THE JOURNAL OF COMPARATIVE NEUROLOGY 403:459–470 (1999)
Morphometric Studies of the Aged Hippocampus: I. Volumetric Analysis in Behaviorally Characterized Rats PETER R. RAPP,1* EDWARD C. STACK,2 AND MICHELA GALLAGHER3 of Aging Laboratories, Fishberg Research Center for Neurobiology, Department of Geriatrics and Adult Development, Mount Sinai School of Medicine, New York, New York 10029–6574 2Department of Biology, Boston College, Chestnut Hill, Massachusetts 02167 3Department of Psychology, Johns Hopkins University, Baltimore, Maryland 21218–2686 1Neurobiology
ABSTRACT The present investigation examined the structural integrity of the aged hippocampus by using computer-aided morphometry to quantify the volume of principal hippocampal circuits in young, mature adult, and aged Long-Evans rats. A key feature of the experimental design was that the status of hippocampal-dependent learning and memory was documented prior to histologic evaluation. The following regions, which were visualized by using Timm staining, were included in the analysis: 1) outer portions of the dentate gyrus molecular layer (OML) innervated by the lateral entorhinal cortex, 2) middle portions of the molecular layer (MML) that receive input from the medial entorhinal cortex, 3) the commissural/associational zone (IML) immediately adjacent to the granule cell layer, and 4) the hilus and mossy fiber projection to the CA3 pyramidal cell field (MF). To identify morphometric changes that emerge during the same segment of the life span as age-related learning impairment, analysis of the volumetric results focused on comparisons between the mature adult group and the aged group. Among the individual regions that were analyzed, age-related decreases in total volume were restricted to the MML. This effect, however, occurred against a background of other, subtle changes that, together, reflected substantial reorganization in the normal balance of hippocampal circuitry. Age-related decreases in the proportion of the molecular layer (ML) that comprises the MML were accompanied by a corresponding increase in relative IML volume. The ratio between the volumes of the MML and the MF also displayed significant age-related decline. Overall, aging affected septal levels of the hippocampus disproportionately, and, with the exception of MML/MF volume ratio, the temporal hippocampus was spared. Finally, the status of spatial learning among the aged animals correlated selectively with decreases in the MML/ML and MML/MF ratios. These results demonstrate that the effects of aging are regionally selective and circuit specific, and they suggest that connectional reorganization may contribute to age-related decline in the computational functions of the hippocampus. J. Comp. Neurol. 403:459–470, 1999. r 1999 Wiley-Liss, Inc. Indexing terms: entorhinal cortex; histological techniques; maze learning; stereology
Hippocampal function is altered substantially during the course of normal aging. Early support for this conclusion includes evidence that the rate of decay of long-term potentiation (LTP; Barnes, 1979) and the amount of stimulation necessary to induce hippocampal kindling (de ToledoMorrell et al., 1984) are increased significantly in aged rats relative to young animals. These neurophysiological changes also correlate with the magnitude of deficits in hippocampal-dependent learning and memory, consistent with the proposal that age-related impairments in cellular
r 1999 WILEY-LISS, INC.
plasticity contribute to cognitive decline (Barnes et al., 1994; de Toledo-Morrell et al., 1988a). Additional insight into the information processing capacities of the aged Grant sponsor: NIH; Grant number: AG09973. *Correspondence to: Peter R. Rapp, Neurobiology of Aging Laboratories, Mount Sinai School of Medicine, Box 1639, One Gustave L. Levy Place, New York, NY 10029–6574. E-mail:
[email protected] Received 19 December 1997; Revised 11 August 1998; Accepted 20 August 1998
460 hippocampus derives from recent studies examining neuronal activity in direct relation to behavior (Barnes, 1998). Barnes et al. (1997), for example, monitored activity simultaneously in large numbers of hippocampal pyramidal cells while young and aged rats explored a figure-8 maze. Location-specific firing patterns were documented twice for each set of neurons, holding the arrangement of spatial cues in the testing environment constant. Ensemble activity in young animals was correlated highly across recording sessions: neurons maintained the same ‘‘map’’ of place-related firing during multiple episodes of maze exploration. In approximately 30% of the recording sessions for aged rats, by comparison, place field activity in one session failed to predict the distribution of firing when the same rat returned to the maze. Related studies demonstrate that, relative to both young subjects and aged rats with intact spatial abilities, hippocampal neurons in aged rats with pronounced spatial learning deficits code only a subset of the stimulus information available to guide navigation (Tanila et al., 1997a). Firing patterns in these rats also exhibit less flexibility under conditions that promote changes in neural representations in younger animals and aged rats with preserved spatial abilities (Tanila et al., 1997b). Together, these findings suggest that changes in the computational properties of the hippocampus are among the factors that contribute to age-related spatial learning impairment. Against this background of declining functional capacities, the gross structural integrity of the aged hippocampus is largely preserved. Recent investigations using sensitive and reliable stereological methods of quantification have established that, in rats (Rapp and Gallagher, 1996; Rasmussen et al., 1996), monkeys (Rapp, 1995; Peters et al., 1996), and humans (West, 1993), the total number of granule cells in the dentate gyrus, and pyramidal neurons in fields CA3 and CA1, remains unchanged over the life span (but see Simic et al., 1997). Moreover, research incorporating behavioral assessment in animal models has provided the additional insight that neuron loss in these areas is not required for the presence of age-related deficits in hippocampal learning and memory. In the rat, for example, hippocampal neuron number is comparable across young subjects and aged animals with or without spatial learning deficits (Rapp and Gallagher, 1996; Rasmussen et al., 1996). The implication of these findings is that electrophysiological markers of aging are not secondary to neuronal loss but, instead, reflect functional changes or reorganization within the surviving architecture of the aged hippocampal formation. Assessing the connectional alterations that might contribute to age-related impairment in hippocampal function is facilitated by the orderly arrangement of afferent cortical input and intrinsic circuitry in this system (Amaral and Witter, 1995). A key feature of this organization is that the majority of direct cortical input, originating in the adjacent entorhinal cortex, synapses onto the dendrites of granule cells in outer portions of the dentate gyrus molecular layer (ML). In rats, this projection follows a topographic arrangement by which input from the lateral entorhinal cortex is directed to the most distal segments of the granule cell dendrites, whereas middle portions of the ML are innervated by projections from neurons in the medial entorhinal cortex. The most proximal portions of the granule cell dendrites, in contrast, receive an intrinsic commissural/associational input arising from cells in the
P.R. RAPP ET AL. polymorphic or hilar region of the dentate gyrus. Output from the granule cells comprises the next link in the largely unidirectional organization of hippocampal circuitry, giving rise to the mossy fiber projection to the CA3 pyramidal cells of the hippocampus proper. These features of hippocampal neuroanatomy are readily apparent in histological material stained by using the Timm method (Zimmer and Haug, 1978), allowing discrete morphometric analysis of pathways with defined connectional characteristics. Moreover, the sensitivity of this approach for examining morphological plasticity has been validated in a number of experimental models, including studies on hippocampal synaptic reorganization after damage to the entorhinal cortex (Cavazos et al., 1992; Steward, 1992). Informed by this background, Coleman et al. (1987) adopted a strategy similar to the present study to quantify potential volumetric alterations over the life span in the hippocampus of Fischer-344 (F344) rats. Although the volume of the individual connectional zones labeled by using the Timm method were not reliably different in rats from 12–37 months of age, significant changes were observed in the relative balance between certain of these areas. Specifically, whereas the total volumes of the perforant path and mossy fiber projections remained relatively stable, the ratio between these measures was affected robustly as a function of age. This effect reflected a subtle, age-related decline in the volume of the perforant path termination zone concomitant with an increase in the volume of the mossy fiber projection. Ratios calculated from the volumes of other connectional zones were comparable across age groups, suggesting that the effects of aging are circuit specific in the rat hippocampus (Coleman et al., 1987). The present investigation was designed to address several unresolved issues concerning the organization of principal circuitry in the aged hippocampus, including the relevance of the observed changes for cognitive decline. Behavioral studies demonstrate that the status of learning and memory supported by the hippocampus varies considerably across aged individuals (Gallagher et al., 1995). These findings establish a useful context for evaluating the functional significance of neural alterations in the aged brain and for detecting markers of brain aging that may not be apparent when comparisons are based on chronological age alone (Rapp and Amaral, 1992). Taking advantage of this approach, our primary aim was to reexamine the effects of aging on hippocampal volume in relation to the status of learning and memory supported by this structure. Current neuroanatomical research has also provided considerable new detail regarding the topographic organization of hippocampal input from the entorhinal cortex (Burwell and Amaral, 1998a,b; Dolorfo and Amaral, 1998), and, guided by these findings, our additional goal was to examine the regional distribution of volumetric change in the aged hippocampus. Toward this end, age-related volumetric change was examined separately in septal and temporal aspects of the hippocampus, corresponding to regions that receive distinct complements of cortical and subcortical input relayed through the entorhinal cortex.
