Molecular Studies in Treponema pallidum Evolution: Toward Reality
Kristin N. Harper1, Bret M. Steiner2, Michael Silverman3 and George J. Armelagos4

1 Population Biology, Ecology, and Evolution Department, Emory University, Atlanta, GA
2 Laboratory Research and Reference Branch, Division of Sexually Transmitted Diseases, U.S. Centers for Disease Control and Prevention, Atlanta, GA
3 University of Toronto, ON, Canada
4 Anthropology Department, Emory University, Atlanta, GA


Recently, Mulligan, Norris and Lukehart’s critical commentary, “Molecular Studies in Treponema pallidum Evolution: Toward Clarity?” accompanied the article “The Origin of the Treponematoses: a Phylogenetic Approach” [1]. Most of the criticisms stemmed from misunderstandings about methodology, terminology, and the manner in which evolution works. Here, we address the concerns they raised in their commentary, as we understand them.

Mulligan et al argue that there is no biological basis for transmission route in the T. pallidum subspecies, since there are examples from old textbooks of syphilis being transmitted through non-venereal routes, such as from infants to wet nurses. However, anecdotal evidence of syphilis being transmitted via non-sexual routes is not evidence that biological adaptations to sexual transmission are absent. For example, there have been documented cases of sleeping sickness and typhoid being sexually transmitted [2,3]. Would Mulligan and coworkers also argue that there is no biological basis for vector transmission in the former or fecal-oral transmission in the latter? They appear to be espousing the Unitarian hypothesis of treponemal evolution: that the mode of transmission in the treponemes is defined primarily by “opportunity.” However, numerous studies, including ours [1] and previous articles by Mulligan and Lukehart [4-6] indicate that the subspecies are genetically distinct and have evolved along divergent paths. In fact, in late 2006, these researchers argued that “molecular data suggest that the three subspecies are legitimately classified as distinct entities” [4]. Since subsequent research has added more data to support this belief, it is curious that the authors now argue against it.

Mulligan et al argue that several subsp. pertenue, or yaws-causing, strains bear the molecular signature of subsp. pallidum, or syphilis-causing, strains and that by grouping these strains with subsp. pallidum strains we have created a phylogeny biased in favor of genetically distinct subspecies. The two strains to which they refer, Haiti B and Madras, were obtained from patients with classical yaws in the mid-20th century. However, recent laboratory work, much of it performed in the Lukehart lab, has demonstrated that they are identical to syphilis-causing strains at all polymorphic sites examined to date [1,6-8]. The most likely explanation for these two aberrant strains is that the original yaws-causing strains were swapped for syphilis strains in the laboratory. This is the hypothesis that the researcher who first handled the Madras strain favored. After noting that syphilis-causing strains were sent in the same shipment as the subsp. pertenue strain, he wrote: “In my hands, Madras was identical to syphilis, including immunity. I was unable to check complete linear descent in Mr. Hanson’s files. Mr. Hanson [the source of the strain] also questioned the strain” [9]. The Haiti B strain was also probably replaced with a laboratory subsp. pallidum strain at some point. Originally, infection with this strain resulted in clinical signs consistent with classic yaws infection in a rabbit model [10]. However, for some time this strain has caused clinical signs identical to those of subsp. pallidum, rather than subsp. pertenue, strains in infected rabbits. Mulligan and colleagues’ belief that Haiti B can infect hamsters, and thus cannot be a syphilis strain, is incorrect. Schell et al showed in 1982 that the standard laboratory strain of syphilis, Nichols, grows in hamsters, demonstrating that hamsters can be infected with subsp. pallidum [11]. In this study, they also demonstrated that infection with the Nichols strain conferred immunity against the Haiti B strain, but not against the Bosnia strain of subsp. endemicum, which causes endemic syphilis or bejel. These data are consistent with Haiti B having been substituted with a subsp. pallidum strain by this time. Unfortunately, strain mishaps are not uncommon. As Dr. Lukehart knows, a subsp. pallidum strain was accidentally substituted for the Gauthier subsp. pertenue strain in the laboratory several years ago, and it took some time to realize there had been an accident [12]. Therefore, there is no reason to presume that the strains Haiti B and Madras cause yaws but bear the molecular signature of subsp. pallidum strains.

