|Home | About | Journals | Submit | Contact Us | Français|
Genomics research provides an unprecedented opportunity for us to probe into the pathogenicity and evolution of the world's most deadly pathogenic bacterium, Yersinia pestis, in minute detail. In our present work, extensive microarray analysis in conjunction with PCR validation revealed that there are considerable genome dynamics, due to gene acquisition and loss, in natural populations of Y. pestis. We established a genomotyping system to group homologous isolates of Y. pestis, based on profiling or gene acquisition and loss in their genomes, and then drew an outline of parallel microevolution of the Y. pestis genome. The acquisition of a number of genomic islands and plasmids most likely induced Y. pestis to evolve rapidly from Yersinia pseudotuberculosis to a new, deadly pathogen. Horizontal gene acquisition also plays a key role in the dramatic evolutionary segregation of Y. pestis lineages (biovars and genomovars). In contrast to selective genome expansion by gene acquisition, genome reduction occurs in Y. pestis through the loss of DNA regions. We also theorized about the links between niche adaptation and genome microevolution. The transmission, colonization, and expansion of Y. pestis in the natural foci of endemic plague are parallel and directional and involve gradual adaptation to the complex of interactions between the environment, the hosts, and the pathogen itself. These adaptations are based on the natural selections against the accumulation of genetic changes within genome. Our data strongly support that the modern plague originated from Yunnan Province in China, due to the arising of biovar orientalis from biovar antiqua rather than mediaevalis.
Yersinia pestis, the causative agent of bubonic and pneumonic plagues, is thought to be one of the most dangerous pathogens in the world. There have been three recorded human plague pandemics, which have claimed hundreds of thousands of lives. Areas where this disease is endemic exist widely in Asia, Africa, and the Americas, where occasional epizootics of animal plague pose great threats to public health (18). Plague has been classified as a reemerging disease by the World Health Organization due to the worldwide increasing incidence of human plague.
Y. pestis can be divided into three biovars, i.e., antiqua, mediaevalis, and orientalis, according to their ability to reduce nitrate and utilize glycerol (1). These three biovars are thought to be responsible for the three major plague pandemics: the Justinian plague, the Black Death, and the modern plague, respectively (1). The third plague pandemic was believed to have originated from Yunnan Province, China, in 1855. It then spread around the world with the aid of modern transportation (25). Human plague has been successfully controlled since the 1950s in China. However, 11 natural plague foci still remain in China, covering more than 277 counties in 19 provinces with an area of more than 1 million km2 (10, 13).
We assumed that the host niche, compound interactions between the environment, the reservoirs, the vectors, and the pathogen, would determine the traits (host range, virulence, biochemical features, genetics contents, etc.) of Y. pestis in a specific geographic region and that strains of Y. pestis from different origins should slightly differ in genome content. However, the challenge was how to track down the genetic differences, how to use these differences as markers for genomotyping, and how to illustrate genomotypic and phenotypic microevolution of Y. pestis. The recently decoded whole-genome sequences of Y. pestis CO92 (17), KIM (5), and 91001 (Y. Song and R. Yang, unpublished data) (accession numbers AE017042, AE017043, AE017044, AE017045, and AE017046) provide the unprecedented opportunity to overcome this challenge. Here, we report our results on the considerable genome dynamics in natural populations of Y. pestis due to gene acquisition and loss, as determined by using DNA microarray-based comparative genomic analysis in conjunction with PCR-based screening. Further, we outlined the parallel microevolution of the Y. pestis genome and propose its intimate link with niche adaptation of Y. pestis in natural foci.
Forty-three strains were used in microarray hybridization (Tables (Tables11 and and2).2). Thirty-six of them are Y. pestis strains that were isolated from 10 plague foci in China and were selected to represent the most abundant Y. pestis diversity associated with adaptive evolution in plague foci. In addition, seven Yersinia pseudotuberculosis strains were included as controls. Two natural isolates of Y. pestis, 91001 and 82009, were used as reference strains in microarray analysis. Y. pestis 91001, a human avirulent strain of biovar mediaevalis, was isolated from a Microtus-related plague focus (focus L [see below]) in China. Y. pestis 82009, a fully virulent strain of biovar orientalis, was isolated from a house mouse-related plague focus (focus F) and was used as an alternative to strain CO92, which is also an orientalis strain. In addition, a total of 260 isolates of Y. pestis (including the previous 36 isolates) from the 10 plague foci were used in PCR analysis.
