Rapid cell mapping using nanoparticles and SERRS

Abstract

Bone marrow-derived immune cells (macrophages) treated with gold and silver nanoparticles before fixation and dye staining have been analysed by multiple wavelength line scanning surface enhanced resonance Raman scattering (SERRS) mapping. The method yields high selectivity and sensitivity within short analysis times, identifying nanoparticle aggregates in secondary lysosomes. Using routine cell stains, the output from fluorescence, Raman and SERRS is quantified at four wavelengths of excitation, demonstrating the potential at longer biologically compatible wavelengths of using nanoparticles with cell stains for superior cell mapping.

---

Introduction

There is a growing need for new methods for cellular analysis that combine high speed with high resolution and sensitivity. The ability of a technique to identify multiple targets within a cell at low concentrations is also highly desirable. Surface enhanced Raman scattering (SERS) is a technique that could offer this and arises when a target molecule is adsorbed onto a suitably roughened metal surface, resulting in an enhancement of the signal intensity (10⁵–10⁶).¹ A significant additional resonance enhancement is observed when the exciting laser line is close to the maximum absorbance of the metal nanoparticle surface plasmon and a suitable vibronic transition of the molecular chromophore. Surface enhanced resonance Raman scattering (sometimes given the distinction ‘SERRS’) is a highly sensitive spectroscopic technique that is finding an increasing number of applications in biodiagnostics including gene probes^2,3 and DNA detection.⁴

Recently, Nie and co-workers demonstrated an elegant application of SERS by the in vivo targeting of tumour cells in mice.⁵ In this case, nanoparticles with embedded dye labels were functionalised with a single-chain variable-fragment (ScFv) antibody to target the over-expression of epidermal growth factor in malignant cells. Importantly, the same study reports the considerable increase in ‘brightness’ in the red near-IR window, achieved by using the SERRS method over quantum dots. Spatial resolution of SERS or SERRS from a variety of samples can be performed by using near-field or confocal scanning,⁶ tip-enhanced Raman scattering (TERS),⁷Raman mapping,⁸ or Raman imaging.⁹ Kneipp et al. have previously used SERS as an effective method to probe cellular compartments¹⁰ and to visualise the distribution of the dye indocyanine green.¹¹ Raman point mapping has recently been shown to be an effective method for the analysis of human osteosarcoma cells using gold nanoparticles and the dye Rhodamine 6G.¹² In this study, we present the application of multiple wavelength fast line scanning SERRS to the analysis of immune system cells treated with staining materials routinely used by life scientists. The use of fast line scanning combined with metallic nanoparticles as the surface for enhancing the Raman scattering produces a highly efficient mapping of cells and reduces the overall collection time for the analysis of a single cell by a factor of 15 compared to single point mapping.

Results and discussion

The cell stains methylene blue and the eosin derivative, known collectively as Giemsa, were selected as the reporter molecules in this study. The cell stain Giemsa was chosen as it comprises two components with similar molar extinction coefficients, methylene blue (MB, λ_max = 660 nm, ε ≈ 75 000 M cm⁻¹) and eosin Y (EY, λ_max = 525 nm, ε ≈ 112 000 M cm⁻¹). MB does not exhibit considerable fluorescence and has a structure suited to Ag and Au surface binding. The fluorescein derivative eosin is fluorescent and has a quantum yield of approx. 0.67. As common materials in cellular and molecular biology with several applications, these dyes are often used as stains to target the phosphate backbone of nucleic acids, particularly in adenine–thymine-rich regions.

A major advantage of the SERRS approach is that the excitation wavelength (λ_ex) can be selected anywhere in the optical range. In addition, it has been recently shown how intelligent dye reporter design¹³ and wavelength selectivity¹⁴ can increase the potential for multiple target detection within a mixture. A range of dyes have been reported that are effective at longer wavelengths using both silver and gold nanoparticles as the enhancing substrate.¹⁵ The limits of detection reported in these examples are significantly lower than using conventional fluorescence methods. Although highly sensitive, the optimal λ_ex for equivalent quantum dot-based approaches is often in the blue or near-UV, especially if multiple targets are to be analysed. Therefore the use of SERRS and λ_ex in the red or near-infra-red (NIR)¹⁶ can reduce background fluorescence and increase sample depth penetration due to lower background scattering intensity (I_scat) since I_scat ∝ 1/λ⁴. The red-NIR spectral window (600–1200 nm) is ideal for analysis of real biological samples as the balance between minimal absorbance from haem groups (shorter λ) and water (longer λIR) is optimal.

