Human cytomegalovirus (CMV) is a common virus that infects almost half of people over 40, usually causing no symptoms. However, immunocompromised individuals can develop life-threatening disease. Once infected with CMV a person remains infected and the virus can reactivate many years after the initial event. Even though so many people have CMV, early events of infection are not well documented or studied. This is probably due to the fact that initial infection is mostly during childhood and asymptomatic. To better understand initial infection and spread, researchers at Fred Hutch in the Vaccine and Infectious Disease Division studied a cohort of 30 high risk infants over time, taking oral samples to test for the presence of CMV DNA, hoping to catch the initial infection stage and study viral spread. Dr. Schiffer and colleagues collected weekly swabs from infants and found 20 with primary infection and more that 130 self-limiting episodes that resulted in what they called transient infection. Transient infection was marked by swabs positive for up to 2 weeks that then returned to baseline after this time (figure 1A). These infections did not lead to long-term infection; thus they were termed transient. An infant was considered infected with a primary infection if the viral load increased and subsequent swabs returned positive for CMV by PCR (median time = 31weeks) (figure 1B). When the group compared infants that became primarily infected versus ones that are only transiently infected, the swabs prior to the onset of primary infection looked similar, suggesting that transient infection probably did not represent early primary infection.
Taking this information into account as well as previous observations the group created mathematical models to identify the early events in CMV infection that would lead to transient and primary infections similar to what was seen in their cohort. To do this, they started with a stochastic ordinary differential equation model that is a appropriate when working with an infection that most likely starts with a small number of events. The model takes into account biological processes like viral replication, spread, viral decay and infected cell death rate. The model does not include immune pressure but this is thought to not occur, since viral growth seems to be unimpeded during the early stages of infection. In this model the R0 (reproductive number) is defined as the average number of cells infected by a single cell (with no target cell limitation) and was calculated in a previous study to be low in primary oral CMV infection. To test the effects of viral spread and number of cells infected on transient versus primary infection, the group varied both R0 and I0, which is the number of cells initially infected with replicating virus. By running the model with various values for R0 and I0 they found that low number of cells infected and viral spread lead to transient infection while higher spread and number of initially infected cells causes primary infection. After testing various combinations, they found that an R0 of 1.1 with a I0 less than 10 lead to transient infection with similar characteristics to what was observed in the infant cohort. The researchers also found that the viral load was mostly a function of the number of cells initially infected. CMV viral loads in the model most closely mimicked the data when initial infectivity was 1 suggesting that most primary infections are initiated in a single cell. In regards to viral spread, the model is closest to the data when R0 was 1.1-1.5. Together this data suggests that infection begins with replication in a few cells and that infectivity is low.
To further test early oral CMV infections, a more complex model was established. This model takes into account the special constraints that tissues have on viral spread. In other words, there is not an endless supply of non-infected cells and thus target cells deplete over time with infection. The model uses a three-dimensional histology of stacked epithelial cells in mucosal tissue. In the model cells can infect their contacting neighbors in a lattice with a depth of 10 cells. This leads to a maximum R0 of 12. In this model each cell is assigned a unique duration of infectivity and thus a different R0, allowing for the expected differences among cells. Like the 1st model this model produced transient infection with a single cell infected (I0 = 1) and low viral spread (low R0). The three-dimensional model emphasizes the idea that transient infection is caused by low cell infectivity and spread. This data suggests that both transient and primary infection are caused by a small number of initially infected cells that that only a small number of viruses reach new target cells and infect them efficiently and that this is inherent of CMV. When asked to summarize the importance of the work and future direction Dr. Mayer, the lead author on the paper stated, “In this study, we identified that CMV, despite infecting nearly all infants in Uganda, is actually transmitted inefficiently. Using data analysis and mathematical modeling, we identified that most CMV infections start with a single virus infecting a single cell and are aborted before taking hold in the infants. Infants eventually get infected due to repeated household exposures. In the future, we wish to identify why infection is so inefficient and will study if this is a feature of the virus or immune system early during infection.”
This study was funded by the National Institutes of Health and the University of Washington Center for AIDS Research.
Mayer BT, Krantz EM, Swan D, Ferrenberg J, Simmons K, Selke S, Huang M-L, Casper C, Corey L, Wald A, Schiffer JT, Gantt S. 2017. "Transient oral human cytomegalovirus infections indicate inefficient viral spread from very few initially infected cells." J Virol. JVI.00380-17. doi: 10.1128/JVI.00380-17