Kutsukake et al 2009 AJP

RESEARCH ARTICLE

Validation of Salivary Cortisol and Testosterone Assays in Chimpanzees

by Liquid Chromatography-Tandem Mass Spectrometry

NOBUYUKI KUTSUKAKE¹, KOKI IKEDA², SEIJIRO HONMA³, MIGAKU TERAMOTO⁴, YUSUKE MORI⁴, IKUO HAYASAKA⁴, RAIN YAMAMOTO⁵, TAKAFUMI ISHIDA⁵, YASUHIRO YOSHIKAWA⁶,

ANDTOSHIKAZU HASEGAWA²

1Department of Evolutionary Studies of Biosystems, The Graduate University for Advanced Studies, Hayama, Kanagawa, Japan

2Department of Cognitive and Behavioral Science, The University of Tokyo, Tokyo, Japan

3Aska Pharma Medical, Kawasaki, Japan

4Chimpanzee Sanctuary Uto, Kumamoto, Japan

5Department of Anthropology, Laboratory of Human Genetics, The University of Tokyo, Tokyo, Japan

6Department of Biomedical Science, The University of Tokyo, Tokyo, Japan

Owing to its high temporal sensitivity, saliva has distinct advantages for measuring steroids, compared with other noninvasive samples such as urine and feces. Here, we report the validity of assaying salivary cortisol (C) and testosterone (T) using liquid chromatography-tandem mass spectrometry (LC- MS/MS) in captive male chimpanzees,Pan troglodytes. For both the C and T concentrations, we found positive relationships between saliva and plasma. The concentrations of C and T in saliva showed clear patterns of diurnal fluctuation, whereas those in urine and feces did not. These results suggest that the salivary steroid concentrations can be regarded as good indicators of circulating steroid levels. We also developed and validated an efficient method for collecting saliva samples from cotton rope. Although rope includes inherent steroid-like compounds and may affect the accuracy of steroid measurements, our rope-washing procedures effectively removed intrinsic steroidal materials. There was a significant association between the C and T concentrations measured from saliva collected from rope licked by the chimpanzees and those measured from saliva collected directly from the mouth. Salivary T values estimated by LC/MS-MS were similar to those measured by radioimmunoassay. The results indicate the usefulness of saliva as a noninvasive steroid measure and that steroids in the saliva of chimpanzees can be accurately measured by LC-MS/MS. Am. J. Primatol. 71:696–706, 2009. r2009 Wiley-Liss, Inc.

Key words: salivary steroids; testosterone; cortisol; chimpanzees; liquid chromatography-tandem mass spectrometry (LC-MS/MS)

INTRODUCTION

Saliva is an excellent physiological medium for the noninvasive investigation of steroid hormone endocrinology in both humans and nonhuman animals [Granger et al., 1999; Gro¨schl, 2008; Kirschbaum & Hellhammer, 1994; Laudenslager et al., 2006; Lewis, 2006]. The salivary steroid concentration is thought to reflect the concentration of plasma free-fraction steroids that are not bound to carrier proteins and are thus biologically active [Gozansky et al., 2005; Lac, 2001]. The steroid concentrations in the urine and feces reflect the blood concentrations of metabolites over time [Palme, 2005] and are useful in investigating cumulative and chronic aspects of endocrine physiol- ogy, whereas the saliva is very sensitive to acute changes in blood hormone concentrations and to diurnal fluctuations. Therefore, researchers can investigate the sequential and immediate changes in internal endocrine physiology by repeatedly collecting saliva samples following a schedule predetermined by

researchers, which cannot be done for urine and feces. Such sequential measurements are indispensable for investigating causal relationships between hormones and behavior. Nevertheless, urine and fecal sampling is commonly used in captivity to measure steroids in nonhuman primates [Laudenslager et al., 2006; Whitten et al., 1998b], because relatively few salivary cortisol (C) [Behringer et al., 2009; Boyce et al., 1995; Cross & Rogers, 2004; Cross et al., 2004; Elder &

Published online 18 May 2009 in Wiley InterScience (www. interscience.wiley.com).

DOI 10.1002/ajp.20708

Received 8 February 2009; revised 12 April 2009; revision accepted 20 April 2009

Additional Supporting Information may be found in the online version of this article.

Correspondence to: Nobuyuki Kutsukake, Department of Evolutionary Studies of Biosystems, The Graduate University for Advanced Studies, Hayama, Kanagawa 240-0193, Japan. E-mail: [email protected]

Menzel, 2001; Fuchs et al., 1997; Hohmann et al., 2009; Kuhar et al., 2005; Lutz et al., 2000; Pearson et al., 2008; Tiefenbacher et al., 2003] and testosterone (T) [von Engelhardt et al., 2000] assays have been developed in nonhuman animals.

In this study, we report the validation of C and T salivary assays using liquid chromatography-tandem mass spectrometry (LC-MS/MS) in the chimpanzee, Pan troglodytes. Previous studies of endocrine physio- logy in captive chimpanzees have commonly measured urinary or fecal steroid concentrations [e.g., Anestis, 2005; Bahr et al., 2000; Hauser et al., 2008a,b; Mohle et al., 2002; Whitten et al., 1998a]. This is the first study measuring salivary steroid concentrations in chimpanzees. The commonly used radioimmunoassay (RIA) and enzyme immunoassay (EIA) are convenient and cost effective. However, LC-MS/MS has a remark- able advantage for measuring small quantities of steroids that cannot be measured using an immuno- assay. LC-MS/MS enables one to assay two or more kinds of steroid simultaneously (see Methods). More- over, the results of the immunoassay are often biased by antibody cross-reactivity to substances structurally similar to the target substances [e.g., Wang et al., 2004]. In this study, we overcame this complication by using LC-MS/MS to determine the salivary C and T concentrations simultaneously.

First, we tested the biological validity of measur- ing the concentrations of C and T in saliva by comparison to the steroid concentrations in plasma. In addition, we assayed the plasma bioavailable testosterone (BAT) concentration, which is the fraction of circulating T that is not bound to sex hormone-binding globulin (SHBG) [Emadi-Konjin et al., 2003]; therefore, BAT is a more accurate reflection of biologically active T than is total plasma T [Morley et al., 2002]. The relationships to the BAT concentration are the best criteria to assess the biological relevance of the steroid concentrations in each medium. Moreover, diurnal fluctuations in the steroid concentrations in saliva were investigated and compared with those in urine and feces. Both C and T exhibit circadian rhythms in humans

[Rose et al., 1972] as well as in chimpanzees [Muller

& Lipson, 2003], with the highest peak in the early morning and a lower nadir from evening to night. Investigation of the diurnal fluctuation will eluci- date whether chimpanzee salivary steroids can be regarded as biologically relevant markers of chim- panzee internal physiology.

