Since the first recorded epidemic of syphilis in 1495, controversy has surrounded the origins of the bacterium Treponema pallidum subsp. pallidum and its relationship to the pathogens responsible for the other treponemal diseases: yaws, endemic syphilis, and pinta. Some researchers have argued that the syphilis-causing bacterium, or its progenitor, was brought from the New World to Europe by Christopher Columbus and his men, while others maintain that the treponematoses, including syphilis, have a much longer history on the European continent.
We applied phylogenetics to this problem, using data from 21 genetic regions examined in 26 geographically disparate strains of pathogenic Treponema. Of all the strains examined, the venereal syphilis-causing strains originated most recently and were more closely related to yaws-causing strains from South America than to other non-venereal strains. Old World yaws-causing strains occupied a basal position on the tree, indicating that they arose first in human history, and a simian strain of T. pallidum was found to be indistinguishable from them.
Our results lend support to the Columbian theory of syphilis's origin while suggesting that the non-sexually transmitted subspecies arose earlier in the Old World. This study represents the first attempt to address the problem of the origin of syphilis using molecular genetics, as well as the first source of information regarding the genetic make-up of non-venereal strains from the Western hemisphere.
For 500 years, controversy has raged around the origin of T. pallidum subsp. pallidum, the bacterium responsible for syphilis. Did Christopher Columbus and his men introduce this pathogen into Renaissance Europe, after contracting it during their voyage to the New World? Or does syphilis have a much older history in the Old World? This paper represents the first attempt to use a phylogenetic approach to solve this question. In addition, it clarifies the evolutionary relationships between the pathogen that causes syphilis and the other T. pallidum subspecies, which cause the neglected tropical diseases yaws and endemic syphilis. Using a collection of pathogenic Treponema strains that is unprecedented in size, we show that yaws appears to be an ancient infection in humans while venereal syphilis arose relatively recently in human history. In addition, the closest relatives of syphilis-causing strains identified in this study were found in South America, providing support for the Columbian theory of syphilis's origin.
Citation: Harper KN, Ocampo PS, Steiner BM, George RW, Silverman MS, Bolotin S, et al. (2008) On the Origin of the Treponematoses: A Phylogenetic Approach. PLoS Negl Trop Dis 2(1): e148. doi:10.1371/journal.pntd.0000148
Academic Editor: Albert Ko, Fundação Oswaldo Cruz, Brazil
Received: February 1, 2007; Accepted: November 15, 2007; Published: January 15, 2008
This is an open-access article distributed under the terms of the Creative Commons Public Domain declaration which stipulates that, once placed in the public domain, this work may be freely reproduced, distributed, transmitted, modified, built upon, or otherwise used by anyone for any lawful purpose.
Funding: This work was supported by a pre-doctoral fellowship from the Howard Hughes Medical Institute and dissertation improvement grants from the National Science Foundation (Award Number 0622399) and the Wenner-Gren Foundation to K.N.H. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing interests: The authors have declared that no competing interests exist.
As Naples fell before the invading army of Charles the VIII in 1495, a plague broke out among the French leader's troops . When the army disbanded shortly after the campaign, the troops, composed largely of mercenaries, returned to their homes and disseminated the disease across Europe ,. Today, it is generally agreed that this outbreak was the first recorded epidemic of syphilis. Although its death toll remains controversial, there is no question that the infection devastated the continent . Because the epidemic followed quickly upon the return of Columbus and his men from the New World, some speculated that the disease originated in the Americas . Indeed, reports surfaced that indigenous peoples of the New World suffered from a similar malady of great antiquity  and that symptoms of this disease had been observed in members of Columbus's crew . In the twentieth century, criticisms of the Columbian hypothesis arose, with some hypothesizing that Europeans had simply not distinguished between syphilis and other diseases such as leprosy prior to 1495 .
It was soon recognized that different varieties of treponemal disease exist. Unlike syphilis, which is caused by the spirochete T. pallidum subspecies pallidum, the other types normally strike during childhood and are transmitted through skin-to-skin or oral contact. All are quite similar with regard to symptoms and progression , but endemic syphilis, or bejel, caused by subsp. endemicum, has historically affected people living in hot, arid climates and yaws, caused by subsp. pertenue, is limited to hot and humid areas. Pinta, caused by Treponema carateum, is the most distinct member of this family of diseases. Once found in Central and South America, this mild disease is characterized solely by alterations in skin color. Today, the debate over the origin of treponemal disease encompasses arguments about whether the four infections are caused by distinct but related pathogens  or one protean bacterium with many manifestations .
Paleopathologists have played a pivotal role in addressing the question surrounding the origin of syphilis. The treponemal diseases, with the exception of pinta, leave distinct marks upon the skeleton and can thus be studied in past civilizations. Paleopathological studies of populations in the pre-Columbian New World show that treponemal disease was prevalent, with cases dating back 7,000 years and increasing over time . In contrast, paleopathological studies of large pre-Columbian populations in Europe and Africa have yielded no evidence of treponemal disease –. However, isolated cases of pre-Columbian treponemal disease from other Old World excavation sites have been reported sporadically . Although these cases have often been met with criticism regarding diagnosis, dating, and epidemiological context, they have convinced some that treponemal disease did exist in the pre-Columbian Old World ,.
The T. pallidum genome is small (roughly 1,000 kilobases) and was sequenced in 1998 . However, comparative genetic studies of T. pallidum – have been rare and relatively small in scope. One reason for this is the difficulty in obtaining non-venereal strains for study. Today only five known laboratory strains of subsp. pertenue, two strains of subsp. endemicum, and no strains or samples of T. carateum survive. Furthermore, it is uncertain whether the disease pinta still exists. No cases have been reported to the World Health Organization from the former endemic countries Mexico or Colombia since 1979 . Similarly, endemic syphilis was eradicated some time ago in its European focus, Bosnia . In Turkey, only one infected family has been reported in the last forty years , and a large survey in the United Arab Emirates revealed only non-active cases of endemic syphilis in the elderly . In the Old World, yaws is still reported but appears limited to a few isolated foci in the Republic of Congo and the Democratic Republic of Congo ,, as well as Indonesia and Timor-Leste, where roughly 5,000 cases are reported annually . In the New World, yaws appears to be constrained to an ever-constricting area of Guyana's interior . Because of the paucity of samples available for experiments, most comparative studies have included very few non-venereal strains.
