J.
Prinz
a,
C.
Heck
abc,
L.
Ellerik
a,
V.
Merk
c and
I.
Bald
*ab
aInstitute of Chemistry, University of Potsdam, Karl-Liebknecht-Str. 24-25, 14469 Potsdam, Germany. E-mail: bald@uni-potsdam.de
bBAM Federal Institute for Materials Research and Testing, Richard-Willstätter Str. 11, 12489 Berlin, Germany
cDepartment of Chemistry + SALSA, Humboldt-Universität zu Berlin, Brook-Taylor-Str. 2, 12489 Berlin, Germany
First published on 11th February 2016
DNA origami nanostructures are a versatile tool to arrange metal nanostructures and other chemical entities with nanometer precision. In this way gold nanoparticle dimers with defined distance can be constructed, which can be exploited as novel substrates for surface enhanced Raman scattering (SERS). We have optimized the size, composition and arrangement of Au/Ag nanoparticles to create intense SERS hot spots, with Raman enhancement up to 1010, which is sufficient to detect single molecules by Raman scattering. This is demonstrated using single dye molecules (TAMRA and Cy3) placed into the center of the nanoparticle dimers. In conjunction with the DNA origami nanostructures novel SERS substrates are created, which can in the future be applied to the SERS analysis of more complex biomolecular targets, whose position and conformation within the SERS hot spot can be precisely controlled.
One of the main advantages of DNA origami templates is the possibility of arranging functional units, e.g. gold nanoparticles (AuNPs), with nm precision, which makes it a predestined technique for the study of plasmonic effects11,12 and surface-enhanced spectroscopies.
Surface-enhanced spectroscopies are based on the enhancement of the electromagnetic field close to metal nanoparticles (NPs) upon excitation of their surface plasmon resonance. This can result in enhanced optical signals such as fluorescence and Raman scattering. Particularly high field enhancement can be generated in the gap between adjacent NPs due to a coupling of the individual surface plasmon resonances.13,14 Thus, to benefit from highest field enhancements it is required to control the position of the AuNPs with respect to the analyte molecules.
Depending on the distance to the plasmonic nanostructure a fluorescent dye can be subject to fluorescence quenching15 or fluorescence enhancement.16 At close distance to the NP surface analyte molecules can also be detected by surface-enhanced Raman scattering (SERS).17–20 SERS is a particularly interesting technique since the Raman signal can be enhanced by many orders of magnitude, which renders the detection of single molecules possible.21,22
Apart from using ultralow analyte concentrations different approaches aiming at the detection of SERS from single molecules have been developed. On the one hand, the bi-analyte method has been introduced by Le Ru et al.,23 which is based on the measurement of a mixture of two substances with distinguishable SERS spectra. Subsequently, single-molecule events have to be asserted by a statistical analysis. The bi-analyte method has been improved by isotopic labelling of dyes resulting in spectral shifts of certain SERS peaks.24 On the other hand, dimers of Au–Ag-core–shell NPs with tailored gap size have been used for single-molecule SERS. Suh et al. presented gap-tailored Au–Ag core–shell nanodumbbells providing enhancement factors (EFs) of the order 1012 as well as gold nanobridged nanogap particles generating EF values between 108–109 for about 90% of the enhancing sites.25,26 In both cases the gap sizes were in the range of 1 nm and the EFs were high enough for the detection of single Cy3 dyes.
The use of DNA origami structures as scaffolds for SERS active nanostructures is particularly attractive due to their versatility with respect to further functionalization. Additionally, the DNA origami technique represents a bottom-up approach, which allows for the production of a large number of plasmonic nanostructures at once. Such processes are much more cost-effective than widely used top-down lithography methods.27 Here, we present DNA–AuNP hybrids that are optimized in various respects in order to increase the SERS sensitivity to a single-molecule level. To the best of our knowledge, this is the first study that combines the DNA origami technique with SERS to detect single molecules. Furthermore, we have estimated EFs for selected nanostructures through direct correlation of AFM and Raman images.
