Article pubs.acs.org/Langmuir
Rational Targeting of Subclasses of Intermolecular Interactions: Elimination of Nonspecific Binding for Analyte Sensing Jordan S. Lane, Joanna L. Richens, Kelly-Ann Vere, and Paul O’Shea* Cell Biophysics Group, Institute of Biophysics, Imaging & Optical Science, School of Life Sciences, University of Nottingham, Nottingham, NG7 2RD United Kingdom ABSTRACT: The ability to target and control intermolecular interactions is crucial in the development of several different technologies. Here we offer a tool to rationally design liquid media systems that can modulate specific intermolecular interactions. This has broad implications in deciphering the nature of intermolecular forces in complex solutions and offers insight into the forces that govern both specific and nonspecific binding in a given system. Nonspecific binding still continues to be a problem when dealing with analyte detection across a range of different detection technologies. Here, we exemplify the problem of nonspecific binding on model membrane systems and when dealing with low-abundance protein detection on commercially available SPR technology. A range of different soluble reagents that target specific subclasses of intermolecular interactions have been tested and optimized to virtually eliminate nonspecific binding while leaving specific interactions unperturbed. Thiocyanate ions are used to target nonpolar interactions, and small reagents such as glycylglycylglycine are used to modulate the dielectric constant, which targets charge−charge and dipole interactions. We show that with rational design and careful modulation these reagents offer a step forward in dissecting the intermolecular forces that govern binding, alongside offering nonspecific binding elimination in detection systems.
■
INTRODUCTION Although all forces of relevance to biological molecules are electrical in nature, they are manifested in a variety of ways. Their ubiquity underlies the processes of biology with Coulombic interactions, dipole interactions, van der Waals forces, and dispersion forces combining to drive any given molecular interaction.1,2 Essential natural processes from cell−cell interactions1 to solvation3 and the stabilization of macromolecular structure1,2 such as protein folding (and misfolding) all have evolved in a biological context by exploiting these manifold molecular interactions. Targeting intermolecular interactions has been undertaken in a number of contexts for many years based on the knowledge of the nature of the interaction and range across the whole spectrum of biological and biomedical research. One of the cornerstones of molecular medicine, for example, seeks to design molecules that specifically target important binding reactions of therapeutic reagents (as agonists or antagonists). In other areas the saltingin/out of proteins as part of a purification protocol4 is an example of targeting particular forces (i.e., reducing or augmenting Coulombic interactions). Surprisingly, however, reports of targeting the many other types of intermolecular interactions in a systematic fashion are rare and tend to be developed for a bespoke process of interest. This shortcoming is addressed in the present paper as we seek to provide a rational basis for sets of conditions that allow the handling of biological molecules in which particular forces are reduced or eliminated altogether in order to assess their contribution to a particular interaction. A coarser approach was used previously with some success to assess © 2014 American Chemical Society
effects of reducing separately the Coulombic and hydrophobic phenomena that take place in cell−cell interactions during the course of cellular aggregation.1,3 In the present report we illustrate our approach by outlining conditions that target several types of intermolecular interaction and so allow an assessment of their importance for any given molecular interaction. Although at the level of a molecular force there is no essential difference between a specific and nonspecific interaction as this simply resides in an observer’s definition, as a practical illustration we demonstrate that it is possible to eliminate the problems of nonspecific binding reactions in a number of analyte assay systems. The approach could also be used in a similar fashion to assess dominant interactions in many processes such as protein folding and ligand−receptor interactions. We utilize two assay systems; the first, developed in our laboratories, is a fluorescence-based membrane binding assay, and the second is a label-free surface plasmon resonance (SPR) detection system. Both are sensitive tools for the identification and characterization of many types of binding interactions5 with many applications in basic and applied research (from environmental6 to clinical diagnostic settings7,8). A major benefit of plasmonic-based detection modalities is that they may be used in a label-free manner for the analytes, reducing both the length of the protocol time and the training level required by the users Received: April 29, 2014 Revised: June 26, 2014 Published: July 21, 2014 9457
dx.doi.org/10.1021/la5016548 | Langmuir 2014, 30, 9457−9465
Langmuir
Article
Table 1. Buffer Compositions Used During the Protein Interaction Study With Their Experimentally determined Bmax and Kd Valuesa fibrinogen
albumin
a
buffer ID
NaCl (mM)
NaSCN (mM)
arginine (mM)
gly-gly (mM)
gly-gly-gly (mM)
Bmax
Kd
Bmax
Kd
PBS NSBB1 NSBB2 NSBB3 NSBB4 NSBB5 NSBB6 NSBB7 NSBB8 NSBB9 NSBB10 NSBB11
137 87 87 87 50 37 0 0 0 0 0 0
0 0 0 0 87 0 87 87 87 50 100 100
0 100 0 0 0 50 100 0 0 87 200 200
0 0 100 0 0 50 0 100 0 100 200 0
0 0 0 100 0 0 0 0 100 0 0 200
30.45 ± 1.65 3.26 ± 0.37 26.24 ± 2.24 47.71 ± 2.24 8.96 ± 2.51 12.02 ± 0.23 1.24 ± 0.23 3.86 ± 1.17 2.55 ± 0.64 57.18 ± 16.79 1.63 ± 0.54 2.13 ± 0.42
19 ± 2.09 0.58 ± 1.29 13.39 ± 2.79 10.43 ± 2.84 16.45 ± 10.25 18.75 ± 1.77 1.00 ± 1.77 7.76 ± 3.17 2.63 ± 0.44 300.1 ± 94.93 6.83 ± 0.49 3.27 ± 0.50
58.1 ± 13.13 2.59 ± 1.02 1.52 ± 0.50 0.88 ± 0.15 1.07 ± 1.13 10.36 ± 1.12 1.74 ± 0.75 3.53 ± 1.64 1.58 ± 0.22 8.98 ± 1.44 2.55 ± 0.68 1.14 ± 0.68
89.81 ± 23.98 3.63 ± 5.32 1.41 ± 3.15 1.24 ± 0.41 8.29 ± 0.46 26.49 ± 4.62 8.71 ± 4.49 19.41 ± 3.43 1.17 ± 0.31 36.7 ± 8.39 3.19 ± 3.29 1.97 ± 0.32
All buffers also contained 2.7 mM KCl, 8.1 mM Na2HPO4, and 1.47 mM KH2PO4 and were adjusted to pH 7.4.
