Article pubs.acs.org/Langmuir
Antibody Adsorption and Orientation on Hydrophobic Surfaces Meredith E. Wiseman and Curtis W. Frank* Department of Chemical Engineering, Stanford University, 381 North-South Mall, Stauffer III, Stanford, California 94305-5025, United States ABSTRACT: The orientation of a monoclonal, anti-streptavidin human IgG1 antibody on a model hydrophobic, CH3-terminated surface (1-dodecanethiol self-assembled monolayer on gold) was studied by monitoring the mechanical coupling between the adsorbed layer and the surface as well as the binding of molecular probes to the antibodies. In this study, the streptavidin antigen was used as a probe for the Fab portions of the antibody, while bacteria-derived Protein G′ was used as a probe for the Fc region. Bovine serum albumin (BSA) acted as a blocking protein. Monolayer coverage occurred around 468 ng/cm2. Below 100 ng/cm2, antibodies were found to adsorb flat-on, tightly coupled to the surface and unable to capture their antigen, whereas the Fc region was able to bind Protein G′. At half-monolayer coverage, there was a transition in the mechanism of adsorption to allow for vertically oriented antibodies, as evidenced by the binding of both Protein G′ and streptavidin as well as looser mechanical coupling with the surface. Monolayer coverage was characterized by a reduced level in probe binding per antibody and an even less rigid coupling to the surface.
1. INTRODUCTION Antibodies, comprising two identical light chains and two identical heavy chains, are known to adopt a distinct, three-lobed structure; two of these identical lobes, the fragment antigenbinding (Fab) regions, contain the antigen-binding sites, while the other consists of the fragment crystallizable (Fc) region and is responsible for recruiting components of the immune system. Whether the antibody adsorbs onto hydrophobic surfaces in an end-on (Fab-up or Fab-down), side-on, or a f lat-on orientation (see Figure 1) is of interest in designing assays that make use of
(Fc-down) orientation becomes favored. This result alone suggests that the natural orientation of antibodies on hydrophobic surfaces is not strictly Fab-up. A priori, one would expect the flat-on orientation to be preferred in conditions where electrostatic interactions are weak, as this would maximize the van der Waals interactions with the surface. In fact, Monte Carlo simulations predict this to be the case.4,5 In conditions where electrostatic forces are important, such as when the antibody has a strong dipole, the ionic strength is low, and the surface is charged, the dipole moment of the antibody tends to determine its orientation on the surface. However, when van der Waals forces dominate (neutral surface, high ionic strength), simulations predict that the flat-on orientation would be favored. One straightforward method for determining the approximate orientation of antibodies is to examine the mass of a monolayer. If a homogeneous, close-packed structure is assumed, the crystallographic dimensions of Fab and Fc fragments can be used to predict the mass loading of monolayers consisting of antibodies in different orientations. An often-cited work by Buijs and colleagues predicts the mass of a flat-on layer to be 2.0 mg/m2, while the mass of an end-on layer would vary depending on the angles between the Fab fragments (2.6 mg/m2 for Fab fragments in a line forming a “T” shape, 3.7 mg/m2 for a “Y” shape, and up to 5.5 mg/m2 when the Fab fragments lie close together and parallel6). Although this can give a good first approximation of orientation, it relies on the measurement method being quantitative and also assumes that the orientation of antibodies in the film is homogeneous. Furthermore, it cannot distinguish between Fab-up or Fab-down, which are both end-on orientations.
Figure 1. Terms used when referring to antibody orientation. The heavy chains are shown in white and the light chains in gray. The Fab fragments are the lobes comprising light and heavy chains, while the Fc fragment is made up of only heavy chains.
passive adsorption of antibodies onto hydrophobic surfaces, such as the commonly used enzyme-linked immunosorbent assay (ELISA); the efficiency of capture would be much greater if the Fab-up orientation were favored. A common protocol for ELISA assays involves passive adsorption of the capture antibody on polystyrene plates, indicating that passive adsorption of antibodies on a hydrophobic surface leads to an orientation or mixture of orientations where the antigen can be captured, i.e., neither flat-on nor Fab-down. However, it has been shown that preadsorbing the bacterial proteins Protein G or Protein A, which bind specifically to the Fc-fragment of the antibody, can increase the efficiency of antigen capture1−3 because the Fab-up © 2011 American Chemical Society
Received: August 8, 2011 Revised: December 15, 2011 Published: December 19, 2011 1765
dx.doi.org/10.1021/la203095p | Langmuir 2012, 28, 1765−1774
Langmuir
Article
anti-β-hCG IgG on hydrophilic silica is to some degree end-on or side-on, as a purely flat-on orientation would not allow for antigen binding after the BSA blocking step. By contrast, Chen and colleagues report a flat-on orientation on a hydrophobic CH3-terminated surface, based on a surface coverage of 2.2 mg/m2 and a binding ratio of ∼0.33, as determined by SPR.14 One compelling recent study that gives information about the possible orientation on hydrophobic surfaces is an enzymatic degradation study by Yu and Ghosh, who showed that polyclonal human IgG passively adsorbed onto a hydrophobic membrane was vulnerable to both pepsin, which degrades the Fc fragment, and papain, which cleaves the Fab fragment.15 They concluded that the antibody bound the membrane through the hinge region and an adjacent portion of the CH2 domain in the Fc fragment, which is equivalent geometrically to a flat-on orientation. In summary, while there are many methods available to acquire information about the orientation of antibodies adsorbed on hydrophobic surfaces, these methods have not yielded a consensus. While the flat-on orientation is the most commonly cited, other studies find that these particular antibodies are able to bind their antigen, which would be sterically impossible in the flat-on state, thus raising questions about the orientation. In this study, we use QCM-D to show that the orientation is a function of surface coverage, thereby explaining why studies looking at low-density, individual antibodies report flat-on orientation while those looking at denser films near monolayer coverage reach different conclusions.
