A hypothesis for the superior sensitivity of wheat germ agglutinin as a neuroanatomical probe

Brain Research, 495 (1989) 131-139

131

Elsevier BRES 14731

A hypothesis for the superior sensitivity of wheat germ agglutinin as a neuroanatomical probe A. Stieber l, S.D. E r u l k a r 2 a n d N . K . G o n a t a s ~ *Division of Neuropathology, University of Pennsylvania, Philadelphia, PA 19104 (U.S.A.) and 2Department of Pharmacology and Mahoney Institute of Neurological Sciences, University of Pennsylvania Medical School, Philadelphia, PA 19104 (U.S.A.) (Accepted 31 January 1989)

Key words: Uncoupling ligand from receptor (CURL); Presynaptic terminal We have shown that a conjugate of wheat germ agglutinin (WGA), with horseradish peroxidase (HRP), is more sensitive than native HRP as a probe of neuroanatomic connections involving the retrograde transport of the lectin. It has also been shown in our laboratory that WGA-HRP remains at the site of injection twice as long as HRP. The purpose of the present morphometric study was to investigate the basis for the higher sensitivity of WGA-HRP over HRP as a retrogradely transported tracer molecule. To do this, we modified the experiment of Heuser and Reese which utilized the tracing of HRP in the frog neuromuscular junction (Heuser, J.E. and Reese, T.S., J. Cell Biol., 57 (1973) 315-344). Instead of using HRP alone, we examined, in double labeling experiments, fluid and adsorptive endocytosis with free HRP and WGA coupled to ferritin (WGA-ferritin) respectively. Immediately after nerve stimulation, both markers are taken up simultaneously into cisternae, and in tubular structures strikingly similar to the described compartment of uncoupling of receptor from ligand (CURL). Frequently, cisternae were connected with putative CURL. This early double labeling of cisternae and putative CURL was followed by the appearance of synaptic vesicles labeled with WGA-ferritin only (72-79%), HRP only (6-11%), and both labels (13-16%). In contrast to the labeling pattern of synaptic vesicles, the majority of cisternae and putative CURL had both labels throughout the duration of the experiments (77-80%). The results of this study indicate that most of WGA-ferritin and HRP are co-localized in cisternae and putative CURL, compartments involved in endocytosis and surface receptor recycling. Therefore, the superior sensitivity of WGA over HRP as a marker of retrograde transport cannot be attributed to a differential distribution of these two tracer molecules within the intraceUular compartment involved in the endocytosis of ligands. These results are consistent with the hypothesis that in the nerve ending, as in the cell body, there is a compartment involved in the recycling of surface receptors. Such receptors may include molecules with terminal sialic acid and N-acetylglucosamine to which WGA binds. This compartment may be involved in numerous cycles of endocytosis of WGA bound to the surface of the nerve ending until the surface bound and extracellular pool of the lectin is exhausted. It remains to be proven whether in the frog presynaptic terminal the compartment identified in this study as 'putative CURL' is indeed involved in the recycling of cell surface receptors.

INTRODUCTION

site of injection twice as long as H R P 23. According

We have reported that a conjugate of the lectin wheat germ agglutinin ( W G A ) with horseradish peroxidase ( H R P ) is more sensitive than H R P in studies involving the retrograde transport of tracer molecules 8. Subsequently, studies from our laboratories and others have shown that W G A and other ligands are superior to H R P i n experiments involving the orthograde and transsynaptic transport of tracer molecules n'23. O n e study from this laboratory has shown that W G A coupled to H R P remains at the

to this study, the superior sensitivity of W G A - H R P over native H R P in n e u r o a n a t o m i c studies may be attributed, in part, to the slower rate of diffusion of W G A - H R P from the injection site. In order to investigate the basis for the higher sensitivity of W G A over H R P as a retrograde transport tracer, we adopted a modification of the H e u s e r - R e e s e experimental protocol, which was initially designed to study the recycling of synaptic m e m b r a n e s at the frog n e u r o m u s c u l a r junction with H R P as the tracer molecule ~2. We introduced one

Correspondence: A. Stieber, Division of Neuropathology, University of Pennsylvania, 454 Johnson Pavilion/G2, Philadelphia, PA 19104, U.S.A. 0006-8993/89/$03.50 © 1989 Elsevier Science Publishers B.V. (Biomedical Division)

