THE JOURNAL OF COMPARATIVE NEUROLOGY 311:425-433 (1991)
Effects of Undernutrition During Early Life on Granule Cell Numbers in the Rat Dentate Gyrus K.S. BED1 Department of Anatomy, University of Queensland, St. Lucia, Brisbane, Australia 4072
ABSTRACT Undernutrition during early life is known to affect the morphology of the hippocampal formation. Recent advances in stereologicaltechniques have made it possible to make relatively unbiased estimates of total cell numbers in well-defined brain regions. It was decided to use these methods to determine the effects of different levels of undernutrition during early postnatal life on the granule cells of the rat dentate gyrus. Male hooded Long Evans rats were undernourished between the 16th day of gestation and 30 postnatal days of age to two different levels. The daily food intake of level-1 and level-2 rats represented about 60 and 40%, respectively, of that eaten by well-fed, age-matched controls. Nutritional rehabilitation of the rats was commenced when they had reached 30 days of age by placing them on an ad libitum diet. Groups of control and experimental rats were killed at 70 and 212 days of age. The Cavalieriprinciple was used to determine the granule cell layer volume within the dentate gyrus, and the "disector" method was used to determine numerical densities of these granule cells. These estimates were used to calculate the total numbers of granule cells. There were between 260,000 and 320,000 granule cells within the dentate gyrus of 70-day-oldcontrol and experimental rats. By 212 days of age, well-fed controls had an average of about 834,000 granule cells. The level-1 and level-2 previously undernourished rats had about 515,000 and 595,000 granule cells, respectively. Two-way analysis of variance procedures showed significant main effects of nutrition and age as well as a significant interaction between them. These results provide strong evidence that a period of undernutrition during early life causes a long-term deficit in the total number of dentate gyrus granule cells. Key words: malnutrition,stereology, morphology, hippoeampus
It is well documented that undernutrition during early life can cause deficits and distortions of brain structure (for reviews see Bedi, '84, '87; Bedi and Warren, '88). These deficits are not diffuse throughout the brain. Some regions show very marked effects (e.g., cerebellum), whereas others (e.g., brainstem) appear to remain unaffected. It is possible that these distortions in brain morphology are related to the timing and severity of the period of undernutrition with respect t o the brain growth spurt (Dobbing and Sands, '71). Individual brain regions almost certainly have different time periods when they are most vulnerable to the effects of undernutrition. The rat dentate gyrus has a morphologically distinct and simple histological structure having only three layers. The central layer is a curved cellular layer packed with granule cell neurons. The dendritic trees of these granule cells arborize superficially into the adjacent molecular layer and form synaptic contacts with incoming fibers from other brain regions. The granule cell axons pass through the more deeply situated polymorphic cell layer and form mossy fiber contacts with the dendritic trees of
o 1991 WILEY-LISS, INC.
the pyramidal cells in the hippocampal formation (Isaacson and Pribram, '75; Cowan et al., '80). The polymorphic zone contains local and projection interneurons (Amaral, '78). The rat dentate gyrus is a region of the brain which contains neurons that are known to develop postnatally. Autoradiographic studies have shown that about 80%of the granule cells arise after birth (Angevine, '65; Schlessinger et al., '75). More recent evidence (Bayer et al., '82; Kaplan and Bell, '83, '84) has shown that granule cells of the rat dentate gyrus continue to be produced well into adulthood. Relatively few quantitative morphological studies have examined the neuronal cells in the hippocampi of undernourished animals (e.g., see Lewis et al., '79; Jordan et al., '82; Ahmed et al., '87). There have, however, been several such studies on normal rats at various ages (e.g., see Bayer and Altman, '75; Bondareff and Geinisman, '76; Geinisman et Accepted May 20,1991. Address reprint requests to K.S. Be&, Dept. of Anatomy, University of Queensland, St. Lucia, Brisbane, Australia 4072.
K.S. BED1
426 al., '78; Bayer, '80; West and Andersen, '80; Bayer et al., '82; Boss et al., '85). Of these only some (e.g., West and Andersen, '80; Bayer et al., '82; Boss et al., '85) have attempted to quantify the total number of granule cells in the dentate gyrus. Recent advances (Gundersen, '86; Braendgaard and Gundersen, '86; Gundersen and Jensen, '87; Gundersen et al., '88; Michel and Cruz-Orive, '88) in stereologicaltechniques have now made it possible to make relatively unbiased estimates of the total number of cells in biological structures. It was decided to utilize these methods to estimate the total number of granule cells in the dentate gyrus of rats previously undernourished for the first 30 days of postnatal life. Two levels of undernutrition were used to assess the possible differences due to differing degrees of severity of the period of undernutrition. The levels of undernutrition chosen represented a moderate and relatively severe degree of nutritional deprivation. Before the rats were killed for this quantitative histological study, the spatial learning capacity of the animals was tested using the Morris ('81) water maze. The results of this behavioral study (Bedi, in preparation) will be presented elsewhere.
MATERIALS AND METHODS Animals The rats used in this study were of an outbred hooded Long Evans strain purchased from the Central Breeding Unit of Monash University, Victoria, Australia. Virgin females were housed overnight with males, and the presence of a vaginal plug the next day was used to determine whether mating had occurred. Mated females were housed in separate cages. These dams were divided at random into three groups on the 16th day of pregnancy. The first group was designated as the control group; the second as the level-1 and the third as the level-2 undernourished groups. At birth, all litters were standardized to contain 8 pups with at least 2 , but not more than 4,females in each litter. The control animals and their offspring were allowed to feed ad libitum throughout the experiment. The amount of food given to the level-1and level-2 dams was restricted from the 16th day of pregnancy and throughout the suckling period to that shown in Table 1. All male pups were weaned on postnatal day 19 by separating them from their mothers and housing them, 3 to a cage. Female pups were discarded at this stage. The amount of food given to level-1and level-2 pups was restricted until they had reached 30 days of age. The exact amount of food given per rat per day in each of TABLE 1. Daily Amounts (g) of Food Given to Rats in the Two Undernourished Groups During Early Development Period2 Gestation G16-Birth Post-Natal PNDO-PND6 PND7-PND13 PND14-PND18 Post-Weaning PND19-PND21 PND22-PND25 PND26-PND30
Le~el-1~ 12
Level-2' 8
17
13
22
18 23
27 3.0
2.0
3.5
2.5
4.0
3.0
'Note that during the gestation and lactation periods, the dam received t h e food ration indicated, whereas after weaning the pups received the food ration indicated (see Methods). 'G = Gestation; PND = postnatal day. 3Level-l undernourished rats. 'Level-2 undernourished rats.
the groups is shown in Table 1.The daily food intake of the level-1 and level-2 rats represented about 60 and 40%, respectively, of that eaten by the well-fed control animals at the same stage. Water was available to all rats ad libitum throughout the experiment. Nutritional rehabilitation of the undernourished pups was commenced when they had reached 30 days of age by placing them on an ad libitum diet. Behavioral testing of some of these rats in a Morris ('81)water maze was carried out when they were between 35 and 65 days of age (Bedi, in preparation). These rats were subsequently killed for histological analysis when they were 70 days of age. Other rats were nutritionally rehabilitated until they were 212 days of age when they were also killed for histological examination. Behavioral tests on these rehabilitated rats was carried out in the few weeks just prior to the time they were killed.
Tissue Groups of control, level-1, and level-2 undernourished rats were given an intraperitoneal injection of 0.1 ml of Heparin and 20 minutes later anesthetized with a mixture (0.1 ml/kg of body weight) of 50:50 (v:v) Ketomine (100mgi ml; Apex Laboratories Pty. Ltd., Australia) and Xylazine (20mg/ml; Coopers Animal Health Australia Ltd., Australia). When the rats were under deep anesthesia, they were perfused intracardiacally with a few mls of isotonic saline followed immediately by about 350 mls of a modified Karnovsky fixative (Ito and Karnovsky, '68). This consisted of a mixture of 2.5% glutaraldehyde and 2% paraformaldehyde in a 0.2 M sodium cacodylate buffer (ph 7.2) to which 0.1 glliter of picric acid had been added. The brains of these animals were removed between 2 and 5 hours after being killed. The "forebrain" (brain minus cerebellum, olfactory lobes, and brainstem) was separated from the remainder of the brain by cutting between the anterior and posterior colliculi in the transverse fissure (Zeeman and Innes, '63). The right cerebral hemisphere from each animal was, in turn, sectioned transversely in a plane just rostral to the anterior commissure, which approximately bisects the hippocampal formation. The portions of hippocampus contained in the tissue on either side of this plane were carefully dissected away from the remainder of the cerebral hemispheres. These portions were, in turn, serially sectioned on a vibrotome (Oxford Instruments, England) at a nominal thickness of 100 pm. To ensure that all regions of the hippocampus had the same chance of being sliced, the position of the first cut was randomized along the interval 0 to 100 pm where 0 represents the tangent to the portion of tissue being cut. The first section to be picked up was chosen, by lottery, between the first two sections cut containing hippocampal tissue. Following this, every alternate section was picked up. These sections were mounted on clean glass histology slides and stained with 0.1% toluidine blue. They were used later (see below) to determine the volume of the granule cell layer in the dentate gyrus using the Cavalieri Principle (Michel and Cruz-Orive, '88). The 100 pm-thick vibrotome sections not picked up and stained with toluidine blue were collected and a random sample of about 8 was processed for embedding in resin suitable for electron microscopy. The area of some of these sections was measured both before and after embedding in resin in order to determine the extent of any shrinkage due to the embedding procedures. The average areal shrinkage
GRANULE CELLS IN DENTATE GYRUS
427
of these sections was measured to be 7.2% due to the embedding procedures.
