Early Life Stress and Inherited Variation in Monkey Hippocampal Volumes FREE

SMALL HIPPOCAMPAL volumes are found in adult humans with recurrent major depression^{1 - 2} and posttraumatic stress disorder.^{3 - 4} Childhood stress increases the risk of developing these mood and anxiety disorders,⁵ and small hippocampal volumes are evident in adult survivors of childhood maltreatment.^{6 - 7} Hippocampal morphology is altered by stress in carefully controlled studies of rodents,^{8 - 11} and small hippocampal volumes in humans have been taken as evidence that stress-related disorders induce hippocampal volume loss.^{2 ,11 - 12} An alternative possibility that cannot be dismissed in the absence of prospective longitudinal research is that small hippocampal volumes are inherited and predispose toward the development of psychiatric disorders that are triggered or aggravated by stress.^{3 ,11}

Aside from a limited number of reports on inherited variation in autonomic activity¹³ and cerebrospinal fluid monoamine levels,^{14 - 15} primate research on stress neurobiology has focused on maternal deprivation.¹⁶ Rhesus macaque monkeys raised without mothers exhibit increased brain dopamine and norepinephrine sensitivity,^{17 - 18} exaggerated hypothalamic-pituitary-adrenal (HPA) responses to stress,^{19 - 20} altered regulation of autonomic activity,²¹ fragmented sleep patterns,²² depressionlike behavior,¹⁸ and excessive consumption of alcohol.²³ Ecologically informed research on maternal availability has likewise identified untoward effects on primate postnatal development. Bonnet macaque monkeys raised by their mothers in variable foraging-demand conditions are impaired in social and emotional development.^{24 - 25} In early adulthood, these same monkeys exhibit elevated cerebrospinal fluid levels of monoamines, somatostatin, and corticotropin-releasing factor (CRF).^{26 - 27}

Studies of stress and hippocampal plasticity in primates have produced conflicting results. Maternal deprivation in rhesus macaque monkeys increases dentate gyrus non–phosphorylated neurofilament protein levels, a neuronal marker of vulnerability in Alzheimer disease, Huntington disease, Parkinson disease, and amyotrophic lateral sclerosis.²⁸ Hippocampal cell damage in adult vervet monkeys is induced by sustained and fatal stress²⁹ or long-term treatment with cortisol implanted in adult hippocampus.³⁰ Yet elderly adult pigtailed macaque monkeys treated long-term with oral cortisol do not respond with diminished volumes or neuron numbers in any hippocampal cell field.³¹ Rhesus macaque monkeys raised in social isolation also fail to show altered hippocampal volumes as determined in vivo at 18 months of age by magnetic resonance imaging (MRI).³²

Here we test for early stress effects and inherited variation in hippocampal volumes in 39 squirrel monkeys. Paternal half-siblings raised apart from one another by different mothers in the absence of fathers were randomized to conditions that disrupted diverse aspects of early maternal care. In one condition on 5 occasions offspring were briefly separated from groups composed of 3 or 4 mother-infant pairs. Intermittent separations consistently elicit short-term elevations in cortisol levels with baseline cortisol levels restored soon after the subsequent social reunions.^{33 - 35} In 2 other postnatal rearing conditions, differences in maternal availability were produced by manipulating the effort required to find food. Body weights and amounts of food consumed in each foraging condition were not significantly different, but high-foraging demand mothers stopped carrying their infant earlier, and these infants displayed modest prolonged increases in cortisol levels throughout the high-demand condition.^{36 - 37}

Following completion of each 12-week condition at 21 weeks of age, all mothers were removed from natal groups after weaning at 36 weeks. At this stage of development squirrel monkeys are no longer reliant on maternal care. Sexual maturity is achieved at 3 years, and the squirrel monkey life span is approximately 21 years.³⁸ As previously described elsewhere in greater detail,³⁹ social behavior, emotional responses, and increases in cortisol levels elicited by removing all mothers after weaning were examined at 36 weeks of age for the study cohort. In early adulthood at 5 years we tested for differences in cortisol-negative feedback regulation of the HPA-axis response to exogenous stimulation by CRF.⁴⁰ Then 5 weeks later hippocampal volumes were determined by high-resolution MRI.

