Brain Research, 122 (1977) 95-112

95

© Elsevier/North-Holland BiomedicalPress, Amsterdam - Printed in The Netherlands

SUBCELLULAR DISTRIBUTION OF RAT BRAIN CORTEX HIGHAFFINITY, SODIUM-DEPENDENT, GLYCINE TRANSPORT SITES

FERNANDO VALDI~S, CARLOS MUI~OZ, ALFREDO FERIA-VELASCO and FERNANDO ORREGO Depto. de Fisiologia y Biofisica, Universidad de Chile, Casilla 6542, Santiago-4 (Chile) Instituto Nacional de Cardiologia, Mexico 7, D.F. and Unidad de Investigacidn Biomedica, I.M.S.S., Guadalajara (Mexico)

(Accepted June 2nd, 1976)

SUMMARY The subcellular distribution of the membrane components, present in rat brain cortex homogenates, that interact with glycine in the presence of sodiumions was studied. The distribution in the primary fractions, as per cent of total binding in the homogenate, was: P1 ('nuclear'), 58 ~o; P2 (large granule), 39 ~ ; Pa (microsomal),

2Yoo. Of the subfractions obtained by centrifuging P1 in a linear 0.32-1.5 M sucrose gradient, only the lighter fraction (PI-III) formed by large myelin fragments was enriched in specific binding activity with respect to P1. The pellet formed by purified nuclei had negligible binding, and fractions of intermediate density had a lower activity than P1. Transient exposure of PI-III to 1.5 M sucrose did not diminish its binding ability. Similarly, in the subfractions obtained by centrifuging P1 in a discontinuous sucrose gradient, only the least dense one, P1-A, that is formed exclusively by large myelin fragments, was enriched with respect to P1. The electron microscopy of these fractions is presented. The Pz subfractions, obtained in a linear 2-18 ~o Ficoll gradient, had the following sodium-dependent activity (counts/min/mg protein, fractions being in the order of decreasing density): pellet, 0; P2-I, 0; P2-II, 450; P~-III, 1770; P2-IV, 4130; unfractionated P2, 880; P2-IV, the least dense fraction being composed mainly of myelin. With P2 subfractions obtained in a discontinuous sucrose gradient (0.32, 0.8 and 1.2 M sucrose layers), it was also found that sodium-dependent glycine binding was only enriched, with respect to P2, in the myelin fraction P2-A. Glycine binding to purified brain cortex myelin was also found to be very high, while binding to non-myelin membranes, obtained during the purification procedure, was only 0-7 ~o of that seen with myelin. These results suggest that high-affinity glycine binding is located in myelin proper, and possibly also in some other glial plasma membranes, but not in nuclei, mito-

96 chondria, endoplasmic reticulum or synaptosomes. The relevance of these findings for interpreting previous reports on high-affinity glycine transport in the central nervous system is analyzed.

INTRODUCTION

Amino acid transport in the CNS has been extensively studied in tissue slices4,16, 17,20,24, synaptosomesa,6,22,29, isolated glial and neuronal cellslO,11,a2 and other preparations2, aS. Recently, it has been shown a6 that glycine interacts with osmotically shocked membrane fragments, present in the large granule (P2) fraction of a rat brain cortex homogenate, in a manner that closely resembles glycine transport by cellcontaining preparations, with respect to kinetics (a high-affinity binding, i.e. Km == 40/~M, was demonstrated), dependency on sodium ions and temperature, amino acid specificity and occurrence of homo- and heteroexchange reactions. The present report is concerned with the subcellular distribution of such transport sites. In addition, the 'nuclear' (P1) fraction of cerebral cortex homogenates, that has been much less studied than the 'large-granule' (P2) fraction, and for which only a fragmentary description is availableS,9,15, is further characterized. The results obtained suggest an alternative interpretation for the discrepancies that exist regarding the presence or absence of high-affinity glycine transport in the cerebral cortex, that has been shown to occur by Peterson and Raghupathy 29 and by us 36, but has not been found by other authors 6, 17,22.

METHODS

Cerebral cortices of 200-250 g Sprague-Dawley rats were dissected following the procedure of Whittaker and Sheridan 4°. Homogenization and preparation of the primary fractions P1 ('nuclear'), P2 (large granule), Pz (microsomal) and Sa (supernatant) was done according to Gray and Whittaker 9, except that the P1 fraction was washed once. Centrifugations were performed at 4 °C in a Sorvall RC-2B or in a Beckman L Centrifuge. The P1 fraction was analyzed by layering 5 ml of the P1 pellet resuspended in 0.32 M sucrose (2.5 ml per g original weight), over a linear 0.32-1.5 M sucrose gradient and centrifuging in a Beckman SW 25.2 rotor at 25,000 rev./min for 1 h. All further subfractions were obtained with this type of rotor. The P1 fraction was also subfractionated, on a discontinuous sucrose gradient, by the procedure of Eichberg et al. 8. When the P1 fraction was studied, the animals were anesthetized with ether and perfused with 0.15 M NaC1 at 35 °C, via the left heart ventricle, until the fluid flowing out of the cut right ventricle appeared blood-free. The animals were then sacrificed as usual. This procedure completely exsanguinated the brains. P2 was subfractionated by layering 5 ml of the pellet resuspended in 0.32 M sucrose on a linear 2-18 ~ Ficoll gradient made in 0.32 M sucrose 1,7, and centrifuging at 25,000 rev./min for 2 h. P2 was also analyzed by the discontinuous sucrose gradient

