THE JOURNAL OF BIOLOGICAL CHEMISTRY 0 1989 by The American Society for Biochemistry and Molecular Biology, h e .

Vof. 264, No. 9, Ism of March 25, pp. 5121-5127,1989 Printed in U.S.A.

The Isolation, Characterization, and Lipid-aggregating Properties of a Citrulline Containing Myelin Basic Protein*

(Received for publication, October 14, 1988)

D. D. Wood and M. A. MoscareIloS From the Research Institute, The Hospital for Sick Children, Toronto, Ontario M5G 1x8, Canada

Human myelin basic protein was fractionated into its various charge isomers by CM52 cation exchange chromato~aphy. Approximately 25-30% of the total charge applied to the column appeared in the void volume. This material termed “C-8,” was further pu- rified by reversed phase high performance liquid chro- matography. Amino acid analyses of C-8 revealed low Arg (7 residue % in C-8 compared to 11-12 residue % in C-I) and increased GIx residues. The low Arg was accounted for by a corresponding amount of citrulline.

Sequence analysis after chemical fragmentation (cyanogen bromide and BNPS-skatole) and enzymatic (cathepsin D and carboxypeptidase S-1) digestion lo- calized the citruIline at residues 25,31,122,130,159, and 170 of the amino acid sequence. The effect of this loss of positive charge on the ability of the protein to aggregate lipid vesicles was demonstrated with vesi- cles composed of phosphatidylcholine (92.2 mol %) and phosphatidyl~rine (7.8 mol %). C-1 was the most ef- fective charge isomer, and C-8 was the least effective. The ability of these charge isomers to aggregate vesi- cles correiated with the net positive charge on each.

Vesicles composed of phosphatidylcholine alone were not aggregated by Iipophilin or any of the charge iso- mers. However, when lipophilin was incorporated into phosphatidylcholine vesicles (50% wlw), small, opti- cally clear suspensions of vesicles were formed. None of C-1, C-2, or C-3 aggregated these vesicles, but C-8 produced rapid vesicle aggregation. Since the substi- tution of citrulline for Arg would generate several relatively long apolar sequences, these would enhance the ability of C-8 to interact with the hydrophobic lipophilin molecule, promoting vesicle aggregation by hydrophobic interactions.

The mechanism by which citruliine is generated in myelin is not known, although enzymatic conversion has been described in other systems. Studies are un- derway to elucidate the mechanism by which this post- translational modification is generated.

The myelin sheath is a multibilayer structure which sur- rounds axons in the central nervous system and i s composed of lipids and proteins in a ratio of about 3:l (1). The protein composition is relatively simple since two proteins, the basic

* This work was supported by a Program Grant from the Medical Research Council of Canada. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “aduertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

f To whom correspondence and reprint requests should be ad- dressed.

protein (BP)’ and lipophilin (a proteolipid) together account for approximately 85% of the total membrane protein. The integrity of the myelin sheath depends in large part on the interactions of these two major proteins with the different lipids in the bilayer.

Lipophilin, which accounts for about 50% of the total myelin protein, has been shown to have an intramembranous orientation by x-ray diffraction (2), freeze-fracture electron microscopy (3), and fluorescent and chemical labelling (4-7). Several intramembranous domains have been postulated on the basis of hydropathy plots (8, 9). Recently, Kahan et al. (10, 11) provided both chemical and fluorescence evidence that lipophilin has at least three intramembranous domains. Lin and Lees (12), using an immunological probe, predicted a fourth membrane-embedded domain. Such hydrophobic in- teractions between lipophilin and membrane lipids contribute to membrane organization and stability and complement the interactions between BP and lipids.

BP, the most extensively studied of the myelin proteins, has been shown to associate primarily with acidic lipids (11, fluidize the bilayer and alter its permeability (13,14), induce the formation of a compact multilayer arrangement of bilayers (15, 16), and promote vesicle aggregation (17, 18). Although the interaction of the BP with lipids appears to be largely electrostatic, hydrophobic interactions are of considerable i m p o ~ n c e in the above mentioned processes (1).

The above mentioned studies highlight the importance of protein-lipid interactions in myelin, however protein-protein interactions must also contribute to the organization of the myelin membrane and compaction of individual bilayers to form a stable multilamellar structure. Several studies have shown that basic protein molecules self-associate (19-21) and may exist in the membrane as a dimer (22). These BP-BP interactions may be important in maintaining the multilayer structure of myelin.

