Laurie A. Achenbach is an associate professor and John D. Coates is an assistant professor in the Department of Microbiology and Center for Systematic Biology, Southern Illinois University, Carbondale.
Links to Other ASM Pages:
Disparity between Bacterial Phylogeny and Physiology
Comparing 16S rRNA sequences to assess relationships can be a powerful tool, but its limitations need to be considered
Laurie A. Achenbach and John D. Coates
Microbial diversity studies and molecular community profiling without culture isolation largely depend on small-subunit ribosomal RNA sequence analyses. Examination of rRNA sequence data from some closely related prokaryotes in the Proteobacteria revealed high sequence similarities associated with physiologically disparate organisms. Although phylogenetic interpretations are possible, physiologic or metabolic characteristics of community members cannot be extrapolated from rRNA sequence data alone.
Analysis of 16S rRNA sequences has long been used to assess the relationships of bacterial species and, indirectly, to predict the physiological characteristics of an organism (Giovannoni et al., J. Bacteriol. 170:720-726, 1987). In many cases, close relatives share metabolic capabilities, and this fact has been used to identify previously unrecognized metabolisms in known bacterial isolates (Coleman et al., Nature 361:436-438, 1993). Arising from this observation is the implication that the phylogeny of an uncultured bacterium can be used to successfully guide the isolation strategy or predict metabolic function in the natural environment. Although there are numerous cases in which close relatives are similar phenotypically, the utility of 16S sequence data in predicting physiology must be viewed with caution. For example the Rhodocyclus group in the b subclass of the Proteobacteria form a very tight cluster based on 16S rDNA sequence analyses but is composed of widely disparate metabolic capabilities including strict anaerobiosis, facultative anaerobiosis, perchlorate respiration, phototrophy, and iron respiration, and 16S sequence data alone cannot be used as a predictive indicator of any of these physiologies.
During the past two years, we have been studying bacteria that can utilize perchlorate as an electron acceptor for anaerobic respiration. These studies have resulted in the isolation of a number of organisms that are members of the Proteobacteria, a major subdivision of the gram-negative bacteria. At present, the Proteobacteria contains five subclasses, a , b , g , d , and e , which are currently defined based on 16S ribosomal RNA gene sequences. Using 16S sequence data, most of the isolates we obtained are closely related to each other and to the bacterial species Rhodocyclus tenuis and Ferribacterium limneticum in the b subclass. The perchlorate-reducing isolates are all motile, nonfermentative facultative anaerobes that couple growth to the reduction of perchlorate, chlorate, or oxygen. R. tenuis is a phototrophic nonsulfur purple bacterium that contains bacteriochlorophyll and is found on soil surfaces and in shallow waters exposed to sunlight. F. limneticum is a strict anaerobic, nonfermenting, dissimilatory Fe(III)-reducer (Cummings et al., Arch. Microbiol. 171:183-188, 1999). To date, its environmental significance is unknown since only a single isolate has been described. Although the perchlorate-reducing bacteria are closely related to these organisms, they exhibit distinct physiologies. None of the perchlorate-reducing isolates can grow by phototrophy or Fe(III)-reduction. By the same token, F. limneticum does not grow by phototrophy or by the reduction of perchlorate, and R. tenuis cannot grow by anaerobic respiration with a broad range of electron acceptors including perchlorate or Fe(III).
Despite their phenotypic differences, these dissimilar microbial groups are phylogenetically closely related to each other (see figure in online version). For example, the perchlorate-reducing strain SIUL 16S gene sequence is 95.1% similar to R. tenuis and is even more closely related to F. limneticum (97.6%). However, strain SIUL is only 93.6% similar to another perchlorate-reducing isolate, strain PS. As the currently accepted definition of genus level relationships in bacteria is 95% or greater 16S rDNA sequence similarity (Amann et al., Microbiol. Rev. 59:143-169, 1995), these two perchlorate-reducing isolates represent two different genera within the Rhodocyclus group despite their phenotypic similarities.
An even more striking example in the b subclass of the Proteobacteria can be found in the perchlorate reducer, strain RCB. Although RCB shares 99.5% 16S rDNA sequence similarity with F. limneticum (see figure in online version) which would imply that they are the same species, strain RCB is not a strict anaerobe and is incapable of dissimilatory Fe(III) reduction, two distinct characteristics of F. limneticum.
Other examples can be found outside the b subclass of the Proteobacteria. One such example is the close phylogenetic relationship of perchlorate-reducing strain WD with Magnetospirillum gryphiswaldense, a member of the a subclass (see figure in online version). All members of the Magnetospirillum genus that have been isolated to date form magnetosomes—an intracellular form of magnetite—when growing microaerophilically on iron-based media, which gives these organisms a unique magnetotactic characteristic. Although strain WD shares 96.3% 16S rDNA sequence identity with M. gryphiswaldense, which would suggest that it is a Magnetospirillum species, strain WD does not produce the fine-grained magnetite characteristic of magnetotactic bacteria (Dennis Bazylinski, personal communication). This observation is supported by a recent report of an isolate, CC-26, which is even more closely related to a Magnetospirillum species (98.3% sequence similarity) and yet is also incapable of producing magnetosomes (Shinoda et al., Appl. Environ. Microbiol. 66:1286-1291, 2000). In addition, none of the Magnetospirillum species tested could grow by dissimilatory perchlorate reduction.
These results show that several distinct phenotypes [phototrophy, chlorate reduction, Fe(III) reduction, strict anaerobiosis, and facultative anaerobiosis] are represented in closely related organisms and that the 16S rDNA sequence data for these organisms is not informative of metabolic ability. If, for example, F. limneticum represented an uncultured organism for which only the 16S sequence was available, one would predict (incorrectly) that this organism would most likely be a facultative anaerobe that is able to reduce perchlorate. The possibility of being misled when attempting to predict physiology based entirely on phylogenetic analysis is borne out by Rhodocyclus-type 16S sequences deposited in GenBank from uncultured and/or undescribed organisms. One such sequence, strain B2 (GenBank accession no. AFO3504) is 96.5% similar to the perchlorate reducer strain SIUL but only 93.1% similar to another perchlorate reducer, strain PS. As strain B2 is 95.3% and 94.3% similar to F. limneticum and R. tenuis, respectively, one cannot predict with confidence which type of metabolism strain B2 represents or even if it is a strict or facultative anaerobe.
In the last two decades, many advances have been made in molecular biology techniques and many of these are being used to assess microbial diversity and population structure in the environment. There can be no doubt of the usefulness of these molecular approaches in identifying the presence of various microorganisms in complex microbial communities. However, with increasing application, there is an alarming trend towards the inference of metabolic functionality of the microbial communities and even individual microbial species identification based entirely on 16S rDNA datasets without supporting isolation or phenotypic characterization. As microbiologists, we are concerned with the fallacy of such data interpretation. As most organisms are capable of a broad range of metabolic pathways, it is still impossible to predict which form of metabolism an organism would use to survive in its natural environment. Thus, those researchers analyzing 16S sequences in order to design a cultivation scheme or simply to associate a metabolic function with an organism in a mixed culture must view the phylogenetic conclusions with appropriate caution.
Email: webmaster@asmusa.org