- Email: [email protected]
Transient spine density increases in the mid-molecular layer of hippocampal dentate gyrus accompany consolidation of a spatial learning task in the rodent
Transient spine density change with learning
Pergamon PII: S0306-4522(00)00182-2
Neuroscience Vol. 99, No. 2, pp. 229–232, 2000 229 䉷 2000 IBRO. Published by Elsevier Science Ltd Printed in Great Britain. All rights reserved 0306-4522/00 $20.00+0.00
www.elsevier.com/locate/neuroscience
TRANSIENT SPINE DENSITY INCREASES IN THE MID-MOLECULAR LAYER OF HIPPOCAMPAL DENTATE GYRUS ACCOMPANY CONSOLIDATION OF A SPATIAL LEARNING TASK IN THE RODENT A. O’MALLEY, C. O’CONNELL, K. J. MURPHY and C. M. REGAN* Department of Pharmacology, The Conway Institute, University College Dublin, Belfield, Dublin 4, Ireland
Abstract—In previous studies, we observed a transient increase in dendritic spine frequency in the molecular layer of the dentate gyrus at 6 h following passive avoidance training [O’Malley A., O’Connell C. and Regan C. M. (1998) Neuroscience 87, 607–613]. To determine if a similar change is associated with spatial forms of learning, we have estimated time-dependent modulations of spine number in the dentate gyrus of the adult rat following water maze training. All animals exhibited significant reductions in the latency to locate the platform over the five training sessions of the single trial (median and interquartile ranges of 60, 8 versus 8, 3 s for trials 1 and 5, respectively) and this improved performance was retained just prior to killing at the 6 h post-training time. The unbiased dissector stereological procedure was used to estimate spine number in serial pairs of ultrathin coronal sections obtained at a point 3.3 mm caudal of Bregma. This analysis revealed a significant learning-associated increase in spine number at the 6 h posttraining time (1.32 ^ 0.18 spines/mm 3) as it was not observed in paired controls exposed to the water maze for a similar swim-time (0.66 ^ 0.11 spines/mm 3). The increase was transient as spine number returned to control levels at the 72 h post-training time. These spine frequency changes are proposed to reflect increased synapse turnover rate and concomitant change in connectivity pattern in the processing of information for long-term storage. 䉷 2000 IBRO. Published by Elsevier Science Ltd. All rights reserved. Key words: axospinous synapses, water maze, dissector, dentate gyrus.
There is evidence to suggest that the process of information storage involves a cascade of events by which enduring neural activity results in transcription factor activation, gene induction, protein synthesis and, ultimately, change in synapse structure, number and/or connectivity pattern. 2 Such structural change may form the basis of memory as the increased synapse elaboration observed following complex environment rearing has been associated with improved spatial learning in rodents. 17,29 Long-term potentiation (LTP), which serves as a paradigm of information processing, has provided evidence for the sequence of ultrastructural events that is believed to accompany memory formation. 28 The transient increase in the frequency of perforated synapses that follows the induction of LTP is succeeded by the elaboration of activated dendritic spines onto axon terminals in the CA1 region of the hippocampal formation. These changes are remarkably large as a two- to threefold increase is observed in the elements quantified. Also, repetitive tetanizations of the perforant path over a four-day period result in an increased ratio of perforated to non-perforated synapses in the mid-molecular layer of the dentate gyrus when determined at 1 h following the last tetanization. 9 However, these changes are smaller, exhibiting an approximate 25% increase, as compared to those observed in the CA1 region. It is probable that these stimulus-induced alterations in synapse morphology represent structural intermediates of increased synapse turnover. 4,19
Transient change in synapse turnover has also been observed following behavioural modification in the intact animal. Within 1 h following avoidance conditioning in the day-old chick the number of axospinous synapses in the intermediate hyperstriatum ventrale (IMHV) are increased by over 70%. 6 This structural modification is transient as over the following 24 h the learning-associated increase in synapse number returns to that observed in the control animals. Change in synapse frequency is also observed in the chick lobus parolfactorius (LPO). 14 In this brain region avoidance training results in a 30% increase in synapse number at 24 h following training but not at the earlier post-training times of 1, 6, and 12 h. The time-dependent sequence of these structural changes closely parallels the involvement of these brain regions in task consolidation. 10 Lesion studies have indicated the IMHV to be the initial repository of the memory trace from which it is translocated to the LPO within the first hour post-training. Surprisingly, there is little evidence for the involvement of synapse frequency change during the acquisition and consolidation of learning in the adult mammal. We have observed a significant, twofold increase in the frequency of axospinous synapses in the mid-molecular layer of the dentate gyrus in a 6 h post-training period following passive avoidance learning in the adult rat. 21 This increase was transient as spine number returned to control levels at the 72 h post-training time, the latest period examined. Although no change in spine frequency was observed in control animals allowed to passively explore the training apparatus, the possibility exists that these structural modifications may be the result of some stress response specific to the trained animal that received the foot shock as the aversive stimulus. To address this issue, dentate spine frequency change was evaluated in animals trained in a water maze spatial paradigm as compared to
*To whom all correspondence should be addressed. Tel.: ⫹ 353-17061557; fax: ⫹ 353-1-2692749. E-mail address: [email protected] (C. M. Regan). Abbreviations: IMHV, intermediate hyperstriatum ventrale; LPO, lobus parolfactorius; LTP, long-term potentiation; NCAM, neural cell adhesion molecule. 229