Another limitation on comparative studies has been the small amount of variation present in the T. pallidum genome. Variability is sufficiently low that the discovery of a single nucleotide polymorphism (SNP) has warranted publication in the past ,,. One study has suggested that most variation between the subsp. pallidum and pertenue genomes lies within the tpr gene family , a family of 12 genes with sequence similarity that make up roughly 2% of the genome. A recent examination of this large gene family demonstrated more extensive variation between strains than had previous studies but also documented an unusually high frequency of intra-gene conversion events . Thus, it is possible that most polymorphism in the T. pallidum genome may be concentrated in genes with limited phylogenetic informativeness.
Our goal in this study was to identify variable sites in the T. pallidum genome and characterize them in as many non-venereal strains as possible, in order to test the hypothesis that syphilis emerged in humankind's recent past, from New World-derived strains of T. pallidum. Recombination can result in a phylogeny different from the true one ,, and it has been shown that gene conversion is an important evolutionary mechanism in one large T. pallidum gene family . For this reason, after sequencing many sites from around the genome, we performed rigorous tests for recombination, then built a phylogeny from the non-recombining SNPs and insertions/deletions (indels) identified. These results, paired with geographic analysis of strains, provide novel information on the history of T. pallidum.
Origin and Preparation of Isolates
Twenty-two human Treponema pallidum strains, one T. pallidum strain collected from a wild baboon, and three T. paraluiscuniculi strains, which are responsible for venereal syphilis in rabbits, were used in this study (Table 1). This included all laboratory strains of subsp. pertenue (n = 5) and subsp. endemicum (n = 2). Guyana is the only known active site of yaws infection in the western hemisphere. In order to provide representation of non-venereal strains from the Americas, subsp. pertenue strains (n = 2) were obtained from indigenous children with clinical evidence of non-venereal treponemal disease during a humanitarian medical mission to protected native reserves deep within the Guyanese interior. These samples were collected from a population that has had very little contact with the outside world, both due to the remoteness of their forest location and to legal restrictions on outsider interference. Ethical approval for the sample collection protocol was obtained from the Ethics Board at Lakeridge Health Centre (Oshawa, Canada) and included obtaining informed consent from patients. Scrapings were taken from active yaws lesions and were immediately deposited in either ethanol or saline. They were kept as cool as possible, but not frozen, for the duration of the two-week medical trip. Upon return to the laboratory, they were kept frozen until DNA was extracted. Additional subsp. pertenue samples (n = 4) came from a Dutch collection of strains destroyed years ago during a freezer breakdown. Although these strains were non-viable, the organisms were kept and a small amount of each was made available for this project. Eleven syphilis strains were chosen for analysis based on geographic and chronological span. This included two strains of uncertain subspecies and origin, Haiti B and Madras, which were originally labeled as subsp. pertenue strains, but appear to be subsp. pallidum strains based on both genetic studies and clinical manifestations in a rabbit model , –,.
DNA was obtained from Treponema organisms grown in rabbit tissue or from clinical specimens. Laboratory isolates of T. pallidum were grown in New Zealand white rabbits, consistent with guidelines set out by the Institutional Animal Care and Use Committee at the U.S. Center for Disease Control and Prevention (CDC). DNA was isolated using the QIAamp mini kit (Qiagen, Valencia, CA) according to the manufacturer's instructions for tissue or fluid preparations, depending on the type of sample.
Whole genome amplification (REPLI-g Midi Kit, Qiagen) was performed on the strains for which only a limited amount of DNA was available: the four subsp. pertenue strains from the Dutch collection and a pallidum strain obtained from a South African clinical specimen. The whole genome amplification product was used as a template for subsequent polymerase chain reactions (PCRs). Unfortunately, amplification of the whole genome of the two strains collected in Guyana could not be performed, due to DNA degradation.
Twenty-one genetic regions (Table 2) were sequenced in all strains except for the two clinical samples from Guyana. These regions were scattered around the genome (Fig. 1), and were chosen based on previously demonstrated polymorphism , –,, implication in pathogenesis, or because they harbored repetitive sequences. Because a very limited amount of considerably degraded DNA was available from the clinical specimens collected in Guyana, only seven polymorphic sites encompassing 17 SNPs could be sequenced in these strains. The sites were chosen based on 1) which appeared to be the most phylogenetically informative at the time of sequencing; and 2) which involved small molecular weight products, easily amplified from damaged DNA. They included IGR(fliG-tp0027), deoD, gpd, tp0618, tprI, cfpA, and tpF-1.
The name of the genes and/or intergenic regions (IGRs) amplified are shown in bold next to their numeric coordinates and a point denoting their location in the genome. The coordinates of the nucleotides sequenced within each gene/IGR can be found in Table 2.
Primers (Table S1) were designed using the programs MacVector (Accelrys, Burlington, MA) and Primer3  and the subsp. pallidum genomic DNA sequences posted on the Los Alamos National Laboratory Bioscience Division's STD Sequence Databases webpage (http://www.stdgen.northwestern.edu). PCR amplifications were performed in 50 µL reactions containing 0.5 µM primers (Invitrogen, Carlsbad, CA), 200 µM GeneAmp dNTPs (Applied Biosystems, Foster City, CA), and 2.5 U AmpliTaq Gold polymerase with Gold Buffer and 3.0 mM MgCl2 (Applied Biosystems). PCR conditions were as follows: One cycle of 94°C for 5 minutes; 35 cycles of 94°C for 30 seconds, primer annealing at the appropriate temperature for 30 seconds (Table S1), and 72°C for 1 minute and 30 seconds; followed by a final extension for seven minutes at 72°C. Standard precautions to avoid DNA contamination were employed, including the use of negative controls, aerosol resistant pipette-tips, and a three-station PCR set-up protocol. In addition, sequences that were especially important in the phylogenetic analysis were confirmed through independent amplifications. Amplicons were purified using a gel extraction kit (Qiagen, Valencia, CA) and sequenced using either an ABI 3100/3700 Automated Capillary DNA Sequencer and the Big Dye Sequencing kit (Applied Biosystems) or the CEQ 8000 Genetic Analysis System and the DTCS Quick Start Kit (Beckman-Coulter, Fullerton, CA). Sequencing was performed at the CDC, SeqWright/Fisher Sequencing Services, or Oregon Health Sciences University's Core Laboratory. Sequences were deposited in GenBank under the accession numbers listed (Table 2).
