ATR-FTIR study of lipopolysaccharides at mineral surfaces

Available online at www.sciencedirect.com

Colloids and Surfaces B: Biointerfaces 62 (2008) 188–198

ATR-FTIR study of lipopolysaccharides at mineral surfaces Sanjai J. Parikh a , Jon Chorover b,∗ a

b

Department of Plant and Soil Sciences, University of Delaware, Newark, DE 19716, United States Department of Soil, Water and Environmental Science, The University of Arizona, Tucson, AZ 85721, United States Received 7 September 2007; received in revised form 27 September 2007; accepted 1 October 2007 Available online 5 October 2007

Abstract Lipopolysaccharides (LPS) are ubiquitous in natural aqueous systems because of bacterial cell turnover and lysis. LPS sorption and conformation at the mineral/water interface are strongly influenced by both solution and surface chemistry. In this study, the interaction of LPS with various surfaces (ZnSe, GeO2 , ␣-Fe2 O3 , ␣-Al2 O3 ) that vary in surface charge and hydrophobicity was investigated using attenuated total reflectance Fourier transform infrared (ATR-FTIR) spectroscopy. The presence of Ca2+ (versus Na+ ) in LPS solutions resulted in aggregate reorientation and increased sorptive retention. ATR-FTIR spectra of Na-LPS systems are consistent with reduced surface affinity and are similar to those of solution phase LPS. Ca-LPS spectra reveal hydrophobic interactions of the lipid A region at the ZnSe internal reflection element (IRE). However, pH-dependent charge controls Ca-LPS sorption to hydrophilic surfaces (GeO2 , ␣-Fe2 O3 , and ␣-Al2 O3 ), where bonding occurs principally via O-antigen functional groups. As a result of accumulation at the solid–liquid interface, spectra of Ca-LPS represent primarily surface-bound LPS. Variable-angle ATR-FTIR spectra of Ca-LPS systems show depth-dependent trends that occur at the spatial scale of LPS aggregates, consistent with the formation of vesicular structures. Published by Elsevier B.V. Keywords: LPS; ATR-FTIR spectroscopy; Depth profiling; Endotoxins

1. Introduction Bacterial adhesion to surfaces is a complex function of the full array of macromolecules resident on the cell surface (e.g., LPS, EPS, teichoic acids, surface proteins, flagella), substratum surface chemistry (e.g., hydrophobicity, surface charge), aqueous environmental conditions (e.g., pH, ionic strength), and the distribution and composition of conditioning films. No single factor exerts full control, and therefore deconvolution of various factors requires a model systems approach. In the case of Gram negative bacteria, the surface interaction of free- and membrane-bound lipopolysaccharides (LPS) certainly represents one of the important molecular-level controls over bacterial adhesion [1–4]. LPS are amphiphilic molecules with a hydrophobic lipid A region embedded in the outer membrane of Gram negative bacteria [5]. Beyond the lipid A is a “core sugar” region, and the O-antigen (Fig. 1). The portion of the molecule comprising the O-antigen is present in “smooth” LPS, whereas it is absent from “rough”

∗

Corresponding author. Tel.: +1 520 626 5635; fax: +1 520 621 1647. E-mail address: [email protected] (J. Chorover).

0927-7765/$ – see front matter. Published by Elsevier B.V. doi:10.1016/j.colsurfb.2007.10.002

LPS. The O-antigen is hydrophilic and extends outward from the intact cell into aqueous solution. It is composed of 20–70 repeating units of three to five sugars [5]. Some bacteria, such as Pseudomonas aeruginosa, possess LPS with O-antigens extending up to 40 nm from the cell surface [6]. Since cell turnover and lysis results in the presence of both “cell-bound” and “free” LPS in natural aquatic systems [7], LPS may promote bacterial adhesion by sorption of either free LPS molecules to surfaces during conditioning film formation, or through cell adhesion mediated by membrane bound LPS [1,3]. It has been suggested that during cell adhesion to negatively charged surfaces, the O-antigen may extend beyond the electrostatic energy barrier and become adsorbed in a secondary minimum in close proximity to the surface [3]. Adhesion of both rough and smooth LPS has been observed to occur on metal oxides [2,8], crystalline calcium silicate hydrate [9,10], ZnSe [11], GeO2 , positively charged lipids and polymers [12], and to bovine lung and tracheal tissue samples [13]. LPS of P. aeruginosa ser 10 LPS (in ultrapure water) was found to bond more strongly to positively charged (aminopropyltriethoxysilane polymers) than to hydrophilic (GeO2 ) or hydrophobic (dipalmitoylphosphatidic acid monolayer) surfaces [12].

S.J. Parikh, J. Chorover / Colloids and Surfaces B: Biointerfaces 62 (2008) 188–198

189

Fig. 1. Schematic diagram of (a) smooth lipopolysaccharide and (b) lipid A [5]. Republished with permission from Parikh and Chorover [11].

Adsorption of free LPS may be mediated by functional groups associated with either hydrophilic or hydrophobic portions of the molecule. However, in free LPS, exposure of the lipid A is limited by LPS amphiphilic properties that promote intermolecular associations and the formation of supramolecular structures above a critical aggregation concentration (CAC) [14–17]. Dynamic light scattering measurements indicate that LPS aggregate sizes (4 mg mL−1 LPS, I of 10 mM, pH 6) range from 325 to 400 nm for Na-LPS and from 400 to 475 nm for Ca-LPS [18]. LPS aggregates have been used above the critical aggregate concentration (CAC) to represent cell-bound forms under the assumption that only the O-antigen is exposed for interaction with environmental surfaces. For example, Jucker et al. [2] measured the adsorption of phosphate-buffered LPS aggregates in various ionic strengths and electrolytes (NaCl, KH2 PO4 , K2 HPO4 ) to surfaces of TiO2 , Al2 O3 , and SiO2 . They found greater adhesion to TiO2 and Al2 O3 surfaces. In some cases irreversible adhesion was observed, particularly for LPS with long O-antigen regions. However, the possible surface interactions of monomeric LPS (in thermodynamic equilibrium with aggregates) and/or the potential restructuring of LPS aggregates that may occur upon association with a surface were not investigated. The relation between surface hydrophilicity and LPS structure also plays a role. For example, P. aeruginosa (PAO1) cells with primarily long O-antigen preferentially adhere to hydrophilic surfaces, whereas cells with shorter O-antigen have a higher affinity for hydrophobic surfaces [1]. Thus, the capability of a cell to mediate O-antigen length might confer a capacity to influence adhesion in dynamic environments. However, adhe-