97 procedure of Whittaker a9, namely, 5 ml of P2 in 0.32 M sucrose was layered on a twostep gradient made of 10 ml of 0.8 M sucrose on top of 10 ml 1.2 M sucrose, and centrifuged at 25,000 rev./min for 2 h. All subfractions were recovered by means of a Beckman fraction recovery system, to which air pressure was applied with a Sigmamotor pump. Liver mitochondria were obtained from a 10 ~ liver homogenate made in 0.32 M sucrose, centrifuged first at 800 × g for 15 min and the supernatant at 10,000 × g for 20 min. The pellet obtained from this second centrifugation was resuspended in 0.32 M sucrose and centrifuged again under the same conditions. This mitochondrial pellet was further processed for measuring glycine binding in the same way as the brain cortex membranes. Cerebral cortex myelin was purified by the method of Norton and Poduslo 25, scaled down for the SW 25.2 rotor. The subcellular fractions were fixed in 2.5 70 glutaraldehyde, 0.1 M sodium cacodylate buffer, pH 7.4, for 1 h at 4 °C al, and postfixed in 1 ~o osmium tetroxide, 0.1 M cacodylate, pH 7.4, also for 1 h at 4 °C z6. After dehydration in graded ethanols, the samples were embedded in Epon-812 for 24 h at 60 °C z3. Thin sections in the silver color range 27 were obtained with a Porter-Blum MT-2 ultramicrotome and stained with uranyl acetate 34 and lead citrate 37. They were observed in a Philips EM-300 or in a Zeiss EM-10 electron microscope. Hypoosmotic treatment of the membranes and the measurement of glycine binding were performed as indicated previously 36. Binding in the presence and absence of sodium, as well as blank values, were performed in triplicate to quintuplicate. All binding measurements were performed for 10 min, except when indicated.

6O

m

g ~

.-.a

40

"~'"

1800 --

1600

ce A

J2oo

~

~.

20 400 n

~ (/)

PI

P2

P3

$3

FRACTION Fig. ]. Distribution o f sodium-dependent glycine binding in primary subcdlular fractions. The nomenclature and method of obtaining the fractions is that of Gray and Whittaker 9. Open bars represent

the percentage of total sodium-dependent binding present in each fraction. Hatched bars indicate specific activity of sodium dependent glycine binding. Each bar represents quintuplicate measurements (single experiment).

98

4000 A

0

m 3000 E E

7

2000

IO00

p,

P,--_I

Pl -I~

FRACTION

Fig. 2. Sodium-dependent glycine binding in P1 subfractions obtained in a linear 0.32-1.5 M sucrose gradient. P1 refers to the unfractionated parent fraction; PI-I is the fraction that equilibrates between 1.15 and 1.5 M sucrose, plus the pellet; PI-II represents material that equilibrates between 0.74 and 1.15 M sucrose, and P~-III is the ivory-white material lighter than 0.74 M sucrose. Results of 3 experiments, each performed in triplicate.

"i 3 0 0 0

.F E O.

2000

o z 21 0al i 000

.J

~

0

m

Pt

PeA

Pt-B

PcC

Fig. 3. Sodium-dependent glycine binding to P1 subfractions obtained in a discontinuous sucrose gradient. Subfractions were obtained by the method of Eichberg et al.L P1-A and P~-B correspond to the fractions that equilibrate at the 0.32-0.8, and 0.8-1.2 M sucrose interphases, respectively. P,-C is the pellet of material denser than 1.2 M sucrose. Results are mean values of 3 experiments performed in triplicate.

99 I

I

I

I

I

60

/ / I J I

i

0

I

i

I

II

40 TIM E

(lice)

Fig. 4. Time-course of sodium-dependent glycine binding to P1-A. Each point is the mean -4- 1 S.D. (5 measurements each). 1 pmole = 5.7 counts/min.

Fig. 5. Fraction Px-C. Nucleus lacking its envelope (upper right corner), chromatin aggregates, swollen mitochondria and numerous empty profiles and small electron-dense particles are seen. Several myelin fragments, probably attached to other structures, are also present. Such myelin fragments are scarce, and this micrograph is not representative in that respect.