of taking and printing standard photographic film. All counts and measurements were carried out on these pairs of micrographs. Between 4 and 8 pairs of micrographs were Estimation of the volume of granule cell layer used for each animal. The final magnification of the microThe volume of the granule cell layer was estimated using graphs was determined with the aid of a stage graticule and the Cavalieri principle (Gundersen and Jensen, ’87; Michel found to be ~ 2 , 0 6 0 . Each micrograph of the granule cell layer was superimand Cruz-Orive, ’88). This involves estimating the area of the compartment of interest (in this case the granule cell posed with a rectangular test area just slightly smaller than layer) in a systematic sample taken from serial sections the granule cell layer profile in the micrograph (Fig. 1).The through the object. The most efficient way of achieving this “forbidden line” rule (Gundersen, ’77) was used in all is to randomly superimpose a lattice of regularly arranged counting procedures. Granule cell neurons were distintest points over each section sampled and counting the guished from glial cells by using the morphological criteria number of points which fall over the profile of the compart- described by Ling et al. (’73).Counts were made of the total ment. This provides an unbiased estimate of the compart- number (Q+) of granule cell nuclear profiles in each test ment area. This, together with knowledge of the mean area, and the number (Q-) of such profiles appearing in one distance between serial sections, can be used to make an micrograph (test section) but not in the corresponding pair (“look-up’’ section). In order to increase efficiency, each unbiased estimate of the volume of the compartment. In the present case, a magnified image of each of the micrograph was used in turn, first as the test section and sampled 100-pm-thick toluidine blue-stained sections was then as the “look-up’’ section (Gundersen, ’86). The thickprojected, in turn, onto a screen using a Riechert projecting nesses of the sections used for these counting procedures microscope. The magnification of the final image of the were assessed either by a resectioning method described section was measured with the aid of a stage graticule and elsewhere (Bedi, ’87) or by interference microscopy (Goldfound to be ~ 9 6 The . image was superimposed with a stein and Hartman-Goldstein, ’74).Estimates of the numertransparent lattice having a regular array of test points 1.5 ical density of granule cells (Nv,) was obtained using the cms apart. Each point represented an area (a) of 24,414 pm2 formula (Sterio, ’84), in the section plane. The number (P) of test points falling Nv, = &-/a. h, on the profile of the granule cell layer was determined for each animal. This, together with measurements of the mean section thickness (t) of the vibrotome sections (deter- where a = the total area of the test section examined. of the granule cell nuclei The mean projected height (Ex) mined by resectioning; see below), was used to determine the volume (V) of the granule cell layer using the Cavalieri was estimated by the formula (Sterio, 19841, relation: B, = Q’ MQ V = PatN/n, The mean diameter (B,) of the granule cell nuclei was where N = the total number of serial sections through the estimated assuming them to be spherical and using the hippocampus and n = the number of these sections sampled formula (Underwood, ’701, for the point counting procedures. These volume estimates were corrected for the degree of shrinkage which occurred = ITg. during the embedding procedures.
q
Estimation of numerical density of granule cells The lOO-p,m-thick vibrotome sections embedded in resin were resectioned on a Riechert OMU4 ultramicrotome to yield 3 or 4 serial 1-pm- or 2-pm-thick semithin sections through the granule cell layer. These sections were stained with toluidine blue and used to (1) measure the exact thickness of the sections (produced by the vibrotome) following any shrinkage due to the embedding procedures, and ( 2 ) to estimate the numerical density of the granule cells using the disector method (Sterio, ’84). The disector method involves examining particle profiles (in this study granule cell nuclei were used as the counting unit) in two serial sections of known distance (h) apart. The number (Q-) of profiles which appear in one section but not the adjacent section in a given area of tissue is determined. This procedure was carried out by taking pairs of light micrographs (e.g., see Figs. 1, 2) of the granule cell layer from identical regions of a pair of serial semithin sections. The micrographs were taken with the aid of a Vickers M17 light microscope fitted with a Hitachi video camera connected to a monitor and a Mitsubishi video copy processor. The video copy processor yielded micrographs on heatsensitive paper immediately without the time and expense
Granule cell neuron number
Estimates of the total number of granule cell neurons in the dentate gyms were calculated by multiplying estimates of granule cell numerical density with estimates of granule cell layer volume.
Statistics The results for each individual animal were first calculated before group means and standard errors were computed. The results were then analyzed by two-way analysis of variance (ANOVA) procedures with posthoc tests where necessary. All statistical analyses were performed on an IBM-compatible computer using a SAS release 6.03 statistics software (SAS Institute Inc., Cary, NC 27512-8000) package.
RESULTS Body and brain weights Table 2 shows the body and brain weights of the control and experimental rats together with the results of a twoway ANOVA of this data. There were significant main effects of age and nutrition for body, cerebral, and cerebella weights (Table 2). The nutrition x age interaction was not
428
Fig. 1. (a)and (b)show light micrographs through the granule cell layer of two serial 2-pm-thick toluidine blue stained sections to illustrate the principle of the disector method. Granule cell nuclear
K.S. BED1
profiles, which appear in one section but not the adjacent section, are marked with an * (see Materials and Methods). Scale bar = 10 pm.
GRANULE CELLS IN DENTATE GYRUS
429
2a
Fig. 2. (a)and (b)show line tracings of the cell profiles seen in the micrographs contained in Figure 1. The granule cell nuclear profiles,
which appear in one section but not the adjacent section, are shaded with cross hatching. Profiles through blood vessels are shaded black.
K.S. BED1
430
TABLE 4. Mean @EM) Estimates of Dentate Gyms Granule Cell Nuclear Diameters (wm)
TABLE 2. Mean (SEM) Body and Brain Weights of Control and Experimental Animals Age (days) Body weight (9) 70
8
212
2595 (6.8) 385.0 (15.8)
9
Cerebral bempisphere weight (g) 70 8 212
9
Cerebellar weight (mg) 70
8
212 ~~
n
Controls
n’
8
220.8 (7.8) 344.5 (7.5)
8
1.26 (0.02) 1.32 (0.03)
n
8
8
Level2
8
8
8
243
Controls
n
Level-1’
n
8
8
9
11.13 10.43) 8.16 (1.02)
8
212
9.93 (0.70) 6.08 (0.68)
193 (41 223 13)
110) 8
n’
70
1.14 iO.02) 1.27 (0.01)
6
212
Age (days)
192.9 (3.8) 323.3 (12.8)
6
1.21 (0.03) 1.30 (0.04)
8
239 (6) 240 (5)
9
Level-1’
6
(12)
7
Results of two-way ANOVA of above data (# = 2,401 Nutrition (df = 1,40) Age Interaction (df = 2,40)
Le~el-2~
6
9.07 (0.25) 6.72 (0.63)
F5= 4.38 (p < 0.039) F = 33.67 (p < 0.0001) F = 0.67 (ns)
’Number of animals. ‘Level-1 undernourished rats. 3Level-2undernourished rats. ‘Degrees of freedom. ’F-values. ns = not significant.
~
Results of two-way analysis of variance of above data Nutrition (2,411
F-Values Age (1,41!
20.53 (p < 0.0001) 5.03 ip < 0.0111) 8.40 (p < 0.001)
230.42 (0 < 0.0001) 19.25 (p < 0.0001! 10.57 lp < 0.0023)
df4 Body Cerebral hemispheres Cerebellum
Interaction 12,411 0.04
(ns)
TABLE 5. Mean (SEM) of Estimates of the Numerical Density of Granule Cells in Granule Cell Layer ( x 103/mm3) Age (days)
n’
Controls
n
Level-I’
n
70
8
8
9
582.1 (18.2) 947 (81)
8
212
640.4 (31.3) 1,556 (174)
0.93 (ns) 2.65 (ns)
’Number of animals. ‘Level-1 undernourished rats. 3Level-2undernourished rats. ‘Degrees of freedom. ns = not significant.
Results of two-way ANOVA of above data Nutrition (df4 = 2,401 idf = 1,401 Age Interaction ldf = 2,40)
n’
Controls
n
Level-1’
n
70
8
8
9
0.450 i0.039) 0.543 10.024)
8
212
0.471 (0.024) 0.537 i0.026)
Results of two-way ANOVA of above data Nutrition idP = 2,40) Age (df = 1,401 Interaction idf = 2,401
7
6
6
626.9 (32.5) 1,124 (129)
F5 = 4.69 (p < 0.0148) F = 54.51 (p < 0.0001) F = 5.41 (p < 0.0084)
’Number of animals. ‘Level-1 undernourished rats 3Level-2undernourished rats. 4Degrees of freedom. ’F-values.
