Some Spaces and Barriers in Postmortem Multiple Sclerosis

493

Some Spaces and Barriers in Postmortem Multiple Sclerosis WALLACE W . TOURTELLOTTE

AND

J U L I U S A. P A R K E R

Departmetit of Neurolog.v, University of Michigan Medical School,

Atiri

Arbor, Michigan ( U S A )

This report will deal with the chemical anatomy of human postmortem “normal” (control) white matter, as well as postmortem normal-appearing white matter and areas of demyelination (plaque tissue) in a primary demyelinating disease (multiple sclerosis). A n attempt will be made to correlate the concentration of selected chemical constituents within an anatomical space in the normal-appearing white matter and in the plaque tissue. Furthermore, some tests to evaluate the blood-brain barrier in postmortem multiple sclerosis plaque tissue will be presented. METHODS

Patient material The left cerebral hemisphere was obtained at autopsy (12 to 36 h postmortem) from 12 patients who died of multiple sclerosis; and 11 patients whose deaths were from other causes (that is, they did not have multiple sclerosis or any other gross or microscopic structural brain damage). The unfixed brain tissue was cut into 3-mm coronal sections at room temperature by an electric slicer. Each section was placed in a plastic bag, labeled, sealed, and frozen; the storage temperature was -90°C. On a thawed section, the white matter from the control brains was dissected (50 to 500 mg samples) ; in addition, similar-sized samples of normal-appearing white matter from multiple sclerosis brains were grossly dissected free of demyelinated plaque tissue, and vice versa. Each dissection was accompanied by a punch biopsy, which was used for histological analysis to confirm the presence or absence of myelin or for estimation of the axon density. Then, the dissected bits, from each multiple sclerosis patient, of normal-appearing white matter were pooled, if insufficient material was available, and the plaque tissue was pooled, if it was necessary. The dissections were weighed and homogenized in enough 0.15 M sodium chloride or water to make 20 percent homogenates or in 20 volumes of 2 : I chloroform-methanol. A portion was centrifuged (25,000 rev./min), and samples of the appropriate supernatant fiuid were taken for the various assays. Hefrrences p. 520-522

494

W. W. T O U R T E L L O T T E

Histological methods

A 16-gauge needle was used to obtain tissue for histological analyis from the centers of the dissected samples. The specimens were fixed in 10% formaldehyde solution (formalin) and labeled by a code system unknown to the neuropathologist. The samples were then embedded in paraffin, sectioned at 8 p , and stained with the pyridinesilver nitrate method of Cajal (Addison, 1950) for axons and with the KliiverBarrera method (1953) for myelin. All the biopsies taken from the dissected plaques showed severe to complete demyelination, whereas those taken from the control white matter or normal-appearing multiple sclerosis white matter samples showed normal myelination. The number of axons per unit volume of tissue was assessed in four areas of a single microscopic section from each sample with the aid of an ocular grid. The grid consisted of squares formed by eight horizontal and eight vertical lines (1 mm apart) and was used at a magnificationof 970. All axons through the entire depth ofsection falling on one or more horizontal lines were counted. A single long axon recrossing the same line or another line was counted as many times as it crossed these lines. This technique eliminates the problem of identifying the individual axons in a field and yields a number (“axonal density index”) related to the number of axons per unit volume. Water and total lipid determination

An aliquot of the homogenate was dried in a 110°C oven. The weight of the tissue in the suspension minus the weight of the dried sample yielded the data for the water content. Two methods were used for obtaining the total lipid content. The dried samples were extracted two times each with chloroform and hexane. The lipid-free sample was redried. The dry weight minus the fat-free dry weight was used to calculate the total lipids. The alternate total lipid procedure was done on a weighed tissue sample homogenized with 20 volumes of chloroform-methanol 2 : 1 . An aliquot was filtered, and the filtrate was washed three times with aqueous potassium chloride and potassium chloride-containing “Folch upper layer” as described previously (Kishimot0 and Radin, 1950). The lower layer, which contains the total lipids, was evaporated to a small volume with a stream of nitrogen, then lyophilized with benzene and weighed. This weight is a little low because the gangliosides, and a small amount of other lipids, enter the upper layers. However, there was good agreement between the two total lipid methods used - the average per cent difference was 0.9 with a range of 0.7 to 1.1 on five replications. Hemoglobin The carboxy-hemoglobin method of Gordon and Nurnberger (1956) was used on the supernatant fluids of homogenates made in water or 0.15 M sodium chloride. The two types of homogenates gave the same results. The extent of contribution of residual blood immunoglobulin-G or albumin in the brain was calculated as follows: The con-

495

POSTMORTEM M U L T I P L E S C L E R O S I S W H I T E M A T T E R

centration of hemoglobin (Hb) per kilogram (kg) of fresh tissue was divided by the concentration per ml of the patient’s blood, and then multiplied by (1-hematocrit/lOO). This value is the ml of serum per kg of brain. The g of immunoglobulin-G or albumin per kg of brain contributed by the blood in the brain was determined by multiplying the nil of serum per kg of brain by the serum concentration of albumin or immunoglobulin-G. This value when subtracted from the immunoglobulin-G or albumin for total brain tissue value yielded the concentration in the brain itself (brain immunoglobulin-G, or albumin corrected for residual blood in the brain). It is possible that a method based on the analysis of hemoglobin such as thecarboxyhemoglobin method of Gordon and Nurnberger is not an appropriate method for determination of total hemoglobin in all postmortem brain tissue. Post-mortem brain tissue, which is acidic (pH 6-6.5), enhances the formation of methemoglobin, which is not detected by the carboxyhemoglobin method. Hence, we have tried the cyanmethemoglobin procedure of Evelyn and Malloy (1938). This method detects both methemoglobin and hemoglobin. We have applied both methods to the supernatant fluid of both control and multiple sclerosis brain tissue. In approximately ten percent of the cases there existed a significant difference. An example of a statistically significant difference(p 5 0.05)of ml of blood per kg of brain (mean standard l error of the mean) was: control white matter (Stant.), carboxyhemoglobin method 3.0 0.32 and cyanmethemoglobin 8.0 f 0.58; however, normal-appearing multiple sclerosis white matter (Wacht) contained 6.0 & 0.31 and 12.0 0.61, and the plaque (Wacht) contained 6.0 f 0.31 and 14.0 0.72, for the carboxyhemoglobin and cyanmethemoglobin method, respectively. When a significant difference was found utilizing the two methods, the bloodcorrected brain values were based on the cyanmethemoglobin method.

*

*

*

Gangliosides Weighed tissue samples were homogenized with 20 volumes of chloroform-methanol 2 : I . An aliquot of the suspension was filtered, and the filtrate was washed three times with aqueous potassium chloride and potassium chloride-containing “Folch upper layer”. The aqueous layers were combined, and removal of contaminating lipids was accomplished by a back-wash with “Folch lower phase”. This was followed by methanolysis of the dried residue of the washed upper layer, and gas chromatographic determination of the ganglioside fatty acids. A mixture of methyl esters of 19 : 0 and 21 : 0 was used as an internal standard (Kishimoto and Radin, 1966). It should be noted that the final data represent moles of ganglioside, rather than of sialic acid. Hydroxyproline The white matter from two control brains (LaJ and Campb) and the normal appearing white matter and plaques from five multiple sclerosis brains (Hi, Bo, Re, Wi and Wa) were used for the analysis of hydroxyproline. Twelve homogenates (enough 0.15 M sodium chloride to make 20 percent homogenates) were prepared. Singlicate analyses were carried out on 250 ,ul of the homogenate by the method of Prockop and Rtfcrviice.> p.

520-.’22

496

W. W. TOURTELLOTTE

Udenfriend (1960) in the laboratory of Dr. A. Sjoerdsma, Chief, Experimental Therapeutics Branch, National Heart Institute, Bethesda, Maryland. Since blood vessel walls contain 9 mg of hydroxyproline per kg (Sjoerdsma, personal communication) we have attempted to correct the brain tissue value for the presence of blood vessel walls. If one-tenth of the volume of blood in the brain per kg is blood vessel walls, then control white matter on the average should contain 45 mg hydroxyproline per kg because of blood vessels (4.0 ml of blood per kg (see Table 11) multiplied by 0.1 and divided by 9 mg per kg), and normal-appearing multiple sclerosis white matter and plaques, 54. The correction for hydroxyproline in plasma was not significant, since plasma contains 10 y/ml. Albumin and immunoglobulin-G The immunochemical procedure of Kabat et al. (1948) was used, except that a biuret reaction was carried out to determine the protein concentration of the immunoprecipate instead of a micro-Kjeldahl procedure. Furthermore, we have scaled down the procedure so that the final color reaction is read in a final volume of 60 pl. Ten pl of antibody reagent (5 mg antibody protein/ml) was added to a 50 pl aliquot of the supernatant fluid obtained after centrifugation of a 20 per cent 0. I5 M sodium chloride homogenate of brain tissue. The immunoprecipitate that results after 18 h of incubation (I h at 36°C and 16 h at 4°C) was washed two times with 100 pl of cold 0.15 M sodium chloride. The precipitate was dissolved in 60 pl of a biuret reagent and read at 540 m p in a DU Beckman spectrophotometer adapted to read 35 pl. The antibody reagents were prepared in rabbits immunized with purified human albumin or immunoglobulin-(; suspended in Freund’s adjuvant. Immunoelectrophoresisof the human serum against the rabbit immune serum was used tocheck the purity of the antibody reagent. Since the procedure of Kabat et al. (I 948) has been standardized for body fluids and not for extracts from solid tissues, it was necessary to test its applicability to 0.15 M sodium chloride brain extracts. Standard solutions of immunoglobulin-G were recovered quantitatively (90 to 110 percent) when added to homogenates of control white matter, as well as to those from normal-appearing white matter and plaque tissue from brains of patients that had died of multiple sclerosis. Bromide Seven multiple sclerosis patients were given 0.64 g of sodium bromide (orally) three times per week for one to three years prior to death to maintain a serum bromide ion concentration that was non-toxic and did not produce pharmacologic effects (2.0 to 5.4 mequiv./l). The postmortem brain tissue was cut into 3-mm coronal sections and stored frozen, as described above. The dissected specimens of normal-appearing white matter and plaque tissue were homogenized in enough water to make a 20 percent suspension. An aliquot of the homogenate was used for dry and fat-free dry weight. Another aliquot (20 to 150 mg) was taken for bromide ion analysis. The organic matter was destroyed by fusion with sodium hydroxide and potassium nitrate as recommended by Brodie and Friedman (1938). The remaining ash was dissolved in

