SEARCH

SEARCH BY CITATION

Keywords:

  • biofilms;
  • sessile cells;
  • phenotypic adaptation;
  • gene expression;
  • stress;
  • antibiotics

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. The search for biofilm-specific gene expression patterns: C. albicans as an example
  5. Stress-induced gene expression in biofilms and phenotypic adaptation
  6. Problems associated with reliably identifying gene expression patterns in biofilms and implications for identifying the response to stress
  7. Outlook
  8. Acknowledgements
  9. References

A better understanding of the genotypic and phenotypic adaptation of sessile (biofilm-associated) microorganisms to various forms of stress is required in order to develop more effective antibiofilm strategies. This review presents an overview of what high-throughput transcriptomic analyses have taught us concerning the response of various clinically relevant microorganisms (including Pseudomonas aeruginosa, Burkholderia cenocepacia and Candida albicans) to treatment with antibiotics or disinfectants. In addition, several problems associated with identifying gene expression patterns in biofilms in general and their implications for identifying the response to stress are discussed (with a focus on heterogeneity in microbial biofilms and the role of small RNAs in microbial group behavior).


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. The search for biofilm-specific gene expression patterns: C. albicans as an example
  5. Stress-induced gene expression in biofilms and phenotypic adaptation
  6. Problems associated with reliably identifying gene expression patterns in biofilms and implications for identifying the response to stress
  7. Outlook
  8. Acknowledgements
  9. References

Biofilms are microbial communities containing sessile cells embedded in a self-produced extracellular polymeric matrix (containing polysaccharides, DNA and other components). In comparison with their planktonic (free-living) counterparts, sessile cells are often much more resistant to various stress conditions (including treatment with antimicrobial agents) and this increased resistance has a considerable impact on the treatment of biofilm-related infections (Fux et al., 2005). Several mechanisms are thought to be involved in biofilm antimicrobial resistance including (1) slow penetration of the antimicrobial agent into the biofilm, (2) changes in the chemical microenvironment within the biofilm, leading to zones of slow or no growth, (3) adaptive stress responses and (4) the presence of a small population of extremely resistant ‘persister’ cells (Mah & O'Toole, 2001; Stewart & Costerton, 2001; Donlan & Costerton, 2002; Gilbert et al., 2002a, b).

In a first part of this review, I will highlight the problems associated with the study of gene expression in biofilms, using a set of studies on the human-pathogenic fungus Candida albicans as an example. Subsequently, I will review the recent literature on differential gene expression in a number of microbial biofilms in response to stress (with a focus on stress related to exposure to antibiotics and reactive oxygen species) and link that to phenotypic adaptation.

The search for biofilm-specific gene expression patterns: C. albicans as an example

  1. Top of page
  2. Abstract
  3. Introduction
  4. The search for biofilm-specific gene expression patterns: C. albicans as an example
  5. Stress-induced gene expression in biofilms and phenotypic adaptation
  6. Problems associated with reliably identifying gene expression patterns in biofilms and implications for identifying the response to stress
  7. Outlook
  8. Acknowledgements
  9. References

Earlier work [reviewed by Sauer (2003), Beloin & Ghigo (2005) and Lazazzera (2005)] indicated that, although gene expression patterns in biofilms often differed remarkably from those in planktonic cells, finding common biofilm gene expression patterns between different studies (even those using the same organisms) was difficult. This was attributed to the minimal overlap between the functions involved in biofilm formation and the fact that subsets of genes expressed in biofilms are also expressed under various planktonic conditions.

Candida albicans is a commensal fungus of healthy human individuals and can cause superficial and systemic infections when the immune defenses are repressed or when the normal microbial flora is disturbed. Candida albicans infections are often associated with the formation of biofilms (Douglas, 2003). A first comprehensive transcriptome analysis of biofilm formation in C. albicans was presented by Garcia-Sanchez et al. (2004). In this study, gene expression in various biofilm model systems (microfermentor, catheter disks and microtiter plate) was compared with the expression in planktonic cultures. Three different strains were tested (SC5314, CAI4 and CDB1) and several environmental parameters (medium flow, glucose concentration, aeration, time and temperature) were varied. Despite the marked differences in the growth conditions, the correlation coefficients for the biofilm–biofilm comparisons were high (between 0.80 and 0.97), while comparing gene expression profiles between planktonic cultures or between biofilms and planktonic cultures resulted in lower correlation coefficients (0.54–0.80 and 0.54–0.81, respectively). Three hundred and twenty five genes were identified as being differentially expressed between biofilm and planktonic conditions (214 genes were activated in biofilms, and 111 were repressed). In this set, genes involved in protein synthesis, amino acid, lipid and nucleotide metabolism, transcription and control of the cellular organization are over-represented. A high fraction of the 214 overexpressed genes are related to the synthesis of amino acids and many of these are homologues of genes that are under the control of Gcn4p in Saccharomyces cerevisiae. Reduced biofilm formation in a C. albicans gcn4/gcn4 mutant confirmed the requirement for a functional Gcn4p for normal biofilm formation. In addition, ALS1 (thought to be involved in adhesion) was identified as the major overexpressed genes in biofilms, while other genes of the ALS family were underexpressed (ALS7) or not differentially expressed at all (ALS5, ALS10). In a second transcriptome study, Murillo et al. (2005) focused on gene expression in the early phases of C. albicans biofilm formation (30–390 min). Forty-one genes were identified as being differentially upregulated in biofilms compared with planktonic cultures, while 25 genes were downregulated in biofilms. Nine of these 41 genes encode proteins involved in sulfur metabolism, suggesting an upregulation of the entire sulfur-assimilation pathway in early biofilm cultures. A second set of genes differentially upregulated in young biofilms is associated with phosphate metabolism. Marchais et al. (2005) identified 117 differentially expressed genes (77 overexpressed in adherent cells and 40 underexpressed). Among the overexpressed genes, 22% played a role in cellular organization and intracellular transport, 10% were involved in amino acid and protein metabolism, 7% in carbohydrate metabolism, 5% in lipid and fatty acid metabolism and 5% in transcription, but the majority (46%) had no known function. Yeater et al. (2007) determined the gene expression profiles in two C. albicans strains grown on two different substrates (denture and catheter) at three different time points (representing early, intermediate and mature biofilms). Two hundred and forty three genes were differentially expressed in either biofilm or planktonic specimens, over the experimental time course, while 191 genes were differentially expressed between biofilm and planktonic cells at the three developmental time points studied. Data from this study indicated that the assimilation of carbohydrates (both through glycolytic and nonglycolytic routes), amino acid metabolism and intracellular transport mechanisms are important in the early phase (6 h) of biofilm formation. During the intermediate phase (12 h), sessile cells have a high energy demand and use specific transporters for amino acids, sugars, ions, oligopeptides and lactate/pyruvate. At the 48-h time point, few genes were differentially expressed.

