|Home | About | Journals | Submit | Contact Us | Français|
Despite the widespread presence of bubonic plague in sylvatic reservoirs throughout the world, the causative agent (Yersinia pestis) evolved in its present form within the last 20,000 years from enteropathogenic Yersinia pseudotuberculosis. Comparison of the genomes from the two species revealed that Y. pestis possesses only a few unique plasmid-encoded genes that contribute to acute disease, whereas this organism has lost about 13% of the chromosomal genes that remain active in Y. pseudotuberculosis. These losses reflect readily detectable additions, deletions, transpositions, inversions, and acquisition of about 70 insertion sequence (IS) inserts, none of which are likely to promote increased virulence. In contrast, major enzymes of intermediary metabolism, including glucose 6-phosphate dehydrogenase (Zwf) and aspartase, are present but not catalytically functional due to the presence of missense mutations. The latter are generally not detectable by the technology of bioinformatics and, in the case of Y. pestis, result in radical changes in the metabolic flow of carbon. As an important consequence, plague bacilli exhibit a stringent low-calcium response characterized by conversion of L-glutamate (and metabolically related amino acids) to L-aspartate with secretion of the latter into supernatant fluid at 37°C in culture media containing Na+ but lacking added Ca2+. This phenomenon also occurs in vivo and likely adversely affects the bioenergetics of host amino acid pools. Curiously, aspartase is functional in all tested enzootic (pestoides) strains of Y. pestis. These isolates are typically restricted to the ancient plague reservoirs of Central Asia and Africa and are fully virulent in members of the rodent Superfamily Muroidea but avirulent in guinea pigs and man. The implications of these findings for the distribution and ecology of Y. pestis could be significant.
Approximately 200 million years ago the yersiniae line carrying the progenitor low Ca2+-response (LCR) virulence plasmid, now termed pCD (calcium dependence) in Yersinia pestis (Ferber and Brubaker 1981), diverged, leading to the emergence of modern Yersinia enterocolitica and Y. pseudotuberculosis (Achtman et al. 1999). These organisms retain slightly different functional forms of this ~70-kb plasmid (Hu et al. 1998), where it is designated pYV (Yersinia Virulence) (Zink et al. 1980), and exhibit both modest gains and losses of chromosomal genes (Chain et al. 2004, 2006, Thomson et al. 2006). Important additions to Y. pseudotuberculosis that occurred during this long period of evolution include the ability to synthesize and incorporate 2,6-dideoxy-hexoses (not present in Y. enterocolitica) into lipopolysaccharide O-groups (Davies 1961), express a catalase-peroxidase termed KatY (Garcia et al. 1999), and colonize regional lymphatic tissue (Avtsyn et al. 1990) as opposed to Y. enterocolitica, which more commonly promotes direct gastrointestinal disease (Bottone 1997). Nevertheless, both of these species remain enteropathogens and, except for the new world serogroup O:8 isolates of Y. enterocolitica (Carter and Collins 1974), seldom cause systemic disease in noncompromised human hosts.
In contrast, Y. pestis (the etiological agent of bubonic plague), which evolved from a serotype O:1b strain of Y. pseudotuberculosis (Skurnik et al. 2000) within the last 5000–15,000 years (Achtman et al. 1999), arguably causes the most severe epidemic disease known to human (Pollitzer 1954, Orient 2004, Kelly 2005). This eyeblink of geological time is far too short to accommodate de novo gene evolution or significant modification and regulation of preexisting genes capable of causing acute infection. However, salient new genetic information was acquired during this period by lateral transfer of two unique plasmids of ~10 and ~110kb termed pPCP (pesticin coagulase plasminogen activator) and pMT (murine toxin), respectively (Ferber and Brubaker 1981). The plasminogen activator of the plague bacillus (Beesley et al. 1967, Sodeinde et al. 1992, Lähteenmäki et al. 1998, Kukkonen et al. 2004) is encoded by pPCP (Ferber and Brubaker 1981) and fraction 1 antigen (F1) of typically encapsulated Y. pestis (Baker et al. 1952) and murine toxin, a phospholipase D (Rudolph et al. 1999), are encoded by pMT (Protsenko et al. 1983, Kutyrev et al. 1986). Nevertheless, these Y. pestis–specific determinants are not always essential for acute disease as judged by the observation that transformation of pPCP into Y. pseudotuberculosis did not significantly increase virulence in mice (Kutyrev et al. 1999), and the absence of this plasmid in certain enzootic isolates of Y. pestis (described below) did not hinder tissue invasiveness (Worsham and Roy 2003). Similarly, F1 is not required for virulence in mice (Burrows 1957), and a fatal human case of plague was caused by an F1-deficient mutant (Winter et al. 1960). These findings strongly suggest that other factors contribute to the acute nature of plague. This paper discusses recent data regarding the importance of the absence of an active aspartase enzyme for the virulence of Y. pestis in mammalian hosts and its potential implications for the distribution and ecology of plague.