Open reading frames containing polymorphisms were translated into protein sequences to examine the resulting amino acid changes. The average number of nucleotide differences between groups, as well as the average pairwise difference for each polymorphic gene, was calculated using DnaSP . All nucleotide substitutions occurring in genic areas were deemed either synonymous or non-synonymous (Table S2). The amino acid substitutions resulting from non-synonymous changes were scored according to three criteria. Amino acids were categorized according to charge, polarity and volume, and Grantham's distance ,. In the first two cases, a substitution resulting in a change from one category to another was considered a radical substitution, while others were considered conservative. In the latter case, substitutions resulting in distances greater than 100, according to Grantham's index, were categorized as radical, others conservative. Substitutions considered radical in at least two of the three categories were scored as radical (Table S2). Secondary DNA structure in the polymorphic region of IGR(fliG-tp0027) was predicted using the program Mfold .
In order to rule out recombination in the areas analyzed, each gene was tested against all of the paralogs present in the sequenced T. pallidum subsp. pallidum Nichols strain genome, using the program RDP2 . This type of search would identify recombination events between the strains sequenced in this study and the paralogs present in the sequenced genome, as well as recombination between the genes in the strains sequenced. In addition, when small stretches of extremely polymorphic DNA were identified, a BLAST search was performed in order to identify possible donor regions involved in intragenomic conversion events (http://www.ncbi.nlm.nih.gov/BLAST/). Because the complete genome sequence was only available for one strain, it is possible that some recombination events, between paralogs or donor regions not present in the sequenced genome, would not be detected. For this reason, the frequency of synonymous and non-synonymous substitutions was examined in highly polymorphic stretches; regions with a high number of synonymous substitutions and multiple substitutions per codon were considered to be possible results of recombination and were excluded from the analysis. The sequence of these regions can be found in Table S3. The complete alignments, constructed as described below and encompassing all polymorphism in which within-gene recombination had been ruled out, were also analyzed using RDP2, in order to rule out large-scale genomic recombination.
In order to construct phylogenetic trees incorporating all variation, an alignment of the concatenated SNPs and indels was created using ClustalX version 1.83 . The order of the regions in the concatenation corresponds to their position in the genome. Modeltest  was used to choose the appropriate model of nucleotide substitution, Kimura's two parameter, and phylogenetic trees were built in *PAUP 4.0 . Both maximum likelihood and maximum parsimony methods were employed, to glean the most information from both parsimony-uninformative traits and indels, respectively. T. paraluiscuniculi was used as an outgroup (i.e. a taxon known to lie outside of the T. pallidum grouping) with which to root the trees and determine the directionality of substitutions. In the alignment, indels were trimmed to one basepair, in order to prevent their greater length from dominating the analysis. One thousand replicates were run to obtain bootstrap support at each node, with starting trees obtained through random, step-wise addition. Tree bisection and reconnection was used for the branch-swapping algorithm. The maximum likelihood tree was chosen for display, with bootstrap support from both methods displayed at nodes. Trees in which trimmed indels were weighted 1/5 the value of substitutions were also created, in light of the faster mutation rate of repeat regions. Because many substitutions were contained in one gene, tp92, trees in which polymorphism in tp92 was not considered were built. In addition, trees built only from synonymous SNPS or non-synonymous SNPs were created using the methods described above.
Of the twenty-one genes sequenced in this study, data from T. paraluiscuniculi was obtained for all but one gene, tprI, which had previously been shown to be absent from the genome of this species . The tp0618 gene could be amplified from T. paraluiscuniculi Strain A, but not from Strains H and M. It is possible that the gene is missing from the genome of the latter two strains or that the sequence in the priming regions has diverged sufficiently to prevent amplification.
No areas likely to result from recombination were identified in formal tests performed in RDP2. However, polymorphic regions in two genes, tp0618 and tp92, contained small tracts of highly variable DNA with an elevated frequency of synonymous nucleotide substitutions and multiple substitutions per codon. No possible donor regions could be identified for these tracts using BLAST searches on the T. pallidum genome. However, because such a high frequency of synonymous substitutions could be explained by recombination, and it was thus not clear how many mutational events were responsible for the polymorphism observed, the polymorphism in tp0618 and four polymorphic regions from tp92 were excluded from the phylogenetic analysis. The tp92 substitutions included in the analysis can be found in polymorphism Table S2, while all substitutions can be found in polymorphism Table S3.
In the remaining 7 kilobases of the T. pallidum genome examined, which were sequenced from 20 scattered regions, a total of 70 SNPs and 12 indels were identified. Most of them are described here for the first time. Twenty-six substitutions occurred between T. pallidum strains (Table 3), amounting to about one substitution per 275 basepairs. This value is likely to significantly overestimate the amount of polymorphism typical of the genome, however. A number of regions were sequenced either because of their previously demonstrated polymorphism or because they were thought likely to contain variation. This may have weighted the regions sampled towards exceptionally polymorphic areas. The region sequenced in tp92, for example, contained 7 of the 24 substitutions observed between T. pallidum strains (Table 3). Roughly two-thirds of the total substitutions observed represented fixed differences between T. pallidum and the outgroup, T. paraluiscuniculi (Table 4). Singletons were rare, accounting for only 7 of the 70 observed substitutions (Table S2). Three of these singletons were found in the Pariaman strain, which was geographically distinct from other subsp. pertenue strains, and no singletons were observed in the other whole genome amplified strains. Thus, it appears that whole genome amplification did not introduce spurious substitutions.
