Introduction

Rhamnolipids are surface-active molecules produced by Pseudomonas aeruginosa and are essential biotechnological products with a wide range of applications in many areas, e.g. cosmetics (emulsifiers), food industry (food formulation ingredients) [25], biomedicine (due to their antiadhesive and antimicrobial properties) [31], agriculture (due to their antimicrobial and antifungal effects) and bioremediation (removal of toxic heavy metals from soils) [14, 24, 29].

Soil washing

Soil washing with rhamnolipids is one of the most often proposed potential strategies for reducing heavy metal toxicity to indigenous soil bacteria (for review see [24]), and it is postulated that removal of metal toxicity includes complexation of heavy metals by rhamnolipids [26].

Bioremediation

Although for remediation of sites polluted with heavy metals, both the potential of rhamnolipids to remove heavy metals from soils as well as the possible toxic effects of rhamnolipids to soil microorganisms should be taken into account. Existing data about the interactions between rhamnolipids−metal complexes and soil (micro)organisms are relatively rare and controversial. Stacey et al. [35] postulated that rhamnolipids from neutral lipophilic complexes with cationic metal ions and enhance the absorption of zinc by plant roots. Conversely, Al-Tahhan et al. Rhamnolipids reduced the cell surface charge of Gram-negative bacteria resulting in increased hydrophobicity and, thus, reduced cadmium uptake.

Similarly, several studies have shown a reduction in heavy metal toxicity to bacteria in the presence of rhamnolipids [20, 33]. On the other hand, Shin et al. [34] showed that the addition of 240 mg l-1 rhamnolipids for an in situ remediations inhibited the phenanthrene degrading bacteria referring to the potential toxic effect of biosurfactants. Indeed, rhamnolipids have been shown to exhibit powerful antibacterial, antifungal and algicidal activities [8, 37]. Thus, to be used for in situ bioremediation, preparations of rhamnolipids should be thoroughly characterized not only concerning their remediation efficiency but also for potential toxic effects pathogenicity.

Objectives of the study

The main aim of this study was to characterize the rhamnolipids produced by Pseudomonas aeruginosa for their inherent toxicity to Gram-negative (Vibrio fischeri, Pseudomonas fluorescens, P. aeruginosa, Escherichia coli) and Gram-positive (Bacillus subtilis) bacteria and their potential to decrease Cd bioavailability and remove Cd- toxicity in aqueous media and soils.

Materials and Methods

1. Materials

CdCl2 (>98%) was obtained from Sigma, Tween 80 from Serva, components of growth media were either from LabM or Sigma, L-rhamnose, orcinol and 1-N- phenylnaphthylamine (1-NPN) were from Sigma-Aldrich. Rhamnolipids were purified from the culture broth of Pseudomonas aeruginosa. A sandy soil (initial concentration of Cd 0.17 mg kg-1) spiked with CdCl2 (1.5, 15, 150, 1500 or 15 000 mg of Cd kg-1) as previously reported [11] was used for the bioavailability studies. Before spiking, the soil was characterized in a certified laboratory and had the following properties: 10.6% of clay, 10.6% of silt, 72.8% of sand, 5.7% of organic matter; 39 g·kg-1 of CaCO3, 3.59 g·kg-1 of N, 0.62 g·kg-1 of P, 0.17 mg kg-1 of Cd; with 2.3 cmol+kg-1 of CEC (cation exchange capacity) and pH of 7.3. Deionized water was used throughout the study.

2. Characterization of the rhamnolipids-producing bacterium Pseudomonas

aeruginosa

In the current study, the 16S rRNA gene of P. aeruginosa DS10-129 was sequenced by DSMZ (German Collection of Microorganisms and Cell Cultures; and the partial sequence of the 16S rRNA gene was deposited in the EMBL nucleotide sequence database with accession number AM419153. The series was compared with those available in EMBL and NCBI databases using the program BLAST to find its closest homologues. The similarity matrix was constructed by pairwise analysis of validated (most completed) sequences using the corrections computed by the Kimura’s 2-parameter model [18]. A phylogenetic tree was created with the partial 16S rRNA gene sequences using a multiple sequence alignment software CLUSTALW [10] by the Neighbour-Joining method [32] and illustrated with the TreeView program. The resulting hierarchical clustering tree was “pruned” to save space, and the closest

relatives were retained.

