Lipase Surface Diffusion Studied by Fluorescence Recovery after

in free solution has, according to the Stokes−Einstein relation, a diffusion coefficient of about 10-6 cm2/s ...... Jervis, E. J.; Haynes, C. A...
1 downloads 0 Views 337KB Size
Download PDF
Langmuir 2005, 21, 11949-11956

11949

Lipase Surface Diffusion Studied by Fluorescence Recovery after Photobleaching Andreas W. Sonesson,†,‡ Thomas H. Callisen,§ Hjalmar Brismar,*,‡ and Ulla M. Elofsson† YKI, Institute for Surface Chemistry, Stockholm, Sweden, Novozymes A/S, Bagsvaerd, Denmark, and Department of Cell Physics, Royal Institute of Technology, SE-106 91 Stockholm, Sweden Received July 1, 2005 We have analyzed surface diffusion properties of a variant of Thermomyces lanuginosa lipase (TLL) on hydrophilic silica and silica methylated with dichlorodimethylsilane (DDS) or octadecyltrichlorosilane (OTS). For this study a novel method for analysis of diffusion on solid surfaces was developed. The method is based on fluorescence recovery after photobleaching using confocal microscopy. When a rectangular area of the sample was photobleached, fluorescence recovery could be analyzed as one-dimensional diffusion, resulting in simplified mathematical expressions for fitting the data. The method was initially tested by measuring bovine serum albumin diffusion on glass, which led to a diffusion coefficient in good correspondence to earlier reports. For the analysis of TLL diffusion, ellipsometry data of TLL adsorption were used to calibrate fluorescence intensity to surface density of lipase, enabling measurements of the diffusion coefficient at different surface densities. The average diffusion coefficient was calculated in two time intervals after adsorption. Mobile fraction and diffusion coefficient were lowest on the OTS surface, when extrapolated to infinite surface dilution. Moreover, the diffusion rate decreased with time on the hydrophobic surfaces. Our observations can be explained by the surface dependence on the distribution of orientations and conformations of adsorbed TLL, where the transition from the closed to the catalytically active open and more hydrophobic structure is important.

1. Introduction Lateral diffusion of proteins at interfaces is important for many biological systems. It leads to enhanced receptorligand recognition in the cell membrane and is relevant for understanding the localization of different membrane receptors.1 The observation that proteins irreversibly adsorbed on solid surfaces indeed can be mobile and diffuse was first reported with bovine serum albumin (BSA) on coverslip glass,2 and the phenomenon has recently been extensively reviewed.3 Lateral mobility also enables efficient packing of an adsorbed protein layer, thereby increasing the surface concentration. It is evident that protein properties, e.g., conformational flexibility and molecular weight, combined with surface properties, e.g., hydrophobicity and charge, affect the diffusion rate, but the exact mechanism is not well explained. Estimated surface diffusion coefficients for different systems are difficult to compare since they have been found to be strongly dependent on the surface density of proteins, which arises from lateral repulsion between adsorbed protein molecules.4 However, the diffusion at infinite surface dilution, D0, is only dependent on the proteinsurface interactions. Therefore, the ideal method is to do * Corresponding author. E-mail: [email protected]. † YKI, Institute for Surface Chemistry. ‡ Royal Institute of Technology. § Novozymes A/S. (1) Axelrod, D.; Ravdin, P.; Koppel, D. E.; Schlessinger, J.; Webb, W. W.; Elson, E. L.; Podleski, T. R. Proc. Natl. Acad. Sci. U.S.A. 1976, 73, 4594-4598. (2) Michaeli, I.; Absolom, D. R.; van Oss, C. J. J. Colloid Interface Sci. 1980, 77, 586-587. (3) Tilton, R. D. Mobility of Biomolecules at Interfaces. In Bioploymers at Interfaces, 2nd ed.; Marcel Dekker: New York, 2003; Vol. 110, pp 221-257. (4) Tilton, R. D.; Gast, A. P.; Robertson, C. R. Biophys. J. 1990, 58, 1321-1326.

repeated measurements on surfaces with different densities of proteins, enabling an extrapolation to zero surface density. The common method to study surface diffusion is fluorescence recovery after photobleaching (FRAP). The technique was first used for studying diffusion in lipid monolayers and bilayers5-7 and is now most widely used in pharmaceutical research8 and cell biology.9,10 The procedure implies photobleaching of a defined area on a fluorescent sample with a high-intensity laser, creating a sharp step in detected fluorescence signal. With time, fluorescent molecules will diffuse into the bleached region and bleached molecules to the unbleached region, and the difference between the two regions will be reduced. The fluorescence recovery is sampled and fitted to the solution of the diffusion equation with the specific boundary- and initial conditions of the bleached area. This means that both the diffusion coefficient D and the mobile and immobile fractions of the diffusive species can be estimated.3 A common bleaching procedure when studying protein diffusion at interfaces is to use two coherent laser beams to form a fringe pattern on the surface, so-called fluorescence recovery after pattern photobleaching (FRAPP).11,12 This can be used in combination with total internal reflection fluorescence (TIRF),13,14 so that only (5) Edidin, M.; Zagyansky, Y.; Lardner, T. J. Science 1976, 191, 466468. (6) Liebman, P. A.; Entine, G. Science 1974, 185, 457-459. (7) Poo, M.; Cone, R. A. Nature 1974, 247, 438-441. (8) Meyvis, T. K. L.; De Smedt, S. C.; Van Oostveldt, P.; Demeester, J. Pharm. Res. 1999, 16, 1153-1162. (9) Reits, E. A. J.; Neefjes, J. J. Nat. Cell Biol. 2001, 3, E145-E147. (10) White, J.; Stelzer, E. Trends Cell Biol. 1999, 9, 61-65. (11) Tanaka, K.; Yu, H. Langmuir 2002, 18, 797-804. (12) Kim, S.; Yu, H. J. Phys. Chem. 1992, 96, 4034-4040. (13) Tilton, R. D.; Robertson, C. R.; Gast, A. P. J. Colloid Interface Sci. 1990, 137, 192-203.