MATERIALS AND METHODS Subjects A total of 23 male, pathogen-free, Long-Evans rats (Charles River Laboratories, Raleigh, NC) served as sub-
HIPPOCAMPAL VOLUME IN AGED RATS jects. Three age groups were studied: young rats at 3–4 months of age (n ⫽ 6), mature adult rats at 9–10 months of age (n ⫽ 7), and aged rats at 27–28 months of age (n ⫽ 10). This design provided a basis for documenting the morphometric effects of advanced age relative to mature adults, independent of a variety of late developmental events that appear to affect hippocampal volume (see Data Analysis, below). Rats were housed singly in a climate-controlled vivarium (25°C) on a 12:12 hour light:dark schedule. Standard laboratory chow and water were available ad libitum in the home cage throughout the experiment. Sentinel screening for a panel of viral antibodies routinely was negative in aged animals from this colony, and no rats that were included in the present investigation had pituitary tumors, kidney necrosis, or other frank pathologies that would be expected to influence the experimental outcome measures. All procedures relating to the maintenance, treatment, and perfusion of experimental animals were approved by Institutional Animal Care and Use Committees and conformed to the guidelines of the National Institutes of Health Guide for the Care and Use of Laboratory Animals.
Behavioral testing procedures Spatial learning was assessed prior to perfusion and histological analysis by using standardized behavioral procedures that were identical to numerous previous experiments (for review, see Gallagher et al., 1995). Briefly, animals were tested on two versions of the Morris water maze: one that required the use of distal cues to learn the position of a camouflaged escape platform and a second version assessing nonspatial, goal-approach learning. The former assessment requires the functional integrity of the hippocampus (Morris et al., 1982) and reveals significant spatial learning impairment in a substantial proportion of aged Long-Evans rats (Gallagher et al., 1993). Subjects were trained for three trials per day, with an intertrial interval of approximately 60 seconds, for a total of 8 consecutive days. The location of the hidden escape platform remained constant, and, on each trial, rats swam for 90 seconds or until they found the platform. The starting location varied randomly across trials among four equidistant points around the perimeter of the apparatus. Every sixth trial (i.e., the last trial on every other day) was a probe test in which the platform initially was retracted to the bottom of the maze for 30 seconds. Probe testing allows assessment of the search strategies used to navigate the maze, yielding measures that are sensitive to aging and experimental hippocampal damage in young rats (Morris et al., 1982; Gallagher et al., 1993). Nonspatial learning was tested subsequently in a single session of six trials in which rats swam to a black escape platform that varied in location across trials and protruded above the surface of the pool. Aged rats from the same colony examined here routinely are unimpaired on the nonspatial cue version of the water maze, indicating that they can swim proficiently and are motivated to escape. Performance was monitored throughout all phases of testing by using a video tracking system designed for this purpose (HVS Imaging, Hampton, United Kingdom). The primary behavioral measure used in the present analysis was a spatial learning index score derived from data collected during the three interpolated probe trials in which the platform initially was unavailable for escape. This graded performance measure, as described in detail
461 elsewhere, reflects the average proximity from the training location of the escape platform during probe testing; accurate searching focused on the target location results in lower learning index scores (Gallagher et al., 1993). This measure effectively distinguishes spatial learning impairment from nonspecific performance deficits that presumably are unrelated to hippocampal function (e.g., motor deficits).
Histological processing Approximately 1 week after the conclusion of behavioral testing, rats were anesthetized deeply with 35% chloral hydrate (i.p.) and were perfused through the ascending aorta according to a modified Timm stain procedure described by Sloviter (1982). Briefly, perfusion was initiated for 5 minutes with a 0.37% sulphide solution in phosphate buffer (PB), pH 7.2, which was delivered by a peristaltic pump at a flow rate of 32 ml/minute. Ten percent formalin in 0.1 M PB, pH 7.4, was then perfused at the same flow rate for a total of 5 minutes. After removal from the skull, brains were stored overnight in the same fixative solution with the addition of 20% glycerin. On the following day, the left hippocampus from each brain was dissected free of the surrounding tissue and placed in 20% glycerin without fixative for 48 hours. All perfusion reagents were maintained at room temperature. Hippocampi were frozen subsequently in an ‘‘extended’’ orientation on pulverized dry ice and stored at ⫺70°C until further processing. This procedure, as detailed in developmental studies by Gaarskjaer (1978), attenuates the curvature of the hippocampus, allowing the preparation of histological sections that are roughly orthogonal to the long axis of the hippocampus throughout nearly its entire rostrocaudal extent. In the present experiments, this greatly simplified the delineation of regional boundaries for quantitative morphometric analysis (see below). Serial histologic sections were cut coronal to the longitudinal axis of the hippocampus on a freezing microtome at a nominal thickness of 30 µm. Two, 1-in-8 series of sections (240-µm spacing) were mounted on acid-washed, gelatincoated slides for Timm staining, and a closely adjacent series was processed by using standard Nissl methods (0.25% thionin). Timm staining was conducted according to Sloviter (1982), with modifications from earlier descriptions. Briefly, the tissue was defatted in chloroform and ethanol (1:1), rehydrated through descending alcohols, and metal-sulphide complexes were labeled with silver [285 mg AgNO3 in 300 ml of a citrate-buffered, gum Arabic suspension with 5.1% (weight/volume) hydoquinone] under light-tight conditions. Staining intensity with this method is temperature sensitive, and, in the present experiment, the development was carried out at 26°C by using a water bath. Two closely adjacent series of sections from each hippocampus were stained for 50 minutes and 60 minutes, respectively. Although both sets of material were used for qualitative, descriptive purposes, all quantitative morphometric data were derived from sections that were stained for 60 minutes. After washing, sections were dehydrated and coverslipped with DPX (BDH Laboratory Supplies, Poole, United Kingdom).
Quantitative volumetric analysis Timm-processed material was viewed under brightfield illumination on a Leitz Medilux microscope (Leica, Inc., Deerfield, IL) interfaced with a CCD color video camera
462
P.R. RAPP ET AL.
