- Email: [email protected]
depression: an ultrastructural study in the hippocampus
Brain Research7 265 (1983) 21-30 Elsevier Biomedical Press
21
Synaptic Correlates of Associative Potentiation/Depression: an Ultrastructural Study in the Hippocampus NANCY L. DESMOND and WILLIAM B. LEVY* Department of Neurological Surgery, Box 420, University of Virginia School of Medicine, Charlottesville, Vii 22908 (U.S.A.) (Accepted September 14th, 1982) Key words: long-term potentiation - - hippocampus - - competition - - associative potentiation/depression
Brief high-frequency trains delivered to the monosynaptic entorhinal cortical input to the dentate gyrus result in both increases and decreases of synaptic strength as a function of whether a particular afferent is active during conditioning (associative potentiation/depression15,8°). The present report concerns the effect of such brief, high-frequency conditioning trains upon the asymmetric synapses of the rat dentate gyrus molecular layer. Only those animals whose responses increased at least 50 ~ following conditioning stimulation were included in the study. Additional animals were used for one-dimensional current source density analyses to localize the activated synaptic region. Double blind scoring procedures were used to classify and quantify electron micrographic data. Asymmetric synapses were scored as a function of their position in the molecular layer, spine head size and shape, and postsynaptic density length. All data were treated as inherently matched comparisons between the conditioned and control sides of each animal. The number of large, concave spine synapses with large postsynaptic densities significantly increases in the central zone of synaptic activation. Bordering this zone are regions with increases in synaptic number following conditioning, primarily due to an increased number of small spine synapses. The increased number of large, concave spine synapses in the central zone is postulated to mediate associative potentiation. The many small spine heads just adjacent to the zone of strongest synaptic activation may reflect synaptic depression evoked at synapses inactive during conditioning. INTRODUCTION Since the inception o f the n e u r o n doctrine, numerous neuroscientists have suspected or assumed that modification o f synaptic connections mediates m e m o r y storage2,6,10,11,18, 24. Recent experiments lend support to this assumption. Morphological alterations o f synapses, as well as apparent numerical changes, occur as a consequence o f experience (e.g. environmental manipulations 9) and o f artificially-induced neuronal use (e.g. electrical stimulationS,S,13,20,21,~7). The initial discovery o f long-term potentiation 4 (LTP) in the h i p p o c a m p u s provides the first clear neurophysiological evidence for a neuronal analog o f m e m o r y storage. T o d a y , this experimental model is an even better analog since L T P as studied at the entorhinal synapses o f the dentate g y m s is an associative p h e n o m e n o n , dependent u p o n the cor* To whom correspondence should be addressed. 0006-8993/83/0000-0000/$03.00 © 1983 Elsevier Science Publishers
related activity o f convergent, m o n o s y n a p t i c excitatory afferents15,17. Here, the c o m b i n e d co-activity o f a presynaptic input and sufficient excitation o f a postsynaptic cell increases the synaptic efficacy o f the particular synapses involved (associative potentiation). The same cell that exhibits potentiation m a y concurrently exhibit the p h e n o m e n o n o f depression, an erasure-like, long-term process that opposes potentiation. In the case o f depression, inactivity o f a particular presynaptic afferent concomitant with strong activity o f converging excitatory afferents decreases the synaptic efficacy o f the synapse formed by the inactive afferent u p o n the postsynaptic cell. Finally, physiological evidence, c o m b i n e d with the inherent controls existing in this monosynaptic, bilaterally projecting system, argue for a synaptic locus (including possibly spine stem changes) for these alterations (see refs. 14, 15, 29, 30 for evidence).
22 Previous investigations correlating morphological changes produced by brief high-frequency trains expected and sought only correlates of potentiation, not of depression. On the other hand, the present investigation seeks correlated morphological changes that could logically mediate both potentiation and depression in the entorhinal cortexdentate gyrus (EC-DG) system. The conclusions of Levy and Steward 15 and Wilson et al.Z9,30 guide the search and interpretations of any anatomical changes observed after conditioning. Taking advantage of the laminated distribution of the EC-DG afferents22, the present study finds morphological correlates that logically explain and may be the cellular basis of associative potentiation/ depression. A preliminary report has appeared 7. MATERIALS AND METHODS
Experiments were performed on 300-400 g, male Sprague-Dawley derived rats (Flow Laboratories). Each animal was initially anesthetized with chloralose (55 mg/kg) and urethane (0.2-0.4 g/kg). Anesthesia was maintained with supplemental doses of urethane as needed over the course of the experiment. Body temperature was maintained at 37 ~ 1 °C.
Electrophysiological methods These were comparable to those methods previously describedlL Twisted wire, bipolar stimulating electrodes were positioned bilaterally in the angular bundle (8.1 mm posterior to bregma, 4.4 mm lateral to midline). Glass recording micropipettes filled with 0.9 ~ (w/v) NaC1 were advanced bilaterally into the anterior portion of the dentate gyrus at a 5° angle away from the midline (2.8 mm posterior to bregma, 1.3-1.5 mm lateral to midline). These micropipettes were positioned between the dorsal and ventral leaves of the dentate gyrus and at its lateral edge so as to avoid tissue damage in the region to be analyzed morphologically. The exposed brain surface was covered with mineral oil until all electrodes were situated at the depth evoking maximal dorsal leaf granule cell population responses. A warm 5 ~ (w/v) Agar solution in 0.9 ~ (w/v) NaC1 was then placed over the skull to minimize edema. Monophasic constant current pulses of 300 /zs
duration were delivered through the angular bundle stimulating electrodes. Test stimuli were delivered alternately through each stimulating electrode once every 15 s, i.e. each angular bundle was stimulated at 1/30 s. Baseline-evoked responses were obtained by application of test stimuli for 15-45 min prior to conditioning. Conditioning stimulation, consisting of 8 trains (1 train/10 s) of 8 pulses each applied at a rate of 400 Hz, was delivered through one stimulating electrode (the 'conditioned' side). Three or 4 sets of 8 conditioning trains were presented at 10 min intervals, depending upon the amount of conditioning observed after the third set of trains. Test stimulation was continued throughout the experiment. Responses were recorded on FM tape for subsequent off-line analysis by a Nicolet model 1070 signal averager. An animal was prepared for the morphological study if the amount of potentiation measured was at least 50 ~. The initial slope of the population EPSP was the response measure analyzed. The effect of conditioning was assessed by comparing averages of 4 responses before and after conditioning. Potentiation was computed as the difference between the percent change of the average response on the conditioned side and the control side. This inherently matched control procedure was continued through all phases of analysis. Four animals which were not prepared for morphological analysis were used for current source density analyses to define the relative distribution of active synapses across the laminae of the dentate molecular layer. The same stereotaxic coordinates were used for stimulating electrode placement here as for the conditioned animals. However, the placement of the recording electrodes was in the region typically selected for anatomical analysis. After establishing a baseline of evoked dentate granule cell population responses, the response was mapped at 20 /~m or 50 /~m (depending on the animal) increments between the dorsal leaf granule cell layer and the hippocampal fissure. An average of 2-5 responses (depending on the animal) at each location was used to obtain the second derivative of the response with respect to distance, or the one-dimensional current density at each location in the proximo-distal axis 28. After mapping, a baseline response was again measured in the granule cell layer
23 to be certain there had not been a shift in response magnitude. Fast Green was injected to localize the recording micropipettes in 2 animals. In the remaining 2 animals, cell firing was used to localize the recording micropipettes to the top of the granule cell layer.
