On the relative roles of background selection and genetic hitchhiking in shaping human cytomegalovirus genetic diversity

Developing an appropriate evolutionary baseline model for the study of SARS-CoV-2 patient samples

Over the past 3 years, Severe Acute Respiratory Syndrome Coronavirus 2 (SARS-CoV-2) has spread through human populations in several waves, resulting in a global health crisis. In response, genomic surveillance efforts have proliferated in the hopes of tracking and anticipating the evolution of this virus, resulting in millions of patient isolates now being available in public databases. Yet, while there is a tremendous focus on identifying newly emerging adaptive viral variants, this quantification is far from trivial. Specifically, multiple co-occurring and interacting evolutionary processes are constantly in operation and must be jointly considered and modeled in order to perform accurate inference. We here outline critical individual components of such an evolutionary baseline model-mutation rates, recombination rates, the distribution of fitness effects, infection dynamics, and compartmentalization-and describe the current state of knowledge pertaining to the related parameters of each in SARS-CoV-2. We close with a series of recommendations for future clinical sampling, model construction, and statistical analysis.

Human herpesvirus diversity is altered in HLA class I binding peptides

Herpesviruses are ubiquitous, genetically diverse DNA viruses, with long-term presence in humans associated with infrequent but significant pathology. Human leukocyte antigen (HLA) class I presents intracellularly derived peptide fragments from infected tissue cells to CD8+ T and natural killer cells, thereby directing antiviral immunity. Allotypes of highly polymorphic HLA class I are distinguished by their peptide binding repertoires. Because this HLA class I variation is a major determinant of herpesvirus disease, we examined if sequence diversity of virus proteins reflects evasion of HLA presentation. Using population genomic data from Epstein–Barr virus (EBV), human cytomegalovirus (HCMV), and Varicella–Zoster virus, we tested whether diversity differed between the regions of herpesvirus proteins that can be recognized, or not, by HLA class I. Herpesviruses exhibit lytic and latent infection stages, with the latter better enabling immune evasion. Whereas HLA binding peptides of lytic proteins are conserved, we found that EBV and HCMV proteins expressed during latency have increased peptide sequence diversity. Similarly, latent, but not lytic, herpesvirus proteins have greater population structure in HLA binding than nonbinding peptides. Finally, we found patterns consistent with EBV adaption to the local HLA environment, with less efficient recognition of EBV isolates by high-frequency HLA class I allotypes. Here, the frequency of CD8+ T cell epitopes inversely correlated with the frequency of HLA class I recognition. Previous analyses have shown that pathogen-mediated natural selection maintains exceptional polymorphism in HLA residues that determine peptide recognition. Here, we show that HLA class I peptide recognition impacts diversity of globally widespread pathogens.

Inferring the distribution of fitness effects in patient-sampled and experimental virus populations: two case studies

We here propose an analysis pipeline for inferring the distribution of fitness effects (DFE) from either patient-sampled or experimentally-evolved viral populations, that explicitly accounts for non-Wright-Fisher and non-equilibrium population dynamics inherent to pathogens. We examine the performance of this approach via extensive power and performance analyses, and highlight two illustrative applications - one from an experimentally-passaged RNA virus, and the other from a clinically-sampled DNA virus. Finally, we discuss how such DFE inference may shed light on major research questions in virus evolution, ranging from a quantification of the population genetic processes governing genome size, to the role of Hill-Robertson interference in dictating adaptive outcomes, to the potential design of novel therapeutic approaches to eradicate within-patient viral populations via induced mutational meltdown.

Quantitative measures of within-host viral genetic diversity

The genetic diversity of virus populations within their hosts is known to influence disease progression, treatment outcome, drug resistance, cell tropism, and transmission risk, and the study of dynamic changes of genetic heterogeneity can provide insights into the evolution of viruses. Several measures to quantify within-host genetic diversity capturing different aspects of diversity patterns in a sample or population are used, based on incidence, relative frequencies, pairwise distances, or phylogenetic trees. Here, we review and compare several of these measures.