In the present work, 4,005 annotated open reading frames (genes) were amplified successfully from Y. pestis 82009 or 91001 by using gene-specific primer pairs. These 4,005 genes included nearly all of the CO92 genes and the genes unique to 91001 (Table (Table3)3) after the exclusion of genes encoding insertion sequence protein, integrase, and transposase. The purified PCR products were spotted in duplicate on CSS-1000 silylated glass slides (CEL) by using a SpotArray72 microarray printing system (Perkin-Elmer Life Sciences) to construct the DNA microarrays. A mixture of equal quantities of 91001 and 82009 genomic DNAs was used as reference DNA. Genomic DNA from each of the natural isolates studied was referred to as test DNA. Cy3- or Cy5-labeled probes were generated by priming of the reference or test DNA with random hexamers and extension with Klenow polymerase (2). The labeled reference and test DNAs were combined to hybridize with the microarrays by dual-fluorescence hybridization (2). All hybridizations were performed in triplicate. The hybridized slides were scanned by using a GenePix 4100A personal microarray scanner (Axon Instruments). The scanning images were processed and the data were further analyzed by using GenePix Pro 4.1 software (Axon Instruments) combined with Microsoft Excel software.
An intensity ratio (test DNA normalized intensity/reference DNA normalized intensity) was recorded for each spot and then was converted to log2.5. The hexa-ratios of each gene were averaged. Spots displaying low hybridization signals (the lowest 10% based on Cy3-normalized medians) were filtered out; spots with bad data because of slide abnormalities were discarded as well. The efficacy of the DNA microarrays was further assessed by the control hybridizations of 82009 DNA versus reference DNA, 91001 DNA versus reference DNA, and reference DNA versus reference DNA. A log value of lower than −1 was taken as defining the absence of a gene in the relevant strains. Ninety-nine percent of the spots gave correct predictions of the presence or absence of the corresponding genes. The remaining 1% of the spots gave false predictions and were rejected from the analysis. In the end, 3,661 genes were included in the data sets, and a log ratio of −1 was taken as the cutoff value throughout the experiments.
All of the difference region (DFR) genes (see below) were used in PCR amplification to validate the deletion events, identified by the microarray methods, in the 36 Y. pestis isolates listed in Table Table1.1. Then, one or more genes were chosen from each DFR to stand for the corresponding DFR, and PCR amplification of the selected genes was performed on the 260 isolates of Y. pestis to screen the distribution of DFRs in these strains. All of the DNAs to be tested were arrayed in 96-well PCR plates. Each primer pair was pretested, with the genomic DNA of strain 91001 or CO92 as a template, to ensure successful amplification. PCR products were analyzed by 1.2% agarose gel electrophoresis with ethidium bromide staining.
Twenty-two genomic regions that were absent in at least one of strains studied were identified by extensive microarray analysis (Table (Table4).4). Each of these regions is referred to as a DFR, a term used in a previous study of genome plasticity in Y. pestis (20). The DFR profiles of the 260 isolates of Y. pestis tested can be assigned to 14 groups; each group was termed a genomovar (Table (Table5).5). In this way, the genomic variability in natural populations of Y. pestis was tracked down successfully and was further used as markers to rationally group homologous isolates of Y. pestis.
The DFRs represent the dynamic regions of the Y. pestis genome in natural populations, demonstrating its characteristics of acquisition or deletion in the adaptive evolution of Y. pestis in natural plague foci. Here we present an overview of the parallel microevolution of the Y. pestis genome in natural populations, with evidence of genome content flux through the acquisition or loss of plasmids and chromosomal segments (Fig. (Fig.11).
Y. pestis is a clone that evolved from Y. pseudotuberculosis 1,500 to 20,000 years ago, shortly before the first known pandemics of human plague (1). Y. pestis has acquired two unique virulence plasmids (pPCP1 and pMT1) in the process of speciation (25). Plasmid pPCP1 encodes the plasminogen activator (Pla), a putative invasin/adhesin that is essential for virulence by the subcutaneous route (4, 12). Plasmid pMT1 encodes murine toxin (Ymt), which has a role in the transmission of plague (9, 22), and the F1 capsular protein provides an additional mechanism to block phagocytosis; this mechanism works differently from that of the type III secretion system encoded by the pCD1 plasmid (6).
There are 21 genomic islands (including the virulence-related pgm locus), probably acquired from other organisms through horizontal gene transfer, on the chromosome of strain CO92 (17). Our experimental data indicate that 18 of them are harbored in both Y. pseudotuberculosis and Y. pestis, while the other 3 (island 09, island 14, and island 15) are present only in Y. pestis (Table (Table6).6). Apparently, these three islands were acquired by Y. pestis during the course of speciation (Fig. (Fig.1).1). All three of these islands encode prophages. Some bacteriophages encode bacterial proteins that enable the bacteria to invade host tissues, avoid the host immune defense, and damage host cells (3). With the integration of the bacteriophage genome into the bacterial chromosome, the virulence factors encoded by the prophage can convert their bacterial host from a nonpathogenic strain to a virulent one or to a strain with increased virulence (3).