Tissue preparation Animals were perfused 10 rain (n = 3) or 60 min (n -- 3) following the last set of conditioning trains. One of the following 3 fixative schedules was applied: (1) 2 ~ (v/v) glutaraldehyde-2 ~ (w/v) paraformaldehyde in 0.14 M sodium cacodylate buffer containing 1.5 m M MgSO4, pH 7.3 (n == 4); (2) the above with 2 ~ (w/v) p~)lyvinylpyrrolidone (n = 1)5; or (3) 1 ~ (w/v) paraformaldehyde-l.25 ~ (v/v) glutaraldehyde followed by 4 ~ (w/v) paraformaldehyde5 °/o / (v/v) glutaraldehyde, both in 0.14 M sodium cacodylate buffer containing 1.5 mM MgSO4, pH 7.3 (n = 1). Neither a buffer rinse, anticoagulants, nor vasodilators were included in any of the perfusion protocols. Following perfusion, the body was refrigerated for 4 h. The brain was then dissected from the skull and was placed in cold fixative overnight. No relevant differences were observed between animals sacrificed after a 10 min or 60 min delay or between the different fixative procedures. Hippocampi were carefully dissected from the cortex and coded. Lamellar 1,z3 sections 200-400 # m thick were cut on a tissue chopper from the rostral aspect of the hippocampus, approximately 1.6-2.4 mm posterior to bregma. Tissue blocks were rinsed in cacodylate buffer, osmicated, dehydrated through an ascending series of graded ethanol solutions, and were embedded in Araldite resin. Thick (1 #m) sections of the dentate gyrus were stained with toluidine blue or Azure II. Ultra-thin sections (80-90 nm thick) on formvar-coated slot grids were stained with methanolic uranyl acetate followed by aqueous lead citrate. Sections were viewed in an Hitachi HU-12A electron microscope at a saturated accelerating voltage of 100 kV. For each animal, two montages were photographed, one from each hippocampus. Each pair of montages was taken from the same anterior-posterior level of the hippocampus and approximately the same medio-lateral location within the dorsal or ventral leaf of the dentate gyrus. The medio-lateral
location of each montage was selected from viewing the corresponding thick sections with the light microscope. Four of the 6 cases included here involved dorsal leaf montages and the remaining 2 cases sampled from the ventral leaf. Each montage was rectangular in shape and contained no disturbing somata or blood vessels. Montages were photographed at 10,000× across the extent of the molecular layer, with a final magnification of 27,500 ×.
Data analysis To ensure unbiased data acquisition and analysis, various procedures were observed consistently. Each hippocampus was assigned a code number which was unrelated to any experimental variables. The code number was used throughout data analysis and was only decoded after the micrographs were quantified. Furthermore, micrographs comprising the control montage and the conditioned montage for each animal were thoroughly intermixed and scored as a single group of plates. The individuals sectioning, photographing, and scoring the tissue had no information on the conditioning history of the tissue. Several criteria helped to identify and to classify synaptic profiles. The necessary and sufficient conditions for synapse identification were: (1) apposed pre- and postsynaptic paramembranous densities where the postsynaptic density was thicker than its presynaptic complement; and (2) the presence of presynaptic vesicles near the presynaptic paramembranous density. A shaft synapse was defined as a dendrite with a density on its cytoplasmic face apposed by a vesicle-containing axon terminal. Spine synapses had a plate on the postsynaptic cytoplasmic face apposed by a vesicle-containing element. These spine profiles contained wispy, filamentous material extending from the density into the interior of the profile, but lacked microtubules and mitochondria. Some also contained a spine apparatus12,25,26. Spine synapses were further categorized by spine head shape as: (1) simple spine heads, when the ratio of the spine head's major and minor axes (parallel and perpendicular to the postsynaptic density) was less than 1.5:1 and when they were not concave (see below). These were generally small, rounded profiles;
24 (2) ellipsoid spine heads, when the axial ratio was equal to or greater than 1.5-1 and the spine heads were convex; (3) U-shaped spine heads which possessed varying degrees of U-ness, ranging from a slight (380 nm) inward dimple at the synaptic apposition to the apparent enclosure of the presynaptic element by the arms of the postsynaptic profile; and (4) spinule spine heads which were shaped like a W and generally possessed a split postsynaptic density26. Identifiable spinule profiles constituted less than 13 ~ of the total number of synapses on the control side. These last 2 categories were grouped together as the concave profile category. Spine profile size was quantified in two ways. Spine profile area and perimeter were measured using a Numonics digital planimeter or a Zeiss MOP-Videoplan image analysis system. Size was also approximated via measurement of the circle within which each profile fit using a Berol RapiDesign metric template (R-2040). This sizing reflects maximal profile diameter. The correlation between 'circle size' (diameter) and the planimeter perimeter of each spine profile was assessed in one animal using the Pearson correlation coefficient (SPSS-6000 version 8.0; ref. 19). For the concave profile category, the median Pearson correlation coefficient between 'circle size' and perimeter was 0.89. For the simple category it was 0.99 and for the ellipsoid category 0.93. Because these correlation coefficients were statistically significant (range of coefficients: 0.76-0.96) and the method more rapid than planimetry, 'circle size' was measured in all animals instead of perimeter. The Zeiss MOP-Videoplan system was used to measure the postsynaptic density (PSD) length for spine head and shaft synapse profiles. The dentate molecular layer varies in its proximodistal extent both within a section from one anteriorposterior level of the hippocampus and between sections from different anterior-posterior levels of the hippocampus. In our samples, the dorsal molecular layer within one section had an average range of 16 in its proximo-distal extent. In order to compensate for this variability and the further variability added by planes of section not perfectly perpendicular to the granule cell layer, the proximo-distal extent of the molecular layer was subdivided into laminae parallel to the granule cell layer. Although thirds were the
traditional divisions of the molecular layer when this study was initiated, we divided the molecular layer into sixths since these sectors were easily converted to thirds, yet offered higher resolution than thirds. The total area of the molecular layer montages was normalized to 1288 #m z. Level 1 refers to the most proximal sector within the molecular layer and level 6 to the most distal region. Level 2 is a mixed zone, containing commissural, associational, and entorhinal afferents to the molecular layer. The number of synapses was estimated per unit area (NA). In this paper, NA represents the number of synapses counted per x square micrometers of analyzed tissue where x represents 1288/zm 2 for the total area or 215/~m 2 for one level of the molecular layer. Changes in NA on the conditioned side (ANA) were calculated using the inherently matched control side of each animal (matched difference) to minimize between animal variability of fixation, anesthetic, etc. When NA and ANA were quantified as a function of spine head profile size, shaft synapses and connected spine profiles were excluded. These two categories together contributed less than 16 ~ of the total synapse number. The matched differences between the conditioned and control sides were statistically analyzed using inherently matched 2-tailed t-tests. In all cases, n equalled 6. RESULTS
Neurophysiology and location of conditioned synapses. The average amount of potentiation of the population EPSP was 58.3 ~ [± 24.2 (S.E.M.)] for the 6 animals included in the electron microscopic analyses. The stimulating electrode placement of this study activates a limited band of EC afferents, i.e. those forming the more proximally located EC-DG synapses 16. One-dimensional CSD analyses (Levy and Steward, unpublished observations) with stimulating electrodes placed directly in the medial EC confirm the proximal molecular layer location of the EC-DG synapses activated by medial EC stimulation. Although CSD maps were not generated in animals used for electron microscopic analysis (to
25 IN
aptic element, only one category shows significant changes in NA with conditioning. For the simple and ellipsoid spine populations, no significant changes in NA exist with conditioning ( - - 5 . 6 ~ and - - 2 . 6 ~ , respectively). However, the concave category shows a substantial (32.0 ~ ) and significant (t = 4.85, P < 0.05) increase of 47.4 concave synapses per 1288 F m 2 on the conditioned sides. Some exemplary concave synaptic profiles observed following conditioning are presented in Fig. 2. A trend (t = 2.71 P < 0.05) exists for an increase of shaft synapses. The increase of 18.6 synapses per 1288 # m s with conditioning reflects a 45.7 ~ change in shaft synapse density.
