| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Department of Civil and Environmental Engineering, Massachusetts Institute of Technology, Cambridge, Massachusetts 02139
* To whom correspondence should be addressed. E-mail: mpolz{at}mit.edu
 |
Abstract |
|---|
 TOP Abstract Literature Cited |
|---|
The last two decades have seen a radical shift in our understanding of microbial diversity. Previously the number of bacterial and archaeal species had been estimated to be in the thousands. It is now generally accepted that the number may actually be as high as several million (Torsvik et al., 2002). This change was brought about by a gradual replacement of diversity estimates based on pure-culture isolation of strains with a determination of diversity based on co-occurring gene sequences, largely ribosomal RNA (rRNA) genes (Head et al., 1998). Although early attempts were made to screen diversity by shotgun cloning of environmental DNA, with subsequent detection and sequencing of rRNA gene inserts (Schmidt et al., 1991), large-scale application of the molecular approach was dependent on PCR protocols that allow the enrichment of rRNA genes from genome mixtures using universal primers (Head et al., 1998). Today, the assessment of the entire diversity of rRNA sequences (ribotypes) coexisting within specific microbial communities has become a realistic possibility due to the ease of PCR implementation and the increased availability of high-throughput sequencing facilities.
Although the exact magnitude of microbial diversity still remains an open question, the PCR-based approach has led to the retrieval of large numbers of sequences from almost any environment examined (Hugenholtz et al., 1998). Thus extensive comparative databases are now available from which patterns of microbial community structure are beginning to emerge. For example, studies of bacterioplankton diversity in the ocean, which represents one of the best-studied environments, have shown that, surprisingly, the major phylogenetic groups in the open ocean and the coastal ocean are similar, despite marked differences in trophic state and habitat quality between these environments (Giovannoni and Rappé, 2000). Such observations have provided important insights; yet they also highlight the major problem of the molecular diversity approach: because very few of the retrieved sequences have closely related cultured representatives available, the ecological role of an organism in question cannot even be guessed (Hugenholtz et al., 1998). Furthermore, even when closely related cultured organisms exist, they can display quite significant genomic, physiological, or metabolic differences (Gray and Head, 2001). New alternatives for diversity studies, such as analysis of large genome fragments retrieved from the environment (Béja et al., 2000) or gene cassette PCR for recovery of complete open reading frames from environmental DNA (Stockes et al., 2001), can enhance our understanding of uncultured organisms. Nonetheless, elucidation of structure-function relationships or niche differentiation of populations within microbial communities remains one of the big challenges in microbial ecology.
During the last few years, molecular diversity studies have been augmented with tracer techniques that allow assignment of biogeochemical function to uncultured microbial populations [recently reviewed by Gray and Head (2001)]. Most notably, combined microautoradiography and in situ hybridization (STAR- or MICROFISH) (Lee et al., 1999; Ouverney and Fuhrman, 1999; Cottrell and Kirchman, 2000) or stable isotope probing (Boschker et al., 1998; Radajewski et al., 2000) allow identification of microbial populations responsible for the metabolism of specific organic compounds. In both cases, environmental samples are incubated with isotopically labeled substrates. In STAR- or MICROFISH, microautoradiography and in situ hybridization are carried out on the same microscope slide with the goal of matching uptake of radiochemicals with phylogenetic identification on the single-cell level. In stable isotope probing, either lipid biomarkers (Boschker et al., 1998) or DNA (Radajewski et al., 2000) are extracted from communities incubated with 13C-labeled compounds. If cells grow on the added compounds, their pool of macromolecules will be isotopically heavy compared to those of metabolically inactive organisms. This makes it possible to identify the organism in one of two ways: (1) by mass spectrometry of labeled "signature" lipids (Boschker et al., 1998); or (2) by separation by ultracentrifugation of community DNA according to mass differences, followed by identification of rRNA genes in the isotopically heavy DNA pool by PCR, cloning, and sequencing (Radajewski et al., 2000).
The above approaches have already produced interesting insights into the ecological roles of uncultured Bacteria and Archaea. For example, using MICROFISH, it was demonstrated that low-temperature Archaea, which represent a dominant group in deep ocean water but are currently known only from rRNA gene clone libraries, readily take up amino acids at low ambient concentrations (Ouverney and Fuhrman, 1999). In another study, incubation of anaerobic sediments with 13C-acetate yielded signature lipids of gram-positive bacteria rather than those of the more readily isolated delta Proteobacteria sulfate-reducing bacteria (Boschker et al., 1998). However, each of the techniques has distinct drawbacks. Stable isotope probing requires very high substrate concentrations and long incubation times, so that the procedure actually resembles enrichment cultures (Radajewski et al., 2000); MICROFISH involves labor-intensive hybridization and microautoradiography, which limits the number of populations whose metabolism can be explored.
We are currently developing a combination of DNA microarrays and radiotracer incubations, the "functional diversity array," as a high-throughput complement to the above methods (Bertilsson and Polz, 2001). DNA microarrays can carry hundreds to thousands of specific nucleic acid probes, which are arrayed in discrete spots. During the hybridization process, these probes capture their specific target from mixtures of templates. If these are either radioactively or fluorescently labeled, the presence or, to a certain extent, the quantity of all specific templates for which probes have been spotted can be ascertained. The application of DNA microarrays for screening and monitoring microbial community structure by arraying rRNA-specific oligonucleotide probes is being explored by a number of laboratories (Cho and Tiedje, 2001; Small et al., 2001; Koizumi et al., 2002). In the functional diversity array, diversity screening is combined with detection of populations responsible for specific transformations in the community (Fig. 1). Samples are spiked with 14C-labeled compounds, leading to incorporation of radionuclides into rRNA of populations that actively metabolize the compound of interest (Bertilsson and Polz, 2001). RNA is subsequently extracted, fluorescently labeled, and hybridized to the microarray, which contains oligonucleotide probes specific for each "ribotype" in the community. Radioactivity in each spot due to hybridization can be determined by either microautoradiography or phosphor-imaging, so that in combination with the fluorescent signal, a specific activity can be estimated for each population (Fig. 1).
