INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Staphylococcus aureus has been recognized as an important pathogen in human disease. S. aureus is a common cause of community-acquired infections, including endocarditis, osteomyelitis, septic arthritis, pneumonia, and abscesses (26, 51). One reason for the occurrence of the ubiquitous infections caused by this pathogen is its ability to adhere to inert surfaces of medical implantable devices through interaction with deposited host factors (13, 17, 19, 20). Another reason is that the organism colonizes biologic substrates, as in endocarditis. On these inert or physiologic surfaces, S. aureus may proliferate as a structured community of bacterial cells enclosed in a self-produced polymeric matrix (10).

Microorganisms like S. aureus living in a biofilm are phenotypically resistant to a large variety of antimicrobial agents (9, 30). Several mechanisms have been put forward to explain antimicrobial resistance and the marked tendency for persistent infection in these settings. (i) Phenotypical resistance of biofilm microbes to antibiotics may be a result of the failure of an agent to penetrate the full depth of the biofilm (30); however, certain compounds have been shown to readily penetrate biofilms (10, 31). (ii) Some of the cells in a biofilm may experience nutrient limitation and therefore exist in a slow-growing or starved state; slow-growing or nongrowing cells display reduced susceptibilities to many antimicrobial agents (3, 4, 9, 11). (iii) In response to growth on a surface, adherent bacteria may express a pattern of genes different from that of their planktonic counterparts (10). It has been demonstrated with Escherichia coli that the levels of gene expression between biofilm and planktonic populations differ markedly (40). Currently, it is unclear whether these differences are a result of a programmed response to growth on a surface, a consequence of altered requirements of nutrients or metabolic product accumulation, and/or a reflection of quorum-sensing mechanisms due to autoregulatory peptide function (22).

The aim of this study was to identify genes in adherent S. aureus populations that are differentially expressed compared to those in their planktonic counterparts. A variety of methods to study differential levels of gene expression in prokaryotes have been described previously (16, 46, 48). These include differential-display PCR, arbitrarily primed PCR, gene fusion, and subtractive and differential hybridization. Furthermore, a number of microarray-based methods for the detection of differentially expressed genes have been described (12, 43). Most of these methods have the disadvantage that large quantities of mRNA are required. Some of these methods, like differential-display PCR, arbitrarily primed PCR, and gene fusion, do not eliminate sequences common to both, a feature that complicates the interpretation of the results and the identification of the differentially expressed genes. Other methods, like the previously described subtractive and differential hybridization techniques, are not capable of eliminating the large amount of rRNA from the total RNA, and complicated steps for mRNA enrichment have to be performed (15, 38, 50). Methods for mRNA enrichment are time-consuming, may result in the loss of some mRNAs, and therefore may reduce the overall sensitivity of the subsequent subtractive technique to detect differences in genes of limited expression and regulatory genes. Microarray methods present an attractive option for investigating differential levels of gene expression of staphylococci in the future (12, 43). However, whole-genome arrays for S. aureus are not yet available and their application is not yet standardized or validated.

In this study a micro-representational difference analysis of cDNA (cDNA micro-RDA) was performed. The protocol is an adaptation of the RDA method first described for applications to eukaryotic genomes by Lisitsyn et al. (25), combined with a phenol emulsion reassociation technique (23, 27). Unlike with previously described cDNA RDA protocols, the method described herein has been successfully applied to gram-positive bacteria and has the advantages that only minimal quantities of RNA are needed and, most importantly, that total RNA can be used. The large amount of rRNA in total RNA does not interfere with the micro-RDA procedure, so an mRNA enrichment prior to subtractive hybridization is not necessary.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Bacterial strains. In order to appropriately test the study question, we needed to use an S. aureus isolate forming a macroscopically visible and stable biofilm. For this purpose, a quantitative assay for biofilm formation (8, 37) with modifications described by Heilmann et al. (18) was employed. After the screening of 20 strains or isolates, S. aureus DSM 20231 was selected.

Culture conditions for S. aureus DSM 20231 was cultured overnight in tryptic soy broth (TSB) supplemented with 0.25% glucose. The threonine-supplemented TSB-glucose medium contained 1% L-threonine. To prepare biofilm cultures, 100 µl of a fresh overnight culture was used to inoculate 50 ml of TSB-glucose in a 100-ml Erlenmeyer flask. The flask was incubated at 37°C for 24 h without agitation. For preparation of a planktonic culture, 80 µl of the same overnight culture used for biofilm cultivation was incubated in 40 ml of TSB-glucose at 37°C for 24 h under constant agitation (150 rpm) in a 100-ml Erlenmeyer flask.