10.1021/la051773+ CCC: $30.25 © 2005 American Chemical Society Published on Web 11/09/2005

11950

Langmuir, Vol. 21, No. 25, 2005

molecules near the surface contribute to the detected fluorescence signal. The aim of this study was to implement a novel FRAP-based method for bleaching and data evaluation to quantify self-diffusion coefficients of adsorbed proteins on solid surfaces. The surface density and time dependence of adsorbed protein lateral mobility was also studied, something that has received very little attention in the literature. To do this, we used confocal laser scanning microscopy (CLSM) for imaging the adsorbed protein layer and bleaching one large rectangular area of the sample so that the process could be evaluated as a one-dimensional diffusion from an initial spatial step in fluorescence. Using a flow cell in the CLSM allowed rinsing of the surface with buffer after adsorption so that the recovery was limited to lateral diffusion, not subsequent adsorption of fluorescent molecules from the bulk solution. To test the method, diffusion measurements of BSA on a glass surface was performed. This system has been studied independently by at least two different groups15,16 using the FRAPP technique. After the FRAP method was implemented, the diffusion properties of the lipase from the fungus Thermomyces (formerly Humicola) lanuginosa (TLL) on solid surfaces were studied. Ellipsometry was used to calibrate fluorescence intensity to relative surface density of lipase. Although the adsorption behavior of lipases is extensively studied,17-19 there are to our knowledge no diffusion studies of any lipase on solid supports. The dynamics of lipases at interfaces is particularly important in technical applications such as detergency. TLL may upon association with a hydrophobic surface undergo a conformational change where a lid domain rolls over and exposes the hydrophobic active site region, a so-called interfacial activation.20 TLL is known to adsorb in different amounts on surfaces with different wettability.21 Moreover, the activity toward a substrate in solution has been strongly dependent on the hydrophobicity of the surface, indicating different orientations of adsorbed TLL on different surfaces.22 Therefore, it is interesting to measure TLL surface diffusion on surfaces of different hydrophobicity. The model surfaces used in this work were untreated, hydrophilic silica, and silica methylated with dicholrodimethylsilane (DDS) or octadecyltrichlorosilane (OTS). 2. Materials and Methods 2.1. Materials. Dichlorodimethylsilane, DDS (catalog no. 8.03452.0250), HCl (25%), H2O2 (30%) NH3 (25%), p-xylene (>99%), chloroform (>99%), and sulfuric acid (95-97%) were from Merck, toluene (>99.5%) was from Fluka, OTS (catalog no.104817-25G) and dimethylformamide (99.9%) were from Sigma-Aldrich. Fluorescein isothiocyanate, FITC (catalog no. F-143, lot. 1173-1), and Alexa Fluor 488 protein labeling kits were purchased from Molecular Probes Europe BV (Leiden, The Netherlands) and Sephadex G-25 M PD-10 columns from (14) Gaspers, P. G.; Robertson, C. R.; Gast, A. P., Langmuir 1994, 10, 2699-2704. (15) Chan, V.; Graves, D. J.; Fortina, P.; McKenzie, S. E. Langmuir 1997, 13, 320-329. (16) Yang, Z.; Galloway, J. A.; Yu, H. Langmuir 1999, 15, 84058411. (17) Hedin, E. M. K.; Hoyrup, P.; Patkar, S. A.; Vind, J.; Svendsen, A.; Fransson, L.; Hult, K. Biochemistry 2002, 41, 14185-14196. (18) Noinville, S.; Revault, M.; Baron, M.-H.; Tiss, A.; Yapoudjian, S.; Ivanova, M.; Verger, R. Biophys. J. 2002, 82, 2709-2719. (19) Yapoudjian, S.; Ivanova, M.; Douchet, I.; Zenatti, A.; Sentis, M.; Marine, W.; Svendsen, A.; Verger, R. Biopolymers 2002, 65, 121-128. (20) Schmid, R. D.; Verger, R. Angew. Chem., Int. Ed. Engl. 1998, 37, 1608-1633. (21) Wannerberger, K.; Arnebrant, T. J. Colloid Interface Sci. 1996, 177, 316-324. (22) Wannerberger, K.; Welin-Klintstrom, S.; Arnebrant, T. Langmuir 1997, 13, 784-790.

Sonesson et al.

