Protein–surfactant interactions at hydrophobic interfaces studied with total internal reflection fluorescence correlation spectroscopy (TIR-FCS)

Journal of Colloid and Interface Science 317 (2008) 449–457 www.elsevier.com/locate/jcis

Protein–surfactant interactions at hydrophobic interfaces studied with total internal reflection fluorescence correlation spectroscopy (TIR-FCS) Andreas W. Sonesson a,b , Hans Blom c,∗ , Kai Hassler c , Ulla M. Elofsson a , Thomas H. Callisen d , Jerker Widengren c , Hjalmar Brismar b a YKI, Institute for Surface Chemistry, Stockholm, Sweden b Department of Cell Physics, Royal Institute of Technology, Stockholm, Sweden c Department of Biomolecular Physics, Royal Institute of Technology, Stockholm, Sweden d Novozymes A/S, Bagsvaerd, Denmark

Received 27 August 2007; accepted 27 September 2007 Available online 6 October 2007

Abstract The aim of this work was to study the dynamics of proteins near solid surfaces in the presence or absence of competing surfactants by means of total internal reflection fluorescence correlation spectroscopy (TIR-FCS). Two different proteins were studied, bovine serum albumin (BSA) and Thermomyces lanuginosus lipase (TLL). A nonionic/anionic (C12 E6 /LAS) surfactant composition was used to mimic a detergent formulation and the surfaces used were C18 terminated glass. It was found that with increasing surfactant concentrations the term in the autocorrelation function (ACF) representing surface binding decreased. This suggested that the proteins were competed off the hydrophobic surface by the surfactant. When fitting the measured ACF to a model for surface kinetics, it was seen that with raised C12 E6 /LAS concentration, the surface interaction rate increased for both proteins. Under these experimental conditions this meant that the time the protein was bound to the surface decreased. At 10 µM C12 E6 /LAS the surface interaction was not visible for BSA, whereas it was still distinguishable in the ACF for TLL. This indicated that TLL had a higher affinity than BSA for the C18 surface. The study showed that TIR-FCS provides a useful tool to quantify the surfactant effect on proteins adsorption. © 2007 Elsevier Inc. All rights reserved. Keywords: Protein–surfactant interactions; TIR-FCS; Lipase; BSA; C12 E6 /LAS; Hydrophobic surface

1. Introduction Interactions between proteins and surfactants at interfaces are of great importance in a variety of fields, such as interfacial enzymology and lipid–protein interactions in biological membranes and in technical applications like food chemistry, detergency formulations and gel electrophoresis [1]. By complexing with proteins in solution [2], surfactants might alter properties such as the conformational stability of the protein structure, the hydrophobicity of the protein surface [3–5] or the catalytic activity of enzymes [6,7]. At solid surfaces surfactants are known to displace adsorbed proteins either by replacement * Corresponding author: Royal Institute of Technology, Albanova University Center, Experimental Biomolecular Physics, SE-106 91 Stockholm, Sweden. E-mail address: [email protected] (H. Blom).

0021-9797/$ – see front matter © 2007 Elsevier Inc. All rights reserved. doi:10.1016/j.jcis.2007.09.089

or by inducing solubilization of the protein if the surfactants are able to bind to the protein structure [1]. Under competitive adsorption, surfactants might compete with the protein for adsorption sites at various interfaces. Commonly used techniques to study protein–surfactant interactions at interfaces are, e.g., surface tension measurements [8–10], ellipsometry [11,12] and Langmuir–Blodgett films imaged by atomic force microscopy [13,14]. Fluorescence correlation spectroscopy (FCS) is yet another technique that can be used to study interactions as well as mobilities and densities with single molecule sensitivity in vitro or in vivo [15,16]. Processes commonly studied with FCS include diffusion, flow, binding, rotation, photo-physics, bleaching and conformational changes [17–19]. In combination with total internal reflection excitation, so-called TIR-FCS, the technique has been used to quantify dynamic events at interfaces like reversible adsorption/desorption kinetics of dye molecules

450

A.W. Sonesson et al. / Journal of Colloid and Interface Science 317 (2008) 449–457

[20,21] and monodisperse polymers [22] to solid surfaces, dye transport in sol–gel films [23], ligand–receptor kinetics [24], antibody diffusion near phospholipid bilayers [25,26] and lateral mobility of proteins in cellular membranes [27]. However, there are to our knowledge no reported TIR-FCS studies of surface kinetics of protein–surfactant interactions. Proteins and surfactants are known to interact both in solution and at interfaces, but little is known how surfactants affect the surface dynamics of proteins. The theoretical basis of TIR-FCS and the first experimental results were demonstrated by Thompson et al. [28,29] already in the early 1980’s. FCS is basically an analytical tool that analyses fluorescence fluctuations from individual analyte molecules. The statistics of the fluctuation carries information about thermodynamic and kinetic properties of the investigated system. Correlation analysis of the fluctuations can therefore deduce molecular kinetics and numbers of individuals participating in a specific dynamic process. Although the FCS technique is sensitive to single molecules, an ensemble of events is measured and correlated in order to yield good statistics. In standard (confocal) FCS, fluorescence is detected from within a microscopic detection volume defined by the focused laserspot in a confocal microscope. However, since the axial extent of the confocal detection volume is about 1–2 µm, total internal reflection excitation is more appropriate to discriminate events happening on or at the vicinity of an interface. The axial extent of the detection volume in TIR-FCS, mainly defined by the evanescent field of the totally reflected laser, is typically smaller than 100 nm. Such reduced detection volumes [30] generate the possibility of studying kinetic processes at interfaces such as cellular membranes, thin films and phospholipid bilayers. The aim of this work was to apply the technique of TIRFCS to study the dynamics of proteins near solid surfaces in the presence or absence of competing surfactants. This technique allows detailed quantification of the kinetics and interaction terms of the investigated system, and thus gives important complementary information to traditional adsorption techniques. The proteins investigated were bovine serum albumin (BSA) and the lipase from the fungus Thermomyces lanuginosus (TLL). Since proteins are known to interact strongly with hydrophobic surfaces [31], glass surfaces silanized with long carbon chains (C18) were used as substrates. A nonionic/anionic (C12 E6 /LAS) surfactant mixture was used to mimic the main fraction in a detergent formulation. Competitive adsorption/desorption kinetics of proteins and surfactants on the hydrophobic surfaces was analyzed and quantified.

