Kodak, Rochester, NY, USA) at −80°C for 5 weeks. Radiolabeled bands were quantified by densitometry scanning.

Cell lines and primary cultures

Human 293T kidney cells and N2a mouse neuroblastoma cells were grown in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% fetal bovine serum. N2a were induced to differentiate by serum withdrawal and addition of 1 mm dibutyryl cAMP (Sigma, Oakville, ON, Canada) during 72 h.

DRG cultures were prepared from E13.5 peripherin transgenic embryos or WT littermates as previously described (Robertson et al. 2001). DRG were dissected, dissociated with trypsin and plated onto glass coverslips (12 mm round), doubly-coated with poly d-lysine and with extracellular matrix (ECM) gel (Sigma), at a density of 20 000 cells/well. DRG were maintained in Neurobasal medium supplemented with B-27 supplement, glutamine (0.5 mm), nerve growth factor (NGF; 15 ng/mL) and penicillin/streptomycin (all from Invitrogen, Burlington, ON, Canada). Cultures that were infected with lentivirus construct were treated with 5-fluorouracil anti-mitotic agent on days 4 and 5 to prevent growth of non-neuronal cells.

Transduction of DRG neurons using a lentiviral vector encoding GFP-tagged NF-M

The enhanced GFP (eGFP)-NF-M plasmid expressing rat NF-M cDNA fused to eGFP under the control of the cytomegalovirus (CMV) promoter has been previously described (Yabe et al. 1999). To construct the corresponding lentiviral plasmid, an AseI-NotI blunt fragment, including the CMV promoter and eGFP-NF-M transgene, was ligated into the NdeI-Acc65I blunt sites of the pTrip plasmid containing the lentiviral genome, including the central Flap sequence (Zennou et al. 2001). Viral stock was produced using co-transfection of 293T cells with the lentiviral plasmid, an encapsidation plasmid (p8.7) and a vesicular stomatitis virus (VSV) envelope plasmid. The amount of p24 capsid protein in the virus stock was measured by an antigen immunoadsorption kit (Beckman Coulter, Roissy, France). Transducing units were determined by incubating 293T with serial dilutions of the viral stock, followed by a quantification of fluorescent cells using fluorescence-activated cell sorting (FACS) analysis. The final viral concentration was 7 × 108 transducing units (TU)/ng of p24 protein. Fourteen-day-old DRG were infected overnight with virus suspension (15 TU/cell). Cells were harvested for protein extraction 36 h later, or fixed at 36 h and 72 h post-infection (4% paraformaldehyde prepared in PBS for 10 min followed by cold methanol for 5 min). Double-immunofluorescence staining was performed with monoclonal anti-GFP (green) and polyclonal anti-peripherin (red fluorescence) antibodies.

Antibodies

Antibodies recognizing actin (monoclonal MAB1501), peripherin (monoclonal MAB1527, polyclonal AB1530), NF-M (NN18 clone) and NF-H (RT97 clone) were purchased from Chemicon International Inc. (Temecula, CA, USA). Anti-NF-H (N52 clone) and anti-NF-L (NR4 clone) were from Sigma. The mouse anti-green fluorescent protein (GFP) was obtained from Molecular Probes (Eugene, OR, USA). The rabbit anti-GFP (ab290) was purchased from Abcam (Paris, France).The peroxidase-conjugated anti-mouse and anti-rabbit (made in goat) were from Jackson Immunoresearch Laboratories (Mississauga, ON, Canada). The biotinylated anti-rabbit IgG (made in goat) was from Vector Laboratories Inc. (Burlington, ON, Canada) and the biotinylated Alexafluor goat anti-rabbit and anti-mouse IgG were from Molecular Probes.

