Introduction

Release of volatile organic compounds (VOCs) by living organisms is non-random temporally (e.g., Dufa et al. 2004; Pause et al. 2004; Steingass et al. 2014; Turlings et al. 1990), and hence is eligible to transmit time-triggered chemosensory cues. In plants, an event-related release of VOCs is, among others, afforded by conjugates, which set free VOCs when adequate enzymatic activity is present, e.g., upon cell damage, bringing together substrates and enzymes from different cellular compartments. A well-known example is in the genus Allium with its S-alk(en)yl-L-cysteine sulfoxides and γ-glutamyl-cysteine-S-conjugates (Block 1992; Starkenmann et al. 2011), although occurrence of cysteine-S-conjugates has been described more generally [e.g., Vitis (Tominaga et al. 1998); Capsicum (Starkenmann and Niclass 2011); Passiflora (Fedrizzi et al. 2012)].

Mammals also use conjugates as one of manifold ways (c.f., Flower 1996) to store or transport small hydrophobic molecules throughout the organism, and these conjugates may be cleaved to release volatile cues (e.g., Brewington et al. 1973; Fabregat et al. 2013; Lopez and Lindsay 1993; Starkenmann et al. 2014; Wagenstaller and Buettner 2013). For instance, in human sweat, conjugates provide an odor release that gains variability, not only related to an individual’s secretory characteristics but also to an individual’s axillary microbiota, and these factors can be interdependent (Harker et al. 2014; Kuhn and Natsch 2009; Leyden et al. 1981; Martin et al. 2010; Troccaz et al. 2004). Indeed, human sweat odor has been shown to emerge from originally odorless apocrine sweat upon enzymatic cleavage of cysteinylglycine-S-conjugates, glutamine-N-α-conjugates, and glucuronides (Natsch et al. 2003, 2006; Shelley et al. 1953; Starkenmann et al. 2005, 2013). To date, however, it remains disputed whether these conjugates are formed during (from glutathionyl conjugates transported by ABCC11; Baumann et al. 2014) or after the apocrine gland secretion process (via opening of odorant-carrying proteins and enzymatic action; Preti and Leyden 2010; Zeng et al. 1996). Odorant-carrying proteins have also been suggested to occur in sweat from other regions (e.g., areolar sweat; Spielman et al. 1995). In contrast, part of the afore-mentioned conjugates, amongst others N-α-3-hydroxy-3-methylhexanoyl-L-glutamine and N-α-3-methylhex-2-enoyl-L-glutamine, are thought to be specific to axillary sweat since they have not been detected in sweat originating from other body regions, nor in plasma or urine (Natsch et al. 2006). However, this notion is challenged by the detection of the above-mentioned glutamine-N-α -conjugates in human amniotic fluid, colostrum, and milk (Hartmann et al. 2012). Further, cysteinylglycine-S-conjugates of 3-methyl-3-sulfanylhexan-1-ol and 3-sulfanylhexan-1-ol occur in human colostrum and milk (Hartmann et al. 2012). Thus, perinatal fluids appear to contain conjugated forms of odorants known to contribute to sweat odor. Yet, while efforts have been made to understand conditions, temporal dynamics, and communicative value of odor formed in the axilla (reviewed by James et al. 2013), it is unknown whether conjugates occurring in amniotic fluid and milk release odorants into a newborn’s environment, and whether these odorants are involved in interpersonal communication.

In adults, cysteine-S-conjugates are cleaved by oral microbiota, leading to a delayed but prolonged olfactory percept compared with consumption of the free odorant itself (Starkenmann et al. 2008). An analogous, temporally defined, odor release during mother-infant interaction would offer opportunities for odor perception in the “perinate” (i.e., the transitional organism 2–3 d before, during, and 2–3 d after birth), especially when occurring concurrently with states of high neonatal learning susceptibility. Highly efficient odor acquisition is thought to happen during and/or shortly after birth due to high neonatal brain norepinephrine levels caused by the effect of uterine contractions on an infant’s skull (Alberts and Ronca 2012; Varendi et al. 2002). In fact, 30 min exposure to cherry or mango odor leads neonates to prefer these odors when exposure occurs during the first hour after delivery, but not when it occurs 12 h after delivery (Romantshik et al. 2007). Further, breastfeeding offers highly reinforcing conditions for odor learning, generating odor preferences persisting as long as weeks (Schleidt and Genzel 1990) or months (Delaunay-El Allam et al. 2010). During both delivery and breastfeeding, enzymes able to cleave the aforementioned amino acid conjugates might be provided by a mother’s or a newborn’s microbiota, which occur within the amnion (e.g., DiGiulio et al. 2008), vagina (e.g., Tannock et al. 1990), milk and breast (e.g., Urbaniak et al. 2014), and in the mouth and skin (e.g., Rotimi and Duerden 1981).