Another way to study orientation is to image antibodies directly. Several high-resolution AFM studies have been able to resolve the individual Fab/Fc domains when very dilute samples are imaged in air after extensive drying.7 However, it is not clear that the dry scans are representative of the behavior in liquid. Unfortunately, good resolution is difficult to achieve in liquid AFM. Xu et al. imaged anti-β-hCG IgG1 on hydrophilic silica in water and found possible protein clusters 30−33 Å in height by 200−500 Å in lateral size; however, there was evidence of tip− protein interaction that might have led to motion of the antibodies on the surface, and resolution was insufficient to resolve individual domains.8 Their study also showed that drying the sample had a large effect on the morphology of the film. By contrast, a subsequent AFM study in air using the same system found no evidence of aggregation at low surface coverage, with aggregation occurring only at higher coverage.9 Because the resolution of AFM in liquid is too poor to image domains, and drying the film may alter its structure, we cannot rely on AFM to give reliable orientational information about the antibodies, only information on surface aggregation that occurs at larger length-scales. Other works have studied orientation by characterizing the protein density as a function of distance away from the surface using neutron reflection. Xu et al., again using anti-β-hCG IgG1 on hydrophilic silica, found that the film structure could best be described by a three-layer model: a thin 10 Å inner layer with a protein volume fraction φp of 0.22 bordered by a dense 30 Å middle layer of fraction φp = 0.42 and then a 25 Å outer layer of φp = 0.10.8 They also found evidence that electrostatic interactions between domains were very important in determining film structure.10 Petrash and colleagues performed a similar study on hydrophobic substrates, finding the first 50 Å closest to the surface to be very dense with a protein fraction of up to 0.8, and then the next 50 Å to be more diffuse, with a value closer to 0.4.11 They interpreted these observations as being a result of a mixture of flat-on adsorbed antibodies leading to the dense layer near the surface, interspersed with some having an end-on orientation that led to the more diffuse outer layer. One promising, direct approach to determining orientation uses TOF-SIMS. Here, the surface of the sample is sputtered, and the amino acids observed on the surface are compared to the known sequence of the antibody to determine which domains of the antibody are on the surface.12 While promising, this method has not yet been fully implemented because of the difficulties in finding uniquely distinguishable patterns of amino acids in different locations on the surface of the antibody. Furthermore, the measurements are performed on dry samples in vacuum, which could alter the film structure. Another way of assessing antibody orientation is to determine which of its domains are physically accessible on the surface. To this end, several groups have measured the antigen binding capacity of anti-β-hCG IgG. Xu and colleagues studied this antibody on hydrophilic silica surfaces by spectroscopic ellipsometry and found the binding ratio of hCG (moles hCG bound per antibody) after a BSA-blocking step to vary with pH but generally decrease with increasing surface coverage, for example from 0.7 at a low coverage of ∼1.4 mg/m2 down to a binding ratio closer to 0.05 at maximum coverage of 2.5 mg/m2.13 This is interpreted as being the result of increased steric hindrance at high surface coverage. Wang et al., using a lower pH, report similar trends of 0.7 at low coverage of 1 mg/m2 down to 0.13 at maximum coverage of 2.7 mg/m2.9 Their findings of relatively high antigen binding capacity at low surface coverage would seem to suggest that the orientation of
2. EXPERIMENTAL SECTION 2.1. Materials. Anti-streptavidin IgG1 (henceforth abbreviated IgG), MW = 142 kDA, pI ∼ 9, was provided by Amgen (Thousand Oaks, CA). Streptavidin (SA), Protein G′ (G′), and bovine serum albumin (BSA) were purchased from Sigma. Protein concentration was determined spectroscopically using extinction coefficients of 1.586 (IgG), 3.1 (SA), 1.277 (G′), and 0.667 (BSA) at 280 nm. Proteins were frozen in small aliquots and diluted upon use in a pH 5, 10 mM acetate buffer, with 100 mM NaCl (sodium acetate trihydrate, glacial acetic acid, sodium chloride purchased from Sigma, and Milli-Q water). 2.2. Quartz Crystal Microbalance with Dissipation (QCM-D). In this technique, the mass adsorbed onto a crystal surface is measured by monitoring changes in its resonant frequency. Briefly, an rf voltage is applied across electrodes deposited onto the surface of an AT-cut quartz crystal, resulting in a thickness shear mode oscillation. The oscillation is driven at the crystal’s fundamental frequency or an overtone. As matter adsorbs onto the surface of the crystal, there is a resultant change in the frequency of the oscillation. For completely rigid films, there exists a linear relationship between the mass adsorbed Δm and the frequency shift:
Δm =
C Δfz z
where C is the Sauerbrey mass sensitivity constant (around −18 ng/cm2 Hz for a 5 MHz crystal in water) and Δfz is the frequency change measured at the zth overtone (only odd overtones can be excited 16). This so-called Sauerbrey relationship is applied both in air and in liquids, although the mass measured in liquids can include water molecules mechanically coupled within the film being studied and therefore reflects a hydrated film mass. During the adsorption of proteins, the resultant film is rarely rigid, so in addition to the resonant frequency f, QCM-D measures a dissipation value D proportional to the inverse of the characteristic time scale τ of the oscillation’s exponential decay:
D=
1 πf τ
Changes in the dissipation value ΔD relative to a baseline established on the clean crystal surface give information about the 1766
dx.doi.org/10.1021/la203095p | Langmuir 2012, 28, 1765−1774
Langmuir
Article
Table 2. Time Course of Solutions Useda
rigidity of the adsorbed film, with a perfectly rigid film having a ΔD value of zero. The Sauerbrey equation is generally applied when the dissipation value is below 1 × 10−6, with more complicated viscoelastic modeling being necessary above that value to accurately determine the adsorbed mass. 2.3. Experimental Setup for QCM-D. Gold-coated quartz crystal sensors were purchased from Q-Sense. These were coated with a selfassembled monolayer of 1-dodecanethiol (DDT, Sigma) by overnight immersion in a 1 mM solution of DDT in ethanol, followed by rinsing in ethanol and drying with nitrogen. The crystals were cleaned between experiments following the manufacturer’s recommended protocol.17 Briefly, they were immersed overnight in 10 mM sodium dodecyl sulfate (SDS), rinsed with water and ethanol, dried with nitrogen, exposed to oxygen plasma for 3 min at 10.5 W, immersed in ammonium peroxide at 80 °C for 5 min, rinsed in water and ethanol, dried with nitrogen, and finally exposed to oxygen plasma for another 3 min. Measurements were performed on the E4 experimental platform from Q-Sense, which consists of four chambers running in parallel under continuous flow driven by a peristaltic pump at a constant flow rate of 100 μL/min at 25 °C. Once the crystals were loaded into the chambers, their resonant frequencies were measured to calculate the mass sensitivity constant C. Crystals with abnormally high total dissipation values in liquid, abnormally large spurs in the frequency sweep, or an amplitude magnitude less than 1 at any resonant frequency were discarded. Buffer was introduced slowly into the chambers, and the direction of flow was reversed quickly to ensure that the baseline was stable and no bubbles had formed on the crystal surface. To protect against drift and the buildup of contaminants on the surface during equilibration, only crystals that demonstrated a stable baseline (slope of less than 2 Hz/h) within 10 min of buffer introduction were used. The experiment was started as soon as a stable baseline was obtained in all chambers, no more than 10 min after the initial introduction of buffer. The time course of experiments where IgG was allowed to adsorb directly to the methylated crystal surface is summarized in Table 1. Every time a
a
a
solution
solute concn, μg/mL
0 10 20 30 50 60 70
buffer IgG buffer BSA buffer probe buffer
0 (varied) 0 25 0 10 0
solution
solute concn, μg/mL
0 5 15 25 35 65 75
buffer SA (G′) buffer IgG buffer SA or G′ (G′ or SA) buffer
0 10 0 100 0 10 0
Orientation of IgG biased by pre-adsorbing SA or G′.
