Inherent sexual attraction through scent contact

As the main components of the HMW fraction consisted of a mixture of different MUP isoforms and their bound ligands [32, 33], these were further separated by anion exchange chromatography into four specific charge group fractions (I - IV) that contained different MUPs. The same four fractions were collected from B6 female urine, although female fractions I and IV contained very low levels of protein (Figure 2, Table 2). MUPs predominated in these fractions. Intact mass analysis by mass spectrometry confirmed that male fractions I and IV contained distinct MUPs expressed predominantly or exclusively by males (masses 18645Da and 18893Da, respectively, corresponding to known Mup genes that show sex-specific expression in the B6 mouse strain [33]). The MUPs in fractions II and III were expressed by both sexes (Figure 2) albeit at different concentrations (Table 2).

Stimulus		Protein (μg/μL)
Combined fractions I-IV	male	3.57
	female	0.47
Fraction I	male	2.96
	female	0.25
Fraction II	male	3.38
	female	1.39
Fraction III	male	3.32
	female	0.59
Fraction IV	male	0.69
	female	0.10

A mixture of these four MUP-containing fractions (excluding all other protein-free fractions) presented at concentrations similar to that in intact urine was highly effective in eliciting much greater time near the male stimulus (Figure 3A). The difference was due to the time under the stimulus but not sniffing rather than to active sniffing behaviour (Additional File 1: Figure S2 shows the breakdown of the time sniffing and not sniffing) and was thus similar to the response to the complete HMW fraction. When each charge fraction was tested separately, fractions II and III failed to stimulate significant preference for the male fraction (Figure 3A), although the MUP concentration was much greater in the male than in the female fraction in both cases (2.4 and 5.6 times greater, respectively; Table 2). Male fractions I and IV both stimulated greater attraction than the equivalent female fractions. As this might have been due to the absence of detectable MUP in female fractions I and IV (which would not be the case for intact urine signals), we tested the attraction to each of these male fractions when paired with female fraction III (the most attractive female fraction when tested against the equivalent male fraction). Although male fraction I contained five times more protein than female fraction III, the male fraction was not more attractive (Figure 3B). By contrast, male fraction IV elicited strong attraction compared to female fraction III (Figure 3B); these male and female fractions contained approximately equal concentrations of urinary protein but very different MUP isoforms.

In house mice, most urinary MUP isoforms are encoded by Mup genes in the central region of the Mup cluster on chromosome 4 where there has been recent rapid expansion of gene-pseudogene pairs [33]. Variation in expression of these MUP isoforms, which differ in sequence from each other by only a few amino acids, underlies the strong individual variation in MUP profiles between wild mice. These central Mup genes encode the urinary MUPs expressed by both sexes as well as the 18645Da isoform in male fraction I. While expression of the 18645Da MUP is male-specific among laboratory strains [33], it is expressed only by some wild mice and male-specificity varies between wild MUP genotypes (Figure 4). By contrast, the 18893Da MUP isoform in male fraction IV is encoded by a gene in the peripheral region of the Mup cluster where there is considerably greater divergence between Mup genes [33]. This protein is variously known as the '18893Da MUP', 'Peripheral region 2 MUP 17', 'Class A MUP 24' or 'MUP20' but may be best referred to according to the Mouse Genome Informatics Database (MGI) nomeclature: (MGI: Mup20, major urinary protein 20, MGI:3651981; Ensembl reference Gene: Mup20, ENSMUSG00000078672). This MUP is consistently expressed in the urine of male but not female wild mice across a large number of genotypes ([32] and unpublished observations). It shows unusually high mobility on reducing gel electrophoresis (Figure 1B), and is responsible for binding most of the male-specific volatile 2-sec-butyl 4,5 dihydrothiazole (thiazole) in mouse urine [32]. In order to highlight its unusual characteristics compared to all other known MUPs, and its role in female sexual attraction (see below), we named this 18893Da MUP as darcin (after Jane Austen's hero in Pride and Prejudice).

