CID Annual Symposium 2023: Posters

Morning Poster Session (10.30-11.10h)

Modelling the effects of virus manipulation of host and vector on plant disease epidemics: does model structure matter?

Elin Falla, Department of Plant Sciences

Plant viruses threaten global food security and are often transmitted by arthropod vectors. Non-persistently transmitted (NPT) plant viruses are characterised by a very short virus retention time in the vector and are transmitted almost exclusively by aphids. Compartmental models using ordinary differential equations to capture the course of an epidemic have been used in plant virus epidemiology for decades. However, the underlying model structure, in which the infective period of vectors is fixed, omits a key feature of non-persistent transmission: probing or feeding on a plant is often what causes an aphid to lose its infectivity. A recent model by Donnelly et al. (2019) captures this behaviour via a Markov chain that tracks the behaviour of individual aphids. We introduce a new compartmental model which replicates this model, while allowing the easy extensibility characteristic of compartmental models. It is comprised of linked Susceptible-Infected models for the plants and aphids, where loss of aphid infectivity is conditioned upon its probing and feeding behaviour, rather than occurring at a fixed rate. This additional biological realism means our model behaves differently to previous compartmental models of NPT viruses, therefore allowing us to more accurately investigate virus transmission dynamics for all NPT systems.

Protection of SARS-CoV-2 vaccination and prior infection against future infection

Peter Kirwan, MRC Biostatistics Unit

• Background: The SARS-CoV-2 Immunity & Reinfection Evaluation (SIREN) study is a prospective cohort study of adult healthcare workers in UK hospitals. We estimated protection of vaccination and prior infection against future infection for this cohort.
• Methods: Participants completed an enrolment questionnaire, routine RT-PCR testing was undertaken every two weeks. Vaccine uptake was high; the majority received BNT162b2. A multi-state Markov model was used to assess hazard of infection and duration of positivity given vaccination, infection history, variant circulation period, and other covariates. Goodness of fit was assessed by comparing expected and observed prevalences over time.
• Results:  38,973 participants were followed between June 2020-February 2022. Most were female, white, and aged 25-64. Hazard of infection was increased during the Omicron period (December 2021 onwards), as compared to the Alpha/Delta period (December 2020-November 2021), but reduced with vaccination, prior infection, older age, and reduced work-related COVID-19 exposure. The mean duration of positivity was 2.3 weeks (95% confidence interval 2.1-2.6 weeks), shorter among those with prior infection compared to naïve participants.
• Discussion: Both vaccination and prior infection were estimated as protective against future infection, although protection waned over time. The lower hazard estimated for older individuals may reflect behavioural factors outside the workplace.

Therapeutic interventions alter ecological interactions in the cystic fibrosis airway microbiota

Pok-Man Ho, Department of Biochemistry

I am interested in understanding better the interactions between microbes in human infection scenarios. For example, two species may be in conflict with one another, or they may synergise one another. In order to tease apart the ecological interactions between species, we have developed a statistical-mathematical pipeline for analysing the microbial population dynamics data in the cystic fibrosis (CF) airways. The airways of people with CF are frequently infected by a mixture of species, including well known pathogens such as Pseudomonas aeruginosa (PA) and Staphylococcus aureus (SA). To investigate this further, time-series data containing information about species presence/absence were obtained for the period 2008-2020 from the UK Cystic Fibrosis Registry. This allowed us to determine the probability distribution of ecological relationship(s) between species along a rolling time frame in a large population of patients. Our data indicate that these ecological relationships are shaped not only by e.g., antibiotic treatments, but also by other forms of medical intervention. We have also been using the pipeline to model the interactions in an experimentally-perturbable in vitro polymicrobial model of the CF airways, allowing us to quantify the ecological impact of mutating key genes, and study, in defined conditions, the influence of antibiotics on inter-species interactions. In summary, we have developed a pipeline which accurately captures the changing nature of inter-species interactions in complex, polymicrobial infection scenarios. Our goal is that this model may potentially inform on treatment regimes which minimize unfavourable (for the patient) microbial interactions.

Estimating the potential number of cases prevented by infant/ toddler immunisation with a MenACWY Vaccine. 