230
A. O’Malley et al.
controls with a similar swim-time. We now demonstrate transient spine frequency change to be a common feature in both spatial and avoidance learning. EXPERIMENTAL PROCEDURES
Training The spatial learning task employed has been described in detail previously. 18 Briefly, the water maze apparatus consisted of a large circular pool (1 m diameter, 80 cm high, temperature 26 ^ 1⬚C) with a platform (11 cm diameter) submerged 1.5 cm below the water surface. Both the pool and the platform were constructed of black polyvinyl plastic and offered no intra-maze cues to guide escape behaviour. The experimental room contained several extra-maze visual cues. During training the platform was hidden in the same quadrant at a point which was 30 cm from the side wall. Each trial started with the rat facing the wall of the maze at one of three defined locations. The time taken by the rat to find the hidden platform within a 60-s period was recorded. On the first trial, rats failing to find the platform within the 60-s period were placed on it for 10 s. Times to reach the platform were measured over five trials with inter-trial intervals of 300 s in the single training session. Animals were tested for recall of platform location at the 6 h post-training time. Paired controls were allowed to swim in the water maze in the absence of the platform for the same period of time as their trained counterparts. A separate control group was comprised of naı¨ve animals that had not been exposed to the training environment. All experimental procedures were approved by the Animal Ethics Committee of the Biomedical Facility at University College Dublin and were carried out by individuals who held the appropriate national license. Microscopy All animals were killed by being anaesthetized with sodium pentobarbitone (60 mg/kg, i.p.) and, following transcardial perfusion with a 4% (w/v) paraformaldehyde/2% (w/v) glutaraldehyde solution at pH 7.4, their brains were removed and kept overnight in the same fixative. A pre-formed mould (Zivac Millar, U.S.A.) was used to make three coronal cuts rostral to the cerebellum and caudal to the prefrontal cortex at a point approximately ⫺3.3 mm with respect to Bregma. The two central slices obtained by this process were fixed separately to the stage of a vibroslicer (Campden Instruments, U.K.), ⫺3.3 mm face uppermost, and sections of approximately 100 mm, both rostral and caudal to the centre cut, were collected into phosphate buffer (0.1 M) in preparation for processing into epoxy resin. Slices were postfixed for 30 min in 0.1% (w/v) osmium tetroxide (Sigma Chemical Co. Ltd, U.K.) in phosphate buffer (0.1 M), dehydrated and flat embedded with epoxy resin (Agar 100; Agar Scientific, U.K.) by routine methods. Following polymerization of the resin, the hippocampus was excised from brain slices and re-embedded in resin filled capsules. Using an ultramicrotome the re-embedded hippocampus was identified, the dentate gyrus isolated, and semi-thin sections taken. In such sections, collected serially and stained with 1% Toluidine Blue (Sigma Chemical Co. Ltd, U.K.), the changing form of the granule cell layer could be discerned. This distinctive change allowed a match in cell layer-form to be sought between that in any given series of sections and that recorded in a pre-defined atlas of the dentate gyrus. This atlas was generated using serial semi-thin sections to a distance of 140 mm posterior to an arbitrary point which was approximately ⫺3.3 mm with respect to Bregma. 20 Once a “frame match” was achieved, 10 serial ultrathin sections of gold interference colour were cut from the block face and collected in pairs on coated slot grids (2 × 1 mm; Agar Scientific, U.K.). Further semithin sections were then taken at a distance great enough to preclude double counting yet sufficiently small to ensure future ultrathin sectioning remained within the desired area. On average two series of ultrathin sections were taken per re-embedded dentate slice. Ultrathin sections were stained using uranyl acetate (5% w/v distilled water) and lead citrate (0.3% w/v in 0.1 M sodium hydroxide). Sections were examined in a JEOL 1200EX electron microscope at an accelerating voltage of 80 kV. Stereology The frequency of dendritic spines in the dentate middle molecular layer region was estimated using the double dissector, an unbiased
Table 1. Learning performance of rats in successive trials of the water maze Trial no.
Time to platform Median (interquartile range)
1 2 3 4 5 Recall
60 (8) 34 (33) 8 (8) 10 (7) 8 (3) 9.5 (28)
The data are the median and interquartile range (n 6) for each trial and the transfer recall test. All values are significantly reduced from the first trial (P ⬍ 0.05; Mann–Whitney U-test).
stereological method. 26 In this method two serial sections are photographed, the first of which is designated the reference section and the second the lookup section. Such photomicrographs were printed to a final of 23,000-fold magnification. Within an unbiased counting frame of area equivalent to 45 mm 2, the number (Q) of all spines present in the reference section but absent in the lookup section was counted. 12 Spines were characterized by a spine neck which was thinner than the synaptic head and in which a cisternal structure was commonly encountered. 22 Due to the thinness of section the necks appeared predominantly to be isolated from the corresponding synapse, the intervening portion of the neck or the synapse often being absent from the plane of section. However, such partial spine emanations from the dendrite were considered predictive of spine heads. The number of spine counts per unit volume was calculated by: Nv Q/ha(fra), where Nv is the numerical density (mm 3) of spines; h the dissector height (the distance between the two dissector planes) which is equal to the thickness of the reference section; and a(fra) the area of the counting frame. 30 Section thickness (h) was determined using the minimum fold technique. 24 Minimum folds were identified and photographed in ultrathin sections (at 7000-fold magnification), and measured on photographic prints of final 17,000-fold magnification. These latter criteria were also used to examine a calibrating grating replica of 1200 lines per mm (Agar Scientific, U.K.), all measurements being adjusted to this standard. Mean section thickness was 60.41 ^ 1.37 nm (mean ^ S.E.M.). For each animal, dissector pairs were chosen until a progressive mean test of Q values consistently showed the standard error of the mean to be less than 10%. 31 This resulted in 367 ^ 21 mm 2 (mean ^ S.E.M.) of dentate middle molecular layer being quantified in each animal. The mean spine count of each animal was then used to establish the final mean ^ S.E.M. for each group. Statistical analysis employed the Student’s t-test, and P ⬍ 0.05 was taken to indicate significant difference. RESULTS
The water maze task was acquired by all animals as the swim latency times became significantly reduced between the first and the fifth trial (median and interquartile ranges of 60, 8 versus 8, 3 s, respectively; P ⬍ 0.05, Mann–Whitney U-test; Table 1). The passive control animals explored the water maze for a time matched to their trained counterparts but in the absence of the platform. The improved performance of the task in the trained animals persisted in the recall trial at 6 h post-training (median escape latency 9.5 s, interquartile range 28 s), indicating that the process of memory consolidation had been initiated. In this study, we employed spine counting in preference to direct quantification of synapses, for example by vesicle content and/or phosphotunstic acid-labelled postsynaptic densities, which presumes the morphology of novel synapses to be invariant in any given post-training period. A spine neck that was thinner than the synaptic head characterized the dendritic spines quantified. In addition, many spines exhibited a cisternal structure in the neck region (Fig. 1). However, in most cases, due to thinness of section, the spine necks were
Transient spine density change with learning
231
Table 2. Quantitation of dentate spine frequency following water maze training Time post-training
– 1h 6h 72 h 6h
Treatment
Spine number/mm 3 Mean ^ S.E.M.
