subordinate relationships between hamsters

Neuroscience Vol. 29, No. 3, pp. 675483, Printedin Great Britain

0306-4522/89 $3.00 + 0.00

1989

Pergamon Press plc 0 1989IBRO

VASOPRESSIN IMMUNOREACTIVITY IN THE ANTERIOR HYPOTHALAMUS IS ALTERED DURING THE ESTABLISHMENT OF DOMINANT/SUBORDINATE RELATIONSHIPS BETWEEN HAMSTERS C.

F.

FERRIS,* J. F. &msoN,t

A. M. MARTIN

and L. F.

ROBERGE

Department of Physiology, University of Massachusetts Medical Center, 55 Lake Avenue North, Worcester, MA 01655, U.S.A. TDepartment of Psychology, College of the Holy Cross, Worcester, MA 01610, U.S.A. Abstract-When paired for 15min periods for 5-8 consecutive days, castrated, testosterone-treated hamsters consistently assumed the dominant status, based on a higher aggression index (18 f 3) and frequency of flank marking (15 k 3) as compared to their castrated, untreated subordinate partners (- 1.3 f 1 and 2.4 f 1, respectively). In addition to these hamsters with established dominant/subordinate relationships, control hamsters with no social interactions were killed, and in all animals the vasopressin level in the anterior hypothalamus-medial preoptic area was assessed by counting vasopressin immunoreactive perikarya following immunocytochemistry, or by radioimmunoassay of vasopressin from tissue punches. In the socialized pairs the subordinate hamsters had a significantly (P < 0.01) lower number of vasopressin staining perikarya in the anterior hypothalamus, specifically the area of the nucleus circularis, than their dominant partners (n = 6 pairs). There was also a significantly (P < 0.001) lower level of vasopressin immunoreactivity in punches taken from the area of the nucleus circularis in subordinate hamsters as compared to their dominant partners (n = 14pairs). However, there were no significant differences in the number of perikarya or the concentration of immunoreactive vasopressin between subordinate and dominant hamsters in the supraoptic nucleus, paraventricular nucleus, suprachiasmatic nucleus or bed nucleus of the stria terminalis. The number of perikarya (n = 5 pairs) and concentration of vasopressin (n = 8 pairs) for all vasopressin immunoreactive sites, including the nucleus circularis, were similar for testosterone-treated and untreated hamsters that remained isolated and not subjected to daily aggressive encounters. The values in both groups of unsocialized hamsters were also similar to those seen in the testosterone-treated dominant hamsters. In summary, the only difference in the vasopressin level between dominant and subordinate hamsters was observed in the area of the nucleus circularis. Furthermore, the only group of hamsters that differed in both the number of vasopressin immunoreactive perikarya and concentration of vasopressin were castrated controls, i.e. subordinates, that were tested daily with highly aggressive testosterone-treated dominant hamsters. These data provide evidence showing (1) that the diminution in vasopressin immunoreactivity in subordinate hamsters occurs in response to repeated social encounters with aggressive dominant hamsters, (2) that a behavioral manipulation may contribute to changes in the neuroanatomy and/or chemistry of the central nervous system, and (3) that the vasopressinergic neurons of nucleus circularis may be involved in the control of vasopressin-dependent flank marking. The lower level of flank marking behavior in subordinate hamsters may reflect, in part, the level of vasopressin synthesis and release from this population of neurons.

Many mammals communicate by disseminating olfactory signals produced by specialized scent glands. Male Golden hamsters (Mesocricetus auratus) have large, scent-producing, sebaceous glands located on the dorsolateral flanks surrounded by tufts of dark pigmented hair. Hamsters disseminate the chemicals produced by these glands by rubbing their flanks against objects in their environment.” This behavior, called flank marking, occurs most frequently during

aggressive encounters between hamsters, and one of its functions is to communicate social status.‘2-‘5 Once a dominant/subordinate relationship is established, the dominant hamster of a pair will flank mark significantly more than its subordinate partner during a social encounter.r3 Vasopressin-sensitive neurons in the anterior hypothalamus-medial preoptic area (AH-MPOA) appear to mediate flank marking behavior.” The microinjection of arginine vasopressin (AVP), but no other

*To whom correspondence should be addressed. Abbreviations: AH-MOPA. anterior hvoothalamus-medial

peptides or the classical neurotransmitters, into this area stimulates robust flank markine behavior.**”

preoptic area; AVP, &nine vaso&ssin; BNST, bed nucleus of the stria terrninalis; ICC, immunocytochemistry; NC, nucleus circularis; PBS, phosphate-buffered saline; PVN, paraventricular nucleus; RIA, radioimmunoassay, SCN, suprachiasmatic nucleus; SON, supraoptic nucleus; TBS, Tris-buffered saline.