POSTMORTEM M U L T I P L E S C L E R O S I S W H I T E MATTER

497

water, and the bromide was oxidized to bromate, followed by iodometric titration. A GilmontR microbiuret calibrated to deliver 0.2 p1 was used to titrate 2 1.0 pg of bromide; the coefficient of variation was 11.7 percent. When 15 pg of bromide were titrated the coefficient of variation fell to 4.4 percent (Haerer et al., 1964). It was also shown that brain tissue from patients who died with multiple sclerosis, but who had not received sodium bromide before death, did not contain interferring substances. Sodium, potassium, and chloride Two g of postmortem white matter were obtained from a control brain (Campb) and also from a multiple sclerosis brain (Wacht), which was separated into normal appearing white matter and plaque tissue. After obtaining the exact weight of the tissue, a 20 percent homogenate in water was made and 0.25 ml was taken for dry and fat free dry weight. Duplicate aliquots of 1.0 ml were ashed at 400°C overnight, and the ash was dissolved in 0.5 ml of 1 N nitric acid. The concentration of sodium and potassium was determined by internal flame photometry. For chloride ion determination it was necessary to neutralize the phosphoric acid generated during ashing, so 0.1 ml of 1.7 M sodium carbonate was added to 2.0 ml of homogenate before the 1.O ml duplicate aliquots were taken for ashing. The ash was dissolved in 0.5 ml of 1 N nitric acid, and the chloride ion was determined in the Cotlove (1958) titrator. (The above analyses were carried out by Dr. J. P. Chandler, Chief of the Core Laboratory, Clinical Research Unit, University of Michigan.) TABLE I

~

SOME H Y P O T H E T I C A L W H I T E M A T T E R S P A C E S

__

Aiiatotriical spaces

Normal Vascular True extracellular Glia (oligodendroglia and astrocytes) Myelin Axoplasm Multiple sclerosis (add following): Expanded true extracellular space? Reactive glia (microglia, lipomacrophages, fibrous astrycotes, myelin forming cells) Perivascular inflammatory cuffs (mostly lymphocytes)

Some possible chemical correlates

Hemoglobin (specific) Electrolytes (chloride ion space) Bromide ion chloride ion space Albumin and immunoglobulin-G do. Total lipids (2&25 ”” not myelin lipids) (cerebrosides may be better indicator) electrolytes Gangliosides (specific) Electrolytes (potassium ion space)

-

-

Electrolytes (chloride ion space) chloride ion space Bromide ion Albumin and immunoglobulin-G do. + Hydroxyproline (“specific” for fibrous astrocytes?) + Immunoglobulin-G synthesis Immunoglobulin-G synthesis

498

W. W. TOURTELLOTTE RESULTS A N D DISCUSSION

Some possible white matter spaces

In tabular form the normal anatomical white matter spaces, and also other spaces possibly introduced by multiple sclerosis, are listed in Table I. The normal anatomical spaces in white matter have been well delineated by light and electron microscopy. The non-porous endothelium of the blood vessels is surrounded by a basement membrane to which glial feet are attached. The glial feet probably encase no more than 80% of the capillary wall (Brightman, 1965). Between plasmalemmas of adjacent glial cells there exists an electron microscopic lucent pericellular space (true extracellular space about 100-200 A wide; Katzman, 1966). However, this space is irregular, for there exist intermembrane fusions between glial feet around blood vessels and many enlarged confluences measuring as much as 400 to 800 A across (Brightman, 1965). This true extracellular space has been estimated to consist of approximately 12 percent of the volume of the white matter (Oldendorf and Davson, 1967). Inside the glial cell exists the myelin sheath. There are also periodic nodes of Ranvier permitting an exposure of the axon to the true extracellular space. On the other hand, the situation is more complex when we examine the light and electron micrographs of the normal-appearing white matter and plaque tissue of multiple slcerosis. Normal-appearing white matter near a plaque shows normal myelin sheaths, an occasional microglial cell with phagocytic activity, and a mildly expanded extracellular space (PCrier and GrCgoire, 1965). Blood vessels in the area of plaques of demyelination have perivascular inflammatory cuffs (mostly lymphocytes and lipophages). Furthermore, at the plaque edge new cells appear, such as microglia, lipophages, reactive astrocytes, and perhaps myelin-forming cells. In the plaque, fibrous astrocytes become prominent and axons without a myelin sheath are present (Adams and Kubik, 1952; PCrier and GrCgoire, 1965). The latter have demonstrated with electron micrographs, in cases studied one to three hour post mortem, an expanded true extracellular space and a thin layer of glial cytoplasm applied to the basement membrane of some of the blood vessels. The possibility exists that the true extracellular space in PCrier and GrCgoire’s material is an artefact introduced by the rupture of glial cells secondary to swelling that occurred during the postmortem period prior to fixation. Also in Table I we have indicated some chemical constituents that are probably exclusively compartmentalized to a particular anatomical white matter space. It is well accepted that the determination of the hemoglobin in brain tissue can yield an estimate of the volume of the blood vascular space if the hematocrit of the circulating blood is known. It is possible that the capillaries produce a streaming of the blood in the brain. Hence, the hematocrit of the blood circulating in the brain may be different from that in a larger vessel from which blood is sampled to determine the hematocrit. If this is the case, as some have suggested (Dobbing, personal communication), the true estimation of the blood vascular space must await the development of an ultramicrohematocrit method for capillary blood.

POSTMORTEM M U L T I P L E S C L E R O S I S W H I T E M A T T E R

499

The true extracellular space and the glial cell space in normal white matter probably have a similar electrolyte profile (sodium, potassium, and chloride). Whether bromide ion is an indicator of extracellular and glial cell chloride ion space in postmortem white matter has yet to be resolved (see section on bromide ion). Perhaps macromolecules such as albumin and immunoglobulin-(; (corrected for contribution by the blood) may reside primarily in the true extracellular space, as has been shown for other macromolecules, such as ferritin (Brightman, 1965). On the other hand, the myelin sheath is rich in lipids, and the total lipid determination can be used as a type of indicator of the myelin sheath space. Our data indicate that in completely demyelinated white matter approximately 20-25 percent (40 g of lipids per kg) of the total lipid of normal-appearing white matter is non-myelin lipid. Perhaps the use of cerebroside as a myelin sheath space indicator would be more appropriate. The cerebroside of white matter is probably located mainly in the myelin (Kishimoto ef al., 1967) and is, therefore, a more specific measure of myelin concentration than total lipids, which include axonal lipids and those naturally in glial and endothelial cells, as well as the myelin lipids. Data we have recently accumulated and that of other investigators strongly suggest that gangliosides of white matter are located primarily i n the axons (Kishimoto et a/., 1967). Within multiple sclerosis lesions the situation is vastly more complicated. If the expanded true extracellular space in plaque tissue, described by Ptrier and GrCgoire, (1965) is not an artifact, the glial cell space is essentially non-existent. However, their electron micrographs show at least in some instances that the basement membranes of the capillaries tranversing the demyelinated areas are still surrounded by glial cytoplasm. Perhaps this can function as a type of regulator of the blood-brain barrier. Hence, we may have electrolytes (sodium, potassium, and chloride ion), as well as bromide ion, albumin, and immunoglobulin-G, as an indicator of this space. In the case of plaques of demyelination an appreciable amount of material (onesixth to one-third of the total dry weight) “forms” sheets or fibers during homogenization (Kishimoto et a/., 1967). Incidentally, these sheets from the plaque tissue do not seem to have been noted by other workers. Possibly it is a conglomeration of fibers derived from proliferated fibrous astrocytes, seen in abundance with the electron microscope (Perier and GrCgoire, 1965). Even though collagen fibers have not been seen in plaques (Harken, personal communication), we have carried out hydroxyproline determination of homogenates, hesitantly suggesting that hydroxyproline may be a chemical correlate for these fibers. Other complications appear to be introduced by multiple sclerosis. Our data suggest that immunoglobulin-(; is synthesized in multifocal areas of the multiple sclerosis brain (Tourtellotte and Parker, 1965; Tourtellotte and Parker, 1966a; Tourtellotte and Parker, 1966b; Tourtellotte and Parker, 1967). We have suggested that the sites for formation are located at the edge of the plaque and in theperivascular inflammatory cuffs (mostly lymphocytes) (Simpson et al; Tourtellotte et al., 1966). Furthermore, it would appear that immunoglobulin-(; can move through the brain from areas of synthesisinto thecerebrospinal fluid(Tourte1lotte and Parker, 1966b). Hence, immunoglobulin-(; concentration compared to that in control white matter, or a ratio of References p . 520-522

500

W. W. T O U R T E L L O T T E

TABLE I1 BLOOD I N POSTMORTEM WHITE M A T T E R

Brain diagnosis

Whole blood (serum) mllkg

Number of brains

Control

10

4.0 =t1.3* (2.6 f 0.7)

Multiple sclerosis Normal appearing white matter

11

4.8 f 2.0 (3.2 i 1.2) 4.9 2.0 (3.1 & 1.3)

Plaque

*

*

Same 11 as above

Mean i s.e.m.

immunoglobulin-G to albumin, is offered as a chemical correlate to the synthetic space of immunoglobulin-G in the multiple sclerosis brain tissue. Blood in postmortem white matter The data in Table 11 show that there is on the average about 4.0 ml of blood per kg of brain in the control white matter and approximately 20 percent more in multiple sclerosis normal-appearing white matter and plaque tissue. This difference was not statistically significant (p < 0.2 and > 0.1). Relationship between lipid, water content, and fat-free dry weight in postmortem white matter The data are shown in Table 111. The control white matter contained 187 g/kg and approximately 16 percent less in the normal-appearing multiple sclerosis white matter (p > 0.05). The lipid content in the completely demyelinated plaques was markedly reduced to a mean value of 44 g/kg. Hence, on the average 143 g/kg represented the TABLE 111 R E L A T I O N S H I P B E T W E E N L I P I D , W A T E R C O N T E N T , A N D F A T F R E E D R Y W E I G H T IN P O S T MORTEM HUMAN BRAINS

Brain diagnosis Control white matter Multiple sclerosis Normal-appearing white matter Plaque

*

Mean f s.e.m.

