Contention:‘Community-scale characterizations based on cultivable microbes are so unrepresentative that they have no utility in contemporary environmental microbial ecology’.
Microbes are, by definition, very small. Individual microbial cells and isolated fungal hyphae are invisible to the unaided human eye, and remained intangible until the invention of microscopes in the 17th Century. Aggregations of microbial cells to form larger-scale and visible entities occur naturally and can be spectacularly manifest, for example as bacterial slimes, myxomycete plasmodia, and fungal basidiocarps. In the history of microbiology, there was an imperative not just simply to visualize microbial cells, but to enumerate them and understand their form, physiology, function and relationships with their environment. This required the isolation, purification and multiplication of cells to provide sufficient biomass to enable analyses. This was conveniently achieved using dispersal and dilution techniques, most commonly in an aqueous medium, followed by modest enrichment on semisolid media, such that ostensibly separated cells proliferated and formed visible colonies; enumeration was then based on these so-called CFUs, in recognition of the fact that it was not necessarily a single cell that was the originator. Such techniques were pioneered by Koch in the 1880s, who first used cut surfaces of boiled potatoes as an enrichment surface, and then adopted hydrated gelatine films, which were better suited to visualization of colonies. Methods then evolved which adopted other gelling agents, most notably agar, and a huge variety of nutritional supplements to such media as it became apparent that different organisms grew more or less effectively on a variety of substrates. Media were devised that were apparently selective to restricted groups or even species of microbe.
These techniques revolutionized both medical and environmental microbiology. In a medical context, the ability to isolate disease-causing microorganisms in a selective manner enabled Koch to formulate and apply his exemplary postulates, which led to significant advances in pathology. In environmental science, a plethora of microorganisms was isolated from soils and waters, which, when characterized, revealed their roles and significance in carbon and elemental cycling, biotic interactions and ecological functioning. The diversity of organisms thus encountered was beguiling, and there was a generally implicit assumption that these were the significant players in the ecosystem, not least because they were amenable to study. However, pioneering soil microbiologists noted that there was a persistent discrepancy between the number of bacterial cells that could be directly observed in smears of diluted soils when viewed using a microscope, and the apparent numbers of bacteria established by dilution-plating. It was generally found that of the order of 1% of the extant number of bacterial cells in soils were expressed in culture systems, whatever the composition of the media adopted or associated growth conditions. This lead to the realization that, as Russell (1923) noted, ‘the number and variety of bacteria existing in the soil is so enormous that themethods of separating out all the different forms and of displaying their characters and functions have proved impractical’.
But this did not curtail the Great Isolationists such as Beijerinck, Winogradsky, Omeliansky, Lipman and Waksman from culturing remorselessly and studying the actions and reactions of their microbial zoos. In the early days, the emphasis was on the study of individual organisms, which was carried out almost exclusively in vitro. This was desirable and crucially important, as it provided the foundation of understanding the basic biology of the microorganisms. It also formed the basis of microbial taxonomy, where growth characteristics, including colony morphology, on particular substrates became taxonomic characters. However, this de-emphasized the fact that in natural situations, all organisms operate in the multi-facetted physical, chemical and biological context of such environments. Whilst complexity of the soil environment at least was appreciated by some, its true extent is only now becoming apparent with the advent of molecular biology and ultra high-resolution imaging techniques.
Since those pioneering days, literally millions of Petri plates and culture tubes have been poured into and pored over, and an uncountable number of microorganisms studied.
The use of cultural enrichment techniques has its place where it is clear that such approaches are appropriate to the circumstances in which they are being applied; for example,
as a means of procuring organisms for fundamental biochemical, physiological, genetic or developmental studies where growth in vitro is a prerequisite to provide adequate biomass, purity of sample and then a means of study;
for screening and isolating organisms for potential biotechnological application where industrial-scale growth in process systems, and hence amenability to growth in such systems, is required;
where selective media are required to demonstrate the presence, and possibly the magnitude, of particular microorganisms in a system, and it has been proven that such expression is representative and relatable to the objectives of the study.
Whilst the pioneers of environmental microbiology recognized that in terms of numbers at least, there are orders-of-magnitude disparities between direct counts and CFUs, they were generally unconcerned about representativeness. This was largely a pragmatic viewpoint, as there were no adequate means then available of testing the assertion. However, there are now many lines of evidence that show the cultivable subsets of microorganisms present in the environment do not reflect accurately the overall properties of the communities from which they are derived. As the number of media which can be used to support growth of microorganisms from environmental samples is practically limitless, and the fraction that grow appears to be contingent on the state of the culture system, it is arguable that it is only ever possible to express a cultivable fraction of an extant community, rather than the cultivable fraction. There are instances of claims to elucidating the ‘total numbers of cultivable bacteria’ in environmental samples, which are absurd.
