At the advent of an era where science and technology is making long-term spaceflight missions to the Moon and Mars achievable, increased health risks are arising that are associated with the sustained stay of astronauts in isolated, confined environments. One of the critical factors to ensure safety, health and performance of the crewmembers is anticipating the risk for infectious disease during space exploration and habitation.
The ubiquitous Gram-negative bacterium Pseudomonas aeruginosa has previously caused respiratory and urinary tract infections in crewmembers of Apollo missions during and/or immediately after spaceflight (Taylor, 1974; Hawkins and Ziegelschmid, 1975). Pseudomonas aeruginosa is typically found in water and soil, and is occasionally part of the healthy human flora. As a result, Pseudomonas sp. have been isolated from surfaces and the potable water system of the International Space Station (ISS), despite stringent disinfection regimes (Bruce et al., 2005; Novikova et al., 2006). Besides being an environmental organism, P. aeruginosa is an important opportunistic pathogen in patients with both primary and acquired immunodeficiencies, accounting for 11–14% of nosocomial infections (Driscoll et al., 2007). Importantly, spaceflight conditions are known to compromise the immune system in multiple fashions (reviewed by Sonnenfeld, 2005). Earth-based models, such as hind limb unloading of rodents and bed-rest studies of human, simulate aspects of the altered immune response in space (reviewed by Sonnenfeld, 2005). Mice exposed tohind limb unloading showed a compromised resistance to infection by the Gram-negative pathogens Klebsiella pneumoniae and P. aeruginosa (Belay et al., 2002; Aviles et al., 2003). Furthermore, previous studies have demonstrated an increased virulence and stress resistance of bacterial pathogens in response to both spaceflight conditions and low shear modelled microgravity (LSMMG), compared with the ground controls (Nickerson et al., 2000; Wilson et al., 2002; 2007; 2008; Lynch et al., 2004).
As opportunities for bacterial experiments aboard the ISS are often limited due to constraints in crew time and safety reasons, specialized lab scale bioreactors have been developed to study certain aspects of microgravity on cellular reactions on Earth. The rotating wall vessel (RWV) bioreactor and the random positioning machine (RPM) are the most commonly used devices for this purpose. In the RWV, cells continuously fall at terminal velocity in a low fluid-shear suspension environment, which is inherent to microgravity (reviewed by Nickerson et al., 2004). In the RPM, cells are subjected to random orientations and accelerations, which is thought to result in the random alteration and thus theoretical reduction of the gravity vector (van Loon, 2007). Although the RPM has rarely been used for bacterial experiments (de Vet and Rutgers, 2007; Mastroleo et al., 2009), many reports exist on its use to study microgravity-analogue-induced changes in liquid cultures of suspended eukaryotic cells, mostly on leucocytes (Schwarzenberg et al., 2000; Boonyaratanakornkit et al., 2005).
The present study aimed at assessing the behaviour of P. aeruginosa PAO1 in response to microgravity-analogue conditions, emphasizing on pathogenicity mechanisms and stress resistance. Therefore, global transcriptional analysis and complementary phenotypic and physiologic assays were performed on P. aeruginosa grown in the RWV and RPM. To the best of our knowledge, this study presents the first global response of P. aeruginosa to spaceflight-analogue conditions, and the first direct comparison for bacteria between the most commonly used ground-based microgravity simulators.