Here, we present structures based on triangular DNA origami substrates that are functionalized with AuNP dimers. The DNA origami substrates are folded during a hybridization process between the M13mp18 ssDNA scaffold strand and 208 suitable ssDNA staple strands.1 By modification of certain staple strands with a capture sequence that protrudes from the DNA triangle anchor points for AuNPs are introduced (see Experimental section for details). Fig. 1 illustrates the attachment process of AuNP dimers to the DNA origami template. Two different strategies (Fig. 1a and b) are pursued that differ in terms of the positions of AuNPs as well as analyte molecules (carboxytetramethylrhodamine (TAMRA), cyanine 3 (Cy3); molecular structures shown in Fig. 1c). In both strategies the attachment process is realized via DNA hybridization between the ssDNA capture sequences (5′-(AAA)8T4-3′) and the ssDNA coating strands (5′-(TTT)4T-SH-3′) covering the AuNPs. In strategy (a) two 40 nm AuNPs are attached to one side of the DNA origami template by three anchor points per particle resulting in structure 1. In that case the AuNPs are covered with a TAMRA-modified sequence (5′-(TTT)4TX-SH-3′; X = TAMRA) and a non-TAMRA-modified thiolated DNA strand used as a spacer to reduce the TAMRA concentration on the AuNP surface. In contrast, in strategy (b) one single analyte molecule (TAMRA or Cy3) is incorporated into the DNA origami template by modification of one staple strand. In the following step two 60 nm AuNPs are attached to different sides of the DNA origami template by four anchor points per particle resulting in structure 3a (TAMRA) or 3b (Cy3). The different DNA–AuNP hybrids (structures 1, 3a, 3b) are further modified by electroless silver deposition yielding DNA-Au–Ag-core–shell hybrids (structures 2, 4a, 4b). For subsequent Raman and AFM studies the hybrid structures are adsorbed on Si substrates.
Single-particle SERS measurements are performed by correlation of AFM images and Raman maps as illustrated in Fig. 2. For SERS experiments a confocal Raman microscope and a 532 nm excitation laser is used resulting in resonant excitation of TAMRA or Cy3. AFM (Fig. 2, blue frame) and SERS images (Fig. 2, red frame) are superimposed by means of a marker on the Si surface. This approach allows for a direct assignment of SERS signals to specific DNA–AuNP hybrid structures. In this way the effect of NP structure modification on the SERS enhancement is revealed.
In Fig. 3 correlated AFM images and SERS spectra for single DNA origami substrates functionalized with a 40 nm AuNP dimer covered with TAMRA-modified DNA (structure 1) are shown. The maximum gap size of approximately 14 nm between the two individual particles is estimated by considering the number of DNA bases located in the DNA origami between the two center positions of the particles (see Fig. S1 in the ESI† for details). Within the remaining structural flexibility of the AuNPs connected to the DNA origami platforms the AuNPs might approach each other slightly during the drying process due to a temporarily increased salt concentration. Thus, the minimal gap size is determined by the DNA coating which is assumed to be 2.5 nm per AuNP. In Fig. 3a AFM images of three individual and well-defined DNA–AuNP hybrids (i, ii, iii) are presented (height profiles shown in Fig. S2 (ESI†)). The corresponding SERS spectra are shown in Fig. 3b (cyan spectra). In all three cases the most prominent TAMRA band at 1652 cm−1 has been detected with an intensity of approximately 5–10 cts (highlighted in grey). For comparison a reference spectrum obtained from single 60 nm AuNPs covered with TAMRA-modified DNA is shown, which contains the characteristic TAMRA bands at 1222, 1360, 1507, 1531, 1568, 1596 and 1652 cm−1 (black spectrum). Since single 40 nm AuNPs covered with TAMRA are found to give extremely weak SERS signals (see Fig. S3 (ESI†)) under the presently applied conditions we conclude that the detected signal for structures i–iii mainly arises from TAMRA molecules located in the hot spot formed in-between the two particles. To estimate the number of TAMRA molecules that contribute to each of the three SERS spectra we assume that one 40 nm AuNP is covered with 430 oligonucleotides.28 Furthermore, half of the DNA coating strands are modified with a TAMRA molecule and we assume about 10% of all coating strands being located in the hot spot. Therefore, a maximum of approximately 40 TAMRA molecules contribute to each SERS spectrum (i–iii) presented in Fig. 3b. It should be emphasized that among all correlated hybrids with structure 1 only those revealing the strongest SERS signals are presented. Thus, the dimers i–iii are expected to have smaller gap sizes than 14 nm.
