**3.1 Nanoparticle characterization**

DLS, which is also known as Photon Correlation Spectroscopy or Quasi-Electron Light Scattering, is a technique used to study the size and size distribution of particles suspended in a liquid. The technique is based on the scattering of light of particles in diffusive random (brownian) motion. The average displacement for the Brownian motion is defined by the translational diffusion coefficient (D). The particle diffusive motion in liquid is size dependent, and a larger particle has a slower motion as compared to a smaller particle. This brownian motion can be investigated by irradiating the sample with a coherent laser and studying the intensity fluctuations of the scattered light (Finsy, 1994).

Particle sizing can be done in several ways. Typically the information retrieved from different techniques is to some extent diverse, as each technique is sensitive to it´s specific properties of the particles. That means that a technique which is based upon the scattering intensity does not deliver the same size or size distribution as a technique that is based upon the projected area or the density of a nanoparticle. For nano sized particles, transmission electron microscopy (TEM) is frequently used in purpose to study the size and the shape of particles. In TEM, the sample preparation together with the measurement is relatively time consuming and furthermore the measurement is limited only to a very small fraction of the sample. This means that a lot of replicates must be studied in order to achieve good statistics. A dynamic light scattering measurement on the other hand, is fast and convenient as it usually takes only a few minutes to perform. Data recording procedure is thus short but the analysis and interpretation requires knowledge and care. DLS measurements must be performed with highly diluted solutions, to avoid multiple scattering phenomenons and misleading artifacts are frequently present in DLS studies.

Particle sizes obtained when measuring with DLS are by default larger than those obtained when analyzing the material with TEM. The size calculated from the translational diffusion

Two-Dimensional Gel Electrophoresis and Mass

Spectrometry in Studies of Nanoparticle-Protein Interactions 337

composed of at least two populations with hydrodynamic radius of about 100 nm and 200- 300 nm respectively. The smallest radius from this DLS study was achieved for Al2O3 dispersed in Milli-Q water and ultrasonicated with a bar homogenizer (radius approx. 30 nm). Also for samples dispersed in PBS, ultrasonication with a bar homogenizer indicates

decrease of the aggregation and introduction of populations with smaller radius.

Fig. 1. Example of the typical information retrieved in a DLS measurement.

and E) respectively.

distributions.

The measurement is performed on Al doped ZnO nanoparticles dissolved in MilliQ water. A) shows the correlation function which has a high correlation (close to 1) for really short lag times but decays to zero with a rate dependent on the particle size distribution in the sample. B) shows the normalized distribution function of the Decay time. C) shows the normalized distribution of the unweighted radius and the corresponding normalized distributions of the mass weighted radius and the number weighted radius are shown in D)

Figure 3 shows the size distribution of Aluminium doped ZnO nanoparticles dispersed in water and PBS. Bar ultrasonication introduces smaller sized populations, in consistence with the results in Figure 2. The hydrodynamic radii of these particles in MilliQ water are around 30 nm as largest according to Figure 3. This could be compared with the information from the supplier that these particles are < 50 nm as measured by TEM. Again hydrodynamic diameter is always larger that the core size of nanocrystals obtained from TEM. In this case the sample is also aggregated in solution, which produces even larger sizes and size

The SiO2 particles are 7 nm sized as primary particles according to the supplier based on calculations using the surface area as measured by the nitrogen adsorption method of Brunauer (Brunauer et al., 1938). The supplier also states that these particles commonly form

coefficient in DLS generally is referred to as the hydrodynamic diameter, e.g the diameter of a sphere having the same diffusion coefficient as the particle, while the size of the nanoparticles obtained from TEM is the core size of the nanoparticle investigated. It should be noted that in many cases when the sample consists of a mixture of nanoparticles with a range of sizes and/or mixture of particle shapes, results should be taken as an estimation only but clearly trends can be observed.

The autocorrelation function of the scattered intensity results in an average value of the product of the intensity at time t and the intensity at a time delay later t+dt. The value obtained from the correlation function is large for short delays, since the intensities are highly correlated. The value will be low for longer delays; i.e. the autocorrelation can be described as a decaying function of time delay. From the autocorrelation function, the diffusion constant D can be determined. Furthermore, by using Stokes-Einstein equation the corresponding size distribution is calculated according to:

$$\mathbf{d} = \mathbf{k}\_{\mathrm{B}} \mathbf{T} / 3 \mathbf{m} \mathbf{p} \mathbf{D} \tag{1}$$

where kB is the Boltzman constant, T the temperature, η the viscosity of the solvent and d the hydrodynamic diameter of the particles. This implies that the temperature must be constant during the measurement and the viscosity of the sample solvent must be known. It should be noted that the formula shown above (Equation 1) is valid only for non-interacting spherically shaped particles, i.e experimental data are fitted to a model assuming spherical particles (Finsy, 1994).