Forty infants sired by fathers that had no contact with their offspring were distributed randomly in groups each composed of 3 or 4 mother-infant pairs. All monkeys were of Guyanese origin (Saimiri sciureus) and were born and maintained at the Stanford University Primate Facility, Stanford, Calif. One monkey from the low-foraging demand condition (see "Experimental Design and Procedures" section) was excluded for reasons unrelated to the study. Twelve fathers and 30 mothers produced the 39 monkeys that constituted the study cohort. Twenty-one mothers each contributed 1 offspring, and 9 mothers produced with different fathers on separate occasions 2 offspring. Two fathers each sired a single offspring, 3 fathers each sired 2 offspring, 2 fathers each sired 3 offspring, 3 fathers each sired 4 offspring, 1 father sired 5 offspring, and 1 father sired 8 offspring. All procedures were conducted in accord with and as required by the Animal Welfare Act, and approved by Stanford University's Administrative Panel on Laboratory Animal Care.

Four natal groups were randomly assigned to each of the following 3 rearing conditions when infants were 10 weeks old (age range, 8-13 weeks).

After completion of these postnatal protocols at 21 weeks of age, all monkeys were housed in standard conditions. Social behavior, emotional reactivity, and increases in plasma cortisol levels elicited by removing all mothers after weaning were examined at 36 weeks.³⁹ Shortly thereafter monkeys were housed with 2 or 3 animals of the same sex from different rearing conditions. Approximately 4 years later in early adulthood (age range, 3.6-5.9 years) a neuroendocrine challenge was administered to test cortisol-negative feedback regulation of the HPA-axis response to exogenous stimulation by CRF.⁴⁰ Then 5 weeks later hippocampal volumes were determined by MRI.

Magnetic resonance imaging was performed using a 1.5-T (General Electric Medical Systems, Milwaukee, Wis) system. Monkeys were scanned under anesthesia induced by a subcutaneous injection of a combination of 20 mg/kg of ketamine hydrochloride, 4 mg/kg of xylazine hydrochloride, and 0.04 mg/kg of atropine sulfate. Body temperatures were maintained within the normal range using a cushioned heating pad. Earplugs provided protection from noises generated by the scanner.

The first scan for each monkey was acquired in the sagittal plane with a 2-dimensional sequential spoiled gradient acquistion pulse sequence: repetition time, 18 milliseconds; echo time, 4 milliseconds; flip angle, 30°; 1 signal averaged; acquisition matrix, 256 × 128 pixels; voxel size, 0.5 × 1.0 × 4.0 mm; and slice thickness, 4 mm. This initial localizer scan was used to standardize head tilt and rotation by assuring that 2 external markers (vitamin E capsules in the meatus of each ear) were aligned in both the coronal and axial planes. The head was repositioned as required, and another sagittal localizer scan was performed. Head pitch was then standardized against the midsagittal image, with the final scan acquired in the coronal plane. The final scan used for hippocampal measurements (Figure 1) was a 3-dimensional inversion recovery prepared fast spoiled gradient acquistion pulse sequence: repetition time, 12 milliseconds; echo time, 3 milliseconds, inversion time, 300 milliseconds; flip angle, 15°; 4 signals averaged; acquistion matrix, 256 × 224 pixels; voxel size, 0.31 × 0.36 × 1.00 mm; and slice thickness,1 mm.

Image processing was performed offline with ANALYZE software (Biomedical Imaging Resource; Mayo Foundation, Rochester, Minn) as previously described for human studies of hippocampal volumes.² To minimize interscan variation a Histogram Match function in ANALYZE was used to normalize gray-scale pixel values for each brain against a single standard. A trained human rater blind to each monkey's identity then measured hippocampal volumes on each brain side.