Exploring the response of opportunistic pathogens to spaceflight conditions is not solely important in the frame of preventing infectious disease in astronauts but also allows for the study of bacteria in physiologically relevant environments which cannot be achieved with conventional technologies. Low fluid-shear conditions, which are inherently linked to microgravity, are encountered by bacteria during their natural course of infection and culturing under such conditions can be a potential tool to unravel novel pathways involved in the infection process (Nickerson et al., 2004; Wilson et al., 2007; 2008; Crabbéet al., 2008). During this study, the transcriptional and physiological response of P. aeruginosa PAO1 to microgravity-analogue conditions (LSMMG and RG) was determined, employing two commonly applied techniques, the RWV and the RPM. While 134 genes were significantly more transcribed in the LSMMG culture conditions of the RWV, only nine genes showed a differential expression in the RPM compared with the control condition. These distinct results prompted us to scrutinize the role of mixing, because a low fluid shear and mixing environment is required to study the indirect effects of microgravity on Earth. The indirect effects of microgravity are induced by microgravity-mediated changes in the extracellular environment (e.g. mechanical forces of the surrounding liquid, nutrients and wastes surrounding the cell) while direct effects refer to the sensing of and responding to changes in g-force as a direct effect of weight (on the cell and cell components) (Albrecht-Buehler, 1991; Klaus et al., 2004). Theoretical assessment of single cell gravitational sensing predicted that forces exerted by cell weight may be too small to be sensed and that indirect effects prevail (Albrecht-Buehler, 1991). Others, however, claim that gravity can be sensed directly at the single cell level (Goldermann and Hanke, 2001). In the RWV, cells are grown in a low fluid-shear environment, which is an indirect effect of microgravity. Low fluid-shear levels in LSMMG have been demonstrated through mathematical calculation (Nauman et al., 2007) and were confirmed with the crystal violet dispersal assay in the present study. The mixing level in NG and RG conditions was found to be higher in comparison with the LSMMG condition in the present study. Correspondingly, eukaryotic monolayer cultures exposed to RG conditions experienced fluid-shear levels up to 44 times the shear in LSMMG (Pardo et al., 2005). Consequently, both low fluid mixing and shear, which are characteristic during spaceflight, may not be accurately represented in the RPM. Based on the conclusions made above, the RWV is believed to be more appropriate to study the LSMMG-induced response of cells in a suspension environment, rather than the RPM. It is, however, important to mention that, besides the low fluid-shear environment encountered in low-Earth orbit, other environmental factors (such as irradiation and vibration) may contribute to the bacterial response to spaceflight conditions. Indeed, the RWV technology allows to reproduce only certain aspects of the microgravity environment, which could at least partly explain differences between LSMMG- and spaceflight-induced bacterial responses. On the other hand, similarities between spaceflight and LSMMG responses can presumably be ascribed to the analogous low fluid-shear conditions in-flight and in the RWV.
This study indicated that the alternative sigma factor AlgU (also known as AlgT or sigma 22 in P. aeruginosa, analogue of RpoE or sigma 24 in other bacteria) plays a regulatory role in the response of P. aeruginosa to LSMMG. The activity of AlgU is regulated by proteins encoded by the algUmucABCD operon (reviewed by Ramsey and Wozniak, 2005). In non-mucoid strains of P. aeruginosa or in the absence of specific environmental stimuli, the cytoplasmic domain of the anti-sigma factor MucA binds AlgU, resulting in a low expression of promoters targeted by AlgU. The periplasmic domain of MucA is associated with MucB, which is a negative regulator of alginate biosynthesis (Wood and Ohman, 2009). The role of MucC remains to be determined. The higher transcription of mucA and mucB in LSMMG can be explained by autoregulation of the algUmucABCD operon by AlgU as a negative feedback mechanism (DeVries and Ohman, 1994; Wood and Ohman, 2009). Alginate production, transcription of heat shock genes through the sigma factor rpoH and other genes under control of AlgU (e.g. dsbA, oprF, phaF, foaA, algR) were increased in LSMMG. Among the LSMMG-induced AlgU-controlled genes, oprF, algR, rpoH and PA0856 have known AlgU-dependent promoters (Firoved et al., 2002). Bacterial alginates are linear exopolysaccharides which consist of O-acetylated β-1,4-linked β-d-mannuronic acid and its C5 epimer α-l-guluronic acid (Evans and Linker, 1973). The only bacterial genera known to produce alginates are Pseudomonas (Evans and Linker, 1973) and Azotobacter (Pindar and Bucke, 1975). Alginate restricts the diffusion of anti-microbial agents and confers resistance to human immune defence mechanisms, by avoiding phagocytic uptake, scavenging reactive oxygen intermediates and suppressing leucocyte function (Baltimore and Mitchell, 1980; Learn et al., 1987; Eftekhar and Speert, 1988; Simpson et al., 1989). AlgU is required for the production of alginate by acting directly on the transcription of the alginate biosynthesis operon, or indirectly by activating the transcriptional regulator AlgR (Martin et al., 1994; Wozniak and Ohman, 1994; Ramsey and Wozniak, 2005). Alginate regulatory gene transcripts (algU, algR, algP) and, correspondingly, alginate itself were shown more abundant in LSMMG. Furthermore, expression of the negative regulator of PHA synthesis PhaF (Hoffmann and Rehm, 2004) was induced. Because alginate and PHA biosynthetic enzymes compete for the common precursor acetyl-CoA (Pham et al., 2004), this finding indicates that acetyl-CoA might be directed towards the production of alginate in LSMMG. Besides the regulatory role of AlgR in alginate biosynthesis, this transcriptional regulator activates and represses the transcription of many other genes (Lizewski et al., 2004). The transcription of AlgR-dependent genes ipbA, lon and hslV was induced in LSMMG. In addition, genes under negative control of AlgR also showed an upregulation in LSMMG (Lizewski et al., 2004). A majority of these genes, however, also belong to the Hfq and/or microaerophilic/anaerobic regulon, which is activated in LSMMG (see below). These findings indicate a very complex, regulatory network in which regulation might occur at transcriptional, post-transcriptional and post-translational level.
