Molecular Methods For Analyzing Uncultured Bacteria

15.2.1 Phylogenetic Analysis of 16S Ribosomal RNA Gene Sequences

The phylogenetic analysis of 16S ribosomal RNA (rRNA) gene sequences3,4 is the most widely used method for cultivation-independent microbiological surveys of environmental samples. DNA is extracted from a sample and is used as template for the specific amplification, by polymerase chain reaction (PCR) with appropriate primer sets, of bacterial 16S rRNA genes. The amplified 16S rRNA genes are cloned and sequenced. The obtained sequences are compared, by phylogenetic methods, with known reference rRNA sequences in order to infer the phylogenetic affiliations of the different bacteria in the environmental sample. Thousands of suitable reference sequences have already been deposited in public databases. This approach is invaluable for microbial ecology and has not only provided insight into the microbial diversity of many different habitats, but has also led to the discovery of novel, yet uncultured prokaryotic lineages.5

However, the analysis of 16S rRNA gene libraries has some pitfalls, which must be considered in order to avoid misinterpretations of the results. A library that consists of amplified and cloned 16S rRNA genes does not often correctly represent the real bacterial biodiversity in an environmental sample. Inefficient extraction of DNA and biases of PCR and of cloning can cause pronounced differences between the frequencies of specific 16S rRNA genes in the library and the frequencies of the corresponding organisms in the sample.6-8 These problems can hamper even the detection of those ecological key players, which are highly abundant in a habitat.9 Therefore, it is recommended to apply different DNA extraction methods on the same sample, to pool the DNA that was isolated by these methods prior to the amplification of 16S rRNA genes, to use different 16S rRNA gene-specific primer sets in parallel PCR experiments, and to use different vectors for cloning.10 Despite these precautions, in most cases, it will not be feasible to obtain the 16S rRNA gene sequences of all bacterial species living in a complex environmental sample. Already a single library established from such a sample can contain thousands of different 16S rRNA genes. The random selection of clones for sequencing is a simple approach to analyze such libraries, which has the disadvantage that the less frequent 16S rRNA genes may not be encountered while the more frequent 16S rRNA genes are sequenced repeatedly. Alternatively, one could use membrane hybridization with specific rRNA-targeted nucleic acid probes or PCR with highly specific primers to screen rRNA gene libraries for members of different phylogenetic lineages. Nucleic acid fingerprinting techniques, as for example restriction fragment length polymorphism (RFLP) analysis11 and denaturing gradient gel electrophoresis (DGGE),12 are other useful tools for systematic analyses of 16S rRNA genes in environmental samples. RFLP analysis can reduce the sequencing effort needed to retrieve a high number of different, already cloned rRNA genes from a library. Genes with identical RFLP patterns are collected in groups, and then only a few representatives of each group are sequenced and further analyzed. DGGE separates different PCR-amplified rRNA gene fragments in polyacrylamide gels with increasing denaturing gradients. In the ideal case, each band in the gel represents one organism in the environmental sample. Single bands can be excised, re-amplified by PCR, and sequenced for phylogenetic analysis. DGGE has the disadvantage that it can separate only fragments of 16S rRNA genes with a maximum size of approximately 500 base pairs. Partial 16S rRNA sequences of this length are of limited use for phylogeny and for the design of rRNA-targeted nucleic acid probes.

Due to the biases of DNA extraction, PCR, and cloning, the quantitative composition of bacterial communities cannot be inferred from the frequencies of different 16S rRNA genes in clone libraries. For the same reason, the intensities of the bands in fingerprints of PCR-amplified rRNA genes do not necessarily correlate with the in situ abundance of the corresponding bacteria. Recently developed quantitative variants of PCR (real-time PCR13,14 and competitive PCR15-17), or PCR-independent methods like quantitative dot blot18,19 and fluorescence in situ hybridization (FISH) (discussed in the next section), should be applied for quantitative analyses of microbial communities.

