Induction of experimental autoimmune neuritis with peripheral myelin ...

Brain (1998), 121, 1895–1902

Induction of experimental autoimmune neuritis with peripheral myelin protein-22 C. M. Gabriel,1,2 R. A. C. Hughes,1 S. E. Moore,2 K. J. Smith1 and F. S. Walsh2 Departments of 1Neurology and 2Experimental Pathology, UMDS, Guy’s Hospital, London, UK

Correspondence to: Dr C. M. Gabriel, Department of Neurology, UMDS, Guy’s Campus, St Thomas’ Street, London Bridge, London SE1 9RT, UK

Summary Two myelin proteins, P2 basic protein and P0 glycoprotein, can induce experimental autoimmune neuritis (EAN), a model of human inflammatory neuropathy. We investigated whether peripheral nerve myelin protein-22 (PMP22), the gene for which is duplicated in hereditary motor sensory neuropathy type 1a, can also induce EAN. PMP22 cDNA produced by the reverse transcriptase– polymerase chain reaction from rat sciatic nerve was expressed in Escherichia coli as a fusion protein with glutathione-S-transferase (GST). Ten Lewis rats were immunized with purified PMP22 fusion protein (50– 100 µg) and eight controls with the same amount of GST. Two additional animals were immunized with each of two peptides (250 µg) of the human PMP22 extracellular sequences. Animals were examined daily until 20 days following immunization, when they underwent neuro-

physiological examination. A serum sample was then taken, prior to perfusion with glutaraldehyde and removal of the sciatic nerves and cauda equina. PMP22-immunized animals developed antibodies to the fusion protein and five out of 10 developed limp tails. No changes were observed in controls immunized with GST or in animals immunized with peptide. The mean compound motor action potentials elicited in the foot muscles by stimulation of the sciatic nerve at the sciatic notch and of the tibial nerve at the ankle were significantly reduced in the PMP22-immunized group (P , 0.05). Spinal roots from the group of animals immunized with PMP22 showed sparse infiltration of mononuclear cells, oedema and demyelination. PMP22 now deserves consideration as an autoantigen in human acute inflammatory demyelinating polyradiculoneuropathy.

Keywords: experimental autoimmune neuritis; PMP22; myelin protein; demyelination; polyneuropathy Abbreviations: CIDP 5 chronic inflammatory demyelinating polyradiculoneuropathy; cDNA 5 complementary DNA; CMAP 5 compound motor action potential; EAN 5 experimental autoimmune neuritis; ECD 5 extracellular domain; GST 5 glutathione-S-transferase; HMSN 1a 5 hereditary motor and sensory neuropathy type 1a; PMP22 5 peripheral myelin protein-22; RT–PCR 5 reverse transcriptase–polymerase chain reaction

Introduction The heterogeneous clinical features of human inflammatory neuropathy may be due to autoimmune responses to different antigens in different patients. Newly discovered myelin proteins deserve investigation as potential autoantigens. Peripheral myelin protein-22 (PMP22) was relatively recently characterized as a 22 kDa transmembrane protein expressed in Schwann cells and compact myelin, and constitutes 2–10% of total peripheral nerve myelin protein (Snipes et al., 1992; Koehler et al., 1996). The gene for PMP22 is responsible for hereditary motor and sensory neuropathy type 1a (HMSN 1a), which is most commonly associated with a tandem duplication of chromosome 17p11.2–12, within which the PMP22 gene is located (Matsunami et al., 1992; Patel et al., 1992; Timmerman et al., 1992; Valentijn et al., 1992b), or due to point mutations within the PMP22 gene itself © Oxford University Press 1998

(Valentijn et al., 1992a; Roa et al., 1993a, b, c). Transgenic rats overexpressing the gene for PMP22 (Sereda et al., 1996) exhibit a hypomyelinating peripheral neuropathy. The function of this protein in myelin remains poorly understood, but it may be involved in the control of the Schwann cell growth cycle (Zoidl et al., 1995), adhesion between Schwann cells and axons (Snipes et al., 1992), intercellular signalling (Snipes et al., 1993) or the maintenance of the structural integrity of myelin. The possible role of PMP22 as an autoantigen has not been considered. Experimental autoimmune neuritis (EAN) is an accurate experimental model of acute inflammatory demyelinating polyradiculoneuropathy, the most common form of Guillain– Barre´ syndrome, in which the target of the autoimmune reaction is believed to be a component of myelin. EAN is