Immunoblotting

Mice were killed by carbon dioxide, then spinal cords were dissected out, frozen in liquid nitrogen and kept at −80°C until protein extraction. DRG were harvested by scraping in PBS, centrifuged at 1000 gfor 10 min and the pellets kept at −80°C until protein extraction. Spinal cord and DRG were lysed in a buffer containing 62.5 mm Tris-HCl pH 6.8, 2% SDS, 10% glycerol and protease inhibitors (aprotinin, leupeptin, pepstatin). Lysed extracts were centrifuged at 13 000 gfor 10 min and protein concentration of supernatant fluids was estimated by the bichinconinic acid assay (Sigma). Proteins (15 µg) were separated on 8% SDS-polyacrylamide gels and transferred electrophoretically to polyvinylene difluoride (PVDF) membranes. Membranes were incubated with primary antibodies in PBS, containing 5% milk and 0.1% Tween 20, overnight at 4°C, followed by 1 h of incubation with the corresponding peroxidase-conjugated secondary antibodies. Signals were detected using Western Lightning Chemiluminescence Reagent (Perkin Elmer, Wodbridge, ON, Canada) and Biomax radiographic film (Kodak, Rochester, NY, USA).

Immunohistological analysis

For DAB (3,3′-diaminobenzidine) staining, mice were anesthetized with xylazine/ketamine and perfused with 4% paraformaldehyde in PBS. The DRG and spinal cord segments were dissected out at the L5 level, post-fixed for 2 h with the fixative, cryoprotected for 2 days in PBS containing 30% sucrose and frozen at −80°C. Sections (20 µm) were cut using a cryostat. Slides were incubated for 30 min in PBS containing 0.3% hydrogen peroxide. After three consecutive washes in PBS, sections were incubated in a blocking solution (10% goat serum, 0.6% gelatin, 0.3% Triton X-100 in PBS) for 1 h. All antibodies were diluted in the blocking solution. Incubation with the polyclonal peripherin antibody (1 : 1000) was perfomed overnight. The anti-peripherin antibody was detected using biotinylated anti-rabbit IgG and the avidin-biotin peroxidase complex (Vector Laboratories Inc.) with DAB (Sigma) as the chromogenic substrate.

For immunofluorescence staining, fixed cells were incubated for 30 min with a blocking solution of PBS containing 3% bovine serum albumin (BSA) and 0.1% Triton X-100. All primary antibodies were diluted (at 1 : 1000) in the blocking solution and incubation was carried out overnight at 20°C. After four consecutive washes in PBS, coverslips were incubated with anti-rabbit AlexaFluor 488 (green) and anti-mouse AlexaFluor 594 (red) antibodies for 1 h at 20°C (diluted in the blocking solution at a final concentration of 1 : 250). Coverslips were mounted in Mowiol 4–88 (Calbiochem, VWR International, Mississauga, ON, Canada) and examined by epifluorescence microscopy.

Cells counts

For motor neuron cell counts, the 20 µm sections of spinal cord were stained with a thionin solution (0.25%). The large motor neurons in the ventral horn were counted each fourth section in a blind manner (total of 30 ventral horn sections per animal). The mean number of motor neurons per ventral horn section was calculated for WT and Per animals aged 6, 12 and 24 months (three animals per group). The means were compared using a one-way analysis of variance followed by a Tukey's post-hoc test.

To analyse the morphological accumulations of peripherin in ventral horn of Per mice, each fourth spinal cord section was immunostained with anti-peripherin antibody. The number of perikaryal peripherin deposits in motor neurons and the number of spheroids were recorded in three Per mice at each time point.

To quantify the phenotypes of the eGFP-NF-M-expressing neurons, images of all transduced DRG neurons were captured for three WT and three Per (derived from three different embryos) with a fixed time exposure of 300 ms for red and 1 s for green fluorescence. These images (total of 200–500 neurons/well) were analyzed for the presence of green neurites. Neurons were scored as possessing eGFP-NF-M in their neurites if they had one or more green neurites at least as long as the soma diameter.