gained from patterns in probe binding, the orientation of the antibody was biased by first depositing G′ or SA onto the surface. The time course of these experiments are described in Table 2. We expect a Fab-up orientation to be favored on the G′ surface18 and an Fc-up or Fab-down orientation on the SA surface. The frequency and dissipation shifts for the four experiments conducted are shown in Figure 2. Solid lines are used when SA was preadsorbed, which should lead to a Fab-down-like orientation, while dashed lines are used when G′ was preadsorbed, which should lead to a Fab-up-like orientation. Once the antibody was deposited on top of the SA or G′ layer, it was assessed for its ability to capture SA (black lines) and G′ (gray lines). Preadsorbing SA leads to a frequency shift of −30.5 Hz, or a Sauerbrey mass of 548 ng/cm2, and a dissipation shift of 0.13 × 10−6 after rinsing. Preadsorbing G′, a smaller protein, leads to a −18.3 Hz shift, or 329 ng/cm2, and a relatively large dissipation change of 0.62 × 10−6. On the SA surface, an average of 407 ng/cm2 IgG was bound with a dissipation shift of 1.39 × 10−6 (relative to the SA base layer after rinsing), whereas on the G′ surface, the coverage was 713 ng/cm2 with a dissipation of 0.53 × 10−6. The adsorption curves for the probe proteins are shown in more detail in Figure 3. The adsorption is shown in units of moles of probe per mole surface IgG, calculated using the amount of antibody on the crystal measured in the previous step. Figure 3A shows the result for G′ (light gray circles in diagram) preadsorbed on the original surface. The stoichiometry of SA:IgG binding at the maximum point is around 0.68, although equilibrium does not appear to have been reached during the adsorption phase. The stoichiometry of G′:IgG binding is lower at 0.38. With the antibody adsorbing onto a layer of predeposited G′, almost 1.8 times as much SA as G′ were able to bind. Figure 3B shows the result with SA preadsorbed onto the original surface (dark gray rectangles in diagram). In this experiment, the drifting baseline indicates slow desorption of the antibody. Beginning around t∼66 min when the probe protein solutions enter the QCM chamber, there was a spike in the frequency shift measured for the trial using the G′ probe. Because this is accompanied by a decrease in the dissipation value, we believe this is due to a film contraction and associated water loss (see discussion in 4.1. Oriented Immobilization of Antibody). Superimposed upon this behavior, we see that more G′ binds than SA: the maximum G′:IgG binding ratio measured is 0.51 and, if the shift due to the water loss spike is taken into account, the value becomes 1.22. As can be seen, when G′ is preadsorbed, the antibody adsorption is biased in such a way that more SA than G′ binds, whereas when SA is preadsorbed, binding of G′ is preferred.
Table 1. Time Course of Solutions Useda start time, min
start time, min
IgG adsorbed directly onto CH3 surface.
solution was to be changed, the flow was stopped 30 s prior in order to exchange the source vials, and then flow was resumed. The time for the new solution to traverse the tubing and reach the chamber once flow was restarted was found to be approximately 85 s. For example, buffer was replaced with an IgG solution at t = 10 min, so the flow was stopped after 530 s to exchange the solutions, flow was resumed at 600 s, and IgG entered the chamber at around 685 s. As a complement to the experiments where IgG was allowed to adsorb directly to the hydrophobic crystal surface, a second set of four experiments was performed in which the orientation of the antibody was biased by preadsorbing either streptavidin or Protein G′, after which the layer of IgG that resulted was interrogated by either SA or G′. Because BSA was found to displace SA, no blocking step was used and, instead, the concentration of antibody was kept high. The time course of these experiments is summarized in Table 2.
3. RESULTS 3.1. Oriented Immobilization of Antibody. To verify that the probe proteins are binding to the expected regions on the antibody, and to show that orientational information can be 1767
dx.doi.org/10.1021/la203095p | Langmuir 2012, 28, 1765−1774
Langmuir
Article
Figure 3. Preadsorbing Protein G′ or streptavidin (SA) biases the orientation of the anti-streptavidin antibody on the surface. (A) With G′ preadsorbed, the antibody adopts a Fab-up-like orientation and is able to capture more SA than G′. (B) With SA preadsorbed, the antibody adopts a Fab-down-like orientation and is able to capture more G′ than SA. The spike around 66 min in the G′ adsorption is associated with water loss and film contraction.
led to more IgG adsorption, and less BSA. In addition, more Protein G′ binds as the bulk solution concentration of IgG increases. It is important to note that no G′ became bound to the control surface lacking IgG, indicating that the interaction between G′ and BSA was successfully screened out by the high salt concentration of the buffer used (see discussion in 4.2.3. BSA as a Blocking Protein). Furthermore, the adsorption of antibody onto the bare DDT surface does not appear to be easily reversible. This is evident in the buffer rinsing step immediately following IgG adsorption. For the 10 μg/mL IgG trial, almost no detectable desorption occurred. For the 1000 μg/mL, only a small amount desorbed despite the high level of coverage. This trend was found to persist over a long period of time. In other experiments, it was found that after an initial small amount of desorption, antibody layers would persist at a constant mass during buffer rinses for hours without desorption. For example, a layer initially weighing 194.6 ng/cm2 decreased upon rinsing to 184.4 ng/cm2 after 12 min, but 150 min later was still 184.1 ng/cm2. Because any desorption of the antibody happened quickly, the surfaces in this experiment were rinsed for only 10 min before BSA blocking to minimize the age of the layer, lessening the time available for any potential surface rearrangements to occur. 3.2.2. IgG Surface Coverage as a Function of Bulk Concentration. In Figure 5, IgG surface concentrations after buffer rinsing are presented as a function of bulk IgG concentration. There is good but not perfect reproducibility between solutions of the same bulk IgG concentration. There appears to be a plateau in the IgG surface coverage at a mass loading of approximately −26 Hz
Figure 2. Frequency (A) and dissipation (B) values for the oriented immobilization of anti-streptavidin IgG. In four separate experiments, represented schematically in panel C, the orientation of the antibody was biased by first preadsorbing the antigen streptavidin (SA, solid lines) or the Fc-binding Protein G′ (G′, dashed lines) onto a CH3terminated surface. The antibody was then assessed for its ability to capture SA (black lines) or G′ (gray lines). Table 2 describes the time course of these experiments.
