Journal of the Neurological Sciences, 1989, 89:289-300 289 Elsevier

JNS 03121

Normal rat serum cytotoxicity against syngeneic oligodendrocytes

Complement activation and attack in the absence of anti-myelin antibodies

N.J. Scolding 1, B.P. Morgan 2, A. Houston 2, A.K. Campbell 2, C. Linington I and D .A .S . Compston 1

l Secn'on of Neurology, Department of Medicine, and 2Department of Medical Biochemistry, University of Wales College of Medicine, Cardiff CF4 4XN (U.K.)

(Received 29 August, 1988) (Revised, received 17 October, 1988) (Accepted 21 October, 1988)

SUMMARY

The role of complement in mediating oligodendrocyte and myelin injury has been investigated by studying the effects of normal adult rat serum on syngeneic cultured neonatal glial cells. Rat serum has cytotoxic activity directed against oligodendrocytes but not astrocytes, the potency of which increases with cell maturation. The effects of heat inactivation, decomplemented rat serum, EGTA treatment, removal of any possible anti-myelin antibody by absorbtion using syngeneic myelin and absence of surface staining for immunoglobulins on serum-treated oligodendrocytes, C9 depletion and reconstitution, and oligodendrocyte staining for surface C9 demonstrate that this cyto- toxicity is mediated by complement via antibody independent activation of the classical pathway and is membrane attack complex dependent. These fmdings significantly extend the previous demonstration of complement activation by extracted myelin, and may have significance for the pathogenesis of demyelinating diseases.

Key words: Oligodendrocytes; Complement activation; Multiple sclerosis; Demyeli- nating diseases

Correspondence to: Professor D. A. S. Compston, University of Cambridge, School of Clinical Medi- cine, Addenbrooke's Hospital, Hills Road, Cambridge CB2 2QQ, U.K.

0022-510X/89/$03.50 © 1989 Elsevier Science Publishers B.V. (Biomedical Division)

290

INTRODUCTION

Several indirect lines of evidence implicate complement in the pathogenesis of human and experimental demyelinating disease. Abnormalities in the concentration of individual complement components in the cerebrospinal fluid (CSF) of patients with episodes of demyelination are consistent with intrathecal complement activation (Compston et al. 1986), and perivascular staining for complement membrane attack complexes (MACs) has been demonstrated in samples from patients with multiple sclerosis (MS) (Compston et al. 1989).

The mechanism of complement activation remains obscure. One possibility is that antibodies to myelin or oligodendrocyte surface components activate the classical pathway after binding to their target antigens - an interpretation which is consistent with intrathecal IgG synthesis (Tourtellotte et al. 1985) and the occurrence of oligoclonal bands in the CSF of patients with MS (Ebers 1984). Furthermore, antibody mediated demyelination in the Lewis rat is complement mediated and depends on f'Lxation of C 1 by the Fc portion of the demyelinating antibody (Linington et al. 1989). However, it is now well established that the antibodies found in the CSF of patients with MS are not directed against surface myelin or oligodendrocyte components (Lubetzki et al. 1986) and another explanation is required for complement activation in demyelinating disease.

In vitro, purified CNS myelin activates complement via the classical pathway in an antibody-independent manner (Cyong et al. 1982; Vanguri and Koski 1982), but these results cannot directly be extrapolated to suggest activation of complement at the surface membrane of intact living oligodendrocytes because of the disruption of the myelin membranes during extraction and, in particular, the exposure of normally internal components.

The demonstration of antibody independent complement activation by intact living oligodendrocytes would, however, resolve the paradox presented by the occur- rence of complement activation in the absence of myelin or oligodendrocyte specific antibodies. We have investigated this possibility using cultures of neonatal rat optic nerve and studying their interaction with normal syngeneic serum as a source of complement. Antibody independent cytotoxieity has been demonstrated which depends on classical pathway complement activation. It will now be important to identify the surface component of the oligodendrocyte-myelin complex responsible for this activity.

METHODS

Preparation of oligodendrocytes Oligodendrocytes were cultured from neonatal rat optic nerve as previously

described (Raft et al. 1983; Scolding et al. 1988).