In relation to enumeration
The disparity in apparent numbers (typically 0.1–5%) between directly-observable cells in environmental samples, particularly soils, and numbers of CFUs observed in the earliest days of environmental microbiology has prevailed. Vital stains and radio-labelling techniques have often demonstrated that many of the cells visible by direct microscopic observation are alive and often metabolically active (e.g. see Torsvik et al., 1997); it is reasonable to postulate that such cells are therefore functionally active in the environmental context. As the greater proportion of such cells are not manifest in culture systems, relating the cultivable fraction to environmental circumstances and any associated functions is unlikely to be informative.
The relative proportions of such cells (i.e. cultivable vs. visualized and alive vs. total visualized) also varies. Hence it is not appropriate to argue that simple conversion factors could be used in compensation. It is pertinent to note that the apparent proportion of cultivable cells appear to be affected by changes in environmental conditions, which informs hypotheses as to the mechanisms behind the phenomenon of noncultivability. The situation is further complicated by the ‘viable but nonculturable’ (VBNC) syndrome, in which apparently culturable microorganisms enter a state following cultivation where they are no longer expressed under ostensibly amenable conditions (e.g. Roszak & Colwell, 1985; Binnerup et al., 1993).
In relation to community structure
Compelling evidence for an overall lack of representativeness in the cultivable fraction comes from genetic analyses which have been only been carried out within the past two decades. The earliest evidence came from the pioneering work of Torsvik et al. (1990), who quantified the complexity of combined DNA derived from entire microbial communities by measuring the melting properties and reassociation (or reannealing) kinetics of such community DNA, relative to that of single bacterial strains. This work suggested that of the order 4000 distinct ‘bacterial genome equivalents’ were present in 180 g forest soil (where the direct counts were 1010 g−1 and cultivable numbers 107 g−1 on soil extract agar). This may have been of a similar order to the then known number of soil-dwelling bacterial species in the global collection of bacterial culture collections. Subsequently, the reassociation kinetics technique was used to measure the relative diversity of a hierarchical series of community DNA samples derived from an arable soil (Ritz et al., 1997). These comprised (i) whole-community DNA obtained by direct extraction from the entire soil, (ii) an extractable bacterial fraction obtained by differential centrifugation of bacterial cells and (iii) a cultivable microbial fraction obtained by growth on nine contrasting media ranging from oligotrophic to copiotrophic, which included both prokaryotes and eukaryotes. Here, it was found that the cultivable subset was approximately two orders of magnitude more complex than a single bacterial strain, the extractable bacterial fraction was approximately three orders of magnitude more complex, and the whole-community DNA was some orders of magnitude more complex still – so much so that it was not really tenable to extrapolate to genome equivalents. This then was further evidence that the diversity of even a cultivable fraction which was strenuously designed to be comprehensive, was of the order of considerably <1% of the indigenous diversity.
These broad-scale analyses were soon superseded by a number of higher-resolution genetic analyses of aquatic and soil communities (e.g. Ward et al., 1990; Borneman et al., 1996; Borneman & Triplett, 1997; McCaig et al., 1999). When broad-scale PCR amplification, cloning and sequencing techniques were applied to such community DNA, the majority of nucleotide sequences, principally targeting ribosomal DNA, which were extracted from such samples were ‘unknown’. This meant that such sequences did not exist in extant databases, which at that time were populated entirely with data relating to cultivable microorganisms. Furthermore, in soil samples particularly, the huge diversity suggested by the reassociation kinetics studies was being confirmed – almost every sequence was different, even where hundreds of clones were analyzed. Rarefaction curves that were new to ecology were being plotted – they were linear, and they did not decline over hundreds of organisms sampled. The implication is that soil biodiversity, particularly at the prokaryotic level, is vastly greater than perhaps even Russell suspected (Curtis & Sloan, 2004). In other studies from a variety of ecosystems, sequences derived from community DNA relating to cultivable bacteria varied from 2% to under 50%, even in cases where cultivable subsets were derived from the same soil as the community DNA was extracted, amplified and sequenced (e.g. Chelius & Triplett, 2001; Kaiser et al., 2001; Smith et al., 2001).
