1. Introduction

Metal nanoparticles (NPs) are attracting attention as optical biomarkers because they exhibit significant advantages over alternative markers. Metal NPs are brighter than chemical fluorophores and quantum dots [1], they are very stable and biocompatible [2,3], and they do not suffer from photobleaching [1,2]. Furthermore, the use of metal NPs as biomarkers benefits from two unique properties: their large scattering cross section, which creates high contrast and permits easy localization, and the strong dependence of their scattering and absorption spectra on the refractive index of their surroundings. These properties make NPs ideal for the spectral interrogation of various biological interactions [1,4–6] and facilitate their use as multi-functional tags which indicate both the location and environment of the target of interest.

One exciting application of metal NP biomarkers is in the development of tools for cancer diagnostics. Loo, et al. [3] demonstrated the use of immunoconjugated NPs as contrast agents for breast cancer. El-Sayed, et al. [2] used anti-EGFR conjugated NPs to bind cancerous epithelial cells. Their research indicates that extinction spectra are broader for cells incubated with non-conjugated NPs, as compared to the conjugated NPs, suggesting that the NPs’ resonance spectrum may be an indicator of binding to the EGFR target. Nanoparticle biomarker studies to date have used NPs either simply for localization, disregarding the NP spectral features [1,3,7], or have measured the extinction spectrum for a large collection of NPs [2], foregoing their localization capabilities. However, the ability to both image and spectrally measure single NPs is of great interest for realizing the potential of metal NP tags. Darkfield microscopy allows such single NP studies, and an epi-illumination darkfield scheme permits NP tagging studies of live cells in culture.

Existing darkfield illumination techniques impose significant sample restrictions. Trans-illuminated darkfield microscopy introduces limits on the sample thickness, requires transparent samples, and restricts access to the sample, which complicates sample manipulation. Furthermore, oil immersion condensers required for high-NA objectives are not compatible with imaging cells in culture. Epi-illumination darkfield objectives designed for surface analysis are currently available, but these are incompatible with index-matching fluids, thus having limited numerical aperture, and high-magnification versions deliver little light through a substrate. Furthermore, this approach requires specialized objectives, which can add significant expense to an existing system. Finally, evanescent wave darkfield methods provide sample illumination only within a small depth range, on the order of a wavelength from the substrate.

In this paper, we present a darkfield scheme that does not suffer from these limitations, permitting the imaging and spectral analysis of NPs in a variety of environments, including cells in culture. We first present a validation of the epi-illumination scheme by analyzing silver nanospheres in various refractive index environments, and characterize the darkfield illumination distribution. We then apply the scheme to the spectral analysis of non-conjugated and anti-EGFR conjugated gold nanospheres interacting with cancer cells in culture.

2. Epi-illumination micro-spectroscopy system

Our darkfield illumination scheme utilizes the microscope’s objective for both illumination and imaging of the sample. Broadband light is delivered through a 200 µm diameter fiber and collimated by a short focal length lens (Fig. 1). The collimated light is converted by a refractive axicon into a ring of light converging at a specified angle (1.3°). The converging light passes through a long focal length (f=140mm) lens and then through a beamsplitter, producing a ring of light at the back focal plane of the microscope objective. The light emerges at the sample side of the objective as a collimated ring of light converging at a high angle. While most light is transmitted through the sample, some is reflected at the glass substrate of the sample chamber. The reflected light returns to the objective at the same angle at which it emerged, so the reflected light re-forms a ring at the objective’s back focal plane. This ring of reflected light continues to diverge as it passes through the beamsplitter and is blocked by a field stop. Although the field stop limits the effective NA of the objective, the reduction of NA to permit darkfield imaging is a trade-off common to the use of trans-illuminated darkfield condensers with high-NA objectives. Of the portion of light delivered to the sample, some is backscattered by the NPs. This backscattered light passes through the center of the field stop to be imaged.

 

Fig. 1. Epi-illumination darkfield scheme showing the incident, reflected, and scattered light paths in relation to the microscope objective and darkfield illumination optics.

Download Full Size | PDF

To enable the study of NPs in various immersion or sample media, we have designed and constructed a high-resolution, high-magnification microspectroscopy system that allows the imaging and spectroscopy of NPs under darkfield illumination[8]. The foundation of our system is a Zeiss Axiovert 200 inverted microscope with both 40× dry Plan-neofluar^® (Zeiss) and 100× oil immersion Plan-neofluar^® (Zeiss) objectives. Illumination for the epi-illumination scheme is provided by an external 250 W xenon source (Oriel).

