Reproductive success
The production of urinary pheromones correlated with male but not female reproductive success (RS; defined in “Materials and methods” section). The most important predictors of male RS were total urinary protein concentration (75%) and social status (69%; Table 1; based on conditional model average sum of weights). The relative importance of age, creatinine, and mass ranged from 23 to 39%; PC ratio (protein:creatinine concentration) was excluded from the model due to collinearity (VIF = 6.97). Total urinary protein concentration during the enclosure phase was positively correlated with RS for males (Spearman R = 0.52, p = 0.01; Fig. 1a), but not females (Fig. 1b). This correlation is explained by the low protein concentration in the urine of non-reproductive males, as it is no longer significant after removing these males from the analysis (R = 0.12, p = 0.62; Supplementary Fig. S2). The median total urinary protein concentration was 5512 µg mL−1 and 5028 µg mL−1 for reproductive and non-reproductive males, respectively (Wilcoxon rank sum test W = 5, p < 0.001; Supplementary Fig. S2).
Reproductive success in relation to urinary protein and social status. Scatterplots show the total urinary protein concentration of males (a) and females (b) in relation to reproductive success. The boxplot (c) shows female and male social status in relation to reproductive success. Light gray coloration of data points and boxes indicate subordinate (S) social status during the enclosures. Black data points and dark gray boxes indicate dominant (D) social status. The black trend line in the scatterplots shows the loess (local regression) fit for non-parametric data (50% of data points to fit Epanechnikov kernel).
The most important predictors of female RS were mean body mass (89%) and social status (76%), whereas age, PC ratio, and total protein and creatinine concentration ranged from 14 to 20% (Supplementary Table S1). Female mean body mass during the enclosure was positively correlated to RS (R = 0.57, p = 0.004). When mean body mass during the enclosure is replaced with initial body mass as a model predictor, the relative influence of social status on female RS is 94%; all other variables ranged from 14 to 34% with initial body mass at 23% (Supplementary Table S1). For both sexes, dominant individuals (male = 12; female = 9) accounted for the majority of reproduction compared to subordinates (male = 12; female = 15; Welch’s t-test post hoc male p = 0.006, female p = 0.01; Fig. 1c). Reproduction in the enclosures resulted in 306 offspring from 51 litters (multiple paternity in 69%; Supplementary Table S2). Mate fidelity was 29% and 8% for males and females, respectively. The non-reproductive mice were all subordinates (male = 5; female = 8).
Male urinary VOC expression during the enclosure phase also correlated with male reproductive success. The explained variance (R2Y) and cross validation score (Q2) of orthogonal partial least-squares (OPLS) models showed a significant correlation between RS and VOC expression of denatured and intact urine (Fig. 2a; denatured: R2Y = 0.54, Q2 = 0.46; intact: R2Y = 0.51, Q2 = 0.39). Two specific urinary volatiles, HMH and TMA, correlated with male RS. In intact urine, peaks corresponding to HMH expression during enclosures were positively correlated to RS (Fig. 2b; R = 0.63, padj < 0.004), but this correlation is weak in denatured urine (R = 0.47, padj = 0.02 (n.s.)). We also confirmed that minor ions of HMH in intact urine were correlated with male RS (8 HMH peaks: R > 0.61, padj < 0.004). TMA was negatively correlated with RS during the enclosure phase, regardless of protein conformation (Fig. 2c; intact: R = − 0.59, padj < 0.004; denatured: R = − 0.55, padj < 0.008). After omitting non-reproductive males, the correlations between reproductive male RS and both HMH and TMA expression were not significant (HMH: R = 0.23, p = 0.36; TMA: R = − 0.12, p = 0.62; Supplementary Fig. S2). Significant differences in HMH and TMA expression were observed when comparing reproductive and non-reproductive males (Wilcoxon rank sum test p < 0.003 for both VOCs; Supplementary Fig. S2).
