Synaptic atrophy in the senescent hippocampus

Mechanisms of Ageing and Development, 9 (1979) 163-171 © Elsevier Sequoia S.A., Lausanne - Printed in the Netherlands

163

SYNAPTIC ATROPHY IN THE SENESCENT HIPPOCAMPUS**

W. BONDAREFF Department of Anatomy, Northwestern University Medical School, Chicago, Illinois 60611 (U.S.A.J (Received November 23, 1977)

SUMMARY Quantitative analyses of electron micrographs have shown a decrease in the number of synapses in the dentate gyrus of the senescent Fischer-344 rat. The loss of synapses, involving both dendritic spines and shafts and axon terminals of more than one population of presynaptic neurons, did not depend upon the antecedent loss of postsynaptic neurons or their dendrites. These findings suggest that the age-related loss of synapses in the dentate gyrus may depend upon an inability of presynaptic elements to maintain the structural integrity of synapses in senescence. It is proposed that a change in the glycoprotein component of presynaptic plasma membranes resulting from a deficiency in axonal transport mechanisms in the septo-hippocampal pathway may underly this presynaptic malfunction. The resulting pe.rtial deafferentation of neurons in the dentate gyrus in senescence appears to be associated with a secondary atrophy of dendrites, which results in a loss of postsynaptic membranes before a loss of postsynaptic neurons can be documented.

INTRODUCTION Senescence is accompanied by changes in the neuronal microenvironment [1]. What causes these changes is not known; whether they depend upon changes in regional blood flow or in the genome of selected populations of neurons or neuroglia is uncertain. They can be defined only in a limited way but in the hippocampus of the senescent rat the neuronal microenvironment is altered by a change in charge density of intercellular channels [2]. These extracellular spaces, too small and too comingled with the finest processes of neuroglia and neurons to be isolated, can be selectively filled with charged metallic compounds such as ruthenium red and visualized electron microscopically [3]. Ruthenium red reacts with anionic macromolecules, perhaps glycoproteins or glycosaminoglycans, which occupy the extracellular channels of the brain. The reaction is *Based on a paper presented at a specialgroup of Symposia entitled, "Frontiers in AgingResearch", arranged by the program committee of the Biological Sciences Section of the Gerontological Society, San Francisco Meeting, November 18-22, 1977. tSupported by Research Grant AG 00383 from the National Institute on Aging.

164 charge-dependent and as ruthenium red becomes bound, extracellular channels become clogged, limiting further penetration. The depth of penetration becomes, therefore, a measure of charge density. It was 28% less in the dentate gyri of 25-, as compared with 3month-old Fischer-344 rats, which suggested an age-related change in charge density of extracellular molecules in the senescent dentate gyrus [2]. Although the volume of the extracellular space has not been studied in the hippocampus, it was found to be diminished significantly in the cerebral cortex of senescent rats [4]. It is likely, therefore, that the change in charge density demonstrated in the dentate gyrus is associated with a decrease in volume of the extracellular compartment, which presumably is associated with an alteration in water-binding capacity of extracellular macromolecules. It has been proposed that this would probably affect the neuronal microenvironment by altering the capability of intercellular channels to transport metabolites [5]. It is, of course, not known how a neuron would react when it is deprived in senescence, in whole or in part, of metabolites normally reaching it through an extracellular compartment whose myriad intercellular channels separate cells and cell processes from one another. A failure of neuronal metabolism can be predicted and a variety of such failures has been reported in various regions of the senescent brain. These include age-related changes in enzymatic activity [6], nucleic acid content [7], fibrillar proteins [8] and transmitter metabolism [9]. Our studies have focused upon the rat dentate gyrus in which there is an apparent lack of cellular pathology in senescence. In the electron microscope, sections from the %

2

4

6

NUMI|R OF SYHAP$|S

NUM|III

OF | Y N A P | l |

(a) (b) Fig. I. Distributions of the number of synapses involving d~adritic shafts (a) and dendritic spinM (b) per square area of the dentate gyrus molecular layer in groups of 3-month old rats (shaded area) and of 25-month old rats (unshaded area) (from Geinisman and Bondareff [II ] ).