Myelin basic protein exists as more than one molecular weight form (21,500, 20,000, 18,500, and 17,300). The M, = 18,500 form, the major isoform in the human, exhibits charge microheterogeneity “in vitro” as a result of post-translational modifications such as loss of C-terminal Arg, deamidation, phosphorylation, and oxidation of methionine to methionine sulfoxide (23-26). These charge isomers can be isolated by cation exchange chromatography using a salt gradient at pH 10.6. The variable amount of protein (20-30%), which does not bind but washes through the column before the gradient begins, was termed “C-8” (16).

In the present report, we describe the purification and ~~

The abbreviations used are: BP, basic protein; MBP, myelin basic protein; SDS, sodium dodecyl sulfate; PAGE, polyacrylamide gel electrophoresis; PC, phosphatidylcholine; PS, phosphatidylserine; HEPES, 4-(2-hy~oxyetbyl)-l-p~perazineethanesulfonic acid; HPLC, high performance liquid chromatography; PTH, phenylthiohydan- toin; BNPS, 2-(2 nitrophenylsuIfenyl)-3-methyI-3’-bromoindolenine.

5121

5122 Properties of C ~ t r ~ ~ l ~ n e C o n t ~ i n i ~ elin in Basic Protein

c h a r a ~ r i ~ t i o n of C-8. We present evidence that at least 6 arginine residues have been converted to the amino acid citrulline possibly as a result of post- translation^ modifica- tions (27), which impart unique lipid vesicle aggregating prop- erties to the protein. The ramifications of these properties for our understanding of myelin are discussed.

MATERIALS AND METHODS

P r e ~ r a t ~ n of Protein-Basic protein was isolated from normal human brain (age 46 years) white matter as described by Cheifetz et ai. (26). The microheteromers, components of BP, were prepared by the method of Chou et aZ. (24) with slight modifications (26). The BP was dissolved in a urea-glycine buffer, pH 9.6; and applied on to a CM52 cellulose cation exchange column equilibrated in a urea-glycine buffer, pH 10.6. The components were eluted from the column using a NaCl gradient (0-0.20 M) and then desalted on a Bio-Gel P-2 column in 0.01 N HCI. The desalted components were dialyzed, lyophilized, and stored at -20 "C.

Lipophilin was isolated from normal human myelin by the method of Gagnon et al. (28). After lyophilization, it was extracted with 5.0% acetic acid at 4 "C for 18 h. After cen t r i fugi~ a t 15,000 rpm for 15 min, the protein solution (residual BP) was dialyzed against water, lyophilized, and stored at -20 "C. The pellet (lipophilin) was washed twice with water, lyophilized, and stored at -20 "C.

Characterization of Component 8 (C-8)"Component 8 was further purified on a Cla reversed phase column (Waters Novapak or gBond- apak) using a trifluoroacetic acid (0.05%)-acetonitrile (0-60%) sol- vent system (Fig. 2). The eluted protein was subjected to SDS-PAGE (29) in order to determine its purity and an alkaline-urea disc gel to assess its charge (30). An aliquot was hydrolyzed in 5.7 N HCl at 110 "C for 24 h. Amino acid analyses were obtained on both the Waters PicoTag system for protein hydrolysates and the Beckman 7300 analyzer using the p h y s i o l o ~ c ~ system. Using the Beckman 890M sequenator, an N-terminal analysis was attempted. A C-ter- minal analysis was obtained using the enzyme carboxypeptidase S-1 (gift of Dr. T. Hoffmann, University of Toronto). Digestion was carried out in a 0.05 M pyridine-formate buffer, pH 4.0, for 18 h at 37 "C. The released amino acids were determined by amino acid analysis in the Waters PicoTag system.

In order to determine the sites of the citrulline residues in the polypeptide chain, a number of peptides were generated and then subjected to sequence analysis using a Beckman 890M sequenator. Cyanogen bromide (CNBr) digestion of C-8 was performed as follows: 2.5 mg of C-8 were dissolved in 500 p1 of 80% formic acid. To this was added 54 gl of 80% formic acid containing 0.6 mg of CNBr. Digestion was carried out at room temperature for 36 h in dark. Water (10 ml) was added to the solution, which was lyophilized, and this step was repeated three times. BNPS-skatole cleavage of C-8 was performed according to the method of Boggs et al. (31). Digestion of C-8 with cathepsin D (1:lOO w/w enzymefC-8) was carried out in a 50 mM ammonium acetate buffer, pH 3.5, for 18 h at 37 "C. The solution was lyophilized, and the residue was dissolved in water and lyophilized again. Before sequencing, the peptides were run on SDS- PAGE (12.5%) to determine the extent of digestion in each case. Endoproteinase Lys C digestion was done in 0.1 M ammonium bicar- bonate, pH 8.0, at an enzyme to C-8 ratio of 1:50 (w/w).