Bone marrow-derived murine macrophages¹⁷ were incubated with gold (13 nm) and silver (35 nm) nanoparticles for 2, 4, 6 and 24 hours at 37 °C, 5% CO₂ before being fixed (100% methanol or 4% paraformaldehyde) and stained (Giemsa or only methylene blue). Analysis of the time course by polarized light microscopy indicated that a substantial number of both the silver and the gold nanoparticles had been internalized by endocytosis. This was confirmed by transmission electron microscopy (TEM, shown in Fig. 1). Two hours post-treatment, nanoparticles were identified as having been endocytosed into transparent vacuoles near the macrophage plasma membrane. Furthermore, by this time, nanoparticles were also located within electron-dense vacuoles towards the centre of the cell and often near the nucleus. The darker matrix vacuoles containing nanoparticles are typical of phagosome maturation as a result of lysosome fusion.

Fig. 1 TEM of macrophage after 2 hours of incubation with gold nanoparticles. (A) Transverse section of a macrophage showing uptake of nanoparticles into vacuoles near the cell surface (transparent matrix) and also in a mature phagosome (darker matrix) near the nucleus in the centre of the cell. (B) Detail is seen in the lower panel, an enlargement of the rectangular area outlined in the upper panel.

The transmission polarized light images shown in Fig. 2(A, A1 and C1) were recorded using polarized white light illumination and an analyzer to view the scattered light from the sample (Eclipse SE2000, Nikon, Japan). The internalized nanoparticles appear coloured in high contrast due to the fact that they scatter light very efficiently at the frequency of the surface plasmon resonance maximum, up to 10⁶–10⁸ times more than an equivalent fluorophore. The dark-field image shown below in Fig. 2(B1) was recorded in the conventional manner using a high NA masked condenser (Leica, Germany). Although this method has less overall light throughput there is significantly less background scattering from the cell walls and other features. Nonetheless, using both contrast methods both Au and Ag nanoparticles were clearly visible within the cytoplasm of the cell. The imaging also indicates that the nanoparticles do not cross the nuclear membrane as this area primarily retains the blue colouration of the stain when viewed in polarized transmission. This distribution was also confirmed by TEM (Fig. 1).

Fig. 2 Nanoparticle time course studies. (A) Transmission polarized light microscope image of a macrophage cell after 24 hours of incubation with citrate-reduced Ag nanoparticles and stained with methylene blue. Areas of colour indicate the presence of nanoparticles scattering the polarized white light at plasmon resonant frequencies. (A1) As (A) except that the incubation time was reduced to 2 hours. (B1) Bright-field image of macrophage cell incubated with silver nanoparticles for 4 hours and stained with Giemsa. (B2) Dark-field image of the same cell as in (B1) recorded with illumination from a masked high NA condenser (0.9). The nanoparticles show up as brightly scattering points in the image. (C1) Macrophage incubated for 6 hours with silver nanoparticles and stained with methylene blue. (C2) Polarized light microscope image of the same cell shown in (C1).

More detailed and sensitive distribution information was obtained by a SERRS analysis of the cells. As the samples contained both nanoparticles and dye labels, strong SERRS spectra could be obtained using suitable combinations of metal nanoparticle, dyeλ_max and λ_ex.¹⁵ The effective combination of these parameters is critical for obtaining the most sensitive results. For example, MB (λ_max = 660 nm) and Giemsa (MB and eosin, λ_max = 525 nm) provided strong SERRS spectra from within cells incubated with silver nanoparticles at 514.5, 632.8, 785 and 830 nm λ_ex.

Conventional Raman point mapping or scanning provides an efficient means of both spectral acquisition and analysis. The method is information-rich as a complete vibrational spectrum is recorded for each pixel in the map as the focused laser spot is rastered across the sample area. Practically the technique is slow, especially if longer accumulation times are required for low concentration samples. Raman line mapping significantly decreases the time required to record an image by line focusing the laser and recording spectra across the full area of the charge coupled device (CCD). Owing to the high efficiency of the SERRS method, minimal exposure times could be applied, thus recording up to 1000 spectra per minute. An equivalent point mapping experiment using the same exposure time would record approx. 30 spectra per minute. Line mapping is compromised by reduced resolution along the laser line, although this can be traded against Y binning and scanning speed. The application of line mapping has several other distinct advantages. Since the laser is (line) defocused and scanned rapidly across the sample, the power density on any one area of the cell is lessened by a factor of 40–60 times, when compared to an equivalent like-for-like response in scanning confocal or point mapping. This reduces the effect of photobleaching which can be observed in identical point mapping experiments. In addition, the effective accumulation time for each point is increased.