Specific environmental stimuli, e.g. heat shock, cell wall inhibitory antibiotics, nutrient limitation, osmotic and oxidative stress, and copper sulfate, are known to induce expression of AlgU in Pseudomonas sp. (Terry et al., 1992; Schurr et al., 1995; Keith and Bender, 1999; Wood and Ohman, 2009). However, the receptor(s) and signal transduction pathway(s) that activate AlgU and its regulon remain to be identified for the LSMMG-associated stimuli. It is interesting to note that several genes induced by the cell wall-disrupting antibiotic d-cycloserine (Wood and Ohman, 2009) were also upregulated by the LSMMG condition. Adopting a 1.5-fold threshold in the LSMMG gene set, 50 genes overlapped between LSMMG and d-cycloserine upregulated genes (P < 0.05, combining the 15 and 60 min exposure transcriptomes), e.g. phaF, foaA, PA2562, PA2486, mucABC. A majority of the latter gene set is dependent or under control of AlgU (44 out of 50) (Wood and Ohman, 2009). Consequently, AlgU might be an important regulator following the sensing of mechanical or chemical cell wall signals.
Another interesting fact is that alginate overproduction (AlgU dependent) of P. aeruginosa has been associated with phosphate limitation (Terry et al., 1992). Interestingly, phosphate was proposed to play a key role in the LSMMG-induced stress resistance of Salmonella enterica (Wilson et al., 2008), as supplementation of LB medium with phosphate nullified the differences in stress resistance between LSMMG and NG conditions. Hypothetically, decreased phosphate availability in LSMMG, due to low homogenization of culture medium, could have activated AlgU and AlgU-controlled pathways. However, other than katA, which is induced upon phosphate limitation (Yuan et al., 2005), genes of the Pho regulon were not differentially regulated at the transcriptional level in the test conditions.
Interestingly, the AlgU regulon for the plant pathogen X. fastidiosa revealed that the conserved RNA binding protein Hfq was positively regulated by AlgU (Shi et al., 2007). The AlgU and Hfq protein sequences of P. aeruginosa PAO1 and X. fastidiosa share 56% and 75% identity respectively. The major physiological functions of Hfq encompass stability control of small regulatory RNAs (sRNAs) and mRNAs as well as positive and negative translational regulation of target mRNAs by sRNAs (reviewed by Valentin-hansen et al., 2004; Majdalani et al., 2005). Also in S. enterica, RpoE has been suggested to positively regulate Hfq transcription, indirectly via the transcriptional activation of RpoH (Bang et al., 2005). On the other hand, Hfq was shown to exert a negative post-transcriptional effect on RpoE in S. enterica, Vibrio cholerae and Escherichia coli (Ding et al., 2004; Figueroa-bossi et al., 2006; Guisbert et al., 2007). Indeed, rpoE itself and RpoE-controlled gene transcription, among which genes encoding heat shock proteins, was induced in hfq mutants of V. cholerae and E. coli. As AlgU and the characteristic AlgU-controlled pathways (heat shock, alginate biosynthesis) were reported not to be affected in an hfq mutant of P. aeruginosa (Schreiber et al., 2006), the stability of algU-mRNA is presumably not under control of Hfq in this bacterium. Hfq has been shown to be involved in the LSMMG and spaceflight microgravity response of S. enterica (Nickerson et al., 2000; Wilson et al., 2007; 2008) and was upregulated when P. aeruginosa PAO1 was exposed to LSMMG in the present study. While for S. enterica, downregulation of Hfq in both LSMMG and spaceflight microgravity led to differential expression of genes under both positive and negative control of Hfq, the upregulation of Hfq for P. aeruginosa in LSMMG suggested mainly a positive regulatory role. Hfq has been shown to be important for virulence and stress resistance of several (opportunistic) pathogens, including P. aeruginosa PAO1, partly through its action on the sigma factor rpoS and the QS system (Sonnleitner et al., 