Modulation of Metabolism and Switching to Biofilm Prevail over Exopolysaccharide Production in the Response of Rhizobium alamii to Cadmium

Heavy metals such as cadmium (Cd2+) affect microbial metabolic processes. Consequently, bacteria adapt by adjusting their cellular machinery. We have investigated the dose-dependent growth effects of Cd2+ on Rhizobium alamii, an exopolysaccharide (EPS)-producing bacterium that forms a biofilm on plant roots. Adsorption isotherms show that the EPS of R. alamii binds cadmium in competition with calcium. A metabonomics approach based on ion cyclotron resonance Fourier transform mass spectrometry has showed that cadmium alters mainly the bacterial metabolism in pathways implying sugars, purine, phosphate, calcium signalling and cell respiration. We determined the influence of EPS on the bacterium response to cadmium, using a mutant of R. alamii impaired in EPS production (MSΔGT). Cadmium dose-dependent effects on the bacterial growth were not significantly different between the R. alamii wild type (wt) and MSΔGT strains. Although cadmium did not modify the quantity of EPS isolated from R. alamii, it triggered the formation of biofilm vs planktonic cells, both by R. alamii wt and by MSΔGT. Thus, it appears that cadmium toxicity could be managed by switching to a biofilm way of life, rather than producing EPS. We conclude that modulations of the bacterial metabolism and switching to biofilms prevails in the adaptation of R. alamii to cadmium. These results are original with regard to the conventional role attributed to EPS in a biofilm matrix, and the bacterial response to cadmium.


Introduction
The exposure of bacterial cells to heavy metals in their environment mediates biological effects, usually through the direct or indirect action of reactive oxygen species [1,2]. In fact, nonredox-reactive metals, such as cadmium, show a high degree of reactivity towards sulfur, nitrogen and oxygen atoms in biomolecules. Cadmium may bind sulfur in essential enzymes, and alter their functions. Many studies have focused on the molecular mechanism of bacterial cell tolerance to cadmium, mainly for the case of species that are resistant to high metal concentrations, such as Stenotrophomonas [3] or Cupriavidus metallidurans (review in [4]). However, cadmium concentrations and its availability in metal contaminated soils are generally low. At low cadmium concentrations, Dedieu et al. [5] studied the interactions of Sinorhizobium meliloti extracellular compounds on cadmium speciation and availability, and Pagès et al. [6] reported on the completely different adaptation mechanisms of phenotypic variants of Pseudomonas brassicacearum in the presence of cadmium. Varied mechanisms account for cadmium detoxication in bacteria, involving exclusion, binding and sequestration. Cadmium is removed from cells by metal efflux transporters [7,8,9], reduced as cadmium sulfide [10], precipitated as insoluble salts [11], immobilized within the cell walls [12], or linked to chelating agents [13,14]. Cell exudates, such as proteins, siderophores and to a minor extent polysaccharide, play a role in the short-term interaction between Sinorhizobium meliloti and cadmium [5,15].
Because the adsorption of cadmium as well as of other metals can be associated with the secretion of exopolysaccharide (EPS) or capsular material [2,16,17], EPSs are considered as potential metal transporters in soil [18,19].
Gram-negative soil bacteria belonging to the commonly named rhizobia are able to produce EPSs with a large diversity of chemical structures [20,21]. These EPSs are the main contributors in legume-rhizobia interactions, leading to nodulation and nitrogen fixation. Eventually, much of what we know about rhizobia and their EPSs arises from studies of their symbiotic interactions with legume plants, whereas their interactions with non-legumes have been neglected. However, rhizobia associate with the roots of non-legumes such as Arabidopsis thaliana [22,23], Helianthus annuus [24] and Brassica napus [25]. Rhizobial EPSs also have functions beyond specific recognition in the nodulation process, such as plant growth promotion of non-legumes [24] or evasion from the defense response of plant legumes during crack entry in roots [26]. We should consider communities of rhizobia and their EPSs as integral and functionally important partners of a diverse plant rhizosphere.
For that purpose, we addressed the question of how Rhizobium alamii [27], an EPS-producing bacterium, responds to the toxic impact of cadmium. R. alamii is a rhizobacterium, isolated from the rhizophere of the sunflower, producing a mucoid and an adhesive EPS [24]. This bacterium colonizes the root system of the sunflower, A. thaliana, and rapeseed, and together with its production of EPS in the rhizosphere, it improves the physical structure of the root adherent soil and plant growth under conditions of hydric stress [23,24]. We studied the mechanisms through which EPS could contribute to the tolerance of R. alamii to cadmium, by using a mutant strain impaired in EPS production (MSDGT) [23]. We investigated cadmium adsorption on the EPS, in the absence or in competition with calcium. Ion cyclotron resonance Fourier transform mass spectrometry (ICRFT/MS) was used to monitor the metabolic perturbations after exposure of R. alamii cells to a cadmium concentration, which slowed down, but did not inhibit cell growth. We discuss the means by which the bacterial cells adjusted both their entire metabolic processes, and their way of life, to limit cadmium damage. We have for the first time shown that cadmium promotes the formation of a biofilm by R. alamii, and that this change occurs independently of the presence of EPS. These results are original with regard to the conventional role attributed to EPS in a biofilm matrix, and the bacterial response to a heavy metal.

