IntroductionDuring lactation, mammalian mothers provide extensive behavioral care to and physiological investment in their offspring, influencing infant developmental trajectories. This maternal care provides information about the world into which the infant is born; milk that mothers supply to their infants is one mechanism of information transfer (German, Dillard, & Ward, 2002; Bernt & Walker, 1999). It is well established across mammals that the energetic value of milk is positively associated with infant growth and development (reviewed in Hinde et al. 2009). Whether or not the composition of maternal milk directly influences infant behavior is less understood (Hinde & Capitanio, 2010). Milk contains numerous non-nutritive bioactive components, including hormones (Akers, 2002; Donovan & Odle, 1994; Rodriguez-Palmero, Koletzko, Kunz, & Jensen, 1999; Schwalm & Tucker, 1978), that may have important effects on the behavioral development of infants. One class of biologically active components in milk is glucocorticoids (corticosterone in rodents; cortisol in humans and nonhuman primates), which have the potential to influence infant behavior through calibration of the hypothalamic-pituitary-adrenal (HPA) axis of infants. Glucocorticoids (GCs) in milk are reflective of serum blood GC concentrations in response to physical, pharmacological, and psychological challenges (Bremel & Gangwer, 1978; Groer, Humenick, & Hill, 1994; Pearlman, 1983), with glucocorticoid concentrations equilibrating rapidly between plasma and milk (Fox, Butler, Everett, & Natzke, 1981; Angelucci, Patacchioli, Scaccianoce, Di Sciullo, Cardillo, & Maccari, 1985; Termeulen, Butler, & Natzke, 1981). In this way, milk may serve as a salient biochemical signal of environmental conditions from mother to infant, entraining infant physiological functioning and patterns of responding.
Glucocorticoid concentrations in milk may affect behavior through effects on broad dispositional characteristics. Temperament is an individual’s consistent pattern of responsiveness that is thought to be partially heritable, biologically based, present early in life, and stable over time (Allport, 1937). The HPA axis has long been identified as an important physiological underpinning of temperament, with both regulation and reactivity of the system providing important information about the relationship between physiological and psychological arousal (Levine, Haltmeyer, Karas, & Denenberg, 1967; Mcewen, 2001; Stansbury & Gunnar, 1994; Suomi, 1991; Mendoza & Mason, 1989; Capitanio, Mendoza, & Bentson, 2004). For example, in comparison to socially inhibited children, uninhibited children have lower morning basal cortisol concentrations (Kagan, Reznick, & Snidman, 1988) and greater lability in HPA axis responsivity to social stress (Gunnar, Tout, De Haan, Pierce, & Stansbury, 1997). Characteristic dispositions to respond show consistency from childhood into adulthood (Gest, 1997) and glucocorticoids contribute to these lasting patterns by strengthening or weakening neural pathways that become canalized over time (Vyas, Mitra, Shankaranarayana Rao, & Chattarji, 2002; Korte, 2001; Casolini, Cigliana, Alema, Ruggieri, Angelucci, & Catalani, 1997). These neural pathways influence the likelihood of future behavioral responses resulting in individual differences in temperament (Gunnar & Quevedo, 2006). Neural changes are influenced by maternal indicators of environmental conditions, such as the quality of maternal behavioral care (Fish, Shahrokh, Bagot, Caldji, Bredy, Szyf, & Meaney, 2004), but may also be influenced through physiological signaling during lactation. To date, however, little is known about how maternal physiological investment during lactation shapes infant behavior and temperament (Hinde & Capitanio, 2010).
Glucocorticoids have been indentified in the milk of several species, including humans (for review, see Alexandrova & Macho, 1983; humans, Kulski & Hartmann, 1981; Patacchioli, Cigliana, Cilumbriello, Perrone, Capri, Alema, Zichella, & Angelucci, 1992; dairy cattle, Tucker & Schwalm, 1977; rodents, Angelucci, Patacchioli, Chierichetti, & Laureti, 1983). Infant mammals have corticosteroid (mineralocorticoid and glucocorticoid) receptors in their intestinal tract which provide a potential pathway for maternal GCs to influence infant physiology following milk ingestion (for review see Pacha, 2000). Glucocorticoid receptors (GRs) are found at higher densities in the intestines during infancy, but intestinal GRs decrease to adult levels after weaning (Henning & Kretchmer, 1973). This is especially striking because GR density increases in other body tissues during development (Henning, Ballard, & Kretchmer, 1975).