Lauren Adams, Department of Veterinary Medicine

In 2018 GSK announced their decision to cease the supply of Hib/MenC vaccine Menitorix®. Therefore, the UK immunisation programme needs to be considered. The quadrivalent MenACWY vaccine is likely to be the most sustainable option and is already used for the teenage immunisation programme, MenW and MenC cases may be controlled indirectly from this. However, since meningococcal disease occurs more frequently in the first 24 months of life vaccination at 3 months or 12 months should also be considered. We estimate the potential number of MenACWY cases prevented in one cohort born in 2025 from MenACWY vaccine given at 3-month or 12-month of age. We then used a transmission model to estimate the indirect effects of the teenage MenACWY programme before adding the direct effects of MenACWY immunisation at either 3 or 12-months. We also simulate the impact of Covid-19 lockdown on transmission. We show that carriage prevalence of MenACWY is well controlled by the teenage vaccination programme. Using the pandemic assumptions and maintained high vaccine uptake we predict that MenACWY carriage prevalence will be <1% by 2025. In all scenarios, we predict that very few cases would be prevented by the addition of an infant/toddler MenACWY programme. Although adding the MenACWY dose at 3-months consistently results in a larger number of cases prevented we predict a maximum of 22 cases prevented in the base case. 

Transmission dynamics of Japanese encephalitis in Bangladesh 

Mariana Perez da Costa de Albuquerque Duque, Department of Genetics

Japanese encephalitis (JE) is a mosquito-borne virus and the most common cause of viral epidemic encephalitis in Asia and Western Pacific where approximately 3 billion people in the world are at risk. The enzootic cycle involves mostly water birds, pigs, and the mosquito vector. Japanese encephalitis is a vaccine-preventable disease, and humans are incidental and dead-end hosts. Although Bangladesh shares borders with other endemic countries, incidence rates are heterogenous across the country, and no vaccination program is in place. To tackle JE, we need to estimate the number of infected individuals, understand individual and community risk factors for infection and map the risk across the country. This project will provide fundamental insights to inform policymakers on vaccine deployment strategies.

Multiplex micro-bead immunoassays for the detection of seroconversion to SARS-CoV-2 in patients and healthcare workers

Andrew Chan, Department of Veterinary Medicine

Measuring antibody responses to multiple antigens from SARS-CoV-2 infection by assays such as ELISA can be time-consuming and resource intensive. Multiplex micro-bead immunoassays allow for rapid assessment of seroconversion and require far fewer target antigens. Here we describe its use in characterising the humoral response to infection and vaccination in patients and healthcare workers from Royal Papworth Hospital, Cambridge, as part of the work carried out by the Humoral Immune Correlates to Covid-19 Consortium (HICC) during the pandemic. 

Pemphigus Vulgaris favours the enrichment of pathogens and resistome profile related to respiratory infections

Luis González, Wellcome Sanger Institute

Pemphigus Vulgaris (PV) is an autoimmune disease that produces oral blisters as the initial manifestation, spreading all around the skin. The primary cause of death in PV patients is systemic and respiratory infections. Previous studies revealed that PV influences oral microbiome composition by favouring the increase of pathobionts and reducing bacterial diversity. However, these studies were limited to 16S rRNA gene profiling, making it impossible to characterise pathogen invasiveness. In this work, we profiled microbial and resistome composition of the mouth microbiota in PV patients using metagenomic shotgun sequencing. We showed that PV favour enriching pathogens related to oral infections and respiratory sepsis. Moreover, we showed that patients with PV had more antimicrobial-resistant genes related to antibiotic respiratory treatment than healthy patients. This study improved our understanding of composition and pathogen invasiveness in the oral microbiome of PV patients. These results will have implications for future large-scale studies looking at the effect of PV in the oral microbiome for developing treatments to modulate the microbiota and selecting antimicrobial treatments to influence the clinical outcomes of this disease.