From the above evidence, we hypothesized human perinates familiarize themselves with human sweat odorants through postnatal, and possibly perinatal, exposure. Whereas maternal sweat odor has been shown to be acquired by 12–18-d-old breast-fed newborns (Cernoch and Porter 1985), so far no studies have investigated neonatal responses to pure sweat odorants. Neonatal perception of these compounds is not only highly interesting in view of a possible chemocommunicative function of these odorants in perinatal fluids, such as colostrum and milk (e.g., Macfarlane 1975; Marlier and Schaal 2005; Russell 1976; Schaal 2005, 2015) but, also, in view of developmental aspects of human sweat odor perception: even if several authors have investigated pleasantness/unpleasantness of axillary and body odors [(e.g., adults (Doty et al. 1978; Troccaz et al. 2009), children (Ferdenzi et al. 2010)], the molecular bases of human sweat odor hedonicity are not yet completely understood.

In regard to this line of thought, we examined neonatal responses to (R)/(S)-3-methyl-3-sulfanylhexan-1-ol (MSH), (R)/(S)-3-sulfanylhexan-1-ol (SH), (E)/(Z)-3-methylhex-2-enoic acid (3M2H), and (R)/(S)-3-hydroxy-3-methylhexanoic acid (HMHA). Neonatal detection of and hedonic response to these odorants were assessed by administering aqueous solutions to 3-d-old human infants. Further, detection thresholds of adults were obtained by using the same presentation technique but with a different response measure, so as to compare adult and neonatal sensitivity to these odorants. These odorants were chosen because they: i) are impactful sweat odor contributors, ii) have, so far, not been detected in animal milk, and iii) might result from enzymatic cleavage of conjugates detected in amniotic fluid and milk. Since we were especially interested in a possible chemocommunicative function of these substances during breastfeeding, we used very low concentrations that were supposed to match the natural concentrations reported for amniotic fluid and milk.

Several hypotheses can be advanced regarding the direction of an infant’s responsiveness to these separate sweat odorants after presumed prenatal exposure through amniotic fluid, and/or postnatal exposure through colostrum, milk, or maternal sweat. If these compounds elicit positive reactions (e.g., oro-facial responses indicating attraction or appetence), they may be inferred to be somehow reminiscent of the original substrate’s smell, an individual compound’s odor being potentially representative of the odor of the whole mixture. In contrast, neutral or even negative responses may indicate that newborns do not attend to, or even avoid, one or several of these isolated compounds; in this case, the odor percepts conveyed by isolated compounds may differ from the percept of the original substrates from which they stem.

Methods and Materials

Participants

The study was conducted at the maternity department of the Dijon University Hospital. It was approved by the local Committee for the protection of persons submitted to experimentation and complied with the Declaration of Helsinki for medical research involving human subjects. The parents were informed about the aims and methods of the research. They all gave written consent to let their infants participate, and were physically present during the experiment. All infants were in optimal health at birth (Apgar score > 8 at 1 min and =10 at 5 and 10 min) and at the time of olfactory testing.

Experiments

Two experiments were conducted to assess neonatal responsiveness to thiols (Exp. 1) or to acids (Exp. 2). In Experiment 1, 52 newborns participated. As 9 newborns woke up before the test was finished, they were excluded from further analyses, and the final sample consisted of 43 infants (40 Caucasian, 2 African, 1 Asian). In Experiment 2, 18 newborns participated, although complete testing could not be achieved for 2 newborns. The final sample consisted of 16 infants (15 Caucasian, 1 African). Characteristics of the participants having undergone the whole test session are described in Table 1.

Table 1 Descriptive Characteristics Of The Newborn Samples Assessed In Experiments 1 And 2
Full size table

Stimuli

The target odorants, 3-sulfanylhexan-1-ol (SH; 98%; Acros Organics, Geel, Belgium), 3-methyl-3-sulfanylhexan-1-ol (MSH), 3-methyl-2-hexenoic acid (3M2H; 90% E), and 3-hydroxy-3-methylhexanoic acid (HMHA; all from aromaLab, Freising, Germany), were checked for odorous contaminants by gas chromatography-olfactometry (Trace Ultra GC, Thermo Finnigan, Dreieich, Germany; capillary column DB-FFAP, 30 m × 0.32 mm, film thickness 0.25 μm, J&W Scientific, Agilent Technologies, Santa Clara, US). Solutions in dichloromethane were injected on-column at 40 °C. After 2 min, the oven temperature was raised by 12 °C/min to 230 °C and held for 5 min. The flow rate of the helium carrier gas was 2.2 ml/min. One to two contaminants were detected in the SH, MSH, and HMHA. Since no contaminant was perceived with higher intensity than the target odorant, we proceeded without further purification.

Based on quantitative data reported in Hartmann et al. (2012; thiol conjugates: < 100 ng/kg human milk, acid conjugates: < 400 µg/kg human milk), maximum hypothetical concentrations of the free thiols/acids in human milk (assuming 100% cleavage of the respective precursors) were calculated for testing in Experiments 1 and 2.

