Stereological Estimation of the TotalNumber of Neurons in the Murine

Hippocampus Using the Optical Disector

IMAN ABUSAAD,1,2 DANIEL MACKAY,1 JINGHUA ZHAO,2 PAUL STANFORD,1

DAVID A. COLLIER,1,2 AND IAN P. EVERALL1,2*1Department of Neuropathology, The Institute of Psychiatry, De Crespigny Park,

Denmark Hill, London SE5 8AF, United Kingdom2Department of Psychological Medicine, The Institute of Psychiatry, De Crespigny Park,

Denmark Hill, London SE5 8AF, United Kingdom

ABSTRACTUsing a stereological method, the optical disector, we examined three inbred strains of

mice (NZB/BINJ, DBA/2, and C57BL/6J) for morphological differences in volume, neuronalnumber, and density of the pyramidal cell and dentate gyrus granule cell layers of thehippocampus. We found significant differences in volume and neuronal number for bothregions between the three strains at 9 weeks of age, but only modest differences in neuronaldensity. The left dentate volume was 90% larger in the NZB strain and 70% greater in the DBAstrain (P , 0.0001), and the left pyramidal cell layer was 144% larger in the NZB strain and150% larger in the DBA strain, than in the B6 strain (P , 0.0001). Neuron number in the leftdentate was 81% greater in NZB and 37% greater in DBA (P , 0.001), and in the leftpyramidal cell layer 118% greater in the NZB and 92% greater in the DBA (P , 0.01).Differences in neuronal density of the left dentate were not significant (P 5 0.060, ns). For theleft pyramidal cell layer, neuronal density was 14% greater in B6 and 34% greater in NZBthan the DBA strain (P 5 0.016). No significant differences were found in left-right laterality,or according to sex. We found that strain accounted for 60% of the variance in hippocampalvolume and 44% of neuron number. These differences thus mainly reflect genetic variation inhippocampal volume and may have important implications for brain evolution, behaviour, andhuman diseases where hippocampal degeneration is involved. J. Comp. Neurol. 408:560–566,1999. r 1999 Wiley-Liss, Inc.

Indexing terms: mouse; stereology; Cavalieri, dentate gyrus; pyramidal cell

The brain is by far the most complex mammalian organ,and of the 80,000 or so genes in the human and murinegenomes (Antequera and Bird, 1993), about 30% arebrain-specific (Adams et al., 1992). The majority of thesegenes are likely to be involved in neurodevelopment,maintenance, and plasticity. Deficits in neuronal numberand migration have been implicated in a number of humandiseases that affect cortical development and function,e.g., dyslexia (Galaburda et al., 1983), autism (Baumanand Kemper, 1985), Rett syndrome (Bauman et al., 1995),epilepsy (Meencke and Janz, 1984), schizophrenia (Kovel-man and Scheibel, 1984, 1986; Conrad et al., 1991; Heck-ers et al., 1991; Scheibel and Conrad, 1993; Woodruff andMurray, 1994), and foetal alcohol syndrome (Miller, 1986).In addition there is evidence that hippocampal atrophy isassociated with Cushing’s syndrome, post-traumatic stressdisorder, chronic depressive illness, severity of cognitivedecline inAlzheimer’s disease (reviewed in McEwen, 1997),

and reduced hippocampal volume appears to be correlatedto stress and cognitive decline in the healthy elderly(Lupien et al., 1998).

However, there is relatively little information inmammals on which genes determine brain structure,particularly the volume, orientation, and density ofneuronal subfields. A variety of human diseases havegross structural abnormalities of the brain, such as agen-esis of the corpus callosum and lissencephaly (Dobyns andTruwit, 1995). However, these are, in general, rare disor-

Grant sponsor: Medical Research Council (UK); Grant number: G9507875.*Correspondence to: I.P. Everall, Department of Neuropathology, The

Institute of Psychiatry, De Crespigny Park, Denmark Hill, London SE58AF, United Kingdom. E-mail: i.everall@iop.bpmf.ac.uk

Received 23 December 1997; Revised 9 November 1998; Accepted 21January 1999

THE JOURNAL OF COMPARATIVE NEUROLOGY 408:560–566 (1999)

r 1999 WILEY-LISS, INC.

ders which result from either single gene mutations,chromosomal deletions, or severe environmental in-sults, and are of limited use in providing information onnormal brain development. Likewise, there are a num-ber of mouse mutations which affect brain structure,particularly hippocampal development (Nowakowski,1986b, 1988), but the majority of these are rare, singlegene mutations which severely disrupt neuronal position-ing.

