2.1.

Materials

Carboxylate-modified ${\mathrm{FluoSpheres}}^{\circledR }$ [100-nm, yellow-green( ${\lambda }_{\mathrm{ex}}=488\mathrm{nm}$ , ${\lambda }_{\mathrm{em}}=515\mathrm{nm})$ , 2% solids] were obtained from Molecular Probes. Poly(styrene sulfonic acid) sodiumsalt [PSS, molecular weight $\left(\mathrm{MW}\right)\sim 1\mathrm{MDal}$ , Sigma) andpoly(allylamine hydrochloride) (PAH, $\mathrm{MW}\sim 200\mathrm{kDal}$ , Sigma) were used in polyelectrolyte film formation. Separate solutions of PAH and PSS $(2\mathrm{mg}∕\mathrm{mL},0.1\text{-}\mathrm{MKCl})$ were prepared from deionized water and stored at room temperature. Tris(4,7-diphenyl-1,10-phenanthroline)ruthenium(II) dichloride $\left(\right[\mathrm{Ru}\left({\mathrm{Ph}}_2\mathrm{phen}{)}_3{\mathrm{Cl}}_2\right]$ , $\mathrm{MW}=1169$ , Fluka) was used as the oxygen-sensitive fluorophore. The molecular structure of the dissociated sensing cation $\left[\mathrm{Ru}\right({\mathrm{Ph}}_2\mathrm{phen}{)}_3{]}^{2+}$ is shown in Fig. 1 . $\left[\mathrm{Ru}\right({\mathrm{Ph}}_2\mathrm{phen}{)}_3{\mathrm{Cl}}_2]$ was dissolved in a $0.11\text{-}\mathrm{mM}$ stock solution (37% methanol, 63% deionized water by volume) containing $6.67\text{-}\mathrm{mM}\mathrm{NaOH}$ .

Fig. 1

Molecular structure of tris(4,7-diphenyl-1,10-phenanthroline)ruthenium(II) $\left[\mathrm{Ru}\right({\mathrm{Ph}}_2\mathrm{phen}{)}_3{]}^{2+}$ (dissociated form of dichloride salt).

Poly(ethyleneimine) (PEI, $\mathrm{MW}\sim 750\mathrm{kDal}$ , Sigma) was dissolved $(2\mathrm{mg}∕\mathrm{mL})$ in type I water for use in delivery studies. Human dermal fibroblasts (HDFs) were obtained from Cascade Biologics. Growth media [Dulbecco’s minimum essential medium (DMEM), low glucose] and fetal bovine serum (FBS) were obtained from Atlanta Biologicals. Plastic microplates (96 wells) were obtained from Fisher Scientific. Hank’s balanced salt solution (HBSS) was obtained from Sigma in powder form and dissolved in type I water.

2.2.

Characterization of $\left[\mathrm{Ru}\right({\mathrm{Ph}}_2\mathrm{phen}{)}_3{]}^{2+}$ Adsorption on Quartz Crystal Microbalance (QCM) Templates

To analyze dye uptake using the insolubility-induced precipitation technique, multilayer films were fabricated on QCM resonators using standard methods.⁷ Alternate PAH/PSS adsorption cycles were monitored using QCM analysis with resonators having evaporated silver electrodes with combined surface area of $0.16{\mathrm{cm}}^2$ and initial resonance frequency of $9\mathrm{MHz}$ (AT-cut, Sanwa Tsusho). During the assembly procedure, each resonator was dipped into the cleaning solution [39% EtOH, 1% KOH, 60% deionized (DI) water] and sonicated in a Branson 1510 ultrasonic bath to create a negative surface charge on the resonator. The resonator was then immersed in a polyelectrolyte solution for $15\min$ to allow complete surface saturation, rinsed twice by immersion in water, and dried with streaming nitrogen. The resonance frequency was then measured using a USI SC-7201 frequency counter. Films with architectures ranging from $\{\mathrm{PAH}∕\mathrm{PSS}{\}}_1$ to $\{\mathrm{PAH}∕\mathrm{PSS}{\}}_5$ were deposited on different QCM resonators. The subscripts 1 and 5 represent the number of bilayers of PAH/PSS deposited. To assess the influence of surface-charge on the immobilization process, films with architectures of $\{\mathrm{PAH}∕\mathrm{PSS}{\}}_1\mathrm{PAH}$ to $\{\mathrm{PAH}∕\mathrm{PSS}{\}}_5\mathrm{PAH}$ were also deposited on QCM resonators. A control sample, with no polyelectrolyte multilayers, was also analyzed. For $\left[\mathrm{Ru}\right({\mathrm{Ph}}_2\mathrm{phen}{)}_3{]}^{2+}$ adsorption, the resonator was immersed in dye stock solution for $30\min$ . The change in mass due to deposition the final polyelectrolyte layer was determined using the frequency counter.