sion of free-LPS to surfaces may be quite different, particularly if LPS aggregates are disrupted and interaction between the lipid A region and a substratum is favorable. Toward the goal of building a molecular-level understanding of initial bacterial cell adhesion at mineral surfaces, the current work involves isolation of the LPS component for in-situ spectroscopy studies. This approach allows us to investigate the influence of substratum surface composition (charge and hydrophobicity), solution chemistry, and LPS aggregation. Specifically, it has been suggested that an increase in LPS aggregation [19–22] might result when Ca2+ – rather than Na+ – is present because of the high stability of Ca2+ complexation with phosphate moieties of the lipid A region. If so, that would be expected to affect surface interactions of the macromolecules [14,19,23,24]. Our previous studies using attenuated total reflectance (ATR) Fourier transform infrared (FTIR) spectroscopy do indeed provide direct evidence of Ca2+ binding to LPS phosphate groups, but also suggest a more complex effect on aggregate structure [11]. The spectra showed increased intensity of phosphate and fatty acid absorbances relative to carbohydrate for Ca-LPS versus Na-LPS samples. These results suggested that Ca2+ ion bonding to LPS phosphate groups in the lipid A region resulted in disruption of LPS aggregates. Our data support the hypothesis of Wang et al. that Ca2+ may disrupt LPS aggregates causing reorientation on calcium silicate hydrate surfaces [9]. Therefore, the objective of this work was to examine the LPS aggregation and reorientation in the presence of surfaces with varying surface chemistry.

190

S.J. Parikh, J. Chorover / Colloids and Surfaces B: Biointerfaces 62 (2008) 188–198

Table 1 pH-dependent charge and hydrophobicity of LPS, IREs, and metal oxide coatings Sample

LPS ZnSe GeO2 ␣-Fe2 O3 ␣-Al2 O3

3.1. LPS preparation

Surface charge and hydrophobicity pH 3

pH 6

pH 9

Hydrophobicity

Neutral-negative Positive Positive Positive Positive

Negative Negative Negative Positive Positive

Negative Negative Negative Negative Neutral

Amphiphilic Hydrophobic Hydrophilic Hydrophilic Hydrophilic

2. Research approach Using attenuated total reflectance ATR-FTIR spectroscopy we probed the behavior of LPS in Na+ (Na-LPS) and Ca2+ (CaLPS) solutions (I = 10 mmol L−1 , pH 3, 6, 9) in contact with ZnSe, GeO2 , ␣-Fe2 O3 , and ␣-Al2 O3 surfaces. These surfaces exhibit a range in surface functional group composition, surface charge properties and hydrophobicity (Table 1). ATR-FTIR spectra report on infrared-absorbing moieties at the liquidIRE interface. An evanescent wave propagates ca. 102 –103 nm (depending on crystal type, incident beam angle, and wavelength) beyond the interface of the IRE and into an aqueous suspension. Mean beam penetration depth (dp ) varies according to Eq. (1): dp =

3. Experimental methods

λ 1/2

2πn1 [(sin2 θeff ) − (n2 /n1 )2 ]

(1)

where λ is wavelength (cm) of incident radiation, n1 and n2 are the refractive indices (RI) for the IRE and sample, respectively, and θ eff is the effective angle of incidence [25]. The RI values for the ZnSe and GeO2 IRE used in the present study are 2.4 (nZnSe ) and 4.0 (nGe ). The RI range for LPS (nLPS ) likely falls in the range reported for lipid A (1.50 for pure lipid A; 1.33 with 90% water) [26], bacterial cells (1.38) [27,28], and proteins (1.5) [29]. The limited wave penetration depth and the fact that the wave intensity decays exponentially with distance from the crystal-solution interface [25] indicate that ATR spectra are biased toward molecules in close proximity to the interface. As a result of θ eff -dependence (Eq. (1)), variable angle (V) ATR-FTIR permits depth profiling of samples. By systematically varying θ eff , dp can be varied over length scales varying up to hundreds of nanometers (Table 2), which is comparable to the size of LPS aggregates as measured by dynamic light scattering [16–18]. In the present study, depth profiling of Na-LPS and CaLPS on GeO2 and ZnSe IREs was conducted via VATR-FTIR spectroscopy. For the VATR cell used in this study (ATRMAX II Variable Angle Horizontal ATR Accessory, PIKE Technologies, Inc.), θ eff was determined according to Eq. (2) [30]:   −1 sin(θfix − θvar ) (2) θeff = θfix − sin n1 where θ fix is the angle of the crystal face (45◦ ), and θ var is the scale angle set on the VATR accessory. The effect of variation in the dp as a function of wavenumber, θ eff , IRE composition is shown in Table 2.

