Preliminary studies and potential applications of localized surface plasmon resonance spectroscopy in medical diagnostics

Transcription

1 Review CONTENTS Optical properties of noble metal nanoparticles Using nanoparticles for biomolecule sensing LSPR sensing of large proteins & antibodies on arrays of nanoparticles Reversibility of the sensor Selectivity of the sensor Detection of disease markers Miniaturization of the sensor Comparison with an industry standard: SPR sensors Outlook & conclusions Expert opinion Five-year view Key issues References Affiliations Author for correspondence Northwestern University, Department of Chemistry, 2145 Sheridan Road, Evanston, IL , USA Tel.: Fax: KEYWORDS: biosensing, disease diagnosis, localized surface plasmon resonance nanosensor, single nanoparticle spectroscopy Preliminary studies and potential applications of localized surface plasmon resonance spectroscopy in medical diagnostics Amanda J Haes and Richard P Van Duyne Miniature optical sensors that specifically identify low concentrations of environmental and biological substances are in high demand. Currently, there is no optical sensor that provides identification of the aforementioned species without amplification techniques at naturally occurring concentrations. Recently, it has been demonstrated that triangular silver nanoparticles have remarkable optical properties and that their enhanced sensitivity to their nanoenvironment has been used to develop a new class of optical sensors using localized surface plasmon resonance spectroscopy. The examination of both model and nonmodel biological assays using localized surface plasmon resonance spectroscopy will be presented in this review. It will be demonstrated that the use of a localized surface plasmon resonance nanosensor rivals the sensitivity and selectivity of, and provides a low-cost alternative to, commercially available sensors. Expert Rev. Mol. Diagn. 4(4), (2004) Advancements in technology due to nanoscale phenomena will become optimized when the chemical and physical properties of materials are more thoroughly understood. Relevant to this work, the potential to develop highly sensitive and specific sensors for biological targets motivates a portion of the research in this field. Previously, results have been presented based on the nanoscale limit of surface plasmon resonance (SPR) sensors. This novel, nanoscale development, termed the localized surface plasmon resonance (LSPR) nanosensor, is a refractive index-based sensing device that relies on the extraordinary optical properties of noble (e.g., Ag, Au and Cu) metal nanoparticles [1 5]. Optical properties of noble metal nanoparticles Noble metal nanoparticles exhibit a strong ultraviolet (UV)-visible (vis) absorption band that is not present in the spectrum of the bulk metal [6 12]. This absorption band results when the incident photon frequency is resonant with the collective oscillation of the conduction electrons and is known as the LSPR. LSPR excitation results in: Wavelength selective absorption with extremely large molar extinction coefficients (~ M -1 cm -1 ) [13] Resonant Rayleigh scattering [14,15] with an efficiency equivalent to that of 10 6 fluorophors [16] The enhanced local electromagnetic fields near the surface of the nanoparticle, which are responsible for the intense signals observed in all surface-enhanced spectroscopies [5,14,17 19] The simplest theoretical approach available for modeling the optical properties of nanoparticles is the Mie theory estimation of the extinction of a metallic sphere in the long wavelength, electrostatic dipole limit. This is explained in EQUATION 1 [20]: E( λ) = 24πN A a ε m λ ln( 10) ε i ( ε r + χε m ) 2 + ε 2 i Future Drugs Ltd. All rights reserved. ISSN

2 Haes & Van Duyne E(λ) = extinction (viz., sum of absorption and scattering) N A = areal density of nanoparticles A = radius of the metallic nanosphere ε m = dielectric constant of the medium surrounding the metallic nanosphere (assumed to be a positive, real number and wavelength independent) λ = wavelength of the absorbing radiation ε i = imaginary portion of the metallic nanoparticle s dielectric function ε r = real portion of the metallic nanoparticle s dielectric function χ describes the polarization factor that corresponds to the aspect ratio of the nanoparticle (equal to 2 for a sphere) It is abundantly clear that the LSPR spectrum of an isolated metallic nanosphere embedded in an external dielectric medium will depend on the nanoparticle radius A, the nanoparticle material (ε i and ε r ) and the nanoenvironment s dielectric constant (ε m ). Furthermore, when the nanoparticles are not spherical, as is always the case in real samples, the extinction spectrum will depend on the nanoparticle s inplane diameter, out-of-plane height and shape (χ). The values for χ increase from 2 (for a sphere) up to and beyond values of 17 for a 5:1 aspect ratio nanoparticle. In addition, many of the samples considered in this work contain an ensemble of nanoparticles that are supported on a substrate. Thus, the LSPR will also depend on interparticle spacing and substrate dielectric constant. Using nanoparticles for biomolecule sensing It is apparent from EQUATION 1 that the location of the extinction maximum of noble metal nanoparticles is highly dependent on the dielectric properties of the surrounding environment. Wavelength shifts in the extinction maximum of nanoparticles can be used to detect molecule-induced changes surrounding the nanoparticle. As a result, there are at least four different nanoparticle-based sensing mechanisms that enable the transduction of macromolecular or chemical-binding events into optical signals based on changes in the LSPR extinction or scattering intensity shifts in LSPR λ max, or both. These mechanisms are: Resonant Rayleigh scattering from nanoparticle labels in a manner analogous to fluorescent dye labels [15,16,21 28] Nanoparticle aggregation [29 33] Charge-transfer interactions at nanoparticle surfaces [1,20,34 37] Local refractive index changes [1 5,38 46] In this review, the refractive index change in the environment near noble metal nanoparticles is demonstrated to be an effective platform for highly sensitive detection techniques. In order to systematically study the LSPR response of noble metal nanoparticles to changes in their dielectric environment, a technique that produces nanoparticles with size and shape monodispersity is required. The chemical