We chose gold nanorods as plasmonic transducers of binding events for two primary reasons. First, we chose gold over silver, even though silver particles exhibit a higher bulk refractive index sensitivity than gold particles of the same shape and size,41
because the greater reactivity of silver as compared to gold makes it less suitable for use in biologically relevant media. Second, we chose nanorods over many other possible shapes because gold nanorods can be conveniently synthesized to exhibit plasmon bands with peak wavelengths ranging from 600 to 900 nm simply by tuning their aspect ratio and size through chemical synthesis methods.35
This wavelength range is attractive for optical biosensing for many reasons. Recent simulations by Miller and Lazarides have shown that the bulk refractive index sensitivity of the plasmon band is linearly correlated with the wavelength of the LSPR peak for particles of a specified composition42
so that nanoparticles with a plasmon band at higher wavelengths are more sensitive to their local environment than those at shorter wavelengths. In view of this finding, we chemically synthesized nanorods with dimensions of 74 nm × 33 nm (l
) with a longitudinal plasmon band with a peak wavelength of 780 nm to maximize their sensitivity to their environment while remaining in the visible wavelength range (). The choice of rods with a scattering peak wavelength centered at ~780 nm is also advantageous because the background absorption and scattering of endogenous chromophores from biological mixtures (e.g., serum and blood) is minimal in this wavelength range.
shows a dark-field micrograph of nanorods immobilized on glass acquired on the microspectroscopy system. The dark-field micrograph shows diffraction-limited spots corresponding to light scattered by individual nanoparticles. shows a typical scattering spectrum of a single nanorod that is obtained on the dark-field microspectroscopy system and shows the high S/N of 80 and low full width at half-maximum (fwhm) of ~60 nm that is obtained for an isolated gold nanorod by single-particle spectroscopy with an acquisition time of 10 s. The fwhm of 60 nm of these nanorods is narrower than that of other nanostructures that also have a plasmon band in this wavelength range such as gold nanoshells,43
which allows for more accurate determination of peak shifts, and is a useful spectral feature of these nanorods as plasmonic transducers.
This figure also highlights an important spectroscopic advantage (albeit at the cost of more complex instrumentation), by comparing the scattering spectrum of a single immobilized gold nanorod (, red curve) with the extinction spectrum of an ensemble of ~108
nanorods from the same synthesis batch immobilized on glass (, blue curve), as measured by a UV–vis spectrophotometer (Cary 300-Bio). The ensemble spectrum exhibits heterogeneous broadening compared to the single Lorentzian peak observed for each resonance of a single nanoparticle. This broadening is simply due to the fact that even the most careful chemical synthesis will yield nanoparticles with a distribution of size and shapes, and as the LSPR behavior of a nanoparticle is strongly dependent upon these structural parameters, this structural inhomogeneity leads to a broadening of the ensemble LSPR peaks. Additionally, the location and amplitude of smaller scattering peaks corresponding to plasmonic resonances other than the longitudinal resonance of the nanorod are observed. The location and magnitude of these minor peaks has been shown to be highly dependent on nanorod end cap geometry and thus tends to vary widely between chemically synthesized nanorods.44
The bulk refractive index sensitivity was determined as shown in the inset to . The sensitivity of the rods used in this study was 261.7 ± 26.9 nm/RIU (n
= 15). This value determined by single nanorod measurements is in good agreement with the previously reported sensitivity of 252 nm/RIU from an ensemble of gold nanorods with the same nominal dimensions.18
These data highlight that rods are attractive as optical transducers that work in a wavelength sensing mode, because their mean bulk refractive index sensitivity of 262 nm/RIU is significantly greater than that of 39 nm diameter gold spheres25
(70 nm/RIU), and gold nanoshells (~140 nm/RIU)30
that have a similar wavelength maximum of the plasmon band around 700 nm.
The large variability noted in refractive index sensitivity is believed to be caused by several reasons. The largest variations are a result of nanoparticle size variations resulting from the chemical synthesis technique. These geometric variations result in a corresponding distribution of nanoparticle sensitivities since nanoparticle sensitivity has been shown to be directly correlated to LSPR wavelength42
and LSPR wavelength is directly determined by nanoparticle shape.8–11,45,46
Additionally, remaining CTAB on the surface of some of the nanorods could affect the measured sensitivity of the nanorod. CTAB was rinsed from the nanorod suspensions by centrifugation and resuspension. If additional rinse cycles were performed, the nanorods would aggregate and fall out of suspension.18
Thus, a critical amount of CTAB is required to keep the nanorods in stable suspension. We believe that small amounts of CTAB remain on some of the particles, restricting the active sensing volume accessible for sensitivity measurements as well as binding experiments.