Experiment 1

The SH was diluted in distilled water to 1 and 10 ng/l, while MSH was taken from a stock solution (990 μg/ml in propylene glycol; 99.5%, Sigma Aldrich, Steinheim, Germany) and diluted in distilled water to 1, 10, and 50 ng/l. Odorless references were distilled water and distilled water containing propylene glycol (“control”; 52 μg/l). To compare target odorants with stimuli of hedonic values known from previous studies (Bingham et al. 2003; Marlier and Schaal 2005; Soussignan et al. 1997), the stimulus set included vanillin (0.01% in distilled water; > 98%, Sigma-Aldrich, Steinheim, Germany) and familiar milk [for a breast-fed infant, its mother’s milk expressed between 1 and 5 min before the test; for a bottle-fed infant, the formula milk consumed since birth (brands: Gallia, Nidal, Modilac, Guigoz)].

Experiment 2

The acids were diluted in distilled water to 30 ng/l and 80 μg/l for 3M2H, and 0.5 ng/l, 50 ng/l, and 200 μg/l for HMHA, corresponding to the maximal and minimal hypothetical concentrations of the acids in colostrum and milk, assuming complete cleavage of the conjugates (see Hartmann et al. 2012). Since the target odorants occur in human sweat, familiar sweat was included as a reference stimulus in addition to distilled water, vanillin, and familiar milk (see above). The sweat stimulus was obtained a few minutes before being tested by turning a glass rod 2–6 times in the maternal axilla and adding 10 μl of distilled water. The mothers were asked previously not to use odorous soap, shower gel, or deodorants on the day of the test.

Procedure and Behavioral State of Newborns

Testing took place in a quiet, dedicated “Baby-lab” (temperature: 23–27 °C) and was conducted by two experimenters. A newborn was seated in a semi-reclining chair surrounded by draperies equilibrating light on each side. Experimenter 1 prepared the stimuli in random order by pipetting 10 μl of a solution on the tip of a 20-cm long glass rod. Experimenter 2, blind to the nature of the stimuli, stood behind the infant and administered the stimuli approximately 1 cm under the nostrils. Following online psychophysiological responses and the infant’s behavior through a TV monitor, Experimenter 1 silently (visually) paced the onset/offset of stimulus administration. Stimulus onset followed a period of at least 10 s, during which an infant had no visible facial movements. Stimulus offset was 10 s after the onset, and inter-stimulus intervals were at least 50 s. The test was considered complete only in the case of all 9 stimuli being applied. Newborns display higher facial reactivity to odors during irregular than during regular sleep (Soussignan et al. 1997). Therefore, infants were required to be in irregular sleep at the beginning of the test, as determined by respiratory and behavioral cues according to Prechtl’s (1974) classification of neonatal behavioral states. Some infants, however, changed to regular sleep during the test (in Exp. 1, 22 newborns, 11 male and 11 female, while in Exp. 2, 4 newborns). Testing was continued in these cases. However, before conducting statistical analyses on these data, we verified that the proportion of newborns having changed to regular sleep was similar for each stimulus (Exp. 1, 5 newborns with water, 6 newborns with MSH at 10 ng/l, SH at 1 ng/l, vanillin, and the control, and 7 newborns with familiar milk, SH at 10 ng/l, MSH at 1 ng/l, and MSH at 50 ng/l; Q 8 = 0.9; P = 0.999; in Exp. 2, 0 newborns with vanillin and HMHA at 200 μg/l, 1 newborn with HMHA at 50 ng/l, 3M2H at 80 μg/l, maternal sweat, and water, and 2 newborns with HMHA at 0.5 ng/l, 3M2H at 30 ng/l, and familiar milk; Q 8 = 6.5; P = 0.59). For Experiment 1, this was verified for both females and males (Q 8 = 4.2, P = 0.84 and Q 8 = 8.0, P = 0.43, respectively), as well as for breast-fed and bottle-fed infants (Q 8 = 6.2, P = 0.62 and Q 8 = 3.7, P = 0.88, respectively).

Autonomic Responses

Breathing was recorded using an 8-channel MacLab data-recording system (ADInstruments Pty Ltd., Castle Hill, Australia) by means of a pneumobelt (Model 1132, Pneumotrace, UFI Instruments, Morro Bay, CA, USA) secured around an infant’s abdomen. The respiratory rate (in beats per minute, bpm) was measured online using the Chart software (version 3.5.2). The dependent variable was computed offline by calculating the mean of data during the 5-s pre-stimulus block (baseline condition), and the subsequent 5-s blocks of the stimulus period (2 × 5 s) and the post-stimulus period (2 × 5 s).

Oro-Facial Responses

The oro-facial reactivity of newborns was recorded by a silent digital video camera (DCR-TRV22E, Sony Corporation, Tokyo, Japan). Oro-facial responses were coded using Observer software (Version 8, Noldus, Wageningen, Netherlands), and immediately classified into positive and negative responses, as follows. Rooting, munching, tongue or lip protrusions, licking, and sucking were considered positive mouthing actions (as defined in Doucet et al. 2007). Facial responses considered as conveying a negative emotion were conservatively coded into three different exclusive clusters of items (items of a cluster that occurred concurrently with items of another cluster were not coded). The first cluster was composed of the following actions: lowered, oblique, or raised and drawn together brow movements [action units (AUs) 1, 4, 1 + 4, 1 + 2 + 4 of the infant version of Ekman and Friesen’s Facial Action Coding System (Oster 2007)]. Nose wrinkling and upper lip rising, associated or not with cheek raising and lids tightening (AUs 9, 10, 9 + 10, 6/7 + 9 + 10), formed the second cluster. The third cluster was composed of 8 AUs: nasolabial furrow deepening (AU 11), lip corner tightening (AU 14), lip corner depressing (AU 15), lower lip depressing (AU 16), chin raising (AU 17), lateral lip stretching (AU 20), and lips tightening and pressing (AUs 23, 24).