2003; Ding et al., 2004; Sittka et al., 2007). The QS system of P. aeruginosa has been extensively reviewed (Diggle et al., 2006; Venturi, 2006). Hfq exerts a positive effect on the Rhl-QS system of P. aeruginosa PAO1, through stabilization of the regulatory RNA rsmY, which in turn inactivates the negative post-transcriptional Rhl-QS regulator RsmA (Sonnleitner et al., 2006). As none of the major QS-regulated genes were differentially expressed above the twofold threshold in the test conditions, the positive effect of Hfq on the Rhl-QS system could have been counterbalanced by the upregulation of the negative QS regulators rsmA and the lon-protease in LSMMG (Pessi et al., 2001; Takaya et al., 2008). In addition, we observed that a negative regulator of the las QS system, i.e. rsaL, was induced in LSMMG compared with NG (Rampioni et al., 2006; 2007). Interestingly, Hfq is a positive regulator of genes involved in the microaerophilic/anaerobic metabolism of P. aeruginosa (anr, hemN, ccoO2, ccoN2), also upregulated in LSMMG. In addition, 16 other genes belonging to the LSMMG-induced Hfq regulon are described as being upregulated in microaerophilic/anaerobic conditions (Table 1). Consequently, it is worthwhile to further explore the role of Hfq in the response of P. aeruginosa PAO1 to oxygen-limiting conditions. Collectively, these findings indicate that hfq is an important regulator of the LSMMG response across the bacterial species border.
A significant number of genes involved in microaerophilic/anaerobic metabolism and in the fermentation of arginine and pyruvate were upregulated in LSMMG conditions. Among others, the anaerobic regulator Anr and genes under control of Anr were induced in LSMMG. Anr induces the transcription of the cell surface chemoreceptor Aer (induced in LSMMG), which detects and responds to the chemotactic ligand oxygen, by redirecting the flagellar movement towards or away from oxygen through communication via chemotactic proteins (reviewed by Kato et al., 2008). Correspondingly, several genes involved in the P. aeruginosa chemotaxis and motility were upregulated in LSMMG compared with NG. Collectively, the above data suggest that different metabolic pathways were activated in LSMMG to ensure growth and survival in this environment with apparent lower oxygen availability. Previous studies have demonstrated that the oxygen level in a shake culture of P. aeruginosa rapidly decreases, leading to an apparent microaerophilic/anaerobic growth environment after 3–4 h of cultivation (Sabra et al., 2002). The DO levels in both LSMMG and NG cultures were also below the detection limit of the oxygen probe after 24 h of cultivation. Interestingly, the oxygen transfer rate, as reflected by the volumetric mass transfer coefficient kLa, was higher in NG than LSMMG cultures. While high amounts of alginate reduced the oxygen transfer from air into LB [in agreement with the study of (Hassett, 1996), the low amounts of alginate quantified in LSMMG and NG could not account for the significant difference in kLa-values. Herewith, the presence of a yet unidentified factor responsible for decreasing the oxygen transfer rate is proposed. This factor is presumably present in the cell fraction and not in the spent culture supernatant. Kim and colleagues (2003) linked a decrease in the oxygen transfer rate in P. aeruginosa cultures to a depletion of iron into the culture medium (Kim et al., 2003). In the present study, no evidence arose that would indicate iron depletion in LSMMG because none of the genes involved in the transcriptional response of P. aeruginosa to iron-limiting conditions (Ochsner et al., 2002) were differentially expressed in LSMMG and NG. Importantly, hemN and anr mutants demonstrated a decreased virulence in the lettuce model of infection (Filiatrault et al., 2006), indicating that growth of P. aeruginosa PAO1 in a microaerophilic or anaerobic environment could be of importance in pathogenicity.