100 RESULTS The distribution of sodium-dependent glycine binding in the primary subcellular fractions (Fig. 1), indicated that the highest binding, both with respect to specific activity and to per cent of total binding, was located in P1. The activity in P9 was noticeably lower, and that in the other fractions rather negligible.

Composition and binding activity of P1 subfractions The P1 fraction was subfractionated by two different density gradient procedures. In the first one, P1 was centrifuged through a linear 0.32-1.5 M sucrose gradient for l h, and three fractions separated. Sodium-dependent glycine binding activity was considerably enriched, with respect to the parent fraction Pl, in the least dense fraction P1-11t (Fig. 2), while the intermediate fraction PI-II had a lower binding activity than P1, and this was still further reduced in the denser fraction PI-I. When PI-III was centrifuged in 0.32 M sucrose (1500 × g, 10 min), the pellet formed resuspended in 1.5 M sucrose and, after 1 h the sucrose diluted again to 0.32 M, recentrifuged and resuspended in distilled water, its glycine binding capacity remained identical to that of another Pa-III aliquot that was processed in parallel, but not exposed to hypertonic sucrose.

Fig. 6. Fraction P1-B. Empty membrane profiles, some of them concentric, and abundant circular or elongated electron-dense bodies are present.

101 This indicated that concentrated sucrose did not inhibit binding, and was, therefore, not responsible for the low binding seen in the more dense fractions. The second fractionation procedure utilized, was that of Eichberg et al. s, in which P1 is centrifuged, for 2 h, through a discontinuous sucrose gradient. Glycine binding to the different subfractions (Fig. 3) also revealed a marked enrichment in the fraction of lower density (P1-A), while a much lower binding was seen in the other two subfractions. The time-course of binding of the Pz-A subfraction was also studied (Fig. 4). During the first 20 sec a very rapid rate could be seen, followed by a linear phase that started at 20 sec, and lasted, at least, until 100 sec. The electron microscopic study of the P1 subfractions revealed the following. Pz-C, the pellet of material denser than 1.2 M sucrose, was composed of a variety of different elements (Fig. 5), the most characteristic of which were disrupted nuclei, mainly present as chromatin aggregates, with the nuclear membranes generally absent. A few mitochondria and shrinked synaptosomes could also be recognized. Very dense bodies and closed empty vesicular profiles, whose origin could not be identified, were the most abundant elements. Broken capillary fragments were seen infrequently, and red blood cells were absent. The latter indicates that the perfusion had been effective. Myelin fragments were seen with some frequency. Since myelin

Fig. 7. Fraction P1-A. Only empty myelin fragments, with a variable degree of membrane disruption and vesiculation, may be seen.

102

Fig. 8. Higher magnification of one of the P1-A myelin fragments. Myelin periodicity (upper right quadrant) and abundant membrane vesiculation is present. has a density much lower than 1.2 M sucrose, these fragments probably represent axons attached to other components of higher density, from which they have not separated during homogenization. P1-B (Fig. 6) is a more homogenous fraction, mainly composed of closed empty membranous profiles, some of them concentric, and abundant electron-dense structures of elongated or circular profile. Some of these electron-dense bodies could be derived from the very long mitochondria frequently seen in dendrites and axons of intact tissue ~8, that are compressed and fragmented during homogenization. A few relatively compact myelin fragments could also be observed. Fraction Px-A, the ivory-colored band that equilibrates between 0.32 and 0.8 M sucrose, was found to consist of a pure population of large myelin fragments, some of up to 15 #m in diameter (Fig. 7, lower left corner). No axoplasm could be detected, and the myelin unit membranes were largely separated from one another and partially disrupted. A considerable degree of membrane vesiculation was also apparent (Figs. 7 and 8). Such elements give this fraction its characteristic lace-like aspect. These structural features were retained when Px-A was submitted to osmotic shock (Fig. 9), under the same conditions as those used prior to the binding assay 36. Our P~-A fraction resembles the equivalent one N-A, obtained from the cerebellum by Tapia el al. 3z,

103

Fig. 9. OsmoticallyshockedP1-Amyelinfragment. Fraction P1-Awas isolatedand osmoticallyshocked under the same conditions used for the bindingstudiess6. Membrane disruption and vesiculation,similar to that seenin unshockedpreparations,is present. with the exception that in P1-A no axoplasm is found, and that the myelin layers are more disrupted. Both these findings are probably a consequence of the more vigorous homogenization conditions of the Gray and Whittaker procedure 9.

