INTRODUCTION

Top Abstract Introduction Methods Results Discussion References

AN INCREASING NUMBER of investigators in biomedical sciences have become interested in identifying the extent and causes of individual differences in cellular, physiological, pathophysiological, and social endpoints. Such insight is critical for understanding variability in susceptibility to a disease and its course as well as to phenomena such as successful social adaptation and aging. Typically, plasma hormones and metabolic factors have been investigated to gain an understanding of relevant processes. Many studies have focused on nonhuman primates, examining variables such as individual differences in dominance rank. Since 1978, R. M. Sapolsky has examined the relationship between multiple physiological endpoints and rank in a population of wild baboons in the Serengeti ecosystem in East Africa. Marked rank-related differences have been observed in adrenocortical and gonadal function, autonomic and immune profiles, cholesterol metabolism, and cardiovascular responsivity (reviewed in Ref. 24). These findings show that in a stable hierarchy, it is the socially subordinate animals that have the less "adaptive" profiles, showing indexes of chronic stress and being more prone toward stress-related pathologies.

In the present report, we examine the relationship in this population between social rank and circulating concentrations of insulin-like growth factor (IGF)-I, a potent growth-promoting and anabolic hormone with major roles in cellular growth and differentiation, protein metabolism and muscle physiology, wound healing, erythropoiesis, and immune stimulation (11, 28). To our knowledge, this is the first such report examining the relationship between dominance status in a social species and IGF-I plasma levels. We observe that social subordinance is associated with suppressed IGF-I concentrations.

	METHODS

Top Abstract Introduction Methods Results Discussion References

Study site and subjects. Research was conducted by R. M. Sapolsky in the Masai Mara National Reserve in Kenya. The study subjects were the males of a troop of wild baboons (Papio anubis) that was originally habituated to observation in 1976. All subjects had already carried out their adolescent transfer into the troop from their natal troop from 1 mo to 9 yr before observation. Males that had undergone a major growth spurt and emergence of secondary sexual characteristics 2 yr previously were termed "subadult." Males that had fully developed secondary sex characteristics and/or had gained no more than 2 kg body wt from the previous year were termed "adults." "Older adults" were identified in two ways. Animals that had been members of the troop as adults and that were now at least 4 yr past that point were considered to be older adults; no individuals in the troop fit that criterion. In addition, aged animals occasionally transfer into troops and are identified as aged because of the combination of their poor coat quality, atrophied musculature, and broken and worn dentition; all four older adults in the study were of this category. Males that had reached adult status at least 4 yr earlier were termed older adults; no such animals were in this troop during the study. A total of 37 different subjects was studied over the years, as individuals transferred into or out of the troop or animals died.

Behavioral data. Data from this study were collected from July to November from 1980 to 1986. The principal method of data collection was focal sampling. Ad libitum observations were made for contextual information only. In addition, births, deaths, injuries, and female reproductive status were recorded daily. Each male under study was sampled for 20 min at a time, from one to three times per day. Equal numbers of samples were collected for all individuals during each annual observation season; the number of such samples ranged from 25 to 45 per season. The initial subject to be observed each day was selected randomly; at the end of that sample, the investigator moved in a randomly chosen direction to the nearest male not yet sampled that day. Later in each season, this pattern was modified to assure that the distribution of sample times throughout the day did not differ among individuals. During observation, all occurrences of social interactions, feedings, and autogrooming were recorded.

Rank in dominance hierarchy. The rank of each individual for each season was determined with the use of approach-avoidance criteria, which were defined to indicate active avoidance on the part of the loser rather than overt aggression on the part of the winner (see Ref. 25). Avoidances, supplants, male-male mounts, and presentations were included in this category. Mounts that occurred in an affiliated context, e.g., in reinforcing a coalition, were excluded. Avoidances and supplants occurred when the losing individual moved >2 m away when the winner approached within 3 m. The interaction was termed a supplant if, after an avoidance interaction, the winner made use of some resource, e.g., a feeding, resting, or grooming site, previously used by the loser. The male that won in the majority of unambiguous interactions was termed the dominant individual in a dyadic relationship. In no case did ambiguous interactions account for >l0% of the total interactions in a dyad. Hierarchies were constructed in which rank was based on the number of males dominated by each male.

