Sexual Dimorphism in Neuronal Number
of the Posterodorsal Medial Amygdala Is
Independent of Circulating Androgens
and Regional Volume in Adult Rats
JOHN A. MORRIS,1CYNTHIA L. JORDAN,1,2
1Neuroscience Program, Michigan State University, East Lansing, Michigan 48824-1101
2Psychology Department, Michigan State University, East Lansing, Michigan 48824-1101
AND S. MARC BREEDLOVE1,2*
The posterodorsal medial amygdala (MePD) in rodents integrates olfactory and phero-
monal information, which, coupled with the appropriate hormonal signals, may facilitate or
repress reproductive behavior in adulthood. MePD volume and neuronal soma size are
greater in male rats than in females, and these sexual dimorphisms are maintained by adult
circulating hormone levels. Castration of adult males causes these measures to shrink to the
size seen in females 4 weeks later, whereas testosterone treatment of adult females for 4
weeks enlarges these measures to the size of males. We used stereological methods to count
the number of cells in the MePD and found that, in addition to the sex difference in regional
volume and soma size, males also have more MePD neurons than do females, yet these
numbers are unaffected by the presence or absence of androgen in adults of either sex. Males
also have more glial cells than do females, but, in contrast to the effects on neuronal number,
the number of glial cells is affected by androgen in the right MePD of both sexes and,
therefore, may contribute to regional volume changes in adulthood in that hemisphere. Thus,
regional volume, neuronal size, and glial numbers vary in the MePD of adult rats in response
to circulating androgens, but neuronal number does not. These results suggest that the sex
difference in neuronal number in the rat MePD may be “organized” by androgens prior to
adulthood, whereas regional volume, neuronal size, and glial numbers can be altered by
androgens in adulthood. J. Comp. Neurol. 506:851–859, 2008.
© 2007 Wiley-Liss, Inc.
Indexing terms: medial amygdala; adult hormone manipulation; neural plasticity; glia; brain
morphology; stereology; testosterone; sexual dimorphism
In mammals, sex differences in brain morphology typi-
cally result from differential exposure to gonadal steroid
hormones during critical periods in development, al-
though steroids also exert effects on the brain during
puberty and in adulthood. As an example, the medial
amygdala is influenced by both perinatal and adult andro-
gen manipulations (Mizukami et al., 1983; Malsbury and
McKay, 1994; Cooke et al., 1999; Stefanova and Ovcharov,
2000). One quadrant of the medial amygdala, the pos-
terodorsal aspect (MePD), is particularly sensitive to an-
drogen manipulations in adult rodents. Adult levels of
circulating androgen maintain the larger volume of the
MePD in male rats compared with females. Castration of
adult males, or testosterone (T) treatment of females,
abolishes the sex difference by decreasing or increasing
MePD volume in males or females, respectively (Cooke et
al., 1999). Although many brain regions are sexually di-
morphic in volume, the MePD appears unusual in that it
is sensitive to hormones in adulthood.
Gonadal steroids may activate androgen receptors (AR)
and/or estrogen receptors (ER) to cause the sexual dimor-
phism in MePD volume. Systemic treatment of adult go-
nadectomized males with estradiol, but not dihydrotestos-
terone (DHT), maintains MePD volume in rats (Cooke et
Grant sponsor: National Institutes of Health; Grant number: NS28421;
Grant number: NS0450195.
*Correspondence to: S. Marc Breedlove, Neuroscience Program, 108
Giltner Hall, Michigan State University, East Lansing, MI 48824-1101.
Received 13 June 2007; Revised 21 August 2007; Accepted 11 September
Published online in Wiley InterScience (www.interscience.wiley.com).
THE JOURNAL OF COMPARATIVE NEUROLOGY 506:851–859 (2008)
© 2007 WILEY-LISS, INC.
al., 2003), yet male rats with a dysfunctional AR exhibit
MePD volume intermediate to that of control males and
females (Morris et al., 2005), indicating that AR also con-
tributes to the masculinization of the MePD. In adulthood,
this region contains abundant neurons expressing AR
and/or ER protein (Roselli, 1991; Li et al., 1997; Yokosuka
et al., 1997) and mRNA (McAbee and DonCarlos, 1998;
Shughrue et al., 1997), suggesting that gonadal hormones
may act directly in the MePD to influence its morphology.
Sex differences in MePD volume may be caused by a
number of integral anatomical components: cell size (soma
and processes), cell number, vasculature, neuropil, and
extracellular space. Although the size of neurons in the
MePD is dimorphic (Bubenik and Brown, 1973; Cooke et
al., 1999; Morris et al., 2005; Hermel et al., 2006) and,
therefore, might account for the difference in MePD vol-
ume, previous work has shown that DHT treatment main-
tains neuronal size in the male MePD, without maintain-
ing volume (Cooke et al., 2003). This dissociation of
neuronal size and regional volume indicates that other
components of the MePD are probably affected by circu-
lating androgen to alter MePD volume. For example, no
one has determined whether the sex difference in adult
medial amygdala volume is accompanied by a sex differ-
ence in neuronal number, nor whether androgen manipu-
lations in adulthood affect the number of neurons. This
possibility is raised because T increases neurogenesis in
the medial amygdala of castrated adult male meadow
voles (Fowler et al., 2003) and prevents apoptosis in the
preoptic area of developing male rats, resulting in a larger
adult volume (Davis et al., 1996).