3.2. Antibody Orientation on Hydrophobic Surface. To gain information about the orientation of an antibody adsorbing onto a bare hydrophobic surface, the binding patterns of its antigen (SA) and Protein G′ were measured at different antibody surface concentrations after a BSA blocking step. The time course of this experiment is described in Table 1. 3.2.1. Typical Adsorption Experiments. Figure 4 shows the results of typical QCM-D experiments. Shaded regions indicate times when a protein solution (IgG, BSA, probe) was present in the QCM chamber; white regions indicate the presence of buffer. Here, IgG at concentrations of 1000 μg/mL, 10 μg/mL, and 0 (control) were used in the antibody adsorption step, leading to IgG frequency shifts around −58, −26, and 0 Hz, respectively. As expected, higher bulk solution concentrations of IgG 1768
dx.doi.org/10.1021/la203095p | Langmuir 2012, 28, 1765−1774
Langmuir
Article
3.2.3. Evolution of Dissipation Value. Changes in the dissipation value of a film can reveal important structural information. The dissipation value D is defined as being proportional to the ratio between the energy lost during one oscillation cycle of the crystal and the total energy of the oscillator. Small dissipation values indicate rigid films, while larger values suggest a more viscoelastic film. In Figure 6, the dissipation values (D)
Figure 4. Typical results of adsorption experiments on a hydrophobic surface. IgG concentrations of 0, 10 μg/mL, and 1 mg/mL were applied to the clean CH3-terminated surface (first shaded region) and then rinsed with buffer. BSA at a concentration of 25 μg/mL was applied as a blocking protein (second shaded region). Protein G′ at 10 μg/mL (third region) is shown here as a probe for the Fc region of the antibody.
Figure 6. Dissipation analysis of IgG adsorption. Dissipation values ΔD are plotted as a function of the calculated Sauerbrey mass for different bulk concentrations of IgG. Values below 1 × 10−6 are considered to lie within the linear Sauerbrey regime. The dissipation change stays close to zero, indicating a rigid film, until the development of half a monolayer. At this point, the slope increases, indicating that a softer adlayer is being formed.
are plotted as a function of the mass loading on the crystal for several representative trials using different bulk concentrations of antibody. The first thing we note is that data from bulk IgG concentrations over 2 orders of magnitude collapse onto the same curve, suggesting that the same mechanism of adsorption is occurring at both high and low bulk concentrations. Furthermore, the relation between D and the mass adsorbed is not linear; instead, the slope increases as the layer of antibody is built up on the surface. Initially, the curve is quite flat, suggesting a very rigid mechanical coupling between the antibody and the surface. Around a half monolayer, the slope becomes greater, indicating a looser mechanical coupling of subsequently adsorbed antibodies with the surface. The slope gets even steeper at just over one monolayer, meaning the antibodies adsorbing in this regime are even less tightly bound. 3.2.4. Molecular Probes. After the antibody was allowed to adsorb to the surface from bulk solutions of varying concentration, the surface was blocked with BSA and then interrogated with the proteins SA or G′, which bind specifically to the antibody. The amounts of SA and G′ that remained after rinsing were used to infer information about the orientation of antibody on the surface. For these values to be meaningful, it was important that there be no interactions between either of the probes and the blocking protein BSA. Although an engineered version of the Fc-receptor Protein G with its BSA-binding domains removed (Protein G′) was used for this study, residual interactions were still found when using a 10 mM pH 5 acetate buffer. For example, when a 10 μg/mL solution of Protein G′ was passed
Figure 5. IgG surface coverage after 10 min adsorption time and rinsing as a function of bulk concentration. An apparent ‘monolayer’ emerges at a coverage of around 468 ng/cm2 as determined by QCM-D. At higher concentrations, significant unwashable adlayers are present beyond monolayer coverage, indicating aggregation on the surface.
(Sauerbrey 468 ng/cm2). The dissipation values of these films show a similar trend, with a shoulder around 0.5 × 10−6 (data not shown). For the purposes of discussion, we define this value (468 ng/cm2) as monolayer coverage. Because our molecular probe method is only sensitive to the surface of the layer being studied, we are more concerned with behavior in the submonolayer coverage regime; however, there appear to be some interesting phenomena once the bulk concentration of antibody increases beyond 100 μg/mL. For concentrations above this, more than one layer of antibody persists on the surface even after rinsing. This may indicate antibody− antibody interactions and possible aggregation, but further experimentation will be necessary to resolve the behavior in this high-surface-concentration regime. 1769
dx.doi.org/10.1021/la203095p | Langmuir 2012, 28, 1765−1774
Langmuir
Article
over a crystal covered with physisorbed BSA, roughly 23 ng/cm2 remained bound to the surface. Because increasing the salt content of the buffer was found to decrease this bound amount, 100 mM NaCl was added to the buffer (corresponding to the concentration at which the amount of G′ binding to BSA was found to be undetectable). The amounts of probe bound after BSA blocking and rinsing are reported as a function of surface IgG concentration in Figure 7. Both probes are clearly captured by the antibody, with
described in Table 2 are presented as well. White squares represent the trials where G′ was preadsorbed (Fab-up), while black triangles represent SA being preadsorbed (Fab-down). In Figure 7B, the G′ binding value for the Fab-down oriented immobilization data point has been extrapolated to compensate for water loss due to film contraction, as discussed in 4.1. Oriented Immobilization of Antibody. It is notable that the data for adsorption onto the bare hydrophobic surface (gray) lie between the values for oriented immobilization (black and white). Figure 7A shows the amount of SA bound as a function of IgG surface coverage. It appears that there is a threshold around 100−150 ng/cm2 below which no binding is detected, and above which the amount bound is roughly linear with surface coverage. In Figure 7B, the amount of G′ bound is shown. In this case, binding increases roughly linearly from zero IgG coverage, where no G′ binding occurs, until a monolayer. In the adsorption of both SA and G′, there is a discontinuity in the slope of the probe bound when monolayer coverage of IgG is achieved. The data expressed in Figure 7 are plotted again in Figure 8, normalized by the amount of antibody on the surface so that the amount of probe on a per-molecule IgG basis is shown as a function of surface coverage. Error bars corresponding to uncertainty in the QCM measurement of ±5 ng/cm2 are included. Since the values at low surface coverage are computed by dividing by progressively smaller numbers, the relative magnitude of this error increases, approaching zero coverage. However, it is still clear that the per-molecule amount of SA bound converges to zero earlier than for G′. For SA, the maximum per molecule IgG binding level occurs just shy of a monolayer, with ∼0.5 SA/IgG. The maximum for G′ is more pronounced, occurring between 200 and 300 ng IgG/cm2 with up to 0.95 G′/IgG. At a monolayer, the binding ratio for G′ is around 0.8 G′/IgG. Again, the probe-binding data for simple adsorption of IgG directly onto the hydrophobic surface lie between the values for oriented immobilization points generated by preadsorbing SA or G′.