Bioactive maternal glucocorticoids are transmitted to mammalian infants via milk ingestion and the effects show sex biases. Angelucci and colleagues delivered radiolabeled corticosterone to adult female rats by oral administration in drinking water, which resulted in increased corticosterone concentrations in both maternal plasma and milk, but within the range of naturally occurring stress-induced cortisol concentrations (Angelucci et al., 1983). After infants ingested the milk of these mothers, labeled corticosterone was found in their gastric contents, plasma, and brains. Infants also displayed higher circulating morning plasma concentrations of corticosterone (Angelucci et al., 1983). Ingesting elevated cortisol concentrations in infancy caused lasting changes to offspring: adult females, whose mothers were administered exogenous corticosterone, had elevated corticosterone concentrations in comparison to female controls, whereas adult males had lower plasma corticosterone concentrations in comparison to male controls (Angelucci et al., 1983).
Glucocorticoids in milk are also implicated in infant behavioral outcomes in both rats and humans. Offspring whose mothers were administered exogenous corticosterone, described above, displayed fewer behavioral indicators of anxiety in response to stress than did controls. This effect was present during infancy, as well as after weaning into adulthood (Angelucci et al., 1985; Catalani, Casolini, Scaccianoce, Patacchioli, Spinozzi, & Angelucci, 2000; Catalani, Casolini, Cigliana, Scaccianoce, Consoli, Cinque, Zuena, & Angelucci, 2002). In humans, infant fear behavior is positively correlated with mother’s plasma cortisol concentrations but only in infants that were breast-fed (Glynn, Davis, Schetter, Chicz-Demet, Hobel, & Sandman, 2007) suggesting that maternal physiology, rather than behavior, is responsible for this effect. To date, however, no studies have directly investigated cortisol concentrations in milk and infant temperament in human or non-human primates.
To better understand the relationship between cortisol in milk and infant temperament we investigated these measures in a non-human primate: the rhesus monkey (Macaca mulatta). Rhesus macaques display substantial individual variation in behavior and temperament (Suomi & Ripp, 1983; Capitanio, Mason, Mendoza, Del Rosso, & Roberts, 2006). Additionally the regulation of their HPA axis has been well characterized (Levine & Wiener, 1988; Clarke, 1993; Gorman, Mathew, & Coplan, 2002) and HPA regulation and responsivity to stress have been found to be affected by early experience (Capitanio, Mendoza, Mason, & Maninger, 2005). The aims of the present study were to determine the correlation between cortisol concentrations in maternal plasma and milk, and the association between milk cortisol concentrations and infant temperament, as assessed in response to a stressful experience.
MethodsSubjectsSubjects were 44 adult female rhesus monkeys (Macaca mulatta) and their infants. Twenty mothers had male offspring (6 primiparous, 14 multiparous) and 24 had female offspring (14 primiparous, 10 multiparous). Infants were between 91–124 days old (mean age=110 days) at the time of the assessment (see below). Dominance rank did not differ between mothers of male and female offspring (Mann-Whitney U = 211.5, p = 0.468) or between primiparous and multiparous mothers (Mann-Whitney U = 223.0, p = 0.665). Prior to assessment, mothers and infants were housed outdoors in half-acre enclosures consisting of 100–150 animals of mixed age- and sex-classes of close kin, distant kin, and non-kin at the California National Primate Research Center (CNPRC). A commercial diet (Outdoor Monkey Lab Diet, PMI Nutrition Int’l, Brentwood, Missouri) was provided twice daily, fruit and vegetable supplements were provided twice weekly, and water was available ad libitum.
Subject RelocationMothers and their offspring were captured from their homecage, separated from one another, and relocated to separate novel environments (between 0800–0900h) for a 25-hour period. Infants were weighed immediately after separation, and were tested in cohorts of five to eight animals. Housing for mothers and for infants was in standard laboratory cages (60cm X 65cm X 79cm, Lab Products, Inc., Maywood, NJ), with all members of a given cohort housed in the same room throughout the 25-hr period. Mothers and infants were housed in separate rooms with no visual, auditory, or olfactory contact possible. At the conclusion of the 25-hour testing period, infants and mothers were reunited in the mother’s holding cage and were housed together for one hour before being returned to their outdoor enclosures.
Infant Biobehavioral AssessmentAll 44 infant subjects were part of an ongoing biobehavioral assessment program at the CNPRC described in detail elsewhere (Capitanio et al., 2005; Golub, Hogrefe, Widaman, & Capitanio, 2009; Hinde & Capitanio, 2010). Over the 25-hr. testing period, behavioral data were collected in a variety of standardized testing situations, including behavioral observations in the novel holding cage, subjects’ recognition memory, and subjects’ responsivity to social stimuli, graded conditions of challenge, and novel objects. Blood samples (1.0 ml) were also collected from unanesthetized infants 2 hours after separation and relocation via femoral venipuncture.
