OUP user menu

Murine scent mark microbial communities are genetically determined

Clare V. Lanyon, Stephen P. Rushton, Anthony G. O'Donnell, Mike Goodfellow, Alan C. Ward, Marion Petrie, Susanne P. Jensen, L. Morris Gosling, Dustin J. Penn
DOI: http://dx.doi.org/10.1111/j.1574-6941.2006.00252.x 576-583 First published online: 1 March 2007


Scent marking in mice allows males to communicate information such as territory ownership, male competitive ability and current reproductive, nutritional, social and health status. It has been suggested that female mice eavesdrop on these olfactory cues, using them as a means of selecting mates with dissimilar major histocompatibility complex (MHC) genes, known as H2 in mice. The mechanisms underpinning MHC-dependent olfactory communication remain unresolved. Using congenic mouse strains and molecular methods we explore the involvement of the microbial communities, a known source of odourants, in scent marks to test the hypothesis that the microbial communities and hence the olfactory signals are genetically determined. Here we show that the indigenous microbial community of murine scent marks is genetically determined. Both background genotype and H2 haplotype influence the community structure of the scent mark flora, removing the possibility that community composition is solely orchestrated by the MHC. Qualitative and quantitative components of the bacterial community associated with MHC haplotype and background genotype were identified. The analyses confirm that the four groups of congenic mice tested are distinguishable on basis of the microbiology of their scent marks alone, strengthening the role of microorganisms in the development of MHC-dependent odours.

  • commensal microbial communities
  • congenic
  • murine
  • major histocompatibility complex


The high degree of polymorphism observed at the major histocompatibility complex (MHC) makes these genes excellent candidates for determining individual odourtypes; genetically-influenced body odour that distinguishes one individual from another (Penn & Potts, 1998; Yamazaki et al., 1998; Penn, 2002). MHC influences the composition of odour compounds in many vertebrates such as fish (Olsen et al., 1998; Aeschlimann et al., 2003), lizards (Olsson et al., 2003), birds (Zelano & Edwards, 2002), mice (Yamazaki et al., 1976), rats (Singh et al., 1990; Schellinck et al., 1995) and humans (Ober et al., 1997). In some species this odour signature acts as a cue for MHC-dependent behaviours such as mate and social preferences (Penn & Potts, 1998). In semi-natural populations and in inbred strains of mice these signals contribute to imprinted mating preferences and frequency of pregnancy blocks that serve to promote heterozygosity (Potts et al., 1991; Penn & Potts, 1999; Beauchamp et al., 2000; Ihara & Feldman, 2003). The underlying mechanisms controlling how MHC genes influence odour remain, however, unresolved.

MHC-mediated odours are thought to be soluble molecules, bound peptides or metabolites produced endogenously and transported by a protein carrier molecule (Penn, 2002). It is suggested that commensal microbial communities volatilize these molecules/peptides that contribute to the production of individual odour or amplify the signal by providing a substrate for selection of odourants by MHC molecules (Yamazaki et al., 1998). As such, the community structure of indigenous microbial communities is shaped either directly by the MHC genes themselves (microbial communities hypothesis; Howard, 1977) or indirectly by utilisation of the molecules, peptides or secondary metabolites (microbial communities-peptide hypothesis; Penn & Potts, 1998). A recent study found that the composition of gastrointestinal microbial communities of mice varies among MHC-congenic strains (Vaahtovuo et al., 2003), supporting the genetic modulation of indigenous microbial communities and thus the involvement of the microbial communities in MHC-mediated odours. Conclusions from experimental tests of the microbial communities/peptide hypothesis are mixed (Singh et al., 1990; Yamazaki et al., 1990; Schellinck et al., 1995). Evidence from two independent studies with germ-free mice and rats is conflicting (Schellinck & Brown, 1992). Furthermore, untrained sniffer animals can distinguish urine from congenic mice but are unable to discriminate haplotypes from those lacking microbial communities (axenic) (Singh et al., 1990; Roser et al., 1991; Schellinck et al., 1995). Yet, when trained for example by water deprivation, the urinary odour of axenic mice can be distinguished (Yamazaki et al., 1990), indicating the salience of MHC cues in the absence of microbial communities, although trained animals will be highly motivated to respond to any signal. As such, indigenous microbial communities may not be necessary for the production of odourtypes per se, but are shown to influence the ability to discriminate urinary odours (Schellinck et al., 1991), suggesting that they are needed to amplify the signal or contribute to MHC-mediated cues (microbial communities-peptide hypothesis).

