Evolutionary changes are not always required for viruses to emerge in new hosts. For example, canine distemper virus has a very wide host range in mammals, naturally infecting marine mammals, lions, black-footed ferrets, and other hosts, and its emergence in these species appears to be limited primarily by contact. However, in other cases emergence requires the evolution of the virus to allow efficient infection and transmission within the new host. The evolution of viruses to allow adaptation to new hosts is still not well understood. The level of genetic variation is important, and most viruses transferred to new hosts are poorly adapted, replicate poorly, and are inefficiently transmitted, so that the greater the rate of variation the more likely a virus is to adapt to the new host. This indicates that cross-species transmission should be more common in rapidly evolving viruses (
12,
24,
41,
147,
149). RNA viruses have error-prone replication (
23), lack a proofreading mechanism, and have rapid replication, short virus generation times, and large virus populations (
22,
82). In contrast, most DNA viruses are less variable and more often associated with virus-host cospeciation (
42,
105). However, the distinctions between RNA and DNA viruses in rates of evolutionary change are not straightforward: some retroviruses (e.g., the simian foamy viruses) show temporal rates of nucleotide substitution far lower (~10
−8 substitutions/site/year) than those seen for other RNA viruses (
119). There is also strong evidence that some RNA viruses have coevolved with specific hosts over long periods (including hantaviruses and arenaviruses), developing a high degree of host specialization (
9,
19,
56,
76,
111). The rates of variation of some DNA viruses may also be underestimated. In particular, the single-stranded DNA viruses (in animals, the
Parvoviridae and
Circoviridae) are more diverse than are other DNA viruses and may evolve at rates similar to those of many RNA viruses (
93,
103,
106,
107,
126).
Viral Fitness Trade-Offs
A fundamental challenge for host-switching viruses that require adaptation to their new hosts is that mutations that optimize the ability of a virus to infect a new host will likely reduce its fitness in the donor host (Fig. and ). The nature of these fitness trade-offs and how they affect cross-species transmission is an important unresolved area of study. Interactions between virus and hosts determine the fitness landscape for the virus, and after a host-switching event combinations of genetic drift and selection will determine the viral genetic variation that remains in the long term. However, only a small proportion of the viral mutational spectrum will exhibit increased fitness, particularly after passing through the population bottlenecks that accompany host switching (
15,
24,
81,
101). The advantageous and deleterious mutations often show complex epistatic interactions that likely have major effects on the rate and progress of adaptation. As one example, in the case of vesicular stomatitis virus, regaining full fitness after host transfer is a complex process involving multiple compensatory changes (
100).
Mode of Virus Transmission
An important constraint influencing emergence and successful host transfer is the mode of virus transmission. For example, arthropod vectors that feed on a range of mammalian hosts can facilitate cross-species viral exposures. However, both phylogenetic and in vitro studies of arboviruses indicate that their levels of variation are relatively constrained compared to what is observed for viruses transmitted by other mechanisms (
62,
128,
136,
154). Those viruses would need to balance the fitness in at least three hosts during the process of adaptation, i.e., the donor and recipient hosts and the vector(s), presenting a difficult challenge to new emergences. Adaptation to interhost transmission by droplet spread, that by sexual inoculation, and that by fecal-oral transmission each represent different adaptational challenges due to host differences and variation in environmental exposure. However little is known about how shedding and infection are controlled in different hosts. For example, it is not clear why influenza A viruses are enteric viruses in their natural avian hosts but mainly infect the respiratory tract in mammals, but this likely influences the host adaptation of the viruses to mammals and the ability to spread efficiently.
Recombination and Reassortment in Viral Evolution Leading to Host Switching
For many viruses, recombination (and its variation seen for viruses with segmented genomes, reassortment) allow the acquisition of multiple genetic changes in a single step and can combine genetic information to produce advantageous genotypes or remove deleterious mutations. Examples of reassortment in disease emergence include the emergence of the 1957 H2N2 and 1968 H3N2 influenza A pandemic viruses, where new avian genome segments were imported into the backbone of 1918-descended H1N1 viruses (
137), as well as the 2003 emergence of the pathogenic Fujian H3N2 influenza strain by interclade reassortment (
43).
The potential for recombination varies among different RNA and DNA viruses. Aside from segmental reassortment, recombination is rare among negative-stranded RNA viruses, while retroviruses such as HIV have high rates of recombination (
20,
52,
108). Recombination between viruses from different primate hosts was associated with human HIV emergence; the possible donor host origins, recombination events, and intermediate host transfers are depicted in Fig. (
55,
67,
102). The SARS CoV appears to have arisen from a recombinant between a bat CoV and another virus (most likely also a bat virus) before infecting humans and carnivore hosts (Fig. ). As described above, part of the receptor binding sequence of this virus may have been acquired by recombination with a group 1 human CoV, which was then selected for more-efficient use of the human ACE2 receptor (Fig. ) (
71).