Results and Discussion
Effect of cadmium on R. alamii growth and EPS production In this study, the minimal inhibitory concentration (MIC) of cadmium (Cd 2+ ) in R. alamii cells was assessed at 44 mM (5 mg.L 21 Cd 2+ ) in a tenfold diluted Tryptic Soy Broth (TSB/10) and as illustrated in Figure 1. The increase of Cd 2+ concentration from 1 mg.L 21 to 2 mg.L 21 almost doubled the lag phase. In our hands, cadmium MIC was 53 mM (6 mg.L 21 ) on R. alamii wt and MSDGT mutant cells on an agarose solidified TSB/10 medium ( Figure 2). 1 H NMR spectra of the EPS isolated from the bacteria, cultured in the absence or presence of cadmium, showed that the chemical structure of the EPS was unmodified (A. Heyraud, personal communication). Cadmium concentrations ranging from 0 to 133 mM (0 to 15 mg.L 21 of Cd 2+ ) did not significantly (p.0.5) modify the amount of EPS synthesized by the wt strain in a RCV mineral medium at pH 6.8, supplemented with glucose ( Figure  S1). The R. alamii behaved like Sinorhizobium meliloti whose EPSs content is unmodified, at pH 7 in response to 10 mM of cadmium nitrate [28].

R. alamii metabonomics in response to cadmium
We carried out metabonomic investigations on cells grown up to the end of the exponential phase, with or without 18 mM (2 mg.L 21 ) Cd 2+ , in order to determine which metabolism pathways were altered by cadmium in R. alamii. We used a screening method based on ion cyclotron resonance Fourier transform mass spectrometry (ICRFT/MS). This technology allows high-precision measurements to be made of a charged mass, within an error range of only a few parts per million. MassTRIX (http://masstrix.org) allows the compounds, detected with a chemically probable structure, to be assigned in the context of restricted metabolite possibilities for a given organism, using the KEGG pathway database: http://www.genome.jp/kegg/ pathway.html [29]. By considering the set of R. alamii metabolites detected in positive and negative ESI of MS, in the experiments with or without cadmium, 1897 putative compounds were common to these conditions, and respectively 936 and 653 probable components were exclusively expressed, in the presence or absence of the metal. Cadmium induced metabolic alterations in major pathways, as summarized in Tables 1 and 2. Cadmium increased the number and level of enzymes and metabolites involved in the sugar metabolism (potentially fructose, glucose, mannose, galactose, cellobiose, inositol, starch and sucrose), phosphorylated intermediates of glycolysis compounds (likely glucose 6-phosphate, fructose 6-phosphate, glucose 1-phosphate), ABC transporters of sugars (potential methyl-galactoside, Dallose, fructose, cellobiose) and the phosphotransferase system which, in bacteria, is the major carbohydrate transport system for incoming sugar substrates, through translocations across the cell membrane. Cadmium ions are admitted into sensitive bacterial cells by the energy-dependent manganese transport systems, where they cause rapid cessation of respiration by binding to sulfhydryl groups in proteins [30,31]. Glycolysis therefore appeared to be an alternative pathway for energy. Under cadmium stress, hexoses were potentially channelled towards a pentose phosphate pathway. In the presence of Cd 2+ , the nucleotide metabolism revealed a modulation of the purine metabolism, with an accumulation of purine-based nucleosides (adenosine, deoxyadenosine, guanine and deoxyguanine) at the same time as a decrease in purine-based nucleotides (AMP and dGMP) as illustrated in Figure 3. This result was confirmed by ultra performance liquid chromatographic (UPLC) analysis ( Figure 4). The decrease in nucleotide contents and the recovery of pyrimidine and purine compounds could be consistent with an adaptation process, through slowing of cell division in cells likely to adapt to the metal toxicity [32]. In the presence of cadmium, the citric acid cycle and tryptophan metabolism were shut off (Table 1) suggesting the cessation of aerobic sugar respiration and amino acid synthesis. A glutathione precursor, c-L-Glutamyl-L-cysteine, was identified under cadmium stress. In response to cadmium, Pseudomonas brassicacearum is found to switch from the citric acid cycle to an anaerobic metabolism [6], and the inhibition of protein and glutathione syntheses are described in the response of R. leguminosarum to cadmium [14,33]. Cadmium activated the biosynthesis of lipopolysaccharide and capsular polysaccharide, which are the outer membrane components of Gram-negative bacteria and are potential metal binding sites [6,12,34]. Pyrophosphate was detected following Cd 2+ treatment only. Pyrophosphate can form crystals with cadmium [35]. The accumulation of inorganic phosphate is a detoxification mechanism reported in Klebsiella aerogenes [36] and is also involved in the regulation of biofilm formation in Pseudomonas fluorescens [37]. Cadmium is also activated the inositol phosphate metabolism that regulates cytoplasmic Ca 2+ and communication, and the cyclic ADP-ribose pathway, which is a  Glycolysis/Gluconeogenesis 0 7 Citrate cycle (TCA cycle) 3 0 Pentose phosphate pathway 0 6 Pentose and glucuronate interconversions 1 4 Fructose, mannose, inositol and galactose metabolism

ABC transporters 13 20
Two-component systems 0 2 Phosphotransferase system 9 1 7 Rhizobium alamii cells were grown up to a late exponential phase, with 2 mg.L 21 of cadmium, as compared to the absence of cadmium. The potential metabolites identified were matched in KEGG pathway using MassTRIX (http:// masstrix.org). doi:10.1371/journal.pone.0026771.t001 secondary messenger for Ca 2+ mobilisation [38]. The structural homology between calcium and cadmium ions might account for the activation of pathways needed to maintain calcium homeostasis and signal transduction. Altogether, these results show that the adaptation of R. alamii cells to cadmium could imply a multifaceted scheme using metal binding, precipitation or export, changes in respiration process, and the retention of cadmium-like metal homeostasis.
Cadmium induces the formation of a biofilm by R. alamii independently of EPS synthesis R. alamii wt and MSDGT strains were grown statically in the presence of cadmium (from 0 to 15 mg.L 21 of Cd 2+ ), in mineral RCV medium supplemented with glucose (for composition see materials and methods) to promote EPS synthesis. Figure 5A shows staining of biofilms formed on the tube walls using crystal violet [39], and Figure 5B represents the quantity of planktonic  A biofilm is a lifestyle able to resist various environmental stressors such as antibiotics [40], or metals such as copper, zinc and lead [41], and nickel [42]. In the present study, we reveal for the first time cadmium-induced biofilm formation by R. alamii cells. Interestingly, wt and EPS-mutant strains showed the same change in growth, from free-swimming to biofilm mode, in response to cadmium, suggesting that biofilm formation, rather than EPS, was a remediation to metal toxicity.