(Optronics Engineering, Goleta, CA) and a NeuroLucida digitizing morphometry system (MicroBrightField, Inc., Colchester, VT). By using a 10⫻ Plan-achromatic objective, live color images of the histological material were displayed on a high resolution video monitor at a final magnification of approximately 200⫻. The following regional boundaries, which are delimited sharply by the Timm method, were traced and digitized for quantitative analysis: 1) outer portions of the dentate gyrus molecular layer (OML) innervated by the lateral entorhinal cortex, 2) middle portions of the molecular layer (MML) that receive projections from the medial entorhinal cortex, 3) the commissural/associational zone (IML) immediately adjacent to the granule cell layer, 4) the granule cell layer (GCL) itself, and 5) the hilus and mossy fiber projection to the CA3 field of the hippocampus (MF). For the latter region, no effort was made to exclude occasional small patches within the pyramidal cell layer that were unstained or to quantify separately supra-, intra-, or infrapyramidal labeling. Starting with one of the two most anterior sections, which was selected on a random basis across brains, every other section through the rostrocaudal extent of each hippocampus was subjected to quantitative analysis. By using this sampling strategy, an average of approximately 18 histological sections per brain were analyzed. Cross-sectional areas were calculated for all of the digitized borders, and the total volume for each region of interest was estimated as the sum of the area measurements multiplied by the distance between the histological sections (i.e., 480 µm). Relative differences between groups were of greater interest than the absolute volume estimates, and the data were not corrected for tissue shrinkage associated with histological processing. All morphometric quantification was conducted blind with respect to chronological age and the outcome of behavioral testing.
Data analysis The rat hippocampus undergoes substantial postnatal development, and previous studies, using methods similar to those used in the present investigation, have documented significant structural change in certain hippocampal subregions during the first year of life (Zimmer and Haug, 1978; Gaarskjaer, 1985; Coleman et al., 1987). Thus, to identify alterations that are associated specifically with advanced age and that emerge during the same segment of the life span as spatial learning impairment, statistical analyses in the present experiments focused on hippocampal volume in aged rats relative to mature adults at 9–10 months of age. Although brains from 3–4-monthold rats were also studied, these data are reported primarily as a replication of earlier normative findings and were not considered in the formal analysis of potential age effects. Differences in hippocampal volume were tested by factorial analysis of variance (ANOVA), exploring the source of significant main effects by subsequent groupwise comparisons (Bonferroni-Dunn). Following an analytic strategy that has been validated in earlier research (Baxter and Gallagher, 1996), a twotiered approach was adopted to relate potential morphometric alterations during aging to the status of learning and memory supported by the hippocampus. First, for all group comparisons of the volumetric results, aged animals were divided into ‘‘unimpaired’’ and ‘‘impaired’’ subgroups, which were distinguished on the basis of the spatial learning index scores derived during behavioral testing. The aged-
Fig. 1. Scatter plot of individual spatial learning index scores derived from probe testing in the water maze-task for young (n ⫽ 6), mature adult (n ⫽ 7), and aged (n ⫽ 10) rats. With this measure, lower values reflect more accurate searching focused on the training location of the escape platform. Note the substantially greater variability in learning scores among aged rats relative to younger animals.
unimpaired group (n ⫽ 5) consisted of 27–28-month-old rats that learned the place version of the water maze within the range of values for a large body of normative data from young animals that were tested under identical conditions. Aged-impaired subjects (n ⫽ 5), by comparison, scored outside this normative range and outside the range of all young and adult subjects tested concurrently in the present study. Volumetric differences that proved statistically significant according to this between-group approach were then explored by linear correlation analysis. These comparisons focused exclusively on data from the aged subjects, testing the specific proposal that the magnitude of volumetric change in the aged hippocampus predicts individual variability in the status of spatial learning. For purposes of photographic documentation, images were acquired with a MicroLumina digital scanning camera (Leaf Systems, Inc., Southborough, MA) interfaced with a Power Macintosh 8600/200 computer (Apple Computer Inc., Cupertino, CA). Aside from contrast and brightness adjustment, photomicrographs in the present report are unretouched.
RESULTS Spatial learning in the aged rat The outcome of behavioral testing was entirely consistent with numerous earlier studies using identical procedures (Gallagher et al., 1995), and only a brief description is provided here. Over the course of training in the spatial version of the water maze, average escape latencies for the aged group were significantly longer than in young and adult animals (repeated-measures ANOVA group effect: F2,18 ⫽ 16.80; P ⬍ 0.0001). Not all aged rats, however, displayed robust spatial learning impairment. Figure 1 presents the learning index scores for individual animals, reflecting average proximity during probe testing from the training location of the hidden escape platform (see Materials and Methods). By using this measure, a search concentrated in the vicinity of the target location results in lower scores. Although the aged group was significantly
HIPPOCAMPAL VOLUME IN AGED RATS
Fig. 2. Low-magnification, digital scanning photomicrographs of Timm staining in an extended hippocampal preparation from a representative adult rat. The regions quantified in the present experiments are readily apparent in sections through septal (top), middle (middle), and temporal (bottom) levels of the hippocampus (for regional borders, see Fig. 3). Increased intensity of Timm staining at caudal levels of the dentate gyrus was observed consistently. Scale bar ⫽ 500 µm.
impaired (F2,20 ⫽ 4.36; P ⬍ 0.05), half of the rats performed comparably to normative control values (index scores ⬍ 245), whereas the remainder scored considerably outside the range of young and adult subjects. The latter groups, however, did not differ according to this measure (P ⬎ 0.1). In contrast to these results for the hippocampaldependent, place version of testing, no age-related deficit was observed on the nonspatial, goal-approach task (P ⬎ 0.1; data not shown). This pattern of results provides an appropriate context for evaluating potential morphometric alteration during aging in relation to individual differences in the status of spatial learning supported by the hippocampus (Gallagher et al., 1995; Baxter and Gallagher, 1996). Because the behavioral data indicated that spatial learning impairment emerges between 9–10 months of age and 27–28 months of age, formal statistical analysis of the volumetric results focused on the adult and aged groups, avoiding the potential confound of developmental events that occur earlier in the life span (Coleman et al., 1987).
Distribution of Timm staining in the extended rat hippocampus: Qualitative observations The overall pattern of Timm staining in the hippocampus of the Long-Evans rat (Figs. 2, 3) was essentially the same as described for other strains (Zimmer and Haug, 1978; Gaarskjaer, 1985), and there were no grossly apparent differences in the quality or intensity of label in
463 material from young and aged animals. Throughout the transverse and rostrocaudal extent of the dentate gyrus, a zone of moderately intense silver labeling occupied approximately the outer one-third of the ML. This band bordered an unstained region in the MML layer that comprised approximately 40% of the total ML width. Neuroanatomical studies document that these sharply delineated outer and middle aspects of the ML correspond to the terminal fields of the lateral and medial entorhinal cortex, respectively (Steward, 1976). The remaining inner portion of the ML, positioned between the easily distinguished dentate gyrus granule cells and the unstained region of medial entorhinal cortical input, also displayed moderately heavy labeling. This narrow band defines the intrinsic, commissural/associational pathway, arising from a collection of polymorphic cell types in the hilus of the dentate gyrus (Amaral and Witter, 1995). Thus, the overall distribution of Timm staining in the ML was trilaminar, with the clearly defined OML and IML demarcating the unstained MML. The final hippocampal component quantified in the present experiments was the well-characterized MF system. This projection is particularly prominent in Timmstained material, appearing as dense label distributed throughout the hilus of the dentate gyrus and along both stratum lucidum and the intrapyramidal course of field CA3. The smaller, infrapyramidal segment of the MF projection also is stained by this method. For the present analysis, these regions were considered together in estimating the total volume of the MF system. The normal curvature of the hippocampus complicates the analysis of organizational features along its rostrocaudal axis. In standard coronal preparations from whole brains, for example, the most caudal or temporal aspects of the hippocampus appear in histological sections cut rostral to those containing more rostral or septal hippocampal levels. This problem largely can be circumvented, however, by physically straightening the hippocampus prior to sectioning, so that the septotemporal organization of the resulting material is preserved (Gaarskjaer, 1978). Figure 2 presents low-power photomicrographs of Timm-stained sections through septal, middle, and temporal levels of the hippocampus in an adult rat, and the rostrocaudal distribution of labeling is illustrated schematically in Figure 3. By comparison with the complex organization seen in whole brain preparations, the pattern of Timm staining in the extended hippocampus was remarkably regular throughout the majority of its longitudinal extent. Whereas the most posterior levels typically retained some residual degree of curvature, regional boundaries of interest were recognized easily and amenable to quantitative morphometric assessment.
