- Email: [email protected]
Viscoelastic properties of adsorbed and cross-linked polypeptide and protein layers at a solid–liquid interface
Journal of Colloid and Interface Science 324 (2008) 55–60
Contents lists available at ScienceDirect
Journal of Colloid and Interface Science www.elsevier.com/locate/jcis
Viscoelastic properties of adsorbed and cross-linked polypeptide and protein layers at a solid–liquid interface Amit K. Dutta, Arpan Nayak, Georges Belfort ∗ Howard P. Isermann Department of Chemical and Biological Engineering, Rensselaer Polytechnic Institute, Troy, NY, USA
a r t i c l e
i n f o
a b s t r a c t
Article history: Received 3 January 2008 Accepted 29 April 2008 Available online 7 May 2008
The real-time changes in viscoelasticity of adsorbed poly(L-lysine) (PLL) and adsorbed histone (lysine rich fraction) due to cross-linking by glutaraldehyde and corresponding release of associated water were investigated using a quartz crystal microbalance with dissipation monitoring (QCM-D) and attenuated total reflection Fourier transform infrared spectroscopy (ATR/FTIR). The kinetics of PLL and histone adsorption were measured through changes in mass adsorbed onto a gold-coated quartz surface from changes in frequency and dissipation and using the Voigt viscoelastic model. Prior to cross-linking, the shear viscosity and shear modulus of the adsorbed PLL layer were ∼3.0 × 10−3 Pa s and ∼2.5 × 105 Pa, respectively, while after cross-linking, they increased to ∼17.5 × 10−3 Pa s and ∼2.5 × 106 Pa, respectively. For the adsorbed histone layer, shear viscosity and shear modulus increased modestly from ∼1.3 × 10−3 to ∼2.0 × 10−3 Pa s and from ∼1.2 × 104 to ∼1.6 × 104 Pa, respectively. The adsorbed mass estimated from the Sauerbrey equation (perfectly elastic) and the Voigt viscoelastic model differ appreciably prior to cross-linking whereas after cross-linking they converged. This is because trapped water molecules were released during cross-linking. This was confirmed experimentally via ATR/FTIR measurements. The variation in viscoelastic properties increased substantially after cross-linking presumably due to fluctuation of the randomly cross-linked network structure. An increase in fluctuation of the viscoelastic properties and the loss of imbibed water could be used as a signature of the formation of a cross-linked network and the amount of cross-linking, respectively. © 2008 Elsevier Inc. All rights reserved.
Keywords: QCM-D Viscoelasticity Cross-link Peptides Proteins
1. Introduction There is increasing interest in cross-linking polypeptides and proteins to increase protein stability [1,2]. Recent evidence has shown that proteins, either from cross-linking [3] or from poly(ethylene glycol) [4] addition, have longer in vivo half-lives and hence are likely be more efficacious provided they retain their activity. Similarly, stabilizing proteins at interfaces are of relevance to bio-fouling, [5] nano-fluidics and protein-based sensor technology [6]. In all these cases, the rheological properties of aggregates or films are important [7]. Protein cross-linking also occurs in biological systems during post-translational modification [8]. Transglutaminases (Tgases) cross-link protein via ε -(γ glutamyl)lysine bonds or through incorporation of primary amines at selected peptide-bound glutamine residues. These cross-linked proteins have high mechanical strength, are resistant to proteolytic degradation, and they accumulate in a number of tissues and processes where such properties are important such as skin,
*
Corresponding author. E-mail address: [email protected] (G. Belfort).
0021-9797/$ – see front matter doi:10.1016/j.jcis.2008.04.065
©
2008 Elsevier Inc. All rights reserved.
hair, and blood clotting [8]. Many researchers have investigated the rheological properties of adsorbed protein layers [9–11]. Though cross-linking proteins in bulk solution to increase their stability has been widely studied, similar studies at solid–liquid interfaces are limited. To monitor the changes in rheological properties during cross-linking in real time at an interface, we have chosen to study a model polypeptide, poly(L-lysine), and a model protein, histone (lysine rich fraction) adsorbed onto a flat gold surface using a quartz crystal microbalance with dissipation monitoring (QCM-D). QCM-D is a simple and highly sensitive mass sensor by which a wide range of interfacial adsorption reactions can be monitored, on a variety of supports, in real time. In a QCM-D, a gold-plated AT-cut piezoelectric quartz crystal is oscillated in shear mode at its fundamental resonant frequency by passing an alternating current through the crystal. The dissipation and resonant frequency of the crystal can be recorded in real time. The principles behind this technique have been reported elsewhere [12]. The measurement of changes in frequency and dissipation using QCM-D allows one to estimate the viscoelasticity of the non-elastic adsorbed layer by using Voigt viscoelastic model [13–15], which has been used to probe the rheological properties of proteins [16,17] and synthetic polymers [18–20] at the solid–liquid interface. Höök et al. [16]
56
A.K. Dutta et al. / Journal of Colloid and Interface Science 324 (2008) 55–60
used this model to study the viscoelasticity of an adsorbed Mytilus edulis foot protein (Mefp-1) film both before and after cross-linking of the adsorbed layer by NaIO4 . Adsorption and viscoelastic properties of type I collagen, in native and heat-denatured forms, on polystyrene was studied by Gurdak et al. [17]. Adsorption kinetics and cross-linking of ethyl(hydroxyethyl) cellulose (EHEC) and hydrophobically modified EHEC at hydrophilic and hydrophobic modified surfaces have been investigated in detail by Hedin et al. [18]. Here, we have used the same technique to study the changes in viscoelasticity of an adsorbed PLL and a histone layer due to cross-linking by glutaraldehyde (GTA). We choose histone as it is a well-studied protein rich in lysine residues for effective crosslinking (MW 21.5 kDa, 24–32% lysine). Core histones of which the histone chosen for this study is one component, assemble to form one octameric nucleosome core particle by wrapping base-pairs of DNA around a protein spool [21]. They also play a role in gene regulation. Our work shows that structural differences between PLL and histone result in differences in viscoelastic properties of the adsorbed layer. Cross-linking of PLL and histone with GTA resulted in an increase in rigidity of the adsorbed polypeptide and protein layers at the solid–liquid interface and an increase in the fluctuation of their constitutive viscoelastic properties.
2.3. Viscoelastic modeling The Voigt element is a parallel combination of a spring and a dashpot to represent the elastic (storage) and inelastic (damping) behavior of a material, respectively. Using the Voigt model, Voinova et al. [13] developed the following equations to describe the viscoelasticity of an adsorbed layer:
f =
Im(β) 2π dq ρq
D = −
(1)
,
Re(β)
π f d q ρq
,
(2)
where
β=
(2π f ηξ1 − i μξ1 )(1 − α exp(2ξ1 d)) , 2π f (1 + α exp(2ξ1 d))
(2π f ηξ1 − i μξ1 + 2π f ηl ξ2 ) , (2π f ηξ1 − i μξ1 − 2π f ηl ξ2 ) (2π f )2 ρ ξ1 = − , μ + i2π f η
α=
i2π f ρl
(3) (4)
(5)
2. Materials and methods
ξ2 =
2.1. QCM-D
d q and ρ q are the quartz thickness and density, respectively. f 0 is the fundamental resonant frequency, and f = nf 0 (with n = 3, 5, and 7 are the overtone numbers). ρ l and η l are the bulk liquid density and viscosity, respectively. Changes in frequency ( f ) and dissipation ( D) for three overtones (3, 5, 7) were used to estimate the thickness (d) (or mass adsorbed) and two viscoelastic parameters, shear modulus (μ) and shear viscosity (η ) of the adsorbed layer using QTools software (Q-Sense). More details on the model have been reported elsewhere [13,14]. This model does not take into account the frequency dependence of viscoelasticity. However, Höök et al. [16] and Larsson et al. [24] have shown that effect of frequency on viscoelasticity is very small.