As reported in the latter part of this paper, we also developed and validated a cotton-rope washing method that allows accurate determination of salivary C and T from saliva samples obtained from rope used by the chimpanzees (Experiment 2). To date, use of a cotton swab or rope as a collecting medium has been widely accepted for C assays in primates [Boyce et al., 1995; Cross & Rogers, 2004; Cross et al., 2004; Elder

& Menzel, 2001; Fuchs et al., 1997; Lutz et al., 2000; Tiefenbacher et al., 2003] as well as in humans [Kirschbaum & Hellhammer, 1994]. However, the cotton sometimes interferes and leads to errors in steroid concentration estimates, probably because some intrinsic steroidal materials cross-react with steroid antibodies in an immunoassay or the steroids adhere to the cotton fibres [Dabbs, 1991; Granger et al., 2004; Shirtcliff et al., 2001]. In Experiment 2, we elaborated the washing method to remove intrinsic steroidal materials from the raw cotton material, and tested the validity of this collection method by comparing the salivary steroid concentrations obtained via wash-treated cotton ropes with matched saliva obtained directly from chimpanzees’ mouths. In the final experiment (Experiment 3), we tested the consistency of estimates of the salivary T assay between the LC-MS/MS method and the RIA method that is commonly used in the previous studies.

METHODS

General: Animals and Housing

Ten adolescent and adult male chimpanzees housed in Chimpanzee Sanctuary Uto (CSU; Misumi, Kumamoto, Japan) were used for the experiments (Table I). The animals were fed monkey pellets supplemented with fresh fruits and vegetables. Water

TABLE I. The male chimpanzees used in the experiments

Experiment Chimpanzee ID Birth year

Age at the time of the experiment^

Weight (kg) Born

1 Yukio 1987 17 66.3 Captive

1 Ken 1983 21 55.6 Captive

1 Mizuo 1989 15 59.5 Captive

1, 2, 3 Kotaro 1993 11 (14) 65 Captive

2 Maruku 1980? 24 58.7 Wild

2 Lucky 1974 30 52.9 Wild

2 Kazuya 1987 17 62.6 Captive

2, 3 Kenji 1984 20 (23) 57.6 Captive

2, 3 Gou 1977? 27 (30) 56.7 Wild

3 George 1979? 25 (28) 45.1 Wild

Age in 2007 shown in parentheses.

was provided ad libitum. All housing was illuminated with indoor room lights from 0800 to 1800 hr and with natural sunlight through windows. Prior approval for this study was obtained from the Screening Commit- tee for Chimpanzee Experiment of CSU (Chimpanzee Sanctuary Uto). This study complied with applicable national laws.

All steroid assays were conducted at Aska Pharma Medical (Kawasaki, Japan).

Experiment 1 Animals

In Experiment 1, we aimed to validate the salivary steroid assays. Experiment 1 was conducted in March 2004. Four males, whose saliva could be collected directly after one week of training (see Supporting Information), were used for the experi- ment (Table I).

Sample collection

Two of the four individuals (Mizuo and Yukio) have been trained to accept venipuncture for anes- thesia and to permit blood collection voluntarily. Blood samples were collected twice (at 0900 and 1300 hr) on the second day of each of two 4-day sampling periods (8 days in total). There was a 3-day interval between the sampling periods. Blood samples were collected in heparinized tubes, and the plasma was separated and frozen at 701C until assayed.

Saliva, urine, and feces were collected three times per day (at 0900, 1300, and 1700 hr) for 8 days, from the animals’ individual cages (base: 1.84  1.91 m²; height: 2.74 m) where males always stay when they are not in an outdoor cage and where they sleep every night. Saliva was collected directly from the mouth of trained individuals by using a poly- propylene syringe (see Supporting Information). All saliva samples were frozen within 1 hr of collection and stored at 501C until assayed.

The urine samples were taken from the urine collected over the time periods between the sampling times. To provide for urine collection, the floor of the animal’s bedroom had a gentle slope with small steps, allowing the urine to flow into a collecting area, and the cages had stainless steel floors with slits through which the urine fell onto a corrugated plastic sheet sloped downward toward a plastic pool. For 15 min at 45-min intervals, an automated vacuum pump sucked the urine into a large beaker. The pooled urine was collected and measured at each sampling time, and 50 mL from each pool were transferred to a poly- propylene tube and stored frozen at _{501C until} assayed. The remainder of the sample was discarded. Thus, each morning sample was 50 mL of the urine pool that had been collected for 16 hr, from 1700 hr the evening before until 0900 hr that day. Similarly, the afternoon and evening samples were taken from urine pools that had been collected for 4 hr, from 0900

to 1300 hr for the afternoon sample and from 1300 to 1700 hr for the evening sample.

The feces present at each sampling time were collected into a plastic bag and treated as one sample for later analysis. The feces were weighed, and an amount of water having the same weight was added. Following homogenization with an Ultra-Turrax homogeniser (IKA, Staufen, Germany), 50 mL of the homogenate were collected into a polypropylene tube and frozen at 501C until assayed.

All the devices used for sampling were thoroughly washed and dried after each use. The bedrooms and cages were cleaned daily, immediately after the morn- ing sampling. Some of the samples were too small for the assay and were excluded from the analysis.

Cortisol extraction

Plasma samples (100 mL) were thawed and diluted with 400 mL of water. The saliva samples were precisely measured between 0.6 and 1.0 mL. For both sample types, 5 mL of 10% benzene:ethyl acetate (5:95, v/v) were added to each sample, and the samples were shaken for 10 min. The organic layer was separated by freezing in a dry-ice/ethanol bath. The extract was evaporated to dryness in a centrifugal vacuum concentrator at 371C and was reconstituted with 200 mL of phosphate-buffered saline (PBS, pH 7.2) containing 0.1% bovine serum albumin (BSA). Each sample (50 mL) was added to a well of a microplate for EIA.

The urine and fecal samples were subjected first to enzyme hydrolysis. For the urine samples, 500 mL of 50 mM acetate buffer (pH 5.0) and 50 mL (5,000 U) of b-glucuronidase (Helix Pomatia catalogue no. 70022923; Roche Diagnostics, Tokyo, Japan) were added to a 200-mL aliquot of each sample. For the fecal samples, an amount of each sample between 10 and 15 g was precisely weighed, 5 mL of water were added, and the mixture was homogenized by vortexing. To 500 mL of each homogenized sample, 500 mL of 50 mM acetate buffer and 100 mL (10,000 U) of b-glucuronidase were added. Both the urine and fecal samples were incubated for 18 hr at 371C. To each hydrolysis sample was added 5 mL of 10% benzene: ethyl acetate (5:95, v/v) and the samples were shaken for 10 min. The organic layer was separated by freezing in a dry-ice/ethanol bath; the organic phase was decanted into another tube. To the obtained phase, 0.5 mL of 0.1 N NaOH was added, followed by vortexing, and the ether phase was separated again by freezing. The extract was evaporated to dryness in a centrifugal vacuum concentrator at 371C and was reconstituted in 2 mL of PBS-BSA buffer. From each sample, 50 mL were used for EIA.