Preparation of Lipid Vesicles-Egg phosphatidylcholine (egg PC) and bovine brain phosphatidylserine (PS) were purchased from Avanti (Birmingham, AL) as dry powders. For the preparation of vesicles, they were dissolved in chloroform at a concentration of 5 mg/ml and stored at -20 "C under nitrogen. Aliquots of the dissolved lipid (92.2 mol % PC and 7.8 mol % PSI were transferred to acid- washed tubes, and the chloroform was evaporated under a stream of nitrogen and further dried by lyophilizing the sample for 30 min. The lipids were suspended in 1 ml of 10 mM HEPES buffer, pH 7.4, containing 100 mM NaC1, 1 mM EDTA, and 0.01% sodium azide and sonicated in a bath sonicator at 20-25 'C until clear. The volume was adjusted to 5 ml with buffer, and the vesicles were centrifuged at 40,000 rpm for 1 h to remove large multilayered vesicles. The phos- pholipid concentration of the supernatant was computed from the phosphorus content measured by the method of Bartlett (32).

P r ~ ~ r a ~ ~ o n of L i ~ p h i ~ i R - L i p ~ Vesicles-Egg PC in chloroform was taken to dryness under nitrogen and further dried by lyophiliza- tion for 30 min. The lipid was suspended in freshly distilled 2-

* P. R. Carnegie, personal communication.

chloroethanol. Lipophilin was wetted with 2-chloroethanol, and then water was added to make the final concentration 20% with respect to water. The protein solution was sonicated in a bath sonicator at 20- 25 "C to clarify (5-10 min). The lipid and protein were combined to

buffer, pH 7.4, containing 10 mM NaCl and 1 mM CaC12. After dialysis, give a ratio of 1:l by weight and then dialyzed against 2 mM HEPES

the volume was adjusted with buffer so that 200 gl contained 100 pg of lipophilin.

~ r e g a ~ i o n of PC Vesicles. Containing 7.8 Mol % PS-For each aggregation assay, 0.32 pmol of phospholipid was used. To this was added 0-100 pg of the protein in a final volume of 500 gl. After 10 min at room temperature, the turbidity of the solution was measured at 450 nm.

0-300 gg of BP components were added to vesicles containing 100 pg Aggregation of P C - L i p o ~ ~ ~ i r i Vesicles-For each aggregation assay,

of PC and 100 pg of lipophilin in a final volume of 500 J. The turbidity of the solutions were measured at 450 nm after 10 min at room temperature.

RESULTS

Separation of Charge Isomers of 3P on CM52 Columns BP was isolated from human brain white matter and frac-

tionated by CM52 cation exchange chromatography, as de- scribed under "Materials and Methods," in order to obtain each of the charge isomers (Fig. U). The most cationic (C- I), eluting last from the column, was preceded by C-2, and C- 2 was preceded by C-3, etc. The material, which eluted before the gradient was applied, was termed C-8 (fractions 5-15).

Characterization of Component 8 Further puri~catio~ of C-8 was achieved by HPLC on a

Waters Cla reversed phase column. A typical chromatogram

0 03 N

D 0

0.8- ..

5 03 N

0 Q

FRACTION NO. (3rnl/Fraction)

FIG. 1. CM62 cation exchange chromatography of human myelin basic protein. A , MBP (75 Am units) was dissolved in 5.0 ml of 0.08 M glycine buffer, pH 9.6, containing 6 M urea. After centrifuging at 15,000 rpm for 20 min, the clear supernatant was applied to a CM52 column (1.0 X 24 em) equilibrated with 0.08 M glycine buffer, pH 10.6, containing 2 M urea. The column was washed in this buffer until the C-8 fraction was obtained. Elution of the other components was carried out with 0.08 M glycine buffer, pH 10.6, containing 2 M urea and a sodium chloride gradient 0-0.20 M. E , MBP (25 Am units) isolated from lyophilized lipophilin was dissolved and applied to the column as described in A.

Properties of Citrulline Containing Myelin Basic Protein 5123

is shown in Fig. 2. The fractions under the peak eluting at 20 min were pooled and lyophilized. On SDS-PAGE (Fig. 3), this protein migrated slightly more slowly than C-1 (lanes 3 and 2, respectively). The amino acid analysis (PicoTag method) of this charge isomer of basic protein is shown in Table I. The results represent the means and standard deviation for six independent analyses. The composition is identical with the expected composition of BP except that Glx is high (3 residues per 100 more) and Arg is low (4 residues per 100 less). This amino acid composition is consistent with the less cationic nature of this protein. Tryptophan (Trp) was detected by intrinsic fluorescence. Since BNPS-skatole cleavage gener- ated only 2 fragments, 1 residue of Trp was assumed.