Aqueous solutions at 300 μg/ml of poly(L-lysine) hydro bromide (molecular weight > 300 kDa, 81356, Sigma, Saint Louis, MO) and 21.5 μg/ml of histone (21.5 kDa, H5505, Sigma) and 100 μg/ml glutaraldehyde (GTA, 340855, Sigma) were prepared in phosphate buffered saline (PBS) at pH 7.4. Concentration of PLL, histone and glutaraldehyde was kept very low so that they do not affect the viscosity of the solution. Polished gold coated AT-cut quartz crystals with fundamental frequencies of ∼5 MHz (QSX301, Q-Sense AB, Göteborg, Sweden) were cleaned by immersion in a 1:1:5 mixture of H2 O2 (30%), NH3 (25%), and distilled water at 60 ◦ C for 20 min followed by exposure to UV/O3 for 10 min. 250 μl of the polypeptide/protein solution were injected through the sensor loop and allowed to adsorb for 40 min. The adsorbed protein on the crystal was then washed with buffer for 20 min and then cross-linked by injecting 250 μl of glutaraldehyde solution for ∼20 min. Approximately 100% of the reactive lysine present in the protein/polypeptide layer was cross-linked due to addition of GTA [22]. Mass and dissipation measurements were performed using a QCM-D (D300, Q-Sense AB, Göteborg, Sweden) in batch mode at 24 ± 0.1 ◦ C. More details about the experimental procedure have been reported elsewhere [23]. 2.2. ATR/FTIR All polypeptide/protein solution spectra were recorded in their respective aqueous buffer solution. A horizontal ATR (Magna-IR 550 Series II, Thermo Nicolet Instruments Corp., Madison, WI) accessory with a trapezoidal germanium crystal having ends cut to 45◦ generated 12 internal reflections. The spectrometer was equipped with a liquid nitrogen-cooled mercury cadmium telluride detector. To reduce the contributions of water vapor and carbon dioxide, the IR system was continuously purged with air from an FTIR Purge gas generator (Model 74-45, Balston, Haverhill, MA). Spectra were collected at a gain of 8 and an aperture of 40 with spectral resolution of 2 cm−1 . All spectra were collected in the 4000–1000 cm−1 range as sets of 256 time-averaged interferograms. Omnic (v 7.0) software (Thermo Nicolet Instruments Corp., Madison, WI) was used to analyze and measure peak heights for different sample spectra.
ηl
.
(6)
3. Results and discussion The adsorption of PLL and histone from PBS at pH 7.4 and 24 ± 0.1 ◦ C onto a gold coated quartz crystal was measured with QCM-D by recording changes in frequency and dissipation with time. In order to compare the data for the three overtones, the decrease in frequency was normalized by their overtone number. The adsorption of PLL and histone resulted in a decrease in the normalized frequency and an increase in the dissipation for the three overtones as shown in Figs. 1a and 1b, respectively. The normalized change in frequency ( f /n) and dissipation ( D) for the three overtones did not overlap. This indicates that the adsorbed layer is not perfectly elastic [16]. If a layer is not perfectly elastic, it dissipates energy and the adsorbed mass is not proportional to change in normalized frequency. Hence, f /n for three overtones does not overlap for a viscoelastic adsorbed layer such as ours. For a perfectly elastic and thin adsorbed layer, researchers have successfully used the Sauerbrey equation [25]. According to this equation, the adsorbed mass, m (ng/cm2 ) is proportional to a normalized decrease in the frequency, f /n (Hz). If the Sauerbrey equation holds, then,
m = C
f n
,
(7)
where C is the mass sensitivity constant (C = 17.7 ng cm−2 Hz−1 ). Due to the viscoelastic nature of the adsorbed layer, we then fit the Voigt viscoelastic model to the data using QTools (Q-Sense)
A.K. Dutta et al. / Journal of Colloid and Interface Science 324 (2008) 55–60
57
(a) (a)
(b)
(b) Fig. 1. Adsorption of (a) poly(L-lysine) and (b) histone onto gold. The normalized change in frequency ( f /n) decreased and dissipation ( D) increased as a function of time for the three overtones (red: n = 3, blue: n = 5, and green: n = 7; f 0 = 5 MHz). The black points describe the Voigt model for the three overtones. dq = 3.34 × 10−4 m; ρq = 2650 kg/m3 ; ρl = 1000 kg/m3 and ηl = 1.0 × 10−3 Pa s [34]. T = 24.0 ± 0.1 ◦ C.
software to obtain estimates of the mass adsorbed and the two viscoelastic fitting parameters, shear viscosity and shear modulus. In order to fit the data, we assumed that the density of the adsorbed layer was 1100 kg/m3 . Note that the density of the adsorbed layer was arbitrarily chosen and was assumed constant. The density of the adsorbed layer was likely to increase with increasing coverage. However, the adsorbed layer density primarily affects the thickness of the adsorbed layer and not the mass adsorbed, or the shear modulus and shear viscosity as demonstrated by Larsson et al. [24]. Varying the density of the adsorbed layer up to 1400 kg/m3 (density of protein [16]) did not significantly change the mass, shear viscosity or shear modulus of the adsorbed layer and thus did not affect the overall conclusions of this work. This approach has been widely used to analyze QCM-D data as reported for the adsorption of proteins [16], DNA [14,26], and polymers [18,19]. Relatively good model fits (black points) for the adsorption of PLL and histone are shown in Figs. 1a and 1b, respectively. The adsorption of PLL and histone onto the gold surface was fast and reached saturation within ∼30–40 min and the values of m obtained from Voigt model were 390–420 and 2510–2580 ng/cm2 , respectively, as shown in Fig. 2. Washing with PBS buffer resulted in desorption of loosely adsorbed PLL from the adsorbed layer and resulted in a positive increase in frequency and a concomitant drop in the adsorbed mass. Adsorption of histone onto the gold surface followed by washing with the PBS buffer did not result in significant desorption of the adsorbed protein indicating that histone adsorbed much more strongly than PLL. Also, the histone layer exhibited 3–4 times the dissipation, D, of the PLL layer (Fig. 1). Adsorbed mass
Fig. 2. Estimation of the mass adsorbed for (a) the PLL and (b) the histone layers. The mass adsorbed (m) calculated from Sauerbrey equation as a function of time for the three overtones (red: n = 3, blue: n = 5, and green: n = 7) is compared with that calculated from Voigt vicoelastic model (black points).