Regional hippocampal volume and relation to behavior: Quantitative analysis Table 1 lists the mean estimated total volume of the OML, MML, IML, GCL, and MF system for the young, adult, aged-unimpaired, and aged-impaired groups. A noteworthy feature of these results is that between-subject variability was low (e.g., standard errors were well under 10% of the group means for most estimates), maximizing sensitivity for detecting age-related volumetric change. In agreement with earlier studies (Zimmer and Haug, 1978; Gaarskjaer, 1985; Coleman et al., 1987), subtle developmental changes were noted in hippocampal morphology, most prominently in the GCL, where total volume increased
Fig. 3. Digitized tracings of Timm-stained sections from an extended hippocampal preparation. Areas that were subjected to quantitative analysis are indicated by shading, and numbers in parentheses represent approximate distance (in millimeters) from the septal extreme of the hippocampus. Note the highly regular appearance of
the Timm-stained areas throughout nearly the entire longitudinal extent of the hippocampus. GCL, granule cell layer; IML, inner molecular layer; MF, mossy fiber system; ML, molecular layer; MML, middle molecular layer; OML, outer molecular layer; sp, stratum pyramidale.
HIPPOCAMPAL VOLUME IN AGED RATS
465
TABLE 1. Mean Total Volume (S.E.) of Hippocampal Regions (mm3 ) Group Young Adult Aged unimp Aged imp
OML
MML*
IML
GCL
MF
2.51 (0.07) 2.46 (0.09) 2.40 (0.18) 2.60 (0.09)
2.71 (0.15) 2.94 (0.08) 2.62 (0.14) 2.59 (0.09)
1.77 (0.11) 1.65 (0.04) 1.73 (0.12) 1.80 (0.04)
1.68 (0.09) 1.91 (0.06) 1.76 (0.09) 1.69 (0.06)
3.72 (0.18) 3.52 (0.14) 3.85 (0.26) 4.02 (0.12)
*Significant main effect of group, P ⱕ 0.05. GCL, granule cell layer; IML, inner molecular layer; MF, mossy fiber system; MML, middle molecular layer; OML, outer molecular layer; unimp, unimpaired; imp, impaired.
more than 13% across young and adult rats. With the exception of this effect, however, the numerical differences observed between these groups were not statistically significant. Subsequent analyses were directed at identifying morphometric alterations that occur later in the life span, coincident with age-related deficits in hippocampal learning and memory. These comparisons revealed that the volume of most components of the hippocampus remained relatively stable between 9–10 months of age and 27–28 months of age. Indeed, the only statistically robust effect occurred in the MML; relative to adults, the volume of the ML innervated by the medial entorhinal cortex was 11– 12% lower in the aged groups (F2,14 ⫽ 3.74; P ⱕ 0.05). The magnitude of this change was unrelated to cognitive status, in that the volume of the MML was comparable in unimpaired and impaired aged rats (P ⬎ 0.1). Morphometric data revealed only numerical trends for the other hippocampal regions, including age-related increases in the average volume of the IML and the MF ranging from 5% to 14% (Table 1). Because the total volume of the ML (i.e., the sum of the OML, MML, and IML measures) was virtually identical across groups, varying from 7.0 mm3 in adults to a mean of 6.9 mm3 in aged rats (P ⬎ 0.1), the presence of a significant age-related decline in MML volume suggests that a complementary increase occurs in other components of this region. This possibility was examined by comparing the ratio of volumes for selected hippocampal components in adult, aged-unimpaired, and aged-impaired rats. One set of ratio measures focused on the proportion of the total ML volume occupied by the OML, MML, and IML (Table 2). Complementing the results noted above, the percentage of ML volume comprising the MML declined significantly with age (F2,14 ⫽ 6.4; P ⱕ 0.01). The absolute magnitude of change was greatest for the aged-impaired rats relative to adult controls, and this difference proved statistically robust in subsequent group-wise comparisons (P ⱕ 0.005). Whereas a marginal decline was also observed in the aged-unimpaired group relative to adults (P ⫽ 0.06), MML/ML volume ratios were statistically equivalent across aged rats distinguished by their behavioral performance (P ⬎ 0.1). In contrast to these results for the zone of medial entorhinal cortical input, the percentage of the ML comprising the OML was not affected reliably as a function of age (P ⬎ 0.1). Although the previous analysis indicated that the absolute volume of the IML was not significantly different across groups, a robust, age-related increase was observed in the ratio of IML volume relative to the total ML (F2,14 ⫽ 7.2; P ⱕ 0.01). Both aged subgroups differed from adult controls according to this measure (P values ⱕ 0.01), and the magnitude of age-related increase in the IML/ML ratio was equivalent in aged-unimpaired and impaired rats (P ⬎ 0.1). Together with the observation that total ML volume remains stable, these findings indicate
that aging in the rat is associated with two complementary shifts in the organization of dentate gyrus circuitry: 1) a decrease in the volume of the ML that receives input from the medial entorhinal cortex, and 2) a corresponding increase in the relative volume of the intrinsic commissural/ association pathway. The MML/ML values were also correlated inversely with IML/ML ratios across the adult and aged rats (r ⫽ ⫺0.53; P ⱕ 0.05), raising the possibility that these alterations are interdependent events. Additional analyses focused on potential age-related alteration in the balance between the major inputs and outputs of the dentate gyrus based on the ratio of volumes for the ML subdivisions relative to the MF system (Table 2). Among these measures, aging was associated with a selective decline in the ratio between the volumes of the MML and MF pathway (F2, 14 ⫽ 14.2; P ⱕ 0.0005). Group comparisons confirmed that the magnitude of this effect did not differ significantly for unimpaired and impaired aged rats (P ⬎ 0.1), and both aged cohorts displayed a reliable difference relative to adult controls (P values ⱕ 0.005). Because an age-related decline was noted for the MML alone (Table 1), and because the first set of analyses failed to detect a change in total MF volume, it is possible that MML/MF ratios were affected only secondarily, independent of any alteration specific to the MF system. The overall pattern of results, however, counts against this interpretation. Age-related reductions in MML/MF ratios ranged from 19% to 23% among the aged groups, substantially exceeding the 11–12% decline documented for the MML alone. This latter measure was also a less robust marker of aging in statistical comparisons; although the trend toward a decline in MML volume was not reliable for the aged-unimpaired group relative to adults, the decreased MML/MF ratio was highly significant (P ⱕ 0.002). Considered together with the results presented in Table 1, these findings suggest that the influence of age on the MML/MF measure reflects the combination of a pronounced decline in MML volume and a more subtle increase in the volume of dentate gyrus output through the MF system. Standard statistics comparing group means (ANOVA) generally failed to reveal significant volumetric differences between aged-unimpaired and impaired rats. Nonetheless, for all parameters that differed as a function of chronological age (i.e., total MML volume and the MML/ ML, IML/ML, and MML/MF volume ratios), the magnitude of effect was numerically greatest for aged rats with spatial learning deficits (Tables 1, 2). These findings leave open the possibility that variability in the morphometric consequences of aging might predict individual differences in hippocampal-dependent learning. This proposal was explored by using a linear correlation analysis, examining potential relationships between the anatomical results and the graded measure of spatial learning ability derived from behavioral testing (i.e., the index scores). Correlations were tested only for morphometric measures that differed between adult and aged rats, minimizing the total number of comparisons and reducing the risk of revealing spurious associations. It is important that the analysis also was restricted to data from the aged rats. By using this approach, our aim was to test the hypothesis that, specifically among the aged individuals, the degree of volumetric alteration in the hippocampus is coupled to the status of spatial learning. This assessment revealed that
466
P.R. RAPP ET AL. TABLE 2. Mean Volume Ratios (S.E.) for Hippocampal Regions
Group Young Adult Aged unimp Aged imp
OML/ML
MML/ML*
IML/ML*
OML/MF
MML/MF*
IML/MF
0.360 (0.006) 0.349 (0.013) 0.355 (0.007) 0.372 (0.007)
0.387 (0.012) 0.417 (0.012) 0.389 (0.006) 0.370 (0.005)**
0.252 (0.010) 0.234 (0.005) 0.256 (0.004)** 0.258 (0.006)**
0.680 (0.020) 0.708 (0.049) 0.623 (0.015) 0.648 (0.022)
0.730 (0.019) 0.841 (0.037) 0.684 (0.010)** 0.644 (0.018)**
0.476 (0.021) 0.470 (0.013) 0.450 (0.008) 0.449 (0.009)
*Significant main effect of group, P ⱕ 0.01. For abbreviations, see Table 1. **P ⬍ 0.02 relative to adult group.