The processes which control neuronal number and migra-tion during different stages of development are of consider-able importance, as they determine neuronal function andhence behaviour (Nowakowski and Rakic, 1981; Nowa-kowski, 1986a; Williams and Herrup, 1988). One approachto understanding these processes is the analysis of varia-tion in neuronal number at maturity. The total neuronnumber in the central nervous system ranges from under300 for small metazoans (Bullock and Horridge, 1965) toabout 85 billion in the human brain (Lange, 1975). Notonly is the absolute number of neurons regulated duringdevelopment, but mechanisms must exist to ensure ad-equate variation in neuron number between membersof a population (Williams and Herrup, 1988). In thisway, evolutionary changes in total or relative neuronnumbers are possible, and the control of this variation willultimately control the rate of brain evolution (Mayer,1963).

Although there has been no systematic investigationof inherited variation in brain structures in normal hu-man populations, evidence from analysis of arthro-pods and annelids (Williams and Herrup, 1988) andthe mouse (Mullen and Herrup, 1979) indicates thatthere is substantial genetically determined variation insize and cell number. In the Purkinje cell degenerationmouse (pcd), a single gene mutation results in the elimina-tion of the entire Purkinje cell population (Mullen, 1977).In the staggerer (sg) mouse, homozygosity for the muta-tion leads to a substantial reduction in the size ofthe cerebellum and has a marked effect on the develop-ment of the Purkinje cell population (Williams and Her-rup, 1988).

Studies by Wimer et al. (1978, 1980) and Wimer andWimer (1982) on the presence and magnitude of geneti-cally associated variability in strains of laboratory miceindicate that there is strong genetic influence on neuronalnumber in the pyramidal cell layer of the hippocampusand granule cell number in the area dentata. Analysis ofretinal ganglion cell number in the mouse with a variety ofstrains also indicates strong genetic control (Rice et al.,1995). There is also evidence of environmental influence onneuron number, because adult mice living in enrichedenvironments develop more hippocampal neurons thanthose living under standard laboratory conditions (Kemper-man et al., 1997a,b).

The mechanisms of control of neuron number mayinvolve both regulation of cell proliferation or cell death.Thus, in the pcd mouse mutant, the gene required toproduce and/or maintain Purkinje cells is absent, whereasin the sg chimera mouse, Purkinje cells are not directlyprogrammed to die, but more granule cells are producedthan can be supported by healthy Purkinje cells. Theresult is increased cell death, leading to a linear relation-ship between Purkinje and granule cell numbers (Herrupand Sunter, 1987).

In the present study, we have examined the relation-ship between neuron number and subfield volume inthe dentate gyrus and pyramidal cell layers of thehippocampus in three strains of common laboratorymice, C57BL/6J (B6), DBA/2 (DBA), and NZB/BINJ(NZB) to determine if there are genetic differences deter-mining these parameters. The use of recombinant or F2hybrid crosses between these strains should ultimatelyallow the characterisation of the gene(s) involved in thedetermination of the neuronal number and hippocampalsize.

MATERIALS AND METHODS

Tissue preparation

Three inbred strains of mice, half of each sex (C57BL/6J[B6], n 5 6; NZB/BINJ [NZB], n 5 8; and DBA/2 [DBA],n 5 8) were obtained from a laboratory supplier (Harlan-Olac Ltd., UK). Mice were kept in standard laboratoryconditions with four to six mice to a cage. At age 9 weeks,the mice were killed by cervical dislocation and the brainexcised. This method was approved as a Schedule 1procedure according to current Home Office regulations onanimal welfare, by the Chairman of the Animal Welfareand Research Advisory Committee at the Institute ofPsychiatry. The brains were placed individually in 10%neutral buffered formalin, pH 7.0 (100 volumes) and left atroom temperature for 6 weeks to fix the tissue. Brainswere then dehydrated in graded ethanol, cleared in xylene,and embedded in paraffin wax in an embedding centre on a24-hour cycle. Each brain was then oriented longitudinallysuch that the dorsal (posterior) surface was adjacent to theblock face. Sections were cut exhaustively in the coronalplane from one end to the other at thickness of 16 µm witha LEITZ base sledge microtome by using disposable steelknives. The sections were floated on a warm water bath,mounted onto silane-coated glass slides, and immediatelydried at 40°C prior to staining. Following random selectionof the first slide, every 10th section in the series wascollected for sampling and stained in Cresyl violet for 6minutes.