3. Isolation, purification and characterization of rhamnolipids

Rhamnolipids were isolated from P. aeruginosa DS10-129 cell-free supernatant. P. aeruginosa was grown on the mineral medium containing (per l water) 20 g of glycerol, 0.7 g of KH2PO4, 2 g of Na2HPO4, 0.4 g of MgSO4•7H2O, 0.01 g of CaCl2 and 0.001 g of FeSO4•H2O and 1% of Yeast Extract. The presence of the rhamnolipids was verified by thin-layer chromatography on silica gel 60 F254, according to Matsufuji et al., [21]. Rhamnolipids were extracted from the cell-free supernatant of the 4-day bacterial culture by centrifugation at 8000 g for 20 minutes at 4°C and the subsequent filtration through a sterile filter (pore size 2 µm). The cell-free culture was acidified to pH 2 with 2 M H2SO4, and the precipitated rhamnolipids were extracted with an equal volume of 2:1 dichloromethane/methanol (liquid-liquid extraction) [38]. The organic phase was dried with anhydrous Na2SO4 to remove excess water and evaporated on a rotary evaporator (Buchi) at 60–70 °C to yield rhamnolipids. The rhamnolipids were dissolved in 0.05 M NaHCO3. The concentration of rhamnolipids was determined using the orcinol assay [6] by mixing 100 µl of diluted solution of rhamnolipids (purified with liquid-liquid extraction) with 900 µl of freshly prepared 0.19% orcinol solution in 53% H2SO4. The mixture was heated at 80oC for 30 min and its absorbance was measured spectrophotometrically at 421 nm. The concentration of rhamnolipids was calculated according to L-rhamnose standard curve (0 to 50 mg l-1) and by multiplying the result with a coefficient of 3.4 obtained from the correlation of pure rhamnolipids/rhamnose [3]. The critical micelle concentration (CMC) was determined by measuring the surface tension of serial dilutions of the rhamnolipids [17]. Fourier Transform Infrared spectrophotometer (FTIR) Perkin Elmer 100 series was used to determine the molecular structure of the rhamnolipids. The cell-free supernatant was acidified to pH 2 by adding drops of 2M Sulphuric acid to precipitate the rhamnolipids. The precipitated rhamnolipids were extracted with an equal volume of 2:1 dichloromethane/methanol. The organic phase was dried with anhydrous Sodium Sulphate (Na2SO4) and evaporated on a rotary evaporator (Buchi, Rota vapour R-200 Germany) set at 60-70°C. Approximately 2-5mg of the concentrated rhamnolipids were analysed with the FTIR spectrophotometer.

4. Luminescent bacterial strains for toxicity and bioavailability assays

The luminescent bacterial strains used for toxicity and bioavailability studies are listed in Table 1. Constitutively luminescent bacterial strains, both natural (Vibrio fischeri) and recombinant strains were used to evaluate the toxicity of the rhamnolipids using bioluminescence inhibition as a toxicity endpoint. Recombinant luminescent Cd- inducible sensor bacteria were used to study the modulatory effect of rhamnolipids on availability of Cd to bacteria. All recombinant bioluminescent bacterial strains, except Pseudomonas aeruginosa DS10-129 (pDNcadRPcadAlux) were constructed previously (Table 1). P. aeruginosa DS10-129 (pDNcadRPcadAlux) was initially constructed as Cd-inducible strain by electroporating a 14,525 bp plasmid pDNcadRPcadAlux [13], which contains bioluminescence-encoding luxCDABE genes under the control of Cd response elements: Cd-regulated promoter (promoter of cadA) and a Cd-binding regulatory protein (CadR), into P. aeruginosa DS10-129 competent cells [3]. Bacteria were plated onto LB agar (10 g of tryptone, 5 g of yeast extract and 5 g of NaCl per 1 l of deionised / distilled? water) containing 50 mg l-1 of tetracycline and the plasmid-containing colonies were selected by luminescence. During further experiments P. aeruginosa DS10-129 (pDNcadRPcadAlux) failed to be induced with Cd (maximum induction below the limit of detection), mostly due to its high background luminescence. However, the high bioluminescence level favoured the use of this strain as a model organism for general toxicity testing. 179