15.2.2 Fluorescence In Situ Hybridization

Fluorescence in situ hybridization (FISH) with rRNA-targeted probes is a cultivation-independent method to identify microorganisms and to visualize microbial cells directly in their habitats.2'20 The ribosomal RNAs are optimal targets for in situ hybridization. These ubiquitous molecules have a highly conserved structure and are essential components of the ribosomes in all living cells. The high number of ribosomes in a microbial cell causes a natural amplification of the probe-conferred fluorescence. A specific rRNA-targeted oligonucleotide probe must be sequence complementary to a short base sequence, which exists only on the ribosomal RNA of the probe target organisms. The nontarget organisms should have a different base sequence at the probe-binding site and in all other regions of the ribosomal RNAs. Functions to automatically design new rRNA-targeted probes and to match probes against known rRNA sequences are offered, for example, by the ARB software package21 and by the Ribosomal Database Project.22 A comprehensive collection of available rRNA-targeted probes is provided by the ProbeBase database.23

FISH with rRNA-targeted probes and phylogenetic analyses complement each other. In the first step, rRNA gene sequences are retrieved from an environmental sample and are analyzed in order to determine the phylogenetic affiliations of the corresponding organisms. In the second step, probes targeting these sequences are designed and are applied in FISH experiments to visualize their target organisms in the original sample. The detection of the probe target cells by FISH confirms that the analyzed rRNA genes belong to organisms, which are really present in the environmental sample, and not to contaminating naked DNA.2

Organisms in environmental samples can be identified with a high degree of certainty by FISH with hierarchically nested rRNA-targeted probes. A nested probe set may contain, for example, one species-specific (probe I), one genus-specific (probe II), and one phylum-specific probe (probe III), which are labeled with different fluorescent dyes. Provided that the target species of probe I belongs to the genus and phylum that is targeted by probes II and III, respectively, all target organisms of probe I must also be detected by the other two probes. Accordingly, all target organisms that are stained by probe II must also be stained by probe III. Cells that are detected only by probe I or II are most likely nontarget organisms of these probes and have been stained due to unspecific hybridization. Nested probes make it possible to detect up to seven related phylogenetic groups in the same FISH experiment.24

FISH has been applied to study the microbial community structures of activated sludge flocs25-28 and to monitor important functional groups of bacteria in wastewater treatment plants, as for example organisms involved in nitrification,9,29-32 in floc formation33-39 or in enhanced biological phosphorus removal.40-44

Sometimes an organism cannot be detected by FISH even if it is clearly detectable by other methods in the same environmental sample. Possible reasons for such failure of FISH are: (i) the abundance of the probe target organism in the sample is below the detection limit of FISH, which is between 103 and 104 cells per ml,9 (ii) the probe-conferred fluorescence is too weak due to a low ribosome content of the probe target cells; (iii) the cell wall of the probe target organism is impermeable for oligonucleotide probes and requires special treatment prior to FISH; (iv) the probe-binding site is blocked by secondary structures of the rRNA or by ribosomal proteins; and (v) the hybridization conditions are too stringent for the applied probe. Furthermore, FISH is severely hampered if a sample contains many autofluorescent or DNA-binding particles. Since ribosomal RNA is a phylogenetic and not a physiological marker, rRNA-targeted FISH alone does not provide insight into the ecophysiology of the detected organisms. An overview of approaches to overcome some limitations of FISH is provided by Wagner et al.45 Extensions of FISH that allow monitoring metabolic activities of uncultured microorganisms are described later in this chapter.

15.2.3 Quantification and 3D Visualization of

Microorganisms by Confocal Laser Scanning Microscopy and Digital Image Analysis

The quantification of uncultured bacteria in environmental samples is a frequently encountered task in microbial ecology. Microbial cells can be stained, for example, by FISH with rRNA-targeted probes, and can then be counted "manually" in an epi-fluorescence microscope. However, this technique is tedious and yields reliable results only when applied on planktonic bacteria, which do not form larger cell aggregates. Manual counting of cells, which are embedded in aggregates, flocs, or biofilms, is difficult and tends to underestimate the cell numbers.26 Bacterial cell clusters are often packed very tightly (Figure 15.1) and cannot be broken up prior to counting.26 Difficulties with the dispersion of cell aggregates also hamper the use of flow cytometry to quantify bacteria in flocs.46,47