Experimental tests of the involvement of the microbial communities in MHC-mediated odours focus predominantly on behaviour and biochemical analysis. To date, analysis focussing on the community structure of indigenous microbial communities has largely been ignored. We investigated whether murine scent mark microbial communities are genetically regulated using molecular approaches and strains of congenic mice. The relative contributions of both MHC and background genotype associated components of the microbial population were determined. The findings support the involvement of indigenous microbial communities in the production of MHC-mediated odour.

Materials and methods


Twenty eight-week-old male mice of each of two background strains, hereafter termed genotype and two MHC haplotypes (H2b and H2d), were obtained from Harlan UK: BALB.B/OlaHsd (hereafter called BALB/c.B), BALB/c.DOlaHsd (BALB/c.D), C57BL/10.ScSnOlaHsd (C57BL/10.B) and C57BL/10.D2/nOlaHsd (C57BL/10.D). Mice were single-housed in MB2 (20 × 10 × 13 cm) cages in mixed strain racks and maintained under controlled temperature (17–28°C) and light conditions (reversed 12 : 12 h light:dark cycle with lights on at 22 : 00 h), with ad libitum access to water, shredded paper nesting materials, cardboard tubes and softwood chewing sticks for enrichment. To minimize the possibility of mice being from the same litter, nine of the 20 mice purchased (of each strain) were selected at random for use in this study. Scent marks were collected when the mice were aged 14–23 weeks.

Scent mark collection

Glass microscope slides (76 × 26 × 1.5 mm thick) were frosted on one side using silicon carbide (carborundum) paper and cut into two strips (76 × 13 mm). Slides were cleaned with Dri-Decon, rinsed in water and oven-dried overnight at 110°C. For scent mark collection, glass slides were placed frosted side up on a clean glass slide (76 × 38 × 1.5 mm thick) resting on a pad (76 × 38 × 5 mm thick) on an upside-down aluminium tin (10 × 5.4 × 3.2 cm deep). A second tin containing a square hole (65 × 30 mm) was placed on top to hold the slides in place and to raise the scent mark area, thereby reducing contamination by materials from the cage bottom. The sampling device was placed in a mouse cage for scent mark collection for 1.5 h at the beginning of the dark phase. Slides were removed from the metal holders, placed in a 40-mL glass vial with a polypropylene screw cap and a Silicone/PTFE septum and stored at −80°C.

DNA extraction and PCR amplification

DNA was extracted from the glass slides using the Fast Prep DNA SPIN kit (Anachem) with modifications. Glass slides were rotated at room temperature with the cell suspension and cell lysing buffer for 30 min, after which the nucleic acids were extracted as described by the manufacturer. Approximately 10–15 ng of template DNA was used in a 30 μL PCR reaction mixture containing: 2 U of DyNAzyme EXT (Finnzymes, FIN-02201), 1X PCR buffer (2.0 mM Tris – HCl, 1.5 mM MgCl2) (Finnzymes), 2.5 mM deoxynucleoside triphosphate mix (Biogene, Cambs, UK), 0.5 μm of both Eub338f (Whiteley & Bailey, 2000) and 530r primers (Muyzer et al., 1993) and sterile Millipure water to a volume of 30 μL. Cycling parameters were 94°C for 10 min followed by 33 cycles of 94°C 1 min, 53°C 1 min and 72°C for 1 min with a final extension of 72°C for 10 min. The amplified products included the insertion of a GC clamp at the 5′ end of the gene (Muyzer & Smalla, 1998). To accommodate any PCR biases all samples were analyzed (extracted, PCR amplified and DGGE separated) in triplicate and treated as individual samples throughout the analysis.