When C-8 was analyzed using a physiological system, we observed the presence of both citrulline and ornithine. Orni- thine is generated from the partial loss of citrulline during acid hydrolysis. The moles of citrulline and ornithine are shown in Table 11. Component 1 had no detectable citrulline or ornithine; C-2, C-3, and C-4 had less than 1 residue/100 residues of amino acids, whereas C-8 had 3.9 residues citrul- line and 1.5 residues of ornithine per 100 residues of amino

O . 1

i o 2b TIME OF ELUTION (rnin)

40

FIG. 2. Chromatography of C-8 on HPLC. C-8 was purified on a reversed phase C,S pBondapak column (Waters Associates) in 0.05% trifluoroacetic acid and a linear acetonitrile gradient (0-60%). The ordinate is A22enm.

18.5- kDo

FIG. 3. SDS-polyacrylamide gel electrophoresis of compo- nents isolated in Fig. 1A. Each sample was dissolved in 0.1% SDS, and 25 pg was applied to a 12.5% SDS gel and run by the method of Laemmli (29). Lane I , isolated MBP applied to CM52 column; lune 2, C-1; lane 3, C-8; lane 4, C-3. In addition to the 18.5-kDa major band, the 17-kDa MBP can be seen in C-3.

TABLE I Amino acid composition of component C-8 (residues/IOO)

For each amino acid analysis, 10 pg of protein were hydrolyzed in liquid phase with 5.7 N HCl at 110 "C for 24 h. After lyophilization, the amino acid mixture was derivatized and resolved on a Water's PicoTag system.

Amino Isolated from Basic protein Isolated from acid basic protein (18.5 kDa) lipophilin

Asx" 7.3 f 0.64 6.5 6.5 Thr 5.4 & 0.26 4.7 5.5 Ser 10.9 -C 0.5 11.2 11.4 G1xb 8.0 f 0.58 5.3' 7.2 Pro 8.0 -C 0.49 7.1 10.0

Ala 7.9 rf: 0.43 7.1 7.7 Val 2.2 f 0.18 2.4 1.9 Y2Cys Met 0.8 f 0.38 1.2 0.33 Ile 2.1 f 0.18 2.4 1.7 Leu 4.9 & 0.33 4.7 4.4 TYr 2.4 f 0.14 2.4 2.0 Phe 4.9 f 0.36 5.3 4.6 LYS 6.3 & 0.78 7.1 6.4 His 6.0 & 0.45 5.9 6.1 Arg 6.9 f 0.68 11.2 8.0 Trpd + 1.0

G ~ Y 16.4 f 0.61 15.3 16.3

a Asx = Asp + Asn. * Glx = Glu + Gln. Amino acids considered different from known composition. The

mean and standard deviations were derived from 6 amino acid anal- yses, of six independent samples.

TRP was detected by intrinsic fluorescence.

TABLE I1 Citrulline analyses on charge isomers from myelin basic protein

The various charge isomers were separated on CM-52 columns a t pH 10.6 as described under "Materials and Methods." After hydrolysis in 5.7 N HCI, a t 110 "C for 24 h, the hydrolysates were analyzed on the Beckman physiological system (Beckman 7300).

Component Citrulline Ornithine Total r e S i d u e S / 1 0 0

c-1 ND" ND c-2 0.13 0.13 0.26 c-3 0.15 0.20 0.35 c-4 0.28 0.40 0.68 C-8 3.90 1.5 5.40

a ND, not detected.

acids. Thus, the low number of arginyl residues/100 in C-8 were accounted for by the total of citrulline plus ornithine. Confirmation of the presence of citrulline in C-8 and identi- fication of the sites was obtained by automated sequence analysis of peptides generated by CNBr and BNPS-skatole cleavage and enzyme digestion.

Since the N terminus of C-8 was blocked, as demonstrated by failure to sequence in the automated Sequencer, chemical cleavages and enzymatic digests were done. Following CNBr cleavage, 3 peptides were generated, the N-terminal peptide, residues 1-21; the large peptide, residues 22-167; and a small C-terminal peptide, 168-170. The sequence of peptide 168- 170 demonstrated that the C terminus as Ma-Arg-citrulline instead of Ala-Arg-Arg. In addition, digestion with carboxy- peptidase S-l released citrulline confirming the C-terminal citrulline. The large peptide (residues 22-167) yielded 44 cycles. The sequence was identical with that of C-1 except citrulline was found to replace Arg at residues 25 and 31. The identification of citrulline was confirmed by preparing the PTH derivative of the pure amino acid. It eluted from the HPLC (15 min) just before the PTH derivative of glycine, whereas the PTH derivative of arginine eluted at 24 min (Fig.