2.3.

Characterization of $\left[\mathrm{Ru}\right({\mathrm{Ph}}_2\mathrm{phen}{)}_3{]}^{2+}$ Adsorption on FluoSpheres ${}^{\circledR }$

Films with architecture $\{\mathrm{PAH}∕\mathrm{PSS}{\}}_3$ were constructed on fluorescent nanoparticle templates using standard methods.¹⁶ Initially, $40\mathrm{\mu }\mathrm{L}$ of nanoparticles $(\sim 1.5\times {10}^{15}\mathrm{particles})$ was used for coating. The LbL assembly process for deposition of ultrathin films on nanoparticles is shown in Fig. 2 . The nanoparticles were suspended alternatively in $1\mathrm{mL}$ of PAH and PSS solutions for 20-min intervals, with three standard rinse cycles in DI water between each polyelectrolyte step. Standard rinse cycles included centrifugation at $13,000\mathrm{rpm}$ for $1\mathrm{h}$ in an Eppendorf 5804R centrifuge and sonication for $1\min$ . After each layer was deposited, a $10\text{-}\mathrm{\mu }\mathrm{L}$ sample of the 1-mL volume of coated nanoparticles was diluted in $2\mathrm{mL}$ DI water, and the surface potential of the modified particles was measured using the ZetaPlus $\zeta$ -potential analyzer (Brookhaven Instruments). Once the desired film architecture was attained, a total of three bilayers of $\{\mathrm{PAH}∕\mathrm{PSS}\}$ , the nanoparticles were suspended in $1\mathrm{mL}$ of $\left[\mathrm{Ru}\right({\mathrm{Ph}}_2\mathrm{phen}{)}_3{]}^{2+}$ stock solution for $24\mathrm{h}$ . Following this period the nanoparticles were rinsed three times in DI water.

Fig. 2

Schematic for sensor production process. Ultrathin films are deposited on 100-nm fluorescent templates using L & L assembly. $\left[\mathrm{Ru}\right({\mathrm{Ph}}_2\mathrm{phen}{)}_3{]}^{2+}$ precipitation into the nanofilms is shown as the final step in sensor fabrication.

To determine the total mass of $\left[\mathrm{Ru}\right({\mathrm{Ph}}_2\mathrm{phen}{)}_3{]}^{2+}$ adsorbed into the polyelectrolyte multilayer films on nanoparticles, the concentration of removed $\left[\mathrm{Ru}\right({\mathrm{Ph}}_2\mathrm{phen}{)}_3{\mathrm{Cl}}_2]$ supernatant was determined using a premeasured absorbance calibration curve. The sample was irradiated using an Ocean Optics LS-1 tungsten halogen lamp, and absorbance data were collected for the diluted supernatant using an Ocean Optics USB2000 spectrophotometer. The absorbance at $463\mathrm{nm}$ was plotted against a $\left[\mathrm{Ru}\right({\mathrm{Ph}}_2\mathrm{phen}{)}_3{]}^{2+}$ absorbance-concentration profile, obtained through successive dilution of the stock $\left[\mathrm{Ru}\right({\mathrm{Ph}}_2\mathrm{phen}{)}_3{\mathrm{Cl}}_2]$ solution, and the concentration of $\left[\mathrm{Ru}\right({\mathrm{Ph}}_2\mathrm{phen}{)}_3{\mathrm{Cl}}_2]$ in the final supernatant was determined. The removed supernatant concentration was subtracted from the loading solution concentration to determine the mass of $\left[\mathrm{Ru}\right({\mathrm{Ph}}_2\mathrm{phen}{)}_3{]}^{2+}$ adsorbed.

To assess the stability of the immobilized dye-nanoparticle system, absorbance data for supernatant of the nanosensor suspension were collected over a 28-day period following ${\mathrm{O}}_2$ studies. This was performed to determine the amount of indicator released from the nanofilms under normal conditions of wet storage or use in aqueous media. The nanosensor solution was centrifuged and the supernatant was removed four times during the 4-week period for absorbance studies. The supernatant absorbance data were collected and compared to the absorbance of the original suspension.