3. Epi-illumination micro-spectroscopy system validation and characterization

The performance of the epi-illumination microspectroscopy system was validated by measuring 80 nm silver nanospheres in various refractive index environments. The NPs were bound to a silanated glass substrate, and the NP side of the substrate was exposed to either air, water, or index-matching oil. In a previous analysis[8] of this type of sample under transillumination darkfield, we found that a weighting factor can provide an effective refractive index for the NP environment which accounts for the substrate and accurately predicts spectral shifts due to changes in media refractive indices. We apply this weighting factor to the spectra acquired with our epi-illumination scheme in order to validate the spectral results.

Darkfield images of the sample in air, water, and oil are presented in Fig. 2. An illumination artifact is observed in the image of the sample in air, but is negligible in the images of the sample in water and oil. This artifact is the result of undesired curvature at the tip of the axicon, which is due to a limitation in the manufacturing process. The curvature causes the axicon tip to refract light over a range of angles, rather than at just the specified angle. When passed through the objective, the light refracted from the tip is focused to the center of the image, and the reflected portion of this light is imaged back through the objective. We have reduced the effects of this artifact by introducing a 1 mm diameter field stop preceding the axicon to prevent light from reaching the axicon tip. This approach diminishes the artifact, but also decreases illumination at the center of the image and introduces haze from diffraction at the edge of the field stop. Both the variation in illumination and the illumination artifact can be characterized and accounted for during analysis, as discussed below. While other solutions to the illumination artifact are being developed, the illumination scheme in its current form provides the desired darkfield illumination over the majority of the field of view (FOV), while providing the advantages outlined above for an epi-illumination darkfield scheme.

 

Fig. 2. Darkfield images of 80 nm Ag NPs in: (a) air, (b) water, and (c) index-matching oil environments. (d) Resonance spectra for the same single NP in each refractive index environment. (e) Experimental versus predicted SPR peak positions as a function of refractive index environment for a weighting factor of 0.58.

Download Full Size | PDF

The resonance spectra for an individual representative NP are presented in Fig. 2(d). Source-correction was performed by using the spectra from adjacent slit positions at which there are no NPs. As shown in Fig. 2(e), the spectral peaks correspond to those expected for an 80 nm particle when using our previously-determined substrate weighting factor of 0.58[8]. The weighting factor determines an effective refractive index, n_eff, according to the following equation: n_eff=α·n_medium+(1-α)·n_substrate, where α is the weighting factor, n_medium is the immersion medium index, and n_substrate is the substrate index. The effective refractive index may be used in standard Mie theory calculations to model NPs on substrates.

As mentioned above, the current epi-illumination scheme exhibits an illumination artifact and reduced illumination at the center of the FOV. These effects may be characterized by mapping the intensity of the darkfield illumination, which is the light that may be imaged only after scattering from the sample. In order to map the intensity of darkfield illumination, images of two standard samples are collected with the system: the image of a spectrally flat, uniformly-scattering standard, and the image of a mirror, which serves as a perfectly reflecting sample. The difference of the two sample measurements provides a map of the darkfield illumination intensity, I_d, according to the following equation: I_d=M_s-M_m=(I_d+I_r)·S-I_r. This may be understood by considering that the light incident on each sample is a combination of darkfield illumination, I_d, and artifact, or reflected illumination, I_r. The scattering standard returns both types of light, whereas the mirror returns only the artifact illumination. As shown in the equation above, when the scattering coefficient, S, is 1, the difference of these two images, M_s-M_r, provides the desired map of darkfield illumination intensity. The resultant illumination intensity map for the whole FOV (shown in Fig. 3(a)) and a cross-section of the illumination intensity corresponding with the slit position of the imaging spectrograph (shown in Fig. 3(b)) reveal that the normalized illumination intensity remains within a factor of 2 of the maximum intensity over 77% of the FOV and 66% of the slit.

 

Fig. 3. (a) Map of darkfield illumination intensity, and (b) plot of illumination intensity variation along line of spectral acquisition.

Download Full Size | PDF

4. Cell study

In order to demonstrate the capabilities of our epi-illumination system for studying NP interactions with cells in culture, we present a preliminary study to image and spectrally analyze anti-EGFR conjugated NPs interacting with cultures of live A-431 cells, which are derived from a human epithelial carcinoma and exhibit significantly elevated EGFR expression [9]. Images and spectra of cells incubated with anti-EGFR conjugated NPs are compared with those of cells incubated without NPs and cells incubated with non-conjugated NPs.

