Gold nanorod / nanosphere clustering by split-GFP fragment assembly for tunable near-infrared SERS detections

We describe the near-field properties of plasmonic nanoclusters made of gold nanorod (AuNR) and gold nanosphere (AuNS) colloids that are assembled using selfcomplementary split-green fluorescence protein (sGFP) fragments. Numerical modeling of the optical responses and field enhancement characteristics for these hybrid AuNR/AuNS heteroclusters were performed (i) as a function of AuNS binding locations along the edges or at the tips of AuNRs, (ii) as a function of the size and number of AuNS per AuNR, and (iii) as a function of cluster geometry and orientation with respect to the major polarization states of light. We show that near-infrared (NIR)-active plasmonic hot spots that provide large SERS enhancement factors of the vibrational fingerprints from the reconstructed GFP-chromophore are consistently obtained for longitudinally polarized-excitation of the AuNR/AuNS nanoassemblies. A set of clusters having sufficiently flexible geometry and good spectral resonance with traditional NIR laser excitations at 785 nm is proposed for NIR SERS detection of the GFP chromophore with enhancement factors in the range of 10-10 folds. This study provides grounds to improve the assembly of hot spot AuNR/AuNS SERS nanoprobes for NIR biosensing using GFP as a Raman reporter. © 2017 Optical Society of America OCIS codes: (160.4236) Nanomaterials; (250.5403) Plasmonics; (240.6695) Surface-enhanced Raman scattering; (310.6628) Subwavelength structures, nanostructures; (000.4430) Numerical approximation and analysis. References and links 1. C. L. Evans and X. S. Xie, “Coherent anti-stokes Raman scattering microscopy: Chemical imaging for biology and medicine,” Annu. Rev. Anal. Chem. 1(1), 883–909 (2008). 2. S. Keren, C. Zavaleta, Z. Cheng, A. de la Zerda, O. Gheysens, and S. S. Gambhir, “Noninvasive molecular imaging of small living subjects using Raman spectroscopy,” Proc. Natl. Acad. Sci. U.S.A. 105(15), 5844–5849 (2008). 3. J. H. Kim, J. S. Kim, H. Choi, S. M. Lee, B. H. Jun, K. N. Yu, E. Kuk, Y. K. Kim, D. H. Jeong, M. H. Cho, and Y. S. Lee, “Nanoparticle probes with surface enhanced Raman spectroscopic tags for cellular cancer targeting,” Anal. Chem. 78(19), 6967–6973 (2006). 4. T. Vo-Dinh, H. N. Wang, and J. Scaffidi, “Plasmonic nanoprobes for SERS biosensing and bioimaging,” J. Biophotonics 3(1-2), 89–102 (2010). 5. J. M. Romo-Herrera, R. A. Alvarez-Puebla, and L. M. Liz-Marzán, “Controlled assembly of plasmonic colloidal nanoparticle clusters,” Nanoscale 3(4), 1304–1315 (2011). 6. T. Chung, T. Koker, and F. Pinaud, “Split-GFP: SERS enhancers in plasmonic nanocluster probes,” Small 12(42), 5891–5901 (2016). 7. A. M. Funston, C. Novo, T. J. Davis, and P. Mulvaney, “Plasmon coupling of gold nanorods at short distances and in different geometries,” Nano Lett. 9(4), 1651–1658 (2009). 8. A. Gulati, H. Liao, and J. H. Hafner, “Monitoring gold nanorod synthesis by localized surface plasmon resonance,” J. Phys. Chem. B 110(45), 22323–22327 (2006). 9. K. S. Lee and M. A. El-Sayed, “Dependence of the enhanced optical scattering efficiency relative to that of absorption for gold metal nanorods on aspect ratio, size, end-cap shape, and medium refractive index,” J. Phys. Chem. B 109(43), 20331–20338 (2005). Vol. 7, No. 9 | 1 Sept 2017 | OPTICAL MATERIALS EXPRESS 3270 #301227 https://doi.org/10.1364/OME.7.003270 Journal © 2017 Received 29 Jun 2017; revised 11 Aug 2017; accepted 11 Aug 2017; published 17 Aug 2017 10. P. B. Johnson and R. W. Christy, “Optical constants of noble metals,” Phys. Rev. B Condens. Matter 6(12), 4370–4379 (1972). 11. J. Kumar, X. Wei, S. Barrow, A. M. Funston, K. G. Thomas, and P. Mulvaney, “Surface plasmon coupling in end-to-end linked gold nanorod dimers and trimers,” Phys. Chem. Chem. Phys. 15(12), 4258–4264 (2013). 12. P. F. Liao and A. Wokaun, “Lightning rod effect in surface enhanced Raman-scattering,” J. Chem. Phys. 76(1), 751–752 (1982). 13. A. McLintock, C. A. Cunha-Matos, M. Zagnoni, O. R. Millington, and A. W. Wark, “Universal surfaceenhanced Raman tags: individual nanorods for measurements from the visible to the infrared (514-1064 nm),” ACS Nano 8(8), 8600–8609 (2014). 14. C. Tabor, D. Van Haute, and M. A. El-Sayed, “Effect of orientation on plasmonic coupling between gold nanorods,” ACS Nano 3(11), 3670–3678 (2009). 15. Y. Wang, K. Lee, and J. Irudayaraj, “SERS aptasensor from nanorod-nanoparticle junction for protein detection,” Chem. Commun. (Camb.) 46(4), 613–615 (2010). 