Stereological methods were used with ANALYZE software to generate unbiased estimates of hippocampal volumes. Sampling parameters were set to yield at least 150 "hits" per measurement, a number previously shown to generate reliable determinations.⁴¹ For sampling purposes, a rigid grid was superimposed on each brain image with grid placement randomly determined by ANALYZE. All grid points falling directly on hippocampal gray matter were identified by the trained human rater. From these determinations ANALYZE generated an unbiased estimate of hippocampal volumes based on the Cavelieri Principle.

Rules for identifying hippocampal volumes were adapted from human protocols.^{2 ,42} The most posterior coronal slice for volumetry was identified when gray matter hippocampus first appeared adjacent to the trigone of the lateral ventricle. Hippocampal gray matter in all coronal slices anterior to this location was bordered superiorly by the fornix-fibria white matter junction, inferiorly by parahippocampal gyrus white matter, medially by subarchnoid spaces of the ambient cistern, and laterally by the cerebrospinal fluid–filled lateral ventricle or temporal horn white matter. The most anterior coronal slice used for volumetry fell at the head of the hippocampus medial to the amygdala in the coronal plane. One or two 1-mm slices anterior to this location were excluded from determinations of hippocampal volumes due to the lack of reliable boundaries for distinguishing amygdala from hippocampus.

To adjust for variation in overall brain size, brain volumes were defined and subsequently measured as all gray and white matter in both hemispheres, including the midbrain superior to the pons. The superior border of the pons was choosen for demarcation because it is easily recognized on MRIs of brain.² Based on the measurements of 2 trained raters independently scoring the same 13 monkey brains, interrater reliabilities expressed as intraclass correlations ranged from 0.90 to 0.97 (left hippocampus, 0.94; right hippocampus, 0.90; and overall brain size, 0.97).

Sex, paternity (offspring grouped by father), and rearing condition main effects for adult hippocampal volumes were examined with repeated-measures analysis of variance (ANOVA) using least squares estimates from general linear models (Systat, Evanston, Ill). Sex, paternity, and rearing condition were between-subjects factors, and hippocampal volume on each brain side was considered the within-subjects factor. Paternity was not analyzed as a random factor because the error terms for sex and rearing condition could not be generated from appropriate interactions in the unbalanced factorial design.³⁹ This did not influence the analysis of paternity since the same error term is used regardless of whether paternity is random or fixed.⁴³

Quantitative estimates of heritability (h²)were generated from 1-way ANOVAs used to evaluate paternal half-sibling effects.⁴⁴ From separate ANOVAs for each measure of interest, the between-father mean square minus the within-father mean square was divided by 3.25 (average number of offspring per father), multiplied by 4 (paternal half-siblings share, on average, 25% of their genome by common descent), and divided by the total variance. Heritabilities resemble intraclass correlations adjusted for genetic relatedness. Under the null hypothesis of no hereditary effect, within- and between-father components of variance are equivalent, and the resulting F ratios approximate 1. As the between-father component of variance increases relative to within-father variance, F ratios grow larger in the half-sibling analysis and the null hypothesis is rejected. All test statistics were evaluated with 2-tailed probabilities (α<.05), and descriptive statistics are presented as mean ± SD.

Analysis of adult hippocampal volumes revealed a brain side main effect (F_1,24 = 6.78, P = .02). Right hippocampal volumes were 4% larger than hippocampal volumes on the left side of the brain. Neither sex nor rearing condition differences were discerned, and none of the brain side interactions was significant, but the repeated measures ANOVA uncovered a paternity main effect (F_11,24 = 2.47, P = .03).

Certain fathers produced monkeys with large hippocampi, other fathers produced monkeys with smaller hippocampi, and similar hippocampal volumes were found among monkeys that shared a common father (Figure 2). Right and left hippocampal volumes were correlated with one another (Figure 3), and these measures were added together to create a total hippocampal volume score. The estimated proportion of genetic variance, ie, heritability, was 54% for total hippocampal volume size (F_11,27 = 2.58, P = .02). Heritabilities for the right (h² = 49%, F_11,27 = 2.34, P = .04) and left hippocampal volumes (h² = 52%, F_11,27 = 2.45, P = .03) were not significantly different. The distribution of total hippocampal volumes was Gaussian in the sample of 39 monkeys (Figure 4), suggesting contributions from multiple additive genetic effects on hippocampal size.