Next, we investigated the binding of streptavidin to a biotin-functionalized gold nanorod. , parts a and b, shows three representative spectra collected from a streptavidin binding experiment. The first spectrum was collected before conjugating biotin to the nanorods, and the second spectrum was collected after biotin was conjugated to the nanorods. The first two spectra are useful diagnostics of the efficiency of biotin conjugation to a gold nanorod. From previously reported ensemble studies, it has been observed that successful biotin conjugation onto nanorods increases the refractive index surrounding the nanoparticles and induces a reproducible red shift in the LSPR peak18,25
The 5.6 nm shift observed after biotin conjugation indicates successful coupling of the biotin to the nanorod. The third spectrum was collected after the nanorod was incubated in a solution of 130 nM streptavidin. The resulting 5.2 nm shift indicated that streptavidin binding occurred on the surface of the biotinylated nanorod. We conclude that the nanorod sensor saturated at this concentration because incubation with higher concentrations of streptavidin resulted in identical LSPR shifts (data not shown).
Figure 3 (a and b) Scattering spectra of a single gold nanorod after sequential incubation in EG3SH/MHA (blue), biotin (red), and 10 nM streptavidin (black). (c and d) Scattering spectra of a single gold nanorod in EG3SH/MHA (blue), biotin (red), and 100 nM streptavidin (more ...)
In order to probe the system detection limit, measurements were taken of biotinylated gold nanorods in streptavidin concentrations of 130, 10, and 1 nM. A mean LSPR centroid shift of 5.29 ± 1.47 nm (95% CI, n = 9) was observed upon introduction of the 130 nM streptavidin solution. An LSPR peak centroid shift of 1.22 ± 0.24 nm (95% CI, n = 8) was measured from 10 nM streptavidin and 0.588 ± 0.32 nm (95% CI, n = 9) for 1 nM streptavidin.
, parts c and d, shows the results of a control experiment using biotin-saturated streptavidin. The spectra shown are of a single gold nanorod before biotin conjugation, after biotin conjugation, and after incubation with biotin-saturated streptavidin. The characteristic ~5 nm LSPR shift resulting from the biotin coupling was observed, consistent with the previous experiment, which indicated that biotin was successfully conjugated to the gold nanorods. However, there was no further shift in the LSPR peak centroid upon incubation with the biotin-saturated streptavidin, which clearly demonstrated that the saturation of biotin-binding sites on the streptavidin prevents binding of streptavidin to the biotin-functionalized nanoparticle surface. These results also suggest that the mixed SAM on the nanorod surface successfully prevented nonspecific adsorption of streptavidin because the streptavidin molecules did not nonspecifically adsorb to the nanorods when their biotin-binding sites were blocked. Together, these results strongly suggest that the measured LSPR shifts observed for streptavidin binding to the biotin-functionalized nanorods are caused by molecular recognition of biotin by streptavidin.
Next, the time-resolved spectral acquisition mode of the microspectroscopy system was used to measure the kinetics of streptavidin binding to biotin-conjugated nanorods. shows a plot of the LSPR scattering peak centroid versus time for single biotin-conjugated nanorods incubated in 130, 10, and 1 nM streptavidin. A rapid 5.10 nm shift in the LSPR scattering peak centroid was observed upon introduction of the 130 nM streptavidin solution. After rinsing with PBS at time t
= 100 min, the peak centroid location remained constant, which demonstrated the irreversibility of the biotin–streptavidin binding and is consistent with the long (~35 h) half-life of the biotin–streptavidin bond.47,48
A steady-state LSPR peak centroid shift of 1.09 nm was observed for streptavidin binding at a concentration of 10 nM. Finally, a shift of 0.48 nm was observed from a biotin-conjugated nanorod after incubation in streptavidin concentration of 1 nM. This shift is close to the limit of detection of this system, which we estimate to be ~0.3 nm.40
Real-time measurement of the LSPR scattering peak centroid shifts of single biotin-conjugated gold nanorods incubated in 130 (black), 10 (blue), and 1 nM (red) streptavidin in PBS.