A single batch of freeze-dried P. aeruginosa serotype 10 LPS (batch 123K4144; Sigma Inc.) was used for all experiments. For FTIR studies, LPS solutions were prepared above the critical aggregate concentration [11] by dissolving 4.0 mg of freezedried LPS in 1.0 g of NaCl or CaCl2 electrolyte solution at an ionic strength (I) of 10 mmol L−1 and pH was adjusted to 3, 6, and 9 using 0.01M HCl or 0.01M NaOH. Samples were vortexed, sonicated for 10 min, and stored overnight at 4 ◦ C prior to reequilibration the following day at room temperature. 3.2. Metal oxides Colloidal alumina (␣-Al2 O3 ) and hematite (␣-Fe2 O3 ) solids were used to coat ATR crystals for FTIR experiments. The ␣Al2 O3 was obtained from Alfa Aesar, who also provided data on mean particle size (1 ␮m) and specific surface area (6–8 m2 g−1 ). A detailed description of the colloidal hematite (␣-Fe2 O3 ) synthesis, based on the method of Schwertmann and Cornell [31], is given in Parikh and Chorover [32]. Ge powder (Sigma–Aldrich) was analyzed as an analogue for the Ge IRE to assess the extent to which the Ge crystal is surface oxidized and (hydr)oxylated. Diffuse reflectance infrared (DRIFT) spectroscopy and Xray diffraction (XRD) were carried out on solids to confirm composition. For DRIFT, samples were diluted with KBr to approximately 10% (w/w) by first gently mixing 39 mg of sample with 30 mg of KBr for 40 s, and then folding in an additional 390 mg of KBr to homogenize the samples. DRIFT spectra were collected using a Nicolet 560 Magna IR spectrometer (Madison, WI) with 400 scans at 4 cm−1 resolution. X-ray diffraction (XRD) patterns were collected using a Scintag XDS 2000 (Scintag, Inc.) with a Cu X-ray source (40 kV and 40 mA), scan speed of 2◦ 2θ min−1 , and a step width of 0.03◦ 2θ. Transmission electron microscopy (TEM) analysis was carried out on samples on 200 mesh copper grids. Carbon coated mica was placed into a drop of suspension and carbon/colloids were floated onto the grid. Samples were observed a 60 kV with a Japanese Electron Optical Laboratories JEM-100CX II electron microscope. 3.3. ATR-FTIR spectroscopy and analysis FTIR spectra were collected using a Nicolet 560 Magna IR spectrometer (Madison, WI). Spectra of LPS were collected using both 45◦ ZnSe and Ge IREs (Spectra-Tech ARK ATR cell). The Ge IRE surface, which is oxidized and (hydr)oxylated upon contact with oxygenated aqueous solution [33], is referred to as GeO2 in this paper (also see Section 4.1 and Fig. 2c). Metal oxide IRE coatings were made by drying the appropriate suspension (6 mL of ␣-Fe2 O3 , 1.96 g L−1 , pH 4; 1 mL ␣-Al2 O3 , 25 g/L, pH 6) on the ZnSe IRE overnight under vacuum (10 mmHg). Spectra of dry metal oxide films were acquired to determine consistency of colloidal coatings and to provide a means for subtraction of their contribution to spectra collected

S.J. Parikh, J. Chorover / Colloids and Surfaces B: Biointerfaces 62 (2008) 188–198

191

Table 2 Depth of penetration at wavenumbers pertinent to LPS samples for ATR-FTIR with ZnSe (refractive index = 2.4) and GeO2 (refractive index = 4.0) IRE Scale angle (◦ )

ZnSe Effective angle (◦ )

30 37 45 52 60 a

38.8 42.9 45.0 47.1 51.2

GeO2 (nm)a

for Depth of penetration selected wavenumbers (cm−1 ) 2920

1240

1060

911 622 552 502 432

2145 1466 1299 1183 1017

2509 1715 1520 1384 1189

Effective angle (◦ )

41.3 43.0 45.0 46.8 48.7

Depth of penetration (nm)a for selected wavenumbers (cm−1 ) 2920

1240

1060

242 232 221 212 204

570 545 520 500 481

667 638 608 585 562

The refractive index of LPS was assumed to be close to that of bacterial samples, and therefore a value of 1.38 was used [27,60].

subsequently. For experiments utilizing the ARK ATR accessory, a 1 mL aliquot of LPS solution (4 mg mL−1 ) was deposited on the appropriate IRE and oxide-coated IRE for each of the solution chemistries discussed above. Spectra were collected at 0, 15, 30, 60, and 120 min after introduction of LPS solution into the ATR cell. VATR-FTIR measurements were made using an ATRMAX II Variable Angle Horizontal ATR Accessory (PIKE Technologies, Inc.) with a 45◦ ZnSe or Ge IRE. Spectra were acquired at scale angles of 30◦ , 37◦ , 45◦ , 52◦ , and 60◦ , and as a function of time (0, 15, 30, 45, 60, 90, and 120 min) after sample (0.6 mL of 4 mg mL−1 LPS, 10 mM, pH 6 in both NaCl and CaCl2 ) introduction. All FTIR spectra presented are averages of 400 scans at a 4 cm−1 resolution (collection time: 495 s) using the corresponding LPS-free electrolyte solution as background. Peak locations were verified via second derivative analysis and peak areas were determined via curve fitting using Grams/AI software (Salem, NH). In some cases, spectral areas of Lorentzian fitted peaks were used in quantitative analysis and peak area ratios.

4.2. ATR-FTIR spectra of interfacial LPS

Fig. 2. DRIFT FTIR spectra of (a) synthesized ␣-Fe2 O3 , and purchased (b) ␣-Al2 O3 and (c) GeO2 .

Distinct differences between ATR-FTIR spectra for LPS in NaCl and CaCl2 were observed for all surfaces studied (Figs. 3–6). IR band assignments are based on [11] and references cited therein. In that work it was shown that spectra for Ca-LPS solutions on a ZnSe surface show time-dependent changes (increased spectral absorbance for 120 min) whereas this was not observed beyond 15 min for Na-LPS samples (i.e., systems reached apparent equilibrium within this time) [11]. A similar, but less dramatic, time-dependency is observed for LPS samples on the GeO2 IRE. However, spectra for Na-LPS and Ca-LPS samples in contact with ␣-Fe2 O3 and ␣-Al2 O3 suggest equilibration (no further changes in spectra) within 15 min (data not shown). Visual examination of spectra allows for qualitative analysis of trends regarding contributions from fatty acid [νas (CH2 )], phosphate [νas (PO2 − )], and carbohydrate [ν(C–O, C–O–C)] moieties. Spectra of Na-LPS show some small effects of surface composition, but strong similarities overall. Typically Na-LPS spectra contain a single broad peak corresponding to [νas (PO2 − ); 1220–1260 cm−1 ] and a relatively large peak in the carbohydrate (O-antigen) region [ν(C–O, C–O–C); ∼1060 cm−1 ]. Fatty acid [(νas (CH2 ); i.e., lipid A)] contributions, relative to νas (PO2 − ) and ν(C–O, C–O–C), are greatest in Na-LPS spectra collected

4. Results 4.1. Metal oxide characterization The DRIFT spectrum of Ge powder purchased from Sigma–Aldrich indicates the presence of both hydroxide and oxide groups (Fig. 2c). A sizable peak at 3357 cm−1 corresponds to ν(OH) and peaks between 600 and 1000 cm−1 result from ν(Ge–O) [34,35]. The identity of colloidal ␣-Al2 O3 was confirmed via XRD. Synthesis of colloidal ␣-Fe2 O3 was confirmed via XRD, TEM, and DRIFT analysis. XRD and TEM have been published previously [32]. DRIFT spectra (Fig. 2a) show prominent peaks at 454 and 458 cm−1 corresponding to the Fe–O vibrations of hematite [36]. DRIFT analysis of the ␣Al2 O3 (Fig. 2b) reveal a broad peak between 560 and 790 cm−1 and additional peaks between 375 and 520 corresponding to Al–O vibrations of corundum [37,38].

