Reduced genetic variation among hosts may favour the emergence of virulent infectious diseases by enhancing pathogen replication and its associated virulence due to adaptation to a limited set of host genotypes. Here, we test this hypothesis using experimental evolution of a mouse-specific retroviral pathogen, Friend virus (FV) complex. We demonstrate rapid fitness (i.e. viral titre) and virulence increases when FV complex serially infects a series of inbred mice representing the same genotype, but not when infecting a diverse array of inbred mouse strains modelling the diversity in natural host populations. Additionally, a single infection of a different host genotype was sufficient to constrain the emergence of a high fitness/high virulence FV complex phenotype in these experiments. The potent inhibition of viral fitness and virulence was associated with an observed loss of the defective retroviral genome (spleen focus-forming virus), whose presence exacerbates infection and drives disease in susceptible mice. Results from our experiments provide an important first step in understanding how genetic variation among vertebrate hosts influences pathogen evolution and suggests that serial exposure to different genotypes within a single host species may act as a constraint on pathogen adaptation that prohibits the emergence of more virulent infections. From a practical perspective, these results have implications for low-diversity host populations such as endangered species and domestic animals.
Genetic variation in host populations can prevent the emergence of infectious diseases in two non-mutually exclusive ways: by directly reducing pathogen transmission rates (e.g. resistant individuals in a contact network block potential routes of transmission), and by creating variable host environments that result in fitness trade-offs to rapidly adapting pathogens . In the latter case, fitness trade-offs arise as a consequence of pathogen adaptive responses that enhance fitness in one host genotype at the cost of fitness in another host genotype. This phenomenon is often referred to as antagonistic pleiotropy.
Low genetic diversity in host populations may reduce fitness trade-offs associated with specialization and promote the expansion of highly virulent pathogens , a phenomenon long appreciated by plant pathologists and termed the monoculture effect [2–7]. Given the practical implications of this phenomenon, it is surprising that only a handful of empirical studies have quantified the role of genetic variability among animal hosts in limiting infectious disease. Numerous serial passage experiments have shown that fitness trade-offs arise as a consequence of pathogen adaptation to a novel host [8,9]. However, most of these studies focus on adaptation to new host species rather than genotypes within a single host. Empirical evidence that genetic diversity in an animal species limits pathogen adaptation and virulence evolution comes primarily from invertebrate studies [10–12]. There is a need to empirically address this hypothesis in vertebrate host–pathogen systems, particularly because their unique adaptive immune system will heavily influence the outcome of host–pathogen interactions.
The mouse-specific pathogen Friend virus (FV) complex is a well-studied host–pathogen model. The term ‘complex’ refers to the presence of two viruses: the replication-competent Friend murine leukaemia virus (F-MuLV) and the replication incompetent spleen focus-forming virus (SFFV). Synergistic interactions between these viruses promote the erythroproliferative disease observed in susceptible mouse genotypes (electronic supplementary material, figure S1) [13–15], and multiple host polymorphisms have been identified that influence the susceptibility of mice to infection [16–19]. Previous work in our laboratory has shown that serial infection of FV complex through a single mouse genotype can promote adaptive responses that enhance fitness in that host genotype at the cost of reduced fitness in other host genotypes [20,21]. Moreover, resistance variability among host genotypes can significantly influence both the magnitude of an adaptive response and the degree of host specialization (i.e. magnitude of fitness trade-offs) exhibited by this pathogen . These data suggest that genetic diversity among hosts may constrain the evolution of more virulent infectious diseases.
Expanding upon these previous results, the experiments described here sought to empirically address the hypothesis that genetic variation among individuals of the same mammalian species, the house mouse (Mus musculus), acts as a constraint on pathogen adaptation and virulence evolution. To that end, we serially passaged FV complex through a series of individuals from the same mouse genotype or through a series of individuals representing different mouse genotypes and compared patterns of viral fitness (i.e. viral titres) and disease virulence that emerged in response to these differing selection regimes. We predicted that serial passage of FV complex in a single host genotype would result in adaptive responses that increased pathogen fitness and virulence and that FV complex serially passaged through a series of genetically distinct hosts would result in significantly lower fitness and virulence. Experimental results demonstrate that serial infection of a diverse array of host genotypes, and even a single round of host alternation, places a powerful constraint on the capacity of FV complex to evolve higher fitness and virulence.