TABLE 3. Mean (SEMI Volume (mm3)of Granule Cell Layer in Dentate Gyrus Age (days)
7
Le~el-2~
Le~el-2~ 0.541 (0.034) 0.526 10.026)
F5 = 0.30 (ns) F = 5.32 (p < 0.0263)
ranged between 0.45 and 0.54 mm3.At 212 days of age, the granule cell layer volume had increased to about 0.54 mm3 for all groups. These changes were reflected in a significant main effect of age but not nutrition, with no significant age x nutrition interaction (Table 3).
F = 0.85 ins)
’Number of animals. ‘Level-1 undernourished rats. 3Level-2undernourished rats. ‘Degrees of freedom. ’F-values. ns = not significant.
significant at the 5% level for any of these variables. This appears to indicate that, overall, nutritional rehabilitation of the undernourished rats did not lead to any significant “catch-up’’ growth. Closer examination of the data revealed that the level-2 undernourished rats generally had lower body and brain weights than level-1 rats, which in turn had lower weights than well-fed controls. By 212 days of age, nutritionally rehabilitated level-1 rats appeared to have cerebral and cerebella weights, which were not significantly different from age-matched controls indicating that some “catch-up” growth may have occurred in the brains of these particular rats. Nutritionally rehabilitated level-2 rats did not show such “catch-up’’ growth in brain weight (Table 2).
Granule cell layer volumes The mean estimates for the volume of the granule cell layer within the dentate gyrus of the various groups of rats are presented in Table 3. At 70 days of age, the volume
Granule cell nuclear diameters The estimates of granule cell nuclear diameters in the various groups of rats and a two-way ANOVA of this data are presented in Table 4. Once again this showed significant main effects of age and nutrition, but no significant interaction between them. Level-1 undernourished rats appeared to have slightly but significantly larger granule cell nuclei than control and level-2 undernourished animals. Within nutritional groups, rats at 212 days of age had significantly smaller granule cell nuclei than 70-day-old animals (Table 4).
Numerical densities of granule cells Table 5 shows the estimates of the numerical densities of dentate gyrus granule cells together with results of a two-way ANOVA of this data. There were significant main effects of age and nutrition, as well as a significant age x nutrition interaction. Examination of the data revealed that 70-day-old rats had 582,000-640,000 granule cells per mm3 of tissue with no obvious significant differences between nutritional groups. The numerical density of granule cells increased in all nutritional groups by 212 days of age. The increase was most pronounced in well-fed control rats, which had as many as 1.5 million granule cells per mm3 of granular layer tissue (Table 5 ) .
GRANULE CELLS IN DENTATE GYRUS
Granule cell number Table 6 shows the estimates of the total numbers of granule cell neurons within the granule cell layer of the dentate gyrus. There were between 260,000 and 320,000 granule cells within the dentate gyrus of 70-day-oldcontrol and experimental rats. By 212 days of age, well-fed controls had an average of about 834,000 granule cells. The level-1 and level-2 previously undernourished rats had 5 15,000 and 595,000 granule cells, respectively, at 212 days of age. Two-way ANOVA showed significant main effects of nutrition and age as well as a significant interaction between them (Table 6). Posthoc analysis showed that there were no significant differencesbetween control and level-1or level-2 undernourished rats at 70 days of age. However, by 212 days of age both previously undernourished groups were significantly lower than age-matched controls. There was also a significant increase with age in the total number of granule cell neurons for all nutritional groups of rats, the greatest effect being observed in the control animals (Table 6).
DISCUSSION The results of my experiments, using recent advances in stereological techniques, provide strong and unequivocal evidence that a period of undernutrition during early life causes a long-term deficit in the total number of neurons in a well-defined and circumscribed region of the brain. Furthermore, the results confirm that granule cell neurons in the dentate gyrus continue to be produced in adult life (Angevine, ’65; Schlessinger et al., ’75; Bayer et al., ’82; Kaplan and Bell, ’83, ’84). In this study more than half the adult cell number seemed to arise after 70 postnatal days of age. This lengthy period of neurogenesis makes granule cells of the dentate gyrus unusual compared with the majority of other neuronal cell types. It may be that a period of undernutrition imposed during early life results in a permanent change in the duration of the cell cycle time of dentate gyrus granule cell neurons, causing fewer of them to be produced in adult life than would otherwise be the case. There is good evidence that this may occur. Lewis et al. (’79),using tritiated thymidine autoradiography, showed that undernutrition during early life can cause the cell cycle time of the dentate gyrus granule cells to be increased. Such an increase would result in fewer neurons being produced in a given period and could over time result in a substantial deficit in total cell number. This offers a reasonable explanation for the observed deficit in the total number of dentate gyrus granule cells as a result of the undernutrition imposed in the present study. It is uncertain from this study whether or not the period of undernutrition during early life could cause a deficit in the number of neurons whose period of neurogenesis is complete either before or soon after birth. In other words, it remains unknown whether undernutrition can cause the loss (by death) of neurons which have already been generated and have migrated to their normal position within the brain. Preliminary studies designed to answer this question with respect to cerebellar Purkinje cells, which in the rat are generated largely before birth, have been carried out in the author’s laboratory. These studies have used the relatively newly developed “fractionator” technique (Gundersen, ’86). We (Campbell, Bedi and Mayhew, in preparation) could not demonstrate any statistically significant loss
431 TABLE 6. Mean (S.E.M.)of the Estimates of Granule Cells in the Dentate Gyrus of Control and Experimental Rats Age (days)
n1
Controls
n
Level-12
n
70
8
261,400 (20,400) 515,000 (51,0001
8
9
302,600 (25,000) 834,000 (98,0001
8
212
7
6
Le~el-2~ 321,700 (26,8001 595,000 (76,000)
Results of two-way analysis of variance of above data Nutrition (dfc = 2,401 F5 = 3.98 (p < 0.02651 Age (df = 1,40) F = 55.89 (p < 0.0001) Interaction (df = 2,401 F = 4.53 (p < 0.0169) ‘Number of animals. ‘Level-1 undernourished rats. 3Level-2undernourished rats. ‘Degrees of freedom. ‘F-values.
of Purkinje cells as a result of a period of undernutrition during early life. Studies on other neuronal cell types are currently in progress. There are a few previous studies which have attempted to estimate dentate gyrus granule cell numbers in normal rats. These previous studies were, of course, carried out before the modern unbiased stereological methods, mentioned above, became generally available. It is nevertheless worth comparing the results of these studies with those obtained in the present experiments. Seress and Pokorny (’81)estimated that 35-day-oldWistar rats had about 635,000 granule cells in the dentate gyrus, whereas Schlessinger et al. (’75) obtained a value of about 630,000 in 28-day-old Holtzman rats. These values were significantly smaller than the values of about 0.99 to 1.19 million obtained by Gaarskjaer (’78)for adult Wistar rats. In a study on 30- to 365-day-old male Wistar rats, Bayer (’82)estimated values between 890,000 to 1,277,000,respectively, for the dentate gyrus granule cell numbers, indicating about a 43% increase between these ages. Bayer et al. (’82)used 3-km-thick serial sections of methacrylate embedded tissue in order to estimate the total volume of granule cell nuclei present within a given animal as well as the average volume of granule cell nuclei. Their method of sampling, which involved estimating the total profile area of all granule cell nuclear profiles (or part profiles) in “strips” of tissue through randomly selected portions of the granule cell layer, took no account of the “edge effect” (Gundersen, ’77). This may well have caused an overestimate of the total granule cell nuclear volume, and hence estimates of total cell number. The situation is further complicated by the fact that in their study average granule cell nuclear volume decreased from about 664 pm3 at 30 days of age to about 562 pm3at 365 days of age. It is known that, with a given random section, large cells have relatively more chance of being sectioned than small ones. In other words, the magnitude of the error due to the “edge effect” may well have been different for the various ages of rats studied due to the alteration in size of the granule cell nuclei. Boss et al. (’85) estimated dentate gyrus granule cell numbers in 30-, 120-, and 365-day-old female Wistar and Sprague Dawley rats. They found that there was an average of between about 710,000 and 1,000,000 granule cells per dentate gyrus and that there was a significant age and strain difference in these estimates. In this study Boss et al. (’85) used serial 15 km-thick cresyl violet stained sections to estimate both the granule cell layer volume and the
K.S. BED1
432
granule cell numerical density. For these latter estimates, the counting unit used to identify the granule cells was the nucleus. As the ratio between the section thickness and the granule cell nuclei must have been in the order of about 2, this may well have caused problems in the cell counts due to overlapping and overprojection effects (Cruz-Orive, '83). West and Anderson ('80) estimated there to be about 2.17 million granule cells within the dentate gyrus of adult rats. This is considerably higher than all the other published estimates. The exact reasons for this are uncertain, but it seems that West and Andersen ('80) obtained their volumetric measures from brains processed differently from those in which the granule cell numerical densities were determined. Differential shrinkage in the two sets of brains, if not correctly accounted for, could, of course, lead to substantial bias in the final results. It can be seen that most of the previous estimates of the total numbers of dentate gyrus granule cells are reasonably close to the values obtained in the present study for adult animals. However, the estimates for the 70-day-old, wellfed rats in this investigation appear t o be somewhat smaller than that obtained for young rats by previous workers (Bayer, '82; Boss et al., '85). The exact reasons for this discrepancy are uncertain. They may well be due to the fact that, as mentioned above, previous researchers only had methods available which were relatively more biased than the recently developed and improved stereological methods (Gundersen, '86). It is also possible that some of the differences could be due to differing strains of rats used in the varying studies. Strain differences in the number of dentate gyrus granule cells have been reported both in rats (Boss et al., '85) and in mice (Wimer and Wimer, '82). Previous estimates for the numerical density of dentate gyms granule cells per unit volume of granule cell layer have ranged between about 500,000 and 1,000,000 per mm3 (West and Andersen, '80; Bayer et al., '82; Boss et al., '85; Ahmed et al., '87). The estimates for the 70-day old rats used in the present study fall within this range. However, the estimate for the 212-day-old animals is considerably higher at about 1.5 million cell per mm3. It is possible that the discrepancies, both in the estimates of numerical densities and total granule cell numbers, could be due to the relatively biased procedures used in the previous studies. In conclusion, it seems that a period of undernutrition during early life can affect the development of granule cells in the dentate gyrus. This results in the presence of fewer granule cells in the dentate gyri of adult rats previously undernourished for the first 30 postnatal days of life. Available evidence seems to indicate that this deficit in total granule cell number may be due to a permanent change in the length of the cell cycle time of these cells brought about directly as a result of a period of undernutrition during early life.