Besides alginate, which can confer a certain degree of stress resistance to P. aeruginosa, several gene transcripts involved in environmental stress tolerance, i.e. heat shock proteins, catalase and universal stress proteins, were significantly more abundant in LSMMG. Importantly, AlgU is involved in the resistance of P. aeruginosa to heat shock (through rpoH regulation) and oxidative stress (Martin et al., 1994; Schurr and Deretic, 1997; Malhotra et al., 2000; Firoved et al., 2002). In accordance with the gene expression data, P. aeruginosa grown in LSMMG showed an increased resistance to heat shock and oxidative stress, while no significant differences in the acid stress tolerance were observed. As alginate is thought to scavenge oxygen radicals (Learn et al., 1987), the higher alginate concentration in LSMMG might have acted as a long-lasting barrier protecting the cells from H2O2. In addition, the induced transcription of katA (encoding a catalase) could have accounted for a higher turnover of H2O2 to H2O and O2 in LSMMG. The increased oxidative stress resistance observed over time might reflect enzymatic degradation of H2O2. Conversely, the increased heat shock tolerance in LSMMG could only be observed after a 15 min exposure time. Indeed, the cellular damage after 60 min exposure to high temperature might have been too important to compensate with the increased chaperone concentration in LSMMG. Correspondingly with the results of this study, Wilson and colleagues (2002) have shown an increased heat shock resistance for S. enterica grown in LSMMG compared with NG. In addition, increased osmotic and acid stress tolerance was demonstrated for S. enterica (Wilson et al., 2002). In contrast, E. coli MG1655 did not show any differences in its stress resistance profile when cultured in LSMMG and NG (Tucker et al., 2007), indicating that the responses between bacterial species might differ significantly.
Overall, the LSMMG-induced response of P. aeruginosa PAO1 shows similarities with the molecular profile of other bacteria studied in spaceflight and LSMMG conditions. More specifically, genes encoding chaperones, citric acid cycle enzymes, ATP synthases, cytochromes and ribosomal subunits have been reported to be upregulated in LSMMG and in space (Wilson et al., 2007; 2008; Leys et al., 2009; Mastroleo et al., 2009). Other LSMMG-induced genes and pathways were unique for each studied bacterium. As differences in low fluid-shear levels are multifactorial and are presumably associated with alterations in mechanical forces, availability of nutrients, oxygen concentrations, etc., the response to each of these shear-related factors may differ or be similar for different bacterial species. Besides genomic differences between bacteria, their preferred niche and how an environmental signal relates to this niche may play a role in these observations.
In conclusion, this study showed for the first time that cultivation of P. aeruginosa PAO1 in the LSMMG environment of the RWV induces a transcriptomic and physiologic response with possible implications for virulence. AlgU was identified as one of the key regulators in the LSMMG-induced response and this alternative sigma factor presumably upregulated the transcription of genes downstream involved in environmental stress resistance, alginate production and microaerophilic/anaerobic metabolism. Moreover, AlgU might control hfq transcription in P. aeruginosa PAO1, which has previously been linked with bacterial response to space and LSMMG conditions. The identification of the sensor which activates the AlgU cascade in LSMMG remains to be determined and is an important element to further unravel the virulence mechanisms of P. aeruginosa on Earth and, potentially, in space. The observations during the present study in combination with a decreased immune responsiveness of astronauts could become problematic during long-term spaceflight missions and should direct future research regarding the risk assessment and prevention of infections caused by this opportunistic pathogen.