Naı¨ve Trained Trained Trained Passive
0.773 ^ 0.092 0.601 ^ 0.092 1.320 ^ 0.178* 0.657 ^ 0.056 0.655 ^ 0.112
*All values are the mean ^ S.E.M. (3 ⱕ n ⱕ 4) and those significantly different (P ⬍ 0.05, Student’s t-test) from naı¨ve are indicated by an asterisk.
Fig. 1. A typical axospinous synapse (asterisk) in the mid-molecular layer of the dentate gyrus. The synapse incorporates a postsynaptic thickening and vesicle-containing presynaptic element. Scale bar 200 nm.
isolated from the corresponding spine head as the intervening portion of the neck or the spine head was absent from the plane of view. Nonetheless, such partial spine emanations from the dendrite were considered predictive of spine heads for the purpose of these studies. Moreover, in all cases, an individual who was unaware of the source of section counted these spine emanations. Axodendritic synapses were identified by the presence of presynaptic vesicles and a postsynaptic thickening, those lacking a spine neck being infrequent in comparison to axospinous synapses that were most numerous. Trained rats exhibited a significantly increased (P ⬍ 0.05, Student’s t-test) spine frequency 6 h after training when compared to naı¨ve animals or values 1 h post-training (Table 2). The naı¨ve level of spines at this point in the mid-molecular layer was 0.77 ^ 0.09 spines/mm 3 while that at 6 h post water maze training proved to be significantly greater at 1.32 ^ 0.18 spines/mm 3. This increase with time after learning did not persist, as is shown by the return to basal levels at 72 h post-training when a spine count of 0.66 ^ 0.06 spines/mm 3 was measured. Passive controls possessed significantly fewer spines (0.66 ^ 0.11 spines/ mm 3) compared to trained animals at the 6 h time point (P ⬍ 0.05, Student’s t-test). Further, this spine density did not differ from naı¨ve levels, indicating that the large increase in spine density at this time is most probably the result of spatial learning. Thus, a post-training, time-dependent increase in spine frequency occurs following both avoidance and spatial learning. DISCUSSION
The experiments reported here imply that among the
sequence of physiological processes initiated following acquisition of a spatial task is modulation of spine number in the mid-molecular layer of the dentate gyrus. Sensory information arrives at the hippocampus via the perforant path entorhinal cortical efferents, 80% of which terminate in the dentate molecular layer as axospinous synapses. 5,13,27 Thus, the observed change in spine number most probably relates to the elaboration of granule cell dendritic connections with these efferent axons. The magnitude of the spine frequency change is not unlike that reported following the induction of LTP or in the chick and rat following avoidance training. 6,21,28 However, the observed transience of the spine frequency change following spatial learning is unique. Previously, quantitative studies on synapse elaboration following spatial learning have failed to find change in the number of axospinous or axodendritic synapses in either the dentate gyrus or CA1 regions of the hippocampal formation in rats killed at six days following the last training session. 23 These observations are not at variance with those reported here as axospinous synapse number in the dentate gyrus returns to baseline level at three days following the last training session. Moreover, this learning-associated transience in synapse modification is also observed following passive avoidance training and occurs within the four- to five-day period thought to be required by the hippocampus for consolidation of tasks such as context fear conditioning. 15,21 In contrast, the enduring change in cortical synapse number that accompanies complex environment rearing or motor learning 3,11,16 reflects the persisting role of this brain area in information storage. The transient spine frequency change that occurs in the 6-h post-training period following either passive avoidance or water maze learning is commensurate with previously described molecular and morphological events. The amnesic action of protein synthesis inhibitors is only effective in the 3–6 h period that follows an avoidance learning task and immediately precedes axospinous growth in the dentate. 25 Moreover, increased expression of ribosomes and microtubules in granule cells and their dendrites is observed in the period coincident with spine growth. 20 There is also evidence for the involvement of the neural cell adhesion molecule (NCAM) in the synaptic rearrangement that accompanies this period of memory consolidation. Intraventricular administration of anti-NCAM at the 6–8 h post-training time produces amnesia in a manner that is consistent with transient synapse formation. 1,7 In this situation memory is retained at the 24 h recall time but is lost when animals are re-tested at 48 h. The most parsimonious explanation given for this effect is that the synapses elaborated are defective in
232
A. O’Malley et al.
NCAM function and fail to compete effectively in the subsequent period of synapse elimination. 8 In summary, the spine growth that accompanies memory formation exemplifies a consolidation process whereby synaptic alteration is dependent on protein synthesis, relies
on intracellular transport mechanisms, and occurs predominantly on dendritic spines.
Acknowledgement—This work was supported by Enterprise Ireland.