Glycine binding to Pz subfractions When sodium-dependent glycine binding to the P2 subfractions, obtained in a linear 2 - 1 8 ~ Ficoll gradient, made in 0.32 M sucrose, was studied (Fig. 10), no binding was detected in the pellet or in P2-I, where most of the free mitochondria sediment 1. Similarly, liver mitochondria are devoid of sodium-dependent glycine binding (not shown). Binding only appeared in fractions of lower equilibrium density, starting with Pz-II, where specific binding activity was lower than in unfractionated P2. Binding activity became enriched, with respect to P2, in P2-III and, especially, in P2-IV, the ivory-white layer on top of the gradient. Synaptosomes are present mainly in fractions I, II and 1111,18, while myelin, that is the principal constituent of fraction IV, can be found, though in lesser amount, in fractions of higher density, including those that equilibrate in 13 ~ Ficoll, i.e. Pe-II is (and our unpublished results). Glycine binding was also studied using the Pz subfractions obtained on a dis-

104

~--I ---~I

~ ~H~-~--~

c 4000

3000

E E o

2000

C 0

~ooo 0

PiP ~-I

~-~ ~-x

~-~

Fraction

Fig. 10. Distribution of sodium-dependent glycine binding in P2 subfractions. These were obtained in a linear 2-18 ~ Ficoll gradient made in 0.32 M sucrose. P2 represents the unfractionated starting material; P2-P, the pellet; P2-I, subfractions that equilibrate between 12.6 and 1 8 ~ Ficoll; P~-II, those between 7.3 and 12.6~; P2-III those between 4.9 and 7.3 ~, and P~-IV, those lighter than 4 . 9 ~ Ficoll. The position of the fractions in the tubes is indicated in the upper drawing, where the relative amounts of material are depicted by means of stippling intensity. From the pellet to fraction III the distribution of material is continuous throughout the gradient, while the ivory-white fraction IV is separated from denser fractions by a layer of fluid devoid of substances in suspension. Results of 4 experiments, each performed in triplicate.

continuous sucrose gradient following Whittaker a9 (Fig. 11). Once again binding was highly enriched in the myelin fraction P1-A; relatively low in Pe-B, where most of the synaptosomes equilibrate, and virtually absent in the mitochondrial fraction P2-C. Purified P2-A was further fractionated, in another discontinuous sucrose gradient (Fig. 11), into two subfractions: P2-A-II, that equilibrates in the interphase between 0.5 and 0.65 M sucrose, and a denser P2-A-III (0.65-0.8 M sucrose interphase). Sodiumdependent glycine binding was slightly enriched in P2-A-II, with respect to P2-A, while that in P2-A-III was noticeably lower, but still quite higher than in P2. We have found that P2-A is a highly purified myelin preparation, composed of fragments of about 0.5-5 #m in diameter, a few of which contain axoplasm. A small contamination with other non-myelin membranes was also present (not shown). Osmotically shocked P2-A myelin also presented a considerable degree of vesiculation (Fig. 12).

Glyeine binding to purified myelin Although the association of sodium-dependent glycine binding with myelinrich fractions, present both in P1 and P~, was evident, it seemed desirable to measure

105

A

c o

3000

Q.

E

O.

2000 Q Z 0

>.J

t000

P2

P2"A

P2"A'# P2-Ag[

P2-B

P2-C

Fig. 11. Sodium-dependent glycine binding to P2 subfractions obtained in a discontinuous sucrose gradient. P~ was subfractionated by the Whittaker procedureaL P2-A and P2-B are the fractions that equilibrate at the 0.32-0.8 and 0.8-1.2 M sucrose interphases, respectively. P~-Cis the pellet denser than 1.2 M sucrose. Results of P2-A, B and C are mean values of 3 experiments (3-5 measurements per fraction, in each of the experiments). In two other experiments, P1-A was isolated, dispersed in 0.32 M sucrose, and placed on a discontinuous gradient formed by layers (8 ml each) of 0.5, 0.65 and 0.8 M sucrose. Centrifugation was in the SW 25.2 rotor at 23,000 rev./min for 2 h. P~-A-II and P2-A-III refer to the bands present at the 0.5-0.65, and 0.65-0.8 M sucrose interphases, respectively. A lesser amount of material was present throughout the 0.65 M sucrose layer. Only a very faint white line of material was seen at the 0.32-0.5 M sucrose interphase, and none in the 0.5 or 0.8 M sucrose layers. No pellet was formed either. Pz-A-1I and III bars represent mean of 5 measurements in each of 2 independent experiments. such binding to myelin purified by procedures specifically designed for such a purpose. We proceeded to purify brain cortex myelin by the procedure o f N o r t o n and Poduslo2L In the final fraction obtained, we could find no evidence o f contamination by other membranes (Fig. 13). Glycine binding to this pure myelin was measured for 1 and 10 min (Table I). In parallel, binding to the non-myelin membranes obtained in the procedure, was also measured. A very high binding to myelin was again present, while that to the myelin-free membranes was only 0-7 ~o to that seen with myelin. The timecourse o f binding to this pure myelin was unusually rapid (Fig. 14). N o true initial velocity could be measured by this procedure, and binding rapidly a p p r o a c h e d equilibrium. This is c o n c o r d a n t with the relatively small differences in binding seen between 1 and 10 min. DISCUSSION