Common Polymorphisms in the Glycoproteins of Human Cytomegalovirus and Associated Strain-Specific Immunity

Human cytomegalovirus (HCMV), one of the most prevalent viruses across the globe, is a common cause of morbidity and mortality for immunocompromised individuals. Recent clinical observations have demonstrated that mixed strain infections are common and may lead to more severe disease progression. This clinical observation illustrates the complexity of the HCMV genome and emphasizes the importance of taking a population-level view of genotypic evolution. Here we review frequently sampled polymorphisms in the glycoproteins of HCMV, comparing the variable regions, and summarizing their corresponding geographic distributions observed to date. The related strain-specific immunity, including neutralization activity and antigen-specific cellular immunity, is also discussed. Given that these glycoproteins are common targets for vaccine design and anti-viral therapies, this observed genetic variation represents an important resource for future efforts to combat HCMV infections.

In vitro evolution of herpes simplex virus 1 (HSV-1) reveals selection for syncytia and other minor variants in cell culture

The large dsDNA virus herpes simplex virus 1 (HSV-1) is considered to be genetically stable, yet it can rapidly evolve in response to strong selective pressures such as antiviral treatment. Deep sequencing has revealed that clinical and laboratory isolates of this virus exist as populations that contain a mixture of minor alleles or variants, similar to many RNA viruses. The classic virology approach of plaque purifying virus creates a genetically homogenous population, but it is not clear how closely this represents the mixed virus populations found in nature. We sought to study the evolution of mixed versus highly purified HSV-1 populations in controlled cell culture conditions, to examine the impact of this genetic diversity on evolution. We found that a mixed population of HSV-1 acquired more genetic diversity and underwent a more dramatic phenotypic shift than a plaque-purified population, producing a viral population that was almost entirely syncytial after just ten passages. At the genomic level, adaptation and genetic diversification occurred at the level of minor alleles or variants in the viral population. Certain genetic variants in the mixed viral population appeared to be positively selected in cell culture, and this shift was also observed in clinical samples during their first passages in vitro. In contrast, the plaque-purified viral population did not appear to change substantially in phenotype or overall quantity of minor allele diversity. These data indicate that HSV-1 is capable of evolving rapidly in a given environment, and that this evolution is facilitated by diversity in the viral population.

Association of CMV genomic mutations with symptomatic infection and hearing loss in congenital CMV infection

Background

Congenital cytomegalovirus (cCMV) infection is the most common congenital infection and a leading cause of long-term neurological and sensory sequelae, the most common being sensorineural hearing loss (SNHL). Despite extensive research, clinical or laboratory markers to identify CMV infected children with increased risk for disease have not been identified. This study utilizes viral whole-genome next generation-sequencing (NGS) of specimens from congenitally infected infants to explore viral diversity and specific viral variants that may be associated with symptomatic infection and SNHL.

Methods

CMV DNA from urine specimens of 30 infants (17 asymptomatic, 13 symptomatic) was target enriched and next generation sequenced resulting in 93% coverage of the CMV genome allowing analysis of viral diversity.

Results

Variant frequency distribution was compared between children with symptomatic and asymptomatic cCMV and those with (n = 13) and without (n = 17) hearing loss. The CMV genes UL48A, UL88, US19 and US22 were found to have an increase in nucleotide diversity in symptomatic children; while UL57, UL20, UL104, US14, UL115, and UL35 had an increase in diversity in children with hearing loss. An analysis of single variant differences between symptomatic and asymptomatic children found UL55 to have the highest number, while the most variants associated with SNHL were in the RL11 gene family. In asymptomatic infants with SNHL, mutations were observed more frequently in UL33 and UL20.

Conclusion

CMV genomes from infected newborns can be mapped to 93% of the genome at a depth allowing accurate and reproducible analysis of polymorphisms for variant and gene discovery that may be linked to symptomatic and hearing loss outcomes.

HCMV Envelope Glycoprotein Diversity Demystified

Human cytomegalovirus (HCMV) is the leading viral cause of congenital birth defects and is responsible for morbidity and mortality in immunosuppressed individuals. Considerable efforts have been deployed over the last decade to develop a vaccine capable of preventing HCMV infection. However, in recent clinical trials, vaccines showed at best modest efficacy in preventing infection. These findings might be explained by the high level of sequence polymorphism at the genomic level. To investigate if genomic variation also leads to antigenic variation, we performed a bioinformatic sequence analysis and evaluated the percentage of conservation at the amino acid level of all the proteins present in the virion envelope. Using more than two hundred sequences per envelope glycoprotein and analyzing their degree of conservation, we observe that antigenic variation is in large part limited to three proteins. In addition, we demonstrate that the two leading vaccine candidates, the pentamer and gB complexes, are well conserved at the amino acid level. These results suggest that despite genomic polymorphism, antigenic variability is not involved in the modest efficacy observed in the recent clinical trials for a HCMV vaccine. We therefore propose that next-generation vaccines should focus on stabilizing and refining the gB domains needed to induce a protective humoral response.