4. DISCUSSION 4.1. Oriented Immobilization of Antibody. Preadsorbing the Fc-binding Protein G′ or the antigen onto the surface was found to influence the orientation of subsequently adsorbed antibody, leading to different patterns in probe binding. Although for convenience these orientations are referred to as Fab-up and Fab-down, in reality the orientations of the protein on the surface are most likely not the canonical Fab-up or Fab-down sketched in Figure 1. Protein G′ is thought to bind the cleft between the CH2 and CH3 domains in the Fc region,19 so preadsorbing G′ onto the surface should bias the antibody to orient with the middle, and not the tip, of its Fc region in contact with the surface. One can imagine that this would result in something closer to side-on rather than Fab-up, depending on the flexibility of the hinge region. A side-on orientation would result in one Fab fragment extending out toward the bulk solution. This is consistent with the observed result that 0.68 antigens are captured per antibody, as opposed to the two that would be predicted in an ideal Fab-up orientation. The side-on orientation also explains why the G′ binding was so high (0.38), because in this orientation the G′ binding site on the other heavy chain would be accessible, although perhaps slightly sterically hindered by adjacent proteins. The low dissipation value of the antibody layer also favors the
Figure 7. Bound amounts of streptavidin (A) and Protein G′ (B) on BSA-blocked layers of IgG as a function of IgG surface density. Results from the oriented immobilization experiment (Figure 2) are included for reference, with the Fab-up-like (G′ preadsorbed) orientation denoted by open squares and the Fab-down-like (SA preadsorbed) orientation by filled triangles.
the amount bound increasing with the amount of antibody present on the surface, indicating that a specific binding interaction is occurring between the antibody and the probe. For both streptavidin (SA) and Protein G′ (G′), the amount bound was reduced to zero when no antibody was present on the surface. In fact, for SA the amount bound at low IgG surface coverage was very slightly negative. Although the amounts are near the limit of detection for QCM-D in liquids, here the mass loss on the crystal surface was due to the slow desorption of BSA from the blocking step. In Figure 7, amounts of probe bound to antibodies adsorbed onto a bare hydrophobic surface are shown in gray, corresponding to the experiment described in Table 1. For the sake of comparison, data from the oriented immobilization experiment 1770
dx.doi.org/10.1021/la203095p | Langmuir 2012, 28, 1765−1774
Langmuir
Article
results from desorption/displacement of protein, as it happens on a time scale much shorter than the adsorption of G′, which can be followed after the spike. It is also unlikely that the spike is an artifact of viscoelastic effects alone. Since viscous losses in the adlayer result in the Sauerbrey mass underestimating the true mass, a sudden decrease in the dissipation value without mass exchange should lead to an increase in the Sauerbrey mass, not a decrease as observed. Viscoelastic modeling of the data (manuscript in preparation) suggests that there is a significant loss in mass associated with this event. We suspect that this is due to a loss of mechanically coupled water during a structural change precipitated by the addition of Protein G′. In the Fab-down orientation, the Fc fragments would be facing out toward the bulk, with their entire surface hydrated (i.e., not in contact with other protein). If they were to change their orientation, either by rotating or associating with neighboring Fc fragments, this would decrease their hydrated surface area and cause a mass loss and dissipation decrease to be registered by QCM-D. Several authors have argued that dehydration is responsible for the mass loss observed in QCM-D during protein conformational changes, most notably Höök et al. with the cross-linking of mussel adhesive protein,20 and more recently Furusawa et al. with calmodulin21 and Lee et al. in the HIV-1 envelope protein gp120.22 Although we suspect that this phenomenon is also occurring here, it is not entirely clear what sort of conformational change is being induced by Protein G′. Further experimentation is necessary to understand the exact nature of the structural change occurring during this step. Although preadsorbing G′ or SA onto the surface does not necessarily lead to canonical all-Fab-up or all-Fab-down orientations, it does systematically bias the orientation of the antibody in such a way that qualitatively different probe binding patterns are observed. In the next section, we use probe binding patterns to determine the orientation of antibodies adsorbed to a bare, CH3-terminated hydrophobic surface. 4.2. Antibody Orientation on Hydrophobic Surface. 4.2.1. IgG Surface Coverage as a Function of Bulk Concentration. Many analyses of protein−surface interactions attempt to describe their systems using simple equilibrium adsorption models such as the Langmuir isotherm. However, these models are rarely applicable in the case of protein adsorption. In our system, for example, multilayer adsorption can occur, and the surface can induce structural changes in the protein. Furthermore, with the QCM-D detection method used in this study, which is sensitive to bulk solution properties as well as changes occurring on the crystal surface, a reliable adsorption isotherm cannot be obtained. To measure surface coverage on the crystal at equilibrium, the protein solution would have to be present in the measurement chamber; however, the viscosity of protein solutions is highly concentration-dependent (affecting the value of the Sauerbrey constant), and high-concentration solutions are non-Newtonian, leading to a non-Sauerbrey response measured by QCM-D that mostly disappears when the solution is replaced with buffer. Given these limitations, and to mitigate the effect of surfaceinduced conformational changes, the time scale of adsorption was kept to a minimum and Sauerbrey calculations were performed using data obtained once the solution in contact with the crystal surface had been exchanged for buffer. For this reason, the data presented in Figure 5 are not an isotherm and instead refer to the amount of protein remaining on the surface after rinsing. Moreover, the plateau emerging at −13 Hz or 468 ng/cm2 does not represent a monolayer in the strictest sense of the word; however, the concept is useful and we believe that this value corresponds to the filling up of all available sites on the
Figure 8. Bound amounts of probe molecule (A, streptavidin; B, Protein G′) on a per-molecule IgG basis. Results from the oriented immobilization experiment (Figure 2) are included for reference, with the Fab-up-like (G′ preadsorbed) orientation denoted by open squares and the Fab-down-like (SA preadsorbed) orientation by filled triangles. Error bars correspond to a frequency shift in the probe measurement of ±5 ng/cm2.