2. Material and methods
(a) Host model
Several genotypes of inbred mice were purchased from Jackson laboratories for use in the experiments described here, namely BALB/cJ, C.C3 (a.k.a. BALB.KK), DBA/2J, A.SW, A/J, C3H/HeJ and 129X1/svj (see the electronic supplementary material, table S1, for more details on mouse strains). Animals were either bred under specific-pathogen-free conditions at the University of Utah or purchased as needed. All experimental animals were females between 8 and 16 weeks of age. The BALB/c and DBA/2J genotypes were chosen as test hosts for two primary reasons. First, we wanted to standardize the host environment across virus stocks to focus on the effect of selection regime on patterns of pathogen fitness and virulence, and using these two genotypes provided us with independent experimental replicates. Second, because of their high sensitivity to FV complex infection, we reasoned that these genotypes would be useful models for readout of low fitness/low virulence viral phenotypes, whereas other more resistant genotypes might clear or be refractory to such infections. Owing to constraints on availability of mouse strains used in five-round experiments (A.SW, BALB.KK), and the requirement of susceptibility to FV complex infection and FV-induced disease, we chose to use the 129X1/svj, A/J and C3H/HeJ strains as alternate host genotypes in two-round serial passage experiments.
(b) Pathogen model
An NIH3T3 fibroblast cell line (3–6a) containing a biological clone of an NB-tropic strain of FV complex (i.e. an NB-tropic strain of F-MuLV complexed with the polycythemia strain of SFFV) was kindly provided by Dr Sandra Ruscetti (NIAID). This biological clone was grown in tissue culture and supernatants were collected to produce stocks of ‘unpassaged’ FV complex, which were subsequently used to initiate all serial passage experiments.
(c) Serial passage of FV complex
All passage lines were initiated by infecting animals via intraperitoneal (i.p.) injection with equivalent volumes (200 µl) of the unpassaged FV complex stock described above. Mouse-to-mouse passage of FV complex consisted of serially passaging FV complex from the spleens of infected animals to a new host via i.p. injection of virus-laden spleen supernatants. Infected spleens were homogenized in 15 ml conical tubes containing an equivalent weight/volume of sterile 1× phosphate-buffered saline (PBS) (0.1 g = 100 µl 1X PBS), centrifuged at 5000 r.p.m. for 5 min at 4°C, and their supernatants were stored at −80°C in 500 µl aliquots until use. Virus stocks were thawed once immediately prior to use. Infected materials were kept on ice during all steps of virus stock preparation. Variability in infectious dose between passage rounds was controlled by infecting animals with equivalent volumes of different virus stocks whose concentrations were adjusted during the w/v dilution of spleen homogenates. Two independent serial passage designs were employed in this study and are described in detail below.
(i) Five-round serial passage experiments
Two independent experiments were conducted (designated BALB/c or DBA/2J after the host genotypes that served as test-phase genotypes). For each experiment, FV complex was serially passaged five times through a series of female animals from the same Mus inbred strain (BALB/c or DBA/2J) or through females from five different Mus strains (figure 1a). Thus, one ‘same host genotype’ and one ‘different host genotypes' virus stock were created for each of the independent BALB/c and DBA/2J experiments. A total of six different mouse strains were used for these five-round serial-passage experiments and all have previously been shown (BALB/c, DBA/2J, 129X1/svj, A.SW, BALB.KK) [20–22], or are shown here (electronic supplementary material figure S2), to be susceptible to FV complex infection. Infected animals were monitored daily, with each round of infection lasting 10–12 days depending on the health status of the animal. To increase the likelihood of successful passage as well as increase volumes of passage stocks, two animals from the same strain were infected per round of infection and their spleen supernatants were pooled. To expand our stock of passaged virus for test-phase infections, 5–10 animals were infected during the fifth passage round. At the test phase, independent cohorts of female DBA/2J and BALB/c animals (n = 8–10) were infected with equivalent volumes of virus stocks derived by serial passage through the same or different host genotypes. Cohorts of animals from each genotype (DBA/2J = 20 animals, BALB/c = 40 animals) were also infected with unpassaged virus to obtain baseline viral fitness and virulence estimates. To further control infectious dose in the test phase, viral stocks were titred using an infectious particle assay and animals received roughly 1500 focus-forming units (FFU) (i.e. infectious particles) per 200 µl i.p. injection. Test-phase infections lasted 10 days.