16. S. Pierrat, I. Zins, A. Breivogel, and C. Sönnichsen, “Self-assembly of small gold colloids with functionalized gold nanorods,” Nano Lett. 7(2), 259–263 (2007). 17. L. Shao, C. Fang, H. Chen, Y. C. Man, J. Wang, and H. Q. Lin, “Distinct plasmonic manifestation on gold nanorods induced by the spatial perturbation of small gold nanospheres,” Nano Lett. 12(3), 1424–1430 (2012). 18. A. F. Bell, X. He, R. M. Wachter, and P. J. Tonge, “Probing the ground state structure of the green fluorescent protein chromophore using Raman spectroscopy,” Biochemistry 39(15), 4423–4431 (2000). 19. M. G. Blaber and G. C. Schatz, “Extending SERS into the infrared with gold nanosphere dimers,” Chem. Commun. (Camb.) 47(13), 3769–3771 (2011). 20. T. Koker, T. Chung, and F. Pinaud, “Self-assembled split-FP/metal nanoclusters as Raman enhancers for molecular and cellular detection,” presented at the 251st American Chemical Society National Meeting, San Diego, CA, USA, 13–17 March 2016. 21. T. Koker, N. Tang, C. Tian, X. Wang, R. Martel, and F. Pinaud, Department of Biological Sciences, University of Southern California, Los Angeles, CA 90089, USA, are preparing a manuscript to be called “Targeted cell imaging by in-situ assembly and activation of hot spot SERS nanoprobes using split-fluorescent protein scaffolds,”. 22. S. H. Seo, B. M. Kim, A. Joe, H. W. Han, X. Chen, Z. Cheng, and E. S. Jang, “NIR-light-induced surfaceenhanced Raman scattering for detection and photothermal/photodynamic therapy of cancer cells using methylene blue-embedded gold nanorod@SiO2 nanocomposites,” Biomaterials 35(10), 3309–3318 (2014). 23. G. von Maltzahn, A. Centrone, J. H. Park, R. Ramanathan, M. J. Sailor, T. A. Hatton, and S. N. Bhatia, “SERScoded gold nanorods as a multifunctional platform for densely multiplexed near-infrared imaging and photothermal heating,” Adv. Mater. 21(31), 3175–3180 (2009).


Introduction
Molecular fingerprinting by Raman scattering holds great potential for single molecule detection and in vivo cellular imaging with visible and NIR excitations [1,2].However, the Raman cross-section of endogenous biomolecules is generally very weak, which impedes high sensitivity Raman detections in biological settings.To overcome this limitation, exogenous Raman nanoprobes with distinct chemical fingerprints can be designed for targeted bioimaging [3,4].These probes exploit the surface plasmons generated by noble metallic nanoparticles (NPs) to induce dramatic enhancements of the Raman cross-section of chromophore reporters by surface-enhanced Raman scattering (SERS).With SERS, the Raman signal intensities of reporters are remarkably increased, with enhancement factors of 10 2 -10 5 fold at the surface of individual NPs and factors of 10 8 -10 10 fold when reporters are positioned within small nanogaps between NP dimers or larger clusters [5].One critical requirement in the design of clustered nanoprobes is to control the size of the nanogap between NPs in order to achieve optimal and reproducible SERS enhancements of the exogenous reporters within plasmonic hot spots.A second imperative is to modulate the photonic responses of the clustered probes towards optical wavelengths that facilitate Raman imaging of biological samples, for instance in the NIR range where cell and tissue background signals are minimized.
Recently, we introduced the use of split-green fluorescent protein fragments (sGFP) as biocompatible Raman reporters and molecular glue to assemble gold nanospheres (AuNSs) into SERS nanocluster probes having 2 nm gaps seeded by the reconstruction of full length GFPs [6].Through numerical modeling we determined that SERS detection of the GFP chromophore imidazolinone/exocyclic C = C mode at 1530 cm −1 could be achieved with enhancement factors of about 10 9 folds for 40 nm AuNSs assembled into linear clusters of 4-10 AuNSs and for a resonant Raman excitation at 633 nm [6].Despite the large red-shift of the maximum near-field enhancement wavelength (λ max ) induced by dipolar plasmon mode couplings in these linear clusters, radiative damping along the AuNS chains results in spectral shift saturation of λ max at 630 nm and restricts the sensitive detection of the GFP chromophore Raman modes to the onset of the NIR spectra (701 nm).We therefore explored the use of other nanomaterials such as gold nanorods (AuNRs) [7,8] to extend and optimize SERS detection of the GFP chromophore in assembled nanoclusters further into the NIR.Indeed, compared to AuNSs, AuNRs have the distinctive advantage of providing wide spectral tunability from the visible to NIR range together with intense plasmonic enhancements due to their large optical cross-sections and convex tips [9].
