2.1.

Materials

Tetrachloroauric acid trihydrate (${\mathrm{HAuCl}}_4·3{\mathrm{H}}_2\mathrm{O}$) (99.9%), trisodium citrate ($>99\%$), poly-l-lysine solution (0.1% in ${\mathrm{H}}_2\mathrm{O}$), and trypsin-Ethylenediaminetetraacetic acid (0.25% in ${\mathrm{H}}_2\mathrm{O}$) were purchased from Sigma Chemicals. Silver nitrate (99.9%) was purchased from RANKEM, India. Dulbecco’s modified Eagle medium (DMEM) with high glucose, GlutaMAX™ Supplement and pyruvate, Antibiotic ($100\times$, penicillin–streptomycin–glutamine), and fetal bovine serum were purchased from Invitrogen. Plasticware for cell culture was purchased from Tarson, India. Millipore deionized water (DI) ($\sim 18.2\mathrm{M}\mathrm{\Omega }$) was used throughout the experiments.

2.2.

Synthesis of Citrate Capped Silver Nanoparticles

AgNPs were synthesized by the Turkevich method.³² Briefly, 17 mg of AgNO₃ was dissolved in 100 mL of DI water. This solution was heated to 100°C and then 40 mg of trisodium citrate was added. Once the color of the solution started changing to pale yellow, it was cooled under tap water. As synthesized, AgNPs were characterized using UV–Vis spectroscopy and transmission electron microscopy (TEM) imaging. A broad surface plasmon resonance (SPR) distribution with a maximum at $\sim 410\mathrm{nm}$ was observed [Fig. 1(a)]. TEM suggests nonuniformity in the size and shape of AgNPs.

Fig. 1

(a) UV–Vis extinction spectrum of citrate capped AgNPs. Inset shows transmission electron microscopy (TEM) image of these nanoparticles. (b) UV–Vis extinction spectrum of citrate capped AuNPs. Inset shows TEM image of these nanoparticles.

2.3.

Synthesis of Citrate Capped Gold Nanoparticles

GNPs were synthesized by the Turkevich method.³² Briefly, 20 mL of 0.3 mM ${\mathrm{HAuCl}}_4$ solution in DI water was heated in a synthesizer (400 rpm). Upon boiling, $240\mu \mathrm{L}$ of 100 mM trisodium citrate solution was added. After 20 min of boiling, the color of the solution changed to wine red. Then the solution was cooled at room temperature. As synthesized, AuNPs were characterized using UV–Vis spectroscopy and TEM imaging. AuNPs of size $\sim 40\mathrm{nm}$ in diameter with SPR around 529 nm were observed [Fig. 1(b)].

2.4.

Bacterial Culture and Sample Preparation

Escherichia coli 739 bacterial culture was grown to an ${\mathrm{OD}}_{600}$ (optical density at 600 nm) value of 1. This culture (5 mL) was washed twice by centrifugation (3000 rpm, 10 min) with DI water. The final pellet was resuspended in the as-prepared AgNP solution and incubated at room temperature for 1 h. After that, this E. coli-AgNP suspension was washed twice with DI water and spotted on a poly-l-lysine-coated ultrasonically cleaned glass slide (1-mm thickness, SCHOTT). After 5 min of incubation, the slide was flushed with DI water to remove unbound bacteria and covered with 0.145-mm thick coverslips (SCHOTT) for imaging purposes.

2.5.

Mammalian Cell Culture and Sample Preparation

HEK293 cells were cultured on poly-l-lysine-coated glass coverslips in six-well plates. After reaching $\sim 80\%$ confluency, cells were washed twice with $1\times$ phosphate-buffered saline (PBS) and 2 mL DMEM containing $100\mu \mathrm{L}$ citrate capped GNPs was added. These cells were incubated overnight inside the incubator at 37°C. After this, the cells were washed thrice with $1\times$ PBS and incubated with paraformaldehyde solution (4% in $1\times$ PBS) for 8 min. Then the cells were washed thrice with $1\times$ PBS and mounted on 1-mm thick glass slides (SCHOTT).

2.6.

UV–Visible Extinction Spectroscopy

Ensemble UV–Vis extinction spectroscopic measurements were performed using PerkinElmer Lambda 25 spectrophotometer in the range of 200 to 1100 nm.

2.7.