An attractive alternative explanation for the severity of plague is the possibility that invasiveness is a function of genetic loss. The major phenotypic differences between Y. pestis and Y. pseudotuberculosis are summarized in Table 1. These include recognition that plague bacilli typically exhibit a longer doubling time than cells of the enteropathogenic yersiniae, especially at host temperature (Brubaker 1972, Perry and Fetherston 1997). This reduced rate of growth is correlated with a requirement for CO2 that is not essential but, if supplied, shortens the division time of Y. pestis to that of Y. pseudotuberculosis provided that other essential nutrients are present (Delwiche et al. 1959, Surgalla et al. 1964). Early studies demonstrated that CO2 is normally fixed by Y. pestis into phosphoenolpyruvate (via phosphoenolpyruvate carboxykinase and an irreversible phosphoenolpyruvate carboxylase) to form oxaloacetate (Baugh et al. 1964) as is typical of many bacterial species. Although this approach did not resolve the basis for the stimulatory role of CO2 for Y. pestis, it did implicate oxalacetate, the immediate precursor of L-aspartic acid, as a critical intermediate. Results of subsequent work (Dreyfus and Brubaker 1978) showed that classical strains of Y. pestis lacked the common prokaryotic enzyme aspartate ammonia-lyase or aspartase (AspA) that converts L-aspartate into fumarate, an intermediate of the tricarboxylic acid cycle. Unrelated parallel work (Mortlock 1962, Mortlock and Brubaker 1962) undertaken on hexose metabolism also demonstrated the absence in typical plague bacilli of glucose 6-phosphate dehydrogenase (Zwf), the initial enzyme of the pentose phosphate pathway (required for production of pentose for nucleic acid synthesis). These two genetic lesions promote a significant lack of fitness, at least in vitro, and likely account in full for the fastidious nature of Y. pestis as opposed to Y. pseudotuberculosis and Y. enterocolitica (Fowler and Brubaker 1994, Brubaker 2005, 2007, Viola et al. 2008).
Comparative annotation of the genomes of Y. pestis (Parkhill et al. 2001, Deng et al. 2002, Zhou et al. 2004a) and Y. pseudotuberculosis (Chain et al. 2004, 2006) revealed that plague bacilli have lost about 13% of all genes that remain functional in Y. pseudotuberculosis. This gene reduction reflects the consequences of chromosomal inversion, transposition, deletion, insertion of ~70 IS elements, and various mechanisms causing altered reading frames. These events account for known alterations between pCD and pYV (Hu et al. 1998), loss mutations promoting a reduced inflammatory response (Kukkonen et al. 2001, Kawahara et al. 2002), and the differences accounting for phenotypic distinctions in nutritional requirements, carbohydrate fermentation, and other biochemical properties (Brubaker 2004). The comparison did not, however, accommodate the critical absence of functional aspA or zwf, and subsequent effort demonstrated that these losses were caused by altered single DNA base pairs causing missense mutations.
A single base transition (T·A→C·G) at amino acid position 155 causes replacement of serine in the active Zwf of Y. pseudotuberculosis with proline in the inactive enzyme of typical Y. pestis (Chain et al. 2004, 2006). It is well established that proline causes distortion of secondary structure, and thus loss of enzyme activity after its introduction into a given enzyme is not surprising. In contrast, inactivity of AspA reflects a single base transversion (G·C→T·A) at amino acid position 363, where valine is replaced in the active enzyme of Y. pseudotuberculosis by the structurally similar, but inactive, leucine of typical isolates of Y. pestis (Chain et al. 2004, 2006). Skepticism regarding the possibility that substitution of one aliphatic amino acid by another solely accounts for loss of catalytic ability was eliminated by demonstrating that restoration of the active valine at position 363 of the AspA of Y. pestis restored full enzymatic activity. This seemingly innocuous change, which significantly reduces turnover (kcat) but not substrate binding (Km), may cause alteration in positioning of the adjacent catalytic residues Glu334 and Gln191 (Viola et al. 2008).