132 modification to their experiment, namely we used a monomeric conjugate of W G A with ferritin together with HRP in double label studiesH( These double labeling experiments have allowed us to gain insight into the co-distribution of a marker of adsorptive endocytosis (WGA-ferritin), and fluid phase endocytosis (HRP), in frog presynaptic terminals. On the basis of the information obtained, we propose a hypothesis to explain the superior sensitivity of WGA, and other ligands, as tracers in neuroanatomic studies. According to this hypothesis, presynaptic terminals in the frog contain a compartment (CURL) for mediating the uncoupling of ligands from receptors and the retrieval of receptors at the cell surface 4. This compartment may be involved in the internalization of receptors to W G A with bound lectin, uncoupling of W G A from its receptors and their recycling to the cell surface for repeated cycles of endocytosis of WGA, until the extracellular pool and surface bound W G A are exhausted. MATERIALS AND METHODS Frogs (Rana pipiens) were used. They were anesthetized with tricaine sulfonate added to the water and the sartorius muscle was exposed. One hundred microliters of wheat germ agglutinin-ferritin (WGA-ferritin) or W G A - H R P was injected into muscle at different sites in a series of 20 injections. The WGA-ferritin was used at a concentration of 50 /~g/ml of W G A in frog Ringer's solution at a pH of 7.2. A monomeric conjugate of WGA-ferritin with glutaraldehyde as the cross-linking agent, was purchased from Polysciences (Warrington, PA 189762590). The conjugate of W G A with HRP was done in our laboratory according to previously described methods ~'5's. As described in detail previously, separation of WGA--HRP from free HRP and free W G A was done by column chromatographys. The method of conjugation of W G A to HRP, used in these experiments, does not modify the specific binding of W G A to N-acetylglucosamine5. Similarly, as reported previously, conjugation of W G A with ferritin with glutaraldehyde as the cross-linking agent does not change the affinity of the lectin for its plasma membrane receptor(s) w. The concentration used was 50 ktg/ml of WGA. The skin incision over the sartorius was closed by

sewing together the skin flaps and the frog was allowed to recover over the next 1.5 h. During this period, frogs were kept at room temperature. This procedure allowed the WGA-ferritin or W G A HRP to reach the endplate region, but it must be noted that as the animal recovered from the anesthesia some movement took place. Thus some of WGA-ferritin or W G A - H R P was internalized and the control levels of the label within the terminal were higher than would be expected if the muscle had been completely quiescent. In order to examine whether residues of free ferritin in the WGA-ferritin conjugate may have contributed to the counts of ferritin particles, free ferritin (10 mg/ml) was injected into sartorius muscle; 2.5 h later, the muscle was excised, fixed and processed for electron microscopy. In a second experiment, free ferritin (110 mg/ml) was injected into the muscle; 2 h later, the nerve was stimulated for 15 min and the muscle was fixed after 15 min rest. In the third experiment of this group, the muscle was excised and bathed in ferritin (10 mg/ml) for 2 h before fixation. Two hours after injection, the sartorius muscle and its nerve were removed and placed in a nerve-muscle chamber. Square wave pulses of 0.5 ms were then applied to the nerve and increased in intensity until the threshold for contraction was reached. The stimuli were then set at values of 4 times threshold; D-tubocurarine (0.02 mM in frog Ringer's solution) and, in several experiments, free horseradish peroxidase (HRP, 10 mg/ml) were added to the chamber; after 15-30 min, the nerve was stimulated with one or two pulses to test whether contraction had been abolished. Invariably, this was the case. When HRP was used, it was present in the bath for a total of 30 min before beginning stimulation. The total time from the end of the W G A ferritin or W G A - H R P injection to the beginning of stimulation was 2.5-3 h. The nerve was then stimulated at 10 Hz for one or 15 min. In control experiments, no stimulation was applied. Some muscles were then fixed immediately in 2% paraformaldehyde and 3% glutaraldehyde in 0.09 M cacodylate buffer pH 7.2 + 20 mM CaCI2 and left overnight at 4 °C. Other muscles rested for 5 or 30 min before fixation. The next day, the preparation was washed for 2 h, with several changes, in 0.09 M cacodylate buffer + 20 mM CaCi 2. Cross-sections,

133 70-100/~m thick, were cut from the middle half of the muscle with a Vibratome (Oxford). The sections were immersed in diaminobenzidine solution (DAB; 0.5 mg/ml in 0.1 M Tris, pH 7.4) without H 2 0 2 and placed in an orbital shaker for 30 min. This was followed by the same solution plus 0.01% H20 2 for 20 min in the dark with shaking. The sections were washed in 0.1 M cacodylate buffer at pH 7.2, postfixed with 2% osmium tetroxide in 0.1 M cacodylate buffer at pH 7.2, washed once, dehydrated with ethanol, and embedded in Araldite. From semithin sections, areas containing endplates were selected for thin sectioning; if H R P was used, only endplates surrounded by intense staining were selected. Since staining sections with uranyl and lead interfered with unambiguous detection of

