• Open Access

Multifunctional Magnetic Optical Sensor Particles with Tunable Sizes for Monitoring Metabolic Parameters and as a Basis for Nanotherapeutics


  • Günter Mistlberger,

    Corresponding author
    1. Institute of Analytical Chemistry and Food Chemistry Graz University of Technology Stremayrgasse 16, 8010 Graz (Austria)
    • Institute of Analytical Chemistry and Food Chemistry Graz University of Technology Stremayrgasse 16, 8010 Graz (Austria).
    Search for more papers by this author
  • Klaus Koren,

    1. Institute of Analytical Chemistry and Food Chemistry Graz University of Technology Stremayrgasse 16, 8010 Graz (Austria)
    Search for more papers by this author
  • Elisabeth Scheucher,

    1. Institute of Analytical Chemistry and Food Chemistry Graz University of Technology Stremayrgasse 16, 8010 Graz (Austria)
    Search for more papers by this author
  • Daniel Aigner,

    1. Institute of Analytical Chemistry and Food Chemistry Graz University of Technology Stremayrgasse 16, 8010 Graz (Austria)
    Search for more papers by this author
  • Sergey M. Borisov,

    1. Institute of Analytical Chemistry and Food Chemistry Graz University of Technology Stremayrgasse 16, 8010 Graz (Austria)
    Search for more papers by this author
  • Armin Zankel,

    1. Institute for Electron Microscopy Graz University of Technology Steyrergasse 17, 8010 Graz (Austria)
    Search for more papers by this author
  • Peter Pölt,

    1. Institute for Electron Microscopy Graz University of Technology Steyrergasse 17, 8010 Graz (Austria)
    Search for more papers by this author
  • Ingo Klimant

    Corresponding author
    1. Institute of Analytical Chemistry and Food Chemistry Graz University of Technology Stremayrgasse 16, 8010 Graz (Austria)
    • Institute of Analytical Chemistry and Food Chemistry Graz University of Technology Stremayrgasse 16, 8010 Graz (Austria).
    Search for more papers by this author


Magnetic optical sensor particles with multifunctional cores and shells are synthesized via a facile nanoprecipitation method and the subsequent modification of the particle shell. The hydrophobic particle core includes optical oxygen indicators, a light harvesting system, photosensitizers, and magnetic nanoparticles. Further functionalities are introduced by modifying the shell with enzymes, antibodies, multiple layers of polyelectrolytes, stimuli-responsive polymers, and luminescent indicator dyes. The hydrodynamic diameter is tunable by varying different precipitation parameters.

1. Introduction

Real-time monitoring and imaging of physiologically important parameters in biological samples is of high interest in both bioengineering and life science.1 Optical sensor particles represent a convenient tool for real-time monitoring of these parameters. Recently, the concept of optical sensor particles was enhanced by the incorporation of magnetic nanoparticles. This enables the operator to trap the sensors at a distinct spot and to guide them to a desired position within the measurement setup.2 Thereby, the signal intensity is increased and optical interferences with the medium are reduced. Furthermore, it was shown that a particle “swarm” follows a moving magnetic separator faster than single particles and that fewer sensor particles were required to achieve the signal intensity of dispersed sensor particles.2a, ,3, 3, 4

A large number of laboratories is nowadays working on the development of multifunctional particles as nanotherapeutics and imaging components. Most of these particles respond to stimuli such as changes in pH, temperature, light, and magnetic field.5 They might also be loaded with optical labeling components, such as quantum dots for imaging purposes.6 Another common modification is the introduction of magnetic nanoparticles for purposes such as magnetic guiding, enrichment, and thermotherapy by AC magnetic fields and the particles' use as a magnetic resonance imaging (MRI) contrast agent.7, 8 While optical sensor particles can be utilized for monitoring changes in oxygenation, pH, temperature, ion concentrations, and ammonia concentrations,9 multifunctional nanoparticles were reported as tools for drug delivery, biosensors, cancer thermotherapy, magnetic drug targeting, and magnetically induced thermal drug release.7, 10 Obviously, a smart combination of both principles (optical sensing and the use of multifunctional polymeric nanoparticles) would lead to novel tools for research and life science.

Here, we present a facile route to multifunctional magnetic optical sensor particles (MF-MOSePs) with tunable hydrodynamic diameters between 50 and 180 nm. The particle cores were equipped with oxygen indicators (in visible and near IR (NIR) range), magnetic nanoparticles, and a photosensitizer. Furthermore, we modified the surface of the MOSePs with an enzyme, polyelectrolytes, a pH-indicator, and a stimuli-responsive polymer. Such a nanodevice can, for example, carry an enzyme and monitor the enzymatic reaction with the help of an incorporated optical sensor.

Different applications require different particle sizes. Whenever the position of particles has to be magnetically controllable in a reasonable time, particles with sizes above ∼100 nm are preferable. On the contrary, smaller magnetic nanoparticles (<60 nm) might enable diagnostic and therapeutic applications (MRI imaging, thermotherapy, temperature-induced drug release, etc.).