Neuron-containing layers

The highly laminated structures of the hippocampalformation provide an ideal location for the assessment ofthe effects of genetic variation on the development of thecentral nervous system. Figure 1 illustrates the hippocam-pal formation of a B6 mouse in coronal section, using aslide from the present study. The coronal plane was chosenarbitrarily and this was used for all the brains studied.Pache et al. (1993) stated that although the orientation isof theoretical importance, it is unlikely to be of practicalimportance in affecting the subsequent stereological analy-sis.

The major subdivisions of the hippocampal formationare the subiculum, the hippocampus, and the dentategyrus (Angevine, 1965). The hippocampus is further subdi-vided into areas CA1, CA2, and CA3. Moving from ven-tricle to pia, the laminae of the hippocampus are thealveus, the stratum oriens, the pyramidal cell layer, thestratum radiatum, and the stratum lacunosum-molecu-lare. In area CA3, there is the additional stratum lucidum,which contains the mossy fibres from the granule cells ofthe dentate gyrus. The main cellular layer of the dentate

STEREOLOGY OF THE MURINE HIPPOCAMPUS 561

gyrus is the horseshoe-shaped granule cell layer. Themolecular layer is external to it, and the hilus (or areaCA4) is the area internal to it. In the present study, weexamined the pyramidal cell layer, in CA1, CA2, and CA3,

and the granule cell layer in the dentate gyrus. The layerscontaining the cell bodies of the neurons of the differenthippocampal components were defined consistently at alllevels of each section series and in all animals.

Fig. 1. Morphology of the C57BL/6J mouse hippocampus. A: Directphotograph of a 16-µM section through the hippocampus of theC57BL/6J mouse in coronal section, as used in the present study.Magnification 2.53. Scale bar 5 3.8 mm. B: A schematic diagram ofthe hippocampus showing the relevant neuroanatomical subdivisions.

CC, corpus callosum; Cing, cingulum; FD, fascia dentata; FH, fimbrihippocampi; G, granule cell layer; H, hilus of the dentate gyrus; Hbl,nucleus lateralis habenulae; Hbm, nucleus medialis habenulae; Hip,hippocampus; L, nucleus lateralis thalami; P, pyramidal cell layer;VIII, ventricus tertius; V lat, ventriculus lateralis.

562 I. ABUSAAD ET AL.

Stereology

Two estimations were performed in all three mice strains:Cavalieri estimator of reference volume, and optical disec-tor estimator of neuronal numerical density. Microscopywas performed on an Olympus BH2 microscope, modifiedfor stereology by attachment to the stage of HeidehainVRZ403 electronic length gauge measuring vertical move-ment through the section. Microscopic field images werecaptured by a JVC TK-1280E colour camera and analysedby stereology software (Kinetic Imaging Ltd., UK) on aDOS-based PC.

The individual volumes of dentate and pyramidal celllayers were determined by using the Cavalieri method. Byusing a 400 cross-hatch point counting grid, a ‘‘hit’’ wasrecorded if the object of interest was on any part of thepoint. Stereology software superimposed the grid on theimage on the monitor and this was applied repeatedly toall sections upon which the whole hippocampus wasvisualized with a low-power objective lens (32.5). Sectionswere systematically sampled with the first section beingpicked randomly from the first ten available and thenevery tenth section picked randomly afterwards. The areaaround each point was calibrated using a stage microme-ter. By knowing the distance (t) between sections andmultiplying that by the area per point (ap), each pointserved as a volume probe. The total volume (Vref) of thehippocampus was determined by counting the number ofpoints (Pi) overlaying the hippocampus in the sampledsections and multiplying that sum (SPi) by the volumeassociated with each point according to the formula:

Vref 5 t.ap. oi51

n

Pi

where: t 5 0.16 mm, equal to average distance from onesection to the next. Calculated by the thickness of thesection (0.016 mm) multiplied by the number of slidesbetween sections (10). ap 5 0.0096 mm2, equal to the areaassociated with one point in the grid. Calculated from thearea of the grid (3.84 mm2) divided by the number of gridpoints (400). Pi 5 number of points hitting the section ofthe hippocampus; n 5 number of equidistant sectionsused.

The number of points overlying the hippocampus in eachof the 1 to ith sampled sections was used to calculate thecoefficient of error (CE) of the volume estimates, as de-scribed by Gundersen et al. (1988a,b). A CE of 5% or less isacceptable, and this level was obtained for all cases in thisstudy.

