Hydrolysis of myelin basic protein in human myelin by terminal complement complexes.

The participation of terminal complement complexes (TCC) in demyelination has been shown in rodent cerebellar cultures. Since TCC modulates activities of various membrane-associated enzymes and increases the level of cellular Ca2+ we investigated whether TCC could activate Ca2+-dependent neutral proteases in myelin that would lead to hydrolysis of myelin basic protein (BP). Addition of antibody and C7-deficient serum plus C7 to sealed myelin vesicles of two to six bilayers caused significant BP hydrolysis compared to the hydrolysis caused by antibody and C7-deficient serum. Significant hydrolysis occurred at the stage of C5b6,7 assembly, which increased in magnitude at the C5b6-8 stage. C5b6-9 formation did not enhance the effect of C5b6-8. BP hydrolysis by C5b6,7 did not require Ca2+ whereas the effect of C5b6-8/C5b6-9 was, in part, Ca2+-dependent. We postulated that TCC formation in myelin membranes causes activation of myelin-associated neutral proteases with subsequent hydrolysis of BP as a consequence of complement peptide insertion and channel formation. Such processes may alter the structure of myelin and augment the action of other inflammatory cells and their products in demyelinating diseases that could ultimately lead to the loss of myelin.

spiralling process of an oligodendrocyte while MAG is a transmembrane protein, thought to be limited to the enfolding myelin adjacent to the axon (4, 5). The biological importance of these proteins is believed to be their roles as structural stabilizers to maintain the compaction of adjacent lamellae of the myelin sheath, in the case of BP, and the myelinaxonal relationship in the case of MAG (6)(7)(8)(9). Recently, hydrolysis of BP and MAG in myelin by neutral proteases (NP) present in myelin or derived from inflammatory cells has been demonstrated (10,11). The NP instrinsic to myelin, initially demonstrated by Reikkinen et al. (12), are now known to consist of many enzymes that are either Caz+-dependent or Ca2+-independent (13)(14)(15).
In the present study we have examined whether activation of the late-acting complement proteins C5-C9 and formation of membrane-reactive terminal complement complexes (TCC) produce hydrolysis of BP possibly by activating the Ca2+-NP in myelin. We and others (16,17) have previously shown that when exposed to whole serum, central nerve myelin directly activates the first component of complement C1 in the absence of myelin-specific antibodies, and such activation proceeds through the entire activation cascade of the classical pathway to form TCC in myelin membranes (18). The sequential assembly of TCC from C5b to C8 or to C9 forms transmembrane channels in membranes (19)(20)(21) that can generate a transient increase in intracellular Ca2+ when a sublytic number of channels are formed on the cell surface (22,23). In addition, C5b-8 and/or C5b-9 channels can cause activation of membrane-associated enzymes such as phospholipases and lipid methyltransferases in the target cells (24)(25)(26)(27)(28). In view of the fact that complement channels modulate the activity of membrane enzymes and cause Ca2+ fluxes, the effects of TCC on BP was investigated with an in uitro model consisting of myelin vesicles resealed in the absence of Ca2+. The assay consisted of incubation of the resealed myelin vesicles (RMV) with purified terminal complement proteins in the presence or absence of externally added Ca2+ and analysis of the remaining intact BP by SDS-PAGE and densitometric measurement. With the use of this model, we obtained evidence that activation of C5-C9 and formation of membrane-reactive TCC on myelin caused hydrolysis of BP in the absence of exogenously added proteases. BP hydrolysis occurred at the C5b-7 stage, and increased to the maximum at the stage of C5b-8 assembly. BP hydrolysis by C5b-7 occurred in the absence of Ca2+ whereas the effect of C5b-8/ C5b-9 was, in part, Ca2+-dependent.
Complement Proteins, Antibodies, and Chemical Reagents-Human C2, C3, C5-C9, were obtained from Cordis. Human C5-C9 were also obtained from Cytotech. C5b6 was generated according to the method described (30) with modifications as follows: human serum, adjusted to pH 6.0, was precipitated with 20% (w/v) sodium sulfate. The precipitate was dissolved, dialyzed against 0.1 M phosphate buffer, pH 7.5, with 55 mM NaCl and applied to a DE52 column to remove C7. The C5 and C6 fractions obtained in a gradient of 55-200 mM NaCl were pooled, dialyzed against VBS2+, and incubated with Zymosan containing alternative pathway C5 convertase. The resulting C5b6 was further purified on a DE52 column and then on hydroxyapatite. Human serum deficient in C7 or C8 was a gift of Dr. H.