2. Results and Discussion

2.1. Synthesis and Particle Structure

The synthetic route to the MF-MOSePs is outlined in Scheme 1 and relies on an emulsifier-free nanoprecipitation method.9a, ,11 After dissolving the polymer (poly(styrene-co-maleic anhydride, PSMA93 with 7% maleic anhydride (MA) and a MW of 224 000 g mol−1 or EF-80 with 11% MA and a MW of 14 400 g mol−1) and an indicator dye in tetrahydrofuran (THF), lipophilic magnetic nanoparticles (L-MNPs) were dispersed in the “cocktail.” Upon the addition of water, which is miscible with THF but is not a solvent for the polymer nor a dispersant for the magnetic nanoparticles, spherical particles formed spontaneously. The unique properties of THF (its solvent and dispersant capabilities for the “cocktail” components and its miscibility with the precipitant) were crucial for the success of this synthetic route. The polarity of other tested solvents was either too high to obtain a homogeneous dispersion of the L-MNP (acetone, dimethylformamide) or too low to allow water miscibility (chloroform, toluene). Notably, no surfactants are used in the process, and consequently, no additional cleaning steps are required to remove emulsifiers that would influence biological systems. After the precipitation, the solvent was evaporated, and the particles shrank to their final size. In Figure 1a the spherical shape of MOSePs is shown. During the precipitation process, the L-MNPs were irreversibly trapped inside the polymeric matrix (Fig. 1b).

Scheme 1.

Synthetic route to multifunctional magnetic optical sensor particles (MF-MOSePs) via a nanoprecipitation technique. After particle, formation the surface was modified with fluorophores, enzymes, layer-by-layer agents, or stimuli-responsive polymers.

Figure 1.

a) Scanning electron microscopy (SEM) image of MOSePs (2% PSMA93, 20% magnetite, z-average = 185 nm). b) Transmission electron microscopy (TEM) image of the core MOSePs used for the layer-by-layer (LbL) experiment (0.5% EF-80, 20% magnetite).

During precipitation, anhydride groups on the surface react with water and form carboxyl groups. This results in a negative zeta potential of –35 mV at a pH > 5 and ensures highly stable dispersions of MOSePs in aqueous media. In contrast, acidifying a particle dispersion resulted in spontaneous aggregation and sedimentation of the particles. Binding 5-aminofluorescein to the surface via a zero-length crosslinking method12 with 1-ethyl-3-[3-dimethylaminopropyl]carbodiimide hydrochloride (EDC) elucidated a ratio of 2.0 ± 0.5 bound fluorescein molecules per 1000 maleic anhydride (MA) molecules or 2.2 ± 0.5 µmol per gram of particles. This means that an average of ∼10 000 fluorescein molecules is bound to a 250 nm particle (the z-average of the modified batch). The applied assay returns the number of accessible carboxyl groups, which is more relevant for covalent surface modifications than the total number of hydrolyzed anhydrides. Fourier transform (FT) IR spectra confirmed the low total carboxy concentration in the particles (Fig. 2). The unmodified matrix polymer EF-80 showed characteristic signals at 2925 cm−1 and a double peak at 1453 and 1493 cm−1 for the aromatic vibrations of the styrene. At 1780 cm−1 the typical signal for the carbonyl bond of the anhydride group can be found.13 The broad [BOND]OH peak at 3000 cm−1 for a hydrolyzed anhydride group is missing in the spectra of both MOSeP samples, with and without L-MNP included. Furthermore, a shift of the carbonyl peak would be expected during hydrolysis.

Figure 2.

IR spectra of the magnetic nanoparticles (“EMG1300”), MOSePs with magnetite, MOSePs without magnetite, and the matrix polymer EF-80 (chemical structure in bottom left corner). No hydrolyzed anhydride groups can be seen in the MOSePs' spectra. This confirms the low 5-aminofluorescein binding rate (2.2 µmol (g particles)−1).

While a hydrophilic surface ensures highly stable aqueous dispersions, the particle core needs to be hydrophobic in order to retain the apolar components inside the matrix. Both the dye and the L-MNPs are trapped via hydrophobic interactions in the polymer. A polar matrix would result in rapid leaching of the dye and magnetic particles out of the core into the medium. Because of the amphiphilic character of the matrix polymer in MOSePs, no leaching of the dye or magnetic particles was observed over a period of several months.

2.2. Precipitation Parameters Influencing Particle Sizes

Fine-tuning of the particle sizes by varying different precipitation parameters enables the control of certain size-dependent properties, such as separation speed, sensor response, and dispersion stability.

Particularly, the polymer concentration was found to be a critical parameter for adjusting the particle size as can be seen in Figure 3a. The values of the z-average (z-av) measured using the dynamic light scattering of four consecutive precipitations under the same conditions resulted in a relative standard deviation below 2% and an average polydispersity index (PDI) of 0.04. The highest concentration of PSMA93 was above the limit for efficient particle production with this method.

Figure 3.

Influences of the polymer concentration along with the a) polymer type, b) drop speed, c) magnetite content, and d) precipitant on the resulting hydrodynamic diameters. All PDIs, except for 1.2% PSMA93, are below 0.1.