Figure 1. Surfaces with different hydrophobicity used in diffusion experiments with TLL: untreated, hydrophilic silica (A), silica methylated with dichlorodimethylsilane (DDS) (B), and silica methylated with octadecyltrichlorosilane (OTS) (C). Amersham Biosciences AB (Uppsala, Sweden). A variant of the lipase from Thermomyces lanuginosa (TLL) was provided by Novozymes A/S (Bagsvaerd, Denmark), and bovine serum albumin, BSA (catalog no. A-4503, lot 129H0913, Mw ) 66 000), was purchased from Sigma. Buffers used were glycine pH 9.0 (10 mM NaCl, 0.05 mM EDTA, 50 mM glycine, 1 mM NaN3), PBS pH 7.0 (10 mM, NaH2PO4 + NaCl), and sodium bicarbonate pH 9 (50 mM, NaHCO3 + Na2CO3). All buffer salts were of analytical grade and all water was of Milli-Q grade. 2.2. Surface Preparations. Coverslip glasses (MenzelGla¨ser, Braunschweig, Germany), used in diffusion studies of BSA, were cleaned in 2:1 (v/v) sulfuric acid/H2O2 at 130 °C for 10 min and then thoroughly rinsed in water. Silica wafers, cut in appropriate sizes for the CLSM flow chamber (13 × 18 mm), were cleaned for 5 min in 80 °C 5:1:1 (v/v/v) H2O/NH3/H2O2, thoroughly rinsed in water, and then cleaned for 5 min in 80 °C 5:1:1 (v/v/v) H2O/HCl/H2O2. The wafers were then rinsed several times with water and ethanol and blown dry with nitrogen. Contact angles for the cleaned silica surfaces were 21 ( 2°. The procedure for creating surfaces of intermediate hydrophobicity with desirable smoothness and homogeneity has been described elsewhere.22,23 Silica surfaces, cleaned as described above, were rinsed in ethanol and p-xylene and immersed in an unstirred solution of 0.1% (v/v) DDS in p-xylene for 20 min. The surfaces were then cleaned in p-xylene and ethanol and stored in ethanol. Surfaces treated in this way had a contact angle of 59 ( 3°. The OTS surfaces were prepared by immersing freshly cleaned wafers in an unstirred solution of 0.5% (v/v) OTS in toluene for 24 h. The wafers were then removed from the silane solution, rinsed in chloroform, and put in a chloroform bath for 2 min. Finally, the wafers were rinsed with ethanol and water and were blown dry with nitrogen. This procedure led to hydrophobic surfaces with a contact angle of 105 ( 1°. The surface roughness was measured with atomic force microscopy to 1.6 ( 0.2 nm. The hydrophilic and silanized silica surfaces can be seen in Figure 1. All surfaces were stored in ethanol and rinsed in both ethanol and water prior to use. 2.3. Protein Labeling. BSA was dissolved in sodium bicarbonate buffer and mixed with FITC in DMF to a molar ratio of approximately 5:3 (FITC/BSA). The mixture was protected from light and incubated at room temperature for 2 h with no stirring. A Sephadex G-25 M PD-10 column preequilibrated with PBS buffer was used to remove unreacted FITC. This led to a degree of labeling of 0.92 mol of FITC/mol of BSA, estimated with a Bio-Rad protein assay (Bio-Rad Laboratories, catalog no. 500006) and UV spectroscopy, FITC ) 196 cm-1 (mg/mL)-1 at 495 nm. Labeled BSA was stored in aliquots at -20 °C for up to 6 months after labeling. Fractions were diluted in PBS to appropriate concentration prior to use. TLL was labeled with Alexa Fluor 488 dyes according to the labeling kit provided by the manufacturer (http://www.probes.com). To determine the protein and fluorophore concentration, the absorbance at 280 and 494 nm was measured, using TLL,280nm (23) Elwing, H.; Welin, S.; Askendal, A.; Nilsson, U.; Lundstro¨m, I. J. Colloid Interface Sci. 1987, 119, 203-210.

Lipase Surface Diffusion Studied by FRAP

Langmuir, Vol. 21, No. 25, 2005 11951

) 38440 cm-1 M-1 and Alexa488,494nm ) 71000 cm-1 M-1, respectively. This led to a degree of labeling of 0.6 mol dye/mol of enzyme. The Alexa Fluor 488 reactive dye has a tetrafluorophenyl (TFP) ester moiety that reacts with primary amines of the enzyme, most often lysine, and has absorption and fluorescence emission maxima at 494 and 519 nm, respectively. Labeled lipase was stored in aliquots at -20 °C for up to 6 months after labeling. Fractions were thawed for 1 h and diluted in glycine buffer to appropriate concentration prior to use. Thawed samples were stored at +6 °C and used for two consecutive days. 2.4. Ellipsometry Experiments. Ellipsometry measurements were performed using a Rudolph thin film ellipsometer, type 436 (Rudolph Research, Fairfield, NJ). A xenon arc lamp with a wavelength of 4015 Å was used as light source at an angle of incidence of 67.7°. The adsorption was measured in situ in a cuvette with a solution volume of 3 mL with continuous stirring under nonflow conditions. A more detailed description of the instrumental setup has been published elsewhere.24 2.5. FRAP Experiments. An inverted Axiovert 100M microscope with a Zeiss LSM Pascal scanner was used for FRAP experiments, with a 40 × 1.3 oil objective and an Ar 488 laser. A flow chamber designed for surfaces of the size 13 × 18 mm was constructed and mounted in the microscope. A 100 µm thick silicon rubber and a coverslip glass sealed the chamber, limiting the volume to about 15 µL. A flow system consisting of a Minipuls 2 peristaltic pump (Gilson, Villiers le Bel, France) in conjunction with a six-port injection valve (Skandinaviska Genetec AB, Go¨teborg, Sweden) was used to pump buffer through the chamber. Protein samples were loaded in a 50 µL injection loop in the valve and could thereafter be injected to the flow system. An InSpeck Green(505/515) microscope image intensity calibration kit with 6 µm microspheres purchased from Molecular Probes Europe BV (Leiden, The Netherlands) was used to control the effect of the Ar 488 laser and the PM tube voltage on the detected fluorescence and verify that there was no drift between experiments. Diffusion measurements were performed by injecting a protein sample to the chamber and stopping the flow during the time of adsorption to the surface. Before recovery measurements, rinsing of the chamber for 2 min with buffer solution took place to ensure that the fluorescence recovery was only due to lateral motion and not subsequent adsorption of protein from solution. A defined rectangular area of the surface (about 60 × 220 µm) was bleached for 2 min, using the Ar 488 laser with 100% intensity. Recovery micrographs were sampled under no-flow conditions every fourth minute for 90 min after bleaching. To compensate for possible tilts and inhomogeneities of the surface, the pinhole was fully opened during the recovery. Moreover, to compensate for a possible drift in focus, a z-stack was recorded at every sampled time so that the picture in focus could be picked out and analyzed. After the first 90 min, the FRAP procedure was repeated on another spot on the same surface. The diffusion coefficient was averaged over 0-90 min after adsorption (first bleaching) and over 90-180 min after adsorption (second bleaching), denoted Dt)0-90 and Dt)90-180, respectively. For analysis of immobile and mobile fractions, a square (about 20 × 20 µm) of the surface was bleached. The average intensity of the bleached square and the average intensity of the surrounding unbleached region were measured until the ratio between the two areas reached a plateau value. All measurements were performed at room temperature. 2.6. Analysis of FRAP Data. The large size of the bleached area made it possible to average and analyze the intensity profile I(x,t) of the middle section as a one-dimensional diffusion process from a step function (Figure 2A,B). I(x,t) is the sum of the contributions from the mobile fraction f and the immobile fraction (1 - f) (eq 1)