molecular recognition reactions upon which the sensor relies allows the virtual elimination of nonspecific binding. These experimental conditions represent, therefore, practical solutions that significantly enhance the specificity of label-free molecular detection, leading to an improved detection limit. The broader implications of this work are that it will facilitate a systematic approach to dissecting the dominant interactions that underlie processes such as protein folding and ligand−receptor interactions, protein−protein associations such as those found in amyloid formation, and other undesirable protein-folding processes.20
when compared to those of alternative detection methods such as ELISA. Additionally, the label-free nature of plasmonic-based biosensor (PBB) techniques provides the scope for the examination of virtually any species of biomolecule including proteins,9 antibodies,10 single nucleotide polymorphisms,11 sugars,12 narcotics,13 peptides,14 small molecules,15 and microRNAs16 given that appropriate capture molecules and surface chemistry for ligand attachment are available. For our purposes the label-free nature of the detection protocols of PBB systems allows us to probe various types of intermolecular interactions and to validate a rational basis for preparing solute systems that reduce or augment particular types of interactions. We demonstrate these capabilities by defining media that allow the total or partial elimination of nonspecific binding problems that compromise many types of analyte sensing systems.17 Utilizing detection systems that measure the addition of mass to a surface requires care to ensure accurate target detection and quantification. The major difficulty encountered arises from nonspecific molecular interactions of nontarget molecules that may be present in the sample under investigation or target molecules that bind to a surface that is not equipped with the analyte-recognition moiety. These phenomena are particularly problematic in diagnostic applications where biomarkers are sought in complex bodily fluids that contain many different macromolecules exhibiting many different types of possible intermolecular interactions.18,19 This results in a reduction of sensitivity and contamination of the sensing surfaces, compromising measurements and often also reducing the useful lifetime of the sensor. To overcome this it is necessary to develop procedures that target these nonspecific interactions while not compromising the desirable (specific) interactions associated with the molecular target. We have previously identified media that prevent the nonspecific binding of some proteins, in particular, serum albumin.17 In many cases19 the reagents we recommended were found to be highly effective, but we also found that some other proteins are less easily manipulated.19 The reason for this is most likely that other nonspecific interactions cumulatively dominate the interaction profile of the proteins. In the present study, therefore, we demonstrate that the selective inclusion of additional reagents in media designed to target noncovalent macromolecular interactions without reducing the specific
■
METHODS
Buffer Compositions. Buffers were based upon PBS and altered to include varying levels of thiocyanate, arginine, glycylglycine (gly-gly), or glycylglycylglycine (gly-gly-gly), as illustrated in Table 1. In addition to the chemicals detailed in Table 1 all buffers contained 2.7 mM KCl, 8.1 mM Na2HPO4, and 1.47 mM KH2PO4. Buffers were all adjusted to pH 7.4. Membrane Preparation and Labeling with Fluoresceinphosphatidylethanolamine (FPE). Egg phosphatidylcholine (PC) was supplied by Lipid Products (U.K.). Membrane vesicles were prepared as previously described.21 Briefly, PC dissolved in chloroform in a roundbottomed flask was dried under a stream of oxygen-free nitrogen gas until a thin film was formed. The lipid film was rehydrated with the appropriate buffer (pH 7.4) from Table 1. The resulting multilamellar solution was frozen and thawed 5 times and finally extruded 10 times through 25-mm-diameter polycarbonate filters with pores 0.1 μm in diameter (Nucleopore Corp.) using an extruder (Lipex Biomembranes Inc.) according to previously described extrusion procedures.22 This resulted in a monodisperse, unilamellar suspension of phospholipid vesicles. FPE was synthesized and used to label the outer bilayer leaflet of the phospholipid vesicles as previously described.23 Briefly, the phospholipid vesicles were incubated with ethanolic-FPE (never more than 0.1% ethanol in the total aqueous volume) at 37 °C for 1 h in the dark. Unincorporated FPE was removed by gel filtration on a PD10 Sephadex column. This procedure leads to the incorporation of approximately 70% of the externally added FPE to the preformed membrane vesicles. Fluorescence Measurements. Fluorescence spectroscopy was conducted on a Fluoromax-4 spectrofluorimeter (HORIBA Jobin Yvon) with emission and excitation wavelengths set at 520 and 490 nm, respectively. Fluorescence changes versus time were recorded following the additions of controlled amounts of either human serum albumin or fibrinogen to suspensions of FPE-labeled phospholipid vesicles (400 μM lipid). Liposomes made and suspended in each of the nonspecific 9458
dx.doi.org/10.1021/la5016548 | Langmuir 2014, 30, 9457−9465
Langmuir
Article
binding buffers (NSBBs) detailed in Table 1 were examined, and appropriate controls whereby protein additions were replaced with buffer additions were undertaken. Data Analysis of Lipid Binding Graphs. Bmax and Kd values were established by a nonlinear regression fit. The differentiation of Kd and Bmax action was determined by confidence levels determined by the oneway ANOVA Fisher’s LSD test against the PBS control and then grouped into categories. Surface Plasmon Resonance (SPR) Measurements. Antibody Attachment. A Reichert Life Sciences (USA) SR7000DC SPR spectrometer with the SPR Autolink data acquisition program was used for all SPR experiments. All experiments were undertaken at 25 °C. A baseline was calibrated by allowing degassed PBS-Tween (PSBT) to flow over an SR7000 mixed self-assembled monolayer gold sensor slide composed of 10% COOH-(PEG)6-C11-SH and 90% OH-(PEG)3C11-SH (Reichert, USA). The slide was activated with 1-ethyl-3-(3(dimethylamino)propyl)carbodiimide (EDC, 0.2 M) and N-hydroxy succinimide (NHS, 0.1 M) followed by the attachment of either an antifibrinogen antibody (Abcam, ab10066), an anti-IL8 antibody (R&D Systems, MAB208), or an IgG isotype control antibody (Genetex, GTX35009) to one channel of the system. No antibody was bound to the second “reference” channel. The remaining active sites in both channels were then capped using ethanolamine (1 M) to prevent the immobilization of protein on the slide surface during subsequent binding reactions. Binding Interactions. Stable baseline readings were established with the appropriate running buffer prior to binding reactions. Solutions of albumin (50 mg/mL) or fibrinogen (5 μg/mL for specific binding reactions, 5−500 μg/mL for nonspecific reactions) diluted in the corresponding running buffer were passed over the immobilized antibody and reference channel for 5 min (20 μL/min) prior to a 5 min buffer dissociation period. Surfaces were regenerated using 5 mM HCl (100 μL/min for 2 min), and a buffer baseline reading was reestablished between each binding reaction. Changes in refractive index units (RIU) due to protein binding were recorded for each of the conditions assessed.