The method used for stereological estimations of neuro-nal numerical density was by the stereological probeOptical Disector (Sterio, 1984). This probe is specificallysensitive to estimating the numerical density and isunbiased, with all neurons being sampled regardless oftheir shape or size. It consists of a planar counting frame ofknown area coupled to an identical counting frame on aparallel plane at a known depth within the tissue. In thisstudy, the depth was 5 µm, taken from within the 16-µm-thick tissue slice. Use of an objective lens of 3100 magnifi-cation with a high numerical aperture (NA: 1.4) allowsoptical focusing through the section, and thus the countingframe points are optically defined within the section. Thedisector was randomly applied on all the sampled sectionsthroughout the length of the hippocampus. Neurons were

only counted if they fell within the three-dimensionalmeasuring volume of the disector and as long as they werenot overlapping the two forbidden lines on the grid (the leftand lower sides). Neurons were identified by a clearnuclear profile containing a nucleolus and Nissl substancein the cell body, as the focus went through the section fromone grid to the other. In granule cells, a nucleolus was notalways visible, but potential double counting was avoidedby always counting sampled cells while moving in oneconsistent direction from the upper to lower optical planein the section. On sampling the next disector field, themicroscope stage was returned to the zero level and theprocess of moving down 5 µm repeated. The neuronaldensity was defined as follows:

Nv 5 (SQ/SPi)/Vdis

where Q is the number of neurons counted within thesampling volume (Vdis) which is determined by multiply-ing the area of the sampling frame by the distance betweenthe two optical grids. Pi is number of disectors applied. Inaddition to providing a useful parameter in itself, the Nvcan be combined with volume (Vref) to estimate the totalnumber of neurons present (Nabs) by the equation:

Nabs 5 (Nn) (Vref).

To estimate the genetic contribution to hippocampalvolume, neuronal density and number, analysis of vari-ance components were calculated using analysis of vari-ance (ANOVA), as implemented in SAS (Statistical Analy-sis System) GLM (General Linear Model) procedure. Totalvolume, neuronal density, and number were used asdependent variables and adjusted by the independentvariables strain, sex, and side. Total volume, neuronaldensity, and number were standardized before performingANOVA so that the variance components of the factorscould be easily estimated. The genetic contribution to totalvolume, neuronal density, and number was defined as thepercentage contribution of the effect of mouse strain to thetotal variance. The type III F-test was used to examine thestatistical significance of the independent variables.

RESULTS

The estimates by left and right side for hippocampalvolume are shown in Table 1, neuronal number in Table 2,and neuronal density in Table 3. The most striking differ-ence was that the B6 strain had the smallest dentate andpyramidal cell layer volumes, whereas the NZB and DBAstrains were similar (Table 1). The area of the left dentatewas 90% larger in the NZB strain and 70% greater in the

TABLE 1. Hippocampal Subfield Volumes in the B6, NZB, and DBA Mice1

Strain

Estimated volumes

L dentate* R dentate L pyramidal* R pyramidal

B62 150.5 (17.8)3 116.0 (8.5) 273.9 (24.4) 272.9 (16.7)NZB 285.7 (13.3) 290.5 (9.5) 670.5 (54.7) 721.9 (48.2)DBA 256.1 (17.2) 268.4 (33.9) 687.0 (59.0) 636.3 (83.6)

1Left (L)- and right (R)-sided strain means for estimated volume (3106 µm3) of thegranule cell layer and pyramidal cell layer in the mouse hippocampus by side, using theCavalieri method.2B6, C57BL/6J; NZB, NZB/BINJ; DBA, DBA/2.3Standard errors are shown in parentheses.*P , 0.0001.

STEREOLOGY OF THE MURINE HIPPOCAMPUS 563

DBA strain compared to B6 strain (P , 0.0001), whereasthe left pyramidal cell layer was 144% larger in the NZBstrain and 150% larger in the DBA strain compared to theB6 strain (P , 0.0001). Comparison of left and right sidesdid not reveal significant differences for the DBA and NZBmice for either the dentate or pyramidal cell layer. How-ever, although the volumes of the left and right pyramidalcell layer were almost identical for the B6 strain, thevolume of the right dentate was measured as 29% smaller.However, when we analyzed the B6 strain separately, thisdifference did not reach statistical significance (t-test,F 5 0.11).

Neuron number in the left dentate was 81% greater inNZB and 37% greater in DBA compared to the B6 strain(P 5 0.0006; Table 2). Neuron number in the left pyrami-dal cell layer was ranked in the same order, with NZBhaving 118% and DBA 92% more neurons than the B6strain (P 5 0.001). There were no significant differencesbetween the left and right sides, with all differences beingless than 8%, with the exception of the dentate of the B6strain, which had 30% fewer neurons on the left side. Thiswas a consequence of the reduced volume of the rightdentate we measured in the B6 strain. However, left-rightdifferences in neuron number did not reach statisticalsignificance in the B6 strain (t-test, F 5 0.16).