The employed polymer type had an effect on both the size of the particles and the upper limit of polymer concentration. The two polymers, EF-80 and PSMA93, differ in the average molecular weight and their MA content, which influences the polarity of the matrix. EF-80, the polymer with shorter chains (weight-average molecular weight, Mw = 14 400 g mol−1 versus 224 000 g mol−1 for PSMA93) and higher MA content (11% versus 7% w/w for PSMA93), resulted in particles which were approximately 30 nm smaller over the whole range of tested concentrations (Fig. 3a, Table 1). The lower the molecular weight the lower the viscosity of the cocktail was, and this favored the formation of smaller particles. Also the higher polymer polarity might have contributed to the formation of smaller particles due to increased precipitant–polymer interactions. Size determination from electron microscopic images resulted in smaller values because the z-average resembles the hydrodynamic diameter of particles in an aqueous dispersion while the particles in the electron microscope were completely dry (Fig. 4; Fig. S1 in the Supporting Information (SI)). Moreover, the z-average represents a statistical value that depends on both the particle volume and number. While high polymer concentrations increased the particle size and distribution width, a narrow size distribution was achieved at low polymer concentrations (Fig. 4).

Table 1. Overview of various precipitation parameters and resulting values for z-average and PDI. The polymer concentration (w(Poly)) is given in w/w solvent. Concentrations of additives such as magnetite and dyes are presented in w/w polymer.
Samplew (Poly)w (Mag) [%]PolymerDirectionVortex [min−1]Flowrate [mL s−1]Precipitantz-av (±s) [nm]PDI (±s)
  • [a]

    Includes 1% (w/w polymer) PdTPTBP dye. Particles were filtered through a 0.8 µm syringe-filter after precipitation.

  • [b]

    In this batch a total of 100 mL of cocktail containing 1% polymer was precipitated with 200 mL of water without additional mixing. These undefined conditions resulted in large particles with a higher PDI. Considering the large amount of particles required for the binding experiments, these properties were acceptable.

CE10.220EF-80CiP12000.50H2O71.5 (±1.2)0.045 (±0.019)
CE20.420EF-80CiP12000.50H2O90.1 (±1.6)0.027 (±0.007)
CE30.620EF-80CiP12000.50H2O108.5 (±1.6)0.029 (±0.009)
CE40.820EF-80CiP12000.50H2O123.0 (±0.8)0.040 (±0.013)
CE50.020EF-80CiP12000.50H2O134.0 (±0.5)0.039 (±0.005)
CE61.220EF-80CiP12000.50H2O147.2 (±0.5)0.076 (±0.005)
CP10.220PSMA93CiP12000.50H2O105.4 (±1.0)0.022 (±0.006)
CP20.420PSMA93CiP12000.50H2O121.5 (±1.2)0.035 (±0.007)
CP30.620PSMA93CiP12000.50H2O134.8 (±1.2)0.035 (±0.011)
CP40.820PSMA93CiP12000.50H2O146.2 (±0.9)0.059 (±0.012)
CP51.020PSMA93CiP12000.50H2O162.9 (±1.9)0.079 (±0.002)
CP61.220PSMA93CiP12000.50H2O180.0 (±4.2)0.198 (±0.058)
PP0.620PSMA93CiP12000.50H2O132.2 (±1.7)0.072 (±0.012)
PE0.620EF-80CiP12000.50H2O115.6 (±2.3)0.068 (±0.007)
M10.60EF-80CiP12000.50H2O135.6 (±1.9)0.049 (±0.006)
M20.65EF-80CiP12000.50H2O132.9 (±3.7)0.037 (±0.006)
M30.610EF-80CiP12000.50H2O123.7 (±1.9)0.051 (±0.014)
M40.620EF-80CiP12000.50H2O115.4 (±0.8)0.044 (±0.011)
M50.630EF-80CiP12000.50H2O110.1 (±1.4)0.042 (±0.020)
D10.620EF-80CiP12000.04H2O132.9 (±1.0)0.050 (±0.025)
D20.620EF-80CiP12000.50H2O126.6 (±1.9)0.046 (±0.024)
D30.620EF-80CiP12002.00H2O110.3 (±2.1)0.076 (±0.007)
V10.620EF-80CiP6000.50H2O120.2 (±2.5)0.034 (±0.024)
V20.620EF-80CiP12000.50H2O116.8 (±1.5)0.040 (±0.007)
V30.620EF-80CiP18000.50H2O118.3 (±2.5)0.063 (±0.025)
PD10.620EF-80CiP12000.50H2O113.9 (±1.6)0.052 (±0.015)
PD20.620EF-80PiC12000.50H2O135.0 (±2.5)0.029 (±0.023)
PR10.620EF-80CiP12000.50H2O115.4 (±0.8)0.044 (±0.011)
PR20.620EF-80CiP12000.50MeOH99.6 (±5.8)0.027 (±0.009)
PR30.620EF-80CiP12000.50EtOH74.9 (±3.9)0.028 (±0.015)
PDT[a]2.020PSMA93PiC12000.50H2O171.1 (±1.2)0.089 (±0.012)
GOX1.020PSMA93CiP12000.50H2O186.0 (±1.7)0.029 (±0.024)
FLU[b]0.60EF-80CiP0.50H2O256.8 (±4.8)0.274 (±0.063)
SR0.520EF-80CiP12000.50H2O100.5 (±0.5)0.021 (±0.004)
LBL0.520EF-80CiP12000.50H2O105.9 (±0.6)0.077 (±0.012)
SEM2.020PSMA93PiC6000.50H2O185.4 (±3.6)0.093 (±0.012)
IrC1.020PSMA93CiP12000.50H2O186.0 (±1.7)0.029 (±0.024)
PdBP0.620PSMA93CiP12000.50H2O124.9 (±3.2)0.100 (±0.024)
PtBP0.620PSMA93CiP12000.50H2O110.3 (±2.1)0.180 (±0.044)
Figure 4.