165 dentate gyri of 25-month old rats were difficult or even impossible to distinguish from those of 3-month old animals. However, quantitative analysis of synaptic numbers in the middle third of the molecular layer demonstrated a significant difference [ 10, 11 ]. There was a 35% decrease in the number per square area of synapses involving dendritic spines in the dentate gyri of 25-month old animals as compared with those 3 months of age (Fig. 1). A remarkably similar loss of synapses was found in the supragranular portion of the molecular layer [12]. Here there was a 35% decrease in the number per square area of synapses involving dendritic shafts and a 24% decrease in the number involving dendritic spines (Table I). In both populations of synapses the mean postsynaptic length (taken as an indication of synaptic size) was measured and although there were differences between synapses involving shafts and spines, the former being somewhat larger than the latter, there were no significant differences between young and senescent animals. A quantitative study of synapses in the cerebellar cortex showed, similarly, a 25% decrease in total numbers of synapses in the senescence. These results differ from those involving the dentate gyrus in that they indicate a slight increase (5.5%) in the number of synapses involving dendritic shafts which is not significant and a significant 33% decrease in the number of synapses involving dendritic spines. In this study of the cerebellar cortex, 25month-old animals were compared with adults 12 months of age (Glick, unpublished data). These three studies of numbers of synapses involve three different synaptic populations. Synapses in the supragranular zone of the dentate gyrus are formed by commissural, associational and probably septal afferents and represent a different population

TABLE 1 NUMBERS OF SYNAPSES PER TISSUE SQUARE AREA IN THE SUPRAGRANULAR ZONE OF THE DENTATE GYRUS OF YOUNG ADULT (3-MONTH-OLD)AND SENESCENT(25-MONTHOLD) RATS (FROM GEINISMANet al. [12] ) Per tissue square area (66 Wn 2) Shafts

Spines

3 months

25 months

3 months

25 months

Mean ± S.E.M. per animal

2.6±0.2 3.5±0.2 2.4±0.2 2.1±0.2 2.5±0.2

1.6±0.1 1.0±0.I 2.0±0.2 2.3±0.2 1.7±0.1

17.0±0.5 14.2±0.6 14.6±0.5 14.7±0.4 13.6±0.6

13.1±0.4 13.6~0.4 11.2±0.4 9.7±0.5 8.2±0.4

Mean ± S.E,M. per group % 3-month-old

2.6±0.2

1.7±0.2

14.8±0.6

11.2±1.0

65.4

75.7

group % loss

Significancelevel

34.6 P < 0.02

24.3 P < 0.02

166 from those in the middle third of the molecular layer, which are formed by entorhinal afferents. Obviously, the loss of synapses involves more than one afferent pathway to the dentate gyrus. It is obvious also that the age-related loss of synapses involves more than the dentate gyrus. We do not know how widespread this phenomenon is or whether it is characteristic of species other than the Fischer-344 rat but loss of synapses in senegcence has been found in the senescent cerebellum (Glick, unpublished data) and in the visual cortex [13]. Synaptic loss was not found in a quantitative study of synaptic numbers in the human cerebral cortex [14]. A generalization as to the greater or lesser vulnerability of synapses involving dendritic spines or shafts is unwarranted but there do appear to be differences in losses of synapses depending upon the afferent pathways involved. It appears, therefore, that whereas synapses involving dendritic shafts and spines are lost to about the same degree in the dentate gyri of senescent rats, synapses involving dendritic spines are preferentially lost in the cerebellar cortex. TABLE II NUMBER OF GRANULE CELLS IN SUPRAGRANULAR ZONE OF THE DENTATE GYRI OF 3AND 25-MONTH-OLDRATS Cell number (per 100 urn) 3 months

25 months

Mean 2 S,E.M. per animal

20.7~1.8 16.5±0.5 19.3±2.6 19.720.4 19.721.4

21.9~1.5 18.220.8 17.2±1.6 17.8±2.3 16.321.0

Mean 2 S.E.M. per group

19.22 0.7

% of 3-month-old group Significance level

18.3 ~ 1.0 95.3 P> 0.5

These findings suggest that the loss of synapses m senescence is not secondary to a loss of neurons giving rise to postsynaptic dendritic shafts or spines. Were it otherwise, a decrease in the number of granule cells in the dentate gyri of senescent rats would be anticipated. But no significant difference was found when the numbers of granule cells were estimated in 3- and 25-month old rats (Table II). In addition, it appears unlikely that the loss of synapses depends directly upon a prior loss of dendritic shafts and/or spines by a constant population of postsynaptic neurons. Although the numbers of dendritic branches and spines have been shown to be diminished in various regions of the senescent rat brain, including the dentate gyrus, the loss of synapses does not appear to depend solely on a loss of dendrites. This is indicated by the fact that synaptic loss in the dentate gyms is significant in terms both of square area tissue of synapses/unit square area of tissue (Table I) and of dendrite unit length (Table III). When all axe-dendritic synapses involving d~ndritic spines within 1.5 tun of longitudinally sectioned dendritic segments were counted

