The integration of molecular tools into veterinary and spatial epidemiology

Abstract

At the interface of molecular biology and epidemiology, the emerging discipline of molecular epidemiology offers unique opportunities to advance the study of diseases through the investigation of infectious agents at the molecular level. Molecular tools can increase our understanding of the factors that shape the spatial and temporal distribution of pathogens and disease. Both spatial and molecular aspects have always been important to the field of infectious disease epidemiology, but recently news tools have been developed which increase our ability to consider both elements within a common framework. This enables the epidemiologist to make inferences about disease patterns in space and time. This paper introduces some basic concepts of molecular epidemiology in a veterinary context and illustrates the application of molecular tools at a range of spatio-temporal scales. Case studies – a multi-state outbreak of Serratia mastitis, a national control program for campylobacteriosis, and evolution of foot-and-mouth-disease viruses – are used to demonstrate the importance of considering molecular aspects in modern epidemiological studies. The discipline of molecular epidemiology is in its infancy and our contribution aims to promote awareness, understanding and uptake of molecular epidemiology in veterinary science.

Introduction

Molecular epidemiology is an inherently interdisciplinary approach to the study of health and disease in human and animal populations. It builds on disciplines such as molecular biology, epidemiology, and population genetics (see glossary in Table 1 for an explanation of terms). Molecular epidemiology is defined as the study of the distribution and determinants of disease – either infectious or non-infectious – through the use of molecular biology methods (Riley, 2004). In this paper we focus on infectious diseases of animals and humans. In general the pathogenesis and epidemiology of diseases may differ both between and within species of viruses, bacteria and parasites (Zadoks and Schukken, 2006). Molecular measures, as techniques of refinement, can offer high resolution answers in relation to questions on disease causation that go beyond the species level and, for infectious diseases, provide insight that is not available with traditional culture methods or species level identification of bacteria, viruses, or parasites (McMichael, 1994, Zadoks and Schukken, 2006).

Molecular tools are becoming more widely available to epidemiologists and offer powerful opportunities to increase our understanding of the epidemiology of important pathogens affecting human and animal health. Technologies to generate molecular typing data and bioinformatic tools to analyse such data are developing rapidly. Early molecular methods mostly used DNA amplification by means of polymerase chain reaction (PCR) or enzymatic DNA restriction to generate DNA fragments that were subsequently separated on gels and visualized as banding patterns (Struelens, 1996, Tenover et al., 1995). These methods, in particular random amplified polymorphic DNA (RAPD) typing and pulsed-field gel electrophoresis (PFGE), were very well suited for comparative analysis of isolates that were collected over short periods of time and a small area. They performed well with regards to convenience criteria such as cost, turn-around time and ease of use. However, performance criteria such as typeability, reproducibility and discriminatory power were variable (Struelens and ESGEM, 1996). In particular, limited reproducibility of banding patterns precluded widespread use as standardized typing methods, possibly with the exception of PFGE and ribotyping in studies of foodborne pathogens (Batt, 1997, Swaminathan et al., 2006). Currently, sequencing of RNA or DNA is routinely used for molecular typing. Sequence data can be available for whole genomes or selected areas, such as specific genes or repetitive elements. A major advantage of sequence data is that it is unambiguous, and can easily be stored and exchanged. Sequence-based methods are library typing methods: every result has universal meaning. In contrast, in the comparative typing methods band sizes need to be expressed relative to each other (Struelens et al., 1998). Many sequence-based typing methods and hybrids of banding pattern and sequences based methods exist, and it would be far beyond the scope of this paper to discuss all of them. The field continues to evolve rapidly, which is illustrated by next generation sequence technology that allows for the assembly of entire microbial genome sequences in a matter of days (Metzker, 2010).

The capacity to type a large number of loci in the same pathogen combined with the development of analytical tools to interpret these genotypes has revolutionised molecular epidemiology (Archie et al., 2009). This is particularly well illustrated by work on RNA viruses (Lemey et al., 2009) and the successful application of multi locus sequence typing (MLST) to a wide and growing range of bacterial pathogens (Urwin and Maiden, 2003). In parallel with MLST and other typing schemes for a growing number of bacterial and protozoal pathogens, several novel analytical tools have been developed to analyse and make inference from sequence data. Examples include STUCTURE (Didelot and Falush, 2007), Clonal Frame (Pritchard et al., 2000), AMOVA (Excoffier et al., 2005), minimum spanning trees (MSTs) (Spratt et al., 2004), the island model (Wilson et al., 2008) and a Bayesian framework for integrated analysis of molecular, spatial and temporal information on infectious agents (Lemey et al., 2009). The development of these analytical approaches to molecular data is rapid and driven by forces such as the availability of molecular data from surveillance systems, which need to be incorporated into epidemiological models and investigations; developments in bioinformatics and evolutionary genetics; and the increasing availability of whole genome sequences.

The value of incorporating evolutionary information into epidemiological studies is illustrated by the emergence of phylodynamics and phylogeography. Where the former primarily focuses on the integration of immunodynamics, epidemiology and evolutionary biology (Grenfell et al., 2004), the latter specifically combines epidemiological, evolutionary and spatial information to understand endemic and epidemic dynamics (Lemey et al., 2009). As molecular tools become more integrated into epidemiology, their application to spatio-temporal studies of disease is likely to increase and lead to an improved understanding of key epidemiological factors such as the interaction between local persistence of pathogens and disease dynamics in time and space (Holmes and Grenfell, 2009).

In this paper, we present three case studies illustrating the application of molecular tools in veterinary epidemiology at a range of spatial and temporal scales, molecular resolution, and analytical complexity. First we present a case study of the integrated use of molecular biology and epidemiological approaches to identify the cause of a multistate outbreak of Serratia mastitis in USA dairy cattle using a comparative typing method. Next, we discuss an enhanced surveillance study over multiple years of the multihost pathogen Campylobacter jejuni using library typing. This example illustrates how molecular epidemiology can be utilized to address epidemiological and public health questions and how spatial and molecular analysis can be combined to elucidate transmission pathways for an endemic disease. Finally we present whole genome based studies and theoretical approaches to the molecular evolution of foot-and-mouth disease (FMD) virus (FMDV) over multiple decades and continents to demonstrate the importance of integrating evolutionary and epidemiological approaches into our understanding of disease transmission. We argue that the discipline of molecular epidemiology is still in its infancy and hope that our contribution will demonstrate the value of molecular tools in addressing epidemiological problems at different spatio-temporal scales.