«Kinetic Investigations of Thiolate Protected Gold Nanoparticles: Protein Interactions, Electron Transfer, and Precursor Formation By Brian N. Turner ...»
(3-2) where C is the bulk concentration of mimic A. A linear plot of τ-1 versus C (in mol/L) yields a slope equal to kf and a y-intercept equal to kr. For a 2:1 binding model, this looks like Figure 26.
Figure 26. Plot of individual time constants determined for each concentration.
Assuming a Langmuir isotherm, a linear plot will give a slope equal to k f and a yintercept equal to kr. Above ~ 2 µM, linearity is broken, and the use of those data points to determine binding coefficients is not feasible. The data for the most concentrated sample, 3.90 µM, is questionable, as it quickly violates the 2:1 binding model (as in Figure 24).
From the data, it appears that the 2:1 Langmuir binding model is not valid for the entire process of mimic interacting with the sensor surface. A plausible explanation is found when the assumptions of the Langmuir isotherm are examined. The model assumes that surface coverage is restricted to a single monolayer and that interactions between adsorbates are negligible and all binding sites are equivalent. The equivalence of binding sites is violated to a certain extent, as there is no unidirectional relationship between the number of PZ molecules and the mass bound to the surface (see Table 8). Deviations were attributed to negative steric effects overtaking positive effects of increasing the total number of binding sites. Steric effects do not, however, explain the data in Figure 26. A more informed explanation was found by breaking the other two assumptions of the Langmuir isotherm. It was, therefore, assumed that the mimics could interact with each other, which would, in turn, lead to multilayer coverage. If true, one would expect that a 2:1 Langmuir model is insufficient. Plausibly, the system follows Langmuirian adsorption up to a certain concentration, which was defined as the point at which linearity is broken. If true, multilayer adsorption is catalyzed by surface immobilization of the first monolayer in a fashion similar to systems that follow a BET isotherm. 130, 131 This theory also explains the reason why it was difficult to achieve saturation of the sensor (indicated by a plateau in the signal) after a somewhat lengthy period of time and use of a significant amount of material. Examining the four lowest concentrations, the plot in Figure 27 is produced.
Figure 27: Plot to determine binding coefficients using the four lowest concentration samples.
If the four lowest concentrations are assumed to follow a Langmuirian isotherm, which they appear to, then, for a 2:1 binding model, it is determined that kf = 6.74 x 102 ± 0.88 x 102 M-1 s-1, kr = 1.84 x 10-4 ± 1.09 x 10-4, and, therefore, when one considers that [ ] (3-3) then Ka = 3.66 ± 2.22 x 106 M-1, and Kd = 292 ± 177 nM. These values are in the range expected for mAb/antigen interaction. The value of Kd suggests that the affinity of PZ for the mimic is between ~100 and ~300 fold weaker than PZ for the F protein (Kd = 1.4 nM).86 The binding is competitive with that of epitope scaffolds constructed by McLellan and coworkers (Kd = 87 to 3950 nM) with the same primary epitope sequence.125 The advantage of this mimic is that it does not require recombinant genetic techniques to produce; rather, it involves straightforward laboratory techniques and automated peptide synthesis equipment. While nanoparticle and peptide synthesis have their own unique challenges, it is suspected that this is much more straightforward and cost effective on the basis of materials costs and training time of mentored students.
The most significant problem that was encountered with this material was its poor solution stability at ambient conditions. After a few weeks the solutions lost their activity and visible amounts of precipitate could be seen in the stock solution. Observed agglomeration of this nature might be inhibited by refrigeration, freezing, or purging with inert gas. Chemical modification is another option that is worth exploring in the future.
Tiopronin carboxylate groups on one nanoparticle can form salt bridges with peptides on another nanoparticle, leading to agglomeration. The effect of salt bridges was suspected on the basis of having worked with similar peptides on tiopronin nanoparticles in the past, and using pH 8.4 borate buffer to avoid precipitation, presumably by deprotonating side chains and allowing for charge shielding. Similar problems with p-mercaptobenzoic acid protected nanoparticles was reported in the literature. 29 Borate buffer was not used here because the nanoparticles were initially stable at pH 7. The use of an alternative water solubilizing ligand, such as 1-mercaptotetraethylene glycol,132 could solve this problem.