A few polymorphic regions analyzed here are of special interest in light of past studies and the paucity of genetic variation described to date. Polymorphism between subsp. pertenue strains occurring in the first 200 basepairs of the tprI gene, demonstrated here, had not been described in previous studies of this gene ,. In the small region of tprI sequenced here, all 3 substitutions documented in T. pallidum strains were non-synonymous and in close proximity (Table 3). Two resulted in radical amino acid substitutions (Table S2). Variation in the cfpA gene had previously been reported in a study comparing the sequence of this gene in the Nichols and Haiti B subsp. pallidum strains . We observed only one of the substitutions reported in this article, a polymorphism occurring among subsp. pallidum strains at position 92 (Fig. 2). However, we discovered one fixed difference between the non-venereal subspecies and subsp. pallidum, at residue 303 (Fig. 2). Much of the polymorphism in tp92 was not analyzed in this study, because it fell within hyper-variable regions that were difficult to align and in which recombination could not be ruled out. Even so, in the regions included, 7 non-synonymous and no synonymous substitutions occurred among T. pallidum strains (Table 3). Four of these substitutions resulted in radical amino acid changes (Table S2). Finally, in IGR(fliG-tp0027) the presence of a SNP followed by a long homonucleotide repeat region (Fig. S1) was documented in subsp. pallidum, but not in the other subspecies. The net polymorphism in the former strains was predicted, on the basis of conformational stability, to form a stem-loop structure in the intergenic region between the oppositely transcribed genes fliG and tp0027. This structure, located between the predicted promoter and transcriptional start site of the operon containing genes tp0027 and tp0028 (Fig. S1), could attenuate transcript levels of these genes. Both of these genes are homologous to tlyC, which is believed to either code for a hemolysin or for a protein that regulates hemolysin production .
A phylogenetic tree (Fig. 3), constructed using maximum likelihood and parsimony methods, demonstrated that all T. pallidum strains fell within a single clade. Within this larger clade, several T. pallidum clades with bootstrap support greater than 90% were identified. Subsp. pallidum and subsp. endemicum strains formed groupings distinct from subsp. pertenue strains. In addition, subsp. pertenue strains CDC-1, CDC-2575, and Ghana formed a clade distinct from the other strains of this subspecies, including those gathered nearby. Several clades with lower bootstrap support were also identified within the pallidum and pertenue subspecies. The subsp. pertenue strains occupy a basal location on the tree, indicating an ancestral position for them in the T. pallidum family. This basal position is supported by the average number of nucleotide differences from the T. paraluiscunili clade, which was lower for the subsp. pertenue strains (50.13) then for the subsp. endemicum and pallidum clade (55.15), or for the individual subsp. endemicum (51.33) and subsp. pallidum (56.00) clades. The genetic distance between subsp. pertenue and endemicum strains was small compared to the distance between these non-venereal subspecies and subsp. pallidum. The terminal position of the subsp. pallidum clade on the tree indicates that it diverged most recently.
This maximum likelihood tree is based on 20 polymorphic regions in the T. pallidum genome. Bootstrap support was estimated with 1,000 replicates in order to assess confidence at branching points and are shown within circles where values are high (>90%). Bootstrap support values for both maximum likelihood and maximum parsimony trees are shown, in that order.
The results obtained from this phylogenetic analysis were robust. Trees in which trimmed indels were weighted one-fifth the value of substitutions were found to be qualitatively indistinguishable from unweighted trees. And because many of the substitutions analyzed here came from the portion of tp92 sequenced, a tree built without this gene was constructed (tree not shown). The major groupings described above were observed. However, without tp92, groupings that appeared in Fig. 3 with low bootstrap support, such as the clade containing the Mexico A and subsp. pallidum strain from South Africa, were not identified. Trees built using only synonymous or non-synonymous substitutions demonstrated that while the subspecies clades could be obtained by analyzing only synonymous substitutions, non-synonymous substitutions were responsible for increased phylogenetic resolution (trees not shown). That is, identification of all within-subspecies clades was dependent on non-synonymous substitutions.
Genetic analysis of the two subsp. pertenue strains collected from indigenous groups in Guyana revealed that they were the closest relatives of modern subsp. pallidum strains identified in this study. Only a subset of seven genetic regions could be analyzed in these strains: IGR(fliG-tp0027), deoD, gpd, tp0618, tprI, cfpA, and tpF-1. These regions contained 17 SNPs at which Old World non-venereal strains differed from subsp. pallidum strains (Fig. 2). At 4 of the 17 SNPs examined, the New World subsp. pertenue strains were found to be identical to subsp. pallidum strains. These 4 SNPs occurred in 2 loci on separate sides of the genome, tprI and gpd. At the remaining 13 sites, the New World pertenue strains were identical to Old World non-venereal strains. A network path was constructed, in which these SNPs were considered in their geographic context (Fig. 4). In addition, polarity of substitutions was determined using data from the previously constructed phylogeny (Fig. 3). The tree indicated that the ancestral state of the gpd gene could be found in Old World subsp. pertenue and endemicum strains. In addition, polymorphism in the tprI gene could be used to divide T. pallidum into four groups: subsp. pallidum, subsp. endemicum, and two smaller groups of Old World subsp. pertenue. The phylogeny indicated that of the two Old World subsp. pertenue groups, the more recently diverged was the CDC-1/CDC-2575/Ghana one, while the other subsp. pertenue group was ancestral. Thus, the 4 SNP sequence present in the majority of Old World subsp. pertenue strains appears to have arisen first. The network path suggests that a series of substitutions led from this first group of Old-World pertenue strains to a second group of African subsp. pertenue strains and to subsp. endemicum strains (Fig. 4). The pattern of substitutions suggest that a hypothetical intermediate strain, arising from either the group II subsp. pertenue strains or endemicum strains, once existed and was a progenitor to both New World subsp. pertenue and to all subsp. pallidum strains. This data also suggests that the New World subsp. pertenue strains belong to a group distinct from the Old World subsp. pertenue strains, occupying a phylogenetic position somewhere between Old World non-venereal strains and modern subsp. pallidum strains.
