Organization of Membrane Proteins in the Intact Myelin Sheath PYRIDOXAL PHOSPHATE AND SALICYLALDEHYDE AS PROBES OF MYELIN STRUCTURE*

Pyridoxal phosphate and salicylaldehyde were used as protein-labeling probes to study the organization of membrane proteins in the intact myelin sheath of the cat dorsal column. Both reagents react with protein amino groups to form Schiff’s bases which can be reduced with NaBSH,. The relatively membrane-impermeant pyridoxal phosphate labels all proteins of the intact myelin except basic protein. This major protein of myelin is labeled only after loss of membrane integrity. The relatively membrane-permeant probe, salicylaldehyde, was then used to establish that the basic protein is truly located on the cytoplasmic side of the myelin bilayer, and not merely sequestered within the multiple lamellar structure of the sheath. All proteins in the intact myelin are readily labeled by this reagent, with the label distribution pattern identical to that of disrupted myelin fragments. These data suggest a model for myelin structure in which the basic protein is the only major protein component located exclusively on the cytoplasmic side of the membrane (the major period zone of the sheath), with the other major proteins disposed wholly, or in part, in the extracellular half of the membrane bilayer (the intraperiod zone). All proteins, although asymmetrically disposed with respect to membrane sidedness, appear to be randomly distributed throughout the lamellae which comprise the sheath.

Myelin is a specialized plasma membrane with a relatively simple protein composition. It is a multilamellar structure that is spirally disposed around an axon, having both extracellular and cytoplasmic surfaces of the membrane in close apposition (1). The solution properties of the major protein components of myelin, the proteolipid protein and the basic encephalitogenic protein have been well characterized; as yet no enzymatic activity has been ascribed to either of them. It seems likely that their function in the membrane is to structurally maintain the unique morphology of myelin.
Our first objective in delineating the role of myelin proteins was to determine their location within the membrane by using membrane-impermeant surface probes. Poduslo and Braun (2) used lactoperoxidase to catalyze the iodination of accessible tyrosine residues of myelin proteins in the intact membrane. The extensive iodination of myelin basic protein only when the structural integrity of the membrane was mechanically disrupted indicated its location to be the cytoplasmic surface of myelin. In this communication we confirm and extend these observations.
Pyridoxal phosphate was first used as an impermeant membrane probe by Rifkin et al. (3). Subsequently, this reagent has been used as a surface probe of several membrane *This investigation was supported by Grant  from the Canadian Medical Research Council.
$ To whom correspondence should be addressed.
systems with varying degrees of success (4-8). Pyridoxal phosphate forms Schiff's bases with protein amino groups at physiological pH. The reduction of these complexes with sodium borotritide serves to introduce a tritium label of high specific radioactivity.
The membrane impermeancy of pyridoxal phosphate lies in its highly hydrophilic nature and its negative charge. Because its carbonyl group reacts principally with the c-amino group of lysines, this reagent provides information on the surface accessibility of an amino acid residue other than those residues (tyrosine and histidine) which are labeled by lactoperoxidase-catalyzed iodination. More importantly with regard to a study of the molecular architecture of myelin, pyridoxal phosphate is a much smaller molecule than lactoperoxidase, a protein of molecular weight 78,600. The extracellular spacing between adjacent lamellae of myelin, often referred to as the intraperiod zone, has been experimentally estimated to be 20 to 30 A in width (9). The diffusion of a molecule the size of lactoperoxidase is known to be very limited in such a 20 to 30 A space (10). The effect of this restricted diffusion of lactoperoxidase should be to iodinate tyrosine residues on mainly the very outer surfaces of myelin sheaths surrounding individual nerve fibers. Since pyridoxal phosphate is a much smaller molecule, its diffusion in the intraperiod zone should not be as severely restricted; thus, it should penetrate deeper into the myelin sheath where it can label sites not accessible to the larger enzyme. Usually 800 nmol of reagent, a 2.fold molar excess with respect to the amount of lysine in 1 mg of myelin protein, was used. The reaction was allowed to proceed for 10 min at room temperature before reducing agent was added. In the case of the experiments with intact myelin, the incubation mixture containing unreacted reagent was removed from the chamber and replaced with fresh buffer prior to reduction.
The disrupted membrane fragments and the purified myelin were reduced by the addition of the borohydride directly to the incubation mixture (1 ml). Reduction was achieved by adding 10 al of tritiated sodium borohydride, freshly prepared in 0.01 N NaOH, to the membrane preparations.
A S-fold excess, with respect to the aldehyde reagent, was found to be adequate. The reaction was complete in 3 min. The labeled dorsal column was immediately rinsed in fresh buffer and myelin was isolated (12). This step served to remove extraneous radioactivity.
' Preliminary experiments which demonstrated the labeling of myelin proteins by pyridoxal phosphate were conducted by Dr. Joseph Poduslo in our laboratory (unpublished results).