1896

C. M. Gabriel et al.

most reproducibly generated by actively immunizing Lewis rats with bovine spinal root myelin. Immunization with the myelin proteins P2 (Kadlubowski and Hughes, 1979), P2 peptides (Rostami et al., 1988; Olee et al., 1989; Hahn et al., 1991) and P0 (Milner et al., 1987) has produced the disease. CD41 T-cell lines against P2 (Linington et al., 1986) or P0 (Linington et al., 1992) peptides, when injected intravenously, will adoptively transfer histologically similar disease to naive recipients. If allowed to survive, some animals with EAN develop either a progressive or a relapsing disease which resembles chronic inflammatory demyelinating polyradiculoneuropathy (CIDP). Histologically, CIDP is similar to EAN but with the eventual development of ‘onion bulbs’, i.e. concentric layers of Schwann cell processes around demyelinated axons, especially in the dorsal roots (Craggs et al., 1986; Adam et al., 1989). Such changes are typically seen in the peripheral nerves in both CIDP and HMSN type 1. T-cell (Khalili-Shirazi et al., 1992) and antibody (KhaliliShirazi et al., 1993) responses to P2 and P0 have been identified in some patients with Guillain–Barre´ syndrome and CIDP, and although their pathogenic importance remains uncertain, complement-fixing antibodies against myelin antigens provide a plausible mechanism of inducing demyelination. Anti-PMP22 antibodies may play a pathogenic role in EAN induced by bovine myelin since titres of IgM antibodies to PMP22 have been found to be particularly high in active disease (Koehler et al., 1996). From this, and in view of its predominant expression in the peripheral nervous system and its role in an inherited demyelinating peripheral neuropathy, PMP22 is a potentially important antigen in human inflammatory neuropathy.

and DNA of positive clones was produced by maxipreparation and sequenced using the Sequenase kit (Amersham, USB, Cleveland, OH, USA). E. coli BL21 (protease-deficient) were transformed with this product. Large-scale bacterial broths were grown to an optical density (290 nm) of 0.7 at 37°C, induced to express PMP22 using the lactose analogue isopropyl β-D-thiogalactoside for 4 h, harvested and lysed in the presence of protease inhibitors (aprotinin and phenylmethylsulphonylfluoride). The lysate was incubated with 1% DDM (N-dodecyl-B-D-maltoside) (Sigma) for 30 min at room temperature, then cleared of cellular debris by centrifugation. The cleared lysate was applied to a column of glutathione Sepharose 4B (Pharmacia Biotech). The column was washed with ice-cold PBS (phosphate-buffered saline) and eluted with 10 mM reduced glutathione in 50 mM Tris–HCl (pH 8). Reduced glutathione and solute were removed by dialysis against distilled water/ 0.01% DDM. Fusion protein was identified by SDS–PAGE analysis, stained with Coomassie Blue and by Western blot (Fig. 1). Proteins were run on a 10% SDS–PAGE, transferred overnight to nitrocellulose paper, blocked in 10% milk powder (Marvel)/0.3% Tween–PBS, incubated at room temperature with the primary antibody at 1 : 100 (goat anti-GST, Pharmacia Biotech, UK), washed in 0.3% Tween–PBS, incubated at room temperature with the secondary antibody (horseradish peroxidase conjugated rabbit anti-goat immunoglobulin, Sigma; St Louis, Mo., USA), washed in 0.3% Tween–PBS and developed with ECL (Amersham).

Material and method Expression and purification of PMP22 protein Total RNA was extracted from freshly frozen adult rat sciatic nerves by the method of Chomczynski and Sacchi (1987) (final concentration of total RNA 5 0.343 µg/µl) and full-length PMP22 cDNA was produced by the reverse transcriptase–polymerase chain reaction (RT–PCR) using the published sequence (Edomi et al., 1993) to select appropriate oligonucleotide primers. Primers, as below, were synthesized on an ABI 391 DNA synthesizer: Antisense: 59 TCATCTCGAGTCCCTCCCTGTGTACCCTATGCACG 39 Sense: 59 GAATTCATGCTTCTACTCTTGTTGGGGATCCTGTTCC 39

The RT–PCR product was cleaned by phenol–chloroform extraction, precipitated overnight at –20°C and digested with Xho1 and EcoR1 (Boehringer, Mannheim, Germany). pGEX4T-1 (27–4580–01) (Pharmacia Biotech) was digested with the same restriction enzymes. Both digests were cleaned using the Qiaex II gel purification method (Qiagen) and the products were ligated at 16°C overnight. Competent Escherichia coli XL1B were transformed with the product,

Fig. 1 Freeze-dried products of affinity purification used for immunization. Total fusion protein (20 µg) (GST–PMP22, left-hand lanes) or GST alone (20 µg) (right-hand lanes) were loaded on to a 10% sodium dodecyl sulphate–polyacrylamide (SDS–PAGE) gel and stained with Coomassie blue (upper plate) or transferred to nitrocellulose and Western blotted with anti-GST antibodies (lower plate). GST–PMP22 fusion protein is indicated by arrowheads. As illustrated in the Coomassie-stained gel, the fusion product was unstable and broke down to smaller products, including GST alone; GST alone was therefore used as a control in all experiments. The two plates are not taken from the same gel (and therefore molecular weight markers are not at identical distances).