192

S.J. Parikh, J. Chorover / Colloids and Surfaces B: Biointerfaces 62 (2008) 188–198

values of νas (CH2 ), νas (PO2 − ), and ν(C–O, C–O–C) for Ca-LPS and Na-LPS on the four different surfaces. With the exception LPS on ZnSe (Fig. 7a), there is a general trend of decreasing absorbance with increasing pH (particularly from pH 6–9). LPS spectra acquired on ␣-Fe2 O3 contain peaks at ∼1407 cm−1 (Fig. 5) corresponding to νs (COO− ). In LPS spectra collected on ␣-Al2 O3 (Fig. 6), a less prominent peak at 1417 cm−1 is also assigned to νs (COO− ). LPS samples on the hydrophobic (ZnSe) surface exhibit a distinct difference in IR absorbance based on the background electrolyte (Fig. 7a), with increased absorbance intensity for Ca-LPS samples. On the hydrophilic surfaces (GeO2 , ␣-Fe 2 O3 , and ␣-Al2 O3 ; Fig. 7b–d) no distinguishable effect of background electrolyte on IR absorbance is observed. It is important to note that the Y-axis scales are not the same for each part of Fig. 6. The greatest absorbance values correspond to low pH samples of Ca-LPS on ZnSe (Fig. 7a) and LPS (Na and Ca) on ␣-Fe2 O3 .

Fig. 3. ATR-FTIR spectra of ser 10 LPS on ZnSe as a function of pH (3, 6, 9) in (a) 10 mmol L−1 NaCl and (b) 10 mmol L−1 CaCl2 .

on ZnSe (Fig. 3a) and ␣-Fe2 O3 (Fig. 5a). An increased contribution of νas (PO2 − ) and ν(C–O, C–O–C) [relative to νas (CH2 )] is observed with decreasing pH for Na-LPS samples. The spectra of Ca-LPS samples are strongly dependent on crystal surface chemistry (Figs. 3b, 4b, 5b, and 6b). Ca-LPS spectra acquired using a ZnSe IRE (Fig. 3b) show relatively small contributions arising from ν(C–O, C–O) (∼1060 cm−1 ). In contrast, when a GeO2 IRE is used for Ca-LPS (Fig. 4b) there is a large ν(C–O, C–O) peak at pH 3. This O-antigen contribution is diminished with increasing pH, but strong νas (PO2 − ) peaks remain. Regardless of pH, νas (CH2 ) peaks remain small (Fig. 4b) and contribute less to the spectra than is the case for Ca-LPS on ZnSe (Fig. 3b). Ca-LPS spectra acquired on ␣-Fe2 O3 -coated ZnSe and ␣-Al2 O3 -coated ZnSe have strong ν(C–O, C–O–C) peaks, more like the GeO2 case. There is also an increase in the ∼1085 cm−1 peak (relative to that at 1060 cm−1 ) with increased pH, attributed to either νas (PO2 − ) or ν(C–O, C–O–C) ring vibrations. In order to show details of all spectra, the Y-axis in Figs. 4–6 are not on a common scale. Fig. 7 shows the actual absorbance

Fig. 4. ATR-FTIR spectra of ser 10 LPS on GeO2 as a function of pH (3, 6, 9) in (a) 10 mmol L−1 NaCl and (b) 10 mmol L−1 CaCl2 . Decreased peak intensity with increased pH was observed, therefore the Y-axis scale is expanded to show spectra details (non-common scale).

S.J. Parikh, J. Chorover / Colloids and Surfaces B: Biointerfaces 62 (2008) 188–198

193

sus the square root of time (t, diffusion-limited sorption model) reveal near linear trends, particularly for Ca-LPS on ZnSe. The goodness of fit (R2 ) for regressions of ν(CH2 , CH3 ) versus t1/2 ranged from 0.93 to 0.97, from 0.80 to 0.94 for νas (PO4 − ), and from 0.94 to 0.99 for ν(C–O, C–O–C). For Ca-LPS samples collected on GeO2 , R2 values for absorbance versus t1/2 ranged from 0.50 to 0.96 for ν(CH2 , CH3 ), from 0.88 to 0.99 for νas (PO4 − ), and from 0.14 to 0.67 for ν(C–O, C–O–C). The slope of each line is plotted against θ eff in Fig. 8. Error bars represent 95% confidence intervals and, if not visible, are smaller than the symbol size. Each point on the graph represents the slope taken for peak area as a function of collection time; n = 7 for each data point. Because of the much lower absorbance of Ca-LPS on GeO2 relative to ZnSe [i.e., ∼10 times for νas (PO4 − ) and ν(CH2 , CH3 ), ∼2 times for ν(C–O, C–O–C)], the slope values for spectra acquired on ZnSe are much greater.

Fig. 5. ATR-FTIR spectra of ser 10 LPS on ␣-Fe2 O3 -coated ZnSe as a function of pH (3, 6, 9) in (a) 10 mmol L−1 NaCl and (b) 10 mmol L−1 CaCl2 . Decreased peak intensity with increased pH was observed, therefore the Y-axis scale is expanded to show spectra details (non-common scale).