Although Raman mapping data are often presented in two dimensions it is important to consider the fact that the spectra are collected from a three-dimensional voxel. For single point Raman mapping, the minimum spot size (S) is given by the approximation: S = 0.61λ_ex/NA, with the confocal aspect ratio being approximately 3 : 1 in this case. Normally, increasing the confocality by a factor of 5 (high confocality) much of the out-of-focus scattering is rejected but the SERRS throughput is reduced dramatically. The SERRS collection efficiency in line mapping mode at varying focal depths was investigated and is shown below in Fig. 3. The sample cell depth (when sandwiched between coverslips) was measured using z-drive focusing to be approximately 25 µm, with 0 µm being set at the coverslip surface. In a standard confocal set-up, the majority of the SERRS signal arises from within an area of the cell +5 to +20 µm above the cover slip plane (set as 0 µm). This study confirms our TEM observations illustrating that the majority of the SERRS signals originated from within the cell rather than being concentrated at receptors on the surface.

Fig. 3 SERRS signal to focal depth profile through a fixed macrophage cell (cellz ≈ 25 µm) [100×/0.75 objective] and schematic of SERRS collection set-up.

Examples of SERRS maps obtained from macrophages treated with nanoparticles, fixed, and Giemsa-stained are shown in Fig. 4. The false colour SERRS maps were created by analysis of the spectrum recorded at each pixel. The SERRS intensity was calculated by integrating the area under a known peak in the SERRS spectra of methylene blue or Giemsa (shown in Fig. 4(A)). Control experiments were performed to better understand the strength and distribution of the SERRS signals. Fig. 4(B) shows the mapped fluorescence response from control cells that have been stained using the same procedure but have not been treated with nanoparticles. The fluorescence response of the (predominantly eosin) component was recorded by measuring the absolute intensity at a fixed point in the Stokes region. The fluorescent signal intensity observed was relatively even across the cells examined, with some cells showing significantly higher intensity from the nucleus. Using 632.8 nm λ_ex, weak resonance Raman scattering (from the MB component) was detected from the stained control cells with the signal again being concentrated around the nucleus (data not shown).

Fig. 4 (A) Individual SERRS spectrum from within cells (632.8 nm, 1 × 1 s, 0.7 mW). This peak is measured relative to the baseline to dictate the intensity of the colour in the SERRS maps. (B) Macrophage cell stained with Giemsa, false colour map based on fluorescence intensity recorded by spectrometer (632.8 nm λ_ex). No nanoparticles were added. (C–C3) Streamline SERRS maps over white light image of macrophages recorded after 4 hours of incubation with Ag nanoparticles (Giemsa stain) (632.8 λ_ex ). (D) Streamline SERRS map of macrophages recorded after 4 hours of incubation with Au nanoparticles (Giemsa stain) (785 nm λ_ex). Originally recorded at 1000× magnification, the effective accumulation time per pixel is 2 seconds.

Analysis of the cells that had been incubated with nanoparticles (NP) revealed that the Raman signals from the MB component of the Giemsa stain were significantly enhanced. In the case of silver nanoparticles at 632.8 nm λ_ex, the intensity recorded in some pixels (500 nm width) of the maps increased by up to 20-fold. In addition, the signal intensity in the SERRS maps is no longer evenly distributed throughout the cytoplasm. Instead, the signals follow similar distribution patterns to the Rayleigh scattering observed using polarized and dark-field contrast (Fig. 2) and TEM (Fig. 1), with higher signals from the cytoplasm and relatively little from the nuclei. The relative fluorescence, resonance Raman and SERRS intensities observed in this study are summarised in Table 1. This highlights the effectiveness of SERRS at longer λ_ex. The UV/visible extinction of Giemsa, relative to the position of each laser excitation is show in Fig. 5 below.

Fig. 5 UV/visible spectrum of Giemsa cell stain showing the relative positions of the laser lines used.

