Importance of LPS structure on protein interactions with Pseudomonas aeruginosa

Colloids and Surfaces B: Biointerfaces 67 (2008) 115–121

Contents lists available at ScienceDirect

Colloids and Surfaces B: Biointerfaces journal homepage: www.elsevier.com/locate/colsurfb

Importance of LPS structure on protein interactions with Pseudomonas aeruginosa Arzu Atabek, Yatao Liu, Paola A. Pinzón-Arango, Terri A. Camesano ∗ Department of Chemical Engineering Worcester Polytechnic Institute, USA

a r t i c l e

i n f o

Article history: Received 13 April 2008 Received in revised form 14 July 2008 Accepted 11 August 2008 Available online 23 August 2008 Keywords: Bacterial protein interactions Atomic force microscopy Bacterial adhesion Biomaterials

a b s t r a c t Atomic force microscopy (AFM) was used to quantify the adhesion forces between Pseudomonas aeruginosa PAO1 and AK1401, and a representative model protein, bovine serum albumin (BSA). The two bacteria strains differ in terms of the structure of their lipopolysaccharide (LPS) layers. While PAO1 is the wild-type expressing a complete LPS and two types of saccharide units in the O-antigen (A+ B+ ), the mutant AK1401 expresses only a single unit of the A-band saccharide (A+ B− ). The mean adhesion force (Fadh ) between BSA and AK1401 was 1.12 nN, compared to 0.40 nN for Fadh between BSA and PAO1. In order to better understand the fundamental forces that would control bacterial–protein interactions at equilibrium conditions, we calculated interfacial free energies using the van Oss–Chaudhury–Good (VCG) thermodynamic modeling approach. The hydrogen bond strength was also calculated using a Poisson statistical analysis. AK1401 has a higher ability to participate in hydrogen bonding with BSA than does PAO1, which may be because the short A-band and absence of B-band polymer allowed the core oligosaccharides and lipid A regions to be more exposed and to participate in hydrogen and chemical bonding. Interactions between PAO1 and BSA were weak due to the dominance of neutral and hydrophilic sugars of the A-band polymer. These results show that bacterial interactions with protein-coated surfaces will depend on the types of bonds that can form between bacterial surface macromolecules and the protein. We suggest that strategies to prevent bacterial colonization of biomaterials can focus on inhibiting these bonds. © 2008 Elsevier B.V. All rights reserved.

1. Introduction The ability to prevent infections on biomaterials is influenced by our understanding of protein–biomaterial interactions. As soon as they are implanted into the body, biomaterials quickly become coated with protein layers that form a conditioning film for bacteria deposition [1,2]. Once the initial cell attachment occurs, bacteria may grow into a biofilm that is very difficult to eradicate, causing failures of biomedical devices that usually require implant removal [3]. Researchers are actively trying to create coatings for catheters or other indwelling devices that will resist biofilm formation [4,5], but the coatings must also resist protein deposition in order to successfully prevent infection. The ubiquitous nature of Pseudomonas aeruginosa makes it an appropriate model organism to characterize bacterial protein interactions. P. aeruginosa is an opportunistic pathogen that colonizes multiple environments by using various natural com-

∗ Corresponding author at: Life Sciences and Bioengineering Center at Gateway Park, 60 Prescott Street, Worcester, MA 01605, USA. Tel.: +1 508 831 5380; fax: +1 508 831 5853. E-mail address: [email protected] (T.A. Camesano). 0927-7765/$ – see front matter © 2008 Elsevier B.V. All rights reserved. doi:10.1016/j.colsurfb.2008.08.013

pounds as energy sources [6]. P. aeruginosa strains cause serious infections, such as in the airway of cystic fibrosis patients [7], ulcerative bacteria keratitis in soft contact lens users [8,9], and can especially cause problems in immunocompromised persons, such as burn victims, cancer patients, and people infected with HIV [10–12]. We selected two strains of P. aeruginosa that differ in the LPS structure, to help us understand the effect of LPS on protein–bacteria interactions. PAO1 is a genetically wellcharacterized serotype O5 wild-type strain [13,14] with a complete LPS structure (i.e. core oligosaccharide, lipid A, O-antigen). A unique feature of its LPS is that it co-expresses two chemically distinct saccharides, termed the A-band and B-band antigens (A+ B+ ) [15]. The A-band is considered a common antigen for P. aeruginosa, while the B-band allows for O-specific serotyping. It is not known conclusively whether the A-band and B-band molecules are present on adjacent LPS molecules [16,17] or on the same LPS molecule [18]. Further, not all LPS molecules of PAO1 will express the Bband at once, with an estimated 8% of core oligosaccharides having attached O-specific polymer [17]. AK1401 (A+ B− ) is a semi-rough mutant strain of PAO1, whose LPS contains only a single repeating unit of the A-band saccharide [19,20]. A-band saccharides are neutral, while B-band molecules have a negative charge, leading to