When the disrupted membrane preparation (the homogenate) was the starting material for labeling, the suspension was diluted with buffer immediately after the reducing step, and washed by centrifugation and resuspension.
Myelin was then isolated (12). Following the labeling of purified myelin fragments, the borotritidetreated membranes were washed five times in buffer and three times in water; the fragments were collected by centrifngation. A membrane preparation treated only with sodium borotritide served as a control for each experiment.
The reducing activity of the sodium borotritide was found to vary considerably from one commercial preparation to another. It was, therefore, routinely assayed spectrophotometrically.
The decrease in optical density at 388 nm, caused by reduction of pyridoxal phosphate, was used to determine the reducing activity.
When concentrations identical to those used to label the membrane preparations were employed, reduction of the carbonyl was complete in 2 min. The amount of nonvolatile tritium remaining in solution after acidification of a sodium borotritide preparation was also used as a measure of purity. The protein solution was prepared for gel electrophoresis by the addition of 0.1 volume of glycerol and 0.01 volume of fl-mercaptoethanol and 0.1% bromphenol blue.
Polyacrylamide Gel Electrophoresis-The discontinuous gel electrophoresis system for separation of myelin proteins described by Greenfield et al. (15) was used with the following modifications.
Gel size was 5 mm (inside diameter) by 11 cm. The acrylamide concentration in the stacking gel was 6% and the SDS concentration was 0.1% in both stacking and separating gels. Normally, 50 to 100 ag of myelin protein were applied to each gel tube. The gels were subjected to a constant voltage of 45 V for 14 hours by which time the tracking dye had migrated off the gel. The gels were removed from the tube by breaking the glass, fixed by several rinses in water/isopropanol/acetic acid (65/25/10), stained for 2 to 18 hours in 0.2% Coomassie blue in water/isopropanol/acetic acid (80/10/10), and destained in 10% acetic acid. Glycoproteins were visualized by the PAS staining method of Fairbanks et al. (16). Normally, 150 pg of protein were applied to gels to be used for PAS staining. The gels were sliced in a Gilson gel slicer into l-or 2.mm slices and solubilized by 0.2 ml of 30% H,O, before addition of scintillation mixture. Counting efficiency was 38 to 40%.

Quantitation of Myelin
Proteins-The quantitation of myelin proteins by spectrophotometric scanning of polyacrylamide gels at 280 nm, using published extinction coefficients for basic protein and proteolipid protein has already been described (2). The results, showing the protein composition of cat spinal cord myelin to be 31% proteolipid protein, 32% basic protein, and 37% remaining proteins, were used to calculate specific activities. It was found by dry weight analysis that 92 ag of cat spinal cord myelin protein are equivalent to 100 pg of bovine serum albumin by the Lowry protein determination. An appropriate correction factor was, therefore, applied to the calculations of specific activity.
The specific activities of the myelin proteins could not be averaged from one experiment to another because of variables such as the size of the dorsal column and the reducing activity of the sodium borotritide. Data pooled from several experiments are, therefore, shown as a percentage distribution of radioactivity for each protein group. This method of data presentation permitted us to obtain some information about the labeling of minor protein -ZThe abbreviations used are: SDS, sodium dodecyl sulfate; PAS, periodic acid-Schiff. components for which specific activities could not readily be determined.

Characteristics
of Labeling Reaction-Our first objective was to characterize the reaction of myelin proteins with pyridoxal phosphate. The possibility that the reactivities of myelin proteins depend on the concentration of pyridoxal phosphate or on the reaction time had to be examined before meaningful comparisons could be made between myelin with one side and both sides of the membrane exposed. One of the characteristic properties of myelin is that it does not form perfectly resealed vesicles when isolated. Consequently, in isolated myelin fragments proteins on either side of the membrane are able to react with pyridoxal phosphate. Therefore, we determined reaction conditions on isolated myelin fragments in order to establish the extent to which all myelin proteins could be pyridoxylated.
The distribution of pyridoxal phosphate amongst the protein groups should correspond approximately to the lysine distribution in these groups, as shown in Table I. Also shown is the distribution of the salicylaldehyde label amongst the myelin protein groups. This distribution is not identical to that of pyridoxal phosphate, which may reflect the hydrophobic character of salicylaldehyde and possibly a greater affinity for proteins which are in a more hydrophobic environment than the basic protein.
The actual amount of pyridoxal phosphate (13 nmol) which is covalently bound to proteins in myelin fragments after reduction with borotritide, in a typical experiment, corresponds to about 3% of the total lysines (440 nmol) present in 1 mg of total protein. Similarly, 1% of the lysines are modified by salicylaldehyde.