5. Analysis of toxicity and Cd bioavailability

Luminescent bacterial strains were either rehydrated from lyophilised culture (Vibrio fischeri) obtained from Aboatox, Turku, Finland or cultivated. V. fischeri was reconstituted in heavy metal MOPS medium (HMM medium) supplemented with 2% NaCl at room temperature for 1 h. The HMM medium contained (per l of deionised / distilled? water): 8.4 g of MOPS buffer, 0.4 g of glucose, 0.22 g of glycerol-2- phosphate, 3.7 g of KCl, 0.54 g of NH4Cl, 0.06g of MgSO4, 0.162 mg of FeCl3. All recombinant bacteria were cultivated freshly by growing the cultures overnight in 3 ml of LB medium [44] supplemented with appropriate antibiotics as in Table 1. The overnight culture was diluted 1:50 with 10–50 ml of HMM medium, grown until OD600 of 0.3 and then diluted to OD600 of ~0.1 prior to test.

5.1. Toxicity testing

To measure toxicity CdCl2 or CdCl2-rhamnolipids mixtures were analyzed by measuring the inhibition of bioluminescence of constitutively luminescent bacterial strains (Table 1). In addition, Pseudomonas aeruginosa DS10- 129(pDNcadRPcadAlux) was used to assess the inhibitory effect of rhamnolipids. The effect of rhamnolipids on viability of bacteria was evaluated by plating the treated bacteria on solidified growth medium (see below). Dilutions of CdCl2 (0.1 – 100 mg l- 1 as final concentrations), rhamnolipids (5 – 200 mg l-1 as final concentrations) or Cd- rhamnolipids mixture (the final concentration of rhamnolipids was 50 mg l-1 or a concentration reducing the light output of bacteria by 20%, indicated below) were prepared in HMM medium or HMM medium supplemented with 2% NaCl (in case of V. fischeri). For toxicity assessment, luminescence inhibition assay was performed in 96-well microplates essentially as in Mortimer et al. [23]. Briefly, 100 ml of the diluted test compound(s) was pipetted into 96-well microplate and 100 ml of the bacterial suspension was automatically dispensed into the wells. Bacteria were incubated at 20°C (V. fischeri) or 30°C (recombinant luminescent bacteria) and the luminescence was continuously recorded during the first 30 seconds of exposure and once after 30 minutes of incubation using Fluoroskan Ascent FL plate luminometer (ThermoLabsystems). Inhibition of bacterial bioluminescence by the tested compounds/mixtures was calculated as percentage of the unaffected control (HMM medium or HMM supplemented with 2% NaCl, respectively). 30-s and 30- min EC50 and EC20 values (the concentration of chemical which reduces the light output of bacteria by 50 or 20% after the respective exposure times) were calculated by linear regression from dose-response curves of the studied compounds. Measurements were performed in three independent assays. Viability of bacteria was assessed after their exposure to 100 mg l-1 of rhamnolipids by plating the bacteria onto LB agar plates containing appropriate antibiotics. Plates were incubated for 24 h at 30°C after which colony forming units (CFU) were counted.

5.2. Bioavailability testing

Availability of Cd (with or without rhamnolipids) to Cd-sensor bacteria (Table 1) both in aqueous environment and in soil-water suspension was analysed as described by Bondarenko et al. [4]. CdCl2 dilutions at final concentrations of 0.01-10 mg l-1 were prepared by rotating soil:water (1:10) aqueous suspensions of Cd-spiked soils at room temperature for 24 h. Non-spiked soil was used as a control for soil assays and deionised / distilled? water served as a control for CdCl2 dilutions. Rhamnolipids were added to CdCl2 dilutions or Cd-spiked soil suspensions to the final concentrations of 10, 20 and 40 mg l-1 . Samples (100 µl) were added to 100 µl of the sensor bacterial culture in HMM medium in 96-well microplates and incubated at 30oC for 2 h. Luminescence was measured with Fluoroskan Ascent FL plate luminometer. Induction of luminescence of sensor bacteria by Cd was calculated as follows:

Induction = LS/LB, where LS is luminescence in the sample (CdCl2, CdCl2-rhamnolipids mixture, soil- water suspension or its mixture with rhamnolipids) and LB is the background luminescence (bacteria in HMM medium added to water or unspiked soil). The concentration of Cd in the sample causing induction of bioluminescence twice above the background value was defined as minimal inducing concentration (defined also as limit of determination (LOD) by Ivask et al. [13]). Bioavailability analyses were performed in three independent assays. 243