116

A. Atabek et al. / Colloids and Surfaces B: Biointerfaces 67 (2008) 115–121

differences in the size and charge properties of the LPS of these two bacteria [17,21]. Atomic force microscopy (AFM) has been used in many biological applications to reveal the nanoscale structure of living microbial cells (bacteria, yeast, fungi), map interaction forces at microbial surfaces, monitor conformational changes of individual membrane proteins, and to study molecular recognition [22–24]. AFM tips can also be modified with bacteria and used to probe bacteria interactions with different substrates such as proteins, epithelial cells and biomaterials [25,26]. In addition to the use of biological force probes in bacterial research, such probes have been developed for other types of cells, such as the amoeba Dictyostelium discoideum [27]. While spores of D. discoideum are around 5 ␮m in diameter, vegetative cells can have diameters greater than 11 ␮m [28]. Thus the larger size and softer nature of these cells requires that extra care be taken when preparing and using cellular probes, and some advice along these lines was provided by Benoit and Gaub [27]. For bacteria, which are generally about a micron in diameter, researchers have been effectively able to immobilize cells to probes and have not seen evidence of bacterial cell deformation during the course of a force measurement [29]. In addition to direct force measurements, such as with the AFM, bacterial adhesion can be described with thermodynamic models that account for the interfacial free energy changes needed to bring the bacterium and substrate from their original interfaces (with aqueous media) to the creation of a new bacteria/substrate interface [30]. The Gibbs free energy of adhesion (Gadh ) can be calculated and related to the energy associated with breaking the interface calculated from AFM pull-off forces [31]. Often, when one wishes to calculate bacterial adhesion at the point of contact (assumed to be 0.157 nm), a thermodynamic model is applied to sum the non-specific Lifshitz–van der Waals (LW) and acid/base (AB) interactions between the bacteria, substrate, and suspending media [32–34]. The most widely used approach is the energy balance calculation proposed by van Oss, Chaudhury, and Good (VCG), which is based on calculation of interfacial tensions from measuring contact angles on bacteria and substrate layers with several probe liquids [30,35,36]. The goal of the present study was to determine how the structure of bacterial LPS affects bacterial adhesion with a protein-coated surface. We used direct adhesion force measurements from AFM to explore the fundamental nature of the interaction forces between bacteria and protein, and to discriminate between the two bacterial strains. We chose bovine serum albumin (BSA), a globular protein [37], to represent a protein that binds mainly through non-specific interactions with bacteria [38,39]. BSA is a common surrogate for human albumin, which is present in blood at about 35–50 mg mL−1 [40]. In some cases, bacteria can also interact with a protein due to specific adhesion, when the bacteria surface possesses the appropriate adhesin or ligand molecule to recognize a bioreceptor [41]. The wild-type strain, PAO1 was less able to participate in hydrogen bonding with BSA because the LPS of this strain is mainly comprised of weakly adhesive, neutral sugars. In contrast, the mutant AK1401 exhibited stronger adhesion with BSA. We suggest that coating of biomaterials with BSA may represent a method to prevent the attachment of wild-type bacteria on implanted devices. 2. Experimental methods 2.1. Protein attachment to clean glass slides and characterization Micro cover glass slides (VWR International, West Chester, PA) were cleaned with a 4:1 mixture of H2 SO4 and H2 O2 in acid solution for 25 min, followed by rinse with ultrapure water (Milli-Q water, Millipore, Billerica, MA).

Clean glass slides were immersed in solutions containing 10 mg mL−1 BSA (Sigma–Aldrich, St. Louis, MO) in HEPES/DTT buffer (50 mM HEPES, 110 mM NaCl, 1 mM dl-dithiothreitol (DTT, Sigma–Aldrich, St. Louis, MO), pH 4.5). DTT was added to the buffer to reduce disulfide bonds, prevent protein aggregation, and maximize bonding to the glass slides [42]. Protein attachment to glass slides was not successful at neutral pH, but was successful at pH 4.5, which is the isoelectric point of BSA. However, it was not desirable to perform AFM experiments with bacteria at pH 4.5. We found that after a 4-h incubation period in protein solution at low pH, the glass slides containing attached BSA could be rinsed with physiological HEPES/DTT buffer, and this buffer solution could be used for all subsequent AFM experiments. Protein layers remained stable for several hours. The addition of DTT was necessary to avoid intra-protein multimers [43]. Without DTT, we found that the BSA always formed large and heterogeneous aggregates, preventing us from having a reproducible surface as a starting material. A Zetasizer Nano ZS (Malvern Instruments, Southborough, MA) was used to measure the zeta potentials (␨) of the protein solutions (5 mg mL−1 BSA in physiological HEPES/DTT buffer (50 mM HEPES, 110 mM NaCl, 1 mM DTT, pH 7.4). The average of six measurements was calculated. 2.2. Bacteria culture conditions Two strains of P. aeruginosa, smooth PAO1 (A+ B+ ) and semirough AK1401 (A+ B− ), were provided by Professor Gerald Pier (Channing Laboratory, Department of Medicine, Brigham and Women’s Hospital, Harvard Medical School). Strains were maintained at 4 ◦ C on tryptic soy agar (TSA). Tryptic soy broth (TSB) was used as the liquid growth media. Cultures were grown in TSB to mid-exponential phase at 37 ◦ C (corresponding to an optical density of 0.9 at 600 nm). 2.3. Tip modification for bacteria–protein interaction force measurements For some experiments, bacteria were attached to silicon AFM tips (CSC38-B type; Mikromasch USA, Wilsonville, OR) using a polyl-lysine (PLL) intermediate coating on the tip. Previous researchers have used PLL [25] or poly-d-lysine [29] to immobilize microbial cells on AFM tips. Each cantilever was treated with 0.1% (w/v) PLL (Sigma–Aldrich, St. Louis, MO) for 30 min and dried for 10 min. A micromanipulator was used to hold and lower the AFM tip into bacteria solution (2 ␮L of cell suspension in physiological HEPES buffer, at ∼9 × 109 cells mL−1 ) for 5 min (strain AK1401) or 20 min (strain PAO1). Each cantilever containing bacteria was used to probe BSA that had been deposited on glass slides. Immediately after the attachment process, the functionalized tip was used to probe protein molecules in physiological HEPES/DTT buffer. Scanning electron microscopy (SEM, Jeol JSM-840) was performed on all tips coated with bacteria after AFM measurements to verify the presence and placement of bacteria on the silicon tip. 2.4. AFM force measurements A Dimension 3100 atomic force microscope with Nanoscope IIIa controller (Veeco Metrology, Inc., Santa Barbara, CA) was used with CSC38-B type silicon cantilevers (Mikromasch USA, Wilsonville, OR). The spring constants of unmodified tips were 0.16 ± 0.08 N/m (n = 30), measured using a thermal method [44]. Previous work has shown that the tip must be modified with a much heavier mass in order to modify the spring constants appreciably. For example, Green et al. showed that a tungsten sphere of 7 ␮m (density of