We report here that male rats have more neurons and
glia in the MePD than do females. Although neuronal
number is unaffected by hormone manipulations in adults
of either sex, glial number is affected in the MePD of the
right hemisphere only. These results further delineate the
cellular mechanisms involved in the hormone-regulated
plasticity of the rat MePD, indicating that a permanent
sex difference in neuronal number is organized earlier in
life and does not contribute to fluctuations in regional
volume in adulthood.
MATERIALS AND METHODS
Sixty-day-old male and female Long Evans rats
(Charles River, Wilmington, MA) were housed three per
cage by gender in standard rat cages with food and water
freely available. Males and females were housed in sepa-
rate rooms. Lights were turned off at 1900 hours and on at
0700 hours. Animal care followed standards set by the
National Institutes of Health and were approved by the
Institutional Animal Care and Use Committee at Michi-
gan State University. After 1 week of acclimation, surger-
ies and hormone capsule implantations were performed
during isoflurane inhalant anesthesia using aseptic pro-
Animals were randomly assigned to one of four groups
of N ? 10 animals each. Females were ovariectomized
(OVX) and implanted s.c. with two Silastic capsules (each
having 20 mm effective release length, 30 mm total
length, i.d. 0.062 inch; o.d. 0.125 inch) containing either
crystalline testosterone (T) or nothing (“blank”) and killed
28 days later. Capsules were incubated for 48 hr in
phosphate-buffered saline (pH 7.4) before implantation.
Males were either castrated or subjected to sham surgery
and then killed 28 days later.
At the end of the experiment, animals were injected IP
with an overdose of sodium pentobarbital (120 mg/kg).
Deep anesthesia was noted by lack of reflexes to tail and
foot pinch as well as a lack of a corneal reflex. Blood was
taken via cardiac puncture for radioimmunoassay. Ani-
mals were then perfused intracardially with 0.9% saline,
followed by 10% neutral buffered formalin (?300 ml/
animal). Gonadectomies and hormone implants were con-
firmed at sacrifice. Brains, spinal cords, preputial glands,
seminal vesicles, and perineal muscles were removed;
placed into the same fixative solution; and, after at least 1
month of fixation, trimmed and weighed.
Plasma testosterone concentrations were measured in
duplicate using the Coat-a-Count Total Testosterone Kit
(Diagnostic Products Corp., Los Angles, CA) and sample
volumes of 50 ?l plasma. The lower limit of detectability
was 0.1 ng/ml, and the intraassay coefficient of variation
After at least 1 month postfixation in buffered formalin,
the brains were placed overnight in 20% phosphate-
buffered sucrose (pH 7.4) at 4°C prior to coronal sectioning
on a freezing sliding microtome set to 40 ?m using Multi-
Brain Technology (Neuroscience Associates). An embed-
ding matrix ensured no lost sections and preserved hemi-
spheric orientation. Brains were sectioned throughout the
entire region of interest and alternate sections mounted
onto gelatin-subbed slides with a random start to ensure
that every section had an equal probability of being chosen
for sampling. Mounted tissue was allowed to air dry,
stained with thionin for Nissl substance, and coverslipped
All analyses were conducted by an investigator blind to
group status. The investigator first measured the regional
volume of the MePD on both sides of the brain. MePD
boundaries were determined as previously described
(Morris et al., 2005) following nomenclature from a stan-
dard rat atlas (Paxinos and Watson, 2005) and using
criteria from Hines et al. (1992) and Canteras et al. (1995).
Furthermore, in defining the MePD borders, at least two
cytoarchitectonic features (size, shape, distribution, orien-
tation, staining intensity, packing density, and heteroge-
neity of cells) were used to distinguish it from the sur-
rounding areas, which allowed the entire reference space
to be accurately and consistently traced (Fig. 1).
StereoInvestigator (MicroBrightField, Colchester, VT)
software was used to estimate the volume of the MePD
and the average perikaryal size (n ? 8 animals per group).
A digital camera on a Zeiss Axioplan 2 compound micro-
scope captured an image containing the area of interest
from which the perimeter of the area was traced in suc-
cessive sections throughout the rostrocaudal axis. Total
volume of the MePD in each hemisphere was calculated by
multiplying the sampling ratio (4) times section thickness
(40 ?m) times the total two-dimensional area traced.
We estimated the size of neuronal cell bodies only and
not apparent glia. For perikaryal estimates, sample re-
The Journal of Comparative Neurology. DOI 10.1002/cne
852 J.A. MORRIS ET AL.
(MePD). Photomicrographs of coronal sections of the MePD in an
adult, gonadally intact male rat (left column) and an adult, ovariec-
tomized female rat without androgen treatment (right column). The
caudalmost portion of the MePD appears as a capsule ventral to the
optic tract (ot) surrounded by the relatively soma-sparse area of the
stria terminalis (st), at a rostrocaudal level intersecting the lateral
ventricle (v) and the anterolateral part of the amygdalohippocampal
transition area (AHiAL). At this extreme end, the MePD is of a similar
Boundaries of the rat posterodorsal medial amygdala
size in males (a) and females (d). More rostrally, the MePD in males
(b) has a triangular or wedge-shaped profile, with a slight reduction in
cell density in the center, which gives the appearance of an inverted
letter “v.” In females (e), the nucleus is considerably smaller and has
a less cuneate appearance. The rostral end of the MePD is still
adjacent to the ventral-lateral aspect of the ot and is more prominent
in males (c) than in females (f). In both sexes, the ovoid intercalated
nucleus of the amygdala is visible to the right of the MePD at this
level. Scale bar ? 250 ?m.