REFERENCES
1. Alexinsky T., Przbyslawski J., Mileusnic R., Rose S. P. R. and Sara S. J. (1997) Antibody to day-old chick brain glycoprotein produces amnesia in adult rats. Neurobiol. Learn. Mem. 67, 14–20. 2. Bailey C. H., Bartsch D. and Kandel E. R. (1996) Toward a molecular definition of long-term memory storage. Proc. natn. Acad. Sci. U.S.A. 93, 13,445–13,452. 3. Black J. E., Isaacs K. R., Anderson B. J., Alcantara A. A. and Greenough W. T. (1990) Learning causes synaptogenesis, whereas motor activity causes angiogenesis, in cerebellar cortex of adult rats. Proc. natn. Acad. Sci. U.S.A. 87, 5568–5572. 4. Carlin R. K. and Siekevitz P. (1983) Plasticity in the central nervous system: do synapses divide? Proc. natn. Acad. Sci. U.S.A. 80, 3517–3521. 5. Cotman C. W. and Nadler J. V. (1978) Reactive synaptogenesis in the hippocampus. In Neural Plasticity (ed. Cotman C. W.). Raven Press, New York. 6. Doubell T. P. and Stewart M. G. (1993) Short-term changes in the numerical density of synapses in the intermediate and medial hyperstriatum ventrale following one-trial passive avoidance training in the chick. J. Neurosci. 13, 2230–2235. 7. Doyle E., Nolan P. M., Bell R. and Regan C. M. (1992) Intraventricular infusions of anti-neural cell adhesion molecules in a discrete posttraining period impair consolidation of a passive avoidance response in the rat. J. Neurochem. 59, 1570–1573. 8. Doyle E., Nolan P., Bell R. and Regan C. M. (1992) Neurodevelopmental events underlying information acquisition and storage. Network 3, 89–94. 9. Geinisman Y., deToledo-Morrell L. and Morrell F. (1991) Induction of long-term potentiation is associated with an increase in the number of axospinous synapses with segmented postsynaptic densities. Brain Res. 566, 77–88. 10. Gilbert D. B., Patterson T. A. and Rose S. P. (1991) Dissociation of brain sites necessary for registration and storage of memory for a one-trial passive avoidance task in the chick. Behav. Neurosci. 105, 553–561. 11. Greenough W. T., Hwang H.-M. F. and Gorman C. (1985) Evidence for active synapse formation or altered postsynaptic metabolism in visual cortex of rats reared in complex environments. Proc. natn. Acad. Sci. U.S.A. 82, 4549–4552. 12. Gundersen H. J. G. (1977) Notes on estimation of the numerical density of arbitrary profiles: the edge effect. J. Microsc. 111, 219–223. 13. Hjort-Simonsen A. and Jeune B. (1972) Origin and termination of the hippocampal perforant path in the rat studied by silver impregnation. J. comp. Neurol. 144, 215–232. 14. Hunter A. and Stewart M. G. (1993) Long-term increases in the numerical density of synapses in the chick lobus parolfactorius after passive avoidance training. Brain Res. 605, 251–255. 15. Kim J. J. and Fanselow M. S. (1992) Modality-specific retrograde amnesia of fear. Science 256, 675–677. 16. Kleim J. A., Kapil V., Ballard D. H. and Greenough W. T. (1997) Learning-dependent synaptic modifications in the cerebellar cortex of the adult rat persist for at least four weeks. J. Neurosci. 17, 717–721. 17. Moser M., Trommold M. and Anderson P. (1994) An increase in dendritic spine density on hippocampal CA1 pyramidal cells following spatial learning in adult rats suggests the formation of new synapses. Proc. natn. Acad. Sci. U.S.A. 91, 12673–12675. 18. Murphy K. J., O’ Connell A. W. and Regan C. M. (1996) Repetitive and transient increases in hippocampal neural cell adhesion molecule polysialylation state following multi-trial spatial training. J. Neurochem. 67, 1268–1274. 19. Nieto-Sampedro M., Hoff S. F. and Cotman C. W. (1982) Perforated postsynaptic densities: probable intermediates in synapse turnover. Proc. natn. Acad. Sci. U.S.A. 79, 5718–5722. 20. O’Connell C., O’Malley A. and Regan C. M. (1997) Transient, learning-induced ultrastructural change in spatially-clustered dentate granule cells of the adult rat hippocampus. Neuroscience 76, 55–62. 21. O’Malley A., O’Connell C. and Regan C. M. (1998) Ultrastructural analysis reveals avoidance conditioning to induce a transient increase in hippocampal dentate spine density in the 6 h post-training period of consolidation. Neuroscience 87, 607–613. 22. Peters A., Palay S. L. and Webster H. D. (1991) The Fine Structure of the Nervous System. Oxford University Press, UK. 23. Rusakov D. A., Davies H. A., Harrison E., Diana G., Richter-Levin G., Bliss T. V. and Stewart M. G. (1997) Ultrastructural synaptic correlates of spatial learning in rat hippocampus. Neuroscience 80, 69–77. 24. Small J. V. (1968) Measurement of section thickness. In Proceedings of the 4th European Congress on Electron Microscopy (ed. Bocciarelli D. S.). Tipografia Poliglotta Vaticana, Roma. 25. Squire L. R. and Barondes S. H. (1972) Variable decay of memory and its recovery in cycloheximide-treated mice. Proc. natn. Acad. Sci. U.S.A. 69, 1416–1420. 26. Sterio D. C. (1984) The unbiased estimation of number and sizes of arbitrary particles using the disector. J. Microsc. 134, 127–136. 27. Steward O. (1976) Topographic organisation of the projections from the entorhinal area to the hippocampal formation of the rat. J. comp. Neurol. 167, 285–314. 28. Toni N., Buchs P.-A., Nikonenko I., Bron C. R. and Muller D. (1999) LTP promotes formation of multiple spine synapses between a single axon terminal and a dendrite. Nature 402, 421–424. 29. Turner A. M. and Greenough W. T. (1985) Differential rearing effects on rat visual cortex synapses. I. Synaptic and neuronal density and synapse per neuron. Brain Res. 329, 195–203. 30. West M. J. and Gundersen H. J. G. (1990) Unbiased stereological estimation of the number of neurons in the human hippocampus. J. comp. Neurol. 296, 1–22. 31. Williams M. A. (1977) Stereological techniques. In Quantitative Methods in Biology (ed. Glauert A. M.), Vol. 6. North Holland Publishing, The Netherlands. (Accepted 7 April 2000)
Pergamon PII: S0306-4522(00)00182-2