Structure-function

studies with different analogs of

vasopressin and oxytocin having pressor (V, -receptor mediated response) or antidiuretic (V,-receptor mediated response) activities show that flank marking is mediated by a V,-like receptor.’ Furthermore, flank

675

676

C. F. FERRISet ul

marking behavior triggered by the microinjection of AVP and flank marking elicited by the odors or aggressive encounters from other hamsters can be prevented by the prior microinjection of V, -receptor antagonists into the AH-MPOA.9,‘2 These data indicate that flank marking, a complex communicative behavior, is controlled by vasopressin-sensitive neurons in a discrete site of the hypothalamus of hamsters. The purpose of the present study was to determine whether the vasopressin system in the AH-MPOA differed between dominant and subordinate hamsters. The vasopressin system in dominant and subordinate members of a pair was examined using two methods: (1) visualization and counting of vasopressin perikarya following immunocytochemistry (ICC) using the unlabeled antibody-enzyme immunoperoxidase technique with primary antiserum directed against vasopressin; and (2) radioimmunoassay (RIA) of vasopressin in discrete punches of different hypothalamic nuclei shown to have positive immunostaining for vasopressin. EXPERIMENTALPROCEDURES In preliminary studies we noted a modest diminution in the concentration of vasopressin in the anterior hypothalamus of gonadally intact subordinate hamsters as compared to their intact dominant partners. Commensurate with this change were differences in aggression and flank marking. However, the levels of aggression and flank marking between these pairs of hamsters with established dominant/subordinate relationships were not always consistent and in some cases subtle. In order to maximize the behavioral differences, dominant/subordinate relationships were established between castrated testosterone-treated hamsters and castrated, testosterone deficient hamsters. Male hamsters (110-120 g) were obtained from Harlan Sprague-Dawley Laboratories (Indianapolis, IN). Within one week of their arrival each hamster was anesthetized with sodium pentobarbital(lO0 mg/kg), castrated, and implanted with a silastic capsule measuring 22 mm in length, with an inner and outer diameter of 0.78 in and 0.125 in respectively. Capsules packed with testosterone (Sigma Chemical, MO, lot No. 65F-0054) or empty controls were implanted beneath the skin above the shoulders. Following surgery hamsters were housed individually in Plexiglas cages, maintained on a reversed light:dark cycle of 14: 10 (light on at 1700 h), and provided with Purina Lab Chow and water ad libittun. After four weeks castrated, testerone-treated hamsters were randomly paired with castrated, untreated hamsters for the purpose of establishing dominant/ subordinate relationships. In order to control for the effect of testerone in the absence of any socialization process, other castrated, testosterone-treated and untreated hamsters were not paired and remained isolated for the duration of 4-5 weeks. Pairs of testosterone-treated and untreated hamsters were maintained for 5-8 days of consecutive testing. Members of a pair were placed together in a neutral arena (50 x 30 x 26cm) for 15-rnin test periods each day. Dominant/ subordinate relationships were determined as previously described.‘**” Briefly, during each test period the number of attacks, bites, retreats, and flank marks made by each hamster were recorded. An attack was scored each time a hamster physically assaulted its partner. Retreats were scored when a hamster moved away from its partner while displaying a submissive posture, i.e. crouched, flank exposed

and hindlimb lifted. A flank mark was scored each time 3 hamster arched its back and rubbed its flank gland region against the sides or corners of the testing arena. A hamster was considered dominant if it had a higher aggressive index (attacks plus bites, minus retreats) than its partner for a majority of the test periods. All tests were performed during the first 4 h of the dark phase under dim, red illumination. Analysis of variance was used to compare possible differences in aggression and flank marking behaviors between dominant and subordinate hamsters From three separate studies 22 of 24 pairs of animals (c. 92%) established dominant:subordinate relattonships The two pairs of hamsters that did not establish clear dominant/subordinate relationships showed little if any aggression or flank marking behavior and were therefore omitted from any further analyses. In all pairs that estahlished relationships the testosterone-treated member was dominant over its castrated. untreated partner. On the samr day of the final test period hamsters were killed and thetr brains prepared for either ICC or RlA