2.4.

Characterization of Sensor Response to Changes in Bulk Oxygen Concentration

To determine the response of the $\left[\mathrm{Ru}\right({\mathrm{Ph}}_2\mathrm{phen}{)}_3{]}^{2+}$ -nanoparticle sensors to changes in ${\mathrm{O}}_2$ concentration, the final suspension of nanosensors (in $2\mathrm{mL}$ DI water) was placed in a cuvette into which ${\mathrm{O}}_2$ and ${\mathrm{N}}_2$ were bubbled into the chamber at random concentrations using individual regulators. To minimize time-correlated nonspecific effects, ${\mathrm{O}}_2$ concentrations were adjusted alternatively higher or lower for each successive measurement. Oxygen measurements were collected from a Unisense PA2000 picoammeter with an OX500 micro-oxygen electrode connected to a PC running LABVIEW with a data acquisition (DAQ) card. An Hg-lamp with a 460-nm emission filter was used to excite the nanosensors using a fiber probe; emission spectra were collected with a fiber probe oriented perpendicularly to the excitation probe, and the spectra were analyzed using an Ocean Optics spectrometer. Three spectra were collected for each oxygen concentration, and the data were averaged and normalized to the maximum peak for green emission $\left(515\mathrm{nm}\right)$ . The normalized fluorescence intensity at $614\mathrm{nm}$ was used as the ratiometric indicator of oxygen.

A similar experiment was performed to evaluate the susceptibility of the sensors to interference by proteins or peptides in the surrounding solution, as might be encountered in the cytosol. Using a separate batch of sensors, which were produced using different loading conditions (ruthenium to nanoparticle concentration) and were stored wet (in DI water) for $>10$ months, sensors were split into separate cuvettes for comparison experiments. Both samples initially comprised nanosensors in DI water, while poly(L-lysine) (PLL) and FBS were added to the second and third samples. For five different oxygen concentrations ranging from 0 to 100%, spectra were measured (three at each concentration), and Stern-Volmer curves were constructed to determine whether the presence of protein and/or peptide resulted in significantly different quenching behavior.

2.5.

Intracellular Delivery

Sensors were produced as described above, using sterilized water and materials, and all procedures were performed under a laminar flow hood. Human dermal fibroblasts (passage six) were cultured in 96-well plates using DMEM supplemented with 10% FBS (V/V). The $\left[\mathrm{Ru}\right({\mathrm{Ph}}_2\mathrm{phen}{)}_3{]}^{2+}$ -nanoparticle sensors were coated with PEI, a cationic polymer frequently used to complex with DNA for transfection, using electrostatic adsorption. PEI-coated sensors suspended in sterilized type I water were added directly to cells preplated and adhered onto 96-well plates. Based on previously reported methods,³⁹ the total number of particles added to each well was varied from ${10}^8$ to ${10}^{10}$ . After 48-h exposure to PEI-coated nanoparticles, cells were rinsed with HBSS, and media was reintroduced before imaging with a Leica TCS SP2 laser scanning confocal microscope. Nomarski phase and fluorescence images were collected for the same focal plane to enable coregistration of cellular boundaries and nanoparticle localization.

3.1.

Mass Deposition Measured as Adsorption onto QCM Resonators

Mass deposited as a function of layer number for the $\{\mathrm{PAH}∕\mathrm{PSS}{\}}_3$ QCM sample is shown in Fig. 3 . A slower rate of film growth is observed for the first two bilayers, which is in agreement with previous reports.⁴⁰ An increase in mass and, therefore, film thickness during the assembly process is evident, with near-linear film growth for the phase after two bilayers. QCM data from other film architectures were observed to show similar trends. The mass increase associated with the $\left[\mathrm{Ru}\right({\mathrm{Ph}}_2\mathrm{phen}{)}_3{]}^{2+}$ adsorption step is also shown in the graph (adsorption cycle 7), though this cannot necessarily be related to a thickness increase due to the penetration of the small molecules into the preadsorbed films.

Fig. 3

Growth of $\{\mathrm{PAH}∕\mathrm{PSS}{\}}_3$ film and $\left[\mathrm{Ru}\right({\mathrm{Ph}}_2\mathrm{phen}{)}_3{]}^{2+}$ adsorption during QCM experiments, with thickness estimated from mass measurements.