H R P stained structures, sections from muscles exposed to WGA-ferritin, W G A - H R P , or free H R P (labeled muscles) were stained for HRP, but not with uranyl or lead. Sections from unlabeled muscles were post-fixed with 2% osmium tetroxide in 0.1 M cacodylate buffer at pH 7.2 and stained with uranyl acetate and lead citrate. Pictures were taken from two sections of each block, avoiding taking pictures of the same endplates. In labeled muscles, only endplates well stained by both labels were used. For quantitative studies, pictures were printed at final magnifications of 45,000-50,000 times. The surface area of sections of endplates was calculated with a L A D D digitizer interfaced with a MINC-11 minicomputer according to a described program 15. Synaptic vesicles, coated vesicles and

Fig. 1. Endplate labeled with WGA-ferritin and HRP; the nerve was stimulated for 15 rain and the muscle was immediately fixed. At this magnification several ferritin labeled synaptic vesicles and a few HRP labeled cisternae, 0.25/~m in average diameter, are seen. Inset: higher magnification showing that the cisternae are indeed labeled with both HRP and ferritin. Arrow, fusion of a vesicle with a cisterna.

134 cisternae were counted on sections from unlabeled muscles. Since coated vesicles could not be reliably identified in muscles labeled with HRP, only synaptic vesicles and cisternae were counted in them. We calculated the density of labeled and unlabeled organelles per tim 2 of the section area of the terminal, and percentages of organelles with each or both labels. For each experiment, means and standard errors of means were calculated. RESULTS

Definition of organelles Fig. 1 is from a section of a sartorius muscle labeled with W G A - f e r r i t i n and H R P ; the nerve was stimulated for 15 min and the muscle was immediately fixed. The endplate contains ferritin labeled synaptic vesicles and large vacuoles with clear centers which will be referred to as cisternae; also, the endplate contains H R P and ferritin labeled synaptic vesicles and cisternae (double labeling). Fig. 2 is from a section of muscle from the same experiment as that shown in Fig. t. In addition to several double labeled cisternae, the endplate contains double labeled tubulovesicular structures, similar to C U R L 4. According to Geuze et al. both the tubulovesicular structures and the cisternae with which they fuse (see Fig. 2), belong to C U R L 4. These tubulovesicular structures were labeled with H R P and W G A - f e r r i t i n in a pattern very similar to

Fig. 2. Endplate from same experiment as Fig. 1. Note double labeled cisternae and tubulovesicular structures (CURL) in continuity with cisternae (arrow). C, cisterna.

that of large cisternae (Fig. 1). For the above reasons, we decided to include measurements of the tubulovesicular structures in the compartment of 'cisternae'. Fig. 3 is from an experiment conducted as follows: free ferritin was injected and the muscle was dissected as with W G A - f e r r i t i n ; the nerve was stimulated for 15 min and rested 30 min prior to fixation. In this figure, a typical e n d o s o m e - r e c e p t o s o m e as described by Willingham and Pastan is shown 22. To be noted is the characteristic fuzzy material on one straight segment of the limiting membrane 22. It should be noted that the electron micrograph in Fig. 3 was selected for the typical illustration of an endosome. This figure is not representative of the average concentration of free ferritin, which was extremely low (see below).

Free ferritin Native ferritin was conjugated with W G A . In order to assess whether contaminants of unconjugated ferritin, present in W G A - f e r r i t i n conjugate, may affect our quantitative studies we conducted the experiments with native, or free ferritin. These experiments with free ferritin showed many particles of ferritin in the extracellular space, but few in cisternae, tubulovesicular structures, e n d o s o m e s receptosomes and synaptic vesicles. The quantitative differences between the internalization of W G A ferritin and free ferritin were so striking that it is safe

Fig. 3. Free ferritin, (100 mg/ml), was injected and the dissected muscle soaked in ferritin. The nerve was stimulated for 15 min and rested 30 min prior to fixation. Arrow: typical endosome with fuzzy coat over one limiting membrane.