synthesis of noble metal nanostructures is often employed for this purpose [47 52]. These approaches are used to prepare high concentrations of a variety of shapes and sizes of nanoparticles with varying degrees of monodispersity and tunable optical properties (FIGURE 1). For solution-phase LSPR-based sensing, signal transduction depends on the sensitivity of the surface plasmon to interparticle coupling. When there are multiple particles in solution that support a localized surface plasmon and are in close proximity (i.e., interparticle spacings less than the nanoparticle diameter), they are able to interact electromagnetically through a dipole coupling mechanism. This broadens and red shifts the LSPR, a change easily detected using UV-vis spectroscopy. Two methods of detection readily lend themselves to monitoring these changes in the position of the LSPR: UV-vis extinction (absorption plus scattering) and resonant Rayleigh scattering spectroscopy. A 1 2 B 20 nm 83 nm nm Extinction nm 40 nm (4) Wavelength (nm) 3 50 nm 6 (2) (5) (6) (3) (1) Figure 1. Tunable Ag and Au nanoparticle solutions fabricated via the citrate reduction of metal salts. (A) Corresponding transmission electron micrographs. (1) The solution consists of homogenous Au nanospheres (13 nm diameters). (2) The solution consists of inhomogenous Ag nanoparticles (nanospheres, trigonal prisms and polygon platelets). (3) The solution consists of Ag nanoparticles (nanospheres, trigonal prisms, and polygon platelets). (4) The solution consists of Ag nanoparticles (trigonal prisms and polygon platelets). (5) The solution consists of Ag nanoparticles (trigonal prisms with rounded tips and polygon platelets). (6) The solution is made up of inhomogenous oblong Ag nanoparticles. (B) Ultraviolet-visible extinction spectra of the corresponding solutions. 528 Expert Rev. Mol. Diagn. 4(4), (2004)

3 Localized surface plasmon resonance spectroscopy Recently, several papers have been published on a gold nanoparticle-based UV-vis technique for the detection of DNA. This colorimetric detection method is based on the change in absorbance spectra (i.e., color) as particles are brought together by the hybridization of complementary DNA strands [30 33,53 55]. The limits of detection (LOD) reported are in the range of tens of femtomoles of target oligonucleotide. These nanoparticle aggregation assays represent a 100-fold increase in sensitivity over conventional fluorescence-based assays [54]. An alternative approach to solution-phase nanoparticle sensing is to synthesize nanoparticles bound to substrates. In this format, signal transduction depends on changes in the nanoparticles dielectric environment induced by solvent or target molecules (not via nanoparticle coupling). One approach to fabricate these nanoparticles is to use nanosphere lithography (NSL) [56], a powerful fabrication technique that inexpensively produces arrays of nanoparticles with controlled shape, size and interparticle spacing. NSL begins with the self-assembly of size-monodisperse nanospheres of diameter, D, to form a 2D colloidal crystal deposition mask. The substrate is prepared so that the nanospheres freely diffuse until they reach their lowest energy configuration. This is achieved by chemically modifying the nanosphere surface with a negative charge that is electrostatically repelled by the negatively charged substrate, such as mica or chemically treated glass. As the solvent (water) evaporates, capillary forces draw the nanospheres together and they crystallize in a hexagonally close-packed pattern on the substrate. As in all naturally occurring crystals, nanosphere masks include a variety of defects that arise as a result of nanosphere polydispersity, site randomness, point defects (vacancies), line defects (slip dislocations) and polycrystalline domains. Typical defect-free domain sizes are in the µm range. Following self-assembly of the nanosphere mask, a metal or other material is then deposited by thermal evaporation, electron beam deposition, or pulsed laser deposition from a collimated source normal to the substrate through the nanosphere mask to a controlled mass thickness, d m. Following metal deposition, the nanosphere mask is removed, typically by sonicating the entire sample in a solvent, leaving behind surface-confined nanoparticles that have a triangular footprint. The optical properties of these nanoparticles can be easily tuned throughout the visible region of the spectrum by changing the size or shape of the nanoparticles (FIGURE 2). LSPR sensing of large proteins & antibodies on arrays of nanoparticles Using NSL, the authors have demonstrated that nanoscale chemosensing and biosensing could be realized through shifts in the LSPR extinction maximum (λ max ) of these triangular silver nanoparticles [1,2,4,43]. Instead of being caused by the electromagnetic coupling of the nanoparticles, these wavelength shifts are caused by adsorbateinduced local refractive index changes in competition with chargetransfer interactions at the surface of nanoparticles. It should be noted that the signal transduction mechanism in this nanosensor is a reliably measured wavelength shift rather than an intensity change as in many previously reported nanoparticle-based sensors. A B C 1 Extinction nm nm (2) 552 nm (1) 432 nm (3) 623 nm Wavelength (nm) 250 nm Figure 2. Tunable Ag nanoparticle substrates fabricated using nanosphere lithography. (A) Photographs of nanoparticle substrates. (B) Atomic force microscopy images of nanoparticle substrates. (C) Ultraviolet-visible extinction spectra of Ag nanoparticle substrates. In all sections, Ag nanoparticle substrate D = 400 nm, d m = 50.0 nm. (1) After thermal in vacuum for 1 h at 600 C, (2) after thermal annealing in vacuum for 1 h at 300 C, and (3) as fabricated (no annealing). As required for all new sensor development, a model system must be chosen to test the capabilities of the sensor with a wellcharacterized system. For that reason, the well-studied biotin/streptavidin system [2], with its extremely high binding affinity (K a ~10 13 M -1 ), and the antigen/antibody couple, biotin/antibiotin [4], were chosen to illustrate the attributes of these LSPR-based nanoscale affinity biosensors. For these experiments, macroscale UV-vis extinction spectra were collected in standard transmission geometry with unpolarized light using a miniature spectrometer. In the streptavidin studies, NSL was used to create surfaceconfined triangular Ag nanoparticles supported on a glass substrate [2]. The Ag nanotriangles have in-plane widths of approximately 100 nm and out-of-plane heights of approximately 51 nm as determined by atomic force microscopy. To prepare the LSPR nanosensor for biosensing events, the Ag nanotriangles are first functionalized with a self-assembled monolayer (SAM) composed of 3:1 1-octanethiol:11-mercaptoundecanoic acid resulting in a surface coverage corresponding