Fig. 4. Scatter plots of the MML/ML (left) and MML/MF (right) volume ratios relative to the spatial learning index scores for aged rats. Decreases in both morphometric parameters were correlated significantly with the magnitude of age-related spatial learning impairment (see text).
the total volume of the MML and the IML/ML volume ratio were not correlated significantly with the learning index scores (r values of 0.11 and 0.17, respectively; P values ⬎ 0.1). In contrast, substantial inverse correlations with spatial learning ability were obtained for the volume ratios of the MML relative to both the total ML (r ⫽ ⫺0.76; P ⱕ 0.01) and the MF system (r ⫽ ⫺0.65; P ⱕ 0.05). Figure 4 illustrates these relationships, plotting the morphometric values for individual aged subjects against their spatial learning index scores. The data for each of these measures were distributed relatively evenly across a broad range, such that decreases in the MML/ML and MML/MF ratios were associated with increasingly poor spatial learning (i.e., higher index scores). Because declining MML volume alone failed to predict behavioral performance, this pattern of results implies that age-related learning and memory impairment is coupled more tightly to shifts in the balance between principal hippocampal circuits. Recent neuroanatomical studies suggest that different septotemporal levels of the rat hippocampus receive distinct complements of neocortical and subcortical input, following the topographic organization of afferent and efferent connections of the entorhinal cortex (Witter et al., 1989; Burwell and Amaral, 1998a,b; Dolorfo and Amaral, 1998). This arrangement might be expected to confer a corresponding functional specialization, and emerging data tend to support this conjecture. It is of particular relevance to the findings reported here that current behavioral and neurophysiological evidence indicates that the dorsal hippocampus plays a more significant role than ventral levels in normal spatial information processing (Moser et al., 1993; Jung et al., 1994). Guided by these observations, the final component of the present analysis examined potential septotemporal differences in the volumetric effects of aging. Volumes were calculated separately for the septal and temporal halves of the hippocampus, focusing on those parameters that revealed reliable age differences in the
overall analysis (i.e., the total MML volume and the IML/ML, MML/ML, and MML/MF ratios). The total number of histological sections comprising the rostrocaudal extent of the hippocampus varied only slightly across animals [adult mean (S.E.) ⫽ 17.1 (0.26); aged mean (S.E.) ⫽ 18.2 (0.33)], and the nine most anterior sections from each brain were taken to define the septal hippocampus. Estimates for the temporal half were calculated from the remaining sections. Variability in section number across individuals primarily reflected differences in the degree of residual curvature at the temporal extreme of the hippocampus, and the approach that we adopted maximized the comparability of the septotemporal areas selected for regional analysis. Although volume estimates derived for an arbitrarily defined portion of a neuroanatomical structure may not be unbiased by strict stereological standards, they do provide a useful relative metric for evaluating potential regional selectivity in the volumetric effects of hippocampal aging. Table 3 presents results of the regional analysis for the adult and aged groups. A noteworthy feature of these findings is that, overall, the magnitude of age-related volumetric change was greatest in the septal hippocampus. Indeed, although no alterations occurred selectively in the temporal division, several measures revealed statistically reliable effects that were confined entirely to the septal hippocampus. This pattern characterized results for the MML volume and the MML/ML volume ratio: these parameters declined with age in the septal hippocampus (MML volume: F2,14 ⫽ 8.16; P ⱕ 0.005; MML/ML ratio: F2,14 ⫽ 11.0; P ⱕ 0.005), but group differences were not reliable at temporal levels (P ⬎ 0.1). Although the degree of septal volumetric change was numerically greatest among aged-impaired rats, MML and MML/ML values failed to differ between the aged subgroups (P values ⬎ 0.1). By comparison, the percentage of ML volume occupied by the commissural/association pathway (i.e., the IML/ML ratio) increased significantly in both rostral (F2,14 ⫽ 9.4; P ⱕ 0.005) and caudal (F2,14 ⫽ 3.8; P ⱕ 0.05) portions of the aged hippocampus. The outcome of subsequent between-group tests, however, revealed that this change was not equally robust across the longitudinal extent of the structure. Specifically, whereas septal hippocampal IML/ML ratios increased in both aged cohorts relative to adult controls (P values ⱕ 0.005), parallel group comparisons for the temporal hippocampus were not statistically significant (P ⬎ 0.1). Against this background of evidence for relative sparing in the caudal hippocampus, substantial and widely distributed, age-related reductions were noted for the MML/MF ratio, ranging from 17% to 26% in aged rats relative to controls (Table 3). This decline was statistically reliable for both septal (F2,14 ⫽ 18.0; P ⱕ 0.0001) and temporal (F2,14 ⫽ 8.2; P ⱕ 0.005) levels and for each of the aged subgroups in comparison with adults (P values ⱕ 0.005). Despite the lack of regional selectivity, the relationship between de-
HIPPOCAMPAL VOLUME IN AGED RATS
467
TABLE 3. Regional Analysis: Mean (S.E.) Total Volume or Volume Ratio for Selected Hippocampal Areas MML Group Adult Aged unimp Aged imp
MML/ML
IML/ML
MML/MF
Septal*
Temporal
Septal*
Temporal
Septal*
Temporal*
Septal*
Temporal*
1.63 (0.04) 1.44 (0.04) 1.28 (0.11)**
1.31 (0.05) 1.19 (0.13) 1.31 (0.11)
0.461 (0.009) 0.422 (0.009)** 0.412 (0.003)**
0.373 (0.016) 0.353 (0.008) 0.340 (0.010)
0.224 (0.004) 0.248 (0.005)** 0.252 (0.007)**
0.243 (0.009) 0.265 (0.004) 0.268 (0.006)
0.989 (0.032) 0.819 (0.019)** 0.734 (0.038)**
0.711 (0.039) 0.566 (0.003)** 0.572 (0.020)**
*Significant main effect of group, P ⱕ 0.05. For abbreviations, see Table 1. **P ⬍ 0.05 relative to adult group.