Reasons why cultivable subsets are not representative have long been discussed. The two main hypotheses are that not all CFUs relate to a single cell and that the culture conditions do not adequately replicate the natural environment in which the microorganisms normally grow and function. Both are likely to be true, with the latter being more logical as the primary reason. Culture conditions are generally very simple and homogeneous, in stark contrast to natural environments. Soils in particular are extremely diverse and heterogeneous in relation to both their chemical and physical constitutions. Gel- and liquid-based media are far removed from the complex porous medium that soil represents. Another crucial factor is that organisms, which have evolved in the context of a community, are affected by the presence and status of other organisms in their vicinity, and it is known that there is a plethora of positive and negative interactions between microorganisms. In culture systems, communities which develop are by definition constituted from organisms that will grow in such a context, and the subsequent development of the community will be in part contingent on the local actions and reactions that develop; in many ways, what is manifest is not so much a reflection of the original community in the environment, but a new ‘local ecology’ within the culture system. Systems continue to be devised that enable hitherto ‘uncultivable’ microorganisms to be grown in vitro, as explained in the companion article, but these are generally targeted toward individual isolates. For the reasons outlined above, it is unlikely that in vitro systems will ever be broadly representative.
In vitro community-level physiological profiling: Biolog
Prior to the advent of molecular biology, bacterial taxonomy was largely based upon the growth properties of bacteria – their ability to utilize certain substrates, or grow under particular environmental conditions. The substrate utilization properties of bacteria were exploited in a system developed by Bochner (1989) and marketed by the Biolog® Corporation, based upon a microtitre-plate technique whereby the utilization profile of a bacterial culture (determined to be a single species or strain) could be readily ascertained, and used to identify the organism by matching to utilization profiles associated with known types. In a laudable piece of lateral thinking, Garland & Mills (1991) adapted the technique to study the ability of soil communities to utilize the substrates present in the Biolog repertoire, by inoculating soil suspensions, rather than cultured isolates, into the system. The technique, termed community-level physiological profiling (CLPP), was widely adopted and has been applied in hundreds of ecological contexts. There is no doubt that the multivariate substrate utilization profiles thus produced show a variety of sensitivities and responses to all manner of environmental circumstances. However, each instance of a Biolog plate is essentially 96 tiny liquid culture vessels, containing a micro-ecology of cultivable microorganisms, with many of the pitfalls that this entails (see Preston-Mafham et al., 2002). Indeed, Garland (1997) acknowledged that the ecological interpretation of CLPP profiles needs due care. This is particularly the case because the method is relatively straightforward to apply and rapidly yields large volumes of data, which have perhaps been as beguiling to contemporary environmental microbiologists as the Petri dish was to their forebears.
In relation to fungi
In this debate, fungi have a special place. The majority of fungi appear to be cultivable in vitro, notwithstanding that some of them are very fastidious. There are some exceptions, which are mainly obligate symbionts, whether mutualistic or antagonistic. However, filamentous fungi have a growth form based upon extending and branching hyphae, which form the mycelium. Mycelia are indeterminate structures that can vary in extent from micrometers to kilometers. If mycelia are disrupted, hyphal fragments can regenerate to form new mycelia. Furthermore, the majority of fungi also reproduce by means of asexual and/or sexual spores; in some cases different forms of spores are produced by one species, such as so-called macro- and micro-conidia of Fusarium spp. The indeterminacy of the mycelium results in a growth form particularly well-adapted for living in physically structured environments such as soil. It also complicates means of characterizing and quantifying fungal communities. In relation to fungi, whilst it could be argued that colonies formed on gel media are at least more likely to be indicative of types present than for prokaryotes, it is very difficult to relate the number of such CFUs to the status of the fungal community in the original environment. This is because the number of CFUs formed will not only relate to the local conditions developing within the microcosm of the culture system, but also the frequency of spores present in the environment and the degree of disruption of the mycelia in the sample preparation procedure. This is compounded by the fact that some (but not all) fungi are extremely prolific spore formers, and the proneness to hyphal disruption and subsequent propensity for regeneration from hyphal fragments is also very variable between types. Hence relating fungal CFUs to a functional context in an ecological sense is very difficult, and ultimately any attempt to quantitatively describe a fungal community using this metric is very limited.
In microbial ecological terms, application of cultural techniques to isolate putatively environmentally-active microorganisms, or to characterize microbial communities in terms of numbers, biomass or composition (i.e. ‘biodiversity’), is not appropriate and no longer admissible! However, old habits die hard. Manuscripts are still being doggedly submitted, and papers still being published, involving community characterization by CFUs on agar media and physiological profiling in vitro (Web of Knowledge hits for ‘CFU and soil’ returns 70, 73 and 72 for 2006, 2005 and 2004, respectively). There are circumstances where such techniques are appropriate. But in contemporary environmental microbial ecology they are very few, and any persistence in their general application will distort thinking. The early soil microbiologists were convinced that they could relate the communities they characterized and enumerated in their Petri dishes to soil fertility. They were thwarted by some naivety and a misunderstanding of the true nature of the system they were studying, and the tools they were using. We now know better.