Here, we assembled a variety of AuNR/AuNS hetero-nanoclusters using sGFP fragments and studied their photonic properties by finite-difference time-domain (FDTD) modeling with the aim of optimizing SERS responses in the NIR.Using experimental transmission electron microscopy (TEM) images for different sizes and shapes of clusters, together with localized mesh refinement computational techniques, we numerically modeled (i) how AuNS binding locations on AuNRs and (ii) how AuNRs/AuNSs orientations impact SERS enhancement factors and λ max under major polarization states of light for each type of cluster.We observed that AuNR/AuNS clusters excited by longitudinally polarized light enable robust NIR SERS responses and that some of these hetero-nanoclusters provide a broad range of assembly tolerance.Under optimal SERS amplification, AuNR/AuNS hetero-nanoclusters have SERS enhancement factors of ~10 9 and GFP seeded nanogaps consistently induce the formation of plasmonic hot spots.Through spectral mapping of each cluster, we propose a set of appropriate AuNR/AuNS assemblies that are suitable for high SERS enhancements of the GFP chromophore C = C vibrational modes at NIR wavelengths.

Numerical modeling and plasmonic properties of AuNRs
To study the nanoscale plasmonic properties of AuNRs and AuNSs assembled using split-GFP fragments, we used a previously described 3D-FDTD modeling environment [6] and verified that it adequately reproduced the photonic responses of AuNRs in aqueous buffers.Numerical modeling was performed on a Lumerical software package (Lumerical FDTD Solution 8.11) using a dielectric permittivity model of Au derived from Johnson and Christy [10].The simulation domain (1x1x1 μm in x, y, and z dimensions) was set with a global mesh size of 2 nm and a refractive index (RI) value of 1.34.GFPs at the surface of AuNRs (cylindrical rods with rounded tips) or AuNSs (spheres) were modeled as 2x4 nm cylindrical shapes having a RI value of 1.41 and an additional local mesh domain (maximum size of 1x1x0.4nm and minimum size of 0.25x0.25x0.25 nm in x, y, and z dimensions) was inserted inside the 2 nm global mesh domain to accurately model these GFP patches.The sizes and shapes of different AuNRs and AuNSs assemblies determined from TEM images (JEOL Jem-2100 (LaB6) microscope operated at 200kV) were used to physically model the shapes and organizations of the nanoclusters within the simulation domain.The nanogaps formed by the assembly of the sGFP fragments were systematically modeled with a 2 nm size, which corresponds to the transversal orientation of a complemented GFP between nanoparticles.
To validate this modeling environment, we first compared the ensemble far-field extinction spectra of 10x40 AuNRs (Aldrich, OD: 1.0, 0.4 nM) experimentally determined with a Beckman Coulter DU 800 spectrometer and the calculated far-field extinction spectra of a similar 10x40 nm individual AuNR having rounded tips with 1 nm radius and modeled from TEM images of AuNRs deposited on TEM grids (Ted Pella Carbon Type-B, 200 mesh, copper) (Fig. 1(a)).Taking into account the spectral widening due to the ensemble size distribution of AuNRs, the experimental extinction spectra was well reproduced by our modeling conditions, with a splitting of the surface plasmon resonance (SPR) into a transverse (~515 nm) and longitudinal mode (~820 nm) typical for anisotropically shaped AuNRs [11] (Fig. 1(b)).Focusing on the NIR active longitudinal mode, we then calculated the near-field electric enhancements and λ max for 10 nm wide AuNRs with lengths varying from 30 to 70 nm (Fig. 1(c)).Calculations were performed for excitations polarized parallel to the long axis of each type of AuNRs and near-field properties were measured at the AuNR tip, a position where lighting rod effects result in the strongest field enhancements [12,13].As shown in Fig. 1(c), the near-field enhancement increased with increasing length of the AuNRs, reaching its maximum for 50 nm-long AuNRs, but it decreased for rods with aspect ratio larger than 5:1 (length:width).On the other hand, λ max red-shifted linearly as a function of AuNR lengths, indicating that it is strongly associated with the localized SPR coupling distance along the AuNR major axis [14].In good agreement with previous observations [14], the slope of this linear red-shift is about 10, consistent with the ability to easily tune λ max over hundreds of nanometers through variation of the AuNR length by a few tens of nanometers.From these numerical investigations, 10 nm wide AuNRs with length between 40 and 50 nm therefore provide the strongest plasmonic near-field enhancements with λ max between 850 and 960 nm, well within the NIR range.

