Chester Clinic

 

Oconnell.fas.harvard.edu

Social Status Predicts How Sex Steroid Receptors
Regulate Complex Behavior across Levels of
Biological Organization
Lauren A. O'Connell and Hans A. Hofmann Institutes for Cellular and Molecular Biology (L.A.O., H.A.H.) and Neuroscience (H.A.H.) and Section ofIntegrative Biology (L.A.O., H.A.H.), The University of Texas at Austin, Austin, Texas 78712 Social status strongly affects behavior and physiology, in part mediated by gonadal hormones,
although how each sex steroid acts across levels of biological organization is not well understood.
We examine the role of sex steroids in modulating social behavior in dominant (DOM) and sub-
ordinate (SUB) males of a highly social fish, Astatotilapia burtoni. We first used agonists and
antagonists to each sex steroid receptor and found that androgens and progestins modulate
courtship behavior only in DOM, whereas estrogens modulate aggressive behavior independent
of social status. We then examined the hormonal and physiological responses to sex steroid re-
ceptor antagonist treatment and uncovered substantial changes in circulating steroid hormone
levels and gonad size only in SUB, not in DOM. Consistent with status-based physiological sensi-
tivities to drug manipulation, we found that neuropeptide and steroid receptor gene expression
in the preoptic area was sensitive only in SUB. However, when we compared the transcriptomes of
males that received either vehicle or an estrogen receptor antagonist, 8.25% of all genes examined
changed expression in DOM in comparison with only 0.56% in SUB. Finally, we integrate behavior,
physiology, and brain gene expression to infer functional modules that underlie steroid receptor
regulation of behavior. Our work suggests that environmentally induced changes at one level of
biological organization do not simply affect changes of similar magnitude at other levels, but that
instead very few key pathways likely serve as conduits for executing plastic responses across mul-
tiple levels. (Endocrinology 153: 1341–1351, 2012)
individual's social status. To function in a social ment, physiology, and molecular responses in the nervous group, every group member integrates information about system. Thus, our present understanding of how physiol- its own internal physiological state with external social ogy, brain gene expression, and social environment regu- information into a behavioral output that ultimately pro- late behavior in social groups is still limited.
motes its fitness. Although behavioral and hormonal re- Sex steroid hormones are excellent candidates for me- sponses within social groups are relatively well under- diating the integration of external and internal informa- stood in mammals and fish (1–5), researchers have only tion into an adaptive behavioral response. These hor- recently begun to examine the molecular events in the mones vary across many physiological contexts such as brain that mediate the behavioral responses of individuals sex, social status, breeding condition, and season (6 –10).
in different social states (6, 7). Despite the importance of Such variability is important in regulating an animal's be- integrating external and internal signals within the brain havioral response to both reproductive and aggressive into subsequent behavioral changes, studies investigating contexts (11). Circulating hormones robustly respond to this integration are limited due to the general difficulty of social stimulation and have been extensively studied in the ISSN Print 0013-7227 ISSN Online 1945-7170 Abbreviations: AR, Androgen receptor; AVT, arginine vasotocin; CA, cyproterone acetate; Printed in U.S.A.
DHT, dihydrotestosterone; DOM, dominant; ER, estrogen receptor; gbw, gram body Copyright 2012 by The Endocrine Society weight; GEE, generalized estimating equations; GSI, gonadosomatic index; IST, isotocin; doi: 10.1210/en.2011-1663 Received August 22, 2011. Accepted November 23, 2011.
11KT, 11-ketotestosterone; 17␣-20␤-P, 17␣-20␤-dihydroprogesterone; POA, preoptic area; PR, First Published Online December 13, 2011 progesterone receptor; qPCR, quantitative PCR; SUB, subordinate.
For editorial see page 1001
Endocrinology, March 2012, 153(3):1341–1351 O'Connell and Hofmann Social Status Predicts Steroid Hormone Action Endocrinology, March 2012, 153(3):1341–1351 context of the androgen challenge response to aggressive compliance with the Institutional Animal Care and Use Com- interactions, a physiological response conserved across mittee at The University of Texas at Austin.
vertebrates (11, 12). Sex steroids integrate these social and Behavior and pharmacology
physiological cues into a molecular response by acting Male A. burtoni chosen for this study were stable in social through either nuclear hormone receptors, which function status for 1 wk before injections. Only one animal per tank was as transcription factors and modulate gene expression (re- manipulated at a given time. Animals were injected ip with 10 ␮l viewed in Ref. 13), or more quickly through membrane- mineral oil per gram body weight (gbw) for 2 d (to allow for bound receptors that trigger a signal transduction cascade within-individual comparisons) and then with a sex steroid re- ceptor agonist or antagonist for the next 2 d. Fish were imme-diately placed back into the home tank after injection. One group To study the proximate mechanisms of male-typical of males used as the control group for all end-point analyses were behavior in a social hierarchy, we use the African cichlid injected with vehicle for all 4 d. Five-minute focal observations fish, Astatotilapia burtoni, a model system in social neu- were conducted between 0800 and 1100 h the following morn- roscience and behavioral genomics (16, 17). A. burtoni ing of each day after injections and for 2 d after injections ended males have two plastic behavioral phenotypes. Brightly (Supplemental Fig. 1, published on The Endocrine Society's Jour-nals Online web site at http://endo.endojournals.org) by an ob- colored dominant (DOM) males aggressively defend ter- server blind to the treatment. This observation paradigm avoided ritories where they court and spawn with females. Subor- nongenomic fast-acting effects of steroid hormones (14). Social dinate (SUB) males are dull in coloration, reproductively behaviors recorded are described elsewhere (22) and include ag- suppressed, and school with females. The brain gene ex- gressive (chasing and biting), sexual (quivering and leading), ter- pression profiles of these two phenotypes differ substan- ritorial (border dispute, carousel, and threats), and subordinate tially (18). Given the opportunity, SUB will attain DOM (fleeing) displays.
We chose the following doses based on a dose-response ex- status in an astonishing transition that is accompanied by periment in DOM (Supplemental Fig. 2) where three doses of a concerted change in gene expression in brain and testes each drug were tested to see which had the largest effect on as well as a rapid response in circulating steroid hormones aggressive and/or courtship behavior: 0.4 ␮g/gbw 17␤-estradiol (6). However, we know relatively little about how sex (Steraloids, Newport, RI; n ⫽ 8 SUB and n ⫽ 10 DOM), 0.13 steroid hormone receptors mediate behavior in relation to ␮g/gbw dihydrotestosterone (DHT; Sigma Chemical Co., St.