Measured mass of $\left[\mathrm{Ru}\right({\mathrm{Ph}}_2\mathrm{phen}{)}_3{]}^{2+}$ adsorbed as a function of film thickness (estimated from adsorbed polyelectrolyte mass using the Sauerbrey equation) is shown in Fig. 4 . $\left[\mathrm{Ru}\right({\mathrm{Ph}}_2\mathrm{phen}{)}_3{]}^{2+}$ adsorption was measurable on the control sample (resonator without film) and samples with a PAH terminal layer. This suggests that some $\left[\mathrm{Ru}\right({\mathrm{Ph}}_2\mathrm{phen}{)}_3{]}^{2+}$ precipitates directly onto the surface of the crystal. However, $\left[\mathrm{Ru}\right({\mathrm{Ph}}_2\mathrm{phen}{)}_3{]}^{2+}$ adsorption increased threefold in PSS-terminated films of greater than $5\mathrm{nm}$ thickness. These findings suggest that $\left[\mathrm{Ru}\right({\mathrm{Ph}}_2\mathrm{phen}{)}_3{]}^{2+}$ adsorption into negatively charged films is an exponential function of film thickness. The data were fit to an exponential relationship using an equation of the form

Fig. 4

Mass of $\left[\mathrm{Ru}\right({\mathrm{Ph}}_2\mathrm{phen}{)}_3{]}^{2+}$ adsorbed as an exponential function of film thickness (estimated from mass measurements). The amount of $\left[\mathrm{Ru}\right({\mathrm{Ph}}_2\mathrm{phen}{)}_3{]}^{2+}$ deposited is shown for both PSS-terminal $(◆)$ and PAH-terminal $(▴)$ nanofilms with different numbers of layers.

In contrast to this exponential relationship observed for PSS-terminated polyelectrolyte nanofilms, negligible changes in mass of dye adsorbed relative to the control sample were measured for films terminated with PAH, regardless of the number of polyelectrolyte layers deposited. This indicates that the use of PAH as a terminal layer serves to limit rather than promote $\left[\mathrm{Ru}\right({\mathrm{Ph}}_2\mathrm{phen}{)}_3{]}^{2+}$ adsorption. Poor adsorption of another ruthenium compound into PAH-terminated films was also reported previously.²³ Poor adsorption is likely caused either by surface-charge or water-affinity differences between the terminal polyelectrolytes.

$\left[\mathrm{Ru}\right({\mathrm{Ph}}_2\mathrm{phen}{)}_3{]}^{2+}$ adsorption in films fabricated using LbL assembly is attributable to a difference in water concentration between the stock $\left[\mathrm{Ru}\right({\mathrm{Ph}}_2\mathrm{phen}{)}_3{\mathrm{Cl}}_2]$ solution and the $\{\mathrm{PAH}∕\mathrm{PSS}{\}}_n$ film. As $\left[\mathrm{Ru}\right({\mathrm{Ph}}_2\mathrm{phen}{)}_3{]}^{2+}$ passes through the film, it is immobilized within it. Farhat also studied water concentration in films,⁴¹ and Radtchenko showed that polyelectrolytes used in LbL assembly promote high local water concentrations in films even in near-100% acetone solutions.³⁵ This is likely due to the presence of the less hydrophobic polyelectrolyte PSS.

The maximum mass of $\left[\mathrm{Ru}\right({\mathrm{Ph}}_2\mathrm{phen}{)}_3{]}^{2+}$ adsorbed to the coated QCM resonator is $330\mathrm{ng}$ , which is approached as film thickness approaches infinity (Fig. 4). Translated into a depth limitation for adsorption (penetration of the dye into the films), the data suggest that $\left[\mathrm{Ru}\right({\mathrm{Ph}}_2\mathrm{phen}{)}_3{]}^{2+}$ adsorption is highly controllable within $10\mathrm{nm}$ . Tedeschi ²⁴ reported a depth limitation for adsorption of an anionic pyrene chromophore (4-PSA) to be $17\mathrm{nm}$ . Differences in depth limitations result from different molecule sizes, different methods of adsorption, or different film compositions. Thus, further control may be easily added by working with alternate materials for the nanofilms, and development of this capability will be a subject of future study.

3.2.