(ii) Two-round serial passage experiments
Two independent experiments were conducted (again denoted as ‘BALB/c’ or DBA/2J’ experiments). Experiments consisted of serially passaging FV complex for two sequential rounds of infection through animals from the same host genotype or through animals representing two different host genotypes (figure 1b). A/J, C3H/HeJ and 129X1/svj inbred Mus strains were chosen for use as alternate host genotypes because they are susceptible to FV complex infection (electronic supplementary material, figure S2). To increase statistical power, a total of 10 independent ‘same host genotype’ and 30 independent ‘differnt host genotypes’ passage lines (10 independent lines for each of our three alternate host genotypes) were created for comparison in the test phase for both the DBA/2J and BALB/c experiments. Spleen supernatants were prepared as described above. Briefly, after the first round of infection, spleen supernatant from one BALB/c or DBA/2J animal was split into aliquots of equivalent volumes and used to infect either another animal of that respective genotype or an animal representing one of the three ‘alternate’ host genotypes. Spleen supernatants were diluted as described above, and all animals during second round infections received between 75 and 150 µl i.p. injections.
At the test phase, independent cohorts of female DBA/2J and BALB/c animals were infected with equivalent doses of virus derived by passage through the same or different host genotypes. To do this, volumes of supernatants used to infect test-phase animals were adjusted based on viral titres of infecting stocks. Specifically, the lowest mean proviral load estimate from spleen supernatants derived from one of the four cohorts of second round animals was designated the ‘target’ dose and the fold difference of all stocks above this value was used as the dilution factor. Viral stock dilutions were made in sterile 1X PBS immediately prior to i.p. injection of test animals. The entirety (approx. 250 µl) of spleen supernatants that had proviral load estimates lower than the target dose was used to infect test-phase animals.
(d) FV complex fitness estimate
An important parameter of pathogen fitness is within-host growth rate, which we estimated by measuring viral titres in infected animals. A quantitative PCR assay was developed using the Lightcycler 2.0 platform (Roche) to measure the number of integrated retroviral genomes (proviruses) in host DNA and has been detailed previously [20,21]. Briefly, after weighing spleens from euthanized infected animals to estimate virulence, organs were suspended in an equivalent weight/volume of sterile 1× PBS and then mechanically homogenized. All work with infected animal tissues was carried out in a certified BSL2 laminar flow hood. Homogenates were kept on ice until being centrifuged at 5000 r.p.m. for 10 min. High-quality genomic DNA was extracted from 100 µl of spleen supernatant using a DNeasy DNA extraction kit (Qiagen). Concentration (ng µl−1) and quality (260/280 ratio of approx. 1.8–2.0) of DNA samples were checked using a Nanodrop spectrophotometer (Thermo Scientific). Stock DNA samples were used to create 20 ng µl−1 sample dilutions for use in quantitative PCR assays (50 ng total gDNA per reaction). Because spleens were homogenized in an equal weight/volume ratio of 1× PBS during tissue processing and because we standardized qPCR reactions to contain 50 ng of genomic DNA, qPCR estimates only reflect proportional differences in viral titres (i.e. viral genomes per copies of GAPD). To account for these dilution effects, we analyse viral fitness data standardized (i.e. multiplied) by spleen weight to approximate absolute loads of virus (i.e. viral genomes per spleen). Both standardized values (main document) and unstandardized values (electronic supplementary material) are reported, but we limit our description of results to standardized data for brevity. Viral load estimates are represented as proviral copies per 1 000 000 host cells (i.e. 106 GAPD copies).
