Francesca L. Short and Janet I. MacInnes As noted in Chapter 1, understanding bacterial pathogenesis starts with considering the notion of disease. It is important to appreciate that “diseases,” and the organisms that cause them, must always be considered in context. The classic way of thinking of this is the host–environment–pathogen triangle (Figure 4.1). Host characteristics such as breed/genotype, age, immune status, and route of infection; environment factors such as temperature, air quality, organic burden, “stresses” like transport; and pathogen properties including species, strain/genotype, number of viable organisms, and presence and expression of specific virulence genes can all play a role in the development of disease. Depending on these factors, bacteria can behave as benign commensals, be opportunistic pathogens, or act as primary pathogens. However, rather than trying to assign a specific label to a particular organism, it can be useful to consider its pathogenic potential as proposed by Arturo Casadevall, where morbidity, mortality, communicability, and time to infection can be taken into account (Casadevall 2017). The mere presence of an organism in a diseased animal does not prove that it is the causative agent of the observed clinical signs. The traditional way of demonstrating whether a particular organism is the etiological agent of the observed pathology was to test Koch’s postulates. Roughly paraphrased, Koch’s postulates state that a microorganism is the cause of a disease if: (i) it is present in large numbers in all animals suffering from the disease but not in healthy animals; (ii) it is possible to isolate the organism from a diseased animal; (iii) the disease can be recreated in another animal following isolation and culture; and, (iv) the experimentally infected animal presents with the same disease and it is possible to reisolate the organism (Figure 4.2). Although it was soon recognized by Koch and others that not all disease‐causing organisms follow Koch’s postulates, it is still arguably the gold standard for the study of bacterial pathogens. Fulfilling Koch’s postulates is most rigorously tested in the natural host, but, for ethical and practical reasons, model systems are often used. Rodents, especially mice, are widely (and not always wisely) employed to model diseases of a very wide range of species. Juveniles are often used as they may be easier to handle and are often more susceptible to infection. Seeder infection systems, where test animals are put in contact with naturally or experimentally infected animals, most closely reproduce natural infection. However, with many bacterial pathogens, exposing animals to cofactors such as viral pathogens prior to challenge, introducing relatively high concentrations of the organism, or introducing bacteria via a route that avoids some of the host defenses (e.g. intratracheal instillation) may be needed to produce clinical signs (Word et al. 2020). With all of these experiments, animal welfare concerns should be a priority. Getting the greatest amount of statistically significant data with the smallest number of animals requires considerable thought. In addition to bacterial counts or other readouts such as gene expression, in most experiments, clinical signs, gross pathology and histopathology, and host gene expression can be used to gain greater insights into pathogenic processes. Recent developments in the refinement of animal use in research, and alternative approaches that reduce the need for such models, are described in detail below. Figure 4.1 Host–environment–pathogen triangle. The ability of a bacterial species to cause disease depends on properties of the strain (e.g. the ability to produce toxins) and host factors, such as the presence of protective antibodies. In addition to a series of complex interactions between the host and the pathogen, environmental factors can also help to tip the balance between health and disease. Source: Figure created with Biorender. As discussed in Chapter 1, the main “steps” in pathogenesis include initial contact and finding a niche in the host (colonization, invasion), acquiring nutrients and replicating, subverting and/or evading host defenses, causing damage, and spreading (within and without). These processes are dynamic and can be influenced by a large number of host, environment, and pathogenic factors. The main classes of virulence factors involved in these processes include adhesins (fimbrial and nonfimbrial), nutrient acquisition systems (e.g. iron or sugar uptake systems), cell‐surface molecules (e.g. capsule, lipopolysaccharide, lipoteichoic acid, outer‐membrane proteins), and toxins. It should be noted that many virulence factors play multiple roles. For example, in addition to playing a role in evading host defenses, bacterial capsules may also be involved in attachment as well as survival in the environment, which may affect transmission. Figure 4.2 Koch’s postulates and molecular Koch’s postulates. (a) Koch’s third postulate (the ability of a putative pathogen to cause damage) can be used to demonstrate the etiological agent of many diseases. (b) The role of many key virulence factors can be demonstrated by showing that the wild‐type and complemented wild‐type organism are able to cause disease, while the isogenic knockout strains are less able to do so. Source: Figure created with Biorender. Concentration of the virulence factor (affected by expression levels but also cell numbers) can also have an effect. For example, the Mannheimia haemolytica Lkt toxin causes cell lysis at high concentrations but induces apoptosis at lower levels. Characterizing the role and importance