Modern methods in subsurface microbiology: in situ identification of microorganisms with nucleic acid probes


Corresponding author. Tel: +49 (421) 2028930; fax: +49 (421) 2028790.


Like many other parts of microbial ecology, subsurface microbiology has entered the molecular age. As one example of various powerful molecular techniques, fluorescently labeled rRNA-targeted nucleic acid probes today allow an in situ identification of individual microbial cells in their natural habitats. The technique relies on the specific hybridization of the nucleic acid probes to the naturally amplified intracellular rRNA. Fluorescently labeled, rRNA-targeted oligonucleotide probes are perfect tools for many areas of microbial ecology since they can monitor specific populations in environmental samples based on constant genotypic features and not on variable phenotypic features like morphology. In case of immobilized communities like biofilms, exact spatial distributions of microorganisms can be analyzed on a micrometer scale. Recent technical improvements have increased the number of potential applications considerably. Today, better fluorescent dyes enable identification of routinely more than 50% of the cells even in oligotrophic aquatic samples in which the visualization of small cells with low numbers of ribosomes had been problematic. This compares favorably with the usually less than 1% of microorganisms which can be characterized based on cultivation-dependent methods.

1Subsurface microbiology in the molecular age

Like many other parts of microbial ecology, subsurface microbiology has entered the molecular age with its clone libraries, phylogenetic trees, and the nucleic acid probe-based quantifications of well-known or ‘new’ microbial species. An early milestone of this era is the 1987 review article of Carl Woese entitled ‘Bacterial Evolution’[1]in which the author summarizes the results of the first decade of comparative 16S rRNA analysis. Due to the fact that any cell contains ribosomes, he and his coworkers presented us an universal framework of stable evolutionary relationships, of which many were unexpected from previous artifical attempts of bacterial systematics, that has since been a solid and constantly extended basis for a phylogenetically correct placing and identification of bacteria. By using an rRNA approach for studying microbial ecology scientists also obey one of biology's golden rules formulated so precisely by Dobzhansky in 1937 “Nothing in biology makes sense without evolution”[2].

There are two larger ribosomal RNA molecules. The ribonucleic acid in the small subunit of the ribosome has for most bacteria a length of about 1,500 nucleotides and is, based on its sedimentation speed, referred to as 16S rRNA. The 23S rRNA of the large subunit of the ribosome has a length of about 3,000 nucleotides. Both are patchworks of regions with higher and lower evolutionary conservation. Consequently, comparative analyses of rRNA sequences can identify so-called signature nucleotides or nucleotide motifs on various taxonomic levels which are perfect targets for an evolutionary based identification [3, 4].

Hybridization techniques now usually rely on the specific binding of nucleic acid probes to such signature regions on the rRNA. By definition these probes are single strands of nucleic acid which have the potential of carrying detectable marker molecules highly specifically to complementary target sequences. Even though probes can be directed to a multitude of target molecules, e.g. metabolic genes, mRNA or plasmids, this minireview is restricted to rRNA-targeted oligodeoxyribonucleotide probes. There are several reasons which make rRNA-targeted oligodeoxyribonucleotides particularly suitable for the in situ identification of microorganisms (for detail see [4]). Firstly, rRNA molecules occur in high copy numbers of usually more than 1,000 in any living cell. Secondly, rRNA sequences can be today routinely retrieved from the samples of interest without prior cultivation. Thirdly, a directed probe design, considering a priori multiple nontarget organisms, is possible due to the huge sequence collections available especially for the 16S rRNA. As a consequence, our knowledge on the secondary structure and the evolutionary conservation of rRNA probe target sites is better for 16S rRNA than for any other potential target molecule.

2Probes and their design

According to their length, nucleic acid probes can be grouped as polynucleotide probes with usually more than 50 nucleotides and as oligonucleotide probes which frequently have around 20 nucleotides. For two main reasons, oligonucleotide probes are today more frequently used than polynucleotide probes. Firstly, they are short enough to allow for single mismatch discrimination of target nucleic acids. Secondly, large quantities of oligonucleotides can be rapidly and inexpensively produced in high quality by solid phase synthesis. Various marker or linker molecules can be introduced during the synthesis. Labels like fluorescein or biotin can be covalently bound in the last cycle to the 5′ end of the oligonucleotide. As an alternative a primary aliphatic aminolinker might be bound to the 5′ end which serves as a versatile binding site for activated labels such as carboxytetramethylrhodamine-isothiocyanate or the N-hydroxy-succinimide esters of digoxigenin or of new fluorescent dyes such as the carbocyanines Cy3 and Cy5 [5]. The labeling of aminolinked oligonucleotide probes and the subsequent purification is a routine method that can be done without expensive equipment [6]. The quality of the probe is crucial for the success of the hybridization assay and should be carefully checked by spectrophotometrical analysis and polyacrylamide gel electrophoresis.