Semi-automated quantification techniques, which combine FISH or other fluorescence staining methods with digital image analysis, are less tedious than manual cell counting and can accurately quantify planktonic cells as well as clustered bacteria in flocs and biofilms. Epifluorescence microscopes equipped with digital cameras and confocal laser scanning microscopes are used to acquire images of fluorescing bacterial cells. Sharp confocal images, which contain only objects located in the current

FIGURE 15.1 Confocal micrograph of aggregated ammonia-oxidizing bacteria (Nitroso-monas sp.) detected in a nitrifying biofilm by FISH with probe S-*-Nsm-0651-a-A-18.30

focal plane, are better input data for image analysis software than pictures recorded by conventional microscopes. Digital image analysis has been applied to quantify soil bacteria, which were labeled by protein staining,48 and to enumerate planktonic cells that were stained by FISH with rRNA-targeted probes.49,50 Shopov et al.51 used image analysis to count planktonic bacteria and cells, which were attached to detritus. Unfortunately, no published image analysis routine is capable of counting the individual cells in dense cell aggregates. These tightly packed cells (Figure 15.1) are not recognized and separated by the algorithms even if the input data are high-quality confocal images. However, aggregated bacteria can be quantified by a stereological method, which does not require single cell recognition. This method determines the volume fraction of a "phase of interest," which is part of a larger reference volume.52 Any microbial population, which has been stained by FISH with a specific rRNA-targeted probe or by another specific fluorescence labeling technique, can be the phase of interest. The reference volume is either the total volume of all Bacteria or the total volume of all microorganisms (Bacteria, archaea, and eukaryotes) in the sample. Most known Bacteria can be labeled by FISH with the EUB338 probe mix.53,54 Unspecific labeling of all microbial cells is achieved by nucleic acid stains like 4',6-diamidino-2-phenylindole (DAPI).55 The biovolume fraction of the specifically labeled population can be manually determined by using an epifluorescence microscope and a grid ocular. In the current microscope field, all grid points that touch the specifically stained population (point set A) and all grid points that touch the total bacterial or microbial biomass (including the specifically stained population, point set B) are counted. Then the ratio of the number of points in set A to the number of points in set B is calculated (this is the fraction of point set A). This procedure is repeated at random positions within the specimen. Finally, the fractions of point set A that were obtained in the evaluated microscope fields are averaged. Provided that enough microscope fields have been processed, this mean fraction of point set A is equivalent to the biovolume fraction of the specifically stained population. This technique works with solitary and with aggregated microbial cells and can be applied without modifications on all kinds of bacterial morphotypes. A software implementation of this method has been combined with FISH and confocal laser scanning microscopy to quantify bacteria in flocs and biofilms.56,57 Recently, Juretschko et al.28 used rRNA sequence analysis, FISH, and the stereological estimation of biovolume fractions to identify, detect, and quantify most bacterial populations in activated sludge from a full-sized industrial wastewater treatment plant. In another recent study, Schmid et al.39 applied the same quantification method to analyze the bacterial communities in poor-settling activated sludge flocs, and correlated the results with physical floc properties. An extension of this stereological approach allows measuring not only biovolume fractions, but also absolute cell concentrations of clustered and of planktonic bacteria by spiking environmental samples with known amounts of reference cells or fluorescent beads ("Spike-FISH"58). Absolute cell concentrations are needed, for example, to compare the abundance of a bacterial population in different environmental samples with different biomass content. Another stereological approach has been applied to quantify the biovolume of filamentous bacteria by image analysis.59

Microbial ecology has numerous other applications for digital image analysis. Bacterial cells can be classified by software into morphotypes.49,60,61 The optimal hybridization conditions for FISH with new rRNA-targeted oligonucleotide probes are determined by measuring the probe-conferred fluorescence.54'62 The fluorescence intensity after FISH is also quantified in order to estimate the ribosome content of microbial cells.63