A. Atabek et al. / Colloids and Surfaces B: Biointerfaces 67 (2008) 115–121

tungsten = 19,250 kg/m3 ) modifies the spring constant of an AFM cantilever from 0.020 to 0.019 N/m [45]. For a lighter material, such as silica (density = 2400 kg/m3 ), even a 14-␮m sphere modified the spring constant from 0.020 to 0.017 N/m. Since the bacterial mass added will be much lower than for either of these spheres, we considered the effect of the added mass on the tip spring constant to be negligible. In the force measurements, the force applied to the cantilever was controlled manually to approximately 50 pN, according to a procedure described by Puech et al. [46]. In this case, the set point was adjusted so that it was just sufficient to keep the cantilever engaged with the sample. All experiments were made at ambient temperature (23 ◦ C). The tip speed used was 50 ␮m/s, corresponding to a loading rate in the range of (4–12) × 106 pN/s, considering the range of the spring constants of the AFM tips. The contact time in each force cycle was 1 ␮s. For soft cells with a slow viscoelastic response or under conditions that can lead to plastic deformation, care must be taken to ensure that adhesion force measurements can be separated from viscoelastic information. This was noted by Benoit and Gaub, who observed plastic deformation when performing AFM studies on human endometrial cells [27]. However, for loading rates, tips speeds, and contact times in a range similar to the values we used, Gaboriaud et al. found that bacterial deformation is fully reversible and the dynamics of bacterial response to a force event with the AFM happen much quicker than the timescale of an AFM measurement [47]. Prior to conducting force measurements, the glass slides containing immobilized protein molecules were imaged in physiological HEPES/DTT buffer using contact mode or tapping mode AFM, in order to appreciate the morphology of the adsorbed protein layers (Supplementary Information Figure S1). Preliminary experiments suggested that we could reproducibly image the protein-coated substrate with either tapping or contact mode, and the morphologies appeared the same. Force interactions on BSA layers were measured with a bare silicon probe first. Next, the protein layers were probed with the bacteria-containing AFM probe. In all force measurements, forces were investigated by recording cantilever deflection on approach to the protein-coated substrate and during retraction of the cantilever from the substrate. The deflection voltage separation distance curves were converted into force (nN) vs. separation (nm) curves [48]. At least two modified AFM tips were used and 30 force profiles were captured per experimental condition. The adhesion forces between the substrate and tip (Fadh ) were measured from analysis of the retraction portions of the 30 force cycles. We counted all peaks in the force profiles, regardless of the distance at which they occurred. Since the adhesion force should be interpreted in the context of the distance where the pull-off event occurred, we also counted the pull-off distances (LD ) corresponding to each pull-off event. 2.5. Thermodynamic calculations of interfacial free energies Using the VCG model, the Gibbs free energy change upon bacteria–substrate adhesion in aqueous media, Gadh , was calculated based on the interfacial tensions ( ij ) for bacteria/substrate, bacteria/aqueous media, and substrate/aqueous media [30,35,36], as Gadh = BS − BA − SA

(1)

where B, S, and A correspond to bacteria, substrate (protein layer), and aqueous media (50 mM HEPES, 110 mM NaCl buffer), respectively. The aqueous media has almost the same interfacial tension of water (72 N/m). The interfacial tension of HEPES was previously measured to be 73 N/m, and this value was used in the present calculations [49]. The addition of NaCl at this concentration has a

117

negligible effect on the interfacial tension (∼0.2 N/m) [50], and was therefore neglected. In the system examined here, the bacteria are on the AFM spherical probe, and the substrate is a glass side coated with BSA. In our control experiments, however, there are no bacteria on the AFM probe, and so the Gibbs free energy of adhesion between the silicon probe (P) and the substrate layer (S, which is protein), is calculated according to Gadh = PS − PA − SA

(2)