Figure 1. (A) Changes in the fluorescence profile of FPE following albumin (HSA; 3 μM) interaction with the membrane surface in different buffer systems. Binding profiles of (B) albumin and (C) fibrinogen to FPE-labeled PC100% liposomes in a variety of different NSBBs. The lipid concentration was 400 μM, and experiments were undertaken at 25 °C. Data represented are mean ± SD for three repeat experiments.
■
RESULTS Protein Binding to Lipid Membrane Systems: Effects of Force-Modifying Solutes on the Binding Extent and Affinity. Studies of the effects of media designed to modify particular types of molecular interactions associated with protein systems that interact with membranes were performed with model lipid systems. Thus, serum proteins were used to challenge FPE-labeled monodisperse, unilamellar, phosphatidylcholine vesicles to determine their binding characteristics in different media designed to target different types of molecular interactions. FPE was used as this is a well-established and reliable fluorescent indicator of molecular interactions with membranes.21,23,24 The fluorescence yield of FPE is modified by the surface electrostatic potential of the membrane, and the binding of molecules that possess a positive or negative charge leads to signal changes that can be used to determine the extent (referred to as the binding capacity, Bmax, in arbitrary units, a.u.) and affinity (Kd) as well as the kinetics of a binding reaction. The addition of CaCl2 to FPE-labeled phospholipid membranes, for example, leads to an increase in the observed fluorescence as Ca2+ becomes adsorbed onto the membrane surface and represents the addition of positive charge25 while the binding of negatively charged albumin molecules leads to a decrease in signal (Figure 1a). These observations are explained based on the wellcharacterized mode of action of the FPE probe21 and are a consequence of the decrease in electronegative (or increase in electropositive) surface potential upon membrane binding of the respective molecular species. Successive additions of protein lead
to a stepwise change in the fluorescence as a consequence of the protein binding to the membrane surface (Figure 1a). Figure 1b,c indicates the membrane binding properties of albumin and fibrinogen in PBS to the membrane surface supplemented with different active species anticipated to affect specific molecular interactions. Thus, these reagents can be used in conjunction with binding studies to explore the magnitude that each intermolecular force may have on the overall interaction of the protein with the membrane. The formulations were made by replacing 137 mM NaCl with reagents that had the same total ionic charge; the reagents were then taken to high ionic concentration to determine the effect of the concentration dependence. The various reagents identified in Table 1 were tested in the manner shown in Figure 1 with the effects of each medium on the measured values of Bmax and Kd of each protein membrane interaction tabulated in Table 1. We have shown previously that sodium thiocyanate is effective at preventing the nonspecific binding of albumin and several other proteins to membrane surfaces but was much less effective with other proteins including fibrinogen.17 Figure 1c outlines some results of the membrane-binding characteristics of fibrinogen. It is shown in Figure 1c that studies in the presence of gly-gly led to a lower membrane affinity for fibrinogen as compared to identical studies made in PBS. This is indicated by the relative changes in both the Bmax and Kd parameters, reduced 9459
dx.doi.org/10.1021/la5016548 | Langmuir 2014, 30, 9457−9465
Langmuir
Article
from 58.1± 13.13 and 89.81 ± 23.98 a.u. in PBS to 1.518 ± 0.50 and 1.414 ± 3.15 a.u., respectively. Thus, both the membranebinding affinity and the extent of binding are affected by the presence of gly-gly. However, no similar effects were found to take place for the membrane interactions of albumin. In contrast, the presence of arginine was shown to be effective against both albumin and fibrinogen to similar extents (Table 1, NSBB1). The use of molecules such as gly-gly and arginine were employed to manipulate the ambient dielectric constant, and although many interactions are affected by this parameter the extent of these effects varies greatly. However such treatments may be used selectively to target interactions that are dominated by such phenomena with the same rationale as salting-in or salting-out during protein preparation (or other colloids). Glygly-gly has a higher positive dielectric constant increment (+128) than gly-gly (+71) and arginine (+62) as shown in Table 2.26 The dielectric increment values are taken from a long-standing calculation created from observing the dielectric constant of a polar solvent (water unless stated) with the addition of smallmolecule reagent. In the cases of amino acids the dielectric increment is linear with concentration and increases with the dipole moment of the solute. It is also noteworthy that the change in the dielectric constant is relatively independent of solvent and temperature.26 Gly-gly-gly was observed to modify the binding parameters of fibrinogen as shown in Figure 1c. Glygly-gly further decreased the Kd and Bmax of fibrinogen binding (Bmax = 0.883 ± 0.15, Kd = 1.243 ± 0.41), effectively eliminating the nonspecific contributions to the observed fibrinogen binding measurements. Gly-gly-gly and thiocyanate together were found to eliminate completely the binding of both proteins to the membrane (Figure 1a; Table 1, NSBB). These data together suggest that the intermolecular interactions that dominate the membrane binding of albumin and fibrinogen are different. Such a statement is perhaps not that surprising as they are very different proteins. The point we wish to emphasize, however, is that our data illustrates that we have the experimental means to determine the relative extent and magnitude of each of the intermolecular forces and their respective influences on macromolecular interactions, in this case, membrane binding. Studies of Nonspecific Binding: Interactions of Albumin with Antibodies and Material Interfaces. The reagent combinations outlined in the previous section that were established to affect protein−membrane interactions were studied with commercially available plasmonic-based measurement systems, in other words, by utilizing plasmonic-based analyte sensing systems known to be compromised from nonspecific binding problems. We seek to demonstrate that by targeting particular molecular interactions we may eliminate the nonspecific binding properties of prospective analytes. The methodology is outlined as follows: Figure 2a indicates a nonspecific binding reaction of albumin over an anti-IL-8 antibody. Albumin solutions and the nonspecific binding buffers exhibit refractive indexes that are greater than those of the media, hence during incubation the time points on any binding curve are hidden. When the protein is washed away, however, a clear change in the baseline compared to the PBS solution was observed. An SR7000 mixed self-assembled monolayer gold sensor slide (composed of 10% COOH-(PEG)6-C11-SH and 90% OH-(PEG)3-C11-SH) was activated by EDC and NHS, the reference channel was bypassed, a monoclonal IL-8 antibody at 5 μg/mL was conjugated to the specific channel, and the whole