Gewurz. Anti-sheep IgM was purchased from Cappel and anti-myelin antibody was prepared by immunizing rabbits with purified rat central nerve myelin according to Ref. 31. Reagents used for SDS-PAGE were obtained from Bio-Rad. "Ca was purchased from DuPont-New England Nuclear and trypsin type 111-S, soybean trypsin inhibitor and trans-epoxysuccinyl-L-leucyl amido (4-guanidino) butane (E-64) were obtained from Sigma.
Preparation of Serum Complement-Normal and C7-deficient human sera were dialyzed against VBS containing 0.02 mM Ca2+ and 1 mM M f l for 2 h at 4 "C with two changes using Spectrapor semipermeable membrane tubing with a molecular weight cut off of 50,000 (Spectrum Medical). Normal serum was also dialyzed against 0.15 mM Ca2+.
Hemolytic Assays-Total hemolytic activity of whole serum. Sheep erythrocytes (E) were sensitized with anti-sheep E IgM (EA), then the EA were resuspended to 1.5 X 108/ml in VBS with 0.15 mM Ca2+ or 0.02 mM Ca2+. Normal sera dialyzed with buffers of the two different Ca2+ concentrations were incubated with the corresponding EA for 1 h at 37 "C. The released hemoglobin was then measured spectrophotometrically at 412 nm.
To assay the hemolytic activities of the late acting C proteins, 0.2 ml of varying dilutions of C5b6, C7, C8, or C9 were titered with 0.1 ml of EA carrying various C intermediates (1.5 X 108/ml) by adding 0.2 ml of the appropriate reagents. EAC1,4b,2a,3b intermediates were made by incubating EA with K76-treated serum as described (32). To form EAC4b,3b, the Clr,s and C2a were removed by incubating EAC1,4b,2a,3b for 2 h at 37 "C with EDTA-VBS. EAC1-7 were made by incubating EA with C8-deficient serum for 30 min at 30 'C. C5b6 was assayed on EAC4b,3b by adding normal serum diluted 40-fold with EDTA-VBS as a source of C7-C9.
Purification of Myelin and Preparation of Resealed Myelin Vesicles-Human spinal cord was obtained at autopsy within 10 h of death from patients who died of nonneurological causes. Central nerve myelin was isolated and purified according to Ref. 33. The purity of the myelin was assessed by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) (34), and the membrane protein was determined by the modified Lowry procedure (35). Membranes were aliquoted and stored in 10 mM EDTA in H,O at -70 "C.
For preparation of RMV, myelin was washed and suspended in sodium medium (36) at a concentration of 8 mg of protein/ml. The myelin was further diluted 10-fold with sodium medium and sealed into vesicles by incubating at 37 "C for 10 min, followed by one more wash in sodium medium. All centrifugations (9980 X g) were performed in an Eppendorf microcentrifuge for &lo min. In all experiments, between 2 and 4 mg wet weight of RMV corresponding to 80-160 pg of protein was used in each assay sample.
Qwntitation of B P in Myelin-The myelin vesicles were dissolved in 0.02 ml of 10% SDS and mixed with 0.08 ml of sample buffer (2% SDS, 5% 2-mercaptoethanol, 20% glycerol, and 62.5 mM Tris-HC1). In most of the experiments, prior to electrophoresis, purified goat IgG was added to each sample at a final concentration of 0.2 mg/ml. The samples were placed in boiling water for 5 min and applied to 12% polyacrylamide gels with 5% stacking gels. Protein separation was performed according to Laemmli (34) with a constant current of 25 mA through the stacking gel and 35 mA through the separating gel. The gels were stained with Coomassie Blue R-250 and scanned in a Hoeffer densitometer equipped with a single-pen Kipp and Zonen chart recorder. The protein bands of intact BP and IgG heavy chain (when IgG was added at the time of electrophoresis) in each lane were quantitated either by measuring the area ,under the peak or by weighing the appropriate portions of the chart paper.
Ultrastructural Morphology of Myelin Vesicles-The myelin vesicles were observed by transmission electron microscopy. Sealed vesicles were centrifuged, and the pellet was fixed in 100 mM cacodylate buffer, pH 7.4, containing 4% glutaraldehyde for 2 h. The samples were further fixed in 1% osmium tetroxide, dehydrated, and cell blocks made at different levels were embedded in Epon 812. Thin sections were cut from all cell blocks. The sections were stained with uranyl acetate and lead citrate, and pictures were taken in a Jeol 10 Cx microscope.