Cortisol EIA

The EIA was performed using commercially available kits (Enzaplate C; Bayer Medical). The

saliva samples were serially diluted in parallel to the C standard curve (r 5 0.99) over a range of 0.19–25 ng/mL. The mean C recovery was 100.3476.34%. The intra and inter-assay coefficients of variation were 8.1 and 6.0%, respectively, for the low-concentration control. The assay sensitivity was 0.039 ng/mL. The cross-reactivities were 14.2% for 21-deoxycortisol, 11.5% for corticosterone, 8.5% for cortisone, and o1.0% for 11-deoxycortisol, deoxy- corticosterone, and aldosterone.

In a previous study using high-performance liquid chromatography (HPLC) immunograms [Palme, 2005] and injected radiolabelled steroids, only a small fraction of authentic C was detectable in the urine, and C was undetectable in chimpanzee feces, with almost all of the injected radiolabelled steroids being present as fecal metabolites [Bahr et al., 2000; Whitten et al., 1998a]. Therefore, the immunoreactive C measured in the urine and feces in this study probably included immunoreactive glucocorticoid me- tabolites in addition to authentic C. Given that most previous studies on chimpanzees that measured the C concentration used the EIA or RIA method with the problem of cross-reactivity [Anestis, 2005; Muehlenbein, 2006; Muller & Lipson, 2003; Muller

& Wrangham, 2004; Whitten et al., 1998b], and to compare the results of the LC-MS/MS method with the commonly used RIA method, we expressed the measured immunoreactive C as the ‘‘C concentra- tion’’ of the urine and fecal samples for convenience.

Testosterone extraction

Plasma samples (100 mL) were mixed with T-d3 (1 ng) as an internal standard and were diluted with 400 mL of water. Saliva samples (0.5–1.0 mL) were also mixed with T-d3 (1 ng) as an internal standard. Diethyl ether (5 mL) was added to each sample, and the mixtures were shaken for 10 min. The organic layer was separated by freezing, evaporated under nitrogen gas at 371C, dried for 1 hr under low pressure, subjected to derivatization (described below), and then assayed by LC-MS/MS. The urine and fecal samples were hydrolyzed with b-glucuronidase as described above for the C assay, except that the extracted ether phase was evaporated under nitrogen gas at 401C and dried for 1 hr under low pressure. The samples were derivatized to increase sensitivity, and were assayed as described for the plasma and saliva samples.

Plasma BAT extraction

Concanavalin A (Amersham Biosciences, Piscataway, NJ) was used to precipitate SHBG and SHBG-bound T from the plasma [Honma et al., 2004]. One hundred microlitres of 5% concanavalin A were added to 200 mL of each plasma sample; the sample was slowly mixed and incubated for 1 hr at room tempera- ture. PBS (1,000 mL) was added to the sample, and it was mixed and centrifuged at 3,000 rpm for 10 min

at 41C. A 1,000-mL aliquot of the supernatant was used for BAT extraction, according to the method described above for T extraction. BAT concentrations measured by this method were highly correlated (r 5 0.96, Po0.001) [Honma et al., 2004] with those measured by the method commonly used in previous studies [Vermeulen et al., 1999].

Testosterone LC-MS/MS

Each sample was mixed with 200 mL of 2% 2-fluoro-1-methylpyridinium-p-toluenesulfonate (Tokyo Kasei Kogyo; Tokyo, Japan) in dichloromethane and 30 mL of 10% triethylamine/dichloromethane solution and then maintained at room temperature for 1.5 hr. The solution was evaporated under nitrogen gas at 371C and dried for 1 hr under low pressure. The derivatised T was extracted by passing the sample through a C18 solid-phase extraction cartridge (Bond Elut^s; Varian, Palo Alto, CA). The sample was dissolved with 20% methanol/water and applied to the cartridge, which had been conditioned continu- ously with 6 mL each of methanol and water. The cartridge was washed with 3 mL of water, 3 mL of 0.3% ammonia, 2 mL of methanol, and 3 mL of 0.01% formic acid:methanol (1:1, v/v). Then, T was eluted with 2.5 mL of 10% formic acid:acetonitrile (2:8, v/v). The eluate was evaporated, and the dried sample was reconstituted with 100 mL of 0.05% formic acid:methanol (6:4, v/v). Sample solution (20 mL) was injected into an LC-MS/MS.

The LC-MS/MS analysis was performed using a Micromass Quattro II triple quadrupole mass spec- trometer (Nihon Waters, Tokyo, Japan) equipped with an electrospray ionization (ESI) interface and coupled to an Agilent HP1100 column. The HPLC column was an Atlantis^s dC18 (pore size, 5 mm; 50 mm  2.1 mm i.d.; Nihon Waters), the column temperature was 401C, and the flow rate was 0.2 mL/min. Gradient elution was performed using a binary solvent mixture, where solvent A was 0.05% formic acid and solvent B was methanol. The gradient program began at 40% B for 2 min, ramped to 60% B at 3 min, returned to 40% B after 3.01 min, and then held for 7 min. The MS-MS spectra were recorded in the positive mode using the ESI inter- face. The transitions were monitored at m/z 380.3 to 253 for T and at m/z 383.2 to 256 for T-d3. The limit of quantification was 5.0 pg. The quality of the measurement was ensured by checking the quality control samples continuously. For samples contain- ing 5, 500, and 1,000 pg of T, the accuracy was 91.6, 90.8, and 99.4%, and the precision (intra-assay coefficient of variance) was 10.9, 3.9, and 0.6%, respectively (n 5 5 for each). The inter-assay coeffi- cient of variance of the low-concentration control (5 pg) was 6.2% (n 5 15; data collected for three days, in which five samples were collected per day). The

comparison between the predicted and observed values showed parallel responses (_{r 5 0.99).}

Experiment 2 Animals

In Experiment 2, we aimed to assess the validity of the assay for steroids from saliva that had been collected from a rope [Lutz et al., 2000; Tiefenbacher et al., 2003] used by the chimpanzees. Experiment 2 was conducted from September to October 2004. We used six individuals trained to allow saliva collection (see Supporting Information and Table I).