5124 Properties of Citrulline Containing Myelin Basic Protein

4). When similar peptides were prepared from C-1, no PTH citrulline was detected.

BNPS-skatole cleavage at the single Trp residue yielded two peptides, a large N-terminal peptide, residues 1-116, with a blocked N terminus which did not sequence and a smaller C-terminal peptide, residues 117-170. Citrulline was identi- fied at Arg positions 122, 130, and 159 in the polypeptide chain. Further sequence data were obtained from a cathepsin D digest of C-8 (which digests BP at the Phe-Phe linkages 44-45 and 89-90). This permitted us to obtain overlapping sequences as well as additional sequences up to Arg (residue 97) and from Phe (residue 90) through the Trp region. No additional citrullines were found. Therefore, from our se- quence data, 5 citrullines have been found internally and 1 at the C-terminal position accounting for a total of 6 residues. The location of citrullines is shown in Fig. 5 which represents the known sequence of BP. Additional overlapping sequence data were obtained by Lys C digestion of cathepsin D peptide 1-44. Peptide 14-44 was sequenced (Table 111).

The Isolation of C-8 from Lipophilin

In the routine isolation of lipophilin, a small amount of BP (about 8% w/w) purified consistently with the lipophilin. After lyophilization, this BP was readily extracted by 5% acetic acid. On CM52 column chromatography, it failed to bind to the cation exchange column like C-8; none of the other charge isomers were present in this material (Fig. 1B). The identification of this protein as C-8 was carried out as follows. On SDS-PAGE, it migrated the same distance as C- 8 from BP and therefore had the same Mr. Further purifica-

0.2- A

t E, e In cu m c

0.2 -

C

0.2

G

J I I 10’ 153 ion i 5 ’ - Elution Time*

G

FIG. 4. HPLC profile of PTH-citrulline. A, pure citrulline was derivitized to the PTH amino acid and run on HPLC reversed phase C,, end-capped column. B, PTH-citrulline (C) eluting just before PTH-glycine (C) in cycle 14 of the BNPS-skatole cleavage of C-8. Glycine, which was identified in the previous two cycles (residues 128 and 129), is decreased over the previous cycle. C, PTH-citrulline (C) was added to the sample from the Sequencer demonstrating the superimposition of the added citrulline with the PTH-citrulline from the Sequencer. D, cycle 14 of BNPS-skatole cleaved C-1 demonstrat- ing that citrulline was not recovered in this cycle compared to the corresponding cycle of C-8 (see B above). Arg was recovered at 24 min.

BASIC PROTEIN SEQUENCE

1

1 3

25

3 7

4 9

6 1

7 3

8 5

9 7

109

121

133

145

157

169

FIG. 5. Amino acid sequence of human MBP. The Arg residues which were replaced by citrulline are circled.

tion on HPLC as described under “Materials and Methods” for C-8, followed by amino acid analysis (Table I), confirmed that it was similar to C-8. This BP has not been characterized further.

Vesicle-aggregating Properties of C-8 Compared to Other Charge Isomers of B P

Phosphatidylcholine-Phosphatidylserine Vesicles-Vesicles were prepared from phosphatidylcholine (PC) and phospha- tidylserine (PS) composed of 92.2 mol % PC and 7.8 mol % PS as described under “Materials and Methods.” The ability of C-8 to aggregate these vesicles was compared to that of other charge isomers C-1, C-2, and C-4. Increasing amounts of protein were added to the vesicle suspensions and aggre- gation was measured at 450 nm. The data are shown in Fig. 6. C-1, the most cationic charge isomer, was the most effective; C-2 was less effective than C-1; and C-4 (3 less positive charges than C-1) was less effective than C-2, while C-8, was the least effective. When compared to C-1, which was arbi- trarily assigned an aggregation value of loo%, the effective- ness of the other charge isomers decreased in linear fashion (Fig. 6, inset) correlating with the net charge of the protein. Phosphatidylcholine-Lipophilin Vesicles-Lipophilin in-

duces aggregation of negatively charged vesicles only (1). In the presence of PC, a neutral phospholipid, little if any aggregation was induced by lipophilin or C-1 from BP.

When lipophilin was incorporated into PC vesicles at a concentration of 50%, little aggregation was seen (A450 = 0.2). Addition of C-1, C-2, or C-3 to this system did not induce vesicle aggregation (Fig. 7). However, addition of C-8 induced vesicle aggregation, suggesting that of the charge isomers in BP, C-8 represents the one which interacts most readily with lipophilin. This interaction between C-8 and lipophilin in lipid vesicles has been observed with five independent prep- arations of C-8, each from a different human brain.