Comparison of Resealed Vesicles and Unsealed Myelin for Susceptibility of B P to Externally Added Trypsin-Resealed and unsealed myelin (4 mg wet weight) were incubated with varying doses of trypsin (13,500 BAEE units/mg of protein) in GVB for 30 min at 30 "C. The enzyme reaction was stopped by adding excess soybean trypsin inhibitor. The mixtures were further incubated for 10 min at 30 "C. Myelin vesicles were washed thoroughly, and the remaining BP was assessed by SDS-PAGE and densitometric scanning as described.
Spontaneous Hydrolysis of B P in RMV with Varying Concentrations of Ca2+-RMV were incubated with sodium medium containing 0-1.5 mM CaC1, for 1 h at 37 "C. The RMV were then centrifuged, washed, and the remaining BP was quantitated.
Scatchard Analysis of ea2+ Binding to RMV-Because external Ca2+ caused BP hydrolysis in RMV (Fig. 31, Ca2+ binding to RMV was analyzed according to Schlatz and Marinetti (37), and the binding affinities were calculated according to Scatchard (38). The conditions were adjusted according to preliminary experiments in which the amount of RMV and the binding time were varied. RMV (6 mg wet weight) in triplicates were incubated with 0-24 mM CaCl, in the presence of 0.32 pCi 45Ca (specific activity, 2.86 Ci/mmol) for 15 min at 37 "C in sodium medium. The vesicles were separated using 0.2pm Gelman membrane filters (Metricel GA-8) in a Millipore 1225 sampling manifold. The filters containing RMV were washed extensively with sodium medium, then the radioactivity on the filters and in the filtrates was measured. Experiments were also performed to test the effect of EGTA on the Ca2+ bound to RMV by the method described above except that one set of RMV were extensively washed with 10 mM EGTA/sodium medium instead of sodium medium.
Requirement of C7 in B P Hydrolysis Mediated by Serum Complement-RMV were sensitized with anti-myelin antibody in sodium medium/EDTA for 20 min at 30 "C and resuspended at a final concentration of 3 mg wet weight of RMV in 0.01 ml of sodium medium containing appropriate amounts of Ca". Antibody-sensitized RMV were then incubated in 0.2 ml of 20% C7-deficient serum with or without C7 in the presence of 0.02 mM Ca2+ for 1 h at 37 "C. The RMV were washed with ice-cold sodium medium and then subjected to SDS-PAGE analysis.
Inhibition of Complement-mediated BP-hydrolysis by E-64"RMV were prepared as described except that E-64, a thiol protease inhibitor (39), was included during resealing. A stock solution of E-64 was made in dimethyl sulfoxide (Me,SO), then diluted in sodium medium to give a final concentration of 100 pM E-64 and 0.05% (v/v) Me2S0. RMV made in the presence of E-64/Me2S0 or Me,SO alone were treated with antibody and C7-deficient serum with or without C7. The final concentration of 100 p M E-64 was maintained throughout the experiment.
The effect of E-64/MezS0 on complement activation was also tested by measuring hemolytic titers of serum complement on sheep EA in the presence and absence of 100 p~ E-64,0.05% Me2S0.