Results

Transgenic mice overexpressing peripherin (Per mice) develop axonal spheroids late in life

Per transgenic mice overexpress the mouse peripherin gene under the control of the 2.5 kb of 3′ untranslated sequences of its own promoter (Beaulieu et al. 1999a). We monitored peripherin protein level in spinal cord extracts of these mice at different ages. As shown in Fig. 1(a), peripherin expression level in Per mice is fivefold higher than endogenous mouse peripherin in WT animals. Similar levels of peripherin protein were detected in spinal cord extracts of 6-, 12- and 24-month-old Per mice (Fig. 1a). These data demonstrate that total peripherin does not accumulate over time in the spinal cord of Per animals.

Figure 1.

 Age-dependent formation of inclusions in Per transgenic mice. Proteins were extracted from spinal cord of 6-, 12- and 24-month-old Per mice and their age-matched control littermates. (a) Western blot analysis was perfomed using monoclonal anti-peripherin and anti-actin antibodies. Immunohistochemical detection of peripherin protein on the ventral horn (b–g) and dorsal horn (h–m) spinal cord sections, and on DRG (n–s) of WT (b–d, h–j, n–p) and Per mice (e–g, k–m, q–s) aged of 6 (b, e, h, k, n, q), 12 (c, f, i, l, o, r) and 24 (d, g, j, m, p, s) months using polyclonal peripherin antibody. Punctated perikaryal inclusions were detected in motor neurons of 6-month-old animals (e, white arrow). Punctated aggregates (white arrows) and large inclusions (black arrow) were observed in motor neuron cell bodies of 12-month-old Per mice (f). Note that some inclusions were seen in proximal segments of axons (f, white arrowhead). Numerous spheroids within neurites (black arrowheads) were detected in 24-month-old Per animals (g). Some spheroids were also detected in the dorsal root spinal nerve of 24-month-old animals (m). Few DRG neurons presented peripherin aggregates in 6- and 12-month-old animals (q and r, arrows). Numerous amorphous aggregates that filled the DRG cytoplasm were detected in 24-month-old Per mice (s, arrows). Bars: 10 µm (b–m), 40 µm (n–s). (t) The number of large motor neurons per ventral horn in L5 spinal cord sections of WT and Per animals was determined. Means were compared by one-way anova followed by Tukey's multiple comparison test. A 35% loss of motor neuron cell body was found in 24-month-old Per animals compared with other groups (**p < 0.01). (u) Percentage of motor neurons with intracellular perikaryal or spheroid peripherin deposits in ventral horn of Per mice at each time point. The proportion of motor neurons containing morphological accumulations increased with age in Per mice.

The total number of motor neurons in WT and Per mice was compared at 6, 12 and 24 months-of-age. The results are shown in (Fig. 1t). The number of large motor neurons per ventral horn in L5 spinal cord section is similar in WT and Per animals at 6 and 12 months-of-age. In contrast, a 35% loss of motor neuron cell body was found in 24-month-old Per animals. These results are consistent with the 35% loss of motor axons we previously reported in L5 ventral roots of 28-month-old Per transgenic animals (Beaulieu et al. 1999a).