6. Analysis of membrane permeability

The enhancement in permeability of Escherichia coli MC1061(pDNlux) cell membranes by rhamnolipids was measured by the uptake of a hydrophobic probe 1- N-phenylnaphthylamine (1-NPN) as described by Helander et al. [12]. As compared to hydrophilic environments, the fluorescence of 1-NPN is significantly enhanced in hydrophobic environments (e.g. membrane phospholipids), rendering it a suitable dye

to probe outer membrane integrity of Gram-negative bacteria [9]. Briefly, 50 µl of 40 µM 1-NPN dye and 50 µl of the surfactants (rhamnolipids or non-ionic chemical surfactant Tween 80 serving as a positive control) in 5 mM HEPES buffer (pH 7.2) were pipetted into black microplates. 5 mM HEPES buffer was used as a negative control. 100 µl of bacterial suspension in 5 mM HEPES buffer were automatically dispensed into each well and the fluorescence was immediately measured (Fluoroskan Ascent FL plate luminometer; excitation/emission filters 350/460 nm). The final concentrations of both surfactants in the test were 10, 40 and 100 mg l-1. The 1-NPN cell uptake factor was calculated as a ratio of fluorescence values of the bacterial suspension in the presence and absence of surfactants.

7. Analysis of free Cd2+ with ion-selective electrode

A Cd-selective electrode (Thermo Orion 96-48 and Thermo Orion 4-star meter; ThermoOrion) was used to measure the free Cd2+ in the aqueous environment and in soil-water suspensions. Before use, the electrode was polished with alumina strips. The inner filling solution of the electrode was replaced at least weekly. CdCl2 solutions at final concentrations of 11.2-1120 mg l-1 were prepared in deionised / distilled? water; 1:10 water suspensions of Cd-spiked soils were rotated for 24 h prior to the test. Rhamnolipids were added to these solutions at final concentrations of 10, 20 and 40 mg l-1 and incubated at 30oC for 2 hours. The concentrations of free Cd2+ were calculated by comparing the results from different test conditions with the response of the electrode to CdCl2 solutions in distilled / deionised? water.

8. Analysis of Cd desorption from soil

The effect of rhamnolipids on desorption of Cd in Cd-spiked soil was analyzed by Atomic Absorption Spectroscopy (Shimadzu, Kyoto, Japan). Rhamnolipids are added at final concentrations of 10, 20 and 40 mg l-1 to 1:10 water suspensions of Cd-spiked soil, incubated for 2 hours and centrifuged at 13,000´g for 5 minutes. The resulting soil-water extracts were acidified with 1% HNO3 and the concentration of Cd in the extracts was analysed.

9. Analysis of rhamnolipids sorption to soil

The sorption of rhamnolipids to soil was determined by comparing the surface tension of rhamnolipids (10, 20 and 40 mg l-1) in distilled / deionised? water and in soil: water (1:10) suspension. Surface tension was measured by the drop weight method [29].

Results

1. Characterization of rhamnolipids-producing strain Pseudomonas aeruginosa

DS10-129

The P. aeruginosa DS10-129 strain used in this work has been isolated previously and characterised for the synthesis of rhamnolipids [28]. Based on the 16S rRNA gene sequences evolutionary relationships of P. aeruginosa DS10-129 were determined. Neighbour-joining analysis showed that 14 compared isolates of Pseudomonas formed four distinct clusters of highly related members (Figure 1) and P. aeruginosa

DS10-129 belonged to cluster I where the similarity in 16S rRNA gene sequence was more than 90% and the difference in nucleotides was 2–43. Homologies between the 16S rRNA gene sequence of P. aeruginosa DS10-129 and the other 13 strains of Pseudomonas compared ranged between 89-99%. Among the strains, the isolate P. aeruginosa DS10-129 showed the highest (99%) sequence similarity with P. aeruginosa B2, a strain capable of degrading lubricant base oil consisting of trimethylolpropaneoleate, and the lowest (89%) with P. anguillispectica FTB-40 and P. frederiksbergensis AJ28. The sequence similarity of P. aeruginosa DS10-129 isolate with the out-group Escherichia coli (accession number X80724) was 83%. 303