+ 20
L. -=
(J 0 0 > .m t~ rt-
- 20
our I
I
I
I
I
1
2
3
4
5
Level
of
Molecular
6
Layer
Fig. 1. Average, normalized current source density (CSD). This CSD map, based upon 2 experiments, illustrates the approximate location of the dentate synapses activated by angular bundle stimulation. Levels 1 through 6 refer to equal divisions of the proximo-distal extent of the molecular layer. Level I begins where the somatic layer ends, and level 6 ends at the hippocampal fissure. Inward current flow, indicating the region of greatest synaptic activation, is centered around level 3, within the more proximal range of the entorhinal projection to the dentate gyrus. avoid the tissue damage associated with multiple electrode penetrations), such maps were generated in 4 separate experiments using the usual electrode sites. The current sink (i.e. active excitatory synapses) is limited to a narrow region (20-80 #m) centered about 40 # m above the cell layer, i.e. in the most proximal zone of the E C - D G projection. Since several sources of inaccuracy remain using this method [e.g. the size of the dye spot, tissue swelling or shrinkage, and the resolution of the CSD method used (i.e. steps between recording sites)], the CSD map may be displaced by 4-50 #m. Fig. 1 depicts the average, normalized current source density for the 4 experiments. The primary region of synaptic activation, or the region of strongest activation, is centered around the lower portion of level 3.
NA as a function of synapse type When NA is subdivided by category of postsyn-
NA as a function of level of the molecular layer With conditioning, changes in the counted synaptic density (ANA = NA conditioned --NA control) were observed as a function of level. The increased NA in level 2 represents a mean change of 26.2 ~ (an increase of 18.8 synapses per 215 /zm 2, t = 2.42, n.s.). In level 4, a similar increase of 2 6 . 6 ~ exists (21.4 synapses per 215 # m 2, t = 2.61, P < 0.05). Most striking, however, is the conspicuous absence of such an increase and, instead, a slight trend for a decrease in NA with conditioning in level 3, the presumed region of primary synaptic activation (cf. Levy and Desmond, in preparation, for stereological considerations of these results). This change is a 3.5 ~ decrease (t ---- 0.37, n.s.). Outside the region of primary synaptic activation, no significant changes in NA are observed. Level 1 shows a 0.07 ~ increase in NA (t = 0.003, n.s.). Small increases of 10.5 ~ and 9.1 ~ exist in levels 5 and 6 respectively (t = 1.27, n.s.; t = 0.93, n.s.). NA as a function of synapse type and level NA also varied as a function of both synapse type and level of the molecular layer (see Fig. 3). The percent changes given below reflect type by type comparisons unless stated otherwise. Simple and ellipsoid spine heads Of note is the contrast between the changes seen in levels 2 and 3. Level 3 shows a decreased synaptic density and level 2 an increased synaptic density for these two categories. The significant changes in level 3 consist of a 44.4 ~ decrease in simple synapses (t
26
Fig. 2. Representative dendritic spine profiles observed in the dentate molecular layer. The concave spine profiles (asterisks) are of various sizes and degrees of concavity and typically increase in number with conditioning. The simple (Sire) and ellipsoid (El) profiles are usually smaller than the concave spine profiles, x 37,000.
27 +
+
Level 4
Level4
12o
20.0 80
10.0
40
ANA
o
1
I
r
i 4.0
10.0
8.0Sh
Sire
El
Con
÷
Level 3
Level3
12.0 •
20.0
8.0-
4010.0
L_J AN A
o
I
I
10.0
Sh
4.0
lJl Sim
El
8.0-
12.0 -
Con
+
-4-
Level 2 16.0 -
20.0 12.0 •
10.0
~l
Level2
8.0
4.0
&N A 10.0
0
4.0 Sh
Sire
El
Con
Fig. 3. Change in the number of counted synapses (ANA) per 215 /zm 2 as a function of profile type and level of the molecular layer. ANA values for shaft (Sh), simple (Sire), ellipsoid (El), and concave (Con) spine profiles are depicted for levels 2, 3, and 4. Positive ANA values reflect more synapses on the conditioned side; negative ANA values indicate fewer synapses on the conditioned side. Of particular interest are the increased NA for the simple and ellipsoid spine categories in level 2 and the increase in NA for the concave category in levels 3 and 4 with conditioning. The changes in NA for the shaft synapse category parallel the changes across level. No statistically significant changes in NA as a function of profile type are observed in levels 1, 5, and 6 with conditioning.
= 3.69, P < 0.05) and 44.6 ~ decrease in ellipsoid synapses (t ~-- 5.49, P < 0.05). In level 2, simple synapses increase by 22.4~o (t ----- 0.90, n.s.) and ellipsoid synapses by 41.2 ~ (t ----- 1.35, n.s.). In level
80
042 02, °38 0,, o~g 0,0 0 2 0 3 , PROFILE
DIAMETER
(pm)
Fig. 4. Change in number of counted synapses (ANA) per 215 /~m2 as a function of profile diameter ~ m ) in levels 2, 3, and 4. If ANA is positive, then more synapses are observed on the conditioned side. A negative ANA value means fewer synapses on the conditioned side. There are few changes in NA in level 2 with conditioning except for a statistically significant increase in small profiles. In level 3, two changes occur - - the number of small profiles decreases and the number of large profiles increases.
4, the changes are of a minor nature for both simple and ellipsoid synapses (4.8 ~ increase, t = 0.30, n.s.; 10.6 ~ decrease, t = 1.46, n.s., respectively).
28
Level 6
+ 80-
Concave spine heads
40Z~NA
increases in NA in levels 2 (87.0 ~ increase, t = 1.82) and 4 (49.4~ increase, t = 0.83).
r--~ll
0 4.0
8.0-
Level 5 + 8.04.0•A.NA
4.0
The concave spine category not only shows an overall increase on the conditioned sides, but also a very large increase specifically within level 3, the zone of primary synaptic activation. Here, there is a 80.9 ~ increase, viz. 22.43 synapses over the control number of 27.73 (t = 4.22, P < 0.05). Another way of evaluating this change in level 3 is to relate it to the total number of synapses. On the control sides, 3 0 . 4 ~ of the total counted synapses are concave while this percentage increases to 5 7 . 6 ~ on the conditioned sides. In level 4, the concave synapses increase 66.2 ~ over the baseline control number (t = 4.85, P < 0.05).
8.0-
Level 1
+ 8.0. 4.0-
~N~
n
0
F-1
U
4.0 80_
'
' ' ' ' '
i'
!
!o'
0.12 0.24 0.36 0.47 059 070 0.82 093 1 5 <0.12 0.~23 0~35 0~16 0~58 0~69 0~'81 0~)2 1.04\ 1"15 >1.15 PROFILE DIAMETER
(pm)
Fig. 5. Change in number of spine profiles (ANA) per 215/~m2 as a function of size (profile diameter,/~m) and level of the molecular layer. In contrast with levels 2, 3, and 4 of the molecular layer where the relative size distribution of spines is altered with conditioning (see Fig. 4), no obvious changes occur in the size distribution of spine profiles in levels 1, 5, and 6.
Shaft synapses Shaft synapse density by level parallels the trend for all synapse types combined, with a nonsignificant decrease in NA Of 2 5 . 4 ~ (t = 0.84) within level 3 surrounded by large, but statistically nonsignificant,
Changes in spine head size Our unpublished observations show that the concave spine head category contains most of the larger (both perimeter and volume) spine heads of the total spine head population sampled (612 spine profiles). Consistent with this observation are the data of Fig. 4, level 3. Furthermore, Fig. 4 implies that the increased spine head size in level 3 is at the apparent expense of the number of small spine profiles. In levels 2 and 4, in contrast, the number of small spine profiles increases following conditioning (for the 0.12-0.30/zm diameter range, t ---- 2.71, P < 0.05 for level 2; t = 1.16, n.s. for level 4). Following conditioning, there are few changes in the relative distribution by profile size in levels 1, 5, and 6 (see Fig. 5).