Validation of the rope wash method

The rope was soaked in distilled water for 15 hr, soaked again in water at 60–701C for 5 hr, and then soaked twice in 70% ethanol at 801C for 3 hr. The washed rope was dried at 801C.

We tested the effectiveness of the wash treat- ment in removing intrinsic steroidal materials. Ropes of untreated natural cotton (n 5 3) and ropes that had been washed as above (n 5 15) were each cut into 10-cm pieces, and the pieces were soaked in distilled water for 15 hr. The water absorbed into the rope was extracted by centrifugation, and 12 mL of the extract were assayed for hormones.

We also evaluated the effect of drink-mix crystals (Pocari Sweat^s; Otsuka Pharmaceutical, Tokyo, Japan), which were used to attract the chimpanzees to the rope (see below), on the steroid assays. After the wash treatment, a rope was disentangled, dipped in a solution of drink-mix crystals, and then desic- cated to create a flavored rope. The intrinsic steroidal materials in the flavored rope (n 5 3) were extracted and assayed as described above.

Saliva collection

Cotton rope was cut into 80-cm pieces, which were divided into several threads and dipped in a solution of powdered drink-mix crystals. After desic- cation, three cotton threads were tied together, attached to a metal carabineer, and suspended from the grid of a cage or bedroom. The chimpanzees were allowed to chew the threads for 5 min (see Supporting Information for photograph). The saliva absorbed into the rope was collected by centrifugation, and the samples (rope samples) were subjected to assay only when a sufficient volume of saliva (41.0 mL) had been collected. Any additional samples obtained were collected separately and assayed. The mean values were used for statistical analysis.

Saliva samples (direct samples) were collected at the same time and from the same animals as the rope samples. Thus, we obtained matched pairs of direct and rope samples (n 5 21; the number of samples/ individual varied from 1 to 5). All samples were frozen at –501C within 1 hr after collection.

Simultaneous C-T LC-MS/MS assay

The simultaneous C-T LC-MS/MS assay is a convenient method for simultaneously quantifying C and T without derivatisation and was developed by one of the authors (SH). In this study, it was used to investigate the effect of drink-mix crystals on steroid concentrations and to compare the rope saliva and direct saliva collection methods. The saliva samples were thawed and measured (0.6–1.0 mL). Internal standards (2 ng of C-d4 and 1 ng of T-d3) and 5 mL of ethyl acetate were added, and the mixture was shaken for 10 min. After centrifugation, the organic layer was separated by freezing, and the extracts were evaporated to dryness in a centrifugal vacuum concentrator at 371C. The extract was reconstituted with 100 mL of 0.05% formic acid:methanol (6:4, v/v), and a 20-ı`L aliquot was subjected to LC-MS/MS.

All the assays for this experiment were conducted with an API 4000 (Applied BioSystems Japan, Tokyo, Japan) because it maintains high detection sensitivity owing to less interference from ion suppression, and the steroids do not have to be derivatized [Yasuda et al., 2008]. The API 4000 was coupled to an Agilent HP1100 HPLC system equipped with an EPI interface and an Xterra MSC18 column (Nihon Waters). The column temperature was 401C, and the flow rate was 0.4 mL/min. Gradient elution was performed with 20 mM ammonium formate (solvent A) and acetonitrile: methanol (10:3, v/v) (solvent B). The gradient program began at 35% solvent B for 0.8 min, ramped to 100% solvent B at 1.5 min, returned to 35% solvent B at 3.01 min, and held for 7 min. The MS-MS spectra were recorded in the positive mode with the EPI interface. The transitions were monitored at m/z 363.3 to 121.2 for C, 367.3 to 121.2 for C-d4, 289.2 to 97.3 for T, and 292.2 to 97.3 for T-d3. The limits of quantification were 5 and 10 pg for T and C, respectively. The quality of the measurement was ensured by checking the quality control samples continuously. For samples containing 5, 500, and 1,000 pg of T, the accuracy was 95.5, 99.4, and 98.2%, and the precision (intra-assay coefficient of variance) was 7.3, 1.2, and 1.9%, respectively. The inter- assay coefficient of variance for the low-concentration control (5 pg) was 0.3%. For samples containing 10, 20, 1,000, and 10,000 pg of C, the accuracy was 83.7, 84.5, 96.3, and 98.3% and the precision (intra-assay coeffi- cient of variance) was 15.8, 16.0, 2.7, and 2.3%, respectively (n 5 5 for each). The inter-assay coefficient of variance for the low-concentration control (20 pg) was 5.8% (n 5 15; data collected for three days, during which five samples were collected per day). For both C and T, comparison between the predicted and observed values showed parallel responses (_{r 5 0.99).}

Experiment 3 Animals

In Experiment 3, we measured the T concentra- tion in saliva by LC-MS/MS and compared the values

to those obtained by RIA. Experiment 3 was conducted in June 2007. Four individuals were used in this experiment (Table I).

Sample collection

We obtained 20 matched pairs of direct and rope samples from four individuals (2–7 pairs from each individual) while the chimpanzees were in individual cages. All samples were frozen at 501C within 1 hr after collection and until assayed.

Simultaneous C– T LC-MS/MS assay

We followed the same procedures as those described for Experiment 2.

Testosterone RIA

The T RIA was performed with a commercially available kit (DPC total testosterone kit; Mitsubishi Kagaku Iatron, Tokyo, Japan). The saliva samples were serially diluted parallel to the T standard curve (r 5 0.99) over a range of 0.20–16 ng/mL. The mean T recovery was 98.872.7%. The intra- and inter-assay coefficients of variation were 15.0% respectively, for the low-concentration standard. The assay sensitiv- ity was 0.02 ng/mL. The antibody cross-reactivities were 2.8–3.4% for 5a-dihydroteststerone and o1.0% for dehydroepiandrosterone, androstendione, an- drosterone, progesterone, and estradiol.

Statistical analysis

The concentrations of steroids in the urine vary with the volume of fluid intake and output and therefore were corrected for creatinine concentration using Jaffe’s method [Taussky, 1954]. All urinary steroid concentrations are expressed relative to creatinine (ng/mg Cr). All fecal concentrations are expressed relative to fecal wet weight (ng/g). The total amount of excreted steroids in the urine and feces was calculated. All steroid concentrations are presented as the mean7SD. In Experiment 2,

concentrations are expressed as pg/cm of rope and the mean7SD.

In the data analyses, we first checked whether the data met the normal distribution. If not, we conducted logarithmic transformation to fit the data to the normal distribution and used parametric tests for analysis. Nonparametric tests were used for data that did not approach normality. The a level was set at 0.05, and all tests were two-tailed.