Our claim for a New World origin of venereal syphilis is based on sequence similarity between the Guyana subsp. pertenue samples and the subsp. pallidum strains. Our sequence comparison consisted of 17 SNPs that differentiated subsp. pallidum from subsp. pertenue and endemicum strains. The Guyana subsp. pertenue samples resembled syphilis-causing strains at four of these SNPs; at most, other non-venereal strains were identical at only one. Mulligan et al object that three of these identical SNPs come from a single gene, tprI. They neglect to mention that the fourth SNP, a synonymous substitution in the gpd gene, is on the other side of the genome and tells the same story. They argue that, "the tprI locus is atypical of the treponemal genome and, thus, not the best choice when trying to resolve the decades-old debate concerning the origin of venereal syphilis." While we clearly explained that the tpr genes may be atypical of the genome in our article, and thus emphasized the importance of the supporting SNP from the gpd gene, in their 2006 paper Drs. Mulligan and Lukehart created phylogenies entirely from the tpr genes and used them to make inferences about the treponemes’ evolutionary past [4]. Much of their discussion in that article centered on the tprI gene in particular, in which they asserted that recombination did not play an important role and in which substitutions “were assumed to evolve in a clocklike manner.” In the commentary, Mulligan et al state that the CDC-1 and CDC-2 strains of subsp. pertenue should have identical genetic signatures at the tprI gene, since they were collected in West Africa at similar times. For this reason, they seem to have changed their mind regarding the suitability of this gene for evolutionary inference. However, if yaws did indeed originate in Africa, then we would expect to find a greater amount of variation in African strains than in those from other locations.


Mulligan and colleagues make the point that the two New World subsp. pertenue samples were taken from one location. In fact, although these two samples were both taken from Guyana, the villages involved are at a significant distance from each other. Travel between them requires a combination of bush plane and boat. The communities therefore do not mix, and, therefore, these are truly two separate samplings.

Mulligan and colleagues’ discussion of our phylogeny is riddled with errors in terminology and misunderstanding of basic evolutionary concepts. This impairs their interpretation of our study and makes it difficult to understand their criticism. They frequently refer to the study of genes that are subject to natural selection or recombination as violating the “assumptions of phylogenetic analysis.” There are no universal assumptions of phylogenetic analysis. The assumptions of each phylogeny depend on the methods used and the problem addressed. Drs. Mulligan and Lukehart must be aware of this fact, having published a paper full of phylogenies based on the members of the recombining tpr gene family [4]. We assume that what they refer to is that in genes subject to these processes, the molecular clock cannot be used. Similarly, Mulligan et al state that because the “strains were collected contemporaneously, the branch lengths should all be approximately equivalent, since a phylogeny reflects only mutational evolution.” They also write that selection violates the “evolution-by-mutation-only assumption of a phylogenetic analysis.” Again, they confuse phylogenetic analysis and the molecular clock, while also asserting that only neutral substitutions result from mutation. We presume that what they meant to say is that the use of the molecular clock assumes that neutral substitutions accumulate in a clockwork manner, but substitutions subject to positive selection may not.

It is our understanding, then, that Mulligan and coworkers’ main concern is that recombination and positive selection are present in our data and complicate the interpretation of the phylogeny’s branch lengths in terms of time elapsed. However, we performed rigorous tests for recombination (p. 5) and for positive selection (using maximum likelihood methods to estimate the ratio of non-synonymous to synonymous substitutions along each branch of the phylogeny), and found evidence of neither. If Mulligan et al assert that there is rampant recombination and selection present in these data, they are obligated to provide evidence to support their assertion. Although they cite the long length of the branch leading to syphilis-causing strains as evidence of one or both of these processes at work, this branch could also be explained by the recent divergence of this subspecies and the time dependence of molecular rate estimates [13], which is consistent with our conclusions.

Mulligan and colleagues’ other main contention is that our phylogeny lacks significant structure. They state that the phylogeny was created from 17 SNPs/INDELS. This is incorrect. Our phylogeny was created from a total of 70 SNPs and 12 INDELS, found in 20 different genetic regions; 26 of these SNPs and 5 of these INDELS occurred between T. pallidum strains. In their reanalysis of our data, they fail to provide their methodology, and thus it is unclear which 17 substitutions were used and which were not to redraw the phylogeny. However, having only analyzed a subset of our data, it is to be expected that the authors did not identify some of the clades that we found in our analysis. They also fail to mention that the relative order of divergence of the subspecies found in our phylogeny was further supported by the average nucleotide difference of each subspecies from T. paraluiscuniculi, the outgroup (p. 9).

Mulligan et al express concern that, in our phylogeny syphilis-causing strains appear to be most closely related to subsp. endemicum strains, while in our network analysis syphilis-causing strains appear most closely related to New World subsp. pertenue strains. But, as they note earlier in their commentary, the New World subsp. pertenue strains were not included in the phylogeny because extensive sequencing could not be performed upon them. Therefore, the phylogeny could not show that syphilis-causing strains were most closely related to New World subsp. pertenue strains. In the network analysis, in which all strains were included, syphilis-causing strains were most closely related to New World subsp. pertenue strains, with subsp. endemicum strains following. Figures three and four in our paper are consistent.

In one of their most surprising objections, Mulligan et al suggest that the resemblance between syphilis-causing strains and New World subsp. pertenue strains may be due to geographic clustering. There is no evidence in any peer-reviewed publication that suggests geographic clustering between subspecies occurs. Are they again proposing the Unitarian hypothesis, which posits no genetic differences between the subspecies, is viable? While their idea is an interesting one, there is absolutely no evidence to support it.