Postsynaptic density length The PSD is thought to indicate the activated region of the synaptic interface. Alterations in PSD size may thus be of functional significance. Table I presents the mean PSD lengths for each synapse type in levels 2, 3, and 4. As may be inferred from the spine head size data above, the U-shaped and spinule spine profiles have the longest measured PSDs within all levels. DISCUSSION One notable result of this study is the increased
29 TABLE I Mean observed length (nm) of postsynaptic densities. PSD length was averaged across 6 animals for each category of synapse type by level of the molecular layer and conditioning history
Shaft Simple Ellipsoid U-shaped Spinule
Conditioned Control Conditioned Control Conditioned Control Conditioned Control Conditioned Control
Level 2
Level 3
Level 4
230 240 152 158 182 196 314 302 349 421
220 180 132 153 170 180 245 250 327 374
200 180 158 146 155 165 220 241 266 363
number of large, concave postsynaptic spine heads, localized to the region of primary synaptic activation, following conditioning. It is easily imagined that these larger synapses with the larger area of their PSDs mediate the enhanced synaptic function that was measured. In fact, the actual spine head perimeter increases are considerably larger than the 'circle size' data (Fig. 4) indicate since the factor converting 'circle size' to perimeter is much greater than :r for the concave spine head category (it is 5.1 for the concave category rather than the 3.14 which nearly approximates the factor for the simple and ellipsoid categories). This result is consistent with the Fifkova and Van Harreveld experimentsS, 27 that used a different fixation method. There, EC stimulation produced increases in the size of dentate spine head profiles in the presumed region of synaptic activation. Another significant result of the present experiments is that numerous small spine profiles are situated in regions just peripheral to the central region of synaptic activation, with its many large concave spine heads. It is possible that these small synapses are newly formed and in the process of 'growing up'. On the other hand, since: (1) synaptic
REFERENCES 1 Andersen, P., Bliss, T. V. P. and Skrede, K. K., Lamellar organization of hippocampal excitatory pathways, Exp. Brain Res., 13 (1971) 222-238. 2 Anderson, J. A., Parallel computation with simple neural networks, Cognition and Brain Theory, 3 (1979) 45-53.
inactivity in the presence of nearby potentiation (ref. 15; Levy and Steward, unpublished observations) produces synaptic depression; and (2) since the conditioning electrode tmdoubtedley does not activate all the synapses within the activated zone, it is also possible that these smaller synapses were larger synapses before conditioning. High-frequency activation of sufficient nearby synapses induced synaptic depression at these synapses, reflected in and mediated by a decreased synaptic area. ACKNOWLEDGEMENTS This work was supported in part by N I H Grant R01-NS15488 to W . B . L . N . L . D . was supported in part by N I H Predoctoral Fellowship MH05677 and by the Department of Neurological Surgery. We thank Dr. O. Steward for his collaboration on the initial observation of concave synapses and for his advice and encouragement throughout the study. We also acknowledge the expert technical assistance of S. Vinsant and T. Gill, the secretarial aid of M. P. Janssen, and the support and encouragement of Dr. John A. Jane.
3 Baudry, M., 0liver, M., Creager, R., Wieraszko, A. and Lynch, G., Increase in glutamate receptors following repetitive electrical stimulation in hippocampal slices, Life Sci., 27 (1980) 325-330. 4 Bliss, T. V. P. and Lomo, T., Long-lasting potentiation of synaptic transmission in the dentate area of the anaesthetized rabbit following stimulation of the perforant path, J. Physiol. (Lond.), 232 (1973) 331-356.
30 5 Bohmann, S. O. and Maunsbach, A. B., Effects on tissue fine structure of glutaraldehyde fixatives, J. ultrastruct. Res., 30 (1970) 195-208. 6 Cooper, L. N., Lieberman, F. and Oja, E., A theory for the acquisition and loss of neuron specificity in visual cortex, BioL Cybernet., 33 (1979) 9-28. 7 Desmond, N. L. and Levy, W. B., Ultrastructural and numerical changes in dendritic spines as a consequence of long-term potentiation, Anat. Rec., 199 (1981) 68A69A. 8 Fifkova, E. and Van Harreveld, A., Long-lasting morphological changes in dendritic spines of dentate granular cells following stimulation of the entorhinal area, J. NeurocytoL, 6 (1977) 211-230. 9 Globus, A., Brain morphology as a function of presynaptic morphology and activity. In A. H. Riesen (Ed.), The Developmental Neuropsychology of Sensory Deprivation, Academic Press, New York, 1975, pp. 9-91. 10 Hebb, D. O., The Organization of Behavior, Wiley, New York, 1949, 335 pp. 11 Kohonen, T., Associative Memory. A System-Theoretical Approach, Springer, New York, 1977, 176 pp. 12 Laatsch, R. H. and Cowan, W. M., Electron microscopic studies of the dentate gyrus of the rat. I. Normal structure with special reference to synaptic organization, J. comp. Neurol., 128 (1966) 359-396. 13 Lee, K. S., Schottler, F., Oliver, M. and Lynch, G., Brief bursts of high-frequency stimulation produce two types of structural change in rat hippocampus, J. NeurophysioL, 44 (1980) 247-258. 14 Levy, W. B., Associative changes at the synapse: LTP in the hippocampus. In W. B. Levy, J. Anderson and S. Lehmkuhle (Eds.), Synaptic Modification, Neuron Selectivity and Nervous System Organization, Lawrence Erlbaum, Hillsdale, N J, in press. 15 Levy, W. B. and Steward, O., Synapses as associative memory elements in the hippocampal formation, Brain Research, 175 (1979) 233-245. 16 McNaughton, B. L. and Barnes, C. A., Physiological identification and analysis of dentate granule cell responses to stimulation of the medial and lateral perforant pathways in the rat, J. comp. NeuroL, 175 (1977) 439-454. 17 McNaughton, B. L., Douglas, R. M. and Goddard, G. V., Synaptic enhancement in fascia dentata: cooperative among coactive afferents, Brain Research, 157 (1978) 277-293. 18 Marr, D., A theory of cerebellar cortex, J. PhysioL
(Lond.), 202 (1969) 437--470. 19 Nie, N. H., Hull, C. H., Jenkins, J. G., Steinbrenner, K. and Bent, D. H., SPSS: Statistical Package for the Social Sciences, 2nd Edn., McGraw-Hill, New York, 1975, 675 pp. 20 Rutledge, L. T., Synaptogenesis" effects of synaptic use. In M. R. Rosenzweig and E. L. Bennett (Eds.), Neural Mechanisms of Learning and Memory, MIT Press, Cambridge, MA, 1976, pp. 329-339. 21 Rutledge, L. T., The effects of denervation and stimulation upon synaptic ultrastructure, J. comp. NeuroL, 178 (1978) 117-128. 22 Steward, O. and Scoville, S. A., Cells of origin of entorhinal cortical afferents to the hippocampus and fascia dentata of the rat, J. comp. NeuroL, 169 (1976) 347-370. 23 Struble, R. G., Desmond, N. L. and Levy, W. B., Anatomical evidence for interlamellar inhibition in the fascia dentata, Brain Research, 152 (1978) 580-585. 24 Tanzi, E., I Fatti i le induzioni nelrodierna istologia del sistema nervoso, Riv. sperim, freniatria medic, leg., 19 (1893) 419-472. 25 Tarrant, S. B. and Routtenberg, A., The synaptic spinule in the dendritic spine: electron microscopic study of the hippocampal dentate gyrus, Tiss. Cell, 9 (1977) 461-473. 26 Tarrant, S. B. and Routtenberg, A., Postsynaptic membrane and spine apparatus: proximity in dendritic spines, Neurosci. Lett., 11 (1979)289-294. 27 Van Harreveld, A. and Fifkova, E., Swelling of dendritic spines in the fascia dentata after stimulation of the perforant fibers as a mechanism of post-tetanic potentiation, Exp. Neurol., 49 (1975) 736-739. 28 Wilson, R. C., Functional Consequences of the Morphological and Physiological Modification of Synapses and Circuits in the Hippocampal Formation of the Rat, Unpublished doctoral dissertation, University of Virginia, 1981. 29 Wilson, R.~C., Levy, W. B. and Steward, O., Functional effects of lesion-induced plasticity: long-term potentiation in normal and lesion-induced temporodentate connections, Brain Research, 176 (1979) 65-78. 30 Wilson, R. C., Levy, W. B. and Steward, O., Changes in the translation of synaptic excitation to dentate granule cell discharge accompanying long term potentiation. II. An evaluation of mechanisms utilizing the dentate gyrus dually innervated by surviving ipsilateral and sprouted crossed temporodentate inputs, J. NeurophysioL, 46 (1981) 339-355.