Edge-coupled AuNR/AuNS nanoclusters assembled by sGFP under longitudinally polarized excitation
To form AuNR/AuNS clusters by specific binding between split-GFP fragments appended to their surface, we functionalized CTAB-stabilized 10x40 nm or 10x50 nm AuNRs (Aldrich) with the large sGFP 1-10 fragments bearing a N-terminal tetracysteine tag and various sizes of citrate-stabilized AuNSs (10-40 nm, Sigma) with the smaller complementary M3 peptide fragments bearing an N-terminal cysteine [6].After 12 hours co-incubation of both functionalized AuNRs and AuNSs, a variety of hetero-nanoclusters formed in solution, some of which are presented in the TEM images of Fig. 2(a).As shown in these examples, the 10 nm AuNSs bind at various positions on the AuNRs, along their edge or close to their tip, forming AuNR/AuNS hetero-nanoclusters similar to those obtained with different surface chemistries [15,16].To understand the different optical properties of these diverse assemblies we systematically studied their near-field photonic responses for incident light longitudinally polarized along the major axis of AuNRs.
Starting with the first TEM image of Fig. 2(a) where 10 nm AuNSs are sequentially bound along the edge of a AuNR, we calculated how the number and binding locations of 10 nm AuNSs impact the near-field enhancement and the spectra λ max at the plasmonic hot spot closest to the tip of each clusters.As shown in Fig. 2(b), the near-field λ max is around 960 nm for a GFP-seeded hot spot formed by a 10 nm AuNS whose center is positioned 5 nm from the tip of a 50 nm long AuNR.This low energy peak is assigned to the bonding dipole-dipole mode between the AuNR and the AuNS [17].Although quadrupolar modes in AuNRs with small aspect ratios can be excited through near-field interactions with small AuNSs when coupling gaps are less than 1.5 nm [17], their influence was not obvious in the spectra because of the large aspect ratio of the 10x50 nm AuNR and the 2 nm gap size generated by the assembled split-GFP fragments.Negligible spectral variations in λ max at this hot spot is observed if additional 10 nm AuNSs are GFP-seeded along the AuNR edge with interparticle separations of 15 nm.However, the degree of electric field enhancement at the first hot spot decreases as additional 10 nm AuNSs are bound to the AuNR, consistent with the expected field interferences by the other hot spot coupling modes formed along the AuNR.Nonetheless, the presence of GFP at the nanogaps systemically provides higher near-field enhancement than for hollow nanogaps, as we have previously reported [6].
To study in more details the electric field spatial distributions within the GFP-seeded nanogaps of different 10 nm AuNSs coupled along AuNRs, cross-sectional field distributions of the clusters were provided in Fig. 2(c).The strongest hot spot intensity is observed for a AuNR with a single 10 nm AuNS close to its tip.While two hot spots are observed when an additional AuNS is added toward the center of the AuNR, this second hot spot shows a much smaller field enhancement that is not aligned with respect to the nanogap axis, but is spatially shifted towards the closest AuNR tip.When a third 10 nm AuNS is bound along the AuNR edge, a third hot spot is formed with a field distribution shifted towards the second AuNR tip and with a near-field coupling significantly stronger than for the central second AuNS.This indicates that the AuNR plasmonic properties strongly influence the field distribution in clusters formed with 10 nm AuNSs, with (i) strong field coupling observed for hot spots that are closest to the AuNR tips due to accumulated charges, and (ii) limited field coupling in the middle of the AuNR.These effects were quantified by calculating the respective SERS enhancement factors within a 2 × 2 × 2 nm volume over each hot spot for a cluster formed by the triple binding of 10 nm AuNSs along a 50 nm AuNR.As shown in Fig. 2(d), the highest SERS enhancement factor is observed for the AuNS closest to the AuNR tip with an enhancement of ~7.0x10 7 fold upon GFP-induced nanogap formation ( + 152% compared to the same size hollow gap).The central and the third hot spots, however, show significantly smaller SERS factors.In order to assess which edge-coupled AuNR/AuNS nanoclusters might provide the best global SERS response when illuminated by longitudinally-polarized light, the total cumulative SERS enhancement factors for clusters formed with AuNRs 40 or 50 nm in length and with 1-3 10 nm in diameter were computed and compared by adding the nearfield contribution from each hot spot.As shown in Fig. 2(e), a single 10 nm AuNS bound next to the tip of a 50 nm AuNR provides the highest SERS enhancement factor (8.8 × 10 7 fold), well within the ~10 7 fold enhancement range previously reported for similar AuNR/AuNS assemblies with 2 nm gap sizes [15].However, increasing the number of hot spots with more AuNSs does not lead to better global enhancements.Indeed, the second and third hot spots do not participate to an increase in total SERS enhancement factor, but in fact tend to reduce slightly field coupling at the strongest hot spot.For the 1:1 AuNR/AuNS dimer, the length of the AuNR plays an important role with regards to the total SERS enhancement factor, with a 3.9 fold increase in SERS enhancement for the 50 nm long AuNRs compared to the 40 nm long ones, consistent with our modeling of non-clustered AuNRs in Fig. 1(c).In addition to this significant increase in SERS enhancement, the 50 nm AuNR provides a more red-shifted λ max (960 nm) than the 40 nm AuNR (860 nm) in such AuNR/AuNS dimers as shown in Fig. 2(f).We note that, in both cases, λ max of the AuNR/AuNS dimers are identical to the λ max of the non-clustered AuNRs, further indicating that (i) the plasmonic properties of the clusters are largely dominated by the AuNR when dimers are formed with small 10 nm AuNSs and (ii) that the cluster near-field spectra can be easily tuned by simply changing the AuNR length in order to attain large SERS enhancements in the NIR.