RNA isolation. After 24 h of cultivation, the S. aureus biofilm was carefully washed with 4°C sterile double-distilled water to remove all cells not adhering to the flask. Subsequently, biofilm cells were resuspended in cold sterile double-distilled water by rapidly flushing the flask with water by using a pipette until no visible biofilm was left on the glass surface. During the whole procedure, the cells were put on ice to prevent RNA degradation. Both the planktonic culture and biofilm cells were centrifuged for 5 min at 3,345 × g at 4°C. Both pellets were resuspended in RLT buffer (Qiagen GmbH, Hilden, Germany) and then mechanically disrupted with 0.1-mm-diameter zirconia-silica beads in a Fast Prep FP120 instrument (Qbiogene, Heidelberg, Germany) at maximum speed for 20 s. RNA isolation from this lysate was performed with an RNeasy mini kit (Qiagen) according to the instructions of the supplier. The isolated total RNA was treated with DNase (Oncor Appligene, Heidelberg, Germany).

cDNA synthesis. cDNA synthesis was performed by random priming with the SuperScript choice system for cDNA synthesis (Life Technologies GmbH, Karlsruhe, Germany). The protocol was performed according to the manufacturer's instructions with the following exceptions. The first-strand synthesis was continued up to 2 h, and the second-strand synthesis was incubated overnight. The double-stranded cDNA was subsequently isolated using a PCR purification kit (Qiagen).

Oligonucleotides. Sequences of adapters and primers used for the micro-RDA were as follows: that of R-Bgl 12 was 5'-GATCTGCGGTGA-3', that of R-Bgl 24 was 5'-AGCACTCTCCAGCCTCTCACCGCA-3', that of J-Bgl 12 was 5'-GATCTGTTCATG-3', that of J-Bgl 24 was 5'-ACCGACGTCGACTATCCATGAAC-A-3', that of N-Bgl 12 was 5'-GATCTTCCCTCG-3', and that of N-Bgl 24 was 5'-AGGCAACTGTGCTATCCGAGGGAA-3' (25). Primer set 1 (R series) was used for amplicon preparation, and primer sets 2 (J series) and 3 (N series) were used alternately for the selective amplifications.

RDA. For amplicon synthesis, both tester and driver cDNAs were digested with DpnII (New England Biolabs GmbH, Frankfurt am Main, Germany). Subsequently, 100 ng of each restricted cDNA population was ligated to 0.5 nmol of the adapter R-BglI2 and 0.5 nmol of the adapter R-Bgl 24. After ligation, tester and driver populations were PCR amplified. Fifty nanometers of template DNA was used for each amplicon PCR. In addition, the PCR mixture contained 67 mM Tris-HCl (pH 8.8), 4 mM MgCl₂, 16 mM (NH₄)₂SO₄, 10 mM -mercaptoethanol, 100 µg of bovine serum albumin per ml, 300 µM deoxynucleoside triphosphate mix, and 15 U of Taq polymerase (Roche Diagnostics GmbH, Mannheim, Germany) (25). For priming, oligonucleotide R-Bgl 24 was used. The reaction mixtures (containing DNA, buffer, and the deoxynucleoside triphosphate mix) were preheated to 72°C, and then the Taq polymerase was added and the fill-in reaction was started. After an incubation period of 10 min, the PCR was started by adding 1.75 µM R-Bgl 24 to the reaction mix (17 cycles of 95°C for 50 s and 72°C for 3 min; the last cycle was followed by an extension at 72°C for 10 min). After amplification, both tester and driver amplicons were redigested with DpnII to cleave the adapters. Digested adapters were always removed with the PCR purification kit (Qiagen). Subsequently, only tester fragments were ligated to J-Bgl oligonucleotides.

View larger version (21K): [in this window] [in a new window]   FIG. 1.   Schematic illustration of the micro-RDA protocol. Broken lines represent the driver population, and solid lines represent the tester fraction. Small solid boxes symbolize the Bgl adapter series. This diagram illustrates cDNA synthesis, amplicon synthesis (A), and the synthesis of DP 1 via subtractive hybridization and selective amplification (B). To generate DP 2 and DP 3, the adapters have to be changed and subtractive hybridization and selective amplification have to be repeated. ss, single stranded; ds, double stranded.

Micro-RDA sensitivity test. To evaluate the sensitivity of micro-RDA, a series of spiked controls containing various amounts of MS2 bacteriophage RNA (Roche Diagnostics) added to a fixed amount of S. aureus driver RNA was used as the tester fraction against driver RNA in the micro-RDA procedure. To prepare a sample containing one copy of the MS2 RNA spike per S. aureus cell, 98 pg of MS2 RNA was added to 5 µg of total S. aureus RNA. This ratio takes into account the molecular weight of MS2 RNA (1.18 × 10⁶ g/mol) and assumes that there is 0.1 pg of total RNA per S. aureus cell. Additional concentration ratios of 5, 10, 50, and 100 copies/cell were prepared accordingly.