There were only modest differences in dentate andpyramidal cell neuronal density, as shown in Table 3. Forthe left dentate, the B6 strain had 9% greater neuronaldensity than the NZB strain and 29% greater than theDBA strain (P 5 0.060, ns). Analysis of the right sideproduced similar values (4% and 23%, respectively) whichwere likewise not significant (P 5 0.062), and there wasalso no significant difference between the left and rightsides. For the left pyramidal cell layer, the B6 strain had14% greater neuronal density than the NZB strain and34% greater neuronal density than the DBA strain

(P 5 0.016). Analysis of the right side produced similarvalues (18% and 37%, respectively; P 5 0.008) with nosignificant difference between the sides.

We next estimated the extent of genetic control ofneuron number and cell density in the hippocampus usinga variance component estimation procedure. There was nosignificant effect of sex for any measure, but strain ac-counted for 60% of the variance in hippocampal volume(F 5 32.52; P , 0.0001), 44% of neuron number (F 5 16.09;P , 0.0001), and 32% of neuronal density (F 5 9.15;P , 0.0006) with the remainder accounted for by variationbetween animals of the same strain.

DISCUSSION

One of the objectives for performing this study was tomeasure strain-specific extremes of neuron number anddensity in each of the subdivisions of the hippocampus ininbred mice, so that the genes involved in controllingneuron number and subfield volume can be mapped usingF2 crosses or recombinant inbred strains. We found signifi-cant differences in hippocampal volume and neuronalnumber between the three strains at 9 weeks of age, withthe NZB and DBA strain being similar and the B6 mice themost divergent, having the smallest dentate gyrus andpyramidal cells layers and the fewest neurons. Thesedifferences were similar to those found by Wimer et al.(1980), who found 35% more pyramidal cells and 48% moretotal volume in the NZB strain in comparison to theC57BL/10J strain. However, the C57BL strain examinedby Wimer et al. (1980; 10J) was not the same as used inthis study (6J) and they used different methodology tocount hippocampal volume and neuronal number. In amore recent study, Kemperman et al. (1997b) examinedneurogenesis and neuronal density, number, and dentatevolume in four inbred strains of mice. Quantitatively, theyreported about twofold fewer granule cells in the dentategyrus of the C57BL/6 (B6) strain (0.24 3 106) than in thepresent study (0.49 3 106 cells) for mice of the same age.Also, the volume of the dentate gyrus granule cell layerestimated by Kemperman et al. (1997b) for C57BL/6(2.63 3 108 µm3) was 75% larger than the volume estimateof the present study (1.5 3 108 µm3). Consequently, ourestimates of neuronal density differ because this is fac-tored from density multiplied by volume. The difference inneuronal number might be accounted for by variation instereological methodology between the two studies.

Using a variance component estimation procedure, wefound that strain accounted for 60% of the variance inhippocampal volume and 44% in neuron number, andalthough neuronal density was relatively constant (Table3), comparison between the three strains still indicated asignificant but modest strain dependent effect (32% ofvariance). The most likely explanation for these strain-dependent differences is direct genetic control of neuronnumber, although indirect genetic effects are possible andindeed environmental effects cannot be formally elimi-nated without embryo transfer or cross-fostering experi-ments. For example, it is possible that there are differ-ences between the mouse strains in gestation or rearingwhich cause neuronal differences. This would still be agenetic effect, although in an indirect or epigenetic sense,because any gene(s) influencing gestation or rearing mightbe maternal, and not transmitted to the offspring.

TABLE 2. Hippocampal Subfield Neuronal Number in the B6, NZB, andDBA Mice1

Strain

Total neuronal number3

Leftdentate**

Rightdentate

Leftpyramidal*

Rightpyramidal

B62 0.493 (0.044)4 0.379 (0.060) 0.427 (0.036) 0.448 (0.039)NZB 0.892 (0.053) 0.936 (0.059) 0.933 (0.087) 1.004 (0.082)DBA 0.675 (0.066) 0.721 (0.122) 0.818 (0.135) 0.755 (0.140)

1Strain means for total neuronal number (3106) in the dentate gyrus and pyramidal celllayer by side(L, left; R, right).2B6, C57BL/6J; NZB, NZB/BINJ; DBA, DBA/2.3Total neuronal number was estimated from the product of the subfield volumeestimated by the Cavalieri method and the neuronal density estimated by the opticaldisector.4Standard errors are shown in parentheses.**P , 0.001; *P , 0.01.