4 Haes & Van Duyne to 0.1 monolayer of carboxylate binding sites [57]. Since the maximum number of alkanethiol molecules per nanoparticle is 60,000 [1], this results in an equivalent of approximately 6000 carboxylate binding sites per nanoparticle. Next, an amine-terminated biotin was covalently attached to the carboxylate groups using 1-ethyl-3,3 (dimethylaminopropyl)carbodiimide (EDC). The number of resulting biotin sites is determined by the yield of the EDC coupling reaction. Since this is likely to be approximately 1 5% efficient [58], one expects there to be only biotin sites per nanoparticle at maximum coverages. In these studies, the λ max of the Ag nanoparticles were monitored during each surface functionalization step. An example of an assay for the detection of streptavidin follows. First, the LSPR λ max of the bare Ag nanoparticles were measured to be nm. To ensure a well-ordered SAM on the Ag nanoparticles, the sample was incubated in the thiol solution for 24 h. After careful rinsing and thorough drying with N 2 gas, the LSPR λ max after modification with the mixed SAM was measured to be nm. The LSPR λ max shift corresponding to this surface functionalization step was a +38 nm wavelength shift. Next, biotin was covalently attached via amide bond formation with a two-unit polyethylene glycol linker to carboxylated surface sites. The LSPR λ max after biotin attachment was measured to be nm, corresponding to an additional +11 nm shift. Given the magnitude of this shift and the expected magnitude of the response (see EQUATION 2, vida infra), this shift is confirmed to arise from an approximately 5% submonolayer coverage of biotin. The LSPR nanosensor has now been prepared for exposure to the target analyte. To ensure that molecular binding had saturated, approximately 25 µl of the molecule sample was exposed to the nanoparticle substrate for 3 h [2]. Exposure to 100 nm streptavidin resulted in LSPR λ max = nm, corresponding to a +27 nm shift for a full monolayer coverage of streptavidin (FIGURE 3). Repeating this experiment with exposure of the nanosensor surface to only 1 pm streptavidin results in a marked decrease of the response to a small but reproducibly detected +4 nm shift. This concentration limit is as sensitive as other commercially available optical sensors. An identical approach was taken to detect antibiotin antibody (sigma). However, instead of exposing the LSPR sensor chip to streptavidin, it was exposed to antibiotin antibody. The binding affinity of antibiotin antibody to biotin is significantly lower ( to M -1 ) than that of biotin/streptavidin [59,60]. In this case, upon exposure to 700 nm antibiotin antibody, a wavelength shift of nm was detected. When the solution concentration was decreased to 700 pm antibiotin antibody, a small but reproducible +2.5 nm LSPR response shift was observed. Clearly, as the concentration of the solution decreases, the LSPR wavelength shift response decreases. This encouraged the authors to determine the full response of the sensor over a wide concentration range. For this reason, variable analyte concentrations were exposed to a biotinylated LSPR chip to test the sensitivity of the system to different molecules. Specifically, the LSPR λ max shift, R, versus [analyte] response curve was 40 Streptavidin B LSPR response (nm) Extinction Rmax = 26.5 nm K a, surf = M -1 LOD < 1 pm Biotin Streptavidin Wavelength (nm) Concentration (M) A Extinction Biotin Antibiotin R max = 38.0 nm K a, surf = M -1 LOD < 700 pm Antibiotin Wavelength (nm) Figure 3. The specific binding of streptavidin (A) and antibiotin antibody (B) to a biotinylated Ag nanobiosensor is shown in the response curves. All measurements were collected in a N 2 environment. The solid line is the calculated value of the nanosensor's response. The left inset represents an LSPR nanosensor experiment in which biotinylated nanoparticles are exposed to 100 nm streptavidin. As shown, the LSPR of biotinylated Ag nanoparticles shifts to the red by 27 nm upon the specific linkage of a monolayer of streptavidin to its surface. The right inset represents an LSPR nanosensor experiment in which biotinylated nanoparticles are exposed to 7 mm antibiotin antibody. As shown, the LSPR of biotinylated Ag nanoparticles shifts to the red by 38 nm upon the specific linkage of a monolayer of antibiotin antibody to its surface. K a,surf : Surface-confined thermodynamic binding constant; LSPR: Localized surface plasmon resonance; LOD: Limit of detection; R max : Saturation response. 530 Expert Rev. Mol. Diagn. 4(4), (2004)

5 Localized surface plasmon resonance spectroscopy measured over the concentration range M < [streptavidin] < M and M < [antibiotin] < M (FIGURE 3) [2,4]. The lines seen in the figure are not a fit to the data. Instead, the line was computed from a response model (a complete analysis of this model is described in [4]). It was found that this response could be interpreted quantitatively in terms of a model involving: 1:1 binding of a ligand to a multivalent receptor with different sites but invariant affinities The assumption that only adsorbate-induced local refractive index changes were responsible for the operation of the LSPR nanosensor The binding curve provides three important characteristics regarding the system being studied. First, the mass and dimensions of the molecules affect the magnitude of the LSPR shift response. Comparison of the data with theoretical expectations yielded a saturation response of R max = 26.5 nm for streptavidin, a 60-kDa molecule, and 38.0 nm for antibiotin antibody, a 150-kDa molecule. Clearly, a larger mass density at the surface of the nanoparticle results in a larger LSPR response. Next, the surface-confined thermodynamic binding constant K a,surf can be calculated from the binding curve and is estimated to be M -1 for streptavidin and M -1 for antibiotin antibody. These numbers are directly correlated to the third important characteristic of the system, the LOD. The LOD is less than 1 pm for