The distribution of peripherin protein in the spinal cord of the Per mice was examined immunohistochemically using peripherin polyclonal antibody at different time points. A diffuse peripherin staining was observed in the perikarya of most of the motor neurons in 6-month-old Per mice. Peripherin punctated inclusions were occasionally observed in the perikaryon of some motor neurons (Fig. 1e, white arrow). No peripherin-positive inclusions were detected in the axonal motor neuron compartment at this age. Several types of peripherin aggregates were detected at the age of 12 months (Fig. 1f), including punctated peripherin inclusions (white arrows) and large inclusions localized on one side of the motor neuron perikarya (black arrow) or obstructing the proximal axons of some motor neurons (white arrow head). In contrast, spheroids were prominent in 18 month- (data not shown) and 24-month-old Per mice in ventral horn neurites (Fig. 1g, black arrow heads). Such inclusions were not observed in age-matched control animals. None of these aggregates was recognized by anti-ubiquitin antibody (data not shown). Quantitative analysis of the morphological accumulations of peripherin was assessed in the ventral horn of 6-, 12- and 24-month-old Per mice. The number of intracellular peripherin deposits in motor neurons and of spheroids in ventral horn of Per mice was recorded at each time point. Results indicated that the proportion of motor neurons containing morphological accumulations increased with age in Per mice (Fig. 1u). Only 5% of motor neurons presented peripherin aggregates in 6-month-old Per mice. In contrast, in 2-year-old transgenic animals, 30% of the remaining motor neurons presented peripherin accumulations. A high number of spheroids occurred in these mice. Thus, the motor neuron loss in Per mice was concomitant with the accumulation of spheroids within the ventral horn of these animals.

We also determined whether peripherin accumulation occurred in DRG neurons of Per mice. Few peripherin aggregates were detectable in the cellular compartment of DRG neurons in 6-month-old animals (Fig. 1q, arrow). Unlike the light diffuse labeling observed in WT DRG, numerous intensely-labelled amorphous aggregates were detected in old Per animals (Fig. 1s). Small spheroids were visible in the dorsal root spinal nerve of 2-year-old animals (Fig. 1m, black arrow heads). These results show that the formation of spheroids, which occurred late in the life of Per mice, was not restricted to motor neurons and may take place in other cell types that overexpress peripherin protein.

Defective axonal transport of NFs proteins in 6 month-old Per mice

We measured axonal transport velocity in motor neurons before appearance of the axonal inclusions in 6-month-old animals (Fig. 2). This was done by monitoring radiolabeled proteins in the sciatic nerve 28 days after injection of [35S]-labeled methionine into the L5 anterior horns of Per mice and age-matched control littermates. Newly synthesized radiolabeled proteins were detected by autoradiography while they travelled in the 3 mm consecutive segments of the sciatic nerve. There was a significant reduction of radiolabeled NF protein in the distal segments of the sciatic nerve in Per mice. NFs were not detectable after the second segment of the sciatic nerve (18 mm from the spinal cord), although they were readily detectable until the fourth segment in WT mice (24 mm). Notably, radiolabeled proteins were concentrated in the 15 mm segment of Per mice compared with the 18–24 mm distribution in WT animals. Densitometry analysis showed that the NF leading peak corresponds to a transport rate of approximately 0.85 mm/day for WT animals. In contrast, axonal transport of NF proteins was impaired in Per mice, with a leading peak corresponding to a rate of about 0.64 mm/day. Transport of tubulin and actin was also altered in Per mice, with reduced levels of radiolabeled proteins in distal nerve segments. Thus, overexpression of peripherin gene in motor neurons impairs the axonal transport of cytoskeleton components in vivo.

Figure 2.

(a) Axonal transport of NF proteins is delayed in Per mice. Slow axonal transport in motor axons of the sciatic nerves of 6-month-old WT and Per mice 4 weeks after intraspinal injection of [35S]-methionine. (b, c) Fluorographs of slow axonal transport profiles. Protein extracts from L5 roots were pooled in one fraction (lane 1) corresponding to a 12 mm length from the spinal cord. Protein extracts from consecutive 3 mm segments of the sciatic nerve were loaded in lanes 2–8. Each lane represents the pool of the sciatic nerves of three animals. NF-H, NF-M, NF-L, tubulin and actin were recognized by their apparent molecular masses of 200, 150, 68, 55 and 45 kDa, respectively. NF-H and NF-M are indicated by arrows. (b) In WT mice, the NF radioactive leading edge (indicated by an arrowhead) occurred at 24 mm. (c) In Per mice, this leading edge occurred at 18 mm. The same results were obtained from two independent experiments. (d, e) Transport of NF-H, NF-M, tubulin and actin quantified by densitometry scanning of fluorograph profiles. Each point is the mean ± SEM of protein values obtained from the six WT (d) and the six Per mice (e) used in this study. The axonal transport of cytoskeletal components is significantly retarded in Per mice.