Table 2. Dielectric Constant Modulator Molecules with Their Associated Dielectric Incrementsa reagent
dielectric increment Molecules
NaCl EtOH glycerol mannitol sucrose phenol urea
0 −2.6 −2.6 −2.6 −7.6 −6.6 +2.7 Amino Acids −10 +22.6 +23.4 +25 +25 +21 +26 +27.8 +20.8 +24.4 +30 +30 +32.2 +37.3 +34.2 +41 +51 +62
glycine anhydride glycine α-alanine L/D-valine L-leucine L/D-proline D-glutamic acid L-aspartic acid D-glutamine L-asparagine N-phenylglycine glycocyamine creatine β-alanine β-aminobutyric acid taurine ornithine arginine Peptides diglycine glycylalanine alanylglycine leucylglycine glycylleucine glycylphenylalanine phenylalanylglycine triglycine leucylglycylglycine tetraglycine pentaglycine hexaglycine heptaglycine
+71 +71.8 +71 +62.5 +72.3 +67 +70.4 +56.7 +128 +159 +215 +234 +290 (in 5.14 molar urea) Double Dipoles
lysylglutamic acid cystinyldiglycine cystinyldidiglycine diglycylcystine
+345 +139 +250 +139
a
This list allows for the categorization of currently known molecules that affect dielectric constants, for further quantification or consideration if molecules are currently used in formulations.
slide surface was blocked by ethanolamine. An IL-8 specific binding profile was then established. The slide and protocol have previously been optimized to minimize nonspecific binding using conventional methods.27−30 Figure 2a indicates that the nonspecific binding of albumin is significantly greater than the IL-8 specific binding measurement, showing that current nonspecific blocking measures are inadequate. 9460
dx.doi.org/10.1021/la5016548 | Langmuir 2014, 30, 9457−9465
Langmuir
Article
Figure 2. (A) Nonspecific binding of albumin at physiological concentration (50 mg/mL) in PBS and specific binding of IL-8 above physiological concentration (10 ng/mL) to the IL-8 antibody-conjugated surface in PBS. The nonspecific binding of albumin is abolished in the presence of NSBB8. Using NSBB8 as a running buffer when a surface has been pretreated and incubated in 50 mg/mL albumin for 30 min, IL-8 NSBB8 shows that a specific IL-8 binding curve can still be reproduced, showing no signs of nonspecific retardation. Experiments were undertaken at 25 °C. (B) Albumin interacts nonspecifically with both the antibody-conjugated (solid) and reference (dashed) channels of an SPR sensor chip. Data are analyzed by deviation from the initial baseline after protein incubation, and the data represents the mean ±SD for three repeat experiments. (C) Altering the chemical composition of the running and sample buffers reduces the nonspecific binding of albumin to the IgG isotype antibody-conjugated (solid) and reference (dashed) channels of an SPR sensor chip.
then validated using NSBB8 as a running buffer, albumin was incubated on the surface under the same conditions as for the nonspecific binding experiments, and an IL-8 sample in NSBB8 was then tested for its specific binding capability. The specific binding level was consistent with a sample of IL-8 PBS binding without the presence of albumin. Within Figure 2a, it is evident that albumin NSBB8 has a higher refractive index than that of albumin PBS; all of the nonspecific binding buffers increase the refractive index by virtue of increased solute concentration. This result showcases the fact that the buffers can negate nonspecific binding while still retaining specific binding where nonspecific fouling would have previously dominated or contributed to false positive effects in label-free systems. To further test the efficacy of the binding buffers, similar studies with the nonspecific binding of fibrinogen were also performed at a series of different concentrations, i.e., 5−500 μg/ mL, and measurements at the boundary values are shown in Figure 3a. The figure indicates that fibrinogen binds nonspecifically to both the isotype antibody ((solid) binding channel) and to the SPR surface ((dashed) reference channel). Figure 3a shows a range of different reagents employed to reduce the nonspecific binding. It is noteworthy that the nonspecific reduction has a differential profile with respect to that of the albumin nonspecific binding; however, this is unsurprising when
Figure 2b shows that in plasmonic-sensing systems albumin binds nonspecifically to a range of different antibodies as well as the SPR surface itself. This is demonstrated by the increased binding in both the “specific” antibody channel (solid) and the SPR surface in the reference channel (dashed). As the signal change is typically greater in the binding channel than in the reference channel (e.g. Figure 2c), this means that the protein is actively binding the core antibody domains as well as circumventing the antibodies and sticking to the surface. Increased nonspecific binding was shown in the presence of both an IL-8-specific antibody and an IgG isotype control antibody, which agrees with previous results.17 Within the IL-8 antibody nonspecific binding study it is possible that there may be an unknown contribution of crossreactivity, hence all further tests were carried out with albumin nonspecific binding and the IgG isotype control antibody, which is devoid of specific binding regions. A range of different reagents were employed to abolish the nonspecific binding present (Figure 2c). Of these, it is shown in Figure 2a that NSBB8 completely eliminates all detectable nonspecific binding of albumin. Under these circumstances the sensitivity of a specific measurement can be much enhanced, provided that checks were made to ensure that NSBB8 does not have a similar effect on the specific binding of the target analyte. The sensing system was 9461
dx.doi.org/10.1021/la5016548 | Langmuir 2014, 30, 9457−9465
Langmuir
Article
Figure 3. (A) Nonspecific binding of fibrinogen to the isotype antibody over a range of concentrations (5−500 μg/mL) in the specific (solid) and reference (dashed) channels. Buffer conditions were then altered to attenuate the nonspecific binding. (B) Exemplar trace of fibrinogen nonspecific binding in PBST and fibrinogen nonspecific binding in NSBB8.