The methods to calculate the interfacial free energies based on contact angles with three probe liquids have been discussed extensively in the literature [36,51,52]. We have followed the same procedures and model equations described in our previous work [53] to calculate the interfacial free energies and Gibbs free energies of adhesion in the present system. 2.6. Hydrogen bond force calculation Statistical analysis of AFM adhesion events can be used to decouple “specific” and “non-specific” forces, based on characterizing the mean and variance of all adhesion events for all bonds formed [54,55]. According to the conventions of terminology proposed by these researchers, the specific forces would mainly involve hydrogen bonding in a system that has no other ligand/receptor bonds. We have previously used this methodology to calculate the hydrogen bond force for Escherichia coli [56,57]. A linear regression is performed on plot of the mean adhesion force for a given condition (mean Fadh ) vs. the variance of the adhesion force for that condition. The slope of the line corresponds to the specific or hydrogen bond component of the interaction. 3. Results 3.1. Verification of tip modification with bacteria The silicon AFM tips were modified with PLL and coated with bacteria to probe bacteria interactions with protein molecules. Several techniques were used to verify that our tip coating procedure was successful. First, we measured the resonant frequency shift of the cantilever before and after each modification step. The resonant frequency of the cantilevers decreased after each treatment (PLL addition and bacteria coating), due to the increased mass on the cantilevers. Typical values found were 12.77 kHz for the uncoated cantilever, 12.28 kHz for a tip coated with PLL, and 7.55 kHz for a tip coated with PLL and bacteria. We also examined the tips by SEM after the conclusion of each experiment, where multiple bacteria cells could be seen (representative image shown in Fig. 1). Similar data were obtained using replicate probes. In addition, the overall shape of the AFM force profiles was a good indicator of whether bacteria were present. For verification of whether the bacteria were detaching from the AFM tip after a certain period of time, the protein molecules were probed with the modified AFM tip after collecting the force profiles for an extended time. There was one instance where the bacteria detached from the tip after 1 h of scanning, as shown by an absence of interactions between the tip and the protein molecules in the force profiles, and this data set was discarded. When this tip was examined with SEM, no bacteria could be seen. Control experiments were also performed to examine the interactions of a PLL-coated tip with the proteins (Supplementary Information Figure S2). These force cycles showed only short-range forces (decay distances of <100 nm) and could clearly be distinguished from bacteria–protein force profiles.

118

A. Atabek et al. / Colloids and Surfaces B: Biointerfaces 67 (2008) 115–121

Fig. 1. SEM image of bacteria attached to the AFM tip.

3.2. Protein interactions with silicon or bacteria-coated AFM tip Force cycles were captured for BSA in physiological HEPES/DTT buffer, starting with a bare silicon AFM tip as a control probe, and then using tips coated with bacteria. Since the geometry of the tip was different in the control vs. the bacterial experiments, we first considered the normalized force (F/R) for each system (Fig. 2A). PAO1 exhibited very weak normalized adhesion forces with BSA. The normalized adhesion forces between AK1401 and BSA were greater and still present after hundreds of nanometers. BSA had the highest normalized adhesion force with silicon, although all of the interactions ended before ∼150 nm. When we further consider how the two strains behave in comparison to one another, the Fadh values and their distribution can be examined in more detail (Figs. 2B and 3A). In this case, it was unnecessary to normalize Fadh by the radius of interaction, since the two bacteria probes were of equal radii. The AK1401–BSA interactions generally show a single large peak or a double peak, with a force that returns to zero after polymer detachment, while the PAO1–BSA interactions show multiple small and indistinct force peaks (Fig. 2B). The mean Fadh values with BSA were 1.12 nN for AK1401 and 0.40 nN for PAO1 (Fig. 3A), and were statistically different from one another, based on the Kruskal–Wallis one-way analysis of variance (ANOVA) test (P ≤ 0.001). The median Fadh values were 1.18 and 0.40 nN, for AK1401 and PAO1, respectively. The similarity of the mean and median values suggests that the adhesion force distributions are close to symmetrical. BSA interactions extended over a longer range with PAO1 than for AK1401 (Fig. 3B), with mean LD values of 341 and 181 nm, respectively. These pull-off distances were statistically different (ANOVA, P ≤ 0.001). Although the pull-off distances were large, they are consistent with values we have observed for these two P. aeruginosa strains under a variety of conditions and with other types of AFM probes [58,59]. 3.3. Thermodynamic interpretation of interaction forces 3.3.1. Interfacial tensions Based on the measured or known contact angle measurements with the three probe liquids, we calculated the interfacial tensions in terms of their LW and AB components. The LW components of the interfacial tensions for PAO1 and AK4101 were slightly lower

Fig. 2. Representative retraction curves: (A) normalized force for a silicon tip interacting with BSA, compared to a bacteria probe (either PAO1 or AK1401) interacting with BSA layer. The force was divided by the radius in this plot to help compare the effects of a bacteria coated probe with an uncoated probe. For the bacteriacoated AFM tip, we used the bacterial radius (600 nm) for the normalization. For the uncoated AFM tip, we used the nominal tip radius (10 nm). (B) More detailed comparison of AFM retraction curves for PAO1 and AK1401. For comparison of the magnitude of the two bacteria probes with each other, it was not necessary to normalize the forces since the probe sizes were identical. For all cases, 30 force profiles were captured in physiological HEPES/DTT buffer, but only three profiles per condition are shown. Probes prepared on different days and used with separately prepared substrates yielded similar results.

than previous reports (Table 1), which have shown that  LW values for more than 140 bacteria were within 10% of 40 mJ/m2 [34,60]. In our study, the values were very similar for PAO1 and AK1401, and also for BSA. The electron donating component of the interfacial tension for AK1401 was much greater than that of PAO1, with values of 1.5 and 0.2 mJ/m2 , respectively. The acid/base component of the interfacial tension was also greater for AK1401 (18.3 mJ/m2 ) than PAO1 (7.5 mJ/m2 ). Therefore, when the total interfacial tensions were calculated, AK1401 had a higher value than PAO1. 3.3.2. Gibbs free energy change upon protein adhesion Using the interfacial tensions presented in Table 1, Gadh values were calculated for the protein/bacteria or protein/silicon systems in aqueous media. All of the Gadh values were positive (Table 2). Protein/substrate interaction was least favorable for BSA interacting with PAO1, followed by AK1401 and then silicon. When the Gadh values were decoupled into their components, we found that the acid/base interaction was dominant compared to the Lifshitz–van der Waals contribution.