As noted above, the LCR is mediated by pCD/pYV. This phenotype is defined as the ability to either initiate vegetative growth at 37°C in vitro in the presence of ~2.5mM Ca2+ without upregulation of a pCD/pYV-encoded type 3 secretion system (T3SS) or to undergo bacteriostasis with full induction of this T3SS. The LCR is especially stringent in typical strains of Y. pestis, and restricted growth in Ca2+-deficient medium is characterized by accumulation of L-aspartic acid in spent culture medium (Dreyfus and Brubaker 1978, Brubaker 2005). It is highly likely that this lost aspartic acid accounts for the prompt shutoff of growth observed for cells of Y. pestis after shift from 26°C to 37°C in Ca2+-deficient medium (Brubaker 2007) since this form of bacteriostasis occurs as a consequence of reduced adenylate energy charge (Zahorchak et al. 1979). Another variable that influences the nutritional response of typical plague bacilli to Ca2+ is the presence of Na+, which may serve as a porter for the exit of aspartic acid at 37°C. Indeed, at slightly acidic pH the organisms exhibit normal full-scale growth if both Na+ and Ca2+ are omitted from the culture medium. Na+ did not, however, significantly modify the parallel upregulation of the attendant T3SS known to be functional in Ca2+-starved cells grown at 37°C (Brubaker 2007). In summary, these findings suggest that the abrupt bacteriostasis observed for typical Ca2+-starved cells of Y. pestis is caused by loss of metabolic carbon released in the form of aspartic acid by at least one Na+-driven export reaction.
As already noted, Zwf initiates the synthesis of ribose from hexose via the pentose phosphate pathway. This reaction probably has little importance in cells of Y. pestis, which typically reside within a closed flea–host cycle where exogenous pentose is readily available; the organisms can also generate ribose from 3-carbon fragments via transketolase–transaldolase rearrangement (Mortlock 1962). However, nicotinamide adenine dinucleotide phosphate (NADPH) is an important secondary product of Zwf that must be generated by alternative means by Y. pestis grown on hexose. Plague bacilli cultivated at 37°C in a Ca2+-deficient chemically defined medium containing glucose undergo marked lysis (Brownlow and Wessman 1960); thus, Zwf-deficiency may also modulate the LCR in these organisms. This phenomenon, of course, is distinct from the typical bacteriostasis with retention of viability that occurs in Ca2+-deficient medium containing Na+, at neutral pH.
Available evidence indicates that the transmission of plague in sylvatic reservoirs is almost exclusively undertaken by fleabite. Reliance upon this method of transfer depends upon a sufficient host bacteremia to ensure infection of the flea vector (Engelthaler et al. 2000, Eisen et al. 2006) and the ability of Y. pestis to kill their current host, thereby assuring that infectious fleas disseminate in search of new hosts. Mortality therefore became an essential part of the life cycle of Y. pestis, and selective pressure caused by this necessity has obviously directed the further evolution of acute disease. As a consequence of this process, plague bacilli became efficient pathogens and have now disseminated from ancestral foci in Central Asia and perhaps Africa to reservoirs throughout the world where they cause lethal disease in many animal species, including humans (Suntsov and Suntsova 2008). The virulence of these potential “epidemic” isolates contrasts markedly with “enzootic” strains of Y. pestis (also termed “pestoides” variants) that remain contained within their original reservoirs (Martinevskii 1969, Anisimov et al. 2004). Enzootic isolates are now classified according to subspecies designation altaica, caucasica, hissarica, ulegeica, and talassica (Anisimov et al. 2004). With the exception of caucasica isolates (which lack pPCP) these enzootic strains possess all known virulence factors of epidemic isolates, and commonly retain the normal ability of Y. pseudotuberculosis to ferment the sugars rhamnose and melibiose. Curiously, they are attenuated in many mammalian species, including guinea pigs and primates, but remain virulent in members of the rodent Superfamily Muroidea and in nature have only been recovered from these rodents or their fleas (Martinevskii 1969, Anisimov et al. 2004). Further study of enzootic variants has recently shown that they sometimes express catalytically active Zwf and always exhibit functional AspA (Bearden et al. 2008).
A comparison of 10 enzootic isolates of Y. pestis with their epidemic counterparts (Bearden et al. 2008) demonstrated that three of the former expressed functional Zwf associated with retention of serine at amino acid position 155 (Table 2). These strains were the Russian pestoides E, F, and G strains, all of subspecies caucasica, that also exhibit other primitive traits, including the absence of pPCP and attendant plasminogen activator. All tested enzootic isolates produced biologically active AspA although its specific activity exhibited significant variation depending upon the amino acid present at amino acid position 363. The value of aspartase activity was essentially identical to that of Y. pseudotuberculosis when valine was retained at this position (the Russian pestoides A, B, C, and D strains) or when serine was substituted for valine (the Russian pestoides E, F, G, and I strains). Curiously, the enzootic Angola and A16 strains from Africa contained phenylalanine at this position and exhibited significantly reduced but nevertheless significant activity (Table 2). Further, cells of the attenuated enzootic microtus biovar isolate from China (Fan et al. 1995) encode valine at amino acid position 363 of AspA (Zhou et al. 2004b) and should therefore exhibit enzymatic activity. These findings indicate that synthesis of enzymatically active AspA is a biomarker of attenuation in nonmuroid hosts. Further, the observations have obvious significance with respect to expression of acute disease as well as the evolution and potential spread of epidemic isolates.