2. Characterization of rhamnolipids produced by Pseudomonas aeruginosa

DS10-129

The presence of rhamnolipids in the P. aeruginosa DS10-129 culture broth was confirmed by thin layer chromatography where two anise aldehyde positive spots (Rf 0.32 and Rf 0.52, corresponding to di- and mono-rhamnolipids) were detected. The concentration of rhamnolipids extracted from the bacterial culture broth was 174 mg l- 1 and the critical micelle concentration (CMC) of the rhamnolipids was 22 mg l-1. FTIR spectroscopy was used to determine the molecular structure of the rhamnolipids (Figure 2). Strong and broad bands of the hydroxyl group (-OH) free stretch due to hydrogen bonding were observed in the region A (3368 cm-1). The presence of carboxylic acid functional group in the molecule was confirmed by the bending of the hydroxyl (O-H) of medium intensity bands in the region D (1455-1380cm-1). The aliphatic bonds CH3, CH2 and C-H stretching with strong bands are shown in regions B and D (2925-2856 and 1455-1380 cm-1). The carbonyl (C=O) stretching was found in the region C (1737cm-1) with strong intensity bands. Two other strong peaks between 1300 cm-1 and 1033 cm-1 in region E due to C-O stretch are characteristic of an ester functional group. The peaks in the range of 1121–1033 cm-1 were also

reported as C-O-C stretching in rhamnose by Pornsunthorntawee et al. [27]. These characteristic adsorption bands together demonstrate the presence of rhamnose rings and long hydrocarbon chains, which are characteristic for rhamnolipids according to Guo et al. [7]. Besides, the comparative study of mono- and di-rhamnolipids has shown the presence of the shoulder around 3006 cm-1 in the spectrum of mono- but not di-rhamnolipid [7]. In the FTIR spectrum of rhamnolipids produced by P. aeruginosa DS10-129 we could observe only a minor shoulder in this region, likely because of the dominance of the di-rhamnolipid in the mixture [29]. Moreover, we noticed stronger bands of pyranyl I sorption band in region F at 918-940 cm-1 and α- pyranyl II sorption band in region G at 838-844 cm-1 that according to Guo et al. [7] also suggested the dominance of the di-rhamnolipid in the mixture.

3. Toxicity of rhamnolipids to bacteria

The short-term inhibitory effect of rhamnolipids on bacteria was analysed by measuring the kinetics of bioluminescence of five bacterial strains (both, Gram- negative and Gram-positive bacteria) upon their exposure to rhamnolipids. Naturally luminescent Vibrio fischeri and recombinant luminescent Escherichia coli, Pseudomonas aeruginosa, Pseudomonas fluorescens and Bacillus subtilis (Table 1) were tested. Among the tested bacterial strains and in the conditions used, the rhamnolipids were least inhibitory to both Pseudomonas species (30-s and 30-min EC50 values 204 and 138–164 mg l-1, respectively) and most inhibitory to V. fischeri (30-s and 30-min EC50 values 89 and 45 mg l-1, respectively) (Table 2). The rhamnolipids inhibited the bacterial bioluminescence from the very first second of exposure (Figure 3) and only slight (up to 2-fold) increase in toxicity was observed when the exposure time was extended to 30 minutes (Table 2). Despite the inhibition of bioluminescence, the tested concentrations of rhamnolipids did not decrease the viability of bacteria: the number of viable cells (analysis of colony forming units on LB agar plates–data not shown) was not decreased after 30 minutes of exposure to 100 mg l-1 rhamnolipids (causing over 50% bioluminescence inhibition in three tested bacteria, Figure 3). It was subsequently shown, that the toxic effect of rhamnolipids was due to the increase in permeability of cell membranes of Gram-negative bacteria. As recorded for the non-ionic chemical surfactant Tween 80, rhamnolipids similarly facilitated the entrance of hydrophobic fluorescent dye 1-N-phenylnaphthylamine (1- NPN) into the cells of Gram-negative E. coli MC1061(pDNlux) (Table 3).