PMP22 induces EAN Quantities of GST alone were prepared and purified by identical methods. Proteins were concentrated by freezedrying overnight and dissolved in PBS to give a final concentration of 1–2 mg/ml.

Preparation of PMP22 peptides Two peptides [first extracellular domain (ECD) peptide, ECD1, 53–64 primary amino acid sequence of human PMP22, Cys-Phe-Ser-Ser-Ser-Pro-Asn-Glu-Trp-Leu-Gln-Ser, and second ECD peptide, ECD2, 117–133, Tyr-Thr-Val-ArgHis-Pro-Glu-Trp-His-Leu-Asp-Ser-Asp-Tyr-Ser-Tyr-Gly] were synthesized on an ABI 431A Peptide Synthesizer on a ‘solid phase’ hydroxymethylphenoxymethyl polystyrene resin using standard Fmoc chemistry. Peptide was cleaved from the resin (with 95% trifluoroacetic acid using 5% ethanedithiol : anisole 1 : 3 (v/v) as scavenger to minimize side reactions with Trp residues in the peptide), filtered, washed in ether, dried and applied in water to a G10 desalting column. Fractions containing peptide were monitored by absorbance at 220 nm and freeze-dried. The purity of the peptide was determined using an HPLC 400 Solvent Delivery system, obtaining readings from an ABI 783A Programmable Absorbance Delivery System. A 5 mg/ml solution of peptide in PBS was used for injection.

1897

heat. The left sciatic and tibial nerves were stimulated with supramaximal electrical stimuli delivered at the sciatic notch and ankle using monopolar needle electrodes (stimulus duration 0.1 ms, 2 3 supramaximal voltage, 1 Hz). Recordings were made of the EMG signal from the dorsum of the left hind foot. The magnitude and latency of the compound motor action potentials (CMAPs) obtained from proximal and distal stimulation were determined, and motor nerve conduction velocity was calculated.

Collection and processing of tissues Under anaesthesia, following EMG recording, blood samples (2–5 ml) were taken from each animal by cardiac puncture and the serum was stored at –70°C. Nine out of 10 PMP22immunized animals, eight out of eight GST-immunized animals, one out of two ECD1-immunized animals and one out of two ECD2-immunized animals were perfused under deep general anaesthesia with glutaraldehyde (3% in 0.1 M phosphate buffer). The cauda equina and left sciatic nerve were removed and processed into resin using a standard protocol for preparation of 1 µm sections stained with thionin acetate and acridine orange.

Serum analysis Induction of EAN (day 0) Male Lewis rats (200–250 g, Harlan, UK) were housed in cages in pairs, and allowed free access to food (rat chow) and water. Animals were immunized under anaesthesia (1.5– 2% halothane in oxygen) with 0.05 ml of an emulsion containing 0.5 mg Mycobacterium tuberculosis and either 50 µg (four animals) or 100 µg (six animals) PMP22 fusion protein, 50 µg (two animals) or 100 µg (six animals) GST, 250 µg ECD1 (two animals) or 250 µg ECD2 (two animals) in incomplete Freund’s adjuvant into the right hind footpad. Animals immunized with PMP22 fusion protein or GST were paired for weight.

Monitoring of animals Rats were weighed and examined daily for 20 days for signs of disease by an observer blind to the immunization protocol. The severity of EAN was scored according to an 18-point protocol (Gabriel et al., 1997) in which one point was given for each sign of disease. The mean clinical score was the mean score for each group on each day.