The Journal of Comparative Neurology. DOI 10.1002/cne
853SEXUAL DIMORPHISM IN MEDIAL AMYGDALA CELL NUMBER
gions were randomly selected by the StereoInvestigator
software, which positioned points within the MePD with-
out bias for location or appearance. On average, the two-
dimensional profile areas of 10 perikarya from each sec-
tion were measured throughout the rostrocaudal extent of
the MePD (sampling 55–95 neurons per hemisphere) and
were traced with a ?100 Plan-NeoFluar, 1.3 N.A oil-
immersion objective. Neuronal perikarya measures were
averaged within each hemisphere, yielding a mean soma
size (area in square micrometers) per hemisphere for each
Neurons were identified by the presence of a distinct
Nissl-stained cytoplasm and nucleolus (Fig. 2). Glia were
identified by the presence of many darkly stained hetero-
chromatic bodies in the nucleus that were interspersed
with strands of thionin-stained material. Such presump-
tive glial nuclei were generally smaller and were not sur-
rounded by a distinct cytoplasmic shell. The smallest cells
that also had a very darkly stained nucleus with several
large stained bodies within the nucleus were categorized
as microglia. Cells that could not be classified with confi-
dence into these three categories were designated as “un-
The optical fractionator method was used for estimating
number of cells (West et al., 1991). This method estimates
total number of objects (Nobj) from the sum of objects
sampled (?N) in a known fraction of the reference space.
The calculation then is: Nobj ? (?N)(1/ssf)(1/asf)(1/tsf),
where section sampling fraction (ssf) is the number of
sections sampled divided by the entire number of sections
through the reference space, area sampling fraction (asf)
is the total area sampled by the array of counting frames
divided by the total area of all sampled sections, and
thickness sampling fraction (tsf) is the height of the dis-
ector divided by the average total section thickness.
Cells in the MePD were counted by the optical fractiona-
tor method (West et al., 1991) with a Plan-NeoFluar ?100
oil-immersion (1.3 N.A.) objective, which allowed us to
distinguish in most cases neurons from glia and to ensure
discrimination of discrete objects. Sampling parameters
were set to allow a coefficient of error (m ? 1; Gundersen,
1999) of no more than 0.10 for each animal, and percent-
age of total observed variance resulting from interanimal
variance to be at least double that of variance from sam-
pling error (n ? 5 per group). A pilot study indicated that
sampling boxes with the following measures were appro-
priate: height, 5 ?m; guard height, 1.5 ?m minimum;
sampling frame area, 625 ?m2; x-y spacing 125 ?m. An
average of 272 cells was counted in each side of each brain.
Average thickness of stained sections for each brain was
estimated by measuring section thickness in every third
counting frame and was used to adjust cell counts for that
brain. Overall average section thickness was 8 ?m. DIC
optics used in a pilot study indicated that potential slicing
artifact was limited to about 1 ?m, which left approxi-
mately 5 ?m of depth for sampling cell counts. Neuronal
nucleoli and glial nuclei were used as unique counting
points. Adobe Photoshop 7.0 was used to generate figures
containing photomicrographs. No processing of images oc-
curred except for resizing and brightness.
MePD as revealed by thionin stain for Nissl. Arrow indicates a neuron; black arrowheads indicate glia
(oligodendrocyte or astrocyte); white arrowhead indicates a microglia. Scale bar ? 10 ?m.
Classification of cell types. Typical cells viewed with a ?100 objective in a sham male adult rat
The Journal of Comparative Neurology. DOI 10.1002/cne
854J.A. MORRIS ET AL.
Results are expressed as mean ? SEM. A three-way
mixed-design analysis of variance (ANOVA), with left and
right sides as a repeated measure, and two independent
factors [sex and androgen status (high in gonadally intact
males and T-treated females, low in castrate males and
blank-treated females)], was conducted for each depen-
dent variable (MePD volume, average perikaryal area,
neuronal number, glial number). One statistically signif-
icant interaction was detected for each of two measures
(soma size and glial number), prompting followup two-way
ANOVAs to confirm trends detected by the three-way
ANOVA. t-Tests comparing sham-operated males and
blank-treated females were used to assess previously re-
ported sex differences in MePD regional volume and soma
size (Cooke et al., 1999; Morris et al., 2005). t-Tests were
also used to assess the well-documented effects of T on the
periphery to confirm androgen manipulations. For all
analyses, a value of 0.05 was used as the significance
criterion, with N representing the number of animals in
Bioassays of T treatment
Testosterone manipulations for 28 days had gross ef-
fects in both males and females (Table 1). Castration of
males significantly decreased the average weight of the
levator ani muscles, and overall body weight (all Ps ?
0.05). T implants in females significantly increased the
average weight of preputial glands (P ? 0.05), but not
overall body weight (P ? 0.17). As expected, circulating T
was beneath the detection limit in castrated males and
blank-treated females. Insofar as T levels did not differ
significantly between sham-castrated males and T-treated
females (P ? 0.10), the treatment appeared to provide
physiologically relevant androgen stimulation to females.