I(x,t) ) f F(x,t) + (1 - f) H(x)

(1)

F(x,t) is the solution of the diffusion equation for a onedimensional diffusion from a step function25 and H(x) is the initial (24) Landgren, M.; Jo¨nsson, B. J. Phys. Chem. 1993, 97, 1656-1660. (25) Weiss, T. F. Diffusion. In Cellular Biophysics: Transport; The MIT Press: Cambridge, MA, 1996; Vol. 1, pp 83-185.

Figure 2. Data analysis of the FRAP experiments. The middle section of the bleached area is averaged and analyzed as a step function (A and B). Using eq 4, the (maximum slope/f)-2 is plotted versus time to extract the diffusion coefficient D (C). The distance per pixel is 0.45 µm. step function, which leads to the following expression of I(x,t) (eq 2)

I(x,t) ) f I0

∫

x/(2Dt)1/2

-∞

2 1 e-y /2 dy + (1 - f) I0H(x) (2) (2π)1/2

Normalizing to the intensity of the unbleached area, I0, and differentiating eq 2 leads to eq 3

∂I(x,t) e-x /4Dt )f + (1 - f) ∂(x) ∂x (4πDt)1/2 2

(3)

Using a median filter operator removes the contribution from the immobile fraction and hence, the maximum slope between the two areas becomes a simple expression of f, D, and t (eq 4)

| | ∂I(x,t) ∂x

) x)0

f (4πDt)1/2

(4)

11952

Langmuir, Vol. 21, No. 25, 2005

Sonesson et al.

Table 1. Diffusion Coefficients and Mobile Fractions of Adsorbed BSA on Glass

protein BSAa BSAb BSAc a

bulk concn, µM

D (cm2/s)

mobile fraction (f)

3.0 ≈1 1.5 3.0

(2.6 ( 2) × 10-10 (n ) 3) (2.2 ( 0.3) × 10-10 (3.3 ( 0.2) × 10-9 (2.7 ( 0.1) × 10-9

0.47 ( 0.04 (n ) 3) 0.59 ( 0.06 0.28 ( 0.02 0.20 ( 0.03

This work. b Yang et al. (ref 16). c Chan et al. (ref 15).

Images and intensity profiles were processed in softwares ImageJ 1.31v and Matlab 6.5. The derivative was calculated with a discrete differentiation operator. By use of linear leastsquares regression, (maximum slope/f)-2 was plotted versus time to estimate D for the different diffusion processes (Figure 2C). The diffusion coefficient at infinite surface dilution, D0, was extrapolated using an exponential fit of estimated D at different surface densities which has been suggested from theoretical work on surface diffusion.26 2.7. Calibration of Intensity to Surface Density. Samples with the same total TLL concentration, 0.3 µM, but with different fractions of the labeled population were used to calibrate fluorescence intensity in the CLSM to surface density of TLL. Samples were injected into the CLSM chamber according to the usual protocol, adsorbed for 300 s, and followed by a 2 min rinse with glycine buffer. The intensity at the surfaces after adsorption of each sample was measured as the average of ten 220 × 220 µm squares. Adsorption results from the ellipsometer gave the maximum possible density, θmax, and the measurements could be used for a calibration graph of intensity vs surface density.