Figure 1 shows five different functions representing the typical information retrieved in DLS in the actual measurement and by using algorithms. The measured data in DLS is the correlation function (Figure 1A). This function holds information about the diffusion of particles in the sample and can be transformed to a graph showing the decay time of light scattering fluctuations (Figure 1B). From the decay time distribution function, the values of the particle radius (Figure 1C) can be calculated using the Stokes-Einstein equation.

The amount of light that is scattered from a particle is dependent on the particle size. According to the Rayleigh theory (Sorensen, 2008), the scattering factor roughly is proportional to the sixth power of the particle size. This means that a small particle scatter light less than a larger particle and thus different weights have to be applied to transform the intensity weighted data to a useful size distribution. These weights can be mass based (Figure 1D) or number based (Figure 1E). Powder based samples that are dispersed in a liquid often are severely aggregated. Ultrasonic baths can be used to decrease the aggregation in the solution. Choice of solvent is also important and it clearly affects the capability of dispersing nanoparticles.

A number of examples of DLS results for commercial particles are shown in Figure 2, 3 and 4. The size distributions of Al2O3 nanoparticles , based upon number weighted fits of the data are shown in Figure 2. A set of samples were dispersed both in Milli-Q water and PBS. The size and size distribution were measured as a function of concentration. According to the supplier these particles are < 50 nm in size as measured by TEM, which should be taken as the core size of the nanocrystals in the Al2O3 material. A lot larger hydrodynamic diameters are achieved in our measurements, which show that the water based sample is

coefficient in DLS generally is referred to as the hydrodynamic diameter, e.g the diameter of a sphere having the same diffusion coefficient as the particle, while the size of the nanoparticles obtained from TEM is the core size of the nanoparticle investigated. It should be noted that in many cases when the sample consists of a mixture of nanoparticles with a range of sizes and/or mixture of particle shapes, results should be taken as an estimation

The autocorrelation function of the scattered intensity results in an average value of the product of the intensity at time t and the intensity at a time delay later t+dt. The value obtained from the correlation function is large for short delays, since the intensities are highly correlated. The value will be low for longer delays; i.e. the autocorrelation can be described as a decaying function of time delay. From the autocorrelation function, the diffusion constant D can be determined. Furthermore, by using Stokes-Einstein equation the

where kB is the Boltzman constant, T the temperature, η the viscosity of the solvent and d the hydrodynamic diameter of the particles. This implies that the temperature must be constant during the measurement and the viscosity of the sample solvent must be known. It should be noted that the formula shown above (Equation 1) is valid only for non-interacting spherically shaped particles, i.e experimental data are fitted to a model assuming spherical

Figure 1 shows five different functions representing the typical information retrieved in DLS in the actual measurement and by using algorithms. The measured data in DLS is the correlation function (Figure 1A). This function holds information about the diffusion of particles in the sample and can be transformed to a graph showing the decay time of light scattering fluctuations (Figure 1B). From the decay time distribution function, the values of the particle radius (Figure 1C) can be calculated using the Stokes-Einstein

The amount of light that is scattered from a particle is dependent on the particle size. According to the Rayleigh theory (Sorensen, 2008), the scattering factor roughly is proportional to the sixth power of the particle size. This means that a small particle scatter light less than a larger particle and thus different weights have to be applied to transform the intensity weighted data to a useful size distribution. These weights can be mass based (Figure 1D) or number based (Figure 1E). Powder based samples that are dispersed in a liquid often are severely aggregated. Ultrasonic baths can be used to decrease the aggregation in the solution. Choice of solvent is also important and it clearly affects the

A number of examples of DLS results for commercial particles are shown in Figure 2, 3 and 4. The size distributions of Al2O3 nanoparticles , based upon number weighted fits of the data are shown in Figure 2. A set of samples were dispersed both in Milli-Q water and PBS. The size and size distribution were measured as a function of concentration. According to the supplier these particles are < 50 nm in size as measured by TEM, which should be taken as the core size of the nanocrystals in the Al2O3 material. A lot larger hydrodynamic diameters are achieved in our measurements, which show that the water based sample is

d = kBT/3πηD (1)

only but clearly trends can be observed.

particles (Finsy, 1994).

capability of dispersing nanoparticles.

equation.

corresponding size distribution is calculated according to:

composed of at least two populations with hydrodynamic radius of about 100 nm and 200- 300 nm respectively. The smallest radius from this DLS study was achieved for Al2O3 dispersed in Milli-Q water and ultrasonicated with a bar homogenizer (radius approx. 30 nm). Also for samples dispersed in PBS, ultrasonication with a bar homogenizer indicates decrease of the aggregation and introduction of populations with smaller radius.