|
Quantification of the radioactive signal on arrays shows that the approach is sensitive enough to detect populations at naturally occurring levels. The use of phosphor-imaging screens, to which the entire array is exposed, allows 14C bound to each spot to be quantified. We have experimentally determined the detection limit for 14C on these screens to be 0.1 DPM for a spot 150 µm in diameter, which consists of about 108 oligonucleotide probes. Assuming that the rRNA molecules fragment to about 300 bp and that only 1% of the oligonucleotide probes will be bound to templates after hybridization, the detection limit is between 102 and 103 cells. This estimate is based on a cellular rRNA content between 1,000 and 10,000 molecules, which is in the range of slow- and fast-growing cells, respectively. Using this procedure, we have been able to specifically detect and differentiate rRNA from sulfate-reducing strains grown on 14C-labeled lactate (Klepac and Polz, unpubl. data).
For the array to represent the actual microbial populations responsible for metabolism of a specific compound, realistic substrate concentrations must be used in the incubations to avoid introducing a major bias in community structure due to selective growth of specific populations. In coastal waters, which we are using as a model ecosystem, we found that even low additions of 14C-labeled organic substrates (representing <3% of the total organic carbon) produced highly radiolabeled rRNA after 7 h incubation at in situ conditions. This incubation time is similar to the average generation time for the entire bacterial community (9 h), and both the uptake rate and the growth yield on the labeled substrates were linear during the incubation, suggesting that there were no major shifts in the microbial community. These tests also showed that the proportion of labeled C allocated to rRNA was strongly dependent on the quality of the substrate (e.g., 12%–19% for adenine, 1.1%–1.3% for acetate). In addition, tests with exponentially growing bacteria in pure culture showed that a constant fraction of the total cellular 14C (average 8% for Vibrio cholera and 17% for E. coli) could be recovered in rRNA after about one cell doubling. Thus, the major advantages of rRNA detection are the linearity of the labeling process and the possible limitation to few cell doublings, which ensure that community structure will be only minimally biased.
The third question, whether rRNA diversity can be ascertained with realistic effort, requires a reexamination of the PCR-based approach. We have recently presented the hypothesis that a simple, PCR-induced artifact may lead to severe overestimation of diversity of rRNA genes (Thompson et al., 2002). During the co-amplification of homologous templates with universal primers, a significant fraction (up to 50%) of products may be present as heteroduplices. These were increasingly prevalent as template diversity increased or primer availability became limiting (Thompson et al., 2002). After cloning, heteroduplex molecules may become subject to mismatch repair by the E. coli MutHLS system. This can theoretically lead to independent repair of each mismatched position since the repair system, in vivo, is directed by hemimethylation (Modrich, 1987), which is absent in PCR products. A model exploring the effects of heteroduplex repair demonstrated that the undirected repair process might be responsible for large overestimation of rRNA diversity (Thompson et al., 2002). For example, a simple system of 2 sequences with 3 shared mismatched positions can result in 8 sequence permutations; for 4 sequences with 10 shared mismatched positions, the number increases to 6136 (Thompson et al., 2002). Although this is a dramatic example, the potential contribution of heteroduplex repair to sequence diversity can easily be avoided by "reconditioning PCR," a low-cycle-number reamplification of a 10-fold diluted, mixed-template PCR product.
Although the exact contribution of heteroduplex repair to diversity estimates is still being analyzed in our laboratory, we have used the modified amplification protocol (reconditioning PCR) to estimate bacterial diversity in a coastal bacterial community. We generated a clone library from amplified 23S rRNA genes, then assessed sequence diversity in the library by a combination of rarefaction analysis and Chao-1 estimators, which are based on capture-recapture statistics (Hughes et al., 2001). The results demonstrated that diversity was relatively moderate, with the number of coexisting sequence types remaining in the low 100s (Acinas, Hunt, Bertilsson, and Polz, unpubl. data). This ongoing analysis is currently complemented with rarefaction of a 16S rRNA gene library derived from the same sample, demonstrating that it may indeed be possible to represent entire communities with reasonable effort on DNA microarrays.
|
Acknowledgments |
|---|
|
Footnotes |
|---|
Present address: Department of Evolutionary Biology, Limnology, Uppsala University, Norbyvägen 20, SE-75236, Uppsala, Sweden. 
The paper was originally presented at a workshop titled Outcomes of Genome-Genome Interactions. The workshop, which was held at the J. Erik Jonsson Center of the National Academy of Sciences, Woods Hole, Massachusetts, from 1–3 May 2002, was sponsored by the Center for Advanced Studies in the Space Life Sciences at the Marine Biological Laboratory, and funded by the National Aeronautics and Space Administration under Cooperative Agreement NCC 2-1266.
|
Literature Cited |
|---|
|
TOPAbstractLiterature Cited |
|---|
This article has been cited by other articles:
|
|
 |
|
  M. D. J. Lynch and R. G. Thorn Diversity of Basidiomycetes in Michigan Agricultural Soils Appl. Envir. Microbiol., November 1, 2006; 72(11): 7050 - 7056. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 E. G. Zoetendal, C. T. Collier, S. Koike, R. I. Mackie, and H. R. Gaskins Molecular Ecological Analysis of the Gastrointestinal Microbiota: A Review J. Nutr., February 1, 2004; 134(2): 465 - 472. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