The ICC followed the procedure previously described by King and coworkers. ” Hamsters were anesthetized with sodium pentobarbital and transcardially perfused with heparinized (1000 units) saline (IO ml) followed bv 500 ml of a phosphate-buffered ‘solution’ of 3.7% formalin: pH 7.1. The brains were removed. blocked, and postfixed in buffered formalin for 3648 h. Coronal sections of 50 pm each were cut using a Vibratome (Lancer). The Vibratome bath was filled with 0.05 M Tris-buffered saline (TBS) (0.6% Trizma base, 0.9% NaCl brought to pH 7.6 with HCI). Sections were washed in 5% dimethyl sulfoxide for IO min. rinsed in TBS, and incubated with a solution of 10% normal goat serum (Amel Products, Inc., NY) in I % hydrogen peroxide for 30min at room temperature to remove endogenous peroxidase. Following three rinses in TBS. sections were incubated in the primary antiserum of vasopressin (Incstar Corp., MN, lot No. 8636015) at 4’C for 72 h. The concentration of vasopressin antiserum was I : 1500 in TBS containing 0.1% gelatin, 0.02% NaN,. and 0.8% Triton-X, pH 7.6 at 4°C. Following incubation with primary antiserum, sections were washed four times for 25 min each in TBS containing 0.1% gelatin, 0.02% NaN, and 0.02% T&on-X tJBS*). Sections were then incubated for 2 hr at room temperature in secondary antiserum of goat anti-rabbit IgG (Amel Products, lot No. F07092) diluted I : 50 in TBS. Following two 30-min washes in TBS., sections were incubated for 1 h in rabbit peroxidase antiperoxidase (Stemberger-Meyer Immunocytochemicals, Inc., MD, lot No. 50%) diluted I:100 in TBS*. Sections were washed in TBS and exoosed to diaminobenzidine (25 mg/ 100 ml TBS + 0.05% hydiogen peroxide, pH 7.6) for 30 min at room temperature, mounted on gelatin coated slides, counterstained with Thionin, and coverslipped with Permount (Fisher Sci.). To control for antibody specificity, vasopressin antiserum (1:1500) was preincubated with vasopressin or oxytocin (Bachem, CA) in concentrations of 1. IO, 100 and 250 FM for 24 h. Four brains from intact male hamsters were serially sectioned, with every third section incubated in either normal antiserum (I : 1500) or antiserum absorbed with either vasopressin or oxytocin at each respective concentration. There was no staining of perikarya in any hypothalamic sites when vasopressin antiserum was preincubated in 100 or 250 PM AVP, although some staining of a few large fibers around the SON persisted in both cases. Preincubation with I and IO PM AVP prevented all staining of perikarya and fibers in the suprachiasmatic nucleus (SCN), nucleus circularis (NC) and bed nucleus of the lamina terminalis (BNST), while light staining could be observed in the supraoptic (SON) and paraventricular nuclei (PVN). Preincubation of vasopressin antiserum with I and IO p M oxytocin caused little discernible change in staining while

