Genomics Should Inform Appropriate Analysis of Taxonomy and Pathogenesis of Rickettsia

(See the Major Article by Paddock et al on pages 849–58.)

In this issue of The Journal of Infectious Diseases (JID), Paddock et al have reported the results of a remarkable study. They have demonstrated that a genetically distinct set of rickettsial isolates from ticks that correlate with patients with a clinically distinct disease, a distinct ecology reflecting tick transmission, and a distinct geographic distribution are a separate clade of Rickettsia rickettsii. The disease, Pacific Coast tick fever, had been recognized as a spotted fever group rickettsiosis. It is autochthonous to California and has a high incidence of eschars, low occurrence of rash, and less severity than Rocky Mountain spotted fever. Their meticulous genomic studies reveal that it is too closely related to R rickettsii to justify creating a separate species but distinct enough to justify establishment of a novel subspecies. This is a useful contribution to the field of infectious diseases.

There is a great spectrum of severity of spotted fever group rickettsioses that is poorly revealed by the burgeoning myriad of named Rickettsia species. The complex nature of spotted fever group rickettsioses has made their diagnosis challenging and often uncertain for clinicians. In the Western Hemisphere, the situation is more comprehensible than in Eurasia where the number of proposed named species is overwhelming for practicing physicians, including those specializing in infectious diseases [1]. Performing a PubMed search with the term for proposed new species names, “Candidatus Rickettsia,” yielded 854 articles. Review of only the first 85 entries provided the names of 30 proposed unique species of Rickettsia, 24 of which are from the Eastern Hemisphere. In the Americas the pathogenic species, R rickettsii and R parkeri, cause diseases that are clinically distinctive and are transmitted by ecologically distinct ticks. In the United States, R amblyommatis and R tillamookensis, which are geographically and ecologically distinct, are associated with a significant incidence of seropositivity among healthy persons in their geographic distributions, perhaps owing to subclinical infections. There is no evidence that other American rickettsiae—R montanensis, R rhipicephali, or R peacockii—cause disease. Other than a few cases of human infection by R massiliae, there is little evidence of introduction of spotted fever group rickettsial pathogens from the Eastern Hemisphere into the Western Hemisphere, raising questions about the genetic analyses underpinning taxonomic placement.

This situation results from too few physicians engaged in investigating rickettsial diseases and too little emphasis by microbiologists in elucidating pathogenic disease mechanisms and the genetic basis for rickettsial virulence. The lack of a taxonomic system that provides this information has arisen because of the difficulty in studying obligately intracellular bacteria that require biological safety level 3 handling. Similarly, it has been assumed that standard taxonomic methods for other prokaryotes do not effectively apply to rickettsiae. Yet gene sequencing is easy and is the basis of modern phylogeny and taxonomy. Thus, resulting schemes created gene sequence rickettsial clades based on very limited depths of genomic analysis, often 6 or fewer incomplete gene sequences [2–4]. This created extensive divisions and proposed individual species without consideration of substantial phenotypic characters, including the lack of evidence for pathogenic or virulence mechanisms, or caution given limited clinical findings based on small numbers of cases [5]. The lack of defined virulence mechanisms also meant that key features required for clinical utility could not be considered and are not engendered with the naming conventions. On review, the complexity of species within the spotted fever group in the Rickettsia genus seems, in part, to be a result of the drive for pathogen discovery. Paradoxically, this is uncoupled from evidence of pathogenicity and appears to assume that all rickettsiae are pathogenic or potentially pathogenic.

In this issue of JID, the taxonomic clarification of the organisms known as Rickettsia strain 364D as R rickettsii subsp californica is a remarkably significant contribution of new knowledge that begins to address many of the problems that exist in current rickettsial taxonomy. The genomic identification as R rickettsii avoids the burden of adding yet another unwarranted species name. Moreover, with the critical analysis of R rickettsii subsp californica comes a new recognition that thorough genomic analysis can provide not only scientific evidence for evolutionary processes, but also evidence for how diverging processes sufficiently distinguish, explain, and raise awareness for human pathogenicity—integration sorely missing from most studies that solely seek to identify “novel” rickettsial species. The authors carefully elucidate the genome sequence with annotations for taxonomic placement using digital DNA-DNA hybridization and average nucleotide identities (orthoANI), and compare genomes by gene synteny, indels, and predicted amino acid sequences of the major surface antigens involved with mouse serotyping, to obtain an accurate placement. Overall, this is the enlarging picture of R rickettsii, reproducibly modified from severely pathogenic strains across all 364D isolates, and suggests some unique properties that diminish but do not result in loss of human pathogenicity.