Table 1 Relative signal intensities from mapping methods

Method	Control^a	Ag NP^b	Au NP^b
a Fixed Giemsa-stained cells not exposed to nanoparticles. b NB : fluorescence and Raman signals were calculated by different digital analysis methods and cannot be compared directly (see Experimental section). c Point mapping only. All others are fast line mapping.
Fluorescence
λ ex 514.5 nm	4165
λ ex 632.8 nm	4012
λ ex 785 nm	833
λ ex 830 nm	—
Resonance Raman
λ ex 514.5 nm	1037
λ ex 632.8 nm	255
SERRS
λ ex 514.5 nm^c		2040	—
SERRS
λ ex 632.8 nm		6426	3366
SERRS
λ ex 785 nm		2635	6613
SERRS
λ ex 830 nm		1037	748

---

The signal-to-background ratio increases with longer λ_ex when using silver nanoparticles. This is due to the weaker background contribution from fluorescence and general light scattering in the optical path at longer wavelengths. The resonance contribution for many vibronic states will decay in an approximately Lorenztian fashion as the wavelength of the excitation source is moved away from the optimum. However, the surface enhancement provided by the nanoparticles is available over larger wavelength ranges and at longer wavelengths. This may be due to the presence of aggregates of nanoparticles contained within vesicles as demonstrated by TEM. Larger aggregates are known to afford longer wavelength surface plasmon resonances in addition to stronger electric field gradient junctions between particles. Examination of Fig. 5 and gold nanoparticles were found to be relatively ineffective SERRS substrates in this study using 514.5 nm λ_ex due to the proximity of the interband absorbance region. However, we found that using the same dye, gold nanoparticles were highly effective at 632.8, 785 and 830 nm λ_ex. It was possible to detect low levels of gold nanoparticles in the stained cells after only short accumulation times (2 s effective per pixel, ca. 6 min per cell using 500 nm Y step size).

Conclusions

Fast line mapping combined with the high Raman scattering efficiency from areas of the sample where nanoparticles are present results in an efficient, highly sensitive analytical technique. The further development of this method using biologically functionalised nanoparticles to target activity in cells is in progress and will be reported shortly. More efficient, accurate and sensitive analysis of the biology of cells, such as macrophages, which play significant roles in infectious and inflammatory disease, will create the potential to identify more and better therapeutic targets for disease intervention.

Experimental

Cell culture

Bone marrow-derived macrophages were obtained from BALB/c mice. Briefly, the femurs were flushed with 5 ml of Dulbecco's Modified Eagle Medium (D-MEM; Invitrogen, Paisley, UK) substituted with 10% v/v heat inactivated fetal bovine serum (Sigma Aldich, Poole, UK), 1% v/v of 2 mM L-glutamine solution, 1% v/v of 100 IU/ml Penicillin-100 µg/ml Streptomycin (PAA Laboratories, GmbH, Austria) and 30% v/v L-cell conditioned media and the resultant cell suspension was incubated at 37 °C and 5% CO₂. The cells were harvested on day 10, suspended in serum-free media (RPMI 1640; Lonza, Belgium) supplemented with 1% v/v of 2 mM L-glutamine solution and 1% v/v of 100 IU/ml Penicillin-100 µg/ml Streptomycin (PAA Laboratories, GmbH, Austria) and seeded at 1 × 10⁴ tp 1 × 10⁵ per well in 24-well sterile tissue culture plates (Iwaki, Japan) to which 13 mm round coverslips (Scientific Laboratory Supplies Ltd, Nottingham, UK) had previously been added. Macrophages were incubated overnight at 37 °C and 5% CO₂ to allow the cells to adhere to the coverslips. Macrophages were washed with serum-free media and incubated at 37 °C and 5% CO₂ with gold or silver colloid (used at 1–10 µl/ml) for 2–24 hours. Colloid was prepared by centrifuging 1 ml aliquots of gold or silver at 13 000 rpm for 15 minutes and resuspending the resultant pellet in 500 µl of molecular grade water (Sigma Aldrich, Poole, UK). Following incubation with colloid, the cells were washed twice with 1× phosphate buffered saline (PBS, pH 7.4) to remove any extracellular colloid before fixation with 4% paraformaldehyde. Macrophages were stained with either methylene blue or a 10% w/v aqueous Giemsa solution (BDH, VWR International Ltd, UK) for 20 minutes and washed with PBS before being allowed to dry and mounted onto labeled slides (26 × 76 × 1.0 mm microscope slides; Thermo Electron Corporation, Cheshire, UK) using DPX Mountant (BDH, VWR International Ltd, UK).