streptavidin and 100 pm for antibiotin antibody. As predicted, the LOD of the nanobiosensor studied is lower for systems with higher binding affinities, such as for the well-studied biotin streptavidin couple, and higher for systems with lower binding affinities as seen in the antibiotin antibody system. Given this information and analysis, similar treatment can be made for virtually any ligand receptor system. It should be noted that the LOD of the system corresponds to the smallest reliable wavelength shift response induced by a given solution concentration. These real LODs are often translated to the commonly reported surface coverage in terms of molecules. For the LSPR nanosensor, the surface coverage detection corresponds to approximately 25 streptavidin molecules/nanoparticle. In previously described experiments, approximately nanoparticles were probed (nanoparticle density approximately nanoparticles/cm 2 with a spot size with a 1-mm radius) and this corresponds to the detection of streptavidin molecules. A clear method to decrease the number of molecules detected would be to decrease the number of nanoparticles probed. This has been recently demonstrated and is discussed later. Reversibility of the sensor In order for LSPR nanobiosensors to fulfill their mandate, they must be biocompatible and work under physiological conditions. Some binding interactions, such as poly-l-lysine to a negatively charged surface, can interact reversibly, while other couples with higher surface binding affinities interact irreversibly. A commercially viable nanobiosensor should be entirely reusable. In the case of this study, this means that the analyte detection must be entirely removable, rendering the sensor reusable. Reusability has a large impact on the cost effectiveness and the simplicity of biosensor use. The reversibility of the LSPR nanobiosensor has been demonstrated by exposing an antibiotin antibody functionalized chip with an excess concentration of biotin [4]. In less than 30 s, the sensor s capability to be reused to detect antibiotin antibody had been fully regenerated. While this experiment is more difficult if the affinity between the ligand and receptor is stronger (as is the case with biotin/streptavidin), these results clearly indicate that an LSPR sensor can be reused multiple times for antigen/antibody interactions. Selectivity of the sensor Although LSPR spectroscopy is a totally nonselective sensor platform, a high degree of analyte selectivity can be conferred using the specificity of surface-attached ligands and passivation of the sensor surface to nonspecific binding. For this reason, a set of control experiments were performed to show that the streptavidin and antibiotin antibody binding to the sensor surface containing no capture ligand (biotin), prebiotinylated streptavidin binding to a sensor surface with biotin and bovine serum albumin in large excess, simulating a clinical sample, binding to a sensor surface with biotin, all produce wavelength shift responses less than that corresponding to the LOD [2]. The positive results found with these model systems were gained after the development of proper surface chemistries and correct rinsing buffers. In order to realize the full potential of this technique, the development of highly specific biomarkers that capture specific analytes will aid in the continued success of the LSPR sensor chip as well as other devices that rely on molecular interactions for a signal transduction. Detection of disease markers Alzheimer s disease is the leading cause of dementia in people over 65 years of age and affects an estimated 4 million Americans [61]. Although first characterized almost 100 years ago by Alois Alzheimer, who discovered brain lesions, now known as plaques, and tangles in the brain of a middle-aged woman who died with dementia in her early fifties [62], the molecular cause of the disease is not understood. Also, an accurate diagnostic test has yet to be developed. However, two inter-related theories for Alzheimer s disease have emerged that focus on the putative involvement of neurotoxic assemblies of a small 42-amino acid peptide known as amyloid-β [63,64]. Although a normal protein catabolite, amyloid β is abundant in Alzheimer s disease brain tissue, where it polymerizes into extremely large amyloid fibrils. Insoluble deposits of amyloid fibrils constitute the proteinaceous cores of Alzheimer s disease plaques. The widely investigated amyloid cascade hypothesis suggests that the amyloid plaques cause neuronal degeneration and, consequently, memory loss and further progressive dementia. In this theory, the amyloid β protein monomers, present in all normal individuals, do not exhibit toxicity until they assemble into amyloid fibrils [65]. However, there is a poor correlation between plaque sites 531

6 Haes & Van Duyne and disease progression, and more recent studies have established that amyloid plaques are not the only toxic form of amyloid-β [66]. The other toxins are known as amyloid-β-derived diffusible ligands (ADDLs). In contrast to the fibrils, ADDLs are small, globular and readily soluble, comprising of 3 24-mer oligomers of the amyloid-β monomer [67]. ADDLs are potent and selective CNS neurotoxins. Perhaps even more significantly, they inhibit mechanisms of synaptic information storage with great rapidity prior to any neurodegenerative consequences [67]. ADDLs have now been confirmed to be greatly elevated in autopsied brains of Alzheimer s disease subjects [68]. Therefore, it would be of great value to develop new tests capable of quantifying ADDLs in patient populations. Such an ultrasensitive method for ADDL/anti-ADDL antibody detection could potentially emerge from LSPR nanosensor technology, providing an opportunity to develop the first clinical laboratory diagnostic test for Alzheimer s disease. Preliminary results indicate that the LSPR nanosensor can be used to aid the diagnosis of Alzheimer s disease [69,70]. In these studies, antibodies that specifically interact with ADDLs were decorated onto the silver nanoparticle surface. The sensor s viability was tested by exposing this surface to model target ADDL molecules. An additional antibody then amplified the LSPR nanosensor s response. After the demonstration of the model assay, the target molecules that were sandwiched in the previous experiment were substituted by cerebral spinal fluid from diseased and nondiseased patients [69]. The fluid was centrifuged