Doodle by Elisa Visher

Abstract:
Bats are reservoirs for the world's most virulent viral zoonoses, including Ebola and Marburg filoviruses, Hendra and Nipah henipaviruses, and SARS and MERS coronaviruses, pathogens which they host without exhibiting symptoms of clinical disease. Recent molecular advances suggest that the evolution of flight may have promoted the development of anti-inflammatory physiologies allowing bats to effectively mitigate oxidative stress accrued during metabolism--with cascading consequences for longevity and viral tolerance. My research investigates ecological and evolutionary questions in the bat-virus system, exploring the mechanisms enabling bat virus persistence at the population level, as well as the impacts of unique bat immune strategies on the probability of between-host viral transmission and the evolution of within-host viral virulence. I combine field studies focused on longitudinally-monitored fruit bat populations in central Madagascar with in vitro experiments in bat tissue culture and theoretical adaptive dynamics approaches to explore questions related to the persistence, evolution, and cross-species emergence of zoonotic viruses from bat reservoirs into human hosts.

Abstract from Cara Brook

Doodle by Alison Feder

Emily Ebel visited the Berkeley EEID seminar from the Petrov Lab at Stanford to present her work on genetic and phenotypic variants that shape malaria fitness in human blood. 

Poem by Nina Sokolov

​Using C. elegans to study the role of host genetics in shaping microbiome structure and function
The gut microbiome contributes to host health and fitness. Phylogenetic analyses demonstrate the importance of evolutionary processes for shaping host-microbiome interactions. However, identifying host genes shaped by the microbiome, and characterizing their involvement in determining microbiome structure and function has been difficult, in particular in vertebrates, where inter-individual variation masks shared patterns. Work in C. elegans offers the opportunity to work with clonal host populations, reducing noise and highlighting gene signatures, and facilitates genetic manipulation to identify relevant genes. In my talk I will describe our work characterizing the worm gut microbiome and the role of host genetics in shaping its structure and function. I will describe in greater detail the identified involvement of the conserved TGFb/BMP pathway in controlling commensal abundance and function, and will consider the implications of this for pathogenic potential of commensal blooms and dysbiosis. 

Abstract by Michael Shapira

Illustration by Elisa Visher

Carly Rozins presented her work on integrating disease models with a 'perception kernel' that allows vectors to move based upon how they are perceiving resources in their surroundings. For a Wordsworth based, poetry style summary:

Pollinators wander lonely as a cloud
And float among the flowers in the Italian nation
They search for nectar among the crowd,
Of their host, a mountain carnation
Whether or not they choose to leave
Depends on what their biology says they can perceive
 
Continuous as the stars that shine
And twinkle in the Milky Way
The carnations upon which the pollinators dine
Are spread out into an array
The pollinators determine their density with a glance
Then to the best patch, they dance
 
As the pollinators move, they
Transmit anther smut from flower to flower
And thus to disease dynamics, perception kernels may
Lend some extra explanatory power
Carly has mathed-and mathed-with much thought
And from her models, insight can be brought
 
So oft when vectors can perceive,
And make decisions
About when to leave
Based upon how to find the best provisions
Smut epidemics will spread faster
Amongst all the flowers in the pasture
 

Summary by Elisa Visher

Illustration by Kacie Ring

Last week, we had Kacie Ring visit us from the Swei Lab at SFSU to present her research on lyme disease in the Bay Area! This week, we've written our summary as a poem!

​
In California, we have a lizard
That cannot survive in those northeastern blizzards.
The black-legged tick likes to feed on its blood

Which carries a toxin that nips Lyme Disease in the bud
Lyme burden is lower where these lizards slither
Eco-prophylaxis: Nature's a wizard!

Poem by Cara Brook

Illustration by Nina Sokolov

Lewis has worked at University of Exeter, UC Berkeley, Emory University, and most recently UGA, to understand when and how we can gain novel insights into bee management by incorporating the ecology and evolution of disease. 