Louis, MO; n ⫽ 8 per social status), 0.125 ␮g/gbw 17␣-20␤- social status and across levels of biological organization dihydroprogesterone (17␣-20␤-P; Steraloids; n ⫽ 8 per social (including, e.g. behavior, physiology, and brain gene status), 1.6 ␮g/gbw ICI182780 (ER antagonist; Sigma; n ⫽ 8 per social status), 0.83 ␮g/gbw cyproterone acetate (AR antagonist; We examined the differential role of sex steroid hor- Sigma; n ⫽ 8 per social status), and 1.6 ␮g/gbw ZK0112993 (PR mones in the regulation of behavior in these two social antagonist; Bayer Schering Pharma AG, Berlin, Germany; n ⫽ 8per social status). Dose-response curves were not conducted in phenotypes with the hypotheses that 1) steroid hormone SUB, because social suppression by DOM males may mask any receptors mediate distinct behavioral components and 2) effect of a drug, and isolating SUB for testing will result in a the gene regulatory actions of sex steroid hormone recep- transition to DOM status within hours, a period too short to tors differ with social status. We first determined how conduct a dose-response curve manipulating nuclear hormone androgen receptors (AR), estrogen receptors (ER), and the receptors. DHT was used, rather than the teleost-specific 11-ketotestosterone (11KT), because DHT binds AR with higher progesterone receptor (PR) modulate social behavior and affinity than 11KT (23, 24). All antagonists have been shown to circulating hormone levels in DOM and SUB within a bind their respective teleost sex steroid receptors [ICI182780 community setting. Additionally, we examined steroid re- binds all three ER (25); cyproterone acetate binds both AR (23); ceptor regulation of gene expression in the preoptic area and ZK0112993 binds PR (26)].
(POA), a brain region that regulates male aggressive and On the last day of behavioral observations, we recorded weight and length of each individual that received an antagonist sexual behaviors in all vertebrates (19, 20).
or only vehicle and drew blood from the dorsal aorta using hep-arinized 26-gauge butterfly infusion sets (Becton Dickinson,Mountain View, CA). Gonads were removed and weighed for Materials and Methods
calculation of the gonadosomatic index (GSI: testes mass dividedby body mass times 100). Brains were rapidly dissected, embed- ded in Tissue-Tek OCT Compound (Sakura Finetek U.S.A., Tor- A. burtoni from a wild-caught stock population were main- rance, CA), fresh frozen on dry ice, and stored at ⫺80 C.
tained in a naturalistic community as previously described (21)with eight males and eight females per 110-liter tank. A total of 34 DOM (length, 4.877 ⫾ 0.1264 cm; weight, 3.341 ⫾ 0.2851 Free testosterone, 17␤-estradiol, and progesterone were mea- g) and 32 SUB (length, 4.615 ⫾ 0.0851 cm; weight, 2.486 ⫾ sured for each individual using ELISA (Enzo Life Sciences, Farm- 0.1289 g) were used in this study. All work was carried out in ingdale, NY); Intraassay variation was 9.43, 2.67, and 4.06%, Endocrinology, March 2012, 153(3):1341–1351 respectively; interassay variation was 4.96, 3.38, and 10.5%, structed from brain-specific and mixed tissue libraries represent- respectively. Plasma samples were diluted 1:30 and processed ing a total of 17,712 cichlid-specific features (30, 69). DOM and as previously described (27).
SUB males were examined in separate loops, and each array wascompetitively hybridized with a vehicle- or ER antagonist- Laser microdissection and RNA isolation
treated sample. Two technical replicates were performed, where Brains were sectioned at 18 ␮m, thaw mounted onto mem- each sample was labeled with both Cy3 and Cy5 to control for brane-covered slides (P.A.L.M. Microlaser Technologies AG, dye bias. The slides were scanned on an Axon 4000B array scan- Bernried, Germany), and hydrated for 1 min in cold 95, 70, and ner using GenePix version 6.0 software (Molecular Devices, 50% ethanol, stained with toluidine blue (0.5% toluidine blue, Sunnyvale, CA) and filtered for microarray printing errors, hy- 1% phenol, 20% ethanol in water) for 2 min, rapidly dehydrated bridization artifacts, and signals with low intensity. Intensities in an ascending ethanol series, incubated in xylene for 5 min, and were lowess-normalized within arrays in R/Bioconductor; only air dried for 2 min. Sections were visualized on a laser-micro- features with average intensities 2 SD or more above average dissection microscope (P.A.L.M. Microlaser Technologies), and background were considered for further analysis (18). All raw an area corresponding to the POA (including parvocellular, mag- and processed data are available at the GEO database (accession nocellular, and some gigantocellular cells) (28) was excised and number GSE28508).
captured with the laser (Supplemental Fig. 3). Some technicalvariation likely resulted from the capture of the gigantocellular POA due to the sporadic distribution of these neurons. Thirty All analyses, with the exception of the microarray data, were microliters of Trizol (Invitrogen, Carlsbad, CA) was added to the conducted in PASW (IBM, Somers, NY). To determine differ- sample and stored at ⫺80 C until further processing. Total RNA ences between DOM and SUB, t tests were used with vehicle- was isolated with Trizol according to the manufacturer's treated animals. In all other cases, statistics for DOM and SUB were run separately. For behavioral data, a generalized estimat-ing equations (GEE) (31) model was used with the behavioral RNA amplification, reverse transcription, and
measure as the dependent variable, and baseline behavior (d 1 quantitative PCR (qPCR)
and 2) was compared with drug treatment (d 4 – 6) as a within-subject variable; the model was run separately for each drug Isolated RNA was subjected to one round of amplification treatment. Hormone measurements and qPCR data were not using the MessageAmp II system (Ambion, Austin, TX) accord- normally distributed and were log-transformed, resulting in nor- ing to manufacturer's instructions. Additionally, samples from mally distributed residuals. Hormone data were analyzed with animals that received either vehicle only or ER antagonist treat- ANOVA using hormone as the dependent variable and drug as ment were divided after the first round of amplification, with one the independent variable followed by a Bonferroni post hoc test.
half being immediately reverse transcribed and the other half qPCR gene expression data from the experimental groups were processed for another round of RNA amplification to be used for compared with the control gene expression levels using ANOVA microarray analysis. Each RNA sample was treated with deoxy- followed by Dunnett's t test, which corrects for multiple testing.
ribonuclease I (Ambion) according to the manufacturer's in- Significance was considered as P ⬍ 0.05. Because most of our structions. The RNA was reverse transcribed using Superscript data were nonparametric, we calculated correlations using III reverse transcriptase (Invitrogen) and gene-specific primers Spearman rank correlation coefficients to explore covariation (see Supplemental Table 1). Positive controls used 1 ng of whole- patterns. The microarray data were analyzed in R/Bioconductor.
brain RNA in place of RNA derived from laser-microdissected Between-group analyses were conducted using the LIMMA samples and in negative controls the reverse transcriptase was package (18, 32).
omitted. Excess primers and salts from the transcription reactionwere removed in Microcon YM30 columns (Millipore, Bedford,MA). An aliquot of the sample was used for Ribogreen (Invit-rogen) analysis to determine total cDNA concentration of each sample. qPCR primers (Supplemental Table 1) were designed toflank exon boundaries using the zebrafish genome as a reference.