The selected items were viewed on a TV screen in slow motion and frame-by-frame to determine onset and offset times with a precision of ±1 video frame (±0.04 s). For each stimulus, the relative duration of positive/negative responses was computed by dividing the value during the 20-s block, comprising the stimulus and post-stimulus period, by the exact duration of the coded video. Inter-observer reliability was assessed between the main coder (not blind to the olfactory stimuli) and a second coder who independently scored randomly selected video clips (Exp. 1, 27 videos; Exp. 2, 42 videos). The second coder was blind to the hypotheses/aims of the study and to the identity of the olfactory stimuli. Spearman correlation coefficients were computed between the two coders for the duration of positive and negative oro-facial actions (Exp. 1, 0.85, and 0.67, respectively; Exp. 2, 0.85, and 0.69, respectively).

Olfactory Testing with Adults

Twenty healthy subjects (age 19–31 years; 10 females) without history of olfactory dysfunction were recruited for each target odorant. The composition of the panel was not identical for each odorant, due to restricted availability of some participants who could participate only in the threshold determination of one of the odorants. A triangle test was carried out with aqueous solutions of the target odorants presented in ascending order of 3-fold dilution steps on 20 cm-long glass rods (SH at 8 ng/l to 160 μg/l, MSH at 52 ng/l to 113 μg/l, 3M2H at 55 μg/l to 360 mg/l, HMHA at 29 ng/l to 5.19 mg/l) against two control stimuli (distilled water). Participants had to choose one of the stimuli as the deviating one. The choice of the odorous glass rod was noted “1” and the choice of a water stimulus as “0”.

Pleasantness then was rated for defined supraliminal concentrations of the odorants (SH at 500 μg/l, MSH at 383 μg/l, 3M2H at 40 mg/l, HMHA at 1.7 mg/l) on a nine point Likert scale (ranging from 1 = extremely unpleasant to 9 = extremely pleasant). These concentrations were chosen as they were readily perceivable by participants. The solutions (10 ml) were presented in brown glass jars (volume = 30 ml, ø ext. = 34 mm).

Statistical Analyses - Respiratory Data

A logarithmic transformation was performed on raw data since they were not normally distributed (Shapiro-Wilk test). Neither age, nor birth weight, nor time elapsed between the last feeding and the test, were consistently correlated with the dependent variables (Pearson test). Analyses of variance (ANOVA) were calculated with the mode of feeding (breast- vs. bottle-fed) and an infant’s sex (male vs. female) as between-subject factors, and olfactory stimulus and test period [5-s interval(s) before, during, and after stimulus administration] as within-subject factors. Fisher’s least significant difference (LSD) test was used for post-hoc multiple comparisons between means.

Statistical Analyses - Oro-Facial Activity Data

Proportions of oro-facially reactive newborns were compared by Cochran’s Q tests on the whole group and, for Exp. 1, on the feeding mode (breast- vs. bottle-fed) and an infant’s sex (male vs. female). McNemar’s tests were calculated post-hoc to compare responsiveness between stimuli.

Durations of facial activity were not normally distributed (Shapiro-Wilk test). Since no transformation resulted in normally distributed data, further analyses were conducted on the raw data. Friedman’s test was used as an omnibus test on the whole group and, for Exp. 1, on the feeding mode and sex (breast- vs. bottle-fed, and male vs. female, respectively). Post-hoc, durations of facial activity in response to different stimuli were compared using Wilcoxon tests.

Results

Experiment 1: Neonatal Reactivity to the Odors of the Thiols (MSH and SH)

Respiratory Rate

A main effect was found for the stimulus period (Fig. 1; F 4,156 = 4.5; P < 0.01). The mean respiratory rate during the 5-s interval prior to stimulus administration was higher compared to the respiratory rate during the fourth 5-s interval after stimulus onset (LSD; P < 0.01). The mean respiratory rate during the 5-s interval following stimulus onset was higher compared with the third and fourth 5-s interval (LSD; P < 0.05 and P < 0.001, respectively). In sum, neonates did not appear to change their respiratory rate in a stimulus-dependent way. Averaged over all stimuli, the respiratory rate decreased after stimulus offset compared with the baseline level and the 5-s interval following stimulus onset.

Fig. 1
figure 1

Mean respiratory rate (± s.e.m.) of neonates over five consecutive 5-s periods in Experiments 1 and 2 (N = 43 and 16, respectively; see Results for significant differences between periods). Pre = baseline period 5 s prior to stimulus administration, S1 and S2 = consecutive 5-s periods during stimulus administration, P1 and P2 = consecutive 5-s periods after stimulus administration

Full size image

Positive Oral Actions

Proportion of Responding Newborns

Between 19 and 37% of newborns evinced positive oral actions, depending on the stimulus (familiar milk, MSH at 10 ng/l: 19%, control, SH at 10 ng/l, MSH at 1 ng/l, MSH at 50 ng/l: 26%, water: 35%, vanillin, SH at 1 ng/l: 37%). Cochran’s Q test did not reach significance for these odorants in the whole group, and also not for feeding mode or sex of newborns.

