OUP user menu

Novel approach using substrate-mediated radiolabelling of RNA to link metabolic function with the structure of microbial communities

Marcell Nikolausz , Márton Palatinszky , Anna Rusznyák , Hans-Hermann Richnow , Uwe Kappelmeyer , Matthias Kästner
DOI: http://dx.doi.org/10.1111/j.1574-6968.2007.00821.x 154-161 First published online: 1 September 2007

Abstract

A novel concept was developed applying radioisotope-labelled substrate incorporation into the biomass. The resulting radiolabelled RNA was used both as an indicator of activity and as a template for gaining structural and functional information about a substrate-utilizing microbial community. Sequences of PCR products are separated via cloning or using molecular fingerprinting techniques. Nucleic acids from predominant clones or the whole molecular fingerprinting pattern are transferred to a membrane and hybridized with the radiolabelled sample RNA. Scanning of the hybridized blots for radioactivity indicates the members involved in the utilization of the substrate. This novel ‘random walk’ approach using radioisotope probing was evaluated in a model community experiment.

Keywords
  • radioisotope probing
  • 14C
  • microbial activity
  • community structure
  • metabolic function

Introduction

Identification of metabolically active microorganisms and the search for the link between metabolic function and taxonomic identity has become a hot topic in microbial ecology. The addition of a substrate labelled with stable isotopes to microbial communities has been shown to be an elegant way to detect and identify microorganisms that can utilize specific compounds in the environment. Stable-isotope probing (SIP) of nucleic acids has become very popular during the last few years for this purpose (Radajewski et al., 2000; Manefield et al., 2002). The isotopically labelled 13C-DNA fraction derived from active substrate-utilizing cells is separated by density-gradient centrifugation. After separation, the identity of the bacteria responsible for the particular degradation and utilization of the stable isotope-labelled substrate can be revealed by 16S rRNA gene sequence analysis from the heavy fraction (Dumont & Murrell, 2005). The drawbacks of the DNA-SIP approach are the high concentration of substrate required and the long incubation time necessary in order to incorporate sufficient 13C into the DNA for effective subsequent separation, which may lead to the spreading of the signal to secondary consumers via cross-feeding (Radajewski et al., 2003). The RNA-SIP approach requires shorter incubation, but a more rigorous separation technique is necessary and only the semi-quantitative comparison of different fractions gives reliable results (Lueders et al., 2004; Whiteley et al., 2006). Nonetheless, SIP approaches have been successfully used for studying a wide range of complex microbial communities (Hutchens et al., 2004; Lu et al., 2005; Kasai et al., 2006; Lueders et al., 2006; Prosser et al., 2006; Singleton et al., 2006); for more details, see recent reviews (Radajewski et al., 2000, 2003; McDonald et al., 2005; Whiteley et al., 2006).

The combination of FISH and microautoradiography has also been used for linking the identity of microbial community members with their function (Lee et al., 1999; Ouverney & Fuhrman, 1999). Microbial communities were incubated with a radioactively labelled substrate, and identification of bacteria was performed by FISH techniques. Incorporation of radiolabelled materials into biomass indicates that the detected cells were metabolically active; however, care should be taken to distinguish signals derived from physically attached but not actually consumed compounds (Nielsen et al., 1999). FISH–microautoradiography has been applied in various studies for complex community samples (Rossello-Mora et al., 2003; Ginige et al., 2004, 2005; Teira et al., 2004). However, such studies require careful design and a preliminary knowledge about the potential key players. Furthermore, only a few groups of microorganisms can be investigated in parallel.

To overcome this limitation, rRNA-targeted oligonucleotide probes were applied in an array format to achieve multiple reverse hybridizations in a so-called isotope array (Adamczyk et al., 2003). Fluorescently labelled total RNA including a radioactive fraction, obtained from environmental samples incubated with a 14C-labelled substrate, was hybridized in the microarrays. Fluorescent signals and radioactivity were detected with a fluorescent scanner and a β imager, respectively. The position of the fluorescent signal provides information about the community structure, while radioactive probe spots provide information about active utilization of the radioactively labelled material. However, the application of the isotope array technique has special equipment demands and depends on the availability of appropriate microarrays with probes of known target organisms (Wagner et al., 2006). Some studies reported only moderate detection limits (Cho & Tiedje, 2002; Bodrossy & Sessitsch, 2004), specificity problems of microarrays (Small et al., 2001) and lack of consistency (Shi et al., 2004). Owing to the high unexplored diversity of environmental samples, oligonucleotide hybridization may be much better for validation of results obtained by other methods than for de novo analysis.