Temperament RatingsInfant temperament ratings were conducted in order to collect an overall “thumbnail” portrait of the animal’s functioning during the entire Infant Biobehavioral Assessment period. At the conclusion of the 25-hr. testing period, observers who performed the testing rated each infant on 16 trait adjectives describing characteristics of temperament (see Table 3 in Golub et al., 2009; Hinde & Capitanio, 2010). Ratings were made using a seven point Likert scale, with a score of 1 reflecting a total absence of the observation of behaviors related to the characteristic and a score of 7 indicating the observation of an extremely large amount of the behaviors reflecting that characteristic. Agreement between independent observers (a difference of no more than 1 scale point) for each trait was significantly greater than chance using chi-square (p<0.000001). Ratings on each adjective were z-scored across all subjects within a given birth year. Exploratory and confirmatory factor analyses of the data from the full sample between 2001 and 2005 (n=1284, described in Golub et al., 2009) revealed four factors (named for the trait adjective with the highest positive loading): Confident (active, bold, confident, curious, playful), Gentle (calm, curious, flexible, gentle), Vigilant (vigilant, not depressed, not tense, not timid), and Nervous (fearful, nervous, timid, not calm, not confident). The trait adjectives preceded by the word “not” reflect a negative loading in the factor analysis. A composite score for each factor was created by adding the z-scores of each trait rating adjective (or the reverse-coded z-score for negative loadings). These scores were then standardized within each factor scale and z-scores were created. Cronbach’s alpha (scale reliability) values, based upon the full sample of 347 animals assessed this year, were 0.91 (Confident), 0.78 (Gentle), 0.87 (Vigilant), and 0.61 (Nervous).
Sample CollectionAfter 3.5–4 hours of milk accumulation, mothers were sedated with ketamine hydrochloride intramuscularly (5–10 mg/kg). Blood was collected on a randomly selected subset of subjects (n=16: 6 with female offspring, 10 with male offspring) via femoral venipuncture into pre-chilled EDTA tubes which were immediately placed in ice. Subjects were then administered exogenous oxytocin (2 IU/kg) intramuscularly for myoepithelial cell contraction and milk let down. Milk was collected by gentle hand stripping of the nipple. To minimize sampling bias (Oftedal, 1984) the mammary was fully evacuated which occurred within 10–15 minutes for all subjects as described elsewhere (Hinde, Power, & Oftedal, 2009).
Cortisol AssaysBlood samples were centrifuged at 3000 RPM for 10 min, and the plasma fraction was removed and frozen at −80°C prior to being assayed for cortisol concentration by coated-tube RIA (Diagnostic Products Corp., Los Angeles, CA). The interassay coefficient of variation (CV) was 7.3% and intra-assay CV was 5.1%.
Milk was briefly vortexed, aliquoted, and frozen at −80°C until thawed for analysis. Once thawed, it was centrifuged at 3000 RPM for 10 min., consistent with methods used in other species (Butler & Des Bordes, 1980; Agrimonti, Frairia, Fornaro, Torta, Borretta, Trapani, Bertino, & Angeli, 1982; Groer, Davis, Casey, Short, Smith, & Groer, 2005; Spencer, Boyd, Cabrera, & Allee, 2003). Centrifugation resulted in the separation of aqueous component from suspended particles in milk and a floating lipid layer. Concentrations of cortisol in the aqueous component were estimated in duplicate using commercial RIA kits (Siemens Medical Solutions Diagnostics, Los Angeles, CA). During assay development we determined that the partial removal of lipids and suspended particles through centrifugation resulted in reduced cortisol concentrations that were 81% of whole milk values, consistent with results in cows (Schwalm & Tucker, 1978), and that cortisol concentrations in milk were approximately 10% of plasma values. Assay procedures were modified as follows: 1) standards were diluted to concentrations ranging from 2.76 to 345 nmol/L; 2) sample volume was increased to 200 μl, and 3) incubation times were extended to 3 h. Serial dilution of samples indicated that the modified assay displays a linearity of .98 and a least detectable dose of 1.3854 nmol/L. All samples were run in one assay with an intra-assay CV of 2.54%.