Duration of Positive Oral Actions

The results are depicted in Fig. 2. Friedman’s test was not significant for the whole group or for feeding mode. Segregated by sex, Friedman’s test was marginally significant for male newborns only (Friedman χ 2 8 = 14.3; P = 0.07). Wilcoxon tests indicated that males showed shorter oral responses to SH at 10 ng/l compared to water (P < 0.01; Fig. 3), vanillin and MSH at 1 ng/l (P < 0.05 in both cases). Further, MSH at 10 ng/l elicited shorter (P < 0.05) oral activity than did water. In summary, stimulus-dependent responses were evident in males only, with MSH at 10 ng/l eliciting shorter oral activity compared to water, and with SH at 10 ng/l eliciting shorter oral activity compared to water, vanillin, and MSH at 1 ng/l.

Fig. 2
figure 2

Medians and quartiles of the relative duration of positive oral (a,c) and negative oro-facial (b,d) movements of neonates in response to stimuli of Experiment 1 (a, b; N = 43) and Experiment 2 (c, d; N = 16). Values >1.5 times the interquartile range are denoted by a circle. See Results for significant differences. W = water, fM = familiar milk, van = vanillin, S = sweat, con = odorless control containing propylene glycol, SH = 3-sulfanylhexan-1-ol, MSH = 3-methyl-3-sulfanylhexan-1-ol, 3M2H = 3-methyl-2-hexenoic acid, HMHA =3-hydroxy-3-methylhexanoic acid

Full size image
Fig. 3
figure 3

Medians and quartiles of the relative duration of positive oral (left) and negative oro-facial (right) movements of neonates in response to stimuli of Exp. 1 for males only (N = 20). Values >1.5 times the interquartile range are denoted by a circle. See Results for significant differences. W = water, fM = familiar milk, van = vanillin, SH = 3-sulfanylhexan-1-ol, con = odorless control containing propylene glycol, MSH = 3-methyl-3-sulfanylhexan-1-ol

Full size image

Negative Oro-Facial Actions

Proportion of Responding Newborns

Depending on the stimulus, 16–35% of newborns displayed negative oro-facial actions (vanillin: 16%, familiar milk, MSH at 10 ng/l: 19%, control, MSH at 1 ng/l: 21%, water, SH at 10 ng/l: 23%, SH at 1 ng/l: 28%, MSH at 50 ng/l: 35%). Cochran’s Q test did not reach significance for these odorants in the whole group, and there was no impact of feeding mode. Segregated by sex, Cochran’s Q test indicated a trend for the effect of the stimulus in males (Q 8 = 12.8, P = 0.12). McNemar’s test indicated that more male newborns showed negative oro-facial actions in response to MSH at 50 ng/l compared to familiar milk (50% vs. 20%; P = 0.08), vanillin, SH at 1 ng/l and MSH at 1 ng/l (50% vs. 15%, P < 0.05 in all cases).

Duration of Negative Oro-Facial Actions

Results are presented in Fig. 2. Friedman’s test was not significant for the tested odorants and also not for feeding mode. Segregated by sex, Friedman’s test indicated a trend for an effect of the stimulus in male newborns (χ 2 8 = 12.9, P = 0.12). Wilcoxon tests indicated that male newborns showed longer negative oro-facial responses to MSH at 50 ng/l, compared to water, MSH at 1 ng/l, and the propylene glycol-containing odorless control (P < 0.05 in all cases; Fig. 3).

In summary, stimulus-specific responsiveness was apparent in male newborns only, with the longest facial activity occurring in response to MSH at 50 ng/l. MSH at 50 ng/l elicited longer negative responses in males compared with the odorless reference and MSH at 1 ng/l.

Adults’ Sensitivity and Hedonic Ratings

Adults chose the odorous stimulus with higher than chance level at concentrations ≥ 1 μg/1 and ≥ 0.7 μg/1 for MSH and SH, respectively (χ 2, P < 0.05 and P < 0.001). Overall, women performed better than men (Fig. 4). Thus, MSH was detected by women at 1 μg/1 (χ 2; P < 0.01), whereas a concentration of 38 μg/1 was necessary for this compound for olfactory detection by men (χ 2; P < 0.05). For SH, 0.7 μg/1 was detected by both men and women (χ 2; P < 0.05 and P < 0.01, respectively) but, with increasing concentrations, men again chose the correct sample only when 55 μg/1 was reached (χ 2; P < 0.01). Supra-liminal concentrations of the odorants were rated to be close to hedonic neutrality [mean ± SD = 5.4 ± 1.7 (MSH), 4.6 ± 1.7 (SH), on a 9-point Likert scale].