Two-dimensional image analysis does not exploit the full potential of confocal laser scanning microscopy. Confocal microscopes allow peeking into complex samples without mechanical sectioning, which could distort three-dimensional structures. Modern confocal microscopes can adjust the z-position of the focal plane with high precision in sub-micrometer intervals and can automatically record vertically stacked "optical sections" through a sample. Such image stacks contain 3D information, which is extracted and quantified by 3D image analysis routines. This approach has proven highly useful to quantify key features of biofilms like the total volume of the biomass, the biofilm roughness, and the substratum coverage.64-66 We are currently developing 3D image analysis algorithms that scan stacks of confocal images and evaluate spatial relationships between probe-defined bacterial populations. This approach could help in discovering ecologically important relationships between uncultured bacteria. For example, the statistically confirmed spatial co-localization of two populations may be used as an indicator for a putative mutualistic symbiosis between these organisms.

Confocal image stacks of probe-stained bacteria are well suited for 3D visualization, which can reveal interesting features like the networks of cavities within the cell clusters of Nitrospira-like bacteria31 (Figure 15.2). Visualization programs should provide high-speed interactive rendering with arbitrary viewing perspectives.

FIGURE 15.2 Three-dimensional reconstruction of a cell aggregate of Nitrospira-like bacteria, which were stained by FISH with probe S-G-Ntspa-662-a- A-18.31 A part of the cell cluster has been removed by software in order to visualize a network of cavities within the microcolony.

Functions for lighting and shading can significantly improve the 3D impression of the rendered images. The quality and information content of 3D visualization is strongly influenced by the applied rendering algorithm. The surfaces of objects can be mathematically transformed to 3D meshes of polygons.67 The polygons are projected to 2D space, are drawn by the program in a color defined by the user, and are shaded according to the directions of their surface normal vectors and according to the position of a virtual light source. Algorithms based on this approach are relatively simple and fast because only the vertices and the surface normal vectors of the polygons are stored and processed. In contrast, volume rendering uses the whole content of an image stack (the empty space as well as the surfaces and the inner parts of objects) for 3D visualization. Such programs can produce very detailed 3D reconstructions, but interactive volume rendering requires fast computers and sophisticated software implementations. The increasing performance of commodity computers makes 3D visualization by volume rendering cheaper and easier to use as a routine tool in floc and biofilm research.

15.2.4 Methods to Determine Ecological Functions of Uncultured Bacteria

The traditional and well-proven methods to study physiological properties of bacterial pure cultures cannot be used to investigate the ecophysiology of yet uncultured microorganisms in environmental samples. This goal can be achieved only by using cultivation-independent molecular tools.

Microautoradiography (MAR)68 detects the uptake and incorporation of substrates by microbial cells without the need to cultivate the organisms. An environmental sample is incubated with a radioactively labeled substrate under defined conditions, as for example temperature, pH, and concentrations of electron donors and acceptors. After the incubation, the microbial cells in the sample are chemically fixed and excess radioactivity is removed by washing the sample with appropriate buffer solutions. Samples containing flocs or biofilm are then sectioned by using a microtome or a cryotome. The sections are applied onto microscope slides or cover slips and are covered with a film emulsion, which solidifies at lower temperatures. During the following exposition, which may last several days or weeks, silver grain formation occurs in the film above radioactively labeled cells. In contrast, the film remains clear above cells that do not contain the radioactive tracer. The MAR experiments are evaluated by bright field and phase contrast microscopy. Microautoradiography offers unique possibilities to monitor, with a single cell resolution, the use of specified substrates by bacteria in environmental samples. However, this technique is not able to identify the radioactively labeled cells. This limitation is overcome by the combination of MAR and FISH with rRNA-targeted probes (FISH-MAR)69,70: The microbial cells are identified by FISH, and the incorporation of radioactively labeled substrates is detected by MAR (Figure 15.3). The high specificity of FISH and the single cell resolution of both FISH and MAR make FISH-MAR a powerful tool for determining the substrate usage spectra of bacteria in environmental samples, which must be amenable to FISH (this requirement hampers the application of FISH-MAR on many soils and sediments).

2-10 ^m section

Cover slip

Black film above cells that incorporated radioactive substrate

0 0

Post a comment