The tight binding of thiazole to darcin considerably extends the release of this volatile pheromone from scent marks over many hours; when artificially displaced from MUPs, thiazole evaporates from scent marks within minutes [8]. Thus, greater attraction to male fraction IV may be due to the binding of thiazole (and possibly other volatile ligands) to darcin. In order to explore this possibility, we compared attraction to male urine that was freshly deposited with attraction to urine deposited 24 h or 7 days prior to testing. Virtually all (> 98%) thiazole is lost 24 h after deposition under similar conditions [32]. If females are attracted by volatile components specifically bound to darcin in male fraction IV, attraction should reduce as scent marks age and volatiles are lost. However, scent age had no significant effects on the strength of attraction to the male sample whether intact urine or HMW stimuli were used in tests (Figure 5) and females continued to show significant attraction even when urine deposits were aged by 7 days.

Attraction to recombinant darcin

Continued attraction to aged male urine does not rule out the possibility that females are sensitive to extremely low levels of thiazole that may still be bound to darcin or, perhaps, to other unidentified components in male MUP fraction IV. In order to test attraction to darcin itself, we constructed an artificial gene, codon optimized to drive high level heterologous expression of the recombinant protein in Escherichia coli (Figure 6 and Additional File 1: Figure S4 showing codon optimized DNA sequence and translation). The recombinant darcin (r-darcin) had atypical mobility on SDS-PAGE (Figure 7C), suggesting similar folding to the native protein. When presented alone, at a physiological level similar to that in normal wild male or B6 strain urine, r-darcin stimulated significant attraction relative to both female urine and to a buffer control [Figure 7A(b, c); Additional File 1: Figure S3 shows the breakdown of time sniffing and not sniffing]. Indeed, response levels were very similar to a B6 intact urine control test [Figure 7A(a)]. Tests of other recombinant MUPs (expressed using the same system - see Additional File 1: Figure S5 - showing codon optimized DNA sequence and translation) confirmed that females showed no attraction either towards recombinant 18694Da MUP (which does not have sex-specific expression) or towards recombinant 18645Da MUP [expressed by male but not female laboratory mice; Figure 7A(d,e)]. Note that response to the 18694Da r-MUP had abnormally high variability [Figure 7A(d)] because one female spent an unusual 85s resting under this stimulus. Excluding this trial, the mean time under the recombinant 18694Da MUP was 6.4 ± 1.3s, very similar to the time normally spent under water, buffer or female urine. This complete absence of response to other recombinant MUPs confirms that attraction to r-darcin was not due to simple unfamiliarity or to a contaminant from the expression system.

Under normal conditions, darcin would not be encountered alone but as a constituent of urine, which also contains a large number of volatile molecules. Airborne volatiles are likely to be important for alerting attention to the presence of a scent and stimulating the close contact investigation that is necessary to detect involatile proteins [8]. As we were testing the inherent attraction that females show in response to contact with a male's scent, trials sometimes had to be excluded because females did not make contact with one or both stimulus locations during the 10 min test. When r-darcin alone was tested against female urine, there was a strong response and most females contacted the stimuli during their normal exploration of the test arena (only 2/15 trials were excluded). However, when r-darcin was tested against just a buffer control, 9/20 trials were excluded due to lack of contact. While this may have been a chance effect, the absence of any urinary volatiles in the arena may also have contributed by failing to stimulate investigation during these trials. In order to confirm that darcin elicits female attraction in the normal context of intact male urine, we also tested sexual attraction to urine from an inbred laboratory strain that expresses this MUP at only trace levels. Several of the laboratory strains derived through the Castle lineage express extremely low and sometimes undetectable levels of darcin [22], with BALB/c males expressing darcin at < 0.5% of their total MUP output (Figures 8, 9A). These classic laboratory strains have been bred in captivity over a considerable number of generations [34], divorced from the normal pressures of sexual selection and the need to attract females through scent. By contrast, adult male wild mice consistently express a high level of darcin (typically 10-20% of MUP expressed similar to B6 males, Figure 9B). As predicted, females were strongly attracted to urine from a random selection of males recently derived from the wild, but failed to spend significantly more time near male BALB/c urine than near female urine [Figure 7B(f,g)]. However, addition of r-darcin to male BALB/c urine at a normal male level stimulated a normal level of attraction [Figure 7B(h)]. They showed no such attraction when other recombinant MUPs were added to male BALB/c urine [Figure 7B(i,j)], confirming that darcin alone stimulates greater attraction to male than to female urine.