Fat-free dry weight glkg

Number of brains

Total lipids glkg

10

187 f 4.0*

706

12.0

107 i 1.5

11

158 f 1.6

136 f 30.6

106 f 5.1

same 11 brains as above

44 f 6.0

850

106

Water glk

* 56.6

* 2.2

POSTMORTEM M U L T I P L E SCLEROSIS W H I T E MATTER

50 1

lipid content of the myelin sheath, and on the average 44 g/kg represented the lipids in other tissue elements, such as glial cells, lipophages, blood vessels and the associated perivascular inflammatory cuffs, and the normal-appearing but demyelinated axons which course through the plaque tissue. The significant increase of the water content (p < 0.05 and > 0.02) in plaque tissue appeared to replace the loss of the myelin. Hence, the tissue volume of the multiple sclerosis white matter may stay normal because a water space moves in when the myelin sheath moves out. These observations may explain in part the clinical observation that brains of multiple sclerosis patients have only a modest reduction of the total brain weight even though there exists a severe degree of demyelination. In our series the 1 I control brains weighed on the average (* SEM) 1300 g (* 36) and the 12 multiple sclerosis brains, I150 (*45). The 11.6 percent reduction in brain weight is statistically significant (p < 0.02 and > 0.01). Hence, we must be cautious about concluding that water replaces myelin loss precisely. On the other hand, the fat-free dry weight (mostly protein) was constant for the three tissues studied, 106-107 g/kg. Ganglioside content of postmortem white matter (Kishimoto et al., 1967) Only the data for stearic acid-containing gangliosides are shown in Table IV, TABLE IV G A N G L l O S l D E C O N T E N T O F POSTMORTEM W H I T E MATTER

Brain diagnosis

Control white matter Multiple sclerosis Normal-appearing white matter Plaque

*

Mean

Number of obser varions 1 1 (3 patients)

3 (2 patients) 8 (Same 2 patients as above)

p M / k g lipid 189

16.0*

223 (126, 238, 306) 126 232

1 1 50 f 96.5

910(720,840, 1170) 4870 f 543.7

s.e.m.

calculated on the basis of brain and lipid weight. Stearate represents about 85 % of the fatty acids in gangliosides, and no notable difference was seen in the relative amounts of the other acids(l6 : 0, 18 : I , 20 : 0). The most striking finding is that the concentration of gangliosides in fresh tissue is rather similar in plaque tissue, normal-appearing multiple sclerosis white matter, and control samples. The P value for the difference between control and plaque concentrations using the student t-test for the mean obtained from each patient was > 0.05. This retention of the ganglioside is emphasized by the comparison with the weight of total lipid. Where control and normal-appearing white matter values range between Rrfrrences p . 520-522

502

W. W. T O U R T E L L O T T E

TABLE V RELATION BETWEEN A X O N A L DENSITY I N D E X A N D G A N G L I O S I D E C O N C E N T R A T I O N

Brain diagnosis

Number of ohservatioris

Axonal density index

Ratio ganglioside concenrratiori per axon density index ~~

Control white matter Multiple sclerosis Normal appearing white matter, Plaque

*

5 (3 patients) 2 (2 patients) 7 (same 2 patients as above)

116

+. 17

13 (58 and 87) 46 4 12

1.4 4 0.15

2.7 (2.2 and 3.2) 2.5 :I: 0.46

Number of axons touching a standard grid per unit volume of tissue.

610 and 1,670 ,umoles/kg lipids, the plaque tissue ranged between 2,540 and 6,890 The relationship between axon densities and gangliosides concentration is shown in Table V. The five control samples counted yielded an average axonal density index of I16 (SEM = f 17), while the two normal-appearing white matter samples examined yielded a lower value, 73 (58 and 87). The density index for seven plaque tissue samples was 26 & 12, which is significantly lower than the control values ( p < 0.01). It would appear from these data that there is roughly a 60% decrease in axon density. Since gangliosides are apparently primarily neuronal substances, it is interesting to compare the ratio, ganglioside concentration in whole tissue divided by the axonal density index. Taking the data from the actual samples counted, the average ratio is 1.4 rt 0.5 in the control samples, 2.7 (2.2 and 3.2) in the normal-appearing white matter and 2.5 & 0.46 in the demyelinated samples. The difference in ratios between control and demyelinated matter is not significant. Even closer agreement is seen in the ratios from the normal-appearing white matter and plaque tissue of the same patients. Our finding that the ratio, ganglioside concentration: axon density, remains relatively unchanged during demyelination suggests that the gangliosides are located in the axon rather than in glial cells or myelin. As yet there is no method for isolating axons in quantity or for histological staining for gangliosides, so the question cannot be settled. Preparations of myelin isolated by centrifugal methods vary in their ganglioside content, and it may be that the gangliosides appear in the myelin as the result of axonal contamination or of ionic binding of liberated gangliosides (Norton and Autilio, 1966). Since multiple sclerotic plaques contain other formed elements besides axons, particularly astrocytes, there is the possibility that some of theplaquegangliosides are derived from these cells. However, Lowden and Wolfe (1964), in a study of pathological human cortex exhibiting markedlossof neurons and heavy astrocytic proliferation,

503

POSTMORTEM M U T L I P L E SCLEROSIS W H I T E MATTER

found a great decrease in ganglioside concentration. From this it would appear that astrocytes contain little or no gangliosides. If the gangliosides of white matter are primarily contained in the axoplasm, determination of the ganglioside concentration in white matter may give a value related to the axon density or the state of the axons. A particularly sensitive index of the state of a section might be obtained if both cerebroside and ganglioside determinations were made. The cerebrosides of white matter are probably located mainly in the myelin and are, therefore, a more specific measure of myelin concentration than total lipids. Such a pair of determinations might aid in distinguishing primary from secondary demyelination. TABLE V I T OT AL H Y D R O X Y P R O L I N E C O N C E N T R A T I O N O F POSTMORTEM W H I T E MATTER

Hyrlroxyproline mglkg Control

Multiple sclerosis

~~

Brain diaprosis

Control LaJ. * Csrnpb. Multiple sclerosis Hinch.

White matter Total

Corrected for blood vessels

62 60

17 15

Bog.

Redd. Wacht. Wi-diffuse dernyelination Mean

61

16

Normal-appearing white matter Total

Corrected for blood vessels

54 82 90 84

0 28 36 30

78

24

Plaque Total

Corrected for blood vessels

148 204 200 118 141 162

94 150 146 64 87 108

* Patient identification

**

See text.

Total hydroxyproline concentration of postmortem white matter. The preliminary data are shown in Table VI. The control and the normal-appearing multiple sclerosis white matter had similar hydroxyproline concentrations on the average, 61 to 78 mg/kg, whereas the plaque tissue had approximately 2.5 times this value, 162. When a type of correction was made for blood vessel walls, which are rich in hydroxyproline (see Methods), there is a further relative increase to approximately 6 times the control value. Further experimentation is necessary to determine whether the increase of hydroxyproline in plaques is due to exclusive subcellular localization in the fibrous astrocytes. If the fibrous material that we have noted on homogenizing plaques of multiple Rifiwnces p . 520-522

504

W. W. T O U R T E L L O T T E

I""]

Fig. 1. Water, sodium, potassium, and chloride content of postmortem white matter. Control patient was Campb and the multiple sclerosis (MS)patient was Wacht.

sclerosis can account for the increase in hydroxyproline noted in the total homogenate, and if these isolated fibers appear like fibers of fibrous astrocytes with electron microscopy, the compartmentalization of hydroxyproline to fibrous astrocytes will be on a sounder experimental base. Sodium, potassium, and chloride content of postmortem white matter The data shown in Fig. 1 agree well with those published by Lowenthal(1961). For control cerebral white matter they found a sodium concentration of 65 to 78 mequiv./ kg and potassium of 56 to 72. Our sodium values were 77, 68 and 76 mequiv./kg for control white matter and normal-appearing multiple sclerosis white matter and plaque tissue, respe:tively. On the other hand, our potassium values were 60, 72, and 57 respectively. To our knowledge human postmortem white matter chloride values have not been reported; our values were 47, 42, and 50 mequiv./kg. It is known that death destroys electrolyte transport mechanisms of the brain (Katzman, 1966). Furthermore, to our knowlegde the concentration of sodium, potassium, and chloride ion concentration in living human white matter obtained by biopsy at the time of surgery is not available in the literature. So to determine the effect of postmortem movement of ions we have used the figures reported by McIlwain (1955) for whole human brain. The sodium, potassium, and chloride ion concentrations were 57,96 and 37 mequiv./kg, respectively. It would appear in general that our postmortem values for sodium and chloride are modestly high and our potassium values very low. Because of the artifact introduced by the postmortem state (Katzman, 1966) no attempt will be made to calculate the sodium, potassium, or chloride ion spaces. Since plaques and normal-appearing white matter contain more water than control white matter the data are expressed on the bases of mequiv./l of tissue water. It was found that control white matter contained 106 mequiv. of sodium ion, the normal-

POSTMORTEM M U L T I P L E S C L E R O S I S W H I T E MATTER

505

appearing multiple sclerosis white matter 92 and the plaque tissue, 91. The same trend was found in chloride ion concentration. The control white matter contained 64 mequiv./l, the normal-appearing multiple sclerosis white matter 58 and the plaque tissue, 60. On the other hand, the potassium ion concentration was 82 mequiv. in control white matter, 97 in normal-appearing multiple sclerosis white matter and moderately reduced in the plaque tissue to 68. No conclusions will be drawn from our meager postmortem data. However, it would appear fromacomparisonof the plaque concentration of sodium ion (mequiv./l of tissue water) to surrounding normal-appearing white matter from the same patient that the concentrations do not differ, even though both are moderately lower than the concentration in the control white matter. This same trend was shown for the chloride ion. Hence, it is possible that the postmortem sodium and chloride ion spaces in multiple sclerosis normal-appearing white matter do not differ from the plaque tissue. On the other hand, itwould appearfrom a comparison of the plaque tissue concentration of potassium ion to surrounding normal-appearing white matter from the same patient that the concentration was markedly lower in the plaque tissue, whereas the control white matter concentration did not reach the value in the normal-appearing multiple sclerosis white matter but was higher than that in the plaque. A partial explanation of the low value of potassium in the plaque tissue compared to the surrounding normal-appearing multiple sclerosis white matter might be that there was a significant reduction of axons which are known to be rich in potassium. Unfortunately, we did not estimate the axon density in this specimen, but data shown in Table V on plaque tissue from other patients indicate that a reduction of axons in a plaque can exist. SOME TESTS O F T H E B A R R I E R