from Sigma-Aldrich. Hexaethylene glycol mono n-dodecylether (C12 E6 ) was from Nikko Chemicals Co., Ltd. (Tokyo, Japan, Lot. No. 6025) and linear alkylbenzene sulfonate (LAS, technical grade with an alkyl chain length average of 12) was from Petresa (Spain). Fluorescein isothiocyanate, FITC (Cat. No. F-143, lot. 1173-1), was purchased from Molecular Probes Europe BV (Leiden, The Netherlands). The buffer used throughout the TIR-FCS experiments was glycine pH 9.0 (10 mM NaCl, 0.05 mM EDTA, 50 mM glycine and 1 mM NaN3 ). All buffer salts were of analytical grade. The dye Rhodamine 6G (Rh6G) was from Lambda Physics (Göttingen, Germany). Stock solution of Rh6G was prepared in spectroscopically pure ethanol, which prior to measurements was diluted in Milli-Q water to nanomolar concentrations. 2.2. FITC labeling of proteins TLL was dissolved in sodium bicarbonate buffer and mixed with FITC dissolved in DMF to a molar ratio of approximately 40:1 (FITC:TLL). The mixture was protected from light and incubated under continuous stirring at room temperature for 2 h followed by overnight incubation at +4 ◦ C. A Sephadex G-25 M PD-10 column (Amersham Biosciences AB, Uppsala, Sweden) with glycine buffer as elution buffer was used to remove unreacted FITC. To determine approximate protein and fluorophore concentrations, the absorbance at 280 and 494 nm was measured, using ε TLL, 280 nm = 38440 cm−1 M−1 and ε FITC, 494 nm = 78000 cm−1 M−1 . This indicated a FITC to TLL ratio of about 1:1. The free dye content in the FITClabeled TLL sample was estimated to about 50% using confocal FCS [17] (data not shown). In the commercial FITC-BSA sample, the free dye content was about 20%. FITC-TLL and FITC-BSA were stored in aliquots at −20 ◦ C and thawed prior to use. 2.3. Surface preparations

2. Materials and methods

Coverslips (Menzel-Gläser, Braunschweig, Germany,  = 22 mm) were cleaned for 5 min in 80 ◦ C 5:1:1 (v/v/v) H2 O:NH3 : H2 O2 , thoroughly rinsed in water and then cleaned for 5 min in 80 ◦ C 5:1:1 (v/v/v) H2 O:HCl:H2 O2 . The wafers were thereafter rinsed several times with water and immersed in an unstirred solution of 0.5% (v/v) OTS in toluene for 24 h. The glass slides were then rinsed in chloroform and put in a chloroform bath for 2 min. Finally, the slides were rinsed with ethanol and water and blown dry with nitrogen. This procedure led to hydrophobic, C18-terminated glass surfaces with a contact angle of 101 ± 2◦ .

2.1. Materials

2.4. TIR-FCS instrumentation

A lipase variant from Thermomyces lanuginosus (TLL) was provided by Novozymes A/S (Bagsvaerd, Denmark). Bovine serum albumin (BSA), BSA–fluorescein isothiocyanate conjugate (FITC-BSA, Cat. No. A9771, 7 mol FITC per mol albumin), dimethylformamide (DMF, 99.9%) and octadecyltrichlorosilane, OTS (Cat. No. 104817-25G), were purchased

The experimental TIR-FCS setup has been described previously for dual-color cross-correlation investigations [32]. A schematic of the experimental set-up is seen in Fig. 1, top. A single line 488 nm linear polarized argon-ion laser (532-BSA04, Melles Griot) emitting 50 mW was expanded to about 3 mm and collimated (lenses L1 and L2). Circular polarization

A.W. Sonesson et al. / Journal of Colloid and Interface Science 317 (2008) 449–457