To assess how changes in the size of AuNSs might further influence near-field coupling and spectra in 1:1 AuNR/AuNS dimers, we varied the diameter of the AuNS from 10 nm to 40 nm when it is positioned 5 nm away from the tip of a 50 nm AuNR.The near-field spectra and cross-sectional electric field distributions as a function of AuNS diameter are shown in Fig. 3(a-b).In the spectra of Fig. 3(a), the near-field λ max gradually red shifts as the AuNS diameter increases, consistent with the expected stronger plasmon/plasmon dipolar coupling at the GFP-seeded nanogap for the larger AuNSs.This indicates that, in addition to modulating λ max in the NIR by adjusting the length of the AuNR, additional fine tuning of λ max can be achieved by employing different sizes of AuNS, in good agreement with previously reported effects of AuNS sizes on the scattering spectra of AuNR/AuNS heterodimers [17].We also observed a gradual enhancement in electric field magnitude for a high energy but minor peak located at ~660 nm.This peak was assigned to a dipolequadrupole hybridized mode, which can be excited through near-field interactions between AuNSs and AuNRs [17].Its growing magnitude as a function of AuNS size reflects increased high order multipolar contributions from the AuNR due to the incremental asymmetry of the hetero-nanoclusters as the AuNSs become bigger.Interestingly, dimers formed with a 20 nm AuNS have the highest near-field enhancement at λ max .To understand why this is the case, the cross-sectional electric field distributions calculated at each near-field λ max are presented in Fig. 3(b).As can be seen, the electric field for the 20 nm AuNS is more confined at the GFP seeding position than for the 30 and 40 nm AuNSs, where the field appears distributed over a larger area around the hot spot.This effect is likely due to the greater local curvature of the 20 nm AuNS compared to bigger AuNSs, which provides a sharper apex at the interface with the AuNR edge and enhances the convergence of electromagnetic waves at the center of the nanogap.
Regarding the spectral tunability of edge-coupled AuNR/AuNS hetero-nanoclusters under longitudinally-polarized light, we note that binding geometry and number of AuNS per AuNR both have limited influence on the near-field λ max , has shown for a variety of 40/10 nm AuNR/AuNS clusters in Fig. 3(c).As discussed previously, this implies that a modulation of λ max is best achieved by employing different sizes of AuNRs and AuNSs to form clusters.On the other hand, the degree of near-field enhancement is strongly dependent (i) on the binding location of a given size AuNS to a AuNR (Fig. 3(c)), (ii) on the length of the AuNR itself, and (iii) on the size of the AuNSs employed.

Edge-coupled AuNR/AuNS nanoclusters assembled by sGFP under transversely polarized excitation
In addition to studying the near-field response of AuNR/AuNS clusters with longitudinally polarized excitation, we also examined their optical behaviors under transversely polarized light.As shown in Fig. 4(a), we numerically modeled the near-field spectra and the SERS enhancement of 40x10 nm AuNRs that are gap-coupled to different sizes of AuNSs via GFPinduced hot spot assembly at the center of the AuNRs.Considering the weak transverse plasmonic mode of individual AuNRs (inset of Fig. 4(b)), gap-induced near-field enhancements under transversely polarized light excitations are expected to be dominated by the AuNS and to scale with the AuNS size in the GFP assembled nanoclusters [6].We therefore assessed the effect of AuNS size on simple 1:1 AuNR/AuNS dimeric clusters by varying the AuNS diameter from 10 nm to 40 nm.As shown in Fig. 4(b), the near-field λ max systematically appears in the visible wavelength range for all the AuNS sizes tested, and can be assigned to the dipole-dipole hybridization between the transverse AuNR mode and that of the AuNSs.The degree of near-field enhancement under transversely polarized light increases substantially with the incremental AuNS diameters, confirming the dominant effect of the nanospheres.For these dimers, small near-field enhancements in the NIR were also observed under longitudinally-polarized light excitation, but those were weaker than for transverse polarization, because longitudinal excitations induce field enhancements that are much weaker at the AuNR center compared to the AuNR tips (Fig. 2(c)) [11].The near-field λ max of the transversely excited AuNR/AuNS clusters is near 630 nm and it slowly red-shifts by a few nanometers in response to increasing AuNS sizes, as shown in Fig. 4(c).This manifestation of increased plasmonic couplings between AuNR and AuNS with increasing AuNS size was confirmed by calculation of the SERS enhancement factors from the central hot spot formed in the AuNR/AuNS dimers.As seen in Fig. 4(d), under transverse polarization of the incident field, SERS enhancement factors increase significantly with increasing sizes of the AuNSs and are further amplified by the presence of GFP at the nanogap, as expected [6].These amplifications of the SERS enhancement factors are maximal for AuNR/AuNS GFP-seeded dimers formed with a 20 nm AuNS ( + 640%, Fig. 4(e)).As discussed in the case of edgecoupled AuNR/AuNS clusters excited by longitudinal polarization (Fig. 3(a-b)), the greater local curvature of the 20 nm AuNS allows efficient near-field enhancements at the interface with the AuNR edge.Yet, a controlled positioning of AuNSs toward the AuNR tip together with the use of longitudinal rather than transverse polarization of the excitation are required to maximize SERS enhancements.