PMP22 fusion protein, GST, ECD1 and ECD2 were suspended at 6 µg protein/ml in PBS. Microtitre wells of ELISA (enzyme-linked immunosorbant assay) plates (Costar) were coated with 50 µl aliquots and incubated overnight at 4°C. Non-specific binding was blocked with 0.5% gelatin in PBS for 1 h at room temperature. Rat serum (50 µl) [in serial dilutions of 1/25–1/3200 in 0.5 mg/ml BSA (bovine serum albumin) in PBS] was added to each well and incubated at room temperature for 1 h. All sera were tested in duplicate. After washing three times in 0.5 mg/ml BSA–PBS, 50 µl of 1/3000 anti-rat immunoglobulin horseradish peroxidase conjugate (Sigma) was added and incubated for 1 h at room temperature. The plates were washed three times with 0.5 mg/ml BSA–PBS, developed with O-phenylenediamine (Sigma) and 0.02% (v/v) H2O2 in citrate buffer. Optical density was determined at 490 nm. The reactions of the animals immunized with PMP22 fusion protein, GST and those of a naive Lewis rat, to total fusion protein were determined before and after absorption of the sera with GST protein. For the absorption, 25 µl serum in 0.5 ml BSA–PBS were incubated with 80 µg GST at 37°C for 40 min, after which the supernatant was removed by centrifugation for ELISA assay.

Electrophysiological assessment A terminal EMG examination was performed on the 12 animals from the groups immunized with 100 µg PMP22 or GST. Rats were anaesthetized (1.5–2% halothane in oxygen) and the rectal and subcutaneous temperatures of the left leg were monitored and maintained at 37.5 6 0.5°C using radiant

Histological analysis Transverse sections of the cauda equina at the level of L4 and of the mid-thigh portion of the sciatic nerve were examined by an observer blinded to the immunizing antigen. Specimens were graded for oedema, axonal degeneration,

1898


demyelination and cellular infiltration on the following scale: 0 5 normal; 1 5 mild, with ,10% of the cross-sectional area/1–9 myelinated nerve fibres affected; 2 5 moderate, with 10–50% of the cross-sectional area/10–50 myelinated fibres affected; 3 5 severe, with .50% the cross-sectional area/.50 myelinated nerve fibres affected.

were significantly lower (P , 0.01, Mann–Whitney U test) in the PMP22-immunized group than in the control group immunized with GST. There was no significant difference in distal motor latencies or nerve conduction velocities (Table 1).

Serological responses to PMP22 Results Induction of clinical signs of disease In the group immunized with PMP22, five out of the 10 animals (two out of four immunized with 50 µg PMP22, three out of six with 100 µg PMP22) developed mild weakness of the tail tip. The onset of weakness was on day 10 after immunization in one animal, day 11 in two and day 13 in two, and weakness persisted throughout the experiment. There was no relationship between time of onset or severity of disease and the dose of PMP22 administered. More severe disease was not seen. Disease was not seen in any of the other groups (Fig. 2).

Reduction of CMAP amplitude The mean proximally and distally evoked sciatic nerve compound motor action potentials (CMAPs) after 20 days

Fig. 2 Mean clinical scores following immunization with either PMP22 fusion protein (10 animals) or GST protein (8 animals).

All animals formed antibodies against the immunizing antigen used in each case (data not shown). In the PMP22 fusion protein-immunized animals, antibody titres against total PMP22–GST fusion protein were unchanged by prior absorption of the sera with GST protein, whereas in the GSTimmunized animals the titre was reduced to background levels (Fig. 3).

Demyelination of the cauda equina Histological specimens from 21 animals were examined. This amounted to 21 sciatic nerves and 108 cauda equina nerve roots. Four out of nine (two out of three immunized with 50 µg PMP22, two out of six with 100 µg PMP22) animals immunized with PMP22 fusion protein showed histological changes in the cauda equina, with changes in a total of seven roots. These consisted of sparse multifocal isolated demyelinated fibres, sometimes with mononuclear cell infiltration and macrophages laden with myelin debris in some roots (grade 1 changes for demyelination and cellular

Fig. 3 Optical density (OD) at 490 nm of sera from animals immunized with PMP22 fusion protein or GST protein and from a naive Lewis rat before (continuous lines) and after (dotted lines) absorption with GST protein.

Table 1 Electrophysiological results in animals immunized with PMP22 fusion protein or GST protein

PMP22-immunized (n 5 6) GST-immunized (n 5 6) P

CMAP* (mV)

CMAP† (mV)

Latency* (ms)

Latency† (ms)

Nerve conduction velocity (m/s)

5.0 (2.2) 8.0 (2.7) ,0.01

4.0 (1.6) 7.3 (2.8) ,0.01

1.3 (0.1) 1.3 (0.07) NS

2.65 (0.22) 2.65 (0.13) NS

43 (5.5) 45 (3.9) NS

Values are given as means (SD). Two-tailed P values are for differences between the groups (Mann–Whitney U test). CMAP 5 compound motor action potential; GST 5 glutathione-S-transferase. *Stimulation of the tibial nerve at the ankle. †Stimulation of the sciatic nerve at the sciatic notch.