The geographical distribution of the endemic treponemal diseases circa 1900 is shown, based on a map created by Hackett . Each polymorphism pattern is linked to the sites where the strains that contain it were gathered. Arrows convey the directionality of change, determined from the previously built phylogenetic tree as described in the text. The four substitutions were located in two genes located on separate sides of the genome, tprI and gpd. The locations from which subsp. pallidum strains were gathered are listed in Table 1.
In the past, a number of different hypotheses regarding the origin of T. pallidum subsp. pallidum, the causative agent of syphilis, have been put forth. Using new data collected in this study, we assess a number of these hypotheses.
The phylogenetic tree created in this study sheds light on the relative order in which the T. pallidum subspecies emerged. Subsp. pertenue strains gathered from central Africa and the South Pacific occupy basal positions on the tree, indicating that they most closely resemble the ancestral pathogen in humans (Fig. 3). The early emergence of these strains in human history is also supported by their low average nucleotide differences from T. paraluiscuniculi and their similarity to the simian strain of T. pallidum, which infects wild baboons. Although the simian strain could not be distinguished from subsp. pertenue strains using the polymorphic data in this study (Fig. 3), another study focusing on the tpr genes found that it was distinct from human T. pallidum strains examined . This evidence is consistent with the hypothesis that yaws is an heirloom disease in humans, one caused by a pathogen that infected our anthropoid ancestors and has evolved with our species . The presence of yaws in wild populations of our closest relatives, gorillas and chimpanzees ,, further supports this theory. However, a more recent cross-species transfer event between humans and non-human primates cannot be ruled out using the available genetic data. It has been shown that inoculation with the simian strain can cause a yaws-like infection in humans , and it is known that infection rates are high in both humans and wild baboons in yaws-endemic areas of West Africa ,. Thus, it is possible that non-human primates serve as a source of human disease, or vice versa. Therefore, in the future it would be desirable to collect T. pallidum strains from various wild, non-human primate species and to sequence them at additional loci. Such information is likely to provide important information concerning the antiquity of yaws in humans.
Subsp. endemicum strains, gathered from the Middle East and the Balkans, diverged from subsp. pertenue strains at some later date, and subsp. pallidum strains diverged most recently, indicating that they emerged relatively recently in human history (Fig. 3). The topology of the tree is consistent with the long-held belief that treponemal disease is very old and has traveled with humans during their migrations, evolving from ancestral subsp. pertenue, in hot, humid regions, into subsp. endemicum as people settled in cooler and dryer areas, and finally into subsp. pallidum ,. Examination of additional variable sites in non-venereal strains from Africa and Asia may aid in pinpointing the trajectory of this pathogen family in the Old World.
The study of two yaws-causing strains from the Americas provides additional clues to understanding the history of syphilis. The genetic analysis of the two subsp. pertenue strains gathered in Guyana demonstrates that they are the closest relatives of venereal syphilis-causing strains identified in this study (Fig. 4). These strains are genetically distinct from Old World subsp. pertenue strains, having diverged more recently than Old-World non-venereal strains. The geographic analysis of strains, paired with the phylogeny, suggests a three-stage model for T. pallidum's dissemination and evolution. First, T. pallidum arose in the Old World, in the form of non-venereal infection, before spreading with humans to the Middle East/Eastern Europe, in the form of endemic syphilis, and then to the Americas, in the form of New World yaws. Second, a T. pallidum strain from the Americas was introduced back into the Old World, probably as a result of the European exploration of the Americas, becoming the progenitor of modern syphilis-causing strains. Third, modern subsp. pallidum strains disseminated from Europe to the rest of the world.
Descriptions of the clinical presentation of yaws in indigenous Guyanese patients  support the “transitional” position of South American subsp. pertenue strains, between Old World non-venereal and subsp. pallidum strains, indicated by the genetic comparison in this study. These yaws patients, who are inhabitants of Guyana's interior, typically present with a chancre (a chronic, painless ulcer with raised margins) similar to the type seen in venereal syphilis, though found in children and in extra-genital locations. This clinical presentation is quite different from the textbook “frambesiform” lesion characteristic of yaws in Africa and Asia. It is possible that this distinctive lesion results from differences in the pathogen genome, although host differences, both cultural and genetic, may also play a role. Though modern travel increases the probability that a strain present in one place may have been recently introduced from another, the isolation of the aboriginal population from which these strains were collected, as well as their unique genetic and clinical characteristics, makes such an event unlikely. It is possible that these strains possess some of the characteristics of subsp. pallidum strains while still retaining many of the features of subsp. pertenue strains, including a non-sexual transmission mode. The condition of the two samples precluded a complete comparative study involving these strains, but the distinct genetic make-up indicated by the available data makes further study of South American subsp. pertenue strains desirable, if additional samples can be gathered in the future.
Although the data suggest that venereal syphilis-causing strains arose recently, from a New World progenitor, the transmission mode of the ancestral bacterium remains unclear. The closest relatives of syphilis-causing strains in this study were non-venereally transmitted. However, because of the disappearance of endemic treponemal disease from South America it was only possible to study two indigenous strains from this continent, gathered in close proximity. It is possible that a strain even more closely related to subsp. pallidum once existed in pre-Columbian South America and was transmitted venereally. Paleopathologists have assessed the age distribution of treponemal infection in pre-Columbian Native American populations, in hopes of determining the mode of transmission in these civilizations. However, because permanent bone remodeling due to treponemal infection usually occurs in tertiary-stage disease, which can take many years to develop, it is difficult to determine at which age individuals typically contracted the infection in past populations, even when finds are abundant . Several possible cases of congenital syphilis, believed by many to occur only in venereal infection, have been reported from pre-Columbian America, but the diagnoses remain tentative . Some researchers have attempted to determine the nature of treponemal infection in past populations through statistical comparison of specific pathologies in the skeletal record. Based on this method, they assert that the treponemal infection present in the Dominican Republic at the time of Columbus's landing was more akin to venereal syphilis than yaws or endemic syphilis . However, this method remains controversial, because of the limited samples from which specific pathology rates have been determined for each disease . Therefore, it is not clear whether venereal syphilis existed in the New World prior to Columbus's arrival. While it is possible that Columbus and his crew imported venereal syphilis from the New World to Europe, it is also possible that the explorers imported a non-venereal progenitor that rapidly evolved into the pathogen we know today only after it was introduced into the Old World. Indeed, analysis of the changing descriptions of venereal syphilis following its appearance in Europe have led many to believe that the pathogen did evolve rapidly after its initial introduction . Given the limitations of the available data, the question of whether the progenitor of modern syphilis-causing strains was venereal or non-venereal may remain unresolved.
