2.1.

Generation and Examination of Gold Nanoparticles

The applied laser-based approach to nanoparticles consists in the ablation of a target in liquid media by intense laser radiation, leading to an ejection of its constituent and the formation of a colloidal nanoparticle solution.^{14, 15} In the study at hand, AuNPs were generated using a femtosecond laser system (Spitfire Pro, Spectra-Physics, Santa Clara, California) delivering $120\text{-}\mathrm{fs}$ laser pulses at a wavelength of $800\mathrm{nm}$ (maximum energy $400\mu \mathrm{J}$ per pulse, beam diameter $4\mathrm{mm}$ ). The pulse energy was fixed to $200\mu \mathrm{J}$ at a repetition rate of $5\mathrm{kHz}$ . The principle setup is described as follows. The laser beam was focused on $5\times 5\text{-}\mathrm{mm}$ gold foil (99.99% purity), thoroughly cleaned and placed on the bottom of a Petri dish filled with $4\text{-}\mathrm{mL}$ bidistilled water. The plate was placed on an axis system that moved at a constant speed of $1\mathrm{mm}{\mathrm{s}}^{-1}$ in a spiral with outer radius of $2\mathrm{mm}$ and inner radius of $0.4\mathrm{mm}$ . Time of irradiation was fixed to $12\min$ corresponding to three spirals.

The resulting colloid was characterized by UV-visual spectroscopy using a Shimadzu 1650 (Shimadzu Europe GmbH, Duisburg, Germany) and TEM (EM 10 C electron microscope, Zeiss, Oberkochen, Germany).^{4, 16} 500 nanoparticles were counted for the determination of the average Feret diameter (Fig. 1 ). AuNP concentration was estimated by weighing (Sartorius M3P-000V001, Sartorius AG, Göttingen, Germany) the gold foil three times prior and after laser ablation with an accuracy of $1\mu \mathrm{g}$ .

Fig. 1

Number- and mass-weighted size distribution of laser-generated gold nanoparticles in water estimated by centrifugation. Insert: scheme of nanoparticle generation by laser ablation in liquids and representative transmission electron micrograph of nanoparticles prior to centrifugation.

2.2.

Nanoparticle Size Separation

Size separation of AuNPs was performed via successive centrifugation steps of $500\mu \mathrm{L}$ of the colloid with a Hellma Universal 320 centrifuge Rotor 1612. The time of centrifugation $\left(t\right)$ was fixed to $15\min$ , and the rotation velocity $\left(v\right)$ was calculated via Eq. 1 (for detailed values see Table 1 ).

Table 1

Centrifugation parameters for nanoparticle size separation ( t=15min each).

The supernatant was separated from the pellet and used for the following centrifugation step. The pellet was redispersed in $100\text{-}\mu \mathrm{L}$ ${\mathrm{H}}_2\mathrm{O}$ , and size distributions and optical properties were characterized by dynamic light scattering and UV-visual spectroscopy, respectively.

Nanoparticle mass concentrations $\left(c_m\right)$ were estimated via recorded intensities at $380\mathrm{nm}$ , which were converted by interpolation from a linear standard calibration curve. The wavelength of $380\mathrm{nm}$ corresponds to the interband transition of gold and is hardly affected by nanoparticle size, shape, and environment. The number concentration $\left(c_n\right)$ was subsequently calculated via Eq. 2 assuming a density of GNPs of $p=19.3\mathrm{g}∕{\mathrm{cm}}^3$ , where $r$ is the mean radius of each group.

Two different size-separated series were established. While one of them was adjusted to constant particle mass concentrations ( $50\mu \mathrm{g}∕\mathrm{mL}$ respective $250\mu \mathrm{M}$ $\mathrm{Au}=\mathrm{CM}$ ), a second series was adjusted to constant particle number concentrations ( $1\ast {10}^{10}$ $\mathrm{AuNP}∕\mathrm{mL}=\mathrm{CN}$ ).

2.3.