Properties of Citrulline Containing Myelin Basic Protein TABLE I11

The amino acid sequence of peptides generated by chemical fragmentation and enzymic digestion of C-8 C-8 from MBP was chemically fragmented with CNBr to generate 3 peptides consisting of peptides 1-21, 22-

167, and 168-170. Since the N terminus is acylated, sequence data were obtained on peptide 22-167 and 168-170. BNPS-skatole cleavage at the single Trp yielded 2 peptides 1-116 and 117-170. Cathepsin D digestion at only two Phe-Phe linkages yielded peptides 1-44, 45-89, 90-170,45-170. Peptide 1-44 was digested with endoproteinase Lys C, and peptide 14-44 was isolated (see "Materials and Methods"). The superscripts represent the actual residues sequenced.

5125

Fragmentation procedure Sequence"

0.4-

0.3-

0 v) 0" 0.2- 0

0.1 -

CNBr Peptide 22-167

Peptide 168-170

Peptide 117-170 BNPS-skatole

Cathepsin D

D * ~ H A @ ~ H G F L P @ H R D T G I L D S I G R F F G G D R G A P K R G S G K D S H H P A R T A H Y G S L P Q K S H 7 7 A168 R @70

G 1 1 7 A E G Q @ P G F G Y G G @ A S D Y K S A H K G F K G V D A Q G T L S K I F K L G G @ D S R S G S P M A R @ ' "

Peptide 45-170 F 4 ' G G D R G A P K R G S G K D S H H P A R T A H Y G S L P Q K S H G R T Q D E N P V V H F F K N I V T P R g 7

Peptide 90-170 F ~ K N I V T P R T P P P S Q G K G R G L S L S R F S W G A E G Q @ P G F G Y G G @ A S D Y K S A H K 1 3 9

Lys c Carboxypeptidase S-1 ,4168 R @I70

Peptide 14-44 Y " L A T A S T M D H A @ H G F L P R H R D T G I L D S I G R F 4

Amino acids are represented by single-letter code. @ indicates argininyl residue replaced by citrulline.

0 20 40 60 do IO0 PROTEIN (pg)

FIG. 6. Vesicle aggregation of phosphatidylcholine/phos- phatidylserine vesicles. PC/PS vesicles containing 92.2% PC and 7.8% PS were prepared as described under "Materials and Methods." To aliquots of the vesicles, increasing amounts of protein were added (0-100 pg). After 10 min at room temperature, theAaso was measured. Inset, the percent aggregation by each component at 100 pg of protein is plotted against the charge isomer.

DISCUSSION

Microheterogeneity of BP was first reported by Martenson and Gaitonde (33). A number of reports followed the original description in which the source of the microheterogeneity was determined in part at least (24). Thus, deamidation, phospho- rylation, and loss of a C-terminal Arg partially explained the microheterogeneity. Deamidation of Gln or Asn would impart one additional negative charge to the molecule and, although not established experimentally, C-2 could arise by this mech- anism. Phosphorylation at a single site would impart 2 nega- tive charges giving rise to C-3 (34). Since C-3 contains 0.2 - - + 0.4 mol of Po4 per mol of BP only, phosphorylation alone does not account for the charge on this isomer (26).

O.' 1 0 . o I

0 50 100 150 200 250 300 PROTEIN (pg)

FIG. 7. Aggregation of phosphatidylcholine vesicles con- taining lipophilin. PC vesicles were prepared with lipophilin incor- porated (50% w/w). To aliquots of this vesicle preparation, increasing amounts of protein were added from 0-300 pg.

Clearly, the source of charge microheterogeneity is complex, e.g. methionine sulfoxide has been identified by NMR studies and may contribute some negative charge (26). The search for other modifications to explain the large number of charge isomers appeared appropriate.

In recent years, we have suggested that the various charge isomers perform definite functional roles in the maintenance of the multilayered structure of myelin. We have found that when human white matter was fractionated after the removal of the compact myelin, several denser and less compact myelin fractions could be isolated (35). Isolation of BP from each of these fractions showed that the BP from denser myelin frac- tions was less cationic than the BP from the compact myelin fraction. On alkaline-urea gels, we were able to demonstrate

5126 Properties of Citrulline Containing Myelin Basic Protein

a decreased amount of C-1 and C-2 and an increased amount of the less cationic components (C-8) in the denser myelin fractions. Lipid x-ray diffraction of a model membrane system consisting of egg phosphatidylglycerol and individual micro- heteromers showed that C-1 was the most effective in inducing the farmation of crystalline multilayer arrangement; C-3 was only 50% as effective and C-8 was ineffective (16). These data suggested to us that each of the charge isomers has a specific role to play in the formation and maintenance of myelin structure, and that C-8 must have a unique role. Since this material was poorly understood, we undertook to purify and characterize it further.