Inconsistencies between dengue infection intensities estimated from serological and passive case surveillance studies

Angkana Huang, Department of Genetics

Force of infection (FOI), an important metric when designing and implementing interventions, can be inferred from multiple data sources. Data from longitudinal serological studies are considered gold standards as they directly measure rates at which naive individuals seroconvert. Through introducing assumptions on gain and sustainment of seropositivity, cross-sectional serological data including individuals of variable age can draw information from differences in risk experienced across birth cohorts to provide estimates of FOI. Adding models to link between infection and generation of cases, FOI can be inferred from age-stratified case counts. Agreement between these estimates, however, are seldom assessed due to the rarity of longitudinal serological data.

Leveraging data from Kamphaeng Phet province, Thailand, a rare instance where all three data sources are available, we found poor concordance between the estimates. We analytically identified assay noise and antibody kinetics as important causes of the discordances and went on to show through simulations that inferences not taking into account these processes can lead to the observed inflation of FOI inferred from seroincidence data and the dampened FOI inferred from seroprevalence data. We are developing models to correct for these causes to reconcile the estimates.

Indoor heat stress mitigation in low-income Kenyan homes with mosquito prevention intervention  

Haiwei Li, Department of Architecture

The impact of global warming on health is increasingly becoming profound, especially in low- and middle-income countries like Kenya. With rapid urbanization and massive low-income housing construction, Kenya is under severe risks of indoor and outdoor overheating, which is commonly seen as an accelerator of transmission and geographical expansion of vector-borne diseases, such as malaria. Many mosquito prevention technologies have been implemented in residential houses, such as eave screening, window mesh, and bed nets. This study uses computational fluid dynamic (CFD) simulation to investigate the indoor heat stress of a typical low-income house with and without mosquito prevention technology in Siaya county in western Kenya. The intention is to propose optimal and applicable heat mitigation solutions, including ceiling insulation, white roof coating, and cross-ventilation for low-income housing with the consideration of mosquito prevention. Results show that houses with mosquito prevention interventions have the potential to reduce heat stress, provided the design of the houses is taken into consideration. The sustainable, resilient, and inexpensive design with retrofitting solutions can benefit policymakers in reducing heat-health burdens while reducing the incidence of vector-borne diseases in the low-income houses of Africa. 

The biofilm matrix of Pseudomonas aeruginosa: a highly ordered compartment for virulence and survival

Rahan Nazeer, Department of Biochemistry

Pseudomonas aeruginosa (PA) is a WHO “critical priority pathogen”. PA is the ultimate opportunist and exhibits a worrying level of multi-drug resistance. The organism is also exceptionally versatile, and readily “rewires” its basic metabolism allowing it to adapt to challenging environments, such as those in the human airways. Not only does PA adapt to its environment; it also adapts the environment to it, using an arsenal of secreted factors to break down organic matter, communicate with other cells, sequester scarce nutrients and kill hostile interlopers.

Perhaps the most basic way in which PA modifies its immediate environment is through close colonisation, mediated by the formation of biofilms. This entails the secretion of an array of biomolecules which form an ordered, protective compartment. This shields the encapsulated cells from immunological factors and antibiotics, making biofilm formation an integral component of PA’s colonisation strategy.

The extracellular matrix which characterises the biofilm is a complex network of proteins, secondary metabolites, nucleic acids and other exopolymers. Using a selection of defined deletions in key biofilm-associated genes, we have been interrogating the matrix structure and proteome, allowing us to understanding better which components interact and how the matrix might function as a discrete extracellular “compartment”. In particular, we are using quantitative proteomics to understand how the "matrixome" differs in composition in the presence and absence of key linchpin protein(s). Moreover, we are using a novel fluorescent dye to monitor how the dynamics of the biofilm structure change over time.