Hydrolysis of B P by C5b-9-RMV were incubated with 0.03 ml of C5b6 (30 pg) at 30 "C for 10 min. C7 (5 pg in 0.02 ml) was then added and incubated for another 15 min at 30 "C. EDTA or Caz+ was present in the incubation medium at a final concentration of 10 or 0.02 mM, respectively. The RMV-C5b6,7 intermediates were centrifuged, then incubated with 0.04 ml of sodium medium containing 5 pg of protein each of C8 and 9 for 5 min at 37 "C. At the end of the 5 min, 0.16 ml of sodium medium with 0.02 mM Ca2+ or 10 mM EDTA were added, and the mixtures were incubated for 55 min at 37 'C. The samples were mixed with 1 ml of cold sodium medium, centrifuged, and processed for SDS-PAGE analysis. As described above a known quantity of goat IgG was added as an internal standard to each assay tube before application on the gel. The IgG heavy chain and BP bands were scanned in each lane, and the ratio of these two bands was obtained. Controls included RMV incubated with sodium medium, 0.02 mM CaZ+ or with preincubated mixtures of C5b6, C7, C8, and C9 which resulted in the loss of their membrane reactivity, (C5b-9)i (40). The (C5b-9)i was also incubated in the presence of Caz+ or EDTA as was the case for the active C5b-9 complex. The effect of externally added Ca'+ on spontaneous hydrolysis of BP in RMV was assessed. The quantity of BP hydrolyzed in the absence of added Ca" was used as 0%. About 20% BP hydrolysis occurred in the presence of 1.5 mM Ca'+ in the medium, and approximately 5% hydrolysis occurred at 0.02 mM Ca' +. C5b6, C5b6,7, C5b6-8, or C566-9 in the Presence of 0.02 mM Ca'+-RMV were incubated with C5b6 or C5b6 followed by C7 as described above and then incubated for 1 h at 37 "C in sodium medium, 0.02 mM Ca'+. Portions of RMV-C5b6,7 were further incubated with C8, or with C8 plus C9 in sodium medium, 0.02 mM Ca'+ for 1 h at 37 "C. RMV treated with Ca2+ alone was used as 0%. C5b-7, previously incubated for 2 h at 37 "C, (C5b6,7);, was also tested. RMV-C5b6,7 were incubated with varying concentrations of C8 in the absence of C9 or in the presence of a constant amount of C9 (8 pg) for 1 h at 37 "C.

RESULTS
Evaluation of RMV as an in Vitro Myelin Model: Quantitation of BP by SDS-PAGE-Varying amounts of porcine BP or human myelin were analyzed by SDS-PAGE and densitometric scanning of the protein bands. As seen in Fig. la   C-treated RMV were washed and the remaining intact BP was quantitated by SDS-PAGE and densitometric analysis. BP hydrolysis by incubation in C7-deficient serum was used as 0%. RMV were made in the presence of 100 p~ E-64/MezS0 (0.05%) and treated as in experiment 1, except that E-64 and Me2S0 were also included in the reaction mixtures. negligible contamination with other organelles in the entire pellet of myelin vesicles.

Ca2+-dependent Spontaneous Hydrolysis of BP in RMV-
The BP hydrolysis in RMV was affected by the concentration of external Ca2+ (Fig. 3). Maximum BP hydrolysis at 1.5 mM Ca2+ was approximately 20%, whereas only 5% hydrolysis was observed at 0.02 mM Ca2+ in 1 h. Such a hydrolysis was most likely due to the inability of RMV to exclude Ca2+ inspite of the presence of Ca2+, Mg2"ATPases in the RMV due to the absence of added ATP (41).
Ca2+ Binding by RMV- Fig. 4A is the Scatchard plot of the binding of Ca2+ to RMV, showing at least two kinds of binding sites for Ca2+. The high affinity binding site could accommodate 12 nmol of Ca2+/mg of protein, and the association constant was calculated to be 9. 5 x lo3 M-'. The low affinity binding sites could bind 499 nmol of Ca2+/mg of protein, and the association constant was 1 X lo2 M-'. The results are representative of two separate experiments with triplicates in each experiment. As shown in Fig. 4B, the binding of Ca2+ to low affinity sites was susceptible to EGTA whereas the Ca2+ bound to high affinity sites on RMV was not displaced by EGTA.
Hydrolysis of BP in RMV by External Proteases-In order to evaluate whether RMV were tightly sealed and right-side out, we measured the degree of BP hydrolysis following addition of trypsin externally. The maximum hydrolysis obtained in RMV was approximately 12% whereas 60% hydrolysis was observed in unsealed myelin (Fig. 5). The maximum   Fig. 3). The hemolytic activities of serum complement were identical whether the serum Ca2+ was 0.02 or 0.15 mM (Fig. 6). Accordingly, the requirement of activated terminal-complement proteins for BP hydrolysis was determined using C7-deficient serum at 0.02 mM Ca2+. Experiment 1 in Table I shows the results of an experiment performed in triplicate. The amount of BP remaining after incubation of RMV with C7-deficient serum in 0.02 mM Ca2+ was used as the 100% value. Approximately 19% BP hydrolysis was observed that could be attributed to the activation of terminal-complement proteins. Experiment 2 in Table I also demonstrates the effect of E-64 on BP hydrolysis under identical conditions. As noted, the 17% BP hydrolysis obtained by C7 reconstitution was completely abolished in the presence of 100 HM E-64 while 0.05% M e 8 0 did not have any effect on the activation cascade of serum complement when tested in hemolytic assays (data not shown). These results indicate that E-64 inhibits endogenous myelin proteases that can be activated by TCC to produce BP hydrolysis. The kinetics of BP hydrolysis in RMV using C8-deficient serum with reconstitution showed that the majority of BP hydrolysis occurred within 10 min, which was followed by a slow decrease in hydrolysis (data not shown). In subsequent experiments 60 min was used as the total incubation period to obtain maximum BP hydrolysis.