In all analyses, we used separate general linear mixed models (GLMMs) [Crawley, 2007; Hruschka et al., 2005; Muehlenbein et al., 2004] to examine the relationships between two values or to compare two data sets. Mixed models allow both fixed and random terms to be fitted to a model. Random terms take into consideration repeated sampling; we included the identity of a chimpanzee and the sample date as random terms. When the estimated variance of a given random term was 0, we excluded that random term from the model.

RESULTS

Experiment 1: Steroid Concentrations in Saliva and Plasma

The mean C and T concentrations in the saliva were approximately 1.5 and 0.8% of those in the plasma (Table II). The BAT was approximately 28.4% of the total T concentration in the plasma. The mean total daily fecal C and T concentrations were 3.0 and 0.5%, respectively, of those in urine.

The steroid concentrations in saliva reflected those in the plasma. There was a positive relation- ship between the salivary C concentration and plasma C concentration (Fig. 1a; GLMM, b 5 0.93, t 5 2.34, P 5 0.034). Similarly, the salivary T concen- tration was positively related to the plasma concen- trations of T (Fig. 1b; GLMM, b 5 0.87, t 5 7.51, P40.001) and BAT (Fig. 1c; GLMM, b 5 0.77, t 5 4.03, P 5 0.003).

TABLE II. Mean concentrations of C and T in plasma, urine, saliva, and feces of male chimpanzees Steroid concentration, mean7SD (n)

Steroid Medium 0900 sample 1300 sample 1700 sample Mean

Cortisol Plasma (ng/mL) 146.0736.3 (4) 107.7743.7 (4) N.A. 126.8742.5 (8) Saliva (ng/mL) 2.3170.69 (30) 1.5470.64 (31) 0.4470.22 (29) 1.4470.94 (86) Urine (ng/mg Cr) 88.04730.51 (32) 108.12751.91 (32) 71.30744.59 (30) 124.6479.82^(31) Feces (ng/g) 7.0372.97 (32) 5.9572.40 (32) 6.4473.43 (29) 3.7771.24^(32) Testosterone Plasma (ng/mL) 3.1970.70 (4) 2.6571.45 (4) N. A. 2.9271.36 (8)

Saliva (pg/mL) 30.5978.92 (29) 24.1977.32 (31) 17.5675.58 (30) 24.0479.02 (86) Urine (ng/mg Cr) 74.48729.35 (32) 63.43732.72 (32) 67.01729.08 (32) 106.5578.83^(32) Feces (ng/g) 0.7070.61 (26) 0.7570.54 (28) 0.7870.59 (26) 0.5270.54^(23)

BAT Plasma (ng/mL) 0.9370.09 (4) 0.7470.23 (4) N. A. 0.8370.23 (8)

Mean urine and fecal steroids excreted in 1 day, mg/day. N.A., data not available.

Experiment 1: diurnal variations

The expected pattern of diurnal fluctuation was found in salivary steroids. The concentrations of salivary C and T varied among the sampling times (Fig. 2 and Table II; GLMM, C: w²5 152.93, Po0.001; T: w²5 56.69, Po0.001). The mean morn- ing concentrations were significantly higher than those measured for the afternoon and evening, and the afternoon concentrations were higher than those for the evening (Po0.01 for all comparisons).

The urinary steroid concentrations also varied among the sampling times (GLMM, C: w²5 27.37, Po0.001; T: w²5 7.89, P 5 0.019), but the pattern of urinary C was different from the expected diurnal fluctuation. Although the mean morning and after- noon C concentrations were significantly higher than those measured for the evening (Po0.01), the C concentrations tended to be higher in the afternoon than in the morning (P 5 0.063). The morning urinary T concentrations were higher than those in the afternoon and evening urine samples (_{P 5 0.02} for both comparisons).

The fecal steroid concentration did not vary according to the sampling time (GLMM, C: w²5 2.98, P 5 0.23; T: w²5 2.91, P 5 0.23).

Experiment 2: Effects of Wash Treatment and Drink-Mix Crystals

The wash treatment removed 96% of the T-like steroids from the raw cotton rope (untreated

cotton rope: 41.49710.81 pg/cm,n 5 3; wash-treated rope: 1.7570.69 pg/cm, n 5 15; GLMM, w²5 26.03, Po0.0001). The addition of the drink-mix crystals did not alter the effect of the wash treatment on T-like intrinsic steroidal material (flavored rope: 1.0670.23 pg/cm, n 5 3; paired t-test, t 5 0.55, P 5 0.64).

According to the simultaneous C-T LC/MS-MS, the mean C-like intrinsic steroidal material concen- tration retained in the flavored rope was 1.7171.89 pg/mL, which was negligible compared with the mean salivary C concentration of around 1 ng/mL.

Experiment 2: differences and relationships between the direct and rope samples

The salivary steroid concentrations in the rope samples reflected those in the direct samples to a considerable extent. There were significant differences in the absolute values of C (direct: 1.0970.88; rope: 0.9570.72 ng/mL;n 5 20 for each; w²5 4.98, P 5 0.026) and T (direct: 17.24712.26; rope: 13.0373.17 pg/mL; n 5 20 for each; w²5 6.11, P 5 0.013) between the direct and rope samples (Fig. 3). Nevertheless, there was a significant relationship between the C concentrations in the direct and rope samples (_{b 5 0.90, t 5 12.26,} Po0.001) and between the T concentrations in the direct and rope samples (b 5 0.26, t 5 4.77, Po0.001).

Fig. 1. Relationship between steroid concentrations in saliva and plasma samples. (a) C, (b) T, (c) BAT. White circles indicate the data for Mizuo and the black circles indicate those for Yukio.

Experiment 3: Comparison of LC-MS/MS and RIA

The salivary T concentrations measured by LC- MS/MS and RIA were similar, and there was a significant relationship between the T concentrations in the direct samples and rope samples as determined by LC-MS/MS (b 5 1.11, w²5 13.93, Po0.001) and by RIA (b 5 1.25, w²5 4.76, P 5 0.029). The salivary T concentrations were similar as measured by LC-MS/MS and RIA in the direct samples (Fig. 4a; b 5 1.01, w²5 30.31, Po0.001) and in the rope samples (Fig. 4b;b 5 0.53, w²5 10.63, P 5 0.001), although the salivary T concentration in the direct samples tended to be lower as determined by LC-MS/MS compared with RIA (w²5 3.45, P 5 0.063; note that there was a significant difference in Experiment 2). RIA did not detect statistical differences in T concentration between the direct and rope samples (w²_{5 1.47,} P 5 0.226).