Distinct roles of sex steroids in a social hierarchy
For each sample, target gene abundance was measured in trip- We used a within-subject design by treating DOM and licate in an ABI PRISM 7900HT real-time PCR cycler (ABI SDS SUB with specific sex steroid receptor agonists or antag- version 2.2.1 software) using SYBR Green (Invitrogen). Stan- onists (Supplemental Fig. 1) and found that sex steroid dard curves were constructed using known dilutions of cDNA,and amplification efficiency was calculated. For each individual, receptors differentially modulate aggression and court- median values from the reference and target gene triplicates were ship in the two phenotypes (Fig. 1). Although teleost fish used to calculate the relative transcript abundance of the target have a single ER␣ and PR, there are two isoforms of AR gene using the mean normalized expression formula of Simon (AR␣ and AR␤) and ER␤ (ER␤a and ER␤b) due to a ge- (29). Each sample was normalized to total cDNA as measured by nome duplication event in this lineage (33). We selected drugs that targeted all three ER isoforms or both ARisoforms.
Microarray samples were prepared as previously described Manipulation of ER changed aggressive behavior, but (18) and hybridized in a loop design (Supplemental Fig. 4) to a not courtship behavior, independent of social phenotype.
19K A. burtoni microarray (GEO platform GPL6416) con- DOM males display two kinds of aggression: chases are

O'Connell and Hofmann Social Status Predicts Steroid Hormone Action Endocrinology, March 2012, 153(3):1341–1351 FIG. 1. Effects of sex steroid receptor manipulation vary by social status. The mean behavioral observations of animals given ER (top row), AR
(middle row), or PR (bottom row) manipulations are shown with agonist (black lines) and antagonist (gray lines) treatments for DOM and SUB
males (n ⫽ 8 per status per treatment). Behavior per 5 min is represented on the vertical axis, and day of treatment is on the horizontal axis. Gray
background shading indicates days of drug administration. Data are represented as the mean ⫾ SEM; GEE: **, P ⬍ 0.0001; *, P ⬍ 0.05.
typically directed toward SUB and females in the school, gressive behavior even under control conditions. ER ma- whereas territorial defense behavior is typically directed nipulation did not affect fleeing behavior in SUB (GEE: toward other DOM in the community. Estradiol treat- estradiol, P ⫽ 0.24; Wald ␹2 ⫽ 1.382; ICI182780, P ⫽ ment increased chases in DOM (GEE, P ⫽ 9.3 ⫻ 10⫺9; 0.86; Wald ␹2 ⫽ 0.033). Courtship and territorial defense Wald ␹2 ⫽ 32.984), whereas treatment with the ER-spe- behavior are almost never observed in SUB and thus are cific antagonist ICI182780 decreased chases (GEE, P ⫽ not reported.
1.4 ⫻ 10⫺6; Wald ␹2 ⫽ 23.284). Estradiol treatment also AR manipulations exclusively affected courtship be- decreased territorial defense behavior (GEE, P ⫽ 0.045; havior and had no effect on aggressive behavior in either Wald ␹2 ⫽ 4.036; data not shown), although ICI182780 social phenotype. Instead of the teleost-specific androgen treatment had no effect (GEE, P ⫽ 0.283; Wald ␹2 ⫽ 11KT, we used the nonaromatizable androgen DHT for 1.151; data not shown). ER manipulations in DOM did pharmacological manipulation of AR, because both 11KT not affect courtship displays (GEE: estradiol, P ⫽ 0.864; and DHT produce similar effects in other teleosts (34, 35), Wald ␹2 ⫽ 0.03; ICI182780, P ⫽ 0.06; Wald ␹2 ⫽ 3.539), yet DHT has a higher binding affinity to both AR␣ and although there is a nonsignificant trend for the antagonist AR␤ (23, 24). DOM treated with DHT increased court- to decrease courtship displays. Aggressive displays (a sum ship displays (GEE, P ⬍ 0.0002; Wald ␹2 ⫽ 13.794), of chases and bites) were also increased in SUB treated whereas treatment with an AR antagonist, cyproterone with estradiol (GEE, P ⫽ 1.8 ⫻ 10⫺4; Wald ␹2 ⫽ 13.993), acetate (CA), decreased courtship behavior (GEE, P ⫽ which was unexpected given the severe social suppression 8.5 ⫻ 10⫺6; Wald ␹2 ⫽ 19.823). Manipulation of AR did SUB are subjected to by DOM. Not surprisingly, treat- not alter chases (GEE: DHT, P ⫽ 0.22; Wald ␹2 ⫽ 1.502; ment of SUB with an ER antagonist did not result in a CA, P ⫽ 0.20; Wald ␹2 ⫽ 1.624) or territorial defense be- behavioral change (GEE, P ⫽ 0.8; Wald ␹2 ⫽ 0.064), pos- havior (GEE: DHT, P ⫽ 0.20; Wald ␹2 ⫽ 1.653; CA, P ⫽ sibly due to a floor effect, because SUB rarely display ag- 0.09; Wald ␹2 ⫽ 2.890; data not shown) in DOM. Similarly,

Endocrinology, March 2012, 153(3):1341–1351 no behavioral change in total aggression (GEE: DHT, P ⫽0.55; Wald ␹2 ⫽ 0.360; CA, P ⫽ 0.55; Wald ␹2 ⫽ 0.355) orfleeing (GEE: DHT, P ⫽ 0.81; Wald ␹2 ⫽ 0.058; CA, P ⫽0.75; Wald ␹2 ⫽ 0.100) was observed in SUB.