side-on explanation, as this would lead to a dense layer in contact with the surface and relatively good mechanical coupling. The conclusion that the orientation of the antibody on a surface containing Fc-receptor protein is tilted and not Fab-up is not without precedent, as Bae et al. conclude that the tilt is a function of antibody surface density,18 which was held constant in our oriented immobilization experiment. By contrast, the ideal Fab-down picture seems reasonably appropriate to describe the antibody adsorbed on a surface covered with antigen. It is possible that some of the antibodies are interacting with only one surface antigen, because having them all bind with exactly two would require the distribution and orientation of streptavidin on the surface to exactly match the geometry of the antibody. However, the SA binding capacity is quite low, which one would not expect if there were any accessible (i.e., not surface-attached) Fab fragments. The unusually high dissipation value of this film is consistent with Fab-down, as the two points of contact with the surface and the end-on orientation of the antibody would lead to poor mechanical coupling. The unexpected mass loss spike upon G′ binding, which occurs very quickly and is associated with a step drop in the dissipation value, is somewhat perplexing. It is unlikely that the mass loss 1771
dx.doi.org/10.1021/la203095p | Langmuir 2012, 28, 1765−1774
Langmuir
Article
the surface and then fill in the gaps with more vertical orientations. Below 200 ng/cm2, the shift in dissipation value of the crystal is almost zero, meaning that this initial layer of adsorbed antibodies is extremely rigidly coupled to the surface. This is what we would expect of the flat-on orientation, with antibodies deposited in a thin and dehydrated layer on the surface and having a maximum number of contact points. We expect the Sauerbrey mass below this threshold of 200 ng/cm2 to be fairly quantitative, as it is measured at an extremely low dissipation value. Coincidentally, 200 ng/cm2 is the predicted mass density for the flat-on orientation.6 Above 200 ng/cm2, there is an increase in slope, meaning that once an initial 200 ng/cm2 flat-on layer of antibodies has adsorbed, newly incident antibodies are oriented in such a way as to cause less rigid mechanical coupling to the surface. Because the adsorption data suggest that monolayer coverage is not achieved until 468 ng/cm2, we believe that this is not due to stacking on top of the previous layer but instead to a filling of interstices with endon or side-on oriented antibodies, or a displacement of flat-on antibodies in favor of more vertical ones. Due to the ability of the full monolayer to capture Protein G′ and the apparently very strong adhesion of the antibody to the surface, we believe the former explanation is more likely, although further study would be necessary to distinguish between the two possibilities. Once the whole surface is filled, any additional antibodies are not directly in contact with the surface at all and are therefore even more loosely coupled to it, leading to yet higher dissipation values and a further increase in slope once monolayer coverage is achieved. 4.2.3. BSA as a Blocking Protein. Despite being commonly used as a blocking protein in ELISA assays and in studies of antibody binding,9,13,14,26 BSA is far from an ideal blocking protein. Its disadvantages most relevant to this study are its relatively large size, its ability to carry charge, and its nonnegligible interactions with the probe Protein G′. However, its defined nature and relatively simple structure make it a better choice than other commonly used blocking protein preparations (nonfat dry milk, casein, or fish gelatin, for example), and its ability to resist displacement by the probe protein makes it a better choice than a nonionic surfactant such as Tween-20. Unlike the smaller surfactant, BSA is relatively large, which is undesirable because it might sterically hinder probe binding sites or fail to block surface gaps smaller than itself (but large enough for Protein G′ to adsorb nonspecifically). The remaining disadvantages of BSA, its surface charge and non-negligible interaction with Protein G′, were minimized by choosing appropriate buffering conditions. A buffer pH of 5 was chosen because it was of interest in the study of the antibody used but also because it is near the isoelectric point of BSA (4.7), thus minimizing the charge buildup on the surface that is due to the blocking protein. An engineered version27 of Protein G, Protein G′, with its BSA-binding domains removed, was used to minimize interactions between that probe and the blocking protein. BSA was also found to adsorb in small amounts to the antibody, as can be seen in Figure 4 where some BSA adheres to the 1 mg/mL IgG surface (at the multilayer coverage of over 1000 ng/cm2). In this experiment, we cannot eliminate the possibility that BSA displaces the antibody on the surface. On the basis of the work of Elgersma and colleagues,28 we believe this would lead to a displacement of at most 10% of the antibody, which would have the effect of decreasing the actual IgG surface concentrations by a similar amount (shifting the “monolayer” line to the right by 10% in Figures 7 and 8). This
surface. Coincidentally, this corresponds to the mass of a lipid bilayer formed on a SiO2-coated crystal as measured by QCM-D,23 which has a thickness around 5 nm. For the sake of analysis, we shall define the Sauerbrey mass of 468 nm/cm2 (1.98 molecules/ 100 nm2 for our antibody) as one monolayer. Calculations by Buijs et al., by comparison, predict a monolayer mass of 200 ng/cm2 for flat-on oriented antibodies, 370 ng/cm2 for the regular end-on orientation, and 550 ng/cm2 for end-on with contracted Fab fragments.6 Our experimental Sauerbrey mass of 468 ng/cm2 lies within the expected range, but because of the nature of the QCM-D measurement method, which gives a hydrated mass and is subject to viscoelastic losses, we do not expect this value to be quantitative. QCM-D’s main limitation in providing accurate determinations of protein surface coverages is that it measures the hydrated mass of the protein film, including the mass of water rigidly coupled to the protein.20,24 On a hydrophilic surface, this can lead to up to a 2-fold discrepancy compared to other methods that measure the ‘dry’ mass,24 although we expect the hydrophobic surface used in this study to lead to a more compact architecture. Another potential caveat of QCM-D is that the Sauerbrey data analysis method used here only applies in the case of uniform, rigid adlayers that are fully mechanically coupled to the surface. This can underestimate the mass present due to viscous losses in the film.25 If the dissipation values are very large and there is poor overlap between successive overtones, these deviations from Sauerbrey behavior can be used to construct a Voigt viscoelastic model of the film that yields information about its thickness, viscosity, and shear modulus. While the data encountered in this study lie mostly within what is typically accepted as the Sauerbrey regime (D < 1 × 10−6 and overlap between successive overtones in frequency shift), preliminary use of a Voigt viscoelastic model suggests that this ‘missing mass’ effect can be significant, especially when large amounts of protein are present on the crystal surface. Given these limitations, we cannot rely solely on the value of the monolayer mass as determined by QCM-D to determine the orientation of our antibodies on the surface. Not only is there an uncertainty associated with that measurement, but more importantly because doing so assumes that the antibodies are all uniformly oriented. In order to resolve this problem, we also analyze the evolution of the dissipation value more closely and employ a molecular probe method to determine more directly what the orientation of the antibodies may be. 4.2.2. Dissipation Analysis Yields Orientational Information. The dissipation value measured by QCM-D gives information about the energy lost during oscillation cycles of the crystal surface. If the dissipation values are large enough and if there is sufficiently little overlap in the frequency output measured at different resonances, viscoelastic models can be used to fit the data and calculate best-fit parameters for the film thickness, viscosity, and shear modulus. In this work, however, the data were in or close to the Sauerbrey regime, so viscoelastic modeling often led to poor fits or complications in the selection criteria for best fit. However small the dissipation values may be, it is still informative to analyze the evolution of the dissipation value as a function of the amount of antibody on the surface (linearly related to the frequency shift according to the Sauerbrey relation) to gain a qualitative understanding of the film structure. The evolution of the dissipation value as the antibody layer is built on the surface, shown in Figure 6, is consistent with a picture of adsorption where antibodies initially adsorb flat-on to 1772
dx.doi.org/10.1021/la203095p | Langmuir 2012, 28, 1765−1774
Langmuir
Article
At monolayer coverage, there is a decrease in slope in the amount of probe bound as a function of surface coverage. The transition corresponds physically to the point at which antibodies adsorbing in excess of a monolayer adsorb onto and block others on the surface, rendering their binding sites less accessible. If these additional antibodies entirely screened out those below, one would expect to see the amount of bound probe to level off completely. Surprisingly, this is not the case; instead there is still a slight rise in the amount of probe bound with additional antibody. Because of this increased binding and because the dissipation value for films in this region is larger, we propose that antibodies in excess of a monolayer form a secondary film or series of aggregates that are more diffuse and hydrated than the first monolayer, resulting in mechanical coupling to the surface that is not as strong and is more susceptible to penetration by other proteins. Because these multilayer films also appear resistant to buffer rinsing, at least on the 10-min time scale measured, they appear to be anchored to the first monolayer by fairly strong intermolecular forces.