Adsorption onto Nanofilm-Coated ${\mathrm{FluoSpheres}}^{\circledR }$

Zeta potential as a function of polyion bilayer number for negatively charged nanoparticles alternatively coated with PAH and PSS is shown in Fig. 5 . After deposition of each consecutive polyelectrolyte layer, the measured voltage reverses direction. This voltage reversal implies a reversal of surface charge. These results, combined with QCM assembly data, verify that films were formed on the surfaces of the nanoparticles. No quantitative conclusions are made from the zeta potential data.

Fig. 5

Zeta potential of polyelectrolyte multilayer films on nanoparticles, as measured during the coating process.

After 24 h of $\left[\mathrm{Ru}\right({\mathrm{Ph}}_2\mathrm{phen}{)}_3{]}^{2+}$ adsorption, only 4.2% of the original $\left[\mathrm{Ru}\right({\mathrm{Ph}}_2\mathrm{phen}{)}_3{\mathrm{Cl}}_2]$ remained in the solution, with the balance immobilized within the multilayers. Assuming no loss of nanoparticles, this translates into an effective $\left[\mathrm{Ru}\right({\mathrm{Ph}}_2\mathrm{phen}{)}_3{]}^{2+}$ concentration within films of approximately $1.2\mathrm{mg}∕\mathrm{mL}$ , or six times the $\left[\mathrm{Ru}\right({\mathrm{Ph}}_2\mathrm{phen}{)}_3{]}^{2+}$ concentration of the original $\left[\mathrm{Ru}\right({\mathrm{Ph}}_2\mathrm{phen}{)}_3{\mathrm{Cl}}_2]$ solution. Taking into account nanoparticle losses during the coating and rinsing process, the $\left[\mathrm{Ru}\right({\mathrm{Ph}}_2\mathrm{phen}{)}_3{]}^{2+}$ density in films on nanoparticles is much higher.

Nanoparticles with $\left[\mathrm{Ru}\right({\mathrm{Ph}}_2\mathrm{phen}{)}_3{]}^{2+}$ adsorbed into multilayer films were analyzed using a transmission electron microscope (TEM) to assess dye immobilization. A TEM image of several nanoparticle sensors is shown in Fig. 6 . After $\left[\mathrm{Ru}\right({\mathrm{Ph}}_2\mathrm{phen}{)}_3{]}^{2+}$ adsorption, the nanoparticles appear round. The TEM results verify that dye adsorption does not produce surface abnormalities, suggesting that the polyelectrolyte films and dye adsorb uniformly over the surface of the nanoparticles.

Fig. 6

TEM image of nanoparticles with $\left[\mathrm{Ru}\right({\mathrm{Ph}}_2\mathrm{phen}{)}_3{]}^{2+}$ adsorbed into surface coatings of multilayer nanofilms.

Absorbance data for the storage supernatant of the nanosensors indicate less than 5% loss of $\left[\mathrm{Ru}\right({\mathrm{Ph}}_2\mathrm{phen}{)}_3{]}^{2+}$ over a 28-day of storage period in DI water. Furthermore, sensors that were rinsed once following 28 days of storage were then stored wet for 9 additional months, and the ruthenium absorbance in the supernatant at the end of that period was negligible. These findings suggest that $\left[\mathrm{Ru}\right({\mathrm{Ph}}_2\mathrm{phen}{)}_3{]}^{2+}$ -nanoparticle nanosensors will not lose significant levels of dye during short-term use, such that ratiometric experimental observations should be reliable and should not be expected to have significant adverse effect on cells. Furthermore, the sensors have a long shelf life and can therefore be made in large quantities to be used at a later time.

3.3.