calculated from the Sauerbrey equation (mSauerbrey ) is lower than that estimated from Voigt model (mVoigt ) for both PLL and histone as shown in Fig. 2. However, the deviation of mVoigt from mSauerbrey for histone is much larger than that for PLL. This indicates that the adsorbed PLL layer is more rigid (or less viscoelastic) than adsorbed histone layer which is also reflected in their shear viscosity and shear modulus values (below). PLL exhibited a small drop in mass during PBS washing and GTA cross-linking (for both models). For histone, no change was observed during PBS washing while cross-linking induced a substantial change for the Voigt fit and a moderate change for the Sauerbrey prediction. The shear modulus and shear viscosity of the adsorbed PLL and histone layers were obtained from fitting the Voigt model to the experimental data and are shown in Figs. 3 and 4, respectively. Even though washing the adsorbed PLL layer with the PBS buffer released ∼80 ng/cm2 of the mass adsorbed (Fig. 2a), the change in viscoelasity of the layer was negligible. Addition of the cross-linker GTA, resulted in cross-linking of the adsorbed layer accompanied by an increase in change in normalized frequency and a drop in dissipation for both PLL and histone (Fig. 1). This resulted in a decrease in mass of the adsorbed layer for both PLL and histone (Fig. 2). This also resulted in a substantial increase in shear viscosity and shear modulus for the adsorbed PLL layer from ∼3 × 10−3 to ∼17.5 × 10−3 Pa s and from ∼2.5 × 105 to ∼2.5 × 106 Pa, respectively. For the histone layer, shear viscosity increased from ∼1.3 × 10−3 to ∼2.0 × 10−3 Pa s and shear modulus changed from ∼1.2 × 104 to ∼1.6 × 104 Pa. The higher shear viscosity and shear modulus of the cross-linked PLL layer as compared with the cross-linked histone layer was likely due to the far greater number (∼1500 lysine/molecule of PLL and ∼40 lysine/molecule of histone) of cross-linkable lysines present in the PLL layer. Since,
58
A.K. Dutta et al. / Journal of Colloid and Interface Science 324 (2008) 55–60
(a)
(b)
Fig. 3. Variation of (a) shear viscosity, η , and (b) shear modulus, μ, as a function of time for adsorption of PLL onto gold during adsorption, washing with PBS and cross-linking with glutaraldehyde (GTA) at 24.0 ± 0.1 ◦ C and pH 7.4. Cross-linking of the adsorbed polypeptide layer with GTA increased its shear viscosity and shear modulus.
(a)
(b)
Fig. 4. Variation of (a) shear viscosity, η , and (b) shear modulus, μ, as a function of time for adsorption of histone onto gold during adsorption, washing with PBS and cross-linking with glutaraldehyde (GTA) at 24.0 ± 0.1 ◦ C and pH 7.4. Cross-linking of the adsorbed protein layer with GTA increased its shear viscosity and shear modulus. Table 1 Effect of assumed density on the mass and viscoelastic properties of the adsorbed layer Assumed density (kg/m3 )
1100 1400
Mass adsorbed after GTA cross-linking (ng/cm2 )
Shear modulus after GTA cross-linking (Pa)
PLL
Histone
PLL
720–760 725–760
∼2.5 × 10 ∼2.2 × 106
230–240 225–235
after cross-linking, the viscoelastic properties for both the layers fluctuated enormously, we attribute this to to concentration variation of the randomly cross-linked network structure [27]. This increase in fluctuation was also observed due to cross-linking of the Mytilus edulis foot protein (Mefp-1) [16] and for polymerization of DNA [14]. This increase in fluctuation of viscoelastic properties could be used as a signature for the formation of a cross-linked network. The decrease in mass after GTA addition is most likely due to the release of water molecules from the compact crosslinked polypeptide/protein layer. This was also reported by Höök et al. [16] for cross-linking of an adsorbed Mefp-1 film by addition of NaIO4 . Reported values of shear viscosity and shear modulus before cross-linking were ∼2 × 10−3 Pa s and ∼1 × 105 Pa, respectively. After cross-linking with NaIO4 , Höök et al. reported that the shear viscosity and shear modulus increased to ∼5.5 × 10−3 Pa s and to ∼4.5 × 105 Pa, respectively. The adsorption of PLL and histone followed by cross-linking with the GTA, demonstrates that QCM-D can be used to study the cross-linking of proteins and other macromolecules with high precision and sensitivity. The normalized change in frequencies for the three overtones was different before cross-linking. However, after cross-linking, they were almost similar for all the three overtones indicating that they followed the Sauerbrey equation better after than before cross-linking because the adsorbed layer became more rigid. Similarly, the dissipation values after cross-linking were very small (<1 × 10−6 ).
Shear viscosity after GTA cross-linking (Pa s) Histone
6
∼1.6 × 10 ∼1.4 × 104 4
PLL
Histone
∼17.5 × 10−3
∼2.00 × 10−3 ∼1.75 × 10−3
∼13.5 × 10−3
The calculated mass adsorbed for the PLL layer from the Sauerbrey equation before cross-linking and buffer wash for the three overtones varied from 240 to 280 ng/cm2 , whereas the mass adsorbed from the Voigt viscoelastic model was estimated to be greater than 390 ng/cm2 . This large deviation indicated that that adsorbed layer was not rigid prior to cross-linking. However, after crosslinking with GTA, the mass adsorbed calculated from Sauerbrey equation varied from 220 to 235 ng/cm2 , whereas that from the Voigt model was 235–245 ng/cm2 . This close agreement between the mass prediction by the Voigt viscoelastic model and the Sauerbrey equation after cross-linking indicates that the layer became more rigid due GTA addition. Similar results were also observed for the adsorbed histone layer. To study the effect of density on the mass and viscoelastic properties of the adsorbed layer, we fitted the experimental data assuming an adsorbed layer density of 1400 kg/m3 . Comparative results are shown in Table 1. 1400 kg/m3 is the density of usual proteins and hence represents an upper bound. The effect of density on mass adsorbed varied from ∼1 to ∼4% whereas, the effect of density on shears modulus and shear viscosity varied from ∼12 to ∼25%. The mass changes due to a density increase from 1100 to 1400 kg/m3 were negligibly small. These small changes in viscoelastic parameters due to change in assumed density does not significantly affect the overall conclusion about the rheological characteristics of the adsorbed layers.
A.K. Dutta et al. / Journal of Colloid and Interface Science 324 (2008) 55–60
59
The ratio | D /( f /n)| has also been used to represent the aggregate viscoelasticity of an adsorbed layer [19,26,28]. The average | D /( f /n)| ratio for the three overtones for the three transition periods, before buffer wash, after buffer wash and after cross-linking, are shown in Fig. 5. The | D /( f /n)| ratio for the PLL layer before buffer wash was 0.087 ± 0.012 × 10−6 Hz−1 . Due to the buffer wash, some loosely adsorbed PLL molecules desorbed from the surface resulting in a decrease of the | D /( f /n)| ratio to 0.062 ± 0.008 × 10−6 Hz−1 . However, cross-linking of the adsorbed PLL layer with GTA lowered the ratio to 0.011 ± 0.001 × 10−6 Hz−1 indicating increased rigidity. In comparison, | D /( f /n)| ratio for the histone layer before buffer wash was 0.136 ± 0.009 × 10−6 Hz−1 which was much higher than that of the adsorbed PLL layer. The lower | D /( f /n)| ratio of PLL suggests that the random coil of PLL lies down flat-on on the gold surface and has less trapped water than the adsorbed histone layer. The tertiary structure of the histone and its resultant layer on the surface likely imbibe a larger amount of trapped water than the PLL layer. The | D /( f /n)| ratio for the histone layer after buffer wash did not change because, there were almost no desorption of histone from gold surface. However, after cross-linking
Fig. 5. Average | D /( f /n)| ratio for the three overtones and for the three transition periods: before buffer wash, after buffer wash and after cross-linking. PLL ((") and solid line) and histone ((2) and dashed lines). The lines are used to indicate the transition from one point to another.
(a)
(b) Fig. 6. ATR/FTIR spectra of (a) PLL and (b) histone prior to cross-linking (without GTA, solid line) and after cross-linking (with GTA, dotted line). Spectra were collected with a gain of 8 and resolution of 2 cm−1 with 256 scans for each sample. T = 24 ◦ C and pH 7.4.