The design of rRNA-targeted oligonucleotide probes is today performed in a computer-assisted fashion. It is essential to work on an updated rRNA data base such as those maintained by the Ribosomal Database Project in the USA [7]or the group around Rupert DeWachter in Antwerpen [8]. Wolfgang Ludwig and his colleagues from our Department have recently developed the software package ARB which greatly facilitates the design of rRNA-targeted oligonucleotide probes [9]. The Probe Design option starts with the selection of individual sequences or sequence clusters in the phylogenetic tree and yields lists of potential probes organized according to many important aspects of probe design. In the second, manual step additional information, e.g., on the in situ accessibility of certain rRNA sites is used to select probes from this list. Then probe sequences are first evaluated with the Probe Check option in ARB. Following synthesis probes are further evaluated experimentally on selected target and non-target bacteria. Probes will only bind correctly under defined hybridization conditions and the rigorous optimization of hybridization and washing conditions is as important as the probe design. Due to the ever growing databases it is important to re-evaluate the actual specificity of ‘old’ probes before using them.

3Principles of in situ hybridization

The principles of in situ hybridization with fluorescently labeled, rRNA-targeted oligodeoxyribonucleotides are quite straightforward [4, 6]. First, the morphology of the cells in the examined sample has to be stabilized and the cell walls and membranes have to be permeabilized for the penetration of the probes. This can be both achieved with fixatives which are usually based on aldehydes and/or alcohols. Subsequently, the probes are applied in an adequate hybridization buffer and incubated at an adequate hybridization temperature (usually between 35 and 50°C) for one to several hours. Washing steps are applied to remove unbound and part of the non-specifically bound fluorescent probe and the sample can subsequently be analyzed by epifluorescence microscopy or flow cytometry. Several probes labeled with spectrally different fluorochromes can be simultaneously used on one sample, e.g. a fluorescein-labeled probe which emits green light upon blue excitation, together with the orange-red carboxytetramethylrhodamine and the dark-red Cy5. The sensitive visualization of the latter requires a CCD camera.

4Sequencing and probing

In situ hybridization with fluorescently labeled, rRNA-targeted oligodeoxyribonucleotides is just one part of the rRNA approach to microbial ecology and evolution [4, 10]. The rRNA approach can be divided in a sequencing and probing part as shown in Fig. 1. Usually rRNA sequences are first retrieved from an environmental sample in a cultivation-independent fashion. rRNA sequences can however also originate from cultures obtained from the sample of interest or from the rRNA databases. The comparative analysis of these sequences allows to estimate the microbial diversity in the examined sample and to design nucleic acid probes which can be used to analyze the original or other samples for the abundance of certain sequences. In situ hybridization thereby links a sequence of a given phylogenetic affiliation, the phylotype, to whole fixed cells, the morphotype.

Figure 1.

Flow chart showing the principle phases of the rRNA approach, sequencing and probing (taken from [4]).

The rRNA approach is today routinely applied for phylogenetic analysis and in situ identification of hitherto uncultured microorganisms [4, 11]. Magnetotactic bacteria, e.g., are an unique group of such bacteria occurring in the microaerobic zone of sediments. They have small internal magnetic particles, so-called magnetosomes, and swim along the lines of the magnetic field. Since several of these morphologically conspicuous bacteria have remained unculturable over the years, the rRNA approach was used in an attempt to phylogenetically classify and identify these interesting bacteria [12]. When we first used a magnetically enriched cell preparation, the retrieved rDNA sequences could not be related to cells containing magnetosomes [13]. Only after a second magnetic enrichment, clones were obtained that could be undoubtedly assigned to distinct magnetotactic morphotypes. This clearly demonstrated the importance of in situ hybridization in assigning retrieved sequences to cell populations in the examined samples.

In order to correctly interpret results of rRNA sequencing studies, the multiple steps involved in today's standard technique for rRNA sequence retrieval have to be considered (Fig. 2). The results will be dependent (i) on the optional cell enrichment, (ii) on the efficacy of DNA extraction, (iii) on the PCR amplification of the different rRNA genes (rDNA), (iv) on the cloning of the different rDNAs, and (v) on the number of clones which are screened and sequenced. Each step and especially the PCR, which is in itself a multistep procedure, can introduce biases. Furthermore, the numbers of genomes per cell and rRNA operons per genome might vary considerably. Therefore, it is not very reasonable to estimate the abundance of a certain organism in a sample from the number of respective clones in the rDNA library. If the sampling and the subsequent steps were performed properly it is at least clear that the retrieved sequence is part of the rRNA diversity in the investigated sample. It is, however, still unclear whether it was retrieved from intact cells or from the pool of free DNA present in many ecosystems. Absolute or relative abundance of a given sequence or organism can much more reliably be determined by hybridization with oligonucleotide probes. Such hybridizations can be performed on various levels (Fig. 3). The most important hybridization techniques for microbial ecology, since they are most direct, are the quantitative dot blot hybridization of extracted nucleic acids as first used by Stahl et al. on rumen samples ([14]; more references on this technique in [4]) and the in situ hybridization of the fixed environmental sample [4].