DRG neurons with elevated peripherin exhibit abnormal distribution of phosphorylated NF-H

To demonstrate further that transport impairment occurs early in the life of Per mice, we compared NF-H expression in DRG neurons isolated from Per and WT mouse embryos. Similar to the increase measured in the adult spinal cord extracts described above, the peripherin level was fivefold higher in extracts from Per DRG cultures compared with WT DRG cultures (Fig. 3a).

Figure 3.

 Decreased levels of hyperphosphorylated NF-H in cultured DRG neurons from Per transgenic embryos. DRG cultures were prepared from WT and Per mouse embryos and maintained for 7–21 days in vitro (DIV). (a) Western blot analysis of protein extracts from DRG at (a) 14 and (b) 21 DIV probed with monoclonal anti-peripherin, anti-actin, anti-NF-L (NR4 clone), anti-NF-H (N52 clone) recognizing both hyper and hypophosphorylated NF-H (referred to as pNF-H and dpNF-H, respectively) and anti-NF-H (RT97 clone) that binds specifically to pNF-H. Duration of the culture is indicated above the corresponding lanes. Equal amounts of total protein were loaded in each lane. Note that hyperphosphorylated NF-H became detectable when Per DRG were maintained for 21 DIV (b, black arrow). Double-immunofluorescence detection of peripherin (c′–n′, green) and NF-H N52 clone (c–h, red), α-tubulin (I–j, red) or NF-M (k–n, red) in WT (c, e, g, i, k, m) and Per (d, f, h, j, l, n) DRG neurons at 7 (c–d, i–j), 14 (e–f, k–l) and 21 (g–h, m–n) DIV. Arrows point to peripherin aggregates. Note the presence of punctate-like (f′, l′), spherical (h′, n′) and amorphous (d′, j′) aggregates in Per DRG neurons. Arrowheads indicate the neurites. Note the absence of NF-H in neurites from Per DRG neurons at 7 (d) and 14 (f) DIV. NF-H is detectable in 21 DIV Per neurites (h, arrowhead). Part of NF-M was retained within the perikaryon at 14 (l) and 21 (n) DIV, as shown by the intense staining in the perikarya. These data are consistent with a delay in NF transport into the Per neurites. Bars: 40 µ m (c–h) and 20 µm (i–n).

In these neurons, expression of non-phosphorylated NF-H is detectable by day 3, whereas the phosphorylated form of NF-H appears after establishment of the neurites and is restricted to this cellular compartment (Foster et al. 1987). We probed 14-day-old DRG extracts by immunoblotting with N52 antibody, which binds independently of the phosphorylation state of NF-H. Two migrating bands corresponding to hypo and hyperphosphorylated NF-H, respectively, were detected in WT DRG cultures (Fig. 3a, lanes 2 and 4). In contrast, only the more rapidly migrating form was detectable in Per mice (Fig. 3a, lanes 1, 3 and 5). The lack of pNF-H in Per cultures maintained for 14 days in vitro (DIV) was confirmed using RT97 antibody that specifically recognizes the phosphorylated form of NF-H. Note that there is relatively more dephosphorylated NF-H in peripherin-overexpressing samples compared with WT, and that we did not find any significant differences in NF-L expression between WT and Per transgenic cultures (Fig. 3a).

These results demonstrated that there is no detectable hyperphosphorylated NF-H in Per DRG neurons at 2 weeks. This was confirmed by double-immunofluorescence stainings performed at 7 and 14 days of culture. Striking differences were observed in the distribution of NF-H in WT and Per DRG neurons. NF-H immunoreactivity was detectable in the cell body and neuritic processes of WT DRG (Figs 3c and e), which is consistent with the transport of NF-H into the neurites. In contrast, NF-H labeling was restricted to the perikaryon of 7- and 14-day-old Per DRG (Figs 3d and f). An α-tubulin staining revealed that no difference in neurite development was observed in the Per DRG compared with the WT cultures (Figs 3i and j). These results showed that NF-H transport was impaired in the Per DRG neurons that had a fully established network.