Figure 4. Specific binding of fibrinogen (5 μg/mL) to an anti-fibrinogen antibody in different buffers. Binding was repeated eight times in PBST (A), NSBB8 (B), and NSBB11 (C), with antibody regeneration of 0.2 M HCl for 30 s between each repeat. (D) Comparison of binding in different buffer media, where the data shown is the mean data collated from seven repeats and the initial binding curves are excluded from averages based on manufacturer guidelines. Dashed lines represent ± standard deviation.
The fibrinogen protein does not affect the refractive index to the same extent as does albumin or the NSBBs; however, the nonspecific binding is still completely eliminated in the same fashion as for albumin.
comparing the proteins where it is obvious that different binding forces will dominate their mode interaction profile. Figure 3b illustrates that NSBB8 is also effective in analogous studies of fibrinogen-mediated nonspecific binding phenomena. 9462
dx.doi.org/10.1021/la5016548 | Langmuir 2014, 30, 9457−9465
Langmuir
Article
Table 3. The Relative Effect of Buffers on Bmax and Kd in Relation to Control PBSa
Effects of Antinonspecific Binding Reagents on Specific Binding Reactions. We indicated in the Introduction section of this paper that the physical parameters that underlie specific and nonspecific interactions arise simply from their subjective definition as they are not objectively different interactions. Thus, conditions established that interfere with specific interactions may also compromise the same interaction under the other subjective classification. This is not always the case, however (see Discussion), but for their practical exploitation, these possibilities need to be addressed. We illustrate these considerations by establishing binding profiles for fibrinogen (5 μg/mL) in PBS (Figure 4a) and two of our reagent combinations (NSBB8, NSBB11; Table 1), where each reagent combination was chosen due to its previous efficacy in blocking nonspecific interactions in the model membrane system, with representative low (NSBB8) and high (NSBB11) ionic strength solutions. The level of specific binding is shown to decrease slightly in the different buffers employed. Figure 4b,c shows that the binding kinetics of the protein is unaffected; hence there are no relative effects on the binding reaction. Figure 4d shows a slight decrease in signal in the presence of both reagent combinations. This is expected due to the removal of the nonspecific contribution to the binding curve. The decrease in NSBB11 is perhaps more pronounced than expected, which suggests that there are diminishing returns with further increments of concentration. The broad conclusions are outlined in the next section with some further aspects developed in the Discussion section. Summary of the Data. The various reagents used to attenuate the effects of nonspecific binding were selected to target different intermolecular forces. The ability of these reagents to alleviate nonspecific binding phenomena, however, is also dependent on the properties of each individual protein. We previously introduced the use of thiocyanate to diminish nonspecific binding by virtue of its large entropy of hydration.20 This has an effect of modulating the water molecules around nonpolar regions of proteins and reducing binding. This is reflected in the reduction of the experimental Bmax value but with a less pronounced impact on the affinity (i.e., Kd in Table 1 for NSBB4). The action of thiocyanate can therefore be understood as a reduction in nonpolar interactions. The dielectric constant modulators (Table 2) increase the dielectric constant (D), which theoretically29 decreases the charge−charge and dipole interactions by a factor of 1/D.31 This has been experimentally corroborated by having an independent effect on the binding affinity and overall binding of a different protein than thiocyanate action. Table 3 demonstrates the differential effects of the different reagents present.
effect on Bmax or Kd fibrinogen
albumin buffer
molecule employed
Bmax
NSBB1 NSBB2 NSBB3 NSBB4 NSBB8 NSBB11
arginine gly-gly gly-gly-gly thiocyanate gly-gly-gly + thiocyanate arginine + gly-gly-gly + thiocyanate
↓↓↓↓
↓↓↓ ↓↓↓↓ ↓↓↓
Kd
Bmax
Kd
↓
↓↓ ↓↓ ↓↓ ↓↓ ↓↓
↓ ↓ ↓ ↓ ↓
a
For the statistical analysis, a one-way ANOVA analysis was carried out, followed by a Fisher’s LSD test, and grouped by statistical power to determine relative effects. (↓) corresponds to 99% confidence, (↓↓) 99.5%, (↓↓↓) 99.9%, and (↓↓↓↓) 99.99%.
trehalose are also known to impart a level of stability for protein preparations perhaps based on its dipolar properties.34,35 The dipolar properties of trehalose, sucrose, and glucose all use water stabilization to prevent ice formation within proteins and hence act as a cryoprotectant.36 Both sucrose and trehalose have been shown to modulate the properties of water associated with bilayers through different mechanisms.37 On the atomistic level, we have emphasized that binding reactions are not really specific or nonspecific, thus interference requires a balance of probabilities in order to select a preferred outcome. In complex protein solutions, such as blood plasma, cerebrospinal fluid, and so forth, each protein will have a unique combination of molecular interactions alongside other determinants such as molecular crowding that cannot be ignored.2,38−40 The main determinants in protein solubility, stability, and viscosity have been identified as charge−charge and electrostatic dipole interactions.31,38 This is unsurprising due to the relative strengths of these interactions, coupled with their impact over distance. Increasing the dielectric constant, by virtue of the molecules shown in Table 2, reduces the proximity energies by 1/D. If the dielectric constant is increased sufficiently, then the only attractive force that remains is dispersion, which can be overcome by detergents.31 In this work, we chose a range of inert dielectric constant modulators (arginine, gly-gly, and gly-gly-gly) to increase the dielectric constant. The impact of these reagents had quite different effects compared to that of thiocyanate (Figure 1b,c). These reagents are also effective in conjunction with one another (Figure 1b,c), as exemplified by the reduction of Bmax, i.e., the relative reduction in the total interaction and an increase in affinity, i.e., as a reduction in Kd (Table 1). The variation between proteins is vast; however, all proteins have the same force determinants. The two exemplar proteins shown here were chosen due to their abundance in blood serum, coupled with their differential force profiles. Thiocyanate has been shown to be greatly effective against albumin,17 however, when further interrogated across the proteome it was seen to be less effective19 for proteins such as fibrinogen. In this paper we have shown that fibrinogen nonspecific binding is effectively attenuated by increasing the dielectric constant. As the dielectric constant impacts charge−charge interactions, it is reasonable to state that these forces dominate its binding interaction profile. Other proteins, such as the aforementioned albumin, may have nonpolar or hydrophilic forces that dominate. Thiocyanate is thought to modulate the surface water of the proteins due to its high entropy of hydration, which will then create an energy barrier which must be overcome for interaction to occur. Specific
■
DISCUSSION In the present paper, we have attempted to demonstrate that by selectively targeting the different intermolecular forces that occur between molecules we may gain some measure of control in reducing undesirable interactions. Some of the reagents we have identified have effects on the dielectric constant (Table 2) while others target other factors. In our hands, agents that have a more dramatic effect on the dielectric constant have been shown to be of use in affecting the protein−surface interactions quite significantly (Table 1). Interestingly, reducing the ambient dielectric constant has been employed in the past (perhaps even unwittingly) through the use of sucrose (−7.6 D) to increase intermolecular forces to form a stable protein matrix that is resilient during −80 °C freezing cycles.32−34 Others such as 9463
dx.doi.org/10.1021/la5016548 | Langmuir 2014, 30, 9457−9465
Langmuir
Article
buffer systems may well aid researchers in dissecting out the most influential parameters in their quests to understand this most important of biological phenomena.