Sensor Response to Changes in Bulk Oxygen Concentration

Normalized $\left[\mathrm{Ru}\right({\mathrm{Ph}}_2\mathrm{phen}{)}_3{]}^{2+}$ -nanoparticle spectra at several ${\mathrm{O}}_2$ concentrations collected during ${\mathrm{O}}_2$ sensitivity experiments are shown in Fig. 7(a) . The top spectrum was collected at 0% ${\mathrm{O}}_2$ , the bottom spectrum was collected at 100% ${\mathrm{O}}_2$ , and the middle spectra were collected at intermediate ${\mathrm{O}}_2$ concentrations. $\left[\mathrm{Ru}\right({\mathrm{Ph}}_2\mathrm{phen}{)}_3{]}^{2+}$ is quenched in the presence of molecular oxygen, so lower relative $\left[\mathrm{Ru}\right({\mathrm{Ph}}_2\mathrm{phen}{)}_3{]}^{2+}$ fluorescence is expected at higher ${\mathrm{O}}_2$ concentrations. The figure reveals that the relative intensity of $\left[\mathrm{Ru}\right({\mathrm{Ph}}_2\mathrm{phen}{)}_3{]}^{2+}$ decreases with increasing concentration of dissolved ${\mathrm{O}}_2$ . The $\left[\mathrm{Ru}\right({\mathrm{Ph}}_2\mathrm{phen}{)}_3{]}^{2+}$ in the sample was quenched by approximately 50% from 0% ${\mathrm{O}}_2$ to 100% ${\mathrm{O}}_2$ . Figure 7(b) is a plot of relative $\left[\mathrm{Ru}\right({\mathrm{Ph}}_2\mathrm{phen}{)}_3{]}^{2+}$ fluorescence $(614\mathrm{nm}$ ) versus oxygen concentration, as determined using the micro-oxygen electrode.

The sensitivity of the nanoparticles to oxygen can be computed using the equation

It is well known that the environment of the dye determines the response behavior, and it is possible that the sensitivity of these systems could be increased by using alternate polyelectrolyte in the films used to adsorb the dye. In addition, photostability depends greatly on the immobilization matrix; it is believed that “caging” (local trapping by the support material) of the highly reactive singlet oxygen produced in the quenching reaction may increase photooxidative degradation rates of the ground-state metal-ligand complexes.⁴⁴ Thus, matrices that allow access of solvent molecules to the complexes generally promote greater stability, and it is logical to predict that the nanofilm-based metal complex entrapment methods may lower photobleaching rates due to short diffusion lengths. Measurements of photobleaching using a fluorescence microscope to illuminate nanosensors with $173\text{-}\mathrm{\mu }\mathrm{W}$ excitation light for $1\mathrm{h}$ showed less than 0.5% change in average fluorescence intensity ratio, suggesting that photobleaching of the sensors will be minimal for the intended application. Note also that the nanofilm properties can also be tailored through selection of materials and assembly conditions, leading to differing solvent and oxygen transport properties, which will affect both sensitivity and photodegradation. Furthermore, as recent reports have shown, additional improvements in sensitivity can be gained by moving to alternative long-lifetime materials such as platinum-based complexes.¹⁰ These strategies will be employed in future studies using the facile approach to sensor production.

The modified Stern-Volmer plot for ratiometric data is given in Fig. 8(a) , where $F_0$ and $F$ are intensities at $614\mathrm{nm}$ , normalized to the emission intensity at $515\mathrm{nm}$ ( $F_0$ is the relative intensity at zero oxygen concentration). The profile exhibits downward curvature, which matched well with a second-order polynomial $(R^2=0.9994)$ , as shown in the figure, where $Q$ is the concentration of ${\mathrm{O}}_2$ . For these data, a first-order quenching constant $\left(A\right)$ of $0.74∕\mathrm{mM}$ , or $0.0128∕\mathrm{Torr}$ , was computed; this linear slope matches well with published values for $\left[\mathrm{Ru}\right({\mathrm{Ph}}_2\mathrm{phen}{)}_3{]}^{2+}$ as oxygen sensors.^{30, 45} However, the downward-curved shape of the Stern-Volmer plot proves that the indicator response deviates from an ideal single-state system. This type of behavior can be attributed to several factors, including spectral overlap between the emission of reference and indicator and possibly two populations of fluorophores with differing quenching constants. The first possibility, spectral overlap, can be minimized by mathematical correction of the ratio calculation or spectral deconvolution (e.g., principal component analysis) using the pure spectra of the two fluorophores. These are not complicated steps, but do require additional postexperiment processing that is not conducive to direct, real-time measurements.

Fig. 8

Stern-Volmer plot for $\left[\mathrm{Ru}\right({\mathrm{Ph}}_2\mathrm{phen}{)}_3{]}^{2+}$ quenching in $\{\mathrm{PAH}∕\mathrm{PSS}{\}}_3$ films on nanoparticles, where $F_0∕F$ refers to the division of the fluorescence intensity ratio (indicator/reference) in the absence of oxygen $\left(F_0\right)$ by the fluorescence intensity ratio at the given oxygen concentration. (a) Raw values and values computed after compensating for spectral overlap. The lines denote second-order polynomial (raw data) and linear (corrected data) fits. Error bars denote one standard deviation of three measurements at each concentration. (b) Stern-Volmer plots for nanosensors with and without ambient proteins, using raw data. Equations are second-order polynomial fits.