to remove large pieces of cellular material; otherwise, the samples were not further modified. In preliminary assays that took less than 2 h to perform, huge differences were observed upon analysis of the two types (diseased and nondiseased) of samples. Miniaturization of the sensor Clearly, an obvious method to improve the limit of detection of the nanoparticle array system would be to reduce the number of nanoparticles probed. A key to exploiting single nanoparticles as sensing platforms is to develop a technique to monitor the LSPR of individual nanoparticles with a reasonable signal-to-noise ratio. UV-vis absorption spectroscopy does not provide a practical means of accomplishing this task. Even under the most favorable experimental conditions, the absorbance of a single nanoparticle is very close to the shot noise-governed LOD. Instead, resonant Rayleigh scattering spectroscopy is the most straightforward means of characterizing the optical properties of individual metallic nanoparticles. Similar to fluorescence spectroscopy, the advantage of scattering spectroscopy lies in the fact that the scattering signal is being detected in the presence of a very low background. The instrumental approach for performing these experiments generally involves using high magnification microscopy coupled with oblique or evanescent illumination of the nanoparticles. Recently, the authors demonstrated that a single silver nanoparticle can be used to detect a submonolayer coverage of streptavidin [25]. In these experiments, chemically synthesized Ag nanoparticles were dispersed on a glass coverslip in a flow cell and a dark field image was collected. After incubation in 10 nm streptavidin, a nm shift was monitored, a response attribute to less than 700 streptavidin molecules. Comparison with an industry standard: SPR sensors During the course of these findings, it was realized that the sensor transduction mechanism of this LSPR-based nanosensor is analogous to that of flat surface, propagating SPR sensors (TABLE 1). For approximately 20 years, SPR sensors copper, gold or silver planar films have been used as refractive indexbased sensors to detect analyte binding at or near to a metal surface. They have also been widely used to monitor a broad range of analyte surface-binding interactions, including the adsorption of small molecules [71 73], ligand receptor binding [74 77], protein adsorption on self-assembled monolayers [78,79], antibody antigen binding [80], DNA and RNA hybridization [81 83], and protein DNA interactions [84]. Just as in LSPR spectroscopy, the sensing mechanism of SPR spectroscopy is based on the measurement of small changes in refractive index that occur in response to analyte binding at or near to the surface of a noble metal (e.g., Au, Ag and Cu) [85]. Chemosensors and biosensors based on SPR spectroscopy possess many desirable characteristics including: Refractive index sensitivity on the order of one part in corresponding to an areal mass sensitivity of approximately 1 10 pg/mm 2 [71,72,86] Long-range sensing length scale determined by the exponential decay of the evanescent electromagnetic field, L z approximately 200 nm [71] Multiple instrumental modes of detection (viz., angle shift, wavelength shift and imaging) [85] Real-time detection on the s time scale for measurement of binding kinetics [72,73,87,88] Lateral spatial resolution on the order of 10 µm, enabling multiplexing and miniaturization, especially using the SPR imaging mode of detection [85] The system is available commercially Important differences between the SPR and LSPR sensors are the comparative refractive index sensitivities and the characteristic electromagnetic field decay lengths. SPR sensors exhibit large refractive index sensitivities (~ nm/riu) [71]. For this reason, the SPR response is often reported as a change in refractive index units. The LSPR nanosensor, on the other hand, has a modest refractive index sensitivity (~ nm/riu) [1]. Given that this is four orders of magnitude smaller in comparison with the SPR sensor, initial assumptions were made that the LSPR nanosensor would be 10,000-times less sensitive than the SPR sensor. However, this is not the case. In fact, the two sensors are very competitive in their sensitivities. The short (and tunable) characteristic electromagnetic field l d, provides the LSPR nanosensor with its enhanced sensitivity [5,18]. These LSPR nanosensor results indicate that the decay length, l d, is approximately 5 15 nm, or 1 3% of the light s wavelength, and depends on the size, shape 532 Expert Rev. Mol. Diagn. 4(4), (2004)

7 Localized surface plasmon resonance spectroscopy Table 1. Comparison between SPR and LSPR sensors. Feature/characteristic SPR LSPR Label-free detection Yes [73,75,82,89] Yes [1,2,4,24] Distance dependence ~1000 nm [71] ~30 nm (size tunable) [5,18] Refractive index sensitivity nm/riu [71,72,74,86] 2 x 10 2 nm/riu [1,18] Modes Angle shift [85] Wavelength shift Imaging Extinction [2] Scattering [24,25] Imaging [24,25] Temperature control Yes No Chemical identification SPR-Raman LSPR-Raman scattering Field portability No Yes Commercially available Yes No Cost US$150, ,000 US$5000 (Multiple particles) US$50,000 (Single nanoparticle) Spatial resolution ~10 µm 10 µm [85,90] One nanoparticle [24,25,91] Nonspecific binding Minimal (determined by surface chemistry and rinsing) [85 87,89,92] Minimal (determined by surface chemistry and rinsing) [2] Real-time detection Time scale = 10-1 to 10 3 s Planar diffusion [72,73,87,88,93] Time scale = 10-1 to 10 3 s, Radial diffusion [24] Multiplexed capabilities Yes [94,95] Yes - possible Small molecule sensitivity Good [72] Better [18] Microfluidics compatibility Yes Possible LSPR: Localized surface plasmon resonance; RIU: Refractive index unit; SPR: Surface plasmon resonance. and composition of the nanoparticles. This differs greatly from the nm decay length, or approximately 15 25% of the light s wavelength, for the SPR sensor [71]. The smallest footprint of the SPR and LSPR sensors also differ. In practice, SPR sensors require at least a 10 µm 10 µm area for sensing experiments. For LSPR sensing, this spot size can be minimized to a large number of individual sensing elements ( nanoparticles for a 2-mm spot size, nanosphere diameter = 400 nm) down to a single nanoparticle (with an in-plane width of ~20 nm) using single nanoparticle techniques [24]. The nanoparticle approach can deliver the same information as the