Following completion of the rearing conditions at 21 weeks of age, all mothers were removed well after weaning at 36 weeks post partum. Plasma cortisol levels in their offspring 1 day later were 146% higher than the preseparation levels assessed with identical procedures (197 ± 9 µg/dL vs 485 ± 27 µg/dL). Significant differences in these relative elevations in cortisol levels were produced by the prior rearing conditions (F_2,24 = 18.11, P≤.001), but these rearing-related differences were not associated with significant differences in hippocampal size (Figure 5).

Paternal half-siblings that responded with larger relative 1-day increases in cortisol levels had smaller adult hippocampal volumes after correcting for sex, rearing condition, and overall brain size (r = −0.58, df = 10, P = .048; Figure 6). Half-sibling group differences in hippocampal volumes were not correlated with differences in overall brain size, and the relation between cortisol and hippocampal volume was absent when the 1-day cortisol measures were analyzed as absolute cortisol concentrations. Average cortisol levels remained higher than baseline 3 and 7 days after the removal of all mothers, but these cortisol levels were not correlated with differences in adult hippocampal size. Half-sibling group differences in hippocampal volumes were also not correlated with subsequent measures of cortisol negative feedback in early adulthood.

In previous studies the subset of monkeys that we separated intermittently prior to weaning differed in assessments of social behavior, emotional reactivity, and cortisol-induced suppression of CRF-stimulated secretion of adrenocorticotropic hormone.^{39 - 40} Rearing-related differences in these monkeys were not found in adult hippocampal volumes. Rhesus monkeys raised in social isolation also fail to show altered hippocampal volumes as determined in vivo by MRI,³² despite striking changes in other brain systems and numerous aspects of behavior.^{17 - 23} Maltreated children with posttraumatic stress disorder do not differ from healthy children in hippocampal volumes,⁴⁵ but show elevated baseline urinary free cortisol and catecholamine concentrations.⁴⁶ Chronic alcohol abuse is known to produce hippocampal atrophy in humans⁴⁷ and rats,⁴⁸ but alcohol abuse was not common in the studies of children with posttraumatic stress disorder, and was absent altogether in monkeys. Early life stress without alcohol abuse may have minimal effects on hippocampal volumes determined in human and nonhuman primates by high-resolution MRI.

Paternal half-sibling group differences were apparent in monkey hippocampal volumes. Certain fathers produced offspring with large hippocampi, other fathers produced offspring with smaller hippocampi, and similar adult hippocampal volumes were found in monkeys that shared a common father. Since paternal half-siblings were raised apart from one another by different mothers in the absence of fathers, phenotypic similarities within half-sibling groups cannot be attributed to shared family environments. Based on a standard half-sibling analysis the estimated heritability was 54% for total hippocampal size. This estimate is not inflated by including in the sample 9 maternal half-sibling pairs.³⁹

High heritabilities for overall brain size have been reported in humans^{49 - 51} and rhesus macaque monkeys,⁵² but little is known about the genetics of variation in regional brain morphologies. Neuroimaging research has identified in human twins a genetic basis for differences in surface features of cerebral cortex,^{49 ,53} corpus callosum size,^{51 ,54} and volumetric measures of other subcortical structures.⁵⁰ Our findings support preliminary evidence in humans⁵⁵ indicating that individual differences in hippocampal volumes are in part determined by genes. The Gaussian distribution of hippocampal volumes in monkeys suggests a polygenic trait, and not the effects of genetic epistasis nor a single major gene. Inherited variation in volumes may reflect heritable differences in hippocampal cell numbers,^{56 - 57} or differences in physiological factors related to in vivo tissue perfusion.