3. Results 3.1. BSA Diffusion. For the BSA experiments, FITClabeled BSA at a concentration of 3 µM was injected to the chamber and allowed to adsorb for 1 h before a 10 s rinse with PBS buffer. FRAP was then performed on the BSA adsorbed on the cover slip glass. The results of both diffusion coefficient D as well as mobile fraction from this study and two earlier reports of BSA diffusion on glass15,16 are given in Table 1. A recovery profile for BSA on glass can be viewed in Figure 3, where the recovery of a small bleached square is shown, reaching its plateau value after 80 min. 3.2. TLL Adsorption. The adsorbed amount of TLL at OTS surfaces in glycine buffer versus time at different concentrations (0.1-3 µM) is shown in Figure 4. The adsorption was monitored during 5000 s and rinsing was performed after 4500 or 40 s. The same plateau value in adsorbed amount, 2.2 mg/m2, was reached for all bulk solution concentrations of TLL. No desorption occurred upon rinsing after 40 s at the 0.3 µM concentration (Figure 4B) or after 4500 s at any concentration. 3.3. Calibration of Intensity to Surface Density. A graph of average intensity as a function of the fraction of the labeled TLL population, with a degree of labeling of 0.6, can be seen in Figure 5. The TLL concentration and adsorption time were chosen to give the maximum adsorbed amount, 2.2 mg/m2, according to the ellipsometer measurements (Figure 4). This corresponded to a surface density θmax ) 0.75, as calculated by using an average TLL radius of 23 Å and a TLL Mw of 31 000.27 All samples used for the calibration graph had the same total protein concentration and therefore led to the same θmax, but the surfaces gave intensities relative to the fraction of labeled TLL. When the fraction of labeled TLL was decreased, (26) Minton, A. P. Biophys. J. 1989, 55, 805-808. (27) Brzozowski, A. M.; Savage, H.; Verma, C. S.; Turkenburg, J. P.; Lawson, D. M.; Svendsen, A.; Patkar, S. Biochemistry 2000, 39, 1507115082.

Figure 3. A recovery profile for a FRAP experiment with BSA on glass. A small area A was bleached at time ) 0 and the intensity sampled until the ratio between the bleached and unbleached area reached plateau value. This kind of graph was used to estimate the mobile population of adsorbed proteins. The values are normalized to the intensity value found before bleaching.

the intensity decreased linearly down to the origin of coordinates, i.e., the value found with only an unlabeled population. The labeled fraction of adsorbed TLL corresponds to a surface coverage directly proportional to the fraction of labeled TLL in the bulk solution. Thus, the fluorescence intensity from the surface can directly be related to the relative surface density, θ/θmax, and could in this way be estimated for every FRAP experiment using the calibration graph. 3.4. TLL Diffusion. FRAP experiments were performed with TLL concentrations of 0.1, 0.3, or 3 µM with adsorption times of 10, 20, or 40 s, followed by a 2 min rinse with glycine buffer. The intensity from the adsorbed proteins was calibrated to surface densities θ/θmax between 0.1 and 1.0. Surfaces with θ/θmax < 0.1 did not have a homogeneous intensity signal, preventing bleaching to a satisfying step function. A representative plot of the (maximum slope/f)-2 of the intensity profile versus time on a hydrophilic silica surface with a relative surface density of 0.1 is seen in Figure 2C. The diffusion coefficient was extracted from the slope of the curve. In Figure 6 the diffusion coefficients extracted from the FRAP measurements at different surface density on hydrophilic silica, DDS, and OTS surfaces are shown. For every type of surface, the diffusion coefficient at infinite surface dilution, D0, was extrapolated for both Dt)0-90 and Dt)90-180 (Figure 6 and Table 2). D0,t)0-90 on the hydrophilic silica surface and the DDS surface was similar, 2.7 × 10-11 cm2/s (R ) 0.94 and R ) 0.91, respectively) but significantly lower on the OTS surface, 0.8 × 10-11 cm2/s (R ) 0.81). There was a noticeable decrease in mobility with time at low surface density on DDS and OTS surfaces, i.e., D0,t)90-180 < D0,t)0-90. There was no noticeable decrease of the slope of

Lipase Surface Diffusion Studied by FRAP

Langmuir, Vol. 21, No. 25, 2005 11953

Figure 4. TLL adsorption (b) on the OTS surface with three different concentrations: 0.1 µM (A), 0.3 µM, (B), and 3 µM (C). With 0.3 µM, rinsing was also performed 40 s after adsorption (∆). All bulk concentrations led to the same plateau value, 2.2 mg/m2, and no desorption occurred due to rinsing after 40 s.

Figure 5. Calibration graph of intensity as function of the fraction of the labeled lipase population with a degree of labeling of 0.6. The linearity through the origin of coordinates suggests that unlabeled and labeled lipase adsorb in the same manner. This graph was used to directly convert fluorescence intensity signal to relative surface density, θ/θmax, of TLL.

the intensity profile when D < 1 × 10-12 cm2/s. Therefore, D < 1 × 10-12 cm2/s was considered as immobile in this study. Mobile fractions were measured on the surfaces with the lowest surface density, and recovery profiles were sampled until a plateau value was reached. This process took 3-6 h, depending on the surface used and the size of the bleached square. The mobile population was lower on OTS surfaces compared to hydrophilic silica and DDS surfaces (Table 1). A control experiment was performed to ensure that no subsequent adsorption from tubes or the inner sides of the flow cell contributed to the recovery signal. After lipase was adsorbed according to the normal protocol, the substrate surface was replaced by a clean surface, and the system was allowed to stand for 8 h to establish that no signal from the surface arising from molecules detached from tubing and chamber walls and readsorbed on the surface could be detected. 4. Discussion The aim of this study was to quantify lipase diffusion on surfaces of different hydrophobicity. An alternative FRAP method instead of the common, but rather com-

Figure 6. Diffusion coefficient of TLL as function of surface density on the untreated, hydrophilic silica surface (A), the DDS surface (B), and the OTS surface (C). D is averaged for both 0-90 min after adsorption (Dt)0-90 ) b) and 90-180 min after adsorption (Dt)90-180 ) ∆). The diffusion coefficient at infinite dilution, D0, is estimated using an exponential fit to the data.