Tip-coupled AuNR/AuNS clusters assembled by sGFP under longitudinally polarized excitation
After studying the near-field properties of edge-coupled AuNR/AuNS nanoclusters, we modeled the optical responses of tip-coupled AuNR/AuNS assemblies, which also formed upon sGFP fragment assembly of AuNRs with AuNSs as shown for the 40x10 nm AuNR and 40 nm AuNS heterodimer of Fig. 5(a).For this dimer, the calculated electric field distribution reveals a plasmonic hot spot at the interface between the AuNR left tip and the AuNS (Fig. 5(c)).Consistent with the fact that the strongest field enhancements in AuNRs take place at their tip [15], the electric field magnitude at this hot spot is much larger than the enhancements observed for the edge-coupled AuNR/AuNS clusters, with a NIR near-field λ max of 910 nm (Fig. 5(e)).A minor peak at ~610 nm corresponding to weakly excited dipolequadrupole hybridized modes in this asymmetric cluster is also present.
Because the lighting rod effect in AuNRs takes place at both tips, we also studied how having both tips coupled to AuNSs would affect the cluster near-field properties, as depicted in the nanodumbbell of Fig. 5(b).As expected, two plasmonic hot spots located at each GFPseeded nanogaps are observed when such nanodumbbell clusters are formed with two 40 nm AuNSs and a 40x10 AuNR (Fig. 5(d)).Both hot spots are symmetric and have similar nearfield properties, including a near-field λ max at 967 nm, which is red-shifted by 57 nm compared to the simpler, tip-coupled AuNS/AuNR dimers (Fig. 5(e)).Indeed, when nearfield modeling of the tip-coupled AuNS/AuNR/AuNS nanodumbbells are performed by modulating the size of AuNS at one of the two gap-coupled AuNR tips (Fig. 5(b,e)), a progressive red-shift of λ max towards the NIR, a broadening of the near-field λ max band and a gradual decrease of the dipole-quadrupole mode minor band are observed as a function of increasing AuNS size.The gradual red-shift of λ max reflects an increased dipolar coupling at the left-hand hot spot that stems from the enhanced polarizability of the nanodumbbell as the AuNS size on the opposite tips becomes bigger.This red-shift is accompanied by a broadening of the λ max band and a slight reduction in near-field enhancement due to larger radiative damping effects as the overall size of the nanodumbbell increases.However, the small lost in near-field enhancement is partially compensated by the increasing symmetry of the nanodumbbell, which results in better dipole-dipole coupling at the expense of dipolequadrupole coupling when the size of the right-hand AuNS is modulated from 10 to 40 nm.As seen in Fig. 5 (e), radiative damping in the assembled nanodumbbells generally results in slightly lower field amplifications at λ max within each hot spot compared to the tip-coupled AuNR/AuNS dimers.Yet, the cumulative effects of both hot spots lead to a significant increase in total SERS enhancement factor, notably as the nanodumbbell becomes more symmetric (Fig. 5 (f)).For instance, while the 3.34x10 8 fold SERS enhancement factor of tipcoupled 40/40 nm AuNS/AuNR dimers is only improved by 1.2 fold upon addition of a 10 nm AuNS at the opposite AuNR tip, formation of a symmetric 40/40/40 nm AuNS/AuNR/AuNS nanodumbbell provides a 5.3 fold increase in SERS factor with a total 1.78x10 9 fold SERS enhancement (Fig. 5 (f)).Interestingly, this value is equivalent to the total SERS enhancement factor we have previously reported for linear nanochains made of four AuNSs assembled by three GFP seeds (1.79x10 9 fold) [6].While such AuNS nanochains allow strong total SERS enhancement of three GFP chromophore at 633 nm [6], the symmetric 40/40/40 nm AuNS/AuNR/AuNS nanodumbbell provides the same level of enhancement with only two GFP-seeded hot spots and at a NIR-shifted resonance wavelength of 967 nm.Thus, in spite of potential complications to achieve tip-only functionalization of AuNRs with AuNSs, nanodumbbells assembled by split-GFP fragments are propitious to the design of clustered nanoprobes for high-sensitivity SERS detection of the GFP chromophore in the NIR spectral range.