We then examined NF-H phosphorylation states in DRG cultures harvested at different time points after plating. Interestingly, phosphorylated NF-H became detectable in Per DRG neurons at 21 DIV (Fig. 3b) concomitantly with the appearance of NF-H immunoreactivity in the neurites (Fig. 3h). A lower NF-M staining was noticeable in Per transgenic neurites (Figs 3l and n) than that detected in WT (Figs 3k and m) at 14 and 21 DIV. Moreover, the strong NF-M staining detected within the cytoplasm indicates that a fraction of NF-M protein was retained within the perikaryon and was not properly transported within the neurites. Altogether, these results demonstrate that there is a delayed transport of NF protein into Per DRG neurites.

Transport of exogenous GFP-tagged NF-M is impaired in Per DRG culture

Finally, we constructed a lentiviral vector expressing eGFP-fused NF-M under the control of the CMV promoter. Lentiviral vectors efficiently transduce a wide variety of cells, including neurons (Naldini et al. 1996). Fluorescence-tagged NF-M has been extensively used in vitro to study several axonal transport concerns, including the form in which NFs are transported (Yabe et al. 1999, 2001a, 2001b), the axonal transport rate (Roy et al. 2000; Wang et al. 2000) and the effect of excitotoxicity on NF axonal transport (Ackerley et al. 2000). Our lentiviral construct was first tested on a N2a neuroblastoma mouse cell line. These cells develop neurites with axonal characteristics after cAMP analog treatment (Shea et al. 1985). Lentivirus-delivered fluorescent NF-M led to the formation of a green typical NF network in these cells that co-localized with endogenous NF-L (data not shown). Then, this lentiviral construct was applied to 14-day-old WT and Per DRG neurons. DRG were fixed 36 h after infection and were processed for double-immunofluorescence staining with anti-GFP and anti-peripherin antibodies. Most of the WT DRG neurons (94%, Fig. 4d) presented long GFP-positive green processes, indicating that exogenous tagged NF-M had been transported within their neurites (Figs 4a and e, white arrows). In contrast, only 23% of the Per DRG exhibited GFP-containing neurites (Fig. 4d). In most of these Per neurons the green fluorescence was restricted to the cell body (Figs 4b and f), despite the presence of bipolar neurites visualized by phase contrast microscopy (Fig. 4b, black arrows). This difference cannot be attributed to a defect in protein synthesis in Per DRG neurons because the level of eGFP-fused NF-M protein was similar in WT and Per-transduced DRG, as shown by western blot analysis (Fig. 4c). Moreover, the total number of GFP-positive neurons was the same in WT and Per cultures (Fig. 4d).

Figure 4.