PMP22 induces EAN infiltration in all abnormal fibres, illustrated in Fig. 4A and B). There was no evidence of a dose-dependent effect on the quantity or severity of histological change. No animals in

1899

any of the other groups showed similar abnormalities. There were no abnormalities in the sciatic nerves in any of the groups.

Fig. 4 (A) Light micrograph showing isolated demyelinated nerve fibres (small arrowheads) and macrophages laden with myelin debris (large arrowheads) in a spinal root of one animal immunized with 50 µg PMP22 fusion protein (bar 5 30 µm). (B) Electron micrograph showing demyelinated axons (arrows) (bar 5 5 µm).

1900


Discussion Induction of EAN by myelin proteins

Amount of myelin protein required for induction of EAN

Human inflammatory neuropathies are likely to be initiated by a heterogeneous group of antigens, and those capable of inducing EAN are regarded as prime candidates. Our findings demonstrate that small amounts of recombinant PMP22 can induce mild EAN in the Lewis rat. Previously, the myelin proteins P2 (Kadlubowski and Hughes, 1979) and P0 (Milner et al., 1987) have been shown to induce EAN, and therefore these are also candidate antigens in Guillain–Barre´ syndrome. B-cell responses to P2, P0 and other myelin proteins were identified in EAN induced by bovine peripheral nerve myelin (Zhu et al., 1994), and T-cell responses and antibodies against these proteins have been found in the serum of some patients with Guillain–Barre´ syndrome and CIDP (Khalili-Shirazi et al., 1992, 1993). However, since these responses are found only in a minority of patients and no pathogenic role for them has been identified, their importance in Guillain–Barre´ syndrome has been questioned (Pette et al., 1994). Other antigens may therefore be more important, and PMP22, which has extracellular domains accessible to immune attack, is an attractive candidate. In our experiments, PMP22 peptides corresponding to the human sequence did not elicit disease in Lewis rats. These peptide sequences were chosen on the basis that they represent the predicted ECDs of the PMP22 protein, and that these regions might include the exposed antigens. In addition, the method of Hopp and Woods (1981, 1983) was used to predict antigenic determinants, and this corresponded to part of the ECD1 molecule. The human and rat protein sequences exhibit 85% homology throughout the whole protein and 75% (ECD1) and 82% (ECD2) in our peptides, and therefore species differences may have been important, although it is possible that higher doses might have elicited disease. Alternatively, these peptides may not be the immunodominant sequences of PMP22, or such peptide fragments may not be appropriately processed by antigen-presenting cells. Certain P2 peptides can actively induce EAN (Olee et al., 1989, 1990; Hahn et al., 1991; Rostami and Gregorian, 1991). Although direct induction of EAN with P0 peptides has not been demonstrated, sequences from both the cytoplasmic and immunoglobulin-like extracellular domains induce strong Tcell responses if used to immunize Lewis rats, and T cells from such animals induce EAN by adoptive transfer (Linington et al., 1992). A number of viral proteins, some of which have been associated with Guillain–Barre´ syndrome, have absolute sequence homology involving five or more amino acids with P0. However, neither peptide sequences of these regions (Adelmann and Linington, 1992) nor other P0 peptides appear to induce EAN actively (Linington et al., 1992). Our peptides are not homologous to any known viral protein sequences, although ECD2 has sequences of five amino acids in common with a Mycoplasma attachment protein and with two E. coli proteins.

The disease induced in our experiments was mild. Immunization with larger amounts of a more concentrated product might induce more severe EAN. However, we were limited in both the total amount and the concentration of protein that could be produced. PMP22 is a particularly hydrophobic molecule with four predicted transmembrane domains (Snipes et al., 1992), and this resulted in difficulties with solubilization. The protein also acts as a growth inhibitor in Schwann cells (Zoidl et al., 1995) and is identical to a fibroblast growth arrest protein (Edomi et al., 1993). We found that the growth of the E. coli transformed by our clone was also inhibited (unpublished observation), which is compatible either with toxicity of the protein in bacteria or with its action as a growth inhibitor in bacteria similar to that seen in Schwann cells and fibroblasts. We cloned our protein from rat sciatic nerve, and the PMP22 with which the rats were immunized was thus homologous. It may be that more severe disease could have been elicited if heterologous protein had been used, as has been reported for EAN induced with whole nerve (Levine and Wenk, 1963). The neuritogenic potential of myelin proteins does not appear to be related to the proportion of that protein present in peripheral nerve. PMP22 has been estimated to constitute 2– 10% of total peripheral nerve myelin protein (Snipes et al., 1992; Koehler et al., 1996). In this study, approximately 25– 50 µg PMP22 (50–100 µg total fusion protein) induced histological changes and mild clinical changes in some animals. By comparison, in EAN in the Lewis rat induced by P2 protein (the quantity of which varies between species but is not more than 15% of total myelin protein) prepared from an acid extract of bovine spinal root myelin, 5 µg was sufficient to induce disease in some animals, although 25 µg was necessary to induce mild disease in all animals (Kadlubowski et al., 1980). However, when EAN was induced with P0 (which constitutes up to 60% of peripheral nerve myelin protein), 200 µg P0 was necessary to induce mild neurological signs in three out of 10 animals and histological change in seven out of 10 (Milner et al., 1987).