Blood collection. Subjects were anesthetized with phencyclidine hydrochloride (~2 mg/kg body wt) from a propelled syringe fired from a blowgun at a distance of 10 m. Phencyclidine is a dissociative anesthetic that does not directly affect steroid secretion in olive baboons (23). Darting occurred between 0700 and 1000. Animals were darted only if no observable individuals were looking at the investigator. Unconscious animals were then immediately removed out of view of other troop members. Animals were not darted if they were ill, had been recently injured, or had been observed in a fight or consortship that morning. Subjects became unconscious 7-12 min after darting, and a 5-ml blood sample was collected from the femoral vein as soon as it could be done safely. Over the different years, the first sample was obtained at an average of 9.2 ± 1.3 min (mean ± SE). Subjects were allowed to recover (typically after an unrelated 4- to 8-h experiment involving the taking of half of a dozen blood samples) and were released near their troop when fully conscious the next day. No loss of habituation to the observer occurs with this protocol. Although anesthetization by darting results in a number of classical endocrine stress responses (e.g., enhanced glucocorticoid secretion) (23), controlled laboratory studies with chair-restrained catheterized baboons indicate that the stressor in this protocol is not the intramuscular injection at time 0 but rather the disorientation and loss of motor controls 5-10 min later, just before unconsciousness. This is insufficient time for the steroid concentrations in the initial blood sample to be changed by the stressor (23). Thus the initial sample is termed "basal." This is also insufficient time for serum IGF levels to change, given that >95% of serum IGF is in the complexed form, which has a half-life of ~18 h (28).

Radioimmunoassays of blood samples. Blood was centrifuged at the study site; serums were separated and frozen in liquid nitrogen. IGF-I and -II were analyzed by radioimmunoassay. The radioimmunoassay for IGF-I was a modification of the previously described method (29). The baboon plasma samples were extracted by acid-ethanol cryoprecipitation to remove IGF-binding proteins and assayed with a polyclonal rabbit antibody (861/5) (5). Acid-ethanol cryoprecipitation was validated by comparison with aliquots chromatographed on Sephadex G-50 in 1 M acetic acid to separate the interfering IGF-binding proteins. The inter- and intra-assay coefficients of variation were 8 and 6%, respectively. Cross-reactivity with IGF-II was <2%. IGF-II was assayed as previously described with the use of the same acid-ethanol cryoprecipitated plasma samples (29). The inter- and intra-assay coefficients of variation were 8 and 6%, respectively. Cross-reactivity with IGF-I was <2%. There was no trend toward IGF-I or -II plasma concentrations declining with increasing length of time frozen (data not shown). In addition, all samples were obtained from subjects by 30 min after darting; the stress of such darting and anesthetization did not alter IGF-I or -II concentrations, as there was no significant change in the concentrations of either as a function of lag time between darting and sample collection (data not shown).

Western ligand blot analysis of IGF-binding proteins. Western ligand blot analysis for IGF-binding proteins was carried out in accordance with the method of Hossenlopp et al. (9a). Three parts of sample were mixed with one part sample buffer and electrophoresed under nonreducing conditions on a 12.5% sodium dodecyl sulfate-polyacrylamide gel in tris(hydroxymethyl)aminomethane (Tris) · HCl, pH 8.3. The separated proteins were blotted onto a nitrocellulose membrane, washed in 3% NP-40 Tris-buffered saline (TBS), and then blocked in 1% bovine fraction V in TBS for 1 h. The nitrocellulose membrane was enclosed in a Seal-O-Meal bag, and 1 × 10⁶ counts per minute of ¹²⁵I-labeled IGF-I in 10 ml blocking buffer were added. The bag was incubated on a rotary shaker for 2 h at room temperature. The membrane was subjected to three washes of TBS containing 0.1% Tween 20, followed by three washes in TBS alone. After drying, the membrane was exposed to Kodak Biomax MS film for 1-3 days at 80°C before development of the film. Gel bands were quantified densitometrically and expressed relative to a 28.7-kDa unglycosylated recombinant human IGF-binding protein run as a marker.

Data analysis and statistics. Data are presented as means ± SE, and statistical tests were those indicated. A statistical problem arises in these studies, owing to the lack of control by the investigators over the departure of study subjects from the troop. In some cases, animals were troop members for only a single year and thus received an annual rank only once and were sampled for hormone values only once. In other cases, males were troop members for multiple years and, because male rank changes among olive baboons over time, had blood samples taken at different ranks. Most appropriately then, data from the single-year and multiple-year animals should be analyzed separately, with repeated-measure tests used on the latter animals, grouped separately according to the number of repeated measures. However, this splitting of the already small population into those smaller categories invariably precludes statistical significance for any category. Thus the most statistically conservative solution that has evolved in these studies has been to treat each annual rank and blood sample for an animal as independent of any subsequent ones.