Many recombinations or reassortments are likely to be deleterious in that they disrupt optimal protein structures or functional gene combinations. For example, the replication proteins of influenza A virus (PA, PB1, and PB2) work as a complex, and altering the combinations through reassortment of genomic segments can reduce replication efficiency and require subsequent adaptation to the combinations of proteins from different sources (
13,
39,
123) (Table ). The HA and NA proteins of influenza A viruses both act on the cell's sialic acid receptors, and complementarity between virus binding (HA) and cleavage (NA) activities is often required for optimal binding to and release from cells expressing different glycan receptors (
109,
118,
132).
| TABLE 4.Amino acid residues that distinguish human and avian influenza virus polymerases identified by comparison of the genome of the human 1918 virus strain with those of other human, avian, swine, and equine virusesa |
Recombination and reassortment may also be important for incremental host adaptation after switching to the new host has occurred. For example, after the 1968 emergence of human H3N2 influenza virus, which contained HA and PB1 gene segments imported from avian viruses, extensive secondary reassortments occurred after transfer, which may have facilitated its further adaptation (
73).
Are Viral Intermediates with Lower Fitness Involved in Host Switching?
The process of virus transfer to a new host is rarely observed directly but can be inferred by comparing viral ancestors in donor hosts with emergent viruses from recipient hosts. If several changes are required to allow host switching, then intermediate viruses would likely be less fit in either the donor or recipient hosts than the parental or descendant viruses (
60) (Fig. ). As mentioned previously, influenza A reassortant viruses carrying single genomic segments from viruses of alternative hosts showed replication in either of those hosts that was lower than that seen for the parental viruses in their original hosts. The adaptation of FPV to dogs also occurred through at least one lower-fitness intermediate, as the first viruses collected from dogs were both less fit in cats than the FPV from which they were derived and less well adapted in dogs than the CPV variants that replaced them (
107,
127).
Crossing any evolutionary “low-fitness valley” for partially adapted viruses can therefore be a key step for virus host switching and may explain the rarity of such transfers: partially adapted viruses would quickly go extinct, as they would be unfit in the donor host and also insufficiently adapted to allow efficient replication and spread in the recipient host (Fig. ). If the transmission rate in the new host population allows virus maintenance, then the length of the period of lower replication and spread would be a function of the number of genetic changes required to gain high transmissibility. In the new host, the virus may not be competing with similar viruses, and if it spreads with an efficiency with a reproductive number (R0) of >1, it could increase its fitness by mutation and selection to propagate epidemically.
Early detection of inefficiently spreading viruses in a new host would provide opportunities for epidemic control. In the SARS CoV outbreak, the first virus that emerged was only inefficiently transmitted by most infected people, and early recognition of the outbreak and institution of active control measures (particularly quarantine) allowed the epidemic to be stopped before the virus could become fully established in humans (
4,
110,
156) (Fig. ). How viruses gain the ability to spread efficiently (so that the
R0 is
![[dbl greater-than sign]](/corehtml/pmc/pmcents/x226B.gif)
1) is a key question in viral emergence, but the mechanisms involved are poorly understood (
68,
124). In addition to optimizing replicative efficiency in cells and tissues, a new virus may have to optimize the intensity of viral shedding from appropriate sites for transmission (e.g., mucosa, respiratory tract, skin, feces, urine, blood, and other tissues), may have to induce sneezing to achieve respiratory shedding, or, for arthropod-transmitted viruses, may have to establish high levels of viremia or replication in vectors (
35,
60,
136). As described above, this process likely requires adaptation to allow passage through host-specific passive barriers at the mucosal surfaces and to avoid early elimination by innate immune responses (
104,
138).
During the early stages of an outbreak, infected individuals who cause a large number of new infections may play a critical amplifying role. Such “superspreading” individuals were documented during the SARS CoV epidemic and during outbreaks of measles and other aerosolized viruses (
75,
89,
135). The determinants of “superspreading” are still poorly understood but may be related to higher levels of virus shedding in some individuals, to host behaviors, and to prolonged times of uncontrolled exposure to susceptible contacts early in the outbreak, before the need for infection control is appreciated (
11,
113). Animal-to-animal or person-to-person transmission has been a difficult subject to investigate experimentally, and we know relatively little about the specific factors that control it for most viruses, particularly during transfers into new hosts. Detailed pathogenesis studies in experimental animals will be required to achieve a better understanding of these factors.