Phylogenetically Mapping the Evolution of the Zika Virus as It Spread Across the World

During the first months of 2016, my research team and I started tracking the genetics of the Zika virus’ spread across the world using Nvector, a tool developed, and currently only in use in our laboratory at the Department of Bioinformatics and Computational Biology at the University of North Carolina, Charlotte. The combination of traditional phylogenetic tools and Nvector allowed us to rapidly perform phylogenetic analyses of the genomic differences and relationships of the Zika virus sequences generated by different research groups around the world, and project the generated phylogenetic trees onto a global map. This approach was pioneered by Daniel Janies, PhD, a Carol Grotnes Belk Distinguished Professor of Bioinformatics and Genomics at University of North Carolina at Charlotte, who had performed similar analyses during prior outbreaks of Middle East Respiratory Syndrome (MERS) and influenza A viruses.

With these analyses, we were able to phylogenetically demonstrate that the viral sequences obtained from Zika as it crossed the Pacific Ocean (and subsequently radiated across northern Latin America and the Caribbean) were unlikely to be derived from the African strain of the virus, which was first described in 1947. The sequences from Brazilian and Pacific Island Zika isolates clearly clustered as a different strain, or clade, and were found to be more closely related to the historic Asian rather than the historic African isolates. This cluster is now usually referred to as the Asian-Pacific-American strain (also known as the Asian strain). Our review of the historic literature as well as subsequent genetic analyses suggests that these African and Asian genetic clusters (or clades) had circulated largely independently for many years prior to the first detection of the virus in the Zika forest of Uganda in 1947.

When analyzing the metadata associated with these various sequences, we realized that most of the Zika sequences isolated in Africa were from either different species of mosquitoes or from non-human primates, rather than from patients, whereas the available Asian sequence accessions were exactly the opposite; most Asian Zika sequences were sourced from human serum-isolated virus. This observation raised concerns about intrinsic selection bias and skewing within the available data, which we highlighted in our study published in Cladistics in December 2016, as the consistency and predictive value of our analyses were dependent on the data available at the time.

In early 2016, one of the most pressing questions surrounding Zika, was why the virus appeared to be causing a (previously unreported) birth defect syndrome in infants, as well as Guillain-Barré syndrome (GBS) in adults. We hypothesized that both syndromes may reflect a type of autoimmunity, which could have been triggered by changes in the Zika genome, and that we could quickly gain insight into this possibility using computational modeling tools. As we were already investigating the phylogenetics and evolution of the Zika virus, it was fairly straightforward to search these evolving viral sequences for changes in predicted Zika virus protein B cell epitopes, and then compare these evolving Zika epitopes to computationally predicted human protein epitope sequences.

With the help of Robert Malone, MD, MS, and Jane Homan, PhD, BVMS, MVSC, and others from Atheric Pharmaceutical and ioGenetics, we identified a set of mutated viral sequences encoding predicted epitopes with homology to epitopes predicted for human neural development-related proteins: NDF4 (Neurogenic differentiation factor 4) and Nav2 (Neural navigator protein 2). Based on this observation, we raised the hypothesis that epitope mimicry may contribute to both congenital Zika syndrome as well as GBS. To better explain, one hallmark of epitope mimicry is when a patient’s proteins become a target of their own adaptive immune response (autoimmunity) after they have been infected by a pathogen, which expresses antigenic proteins with similarities to proteins that are normally produced by the patient’s own cells. Clinical autoimmunity can occur when virus proteins are similar to human proteins; antibodies produced to fight the virus can end up fighting not only the virus but also the human host.

By using computer software tools to predict and compare Zika viral B cell epitopes likely to be recognized by the human immune system, identifying which of these epitopes corresponded to viral genetic changes during evolution and spread across the Pacific to Brazil, and then comparing these mutated epitopes to the entire predicted human B epitope proteome, we were able to identify two sets of mutated sequences (with homology to human proteins NDF4 and NAV2) which were specific to the evolving Asian-Pacific-American strain. This is important because it may help explain a change of behavior of the disease consequent to changes in computer-predicted B epitopes. Human neurogenic differentiation factor 4 (NeuroD4 or NDF4 UniProtKB —Q9HD90) is a basic helix-loop-helix (bHLH) transcription factor that is involved in neurogenesis and control of neuronal differentiation. Neural navigator protein 2 (NAV2 UniProtKB—Q8IVL1), a voltage gated sodium channel, is also involved in neuronal development, specifically in the development of different sensory organs. NAV2 is expressed in lung, heart, dorsal root ganglia, and Schwann cells in the peripheral nervous system, and in the central nervous system expression is concentrated in the circumventricular organs involved in body-fluid homeostasis. Among other activities, NAV2 affects cell migration and cytoskeletal functions by participating in regulation of microtubule dynamics. Epitope mimics present in Zika may act directly to interfere with cellular targets of these proteins, or may interact indirectly by eliciting autoimmune responses.

In our analysis, we also evaluated the structure of the Zika virus untranslated regions (UTR’s), which are important regulatory sequences located next to the coding sequence of the viral polyprotein. These sequences control both Zika virus replication and protein expression. By evaluating the pattern of genetic changes in the Zika genome as it crossed the Pacific, we identified conserved mutations which predict significant changes in both of these regulatory regions. Within the upstream sequences (5’UTR), one particular change we found was in the sequence flanking part of the Zika genome coding for the start of the Zika polyprotein.

In the genetic control sequences at the opposite end of the Zika genome, a Musashi Binding Element (MBE) consensus sequence was identified. Musashi-1 and -2 are RNA binding proteins involved in the control of RNA translation, and are often implicated in stem cell replication and differentiation in both developing brain and during spermatogenesis. We were able to document mutations in the MBE region which were acquired and then preserved as the virus moved across the Pacific. Biochemical calculations based on previously published models indicate that these conserved mutations may have altered Musashi protein binding affinity to these sequences, which could help explain some differences in the pattern of human cell infection by the outbreak version of the Zika virus.

So, what’s next? At this point, all our analyses are purely computational, and so the next step is to functionally test the significance of the observations and hypotheses using different viruses, cultured cells, and animal models. In addition, this work is being used by Atheric Pharmaceutical and ioGenetics to help design experiments and interpret a wide range of experimental data, including studies involving human serum samples from patients that have been infected with Zika, and from patients who have developed autoimmune diseases after being infected. The computer predictions are being used to help design Zika vaccine candidates as well as methods for safety testing those candidates. The discoveries from this computer modeling and the analyses are even helping researchers interpret the importance of various mechanisms of action associated with different drugs which inhibit Zika virus replication, and which may be acting in part via Wnt-beta catenin and Notch/Numb signaling — both of which are associated with Musashi regulation.

To read the full study, click here.

Adriano de Bernardi Schneider is a Brazilian biologist and researcher who has been involved in the fight against the Zika virus in the United States since the last epidemic in Brazil in 2015. He acquired a Master of Science degree in Crop Science from the Federal University of Rio Grande do Sul, Brazil in 2012. He is currently pursuing his PhD in Bioinformatics and Computational Biology at UNC Charlotte. His research focuses on evolutionary biology of arboviruses, mainly Zika and Chikungunya viruses. In 2016, he published several studies that have helped advance the understanding of Zika virus evolution. In that same year he, in a joint effort with researchers from Boston, won a series of awards, such as the International Development Innovation Network (IDIN) — MIT ­– Awardee of Microgrant, for the creation of a mechanism to combat mosquitoes in South America.