interaction is normally orders of magnitude stronger than nonspecific interactions and hence would not be diminished.2 With a combination of both dielectric constant modulation and thiocyanate, the forces that account for nonspecific binding will be attenuated across the proteome.31 We have shown that with careful implementation of the different reagents a medium can be created that selectively modulates each of the aforementioned interactions in order to eliminate some interactions without compromising the desirable, i.e., specific interactions. Thus, a combination of thiocyanate and glycylglycylglycine at particular concentrations, acting together, appear to represent an optimal cocktail that reduces undesirable interactions while not affecting the desirable interactions (Table 1; NSBB8, Figures 2a and 4b). The desirable specific interactions in analyte detection systems are more resilient than nonspecific binding interactions. This is due to the well-characterized mode of antibody binding taking place in solvent-free pockets.23,41 This means that the specific antibody interactions could be much less perturbed by the antinonspecific binding reagents. The assembly of suitable media for use in a given assay requires a number of considerations. This is evident with NSBB11; the reagent concentrations were increased to test the effect of ionic strength and active reagent concentration. When employed in studies of fibrinogen−membrane interactions, the results were promising (Figure 1b,c). When employed in the SPR system to target nonspecific fibrinogen binding, however, the buffer was shown to increase surface and antibody binding with a significant apparent decrease in specific fibrinogen binding (Figure 4c). Thus, the elevated ionic strength appears to promote nonspecific binding while the ionic strength in the other buffers is minimized by using media in which the ionic contribution of the specified active reagent is replaced by exchange with another dominant anion, i.e., Cl−. Thus, the utilization of either of these species alone or in conjunction creates novel systems that can be utilized and customized in order to target specific intermolecular forces. Several technologies are limited by nonspecific binding; this has been shown not only in our work20 but also across a range of different assay technologies, such as Luminex,42 ELISA,43 and SPR.44 PBB technologies typically use sensor chips optimized to limit as best as possible nonspecific binding.27−30 These sensor chips, however, still have significant limitations due to the PEGylation used on the surface being susceptible to thermal and oxidative degradation;45−48 over time this results in decreased performance in preventing nonspecific adsorption.7 In our hands a commercial SPR system exhibits some nonspecific detection limits in accord with published reports.49,50 Our buffer systems, however, allow the virtual elimination of nonspecific binding across different systems and different sensor chips within SPR, opening up new opportunities for improving detection limits. The use of a preventative buffer system also has advantages over expensive pretreatment steps. The use of pretreatment steps, such as albumin or plasma reduction kits, may introduce a number of additional side effects51,52 and is undesirable. A rationally designed and tested buffer system offers an exciting step forward in the utilization of modern tools by effective targeting and modulation of specific intermolecular forces, creating a customizable system that is able to eliminate nonspecific binding in a range of systems and across different forces that can be effective against the majority of the proteome, along with aptamers. Similarly for identifying dominant molecular interactions within protein−protein interactions or intramolecular parameters associated with protein folding, our
■
AUTHOR INFORMATION
Corresponding Author
*E-mail:
[email protected]. Tel: +44(0) 1159513209. Fax: +44(0)1158466580. Notes
The authors declare no competing financial interest.
■
REFERENCES
(1) Jones, L.; Oshea, P. The Electrostatic Nature of the Cell-Surface of Candida-Albicans - A Role in Adhesion. Exp. Mycol. 1994, 18, 111−120. (2) Israelachvili, J. N. Intermolecular and Surface Forces, 3rd ed.; Academic Press: San Diego, CA, 2011; pp 1−674. (3) Hobden, C.; Teevan, C.; Jones, L.; Oshea, P. Hydrophobic Properties of the Cell- Surface of Candidia Albicans - A Role in Aggregation. Microbiology 1995, 141, 1875−1881. (4) Maurer, R. W.; Sandler, S. I.; Lenhoff, A. M. Salting-in Characteristics of Globular Proteins. Biophys. Chem. 2011, 156, 72−78. (5) Mullett, W. M.; Lai, E. P.; Yeung, J. M. Surface Plasmon ResonanceBased Immunoassays. Methods 2000, 22, 77−91. (6) Enrico, D.; Manera, M. G.; Montagna, G.; Cimaglia, F.; Chiesa, M.; Poltronieri, P.; Santino, A.; Rella, R. SPR Based Immunosensor for Detection of Legionella Pneumophila in Water Samples. Opt. Commun. 