In conclusion, a strong binding HRSV F protein mimic has been synthesized using straightforward and relatively cost effective techniques that can be taught to a novice science student. Mimic A performs nearly as well as the state of the art subunit mimic (Kd = 87 nM),125 but not as well as the native protein antigen (Kd = 1.4 nM), for palivizumab binding.86 The performance of this mimic could be improved by optimizing the size of the mimic, experimenting with bidentate attachment on different sizes of nanoparticles, modifying the protecting ligand (1-mercaptotetraethylene glycol),132 and optimizing the peptide loading. Given this performance, the mimic should be investigated as a novel subunit vaccine, which is needed in the clinic during the RSV season. This novel material could also be studied as a non-infectious calibration standard for a number of HRSV immunosensors.
The QCM platform that has been developed, while novel and cost-effective, underperforms the SPR in a number of regards. One issue that was not formally raised in the previous discussion is the lack of a stable signal. Often times the signal will drift, and has to be mathematically corrected. Furthermore, it is also problematic that the QCM signal tracks with temperature, which may be the source of the drift.133 Additionally, the difficulty in keeping air bubbles out of the solution leads too often to useless data, increasing the number of experiments to reach a sufficient N value for most studies. The immobilization strategy used is too inconsistent, and is suspected to be responsible for the variance in detection data. That being said, the QCM was effective at detecting the HRSV virion and the nanoparticle mimic with sufficient specificity.
I would like to thank Dr. David Wright for the suggestion to use borate buffer for nanoparticle stabilization. I would also like to re-highlight the contributions to this project from Dr. Yibin Zhang. Furthermore, I would like to acknowledge various nanoparticle synthesis advice gathered along the way from Dr. Brian Huffman, Brian Hixson, and Dr. Kellen Harkness. Also, I would like to thank Dr. Brian Huffman, Dr.
Leslie Hiatt, and Dr. R. Gerald Keil for their advice on nanoparticles and QCM immunosensing experiments.
SECM OF “WIRED” GOLD NANOPARTICLES TO DETERMINE ELECTRON
TRANSFER KINETICS: PROOF OF CONCEPT FOR POTENTIAL COMPONENTS
IN NANOMOLECULAR CIRCUITS
The current chapter continues with the theme of kinetic studies of gold nanoparticles, but turns the attention from matters of clinical interest to the realm of materials science, as well as moving away from more classical understandings of kinetics to quantum mechanically defined electrons. More specifically, the practical application associated with the experiments that follow lies within the field of molecular electronics, where electron transfer rates are among attributes that define what a given material can and cannot accomplish. The chapter begins with small gold nanoparticles, as in the preceding chapters, but now functionalized with ligands not amenable to aqueous solution. The ligands used are interesting and potentially useful for their electronic properties; straight chain alkanethiols as insulators and a relatively new class of molecule deemed “molecular wires” which are known for their metal-like conductivity or, more accurately, resonant tunneling. The nanoparticles themselves are known for their electronic properties, and therefore, the combination of the two should yield interesting nanomaterials for integration into miniaturize and efficient electronic devices. In the following pages, the electronic properties of these materials and how they fit into the field of molecular electronics, the way in which they can be brought together into a single unit, and characterization of the electronic properties of these nanoparticle conjugates in solution by the scanning electron microscope are discussed. It was hypothesized that integrating a “molecular wire” into the protecting monolayer of otherwise insulated gold nanoparticles should increase the electron transfer rate, as conductance happens faster, or more often, than tunneling.
Molecular Electronics In a 2000 editorial in Nature, Daniel Feldheim suggested that more research into the effect of structure on electronic properties of nanomaterials in circuits would lead to the discovery of new electronic properties and new applications. 134 Specifically, he cites applications such as the development of miniaturized devices at a rate that smashes Moore’s law, energy efficient memory storage devices, and extremely sensitive chemical and biological sensors. In the decade that followed, a great number of advances have been made towards these ends, even if molecular and nanoscale electronics are not yet commercially available. These technological advances, along with the excitement surrounding the versatility of gold nanoparticles as previously discussed, motivated the current study which marries the two fields of molecular electronics and nanoparticle science.