 Transport of exogenous eGFP tagged NF-M is impaired in Per DRG culture. Double-immunofluorescence detection of peripherin (red) and GFP (green) in (a) WT and (b) Per 14 DIV DRG neurons 36 h after infection using a lentivirus vector expressing eGFP-fused NF-M. Note that the tagged NF-M was fully transported into the WT neurites (a, white arrows) whereas it remained concentrated in the soma of Per neurons (b), despite the presence of clear bipolar neurites on the phase contrast image (b, black arrows). (c) Western blot analysis of DRG protein extracts 36 h after infection with the lentivirus. Proteins from a non-infected control were loaded in lane 1. The membrane was probed with polyclonal anti-GFP, monoclonal anti-peripherin and anti-actin antibodies. The level of eGFP fused-NF-M protein was similar in WT- and Per-transduced DRG. (d) The number of neurons presenting or not GFP-positive neurites was recorded in three different wells (from three different embryos) of WT and Per cultures 36 and 72 h post-infection with the lentivirus. Results are means ± SEM. Means were compared using a Student's t-test (***p < 0.001). The proportion of neurons presenting GFP-positive neurites was calculated. Double-immunofluorescence detection of peripherin (red) and GFP (green) in WT (e, g) and Per (f, h) DRG neurons 36 h (e, f) and 72 h (g, h) after infection with the lentivirus vector. Images are presented at low magnification to show the preponderance of the phenotypes. The eGFP-NF-M protein was detected in the neurites of WT DRG 36 h post-infection (e, white arrows). In contrast, the majority of infected neurons in Per cultures had no green neurites at this time point. Most of the Per DRG neurites are GFP-positive at 72 h post-infection (h, white arrows). Bars: 30 µm (a, b), 60 µm (e–h).

Interestingly, the proportion of Per neurons with green neurites was increased 72 h after the infection. The GFP was readily detectable in most of them (71%) at this time point (Figs 4d and h). As the number of GFP-expressing neurons remained the same at 72 h post-infection in WT and Per cultures, this finding does not result from the specific death of congested neurons in Per cultures. Rather, these data indicate that the transport of newly synthesized NF-M was delayed in Per DRG neurites.

Discussion

The data presented here show that overproduction of mouse peripherin protein (type III IF protein) resulted in a marked alteration of axonal transport of NF proteins (type IV IF protein) in motor neurons of adult Per transgenic mice. Our analysis of slow axonal transport in the Per mice demonstrated that newly synthesized NFs were translocated from neuronal perikarya into axons, but that the slow axonal transport was retarded compared with that of WT mice.

In an attempt to substantiate the defect in NF transport further, we analyzed NF-H expression in DRG neurons isolated from Per mice. Results from these DRG culture experiments were consistent with the in vivo findings. The amount of hyperphosphorylated NF-H was lower in 14-day-old DRG neurons from Per embryos than in those prepared from WT embryos. Immunofluorescence experiments showed a lack of hyperphosphorylated NF-H in neurites of Per DRG. This impairment was not due to a defect in neurite development. Rather, the delayed appearance of hyperphosphorylated NF-H in the Per DRG cultures was probably the consequence of a delayed transport of NF subunits. It may also be that overexpression of peripherin induces inappropriate activation of kinases or phosphatases, which could have led to the delayed appearance of hyperphosphorylated NF-H we observed in DRG cultures. Finally, we used a lentivirus construct to compare the distribution of newly synthesized eGFP-tagged NF-M in WT and Per DRG neurons. The transport of fluorescent NF-M was delayed in Per DRG neurons compared with the WT DRG neurons. The combined results demonstrate a general delay in the transport of NF subunits in neurons with an excess of peripherin. One plausible explanation for this phenomenon is that peripherin may compete with NF-L for assembly with NF-M and NF-H subunits, resulting in oligomeric IF structures that are less suitable for the intracellular transport machinery.

The NF subunit transport defects observed in the adult Per mice may account, in part, for the age-dependent progressive accumulation of peripherin inclusions in the perikarya and axons of these animals. Several types of aggregates have been observed in DRG neurons with elevated peripherin in culture (Robertson et al. 2001). These aggregates were restricted to the perikarya and were never observed in the neurites (Robertson et al. 2001). Similarly, we observed several types of perikaryal inclusions containing peripherin in spinal motor neurons of 12-month-old Per animals (Fig. 1f). Axonal spheroids were detected later at the age of 24 months in motor and sensitive track. Their formation could therefore be the result of the sustained impairment in slow axonal transport that we observed in young animals. Overall, our results suggest that accumulation of inclusions in Per mice is not primarily attributable to the physical obstruction of axonal transport. However, the mechanism by which the IF aggregates occur in these animals is not clear. Previous investigations have shown that peripherin is mainly transported by SCa in the sciatic nerve (Filliatreau et al. 1988; Chadan et al. 1994), and that the amount of transported peripherin is increased in response to axotomy (Chadan et al. 1994). Peripherin has a molecular weight of 57 kDa, making it difficult to distinguish from tubulin in one-dimensional polyacrylamide gel electrophoresis. Nevertheless, it is conceivable that the motor responsible for transporting neuronal IFs in axons is sequestered by excessive amounts of peripherin in Per mice.