451

the microscope objective and the dichroic mirror were moved in one block by a linear translator (TS) to adjust the lateral position of the laser beam entering the objective. In this way, the excitation angle could be adjusted without altering the optical path length between the focusing lens and the objective. During measurements the angle was always fixed to a value of about 66◦ . The configuration is easily changed to conventional confocal epi-illumination by removing the focusing lens (L3) and centering the laser beam with the translator (TS). The emitted fluorescence was collected with the same high NA objective, focused with a tube lens (TL), and detected by a fiber-coupled avalanche photodiode (SPAD, SPCM-AQR-14FC, PerkinElmer Optoelectronics). A pinhole in front of the detector reduced the axial extent of the detection volume. The pinhole was realized by using a multimode fiber (Fibertech, Berlin, Germany) with a core diameter of 50 µm situated in a translator for lateral and axial adjustments. A beamsplitter plate (BSP) guided about ten percent of the fluorescent light to an electron multiplying CCD camera (Andor LucaEM ) for visualization of interactions on the hydrophobic C18 surfaces. To block back-reflected and scattered laser light band-pass filters (EMF, HQ 532/70, Chroma Tech. Corp., USA) were put in front of the fiber and the camera. The detector signal was finally correlated online using an ALV-5000/E correlator (ALV Erlangen, Germany) linked to a standard PC. Samples were mixed just prior to pipetting them onto the C18-terminated glass. Autocorrelation functions (ACFs) were detected after approximately 30 s, if not stated otherwise. 2.5. TIR-FCS data analysis Fig. 1. (Top) Detailed scheme of the objective-type TIR-FCS setup. L1–L3: lenses; TS: translator; DM: dichroic mirror; OL: objective lens; EMF: emission filter; M: mirror; TL: tube lens; BSP: beamsplitter plate. (Bottom) Normalized TIR-FCS autocorrelation functions of free FITC and Rh6G on the hydrophobic C18 surfaces. (a) The correlation function of FITC shows a clear contribution from triplet kinetics (T , τt ), free diffusion (GF , τz , w) and surface interactions (GB , RB ). (b) The autocorrelation function of Rh6G is dominated by surface interactions. The fit to Eq. (1) gave the following values for FITC: T = 0.43, τt = 1.8 µs, GF = 0.17, τz = 5.8 µs, w = 0.19 (fixed), GB = 0.06, RB = 300 s−1 ; and for Rh6G: T = 0.08, τt = 2.1 µs, GF = 0.06, τz = 6.1 µs, w = 0.19 (fixed), GB = 0.75, RB = 170 s−1 (irradiance ∼2000 W/cm2 ). Inset: molecular structures of the two dyes; FITC is negatively charged and Rh6G positively charged in the studied Milli-Q water (pH = 7).

was achieved with a broadband mica λ/4 wave plate from OptoSigma. The laser power was controlled by a continuously neutral density wheel with maximum optical density of 4.0 (Thorlabs Inc., USA). An achromatic lens (L3) focused the laser beam into the back-focal plane of an oil immersion objective (OL, α-Plan-Fluar 100× NA1.45, Carl Zeiss), which resulted in a circular spot with a radius of 8 µm (FWHM) at the glass/water interface. A beam offset of about 2.5 mm from the optical axis resulted in super-critical angle illumination, i.e., evanescent field excitation. The 488 nm laser line was reflected into the objective using a dichroic mirror (DM, F495Di02, Semrock, Rochester, USA). 70% of the incident power was delivered to the glass/water interface as measured by a power meter. The lens focusing onto the back-focal plane of

The measured autocorrelation functions (ACFs) were fitted to a model based on the theoretical description of surface interaction kinetics and axial diffusion in the evanescent field by Thompson and co-workers [33,34]. Triplet state kinetics and transversal diffusion through the laterally restricted detection volume were included in the model [35,36]. The expression for the analytical ACF thus reports about photophysical kinetics, free diffusion and surface interactions and reads    −1  τ T 2 τ exp − GF 1 + w g(τ ) = 1 + 1 + 1−T τt τz       τ τ τ × 1− exp erfc 2τz 4τz 4τz 

4τ + (1) + GB exp(−RB τ ) . πτz T here denotes the fraction of fluorescent molecules in the triplet state and τt its relaxation time [35,36]. The axial diffusion time is defined as τz = h2 /4D where D is the diffusion coefficient and h is the axial extent of the detection volume, which is approximately equal to penetration depth of the evanescent field. w = h/r describes a scaling factor for the detection geometry, where r is the lateral extent of the lateral Gaussian intensity distribution. As h and the pinhole size were constant w could easily be fixed [35]. GF and GB are

452

A.W. Sonesson et al. / Journal of Colloid and Interface Science 317 (2008) 449–457

functions of the fluorescent molecules in solution and fluorescent molecules bound to the surface, respectively [24]. RB is the surface interaction rate, which depends on the association and dissociation rate constants [28]. To fit the experimental data a multidimensional least-square algorithm written in Matlab (Mathworks Inc., USA) was used to deliver the unknown variables in Eq. (1). 60-s measurements repeated 2–3 times were performed for each investigated system. The fraction of B = GB /(GB + GF ) the ACF describing surface interaction G and surface interaction rate RB were finally analyzed and compared for each investigated system. Depending on the system analyzed, the variables GF , GB , and RB carry slightly different information: (i) Under reversible adsorption/desorption kinetics, fluorescent molecules in solution of concentration Af are in equilibrium with the density of fluorescent molecules on the surface Cf and the density of unoccupied binding sites B. This reaction mechanism may be written as Af +B  Cf , and is driven forward by the association rate constant kaf and backward by the dissociation rate constant kdf . The fraction of unoccupied binding sites at the surface is then given by β = B/(B + Cf ). Assuming an average of NFf fluorescent molecules in solution and an average of NBf surface-bound molecules gives [28,29] GF = γ NFf /(NFf + NBf )2 , GB = βNBf /(NFf + NBf )2 , RB = kaf Af + kdf ,

(2)

where γ is a geometrical correction factor for the detection volume [17]. (ii) If the molecules bind irreversibly to the surface, the surface interaction rate RB is affected. Since there is no dissociation from the surface kd is replaced by a photobleaching rate giving RB = kaf Af + 1/τbleach ,

(3)

where τbleach is the photobleaching time of the dye. (iii) When the solution also contains a concentration of nonfluorescent molecules An , a competition for binding sites will take place, forming densities of fluorescent Cf and nonfluorescent Cn surface bound molecules. The reaction mechanism for the later case can be written as An + B  Cn and the fraction of unoccupied binding sites can be written as β = B/(B + Cf + Cn ), which gives [24,29,34,37]

GB = 1 − η(1 − β) NBf /(NFf + NBf )2 , (4) where η is the fraction of surface-bound molecules that are fluorescent. In the case that there is a large excess of nonfluorescent molecules such that the association rate kan An  kaf Af , kdn , kdf , the surface interaction rate merely reports on the dissociation process of the fluorescent molecules: RB = kdf .