Thus, darcin acts as a sex pheromone that reliably elicits female sexual attraction to a male's urine scent, doubling the time spent near a male compared to a female urine mark. Unlike most mammalian scents that have been termed 'pheromones', we have shown that darcin meets the criteria widely accepted for the definition of the term pheromone [1, 5]: it is a species-specific molecule, produced deliberately for scent signalling, and elicits a clear behavioural response that is inherent and highly consistent across the large number of genetically variable wild-derived animals used in this study. It is also equally as effective in eliciting female attraction whether encountered in urine or presented alone. We expressed darcin and other MUPs as recombinant proteins to ascertain the role of these proteins in the absence of natural ligands. We cannot exclude the possibility that r-darcin carries exactly the same thiazole in the same stereochemical configuration as the mouse, but this is extremely unlikely. First, the ligand 2-sec-butyl 4,5 dihydrothiazole is not a central metabolite (the biosynthesis is not known) but is unlikely to be a metabolic pathway that is essential for life. Second, a comprehensive search of the metabolic pathways of E. coli reveals that the only thiazole known in E. coli metabolism is that related to biotin synthesis, an essential metabolic function that has been lost in mammals (hence it is a vitamin). Third, as one of the other recombinant MUPs tested (18694Da MUP) also binds thiazole in mouse urine [32] it provided a good control for putative contaminant ligands from E. coli, and clearly stimulated no response. Thus the response appears to be to darcin alone rather than to a darcin-ligand complex.

Although the response to darcin is inherent and highly consistent, this does not mean that female sexual attraction to male mice is inflexible. Nor does this mean that other components of male urine scent have no role to play in sexual attraction. Importantly, female attraction is to the individual owner of the scent encountered, not to males in general [17, 31]. As darcin is a single protein that is not polymorphic between males, it has no capacity itself for providing the individual-specific scent signatures that females need to recognize particular males, which is an essential component of selective mate choice. Instead, females learn an attraction to individual-specific airborne urinary volatiles on contact with male urine [17].

Learned attraction to individual-specific airborne volatiles

To test whether it is contact with darcin in male urine that stimulates this learned attraction to individual-specific airborne volatiles, we provided females with the opportunity to learn odours by pre-exposing them to a male and female urine stimulus prior to testing. We then tested female attraction to airborne volatiles from a male versus female urine stimulus when they were unable to contact the stimuli during the test. Using urine from males that expressed normal adult levels of darcin, females learned an attraction to airborne urinary volatiles if they had prior physical contact before the test either with urine from the same individual male [using stimuli from genetically diverse wild males, Figure 10A(a)], or with urine from a genetically identical male [Figure 10A(c)]. They did not learn any attraction if they were exposed only to the airborne urinary volatiles of the male before the test without any contact with involatile cues [Figure 10B(d,e)]. We also confirmed that females learned the specific airborne odours of the individual male on contact with male urine, as females pre-exposed to contact with urine from one male showed no subsequent attraction to airborne odours that came from a different genetically distinct individual male in the test [Figure 10A(b)].

By contrast, prior contact with urine from BALB/c laboratory males, which express only trace levels of darcin (see above), failed to elicit any learned attraction to the males' airborne volatiles [Figure 10C(f)]. However, when r-darcin was added to male BALB/c urine to a level expressed by normal males, contact elicited a strong learned attraction to airborne volatiles from the stimulus [Figure 10C(g)]. This was not because r-darcin altered the airborne volatile profile such that the volatiles themselves became more attractive to females: a control test confirmed that airborne volatiles from this stimulus were not attractive if females had prior contact only with water [Figure 10C(h)]. Instead, contact with darcin stimulates females to learn the individual-specific airborne odour associated with that male's scent and subsequently show attraction only to that airborne odour; contact with r-darcin in BALB/c male urine led to no attraction to airborne odours from males with different individual airborne odour signatures [Figure 10C(i)]. The individual urinary volatile profiles of mice are complex, and are influenced by many genes including MHC [35, 36], as well as by non-genetic differences such as bacterial microflora and diet [37, 38]. Darcin itself may also have some influence on the pattern of airborne volatiles learned on contact: while contact with BALB/c male urine plus r-darcin resulted in strong attraction to volatiles from this same stimulus, it resulted in only weak attraction to airborne volatiles from BALB/c male urine without r-darcin [Figure 10C(g,j)]. As darcin binds and slowly releases the volatile thiazole (and possibly other volatiles) from male scent marks, it is likely to have induced a slight mismatch between volatiles in the manipulated and unmanipulated BALB/c urine. Finally, to confirm that learned attraction to airborne urinary volatiles is a specific response to darcin, other recombinant MUPs added to male BALB/c urine failed to stimulate any learned attraction to airborne volatiles from the stimulus [Figure 10C(k,l)].