ACKNOWLEDGMENTS This work was supported by a grant from the National Health and Medical Research Council of Australia. I thank D. Murray for providing research assistance during this project.
LITERATURE CITED Ahmed, M.G.E., K.S. Bedi, M.A. Warren, andM.M. Kame1 (1987) Effectsofa lengthy period of undernutrition from birth and subsequent nutritional
rehabilitation on the synapse-to-granule cell neuron ratio in the rat dentate gyrus. J. Comp. Neurol. 263~146-158. Amaral, D.J. (1978) A Golgi study of cell types in the hilar region of the hippocampus in the rat. J. Comp. Neurol. 182:851-914. Angevine, J.B. (1965) Time of neuron origin in the hippocampal region. Exp. Neurol. (Suppl)2:l-70. Bayer, S.A. (1980) Development of the hippocampal region in the rat. 11. Morphogenesis during embryonic and early postnatal life. J. Comp. Neurol. 190:115-134. Bayer, S.A., and J. Altman (1975) Radiation-induced interference with postnatal hippocampal cytogenesis in rats and its long term effects on the acquisition of neurons and glia. J. Comp. Neurol. 163:l-20. Bayer, S.A., J.S. Yackel, and P.S. Puri (1982) Neurons in the rat dentate gyms granular layer substantially increase duringjuvenile and adult life. Science216:890-892. Bedi, K.S. (1984) Effects of undernutrition on brain morphology: A critical review of methods and results. In D.G. Jones (ed): Current Topics in Research on Synapses. New York: Alan R. Liss, pp. 93-163. Bedi, K.S. (1987) Lasting neuroanatomical changes following undernutrition during early life. In J. Dobbing (ed): Early Nutrition and Later Achievement. London: Academic Press, pp. 1-49. Be&, K.S. (1987) A simple method of measuring the thickness of semi-thin and ultra-thin sections. J. Microsc. 148:107-111. Bedi, K.S., and M.A. Warren (1988) Effects of nutrition on cortical development. In A. Peters and E.G. Jones (eds): Cerebral Cortex. New York: Plenum Press, 441478. Bondareff, W., and Y. Geinisman (1976) Loss of synapses in the dentate gyms of the senescent rat. Am. J. Anat. 145t129-136. Boss, B.D., G.M. Peterson, and W.M. Cowan (1985) On the number of neurons in the dentate gyms of the rat. Brain Res. 338:144-150. Braendgaard, H., and H.J.G. Gundersen (1986) The impact of recent stereological advances on quantitative studies of the nervous system. J. Neurosci. Methods 18:39-78. Cowan, W.M., B.B. Stanfield, and K. Kishi (1980) The development of the dentate gyrus. In A.A. Mascona and A. Monroy (eds): Current Topics in Developmental Biology, Vol. 15. New York: Academic Press. Cruz-Orive, L.M. (1983) Distribution-free estimation of sphere size distributions from slabs showing over-projectionand truncation, with a review of previous methods. J. Microsc. 131:265-290. Dohbing, J. and J. Sands (1971) Vulnerability of developing brain: IX. The effect of nutritional growth retardation on the timing of the brain growth spurt. Biol. Neonate. 19r363-378. Gaarskjaer, F.B. (1978) Organization of the mossy fiber system of the rat studied in extended hippocampi. I. Terminal area related to number of granule and pyramidal cells. J. Comp. Neurol. 178:49-72. Geinisman, Y., W. Bondareff, and J.T. Dodge (19781 Dendritic atrophy in the dentate gyrus ofthe senescent rat. Am. J. Anat. 152:321-330. Goldstein, D.J., and I.J.Hartmann-Goldstein (1974) Accuracy and precision of a scanning and integrating microinterferometer. J. Microsc. 102:143164. Gundersen, H.J.G. (19771 Notes on the estimation of the numerical density of arbitrary particles: The edge effect. J. Microsc. 11It219-223. Gundersen, H.J.G. (1986) Stereology of arbitrary particles. A review of unbiased number and size estimators and the presentation of some new ones, in memory of William R. Thompson, J. Microsc. 143:3-45. Gundersen, H.J.G., and E.B. Jensen (1987) The efficiency of systematic sampling in stereology and its prediction. J. Microsc. 147:229-263. Gundersen, H.J.G., P. Bagger, T.F. Bendsten, S.M. Evans, L. Korbo, N. Marcussen, A. Moller, K. Nielsen, J.R. Nyengaard, B. Pakkenberg, F.B. Sorensen, A. Vesterby, and M.J. West (1988) The new stereological tools: Disector, fractionator, nucleator and point sampled intercepts and their use in pathological research and diagnosis. Acta Path. Microbiol. et Immunol. Scand. 962357-881. Isaacson, R.L., and K.H. Pribram (1975) The Hippocampus, Vol. 1, Structure and Development. New York Plenum Press. Ito, S., and M.J. Karnovsky (1968) Formaldehyde-glutaraldehyde fixatives containing trinitro compounds. J. Cell Biol. 39t168a. Jordan, T.C., H.F. Howells, N. McNaughton, and P.L. Heatlie (1982) Effects of early undernutrition on hippocampal development and function. Res. Exp. Med. (Berl.) 180:201-207. Kaplan, M.S., and D.H. Bell (1983) Neuronal proliferation in the 9-monthold rodeneradioautographic study of granule cells in the hippocampus. Exp. Brain Res. 52:l-5. Kaplan, M.S., and D.H. Bell (1984) Mitotic neuroblasts in the 9-day-oldand 11-month-old rodent hippocampus. J. Neurosci. 4: 1429-1441.
GRANULE CELLS IN DENTATE GYRUS Lewis, P.D., A.J. Patel, and R. Balazs (1979) Effect of undernutrition on cell generation in the rat hippocampus. Brain Res. 168:186-189. Ling, E.A., J. Peterson, A. Privat, S. Mori, and C. Leblond (1973) Investigation of glial cells in semi-thin sections. I. Identification of glial cells in the brains of young rats. J. Comp. Neurol. 149t43-72. Michel, R.P., and L.M. Cruz-Orive (1988) Application of the Cavalieri Principle and vertical sections method to lung: Estimation of volume and pleural surface area. J. Microsc. 150:117-136. Morris, R.G.M. (1981) Spatial localisation cues does not require the presence of local cues. Learn. Motiv. 12239-260. Schlessinger, A.R., W.M. Cowan, and D.I. Gottlieb (1975) An autoradiographic study of the time of origin and pattern of granule cell migration in the dentate g y m s of the rat. J. Comp. Neurol. 159:149-176.
433 Seres, L., and J. Pokorny (1981) Structure of the granular layer of the rat dentate gyrus. A light microscopic and Golgi study. J. Anat. 133t181195. Sterio, D.C. (1984) The unbiased estimation of number and sizes of arbitrary particles using the disector. J. Microsc. 134t127-136. West, M.J., and A.H. Andersen (1980) An allometric study of the area dentata in the rat and mouse. Brain Res. Rev. 2317-348. Wimer, R.E., and C.C. Wimer (1982) A biometrical-genetic analysis of granule cell number in the area dentata of house mice. Develop. Brain Res. 2129-140. Zeeman, W., and J.R.M. Innes (1963) Craigie’s Neuroanatomy of the Rat. New York: Academic Press.