Figure 3 shows the size distribution of Aluminium doped ZnO nanoparticles dispersed in water and PBS. Bar ultrasonication introduces smaller sized populations, in consistence with the results in Figure 2. The hydrodynamic radii of these particles in MilliQ water are around 30 nm as largest according to Figure 3. This could be compared with the information from the supplier that these particles are < 50 nm as measured by TEM. Again hydrodynamic diameter is always larger that the core size of nanocrystals obtained from TEM. In this case the sample is also aggregated in solution, which produces even larger sizes and size distributions.

The SiO2 particles are 7 nm sized as primary particles according to the supplier based on calculations using the surface area as measured by the nitrogen adsorption method of Brunauer (Brunauer et al., 1938). The supplier also states that these particles commonly form

Two-Dimensional Gel Electrophoresis and Mass

Fig. 4. DLS measurements of SiO2 nanoparticles.

are present.

homogenizer.

Spectrometry in Studies of Nanoparticle-Protein Interactions 339

some hundreds of nanometer long chainlike branched aggregates. The size distributions obtained in our DLS measurements are presented in Figure 4. For all three concentrations in Milli-Q water the particle size is below 100 nm in radius. When SiO2 particles are dispersed in PBS (Figure 4B), large aggregates are definitely present, which are partly removed when ultrasonicated with a bar homogenizer. Multiple scattering, surface charge on the nanoparticles and water solubility should be considered when further evaluating these data. Furthermore, it is known that these specific samples (SiO2) are inhomogenous and the sample is thus far from ideal, i.e does not contain spherical shaped particles. It is shown in previous studies that long chainlike branched aggregates

The normalized distribution of the number weighted radii of particles in SiO2 nanopowder diluted to different concentrations in A) MilliQ and B) PBS as measured by DLS. As marked

In conclusion, it could be said that size and size distribution of nanoparticle samples could be estimated by DLS. Valuable information as the trends in size distribution connected to sample preparation methods and choice of solvent can be obtained. Sample preparation methods are indeed very important as well as choice of solvent. Care should be taken when choosing fitting model and the model-inbuilt parameters. Inhomogeneous samples are less straight forward to analyze. Presence of aggregates is easily detected. In summary information obtained from DLS is important for everybody that is doing research on nanoparticles in liquids. The numbers given as product information i.e. the size and size distribution, are often relevant for the core-size of the nanocrystals within the material. However the nanoparticles are most often not soluble to that extent. Nanoparticles obtained in dry state and then dispersed in liquid usually form aggregates as shown in this study.

in the figure, two different kinds of ultrasonic treatment were used to decrease the aggregation before performing the measurement; either an ultrasonic bath or a bar

Fig. 2. DLS measurements of Al2O3 nanoparticles.

The normalized distribution of the number weighted radii of particles in Al2O3 nanopowder diluted to different concentrations in A) MilliQ and B) PBS as measured by DLS. As marked in the figure, two different kinds of ultrasonic treatment were used to decrease the aggregation before performing the measurement; either an ultrasonic bath or a bar homogenizer.

Fig. 3. DLS measurements of ZnO (Aluminium 6% doped) nanoparticles. The normalized distribution of the number weighted radii of particles in ZnO (Aluminium 6% doped) nanopowder diluted to different concentrations in A) MilliQ and B) PBS as measured by DLS. As marked in the figure, two different kinds of ultrasonic treatment were used to decrease the aggregation before performing the measurement; either an ultrasonic bath or a bar homogenizer.

The normalized distribution of the number weighted radii of particles in Al2O3 nanopowder diluted to different concentrations in A) MilliQ and B) PBS as measured by DLS. As marked in the figure, two different kinds of ultrasonic treatment were used to decrease the aggregation

before performing the measurement; either an ultrasonic bath or a bar homogenizer.

Fig. 3. DLS measurements of ZnO (Aluminium 6% doped) nanoparticles.

bath or a bar homogenizer.