Zakikhany et al. (2007) and Nett et al. (2009) took the study of gene expression in C. albicans biofilms to the next level by performing transcriptome analyses on biofilms grown in more elaborate model systems that more closely mimic human infections. Zakikhany and colleagues compared the expression in sessile C. albicans cells grown on reconstituted human oral epithelium (RHE) for various time points (1–24 h postinoculation) with that in planktonic cells (grown to the midexponential phase). It turned out that 15% of the approximately 4300 reliably expressed genes were ≥2-fold upregulated at one or more time points. One hour postinoculation, 164 genes were upregulated, of which 29 were only upregulated at this time point. The majority of these ‘early-only’ genes (21/29) had no known function, while others were involved in cellular functions such as transcription. Thirty-eight genes were significantly overexpressed throughout the entire experiment (1–24 h). Several of these were hyphae specific or at least hyphae associated (including HWP1 and ALS3), indicating that contact with the epithelial cells induces hyphae formation. Identification of genes that were only upregulated in later stages (12 or 24 h postinoculation) showed that these were mainly involved in metabolic functions and suggested a shift toward the use of molecules other than glucose as a carbon source (e.g. lipid-derived two-carbon compounds). Interestingly, when the results were compared with the results obtained with mRNA recovered from 11 HIV-positive patients with pseudomembranous candidiasis, 38 genes that were increased at all time points in the RHE model also showed an increased expression in the patient samples. These 38 genes included hyphae-associated genes (including HWP1 and ALS3) as well as genes involved in the utilization of two-carbon compounds via the glyoxylate cycle (Zakikhany et al., 2007). In the study of Nett and colleagues, biofilms were grown on catheters inserted into the jugular vein of rats (Andes et al., 2004). Samples taken from these central venous catheters at two time points (12 h, intermediate growth and 24 h, mature) were compared with in vitro grown planktonic cells. One hundred and twenty four genes were upregulated in biofilms at both time points, compared with the expression in planktonic cells. The majority of these genes were involved in transcription and protein synthesis (13%), energy and metabolism (12%), carbohydrate synthesis and processing (10%) and transport (6%), while 35% of the 124 genes had no function assigned to them. Twenty-seven genes were downregulated at both time points; 30% of these genes were involved in DNA processing.

Besides the above-described transcriptomics studies, several research groups have used proteomics to identify differentially expressed proteins. Thomas et al. (2006) identified nine differentially expressed cell-surface-associated proteins; seven [Hsp70, pyruvate decarboxylase, inositol-1-phosphate synthase, enolase (ENO1), O-acetylhomoserine O-acetylserine sulfhydrylase (MET15), alcohol dehydrogenase 1 and inosine-5′-monophosphate dehydrogenase] were overexpressed, while two (alcohol dehydrogenase 2 and malate dehydrogenase) were downregulated in biofilms. Interestingly, not a single surface-associated protein was identified as being solely expressed in sessile or planktonic cells. Nineteen proteins were significantly overexpressed in C. albicans biofilms grown in 24-well microtiter plates, compared with planktonic cultures, and in contrast to the results obtained by Thomas et al. (2006), ENO1 was twofold underexpressed. Highly significant overexpression was observed for citrate synthase (14.45-fold), and several proteins involved in oxidative stress, including alkyl hydroperoxide reductase AHP1 and several other reductases (GRP2, MCR1, TSA1, PST1 and TRX1), were also overexpressed. Proteomics has also been used for a three-way comparison of planktonic yeast cells, planktonic hyphae and sessile cells (Martinez-Gomariz et al., 2009). One hundred and seventy-five cytoplasmic and 70 cell surface-associated proteins were differentially expressed between sessile and planktonic yeast cells, while these numbers were 218 and 51, respectively, when sessile cells were compared with planktonic hyphae. The fold over- or underexpression varied considerably depending on the comparison made. For example, MET15 was downregulated in biofilms when compared with planktonic yeast cells, but upregulated when biofilms were compared with planktonic hyphae, confirming that morphology is an important factor. Further complicating the comparison of protein expression is the presence of various isoforms of the same protein. For example six isoforms of pyruvate decarboxylase were identified by Martinez-Gomariz and colleagues: isoforms 1, 2, 5 and 6 are underexpressed in biofilms compared with planktonic yeast cells, while isoforms 3 and 4 are overexpressed.

A detailed analysis of the results obtained in the studies summarized above reveals that, although generally representatives of particular classes of genes are differentially expressed between planktonic and sessile cells (Fig. 1), there is very little overlap between C. albicans genes identified as differentially expressed in different studies and the same is true for other microorganisms. The observation that the experimental conditions for culturing the cells before RNA extraction are often variable (Table 1) offers a first explanation. There are a wide range of biofilm model systems available, and few studies have used the same model system. Similarly, planktonic cells are cultured in a variety of ways (Table 1). In addition, there is growing evidence that there are marked differences in gene expression between different stages of biofilm formation, and as such, the comparison of gene expression profiles obtained in a particular model system at a particular time point with an expression profile obtained at another time point in another model system is probably of little relevance. In addition, while most studies with C. albicans were carried out with the reference isolate SC5314, a wider variety of isolates have been included in this kind of studies for other organisms. For example, for Escherichia coli strains MG1655 (Schembri et al., 2003; Ito et al., 2009a, b), TG and TG1 (Beloin et al., 2004), JM109 and ATCC 25404 (Ren et al., 2004), BW25113 (Domka et al., 2007) and PHL628 (Junker et al., 2007) have been used, as well as clinical isolates recovered from asymptomatic bacteriuria (Hancock & Klemm, 2007). Although several of these strains are listed as ‘K12’, subtle differences between them may confound the comparison of gene expression data. It is important to keep this in mind when looking for genotypic and/or phenotypic adaptation to stress in sessile cells, as the differential expression of particular genes due to differences in the environmental conditions in the test and control situation may introduce bias and lead to erroneous conclusions.

image

Figure 1. Candida albicans genes differentially expressed between planktonic and sessile cells, as identified in various studies. Data are expressed as fraction of the total number of differentially expressed genes identified in each study. The total number of genes identified as differentially expressed is also indicated.

Download figure to PowerPoint

Table 1.   Growth conditions for sessile and planktonic Candida albicans SC5314 cells as reported in various transcriptomic and proteomic studies
ReferenceBiofilm modelPlanktonic modelMediumFlowTemperature (°C)Sampling point
  • *

    YNB, Yeast–nitrogen base growth medium.