Effects of Undernutrition During Early Life on Granule Cell Numbers in the Rat Dentate Gyrus K.S. BED1 Department of Anatomy, University of Queensland, St. Lucia, Brisbane, Australia 4072
ABSTRACT Undernutrition during early life is known to affect the morphology of the hippocampal formation. Recent advances in stereologicaltechniques have made it possible to make relatively unbiased estimates of total cell numbers in well-defined brain regions. It was decided to use these methods to determine the effects of different levels of undernutrition during early postnatal life on the granule cells of the rat dentate gyrus. Male hooded Long Evans rats were undernourished between the 16th day of gestation and 30 postnatal days of age to two different levels. The daily food intake of level-1 and level-2 rats represented about 60 and 40%, respectively, of that eaten by well-fed, age-matched controls. Nutritional rehabilitation of the rats was commenced when they had reached 30 days of age by placing them on an ad libitum diet. Groups of control and experimental rats were killed at 70 and 212 days of age. The Cavalieriprinciple was used to determine the granule cell layer volume within the dentate gyrus, and the "disector" method was used to determine numerical densities of these granule cells. These estimates were used to calculate the total numbers of granule cells. There were between 260,000 and 320,000 granule cells within the dentate gyrus of 70-day-oldcontrol and experimental rats. By 212 days of age, well-fed controls had an average of about 834,000 granule cells. The level-1 and level-2 previously undernourished rats had about 515,000 and 595,000 granule cells, respectively. Two-way analysis of variance procedures showed significant main effects of nutrition and age as well as a significant interaction between them. These results provide strong evidence that a period of undernutrition during early life causes a long-term deficit in the total number of dentate gyrus granule cells. Key words: malnutrition,stereology, morphology, hippoeampus
It is well documented that undernutrition during early life can cause deficits and distortions of brain structure (for reviews see Bedi, '84, '87; Bedi and Warren, '88). These deficits are not diffuse throughout the brain. Some regions show very marked effects (e.g., cerebellum), whereas others (e.g., brainstem) appear to remain unaffected. It is possible that these distortions in brain morphology are related to the timing and severity of the period of undernutrition with respect t o the brain growth spurt (Dobbing and Sands, '71). Individual brain regions almost certainly have different time periods when they are most vulnerable to the effects of undernutrition. The rat dentate gyrus has a morphologically distinct and simple histological structure having only three layers. The central layer is a curved cellular layer packed with granule cell neurons. The dendritic trees of these granule cells arborize superficially into the adjacent molecular layer and form synaptic contacts with incoming fibers from other brain regions. The granule cell axons pass through the more deeply situated polymorphic cell layer and form mossy fiber contacts with the dendritic trees of
o 1991 WILEY-LISS, INC.
the pyramidal cells in the hippocampal formation (Isaacson and Pribram, '75; Cowan et al., '80). The polymorphic zone contains local and projection interneurons (Amaral, '78). The rat dentate gyrus is a region of the brain which contains neurons that are known to develop postnatally. Autoradiographic studies have shown that about 80%of the granule cells arise after birth (Angevine, '65; Schlessinger et al., '75). More recent evidence (Bayer et al., '82; Kaplan and Bell, '83, '84) has shown that granule cells of the rat dentate gyrus continue to be produced well into adulthood. Relatively few quantitative morphological studies have examined the neuronal cells in the hippocampi of undernourished animals (e.g., see Lewis et al., '79; Jordan et al., '82; Ahmed et al., '87). There have, however, been several such studies on normal rats at various ages (e.g., see Bayer and Altman, '75; Bondareff and Geinisman, '76; Geinisman et Accepted May 20,1991. Address reprint requests to K.S. Be&, Dept. of Anatomy, University of Queensland, St. Lucia, Brisbane, Australia 4072.
K.S. BED1
426 al., '78; Bayer, '80; West and Andersen, '80; Bayer et al., '82; Boss et al., '85). Of these only some (e.g., West and Andersen, '80; Bayer et al., '82; Boss et al., '85) have attempted to quantify the total number of granule cells in the dentate gyrus. Recent advances (Gundersen, '86; Braendgaard and Gundersen, '86; Gundersen and Jensen, '87; Gundersen et al., '88; Michel and Cruz-Orive, '88) in stereologicaltechniques have now made it possible to make relatively unbiased estimates of the total number of cells in biological structures. It was decided to utilize these methods to estimate the total number of granule cells in the dentate gyrus of rats previously undernourished for the first 30 days of postnatal life. Two levels of undernutrition were used to assess the possible differences due to differing degrees of severity of the period of undernutrition. The levels of undernutrition chosen represented a moderate and relatively severe degree of nutritional deprivation. Before the rats were killed for this quantitative histological study, the spatial learning capacity of the animals was tested using the Morris ('81) water maze. The results of this behavioral study (Bedi, in preparation) will be presented elsewhere.
MATERIALS AND METHODS Animals The rats used in this study were of an outbred hooded Long Evans strain purchased from the Central Breeding Unit of Monash University, Victoria, Australia. Virgin females were housed overnight with males, and the presence of a vaginal plug the next day was used to determine whether mating had occurred. Mated females were housed in separate cages. These dams were divided at random into three groups on the 16th day of pregnancy. The first group was designated as the control group; the second as the level-1 and the third as the level-2 undernourished groups. At birth, all litters were standardized to contain 8 pups with at least 2 , but not more than 4,females in each litter. The control animals and their offspring were allowed to feed ad libitum throughout the experiment. The amount of food given to the level-1and level-2 dams was restricted from the 16th day of pregnancy and throughout the suckling period to that shown in Table 1. All male pups were weaned on postnatal day 19 by separating them from their mothers and housing them, 3 to a cage. Female pups were discarded at this stage. The amount of food given to level-1and level-2 pups was restricted until they had reached 30 days of age. The exact amount of food given per rat per day in each of TABLE 1. Daily Amounts (g) of Food Given to Rats in the Two Undernourished Groups During Early Development Period2 Gestation G16-Birth Post-Natal PNDO-PND6 PND7-PND13 PND14-PND18 Post-Weaning PND19-PND21 PND22-PND25 PND26-PND30
Le~el-1~ 12
Level-2' 8
17
13
22
18 23
27 3.0
2.0
3.5
2.5
4.0
3.0
'Note that during the gestation and lactation periods, the dam received t h e food ration indicated, whereas after weaning the pups received the food ration indicated (see Methods). 'G = Gestation; PND = postnatal day. 3Level-l undernourished rats. 'Level-2 undernourished rats.
the groups is shown in Table 1.The daily food intake of the level-1 and level-2 rats represented about 60 and 40%, respectively, of that eaten by the well-fed control animals at the same stage. Water was available to all rats ad libitum throughout the experiment. Nutritional rehabilitation of the undernourished pups was commenced when they had reached 30 days of age by placing them on an ad libitum diet. Behavioral testing of some of these rats in a Morris ('81)water maze was carried out when they were between 35 and 65 days of age (Bedi, in preparation). These rats were subsequently killed for histological analysis when they were 70 days of age. Other rats were nutritionally rehabilitated until they were 212 days of age when they were also killed for histological examination. Behavioral tests on these rehabilitated rats was carried out in the few weeks just prior to the time they were killed.
Tissue Groups of control, level-1, and level-2 undernourished rats were given an intraperitoneal injection of 0.1 ml of Heparin and 20 minutes later anesthetized with a mixture (0.1 ml/kg of body weight) of 50:50 (v:v) Ketomine (100mgi ml; Apex Laboratories Pty. Ltd., Australia) and Xylazine (20mg/ml; Coopers Animal Health Australia Ltd., Australia). When the rats were under deep anesthesia, they were perfused intracardiacally with a few mls of isotonic saline followed immediately by about 350 mls of a modified Karnovsky fixative (Ito and Karnovsky, '68). This consisted of a mixture of 2.5% glutaraldehyde and 2% paraformaldehyde in a 0.2 M sodium cacodylate buffer (ph 7.2) to which 0.1 glliter of picric acid had been added. The brains of these animals were removed between 2 and 5 hours after being killed. The "forebrain" (brain minus cerebellum, olfactory lobes, and brainstem) was separated from the remainder of the brain by cutting between the anterior and posterior colliculi in the transverse fissure (Zeeman and Innes, '63). The right cerebral hemisphere from each animal was, in turn, sectioned transversely in a plane just rostral to the anterior commissure, which approximately bisects the hippocampal formation. The portions of hippocampus contained in the tissue on either side of this plane were carefully dissected away from the remainder of the cerebral hemispheres. These portions were, in turn, serially sectioned on a vibrotome (Oxford Instruments, England) at a nominal thickness of 100 pm. To ensure that all regions of the hippocampus had the same chance of being sliced, the position of the first cut was randomized along the interval 0 to 100 pm where 0 represents the tangent to the portion of tissue being cut. The first section to be picked up was chosen, by lottery, between the first two sections cut containing hippocampal tissue. Following this, every alternate section was picked up. These sections were mounted on clean glass histology slides and stained with 0.1% toluidine blue. They were used later (see below) to determine the volume of the granule cell layer in the dentate gyrus using the Cavalieri Principle (Michel and Cruz-Orive, '88). The 100 pm-thick vibrotome sections not picked up and stained with toluidine blue were collected and a random sample of about 8 was processed for embedding in resin suitable for electron microscopy. The area of some of these sections was measured both before and after embedding in resin in order to determine the extent of any shrinkage due to the embedding procedures. The average areal shrinkage
GRANULE CELLS IN DENTATE GYRUS
427
of these sections was measured to be 7.2% due to the embedding procedures.