Neuroscience Vol. 99, No. 2, pp. 229–232, 2000 229 䉷 2000 IBRO. Published by Elsevier Science Ltd Printed in Great Britain. All rights reserved 0306-4522/00 $20.00+0.00
www.elsevier.com/locate/neuroscience
TRANSIENT SPINE DENSITY INCREASES IN THE MID-MOLECULAR LAYER OF HIPPOCAMPAL DENTATE GYRUS ACCOMPANY CONSOLIDATION OF A SPATIAL LEARNING TASK IN THE RODENT A. O’MALLEY, C. O’CONNELL, K. J. MURPHY and C. M. REGAN* Department of Pharmacology, The Conway Institute, University College Dublin, Belfield, Dublin 4, Ireland
Abstract—In previous studies, we observed a transient increase in dendritic spine frequency in the molecular layer of the dentate gyrus at 6 h following passive avoidance training [O’Malley A., O’Connell C. and Regan C. M. (1998) Neuroscience 87, 607–613]. To determine if a similar change is associated with spatial forms of learning, we have estimated time-dependent modulations of spine number in the dentate gyrus of the adult rat following water maze training. All animals exhibited significant reductions in the latency to locate the platform over the five training sessions of the single trial (median and interquartile ranges of 60, 8 versus 8, 3 s for trials 1 and 5, respectively) and this improved performance was retained just prior to killing at the 6 h post-training time. The unbiased dissector stereological procedure was used to estimate spine number in serial pairs of ultrathin coronal sections obtained at a point 3.3 mm caudal of Bregma. This analysis revealed a significant learning-associated increase in spine number at the 6 h posttraining time (1.32 ^ 0.18 spines/mm 3) as it was not observed in paired controls exposed to the water maze for a similar swim-time (0.66 ^ 0.11 spines/mm 3). The increase was transient as spine number returned to control levels at the 72 h post-training time. These spine frequency changes are proposed to reflect increased synapse turnover rate and concomitant change in connectivity pattern in the processing of information for long-term storage. 䉷 2000 IBRO. Published by Elsevier Science Ltd. All rights reserved. Key words: axospinous synapses, water maze, dissector, dentate gyrus.
There is evidence to suggest that the process of information storage involves a cascade of events by which enduring neural activity results in transcription factor activation, gene induction, protein synthesis and, ultimately, change in synapse structure, number and/or connectivity pattern. 2 Such structural change may form the basis of memory as the increased synapse elaboration observed following complex environment rearing has been associated with improved spatial learning in rodents. 17,29 Long-term potentiation (LTP), which serves as a paradigm of information processing, has provided evidence for the sequence of ultrastructural events that is believed to accompany memory formation. 28 The transient increase in the frequency of perforated synapses that follows the induction of LTP is succeeded by the elaboration of activated dendritic spines onto axon terminals in the CA1 region of the hippocampal formation. These changes are remarkably large as a two- to threefold increase is observed in the elements quantified. Also, repetitive tetanizations of the perforant path over a four-day period result in an increased ratio of perforated to non-perforated synapses in the mid-molecular layer of the dentate gyrus when determined at 1 h following the last tetanization. 9 However, these changes are smaller, exhibiting an approximate 25% increase, as compared to those observed in the CA1 region. It is probable that these stimulus-induced alterations in synapse morphology represent structural intermediates of increased synapse turnover. 4,19
Transient change in synapse turnover has also been observed following behavioural modification in the intact animal. Within 1 h following avoidance conditioning in the day-old chick the number of axospinous synapses in the intermediate hyperstriatum ventrale (IMHV) are increased by over 70%. 6 This structural modification is transient as over the following 24 h the learning-associated increase in synapse number returns to that observed in the control animals. Change in synapse frequency is also observed in the chick lobus parolfactorius (LPO). 14 In this brain region avoidance training results in a 30% increase in synapse number at 24 h following training but not at the earlier post-training times of 1, 6, and 12 h. The time-dependent sequence of these structural changes closely parallels the involvement of these brain regions in task consolidation. 10 Lesion studies have indicated the IMHV to be the initial repository of the memory trace from which it is translocated to the LPO within the first hour post-training. Surprisingly, there is little evidence for the involvement of synapse frequency change during the acquisition and consolidation of learning in the adult mammal. We have observed a significant, twofold increase in the frequency of axospinous synapses in the mid-molecular layer of the dentate gyrus in a 6 h post-training period following passive avoidance learning in the adult rat. 21 This increase was transient as spine number returned to control levels at the 72 h post-training time, the latest period examined. Although no change in spine frequency was observed in control animals allowed to passively explore the training apparatus, the possibility exists that these structural modifications may be the result of some stress response specific to the trained animal that received the foot shock as the aversive stimulus. To address this issue, dentate spine frequency change was evaluated in animals trained in a water maze spatial paradigm as compared to
*To whom all correspondence should be addressed. Tel.: ⫹ 353-17061557; fax: ⫹ 353-1-2692749. E-mail address: [email protected] (C. M. Regan). Abbreviations: IMHV, intermediate hyperstriatum ventrale; LPO, lobus parolfactorius; LTP, long-term potentiation; NCAM, neural cell adhesion molecule. 229