11954

Langmuir, Vol. 21, No. 25, 2005

Sonesson et al.

Table 2. Diffusion Coefficients and Mobile Fractions of Adsorbed TLL protein TLL TLL TLL

surface

D0,t)0-90 (cm2/s)

D0,t)90-180 (cm2/s)

mobile fraction (f)a

Si 2.7 × 10-11 2.0 × 10-11 0.58 ( 0.07 (n ) 3) Si-DDS 2.7 × 10-11 1.1 × 10-11 0.63 ( 0.04 (n ) 2) Si-OTS 0.8 × 10-11 0.5 × 10-11 0.45 ( 0.04 (n ) 2)

a n ) the number of measurements performed to determine the mobile fraction.

plicated, FRAPP or TIRF-FRAPP techniques was developed. Some basic advantages in using this method could be seen. First of all, the data analysis when seeing the process as a one-dimensional diffusion from a step between bleached and unbleached areas is more straightforward compared to evaluating the fringe pattern of FRAPP data. Thus, the presented method involves data to be fitted only to linear expressions. Similar bleaching procedures with subsequent data evaluation to extract diffusion coefficients have been reported,28,29 but they also involved the use of more complex curve fitting operations. Compared to using a TIRF, the CLSM has the ability to image the adsorbed protein layer before bleaching so that inhomogeneities in the surface caused by contamination etc. can be avoided. Here, the diffusion was averaged over a distance of 100200 µm, which is well in the same range as FRAPP, that measures diffusion on a 100-500 µm distance of the surface,11,15 or TIRF-FRAPP, that illuminates about a 1 mm region of the surface.13 Thus, the region used to estimate D with the CLSM method should be well representative for the surface. There are, however, some potential limitations to the method. The one-dimensional approximation is only valid for systems of very slow diffusion, so that proteins from the edges of the bleached square do not interfere with the analyzed region. Estimation of the time it takes for diffusion to occur over a certain distance, the following approximation can be used (eq 5)25

t1/2 ) x1/22/D

(5)

t1/2 is the time for half of the particles to travel at least the distance x1/2. With t1/2 ) 90 min and the maximum diffusion coefficient found in this study (D ) 2 × 10-10 cm2/s), x1/2 is equal to 10 µm. Hence, for diffusion processes at this rate, the bleached square of 60 × 220 µm can be considered of infinite size. The bleaching laser beam has a Gaussian intensity profile resulting in a Gaussianshaped bleach profile.30 Scanning of successive lines to form the rectangular bleached region results in a summation of several Gaussian-shaped bleach profiles. However, the integrated effect at the edge of the bleached region results in a bleach profile of the same shape as the diffusion itself. With a radial resolution of 250 nm, this effect corresponds to the effect of diffusion during 8 s of a substance with diffusion coefficient 10-11 cm2/s. Combined with the diffusion that takes place during the 2 min bleaching period, the initial intensity profile over the edge differs from the ideal step function. This will lead to a broadening of the Dirac function (∂x) in the differentiated intensity profile (eq 3). However, as a consequence of the selected pixel size (450 nm) and after median filtering of the data, the broadened Dirac function will have no influence on the result of the data analysis. (28) Yuan, Y.; Velev, O. D.; Lenhoff, A. M. Langmuir 2003, 19, 37053711. (29) Jervis, E. J.; Haynes, C. A.; Kilbourn, D. G. J. Biol. Chem. 1997, 272, 24016-24023. (30) Braeckmans, K.; Peeters, L.; N., S. N.; De Smedt, S. C.; Demeester, J. Biophys. J. 2003, 85, 2240-2252.