Ecological and Evolutionary Processes Shaping Viral Genetic Diversity

The contemporary genomic diversity of viruses is a result of the continuous and dynamic interaction of past ecological and evolutionary processes. Thus, genome sequences of viruses can be a valuable source of information about these processes. In this review, we first describe the relevant processes shaping viral genomic variation, with a focus on the role of host–virus coevolution and its potential to give rise to eco-evolutionary feedback loops. We further give a brief overview of available methodology designed to extract information about these processes from genomic data. Short generation times and small genomes make viruses ideal model systems to study the joint effect of complex coevolutionary and eco-evolutionary interactions on genetic evolution. This complexity, together with the diverse array of lifetime and reproductive strategies in viruses ask for extensions of existing inference methods, for example by integrating multiple information sources. Such integration can broaden the applicability of genetic inference methods and thus further improve our understanding of the role viruses play in biological communities.

Genotypic and Phenotypic Diversity of Herpes Simplex Virus 2 within the Infected Neonatal Population

More than 14,000 neonates are infected with herpes simplex virus (HSV) annually. Approximately half display manifestations limited to the skin, eyes, or mouth (SEM disease). The rest develop invasive infections that spread to the central nervous system (CNS disease or encephalitis) or throughout the infected neonate (disseminated disease). Invasive HSV disease is associated with significant morbidity and mortality, but the viral and host factors that predispose neonates to these forms are unknown. To define viral diversity within the infected neonatal population, we evaluated 10 HSV-2 isolates from newborns with a range of clinical presentations. To assess viral fitness independently of host immune factors, we measured viral growth characteristics in cultured cells and found diverse in vitro phenotypes. Isolates from neonates with CNS disease were associated with larger plaque size and enhanced spread, with the isolates from cerebrospinal fluid (CSF) exhibiting the most robust growth. We sequenced complete viral genomes of all 10 neonatal viruses, providing new insights into HSV-2 genomic diversity in this clinical setting. We found extensive interhost and intrahost genomic diversity throughout the viral genome, including amino acid differences in more than 90% of the viral proteome. The genes encoding glycoprotein G (gG; US4), glycoprotein I (gI; US7), and glycoprotein K (gK; UL53) and viral proteins UL8, UL20, UL24, and US2 contained variants that were found in association with CNS isolates. Many of these viral proteins are known to contribute to cell spread and neurovirulence in mouse models of CNS disease. This report represents the first application of comparative pathogen genomics to neonatal HSV disease.IMPORTANCE Herpes simplex virus (HSV) causes invasive disease in half of infected neonates, resulting in significant mortality and permanent cognitive morbidity. The factors that contribute to invasive disease are not understood. This study revealed diversity among HSV isolates from infected neonates and detected the first associations between viral genetic variations and clinical disease manifestations. We found that viruses isolated from newborns with encephalitis showed enhanced spread in culture. These viruses contained protein-coding variations not found in viruses causing noninvasive disease. Many of these variations were found in proteins known to impact neurovirulence and viral spread between cells. This work advances our understanding of HSV diversity in the neonatal population and how it may impact disease outcome.

The Human Cytomegalovirus Strain DB Activates Oncogenic Pathways in Mammary Epithelial Cells

Background

Human cytomegalovirus (HCMV) establishes a persistent life-long infection and increasing evidence indicates HCMV infection can modulate signaling pathways associated with oncogenesis. Breast milk is an important route of HCMV transmission in humans and we hypothesized that mammary epithelial cells could be one of the main cellular targets of HCMV infection.