Male reproductive success in relation to VOC expression. OPLS scores plot of reproductive success based on candidate MS-data derived from denatured male urine collected during the enclosure phase (a). The x-axis of the scores plot is the predictive component (t1) of the RS response variable. The y-axis is the first orthogonal component (to1). Data points for dominant and subordinates are labeled D and S, respectively. Coloration of the data points indicates the range of RS measured for males; high and low RS range from red to blue, respectively. The Spearman rank correlation of HMH and Trimethylamine expression (b and c, respectively) with RS shown for intact urine from dominant (black) and subordinate (gray) males during the enclosure phase. The black trend line in the scatterplots shows the loess fit for non-parametric data (50% of data points to fit Epanechnikov kernel).
Male RS was correlated with both pheromone excretion and social status, and therefore, we examined dominants and subordinates separately and re-ran OPLS models to isolate the effect of VOC expression on reproduction. VOC expression and RS did not correlate among dominant males (OPLS model Q2 < 0, p > 0.05), whereas the VOC expression of intact urine from subordinate males was strongly correlated with RS, and to a lesser degree in denatured urine (intact: R2Y = 0.75, Q2 = 0.64; denatured: R2Y = 0.59, Q2 = 0.49). We found a correlation of subordinate male RS with HMH expression (R = 0.71, p = 0.01), and a negative correlation with TMA (R = − 0.70, p = 0.01), though neither were significant after Bonferroni adjustment for multiple comparisons (refer to “Materials and methods”).
Among females, we found no significant associations between VOC expression during the enclosure phase and RS (OPLS models: R2Y and Q2 p > 0.05; Supplementary Table S3). We also examined whether VOC expression before enclosure phase could predict RS, but OPLS models based on the female and both male MS-datasets showed no significant correlations (R2Y and Q2 p > 0.05).
Male urinary proteins
Male urinary protein excretion in seminatural conditions depended upon social status. Urinary PC ratio (ln transformed) of dominant males significantly increased over time and became higher during the enclosure phase than before (pairwise Tukey post hoc p < 0.04; Supplementary Table S4; Fig. 3a). In contrast, the PC ratio of subordinate males did not vary throughout the experiment (post hoc p > 0.41). Linear mixed effects (LME) modelling reveals that the factors of social status (F1,76 = 4.3, p = 0.04), time point (F4,76 = 5.3, p < 0.001), and their interaction (F4,76 = 3.3, p = 0.01) all had a significant effect on PC ratio. Age had a marginal effect on PC ratio (F1,76 = 3.26, p = 0.07), but not body mass (F1,76 = 0.5, p = 0.47). Male urinary creatinine concentration (ln transformed), as with PC ratio, changed after release into the enclosures, depending upon social status. Creatinine concentration significantly decreased in dominant male urine during the enclosure phase compared to before (post hoc p < 0.02; Supplementary Table S4; Fig. 3b), whereas PC ratio increased. Urinary creatinine concentration of subordinate males did not vary significantly throughout the experiment (post hoc p > 0.45). The factors of social status (F1,76 = 5.4, p = 0.02), time point (F4,76 = 4.3, p = 0.004), and their interaction (F4,76 = 3.1, p = 0.02) all had significant effects on urinary creatinine concentration, but not age or body mass. The LME model of total urinary protein concentration showed a significant increase over time for both social status groups (F4,76 = 15.0, p < 0.001; Supplementary Table S4; Fig. 3c), but was not associated with social status, age, or body mass (all p > 0.12).
Expression of urinary protein in relation to enclosure phase. Line graphs of PC ratio (ln transformed, a), urinary creatinine concentration (µg mL−1 (ln transformed, b)), and total urinary protein concentration (µg mL−1, c). Solid and dashed lines indicate males and females, respectively. Black and gray color indicate dominant (D) and subordinate (S) social status, respectively. Note, February is the before enclosure phase measurement; all other time points were during. Error bars are ± 1 SEM.