Transmission Electron Microscopy

High resolution transmission electron microscopic measurements were performed using a JEOL 3010, 300 kV instrument. Carbon-coated copper grids were used for measurements. Samples were spotted by drop casting followed by ambient air drying.

2.8.

Confocal Raman Scattering Microspectroscopy

Raman microspectroscopic measurements were performed using WiTec GmbH equipment. Frequency doubled Nd:YAG dye laser (532 nm) with a maximum output power of 40 mW was used for excitation of the sample. The laser was focused onto the sample using a $100\times$ oil immersion objective (UPLFLN, Olympus). The signal, after passing through a 532 nm super-notch filter, was dispersed using a grating ($600\text{groves}{\mathrm{mm}}^{-1}$) onto a charge-coupled device (CCD). Spectral images were scanned using the sample mounted on a piezo stage.

2.9.

Plasmonic and Raman Scattering Microspectroscopy

For plasmonic and Raman scattering microspectroscopy (PRSMS) measurements, an attachment was designed to use a Cytoviva™ high-resolution dark field condenser (oil immersion) and $100\times$ oil immersion objective (UPLFLN, Olympus) in the above-mentioned confocal Raman microspectroscope. For white light illumination (400 to 1000 nm), a L1090-Halogen Lamp from International Light Technologies Inc. was used. The other details of the instrument for Raman microspectroscopic measurements are the same as above.

3.1.

Setup for Combined Plasmonic and Raman Scattering Microspectroscopy

To accomplish correlated DFSMS and RSMS measurements of biological samples using the same setup, we attached a high-resolution dark field condenser with a conventional point-scan RSMS setup. Schematic and photographs of the actual setup are shown in Fig. 2. In the PRSMS setup, a dark field condenser was used to illuminate samples in the transmitted dark-field configuration. For the DFSMS of samples, a visible–near-infrared (VNIR) light source with a wavelength range from 400 to 1000 nm was used to excite the plasmonic modes of NMNPs. For RSMS, a 532-nm laser was used to excite the Raman spectra and measurements were performed in the reflection mode. Samples sandwiched between a coverslip and glass slide were placed on a piezoelectric scanner and the scattered light was collected using a $100\times$ oil immersion objective. Substantial signal strength in conjunction with confocality was achieved using a $100\text{-}\mu \mathrm{m}$ optical fiber. This fiber was connected to a VNIR spectrometer equipped with appropriate gratings to resolve the spectra. Since for DFSMS a wide range of wavelengths has to be scanned, a grating with $150\text{grooves}/\mathrm{mm}$ was used. For RSMS, a shorter wavelength range but better resolution is necessary. Hence, to resolve RSMS spectra, a grating with $600\text{grooves}/\mathrm{mm}$ was used. It is important to mention that despite the use of wide field illumination in DFSMS, the method of signal collection was confocal due to the configuration of the RSMS setup. Such a signal collection in DFSMS has its pros and cons, which are discussed below.

Fig. 2

Schematic of the setup for correlated plasmonic and Raman scattering microspectroscopy (PRSMS) measurements. Briefly, it consists of a high-resolution dark field condenser attached with conventional point-scan confocal RSMS setup. Inset shows a dark field condenser (red arrow, top left) along with the designed attachment (blue arrow, down right).

3.2.

Images of AgNP Treated Bacteria Collected with Different Configurations

Figure 3(a) shows a dark field image of bacteria (E. coli 739) treated with citrate capped silver nanoparticles. This image was captured using a CCD camera of the PRSMS setup. In such an image, it appears that only borders (cell wall) of the bacteria scatter substantial light, whereas the central portion appears transparent. This is because the wide field image strongly reflects the relative intensity variations which occur due to light collection from all focal planes. But when this sample was measured for DFSMS using the PRSMS setup, the spectral image was captured pixel by pixel using an optical fiber which introduces confocality. Such a confocal signal collection produces an image constructed from the light collected only from a particular focal plane. This image has a diffused appearance as shown in Fig. 3(b). To eliminate the possibility of defocused imaging in the PRSMS setup, care was taken to focus the desired plane of imaging by using a laser spot for confocal RSMS. Figure 3(c) shows a line profile of the scattering intensity along the yellow line in Fig. 3(b). Line profile shows the maximum at the center similar to the airy disc pattern and confirms that the particle was properly focused.

