Recent progresses in optical microscopic techniques remarkably benefit bio-imaging applications. For example, by means of confocal microscopy,1, 2 stimulated emission depletion microscopy,3, 4 stochastic optical reconstruction microscopy5, 6 and others,7 these techniques enable to retrieve three-dimensional images and even to reconstruct sub-wavelength resolutions beyond the diffraction limit. Among these cutting-edge optical microscopic techniques aforementioned, a critical step requires fluorescent labeling, which is often detrimental to live cells and more critically, could affect the physiology of the cells by means of mechanotransduction. Therefore, surface plasmon resonance microscopy (SPRM), a label-free technique that images the refractive index variation of the local dielectric environment situated in the vicinity to the metal film,8–10 promises a solution to investigate the effects of biophysical stimuli exerted on cells and how cells respond to such cues in a real-time fashion. Although label-free and extremely sensitive, the SPRM still encounters several intrinsic issues–for example, the demand of optical couplers including prisms and gratings, limited operation frequency ranges typically within visible and the foremost shallow detection distances within a couple of hundreds of nanometers to impede intracellular investigation.11
As a consequence, to meet the requirement of label-free, coupler-free, scalable and intracellular bio-imaging, here we present a plasmonic microscopic platform by employing multi-mode resonances in spilt-ring resonators (SRRs). The SRRs are artificially constructed sub-wavelength structures, which allows negative magnetic permeability,12 high-frequency magnetism13 and other unprecedented electromagnetic properties14–16 based on their collective plasmonic resonances. In fact, the resonance condition of the SRR significantly depends on their local dielectric environment, so that the SRRs can be readily employed as refractive-index (RI) sensors.17–23 Recently, we further manifested the multi-mode plasmonic resonances in the SRRs, in which the lower-order modes possess greater sensitivity associated with stronger localized electromagnetic field leading to shorter detection lengths within five hundreds nanometers, whereas the higher-order modes present mediate sensitivity with micron-scale detection lengths to allow intracellular bio-events detection.23 These unique characteristics of the SRR structure not only enable a multi-functional plasmonic biosensor to preserve the merits of the conventional SPR technique (e.g., label-free, excellent sensitivity, quick and real-time diagnose, detection of refractive index variations), but further promise to achieve a coupler-free, scalable and intracellular bio-imaging platform.
The designed SRR samples were fabricated by standard e-beam lithographic and lift-off processes as shown in Figure 1a, in which one sample contains 10 × 10 unit cells, and each unit cell consists of 5 × 5 SRRs. All SRR unit cells contain exactly identical SRR pattern from cell to cell. The parameters of designed SRRs are shown in Figure 1b. Besides, transmission and reflection were characterized by a Fourier-transform infrared spectrometer (Vertex 80V) equipped with an infrared microscope (Bruker Hyperion 2000) in the wavenumber range of 400-8600 cm−1, and the corresponding mid-IR images were captured by a focal-planar-array (FPA) detector. All measured spectra have been normalized with respect to the reflection spectra of an aluminum mirror. The reflection spectrum of the SRRs at normal incidence presents three reflectance peaks, as shown in Figure 1c, in which the resonance wavelengths are consistent with the standing-wave plasmonic resonance (SWPR) model,24
where L is the total length of an SRR defined in Figure 1b, λm is the resonance wavelength, m is the resonance mode, neff is the effective refractive index of the dielectric environment and λo depends on the geometric structure. Notice the multiple reflectance peaks are labeled as two sets of plasmon resonance modes: 1L, 3L (asymmetry cases) and 2L (symmetry case) with respect to two orthogonal E-field polarizations (EL and EL) as shown in the inset of Figure 1c. We observe that the perpendicularly polarized wave (EL) excites odd modes (i.e., mode 1L at 2550 cm−1 and mode 3L at 6900 cm−1), and the horizontally polarized wave (EL) excites even modes (i.e., mode 2 at 5700 cm−1), respectively. Next we mapped the fundamental plasmonic image of the SRR by using a focal planar array (FPA) detector, and the collected plasmonic image about the SRR sample by evaluating the reflectance intensity from 2200 to 2800 cm−1 is displayed in false color, as shown in the upper panel of Figure 1d. We find that the spectral signal emerges stronger within the SRR region due to the fundamental plasmonic resonance from the SRR, which serve as a pixel array in the infrared region. To observe the contrast of plasmonic images, a thin layer of PMMA was spin-coated on the SRR samples and the result is shown in the lower panel of Figure 1d. By comparing with these two panels in Figure 1d, we successfully detected an obvious contrast of the plasmonic images, which stems from the refractive index variation of the local dielectric environment on this SRR platform.