  • RHE, Reconstituted human epithelium.

  • YPD, Yeast–peptone–dextrose growth medium.

  • §

    § PMMA, poly-methyl-methacrylate.

Garcia-Sanchez et al. (2004)Microfermentor YNB*+0.4% glucoseContinuous3748 h
Microtiter plate YNB+0.4% glucoseLimited3748 h
Catheter disks YNB+0.4% glucoseLimited3772 h
Microfermentor YNB+2% glucoseContinuous3072 h
Microfermentor YNB+0.4% glucoseContinuous3772 h
 FlaskYNB+0.4% glucoseLimited3720 h
 FlaskYNB+0.4% glucoseLimited3748 h
 FlaskYNB+2% glucoseLimited3048 h
 Microtiter plateYNB+0.4% glucoseLimited3748 h
Murillo et al. (2005)Polystyrene Petri dishPolypropylene flaskF12 mediumLimited3730, 90, 150, 270 and 390 min
Thomas et al. (2006)Tissue culture flasksLiquid mediumRPMI-1640Limited3724 h
Yeater et al. (2007)Denture acrylic coated with salivaMicrotiter plateYNB+50 mM glucoseLimited376, 12 and 48 h
Zakikhany et al. (2007)RHE  Limited371, 3, 6, 12 and 24 h
 Liquid mediumYPDLimited37Mid-log phase
Seneviratne et al. (2008)Microtiter plateLiquid mediumYNB+100 mM glucoseLimited3748 h
Nett et al. (2009)Rat central venous catheter  Continuous3712 and 24 h
 Liquid mediumYPDLimited3712 and 24 h
Martinez-Gomariz et al. (2009)PMMA§ strips YPD+50 mM glucoseContinuous3748 h
 Liquid mediumYPD+50 mM glucose 3748 h
 Liquid mediumYPD+50 mM glucose 371.5–2 h

Stress-induced gene expression in biofilms and phenotypic adaptation

  1. Top of page
  2. Abstract
  3. Introduction
  4. The search for biofilm-specific gene expression patterns: C. albicans as an example
  5. Stress-induced gene expression in biofilms and phenotypic adaptation
  6. Problems associated with reliably identifying gene expression patterns in biofilms and implications for identifying the response to stress
  7. Outlook
  8. Acknowledgements
  9. References

Adaptation of sessile Pseudomonas aeruginosa cells to exposure to antibiotics

Pseudomonas aeruginosa was one of the first organisms in which gene expression in biofilms was studied, but surprisingly, when Whiteley et al. (2001) compared gene expression levels between cells grown on granite pebbles in a chemostat and cells grown in a liquid culture medium in the same chemostat, very few genes showed differential expression. When gene expression in untreated sessile P. aeruginosa PAO1 cells was compared with the expression in sessile cells treated with high doses of tobramycin [seven times the minimal inhibitory concentration (MIC) for planktonic cells], only 20 genes were differentially expressed (14 were activated and six were repressed). Ten of these genes code for hypothetical proteins with no known function; two additional genes code for hypothetical proteins of a Pf1-like bacteriophage. Upregulated genes include those involved in stress response (dnaK, groES) and efflux systems, while downregulated genes include both hypothetical phage proteins as well as the β-subunit of urease (Table 2). The tolA gene, whose product affects the lipopolysaccharide structure in such a way that the outer membrane has a decreased affinity for aminoglycoside antibiotics, was overexpressed in untreated sessile cells compared with planktonic cells, possibly leading to decreased aminoglycoside susceptibility in biofilms. Genes encoding cytochrome c oxidases (subunits I, II and III, encoded by PA0106, PA0105 and PA0108, respectively), on the other hand, were downregulated (2.7–2.9-fold) in untreated sessile cells when compared with planktonic cells. As cytochrome c oxidase is the terminal electron acceptor during aerobic growth and as aminoglycoside transport is coupled with terminal electron transport (Bryan et al., 1980), this downregulation is likely to confer reduced susceptibility as well.

Table 2.   Selection of Pseudomonas aeruginosa genes identified as being differentially expressed in treated vs. untreated sessile cells
GeneFunctionTreatmentUp (+)/ down (−)References
  1. Genes in bold were identified as differentially expressed in more than one study.

PA0105-0106Cytochrome c oxidase subunitsAzithromycinGillis et al. (2005)
PA0108Cytochrome c oxidase subunitAzithromycinGillis et al. (2005)
PA0376RpoHImipenem+Bagge et al. (2004)
PA0721Hypothetical protein (phage Pf1)TobramycinWhiteley et al. (2001)
PA0725Hypothetical protein (phage Pf1)TobramycinWhiteley et al. (2001)
PA0762-0764Alginate biosynthesisTobramycin+Anderson et al. (2008)
PA0762-0765Alginate biosynthesisImipenem+Bagge et al. (2004)
PA0913MgtETobramycin+Anderson et al. (2008)
PA0996-1000PQS biosynthesisTobramycinAnderson et al. (2008)
PA1078-1081Flagellar synthesisImipenem+Bagge et al. (2004)
PA1163NvdBTobramycin+Mah et al. (2003)
PA1172-1175Periplasmic nitrate reductaseAzithromycinGillis et al. (2005)
PA1178-1180OprH, PhoP, PhoQImipenem+Bagge et al. (2004)
PA1431RsaLAzithromycinGillis et al. (2005)
PA1432LasIAzithromycinGillis et al. (2005)
PA1541Probable drug-efflux proteinTobramycin+Whiteley et al. (2001)
PA1703-1722TT3S-related genesAzithromycin+Gillis et al. (2005)
PA1874-1877Efflux pumpTobramycin+Zhang & Mah (2008)
PA2191Adenylate cyclase ExoYAzithromycin+Gillis et al. (2005)
PA2367Hypothetical proteinImipenem+Bagge et al. (2004)
 AzithromycinGillis et al. (2005)
PA2703Hypothetical proteinTobramycin+Whiteley et al. (2001), Bagge et al. (2004)
PA3049Ribosome modulation factorTobramycin+Whiteley et al. (2001)
 AzithromycinGillis et al. (2005)
PA3540-3551Alginate biosynthesisImipenem+Bagge et al. (2004)
PA3574Probable transcriptional regulatorTobramycin+Whiteley et al. (2001)
PA3819Conserved hypothetical proteinTobramycin+Whiteley et al. (2001), Bagge et al. (2004)
PA3841-3843Exoenzyme S & chaperoneAzithromycin+Gillis et al. (2005)
PA3920Probable metal-transporting P-type ATPaseTobramycin+Whiteley et al. (2001)
PA4110AmpCImipenem+Bagge et al. (2004)
PA4306Type IVb pilinImipenemBagge et al. (2004)
 AzithromycinGillis et al. (2005)
PA4386GroESTobramycin+Whiteley et al. (2001)
PA4407-4417Cell wall biosynthesis, cell divisionImipenem+Bagge et al. (2004)
PA4525-4526Pili biosynthesisImipenemBagge et al. (2004)
PA4597-4599MexCD-OprJ efflux pumpAzithromycin+Gillis et al. (2005)
PA4635Hypothetical MgtC proteinTobramycin+Anderson et al. (2008)
PA4761DnaKTobramycin+Whiteley et al. (2001)
PA4825MgtATobramycin+Anderson et al. (2008)
PA4867Urease β subunitTobramycinWhiteley et al. (2001)
PA5040-5044Pili biosynthesisImipenemBagge et al. (2004)
PA5170Arginine–ornithine antiporterImipenem,+Bagge et al. (2004)
 Azithromycin+Gillis et al. (2005)
PA5216Permease of ABC Fe transporterTobramycinAnderson et al. (2008)
 Azithromycin+Gillis et al. (2005)
PA5253-5255Alginate biosynthesisTobramycin+Anderson et al. (2008)
PA5261-5262Alginate biosynthesisImipenem+Bagge et al. (2004)
 Tobramycin+Anderson et al. (2008)
PA5348Probable DNA-binding proteinTobramycin+Whiteley et al. (2001)