Amino acid analyses in the protein hydrolysate system consistently yielded low Arg values (6-7 less Arg/mol of BP). By carrying out the amino acid analyses in the physiological system, we were able to account for all the lost Arg as citrulline, a known post-translational modi~cation of Arg (27). Since the conversion of Arg to citrulline would result in a loss of 1 positive charge per mol of Arg converted, this modification accounts for the much less cationic nature of C- 8.

Although citrulline has been reported in proteins since 1958, mainly those of epidermal origin (27), and in myelin (36), its function in these proteins is unknown. In myelin, two lines of evidence suggest that some of C-8 is specifically involved in protein-protein interactions with lipophilin. The two lines of evidence are found in (i) the vesicle aggregation studies and (ii) the co-purification of C-8 with lipophilin which was extracted only after lyophilization of the isolated lipophilin. Thus, when lipid vesicles were prepared containing lipophilin and phosphatidylcholine (50% w/w), small vesicles were formed (7) and the suspension was optically clear (A46o = 0.2). Addition of C-1, C-2, or C-3 had no effect on the A4m and therefore failed to induce vesicle aggregation. However, addition of C-8 induced rapid vesicle aggregation. Since none of the charge isomers aggregated vesicles of phosphatidyicho- line alone, the presence of lipophilin in the vesicles was a necessary prerequisite for vesicle aggregation by C-8.

Vesicle aggregation by BP has been shown to be a two-step process (18). In the first step, the BP binds to the vesicles followed by protein-protein interactions forming vesicle ag- gregates, Since excess BP did not inhibit vesicle aggregation (in contrast to polylysine-induced aggregation, in which ex- cess polylysine prevented aggregation of PGfPC vesicles), these authors concluded that multiple hydrophobic interac- tions between BP molecules was responsible (18). In the present study, hydrophobic interactions between lipophilin and C-8 appear to be a reasonable explanation of the vesicle aggregation data. The C-8 molecule which is considerably less charged than C-1 can form multiple hydrophobic interactions with lipophilin, cross-linking the vesicles in a highly cooper- ative manner (Fig. 7). Indeed, the replacement of Arg by citrulline would generate relatively long apolar sequences e.g. Gly (residue 120)-Ser (residue 132) which may present an excellent hydrophobic surface for interaction with lipophilin.

In a model of the multilamellar structure of myelin, a role for protein-protein interactions between proteolipid and BP was postulated by Rumsby (37,38) at the cytoplasmic surfaces which was supported by the isolation of proteolipid-BP dimers from cat spinal cord by using cross-linking reagents (5). NO attempt was made at that time to identify the BP in the cross- linked product with one of the charge isomers. It is tempting to speculate that the BP was C-8. A reinvestigation of cross- linked products in myelin, with cleavable cross-linking re- agents, is underway.

The mechanism through which citrulline arises in proteins

is not understood. An arginine-specific deiminidase, which converts arginine to citrulline has been reported (27). Since the BP isolated from human brain could be modified as a result of postmortem autolysis, we have recently isolated BP from rats of various ages, CM52 cation exchange chromatog- raphy as described under “Materials and Methods” estab- lished that none of the BP from neonatal rats bound to the cation exchanger, i.e. all of it appeared in the C-8 fraction, and all of it migrated as C-8 on a~kal~ne-urea gels at p H 10.6: These animal studies suggest that less cationic forms of BP may have functional roles during development. Their persist- ence into the adult brain may be related to specific functions for the various charge isomers in the final, functional myelin structure, e.g. since they interact less avidly with lipids than the most cationic species, they may represent the dominant form of BP in less compact myelin as suggested by earlier studies (35). In these studies, in which three myelin-contain- ing fractions were isolated from the residue remaining after isolation of compact myelin, the fraction of highest density had none of the most cationic components (C-1, C-2), but relatively more of the less cationic components (C-8), as determined on alkaline gel electrophoresis. At this time, ci- trulline analyses were not done. Re-examination of this ma- terial is currently underway.