As will come as no surprise to most readers, bee populations (both managed and wild) are in rapid decline. This is due in part to pesticide use, changing land use patterns, and pest/pathogen expansion. In the case of monocropping, there is often a large burst of flowering (i.e. resources) at one point in time, followed by an absence of flowers for the rest of the season. For honey bee managers, this means they need many hives to take full advantage of the bloom, but are then stuck with hungry bees once the flowering is finished. This has, of course, led to movement of bees around the landscape. Such movement has had predictable impacts on the spread of pests and pathogens, both for the managed bees and for the wild populations with which they share habitat.

Approximately 3/4 of the Honeybees in North America are moved around the country on trucks. Such industrial-scale bee keeping has brought with it numerous new challenges, especially in terms of disease spread. However, the silver lining of this is that new findings and emerging best practice in bee keeping can be rapidly implemented. In a recent ecological model (see here: https://besjournals.onlinelibrary.wiley.com/doi/full/10.1111/1365-2664.13461), Lewis and colleagues explored how hive orientation might impact pathogen spread by examining the impact of spatial structure on R0 (the number of secondary infections that result, on average, from a given infection in a susceptible population). As expected, prevalence (proportion infected) increases with increasing R0. They then find that the least intensified apiary (a linear array of about nine hives) has a lower predicted prevalence across the spectrum of R0 values than the intensified apiary (with 100s of hives in a grid), but the impact is not particularly strong. This is especially true at higher R0 values (where most estimates for bee pathogens exist). So, it seems like crowding alone can not explain the increased spread of disease being observed. 

Now broadening the focus beyond whether a bee is infected or not, Lewis turned his attention towards severity of infection. The problem, however, is that this will be affected by numerous inter-related factors, including pesticides, stress, and ecology. As a first test of this, Lewis brought feral (no management), traditionally managed, and industrially managed bees into a common garden for a year, then screened for viruses. He found some grouping of viruses by beekeeping history and higher viral titres on average in highly managed bees. However, feral bees carried much higher titres of many particular viruses. 

Finally, in terms of the much-hated varroa mites, which have recently emerged and wreaked havoc across the globe, previous work has shown that the presence/absence of the mite changes the prevalence of viral pathogens. As varroa spread across New Zealand, deformed wing virus increased in prevalence and severity (number of copies observed per bee). A similar pattern was observed in Hawaii. Interestingly, strain diversity rapidly decreased as well, leading to a single highly virulent strain dominating the landscape. This is most likely due to the success of this virus in the mite itself (see Campbell et al. 2016 for an elegant demonstration of this).

Overall, it seems intuition does not get us far in this system - in part because of the many factors interacting to shape the ecology and evolution of disease in managed settings. Thank goodness for guidance from theory, diligent researchers, and people like Lewis who choose to spend their intellect and time focusing on solving these critical questions!