The RWV and RPM were used in the present study to grow P. aeruginosa PAO1 in microgravity-analogue conditions. The RWV is a widely used methodology to study bacterial behaviour in conditions of LSMMG. These conditions are achieved through continuous rotation around a horizontal axis of cylindrical bioreactors that are completely filled with growth medium (Nickerson et al., 2004). Counteracting sedimentation of bacterial cells at adequate rotation speed in the RWV leads to maintained cell suspension within a restricted orbit. The fluid-shear force in a RWV bioreactor is calculated to be less than 0.001 Pa (Gao et al., 1997; Nauman et al., 2007). A RWV bioreactor rotating along the vertical axis is conventionally used as a NG control. The RPM mimics aspects of microgravity in 3-D through rotation around two orthogonal axes. The system is comprised of an inner frame (containing the attached sample) and an outer frame, which is independently rotating but linked to the inner frame. The rotation of the inner and outer frame occurs within a restricted 3-D volume at a random orientation and time interval using software developed by the Dutch Space Agency. A theoretical reduction of the g-vector is aimed to be obtained over time, resulting in an average g-vector of 10−2–10−3 (reviewed by van Loon, 2007). The RPM does, however, not remove the constant unidirectional force that gravity exerts on the cells.
Bacterial strain, culture medium and conditions
The wild-type P. aeruginosa PAO1 strain (ATCC 15692) was used in this study and all cultures were grown in Lennox L Broth Base (LB) (Life Technologies) at 28°C. An overnight shaking culture (125 r.p.m.) of P. aeruginosa in LB was washed and diluted in 0.85% NaCl solution to an OD600 of 1. This bacterial suspension was used to inoculate fresh LB medium at a final concentration of 10−4 cfu ml−1. Synthecon RWV bioreactors (50 or 10 ml) were filled with inoculated medium so that no headspace (i.e. no bubbles) was present. Other than for stress-resistance assays, for which 10 ml capacity bioreactors were used, RWV bioreactors with a capacity of 50 ml were adopted for all experiments. Identical bioreactors were mounted in triplicate on (i) a RWV device in vertical position (LSMMG) (Cellon); (ii) a RWV device in horizontal position (NG) and (iii) the centre of the inner RPM frame (RG) (Fokker Space), and placed in a large humidified (70–80% relative humidity) culture chamber, to avoid evaporation of culture medium through the gas-permeable membrane at the back of each vessel. A 25 r.p.m. rotation speed was adopted for the RWV cultures, while RPM cultures were randomly rotated at 10 r.p.m. (60° s−1), according to the instructions of the manufacturer. Bacteria were grown in the three described test conditions for 24 h. After 24 h of cultivation, the content of every bioreactor was gently mixed by pipetting and divided into several aliquots. Ten millilitres of culture from each growth condition was immediately fixed with RNA Protect Reagent (Qiagen), following the manufacturer's instructions, and fixed cell pellets were frozen at –20°C until RNA extraction. Samples were immediately exposed to different stresses or stored at –20°C for alginate quantification and oxygen transfer rate determination (see below). The bacterial density was assessed by viable count on LB agar (Invitrogen).
RNA extraction, labelling and Affymetrix GeneChip analysis
The extraction of total RNA was performed with the Total RNA Isolation System (Promega). RNA quality and quantity were assessed using the Agilent Bioanalyzer 2100 electrophoresis system and the Nanodrop ND-1000 spectrophotometer (NanoDrop technologies) respectively. cDNA synthesis and further processing for the transcriptomic analysis were performed according to the protocol of the GeneChip® manufacturer (Affymetrix). Ten micrograms of total RNA per sample was used for cDNA synthesis and a Poly-A RNA control was added to the RNA sample using the GeneChip eukaryotic poly-A control kit (Affymetrix). The synthesis of cDNA was performed using denatured total RNA (70°C for 10 min), 13 ng µl−1 random hexamer primers (Roche), 0.5 U µl−1 SUPERase In (Ambion), 25 U µl−1 SuperScript II reverse transcriptase (Invitrogen), 0.5 nmol µl−1 deoxynucleoside triphosphates in 1× first strand buffer with 10 nmol µl−1 dithiothreitol. The latter mix was placed in a Thermocycler operated at 25°C for 10 min, 37°C for 60 min, 42°C for 60 min and 70°C for 10 min. The remaining RNA was degraded by treatment with 1 N NaOH, incubation for 30 min at 65°C and subsequent neutralization with 1 N HCl. cDNA was purified using a MiniElute PCR purification kit (Qiagen), where after the concentration of cDNA was assessed using the Nanodrop ND-1000 spectrophotometer (NanoDrop technologies). Eight micrograms of cDNA was