To further improve the Raman signal enhancement an additional Ag layer is grown on the AuNPs which is expected to result in an increase of the electromagnetic field enhancement for two reasons: (a) Ag exhibits a better enhancement performance in the visible range of the electromagnetic spectrum,29 (b) the gap size between the two NPs is reduced upon the silver shell growth. In order to compare the effect of the silver shell, SERS and AFM data of selected nanostructures have been collected before and after the silver enhancement process (Fig. 4). In Fig. 4a an AFM image of one representative hybrid structure with silver shell is shown. The DNA origami triangle is most probably hidden beneath the Au–Ag core–shell structure in the AFM image. Since the vertical resolution in AFM images is higher than the lateral resolution by a factor of up to 100 the shell thickness can be determined from the height difference in the associated cross sections. The height profile shown in Fig. 4b indicates a thickness of the silver shell of approximately 10 nm in vertical direction (see height profiles of analogue dimers without silver shell in Fig. S2 (ESI†) for comparison). In Fig. 4c the corresponding SERS spectra are shown demonstrating an overall increase of the SERS intensity after electroless silver deposition (grey spectrum). In addition to the bands at 1360 cm−1 and 1652 cm−1 two other characteristic spectral features for TAMRA at 1507 cm−1 and 1531 cm−1 become visible (highlighted in grey). Principally, the effect of photobleaching is observed upon consecutive laser exposures under the here applied conditions. Therefore, the detected SERS signals after silver deposition are expected to result even from a smaller number of TAMRA molecules compared to the initial measurement. Although this experiment demonstrates an increase of the SERS intensity upon silver enhancement the origin of the SERS signals in structure 2 is difficult to reveal since the dye molecules are completely embedded in the silver shell. This may lead to charge transfer processes between Ag and the TAMRA molecules resulting in possible contributions from chemical enhancement. Moreover, with regard to the silver shell thickness of approximately 10 nm also DNA strands from the AuNP coating as well as from the DNA origami template might be embedded in the silver shell. Although the Raman cross section for TAMRA is considerably higher than the Raman cross sections for DNA for some hybrids clear SERS bands arising from the DNA can be detected (see Fig. S4 and Table S1 (ESI†) for an example). This observation is ascribed to two effects: on the one hand, the number of individual DNA bases incorporated in the silver shell is significantly higher in comparison to the number of TAMRA molecules. On the other hand, the two AuNPs are fused together upon silver shell growth resulting in a rod-like plasmonic particle. This can also be seen in scanning electron microscopy (SEM) images shown in Fig. S5a.† Since metal rods are known to provide the maximum electromagnetic enhancement at the particle tips30 new hot spots are created that provide better SERS enhancement for the DNA located at the end of the tips than for the embedded TAMRA molecules. Consequently, core–shell systems with reduced silver shell thicknesses are necessary in order to reduce the amount of embedded DNA in the shell as well as to avoid the relocation of hot spots. To improve the control of the shell size another silver enhancement kit (HQ silver, Nanoprobes) was used for subsequent experiments since it is characterized by a thickening agent to retard the deposition rate.
To explore the suitability of DNA origami based SERS substrates for the detection of single molecules the design was optimized taking the following aspects into account (structures 3 and 4; strategy shown in Fig. 1b): first, placing the single dye molecule with high accuracy in the hot spot is a crucial condition since the EF decreases strongly within distances of a few nm away from the NP surface.31 This condition is fulfilled by attaching the two individual AuNPs to opposite sides of the DNA origami and placing the dye molecule into the axis in between the AuNPs. Moreover, in this arrangement the initial gap size is reduced to 7 nm (assuming 2 nm thickness for the DNA origami template and 2 × 2.5 nm thickness for the DNA coating surrounding the AuNPs). Second, finite-difference time-domain (FDTD) calculations revealed that Au–Ag-core–shell NP dimers (2.5 nm Ag shell and 2 nm gap size) with 60 nm Au cores show a superior electromagnetic field enhancement compared to 40 nm or 80 nm Au cores (see Fig. 5c and S7 (ESI†)). Consequently, the initial size of the AuNPs was increased to 60 nm and a thin Ag coating was added, which further reduces the gap (see Fig. S5b and S5c†). Third, an additional (-fourth) anchor point per AuNP was introduced to the DNA origami template in order to reduce the flexibility of the AuNP attachment position.