Cadmium binds to the EPS of R. alamii
The chemical structure of R. alamii EPS contains carboxylic and hydroxyl functions [43] that may bind metals. The biosorption of metal ions on the EPS of R. alamii has been measured in a MOPS buffer and in a calcium-containing medium, and has been described using the Langmuir and Freundlich isotherms [44]. The parameters calculated from these two models are summarized in Table 3. The Langmuir model, based on the assumption of a monolayer adsorption onto a solid surface with a defined number of identical sites, gave the best fit of equilibrium adsorption data measured in a MOPS buffer. The Freundlich isotherm, which is an empirical model used to describe heterogeneous systems, best fitted the biosorption of cadmium in the presence of an excess of calcium ions. Both models showed that the EPS of R. alamii actually bound cadmium. An apparent distribution coefficient of 3480 L.mg 21 was found at an EPS concentration of 0.125 g.L 21 and a maximum biosorption capacity of 11 mg of cadmium per g of EPS was determined. Both models showed that calcium, an abundant ion in soil solutions, competed with cadmium and reduced its sorption. At the highest EPS concentration tested (1 g.mL 21 ), the EPS of R. alamii showed the lowest ability to bind Cd 2+ (apparent distribution coefficient of 387 L.mg 21 and maximum biosorption capacity of 6 mg of cadmium per g of EPS), showing that polysaccharide chain-chain interactions could preclude the binding of cadmium.

Cadmium and biofilm matrix imaging by confocal laser scanning microscopy
In Figure 6, the structure of cell organization, single cells, microcolonies and biofilms, from R. alamii wt and the MSDGT mutant, were examined after 3 days and 5 days of incubation at 30uC. The constitutive expression of GFP allowed the bacterial cells to be imaged. EPS was labelled with a fluorescent lectin (left lane only). Figure 6 shows typical pictures of the development of R. alamii wt and MSDGT mutant cells on a membrane, in the presence (2 mg.L 21 ) or absence of cadmium. At 3 days of growth in the absence of cadmium, the R. alamii wt and MSDGT mutant colonized the surface, in the form of microcolonies ( Figure 6A), or scattered single cells ( Figure 6E), respectively. After 5 days of growth, R. alamii wt and MSDGT formed spread biofilms with a loose architecture ( Figure 6B and 6F). However, cadmium already induced the formation of condensed biofilms of both strains, after 3 days of incubation (Figs. 6C and 6G). After 3 and 5 days, in the presence of cadmium, EPS-producing R. alamii wt cells formed globular biofilms expanded in the z direction (Fig. 6D), whereas the MSDGT mutant formed broad, flat and dense biofilms ( Figure 6H). We conclude that the EPS production of R. alamii enabled the spatial propagation of the cells, and the construction of a 3D-spread biofilm matrix. This observation corroborates the contribution of bacterial exopolymers to the biofilm architecture [45,46]. The localization of EPS in R. alamii biofilms grown in the presence of cadmium did not occur more frequently than in the absence of the metal, which confirms that R. alamii does not modulate EPS production in response to cadmium toxicity.
The view of the role of rhizobia and their EPS being restricted mainly to symbiotic interactions with legumes is insufficient to gain insight into their physiology, and ability to adapt to environmental fluctuations. Altogether, our data has revealed some unique results, related to the metabolic response and lifestyle of R. alamii, and to the function of the EPS in response to cadmium. The belief that EPSs are the main players in metal tolerance of bacteria is persistent in microbiology, mainly in environmental science literature. Our experiments show that the bacterial polysaccharide (EPS) of R. alamii can bind cadmium, although this is not the way that these bacteria use to adapt to the metal toxicity. Changes of lifestyle, from planktonic to biofilm growth, and altering the metabolism, are the means to escape metal toxicity by wt cells as well as by mutant cells that no longer produce EPS. Further studies are being designed to determine the role of R. alamii and its EPS on plant cadmium uptake in the rhizosphere.