60
A.K. Dutta et al. / Journal of Colloid and Interface Science 324 (2008) 55–60
of the adsorbed histone layer, the | D /( f /n)| ratio decreased to 0.028 ± 0.005 × 10−6 Hz−1 indicating increased rigidity and some loss of water. The cross-linking behavior of PLL and histone and corresponding changes in their associated water content were also investigated using ATR/FTIR. To analyze the pure protein spectra, corresponding buffer (i.e., GTA in buffer spectra for GTA addition) and water vapor spectra were subtracted from the protein spectra in aqueous solutions [29]. Water molecules exhibit two strong vibrational modes—a symmetric OH stretching and an H–O–H bending [30]. The peaks at ∼3375 and ∼1639 cm−1 are assigned to OH stretching and H–O–H bending vibrations [31,32], respectively, and are attributed to associated water molecules [33]. The calculated heights of the peaks at ∼3375 and ∼1639 cm−1 for PLL without GTA were 0.492 and 0.200, respectively (Fig. 6a). After addition of GTA, these values decreased to 0.188 and 0.081, respectively. For histone, these heights before addition of GTA were 0.384 and 0.157, respectively, and decreased only slightly to 0.374 and 0.156, respectively, after addition of GTA (Fig. 6b). The decrease in absorbance of water bands indicates release of associated water molecules from the protein/polypeptide. The larger drop in intensities in PLL as compared with histone suggests that more cross-linking was possible in PLL than in histone as also demonstrated by their changes in shear modulus and shear viscosities (Figs. 3–5). 4. Conclusions We probed the cross-linking behavior of a protein and a polypeptide at a solid–liquid interface in the presence of glutaraldehyde using QCM-D and ATR/FTIR. While the frequency and dissipation data obtained from QCM-D provided a measure of the changes in viscoelasticity of the two layers, the FTIR data gave us a qualitative measure of the changes in associated water content. The data obtained by these two independent analytical methods correlate well with each other. The frequency and dissipation measurements allowed us to probe the structural differences between the PLL peptide layer with the histone protein layer prior to and after cross-linking. Using a combined average parameter, | D /( f /n)| ratio, the value for PLL layer was lower, and hence more rigid, than the histone layer due to greater cross-linking. After cross-linking, the rigidity increased further whereby the PLL adsorbed layer was higher than that of the histone adsorbed layer due to more lysine crosslinking groups in PLL. The higher viscoelasticity and large loss of mass resulting from glutaraldehyde cross-linking for the adsorbed histone layer indicated release of large amounts of trapped water from the histone layer at the solid–liquid interface. These results also showed that the Sauerbrey equation could be used to convert the change in frequency to adsorbed mass after cross-linking where the adsorbed layer was rigid but not prior to cross-linking where it is more viscoelastic. Also after cross-linking, the fluctuation in viscoelastic properties of the adsorbed layer increased. This
increase in fluctuation was also reported for cross-linking of Mefp1 [16] and the polymerization of DNA [14]. We propose that an increase in fluctuation of the viscoelastic properties and a loss of imbibed water could be used as a signature of the formation of a cross-linked network and the amount of cross-linking, respectively. Acknowledgments This work was supported by the National Institute of Health (R01GM071329) and US Department of Energy (DE-FG02-90ER14114 and DE-FG02-05ER46249). References [1] L. Cao, F. Rantwijk, R. Sheldon, Org. Lett. 2 (2000) 1361. [2] F. López-Gallego, L. Betancor, C. Mateo, A. Hidalgo, N. Alonso-Morales, G. Dellamora-Ortiz, J.M. Guisán, R. Fernández-Lafuente, J. Biotechnol. 119 (2005) 70. [3] J. Kupec, K. Charvátová, M. Navrátil, V. Kˇresálek, M. Kˇresálková, J. Polym. Environ. 11 (2003) 93. [4] M. Diwan, T.G. Park, J. Control. Release 73 (2001) 233. [5] M. Karlsson, J. Ekeroth, H. Elwing, U. Carlsson, J. Biol. Chem. 280 (2005) 25558. [6] N.C. Vercillo, K.J. Herald, J.M. Fox, B.S. Der, J.D. Dattelbaum, Protein Sci. 16 (2007) 362. [7] A.I. Romoscanu, R. Mezzenga, Langmuir 21 (2005) 9689. [8] M. Griffin, R. Casadio, C.M. Bergamini, Biochem. J. 368 (2002) 377. [9] J.M.R. Patino, S.E.M. Ortiz, C.C. Sánchez, M.R.R. Niño, M.C. Añón, J. Colloid Interface Sci. 268 (2003) 50. [10] 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. [11] E. Dickinson, Colloids Surf. B 15 (1999) 161. [12] M. Rodahl, F. Höök, A. Krozer, P. Brzezinski, B. Kasemo, Rev. Sci. Instrum. 66 (1995) 3924. [13] M. Voinova, M. Rohdal, M. Jonson, B. Kasemo, Phys. Scri. 59 (1999) 391. [14] G. Stengel, F. Höök, W. Knoll, Anal. Chem. 77 (2005) 3709. [15] A. Domack, O. Prucker, J. Rühe, D. Johannsmann, Phys. Rev. E 56 (1997) 680. [16] F. Höök, B. Kasemo, T. Nylander, C. Fant, K. Sott, H. Elwing, Anal. Chem. 73 (2001) 5796. [17] E. Gurdak, J. Booth, C.J. Roberts, P.G. Rouxhet, C.C. Dupont-Gillain, J. Colloid Interface Sci. 302 (2006) 475. [18] J. Hedin, J.-E. Löfroth, M. Nydén, Langmuir 23 (2007) 6148. [19] A.K. Dutta, G. Belfort, Langmuir 23 (2007) 3088. [20] E.F. Irwin, J.E. Ho, S.R. Kane, K.E. Healy, Langmuir 21 (2005) 5529. [21] K. Luger, A.W. Mäder, R.K. Richmond, D.F. Sargent, T.J. Richmond, Nature 389 (1997) 251. [22] C. Marquié, A.M. Tessier, C. Aymard, S. Guilbert, Nahrung 42 (1998) 264. [23] F. Höök, M. Rodahl, P. Brzezinski, B. Kasemo, Langmuir 14 (1998) 729. [24] C. Larsson, M. Rodahl, F. Höök, Anal. Chem. 75 (2003) 5080. [25] G. Sauerbrey, Z. Physica 155 (1959) 206. [26] T.H. Nguyen, M. Elimelech, Biomacromolecules 8 (2007) 24. [27] Y. Rabin, R. Bruinsma, Europhys. Lett. 20 (1992) 79. [28] A.K. Dutta, A. Nayak, G. Belfort, Macromolecules 41 (2008) 301. [29] A. Sethuraman, G. Belfort, Biophys. J. 88 (2005) 1322. [30] D. Eisenberg, W. Kauzmann, The Structure and Properties of Water, Oxford Univ. Press, New York, 1969. [31] N.I.E. Shewring, T.G.J. Jones, G. Maitland, J. Yarwood, J. Colloid Interface Sci. 176 (1995) 308. [32] F.O. Libnau, O.M. Kvalheim, A.A. Christy, J. Toft, Vibr. Spectrosc. 7 (1994) 243. [33] S. Ziming, K.G. Peter, N.H. David, Polyhedron 17 (1998) 1511. [34] M. Rodahl, B. Kasemo, Sens. Actuators A 54 (1996) 448.