The extent of incorporation, measured in terms of specific radioactivities, can be increased when the ratio of pyridoxal phosphate to protein is increased (Fig. 1). However, the relative distribution of label amongst the three groups of proteins remains constant over the complete range of concentrations tested (see also Fig. 2, Bars I, 3, and 4). For most experiments 800 nmol of pyridoxal phosphate/mg of protein was an adequate ratio.
The membrane surface area in the dorsal column which is accessible to pyridoxal phosphate was estimated to be about 6% of that area which is accessible in isolated myelin fragments. This was done by comparing the specific radioactivities of the labeled proteins in each experiment (Fig. 6~). Thus, the possibility had to be considered that the effective ratio of pyridoxal phosphate to accessible sites in the intact myelin sheath was as much as 15 times greater than that in fragmented myelin. In order to demonstrate that comparisons between the two preparations are valid, we subjected myelin fragmenm to a 30-fold greater amount of pyridoxal phosphate. A comparison of Bars 4 and 5 in Fig. 2 illustrates that the distribution of the label among six categories of myelin proteins remains unaltered under these conditions. Bars 1 and 6 establish that a large increase in the borohydride concentration has no effect on label distribution.
A time course study of the labeling reaction with pyridoxal phosphate revealed that myelin proteins are maximally labeled in 10 min. A comparison of Bars 1 and 2 in Fig. 2 illustrates that the extent of labeling of myelin fragments with pyridoxal phosphate at 10 and 60 min is identical and that the distribution of label among the different categories of proteins is also identical. Preceding the basic protein band in the gel is a shoulder of unknown origin which also has some tritium associated with it. This component is not always seen, and it could be a lipid.SDS complex which migrates close to the gel front, or it could be one of the lower molecular weight forms of the proteolipid protein as suggested by Chan and Lees (20). When a comparison is made between intact myelin (Fig. 5) and either the isolated myelin fragments (Fig. 4) or the dorsal column homogenate (Fig.  3), the obvious and only major difference is the absence of label from the basic protein band of intact myelin. This is also readily apparent in Fig. 6a  protein, approaching 50% of that in the readily labeled proteolipid protein. However, preincubation of another dorsal column for 50 min in buffer, followed by the usual 10 min incubation with pyridoxal phosphate, produced results which were identical to the labeling experiment with pyridoxal phosphate present for the full 60 min. This strongly suggested that the labeling of basic protein during the longer incubations is mostly due to the deterioration of the myelin sheath and possibly only to a small extent to the permeancy of the reagent. This observation is strengthened by the appearance in the gels of low molecular weight peptides which arise from proteolysis of the myelin proteins, products which are not observed when the labeling experiment is restricted to 10 min.
Labeling of Intact Myelin and Myelin Fragments with Salicylaldehyde-The negligible labeling by pyridoxal phosphate of the basic protein in intact myelin can be explained in two ways: (a) the protein is located on the cytoplasmic surface of the membrane bilayer matrix, or (b) the protein is located only in the innermost lamellae of the multilamellar myelin sheath, regardless of its disposition in the bilayer. Pyridoxal phosphate can diffuse in the intact dorsal column not only around the bundles of myelinated axons but also between adjacent lamellae of each individual myelin sheath, i.e. in the intraperiod zone. Since short incubation periods (10 min) with intact myelin must be used to avoid the proteolysis that occurs over longer periods of time, the diffusion of pyridoxal phosphate in the intraperiod zone, between the lamellae of the myelin sheath, will necessarily be limited to the outermost layers. Thus, short labeling periods will result in preferential labeling of only the outermost layers of the intraperiod zone.
In order to discriminate between these two possibilites, we sought an analog of pyridoxal phosphate, of similar molecular dimensions, which would be membrane-permeant and would label the basic protein in intact myelin if this protein is on the cytoplasmic surface of the membrane. On the other hand, if the protein is located only in the innermost lamellae of the sheath, on the extracellular membrane surface, then it should be labeled no more extensively by the analog than by pyridoxal phosphate in short incubations because of the similar diffusion properties of the two reagents. Furthermore, if the myelin proteins are uniformly distributed throughout the lamellae (although asymmetrically distributed with respect to membrane sidedness), then labeling of intact myelin and isolated myelin fragments with a membrane permeant probe should produce identical labeling patterns.
Pyridoxal seemed, at first, to be an ideal choice for a membrane-permeant probe of similar size and specificity to pyridoxal phosphate. However, like pyridoxal phosphate, it is still a fairly hydrophilic reagent which is only slightly soluble in apolar solvents such as ether and chloroform/methanol. A much more suitable reagent is salicylaldehyde, which also forms Schiff's bases with protein amino groups. Although it is very soluble in ether and benzene and only sparingly soluble in water, it is still reactive with polar lysine residues as evidenced by its complete reaction with all the lysine residues in cytochrome c (11). Accordingly, the labeling pattern was