of taking and printing standard photographic film. All counts and measurements were carried out on these pairs of micrographs. Between 4 and 8 pairs of micrographs were Estimation of the volume of granule cell layer used for each animal. The final magnification of the microThe volume of the granule cell layer was estimated using graphs was determined with the aid of a stage graticule and the Cavalieri principle (Gundersen and Jensen, ’87; Michel found to be ~ 2 , 0 6 0 . Each micrograph of the granule cell layer was superimand Cruz-Orive, ’88). This involves estimating the area of the compartment of interest (in this case the granule cell posed with a rectangular test area just slightly smaller than layer) in a systematic sample taken from serial sections the granule cell layer profile in the micrograph (Fig. 1).The through the object. The most efficient way of achieving this “forbidden line” rule (Gundersen, ’77) was used in all is to randomly superimpose a lattice of regularly arranged counting procedures. Granule cell neurons were distintest points over each section sampled and counting the guished from glial cells by using the morphological criteria number of points which fall over the profile of the compart- described by Ling et al. (’73).Counts were made of the total ment. This provides an unbiased estimate of the compart- number (Q+) of granule cell nuclear profiles in each test ment area. This, together with knowledge of the mean area, and the number (Q-) of such profiles appearing in one distance between serial sections, can be used to make an micrograph (test section) but not in the corresponding pair (“look-up’’ section). In order to increase efficiency, each unbiased estimate of the volume of the compartment. In the present case, a magnified image of each of the micrograph was used in turn, first as the test section and sampled 100-pm-thick toluidine blue-stained sections was then as the “look-up’’ section (Gundersen, ’86). The thickprojected, in turn, onto a screen using a Riechert projecting nesses of the sections used for these counting procedures microscope. The magnification of the final image of the were assessed either by a resectioning method described section was measured with the aid of a stage graticule and elsewhere (Bedi, ’87) or by interference microscopy (Goldfound to be ~ 9 6 The . image was superimposed with a stein and Hartman-Goldstein, ’74).Estimates of the numertransparent lattice having a regular array of test points 1.5 ical density of granule cells (Nv,) was obtained using the cms apart. Each point represented an area (a) of 24,414 pm2 formula (Sterio, ’84), in the section plane. The number (P) of test points falling Nv, = &-/a. h, on the profile of the granule cell layer was determined for each animal. This, together with measurements of the mean section thickness (t) of the vibrotome sections (deter- where a = the total area of the test section examined. of the granule cell nuclei The mean projected height (Ex) mined by resectioning; see below), was used to determine the volume (V) of the granule cell layer using the Cavalieri was estimated by the formula (Sterio, 19841, relation: B, = Q’ MQ V = PatN/n, The mean diameter (B,) of the granule cell nuclei was where N = the total number of serial sections through the estimated assuming them to be spherical and using the hippocampus and n = the number of these sections sampled formula (Underwood, ’701, for the point counting procedures. These volume estimates were corrected for the degree of shrinkage which occurred = ITg. during the embedding procedures.
q
Estimation of numerical density of granule cells The lOO-p,m-thick vibrotome sections embedded in resin were resectioned on a Riechert OMU4 ultramicrotome to yield 3 or 4 serial 1-pm- or 2-pm-thick semithin sections through the granule cell layer. These sections were stained with toluidine blue and used to (1) measure the exact thickness of the sections (produced by the vibrotome) following any shrinkage due to the embedding procedures, and ( 2 ) to estimate the numerical density of the granule cells using the disector method (Sterio, ’84). The disector method involves examining particle profiles (in this study granule cell nuclei were used as the counting unit) in two serial sections of known distance (h) apart. The number (Q-) of profiles which appear in one section but not the adjacent section in a given area of tissue is determined. This procedure was carried out by taking pairs of light micrographs (e.g., see Figs. 1, 2) of the granule cell layer from identical regions of a pair of serial semithin sections. The micrographs were taken with the aid of a Vickers M17 light microscope fitted with a Hitachi video camera connected to a monitor and a Mitsubishi video copy processor. The video copy processor yielded micrographs on heatsensitive paper immediately without the time and expense
Granule cell neuron number
Estimates of the total number of granule cell neurons in the dentate gyms were calculated by multiplying estimates of granule cell numerical density with estimates of granule cell layer volume.
Statistics The results for each individual animal were first calculated before group means and standard errors were computed. The results were then analyzed by two-way analysis of variance (ANOVA) procedures with posthoc tests where necessary. All statistical analyses were performed on an IBM-compatible computer using a SAS release 6.03 statistics software (SAS Institute Inc., Cary, NC 27512-8000) package.
RESULTS Body and brain weights Table 2 shows the body and brain weights of the control and experimental rats together with the results of a twoway ANOVA of this data. There were significant main effects of age and nutrition for body, cerebral, and cerebella weights (Table 2). The nutrition x age interaction was not
428
Fig. 1. (a)and (b)show light micrographs through the granule cell layer of two serial 2-pm-thick toluidine blue stained sections to illustrate the principle of the disector method. Granule cell nuclear
K.S. BED1
profiles, which appear in one section but not the adjacent section, are marked with an * (see Materials and Methods). Scale bar = 10 pm.
GRANULE CELLS IN DENTATE GYRUS
429
2a
Fig. 2. (a)and (b)show line tracings of the cell profiles seen in the micrographs contained in Figure 1. The granule cell nuclear profiles,
which appear in one section but not the adjacent section, are shaded with cross hatching. Profiles through blood vessels are shaded black.
K.S. BED1
430
TABLE 4. Mean @EM) Estimates of Dentate Gyms Granule Cell Nuclear Diameters (wm)
TABLE 2. Mean (SEM) Body and Brain Weights of Control and Experimental Animals Age (days) Body weight (9) 70
8
212
2595 (6.8) 385.0 (15.8)
9
Cerebral bempisphere weight (g) 70 8 212
9
Cerebellar weight (mg) 70
8
212 ~~
n
Controls
n’
8
220.8 (7.8) 344.5 (7.5)
8
1.26 (0.02) 1.32 (0.03)
n
8
8
Level2
8
8
8
243
Controls
n
Level-1’
n
8
8
9
11.13 10.43) 8.16 (1.02)
8
212
9.93 (0.70) 6.08 (0.68)
193 (41 223 13)
110) 8
n’
70
1.14 iO.02) 1.27 (0.01)
6
212
Age (days)
192.9 (3.8) 323.3 (12.8)
6
1.21 (0.03) 1.30 (0.04)
8
239 (6) 240 (5)
9
Level-1’
6
(12)
7
Results of two-way ANOVA of above data (# = 2,401 Nutrition (df = 1,40) Age Interaction (df = 2,40)
Le~el-2~
6
9.07 (0.25) 6.72 (0.63)
F5= 4.38 (p < 0.039) F = 33.67 (p < 0.0001) F = 0.67 (ns)
’Number of animals. ‘Level-1 undernourished rats. 3Level-2undernourished rats. ‘Degrees of freedom. ’F-values. ns = not significant.
~
Results of two-way analysis of variance of above data Nutrition (2,411
F-Values Age (1,41!
20.53 (p < 0.0001) 5.03 ip < 0.0111) 8.40 (p < 0.001)
230.42 (0 < 0.0001) 19.25 (p < 0.0001! 10.57 lp < 0.0023)
df4 Body Cerebral hemispheres Cerebellum
Interaction 12,411 0.04
(ns)
TABLE 5. Mean (SEM) of Estimates of the Numerical Density of Granule Cells in Granule Cell Layer ( x 103/mm3) Age (days)
n’
Controls
n
Level-I’
n
70
8
8
9
582.1 (18.2) 947 (81)
8
212
640.4 (31.3) 1,556 (174)
0.93 (ns) 2.65 (ns)
’Number of animals. ‘Level-1 undernourished rats. 3Level-2undernourished rats. ‘Degrees of freedom. ns = not significant.
Results of two-way ANOVA of above data Nutrition (df4 = 2,401 idf = 1,401 Age Interaction ldf = 2,40)
n’
Controls
n
Level-1’
n
70
8
8
9
0.450 i0.039) 0.543 10.024)
8
212
0.471 (0.024) 0.537 i0.026)
Results of two-way ANOVA of above data Nutrition idP = 2,40) Age (df = 1,401 Interaction idf = 2,401
7
6
6
626.9 (32.5) 1,124 (129)
F5 = 4.69 (p < 0.0148) F = 54.51 (p < 0.0001) F = 5.41 (p < 0.0084)
’Number of animals. ‘Level-1 undernourished rats 3Level-2undernourished rats. 4Degrees of freedom. ’F-values.