Location of transport sites The highest specific activity o f sodium-dependent, high-affinity, glycine binding was f o u n d associated with the myelin-rich fractions present in P1 and P2, obtained, in each case, by two different fractionation procedures. The same was true when purified

106

Fig. 12. Osmotically shocked P2-A myelin fragment. Fraction P2-A was isolated and submitted to osmotic shock, as performed prior to binding studies:~. This micrograph is selected, and shows an unusual degree of membrane vesiculation. myelin was used. This leads to the first question: are these glycine transport sites associated with myelin proper, or is it with some other membrane components that contaminate the myelin-rich fractions? The results presented clearly indicate that the glycine sites are not associated with purified nuclei, free mitochondria, including those from liver, or fragmented endoplasmic reticulum, that sediments in the P3 fraction 9. The question thus becomes restricted to what extent are the myelin-rich fractions contaminated by synaptosomes, that are fairly active in amino acid transport, or by other neuronal or glial plasmalemmae. In the primary fractions, synaptosomes are mainly concentrated in P2, with some of the smaller ones also present in P3, and very few in P19. This shows no correlation with glycine binding that is highest in P1, intermediate in P2 and very low in P3. Similarly, in the P2 subfractions obtained in a Ficoll gradient, synaptosomes are present in P2-I, Pz-lI and Pz-IlI, and virtually absent from P2-IV 1,18 (and unpublished), while glycine binding is absent from P2-I, and is enriched, with respect to P2, only in P2-III and P2-IV. In the subfractions of Pz obtained in a discontinuous sucrose gradient, synaptosomes are mainly located in P2-B 9, where glycine binding is less than in unfractionated P2. The binding present in P2-B could, in fact, be explained by a contamination of about

107

Fig. 13. Myelin purified by the procedure of Norton and Poduslo zS. Myelin fragments of varying degrees of compactness may be seen. Membrane vesiculation is relatively scarce.

TABLE I

Sodium-dependent glycine binding to purified myelin and to non-myelin membranes Myelin was purified by the method of Norton and Poduslo 2s. Non-myelin membranes refer to those obtained in the pellet in step 1 of the procedure. These membranes received the same number of osmotic shocks as myelin. Results of two experiments, each performed in triplicate. Mean deviations were 4.2 in A, and 11.6 ~ in B.

Glycine bound (counts/min/mg protein) (A) 1-min binding Myelin Non-myelin membranes (B) 10 rain binding Myelin Non-myelin membranes

2756 206 4070 0

108

-'2 o

40

E C~ Z 0 m )-J (.9

20

i

I

I

I

40

80 TIME

I

I

t20

(sec)

Fig. 14. Time-course of sodium-dependent glycine binding to purified myelin. Myelin was purified by the procedure of Norton and Poduslo2~. Values are means ± S.D. (N = 5 per point). 5 ~o, on a protein basis, with myelin. Binding was highest in P~-A, where synaptosomes are virtually absent 9. This indicates that essentially no correlation exists between glycine binding and presence of synaptosomes. The same seems to hold true for nonmyelinated axon fragments, that although less well studied than synaptosomes are known to be concentrated in P2-B 21. The distribution of other membrane fractions is much less understood, since numerous membrane profiles of unknown origin are present in a number of the P1 and P2 subfractions. We are inclined to believe, however, that the glycine transport sites are associated with myelin proper because of the following. (1) Glycine binding and presence of myelin correlate extremely well in all the primary and secondary fractions, obtained by different procedures. (2) No contamination was observed by us in the P1-A or in the myelin purified by the procedure of Norton and Poduslo2L However, a very small degree of contamination could have escaped detection because of sampling procedures. Such contamination was found in P2-A, but when this was subfractionated, specific binding activity was reduced in the denser myelin subfraction, Pz-A-III, where the contaminants are known to sediment 42, while binding was increased in the lighter subfraction, that represents a more pure specimen 42. (3) If glycine binding were due to a contaminant present in the myelin fractions, two situations could occur: first, part of the contaminant would be present in the myelin fractions, but most of it would appear in a different fraction. This should lead to a bimodal distribution of glycine binding. It is clear that both in the P1 and P2 subfractions glycine binding shows an unimodal distribution, with its peak in the least dense fraction. The second situation is that the glycine sites were entirely present in a fraction that coincides in buoyant