A. Atabek et al. / Colloids and Surfaces B: Biointerfaces 67 (2008) 115–121

119

Table 2 Components of the Gibbs free energy of adhesion, calculated from contact angle measurements

P. aeruginosa PAO1 Probe– BSA P. aeruginosa AK1401 probe–BSA Silicon–BSA

Fig. 3. Adhesive force (A) and pull off distance (B) distributions of protein–bacteria interactions. BSA was probed with P. aeruginosa AK1401 and P. aeruginosa PAO1coated silicon AFM tips in physiological HEPES/DTT buffer.

3.4. Hydrogen bond forces According to the Poisson statistical analysis, the hydrogen bond force for AK1401 interacting with BSA was 0.44 nN, compared to 0.31 nN for PAO1–BSA (Supplementary Information Figure S3). 4. Discussion 4.1. VCG thermodynamic modeling The advantage of employing the VCG thermodynamic model ing approach is that calculated surface free energies can be decou-

AB Gadh (mJ/m2 )

LW Gadh (mJ/m2 )

Gadh (mJ/m2 )

59.20

−0.63

58.58

47.82

−0.29

47.53

25.57

−1.01

27.56

pled into their fundamental components, which allows for a way to distinguish between the interactions of the two bacterial strains with the protein or control surface. The LW components of the interfacial tension were similar for both bacteria strains and for BSA, but differences were noted in the AB contributions. The lower  + value for PAO1 compared to AK1401 suggests that AK1401 has a greater ability to participate in electron acceptor interactions, such as hydrogen bonding. Qualitatively, the direction predicted by the Gadh values agrees with the AFM-measured forces, in that the system with the most adhesion (BSA–silicon) had the least positive Gadh value; and the system with the least adhesion (BSA–PAO1) had the most positive Gadh value. Since all of the Gibbs free energy changes were positive, this calculation helps to show that bacteria do not prefer to be attached to protein, in comparison to the molecular interactions between bacteria–bacteria and protein–protein. While we cannot directly relate this to AFM data, where an adhesion force occurs due to forced mechanical contact and is actually representing a detachment event, we can look at the components of the interfacial and Gibbs free energies to help understand which fundamental forces govern the behavior, and we can look at the relative behavior of the interfacial parameters for the two bacterial strains. In a quantitative sense, there is disagreement between the Gibbs free energy values and the observed phenomena, in that we predict positive values but adhesion still occurs. In part, this is due to the nature of an AFM experiment. During the approach of an AFM tip to the substrate, repulsion always is observed. After contact between the bacterium and opposing surface, there is always one or more local maxima that correspond to adhesion forces. Therefore, AFM experiments measuring microbial adhesion are designed to demonstrate both attraction and repulsion events, as has been noted previously by us and other groups [61,62]. However, others have observed that even systems not making use of AFM are unable to show quantitative agreement between thermodynamic calculations of interaction energies and bacterial attachment behavior. In some experiments with hydrophilic surfaces, lactobacilli still attached to the glass surfaces of a parallel plate flow chamber, even though positive free energy values were predicted [61]. Others have also noted that the complex nature of bacterial-surface

Table 1 Zeta potentials, contact angles, and interfacial tension components for Pseudomonas aeruginosa PAO1 and AK1401, and bovine serum albumin  (mV)

Contact angle (◦ ) W

PAO1a , b AK1401a , b BSAc Silicond

−13.1 −18.9

Parameter value (mJ/m2 ) D

F

−

+

 AB

 total

26.0 23.6

66.8 56.1

0.2 1.5

7.5 18.3

33.5 41.9

35.9 ± 1.6

25.3 ± 7.2

0.2 ± 0.3

4.50

40.5 ± 5.3



31 ± 1 34 ± 1

45 ± 2 52 ± 1

51 ± 1 46 ± 1

58.0 ± 5.0

47.0 ± 3.0

40.0 ± 5.0

LW

All interfacial tensions were calculated based on the contact angle values. a Contact angle measurements for P. aeruginosa PAO1 and AK1401 were measured in the present study. b The zeta potential measurements of the bacteria were also reported in our previous work [58,59]. c Contact angles were taken from the literature for BSA [72]. d Contact angles were taken from the literature for silicon [73]. For this surface only, the probe liquids were water, diiodomethane, and ethylene glycol. It is considered acceptable to substitute formamide and ethylene glycol, as long as a consistent set of parameters is used for the interfacial tension calculations. Therefore, the interfacial tensions reported for silicon are also based on the use of ethylene glycol contact angles.