21
Synaptic Correlates of Associative Potentiation/Depression: an Ultrastructural Study in the Hippocampus NANCY L. DESMOND and WILLIAM B. LEVY* Department of Neurological Surgery, Box 420, University of Virginia School of Medicine, Charlottesville, Vii 22908 (U.S.A.) (Accepted September 14th, 1982) Key words: long-term potentiation - - hippocampus - - competition - - associative potentiation/depression
Brief high-frequency trains delivered to the monosynaptic entorhinal cortical input to the dentate gyrus result in both increases and decreases of synaptic strength as a function of whether a particular afferent is active during conditioning (associative potentiation/depression15,8°). The present report concerns the effect of such brief, high-frequency conditioning trains upon the asymmetric synapses of the rat dentate gyrus molecular layer. Only those animals whose responses increased at least 50 ~ following conditioning stimulation were included in the study. Additional animals were used for one-dimensional current source density analyses to localize the activated synaptic region. Double blind scoring procedures were used to classify and quantify electron micrographic data. Asymmetric synapses were scored as a function of their position in the molecular layer, spine head size and shape, and postsynaptic density length. All data were treated as inherently matched comparisons between the conditioned and control sides of each animal. The number of large, concave spine synapses with large postsynaptic densities significantly increases in the central zone of synaptic activation. Bordering this zone are regions with increases in synaptic number following conditioning, primarily due to an increased number of small spine synapses. The increased number of large, concave spine synapses in the central zone is postulated to mediate associative potentiation. The many small spine heads just adjacent to the zone of strongest synaptic activation may reflect synaptic depression evoked at synapses inactive during conditioning. INTRODUCTION Since the inception o f the n e u r o n doctrine, numerous neuroscientists have suspected or assumed that modification o f synaptic connections mediates m e m o r y storage2,6,10,11,18, 24. Recent experiments lend support to this assumption. Morphological alterations o f synapses, as well as apparent numerical changes, occur as a consequence o f experience (e.g. environmental manipulations 9) and o f artificially-induced neuronal use (e.g. electrical stimulationS,S,13,20,21,~7). The initial discovery o f long-term potentiation 4 (LTP) in the h i p p o c a m p u s provides the first clear neurophysiological evidence for a neuronal analog o f m e m o r y storage. T o d a y , this experimental model is an even better analog since L T P as studied at the entorhinal synapses o f the dentate g y m s is an associative p h e n o m e n o n , dependent u p o n the cor* To whom correspondence should be addressed. 0006-8993/83/0000-0000/$03.00 © 1983 Elsevier Science Publishers
related activity o f convergent, m o n o s y n a p t i c excitatory afferents15,17. Here, the c o m b i n e d co-activity o f a presynaptic input and sufficient excitation o f a postsynaptic cell increases the synaptic efficacy o f the particular synapses involved (associative potentiation). The same cell that exhibits potentiation m a y concurrently exhibit the p h e n o m e n o n o f depression, an erasure-like, long-term process that opposes potentiation. In the case o f depression, inactivity o f a particular presynaptic afferent concomitant with strong activity o f converging excitatory afferents decreases the synaptic efficacy o f the synapse formed by the inactive afferent u p o n the postsynaptic cell. Finally, physiological evidence, c o m b i n e d with the inherent controls existing in this monosynaptic, bilaterally projecting system, argue for a synaptic locus (including possibly spine stem changes) for these alterations (see refs. 14, 15, 29, 30 for evidence).
22 Previous investigations correlating morphological changes produced by brief high-frequency trains expected and sought only correlates of potentiation, not of depression. On the other hand, the present investigation seeks correlated morphological changes that could logically mediate both potentiation and depression in the entorhinal cortexdentate gyrus (EC-DG) system. The conclusions of Levy and Steward 15 and Wilson et al.Z9,30 guide the search and interpretations of any anatomical changes observed after conditioning. Taking advantage of the laminated distribution of the EC-DG afferents22, the present study finds morphological correlates that logically explain and may be the cellular basis of associative potentiation/ depression. A preliminary report has appeared 7. MATERIALS AND METHODS
Experiments were performed on 300-400 g, male Sprague-Dawley derived rats (Flow Laboratories). Each animal was initially anesthetized with chloralose (55 mg/kg) and urethane (0.2-0.4 g/kg). Anesthesia was maintained with supplemental doses of urethane as needed over the course of the experiment. Body temperature was maintained at 37 ~ 1 °C.
Electrophysiological methods These were comparable to those methods previously describedlL Twisted wire, bipolar stimulating electrodes were positioned bilaterally in the angular bundle (8.1 mm posterior to bregma, 4.4 mm lateral to midline). Glass recording micropipettes filled with 0.9 ~ (w/v) NaC1 were advanced bilaterally into the anterior portion of the dentate gyrus at a 5° angle away from the midline (2.8 mm posterior to bregma, 1.3-1.5 mm lateral to midline). These micropipettes were positioned between the dorsal and ventral leaves of the dentate gyrus and at its lateral edge so as to avoid tissue damage in the region to be analyzed morphologically. The exposed brain surface was covered with mineral oil until all electrodes were situated at the depth evoking maximal dorsal leaf granule cell population responses. A warm 5 ~ (w/v) Agar solution in 0.9 ~ (w/v) NaC1 was then placed over the skull to minimize edema. Monophasic constant current pulses of 300 /zs
duration were delivered through the angular bundle stimulating electrodes. Test stimuli were delivered alternately through each stimulating electrode once every 15 s, i.e. each angular bundle was stimulated at 1/30 s. Baseline-evoked responses were obtained by application of test stimuli for 15-45 min prior to conditioning. Conditioning stimulation, consisting of 8 trains (1 train/10 s) of 8 pulses each applied at a rate of 400 Hz, was delivered through one stimulating electrode (the 'conditioned' side). Three or 4 sets of 8 conditioning trains were presented at 10 min intervals, depending upon the amount of conditioning observed after the third set of trains. Test stimulation was continued throughout the experiment. Responses were recorded on FM tape for subsequent off-line analysis by a Nicolet model 1070 signal averager. An animal was prepared for the morphological study if the amount of potentiation measured was at least 50 ~. The initial slope of the population EPSP was the response measure analyzed. The effect of conditioning was assessed by comparing averages of 4 responses before and after conditioning. Potentiation was computed as the difference between the percent change of the average response on the conditioned side and the control side. This inherently matched control procedure was continued through all phases of analysis. Four animals which were not prepared for morphological analysis were used for current source density analyses to define the relative distribution of active synapses across the laminae of the dentate molecular layer. The same stereotaxic coordinates were used for stimulating electrode placement here as for the conditioned animals. However, the placement of the recording electrodes was in the region typically selected for anatomical analysis. After establishing a baseline of evoked dentate granule cell population responses, the response was mapped at 20 /~m or 50 /~m (depending on the animal) increments between the dorsal leaf granule cell layer and the hippocampal fissure. An average of 2-5 responses (depending on the animal) at each location was used to obtain the second derivative of the response with respect to distance, or the one-dimensional current density at each location in the proximo-distal axis 28. After mapping, a baseline response was again measured in the granule cell layer
23 to be certain there had not been a shift in response magnitude. Fast Green was injected to localize the recording micropipettes in 2 animals. In the remaining 2 animals, cell firing was used to localize the recording micropipettes to the top of the granule cell layer.