4. Effect of rhamnolipids on toxicity, bioavailability and concentration of free

Cd2+ in aqueous environment

4.1. The effect of rhamnolipids on Cd toxicity

Acute effects of CdCl2 solutions on the bioluminescence of four different bacterial strains were analysed. In contrast to the rhamnolipids, whose effects on bacterial bioluminescence occurred during the first seconds of exposure (Figure 3), the toxic effect of Cd on bacteria was observed only after 30 minutes of exposure. The 30-min EC50 of CdCl2 solutions were (mg Cd l-1): 0.16 for E. coli, 0.49 for B. subtilis, 0.96 for P. fluorescens and 4.4 for V. fischeri. Addition of sub-toxic concentrations of rhamnolipids (50 mg l-1 and EC20 level: 20 mg l-1 for B. subtilis and V. fischeri, 35 mg l-1 for E. coli and 70 mg l-1 for P. fluorescens) significantly mitigated the toxic effect of cadmium for all the used Gram-negative strains (Figure 4 A, C, D; Table 4). The most remarkable reduction of Cd toxicity was observed in case of E. coli followed by V. fischeri and P. fluorescens (50 mg l-1 of rhamnolipids reduced the toxic effect of Cd by 10, 4.8 and 2.3-fold respectively; Table 4 and Figure 4). Surprisingly, no mitigating effect of rhamnolipids on Cd toxicity was observed in case of Gram- positive bacterium B. subtilis (Figure 4 B), indicating that the effect of rhamnolipids may be related to the structure of the bacterial cell wall. 374

4.2. The effect of rhamnolipids on Cd bioavailability

Similarly to Cd toxicity, Cd bioavailability as measured with Cd-inducible bacterial strains, clearly decreased for the Gram-negative E. coli in the presence of rhamnolipids (Figure 5 A). The minimal inducing concentration of Cd for E. coli MC1061(pSLzntR/ pDNPzntAlux) decreased by 3 and 5.6-fold in the presence of 20 and 40 mg l-1 of rhamnolipids, respectively (Table 4). However, conversely to Gram- negative E. coli, the apparent availability of Cd to Gram-positive sensor B. subtilis BR151(pcadCPcadAlux) even increased (Figure 5 B, Table 4) showing again the dissimilar effect of rhamnolipids on the availability of Cd to bacterial cells with different cell wall structure. 385

4.3. The effect of rhamnolipids on free Cd ion concentration

A Cd-selective electrode was used to analyse the effect of rhamnolipids on the concentration of free Cd ions. Unfortunately, this electrode does not allow the

measurement of low concentrations of Cd (0.001–0.03 mg l-1) that were inducing the sensor bacteria. However, significant (about 95%) reduction in the amount of free cadmium by rhamnolipids (40 mg l-l) in aqueous environment containing 11.2 mg Cd2+ l-1 was observed (Table 4) further proving the strong metal complexing ability of rhamnolipids. At the same time, 70% and 9% reduction in free Cd2+ was observed in solutions containing 112 mg and 1120 mg of Cd2+ per l-1 (Table 4) showing the clear concentration-dependent saturation of metal complexing by rhamnolipids. 396

5. Effect of rhamnolipids on desorption, bioavailability and concentration of free

Cd2+ in soil

As one of the potential applications of rhamnolipids is to bind heavy metals from polluted soils, we investigated the effect of rhamnolipids on the bioavailability of Cd in soil-water suspensions. In the absence of rhamnolipids, 1.5% of the total Cd was available to E. coli sensor bacteria in the studied soil (calculation based on Tables 4 and 5) being in accordance with our previous studies, where median available fraction of Cd to recombinant sensor bacteria was around 1% [12,13,16]. Upon addition of

subtoxic concentrations of rhamnolipids (10 – 40 mg l-1) to the Cd-spiked soil suspensions, the analysis using the Cd-sensor bacteria showed that the bioavailable fraction of Cd in the soil was up to 2.5-fold decreased, i.e., up to 2.5-fold higher concentrations of Cd were required for the induction of sensor bacteria in the presence of rhamnolipids (Table 5) whereas the highest tested concentration of rhamnolipids (40 mg l-1) was most efficiently decreasing the bioavailable fraction of Cd. Surprisingly, the results were different from those observed in aqueous media as upon addition of rhamnolipids the bioavailability of Cd in soil suspensions was decreased for both sensors, Gram-negative E. coli MC1061(pSLzntR/ pDNPzntAlux) and Gram- positive B. subtilis BR151(pcadCPcadAlux).