230
A. O’Malley et al.
controls with a similar swim-time. We now demonstrate transient spine frequency change to be a common feature in both spatial and avoidance learning. EXPERIMENTAL PROCEDURES
Training The spatial learning task employed has been described in detail previously. 18 Briefly, the water maze apparatus consisted of a large circular pool (1 m diameter, 80 cm high, temperature 26 ^ 1⬚C) with a platform (11 cm diameter) submerged 1.5 cm below the water surface. Both the pool and the platform were constructed of black polyvinyl plastic and offered no intra-maze cues to guide escape behaviour. The experimental room contained several extra-maze visual cues. During training the platform was hidden in the same quadrant at a point which was 30 cm from the side wall. Each trial started with the rat facing the wall of the maze at one of three defined locations. The time taken by the rat to find the hidden platform within a 60-s period was recorded. On the first trial, rats failing to find the platform within the 60-s period were placed on it for 10 s. Times to reach the platform were measured over five trials with inter-trial intervals of 300 s in the single training session. Animals were tested for recall of platform location at the 6 h post-training time. Paired controls were allowed to swim in the water maze in the absence of the platform for the same period of time as their trained counterparts. A separate control group was comprised of naı¨ve animals that had not been exposed to the training environment. All experimental procedures were approved by the Animal Ethics Committee of the Biomedical Facility at University College Dublin and were carried out by individuals who held the appropriate national license. Microscopy All animals were killed by being anaesthetized with sodium pentobarbitone (60 mg/kg, i.p.) and, following transcardial perfusion with a 4% (w/v) paraformaldehyde/2% (w/v) glutaraldehyde solution at pH 7.4, their brains were removed and kept overnight in the same fixative. A pre-formed mould (Zivac Millar, U.S.A.) was used to make three coronal cuts rostral to the cerebellum and caudal to the prefrontal cortex at a point approximately ⫺3.3 mm with respect to Bregma. The two central slices obtained by this process were fixed separately to the stage of a vibroslicer (Campden Instruments, U.K.), ⫺3.3 mm face uppermost, and sections of approximately 100 mm, both rostral and caudal to the centre cut, were collected into phosphate buffer (0.1 M) in preparation for processing into epoxy resin. Slices were postfixed for 30 min in 0.1% (w/v) osmium tetroxide (Sigma Chemical Co. Ltd, U.K.) in phosphate buffer (0.1 M), dehydrated and flat embedded with epoxy resin (Agar 100; Agar Scientific, U.K.) by routine methods. Following polymerization of the resin, the hippocampus was excised from brain slices and re-embedded in resin filled capsules. Using an ultramicrotome the re-embedded hippocampus was identified, the dentate gyrus isolated, and semi-thin sections taken. In such sections, collected serially and stained with 1% Toluidine Blue (Sigma Chemical Co. Ltd, U.K.), the changing form of the granule cell layer could be discerned. This distinctive change allowed a match in cell layer-form to be sought between that in any given series of sections and that recorded in a pre-defined atlas of the dentate gyrus. This atlas was generated using serial semi-thin sections to a distance of 140 mm posterior to an arbitrary point which was approximately ⫺3.3 mm with respect to Bregma. 20 Once a “frame match” was achieved, 10 serial ultrathin sections of gold interference colour were cut from the block face and collected in pairs on coated slot grids (2 × 1 mm; Agar Scientific, U.K.). Further semithin sections were then taken at a distance great enough to preclude double counting yet sufficiently small to ensure future ultrathin sectioning remained within the desired area. On average two series of ultrathin sections were taken per re-embedded dentate slice. Ultrathin sections were stained using uranyl acetate (5% w/v distilled water) and lead citrate (0.3% w/v in 0.1 M sodium hydroxide). Sections were examined in a JEOL 1200EX electron microscope at an accelerating voltage of 80 kV. Stereology The frequency of dendritic spines in the dentate middle molecular layer region was estimated using the double dissector, an unbiased
Table 1. Learning performance of rats in successive trials of the water maze Trial no.
Time to platform Median (interquartile range)
1 2 3 4 5 Recall
60 (8) 34 (33) 8 (8) 10 (7) 8 (3) 9.5 (28)
The data are the median and interquartile range (n 6) for each trial and the transfer recall test. All values are significantly reduced from the first trial (P ⬍ 0.05; Mann–Whitney U-test).
stereological method. 26 In this method two serial sections are photographed, the first of which is designated the reference section and the second the lookup section. Such photomicrographs were printed to a final of 23,000-fold magnification. Within an unbiased counting frame of area equivalent to 45 mm 2, the number (Q) of all spines present in the reference section but absent in the lookup section was counted. 12 Spines were characterized by a spine neck which was thinner than the synaptic head and in which a cisternal structure was commonly encountered. 22 Due to the thinness of section the necks appeared predominantly to be isolated from the corresponding synapse, the intervening portion of the neck or the synapse often being absent from the plane of section. However, such partial spine emanations from the dendrite were considered predictive of spine heads. The number of spine counts per unit volume was calculated by: Nv Q/ha(fra), where Nv is the numerical density (mm 3) of spines; h the dissector height (the distance between the two dissector planes) which is equal to the thickness of the reference section; and a(fra) the area of the counting frame. 30 Section thickness (h) was determined using the minimum fold technique. 24 Minimum folds were identified and photographed in ultrathin sections (at 7000-fold magnification), and measured on photographic prints of final 17,000-fold magnification. These latter criteria were also used to examine a calibrating grating replica of 1200 lines per mm (Agar Scientific, U.K.), all measurements being adjusted to this standard. Mean section thickness was 60.41 ^ 1.37 nm (mean ^ S.E.M.). For each animal, dissector pairs were chosen until a progressive mean test of Q values consistently showed the standard error of the mean to be less than 10%. 31 This resulted in 367 ^ 21 mm 2 (mean ^ S.E.M.) of dentate middle molecular layer being quantified in each animal. The mean spine count of each animal was then used to establish the final mean ^ S.E.M. for each group. Statistical analysis employed the Student’s t-test, and P ⬍ 0.05 was taken to indicate significant difference. RESULTS
The water maze task was acquired by all animals as the swim latency times became significantly reduced between the first and the fifth trial (median and interquartile ranges of 60, 8 versus 8, 3 s, respectively; P ⬍ 0.05, Mann–Whitney U-test; Table 1). The passive control animals explored the water maze for a time matched to their trained counterparts but in the absence of the platform. The improved performance of the task in the trained animals persisted in the recall trial at 6 h post-training (median escape latency 9.5 s, interquartile range 28 s), indicating that the process of memory consolidation had been initiated. In this study, we employed spine counting in preference to direct quantification of synapses, for example by vesicle content and/or phosphotunstic acid-labelled postsynaptic densities, which presumes the morphology of novel synapses to be invariant in any given post-training period. A spine neck that was thinner than the synaptic head characterized the dendritic spines quantified. In addition, many spines exhibited a cisternal structure in the neck region (Fig. 1). However, in most cases, due to thinness of section, the spine necks were
Transient spine density change with learning
231
Table 2. Quantitation of dentate spine frequency following water maze training Time post-training
– 1h 6h 72 h 6h
Treatment
Spine number/mm 3 Mean ^ S.E.M.