difference does not impact our conclusions in this study given that the surface concentrations as measured by QCM-D are already expected to be inexact (see discussion in 4.2.1). 4.2.4. Probe Binding Analysis. Given the inherent limitations of QCM-D in providing a quantitative measure of adsorbed mass and the confounding effects of the blocking protein BSA, patterns in the amount of probe bound as a function of IgG surface coverage should be considered qualitative. Still, even these qualitative patterns yield much structural information about the antibody layer. Most importantly, the fact that both probes can be captured at similar IgG surface coverages means that we can immediately eliminate both Fab-up and Fab-down orientations, as these should not display accessible Protein G′ or streptavidin binding sites, respectively. Furthermore, since the bound amounts lie between the values for the oriented immobilization experiments, we can eliminate those as well (side-on and Fab-down). There are other significant qualitative trends in the data. At low coverage, there is strictly no SA binding observed, whereas the G′ binding is only slightly reduced. This observation is consistent with the proposed model where adsorption initially occurs flat-on at low IgG surface coverages. In this case, the SA binding site in the variable region would be close enough to the surface and/or adjacent to BSA blocking protein that the relatively large SA antigen would not be bound due to steric interference. Protein G′, which is smaller, would apparently still have space to bind in this configuration. Not until the maximum number of flat-on oriented antibodies is achieved and interstices are filled in with more vertically oriented IgG would capture of SA begin, once the Fab regions are no longer all in contact with the surface. The binding ratios for both probe molecules are less than the maximum of 2.0 predicted in solution. While the exact numbers are not expected to be quantitative, it is significant to note that the values for SA are in general half those found for G′. While this could be interpreted as indicating an orientation tending toward Fab-down with the Fc fragments on average more accessible than Fab, it seems more likely that this is simply because the smaller size of G′ makes it able to bind in crowded conditions where the larger SA would be sterically hindered. Other studies have explored antigen binding capacity as a function of antibody surface concentration. As mentioned previously, papers by Xu and Wang found the binding capacity of anti-β-hCG after a BSA-blocking step to decrease with increasing surface coverage,9,13 which they explain in terms of steric hindrance. These findings are qualitatively different from ours, where we observe an increase in the antigen binding capacity for SA in the same regime of IgG surface concentration. Because both antigen and antibody are different and because those studies were conducted on a hydrophilic as opposed to a hydrophobic surface, one does not a priori expect the results to agree with ours, but it is intriguing to find such opposite trends. Because their data span a relatively small window of antibody surface concentrations and do not include a secondary probe, it is difficult to draw conclusions about the relative orientations of the anti-β-hCG versus the antistreptavidin antibodies studied in this work. Another study by Chen and colleagues using a different anti-hCG antibody did not look at the dependence on surface coverage but does report a binding ratio of ∼0.33 at 220 ng/cm2 and concludes that this represents a flat-on orientation on a hydrophobic CH3-terminated surface.14 This would be consistent with our finding that flat-on adsorption is favored on the hydrophobic surface, and we report a very similar antigen binding capacity at that surface coverage.
5. CONCLUSION In this study, the orientation of an antibody on a hydrophobic surface was examined by monitoring its adsorption and that of its ligands via QCM-D. We found that the antigen streptavidin (SA), which binds the variable region in the Fab domain, and the Fc domain-binding Protein G′ can be used together as probes for antibody orientation. Preadsorbing G′ caused the antibody to preferentially adsorb side-on, while preadsorbing SA caused it to favor a Fab-down configuration. On a bare hydrophobic surface, antibody adsorption appears irreversible, with monolayer coverage at approximately 468 ng/cm2 (Sauerbrey mass). High bulk concentrations led to persistent multilayer coverage. Analyzing the evolution of the dissipation value with surface coverage below one monolayer shows a close mechanical coupling of the antibody layer at low coverage indicating a flat-on adsorption, with looser coupling at higher coverage indicating a switch to more end-on or side-on type adsorption. This interpretation is supported by the patterns in probe binding, which show reduced binding per IgG molecule at low surface coverage. For SA there was a threshold of ∼100 ng IgG/cm2 below which no detectable SA binding occurred, and then the amount bound became approximately linear with IgG coverage until a maximum per molecule value of 0.45 SA/IgG was attained, occurring just shy of a monolayer. Protein G′ binding was roughly linear with IgG coverage, rising to a maximum per molecule IgG of 0.95 G′/IgG at around 250 ng IgG/cm2. These data suggest that the adsorption initially occurs in a flat-on orientation when the antibody has ample room on the surface. Once the surface begins to be crowded, adsorption occurs in a mixture of vertical orientations, leading to looser mechanical coupling as well as the exposure of both antigen and G′ binding sites.
■
AUTHOR INFORMATION
Corresponding Author
*Tel: +1 650 723 4573. Fax: +1 650 723 9780. E-mail: Curt.
[email protected].
■
ACKNOWLEDGMENTS The anti-streptavidin antibody, as well as support, was graciously provided by Amgen. M.W. was supported in part by the NIH Biotechnology Training Grant (5-T32-GM008412-17). 1773
dx.doi.org/10.1021/la203095p | Langmuir 2012, 28, 1765−1774
Langmuir
■
Article
(20) Hook, F.; Kasemo, B.; Nylander, T.; Fant, C.; Sott, K.; Elwing, H. Variations in coupled water, viscoelastic properties, and film thickness of a Mefp-1 protein film during adsorption and cross-linking: A quartz crystal microbalance with dissipation monitoring, ellipsometry, and surface plasmon resonance study. Anal. Chem. 2001, 73 (24), 5796−5804. (21) Furusawa, H.; Komatsu, M.; Okahata, Y. In Situ Monitoring of Conformational Changes of and Peptide Bindings to Calmodulin on a 27 MHz Quartz-Crystal Microbalance. Anal. Chem. 2009, 81 (5), 1841−1847. (22) Lee, H. S.; Contarino, M.; Umashankara, M.; Schon, A.; Freire, E.; Smith, A. B.; Chaiken, I. M.; Penn, L. S. Use of the quartz crystal microbalance to monitor ligand-induced conformational rearrangements in HIV-1 envelope protein gp120. Anal. Bioanal. Chem. 2010, 396 (3), 1143−1152. (23) Keller, C. A.; Kasemo, B. Surface specific kinetics of lipid vesicle adsorption measured with a quartz crystal microbalance. Biophys. J. 1998, 75 (3), 1397−1402. (24) Hook, F.; Voros, J.; Rodahl, M.; Kurrat, R.; Boni, P.; Ramsden, J. J.; Textor, M.; Spencer, N. D.; Tengvall, P.; Gold, J.; Kasemo, B. A comparative study of protein adsorption on titanium oxide surfaces using in situ ellipsometry, optical waveguide lightmode spectroscopy, and quartz crystal microbalance/dissipation. Colloids Surf., B 2002, 24 (2), 155−170. (25) Voinova, M. V.; Jonson, M.; Kasemo, B. ’Missing mass’ effect in biosensor’s QCM applications. Biosens. Bioelectron. 2002, 17 (10), 835−841. (26) Chen, H. X.; Huang, J. Y.; Lee, J.; Hwang, S.; Koh, K. Surface plasmon resonance spectroscopic characterization of antibody orientation and activity on the calixarene monolayer. Sens. Actuators, B 2010, 147 (2), 548−553. (27) Goward, C. R.; Murphy, J. P.; Atkinson, T.; Barstow, D. A. Expression and Purification of a Truncated Recombinant Streptococcal Protein-G. Biochem. J. 1990, 267 (1), 171−177. (28) Elgersma, A. V.; Zsom, R. L. J.; Lyklema, J.; Norde, W. Adsorption Competition between Albumin and Monoclonal Immunogammaglobulins on Polystyrene Lattices. J. Colloid Interface Sci. 1992, 152 (2), 410−428.