120

A. Atabek et al. / Colloids and Surfaces B: Biointerfaces 67 (2008) 115–121

interactions cannot be accounted for with the thermodynamic model [63]. 4.2. Nature and role of bacteria LPS on adhesive forces with BSA The interactions between BSA and the molecules in the bacteria LPS can be due to a number of factors, including steric forces, electrostatic double layer interactions, specific binding, as well as LW and AB forces. The adhesion of the mutant AK1401 to BSA was much greater than that of the wild-type strain, PAO1. PAO1 has smooth LPS, and expresses two chemically and antigenically distinct saccharides, termed the A-band and B-band antigens, but also expresses a longer A-band polymer than strain AK1401 [15,17,21]. B-band polymer may only be present in 8% of the core oligosaccharide molecules [17]. The distribution of A-band and B-band on the cell surface is still not known clearly, as to whether the LPS types are randomly distributed or present in distinct domains [64], or if they are on adjacent [16,17] or on the same LPS molecule [18]. AK1401 differs from the parent in that it does not express the B-band Oantigen, and it only expresses one polymer unit (of the A-band type) [19,20]. When extracted, the other components of the LPS of these two strains are similar according to NMR and chemical analysis of the inner core of the LPS molecule for PAO1 and AK1401 [65], and lipid A is also known to be conserved among bacteria strains. However, due to the covering effect of the B-band and also due to the longer repeating A-band polymer on PAO1, the inner core and lipid A regions of AK1401 are more exposed to the surface. In previous work, the exposure of inner core regions on AK1401 helped explain the adhesion forces between PAO1 and AK1401 and various sources of natural organic matter [58]. Hydrogen bond forces were greater for AK1401 interacting with BSA than for PAO1 interacting with the protein, according to the Poisson statistical analysis. Also, based on the thermodynamic modeling, AB interactions appear to dominate the adhesion behavior of PAO1 and AK1401 with BSA, which also is a manner of expressing that hydrogen bonding is dominant. In a previous AFM study, SAMs with different terminal groups were functionalized onto an AFM tip, and the adhesion forces with BSA were measured [66]. The authors found that the adhesion forces were high between BSA and SAMs terminating in CH3 (4.5 nN), NH2 (0.8 nN), and OH (0.3 nN). No adhesion was observed between COOH-terminated SAMs and BSA. High adhesion between amino acids and CH3 -SAMs was attributed to strong hydrophobic interactions. In our study, the core oligosaccharide (for both strains) and the O-antigen (for PAO1) both have sugars rich in CH3, such as l-alanine, l-rhamnose, and N-acetyl-d-fucosamine, and the core contains phosphate [65]. The core oligosaccharide and O-antigen also contain amino derivatives of simple sugars, such as d-galactosamine [67] and 2-amino-2deoxy-d-galactose [65]. The molecules in the A-band also contain rhamnose sugars (with a terminal CH3 group), and neutral polysaccharides with mainly OH groups, such as ribose and glucose, but do not contain any amino sugars. In our conceptual model, when the BSA layer is contacted by the AK1401 probe, the BSA sees mainly the effect of the core and lipid A region. Amino and methyl groups from the core, along with the phosphate from the lipid A, have a strong interaction with BSA, and the result is that high adhesion forces are observed. When BSA is contacted by the PAO1 probe, the longer A-band polymer appears to block the interaction with the core and lipid A. Therefore, BSA perceives mainly the neutral sugars of the A-band polymer, and the adhesion forces are weak. This seems reasonable considering the previous AFM studies on BSA [66], and also considering that serum albumin is the principle carrier of fatty acids and serves as a transport protein [68], so it has a high affinity for lipid molecules [69,70]. We might have expected the interaction of BSA with PAO1

to be stronger because of the presence of B-band polymers, which have CH3 and NH2 groups. However, since only a small number of LPS molecules are actually expressing the B-band, and we are using a very large AFM probe with multiple bacteria on it, we apparently did not detect the effect of these functional groups from the B-band. Although we cannot be certain that hydrogen bonds are the only type of interaction occurring, some evidence suggests that other types of forces can be discounted in this system. For example, electrostatic interactions do not explain the trends we saw in adhesion with BSA. BSA is negatively charged at this pH and AK1401 is more negatively charged than PAO1 [58,64]. Therefore, if electrostatic interactions were dominant, we would have expected less adhesion between AK1401 and BSA. Besides, our experiments were conducted at a high ionic strength and so it is likely that electrostatic interactions were suppressed under these conditions. It is also interesting that AK1401 presents a more electrostatically negative surface, according to the zeta potential (Table 1), even though AK1401 has only neutral O-antigen polysaccharides on its surface at microscopic scale. PAO1, with negatively charged polymers in its O-antigen, has a less electrostatically negative surface than AK1401. Since the core-lipid A region of strain AK1401 is more exposed, negatively charged carboxyl groups from the core and lipid A regions might be more important than negatively charged Bband molecules and these groups might influence the interactions of strain AK1401 with BSA by binding to NH2 groups in the protein molecules. The low density of B-band polymers on PAO1 can also help explain this result. Steric interactions also can be ruled out as having a large role in determining the interactions with BSA in this study. Unlike other AFM studies, where we have used a silicon or silica probe to measure forces on P. aeruginosa PAO1 and AK1401 [58,59], the bacteria in this system were immobilized on an AFM tip. We have many bacteria on the tip and they are in close proximity to one another. We think that the constraints of this geometry prevented us from observing any long-range repulsion in AFM approach curves (not shown), which have been attributed to steric forces in our prior work. However, in other types of experimental systems, researchers have noted the importance of electrostatic and steric factors, as well as biological factors such as bacterial cell motility, in influencing the deposition behavior of P. aeruginosa to an alginate-covered surface [71]. 5. Conclusions In summary, we found that hydrogen bonds control the association between P. aeruginosa and protein-coated surfaces. The absence of B-band saccharides and the shorter A-band unit on strain AK1401 allows for the lipid A and core region to be more exposed than in the parent strain, PAO1. The lipid A and core region have strong affinity for BSA due to hydrogen bonding. We did not find that electrostatic or steric interactions were dominant in controlling P. aeruginosa interactions with BSA, although those factors can be important in other study designs or applications. Since P. aeruginosa generally consists of neutral polysaccharides in its LPS, which interact weakly with BSA, these results suggest that treatment of a surface with BSA may inhibit bacterial colonization with wild-type P. aeruginosa. Acknowledgements This research was supported by National Science Foundation (BES-0238627). Several researchers provided experimental assistance and helpful discussions throughout the course of these