(5)

The dissociation constant kdf is replaced by the photobleaching rate if the binding of the fluorescent molecules is irreversible. 3. Results 3.1. Fluorophore interactions with hydrophobic surfaces Interactions of two different dye molecules, fluorescein isothiocyanate (FITC) and Rhodamine 6G (Rh6G), with the C18 modified glass surface were studied. 20 nM samples dissolved in Milli-Q water were used. The resulting autocorrelation functions, g(τ ) − 1 normalized to one at τ = 10−6 of these measurements are displayed in Fig. 1, bottom. For Rh6G, the term representing surface binding dominated the ACF as seen by the large part at longer correlation times (τ > 1 ms). FITC, on the other hand, generated an autocorrelation function where the surface interaction fraction was much lower. In this case all dynamic processes in Eq. (1) were clearly visible: the on-off transition to the metastable triplet state (τ < 5 µs [38]), the free axial and lateral diffusion (5 µs < τ < 200 µs) and finally the surface interaction at longer correlation times. B = 0.93 for Rh6G and Comparing the two dyes generated G  GB = 0.26 for FITC. The surface interaction rate for Rh6G and FITC were deduced to RB = 170 s−1 and RB = 300 s−1 , respectively. 3.2. Effect of laser power for detecting protein–surface interactions The influence of instrumental settings when detecting protein–surface interactions was studied. 20 nM FITC-labeled TLL samples were used and the results are shown in Fig. 2, left. When using a low laser power (i.e., 2 mW, irradiance ∼700 W/cm2 ) a distinct contribution in the ACF from protein– surface binding was clearly visible. When performing the measurement at 3 mW (irradiance ∼1000 W/cm2 ) almost no surB face bound TLL was detected in the ACF. In this case G was less than 0.03 compared to 0.19 at the lower power. Moreover, with high laser power a higher correlation was seen at τ = 10−6 s. The surface interaction rate for TLL was RB = 50 s−1 and RB = 160 s−1 at the low and high power, respectively. To fit the data in Fig. 2, left, an average axial diffusion time of τz = 15 µs was assumed for the 50/50 mixture of free FITC and the FITC-labeled TLL. This value was deduced with the calibration measurement of the dye in Fig. 1, bottom, and assuming that the change in diffusion time is proportional to the mass raised to a power of one third [18] (τTLL ∝ τFITC [MTLL /MFITC ]1/3 ⇒ τTLL = 25 µs with τFITC = 5.8 µs, MFITC ≈ 0.4 kDa and MTLL ≈ 31 kDa). 3.3. Effect of incubation time and excess protein concentration The effect of incubation time on the C18 modified glass surfaces was studied with 20 nM TLL samples. For TLL, the effect of measuring the ACF immediately after addition

A.W. Sonesson et al. / Journal of Colloid and Interface Science 317 (2008) 449–457

453

Fig. 2. (Left) Autocorrelation functions for 20 nM FITC labeled TLL on a C18 surface using two different laser powers. 2 mW (irradiance ∼700 W/cm2 ): T = 0.31, τt = 2.0 µs, GF = 0.7, τz = 15 µs (fixed), w = 0.19 (fixed), GB = 0.16, RB = 50 s−1 ; 3 mW (irradiance ∼1000 W/cm2 ): T = 0.34, τt = 1.9 µs, GF = 1.4, τz = 15 µs (fixed), w = 0.19 (fixed), GB = 0.04, RB = 160 s−1 . (Right) Normalized autocorrelation function for 20 nM FITC labeled TLL measured 0 and 30 min after initial adsorption. 0 min (irradiance ∼700 W/cm2 ): T = 0.3, τt = 2.1 µs, GF = 0.7, τz = 15 µs (fixed), w = 0.19 (fixed), GB = 0.17, RB = 50 s−1 ; 30 min (irradiance ∼700 W/cm2 ): T = 0.3, τt = 2.0 µs, GF = 1.3, τz = 15 µs (fixed), w = 0.19 (fixed), GB = 0.06, RB = 110 s−1 .

unlabeled BSA, an axial diffusion time of τz = 27 µs was derived (as a weighted average of the diffusion times of free FITC and BSA, with MBSA ≈ 66 kDa ⇒ τBSA = 32 µs). 3.4. Protein–surface interactions in the presence of surfactants

Fig. 3. Normalized autocorrelation function for 20 nM FITC labeled BSA without and with excess of unlabeled BSA. No excess (irradiance ∼700 W/cm2 ): T = 0.33, τt = 2.0 µs, GF = 0.52, τz = 27 µs (fixed), w = 0.19 (fixed), GB = 0.18, RB = 8 s−1 ; 1 µM excess (irradiance ∼350 W/cm2 ): T = 0.24, τt = 2.3 µs, GF = γ /NFf = 0.7, τz = 42 µs, w = 0.19 (fixed), fitted only with photophysical kinetics and free diffusion parts in Eq. (1). The geometrical γ factors for the present TIR-FCS setup is 0.29 [35].

of the sample compared to waiting 30 min is displayed in Fig. 2, right. After incubation, very little protein–surface inB = 0.20 B = 0.04 as compared to G teraction was seen, G initially. The surface interaction rate immediately after adsorption was RB = 50 s−1 , waiting 30 min generated RB = 110 s−1 . Additionally, binding of 20 nM labeled BSA to the C18 surface was studied with and without excess of unlabeled BSA. Fig. 3 shows that with addition of 1 µM unlabeled BSA, the term in the ACF representing surface binding was no longer resolvable. Without excess BSA it was found that the surface interaction rate, RB = 8 s−1 (estimated immediately after adsorption), was about six times lower for BSA than for TLL. We also noted that the diffusion time of BSA increased with the addition of excess BSA. Fitting the data displayed in Fig. 3, which was obtained for an excess of 1 µM unlabeled BSA gave an axial diffusion time of τz = 42 µs. However, with no excess of