Preparation of Gold Nanoparticles in Dispersion for Laser Scanning Confocal Microscopy Analysis

Colloidal AuNPs of each of the size fractions of both series were analyzed in $0.5\text{-}\mu \mathrm{L}$ $\mathrm{dd}{\mathrm{H}}_2\mathrm{O}$ resting in a $5\text{-}\mu \mathrm{L}$ silicon oil cushion. All colloids were sealed under a cover slip. Imaging was performed within two hours after preparation.

Thus the number of particles included in each size fraction within CM increased with decreasing particle size (Table 2 ). In CN, particle fractions consist of decreasing masses with smaller sizes (Table 2).

Table 2

Size-dependent reciprocal relationships of AuNP mass and particle number.

2.4.

Cell Culture and Preparation for Laser Scanning Confocal Microscopy Analysis

The cells used in the coincubation studies were derived from a bovine immortalized endothelial cell line (GM7373) obtained from the Cell Culture Collection, Institute of Infectology, FLI, Riems, Germany). Cell growth conditions and postcoincubation treatment corresponded to previously described trials.⁴ For LSCM analyses, these prepared cells were mixed 1:4 with Vectashield (Axxora, Loerach, Germany) and mounted on a slide within a paper reinforcement ring sealed by a cover slip and nail varnish.

2.5.

Gold Nanoparticle Coincubation Assays

Coincubation assays were performed using AuNPs in a final concentration of $50\mu \mathrm{M}$ . After seeding, cells were allowed to attach and were then exposed to the nanoparticles for $48\mathrm{h}$ . The added AuNPs displayed either the complete size range with an average particle diameter of $15\mathrm{nm}$ or separated size fractions of $40\text{to}60\mathrm{nm}$ and $60\text{to}80\mathrm{nm}$ , respectively. Cells incubated without AuNPs for $48\mathrm{h}$ were used as controls. All experiments were carried out in duplicates and repeated three times.

2.6.

Imaging by Laser Scanning Confocal Microscopy

Visualization was carried out with a confocal laser scanning system LSM510 at an Axioplan 200 (Carl Zeiss MicroImaging GmbH, Jena, Germany) within the spectrum of visible light. Laser wavelengths of $514\mathrm{nm}$ (argon, $15\mathrm{mW}$ ), $543\mathrm{nm}$ (helium neon green, $1\mathrm{mW}$ ), and $633\mathrm{nm}$ (helium neon red, $5\mathrm{mW}$ ) were used to excite the SPR or luminescence of the colloidal AuNPs and the AuNPs in cultured bovine cells. Light scattering for each of the excitation wavelengths was recorded in multitracking mode in combination of four separate detection channels within fixed spectral bands. Complete configurations are summarized in Table 3 . The colloidal AuNPs were investigated dispersed in bidistilled water using the configurations described in Table 3. After focusing the edges of the silicone oil cushion, a volume of $10\mathrm{pL}$ centered within the liquid volume, avoiding reflections at the surfaces by exclusion of $2\text{to}3\mu \mathrm{m}$ next to each glass, was imaged. The focus of cells was adjusted using the differential interference contrast (DIC) contrast as reference.

Table 3

Configurations for surface plasmon resonance (SPR) and luminescence detection of AuNPs. Laser power is measured at the object level with 100% laser output. Laser power at 0.54 has performance at 5.9A.

2.7.

Statistics

Statistical analyses were performed with the software SigmaStat 3.0 (StatCon, Witzenhausen, Germany). Quantification of LSCM data was performed on five repetitive measurements for the colloidal nanoparticles and ten fields of view for each of the size groups of the AuNPs in cells. Data of reflecting pixel numbers were normalized by square root transformation and were applied to one-way analysis of variance with the pair-wise multiple test by Holm-Sidak $(p=0.05)$ .

2.8.

Mass Spectroscopic Quantification of Gold in Bovine Immortalized Cells

The summarized analysis of the gold concentration in the cells was performed by inductively coupled plasma mass spectrometry (ICPMS) by a certified laboratory (Wessling GmbH, Hannover, Germany). Samples of $200\mu \mathrm{L}$ containing approximately $3\times {10}^6$ cells were analyzed. The particle numbers $\left(\mathit{Np}\right)$ were calculated using the sample-specific cell numbers $\left(n_c\right)$ and the mass of gold within the sample $\left(m_{\mathrm{au}}\right)$ for each group using Eq. 3. $P_m$ refers to the mean particle masses within the groups.