The effect of rhamnolipids on the mobility of Cd in soil was even more complex. In general, the Cd added to the soils remained very strongly bound and for example, only 0.13% of the total Cd was desorbed from soil containing 1.5mg Cd kg-1in the current leaching conditions (Table 5). Addition of rhamnolipids facilitated the desorption of Cd in less contaminated soils: Cd desorption in 1.5 mg Cd kg-1 spiked soil was two- fold increased in the presence of 40 mg l-1 of rhamnolipids (Table 5). However, no increase in Cd desorption upon addition of rhamnolipids was observed in soils with environmentally not relevant high Cd content (15 000 mg kg-1) showing again that the

effect of rhamnolipids on complexation of Cd is concentration dependent. Due to the detection limit of the Cd-selective electrode, the amount of free Cd ions was not possible to measure in water extracts of less polluted soils (1.5-150 mg Cd kg-1), where desorption of Cd by rhamnolipids was detected by AAS (Table 5). The concentration of both, desorbed and free Cd in soil polluted at 15 000 mg Cd kg-1 was about 300 mg Cd kg-1 (2% of the total) and practically not dependent on the amount of rhamnolipids added (Table 5). 430

Discussion

The potential of rhamnolipids to be used for soil washing due to their ability to decrease the toxicity of heavy metals to soil microbes has been widely acknowledged [20, 33]. However, to be used for soil remediation, the addition of rhamnolipids should not adversely affect soil microorganisms, as microbes play the key role in the mineralization of biological components and in biogeochemical cycles. Previous studies have shown that the properties of rhamnolipids are often determined by their structure (for example the number of hydrophilic carboxyl groups being primary sites for complex formation with metal ions) and proportion of different types of rhamnolipids in the mixture [26, 35]. In order to determine these parameters, FTIR analysis was performed, which showed that P. aeruginosa DS10-129 produced a mixture of mono- and di-rhamnolipids where the latter was the predominant species. 443

The toxicity of rhamnolipids to luminescent Gram-negative and Gram-positive

bacteria

The reduction of light output of naturally luminescent bacteria Vibrio fischeri is a reflection of inhibition in bacterial metabolic activity and proportional to the toxicity of the test sample [5]. The photobacterial luminescence inhibition test for the toxicity evaluation (the standard protocol applies 5-30 min exposure times) has been used in our laboratory for the characterisation of various types of chemicals and environmental samples since 1993 [15]. In the current study, however, we observed that the effect of rhamnolipids on bacterial bioluminescence was evident already from the very first seconds of exposure (Figure 3) suggesting a disturbance of cellular energetic metabolism [22] and showing that rhamnolipids were most probably interfering with the normal function of bacterial cell membranes. The membrane permeabilising effect of rhamnolipids was confirmed by fluorescent 1-NPN dye. In

the bioluminescence inhibition assay with five bacterial strains, the 30-min EC50 values of rhamnolipids ranged from 45 to 167 mg l-1 (Table 2), which exceeded the CMC of the rhamnolipid mixture (22 mg l-1). It is interesting to note, that the EC50 of the rhamnolipids for P. aeruginosa DS10-129 (EC50=138 mg l-1; Table 2) was comparable to the highest concentration of rhamnolipids in the culture broth of this strain (174 mg l-1). However, even if inhibiting the bacterial bioluminescence, the tested concentrations of rhamnolipids were not bactericidal. On the contrary, the concentration of rhamnolipids, which inhibited the bioluminescence of the rhamnolipid-producer strain P. aeruginosa DS10-129, even remarkably stimulated the bioluminescence of this strain after 30-min exposure (Figure 3) and may show the adaptation of this strain to permeabilising effects of rhamnolipids. This emphasizes the action of rhamnolipids via reversible modulation of bacterial membranes. 469