To test the method, we used BSA, a well-studied protein and one of very few for which diffusion data from different methods exist. The diffusion coefficient for this protein has been estimated to be in the range 10-10 to 10-8 cm2/s3, depending on the surface used. In this work, the surface diffusion of BSA was analyzed on coverslip glass in a flow chamber for the CLSM. The protocol involved a 1-h adsorption of 3 µM FITC-BSA, followed by rinsing with PBS buffer. With the CLSM-FRAP method, D for BSA on glass could be estimated to (2.6 ( 2) × 10-10 cm2/s with a mobile population of 0.46 (Table 1). There are at least two other independent reports of BSA diffusion on glass, with similar experimental procedures. Yang et al. (1999) soaked their coverslip glass in1 µM FITC-BSA solution for 1 h, rinsed with buffer and mounted the surface in a Teflon chamber with maintained saturated water vapor pressure, and analyzed the surface diffusion after 24 h with FRAPP. Chan et al. (1997) used a probe-clip fluid cell with adsorption of 1.5-3 µM FITC-BSA for 1 h and analyzed the diffusion without rinsing, using FRAPP. The results on BSA diffusion from these reports are summarized together with the result from this work in Table 1. As seen, the results by Yang et al. correlate well with the results from this study, whereas the work by Chan et al. gave faster D and lower mobile fractions. The deviations between the results probably arise from differences in methodology. Rinsing of the surface with buffer after adsorption ensures that no subsequent exchange process could occur between bleached BSA on the surface and FITC-BSA in the bulk in contrast to when FRAP is performed with labeled BSA still in the solution. Furthermore, differences in surface density could arise even though the adsorption conditions were similar. Such differences could lead to different diffusion rates, since BSA diffusion on poly(methyl methacrylate) (PMMA) surfaces has a documented dependence on surface density.4 In conclusion, the data support that the CLSM method for bleaching and data analysis presented could estimate D in the same range as the FRAPP technique. After the previously described FRAP method was established, the diffusion of TLL on solid surfaces with different hydrophobicity was analyzed. The adsorption of TLL to OTS surfaces is a very fast process, and also at low concentrations (0.1-0.3 µM) an adsorbed amount corresponding to the plateau of the adsorption isotherm was reached within 300 s (Figure 4). Thus, to be able to study the diffusion of TLL at various surface densities (θ), a shorter adsorption time was needed. However, on the steep increasing part of the adsorption curve, the resulting adsorbed amount will be very sensitive to small variations in concentration and adsorption times. Therefore, the fluorescence intensity of the surface was converted to surface density of TLL by means of a calibration graph (Figure 5). Ellipsometry data ensured that no desorption occurred during rinsing and that the adsorption of 0.3 µM lipase for 300 s used for this graph led well into the plateau value of the adsorbed amount (Figure 4). Since the calibration graph decreased linearly down to the value found with only the unlabeled population, both populations should have the same adsorption behavior when the bulk concentration of lipase was 0.3 µM. Hence, a direct way to translate the fluorescence intensity in the CLSM to a relative surface density, θ/θmax, could be established, where θmax corresponds to the maximum adsorbed amount, 2.2 mg/m2, found with the ellipsometer. It is well-known that the catalytic activity of TLL is strongly modulated by adsorption at interfaces and

Lipase Surface Diffusion Studied by FRAP

subsequent conformational changes.20,27,31 To what extent the activity of TLL relates to the mobility of the enzyme is not yet well understood. The surfaces used in this study were model surfaces and not substrates for TLL activity. The TLL diffusion on those surfaces was estimated to be a very slow process (Figure 6 and Table 2). Compared to earlier reports of diffusion on solid surfaces, the TLL diffusion coefficient were in the same range, 10-11 cm2/s, as found for cellulases on cellulose29 but more than 1 magnitude of order lower compared to collagenase on the substrate FALGPA14 or BSA on a variety of surfaces3 (Table 1). The diffusion of a globular protein of similar size (radius 23 Å) in free solution has, according to the Stokes-Einstein relation, a diffusion coefficient of about 10-6 cm2/s while the diffusion rate in biological membranes has been reported in the range between 10-7 and 10-8 cm2/s.32 However, the diffusion process on a solid surface is quite different from diffusion in a membrane, since the latter is more comparable to diffusion in a two-dimensional liquid.33 There was an approximately 10-fold decrease in diffusion rate when θ/θmax increased from 0.1 to 0.7 on the hydrophilic silica and the DDS surface (Figure 6). A similar, but less pronounced, trend was also seen on the OTS surface. The strong dependence on surface density is in good correspondence to theoretical predictions of diffusion of a single species on a homogeneous surface with one type of adsorption site.26,34 Moreover, Brownian dynamics studies on adsorbed proteins support those predictions,26,35 i.e., that diffusion is a decreasing function of surface density due to intermolecular repulsion and sterical hindrance. This behavior has also been experimentally observed by Tilton et al.,4 where the diffusion coefficient of irreversibly adsorbed BSA on poly(methyl methacrylate) (PMMA) decreased by a factor of 10 when the area fraction covered by protein increased from 0.10 to 0.69. Aside from the dependence of surface density, the hydrophobicity of the surface had a noticeable effect on surface diffusion of TLL (Figure 6 and Table 2). On the hydrophilic silica and the DDS surface, D0,t)0-90 and mobile fractions were similar. However, approaching infinite surface dilution, the diffusion rate diminished with time on the DDS surface, i.e., D0,t)90-180 was significantly lower. The same trend was seen on the OTS surface and both D0,t)0-90 and D0,t)90-180 as well as mobile fraction were significantly lower compared to the other surfaces. This means that, apart from the surface density, TLL mobility was affected by both the wettability of the surface and the time after adsorption. The difference in diffusion behavior on the different model surfaces could be explained by the surface dependence on the distribution of orientations and conformations of adsorbed TLL. On the very hydrophobic OTS surface, TLL is likely to expose the hydrophobic active site region; this altered conformation may in turn be responsible for the low mobility.17,18 Moreover, a more narrow distribution function of orientations of the adsorbed state of TLL is likely to emerge (i.e., a larger fraction of TLL is adsorbed in a low-mobility orientation with the open active site region toward the surface). On the more hydrophilic surfaces, the orientation distribution function could be broader with a diverse number of possible orientations, some being more mobile than the orientation (31) Martinelle, M.; Holmquist, M.; Hult, K. Biochim. Biophys. Acta 1995, 1258, 272-276. (32) Vaz, W. L. C.; Goodsaid-Zalduondo, F.; Jacobson, K. FEBS Lett. 1984, 174, 199-207. (33) Singer, S. J.; Nicolson, G. L. Science 1972, 175, 720-731. (34) Scalettar, B. A.; Abney, J. R.; Owicki, J. C. Proc. Natl. Acad. Sci. U.S.A. 1988, 85, 6726-6730. (35) Ravichandran, S.; Talbot, J. Biophys. J. 2000, 78, 110-120.