Therefore, a novel approach is proposed in order to overcome the above-mentioned drawbacks of the present isotope probing and to provide a less rigorous and equipment-demanding alternative to the isotope array technique. The new method involves substrate-mediated radioisotope labelling of a microbial community, using the labelled rRNA both as a template for 16S rRNA gene-based community structure analysis and as an activity indicator for hybridization and for visualizing the amplicons obtained from active cells. A proof of principle obtained from a model community is presented here.

Materials and methods

Bacterial strains, cultivation and isotope-labelling experiment

The model community was established using six bacterial strains. A pure culture of Pseudomonas stutzeri DSM50238 (Ps) was grown for 36 h in liquid mineral medium (Hartmans et al., 1989) in a final volume of 5 mL using 14C-labelled acetate (14CH3COOH, 10–20 mCi of C mmol−1, Sigma) as a sole carbon source [25 µCi mL−1 (or 325 kBq mL) amended to 4 mg mL−1 final concentration] to achieve a high level of labelling. The incorporation of radioactive carbon into the rRNA was checked by scintillation counting and autoradiography of membrane-spotted total RNA (see below). The nonlabelled members of the model community consisted of the closely related strains Pseudomonas putida (Pp) (96% 16S rRNA gene sequence similarity to Ps), moderately related Thauera aromatica (Ta, 87% similarity) Azoarcus sp. DSM9506 (Az, 86%) and the distantly related Bacillus subtilis DSM402 (Bs, 84%) and Escherichia coli DSM5695 (Ec, 84%). For the maintenance of the strains, nutrient agar (DSMZ medium 1) was used. The establishment of the model community was carried out at a nucleic acid level using PCR amplicons or single-stranded DNA (ssDNA) instead of mixing a defined amount of cells in order to avoid potential bias due to differential RNA isolation or PCR (Sipos et al., 2007).

Nucleic acid isolation, PCR, asymmetric PCR

For RNA work, all reagents and processes were prepared in/with either sterile, disposable, guaranteed nuclease-free labware or glass and metalware that had been baked at 280°C for 4 h. All solutions were prepared with water that had been treated with 0.1% (v/v) diethyl pyrocarbonate (DEPC) solution (overnight incubation at 37°C, followed by autoclaving).

Two 2 mL liquid cultures of Ps from the isotope-labelling experiment were pelleted by centrifugation at 10 000 g for 10 min. Total RNA isolation was carried out using the RNeasy Mini Kit (Qiagen) as described by the manufacturer, with the following modifications. In order to achieve highly efficient cell lysis, bacterial cells were disrupted mechanically using the bead beater of a FastPrep system (Qbiogene, Carlsbad, CA). The pellets were resuspended in 700 µL of RLT buffer from an RNeasy Mini Kit, transferred into the FastPrep tube containing the beads (lysing matrix A) and then shaken in the FastPrep machine for 30 s at speed 5. Further steps followed the manufacturer's protocol with on-column DNAse treatment to avoid DNA contamination. RNA was eluted twice in succession with 35 µL of RNAse-free distilled water. Three microliters of eluted total RNA was transferred into 10 mL scintillation fluid (UltimaGold, Packard BioScience) and were analyzed for 5 min in a liquid scintillation counter (Wallac WinSpectral α/β counter 1414, PerkinElmer). The radioactivity incorporation ranged between 660 and 1140 disintegration per minute µL−1 of eluted RNA.

DNA was isolated from all strains using the DNAeasy tissue kit (Qiagen) according to the manufacturer's instruction for Gram-positive cells. DNA was eluted in 120 µL of RNAse-free distilled water. PCR was performed in a final volume of 50 µL on a Mastercycler gradient (Eppendorf, Hamburg, Germany) with 27F and 1378R primers (Heuer et al., 1997) with all strains separately. PCR for denaturing gradient gel electrophoresis (DGGE) was carried out with GC968F (Nubel et al., 1996) and 1378R primers carrying a 6-carboxy-2′,4,4′,5′,7,7′-hexachlorofluorescein (HEX) fluorophore at the 5′ end. The conditions for PCR amplifications were as follows: initial denaturation at 95°C for 15 min, followed by 30 cycles of primer annealing at 51°C for 30 s, chain extension for 50 s at 72°C, denaturation for 30 s at 95°C and a final extension at 72°C for 30 min. PCR products were visualized with UV excitation after gel electrophoresis (1.2% agarose) and ethidium-bromide staining. Quantification was performed using quantity one software (Bio-Rad) and MassRuler DNA ladder mix (Fermentas) for comparison.