DISCUSSION

For the first time, we report the use of saliva samples, which can be obtained from chimpanzees by a noninvasive method, for the measurement of steroid concentrations. Salivary steroid concentra- tions are highly sensitive to acute or episodic changes [Laudenslager et al., 2006]. Therefore, for deter- mining steroid concentrations, saliva has distinct advantages over feces or urine, which have been used in previous studies of captive chimpanzees. We determined the steroid concentrations using LC-MS/ MS, which can measure small quantities of salivary

steroids directly and has better sensitivity than EIA or RIA. LC-MS/MS also avoids any possibility of cross- reactivity, which is a major disadvantage of EIA and RIA [e.g., Wang et al., 2004]. Furthermore, our LC-MS/MS method reported here allowed us to measure the T and C concentrations simultaneously (Experiment 3), which cannot be done with an immunoassay (the simultaneous measurement of T and C by LC-MS/MS cost about 20,000 JPY per sample in this study).

The results from Experiment 1 validated the use of saliva as a bio-marker for chimpanzee endo- crine physiology. First, there were consistent positive relationships between salivary and circulating C and T concentrations. In particular, the positive relation- ship between salivary T and circulating BAT indi- cated that salivary T reflects the biologically active portion of circulating T. In humans, salivary T correlates with BAT and reflects the free T level [Morley et al., 2006]. The mean C concentration in saliva compared with that in plasma (1.5%) was high relative to the values for T (0.8%), which agrees with the difference between water-soluble C and poorly water-soluble T. Moreover, the concentrations of salivary C in chimpanzees (1.44 ng/mL; Table II) were comparable with previously reported values in bonobos (Pan paniscus; adult and juvenile males and females: 0.53–1.67 ng/mL) [Hohmann et al., 2009], in orangutans (Pongo pygmaeus; adult female: 1.11 ng/mL) [Elder & Menzel, 2001], and in humans (male: 3.52; female: 1.75 ng/mL) [Kirschbaum et al., 1992]. Second, the expected pattern of diurnal fluctuation was evident in saliva. This pattern was

Fig. 2. An example of diurnal variation in salivary, urinary, and fecal steroids in one male (Mizuo) for eight days (horizontal axis). The three squares connected by a line correspond to the three samples collected in a day. From the left of each graph, the three squares indicate the morning (0900: black squares), the afternoon (1300: grey squares), and the evening (1700: white squares) samples, respectively.

similar to the well-known pattern of diurnal fluctua- tion in humans [Rose et al., 1972]. We also found diurnal variation in urinary C, but the pattern differed from the expected pattern in that the highest concentration was measured in the afternoon sam- ples, rather than in the morning or evening samples. Given the time delay in the excretion of plasma C into urine [Bahr et al., 2000], it is possible that the low concentration of urinary C in the morning sample reflects the low plasma C concentration of the previous night. Combined, these data indicate that salivary steroids are highly sensitive to temporal changes in blood C and T concentrations.

In Experiment 2, we developed an effective method for sampling saliva with a cotton rope and validated its use in the steroid assay. The T-like steroids present in cotton rope had previously prevented its use for collecting saliva to be analyzed for T [Dabbs, 1991; Granger et al., 2004; Shirtcliff et al., 2001], but our procedure for washing the rope removed 96% of the T-like steroids. The concentra- tion of C-like steroids measured in flavored rope was negligible, confirming that the use of cotton does not

affect the C assay. The effectiveness of this washing method was confirmed by the significant relation- ships for C and T between the direct and rope samples, despite the fact that the use of the rope method reduced the small proportion of salivary steroids. These results suggest that the salivary C and T concentrations measured in samples obtained by the rope method are reliable for estimating chimpanzee steroid physiology.

The salivary T concentration determined by LC-MS/MS was similar to that measured by RIA; however, RIA did not detect the difference in sali- vary T between the rope and direct samples. On the contrary, LC-MS/MS detected the statistical differences in Experiment 2 and a tendency in Experiment 3. Thus, LC-MS/MS may be superior to RIA for measuring the salivary T concentration in saliva samples obtained by the rope method.

In summary, the results of our experiments using LC-MS/MS show the utility of saliva for measuring steroids in chimpanzees. Recently, the

Fig. 3. Relationships of (a) C and (b) T between samples collected directly from the mouth (direct sample) and by the rope method (rope sample). The dashed line indicates the case in which the steroid concentration from the direct sample and the rope sample coincide.

Fig. 4. Relationship of salivary T concentrations determined by LC-MS/MS and RIA for (a) saliva samples taken directly from the mouth (direct sample) and (b) saliva samples taken by the rope method (rope sample). The dashed line indicates the case in which the steroid concentration from the direct sample and the rope sample coincide.

use of saliva as a medium to investigate internal endocrine physiology has become common in both human and nonhuman animal research in a wide range of contexts (e.g., clinical diagnosis in humans and reproductive physiology in endangered animals) [Granger et al., 2004; Gro¨schl, 2008; Lewis, 2006]. We believe that the biological validity of measuring salivary steroids and the sample collection techni- ques developed here can be applied to other primates and nonprimates living in various conditions (captive, free-ranging, wild) to measure acute and episodic changes in steroids. Although previous studies in primates using salivary hormones took advantage of relatively easy sample collection methods, without incurring stress to the study subjects, another important advantage of this method is that researchers are able to investigate sequential changes in the hormone dynamics by comparing the concentration before and after an event or behavior [e.g., Cross et al., 2004; Fuchs et al., 1997; Hohmann et al., 2009]. This type of research design, which has been used in our other experiments in chimpanzees, will enable us to go beyond correlation approaches and provides a powerful test for investigating causal relationships between behaviors and hormones.

ACKNOWLEDGMENTS

This research was financially supported by the JSPS 21st Century COE program ‘‘Center for Evolutionary Cognitive Sciences at the University of Tokyo’’ from the Ministry of Education, Culture, Sports, Science and Technology of Japan, Program of Gerontological Research at the University of Tokyo, and Hayama Center for Advanced Studies at the Graduate University for Advanced Studies. This study was approved by the Screening Committee for Chimpanzee Experiment of CSU (Chimpanzee Sanc- tuary Uto) and was conducted under the Japanese laws.

REFERENCES

Anestis SF. 2005. Behavioral style, dominance, rank, and urinary cortisol in young chimpanzees (Pan troglodytes). Behaviour 142:1245–1268.

Bahr NI, Palme R, Mohle U, Hodges JK, Heistermann M. 2000. Comparative aspects of the metabolism and excretion of cortisol in three individual nonhuman primates. Gen Comp Endocrinol 117:427–438.