Tolerance of Rhizobium alamii to cadmium
The wt and MSDGT strains were grown in a ten-fold diluted tryptic soy broth (TSB/10), or in a RCV supplemented with glucose (2 g.L 21 ) and various concentrations of cadmium (0 to 133.4 mM; 0 to 15 mg.L 21 of Cd 2+ ). The OD 630 nm was measured over time from triplicate samples and reflected planktonic cell concentrations.
The R. alamii wt cells were grown up to an early stationary phase in TSB/10 supplemented with 0 or 2 mg.L 21 of cadmium nitrate. The cells were collected at OD 600 nm near to 0.07. Cytosolic cellular extracts were extracted in 50/50 methanol/ water, in an ultrasonic bath for 15 min. The pellets were centrifuged at 10000 rpm for 15 min, and the supernatant was analyzed using a Bruker-Daltonics APEXQ 12 Tesla ICR-FT mass spectrometer (Bremen, Germany). The samples were introduced by macrospray infusion, at a flow rate of 120 mL.h 21 , ionised in negative and positive electrosprays (ESI), and 256 scans were accumulated over a broadband range of masses (m/z 150-2000). The instrument was externally calibrated on clusters of arginine every measurement day, and the mass spectra were internally calibrated with phtalate diesters in positive ESI, and with fatty acids in negative ESI, thus ensuring a maximum error of 100 ppb. For the ICR-FT MS experiments, three replicates were analyzed for each condition ( Figure S2), and the masses common to all three measurements were selected in order to generate a list of potential compounds. From the full set of detected compounds, we focused on those having a suggested attribution for the chemical structure with a difference between the detected and calculated masses of less than 63 ppm, with detected peaks preferentially confirmed by the presence of 18 0, 15 N or 13 C isotopes. The peaks exceeding a threshold signal-to-noise ratio of 3 were exported to peak lists, and were submitted to a metabolite-annotation web interface, MassTRIX (http://masstrix. org). MassTRIX processes the submitted mass peak list by comparing the input experimental masses with all of the compounds recorded in the Kyoto Encyclopedia Genes and Genome (KEGG) chemical compound database, using Rhizobium leguminosarum as the model organism. The MassTRIX annotation of metabolites was used to highlight differences in metabolic pathways between the R. alamii cells incubated with or without Cd 2+ 2 mg.L 21 . Moreover, the compounds identified by ICR-FT/MS were matched with the KEGG reaction components. A full table of metabolites was added as supporting information  Tables S1, S2, S3, S4. The raw data and the position of the potential metabolites in KEGG pathways can be consulted at http://metabolomics.helmholtz-muenchen.de/masstrix2, section ''job status'', job access numbers 10031710000016906 and 10031709591316561 for wild type cells [Cd 2+ ] 2 or 0 mg/L in negative mode, and 10031709545915503 and 10031709560715818, for wt cells [Cd 2+ ] 2 or 0 mg/L in positive mode.
The analyses of pyrimidine and purine metabolites, with or without cadmium culture conditions, were performed by ULPC-MS on three independent replicates, which were characterised by similar profiles, indicating a high reproducibility of the three experiments ( Figure S3 and Figure S4).

Adhesion of bacterial cells
The R. alamii wt and MSDGT mutant were statically grown to a stationary phase, in polypropylene tubes containing RCVmedium, supplemented with 2 g.L 21 glucose and cadmium nitrate (0 to 133.4 mM; 0 to 15 mg.L 21 of Cd 2+ ). Bacterial adhesion to solid surfaces and the formation of biofilms were monitored using a crystal violet staining-based protocol adapted from [39]. Breafly, the planktonic bacteria were removed from the tubes. Washing was performed with sterile water. After careful remove of the water, staining was performed with 2 mL of 1% crystal violet solution, 15 min at room temperature. The crystal violet solution was gently removed and successive washings were performed with water. Each tube was inverted and gently tapped on paper towels to remove any excess liquid and allow tubes to air-dry. 2 mL of 100% ethanol was added to each tube and OD 595 was measured in a cuvette on a spectrophotometer.

Data treatment
Statistical analyses (ANOVA) were made using version XV of the Statgraphics Centrion software.