IF-containing inclusions are observed in other mouse models of motor neuron disease. Overexpression of the human NF-H gene in mice provokes several axonal defects, including a decline in NF transport, decrease in conduction velocity and atrophy of axons, which are not accompanied by motor neuron loss (Cote et al. 1993; Collard et al. 1995; Kriz et al. 2000). Inclusions observed in these NF-H mice are large and restricted to the perikaryon. In contrast, inclusions in the peripherin overexpressor are of small size and are present in cell bodies and axons. We observed here that axonal spheroids accumulated late in life in Per mice. This accumulation is correlated with the beginning of missing motor neuron axons that we previously reported in these animals (Beaulieu et al. 1999a). This would support the view that axonal spheroids may be more deleterious than perikaryal IF inclusions for neuronal function.

NF and peripherin accumulations in perikaryal and axonal inclusions are also observed in transgenic mice overexpressing mutant forms of the Cu/Zn superoxide dismutase (SOD1) gene (Gurney et al. 1994; Tu et al. 1996), the causative genetic defect found in about 20% of the familial cases of ALS (Deng et al. 1993; Rosen 1993). These transgenic mice develop a disease similar to the human one, leading to severe paralysis and premature death (Gurney et al. 1994; Ripps et al. 1995; Wong et al. 1995; Bruijn et al. 1997). Defects in axonal transport of NFs is observed long before the onset of the disease in these mice, suggesting that this perturbation could participate in the neurodegenerative process (Zhang et al. 1997; Williamson and Cleveland 1999). This impairment of slow axonal transport originates in proximal axons (Sasaki et al. 2004) and persists until the end of the disease (Zhang et al. 1997). Interestingly, we recently reported the presence of an abnormal and toxic splicing variant of peripherin in motor axons of mutant SOD1 transgenic mice (Robertson et al. 2003). Nonetheless, the absence of peripherin gene expression in mutant SOD1 mice did not affect disease progression, suggesting that peripherin is not a contributing factor in disease caused by SOD1 mutations (Lariviere et al. 2003).

The data presented here demonstrate that sustained up-regulation of peripherin expression can provoke defective transport of NF proteins, a phenomenon that may account for the progressive formation of ALS-like spheroids in axons. The axonal transport dysfunction in Per mice reported in this study occurs months before any axonal loss and at least one year before any motor neuron degeneration. This underlines the remarkable capability of motor neurons to overcome such axonal transport impairment. However, a sustained overexpression of peripherin appears detrimental during aging. This is perhaps related to the development of abnormal IF inclusions in the axons that might interfere with trafficking of vital organelles, ultimately compromising the survival of motor neurons.

Acknowledgements

We are very grateful to Dr Thomas Shea (University of Massachusetts, Lowell, USA) for his generous gift of eGFP-NFM plasmid and to Dr Pierre Charneau (Pasteur Institute, Paris, France) for providing the pTrip plasmid. We thank Pascale Hince for technical assistance. This project was supported by the National Institute of Health (NIH) and the Canadian Institutes of Health Research (CIHR). J-PJ is a Canada Research Chair in Neurodegeneration. SM was supported by the Fondation pour la Recherche Médicale and was awarded the Fondation Liliane Bettencourt-Schueller prize. JR is funded by The Motor Neurone Disease Association, the CIHR and The American ALS Association. RL was a recipient of a FRSQ-FCAR-Health studentship.

Ancillary