2013, 294, 420−426. (7) Richens, J. L.; Urbanowicz, R. A.; Lunt, E. A.; Metcalf, R.; Corne, J.; Fairclough, L.; O’Shea, P. Systems Biology Coupled With Label-Free High-Throughput Detection as a Novel Approach for Diagnosis of Chronic Obstructive Pulmonary Disease. Respir. Res. 2009, 10, 29. (8) Sciacca, B.; Francois, A.; Klingler-Hoffmann, M.; Brazzatti, J.; Penno, M.; Hoffmann, P.; Monro, T. M. Radiative-Surface Plasmon Resonance for the Detection of Apolipoprotein E in Medical Diagnostics Applications. Nanomed.: Nanotechnol., Biol., Med. 2013, 9, 550−557. (9) Ladd, J.; Taylor, A. D.; Piliarik, M.; Homola, J.; Jiang, S. Label-Free Detection of Cancer Biomarker Candidates Using Surface Plasmon Resonance Imaging. Anal. Bioanal. Chem. 2009, 393, 1157−1163. (10) de Boer, A. R.; Hokke, C. H.; Deelder, A. M.; Wuhrer, M. Serum Antibody Screening by Surface Plasmon Resonance Using a Natural Glycan Microarray. Glycoconj. J. 2008, 25, 75−84. (11) Feriotto, G.; Ferlini, A.; Ravani, A.; Calzolari, E.; Mischiati, C.; Bianchi, N.; Gambari, R. Biosensor Technology for Real-Time Detection of the Cystic Fibrosis W1282X Mutation in CFTR. Hum. Mutat. 2001, 18, 70−81. (12) Suda, Y.; Arano, A.; Fukui, Y.; Koshida, S.; Wakao, M.; Nishimura, T.; Kusumoto, S.; Sobel, M. Immobilization and Clustering of Structurally Defined Oligosaccharides for Sugar Chips: an Improved Method for Surface Plasmon Resonance Analysis of ProteinCarbohydrate Interactions. Bioconjugate Chem. 2006, 17, 1125−1135. (13) Klenkar, G.; Liedberg, B. A Microarray Chip for Label-Free Detection of Narcotics. Anal. Bioanal. Chem. 2008, 391, 1679−1688. (14) Lung, F. D.; Chen, C. H.; Liou, C. C.; Chen, H. Y. Surface Plasmon Resonance Detection of Interactions Between Peptide Fragments of N-Telopeptide and its Monoclonal Antibodies. J. Pept. Res. 2004, 63, 365−370. (15) Fu, E.; Chinowsky, T.; Nelson, K.; Johnston, K.; Edwards, T.; Helton, K.; Grow, M.; Miller, J. W.; Yager, P. SPR Imaging-Based Salivary Diagnostics System for the Detection of Small Molecule Analytes. Ann. N.Y. Acad. Sci. 2007, 1098, 335−344. (16) Fang, S.; Lee, H. J.; Wark, A. W.; Corn, R. M. Attomole Microarray Detection of MicroRNAs by Nanoparticle-Amplified SPR Imaging Measurements of Surface Polyadenylation Reactions. J. Am. Chem. Soc. 2006, 128, 14044−14046. (17) Richens, J. L.; Lunt, E. A. M.; Sanger, D.; McKenzie, G.; O’Shea, P. Avoiding Nonspecific Interactions in Studies of the Plasma Proteome: 9464
dx.doi.org/10.1021/la5016548 | Langmuir 2014, 30, 9457−9465
Langmuir
Article
Practical Solutions to Prevention of Nonspecific Interactions for LabelFree Detection of Low-Abundance Plasma Proteins. J. Proteome Res. 2009, 8, 5103−5110. (18) Richens, J. L.; Vafadar-Isfahani, B.; Vere, K.-A.; Ball, G.; Kalsheker, N.; Rees, R.; Bajaj, N.; O’Shea, P.; Morgan, K. The future role of biomarkers in Alzheimer’s disease diagnostics. In Genetic Variants in Alzheimer’s Disease; Morgan, K., Carrasquillo, M. M., Eds; Springer: New York, 2013. (19) Rees, J. S.; Lilley, K. S. Method for Suppressing Non-Specific Protein Interactions Observed With Affinity Resins. Methods 2011, 54, 407−412. (20) Dobson, C. M.An Overview of Protein Misfolding Diseases. In Protein Folding Handbook; Buchner, J., Kiefhaber, T., Eds.; Wiley-VCH: Weinheim, Germany, 2005. (21) Wall, J.; Ayoub, F.; Oshea, P. Interactions of Macromolecules with the Mammalian-Cell Surface. J. Cell Sci. 1995, 108, 2673−2682. (22) Mayer, L. D.; Hope, M. J.; Cullis, P. R. Vesicles of Variable Sizes Produced by a Rapid Extrusion Procedure. Biochim. Biophys. Acta 1986, 858, 161−168. (23) Wall, J.; Golding, C. A.; Vanveen, M.; Oshea, P. The use of Fluorsceinphosphatidylethanolamine (FPE) as a Real-Time Probe for Peptide Membrane Interactions. Mol. Membr. Biol. 1995, 12, 183−192. (24) Wall, J. S.; Ayoub, F. M.; O’, P. S. A Study of the Interactions of an Immunoglobulin Light Chain with Artificial and B-Lymphocyte Membranes. Front Biosci. 1996, Aug 1, 46−58. (25) Cladera, J.; Martin, I.; Ruysschaert, J. M.; O’Shea, P. Characterization of the Sequence of Interactions of the Fusion Domain of the Simian Immunodeficiency Virus with Membranes - Role of the Membrane Dipole Potential. J. Biol. Chem. 1999, 274, 29951−29959. (26) Cohn, E. J.; Edsall, J. T. Proteins, Amino Acids and Peptides as Ions and Dipolar Ions; Reinhold Publishing Corporation: New York, 1943; pp 145−154. (27) Fan, X.; Lin, L.; Messersmith, P. B. Cell Fouling Resistance of Polymer Brushes Grafted From Ti Substrates by Surface-Initiated Polymerization: Effect of Ethylene Glycol Side Chain Length. Biomacromolecules 2006, 7, 2443−2448. (28) Kenausis, G. L.; Voros, J.; Elbert, D. L.; Huang, N. P.; Hofer, R.; Ruiz-Taylor, L.; Textor, M.; Hubbell, J. A.; Spencer, N. D. Poly(Llysine)-g-Poly(Ethylene Glycol) Layers on Metal Oxide Surfaces: Attachment Mechanism and Effects of Polymer Architecture on Resistance to Protein Adsorption. J. Phys. Chem. B 2000, 104, 3298− 3309. (29) Pasche, S.; Voros, J.; Griesser, H. J.; Spencer, N. D.; Textor, M. Effects of Ionic Strength and Surface Charge on Protein Adsorption at PEGylated Surfaces. J. Phys. Chem. B 2005, 109, 17545−17552. (30) Pasche, S.; De Paul, S. M.; Voros, J.; Spencer, N. D.; Textor, M. Poly(L-lysine)-Graft-Poly(Ethylene Glycol) Assembled Monolayers on Niobium Oxide Surfaces: A Quantitative Study of the Influence of Polymer Interfacial Architecture on Resistance to Protein Adsorption by ToF-SIMS and in Situ OWLS. Langmuir 2003, 19, 9216−9225. (31) Laue, T. Proximity Energies: a Framework for Understanding Concentrated Solutions. J. Mol. Recog. 