As in previous studies, the volume of the MePD was
greater in males than in females and, in both males and
females, was greater in the right hemisphere than in the
left and was affected by adult androgen levels (Fig. 3A).
The three-way ANOVA revealed these differences as sig-
nificant main effects of sex (male greater than female; P ?
0.0001), of androgen (greater in the presence of androgen
than absence of androgen; P ? 0.003), and of hemisphere
(right greater than left; P ? 0.005), with no significant
interaction terms. Hence no post hoc comparisons for re-
gional volume were called for, although examination of the
data suggests that the left MePD of males may be less
responsive to androgen than the right (Fig. 3A). We have
twice reported a sex difference in MePD volume in gonad-
ally intact rats. A post hoc comparison of the closest com-
parison groups available from this study (sham males and
untreated ovariectomized females) also revealed volumet-
TABLE 1. Average Weights of the Body, Preputial Glands, Seminal
Vesicles, and Bulbocavernosus/Levator Ani Muscles (BC/LA) for Each
Group ? SEM1
(g) BC/LA (g) T (ng/ml)
OVX female ? T
OVX female ? B
462 ? 14
382 ? 9.1
319 ? 8.8
297 ? 13
0.21 ? 0.01
0.12 ? 0.02
0.24 ? 0.01
0.12 ? 0.01
1.99 ? 0.14
0.30 ? 0.03
1.65 ? 0.05
0.75 ? 0.05
2.68 ? 0.43
1.79 ? 0.28
1Plasma testosterone (T) levels were detectable only in sham males and OVX females
given T, which were not significantly different from one another. ND, below limit of
detection of 0.1 ng/ml. For each measure within each sex, hormone manipulation
(castration in males, T treatment in females) had a significant effect (t-tests, all Ps ?
0.05), with the exception of body weight in females.
size in adult rats. A: As seen in previous studies, manipulations of
androgen in adult rats alter the regional volume of the posterodorsal
medial amygdala (MePD). Males were either castrated (low androgen)
or subjected to sham surgery (high androgen), whereas females were
ovariectomized and given capsules containing either testosterone
(high androgen) or nothing (low androgen). Three-way ANOVA re-
vealed a sex difference (males ? females; main effect of sex; P ?
0.0001), asymmetry (right ? left; main effect of hemisphere; P ?
0.005) and a response to androgen manipulation (main effect of hor-
mone; P ? 0.0003), with no significant interactions. B: Neuronal soma
size in the rat MePD also responded to androgen manipulations (main
effect of androgen; P ? 0.0005), but there was no evidence of asym-
metry (P ? 0.2). Although there was not a significant sex difference
overall, there was an interaction of sex and androgen status (P ?
0.02), because soma size was more responsive in females than in
males. The previously reported sex difference in MePD soma size in
gonadally intact animals is represented here by the larger somata in
sham males compared with females receiving no androgen.
Androgen regulates medial amygdala volume and soma
The Journal of Comparative Neurology. DOI 10.1002/cne
855SEXUAL DIMORPHISM IN MEDIAL AMYGDALA CELL NUMBER
ric differences in both the left and the right MePD (t-tests;
Ps ? 0.05), presumably a result of higher circulating an-
drogen in the males than in the females (Table 1).
Neuronal soma size
Also as in previous studies, there was no laterality of
MePD soma size, but this measure was affected by adult
hormone levels in both sexes (Fig. 3B). The ANOVA re-
vealed these effects as a significant main effect of andro-
gen (P ? 0.001), with no significant main effect of lateral-
ity nor any significant interaction of any factors with
laterality. There was no significant main effect of sex,
indicating that adult androgen status is more important
than sex in determining soma size. There was a significant
sex-by-androgen interaction (P ? 0.02), because androgen
status had a greater effect in females than in males. We
previously reported sex differences in MePD soma size in
gonadally intact rats and in the present study found that
the somata are larger in sham males than untreated fe-
males in both the left and the right MePD (Ps ? 0.05,
t-tests), presumably a result of higher circulating andro-
gen in the males than in the females. The absence of
laterality in MePD soma size, despite the robust asymme-
try in regional volume, indicates that the asymmetry in
regional volume cannot be attributed to asymmetry in
soma size. Sham males have also been reported to have
equal-sized neurons across hemispheres, despite asymme-
try in regional volume.
Males have more MePD neurons than do females (main
effect of sex P ? 0.002), and there are more neurons in the
left hemisphere than the right in both sexes (main effect of
side; P ? 0.0001; Fig. 4A). However, these sex differences
and laterality in neuronal number are unaffected by adult
androgen status, as there was no main effect of androgen
status (P ? 0.5) nor any statistically significant interac-
tion terms (Ps ? 0.13). Thus, the left MePD is smaller in
volume than the right MePD, yet the left contains more
neurons than the right, and this dissociation is seen in
both sexes. Furthermore, the absence of androgenic effects
on neuronal number indicates that other factors must
underlie the effects of androgen on regional volume in the
MePD of adult rats and is another dissociation of neuronal
number and regional volume.