671

Dominance and vasopressin there was a small diminution (c. l&15%) in staining following absorption with 100 or 250 nM oxytocin. Vasopressin-immunoreactive perikarya and processes were visualized as the brown reaction product of diaminobenzidine. Cells were counted under light microscopy at 100 x magnification. Four bilateral areas containing vasopressinergic neurons were counted and compared: SON, PVN, NC, and BNST. The SON included all perikarya associated with optic tracts and did not include the retrochiasmatic region. The PVN included all cells found along the dorsomedial aspect of the third ventricle beginning at the rostra1 border of the SCN and ending caudally at the ventromedial nuclei. The PVN also included perikarya located around the posterior perifomical nuclei. The NC contained perikarya situated in the anterior hypothalamus, dorsal to the SCN and midway between PVN and SON. The BNST included cells extending between the anterior commissure and rostra1 paraventricular nuclei situated dorsal and lateral to the fornix. The SCN was omitted from these analyses because the perikarya were small, few in number, and not easily distinguished from the dense stippling of vasopressin immunoreactivity characteristic of the SCN. A cell was counted if any part of the nucleus was visible. There was no attempt to discern between light, medium or dense staining perikarya. Analysis of variance was used to compare possible differences in mean number of perikarya in the various nuclei. Radioimmunoassay Hamsters were killed by decapitation, blood collected, and the brains rapidly removed and frozen on dry ice. The brains were positioned ventral side up in a Plexiglas block (Zivic Miller Labs, Inc., PA) designed with slits fitted for a razor blade and spaced 1 mm apart for the purpose of rapidly cutting uniform sections of whole brain. With this technique 1000~nm frozen sections were obtained that included the rostra1 and caudal borders of the optic chiasm. With the aid of a dissecting microscope a single midline punch measuring 250 pm in diameter was collected from the area of the SCN. The SCN punch was followed by bilateral punches of 250 pm each from the area of the NC. The SON was collected from bilateral punches measuring 500 pm in diameter while the PVN was collected from a single punch measuring 500nm in diameter. Each punch from each nuclei was immediately transferred to the tip of a sonicator (Heat Systems-Ultrasonics, Inc., NY). The tissue was dipped into a microfuge tube containing 250 ~1 of ice cold acid/acetone (1 N HCL/acetone 3: 100, v/v) and sonicated for 5 s. Fifty microliters of each homogenate was saved for protein analysis, while the remaining sample was immediately dried without transfer from the microfuge tube in a vacuum centrifuge (Savant Instruments, NY). All samples with the exception of the SON were reconstituted in 800 nl of assay buffer (0.15 M NaCl, 0.05 M NaH,PO, and Na,HPO,, pH 7.4, 0.1% gelatin, 0.02% NaN,), and duplicate volumes of 350~1 were transferred to I2 x 75 plastic tubes for RIA. The SON samples were reconstituted in 1000~1 of assay buffer and duplicate volumes of 25 ~1 transferred to plastic tubes for RIA. Synthetic AVP (Bachem, CA) was iodinated using a chloramine-T procedure. Four micrograms of AVP and 0.5 mCi Na ‘25I were mixed in 50 ~1 of phosphate buffer (0.5 M, pH 7.5). The reaction was initiated by the addition of 40~1 of chloramine-T (0.2mg/ml in 0.5-M phosphate buffer, Sigma Chemical. MO). vortexed for 60 s. and followed by- the addition of 0.5 ml of 0.1% bovine serum albumin prepared in 0.05 M phosphate buffer. The [‘Z51]AVP was isolated by high pressure liquid chromatograuhv usine a reverse phase -C-l8 column (3.9 x 3OOmmj run a‘i I.5 ml/min in 0.1% TFA and eluted with a linear gradient of 40% acetonitrile over 60 min. The vasopressin antisera was obtained from Calbiochem-Behring, CA (lot No. 393079) and is reported to have less than I % cross-reactivity

with arginine vasotocin, oxytocin, and Lys*-vasopressin. The antisera was used at a final dilution of c. 1: 400,000in an assay volume of 500~1. The trace, antibody, and unknown sample were incubated for 18-24 h at 4°C. Free trace was separated from bound by the addition of 1.0 ml of 1: 2 dilution of a stock suspension of charcoal (2.5%) and dextran T-70 (0.25%) in assay buffer to each tube. The supematant containing the bound fraction was counted in a multiwell gamma counter (Nuclear Enterprises, Scotland). The sensitivity of the assay was between 0.5 and 1 pg as defined by the 90% maximum binding or 68 pg as defined by the 50% maximum binding. The intraassay coefficient of variability was 6.5% and the interassay coefficient of variability was 7.3%. The recovery of synthetic AVP added to 250~1 of acid/acetone in microfuge tubes prior to sonication, in doses of 5, 10 and 20 pg, ranged between 7682%. Each measure of AVP was normalized to pg of protein. The amount of protein was determined by spectrophotometry using a protein assay kit (Bio-Rad, Richmond, CA). The concentrations of AVP in each punch for the dominant/subordinate members of a pair and isolated hamsters were compared by analysis of variance. Testosterone was radioimmunoassayed using a kit obtained from Radioassay Systems Laboratories, CA. Briefly, serum testosterone was isolated by ether extraction. The ether extract was evaporated and subsequently assayed in duplicate. The intraassay and interassay variations were approximately 5 and 15%, respectively. RESULTS When tested together for 15-min periods for 5-8 consecutive days, testosterone-treated hamsters displayed significantly (F,,, = 44.1, P < 0.001) more aggression than their castrated, untreated partners (Fig. 1). In all pairs that established dominant/ subordinate relationships (n = 22), the testosteronetreated hamsters had a higher aggression index and were assigned the dominant status. Similarly, the amount of flank marking was significantly (F,,, = 29.5, P < 0.001) higher in testosterone-treated hamsters (Fig. 1). In every pair the testosterone-treated hamster displayed more flank marking behavior than its untreated subordinate partner. The plasma levels of testosterone in treated and untreated hamsters were 6.8 + 1.2 and 0.2 f 0.03 ng/ml, respectively, as compared to 3.1 f 0.32 ng/ml in gonadally intact