135

TABLE I

TABLE III

Effect of nerve stimulation on numbers of synaptic vesicles, coated vesicles and cisternae

Effect of double labeling on response of synaptic vesicles and cisternae to stimulation

Unstimulated ± S.E.M.*

Synaptic vesicles/#m 2 Coated vesicles//~m2 Cisternae//tm2

69.5 ± 3.1 0.19 + 0.03 0.54 ± 0.07

Fixed immediately after 15minstimulation**

P

28.8 ± 2.5 0.41 ± 0.10 1.74 ± 0.18

<0.01 0.04 <0.01

* Average of 4 separate experiments. ** Average of 2 separate experiments. to assume that most, if not all, of the ferritin label seen in experiments with W G A - f e r r i t i n is due to the binding of the lectin to its receptor(s) ~8. Effect o f nerve stimulation on numbers o f synaptic vesicles, coated vesicles and cisternae In these experiments we counted synaptic vesicles, c o a t e d vesicles and cisternae, before (zero

time) and 15 min after nerve stimulation at 10 Hz (Table I). O u r results are consistent with those of H e u s e r and R e e s e , who r e p o r t e d a decrease in the n u m b e r of synaptic vesicles and an increase in the n u m b e r of c o a t e d vesicles and cisternae in presynaptic terminals after nerve stimulation 12. W G A - f e r r i t i n is a better label than W G A - H R P Since in these experiments we used W G A - f e r r i t i n

instead of W G A - H R P we investigated first the sensitivity of the ferritin conjugate with W G A . E x p e r i m e n t s designed to d e t e r m i n e the most efficient label, as defined by the percentage of synaptic

Total synaptic vesicles//~m2 Total cisternae//~mz

Unstimulated

Stimulated (15 rain)

P

95.4 ± 11.0 2.14 ± 0.59

30.7 ± 3.8 5.24 ± 0.62

<0.01 <0.01

vesicles showing the m a r k e r u n d e r similar conditions of stimulation, showed a g r e a t e r percentage of vesicles labeled with W G A - f e r r i t i n c o m p a r e d to W G A - H R P (Table II). Value (P) derived from Student's t-test showed that the differences between percentages of synaptic vesicles labeled with W G A H R P and W G A - f e r r i t i n at 0 time were not statistically valid (Table II) but all o t h e r differences were statistically valid (Tables I - I I I ) . It should be e m p h a sized that the precision and the quantitative nature of the detection of ferritin particles m a y have contributed to the superiority of W G A - f e r r i t i n over W G A - H R P as an endocytic marker. On the o t h e r hand, evaluations of the density of the diffusable H R P reaction product are subjective and semiquantitative at best 3'17. F r o m these experiments (Table II), we concluded that the use of W G A - f e r r i t i n as the m a r k e r for adsorptive phase of endocytosis is valid. F u r t h e r m o r e , W G A - f e r r i t i n is c o m p a t i b l e with double label studies with HRP, while W G A H R P is not.

TABLE IV Percentages of labeled and unlabeled synaptic vesicles

TABLE II Comparison of sensitivity of WGA-HRP and WGA-ferritin

Stim., duration of stimulation: 0, 1 and 15 min. 15 min, 5 rest, 15 rain stimulation, followed by 5 min rest; 15 min, 30 rest, 15 min stimulation, 30 min rest before fixation.

15 min, specimens fixed immediately after 15 min of stimulation; 15 S/30R, nerve stimulated for 15 min and specimen fixed after rest of 30 min.

Stim.

% WGAferritin

% Free HRP

% Double labeled

% Unlabeled

0time lmin 15min 15min 5rest 15min 30rest

3.9±0.8 7.7±0.8 27.2±2.8

0.1±0.1 0.5±0.2 2.1±0.6

0.1±0.3 1.4±0.2 5.7±1.5

95.9±0.8 90.4±0.9 64.9±3.7

34.2±2.4

5.3±0.8

6.4±0.7

54.1±2.7

34.0±2.6

6.3±1

6.4±0.8

53.2±2.5

% Synaptic vesicles labeled

WGA-HRP WGA-~rritin P

Zerot~e

15mm

15S/30R

1.9±1.1 3.9±0.8 0.15

2.5±0.4 32.9±3.1 <0.01

11.2±1.7 40.4±3.4 <0.01

136 TABLE V

Percentages of labeled synaptic vesicles and numbers of synaptic vesicles and endplates counted Stim.