Immunofluorescence Oiigodendrocytes were identified by indirect immunofluorescence, using

monoclonal mouse anti-galactocerebroside (GalC) antibody (Ranscht et al. 1982), and

291

polyclonal rabbit antibodies against MBP (Sternberger et al. 1978) prepared in our laboratory, and proteolipid protein (PLP; Agrawal et al. 1977). Astrocytes were identi- fied by polyclonal rabbit anti-glial fibrillary acidic protein (GFAP) antibody (Bignami et al. 1972). Primary antibody binding was visualised using either FITC (fluorescein isothiocyanate) labelled sheep anti-mouse IgG antibody or TRITC (tetramethyl- rhodamine isothiocyanate) labelled goat anti-rabbit IgG antibody (Sigma). The cover- slips were mounted with Citifluor (City University, London) to prolong fluorescence (Johnson et al. 1982). Culture purity was assessed by counting at least 200 cells on each of 4 or more coverslips.

Serum Blood obtained by cardiac puncture from ether anaesthetised adult Wistar rats

was left at 0°C to clot for 30 min and centrifuged (1000 x g for 30 min).

Cytotoxicity experiments Fresh serum was added to cultured oligodendrocytes. After incubating at 37 °C

in 5 ~o CO2 for 20 min (unless otherwise stated) coverslips were taken out, washed gently by dipping in fresh culture medium, and re-immersed in fresh culture medium pre-equilibrated with the above culture conditions. During the course of cytotoxicity experiments, serum from over 30 rats was tested at dilutions ranging from 20 to 1 ~ .

Cell damage was assessed using propidium iodide (PI) (Parks et al. 1986), a poorly fluorescent substance which when intercalated with DNA becomes highly fluorescent. Healthy cells exclude it, whereas damaged cells become permeable and therefore develop fluorescent nuclei. PI was added to a fmal concentration of 100 #mol; the cells were studied using a Zeiss inverted fluorescence microscope and photographed with a Contax camera. 100-200 ceils per coverslip were counted, and the percentage showing PI uptake was calculated.

Decomplementation of rats by cobra venom factor (CVF) CVF was isolated from lyophilised cobra venom (Naja naja kaouthia, Sigma)

(Vogel and Mueller-Eberhard 1984). The purified protein was free of phospholipase A2 activity and was functionally active in vitro and in vivo. Serum was taken from adult Wistar rats 24 h after a single intra-peritoneal dose (750 #g/kg) of cobra venom factor, which caused complete elimination of complement activity for at least 48 h (Linington et al. 1988).

Preparation of rat C9, anti-C9 antibody, and C9 depleted serum C9 was purified from rat serum (Serotech Ltd., Kiddington) by a modification

of the method for human C9 (Biesecker and Muller-Eberhard 1980). Briefly, C9 was precipitated from serum by polyethylene glycol 4000 (10.5-25 ~o) and the resolubilised pellet sequentially chromatographed on lysine-sepharose, DEAE-sephacel, hydroxy- apatite and ultrogel ACA 34. The f'mal product was pure as assessed by SDS-PAGE, and retained full haemolytic activity.

292

Antiserum to rat C9 Antiserum was raised by immunising rabbits with purified C9 using standard

methods. The IgG fraction of this antiserum was purified by sodium sulphate precipi- tation and immobilised on CNBr-activated sepharose 4B (Pharmacia).

C9 depleted serum This was prepared using the above antibody. Serum treated with 4.6 mmol EDTA

to prevent non-specific complement activation was added to anti-C9 antibody immobilised on sepharose 4B and the mixture agitated for 15 min at 4°C, then centrifuged at 2000 x gfor 1 min. The supernatant serum was removed and calcium and magnesium restored to physiological concentrations prior to use.

RESULTS

Immunocytochemieal characterisation of cultures Using immunofiuorescence the cultures were found to comprise a very pure

population of oligodendrocytes, especially in the first week. Later, astrocytes and fibroblasts derived from the meninges around the optic nerves proliferated and oligo- dendrocyte purity gradually decreased. Thus after 3 days in vitro (d.i.v.) 93 ~o (SD 3.5) of cells were GalC positive and had oligodendrocyte morphology, and 3 ~o (SD 3.4) were GFAP positive astrocytes. After 6 d.i.v 85 ~ (SD 1.0) cells were GalC positive and 3 9o GFAP positive, and after 9 d.i.v 72~o (SD 3.2) were GalC positive and 17~o GFAP positive. At this stage GalC positive cells also expressed markers of more mature oligodendrocytes such as MBP and PLP.