Behringer V, ClauX W, Hachenburger K, Kuchar A, Mo¨stl E, Selzer D. 2009. Effect of giving birth on the cortisol level in a bonobo groups’ (Pan paniscus) saliva. Primates 50:190–193. Boyce WT, Champoux M, Suomi SJ, Gunnar MR. 1995. Salivary cortisol in nursery-reared rhesus monkeys: reactivity to peer interactions and altered circadian activity. Dev Psychobiol 28:257–267.

Crawley MJ. 2007. The R book. Chichester: John Wiley & Sons, Ltd. 950p.

Cross N, Rogers LJ. 2004. Diurnal cycle in salivary cortisol levels in common marmosets. Dev Psychobiol 45:134–139.

Cross N, Pines MK, Rogers LJ. 2004. Saliva sampling to assess cortisol levels in unrestrained common marmosets and the effect of behavioral stress. Am J Primatol 62:107–114. Dabbs Jr JM. 1991. Salivary testosterone measurements:

collecting, storing, and mailing saliva samples. Physiol Behav 49:815–817.

Elder CM, Menzel CR. 2001. Dissociation of cortisol and behavioral indicators of stress in an orangutan (Pongo pygmaeus) during a computerized task. Primates 42: 345–347.

Emadi-Konjin P, Bain J, Bromberg IL. 2003. Evaluation of an algorithm for calculation of serum bioavailable testosterone (BAT). Clin Biochem 36:591–596.

Fuchs E, Kirschbaum C, Benisch D, Bieser A. 1997. Salivary cortisol: a non-invasive measure of hypothalamo-pituitary- adrenocortical activity in the squirrel monkey, Saimiri sciureus. Lab Anim 31:306–311.

Gozansky WS, Lynn JS, Laudenslager ML, Kohrt WM. 2005. Salivary cortisol determined by enzyme immunoassay is preferable to serum total cortisol for assessment of dynamic hypothalamic-pituitary-adrenal axis activity. Clin Endo- crinol 63:336–341.

Granger DA, Schwartz EB, Booth A, Arentz M. 1999. Salivary testosterone determination in studies of child health and development. Horm Behav 35:18–27.

Granger DA, Shirtcliff EA, Booth A, Kivlighan KT, Schwartz EB. 2004. The trouble with salivary testosterone. Psychoneuroendocrinology 29:1229–1240.

Gro¨schl M. 2008. Current status of salivary hormone analysis. Clin Chem 54:1759–1769.

Hauser B, Deschner T, Boesch C. 2008a. Development of a liquid chromatography–tandem mass spectrometry method for the determination of 23 endogenous steroids in small quantities of primate urine. J Chromatogr B Biomed Appl 862:100–112.

Hauser B, Schulz D, Boesch C, Deschner T. 2008b. Measuring urinary testosterone levels of the great apes—problems with enzymatic hydrolysis using Helix pomatia juice. Gen Comp Endocrinol 158:77–86.

Hohmann G, Mundry R, Deschner T. 2009. The relationship between socio-sexual behavior and salivary cortisol in bonobos: tests of the tension regulation hypothesis. Am J Primatol 71:223–232.

Honma S, Kawabe K, Misawa M, Okuyama M, Kumagai Y, Okazaki Y. 2004. A simple method for the estimation of bio-available testosterone in serum. Clin Endocrinol (Hormone to Rinsho, in Japanese) 52:829–844.

Hruschka DJ, Kohrt BA, Worthman CM. 2005. Estimating between- and within-individual variation in cortisol levels using multilevel models. Psychoneuroendocrinology 30: 698–714.

Kirschbaum C, Hellhammer DH. 1994. Salivary cortisol in psychoneuroendocrine research: recent developments and applications. Psychoneuroendocrinology 19:313–333. Kirschbaum C, Wu¨st S, Hellhammer DH. 1992. Consistent sex

differences in cortisol responses to psychological stress. Psychosom Med 54:648–657.

Kuhar CW, Bettinger T, Laudenslager M. 2005. Salivary cortisol and behaviour in an all-male group of western lowland gorillas (Gorilla g. gorilla). Anim Welf 14:187–193. Lac G. 2001. Saliva assays in clinical and research biology.

Pathol Biol 49:660–667.

Laudenslager ML, Bettinger T, Sackett G. 2006. Saliva as a medium for assessing cortisol and other compounds in nonhuman primates: collection, assays, and examples. In: Sackett G, Ruppenthal G, editors, Nursery rearing of nonhuman primates in the 21st century. New York: Plenum Publishers. p 403–427.

Lewis JG. 2006. Steroid analysis in saliva: an overview. Clin Biochem Rev 27:139–146.

Lutz CK, Tiefenbacher S, Jorgensen MJ, Meyer JS, Novak MA. 2000. Techniques for collecting saliva from awake, unrestrained, adult monkeys for cortisol assay. Am J Primatol 52:93–99.

Mohle U, Heistermann M, Palme R, Hodges JK. 2002. Characterization of urinary and fecal metabolites of testos- terone and their measurement for assessing gonadal endocrine function in male nonhuman primates. Gen Comp Endocrinol 129:135–145.

Morley JE, Patrick P, Perry HM. 2002. Evaluation of assays available to measure free testosterone. Metabolism 51:554–559. Morley JE, Perry HM, Patrick P, Dollbaum CM, Kells JM. 2006. Validation of salivary testosterone as a screening test for male hypogonadism. Aging Male 9:165–169.

Muehlenbein MP. 2006. Intestinal parasite infections and fecal steroid levels in wild chimpanzees. Am J Phys Anthropol 130:546–550.

Muehlenbein MP, Watts DP, Whitten P. 2004. Dominance rank and fecal testosterone levels in adult male chimpan- zees (Pan troglodytes schweinfurthii) at Ngogo, Kibale National Park, Uganda. Am J Primatol 64:71–82.

Muller MN, Lipson SF. 2003. Diurnal patterns of urinary steroid excretion in wild chimpanzees. Am J Primatol 60:161–166.

Muller MN, Wrangham RW. 2004. Dominance, cortisol and stress in wild chimpanzees (Pan troglodytes schweinfurthii). Behav Ecol Sociobiol 55:332–340.

Palme R. 2005. Measuring fecal steroids: guidelines for practical application. Ann N Y Acad Sci 1046:75–80. Pearson BL, Judge PG, Reeder DM. 2008. Effectiveness of

saliva collection and enzyme-immunoassay for the quanti- fication of cortisol in socially housed baboons. Am J Primatol 70:1145–1151.

Rose RM, Kreuz LE, Holaday JW, Sulak KJ, Johnson CE. 1972. Diurnal variation of plasma testosterone and cortisol. J Endocrinol 54:177–178.