Cloning and sequencing of the DPs. The DPs were cloned into the pCRII or pCR2.1 vector (Invitrogen BV/Novex, Groningen, The Netherlands), and the plasmids were transformed into E. coli InvF' cells (Invitrogen BV/Novex). For plasmid isolation E. coli InvF' cells containing the pCRII or the pCR2.1 vector were grown overnight in 5 ml of Luria-Bertani medium supplemented with 100 µg of ampicillin per ml at 37°C with constant agitation. The inserts were sequenced using the T7 or M13 reverse primer on a Li-Cor 4000 sequencer (MWG-Biotech GmbH, Ebersberg, Germany). Resulting sequences were compared to the S. aureus genome at The Institute for Genomic Research Microbial Database and at the Sanger Center using the BLAST program (1) and to sequences in the EMBL procaryote library using the Fasta program (36).

Northern blot analysis. The RNA was electrophoresed through 1.5% agarose-0.66 M formaldehyde gel in MOPS (morpholinepropanesulfonic acid) running buffer and transferred to a nylon membrane (Nytran N; Schleicher & Schuell GmbH, Dassel, Germany) by alkaline downward blotting (7). Gene-specific DNA probes were amplified using the Bgl 24 oligonucleotide ligated to the DP (N-Bgl 24, DP 2; J-Bgl 24, DP 3). Nonradioactive labeling was carried out using a PCR DIG Probe Synthesis Kit (Roche Diagnostics) for the synthesis of DNA probes and a DIG RNA Labeling Kit (SP6/T7; Roche Diagnostics) for RNA probes. Hybridization and detection conditions were according to information from the supplier (Boehringer Mannheim).

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

The micro-RDA technique. As a first step, the RDA protocol first described by Lisitsyn et al. (25) had to be modified and validated for the S. aureus system. The micro-RDA technique consists of five major elements: isolation of total RNA, cDNA synthesis, amplicon synthesis, subtractive hybridization, and selective amplification (Fig. 1).

View larger version (83K): [in this window] [in a new window]   FIG. 2.   Micro-RDA sensitivity test. Different amounts, ranging from 1 copy per cell up to 100 copies per cell, of a known RNA species (MS2 RNA) was mixed with total RNA of S. aureus. In the subsequent micro-RDA, this mixture was used as the tester and total S. aureus RNA was used as the driver. DP 3 was electrophoresed through a 2% agarose gel and stained with ethidium bromide. Fragments typical of the MS2 amplicon could be detected in the spike with a ratio of minimally 5 copies per cell. Lane 1, the S. aureus amplicon (MS2 RNA) at 100 copies/cell in RNA (3-h culture); lanes 2 to 6, DP 3 (MS2 RNA spike in RNA [3-h culture]) at 1 copy of MS2/cell and 5, 10, 50, and 100 copies of MS2/cell, respectively; lane 7, the MS 2 amplicon; lanes M, molecular size markers.

RDA of biofilm versus growing planktonic populations. Using the modified cDNA micro-RDA technique described here, biofilm RNA was used as the tester and planktonic RNA from S. aureus cells cultivated in TSB-glucose medium was used as the driver (Fig. 3, lanes 1 to 3). Subsequently, tester and driver assignment of the amplicons was reversed (Fig. 3, lanes 4 to 6). Amplicons as well as DPs 1 and 2 were separated in an agarose gel. After one round of micro-RDA, no significant differences between the banding patterns of the amplicon and DP 1 could be detected in either subtraction; only bands corresponding to rRNA were visible (Fig. 3, lanes 1 and 2 and lanes 4 and 5). To amplify fragments of differentially expressed genes, a second round of micro-RDA was needed. As was observed in the micro-RDA sensitivity test, smaller fragments were favored during the selective PCR cycles: fragments ranging in size from 100 to 450 bp were selectively enriched. Shotgun cloning and sequencing of DP 2 obtained by subtraction using biofilm-derived RNA as the tester and planktonic-population-derived RNA as the driver population allowed us to identify multiple fragments of a threonyl-tRNA synthetase, a triosephosphate isomerase, and the 23S rRNA from S. aureus, as well as a single fragment with 60% homology to the phosphoglycerate mutase of Bacillus subtilis (Table 1). Northern blot analysis confirmed that the threonyl-tRNA synthetase (Fig. 4A), the phosphoglycerate mutase (Fig. 4B), and the triosephosphate isomerase (Fig. 4C) were differentially expressed in biofilm and planktonic cultures of S. aureus.