Different concentrations (1, 10 and 500 µM) of C12 E6 /LAS were let to interact with 20 nM BSA samples. The effect on the ACF when adding 1 and 10 µM surfactants is displayed in Fig. 4. With increasing C12 E6 /LAS concentration BSA adsorption to the C18 surface decreased, as seen by the lower fraction at τ ≈ 0.1 s in the ACF. At 10 µM surfactant concentration no surface interaction at all was visible in the ACF. The effect of adding C12 E6 /LAS to the BSA solution was also visible in video sequences taken at the surface (two snapshots are displayed in the inset of Fig. 4). When no surfactant was present in the bulk solution, fluorescent bursts from FITC-BSA molecules interacting with the surface were clearly visible (i.e., intense white spots). With 500 µM C12 E6 /LAS present these binding events and fluorescent bursts were no longer visible (i.e., no intense white spots). The imaged bulk intensity was instead more pronounced, indicating that almost all FITC labeled BSA were found in the solution. This imaging analysis supports the ACF analysis seen in Fig. 4. Moreover, with 500 µM C12 E6 /LAS the axial diffusion time of BSA increased slightly over 50% as compared to when no surfactants were present. For TLL, the effect of surfactants on the surface interaction was similar as found with BSA. The autocorrelation functions with only TLL and the addition of 1 or 10 µM C12 E6 /LAS are displayed in Fig. 5. With increasing surfactant concentration, the term in the ACF representing surface binding decreased. However, in contrast to BSA, the surface binding interactions were still visible at 10 µM C12 E6 /LAS. Table 1 summarizes and B and RB values for BSA and TLL as a function compares G of surfactant concentration. For both proteins, with increasing B decreased. In the lipase system, C12 E6 /LAS concentration G B was initially 0.21 decreasing to 0.15 with 1 µM C12 E6 /LAS G B decreased and was 0.11 at 10 µM C12 E6 /LAS. With BSA G

454

A.W. Sonesson et al. / Journal of Colloid and Interface Science 317 (2008) 449–457

Fig. 4. Normalized autocorrelation functions for 20 nM FITC labeled BSA on a C18 surface without and with 1 and 10 µM C12 E6 /LAS, respectively. No surfactant (irradiance ∼700 W/cm2 ): T = 0.33, τt = 2.0 µs, GF = 0.52, τz = 27 µs (fixed), w = 0.19 (fixed), GB = 0.18, RB = 8 s−1 ; 1 µM surfactant (irradiance ∼350 W/cm2 ): T = 0.27, τt = 2.2 µs, GF = 0.66, τz = 27 µs (fixed), w = 0.19 (fixed), GB = 0.12, RB = 11 s−1 ; 10 µM surfactant (irradiance ∼350 W/cm2 ): T = 0.24, τt = 2.2 µs, GF = γ /NFf = 0.76, τz = 26 µs, w = 0.19 (fixed), fitted only with photophysical kinetics and free diffusion in Eq. (1). Inset: Snapshots from the video sequence of the TIR-illuminated region on C18 surfaces. (Top) FITC-BSA with no C12 E6 /LAS; (bottom) FITC-BSA with 500 µM C12 E6 /LAS. The frame time of each image was 26 ms. Table 1 B = GB /(GB + GF ) and surface Normalized surface interaction amplitudes G interaction rates RB for BSA and TLL as a function of C12 E6 /LAS concentration. The protein–surfactant systems were interpreted according to case (iii) in the TIR-FCS data analysis section

Fig. 5. Normalized autocorrelation functions for 20 nM FITC-labeled TLL on a C18 surface without surfactant and with 1 and 10 µM C12 E6 /LAS, respectively. No surfactant (irradiance ∼350 W/cm2 ): T = 0.28, τt = 2.2 µs, GF = 0.70, τz = 15 µs (fixed), w = 0.19 (fixed), GB = 0.19, RB = 40 s−1 ; 1 µM C12 E6 /LAS (irradiance ∼350 W/cm2 ): T = 0.27, τt = 2.1 µs, GF = 0.90, τz = 15 µs (fixed), w = 0.19 (fixed), GB = 0.16, RB = 60 s−1 ; 10 µM C12 E6 /LAS (irradiance ∼350 W/cm2 ): T = 0.29, τt = 2.2 µs, GF = 1.1, τz = 15 µs (fixed), w = 0.19 (fixed), GB = 0.14, RB = 80 s−1 .

from 0.26 without surfactant to about 0.15 at 1 µM C12 E6 /LAS. At the highest surfactant concentration, no BSA surface interactions were distinguishable. The observed surface interaction rate RB was higher for TLL compared to BSA and changed slightly with increased C12 E6 /LAS concentration. For TLL the rate was about 60 and 80 s−1 at 1 and 10 µM surfactant concentration, respectively. For BSA, RB was deduced to 10 s−1 at 1 µM C12 E6 :LAS. 4. Discussion The aim of this study was to investigate protein–surfactant interactions on hydrophobic surfaces with total internal reflection fluorescence correlation spectroscopy (TIR-FCS). Apply-

Protein

C12 E6 /LAS (µM)

B G

RB (s−1 )

Bovine serum albumin (BSA)

0 1 10

0.26 0.15 n.a.

8 11 n.a.