Two types of experiments to test the blood-white matter or plaque tissue barrier in postmortem brain tissue will be described. A foreign anion, bromide, which is known to penetrate the brain slowly and come to a steady-state concentration below the blood level of approximately 0.3 (Haerer et al., 1964), was administered orally ante mortem, and the distribution of the ion was determined post mortem in different areas of brain coronal sections. The second test used was the distribution in brain coronal sections of endogenous serum proteins, namely, albumin and immunoglobulin-G. Both of these proteins penetrate the blood-brain barrier exceedingly slowly, if at all (Barlow, 1964). Bromide ion concentration in postmortem multiple sclerosis white matter and plaque tissue Some of the data are presented in Table VII. It can be seen that all the areas determined to be plaques of demyelination or diffuse demyelination by gross microscopic inspection had a low lipid content; conversely, the normal-appearingwhite matter had a much higher concentration. The water concentration of the dissected areas are also given. In general the plaque tissue and areas of diffuse demyelination had a higher water concentration than the normal-appearing white matter. References p . 520-522

506

W. W. TOURTELLOTTE

TABLE V I I BROMIDE I O N C O N C E N T R A T I O N I N POSTMORTEM MU LTIP LE SCLEROSIS W H I T E MATTER ~~~

Bromide ion concentration Multiple sclerosis patient identification

BO Normal-appearing white matter 1

2 3 Plaque 1 2 3 BU Normal-appearing white matter 1

2 3 Plaque 1 2 WR Normal-appearing white matter 1

2 3 4 5

6 7 8 9 Plaquc 1 2 3 4 5

Lipids glkg

Water glkg

Serum mequiv.11

CSF mequiv./l

2.0

0.6

Brain mequiv./l tnequiv.lkf of tissue water

207 169 I 60

689 736 742

0.32 0.42 0.17

0.46 0.57 0.23

52 41 38

856 869 882

0.39 0.39 0.72

0.45 0.45 0.82

196 187 181

704 705 709

0.41 0.41 0.31

0.58 0.58 0.44

45 27

874 894

0.20 0.42

0.23 0.47

212 188 186 180 174 169 158 142 142

672 725 738 738 718 751 770 775 784

0.61 0.66 0.70 0.62 0.61 0.77 0.61 0.84 0.70

0.9 1 0.91

26 26 20 18 7

882 890 814 878 890

0.62 0.38 0.71 0.30 0.80

2.5

2.9

not done

0.9

0.95

0.84 0.85 1.02 0.79 1.08 1.01 0.70 0.43 0.87 0.34 0.90

507

POSTMORTEM M U L T I P L E SCLEROSIS W H I T E MATTER

T A B L E V l I (ctnd) B R O M I D E I O N C O N C E N T R A T I O N IN P O S T M O RTEM M U L T I P L E S C L E R O S I S W H I T E M A T T E R