PR manipulations showed effects similar to those of AR, as we observed changes in courtship behavior dis-played by DOM. We used 17␣-20␤-P as a PR agonistbecause it cannot be converted readily into testosteroneand has higher binding affinity to the teleost PR than pro-gesterone (24). DOM treated with 17␣-20␤-P increasedcourtship displays (GEE, P ⫽ 0.040; Wald ␹2 ⫽ 4.233),whereas the PR antagonist ZK112993 decreased court-ship displays (GEE, P ⫽ 0.041; Wald ␹2 ⫽ 4.168). PRmanipulation in DOM did not affect chasing (GEE: 17␣-20␤-P, P ⫽ 0.148; Wald ␹2 ⫽ 2.091; ZK112993, P ⫽0.053; Wald ␹2 ⫽ 3.751) or territorial defense behavior(GEE: 17␣-20␤-P, P ⫽ 0.08; Wald ␹2 ⫽ 3.100;ZK112993, P ⫽ 0.752; Wald ␹2 ⫽ 0.100; data notshown). Similar to DOM, total aggression in SUB did notchange with PR manipulation (GEE: 17␣-20␤-P, P ⫽0.14; Wald ␹2 ⫽ 2.173; ZK112993, P ⫽ 0.68; Wald ␹2 ⫽0.172). Interestingly, SUB treated with 17␣-20␤-P dis-played less fleeing behavior (GEE, P ⫽ 0.001; Wald ␹2 ⫽11.195), but their behavior did not change after treatmentwith the PR antagonist (GEE, P ⫽ 0.52; Wald ␹2 ⫽ 0.409).
Dissociation of behavior from hormones and
physiology
To investigate whether our manipulations of sex ste- roid receptor signaling affected circulating levels of sexsteroid hormones in relation to social status, we measuredfree circulating 17␤-estradiol, testosterone, and proges-terone in the plasma of animals that received a receptor FIG. 2. Physiology and circulating hormone levels change in SUB but
antagonist or vehicle only (Fig. 2); individuals that re- not DOM males after steroid receptor manipulation. Free 17␤-estradiol, testosterone, and progesterone levels, as well as GSI, are ceived an agonist were not used for hormone measure- depicted by box and whisker plots for DOM (left panels) and SUB (right ments or gene expression studies. As expected, control panels) males (n ⫽ 8 per status per treatment). Each box represents DOM had higher circulating testosterone (t groups receiving a sex steroid receptor antagonist (x-axis), and the gray 3.7; P ⫽ 0.005) and 17␤-estradiol (t ⫽ background bar depicts the first and third quartiles for control animals.
3.8; P ⫽ 0.003) levels than Letters indicate significance between groups; *, significant difference control SUB as previously reported (36). Progesterone lev- from control with Bonferroni post hoc test.
els did not differ between social states, although there wasa nonsignificant trend to higher levels in DOM (t progesterone, F(3,30) 3.471, P ⫽ 0.03]. Specifically, SUB 2.053; P ⫽ 0.065). Surprisingly, we found that in DOM treated with ER antagonists had lower levels of 17␤-es- treated with a receptor antagonist, circulating levels did tradiol (Dunnett's t test P ⫽ 0.017) and testosterone (Dun- not change for any of the sex steroids compared with con- nett's t test P ⫽ 0.032) compared with controls. Thus, on trols [ANOVA: 17␤-estradiol, F 1.796, P ⫽ 0.169; a physiological level, SUB responded even though they 0.289, P ⫽ 0.83; progesterone, showed little behavioral response to steroid receptor ma- 0.289, P ⫽ 0.436], despite the clear behavioral nipulations within a social community.
responses we observed to receptor antagonist manipula- The GSI, a measure of relative testes mass, was higher in tion. Even more striking, SUB displayed distinct changes in control DOM compared with control SUB (t 2.3; P ⫽ circulating hormone levels with treatment of receptor an- 0.039). Steroid receptor antagonists did not affect the GSI of tagonists [ANOVA: 17␤-estradiol, F 7.106, P ⫽ 0.231; P ⫽ 0.853], yet in SUB, ER antagonist 4.23 ⫻ 10⫺4; testosterone, F 7.106, P ⫽ 0.001; treatment resulted in a significant reduction in GSI [F(3,32)

O'Connell and Hofmann Social Status Predicts Steroid Hormone Action Endocrinology, March 2012, 153(3):1341–1351 FIG. 3. Expression of candidate genes in the POA. A, Expression levels of steroid hormone receptors and neuropeptides in control treated DOM
and SUB males are shown as box and whisker plots. B, Changes in gene expression after treatment of an ER (left panel), AR (middle panel), or PR
(right panel) antagonist in DOM (top row) and SUB (bottom row) males is shown with colored box and whisker plots. Gray background bars depict
the first and third quartiles for control animals. *, P ⬍ 0.05, Dunnett's t test.
sured mRNA levels of all sex steroid receptors as well as SEM: ICI182780 0.391 ⫾ 0.080; control 0.872 ⫾ 0.066], a re- the nonapeptides arginine vasotocin (AVT) and isotocin markable physiological change in such a short time.
(IST), because all these neuroendocrine pathways playimportant roles in modulating social behavior across Sex steroid receptor-mediated gene expression
vertebrates (reviewed in Ref. 37).
impinges on social status
To our surprise, control and experimental groups of To better understand how sex steroid receptors reg- DOM did not differ in the expression of any of these ulate social behavior in a hierarchy, we used qPCR to candidate genes despite the robust changes in social be- analyze the expression of these receptors and neuro- havior we had observed, whereas in SUB, mRNA levels peptides in the POA of males that received either a sex of several candidate genes differed significantly be- steroid receptor antagonist or vehicle (Fig. 3). We mea- tween control and experimental groups. These results Endocrinology, March 2012, 153(3):1341–1351 are consistent with our findings described above of an An even more striking picture emerged in SUB, where apparent dissociation, at least in DOM, of the behav- the P value distribution was surprisingly devoid of small ioral from the physiological responses to sex steroid values (Supplemental Fig. 5B), suggesting a genome- antagonist treatment. However, it should be noted that wide suppression of expression variation in this pheno- microdissection of the POA did not include all gigan- type. At a P ⬍ 0.05 threshold, only 48 (0.59%) of the tocellular neurons and thus may have masked potential features were differentially regulated between the con- differences (see also Ref. 34). In SUB, PR and IST levels trol and ER antagonist groups (LIMMA, P ⬍ 0.05), were sensitive to antagonist treatment [PR: ANOVA, even though SUB demonstrated significant changes in 5.321 P ⫽ 0.006; IST: ANOVA: F(3,28) physiological and candidate gene expression measures, 3.820, P ⫽ 0.022]. Specifically, SUB given an ER, AR, as described above. Finally, the POA transcriptome re- or PR antagonist decreased expression of PR (Dunnett's sponses to ER perturbation showed very little overlap t test: ER antagonist P ⫽ 0.038; AR antagonist P ⫽ between DOM and SUB, because only four genes were 0.005; PR antagonist P ⫽ 0.006), whereas expression of regulated at P ⬍ 0.05 by ER in both datasets (Supple- IST increased after exposure to the AR antagonist (Dun- mental Table 2). These sequences represent novel genes nett's t test P ⫽ 0.02).