Immune responses to PMP22 protein We have shown that antibodies to the PMP22 portion of the fusion protein are produced in immunized animals. Although such antibodies are not necessarily pathogenic, Koehler et al. (1996) have recently correlated high titres of IgM antibodies against PMP22 with clinical disease in EAN induced with bovine PNS myelin. These authors suggest that this immune response is partly directed against the HNK-1 epitope (a glycosylated site known to be capable of eliciting T-cellindependent antibody production) of PMP22. Such a mechanism could not be the case in our experiments, since the bacterially produced PMP22 fusion protein does not undergo post-translational glycosylation and the fusion protein does not therefore include this moiety.

PMP22 induces EAN We were surprised that absorption of sera with GST in animals immunized with the PMP22 fusion protein did not result in at least a slight reduction of the titre of antibodies against this protein, since GST clearly stimulated an antibody response in control animals. A possible explanation is that the response against the PMP22 moiety of the fusion protein was much greater than that against the GST portion. Alternatively, the conformation of GST might be altered when produced as a fusion protein, reducing its antigenicity or preventing recognition of pure GST by antibodies directed against the whole fusion protein. Animals immunized with PMP22 peptides developed a strong antibody response, but did not develop clinical or histological evidence of disease, making it unlikely that, acting alone, antibodies against these extracellular portions of human PMP22 are pathogenic in the rat. However, conformational changes implicit in producing a peptide distinct from its surrounding amino acids could influence the antigenic behaviour of that peptide, or the human and rat sequence differences may be sufficient to prevent the human peptides from generating pathogenic antibodies or T-cell responses to rat myelin. Alternatively, other portions of the whole PMP22 protein may be important in the production of pathogenic antibodies.

Demyelination of the cauda equina Histological changes were mild, but were seen only in those animals immunized with the PMP22 fusion protein. These changes were qualitatively similar to those seen in mild disease produced by either bovine myelin or fractions of other myelin proteins (Lampert, 1969; Powell et al., 1983; Hughes and Powell, 1984). The animals were sampled after 20 days and study of the early stages of demyelination will require further experimentation. Only three of the five animals that developed clinical signs also showed histological changes. In the remaining animals, the absence of histological change could be explained by the sampling error inherent in the histological examination of a disease characterized by multifocal lesions. Since the severity of EAN is known to correlate positively with the dose of myelin used for immunization (Hahn et al., 1988), we predict that more widespread pathological changes would be seen if rats were immunized with a greater quantity of PMP22.

Electrophysiological changes The main target of the autoimmune reaction in EAN is the myelin sheath, and in severe disease evidence of demyelination is characteristically detected on electrophysiological assessment (Heininger et al., 1986). Although characteristic signs of demyelination (e.g. prolongation of distal motor latencies, slowing of nerve conduction velocity and dispersion of the nerve action potential) were not seen in this study, a reduction in CMAP was observed whether it was induced proximally or distally. Such a reduction is usually evidence of

1901

Wallerian degeneration, which is thought in EAN to be due to axonal damage caused by toxic products released in a severe inflammatory reaction (Hahn et al., 1988; Hughes et al., 1990). However, our animals showed only mild inflammation. Alternatively, the electrophysiological results could be explained by distal axonal degeneration or demyelination in terminal motor nerves which should be included in future histological examination. In conclusion, immunization with homologous PMP22 induces mild inflammatory radiculopathy in the Lewis rat. These results warrant further examination of the role of antibody and T cell responses against PMP22 in both EAN and human inflammatory neuropathies

Acknowledgements We wish to thank the Electron Microscope Unit at UMDS, Guy’s Hospital, for their expert help. C.M.G. is funded by the British Brain and Spine Foundation