We indirectly measured urinary MUP20 production based on liver RNA transcription 14 days after the enclosure phase; however, the LME model average of hepatic Mup20 gene expression showed no association with social status, RS, or total urinary protein or creatinine concentration in male mice. Predictor importance ranged from 29 to 14%, suggesting a weak, non-significant correlation between Mup20 transcription and age (29%, R = 0.21, p = 0.35), as well as RS (27%, R = − 0.25, p = 0.28; Supplementary Table S1). Social status was the least important predictor of Mup20 transcription (14%). A similar pattern was observed when the response variable was absolute hepatic Mup20 transcription. Predictor importance ranged from 25 to 14% with age and RS as the most important (both 25%; Supplementary Table S1) and social status the least.
Male urinary VOCs
We used OPLS models to examine correlations between protein concentration and VOC expression in male urine. Total protein in denatured urine during the enclosures showed a stronger correlation with VOC expression than intact urine both before (denatured: R2Y = 0.68, Q2 = 0.63; intact: R2Y = 0.40, Q2 = 0.22; Fig. 4a) and during the enclosures (denatured: R2Y = 0.89, Q2 = 0.62; intact: R2Y = 0.38, Q2 = 0.15; Fig. 4b). Regardless of urinary protein conformation, HMH peaks correlate with protein concentration of urine collected before the enclosures (intact: Pearson R = 0.67, padj < 3.8E−3; denatured: R = 0.77, padj < 0.005). Other pre-enclosure correlations between VOCs and urinary protein concentration depended on conformation, including SBT from denatured urine (R = 0.74, padj < 0.005) and TMA from intact urine (R = 0.21, padj < 3.8E−3). No peaks correlated with total protein concentration of intact or denatured urine during the enclosures (padj > 0.003).
Male VOC expression in relation to urinary protein concentration. OPLS scores plots of total urinary protein concentration based on candidate MS-data derived from denatured male urine (n = 23) collected before (a) and during (b) the enclosure phase. The x-axis of the scores plot is the predictive component (t1) and the y-axis is the first orthogonal component (to1). Data points for dominant and subordinate males are labelled D and S, respectively. Coloration of the data points indicate the range of urinary protein concentration (µg mL−1); high and low concentration range from red to blue, respectively.
We tested whether the expression of VOCs in standard conditions predicted male social status during the enclosure phase. The discriminant analysis (OPLS-DA) of VOC expression in denatured urine collected before the enclosure phase did not reliably discriminate males that became dominant during the enclosure phase (Fig. 5a; full MS-data: R2Y = 0.5, Q2 = − 0.121, misclassification rate (mcr) = 0.17; candidate MS-data: R2Y = 0.311, Q2 < − 0.01, mcr = 0.26; Fig. 5b). Furthermore, the VOC peak expression and total ion chromatogram (TIC) intensity of pre-enclosure urine did not significantly differ based on the social status the individual obtained during the enclosure phase (Welch’s t-test of TIC: full MS-data p = 0.54; candidate MS-data p = 0.55).
VOC expression in relation to social status. OPLS-DA scores plots of social status before (a,b) and during the enclosures (c,d) based on the full MS-data (a,c) and the candidate MS-data (b,d) derived from denatured male urine (n = 23).The x-axis of the scores plot is the predictive component (t1) and the y-axis is the first orthogonal component (to1). Data points for dominant (black) and subordinates (light gray) are labelled D and S, respectively. The boxplots show differential expression of HMH and 4-methyl-6-hepten-3-one (e and f, respectively) in intact urine for dominant (D) and subordinate (S; dark and light gray, respectively) males and females (lattice; n = 24) at both enclosure phases. Different letters above the boxplots denote significant differences.