Subjects and urine donors

The subjects were 469 captive bred adult female Mus musculus domesticus (F0-F2) aged 3-12 months, from a colony derived from wild ancestors captured from five different populations in the northwest of England, UK. Most females (77%) were used in only a single test over a total of 614 trials, but a small number were used in two (17%), three (4%) or four (2%) tests, each involving different scent stimuli, with 2-16 weeks between successive tests. Females were housed in 45 × 28 × 13 cm cages (MB1, North Kent Plastics, Rochester, UK) in single-sex small family groups (2-4 sisters per cage during the test period) in a different room from the urine donors. As we have previously shown that females show very similar attraction responses to urine stimuli from unfamiliar males whether completely naïve to adult male scents or after natural social and sexual experience [17], subjects were not isolated from contact with other adult male scents in the animal unit and soiled bedding from captive bred wild males was regularly added to female cages during the testing period to ensure normal oestrus cycling.

The urine donors were 17 male and 14 female C57BL/6 laboratory mice (Harlan, Loughborough, UK) aged 60 days - 9 months, 12 male and 17 female BALB/c mice (Harlan, UK) aged 4-9 months, and 33 captive bred adult male house mice derived from the same colony as the female subjects but unrelated. Males were housed singly in 43 × 11.5 × 12 cm cages (M3, North Kent Plastics, UK) to ensure that there were no subordinate males whose scent was unattractive to females. Females were housed in single sex small groups (2-4) in 45 × 28 × 13 cm cages. Urine was collected by holding the donor mouse by the scruff of the neck over a clean 1.5 ml Eppendorf tube. Urine from 5-8 individual donors of the same strain and sex was pooled for testing, using different combinations of donors to create each stimulus pool. Pooling also provided a standard female control stimulus that combined urine from donors in different stages of the oestrus cycle rather than reflecting a particular stage. Urine samples from individual captive-bred wild donor males were not pooled as each has a different individual genetic signature. Urine was collected up to 1-2 weeks prior to testing and stored at -20°C until use. A single large pool was used in each fractionation experiment to ensure that fractions were identical between behavioural replicates and biochemical analyses.

Throughout, all animals were housed on a reversed 12:12 h light cycle with lights off at 0800 h, and were maintained on Corn Cob Absorb 10/14 substrate with paper wool nest material and ad libitum access to water and food (Lab Diet 5002 Certified Rodent Diet, Purina Mills, MO, USA). Cardboard tubes and red plastic mouse houses (Techniplast, NJ, USA) were provided for home cage enrichment.

Assay of attraction to male scent stimuli

Three days prior to each trial, soiled nest material and substrate from wild-derived males was introduced into the subject female's home cage to induce oestrus during the preference test [64]. Previous studies using this procedure on mice in our colony has shown that over 95% of females were likely to be in oestrus or proestrus during a test [31] and show a robust attraction to a male urine stimulus compared to an equivalent female stimulus when allowed to contact the urine stimuli [17]. All testing was carried out during the dark phase of the light cycle under dim red lighting.

Tests were conducted in a clean 45 × 28 × 13 cm arena (MB1 cage base fitted with a perforated Perspex lid [17]) to which females were familiarized for 30 min immediately before each test. In most tests, females were presented with a choice between an unfamiliar male urine stimulus and an equivalent female control stimulus that they could contact, placed within two 55 mm diameter circles that were 25 cm apart on the underside of the Perspex arena lid. Test stimuli were deliberately placed in the central zone of the test arena, where wild mice normally spend little time, so that increased time in these sites reflected attraction to the scent. In order to mimic two typical mouse urine scent marks, 10 μL of each test stimulus was streaked on to a 55 mm diameter glass microfibre filter (GMF) cut in half. These were then stuck onto the lid with Sellotape (GMF were used because the release kinetics of urinary volatiles has been characterized previously from this substrate [32, 65]). The position of the male and female control stimulus was randomized but balanced to ensure that an equal number of each stimulus type was presented on each side. Trials lasted 10 min and female behaviour towards the two stimuli was recorded remotely on DVD in a neighbouring laboratory. Transcription of the DVD recordings was carried out blind to the relative position of the two stimuli during each trial which were marked as A and B.