AL&TWSiO”

Flank MarkIng

Fig. 1. Aggression and flank marking for castrated, testosterone-treated hamsters and their castrated, untreated partners. Vertical lines denote S.E.M. lP < 0.001.

C. F. FERRISet al.

678

1

T

PVN

Fig. 2. Numbers of vasopressin immunoreactive perikarya in the bed nucleus of the stria terminals (BNST), nucleus circularis (NC), paraventricular nucleus (PVN), and supraoptic nucleus (SON) in dominant, testosterone-treated (castrate + T) and subordinate, untreated (castrate) hamsters. Vertical lines denote S.E.M. *P < 0.01.

hamsters maintained under similar environmental conditions. It should be noted that all behavioral tests were performed 28-35 days after castration. Initially all animals weighed between 110 and 120 g; however, following castration untreated hamsters became progressively heavier until on the last day of testing they were significantly (F,,M = 14.2, P < 0.001) heavier (135 f 2.6 g) than the testosterone-treated hamsters (123 f 1.6 g). The vasopressin immunoreactive perikarya from SON, PVN, NC and BNST were counted under light-field microscopy. The mean number of perikarya from each region from six pairs of dominant/ subordinate hamsters are shown in Fig. 2. There were significantly fewer (F,,,, = 9.3, P < 0.01) vasopressin immunoreactive perikarya in the NC in subordinate hamsters as compared to their dominant partners. There were no significant differences in the number of vasopressin immunoreactive perikarya between dominant and subordinate animals for the PVN, SON or BNST. A comparison of the vasopressin immunoreactivity at the level of the anterior hypothalamus between a dominant, testosterone-treated hamster and its subordinate, untreated partner is shown in Fig. 3. Vasopressin-positive staining can be seen in the SON, PVN, SCN, and NC. These sections photographed under dark-field reveal qualitative differences in the intensity of staining between the two hamsters with an established social hierarchy. For instance, the

subordinate hamster appears to have less staining m the PVN and NC while its dominant partner shows less staining in the SCN. However, in all cases these differences were only consistent for the area of the NC. In both hamsters vasopressin immunoreactive processes appear to project from the PVN and SON toward the area of the NC in the anterior hypothalamus. The qualitative differences in vasopressin staining perikarya and processes in the NC between this pair of dominant/subordinate hamsters are better seen at the higher magnification shown in Fig. 4. The intensely staining perikarya appear orange in color while the processes are white. These photomicrographs reveal a dense accumulation of processes in the area of the NC of the dominant hamster with fewer processes in the same region of its subordinate partner. The concentration of vasopressin was determined in punches taken from PVN, SON, SCN, and NC in 14 pairs of testosterone-treated and untreated hamsters with established dominant/subordinate relationships. The position and size of these punches are shown in Fig. 5. The only consistent and significant difference (F,,,, = 14.9, P < 0.001) in vasopressin levels between dominant and subordinate partners for each site was found in the NC (Fig. 6). The mean concentration of vasopressin in the NC of subordinate hamsters was over two-fold less than that of their dominant partners. In 13 pairs the subordinate hamsters had less vasopressin in the NC than their dominant partners, while in the remaining one pair the levels were comparable. There were no consistent or significant differences between hamsters in the levels of vasopressin in any other nuclei. Testosterone-treated and untreated hamsters, individually housed for 45 weeks without any social interaction, were examined to determine whether the difference in the vasopressin system between dominant and subordinate members of a pair was due to the presence or absence of testosterone, and unrelated to the development of a social hierarchy. There were no significant differences in the number of immunoreactive perikarya or the concentration of vasopressin in any sites including the NC (see Table 1). Clearly, these data show that the decrease in the vasopressin immunoreactivity seen in untreated, subordinate hamsters was not due to only the absence of testosterone, because the number of cells and the concentration of vasopressin for untreated, isolated hamsters were comparable to those for testosteronetreated hamsters with and without social interactions.