4.3. VATR-FTIR spectroscopy of LPS on ZnSe and GeO2 VATR spectra of NaCl-LPS and CaCl2 -LPS solutions on ZnSe and GeO2 crystals are shown in Fig. 8. In order to quantify spectral differences, Lorentzian peaks were fit to spectra for peak area quantification. These areas and area ratios of major functional groups reveal slight depth-dependent changes for CaLPS solutions on ZnSe and GeO2 . No time-dependent changes were observed for Na-LPS solutions at any of the angles examined. Ca-LPS samples on ZnSe exhibit increased contributions from ν(CH2 , CH3 ) and νas (PO4 − ) at both the high and low angles, with the greatest contribution from ν(C–O, C–O–C) occurring at 45◦ . This trend is observed less dramatically for Ca-LPS samples on GeO2 . Greater relative ν(C–O, C–O–C) intensities are observed for Ca-LPS samples on GeO2 than on ZnSe. Unlike NaCl-LPS solutions, Ca-LPS spectra exhibit increased ν(CH2 , CH3 ) and νas (PO4 − ) peak areas with increased time (e.g., Fig. 9 insets). Plots of major LPS band intensities ver-

Fig. 6. ATR-FTIR spectra of ser 10 LPS on ␣-Al2 O3 -coated ZnSe as a function of pH (3, 6, 9) in (a) 10 mmol L−1 NaCl and (b) 10 mmol L−1 CaCl2 . Decreased peak intensity with increased pH was observed, therefore the Y-axis scale is expanded to show spectra details (non-common scale).

194

S.J. Parikh, J. Chorover / Colloids and Surfaces B: Biointerfaces 62 (2008) 188–198

Fig. 7. Raw IR absorbance values corresponding to νas (CH2 ), νas (PO2 − ), and ␯(C–O, C–O–C) for Na-LPS (solid symbols) and Ca-LPS (open symbols) collected on (a) ZnSe, (b) GeO2 , (c) ␣-Fe2 O3 , and (d) ␣-Al2 O3 as a function of solution pH.

5. Discussion

5.1. Chemical properties of surfaces

This study indicates that adsorbent surface chemistry and solution cation composition both influence how LPS aggregates interact with surfaces. The data suggest a strong influence of substratum surface chemistry (e.g., hydrophobicity, pHdependent charge) on the molecular mechanisms of LPS binding/orientation at the solid/liquid interface for systems with Ca2+ present. Additionally, LPS interaction with Ca2+ results in aggregate reorientation and increased retention on all surfaces studied. LPS aggregates in NaCl solutions are stabilized via H-bonding between saccharide groups [23], whereas the formation of Ca-phosphate bonds in the lipid A region induce aggregate disruption [9] and favor reorientation on a solid surface. These effects are consistent with our previous findings, where Ca-LPS aggregates reoriented on a ZnSe IRE [11]. In the current study, substrata with varying surface chemistry were employed and distinctly different mechanisms of Ca-LPS retention were observed to be dependent on surface hydrophobicity.

The ZnSe IRE is relatively hydrophobic [12,39], whereas GeO2 is considered a hydrophilic IRE [40] because of the formation of a hydroxylated interface similar to that observed on silica. The metal oxides (␣-Fe2 O3 and ␣-Al2 O3 ) are also relatively hydrophilic due to the presence of polar surface hydroxyl groups. These surfaces all exhibit pH-dependent charge, which influence the LPS spectra. The point of zero net proton charge (PZNPC) for ZnSe is <4 [41]. Although a PZNPC value for GeO2 could not be found in the literature, it is estimated to <4 based on the PZC of SiO2 , which comprises similar metaloxygen bond valence, and which has an IEP between 2.0 and 2.5 [42,43]. Furthermore, Gun’ko et al. [43] measured the influence on isoelectric point of adhering highly dispersed GeO2 colloids (up to 20% by mass) to the surface of fumed SiO2 and reported small effects. The PZNPCs for the Fe and Al oxides are 8.0–8.5 for ␣-Fe2 O3 [44], and 9.1 for ␣-Al2 O3 [42]. ATR-FTIR experiments were carried out at pH 3, 6, and 9 to probe a range in surface charge of the IREs and metal oxides. A summary of the

S.J. Parikh, J. Chorover / Colloids and Surfaces B: Biointerfaces 62 (2008) 188–198

195

Fig. 8. VATR-FTIR spectra of ser 10 LPS (pH 6) on (a) ZnSe in 10 mmol L−1 NaCl, (b) ZnSe in 10 mmol L−1 CaCl2 , (c) GeO2 in 10 mmol L−1 NaCl, and (d) GeO2 in 10 mmol L−1 CaCl2 . The effective angles are labeled below spectra. Due to the variation in IR absorbance as a function of depth of penetration the Y-axis has expanded to show detail (non-common scale). Spectra collected at 120 min.

pH-dependent charge and hydrophobicity of LPS and surfaces is given in Table 1. 5.2. ATR-FTIR: LPS interactions with ZnSe, GeO2 , α-Fe2 O3 , and α-Al2 O3 Although, striking differences between the spectra of Ca-LPS and Na-LPS are apparent, similarities in the IR absorbance trends are also observed (Fig. 7). LPS interaction with hydrophilic surfaces (GeO2 , ␣-Fe2 O3 , ␣-Al2 O3 ) exhibits decreased IR absorbance with increasing pH, particularly for ν(C–O, C–O–C) and νas (PO4 − ). This trend is consistent with previous studies that showed decreased peak intensity of bacterial polysaccharides upon proton dissociation [45]. The high IR absorbance values for Ca-LPS on ZnSe and LPS (Ca and Na) on ␣-Fe2 O3 are consistent with peak amplifications resulting from sorption to the solid surface [11,45–48]. Surfactant adsorption to surfaces is controlled by ion exchange, ion pairing, hydrophobic bonding, polarization of π electrons, and dispersion forces