The results of this molecular study clarify some findings of skeletal biology while obfuscating others. The virtual absence of syphilitic lesions from Pre-Columbian Old World skeletons can be explained simply in the context of this data; syphilis did not exist in these areas until the Renaissance. On the other hand, the absence of lesions typical of yaws or endemic syphilis in these areas is puzzling in light of the genetic data. If the non-venereal treponematoses arose in the Old World long before syphilis, then why isn't there more evidence of their presence? If yaws was the first form of treponemal disease, as indicated by our study, and was limited to hot, humid areas, we would expect preservation of ancient, affected remains to be poor. This may contribute to the paucity of skeletal finds.
In light of the hypotheses regarding the rapid evolution of subsp. pallidum in Renaissance Europe and the lack of knowledge concerning the genetic basis for the different clinical manifestations of the treponematoses, any evidence of positive selection or functional change in the T. pallidum genome would be of great interest. A comparison of trees constructed with only synonymous or non-synonymous substitutions emphasized the role of non-synonymous substitutions in differentiating T. pallidum strains. Ten of the 14 non-synonymous substitutions observed between T. pallidum strains occurred in just 2 genes: tprI and tp92. Similarly, 6 of 9 radical amino acid substitutions observed between T. pallidum strains occurred in these 2 genes (Table S2). Given the evidence for the role of the TprI and Tp92 proteins in pathogenicity ,,, the substitutions clustered in the regions of these genes that were sequenced may hint at positive selection. Sequencing the entirety of these genes in many strains of T. pallidum may be worthwhile, in order to better assess this possibility. Similarly, since tp0027 and tp0028 are both homologous to tlyC, a gene that either encodes a hemolysin or regulates a cryptic one in E. coli , a difference in transcript level between subspecies could affect pathogenesis. To this end, a regulatory function for the predicted stem-loop structure in IGR(fliG-tp0027) in subsp. pallidum could be tested. Transcript levels of tp0027 could be characterized in the different subspecies; in addition, the two IGR variants discovered in this study could be placed upstream of a reporter gene in a genetically tractable bacterium, such as E. coli, in order to directly examine the effect of the stem-loop structure on transcript levels.
Our conclusions regarding the history of T. pallidum differ from those drawn in a recent comparative study of the tpr gene family , in which it was asserted that the times of emergence for the pathogens that cause endemic syphilis, yaws, and syphilis were similar and dated to sometime later than the emergence of modern humans but earlier than the Renaissance. We propose several reasons for why the conclusions arrived at in the two studies diverge. The strains collected from Guyana in this study played an integral role in our analysis, because they were most closely related to syphilis-causing strains and helped establish the geographic trajectory of T. pallidum evolution. In addition, in the genetic regions we examined, recombination was much less common than in the majority of the tpr genes, if it was present at all. Our data suggest that the T. pallidum genome is evolving in a largely clonal manner amenable to phylogenetic analysis, with the exception of frequent intra-gene conversion events confined within the members of the tpr gene family . Finally, because T. paraluiscuniculi was extremely similar to T. pallidum at all but one of the sequences examined (tprI is missing in the former species), it was easy to create alignments and to assess the directionality of mutations in this study. The evolutionary pathway between the tpr genes of T. paraluiscuniculi and T. pallidum is more complicated and much harder to interpret.
This study has some obvious limitations. The level of polymorphism found between T. pallidum strains is quite low, as suggested by previous studies. For this reason, the level of resolution in the phylogenetic tree is relatively poor. It is likely that polymorphism data from the rest of the genome will clarify the topology of this tree, including the relationship between subsp. endemicum and subsp. pallidum strains, which are grouped together in this study with low bootstrap support, and the position of subsp. pertenue strains, which occupy a basal position on the tree but group together only by default. Similarly, the close relationship between subsp. pallidum strains and New World subsp. pertenue strains described in this paper hinges on the analysis of only 4 SNPs. Three of these SNPs were found in a single gene, tprI. The close proximity of these non-synonymous SNPs within the gene, the non-linear relationships of the strains indicated by the substitutions (Fig. 4), and the evidence that the TprI protein may be involved in pathogenesis, suggest that tprI may not be evolving neutrally. For this reason, it is unlikely that these SNPs have accumulated in a clockwork manner. Instead, the information that can be gleaned from this gene is limited to the relative order in which evolutionary events occurred.
The large-scale comparative genetic studies possible on the pathogens that cause diseases such as malaria and anthrax will never be possible in T. pallidum, because of the disappearance of the non-venereal treponematoses and the strains that cause them ,. The prevalence of yaws in Guyana, the last country in South America in which yaws has been documented in recent years, has decreased annually, and surveys carried out by our group in 2006 and 2007 in endemic yaws territory demonstrated no active cases of the disease. Analysis of South American strains is necessary in order to assess the relationship between subsp. pallidum and non-venereal treponemal strains. Because it is not clear whether an opportunity to examine such strains will arise again, the results presented in this paper are of special importance in the debate over the origins of treponemal disease.
In conclusion, in this study we found that syphilis-causing strains evolved relatively recently in human history and that the closest relatives of subsp. pallidum were yaws-causing strains from the New World. When this genetic data is combined with extensive documentary evidence that syphilis appeared in Europe for the first time around 1495  and the apparent absence of skeletal signs of syphilis in pre-Columbian Europe and North Africa, the Columbian hypothesis for syphilis's origin gains new strength.
Predicted stem-loop structure observed in all Treponema pallidum subsp. pallidum strains in IGR(fliG-tp0027). All nucleotides preceding the transcription start site are shown, with the predicted polymerase binding site in red and the putative hemolysin genes tp0027 and tp0028 following the intergenic region. The homonucleotide repeat that allows this stem-loop formation is absent in subsp. endemicum and pertenue strains, and its presence may affect transcription of the genes ahead.