While screening a library of approximately 4000 random P. aeruginosa PA14 transposon insertion mutants, Mah et al. (2003) identified a mutant that had decreased tobramycin susceptibility when grown in biofilms, but was otherwise indistinguishable from the wild-type strain (i.e. no differences in tobramycin susceptibility when grown planktonically). The mutation was mapped to PA1163 (ndvB), coding for a periplasmic glucosyltransferase required for the synthesis of cyclic-β-(1,3)-glucans. Through a series of elegant experiments, the authors were able to demonstrate that the cyclic glucans synthesized by ndvB can sequester various antibiotics (including tobramycin, gentamycin and ciprofloxacin) and as such interfere with the movement of the antibiotics through the periplasmic space. Semi-quantitative PCR confirmed that ndvB is preferentially expressed in sessile cells. In addition, further screening of this Tn5 insertion mutant bank resulted in the identification of a novel efflux pump (PA1874–PA1877) that was more highly expressed in biofilm cells than in planktonic cells and contributed to the increased resistance of sessile populations to tobramycin, gentamycin and ciprofloxacin (Zhang & Mah, 2008) (Table 2).

In P. aeruginosa biofilms treated with 1 μg mL−1 of the β-lactam antibiotic imipenem (a concentration below the MIC), 336 genes were induced or repressed at least twofold (Bagge et al., 2004). Not surprisingly, ampC (encoding a chromosomal β-lactamase) showed the strongest differential expression (150-fold on day 3). Several genes involved in alginate biosynthesis (including the algD to algA cluster and the algU-mucABC gene cluster) were also upregulated, while in younger biofilms treated with a subinhibitory concentration of imipenem, downregulation of motility-associated genes (flgC to flgI cluster, pilA, pilB, pilM to pilQ) was observed. The upregulation of alginate-related genes was associated with a drastic (up to 20-fold) increase in alginate production. Imipenem treatment also resulted in significant differences in biofilm structure, with treated biofilms containing more biomass per area and being thicker, but having a smoother surface, leading to a lower surface-to-volume ratio. The overexpression of ampC and genes involved in alginate biosynthesis probably allows the more efficient neutralization of imipenem: the AmpC β-lactamase is secreted in membrane vesicles and the accumulation of this enzyme in the matrix allows the rapid hydrolysis of β-lactams as they penetrate the matrix.

Exposure of P. aeruginosa PAO1 biofilms to sub-MIC levels of azithromycin (2 μg mL−1) for 4 days resulted in the differential expression (≥5-fold difference) of 274 genes compared with untreated control biofilms (Gillis et al., 2005). Several of the upregulated genes encode resistance-nodulation-cell division (RND) efflux pumps, including mexC (94.8 ×), oprJ (19.3 ×), nfxB (14.5 ×), mexD (12.7 ×) and oprN (6.7 ×). The expression of mexAB-oprM genes was not altered, but these genes are constitutively expressed at high levels already. By creating RND efflux pump mutants and transcriptional fusions, Gillis et al. (2005) showed that the mexAB-oprM and mexCD-oprJ RND efflux pumps are required for the formation of azithromycin-resistant P. aeruginosa biofilms. Also, the various efflux pumps showed different expression patterns: while mexA was expressed continuously throughout the biofilm regardless of the presence of azithromycin, mexC was expressed only in biofilms (but not in planktonic cells) in the presence of azithromycin and expression levels appeared to be the highest in the central parts of the biofilm [it should be noted that in an earlier study, the expression of mexAB-oprM and mexCD-oprJ was found to be the highest at the biofilm substratum, and not the center (de Kievit et al., 2001)].

Interestingly, genes PA0105, PA0106 and PA0108 (encoding cytochrome c oxidase subunits) were significantly downregulated in response to azithromycin treatment, suggesting that there may be a coupling between electron transport and susceptibility to macrolides as already observed for tobramycin (Whiteley et al., 2001) (Table 2).

When P. aeruginosa PA14 biofilms formed on cystic fibrosis-derived airway epithelial cells are treated with 500 μg mL−1 tobramycin (approximately half of the minimum bactericidal concentration under these conditions) for 30 min, 338 transcripts were upregulated and 500 were downregulated (Anderson et al., 2008). Tobramycin treatment reduced the virulence of the bacteria toward the epithelial cells and several virulence-related genes were downregulated. Conversely, several genes involved in alginate biosynthesis were upregulated (algU, mucA, algZ), but as core alg biosynthetic genes were not upregulated, it is uncertain whether this leads to increased alginate production. The transcript levels for most resistance-related genes were only slightly altered (PA1541, mexB, mexR) or remained unchanged, suggesting that the expression of other, yet unknown, factors is important for resistance under these conditions.