Fig. 4
figure 4

Sensitivity of adults to 3-sulfanylhexan-1-ol, 3-methyl-3-sulfanylhexan-1-ol, 3-methyl-2-hexenoic acid, and 3-hydroxy-3-methylhexanoic acid. Percentage of women (N = 10; black bars) and men (N = 10; white bars) choosing the correct glass rod in a triangle test. (χ 2, comparison with chance level: ‡ = P < 0.001, ** = P < 0.01, * = P < 0.05)

Full size image

Experiment 2: Neonatal Reactivity to the Odors of the Acids (3M2H and HMHA)

Respiratory Rate

The main effect of the stimulus period was significant (Fig. 1; F 4,60 = 3.7; P < 0.01). The mean respiratory rate was lower during the fourth than during the first and second 5-s interval after stimulus onset (LSD; P < 0.05 and 0.01, respectively). Baseline respiratory rate was higher compared with respiratory rate during the third and fourth 5-s interval after stimulus onset (LSD; P < 0.05 and 0.01, respectively). In summary, an overall decrease of respiratory rate after stimulus onset was observed, but no stimulus-specific respiratory pattern was apparent.

Positive Oral Actions

Proportion of Responding Newborns

Depending on the stimulus, between 19 and 50% of newborns evinced positive oral actions (3M2H at 80 μg/l, HMHA at 200 μg/l: 19%, water, vanillin, 3M2H at 30 ng/l: 31%, maternal sweat, HMHA at 0.5 and 50 ng/l: 38%, familiar milk: 50%). Cochran’s Q test did not reach significance on these percentages (Q 8 = 5.5; P = 0.7).

Duration of Positive Oral Actions

Results are depicted in Fig. 2. Friedman’s test did not reach significance (Friedman χ 2 8 = 7.9; P = 0.45), indicating that the different stimuli were not discriminated in terms of positive oral movements.

Negative Oro-Facial Actions

Proportion of Responding Newborns

Depending on the stimulus, between 38 and 81% of newborns showed negative oro-facial actions (vanillin: 38%, sweat: 50%, water, 3M2H at 80 μg/l, HMHA at 0.5 ng/l: 56%, familiar milk, 3M2H at 30 ng/l, HMHA at 200 μg/l: 63%, HMHA at 50 ng/l: 81%). Cochran’s Q test did not yield significant differences among these response frequencies across stimuli (Q 8 = 7.5; P = 0.49).

Duration of Negative Oro-Facial Actions

Friedman’s test was marginally significant (Friedman χ 2 8 = 13.8; P = 0.09). Newborns evinced longer negative oro-facial actions in response to HMHA at 50 ng/l compared with HMHA at 0.5 ng/l, familiar milk, maternal sweat, and vanillin (Wilcoxon; P < 0.05 for the former, P < 0.01 for the latter two stimuli; c.f., Fig. 2). Further, HMHA at 0.5 ng/l led to negative oro-facial actions of longer duration than did vanillin (Wilcoxon; P < 0.05).

In summary, up to 81% of neonates evinced negative oro-facial expressions. Stimuli were discriminated in terms of duration of negative reactivity. Thus, HMHA at 50 ng/l, but not at 200 μg/l, elicited higher responsiveness than HMHA at 0.5 ng/l, familiar milk, sweat, and vanillin, and HMHA at 0.5 ng/l led to higher responsiveness than vanillin.

Adults’ Sensitivity and Hedonic Ratings

Adults chose the odorous stimulus with higher than chance level at concentrations ≥ 40 mg/l and ≥ 576 μg/l for 3M2H and HMHA, respectively (χ 2; P < 0.05 and P < 0.001). For both compounds, women performed better than men (Fig. 4). Thus, for 3M2H, 120 mg/l was detected by women (χ 2; P < 0.05), whereas 360 mg/l was necessary for detection by men (χ 2; P < 0.01). For HMHA, a concentration of 576 μg/l was sufficient for above-chance detection by women, whereas 1.7 mg/l was necessary for detection by men (χ 2; P < 0.001 in both cases). Finally, supra-liminal concentrations of odorants were rated to be unpleasant [mean ± SD = 3.3 ± 1.8 (HMHA)] or close to neutral [mean ± SD = 4.2 ± 1.7 (3M2H)] on a 9-point Likert scale.

Discussion

The sensitivity and hedonic responsiveness to low concentrations of MSH and SH (Experiment 1), and HMHA and 3M2H (Experiment 2) were assessed in 3-d-old human infants. Upon delivering the odorants, as well as odorous and odorless reference stimuli, the infants’ respiratory rates and facial responsiveness were recorded.