109 density with myelin. However, since several of the myelin fractions are highly pure, this isopycnic contaminant could only be present in very small amounts, and an enormous capacity to transport amino acids should have to be postulated for it. This is unwarranted. (4). In the fractionation procedures utilized, myelin is separated from other fractions because of two different properties: during differential centrifugation, myelin becomes distributed because of size; the larger fragments sedimenting in P1 and the smaller ones in P2. However, when P1 or P2 are subfractionated in density gradients, myelin now separates from other components because of differences in buoyant density. Myelin purification is achieved in the Norton and Poduslo 25procedure by taking advantage of both size and density properties. The present results have indicated that the specific binding activity in all the presumed pure myelin fractions, P1-A, PI-III, Pz-IV and purified myelin are essentially similar. If the binding in all these different myelin fragments were due to a contaminant, such a contaminant should coincide with myelin not only in density, but also in size distribution. Such an alternative is extremely unlikely. This similarity in binding found in the different myelin-rich fractions also suggests that the binding process is the same in all these fractions. However, the more rapid binding time-course seen in P1-A and purified myelin, relative to that in P236, does not allow initial velocity measurements, and, thus, a detailed kinetic analysis which could prove this. The differences in binding velocity between fractions may only represent differences in the physical state of myelin, as is evident in the electron micrographs, and not in the transport process as such. The arguments given above suggest that the presence of glycine transport sites in myelin is likely to represent a real phenomenon. This raises a related but different question: are these high-affinity sites present only in myelin? We believe this cannot be answered at present, because of the complexity of the fractions of higher density than myelin, were glycine binding is also found, and where myelin is present together with a variety of other unidentified membrane fractions. However, it would be rather unusual that oligodendroglial cells possessed such transport sites in part of their plasma membranes, (i.e. those that form the myelin envelope), but not in the rest of its surface.

Presence of high-affinity glycine transport in brain cortex The occurrence of high-affinity (Kin 20-40/zM) glycine transport in the rat cerebral cortex has been described by Peterson and Raghupathy 29 and by us 36, but has not been found by others6,17, ez. The reasons for this discrepancy, we believe, are multiple. It has by now become recognized that all amino acid transport in rodent brains has a marked requirement for sodium ions 20, and that high-affinity amino acid transport cannot be seen when substantial amounts of endogenous amino acids are present 6. Such a situation occurs when crude homogenates are used, and probably also when transport is studied in chopped tissue preparations, that loose into the incubation medium a high proportion of their endogenous amino acidsL It is also possible that in brain slices, the presence of large unstirred layers may artifactually increase transport Kms 41. When purified synaptosomes or membrane fragments are used, the presence of unstirred layers or of released amino acids, that do not penetrate the

110 density gradients, should not greatly interfere. Bennett et al. 6, using purified brain cortex synaptosomes (the PzB fraction of Whittaker), have found high-affinity transport for most amino acids, with the exception of glycine. Peterson and Raghupathy 29, however, using synaptosomes purified by a different procedure 19, have clearly shown that high-affinity glycine transport is present in a fraction that equilibrates between 3 and 13 ~ Ficotl. We believe this apparent contradiction can be solved in the light of the present results. The finding of Bennett et al. ~ that no high-affinity binding is present in synaptosomes is in agreement with our results, that also indicate its absence in these particles. The synaptosomal preparation of Peterson and Raghupathy 29, however, was purified by sedimentation through 3 %o Ficoll, 0.32 M sucrose 19. Myelin is also known to sediment, in large amounts, through that type of layer ls,3s. We, therefore, suggest that in this latter preparation, high-affinity glycine binding was present in myelin and not in synaptosomes. In the same manner, we suggest that high-affinity glycine transport, that is conspicuous in slices4,17,24 and synaptosomes 2~ derived from the spinal cord, may only be a reflection of the much higher myelin content of this CNS region, compared with cerebral cortex. Such higher myelin content should contaminate spinal cord synaptosomes to a considerable degree. In this respect it is of interest that, following incubation of spinal cord slices with [aH]glycine, an intense autoradiographic activity is seen inside some small myelinated axons and glial cells12,13. Also, after incubating a spinal cord homogenate with [14C]glycine, (2 × 10-z M) and centrifuging it in a linear sucrose gradient, relatively large amounts of glycine were found to be associated with particles of density similar to myelin4. It was suggested that this might be due to synaptosomes trapped by myelin4. It has been shown, however, that synaptosomes do not interact with myelin in sucrose, or Ficoll-sucrose gradients 7. This indicates that glycine was most likely present in myelin proper. Honegger et al. ~4, after incubating cat spinal cord slices with 5 × 10-7 [3H]glycine, homogenized the tissue and analyzed the subcellular distribution of [3H]glycine in a 10-step sucrose gradient. It was found by them that about 80% of glycine was present in fractions of density lower than that ofsynaptosomes, a considerable proportion of it probably being myelin ; this being quite compatible with the present results.