Size histograms from SEM images of EF-80 particles with varying polymer concentration in the cocktail (see Table 1). The scale bar is 200 nm in all images.

The effect of the precipitation direction, i.e., “cocktail into precipitant” (CiP) or “precipitant into cocktail” (PiC), can be seen in Figure S2 in the SI. Adding the cocktail to the precipitant (CiP) ensures a fast solvent displacement, and consequently, smaller particles are formed. In the opposite direction (PiC), the solvent concentration in the cocktail is decreasing slowly, which gives the polymer cocktail more time to form particles with a lower surface:volume ratio. This effect can be explained by the decreased contact area between polymer and aqueous phase.

Another parameter influencing the particle size was the flow rate at which the cocktail was injected into the precipitant (Fig. 3b; Fig. S3 (SI)). Here, a higher flow rate (2 mL s−1) decreased the particle sizes by 25 nm compared to the dropwise addition of the cocktail (0.04 mL s−1).

The effect of a varying magnetite concentration indicated the role of the L-MNPs in the particle formation process. Increasing the magnetite concentration from 0% to 30% (w/w polymer) resulted in a stepwise reduction of the particles' z-average from 136 to 110 nm (Fig. 3c, Table 1; Fig. S4 (SI)). The TEM image in Figure 1b also shows the inorganic particles close to the surface. The L-MNPs might act as seeds for the polymeric particles. The higher the magnetite concentration is, the more seeds are available and therefore the particle size decreases.

Finally, the particle size can be influenced by the polarity of the precipitant. We investigated the z-average of particles resulting from precipitations where the cocktail was injected into water, methanol, and ethanol. Figure 3d and S5 (SI) show a significant drop in particle size with decreasing precipitant polarity. However, precipitation with n-propanol resulted in separated precipitation of the magnetic nanoparticles and polymeric nanoparticles. Acetone did not act as a precipitant and resulted in a clear, nonscattering dispersion of the L-MNPs in the polymer solution.

Changing the vortex speed from 600 to 1800 min−1 did not change the particle sizes significantly. Apparently, the mixing due to cocktail injection was sufficient for fast solvent displacement.

Swelling of 110 nm EF-80 MOSePs in mixtures of THF and water resulted in a 7%, 15%, 20% and 26% increase of particle diameter for 20%, 30%, 40% and 50% (v/v) THF in water. The swelling was reversible, and the particles returned to their original size after THF evaporation.

In conclusion, the particle sizes can be adjusted to fit the required properties within a certain range. This is necessary as different particle properties are required for different applications. Larger particles (>100 nm) usually separate faster in a magnetic field, but have higher PDIs. Very small particles (<60 nm) are difficult to separate from dispersion, but have higher surface-to-volume ratios, ensure higher dispersion stability, and might be suitable for in vivo applications such as thermotherapy, thermal drug release, magnetic particle tracking and MRI imaging.

2.3. Functionalities Included in the Particle Core

2.3.1. Magnetism

Magnetic nanoparticles included in the core make MOSePs magnetically controllable. The in situ production of a sensor spot, for instance, increases the signal intensity of optical sensor particles and the response speed to a changing magnetic field.2a We used this technique for sensor characterization. MOSePs dispersed in a buffer solution were collected at the side-wall of the vessel by a magnetic separator with an optical window. An optical fiber pointed through the center of the separator directly onto the particle spot (Fig. 5a). By this, the required amount of sensor particles is significantly reduced compared to a plain sensor dispersion.

Figure 5.

a) Fluorescence image of MOSePs (0.6% EF-80 in the cocktail; 20% magnetite, 1% Ir(CS)2(acac), both w/w polymer; z-average = 105 nm) dispersed in water before and after being captured in front of an optical fiber with a specially designed magnetic separator. b) Sedimentation and magnetic separation speed of MOSePs (2% PSMA93 in the cocktail, 20% magnetite w/w polymer, z-average = 192 nm) in an aqueous dispersion followed by measuring the optical density at 600 nm (OD600).

The separation speed is a function of various factors, including particle size, magnetite content of the particles, and volume of the dispersion from where the particles are separated. Fine-tuning these parameters allows the priority to be set to separation speed or dispersion stability. On the one hand, the hydrophilic surface ensures highly stable aqueous dispersions. On the other hand, if a fast separation is required, MOSePs (z-av = 192 nm, 20% magnetite) are quantitatively separated from a 3 mL aqueous dispersion (5 mg mL−1) with the aid of a strong magnet within 30 min (Fig. 5b). The separation inside microfluidic devices is certainly faster, because of the reduced volume and distances. The time required for separating MOSePs is significantly higher than for particles in the micrometer range,3 but acceptable for biological applications such as the investigation of the surface oxygenation of biofilms.