Table 1. Mean (f S.E.M.) immunoreactive perikarya and AVP concentrations @g/pg protein) in vasopressin immunoreaetive areas from isolated hamsters

Site Treatment

SON

PVN

NC

BNST

Ce11numbez

Testosterone (n = 5) Control (n = 5)

1931 k 110 1718 f 106

487 f 96 411 k32

64*4 65 k 6

61 +5 57 + 8

AVP level

Testosterone (n = 8) Control (n = 8)

140*38 174 * 49

12.5 k 4 11.8*3

9.4* 3 7.8 k 2

24.1 k 6 25.4 k 4

~~~UR~~NATE

PARTNER

2mm Fig. 3. Vasoprcssin immunoreactivity in the area of the anterior hypothalamus for a dominant, testosterone-treated hamster and its subordinate, untreated partner. Shown are 50-pm s&at sections ~hoto~aph~ under dark field at x40 magnification. T&e arrows denote the location of the respective nuclei.

DOMINANT

Fig.

4. Vasopressin nant:subordinate

PARTNER

SUBORDINATE

immunoreactivity in the area of the nucleus circularis (arrow) at higher magnification pair shown in Fig. ? but photographed

PARTNER

for

the dome-

( y 1OCl)

Dominance and vasopressin

t%l

Fig. 5. Photograph of a ~~~rn-t~~k section through the brain of a hamster showing the tocation and size of the punches in the hypothalamus. The arrows indicate the focation of the punches aimed at the SON, PVN, SCN, and NC.

DISCUSSION

Since flank marking is de~nden? upon the vasopressin-sensitive neurons in a localized site of the AH-MPOA, it was the aim of these studies to determine whether the levels of vasopressin in this region were different between testosterone-treated dominant hamsters and their untreated subordinate

NG

Fig. 6. Concentration of arginine vasopressin {AVP) in punches taken from the NC, SCN, PVN, and SON of dominant, testosterone-treated hamsters (castrate + T) and their subordinate, untreated partners (castrate). Vertical lines denote S.E.M. *P < 0.001.

partners. Our data show that domi~ant/subord~nate ~la~onships were easily established when testosterone-treated hamsters were paired with untreated hamsters. In all cases testosterone-treated hamsters assumed the dominant status, displaying intense aggression and flank marking behavior during repeated social encounters with their untreated, subordinate partners. Subordinate hamsters showed little aggression and seldom Aank marked in the presence of their dominant partners. These data corroborate other reports showing a positive correlation between the level of testosterone and aggressive behavior in several different species.3~4~7~‘8~21~22 Associated with these disparities in behavior between dominant and subordinate hamsters were differences in the levels of vasopressin in the AH-MPOA, specifically the NC. The subordinate hamsters had significantly fewer numbers of vasapressin immunoreactive perikarya and an ostensibly lower density of vasopressin fibers in the area of the NC than their dominant partners. These differences in vasopressin immunoreaetive neurons and vasopressin con~nt~tjons between dominant and subordinate hamsters appeared to be limited to the NC because vasopressin levels in the SON, PVN, BNST, and SCN did not differ between both groups.

682

C.

F.

FERRISef ai

While the number of vasopressin perikarya and fibers have been reported to diminish in rats castrated for up to 15 weeks,’ the absence of testosterone per se did not appear to be responsible for the marked decrease in the vasopressin system in the NC of subordinate hamsters as compared to their dominant partners. When testosterone-treated hamsters and untreated hamsters were isolated following castration and not allowed to develop dominant/subordinate relationships there were no significant differences in their vasopressin systems when analysed by either ICC or RIA. In addition, all socially isolated hamsters, whether testosterone-treated or untreated controls, did not differ from testosterone-treated hamsters that became dominant. Therefore, the diminution in vasopressin immunoreactivity only occurred in the subordinate hamsters lacking testosterone and confronted daily with a dominant and aggressive partner. These data suggest that the process of socialization is responsible, in part, for the differences in vasopressin that occur during the development of a social hierarchy over the course of multiple encounters between conspecifics. From these results we speculate that the decrease in vasopressin immunore~tivity that o=urs in untreated socialized hamsters is a result of an interaction between two variables: (1) the lack of testosterone, and (2) stimuli resulting from interactions with an aggressive, dominant conspecific. Whether these observations would generalize to intact hamsters with established dominance hierarchies