Significance of amino acid transport by myelin Autoradiographic studies have given clear evidence that amino acids, injected systemicallyz3 or intraaxonally 2, can readily penetrate myelin sheaths in zones removed from nodes of Ranvier. More recent studies also suggest that myelin proteins in the optic tract may be synthesized from amino acids supplied by axons 3°. This transport activity of myelin, that has been largely overlooked, has been suggested to be important for axonal nutrition 33, and also, in an opposite manner, for maintaining myelin integrity with amino acids supplied by the axoplasm 2. The present results cannot, however, decide between these possibilities. Nevertheless, the finding that such transport shows high-affinity kinetics may have some functional implications. Namely, the passage of amino acids through the myelin sheath should lead to progressively lower amino acid concentrations as the molecules reach deeper myelin layers. Under such circumstances low-affinity transport, that may be quite efficient for transport through

111 single u n i t m e m b r a n e s , becomes inefficient for securing significant solute fluxes, while a high-affinity t r a n s p o r t system seems well a d a p t e d for this function. ACKNOWLEDGEMENTS We are grateful to O. Petit a n d J. Salas for their technical help, to the N a t i o n a l Institute of M e n t a l Health, C O N I C Y T a n d Universidad de Chile for their financial support, a n d to Dulce M a r i a Ortiz a n d M a r i a Luisa G a l i n d o for secretarial help.

REFERENCES 1 Abdel-Latif, A. A., A simple method for isolation of nerve ending particles from rat brain, Biochim. biophys. A cta ( Amst. ) , 121 (1966) 403-406. 2 Alvarez, J. and Cben, W. Y., Injection of leucine into a myelinic axon: incorporation in the axoplasm and transfer to associated cells, ActaphysioL lat.-amer., 22 (1972) 266-269. 3 Aprison, M. H. and McBride, W. J., Evidence for the net accumulation of glycine into a synaptosomal fraction isolated from the telencephalon and spinal cord of the rat, Life Sci., 12 (1973) 449-458 4 Balcar, V. J. and Johnston, G. A. R., High-affinity uptake of transmitters: studies on the uptake of L-aspartate, GABA, L-glutamate and glycine in cat spinal cord, J. Neurochem., 20 (1973) 529-539. 5 Balcar, V. J. and Johnston, G. A. R., Liberation of amino acids during the preparation of brain slices, Brain Research, 83 (1975) 173-175. 6 Bennett, J. P., Mulder, A. H. and Snyder, S. H., Neurochemical correlates of synaptically active amino acids, Life Sci., 15 (1974) 1045-1056. 7 Doyle, L. C. and Cotman, C. W., Analysis of myelin-synaptosomal interactions in sucrose and Ficoll-sucrose gradients, Analyt. Biochem., 49 (1972) 29-36. 8 Eichberg, J., Whittaker, V. P. and Dawson, R. M. C., Distribution of lipids in subcellular particles of guinea-pig brain, Biochem. J., 92 (1964) 91-100. 9 Gray, E. G. and Whittaker, V. P., The isolation of nerve endings from brain: an electron-microscopic study of cell fragments derived by homogenization and centrifugation, J. Anat. (Lond.), 96 (1962) 79-88. 10 Hamberger, A., Amino acid uptake in neuronal and glial cell fractions from rabbit cerebral cortex, Brain Research, 31 (1971) 169-178. 11 Henn, F. A. and Hamberger, A., Glial cell function : uptake of transmitter substances, Proc. nat. Acad. Sci. (Wash.), 68 (1971) 2686-2690. 12 Htikfelt, T. and Ljungdahl, A., Light and electron microscopic autoradiography on spinal cord slices after incubation with labeled glycine, Brain Research, 32 (1971) 189-194. 13 HOkfelt, T. and Ljungdahl, A., Application of cytochemical techniques to the study of suspected transmitter substances in the nervous system. In E. Costa and P. Greengard (Eds.), Advances in BiochemicalPsychopharmacology, VoL 6, Raven Press, New York, 1972, pp. 1-36. 14 Honegger, C. G., Krepelka, L. M., Steinmann, V. and Von Hahn, H. P., Distribution of 3Hglycine and 3H-L-glutamate in synaptosomal subpopulations after in vitro uptake into cat dorsal and ventral spinal cord slices, Experientia (Basel), 30 (1974) 369-371. 15 Horne, R. W. and Whittaker, V. P., The use of negative staining method for the electron-microscopy study of subcellular particles from animal tissues, Z. Zellforsch., 58 (1962) 1-16. 16 Joanny, P., Barbosa, E., Hillman, H. and Corriol, J., The uptake and etttux of glycine from rat cerebral-cortex slices, Biochem. J., 125 (1971) 255-260. 17 Johnston, G. A. R. and Iversen, L. L., Glycine uptake in rat central nervous system slices and homogenates: evidence for different uptake systems in spinal cord and cerebral cortex, J. Neurochem., 18 (1971) 1951-1961. 18 Jo6, F. and Karnushina, I., Morphometric assessment of the composition of the synaptosomal fractions obtained by the use of Ficoll gradients, J. Neurochem., 24 (1975) 839-840. 19 Kurokawa, M., Sakamoto, T. and Kato, M., Distribution of sodium-plus-potassium-stimulated adenosine-triphosphatase activity in isolated nerve-ending particles, Biochem. J., 97 (1965) 833-844.