Investigation of the optical properties of the system revealed spectral overlap with the ruthenium emission from both the yellow/green fluorescent nanoparticle emission as well as a small contribution from to the arc lamp due to the fiber optic sampling system that collects significant reflected light (this is noticeable from a small peak appearing at $575\mathrm{nm}$ in the measured fluorescence emission spectrum). The normalized value of this offset at $614\mathrm{nm}$ was calculated to be 17.5% of the fluosphere intensity at $515\mathrm{nm}$ . In contrast, the contribution to the 515-nm intensity from the source and ruthenium was found to be negligible. Thus, the fluorescence ratios were recalculated after subtracting 0.175 from the normalized intensity at $614\mathrm{nm}$ . Following this adjustment, the Stern-Volmer curve was found to be highly linear $(R^2=0.9996)$ with a quenching constant of $K_{\mathrm{SV}}=1.06∕\mathrm{mM}$ . Because of this excellent fit, consideration of more complex (e.g., two-state^{29, 46}) models for the system behavior was not necessary. Additionally, by applying this correction factor, the new sensitivity for the system was calculated to be $S=60\%$ , an improvement in apparent sensitivity of more than 30%. Thus, removal of the static spectral overlap due to the reference emission and source-specific offset value reveals that the indicator, immobilized on fluorescent templates in a facile manner within a self-assembled nanofilm, is in an environment that provides sensitivity similar to that observed for more complex approaches.^{9, 43}

The sensitivity of the system to the biological environment is also worthy of consideration. The results of comparative quenching experiments with and without poly(lysine) present [Fig. 8(b)] suggest minimal interference by the ambient peptide. The first-order coefficient for the power-law curve was identical, while the second-order coefficient decreased by approximately 30%. Similar results were observed for nanosensors in solution containing serum proteins; the emission spectra were slightly altered, but the Stern-Volmer plots were nearly identical to that calculated for the sensors with and without poly(lysine). The low susceptibility of the sensors to the presence of peptide/protein is likely because the dye is mainly buried beneath an outer layer of poly(ethyleneimine) that was applied after adsorption of the dye. Note also that the sensors used for these experiments [Fig. 8(b)] came from a different batch of sensors than those in the initial oxygen sensitivity experiments [Figs. 7, and 8(a)], where different relative concentrations of ruthenium to nanoparticles were used; therefore, the curves in the inset should be compared directly to one another, but not to the curve in the main figure.

Fig. 7

Results of ${\mathrm{O}}_2$ sensitivity experiment: (a) normalized fluorescence spectra $({\lambda }_{\mathrm{ex}}=460\mathrm{nm})$ for varying ${\mathrm{O}}_2$ concentration and (b) normalized $\left[\mathrm{Ru}\right({\mathrm{Ph}}_2\mathrm{phen}{)}_3{]}^{2+}$ emission as a decreasing function of ${\mathrm{O}}_2$ concentration. Error bars denote one standard deviation for three measurements.

3.4.

Intracellular Delivery

Representative confocal fluorescence microscope and Nomarski phase images of PEI-coated nanoparticle sensors in human dermal fibroblasts are shown in Fig. 9 . After a 48-h exposure, it was observed that a large fraction of nanoparticles were endocytosed by cells prior to rinsing. After washing with HBSS, noninternalized particles were removed and all remaining particles appeared to remain localized within the intracellular space. In many cases, the particles were not uniformly distributed in the cytoplasm, but rather became congregated in the perinuclear region. The human dermal fibroblasts remained viable, as determined with a propidium iodide test, and continued to grow and divide at consistent rates for 2 weeks following nanosensor introduction. The cells were passaged multiple times using standard techniques, and no deleterious effects of the nanosensors were observed. These findings indicate that internalization of PEI-coated nanosensors does not negatively interfere with cellular functions.

Fig. 9

PEI-coated nanoparticle sensors delivered to human dermal fibroblasts: (a) confocal fluorescence image obtained by collecting 520 to 600-nm emission due to 488-nm excitation, (b) Nomarski phase contrast transmission image of HDFs using a 488-nm laser, and (c) overlay of fluorescence and phase images.