AuNR/AuNS nanoclusters for NIR SERS detection of the GFP chromophore Raman signature at 785 nm excitation
To assess which AuNR/AuNS cluster would provide, theoretically, the highest NIR SERS enhancement of GFP Raman-active vibrational modes at each hot spot, we first compared the near-field λ max , its full width half maximum (FWHM) and the total SERS enhancement at λ max for a variety of GFP-assembled nanoclusters.As shown in Fig. 6 (a), these include a 40 nm AuNS homodimer, a 40x10/10 nm edge-coupled AuNR/AuNS heterodimer, a 40x10/40 nm tip-coupled AuNR/AuNS heterodimer and a 40/40x10/40 nm tip-coupled AuNS/AuNR/AuNS nanodumbell, all under longitudinally polarized light excitation.Comparisons were done with respect to the position of a popular NIR laser excitation wavelength at 785 nm (λ exc ) and with respect to the expected Stokes-shifted wavelength (892 nm) that is Raman scattered at 1530 cm −1 by the imidazolinone/exocyclic C = C vibrational mode of the GFP chromophore [18] (λ vib-GFP ).As seen in Fig. 6 (a), the λ max (589 nm) and the FWHM of the AuNS homodimer are clearly off resonance with λ exc and λ vib-GFP , which suggests a limited potential for NIR SERS enhancement of the GFP fingerprint despite a reasonable total SERS factor of 1.6x10 8 fold at λ max .In comparison, the edge-coupled AuNR/AuNS heterodimer has a λ max (860 nm) and a FWHM allowing better resonance with λ exc and λ vib-GFP , indicating that it is likely to provide good NIR enhancement at λ vib-GFP , in spite of a lower SERS enhancement factor than the AuNS homodimer at λ max .Both the tip-coupled AuNR/AuNS heterodimer and the AuNS/AuNR/AuNS nanodumbell have λ max (905 and 967 nm respectively) and FWHM that are largely red-shifted compared to λ exc .Although there are some spectral overlaps with λ vib- GFP , and their total SERS enhancement factor at λ max are high, the pre-resonance excitation of these clusters at 785 nm is unlikely to result in an efficient SERS enhancement of the GFP Raman signature.
To verify these initial assumptions, we calculated the total SERS enhancement factor of the GFP chromophore C = C vibrational mode for each nanocluster.The SERS enhancement was approximated as previously described [6,19] where, for each nanogap, E(ω exc )/E 0 (ω exc ) is the enhanced field at the 785 nm laser excitation and E(ω vib-GFP )/E 0 (ω vib-GFP ) is the enhanced field at the Stokes-shifted wavelength of the GFP chromophore C = C mode, here 892 nm.As seen in Fig. 6(b), the 1530 cm −1 GFP chromophore C = C mode is only weekly enhanced by 4.26x10 5 fold for the 40 nm AuNS homodimer, as predicted for a 785 nm excitation in post-resonance with respect to λ max .However, the GFP Raman fingerprint is significantly more enhanced with a factor of 1.37x10 7 fold for the edge-coupled AuNR/AuNS heterodimer because the 785 nm excitation is in partial resonance with the cluster λ max band.As anticipated, pre-resonance excitation of both the tip-coupled AuNR/AuNS heterodimer and the AuNS/AuNR/AuNS nanodumbell resulted in less efficient SERS enhancement of the GFP chromophore, with SERS enhancement factors of 3.92x10 6 fold and 7.51x10 5 fold respectively.These values are significantly lower than the expected SERS enhancements of 3.3x10 8 and 1.79x10 9 fold at λ max for the AuNR/AuNS heterodimer and the nanodumbell, respectively.Further red-shifted excitation, for instance with a 980 nm laser line, is likely to provide an increase in GFP SERS enhancement factor by 2-3 orders of magnitude for these nanoclusters.

Conclusion
We have shown that 10x40 or 10x50 nm AuNRs and various sizes of AuNSs can assemble into a variety of AuNR/AuNS heteroclusters using surface-appended complementary sGFP fragments as molecular glue.Our numerical modeling indicates that the reconstruction of a full GFP at the interface between AuNRs and AuNSs provides means to create plasmonic hot spots where plasmon/plasmon couplings between the two nanomaterials induce large nearfield and SERS enhancements in the NIR under appropriate polarized excitations.Indeed, while transversely polarized excitation of AuNR/AuNS clusters results in AuNS-dominated near-field responses in the visible spectral range, efficient near-field enhancements in the 800-1100 nm NIR range can be achieved when heteroclusters containing 40-50 nm long

Fig. 1 .