SPR sensor, thereby minimizing its pixel size to the sub-100 nm regimen. Due to the lower refractive index sensitivity, the LSPR nanosensor requires no temperature control, whereas the SPR sensor (with a large refractive index sensitivity) does. The final and most dramatic difference between the LSPR and SPR sensors is cost. Commercialized SPR instruments can vary between US$150, ,000 (plus supplies), whereas the prototype and portable LSPR system costs less than US$5000 (plus supplies). However, a unifying relationship between these two seemingly different sensors is that both sensors overall response can be described using EQUATION 2 [71]: λ max m n 1 e ( 2d) l d ( = [ ]) where λ max is the wavelength shift response, m is the refractive index sensitivity, n is the change in refractive index induced by an adsorbate, d is the effective adsorbate layer thickness, and l d is the characteristic electromagnetic field decay length. It is important to note that for planar SPR sensors, this equation quantitatively predicts an adsorbate s affect on the sensor. When applied to the LSPR nanosensor, this exponential equation approximates the response for adsorbate layers but does not provide a fully quantitative explanation of its response [5,18]. Similar to the SPR sensor, the LSPR nanosensor s sensitivity was realized to arise from the distance dependence of the average induced square of the electric fields that extend from the nanoparticle s surface. This work provides important first steps towards the unified view of LSPR and SPR spectroscopies. Outlook & conclusions Briefly looking to the future, a reasonable extrapolation of the current data leads us to expect that by optimizing these sizeand shape-tunable nanosensor materials and by using single nanoparticle spectroscopic techniques, it will be possible to: Reach sensitivities of a few molecules, perhaps even a single molecule, per nanoparticle sensor element Reduce the time scale for real-time detection and the study of protein binding kinetics by two to three orders of magnitude, 533

8 Haes & Van Duyne since nanoparticle sensor elements will operate in a radial rather than planar diffusion mass transport regime Implement massively parallel bioassays for high-throughput screening applications while maintaining extremely low sample volume requirements Finally, it is pointed out that LSPR nanosensors can be implemented using extremely simple, small, light, robust, low-cost equipment for unpolarized, UV-vis extinction spectroscopy in transmission or reflection geometry. The instrumental simplicity of the LSPR nanosensor approach is expected to greatly facilitate field-portable environmental or point-of-care medical diagnostic applications. Expert opinion The use of nanoparticles for the highly sensitive and selective detection of biomolecules has been demonstrated with both model systems and complex human samples. Additionally, it has been shown that this assay can be clearly minimized to one nanoparticle via the implementation of dark field microscopy. The next challenge for sensor development lies in the capability of formatting the sensor into an array format and integrating this into a microfluidic chip. By improving the already established (and commercially available) SPR sensing device, this is foreseeable in the near future. The choice of nanoparticle size and shape will be critical for the optimization of the sensor s response. Future studies will continue to analyze complex biological samples and if fully realized, will provide vital information regarding disease understanding. Five-year view To date, most research in the area of nanoscience has been applied towards an understanding of the fundamental science involved rather than the implementation of successful applications in nanotechnology. During the next 5 years, this will change. The development of nanodevices, including nanosensors that are highly sensitive and selective (give low false positives and negatives), will provide a major improvement over current technologies. Instruments that provide high-throughput screening, for example, array imaging and identification of biomarkers (i.e., Raman tags), for drug discovery and disease diagnosis will uncover information vital to the understanding and treatment of disease that may lead to the design of better drug candidates for its treatment or prevention. These accomplishments will be met via the achievement of single molecule detection, incorporation of molecular identification using Raman or surface-enhanced Raman spectroscopy, implementation of in vivo sensing using labeled nanoparticles and introduction of microfluidics into the chip design. By overcoming the current obstacles in device development, this nanotechnology will become available to hospitals and pharmaceutical companies, as well as to any laboratory, at an affordable price Acknowledgements The authors acknowledge the support of the Nanoscale Science and Engineering Initiative of the National Science Foundation (NSF) under NSF Award Number EEC Any opinions, findings and conclusions or recommendations expressed in this material are those of the authors and do not necessarily reflect those of the NSF. AJ Haes also wishes to acknowledge the American Chemical Society Division of Analytical Chemistry and DuPont for a graduate fellowship. The authors are grateful for useful discussion, technical support and the expert assistance provided by Lei Chang, William Klein, Adam McFarland, George Schatz, Karen Shafer-Peltier and Shengli Zou. Key issues Silver nanoparticles are extremely sensitive to small changes in their surrounding dielectric environment. For this reason, their surfaces can be chemically functionalized and used to provide highly sensitive and selective detection of biological targets. These localized surface plasmon resonance (LSPR) nanosensors have been demonstrated as successful binding affinity sensors for the detection of proteins and antibodies and in the future, for DNA, RNA and peptide nucleic acids. The continued development of surface chemistries that provide high specificity of molecular interactions and a minimum of nonspecific interactions is required for the continued development of a chip-based LSPR sensor. Along these lines, the development of more precise (or more general) biomarkers for molecular interactions will contribute to this success. The implementation of microfluidic channels into chip design will greatly increase the capabilities of the chip s detection. Improvement of nanoparticle adhesion to the sensor substrate is vital for the long-term viability of the sensor. Current approaches provide sufficient nanoparticle adhesion but this should be improved prior to the development of a commercialized chip. To provide an affordable product for any laboratory, the integration of well plates with LSPR chips is being developed. Continued development of large arrays for multiplexed analysis and associated imaging modalities is vital for the widespread applicability of the LSPR nanosensor. 534 Expert Rev. Mol. Diagn. 4(4), (2004)