of specific virulence factors can be confounded by the fact that many systems are “redundant.” For example, Clostridium perfringens strains may encode more than a dozen toxins in various combinations, while some Escherichia coli strains have multiple adhesins. In the presence of such duplications, inactivation of a single gene may show no effect on pathogenesis even though the gene product plays some role. Often, multiple mutations or heterologous expression may be required to convincingly show virulence roles in such circumstances. Identifying and characterizing virulence factors is further complicated by the fact that there are multiple regulatory systems that alter gene expression in response to a myriad of host and environmental factors. Accordingly, their importance may differ depending on the specific host and/or environmental condition. Many different and complementary approaches can be used to identify bacterial genes that are important for infection. As noted below, some virulence factors might be predicted by virtue of an obvious phenotype such as strong hemolytic or urease activities; others might be inferred by virtue of their similarity to a known virulence factor in a well‐characterized pathogen. Regardless of the initial method for identification, fulfilling “molecular Koch’s postulates” (Figure 4.2) has become the gold standard for the confirmation of virulence factor genes. Proposed by Stanley Falkow in 1988, the postulates state that: And alternatively: Noting that genetic systems were not available for many species, Falkow suggested that the induction of specific antibody to a defined gene product should also be an acceptable alternative to the molecular Koch’s postulates. Like the traditional Koch’s postulates, molecular Koch’s postulates have some limitations (see below), but nevertheless provide a very useful approach to understanding bacterial pathogenesis (Falkow 1988, 2004). Many genes with important roles in pathogenesis fall short of fulfilling molecular Koch’s postulates, particularly the presence of the gene only in pathogenic members of a species. These can be genes coding for factors that regulate expression or are required for the activity of “true” virulence factors, or they can encode factors that provide adaptive advantages both inside and outside the host (e.g. polysaccharide capsules). These genes can instead be referred to as virulence‐associated genes (Wassenaar and Gaastra 2001). The majority of experimental approaches described in this chapter are useful for the study of both virulence and virulence‐associated genes, and both gene categories can be useful as targets for new diagnostics or therapeutics. An alternative working definition of a virulence factor, based on Koch’s molecular postulates but less stringent, is one encoded by a gene that reduces fitness in the host when mutated but does not affect growth in vitro. Another limitation of Koch’s molecular postulates is the use of single gene (isogenic) mutations. In early studies, researchers would screen random transposon libraries (e.g. with Tn5) to find mutants with the genes of interest interrupted. This approach had a number of weaknesses (see below). Following these early studies, a variety of methods were developed that allowed the generation of targeted unmarked isogenic strains (e.g. Pei et al. 2007; Bossé et al. 2009; Holden et al. 2020). However, the availability of techniques for targeted mutagenesis remains a limiting factor for studies of some bacterial pathogens. There are increasing pressures to reduce or eliminate the use of vertebrate animals in the study of bacterial pathogens. To achieve this, it is important to perform well‐designed, rigorous experiments, and to use alternative models where possible. Having a testable hypothesis and statistically sound experimental approach is especially important when conducting animal studies (Gosselin 2019). Pilot studies using a small number of animals (around six or fewer) are often done to establish the dose that will cause disease in a significant number of animals. LD50/survival curves typically have a sigmoid shape so starting with relatively high doses (and reducing them in subsequent challenges if necessary) can help to limit the number of animals used. A dose that can reproduce disease/signs in 30–70% of the animals challenged is usually used when testing for the pathogenic potential of a strain or for the role of a putative virulence factor (see below) or to evaluate the efficacy of vaccines or antimicrobials. Although a common metric in the past, death is now not seen as an appropriate endpoint. For animal welfare reasons, body temperature or combined clinical scores are increasingly being used to prompt humane euthanasia rather than having natural death as the endpoint and the data are presented as a percentage survival rather than LD50. The use of well‐trained, double‐blinded observers who systematically collect data from randomized subjects according to an established rubric can help to ensure reliable results are obtained. A growing number of alternatives to traditional infection models are being employed to compare the relative pathogenicity of organisms and to investigate putative virulence factors. Approaches such as the use of ligated gut loops or dermal inoculations still involve the use of animals, but reduce the numbers required (Westerman et al. 2021). Other methods such as signature‐tagged mutagenesis (STM; see below), where multiple mutants can be tested at the same time, also reduce animal numbers. With some bacterial pathogens, it may be possible to use blood cells (e.g. macrophages) or tissue explants from the natural host