The acquisition of a number of plasmids and islands may have induced Y. pestis to evolve rapidly from Y. pseudotuberculosis to a newly emerged pathogen that not only is able to parasitize insects in part of its life cycle but also is highly virulent to rodents and humans, causing pandemics of a systemic and often fatal disease. It differs dramatically from its ancestor Y. pseudotuberculosis, which causes only nonfatal gastrointestinal disease in similar hosts. Y. pseudotuberculosis, harboring the hemin storage (hms) locus, insecticidal toxins, iron uptake systems, and secretion systems, has the potential to attack mammals, causing systemic infection, and to be transmitted by fleas. At a certain stage of human history, a change of the natural, social, or economic environment, probably caused by a change in human population or behaviors, might have led to a dramatic increase of the population size of a certain rodent (1). This boom in the rodent population might have triggered the speciation of Y. pestis from Y. pseudotuberculosis as a directional natural selection. Y. pseudotuberculosis is found widely in the environment and is a common cause of animal infections. The bacteria can invade rodents suffering from cold, famine, or illness due to drastic in-species competition or unfavorable environment, and then it enters into the bodies of fleas through flea biting (1). Y. pseudotuberculosis shares a niche with other microorganisms in rodents and fleas, and thus horizontal gene transfer may occur randomly and the beneficial events of gene transfer would be stabilized by vertical inheritance under natural selection. The stepwise acquisition of several genomic islands and plasmids ignited the emergence of Y. pestis in the end.
There was an accumulation of pseudogenes in Y. pestis during the course of speciation (Fig. (Fig.1),1), which is the outcome of the switch of Y. pestis from an enteric lifestyle to a mammalian blood-borne lifestyle (17). For example, yadA and inv are both inactive in Y. pestis but encode functional adhesin and invasion in Y. pseudotuberculosis, enabling the enteropathogen to specifically adhere to surfaces of the host intestines and invade the lining epithelial cells (21-24). The lipopolysaccharide (LPS) of Y. pseudotuberculosis has been shown to possess an O antigen, which is an essential virulence determinant (19), while Y. pestis expresses rough LPS lacking the O antigen, due to the inactivation of several genes in the O-antigen gene cluster (11). Smooth LPS production may be unnecessary for Y. pestis virulence, and the metabolic burden has been alleviated by the inactivation of the O-antigen biosynthetic operon (16).
The third human plague pandemic, beginning in the mid-19th century in China's Yunnan Province and then spreading globally, eventually affected more than 60 countries and regions in Asia, Europe, the Americas, and Africa (1). It is believed that the third pandemic was caused by the emergence of the orientalis strains (25). Indeed, all of the strains isolated from plague focus F in Yunnan Province and its neighboring regions are orientalis strains. All of the orientalis strains tested in this study fell into genomovar 09, which has evolved from genomovar 07 (biovar antiqua) by acquiring DFR 13, which is specific for the orientalis strains (Fig. (Fig.1).1). This strongly supports the notion that biovar orientalis directly evolved from antiqua rather than arising from mediaevalis (11). This fact also suggests that horizontal gene acquisition may play a key role in dramatic evolutionary segregation within species.
For reasons that are probably due to human population density and animal husbandry practices, Y. pestis strains causing the modern plague have been considered to have emerged from China. The data presented here confirm that the modern plague originated from Yunnan Province in China, due to the arising of strains of biovar orientalis from biovar antiqua rather than mediaevalis. Genomovar 09 of Y. pestis seems to be the oldest ancestor of biovar orientalis strains.
DFR 13 (island 15) encodes a filamentous prophage that is integrated into the chromosomal dif locus (7). This conserved prophage was also found in the high-virulence clone Escherichia coli O18:K1:H7, with high homology at nucleotide level (7). E. coli O18:K1:H7, the most virulent among all K1 strains in animal models of sepsis and meningitis, is responsible for almost all cases of neonatal meningitis in the United States and for the preponderance of uncomplicated cystitis cases in North American women. The acquisition of this prophage seems to be linked to the differential disease potential of E. coli O18:K1:H7 (7). Likewise, Y. pestis biovar orientalis, believed to be newly emerged and to be responsible for the third human plague pandemic, possesses major genomic differences compared with its antiqua progenitor.
In contrast to selective genome expansion by horizontal gene acquisition, genome reduction occurs through the loss of DFRs (Fig. (Fig.1).1). The lost genes are always nonessential to bacterial survival (unable to provide a selective benefit of bacterial growth efficiency or fitness in the host) in a specific host niche; that is, if genes are rendered useless due to redundancy within the host niche, then loss bias occurs (14). The parallel loss of DFRs in Y. pestis genomes leads to the discrete segregation between the progenitor and offspring strains, and this genome reduction gradually causes the offspring strains to inhabit a more specific host niche, not overlapping with its progenitor.