Fig. 5. Scatter plots of the MML/MF volume ratios for septal (left) and temporal (right) halves of the hippocampus relative to the spatial learning index scores for aged rats. This morphometric marker of aging correlated only with spatial learning capacity in the septal hippocampus.
creases in the MML/MF ratio and behavioral performance, as documented previously for the hippocampus as a whole (Fig. 4), was evident only in the septal hippocampus. Figure 5 illustrates these results, demonstrating that declining MML/MF values in the rostral half of the hippocampus were correlated inversely with spatial learning index scores among the aged rats (r ⫽ ⫺0.63; P ⫽ 0.05). Although the MML/ML ratio measure derived for the entire hippocampus also was correlated inversely with the index scores, this parameter was not associated reliably with spatial learning when the rostral and caudal hippocampus were considered separately. This pattern of results suggests that volumetric change preferentially targets septal portions of the aged rat hippocampus and that, among these alterations, the status of hippocampaldependent learning is coupled to the balance of principal inputs and outputs of the dentate gyrus.
DISCUSSION Summary of quantitative observations: Aging disrupts the normal organization of principal hippocampal circuitry The present results document that the rat hippocampus undergoes significant volumetric alteration during the course of normal aging. The specific pattern of change is characterized by substantial regional selectivity, and, among the various hippocampal components examined, age-related decreases in total volume were restricted to the zone of termination of the medial entorhinal cortex (i.e., the MML). Although the magnitude of this effect was modest numerically, decreases in MML volume occurred in parallel with other subtle changes that, together, are positioned to alter substantially the normal circuit organization of the hippocampus. Age-related decreases in the percentage of total ML volume comprising the MML, for
example, were accompanied by a corresponding increase in the proportion of the ML occupied by the intrinsic commissural/association pathway (i.e., the IML). The ratio between the MML volume and the MF output of the dentate gyrus also declined significantly, providing additional morphometric evidence for a shift in the balance between afferent cortical drive and intrinsic hippocampal processing. Overall, aging disproportionately affected septal levels of the hippocampus, and, with the exception of decreases in the MML/MF ratio, the temporal hippocampus largely was spared. A central feature of our experimental design is that the status of learning mediated by the hippocampus was documented for all subjects, establishing a basis for relating these findings to the morphometric consequences of aging. By using this approach, the magnitude of spatial learning impairment among the aged individuals selectively correlated with decreases in the MML/ML and MML/MF ratios. These results suggest that morphological alterations associated with cognitive aging involve a reorganization in the relative balance of cortical afferents and intrinsic hippocampal connectivity, providing a potential structural account of age-related changes in the information coding properties of hippocampal neurons.
Comparison with earlier findings Methodological considerations complicate direct comparisons with previous normative findings on hippocampal volume. The estimates reported here for young LongEvans rats, for example, averaged 22% less than corresponding values from 4-month-old Wistar rats (Slomianka et al., 1992) and approximately 34% less than results for young F344 rats (Coleman et al., 1987). Although strain differences may account partly for this pattern of results (West, 1990), technical factors almost certainly play an important role. Specifically, whereas the present study took advantage of a modified Timm method, with which animals are perfused with a sodium-sulphide solution followed by formalin (Sloviter, 1982), previous investigations have relied on traditional procedures, omitting aldehyde fixation (Coleman et al., 1987; Slomianka et al., 1989, 1992). These procedures would be expected to cause substantially different degrees of tissue shrinkage; accordingly, the absolute volume data may not be strictly comparable across reports. Remarkable consistency is apparent, however, in the relative organization of regional hippocampal volumes. The proportion of total ML volume comprising the perforant path termination zone, for example, averaged 74.7% for young rats in the present study compared with values of 77.2% and 72.6% for 4-month-old Wistar rats and F344 rats, respectively [the total volume of the perforant path projection zone was used for this comparison, because previous studies have not quantified OML and MML volumes separately]. Reported volume ratios for the perforant path relative to the MF system are also in close agreement, ranging from 1.4 for young
468
P.R. RAPP ET AL.
animals in our analysis to 1.4 and 1.5 in previous investigations [the hilus and mossy fiber values reported by Slomianka et al. (1989) and Coleman et al. (1987) were summed for this comparison, providing a measure comparable to the MF parameter derived in the present analysis]. Assessed by these measures, the normative data reported here fundamentally are in accord with earlier findings on the volumetric organization of the rat hippocampus. Our results are also broadly consistent with available evidence concerning developmental and age-related alterations in hippocampal volume. Data reported by Coleman et al. (1987) indicate that the rat hippocampus undergoes a variety of volumetric modifications during the first year of life, presumably as a consequence of postnatal neurogenesis and the elaboration of synaptic connectivity. The present analysis revealed a similar pattern, documenting subtle numerical differences between young (3–4 months) and adult (9–10 months) rats for several morphometric measures. Whether or not these effects are linked to subtle developmental changes in learning and memory supported by the hippocampus remains an important topic for systematic study. To compare the effects of advanced age across investigations, it is important to recognize that previous volumetric assessments of Timm-stained material have treated the perforant path projection to the ML as a whole, without distinguishing regions of medial and lateral entorhinal cortex input. By using this approach, Coleman et al. (1987) concluded that the only reliable difference in rats ranging from 12 months to 37 months of age is a change in the ratio of the perforant path volume relative to the MF system. The present analysis replicated this finding (i.e., a significant age effect was observed for the ratio of the summed OML and MML measures relative to MF volume: F2,14 ⫽ 7.1; P ⱕ 0.01) but provided the additional insight that perforant path alterations largely are restricted to the zone of medial entorhinal cortex input, in the MML. Indeed, this regional selectivity may be an important factor accounting for the lack of age-related change in other parameters examined by Coleman et al. Analyzing the perforant path zone as a single, undifferentiated region, for example, could easily obscure a modest volume decline confined to the MML. Consistent with this possibility, and confirming earlier results, total perforant path volume in the present study failed to differ as a function of age (P ⬎ 0.1). Although this account points to a potential consensus concerning the volumetric effects of hippocampal aging, one exception is that the increased IML/ML ratio documented here was not observed in earlier research using F344 rats (Coleman et al., 1987). Experiments examining the areal distribution of kainate receptor binding in the IML, however, have provided independent confirmation that expansion of the commissural/associational zone is a robust feature of aging in Long-Evans rats (Nagahara et al., 1993; Nicolle et al., 1996). Exploiting such strain-dependent effects of aging could provide a valuable strategy for defining the influence of volumetric reorganization on the information processing capacities of the aged hippocampus.