(e) Virulence estimates
Infection by FV complex of susceptible animals results in an erythroproliferative disorder (electronic supplementary material, figure S1). Major symptoms associated with acute disease caused by FV complex infection are the development of gross organ enlargement (-megaly) due to clonal expansion of virally infected erythroblasts in the spleen and liver (primary and secondary sites of terminal haematopoeisis, respectively), as well as elevated haematocrit values due to the overabundance of RBCs in the blood (i.e. polycythemia)  (electronic supplementary material, figure S1). Thus, to estimate disease associated with acute FV complex infection, animals were euthanized and their spleens and livers were removed and weighed to the nearest 100th of a gram. Haematocrit was measured as the percentage of red blood cells (RBCs) per volume of whole blood. Splenomegaly is used as our primary estimate of virulence for two reasons. First, the spleen is the primary site of FV complex replication . Second, hepatomegaly is only weakly correlated with F-MuLV titres in our experiments and haematocrit is not correlated with F-MuLV titres at all (electronic supplementary material, figure S3). A total of 75 µl of whole blood was drawn from the femoral artery of euthanized animals into a heparinized haematocrit tube and spun at 2500 r.p.m. for 5 min in a centrifuge. The length of the RBC pack was divided by the total length of volume in the haematocrit tube (measured to the nearest millimetre) to obtain the haematocrit value (%RBCs) for each animal. Splenomegaly and hepatomegaly are correlated estimates of virulence, whereas polycythemia is not correlated with either splenomegaly or hepatomegaly.
(f) Statistical analysis
All statistical analyses were carried out using Prism v. 5 (Graphpad) or JMP v. 9.0 (SAS). Owing to a lack of normality and/or unequal variance in many datasets, a non-parametric Mann–Whitney U-test was used for all pair-wise statistical comparisons. Linear correlation analysis was performed on virulence and proviral load estimates (genomes per 106 GAPD copies).
(a) Five-round serial passage experiments
Thirty-seven experimental infections were conducted during the test phase of five-round serial passage experiments. Serial infection of animals from the same host genotype resulted in significant increases in viral fitness compared with unpassaged control virus in both the BALB/c (40-fold increase) and DBA/2J experiments (20-fold increase; figure 2a; electronic supplementary material, figure S4). Serial infection of different host genotypes led to a significant increase in fitness over unpassaged virus in the BALB/c experiment (Mann–Whitney U-test, p = 0.0002), but the magnitude of fitness increase (4-fold) was significantly lower than that observed after serial infection of the same host genotype (figure 2a; electronic supplementary material, figure S4). Serial infection of different hosts in the DBA/2J experiment did not result in any change in fitness compared to unpassaged virus (Mann–Whitney U-test, p = 0.95; figure 2a; electronic supplementary material, figure S4).
We next contrasted patterns of disease severity resulting from infection with FV complex stocks derived from our two selection regimes. Across all infected animals, there is a significant positive correlation between F-MuLV fitness and splenomegaly up to about 1 g of spleen, after which the relationship plateaus (R2 = 0.34, ANOVA, F1,175 = 87.9, p < 0.0001; electronic supplementary material, figure S5). In comparison with unpassaged virus, splenomegaly was significantly increased during infection with virus stocks derived by serial passage through individuals from the same host genotype, but was not different (BALB/c experiment) or was significantly reduced (DBA/2J experiment) in animals infected with virus stocks derived by passage through different host genotypes (figure 2b; electronic supplementary material, figure S6).
During five-round serial passages, we observed that exposure to the 129X1/svj genotype in passage round 4 was associated with a significant decline in FV-induced virulence (splenomegaly) in both passage lines (electronic supplementary material, figure S7). This effect persisted in subsequent infections even when the virus was exposed to the highly susceptible BALB/c genotype in the DBA/2J alternating passage line (electronic supplementary material, figure S7a). Thus, the reduction in F-MuLV fitness and FV complex-induced disease observed after infection with alternately passaged FV complex could have been due to the 129X1/svj genotype. To address this possibility, we followed up five-round serial passage experiments with a series of two-round experiments where FV complex was alternated through a single host genotype.