Fig. 3

(a) Dark field image captured using CCD camera of the PRSMS setup. (b) Pseudo-colored dark field scattering microspectroscopy (DFSMS) image collected using the PRSMS setup for the same area. Red arrows indicate the presence of AgNPs which exhibit sharp SPRs. A dotted yellow line shows pixels for which scattering intensity profile is shown in (c). Scale bar is $1\mu \mathrm{m}$ for images (a) and (b).

We also observed that when DFSMS images are presented as intensity scaled images, it does not reveal information about locating the NMNPs in complex environments. However, the sharp and nearly monochromatic scattering features of NMNPs allow them to be distinguished as bright and distinct colored spots. This can be achieved by presenting DFSMS images as pseudo-RGB color coded images [Fig. 3(b)]. In pseudo-colored DFSMS images, we have assigned 552-, 615-, and 688-nm wavelengths to blue, green, and red colors, respectively. With this, particles can be clearly located as shown in Fig. 3(b) by red arrows. On this background, we have performed PRSMS measurements on two different systems. These measurements combine our ability to locate NMNPs using DFSMS and colocalizing the same with biomolecules using RSMS.

3.3.

Colocalization of AgNPs and Biomolecules in E. Coli

A previous study¹⁰ by our group using RSMS has shown that AgNPs interact specifically with the bacterial DNA. It was observed that adenine (A), guanine (G), and cytosine (C) exhibit substantial SERS intensities while thymine (T) shows only limited SERS enhancement. This is because the interactions between AgNPs and A, G, and C are stronger than those with T. This specificity of interaction is due to the interaction between AgNPs and the exocyclic nitrogen present in A, G, and C. In the aforementioned study, DFSMS was used separately to show the attachment of AgNPs to the bacteria.¹⁰ Here, in the present study, we use PRSMS to colocalize the position of AgNPs with the Raman signatures of biomolecules. Figure 4 shows the consolidated data of these measurements.

Fig. 4

Images of AgNP treated bacteria collected using PRSMS setup in different configurations. (a) Dark field image captured using CCD camera. (b) Pseudo-colored DFSMS image. (c) Intensity scaled confocal RSMS image and corresponding color scale are shown. (d) Overlay of the clustered RSMS image with dark field image. (e) Scattering spectra of AgNPs 1 to 5 are shown along with the scattering spectrum of bacteria. (f) Averaged cluster spectra corresponding to image (d) are shown in matching colors. Scale bar is $1\mu \mathrm{m}$ in all the images.

Figure 4(a) shows the dark field image captured using a CCD camera with the PRSMS setup. In the region of interest (ROI) selected for imaging, we can see AgNPs labeled 1, 2, 3, and 4 are attached on the surface of bacteria and particle 5 is on the glass slide. The next image shows a pseudo-colored DFSMS image [Fig. 4(b)] for the same ROI. Scattering spectra collected from particle positions are shown in Fig. 4(e). Based on the scattering spectra, it can be observed that particle 1 is possibly a rod or prism-shaped silver nanoparticle;³³ whereas the other particles are relatively smaller and scatter light at lower wavelengths. Among the marked AgNPs, particles 2 and 3 scatter more light as compared to particles 4 and 5. This suggests that AgNPs 2 and 3 are relatively larger particles. Also as compared to AgNP 1, other particles show a broadening in their scattering peak, implying adsorption of biomolecules on their surfaces. Figure 4(c) shows the intensity scaled RSMS image of the same ROI shown in Fig. 4(a). It suggests a maximum intensity enhancement in the vicinity of AgNPs 2 and 4. To understand this data better, we have done cluster analysis of the RSMS image. Cluster analysis helped in grouping and color coding the Raman spectra on the basis of their similarities. Then this color coded image was overlaid with the dark field image as shown in Fig. 4(d). Averages of the spectral clusters are shown in Fig. 4(f) where it can be seen that although some intensity enhancement was observed on AgNP 2 and bacteria at the bottom, actual SERS activity was observed only in the vicinity of AgNP 4 (red colored Raman spectrum). Considering the scattering spectrum of AgNP 4, it can be concluded that in this case, SERS activity is due more to the electronic resonance-charge transfer between the AgNP and the molecule in its vicinity than the electric near-field enhancement. These observations suggest that although many AgNPs attach to the bacteria, all particles do not exhibit SERS activity. This is due to the dependence of SERS activity on the site of molecular adsorption on the surface of the nanoparticle as well as the size and shape of the nanoparticle. Assignments of Raman bands corresponding to the molecule in the vicinity of AgNP 4 are given in Table 1. Most of these bands were found to be due to DNA. These results are in good agreement with our previous observations on the interaction between AgNPs and bacteria.¹⁰ For a further demonstration of the applicability of PRSMS setup with mammalian cells, we have performed measurements on HEK293 cells treated with citrate capped AuNPs. These observations are discussed in Sec. 3.4.