Tissue preparation Animals were perfused 10 rain (n = 3) or 60 min (n -- 3) following the last set of conditioning trains. One of the following 3 fixative schedules was applied: (1) 2 ~ (v/v) glutaraldehyde-2 ~ (w/v) paraformaldehyde in 0.14 M sodium cacodylate buffer containing 1.5 m M MgSO4, pH 7.3 (n == 4); (2) the above with 2 ~ (w/v) p~)lyvinylpyrrolidone (n = 1)5; or (3) 1 ~ (w/v) paraformaldehyde-l.25 ~ (v/v) glutaraldehyde followed by 4 ~ (w/v) paraformaldehyde5 °/o / (v/v) glutaraldehyde, both in 0.14 M sodium cacodylate buffer containing 1.5 mM MgSO4, pH 7.3 (n = 1). Neither a buffer rinse, anticoagulants, nor vasodilators were included in any of the perfusion protocols. Following perfusion, the body was refrigerated for 4 h. The brain was then dissected from the skull and was placed in cold fixative overnight. No relevant differences were observed between animals sacrificed after a 10 min or 60 min delay or between the different fixative procedures. Hippocampi were carefully dissected from the cortex and coded. Lamellar 1,z3 sections 200-400 # m thick were cut on a tissue chopper from the rostral aspect of the hippocampus, approximately 1.6-2.4 mm posterior to bregma. Tissue blocks were rinsed in cacodylate buffer, osmicated, dehydrated through an ascending series of graded ethanol solutions, and were embedded in Araldite resin. Thick (1 #m) sections of the dentate gyrus were stained with toluidine blue or Azure II. Ultra-thin sections (80-90 nm thick) on formvar-coated slot grids were stained with methanolic uranyl acetate followed by aqueous lead citrate. Sections were viewed in an Hitachi HU-12A electron microscope at a saturated accelerating voltage of 100 kV. For each animal, two montages were photographed, one from each hippocampus. Each pair of montages was taken from the same anterior-posterior level of the hippocampus and approximately the same medio-lateral location within the dorsal or ventral leaf of the dentate gyrus. The medio-lateral
location of each montage was selected from viewing the corresponding thick sections with the light microscope. Four of the 6 cases included here involved dorsal leaf montages and the remaining 2 cases sampled from the ventral leaf. Each montage was rectangular in shape and contained no disturbing somata or blood vessels. Montages were photographed at 10,000× across the extent of the molecular layer, with a final magnification of 27,500 ×.
Data analysis To ensure unbiased data acquisition and analysis, various procedures were observed consistently. Each hippocampus was assigned a code number which was unrelated to any experimental variables. The code number was used throughout data analysis and was only decoded after the micrographs were quantified. Furthermore, micrographs comprising the control montage and the conditioned montage for each animal were thoroughly intermixed and scored as a single group of plates. The individuals sectioning, photographing, and scoring the tissue had no information on the conditioning history of the tissue. Several criteria helped to identify and to classify synaptic profiles. The necessary and sufficient conditions for synapse identification were: (1) apposed pre- and postsynaptic paramembranous densities where the postsynaptic density was thicker than its presynaptic complement; and (2) the presence of presynaptic vesicles near the presynaptic paramembranous density. A shaft synapse was defined as a dendrite with a density on its cytoplasmic face apposed by a vesicle-containing axon terminal. Spine synapses had a plate on the postsynaptic cytoplasmic face apposed by a vesicle-containing element. These spine profiles contained wispy, filamentous material extending from the density into the interior of the profile, but lacked microtubules and mitochondria. Some also contained a spine apparatus12,25,26. Spine synapses were further categorized by spine head shape as: (1) simple spine heads, when the ratio of the spine head's major and minor axes (parallel and perpendicular to the postsynaptic density) was less than 1.5:1 and when they were not concave (see below). These were generally small, rounded profiles;
24 (2) ellipsoid spine heads, when the axial ratio was equal to or greater than 1.5-1 and the spine heads were convex; (3) U-shaped spine heads which possessed varying degrees of U-ness, ranging from a slight (380 nm) inward dimple at the synaptic apposition to the apparent enclosure of the presynaptic element by the arms of the postsynaptic profile; and (4) spinule spine heads which were shaped like a W and generally possessed a split postsynaptic density26. Identifiable spinule profiles constituted less than 13 ~ of the total number of synapses on the control side. These last 2 categories were grouped together as the concave profile category. Spine profile size was quantified in two ways. Spine profile area and perimeter were measured using a Numonics digital planimeter or a Zeiss MOP-Videoplan image analysis system. Size was also approximated via measurement of the circle within which each profile fit using a Berol RapiDesign metric template (R-2040). This sizing reflects maximal profile diameter. The correlation between 'circle size' (diameter) and the planimeter perimeter of each spine profile was assessed in one animal using the Pearson correlation coefficient (SPSS-6000 version 8.0; ref. 19). For the concave profile category, the median Pearson correlation coefficient between 'circle size' and perimeter was 0.89. For the simple category it was 0.99 and for the ellipsoid category 0.93. Because these correlation coefficients were statistically significant (range of coefficients: 0.76-0.96) and the method more rapid than planimetry, 'circle size' was measured in all animals instead of perimeter. The Zeiss MOP-Videoplan system was used to measure the postsynaptic density (PSD) length for spine head and shaft synapse profiles. The dentate molecular layer varies in its proximodistal extent both within a section from one anteriorposterior level of the hippocampus and between sections from different anterior-posterior levels of the hippocampus. In our samples, the dorsal molecular layer within one section had an average range of 16 in its proximo-distal extent. In order to compensate for this variability and the further variability added by planes of section not perfectly perpendicular to the granule cell layer, the proximo-distal extent of the molecular layer was subdivided into laminae parallel to the granule cell layer. Although thirds were the
traditional divisions of the molecular layer when this study was initiated, we divided the molecular layer into sixths since these sectors were easily converted to thirds, yet offered higher resolution than thirds. The total area of the molecular layer montages was normalized to 1288 #m z. Level 1 refers to the most proximal sector within the molecular layer and level 6 to the most distal region. Level 2 is a mixed zone, containing commissural, associational, and entorhinal afferents to the molecular layer. The number of synapses was estimated per unit area (NA). In this paper, NA represents the number of synapses counted per x square micrometers of analyzed tissue where x represents 1288/zm 2 for the total area or 215/~m 2 for one level of the molecular layer. Changes in NA on the conditioned side (ANA) were calculated using the inherently matched control side of each animal (matched difference) to minimize between animal variability of fixation, anesthetic, etc. When NA and ANA were quantified as a function of spine head profile size, shaft synapses and connected spine profiles were excluded. These two categories together contributed less than 16 ~ of the total synapse number. The matched differences between the conditioned and control sides were statistically analyzed using inherently matched 2-tailed t-tests. In all cases, n equalled 6. RESULTS
Neurophysiology and location of conditioned synapses. The average amount of potentiation of the population EPSP was 58.3 ~ [± 24.2 (S.E.M.)] for the 6 animals included in the electron microscopic analyses. The stimulating electrode placement of this study activates a limited band of EC afferents, i.e. those forming the more proximally located EC-DG synapses 16. One-dimensional CSD analyses (Levy and Steward, unpublished observations) with stimulating electrodes placed directly in the medial EC confirm the proximal molecular layer location of the EC-DG synapses activated by medial EC stimulation. Although CSD maps were not generated in animals used for electron microscopic analysis (to
25 IN
aptic element, only one category shows significant changes in NA with conditioning. For the simple and ellipsoid spine populations, no significant changes in NA exist with conditioning ( - - 5 . 6 ~ and - - 2 . 6 ~ , respectively). However, the concave category shows a substantial (32.0 ~ ) and significant (t = 4.85, P < 0.05) increase of 47.4 concave synapses per 1288 F m 2 on the conditioned sides. Some exemplary concave synaptic profiles observed following conditioning are presented in Fig. 2. A trend (t = 2.71 P < 0.05) exists for an increase of shaft synapses. The increase of 18.6 synapses per 1288 # m s with conditioning reflects a 45.7 ~ change in shaft synapse density.