Besides magnetic guiding, the magnetic properties of MOSePs might enhance the contrast of MRI images and allow the thermal therapy of cancer via AC magnetic fields.10a, 8a, ,14 The location of magnetic particles is also detectable inside biological tissues using sensitive superconducting quantum interference devices (SQUIDs).15 This multifunctionality of the magnetic nanoparticles enables the application of MOSePs with sizes below 60 nm, which would prohibit a separation in a reasonable time frame for conventional sensor applications.

2.3.2. Optical Sensing

To demonstrate the flexibility of this particular method for incorporating lipophilic indicator dyes into the core, we tested a range of different oxygen-sensitive indicator dyes. For the production of magnetically controllable optical sensor particles, we incorporated oxygen indicators such as iridium(III) acetylacetonato-bis(3-(benzothiazol-2-yl)-7-(diethylamino)-coumarin) (Ir(CS)2(acac)), palladium(II) and platinum(II) meso-tetra(4-fluorophenyl) tetrabenzoporphyrin (PdTPTBPF, PtTPTBPF) and palladium(II) meso- tetraphenyltetrabenzoporphyrin (PdTPTBP). While Ir(CS)2(acac) proved to be an ultrabright oxygen optode for the production of sensors with a high dynamic range (pO2 = 0–1000 hPa),3, 16 the benzoporphyrin dyes PtTPTBPF, PdTPTBP, and PdTPTPBF efficiently absorb red light and emit in the NIR range. This is especially useful for biological applications because of the increased penetration depth and reduced background due to scattering and autofluorescence.

Fitting the Stern–Volmer plot of Ir(CS)2(acac)-stained MOSePs (“IrC” in Table 1) with the simplified two-site model (Eq. 2),17 where one part of the dye is assumed to be unquenchable, i.e., its KSV2 = 0, resulted in a KSV1 = 0.0038 hPa−1 with a quenchable fraction of P = 0.94 (Fig. 6a).

equation image((1))
equation image((2))

where τ is the luminescence decay time, and I the luminescence intensity. The index zero indicates that the concentration of the quencher (e.g., oxygen) is zero. The almost linear correlation between τ0/τ versus pO2 is useful for practical applications (calibration, dynamic range). High signal intensities are important for oxygen imaging with thin particle layers or small sensor spots. Moreover, the high brightness of the incorporated dye overcomes the problem of the highly light-absorbing L-MNPs. In the case of weak fluorescent dyes and excitation in the UV, the dark color of the magnetite can cause problems by absorbing both excitation and emission light.18

Figure 6.

Stern–Volmer plots and absorption and emission spectra of oxygen indicators with different dynamic ranges and spectral properties. Chemical structures of the dyes can be seen in Figure S7 in the SI.

Finally, due to the small particle sizes, the response to a changing oxygen concentration is fast which enables real-time monitoring of dissolved oxygen. The response time of a particle spot to a rapid change in oxygen concentration was measured with a t90 of 1.4 s (Fig. 7; t90 represents the time in which the signal reaches 90% of its equilibrium value). This time is, however, limited by the diffusion inside the dense particle layer, whereas the response of single particles is most probably much faster. Nevertheless, a t90 of 1.4 s enables real-time monitoring of most biological processes.

Figure 7.

Response time of Ir(CS)2(acac)-stained MOSePs (1% w/w PSMA93; 20% magnetite, 1% dye, both w/w polymer; z-average = 186 nm) in a solution of 100 mM glucose upon addition of 400 U mL−1 glucose oxidase (GOX).

For long-term oxygen monitoring in biological samples, benzoporphyrin dyes such as PtTPTBPF and PdTPTBPF can outperform the iridium coumarin dyes due to their higher photostability and phosphorescence emission in the NIR range of the spectrum.19 Moreover, the absorption of the incorporated magnetite nanoparticles is significantly lower in the red part of the spectrum, and therefore, the emission intensity is further increased.

We produced MOSePs with PtTPTBPF for oxygen monitoring from 0–100% air saturation (KSV1 = 0.016 hPa−1, P = 0.92, Fig. 6b, “PdBP” in Table 1) and MOSePs with PdTPTBPF for trace oxygen monitoring (KSV1 = 0.067 hPa−1, P = 0.89, Fig. 6c, “PtBP” in Table 1).

Recently, the suitability of light harvesting systems to enhance the brightness of optical sensors was reported.20 Due to the efficient absorption of excitation light by the harvesting dyes, the fraction of light lost due to magnetite absorption is further reduced. The incorporation of the oxygen indicator platinum(II) meso(2,3,4,5,6-pentafluoro)phenyl porphyrin (PtTFPP) together with an antenna dye (macrolex yellow) in the MOSeP core resulted in oxygen sensor particles with an improved excitation spectrum (Fig. 8). It was possible to increase the emission intensity at 650 nm significantly by changing the excitation wavelength from 505 to 465 nm. Besides the signal enhancement due to increased absorption by the acceptor dye compared to the magnetite, this light harvesting system is suitable for the excitation with extremely bright blue light-emitting diodes (LEDs).