Denaturing gradient gel electrophoresis (DGGE)

Mouse-specific, 16S rRNA gene sequence profiles were produced by separation of the heterogeneous PCR products using DGGE. DGGE was performed using the DCode system (Bio-Rad, Hercules, CA) as described previously (Muyzer et al., 1993). A reference marker, constructed using clones of marker bacteria, was included on each gel to verify the continuity of the gradient and to facilitate comparison between patterns. Gels were stained for 1 h in 1X TAE buffer containing Sybr-Green (molecular probes) before digitization.

Gel processing and pattern recognition by principal component analysis

DGGE gels were processed and digitized using bionumerics gel analysis software (Applied Maths, Sint-Martens-Latem, Belgium). Lanes were aligned relative to the internal markers and the intensity of the bands measured (both manually and automatically using software algorithms) to provide a matrix of 30 quantitative variables (band intensities) that described the scent mark microbial communities of each of the 36 mouse samples. Multivariate pattern recognition was done using principal components analysis (PCA) in the simca-p software package (version 10.0, Umetrics AB, Umeå, Sweden). Before PCA, the data were mean centred, a process equivalent to subtracting the mean gel profile from all other DGGE profiles in the data set and carrying out subsequent analyses on the mean-subtracted data (Keun et al., 2002).

Discriminant analysis

DGGE band intensity characteristics for each individual mouse were used as mouse-group descriptor variables in a linear stepwise multiple discriminant analysis (sMDA) (James, 1985) in an attempt to assess the extent to which the congenic strains (both genotype and H2 haplotype individually and collectively) could be distinguished on the basis of their microbial communities. It was assumed that the prior probability of a mouse belonging to any of the groups was equivalent and set at 0.25. Band intensity data were arc-sin transformed before analysis. The discriminating power of each band was assessed using the conditional F-ratio, which tests the difference between each mouse group, with respect to each variable, conditional on those bands being already included in the discriminant function. It is equivalent to the F-ratio used in an analysis of covariance testing group differences using all variables currently included as covariates. Bands were added sequentially to the classification until inclusion of any further band produced an F-ratio that was not significant at P<0.05. The leaving-one-out estimate of error was then used to assess classification error based on the final group of selected band variables. In this, individual mice were left out sequentially. The analysis was then repeated on the smaller data set and the new discriminant functions used to assign the excluded mouse to the groups. The error rate was the proportion of mis-assignments out of the total number of leaving-one-out runs (in this case the total number of misclassifieds/36). Four analyses were undertaken. The first assessed the extent to which the four congenic mouse strains (BALB/c.B, BALB/c.D, C57BL/10.B, C57BL/10.D) could be distinguished using their DGGE band characteristics. The second analysis analysed the discrimination of the genotypes. The third considered separation of the haplotypes and the fourth, the extent to which the congenic strains could be distinguished when genetically dissimilar.


Figure 1 shows the scent mark DGGE profiles from four congenic mice strains differing both in background genotype and in H2 haplotype: BALB/c.B, BALB/c.D, C57BL/10.B and C57BL/10.D consisting of nine males selected from a stock of 20 at random. DGGE profiles were compared between and within congenic strains to determine whether MHC genes or background genes influence the composition of microbial communities in scent marks. The individual impact of both H2 haplotype and background genotype was determined through strain pair comparisons (Table 1). DGGE profiles were digitized before analysis (Applied Maths). A data matrix consisting of a total of 248 identified bands assigned to 30 variables (band classes) through band matching algorithms was subjected to PCA (Fig. 2); an exploratory technique used to reduce dimensionality of a data set. The data were analyzed semiquantitatively using total band volume as the measure.


Composite PCR-DGGE profiles representing the bacterial diversity in scent marks generated from individual ♂ mice from four congenic strains (BALB/c.B, BALB/c.D, C57BL/10.B and C57BL/10.D). Mice were housed in mixed rack strains and were selected at random from a purchased stock of 20 individuals. The loadings of the samples do not represent the actual DGGE loadings or the housing arrangement.