Oligodendrocyte-specific cytotoxicity of fresh rat serum Serum exerted a cytotoxic effect on oligodendrocytes (Fig. 1). No cytotoxic effect

on contaminating astrocytes or other cells was observed. Astrocytes in mixed cultures grown from whole neonatal rat brain (Chapman and Rumsby 1982) were similarly not affected by serum. No PI uptake or morphological changes were observed in cells exposed to PI alone.

Serum cytotoxicity was reduced by a single freeze-thaw cycle or by leaving serum on ice for over 4 h. The extent of reduction varied considerably with different batches of serum. Fresh serum was therefore used for all experiments, though rat serum from commercial sources (Serotech) was found to exhibit significant (but less potent) cytotoxic activity (not shown).

The sequence of morphological changes was studied by adding 10~o serum to 9 d.i.v oligodendrocytes in culture wells on the microscope stage (Fig. 2). First, the cell body and nucleus were seen to swell and the cytoplasm became granular. The cell processes then degenerated, each becoming pinched off'into a series of'beads', and the cell body eventually became completely globular. By this stage, adherence to the coverslip weakened, the cell body being seen to tremble with minor vibrations of the stage. The time course of oligodendrocyte permeabilisation to PI was studied by taking

293

Fig. 1. Offal cells after exposure to fresh ra t serum. An oligodendrocyte (top left) is shown to have a densely staining PI positive nucleus (a: double exposure, phase contrast + PI optics) and confirmed to be Gal C positive (b: fluorescein optics). The morphology of the processes has been considerably disrupted, though the cell remains adherent to the coverslip. The Gal C negative cells (bottom right) are probably astrocytes

and have PI negative nuclei. ( x 400.)

coverslips out of the culture medium/serum mixture after various incubation times, washing gently several times in fresh culture medium, and examining cells in the presence of PI. Approximately 50~ oligodendrocytes were found to be PI positive after 15 rain exposure to serum; 9 5 ~ after 20 min.

Oligodendrocyte maturation increases susceptibility to serum damage. The effect of serum on oligodendrocytes at various stages of in vitro development was examined in detail by dose-response experiments. Oligodendrocytes showed increasing susceptibility to serum cytotoxicity over the period 2 -6 d.i.v (Fig. 3). Thus 10~ fresh rat serum rendered 8.5% (SD 1.0) of 2 d.i.v oligodendrocytes PI-permeable, 63 % (SD 5.0) at 4 d.i.v, and 97% (SD 3.1) at 6 d.i.v. There was no increase in sensitivity beyond 6 d.i.v.

Initial characterisation of cytotoxic factor Molecular weight. Using ultracentrifugation with a 30 K Centricon the cytotoxic

factor was found to be > 30 kDa in molecular weight. Relationship to arachidonic acid derivatives. Fresh serum obtained from rats

pretreated with intraperitoneal indomethacin (15 mg/kg for 3 days) had cytotoxic

294

Fig. 2. Oligodendrocytes exposed to serum. Two oligodendrocytes have been followed using phase contrast (a,b) and PI optics (c,d), and are shown prior to exposure to serum (a,c) then after 40 minutes' exposure

(b,d). See text for details. ( × 400.)

activity equal to normal serum. These results excluded arachidonic acid derivatives and also superoxides from a role in mediating serum cytotoxicity, the latter having a low molecular weight and also being too short lived. The effect of heat treatment (see below) and activity even at 5 Yo dilution suggested that non-specific osmotic or toxic handling effects were not responsible. These results were, however, consistent with a role for complement.

Relationship of the cytotoxic factor to complement Heat treatment. Cytotoxic activity was entirely abolished by heating serum at

56 °C for 30 min. CVF treatment of rats. Serum from decomplemented rats had no effect on

oligodendrocytes (Fig. 4) even at a dilution of 50Yo (not shown). Oligodendrocytes pre-incubated with anti-GalC antibody were also unaffected by serum (at 20%) from CVF-treated rats.