is unclear. In preliminary studies with intact pairs we observed that subordinate hamsters had a modest decrease in the number of immunoreactive perikarya and concentration of vasopressin in the NC as compared to their dominant partners. However, these differences in the vasopressin system were less dramatic and consistent than in our present study. It is important to note that when testing intact hamsters behavioral differences between dominant and subordinate partners are also less dramatic. Intact hamsters develop dominance. hierarchies gradually, and rarely does a member of a pair immediately assume the dominance status. It is unknown whether the process of subordination in hamsters described in the present studies causes a progressive decline in the level of testosterone and androgenic activity as it does in other If there is a decline in the level of testosspecies. 17sM terone in intact hamsters in response to the daily stress of a dominant aggressor than there would be correspondingly less aggression and flank marking behavior. Hence the development of dominance hierarchies between intact hamsters begins with an aggressive encounter, and presumably with each subsequent interaction the level of testosterone in the subordinate hamsters would diminish. Testosteronedependent behaviors such as aggression and flank marking would also decrease in frequency. In turn the continual pairing with a dominant aggressive partner

and the lowering of testosterone levels would together contribute to a decrease in vasopressin immunoreactivity in the area of the hypothalamus known to control flank marking behavior, i.e. the anterior hypothalamus. This proposed sequence of events starts with a behavior affecting endocrine status. Subsequently, the altered endocrine status in conjunction with the behavioral interaction affects the area of the central nervous system that controls the behavior. The specific mechanisms of how testosterone and behavioral subjugation are affecting the synthesis, storage, and release of vasopressin are undoubtedly complex. A contributing factor that has yet to be examined is how the release of adrenal corticosteroids most likely accompanying the stress experienced by the subordinate hamsters would impact on this system. The diminution in vasopressin immunoreactivity that occurs in subordinate hamsters appears to be limited to the neurons of the nucleus circularis and the fibers in the surrounding anterior hypothalamus, a region shown to be involved in the control of flank marking behavior. The nucleus circularis was originally described as an accessory supraoptic nucleus that fails to migrate laterally along the optic chiasm with the majority of supraoptic neurons during development.* This association with the supraoptic nucleus has presupposed a neurosecretory role for these neurons. However, the level of vasopressin immunoreactivity in the NC was significantly altered in subordinate hamsters while the neurosecretory neurons of the SON and PVN that contribute to the hypothalamo-neurohypophyseal tract appeared to be unchanged. This finding corroborates previous work showing no change in Gomori-positive staining in the NC following prolonged dehydration,” and argues against a purely neurosecretory role for the vasopressinergic neurons in the NC. It is possible that these neurons have axon collaterals that extend to the neurohypophysis while other branches synapse with neurons in the immediate area of the anterior hypothalamus. However, it is unlikely that the dense accumulation of fibers that appear to course through and around the nucleus circularis all originate from vasopressinergic perikarya in the NC. The average number of vasopressin immunoreactive neurons in the NC (c. 60 cells) comprises approximately 50% of the total cell population of the NC. Moreover. the immunoreactive fibers from these neurons are short and tine, and appear to be confined to the immediate area around the NC. Thus the anterior hypothalamus may receive additional input from vasopressinergic neurons, possibly from an extrahypothalamic site,$ that are involved in the control of flank marking behavior. Data from this study suggest that the vasopressin neurons of the NC may be one source of the vasopressin neurotransmitter implicated in the control of flank marking. Subordinate hamsters, which display very little flank marking behavior, have a lower concentration of VP and a fewer number of VP

Dominance and vasopressin