3.1.

Physicochemical Characterization of Gold Nanoparticles

The laser-generated aqueous gold colloid is characterized by a broad size distribution, covering the range of $3\text{to}200\mathrm{nm}$ with an average diameter of $15\mathrm{nm}$ as determined by transmission electron microscopy (Fig. 1, insert). The size classification by centrifugation reveals that the mass-weighted distribution is centered at around $50\mathrm{nm}$ , while the resulting number-weighted distribution shifts to smaller nanoparticles (Fig. 1). More than 80% of the nanoparticles seem to have a size of less than $20\mathrm{nm}$ .

In accordance with the Mie theory,¹⁷ UV-visual spectra of the size fractions reveal a red shift and a broadening of the SPR with increasing mean diameter (Fig. 2 ). By normalizing the size fractions to a constant nanoparticle number of ${10}^{10}\mathrm{AuNP}∕\mathrm{mL}$ , one additionally observes a significant intensity enhancement of the SPR for bigger nanoparticles. The SPR of nanoparticles with a diameter of less than $20\mathrm{nm}$ is thus hardly visible. This fact can be explained by the higher extinction coefficient of bigger nanoparticles and is even detectable when adjusting the fractions to a constant nanoparticle mass, although the nanoparticle number is significantly lowered with increasing diameters (Table 2).

Fig. 2

UV-visual spectra of gold nanoparticle size fractions after centrifugation normalized to constant nanoparticle mass concentrations of $50\mu \mathrm{g}∕\mathrm{mL}$ (CM, left) and constant nanoparticle number concentrations ( $\mathrm{CN}={10}^{10}\mathrm{AuNP}∕\mathrm{mL}$ , right).

3.2.

Visualization of Size-Restricted Colloidal Gold Particles by Laser Scanning Confocal Microscopy

Representative images taken of a volume of about $10\mathrm{pL}$ $(31.3\times 31.3\times 10.21\mu \mathrm{m})$ of colloidal AuNPs of each size group of CN and CM are shown in Figs. 3, 3, 3, 3 and Figs. 3, 3, 3, 3, respectively. The number of particles contained in the visualized volume of about $10\mathrm{pL}$ of each fraction of CM was calculated using the average diameter and is added in Figs. 3, 3, 3, 3. As the particles were not fixed on the glass surface, the appearing irregular arrays of pixels could reflect the Brownian motion of diffusing particles. Time series after single laser excitation, reduced field of view $\left(2.4\mathrm{pL}\right)$ , and short exposure times $3\mu \mathrm{s}∕\text{pixel})$ verified this assumption.¹⁸ As these arrays reflect single particles, counting of the pixels provides a good indication of particles accessible to objective image analysis although the pixel numbers are higher than the number of particles contained in the sample.

Fig. 3

Representative images of size-restricted AuNPs in $10\text{-}\mathrm{pL}$ volumes $(31.3\times 31.3\times 10.21\mu \mathrm{m})$ . (a) through (e) show images from mass-constant dispersions (CM). (f) through (j) show all colloids which were set to an equal number of AuNPs independent of their size (CN). Size of AuNPs was $>100\mathrm{nm}$ in (a) and (f); $60\text{to}80\mathrm{nm}$ in (b) and (g); $40\text{to}60\mathrm{nm}$ in (c) and (h), and $20\text{to}40\mathrm{nm}$ in (d) and (i). Excitation was performed at 514, 543, and $633\mathrm{nm}$ with the corresponding SPR shown in blue, green, and red, respectively. Pixel size is $80\times 80\times 700\mathrm{nm}$ , and pixel time is $5.44\mu \mathrm{s}$ . The box plots in (e) and (j) represent the median with 25 and 75% quartils of reflecting pixels in each of the size classes after excitation at $514\mathrm{nm}$ , scale $1\mu \mathrm{m}$ . (Color online only.)