View larger version (78K): [in this window] [in a new window]   FIG. 3.   Micro-RDA of sessile versus planktonic S. aureus populations grown in unsupplemented medium. The biofilm-forming and planktonic populations were grown in TSB-glucose medium. The amplicons and DPs were electrophoresed through a 2% agarose gel and stained with ethidium bromide. Lanes 1 to 3, amplicon and DPs of micro-RDA using biofilm-derived RNA as the tester and plankton-derived RNA as the driver; lanes 4 to 6, amplicon and DPs of micro-RDA using plankton-derived RNA as the tester and biofilm-derived RNA as the driver; lane 1, amplicon biofilm; lane 4, amplicon planktonic culture; lanes 2 and 5, DP 1; lanes 3 and 6, DP 2; lane M, molecular size markers.

View larger version (59K): [in this window] [in a new window]   FIG. 4.   Northern blot analysis of S. aureus DSM 20231 RNA derived from biofilm (lanes 1 and 3) and planktonic (lanes 2 and 4) cultures probed with digoxigenin-labeled DPs (A to E) obtained by micro-RDA. Lanes 1 and 2, RNAs derived from cells cultivated in TSB-glucose medium; lanes 3 and 4, RNA derived from cells cultivated in TSB-glucose medium supplemented with 1% threonine. (A) DP 11/10 (fragment of the threonyl-tRNA synthetase); (B) DP P4 (with homology to the phosphoglycerate mutase of B. subtilis); (C) DP P9 (fragment of the triosephosphate isomerase); (D) DP 11/6 (with homology to the alcohol dehydrogenase of Zymomonas mobilis); (E) DP 11/1 (homology to clpC of B. subtilis).

View larger version (23K): [in this window] [in a new window]   FIG. 5.   Micro-RDA of sessile versus planktonic S. aureus populations grown in threonine-supplemented medium. Lanes 1 to 4, amplicon and DPs of micro-RDA using biofilm-derived RNA as the tester and planktonic-culture-derived RNA as the driver. The amplicon and DPs were electrophoresed through a 2% agarose gel and stained with ethidium bromide. Lane 1, biofilm amplicon; lane 2, DP 1; lane 3, DP 2; lane 4, DP 3; lanes M, molecular size markers.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

The aim of this study was to identify genes which are typically expressed in sessile S. aureus populations in contrast to the genes expressed in their planktonic counterparts. To achieve this goal, a positive-selection RDA method was chosen. Hitherto, adaptations of this method were used only for genomic subtractions (25), for eukaryotic cDNA subtractions (21, 34), and for cDNA subtraction of the gram-negative bacteria Neisseria meningitidis (2) and Pseudomonas aeruginosa (52). In this study we have further developed a cDNA micro-RDA (35) for use with gram-positive bacteria. In contrast to the previously described cDNA subtraction methods, the micro-RDA technique described here eliminates most of the highly abundant rRNAs without the use of methods for enriching for transcribed gene fragments in total RNA or total cDNA (15, 38, 50) or removing the rRNA by spiking the driver fraction with rRNA molecules with its inherent difficulty to adapt this method to other bacterial species (2). This marks significant progress in the feasibility, sensitivity, and ease of subtractive methods for the analysis of prokaryotic gene expression.

To achieve the high stringency necessary for the removal of all rRNA molecules, two modifications to the original method were critical. First, the amount of driver DNA in the hybridization reaction mixture was increased 10-fold after each subsequent selective enrichment. Second, to ensure complete hybridization, a phenol emulsion reassociation step was introduced. The PERT increases the rate of hybridization by decreasing the aqueous volume (46). The reduction of the aqueous volume also allows a significant reduction in the amount of tester cDNA required: only 1 ng of tester DNA was used during subtractive hybridization. Stable phenol emulsion was maintained by thermal cycling (27). Since phenol depresses the T_m of DNA duplexes (23), raising the temperature to 65°C not only allows one to maintain the phenolic emulsion but also causes nonhomologous DNA duplexes to disassociate, thereby reducing the number of false-positive DPs.

In the RDA technique described earlier (2), four rounds of conventional RDA had to be performed. In micro-RDA, only two rounds of subtractive hybridization and selective amplification appear to be sufficient for generation of specific DPs. A third round of micro-RDA may lead to a further reduction in the number of false-positive fragments, but on the other hand, this may increase the loss of rare transcripts during the steps of adapter changes.