In Fig. 5a correlated Raman spectra and AFM images for structure 4a (TAMRA as analyte molecule) are presented. Initially, an experiment with two TAMRA molecules placed into the gap has been performed (dimer i). In the corresponding AFM image the triangular DNA origami is no longer visible which is valid for all hybrids with structure 3 or 4. This is due to the fact that the AuNP dimers are immobilized on the Si surface, while the DNA triangle can no longer lie flatly on the surface resulting in a twisted arrangement of the hybrid structures (see AFM image of a corresponding dimeric structure with two 40 nm AuNPs in Fig. S6†). The corresponding SERS spectrum for dimer i (light cyan spectrum) exhibits a weak band at 1652 cm−1 (highlighted in grey) which originates from a maximum of both TAMRA molecules located in the hot spot. Furthermore, the same peak was detected for only one TAMRA molecule incorporated in the DNA origami substrate (dimer ii). Additionally, a second weak band that is assigned to TAMRA is visible at 1507 cm−1 (also highlighted in grey). These experiments demonstrate that the detection of TAMRA at a single-molecule level is possible using the optimized hybrid structure 4a. However, the use of TAMRA suffers from very low SERS intensities (with signal-to-noise ratios of 3.6–6.3 for the strongest TAMRA signals) and the fact that the detection is mainly based on the presence of a single peak at 1652 cm−1.
In order to confirm the single-molecule detection of the analyte molecule the same experiment was repeated with one Cy3 molecule incorporated into the DNA–Au–Ag hybrid structure instead of TAMRA (structure 4b, Fig. 5b). The comparison of both single-molecule SERS spectra reveals that most spectral peak positions coincide for dimer iii and iv. All these bands at 1167, 1354, 1384, 1470, 1495 and 1588 cm−1 (highlighted in grey) are also visible in the reference (black spectrum) and thus can be assigned to Cy3. Some spectral features matching the reference spectrum are only present either in the SERS spectrum of dimer iii or in the one for dimer iv (marked with dashed lines). In the case of dimer iii (light blue spectrum) an additional band at 1124 cm−1 as well as overlapping bands in the range between 1270–1310 cm−1 have been detected, which arise also from the Cy3 molecule. The band at 1553 cm−1 is only visible as a shoulder in the reference spectrum. Another reference spectrum for Cy3, which confirms the presence of that band can be found in the ESI (Fig. S8†). In contrast, the SERS spectrum for dimer iv (dark blue spectrum) reveals a spectral feature at 1230 cm−1 arising from Cy3 that is only very weakly present in the spectrum of dimer iii. The observed differences between the recorded spectra (absence or presence of certain bands as well as intensity ratios) are ascribed to slight conformational variations in every individual structure. On the one hand, the position of the dye relative to the axis of the dimer is certainly different for every hybrid structure, which might lead to preferential enhancement of specific bands. On the other hand, the size ratio between Au core and Ag shell slightly differs which has been reported to have an effect on the plasmon coupling.32 The high background in the SERS spectrum of dimer iv (Fig. 5b, dark blue line) which is visible for some nanostructures is most likely caused by some residues from the Ag enhancement solution. In summary, the experiment shows that using Cy3 as analyte molecule gives a clear evidence of single-molecule SERS for two reasons: (a) the intensity of the SERS bands is considerably higher than for TAMRA, (b) the detection of a single Cy3 molecule is based on at least six bands (detected for both dimers iii and iv).
In order to find out whether single-molecule SERS sensitivity can be expected for the structures 4a and 4b EFs have been estimated, which are generally based on a comparison of the SERS signal with that of the same molecule in normal Raman conditions.31 Recording a normal Raman spectrum of TAMRA and Cy3 at 532 nm is hampered by a strong fluorescence background. Thus, the characterization of a reference system with known EF is necessary. Since SERS spectra of a single 60 nm AuNP covered with dye containing DNA (sequence: 5′-(TTT)4TX-SH-3′; X = TAMRA or Cy3) show clear bands with an acceptable signal-to-noise ratio this system was chosen to serve as a reference, whose EF was reported previously.33 First, the number of dye molecules per NP has been determined using a fluorescence based approach following the protocol of Hurst et al.34 Second, the average SERS intensity for a single AuNP has been determined by correlating AFM and SERS data for approximately 15 TAMRA and 15 Cy3 labeled NPs. In the case of TAMRA the intensities of the band at 1652 cm−1 have been averaged whereas for Cy3 the bands at 1470 cm−1 and at 1588 cm−1 serve as references. Third, the EF for a single 60 nm AuNP was assumed to be 9 × 105 according to Hong and Li.33 Relating the EF from the literature to the reference system and in turn to all individual hybrids shown in Fig. 5 reveals an estimated EF for every single dimer i–iv (see Fig. S9 and S10 (ESI†) for detailed information). All relevant values used for the EF estimation are summarized in Table 1. It turned out, that the EFs for the Cy3 containing systems (dimer iii and iv) are of the order of 1010, that is one order of magnitude higher than the EFs for the systems with TAMRA (dimer i and ii). This can be explained by a higher Raman cross section of Cy3 in comparison to TAMRA within the hot spot.