Modulatory effect of rhamnolipids on the toxicity and bioavailability of Cd to

luminescent Gram-negative and Gram-positive bacteria in aqueous media

Cadmium inhibited the luminescence of bacteria at sub mg l-1 level in case of all recombinant luminescent bacterial strains and was somewhat less toxic to naturally luminescent V. fischeri (30 min EC50 4.4 mg l-1). The latter effect could be related to different cadmium speciation in the test solution containing 2% NaCl. Indeed, Villaescusa et al. [36] have shown that the toxicity of Cd to V. fischeri was remarkably increased at lower NaCl concentrations in the test medium. Addition of the subtoxic concentrations of rhamnolipids remarkably (up to 10-fold) reduced the toxic effect of Cd to Gram-negative bacterial strains. The remarkable (up to 95%) decrease in amount of free Cd ions in the presence of rhamnolipids was also shown by Cd-selective electrode (Table 4). Indeed, the complexing effect of rhamnolipids on heavy metal ions has been shown previously [26]. According to Nitschke and Costa [25], the optimal value of the rhamnolipids-Cd complexation ratio is 2 mol of rhamnolipids per mol of Cd. Thus, as the concentrations of rhamnolipids (10-40 mg l-1) in the test greatly exceeded the bioavailable and toxic concentrations of Cd (0.0014-4.4 mg l-1), the decrease in cadmium availability and toxicity to Gram- negative bacteria was at least partly caused by the reduction of free Cd-ions resulting from Cd complexation by rhamnolipids. On the other hand, despite the complexation, the availability of Cd to Gram-positive B. subtilis increased, which resulted in the equal toxicity of Cd in the presence or absence of rhamnolipids (Figure 4, Table 4). A

similar trend was demonstrated with Cd sensor strains: Gram-negative E. coli and Gram-positive B. subtilis. Rhamnolipids decreased the Cd availability to E. coli but even increased to B subtilis (Table 4).

The modulatory effect of rhamnolipids on membranes of Gram-negative bacteria in the presence of low concentrations of rhamnolipids have been previously shown and include the release of negatively charged lipopolysaccharides (LPS) resulting in the reduction of the overall cell charge and increase in cell surface hydrophobicity leading to better protection from toxic cationic compounds due to their decreased uptake [1, 30, 33]. As Gram-positive bacteria differ from Gram-negative bacteria in having only one membrane not containing LPS in their cell envelope, these remarkable changes in cell surface hydrophobicity are theoretically not possible and could explain the differences in Cd availability and toxicity to Gram-negative and Gram-positive bacteria. Thus, the modulation of Cd toxicity by rhamnolipids is obviously not only due to the complexation of Cd and reduction in its bioavailability but rather due to the combination of the interplay between the direct complexation of Cd and the effects on bacterial membranes that may modulate the net uptake of Cd. 507

Modulatory effect of rhamnolipids on the mobility and availability of Cd in soils

In soils, the effect of rhamnolipids on the mobility of Cd was even more complex: rhamnolipids (10 – 40 mg l-1) caused additional desorption of Cd from Cd-polluted (1.5 -150 mg Cd kg-1) soils (Table 5). However, emphasize must be made that only a minor fraction of Cd (e.g., 0.13% of the total Cd in soil containing1.5 mg kg-1 Cd) was desorbed from soil in the absence of rhamnolipids. This is in agreement with our previous studies on 60 heavy metal polluted agricultural soils where median water extractability of Cd was 0.2% [16]. The addition of 40 mg l-1 rhamnolipids increased the desorption twice. Our further experiments showed that this additionally desorbed fraction of Cd remained complexed with rhamnolipids and was not available to Cd- sensor bacteria. Interestingly, availability of Cd in soil was decreased to both, Gram- negative Cd sensor E. coli MC1061(pSLzntR/ pDNPzntAlux) and Gram-positive sensor B. subtilis BR151(pcadCPcadAlux) suggesting that in this environment, the biological effect (possible alterations of bacterial membranes) of rhamnolipids was not significant and had no influence on the bioavailability of cadmium. One explanation for dissimilar behaviour of Cd in aqueous solution and in soil could be the

less effective concentration of rhamnolipids in soil-water suspension compared to that

of the aqueous environment due to the sorption of rhamnolipids to soil particles (57, 34 and 14% of the rhamnolipids at concentrations 10, 20 and 40 mg rhamnolipids l-1 in soil-water suspension were sorbed to soil, respectively). 528

Conclusion

In this paper we showed that recombinant luminescent strains of various Gram- negative and Gram-positive bacteria can be very useful in mechanistic analysis of complex environmental problems, especially if bioavailability of heavy metals in various types of soils is concerned. Indeed, after combining the constitutively luminescent and Cd –sensing recombinant bacteria this paper is the first to report on dissimilar effects of rhamnolipids on heavy metal toxicity and bioavailability to Gram-positive and Gram-negative bacteria (due to their different outer cell wall structure) and indicates that rhamnolipids may modulate the bioavailability and toxicity of Cd to bacteria either by complexation of Cd or by effects on bacterial cell membranes. 540