EGTA treatment. In order to establish the mode of complement activation EGTA (ethylene glycol-bis[fl-aminoethyl ether]-N,N,N',N'-tetraacetic acid) was used

295

l(m

=_

I 20 10 5 2 1

PERCENT SERUM (v/v)

Fig. 3. Serum cytotoxicity increases with oligodendrocyte maturation. The toxicity of serum, indicated by the proportion of cells permeabilised to PI, is shown to be greater with 6 d.i.v oligodendroeytes (1) than 4 d.i.v cells (O), which are in turn more susceptible than at 2 d.i.v (D). Each point represents the mean

( + SD) from at least 3 coverslips; at least 100 process bearing cells per coverslip were counted.

selectively to inhibit the classical pathway. Oligodendrocytes (6 d.i.v) were pre- incubated in culture medium supplemented with 10 mmol EGTA and 10 mmol MgC12 for 30 rain at 37 °C. In control experiments, no cells became PI permeable as a result of this procedure. 20~ serum with 10 mmol EGTA and 10 mmol MgC12 was then added. Untreated control serum rendered 95 ~o (SD 3.9) cells PI permeable, compared with 6.89/0 (SD 3.1) by EGTA-treated serum.

Role of antimyelin antibody in complement activation. In order to remove any possible antimyelin antibody, serum (treated with 4.6 mmol EDTA to prevent comple- ment activation) was incubated for 15 rain at 0 °C with an equal volume of syngeneic CNS myelin (10 g/1 in EDTA-phosphate-buffered saline; Norton and Poduslo 1973). This incubation was repeated, calcium and magnesium restored to physiological con- centrations, and the serum added to 6 d.i.v oligodendrocytes. Myelin-adsorbed serum rendered 77 ~o (SD 4.8) oligodendrocytes PI permeable, compared with 78 ~o (SD 0.14) by control serum (EDTA-treated, then calcium and magnesium restored). To investi- gate further any possible role for antibody in serum cytotoxicity, oligodendrocytes were exposed to EDTA-treated serum both before and after incubation with myelin, then stained with FITC-conjugated goat anti-rat IgG antibody; no antibody was detectable on the oligodendrocyte surface,

Removal and reconstitution of serum C9. C9 depleted serum caused minimal oligodendrocyte permeabilisation (Fig. 4); reconstitution with purified rat C9 restored cytotoxic activity.

Presence of surface C9 on oligodendrocytes exposed to serum. Membrane bound C9 was demonstrated on oligodendrocytes after sublytic exposure to serum by indirect

296

l O e

O

g

5£

L

~J

A B C D E F

Fig. 4. Complement dependence of serum cytotoxicity. In control studies the percentage of PI-positive oligodendrocytes in normal culture medium was low (A), and near total permeabilisation occurred when 6 d.i.v cells were exposed to 10~o serum (B). As a control for the C9 depletion experiment, serum treated with bovine antiparathormone antibody was found to have only slightly reduced cytotoxieity (C), whereas serum depleted of C9 by treatment with anti-C9 antibody had very little activity (D). Cytotoxicity was restored to C9 depleted serum by reconstitution using pure rat C9 (E). Finally, cytotoxicity was absent from

deeomplemented serum (20Fo) from CVF treated rats (F). See text for details.

Fig. 5. Surface C9 stainingofoligodendrocytes after subtytic exposure to serttm. Cells (6 d.i.v) were exposed to 10% serum for 12 min, then washed and double-labelled using the rabbit anti-C9 antibody (visualised with fluorescein conjugated anti-rabbit IgG antibody) and mouse monoclonal anti-GalC antibody (visualised using rhodamine conjugated anti-mouse IgG antibody). Surface C9 (fluoreseein optics) is shown; visualisa-

tion with rhodamine optics confirmed the cells to be Gal C positive oligodendroeytes (not shown). ( x 400.)

297

immunofluorescence using rabbit anti-C9 antibody (Fig. 5). C9 was not detected on GalC negative cells or oligodendrocytes not exposed to serum; FITC conjugate did not significantly stain serum-treated oligodendrocytes when the fh'st layer of antibody (anti-C9) was omitted.

DISCUSSION

We have demonstrated that normal rat serum is specifically cytotoxic to cultured rat oligodendrocytes, and that this activity depends on complement activation and is membrane attack complex dependent. The evidence includes the lack of toxicity of CVF treated and C9 depleted serum, restoration of cytotoxicity to the latter by reconstitution with purified C9, and the demonstration of C9 on the surface of oligodendrocytes after exposure to serum.