TABLE 3. Hippocampal Subfield Numerical Neuronal Density in the B6,NZB, and DBA Mice1

Strain

Numerical density3

Left dentateNS Right dentate Left pyramidal* Right pyramidal

B62 3.40 (0.36)4 3.19 (0.30) 1.58 (0.10) 1.63 (0.07)NZB 3.12 (0.11) 3.21 (0.13) 1.38 (0.03) 1.38 (0.03)DBA 2.64 (0.17) 2.61 (0.17) 1.18 (0.11) 1.19 (0.12)

1Strain means for neuronal density (3106 mm23) in the dentate and pyramidal celllayer, by side.2B6, C57BL/6J; NZB, NZB/BINJ; DBA, DBA/2.3Numerical density was estimated by the optical disector.4Standard errors are shown in parentheses.NSP 5 0.06; *P 5 0.016.

564 I. ABUSAAD ET AL.

In general, neuronal number changes proportionally tohippocampal volume, as illustrated by the relatively stableneuronal density between the strains. The modest varia-tion in neuronal density that Kemperman et al. (1997b)found is consistent with our data, supporting the hypoth-esis that hippocampal neuronal density is not substan-tially polymorphic in mice, although there is still a signifi-cant strain dependent component.

This study of neuronal number indicates that the pyra-midal and the granule cell layers of the dentate gyrus aregenetically differentiated in the three inbred strains of themice examined. This may prove to be a common character-istic of neuronal populations exhibiting separable geneticdetermination of number. Neuron number and hippocam-pal volume might be transmitted through one or moreautosomal genes, because we found no evidence for sexualdifferentiation. This might involve a single gene for eachsubfield, or multiple genes that have additive or multiplica-tive epigenetic interaction. Studies of genetically segregat-ing stocks selected for neuron number would be required todetermine the number of those genes. However, becausestereology is an invasive process and monozygous off-spring appear to be extremely rarely produced in mice(McLaren et al., 1995), selective breeding is not obviouslyfeasible.

These findings are in the same areas of the hippocampusaffected in the dreher (dr) mutant mouse. In dr/dr mice,both neuronal cell proliferation and migration of thedentate gyrus and pyramidal cell layers are affected, withportions of the granule cell layer and infrapyramidallayers of the dentate gyrus absent. These abnormalitiesappear to be secondary to disruptions of the radial glialfibre system and neuronal migration (Sekiguchi et al.,1994). However, in the C57BL/6J, NZB, and DBA/2 mice,the hippocampus appears histologically normal, with noindications of abnormal positioning of neurons, indicatingthat the strain differences in volume and neuronal numberwe see result from normal polymorphic variation in brainstructure. There is also evidence that environmental fac-tors play a role in neuroanatomical plasticity, becauseadult mice living in an enriched environment develop moredentate gyrus hippocampal neurons than those livingunder standard laboratory conditions (Kempermann et al.,1997a,b). However, the increase in neurons and dentategyrus volume was smaller, at 15%, than the between-strain differences we observe.

The discovery of the gene(s) that control this differenceare likely to be of fundamental importance in neurodevel-opment and evolution of brain function. Not only is theabsolute number of neurons regulated during develop-ment, but also mechanisms must exist to ensure adequatevariation in neuron number between members of a popula-tion (Williams and Herrup, 1988). In this way, evolution-ary changes in total or relative neuron numbers arepossible, and the control of this variation will ultimatelycontrol the rate of brain evolution (Mayer, 1963). The useof genetic crosses, such as recombinant inbred strains orF2 crosses between mice will allow the mapping of genescontrolling neuron number. Neuron number may also haveimportant consequences for behaviour, because for ex-ample, water-maze learning and the effects of cholinergicdrugs have been correlated with high and low hippocampalpyramidal cell counts in NZB/BINJ and C57BL/10J mice(Symons et al., 1988).