fragmented to a size range of 50–200 nucleotides with DNase I (Promega) for 10 min at 37°C where after the enzyme was inactivated at 99°C for 10 min. The fragmented cDNA was verified by loading a sample on a 2% agarose gel containing 0.1‰ (v/v) GelRed (Biotium). Fragmented cDNA was end-labelled with biotin-dUTP using the GeneChip® DNA labelling reagent (Affymetrix) and terminal deoxynucleotidyl transferase (Promega) in 1× reaction buffer (Promega) at 37°C for 60 min. The reaction was stopped using 0.5 M EDTA. Biotinylated cDNA was hybridized to GeneChip®P. aeruginosa Genome arrays (Affymetrix) by incubation in a GeneChip® hybridization oven 640 at 50°C with orbital shaking at 60 r.p.m. for 16 h in the following MES-buffer [2-(N-morpholino) ethanesulfonic acid]: 0.1 µg µl−1 herring sperm DNA (Promega), 0.5 µg µl−1 bovine serum albumine (Invitrogen), 7.8% dimethyl sulfoxide (Sigma-Aldrich), 50 mM B2 control oligo (Affymetrix) in 1× hybridization buffer (made according to instructions of Affymetrix GeneChip®). Microarray slides were washed according to the manufacturer's protocol and stained using a fluidics station 450 (Affymetrix) and applying the Midi_euk2v3-450 protocol. Briefly, the staining procedure was comprised of the following steps: (i) binding of streptavidin (Invitrogen) to the biotin-labelled hybridized cDNA; (ii) binding of a biotin-conjugated streptavidin antibody (Labconsult) and (iii) binding of R-phycoerythrin-conjugated streptavidin (Invitrogen). Microarrays were scanned at the emission wavelength of 570 nm where after the data analysis took place. Raw Affymetrix data were pre-processed using the Affy package in BioConductor as follows: (i) background correction based on the robust multichip average (RMA) convolution model (Irizarry et al., 2003); (ii) quantile normalization to make expression values from different arrays more comparable (Bolstad et al., 2003) and (iii) summarization of multiple probe intensities for each probe set to one expression value per gene using RMA (Irizarry et al., 2003). To test for differential expression between the different microgravity-analogue conditions (LSMMG and RG) and the reference condition (NG), the Bayesian adjusted t-statistics was used as implemented in the LIMMA package (version 2.5.15) (Smyth, 2004). P-values were corrected for multiple testing using the Benjamini and Hochberg's method to control the false discovery rate (Benjamini and Hochberg, 1995). Only fold change ratios with adjusted P-values below 0.05 were included as statistically significant. Microarray analysis was performed on all three biological replicates. Based on quality control analysis of the different samples, we identified one of the biological replicates of LSMMG as an outlier, which was discarded for further analysis. The full description of the array analysis platform and the complete array data have been deposited at the Gene Expression Omnibus website ( http://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?token=lhyzzsimmiqwwfe&acc=GSE16970) under accession number GSE16970.
Quantitative real-time PCR analysis
RNA extraction, reverse transcription, primer design, RNA and cDNA quality control, and qRT-PCR analysis were performed as described previously (Crabbéet al., 2008). The list of qRT-PCR primers used in this study is presented in Table S1. Data were normalized using the rrnB gene as an internal control (Crabbéet al., 2008; Wilson et al., 2008), according to the method of Livak and colleagues (Livak and Schmittgen, 2001). The experiment was performed at least in biological triplicate and technical duplicate.
Quantification of uronic acids
The removal of cell material and subsequent isolation of uronic acids from 5 ml supernatants was performed as previously described (Crabbéet al., 2008). Following uronic acid extraction, the quantification was done by applying the modified carbazole assay of Knutson and Jeanes (1968), adapted by May and Chakrabarty (1994). A standard curve of sodium alginate (Fluka), ranging from 10 to 1000 µg ml−1, was included in the experiment.
For all stress assays, bacteria were immediately subjected to the respective stress upon removal from the RWV bioreactors. Specifically, 100 µl of bacterial culture was diluted 10-fold in 0.85% NaCl solution containing 0.05% (v/v) HCl (pH = 3.3) for acid stress or 0.88 M H2O2 for oxidative stress respectively. For thermal stress-resistance assays, bacteria were diluted 10-fold in 0.85% NaCl and exposed to 50°C using a heating block. After an exposure period of 15 or 60 min, cultures were resuspended in 0.85% NaCl and viable counts were determined. The survival percentage was calculated by relating the colony forming units (cfu ml−1) at time zero (i.e. before addition of the stress) to the cfu ml−1 after exposure to the respective stress.