A. Atabek et al. / Colloids and Surfaces B: Biointerfaces 67 (2008) 115–121

studies, including Jochen Weiss, Thrandur Helgason, Satya S. Shivkumar, Xiaoshu Dai, José Argüello, Laila Abu-Lail, Ray J. Emerson IV, and Joshua Strauss. Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at doi:10.1016/j.colsurfb.2008.08.013. References [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14]

[15] [16] [17] [18] [19] [20] [21] [22] [23] [24] [25] [26] [27] [28] [29] [30] [31] [32]

A.G. Gristina, Science 237 (1987) 1588–1595. P. Brokke, J. Dankert, J. Carballo, J. Feijen, J. Biomater. Appl. 5 (1991) 204–226. S.O. Heard, Ann. Acad. Med. Singapore 30 (2001) 419–429. F. Boulmedais, B. Frisch, O. Etienne, P. Lavalle, C. Picart, J. Ogier, J.-C. Voegel, P. Schaaf, C. Egles, Biomaterials 25 (2004) 2003–2011. B. Gottenbos, H.C. Van der Mei, F. Klatter, P. Nieuwenhuis, H.J. Busscher, Biomaterials 23 (2002) 1417–1423. J.B. Lyczak, C.L. Cannon, G.B. Pier, Microbes Infect. 2 (2000) 1051–1060. A. Bush, Paediatr. Respir. Rev. 3 (2002) 82–88. X. Wu, S.K. Gupta, L.D. Hazlett, Curr. Eye Res. 14 (1995) 969–977. S.A. Klotz, R.P. Misra, S.I. Butrus, Cornea 9 (1990) 266–270. G.P. Bodey, R. Bolivar, V.J. Fainstein, L. Jadeja, Rev. Infect. Dis. 5 (1983) 279–313. I. Chatzinikolaou, D. Abi-Said, G.P. Bodey, K.V.I. Rolston, J.J. Tarrand, Arch. Intern. Med. 160 (2000) 501–509. M.G. Schuster, A.H. Norris, AIDS 8 (1994) 1437–1441. M. Fletcher, in: M. Fletcher (Ed.), Bacterial Adhesion: Molecular and Ecological Diversity, Wiley-Liss, Inc., New York, 1996. C.K. Stover, X.Q. Pham, A.L. Erwin, S.D. Mizoguchi, P. Warrener, M.J. Hickey, F.S.L. Brinkman, W.O. Hufnagle, D.J. Kowalik, M. Lagrou, R.L. Garber, L. Goltry, E. Tolentino, S. Westbrock-Wadman, Y. Yuan, L.L. Brody, S.N. Coulter, K.R. Folger, A. Kas, K. Larbig, R. Lim, K. Smith, D. Spencer, G.K.S. Wong, Z. Wu, I.T. Paulsen, J. Reizer, M.H. Saier, R.E.W. Hancock, S. Lory, M.V. Olson, Nature 406 (2000) 959–964. R.E. Hancock, A.M. Carey, J. Bacteriol. 140 (1979) 901–902. M.Y. Lam, E.J. McGroarty, A.M. Kropinski, L.A. MacDonald, S.S. Pedersen, N. Hoiby, J.S. Lam, J. Clin. Microbiol. 27 (1989) 962–967. M. Rivera, L.E. Bryan, R.E. Hancock, E.J. McGroarty, J. Bacteriol. 170 (1988) 512–521. K. Hatano, J.B. Goldberg, G.B. Pier, J. Bacteriol. 175 (1993) 5117–5128. D. Berry, A.M. Kropinski, Can. J. Microbiol. 32 (1986) 436–438. J. Lightfoot, J.S. Lam, J. Bacteriol. 173 (1991) 5624–5630. M. Rivera, E.J. McGroarty, J. Bacteriol. 171 (1989) 2244–2248. P. Hinterforfer, Y.F. Dufrêne, Nat. Methods 3 (2006) 347–355. F. Gaboriaud, S. Bailet, E. Dague, F. Jorand, J. Bacteriol. 187 (2005) 3864–3868. R. Yongsunthon, V.G.J. Fowler, B.H. Lower, F.P.I. Vellano, E. Alexander, L.B. Reller, G.R. Corey, S.K. Lower, Langmuir 23 (2007) 2289–2292. A. Touhami, M.H. Jericho, J.M. Boyd, T.J. Beveridge, J. Bacteriol. 188 (2006) 370–377. T.L. Cail, M.F. Hochella, Geochim. Cosmochim. Acta 69 (2005) 2959–2969. M. Benoit, H.E. Gaub, Cells Tissues Organs 172 (2002) 174–189. J.T. Bonner, E.B. Frascella, Biol. Bull. 104 (1953) 297–300. S.K. Lower, C.J. Tadanier, M.F. Hochella Jr., Geochim. Cosmochim. Acta 64 (2000) 3133–3139. C.J. van Oss, Interfacial Forces in Aqueous Media, Marcel Dekker, New York, 1994. E.R. Beach, W.D. Tormoen J., J. Adhes. Sci. Technol. 16 (2002) 845–868. G. Chen, K.A. Strevett, Environ. Microbiol. 2001 (2001) 237–245.