Bromide ion concentration

Multiple sclerosis potient irlentificritioti

Lip ids Rlk

Wuter g/kg

Seriini nieqiiiv./l

Brain

CSF nwquiv./l

tnequiv./kg

tnequivJ1 of tissue n’oler

.~

~~~

--

--

HE Normal-appearing white matter I 2 3 4 5 6 7 8 9 Plaque 1

2 3 4 5

WA Normal appearing white matter Plaque HO Normal appearing white matter Diffuse demyelination I 2 3 4 5 Plaque 1

-

~

3.6

1.1

217 215 213 20 I 196 I82 I47 141 123

674 685 683 685 717 724 756 767 790

0.84 0.88 0.89 0.92 0.90 0.69 0.89 0.76 0.92

1.25 1.28 1.30 0.74 1.26 0.95 0.85 0.99 1.16

37 33 31 28 24

849 877 870 894 882

1.26 1.12 1.18 0.99 1.22

1.49 1.28 1.36 1.11 1.38

147 24

778 880

0.94 0.78

1.19 0.89

131

77 5

1.83

2.36

95 90 86 78 55

818 833 82 I 828 864

2.04 1.42 1.70 1.84 I .71

2.49 1.70 2.07 2.22 1.98

28 24 16

89 1 880 906 907 916 918 910 914

1.86 2.32 2.20 2.34 1.84 1.82 1.98 1.95

2.08 2.64 2.43 2.58 2.01 1.98 2.18 2.14

10

7 7 5 2

4.0

5.4

1.8

3.8

It can also be seen that a direct relationship between the serum concentration of bromide ion and the concentration in the cerebrospinal fluid and in the brain existed. I t is reasonable to assume that the brain bromide ion concentrations are probably Rrfrnvi~r3.sp.: 520-522

508

W. W. TOURTELLOTTE

Fig. 2. Bromide ion concentration in multiple sclerosis white matter.

steady-state ion concentrations. The oral medication was given at regular intervals (three times per week) for no less than one year. Furthermore, for any given patient the serum bromide levels were maintained at a constant level. For example, patient WA had serum bromide ion concentrationsof 3.9 mequiv./l in August, 4.0 in September, and 4.2 in December, 1964 and 3.8 in April 1965 and 4.0 in June 1965 (12 h post mortem). Furthermore, the ratio of serum to cerebrospinal fluid was reasonably constant over this period of time too (2.2,2.3,2.2,2.4,2.2, and 2.0). In the majority of the cases the brain concentration of bromide ion is equal to or approaches that in the cerebrospinal fluid. In Fig. 2 the concentration of the bromide ion per liter of tissue water in a dissected area was plotted against its concentration of lipids per kg for each patient shown in Table VII. The type of tissue, plaque of demyelination, diffuse demyelination or normal-appearing white matter are also shown. Inspection of the plotted data reveal no obvious bromide ion concentration differences between the plaque tissue and the areas of diffuse demyelination when compared to the normal-appearing white matter for any given patient. Fig. 3 presents a photograph of a dissection plan of a coronal section of a multiple sclerosis patient (HO) in order to show the relationship between the bromide ion concentration in plaques and the adjacent white matter, as well as several selected gray matter areas. The data for each dissection are shown in Table VIII. Inspection of the bromide ion concentration per liter of tissue water for plaque tissue was not different from that in the surrounding normal-appearing white matter. In most dissected regions the brain bromide ion concentration approached or equaled the cerebrospinal fluid concentration. A notable exception to this is thevalues obtained for cerebral cortex,

509

POSTMORTEM M U L T I P L E S C L E R O S I S W H I T E M A T T E R

TABLE VIII BROMIDE ION C O N C E N T R A T I O N I N DISSECTED P A R T S OF T H E C O R O N A L S E C T I O N S H O W N I N FIG. 3

Gross and microscopic description

Dissection number see Fig. 3

Bromide ion

Lipids

Water

glkg

glkg

rnequiv.lkg

mequiv.11 of tissue water

28 59 106

872 841 794

1.96 1.88 1.61

2.25 2.23 2.03

1

Plaque in corpus callosum Plaque in corpus callosum Corpus callosum adjacent to plaque 1 and 2

5

Periventricular plaque Deep portion of periventricular plaque Periventricular plaque

82 62

818 838

1.86 2.21

2.28 2.64

26

874

1.85

2.12

15

Shadow plaque White matter adjacent to shadow plaque 15

87 38

813 762

1.71 2.10

2.10 2.76

21 22

Plaque White matter around plaque 21 White matter adjacent to 22

48 75

852 825

1.94 1.63

2.28 1.98

149

751

1.77

2.36

2 3

10 4 24

23 I1 12

Plaque White matter around plaque 11 White matter adjacent to 12 White matter adjacent to 12

48 90

852 810

1.96 I .94

2.30 2.39

I49 149

75 1 735

1.70 1.58

2.26 2.15

Plaque in caudate nucleus Anterior internal capsule Plaque in putamen Putamen

84 129 97 97

816 771 803 803

1.37 1.39 1.44 1.30

1.68 1.80 1.80 1.62

Cerebral cortex plaque Deep portion of cerebral cortex plaque 18 Cerebral cortex adjacent to plaques 18 and 19

58 63

799 820

2.59 2.65

3.24 3.23

55

742

2.32

3.20

20

Cerebral cortex plaque

63

810

2.60

3.1 1

16

Arachnoid membrane

48

852

2.05

2.41

I3 14 6 7 8 9 18 19 17

which in all cases exceeded the cerebrospinal fluid value as well as that for the basal ganglia. Again the data obtained on this patient show no obvious bromide ion concentration difference between the plaque tissue and the surrounding normal-appearing white matter. An examination of the bromide steady-state ratio (RCSF = concentration in the cerebrospinal fluid/concentration in serum) for each multiple sclerosis patient studied (data presented in Tables VII and VIII) revealed a direct relationship between R C ~ F R&rmci.s p. 520-522

510

W. W. T O U R T E L L O T T E

Fig. 3. Legend is o n the photograph.

values and the serum bromide level. Serum bromide levels of 2.0 to 3.6 mequiv./l, 4.0 to 4.8, and 5.4 gave RCSFvalues of 0.31,0.45 and 0.7, respectively. This result may be explained by a mechanism recently proposed by Bito, Bradbury and Davson (1966). Their data supported the hypothesis that bromide ions passively diffuse across the blood-brain barrier and that the low steady-state distribution ratio (RCYF= 0.7) found for bromide ion was due to an active process in the choroid plexus that tends to drive bromide ion out of the cerebrospinal fluid. If this hypothesis were true, it might be predicted that the higher the serum bromide ion concentration, the greater the inhibition of the active transport system, perhaps by saturating the limited number of carrier sites (Bito e? al., 1966); hence, the bromide ion is not pumped out of the cerebrospinal fluid, so the cerebrospinal fluid bromide ion concentration rises and the R ~ s p increases. Unfortunately, Bito et al. (1966) studied only one dosage level, which was very high compared to our doses on a per kg body weight base. They gave an intravenous dose of sodium bromide of approximately 7.5 mequiv./kg, whereas we used initially 1.25 mequiv./kg and a maintenance dose of 0.1 mequiv./kg three times a week. Since it would appear from our data that higher serum bromide levels inhibit the active transport of bromide ion out of the cerebrospinal fluid, it is reasonable to assume that Bito e? al. were studying the effect of inhibitors and the size of the bromide ion space on an active transport system initially inhibited by a high serum bromide level. Because we have studied postmortem tissue we have purposely avoided a discussion of the calculation of electrolyte spaces. However, it has occurred to us after a comparison of calculated postmortem bromide and chloride ion spaces that the bromide space

POSTMORTEM MULTIPLE SCLEROSIS W H I T E MATTER

-

51 1

WHITE MAnER

C

'E

* M + 2 S.D.

*

M e a n + 2 x Standard Deviation

Fig. 4. Stratification of albumin concentration of postmortem white matter corrected for blood albumin. M.S. = multiple sclerosis.

was larger. The postmortem chloride: ion space (100 x postmortem tissue chloride ion concentration mequiv./l of tissue water divided by 101 mequiv. chloride ion/kg of postmortem cerebrospinal fluid). The postmortem chloride space values were 57, 5 I and 53 percent for specimens of control white matter, normal-appearing multiple sclerosis white matter, and plaque tissue, respectively. On the other hand, the postmortem bromide spaces (100 x postmortem tissue bromide ion concentration mequiv./l of tissue water divided by bromide ion concentration of the cerebrospinal fluid) were larger. The mean and range were obtained from the data in Tables VII and VIII. The values were: 84 percent (63-103) and 83 percent (47-1 20) for specimens of normal-appearing multiple sclerosis white matter and plaque tissue, respectively. Perhaps, when bromide ion is administered for a long period of time (years) resulting in serum bromide ion concentrations ranging from 2 to 5.4mequiv. per 1 the bromide ion penetrates into more than the chloride ion space. Bito et al. in their experiments found at 24 hours that the bromide and chloride ion spaces were identical. Furthermore, inpatient HO (see table 8) we found that the bromide ion concentrationin thecortical mantleactually exceeded that in the cerebrospinal fluid.

Relationships between Albumin and Immunoglobulin-G in Postmortem Control White Matter, Normal-Appearing Multiple Sclerosis White Matter and Plaque Tissue and Cerebrospinal Fluid Fig. 4 presents the albumin concentration (corrected for serum albumin) for each patient's brain region dissected. The mean control white matter concentration was R&rcnccs

p.

520-522

512

W. W. TOURTELLOTTE

M t 21.0. , A W N + 2SlANDARO DNlAIlON OF CONTROI “M;-NYAN of VAWtSM+ZS.D.

Fig. 5. Stratification of immunoglobulin-G concentration of postmortem white matter corrected for

blood immunoglobulin-G.

identical to that in the normal-appearing multiple sclerosis white matter, 152 mg/kg (& 28 = S.E. of the mean) and 145 f 22, respectively. On the other hand, the plaques of demyelination had a higher mean value (190 f 32), with two patients outside the mean plus two standard deviation level. However, the elevated value found in plaques was not statistically significantly different from the surrounding normal-appearing white matter or control white matter@ < 0.4 and > 0.3). Moreover, the concentration of albumin in control white matter, normal-appearing multiple sclerosis white matter, and plaque tissue were very similar, if calculated on the basis of mg/l of tissue water The values were: 215 &- 39.6, 197 f 30.0 and 212 f 35.6, respectively. Fig. 5 presents the immunoglobulin-(; concentrdtion (corrected for serum immunoglobulin-G) for each patient’s brain region dissected. The control white matter value was 162 f 26 mg/kg, the normal-appearing multiple sclerosis white matter, 433 135 and the plaques of demyelination, 434 f 98. There was a statistical difference between the control white matter value and the multiple sclerosis specimens (p < 0.01). On the other hand, the immunoglobulin-G concentration was the same for the normalappearing multiple sclerosis white matter and plaque tissue. Inspection of the frequency distribution of immunoglobulin-G data from the multiple sclerosis specimens revealed that there are probably two modes. The first mode ((‘MI”) was calculated on values which fell below the control white matter mean plus two standard deviations and the second one (“Mz”) was calculated on values which were above this cut-off point. There was no statistical difference between “MI” and the control white matter, whereas “Mz” was statistically highly different (p <

POSTMORTEM M U L T I P L E SCLEROSIS WHITE MATTER

513

WHITE MATTER With blwd

600 -T

Ratio

s.e.m.

Corrected for blwd

T

0.08

0.34

0.26

0.19

0.80

0.48

Fig. 6. Immunoglobulin-G and albumin concentration in postmortem white matter. Ten controls and 1 1 multiple sclerosis patients. y G = immunoglobulin-G; C = control; MS = normal-appearing multiple sclerosis white matter; P = plaque tissue; s.e.m. = standard error ofthe mean.

0.003). “M2” was the same for normal-appearing multiple sclerosis white matter and plaque tissue. Fig. 6 presents the mean ratio of immunoglobulin-G over albumin. The serumcorrected ratio was 1.1 for control white matter, 3.0 for normal-appearing multiple sclerosis white matter and 2.3 for the plaque tissue. Therefore, on the average there was a significantly different distribution of immunoglobulin-G and albumin in the soluble protein fraction of multiple sclerosis brains for the regions studied. Furthermore, this redistribution in multiple sclerosis was due to an increase of immunoglobulin-G and not to a decrease of albumin. The possible factors to explain an increase of immunoglobulin-G in the multiple sclerosis brain tissue are as follows: ( I ) Hyperemia of the multiple sclerosis brain, hence, introducing serum immunoglobulin-G ; (2) Hypergammaglobulinemia with transudation of immunoglobulin-G into the brain; (3) Decrease of the blood-brain barrier to immunoglobulin-G; (4) Synthesis of immunoglobulin-G in the brain itself; ( 5 ) Selective secretion of immunoglobulin-G in the brain itself; (6) Combination of 1 through 5 . Some experiments have been done to determine which of these factors might be responsible for the high immunoglobulin-G in multiple sclerosis brain tissue. Hyperemia does not exist in multiple sclerosis brain tissue (see data in Table 11). Furthermore, multiple sclerosis patients do not have hypergammaglobulinemia (Tourtellotte and Parker, 1965). The albumin concentration whose MW is 40,000 was not increased in multiple sclerosis brain tissue (see data in Fig. 4); hence, it is reasonable to assume that the blood-brain barrier could be intact for larger molecules like immunoglobulin-G, whose MW is 140,000.The possibility exists that immunoglobulin-G is selectively secreted into the multiple sclerosis brains. No experimental data References p . 520-522