Asymmetric PCR was performed using 100 ng of purified (Qiagen, QIAquick PCR Purification Kit) PCR product from the previous reaction, obtained with 27F and 1378R primers, as a template. PCR conditions were the same as described above for standard PCR, but only one reverse primer [907R (Muyzer et al., 1995) or 1378R] was applied and the cycle number was increased to 40. This reaction mainly generates ssDNA, antisense compared with the 16S rRNA gene. The product was purified using the QIAquick PCR purification Kit (Qiagen).

Dot-blot hybridization

Asymmetric PCR products were denatured at 98°C for 10 min and then transferred immediately onto ice. Denatured samples (45 µL each) were transferred to positively charged nylon membranes (Boehringer Mannheim) using a Bio-Dot microfiltration unit (Bio-Rad, Munich, Germany) with transfer medium consisting of 6 × SSC buffer (20 × SSC is 3 M NaCl and 0.3 M sodium citrate in DEPC-treated water) according to the manufacturer's recommendations. DNA was immobilized onto the membrane by exposing it under UV light (302 nm, 100 mJ cm−2) for 2 min.

Prehybridization and hybridization were performed at 45°C in a Mini hybridization oven MK II (Hybaid, Teddington, UK) in 6 mL of hybridization solution containing 50% (v/v) formamide, 0.03% (w/v) sodium dodecylsulfate (SDS), 30 mg mL−1 powdered milk and 0.3 mg mL−1 salmon sperm in 5 × SSC solution. After 4 h of prehybridization of the membrane, denatured sample RNA (98°C for 10 min) was added and further hybridized for 16 h. After hybridization, the filter was washed twice at room temperature with 25 mL of 2 × SSC containing 1% (w/v) SDS and twice at 68°C with 25 mL of 0.1 × SSC containing 1% SDS, for 15 min each time.

DGGE analysis, membrane transfer and hybridization

DGGE was carried out according to the protocol of Muyzer (1997) using the INGENY PhorU System (Ingeny International, the Netherlands). 16S PCR products of strains (1 µg) generated by GC968F and 1378R(−HEX) primers separately or a mixture of all amplicons (700 ng each) were directly applied to 6% (w/v) polyacrylamide gels that contained denaturing gradients between 30% and 60% [7 M urea and 40% formamide (v/v) as 100% denaturants] and were separated by electrophoresis for 5 h at a constant temperature (60°C) in 1 × TAE buffer at 200 V. Gels with HEX-labelled products were scanned without any additional staining using a Molecular Imager FX (Bio-Rad) laser scanner set to measure HEX fluorescence. Two holes served as a position marker labelled with paper triangles for comparing fluorescent scans with subsequent radioscans.

The gel was then electroblotted onto a positively charged nylon membrane (Boehringer Mannheim) using a Multiphor II electroblotter (Pharmacia Biotech) for 1 h at 150 mA. The efficiency of transfer was checked by scanning the DGGE gels after the electroblotting for residual HEX-labelled products. The position of the marker holes was indicated on the membrane by ink. The membrane was subsequently incubated for 15 min on a pad of gel-blotting papers (Schleicher&Schuell) soaked in 0.4 M NaOH for 15 min to denature the DNA. The membrane was washed twice in 0.2 × SSC solution and the DNA was immobilized by UV exposure. Hybridization with radioactively labelled sample RNA was performed as described above for dot-blot hybridization.

Autoradiography

Dried membranes were incubated at room temperature for various times, tightly pressed to a Kodak Storage Phosphor Screen-K (Eastman Kodak) using an aluminum exposure cassette (Bio-Rad). The latent image was scanned using a Molecular Imager FX scanner (Bio-Rad) with 532 nm wavelength excitation; the emitted light at 390 nm was detected by a photomultiplier at 50 µm resolution. The resulting images were recorded and processed by quantity one software. The position markers were labelled with isotopically labelled RNA before scanning in order to enable comparison with fluorescent scans.