	RESULTS

Top Abstract Introduction Methods Results Discussion References

There was a highly significant relationship between rank and IGF-I profiles, in that social subordinance was associated with marked suppression of IGF-I concentrations (Fig. 1; P < 0.007). A particularly striking example of this is shown in the longitudinal data from the sole male that was a group member throughout the entire study period (Fig. 2).

View larger version (17K): [in this window] [in a new window]   Fig. 1.   Relationship between dominance rank and circulating insulin-like growth factor I (IGF-I) concentrations () as described by equation: y = 21.3x + 665. There was a significant decrease in IGF-I concentrations with social subordinance (F = 7.812; degrees of freedom = 1/53; P < 0.007; r = 0.35).

View larger version (10K): [in this window] [in a new window]   Fig. 2.   Longitudinal relationship between rank and IGF-I concentrations in 1 individual baboon, male 261. A: rank of male 261 from 1980 through 1986; B: IGF-I concentrations in same individual during that time.

Potentially, this rank effect might have been confounded by age, because there was a significant decline in IGF-I concentrations among older adults, an age group that is disproportionately low ranking in the hierarchy (Table 1). However, a highly significant relationship between low rank and suppression of IGF-I concentrations still occurred after elimination of the data taken from the four older adults [y = 18.7x + 658; F = 5.757; degrees of freedom (df) = 1/49; P < 0.02] and even after restriction of analysis to only adults (i.e., elimination of both older adults and subadults) (y = 21.9x + 663; F = 4.27; df = 1/39; P < 0.05).

In this population of baboons, social subordinance is associated with elevated basal concentrations of glucocorticoids (14 ± 2 µg/dl cortisol in the higher-ranking half of the hierarchy; 22 ± 3 µg/dl in the lower-ranking half). This pattern has been observed repeatedly in this population (reviewed in Ref. 1) and is thought to reflect the physical and psychological stress of everyday life for a subordinate individual. We next investigated whether there might be a relationship between the elevated cortisol concentrations and suppressed IGF-I concentrations of subordinates. However, we observed no such relationship [IGF-I concentration (in ng/ml) = 6.69 × cortisol concentration (in µg/dl) + 660.7; r² = 0.06, not significant]. In addition, there was no relationship between IGF-I concentrations and either basal testosterone concentrations or body weight (Table 2).

In contrast to the pattern seen with IGF-I, the concentrations of neither IGF-II nor of IGF-binding protein varied as a function of social rank (Table 3; Fig. 3).

View larger version (57K): [in this window] [in a new window]   Fig. 3.   Representative Western blot showing IGF-binding proteins running at 43 and 41 kDa with recombinant, unglycosylated human binding protein marker running at 28.7 kDa. Lanes 1-8 represent samples from subjects, with low-ranking animals in odd lanes and high-ranking animals in even lanes; lane 9 contains recombinant marker (M).

	DISCUSSION

Top Abstract Introduction Methods Results Discussion References

We have observed a novel correlation between circulating IGF-I levels and social rank in a wild population of male primates. There was no rank relationship with IGF-II concentrations, which was not surprising, given that its actions are most relevant to fetal development. We also did not observe a relationship between rank and quantity of IGF-binding proteins.

Because of the lack of data regarding IGF-I concentrations in baboons, it is not possible to know whether the correlation is due to elevation in dominant animals or suppression in subordinates. However, the latter is more likely, since there are a variety of abnormalities in levels of other circulating factors in subordinate individuals in many species (reviewed in Ref. 24). A critical point is whether the differences in IGF-I profiles contributed to the rank or whether the reverse holds. Although we cannot answer this, studies manipulating groups of captive primates demonstrate that the endocrine markers of rank are a result, rather than a predictor, of rank (24). A priori, a number of other factors may have given rise to the IGF-I difference that we have observed.

First, in stable dominance hierarchies, subordinance has been associated with elevated basal levels of glucocorticoids in a broad variety of species (reviewed in Ref. 24), reflecting the physical and/or psychological stressors of subordinance. Among these wild primates, subordinates have ~50% higher basal cortisol concentrations than do dominant males. Some investigators have shown that glucocorticoids suppress IGF-I mRNA expression in the liver (the source of circulating IGF-I) (14); however, others have shown that glucocorticoids induce IGF-I in cultured hepatocytes (4) and increase circulating IGF-I concentrations (16), whereas others report no change or a decrease (8, 15, 32). We observed no significant relationship between basal cortisol concentrations and IGF-I levels. (However, it should be noted that there was a mild trend in the direction of higher cortisol concentrations being associated with lower IGF-I concentrations; more importantly, because there was only a single cortisol determination, and adrenocortical secretion can be episodic, the question of a glucocorticoid-IGF-I link remains somewhat open in this study population.)