Structural basis of volumetric alteration in the aged hippocampus: Neuron loss vs. synaptic reorganization A wide variety of age-related events might influence hippocampal volume, including neuronal degeneration,
synapse loss, dendritic remodeling, and changes in resident glial populations. Current evidence, however, suggests that many morphometric parameters are substantially more resistant to the effects of aging than previously thought. Across a series of independent investigations using modern stereological methods of quantification, recent findings in rats (Rapp and Gallagher, 1996; Rasmussen et al., 1996), monkeys (Rapp, 1995; Peters et al., 1996), and humans (West, 1993) indicate that total neuron number in the granule cell layer and the pyramidal cells fields of the hippocampus is remarkably well preserved during normal aging. It is important to note that one of these recent studies in rats examined young and aged brains from the same study population that was used in the current investigation (Rapp and Gallagher, 1996). In the monkey, sparing has also been noted for neurons that originate the perforant path projection to the dentate gyrus, in layer II of the entorhinal cortex (Rapp, 1995; Gazzaley et al., 1997). Combined with parallel results from studies of nonneuronal cell types (Rapp et al., in preparation), these findings significantly constrain potential accounts of hippocampal aging, demonstrating that volumetric change occurs against a background of largely preserved neuronal and glial cell number. Several lines of evidence converge on the idea that synaptic reorganization may contribute to volumetric alteration in the aged hippocampus. Quantitative ultrastructural studies have documented significant age-related synapse loss in the ML of the rat dentate gyrus (Geinisman et al., 1992). It is interesting to note that the magnitude of this effect in the termination zone of the medial entorhinal cortex (i.e., the MML) is greatest among aged subjects with documented memory deficits and impaired hippocampal cellular plasticity (de Toledo-Morrell et al., 1988b). Experimental models of morphological plasticity in young rats extend these observations, providing a basis for predicting the influence of synapse loss on the volumetric organization of the aged hippocampus (Steward, 1989). Deafferentation of granule cell dendrites following entorhinal cortex damage leads to a cascade of alterations, including pronounced atrophy in dennervated outer portions of the ML and a more limited expansion of both the commissural/ associational pathway and mossy fiber system. A qualitatively similar pattern was observed in the present analysis, consisting of an age-related decline in ML volume that receives input from the medial entorhinal cortex and a relative increase in IML and MF projections. Of course, the degree of synapse loss that accompanies normal aging is far less than that following direct ablation of the entorhinal cortex, and, as might be expected, age-related volumetric reorganization was correspondingly subtle. Together, these qualitative parallels across experimental settings suggest that age-related synapse loss may contribute to volume decline in the MML and induce a reactive expansion of intrinsic hippocampal circuitry. An interesting implication of this account is that the capacity for morphological plasticity appears at least partly preserved in the aged rat, and the hippocampus retains the ability to mount a compensatory response consequent to the loss of input from the entorhinal cortex. Other findings worth noting in the context of this proposal indicate that the mossy fiber content of the opioid peptide dynorphin increases with age and that this effect is correlated inversely with extracellular glutamate levels (Jiang et al., 1989; Zhang et al., 1991). The findings reported here suggest that structural reorga-
HIPPOCAMPAL VOLUME IN AGED RATS nization, involving a relative expansion of the mossy fiber terminal zone, may contribute to the elevation in dynorphin neuropeptide content observed in the aged rat hippocampus.
Functional implications A central aim of the present experiments was to evaluate the morphometric consequences of aging in relation to the status of learning and memory supported by the hippocampus. Although the volumetric effects documented here were highly selective in terms of regional distribution, spatial learning impairment appears more tightly coupled to a distributed reorganization of hippocampal circuitry than to alterations in any single region. Indeed, across a wide variety of measures, correlations with behavior emerged exclusively for parameters involving the volumetric relationship between multiple hippocampal components. Specifically, although the total volume of the MML declined significantly with age, this change alone failed to predict the magnitude of spatial learning impairment among aged rats. Instead, the age-related loss of MML volume was correlated only with hippocampal learning when considered in relation to the total volume of the dentate gyrus ML (i.e., the MML/ML measure) or the mossy fiber system (i.e., the MML/MF ratio). The regional distribution of these changes is also informative, suggesting that age-related volumetric reorganization associated with cognitive aging predominantly involves a shift in the relative balance between afferent cortical input and intrinsic hippocampal circuitry. By this account, modification in the information coding properties of aged pyramidal cells (Barnes et al., 1997; Shen et al., 1997; Tanila et al., 1997a,b) partly may reflect structural rearrangement at an earlier stage of hippocampal processing, including a decline in the fidelity of cortically derived information combined with a relative enhancement of dentate gyrus output via the mossy fiber system. Volumetric alterations coupled to the magnitude of spatial learning impairment, like the overall effects of aging, preferentially targeted the septal half of the hippocampus. Recent neuroanatomical findings indicate that relatively distinct complements of input are directed to septal and temporal levels of the dentate gyrus, following the topographic organization of afferent and efferent connections of the entorhinal cortex (Burwell and Amaral, 1998a,b; Dolorfo and Amaral, 1998). A key finding with respect to the present results is that the vast majority of neocortical sensory and polymodal associational information is conveyed to septal portions of the dentate gyrus. Other entorhinal inputs are directed disproportionately to temporal levels, including prominent projections from the amygdala and olfactory information relayed from the piriform cortex. The functional significance of this neuroanatomical organization remains to be delineated fully, taking into account the intrinsic, associational networks that connect different levels of both the hippocampus and adjacent cortical areas (Witter et al., 1989). Nonetheless, recent evidence has begun to reveal at least a coarse functional specialization along the septotemporal axis of the hippocampus. Lesion studies, for example, demonstrate that normal spatial learning critically depends on the integrity of the septal hippocampus and that damage restricted to more temporal levels has relatively little effect on this capacity (Moser et al., 1993). Neurophysiological findings in intact rats provide additional support,
469 indicating that the incidence and spatial tuning of locationspecific firing is greatest among pyramidal neurons in the septal hippocampus (Jung et al., 1994). Although these findings converge on a role for the rostral hippocampus in spatial learning, the cognitive contributions of the caudal hippocampus remain largely unspecified. Progress on this issue could inform current research substantially, enabling efforts to relate the effects of aging across multiple levels of behavioral and neurobiological analysis. The findings outlined here, for example, are consistent with the view that age-related circuit reorganization in the septal hippocampus contributes to the altered informationcoding properties of neurons in this region and that these changes influence spatial learning supported by the septal hippocampus. On the basis of our volumetric observations, however, it would be of considerable interest to determine whether cognitive processes mediated by the temporal hippocampus are more resistant to aging and whether the behavioral physiology of neurons in this region is similarly spared. The regional selectivity of volumetric aging in the hippocampus could also serve as a useful guide for further research on age-related alterations in synaptic connectivity. It remains to be determined, for example, whether synapse loss in the aged dentate gyrus displays the same septotemporal gradient of vulnerability documented in the present study by volumetric analysis. If so, then this information could be exploited as a basis for defining the constellation of local factors and distal influences that are responsible for the maintenance or loss of synaptic integrity during aging. Systematic quantification is also needed to establish whether the regional distribution of synapse loss accounts for the pattern of volumetric alterations associated with cognitive aging. Although age-related reductions in synapse density have been reported in many areas of the hippocampus (i.e., the OML, MML, IML, and stratum radiatum of CA1), the relationship between these changes and the status of hippocampal-dependent learning and memory has been examined thoroughly only for the MML (de Toledo-Morrell et al., 1988a). The findings reported here, however, predict that cognitive decline may be coupled more tightly to a distributed pattern of synapse loss, reflecting a shift in the relative balance of afferent cortical input and intrinsic hippocampal circuitry.
ACKNOWLEDGMENTS The expert technical assistance of Dama Morales and Janet Weber is gratefully acknowledged. The authors also thank Patrick Hof and John Morrison for helpful comments on an early version of the paper.
LITERATURE CITED Amaral DG, Witter MP. 1995. Hippocampal formation. In: Paxinos G, editor. The rat nervous system (2nd Edition). San Diego: Academic Press. p 443–493. Barnes CA. 1979. Memory deficits associated with senescence: a neurophysiological and behavioral study in the rat. J Comp Physiol Psychol 93:74–104. Barnes CA. 1998. Spatial cognition and functional alterations of aged rat hippocampus. In: Wang E, Snyder S, editors. Handbook of the Aging Brain. New York: Academic Press. p 51–66. Barnes CA, Treves, A, Rao G, Shen J. 1994. Electrophysiological markers of cognitive aging: region specificity and computational consequences. Semin Neurosci 6:359–367.