Methods

The infectivity of primary human mammary epithelial cells (HMECs) was assessed following infection with the HCMV-DB strain, a clinical isolate with a marked macrophage-tropism. The impact of HCMV-DB infection on expression of p53 and retinoblastoma proteins, telomerase activity and oncogenic pathways (c-Myc, Akt, Ras, STAT3) was studied. Finally the transformation of HCMV-DB infected HMECs was evaluated using soft agar assay. CTH cells (CMV Transformed HMECs) were detected in prolonged cultures of infected HMECs. Tumor formation was observed in NOD/SCID Gamma (NSG) mice injected with CTH cells. Detection of long non coding RNA4.9 (lncRNA4.9) gene was assessed in CTH cells, tumors isolated from xenografted NSG mice and biopsies of patients with breast cancer using qualitative and quantitative PCR.

Results

We found that HCMV, especially a clinical strain named HCMV-DB, infects HMECs in vitro. The clinical strain HCMV-DB replicates productively in HMECs as evidenced by detection of early and late viral transcripts and proteins. Following infection of HMECs with HCMV-DB, we observed the inactivation of retinoblastoma and p53 proteins, the activation of telomerase activity, the activation of the proto-oncogenes c-Myc and Ras, the activation of Akt and STAT3, and the upregulation of cyclin D1 and Ki67 antigen. Colony formation was observed in soft agar seeded with HCMV-DB-infected HMECs. Prolonged culture of infected HMECs resulted in the development of clusters of spheroid cells that we called CTH cells (CMV Transformed HMECs). CTH cells when injected in NOD/SCID Gamma (NSG) mice resulted in the development of tumors. We detected in CTH cells the presence of a HCMV signature corresponding to a sequence of the long noncoding RNA4.9 (lncRNA4.9) gene. We also found the presence of the HCMV lncRNA4.9 sequence in tumors isolated from xenografted NSG mice injected with CTH cells and in biopsies of patients with breast cancer using qualitative and quantitative PCR.

Conclusions

Our data indicate that key molecular pathways involved in oncogenesis are activated in HCMV-DB-infected HMECs that ultimately results in the transformation of HMECs in vitro with the appearance of CMV-transformed HMECs (CTH cells) in culture. CTH cells display a HCMV signature corresponding to a lncRNA4.9 genomic sequence and give rise to fast growing triple-negative tumors in NSG mice. A similar lncRNA4.9 genomic sequence was detected in tumor biopsies of patients with breast cancer.

•

The infection of primary human mammary epithelial cells (HMECs) with the HCMV-DB strain results in a pro-oncogenic cellular environment.

•

HCMV-DB transforms primary HMECs in vitro as measured by a soft agar assay.

•

Prolonged culture of HMECs infected with HCMV-DB results in the appearance of clusters of spheroid cells that we called CTH cells (CMV Transformed HMECs).

•

CTH cells when injected in NOD/SCID Gamma mice resulted in the development of breast tumor.

•

The HCMV lncRNA4.9 sequence was detected in CTH cells, in tumors isolated from xenografted NSG mice injected with CTH cells and in biopsies of patients with breast cancer.

Research in Context: Worldwide breast cancer is the most common cancer diagnosed among women. Etiological factors involved in breast cancer include genetic and environmental risk factors and among these latter viruses could be involved with close to one-fifth of all cancers in the world caused by infectious agents. We found that the cytomegalovirus strain DB, a member of the herpesvirus family, activates oncogenic pathways in infected mammary epithelial cells, transforms these cells in culture and favors the appearance of tumors in xenografted mice. Our findings might lead to a better understanding of the pathogenesis of breast cancer.

On the Demographic and Selective Forces Shaping Patterns of Human Cytomegalovirus Variation within Hosts

Human cytomegalovirus (HCMV) is a member of the β -herpesvirus subfamily within Herpesviridae that is nearly ubiquitous in human populations, and infection generally results only in mild symptoms. However, symptoms can be severe in immunonaive individuals, and transplacental congenital infection of HCMV can result in serious neurological sequelae. Recent work has revealed much about the demographic and selective forces shaping the evolution of congenitally transmitted HCMV both on the level of hosts and within host compartments, providing insight into the dynamics of congenital infection, reinfection, and evolution of HCMV with important implications for the development of effective treatments and vaccines.

