It is well established that BVDV isolates show considerable variation in nucleotide sequence, even within subgenotypes. However, it has not been determined how and when this genetic variation arises. It was proposed that variability in nucleotide sequence arises primarily during acute infections with immune pressure being the major driving force behind changes in the immunogenic structural proteins . Changes in the nonstructural proteins come about more slowly and are due primarily to the error-prone RNA-dependent RNA polymerase.
Previous studies of genetic change in BVDV using sequencing of portions of the 5’ UTR and E2 protein of the BVDV genomic RNA revealed that nucleotide substitutions were introduced slowly [5, 16]. The choice of the E2 protein in these earlier studies, based on data presented here, was an appropriate choice for analysis of genetic changes because of the greater number of changes. However, the small size of the genomic region analyzed provided only limited analysis. In our earlier study , we found that sequencing of the entire ORF provided a more complete overview. Furthermore, this showed that more nucleotide substitutions were found in the BVDV genomic RNA during a single ‘in vivo passage’ in the establishment of a persistent infection in a calf in utero than in a series of acute infections caused by a single strain of BVDV over a large geographic region and greater than a year’s time. These data indicated that infection of pregnant animals may be the greater source of genetic variation. Unfortunately, this analysis did not answer the question of when the changes were introduced in pregnant cattle; specifically, whether they occurred during the acute infection of the dam or in crossing the placenta and infection of the fetus. To address this, the ORF of viruses isolated from the acute infections of the dams from this previous study were sequenced and compared to the sequences of the progenitor and progeny PI viruses to determine when and to what extent that variation in BVDV genomic sequences occurred. This clearly demonstrated that the majority of the genetic changes were introduced during the acute infection of the dam. Within six days of exposure to the progenitor PI calf, the virus isolated from serum possessed the majority of the changes identified in the progeny PI virus. The remaining two to three nucleotide changes found in the progeny PI viruses were introduced during the infection of the fetus. Importantly, the acute phase virus was altered before the infection of the fetus. Also, based on the number of changes in the genome of the progeny PI from that of the acute phase virus from the dam, it appears that the fetus can be considered a non-pregnant host, thus exhibiting low numbers of additional nucleotide substitutions. One particularly surprising finding in this study was the number of nucleotide changes that were introduced in the progeny PI virus infecting the calf of the PI dam. It was expected that the virus of the dam would infect the fetus with little resulting genetic change because of lack of adaptive immunity in the dam or fetus, and the large amount of virus that is continuously present. It is not clear at this point why this occurred.
The data presented here revealed that nucleotide substitutions occurred nearly randomly throughout the genomic RNA of BVDV with nonsynonymous changes being more limited in the regions of the genome where they may occur (Figure2). The occurrence of nonsynonymous changes was biased to specific genomic regions, primarily the structural proteins, where amino acid changes are more easily tolerated. There appeared to be specific locations where change was not tolerated, most likely to conserve critical functions of the proteins. Examination of the types of nucleotide substitutions showed that there is not a great difference in the numbers of the four different types of transitions (Table3). Transitions outnumbered transversions by more than 6:1, implicating misincorporation of nucleotides by the RNA-dependent RNA polymerase [17, 18]. When the total numbers of nucleotide substitutions were compared between acute viruses isolated from pregnant animals and acute viruses from non-pregnant animals, there was a 3.3 fold increase in nucleotide substitutions in pregnant animals over non-pregnant animals with a range of 2 to 6 fold. In the non-pregnant animals, the BVDV isolates had a mutation rate of 6 x 10-4 substitutions per nucleotide, while substitution rates in the pregnancy-associated viruses were 1.9 x 10-3 substitutions per nucleotide, or an average of 3.3 fold higher. This clearly shows that there was some mechanism present in pregnant animals that brought about or allowed greater genetic variability to be introduced into the BVDV genome.
There may be two possible reasons for the observation of a greater number of nucleotide changes in the genomic RNAs of BVDV in pregnant animals. First, there may be a protective mechanism that is more active in pregnant cattle that introduces mutations into the BVDV RNA genome in an attempt to stop or limit the infection. This type of mechanism that may resemble the APOBEC cytosine deaminases, may enzymatically alter nucleotides to introduce lethal genetic changes in a manner similar to those introduced into the genomes of lentiviruses and hepatitis B virus [19–24]. Analysis of the form of nucleotide changes found in BVDV strains (Table3) indicated that this type of enzymatic mutagenesis was unlikely to be the source of genetic change in these viruses. There was not a preponderance of changes that would result from cytosine to uridine changes. In fact, most changes were transitions that resembled the misincorporation of nucleotides by the RNA-dependent RNA polymerase where guanosine may bond with uridine, similarly to that reported in poliovirus, foot and mouth disease virus, and cucumber mosaic virus [17, 18, 25].
The second and the more probable mechanism lies in the BVDV quasispecies of the infecting progenitor PI calf and the genetic bottleneck created by crossing protective barriers of the naïve animal  which may vary based on reproductive status. Multiple physical layers must be breached in order to establish a productive infection. In the case of BVDV, the mucous secretions covering the nasal mucosa must be penetrated, followed by infection of the underlying cells. The innate immune mechanisms of these cells must be overcome, followed by penetration of the virus into the underlying cell layers. Finally, spread into the vascular system must be achieved for widespread dissemination. These physical and immunological barriers take a considerable toll on the infecting virus where only a small number survive, acting to reduce quasispecies genetic diversity in a stochastic manner. With each virus in the quasispecies differing from the population consensus sequence by only a few nucleotides, infection by a small number of viruses would, in all likelihood, result in a new, slightly different population consensus. Two models were proposed by Pfeiffer and Kirkegaard  that may explain genetic change in BVDV. The “tough-transit” model proposes that viruses have a difficult time passing the natural barriers and innate immunity of the host and the few viruses that succeed in bypassing the barriers establish the new viral population. The “burned-bridges” model proposes that viruses have little difficulty bypassing the protective barriers and the first to do so quickly establish an antiviral state that inhibits further infection by following viruses. The data presented here suggests that the “tough-transit” model may be more applicable. BVDV normally have difficulty passing the natural barriers, as evidenced by the few changes observed in viruses from non-pregnant animals. The few surviving viruses would quickly amplify and, depending on the sequence of these viruses, establish a new population consensus sequence. In pregnant animals, where immunity is already altered because of the pregnancy, the innate immune system may be somewhat leaky or less effective at stopping infection, allowing a greater number of infecting virus particles to establish a productive infection. This results in a population consensus with a larger number of nucleotide differences from the population mean of the infecting animal, all being dependent on the number of infecting viruses and which and how many became prevalent in the new quasispecies. This view was supported by the findings that the acute virus isolated from a pregnant dam already possessed a large number of nucleotide changes from the progenitor PI virus while the progeny PI virus showed only 2 or 3 nucleotide changes from the acute phase virus, resulting from infection of a non-pregnant animal (fetus) with intact innate immunity. Also, the viruses from the acutely infected steers (Figure1b) possessed few additional changes.
This study provided results that defined where and when nucleotide changes are generated or introduced into BVDV genomic RNAs. However, to understand the mechanism of genetic change in greater detail, a deeper analysis of the diversity in the viral quasispecies present in each virus population is needed. Future experiments will require deep sequencing of each quasispecies to look for the presence and prevalence of the specific mutations found in the new quasispecies to provide a more detailed understanding of genetic change in BVDV.