% WGA-ferritin

% Free HRP

% Double labeled

No. SV

No. endplates

0 time 1 rain 15 min 15 min, 5 rest 15 min, 30 rest

96.5 + 3.1 79.8 + 2.9 77.7 _+3.2 74.6 + 2.3 72.7 + 2.9

2.0 _+3.0 5.5 _+ 1.8 6.0 _+ 1.4 11.4 _+ 1.8 13.5 -+ 2.5

1.5 + 0.8 14.6 + 2.4 16.3 _+2.5 t3.9 _+ 1.2 13.8 + 1.5

4946 4311 3763 6418 4326

39 46 55 88 47

Double labeling shows that markers o f both adsorptive and fluid phase endocytosis are incorporated in the terminal following stimulation Table I11 shows that the procedure used for double labeling may enhance the response of the synaptic vesicles and cisternae to stimulation. Thus, unstimulated nerve endings had about 70 synaptic vesicles and cisternae per ,um2 but unstimulated endings exposed to W G A - f e r r i t i n and H R P had 97 organelles per,um 2 (Tables I and III). This apparent effect of the injections of W G A - f e r r i t i n on the n u m b e r of organelles in nerve endings may have been caused by the m o v e m e n t of limbs as the animals woke up. Comparison between Tables I and III show a similar decrease of synaptic vesicles and increase of cisternae after stimulation. Double labeling shows that the majority o f cisternae are labeled with H R P and W G A - f e r r i t i n while the majority o f synaptic vesicles are labeled only with WGA-ferritin Percentage counts of synaptic vesicles labeled after nerve stimulation showed that about 50% of synaptic vesicles were labeled with HRP, W G A ferritin or both labels. Furthermore, 72-80% of

labeled synaptic vesicles contained W G A - f e r r i t i n (Tables IV and V). In contrast, after nerve stimulation 9 1 - 9 4 % of cisternae were labeled with either HRP, W G A - f e r r i t i n or both labels, and the majority of labeled cisternae ( 7 6 - 8 0 % ) contained both labels (Tables VI and VII). Since W G A - f e r r i t i n was injected 2 h before the excision of the nerve and sartorius muscle, and the frog was moving during this time, some cisternae were labeled with W G A ferritin even before any experimental stimulation was applied. Even after a brief stimulation the n u m b e r of cisternae labeled only with W G A - f e r r i t i n decreased while the n u m b e r of cisternae containing both labels increased (Tables VI and VII). This finding strongly suggests that with stimulation, both adsorptive and fluid phase markers are simultaneously internalized into newly formed cisternae rather than into those already containing only W G A - f e r r i t i n , which apparently do not communicate with newly formed doubly labeled cisternae. It is clear that when both labels are present during stimulation, they are taken up into cisternae simultaneously; after just 1 min of stimulation, 77% of all labeled cisternae took up both markers (Table VII).

TABLE VII TABLE VI

Percentages of labeled eisternae and numbers of cisternae counted

Percentages of labeled and unlabeled cisternae Stim.

% WGAferritin

0time lmin 15min 15 rain, 5 rest 15mira 30 rest

43.2+7.0 15.0+4.1 15.1+2.7

% Free HRP

% Double labeled

2.7+2.1 17.1+4.1 6.1 + 1.3 70.0+4.6 5.9_+2.1 71.8+4.2

% Unlabeled

Stim.