SERRS Spectroscopy and line mapping

SERRS mapping was performed using Renishaw InVia Raman inverted and upright microscope systems (Renishaw, Wotton-under-Edge, UK), equipped with 100×/0.75 and 50×/0.5 long working distance objectives. Four excitation sources were used at 514.5 nm (ca. 6 mW), 632.8 nm (HeNe, ca. 30 mW), 785 nm (ca. 180 mW) and 830 nm (diodes, ca. 170 mW). Power was attenuated using neutral density filters to that described for each experiment. Line mapping was performed using a Streamline Raman mapping system (Renishaw, UK). The reduction in power density using the line focusing optics was between 40 to 60 times depending on the wavelength of excitation used. SERRS maps were coloured by integrating the area under single characteristic peaks. Fluorescence intensity is recorded as intensity at a fixed point in the Stokes region normally free from Raman bands (e.g.ca. 1700–1900 cm⁻¹). The detector used at 632.8, 785 and 830 nm was a deep-depletion RenCAM CCD. SERRS intensities were corrected in Table 1 for objective NA and for the wavelength dependent response of the detector and grating using a calibration source. Two gratings were used in the experiments for spectral dispersion. These were 1800 lines/mm (514.5 and 632.8 nm) and 1200 lines/mm (632.8, 785 and 830 nm). The response was normalized using a combination of a white light and neon calibration source. This allowed a direct comparison of the efficiency of the technique at four wavelengths.

Dark-field and polarized light microscopy

Dark-field microscope images were recorded using an inverted microscope equipped with a masked condenser NA = 0.9 (S1; Leica, Germany) and long working distance objectives with appropriate NA values. Polarized light images were recorded using an inverted microscope (Eclipse TE2000, Nikon, Japan) equipped with a polarized xenon arc white light source. The transmitted light was collected using a ∞60× objective (Nikon, Japan) and an analyzing polarizer. The total optical magnification in the original images was 900×.

Transmission electron microscope imaging

Cells were harvested after nanoparticle incubation, fixed in phosphate buffered 2% glutaraldehyde for 40 mins, followed by aqueous 1% osmium tetroxide and 2% uranyl acetate en bloc staining, dehydration in ethanol and finally embedding in Epon/Araldite resin. After polymerizing at 60 °C for 48 hours, sections were cut and imaged after lead and uranyl staining. Images were recorded on a 2K Proscan CCD on a Zeiss 912 AB energy filtering transmission EM, contrast enhanced using zero-loss imaging mode.

References

1. M. Moscovits, Rev. Mod. Phys., 1985, 57(3), 783.
2. T. Vo-Dinh, K. Houck and D. L. Stokes, Anal. Chem., 1994, 66(20), 3379.
3. L. R. Allain and T. Vo-Dinh, Anal. Chim. Acta, 2002, 469, 149.
4. Y. C. Cao, R. Jin and C. A. Mirkin, Science, 2002, 297(5586), 1536.
5. X. Quian, X.-H. Peng, D. O. Ansari, Q. Yin-Goan, G.-Z. Chen, D. M. Shin, L. Yang, A. N. Young, M. D. Wang and S. Nie, Nat. Biotechnol., 2008, 26(1), 83.
6. P. Anger, A. Feltz, T. Berghaus and A. J. Meixner, J. Microsc. (Oxf.), 2005, 209(3), 162.
7. B. Ren, B. Pettinger, G. Picardi, R. Schuster and G. Ertl, Angew. Chem., Int. Ed., 2005, 44(1), 139.
8. K. Kneipp, A. S. Haka, H. Kneipp, K. Badizadegan, N. Yoshizawa, C. Boone, K. E. Shafer-Peltier, J. T. Motz, R. R. Dasari and M. S. Feld, Appl. Spectrosc., 2002, 56(2), 150.
9. M. B. Wabuyele, F. Yan, G. D. Griffin and T. Vo-Dinh, Rev. Sci. Instrum., 2005, 76, 63710.
10. J. Kneipp, H. Kneipp, M. McLaughlin, D. Brown and K. Kneipp, Nano Lett., 2006, 6(10), 2225.
11. J. Kneipp, H. Kneipp, W. L. Rice and K. Kneipp, Anal. Chem., 2005, 77(8), 2381.
12. H.-W. Tang, X. B. Yang, J. Kirkham and D. A. Smith, Anal. Chem., 2007, 79(10), 3646.
13. R. J. Stokes, A. Ingram, J. Gallagher, D. R. Armstrong, W. E. Smith and D. Graham, Chem. Commun., 2008, 567.
14. K. Faulds, F. McKenzie and D. Graham, Angew. Chem., Int. Ed., 2007, 46, 1829.
15. R. J. Stokes, A. Macaskill, P. J. Lundahl, K. Faulds, W. E. Smith and D. Graham, Small, 2007, 3(9), 1593.
16. K. Kneipp, H. Kneipp, I. Itzkan, R. R. Dasari and M. S. Feld, Chem. Rev., 1999, 99, 2957.
17. J. Caamano, J. Alexander, L. Craig, R. Bravo and C. A. Hunter, J. Immunol., 1999, 163, 4453.

---