Comparing the data reported in the various studies revealed that very few differentially expressed genes are common between the different studies (Table 2). Analysis of the expression data reported by Whiteley et al. (2001) and Bagge et al. (2004) revealed that only PA2703 (encoding a hypothetical protein) and PA3819 (encoding a hypothetical membrane protein) are overexpressed as a result of both tobramycin and imipenem treatment (Table 2). The only two genes that were upregulated by imipenem (Bagge et al., 2004) and tobramycin (cystic fibrosis-derived airway epithelial cell model, Anderson et al., 2008) (PA5261 and PA5162) are both involved in alginate biosynthesis. Also, when a treatment with imipenem (Bagge et al., 2004) is compared with treatment with azithromycin (Gillis et al., 2005), two genes are found to be regulated in the same way: while PA4306 (encoding a type IVb pilin) is downregulated as a consequence of both treatments, PA5170 (encoding an arginine-ornithine antiporter) is upregulated (Table 2). When comparing these studies, it also becomes obvious that the expression of particular genes can be induced or repressed, depending on the antibiotic used (Table 2). PA2367 is downregulated by azithromycin and it is upregulated by imipenem. Similarly, PA3049 is downregulated by azithromycin and upregulated by tobramycin, while PA5216 is downregulated by tobramycin and upregulated by azithromycin (Table 2).

General stress response in E. coli biofilms

The studies by Schembri et al. (2003), Beloin et al. (2004), Ren et al. (2004), Domka et al. (2007) and Hancock & Klemm (2007) revealed that stress-related genes are often overexpressed in sessile E. coli populations compared with planktonic cultures, even in the absence of antibiotics (Wood, 2009). When comparing 40-h-old E. coli biofilms grown in a flow cell with exponentially growing planktonic cultures, Schembri et al. (2003) noted that 46% (30/65) of rpoS-controlled genes were differentially expressed during biofilm growth (most were upregulated) and an rpoS mutant turned out to be incapable of forming a biofilm in the flow system. In addition, yeaGH were also overexpressed; these genes are rpoS-regulated in Salmonella enterica and may also be associated with a stress response. Ito et al. (2008, 2009b) confirmed that rpoS-mediated stress responses contribute to biofilm-specific phenotypes (including ampicillin resistance). Also, in 8-day-old E. coli TG1 biofilms grown in a microfermentor, stress-related genes were upregulated, including SOS response genes, chaperones, general stress response genes, heat shock proteins and genes involved in DNA repair and envelope stress response (Beloin et al., 2004). This last group of genes includes cpxAR (sensor-regulator components of the cpx two-component system) and the phage shock protein operon (pspABCDE), although no biofilm-related phenotype was obvious in a psp operon mutant. In addition, a TG1 recA mutant was no longer capable of forming mature biofilms, confirming the importance of stress responses in biofilm formation. In E. coli biofilms grown on glass wool, stress genes are also induced, including hslS, hslT, hha, soxS and b1112 (Ren et al., 2004). hslST are involved in response to heat shock and superoxide stress, while soxS is involved in the response to superoxide. Gene b1112 (also known as ycfR or bhsA), encoding a putative outer membrane protein, plays an important role in stress response and biofilm formation as it mediates the stress response by a mechanism that involves increased synthesis of the signal molecule indole (Zhang et al., 2007; Wood, 2009). Cells in urine-grown biofilms formed by isolates recovered from asymptomatic bacteriuria cases also exhibit an overexpression of stress genes (Hancock & Klemm, 2007). Among the most upregulated genes are cold and heat shock proteins including cpsAGH and hslS, and soxS, yfiD and pphA. The temporal data from Domka et al. (2007) revealed that various cold and heat shock proteins (cspABFGI) were upregulated in young, but not in older biofilms. Whether, to what extent and how these general stress genes protect E. coli biofilms remains to be determined.

Adaptation of sessile P. aeruginosa and Burkholderia cenocepacia cells to oxidative stress

In several Gram-negative bacteria, coordinated regulation of many genes associated with oxidative stress is mediated by the transcriptional regulator OxyR (Ochsner et al., 2001; Zheng et al., 2001). In P. aeruginosa, oxidized OxyR increases the expression of ahpCF and katB (both encoding cytoplasmic enzymes) and of ahpB (encoding a periplasmic enzyme) (Ochsner et al., 2001). Panmanee & Hassett (2009) recently showed that these OxyR-controlled antioxidant enzymes play differential roles in planktonic and sessile P. aeruginosa cells. While exposure to H2O2 results in the upregulation of the katB gene in planktonic cells, no such upregulation is observed in sessile cells. In contrast, the treatment of planktonic cultures with H2O2 does not result in a differential expression of ahpC, while this gene is significantly upregulated in sessile cells treated with high (25 mM) H2O2 concentrations. A possible explanation for this is that, due to iron starvation, the catalase activity in biofilm cells is extremely low, making the increased expression of ahpCF a necessity for survival under these growth conditions (Panmanee & Hassett, 2009).

Burkholderia cenocepacia is a Gram-negative bacterium that is well known for causing respiratory infections in individuals with cystic fibrosis (Coenye & Vandamme, 2003; Mahenthiralingam et al., 2008). Most B. cenocepacia strains readily form biofilms on various surfaces, and sessile B. cenocepacia cells are highly resistant against antibiotics and disinfectants (Peeters et al., 2008, 2009). While studying the resistance of sessile B. cenocepacia cells against disinfection procedures implemented in various infection control guidelines, it was noticed that these sessile cells are highly resistant against H2O2 and NaOCl (Peeters et al., 2008). This observation not only has implications for infection control practices, but, as these oxidative agents are being produced by neutrophils as part of the endogenous defense against microorganisms (MacDonald & Speert, 2007), may also have implications for pathogenesis. When the transcriptional response of treated vs. untreated B. cenocepacia biofilms was compared, it was observed that the exposure to H2O2 and NaOCl resulted in an upregulation of 315 (4.4%) and 386 (5.4%) genes, respectively (Peeters et al., 2010). Transcription of 185 (2.6%) and 331 (4.6%) genes was decreased in response to H2O2 or NaOCl treatments, respectively. Not surprisingly, many of the upregulated genes in the treated biofilms are involved in (oxidative) stress responses, emphasizing the importance of the efficient neutralization and scavenging of reactive oxygen species. In addition, multiple upregulated genes encode proteins that are necessary to repair reactive oxygen species-induced cellular damage. Similar to what was observed for P. aeruginosa, ahpC and ahpF were highly upregulated, while katB was only modestly upregulated (upregulations of 41.3-, 15.5- and 1.8-fold, respectively, after 30 min of treatment with H2O2) (Peeters et al., 2010). However, biofilms formed by a B. cenocepacia katB mutant (which still contains a functional ahpCF) were nevertheless highly susceptible to H2O2, and there is already substantial expression of katB in untreated biofilms. This clearly indicates that, unlike in P. aeruginosa, this catalase is crucial for the protection of sessile cells against exogenous H2O2, although its expression is not increased following exposure to reactive oxygen species. Treatments with H2O2 or NaOCl also resulted in the increased transcription of several organic hydroperoxide resistance (ohr) genes, including BCAS0085. Interestingly, in addition to the upregulation of BCAS0085 (49.3-fold), a marked increase in the expression of BCAS0086 (encoding an exported lipase) was also observed (96.6-fold), probably due to the cotranscription of both genes. As a result of the marked overexpression of BCAS0086, an increased extracellular lipase activity was observed in treated biofilms. BCAS0085 and BCAS0086 orthologues in other Burkholderia genomes are organized in a similar operon-like manner, and increased lipase activity was also observed in the supernatant of H2O2-treated biofilms of B. cenocepacia C5424, HI2424 and AU1054, Burkholderia multivorans LMG 17588, Burkholderia ambifaria LMG 19182 and Burkholderia dolosa AU0158 (Peeters et al., 2010). It remains to be determined whether this increased lipase activity has a protective effect or is merely the consequence of the cotranscription of a lipase-encoding gene.