Detection of the Sweat-Related Odor Stimuli

In line with the results obtained by Loos et al. (2014), respiratory rate was not discriminative for the olfactory stimuli. Oro-facial reactivity raised no strongly significant effects in the whole sample of newborns, with the odorous reference compounds (familiar milk, vanillin, and sweat) not being discriminated from the odorless stimulus. However, olfactory discrimination became apparent in male newborns exposed to MSH and SH. More precisely, males showed increased facial responsiveness to MSH at 10 and 50 ng/l, and to SH at 10 ng/l, relative to that to water. These results point to neonatal orthonasal detection of the odors of these thiols. In Experiment 2, HMHA at 50 ng/l led to increased facial responsiveness compared to HMHA at 0.5 ng/l, indicating neonatal detection of the acid. However, HMHA at 200 μg/l was not discriminated from HMHA at 0.5 ng/l. This finding, that the highest concentration induces a less intense behavioral response compared to intermediate concentrations, is reminiscent of a previous result with 5α-androst-16-en-3-one in newborns (Loos et al. 2014). The underlying reason is unclear. One may speculate about a non-linear concentration-response relation, as reported for several behavioral and perceptual brain processes (Burke et al. 2012; Coureaud et al. 2004; Nielsen et al. 2011).

With adults, concentrations in the μg/l range or higher were necessary for above chance olfactory detection of the acids and thiols in a triangle test. These values correspond approximately to threshold concentrations published so far for adults. MSH has been reported to be detected at 10 ppb in mineral oil (Hasegawa et al. 2004), SH retronasally at 22 ng/l in water (Starkenmann et al. 2008), HMHA, depending on the enantiomer, at 0.1 (S) or 10 ppm (R) in mineral oil (Hasegawa et al. 2004), and (E)-3M2H at about 1 ppm in mineral oil (Ferdenzi et al. 2015; Wysocki et al. 1993). In contrast with results reported by Troccaz et al. (2009), who found no gender variation of odor detection thresholds for MSH, HMHA, and 3M2H, women appeared to be more sensitive than men in the present study. Further investigations with a greater number of adult participants appear necessary to corroborate a possible difference in sensitivity between sexes.

Highly diluted odorants that elicited a behavioral response in neonates were not detected by adult participants in triangle tests, which could point to higher neonatal sensitivity. However, different response measures were used for infants and adults, which may account for divergent outcomes. Few investigators have assessed odor detection thresholds of neonates. Engen (1965) and Rovee (1969) determined infant detection thresholds to a homologous series of n-alcohols. Unfortunately, the threshold of adults was reported for propanol only, but was a factor of 10 lower compared to the average threshold for neonates. This lower sensitivity of neonates (at least to propanol) contrasts with the results obtained in the present study. Clearly, studies are needed to elucidate the developmental course of human olfactory and nasal trigeminal sensitivities, not only in neonates, but also in young children whose olfactory detection thresholds have been assessed only for a few odorants, with varying outcomes [e.g., phenylethanol (Poncelet et al. 2010), tetrahydrothiophene and (R)-carvon (Monnery-Patris et al. 2009), butanol (Lehrner et al. 1999), trimethylamine (Solbu et al. 1990)].

In summary, the results indicate neonatal orthonasal detection of HMHA at 0.5 ng/l and 50 ng/l, and detection by male newborns of SH at 10 ng/l, and of MSH at 10 and 50 ng/l. Since neonatal responses to 3M2H were not different from those to other stimuli, detection of the odor of this acid at concentrations used here remains uncertain. Adults detected all four compounds at the μg/l level, indicating lower adult sensitivity to MSH, SH, and HMHA. However, to ascertain lower adult olfactory sensitivity to individual odorants more fully, methods are needed that determine, in a comparable manner, individual olfactory detection thresholds in neonates and adults.

Hedonic Response to Sweat-Related and Other Odor Stimuli

The oral appetitive movements of newborns were not systematically increased by the odors of the thiols. Instead, male newborns evinced shorter oral movements to SH at 10 ng/l compared to water, vanillin, and SH at 1 ng/l. Further, longer oral movements were elicited by water compared to MSH at 10 ng/l. Thus, it appears that the odor of the thiols reduced the duration of appetitive oral movements in male newborns. MSH at 50 ng/l was even more efficient in eliciting negative oro-facial responses than several other stimuli, in terms of both percentage of responding male newborns and the duration of facial expressions, indicating a negative hedonic valence of its odor. No such response was evident in female newborns, suggesting a sex-dependent neonatal response to a human sweat odorant (c.f., Loos et al. 2014). Although female and male neonates behaved differently during certain double-choice tests (e.g., Delaunay-El Allam et al. 2006; Porter et al. 1991), we are not aware of a general sex-dependent difference in neonatal facial reactivity toward odors.

Female human sweat has been reported to contain, on average, a higher amount of MSH-conjugates than male sweat (Troccaz et al. 2009). However, no data are available exploring the influence of an infant’s sex on the occurrence of thiols and their conjugates in amniotic fluid and human milk. Thus, one may only speculate whether the difference between male and female responses is due to different neonatal hormonal levels (Doty and Cameron 2009; Scott et al. 2009), or based on different rates of pre- and/or post-natal exposure to these compounds.

The odor of the acids also did not enhance appetitive oral movements in human newborns. Instead, longer negative oro-facial expressions were shown in response to HMHA at 50 ng/l compared to HMHA at 0.5 ng/l and familiar milk, and in response to HMHA at 0.5 ng/l compared to vanillin. It can thus be hypothesized that HMHA conveys a negative valence for infants, whereas 3M2H was not detected or of neutral valence under the present conditions.