Naı¨ve Trained Trained Trained Passive
0.773 ^ 0.092 0.601 ^ 0.092 1.320 ^ 0.178* 0.657 ^ 0.056 0.655 ^ 0.112
*All values are the mean ^ S.E.M. (3 ⱕ n ⱕ 4) and those significantly different (P ⬍ 0.05, Student’s t-test) from naı¨ve are indicated by an asterisk.
Fig. 1. A typical axospinous synapse (asterisk) in the mid-molecular layer of the dentate gyrus. The synapse incorporates a postsynaptic thickening and vesicle-containing presynaptic element. Scale bar 200 nm.
isolated from the corresponding spine head as the intervening portion of the neck or the spine head was absent from the plane of view. Nonetheless, such partial spine emanations from the dendrite were considered predictive of spine heads for the purpose of these studies. Moreover, in all cases, an individual who was unaware of the source of section counted these spine emanations. Axodendritic synapses were identified by the presence of presynaptic vesicles and a postsynaptic thickening, those lacking a spine neck being infrequent in comparison to axospinous synapses that were most numerous. Trained rats exhibited a significantly increased (P ⬍ 0.05, Student’s t-test) spine frequency 6 h after training when compared to naı¨ve animals or values 1 h post-training (Table 2). The naı¨ve level of spines at this point in the mid-molecular layer was 0.77 ^ 0.09 spines/mm 3 while that at 6 h post water maze training proved to be significantly greater at 1.32 ^ 0.18 spines/mm 3. This increase with time after learning did not persist, as is shown by the return to basal levels at 72 h post-training when a spine count of 0.66 ^ 0.06 spines/mm 3 was measured. Passive controls possessed significantly fewer spines (0.66 ^ 0.11 spines/ mm 3) compared to trained animals at the 6 h time point (P ⬍ 0.05, Student’s t-test). Further, this spine density did not differ from naı¨ve levels, indicating that the large increase in spine density at this time is most probably the result of spatial learning. Thus, a post-training, time-dependent increase in spine frequency occurs following both avoidance and spatial learning. DISCUSSION
The experiments reported here imply that among the
sequence of physiological processes initiated following acquisition of a spatial task is modulation of spine number in the mid-molecular layer of the dentate gyrus. Sensory information arrives at the hippocampus via the perforant path entorhinal cortical efferents, 80% of which terminate in the dentate molecular layer as axospinous synapses. 5,13,27 Thus, the observed change in spine number most probably relates to the elaboration of granule cell dendritic connections with these efferent axons. The magnitude of the spine frequency change is not unlike that reported following the induction of LTP or in the chick and rat following avoidance training. 6,21,28 However, the observed transience of the spine frequency change following spatial learning is unique. Previously, quantitative studies on synapse elaboration following spatial learning have failed to find change in the number of axospinous or axodendritic synapses in either the dentate gyrus or CA1 regions of the hippocampal formation in rats killed at six days following the last training session. 23 These observations are not at variance with those reported here as axospinous synapse number in the dentate gyrus returns to baseline level at three days following the last training session. Moreover, this learning-associated transience in synapse modification is also observed following passive avoidance training and occurs within the four- to five-day period thought to be required by the hippocampus for consolidation of tasks such as context fear conditioning. 15,21 In contrast, the enduring change in cortical synapse number that accompanies complex environment rearing or motor learning 3,11,16 reflects the persisting role of this brain area in information storage. The transient spine frequency change that occurs in the 6-h post-training period following either passive avoidance or water maze learning is commensurate with previously described molecular and morphological events. The amnesic action of protein synthesis inhibitors is only effective in the 3–6 h period that follows an avoidance learning task and immediately precedes axospinous growth in the dentate. 25 Moreover, increased expression of ribosomes and microtubules in granule cells and their dendrites is observed in the period coincident with spine growth. 20 There is also evidence for the involvement of the neural cell adhesion molecule (NCAM) in the synaptic rearrangement that accompanies this period of memory consolidation. Intraventricular administration of anti-NCAM at the 6–8 h post-training time produces amnesia in a manner that is consistent with transient synapse formation. 1,7 In this situation memory is retained at the 24 h recall time but is lost when animals are re-tested at 48 h. The most parsimonious explanation given for this effect is that the synapses elaborated are defective in
232
A. O’Malley et al.
NCAM function and fail to compete effectively in the subsequent period of synapse elimination. 8 In summary, the spine growth that accompanies memory formation exemplifies a consolidation process whereby synaptic alteration is dependent on protein synthesis, relies
on intracellular transport mechanisms, and occurs predominantly on dendritic spines.
Acknowledgement—This work was supported by Enterprise Ireland.