REFERENCES
(1) Klonisch, T.; Panayotou, G.; Edwards, P.; Jackson, A. M.; Berger, P.; Delves, P. J.; Lund, T.; Roitt, I. M. Enhancement in antigen binding by a combination of synergy and antibody capture. Immunology 1996, 89 (2), 165−171. (2) Neubert, H.; Jacoby, E. S.; Bansal, S. S.; Iles, R. K.; Cowan, D. A.; Kicman, A. T. Enhanced affinity capture MALDI-TOF MS: Orientation of an immunoglobulin G using recombinant protein G. Anal. Chem. 2002, 74 (15), 3677−3683. (3) Kausaite-Minkstimiene, A.; Ramanaviciene, A.; Kirlyte, J.; Ramanavicius, A. Comparative Study of Random and Oriented Antibody Immobilization Techniques on the Binding Capacity of Immunosensor. Anal. Chem. 2010, 82 (15), 6401−6408. (4) Zhou, J.; Chen, S. F.; Jiang, S. Y. Orientation of adsorbed antibodies on charged surfaces by computer simulation based on a united-residue model. Langmuir 2003, 19 (8), 3472−3478. (5) Zhou, C.; Friedt, J. M.; Angelova, A.; Choi, K. H.; Laureyn, W.; Frederix, F.; Francis, L. A.; Campitelli, A.; Engelborghs, Y.; Borghs, G. Human immunoglobulin adsorption investigated by means of quartz crystal microbalance dissipation, atomic force microscopy, surface acoustic wave, and surface plasmon resonance techniques. Langmuir 2004, 20 (14), 5870−5878. (6) Buijs, J.; Lichtenbelt, J. W. T.; Norde, W.; Lyklema, J. Adsorption of Monoclonal IgGs and Their F(Ab′)(2) Fragments onto Polymeric Surfaces. Colloids Surf., B 1995, 5 (1−2), 11−23. (7) San Paulo, A.; Garcia, R. High-resolution imaging of antibodies by tapping-mode atomic force microscopy: Attractive and repulsive tip-sample interaction regimes. Biophys. J. 2000, 78 (3), 1599−1605. (8) Xu, H.; Zhao, X. B.; Grant, C.; Lu, J. R.; Williams, D. E.; Penfold, J. Orientation of a monoclonal antibody adsorbed at the solid/solution interface: A combined study using atomic force microscopy and neutron reflectivity. Langmuir 2006, 22 (14), 6313−6320. (9) Wang, X. Q.; Wang, Y. N.; Xu, H.; Shan, H. H.; Lub, J. R. Dynamic adsorption of monoclonal antibody layers on hydrophilic silica surface: A combined study by spectroscopic ellipsometry and AFM. J. Colloid Interface Sci. 2008, 323 (1), 18−25. (10) Xu, H.; Zhao, X. B.; Lu, J. R.; Williams, D. E. Relationship between the structural conformation of monoclonal antibody layers and antigen binding capacity. Biomacromolecules 2007, 8 (8), 2422−2428. (11) Petrash, S.; Cregger, T.; Zhao, B.; Pokidysheva, E.; Foster, M. D.; Brittain, W. J.; Sevastianov, V.; Majkrzak, C. F. Changes in protein adsorption on self-assembled monolayers with monolayer order: Comparison of human serum albumin and human gamma globulin. Langmuir 2001, 17 (24), 7645−7651. (12) Wang, H.; Castner, D. G.; Ratner, B. D.; Jiang, S. Y. Probing the orientation of surface-immobilized immunoglobulin G by time-offlight secondary ion mass spectrometry. Langmuir 2004, 20 (5), 1877−1887. (13) Xu, H.; Lu, J. R.; Williams, D. E. Effect of surface packing density of interfacially adsorbed monoclonal antibody on the binding of hormonal antigen human chorionic gonadotrophin. J. Phys. Chem. B 2006, 110 (4), 1907−1914. (14) Chen, S. F.; Liu, L. Y.; Zhou, J.; Jiang, S. Y. Controlling antibody orientation on charged self-assembled monolayers. Langmuir 2003, 19 (7), 2859−2864. (15) Yu, D. Q.; Ghosh, R. Method for Studying Immunoglobulin G Binding on Hydrophobic Surfaces. Langmuir 2010, 26 (2), 924−929. (16) Janshoff, A.; Galla, H. J.; Steinem, C. Piezoelectric mass-sensing devices as biosensors - An alternative to optical biosensors? Angew. Chem., Int. Ed. 2000, 39 (22), 4004−4032. (17) Methods and Protocols. In Q-Sense Methods; Q-Sense AB: Västra Frölunda, Sweden, 2007. (18) Bae, Y. M.; Oh, B. K.; Lee, W.; Lee, W. H.; Choi, J. W. Study on orientation of immunoglobulin G on protein G layer. Biosens. Bioelectron. 2005, 21 (1), 103−110. (19) Kato, K.; Lian, L. Y.; Barsukov, I. L.; Derrick, J. P.; Kim, H. H.; Tanaka, R.; Yoshino, A.; Shiraishi, M.; Shimada, I.; Arata, Y.; Roberts, G. C. K. Model for the Complex between Protein-G and an Antibody Fc Fragment in Solution. Structure 1995, 3 (1), 79−85. 1774
dx.doi.org/10.1021/la203095p | Langmuir 2012, 28, 1765−1774