View this table:

Mouse model system showing strain combinations used to compare the PCR-DGGE profiles of scent mark bacteria and examine the individual influence of background genotype (BALB and C57BL10) and MHC haplotype (H2b/d>)on the commensal scent mark flora

Combined contribution of background genotype, H2 haplotype- Test 1
C57BL10.B (H2b) vs. C57BL10.D (H2d) vs. BALB.B (H2b) vs. BALB/c.D (H2d)
Same background genotype, different H2 haplotype – Test 2
C57BL10.B (H2b) vs. C57BL10.D (H2d)
BALB.B (H2b) vs. BALB/c.D (H2d)
Different background genotype, same H2 haplotype – Test 3
C57BL10.B (H2b) vs. BALB.B (H2b)
C57BL10.D (H2d) vs. BALB/c.D (H2d)
Different background genotype, different H2 haplotype – Test 4
C57BL10.B (H2b) vs. BALB/c.D (H2d)
C57BL10.D (H2d) vs. BALB.B (H2b)

Principal component analysis (PCA) of all DGGE profiles from BALB/c.D (▪), BALB/c.B (▲), C57BL/10.B (●) and C57BL/10.D (◆) mice. The eigen values for the first PC was ë 6.53 and for the second PC ë 3.66. The plot shows that each of the four strains had distinct scent mark flora, which were fairly homogeneous between within-mouse strains, suggesting that scent mark microbial communities are genetically regulated. Specifically the data discriminated according to background genotype (BALB mice separated from C57BL/10 mice) along PC 1 (29.2%) and H2bd haplotype along PC 2 (16.7%).

Figure 2 shows that DGGE profiles can be readily distinguished between the four congenic strains along both the first (x axis) and second (y axis) principal component, indicating that murine scent mark flora between the congenic strains are unique. Furthermore, the within-strain differences are far less than the between-strain differences. This is particularly evident between mice of the BALB/c.D strain whose scent mark flora was predominantly homogeneous in composition. Analysis of the second PC shows discrimination of the microbial communities according to H2 haplotype between the C57BL/10.B vs. C57BL/10.D and the BALB/c.B vs. BALB/c.D strains. Separate PCA analysis of the individual influence of background genotype and H2 haplotype (Table 2, strain comparisons) on the microbial community revealed that both H2 haplotype (C57BL10.B vs. C57BL10.D and BALB.B vs. BALB/c.D comparison) and background genotype (C57BL10.B vs. BALB.B and C57BL10.D vs. BALB/c.D comparison) discriminated along the first PC (data not shown), indicating that individually both genetic components drive the structure of murine scent mark flora. Separate analysis of C57BL10.B vs. BALB/c.D and C57BL10.D vs. BALB.B (data not shown) showed that differences between the microbial communities are greatly reduced when the mice compared are genetically dissimilar, i.e. when there are no similarities between background genotype and MHC haplotype. The reduced discrimination between these strain comparisons corroborates the role of MHC haplotype and background genes in shaping microbial communities.

View this table:

Output from linear stepwise discriminant function analysis on the 2 x 2 strain comparisons

Congenic strainBALB H2d
C57BL/10 H2d19, 18, 21 Different genotype, same H2 TEST 3C57BL/10 H2d
BALB H2b1, 23, 29, 4, 8, 3, 23, 13 Different H2, same genotype TEST 211, 12, 19, 17, 6 Different genotype, different H2 TEST 4BALB H2b
C57BL10 H2b14, 26, 6, 23, 18, 30 Different genotype, different H2 TEST 4Not testable Different H2, same genotype TEST 21, 28, 30, 16, 11 Different genotype, same H2 TEST 3
  • The bands identified in bold (in order of significance) are those responsible for the discrimination between the strain comparisons

Stepwise multiple discriminant function analysis (sMDA) was used to determine the relative contributions of both background genotype and H2 haplotype (individually and collectively) to the observed differences in scent mark flora (Table 2). In all tests the leaving-out estimate of error was zero as none of the mice were mis-classified. Twelve variables were identified as significant discriminators between the four congenic strains. These components in order of inclusion in the model are given in Table 2. Post hoc reclassification of the original data led to a 100% re-assignment of the individual mice to their respective groups on the basis of the 12 identified variables. Eight variables (in combination) were significant discriminators between groups of mice with the BALB background genotypes (H2bd). sMDA of the C57BL/10 genotypes was not feasible due to the low number (2/24) of common bands between haplotypes.