167 TABLE III NUMBERS OF SYNAPSES PER UNIT LENGTH OF DENDRITE IN THE SUPRAGRANULAR ZONE OF THE DENTATE GYRUS OF YOUNG ADULT (3-MONTH-OLD) AND SENESCENT (25MONTH-OLD) RATS (FROM GE!NISMAN et al. [12} ) Per dendrite unit length {10 Izm) Shafts

Mean -+ S.E.M. per animal

Mean ± S.E.M. per group

Spines

3 months

25 months

3 months

25 months

6.3 ± 0.9

4.7 -+ 0.9

4.2 ± 0.6

3.5 ± 0.6

(n = 248)

(n = 195)

8.4 ± 1.3 (n = 253)

3.1 ± 0.6 (n = 239)

3.7 ± 0.8

3.6 -+ 0.6

7.2 ± 1.1 (n = 265)

3.2 ± 0.6 (n = 230)

7.7 ± 1.4

3.7 ± 0.6

6.2 -+0.8 (n = 300)

5.4 +- 1.0 (n = 225)

7.4 ± 0.9

3.5 ± 0.8

7.0 ± 1.0 (n = 320)

4.2 +-0.9 (n = 211)

6.0 ~ 1.I

3.3 ± 0.7

6.8 ± 0.2

4.1 -+ 0.4

5.8 ± 0.8

3.5 * 0.1

% 3-month-old group

60.4

60.3

% loss

39.6

39.7

Significance level

P < 0.001

P < 0.05

along 10 tim-long segments o f dendrites, the number o f synapses per unit dendrite length was found to be significantly lower in 25-month old rats (57.6 + 4.3) than in 3-month old rats (89.4 ± 4.0) [15]. This means that the loss o f synapses in the senescent brain cannot depend exclusively upon the prior loss o f dendrites. The loss o f synapses and the loss o f dendrites in the senescent dentate gyrus appear, then, to be related, at least in the supragranular zone, but there is good reason to suggest that presynaptic elements fail first. It seems, reasonable to suggest that age-related loss o f synapses represents a process o f partial deafferentation resulting primarily from an inadequacy o f presynaptic components to maintain the structural integrity o f synapses during senescence. The maintenance o f structural integrity has proved difficult to demonstrate. It is, o f course, well-known that synaptic membranes contain glycoproteins and that these are synthesized in neuronal perikarya and transported b y axonal transport to plasma membrane sites in synapses where they are incorporated. An examination o f glycoprotein synthesis and transport to dentate gyrus synapses seemed to be a logical place to begin a study o f mechanisms involved in the loss o f synapses in senescence. Because 3(H)-fucose is known to be incorporated into giycoproteins exclusively, it was injected into the septal nuclei o f 3- and 25-month old Fischer-344 rats and at various

168 SA TCA-ffTA INSOLUILE FRACTION

I

.

80C

_/

S

|0

1S

20

2S

80

rain

Fig. 2. Time course of transport of 3ffl)-fueose-labeled glyeoproteins to dentate gyrus, DG, (upper pair

of curves) and neoeortex, NC, (lower pair of emwes)m 3-month.old (---) and 25-month-old ( ~ ) rats (from Geinisman et al. [ 16] ).

time intervals after injection the specific activities of TCA-PTA insoluble fractions of the septal nuclei and the hippocampal formation were determined by scintillation counting [16]. The rate of 3(H)-fucose incorporation into the septal nuclei was found to be linear with time and there was no significant difference in the rate of incorporation in young adult and senescent rats. There were, however, significant differences in the arrival time and the amount of 3(H)-fucose-labeled glycoprotein (Le., in the amount of radioactivity recovered in the TCA-PTA insoluble fraction) transported to the dentate gyrus in the young adult and senescent animals (Fig. 2). Rostra], middle and caudal sections of the dentate gyrus were analyzed for TCA-PTA insoluble radioactive material. In the rostral section the first appearance of 3(I-I)-fucose-labeled glycoprotein in an amount significantly greater than that of a background region (neighboring neocortex) was found 20 rain after 3(H)-fueose injection irtto the septal nuclei of the 3-month-old rats. By 25 min rostral sections of the dentate gyrus of senescent rats also contained amounts of radioactivity significantly greater than in the adjacent neocortex. From these data the rate of axonal transport of labeled glycoprotein in the septo-hippocampal pathway was estimated to be 288 mm/day in young adults and 221 mm/day in senescent rats. Rates of axonal transport in excess of 100 mm/day are generally considered to depend upon an interaction between microtilarnents and microtubules similar to that which underlies the contraction of skeletal muscle. The interaction which generates a motile force, is energy dependent and requires an adequately functioning metabolic machine. A rate of axonal transport in the senescent brain, which is 25% less than that found in young adults, implies a failure of the machine which, although it affects glyco-