Interestingly, and probably due to differences in size or arrangement, a part of single objects in the samples seem to reflect selectively one or the other excitation wavelength, while others show colocalization of reflection of the different excitation wavelengths. Thus, the quantification of the numbers of reflecting pixels was performed for each of the three detection bandpasses and in both series of size-separated colloidal AuNPs.¹⁸ The SPR was shown previously quantified for all three reflection channels of CM (see Fig. 3 of Ref. 18). Exemplarily, the data from the excitation at $514\mathrm{nm}$ are shown [Figs. 3 and 3] as they accounted for the highest number of pixels due to the proximity to the reflection maximum of gold.

Dispersions with constant particle mass $(500\mathrm{fg}\mathrm{Au}∕10\mathrm{pL})$ contain an increasing number of particles with decreasing diameter. Microscopic visualization reveals this number increase for the $100\text{-}\text{to}80\text{-}\mathrm{nm}$ (data not shown) and $80\text{-}\text{to}60\text{-}\mathrm{nm}$ samples compared to AuNPs $>100\mathrm{nm}$ . However, patches of AuNPs $>100\mathrm{nm}$ [Fig. 3] appear bigger than the $80\text{-}\text{to}60\text{-}\mathrm{nm}$ sized particles [Fig 3], resulting in a higher pixel count for the $>100\text{-}\mathrm{nm}$ fraction [Fig. 3]. This probably reflects the size difference, as the reflection of particles in the $>100\text{-}\mathrm{nm}$ fraction might occupy more than one pixel. Particles $40\text{to}60\mathrm{nm}$ in diameter only provide a reduced number of reflective spots [Fig. 3], and from the particles $20\text{to}40\mathrm{nm}$ only single objects contained signal intensities sufficiently high for microscopic detection, despite the 2150 particles included in the visualized volume [Table 2 and Fig. 3]. From AuNPs smaller than $20\mathrm{nm}$ , no reflection could be recorded (data not shown). Thus, the reduced extinction of particles smaller than $60\mathrm{nm}$ measured by UV-vis spectroscopy (Fig. 2) can be correlated with a loss in scattering cross section, confirmed via limited microscopic imaging of these particles. Interestingly, all these findings result in a rather continuous decrease of pixel counts with decreasing particle diameter within CM. In contrast, a sharp drop of reflection was recorded between 60- to 80- and $40\text{-}\text{to}60\text{-}\mathrm{nm}$ sized particles in all three evaluated spectral bands in CN, confirming the incomplete detection of single particles with a diameter $<60\mathrm{nm}$ . In the CM, this reduction of the scattering cross sections with reduced particle size is counteracted by the higher particle numbers and higher proximity of the single particles that enhanced the total reflection. However, it must be recognized that in this case, a bulk of particles adds up to one reflective spot. Furthermore, in the smaller particle groups [Figs. 3 and 3], reflection was dominant at $543\text{-}\mathrm{nm}$ excitation (green), indicating preferential detection of aggregated particles having a phase shift of their SPR to longer wavelengths.^{1, 2, 6, 12, 19, 20} In both series, only the group of $60\text{-}\text{to}80\text{-}\mathrm{nm}$ sized AuNPs [Fig. 3 and 3] shows intensive reflection in all three bandpasses. In groups of particles $>100\mathrm{nm}$ [Figs. 3 and 3], the red band provides the largest clusters, although the total number of pixels reflecting in this bandpass comprises only about 10 to 20% compared to 514- and $543\text{-}\mathrm{nm}$ excitation.

To conclude, visualization of colloidal AuNPs seems to be possible down to particle sizes of $60\mathrm{nm}$ , while for fixed nanoparticles the visualization of single $40\text{-}\text{to}60\text{-}\mathrm{nm}$ sized AuNPs has been reported. Additionally, a broadened absorption spectrum was seen for these surface-spotted AuNPs as described by others.^{2, 3} Doublets of single spherical AuNPs are frequently detectable in TEM images^{9, 21} and could be suggested from the shape of reflective spots we observed from surface-spotted particles in our study.¹⁸ This arrangement could also explain the improved imaging of fixed particles in other studies.^{9, 12} Furthermore, spectral luminescence of AuNPs (size $50\mathrm{nm}\pm 15\mathrm{\%}$ ), spotted on a glass slide, was previously reported by Geddes ⁹ However, from our data, luminescence could not be visualized from colloids of the $60\text{-}\text{to}80\text{-}\mathrm{nm}$ sized group.