% WGAferritin

36.9+6.2 8.9+3.2 7.1_+3.2

0time 1 min 15 min 15 min, 5rest 15 min, 30 rest

68.5+8.1 16.4 + 4.3 16.2 + 3.6

8.4 -+ 2.6

9.7 + 2.2

74.2+ 4.2

7.6 _+3.1

12.0+5.1

10.2+2.7

71.1 _+5.6

6.7+_3.0

% Free HRP

% Double labeled

No. cisternae

4.3+0.5 6.7 + 1.4 6.4 + 2.3

27.1+7.4 76.9 + 4.3 77.4 + 3.9

111 247 644

9.1+3.4

10.5+2.7

80.3+3.9

380

12.8+5.4

11.0+2.8

76.1+5.8

225

137 In conclusion, the majority of the elements of the compartment involved in endocytosis and receptor recycling, i.e. cisternae and 'putative C U R L ' , are labeled and contain both labels, while the majority of synaptic vesicles are not labeled (Tables IV and VI). X2-tests of the percent of synaptic vesicles or cisternae containing each or both labels at successive times were statistically significant only between zero time and 1 min stimulation (P < 0.1) (Tables V and VII). When unlabeled organelles were included in the statistical analysis the results were significant (P < 0.01) between 1 and 15 min stimulation for synaptic vesicles and between zero time and 1 min stimulation for cisternae (Tables IV and VI). Endosomes have been seen at motor nerve terminals, although rarely In only 5 out of the hundreds of endplates examined, structures identical to endosomes as described by Willingham and Pastan were identified (Fig. 3) 22. They were large vesicular structures averaging 0.25-0.5/~m in diameter with one straight edge that had fuzzy material attached to it. Fig. 3 shows that this structure could incorporate free ferritin. In contrast to the rarity of 'typical' endosomes or receptosomes described previously, the occurrence of structures that strikingly resembled the compartment of uncoupling of receptor and ligand (CURL) was quite frequent. These tubulovesicular structures were labeled with both WGA-ferritin and H R P in the same manner as the cisternae, and were thus included in the quantitative studies in the cisternal compartment. These results strongly suggest that the motor nerve terminal has the compartments needed for receptor recycling 4"1°. The rarity of 'endosomesreceptosomes' may be explained by the fact that only a 'fortunate' section of large empty looking cisternae may include its characteristic straight segment with the 'fuzzy' coat (Fig. 3); since most of the double labeled cisternae have the same diameter as endosomes, they may indeed be endosomes sectioned at a level not including the characteristic 'fuzzy' segment.

DISCUSSION Ligands conjugated with various marker mole-

cules (HRP, ferritin, colloidal gold) have been used in ultrastructural studies of adsorptive or receptor mediated endocytosis. Several factors such as the valency of the ligand, the methods of conjugation, the choice of the marker molecule, and the diffusion of the HRP reaction product of the ligand-marker complex affect the routes of internalization and/or detection of the marker molecule. The enzyme H R P has been used in numerous studies as a marker of fluid phase endocytosis. Using a quantitative assay for H R P we have found that cultured murine neuroblastoma cells (neuro-2A line) internalized 100-200 times more H R P conjugated to ricin than free H R P 9. In contrast, Chan et al. found that cultured mouse neuroblastoma cells (clone N18) internalized more isoenzyme C of H R P than other peroxidases or cationized ferritin 2. Because of the known diffusion of the H R P reaction product, quantification of its uptake with morphometric studies, as done by Chan et al., is approximate. Therefore, the above cited discrepancies may reflect real differences of the uptake of H R P isoenzyme C, (which was not used in our studies), the different cell lines used, or methodologic limitations in the quantification of HRP. Our studies have also shown that markers of adsorptive endocytosis undergo endocytosis in the Golgi apparatus while free H R P is not internalized into the Golgi apparatus 5-7'1°'2°'21. Lavail and Margolis have shown that iodinated W G A is transported in the chick optic tectum at significantly slower rates than W G A - H R P or H R P 13. We have shown with quantitative assays that iodinated W G A was 13 times more efficient than iodinated W G A - H R P as a marker of transport from the rat retina to the geniculate and 3 times more efficient as a marker of transsynaptic transport to the cortex t9. We have also shown that while W G A HRP is internalized in the Golgi apparatus of rat pheochromocytoma (PC 12), WGA-ferritin is not 1°. There is little doubt that covalent linkage of W G A with HRP, ferritin or other markers affects the efficiency of the transport of the conjugate and even its intracellular pathway. Furthermore, in previous experiments we have shown that iodination of W G A and its subsequent covalent linkage with H R P with glutaraldehyde reduces the carbohydrate affinity of the lectin. Only 27% of commercially available iodinated W G A bound to NAcGlu-agarose, and

138 49% of iodinated WGA, coupled to HRP, bound to NAcGlu-agarose ~'~. The conjugate of W G A - H R E in addition to being more sensitive than HRP, stays longer at the site of injection and diffuses less than HRP 8'23. While the WGA-ferritin used in these experiments as the membrane marker is more suitable than a fluid phase marker such as free HRP, it cannot be regarded as being bound necessarily and irreversibly to the same membrane. Since the lectin is not covalently bound to its 'receptor', it is possible that after endocytosis, WGA-ferritin may bind to another molecule which may not have been attached to the same membrane originally. W G A is covalently bound to ferritin; therefore it is unlikely that W G A dissociates from ferritin within the synaptic terminal. Our results show that both the monomeric W G A ferritin and free H R P are taken up simultaneously during endocytosis into the compartments described as cisternae. A similar labeling with both fluid phase and adsorptive endocytosis markers has been observed in endosomes of pheochromocytoma cells (PC 12)H( However, this double labeling of cisternae (endosomes?) is followed by the appearance of synaptic vesicles labeled predominantly with W G A ferritin only (73% of those labeled; 40% of all synaptic vesicles) (Table V). Free HRP was seen in only 13% of the synaptic vesicles, a value similar to that reported by Meshul and Pappas 16. These results may be interpreted to indicate that at least two different populations of synaptic vesicles 'pinch off' from the membranes of double labeled cisternae, namely those enriched in 'receptors' to W G A (containing WGA-ferritin) and those poor in ~receptors' to W G A containing only HRP. Alternatively, the differences between WGA-ferritin and HRP labeled synaptic vesicles may be explained by the higher density of WGA-ferritin compared to HRP, which facilitates the detection of WGA-ferritin. At the commencement of the experiment (0 time), 43% of all cisternae or 68% of all labeled cisternae contained only WGA-ferritin (Tables VI and VII). The percentages of cisternae labeled only with WGA-ferritin dropped sharply after nerve stimulation for 1 min and remained at approximately similar low levels during the entire experiment, while following nerve stimulation most cisternae were double labeled (Tables VI and VII). These