In order to test sexual attraction to airborne urinary volatiles alone, and the influence of prior contact with male-specific urine components, females were first pre-exposed for 30 min in their home cage to a 55 mm GMF cut in half and streaked with 10 μL of each test stimulus (or water). Filter papers were held outside a mesh sphere that allowed full nasal contact with the test stimuli, or inside the mesh sphere for pre-exposure to airborne volatiles only, following the procedure in [17]. Females were then tested in the arena as described above except that fresh test stimuli were suspended 2 cm above the perforated Perspex lid using a 5 cm diameter Perspex cylinder to ensure that mice could not physically contact the test stimuli during the test.

In order to assess sexual attraction, we tested whether females spent more total time under the male rather than the female or control stimulus (nose within the 55 mm diameter circle in which a test stimulus was presented) during the 10 min trials. Time under a stimulus consisted of time spent sniffing or not sniffing the stimulus, each of which were recorded separately to provide further insight into the female's response. Time spent sniffing up at the cage lid within each 55 mm circle indicates the female's interest in gaining further information from a scent source and is part of the information gathering process, thus most sniffing occurs in the first few minutes of a test. Time under the stimulus when not directly investigating the stimulus (defined as the subject's nose being within the 55 mm diameter circle but not sniffing up at the lid) reflects non-investigatory attraction to spend time close to the scent source, and is thus likely to be a more robust measure of attraction [17, 31]. In most cases, log transformation (s+1) of behaviour durations approximated normality (Kolmogorov-Smirnov and Shapiro-Wilks tests, P > 0.05) and allowed parametric analyses (matched pair t-tests to assess sexual attraction, parametric ANOVA to compare the log difference in response to the paired stimuli between tests). Non-parametric tests were used when log transformation did not approximate normality (Wilcoxon matched pair signed-ranks test). As it is widely established that females are more attracted to male than to female urine stimuli, and we needed to be conservative in concluding that manipulation of a urine signal eliminates this typical attraction to male urine, one-tailed statistical tests of greater attraction to male than to female or control stimuli were used throughout. In tests with scent contact, we also confirmed that the response observed after the first close sniff at a stimulus (when females may have gained involatile information from the scent) was the same as the response over the whole trial, and that time under a stimulus did not differ between the two stimuli prior to this first close contact sniff (data not presented).

Fractionation of urine stimuli

Samples were separated into HMW and LMW components using Vivaspin 500 centrifugal concentrators (Vivascience, Sartorious Group, Aubagne, France) with a 3000Da molecular weight cut-off. Vivaspins were washed twice using ddH₂O (500 μL added and centrifuged at 13800 rpm for 15 min). Vivaspins were emptied of water and urine (500 μL) was added and centrifuged at 13800 rpm for 15 min. The LMW component was then pipetted into a clean capped 1.5 mL Eppendorf tube, labelled and stored at -20°C until use. Urine was added to the Vivaspin to make up to 500 μL and the process repeated. The HMW and LMW components were pipette into separate capped Eppendorf tubes, labelled and stored at -20°C until use. Protein concentrations of each fraction were determined with the Coomassie plus protein assay reagent kit (Pierce, UK).

Anion exchange chromatography

Purification of individual MUP peaks from C57BL/6 urine was achieved by high-resolution strong anion exchange chromatography on either the Duo-Flow (Biorad Laboratories, CA, USA) or the Äkta (GE Lifesciences, Bucks, UK) chromatography platforms. Both systems were fitted with a Resource Q column (GE Lifesciences, V_t = 6 mL), equilibrated with several column volumes of 50 mM MES buffer, pH 5.0. Pooled urine was desalted on 5 mL Sephadex G-25 columns previously equilibrated with MES buffer, pH 5.0, then passed through a 0.45 μm filter. Typically 1 mL (~10 mg) of desalted protein was applied to the column. Bound protein was eluted from the column using a linear salt gradient of 0 - 1 M NaCl. Fractions (3 mL) were collected and fractions corresponding to individual MUP peaks were pooled then concentrated and desalted to deionised water (~0.2-0.3 mL final volume) using Vivaspin centrifugal concentrators (3 kD molecular weight cut off, Vivascience, Hannover, Germany). Final protein concentrations were obtained by Coomassie dye binding assay. Purity and mass was confirmed by native and SDS-PAGE as well as electrospray ionisation mass spectroscopy.