We implemented an antibody conjugation protocol for 60 nm Au spheres which we modified from protocols published for 12 nm [7] and 30 nm [2] Au spheres. For this protocol, 60 nm diameter Au colloid (Ted Pella, Inc.) was diluted in 20mM HEPES buffer at a ratio of 1 mL colloid : 125 µL HEPES. Separately, anti-EGFR (Sigma) was diluted in 20mM HEPES buffer to prepare a 3% (v/v) anti-EGFR solution. 1 mL of the 3% anti-EGFR was added to 1.125 mL of the diluted colloid and allowed to conjugate at room temperature for 20 minutes. In order to verify antibody attachment, 100 µL of the resulting conjugated colloid was removed and mixed with 10 µL of 10% NaCl. After verifying that the NaCl did not produce a color change in the colloid (a color change indicates NP aggregation due to incomplete antibody coverage), 200 µL of 1% PEG was added to the remaining ~2 mL of conjugated NPs and allowed to interact for 10 min. At the end of this interaction period, the solution was centrifuged at 6000 RPM for 30 min. The supernatant was then withdrawn, and the NP pellet was re-suspended in 1 mL of phosphate buffered saline.

A-431 cells were incubated at 37°C, 5% CO₂ in chambered coverglasses (Lab-Tek) with media prepared for A-431 cells (90% Dulbecco’s modified Eagle’s medium, 10% fetal bovine serum, supplemented with 1% penicillin streptomycin). Each chamber received cells suspended in 0.5 mL media. The chambers then received either an additional 1.0 mL media, or 0.5 mL media and 0.5 mL non-conjugated 60 nm Au NPs. The chambers were incubated overnight. After the incubation period, both sets of cells were analyzed on the microspectroscopy system. Then, 0.3 mL of the conjugated NP solution was added to the chamber of cells in pure media, and the cells were incubated for an additional hour. After this incubation period, the media on the cells was withdrawn, the cells were washed with 1.5 mL fresh media, the media was again withdrawn, and 1.5 mL fresh media was again added. The cells were then returned to the incubator for 1 additional hour before analysis on the microspectroscopy system.

 

Fig. 4. Darkfield images of (a) cells incubated in pure media [1s acquisition time], (b) cell incubated in media with non-conjugated NPs [0.4s acquisition time], and (c) cells incubated with anti-EGFR conjugated NPs [0.4s acquisition time]. Note the increased scattering intensity due to the NPs and the more distinct boundaries for the cells incubated with anti-EGFR NPs, which are expected to localize preferentially on cell surfaces where EGFR is expressed.

Download Full Size | PDF

5. Results and discussion of cell study

Images of the cells in each incubation condition are shown in Fig. 4. The image of cells in pure media was acquired with an integration time of 1 s, while the images of cells with NPs were acquired with an integration time of 0.4 s. These images provide two important results. First, the cells in pure media scatter very little light in comparison with the cells incubated with NPs. This validates the assumption that the colorful scattering of the cells with NPs is from the NPs themselves, rather than from subcellular structures. Secondly, cell boundaries are more distinct in cells with conjugated NPs relative to those with non-conjugated NPs. This observation is likely explained by the fact that the conjugated NPs are bound to the cell surface, which provides a distinct boundary, while the non-conjugated NPs are dispersed throughout the cytoplasm.

In order to determine if the NPs are colocalized with the cells as expected, we took advantage of epi-illumination’s compatibility with dual-mode imaging. Although a number of secondary types of measurement may be performed in conjunction with epi-darkfield, we introduced concurrent brightfield illumination to demarcate cell boundaries. As shown in Fig. 5, the contributions from darkfield and brightfield may be independently controlled, to provide either full darkfield (Fig. 5(a)), full brightfield (Fig. 5(c)), or a mixture of the two (Fig. 5(b)). These images indicate that the NPs are colocalized with the cells and illustrate a powerful advantage of epi-illuminated darkfield microscopy.