Figure 8.

Excitation spectra of PtTFPP (1% w/w polymer) and a light harvesting system consisting of macrolex yellow (2%) and PtTFPP (1% w/w polymer) incorporated in MOSePs (emission wavelength, λem = 650 nm).

2.3.3. Singlet Oxygen Production

Besides the oxygen sensing function of palladium and platinum porphyrins, such dyes are known as efficient singlet oxygen producers. PdTPTBP for instance has a high molar absorption coefficient (λmax (ε) = 629 nm (173 000 L mol−1 cm−1 and moderate quantum yield (0.21).19 Its phosphorescence is almost completely quenched at air saturation as shown in Figure 9a, and the quantum yield for singlet oxygen production of platinum- and palladium porphyrins is usually close to unity.21 An efficient singlet oxygen production, however, also depends on efficient contact of the photosensitizer in its excited state with oxygen. Moreover, the produced singlet oxygen requires rapid transport to the surface in order to avoid deactivation inside the polymer matrix. Small particle diameters result in a high specific surface area and short diffusion distances of both triplet and singlet oxygen.

Figure 9.

a) Luminescence spectra of PdTPTBP-stained MOSePs (“PDT” in Table 1) in oxygen-free and air-saturated medium. b) Singlet oxygen production of the same MOSePs compared to a 1 µm foil containing the same amount of dye.

Lai et al. recently reported on the successful utilization of iridium complexes for the generation of singlet oxygen while simultaneously imaging the localization of the particles and utilizing the magnetic properties of the composite particles for MRI imaging.6b We investigated the singlet oxygen production efficiency of PdTPTBP-doped MOSePs (“PDT” in Table 1) upon illumination with a xenon lamp filtered through a 590 nm long-pass filter. In comparison to a 1 µm thick foil with the same amount of dye, matrix material, and light intensity, the PdTPTBP-doped MOSePs resulted in a 60 times higher production rate (Fig. 9b). Red light excitation is beneficial for a potential application in biological samples. The incorporation of PdTPTBP as trace oxygen sensor and singlet oxygen producer in the MOSeP cores results in particles capable of simultaneous photodynamic therapy (PDT) and oxygen monitoring. Oxygenation is a prerequisite for an efficient photodynamic therapy, but at the same time the oxygenation is also a parameter characterizing the metabolic state of a tissue.

2.4. Modifications of the Shell

As mentioned above, MOSePs possess a number of carboxyl groups at the surface. Besides the stabilization of aqueous particle dispersions, carboxyl groups offer the possibility of surface modifications.

2.4.1. Covalently Bound Indicator Dyes or Fluorescent Labels

Polar indicator dyes, such as pH-indicators require a proton-permeable matrix. In addition, if leaching of the dye occurs, a covalent attachment might be favored over the simple enclosure in the polymer network. As an example, we bound aminofluorescein to surface carboxyl groups after activation by EDC. The spectra of the modified MOSePs at varying pH can be seen in Figure 10a. Plotting the emission at 519 nm in correlation with the pH value resulted in an apparent pKa of 7.3 (Fig. 10b), a value for sensors commonly used in physiological applications. An inclusion of lipophilic fluorescein derivatives in the lipophilic core—a common technique for the production of pH sensors in hydrogel particles—would prohibit the interaction of protons with the indicator.

Figure 10.

MOSePs coated with 5-aminofluorescein resulted in particles with a changing fluorescence emission at 519 nm (excitation wavelength, λex = 480 nm). The observed pKa of this system is 7.3.

2.4.2. Binding of Enzymes and Other Proteins

Through the same zero-length crosslinking method used for the fluorescein binding, it is possible to link an enzyme or protein with an accessible amino group to the MOSePs' surface. Here, we used glucose oxidase (GOX) as a model enzyme to demonstrate the linking of an enzyme in its active form to oxygen-sensitive MOSePs. To prove that the GOX retained its activity after the immobilization, we monitored the oxygen consumption at different glucose concentrations (Fig. 11a). The modified MOSePs were capable of consuming glucose and oxygen while simultaneously monitoring the change in oxygenation with the incorporated indicator (Ir(CS)2(acac)). For this experiment, the particles were collected in front of an optical fiber in a glass tube with the help of a magnetic separator with an optical window,2a and they were flushed with different glucose solutions. Binding of other proteins for increased biocompatibility or the introduction of recognition patterns can be carried out in a similar manner.

Figure 11.

Surface modifications of MOSePs. a) Coupling GOX to oxygen-sensitive MOSePs basically resulted in magnetic glucose sensors. b) The zeta potentials during the coating of four positive and four negative polyelectrolyte layers are plotted. c) The modification of MOSePs with a pNIPAM shell resulted in temperature-sensitive core–shell particles.