Summary by Britt Koskella

Doodle by Elisa Visher

This October, post-doc Saki Takahashi visited our seminar from across the bay to present her work on "Characterizing re-emergence patterns of dengue in Brazil". Saki recently started her post-doc after finishing up her PhD with Jess Metcalf at Princeton. Saki is broadly interested in how immunity drives transmission dynamics in multi-strain pathogen systems. For many infectious diseases, we know that hosts develop immunity after infection. However, this immunity can vary in how long-lived it is, how strong it is, and how broad it is, with all of these variables shaping how immunity influences disease dynamics.
Saki uses modelling techniques to elucidate patterns of transmission from epidemiological data. Specifically, Saki build these models with age-structured case incidence data, as it can be especially informative in teasing apart patterns of immunity, strain dynamics, and transmission. This age-structured data can be important because shifts in the age distribution of cases can indicate changes in the distribution of susceptibles and the force of infection of different strains. 
For the project that she presented to us at the Berkeley EEID seminar, Saki looked at dengue dynamics in Brazil. Dengue is an arboviral disease that has recently re-emerged in Brazil. Globally, it is the most rapidly spreading arboviral disease because of its multiple serotypes, increased dispersion of its vector, urbanization, and global travel. Dengue has four antigenically distinct but interacting serotypes.  Humans can be infected by multiple of these serotypes because infection with each strain does not offer complete immunity against others. However, previous infection with a different stain of dengue does change the clinical symptoms of infection. People normally have worse symptoms during their second infection (including dengue fever and dengue hemorrhagic fever), after generally asymptomatic primary infections.
In Thailand, where all four strains are endemic, dengue is generally considered a childhood disease because each strain’s force of infection is high enough that people tend to be infected by multiple strains and have clinically symptomatic dengue infection during childhood. However, Thailand has recently had a demographic transition with people living longer and having fewer children, which has decreased this force of infection and led to the average age of cases being older. This illustrates how the age-structured case data that Saki uses can be hugely informative for understanding the transmission dynamics of multi-strain pathogens like dengue.
Saki, therefore, turned her attention to using such age-structured data to model dengue dynamics in Brazil, where dengue has been re-emerging since 1986 due to the re-emergence of its mosquito host, Aedes aegypti, in the 1970s. Recently, in 2007, clinical cases of dengue have shifted from mostly being seen in adults to mostly being seen in children. Saki therefore wanted to ask whether she could build a model to capture dengue’s reemergence and temporal variation of transmission in Brazil and use it to identify factors that could explain the shift in the age-distribution of cases.
Saki and colleagues built models fit to age-structured epidemiological data which used time varying forces of infection to allow for the unequal circulation of serotypes over time, accounted for the different clinical symptoms of primary and secondary infections, and allowed for differences in reporting rates. The models could also capture finer spatial scales, such as regions within states. From these models, they were able to test between hypotheses to clarify possible causes of the demographic shift in dengue epidemiology over time. They find good support that the shift in the age distribution of cases was driven by the re-emergence of the dengue 2 serotype in 2007, especially in Brazil’s north east states. Basically, before the re-emergence of dengue 2, children likely had asymptomatic first infections, but did not have secondary infections leading to clinical manifestations until adulthood. After 2007, the renewed circulation of dengue 2 led to increased forces of infection and individuals were getting infected with their second serotype of dengue during childhood.
In summary, data on the age distribution of cases can be used to yield insights into population susceptibility and disease dynamics. Models using this data suggest that the 2007 shift in the age distribution of cases in Brazil was due to the re-emergence of dengue 2. This may suggest that age-dependent clinical guidelines should developed to reflect changing age burdens.
Finally, Saki thinks that one of the next steps for this project is to validate the model with environmental variation data, more epidemiological surveillance data, and more orthogonal data. She also wants to do simulations and modelling on time to endemicity to get an understanding of the epidemiological transmission from reemergence to endemicity.

Summary by Elisa Visher

Illustration by Alison Feder

Evolutionary tradeoffs in an insect-virus model system
Last week, graduate student Elisa Visher of the Boots lab (link: https://bootslab.org/) presented details related to her most recent experiment at our Ecology and Evolution of Infectious Diseases seminar. Elisa works in an experimental evolution model system which Mike developed as a graduate student back in the UK. In this system, Elisa manipulates interactions between a host, the Indian meal moth (Plodia interpunctella) and its species-specific pathogen, Plodia interpuntella granulosis virus (PiGV) to explore evolutionary tradeoffs between investment in resistance and growth on the part of the moth and infectivity and productivity on the part of the virus.
 