Mulligan et al conclude by asking a fascination question: “How could the limited divergence between Treponema species and subspecies give rise to the observed differences in pathogenesis?” This is the question that drives the ongoing work in our laboratories. However, having argued earlier that there is no biological basis for transmission route in the T. pallidum subspecies, we do not understand how Mulligan et al can now assert that variation in the tpr genes most likely underlies biological differences between the subspecies. The authors call for additional sequencing of strains, which we wholeheartedly support. They also propose that more effort be put into sequencing archival strains, in particular into sequencing ancient treponemal DNA. Sadly, the recent literature questions our ability to do this. Mulligan and Lukehart’s previous paper , in which DNA was amplified from a 200-year-old bone sample from Easter Island, is cited. However, amplification of ancient treponemal DNA has never been replicated, and, in their study, Mulligan and colleagues did not provide data that could be used to assess its quality, such as the sequence of individual clones reflecting damage consistent with an age of 200 years. Attempts to amplify ancient treponemal DNA have met with no success, even when a variety of medically diagnosed samples from different geographies and different temporal periods, including the twentieth century, were assessed [14-16]. It has long been suspected that difficulty in identifying treponemes inside diagnostic bone lesions is attributable to the relative absence of the bacteria in tertiary-stage infection [17]. Last year, a rabbit model was used to obtain empirical evidence that, after initial infection, it is difficult to amplify treponemal DNA from bone [16]. Because of the invasiveness of attempting to amplify treponemal aDNA, as well as the method’s poor track record, the scientific community is increasingly skeptical of the utility of this approach.

Mulligan and coworkers misrepresent the most basic facts in our paper, such as the number of regions that we included in our phylogeny. Their failure to accurately present the most basic evolutionary concepts, such as the molecular clock and the uses of phylogenetics, introduces unnecessary confusion. In reality, recent molecular studies have clarified the evolutionary history of the treponemes while providing new questions for future studies [1,5].




References

1. Harper K, Ocampo P, Steiner B, George R, Silverman M, et al. (2008) On the origin of the treponematoses: a phylogenetic approach. PLoS Neglected Tropical Diseases 2: e148.
2. Rocha G, Martins A, Gama G, Brandao F, Atouguia J (2004) Possible cases of sexual and congenital transmission of sleeping sickness. Lancet 363: 247.
3. Reller M, Olsen S, Kressel A, Moon T, Kubota K, et al. (2003) Sexual transmission of typhoid fever: a multistate outbreak among men who have sex with men. Clinical Infectious Diseases 37: 141.
4. Gray R, Mulligan C, Molini B, Sun E, Giacani L, et al. (2006) Molecular evolution of the tprC, D, I, K, G, and J Genes in the Pathogenic Genus Treponema. Molecular Biology and Evolution 23: 2220-2233.
5. Centurion-Lara A, Molini B, Godornes C, Sun E, Hevner K, et al. (2006) Molecular differentiation of Treponema pallidum subspecies. Journal of Clinical Microbiology 44: 3377-3380.
6. Centurion-Lara A, Castro C, Castillo R, Shaffer J, Voorhis W, et al. (1998) The flanking region sequences of the 15-kDa lipoprotein gene differentiate pathogenic treponemes. Journal of Infectious Diseases 177: 1036-1040.
7. Cameron C, Castro C, Lukehart S, Voorhis W (1999) Sequence conservation of glycerophosphodiester phosphodiesterase among Treponema pallidum strains. Infection and Immunity 67: 3168-3170.
8. Cameron C, Lukehart S, Castro C, Molini B, Godornes C, et al. (2000) Opsonic potential, protective capacity, and sequence conservation of the Treponema pallidum subspecies pallidum Tp92. Journal of Infectious Diseases 181: 1401-1413.
9. Clark W (1967) Laboratory notebook 33. Atlanta, GA.
10. Turner T, Hollander D (1957) Biology of the treponematoses. Geneva: World Health Organization.
11. Schell R, Azadegan A, Nitskansky S, LeFrock J (1982) Acquired resistance of hamsters to challenge with homologous and heterologous virulent treponemes. Infection and Immunity 37: 617-621.
12. Lukehart S (2005) Personal communication.
13. Ho S, Shapiro B, Phillips M, Cooper A, Drummond A (2007) Evidence for time dependency of molecular rate estimates. Systematic Biology 56: 515-522.
14. Barnes I, Thomas M (2006) Evaluating bacterial pathogen DNA preservation in museum osteological collections. Proc Biol Scie 273: 645-653.
15. Bouwman A, Brown T (2005) The limits of biomolecular palaeopathology: ancient DNA cannot be used to study venereal syphilis. J Archaeol Sci 32: 703-713.
16. von Hunnius T, Yang D, Eng B, Waye J, Saunders S (2007) Digging deeper into the limits of ancient DNA research on syphilis. Journal of Archaeological Science 34: 2091-2100.
17. Donoghue H, Spigelman M (2006) Pathogenic microbial ancient DNA: a problem or an opportunity. Proc Biol Sci 273: 641-642.