Overview

The underlying theme of our research is to use mathematical models as a tool to develop hypotheses from existing datasets and to work closely with our laboratory and clinic-based collaborators to develop studies which can validate or refute these hypotheses. While the traditional focus of our group has been on herpes simplex virus-2 and HIV infections in humans, our toolset and scientific approach is broadly applicable across much of infectious diseases and immunology research. In recent years, the scope of our collaborations has expanded to cover multiple other pathogens and to include data from in vitro and animal models of infection. While our goal is often to optimize therapeutic and prevention strategies, we are also interested in addressing basic science questions within immunology and virology. Most recently, we have been deeply involved in modeling SARS-CoV-2 infection with a focus on therapeutics, vaccines, and epidemiology.

HIV cure is a major unmet medical need. Current antiretroviral therapies (ART) effectively eliminate HIV replication in CD4+ T cells, prevent progression to AIDS, and allow a normal lifespan. However, ART does not eradicate persistent viruses in an anatomically dispersed reservoir of latently infected cells. As a result, infected persons must take ART for their entire lifetime. We are working collaboratively to optimize multiple new technologies designed to eradicate the HIV reservoir.

We used a mathematical model to predict that after a year of ART, nearly all cells in the HIV reservoir are generated via CD4+ T cell proliferation rather than persistent low-level replication of the virus. In latently infected cells, HIV integrates within human chromosomal DNA and is therefore precisely copied each time the cell divides. We therefore developed the hypothesis that lymphocyte anti-proliferative agents may decrease the size of the reservoir. 

We are also developing testable mathematical models to optimize dosing of other potential cure technologies including virally-delivered DNA cleavage enzymes designed to target and terminally mutate viral DNA within infected cells; autologous stem cell transplantation with genetically modified cells which are resistant to HIV; neutralizing antibodies and synthetic peptides that prevent infection of new cells; and CAR T cells that eliminate HIV-infected cells. Our ultimate goal is to assist in the strategic development of combination therapies, which would allow infected persons to remain healthy off of ART and to pose no transmission risk to their sexual partners.

Other ongoing projects consider the role of the immune system in shaping how HIV evolves in a person over time and simulations of clinical trials of antibody infusions which intend to prevent incident HIV infection.

We have been deeply involved in considering multiple facets of SARS-CoV-2 pathogenesis and epidemiology. We developed mathematical models which capture the timing and intensity of the immune response against viral replication in the upper airway and used these models to predict that early treatment is more likely to result in improved clinical outcomes.

We combined models of observed shedding kinetics with measures of SARS-CoV-2 spread in the general population to estimate viral loads required for transmission, and to explain the observed predisposition of the virus towards super-spreader events. This model served as a platform to estimate how even imperfect masks could dramatically limit the rate of SARS-CoV-2 spread in the population.

We have also developed models intended to recapitulate the epidemic in King County and used these to estimate the future impact of different vaccine strategies.

Based on the tragedy that we have all endured, and the many evident failures in our national pandemic response, a long-term goal of the group is to use models to improve our response to the inevitable next pandemic.

Herpes simplex virus-2 (HSV-2) is the leading cause of genital ulcers worldwide, an important risk factor for HIV acquisition and transmission, and a cause of severe disease in immunosuppressed individuals and infants. No effective, licensed vaccine exists for HSV-2 in humans, and existing antiviral therapies are only partially effective. Accessibility of the skin and mucosal tissues where HSV-2 replicates has allowed for frequent sampling to measure viral levels and immune responses, enabling the completion of complex and informative studies of this fascinating virus.

Through a strategic combination of human studies, animal models, and mathematical modeling, our group has advanced scientific understanding of the complex interactions between HSV-2 and the host. Specifically, we demonstrated that viral shedding episodes occur almost weekly but are extraordinarily variable over time and space. HSV-2 is often eliminated in just a few hours and is shed without symptoms, but some episodes cause uncomfortable lesions and persist for more than a week. The immune response, characterized by dense sheets of CD8+ and CD4+ T cells, is intense but highly localized to micro-regions where viral levels were previously high. Our mathematical models help explain high episode frequency and diversity, as well as the observed spatial features of shedding and the immune response. These studies will help identify the density and type of T cells necessary for an effective vaccine.

Antiviral therapy achieves almost complete effectiveness for many viral infections such as HIV and hepatitis C. Yet, treatment for other viral infections such as SARS-CoV-2, cytomegalovirus (CMV) and HSV-2 is characterized by partial efficacy and substantial variability in therapeutic response among infected persons. We developed models to explain the phenomenon of breakthrough viral shedding while on antiviral agents, incorporating virologic, immunologic, pharmacokinetic, and pharmacodynamic parameters to more accurately predict antiviral agent activity. Our models are highly flexible to simulate various viral infections and account for both the potency of the therapy and the effectiveness of the underlying host immune response. This approach allows us to accurately simulate clinical trials and predict the relative effectiveness of different dosage regimens, with the goal of dose optimization. 