References Adam AM, Atkinson PF, Hall SM, Hughes RA, Taylor WA. Chronic experimental allergic neuritis in Lewis rats. Neuropathol Appl Neurobiol 1989; 15: 249–64. Adelmann M, Linington C. Molecular mimicry and the autoimmune response to the peripheral nerve myelin P0 glycoprotein. Neurochem Res 1992; 17: 887–91. Chomczynski P, Sacchi N. Single step method of RNA isolation by acid guanidinium thiocyanate-phenol-chloroform extraction. Analyt Biochem 1987; 162: 156–90. Craggs RI, Brosnan JV, King RH, Thomas PK. Chronic relapsing experimental allergic neuritis in Lewis rats: effects of thymectomy and splenectomy. Acta Neuropathol (Berl) 1986; 70: 22–9. Edomi P, Martinotti A, Colombo MP, Schneider C. Sequence of human GAS3/PMP22 full-length cDNA. Gene 1993; 126: 289–90. Gabriel CM, Gregson NA, Redford EJ, Davies M, Smith KJ, Hughes RA. Human immunoglobulin ameliorates rat experimental autoimmune neuritis. Brain 1997; 120: 1533–40. Hahn AF, Feasby TE, Steele A, Lovgren DS, Berry J. Demyelination and axonal degeneration in Lewis rat experimental allergic neuritis depend on the myelin dosage. Lab Invest 1988; 59: 115–25. Hahn AF, Feasby TE, Wilkie L, Lovgren D. P2-peptide induced experimental allergic neuritis: a model to study axonal degeneration. Acta Neuropathol (Berl) 1991; 82: 60–5. Heininger K, Stoll G, Linington C, Toyka KV, Wekerle H. Conduction failure and nerve conduction slowing in experimental allergic neuritis induced by P2-specific T-cell lines. Ann Neurol 1986; 19: 44–9. Hopp TP, Woods KR. Prediction of protein antigenic determinants from amino acid sequences. Proc Natl Acad Sci USA 1981; 78: 3824–8. Hopp TP, Woods KR. A computer program for predicting protein antigenic determinants. Mol Immunol 1983; 20: 483–9.

1902


Hughes RAC. Experimental allergic neuritis. In: Hughes RAC. Guillain–Barre´ syndrome. London: Springer-Verlag; 1990. p. 49–81. Hughes RA, Powell HC. Experimental allergic neuritis: demyelination induced by P2 alone and non-specific enhancement by cerebroside. J Neuropathol Exp Neurol 1984; 43: 154–61. Kadlubowski M, Hughes RA. Identification of the neuritogen for experimental allergic neuritis. Nature 1979; 277: 140–1. Kadlubowski M, Hughes RA, Gregson NA. Experimental allergic neuritis in the Lewis rat: characterization of the activity of peripheral myelin and its major basic protein P2. Brain Res 1980; 184: 439–54. Khalili-Shirazi A, Hughes RA, Brostoff SW, Linington C, Gregson N. T cell responses to myelin proteins in Guillain–Barre´ syndrome. J Neurol Sci 1992; 111: 200–3. Khalili-Shirazi A, Atkinson P, Gregson N, Hughes RA. Antibody responses to P0 and P2 myelin proteins in Guillain–Barre´ syndrome and chronic idiopathic demyelinating polyradiculoneuropathy. J Neuroimmunol 1993; 46: 245–51. Koehler NK, Martin R, Wietholter H. The antibody repertoire in experimental allergic neuritis: evidence for PMP-22 as a novel neuritogen. J Neuroimmunol 1996; 71: 179–89. Lampert PW. Mechanism of demyelination in experimental allergic neuritis. Electron microscopic studies. Lab Invest 1969; 20: 127–38. Levine S, Wenk EJ. Allergic neuritis induced in rats without the use of mycobacteria. Proc Soc Exp Biol Med 1963; 113: 898–900. Linington C, Mann A, Izumo S, Uyemura K, Suzuki M, Meyermann R, et al. Induction of experimental allergic neuritis in the BN rat: P2 protein-specific T cells overcome resistance to actively induced disease. J Immunol 1986; 137: 3826–31. Linington C, Lassmann H, Ozawa K, Kosin S, Mongan L. Cell adhesion molecules of the immunoglobulin supergene family as tissue-specific autoantigens: induction of experimental allergic neuritis (EAN) by P0 protein-specific T cell lines. Eur J Immunol 1992; 22: 1813–7. Matsunami N, Smith B, Ballard L, Lensch MW, Robertson M, Albertsen H, et al. Peripheral myelin protein-22 gene maps in the duplication in chromosome 17p11.2 associated with Charcot–Marie– Tooth 1A. Nat Genet 1992; 1: 176–9. Milner P, Lovelidge CA, Taylor WA, Hughes RA. P0 myelin protein produces experimental allergic neuritis in Lewis rats. J Neurol Sci 1987; 79: 275–85. Olee T, Weise M, Powers J, Brostoff S. A T cell epitope for experimental allergic neuritis is an amphipathic alpha-helical structure. J Neuroimmunol 1989; 21: 235–40. Olee T, Powell HC, Brostoff SW. New minimum length requirement for a T cell epitope for experimental allergic neuritis. J Neuroimmunol 1990; 27: 187–90. Patel PI, Roa BB, Welcher AA, Schoener-Scott R, Trask BJ, Pentao L, et al. The gene for the peripheral myelin protein PMP-22 is a candidate for Charcot–Marie–Tooth disease type 1A. Nat Genet 1992; 1: 159–65. Pette M, Linington C, Gengaroli C, Grosse-Wilde H, Toyka KV, Hartung HP. T lymphocyte recognition sites on peripheral nerve myelin P0 protein. J Neuroimmunol 1994; 54: 29–34.