Reference stimuli were not discriminated from each other. However, several authors have shown that the odor of familiar milk is perceived by newborns, and elicits orientation and increased mouthing toward the stimulus (e.g., Marlier and Schaal 1997; Mizuno et al. 2004). Vanillin, too, has been reported to elicit increased mouthing (Steiner 1979), although with high inter-individual variability (Soussignan et al. 1997). In contrast, Doucet et al. (2009) did not observe increased non-nutritive sucking with familiar milk and vanillin, in line with the present results. In view of this conflicting evidence, parameters that influence the mouthing-sucking response to familiar milk (e.g., duration and multimodal context of presentation or prandial state) should be considered in more detail in future experiments.

In summary, the results of the present study point to a neutral or negative hedonic valence of the target odorants, which is in accordance with hedonic ratings from adults. Further, in the present experimental conditions, no clear-cut hedonic valence of the odors of vanillin, familiar milk, and sweat could be inferred from neonatal responses.

Negative Responses to Human Sweat Odorants

The present findings did not support the prediction (see Introduction) of a positive reactivity of newborns to the target stimuli due to olfactory learning in and/or ex utero. Possible explanations for the observed negative responses have been discussed already in a recent study on the detection and hedonic valence of the odor of androstenone in newborns (Loos et al. 2014), and only additional facets will be developed here.

First, in sweat, HMHA, MSH, and SH have mainly the S-configuration (R:S ratio = 28:72 for HMHA and MSH; Hasegawa et al. 2004; Starkenmann et al. 2005; Troccaz et al. 2004). Since odor quality and threshold can depend on the investigated enantiomer (Begnaud et al. 2006; Hasegawa et al. 2004; Starkenmann 2016; Troccaz et al. 2004), and no specific enantiomeric ratio was presented here, a novel olfactory percept might have been created in the human infant, leading to a neophobic response.

Second, release of considerable amounts of free odorants has been reported for human sweat [between 2 and 200 μg/l of (E)-3M2H in whole body sweat (Gordon et al. 1973), 100 μg/l of 3M2H in whole body sweat (Smith et al. 1969), and 2.2 μg/l of MSH in incubated underarm sweat (Troccaz et al. 2004)]. A neonate should thus detect the target odorants when smelling its mother’s sweat. In fact, even if no specific response to familiar sweat was observed here, it has been demonstrated elsewhere that neonates perceive maternal sweat (e.g., Cernoch and Porter 1985). Bottle-fed and breast-fed infants preferentially orient toward breast odor rather than the axillary odor of an unfamiliar breastfeeding mother [Makin and Porter (1989); although this might not hold for an infant’s own mother, Porter et al. (1992)]. Thus, axillary odor appears to be less attractive to human infants than breast odor; the neutral or aversive response found here might be interpreted in the context of an association of the target compounds with sweat.

In summary, intrinsic trigeminal properties of the odorants, perceptual novelty of the stimuli, or learned association of the stimuli with human sweat odor may explain the neutral or even aversive responses of human neonates to MSH, SH, and HMHA observed here.

Limitations

The interpretation of responses of preverbal organisms is not always unequivocal. For instance, it remains uncertain whether higher oral activity to MSH at 1 ng/l or lower activity to MSH at 10 ng/l underlies discriminative responsiveness to these two stimuli. This difficulty in interpreting results is exacerbated by the fact that no stimulus induced longer positive oral movements compared to the odorless reference. Thus, we cannot exclude the present experimental design having a low sensitivity for detection of positively valenced odorants.

Further, the neonates born by cesarean section were heterogeneous in terms of labor experience (cesarean sections with and without labor). Future studies should verify whether the experience of labor impacts the neonatal behavioral response to the investigated stimuli (c.f., Varendi et al. 2002). Finally, the coding procedure used in the present study does not allow a fine-grained analysis of the occurrence and duration of individual action units, nor of combinations thereof. It cannot be excluded that a more detailed analysis of individual facial action units might reveal further stimulus discrimination by neonates.

Outlook

In the present study, target odorants were studied in concentrations corresponding to their presumed concentrations in amniotic fluid and milk. It came out, however, that adult detection thresholds were several magnitudes higher. In future studies, it would be of interest to investigate neonatal responses to target odorants at adult threshold concentrations. Further, future studies should consider measuring a mother’s levels of odorants in order to account for individual differences in odor exposition of neonates, which might influence odor sensitivity to these compounds. Inter-individual differences could be maximized by a cross-cultural study design including different ethnicities and genotypes (Martin et al. 2010; Prokop-Prigge et al. 2016).

Conclusion

Neonatal facial responses to target odorants and reference stimuli led us to conclude that at least three of the four administered target odorants were detected by human newborns at the very low concentrations used here. This study lends further evidence that odorants occurring in complex mixtures in the perinatal environment are not automatically perceived as hedonically positive by an infant, especially when administered in pure form. Finally, our results point to a sex-dependent neonatal response to MSH and SH, and a higher olfactory sensitivity to these odorants in newborns compared to adults. These outcomes call for further investigations on developmental aspects of human sweat odor perception and on the quantitative occurrence of human sweat odorants in various body fluids.