Paternal half-siblings with small adult hippocampal volumes responded to the removal of all mothers after weaning with larger relative 1-day elevations in cortisol levels. In humans with mood and anxiety disorders small hippocampal volumes have been taken as evidence that excessive stress levels of cortisol induce hippocampal volume loss.^{2 ,11 - 12} But in monkeys large rearing-related differences in cortisol levels elicited by removing all mothers after weaning did not produce differences in hippocampal size. An alternative explanation for the observed correlation is that small hippocampal volumes are inherited and predispose toward excessive stress levels of cortisol through diminished neuroendocrine regulation. The hypothesis that genes affect cortisol levels by acting on aspects of hippocampal morphology is consistent with evidence that the hippocampus plays a role in suppressing the HPA-axis stress response.^{58 - 59} Various measures of cortisol negative feedback in monkeys were not, however, correlated with hippocampal size.

A limitation of our finding heritable differences in monkey hippocampal volumes is that heritabilities are specific to the population and circumstance in which they are assessed. Genetically diverse populations in homogeneous environments demonstrate larger heritabilities than do inbred populations in diverse environments.⁴⁴ Very few parents that produced the study cohort shared a common mother or father, but extended family pedigrees could not be determined from monkey breeding colony records. The generality of our findings should therefore be tested in studies of other populations.

The absence of postnatal rearing effects must likewise be considered with caution. Rodent research has convincingly demonstrated that stress or glucocorticoids alone induce altered regulation of hippocampal serotonin receptor expression,⁶⁰ apical dendritic atrophy of CA3 pyramidal cells,⁸ suppression of neurogenesis, and decreased survival of newborn granule cells.¹⁰ Our failure to uncover neuroimaging-based differences does not rule out the possibility that microstructural plasticity occurs in the primate hippocampus.^{28 - 30} There are, in fact, excellent reasons to expect subcellular plasticity in monkeys.

Following completion of the rearing conditions and well after weaning at 36 weeks of age, the monkeys we separated intermittently responded to the removal of all mothers with smaller increases in cortisol levels, fewer distress peep-calls, and more time spent near peers.³⁹ Monkeys from the low-foraging and high-foraging demand conditions did not differ on any of these measures. In early adulthood at 5 years of age, only the intermittently separated monkeys showed signs of enhanced negative feedback regulation of the HPA-axis response to stimulation by exogenous CRF.⁴⁰ These findings parallel studies indicating that in rats brief postnatal intermittent social stress diminishes emotionality and HPA-axis reactivity throughout adolescence and adulthood.^{61 - 62} In rats blunted HPA-axis stress responses are mediated by enhanced negative feedback regulation resulting from increased glucocorticoid receptor expression in adult hippocampus.⁶² Experience-dependent augmentation of glucocorticoid receptor densities might likewise account for enhanced negative feedback described elsewhere for the intermittently separated monkeys.⁴⁰

A final aspect of this study that warrants comment concerns the lack of long-lasting foraging demand effects on squirrel monkey brain and behavior. Squirrel monkey mothers in the high-foraging demand condition consistently exhibit increased cortisol levels relative to mothers in the low-foraging demand condition where food is easy to find.³⁷ High-foraging demand condition mothers stop carrying infants earlier, but continue to demonstrate otherwise normal nursing patterns.³⁷ By selectively accelerating certain aspects of development, squirrel monkey mothers apparently spare their offspring from abnormal outcomes previously reported for bonnet macaque monkeys raised by mothers under variable-foraging demands.^{25 - 27} The high-foraging demand condition does not subsequently alter social or emotional aspects of behavior, HPA-axis stress physiology, or adult hippocampal volumes.

Accepted for publication June 26, 2001.

This work was supported in part by the Nancy Pritzker Network, New York, NY, and Public Health Service grant MH47573 from the National Institute of Mental Health, Bethesda, Md.

We thank Blake Mobley and Nadia Sachs for assistance with MRI processing, and Robert Sapolsky, PhD, and Bruce McEwen, PhD, for thoughtful comments on this research.

Corresponding author: David M. Lyons, PhD, Department of Psychiatry and Behavioral Science, 1201 Welch Rd, Medical School Laboratory Surge Bldg, Room P104–Mail Code 5485, Stanford University School of Medicine, Stanford, CA 94305-5485 (e-mail: dmlyons@stanford.edu).