In addition to providing basic information on neurodevel-opment and behaviour, studies of genetically controlledvariation in neuron number may provide an importantinsight into human disease processes. The role of normalvariation in brain structure in human disease is not widelyexplored, but neurodevelopmental abnormalities in areassuch as the entorhinal cortex, hippocampus, and temporallobes are probable in a variety of syndromes with geneticaetiology, including schizophrenia (Conrad et al., 1991;Heckers et al., 1991; Scheibel and Conrad, 1993; Woodruffand Murray, 1994) and temporal lobe epilepsy (Scheibel,1991). Variation in hippocampal volume and neuron den-sity may also be involved in neurodegenerative diseasessuch as Alzheimer’s, where hippocampal cell loss is associ-ated with cognitive decline (deLeon et al., 1997). This effectmay also be linked to the neurobiological effects of stress,because elevated cortisol has been correlated with reduc-tions in hippocampal volume and spatial memory loss innormal elderly humans (Lupien et al., 1998). The charac-terisation of genetic loci controlling hippocampal volumeand neuron number may provide important candidategenes for a variety of neuropsychiatric disorders.

ACKNOWLEDGMENTS

We are grateful to the Medical Research Council forfunding this study (grant number G9507875 to DAC), andto Professor R. Plomin, SGDP Centre, the Institute ofPsychiatry for facilitating this work. We thank Dr. T. Liand Dr. M.-W. Lin, Institute of Psychiatry, and Dr. P.Scriven, Guy’s Hospital, London, for valuable help.

LITERATURE CITED

Adams MD, Dubnick M, Kerlavage AR, Moreno R, Kelley JM, UtterbackTR, Nagle JW, Fields C, Venter JC. 1992. Sequence identification of2,375 human brain genes. Nature 355:632–634.

Angevine JB Jr. 1965. Time of neuron origin in the hippocampal region: anautoradiographic study in the mouse. Exp Neurol Suppl 2:1–71.

Antequera F, Bird A. 1993. Number of CpG islands and genes in human andmouse. Proc Natl Acad Sci USA 90:11995–11999.

Bauman M, Kemper TL. 1985. Histoanatomic observations of the brain inearly infantile autism. Neurology 35:866–874.

Bauman ML, Kemper TL, Arin DM. 1995. Pervasive neuroanatomicabnormalities of the brain in three cases of Rett’s syndrome. Neurology45:1581–1586.

Bullock TH, Horridge GA. 1965. Structure and function in the nervoussystem of invertebrates. San Francisco: Freeman.

Conrad AJ, Abebe T, Austin R, Forsythe S, Scheibel AB. 1991. Hippocampalpyramidal cell disarray in schizophrenia as a bilateral phenomenon.Arch Gen Psychiatry 48:413–417.

de Leon MJ, George AE, Golomb J, Tarshish C, Convit A, Kluger A, De SantiS, McRae T, Ferris SH, Riesberg B, Ince C, Rusinek H, Bobinski M,Quinn B, Miller DC, Wisenieski HM. 1997. Frequency of hippocampalformation atrophy in normal ageing and Alzheimer’s disease. NeurobiolAgeing 18:1–11.

Dobyns WB, Truwit CL. 1995. Lissencephaly and other malformations ofcortical development update. Neuropediatrics 26:132–147.

Galaburda AM, Sherman GF, Geschwind N. 1983. Developmental dyslexia:third consecutive case with cortical anomalies. Soc Neurosci Abstr9:940.

Gundersen HJ, Bagger P, Bendtsen TF, Evans SM, Korbo L, Marcussen N,Moller A, Nielsen K, Nyengaard JR, Pakkenberg B. 1988a. The newstereological tools: dissector, fractionator, nucleator and point sampledintercepts and their use in pathological research and diagnosis. APMIS96:857–881.

Gundersen HJ, Bendtsen TF, Korbo L, Marcussen N, Moller A, Nielsen K,Nyengaard JR, Pakkenberg B, Sorensen FB, Vesterby A. 1988b. Somenew, simple and efficient stereological methods and their use inpathological research and diagnosis. APMIS 96:379–394.

STEREOLOGY OF THE MURINE HIPPOCAMPUS 565

Heckers S, Heinsen H, Geiger B, Beckmann H. 1991. Hippocampal neuronnumber in schizophrenia. A stereological study. Arch Gen Psychiatry48:1002–1008.

Herrup K, Sunter K. 1987. Numerical matching during cerebellar develop-ment: quantitative analysis of granule cell death in staggerer mousechimeras. J Neurosci 7:829–836.

Kempermann G, Kuhn HG, Gage F. 1997a. More hippocampal neurons inadult mice living in an enriched environment. Nature 386:493–495.

Kempermann G, Kuhn HG, Gage F. 1997b. Genetic influence on neurogen-esis in the dentate gyrus of adult mice. Proc Natl Acad Sci USA94:10409–10414.

Kovelman JA, ScheibelAB. 1984.Aneurohistological correlate of schizophre-nia. Biol Psychiatry 19:1601–1621.