Thermomyces lanuginosus (TLL)

0 1 10

0.21 0.15 0.11

40 60 80

ing TIR-FCS requires that the species of interest are fluorescent or, when studying single proteins, that they are labeled with fluorescent dyes. However, although using size exclusion columns to separate unreacted dye from protein-conjugated dye after chemical labeling reactions, a portion of free dye was always present in the filtered solution. Free dye can also be generated if the labeled protein sample is biologically unstable and decomposes with time. Since FCS is sensitive to fluctuation from individual molecules, the labeled proteins as well as the free dye populations will contribute to the autocorrelation function. Calibration measurements with only free dyes are therefore always performed in FCS. Two commonly used dyes, fluorescein isothiocyanate (FITC) and Rhodamine 6G (Rh6G) were used in this study. Although Rh6G is superior to FITC in terms of quantum yield and photostability [38–40], FITC-labeled proB of the free teins were chosen due to the comparable low G dye on the C18 surfaces. The low contribution of surface kinetics to the ACF ensured that the binding kinetics of the free dye did not mask the contribution from the proteins. In addition, the difference in surface interaction rates for TLL (RB = 50 s−1 ) and BSA (RB = 8 s−1 ) compared to FITC (RB = 300 s−1 ) ensured that the free dye contributions to the surface binding term B and lower surface in the ACF were minimized. The higher G interaction rate RB for Rh6G are explained by a smaller bleach-

A.W. Sonesson et al. / Journal of Colloid and Interface Science 317 (2008) 449–457

ing rate and a higher tendency to interact with the surface. The latter is reasonable due to the structure of Rh6G, which has long hydrophobic carbon chains attached to the fluorescent aromatic structure. Due to the entropic penalty of hydrating these parts, the hydrophobic fluorophore will aggressively adsorb on C18 surfaces [31,41] and Rh6G will therefore disturb the surface binding of labeled proteins. Rh6G interactions with C18 surfaces have earlier been studied thoroughly with TIR-FCS [20,21]. In order to detect protein-surface interactions at the hydrophobic glass interface, it was found that the experimental conditions in the TIR-FCS setup were of great importance. Photobleaching of dye molecules often becomes a problem when studying labeled proteins on surfaces or in cellular compartments [42]. Since the protein adsorption process is thought to be mainly irreversible [31], bound proteins are subject to longer exposures of the laser and will thus have a higher probability to be photobleached [39,40]. This was evident when using two different laser powers with FITC-labeled TLL proteins. At low power (irradiance ∼700 W/cm2 ) the surface interaction amplitude was visible in the ACF, but with a higher power (irradiance ∼1000 W/cm2 ) the dyes attached to the proteins were photobleached and gave only a limited contribution to the ACF (Fig. 2, left). Hence, extreme photobleaching leads to a situation where only fluorescently labeled proteins in the solution can be seen in the ACF, which was clearly visible in this work. To reduce the effect of photobleaching the irradiance was always kept at or below 700 W/cm2 , which allowed discrimination of protein–surface interactions. As already stated by Thompson et al. [24], the amplitude of the autocorrelation function in TIR-FCS should always be lower when surface interactions are visualized as compared to free diffusion. A larger fraction of FITC dyes were also excited to the triplet state at higher irradiances thereby increasing the ACF amplitude [38]. Choosing suitable instrumental settings that optimize and highlight the dynamic process under investigation is therefore always important in fluorescence correlation experiments. It was clearly shown in this study that TIR-FCS has high potential as a technique to study protein–surface interactions. For irreversible surface kinetics, assumed for proteins binding to the C18 surface, the fraction of un-occupied binding sites will change over time. This was seen when studying protein– surface interaction immediately after addition of TLL and after 30 min of incubation time (Fig. 2, right). After 30 min incubation, the surface interaction was no longer visible in the ACF, explained by a covered surface that hindered new binding of unbleached TLL molecules. Stated differently, irreversible binding of proteins makes the fraction of un-occupied binding sites β approach small values, which in turn makes the term in the ACF representing surface binding, GB , minimal. The temporal dependence of β therefore makes it hard to deduce the individual components, NFf and NBf in Eq. (2). In order to minimize the influence that protein packing have on the autocorrelation function, measurements were always performed during the first 1–3 minutes after pipetting sample onto the C18terminated glass.