TABLE 3. Mean (SEMI Volume (mm3)of Granule Cell Layer in Dentate Gyrus Age (days)
7
Le~el-2~
Le~el-2~ 0.541 (0.034) 0.526 10.026)
F5 = 0.30 (ns) F = 5.32 (p < 0.0263)
ranged between 0.45 and 0.54 mm3.At 212 days of age, the granule cell layer volume had increased to about 0.54 mm3 for all groups. These changes were reflected in a significant main effect of age but not nutrition, with no significant age x nutrition interaction (Table 3).
F = 0.85 ins)
’Number of animals. ‘Level-1 undernourished rats. 3Level-2undernourished rats. ‘Degrees of freedom. ’F-values. ns = not significant.
significant at the 5% level for any of these variables. This appears to indicate that, overall, nutritional rehabilitation of the undernourished rats did not lead to any significant “catch-up’’ growth. Closer examination of the data revealed that the level-2 undernourished rats generally had lower body and brain weights than level-1 rats, which in turn had lower weights than well-fed controls. By 212 days of age, nutritionally rehabilitated level-1 rats appeared to have cerebral and cerebella weights, which were not significantly different from age-matched controls indicating that some “catch-up” growth may have occurred in the brains of these particular rats. Nutritionally rehabilitated level-2 rats did not show such “catch-up’’ growth in brain weight (Table 2).
Granule cell layer volumes The mean estimates for the volume of the granule cell layer within the dentate gyrus of the various groups of rats are presented in Table 3. At 70 days of age, the volume
Granule cell nuclear diameters The estimates of granule cell nuclear diameters in the various groups of rats and a two-way ANOVA of this data are presented in Table 4. Once again this showed significant main effects of age and nutrition, but no significant interaction between them. Level-1 undernourished rats appeared to have slightly but significantly larger granule cell nuclei than control and level-2 undernourished animals. Within nutritional groups, rats at 212 days of age had significantly smaller granule cell nuclei than 70-day-old animals (Table 4).
Numerical densities of granule cells Table 5 shows the estimates of the numerical densities of dentate gyrus granule cells together with results of a two-way ANOVA of this data. There were significant main effects of age and nutrition, as well as a significant age x nutrition interaction. Examination of the data revealed that 70-day-old rats had 582,000-640,000 granule cells per mm3 of tissue with no obvious significant differences between nutritional groups. The numerical density of granule cells increased in all nutritional groups by 212 days of age. The increase was most pronounced in well-fed control rats, which had as many as 1.5 million granule cells per mm3 of granular layer tissue (Table 5 ) .
GRANULE CELLS IN DENTATE GYRUS
Granule cell number Table 6 shows the estimates of the total numbers of granule cell neurons within the granule cell layer of the dentate gyrus. There were between 260,000 and 320,000 granule cells within the dentate gyrus of 70-day-oldcontrol and experimental rats. By 212 days of age, well-fed controls had an average of about 834,000 granule cells. The level-1 and level-2 previously undernourished rats had 5 15,000 and 595,000 granule cells, respectively, at 212 days of age. Two-way ANOVA showed significant main effects of nutrition and age as well as a significant interaction between them (Table 6). Posthoc analysis showed that there were no significant differencesbetween control and level-1or level-2 undernourished rats at 70 days of age. However, by 212 days of age both previously undernourished groups were significantly lower than age-matched controls. There was also a significant increase with age in the total number of granule cell neurons for all nutritional groups of rats, the greatest effect being observed in the control animals (Table 6).
DISCUSSION The results of my experiments, using recent advances in stereological techniques, provide strong and unequivocal evidence that a period of undernutrition during early life causes a long-term deficit in the total number of neurons in a well-defined and circumscribed region of the brain. Furthermore, the results confirm that granule cell neurons in the dentate gyrus continue to be produced in adult life (Angevine, ’65; Schlessinger et al., ’75; Bayer et al., ’82; Kaplan and Bell, ’83, ’84). In this study more than half the adult cell number seemed to arise after 70 postnatal days of age. This lengthy period of neurogenesis makes granule cells of the dentate gyrus unusual compared with the majority of other neuronal cell types. It may be that a period of undernutrition imposed during early life results in a permanent change in the duration of the cell cycle time of dentate gyrus granule cell neurons, causing fewer of them to be produced in adult life than would otherwise be the case. There is good evidence that this may occur. Lewis et al. (’79),using tritiated thymidine autoradiography, showed that undernutrition during early life can cause the cell cycle time of the dentate gyrus granule cells to be increased. Such an increase would result in fewer neurons being produced in a given period and could over time result in a substantial deficit in total cell number. This offers a reasonable explanation for the observed deficit in the total number of dentate gyrus granule cells as a result of the undernutrition imposed in the present study. It is uncertain from this study whether or not the period of undernutrition during early life could cause a deficit in the number of neurons whose period of neurogenesis is complete either before or soon after birth. In other words, it remains unknown whether undernutrition can cause the loss (by death) of neurons which have already been generated and have migrated to their normal position within the brain. Preliminary studies designed to answer this question with respect to cerebellar Purkinje cells, which in the rat are generated largely before birth, have been carried out in the author’s laboratory. These studies have used the relatively newly developed “fractionator” technique (Gundersen, ’86). We (Campbell, Bedi and Mayhew, in preparation) could not demonstrate any statistically significant loss
431 TABLE 6. Mean (S.E.M.)of the Estimates of Granule Cells in the Dentate Gyrus of Control and Experimental Rats Age (days)
n1
Controls
n
Level-12
n
70
8
261,400 (20,400) 515,000 (51,0001
8
9
302,600 (25,000) 834,000 (98,0001
8
212
7
6
Le~el-2~ 321,700 (26,8001 595,000 (76,000)
Results of two-way analysis of variance of above data Nutrition (dfc = 2,401 F5 = 3.98 (p < 0.02651 Age (df = 1,40) F = 55.89 (p < 0.0001) Interaction (df = 2,401 F = 4.53 (p < 0.0169) ‘Number of animals. ‘Level-1 undernourished rats. 3Level-2undernourished rats. ‘Degrees of freedom. ‘F-values.
of Purkinje cells as a result of a period of undernutrition during early life. Studies on other neuronal cell types are currently in progress. There are a few previous studies which have attempted to estimate dentate gyrus granule cell numbers in normal rats. These previous studies were, of course, carried out before the modern unbiased stereological methods, mentioned above, became generally available. It is nevertheless worth comparing the results of these studies with those obtained in the present experiments. Seress and Pokorny (’81)estimated that 35-day-oldWistar rats had about 635,000 granule cells in the dentate gyrus, whereas Schlessinger et al. (’75) obtained a value of about 630,000 in 28-day-old Holtzman rats. These values were significantly smaller than the values of about 0.99 to 1.19 million obtained by Gaarskjaer (’78)for adult Wistar rats. In a study on 30- to 365-day-old male Wistar rats, Bayer (’82)estimated values between 890,000 to 1,277,000,respectively, for the dentate gyrus granule cell numbers, indicating about a 43% increase between these ages. Bayer et al. (’82)used 3-km-thick serial sections of methacrylate embedded tissue in order to estimate the total volume of granule cell nuclei present within a given animal as well as the average volume of granule cell nuclei. Their method of sampling, which involved estimating the total profile area of all granule cell nuclear profiles (or part profiles) in “strips” of tissue through randomly selected portions of the granule cell layer, took no account of the “edge effect” (Gundersen, ’77). This may well have caused an overestimate of the total granule cell nuclear volume, and hence estimates of total cell number. The situation is further complicated by the fact that in their study average granule cell nuclear volume decreased from about 664 pm3 at 30 days of age to about 562 pm3at 365 days of age. It is known that, with a given random section, large cells have relatively more chance of being sectioned than small ones. In other words, the magnitude of the error due to the “edge effect” may well have been different for the various ages of rats studied due to the alteration in size of the granule cell nuclei. Boss et al. (’85) estimated dentate gyrus granule cell numbers in 30-, 120-, and 365-day-old female Wistar and Sprague Dawley rats. They found that there was an average of between about 710,000 and 1,000,000 granule cells per dentate gyrus and that there was a significant age and strain difference in these estimates. In this study Boss et al. (’85) used serial 15 km-thick cresyl violet stained sections to estimate both the granule cell layer volume and the
K.S. BED1
432
granule cell numerical density. For these latter estimates, the counting unit used to identify the granule cells was the nucleus. As the ratio between the section thickness and the granule cell nuclei must have been in the order of about 2, this may well have caused problems in the cell counts due to overlapping and overprojection effects (Cruz-Orive, '83). West and Anderson ('80) estimated there to be about 2.17 million granule cells within the dentate gyrus of adult rats. This is considerably higher than all the other published estimates. The exact reasons for this are uncertain, but it seems that West and Andersen ('80) obtained their volumetric measures from brains processed differently from those in which the granule cell numerical densities were determined. Differential shrinkage in the two sets of brains, if not correctly accounted for, could, of course, lead to substantial bias in the final results. It can be seen that most of the previous estimates of the total numbers of dentate gyrus granule cells are reasonably close to the values obtained in the present study for adult animals. However, the estimates for the 70-day-old, wellfed rats in this investigation appear t o be somewhat smaller than that obtained for young rats by previous workers (Bayer, '82; Boss et al., '85). The exact reasons for this discrepancy are uncertain. They may well be due to the fact that, as mentioned above, previous researchers only had methods available which were relatively more biased than the recently developed and improved stereological methods (Gundersen, '86). It is also possible that some of the differences could be due to differing strains of rats used in the varying studies. Strain differences in the number of dentate gyrus granule cells have been reported both in rats (Boss et al., '85) and in mice (Wimer and Wimer, '82). Previous estimates for the numerical density of dentate gyms granule cells per unit volume of granule cell layer have ranged between about 500,000 and 1,000,000 per mm3 (West and Andersen, '80; Bayer et al., '82; Boss et al., '85; Ahmed et al., '87). The estimates for the 70-day old rats used in the present study fall within this range. However, the estimate for the 212-day-old animals is considerably higher at about 1.5 million cell per mm3. It is possible that the discrepancies, both in the estimates of numerical densities and total granule cell numbers, could be due to the relatively biased procedures used in the previous studies. In conclusion, it seems that a period of undernutrition during early life can affect the development of granule cells in the dentate gyrus. This results in the presence of fewer granule cells in the dentate gyri of adult rats previously undernourished for the first 30 postnatal days of life. Available evidence seems to indicate that this deficit in total granule cell number may be due to a permanent change in the length of the cell cycle time of these cells brought about directly as a result of a period of undernutrition during early life.