[49,50]. These forces typically involve monomers rather than aggregates [51]. LPS amphiphile adsorption to surfaces likely involves these mechanisms to varying degrees. As will be discussed below, hydrophobic interactions dominate LPS-ZnSe spectra, whereas pH-dependent charge properties are significant for LPS interactions with the more hydrophilic surfaces. 5.2.1. Na-LPS interactions Na-LPS spectra show small dependence on surface composition (Figs. 3–6). Trends in pH are similar, with increased ν(C–O, C–O–C) and νas (PO4 − ) peaks relative to νas (PO4 − ) at lower pH values. Also, pH 3 spectra show a sharp peak at ∼1060 cm−1 that masks other peaks on the region. The Oantigen is more susceptible to changes in conformation than the lipid A and core regions [14,52]. Variation in the Na-LPS spectra as a function of pH results from conformational changes due to protonation/dissociation of ionizable functional groups in the O-antigen region. The lack of influence of different IRE and metal oxide coatings indicate minimal interaction between

196

S.J. Parikh, J. Chorover / Colloids and Surfaces B: Biointerfaces 62 (2008) 188–198

Fig. 9. VATR-FTIR data for P. aeruginosa ser 10 LPS in 10 mM CaCl2 at pH 6 showing the regression slope of peak areas versus (time)0.5 at five different angles for fatty acid, phosphate, and carbohydrate moieties on (a) ZnSe and (b) GeO2 IRE. Error bars represent 95% confidence intervals and n = 7 for each point, representing data collection at seven different times. The insets are given as examples of data used to for linear regression analysis. This data represent the three points corresponding to the three functional groups examined at an effective angle of 45◦ .

Na-LPS and the selected surfaces. The only evidence of surface interactions for Na-LPS is with ␣-Fe2 O3 , and, to a lesser extent, with ␣-Al2 O3 , where the 1409 cm−1 [νs (COO)] likely represents bonding between the O-antigen chain and Fe(III) or Al(III). This peak is smallest at pH 9 where ␣-Fe2 O3 , carries a negative charge and ␣-Al2 O3 is approximately neutral. Adhesion of negatively charged rough LPS to a positively charged surface (aminopropyltriethoxysilane polymers) in ultrapure water has been previously observed [12]. The lack of change in ATR spectra due to the presence of different surfaces gives further evidence that Na-LPS aggregates are stable in solution and do not re-assemble at the solid–liquid interface. The spectra of Na-LPS may represent solution phase spectra. Due to smaller penetration depths for the GeO2 IRE (Table 2) lower absorbance values are expected.

5.2.2. Ca-LPS interactions Ca-LPS spectra are sensitively dependent on surface composition (Figs. 3–6); differences in IR spectra are observed for Ca-LPS on ZnSe compared to GeO2 , ␣-Fe2 O3 and ␣-Al2 O3 . For ZnSe, the most hydrophobic surface used, spectra show diminished ν(C–O, C–O–C) peaks relative to ␯as (PO4 − ) and ν(CH2 , CH3 ), which are functional groups within the lipid A region. Under these conditions LPS aggregate stability is apparently decreased by Ca2+ binding to PO4 − in the lipid A region. This is in agreement with previously published studies on rough LPS [9,24,53]. We postulate that Na-LPS aggregates interact with the ZnSe IRE, whereas Ca-LPS aggregate disruption results in LPS monomer interaction via hydrophobic mechanisms. Since it is assumed that micelles comprising amphiphiles do not adsorb to hydrophobic surfaces [49], the release of monomers from LPS aggregates to the surface is likely the key factor governing adsorption. The importance of surface hydrophobicity for LPS adhesion has been demonstrated for membrane-bound rough LPS [3]. Adhesion of membrane-bound LPS [1] and LPS in ultrapure water [12] was low for hydrophobic surfaces. Direct comparisons between these studies and the free Na-LPS and Ca-LPS surface interactions in this study cannot be made due the effect of cation valence on LPS aggregation and surface interactions. The FTIR spectra indicate that Ca2+ disrupts LPS aggregation regardless of IRE surface composition; however differences in the interaction of the lipid A and O-antigen regions with surfaces are apparent (Figs. 3–5). GeO2 , ␣-Fe2 O3 , and ␣-Al2 O3 are hydrophilic and possess different pH-dependent charge properties (Table 1). The hydrophilic nature of these surfaces translates into greater interaction with O-antigen chains, as evidenced by increased ν(C–O, C–O–C) peaks, compared to the case for ZnSe spectra. The influence of pH on Ca-LPS spectra collected on ZnSe is minimal as hydrophobic interactions primarily control the interactions. However, the pH-dependent charge on hydrophilic surfaces, and Ca-LPS, greatly influence the collected spectra. Bonding of the hydrophilic region of amphiphilic molecules to hydrophilic surfaces may occur via ion exchange, ion pairing, polarization of π electrons, dispersion forces (van der Waals) [49], and hydrogen bonding [3]. Ca-LPS spectra on GeO2 have relatively large ν(C–O, C–O–C) contributions at pH 3, whereas GeO2 is positively charged. Decreased interactions between the O-antigen ν(C–O, C–O–C) rich and the GeO2 IRE are observed at higher pH values (negative charge on GeO2 ). The pH-dependent charge properties of the O-antigen have been previously discussed (see Section 5.2). Both ␣-Fe2 O3 and ␣-Al2 O3 are positively charged to higher pH values, and therefore larger contributions from the O-antigen (negative charge) are observed in these spectra (Figs. 4 and 5). Increased contributions arising from ν(CH2 , CH3 ) are observed in ␣-Fe2 O3 at pH 3 and 6 (Fig. 6). The high affinity of Fe(III) for PO4 − [54–56] may play a role in the increased ν(CH2 , CH3 ) signal. The ␣-Fe2 O3 surface is positively charged (pH 3, 6) and P–O–Fe binding between the lipid A region and the surface may cause the fatty acid region to migrate in close proximity to the IRE, thus increasing ν(CH2 , CH3 ) absorbance. This