results suggest that during and after nerve stimulation, WGA-ferritin and HRP are internalized into newly formed cisternae rather than into the 'old' ones containing only WGA-ferritin. The alternative explanation, namely that the decrease of the percentages of cisternae labeled only with W G A ferritin is due to the internalization of HRP into them, is unlikely. If the 'old' cisternae, initially containing only WGA-ferritin, after stimulation were filled with HRP, a progressive decrease of cisternae filled only with WGA-ferritin might be expected. This was not the case (Tables VI and VII). Endosomes (receptosomes) and/or C U R L as defined by Willingham and Pastan, Maxfield, and by Geuze et al., have not been described as such in nerve terminals to our knowledge 4'~4"22. However, the large cisternae with clear empty centers and tubular structures have been described in nerve endings12.16. Several pieces of evidence suggest that the cisternae and tubulovesicular structures may function as endosomes and CURL: first, their early appearance with stimulation and decrease in number on cessation of stimulation; second, they are double labeled when both labels are present, as is the case with endosomes and tubulovesicular structures of PC 12 cellsS; third, the appearance of some structures with morphological characteristics identical to endosomes (Fig. 3). Our results suggest that the motor nerve terminal has the requisite organelles for receptor recycling. These experiments offer an explanation for the higher sensitivity of W G A - H R P over HRP in studies of neuronal connectivity. The explanation is based on the hypothesis that the cisternae-tubular vesicular structures (Figs. 1 and 2) are indeed compartments for uncoupling of ligands from receptors, which subsequently return to the cell surface for repeated cycles of endocytosis of W G A until the pool of surface bound and extracellular W G A is exhausted. Our results show that both H R P and WGA-ferritin colocalize in putative CURL. We propose that the slow diffusion of W G A - H R P (and probably WGA-ferritin) from the injection site makes the conjugate available for numerous cycles of endocytosis in the putative C U R L compartment, while the diffusable H R P is not 23. Thus, much more WGA-ferritin (or W G A - H R P ) than HRP can be

139

internalized at the nerve ending. Double labeling studies for both ligand and receptor, and determination of pH in putative C U R L are needed to establish whether presynaptic terminals are indeed endowed with C U R L 4A4.

ACKNOWLEDGEMENTS We thank Ms. Judith A. Westley for typing the manuscript and editorial comments. We wish to thank Mary O ' B r i e n for technical assistance. We also acknowledge Grants NS05572, NS12211 and a grant from the Muscular Dystrophy Association.