Afternoon Poster Session (14.45 - 15.30h)

Incorporating the genetic diversity of Klebsiella pneumoniae into antibody screening using novel high-throughput microscopy 

Peter Brian Gallagher, Department of Medicine

Klebsiella pneumoniae (Kp) presents a major threat to global health. However, no vaccine currently exists that can prevent Kp infection or reduce virulence. One reason for this may be the vast genetic of Kp. To investigate how this diversity impacts antibody response, a collection of 175 isolates was procured and characterised. Outer-membrane vesicles (OMVs) were harvested from 5 isolates and used to immunise mice. The resulting 5 groups of polyclonal sera were harvested, characterised, and screened against live isolates using high-content imaging (HCI). We found that the OMV-derived polyclonal antibodies were cable of both homotypic and heterotypic binding to extracted OMVs and isolates. Screening all 175 isolates against the 5 serum groups and a mixed serum group using HCI revealed relatively few isolates could be strongly bound by the polyclonal serum. This study represents the first use of HCI to screen an antibody response against a collection of Kp isolates. We have demonstrated that OMV-derived antibodies are capable of binding a number of unrelated isolates, however, this is likely to be a small section of the global Kp population. Future work will focus on investigating isolate characteristics influencing antibody binding, to increase isolate coverage.

Siderophore production by aquatic isolates of selected Gram-negative bacteria.

Alazhar Huned Colombowala, CIMR / Anglia Ruskin University

Antibiotic resistance is becoming increasingly urgent, leading to research into mechanisms to transport antibiotics into the bacterial cell. An example of this is exploiting siderophore-dependent pathways for the uptake of iron. In this study, we investigated aquatic environments for bacteria-producing siderophores. 110 bacterial isolates were acquired and assessed for siderophore production using the chrome azural S assay. Taxonomical identification using 16S rRNA and whole genome sequencing (WGS) identified an Enterobacter species, as a promising isolate, showing high siderophore production.  The study also investigated the characterisation of the siderophore using high-performance liquid chromatography (RP-HPLC) and liquid chromatography-mass spectrometry (LC-MS). The results suggest the siderophore produced was aerobactin or a highly similar ligand. However, Enterobacter spp. possesses biosynthetic loci for two distinct metallophores namely hydroxamate (aerobactin) and catecholate (enterobactin), surprisingly, RP-HPLC-LCMS did not reveal the presence of enterobactin in the culture supernatant of the Enterobacter isolate. The study continues to investigate siderophore production by purifying siderophores at various time points and growth phases to determine bacterial isolates capabilities for producing different siderophores. Furthermore, we investigate the efficiency of the siderophores when provided as xenosiderophores to other strains. The work aims to understand the transport and the use of siderophore(s) as potential therapeutics.

Growth, carbon utilization and transcriptional responses to bile, bicarbonate and mucin for a genetically diverse collection of Enterotoxigenic Escherichia coli

Anne Bishop, Wellcome Sanger Institute

"Enterotoxigenic Escherichia coli (ETEC) strains carry heat-labile (LT) and/or heat-stable (ST) toxins and colonisation factors (CFs), causing >80 million episodes of human diarrhoea globally per year.  Genomic information reveals a great deal of ETEC variation, with ten major lineages. Gene expression has focused on three CF types.

We compared growth for nine diverse ETECs with bile, mucin or bicarbonate and using Biolog phenotypic arrays.  For E1779 (CS5+CS6 LT+STh) we compared transcriptional responses to bile, mucin or bicarbonate.

We find intra-lineage variation in ETEC growth: lineage 1 strain E925 (CS1+CS3+CS21 LT-STh) struggles to grow in minimal medium compared to E1739 (CS1+CS3 LT-STh) and is deficient in glycerol utilization; lineage 3 strain E36 (CFA/I+CS21 LT-STh) struggles to grow with bicarbonate compared to E2980 (CS7-LT).

With E1779 RNA-Seq we catalogued ETEC responses to mucin for the first time, bicarbonate for the first CS5+CS6 strain and confirmed bile responses for E1779 are similar to other CS5+CS6 strains.  E1779 shows large shifts in gene expression in response to mucin or bicarbonate and select responses, biased towards predicted surface and secreted proteins, to bile.

These data underpin ongoing RNA-Seq comparing nine diverse ETECs to reveal evidence of evolutionary adaptations and whether potentially protective antigens are transcribed."

Ornithobacterium hominis: Interactions with the human host and microbiome

Susannah Salter, Department of Veterinary Medicine

Ornithobacterium hominis is a recently discovered bacterial species that colonises the nasopharynx. It appears to have a skewed global distribution, identified almost exclusively in Africa, South America, South East Asia and Oceania so far. Populations with O.hominis carriage also tend to have a high burden of respiratory diseases such as pneumonia and chronic rhinosinusitis. 