Shirtcliff EA, Granger DA, Schwartz E, Curran MJ. 2001. Use of salivary biomarkers in biobehavioral research: cotton-based sample collection methods can interfere with salivary immunoassay results. Psychoneuroendocrinology 26:165–173.

Taussky HH. 1954. A microcolorimetric determination of creatinine in urine by the Jaffe reaction. J Biol Chem 208:220–222.

Tiefenbacher S, Lee B, Meyer JS, Spealman RD. 2003. Noninvasive technique for the repeated sampling of salivary free cortisol in awake, unrestrained squirrel monkeys. Am J Primatol 60:69–75.

Vermeulen A, Verdonck L, Kaufman JM. 1999. A critical evaluation of simple methods for the estimation of free testosterone in serum. J Clin Endocrinol Metab 84: 3666–3672.

von Engelhardt N, Kappeler PM, Heistermann M. 2000. Androgen levels and female social dominance in Lemur catta. Proc R Soc Lond B 267:1533–1539.

Wang C, Catlin DH, Demers LM, Starcevic B, Swerdloff RS. 2004. Measurement of total serum testosterone in adult men: comparison of current laboratory methods versus liquid chromatography-tandem mass spectrometry. J Clin Endocrinol Metab 89:534–543.

Whitten PL, Brockman DK, Stavisky RC. 1998a. Recent advances in noninvasive techniques to monitor hormone–- behavior interactions. Am J Phys Anthropol Suppl 27:1–23. Whitten PL, Stavisky R, Aureli F, Russell E. 1998b. Response of fecal cortisol to stress in captive chimpanzees (Pan troglodytes). Am J Primatol 44:57–69.

Yasuda M, Honma S, Furuya K, Yoshii T, Kamiyama Y, Ide H, Muto S, Horie S. 2008. Diagnostic significance of salivary testosterone measurement revisited; using liquid chromato- graphy/massspectrometry and enzyme-linked immunosor- bent assay. J Mens Health 5:1–8.

Supplementary material for “Validation of salivary cortisol and testosterone

assays in chimpanzees by liquid chromatography-tandem mass spectrometry”

1 2 3

4

Comparison of saliva collection methods

Aim

5 6 7 8

We report the efficiency of three methods for collecting saliva from

chimpanzees.

Methods

9 10 11 12 13 14 15 16 17 18

There was a 1-week training period during which each chimpanzee became

accustomed to the apparatus and the collection procedures. We used 17 individuals in

this experiment; the number of individuals used for each method varied (Supplemental

Table 1). Two methods of collecting saliva had been used in previous studies [Lutz et

al., 2000; Tiefenbacher et al., 2003], and we modified the size and strength of the

materials to fit the chimpanzees and their captive conditions. When >1.0 mL of saliva

was collected, the case was counted as a success. When the individual had no interest

in the apparatus or the volume of saliva was insufficient, the case was counted as a

failure. No apparatus was broken by the chimpanzees.

Direct method: Saliva samples were collected directly from the mouth by

using a polypropylene syringe (Supplemental Fig. 1), after the animals had been

trained to lick a powdered drink-mix crystal (Pocari Sweat®; Otsuka Pharmaceutical

Co., Ltd., Tokyo, Japan) solution from the tip of the syringe. In the next stage, the

chimpanzees were trained not to respond to the direct-saliva collection procedure.

19 20 21 22 23

Screen method: Six sheets of cotton gauze (8 × 8 cm) were put between two

pieces of stainless mesh (8 × 8 cm), which were fixed between two pieces of Plexiglass

(20 × 10 cm) with a hole (diameter, 6 cm) at the centre to allow the chimpanzees

access to the gauze. To add flavour, the gauze had been dipped in an aqueous solution

of powdered drink-mix crystals and then desiccated. The apparatus was attached to the

grid of the cage or bedroom. The chimpanzees were allowed to lick the gauze for 15

min, then the apparatus was detached, and the saliva contained in the gauze was

collected by centrifugation.

24 25 26 27 28 29 30 31

Rope method: We first tested natural cotton and polyvinyl alcohol (PVA) fibre

32

(Kuralon®; Kuraray Co., Ltd., Kurashiki, Japan) as rope materials. PVA fibre is a

synthesized filament with a texture and absorbing capacity similar to those of natural

cotton and is commercially available as a slender rope. However, the LC-MS/MS data

showed that PVA created an extremely high level of background noise that completely

concealed the testosterone peak (data not shown). Thus, we dismissed PVA as a rope

material and used only cotton rope.

1 2 3 4 5 6 7

Data analysis

8 9 10 11 12 13

We compared the proportion of successful individuals by the _χ

test. We used

a General Linear Mixed Model to compare the volume of saliva collected/minute

among the three methods. The identity of the individuals and the sampling date were

set as random terms.

Results

14 15 16 17 18 19 20 21 22 23

The rope method was the most efficient method for collecting saliva samples.

There was no difference in the proportion of successful individuals among the three

methods ( _χ

= 2.72, p = 0.256; Supplemental Table 1); however, the volume of saliva

collected varied among the methods (Supplemental Fig. 2; _χ

= 18.42, p < 0.001). The

rope method was more efficient than the other two methods (p < 0.01), and the direct

method was more efficient than the screen method (p < 0.01).

Supplemental Table 1

1 2 3 4

Results of the training of male chimpanzees to allow the collection of saliva by each

method.

ID Birth year Direct method Screen method Rope method

Yukio 1987 Yes -- --

Ken 1983 Yes -- --

Izou 1978 No Yes Yes

Mizuo 1989 Yes -- --

Kotaro 1993 Yes -- --

Maruku 1980? No Yes Yes

Lucky 1974 No No Yes

Lennon 1970 No Yes Yes

Kazuya 1987 No Yes Yes

Takashi 1988 No Yes Yes

James 1993 No No No*

Kanao 1990 No Yes No*

Mikota 1992 No Yes Yes

Norihei 1987 No No No*

Minato 1992 No No No*

Kenji 1984 No Yes Yes

Gou 1977? No No Yes

# trained 17 13 13

# of successes 4 8 9

% (successes/trained) 23.5% 61.5% 69.2%

Yes, successful collection

5 6 7 8

No, failed collection

* Saliva samples could not be collected by the rope method after 1 week of training but

could be collected after additional training.

1 2 3 4 5

Supplemental Fig. 1. Methods for collecting saliva samples. (a) A saliva sample being

collected directly from the mouth of a trained chimpanzee. (b) A chimpanzee chewing

the cotton rope used for collecting a saliva sample.

(a)

6 7 8

_(b)

9

1 2 3 4

Supplemental Fig. 2. Comparison of the saliva volume collected per minute among

the three different methods. Data are means ± SE of the samples.

5 6 7 8