(b) Two-round serial passage experiments
There were a total of 80 experimental infections during the test phase of two-round serial passage experiments. In the second passage round, F-MuLV fitness and virulence was significantly reduced during infection of alternate A/J, 129X1/svj and C3H/HeJ host genotypes compared with a second round of infection in the same BALB/c or DBA/2J host genotypes (electronic supplementary material, figure S8). These dramatic reductions in viral fitness and virulence were also observed in test-phase infections. Serial infection of the same host genotype resulted in significant increases in viral fitness in both experiments (BALB/c = 16-fold increase; DBA/2J = 22-fold increase) compared with unpassaged control virus (figure 3a; electronic supplementary material, figure S9). By contrast, serial infection of different host genotypes resulted in a general decrease in FV complex fitness compared with unpassaged virus (BALB/c experiment-average approximately 4-fold decrease; DBA/2J experiment-average approximately 25-fold decrease; figure 3a; electronic supplementary material, figure S9). These results were correlated with patterns of virulence. Compared to unpassaged virus, serial infection of the same host genotype led to a significant 6-fold (BALB/c experiment) and 4-fold (DBA/2J experiment) increase in splenomegaly, while serial infection of different host genotypes resulted in an average 1.3-fold (BALB/c experiment) and 2.6-fold (DBA/2J experiment) reduction in splenomegaly (figure 3b; electronic supplementary material, figure S10). Because similar effects emerged in response to alternation through 129X1/svj, A/J and C3H/HeJ genotypes, the five-round result is unlikely to be explained by some unique attribute of the 129X1/svj genotype.
(c) Host diversity impedes the pathogenic determinant of FV complex
Given the importance of the replication-defective SFFV virus in promoting acute infection (electronic supplementary material, figure S1), we reasoned that the failure of FV complex to promote splenomegaly after serial passage through different host genotypes might be due to defects in SFFV replication. This is further suggested by the fact that despite substantial overlap in the range of F-MuLV fitness among the three selection regimes (electronic supplementary material, figure S5, inset), resulting patterns of virulence were substantially different (figures 2b and 3b; electronic supplementary material, figures S5, S6 and S10). Analysis of SFFV titres in test-phase samples from our two-round serial passage experiments revealed that serial passage through different host genotypes resulted in undetectable SFFV titres (electronic supplementary material, figure S11). This was unexpected given that all three host genotypes used for alternation in two-round serial passage experiments (A/J, 129svj, and C3H/HeJ) were confirmed to be susceptible to viral replication and susceptible to SFFV-induced splenomegaly (electronic supplementary material, figure S2). It is possible that differences in the production or quality of antiviral factors derived from the previous host that are co-transferred along with virus into new hosts could confound our results. Experiments to address this possibility did not support this explanation (electronic supplementary material, figure S12a). We also tested whether co-transfer of syngeneic (self) versus allogeneic (foreign) spleen tissues impacted viral fitness or virulence (electronic supplementary material, figure S12b), the availability of host target cells in the spleen (electronic supplementary material, figure S13a) or relevant immune system parameters (electronic supplementary material, figure S13b). Tissue transplant did not affect any of these parameters.
Previous serial passage experiments have shown that pathogen adaptation leads to increased fitness and virulence in the host-of-passage, and attenuation of these traits in former host species . These results are consistent with the argument that pathogen fitness constraints arise as a consequence of adaptation to specific host species, but importantly they also imply that variable host genotypes within a species might impede pathogen adaptation and virulence evolution. Collectively, results from our experiments demonstrate that serial exposure of a viral pathogen to different host genotypes within a single species impedes the emergence of high fitness/high virulence viral phenotypes compared with the same number of infections among genetically identical hosts. Even a single round of host alternation was sufficient to produce this constraint (figure 3; electronic supplementary material, figures S9 and S10). To the best of our knowledge, this is the first time this has been demonstrated in a vertebrate host species.
The best empirical evidence in animals that genetic diversity within a host species negatively impacts pathogen adaptation and virulence evolution comes from studies using invertebrate host–pathogen systems. Serial passage of the fungal pathogen Metarhizium anisopliae through individuals of the leafcutter ant species Acromyrmex echinatior showed that increasing diversity among ant hosts was associated with reduced virulence and increased likelihood of extinction of the pathogen . Recently, using an elegant series of experiments, Morran et al.  demonstrated that outcrossing lines of the nematode Caenorhabditis elegans produced less virulent strains of the coevolving bacterial pathogen Serratia marcescens than obligately selfing C. elegans lines . These experiments support the hypothesis that diversity-promoting mechanisms like sexual reproduction may have evolved as mechanisms to counter the negative effects of rapid pathogen adaptation and virulence evolution [24,25]. This has become known as the sex-against-virulence hypothesis, and our data provide additional support for this hypothesis in vertebrates.