Experimental evolution in a host-pathogen system typically involves growing a generation of pathogens on a specific host, then transferring this pathogen to a new generation of hosts, such that over time, the experimenter can quantify adaptations of the pathogen to the host environment. In the Boots lab, the experimenter (in this case Elisa) infects a generation of Plodia larvae by supplying them with virus-infected food. At timepoint one, called the first ‘passage’ in a ‘serial passage’ experiment, Elisa mixes virus in sugar and puts it on a petri dish for the larval meal moths to consume. PiVG is an obligate killer which is transmitted via larval cannibalism, so for the next passage, Elisa collects dead cadavers from the first passage, mashes them up, and mixes them in food for the second generation. Over time, the pathogen is influenced by natural selection in that it competes with other genotypes in the virus population. By being the virus genotype that (a) infects the most cadavers and (b) produces the most virions when the insects get ground up, one virus genotype can outcompete the others to be propagated into the future. The Plodia, by contrast, are under selection to resist this virus in order to survive.
 
In past work (i.e. his PhD), Mike has demonstrated that Plodia will develop resistance to PiGV, though at a cost of longer development times (Boots and Begon 1993), meaning that the more effective a larvae is at avoiding infection, the longer it takes to grow into a moth. Mike has further shown that these costs are magnified when resources available to the host are scarce (Boots 2011). More recently, Lewis Bartlett of the Boots lab developed twelve distinct inbred lines of Plodia to demonstrate that this tradeoff between virus resistance and host growth rate is a constitutive trait with a genetic basis (Bartlett et al. 2018).
 
For one chapter of her PhD, Elisa is expanding on these themes, using Lewis’s inbred lines, to show that PiGV can evolve to specialize on distinct genotypes of Plodia. In a serial passage experiment, Elisa evolved the same stock of virus across three different Plodia genotypes for nine generations, then assessed each evolved virus’s fitness on this familiar genotype and the other two genotypes across the time course of serial passage. Critically, Elisa measured “viral fitness” in two different ways—both in terms of a virus’s capacity to infect a Plodia host, as well as its virion productivity after infection. Two of the viruses she passaged evolved to be “infection specialists” for their host genotype, meaning that they were more effective at infecting the host they had evolved with than the other two genotypes. The third virus, by contrast, evolved to be a “productivity specialist,” showing no difference in its ability to infect other Plodia genotypes at the end of the experiment but producing many more virions post-infection when assayed on its host genotype.
 
In ongoing work, Elisa is using this same system to test the so-called ‘monoculture effect’ to explore whether more diverse host environments made up of multiple different Plodia genotypes select for more generalist viruses which can infect hosts more broadly but may be less virulent. In agricultural systems, we fear that the absence of genetic diversity allows for pathogens to specialize on a single host and attain high levels of virulence which result in crop destruction. In the future, Elisa also plans to manipulate the physical arrangement of Plodia genotypes in a microcosm to further explore the effect of spatial structure on these evolutionary tradeoffs in a host-pathogen system. We are excited to hear what she finds out next!

Summary by Cara Brook

Illustration by Elisa Visher

​A couple of weeks ago, we held a special early session of the Berkeley EEID seminar so that Professor Ayesha Mahmud from Demography could give a presentation on her research. Ayesha’s research program uses demography methods to study infectious disease spread. At the seminar, she presented her research on the spread of Chikungunya in Dhaka City, Bangladesh.
 
Chikungunya is a viral disease transmitted by Aedes mosquitos that is spreading around the world. It doesn’t cause a large number of fatalities, but can leave people who have been infected, especially the elderly, with long term symptoms. Ayesha focused her talk on work that she has done modelling the 2017 outbreak of Chikungunya in Dhaka City, Bangladesh. Dhaka City is one of the densest megacities in the world with about 18 million people in 1300 sq km. This dense population, along with rapid urbanization, regular monsoons, and almost no vector control, has made Dhaka City susceptible to regular outbreaks of mosquito-borne diseases. There have been frequent dengue outbreaks since 2000 and Chikungunya outbreaks in 2011 and 2017.
 
During the 2017 Chikungunya outbreak in Dhaka, hospitals reported 13,176 cases, but this record is likely an underestimate because hospital reporting was not mandatory. Ayesha therefore based her models on a set of data from collaborators who did household surveys asking about recent symptoms in an effort to get better data on Chikungunya infection rates. They found that 77% of the households surveyed had symptoms, 51% of these individuals had antibodies, and 70% of those had evidence of recent infection. That means that nearly 50% of the population had been recently infected with Chikungunya.
 