The missing reflection of particles sized smaller than $20\mathrm{nm}$ and severe reduction of visibility of particles between 20 and $40\mathrm{nm}$ is in agreement to the very low scattering previously described.^{1, 12} Measuring the SPR-enhanced absorption would therefore be necessary to visualize these size ranges below $40\mathrm{nm}$ for quantification.¹²

3.3.

Quantification of Titrated $60\text{-}\text{to}80\text{-}\mathrm{nm}$ Colloidal Gold Nanoparticles by Laser Scanning Confocal Microscopy

To verify whether single particles or clusters of particles were imaged, a volume of $10\mathrm{pL}$ $(31.4\times 31.4\times 10.2\mu \mathrm{m})$ of a dilution series from the AuNPs $\left(60\text{to}80\mathrm{nm}\right)$ was excited with laser wavelengths at 514, 543, and $633\mathrm{nm}$ , and the reflecting pixels were counted. The stock dispersion of $100\mathrm{fg}∕10\mathrm{pL}$ was diluted down to $1\mathrm{fg}∕10\mathrm{pL}$ in six steps. In the $1\text{-}\mathrm{fg}∕10\mathrm{pL}$ dilution, only one out of ten fields of view provided a signal from four AuNPs. The detection gain was increased by 15% to allow for detection of very low reflecting spots. Therefore, this dilution step was not included in the statistical analysis; however, the data for this group are shown in Figs. 4 and 4 . However, as the particle mass of single particles sized $70\mathrm{nm}$ is $3.47\mathrm{fg}$ , the detection of four reflecting spots in one out of ten fields of view fits nicely with the expected appearance of one particle in about $40\mathrm{pL}$ of this dilution, and proved the detection of single particles of $60\text{-}\text{to}80\text{-}\mathrm{nm}$ size by confocal microscopy.

Fig. 4

(a), (b), and (c) are representative images and (d) and (e) are statistical ANOVA results of square root transformed reflecting pixel numbers in dilution series of AuNPs $\left(60\text{to}80\mathrm{nm}\right)$ . (a) represents a particle concentration of $100\mathrm{fg}∕10\mathrm{pL}$ , (b) stands for $10\mathrm{fg}∕10\mathrm{pL}$ , and (c) for $2\mathrm{fg}∕10\mathrm{pL}$ . The square root transformed pixel numbers (d) and (e) correspond to the expected reduction of pixels from separate detection of single nanoparticles only in the more diluted samples. Pixel size is $80\times 80\times 700\mathrm{nm}$ , pixel time is $4.32\mu \mathrm{s}$ , image is $25\times 25\times 10\mu \mathrm{m}$ , and scale $1\mu \mathrm{m}$ . Asterisks show colloids of the $1\text{-}\mathrm{fg}$ dilution that provided reflection only after increased detection gains and were excluded from the statistics.

The minimum pixel size from $80\times 80\mathrm{nm}$ up to $220\times 220\mathrm{nm}$ applied in the current study is not suited to reflect the true size differences of AuNPs between $40\text{to}100\mathrm{nm}$ . Moreover, a summary of pixels at one position might reflect agglomerates of AuNPs^{1, 2} or the pathway of a single particle moving freely in aqueous environment during acquisition (Fig. 3). Thus, single particle detection is most accurate for AuNPs of $60\text{to}80\mathrm{nm}$ , if the interparticle distance is high enough to separate the signals of two individual AuNPs (Fig. 4).

3.4.