Contents lists available at ScienceDirect
Journal of Colloid and Interface Science www.elsevier.com/locate/jcis
Viscoelastic properties of adsorbed and cross-linked polypeptide and protein layers at a solid–liquid interface Amit K. Dutta, Arpan Nayak, Georges Belfort ∗ Howard P. Isermann Department of Chemical and Biological Engineering, Rensselaer Polytechnic Institute, Troy, NY, USA
a r t i c l e
i n f o
a b s t r a c t
Article history: Received 3 January 2008 Accepted 29 April 2008 Available online 7 May 2008
The real-time changes in viscoelasticity of adsorbed poly(L-lysine) (PLL) and adsorbed histone (lysine rich fraction) due to cross-linking by glutaraldehyde and corresponding release of associated water were investigated using a quartz crystal microbalance with dissipation monitoring (QCM-D) and attenuated total reflection Fourier transform infrared spectroscopy (ATR/FTIR). The kinetics of PLL and histone adsorption were measured through changes in mass adsorbed onto a gold-coated quartz surface from changes in frequency and dissipation and using the Voigt viscoelastic model. Prior to cross-linking, the shear viscosity and shear modulus of the adsorbed PLL layer were ∼3.0 × 10−3 Pa s and ∼2.5 × 105 Pa, respectively, while after cross-linking, they increased to ∼17.5 × 10−3 Pa s and ∼2.5 × 106 Pa, respectively. For the adsorbed histone layer, shear viscosity and shear modulus increased modestly from ∼1.3 × 10−3 to ∼2.0 × 10−3 Pa s and from ∼1.2 × 104 to ∼1.6 × 104 Pa, respectively. The adsorbed mass estimated from the Sauerbrey equation (perfectly elastic) and the Voigt viscoelastic model differ appreciably prior to cross-linking whereas after cross-linking they converged. This is because trapped water molecules were released during cross-linking. This was confirmed experimentally via ATR/FTIR measurements. The variation in viscoelastic properties increased substantially after cross-linking presumably due to fluctuation of the randomly cross-linked network structure. An increase in fluctuation of the viscoelastic properties and the loss of imbibed water could be used as a signature of the formation of a cross-linked network and the amount of cross-linking, respectively. © 2008 Elsevier Inc. All rights reserved.
Keywords: QCM-D Viscoelasticity Cross-link Peptides Proteins
1. Introduction There is increasing interest in cross-linking polypeptides and proteins to increase protein stability [1,2]. Recent evidence has shown that proteins, either from cross-linking [3] or from poly(ethylene glycol) [4] addition, have longer in vivo half-lives and hence are likely be more efficacious provided they retain their activity. Similarly, stabilizing proteins at interfaces are of relevance to bio-fouling, [5] nano-fluidics and protein-based sensor technology [6]. In all these cases, the rheological properties of aggregates or films are important [7]. Protein cross-linking also occurs in biological systems during post-translational modification [8]. Transglutaminases (Tgases) cross-link protein via ε -(γ glutamyl)lysine bonds or through incorporation of primary amines at selected peptide-bound glutamine residues. These cross-linked proteins have high mechanical strength, are resistant to proteolytic degradation, and they accumulate in a number of tissues and processes where such properties are important such as skin,
*
Corresponding author. E-mail address: [email protected] (G. Belfort).
0021-9797/$ – see front matter doi:10.1016/j.jcis.2008.04.065
©
2008 Elsevier Inc. All rights reserved.
hair, and blood clotting [8]. Many researchers have investigated the rheological properties of adsorbed protein layers [9–11]. Though cross-linking proteins in bulk solution to increase their stability has been widely studied, similar studies at solid–liquid interfaces are limited. To monitor the changes in rheological properties during cross-linking in real time at an interface, we have chosen to study a model polypeptide, poly(L-lysine), and a model protein, histone (lysine rich fraction) adsorbed onto a flat gold surface using a quartz crystal microbalance with dissipation monitoring (QCM-D). QCM-D is a simple and highly sensitive mass sensor by which a wide range of interfacial adsorption reactions can be monitored, on a variety of supports, in real time. In a QCM-D, a gold-plated AT-cut piezoelectric quartz crystal is oscillated in shear mode at its fundamental resonant frequency by passing an alternating current through the crystal. The dissipation and resonant frequency of the crystal can be recorded in real time. The principles behind this technique have been reported elsewhere [12]. The measurement of changes in frequency and dissipation using QCM-D allows one to estimate the viscoelasticity of the non-elastic adsorbed layer by using Voigt viscoelastic model [13–15], which has been used to probe the rheological properties of proteins [16,17] and synthetic polymers [18–20] at the solid–liquid interface. Höök et al. [16]
56
A.K. Dutta et al. / Journal of Colloid and Interface Science 324 (2008) 55–60
used this model to study the viscoelasticity of an adsorbed Mytilus edulis foot protein (Mefp-1) film both before and after cross-linking of the adsorbed layer by NaIO4 . Adsorption and viscoelastic properties of type I collagen, in native and heat-denatured forms, on polystyrene was studied by Gurdak et al. [17]. Adsorption kinetics and cross-linking of ethyl(hydroxyethyl) cellulose (EHEC) and hydrophobically modified EHEC at hydrophilic and hydrophobic modified surfaces have been investigated in detail by Hedin et al. [18]. Here, we have used the same technique to study the changes in viscoelasticity of an adsorbed PLL and a histone layer due to cross-linking by glutaraldehyde (GTA). We choose histone as it is a well-studied protein rich in lysine residues for effective crosslinking (MW 21.5 kDa, 24–32% lysine). Core histones of which the histone chosen for this study is one component, assemble to form one octameric nucleosome core particle by wrapping base-pairs of DNA around a protein spool [21]. They also play a role in gene regulation. Our work shows that structural differences between PLL and histone result in differences in viscoelastic properties of the adsorbed layer. Cross-linking of PLL and histone with GTA resulted in an increase in rigidity of the adsorbed polypeptide and protein layers at the solid–liquid interface and an increase in the fluctuation of their constitutive viscoelastic properties.
2.3. Viscoelastic modeling The Voigt element is a parallel combination of a spring and a dashpot to represent the elastic (storage) and inelastic (damping) behavior of a material, respectively. Using the Voigt model, Voinova et al. [13] developed the following equations to describe the viscoelasticity of an adsorbed layer:
f =
Im(β) 2π dq ρq
D = −
(1)
,
Re(β)
π f d q ρq
,
(2)
where
β=
(2π f ηξ1 − i μξ1 )(1 − α exp(2ξ1 d)) , 2π f (1 + α exp(2ξ1 d))
(2π f ηξ1 − i μξ1 + 2π f ηl ξ2 ) , (2π f ηξ1 − i μξ1 − 2π f ηl ξ2 ) (2π f )2 ρ ξ1 = − , μ + i2π f η
α=
i2π f ρl
(3) (4)
(5)
2. Materials and methods
ξ2 =
2.1. QCM-D
d q and ρ q are the quartz thickness and density, respectively. f 0 is the fundamental resonant frequency, and f = nf 0 (with n = 3, 5, and 7 are the overtone numbers). ρ l and η l are the bulk liquid density and viscosity, respectively. Changes in frequency ( f ) and dissipation ( D) for three overtones (3, 5, 7) were used to estimate the thickness (d) (or mass adsorbed) and two viscoelastic parameters, shear modulus (μ) and shear viscosity (η ) of the adsorbed layer using QTools software (Q-Sense). More details on the model have been reported elsewhere [13,14]. This model does not take into account the frequency dependence of viscoelasticity. However, Höök et al. [16] and Larsson et al. [24] have shown that effect of frequency on viscoelasticity is very small.