To further demonstrate label-free and intracellular plasmonic imaging, we cultured and collected human bone marrow-derived mesenchymal stem cells (hMSCs),25, 26 and then grew the collected hMSCs on the top of the SRR substrate. The hMSCs grown on the SRR substrate exhibited both the typical flat and spindle-shaped morphology, similar to what were observed on culture flasks. Such fibroblast-like morphology remaining unchanged after cells fixation, indicates the bio-compatibility of the SRR substrate. The detailed process of incubating the hMSCs is described in the section of Experimental.
The signals of biochemical components in hMSCs mainly locate at the wavenumber of 600-1800 cm−1 (amide group) and 2800-3200 cm−1 (lipid group), as shown in Figure 2a. Notice that here we carefully designed the SRR samples to avoid overlapping their operation frequencies with the intrinsic absorption signals of the biochemical components. For examples, by manipulating the dimension of the SRR,24 we can control the fundamental resonance of the SRR samples with the hMSCs cultured atop within the wavenumber 1800-2400 cm−1, which does not overlap with the functional group signals of hMSCs as shown in Figure 2b. In fact, the signals of hMSCs' main biochemical components such as proteins, carbohydrates and waxes/lipids of hMSCs can be identified by their characteristic absorption, in which the frequencies of these intrinsic signals are fixed and their intensities are weak. Yet, for bio-sensing and bio-imaging applications, one requires shiftable signals for quantitative analysis and certainly stronger signals for better performance. Now by employing our scalable SRR platform to probe the refractive index variation of the analyte, we can secure the shifted and stronger (∼10 fold greater than lipid's absorbance signal) plasmonic signals as these biochemical components appear in the vicinity of the SRR substrate. In short, differentiating with characterizing absorption signal, the plasmonic resonance signal resting on the SRR demonstrated the advantages of stronger signal intensity, scalable resonance position, and early stage detection.27
Both the reflection spectra of the fundamental mode about the bare SRR sample and the SRR sample with the cultured hMSCs atop are shown in Figure 2c, labeled by blue and green curves, respectively. The result responds a significant red shift in reflectance peaks due to variation of dielectric environment, indicating that this SRR plasmonic sensor is very sensitive beyond LSPR sensors.23 The sensitivity (i.e., Δλ/Δn) for the 1st resonance mode of our designed SRRs plasmonic biosensor is about 2700 nm RIU−1. In comparison with other refractive index (RI) biosensors such as surface plasmon polariton (SPP) biosensors and localized surface plasmon resonance (LSPR) biosensors, our designed SRR plasmonic biosensor possesses comparable or even better performance, for example, the sensitivity of prism coupler-based surface plasmon polariton (SPP) biosensors in wavelength interrogation ranges from 970 to 13800 nm RIU−1, depending on the resonance wavelength,28, 29 which is comparable with our SRR plasmonic biosensor. Nevertheless, these SPP biosensors require optical couplers and present shallow detection distance (typically shorter than a couple of hundred nanometers), so that their applications in bio-imaging turn to be limited; besides, the sensitivity of localized surface plasmon resonance (LSPR) biosensors is from 120 nm RIU−130 to 500 nm RIU−1,31 which performs much worse than the demonstrated SRR plasmonic biosensor. This SRR plasmonic biosensor performs similar to the hybridization of surface plasmon polariton (SPP) and localized surface plasmon resonance (LSPR)–the former is non-radiative and owns better sensitivity, the latter is radiative and possesses poor sensitivity. For instance, the lower-order modes resemble SPP, which are more sensitive but with shorter sensing depths of sub-micron scales due to the non-radiative nature of SPP; in contrast, the higher-order modes favor LSPR to demonstrate less sensitivity yet greater sensing distances up to micron scales due to the radiative nature instead.23 Such a significant change of resonance peaks due to the local attachment of cells has not been well studied in the metamaterial community. We believe that our biomolecular sensing measurements demonstrate the possibility for sensitive detection and correlation between the peak shift and the length in vertical direction from the SRR surface to inner structure of cells.