2012, 25, 165−173. (32) Wang, B. Q.; Tchessalov, S.; Cicerone, M. T.; Warne, N. W.; Pikal, M. J. Impact of Sucrose Level on Storage Stability of Proteins in FreezeDried Solids: II. Correlation of Aggregation Rate with Protein Structure and Molecular Mobility. J. Pharm. Sci. 2009, 98, 3145−3166. (33) Wang, B. Q.; Tchessalov, S.; Warne, N. W.; Pikal, M. J. Impact of Sucrose level on Storage Stability of Proteins in Freeze-Dried Solids: I. Correlation of Protein-Sugar Interaction With Native Structure Preservation. J. Pharm. Sci. 2009, 98, 3131−3144. (34) Jovanovic, N.; Bouchard, A.; Hofland, G. W.; Witkamp, G. J.; Crommelin, D. J. A.; Jiskoot, W. Distinct Effects of Sucrose and Trehalose on Protein Stability During Supercritical Fluid Drying and Freeze-Drying. Eur. J. Pharm. Sci. 2006, 27, 336−345. (35) Singh, K. J.; Roos, Y. H. State Transitions and Freeze Concentration in Trehalose-Protein-Cornstarch Mixtures. LWT−Food Sci. Technol. 2006, 39, 930−938.
(36) Te, J. A.; Tan, M. L.; Ichiye, T. Solvation of glucose, trehalose, and sucrose by the soft-sticky dipole-quadrupole-octupole water model. Chem. Phys. Lett. 2010, 491, 218−223. (37) Luzardo, M. D.; Amalfa, F.; Nunez, A. M.; Diaz, S.; de Lopez, A. C. B.; Disalvo, E. A. Effect of trehalose and sucrose on the hydration and dipole potential of lipid bilayers. Biophys. J. 2000, 78, 2452−2458. (38) IsraelachviliJ. N. Strong Intermolecular Forces: Covalent and Coulomb Interactions. Intermolecular and Surface Forces, 3rd ed.; Academic Press: San Diego, CA, 2011; pp 53−70 (39) Israelachvili, J.; Ruths, M. Brief History of Intermolecular and lntersurface Forces in Complex Fluid Systems. Langmuir 2013, 29, 9605−9619. (40) Makowski, L.; Rodi, D. J.; Mandava, S.; Minh, D. D. L.; Gore, D. B.; Fischetti, R. F. Molecular Crowding Inhibits Intramolecular Breathing Motions in Proteins. J. Mol. Biol. 2008, 375, 529−546. (41) Sinha, N.; Smith-Gill, S. J. Molecular Dynamics Simulation of a High-Affinity Antibody-Protein Complex - The Binding Site is a Mosaic of Locally Flexible and Preorganized Rigid Regions. Cell Biochem. Biophys. 2005, 43, 253−273. (42) Richens, J. L.; Urbanowicz, R. A.; Metcalf, R.; Corne, J.; O’Shea, P.; Fairclough, L. Quantitative Validation and Comparison of Multiplex Cytokine Kits. J. Biomol. Screen. 2010, 15, 562−568. (43) Klaver, A. C.; Patrias, L. M.; Coffey, M. P.; Finke, J. M.; Loeffler, D. A. Measurement of anti-A beta 1-42 antibodies in intravenous immunoglobulin with indirect ELISA: The problem of nonspecific binding. J. Neurosci. Methods 2010, 187, 263−269. (44) Masson, J.-F.; Battaglia, T. M.; Cramer, J.; Beaudoin, S.; Sierks, M.; Booksh, K. S. Reduction of Nonspecific Protein Binding on Surface Plasmon Resonance Biosensors. Anal. Bioanal. Chem. 2006, 386, 1951− 1959. (45) Han, S.; Kim, C.; Kwon, D. Thermal/Oxidative Degradation and Stabilization of Polyethylene Glycol. Polymer 1997, 38, 317−323. (46) Han, S.; Kim, C.; Kwon, D. Thermal-Degradation of Poly(Ethyleneglycol). Polym. Degrad. Stab. 1995, 47, 203−208. (47) Reich, L.; Stivala, S. S. Autoxidation of Poly(Alkylene Glycols) in Solution. J. Appl. Polym. Sci. 1969, 13, 977−988. (48) Mkhatresh, O. A.; Heatley, F. A Study of the Products and Mechanism of the Thermal Oxidative Degradation of Poly(Ethylene Oxide) Using H-1 and C-13 1-D and 2-D NMR. Polym. Int. 2004, 53, 1336−1342. (49) Furuya, M.; Haramura, M.; Tanaka, A. Reduction of Nonspecific Binding Proteins to Self-Assembled Monolayer on Gold Surface. Bioorg. Med. Chem. 2006, 14, 537−543. (50) Bolduc, O. R.; Masson, J.-F. Monolayers of 3-Mercaptopropylamino Acid to Reduce the Nonspecific Adsorption of Serum Proteins on the Surface of Biosensors. Langmuir 2008, 24, 12085−12091. (51) Ning, Y.; Wang, Y.; Li, Y.; Hong, Y.; He, F.; Sun, Q.; Li, M. Development and Assessment of a High Abundant Human Plasma Protein Depletion Kit Using Monoclonal Antibodies Against Native Serum Proteins. Mol. Cell. Proteomics 2005, 4, S215−S215. (52) Granger, J.; Siddiqui, J.; Copeland, S.; Remick, D. Albumin Depletion of Human Plasma Also Removes Low Abundance Proteins Including the Cytokines. Proteomics 2005, 5, 4713−4718.
9465
dx.doi.org/10.1021/la5016548 | Langmuir 2014, 30, 9457−9465