The normalized distribution of the number weighted radii of particles in ZnO (Aluminium 6% doped) nanopowder diluted to different concentrations in A) MilliQ and B) PBS as measured by DLS. As marked in the figure, two different kinds of ultrasonic treatment were used to decrease the aggregation before performing the measurement; either an ultrasonic

Fig. 2. DLS measurements of Al2O3 nanoparticles.

some hundreds of nanometer long chainlike branched aggregates. The size distributions obtained in our DLS measurements are presented in Figure 4. For all three concentrations in Milli-Q water the particle size is below 100 nm in radius. When SiO2 particles are dispersed in PBS (Figure 4B), large aggregates are definitely present, which are partly removed when ultrasonicated with a bar homogenizer. Multiple scattering, surface charge on the nanoparticles and water solubility should be considered when further evaluating these data. Furthermore, it is known that these specific samples (SiO2) are inhomogenous and the sample is thus far from ideal, i.e does not contain spherical shaped particles. It is shown in previous studies that long chainlike branched aggregates are present.

Fig. 4. DLS measurements of SiO2 nanoparticles.

The normalized distribution of the number weighted radii of particles in SiO2 nanopowder diluted to different concentrations in A) MilliQ and B) PBS as measured by DLS. As marked in the figure, two different kinds of ultrasonic treatment were used to decrease the aggregation before performing the measurement; either an ultrasonic bath or a bar homogenizer.

In conclusion, it could be said that size and size distribution of nanoparticle samples could be estimated by DLS. Valuable information as the trends in size distribution connected to sample preparation methods and choice of solvent can be obtained. Sample preparation methods are indeed very important as well as choice of solvent. Care should be taken when choosing fitting model and the model-inbuilt parameters. Inhomogeneous samples are less straight forward to analyze. Presence of aggregates is easily detected. In summary information obtained from DLS is important for everybody that is doing research on nanoparticles in liquids. The numbers given as product information i.e. the size and size distribution, are often relevant for the core-size of the nanocrystals within the material. However the nanoparticles are most often not soluble to that extent. Nanoparticles obtained in dry state and then dispersed in liquid usually form aggregates as shown in this study.

Two-Dimensional Gel Electrophoresis and Mass

Fig. 5. Plasma protein binding profiles of different nanoparticles.

amyloid fibrils, a process which plays a major role in pathology.

Four different nanoparticles; Al2O3, CNO, ZnO (\* Aluminium-doped 6%) and SiO2, were mixed with human plasma and isolated through ultracentrifugation. The protein contents were then separated by 2-DE and silver stained. Bound proteins were identified by MS as

The inflammatory marker Fibrinogen was found to interact with SiO2, Al2O3 and CNO but interestingly not with ZnO (Fig 5). Fibrinogen binds foreign substances in the circulation and promotes macrophage activation - a mechanism that may result in retention of particle/protein complexes in the intima with accompanying cardiovascular complications (Shulz et al., 2005). On the other hand, Lysozyme C, a well known anti-bacterial agent, was only found on Al2O3 and ZnO. Under the present condition, it was unfortunately not determined if the binding to ZnO occurred due to the 6% Al doping of the ZnO. Lysozyme has previously been found to interact with nano-TiO2 particles (Xu et al., 2010). They reported that the coexistence of nano-TiO2 particles and Lysozyme resulted in the transition of Lysozyme conformation from α-helix into β-sheet secondary structure and a substantial inactivation of Lysozyme. Moreover the β-sheets are able to induce the formation of

investigated.

shown in Table 1.

Spectrometry in Studies of Nanoparticle-Protein Interactions 341

free drug (Ferris et al. 2011). What demand further studies though, is the fate of the silica particles in a longer perspective, taken up by tumor cells as well as other cells. A third transport protein, the thyroxin transporting protein Transthyretin, also known to bind toxic components in the blood stream (Hamers et al., 2011), was found to interact with CNO and SiO2. This finding confirms a previous study that pointed out that silica interaction (inhaled) with plasma Transthyretin is contributing to the stabilization of fibroids in rat lungs (Kim et al., 2005). The possible effects of CNO interaction with Transthyretin remains to be

#### **3.2 Nanoparticle-plasmaprotein interactions**

In a previous study performed by us, the inflammatory response in human monocyte derived macrophages after exposure to wear particles generated from the interface of studded tires and granite containing pavement (Karlsson et al. 2011) was investigated. Particle characterization showed that dominating peaks in the EDX spectra were Silica and Aluminium. Particles of nanosize were also present (SMPS), but it was not possible to characterize them due to low abundance. As a result of their very small diameter (< 0.1 µm), inhaled nanoparticles are believed to be predominantly agglomerated and deposited in the periphery of the lungs, where they interact with cells such as macrophages and epithelial cells (Beck-Speier et al. 2005) but they may also translocate into the circulation, which is a critical step, since their fate *in vivo* in not known. Investigating plasma protein-nanoparticle interactions with a toxico-proteomic approach is a useful tool to improve our knowledge about the effects of nanoparticles of different origin, size and surface properties in biological systems.