Milk Fat and Protein AssaysComposition analyses of proximate milk constituents were conducted in the Nutrition Laboratory at the Smithsonian National Zoological Park in Washington DC using standard methods described elsewhere (Hinde et al., 2009; Oftedal & Iverson, 1995; Milligan, Gibson, Williams, & Power, 2008; Power, Verona, Ruiz-Miranda, & Oftedal, 2008). To determine fat, total lipids were measured by sequential extractions with ethanol, diethyl ether and petroleum ether by a micro modification of the Rose-Gottleib procedure. The amount of nitrogen in each sample was determined using a CHN elemental gas analyzer (Model 2400, PerkinElmer, Norwalk, CT) using a 2-s oxygen burst to promote complete combustion. Crude protein was estimated as 6.38*nitrogen. These methods have been validated at the SNZP Nutrition Laboratory using both fresh cow milk and powdered cow milk from the National Institute of Standards and Technology (Hinde et al., 2009).
Data analysisBivariate correlations were conducted to compare cortisol concentrations in maternal milk to cortisol concentrations in maternal plasma, maternal milk components, offspring plasma, and offspring temperament factor scores. A Bonferroni correction for multiple comparisons was applied to the correlations between cortisol concentrations in milk and offspring temperament factor scores in order minimize the probability of Type I error, α′ = 0.00625 = α/k, where k is the number of tests (8: 4 each for males and females). The significance of the difference between correlations was computed as described in Blalock (1972). Partial correlations were also conducted to assess the unique contribution of cortisol in maternal milk with infant temperament, controlling for other components in milk. A t-test was performed to compare the cortisol concentrations in the milk and plasma of mothers of sons and daughters. There were no differences based on maternal parity in maternal cortisol concentrations in milk or infant temperament factor scores, therefore groups were combined in the final analyses.
DiscussionStress-induced cortisol concentrations in milk were correlated with maternal plasma cortisol concentrations and were in a range comparable to those found in saliva, indicating that cortisol found in milk is likely free rather than bound to CBG or albumin (Umeda, Hiramatsu, Iwaoka, Shimada, Miura, & Sato, 1981). Additionally, naturally occurring variation of endogenous glucocorticoid concentrations in maternal milk in response to stress was associated with one dimension of offspring temperament, and showed sex-specific effects. This suggests that the transmission of biochemical markers from mother to infant through the consumption of milk may not only influence infant physiology (Angelucci et al., 1983), but may further influence characteristics of infant temperament with important, and differential, consequences for sons and daughters.
For sons, Confident factor scores were significantly and positively correlated with maternal cortisol concentrations in milk, whereas there were no correlations between daughters’ temperament factor scores and cortisol in maternal milk. High scores on the Confident factor reflect an infant that has a bold, active, curious and playful approach during a stressful experience. In rodents there are sex-dependent effects on HPA axis regulation in infants exposed to increased maternal cortisol concentrations through milk (Catalani et al., 2000; Catalani et al., 2002), which may underlie the differences in behavioral responsivity observed here. Males may be more sensitive than females to experimentally induced variability of their mother’s HPA axis (Parker, Buckmaster, Sundlass, Schatzberg, & Lyons, 2006). Males exposed to pharmacologically elevated GC concentrations ingested through milk show an increase in both mineralocorticoid (MR) and glucocorticoid receptors (GR) in the hippocampus in comparison to controls (Catalani et al., 2000; Casolini, Domenici, Cinque, Alema, Chiodi, Galluzzo, Musumeci, Mairesse, Zuena, Matteucci, Marano, Maccari, Nicoletti, & Catalani, 2007; Casolini et al., 1997), but no differences in hippocampal MR or GR densities have been observed in females exposed to increased cortisol through milk (Catalani et al., 2002). In this way exposure to cortisol concentrations in milk may lead to sex-dependent effects that contribute to lasting changes in physiological functioning by entraining neuroendocrine regulation in males but not females. This post-natal vulnerability to maternal influence may provide an opportunity for mothers to fine-tune the quality of sons during development.