TAMRA | ||||
---|---|---|---|---|
System | I (1652 cm−1)/cts | Nr. TAMRA/system | I (1652 cm−1)/cts (per TAMRA) | EF |
60 nm AuNP [TAMRA] (reference) | 28 | 4805 ± 556 | 5.83 × 10−3 | 9 × 105a |
Dimer i | 3 | 2 | 1.5 | 2 × 108 |
Dimer ii | 3 | 1 | 3.0 | 5 × 108 |
Cy3 | ||||
---|---|---|---|---|
System | I (1470 cm−1)/cts | Nr. Cy3/system | I (1470 cm−1)/cts (per Cy3) | EF |
60 nm AuNP [Cy3] (reference) | 30 | 3745 ± 672 | 8.01 × 10−3 | 9 × 105a |
Dimer iii | 119 | 1 | 119 | 1 × 1010 |
Dimer iv | 130 | 1 | 130 | 1 × 1010 |
In order to confirm the experimentally estimated EFs FDTD simulations for a single 60 nm AuNP (Fig. 5c, top) as well as for a 60 nm AuNP dimer (2.5 nm Ag shell, gap size 2 nm) (Fig. 5c, below) have been performed assuming excitation with 532 nm. In the case of the dimer the area with EFs in the range of 107 or higher is expanded over approximately 5 nm in y direction reaching a maximal EF of 1010. Thus, the localization of a single dye molecule incorporated into the DNA origami structure within the hot spot is highly probable. Moreover, the experimentally estimated EFs coincide with the simulated ones. However, the EFs revealed by FDTD simulations are only based on the electromagnetic enhancement and do not consider additional effects such as the contribution due to resonant excitation of the analyte which can result in an increase of the intensity up to 5 orders of magnitude.35 On the other hand, the simulation is based on the excitation with light polarized along the dimer axis. Since in the experiment non-polarized light was used the simulation overestimates this contribution. Additionally, the experimentally estimated EFs are based on the EF from ref. 33 which has been determined using a 647 nm laser for excitation. Since the EF is directly correlated to the excitation wavelength the listed values only represent a first approximation. However, it should be noted that for 532 nm excitation the electromagnetic field enhancement is higher (see FDTD simulations in Fig. S11 (ESI†) for 647 nm excitation) resulting in a total EF of a single 60 nm AuNP that is about 10 times higher. Fig. S12† shows the simulated absorption and scattering spectra of the Au–Ag core–shell nanostructures, which exhibit maximum cross sections at 549 nm and 561 nm, respectively, corresponding roughly to the excitation laser used in the current experiment (532 nm).
The novel SERS substrates presented here are highly promising for biosensing applications. For instance, single proteins or DNA strands can be placed in the hot spot with high local precision in order to investigate e.g. protein folding43 or DNA strand break events. Furthermore, the plasmon resonance of the Au–Ag-core–shell NPs can be easily tuned by adjusting the size ratio between core and shell32 which enables the creation of tailored nanostructures for individual applications. Extending the dimeric structures by an additional NP might result in even higher EFs, especially in the case of so called nanolenses consisting of three differently sized AuNPs with specific size ratios.44,45 Consequently, single-molecules with much lower Raman cross-sections might become detectable using DNA–NP hybrid structures in the future.
Footnote |
† Electronic supplementary information (ESI) available: Additional information about materials and methods, designs of DNA origami templates, height profiles, additional SERS spectra, assignment of DNA bands, SEM images, additional AFM images, FDTD simulations, additional reference spectra for Cy3 and detailed description of EF estimation, simulated absorption and scattering spectra. See DOI: 10.1039/c5nr08674d |
This journal is © The Royal Society of Chemistry 2016 |