514

W. W. T O U R T E L L O T T E

Fig. 7. Coronal section ( 3 mm thick) through the posterior portion of the lenticular nucleus of a control cadaver (Campb). The pattern of the dissection is outlined on the right photograph. The superimposed numbers within the outline of the dissection are the concentration of imrnunoglobulin-G (corrected for serum imrnunoglobulin-G) in mg/kg.

exists to disprove this, nor any to support it. Hence, we have elected to mention it only as an outside possibility. Based on the above observations, we have postulated that in the majority of patients who die with multiple sclerosis there exist multifocal areas of synthesis of immunoglobulin-G. We have discussed the possible sites of production of immunoglobulin-G in multiple sclerosis brain tissue (Tourtellotte et a/., 1966). The data supported the hypothesis that proliferation of perivascular cells (mostly lymphocytes) around capillaries and venules is sufficient and probably necessary to produce an elevation of irnmunoglobulin-G in brain tissue from patients with multiple sclerosis. Furthermore, the results also indicated that cells in the advancing margin of a plaque other than perivascular mononuclear cells might produce immunoglobulin-(?. On the other hand, evidence for synthesis of immunoglobulin-G in some multiple sclerosis brain tissue was not found. It has been established from numerous histopathological studies (Adams and Kubik, 1952; Ibrahim, 1965; McAlpine et a/.. 1955) that plaques can be empirically classified into early and old types and old plaques can be further subdivided into active and inactive forms. We are postulating that an elevated immunoglobulin-G value in a multiple sclerosis brain is a chemical indicator of an active plaque of multiple sclerosis. It is possible that the normal-appearing multiple sclerosis white matter surrounding an active plaque has an elevated immunoglobulin-G value because the immunoglobulin-(; synthesized in an active plaque has diffused into it. To test this possibility

POSTMORTEM M U L T I P L E S C L E R O S I S W H I T E M A T T E R

515

Fig. 8. Coronal section ( 3 nini thick) through the posterior portion of the lenticular nucleus of a cadaver with multiple sclerosis (Wacht). I t is essentially the same level of the brain as shown in Fig. 7. On thc left photograph note the large plaque of demyelination which extends from the ventricles deep i n t o the white matter and into the insular cortex. Dorsally there are two cerebral cortex plaques which were ccmbined for analysis (937 on the right photograph). The pattern of the dissection is outlined on the right photograph. The superimposed numbers within the outline of the dissection are the concentration of immunoglobulin-G (corrected for serum immunoglobulin-G) in rng/kg.

we have dissected coronal sections from a control and a multiple sclerosis brain like a mosaic along anatomico-pathological land marks and analyzed the dissected specimens for immunoglobulin-G concentration. Fig. 7 shows the distribution in a control coronal section through the posterior part of the lenticular nucleus. It can be seen on the right hand photograph that concentration of immunoglobulin-(; was rather uniform across the section. On the other hand, when a comparable coronal section from a multiple sclerosis patient was studied, the abnormally elevated immunoglobulin-G appears to be most concentrated in the largest plaque itself (2,072, 2,237 and 2,352 and 1,892 mg/kg) or the periventricular plaques (2, 040 and 2,366) and less in the cerebral cortex plaques (937, 859 and 703). On the other hand, in the surrounding normal-appearing white matter of the largest plaque the value fell to about half of the value (1,064 and 918). Furthermore, the more distant normal-appearing white matter had even lower values, for example, 613 and 367. To carry the idea of diffusion of immunoglobulin-(? through multiple sclerosis brain tissue further, we have correlated the concentration of immunoglobulin-G in multiple sclerosis brain tissue and the concentration of this globulin in the cerebrospinal fluid (Tourtellotte and Parker, 1966). The data are shown in Fig. 9. The tentative conclusion drawn from these results was that in patients with multiple sclerosis, the inReliwrrc cs p. .52O-F22

516

W. W. T O U R T E L L O T T E

150-

a n . -. r 10062

.

B

0

4 .

0

. I

a

1

Control M I 3 SD.

Fig. 9. A correlogram of the immunoglobulin-G concentration in premortal cerebrospinal fluid and the same multiple sclerosis pateint’s plaque tissue immunoglobulin-G concentration (corrected for blood immunoglobulin-G), yG = immunoglobulin-G; M = mean; S.D. = standard deviation.

crease in immunoglobulin-(; in the cerebrospinal fluid is a reflection of an excess of this globulin in the brain. Brightman (1965) has shown by electronmicrographs that ferritin (MW 500,000) can move within the brain between the tissue pericellular spaces (approximately 200 A wide). Hence, it is reasonable that immunoglobulin-(; (MW of 140,000) could diffuse through the same spaces. COMMENTS

The dissection of the plaque materialin theaboveanalyses were performed on plaques easily discernible by inspection. Also the edge of the plaque was grossly:cut inlorder to exclude the surrounding normal-appearing white matter. Hence, the most active areas of demyelination, which are at the edge of a chronic plaque, were probably not taken for analysis. Hence, it is also reasonable to presume that if a small amount of “active” demyelinating tissue were included in the dissection that its constituents would not significantly dilute the larger area of chronic demyelination taken for analysis. Furthermore, the biopsies of the plaques taken for microscopic histology were from the center of the plaque. They all showed severe demyelination. Hence, it is our impression that the above analyses were done on chronic plaques with insignificant contamination from “active” demyelinating tissue. Therefore, the data from this study done on postmortem tissue suggest the conclusion that chronic plaques of multiple sclerosis do not have an altered blood-brain barrier to albumin and to bromide ion. Broman (1947) perfused a multiple sclerosis brain after death with trypan blue and stained areas of demyelination, but not normal-appearing white matter. He concluded from this study that the blood-brain was decreased in plaque tissue of multiple sclerosis. However, in a subsequent publication (1964) the picture appears complicated.

POSTMORTEM M U L T I P L E S C L E R O S I S W H I T E MATTER

517

The trypan blue staining of the postmortem sclerotic plaque appeared uniform; there was no accumulation of the dye either in the parenchyma or in the vascular endothelium. This is a unique finding according to Broman, and it is the only known example in his experience of a blood-brain barrier damage without a coexisting lesion of the vascular wall. He thinks it may be consistent with a primary lesion of the perivascular glia. The electron micrographs presented by PCrier and GrCgoire (1965) may help in explaining the trypan blue staining of the plaques described by Broman in postmortem plaques. I t is possible that the trypan blue is contained in an expanded extracellular space. Furthermore, Perier and GrCgoire presented an electron micrograph that showed an intact layer of endothelial cytoplasm, a continuous basement membrane, and a discontinuous thin layer of glial cytoplasm applied to the outer surface of the vessel basement membrane. Perhaps the trypan blue moved through the vessel wall not covered by glial cytoplasma into the true extracellular space. This hypothesis places the normal exclusion of trypan blue from the brain at the level of the perivascular glial feet or at the level endothelial cells of the vessel wall influenced in some way by the perivascular glial feet. Our data indicate that in postmortem brain tissue the concentration of various ions (sodium, chloride, bromide) and albumin were normal in the chronic plaques when compared to surrounding normal-appearing white matter and control white matter. In view of the findings of PCrier and GrCgoire (1965) that the expanded extracellular space in multiple sclerosis plaques could be in direct contact with the basement membrane of endothelial vessel walls without a continuous layer of perivascular glial feet would suggest that sodium, chloride, bromide, and albumin concentrations are determined by the intact vessel wall. On the other hand, the low potassium concentration of the plaque tissue may indicate that the perivascular glial cells may somehow be involved in the maintenance of the extravascular tissue concentration. However, the posybility that we suggested earlier that the decrease of axoplasm in plaques which are rich in potassium ions is also a possibility. Of interest is the study of Gonsette and AndrC-Balisaux (1965). They injected radioactive phosphorus (3") intravenously into three patients prior to death (they studied the same patients as PCrier and GrCgoire). They then cut coronal sectionsof the brain as rapidly as possible (usually one to three hours post mortem) which showed plaques of demyelination and prepared radioautographs. They found that chronic intracerebra1 gliotic plaque tissue did not pick up the isotope; on the other hand, periventricular plaques did mark the autoradiographs. They suggested that 32P could move from the cerebrospinal fluid into the periventricular plaque, whereas it could not from the blood to the brain. The two most popular etiologies of multiple sclerosis are that it is due to autoimmunity (McAlpine et al., 1965) or a slow virus infection (Palsson et al., 1966). Perhaps in multiple sclerosis there is a destruction of the myelin or of the oligodendroglia cell which maintains the health of the myelin by an immunological mechanism, i.e., the multiple sclerosis patient is allergic to his own myelin, or the oligodendroglia related to myelin is diseased by a virus. The discovery that the majority of patients who die References p.lS20-522

518

W. W . T O U R T E L L O T T E

with multiple sclerosis have an elevated immunoglobulin-G in the brain supports both of these possible etiologies. The fact that the albumin content was normal but the immunoglobulin-G was elevated suggested that the elevated immunoglobulin-G was synthesized in the brain. Other experiments on postmortem multiple sclerosis brain tissue supported the hypothesis that multifocal areas of synthesis of immunoglobulin-G which are probably located in active plaques of demyelination and in the accompanying perivascular cuffs of lymphoid cells, were responsible for the increased values (Tourtellotte et a/., 1966). Moreover, our results imply that the increase of immunoglobulin-(; in the cerebrospinal fluid is the result of diffusion through the normalappearing white matter of an excess of this globulin in the brain. Extrapolation of the results of this postmortem study to the living brain must be done with some hesitation. Numerous artifacts could be introduced during the agonal state of death and the postmortem period. For example, death destroys potassium transport mechanisms and postmortem cerebrospinal fluid potassium rises at a constant. rate while the brain potassium falls (Mason et nl., 1951; Naumann, 1958; Schain, 1964). Furthermore, the limitations of getting sufficient material, such as biopsies of plaque tissue from living multiple sclerosis patients, leaves us in an experimental position to have to wait for biopsy specimens obtained at cryothalamotomy for treatment of intention tremor, which will include both normal-appearing white matter and plaque tissue. SUMMARY

I . The mean values of the vascular space, estimated by a carboxyhemoglobin method, in postmortem control white matter was statistically the same as normal-appearing multiple sclerosis white matter and the plaque tissue; the mean values and standard error of the means were: 4.0 5 1.3, and 4.8 2.0 and 4.9 f 2.0 ml/kg, respectively. 2. After complete myelin loss in postmortem multiple sclerosis white matter the total lipid value was 44 f 6.0 g/kg, whereas the normal-appearing multiple sclerosis 4.0. white matter was 158 4 1.6 and the control white matter, 187 3. When there was myelin loss in the postmortem white matter it was replaced by the water. The control white matter contained 706 f 12.0 g/kg, whereas the normalappearing white matter had 736 & 30.6 and the plaque tissue 850 rt 56.6 4. On the other hand, the fat-free dry weight of the postmortem white matter (mostly protein) was constant. The control white matter had a value of 107 i 1.5 g/kg and the normal-appearing multiple sclerosis white matter 106 & 5.1 and the plaque 2.2. l tissue, 106 5. Ganglioside concentration of postmortem white matter, which is probably compartmentalized to the axoplasm, was significantly lower in plaques. The value for control white matter was 189 & 16.0 pmoleslkg, whereas the normal-appearing multiple sclerosis white matter was 223 (range = 126 + 306), and plaque tissue, 126 23.2. However, the ganglioside concentration divided by the axonal density index was the same for the three types of tissue studied. 6. The total hydroxyproline concentration of postmortem white matter, which may