In purpose to mimic a potential exposure to airborn nanoparticles translocated into the circulation, commercial SiO2 and Al2O3 were mixed with plasma proteins. As comparison, commercial ZnO (Al-doped 6%) and a non-metal oxide; single walled Carbon Nanotubes was used. All preparations were performed in triplicates with three different subjects exposed to each type of particles. The protein patterns resulting from the three different exposures of commercial SiO2, Al2O3, ZnO and CNO respectively were identical.

Particle characterization and estimation of particle agglomeration prior to exposure is crucial. In a recent study of nanoparticle-plasma protein interactions (Deng et al. 2009), the DLS spectra indicated large agglomerates prior to plasma protein exposure. Most likely, and in line with the authors suggestions (Deng et al. 2009), complexes with hydrodynamic size of 10000-100 000 nm do not result in the same protein patterns as the interactions of smaller particles/agglomerates and plasma proteins. In our optimized protocol, with different particle origin, less gentle bar sonication instead of in water bath and thereby reduced hydro dynamic sizes of the agglomerates (Fig 2-4) - an altered pattern of interacting proteins was indeed found (Figure 5, Table 1).

Interestingly, the interaction of SiO2 and CNO with plasma proteins resulted in very similar protein patterns despite their different properties (Fig 5). It has to be stated though, that the fate of CNO in the lung may not be translocation into the circulation due to the tube like structure, but CNO is also of interest for medical applications (Wu et al. 2011).

The transport proteins Albumin and Alpha-2-HS-glycoprotein interacted with SiO2**,** CNO and ZnO but not Al2O3 while Transferrin interacted with SiO2, Al2O3 and notably also CNO but not ZnO. Supporting our results, the binding of albumin to single walled CNO has previously been described to promote uptake by the scavenger receptor in RAW cells (Dutta et al., 2007). In line, intravenous administration of CNO has in a different study resulted in high localization in the liver (Cherukuri et al., 2006). Another possible way for nanoparticles into the cells, are as Transferrin/particle complexes that are able to enter the cells via the Transferrin receptor. The Transferrin receptor is an interesting and relevant target in cancer research since its expression is increased in tumor cells. In a recent study, Transferrin covalently attached to silica nanoparticles carrying a hydrophobic drug caused an increase in mortality of the targeted cancer cells compared to cells exposed to nontargeted particles and

In a previous study performed by us, the inflammatory response in human monocyte derived macrophages after exposure to wear particles generated from the interface of studded tires and granite containing pavement (Karlsson et al. 2011) was investigated. Particle characterization showed that dominating peaks in the EDX spectra were Silica and Aluminium. Particles of nanosize were also present (SMPS), but it was not possible to characterize them due to low abundance. As a result of their very small diameter (< 0.1 µm), inhaled nanoparticles are believed to be predominantly agglomerated and deposited in the periphery of the lungs, where they interact with cells such as macrophages and epithelial cells (Beck-Speier et al. 2005) but they may also translocate into the circulation, which is a critical step, since their fate *in vivo* in not known. Investigating plasma protein-nanoparticle interactions with a toxico-proteomic approach is a useful tool to improve our knowledge about the effects of nanoparticles of different origin, size and surface properties in biological systems. In purpose to mimic a potential exposure to airborn nanoparticles translocated into the circulation, commercial SiO2 and Al2O3 were mixed with plasma proteins. As comparison, commercial ZnO (Al-doped 6%) and a non-metal oxide; single walled Carbon Nanotubes was used. All preparations were performed in triplicates with three different subjects exposed to each type of particles. The protein patterns resulting from the three different

exposures of commercial SiO2, Al2O3, ZnO and CNO respectively were identical.

Particle characterization and estimation of particle agglomeration prior to exposure is crucial. In a recent study of nanoparticle-plasma protein interactions (Deng et al. 2009), the DLS spectra indicated large agglomerates prior to plasma protein exposure. Most likely, and in line with the authors suggestions (Deng et al. 2009), complexes with hydrodynamic size of 10000-100 000 nm do not result in the same protein patterns as the interactions of smaller particles/agglomerates and plasma proteins. In our optimized protocol, with different particle origin, less gentle bar sonication instead of in water bath and thereby reduced hydro dynamic sizes of the agglomerates (Fig 2-4) - an altered pattern of interacting proteins was

Interestingly, the interaction of SiO2 and CNO with plasma proteins resulted in very similar protein patterns despite their different properties (Fig 5). It has to be stated though, that the fate of CNO in the lung may not be translocation into the circulation due to the tube like