Analysis of 16S rRNA gene data from South Africa suggests that O.hominis is carried as part of a highly conserved microbial community, and that some of these taxa are acquired prior to O.hominis colonization. Preliminary work in vitro demonstrates that O. hominis has an inhibitory effect on the growth of some nasal bacteria such as Streptococcus pneumoniae, and a positive impact on Moraxella species. 

O.hominis also produces a toxin structurally similar to the Pasteurella mitogenic toxin (PMT). PMT can cause diverse injury to the animal host including atrophic rhinitis and necrotic lesions. Prior work shows that the O.hominis toxin has a comparable mode of action and host cell receptor(s) as PMT, but despite the similarity of these two toxins there is no record of PMT-like disease in people carrying O.hominis. We therefore plan to investigate how the O.hominis toxin interacts with human cells in vitro. 

Seasonal and spatial patterns of henipavirus seroprevalence in Straw-colored Fruit Bats (Eidolon helvum)

Maya Juman, Department of Veterinary Medicine

A number of bat species are known reservoirs of emerging zoonoses, including the Straw-Colored Fruit Bat (Eidolon helvum), a common migratory African bat. Its massive geographic range and highly mobile, nomadic lifestyle create ample opportunities for human contact and subsequent viral spillover, a risk which is exacerbated by the harvesting of this species by bushmeat hunters. It is the only known host of Kumasi virus (GHV), the first henipavirus identified outside of Australia and Asia, with unknown zoonotic potential. To assess the seasonal and spatial dynamics of viral circulation in this species, we conducted a study measuring GHV antibodies across two years and four roosts in Ghana. We uncovered consistent seasonal patterns of GHV seroprevalence across rural and urban roosts. Furthermore, recapture and longitudinal sampling in a captive colony revealed that peaks in seroprevalence are likely associated with reproductive seasonality in female bats, which pass maternal antibodies on to their pups. These results constitute the first preliminary indicators of higher-risk periods of possible GHV spillover in Ghana. Further work is required to assess human seroprevalence as well as compare these seasonal patterns to those of other paramyxoviruses and filoviruses circulating in E. helvum.

Exploring dengue virus genetic diversity and fitness in Thailand

Loréna Duret, Department of Genetics

Dengue fever is a mosquito-borne viral disease that spreads mostly in tropical regions, caused by the dengue virus (DENV). Subclinical infections and co-circulating lineages commonly hide the underlying dynamics of dengue fever from surveillance systems. Therefore, tracking the DENV lineage composition provides a window to better understand the epidemic drivers. Models informed by pathogen sequences can help. Here we use 1949 whole-genome sequences from dengue virus from Bangkok (1974-2014) to shed light on within-serotype composition and quantify the fitness of sub-lineages. We reconstruct timed-phylogenetic trees with Bactdating. We apply a recently developed index which quantifies sequence-specific diversity representativeness, which is linked to fitness. We use this framework to define lineage based on their relative fitness, agnostically from any previous definition. Through the analysis of this index, we find that dengue genotypes in Thailand exhibit a substantial genetic diversity, with several lineages co-circulating within any genotype. We quantify the fitness of each lineage within the population. We find that some lineages are substantially fitter than others. This study shed light on the diversity within dengue genotypes in Thailand and quantifies the fitness of each sub-lineage. It provides an exciting avenue to assess the contribution of those lineages to the disease burden.

A modelling approach to map the risk of HLB in the Iberian Peninsula

John Ellis, Department of Plant Sciences

Huanglongbing (HLB) is a devastating citrus disease, currently found in Asia, Africa and North and South America. At present, no cases of HLB have been found in Europe, but in the past decade one of the disease vectors, the African citrus Psyllid (AfCP), has been found in several locations in North-Western Spain and Portugal. The presence of an established vector population means there is a high risk of transmission between citrus if HLB is subsequently introduced.
We present the findings of a computational model of vector and pathogen spread in the Iberian Peninsula. Citrus density and climate in residential areas and commercial orchards influence the pattern of spread. Most vectors disperse locally and are dependent on the availability of citrus plants, but we also account for long-distance dispersal via mechanisms such as wind or human transportation. Using the current estimated distribution of AfCP, results often show a pattern of slow growth of the psyllid in the North-West. However, after an introduction of psyllid into the densely populated commercial citrus regions in the South or East, the population quickly increases. There is subsequently a high risk of rapid spread of HLB upon the introduction of an infected plant in this region.