Visibility of Gold Nanoparticles after $48\text{-hours}$ of Coincubation within Bovine Immortalized Cells and Mass Spectrometric Validation

The potential of conventional confocal microscopy to quantify AuNPs was reinvestigated in bovine immortalized cells that had been coincubated with AuNPs for $48\mathrm{h}$ . Reflection intensity increased in the cellular environment, and the spectral characteristics were shifted to longer wavelengths. Luminescence was not recorded after excitation at 543 or $633\mathrm{nm}$ from colloidal AuNPs, but was apparent from AuNPs within cells (Fig. 5 ). Although due to the cellular background, the detection offset was raised for the discrimination of AuNPs within cells and the gain reduced, the recorded intensities of reflection increased for all bands of detection.

Fig. 5

Size restricted AuNPs ( $10\text{-}\mathrm{mM}$ Au) of a mixed sample with mean particle size of $60\text{to}80\mathrm{nm}$ [(a) and (d)], size selected AuNPs of $40\text{to}60\mathrm{nm}$ [(b) and (e)], and of $15\mathrm{nm}$ [(c) and (f)] are visualized as colloids in $\mathrm{dd}{\mathrm{H}}_2\mathrm{O}$ [(a) and (c)] and after $48\mathrm{h}$ of coincubation in GM7373 cells [(d), (e), and (f)]. Pixel size is $90\times 90\times 600\mathrm{nm}$ , pixel time is $1.28\mu \mathrm{s}$ , and image is $33\times 33\times 12\mu \mathrm{m}$ . The figures show the overlay of four imaged channels each. The dispersed AuNPs were visualized in reflection bands after excitation at 543 and $633\mathrm{nm}$ , respectively, and two luminescence bands (at 543-, and $633\text{-}\mathrm{nm}$ excitation) as described in Table 3. However, exclusively SPR was visualized from the dispersed AuNPs of all size groups resulting in the green dominated false color images. The AuNPs in cells [(d), (e), and (f)] are shown as an overlay of reflection after $543\text{-}\mathrm{nm}$ excitation, and luminescence after $543\text{-}\mathrm{nm}$ and $633\text{-}\mathrm{m}$ excitation added to the DIC transmission defining the position of the single cells. The orthogonal views at a single optical slice of the series of $900\text{-}\mathrm{nm}$ -thick optical sections through a single cell (g) in comparison to the 3-D projection of this cell (h) indicate the within-cell location of several nanoparticles, scale $5\mu \mathrm{m}$ . (Color online only.)

Imaging conditions were optimized for dispersed AuNPs achieving 100 to 200 gray values as background and avoiding intensities above the differentiation range. Typical intensity distributions are demonstrated in supplementary information (Figs. 6 and 7 ). Only pixels with intensities above 700 were included in the quantification to ensure a signal-to-noise ratio (SNR) above 3. Signals with low intensities were recorded typically above 1000, but most spots reached intensities of 2000 to 3500 from 4096 gray values.

Fig. 6

Intensity profiles of three channels from the volume of $10\mathrm{pL}$ , including dispersed $60\text{-}\text{to}80\text{-}\mathrm{nm}$ AuNPs of Fig. 5. The green line represents SPR after excitation at $514\mathrm{nm}$ . The background values of the blue and red lines were recorded in the luminescence mode after 514- and $633\text{-}\mathrm{nm}$ excitation. (Color online only.)

Fig. 7

Intensity profiles of four channels from a single optical slice of a Z-series through cells demonstrated as 3-D projection in Fig. 5. The green line represents SPR after $514\text{-}\mathrm{nm}$ excitation. Within the cells, additional signals were recorded from the same AuNPs in luminescence after excitation at 514 and $633\mathrm{nm}$ . The gray line corresponds to the DIC transmission. (Color online only.)

The localization of AuNPs in cells appears to inhibit the free movement of AuNPs as seen in colloids, and thus enables counting of reflecting objects instead of pixels in cells [Figs. 5, 5, 5]. This confirms previous data where the enclosure of single AuNPs in distinct electron dense structures after diffusion in the cells was demonstrated.⁴ Moreover, a high degree of colocalization for the different reflection bands was obvious in contrast to the dispersed particles. Although reflection bands of 505 to 545 and $>585\mathrm{nm}$ were recorded for the dispersed AuNPs as well as from the cells, AuNPs were provided only from cell reflection signals and additional detectable luminescence signals from the longer wavelengths, indicating a further red shift of the SPR due to the increased refractive index of the cellular environment.