REFERENCES 1 Avrameas, S. and Ternynck, T., Peroxidase-labelled antibody and Fab conjugates with enhanced intracellular penetration, lmmunochemistry, 8 (1971) 1175-1179. 2 Chan, K.Y., Bunt, A.H. and Haschke, R.H., Endocytosis and compartmentalization of peroxidases and cationized ferritin in neuroblastoma cells, J. Neurocytol., 9 (1980) 381-403. 3 Courtoy, P.J., Picton, D.H. and Farquhar, M.G., Resolution and limitations of the immunoperoxidase procedure in the localization of extracellular matrix antigens, J. Histochem. Cytochem., 31 (1983) 945-951. 4 Geuze, H.T., Slot, J.W., Strous, J.A.M., Lodish, H.E and Schwartz, A.L., Intracellular site of asialoglycoprotein receptor-ligand uncoupling: double label immunoelectron microscopy during receptor-mediated endocytosis, Cell, 32 (1983) 277-287. 5 Gonatas, N.K. and Avrameas, S., Detection of plasma membrane carbohydrates with lectin peroxidase conjugates, J. Cell Biol., 59 (1973) 436-443. 6 Gonatas, N.K., Stieber, A., Kim, S.U., Graham, D.I. and Avrameas, S., Internalization of neuronal plasma membrane ricin receptors into the Golgi apparatus, Exptl. Cell Res., 94 (1975) 426-431. 7 Gonatas, N.K., Kim, S.U., Stieber, A. and Avrameas, S., Internalization of lectins in neuronal GERL, J. Cell Biol., 73 (1977) 1-13. 8 Gonatas, N.K., Harper, C., Mizutani, T. and Gonatas, J.O., Superior sensitivity of conjugates of horseradish peroxidase with wheat germ agglutinin for studies of retrograde axonal transport, J. Histochem. Cytochem., 27 (1979) 728-734. 9 Gonatas, J., Stieber, A., Olsnes, S. and Gonatas, N.K., Pathways involved in fluid phase and adsorptive endocytosis in neuroblastoma, J. Cell Biol., 87 (1980) 579-588. 10 Gonatas, N.K., Stieber, A., Hickey, W.E, Herbert, H.S. and Gonatas, J.O., Endosomes and Golgi vesicles in adsorptive and fluid phase endocytosis, J. Cell Biol., 99 (1984) 379-390. 11 Gonatas, N.K. and Gonatas, J.O., Lectin and toxinperoxidase conjugates as sensitive markers for the study of neuronal connectivity. In S. Avrameas et al. (Eds.), lmmunoenzyrnatic Techniques, Elsevier, Amsterdam, 1983, pp. 111-120. 12 Heuser, J.E. and Reese, T.S., Evidence for recycling of synaptic vesicle membrane during transmitter release at the

frog neuromuscular junction, J. Cell Biol., 57 (1973) 315-344. 13 Lavail, H.J. and Margolis, T.P., The anterograde axonal transport of wheat germ agglutinin as a model for transcellular transport in neurons. In R.S. Smith and M.A. Bisby (Eds.), Axonal Transport, Liss, New York, 1986, pp. 311-326. 14 Maxfield, ER., Weak bases and ionophores rapidly and reversibly raise the pH of endocytic vesicles and cultured mouse fibroblasts, J. Cell Biol., 95 (1982) 676-681. 15 Merten, C., Stieber, A., Grand, E., Gonatas, J.O. and Gonatas, N.K., Use of a minicomputer for quantitative ultrastructural autoradiography, J. Histochem. Cytochem., 29 (1980) 585-587. 16 Meshul, C.K. and Pappas, G.D., The relationship of pinocytosis and synaptic vesicles at the frog neuromuscular junction, Brain Research, 290 (1984) 1-18. 17 Messing, A., Stieber, A. and Gonatas, N.K., Resolution of diaminobenzidine for the detection of horseradish peroxidase on surfaces of cultured cells, J. Histochem. Cytochem., 33 (1985) 837-839. 18 Monsigny, M., Roche, A.-C., Sene, C., Maget-Domon, R. and Delmotte, F., Sugar-lectin interactions: how does wheat-germ agglutinin bind sialoglyco-conjugates, Eur. J. Biochem., 104 (1980) 147-153. 19 Rhodes, C.H., Gonatas, J.O. and Gonatas, N.K., A quantitative comparison of the efficiency of orthograde axonal transport and transsynaptic transport of iodinated (125I) wheat germ agglutinin (I-WGA) and horseradish peroxidase labelled (I-WGA-HRP) in the rat visual system, Brain Research, 336 (1985) 376-380. 20 Rhodes, C.H., Stieber, A. and Gonatas, N.K., A quantitative electron microscopic study of the intracellutar localization of wheat-germ agglutinin in retinal neurons, J. Comp. Neurol., 254 (1986) 287-296. 21 Trojanowski, J.Q. and Gonatas, N.K., A morphometric study of the endocytosis of wheat-germ agglutinin-horseradish peroxidase conjugates by retinal ganglion cells in the rat, Brain Research, 272 (1983) 201-210. 22 Willingham, M.C. and Pastan, I., Endocytosis and exocytosis: current concepts of vesicle traffic in animal cells, Int. Rev. Cytol., 92 (1984) 51-85. 23 Wan, X.-C.S., Trojanowski, J.Q. and Gonatas, J.O., Cholera toxin and wheat germ agglutinin conjugates as neuroanatomical probes. Their uptake and clearance, transganglionic and retrograde transport and sensitivity, Brain Research, 243 (1982) 215-224.