Previous research has demonstrated associations between early life exposure to elevated glucocorticoid concentrations and characteristics of confidence, consistent with the results presented here. Infant monkeys exposed to brief intermittent separation stress, and the resultant periodic activation of the HPA axis, are more curious and exploratory in novel situations later in life (Parker, Buckmaster, Schatzberg, & Lyons, 2004). Similarly, rodent pups exposed to elevated GC concentrations in milk, increased pharmacologically to stress-induced levels, exhibit fewer behavioral indicators of anxiety in response to novelty from infancy to adulthood than those that were not exposed to increased GCs (Catalani et al., 2002; Catalani et al., 2000). Our results suggest that individual differences in maternal milk GC concentrations in response to stress are associated with individual differences in sons’ behavioral responsivity to a novel situation, with greater GC concentrations in milk corresponding to a son that is more confident. Maternal indicators of environmental condition may play an important role in this trait specifically due to its significance in the life-history of males. Male rhesus macaques emigrate from their natal group (Pusey & Packer, 1986) and attain rank in a new social group based partially on personality characteristics, whereas female rhesus monkeys remain in their natal group and inherit rank from their mother (Walters & Seyfarth, 1986). Indeed, adolescent boldness and confidence positively predict adult rank attainment in male vervet monkeys (Fairbanks, Jorgensen, Huff, Karin, Yung-Yu, & Mann, 2004). For this reason the Confident temperament factor may be particularly sensitive to maternal hormones in a way that the other temperament factors - Vigilant, Gentle, and Nervous –are not.
In conjunction with the sex-biased effects of maternal cortisol on aspects of infant temperament, mothers of males had higher cortisol concentrations in milk than did mothers of females, but determining causal explanations of this difference remains difficult. Among polygynous mammals, mothers in better condition are predicted to bias investment toward sons because of their greater potential reproductive output compared to daughters (Trivers & Willard, 1973; reviewed in rhesus macaques in Bercovitch, Widdig, & Nürnberg, 2000; Hinde, 2009). Maternal condition has been assessed through the examination of glucocorticoid concentrations (Love, Chin, Wynne-Edwards, & Williams, 2005) as glucocorticoids play a prominent role in energy balance (Dallman, Strack, Akana, Bradbury, Hanson, Scribner, & Smith, 1993; Munck & Naray-Fejes-Toth, 1994). In our population of rhesus macaques, however, mothers of sons and daughters do not differ in their condition and presumably pay the same energetic costs to rear offspring; they are equally likely to get pregnant and produce a surviving infant in the subsequent birth season (Hinde, 2009). Further, mothers of sons and daughters in the present study did not differ in plasma cortisol concentrations, suggesting that sex-biased differences in cortisol concentrations in milk are not due to differences in maternal condition and may be specific to mammary function or offspring development. We also found that cortisol in milk was positively correlated with energetic components in milk, specifically protein and fat concentrations, which are higher in the milk produced for sons (Hinde, 2007). Energetic aspects of milk have been shown to predict infant behavior (Hinde & Capitanio, 2010); when controlling for these factors, however, the association between cortisol concentration in maternal milk and Confident factor scores in sons remained. Evidence from several species indicates that male infants assimilate and metabolize maternal milk energy differently than do female infants (see Hinde, 2009 for review) and cortisol may be an important factor in calibrating metabolic processes (German et al., 2002; Rodriguez-Palmero et al., 1999).
This study was part of a larger project, and consequently, there are several limitations in experimental design and control. The measures of cortisol reported here (for both mothers and infants) were collected at a single time point and do not reflect basal levels or baseline concentrations; these measures reflect cortisol concentrations after activation of the HPA axis in response to separation and relocation stress. However, the HPA axis shows trait-like consistency (Stansbury & Gunnar, 1994); therefore maternal cortisol concentrations in milk are also likely to be trait-like and show consistency across a variety of physiological and psychological stressors. Due to testing parameters, samples were collected at different times for infants and mothers (two and four hours post-separation, respectively); the lack of relationship between maternal cortisol concentrations in milk and offspring plasma cortisol concentrations may be due, in part, to this. Moreover, mothers and infants were separated at the time of testing, so transmission of milk from mother to offspring was not possible. Finally, correlations between cortisol in milk and dimensions of infant temperament may reflect underlying similarities between mothers and sons in behavioral and physiological reactivity, perhaps as a result of sons’ direct inheritance of x-linked genes associated with HPA axis functioning from their mother (e.g. Monoamine Oxidase A gene (Jabbi, Korf, Kema, Hartman, Van Der Pompe, Minderaa, Ormel, & Den Boer, 2007)), rather than the direct transmission of maternal signals to offspring through milk from their mother. This may be addressed through longitudinal and cross-fostering studies that are better able to tease apart the influence of genetics and the environment.
The data presented here represent an important first step in understanding the role of maternal glucocorticoids in milk and the potential consequences for one dimension of infant temperament. Our results suggest that confidence in males may be more strongly associated with their mother’s physiological functioning than is the confidence of daughters. This relationship provides a potential physiological mechanism through which mothers may fine-tune their sons’ development during early life. Future studies should further investigate individual differences in concentrations of cortisol in milk over lactation, its relationship to regulation of the HPA axis in offspring, and the consequences for offspring temperament.