The effect of EGTA, together with the cell specificity of toxicity, implies that complement is activated by oligodendrocytes via the classical pathway. Furthermore, myelin adsorption does not reduce serum cytotoxicity, and surface antibody is not found on oligodendrocytes exposed to EDTA-treated (non-lytic) serum; it is therefore sug- gested that complement activation is antibody independent. Antibody independent classical pathway complement activation by purified CNS myelin has previously been demonstrated (Cyong et al. 1982; Vanguri et al. 1982), but in these reports disruption of the myelin membranes during extraction would have exposed internal components which might have been responsible - indeed Cyong et al. suggested that it was the internal protein MBP that activated complement. These results with myelin cannot therefore be extrapolated to intact cells, and we are not aware of any studies showing complement activation by living oligodendrocytes.

The increasing potency of serum cytotoxicity with in vitro oligodendrocyte age is unlikely to be explained by a yield of fragile oligodendrocytes increasingly vulnerable to non-specific insult including activated complement, because the trend of increasing sensitivity to complement attack ceases at 6 d.i.v. We have maintained morphologically normal PI negative oligodendrocytes in culture for 30 days. Although we have no direct evidence, a more likely explanation is that a membrane component responsible for complement activation is linked to ollgodendrocyte maturation. It is during this first week of cell development that the various identified oligodendrocyte markers sequen- tially appear (Bologa 1983); in the culture system we used, GalC, MBP and PLP are only demonstrable immunocytochemically after 2, 3 and 5 days, respectively; meanwhile the number ofoligodendrocytes still expressing the A2B5 progenitor-marker diminishes. The density of a putative complement activating component or receptor might similarly by very low in 1-2-day-old oligodendrocytes, increasing over the next few days to account for the changing susceptibility to serum. Alternatively, a complement-inhibiting factor present on immature cells might have disappeared; some mammalian cells do carry surface components which protect against autologous complement attack (Blaas et al. 1988). The ability of purified murine CNS myelin to activate complement by the classical pathway also increases with age (Tanaka and Cyong 1985).

298

Considering the wealth of data concerning the in vitro effects of sera from animals with EAE and from MS patients (Seil 1977), it might be considered surprising that complement mediated oligodendrocyte damage by normal serum has not previously been described. A number of explanations can be offered. First, most such studies have used organotypic CNS cultures rather than dissociated oligodendrocytes, and have assessed either demyelination or inhibition of myelin formation, rather than directly observing oligodendrocyte injury. Secondly, very few studies have used syngeneic systems. We have preliminary data suggesting that normal guinea pig, mouse, and human sera have very little effect on rat oligodendrocytes. Finally, in most previous reports sera were stored frozen prior to testing: in our experiments this significantly decreased cytotoxicity. A toxic effect of normal human serum on rat oligodendrocytes has been shown (Hirayama et al. 1986; Suzumura et al. 1986), and human serum also demyelinates rat optic nerves in vivo (Sergott et al. 1985). Normal rat serum injected into the ventricles of adult rats causes the release of low density membrane fragments from myelin sheaths (Konat et al. 1986). The mechanisms responsible for these effects remain obscure.

Our extension of earlier observations on complement activation by purified myelin showing the same property in intact living oligodendrocytes, which renders them susceptible to MAC attack, provides a mechanism which might account for the observed cytotoxic activity of animal sera on nervous tissue cultures. These findings may therefore link two apparently disparate areas of research into demyelinating disease pathogenesis.

There is no reason to believe this phenomenon to be peculiar to the rat, and our findings may therefore suggest a general, antibody independent role for complement in the generation of tissue damage in demyelinating diseases. Complement activation would not only produce MACs causing myelin/oligodendrocyte complex damage, but would also influence the inflammatory process both through the release of complement by-products such as C3a and C5a, and the liberation of leukotrienes from the oligo- dendrocyte membrane during complement injury (Shirazi et al. 1987). The importance of the blood-brain barrier (BBB) in controlling exposure of oligodendrocytes to serum complement is implicit in this hypothetical role for complement. BBB breakdown occurs in T cell mediated EAE in the Lewis rat (Goldmuntz et al. 1986); recent reports establish that clinically significant demyelination does occur in this disease (Heininger et al. 1988; Pender 1988) and the requirement for complement in the pathogenesis has been shown (Linington et al. 1989). Moreover, there is now considerable evidence indicating that BBB damage occurs early in the evolution of human and experimental demyelinating diseases and may be a primary initiating event (Daniel et al. 1983; Adams et al. 1985; Lightman et al. 1987; Tanaka et al. 1987; Tsukada et al. 1987).