REFERENCES
1. Alexinsky T., Przbyslawski J., Mileusnic R., Rose S. P. R. and Sara S. J. (1997) Antibody to day-old chick brain glycoprotein produces amnesia in adult rats. Neurobiol. Learn. Mem. 67, 14–20. 2. Bailey C. H., Bartsch D. and Kandel E. R. (1996) Toward a molecular definition of long-term memory storage. Proc. natn. Acad. Sci. U.S.A. 93, 13,445–13,452. 3. Black J. E., Isaacs K. R., Anderson B. J., Alcantara A. A. and Greenough W. T. (1990) Learning causes synaptogenesis, whereas motor activity causes angiogenesis, in cerebellar cortex of adult rats. Proc. natn. Acad. Sci. U.S.A. 87, 5568–5572. 4. Carlin R. K. and Siekevitz P. (1983) Plasticity in the central nervous system: do synapses divide? Proc. natn. Acad. Sci. U.S.A. 80, 3517–3521. 5. Cotman C. W. and Nadler J. V. (1978) Reactive synaptogenesis in the hippocampus. In Neural Plasticity (ed. Cotman C. W.). Raven Press, New York. 6. Doubell T. P. and Stewart M. G. (1993) Short-term changes in the numerical density of synapses in the intermediate and medial hyperstriatum ventrale following one-trial passive avoidance training in the chick. J. Neurosci. 13, 2230–2235. 7. Doyle E., Nolan P. M., Bell R. and Regan C. M. (1992) Intraventricular infusions of anti-neural cell adhesion molecules in a discrete posttraining period impair consolidation of a passive avoidance response in the rat. J. Neurochem. 59, 1570–1573. 8. Doyle E., Nolan P., Bell R. and Regan C. M. (1992) Neurodevelopmental events underlying information acquisition and storage. Network 3, 89–94. 9. Geinisman Y., deToledo-Morrell L. and Morrell F. (1991) Induction of long-term potentiation is associated with an increase in the number of axospinous synapses with segmented postsynaptic densities. Brain Res. 566, 77–88. 10. Gilbert D. B., Patterson T. A. and Rose S. P. (1991) Dissociation of brain sites necessary for registration and storage of memory for a one-trial passive avoidance task in the chick. Behav. Neurosci. 105, 553–561. 11. Greenough W. T., Hwang H.-M. F. and Gorman C. (1985) Evidence for active synapse formation or altered postsynaptic metabolism in visual cortex of rats reared in complex environments. Proc. natn. Acad. Sci. U.S.A. 82, 4549–4552. 12. Gundersen H. J. G. (1977) Notes on estimation of the numerical density of arbitrary profiles: the edge effect. J. Microsc. 111, 219–223. 13. Hjort-Simonsen A. and Jeune B. (1972) Origin and termination of the hippocampal perforant path in the rat studied by silver impregnation. J. comp. Neurol. 144, 215–232. 14. Hunter A. and Stewart M. G. (1993) Long-term increases in the numerical density of synapses in the chick lobus parolfactorius after passive avoidance training. Brain Res. 605, 251–255. 15. Kim J. J. and Fanselow M. S. (1992) Modality-specific retrograde amnesia of fear. Science 256, 675–677. 16. Kleim J. A., Kapil V., Ballard D. H. and Greenough W. T. (1997) Learning-dependent synaptic modifications in the cerebellar cortex of the adult rat persist for at least four weeks. J. Neurosci. 17, 717–721. 17. Moser M., Trommold M. and Anderson P. (1994) An increase in dendritic spine density on hippocampal CA1 pyramidal cells following spatial learning in adult rats suggests the formation of new synapses. Proc. natn. Acad. Sci. U.S.A. 91, 12673–12675. 18. Murphy K. J., O’ Connell A. W. and Regan C. M. (1996) Repetitive and transient increases in hippocampal neural cell adhesion molecule polysialylation state following multi-trial spatial training. J. Neurochem. 67, 1268–1274. 19. Nieto-Sampedro M., Hoff S. F. and Cotman C. W. (1982) Perforated postsynaptic densities: probable intermediates in synapse turnover. Proc. natn. Acad. Sci. U.S.A. 79, 5718–5722. 20. O’Connell C., O’Malley A. and Regan C. M. (1997) Transient, learning-induced ultrastructural change in spatially-clustered dentate granule cells of the adult rat hippocampus. Neuroscience 76, 55–62. 21. O’Malley A., O’Connell C. and Regan C. M. (1998) Ultrastructural analysis reveals avoidance conditioning to induce a transient increase in hippocampal dentate spine density in the 6 h post-training period of consolidation. Neuroscience 87, 607–613. 22. Peters A., Palay S. L. and Webster H. D. (1991) The Fine Structure of the Nervous System. Oxford University Press, UK. 23. Rusakov D. A., Davies H. A., Harrison E., Diana G., Richter-Levin G., Bliss T. V. and Stewart M. G. (1997) Ultrastructural synaptic correlates of spatial learning in rat hippocampus. Neuroscience 80, 69–77. 24. Small J. V. (1968) Measurement of section thickness. In Proceedings of the 4th European Congress on Electron Microscopy (ed. Bocciarelli D. S.). Tipografia Poliglotta Vaticana, Roma. 25. Squire L. R. and Barondes S. H. (1972) Variable decay of memory and its recovery in cycloheximide-treated mice. Proc. natn. Acad. Sci. U.S.A. 69, 1416–1420. 26. Sterio D. C. (1984) The unbiased estimation of number and sizes of arbitrary particles using the disector. J. Microsc. 134, 127–136. 27. Steward O. (1976) Topographic organisation of the projections from the entorhinal area to the hippocampal formation of the rat. J. comp. Neurol. 167, 285–314. 28. Toni N., Buchs P.-A., Nikonenko I., Bron C. R. and Muller D. (1999) LTP promotes formation of multiple spine synapses between a single axon terminal and a dendrite. Nature 402, 421–424. 29. Turner A. M. and Greenough W. T. (1985) Differential rearing effects on rat visual cortex synapses. I. Synaptic and neuronal density and synapse per neuron. Brain Res. 329, 195–203. 30. West M. J. and Gundersen H. J. G. (1990) Unbiased stereological estimation of the number of neurons in the human hippocampus. J. comp. Neurol. 296, 1–22. 31. Williams M. A. (1977) Stereological techniques. In Quantitative Methods in Biology (ed. Glauert A. M.), Vol. 6. North Holland Publishing, The Netherlands. (Accepted 7 April 2000)