2.4.3. Surface Modification by the Layer-by-Layer (LbL) Technique

The initially negatively charged surface of the MOSeP at neutral pH allows for surface modifications with charged species (LbL technique). For a proof of concept, we coated the surface alternately with four layers of poly(diallyldimethylammonium chloride) (+) and four layers of polystyrene sulfonate (−). The zeta potentials and sizes were measured after every step and elucidated the alternating charges from −32 to +37 mV (Fig. 11b). Throughout the coating procedure, the hydrodynamic diameters increased slightly from 105 to 130 nm, compared to blank diameters which increased from 105 to 115 nm. The increasing blank diameter is due to a slight aggregation occurring during the repeated separation and washing steps. However, the SEM image of coated MOSePs shows spherical particles without significant aggregation or change of particle shape (Fig. S6 (SI)).

2.4.4. Particle Coating with Stimuli-Responsive Polymers

These polymers generally respond to changing physical parameters such as pH and temperature,5a, ,22 but also to light, radiation, or chemical stimulators. We chose the hydrogel poly-N-isopropylacrylamide (pNIPAM) as one of the most studied representatives of stimuli-responsive polymers to prove the possibility of producing stimuli-responsive MOSePs (SR-MOSePs). The polymerization was accomplished by polymerizing N-isopropylacrylamide and N,N'-methylenebisacrylamide in the presence of MOSePs as seeds. The resulting particles were magnetic, oxygen-sensitive (Ir(CS)2(acac), τ0/τair = 1.5), and they reversibly changed their size with the temperature (Fig. 11c). Such stimuli-responsive MOSePs would provide an optimal basis for magnetic drug carriers23 releasing a drug at a distinct position and simultaneously measuring the effect of this event in real-time at the very place of the release.

2.4.5. Potential Other Modifications

In analogy to the modifications mentioned above (Sections 2.4.1–4), MOSePs represent a basis for the modification with biodegradable polymers or affinity ligands. Biodegradable polymers in combination with magnetic particles are used for targeted drug delivery.10f,24 Tumor-specific antibodies bound to the surface might reduce the damaging of healthy cells due to drug delivery, PDT, or hyperthermia inducible by MOSePs.

3. Conclusions

A polymeric multifunctional optical nanosensor platform was developed. We successfully demonstrated the application of MF-MOSePs as real-time oxygen sensors, in situ optical biosensors, magnetic PDT agents, magnetic pH sensors, and stimuli-responsive magnetic optical sensors, and we proposed other potential applications. Due to the versatility of the platform, the presented particles can be easily modified to match the requirements of a wide range of scientific and clinical applications. For in vivo applications, a further size reduction might be necessary to allow renal excretion. For in vitro applications, however, the presented MOSePs represent a ready-to-use, multipurpose platform.

4. Experimental


PSMA93 (7% maleic anhydride; Mw = 224 000 g mol−1), GOX (from Aspergillus niger, lyophilized powder), poly(diallyldimethylammonium chloride) (20% in water), poly(sodium 4-styrenesulfonate) (20% in water), N-isopropylacrylamide, N,N'-methylenebisacrylamide, potassium peroxodisulfate, 5-aminofluorescein, N,N-dimethyl-4-nitrosoanilin, imidazol, and sodium dodecylsulfate were purchased from Sigma. EDC was purchased from TCI Europe. Poly(styrene maleic anhydride) with 11% MA and a molecular weight of 14 400 g mol−1 (EF-80) was generously provided by Sartomer Europe. L-MNP (polymer-coated magnetite nanoparticles “EMG1300” from MNP-kit) were purchased from FerroTec GmbH. THF, sodium chloride, phosphate buffer solutions, and glucose were obtained from Carl Roth GmbH. PtTFPP was bought from Frontier Scientific and Macrolex Yellow from Simon & Werner GmbH (Flörsheim, Germany). Nitrogen, oxygen, synthetic air, and test gas with 1% oxygen (all of 99.999% purity) were purchased from Air Liquide. Ir(CS)2(acac), PdTPTBPF, and PtTPTBPF were prepared in our laboratory as reported elsewhere 16, 19. PdTPTBP was prepared according to Finikova et al. 25.

Absorption spectra were measured at a Cary 50 UV–vis spectrophotometer (Varian Inc., USA). Emission spectra were acquired on a Hitachi F-7000 fluorescence spectrometer (Hitachi Inc.) equipped with a red-sensitive photomultiplier R 928 from Hamamatsu. The phase shifts for the oxygen measurements were recorded with a 2 mm optical fiber and a fiber optic phase fluorimeter equipped with a blue LED (470 nm) for excitation and a 550 nm long-pass filter for the emission. The modulation frequency was adjusted to 20 kHz. Alternatively, for PdTPTBPF and PtTPTBPF, a 630 nm LED (Roithner Laser Technik) was modulated with a two-phase lock-in amplifier (SR830, Stanford Research Inc.). A bifurcated fiber bundle was used to guide the excitation light (filtered through a Calflex filter, Linos) to the sample and the luminescence back to the detector after being filtered through an RG9 (Schott) glass filter. Luminescence was detected with a photomultiplier tube (PMT; H5701-02, Hamamatsu). The modulation frequencies were adjusted to 5 kHz for PtTPTBPF and 0.7 kHz for PdTPTBPF. For the calibrations the suspensions were purged with different ratios of nitrogen and oxygen, synthetic air, or test gas (1% O2) adjusted by a gas mixing device at a flow rate of 200 mL min−1. Particle sizes and zeta potentials were measured with a particle size analyzer Zetasizer Nano ZS. FT-IR-spectra were recorded with a Perkin-Elmer Spectrum One instrument equipped with an attenuated total reflectance (ATR) unit. SEM images were recorded on a Zeiss Ultra 55 equipped with a field-emission gun (FEG). A drop of MOSeP suspension was placed on a silicon wafer. After evaporation of the dispersant, the samples were coated with a thin chromium layer to avoid specimen charging. TEM images were recorded on a Philips CM 20 microscope equipped with a LaB6 filament. A drop of MOSeP suspension was placed on a standard copper grid, and the dispersant was evaporated under a slight stream of air.