ACKNOWLEDGMENTS This work was supported by a grant from the National Health and Medical Research Council of Australia. I thank D. Murray for providing research assistance during this project.
LITERATURE CITED Ahmed, M.G.E., K.S. Bedi, M.A. Warren, andM.M. Kame1 (1987) Effectsofa lengthy period of undernutrition from birth and subsequent nutritional
rehabilitation on the synapse-to-granule cell neuron ratio in the rat dentate gyrus. J. Comp. Neurol. 263~146-158. Amaral, D.J. (1978) A Golgi study of cell types in the hilar region of the hippocampus in the rat. J. Comp. Neurol. 182:851-914. Angevine, J.B. (1965) Time of neuron origin in the hippocampal region. Exp. Neurol. (Suppl)2:l-70. Bayer, S.A. (1980) Development of the hippocampal region in the rat. 11. Morphogenesis during embryonic and early postnatal life. J. Comp. Neurol. 190:115-134. Bayer, S.A., and J. Altman (1975) Radiation-induced interference with postnatal hippocampal cytogenesis in rats and its long term effects on the acquisition of neurons and glia. J. Comp. Neurol. 163:l-20. Bayer, S.A., J.S. Yackel, and P.S. Puri (1982) Neurons in the rat dentate gyms granular layer substantially increase duringjuvenile and adult life. Science216:890-892. Bedi, K.S. (1984) Effects of undernutrition on brain morphology: A critical review of methods and results. In D.G. Jones (ed): Current Topics in Research on Synapses. New York: Alan R. Liss, pp. 93-163. Bedi, K.S. (1987) Lasting neuroanatomical changes following undernutrition during early life. In J. Dobbing (ed): Early Nutrition and Later Achievement. London: Academic Press, pp. 1-49. Be&, K.S. (1987) A simple method of measuring the thickness of semi-thin and ultra-thin sections. J. Microsc. 148:107-111. Bedi, K.S., and M.A. Warren (1988) Effects of nutrition on cortical development. In A. Peters and E.G. Jones (eds): Cerebral Cortex. New York: Plenum Press, 441478. Bondareff, W., and Y. Geinisman (1976) Loss of synapses in the dentate gyms of the senescent rat. Am. J. Anat. 145t129-136. Boss, B.D., G.M. Peterson, and W.M. Cowan (1985) On the number of neurons in the dentate gyms of the rat. Brain Res. 338:144-150. Braendgaard, H., and H.J.G. Gundersen (1986) The impact of recent stereological advances on quantitative studies of the nervous system. J. Neurosci. Methods 18:39-78. Cowan, W.M., B.B. Stanfield, and K. Kishi (1980) The development of the dentate gyrus. In A.A. Mascona and A. Monroy (eds): Current Topics in Developmental Biology, Vol. 15. New York: Academic Press. Cruz-Orive, L.M. (1983) Distribution-free estimation of sphere size distributions from slabs showing over-projectionand truncation, with a review of previous methods. J. Microsc. 131:265-290. Dohbing, J. and J. Sands (1971) Vulnerability of developing brain: IX. The effect of nutritional growth retardation on the timing of the brain growth spurt. Biol. Neonate. 19r363-378. Gaarskjaer, F.B. (1978) Organization of the mossy fiber system of the rat studied in extended hippocampi. I. Terminal area related to number of granule and pyramidal cells. J. Comp. Neurol. 178:49-72. Geinisman, Y., W. Bondareff, and J.T. Dodge (19781 Dendritic atrophy in the dentate gyrus ofthe senescent rat. Am. J. Anat. 152:321-330. Goldstein, D.J., and I.J.Hartmann-Goldstein (1974) Accuracy and precision of a scanning and integrating microinterferometer. J. Microsc. 102:143164. Gundersen, H.J.G. (19771 Notes on the estimation of the numerical density of arbitrary particles: The edge effect. J. Microsc. 11It219-223. Gundersen, H.J.G. (1986) Stereology of arbitrary particles. A review of unbiased number and size estimators and the presentation of some new ones, in memory of William R. Thompson, J. Microsc. 143:3-45. Gundersen, H.J.G., and E.B. Jensen (1987) The efficiency of systematic sampling in stereology and its prediction. J. Microsc. 147:229-263. Gundersen, H.J.G., P. Bagger, T.F. Bendsten, S.M. Evans, L. Korbo, N. Marcussen, A. Moller, K. Nielsen, J.R. Nyengaard, B. Pakkenberg, F.B. Sorensen, A. Vesterby, and M.J. West (1988) The new stereological tools: Disector, fractionator, nucleator and point sampled intercepts and their use in pathological research and diagnosis. Acta Path. Microbiol. et Immunol. Scand. 962357-881. Isaacson, R.L., and K.H. Pribram (1975) The Hippocampus, Vol. 1, Structure and Development. New York Plenum Press. Ito, S., and M.J. Karnovsky (1968) Formaldehyde-glutaraldehyde fixatives containing trinitro compounds. J. Cell Biol. 39t168a. Jordan, T.C., H.F. Howells, N. McNaughton, and P.L. Heatlie (1982) Effects of early undernutrition on hippocampal development and function. Res. Exp. Med. (Berl.) 180:201-207. Kaplan, M.S., and D.H. Bell (1983) Neuronal proliferation in the 9-monthold rodeneradioautographic study of granule cells in the hippocampus. Exp. Brain Res. 52:l-5. Kaplan, M.S., and D.H. Bell (1984) Mitotic neuroblasts in the 9-day-oldand 11-month-old rodent hippocampus. J. Neurosci. 4: 1429-1441.
GRANULE CELLS IN DENTATE GYRUS Lewis, P.D., A.J. Patel, and R. Balazs (1979) Effect of undernutrition on cell generation in the rat hippocampus. Brain Res. 168:186-189. Ling, E.A., J. Peterson, A. Privat, S. Mori, and C. Leblond (1973) Investigation of glial cells in semi-thin sections. I. Identification of glial cells in the brains of young rats. J. Comp. Neurol. 149t43-72. Michel, R.P., and L.M. Cruz-Orive (1988) Application of the Cavalieri Principle and vertical sections method to lung: Estimation of volume and pleural surface area. J. Microsc. 150:117-136. Morris, R.G.M. (1981) Spatial localisation cues does not require the presence of local cues. Learn. Motiv. 12239-260. Schlessinger, A.R., W.M. Cowan, and D.I. Gottlieb (1975) An autoradiographic study of the time of origin and pattern of granule cell migration in the dentate g y m s of the rat. J. Comp. Neurol. 159:149-176.
433 Seres, L., and J. Pokorny (1981) Structure of the granular layer of the rat dentate gyrus. A light microscopic and Golgi study. J. Anat. 133t181195. Sterio, D.C. (1984) The unbiased estimation of number and sizes of arbitrary particles using the disector. J. Microsc. 134t127-136. West, M.J., and A.H. Andersen (1980) An allometric study of the area dentata in the rat and mouse. Brain Res. Rev. 2317-348. Wimer, R.E., and C.C. Wimer (1982) A biometrical-genetic analysis of granule cell number in the area dentata of house mice. Develop. Brain Res. 2129-140. Zeeman, W., and J.R.M. Innes (1963) Craigie’s Neuroanatomy of the Rat. New York: Academic Press.