Adaptation of sessile C. albicans cells to exposure to antifungal agents

The molecular mechanisms of antifungal resistance in C. albicans have been studied extensively and changes in the expression of genes have been reported frequently in resistant clinical isolates (White, 1997; White et al., 1998; Sanglard, 2002). Azole antifungal drugs (including fluconazole, miconazole and itraconazole) target the P450 mono-oxygenase encoded by the ERG11 gene. This enzyme is involved in the conversion of lanosterol into ergosterol by mediating 14-α-demethylation, a key step in ergosterol biosynthesis (White et al., 1998). Resistance to fluconazole, the most commonly used antifungal agent, is associated with overexpression of ERG11, but changes in the expression of other ERG genes (including ERG3 and ERG25) have also been associated with azole resistance (Franz et al., 1998; Lopez-Ribot et al., 1998; Henry et al., 2000). In addition, in fluconazole-resistant isolates, genes encoding efflux pumps (including MDR1, CDR1 and CDR2) are often upregulated, resulting in increased efflux (Lopez-Ribot et al., 1998; White et al., 2002; Rogers & Barker, 2003). Polyene antifungal agents, including amphotericin B and nystatin, are fungicidal and bind to ergosterol in the fungal cell membrane, leading to membrane damage and oxidative stress (Brajtburg et al., 1990; Beggs, 1994). In vitro exposure of planktonic cells to amphotericin B often leads to a repression of ERG3 and ERG11 expression and a concomitant decrease in ergosterol levels in the membrane, indicating that changes in the sterol composition are important for amphotericin B resistance in C. albicans (Liu et al., 2005). Furthermore, changes in the expression of genes involved in β-1,6-glucan biosynthesis (including SKN1 and KRE1) have also been proposed as a resistance mechanism against polyene antifungals (Gale, 1986; Mio et al., 1997; Liu et al., 2005).

Antifungal resistance in C. albicans biofilms is a complex phenomenon, and like in planktonic cells, multiple mechanisms appear to be involved (Kuhn & Ghannoum, 2004). It was reported that efflux pumps are highly expressed in young biofilms (Ramage et al., 2002; Mukherjee et al., 2003; Mateus et al., 2004), even in the absence of an antifungal agent. However, the expression of genes encoding efflux pumps (CDR and MDR family) seems to be model system and/or strain dependent as CDR and MDR genes were not found to be overexpressed in the transcriptome studies of Garcia-Sanchez et al. (2004) and Murillo et al. (2005). Nevertheless, some genes (including QDR1 and CDR4) appeared to be overexpressed in the study by Yeater et al. (2007) and other genes (including CDR2 at 12 h and MDR1 at 12 and at 24 h) were overexpressed in the in vivo model described by Nett et al. (2009). Reduced ergosterol levels (combined with increased levels of other sterols) also provide a possible resistance mechanism in biofilms (Mukherjee et al., 2003) and changes in the expression levels of ERG genes were observed in several studies (Yeater et al., 2007; Nett et al., 2009). These changes probably lead to changes in the sterol composition of the cell membrane and may have a profound impact on antifungal resistance. Khot et al. (2006) and LaFleur et al. (2006) showed that resistant subpopulations (persisters) are present in C. albicans biofilms. Using untreated biofilms, Khot et al. (2006) compared the less-resistant, shear-removed, fraction of the biofilm with the basal blastospore subpopulation. In the latter, a marked downregulation of the ERG1 gene was observed, probably resulting in an overall downregulation of the ergosterol biosynthesis (remarkably, the expression of ERG11 was not altered). SKN1 and KRE1 were markedly upregulated in this resistant subpopulation. These changes in gene expression likely contributed to the observed amphotericin B resistance. When C. albicans biofilms in various stages of growth were treated with very high doses of fluconazole, an overexpression of genes involved in the ergosterol biosynthesis (ERG1, 3, 11 and 25) was observed, whereas after exposure to amphotericin B, an upregulation of SKN1 and KRE1 was observed. The transcriptional changes in sessile C. albicans cells in the presence of an antifungal agent likely result in an upregulation of the associated biosynthetic pathways, thereby contributing to a resistant biofilm phenotype. These data suggested that young and mature biofilms show a rapid and antifungal-specific transcriptional response to exposure to antifungal agents. This drug-specific molecular adaptation could help to explain the high resistance of C. albicans biofilms toward antifungal agents (Nailis et al., 2010).

Role of differential expression of phage-related genes

Overexpression of phage-related genes in sessile cells compared with planktonic cells and/or increased expression in response to stress has been observed in several species. The most highly overexpressed P. aeruginosa PAO1 genes in the study of Whiteley et al. (2001) were proteins from a Pf1-like bacteriophage (now designated Pf4; Webb et al., 2004), and this was confirmed by a 100–1000-fold greater abundance of phage particles in the biofilm reactor compared with planktonic cultures. In Bacillus subtilis, 17 genes involved in the production of the defective prophage PBSX are overexpressed in biofilms (Stanley et al., 2003). In B. cenocepacia biofilms, a prolonged treatment (30 or 60 min) with H2O2 resulted in an increased transcription of genes belonging to a BcepMu prophage (BCAS0540–BCAS0554), located on one of the B. cenocepacia genomic islands (genomic island 14) (Peeters et al., 2010). One of these genes (BCAS0547, encoding a putative DNA-binding phage protein) was also found to be upregulated during growth in cystic fibrosis sputum (Drevinek et al., 2008).

Bacterial stress responses can increase the mobility of bacteriophages (reviewed by Miller, 2001), and it has been proposed that prophage production may play a role in generating genetic diversity in the biofilm (e.g. the production of Pf4 in P. aeruginosa biofilms is correlated with the emergence of small-colony variants) (Webb et al., 2004). When faced with unstable environmental conditions, communities are protected by diversity, a principle known as the ‘insurance hypothesis’ (Boles et al., 2004); and the diversity generated by the induction of prophages may contribute to biofilm resilience.