ACKNOWLEDGEMENTS

We thank Dr. Mark Noble for advice on oligodendrocyte culture techniques, Drs. M.L. Cuzner and T.V. Waehneldt for the gifts of anti-GFAP antibody and anti-PLP antibody respectively, and S. Frith, S. Piddlesden and I. Lafaffian for their technical

299

assistance. This work was financed by the Multiple Sclerosis Society of Great Britain and Northern Ireland.

REFERENCES

Adams, C.W.M., R. Poston, S.J. Buk, et al. (1985) Inflammatory vasculitis in multiple sclerosis. J. Neurol. ScL, 69: 269-283.

Agrawal, H.G., B.I.C. Hartman, W.T. Shearer, etal. (1977) Purification and immunohistochemicai localisation of rat brain myelin proteolipid protein. I. Neurochem., 28: 495-508.

Biesecker, G. and H.J. Muller-Eberhard (1980) The ninth component of human complement: purification and physicochemieal eharaeterisation. J. lmmunol., 124: 1291-1296.

Bignami, A., L.F. Eng, D. Dahl and C.T. Uyeda (1972) Localisation of the glial fibrillary acidic protein in astrocytes by immunofluorescence. Brain Res., 43: 429-435.

Blaas, P., B. Berger, S. Weber, H. H. Peters and G. Hansch (1988) Paroxysmal nocturnal haemoglobinuria: enhanced stimulation of platelets by the terminal complement complex is related to the lack of C8 binding protein in the membrane. J. Immunol., 140: 3045-3051.

Bologa, L. (1985) Oligndendrocytes: key cells in myelination and target in demyelinating diseases. J. Neurosci. Res., 14: 1-20.

Chapman, J. and M.G. Rumsby (1982) A syringe modification for the simple and rapid dissociation of post-natai rat cerebral tissue for preparing primary cultures of mixed giial cells. Neurosci. Leg., 34: 307-313.

Compston, D.A.S., B.P. Morgan, D. Oleesky, et ai. (1986) Cerebrospinal fluid C9 in demyelinating disease. Neurology, 36: 1503-1506.

Compston, D.A.S., B.P. Morgan, A.K. Campbell, P. Wilkins, G. Cole, N.D. Thomas and B. Jasani (1989) Immunocytochemieal localisation of the terminal complement complex in multiple sclerosis. Neuropath. AppL Neurobiol., in press.

Cyong, J-C., S.S. Witkin, B. Reiger, et al. (1982) Antibody-independent complement activation by myelin via the classical complement pathway. J. Exp. Med., 155: 587-597.

Daniel, P.M., D.K.C. Lam and O.E. Pratt (1983) Relationship between the increase in the diffusional permeability of the blood-CNS barrier and other changes during the development of experimental allergic encephalomyelitis in the Lewis rat. J. Neurol. Sci., 60: 367-376.

Ebers, G.C. (1984) Oligoclonal banding in multiple sclerosis. Ann. N. Y. Acad. Sci., 436: 206-212. Goldmuntz, E. A., C. F. Brosnan and W.T. Norton (1986) Prazosin treatment suppresses increased vascular

permeability in both acute and passive transferred EAE in the Lewis rat. J. lmmunol., 137: 3444-3450. Hirayama, M., R.P. Lisak and D.H. Silberberg (1986). Serum mediated oligodendrocyte cytotoxicity in

multiple sclerosis patients and controls. Neurology, 36: 276-278. Heininger, K., W. Fierz, B. Schafer, et al. (1988) Electrophysiological investigations in adoptively trans-

ferred experimental allergic encephalomyelitis in the Lewis rat. Brain, in press. Johnson, G.D., R.S. Davidson, K.C. McNamee, et al. (1982) Fading of immunofluorescence during

microscopy: a study of the phenomenon and its remedy. J. Immunol. Methods, 55: 231-242. Konat, G., N.H. Diener and H. Offner (1986) Myelin changes in the rat CNS following intraventricular

injection of serum. Experientia, 42: 37-39. Lightman, S., W.I. McDonald, A.C. Bird, et al. (1987) Retinal venous sheathing in optic neuritis. Its