perikarya in the area of the NC, as compared to their dominant partners. The decrease in vasopressin immunoreactivity in subordinate hamsters may be due to a decrease in peptide synthesis. At this time the question of whether the change in the vasopressin system that follows the lack of testosterone and social subjugation affects a subordinate hamster’s ability to flank mark and perform other aggressive behaviors remains unclear.

683

Acknowledgements-This work was supported by NIH grant NS-23557 to CFF and Veteran’s Administration Research Funds awarded to JFA. We would like to thank Dr Joan King and the members of her laboratory at Tufts University School of Medicine for their instruction on the performance and analyses of immunocytochemistry. We thank Dr Elliott Alber’s laboratory at Georgia State University for the testosterone determinations and Peggy Mathews for her secretarial assistance.

REFERENCES 1. Albers H. E., Pollock J., Simmons W. H. and Ferris C. F. (1986) A Vl-like receptor mediates vasopressin-induced flank marking behavior in hamster hypothalamus. J. Neurosci. 6, 20852089. 2. Bandaranayake R. C. (1971) Morphology of the accessory neurosecretory nuclei and of the retrochiasmatic part of the supraoptic nucleus of the rat. Acta anar. 80, 14-22. 3. Beeman E. A. (1947) The effect of male hormones on aggressive behavior in mice. Physiol. Zool. 20, 373-405. 4. Bevan J. M., Bevan W. and Willaims B. F. (1958) Spontaneous aggressiveness in young castrate C3H mice treated with three dose levels of testosterone. Physiol. Zool. 31, 284-288. 5. Caffe A. R. and Van Leeuwen F. W. (1983) Vasopressin-immunoreactive cells in the dorsomedial hypothalamic region, medial amygdaloid nucleus and locus coeruleus of the rat. Cell Tiss. Res. 233, 23-33. 6. DeVries G. J., Buijs R. M., Van Leeuwen F. W., Gaffe A. R. and Swaab D. F. (1985) The vasopressinergic innervation of the brain in normal and castrated rats. J. camp. Neurol. 233, 236254. 7. Drickamer L. C., Vandenbergh J. G. and Colby D. R. (1973) Predictors of dominance in male Golden hamsters (Mesocricetus auratus). Anim. Behav. 21, 557-563. 8. Ferris C. F., Albers H. E., Wesolowski S. M., Goldman B. D. and Leeman S. E. (1984) Vasopressin injected into the hypothalamus triggers a stereotypic behavior in golden hamsters. Science 224, 521-523.

9. Ferris C. F., Pollock J., Albers H. E. and Leeman S. E. (1985) Inhibition of flank-marking behavior in golden hamsters by microinjection of a vasopressin antagonist into the hypothalamus. Neurosci. Left. 55, 239-243. 10. Ferris C. F. and Albers H. E. (1985) Further studies of vasopressin as a chemical messenger in the expression of flank marking behavior in Golden hamsters. In Washington Spring Symposium on Neural and Endocrine Peptides and Receptors, 1985, Abstr. 138. 11. Ferris C. F., Meenan D. M. and Albers H. E. (1986) Microinjection of kainic acid into the hypothalamus of golden hamsters prevents vasopressin-dependent flank-marking behavior. Neuroendocrinology 44, 112-I 16. 12. Ferris C. F., Meenan D. M., Axelson J. F. and Albers H. E. (1986) A vasopressin antagonist can reverse dominant/subordinate behavior in hamsters. Physiol. Behav. 38, 135138. 13. Ferris C. F., Axelson J. F., Shinto L. H. and Albers H. E. (1987) Scent marking and the maintenance of dominant/subordinate status in male golden hamsters. Physiol. Behav. 40, 661664. 14. Huck U. W., Lisk R. D. and Gore A. C. (1985) Scent marking and mate choice in Golden hamsters. Physiol. Behav. 35, 389-393. 15. Johnston R. E. (1975) Scent marking by male Golden hamsters (Mesocricetus auralus): I. Effects of odors and social encounters. Z. Tierpsychol. 37, 75-98. 16. King J. C., Lechan R. M., Kugel G. and Anthony E. L. P. (1983) Acrolein: a fixative for immunocytochemical localization of peptides in the CNS. J. Histochem. Cytochem. 31, 62-68.

17. Lloyd J. A. (1971) Weights of testes, thymi, and accessory reproductive glands in relation to rank in paired and grouped house mice. Proc. Sot. exp. Biol. Med. 137, 19-21. 18. Payne A. P. and Swanson H. H. (1972) The effect of sex hormones on the agonistic behavior of the male Golden hamster (Mesocricetus auratus Waterhouse). Physiol. Behav. 8, 687691. 19. Peterson R. P. (1966) Magnocellular neurosecretory centers in the rat hypothalamus. J. camp. Neural. 128, 181-190. 20. Rose R. M., Bernstein I. S. and Gordon T. P. (1975) Consequences of social conflict on plasma testosterone levels in rhesus monkeys. Psychosom. Med. 37, 5&61. 21. Sapolsky R. M. (1982) The endocrine stress-response and social status in wild baboon. Harm. Behuv. 16, 279-292. 22. Vendenbergh J. G. (1971) The effects of gonadal hormones on the aggressive behaviour of adult Golden hamsters (Mesocricetus auratus). Physiol. Behav. 19, 589-594. (Accepted 11 Augusf 1988)