for which no annotations could be found in the ER regulation of the social transcriptome
Because perturbation of ER exhibited consistent and significant effects on aggression in both DOM and SUB, Integration of genes, physiology, and behavior
we then asked to which extent the POA gene network To examine the relationship between behavior, hor- regulated by ER differed between social phenotypes. To mone levels, gonadal state, and gene expression in investigate the molecular consequences of ER perturba- DOM and SUB, we used the network analysis platform tion on a genomic scale, we compared the POA transcrip- Cytoscape (38) to create association networks based on tomes of the individuals from the behavioral trials above Spearman correlation coefficients (Supplemental Ta- after administration of either ER antagonist or vehicle. In bles 3 and 4) between measures of behavior, physiology, DOM, 12,676 (71.6%) of 17,712 array features with hormones, and POA candidate gene expression of neu- cichlid sequences yielded above background intensities, roendocrine genes determined by qPCR in DOM (n ⫽ whereas in SUB, there were only 8,132 (45.9%) such fea- 32) and SUB (n ⫽ 32). As can be seen in Fig. 4, only four tures, indicating that the POA expresses about 56% more significant correlations are shared between DOM and genes in DOM compared with SUB (␹2 ⫽ 2405.002; P ⬍ SUB males: ER␣ and ER␤a are highly correlated, as are 0.001). This difference is not a technical artifact, because AVT and IST, 17␤-estradiol levels and GSI, and 17␤- normalized intensity distributions (as a measure of dy- estradiol and progesterone levels.
namic range in gene expression) did not differ substan- This analysis enabled us to propose network modules tially between the DOM and SUB experiments (Supple- that may contribute to specific aspects of A. burtoni mental Fig. 5A). We then compared (separately for DOM sociality, based on what is known about behavior, phys- and SUB) the POA transcription profiles of ER antagonist- iology, and life history of this species. In DOM, border treated animals with those of vehicle controls, although disputes and threat displays are directed only toward none of the differences survived false-discovery correction other DOM (39). Our results suggest that these behav- for P ⬍ 0.05 in either social phenotype, which is not sur- ior patterns are tightly regulated by ER␤a and ER␤b prising given such a subtle perturbation. However, when signaling and together form a territory defense module.
we examined the P value distributions for DOM and SUB, The most striking differences between DOM and SUB we discovered that small P values were considerably over-represented in the DOM dataset (Supplemental Fig. 5B), were the interactions of hormones and physiology with indicating widespread gene regulation. We chose P ⬍ 0.05 behavior and expression of sex steroid receptors in the as an acceptable significance threshold, because our aim POA. In DOM, testosterone, in association with chas- was to compare broad-scale patterns of gene regulation ing and courtship, represents a community interaction across phenotypes. Unexpectedly, despite the almost com- module, because these behavior patterns are directed plete absence of physiological or candidate gene expres- exclusively at the community of females and SUB. How- sion responses to ER antagonist treatment in DOM males, ever, in SUB, testosterone is associated with the other 1,047 (8.25%) of the 12,676 array features that provided steroid hormones and GSI as well as AR subtype ex- signal above background were differentially regulated pression, suggesting that circulating steroid hormones between control vs. ER antagonist (LIMMA, P ⬍ 0.05).
are reflective of gonadal state. Finally, we found strong

O'Connell and Hofmann Social Status Predicts Steroid Hormone Action Endocrinology, March 2012, 153(3):1341–1351 Distinct roles for steroid hormone pathways in
social behavior
Androgens and estrogens have been intensively studied for decades in the context of sexual behavior and aggres-sion because they dynamically influence behavior (40 –42). Within teleosts, the role of androgens in regulatingsexual behavior appears to be consistent across species(43, 44). Of particular relevance to our study is recentwork in mammals by Juntti et al. (45), who showed thatmice with a conditional neuronal AR knockout displayeddeficits in sexual behavior, yet the number of aggressivedisplays did not differ compared with wild-type mice. Re-search in other nonmammalian tetrapods also suggeststhat AR is important in male-typical sexual behavior, in-cluding amphibians (46), birds (47), and reptiles (48). Theconversion of testosterone to estrogen by aromatase is nec-essary for male aggression in mammals and birds (49, 50)as well as for reproductive behavior in some species (51).
Male ER␣ knockout mice also display deficits in sexualbehavior (52), although this may be due to organizationaleffects of estrogen in the developing male brain (53).
Estrogenic and androgenic modulation of behavior varies based on environmental cues (9, 54). However, inA. burtoni males, which breed year-round, variation in ARand ER modulation of behavior likely arises from socialrather than seasonal cues. We have shown here that tes-tosterone modulates sexual behavior in DOM and thatestrogens regulate aggression independent of social status.
Although we cannot rule out the conversion of DHT to3␤-diol (55), we are confident that the androgenic regu-lation of sexual behavior is AR dependent, because theeffects of the AR antagonist were opposite to those ofDHT. We obtained these results using males that were wellestablished in their respective social states and by treatingthem with an agonist or antagonist for several days. How-ever, when SUB are provided with an opportunity to as-cend in social status, aggressive behavior and androgensrise within 30 min of transition, although estrogens do notincrease for several days (7). Androgens may play a short- FIG. 4. Integrating across levels of biological organization: behavior,
term role in modulating aggressive behavior during social physiology, hormones, and gene expression. A covariance network isrepresented in DOM (top) and SUB (bottom) males. Edges represent instability (see also Refs. 8 and 11), whereas estrogens may significant positive (red) or negative (blue) correlations between maintain aggressive motivation over the long term if and behavior (yellow), POA gene expression (green), and physiology when the social environment has stabilized. These differ- (purple). Edges highlighted in gray are common across social states,and dashed lines did not survive false discovery rate correction.
ent temporal roles of androgens and estrogens may also be Functional modules are highlighted in orange. T, Testosterone; E, a consequence of the relatively slow up-regulation of brain estradiol; P, progesterone.
aromatase by AR (56). Interestingly, we found that treat-ment of SUB with DHT did not change aggressive behav- correlations between 17␤-estradiol, progesterone, GSI, ior, although different results may have been obtained if and feeding in DOM, which could be indicative of a SUB were treated with 11KT rather than DHT. Further- module governing reproductive physiology.
more, the dose used to treat SUB was determined based on Endocrinology, March 2012, 153(3):1341–1351 a dose-response curve conducted in DOM, and therefore social opportunities to ascend to DOM status, as previous an alternate dose may also have produced a different be- work has shown a rapid increase in circulating androgens havioral response. Nonetheless, our results suggest that to in response to an opportunity to become territorial (6).