121

[33] B.F. Smets, D. Grasso, M.A. Engwall, B.J. Machinist, Colloids Surf. B 14 (1999) 121–139. [34] P.K. Sharma, K.H. Rao, Adv. Colloid Interf. Sci. 98 (2002) 341–463. [35] C.J. van Oss, R.J. Good, M.K. Chaudhury, J. Colloid Interf. Sci. 111 (1986) 378–390. [36] C.J. van Oss, M.K. Chaudhury, R.J. Good, Chem. Rev. 88 (1988) 927–941. [37] T.J. Peters, All about Albumin: Biochemistry, Genetics and Medical Applications, Academic Press Inc., Orlando, 1996. [38] M.A. Lantz, S.P. Jarvis, H. Tokumoto, T. Martynski, T. Kusumi, C. Nakamura, J. Miyake, Chem. Phys. Lett. 315 (1999) 61–68. [39] J. Shuter, V.B. Hatcher, F.D. Lowy, Infect. Immun. 64 (1996) 310–318. [40] M.I. Ferguson, E.M. Scott, P.S. Collier, APMIS 100 (1992) 876–882. [41] R.A. Hull, R.E. Gill, P. Hsu, B.H. Minshew, S. Falkow, Infect. Immun. 33 (1981) 933–938. [42] W.W. Cleland, Biochemistry 3 (1964) 480–482. [43] A.S. Determan, B.G. Trewyn, V.S.-Y. Lin, M. Nilsen-Hamilton, B. Narasimhan, J. Control. Release 100 (2004) 97–109. [44] N.A. Burnham, X. Chen, C.S. Hodges, G.A. Matei, E.J. Thoreson, C.J. Roberts, M.C. Davies, S.J.B. Tendler, Nanotechnology 14 (2003) 1–6. [45] C.P. Green, H. Lioe, J.P. Cleveland, R. Proksch, P. Mulvaney, J.E. Sader, Rev. Sci. Instrum. 75 (2004) 1988–1996. [46] P.-H. Puech, K. Poole, D. Knebel, D.J. Muller, Ultramicroscopy 106 (2006) 637–644. [47] F. Gaboriaud, M.L. Gee, R. Strugnell, J.F.L. Duval, Langmuir, in press. [48] W.A. Ducker, T.J. Senden, R.M. Pashley, Langmuir 8 (1992) 1831–1836. [49] A.P.J. Middleberg, L. He, A.F. Dexter, H.-H. Shen, S.A. Holt, R.K. Thomas, J. R. Soc. Interf. 5 (2008) 47–54. [50] J.J. Jasper, J. Phys. Chem. Ref. Data I (1972) 841–1010. [51] C. Della Volpe, S. Siboni, J. Colloid Interf. Sci. 195 (1997) 121–136. [52] C. Della Volpe, S. Siboni, D. Maniglio, M. Morra, C. Cassinelli, M. Anderle, G. Speranza, R. Canteri, C. Pederzolli, G. Gottardi, S. Janikowska, A. Lui, J. Adhes. Sci. Technol. 17 (2003) 1425–1456. [53] Y. Liu, J. Strauss, T.A. Camesano, Langmuir 23 (2007) 7134–7142. [54] J.M. Williams, T. Han, T.P. Beebe Jr., Langmuir 12 (1996) 1291–1295. [55] T. Han, J. Williams, T.P. Beebe Jr., Anal. Chim. Acta 307 (1995) 365–376. [56] N.I. Abu-Lail, T.A. Camesano, Langmuir 22 (2006) 7269–7301. [57] N.I. Abu-Lail, T.A. Camesano, Langmuir 24 (2008) 4420. [58] L.I. Abu-Lail, A. Atabek, T.A. Camesano, Environ. Sci. Technol. 41 (2007) 8031–8037. [59] A. Atabek, Investigating bacterial outer membrane polymers and bacterial interactions with organic molecules using atomic force microscopy, Chemical Engineering M.S. Thesis, Worcester Polytechnic Institute, Worcester, MA, 2006. [60] H.C. van der Mei, R. Bos, H.J. Busscher, Colloids Surf. B 11 (1998) 213–221. [61] V. Vadillo-Rodriguez, H.J. Busscher, H.C. van der Mei, J. de Vries, W. Norde, Colloids Surf. B 41 (2005) 33–41. [62] J.M. Pouliot, I.N.-P. Walton, M.L.I. Abu-Lail, T.A. Camesano, Biomacromolecules 6 (2005) 1122–1131. [63] G.S.K.A. Chen, J. Colloid Interf. Sci. 261 (2003) 283–290. [64] S.A. Makin, T.J. Beveridge, Microbiology 142 (1996) 299–307. [65] I. Sadovskaya, J.R. Brisson, J.S. Lam, J.C. Richards, E. Altman, Eur. J. Biochem. 255 (1998) 673–684. [66] S. Kidoaki, T. Matsuda, Langmuir 15 (1999) 7639–7646. [67] Y.A. Knirel, E.V. Vinogradov, N.A. Kocharova, R.A. Paramonov, N.K. Kochetkov, B.A. Dmitriev, E.S. Stanislavsky, B. Lanyi, Acta Microbiol. Hung. 35 (1988) 3–24. [68] S. Curry, H. Mandelkow, P. Brick, N. Franks, Nat. Struct. Biol. 5 (1998) 827–835. [69] G.-L. Friedli, Interaction of SWP with bovine serum albumin (BSA). PhD Dissertation, University of Surrey, Guildford, 1996. [70] J. Grdadolnik, Internet J. Vib. Spectro. 6 (2002) 6. [71] A.J.E.M. de Kerchove, Appl. Environ. Microbiol. 73 (2007) 5227–5234. ´ [72] T. Białopiotrowicz, B. Janczuk, J. Surfact. Deterg. 4 (2001) 287–292. [73] E.M. Harnett, J. Alderman, T. Wood, Colloids Surf. B: Biointerf. 55 (2007) 90–97.