The idea of using thiolate protected gold nanoparticles in nanomolecular electronics is not a new one. For instance, in the same year, Schiffrin and coworkers reported on a molecular switch that consisted of a single gold nanoparticle with 60 redox active bipyridine ligands attached to it and a gold electrode. A schematic of this device is shown in Figure 28.
Figure 28: Nanoparticle-bipyridine redox switch.135 Copyright 2000, Nature Publishing Group, reused with permission.
Self-assembled dinonanethiol-bipyridine molecules (molecule (1) in Figure 28) on gold attach the nanoparticle to the surface, and their redox state can be tuned by controlling the voltage. An STM tip measures a current-voltage ratio of about 30 nA when the bipryidine is in its reduced state, but decreases markedly when it is in its oxidized state.
When the bipyridine ligand is replace with hexanedithiol, an insulator and non-redox active compound, negligible current is measured at the STM tip. The proposed low voltage molecular switch, similar to a Schmitt trigger, was estimated to require a mere 30 electrons to operate, only 60 redox active molecules, and occupied a volume of about 9 x 9 nm. As Feldheim points out, however, this device could never fill the role of a Schmitt trigger as it operates too slowly and the gain is too low to be considered commercially feasible.134 (A Schmitt trigger is a switching circuit that changes an output signal when a positive feedback voltage input signal is sufficiently high.) Instead, he suggested that biosensors are a feasible application area for these types of devices.
Recently, Blum and coworkers showed that nanoparticle-molecular wire networks can be assembled with extraordinary spatial regularity using well defined cowpea mosaic virions (CPMV).136 In this study, CPMV is genetically engineered to include cysteine residues at repeated amino acid positions that are fixed by the folding in the virus to well defined spatial coordinates. Addition of 5 nm gold nanoparticles allows for binding at each cysteine residue with 2 and 3 nm gaps between nearest and second nearest neighbor nanoparticles, respectively. The gaps are then filled with 3 nm di-Pt molecular wires and 2 nm OPV molecular wires (defined in Figure 29).
Figure 29: Cowpea mosaic virus-gold nanoparticle-molecular wire conjugate molecular electronic nanosensor. (a) Model of EF–CPMV based on crystallographic data. Cysteine mutation sites shown in white. (b) Model of EF–CPMV with bound 5 nm gold nanoparticles. (c) Model of EF–CPMV showing molecular network of gold nanoparticles with di-Pt (shown in brown) and OPV molecules (shown in white). Conducting molecules used to assemble the nanosensor. di-Pt; 1,4-C6H4 [trans-(4-AcSC6H4C CPt(PBu3)2C C]2 is 3 nm long, and OPV; oligophenylene-vinylene is 2 nm long.136 Copyright 2011, Elsevier, used with permission.
Lysine residues are then biotinylated, resulting in an avidin capturing nanosensor. The sensing is accomplished by measuring current-voltage curves of these nanosensors before and after avidin exposure. The proximity of charges on various avidin proteins changes the conductance of the overall nanoparticle-molecular network in a way that tracks with the protein charge. The sensors exhibited good specificity when compared to nonbiotinylated controls.
Given this background literature in nanomolecular electronics, it would be useful to describe the electronic properties of the separate components discussed in the above devices: gold nanoparticles and molecular wires.
Electrochemical Properties of Gold Nanoparticles Gold nanoparticles have often been cited for having interesting and useful electrochemical properties. One of the more interesting properties gold nanoparticles possess, in this regard, is their ability to act as a nanoscale capacitor; that is, they are small and soluble like molecules but can undergo a relatively great many redox events within a given voltage window with respect to a small molecule. Using square wave voltammetry, Hicks and coworkers were able to generate voltammograms displaying a great number of charging and discharging events for highly monodisperse hexanethiolate protected gold nanoparticles (AuHexanethiolate). 137 Figure 30 shows a similar experiment performed in our lab.138 Figure 30: Square wave voltammogram of AuHexanethiolate nanoparticles. 138 Within the voltage range shown (-0.9V to +1.6V), the nanoparticle stores and reversibly discharges 10 electrons.