9 Localized surface plasmon resonance spectroscopy References Papers of special note have been highlighted as: of interest of considerable interest 1 Malinsky MD, Kelly KL, Schatz GC, Van Duyne RP. Chain length dependence and sensing capabilities of the localized surface plasmon resonance of silver nanoparticles chemically modified with alkanethiol self-assembled monolayers. J. Am. Chem. Soc. 123(7), (2001). Introduces the concept of refractive index-based sensing on ordered nanoparticle arrays. 2 Haes AJ, Van Duyne RP. A nanoscale optical biosensor: sensitivity and selectivity of an approach based on the localized surface plasmon resonance spectroscopy of triangular silver nanoparticles. J. Am. Chem. Soc. 124(35), (2002). Demonstrates that by detecting small changes in the dielectric environment of silver nanoparticle array substrates, they can be used as sensitive and selective affinity biosensors. 3 Haes AJ, Van Duyne RP. Nanosensors enable portable detectors for environmental and medical applications. Laser Focus World 39, (2003). 4 Riboh JC, Haes AJ, McFarland AD, Yonzon CR, Van Duyne RP. A nanoscale optical biosensor: real-time immunoassay in physiological buffer enabled by improved nanoparticle adhesion. J. Phys. Chem. B 107, (2003). Demonstrates that the localized surface plasmon resonance (LSPR) biosensor can be used (and reused) for immunoassays in solution. 5 Haes AJ, Zou S, Schatz GC, Van Duyne RP. A nanoscale optical biosensor: the long range distance dependence of the localized surface plasmon resonance of noble metal nanoparticles. J. Phys. Chem. B 108(1), (2004). Demonstrates that the sensitivity of the silver nanoparticle biosensor arises from the size-tunable average induced electric fields that surround the nanoparticles. 6 Haynes CL, Van Duyne RP. Nanosphere lithography: a versatile nanofabrication tool for studies of size-dependent nanoparticle optics. J. Phys. Chem. B 105(24), (2001). Reviews the technique of nanosphere lithography for the fabrication of nanoparticle arrays with distinct LSPR. 7 Mulvaney P. Not all that s gold does glitter. MRS Bulletin 26(12), (2001). 8 El-Sayed MA. Some interesting properties of metals confined in time and nanometer space of different shapes. Acc. Chem. Res. 34(4), (2001). 9 Link S, El-Sayed MA. Spectral properties and relaxation dynamics of surface plasmon electronic oscillations in gold and silver nano-dots and nano-rods. J. Phys. Chem. B 103(40), (1999). 10 Kreibig U, Gartz M, Hilger A, Hovel H. Optical investigations of surfaces and interfaces of metal clusters. In: Advances in Metal and Semiconductor Clusters. Duncan MA (Ed.), JAI Press, Inc., CA, USA, (1998). 11 Mulvaney P. Surface plasmon spectroscopy of nanosized metal particles. Langmuir 12(3), (1996). 12 Kreibig U. Optics of nanosized metals. In: Handbook of Optical Properties. Hummel RE, Wissmann P (Eds), CRC Press, FL, USA, (1997). 13 Jensen TR, Malinsky MD, Haynes CL, Van Duyne RP. Nanosphere lithography: tunable localized surface plasmon resonance spectra of silver nanoparticles. J. Phys. Chem. B 104(45), (2000). 14 Michaels AM, Nirmal M, Brus LE. Surface enhanced Raman spectroscopy of individual rhodamine 6G molecules on large Ag nanocrystals. J. Am. Chem. Soc. 121(43), (1999). 15 Schultz S, Smith DR, Mock JJ, Schultz DA. Single-target molecule detection with nonbleaching multicolor optical immunolabels. Proc. Natl Acad. Sci. USA 97(3), (2000). Introduces the use of protein-conjugated single gold nanoparticles for the ultrasensitive detection of various biological assays. 16 Yguerabide J, Yguerabide EE. Light-scattering submicroscopic particles as highly fluorescent analogs and their use as tracer labels in clinical and biological applications II. Experimental characterization. Anal. Biochem. 262, (1998). Demonstrates that silver and gold nanoparticles have a scattering efficiency equivalent to that of 10 6 fluorophores. 17 Schatz GC, Van Duyne RP. Electromagnetic Mechanism of Surface-Enhanced Spectroscopy. Chalmers JM, Griffiths PR (Eds), Wiley, NY, USA (2002). 18 Haes AJ, Zou S, Schatz GC, Van Duyne RP. Nanoscale optical biosensor: short range distance dependence of the localized surface plasmon resonance of noble metal nanoparticles. J. Phys. Chem. B 108(22), (2004). 19 Nie S, Emory SR. Probing single molecules and single nanoparticles by surfaceenhanced Raman scattering. Science 275(5303), (1997). 20 Kreibig U, Vollmer M. Cluster Materials. Springer-Verlag, Heidelberg, Germany (1995). 21 Taton TA, Lu G, Mirkin CA. Two-color labeling of oligonucleotide arrays via size-selective scattering of nanoparticle probes. J. Am. Chem. Soc. 123(21), (2001). Demonstrates that two different oligonucleotide-functionalized gold nanoparticle probes can be used to detect complementary sequences in solution. 22 Sonnichsen C, Franzl T, Wilk T et al. Drastic reduction of plasmon damping in gold nanorods. Phys. Rev. Lett. 88, / /4 (2002). 23 Bao P, Frutos AG, Greef C et al. High-sensitivity detection of DNA hybridization on microarrays using resonance light scattering. Anal. Chem. 74(8), (2002). Uses the plasmon resonance of noble metal nanoparticles to optically detect DNA in a microarray format. 24 McFarland AD, Van Duyne RP. Single silver nanoparticles as real-time optical sensors with zeptomole sensitivity. Nano. Lett. 3, (2003). Demonstrates that single silver nanoparticles can be used to detect less than 60,000 alkanethiol molecules in real time. 25 Van Duyne RP, Haes AJ, McFarland AD. Nanoparticle optics: sensing with nanoparticle arrays and single nanoparticles. Proc. Int. Soc. Opt. Eng. (SPIE) 5223, (2003). 26 Penn SG, He L, Natan MJ. Nanoparticles for bioanalysis. Curr. Opin. Chem. Biol. 7(5), (2003). Reviews the applications of nanoparticles for biomolecule detection. 27 Schultz DA. Plasmon resonant particles for biological detection. Curr. Opin. Biotechnol. 14(1), (2003). 28 Oldenburg SJ, Genick CC, Clark KA, Schultz DA. Base pair mismatch recognition using plasmon resonant particle labels. Anal. Biochem. 309(1), (2002). Identifies nucleotide polymorphs using the optical properties of nanoparticles and a DNA microarray format. 29 Connolly S, Cobbe S, Fitzmaurice D. Effects of ligand-receptor geometry and stoichiometry on protein-induced aggregation of biotin-modified colloidal gold. J. Phys. Chem. B 105(11), (2001)