to provide a measure of the role and importance of some virulence factors, especially toxins and adhesins (Bohl et al. 2021). For the study of intracellular pathogens, gentamycin protection studies can be a powerful tool to study adhesion, invasion, and intracellular replication (Vohra et al. 2019). Organ explants and primary cell cultures have also been successfully used for some studies (Brassard et al. 2001). In addition, immortalized cells lines have been widely used in the study of bacterial pathogenesis (McWhorter et al. 2021). Interestingly, cell lines from the natural host species are not always used, e.g. Vero (African green monkey) cells have been frequently used to study bovine E. coli (Menge 2020). Recently, organoid model systems have been developed to help bridge the gap between cell culture and animal experiments. Organoids are generated from stem cells and mimic the organ development process in vitro, giving rise to self‐organized three‐dimensional structures that recapitulate many features (e.g. cell types, interactions, and topology) of the organ of interest (Kim et al. 2020). Generating and culturing organoids is technically challenging and can be expensive, as the specific growth factors and signals involved in cell differentiation and organ development must be supplied in a precise order to recapitulate the natural development process. In addition, generation of organoids requires stem cells from the animal of interest. Nevertheless, these systems offer great advantages for studying specific details of host‐pathogen interactions without using animals. For example, organoid systems can recapitulate species‐specific interactions that may be absent in surrogate animal hosts (e.g. mice) or be used to assess host genotype effects on pathogen interactions without generating transgenic animals. However, as single‐organ systems, these models cannot recapitulate features of animal infection such as immune system interactions or dissemination from the infection site to other tissues. Organoid systems have been generated for different domestic animals and organ types, and there are limited studies of these organoids in the context of bacterial infections (Beaumont et al. 2021; Kar et al. 2021). However, it is likely that organoid systems will be more widely used in coming years to address specific questions of bacterial pathogenesis in animals. The use of vertebrate animals can be completely eliminated with studies that employ the Caenorhabditis elegans nematode (Paudyal et al. 2019), Drosophila melanogaster fruit fly (Guzman et al. 2021), or larvae of the Galleria mellonella greater wax moth (Dinh et al. 2021). However, considerable care is needed to interpret how closely such models replicate natural infection. Researchers have also tried to study bacterial virulence factors using in vitro conditions that mimic in vivo (host) conditions. For example, studies have been performed in which iron is chelated, or in which the bacterium is exposed to host factors such as low pH, bile, or surfactants (Metcalf and MacInnes 2007; Haines et al. 2015). Regardless of the method employed to study pathogenesis, the strain and the host–model system used can have important effects on reproducibility and interpretation. In addition to the above methods, indirect methods are sometimes used to predict the role of an organism in a particular disease. For example, if populations of animals that have been vaccinated, or which have high levels of antibody, are resistant to a particular infection or there is a high correlation between the presence of an organism and disease, inferences may be drawn about causality. Thoughtful experimental design is essential for researchers to generate robust and objective findings. There is no single ideal approach, as this will vary depending on the experiment type; however, some principles apply broadly. Generally, an experiment will aim to test the effect of an independent variable (e.g. presence of a specific gene) on a dependent variable (e.g. virulence). These are set out in the hypothesis of the experiment. For example, a researcher could set out to test the hypothesis “Pathogenesis of Bacillus anthracis requires the anthrax toxin,” in which pathogenesis is the dependent variable and the presence of anthrax toxin is the independent variable. Using good experiment design strategies such as appropriate controls, replication, and statistical analyses makes it possible to support or reject a hypothesis with high confidence. Although a full discussion of all the factors that contribute to good experimental practice in different scenarios is outside the scope of this chapter, the main principles will be summarized here with reference to a laboratory‐based bacterial phenotypic experiment. Experimental controls are needed to minimize or exclude the effects of extraneous factors on results and are needed for both quantitative and qualitative experiments. Negative controls confirm that the presence of a particular signal is a result of the bacterial phenotype being measured and not due to contamination or error, while positive controls confirm that the phenotype of interest can be detected in the experimental setup being used. For example, a quantitative assay for measuring the amount of capsule produced by a bacterial strain should include a negative control of water or growth medium, a positive control of a well‐characterized bacterial strain known to produce capsule in the growth conditions used, and another positive control of purified capsule (or the specific capsule component detected by the assay). In this case, the two positive controls detect different potential confounding