Finally, we used our proposed platform to construct bio-image of hMSCs based on the fundamental resonance signal (1L mode) of SRR at the wavenumber of 1850-2400 cm−1 that fits in the detection range of the FPA detector (900-3600 cm−1). Both conventional optical microscopic and confocal optical microscopic images were also presented as controlled comparison. Figure 3a shows the conventional optical microscopic image of hMSCs grown on the SRRs samples and the black part in the background refers to the SRRs structure. In this case, we cannot reveal any detail of the inner nucleus and organelles without the labeling process. Figure 3b shows the confocal fluorescent optical microscopic image of hMSCs, in which we can observe the nuclei of hMSCs, the purple part labeled by 4′-6-diamidino-2-phenylindole (DAPI). Nevertheless, such a labeling process is typically expensive and time consuming, impeding the practical application of real-time diagnosis. Besides, for the biological reaction sensitive to the three-dimensional structure of biomolecules, the labeling process of introducing fluorescent markers to the target will radically affect the analytic result.
As a consequence, we construct an intracellular image of hMSCs by the SRR platform instead, which does not require the labeling process aforementioned, but directly detects the change of plsamonic resonance of the SRR fluctuated by the local attachment of the targeting bio-agents. The result was shown in Figure 3c, displaying the plasmonic image of hMSCs cultured on the SRRs sample by evaluating the reflectance intensity within the wavenumber 1850-2400 cm−1. Clearly, we observe evident intracellular contrast resting on the refractive index distribution of hMSCs in our system. For example, the nucleus, a membrane-bound organelle that is densely comprised of nucleic acids and proteins associated with a substantially higher refractive index than the surrounding cytoplasm,32 can be thus easily visible as presented in red corresponding to the greatest shift of resonance frequencies and the strongest reflection intensity. Besides, the other colored parts mainly refer to the cytoplasm, corresponding to the smaller shift of resonant frequencies stemming from the lower refractive index of the cytoplasm.28 In short, the morphologic observation by the label-free SRRM exhibits the similar result to the confocal fluorescent optical microscopy image (as shown in Figure 3b). Note that we cannot identify individual organelles within the cytoplasm of hMSCs either, which is limited by the resolution of our FPA detector and the diffraction of infrared. The spatial resolution can be significantly improved by employing asymmetrically coupled SRRs structures to enhance the Q-factor of the plasmonic resonance.33, 34
In summary, we present a first-ever intracellular plasmonic imaging by exciting multi-mode resonances in spilt-ring resonators. Human bone marrow-derived mesenchymal stem cells (hMSCs) are the target to be observed in our platform. Our study successfully demonstrate the feasibility of using SRRM for constructing the refractive index distribution of hMSCs to achieve intracellular bio-imaging platform and meanwhile, obtaining the information of functional groups from the target cells. The demonstrated SRRM possesses the key advantages beyond other optical microscopy such as label-free and real-time diagnosis (vs. fluorescent and Raman scattering techniques), coupler-free to avoid the issues of coupling oil leakage and dispersion, great detection lengths (vs. SPP techniques), and scalable operation frequencies (vs. LSPR techniques) in particular in IR regimes to prevent strong absorption from bio-agents, providing the possibility for the live cells imaging technique, including the observation of cellular proliferation and differentiation process.