Hydrolysis of B P in RMV by C5b-9"Since formation of C5b-9 complexes in membranes does not require Ca2+ (42) and terminal C proteins do not possess protease activity (43), the use of purified C5b6-C9 has an advantage over the use of serum complement. C5b6-C9 preincubated to form (C5b-9); was used as control. Fig. 7 shows representative SDS-PAGE in triplicate and densitometric scans of RMV treated with C5b-9 or buffer which shows the separation of BP and IgG heavy and light chain bands. Fig. 8, derived from two separate experiments, shows that formation of C5b-9 produced specific BP hydrolysis in the range of 30% in the presence of Ca2+ and 12% hydrolysis in the presence of EDTA.
Effect of C5b6-7, C5b6-8, and C5b-9 on B P Hydrolysis in RMV- Fig. 9 shows the effect of different TCC on BP hydrolysis in RMV. The specific effect of each intermediate was determined using BP hydrolysis occurring in the presence of Ca2+ alone as 0%. C5b6-7 complexes caused significant hydrolysis of BP. Addition of C8 to RMV-C5b6,7 resulted in further enhancement of hydrolysis. Interestingly, addition of C9 did not enhance the effect of C5b6-8. C5b6 had a minimal effect which varied among different experiments from 0 to 12%. Table I1 summarizes the results of four different experiments. Although the absolute values varied depending on the batch of RMV, the relative effects of the various intermediates were similar. The role of C9 in C5b-9-mediated BP hydrolysis in myelin was further evaluated with varying amounts of C8 on RMV-C5b6,7 plus excess C9. As seen in Fig. 10, C5b6-8 produced dose-dependent BP hydrolysis with increasing C8. Addition of C9 did not enhance BP hydrolysis even under the condition where the C9/C8 molar ratio was 8:l. Since C5b-8/ C5b-9-mediated hydrolysis of BP in RMV occurred both in the presence and absence of Ca2+ (Fig. 8), and C5b6-7 does not form membrane pores, the requirement of Ca2+ at the C5b6,7 stage was examined. The results shown in Table I11 indicate that BP hydrolysis by C5b6,7 does not require addition of external Ca2+.

DISCUSSION
The presence of proteolytic enzymes in central nervous tissue including myelin-associated NP and their effects on BP have been well documented (10,(12)(13)(14)(15). Myelin basic protein is the most extensively studied substrate (44), probably because BP is sensitive to most proteolytic enzymes and its abundance and location in myelin allow reliable quantitative analysis. In spite of the apparent functional association of the proteases and BP in myelin and the high susceptibility of BP to proteolysis, the reported catabolic rate of BP in vivo is small, with half-lives varying from 20 days to more than 100 days (45). This indicates tightly modulated homeostasis of enzyme activities in the brain. In this context it is significant to note that increased protease activity (3,44,46) and loss of myelin proteins, especially BP (2,3,47) and MAG (47) have been observed in the affected white matter of multiple sclerosis patients and rodents with experimental allergic encephalomyelitis. These biochemical changes are found concomitantly with structural alterations such as separation and vesiculation of myelin lamellae observed during early stages of inflammatory demyelination (1).
