Chlamydia trachomatis in the Age of the Genome: Application of Molecular Genotyping to Improve Our Understanding of the Immunopathogenesis of Chlamydia Genital Tract Disease

The rapid advances in deoxyribonucleic acid (DNA) sequencing technology over the last few years mean that we now stand at the brink of a deluge of information. Will we sink or swim? In this edition, Christerson et al.1 attempt to investigate how their in-house molecular typing system for Chlamydia trachomatis can be used to advance our understanding of the immunobiology of chlamydia genital tract infection in human beings. They use a molecular typing scheme and apply this to 70 isolates from patients with and without disease. No association between their molecular typing profiles and symptomatology was identified, and they conclude that we need to study both bacterial genetics as well as host immunogenetics to better understand the clinical course of infection (Christerson et al.1). Given the study size and the problems in accurately characterizing disease status (phenotype) and the approach to interpreting the observed genetic variation, we would argue that this is not surprising. In this editorial, we consider what type of molecular typing data and epidemiologic studies will be required to determine the relationship between C. trachomatis genotype, human genotype, and disease outcome. Developing a robust typing scheme for distinguishing chlamydia strains is the first step toward understanding the population genetic structure of this organism. Presently, we have a number of potentially workable typing schemes and a small number of reasonably well-characterized isolates.2 Multilocus Sequence Typing (MLST) is a method of characterizing genetic variation between isolates.3 Typically, schemes are based on 7 housekeeping genes and are highly discriminatory, they are designed to investigate evolutionary origins of bacteria. Online databases for >20 bacterial species have been created, representing thousands of isolates (www.mlst.net). Whole-genome sequencing may eventually supplant MLST, but presently MLST remains inexpensive and more practical for most laboratories and offers the most accurate and portable method of typing bacterial isolates. There are 4 published approaches to molecular typing schemes for chlamydia which address the problem of categorizing strains of infection in different ways. These techniques include an ompA variable number tandem repeats typing scheme,4 an MLST scheme on the basis of 7 housekeeping genes designed to elucidate overall population structure of the whole chlamydiae family and a scheme based on carefully selected housekeeping genes.5,6 By contrast, the scheme used by Christerson et al.1 is based on variation at 5 genes that have been chosen to ensure high resolution when asserting phylogenetic difference.7 Summary of existing typing schemes is given in Table 1. If such approaches are to be applied more widely, we need to reach a consensus on what the most useful markers are and to create an online database for Chlamydia trachomatis. Efforts have been made to develop such a system8; however, the molecular typing schemes available presently have been based on “best guesses,” using very limited amounts of genome sequence data. These approaches are not universal typing systems, but designed for different and specific applications and none have been subjected to rigorous independent evaluation. A precise, universal typing system based on single nucleotide polymorphisms (SNPs) is likely to be the ultimate tool, although the very small size of the chlamydial genome coupled to rapid advances in genome amplification and genome sequencing technology mean that it may soon be possible to sequence the complete C. trachomatis genome directly from clinical samples.TABLE 1: Summary of Chlamydia trachomatis Typing SchemesMLST and other sequence typing scheme data can be easily analyzed using eBURST to visualize recent evolutionary relationships between bacterial isolates, provided a significant number of markers are included9 (www.mlst.net). We show the eBURST population snapshot diagram in Figure 1 for the unique isolates described in Christerson et al.1 (including those from Justrand et al.10), highlighting sequence types (STs) common to both the samples in pink (using the “comparative” option with a group definition of 4/5 loci shared). In contrast to the findings reported by Christerson et al.,1 the eBURST diagram shows that the Swedish samples are distributed across the Dutch samples, even given the small number of samples. Furthermore 5 of a total of 60 unique STs (19 Swedish and 46 Dutch) were found in both the samples, which is unexpected if this system is designed for identifying very recent transmission.Figure 1.: Distribution of 60 unique STs of Chlamydia trachomatis from Sweden and the Netherlands (data derived from Christerson et al.1 and Jurstrand et al.10). There are 60 unique STs, coloured according to study: Dutch sample only (41 unique STs) are coloured black, Swedish sample only (14 unique STs) green, and both (5 unique STs) pink. Five STs present in both the samples have been labeled according to the Dutch nomenclature (Christerson et al.1) for ease of representation. They are as follows: Dutch (Swedish): 12 (3, 743, 3, 888), 30 (3, 813), 82 (4, 667, 4, 822), 100 (4, 376), and 133 (4, 368).Substantive genetic and phenotypic differences have emerged for the 2 major C. trachomatis pathobiotypes associated with either trachoma or sexually transmitted diseases, but differences within the sexually transmitted disease group have not yielded reliable disease severity attributes.11 Caldwell et al. demonstrated that all ocular trachoma isolates tested have inactivating mutations in the tryptophan synthase gene, whereas all genital isolates encode a functional enzyme.12 Moreover, functional enzyme activity was directly correlated to IFN-γ resistance through an