An emerging picture

From the above examples, it is clear that sessile cells have various ways of coping with the stress imposed on them by treatment with antibiotics or disinfectants.

A first defense mechanism is the upregulation of genes encoding efflux pumps, resulting in an increased efflux of the antimicrobial agent. In some organisms, particular efflux pumps appear to be biofilm specific. The increased production of enzymes that can degrade antibiotics or reactive oxygen species is an important defense mechanism in various bacteria. While some of these enzymes appear to be equally important for protecting planktonic and sessile cells (e.g. katB in B. cenocepacia), some appear to be biofilm specific (e.g. ahpCF in P. aeruginosa). Phenotypic adaptations resulting in reduced transport of antimicrobial agents in biofilms and/or reduced permeability of the cell have also been reported. These include the increased production of the extracellular matrix (e.g. alginate) and/or other components that can sequester antibiotics (e.g. cyclic glucans) and the differential expression of genes affecting cellular uptake (e.g. tolA). Finally, altering the expression of genes coding for the target of the antimicrobial agent (e.g. ERG genes in C. albicans) and/or activating alternative pathways can also result in decreased susceptibility.

Interestingly, in various organisms, the expression of genes thought to be involved in stress resistance is altered in sessile cells compared with planktonic cells, even in the absence of the stress, leading to the ‘innate resistance’ of sessile cells. Examples include the upregulation of several genes coding for efflux pumps in C. albicans, the upregulation of tolA in P. aeruginosa, the downregulation of cytochrome c oxidase genes in P. aeruginosa and the upregulation of heat shock proteins in E. coli. Generating diversity by the induction of prophages may also contribute to the intrinsic resistance of biofilm populations.

Problems associated with reliably identifying gene expression patterns in biofilms and implications for identifying the response to stress

  1. Top of page
  2. Abstract
  3. Introduction
  4. The search for biofilm-specific gene expression patterns: C. albicans as an example
  5. Stress-induced gene expression in biofilms and phenotypic adaptation
  6. Problems associated with reliably identifying gene expression patterns in biofilms and implications for identifying the response to stress
  7. Outlook
  8. Acknowledgements
  9. References

Heterogeneity in microbial biofilms

It is a common misconception that all cells in a biofilm are exposed to the same conditions. In contrast, differences in metabolic activities combined with differences in the transport of molecules in a biofilm result in gradients of nutrients, oxygen, signaling molecules and metabolic end products. As a result of these gradients, considerable structural, chemical and biological heterogeneity can be found within a biofilm (Stewart & Franklin, 2008). For example, tomographic fluorescence imaging using silica nanoparticle sensors showed that within an E. coli biofilm, pH values can vary from 5 to >7, due to the low rates of diffusion of acidic metabolites or accumulation of fermentation products in oxygen-limited parts of the biofilm (Hidalgo et al., 2009). As a consequence of this diversity, harvesting entire biofilm populations will only allow the identification of genes as being differentially expressed if these genes are uniquely expressed in biofilms and will result in an ‘average’ picture of gene expression (Stewart & Franklin, 2008). Unfortunately, few alternatives are at our disposal. Reporter genes fused to promoter regions of a gene of interest can be used to microscopically monitor the expression of that gene in a biofilm (Stewart & Franklin, 2008). A recent example of such a study is that of Ito et al. (2009a), who used an rpoS-gfp transcriptional fusion mutant to monitor rpoS expression in E. coli biofilms. Their results confirmed the existence of localized expression profiles, with rpoS being expressed in the majority of cells in the early phases of biofilm formation, while in the later stages of biofilm formation, rpoS expression appeared to be limited to cells at the outside of the biofilm. Although useful, this approach requires the use of genetically manipulated microorganisms and is at present not suitable for the simultaneous analysis of a large number of genes. Lenz et al. (2008) described the use of laser capture microdissection microscopy to recover cells from spatially resolved sites within biofilms. RNA can then be isolated from these cells, allowing the study of gene expression by real-time quantitative PCR. Their proof-of-concept study confirmed that this approach is feasible and demonstrated that mRNA levels for particular genes are not uniform throughout the biofilm. The issue of heterogeneity is particularly relevant for C. albicans, which has multiple morphological forms (yeast, hyphae, pseudohyphae) (Calderone & Fonzi, 2001). The fraction of filaments in a biofilm is highly dependent on the biofilm model system and the stage of biofilm formation (Nailis et al., 2009) and as a number of genes are considered to be hyphae specific (or at least hyphae associated), including ALS3 and HWP1 (Hoyer et al., 1998; Sundstrom, 2002), interpretation of the differential expression of genes under conditions that affect filamentation should take this into account.

It should be pointed out that in planktonic cultures, there can also be considerable heterogeneity. Laser-diffraction particle-size scanning and microscopy of ‘planktonic’ cultures of P. aeruginosa indicated that up to 90% of the entire culture was present in aggregates of 10–400 μm, rather than as individual cells, and these planktonic cultures are actually more similar to ‘suspended biofilms’ (Schleheck et al., 2009). How this growth phenotype influences gene expression is at present unclear, but this observation illustrates that a careful validation of both model systems (biofilm and planktonic) before comparing gene expression is warranted.