455

In Fig. 3, the autocorrelation functions for 20 nM BSA without and with excess of 1 µM unlabeled BSA are shown. It was evident that no surface interaction of FITC-labeled BSA was seen when adding a large amount of unlabeled BSA. This is explained by competitive (irreversible) binding to the hydrophobic surface. Assuming equal binding kinetics for the labeled and unlabeled BSA, only a very small fraction of labeled BSA will be bound to the surface. Nonfluorescent BSA, being in high excess, will occupy almost all binding sites, thereby masking the term in the ACF representing surface binding. With 1 µM excess of unlabeled BSA, the ACF could be fitted to an analytical model describing only free diffusion, i.e. the second part of Eq. (1). The increase in the free diffusion time with excess BSA might be explained by an increase in viscosity (Fig. 3). When comparing TLL and BSA adsorption under equal conditions (immediately after addition and with no excess), it was found that TLL had a shorter correlation time; the ACF reaching zero at τ ≈ 0.1 s as compared too τ ≈ 1 s for BSA. Thus, the surface interaction rate RB for TLL was found to be higher. Since both proteins were labeled with the same fluorophore and thus the bleaching times (or bleaching rates) were equal this implies that TLL had a higher association rate. The quantification of the rate constant, kaf , was however never attempted in this study due to uncertainties in determining the concentration of FITC-TLL in the TIR-FCS detection volume (Eq. (3)). It was seen in the ACFs in Fig. 4 that BSA could be competed off from the C18 surface in the presence of surfactants. The surfactants studied were a mixture of nonionic C12 E6 and anionic LAS molecules (1:2 mol%, M ≈ 0.4 kDa). This system served as a model system for the main fraction in common household detergent formulations, and has been found to have a critical micelle concentration (cmc) of 100–200 µM in the glycine pH 9 buffer used [43]. At a concentration of 10 µM C12 E6 /LAS, no BSA surface interaction was detectable in the ACF. Hence, at this surfactant concentration, all BSA was completely competed off from the hydrophobic surface. The same observation has been made with, e.g., C12 E5 —displacement of other proteins under competitive adsorption conditions [9,11]. These studies suggested that the binding process was dominated by protein interactions at low C12 E5 concentration, but as the surfactant concentration was increased, the protein was completely hindered to adsorb. The competition of BSA off the hydrophobic surface was also clearly visible in the captured video-sequences of the TIR-illuminated area (cf. inset in Fig. 4). At low surfactant concentration protein binding events (bright spots) were clearly visible in each frame, but at 500 µM C12 E6 /LAS no binding events were seen. At 500 µM C12 E6 /LAS, which is above the critical micelle concentration of the surfactant system, the axial diffusion time increased about 50%. This trend indicated an increased viscosity or protein– surfactant complexes diffusing through the evanescent field. When studying TLL and C12 E6 /LAS competition, it was found that the surfactants also competed TLL off the hydrophobic surfaces (Fig. 5). However, at a surfactant concentration of 10 µM, the TLL surface interaction process was still visB compared to ible in the system, which resulted in a larger G BSA at the highest surfactant concentration. This suggested that

456

A.W. Sonesson et al. / Journal of Colloid and Interface Science 317 (2008) 449–457

TLL was more difficult to compete off the surface compared to BSA. Adsorption studies with TLL have previously shown that the protein adsorbs in large amount to hydrophobic surfaces [43–46] due to the hydrophobic active site region in the TLL structure. Moreover, the surface interaction rates, RB , increased slightly with surfactant concentration for both proteins, being most visible for TLL. The small amphiphilic surfactants, being in large excess (i.e., nearly saturating the surface binding sites), have a large association rate for the hydrophobic surface. Therefore kan An  kaf Af , kdn , kdf and the surface interaction rates RB equal the dissociation rate constants kdf of the proteins (Eq. (5)). The results indicated that the average time the proteins remained bound to the surface decreased in the presence of C12 E6 /LAS. Table 1 summarizes the surface dynamics of both protein–surfactant systems. It has earlier been shown with fluorescence recovery after photobleaching (FRAP) that the studied surfactant system boosted the lateral surface mobility of adsorbed TLL molecules on C18 surfaces [43]. This was explained by a surfactant-induced redistribution of lipases due to a desorption–rebinding process, i.e., lipases displaced from the surface by surfactants could be re-adsorbed to the surface without diffusion into the bulk. The data presented in this work supported such a model, i.e., that the presence of surfactants in the solution increased the surface interaction rate for both TLL and BSA, and thus decreased the average time the proteins were bound to the C18 surface. Hence, this effect could contribute to a higher apparent lateral long-range surface diffusion coefficient measured in the FRAP experiments. Both TIR-FCS and FRAP studies give an increased understanding of the dynamics of protein–surfactant systems at interfaces. This might help to further modify or engineer enzymes that can work optimally in surfactant rich environments. 5. Summary Protein–surfactant interactions on hydrophobic (C18-terminated) glass surfaces could be investigated with total internal reflection fluorescence correlation spectroscopy (TIR-FCS). Surface kinetics of two proteins, bovine serum albumin (BSA) and Thermomyces lanuginosus lipase (TLL), with a competing surfactant system (C12 E6 /LAS) could be quantified. It was found that with higher surfactant concentrations, the term in the autocorrelation function (ACF) representing surface binding of proteins decreased and that the surface interaction rate, in this case representing the protein dissociation rate, increased. This suggested that proteins were competed off the surface by the surfactants and that the average time the proteins were bound to the surface decreased. The results also indicated a higher affinity for TLL compared to BSA for the hydrophobic surface. At 10 µM C12 E6 /LAS concentration, lipase surface interactions were still visible, whereas for BSA, no contribution of binding kinetics to the ACF was seen. Acknowledgments The Swedish foundation for strategic research (BioX project), the European Community (Spotlite project) and Novozymes