In view of our previous observations that activation of C5-C8 and/or C9 is a required process for antibody-dependent, complement-mediated demyelination in rodent cerebellar explants (311, and SC5b-9 complexes are present in the cerebrospinal fluid of patients with Guillain-Barre syndrome and multiple sclerosis (48), we have investigated whether C5b-9 attack on myelin can lead to BP hydrolysis, possibly by promoting enhanced activation of Ca2+-dependent-NP in myelin. The effect of TCC on hydrolysis of BP in multilamellar myelin vesicles was explored because C5b-8 and C5b-9 are known to produce a transient increase in cellular Ca2+ (22,23). For this purpose, we developed the RMV model as a myelin target and assayed BP hydrolysis quantitatively following C5b-9 attack. Although BP in RMV is relatively well protected from external proteases (Fig. 5), Ca2+ alone in the medium caused noticeable BP hydrolysis. We attribute this Caz+-dependent, spontaneous BP hydrolysis in RMV to the Ca2+-binding property of the myelin vesicles, which was demonstrated by Scatchard analysis (Fig. 4A). The binding studies of Ca2+ to rat brain membranes by Hemminki (49) also showed two types of Ca2+-binding sites for rat myelin. Gangliosides in myelin bind Ca2+ with high affinity, as demonstrated with the use of purified gangliosides (50,51) and have been postulated as Ca2+ receptors (52,53). It was further suggested that the ganglioside molecule might act as a Ca2+ carrier to deliver Ca2+ to enzymes intrinsic to the membrane bilayer (53). Our finding that Ca2+ that is bound to RMV with high affinity cannot be displaced by EGTA (Fig. 4B) suggested the translocation of bound Ca2+, at least through the outer lipid bilayer. This could explain how the Ca2+-NP, whose Ca2+binding sites are not exposed, might be activated by Ca2+ alone in the absence of membrane disruption. Such Ca2+dependent spontaneous BP hydrolysis would be expected to be minimal in uiuo because myelin contains functionally active, ATP-dependent, Ca", M$+-ATPases in the range of 2.2 pmol Pi/h/mg protein that can maintain Caz+ homeostasis (41). The effect of TCC on BP hydrolysis was shown by the reconstitution experiments with C7-deficient serum conducted at 0.02 mM Ca2+ (Table I). The BP remaining after treatment with the deficient serum in the presence of Ca2+ without reconstitution was used as the 100% value. The most conclusive evidence that C5b-9 can mediate hydrolysis of BP was obtained in experiments with purified C components. As shown in Fig. 8, BP hydrolysis occurred both in a Ca2+dependent and Ca*+-independent manner. The BP hydrolyzed by C5b-9 varied between 30 and 50% (Figs. [8][9][10]. This amount of BP loss is significant in view of the multilamellar nature of RMV and restriction of C5b-9 formation to the outermost myelin lamellae of RMV. When the effect of each step of TCC formation was evaluated, maximum hydrolysis was observed at the C5b-8 stage with no further increment when C9 was added (Figs. 9, 10, and Table 11). Interestingly, the significant hydrolysis observed with C5b6,7 (Fig. 9, Tables I1 and 111) was not enhanced by Ca2+ in contrast to the effect of C5b6-9 (Fig. 8). It is possible that membrane perturbation by inserted C5b-7 could activate Ca2+-independent NP or enhance the interaction between Ca2+-NP and Ca2+ bound to myelin because C5b-7 interacts hydrophobically with membrane bilayers (21,54) and stimulates the elimination of C5b6,7 from the surface of nucleated cells (55), and TCC causes reorganization of lipid bilayers (56). Of additional interest is that (C5b6,7), complexes formed by incubating C5b6 and C7 for as long as 2 h at 37 "C caused more BP hydrolysis than similarly inactivated C5b6-9, which can be attributed to the residual hemolytic activity of (C5b6,7), (data not shown). It appears that complete loss of hemolytic activity of C5b6,7 occurs most efficiently when all the terminal components and/or S protein are present during incubation. The experiments which showed the maximum hydrolysis with C5b-8 (Figs. 9, 10) are interesting, as we and others have observed that other TCC-mediated effects such as leukotriene B 4 release from rat oligodendrocytes (27) and human neutro-phils (26) occur at the C5b-8 stage without any additional effect by C9.
At present we can speculate that activation of C5-C9 on myelin could activate intrinsic proteolytic enzymes, causing BP hydrolysis. This speculation is based on experiments using thiol protease inhibitors such as E-64 (experiment 2 in Table  I), which has been shown to inhibit BP hydrolysis by endogenous Ca2+-NP in human myelin (13). Breakdown of BP and other myelin proteins including MAG, if extensive, could result in structural deterioration of myelin that would make the myelin vulnerable to further attack by inflammatory exudate surrounding the damaged myelin (57). Thus, the initial hydrolysis of BP in myelin, mediated by immune effectors such as C5b-9, may serve as a crucial biochemical step to initiate the autocatalytic process of demyelination.