Males have more MePD glial cells than do females, in
both hemispheres (main effect of sex; P ? 0.003; and no
significant interaction of sex with any other factor). The
number of glial cells is greater on the right than on the left
(main effect of hemisphere; P ? 0.001), and there was no
significant main effect of androgen status (P ? 0.4). How-
ever, there was a significant interaction of androgen sta-
tus and hemisphere (P ? 0.03), which prompted a followup
analysis of the left and right hemispheres separately (Fig.
4B). Two-way ANOVA (sex and androgen status as inde-
pendent factors) of glial numbers in the left hemisphere
revealed a marginally significant sex difference (P ? 0.06)
but no effect of androgen. In contrast, in the right hemi-
sphere, there was both a sex difference (main effect; P ?
0.003) and an effect of androgen (P ? 0.01), with no inter-
action. Thus high androgen is associated with a greater
number of glial cells in the MePD in the right hemisphere
only, and this effect is seen in both sexes.
We considered the density of glia by treatment group,
using the cell counts and regional volume measures re-
ported above. These indicated that the right MePD con-
tains a higher density of glia, and the left MePD contains
a higher density of neurons, in all treatment groups. Over-
all, there does not appear to be a significant sex difference
in neuronal density, but OVX females that receive no T
have a higher density of glia than sham males. This result
seems to confirm the report that female rats display a
greater density of glial fibrillary acidic protein (GFAP)
staining than males in the MePD (Rasia-Filho et al.,
Other cell types
Our sampling parameters were not designed to obtain
reliable counts of rarely encountered microglia or “un-
MePD. A: Males have more MePD neurons than females (main effect
of sex; P ? 0.002), and there are more neurons in the left MePD than
the right (main effect of hemisphere; P ? 0.0001), but there was no
effect of androgen manipulations in either sex or in either hemisphere
(no significant interactions; Ps ? 0.14). B: Males also have more glial
cells in the MePD than do females (main effect of sex; P ? 0.003), and
there are more glia in the right MePD than the left (main effect of
hemisphere; P ? 0.001). However, there was an interaction of andro-
gen status and hemisphere (P ? 0.03). Subsequent separate analyses
of the left and right MePD revealed that androgen is associated with
more glia in both sexes, but only in the right MePD.
Androgen has no effect on neuronal number in the adult rat
The Journal of Comparative Neurology. DOI 10.1002/cne
856J.A. MORRIS ET AL.
known” cells, but the mean estimates for the number of
microglia suggest that, if there is a difference, it also
favors males (sham males ? 3,428 ? 427, castrated
males ? 5,118 ? 576, T-treated females ? 2,720 ? 323,
blank-treated females ? 2,605 ? 260). Likewise, the esti-
mated number of unknown cells was only slightly greater
in males than in females (sham males ? 8,148 ? 1,491,
castrated males ? 11,833 ? 1,296, T-treated females ?
8,134 ? 1,184, blank-treated females ? 6,678 ? 1,424).
Because the mean estimates of microglia and unidentified
cells are greater in males than in females, the sex differ-
ences in numbers of neurons and glia favoring males can-
not be accounted for by misclassification of cells.
The adult rat MePD shows a remarkable level of plas-
ticity that is regulated by gonadal hormones. We found
that the adult female rat MePD responds to T treatment
with increases in volume and neuronal size in both hemi-
spheres and number of glia in the right hemisphere. The
male MePD also responds to gonadal hormones: adult
castration leads to reductions in regional volume and neu-
ronal size in both hemispheres and the number of glia in
the right hemisphere. Unlike the other measures, how-
ever, the number of neurons in males and females was not
affected by changes in adult androgen. These results limit
hypotheses about cellular mechanisms underlying andro-
gen’s effects on MePD regional volume in adult rats and
how hormonally induced plasticity of this nucleus may
affect reproductive behavior. In turn, adult behaviors that
induce hormonal variations, such as exposure to an es-
trous female and social conflict, may influence the regional
volume and neuronal size in this nucleus but are unlikely
to affect neuronal number.
Because adult hormone manipulations can abolish sex
differences in MePD volume (Cooke et al., 1999), one
might have expected that the number of neurons was
equivalent in the two sexes and that adult fluctuations in
volume might reflect changes in neuropil that simply alter
the density of neurons within the nucleus. However, we
found that males in fact have more MePD neurons than do
females in adulthood, no matter what the androgen status
of the animals. Thus castration of adult males causes the
MePD to shrink in volume, without any loss of neurons,
whereas androgen treatment of females enlarges MePD
volume, without adding to neuronal number. Change in
regional volume cannot be explained by changes in neu-
ronal number, so other factors must contribute to volu-
metric changes. We found evidence that androgen affects
glial number, but only in the right hemisphere. In the
present study, changes in MePD soma size generally re-
flected changes in volume, so soma size may contribute to
some of the volume changes we detected. However, the
persistent asymmetry in MePD volume stands in contrast
to a lack of any asymmetry in MePD neuronal somata size.
