A Link Between Virulence and Ecological Abundance in Natural Populations of Staphylococcus aureus

See allHide authors and affiliations

Science  06 Apr 2001:
Vol. 292, Issue 5514, pp. 114-116
DOI: 10.1126/science.1056495


Staphylococcus aureus is a major cause of severe infection in humans and yet is carried without symptoms by a large proportion of the population. We used multilocus sequence typing to characterize isolates of S. aureus recovered from asymptomatic nasal carriage and from episodes of severe disease within a defined population. We identified a number of frequently carried genotypes that were disproportionately common as causes of disease, even taking into account their relative abundance among carriage isolates. The existence of these ecologically abundant hypervirulent clones suggests that factors promoting the ecological fitness of this important pathogen also increase its virulence.

Staphylococcus aureus is one of the most important bacterial pathogens of humans. The continuing burden of community- and hospital-acquired S. aureus disease, including serious endovascular, wound, bone, and joint infections (1, 2), is a major public health concern. This concern is heightened by the increased prevalence of antibiotic-resistant strains such as methicillin-resistant S. aureus (MRSA) and glycopeptide-insensitive S. aureus(GISA) (3). Although up to 30% of the population of the United Kingdom carry S. aureus in their nostrils without symptoms, the annual reported incidence of bacteremia is less than 0.02% (4). Here, we address the question of whether all S. aureus are equally virulent, and infection is purely opportunistic, or whether invasive disease is primarily caused by a subset of particularly virulent genotypes unrepresentative of the carriage population as a whole.

For several pathogens that are carried asymptomatically, it has been shown that certain clones responsible for cases of invasive disease may also be abundant in the carriage population (5, 6). However, in the absence of a representative population framework, it is not possible to deduce whether these clones are atypically virulent, or whether the apparent association between specific clones and disease simply reflects a higher rate of dissemination within the carriage population. To place isolates recovered from invasive S. aureus disease within the broader context of the species as a whole, we compared 61 bacterial isolates from patients with serious community-acquired invasive disease with 179 isolates recovered from the nostrils of healthy individuals living in the same community. We also studied isolates from 94 contemporaneous cases of hospital-acquired disease occurring in hospitals serving the study population, representing a different and clinically important epidemiological setting. All isolates were collected within Oxfordshire between 1997 and 1998 during a prospective case-control study to define host and bacterial factors associated with endemic invasive S. aureus disease (7).

We used multilocus sequence typing (MLST) to compare the isolates (8). Alleles at seven unlinked housekeeping loci are identified by sequencing ∼450–base pair internal fragments of the genes (9, 10), and the sequence type (ST) of an isolate is defined by the alleles at the seven loci. There is an average of 22.3 alleles per locus; hence, MLST could potentially resolve >1 billion STs. A clone is defined as a set of isolates identical at all seven loci.

The 334 isolates belonged to 187 STs; 26 STs were represented by more than one isolate (11). The carriage isolates were significantly more diverse than the disease isolates (12). Forty-eight percent (29/61) of community-acquired disease isolates belonged to just five STs, a degree of clonality consistent with the findings of a previous large study of disease isolates conducted using multilocus enzyme electrophoresis (13), whereas the five most common carriage STs account for just 14% (25/179) of the carriage isolate population. Thus, in the community, isolates causing disease are not drawn randomly from the carriage population.

Most of the isolates fall into clusters of closely related STs, or clonal complexes, and it is likely that isolates in each of these clusters have descended from a single ancestral genotype. To explore this possibility, we used an algorithm that first defines clonal complexes as groups in which each isolate is identical to at least one other isolate at five or more of the seven loci (14) (Fig. 1). In each of the 12 major clonal complexes, we have identified what we believe to be the “ancestral genotype,” assigned as the ST differing from the highest number of other STs within the complex at only one locus. The ancestral genotype also corresponded to the largest clone in the clonal complex in 11 out of 12 cases, providing independent support for these assignments (15). After identifying putative ancestral genotypes, “single-locus variants” (SLVs) were identified. These are assumed to be direct descendents of ancestral genotypes, having undergone changes at a single locus, either by point mutation or recombination, but having remained identical to the ancestral genotype at the other six loci (16, 17). A comparison of the frequencies of variant alleles within SLVs with their predicted ancestral counterparts provides further strong support for the ancestral assignments (18). In summary, a clonal complex is a set of genotypes derived in the recent past from a common ancestor. Clonal ancestors, and direct descendents of these ancestors, can be identified with some confidence within each clonal complex.

Figure 1

Diagram of clonal complexes. Each number represents an MLST sequence type (ST). Where an ST is represented by multiple isolates, the number of isolates with that ST are shown in parentheses. Green numbers denote nasal carriage isolates, red numbers, community-acquired invasive disease isolates, and blue numbers, hospital-acquired disease isolates. No inferences are made concerning the relations between clonal complexes. The central circle of each clonal complex contains the ancestral clone of each clonal complex. Single-locus variants (SLVs) of an ancestral clone lie within the next (solid line) concentric circle, and double-locus variants within the outer (dotted line) circle. A solid straight line between two STs denotes a single-locus difference between them, a dashed straight line, a double-locus difference. Three pairs of related STs where the ancestral genotype cannot be predicted are also shown. Singletons are isolates possessing STs that differ from those of all other genotypes at >2 loci. In two of the clonal complexes, some SLVs were assigned as secondary ancestral clones because they differed by a single locus from at least two other genotypes that had not already been assigned as SLVs. Secondary ancestral clones and associated clonal variants were treated as primary ancestral clones in the analysis and were assigned in rank order according to the number of SLVs they define (for further details see the BURST readme file The clonal complexes are named according to the ST of the primary ancestral genotype, but with the prefix “CC” (for clonal complex). Isolates of ST1 (MSSA) and ST36 (EMRSA-16) are being sequenced at the Sanger Centre (23).