Sodium dodecyl sulphate polyacrylamide gel electrophoresis

SDS-PAGE was performed as described by Laemmli [66]. Concentrated anion exchange fractions were mixed 1:1 with reducing sample buffer (100 mM DTT) and then heated for 5 min at 95°C. All samples were run at a constant voltage of 200 V on 15% gels. Protein bands were visualized with Coomassie brilliant blue stain.

Electrospray ionization mass spectrometry (ESI/MS)

Concentrated anion exchange separated MUP peaks were centrifuged at 11,000 g for 10 min. All analyses were performed on an Easy-nLC nano high performance liquid chromatography system (HPLC; Proxeon Biosystems, Odense, Denmark) coupled to a QToF micro mass spectrometer (Waters, Manchester, UK), fitted with an ESI source. Samples (10 μL) were desalted and concentrated on a C4 reverse phase trap (LC packings). MUPs were eluted at a flow rate of 0.8 μL/min using repeated 0 - 100% acetonitrile gradients. Data were gathered between 800 and 1600 Th and were processed and transformed to a true mass scale using MaxENT 1 maximum entropy software (Waters Micromass, Massachusetts, USA). All data sets were processed at 0.25 Da/channel over a mass range of 18400 -19000 Da, and a peak width of 0.75 Da was used to construct the damage model. The instrument was calibrated initially using a 500 fmol/μL solution of Glu fibrinopeptide (Sigma, Munich, Germany) in 50% acetonitrile and 0.2% formic acid and calibrated during runs with a 1 pmol/μL solution of horse heart myoglobin (Sigma), 2 mM DTT in 0.2% formic acid. All water used was HPLC grade (VWR, Pennsylvania, USA)

Gas chromatography/MS

Hexane extraction of urine and analysis of volatiles was performed according to the methodology in [32], using a Thermo Fisher Scientific PolarisQ GC/MS.

Expression and purification of recombinant MUPs

The primary sequence of darcin was obtained from the sequence located under accession numbers NP_001012323/XP_355497 and was used to direct de novo gene synthesis for maximal expression in E. coli. The gene was codon optimised for expression in E. coli and cloned into pET28b via NcoI and XhoI restriction sites (Entelechon GmBH, Regensburg, Germany). The plasmid was used to transform BL21(λ)DE3 cells and darcin expressed in Luria Broth containing kanamycin (30 μg/mL). At OD_{600 nm} of between 0.6-0.8, the expression of recombinant darcin (r-darcin) was induced by the addition of isopropyl β-D-1 thiogalactopyranoside to a final concentration of 1 mM. Five hours post-induction, cells were harvested by centrifugation at 2000 × g and the cell pellets stored at -20°C prior to further purification. Harvested cells were lysed using Bugbuster protein extraction reagent (Novagen, Nottingham, UK) containing Complete™ EDTA-free protease inhibitor cocktail (Roche, Burgess Hill, UK). Darcin, present in the soluble fraction of the bacterial cell lysate was purified by virtue of the hexahistidine tag on nickel affinity columns according to manufacturers protocols (Novagen). Column fractions containing r-darcin were pooled and dialysed against 50 mM phosphate, 20 mM NaCl pH7.4. This preparation was used without further processing. The purity of r-darcin was assessed by SDS-PAGE analysis and protein concentration was determined by protein assay. Similar workflows were applied to express the 18645Da MUP (accession GB/AAH91744.1) and 18694Da MUP (accession NP_001157998.1).

Recombinant darcin and other recombinant MUPs were tested alone (11 μg in 10 μL buffer) or added to stimulus urine samples (11 μg to 10 μL BALB/c male urine) in order to mimic the natural concentration of darcin observed in B6 males (approximately 10-14% of total MUP) and in wild male mice [22, 32]. To ensure that both the manipulated urine and control stimulus were treated equally, an equivalent volume of 50 mM phosphate, 20 mM NaCl buffer, pH7.4 was added to the corresponding control stimulus.