There was a strong association between male social status and urinary VOC expression during the enclosure phase. The OPLS-DA of full MS-data showed robust separation of dominant and subordinate males based on VOC expression of denatured urine collected during the enclosures (R2Y = 0.79, Q2 = 0.65, mcr = 0.04; Fig. 5c). The denatured urine model of candidate MS-data also showed separation by social status but to a lesser degree (R2Y = 0.62, Q2 = 0.51, mcr = 0.13; Fig. 5d). The models of intact urine VOC expression also discriminate social status but to a lesser degree than the denatured urine models (intact:full: R2Y = 0.69, Q2 = 0.41, mcr = 0.17; intact:candidate: R2Y = 0.55, Q2 = 0.51, mcr = 0.17; Supplementary Table S3). In models of full MS-data, one peak in intact urine and 88 peaks in denatured urine were upregulated in dominant males. The peaks correspond to HMH in denatured urine (mean difference = 1.2E6, Wilcoxon rank-sum post hoc padj < 5.5E−4; Fig. 5e), and 4-methyl-6-hepten-3-one in both urinary protein conformations (denatured: mean difference = 4.6E4, padj < 3.5E−4; intact: mean difference = 1.0E4, padj < 4.4E−4; Fig. 5f). Details for differentiating the spectra of these compounds is in Supplementary Fig. S3. Based on full MS-data, dominant males have a higher TIC intensity than subordinates when comparing denatured urine (mean difference = 1.3E7, p = 0.02), whereas this pattern was not significant for intact urine (mean difference = 7.8E6, p = 0.2). In models of candidate MS-data, peaks that correspond to HMH were upregulated in dominant male intact and denatured urine. The TIC intensity of candidate MS-data did not differ between dominant and subordinate males, regardless of urine conformation (intact urine p = 0.62; denatured urine p = 0.28).
Female urinary proteins
Female mice showed a significant increase in protein excretion (PC ratio) after being released in the enclosures regardless of their social status (Fig. 3a). We observed a significant effect of time point on female PC ratio (LME: PC ratio (ln transformed): F4,75 = 3.3, p = 0.02; Supplementary Table S4), but not for social status, age, body mass, or status:time point interaction (all p > 0.55). Time point also had a strong effect on the LME model of total urinary protein concentration (Fig. 1c; F4,75 = 9.9, p < 0.001; Supplementary Table S4). Female mice significantly upregulated total protein concentration and PC ratio during the enclosure phase (Feb-Mar pairwise Tukey post hoc comparison for both D and S p < 0.001; Fig. 3a,c). Age and body mass had a marginal effect on urinary protein concentration in females (age: F1,75 = 3.6, p = 0.06; mass: F1,75 = 2.8, p = 0.09), but not social status or status:time point interaction (all p > 0.34). The LME of urinary creatinine concentration (ln transformed) was not significantly affected by the model variables (all p > 0.18; Supplementary Table S4), and although stochastic, mean values did not vary significantly between time points (Fig. 3b).
Female urinary VOCs
Total urinary protein concentration was correlated with VOC expression in denatured female urine, as observed for males but to a lesser extent for female urine (Supplementary Table S3). Total protein concentration of denatured urine collected during the enclosures showed a slightly stronger correlation to VOC expression compared to before the enclosures (before: R2Y = 0.68, Q2 = 0.44 during: R2Y = 0.71, Q2 = 0.28). A positive correlation with total protein concentration was observed for 10 peaks before and 2 peaks during the enclosure phase; the VOC(s) to which the peaks correspond were not identified. The OPLS models of female urine examining intact total protein concentration, or PC ratio and creatinine of both intact and denatured urine did not correlate with VOC expression regardless of enclosure phase (p > 0.05; Supplementary Table S3).
Unlike males, VOC expression was not associated with social status in females, regardless of protein conformation and enclosure phase. The OPLS-DA of full MS-data moderately discriminate social status with low predictive ability in denatured female urine (R2Y = 0.52, Q2 = 0.37, mcr = 0.04), and to a lesser extent in intact urine (R2Y = 0.47, Q2 = 0.19, mcr = 0.17). For both intact and denatured urine analyses, there were no significant differences in peak intensity based on social status. The TIC intensity was slightly higher for subordinate females during the enclosures, but this difference was not significant (intact:D mean TIC = 2.5E7, intact:S mean TIC = 3.2E7, p = 0.14; denatured:D mean TIC = 2.3E7, denatured:S mean TIC = 2.8E7, p = 0.36). The OPLS-DA of denatured female urine before enclosure phase was not related to social status (R2Y = 0.46, Q2 < 0.01, mcr = 0.20). There were no expression differences in specific peaks and females that became subordinate during enclosures showed a slightly higher TIC intensity than dominants, though this difference was not significant (before:S mean TIC = 1.3E7, before:D mean TIC = 1.2E7, p = 0.65). With regard to specific female pheromones, the peaks corresponding to 2-heptanone did not correlate with female RS, social status, or urinary protein excretion (R2Y and Q2 p > 0.05; Supplementary Table S3). Two other female pheromones, IBA and 2,5-dimethylpyrazine, were not detected in any samples.