Results and discussion

A novel approach for linking functional and structural information of microbial communities was developed on the basis of radioisotope probing (RIP). The method combines the ‘random walk’ methods used for the exploration of microbial diversity and the radiolabelling of nucleic acids via substrate utilization. The outline of this novel concept can be seen in Fig. 1. The proposed method involves (1) substrate-mediated radioisotope labelling of an active microbial community, using the resulting labelled rRNA both (2) as a template for 16S rRNA gene-based structure analysis and (3) as an indicator for hybridization and for visualizing the amplicons obtained from active cells. Two general approaches can be applied: (I) antisense ssDNA from clone libraries or (II) molecular fingerprinting patterns of PCR products generated from the environmental samples can be blotted on membranes and hybridized with the labelled sample rRNA. The radioactive signal indicates the bands or clones representing active utilizing members of the community. Phylogenetic information can be obtained from the detailed sequence analysis of the clone library or by sequencing the band of interest.

1

Outline of the proposed RIP approach in combination with DGGE and cloning and sequencing of 16S rRNA gene products: (I) combined with the cloning and sequencing approach; (II) combined with a molecular fingerprinting technique, DGGE.

In order to prove the principle of the method, a model community was established using PCR amplicons from six strains of bacteria rather than by mixing cells or using a real artificial community for the cultivation and labelling experiment. In this way, bias associated with PCR (von Wintzingerode et al., 1997) and differences in genomic properties (Farrelly et al., 1995) could not influence the final result. Moreover, it was possible to rule out the potential bias caused by the cross-feeding of the processed substrate and the usage of the same substrate by various microorganisms from the experimental setup (Radajewski et al., 2003). Because dot-blot and DGGE-blot hybridizations with oligonucleotide probes are standard techniques, the main focus of this study was the investigation of the specificity of hybridization of the labelled rRNA to ssDNAs and denatured PCR products (DGGE pattern) blotted on a membrane. One of the strains (Ps) was labelled by cultivating it on a mineral medium containing 14C-labelled acetate as the sole carbon source. The isotope label appears quickly in the rRNA fraction of nucleic acids and indicates the metabolic activity of the cells (Lee et al., 1999; Adamczyk et al., 2003). The rRNA-based investigations of diversity usually consist of several in vitro amplification steps (e.g. reverse transcriptase PCR, cloning) whereby the radioactive label would be lost. However, the native rRNA can be investigated directly via hybridization (Amann et al., 1995). Antisense ssDNA has a high hybridization potential to RNAs (Wetmur, 1991); therefore, ssDNAs obtained from a clone library generated from 16S rRNA gene PCR products can be effectively used for sorting 16S rRNA gene fraction from the environmental communities.

(I) In the present experiment, ssDNAs from pure cultures corresponding to products from a clone library were dotted on a membrane. Total RNA from the isotope labelling experiment, which represents the RNAs from the active members of a community, was then hybridized to this membrane. As expected, a strong radioactive signal was detected at the matching position (ssDNA dot from Ps), while slight cross-hybridization was observed with nucleic acids from closely and moderately related microorganisms (Fig. 2). A stronger signal was observed, with the longer ssDNA product obtained with 1378R primer compared with the shorter one generated with the 907R primer. Asymmetric PCR (Sturzl & Roth, 1990) applied in this study for in vitro ssDNA synthesis provides an easy way of generating a template for hybridization to rRNAs. Alternative means of single-stranded probe generation could include in vitro transcription (Melton et al., 1984) or magnetic bead separation of one strand of the PCR products (Bertilsson et al., 2002). Although the assay is capable of detecting the potential key players from a community even at a species level, closely related species with high 16S rRNA gene sequence similarity may cause problems due to cross-hybridization, which is also a drawback of all other hybridization-based detection methods.

2

Approach I: RIP experiment with Pseudomonas stutzeri (Ps) labelled with 14C acetate as the sole carbon source. After labelling the cells, total RNA was isolated and hybridized to a nylon membrane containing spots of ssDNA. (a) The arrangement of the membrane: asymmetric PCR products generated with primers 907R (upper row) and 1378R (lower row) obtained from the pure cultures of the model community (Ta, Thauera aromatica; Bs, Bacillus subtilis; Ec, Escherichia coli; Ps, Pseudomonas stutzeri; Az, Azoarcus sp.) were transferred to the membrane. (b) The membrane was incubated with a storage phosphor screen for 24 h and the latent image was scanned for detection of radioactive signals.