significance for the pathogenesis of multiple sclerosis. Brain, 110: 405-414. Linington, C., B.P. Morgan, N.J. Scolding, et al. (1989) The role of complement in the pathogenesis of

experimental allergic encephalitis. Brain, in press. Lubetzki, C., P. Lombrail, J. J. Hauw and B. Zalc (1986) Multiple sclerosis: rat and human oligodendrocytes

are not the target for cerebrospinal fluid immunoglobulins. Neurology, 36: 524-528. Norton, W.T. and S. E. Poduslo (1973) Myelination in rat brain: method of myelin isolation. J. Neurochem.,

21: 749-758. Parks, D.R., L.L. Lanier and L.A. Herzenberg (1986) Propidium iodide labelling of dead cells. In: Wier,

D.M. (Ed.) Handbook of Experimental Immunology, 4th ed., Vol. 1, Blackwell Scientific, Oxford, pp. 13-14.

Pender, M.P. (1988) The pathophysiology of myelin basic protein-induced experimental allergic encepha- lomyelitis in the Lewis rat. J. NeuroL Sci., 86: 277-289.

Raft, M.C., R.H. Miller and M. Noble (1983) A gliai cell precursor that develops in vitro into an astrocyte or an oligodendrocyte depending on culture medium. Nature, 303: 390-396.

300

Ranscht, G., P.A. Clapshaw, J. Price et al. (1982) Development of oligodendrocytes and Schwann ceils studied with a monoclonal antibody against galactocerebroside. Proc. Natl. Acad. Sci. USA, 79: 2709-2713.

Scolding, N.J., C. Linington, S.J. Frith, et al. (1988) Myelin-oligodendrocyte glycoprotein is a surface marker of oligodendrocyte maturation. J. Neuroimmunol., in press.

Seil, F.J. (1977) Tissue culture studies of demyelinating disease. Ann. Neurol., 2: 345-355. Sergott, R.C., M.J. Brown, R.M.-D. Polenta, et al. (1985) Optic nerve demyelination induced by human

serum: Patients with multiple sclerosis or optic neuritis and normal subjects. Neurology, 35: 1438-1442. Shirazi, Y., D.K. Imagawa and M.L. Shin (1987) Release of leukotriene B4 from sublethally injured

oligodendrocytes by terminal complement complexes. J. Neurochem., 48: 271-278. Sternberger, N.H., Y. Itoyama, M.W. Kies and H. Webster (1978) Myelin basic protein demonstrated

immunocytochemically in oligodendrocytes prior to myelin sheath formation. Proc. Natl. Acad. Sci. USA, 75: 2521-2524.

Suzumura, A., R.P. Lisak and D.H. Silberberg (1986) Serum cytotoxicity to oligodendrocytes in multiple sclerosis and controls: Assessment by Cr release assay. J. Neuroimmunol., 11: 137-147.

Tanaka, M. and J.C. Cyong (1985) The development of complement activating ability as an age related factor in murine brains. Microbiol. lmmunol., 29: 1219-1227.

Tanaka, Y., N. Tsukada, C.S. Koh and N. Yanagisawa (t987) Anti-endothelial cell antibodies and circu- lating immune complexes in the sera of patients with multiple sclerosis. J. Neuroimmunol., 17: 49-59.

Tourtellotte, W.W., S.M. S taugaitis, M.J. Walsh, et al. (1985) The basis of intra-blood brain barrier IgG synthesis. Ann. Neurol., 17: 21-27.

Tsukada, N., C. S. Koh, N. Yanagisawa, A. Okano, W.M.H. Behan and P.O. Behan (1987) A new model for multiple sclerosis: chronic experimental allergic encephalomyelitis induced by immunisation with cerebral endothelial cell membrane. Acta. Neuropathol., 73: 259-266.

Vanguri, P., C.L. Koski, B. Silverman and M.L. Shin (1982) Complement activation by isolated myelin: activation of the classical pathway in the absence of myelin-specific antibodies. Proc. Natl. Acad. Sci. USA, 79: 3290-3294.

Vogel, C.-W. and H.J. Mueller-Eberhard (1984) Cobra venom factor: improved method for purification and biochemical characterisation. J. lrnmunol. Methods, 73: 203-220.