Impacts of Genome-Wide Analyses on Our Understanding of Human Herpesvirus Diversity and Evolution

Until fairly recently, genome-wide evolutionary dynamics and within-host diversity were more commonly examined in the context of small viruses than in the context of large double-stranded DNA viruses such as herpesviruses. The high mutation rates and more compact genomes of RNA viruses have inspired the investigation of population dynamics for these species, and recent data now suggest that herpesviruses might also be considered candidates for population modeling. High-throughput sequencing (HTS) and bioinformatics have expanded our understanding of herpesviruses through genome-wide comparisons of sequence diversity, recombination, allele frequency, and selective pressures. Here we discuss recent data on the mechanisms that generate herpesvirus genomic diversity and underlie the evolution of these virus families. We focus on human herpesviruses, with key insights drawn from veterinary herpesviruses and other large DNA virus families. We consider the impacts of cell culture on herpesvirus genomes and how to accurately describe the viral populations under study. The need for a strong foundation of high-quality genomes is also discussed, since it underlies all secondary genomic analyses such as RNA sequencing (RNA-Seq), chromatin immunoprecipitation, and ribosome profiling. Areas where we foresee future progress, such as the linking of viral genetic differences to phenotypic or clinical outcomes, are highlighted as well.

Refining the Use of Linkage Disequilibrium as a Robust Signature of Selective Sweeps

During a selective sweep, characteristic patterns of linkage disequilibrium can arise in the genomic region surrounding a selected locus. These have been used to infer past selective sweeps. However, the recombination rate is known to vary substantially along the genome for many species. We here investigate the effectiveness of current (Kelly's [Formula: see text] and [Formula: see text]) and novel statistics at inferring hard selective sweeps based on linkage disequilibrium distortions under different conditions, including a human-realistic demographic model and recombination rate variation. When the recombination rate is constant, Kelly's [Formula: see text] offers high power, but is outperformed by a novel statistic that we test, which we call [Formula: see text] We also find this statistic to be effective at detecting sweeps from standing variation. When recombination rate fluctuations are included, there is a considerable reduction in power for all linkage disequilibrium-based statistics. However, this can largely be reversed by appropriately controlling for expected linkage disequilibrium using a genetic map. To further test these different methods, we perform selection scans on well-characterized HapMap data, finding that all three statistics-[Formula: see text] Kelly's [Formula: see text] and [Formula: see text]-are able to replicate signals at regions previously identified as selection candidates based on population differentiation or the site frequency spectrum. While [Formula: see text] replicates most candidates when recombination map data are not available, the [Formula: see text] and [Formula: see text] statistics are more successful when recombination rate variation is controlled for. Given both this and their higher power in simulations of selective sweeps, these statistics are preferred when information on local recombination rate variation is available.

Characterizing human cytomegalovirus reinfection in congenitally infected infants: an evolutionary perspective

Given the strong selective pressures often faced by populations when colonizing a novel habitat, the level of variation present on which selection may act is an important indicator of adaptive potential. While often discussed in an ecological context, this notion is also highly relevant in our clinical understanding of viral infection, in which the novel habitat is a new host. Thus, quantifying the factors determining levels of variation is of considerable importance for the design of improved treatment strategies. Here, we focus on such a quantification of human cytomegalovirus (HCMV) - a virus which can be transmitted across the placenta, resulting in foetal infection that can potentially cause severe disease in multiple organs. Recent studies using genomewide sequencing data have demonstrated that viral populations in some congenitally infected infants diverge rapidly over time and between tissue compartments within individuals, while in other infants, the populations remain highly stable. Here, we investigate the underlying causes of these extreme differences in observed intrahost levels of variation by estimating the underlying demographic histories of infection. Importantly, reinfection (i.e. population admixture) appears to be an important, and previously unappreciated, player. We highlight illustrative examples likely to represent a single-population transmission from a mother during pregnancy and multiple-population transmissions during pregnancy and after birth.

On the Analysis of Intrahost and Interhost Viral Populations: Human Cytomegalovirus as a Case Study of Pitfalls and Expectations