The transport proteins Albumin and Alpha-2-HS-glycoprotein interacted with SiO2**,** CNO and ZnO but not Al2O3 while Transferrin interacted with SiO2, Al2O3 and notably also CNO but not ZnO. Supporting our results, the binding of albumin to single walled CNO has previously been described to promote uptake by the scavenger receptor in RAW cells (Dutta et al., 2007). In line, intravenous administration of CNO has in a different study resulted in high localization in the liver (Cherukuri et al., 2006). Another possible way for nanoparticles into the cells, are as Transferrin/particle complexes that are able to enter the cells via the Transferrin receptor. The Transferrin receptor is an interesting and relevant target in cancer research since its expression is increased in tumor cells. In a recent study, Transferrin covalently attached to silica nanoparticles carrying a hydrophobic drug caused an increase in mortality of the targeted cancer cells compared to cells exposed to nontargeted particles and

structure, but CNO is also of interest for medical applications (Wu et al. 2011).

**3.2 Nanoparticle-plasmaprotein interactions** 

indeed found (Figure 5, Table 1).

free drug (Ferris et al. 2011). What demand further studies though, is the fate of the silica particles in a longer perspective, taken up by tumor cells as well as other cells. A third transport protein, the thyroxin transporting protein Transthyretin, also known to bind toxic components in the blood stream (Hamers et al., 2011), was found to interact with CNO and SiO2. This finding confirms a previous study that pointed out that silica interaction (inhaled) with plasma Transthyretin is contributing to the stabilization of fibroids in rat lungs (Kim et al., 2005). The possible effects of CNO interaction with Transthyretin remains to be investigated.

Fig. 5. Plasma protein binding profiles of different nanoparticles. Four different nanoparticles; Al2O3, CNO, ZnO (\* Aluminium-doped 6%) and SiO2, were mixed with human plasma and isolated through ultracentrifugation. The protein contents were then separated by 2-DE and silver stained. Bound proteins were identified by MS as shown in Table 1.

The inflammatory marker Fibrinogen was found to interact with SiO2, Al2O3 and CNO but interestingly not with ZnO (Fig 5). Fibrinogen binds foreign substances in the circulation and promotes macrophage activation - a mechanism that may result in retention of particle/protein complexes in the intima with accompanying cardiovascular complications (Shulz et al., 2005). On the other hand, Lysozyme C, a well known anti-bacterial agent, was only found on Al2O3 and ZnO. Under the present condition, it was unfortunately not determined if the binding to ZnO occurred due to the 6% Al doping of the ZnO. Lysozyme has previously been found to interact with nano-TiO2 particles (Xu et al., 2010). They reported that the coexistence of nano-TiO2 particles and Lysozyme resulted in the transition of Lysozyme conformation from α-helix into β-sheet secondary structure and a substantial inactivation of Lysozyme. Moreover the β-sheets are able to induce the formation of amyloid fibrils, a process which plays a major role in pathology.

Two-Dimensional Gel Electrophoresis and Mass

MS/MS (Figure 6B) to further confirm the identity.

2-DE.

uptake.

upper right corner.

exposure and toxicity of nanomaterials (Higashisaka et al., 2011).

Spectrometry in Studies of Nanoparticle-Protein Interactions 343

this study has not been described previously but an increase of Haptoglobin has been reported in a study investigating acute phase proteins as biomarkers for predicting the

At last, the HDL associated Apo A-I, with well known anti-endotoxin activity (Henning et al., 2006) and receptor interaction properties was found in all preparations but was most abundant after SiO2 exposure. Apo A-I may be acting as a scavenger clearing the particles from the blood stream via the scavenger class B-I receptor (SRBI). The SRBI receptor is mainly located on the liver and plays an important role in cholesterol efflux (Verger et al., 2011) but is also present on other cells (Mooberry et al. 2010). Apolipoproteins in general are of interest for the pharmaceutical industry as carriers of nanoparticle bound drugs for brain uptake (Kreuter et al., 2005) since apo E and apo B-100 are taken up by the cells via receptor mediated endocytosis. The hypothesis is that the nano-particle/apolipoprotein complex mimics the natural lipoprotein particle. The identity of Apo A-I was, as the other proteins, confirmed by peptide mass fingerprinting using MALDI TOF MS (Table 1) and one dominating Apo A-I peptide in the MS spectra (Figure 6A) was in addition sequenced by

Fig. 6. Identification of nanoparticle bound apoliprotein A-I with mass spectrometry after