Synthesis of MOSePs (“CE3”)

In a typical synthesis EF-80 (3 mg), L-MNP (0.6 mg) and dye (0.03 mg) were dissolved/dispersed in dry THF (0.5 g) under ultrasonication. This “cocktail” was then added to deionized water (4.5 mL) under vortexing (1200 min−1). Particle precipitation occurred immediately. Under a stream of air, THF was evaporated from the mixture over a period of 25 min, and the resulting particles were washed twice with deionized water by magnetic separation. Occasionally occurring aggregates were removed by filtration through a syringe filter (Rotilabo, 0.8 µm). The particles obtained under these conditions had a hydrodynamic diameter of 109 nm (Table 1).

For particles with other diameters, their syntheses were adjusted as stated in Table 1.

Singlet Oxygen Assay

The assay was based on the procedure published by Kraljic and Mohsni 26. Briefly, a phosphate buffer solution (2.5 mL, pH 7.4, ionic strength = 0.05 M) containing N,N-dimethyl-4-nitrosoanilin (5 × 10−5M), and imidazol (8 × 10−3M) with a 10 µm foil (2.5 cm × 1.4 cm) containing 55 µg PdTPTBP (“foil”) or a 5.5 mg PdTPTBP (1%, w/w)-stained MOSePs (“PDT-MOSePs,” “PDT” in Table 1) were illuminated with a xenon lamp filtered through a 590 nm long-pass filter (excitation via only the Q-band) for up to 300 min. As a blank, only the assay components were illuminated. The production of singlet oxygen was controlled by measuring the absorption at 440 nm of the supernatant every 5 min after separating the MOSePs and sensor foil.

Surface Modification of MOSePs with GOX

MOSePs (5 mg) were dispersed in phosphate buffer (2 mL, 0.05 M, pH 7.0). The dispersion was then incubated with EDC (2 mg) for 10 min to activate the carboxyl groups. For binding, GOX (5 mg) was added and the binding reaction was carried out for 5 h on a rotation mixer at 50 min−1. The resulting particles were magnetically separated and washed 4 times with phosphate buffer for further investigation.

Surface Modification of MOSePs with 5-Aminofluorescein

MOSePs (200 mg, “FLU” in Table 1) were dispersed in 2-(N-morpholino)ethanesulfonic acid (MES) buffer (10 mL, 0.1 M, pH 4.5). The dispersion was then incubated with EDC (30 mg) and 5-aminofluorescein (10 mg) in a rotation mixer for 2 h. The resulting particles were magnetically separated and washed four times with ethanol, three times with water, and redispersed in water for further investigation.

LbL Coating of MOSePs

In a glas vial, MOSePs (5 mg, “LBL,” Table 1, zeta potential = −32.4 ± 0.6 mV) were incubated with a 0.8% (w/w) solution of the two different polyelectrolytes poly(diallyldimethylammonium chloride) and poly(sodium 4-styrenesulfonate) (2 mL) in alternating manner. After each incubation step, the particles were magnetically separated and washed three times with deionized water in the ultrasonic bath.

Coating of MOSePs with a pNIPAM Shell (SR-MOSePs)

In a typical procedure, MOSePs (15 mg, “SR” in Table 1), N-isopropylacrylamide (50 mg), N,N′-methylenebisacrylamide (1.7 mg), and sodium-dodecylsulfate (3.6 mg) were dissolved in deionized (9 mL) water and heated to 65 °C under nitrogen. After the addition of potassium peroxodisulfate (9.0 mg dissolved in 1 mL of water), the NIPAM was allowed to polymerize under continued stirring for 3 h. Afterwards, the particles were cleaned repeatedly with deionized water and dispersed in water for further investigations.

Measurement of the 5-Aminofluorescein Binding Capacity

5-Aminofluorescein-modified MOSePs (100 mg, “FLU,” see Table 1) were dissolved in THF (3 mL), and the concentration was calculated using the absorption and extinction coefficient at 483 nm.


We gratefully thank DI S. Fladischer and DI C. Gspan from the FELMI-ZFE Graz for their help to record the TEM images. We also acknowledge F. Adanitsch and J. Flock for their excellent technical support and Dr. Ute Daschiel for her help performing FT-IR measurements. This work was supported by the Austrian Science Fund FWF (project P 21192-N17). Supporting Information is available online from Wiley InterScience or from the author.