(0.04 MB DOC)
Primers and annealing temperatures used in this study.
(0.07 MB DOC)
(0.07 MB XLS)
Polymorphism table including excluded regions.
(0.09 MB XLS)
We thank F. Heffron, J. Thomas, and S. Altizer for the use of their laboratories; Ve'ahavta (The Canadian Jewish Humanitarian Relief Organization) for collecting strains from Guyana; The Guyana Ministry of Health; L. Schouls, R. Limberger, and S. Lukehart for additional strains used in this project; J. Logsdon, A. Case, T. Schlenke, E. Holmes, and K. Bagamian for comments and help with a previous draft; B. Levin for logistical support; and B. Stoner-Duncan and L. Waller for help with graphics. The findings and conclusions in this paper are those of the authors and do not necessarily represent the views of the U.S. Centers for Disease Control and Prevention.
Conceived and designed the experiments: NS KH BS GA. Performed the experiments: KH BS RG. Analyzed the data: KH. Contributed reagents/materials/analysis tools: KH BS RG MS SB AP GA. Wrote the paper: KH PO BS MS SB AP GA.
- 1. Fracastoro G (1546) De contagione et contagiosis morbis et eorum curatione. Wright W, translator. New York: GP Putnam's Sons.
- 2. Fernandez de Oviedo y Valdes G (1526) Sumario de la natural historia de las Indias. Mexico: Fondo de Cultura Economico.
- 3. Diaz de Isla R (1539) Treatise on the serpentine malady, which in Spain is commonly called bubas, which was drawn up in the hospital of All Saints in Lisbon.
- 4. Quetel C (1990) The history of syphilis. Baltimore: Johns Hopkins Press.
- 5. Las Casas B (1530) Apologetica Historica. Madrid: Allanza Editorial.
- 6. Holcomb R (1934) Christopher Columbus and the American origin of syphilis. United States Naval Medical Bulletin 32: 401–430.
- 7. Antal G, Lukehart S, Meheus A (2002) The endemic treponematoses. Microbes and Infection 4: 83–94.
- 8. Cockburn T (1961) The origin of the treponematoses. Bulletin of the World Health Organization 24: 221–228.
- 9. Hudson E (1963) Treponematosis and Anthropology. Annals of Internal Medicine 58: 1037–1048.
- 10. Cook D, Powell M (2005) Piecing the puzzle together: North American treponematosis in overview. In: Powell M, Cook D, editors. The myth of syphilis: the natural history of treponematosis in North America. Gainesville: University Press of Florida. pp. 442–479.
- 11. Moller-Christensen V (1967) Evidence of leprosy in earlier peoples. In: Brothwell D, Sandison A, editors. Diseases in antiquity. Springfield: C.C. Thomas. pp. 295–306.
- 12. Baker B, Armelagos G (1988) The origin and antiquity of syphilis. Current Anthropology 29: 729–732.
- 13. Stloukal M (1988) Comment: The origin and antiquity of syphilis. Current Anthropology 29: 727.
- 14. Dutour O, Palfi G, Berato J, Brun J (1994) L'origine de la syphilis en Europe: avant ou apres 1493? Toulon, France: Centre Archeologique du Var.
- 15. Meyer C, Jung C, Kohl T, Poenicke A, Poppe A, et al. (2002) Syphilis 2001–a palaeopathological reappraisal. Homo 53: 39–58.
- 16. Brothwell D (2005) North American treponematosis against the bigger world picture. In: Powell M, Cook D, editors. The myth of syphilis: the natural history of treponematosis in North America. Gainesville: University Press of Florida. pp. 480–496.
- 17. Fraser C, Norris S, Weinstock G, White O, Sutton G, et al. (1998) Complete genome sequence of Treponema pallidum, the syphilis spirochete. Science 281: 375–388.
- 18. Noordhoek G, Wieles B, Sluis J, Embden J (1990) Polymerase chain reaction and synthetic DNA probes: a means of distinguishing the causative agents of syphilis and yaws? Infection and Immunity 58: 2011–2013.
- 19. Centurion-Lara A, Sun E, Barrett L, Castro C, Lukehart S, et al. (2000) Multiple alleles of Treponema pallidum repeat gene D in Treponema pallidum isolates. Journal of Bacteriology 182: 2332–2335.
- 20. Cameron C, Castro C, Lukehart S, Voorhis W (1999) Sequence conservation of glycerophosphodiester phosphodiesterase among Treponema pallidum strains. Infection and Immunity 67: 3168–3170.
- 21. 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.
- 22. 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.
- 23. 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.
- 24. Falabella R (1994) Nonvenereal treponematoses: yaws, endemic syphilis, and pinta. Journal of the American Academy of Dermatology 31: 1075.
- 25. Arslanigic N, Bokonjic M, Macanovic K (1989) Eradication of endemic syphilis in Bosnia. Genitourinary Medicine 65: 4–7.
- 26. Yakinci C, Ozcan A, Aslan T, Demirhan B (1995) Bejel in Malatya, Turkey. Journal of Tropical Pediatrics 41: 117–120.
- 27. Nsanze H, Lestringant G, Ameen A, Lambert J, Galadari I, et al. (1996) Serologic tests for treponematoses in the United Arab Emirates. International Journal of Dermatology 35: 800–801.
- 28. Bhattacharya S (2004) Flesh-eating disease threatens Congo's pygmies. NewScientistcom News Service. Available: http://www.newscientist.com/article.nsiddn4967. Accessed 11 November 2006.
- 29. Story F (2006) Yaws prevalence study in the DRC. Epicentre. Available: http://www.epicentre.msf.org/about-us/folder.news/news_item.2005-04-14.4533160824/. Accessed 27 November 2006.
- 30. WHO (2006) Communicable diseases: yaws. Available: http://www.searo.who.int/en/Section10/Section2134.htm. Accessed 27 November 2006.
- 31. Scolnik D, Aronson L, Lovinsky R, Toledano K, Glazier R, et al. (2003) Efficacy of a targeted, oral penicillin-based yaws control program among children living in rural South America. Clinical Infectious Diseases 36: 1232–1238.