REFERENCES

1. Boggs, J. M., Moscarello, M. A., and Papahadjopoulos, D. (1982) in Lipid-Protein Interactions [Jost, P. C., and Griffith, 0. €I., eds) Vol2, pp. 1-51, John Wiley & Sons, New York

2. Brady, G. W., Birnbaum, P. S., Moscarello, M. A., and Papahad- jopoulos, D. (1979) Biophys. J. 26, 23-42

3. Vail, W. J., Papahadjopoulos, D., and Moscarello, M. A. (1974) Biochim. Biophys. Acta 345,463-467

4. Feinstein, M. B., and Felsenfeld, H. (1975) Biochemistry 14,

5. Golds, E. E., and Braun, P. E. (1976) J. Biol. Chem. 261,4729-

6. Wood, D. D., Epand, R. M., and Moscarello, M. A. (1977)

7. Wood, D. D., Boggs, J. M., and Moscarello, M. A. (1980) Neuro-

8. Stoffel, W., Giersiefen, H., Hillen, H., Schroeder, W., and Tung-

9. Laursen, R. A., Samiullah, M., and Lees, M. B. (1984) Proc. Nutl.

10. Kahan, I., and Moscarello, M. A. (1985) Biochemistry 24, 538-

11. Kahan, I., Epand, R. M., and Moscarello, M. A. (1986) Biochem-

12. Lin, L.-F. H., and Lees, M. B. (1985) J. BWL C k m . 260,4371-

13. Papahadjopo~os, D., Moscarello, M. A., Eylar, E. H., and Isac,

14. Cheifetz, S.. Bortgs. J. M.. and Moscarello, M. A. (1985) Biochem-

3049-3056

4735

Biochim. Biophys. Acta 467,120-129

chem. Res. 6,745-755

gal, B. (1985) Biol. C k m . Hoppe-Seyler 366,627-635

Acud. Sci. U. S. A. 81,2912-2916

544

istry 26,562-566

4377

T. (1975) Biochim. Biophys. Acta 401,317335

istry 24,5170-5175

Moscarello. M. A. (1981) Biophys. J. 34,345-350 15. Brady, G. W., Murthy, N. S., Fein, D. B., Wood, D. D., and

16. Brady, G. W:, Fein, D. B., Wood, D. D., and Moscarello, M. A. (1985) Biochem. Biophys. Res. Commun. 126,1161-1165

17. Young, E’. R., Vacante, D. A., and Snyder, W. R. (1982) J. Am. Chem. soe. 104,7287-7291

18. Lampe, P. D., Wei, G. J., and Nelsestuen, G. L. (1983) Biochem-

19. Smith, R. (1980) Biochemistry 19, 1826-1831 20. Smith, R. (1982) Biochemistry 21,2697-2701 21. Moskaitis, J. E., Shriver, L. C., and Campagnoni, A. T. (1987)

22. Smith, R. (1977) Biochim. Biophys. Acta 470, 170-184 23. Deibler, G. E., Martenson, R. E., Kramer, A. J., Kies, M. W., and

Miyamoto, E. (1975) J. Biol. Chem. 260,7931-7938 24. Chou, F. C.-H., Chou, C.-N. J., Shapira, R., and Kibler, R. F.

istry 22,1594-1599

N e u r ~ k m . Res. 12,409-417

J. Ramwani and M. A. Moscarello, unpublished data.

25.

26.

27.

28.

29. 30.

31.

Properties of Citrulline Containing Myelin Basic Protein 5127 (1976) J. Biol. Chem. 261,2671-2676

(1977) J. Neurochem. 28,1051-1059

Biochem. Biophys. 233,151-160

Chou, F. C.-H., Chou, C.-H. J., Shapira, R., and Kibler, R. F.

Cheifetz, S., Moscarello, M. A., and Deber, C. M. (1984) Arch.

Rothnagel, J. A., and Rogers, G. E. (1984) Methods Enzymol.

Gagnon, J., Finch, P. R., Wood, D. D., and Moscarello, M. A.

Laemmli, U. K. (1970) Nature 227,680-685 Deibler, G. E., Martenson, R. E., and Kies, M. W. (1972) Prep.

Boggs, J. M., Wood, D. D., and Moscarello, M. A. (1981) Bio-

107,624-631

(1971) Biochemistry 10,4756-4763

Biochem. 2,139-165

chemistry 20, 1065-1073

32. 33.

34.

35.

36.

37. 38.

Bartlett, G. R. (1959) J. Biol. Chem. 234, 466-468 Martenson, R. E., and Gaitonde, M. K. (1969) J. Neurochem. 16,

Cheifetz, S., and Moscarello, M. A. (1985) Biochemistry 24,1909-

Cruz, T. F., and Moscarello, M. A. (1985) J. Neurochem. 44,

Finch, P. R., Wood, D. D., and Moscarello, M. A. (1971) FEES

Rumsby, M. G. (1978) Biochem. SOC. Trans. 6,448-462 Rumsby, M. G. (1987) A Multidisciplinary Approach to Myelin

Diseases (Crescenzi, G. S., ed) pp. 111-132, Plenum Publishing Corp., New York

333-347

1914

1411-1418

Lett. 15,145-148