Rickettsiology has made little progress in understanding the pathogenic mechanisms of the associated diseases and their clinical manifestations [6], part of the key phenotypic characters useful in taxonomy. Historically, Rickettsia species were delineated by “mouse serotyping,” which became the basis for species delineation. The availability of molecular tools, including gene sequencing, provided an alternative approach. Coupled with the mouse serotyping, this was used to create a mathematical formula using sequences of 5 rickettsial genes (gene sequence–based criteria) to predict whether an unidentified Rickettsia isolate or clinical/environmental sample was likely to be genetically different enough from species defined by mouse serotyping, to be considered a “new” species [2]. This strategy and related multilocus sequencing approaches yielded an extraordinarily large number of Candidatus Rickettsia species substantially lacking phenotypic characters and poorly or not aligned with clinical disease characteristics. With whole genome sequencing, a new approach toward taxonomic classification was proposed [7]. However, the veracity of the proposal was predicated on the species “standards” established by mouse serotyping as modified by newly added “species” identified using gene sequence–based approaches. The result, defined by orthoANI analysis using these criteria [8], was predictably similar to the existing gene sequence–based approaches.

To this end, an analysis of Alphaproteobacteria type strain whole genomes identifies spotted fever Rickettsia genomes that, when subject to accepted species and subspecies criteria delineations, demonstrate that a significant revision is needed [9]. Using the principles applied to improving taxonomy in the Alphaproteobacteria class results in a range of emended descriptions and new subspecies. This further predicts a reclassification of accepted Rickettsia species into a significantly collapsed structure within which existing bacteria of both known and unknown pathogenicity are included. A phylogenetic tree of 72 selected Rickettsia species with whole genome sequences available in the National Center for Biotechnology Information Genome, analyzed using defaults as for Alphaproteobacteria taxonomy, resulted in resolution of 15 species, including a large single-species cluster encompassing R rickettsii, Candidatus R philippii (the subject of the current editorial), R peacockii, R africae, R slovaca, R honei, R parkeri, R sibirica, R conorii, R japonica, R fourneri, R montanensis, and R gravesii as new subspecies clades. Yet, this approach still does not adequately accommodate human pathogenicity as a key phenotypic attribute. A notable example of the genetic basis for pathogenicity is the demonstration that a virulent strain of R rickettsii differs from an attenuated strain and from nonpathogenic R montanensis in the kinetics and magnitude of interferon beta secretion by infected endothelial cells determined by the rickettsial proteins RARP2 and RapL [10]. Endothelial cells infected with R montanensis undergo immediate lysis, endothelial cells infected with attenuated R rickettsii undergo early lysis, and endothelial cells infected with virulent R rickettsii undergo delayed death by apoptosis. The result is greater growth of virulent R rickettsii in endothelial cells because the cells remain alive longer. Accordingly, in the projected alternative taxonomic classification using proposed criteria for Alphaproteobacteria [9], R montanensis could be clustered with R rickettsii as subspecies montanensis. Similarly, there was a report of Rocky Mountain spotted fever caused by a “newly recognized spotted fever group rickettsia,” “Rickettsia sp CA6269,” initially named Candidatus R lanei [11], which was established based on partial or complete sequences of 5 to 6 genes [12]. Here, the authors note a lack of consensus criteria for taxonomic classification of Rickettsia species, and urge taxonomic classification also based on a range of new isolates, whole genome sequences, and their phenotypic features, an approach we applaud. The addition of a complete genome sequence for “Rickettsia sp CA6269” would greatly enhance the opportunity to connect genotype with phenotype. However, for rickettsial classification, in far too many instances authors have used gene or multilocus sequence typing approaches predicated on limited numbers of sequenced genes, or proposed the use of a whole genome taxonomic classification standardized by the gene sequence–based approaches [2, 7]. We argue that this is fundamentally flawed given the very close relationship of “Rickettsia sp CA6269” and “Rickettsia strain 364D” to R rickettsii and the proposed emendations recommended by both Hördt et al and Chung et al [9, 13]. While this will not alleviate the lack of understanding of the biological basis for rickettsial virulence or severe disease, it will provide an understandable taxonomic structure and firm basis for determining the requirement for potentially aggressive care and specific antimicrobial approaches that confound many current healthcare practitioners.

Is any of this of real consequence for science and human health? Currently in the northern states of Mexico bordering the United States, there have been thousands of cases of Rocky Mountain spotted fever with hundreds of deaths, a high proportion of them affecting children, with a case fatality rate of 40% [14]. The case fatality rate of infection with R rickettsii in Brazil is 50% despite the ready availability of doxycycline treatment. The case fatality rate of Rocky Mountain spotted fever in the United States in the preantibiotic era was 23% and is only 4% today. Genetically there are no identified differences that separate the strains in Mexico and South America from strains in the United States, much less determination of the mechanism of increased pathogenicity. Further elucidation of why some rickettsiae cause disease, and their pathogenic mechanisms, would complement such carefully elucidated characterization of strains of Rickettsia with distinct clinical and ecological differences as done in the outstanding report by Paddock et al.

Notes

Disclaimer. The opinions expressed herein are those of the author(s) and are not necessarily representative of those of the Uniformed Services University of the Health Sciences; the United States Department of Defense; or the United States Army, Navy, or Air Force.

References

Author notes