indole rescue mechanism.12 Hence, a strong selective pressure exists for genital strains to maintain a functional synthase. They speculate that this relationship involves the production of indole by components of the vaginal microbial flora, allowing chlamydiae to escape IFN-γ-mediated eradication and thus establish persistent infection.12 Lactobacilli do not produce indole but many of the other microorganisms present in the vaginal flora do.12 Many of these microbes are associated with bacterial vaginosis (BV) in which they are present in much higher numbers than in the normal, Lactobacillus-dominant vaginal flora.12–14 However, it remains unclear how indole production by BV-associated microorganisms might influence the disease. The microorganisms associated with BV have long been linked with pelvic inflammatory disease (PID) but how they interact with chlamydia in producing PID remains unknown and although BV may be associated with the acquisition of chlamydia, the relationship is complex as it is probably influenced by the composition of the microbiologic flora.13,14 Although indole production may play a role, there are other potential mechanisms including the production of mucinases by BV-associated microorganisms.15–17 A number of other candidate C. trachomatis virulence factors have been identified, including the polymorphic outer membrane autotransporter family of proteins, the putative large cytotoxin, type III secretion effectors, stress response proteins, and proteins or other regulatory factors produced by the cryptic plasmid.11 To investigate the association of potential virulence factors with disease, clinical samples will not only need to include information on the genotype of the infecting bacteria but also the presence of other sexually transmitted bacteria including BV,12 host symptoms, disease phenotype, genotype, and sociodemographic or environmental factors. Even with this information, there remain technical challenges in fully understanding the complex interaction between host, pathogen, and disease phenotype. Chlamydia often occurs asymptomatically, effective treatment is available, and clinical outcomes such as tubal factor infertility may occur years after the initial infection has resolved.18,19 Symptoms of chlamydia are nonspecific and may have other causes such as BV or candidiasis.19 Even the clinical diagnosis of PID lacks sensitivity and specificity.19 Urethritis in men and possibly cervicitis in women can be more accurately characterized, however, presently these conditions can only be reliably diagnosed using invasive sampling.19–22 Failure to characterize clinical samples according to disease status is likely to attenuate the effect size and increase the numbers required to demonstrate an association. Ultimately, we wish to draw together information from molecular typing and clinical epidemiology to better understand the natural history of chlamydia and the specific relationship between bacterial and host genetic factors and disease severity, to be able to more accurately predict an individual's risk of disease progression. There are numerous regions of the human genome which potentially interact with the immune response23 and an unknown number which could potentially affect chlamydial attachment, entry, and replication in vivo, which could influence susceptibility and disease after chlamydia exposure. It has been suggested for several years that genetic or immunologic variation between human hosts may play a role in the progression of chlamydial disease.24 Despite examining human leukocyte antigen class I and II variants and functional polymorphisms in cytokine and cellular receptor genes, in relation to chlamydia-related outcomes, no specific alleles or polymorphisms have been identified that reliably predict pathology.25,26 Perhaps, this is not surprising given the complexity and redundancy in biologic pathways which allow multiple genetic influences to affect outcome.23 Moreover, most genetic variations are SNP which do not abrogate production of the protein or generate a dysfunctional protein.23 Rather they are found in putative regulatory areas and affect the rate of transcription or the stability of the mRNA product, creating a more subtle phenotype.23 There is now the technology to investigate the association of 100,000s of SNPs with disease and even whole genome analysis.27,28 But as the strength of associations with common genetic variation are often weak, it is likely that (in the absence of rare, large effects or nonadditive gene * gene/gene * environment interaction) many thousands of characterized specimens are required to demonstrate this. With strong associations smaller studies may suffice but this will be dependent on the frequency of the genetic variation in the population. For example, a study genotyping over 700 HIV-1 positive and negative individuals demonstrated that the uncommon chemokine receptor CCR-5 mutant (approximately 1% of whites) reduces susceptibility to infection with macrophage tropic HIV-1 strains.29 Typing of chlamydia has lagged significantly behind that of other microorganisms such as Neisseria meningitidis or Staphylococcus aureus, so we have a clear opportunity to learn the lessons from earlier work. Eventually, we may be able to relate specific bacterial and host genotypes with disease outcome. Christerson et al.1 have attempted to go beyond “see what's there” and relate their STs to symptoms of chlamydial infection, but this is a very tenuous first step on a long road. Given the likely numbers required, we believe multicentre collaborative studies will be needed using an agreed common protocol, thus ensuring the collection of accurately characterized clinical specimens which can then be rigorously interrogated. To understand fully the cause of the conditions associated with chlamydia, we need multidisciplinary expertise, carefully designed studies, and appropriately designed analyses.