Second, caloric deprivation suppresses IGF-I concentrations (6, 10, 18, 19). However, in nondrought circumstances, dominant and subordinate baboons in savanna ecosystems have equal levels of food intake, of exercise, and of (extremely low) fat deposition, arguing against nutritional differences mediating this rank difference. Furthermore, there was no correlation between IGF-I concentrations and body weight. As the sole possible dietary mediator, subordinate individuals typically do not have access to meat (which is either predated or scavenged), and decreased protein intake can suppress IGF-I levels (10, 19). However, meat accounts for <1% of the diet, even among the most dominant of baboons, making it unlikely to be an important variable.

Third, the rank relationship is unlikely to reflect age-specific anabolism. Although levels were significantly lower among aged animals, as reported previously (2, 21), the rank relationship still held after exclusion of them. Furthermore, even the youngest subjects were past their adolescent growth spurt, a time of enhanced IGF-I concentrations. Testosterone elevates circulating IGF-I concentrations and is thought to be relevant to the male adolescent growth spurt (9, 27). However, among these subadults and adults, we did not observe a relationship between testosterone and IGF-I concentrations.

Fourth, genetic factors were unlikely to contribute. As noted, rank in male olive baboons changes over time, typically peaking at prime age. The same individuals often shifted between the high- and low-ranking cohorts at different times in this multiyear study (as personified by male 261 in Fig. 2). Thus neither rank nor IGF-I concentrations represent stable genetic traits over the lifetime.

Fifth, differences in the clearance and catabolism of IGF-I from the circulation could account for this rank relationship; we have no data to address this issue, however.

Last, both growth hormone and insulin can increase IGF-I concentrations. Insulin concentrations do not differ by rank among these animals (unpublished data); whether such differences occur in growth hormone levels is not known. The relationship between stress (as a possible marker of subordinance) and growth hormone concentrations in primates is complex, reflecting the duration and intensity of the stressor (22). Were one to observe higher growth hormone levels in dominant males (thus perhaps elevating IGF-I levels), one might expect such males to be leaner, a pattern not observed among these baboons. However, few studies have examined the metabolic consequences of small differences in growth hormone profiles, making such a prediction about body leanness speculative.

Thus a number of likely mediators of the rank-IGF-I relationship can be eliminated, with the most plausible explanation, higher growth hormone concentrations in dominant animals, being untested. The magnitude of the IGF-I-rank effect was considerable. Moreover, because there were no rank differences in levels of IGF-binding proteins, this rank difference in total IGF-I levels should translate into differences in free hormone concentrations as well (28). What are the possible consequences of these differences? Age-matched subjects do not differ in body weight by rank among these animals. Given the leanness of these animals, it is thus unlikely that rank difference in levels of IGF-I would manifest itself in heavier muscle mass in dominant males. As another possibility, the higher IGF-I levels in dominant males might be associated with more rapid wound healing and/or resistance to sepsis (17). There is the tacit assumption among many primatologists that such rank differences occur, along with hints of rank-related differences in immunologic parameters [for example, in this population, subordinate males have fewer circulating lymphocytes (24)]. However, no studies have explicitly shown rank-related differences in the physiology of wound healing in primates. Finally, IGF-I elevates high-density lipoprotein (HDL)-to-low-density lipoprotein (LDL) ratios in numerous species (3), and we have observed that dominant male baboons do indeed have higher HDL-to-LDL ratios (25). However, the IGF-I effect appears to be through reduction of LDL levels, whereas the rank effect that we observed emerged from an increase in HDL levels in dominant males. Thus the physiological consequences of these IGF-I-rank differences are not clear. In addition to the above mechanisms, IGF-I can facilitate many other biochemical responses by virtue of the fact that most cells have IGF-I (type 1) receptors. For example, IGF-I is a synergist with thyroid-stimulating hormone, gonadotropins, adrenocorticotropic hormone, and parathyroid hormone; regulates calcium metabolism via intestinal calcium absorption and 1,25(OH)₂D₃ production; increases glomerular filtration rate; and potentiates antigenic activation of T lymphocytes (28, 31). Thus this rank difference may have some as yet uncovered functional consequences.

In conclusion, we observe a positive correlation between high social status and IGF-I concentrations in a wild primate population. This correlation is striking, in that one can rule out many modulating factors typical in human studies, such as smoking, alcohol consumption, or other forms of substance abuse, genetic heterogeneity, or obesity. Future studies with this population will focus on elucidating what physiological implications there may be of this rank difference.