Furthermore, in previous studies, treatment with the var-
ious metabolites of testosterone can dissociate soma size
from volume (Cooke et al., 2003). By elimination, these
results together suggest that changes in synaptic neuro-
pil, including the dendrites of neurons and processes of
glia, probably underlie the bulk of change in MePD vol-
ume in adult rats. Reports of hormone-induced changes in
MePD dendritic structure lend credence to this idea (Go-
mez and Newman, 1991; Rasia-Filho et al., 2005; Hermel
et al., 2006). Cooke et al. (2007) report a sex difference in
the proportion of MePD volume occupied by dendritic pro-
cesses in prepubertal rats, consistent with this idea. They
also found more neurons in males than in females in the
right MePD of prepubertal animals but no sex difference
in neuronal number on the left. Thus the sex difference in
the left MePD may arise during or after puberty, which
would suggest a dramatic reorganization during this pe-
riod of ontogeny.
There is also an interesting and pervasive dissociation
of regional volume and neuronal number in these data.
The MePD has a larger volume on the right than the left,
yet there are more neurons on the left than on the right.
We found this pattern of results in both sexes, and andro-
gen manipulations had no effect on neuronal number.
Again, one might have expected that the hemisphere con-
taining a larger MePD would also contain more MePD
neurons. It seems clear that the several components that
contribute to the volume of a brain nucleus (neuronal
number, glial number, neuronal soma size, etc.) are inde-
pendent, at least in the rat MePD.
The sex difference in neuronal number within the
MePD is unaffected by adult androgen manipulations, so
presumably T or its metabolites affect neuronal number
earlier in life. Cooke and Woolley (2005) showed that, in
pre-pubertal rats at 25–29 days of age (when circulating
androgens are equivalent in males and females), soma
size, cell density, and neuronal number in the left MePD
were not sexually differentiated, but volume was, being
significantly larger in males. If the male amygdala con-
tains more neurons only after puberty, it may be that
rising levels of androgens promote increases in neuronal
size and number in pubertal males. Though speculative,
this would explain why, in humans, amygdalar volume
increases faster in boys than in girls between ages 4 and
18 (Giedd et al., 1997; Merke et al., 2003). Furthermore,
neurogenesis has been reported in the amygdala of young
adult male monkeys (Bernier et al., 2002). Conversely, if
female rats have fewer neurons after puberty than before,
then androgens may prevent apoptosis in male rats to
create the adult sex difference in neuronal number. In any
case, future studies could ask when the sex difference in
neuronal number arises in the developing MePD and
whether androgen organizes this characteristic.
There is also the question of how such differences in
neuronal number emerge. What is the cellular mechanism
involved? Studies of cell number in the MePD during
ontogeny might indicate whether the sex difference in
neuronal number arises because of sex differences in neu-
rogenesis, neuronal migration, neuronal differentiation,
and/or apoptosis. Although we saw no change in neuronal
number in the adult MePD, this does not address whether
neurogenesis or apoptosis occurs in adults. If neurogen-
esis and apoptosis are ongoing and in balance in adults,
our counts of the net number of neurons would not detect
those processes. Although castration can affect neuronal
proliferation in the MePD of adult male meadow voles
(Fowler et al., 2003), there may be species differences in
neurogenesis, neurodegeneration, or both.
Overall, the present findings add to the literature indi-
cating that morphologic lateral asymmetry is common in
the amygdala (Cooke et al., 2003; Morris et al., 2005).
Although androgen receptor distribution has been found
to be asymmetrical in adult rat hippocampus (Xiao and
Jordan, 2002), no such laterality was reported in the me-
The Journal of Comparative Neurology. DOI 10.1002/cne
857SEXUAL DIMORPHISM IN MEDIAL AMYGDALA CELL NUMBER
dial amygdala of mice (Lu et al., 1998). The human amyg-
dala has also been reported to be asymmetric in terms of
both structure (Murphy, 1986) and function (Phillips et
al., 2001; Canli et al., 2002; McClure et al., 2004; Cahill et
al., 2004; Hamann et al., 2004).
For glial number, only the right side of the MePD is
sensitive to hormones. T-treated females had more glia
than blank-treated females only in the right MePD. Con-
versely, castrated males have fewer glia than gonadally
intact males only in the right MePD. These results sug-
gest that the right MePD is more sensitive than the left to
androgen’s effects on glial number. Whether this in-
creased sensitivity in the right hemisphere is due to asym-
metry in steroid receptors, metabolic enzymes (such as
aromatase), or some other factor has yet to be addressed.
Our results indicate that male rats have more glial cells
in the MePD than do female rats. However, visualizing
astroctyes in the rat MePD by GFAP immunoreactivity,
which is expressed by most (but not all) astrocytes (Eng et
al., 2000), reveals a sex difference in the opposite direc-
tion, with females showing a greater density of GFAP
staining than do males (Rasia-Filho et al., 2002). GFAP
marks only astrocytes, and our counts undoubtedly in-
clude both astrocytes and oligodendrocytes, so our find-
ings of a higher glial density in the female MePD could
account for the sex difference in GFAP staining seen by
Rasia-Filho et al. (2002). On the other hand, it is possible
that, despite the fewer glia found in females compared
with males, the glia in females may have longer or more
branched processes than those in males. In fetal rat hy-
pothalamic cultures, astrocytic process elaboration, but
not number, is increased by estradiol (Garcia-Segura et
al., 1989). Other studies have shown that GFAP immuno-
reactivity in the dentate gyrus of the hippocampus and the
MePD correlates with estrogen levels across the female’s
estrous cycle (Luquin et al., 1993; Martinez et al., 2006). It
will be important in future studies to count the number of
GFAP-expressing cells in the MePD both to validate the
identification of glia in this study and to probe the relative
contributions of astrocytes vs. other glia to our counts, as
well as any responses to steroid manipulations in adult-
hood. Although astrocytes have been shown to express
steroid receptors in some parts of the rat brain (Lorenz et
al., 2005; Tabori et al., 2005), we do not know based on the
current data which type of glia is increased in response to
androgen treatment of adult females, nor do we know
whether androgen is acting directly on glial cells or glial
precursors, or acting on some other cell type, including
neurons, to increase glial numbers indirectly in the right
MePD. Finally, it is possible that the T treatment did not
actually increase the number of glia in the right MePD but
somehow altered glia morphology to make the glia more
detectable in our Nissl-stained preparation. Thus any con-
clusions about the mechanisms of change in the number of
glia found in the MePD must remain tentative.