Figure 2.

Flow chart showing the different possibilities to characterize an environmental sample by comparative rRNA sequence analysis (taken from [4]).

Figure 3.

Flow chart showing the different options of using rRNA-targeted nucleic acid probes to analyze an environmental sample by hybridization techniques (taken from [4]).

For quantitative dot blot hybridization, the samples of interest are treated harshly, e.g., by bead-beating in the presence of phenol [14], to maximize cell lysis and release of nucleic acids. The relative abundance of a given rRNA sequence can be accurately determined relative to total rRNA. Changes in relative abundance of a given rRNA might reflect changes in the abundance of the cell population of interest or in the cellular rRNA content. For several organisms, a direct linear relationship has been shown between growth rate and cellular ribosome content [4]. The relative abundance of an rRNA can be interpreted as the relative importance of a defined species in terms of actual metabolic activity or potential metabolic activity. It cannot be transformed to cell numbers.

In situ hybridization relies on the preservation of cells at their sites of action. It allows for exact enumeration and localization of defined microbial populations in respect to the biotic and abiotic environment. The numbers determined can, however, be compromised by difficulties in the cell permeabilization or by a lack of sensitivity for visualization of all cells. Detected cells are either counted in absolute numbers, which can be in a straightforward way converted to biomass, or enumerated relative to total cell counts as determined by the DNA intercalating stain DAPI. In this process, it is important to use positive and negative control probes to determine the fraction of cells accessible for in situ detection and the fractions of non-specifically stained and autofluorescent cells or particles [4]. In order to apply in situ hybridization with fluorescently labeled rRNA probes in a reasonable way, the fraction of detectable cells should be high, preferentially above 50% of the DAPI-stained microbial cells. Dot blot and in situ hybridization should not be regarded as competing but as complementary methods.

5Applications of in situ identification of microorganisms with nucleic acid probes

The following short review of some applications of in situ hybridization with fluorescently labeled, rRNA-targeted oligodeoxyribonucleotides is restricted to examples that might be of special relevance to subsurface microbiology. Many of the microorganisms dwelling in the subsurface are attached to surfaces and therefore the in situ localization of cells in these so-called biofilms is of high importance. However, microorganisms will also occur as planktonic, free-swimming cells in the subsurface aquifers. The subsurface water is frequently oligotrophic and the planktonic bacteria might, consequently, experience nutrient limitation. It might therefore be of interest for the subsurface microbiologist how in situ hybridization with fluorescently labeled, rRNA-targeted oligodeoxyribonucleotides performs in oligotrophic aquatic systems.

5.1In situ visualization of defined cell populations in biofilms

Molecular and microscopic identification of defined bacterial populations in multispecies biofilms was first achieved in an anaerobic fixed-bed bioreactor [15]. By in situ hybridization with a group-specific oligonucleotide, two morphologically distinct populations of gram-negative sulfate reducing bacteria (thick and thin vibrios) could be visualized and the rapid colonization of newly inserted glass slides could be monitored. Based on retrieved rRNA sequences, specific probes were designed differentiating a Desulfuromonas- and a Desulfovibrio-related population [15]. The latter probe was used to monitor enrichments on media optimal for the closest culturable relatives as identified by comparative sequence analysis [16]. Though the enrichments were dominated by probe positive cells, only three out of 30 isolates were probe-positive. This nicely illustrates how the rRNA approach can support directed cultivation attempts.

In a more recent study, a set of probes specific for the genus Paracoccus was used to identify the respective population in the biofilms of a denitrifying sandfilter [17]. A multiple probe approach was used to assure specific identification of paracocci and by confocal laser scanning microscopy it was possible to analyze their three-dimensional distribution (Fig. 4A,B).

Figure 4.