With two minor exceptions, all clonal complexes contained isolates recovered from community-acquired disease, nosocomial disease, and asymptomatic carriage. Thus, invasiveness is not a property of a few rare genotypes with unusually high virulence, and previous observations that particular pathogenic clones are commonly carried (5, 6) can be extended to the bacterial population as a whole. Clones that are most virulent within the community have also become common causes of disease within hospitals, in many cases having acquired resistance to multiple drugs in response to the hospital environment. Both methicillin-susceptible and MRSA isolates that cause hospital-acquired infections typically have ancestral genotypes, and there is also evidence for a similar trend in GISA isolates (19).

However, the crucial observation (Table 1) is that a significantly higher proportion of the disease isolates had an ancestral genotype [114/155 (74%), compared with 39/179 (22%) in the carriage sample, odds ratio for disease 10 (95% confidence intervals or CIs 5.5 to 18)]. Thus, even after taking into account their ecological abundance, isolates with ancestral genotypes (ancestral clones) remain disproportionately associated with disease. Furthermore, as ancestral clones have diversified by point mutation and recombination to form clonal complexes, there appears to have been an associated loss of virulence. Such an effect is more likely to be the result of recombination than point mutation, as recombina- tion may also change loci neighboring the MLST genes. Evidence of an association between recombination and loss of virulence is apparent in the data; SLVs recovered from asymptomatic carriage were more likely to have arisen by recombination than those recovered from invasive disease (P < 0.001) (18). It is interesting that this loss of virulence was much more closely associated with recombination at arcC, tpi, or pta, than with recombination at the other four loci, indicating that there may be virulence factors closely linked to these loci.

Table 1

Distribution of disease-causing and nasal carriage isolates within clonal complexes (see Fig. 1). A 4 by 2 Fisher's exact test comparing the distribution within the ancestral clones, single-locus variants, double-locus variants, and satellite strains of community-acquired disease with the distribution in nasal carriage isolates is highly significant (P < 0.0001). However, there is no statistical difference between isolates recovered from community- and hospital-acquired disease (P = 0.86).

View this table:

Why should the founding genotypes of clonal complexes be the most virulent, and why do clonal complexes exist at all? The presence of clones is often interpreted as reflecting low rates of recombination, but this is unlikely to be the case in S. aureus as we have shown that an allele is approximately eight times more likely to change by recombination than by point mutation (18). Alternatively, it is possible that ancestral genotypes carry a strong selective advantage, such that they spread sufficiently quickly to outrun the diversifying effects of recombination to become an observable clone (20, 21).

If ancestral clones are both ecologically successful and disproportionately more likely to cause invasive disease, there may be a causal relation between fitness and virulence. In support of this, nasal carriage isolates with genotypes identical to those of invasive disease isolates were more likely to be recovered from both nostrils than those with genotypes unique to the carriage population (95% and 70%, respectively; P = 0.002). Similarly, carriage isolates within ancestral clones were more likely to colonize both nostrils than those not belonging to an ancestral clone (90% versus 71%, P = 0.01). These observations suggest that isolates corresponding to a virulent genotype and those belonging to ancestral clones are more successful colonizers than other genotypes.

Within clonal complexes the odds ratio for disease between ancestral clones and their putative descendants is 9.4 (95% CI, 5.0 to 17.6; P < 0.0001); that for colonization of both nostrils is 3.7 (1.1 to 12.5, P = 0.02). We speculate that greater virulence and greater propensity for colonization are pleiotropic effects of the same genetic change, and that, as ancestral clones diversify by point mutation and recombination to form clonal complexes, there is an associated loss of virulence and ecological fitness. As differential colonization of one or both nostrils is only an indirect indication of fitness, loss of fitness with clonal diversification cannot be demonstrated directly from the data.

As invasive disease is relatively very rare and unlikely to contribute to transmission to new hosts, enhanced virulence itself is unlikely to explain the ecological success of ancestral clones. An alternative explanation is that genetic factors promoting aggressive colonization also cause localized tissue damage, providing an increased likelihood of access to the blood stream and, hence, invasive disease. Such a link has been drawn between attachment factors (fimbriae) inEscherichia coli and urinary tract infection (22).

We conclude that clonal complexes arise in the natural staphylococcal population despite high rates of recombination, probably because the founding genotypes of these complexes carry a strong selective advantage. These founding clones are initially highly virulent; however, as the clones diversify (predominantly via recombination), the ability to cause invasive disease declines rapidly. This analysis highlights the importance when studying commonly carried bacterial pathogens of placing disease isolates in the context of the epidemiologically relevant carriage population. In the case of S. aureus, the results suggest that hypervirulent clones are abundant in the bacterial carriage population and that S. aureus is not solely an opportunistic pathogen.

  • * To whom correspondence should be addressed. E-mail:{at}

  • Present address: Department of Biology and Biochemistry, University of Bath, Bath, BA2 7AY, UK.

  • Present address: Department of Infectious Disease Epidemiology, Imperial College School of Medicine, St. Mary's Campus, Norfolk Place, London, W2 1PG, UK.


View Abstract

Stay Connected to Science

Navigate This Article