Langmuir, Vol. 21, No. 25, 2005 11955

with the lid region to the surface. This explanation is supported by previous activity experiments22 of adsorbed TLL toward p-nitro phenyl acetate in solution. The TLL activity was found to increase with increased surface wettability, suggesting that a larger fraction of TLL was adsorbed with the active site oriented outward on more hydrophilic surfaces. The decrease in diffusion rate with time on the DDS and OTS surfaces is probably due to a decrease in mobile fraction with time, which would mean that the distribution function of orientations became narrower and was displaced toward lower mobility with time. Another possibility would be that the threedimensional structure is altered with time, since surfaceinduced conformational changes have been recorded for several globular proteins.18 Especially on surfaces with low protein density, this might lead to partial denaturation of the protein structure with more points of contact to the surface, increasing the adsorption affinity and decreasing the mobility.36 The degree of labeling of TLL (0.6) should ensure that the distance between fluorophores in the adsorbed protein layer was enough to prevent any concentration quenching, i.e., that the interflourophore distance exceeded the Fo¨rster energy transfer radius, which is in the nanometer range.3 This could otherwise be a problem with overlabeled protein populations, disrupting the proportionality between fluorescence intensity to protein concentration. In this study, if the unlabeled protein had a faster diffusion rate than the labeled, the mobile population would have been underestimated and vice versa if the labeling procedure led to a more mobile species of the protein. However, irrespective of the degree of labeling, there will be a distribution of fluorophore/protein ratios resulting in a heterogeneous mixture of species on the surface. Hence, it is unlikely that the degree of labeling has any importance for the results presented in this work. Recovery profile graphs for TLL reach a plateau value of 0.45 for the OTS surface and about 0.6 for the DDS and hydrophilic silica surface (Table 2). Incomplete recovery profiles in FRAP experiments are usually explained by the presence of two adsorption states, meaning that the immobile population of the enzyme adsorbs directly with an orientation of low mobility.3 However, the decrease in mobility with time discovered on the OTS and DDS surface gives rise to another explanation. The incomplete recovery found in the recovery profiles could simply be due to a redistribution of TLL orientations and conformations on the surface, shifting toward lower surface mobility with time. Recovery profiles in this work were only sampled on the surfaces with the lowest TLL densities, and hence the highest diffusion rate, but nothing indicates that it should be different on surfaces with a higher surface density. It is difficult to conclude what mechanism is behind the TLL diffusion. Reported factors that influence the protein diffusion process on solid surfaces are mainly the affinity of the protein for the surface, the flexibility of the protein, and the organization of the protein layer formed.3 Another possibility is a desorption/readsorption migration. The proteins would then alternately fully desorb and readsorb, thereby facilitating a motion across the surface. However, no desorption of TLL could be seen upon rinsing the surface with buffer, and thus, this mechanism is unlikely to occur in this case. It is often thought that the actual migration is due to segmental movement. A more flexible protein structure would then be more diffusive on hard surfaces compared to a rigid structure. Comparing the structure (36) Norde, W. Proteins at solid surfaces. In Physical chemistry of biological interfaces; Marcel Dekker: New York, 2000; pp 115-135.

11956

Langmuir, Vol. 21, No. 25, 2005

of BSA and TLL, BSA has a low native state stability and is known to undergo secondary and tertiary conformation changes upon adsorption.37 This could facilitate a crawling motion on solid surfaces. TLL, on the other hand, is a quite rigid molecule except for the flexibility of the lid domain. As an example, Noinville et al.,18 concluded with FTIR that BSA, when adsorbed onto OTS surfaces, loses 13% of its helical domains while the helix content of the TLL lipase is unaffected. Hence, the lower mobility of TLL compared to BSA recorded in this work is not surprising, taking in account both the structural stability of TLL and the strong interaction between the open conformation and the hydrophobic surface. A rolling migration is a tentative diffusion mechanism of TLL. 5. Conclusions The diffusion of a variant of TLL on surfaces with varied degree of hydrophobicity could be quantified using FRAP. A mathematically straightforward estimation of the diffusion coefficient D of protein diffusion on solid surfaces was obtained using FRAP with a CLSM and analyzing the process as a one-dimensional relaxation of a step function. Due to the geometry of the system, even a rudimentary CLSM, with limited possibilities in defining the photobleached area of the sample, could be used for (37) Haynes, C. A.; Norde, W. Colloids Surf., B 1994, 2, 517-566.

Sonesson et al.

quantitative FRAP studies. Moreover, the CLSM gave the possibility to image the adsorbed proteins prior to bleaching, so that a representative area of the surface could be selected. Comparing the diffusion of BSA on coverslip glass with earlier work revealed that our method could estimate D in the same range as using the FRAPP technique. A calibration graph was used to convert the detected fluorescence intensity to surface density of TLL. Data on TLL surface diffusion revealed that the diffusion was dependent on both the surface density and the surface wettability. Low density and a hydrophilic surface gave rise to the fastest diffusion rate whereas more densely packed enzyme layers in combination with a hydrophobic surface rendered TLL immobile. This could be explained by increased intermolecular repulsion and sterical hindrance at increased surface density of protein in combination with a surface dependence on the distribution of orientations and conformations of adsorbed TLL. The diffusion of TLL decreased with time on the hydrophobic surfaces, probably due to a decrease in the mobile fraction of TLL with time originating from a redistribution of adsorbed states of TLL, with a growing population of TLL adsorbed with the active site region facing the surface. Acknowledgment. Novozymes A/S is gratefully acknowledged for financial support of the project. LA051773+