Formal proof that active AspA in Y. pestis promotes attenuation will be, of course, dependent upon experiments involving salient gene exchange. Nevertheless, this correlation is presently absolute; thus, it is appropriate to consider possible mechanisms where genetic loss of AspA could enhance the ability to cause acute disease. The physiological role of AspA is to assure return of the four carbons comprising the backbone of aspartic acid (generated via catabolic transamination of L-glutamate with oxalacetate) into the tricarboxylic acid cycle in the form of fumarate (Halpern and Umbarger 1960). This mechanism assures the efficient flow of carbon by preventing accumulation of catabolic end-products other than CO2 and NH4+ (Table 3). As such, nearly ubiquitous AspA assures that a large number of saprophytic bacteria capable of its expression remain capable of efficient metabolism during growth in austere environments. Plague bacilli, however, typically reside in the closed, highly enriched, and protected environments of the host and flea vector where efficient processes of catabolism are unnecessary. Indeed, examination of mRNA transcribed during steady-state growth at 37°C showed that the catabolic process used by Y. pestis in the host is inefficiently regulated and largely wasteful if not prodigal (Motin et al. 2004). Accordingly, plague bacilli may have no need for the tempering influence of AspA on catabolism but rather proceed to consume and destroy host nutrient pools quickly and with abandon.
The ability of epidemic strains of Y. pestis to accumulate metabolic aspartate may have an additional sinister effect. Amino acids other than L-glutamate also serve as a source of carbon for synthesis of aspartate during starvation for Ca2+. These include metabolically linked L-proline as well as L-asparagine, which is in equilibrium with L-glutamine in the host (Soupart 1962) and rapidly and stoichiometrically undergoes conversion to L-aspartate in vitro (Dreyfus and Brubaker 1978). This event would be highly deleterious to the host should it occur in vivo because the largest amino acid components of mammalian serum and cytoplasm are L-glutamate, L-glutamine, and L-asparagine, whereas L-aspartate comprises the smallest (Soupart 1962). Reversal of this ratio by secretion of metabolic aspartate would seriously affect the metabolic equilibrium and bioenergetics of these pools by the host, an event that might be especially deleterious during the terminal stages of infection. In this context, it may be significant that only a few other aerobic bacteria also lack AspA, but this group includes Mycobacterium tuberculosis (Fleischmann et al. 2002), Francisella tularensis (Larsson et al. 2005), and rickettsiae (Andersson et al. 1998). The latter, like epidemic isolates of Y. pestis, also convert exogenous L-glutamate to L-aspartate, which is released into the culture medium (Bovarnick and Miller 1950).
Expression of AspA by and attenuation of enzootic isolates of Y. pestis still renders these strains fully capable of killing muroid rodents, suggesting that they possess some factor (other than that associated with AspA-deficiency) that contributes to this process. A likely possibility is pMT-encoded murine toxin, which is highly lethal in muroids but not essential for virulence of epidemic strains of Y. pestis in mice, although it does appear to be important for survival of plague bacteria in the midgut of the flea vector (Hinnebusch et al. 2000, Hinnebusch et al. 2002). It seems possible that the raison d'être for AspA in enzootic strains of Y. pestis is no more than that in the enteropathogenic yersiniae where it facilitates efficient amino acid catabolism in minimal environments. These strains are often, but not always limited, to grasslands and similar habitats containing burrowing muroid rodents (e.g., gerbils and voles), and their maintenance may depend in part on survival in soils including those surrounding decomposed host carcasses (Ayyadurai et al. 2008, Eisen et al. 2008).
Finally, the paucity of available enzootic isolates underscores the importance of further studies examining the mechanisms for the ecological persistence of these genetically and phenotypically diverse strains in sylvatic foci. Moreover, the behavior of these strains in fleas, particularly as it relates to plague virulence, pathogenesis, host immunity, and focal distribution, warrants further investigation. Needless to say, construction and virulence testing of AspA-positive epidemic isolates and murine toxin–deficient enzootic isolates should resolve many of the issues raised in this review.
This work was sponsored by the NIH/NIAID (National Institutes of Health/National Institute of Allergy and Infectious Diseases) Regional Center of Excellence for Bio-defense and Emerging Infectious Diseases Research (RCE) Program. R.R.B wishes to acknowledge membership within and support from the Region V “Great Lakes” RCE (NIH Award 1-U54-AI-057153).
No competing financial interests exist.