S.J. Parikh, J. Chorover / Colloids and Surfaces B: Biointerfaces 62 (2008) 188–198

is observed to a much smaller degree for Ca-LPS on ␣-Al2 O3 (Fig. 5). 5.3. VATR spectra of Na-LPS and Ca-LPS Systematic probing of the aqueous solution at various distances from the IRE permits evaluation of LPS functional group arrangement at the solid/liquid interface. Reorientation of CaLPS results in a depth-dependent arrangement at this interface, whereas Na-LPS spectra are unchanged as function of distance from the IRE. The sorption of polymeric molecules to surfaces can be rate-limited by diffusion to the interface or by surface induced transitions of conformation and/or orientation [57]. Plots of Ca-LPS peak areas, for ν(CH2 , CH3 ), νas (PO2 − ), ν(C–O, C–O–C) versus the square-root of time for five angles of IR beam incidence (all conducted as separate experiments) show a near linear increase (R2 near unity, e.g., Fig. 9 insets) indicating Ca-LPS interactions with ZnSe are consistent with diffusion-limited kinetics [58,59]. The slopes of these regressions are plotted against angle of incidence in Fig. 9a. Similar trends are observed for Ca-LPS on GeO2 (Fig. 9b), however slopes are much smaller and R2 values show greater variation and are consistently near unity only for νas (PO2 − ). The lack of correlation indicates Ca-LPS aggregates do not exhibit the same rate-limited behavior upon adsorption to the GeO2 IRE. VATR data are collected at pH 6, where interactions between Ca-LPS and GeO2 are minimal (Fig. 3) and show PO4 − interactions with GeO2 (not observed for Na-LPS). Data for Ca-LPS on ZnSe reveal different slopes for the three functional groups (Fig. 9a inset), but have similar trends in θ eff for ν(CH2 , CH3 ) and νas (PO2 − ). At low angle (high dp ), steep slopes for ν(CH2 , CH3 ) and νas (PO2 − ) indicate strong timedependence. Slopes decrease with increasing angle followed by a slight increase at shallowest depth measured (θ eff = 51.2◦ ). Although we cannot determine the precise orientation of CaLPS at the ZnSe IRE interface, depth-dependent variation in LPS structure is apparent. Previous attempts to determine LPS orientation via polarized ATR-FTIR were inconclusive [11], likely stemming from problems arising from functional group heterogeneity of large molecules [12]. It is interesting to note that peak areas corresponding to ν(C–O, C–O–C) of O-antigen remain relatively unchanged, whereas lipid A peak areas vary with penetration depth. This suggests that variation of dp is monitoring different layers of the aggregate structure, each of which may have different organization of LPS monomers and associated subunits. Importantly, the length scale over which depth-dependent changes in IR spectra are significant (hundreds of nanometers) is consistent with the size of aggregates measured by DLS [18]. 6. Conclusions The influence of surface hydrophobicity, charge, and chemical composition on ATR-FTIR spectra of Na-LPS and Ca-LPS at the solid–water interface were investigated. The interfacial behavior of LPS is strongly influenced by the cations present. Greater interaction between Ca-LPS (versus NaCl) and sur-

197

faces (ZnSe, GeO2 , ␣-Fe2 O3 , ␣-Al2 O3 ) was observed. Na-LPS spectra were similar irrespective of surface composition, indicating weaker surface interaction and spectra were consistent with solution phase behavior. Ca-LPS interactions with ZnSe are controlled primarily by hydrophobic interactions (lipid A moieties), whereas pH-dependent surface charge plays a key role in Ca-LPS reactions (O-antigen moieties) on more hydrophilic surfaces (GeO2 , ␣-Fe2 O3 , ␣-Al2 O3 ). Increased interactions were observed at lower pH values, where surfaces are generally positively charged. VATR-FTIR spectra indicate variation in Ca-LPS structure as a function of distance from ZnSe and GeO2 IRE surfaces (∼500–2500 nm). However, the specific orientation of LPS molecules could not be determined unambiguously. Na-LPS spectra exhibit minimal depth-dependent changes, revealing limited surface interactions with GeO2 and ZnSe IREs at pH 6. Acknowledgement This research was supported by the National Science Foundation CRAEMS program (Grant CHE-0089156). References [1] S.A. Makin, T.J. Beveridge, Microbiol.-UK 142 (1996) 299. [2] B.A. Jucker, H. Harms, A.J.B. Zehnder, Colloids Surf. B: Biointerf. 11 (1998) 33. [3] B.A. Jucker, H. Harms, S.J. Hug, A.J.B. Zehnder, Colloids Surf. B: Biointerf. 9 (1997) 331. [4] I.W. Sutherland, Annu. Rev. Microbiol. 39 (1985) 243. [5] G. Seltmann, O. Holst, The outer membrane of the Gram-negative bacteria and their components, in: G. Seltmann, O. Holst (Eds.), The Bacterial Cell Wall, Springer-Verlag, Berlin, Germany, 2002, pp. 31–66. [6] J.S. Lam, L.L. Graham, J. Lightfoot, T. Dasgupta, T.J. Beveridge, J. Bacteriol. 174 (1992) 7159. [7] E.T. Rietschel, T. Kirikae, F.U. Schade, U. Mamat, G. Schmidt, H. Loppnow, A.J. Ulmer, U. Zahringer, U. Seydel, F. Dipadova, M. Schreier, H. Brade, FAESB J. 8 (1994) 217. [8] B.A. Jucker, A.J.B. Zehnder, H. Harms, Environ. Sci. Technol. 32 (1998) 2909. [9] Q. Wang, J.P. Zhang, T.R. Smith, W.E. Hurst, T. Sulpizio, Colloids Surf. B: Biointerf. 44 (2005) 110. [10] J.P. Zhang, Q. Wang, T.R. Smith, W.E. Hurst, T. Sulpizio, Biotechnol. Prog. 21 (2005) 1220. [11] S.J. Parikh, J. Chorover, Colloids Surf. B: Biointerf. 55 (2007) 241. [12] G. Reiter, M. Siam, D. Falkenhagen, W. Gollneritsch, D. Baurecht, U.P. Fringeli, Langmuir 18 (2002) 5761. [13] S.E. Paradis, D. Dubreuil, S. Rioux, M. Gottschalk, M. Jacques, Infect. Immun. 62 (1994) 3311. [14] U. Seydel, H. Labischinski, M. Kastowsky, K. Brandenburg, Immunobiology 187 (1993) 191. [15] C.A. Aurell, A.O. Wistr¨om, Biochem. Biophys. Res. Commun. 253 (1998) 119. [16] N.C. Santos, A.C. Silva, M. Castanho, J. Martins-Silva, C. Saldanha, Chembiochem 4 (2003) 96. [17] A. Bergstrand, C. Svanberg, M. Langton, M. Nyd´en, Colloids Surf. B: Biointerf. 53 (2006) 9. [18] S.J. Parikh, A Spectroscopic Study of Bacterial Polymers Mediating Cell Adhesion and Mineral Transformations, Department of Soil, Water and Environmental Science, University of Arizona, Tucson, 2006, p. 270. [19] D. Naumann, C. Schultz, A. Sabisch, M. Kastowsky, H. Labischinski, J. Mol. Struct. 214 (1989) 213. [20] M. Schindler, M.J. Osborn, Biochemistry 18 (1979) 4425. [21] L.P. Li, R.G. Luo, Biotechnol. Tech. 12 (1998) 119.