A/S is acknowledged for financial support of the project. A personal scholarship awarded by Wenner–Grenska sällskapet to K.H. is also acknowledged. References [1] M. Wahlgren, C.A.C. Karlsson, S. Welin-Klintstroem, in: M. Malmsten (Ed.), Biopolymers at Interfaces, vol. 110, second ed., Marcel Dekker, New York, 2003, pp. 321–343. [2] K.P. Ananthapadmanabhan, in: E.D. Goddard, K.P. Ananthapadmanabhan (Eds.), Interactions of Surfactants with Polymers and Proteins, CRC Press, Boca Raton, 1993, pp. 319–365. [3] N.J. Turro, X.-G. Lei, K.P. Ananthapadmanabhan, M. Aronson, Langmuir 11 (1995) 2525–2533. [4] A.S. Antipova, M.G. Semenova, L.E. Belyakova, M.M. Il’in, Colloids Surf. B Biointerfaces 21 (2001) 217–230. [5] R. Miller, V.B. Fainerman, A.V. Makievski, J. Krägel, D.O. Grigoriev, V.N. Kazakov, O.V. Sinyachenko, Adv. Colloid Interface Sci. 86 (2000) 39–82. [6] L. Brady, A.M. Brzozowski, Z.S. Derewenda, E. Dodson, S. Tolley, J.P. Turkenburg, L. Christiansen, B. Huge-Jensen, L. Nørskov, L. Thim, U. Menge, Nature 343 (1990) 767–770. [7] D.N. Rubingh, Curr. Opin. Colloid Interface Sci. 1 (1996) 598–603. [8] P. Messina, G. Prieto, V. Dodero, M.A. Cabrerizo-Vilchez, J. MaldonadoValderrama, J.M. Ruso, F. Sarimento, Biopolymers 82 (2006) 261–271. [9] R.J. Green, T.J. Su, J.R. Lu, J. Webster, J. Penfold, Phys. Chem. Chem. Phys. 2 (2000) 5222–5229. [10] J. Krägel, M. O’Neill, A.V. Makievski, M. Michel, M.E. Leser, R. Miller, Colloids Surf. B Biointerfaces 31 (2003) 107–114. [11] M.C. Wahlgren, S. Welin-Klintstroem, T. Arnebrant, A. Askendahl, H. Elwing, Colloids Surf. B Biointerfaces 4 (1995) 23–31. [12] S. McClellan, E.I. Franses, Langmuir 21 (2005) 10148–10153. [13] A.P. Gunning, A.R. Mackie, P.J. Wilde, V.J. Morris, Langmuir 15 (1999) 4636–4640. [14] A.R. Mackie, A.P. Gunning, P.J. Wilde, V.J. Morris, Langmuir 16 (2000) 8176–8181. [15] P. Schwille, Cell Biochem. Biophys. 34 (2001) 383–408. [16] S. Maiti, H. Ulrich, W.W. Webb, Proc. Natl. Acad. Sci. USA 94 (1997) 11753–11757. [17] N.L. Thompson, in: J. Lakowicz (Ed.), Topics in Fluorescence Spectroscopy, vol. 1, Plenum Press, New York, 1991. [18] R. Rigler, E.L. Elson, Fluorescence Correlation Spectroscopy, SpringerVerlag, Berlin, 2001. [19] J. Widengren, U. Mets, in: C. Zander, J. Enderlein, R. Keller (Eds.), Single Molecule Detection in Solution, Wiley–VHC, Berlin, 2002. [20] R.L. Hansen, J.M. Harris, Anal. Chem. 70 (1998) 4247–4256. [21] R.L. Hansen, J.M. Harris, Anal. Chem. 70 (1998) 2565–2575. [22] K.S. McCain, P. Schluesche, J.M. Harris, Anal. Chem. 76 (2004) 930–938. [23] K.S. McCain, J.M. Harris, Anal. Chem. 75 (2003) 3616–3624. [24] A.M. Lieto, R.C. Cush, N.L. Thompson, Biophys. J. 85 (2003) 3294– 3302. [25] T.E. Starr, N.L. Thompson, J. Phys. Chem. B 106 (2002) 2365–2371. [26] J.K. Pero, E.M. Haas, N.L. Thompson, J. Phys. Chem. B 110 (2006) 10910–10918. [27] Y. Ohsugi, K. Saito, M. Tamura, M. Kinjo, Biophys. J. 91 (2006) 3456– 3464. [28] N.L. Thompson, T.P. Burghardt, D. Axelrod, Biophys. J. 33 (1981) 435– 454. [29] N.L. Thompson, D. Axelrod, Biophys. J. 43 (1983) 103–114. [30] H. Blom, M. Gösch, Curr. Pharm. Biotech. 7 (2006) 51–66. [31] W. Norde, in: M. Malmsten (Ed.), Biopolymers at Interfaces, vol. 110, second ed., Marcel Dekker, New York, 2003. [32] M. Leutenegger, H. Blom, J. Widengren, C. Eggeling, M. Gösch, R.A. Letgeb, T.J. Lasser, J. Biomed. Opt. 11 (2006) 040502-040501/040503. [33] T.E. Starr, N.L. Thompson, Biophys. J. 80 (2001) 1575–1584. [34] A.M. Lieto, N.L. Thompson, Biophys. J. 8 (2004) 1268–1278. [35] K. Hassler, M. Leutenegger, R. Rigler, R. Rao, R. Rigler, M. Gösch, T. Lasser, Opt. Exp. 13 (2005) 7415–7423.

A.W. Sonesson et al. / Journal of Colloid and Interface Science 317 (2008) 449–457

[36] K. Hassler, T. Anhut, R. Rigler, M. Gösch, T. Lasser, Biophys. J. 88 (2005) L01–L03. [37] N.L. Thompson, Biophys. J. 38 (1982) 327–329. [38] J. Widengren, U. Mets, R. Rigler, J. Phys. Chem. 99 (1995) 13368–13379. [39] J. Widengren, R. Rigler, Bioimaging 4 (1996) 149–157. [40] C. Eggeling, J. Widengren, R. Rigler, C.A.M. Seidel, Anal. Chem. 70 (1998) 2651–2659. [41] C.A. Haynes, W. Norde, Colloids Surf. B Biointerfaces 2 (1994) 517– 566.

457

[42] J. Widengren, P. Thyberg, Cytometry 68A (2005) 101–112. [43] A.W. Sonesson, U.M. Elofsson, H. Brismar, T.H. Callisen, Langmuir 22 (2006) 5810–5817. [44] K. Wannerberger, T. Arnebrant, J. Colloid Interface Sci. 177 (1996) 316– 324. [45] A.W. Sonesson, T.H. Callisen, H. Brismar, U.M. Elofsson, Langmuir 21 (2005) 11949–11956. [46] A.W. Sonesson, T.H. Callisen, H. Brismar, U.M. Elofsson, Colloids Surf. B Biointerfaces 54 (2007) 236–240.