POSTMORTEM M U L T I P L E S C L E R O S I S W H I T E MATTER

519

reflect the presence of blood vessel walls and possibly the filamentous portion of the fibrous astrocytes, was 61 mg/kg (range = 60 -+ 62) for the control white matter, 78 (54 + 90) for the normal-appearing multiple sclerosis white matter and 162 ( I 18 -+ 204) for the plaque tissue. After a type of correction was made for blood vessel walls the values were 16, 24 and 108, respectively. 7. Postmortem white matter concentration of sodium ion did not appear to differ in the control white matter, normal-appearing multiple sclerosis white matter, or plaque tissue. The values were, respectively, 106, 92, and 91 mequiv./l of tissue water. These values are modestly high when compared to values available for fresh tissue. 8. The postmortem white matterconcentration of potassium ion appears to be lower in plaque tissue. This may reflect the decrease ofaxoplasm in this tissue. The values obtained for control white matter, normal-appearing multiple sclerosis white matter, and plaque tissue are as follows: 82, 97, and 68 mequiv./l of tissue water, respectively. These values are very lowwhencomparedtothe concentrationavailableforfresh tissue. 9. Postmortem white matter concentration of chloride ion did not appear to differ in control white matter, normal-appearing multiple sclerosis white matter, or plaque tissue. The values were, respectively, 64, 58, and 60 mequiv./l. of tissue water. These values are modestly high when compared to values available for fresh tissue. 10. After giving sodium bromide orally for one to three years before death, the bromide ion concentration did not appear to have a different distribution in the postmortem plaque t issue when compared to theconcentrations in the surrounding normalappearing white matter. The concentration of the bromide ion in the tissue and cerebrospinal fluid, which was very similar, was dependent on the serum bromide ion level, i.e., the higher the serum level, the higher the cerebrospinal fluid and tissue concentration. For example, serum bromide levels of 2.0 to 3.6,4.0to 4.8, and 5.4 mequiv./l gave cerebrospinal fluid (and tissue) values of 0.6 to 1 . 1 , 1.8 to 2.2 and 3.8, respectively. We have suggested that this result could be best explained on the basis that bromide ion gets into the brain by passive diffusion. Furthermore, an examination of the bromide steady-state ratio (RcsF-concentration in the cerebrospinal fluid divided by the concentration in the serum) for each multiple sclerosis patient studied revealed a direct relationship between R C ~ Fvalues and the bromide level. Serum bromide F of 0.31, 0.45 and 0.7, levels of 2.0 to 3.6, 4.0 to 4.8 and 5.4 mequiv./l gave R C ~values respectively. We have suggested that this result (the elevation of the steady-state ratio by rising serum bromide ion concentration) could be best explained on the basis that higher serum bromide ion concentrations are inhibiting the active secretion by the choroid plexus of the bromide ion out of the cerebrospinal fluid into the serum. 11. The albumin concentration in postmortem control white matter corrected for serum albumin is not statistically different from that in normal-appearing multiple sclerosis white matter of plaque tissue. The values were as follows: 152 & 28, 142 i 22 and 190 $ 32 mg/kg, respectively. Moreover, the concentration of albumin in control white matter, normal-appearing multiple sclerosis white matter and plaque tissue was very similar if calculated on the basis of mg/l of tissue water. The values were 215 f- 39.6, 197 f 30.0, and 212 =k 35.6 respectively. Rcfer.rpnrmp . 520-522

520

W. W. TOURTELLOTTE

12. On the other hand, immunoglobulin-(? concentration was increased in the

majority of the postmortem brains of multiple sclerosis patients (64 percent) when compared to control postmortem brain tissue. The values corrected for serum immunoglobulin-(; for control white matter, normal-appearing multiple sclerosis white matter, and plaque tissue were, respectively, 162 26, 433 f 135, and 434 & 98 mg/kg. Furthermore, inspection of the frequency distribution of the immunoglobulin-<; data suggested there were two types of multiple sclerosis patients brains - the majority had an increased immunoglobulin-G concentration and the minority a normal concentration. We have suggested that those multiple sclerosis brains with an increased concentration of immunoglobulin-<; have multifocal areas of synthesis of immunoglobuin-G (advancing margin of the plaque tissue and/or perivascular cuffs of mononuclear cells). Perhaps an elevated immunoglobulin-(; concentration in the brain is a chemical indicator of an active case of multiple sclerosis. Some evidence was presented to support the hypothesis that immunoglobulin-G can diffuse through the brain into the cerebrospinal fluid. A mosaic type of dissection of a coronal section of a multiple sclerosis brain showed that immunoglobulin-G concentration was highest in the plaque tissue and fell to lower concentrations in the surrounding normal-appearing white matter. Moreover, a correlogram was presented between the concentration of immunoglobulin-G in the cerebrospinal and the plaque tissue. It was found that in patients with multiple sclerosis the increase in immunoglobulinG in the cerebrospinal fluid was a reflection of an excess of this globulin in the brain. 13. The data from this study done on postmortem tissue suggests the conclusion that chronic plaques of multiple sclerosis do not have an altered blood-brain barrier to albumin, bromide, sodium, potassium, and chloride ions. On the other hand, there may be multifocal areas of synthesis of immunoglobulin-G in the majority of multiple sclerosis brains, and the excess globulin diffuses from the active sites of synthesis (advancing margin of the plaque and/or perivascular cuffs of mononuclear cells) into the surrounding normal-appearing white matter and cerebrospinal fluid. 14. Extrapolation of the results of this postmortem study to the living brain must be done with some hesitation. ACKNOWLEDGEMENT

This work was supported in part by the National Multiple Sclerosis Society (272-5), National Institutes of Health (NB 05388-03) and the Kenneth H. Campbell Foundation for Neurological Research, Grand Rapids, Michigan, U.S.A. REFERENCES C. S. (1952) The morbid anatomy of the demyelinative diseases. Amer. J . ADAMS,R. D . AND KUEIIK, Med., 12, 510-546. ADDISON, W. H. F. (1950) Neurological Technique. R. McC. Jones (Ed.). McClung's Handbook of Microscopical Technique. New York, Paul B. Hoeber, Inc. (pp. 346388). BARLOW, C. F. (1964) Clinical aspects of the blood-brain barrier. Ann. Rev. Med., 15, 187-202. BITO,L. Z., BRADBURY, M. W. B., AND DAVSON, H., (1966) Factors affecting the distribution of iodide and bromide in the central nervous system. J. Physiol., 185, 323-354.

POSTMORTEM MUTLIPLE SCLEROSIS W H I T E MATTER

52 1

BRIGHTMAN, M. W. (1965) The distribution within the brain of ferritin injected into cerebrospinal fluid compartments. 11. Parenchymal distribution. Amer. J. Anat.. 117, 193-219. BRODIE, B. B. AND FRIEDMAN, M. M. (1938) The determination of bromide in tissues and biological fluids. J. bid. Chem., 124, 51 1-518. BROMAN, T. ( I 946) Blood-brain barrier damage in multiple sclerosis. Supravital test-observations. Acta Neurol. Scand., 40, Suppl. 10, 21-24. -, (1947) Supravital analysis of disorders in the cerebral vascular permeability. 11. Two cases of multiple sclerosis. Acta Psychiat. Neurol., Suppl., 26, 58-71. COTLOVE, E., TRANTHAM, H. V. AND BOWMAN, R.L. (1958) An instrument and method for automatic, rapid, accurate, and sensitive titration of chloride in biologic samples. J. Lab. Clin. Med., 51, 461-468.

EVELYN, K. A. A N D MALLOY, H. T. (1938) Microdetermination of oxyhemoglobin, methemoglobin, and sulfhemoglobin in a single sample of blood, J . biol. Chem., 126, 655-662. GONSEWE,R. AND ANDRE-BALISAUX, G. (1965) La permdabilite des vaisseaux cbrebraux. IV. Etude des lesions de la barriere hemato-encdphalique dans la scldrose en plaques. Acta Neurol. Psychiat. Belg., 65, 19-34. GORDON,M. W. AND NURNBERGER, J. I. (1956) Estimation of whole blood in tissue homogenates. J. Histocheni. Cytochem., 4, 84-85. HAERER, A. F., TOURTELLOTTE, W. W.,RICHARD, K. A., GUSTAFSON, G. M. AND BRYAN,E.R., (1964) A study of the blood-cerebrospinal fluid-brain barrier in multiple sclerosis. I. Blood-cerebrospinal fluid barrier to sodium bromide. Neurology, 14. 345-354. IBRAHIM, M. Z. M. (1965) Neuroglia and demyelination, Neurohistochemistry, C. W. M. Adams (Ed.), Amsterdam, Elsevier, pp. 454464. KABAT,E. A.. GLUSMAN, M. AND KNAUB, V. (1948) Quantitative estimation of the albumin and gamma globulin in normal and pathologic cerebrospinal fluid by immunochemical methods. Amer. J. Med., 4, 653-663. KATZMAN, R. (1966) Effect of electrolyte disturbance on the central nervous system. Ann. Rev. Med., 17, 197-212. KISHIMOTO, Y . A N D RADIN,N. S. (1966) Determination of brain gangliosides by determination of ganglioside stearic acid. J . Lipid Res., 7 , 141-145. KISHIMOTO, Y.,RADIN,N. S., TOURTELLOTTE, W. W., PARKER,J. A. AND ITABASHI, H. H. (1967) Gangliosides and glycerophospholipids in multiple sclerosis white matter. Arch. Neurol., 16,4454. KLUVER,H. AND BARRERA, E. (1953) A method of the combined staining of cells and fibers in the nervous system. J. Neuropathol. exptl. Neurol.. 12, 400403. LOWDEN, J. A., A N D WOLFEL. S. (1964) Studies on brain gangliosides. 111. Evidence for the location of gangliosides specifically in neurones. Canad. J. Eiochern., 42, 1587-1 594. LOWENTHAL, A. (1961) Ddterminations de la teneur du systeme nerveux central en materie skhe, potassium et sodium. Chemical Pathology ofthe Nervous System. Proc. 3rd Intern. Neurochemical Symposium, Strasbourg, 1958, J. Folch-Pi (Ed.). New York, Pergamon, pp. 299-306. MASON,J. K., KLYNE,W. AND LENNOX, B. (1951) Potassium levels on the cerebrospinal fluid after death. J. Clin. Pathol., 4, 231-233. MCALPINE,D., COMPSTON, N. D. AND LUMSDEN, C. E. (1955) Multiple Sclerosis. Edinburgh and London, Livingstone. C. E. A N D ACHESON,E. D. (1965) Multiple Sclerosis-A Reappraisal. MCALPINE,D., LUMSDEN, Baltimore, Williams KK Wilkins. MCILWAIN, H. (1955) Biochemistry and the Central Nervous System. Boston, Little, Brown and CO. NAUMANN, H. N. (1958) Cerebrospinal fluid electrolytes after death. Proc. SOC.exptl. Biol. Med., 98, 16-18.

NORTON,W. T. A N D AUTILIO,L. A. (1966) The lipid composition of purified bovine brain myelin. J . Neurochem., 13, 213-222. OLDENHOF, W. H. A N D DAVSON, H. (1967) Brain extracellular space and the sink action of cerebrospinal fluid. Arch. Neurol., 17, 196-205. PALSSON,P. A., PATTISON,I. H. AND FIELD,E. J. (1966) Transmission experiments with multiple sclerosis, Slow, latent, and temperate virus infections, in D. C. Gajdusek, C. J. Giggs, Jr. and M. Alpers (Eds.). Washington, D.C., United States Government Printing Office, (pp. 49-54) Public Health Service Publication No. 1378. P ~ R I E0. R , A N D G R ~ C O I RA. E ,( 1965) Electron microscopic features of multiple sclerosis lesions. Brain, 88, 937-952.

522

W. W.T O U R T E L L O T T E

PROCKOP,D. J. AND UDENFRIEND, S. (1960) A specific method for the analysis of hydroxyproline in tissues and urine. .4tial. Biochem., 1, 228-239. SCHAIN,R. J. (1964) Cerebrospinal fluid and serum cation levels. Arch. Neurol., 11, 330-333. SIMPSON, J. F., KOKMEN, E. A N D TOURTELLOTTE, W. W. Multiple sclerosis: Identification of immunoglobulin-(; in plaques by immunofluorescence. (to be published) TOURTELLOTTE, W. W. AND PARKER, J. A. (1965) Distribution and subfractionation of immunoglobulins in patients with multiple sclerosis. Trans. Amer. Neurol. Ass., 90, 107-1 12. -, (1966a) Immunoglobulins in multiple sclerosis white matter. J . Neuropathol. expll. Neurol., 25, 167-169. -, (1966b) Multiple sclerosis: Correlation between immunoglobulin-(? in cerebrospinal fluid and brain. Science, 151, 10441046. -, (1967) Multiple sclerosis brain gamma2 globulin and albumin. Nature, 214,683-686. TOURTELLOTTE, W. W., PARKER, I. A. AND ITABASHI,H. H. (1966) Source of elevation of gamma globulin in brain from patients with multiple sclerosis. Trans. Amer. Neurol. Ass., 91, 351-352. WENDER, M. AND HIEROWSKI, M. (1960) The concentration of electrolytes in the developing nervous system with special reference to the period of myelination. J . Neurochem., 5 , 105-108.