169 protein transport, appears not to affect uptake or incorporation of a(H)-fucose. It does not, therefore, indicate necessarily a difference in the amount or composition of the glycoproteins transported to synapses in the senescent brain. Actually the amount of labeled glycoprotein transported in senescence to the dentate gyrus during a 30 min period was shown to be about two times less than at age 3-months and there is certainly ample reason to suspect a qualitative and/or quantitative difference in the glycoprotein composition of presynaptic plasmic membranes. It has not been possible to demonstrate age-related changes in the presynaptic membranes of dentate gyrus synapses and recent attempts to do so have been stymied by the unavailability of a suitable model. In lieu of a more suitable model, giant neurons were isolated from the lateral vestibular nuclei of young and old rat brain stems and used in an initial attempt to define age-related differences in synapses [17]. These isolated nerve cells are capable of carrying of normal neuronal functions in vitro. Possible agerelated difference in their surface glycoproteins were sought by examining their capacity to bind with Concanavalin A, a lectin well-known to bind to specific carbohydratecontaining sites in plasma membranes. It was coupled with fluorescamine and the intensity and distribution of fluorescence in the membrane was determined by means of microspectrofluorometry. Neurons isolated from senescent animals were found to be significantly more fluorescent than those from young adults and patches or caps of fluorescence were found on neurons senescent rats while neurons from young rats bound Concanavalin A in an essentially uniform manner. These data suggest an age-related difference in the amount or configuration of plasma membrane associated glycoproteins. It is not yet possible to characterize this difference adequately, but it is known that prior incubation with neuroaminidase or trypsin effects it to a remarkable degree. Enzyme treatment prior to incubation in Concanavalin A had little effect on the intensity or distribution of Concanavalin bound to the plasma membranes of senescent nerve cells but it significantly changed the amount binding to cells isolated from young adults. In the treated young group there was a significant increase in the intensity of fluorescence and the pattern of binding changed to resemble that characteristic of neurons isolated from senescent rats. These findings suggest that the difference in plasma membrane associated glycoproteins in young and senescent animals may depend upon sialic acid which may mask Concanavalin receptors on plasma membranes of young animals. The time course of synaptic loss in the dentate gyrus is not known. In Fisher-344 rats it is apparent by the 24th or 25th month at which time no vestiges of synaptic atrophy are apparent. Synaptic loss is, however, coincident at 25 months with a loss of dendrites. But the loss of synapses has been shown to vary independently of dendrites [12] and loss of dendrites appear, therefore, to be a sequella of synaptic loss. Synapses appear to diminish in number in the senescent dentate gyrus because certain populations of presynaptic nerve cells lose the capacity to maintain the integrity of presynaptic plasma membranes. This results in a partial deafferentation comparable to that which has been shown to follow a variety of experimental and surgical procedures [18]. The process of partial deafferentation, resulting from the process of aging is associated with a loss in volume of the total dendritic compartment in the dentate gyrus [15].

170 By means of electron microscopic morphometry it is possible to demonstrate a significant decrease in the total volume fraction o f dendrites in the supragranular zone o f the dentate gyrus in 25-month-old Fischer-344 rats [15]. It is not clear from these data or from those o f light microscopic study o f Golgi preparation, whether this loss o f volume fraction represents a loss, thinning and/or shortening o f dendrites. The fact that the decrease in volume fraction in senescence is in excess of the reduction of dendritic profiles counted in electron micrographs suggests that the loss in dendritic volume fraction is the consequence o f a loss o f smaller, higher order dendrites. Light microscopic studies by others which demonstrate a loss o f dendritic spines [ 19] and what appears to be a loss o f more peripheral portions [20] o f dendritic branches in cerebral cortex of the Fischer-344 rat, are also indicative of a loss o f higher order dendrites. In the dentate gyrus, therefore, it is suggested that dendritic atrophy resulting from an age-related deafferentation begins in the peripheral portions o f dendrites and proceeds proximally towards the perikaryon o f the postsynaptic neuron. A correlative decrease in the surface area o f dendrites remaining in the dentate gyrus o f the senescent rat, which has been shown [ 15], is associated with a loss of postsynaptic surface before there is any significant change in the number o f postsynaptic granule cells. The consequence should be a loss o f interneuronal connections, which may well have functional correlates, before any change in neuronal morphology or number becomes apparent.