198

S.J. Parikh, J. Chorover / Colloids and Surfaces B: Biointerfaces 62 (2008) 188–198

[22] L.P. Li, R.G. Luo, Sep. Sci. Technol. 34 (1999) 1729. [23] R.T. Coughlin, A.A. Peterson, A. Haug, H.J. Pownall, E.J. Mcgroarty, Biochim. Biophys. Acta 821 (1985) 404. [24] S. Obst, M. Kastowsky, H. Bradaczek, Biophys. J. 72 (1997) 1031. [25] F.M.J. Mirabella, Appl. Specctrosc. Rev. 21 (1985) 45. [26] U. Seydel, M. Oikawa, K. Fukase, S. Kusumoto, K. Brandenburg, Eur. J. Biochem. 267 (2000) 3032. [27] M. Jonasz, G. Fournier, D. Stramski, Appl. Opt. 36 (1997) 4214. [28] A. Katz, A. Alimova, M. Xu, P. Gottlieb, E. Rudolph, J.C. Steiner, R.R. Alfano, Opt. Lett. 30 (2005) 589. [29] K.K. Chittur, Biomaterials 19 (1998) 357. [30] M.R. Pereira, J. Yarwood, J. Polym. Sci. Part B-Polym. Phys. 32 (1994) 1881. [31] U. Schwertmann, R.M. Cornell, Iron Oxides in the Laboratory: Preparation and Characterization, Wiley-VCH, Weinheim, 1991. [32] S.J. Parikh, J. Chorover, Langmuir 22 (2006) 8492. [33] C. Mui, J.P. Senosiain, C.B. Musgrave, Langmuir 20 (2004) 7604. [34] Y.F. Mei, G.G. Siu, X.H. Huang, K.W. Cheah, Z.G. Dong, L. Fang, M.R. Sheng, X.L. Wu, X.M. Bao, Phys. Lett. A 331 (2004) 248. [35] S.M. Abo-Naf, H. Darwish, M.M. El-Desoky, J. Mater. Sci. - Mater. Electron. 13 (2002) 537. [36] U. Schwertmann, R.M. Taylor, Iron oxides, in: A. Klute (Ed.), Methods of Soil Analysis, Part 1—Physical and Mineralogical Methods, SSSA, Madison, WI, 1989, pp. 379–438. [37] J.A. Gadsen, Infrared Spectra of Minerals and Related Organic Compounds, Butterworths, London, UK, 1975. [38] M.N. Danchevskaya, S.N. Torbin, Y.D. Ivakin, G.P. Muravieva, J. Phys.: Condens. Matter 16 (2004) S1187. [39] Y.P. Song, J. Yarwood, J. Tsibouklis, W.J. Feast, J. Cresswell, M.C. Petty, Langmuir 8 (1992) 262.

[40] T. Snabe, S.B. Petersen, J. Biotechnol. 95 (2002) 145. [41] L.D. Tickanen, M.I. Tejedor-Tejedor, M.A. Anderson, Langmuir 13 (1997) 4829. [42] D.L. Sparks, Environmental Soil Chemistry, Academic Press, Inc., San Diego, CA, 1995. [43] V.M. Gun’ko, V.I. Zarko, V.V. Turov, R. Leboda, E. Chibowski, V.V. Gun’ko, J. Colloid Interface Sci. 205 (1998) 106. [44] G. Sposito, The Chemsitry of Soils, Oxford University Press, New York, 1989. [45] A. Omoike, J. Chorover, K.D. Kwon, J.D. Kubicki, Langmuir 20 (2004) 11108. [46] A. Omoike, J. Chorover, Biomacromolecules 5 (2004) 1219. [47] D. Peak, R.G. Ford, D.L. Sparks, J. Colloid Interface Sci. 218 (1999) 289. [48] S.J. Hug, J. Colloid Interface Sci. 188 (1997) 415. [49] S. Paria, K.C. Khilar, Adv. Colloid Interface Sci. 110 (2004) 75. [50] P. Somasundaran, L. Huang, Adv. Colloid Interface Sci. 88 (2000) 179. [51] J.C. Griffith, A.E. Alexander, J. Colloid Interface Sci. 25 (1967) 311. [52] M. Kastowsky, T. Gutberlet, H. Bradaczek, J. Bacteriol. 174 (1992) 4798. [53] R.T. Coughlin, A. Haug, E.J. Mcgroarty, Biochemistry 22 (1983) 2007. [54] M.I. Tejedor-Tejedor, M.A. Anderson, Langmuir 6 (1990) 602. [55] P. Persson, N. Nilsson, S. Sjoberg, J. Colloid Interface Sci. 177 (1996) 263. [56] Y. Arai, D.L. Sparks, J. Colloid Interface Sci. 241 (2001) 317. [57] M.A. Brusatori, P.R. van Tassel, J. Colloid Interface Sci. 219 (1999) 333. [58] R.S. Kookana, L.A.G. Aylmore, R.G. Gerritse, Soil Sci. 154 (1992) 214. [59] O.H. Jacobsen, P. Moldrup, C. Larsen, L. Konnerup, L.W. Petersen, J. Hydrol. 196 (1997) 185. [60] A. Katz, A. Alimova, M. Xu, E. Rudolph, M.K. Shah, H.E. Savage, R.B. Rosen, S.A. Mccormick, R.R. Alfano, IEEE J. Sel. Top. Quan. El. 9 (2003) 277.