(II) Molecular fingerprinting techniques have been developed for the fast comparison of numerous samples in order to circumvent the time-consuming cloning approach. The possibility of combined application of radioisotope probing and DGGE was investigated. A fingerprinting of the model community was established by mixing PCR products from the pure cultures and separating the product on a denaturing gradient gel in parallel with the singleplex amplicons (Fig. 3). The radioisotopically labelled RNA was then hybridized to the DNA transferred to a nylon membrane from the DGGE pattern. Once again, the strongest signal was obtained from the Ps-specific band from the pattern and the two lanes containing only Ps amplicons (Fig. 2b). Slight nonspecific hybridization was observed with the closely related Pp and moderately related Ta PCR products. Single-stranded conformation polymorphism (Schwieger & Tebbe, 1998) patterns may also be used; in theory, they are better templates for hybridization with 16S rRNA in the case of patterns containing the antisense ssDNA.

3

Approach II: RIP experiment combined with DGGE. Radioisotopically labelled RNA obtained from a Ps culture was hybridized to a membrane containing a blot of DGGE-separated PCR amplicons. (a) DGGE pattern of the model community (Mix) consisting of Ps, Pseudomonas stutzeri; Pp, Pseudomonas putida; Ta, Thauera aromatica; Ec, Escherichia coli; Bs, Bacillus subtilis; Az, Azoarcus sp. This complex product was run in parallel with PCR products of the singleplex amplicons of the respective strains. Two holes serving as position markers were labelled with paper triangles. (a) DGGE gel was scanned for fluorescent HEX labelling. (b) The gel was electroblotted to a membrane and hybridized with 14C-labelled RNA isolated from Ps. After washing the membrane, it was incubated for 12 h tightly pressed to a storage phosphor screen. The latent image of the phosphor screen was scanned with a fluorescent scanner. Dark bands indicate the presence of hybridized 14C-labelled RNA. The position of the marker holes was labelled by radioactive RNA.

The two model-community experiments proved that the combination of radioisotope probing with the DGGE or cloning approaches makes the determination of substrate-utilizing members possible. The only specific equipment demand is the fluorescent scanner for the readout of the latent image from the storage phosphor screen. The signals on the dried membranes have extreme long-term stability and do not pose significant health risk due to the weak β-emitting nature of 14C carbon, which is far below the regular safety threshold of radioisotopes [e.g. 270 µCi (107 Bq) for 14C isotope in Germany]: the membranes can be stored in simple envelopes for decades. The storage phosphor technology can detect 14C-labelled samples at a sensitivity level of around 25 disintegrations per mm2 and has a linear dynamic range providing a more sensitive alternative to X-ray films (Johnston et al., 1990).

However, it must be noted that several biases may be associated with the radioisotope probing approach which was not investigated in the present experimental setup. Differences in cell lyses and in efficiency of RNA extractions may distort the results; therefore, a combination of different cell disruption techniques to achieve representative amount and distribution of RNA has to be accentuated. Another potential source of bias is the differential PCR amplification using primers hybridizing to the highly conserved regions of the 16S rRNA genes (Reysenbach et al., 1992), which may cause the key player not to be represented in a clone library if it is numerically not abundant and/or if the applied primers have mismatches with the target sequences at certain positions. Nevertheless, a recent study suggested that using a low annealing temperature for PCR can reduce preferential amplification. (Sipos et al., 2007). The cross-feeding of the metabolites of the labelled substrate or the predation of the target bacteria is an additional limitation that is inherent in all tracer experiments. Targeting rRNA by radioisotope probing instead of DNA can minimize this effect but cannot totally exclude it in case of investigating slow-growing bacteria with relatively slow turnover of ribosomes.

Even though the application of molecular techniques has revolutionized environmental microbiology, the toolkit currently available is far from being perfect and complete in terms of functional analysis and assessment of the activity of nonculturuble members of microbial communities. The proposed approaches complement the existing molecular toolbox and provide a new method in answering key questions in microbiology. The combination of sequencing of 16S rRNA gene-based clones or DGGE bands with RIP permits the rapid identification of the active metabolizing microorganisms in a real ‘random walk’ approach.

Acknowledgements

This project was financially supported by Helmholtz Centre for Environmental Research — UFZ. Anna Rusznyák was funded by an EU Marie Curie short-term fellowship (AXIOM contract number: MEST-CT-2004-008332). The authors would like to thank the team of the Institute for Interdisciplinary Isotope Research, Leipzig, for assistance with the radioisotope work.

Footnotes

  • Editor: Elizabeth Baggs

References

View Abstract