Reconstructing epidemic spread and inferring host type across phylogeny

Collaborators at the Istituto Zooprofilattico Sperimentale delle Venezie (IZSVe) and I have been working to investigate characteristics of avian influenza epidemics.  We have recently published a paper in Transboundary and Emerging Diseases titled “Spatiotemporal reconstruction and transmission dynamics during the 2016-17 H5N8 highly pathogenic avian influenza epidemic in Italy”. The focus of this paper was to use integrated analyses of genetic sequences and associated metadata to learn about how the epidemic dispersed spatially and the host types involved in such movements.

We used the dates and locations associated with viruses isolated from 83 domestic cases and from 12 wild birds collected by IZSve field epidemiologists to simultaneously reconstruct virus phylogeny and the movements and dates associated with branches of the phylogeny. Among other features, this revealed a highly skewed distribution of movements with many shorter localised movements interspersed with occasional long distance dispersal movements.

This map shows a reconstruction of the spread of the genetic clade ‘Italy-B’ that was mostly restricted to the north-west of Italy (ancestor starred, red rings = wild bird), though also experienced dispersal to Eastern Italy followed by a long-distance southwards movement later in the epidemic:Figure_map

Reconstructing host type using evolutionary distances

We also wanted to reconstruct host type across the phylogeny in order to differentiate clusters of domestic-to-domestic transmission from incursions across the wildlife-domestic interface. This was challenging because of different sampling intensities in domestic and wild birds: a high proportion of domestic cases should be represented in the data whereas in contrast, the epidemic in wild birds was probably greatly undersampled. This meant that approaches such as ancestral state reconstruction would likely overestimate the proportion of internal nodes associated with domestic birds.

To estimate the likelihood of unobserved or cryptic wild bird-mediated transmission across the phylogeny we used an approach based on evolutionary distances – branch lengths of time-scaled phylogenies estimated using a molecular clock model in BEAST.

  1. For each domestic case we first identified a putative domestic ancestor as the phylogenetically closest case from an earlier date. For a cluster of domestic cases below, blue arrows identify putative domestic ancestors. putative_ancestor
  2. Phylogenetic (patristic) distances between domestic cases and putative domestic ancestors were calculated by summing the lengths of branches separating them.
  3. Values within the observed range of these phylogenetic distances were considered as possible thresholds. Moving from tips to the root, internal nodes were used to infer host type at internal nodes. Internal nodes with both descendants being domestic cases with distances to putative domestic ancestors below the threshold were considered domestic, otherwise wild bird was inferred.
  4. Using distances estimated across a range of phylogenies, the range of potential thresholds at which internal nodes were recorded as wild bird or domestic  was recorded. This is shown for a section of the outbreak phylogeny below.Figure_web_post
  5. The transmission type associated with each branch of the viral phylogeny could therefore be estimated as being either 1) wild bird-to-wild bird, 2) wild bird-to-domestic, or 3) domestic-to-domestic.

Using this approach, domestic-to-domestic transmission was estimated to be restricted to the second half of the epidemic under most possible threshold values, something which matched conclusions from epidemiological investigations. Using a threshold value that corroborated epidemiological investigations, we looked at the geographical distances estimated for branches associated with each transmission type.trans_gcd_boxplot

Where large geographic distances were estimated to have been covered in branches of the tree where evolutionary distances indicated a cluster of domestic-to-domestic spread (i.e. domestic cases on nearby dates separated by small genetic distances), we hypothesised that such instances may be characteristic of human-mediated movements. This approach highlighted branches linking a cluster of domestic cases associated with travel to or from a live bird market.

Molecular basis of immune escape by A(H9N2) avian influenza viruses

I’ve been working with collaborators at the University of Glasgow and the Avian Influenza Group at the Pirbright Institute on the genetic basis of antigenic variation among avian influenza viruses. We’ve uploaded a manuscript to bioRxiv on “The molecular basis of A(H9N2) avian influenza viruses”. Thomas Peacock and I share the lead authorship on this work, we’ve used a combination of experimental and computational approaches to explore the genetic basis of immune escape by these viruses.

A(H9N2) viruses are an economically important pathogen of poultry across much of Asia, the Middle East, and North and West Africa. They play a crucial role at the human-animal interface, particularly in China where they are the most common flu subtype in poultry. A(H9N2) viruses pose a threat to human health, both as a zoonotic agent in their own right, but also as an important donor of genes to novel reassortant viruses that may infect humans.

Like other flu subtypes, vaccine effectiveness is persistently challenged by the emergence of novel antigenic variants. Current understanding of antigenic variation among A(H9N2) viruses is largely derived from a handful of monoclonal antibody (mAb) escape mutant studies. These studies have identified a variety of amino acid substitutions that allow mutant viruses to grow in the presence of an antibody that targets a specific area of the virus.

Our main findings:

  1. Many mAb escape mutations are absent, or very rare, among sequenced A(H9N2) viruses.
  2. Several others had no significant effect on chicken antisera binding (mutations that result in extra glycosylation were a notable exception).
  3. Modelling antigenic and genetic data from circulating viruses identifies several novel amino acid substitutions that could explain antigenic variation in the field.
  4. Substitutions enabling immune escape by increasing glycosylation or receptor-binding avidity had the largest impacts on chicken antisera binding.
  5. Modelling and examination of sequence data suggests that of these two mechanisms of immune escape, modulation of avidity likely plays a greater role in evolution in nature.

Flu virus: egg-adaptation, dN/dS ratios and phylogeny

In recent years there has been increasing attention paid to the consequences of egg-adaptation of influenza viruses. To remain effective, the flu vaccine is regularly updated to ensure that it remains a good antigenic match to circulating viruses. The H3N2 component of the vaccine has recently been less effective than expected, despite the vaccine strain being well-matched to circulating strains. This is partly due to changes that occur when the vaccine is grown in chicken eggs.

Continue reading “Flu virus: egg-adaptation, dN/dS ratios and phylogeny”

Protein structure and antigen attractiveness

Great efforts are made to understand what makes particular areas of a pathogen protein attractive to the immune system. A better understanding of the biophysical and structural features that underpin antibody-recognition of antigens may allow us to infer the relative importance of the different areas (epitopes) recognised by antibodies and to predict which mutations are most likely to result in new pathogen strains able to evade pre-existing immunity.

Continue reading “Protein structure and antigen attractiveness”