Role of small RNAs (sRNA) in microbial group behavior

sRNA-mediated post-transcriptional control at the mRNA or the protein level plays a pivotal role in mediating bacterial adaptation to changing conditions (Papenfort & Vogel, 2009; Waters & Storz, 2009). The regulation exerted by sRNAs is often negative, as protein levels are repressed through translational inhibition, mRNA degradation or both. Most require the RNA chaperone Hfq to facilitate RNA–RNA interactions and to stabilize unpaired sRNAs. A given sRNA can regulate multiple targets and this means that a single sRNA can globally modulate a particular physiological response in much the same manner as a conventional transcription factor, but at the post-transcriptional level (Papenfort & Vogel, 2009; Vogel, 2009; Waters & Storz, 2009). Modeling studies have clearly indicated that, when a fast response to external signals is required (like in the case of a stress response), sRNA-based regulation is advantageous over protein-based regulation. sRNAs are also better than transcription factors in filtering out the noise in input signals. Taken together, the data from modeling studies suggest that there is a particular ‘niche’ for sRNAs in allowing the quick and reliable transition between distinct states (Levine et al., 2007; Shimoni et al., 2007; Mehta et al., 2008). Conventional transcriptomic analyses rely on microarray-based identification of gene expression and are inherently biased as only expression levels of genes for which probes are included on the array can be measured. As our knowledge of the occurrence of sRNAs in various organisms is still limited, the number of probes directed against intergenic regions (containing sRNAs) is often small, precluding the identification of transcripts arising from intergenic regions. In addition, reverse transcription of sRNAs is often suboptimal (due to their small size and pronounced secondary structure) and probe labeling can also be hampered by the intrinsic structure of the sRNA (Hüttenhoffer & Vogel, 2006; Sharma & Vogel, 2009). Nevertheless, a limited number of studies have focused on the potential role of sRNAs in biofilm formation and phenotypic adaptation to stress. One of the bacterial regulatory systems involving sRNA is the carbon storage regulator (Csr) system (Romeo, 1998). CsrA is a sRNA-binding protein that represses the expression of many stationary-phase genes, while inducing the expression of exponential-phase pathways (including glycogen synthesis and catabolism, glycolysis and gluconeogenesis). The second component of the Csr system is the sRNA CsrB. CsrB can bind 18 CsrA molecules simultaneously and as such antagonizes the effect of CsrA (Romeo, 1998). Jackson et al. (2002b) showed that in E. coli, biofilm formation is increased in a csrA mutant and that there is no biofilm formation in a csrB mutant. CsrB and CsrC sRNAs modulate protein activity by mimicking mRNA and sequester away the CsrA protein from mRNA leaders. Moreover, induction of csrA expression induces biofilm dispersal. Additional studies have shown that the role of CsrA is consistent under diverse growth conditions and in a variety of enterobacterial strains and species (Jackson et al., 2002a; Agladze et al., 2003). The link between the csrA/B system and biofilm formation was found to be the cell-bound polysaccharide adhesin poly-β-1,6-N-acetyl-glucosamine (PGA) (Wang et al., 2005), as CsrA post-transcriptionally represses the gene required for PGA production, while there is also an indirect repression through the inhibition of glgCAP expression (necessary for the stationary-phase carbon flux into glycogen and subsequent conversion to glucose-1-phosphate required to generate a PGA precursor). In addition, the expression of luxS in E. coli (encoding the key enzyme in the biosynthesis of the autoinducer-2 quorum-sensing molecule) is negatively regulated by the sRNA CyaR (De Lay & Gottesman, 2009). This downregulation results in a decreased AI-2 production; under glucose-limited conditions, this system probably decreases biofilm formation while increasing planktonic behavior and as such may trigger the organisms to move in search of nutrients. Also, in P. aeruginosa, social behavior is coregulated by sRNA molecules (Heurlier et al., 2004; Kay et al., 2006; Lapouge et al., 2008; Lucchetti-Miganeh et al., 2008). RsmA (an sRNA-binding regulatory protein) negatively controls the production of N-acyl homoserine lactone-signaling molecules, but this RsmA-based repression is antagonized by the GacA-dependent sRNAs RsmY and RsmZ. In addition, an rsmY rsmZ double mutant shows enhanced biofilm formation compared with the wild type, suggesting that both genes jointly influence biofilm formation.

Recently, a significant upregulation of the transcriptional activity stemming from intergenic regions was noted when B. cenocepacia J2315 biofilms were treated with oxidizing agents (Peeters et al., 2010). Treatment with H2O2 or NaOCl resulted in the upregulation of 37 and 56 intergenic regions, respectively, compared with untreated biofilms. Several of these intergenic regions were located in the close proximity of genes with a similar expression pattern, suggesting cotranscription. However, other intergenic regions demonstrated markedly different expression patterns compared with their flanking genes and the basal expression levels of several of these regions were high. Several of these putative sRNAs were previously predicted using an in silico approach (Coenye et al., 2007), while others were found to be differentially expressed in B. cenocepacia grown in sputum (Drevinek et al., 2008) or under soil-like conditions (Yoder-Himes et al., 2009). While the function of most of these putative sRNAs remained elusive, one had a marked similarity to the 6S RNA gene consensus structure, indicating its potential involvement in regulating gene expression.

Outlook

  1. Top of page
  2. Abstract
  3. Introduction
  4. The search for biofilm-specific gene expression patterns: C. albicans as an example
  5. Stress-induced gene expression in biofilms and phenotypic adaptation
  6. Problems associated with reliably identifying gene expression patterns in biofilms and implications for identifying the response to stress
  7. Outlook
  8. Acknowledgements
  9. References

Traditionally, microarrays are used to identify changes in gene expression in high-throughput analyses, but there are several drawbacks associated with their use. Probably the most relevant drawback is that this approach is inherently biased (i.e. you can only measure what is known and hence represented on the array). This can be circumvented using high-throughput parallel sequencing (RNA sequencing). This novel, unbiased, approach will not only reveal changes in the expression level of protein-coding genes, but will also lead to the discovery of changes in sRNA expression. Several sequencing technologies are currently available, including pyrosequencing (454 sequencing) and Illumina ‘sequencing-by-synthesis’ (Mardis, 2008; Shendure & Hanlee, 2008; Petterson et al., 2009). These techniques present a vast improvement over microarray-based transcriptome analysis, but still rely on the generation of cDNA before sequencing, which may be the source of various types of errors. Ozsolak et al. (2009) recently described an entirely novel approach called ‘direct RNA sequencing’. Direct RNA sequencing is based on Helicos BioSciences' ‘True Single Molecule Sequencing’ technology and allows the sequencing of femtomole quantities of RNA without the need for prior cDNA generation. This approach would allow the unbiased whole-transcriptome analysis of a low number of cells and would provide a snapshot of the response in various parts of the biofilms.

Despite advances in transcriptomics, the challenge for the future will remain the same, i.e. to link the changes in gene expression to phenotypic changes and (1) to determine whether differential gene expression really results in an observable altered phenotype and (2) to determine whether this differential gene expression and the resulting phenotype are attributable to the stress conditions applied.

Acknowledgements

  1. Top of page
  2. Abstract
  3. Introduction
  4. The search for biofilm-specific gene expression patterns: C. albicans as an example
  5. Stress-induced gene expression in biofilms and phenotypic adaptation
  6. Problems associated with reliably identifying gene expression patterns in biofilms and implications for identifying the response to stress
  7. Outlook
  8. Acknowledgements
  9. References

I wish to thank BOF-UGent, the Fund for Scientific Research-Flanders and Cystic Fibrosis Foundation Therapeutics Inc. for financial support. I also wish to thank colleagues and coworkers (past and present) for their collaboration and support. I apologize to the colleagues whose work I was not able to cite due to space constraints.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. The search for biofilm-specific gene expression patterns: C. albicans as an example
  5. Stress-induced gene expression in biofilms and phenotypic adaptation
  6. Problems associated with reliably identifying gene expression patterns in biofilms and implications for identifying the response to stress
  7. Outlook
  8. Acknowledgements
  9. References