Sexual dimorphism of chemosensory signals
Total urinary protein concentration and PC ratio increased significantly during the enclosure phase in both sexes (generalized mixed model (GLMM)); protein concentration Χ2 = 28.1, φ = 0.77, p < 0.001; PC ratio Χ2 = 28.6, φ = 0.77, p < 0.001; creatinine concentration Χ2 = 4.6, φ = 0.31, p = 0.3 (n.s.); Supplementary Table S5). Overall, the mean values of PC ratio and both protein and creatinine concentration were significantly greater for males than females (all p < 0.001). There was a significant sex:housing interaction on urinary protein (Χ2 = 43.8, φ = 0.96, p < 0.001) and creatinine concentration (Χ2 = 9.1, φ = 0.44, p = 0.002), and a marginal effect on PC ratio (Χ2 = 3.7, φ = 0.28, p = 0.053). The interaction result indicates greater sex differences in protein concentration in standard housing conditions (M:F ratio = 8.5; Supplementary Table S5) compared to seminatural enclosure conditions (M:F ratio = 5). Similarly, the degree of sexual dimorphism in urinary creatinine decreased from before (M:F ratio = 1.7) to during enclosure phase (M:F ratio = 1).
Sexual dimorphism in urinary volatiles was discernible after controlling for protein conformation and enclosure phase. OPLS-DA of intact urine better discriminate the sexes before rather than during enclosures (before: R2Y = 0.87, Q2 = 0.62, mcr = 0.04; during: R2Y = 0.82, Q2 = 0.7, mcr = 0.09; Supplementary Table S3). The expression of 82 peaks representing IT, SBT, TMA, and HMH (Fig. 5e) showed a male bias in pre-enclosure intact urine. During the enclosures, we observed a sex-biased expression of 74 peaks (female:male bias 8:66) in intact urine. Peaks representing TMA and SBT were upregulated in males during the enclosure phase, while females upregulated 4-methyl-6-hepten-3-one, which was also upregulated in the denatured urine of dominant males (Fig. 5f). Male TIC intensity of intact urine was greater than female TIC intensity before (mean difference = 1.7E7, p < 0.001) and during the enclosure phase (mean difference = 1.4E7, p < 0.001). As observed with urinary protein levels, the sexual dimorphism of intact urine TIC intensity was significantly greater before compared to during the enclosure phase (M:F before = 2.1; M:F during = 1.5; Χ2 = 11.6, φ = 0.71, p < 0.001; Supplementary Table S5).
OPLS-DA of sexual dimorphism are improved when analyzing denatured versus intact urine. Sex discrimination based on VOC expression of denatured urine is more accurate during than before the enclosure phase (before: R2Y = 0.72, Q2 = 0.51, mcr = 0.04; during: R2Y = 0.89, Q2 = 0.84, mcr = 0.04; Supplementary Table S3). The expression of 88 peaks representing 4-methyl-6-hepten-3-one, HMH, and TMA showed a male bias in pre-enclosure denatured urine. During the enclosures, we observed male-biased expression of 76 peaks, with upregulations of DHB, IT, SBT, and TMA in denatured urine. Male TIC intensity of denatured urine was greater than female TIC intensity before (mean difference = 2.1E7, p < 0.001) and during the enclosure phase (mean difference = 2.1E7, p < 0.001). Consistent with the intact urine result, the sexual dimorphism of denatured urine TIC intensity significantly decreased during the enclosure phase (M:F before = 2.6; M:F during = 1.8; Χ2 = 7.9, φ = 0.59, p = 0.005; Supplementary Table S5).
Source: Ecology - nature.com