Fig. 1.FDTD modeling of gold nanorods.(a) TEM image of AuNRs (40 nm length and 10 nm width).Scale bar: 50 nm.(b) Comparison between the experimental extinction spectrum for a solution of 40x10 nm AuNRs and the calculated extinction spectrum of a single AuNR.(c) Plasmonic near-field maximum wavelength and the degree of near-field electric enhancement as a function of AuNR length from 30 nm to 70 nm for a fixed 10 nm AuNR width.Incident electric field is 1 (V/m).The near-field maximum wavelength red-shifts linearly as a function of AuNR length with a slope of about 10 (red curve).

Fig. 2 .
Fig. 2. Near-field responses of edge-coupled AuNR/AuNS nanoclusters illuminated by longitudinally polarized light.(a) TEM images of AuNR/AuNS clusters assembled with sGFP fragments.Scale bars: 20 nm.(b) Near-field spectra as a function of the number of 10 nm AuNSs bound to a 50 nm AuNR.(c) Cross-sectional field enhancement distribution for 50 nm AuNRs clusters with 1-3 AuNSs.(d) Local SERS enhancement factor at each hot spot for a 50 nm AuNR with three 10 nm AuNSs having GFP-seeded or hollow nanogaps.(e) Comparison of total SERS enhancement factors for clusters formed with AuNRs 40 or 50 nm in length and with 1-3 AuNSs 10 nm in diameter.(f) Near-field spectra comparison for AuNR/AuNS clusters with 40 or 50 nm AuNRs in the presence or absence of one 10 nm AuNS and for GFPseeded or hollow nanogaps.

Fig. 3 .
Fig. 3. Spectral tunability of edge-coupled AuNR/AuNS clusters formed by split-GFP fragment assembly under longitudinally polarized excitation.(a) maximum wavelength spectra as a function of AuNS diameter.(b) Cross-sectional electric field distribution as a function of AuNS diameter.(c) Near-field spectra and enhancements within GFP-seeded hot spots for various AuNR/AuNS hetero-nanoclusters formed with a 40x10 nm AuNR and different 10 nm AuNSs.Inset arrows are monitoring point for near-field spectra.

Fig. 4 .
Fig. 4. Near-field responses of edge-coupled AuNR/AuNS hetero-nanoclusters under transversely polarized excitation.(a) TEM images of a AuNR/AuNS dimers and corresponding schematic of dimeric AuNR/AuNS clusters with varying AuNS diameter.Scale bars: 10 nm.(b) Near-field wavelength spectra calculated at GFP-seeded plasmonic hot spot for edgecoupled AuNR/AuNS clusters as a function of AuNS size and of light polarization.Inset: Near-field spectrum of a 40x10 nm AuNR.(c) Near-field maximum wavelength as a function of AuNS diameter.(d) SERS enhancement factor as a function of AuNS diameter and GFP seeding for heterodimers formed with a 40 nm AuNR.(e) Percentage in additional SERS enhancement induced by GFP seeding in AuNR/AuNS dimers as a function of AuNS size.

Fig. 5 .
Fig. 5. Tip-coupled AuNR/AuNS clusters under longitudinally polarized excitation.(a) TEM image of a tip-coupled AuNR/AuNS dimer formed by a 40x10 nm AuNR and a 40 nm AuNS.Scale bar: 40 nm.(b) Schematic of a 40/40x10/40 nm AuNS/AuNR/AuNS nanodumbbell, an extended cluster of the tip-coupled dimer in (a).(c) Calculated cross-sectional electric field distribution of the tip-coupled dimer in (a) at a near-field wavelength of 905 nm.Scale bar: 40 nm.(d) Calculated cross-sectional electric field distribution of the gold nanodumbbell in (b) at the maximum near-field wavelength of 967 nm.Scale bar: 40 nm.(e) Near-field wavelength spectra monitored at the left-hand GFP-seeded plasmonic hot spot in gold nanodumbbells when the size of the left-hand AuNS is fixed at 40 nm and the size of the right-hand AuNS is changed from 10 nm to 40 nm.(f) Total SERS enhancement factor from both hot spot in 40/40x10/y AuNS/AuNR/AuNS nanodumbbells as a function of right-hand AuNS diameter.

Fig. 6 .
Fig. 6.SERS detection of the GFP chromophore fingerprints for different AuNR/AuNS nanoclusters under longitudinally polarized excitation at 785 nm.(a) Comparison of the nearfield properties for a variety of GFP-seeded clusters.The position of a 785 nm laser line (red line) and that of the Stokes-shifted wavelength scattered by the chromophore imidazolinone/exocyclic C = C vibrational mode (red dash) are shown.(b) Comparison between the total SERS enhancement factor of the GFP chromophore C = C mode and the total SERS enhancement factor at λ max for different AuNS/AuNS and AuNR/AuNS nanoclusters.