Aqueous solutions at 300 μg/ml of poly(L-lysine) hydro bromide (molecular weight > 300 kDa, 81356, Sigma, Saint Louis, MO) and 21.5 μg/ml of histone (21.5 kDa, H5505, Sigma) and 100 μg/ml glutaraldehyde (GTA, 340855, Sigma) were prepared in phosphate buffered saline (PBS) at pH 7.4. Concentration of PLL, histone and glutaraldehyde was kept very low so that they do not affect the viscosity of the solution. Polished gold coated AT-cut quartz crystals with fundamental frequencies of ∼5 MHz (QSX301, Q-Sense AB, Göteborg, Sweden) were cleaned by immersion in a 1:1:5 mixture of H2 O2 (30%), NH3 (25%), and distilled water at 60 ◦ C for 20 min followed by exposure to UV/O3 for 10 min. 250 μl of the polypeptide/protein solution were injected through the sensor loop and allowed to adsorb for 40 min. The adsorbed protein on the crystal was then washed with buffer for 20 min and then cross-linked by injecting 250 μl of glutaraldehyde solution for ∼20 min. Approximately 100% of the reactive lysine present in the protein/polypeptide layer was cross-linked due to addition of GTA [22]. Mass and dissipation measurements were performed using a QCM-D (D300, Q-Sense AB, Göteborg, Sweden) in batch mode at 24 ± 0.1 ◦ C. More details about the experimental procedure have been reported elsewhere [23]. 2.2. ATR/FTIR All polypeptide/protein solution spectra were recorded in their respective aqueous buffer solution. A horizontal ATR (Magna-IR 550 Series II, Thermo Nicolet Instruments Corp., Madison, WI) accessory with a trapezoidal germanium crystal having ends cut to 45◦ generated 12 internal reflections. The spectrometer was equipped with a liquid nitrogen-cooled mercury cadmium telluride detector. To reduce the contributions of water vapor and carbon dioxide, the IR system was continuously purged with air from an FTIR Purge gas generator (Model 74-45, Balston, Haverhill, MA). Spectra were collected at a gain of 8 and an aperture of 40 with spectral resolution of 2 cm−1 . All spectra were collected in the 4000–1000 cm−1 range as sets of 256 time-averaged interferograms. Omnic (v 7.0) software (Thermo Nicolet Instruments Corp., Madison, WI) was used to analyze and measure peak heights for different sample spectra.
ηl
.
(6)
3. Results and discussion The adsorption of PLL and histone from PBS at pH 7.4 and 24 ± 0.1 ◦ C onto a gold coated quartz crystal was measured with QCM-D by recording changes in frequency and dissipation with time. In order to compare the data for the three overtones, the decrease in frequency was normalized by their overtone number. The adsorption of PLL and histone resulted in a decrease in the normalized frequency and an increase in the dissipation for the three overtones as shown in Figs. 1a and 1b, respectively. The normalized change in frequency ( f /n) and dissipation ( D) for the three overtones did not overlap. This indicates that the adsorbed layer is not perfectly elastic [16]. If a layer is not perfectly elastic, it dissipates energy and the adsorbed mass is not proportional to change in normalized frequency. Hence, f /n for three overtones does not overlap for a viscoelastic adsorbed layer such as ours. For a perfectly elastic and thin adsorbed layer, researchers have successfully used the Sauerbrey equation [25]. According to this equation, the adsorbed mass, m (ng/cm2 ) is proportional to a normalized decrease in the frequency, f /n (Hz). If the Sauerbrey equation holds, then,
m = C
f n
,
(7)
where C is the mass sensitivity constant (C = 17.7 ng cm−2 Hz−1 ). Due to the viscoelastic nature of the adsorbed layer, we then fit the Voigt viscoelastic model to the data using QTools (Q-Sense)
A.K. Dutta et al. / Journal of Colloid and Interface Science 324 (2008) 55–60
57
(a) (a)
(b)
(b) Fig. 1. Adsorption of (a) poly(L-lysine) and (b) histone onto gold. The normalized change in frequency ( f /n) decreased and dissipation ( D) increased as a function of time for the three overtones (red: n = 3, blue: n = 5, and green: n = 7; f 0 = 5 MHz). The black points describe the Voigt model for the three overtones. dq = 3.34 × 10−4 m; ρq = 2650 kg/m3 ; ρl = 1000 kg/m3 and ηl = 1.0 × 10−3 Pa s [34]. T = 24.0 ± 0.1 ◦ C.
software to obtain estimates of the mass adsorbed and the two viscoelastic fitting parameters, shear viscosity and shear modulus. In order to fit the data, we assumed that the density of the adsorbed layer was 1100 kg/m3 . Note that the density of the adsorbed layer was arbitrarily chosen and was assumed constant. The density of the adsorbed layer was likely to increase with increasing coverage. However, the adsorbed layer density primarily affects the thickness of the adsorbed layer and not the mass adsorbed, or the shear modulus and shear viscosity as demonstrated by Larsson et al. [24]. Varying the density of the adsorbed layer up to 1400 kg/m3 (density of protein [16]) did not significantly change the mass, shear viscosity or shear modulus of the adsorbed layer and thus did not affect the overall conclusions of this work. This approach has been widely used to analyze QCM-D data as reported for the adsorption of proteins [16], DNA [14,26], and polymers [18,19]. Relatively good model fits (black points) for the adsorption of PLL and histone are shown in Figs. 1a and 1b, respectively. The adsorption of PLL and histone onto the gold surface was fast and reached saturation within ∼30–40 min and the values of m obtained from Voigt model were 390–420 and 2510–2580 ng/cm2 , respectively, as shown in Fig. 2. Washing with PBS buffer resulted in desorption of loosely adsorbed PLL from the adsorbed layer and resulted in a positive increase in frequency and a concomitant drop in the adsorbed mass. Adsorption of histone onto the gold surface followed by washing with the PBS buffer did not result in significant desorption of the adsorbed protein indicating that histone adsorbed much more strongly than PLL. Also, the histone layer exhibited 3–4 times the dissipation, D, of the PLL layer (Fig. 1). Adsorbed mass
Fig. 2. Estimation of the mass adsorbed for (a) the PLL and (b) the histone layers. The mass adsorbed (m) calculated from Sauerbrey equation as a function of time for the three overtones (red: n = 3, blue: n = 5, and green: n = 7) is compared with that calculated from Voigt vicoelastic model (black points).