Powell HC, Braheny SL, Myers RR, Rodriguez M, Lampert PW. Early changes in experimental allergic neuritis. Lab Invest 1983; 48: 332–8. Roa BB, Garcia CA, Suter U, Kulpa DA, Wise CA, Mueller J, et al. Charcot–Marie–Tooth disease type 1A: association with a spontaneous point mutation in the PMP22 gene. N Engl J Med 1993a; 329: 96–101. Roa BB, Dyck PJ, Marks HG, Chance PF, Lupski JR. Dejerine–Sottas syndrome associated with point mutation in the peripheral myelin protein 22 (PMP22) gene. Nat Genet 1993b; 5: 269–73. Roa BB, Garcia CA, Pentao L, Killian JM, Trask BJ, Suter U, et al. Evidence for a recessive PMP22 point mutation in Charcot–Marie– Tooth disease type 1A. Nat Genet 1993c; 5: 189–94. Rostami A, Gregorian SK. Peptide 53–78 of myelin P2 protein is a T cell epitope for the induction of experimental autoimmune neuritis. Cell Immunol 1991; 132: 433–41. Rostami AM, Ventura E, Kimura H, Brown MJ, Pleasure DE. Induction of severe experimental allergic neuritis (EAN) with a synthetic peptide corresponding to the 53–78 amino acid sequence of the myelin P2 protein [abstract]. Neurology 1988; 38 Suppl 1: 375. Sereda M, Griffiths I, Puhlhofer A, Stewart H, Stewart HR, Rossner MJ, et al. A transgenic rat model of Charcot–Marie–Tooth disease. Neuron 1996; 16: 1049–60. Snipes GJ, Suter U, Welcher AA, Shooter EM. Characterization of a novel peripheral nervous system myelin protein (PMP-22/SR13). J Cell Biol 1992; 117: 225–38. Snipes GJ, Suter U, Shooter EM. Human peripheral myelin protein22 carries the L2/HNK-1 carbohydrate adhesion epitope. J Neurochem 1993; 61: 1961–4. Timmerman V, Nelis E, Van Hul W, Nieuwenhuijsen BW, Chen KL, Wang S, et al. The peripheral myelin protein gene PMP-22 is contained within the Charcot–Marie–Tooth disease type 1A duplication [published erratum appears in Nat Genet 1992; 2: 84]. Nat Genet 1992; 1: 171–5. Valentijn LJ, Baas F, Wolterman RA, Hoogendijk JE, van den Bosch NH, Zorn I, et al. Identical point mutations of PMP-22 in Trembler-J mouse and Charcot–Marie–Tooth disease type 1A. Nat Genet 1992a; 2: 288–91. Valentijn LJ, Bolhuis PA, Zorn I, Hoogendijk JE, van den Bosch N, Hensels GW, et al. The peripheral myelin gene PMP-22/GAS-3 is duplicated in Charcot–Marie–Tooth disease type 1A. Nat Genet 1992b; 1: 166–70. Zhu J, Link H, Weerth S, Linington C, Mix E, Qiao J. The B cell repertoire in experimental allergic neuritis involves multiple myelin proteins and GM1 [see comments]. J Neurol Sci 1994; 125: 132–7. Comment in: J Neurol Sci 1995; 128: 111. Zoidl G, Blass-Kampmann S, D’Urso D, Schmalenbach C, Mu¨ller HW. Retroviral-mediated gene transfer of the peripheral myelin protein PMP22 in Schwann cells: modulation of cell growth. EMBO J 1995; 14: 1122–8.

Received January 22, 1998. Revised May 5, 1998. Accepted June 10, 1998