In our study, each genomovar is confined in a specific geographic region, commonly a plague focus or a part of a focus with a unique set of natural environment, reservoirs, and vectors (Table (Table7).7). Most of the geographic regions with different primary reservoirs have unique genomovars. Sometimes there is more than one genomovar in a single focus within a single primary reservoir, but each of the genomovars corresponds to a unique set of natural environment and primary vector(s). The microevolution of the genomovars is consistent with the expansion of plague foci. Hence, we present a paradigm of the transmission, colonization, and expansion of Y. pestis in China (Fig. (Fig.2).2). Y. pestis strains from Central Asia and the Far East of Russia may have migrated into the Xinjiang-Tibet region and northeast China, respectively, and then separated into three branches with the expansion of plague foci.
Plague is a typical natural focus-based disease. The long-term existence of Y. pestis in a natural focus is accompanied by its interactions with the animal reservoirs and flea vectors. There is a specific natural environment in a defined geographic region (a plague focus or a part of the focus), which ultimately determines the food chain-based relationship of Y. pestis, reservoirs, and vectors. The unique set of hosts (reservoirs and vectors) in the specific natural environment and their interactions with Y. pestis, commonly termed the host niche, determine the existence and also the type of Y. pestis.
It makes sense to say that the origin of ancestral Y. pestis is associated with only one kind of rodent, but Y. pestis has the potential to be transmitted to other species of animals. Once the bacteria are exposed to new animals in new geographic regions by animal contact or vector-borne routes, new host niches come gradually into being. The genetic variations, including gene acquisition, gene loss, point mutation, and genome rearrangement, occur randomly in the genome of Y. pestis. When Y. pestis temporarily colonizes a new host niche, the specific host niche acts as a constant and directional natural selection, leading to the stabilization and vertical inheritance of the beneficial genetic variations in the genome of the Y. pestis colonized in it, which we may define as directional microevolution of the genome. A specific host niche determines not only the long-term existence of Y. pestis but also the genomovar of Y. pestis itself; that is, the expansion of plague foci is a course of the stepwise adaptation of Y. pestis to the new host niches.
The host niches in different natural plague foci, each as a unique natural selection, direct the parallel adaptation of Y. pestis to the corresponding hosts and environment, which is a course of pathogen generalization from a newly emerged single-host species to a multihost pathogen. We can also say that the parallel adaptation to various niches drives Y. pestis strains to diversify into different biovars or genomovars. Certain genomovars are limited to certain geographic regions (or host niches) through a course of within-species segregation, or so-called pathogen specialization, with the advantage of avoiding niche overlapping.
Our study revealed that the genome of Y. pestis, a newly emerged pathogen, is at an intermediate stage of genetic flux, with evidence of selective genome expansion by horizontal acquisition of plasmids or chromosomal islands and genome reduction by loss of DNA regions. The revealed genome dynamics in natural populations of Y. pestis offer an unprecedented opportunity to establish the link between bacterial genome microevolution and niche adaptation under a Darwinian framework. The transmission, colonization, and expansion of Y. pestis in natural foci constitute a parallel, directional, and gradual process of adaptation to the complex of interactions between the environment, the hosts, and the pathogen, which is based on natural selection against the accumulation of small changes within genome.
While this study was in preparation, Hinchliffe et al. described genomic comparisons of Y. pestis and Y. pseudotuberculosis strains performed by using a CO92 gene-specific microarray (8). They identified dozens of DNA loci that were absent or divergent in more than one of the tested strains of Y. pestis and Y. pseudotuberculosis. Quite a number of these results are shared by our study; e.g., DFR 06, DFR 07, DFR 09, DFR 10, DFR 13, DFR 16, and DFR 17 were also found here to be absent from the relevant Y. pestis strains tested by Hinchliffe et al. One major difference of our study is that we selected isolates representing the natural populations of Y. pestis in confined areas that cover a number of typical plague foci. The extensive distribution of typical natural foci, the long-term collection of Y. pestis isolates, and the existing work on the ecology and epidemiology of plague in China enable us to theorize on the bacterial genome microevolution and its links to niche adaptation, taking advantage of the descriptive interpretation of the genomic differences in strains of Y. pestis and Y. pseudotuberculosis from diverse origins. Another difference is that the whole-genome sequencing data for Y. pestis 91001 allowed us to construct a mixed DNA microarray containing a number of genes that are absent in strain CO92, enabling us to capture many more details in the process of genome microevolution.
We are grateful to Qi Guo for very fruitful discussions and suggestions. We express our respect to Chinese researchers for their excellent work on the ecology and epidemiology of the plague in China.
Financial support for this work came from the National High Technology Research and Development Program of China (Program 863, no. 2001-AA223061) and the National Natural Science Foundation of China (no. 30371284).