Figure 6 presents spectra acquired from the cells with non-conjugated NPs (shown in Fig. 4(b)) and conjugated NPs (shown in Fig. 4(c)) using our epi-illumination darkfield system. The spectrum from one position in the cell incubated with non-conjugated NPs is presented in Fig. 6(a). The spectrum is complex, such that the NP resonance is not readily apparent. The origin of this spectral complexity is under investigation, but may be due to a number of factors, including multiple-NP interactions and interference effects. Upon averaging all of the spectra acquired from a line passing through a cell, as shown in Fig. 6(b), the NP resonance becomes clearer. After acquiring and averaging acquisitions from multiple lines through the cell, the NP resonance becomes even more apparent, as shown in Fig. 6(c). One feature of interest in Fig. 6(c) is the rise in the scattering intensity at long wavelengths. To investigate this phenomenon, spectra from cells in pure media were acquired, as shown in Fig. 6(d). The average spectra from such cells show the same feature, indicating that the feature is not due to the NPs themselves. Therefore, the spectra of cells in pure media may be subtracted from the spectra of cells with NPs to isolate the NP contribution to the spectra. The result of this baseline correction for a representative cell incubated with non-conjugated NPs is shown in Fig. 6(e), and that of a cell incubated with conjugated NPs is shown Fig. 6(f). Note that for the cell incubated with non-conjugated NPs, each spectral acquisition and the final NP spectrum are presented, while for the cell incubated with conjugated NPs, only the final NP spectrum from the same analysis is presented.

 

Fig. 5. Demonstration of dual-mode imaging with epi-illuminated darkfield for registration of NPs within cells. Each image is a single acquisition of the microscope FOV, under different illumination conditions, and not an overlay of images.

Download Full Size | PDF

 

Fig. 6. Spectra of cells acquired with microspectroscopy system. (a) The spectrum from a single site in a cell incubated with non-conjugated NPs is complex, and the resonance peak is not apparent. (b) The resonance peak becomes more apparent when the spectra from a single acquisition are averaged. (c) The peak is clearly seen when 6 acquisitions from the same cell are averaged. (d) The ramp in scattering intensity toward longer wavelengths is present in cells without NPs and may therefore be subtracted from the spectra of cells with NPs to isolate the scattering spectrum due to the NPs. The results of this baseline correction are displayed for (e) a cell incubated with non-conjugated NPs and (f) a cell incubated with conjugated NPs.

Download Full Size | PDF

A broader spectrum is seen for the non-conjugated NPs upon comparison of the baseline-corrected spectra from the cells incubated with non-conjugated (Fig. 6(e), FWHM=185 nm) vs. conjugated (Fig. 6(f), FWHM=127 nm) NPs. The observation of broadening in the unconjugated case is consistent with the findings of El-Sayed, et al. [2] for extinction spectra of NPs incubated with cancer cells. Additionally, the resonance peaks for the non-conjugated (Fig. 6(e), λ_peak=594 nm) and conjugated cases (Fig. 6(f), λ_peak=575 nm) differ, revealing a shift toward longer wavelengths in the non-conjugated case. Although these data are preliminary and must be validated on larger sample sets, the spectral differences noted here may be due to the variety of refractive index environments (leading to spectral broadening) and possible NP interactions (leading to higher-wavelength resonances) within the cytoplasm, as compared to the relatively consistent refractive index environment and fixed NP localization on the cell surface. More importantly, these spectra illustrate the potential of spectral analysis for functional investigation of NP interactions with cells in culture.

6. Conclusion

We have demonstrated the ability to provide epi-illumination darkfield through the imaging objective for the imaging and spectral analysis of NPs in various environments. We have validated this technique with an analysis of silver nanospheres on a glass substrate in various refractive index environments, and the results have been shown to match our previously detailed theoretical model [8]. We have characterized the distribution of darkfield illumination, finding the regions of highest incident intensity. Finally, we have shown that this technique is particularly useful for spectral analysis of scattering from NPs incubated with cells in culture, producing average scattering spectra which are consistent with extinction measurements made in previous studies [2].

We are extending this preliminary study to a more detailed analysis of NP interaction with cells in culture. Cell lines of varying EGFR expression will be evaluated, additional controls (e.g., NPs conjugated with an antibody not expected to be present in cultures of cancer cells) will be added, the spectra from dense and sparse NP populations will be compared, and a greater number of cells will be analyzed to determine the statistical significance of the findings.

In conclusion, when combined with darkfield microscpectroscopy, NP tags have the potential to reveal both spatial location and local environment of a target of interest. This multi-functional capability, combined with their high scattering cross section and biocompatibility, makes metal NPs a valuable tool in cell studies. The development of our darkfield illumination technique has the potential to greatly advance the application of single metal NPs as biological tags.