A: Peptide mass fingerprint spectrum obtained by MALDI-TOF mass spectrometry after endoproteinase GluC digestion. Asterisks represent peaks corresponding to peptide masses of apo A-I. B: Sequencing of a triply charged peptide (m/z 409.9) corresponding to position 194-203 of apo A-I by collision induced disassociation (CID) in a tandem mass spectrometer. The amino acid sequence with ions corresponding to the different fragments is shown in the

Overall the binding of plasma proteins to nanoparticles, based on previous and our findings, seems to vary with origin, surface properties, size and thereby also diameter of agglomerates. Particle characterization prior to exposure for plasma proteins or cells is therefore extremely important to receive reliable results that are possible to interpret. Interacting proteins under the present conditions are dominated by proteins involved in the immune defense and reverse transport to the liver but notably also proteins mediating brain


Table 1. Identification of nanoparticle bound plasma proteins by peptide mass fingerprinting after 2-DE.

The table shows identified proteins with Uniprot accession number, isoelectric point (pI), molecule weight (Mw), number of peptide masses matched, sequence coverage and MOWSE score. a Isoelectric point (pI) and molecular weight (Mw) as estimated on gels. b Matched peak masses with a mass error tolerance of 75 ppm.

Furthermore, some antigen binding proteins; IgKC, DKFZ and IgLC, were also found to interact with Al2O3, CNO and SiO2 but were not detectable on ZnO (Fig 5). Immunoglobulins are able to activate the complement system but they also often represent unspecific binding during protein purification caused by insufficient washing. Our results compared to previous findings indicates that increased hydrodynamic size of nanoparticle agglomerates seems to correlate to increased amounts of immunoglobulins. Overall ZnO was not binding as many proteins as the other particles and may even bind less without being Al-doped but an interesting finding in the ZnO preparation was a protein only described on transcript level and highly similar to the protein Amyloid β A4 precursor. This family of proteins acts as chelators of metal ions such as iron and zinc. They are also able to induce histidinebridging between beta-amyloid molecules resulting in beta-amyloid-metal aggregates and it has been reported that extracellular zinc-binding increases binding of heparin to Amyloid β A4 (Uniprot 2011).

The protease inhibitor Alpha-1-antitrypsin, was found in the CNO, SiO2 and ZnO preparations but not in Al2O3 while another antioxidant, Haptoglobin was found only on CNO and SiO2. To our knowledge, the binding of Alpha-1-antitrypsin to the nanoparticles in

**4** Fibrinogen γ Al2O3 P02679 5.4 50000 15 32 1.39e+6 **5** Fibrinogen β Al2O3, CNO P02675 7.0 55000 17 37 5.83e+8

**7** Lysozyme C Al2O3, ZNO P61626 9.4 15000 7 33 12888

**9** Haptoglobin CNO, SiO2 P00738 5.1 46000 8 20 2115 **10** DKFZ CNO Q6N096 8.3 55000 14 34 1.10e+7

Transthyretin CNO, SiO2 P02766 5.4 16000 7 51 69267 Ig KC CNO Q6PJF2 8.0 30000 10 55 409936 ApoA-IV SiO2, ZNO P06727 5.1 45000 10 23 72729 Igγ SiO2 P01859 7.7 55000 7 23 79647 Igγ-1 chain SiO2 P01857 8.5 35000 12 40 3.74e+7 Amyloid βA4 ZNO B4DJT9 5.2 60000 20 15 53.1

Table 1. Identification of nanoparticle bound plasma proteins by peptide mass

Matched peak masses with a mass error tolerance of 75 ppm.

The table shows identified proteins with Uniprot accession number, isoelectric point (pI), molecule weight (Mw), number of peptide masses matched, sequence coverage and MOWSE score. a Isoelectric point (pI) and molecular weight (Mw) as estimated on gels. b

Furthermore, some antigen binding proteins; IgKC, DKFZ and IgLC, were also found to interact with Al2O3, CNO and SiO2 but were not detectable on ZnO (Fig 5). Immunoglobulins are able to activate the complement system but they also often represent unspecific binding during protein purification caused by insufficient washing. Our results compared to previous findings indicates that increased hydrodynamic size of nanoparticle agglomerates seems to correlate to increased amounts of immunoglobulins. Overall ZnO was not binding as many proteins as the other particles and may even bind less without being Al-doped but an interesting finding in the ZnO preparation was a protein only described on transcript level and highly similar to the protein Amyloid β A4 precursor. This family of proteins acts as chelators of metal ions such as iron and zinc. They are also able to induce histidinebridging between beta-amyloid molecules resulting in beta-amyloid-metal aggregates and it has been reported that extracellular zinc-binding increases binding of heparin to Amyloid β