REFERENCES 1 F. O. Schmitt and F. E. Sampson, Jr., Brain cell microenvironment, Neurosci. Res. Prog. Bull., 7 (1969) 277--417. 2 W. Bondareff and S. Lin-Liu, Age-related change in the neuronal microenvironment: penetration of ruthenium red into extraceUular space of brain in young adult and senescent rats, Am. J. Anat., 148 (1977) 57-64. 3 W. Bondareff, An intercellular substance in rat cerebral cortex: Submicroscopic distribution of ruthenium red,Anat. Rec., 157 (1967) 527-536. 4 W. Bondareff and R. Narotzky, Age changes in the neuronal mieroenvironment, Science, 176 (1972) 1135-1136. 5 W. Bondareff, R. Narotzky and A. Routtenberg, lntrastriatal spread of cateeholamines in senescent rats, J. Gerontol., 26 (1971) 163-167. 6 W. Meier-Ruge, K. Reiehlmeirer and P. lwangoff, Enzymatic and enzyme histochemical changes of the aging animal brain and consequences for experimental pharmacology on aging, in R. D. Terry and S. Gershon (eds.), Neurobiology of Aging, Raven Press, New York, 1976, pp. 379-388. 7 U. Ringborg, Composition and content of RNA in neurons of rat hippocampus at different ages, Brain Res., 2 (1966) 296-298. 8 K. Igbal, I. Grundke-Igbal, H. M, Wisniewski, 1. K. Korthals and R. D. Terry, in R. D. Terry and S. Gershon (eds.), Neurobiology o f Aging, Raven Press, New York, 1976, pp. 351-360. 9 E. MeGeer and P. L. McGeer, Neurotransm: her metabolism in the aging brain, in R. D. Terry and S. Gershon (eds.), Neurobiology o f Aging, Raven Press, New York, 1976, pp. 389--404. 10 W. Bondareff and Y. Geinisman, Loss of synapses in the dentate gyrus of the senescent rat, Am. Y. Anat., 145 (1976) 129-136. 11 Y, Geinisman and W. Bondareff, Decrease in the number of synapses in the senescent brain: a quantitative electron microscopic analysis of the dentate gyrus molecular layer in the rat, Mech. AgeingDev., 5 (1976) 11-23. 12 Y. Geinisman, W. Bondareff and J. T. Dodge, Partial deafferentation of neurons in the dentate gyrus of the senescent rat, Brain Res., 134 (1977) 541-545.

171 13 M. L. Feldman, Aging changes in the morphology of cortical dendrites, in R. D. Terry and S. Gershon (eds.), Neurobiology of Aging, Raven Press, New York, 1976, pp. 211-228. 14 B. G. Crag,g, The density of synapses and neurons in normal, mentally defective and aging human brains, Brain, 98 (1975) 81-90. 15 Y. Geinisman, W. Bondareff and J, T. Dodge, Dendritic atrophy in the dentate gyrus of the senescent rat, Am..I. Anat., (1978) in press. 16 Y. Geinisman, W. Bondareff and A. Telser, Transport of a(H)-fucose-labeled glycoproteins in the septo-hippocampal pathway of young adult and senescent rats, Brain Res., 125 (1977) 182-186. 17 K. D. Bennett and W. Bondareff, Age-related differences in binding of Concanavalin A to plasma membranes of isolated neurons, Am. Z Anat., 150 (1977) 175-184. 18 A. Globus, Brain morphology as a function of presynaptic morphology and activity, in H. Riesen (ed.), The Developmental Neuropsychology of Sensory Deprivation, Academic Press, New York, 1975, pp. 9-91. 19 M. L. Feldman and C. Dowd, Loss of dendritic spines in aging cerebral cortex, Anat. EmbryoL, 148 (1975) 279-301. 20 D. W. Vaughan, Age-rela.ted deterioration of pyramidal cell basal dendrites in the rat auditory cortex, Z Comp. NeuroL, 171 (1977) 501-516.