Intrahost and interhost assessments of viral diversity are often treated as measures of separate and distinct evolutionary processes, with numerous investigations reporting seemingly incompatible results between the two. For example, in human cytomegalovirus, the nucleotide diversity estimates are 10-fold higher for interhost data, while the number of segregating (i.e., polymorphic) sites is 6-fold lower. These results have been interpreted as demonstrating that sampled intrahost variants are strongly deleterious. In reality, however, these observations are fully consistent with standard population genetic expectations. Here, we analyze published intra- and interhost data sets within this framework, utilizing statistical inference tools to quantify the fitness effects of segregating mutations. Further, we utilize population level simulations to clarify expectations under common evolutionary models. Contrary to common claims in the literature, these results suggest that most observed polymorphisms are likely nearly neutral with regard to fitness and that standard population genetic models in fact well predict observed levels of both intra- and interhost variability. IMPORTANCE With the increasing number of evolutionary virology studies examining both intrahost and interhost patterns of genomic variation, a number of seemingly incompatible results have emerged, revolving around the far greater level of observed intrahost than interhost variation. This has led many authors to suggest that the great majority of sampled within-host polymorphisms are strongly deleterious. Here, we demonstrate that there is in fact no incompatibility of these results and, indeed, that the vast majority of sampled within-host variation is likely neutral. These results thus represent a major shift in the current view of observed viral variation.

The Impact of Linked Selection in Chimpanzees: A Comparative Study

Levels of nucleotide diversity vary greatly across the genomes of most species owing to multiple factors. These include variation in the underlying mutation rates, as well as the effects of both direct and linked selection. Fundamental to interpreting the relative importance of these forces is the common observation of a strong positive correlation between nucleotide diversity and recombination rate. While indeed observed in humans, the interpretation of this pattern has been difficult in the absence of high-quality polymorphism data and recombination maps in closely related species. Here, we characterize genetic features driving nucleotide diversity in Western chimpanzees using a recently generated whole genome polymorphism data set. Our results suggest that recombination rate is the primary predictor of nucleotide variation with a strongly positive correlation. In addition, telomeric distance, regional GC-content, and regional CpG-island content are strongly negatively correlated with variation. These results are compared with humans, with both similarities and differences interpreted in the light of the estimated effective population sizes of the two species as well as their strongly differing recent demographic histories.

Finding the Genomic Basis of Local Adaptation: Pitfalls, Practical Solutions, and Future Directions

Uncovering the genetic and evolutionary basis of local adaptation is a major focus of evolutionary biology. The recent development of cost-effective methods for obtaining high-quality genome-scale data makes it possible to identify some of the loci responsible for adaptive differences among populations. Two basic approaches for identifying putatively locally adaptive loci have been developed and are broadly used: one that identifies loci with unusually high genetic differentiation among populations (differentiation outlier methods) and one that searches for correlations between local population allele frequencies and local environments (genetic-environment association methods). Here, we review the promises and challenges of these genome scan methods, including correcting for the confounding influence of a species' demographic history, biases caused by missing aspects of the genome, matching scales of environmental data with population structure, and other statistical considerations. In each case, we make suggestions for best practices for maximizing the accuracy and efficiency of genome scans to detect the underlying genetic basis of local adaptation. With attention to their current limitations, genome scan methods can be an important tool in finding the genetic basis of adaptive evolutionary change.

On the importance of skewed offspring distributions and background selection in virus population genetics

Many features of virus populations make them excellent candidates for population genetic study, including a very high rate of mutation, high levels of nucleotide diversity, exceptionally large census population sizes, and frequent positive selection. However, these attributes also mean that special care must be taken in population genetic inference. For example, highly skewed offspring distributions, frequent and severe population bottleneck events associated with infection and compartmentalization, and strong purifying selection all affect the distribution of genetic variation but are often not taken into account. Here, we draw particular attention to multiple-merger coalescent events and background selection, discuss potential misinference associated with these processes, and highlight potential avenues for better incorporating them into future population genetic analyses.

The population genetics of drug resistance evolution in natural populations of viral, bacterial and eukaryotic pathogens

Drug resistance is a costly consequence of pathogen evolution and a major concern in public health. In this review, we show how population genetics can be used to study the evolution of drug resistance and also how drug resistance evolution is informative as an evolutionary model system. We highlight five examples from diverse organisms with particular focus on: (i) identifying drug resistance loci in the malaria parasite Plasmodium falciparum using the genomic signatures of selective sweeps, (ii) determining the role of epistasis in drug resistance evolution in influenza, (iii) quantifying the role of standing genetic variation in the evolution of drug resistance in HIV, (iv) using drug resistance mutations to study clonal interference dynamics in tuberculosis and (v) analysing the population structure of the core and accessory genome of Staphylococcus aureus to understand the spread of methicillin resistance. Throughout this review, we discuss the uses of sequence data and population genetic theory in studying the evolution of drug resistance.