ascend in social status, SUB need both an androgen surge Additionally, several interactions are shared by DOM and as well as a social opportunity, such as a vacant territory SUB (e.g. AVT and IST or ER␣ and ER␤a), suggesting that and/or the removal of suppression by DOM males.
some genes are coregulated in a similar manner indepen- The role of progesterone in the adult male brain is less dent of social status. Finally, although this kind of net- understood than the actions of androgens and estrogens, work analysis is clearly useful for integrating data across although several studies indicate that PR is important in levels of biological organization to visualize large-scale male-typical sexual behavior in mammals (57, 58) and biological patterns, it is important to emphasize that cor- reptiles (59). Our results also show that PR facilitates relation of nodes does not necessarily imply causation and courtship behavior in A. burtoni DOM. In SUB, however, that it will be necessary to experimentally manipulate the treatment with a PR antagonist unexpectedly decreased proposed modules to fully characterize their function in fleeing behavior, which might suggest that the perception regulating behavior in a complex social environment.
or evaluation of social (i.e. threatening) cues is modulated We found several interesting patterns with steroid re- by PR. Work in humans has shown that progesterone, ceptor antagonist treatment and downstream gene expres- along with cortisol, increases in avoidance situations that sion in the POA, particularly the suppression of IST by AR signify fear of rejection (60). Furthermore, treatment with in SUB. Neuropeptides of the oxytocin family have drawn 17␣-20␤-P has antianxiety effects in rodents (61). It is thus a lot of attention in the mammalian literature by regulating possible that PR regulates social cognition in a manner trust (63) and affiliative behavior (64). In A. burtoni, SUB that is conserved across vertebrates, although more studies school (i.e. affiliate) with other SUB and females, yet when across a wide range of taxa will be needed to test this given the opportunity to ascend to dominance (a relatively hypothesis. The work in humans also points to the im- solitary station), schooling behavior is suppressed, possi- portance of the stress axis in social status. Although we did bly via down-regulation of the IST pathway. This idea is not measure cortisol levels or glucocorticoid receptor ex- consistent with a study in goldfish that demonstrated a pression, this pathway does indeed play a pivotal role in A. role for IST in regulating social approach behavior (65).
burtoni behavior (62).
Another interesting relationship is that of androgens, estrogens, and expression of their receptors in the POA, Integrating brain gene expression with behavior
which seem to play very different roles depending on an individual's position in the social hierarchy. SUB have We have used a covariation network analysis to orga- lower expression of ER compared with DOM, and the nize data across levels of biological organization and pro- lack of a transcriptome response in SUB compared with pose functional modules that allow us to generate new DOM follows this pattern. Inhibition of ER mechanisms hypotheses about the neuroendocrine and molecular basis in SUB after administration of an ER antagonist, in com- of social behavior. Here we discuss correlations between bination with low circulating testosterone levels, and the behavior, physiology, and gene expression in the POA, a resulting absence of brain AR activation, appears to lead brain region highly conserved in function across verte- to a remarkable genome-wide suppression of both tran- brates (20), although these patterns may well differ when scriptional activity and variation in the POA.
other brain regions are analyzed. Our results suggest dis-tinct modules that may govern different aspects of behav- Searching for patterns across levels of biological
ior within a social group. DOM display behavior patterns toward other DOM that are closely associated with ex- In an era when molecular tools have become available pression of ER, which is supported by our pharmacolog- for nontraditional model systems to explore the under- ical results where ER manipulations altered these territory pinnings of complex behavioral phenotypes, few research- defense behaviors. Interactions with other community ers have analyzed a particular phenotype across many lev- members, including females and SUB, appear to involve a els of biological organization. Our results show for the different module that is dominated by androgens. We hy- first time how in individuals of the same species, biological pothesize that this is due to the dynamic response of an- responses to a perturbation can differ across levels of bi- drogens to social cues (6, 11, 12) that may then modulate ological organization as well as between phenotypes. We courtship or chasing behavior. SUB are quite different, in are not aware of any other study to date that has integrated that their hormones are tightly linked to gonadal state and behavior, hormones, physiology, candidate gene expres- AR expression, which may enable them to quickly seize sion, and transcriptome profiling within the same individ- O'Connell and Hofmann Social Status Predicts Steroid Hormone Action Endocrinology, March 2012, 153(3):1341–1351 uals, yet a complete understanding of the mechanisms un- social status in the African cichlid, Astatotilapia burtoni. Behav derlying behavioral variation across individuals might Brain Res 166:291–295 3. Burmeister SS, Jarvis ED, Fernald RD 2005 Rapid behavioral and
require the kind of integration across multiple levels of genomic responses to social opportunity. PLoS Biol 3:e363 biological organization we have attempted here. Further- 4. Greenwood AK, Wark AR, Fernald RD, Hofmann HA 2008 Ex-
more, given that mRNA and protein need not always be pression of arginine vasotocin in distinct preoptic regions is associ-ated with dominant and subordinate behaviour in an African cichlid concordant (66), it will be important to include analyses of fish. Proc Biol Sci 275:2393–2402 the proteome and small RNAs in future studies as well.
5. Gesquiere LR, Learn NH, Simao MC, Onyango PO, Alberts SC,
Our results show that phenotypic changes at one level Altmann J 2011 Life at the top: rank and stress in wild male baboons.
Science 333:357–360
of biological organization do not predict changes of com- 6. Maruska KP, Fernald RD 2010 Behavioral and physiological plas-
parable magnitude at another level, although it is impor- ticity: rapid changes during social ascent in an African cichlid fish.
tant to note that gene expression may be more reflective of Horm Behav 58:230 –240 7. Oliveira RF, Hirschenhauser K, Carneiro LA, Canario AV 2002
behavior than circulating hormone levels. Our expecta- Social modulation of androgen levels in male teleost fish. Comp tion at the beginning of this study was that a behavioral Biochem Physiol B Biochem Mol Biol 132:203–215 response to sex steroid receptor manipulations would be 8. Trainor BC, Rowland MR, Nelson RJ 2007 Photoperiod affects
accompanied by corresponding changes in circulating estrogen receptor ␣, estrogen receptor ␤ and aggressive behavior.