Discrimination between groups of mice with same H2 haplotype but different background genotype (test 3) revealed that in the case of the H2b haplotypes, five band classes were significant discriminators between the strains, whereas four were identified as discriminators of the H2d haplotypes (Table 2). In total, 20 of the 30 band classes detected were found to be significant discriminators between the congenic strains (P< 0.05). Six of the 20 variables identified as being statistically significant were associated with H2 haplotype, whereas only four were associated with background genetics, suggesting that H2 has a larger contribution to the discrimination observed amongst the congenic strains. The genetic background contributes to the observed discrimination but influences different bacterial bands to that of MHC. The association of five of the variables could not be attributed to either background genotype or H2 haplotype as they were ubiquitous to test 2 and test 3 (Table 1). As such it is unknown whether these variables are due to a combined influence or are common to both MHC and background genetics.


The results of this study demonstrate that murine scent mark microbial communities are genetically regulated. The four congenic strains tested (BALB/c.B, BALB/c.D, C57BL/10.B, and C57BL/10.D) have discrete microbial communities varying in both qualitative and quantitative components, corroborating previous findings on the genetic determination of faecal indigenous microbial communities (Vaahtovuo et al., 2001, 2003). Mice were housed in mixed strain racks and acclimatized to the shared environment for a minimum of 6 weeks before sampling, thus reducing the potential contamination from exogenous microbial communities. As such the differences observed in the scent mark communities can be attributed to host genetics. The within-strain homogeneity in the scent mark microbial communities of individually housed mice substantiates this attribution.

The MHC constitutes c. 0.5% of the genome and accounts for 50% of genetically determined odourtypes among inbred strains of mice (Eggert et al., 1998; Yamazaki et al., 1998). As such the production of chemosignals is not exclusively co-ordinated by the MHC genes (Eggert et al., 1998; Roberts & Gosling, 2003). Both background genotype and H2 haplotype influence the community structure of the scent mark flora, removing the possibility that community composition is solely orchestrated by the MHC, corroborating chemical tests of chemosensory individuality. This study provides evidence of a small number of MHC-influenced and background genotype bacterial components of the scent mark community. The number of significant bands contributing to the discrimination between congenic strains was greater for MHC than for background genotype. However, neither MHC nor background-associated components of the bacterial community dominated in the discrimination between the congenic strains, suggesting that differences in scent mark flora are codependent on both MHC and background genotype influences. This is further supported by data from both chemical and behavioural studies of MHC-mediated odour (Beauchamp et al., 1990; Eggert et al., 1996; Yamazaki et al., 1998) which indicate that either MHC and background genetic influences are of the same amplitude or that both influences are not independent of each other. As such MHC-dependent signals cannot occur autonomously. Our results support the view that murine odourtypes are composed of both MHC and background influences that are interdependent (Eggert et al., 1996). As scent marks provide a mechanism for phenotypic discrimination, allowing animals such as mice to select a mate based upon their MHC dissimilarity, the finding that scent mark bacteria are genetically determined and that bacteria are a known source of odour molecules, substantiates the role of microbial communities in the production of MHC-mediated odour.