Results from our experiments are consistent with theoretical and empirical work demonstrating that genetic diversity in host populations can limit the spread and severity of infectious diseases [11,26–35]. Theoretical models have argued that when virulence is correlated with pathogen replication, within-host growth rates are predicted to be optimal at just below the point where host immunity can no longer control infection , or at some intermediate level determined by the average resistance among hosts in a population . This maximizes transmission rates and ultimately pathogen fitness. These models argue that host resistance against infection is a major constraint on virulence evolution. Studies of virulence evolution of rodent malaria (Plasmodium chabaudi) in response to infection of immunocompromised or immunocompetent hosts have shown that the relative abundance of virulent clones is determined by how host immunity influences competition with avirulent competitors [36–38]. Using the same pathogen model, Grech et al.  demonstrated that pathogen growth rates are strongly influenced by host genotype (i.e. they are host genotype-dependent), and an earlier study using an ancestral and C57Bl/6 mouse-adapted line of P. chabaudi provided similar conclusions . Host genotype-dependent pathogen fitness and virulence patterns have also been demonstrated previously with our Mus-FV model system [20–22]. Experiments from our laboratory [20–22] also provide evidence to support the argument that pathogen fitness trade-offs that emerge as a consequence of host-specific adaptation can influence virulence evolution. This contrasts with previous work in P. chabaudi where serial passage in C57Bl/6 mice did not result in virulence trade-offs in other mouse genotypes . Many variables could account for this discrepancy (differing passage methodologies, different pathogens, etc.), and more studies in different host–pathogen systems are needed to establish the generality of this phenomenon. However, collectively these results indicate that host resistance can influence virulence evolution by affecting within-host pathogen growth rates through direct elimination of pathogens, by modulating competitive interactions among pathogen variants or by creating fitness trade-offs that will impact the dynamics of subsequent infections.
Multiple host loci are known to influence susceptibility to FV complex infection and disease, and numerous polymorphisms have been characterized among mouse strains (electronic supplementary material, table S2) [17,18,41]. Broadly, these ‘resistance elements' can be characterized as restriction factors (retrovirally derived host-encoded genetic elements that interfere with the virus life cycle), or host genes whose products directly impact immunity against viral infection or that control haematopoiesis [17,19]. Results from our experiments indicate that serial infection of different host genotypes disrupted the erythroproliferative potential of FV complex by severely restricting SFFV replication. Consequently, infection by alternately passaged virus stocks was rendered avirulent even in highly susceptible DBA/2J and BALB/c host genotypes. The severity of this phenotype suggests that perhaps the genotypes used for alternation possessed resistance elements that abolished the ability of SFFV to replicate and be passaged to susceptible BALB/c or DBA/2J hosts in the test phase of infection. Two lines of evidence argue against this. First, mouse strains were chosen for these experiments based on the presence of susceptibility alleles at loci (e.g. FV2, FV1, FV4) that would render them permissive to FV complex infection and SFFV-induced erythroblastosis (electronic supplementary material, table S2). Second, in this (electronic supplementary material, figure S2) and previous experiments from our laboratory [20–22], we have shown that FV complex is able to both replicate and cause disease in all of the host genotypes used in this study.
The production of FV-specific neutralizing antibodies is important for anti-FV immunity and could explain our result if A/J, 129X1/svj and C3H/HeJ mice have potent neutralizing antibody responses, whereas BALB/c or DBA/2J animals do not. While addressing this possibility is beyond the scope of this study, we do know that at least the BALB/c, A/J and 129X1/svj mouse strains carry the susceptibility allele at the Rfv3 locus, which is associated with weaker neutralizing antibody responses (electronic supplementary material, table S2). Additionally, immune responses generated against non-self proteins that are co-transferred in spleen homogenates with virus might contribute to lower viral titres. Results from experiments to address this possibility did not support this argument (electronic supplementary material, figures S12 and S13) and are consistent with previous work demonstrating that allograft responses do not prime immune responses against FV complex infection . Thus, we are currently unable to explain loss of SFFV below levels of detection for our qPCR assay. Perhaps fitness trade-offs that emerge during host alternation (or another currently unidentified resistance element) drop F-MuLV or SFFV titres below a critical multiplicity of infection threshold that blocks disease progression. The striking nature of this phenotype warrants further investigation.