The impressive SNR and stability of the reflection compared to the fading of fluorescence dyes is further augmented if the unfixed cells, washed only once in phosphate buffered saline (PBS), were investigated (Fig. 8 ). Low excitation of autofluorescence within the reflection bandpass enabled similar clear differentiation of the AuNPs within the cells.

Fig. 8

Cells coincubated with $60\text{-}\text{to}80\text{-}\mathrm{nm}$ sized AuNPs for $48\mathrm{h}$ . Visualization was performed from unfixed cells washed once in PBS and mounted in Vectashield. Scale $5\mu \mathrm{m}$ .

After the $48\text{-}\mathrm{h}$ coincubation period with AuNPs sized $60\text{to}80\mathrm{nm}$ , a median of 11.6 particles per cell was found in 100 cells (Table 4 ). Coincubation of cells with the mixed size sample (average particle diameter $15\mathrm{nm}$ ) and AuNPs sized $40\text{to}60\mathrm{nm}$ resulted in particle numbers with a median of 13.8 and 9.1 particles per cell, respectively. As an independent verification of the microscopically counted number of AuNPs inside cells, gold mass spectrometry was performed from a fraction of the same cell cultures. The number of particles recalculated from the total mass of gold found in approximately $3\times {10}^6$ cells confirmed that the particle counts for the size group of $60\text{to}80\mathrm{nm}$ , retrieved from the microscopic evaluation, were accurate (Table 4). These data also revealed by this visualization independent measure that particles smaller than $60\mathrm{nm}$ were only incompletely visualized by confocal microscopy. A mean of nine particles could be counted, but a mean 27 particles per cell had been calculated from the mass spectrometry in the sample incubated with particles of $40\text{to}60\mathrm{nm}$ diameter. Especially for the size group of about $15\mathrm{nm}$ , only a minute number of the AuNPs that penetrated the cells were detected microscopically by their scattering cross sections. Only 13.8 spots per cell were visualized; however, 8634 particles per cell must have been present within the cells according to the mass spectrometric data recorded from the same cell culture (Table 4).

Table 4

Mass spectrometric and microscopic quantification of gold (Au) in cells coincubated with AuNPs of different size classes. The asterisk indicates cells had been counted in individual samples after coincubation in uniform 3miocells∕ml and final processing.

Taking into account that the colloid AuNPs could not be detected if they had been smaller than $40\mathrm{nm}$ , it is somewhat surprising that we saw any particles in the mixed AuNP group with a mean of $15\mathrm{nm}$ diameter. One possible explanation is that the detected particles could be the few AuNPs of bigger size, which are to be expected in this group. On the other hand, signals detected with the microscope could also have come from agglomerated nanoparticles.

It was also surprising that ICPMS analysis found such tremendous amounts of particles with mean diameters of $15\mathrm{nm}$ inside cells. Previous studies²² have indicated the best penetration efficiency for $50\text{-}\mathrm{nm}$ AuNPs compared to smaller ones. However, in those cases, no visualization independent measure of AuNPs was evaluated. Nevertheless, our previous studies revealed diffusion as the most probable mechanism for cellular uptake of unconjugated, laser-generated AuNPs.⁴ The different surface characteristics of laser-generated AuNP in contrast to chemically synthesized ones might also be responsible for these differences in cell penetration.

Diffusion, a main pathway for cell penetration, was not discussed yet for AuNPs. Instead, endosomal pathways were presumed as uptake mechanisms of these gold nanoparticles into the cells. Taylor ⁴ demonstrated diffusion as the most feasible mechanism for cell penetration of the unconjugated AuNPs. As all the size-restricted samples of unconjugated laser-generated AuNPs appear in small discrete signals throughout the cells, the presence of oxidized ${\mathrm{Au}}^{+}$ and ${\mathrm{Au}}^{3+}$ providing a positive surface charge might be responsible for this unique behaviour.