Eur J Neurosci 26:207–218 hormone levels, candidate gene expression, and transcrip- 9. Goymann W 2009 Social modulation of androgens in male birds.
tome activity, regardless of social status. However, we Gen Comp Endocrinol 163:149 –157 found marked differences in DOM and SUB responses at 10. Wingfield JC, Hegner RE, Dufty AM, Jr., Ball GF 1990 The "chal-
lenge hypothesis": theoretical implications for patterns of testoster- various levels. These discrepancies could, in part, be ex- one secretion, mating systems, and breeding strategies. Am Nat 136: plained by differences in gene expression of sex steroid receptors in some brain regions (67) or in the testes (68), 11. Hirschenhauser K, Oliveira RF 2006 Social modulation of andro-
gens in male vertebrates: meta-analyses of the "challenge hypothe- by effects of the community structure on the behavioral sis." Anim Behav 71:265–277 expressions of DOM and SUB males, or most likely a com- 12. Gleason ED, Fuxjager MJ, Oyegbile TO, Marler CA 2009 Testos-
bination of all these factors.
terone release and social context: when it occurs and why. FrontNeuroendocrinol 30:460 – 469 13. Wahli W, Martinez E 1991 Superfamily of steroid nuclear receptors:
positive and negative regulators of gene expression. FASEB J 14. Remage-Healey L, Bass AH 2006 A rapid neuromodulatory role for
We thank Lin Huffman, Aubrey Kelly, Mary Ramsey, Deena steroid hormones in the control of reproductive behavior. Brain Res1126:27–35 Walker, and Keith Whitaker for helpful comments on earlier 15. Mani SK, Portillo W, Reyna A 2009 Steroid hormone action in the
versions of this manuscript; Lin Huffman, Suzy Renn, and Anna brain: cross-talk between signalling pathways. J Neuroendocrinol Sessa for valuable advice on the microarray analysis; and mem- bers of the Hofmann laboratory for discussions. The PR antag- 16. Hofmann HA 2003 Functional genomics of neural and behavioral
plasticity. J Neurobiol 54:272–282 onist was a kind gift from Bayer Schering Pharma AG, Berlin 17. Robinson GE, Fernald RD, Clayton DF 2008 Genes and social be-
havior. Science 322:896 –900 18. Renn SC, Aubin-Horth N, Hofmann HA 2008 Fish and chips: func-
Address all correspondence and requests for reprints to: Dr.
tional genomics of social plasticity in an African cichlid fish. J Exp Hans A. Hofmann, Section of Integrative Biology, The Univer- Biol 211:3041–3056 sity of Texas at Austin, 1 University Station C0930, Austin, 19. Newman SW 1999 The medial extended amygdala in male repro-
Texas 78712. E-mail: [email protected].
ductive behavior. A node in the mammalian social behavior net-work. Ann NY Acad Sci 877:242–257 This work was supported by National Science Foundation 20. O'Connell LA, Hofmann HA 2011 The vertebrate mesolimbic re-
Doctoral Dissertation Improvement Grant 1011253 to L.A.O., ward system and social behavior network: a comparative synthesis.
NSF Grant 0843712, the Alfred P. Sloan Foundation, and a J Comp Neurol 519:3599 –3639 Dwight W. and Blanche Faye Reeder Centennial Fellowship in 21. Munchrath LA, Hofmann HA 2010 Distribution of sex steroid hor-
mone receptors in the brain of an African cichlid fish, Astatotilapia Systematic and Evolutionary Biology and Institute for Cellular burtoni. J Comp Neurol 518:3302–3326 and Molecular Biology Fellowship to H.A.H.
22. Fernald, RD 1977 Quantitative behavioural observations of Hap-
Disclosure Summary: The authors have nothing to disclose.
lochromis burtoni under semi-natural conditions. Anim Behav 25:643– 653 23. Wells K, and Van Der Kraak G 2000 Differential binding of endog-
enous steroids and chemicals to androgen receptors in rainbow trout and goldfish. Environ Toxicol Chem 19:2059 –2065 24. Sperry TS, Thomas P 1999 Characterization of two nuclear andro-
1. Newman ML, Sellers JG, Josephs RA 2005 Testosterone, cognition,
gen receptors in Atlantic croaker: comparison of their biochemical and social status. Horm Behav 47:205–211 properties and binding specificities. Endocrinology 140:1602–1611 2. Parikh VN, Clement TS, Fernald RD 2006 Androgen level and male
25. Hawkins MB, Thomas P 2004 The unusual binding properties of the
Endocrinology, March 2012, 153(3):1341–1351 third distinct teleost estrogen receptor subtype ER␤a are accompa- Schlinger BA 2007 Androgen and the elaborate courtship behavior
nied by highly conserved amino acid changes in the ligand binding of a tropical lekking bird. Horm Behav 51:62– 68 domain. Endocrinology 145:2968 –2977 48. Tokarz RR 1986 Hormonal regulation of male reproductive behav-
26. Pinter J, Thomas P 1997 The ovarian progestogen receptor in the
ior in the lizard Anolis sagrei: a test of the aromatization hypothesis.
spotted seatrout, Cynoscion nebulosus, demonstrates steroid spec- Horm Behav 20:364 –377 ificity different from progesterone receptors in other vertebrates.