Serial passage experiments are valuable in addressing the role of fitness trade-offs on virulence evolution [8,43], but they have important limitations that should be acknowledged . First, relaxed costs of transmission inherent in serial passage experiments may result in levels of adaptation and virulence unachievable under natural settings . However, this is unlikely to explain differences in the trajectory of pathogen adaptation and virulence evolution in our experiments because control and alternating passages were transmitted in the same manner. Second, while murine leukaemia viruses have been shown to circulate in wild mouse populations , the rapid adaptation and virulence increases in FV complex (a spontaneously occurring laboratory isolate ) shown here and in previous studies from our laboratory [20–22] might not represent a natural pathogen response. Third, serial passage of FV complex promotes expansion of the defective SFFV component of FV complex that drives virulence. Severe mortality associated with infection by highly virulent oncogenic retroviruses in nature would probably preclude their persistence and expansion in host populations. However, notable examples of ecologically relevant retroviruses include simian and human immunodeficiency viruses, feline immunodeficiency virus and feline leukaemia virus , as well as the emerging koala retrovirus . Thus, while serial passage experiments are a useful first step for exploring the role of host genetic variation as an impediment to virulence evolution, results must be interpreted in light of the limitations to these models.
Inbred mice are powerful research tools because their inbred status removes the confounding effect of genetic variation for studies on the physiologic response to a stimulus, and because comparisons of different strains (and their relevant crosses) have been instrumental in our understanding of the functional significance of host polymorphisms. In this study, different mouse strains were used to model different genotypes present in a population of a single host species. However, two caveats must be made. First, it is difficult to know at what level variability among different mouse strains approximates genetic variability among individuals in a wild Mus population. Given that the pedigrees of most classic inbred mouse strains share a common and very recent ancestry [50,51], one could argue that our genetic diversity might be on the low end of natural populations, making our results conservative. Second, inbred mice are genetic mosaics of multiple Mus subspecies (M. m. domesticus, M. m. musculus, M. m. castaneus) [52,53], and most of these strains were selected based on the presence of specific traits. Thus, the suite of polymorphisms relevant to FV complex infection and the penetrance of these traits among strains may not represent the natural condition. Ultimately, we believe that experimental evolution studies using wild mice housed under semi-natural conditions will help establish the ecological relevance of the findings in this study.
Results from our experiments are of practical significance because they highlight the potential risks associated with low genetic diversity in host populations. Endangered species and livestock may be particularly susceptible. In the case of endangered species, enhanced disease virulence associated with low diversity among founding individuals might contribute to the high failure rate of reintroductions. Intensive livestock farming practices house related individuals under high density [54,55]. Results from serial passage studies suggest that low genetic diversity among individuals and reduced barriers to transmission may rapidly promote the emergence of more virulent infectious diseases. Intensive farming practices are of particular concern for two reasons. First, virulence evolution in this setting could jeopardize global food security as exemplified by the huge costs associated with periodic epidemics of foot-and-mouth disease virus in livestock . Second, more severe/frequent disease outbreaks might lead to increased use of antimicrobial drugs that could exacerbate the evolution of microbial resistance . Genetic management of endangered species and livestock herds (i.e. manipulating genetic diversity to augment resistance to infection) may be a useful strategy for limiting these risks.
The use of animals in all experiments was in strict adherence to federal regulations as well as the guidelines for animal use set forth by the University of Utah Institutional Animal Care and Use Committee (protocol#08–10017).
Full datasets associated with this manuscript are provided as electronic supplementary material, S2).
This work was supported by National Science Foundation Grant (DEB 0918969) and a National Institute of Health grant (R01-GM109500) to W.K.P. and F.R.A. J.L.K. was supported by a National Science Foundation Doctoral Dissertation Improvement grant (DEB 0910052) and a National Institute of Allergic and Infectious Diseases training grant (T32AI055434). J.L.K. was also supported by a National Science Foundation Educational Outreach grant (DGE 08–41233).
We would like to thank Earl A. Middlebrook and Dr James S. Ruff for their useful comments during manuscript writing. We would like to especially thank Dr Sandra Ruscetti for her valuable insights on working with FV complex. Dr Ruscetti was also instrumental in the development of tissue culture methods and techniques for assaying viral fitness by providing cell lines, antibodies and protocols. Finally, we thank Linda Morrison for her assistance in the design and optimization of assays.
- Received June 25, 2014.
- Accepted October 17, 2014.
- © 2014 The Author(s) Published by the Royal Society. All rights reserved.