470 Barnes CA, Suster MS, Shen J, McNaughton BL. 1997. Multistability of cognitive maps in the hippocampus of old rats. Nature 388:272–275. Baxter MG, Gallagher M. 1996. Neurobiological substrates of behavioral decline: models and data analytic strategies for individual differences in aging. Neurobiol Aging 17:491–495. Burwell RD, Amaral DG. 1998a. Cortical afferents of the perirhinal, postrhinal, and entorhinal cortices of the rat. J Comp Neurol 398:179– 205. Burwell RD, Amaral DG. 1998b. The perirhinal and postrhinal cortices of the rat: interconnectivity and connections with the entorhinal cortex. J Comp Neurol 391:293–321. Cavazos JE, Golarai G, Sutula TP. 1992. Septotemporal variation of the supragranular projection of the mossy fiber pathway in the dentate gyrus of normal and kindled rats. Hippocampus 2:363–372. Coleman PD, Flood DG, West MJ. 1987. Volumes of the components of the hippocampus in the aging F344 rat. J Comp Neurol 266:300–306. de Toledo-Morrell L, Morrell F, Fleming S. 1984. Age-dependent deficits in spatial memory are related to impaired hippocampal kindling. Behav Neurosci 98:902–907. de Toledo-Morrell L, Geinisman Y, Morrell F. 1988a. Age-dependent alterations in hippocampal synaptic plasticity: Relation to memory disorders. Neurobiol Aging 9:581–590. de Toledo-Morrell L, Geinisman Y, Morrell F. 1988b. Individual differences in hippocampal synaptic plasticity as a function of aging: behavioral, electrophysiological and morphological evidence. In: Neural plasticity: a lifespan approach. New York: Alan R. Liss, Inc, p 283–328. Dolorfo CL, Amaral DG. 1998. Entorhinal cortex of the rat: topographic organization of the cells of origin of the perforant path projection to the dentate gyrus. J Comp Neurol 398:25–48. Gaarskjaer FB. 1978. Organization of the mossy fiber system of the rat studied in expended hippocampi. I. Terminal area related to number of granule and pyramidal cells. J Comp Neurol 178:49–72. Gaarskjaer FB. 1985. The development of the dentate area and the hippocampal mossy fiber projection of the rat. J Comp Neurol 241:154– 170. Gallagher M, Burwell R, Burchinal M. 1993. Severity of spatial learning impairment in aging: development of a learning index for performance in the Morris water Maze. Behav Neurosci 107:618–626. Gallagher M, Nagahara, AH, Burwell RD. 1995. Cognition and hippocampal systems in aging: animal models. In: McGaugh JL, Weinberger N, Lynch G, editors. Brain and Memory: Modulation and Mediation of Neuroplasticity. New York: Oxford University Press. p 103–126. Gazzaley AH, Thakker MM, Hof PR, Morrison JH. 1997. Preserved number of entorhinal cortex layer II neurons in aged macaque monkeys. Neurobiol Aging 18:549–553. Geinisman Y, de Toledo-Morrell L, Morrell F, Persina IS, Rossi M. 1992. Age-related loss of axospinous synapses formed by two afferent systems in the rat dentate gyrus as revealed by the unbiased stereological disector technique. Hippocampus 2:437–444. Jiang H-K, Owyang V, Hong J-S, Gallagher M. 1989. Elevated dynorphin in the hippocampal formation of aged rats: relation to cognitive impairment on a spatial learning task. Proc Natl Acad Sci USA 86:2948–2951. Jung MW, Wiener SI, McNaughton BL. 1994. Comparison of spatial firing characteristics of units in dorsal and ventral hippocampus of the rat. J Neurosci 14:7347–7356. Morris RGM, Garrud P, Rawlins JNP, O’Keefe JO. 1982. Place navigation impaired in rats with hippocampal lesions. Nature 297:681–683. Moser E, Moser M, Anderson P. 1993. Spatial learning impairment parallels the magnitude of dorsal hippocampal lesions, but is hardly present following ventral lesions. J Neurosci 13:3916–3925. Nagahara AH, Nicolle MM, Gallagher M. 1993. Alterations in [3H]-kainate receptor binding in the hippocampal formation of aged Long-Evans rats. Hippocampus 3:269–277. Nicolle MM, Bizon JL, Gallagher M. 1996. In vitro autoradiography of ionotropic glutamate receptors in hippocampus and striatum of aged
P.R. RAPP ET AL. Long-Evans rats: relationship to spatial learning. Neuroscience 74:741– 756. Peters A, Rosene DL, Moss MB, Kemper TL, Abraham, CR, Tigges J, Albert MS. 1996. Neurobiological bases of age-related cognitive decline in the rhesus monkey. J Neuropathol Exp Neurol 55:861–874. Rapp PR. 1995. Cognitive neuroscience perspectives on aging in nonhuman primates. In: Nakajima T, Ono T, editors. Emotion, memory and behavior. Tokyo: Scientific Societies Press. p 197–211. Rapp RP, Amaral DG. 1992. Individual differences in the cognitive and neurobiological consequences of normal aging. TINS 15:340–345. Rapp RP, Gallagher M. 1996. Preserved neuron number in the hippocampus of aged rats with spatial learning deficits. Proc Natl Acad Sci USA 93:9926–9930. Rasmussen T, Schliemann T, Sorensen JC, Zimmer J, West MJ. 1996. Memory impaired aged rats: no loss of principal hippocampal and subicular neurons. Neurobiol aging 17:143–147. Shen J, Barnes CA, McNaughton BL, Skaggs WE, Weaver KL. 1997. The effect of aging on experience-dependent plasticity of hippocampal place cells. J Neurosci 17:6769–6782. Simic G, Kostovic I, Winblad B, Bogdanovic N. 1997. Volume and number of neurons of the human hippocampal formation in normal aging and Alzheimer’s disease. J Comp Neurol 379:482–494. Slomianka L, Rungby J, West MJ, Danscher G, Andersen AH. 1989. Dose-dependent bimodal effect of low-level lead exposure on the developing hippocampal region of the rat: a volumetric study. Neurotoxicology 10:177–190. Slomianka L, Rungby J, Edelfors S, Ravn-Jonsen A. 1992. Late postnatal growth in the dentate area of the rat hippocampus compensates for volumetric changes caused by early postnatal toluene exposure. Toxicology 74:203–208. Sloviter RS. 1982. A simplified Timm stain procedure compatible with formaldehyde fixation and routine paraffin embedding of rat brain. Brain Res Bull 8:771–774. Steward O. 1976. Topographic organization of the projections from the entorhinal area to the hippocampal formation of the rat. J Comp Neurol 16:285–314. Steward O. 1989. Reorganization of neuronal connections following trauma: principles and experimental paradigms. J Neurotrauma 6:99–152. Steward O. 1992. Lesion-induced synapse reorganization in the hippocampus of cats: sprouting of entorhinal, commissural/associational, and mossy fiber projections after unilateral entorhinal cortex lesions, with comments on the normal organization of these pathways. Hippocampus 2:247–268. Tanila H, Shapiro M, Gallagher M. Eichenbaum H. 1997a. Brain aging: changes in the nature of information coding by the hippocampus. J Neurosci 17:5155–5166. Tanila H, Sipila P, Shapiro M, Eichenbaum H. 1997b. Brain aging: imparied coding of novel environmental cues. J Neurosci 17:5167–5174. West MJ. 1990. Stereological studies of the hippocampus: a comparison of the hippocampal subdivisions of diverse species including hedgehogs, laboratory rodents, wild mice and men. Progr Brain Res 83:13–36. West MJ. 1993. Regionally specific loss of neurons in the aging human hippocampus. Neurobiol Aging 14:287–293. Witter MP, Groenewegen HJ, Lopes da Silva FH, Lohman AHM. 1989. Functional organization of the extrinsic and intrinsic circuitry of the parahippocampal region. Progr Neurobiol 33:161–253. Zhang W-Q, Mundy WR, Thai L, Hudson PM, Gallagher M, Tilson HA, Hong JS. 1991. Decreased glutamate release correlates with elevated dynorphin content in the hippocampus of aged rats with spatial learning deficits. Hippocampus 1:391–398. Zimmer J, Haug F-MS. 1978. Laminar differentiation of the hippocampus, fascia dentata and subiculum in developing rats, observed with the Timm sulphide silver method. J Comp Neurol 179:581–618.