+ 20
L. -=
(J 0 0 > .m t~ rt-
- 20
our I
I
I
I
I
1
2
3
4
5
Level
of
Molecular
6
Layer
Fig. 1. Average, normalized current source density (CSD). This CSD map, based upon 2 experiments, illustrates the approximate location of the dentate synapses activated by angular bundle stimulation. Levels 1 through 6 refer to equal divisions of the proximo-distal extent of the molecular layer. Level I begins where the somatic layer ends, and level 6 ends at the hippocampal fissure. Inward current flow, indicating the region of greatest synaptic activation, is centered around level 3, within the more proximal range of the entorhinal projection to the dentate gyrus. avoid the tissue damage associated with multiple electrode penetrations), such maps were generated in 4 separate experiments using the usual electrode sites. The current sink (i.e. active excitatory synapses) is limited to a narrow region (20-80 #m) centered about 40 # m above the cell layer, i.e. in the most proximal zone of the E C - D G projection. Since several sources of inaccuracy remain using this method [e.g. the size of the dye spot, tissue swelling or shrinkage, and the resolution of the CSD method used (i.e. steps between recording sites)], the CSD map may be displaced by 4-50 #m. Fig. 1 depicts the average, normalized current source density for the 4 experiments. The primary region of synaptic activation, or the region of strongest activation, is centered around the lower portion of level 3.
NA as a function of synapse type When NA is subdivided by category of postsyn-
NA as a function of level of the molecular layer With conditioning, changes in the counted synaptic density (ANA = NA conditioned --NA control) were observed as a function of level. The increased NA in level 2 represents a mean change of 26.2 ~ (an increase of 18.8 synapses per 215 /zm 2, t = 2.42, n.s.). In level 4, a similar increase of 2 6 . 6 ~ exists (21.4 synapses per 215 # m 2, t = 2.61, P < 0.05). Most striking, however, is the conspicuous absence of such an increase and, instead, a slight trend for a decrease in NA with conditioning in level 3, the presumed region of primary synaptic activation (cf. Levy and Desmond, in preparation, for stereological considerations of these results). This change is a 3.5 ~ decrease (t ---- 0.37, n.s.). Outside the region of primary synaptic activation, no significant changes in NA are observed. Level 1 shows a 0.07 ~ increase in NA (t = 0.003, n.s.). Small increases of 10.5 ~ and 9.1 ~ exist in levels 5 and 6 respectively (t = 1.27, n.s.; t = 0.93, n.s.). NA as a function of synapse type and level NA also varied as a function of both synapse type and level of the molecular layer (see Fig. 3). The percent changes given below reflect type by type comparisons unless stated otherwise. Simple and ellipsoid spine heads Of note is the contrast between the changes seen in levels 2 and 3. Level 3 shows a decreased synaptic density and level 2 an increased synaptic density for these two categories. The significant changes in level 3 consist of a 44.4 ~ decrease in simple synapses (t
26
Fig. 2. Representative dendritic spine profiles observed in the dentate molecular layer. The concave spine profiles (asterisks) are of various sizes and degrees of concavity and typically increase in number with conditioning. The simple (Sire) and ellipsoid (El) profiles are usually smaller than the concave spine profiles, x 37,000.
27 +
+
Level 4
Level4
12o
20.0 80
10.0
40
ANA
o
1
I
r
i 4.0
10.0
8.0Sh
Sire
El
Con
÷
Level 3
Level3
12.0 •
20.0
8.0-
4010.0
L_J AN A
o
I
I
10.0
Sh
4.0
lJl Sim
El
8.0-
12.0 -
Con
+
-4-
Level 2 16.0 -
20.0 12.0 •
10.0
~l
Level2
8.0
4.0
&N A 10.0
0
4.0 Sh
Sire
El
Con
Fig. 3. Change in the number of counted synapses (ANA) per 215 /zm 2 as a function of profile type and level of the molecular layer. ANA values for shaft (Sh), simple (Sire), ellipsoid (El), and concave (Con) spine profiles are depicted for levels 2, 3, and 4. Positive ANA values reflect more synapses on the conditioned side; negative ANA values indicate fewer synapses on the conditioned side. Of particular interest are the increased NA for the simple and ellipsoid spine categories in level 2 and the increase in NA for the concave category in levels 3 and 4 with conditioning. The changes in NA for the shaft synapse category parallel the changes across level. No statistically significant changes in NA as a function of profile type are observed in levels 1, 5, and 6 with conditioning.
= 3.69, P < 0.05) and 44.6 ~ decrease in ellipsoid synapses (t ~-- 5.49, P < 0.05). In level 2, simple synapses increase by 22.4~o (t ----- 0.90, n.s.) and ellipsoid synapses by 41.2 ~ (t ----- 1.35, n.s.). In level
80
042 02, °38 0,, o~g 0,0 0 2 0 3 , PROFILE
DIAMETER
(pm)
Fig. 4. Change in number of counted synapses (ANA) per 215 /~m2 as a function of profile diameter ~ m ) in levels 2, 3, and 4. If ANA is positive, then more synapses are observed on the conditioned side. A negative ANA value means fewer synapses on the conditioned side. There are few changes in NA in level 2 with conditioning except for a statistically significant increase in small profiles. In level 3, two changes occur - - the number of small profiles decreases and the number of large profiles increases.
4, the changes are of a minor nature for both simple and ellipsoid synapses (4.8 ~ increase, t = 0.30, n.s.; 10.6 ~ decrease, t = 1.46, n.s., respectively).
28
Level 6
+ 80-
Concave spine heads
40Z~NA
increases in NA in levels 2 (87.0 ~ increase, t = 1.82) and 4 (49.4~ increase, t = 0.83).
r--~ll
0 4.0
8.0-
Level 5 + 8.04.0•A.NA
4.0
The concave spine category not only shows an overall increase on the conditioned sides, but also a very large increase specifically within level 3, the zone of primary synaptic activation. Here, there is a 80.9 ~ increase, viz. 22.43 synapses over the control number of 27.73 (t = 4.22, P < 0.05). Another way of evaluating this change in level 3 is to relate it to the total number of synapses. On the control sides, 3 0 . 4 ~ of the total counted synapses are concave while this percentage increases to 5 7 . 6 ~ on the conditioned sides. In level 4, the concave synapses increase 66.2 ~ over the baseline control number (t = 4.85, P < 0.05).
8.0-
Level 1
+ 8.0. 4.0-
~N~
n
0
F-1
U
4.0 80_
'
' ' ' ' '
i'
!
!o'
0.12 0.24 0.36 0.47 059 070 0.82 093 1 5 <0.12 0.~23 0~35 0~16 0~58 0~69 0~'81 0~)2 1.04\ 1"15 >1.15 PROFILE DIAMETER
(pm)
Fig. 5. Change in number of spine profiles (ANA) per 215/~m2 as a function of size (profile diameter,/~m) and level of the molecular layer. In contrast with levels 2, 3, and 4 of the molecular layer where the relative size distribution of spines is altered with conditioning (see Fig. 4), no obvious changes occur in the size distribution of spine profiles in levels 1, 5, and 6.
Shaft synapses Shaft synapse density by level parallels the trend for all synapse types combined, with a nonsignificant decrease in NA Of 2 5 . 4 ~ (t = 0.84) within level 3 surrounded by large, but statistically nonsignificant,
Changes in spine head size Our unpublished observations show that the concave spine head category contains most of the larger (both perimeter and volume) spine heads of the total spine head population sampled (612 spine profiles). Consistent with this observation are the data of Fig. 4, level 3. Furthermore, Fig. 4 implies that the increased spine head size in level 3 is at the apparent expense of the number of small spine profiles. In levels 2 and 4, in contrast, the number of small spine profiles increases following conditioning (for the 0.12-0.30/zm diameter range, t ---- 2.71, P < 0.05 for level 2; t = 1.16, n.s. for level 4). Following conditioning, there are few changes in the relative distribution by profile size in levels 1, 5, and 6 (see Fig. 5).