Dissolved oxygen determination in RWV bioreactors
The measurement of the DO concentration in the culture media in RWV bioreactors was performed using a Knick KNI913 oxygen metre. After cultivation of bacteria for 24 h in the RWV bioreactors, the DO probe was inserted into the culture medium through the main port of the RWV bioreactor and dissolved O2 readings were recorded.
Oxygen transfer rate assessment in LSMMG and NG cultures
Aliquots (10 ml) of LSMMG- and NG-grown cultures of P. aeruginosa PAO1 were frozen at –20°C for at least 3 weeks to sterilize the culture in a non-invasive manner. After thawing, plating on solid LB medium revealed no viable cells. The use of complete cultures (i.e. non-viable cells and culture medium) was adopted to study the influence of both cell-bound and soluble factors on the oxygen transfer rate. Ten millilitres of thawed cultures were transferred into 50 ml glass measure cups. A continuous stirring was established using a small magnet, rotating at a speed of 200 r.p.m. The transfer of oxygen from the air into the culture was assessed by applying the ‘Dynamic method’ (reviewed by Garcia-Ochoa and Gomez 2009), taking into account an Oxygen Uptake Rate of zero (due to the absence of biologic oxygen consuming elements). In the initial phase, stirred cultures were omitted of oxygen by bubbling argon gas into the liquid until an oxygen concentration lower than the detection limit of the probe (0.1 mg l−1) was measured. In a second phase, the liquid was put in contact again with ambient air, at a continuous stirring rate, and the increase in oxygen concentration as a function of time was measured until saturation was reached. The oxygen mass transfer rate is described as proportional to the latter concentration gradient, reflected in the volumetric mass transfer coefficient kLa. The kLa is determined based on the following equation:
where CL is the DO concentration in the bulk liquid, C* the oxygen saturation concentration in the bulk liquid and t the time. The resulting curves are sigmoidal and are modelled by the 3-parameter Gompertz model, providing a regression coefficient of at least 0.999. The saturation concentration in the culture liquid of each test condition is extracted from the Gompertz equation, as described by Zwietering and colleagues (1990). The latter value is used as C* in Equation 1 in order to plot ln (1 – CL/C*) as compared with time. The obtained curve was fitted again with the 3-parameter Gompertz model, which allowed the calculation of the lag phase (i.e. the time to initiate exponential oxygen increase in the bulk liquid) and the slope (i.e. kLa). The oxygen transfer rate experiment was performed in technical duplicate for each biological replicate. In addition, the oxygen transfer rate was determined for sterile LB medium supplemented with 0, 1, 5 and 10 mg ml−1 sodium alginate (Fluka) in LB medium (all performed in triplicate).
Assessment of fluid-mixing in LSMMG, NG and RG
RWV bioreactors were completely filled with distilled water and air bubbles were removed to avoid undesired fluid-shear. A 1 ml syringe containing 100 µl 0.03% crystal violet (Merck) was mounted on one sample port of the bioreactor. The RWV bioreactors were placed on the RWV and RPM devices to obtain the LSMMG, NG and RG test positions. After 10 min of rotation, allowing equilibration of the fluid relative to the bioreactor, approximately 50 µl of crystal violet was gently injected into the bioreactor while it was rotating. The dispersal of the crystal violet solution into the water was monitored with online video recording (see supplementary movie files).
Statistical analysis and bioinformatics
All experiments were performed in triplicate, unless stated otherwise. The statistical determination of significance (α = 0.05) was calculated with Microsoft Office Excel 2003 using a two-sample Student's t-test on the biological replicates of each experimental condition. For statistical analysis of three samples, the stats-package in R 2.9.0 was used by performing analysis of variance (aov), followed by a Tukey multiple comparison test (TukeyHSD). Data were checked for normality (Shapiro-Wilk test) and homoscedasticity (Bartlett test). Calculation of the statistical significance of the overlap between the present microarray gene sets and previously published microarray data was performed by applying the hypergeometric distribution method (Fury et al., 2006).