With this data, Ayesha and team were able to explore the spatial and temporal heterogeneity in disease burden. This scale of data was especially important as one of the big concerns during the 2017 epidemic was that the outbreak would spread out of Dhaka city into neighboring towns because Eid was right around one of the two peaks in incidence in the city. Using cellphone data, Ayesha was able to track when people were leaving Dhaka city and where they were going. As expected, there were small spikes in city residents travelling over the weekends, with much larger spikes for the Eid holidays.
 
Using a modelling framework with this mobility, demography, and infection data, Ayesha and team were able to make predictions about disease dynamics across the country. The model showed that Chikungunya was likely exported out of Dhaka in a spatial pattern that differed from a standard gravity model as there were certain patterns of travel between different cities in Dhaka. The model did not pick up the double peak in incidence, but the combination of an SEIR and mosquito model did pick up the epidemic curve. The patterns predicted by the model also matched some of the limited reporting from cities outside of Dhaka, which also tended to peak around the Eid holidays. Now, Ayesha is combining these models of case importation risk with metrics like population and mosquito densities in these other cities to predict epidemic risks. They are also looking at data from a Dengue outbreak in Dhaka this year that had much better reporting to see if mobility data can again make some predictions about disease spread.
 
Overall, Ayesha’s talk gave us a great idea about how different types of data can be combined with models to track disease dynamics and impressed upon us how important movement patterns and mass-movement events (like holidays) could be for disease transmission.

Summary Elisa Visher

Illustration by Elisa Visher

 ​Last month, Chris Hoover presented his research during the Berkeley EEID seminar series. Chris is a PhD student with Dr. Justin Remais at UC Berkeley working on optimal control strategies of Neglected Tropical Diseases (NTD’s) with applications to schistosomiasis. Schistosomiasis is the second most devastating parasitic disease worldwide and caused by parasitic flatworms called schistosomes. The disease is spread by fresh water contaminated with parasites that are released by freshwater snails. There is currently no vaccine to treat the disease but global control efforts have primarily on massive drug administration (MDA) often targeting school aged children. While MDA reduces morbidity, it many regions it has been inefficient in reducing transmission resulting in rebounds of the disease. Recent evidence suggests that MDA combined with snail control efforts is the most cost-effective intervention strategy.
 
With this background in mind, Chris models the effect of worm population dynamics on schistosomiasis transmission dynamics with the ultimate goal of designing effective control strategies. Some important features of the Susceptible-Exposed-Infection model includes environmental transmission, parasite burden, and density dependent processes. Infection is modelled as the population mean worm burden. Negative (i.e. crowding) and positive (i.e. mate limitation) density dependent processes appear in the force of infection of the intermediate snail host. Using these features of the model, one can estimate the breaking point threshold of the population size (i.e. endemic equilibrium) below which the population gets reduced to zero (i.e. elimination). The breaking-point of the mean worm burden is analogous to the herd-immunity threshold as the host becomes resistant to the infection when the worm burden falls below the threshold. An additional feature of the model is that we can manipulate the exposure parameter influencing transmission that we can intervene on. ​By reducing the exposure contamination parameter and the mean worm burden, reasonable levels of MDA coverage can lead to the breaking-point (i.e. control). Current work seeks to understand how demographic stochasticity impacts intervention strategies, such as small population sizes around the breaking-point.
 
The theoretical framework developed by Chris and collaborators is also very informative for real-world infectious diseases. For example, China has been successful in reducing schistosomiasis transmission using an integrated approach that includes snail habitat reduction, improvements in agricultural practices, and MDA. The work presented by Chris is an excellent example of applying well-grounded ecological theory to the management and control of infectious diseases.   

Summary by Senay Yitbarek