Despite the stringent conditions, micro-RDA is as sensitive as conventional RDA (34). The micro-RDA sensitivity test confirmed that even low-copy-number mRNAs could be detected (Fig. 2). Smaller fragments hybridize more efficiently and are amplified more efficiently by PCR. The amplification efficiencies of PCR products are also affected by the nucleotide sequence of the DNA fragment (49). Thus, the probability of identifying differentially expressed genes depends also on the chosen restriction enzyme. A frequently cutting restriction enzyme like the 4-base cutter DpnII should be used during amplicon synthesis. Another possibility for identifying rare transcripts would be the spiking of the driver fraction with highly abundant tester fragments. Apart from high-stringency subtraction conditions, in our study additional factors may have contributed to a limited number of detected differentially expressed transcripts. Planktonic or biofilm populations may consist of subpopulations with variant expression phenotypes as a result of the localization of the bacteria or environmental factors such as pH and oxygen levels, resulting in microheterogeneity of the subpopulations. Furthermore, despite the preparative precautions, some cross-contamination of the biofilm population with bacteria from the planktonic population might have occurred during the harvest and might suffice for removal of different transcripts under high-stringency conditions. Therefore, the five genes identified in our study do not necessarily or likely comprise all genes differently expressed in an S. aureus biofilm population. However, our analysis allows recognition of differently expressed transcripts whose identification would have been difficult to achieve using targeted expression analysis approaches.

In this study we identified five genes which are differentially expressed in biofilm and planktonic populations of S. aureus. Three of the upregulated genes, those for phosphoglycerate mutase, triosephosphate isomerase, and alcohol dehydrogenase, encode enzymes of the glycolysis or fermentation pathway. It is known that oxygen is limited in deeper layers of biofilm (53), and it has been shown for E. coli that the gene expression pattern within a biofilm is altered due to oxygen limitation (40). Thus, the upregulation of these three enzymes may be a consequence of oxygen limitation within the biofilm. On the other hand, it has been shown that glycolytic enzymes possess additional properties when they are located on the cell surface. For example the -enolase of streptococci display a strong plasmin(ogen) binding activity on a streptococcal surface (33). In S. aureus, a surface-associated transferrin-binding protein was identified as the glycolytic enzyme glyceraldehyde-3-phosphate dehydrogenase (GAPDH) (28, 29). Thus, it is intriguing to speculate that, in biofilms, the expression of the glycolytic enzymes may be regulated by stimuli other than oxygen tension and may reflect their putative roles in a complex microbial population. Previously, it was shown that the glycolytic enzymes, including the triosephosphate isomerase, the phosphoglycerate mutase, and the -enolase, are clustered in an operon. This operon is conserved in Staphylococcus, Bacillus, and Lactobacillus spp., and Northern blot analysis indicated the existence of multiple mRNA species (B. Modun, J. Morrissey, A. Cockayne, and P. Williams, Abstr. 9th Int. Symp. Staphylococci Staphylococcal Infect., abstr. 155, 2000). Despite highly stringent conditions (68°C, buffer containing 50% formamide), Northern analysis of the triosephosphate isomerase showed several hybridization products in the biofilm RNA but not in the planktonic culture (Fig. 4C). This finding may indicate the existence of multiple mRNA species of this gene during biofilm formation.

Furthermore, it was shown that threonyl-tRNA synthetase is upregulated in an S. aureus biofilm. It has been proposed that at least some of the cells in a biofilm may experience nutrient limitation and therefore exist in a slow-growing or starved state (3, 4, 9). Since the expression of this aminoacyl-tRNA synthetase is downregulated in biofilms after threonine supplementation, it seems to be possible that threonine starvation is the reason for this finding.

The fifth gene that is upregulated in S. aureus biofilms is the clpC homologue. It encodes the ClpC ATPase, a general stress protein, and can be found in a large variety of prokaryotic and eukaryotic organisms. It has multiple functions, e.g., it participates in the degradation of misfolded proteins and is involved in sporulation, cell division, and the regulation of the competence genes and several virulence factors (5, 24, 39). In degradation, ClpC acts as the ATPase partner of ClpP protease. Since it has been shown that clpP is essential for biofilm formation in Pseudomonas fluorescens (32), it is plausible that this gene also plays an important role in the biofilm formation of S. aureus.

In conclusion, this study details an adaptation of RDA that enables highly specific detection of differentially expressed mRNA species in two different S. aureus populations. Five genes specifically expressed in an S. aureus biofilm were identified. Further analysis of the functions of these genes may provide insights into environmental sensing and metabolism within staphylococcal biofilms. Micro-RDA proved to be a valuable method for detection of differentially expressed genes in complex microbial systems.