calculated from the Sauerbrey equation (mSauerbrey ) is lower than that estimated from Voigt model (mVoigt ) for both PLL and histone as shown in Fig. 2. However, the deviation of mVoigt from mSauerbrey for histone is much larger than that for PLL. This indicates that the adsorbed PLL layer is more rigid (or less viscoelastic) than adsorbed histone layer which is also reflected in their shear viscosity and shear modulus values (below). PLL exhibited a small drop in mass during PBS washing and GTA cross-linking (for both models). For histone, no change was observed during PBS washing while cross-linking induced a substantial change for the Voigt fit and a moderate change for the Sauerbrey prediction. The shear modulus and shear viscosity of the adsorbed PLL and histone layers were obtained from fitting the Voigt model to the experimental data and are shown in Figs. 3 and 4, respectively. Even though washing the adsorbed PLL layer with the PBS buffer released ∼80 ng/cm2 of the mass adsorbed (Fig. 2a), the change in viscoelasity of the layer was negligible. Addition of the cross-linker GTA, resulted in cross-linking of the adsorbed layer accompanied by an increase in change in normalized frequency and a drop in dissipation for both PLL and histone (Fig. 1). This resulted in a decrease in mass of the adsorbed layer for both PLL and histone (Fig. 2). This also resulted in a substantial increase in shear viscosity and shear modulus for the adsorbed PLL layer from ∼3 × 10−3 to ∼17.5 × 10−3 Pa s and from ∼2.5 × 105 to ∼2.5 × 106 Pa, respectively. For the histone layer, shear viscosity increased from ∼1.3 × 10−3 to ∼2.0 × 10−3 Pa s and shear modulus changed from ∼1.2 × 104 to ∼1.6 × 104 Pa. The higher shear viscosity and shear modulus of the cross-linked PLL layer as compared with the cross-linked histone layer was likely due to the far greater number (∼1500 lysine/molecule of PLL and ∼40 lysine/molecule of histone) of cross-linkable lysines present in the PLL layer. Since,
58
A.K. Dutta et al. / Journal of Colloid and Interface Science 324 (2008) 55–60
(a)
(b)
Fig. 3. Variation of (a) shear viscosity, η , and (b) shear modulus, μ, as a function of time for adsorption of PLL onto gold during adsorption, washing with PBS and cross-linking with glutaraldehyde (GTA) at 24.0 ± 0.1 ◦ C and pH 7.4. Cross-linking of the adsorbed polypeptide layer with GTA increased its shear viscosity and shear modulus.
(a)
(b)
Fig. 4. Variation of (a) shear viscosity, η , and (b) shear modulus, μ, as a function of time for adsorption of histone onto gold during adsorption, washing with PBS and cross-linking with glutaraldehyde (GTA) at 24.0 ± 0.1 ◦ C and pH 7.4. Cross-linking of the adsorbed protein layer with GTA increased its shear viscosity and shear modulus. Table 1 Effect of assumed density on the mass and viscoelastic properties of the adsorbed layer Assumed density (kg/m3 )
1100 1400
Mass adsorbed after GTA cross-linking (ng/cm2 )
Shear modulus after GTA cross-linking (Pa)
PLL
Histone
PLL
720–760 725–760
∼2.5 × 10 ∼2.2 × 106
230–240 225–235
after cross-linking, the viscoelastic properties for both the layers fluctuated enormously, we attribute this to to concentration variation of the randomly cross-linked network structure [27]. This increase in fluctuation was also observed due to cross-linking of the Mytilus edulis foot protein (Mefp-1) [16] and for polymerization of DNA [14]. This increase in fluctuation of viscoelastic properties could be used as a signature for the formation of a cross-linked network. The decrease in mass after GTA addition is most likely due to the release of water molecules from the compact crosslinked polypeptide/protein layer. This was also reported by Höök et al. [16] for cross-linking of an adsorbed Mefp-1 film by addition of NaIO4 . Reported values of shear viscosity and shear modulus before cross-linking were ∼2 × 10−3 Pa s and ∼1 × 105 Pa, respectively. After cross-linking with NaIO4 , Höök et al. reported that the shear viscosity and shear modulus increased to ∼5.5 × 10−3 Pa s and to ∼4.5 × 105 Pa, respectively. The adsorption of PLL and histone followed by cross-linking with the GTA, demonstrates that QCM-D can be used to study the cross-linking of proteins and other macromolecules with high precision and sensitivity. The normalized change in frequencies for the three overtones was different before cross-linking. However, after cross-linking, they were almost similar for all the three overtones indicating that they followed the Sauerbrey equation better after than before cross-linking because the adsorbed layer became more rigid. Similarly, the dissipation values after cross-linking were very small (<1 × 10−6 ).
Shear viscosity after GTA cross-linking (Pa s) Histone
6
∼1.6 × 10 ∼1.4 × 104 4
PLL
Histone
∼17.5 × 10−3
∼2.00 × 10−3 ∼1.75 × 10−3
∼13.5 × 10−3
The calculated mass adsorbed for the PLL layer from the Sauerbrey equation before cross-linking and buffer wash for the three overtones varied from 240 to 280 ng/cm2 , whereas the mass adsorbed from the Voigt viscoelastic model was estimated to be greater than 390 ng/cm2 . This large deviation indicated that that adsorbed layer was not rigid prior to cross-linking. However, after crosslinking with GTA, the mass adsorbed calculated from Sauerbrey equation varied from 220 to 235 ng/cm2 , whereas that from the Voigt model was 235–245 ng/cm2 . This close agreement between the mass prediction by the Voigt viscoelastic model and the Sauerbrey equation after cross-linking indicates that the layer became more rigid due GTA addition. Similar results were also observed for the adsorbed histone layer. To study the effect of density on the mass and viscoelastic properties of the adsorbed layer, we fitted the experimental data assuming an adsorbed layer density of 1400 kg/m3 . Comparative results are shown in Table 1. 1400 kg/m3 is the density of usual proteins and hence represents an upper bound. The effect of density on mass adsorbed varied from ∼1 to ∼4% whereas, the effect of density on shears modulus and shear viscosity varied from ∼12 to ∼25%. The mass changes due to a density increase from 1100 to 1400 kg/m3 were negligibly small. These small changes in viscoelastic parameters due to change in assumed density does not significantly affect the overall conclusion about the rheological characteristics of the adsorbed layers.
A.K. Dutta et al. / Journal of Colloid and Interface Science 324 (2008) 55–60
59
The ratio | D /( f /n)| has also been used to represent the aggregate viscoelasticity of an adsorbed layer [19,26,28]. The average | D /( f /n)| ratio for the three overtones for the three transition periods, before buffer wash, after buffer wash and after cross-linking, are shown in Fig. 5. The | D /( f /n)| ratio for the PLL layer before buffer wash was 0.087 ± 0.012 × 10−6 Hz−1 . Due to the buffer wash, some loosely adsorbed PLL molecules desorbed from the surface resulting in a decrease of the | D /( f /n)| ratio to 0.062 ± 0.008 × 10−6 Hz−1 . However, cross-linking of the adsorbed PLL layer with GTA lowered the ratio to 0.011 ± 0.001 × 10−6 Hz−1 indicating increased rigidity. In comparison, | D /( f /n)| ratio for the histone layer before buffer wash was 0.136 ± 0.009 × 10−6 Hz−1 which was much higher than that of the adsorbed PLL layer. The lower | D /( f /n)| ratio of PLL suggests that the random coil of PLL lies down flat-on on the gold surface and has less trapped water than the adsorbed histone layer. The tertiary structure of the histone and its resultant layer on the surface likely imbibe a larger amount of trapped water than the PLL layer. The | D /( f /n)| ratio for the histone layer after buffer wash did not change because, there were almost no desorption of histone from gold surface. However, after cross-linking
Fig. 5. Average | D /( f /n)| ratio for the three overtones and for the three transition periods: before buffer wash, after buffer wash and after cross-linking. PLL ((") and solid line) and histone ((2) and dashed lines). The lines are used to indicate the transition from one point to another.
(a)
(b) Fig. 6. ATR/FTIR spectra of (a) PLL and (b) histone prior to cross-linking (without GTA, solid line) and after cross-linking (with GTA, dotted line). Spectra were collected with a gain of 8 and resolution of 2 cm−1 with 256 scans for each sample. T = 24 ◦ C and pH 7.4.