Modeling approaches are also powerful for simulating the activity of infection prevention modalities. To this end, we are developing mathematical models to better understand emerging outcomes of the broadly neutralizing antibody VRC01 for preventing HIV infection, for dosage regimens currently under evaluation in the HIV Vaccine Trials Network (HVTN) sponsored Antibody Mediated Prevention (AMP) trials.

We are extremely interested in furthering this clinical trial simulation strategy towards viruses with global epidemic potential such as Ebola virus, Lassa viruses and the known human coronaviruses.

Tissue resident T cells (T_RM) are a memory T cell subset critical for early control of viral infections. T_RM remain lodged in tissue long after infection is eliminated and exist in disequilibrium with T cells that circulate in blood. T_RM are present at portals of viral entry, efficiently surveil for infected cells by trafficking within the tissue, proliferate locally when identifying infection, rapidly kill infected cells, and simultaneously express an antiviral alarm which can protect surrounding cells.

A major challenge in the field is linking highly mechanistic observations from animal infection models to observational data from human studies. We are employing complementary tools to understand the duration and mechanisms of protection that CD4+ and CD8+ T_RM provide in tissue, specifically to inform development of realistic 3-D mathematical models of the infection microenvironment. Human biopsy studies allow us to characterize the spatial nature of T_RM and antiviral cytokine dynamics, while studies in a mouse mucosal HSV-2 infection model that we helped develop will elucidate protective parameters of CD4+ and CD8+ T_RM including density, movement and killing rate. Our ultimate goal is to identify threshold densities and phenotypic characteristics of T_RM necessary for rapid viral containment following therapeutic or prophylactic vaccination.

Transmission of viral infection from person to person is difficult to observe in great detail, particularly in children. Through studies focused on herpes virus infections (herpes simplex virus 1, Epstein-Barr virus, cytomegalovirus, human herpes virus 6, and human herpes virus 8) in a cohort of Ugandan infants, mothers and siblings, we quantify and model transmission of these viruses. We used statistical and mathematical models to compare patterns of viral shedding between infants, siblings and mothers. These analyses provide a unique window into the maturation of the immune response against each of these viral pathogens with age. We are also using models to identify fundamental and virus-specific features of viral transmission, including viral load necessary for transmission, risk factors for transmission, viral incubation period, and infectivity of siblings versus mothers for each virus. 

We most recently applied similar techniques to SARS-CoV-2 studies as well. We also helped conceive a study to examine the dynamics governing HHV8 transmission in Uganda. HHV8 is a major driver of Kaposi’s Sarcoma in this region.

The traditional “one microbe, one disease” view is being reimagined, as it becomes increasingly clear that thousands of bacterial species inhabit different anatomic regions of a healthy human body. Perturbation of this complex bacterial fingerprint, called the microbiome, can lead to disease or health improvements. Rapid shifts in microbial diversity and abundance within the human vagina often correlate with development of bacterial vaginosis (BV), a condition associated with discharge and increased STI risk. Yet, dramatic changes in microbial levels also sometimes occur following intercourse and menses, without necessarily leading to BV. We studied the bacterial population dynamics of BV in order to correlate measures of species diversity and abundance with development of BV. Since microbial shifts occur within hours, we are using detailed study protocols in conjunction with mathematical modeling to assess whether key nutrients allow certain species to outcompete others, and to identify which species are critical for altering and maintaining bacterial composition. This information may inform development of more strategic approaches towards treating BV.

Viral infections are a major cause of mortality following stem cell transplantation. By measuring double-stranded DNA (dsDNA) virus kinetics in blood during the high-risk weeks immediately following myeloablative therapy, we helped demonstrate that multi-virus reactivation occurs frequently and persistently in patients during this period, and that cumulative dsDNA viral burden is associated with increased mortality. Our goal is to determine whether novel antiviral therapies might improve outcomes for transplant recipients. We are interested in performing similar studies for respiratory viruses, which are also a potentially lethal complication following transplantation.

A new area of interest is characterizing the nature of the CMV-specific immune response following allogeneic stem cell transplant. Despite relatively effective antiviral therapies, CMV remains a key driver of post-transplant outcomes including mortality, graft-versus host disease and development of other infections. New tools allow definition of the CMV-specific T cell response at the single cell level including discrimination of cells according to donor or recipient origin, T cell subset, and specificity for the virus. We intend to model the dynamics of these different subsets as they re-emerge after transplant, in reference to detectable reactivating virus.