Our androgen manipulations did not affect MePD vol-
ume as dramatically as in a previous study (Cooke et al.,
1999), where mean volumes in gonadally intact males
were not significantly larger than in females given andro-
gen. In the present study, the sex difference in volume,
seen as a main effect of sex in ANOVA, persisted even
when controlling for androgen (Fig. 3A). It is possible that
providing more androgen and/or extending the androgen
treatment (Cooke et al. provided 30 days of T, and we
provided 28) might have more closely duplicated the re-
sults of the previous study. In any case, post hoc tests of
MePD volume in the right hemisphere (the more
androgen-responsive side of the brain) show no significant
difference between castrated males and females given an-
drogen. Therefore, both studies demonstrate that the sex
difference in MePD volume can be eliminated by manip-
ulations of androgen in adulthood.
This plasticity of the adult MePD in rats continues to
offer a model system for studying structure/function rela-
tionships in hormone-sensitive brain areas. Future stud-
ies should address directly whether adult hormone manip-
ulations affect dendritic and/or synaptic structures in the
MePD, the identity of glial cells affected, and the contri-
butions each of these changes to alterations in behavior of
Bernier PJ, Bedard A, Vinet J, Levesque M, Parent A. 2002. Newly gen-
erated neurons in the amygdala and adjoining cortex of adult primates.
Proc Natl Acad Sci U S A 99:11464–11469.
Bubenik GA, Brown GM. 1973. Morphologic sex differences in primate
brain areas involved in regulation of reproductive activity. Experientia
Cahill L, Uncapher M, Kilpatrick L, Alkire MT, Turner J. 2004. Sex-
related hemispheric lateralization of amygdala function in emotionally
influenced memory: an fMRI investigation. Learn Mem 11:261–266.
Canli T, Desmond JE, Zhao Z, Gabrieli JD. 2002. Sex differences in the
neural basis of emotional memories. Proc Natl Acad Sci U S A 99:
Canteras NS, Simerly RB, Swanson LW. 1995. Organization of projections
from the medial nucleus of the amygdala—a Phal study in the rat.
J Comp Neurol 360:213–245.
Cooke BM, Woolley CS. 2005. Sexually dimorphic synaptic organization of
the medial amygdala. J Neurosci 25:10759–10767.
Cooke BM, Tabibnia G, Breedlove SM. 1999. A brain sexual dimorphism
controlled by adult circulating androgens. Proc Natl Acad Sci U S A
Cooke BM, Breedlove SM, Jordan CL. 2003. Both estrogen receptors and
androgen receptors contribute to testosterone-induced changes in the
morphology of the medial amygdala and sexual arousal in male rats.
Horm Behav 43:336–346.
Cooke BM, Stokas MR, Woolley CS. 2007. Morphological sex differences
and laterality in the prepubertal medial amygdala. J Comp Neurol
Davis EC, Popper P, Gorski RA. 1996. The role of apoptosis in sexual
differentiation of the rat sexually dimorphic nucleus of the preoptic
area. Brain Res 734:10–18.
Eng LF, Ghirnikar RS, Lee YLG. 2000. Glial fibrillary acidic protein:
GFAP—thirty-one years (1969–2000) Neurochem Res. 25(9-10):1439–
Fowler CD, Freeman ME, Wang ZX. 2003. Newly proliferated cells in the
adult male amygdala are affected by gonadal steroid hormones. J Neu-
Garcia-Segura LM, Torres-Aleman I, Naftolin F. 1989. Astrocytic shape
and glial fibrillary acidic protein immunoreactivity are modified by
estradiol in primary rat hypothalamic cultures. Brain Res Dev Brain
Giedd JN, Castellanos FX, Rajapakse JC, Vaituzis AC, Rapoport JL. 1997.
Sexual dimorphism of the developing human brain. Prog Neuropsy-
Gomez DM, Newman SW. 1991. Medial nucleus of the amygdala in the
adult Syrian hamster—a quantitative Golgi analysis of gonadal
hormonal-regulation of neuronal morphology. Anat Rec 231:498–509.