DISCUSSION M. BRIGHTMAN: It would seem to me that much of the extracellular space, at least so identified by electronmicroscopy, is intracellular. Would not you expect to see more gliosis in the plaque? We also have thought about the possibility that the extracellular space was W. W. TOURTELOTTE: intraglial and that other “spaces” are an artifact of preparation. We have tried v x y hard to get a plaque with multiple sclerosis for electronmicroscopy study. However, to get it one hour post rnortem is really very difficult. As a matter of fact, until I can use Lowry’s techniques and get the analyses down to the level where 50 micrograms of tissue can be examined with some confidence that it is reasonably homogeneous tissue, I won’t really be able t o answer all the questions. A. LOWENTHAL: I will add two things to what Dr. Tourtellotte said. Firstly, I think that if you have a reduction of the dry weight in the plaques or in some material near the plaqu:s, then the possibility exists that the white matter of the plaques has been replaced by cells. In fact, the dry weight of the plaque in your experiments is similar to that of grey matter. If you have a glial reaction in those plaques, which means that there is grey matter in such plaques, then you can explain a lot of your results. The second point concerns the results showing that the lesion in multiple sclerosis can be found not only at the level of the plaque but also in some other tissue in white matter, appearing as normal white matter. This has been shown repeatedly during the last years, and I think that your results are important.

W. W. TOURTELOTTE: I would like to speak about both of these points, because we are giving them a lot of thought, and to tell you about the controls we have carried out. All the material that we dissect for a plaque is a punch biopsy made with a 16 gauge needle, and there is no chance at all that grey matter and plaque material could have been analyzed together in this. We feel very safe about this. The fact that the dry weight of plaque material in grey matter is similar to that of white matter can be explained on the basis of biological variation. So you have control punch biopsies in order to make sure that you are in the spotted area. As far as normal-appearing white matter is concerned, I think we are all aware of the fact that by gross dissection a lot of things are thrown away. These plaques are very small, so I think that with Lowry’s technique and a guide slide to tell exactly what is being dissected out, we will be able to have more confidence in what normal-appearing white matter is truly illustrating. A. LOWENTHAL: You have no histological control of that normal-appearing white matter?

Yes, every bit of our material is obtained by punch biopsy. We have stained the W. W. TOURTELOTTE: material with Luxol-fast blue, and we were able to show the presence of myelin sheath.

POSTMORTEM M U L T I P L E SCLEROSIS W H I T E MATTER

523

I. KLATZO:Dr. Tourtelotte’s finding that in plaques you have primarily an increase in y-globulin, whereas the albumin seems to be the same as in normal white matter, is slightlypuzzling tome. I think that Bronian was the first who actually demonstrated the vascular permeability disturbances in multiple sclerosis plaques. I would like to ask Dr. Tourtelotte: does he feel that there are no permeability disturbances in such plaques? Otherwise. would you have to assume that this y-globulin is produced by the brain tissue itself? If you have a vascular permeability disturbance, the albumin definitely spreads much more easily then the 1’-globulin. In multiple sclerosis you have two possibilities: either the serum proteins leave the blood vessels around the plaques or in the plaque, or they permeate from the cerebral spinal fluid. In both instances I think there is enough experimental evidence to indicate that albumin will go in to a much greater extent. Thus, for me it is very puzzling if you have to assume that the 7-globulin is produced in the multiple sclerosis plaques. The one thing which would be a source of such a production would be inflammatory cells, especially the plasma cells. However, it is well known that in multiple sclerosis plaques this inflammatory reaction is coincidential, and usually not terribly impressive. The brief answer to these complex questions goes something like this: We have W. W. TOURTELOTTE: inferred that because the albumin is the same in the different materials we have looked at, that this is evidence that we don’t have a marked leak from the vascular space into the brain tissue. Now I am sure that this is going to depend on the activity of the plaque, because there are many plaques that are dead, so to speak. The possibility that there could be a leak at some time in the course of an active lesion is certainly a reality. In that case the albumin may be moving into the tissue from the blood vascular space. And if it moves, the ;!-globulin could be expected to move. To bring up the question of what Broman showed when he perfused the brains with trypan blue - i t is true that some of his plaques did stain with trypan blue. That was his first publication. His second publication showed that this was a special type of diffusion from the vascular space to the brain. I t was different from the diffusion of trypan blue in mechanical lesions of the brain, where there is a true breaking of the blood vessels, after which the trypan blue moves out. He is very unspecific about the special type of moving out of trypan blue from the vascular space to the brain. Maybe somebody from his laboratory could tell me more what it means. As far as the sources of y-globulin are concerned, we are postulating two reasonable sources: 1 think that where there is active demyelination it is possible that the microglia are making y-globulin. This is a new thought because nobody has rcally been able to establish this. It is more reasonable to think in ternis of perivascular cuffs, which are very common in this disorder. In the perivascular cuffs, it is true, there are no plasma cells, but the latest thought is that you don’t need plasma cells to make jqlobulin anymore; it can be made by small lymphocytes. About the problem that albumin moves through the brain more easily than y-globulin: I have learned a lot at this conference, and I think maybe this will help us to design some experiments to make intracerebral injections of albumin and y-globulin in the guinea pig brain, using different antibodies, and to follow the guinea pig-y-globulin, as well as the human y-globulin we will inject

B. JOHANSSON:I just want to add something to what Dr. Tourtelotte said yesterday about a case of multiple sclerosis and jaundice. Recently we had a similar case. We looked at very thin sections with the fluorcscence microscope. The striking thing was the very big difference in the plaques. In severeal of the plaques you could see the staining of the vascular wall. For example, in one plaque thecentral vein stained very nicely and in others you could definitely observe that the bilirubin had left the vascular wall. I cannot define the relationship between the age of the plaque and the difference in tracers, because the histopathological sections are not yet ready.

D. B. TOWER: I have two points. One is related to the sodium values that were shown. I think Dr. Tourtelotte will agree with me and I think it should be emphasized that one cannot necessarily draw conclusions from the values, since they are postmortem. Differences which depended upon pump-niechanisms would be obliterated at this time, so that it is a little confusing to try to draw much of a conclusion from such data. And in regard to the ;,-globulin, it has been shown in the literature that there is an immunologically distinct component of the ;!-globulin complex in the cerebrospinal fluid, a component which is not found in the gcneral circulation. I think this factor should be recognized and perhaps investigated in this type of study.

524

W. W. T O U R T E L L O T T E

W. W. TOURTELOTTE: I am certainly more aware than anybody here that we are dealing with postmortem material. Our efforts to get good cerebral biopsies have already been discussed this morning. It is very difficult, and as far as the specific y-globulin is concerned, this appears in normal CSF, but is no worse in multiple sclerosis CSF. As a matter of fact, the y-globulin fraction still reacts with regular y-globulin reagent. It is mostly an electrophoretic difference: the antibody is immunologically still like all the others, but yet it has a little different mobility. 0. STEINWALL: I will restrict myself to a comment on the trypan blue and postmortem perfusion which Broman performed a number of years ago. He used trypan blue in saline and found out later that this reacts with the tissue and enters the tissue in quite another way than if you use a tracer bound to a protein. In fact it moves, as far as we could detect, much slower. We have just taken it up again and will try to use it even in human material in which we can compare this new technique with what you have found. G. LEVI:Just a brief answer to Dr. Lowenthal and Dr. Tourtelotte. Amaducci showed in plaques an increase in the number of glial cells, and a corresponding increase in proteins.

M. BRIGHTMAN: I would like to make an additional remark about the dynamics of spaces in the brain. We don’t know what the viscosity is of the fluid in these narrow clefts, or whether a peristaltic-like movement actually does occur in vivo, although one can show movement in at least one cell; the oligodendrocyte, in tissue culture. Such effects as diffusion, versus a passive bulk movement between the clefts, represent just a few of the unknown factors in fluid movement, so what could be said here would be pure guess. I don’t know how you could hope to get some pictorial “static evidence” of peristaltic-like membranes, even if they work in phase. H. M. PAPPIUS: I would like to mention one of my own experiments. I was asked to make an investigation of the effect of intracranial infusion of methotrexate. We studied the water content and the electrolytes in the brains of monkeys that had had long methotrexate infusions. These animals were not killed at any specific time, but they died, and the analysis was carried out at variable times after death. I could correlate very well the water content of both the white matter and the cortex and the increase in sodium with the time that elapsed between the estimated time of death of the animal and the time that the tissue was taken out of the cranium. I think the maximum time in our case was six hours. By twelve hours you may be reaching a point where all the sodium that could get in has gotten in, and it says very little about the state of affairs at time of death. A. LOWENTHAL: I want to say something about the comment of Dr. Klatzo: In soluble protein extracts of white matter of multiple sclerosis patients, there is no increase of albumin, but a decrease of y-globulin. Dr. Tower was speaking of a specific y-globulin in multiple sclerosis, bound or free. This is still a controversial matter. I don’t think we can accept the evidence about a specific y-globulin in normal CSF or in CSF of multiple sclerosis patients. This problem could only be solved by isolating such a y-globulin and studying it immunologically. As to the question of the electrolytes, I think that there are changes in electrolyte distribution in the brain that occur post mortem. Thus, when one compares biopsy material and postmortem material there are differences.

H. M. PAPPIUS: I found no difference between normal animals and methotrexate perfused animals. The increase in water and sodium occurred in both, depending on the time that had elapsed since the death of the animal. A. LOWENTHAL: Yes, but in multiple sclerosis there are differences.

I. KLATZO:Just for the record: about the pulsation and possible movements in the CSF. It is true that it would be very difficult to show pulsation by electronmicroscopy. However, we have the evidence, from electronmicroscopy, that there is a movement of cell membrances; I am referring specifically to the evidence of pinocytosis that Dr. Brightman so beautifully showed in many of his photographs. Now pinocytosis implies invagination of cell membranes; in other words, there must be a movement incorporating these invaginated vacuoles.

POSTMORTEM M U L T I P L E SCLEROSIS WHITE MATTER

525

K. A. C. ELLIOIT:Dr. Tourtelotte, I think this is the end of the discussion of your paper. Would you like to tidy anything up?

W. W. TOURTELOTTE: No, I would only like to say that I think that we are certainly the first to appreciate that this is postmortem tissue. I have the feeling that only micro-determinations could settle the problem that all plaques are not the same. Then we may be able to find a quantitative increase in perivascular cuffs, which might again support the idea that the brain tissue now has a reticuloendothelium type of cell in it, and can have multifocal areas of generation of y-globulin. Then we have the problem of whether this macro-molecule can move through the brain. The theory that albumin itself is metabolized and moves back into the blood vessel is certainly an interesting idea. I don’t know how to test it right now; it is a new idea; but we certainly should test that too.