- 32. Weinstock G, Nooris S, Sodergren E, Smajs D (2002) Identification of virulence genes in silico: infectious disease genomics. In: Brogden K, Roth J, Stanton T, Bolin C, Minion F et al., editors. Virulence mechanisms of bacterial pathogens. 3rd ed. Washington, DC: ASM Press. pp. 251–261.
- 33. Posada D, Crandall K (2002) The Effect of Recombination on the Accuracy of Phylogeny Estimation. Journal of Molecular Evolution 54: 396–402.
- 34. Schierup M, Hein J (2000) Consequences of Recombination on Traditional Phylogenetic Analysis. Genetics 156: 879–891.
- 35. Clark W (1967) Laboratory notebook 33. Atlanta, GA.
- 36. Izard J, Samsonoff W, Kinoshita M, Limberger R (1999) Genetic and structural analyses of cytoplasmic filaments of wild-type Treponema phagedenis and a flagellar filament-deficient Mutant. Journal of Bacteriology 181: 6739–6746.
- 37. Rozen S, Skaletsky H (2000) Primer3 on the WWW for general users and for biologist programmers. Bioinformatics Methods and Protocols: Methods in Molecular Biology. Totowa, NJ: Humana Press. pp. 365–386.
- 38. Rozas J, Rozas R (1999) DnaSP version 3: an integrated program for molecular population genetics and molecular evolution analysis. Bioinformatics 15: 174–175.
- 39. Zhang J (2000) Rates of conservative and radical nucleotide substitutions in mammalian genes. Journal of Molecular Evolution 50: 56–68.
- 40. Grantham R (1974) Amino acid difference formula to help explain protein evolution. Science 185: 862–864.
- 41. Zuker M (2003) Mfold web server for nucleic acid folding and hybridization prediction. Nucleic Acids Research 3: 3406–3415.
- 42. Martin D, Williamson C, Posada D (2005) RDP2: recombination detection and analysis from sequence alignments. Bioinformatics 21: 260–262.
- 43. Thompson J, Gibson T, Plewniak F, Jeanmougin F, Higgins D (1997) The ClustalX windows interface: flexible strategies for multiple sequence alignment aided by quality analysis tools. Nucleic Acids Research 24: 4876–4882.
- 44. Posada D, Crandall K (1998) Modeltest: testing the model of DNA substitution. Bioinformatics 14: 817–818.
- 45. Swofford D (2003) PAUP*. Phylogenetic analysis using parsimony (*and other methods). Version 4. Sunderland, MA: Sinauer Associates.
- 46. Giacani L, Su E, Hevner K, Molini B, Voorhis WV, et al. (2004) Tpr Homologs in Treponema paraluiscuniculi Cuniculi A Strain. Infection and Immunity 72: 6561–6576.
- 47. Centurion-Lara A, Molini B, Godornes C, Sun E, Hevner K, et al. (2006) Molecular differentation of Treponema pallidum subspecies. Journal of Clinical Microbiology 44: 3377–3380.
- 48. Hsu T, Hutto D, Minion F, Zuerner R, Wannemuehler M (2001) Cloning of a Beta-Hemolysin Gene of Brachyspria (Serpulina) hyodysenteriae and Its Expression in Escherichia coli. Infection and Immunity 69: 706–711.
- 49. Cockburn T (1967) The Evolution of Human Infectious Diseases. In: Cockburn T, editor. Infectious Diseases: Their Evolution and Eradication. Springfield, IL: Charles C. Thomas. pp. 84–107.
- 50. Levrero F, Gatti S, Gautier-Hion A, Menard N (2007) Yaws disease in a wild gorilla population and its impact on the reproductive status of males. American Journal of Physical Anthropology 132: 568–575.
- 51. Lovell N, Jurmain R, Kilgore L (2000) Skeletal evidence of probably treponemal infection in free-ranging African apes. Primates 41: 275–290.
- 52. Smith J, David N, Indgin S, Israel C, Levine B, et al. (1971) Neuro-opthalmological study of late yaws and pinta. II. The Caracas Project. British Journal of Venereal Diseases 47: 226–251.
- 53. Fribourg-Blanc A, Mollaret H (1969) Natural Treponematosis of the African Primate. Primates in Medicine 3: 113–121.
- 54. Fribourg-Blanc A, Niel G, Mollaret H (1963) Note Sur Quelques Aspects Immunologiques Du Cynocephale Africain. Bulletin de la Societe de Pathologie Exotique et de Ses Filiales 56: 474–485.
- 55. Hudson E (1965) Treponematosis and Man's Social Evolution. American Anthropologist 67: 885–901.
- 56. Cook D, Powell M (2005) Piecing the Puzzle Together: North American Treponematosis in Overview. In: Cook DaMP, editor. The Myth of Syphilis. Gainesville, FL: University Press of Florida. pp. 442–479.
- 57. Rothschild B, Calderon F, Coppa A, Rothschild C (2000) First European Exposure to Syphilis: The Dominican Republic at the Time of Columbian Contact. Clinical Infectious Diseases 31: 936–941.
- 58. Giacani L, Sambri V, Marangoni A, Cavrini F, Storni E, et al. (2005) Immunological evaluation and cellular location analysis of the TprI Antigen of Treponema pallidum subsp. pallidum. Infection and Immunity 73: 3817–3822.
- 59. Giacani L, Hevner K, Centurion-Lara A (2005) Gene organization and transcriptional analysis of the tprJ, tprI, tprG, and tprF loci in Treponema pallidum strains Nichols and Sea81-4. Journal of Bacteriology 187: 6084–6093.
- 60. Anselmi M, Moreira J, Caicedo C, Guderian R, Tognoni G (2003) Community participation eliminates yaws in Ecuador. Tropical Medicine and International Health 8: 634–638.
- 61. WHO (2006) Yaws elimination in India: a step towards eradication. Available: http://www.whoindia.org/EN/Section210/Section424.htm. Accessed 27 November 2006.
- 62. Hackett C (1976) Diagnostic Criteria of Syphilis, Yaws, and Treponarid (Treponematoses) and of Some Other Diseases in Dry Bones. Berlin: Springer-Verlag.