Postsynaptic density length The PSD is thought to indicate the activated region of the synaptic interface. Alterations in PSD size may thus be of functional significance. Table I presents the mean PSD lengths for each synapse type in levels 2, 3, and 4. As may be inferred from the spine head size data above, the U-shaped and spinule spine profiles have the longest measured PSDs within all levels. DISCUSSION One notable result of this study is the increased
29 TABLE I Mean observed length (nm) of postsynaptic densities. PSD length was averaged across 6 animals for each category of synapse type by level of the molecular layer and conditioning history
Shaft Simple Ellipsoid U-shaped Spinule
Conditioned Control Conditioned Control Conditioned Control Conditioned Control Conditioned Control
Level 2
Level 3
Level 4
230 240 152 158 182 196 314 302 349 421
220 180 132 153 170 180 245 250 327 374
200 180 158 146 155 165 220 241 266 363
number of large, concave postsynaptic spine heads, localized to the region of primary synaptic activation, following conditioning. It is easily imagined that these larger synapses with the larger area of their PSDs mediate the enhanced synaptic function that was measured. In fact, the actual spine head perimeter increases are considerably larger than the 'circle size' data (Fig. 4) indicate since the factor converting 'circle size' to perimeter is much greater than :r for the concave spine head category (it is 5.1 for the concave category rather than the 3.14 which nearly approximates the factor for the simple and ellipsoid categories). This result is consistent with the Fifkova and Van Harreveld experimentsS, 27 that used a different fixation method. There, EC stimulation produced increases in the size of dentate spine head profiles in the presumed region of synaptic activation. Another significant result of the present experiments is that numerous small spine profiles are situated in regions just peripheral to the central region of synaptic activation, with its many large concave spine heads. It is possible that these small synapses are newly formed and in the process of 'growing up'. On the other hand, since: (1) synaptic
REFERENCES 1 Andersen, P., Bliss, T. V. P. and Skrede, K. K., Lamellar organization of hippocampal excitatory pathways, Exp. Brain Res., 13 (1971) 222-238. 2 Anderson, J. A., Parallel computation with simple neural networks, Cognition and Brain Theory, 3 (1979) 45-53.
inactivity in the presence of nearby potentiation (ref. 15; Levy and Steward, unpublished observations) produces synaptic depression; and (2) since the conditioning electrode tmdoubtedley does not activate all the synapses within the activated zone, it is also possible that these smaller synapses were larger synapses before conditioning. High-frequency activation of sufficient nearby synapses induced synaptic depression at these synapses, reflected in and mediated by a decreased synaptic area. ACKNOWLEDGEMENTS This work was supported in part by N I H Grant R01-NS15488 to W . B . L . N . L . D . was supported in part by N I H Predoctoral Fellowship MH05677 and by the Department of Neurological Surgery. We thank Dr. O. Steward for his collaboration on the initial observation of concave synapses and for his advice and encouragement throughout the study. We also acknowledge the expert technical assistance of S. Vinsant and T. Gill, the secretarial aid of M. P. Janssen, and the support and encouragement of Dr. John A. Jane.
3 Baudry, M., 0liver, M., Creager, R., Wieraszko, A. and Lynch, G., Increase in glutamate receptors following repetitive electrical stimulation in hippocampal slices, Life Sci., 27 (1980) 325-330. 4 Bliss, T. V. P. and Lomo, T., Long-lasting potentiation of synaptic transmission in the dentate area of the anaesthetized rabbit following stimulation of the perforant path, J. Physiol. (Lond.), 232 (1973) 331-356.
30 5 Bohmann, S. O. and Maunsbach, A. B., Effects on tissue fine structure of glutaraldehyde fixatives, J. ultrastruct. Res., 30 (1970) 195-208. 6 Cooper, L. N., Lieberman, F. and Oja, E., A theory for the acquisition and loss of neuron specificity in visual cortex, BioL Cybernet., 33 (1979) 9-28. 7 Desmond, N. L. and Levy, W. B., Ultrastructural and numerical changes in dendritic spines as a consequence of long-term potentiation, Anat. Rec., 199 (1981) 68A69A. 8 Fifkova, E. and Van Harreveld, A., Long-lasting morphological changes in dendritic spines of dentate granular cells following stimulation of the entorhinal area, J. NeurocytoL, 6 (1977) 211-230. 9 Globus, A., Brain morphology as a function of presynaptic morphology and activity. In A. H. Riesen (Ed.), The Developmental Neuropsychology of Sensory Deprivation, Academic Press, New York, 1975, pp. 9-91. 10 Hebb, D. O., The Organization of Behavior, Wiley, New York, 1949, 335 pp. 11 Kohonen, T., Associative Memory. A System-Theoretical Approach, Springer, New York, 1977, 176 pp. 12 Laatsch, R. H. and Cowan, W. M., Electron microscopic studies of the dentate gyrus of the rat. I. Normal structure with special reference to synaptic organization, J. comp. Neurol., 128 (1966) 359-396. 13 Lee, K. S., Schottler, F., Oliver, M. and Lynch, G., Brief bursts of high-frequency stimulation produce two types of structural change in rat hippocampus, J. NeurophysioL, 44 (1980) 247-258. 14 Levy, W. B., Associative changes at the synapse: LTP in the hippocampus. In W. B. Levy, J. Anderson and S. Lehmkuhle (Eds.), Synaptic Modification, Neuron Selectivity and Nervous System Organization, Lawrence Erlbaum, Hillsdale, N J, in press. 15 Levy, W. B. and Steward, O., Synapses as associative memory elements in the hippocampal formation, Brain Research, 175 (1979) 233-245. 16 McNaughton, B. L. and Barnes, C. A., Physiological identification and analysis of dentate granule cell responses to stimulation of the medial and lateral perforant pathways in the rat, J. comp. NeuroL, 175 (1977) 439-454. 17 McNaughton, B. L., Douglas, R. M. and Goddard, G. V., Synaptic enhancement in fascia dentata: cooperative among coactive afferents, Brain Research, 157 (1978) 277-293. 18 Marr, D., A theory of cerebellar cortex, J. PhysioL
(Lond.), 202 (1969) 437--470. 19 Nie, N. H., Hull, C. H., Jenkins, J. G., Steinbrenner, K. and Bent, D. H., SPSS: Statistical Package for the Social Sciences, 2nd Edn., McGraw-Hill, New York, 1975, 675 pp. 20 Rutledge, L. T., Synaptogenesis" effects of synaptic use. In M. R. Rosenzweig and E. L. Bennett (Eds.), Neural Mechanisms of Learning and Memory, MIT Press, Cambridge, MA, 1976, pp. 329-339. 21 Rutledge, L. T., The effects of denervation and stimulation upon synaptic ultrastructure, J. comp. NeuroL, 178 (1978) 117-128. 22 Steward, O. and Scoville, S. A., Cells of origin of entorhinal cortical afferents to the hippocampus and fascia dentata of the rat, J. comp. NeuroL, 169 (1976) 347-370. 23 Struble, R. G., Desmond, N. L. and Levy, W. B., Anatomical evidence for interlamellar inhibition in the fascia dentata, Brain Research, 152 (1978) 580-585. 24 Tanzi, E., I Fatti i le induzioni nelrodierna istologia del sistema nervoso, Riv. sperim, freniatria medic, leg., 19 (1893) 419-472. 25 Tarrant, S. B. and Routtenberg, A., The synaptic spinule in the dendritic spine: electron microscopic study of the hippocampal dentate gyrus, Tiss. Cell, 9 (1977) 461-473. 26 Tarrant, S. B. and Routtenberg, A., Postsynaptic membrane and spine apparatus: proximity in dendritic spines, Neurosci. Lett., 11 (1979)289-294. 27 Van Harreveld, A. and Fifkova, E., Swelling of dendritic spines in the fascia dentata after stimulation of the perforant fibers as a mechanism of post-tetanic potentiation, Exp. Neurol., 49 (1975) 736-739. 28 Wilson, R. C., Functional Consequences of the Morphological and Physiological Modification of Synapses and Circuits in the Hippocampal Formation of the Rat, Unpublished doctoral dissertation, University of Virginia, 1981. 29 Wilson, R.~C., Levy, W. B. and Steward, O., Functional effects of lesion-induced plasticity: long-term potentiation in normal and lesion-induced temporodentate connections, Brain Research, 176 (1979) 65-78. 30 Wilson, R. C., Levy, W. B. and Steward, O., Changes in the translation of synaptic excitation to dentate granule cell discharge accompanying long term potentiation. II. An evaluation of mechanisms utilizing the dentate gyrus dually innervated by surviving ipsilateral and sprouted crossed temporodentate inputs, J. NeurophysioL, 46 (1981) 339-355.