The cellular and molecular mechanisms by which these class I and II molecules, which are ‘designed’ for peptide presentation, shape the structure of the indigenous microbial communities remain unclear. MHC molecules are soluble glycoproteins that are expressed endogenously (Penn & Ilmonen, 2005). The molecules and/or the bound peptides are transported to the preputial, coagulating, axillary region or other microbe-harbouring glands (Penn, 2002). Independent of odour production, the MHC molecules, peptides or metabolites could provide a discriminating carbon source for opportunistic bacterial metabolism, thus shaping the community structure. This is supported by the microbial community/peptide hypothesis where the indigenous microbial communities volatilize MHC-components into odoriferous compounds (Penn & Potts, 1998). Moreover, class I and II MHC molecules are highly polymorphic transmembrane glycoproteins that control immunological self/nonself recognition. Through antigen presentation, MHC molecules control all specific immunological responses – both cell- and antibody-mediated – and as such influence resistance and susceptibility to infectious and autoimmune diseases (Penn & Ilmonen, 2005). An immune elimination of indigenous bacterial species, specific to the congenic strain, would account for the observed differences in scent mark flora. Although little is known about the effect of MHC on antibacterial responses (Toivanen et al., 1993; Toivanen et al., 2001), this suggestion supports the microbial communities hypothesis that indigenous bacterial populations are directly shaped by the MHC genes themselves. Alternatively, the creation of the congenic strains (Penn, 2002) could also explain the observed differences in the microbial communities. The BALB/c.B haplotype was created by crossing BALB/c.D with C57BL/10J (H2b) and then backcrossing with BALB/c.D for 13 generations. The d haplotype of C57BL/10 was created by crossing with DBA/2 (H2d) and then backcrossing with C57BL/10.B for 11 generations (Klein et al., 1983). This could lead to a steady accumulation of random mutations, resulting in unexpected phenotypic variations. However, the data do not support this, as components of the scent mark bacteria are identified as MHC-associated and are required to orchestrate the overall discrimination in the data, thus reducing the possibility that the differences are solely due to evolutionary divergences. Future work using F2 segregants from the following crosses – BALB/c.B × BALB/c.D and C57BL/10/B × C57BL/10.D – would randomize potential background variations, thus controlling for these possible variations.

Demonstration of the combined genetic influence on murine scent mark community structure corroborates previous behavioural and biochemical tests of MHC-mediated odours and as such supports the microbial communities/peptide hypotheses. Other hypotheses proposed (Penn & Potts, 1998) cannot be ruled out. The microorganisms might be involved in the degradation of MHC molecules, leading to the excretion of the volatilized molecules themselves [MHC molecule hypothesis (Singh et al., 1987)], the peptides bound to them [peptide hypothesis (Singer et al., 1997)] or to the replacement of their peptides by volatile compounds (carrier hypothesis Pearse-Pratt et al., 1992).

In conclusion, the data confirm that the four groups of congenic mice are distinguishable on basis of the microbiology of their scent marks alone and that specific components of the bacteria respond differently to the MHC and unknown loci in the background genetics. Thus we propose that the acquisition of microbial communities is not random but controlled by a genetic element. However, the role of the environment in scent mark community dynamics remains unknown. Identification of the specific components via PCR-sequence analysis or bacteria-specific PCR analysis would further elucidate the response of the indigenous microbial communities under genetic control. Further studies should determine whether the results obtained can be corroborated using other combinations of inbred strains or F2 segregants. Additionally, increasing sample size could strengthen data trends through the inclusion of new observations, variables and thus variation in the data. However, the observed variation determined by the PCA (Fig. 2) is actual variation derived from analysis of the data and as such the trends are compelling. To substantiate the claims for the involvement of the microbial communities in the production of MHC-mediated odours, analysis of the odour profiles derived from scent marks combined with the microbial community analysis would yield important clues regarding the mechanisms underpinning chemosensory individuality.


We thank S. Jenkins and I. Waite for laboratory assistance, our colleagues Dr S.C Roberts, Prof. A. Clare, Dr H. Wang and Dr S. Savelev. We thank staff at the Comparative Biology Centre for animal husbandry services, G.M. Vallons for assistance in sample collection, C. Richardson and P. Flecknell for veterinary supervision and H. Wang, S. Savelev, B. Smith, K.J. Madden and J.F. Marshall for help in developing the scent mark collection device. This work was supported by ARO grant DAAD19-03-1-0215. Opinions, interpretations, conclusions and recommendations are those of the authors and are not necessarily endorsed by the United States Government. Approved for public release, distribution unlimited.


  • Editor: Julian Marchesi


View Abstract