60
A.K. Dutta et al. / Journal of Colloid and Interface Science 324 (2008) 55–60
of the adsorbed histone layer, the | D /( f /n)| ratio decreased to 0.028 ± 0.005 × 10−6 Hz−1 indicating increased rigidity and some loss of water. The cross-linking behavior of PLL and histone and corresponding changes in their associated water content were also investigated using ATR/FTIR. To analyze the pure protein spectra, corresponding buffer (i.e., GTA in buffer spectra for GTA addition) and water vapor spectra were subtracted from the protein spectra in aqueous solutions [29]. Water molecules exhibit two strong vibrational modes—a symmetric OH stretching and an H–O–H bending [30]. The peaks at ∼3375 and ∼1639 cm−1 are assigned to OH stretching and H–O–H bending vibrations [31,32], respectively, and are attributed to associated water molecules [33]. The calculated heights of the peaks at ∼3375 and ∼1639 cm−1 for PLL without GTA were 0.492 and 0.200, respectively (Fig. 6a). After addition of GTA, these values decreased to 0.188 and 0.081, respectively. For histone, these heights before addition of GTA were 0.384 and 0.157, respectively, and decreased only slightly to 0.374 and 0.156, respectively, after addition of GTA (Fig. 6b). The decrease in absorbance of water bands indicates release of associated water molecules from the protein/polypeptide. The larger drop in intensities in PLL as compared with histone suggests that more cross-linking was possible in PLL than in histone as also demonstrated by their changes in shear modulus and shear viscosities (Figs. 3–5). 4. Conclusions We probed the cross-linking behavior of a protein and a polypeptide at a solid–liquid interface in the presence of glutaraldehyde using QCM-D and ATR/FTIR. While the frequency and dissipation data obtained from QCM-D provided a measure of the changes in viscoelasticity of the two layers, the FTIR data gave us a qualitative measure of the changes in associated water content. The data obtained by these two independent analytical methods correlate well with each other. The frequency and dissipation measurements allowed us to probe the structural differences between the PLL peptide layer with the histone protein layer prior to and after cross-linking. Using a combined average parameter, | D /( f /n)| ratio, the value for PLL layer was lower, and hence more rigid, than the histone layer due to greater cross-linking. After cross-linking, the rigidity increased further whereby the PLL adsorbed layer was higher than that of the histone adsorbed layer due to more lysine crosslinking groups in PLL. The higher viscoelasticity and large loss of mass resulting from glutaraldehyde cross-linking for the adsorbed histone layer indicated release of large amounts of trapped water from the histone layer at the solid–liquid interface. These results also showed that the Sauerbrey equation could be used to convert the change in frequency to adsorbed mass after cross-linking where the adsorbed layer was rigid but not prior to cross-linking where it is more viscoelastic. Also after cross-linking, the fluctuation in viscoelastic properties of the adsorbed layer increased. This
increase in fluctuation was also reported for cross-linking of Mefp1 [16] and the polymerization of DNA [14]. We propose that an increase in fluctuation of the viscoelastic properties and a loss of imbibed water could be used as a signature of the formation of a cross-linked network and the amount of cross-linking, respectively. Acknowledgments This work was supported by the National Institute of Health (R01GM071329) and US Department of Energy (DE-FG02-90ER14114 and DE-FG02-05ER46249). References [1] L. Cao, F. Rantwijk, R. Sheldon, Org. Lett. 2 (2000) 1361. [2] F. López-Gallego, L. Betancor, C. Mateo, A. Hidalgo, N. Alonso-Morales, G. Dellamora-Ortiz, J.M. Guisán, R. Fernández-Lafuente, J. Biotechnol. 119 (2005) 70. [3] J. Kupec, K. Charvátová, M. Navrátil, V. Kˇresálek, M. Kˇresálková, J. Polym. Environ. 11 (2003) 93. [4] M. Diwan, T.G. Park, J. Control. Release 73 (2001) 233. [5] M. Karlsson, J. Ekeroth, H. Elwing, U. Carlsson, J. Biol. Chem. 280 (2005) 25558. [6] N.C. Vercillo, K.J. Herald, J.M. Fox, B.S. Der, J.D. Dattelbaum, Protein Sci. 16 (2007) 362. [7] A.I. Romoscanu, R. Mezzenga, Langmuir 21 (2005) 9689. [8] M. Griffin, R. Casadio, C.M. Bergamini, Biochem. J. 368 (2002) 377. [9] J.M.R. Patino, S.E.M. Ortiz, C.C. Sánchez, M.R.R. Niño, M.C. Añón, J. Colloid Interface Sci. 268 (2003) 50. [10] 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. [11] E. Dickinson, Colloids Surf. B 15 (1999) 161. [12] M. Rodahl, F. Höök, A. Krozer, P. Brzezinski, B. Kasemo, Rev. Sci. Instrum. 66 (1995) 3924. [13] M. Voinova, M. Rohdal, M. Jonson, B. Kasemo, Phys. Scri. 59 (1999) 391. [14] G. Stengel, F. Höök, W. Knoll, Anal. Chem. 77 (2005) 3709. [15] A. Domack, O. Prucker, J. Rühe, D. Johannsmann, Phys. Rev. E 56 (1997) 680. [16] F. Höök, B. Kasemo, T. Nylander, C. Fant, K. Sott, H. Elwing, Anal. Chem. 73 (2001) 5796. [17] E. Gurdak, J. Booth, C.J. Roberts, P.G. Rouxhet, C.C. Dupont-Gillain, J. Colloid Interface Sci. 302 (2006) 475. [18] J. Hedin, J.-E. Löfroth, M. Nydén, Langmuir 23 (2007) 6148. [19] A.K. Dutta, G. Belfort, Langmuir 23 (2007) 3088. [20] E.F. Irwin, J.E. Ho, S.R. Kane, K.E. Healy, Langmuir 21 (2005) 5529. [21] K. Luger, A.W. Mäder, R.K. Richmond, D.F. Sargent, T.J. Richmond, Nature 389 (1997) 251. [22] C. Marquié, A.M. Tessier, C. Aymard, S. Guilbert, Nahrung 42 (1998) 264. [23] F. Höök, M. Rodahl, P. Brzezinski, B. Kasemo, Langmuir 14 (1998) 729. [24] C. Larsson, M. Rodahl, F. Höök, Anal. Chem. 75 (2003) 5080. [25] G. Sauerbrey, Z. Physica 155 (1959) 206. [26] T.H. Nguyen, M. Elimelech, Biomacromolecules 8 (2007) 24. [27] Y. Rabin, R. Bruinsma, Europhys. Lett. 20 (1992) 79. [28] A.K. Dutta, A. Nayak, G. Belfort, Macromolecules 41 (2008) 301. [29] A. Sethuraman, G. Belfort, Biophys. J. 88 (2005) 1322. [30] D. Eisenberg, W. Kauzmann, The Structure and Properties of Water, Oxford Univ. Press, New York, 1969. [31] N.I.E. Shewring, T.G.J. Jones, G. Maitland, J. Yarwood, J. Colloid Interface Sci. 176 (1995) 308. [32] F.O. Libnau, O.M. Kvalheim, A.A. Christy, J. Toft, Vibr. Spectrosc. 7 (1994) 243. [33] S. Ziming, K.G. Peter, N.H. David, Polyhedron 17 (1998) 1511. [34] M. Rodahl, B. Kasemo, Sens. Actuators A 54 (1996) 448.