10 Haes & Van Duyne 30 Storhoff JJ, Lazarides AA, Mucic RC et al. What controls the optical properties of DNA-linked gold nanoparticle assemblies? J. Am. Chem. Soc. 122(19), (2000). 31 Storhoff JJ, Elghanian R, Mucic RC, Mirkin CA, Letsinger RL. One-pot colorimetric differentiation of polynucleotides with single base imperfections using gold nanoparticle probes. J. Am. Chem. Soc. 120(9), (1998). Demonstrates that a single base mismatch in a polynucleotide target can be detected using nucleotide-modified 13-nm gold nanoparticles. 32 Elghanian R, Storhoff JJ, Mucic RC, Letsinger RL, Mirkin CA. Selective colorimetric detection of polynucleotides based on the distancedependent optical properties of gold nanoparticles. Science 227(5329), (1997). Demonstrates that the optical properties of gold nanoparticles can be used to monitor DNA hybridization. 33 Mirkin CA, Letsinger RL, Mucic RC, Storhoff JJ. A DNA-based method for rationally assembling nanoparticles into macroscopic materials. Nature 382(6592), (1996). 34 Hilger A, Cuppers N, Tenfelde M, Kreibig U. Surface and interface effects in the optical properties of silver nanoparticles. Eur. Phys. J. D 10(1), (2000). 35 Henglein A, Meisel D. Spectrophotometric observations of the adsorption of organosulfur compounds on colloidal silver nanoparticles. J. Phys. Chem. B 102(43), (1998). 36 Linnert T, Mulvaney P, Henglein A. Surface chemistry of colloidal silver: surface plasmon damping by chemisorbed iodide, hydrosulfide (SH-) and phenylthiolate. J. Phys. Chem. 97(3), (1993). 37 Kreibig U, Gartz M, Hilger A. Mie resonances. Sensors for physical and chemical cluster interface properties. Ber. Bunsen-Ges. 101(11), (1997). 38 Nath N, Chilkoti A. A colorimetric gold nanoparticle sensor to interrogate biomolecular interactions in real time on a surface. Anal. Chem. 74(3), (2002). Demonstrates that low concentrations of biomolecules can be detected on a surface modified with randomly attached gold nanoparticles. 39 Eck D, Helm CA, Wagner NJ, Vaynberg KA. Plasmon resonance measurements of the adsorption and adsorption kinetics of a biopolymer onto gold nanocolloids. Langmuir 17(4), (2001). 40 Okamoto T, Yamaguchi I, Kobayashi T. Local plasmon sensor with gold colloid monolayers deposited upon glass substrates. Opt. Lett. 25(6), (2000). 41 Himmelhaus M, Takei H. Cap-shaped gold nanoparticles for an optical biosensor. Sens. Actuators B 63(1 2), (2000). 42 Takei H. Biological sensor based on localized surface plasmon associated with surface-bound Au/polystyrene composite microparticles. Proc. Int. Soc. Opt. Eng. (SPIE) 3515, (1998). 43 Haes AJ, Van Duyne RP. A highly sensitive and selective surface-enhanced nanobiosensor. Mat. Res. Soc. Symp. Proc. 723, O3.1.1 O3.1.6 (2002). 44 Haes AJ, Van Duyne RP. Nanoscale optical biosensors based on localized surface plasmon resonance spectroscopy. Proc. Int. Soc. Opt. Eng. (SPIE) 5221, (2003). 45 Englebienne P. Use of colloidal gold surface plasmon resonance peak shift to infer affinity constants from the interactions between protein antigens and antibodies specific for single or multiple epitopes. Analyst 123(7), (1998). 46 Englebienne P. Synthetic materials capable of reporting biomolecular recognition events by chromic transition. J. Mat. Chem. 9(5), (1999). 47 Lee PC, Meisel D. Adsorption and surfaceenhanced Raman of dyes on silver and gold sols. J. Phys. Chem. 86(17), (1982). 48 Grabar KC, Freeman RG, Hommer MB, Natan MJ. Preparation and characterization of Au colloid monolayers. Anal. Chem. 67(4), (1995). 49 Sun Y, Xia Y. Synthesis and characterization of metal nanostructures with hollow interiors. Proc. Int. Soc. Opt. Eng. (SPIE) 5221(Plasmonics: metallic nanostructures and their optical properties), (2003). 50 Sun Y, Mayers B, Xia Y. Transformation of silver nanospheres into nanobelts and triangular nanoplates through a thermal process. Nano. Lett. 3(5), (2003). 51 Jin R, Cao YC, Hao E et al. Controlling anisotropic nanoparticle growth through plasmon excitation. Nature 425(6957), (2003). 52 Jin R, Cao YW, Mirkin CA et al. Photoinduced conversion of silver nanospheres to nanoprisms. Science 294(5548), (2001). 53 Mucic RC, Storhoff JJ, Mirkin CA, Letsinger RL. DNA-directed synthesis of binary nanoparticle network materials. J. Am. Chem. Soc. 120, (1998). 54 Taton TA, Mirkin CA, Letsinger RL. Scanometric DNA array detection with nanoparticle probes. Science 289(5485), (2000). Describes a method to analyze DNA arrays using oligonucleotidemodified gold nanoparticles and a flat-bed scanner. 55 Storhoff JJ, Elghanian R, Mirkin CA, Letsinger RL. Sequence-dependent stability of DNA-modified gold nanoparticles. Langmuir 18(17), (2002). 56 Hulteen JC, Van Duyne RP. Nanosphere lithography: a materials general fabrication process for periodic particle array surfaces. J. Vac. Sci. Technol. A 13, (1995). 57 Bain CD, Whitesides GM. Formation of two-component surfaces by the spontaneous assembly on monolayers on gold from solutions containing mixtures of organic thiols. J. Am. Chem. Soc. 110(19), (1988). 58 Hermanson GT. Bioconjugate Techniques. Academic Press, CA, USA (1996). 59 Lynch NJ, Kilpatrick RK, Carbonell RG. Aggregation of ligand-modified liposomes by specific interactions with proteins. II: biotinylated liposomes and antibiotin antibody. Biotechnol. Bioeng. 50, (1996). 60 Adamczyk M, Mattingly PG, Shreder K, Yu Z. Surface plasmon resonance (SPR) as a tool for antibody conjugate analysis. Bioconj. Chem. 10, (1999). 61 Facts: About Alzheimer s disease Alzheimer s disease and Related Disorders Association, Inc., MD, USA (2004). 62 Alzheimer A, Stelzmann RA, Schnitzlein HN, Murtagh FR. An English translation of Alzheimer s 1907 paper, Uber eine eigenartige Erkankung der Hirnrinde. Clin. Anat. 8(6), (1995). 63 Hardy JA, Higgins GA. Alzheimer s disease: the amyloid cascade hypothesis. Science 256(5054), (1992). 64 Klein WL, Krafft GA, Finch CE. Targeting small Aβ oligomers: the solution to an Alzheimer s disease conundrum? Trends Neurosci. 24(4), (2001). 536 Expert Rev. Mol. Diagn. 4(4), (2004)