Reconstructing individual transmission events to quantify effects of processes involved in SARS-CoV-2 spread in Thailand

Natcha C. Jitsuk, Department of Genetics

Thailand is the first country to report SARS-CoV-2 infected cases outside mainland China. The first transmission wave in January 2020 was suppressed in six months through an extensive country-wide mobility restriction. The country carried on with strict international border control measures to throttle new importation while domestic travel resumed. Responses to subsequent waves were localised to areas/subpopulations identified through large-scale contact tracing and genomic analyses to reduce the economic burden. The less stringent control measure likely contributed to the longer time to suppression of the second wave (Alpha-variant) and the uncontrolled transmission of Delta- and Omicron-variants. However, with multiple other changing factors including differences in transmission efficiency of the variants, changes in immunity profiles of the population (through infection and vaccination), and heterogeneity in control measures across places and time, the true efficacy of such strategies is hard to discern. Here, we apply a spatiotemporally structured phylogenetic framework on geolocated COVID-19 sequences from Thailand (N = 9,513, 2021-2022) to explicitly quantify the effects of different factors on transmission. The application of this framework which has been developed to reconstruct individual transmission events in an endemic pathogen system (dengue) to SARS-CoV-2 will generalise its utility to emerging pathogen systems.

Effect of COVID-19 lockdown on sexual health outcomes

Tasnuva Tabassum/Pantelis Samartsidis, MRC Biostatistics Unit

HIV and STI testing and diagnosis patterns among individuals attending sexual health services (SHS) were unavoidably disrupted by the March 2020 nationwide COVID-19 lockdown. In-person attendance was only feasible for those most at need, with some attendances being reconfigured to be online / remote. While changes in attendance, testing and diagnosis rates at the time of the lockdown can be readily quantified, attribution of those changes to the lockdown requires deeper understanding of the trends in these rates over time pre-lockdown and of potentially unobserved factors affecting both the pre-lockdown trends and any changes due to lockdown. In particular, we assume that three key unobserved factors drive changes in SHS attendance and HIV/STI testing and diagnosis: risk behaviour, healthcare-seeking behaviour, and SHS availability. Using Gaussian processes to model these unobserved factors pre-lockdown and a hierarchical model to relate observed attendance, testing and diagnosis outcomes to the latent factors, we estimate both the trends in these outcomes pre-lockdown; and how changes in these trends during lockdown might be mediated by changes in the latent factor trends. 

Bayesian combination of longitudinal studies to estimate the duration of SARS-CoV-2 PCR positivity in the general population

Joshua Blake, MRC Biostatistics Unit

The duration of SARS-CoV-2 PCR positivity is vital to inferring the incidence of infection from prevalence data and informing the public health response. Current estimates do not provide a full distribution and are from subsets of the population (eg: healthcare workers or hospitalised patients). We utilise two longitudinal studies: the first study (the Assessment of Transmission and Contagiousness in Contacts of COVID-19, ATACCC) provides granular data (daily testing), however, suffers from short follow-up and an inadequate sample size; the second study (the Coronavirus Infection Survey, CIS) provides coarse data, however, has a large sample size and long-term follow-up. We aim to combine these datasets to provide unbiased estimates of the duration distribution in the general population.

We use different statistical models for each dataset, and then use transformed posterior estimates of the ATACCC analysis as a prior distribution in the CIS analysis. Using different models initially is desirable due to the differing structures of the studies and needing to analyse the CIS data within a compute-constrained trusted research environment. To enable the combination, we develop a hierarchical prior which respects the posterior correlations from the first analysis and allows incorporation of subjective model uncertainty to accurately represent our prior beliefs.