Kovelman JA, Scheibel AB. 1986. Biological substrates of schizophrenia.Acta Neurol Scand 73:1–32.

Lange W. 1975. Cell number and cell density in the cerebellar cortex of manand some other mammals. Cell Tissue Res 157:115–124.

Lupien SJ, de Leon M, deSanti S, Convit A, Tarshish C, Nair NPV, ThakurM, McEwen B, Hauger RL, Meany MJ. 1998. Cortisol levels duringhuman ageing predict hippocampal atrophy and memory deficits.Nature Neurosci 1:69–73.

Mayr E. 1963. Animal species and evolution. Cambridge, MA: BelknapPress/Harvard University Press.

McEwen BS. 1997. Possible mechanisms for atrophy of the human hippo-campus. Mol Psychiatry 2:255–262.

McLaren A, Molland P, Signer E. 1995. Does monozygotic twinning occur inmice? Genet Res 66:195–202.

Meencke HJ, Janz D. 1984. Neuropathological finding in primary general-ized epilepsy: a study of eight cases. Epilepsia 25:8–21.

Miller MW. 1986. Effects of alcohol on the generation and migration ofcerebral cortical neurons. Science 233:1308–1311.

Mullen RJ. 1977. Site of gene action and Purkinje cell mosaicism in pcd andnormal chimeric mice. Nature 270:245–247.

Mullen RJ, Herrup K. 1979. Chimeric analysis of mouse cerebellar mutants. In:Breakfield XO, editor. Neurogenetics: a genetic approach to the centralnervous system. New York: Elsevier-North Holland. p 271–297.

Nowakowski RS. 1986a. How do genes influence behavior? Some examplesfrom mutant mice. In: Smith SD, editor. Genetics and learning disabili-ties. San Diego, CA: College Hill Press. p 125–151.

Nowakowski RS. 1986b. Neuronal migration in the hippocampal in the lamina-tion defect (Hld) mutant mouse. In: Marthy, HJ, editor. Cellular andmolecular control of direct cell interaction. NY: Plenum Press. p 133–154.

Nowakowski RS. 1988. Development of the hippocampal formation inmutant mice. Drug Dev Res 15:315–336.

Nowakowski RS, Rakic P. 1981. The site of origin and route and rate ofmigration of neurons to the hippocampal region of the rhesus monkey. JComp Neurol 196:129–154.

Pache JC, Roberts N, Vock P, Zimmermann A, Cruz-Orive LM. 1993.Vertical LM sectioning and parallel CT scanning designs for stereology:application to human lung. J Micros 170:9–24.

Rice DS, Williams RW, Goldowitz D. 1995. Genetic control of retinal projectionsin inbred strains of albino mice. J Comp Neurol 354:459–469.

Scheibel AB. 1991. Are complex partial seizures a sequela of temporal lobedysgenesis? Adv Neurol 55:59–77.

Scheibel AB, Conrad AS. 1993. Hippocampal dysgenesis in mutant mouse andschizophrenic man: is there a relationship? Schizophr Bull 19:21–33.

Sekiguchi M, Abe H, Shimai K, Huang G, Inoue T, Nowakowski RS. 1994.Disruption of neuronal migration in the neocortex of the dreher mutantmouse. Brain Res Dev Brain Res 77:37–43.

Sterio DC. 1984. The unbiased estimation of number and sizes of arbitraryparticles using the disector. J Microsc 134:127–136.

Symons JP, Davis RE, Marriott JG. 1988. Water-maze learning and effectsof cholinergic drugs in mouse strains with high and low hippocampalpyramidal cell counts. Life Sci 42:375–383.

Williams RW, Herrup K. 1988. The control of neuron number. Ann RevNeurosci 11:423–453.

Wimer RE, Wimer CC. 1982. A biometrical-genetic analysis of granulecell number in the areas dentata of house mice. Dev Brain Res2:129–140.

Wimer RE, Wimer CC, Chernow CR, Balvanz BA. 1980. The geneticorganisation of neuron number in the pyramidal cell layer of hippocam-pal regio-superior in house mice. Brain Res 196:59–77.

Wimer RE, Wimer CC, Vaughn JE, Barber RP, Balvanz BA, Chernow CR.1978. The genetic organisation of neuron number in Ammon’s horns ofhouse mice. Brain Res 118:219–243.

Woodruff PWR, Murray RM. 1994. The aetiology of brain abnormalities inschizophrenia. In: Ancill R, editor. Exploring the spectrum of psychosis.New York: John Wiley and Sons. p 95–114.

566 I. ABUSAAD ET AL.