Gundersen HJ, Jensen EB, Kieu K, Nielsen J. 1999. The efficiency of
systematic sampling in stereology reconsidered. J Microsc 193(3):199–
Hamann S, Herman RA, Nolan CL, Wallen K. 2004. Men and women differ
in amygdala response to visual sexual stimuli. Nat Neurosci 7:411–
Hermel EE, Ilha J, Xavier LL, Rasia-Filho AA, Achaval M. 2006. Influence
of sex and estrous cycle, but not laterality, on the neuronal somatic
The Journal of Comparative Neurology. DOI 10.1002/cne
858J.A. MORRIS ET AL.
volume of the posterodorsal medial amygdala of rats. Neurosci Lett
Hines M, Allen LS, Gorski RA. 1992. Sex differences in subregions of the
medial nucleus of the amygdala and the bed nucleus of the stria
terminalis of the rat. Brain Res 579:321–326.
Li X, Schwartz PE, Rissman EF. 1997. Distribution of estrogen receptor-
beta-like immunoreactivity in rat forebrain. Neuroendocrinology 66:
Lorenz B, Garcia-Segura LM, Doncarlos LL. 2005. Cellular phenotype of
androgen receptor-immunoreactive nuclei in the developing and adult
rat brain. J Comp Neurol 492:456–468.
Lu SF, McKenna SE, Cologer-Clifford A, Nau EA, Simon NG. 1998. An-
drogen receptor in mouse brain: sex differences and similarities in
autoregulation. Endocrinology 139:1594–1601.
Luquin S, Naftolin F, Garcia-Segura LM. 1993. Natural fluctuation and
gonadal hormone regulation of astrocyte immunoreactivity in dentate
gyrus. J Neurobiol 24:913–924.
Malsbury CW, McKay K. 1994. Neurotrophic effects of testosterone on the
medial nucleus of the amygdala in adult male rats. J Neuroendocrinol
Martinez FG, Hermel EE, Xavier LL, Viola GG, Riboldi J, Rasia-Filho AA,
Achaval M. 2006. Gonadal hormone regulation of glial fibrillary acidic
protein immunoreactivity in the medial amygdala subnuclei across the
estrous cycle and in castrated and treated female rats. Brain Res (in
McAbee MD, DonCarlos LL. 1998. Ontogeny of region-specific sex differ-
ences in androgen receptor messenger ribonucleic acid expression in
the rat forebrain. Endocrinology 139:1738–1745.
McClure EB, Monk CS, Nelson EE, Zarahn E, Leibenluft E, Bilder RM,
Charney DS, Ernst M, Pine DS. 2004. A developmental examination of
gender differences in brain engagement during evaluation of threat.
Biol Psychiatry 55:1047–1055.
Merke DP, Fields JD, Keil MF, Vaituzis AC, Chrousos GP, Giedd JN. 2003.
Children with classic congenital adrenal hyperplasia have decreased
amygdala volume: potential prenatal and postnatal hormonal effects.
J Clin Endocrinol Metab 88:1760–1765.
Mizukami S, Nishizuka M, Arai Y. 1983. Sexual difference in nuclear
volume and its ontogeny in the rat amygdala. Exp Neurol 79:569–575.
Morris JA, Jordan CL, Dugger BN, Breedlove SM. 2005. Partial demascu-
linization of several brain regions in adult male (XY) rats with a
dysfunctional androgen receptor gene. J Comp Neurol 487:217–226.
Murphy GM. 1986. The human medial amygdaloid nucleus—no evidence
for sex difference in volume. Brain Res 365:321–324.
Paxinos G, Watson W. 2005. The rat brain in stereotaxic coordinates.
Phillips ML, Medford N, Young AW, Williams L, Williams SC, Bullmore
ET, Gray JA, Brammer MJ. 2001. Time courses of left and right
amygdalar responses to fearful facial expressions. Hum Brain Mapp
Rasia-Filho AA, Xavier LL, dos Santos P, Gehlen G, Achaval M. 2002. Glial
fibrillary acidic protein immunodetection and immunoreactivity in the
anterior and posterior medial amygdala of male and female rats. Brain
Res Bull 58:67–75.
Roselli CE. 1991. Sex differences in androgen receptors and aromatase
activity in microdissected regions of the rat brain. Endocrinology 128:
Shughrue PJ, Lane MV, Merchenthaler I. 1997. Comparative distribution
of estrogen receptor-alpha and -beta mRNA in the rat central nervous
system. J Comp Neurol 388:507–525.
Stefanova N, Ovcharov V. 2000. Sexual dimorphism of the bed nucleus of
the stria terminalis and the amygdala. New York: Springer. p 78.
Tabori NE, Stewart LS, Znamensky V, Romeo RD, Alves SE, McEwen BS,
Milner TA. 2005. Ultrastructural evidence that androgen receptors are
located at extranuclear sites in the rat hippocampal formation. Neuro-
West MJ, Slomianka L, Gundersen HJ. 1991. Unbiased stereological esti-
mation of the total number of neurons in the subdivisions of the rat
hippocampus using the optical fractionator. Anat Rec 231:482–497.
Xiao L, Jordan CL. 2002. Sex differences, laterality, and hormonal regu-
lation of androgen receptor immunoreactivity in rat hippocampus.
Horm Behav 42:327–336.
Yokosuka M, Okamura H, Hayashi S. 1997. Postnatal development and
sex difference in neurons containing estrogen receptor-alpha immuno-
reactivity in the preoptic brain, the diencephalon, and the amygdala in
the rat. J Comp Neurol 389:81–93.
The Journal of Comparative Neurology. DOI 10.1002/cne
859SEXUAL DIMORPHISM IN MEDIAL AMYGDALA CELL NUMBER