factors – the former will detect any problems in bacterial growth or capsule extraction procedures, while the latter will detect any problems in the chemistry of the capsule measurement assay. Replication provides information on the variability and reproducibility of an observation and is needed to correctly interpret the magnitude and significance of a measured effect. Biological replicates are derived from independent animals, bacterial cultures, or other biologically distinct entities that show variation, while technical replicates are repeated measures of the same biological sample and report variability due to noise associated with particular experimental measures and protocols. Use of biological replicates is essential in microbiological experiments, while technical replicates may or may not be required depending on the experiment. Quantitative experiments need to be analyzed using appropriate statistical tests to determine whether an observed effect is greater than what would be expected by chance (Olsen 2014). There is a wealth of information on which statistical tests are most appropriate in different circumstances. A challenge in experimental microbiological research is that often the number of measured replicates is insufficient to determine whether data are parametric (normally distributed) or non‐parametric. In these cases, prior knowledge about the variable being measured can be useful. For example, measurements like cell size are likely to be approximately normally distributed, while measurements such as viable cell counts and ratios are not. For most experiments, meeting the above principles of good experiment design – including appropriate controls, biological replication, and correct statistical analysis – is achievable. In some cases, for example, when using new or particularly expensive techniques (such as some of those covered in the next section), or rare clinical samples, it may not be possible to meet all these requirements, though results from experiments with such constraints can still be valuable if interpreted carefully and validated using different methodologies. Rigorous research goes beyond well‐designed single experiments (Casadevall and Fang 2016). It will also address the generalizability of a finding (for example, does gene X contribute to phenotype Y in more than one bacterial strain background, or in different growth conditions?), seek to use validation with an independent methodology (for example a finding that expression of gene X increases under antibiotic stress could be measured using both quantitative polymerase chain reaction (PCR) and reporter gene experiments), and interpret the findings in a logical and honest way. The single virulence factor model set out by Molecular Koch’s postulates does not apply to every bacterial pathogen. Indeed, for many bacteria, infectivity depends on many different factors that may tip the probability one way or another. Furthermore, there is often redundancy between virulence factors. For virulence factors and virulence‐associated factors alike, there are many different methods that can be used for their identification, validation, and mechanistic characterization. In particular, advances in next‐generation sequencing have led to a dramatic increase in the ease with which questions of pathogenesis can be addressed at a genome‐wide scale. Though the range of advanced methodologies has exploded in recent years, it is important to remember that, whether a study draws on classical microbiological techniques or a new high‐throughput technology, the concepts underpinning the research are the same. Potential determinants of pathogenesis in bacteria can be identified through three broad approaches: comparison of pathogenic and non‐pathogenic strains, detailed observation of the molecular events that happen during an infection, and unbiased disruption of bacterial genes followed by tracking of their effects. Common experimental methods, and where they fit within these three broad approaches, are summarized in Table 4.1 and illustrated in Figure 4.3. Table 4.1 Three broad approaches to virulence gene discovery.
4
Experimental Approaches to Understanding Pathogenesis
Introduction
Testing Koch’s Postulates
Virulence Factors and Main Steps in Pathogenesis
Molecular Koch’s Postulates
Refining Animal Usage
Experimental Approaches for Virulence Factor Discovery and Validation
Principles of Experimental Design
Approaches for Virulence Factor Discovery
Method
Overview
Developed
Stage
Target
Advantages and limitations
Comparative
Genome‐wide association tests
Whole‐genome sequences are compared across a large collection of isolates, statistical methods identify those associated with pathogenic strains
Accessible since 2010s
Discovery
DNA
Requires a large collection of genomes and reliable metadata about infection type and severity; can identify signals for genes of unknown function
Whole‐genome sequencing and functional annotation
Sequencing of pathogen genome and bioinformatic inference of genes of interest, pathogenicity islands, etc.
Accessible since 2010s
Discovery
DNA
Does not require large genome collection, heavily dependent on accuracy of existing databases for comparison. Difficult to select “typical” strains
Differential hybridization
Looks for presence/absence of target genes in different strains
Early
Discovery
DNA
Phenotype testing
Investigates phenotypic differences between pathogenic and non‐pathogenic strains
Early
Discovery
Phenotype
Accessible, can be labor intensive, requires a strain collection with well‐known pathogenicity differences 
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
Get Clinical Tree app for offline access