49. Schlinger BA, Callard GV 1990 Aromatization mediates aggressive
J Steroid Biochem 60:113–119 behavior in quail. Gen Comp Endocrinol 79:39 –53 27. Kidd CE, Kidd MR, Hofmann HA 2010 Measuring multiple hor-
50. Trainor BC, Kyomen HH, Marler CA 2006 Estrogenic encounters:
mones from a single water sample using enzyme immunoassays. Gen how interactions between aromatase and the environment modulate Comp Endocrinol 165:277–285 aggression. Front Neuroendocrinol 27:170 –179 28. Braford Jr MR, Northcutt RG 1983 Organization of the dienceph-
51. Adkins-Regan E, Garcia M 1986 Effect of flutamide (an antiandro-
alon and pretectum of ray-finned fishes. In: David RG, Northcutt gen) and diethylstilbestrol on the reproductive behavior of Japanese RG, eds. Fish neurobiology. Vol 2. Ann Arbor, MI: University of quail. Physiol Behav 36:419 – 425 Michigan Press; 117–163 52. Wersinger SR, Sannen K, Villalba C, Lubahn DB, Rissman EF, De
29. Simon P 2003 Q-Gene: processing quantitative real-time RT-PCR
Vries GJ 1997 Masculine sexual behavior is disrupted in male and
data. Bioinformatics 19:1439 –1440 female mice lacking a functional estrogen receptor alpha gene. Horm 30. Salzburger W, Renn SC, Steinke D, Braasch I, Hofmann HA, Meyer
Behav 32:176 –183 A 2008 Annotation of expressed sequence tags for the East African
53. Kudwa AE, Michopoulos V, Gatewood JD, Rissman EF 2006 Roles
cichlid fish Astatotilapia burtoni and evolutionary analyses of of estrogen receptors alpha and beta in differentiation of mouse cichlid ORFs. BMC Genomics 9:96 sexual behavior. Neuroscience 138:921–928 31. Liang K, Zeger SL 1986 Longitudinal data analysis using generalized
54. Wingfield JC, Lynn S, Soma KK 2001 Avoiding the ‘costs' of tes-
linear models. Biometrika 73:13–22 tosterone: ecological bases of hormone-behavior interactions. Brain 32. Smyth GK 2005 Limma: Linear models for microarray data. In:
Behav Evol 57:239 –251 Gentleman R, Carey V, Dudoit S, Irizarry R, Huber W, eds. Bioin- 55. Mouriec K, Gueguen MM, Manuel C, Percevault F, Thieulant ML,
formatics and computational biology solutions using R and Biocon- Pakdel F, Kah O 2009 Androgens upregulate cyp19a1b (aromatase
ductor. New York, NY: Springe; 397– 420 B) gene expression in the brain of zebrafish (Danio rerio) throughestrogen receptors. Biol Reprod 80:889 – 896 33. Hoegg S, Brinkmann H, Taylor JS, Meyer A 2004 Phylogenetic
56. Abdelgadir SE, Resko JA, Ojeda SR, Lephart ED, McPhaul MJ, Roselli
timing of the fish-specific genome duplication correlates with the CE 1994 Androgens regulate aromatase cytochrome P450 messenger
diversification of teleost fish. J Mol Evol 59:190 –203 ribonucleic acid in rat brain. Endocrinology 135:395–401 34. Zakon HH, Dunlap KD 1999 Sex steroids and communication sig-
57. Witt DM, Young LJ, Crews D 1994 Progesterone and sexual be-
nals in electric fish: a tale of two species. Brain Behav Evol 54:61– 69 havior in males. Psychoneuroendocrinology 19:553–562 35. Margiotta-Casaluci L, Sumpter JP 2011 5␣-Dihydrotestosterone is
58. Phelps SM, Lydon JP, O'malley BW, Crews D 1998 Regulation of
a potent androgen in the fathead minnow (Pimephales promelas).
male sexual behavior by progesterone receptor, sexual experience, Gen Comp Endocrinol 171:309 –318 and androgen. Horm Behav 34:294 –302 36. Maruska KP, Fernald RD 2010 Steroid receptor expression in the
59. Lindzey J, Crews D 1988 Effects of progestins on sexual behaviour
fish inner ear varies with sex, social status, and reproductive state.
in castrated lizards (Cnemidophorus inornatus). J Endocrinol 119: BMC Neurosci 11:58 37. Goodson JL, Thompson RR 2010 Nonapeptide mechanisms of so-
60. Wirth MM, Schultheiss OC 2006 Effects of affiliation arousal (hope
cial cognition, behavior and species-specific social systems. Curr of closeness) and affiliation stress (fear of rejection) on progesterone Opin Neurobiol 20:784 –794 and cortisol. Horm Behav 50:786 –795 38. Kohl M, Wiese S, Warscheid B 2011 Cytoscape: software for visu-
61. Picazo O, Ferna´ndez-Guasti A 1995 Anti-anxiety effects of proges-
alization and analysis of biological networks. Methods Mol Biol terone and some of its reduced metabolites: an evaluation using the burying behavior test. Brain Res 680:135–141 39. Baerends GP, Baerends-van Roon JM 1950 An introduction to the
62. Fox HE, White SA, Kao MH, Fernald RD 1997 Stress and domi-
study of the ethology of cichlid fishes. Behavior Suppl 1:1–242 nance in a social fish. J Neurosci 17:6463– 6469 40. Beach FA 1948 Hormones and behavior. New York: Paul B. Hoeber
63. Kosfeld M, Heinrichs M, Zak PJ, Fischbacher U, Fehr E 2005 Oxy-
41. Lehrman DS 1965 Interaction between internal and external envi-
tocin increases trust in humans. Nature 435:673– 676 ronments in the regulation of the reproductive cycle of the ring dove.
64. Insel TR, Shapiro LE 1992 Oxytocin receptor distribution reflects
Sex and Behavior. Beach FA, ed. New York: Wiley; 355–380 social organization in monogamous and polygamous voles. Proc 42. Ball GF, Balthazart J 2004 Hormonal regulation of brain circuits
Natl Acad Sci USA 89:5981–5985 mediating male sexual behavior in birds. Physiol Behav 83:329 –346 65. Thompson RR, Walton JC 2004 Peptide effects on social behavior:
43. Salek SJ, Sullivan CV, Godwin J 2001 Courtship behavior of male
effects of vasotocin and isotocin on social approach behavior in male white perch, Morone americana: evidence for control by androgens.
goldfish (Carassius auratus). Behav Neurosci 118:620 – 626 Comp Biochem Physiol A Mol Integr Physiol 130:731–740 66. Vogel C, Abreu Rde S, Ko D, Le SY, Shapiro BA, Burns SC, Sandhu
44. Sebire M, Allen Y, Bersuder P, Katsiadaki I 2008 The model anti-
D, Boutz DR, Marcotte EM, Penalva LO 2010 Sequence signatures
androgen flutamide suppresses the expression of typical male stick- and mRNA concentration can explain two-thirds of protein abun- leback reproductive behaviour. Aquat Toxicol 90:37– 47 dance variation in a human cell line. Mol Syst Biol 6:400 45. Juntti SA, Tollkuhn J, Wu MV, Fraser EJ, Soderborg T, Tan S,
67. Burmeister SS, Kailasanath V, Fernald RD 2007 Social dominance
Honda S, Harada N, Shah NM 2010 The androgen receptor governs
regulates androgen and estrogen receptor gene expression. Horm the execution, but not programming, of male sexual and territorial Behav 51:164 –170 behaviors. Neuron 66:260 –272 68. Maruska KP, Fernald RD 2011 Plasticity of the reproductive axis
46. Behrends T, Urbatzka R, Krackow S, Elepfandt A, Kloas W 2010
caused by social status change in an african cichlid fish. II. Testicular Mate calling behavior of male South African clawed frogs (Xenopus gene expression and spermatogenesis. Endocrinology 152:291–302 laevis) is suppressed by the antiandrogenic endocrine disrupting 69. Renn SCP, Aubin-Horth N, Hofmann HA 2004 Biologically mean-
compound flutamide. Gen Comp Endocrinol 168:269 –274 ingful expression profiling across species using heterologous hybrid- 47. Fusani L, Day LB, Canoine V, Reinemann D, Hernandez E,
ization to a cDNA microarray. BMC Genomics 5, 42

Source: http://oconnell.fas.harvard.edu/files/oconnell/files/oconnellhofmann2012.pdf