112 20 Lajtha, A. and Sershen, H., Inhibition of amino acid uptake by the absence of Na ~ in slices of brain, J. Neurochem., 24 (1975) 667 672. 21 Lemkey-Johnston, N. and Dekirmenjian, H., The identification of fractions enriched in nonmyelinated axons from rat whole brain, Exp. Brain Res., 11 (1970) 392-410. 22 Logan, W. J. and Snyder, S. H., High affinity uptake systems for glycine, glutamic and aspartic acids in synaptosomes of rat central nervous tissues, Brain Research, 42 (1972) 413-431. 23 Luft, J. H., Improvements in epoxy resin embedding methods, J. biophys, biochem. Cytol., 9 (1961) 409--414. 24 Neal, M. J., The uptake of 14C-glycine by slices of mammalian spinal cord, J. Physiol. (Lond.), 215 (1971) 103-117. 25 Norton, W. T. and Poduslo, S. E., Myelination in rat brain : method of myelin isolation, J. Neurochem., 21 (1973) 749-757. 26 Palade, G. E., A study of fixation for electron microscopy, J. exp. Med., 95 (1952) 285-298. 27 Peachy, L. D., Thin sections. I. A study of section thickness and physical distortion produced during microtomy, J. biophys, biochem. Cytol., 4 (1958) 233-242. 28 Peters, A., Palay, S. L. and Webster, H. de F., The Fine Structure of the Nervous System, Harper and Row, New York, 1970, 198 pp. 29 Peterson, N. A. and Raghupathy, E., Developmental transitions in uptake of amino acids by synaptosomal fractions isolated from rat cerebral cortex, J. Neurochem., 21 (1973) 97-110. 30 Prensky, A. L., Fujimoto, K. and Agrawal, H. C., Are myelin proteins synthesized in retinal ganglion cells?, J. Neurochem., 25 (1975) 883-887. 31 Sabatini, D. D., Bensch, K. and Barrnett, R. J., Cytochemistry and electron microscopy. The preservation of cellular ultrastructure and enzymatic activity by aldehyde fixation, J. Cell Biol., 17 (1963) 19-58. 32 Schrier, B. K. and Thompson, E. J., On the role of glial cells in the mammalian nervous system, J. biol. Chem., 249 (1974) 1769-1780. 33 Singer, M. and Salpeter, M. M., The transport of 3H-l-histidine through the Schwann and myelin sheath into the axon, including a reevaluation of myelin function, J. Morphol., 120 (1966) 281-316. 34 Stempak, J. G. and Ward, R. T., An improved staining method for electron microscopy, J. Cell Biol., 22 (1964) 697-701. 35 Tapia, R., Haj6s, F., Wilkin, G., Johnson, A. L. and Balgtzs, R., Subcellular fractionation of rat cerebellum: an electron microscopic and biochemical investigation. II. Resolution of morphologically characterized fractions, Brain Research, 70 (1974) 285-299. 36 Valdds, F. and Orrego, F., Interaction of glycine with brain-cortex membrane fragments: kinetics, ionic requirements and amino acid specificity, Brain Research, 97 (1975) 277-290. 37 Venable, J. H. and Coggeshall, R. E., A simplified lead citrate stain for use in electron microscopy, J. CellBiol., 25 (1965) 407-408. 38 White, T. D. and Keen, P., The role of internal Na + and K + on the uptake of (aH) noradrenaline by synaptosomes prepared from rat brain, Biochim. biophys. Acta (Amst.), 196 (1970) 285-295. 39 Whittaker, V. P., The isolation and characterization of acetylcholine containing particles from brain, Biochem. J., 72 (1959) 694-706. 40 Whittaker, V. P. and Sheridan, M. N., The morphology and acetylcboline content of isolated cerebral cortical synaptic vesicles, J. Neurochem., 12 (1965) 363-372. 41 Winne, D., Unstirred layer, source of biased Michaelis constant in membrane transport, Biochim. biophys. Acta (Amst.), 298 (1973) 27-31. 42 Zimmermann, A. W., Quarles, R. H., Webster, H. de F., Matthieu, J.-M. and Brady, R. O., Characterization and protein analysis of myelin subfractions in rat brain: developmental and regional comparisons, J. Neuroehem., 25 (1975) 749-757.