Table 1

Assignments of Raman bands of AgNP treated Escherichia coli sample.

3.4.

Colocalization of AuNPs and Biomolecules Inside HEK293 Cells

Figure 5(a) shows the dark field image of HEK293 cells treated with citrate capped AuNPs of $\sim 40\mathrm{nm}$ in diameter. The Z scanning (focus stacking) measurements performed on these cells have shown that the AuNPs under investigation were within the cells. To prove this, a movie compiled of a Z scanned image stack is shown in Video 1. Figure 5(b) shows a pseudo-colored DFSMS image of the ROI shown in Fig. 5(a). Corresponding scattering spectra shown in Fig. 5(d) suggest that particle 1 has a sharp plasmon resonance and relatively less FWHM, which implies that it is a single AuNP. Other particles (2 and 3) show scattering signatures of aggregates (multiple broad but sharp SPRs). The scattering signal from cellular components exhibits a broad spectrum as well. When the ROI shown in Fig. 5(a) was scanned for RSMS, particle 1 exhibited the most SERS activity [Figs. 5(c) and 5(e)] whereas an aggregate (particle 2) exhibits only a moderate enhancement in the Raman signal as compared to the background. This observation suggests that, although due to the plasmonic coupling effect aggregates are expected to exhibit better SERS activity than single nanoparticles, SERS activity is highly specific to the site of molecule adsorption. Hence, such PRSMS studies are necessary for additional confirmation about the nature of the nanoparticle under investigation. When diagnostic or sensing techniques are solely based on SERS activity or the expected hybridization of NMNPs by specific biomolecules, colocalization using PRSMS is necessary to prove the presence of moieties and nanostructures of interest. In this particular measurement, assignments for Raman bands are given in Table 2. A major band was assigned to protein molecules (1600 to $1700{\mathrm{cm}}^{-1}$) and another band between 1050 and $1125{\mathrm{cm}}^{-1}$ was assigned to DNA and RNA O-P-O stretching. It has been reported by previous studies that such an appearance of the Raman band is commonly found for mammalian cells undergoing apoptosis.²⁵

Fig. 5

Images of AuNP-treated HEK293 cells collected using PRSMS setup at different configurations. (a) Dark field image captured using CCD camera. Region of interest (ROI) scanned for PRSMS is shown by the square. Image is taken at such focus that nanoparticles can be seen clearly. Cells can be seen clearly in a movie made of Z stacking of images (Video 1). (b) Pseudo-colored DFSMS image of the ROI shown in (a). Particles are numbered as 1, 2, and 3. Scale bar is 1 micron. (c) Intensity scaled RSMS image. (d) Scattering spectra of AuNPs 1 to 3 are shown along with the scattering spectrum from cellular components. (e) Inset of this image shows RSMS image of particle 1 when it was rescanned. Scale bar is $0.8\mu \mathrm{m}$. Raman spectra corresponding to particles 1 (red) and 2 (blue) from RSMS image (c) are shown along with the Raman spectrum of particle 1 after rescanning (black) (Video 1, MPEG, 0.4 MB).[DOI: http://dx.doi.org/10.1117/1.JBO.20.4.046011.1].

Table 2

Assignments of Raman bands of AuNP treated HEK293 sample.

Further, from our observations, it can be seen that when the same ROI containing particle 1 was scanned separately for RSMS [Fig. 5(e), black trace], the Raman spectrum did not repeat with high fidelity upon rescanning [Fig. 5(e), red trace], although in both these measurements, the instrumental parameters were kept exactly the same. This may be due to the complexity of the biological samples and sensitivity of the measurements to the surrounding environment. Hence, we suggest that in biosensing measurements, it is first necessary to optimize experimental parameters using standard samples of similar types and then perform measurements with the experimental samples.