In situ identification of specific bacterial populations by hybridization with fluorescently labeled, rRNA-targeted oligonucleotide probes. (A) A depth-coded profile of the spatial arrangement of Paracoccus sp. cell aggregates after hybridization with TRITC-labeled probe PAR651. Two-dimensional reconstruction of 32 optical sections recorded at intervals of 0.9 μm. (B) A double identification of Paracoccus sp. cells with two genus-specific probes. Hybridization with TRITC-labeled probe PAR651 and FLUOS-labeled probe PAR1457. TRITC-conferred fluorescence displayed in red, FLUOS-conferred fluorescence in green. Single xy-images of the same optical section are combined. (C–F) In situ identification of bacteria in a water sample taken from Lake Gossenköllesee by a combination of hybridization with CY3-labeled probes, and DAPI staining. Identical microscopic fields have been visualized with an epifluorescence microscope using filter sets specific for DAPI (left) and CY3 (right). (C,D) Hybridization with probe EUB338 specific for Bacteria. (E,F) Hybridization with probe BET42a specific for the beta-subclass of Proteobacteria. Bar, 10 μm (applies to C–F).

For the analysis of the community structure in activated sludge flocs, which can be regarded as mobilized biofilms, a so-called top-to-bottom approach has been used. In this approach, oligonucleotide probes specific for phylogenetic groups on different taxonomic levels are applied in an ordered way starting with the more general ones and refining the analysis finally to the genus and or species levels [3, 4]. Already the first round of hybridizations with oligonucleotides specific for the domains Bacteria and Archaea yielded important results. The majority of cells hybridized strongly with a bacterial probe [18]. This indicated not only in situ dominance of members of the domain Bacteria but also that is was possible to use in situ hybridization. Subsequently, probes specific for the alpha-, beta, gamma-subclasses of Proteobacteria and other bacterial groups were used to characterize the community structures by in situ hybridization. Comparison of the in situ probing results with probing of the heterotrophic bacteria isolated on nutrient-rich agar-plates clearly showed the inadequacy of cultivation-dependent techniques for analyzing the community structure of activated sludge [18]. Whereas cells hybridizing with the beta-subclass proteobacterial probe were most abundant in situ, most of the colonies could be classified as gamma-subclass Proteobacteria by probing. Further hybridizations with a genus-specific probe for Acinetobacter revealed that, although between 30 and 60% of the examined colonies were positive, in situ abundance of cells hybridizing with the same probe was only between 1 and 10%[19]. Likely due to their high plating efficacy on nutrient-rich agar, the abundance of Acinetobacter spp. in wastewater treatment plants has been severely overestimated.

One of the most important features of in situ identification in biofilm or activated sludge is that various cell populations can be simultaneously visualized in their arrangement to each other. Recently, we succeeded in using a combination of confocal laser scanning microscopy and three differently labeled rRNA-targeted oligonucleotide probes for the simultaneous visualization of seven distinct phylotypes of quite closely related beta1-subgroup proteobacteria in one activated sludge floc [20]. This clearly demonstrated the high genetic diversity present in many natural samples.

5.2Oligotrophic water samples

An important advantage of the top-to-bottom approach is the fact that as soon as ready-to-use probe sets have been developed, they can be applied to a variety of different ecosystems. In oligotrophic waters like in sediment and soil samples, the application of fluorescently labeled rRNA-targeted oligonucleotide probes for in situ identification of bacteria has long been hampered by low detection rates likely caused by the small size and relatively low ribosome content of the target cells. Glöckner and coworkers recently developed a protocol for freshwater samples using probes labeled with the new, highly sensitive carbocyanine dye CY3 [5]on cells that had been concentrated on white polycarbonate filters [21]. Detection rates achieved on regular epifluorescence microscopes (Zeiss, Jena, Germany) with optimized filter sets (Chroma, Brattleboro, VT, USA) in different lake water samples, which had been low before, were increased to 50% or more of the cells stained with DAPI (Fig. 4C–F). This improved protocol was successfully used to analyze the bacterial community structure in the winter cover and the pelagic zone of an ultraoligotrophic high mountain lake [22]. The fraction of DAPI-stained cells detectable after hybridization with a bacterial probe ranged from 40 to 81%. Interestingly, detection rates and probe-conferred fluorescent signals were also high in areas where the secondary production and the fraction of INT-positive cells was low. This clearly showed that the in situ hybridization is now sensitive enough to detect a reasonable fraction of the cells even in very oligotrophic environments.

6Future perspectives for in situ hybridization in subsurface microbiology

The applications discussed above clearly indicate that in situ hybridization with fluorescently labeled rRNA-targeted probes has reached a level which allows the direct, cultivation-independent analysis of most members of the natural microbial communities present in the subsurface. Such studies could yield new insights in the true abundance of well known or new bacteria and increase our knowledge on the microbial ecology of these interesting ecosystems.


I thank my Ph.D. students and my technician Sibylle Schadhauser for their excellent work and Dr. Wolfgang Ludwig, Prof. Karl-Heinz Schleifer and Prof. David A. Stahl for all their support over the last years. The original work has been supported by grants from the Deutsche Forschungsgemeinschaft (Am 73/2-3), the Freistaat Bayern (FORBIOSICH, FORGEN) and the European Community (HRAMI projects).