The protease inhibitor Alpha-1-antitrypsin, was found in the CNO, SiO2 and ZnO preparations but not in Al2O3 while another antioxidant, Haptoglobin was found only on CNO and SiO2. To our knowledge, the binding of Alpha-1-antitrypsin to the nanoparticles in

chain Al2O3 Q0KKI6 7.4 <sup>30000</sup> 5 32 2779

**(Da)**

SiO2, ZNO P02647 5.2 <sup>24500</sup> 28 74 9.65e+10

ZNO P01009 5.0 <sup>55000</sup> 31 63 5.36e+18

ZNO P02765 4.8 <sup>55000</sup> 6 14 681

**Matched Peaks***<sup>b</sup>*

P02768 6.0 68000 30 47 4.20e+14

Q53H26 6.7 77000 43 65 1.23e+22

**Sequence Coverage (%)**  **MOWSE Score** 

**Uniprot AccessionNumber <sup>p</sup>***Ia* **Mw***<sup>a</sup>*

**Found in Nano particle** 

SiO2

SiO2

**Number in** 

**Figure 2 Protein** 

**<sup>6</sup>**Ig Light

fingerprinting after 2-DE.

A4 (Uniprot 2011).

**<sup>1</sup>**ApoA-I Al2O3, CNO,

**<sup>2</sup>**Albumin Al2O3, CNO,

**<sup>3</sup>**Transferrin Al2O3, CNO,

**<sup>8</sup>**α1-AT CNO, SiO2,

**<sup>11</sup>**Alpha-2-HS CNO, SiO2,

this study has not been described previously but an increase of Haptoglobin has been reported in a study investigating acute phase proteins as biomarkers for predicting the exposure and toxicity of nanomaterials (Higashisaka et al., 2011).

At last, the HDL associated Apo A-I, with well known anti-endotoxin activity (Henning et al., 2006) and receptor interaction properties was found in all preparations but was most abundant after SiO2 exposure. Apo A-I may be acting as a scavenger clearing the particles from the blood stream via the scavenger class B-I receptor (SRBI). The SRBI receptor is mainly located on the liver and plays an important role in cholesterol efflux (Verger et al., 2011) but is also present on other cells (Mooberry et al. 2010). Apolipoproteins in general are of interest for the pharmaceutical industry as carriers of nanoparticle bound drugs for brain uptake (Kreuter et al., 2005) since apo E and apo B-100 are taken up by the cells via receptor mediated endocytosis. The hypothesis is that the nano-particle/apolipoprotein complex mimics the natural lipoprotein particle. The identity of Apo A-I was, as the other proteins, confirmed by peptide mass fingerprinting using MALDI TOF MS (Table 1) and one dominating Apo A-I peptide in the MS spectra (Figure 6A) was in addition sequenced by MS/MS (Figure 6B) to further confirm the identity.

Fig. 6. Identification of nanoparticle bound apoliprotein A-I with mass spectrometry after 2-DE.

A: Peptide mass fingerprint spectrum obtained by MALDI-TOF mass spectrometry after endoproteinase GluC digestion. Asterisks represent peaks corresponding to peptide masses of apo A-I. B: Sequencing of a triply charged peptide (m/z 409.9) corresponding to position 194-203 of apo A-I by collision induced disassociation (CID) in a tandem mass spectrometer. The amino acid sequence with ions corresponding to the different fragments is shown in the upper right corner.

Overall the binding of plasma proteins to nanoparticles, based on previous and our findings, seems to vary with origin, surface properties, size and thereby also diameter of agglomerates. Particle characterization prior to exposure for plasma proteins or cells is therefore extremely important to receive reliable results that are possible to interpret. Interacting proteins under the present conditions are dominated by proteins involved in the immune defense and reverse transport to the liver but notably also proteins mediating brain uptake.

Two-Dimensional Gel Electrophoresis and Mass

Fig. 8. Subfractionation of plasma with regard to HDL

Plasma (left), HDL purified according to apo A-I with immunoaffinity chromatography (middle) and HDL isolated according to density with ultracentrifugation (right) were separated with 2-DE and silver stained. Proteins were identified with mass spectrometry.

Gel-separated proteins can be visualized by several commonly used staining methods, including dyes (e.g. Coomassie Brilliant Blue and colloidal Coomassie), metals (e.g. silver staining) and fluorescent probes (e.g. Sypro staining and Cy-dyes) (Rabilloud, 2000). The stains interact differently with the proteins and have different limitations with regard to sensitivity, linear range, compatibility with mass spectrometry and type of proteins that stain best. In general, for staining of complex protein samples, silver staining can be considered the most sensitive technique (1-5 ng protein) and Coomassie Brilliant blue the least (50-100 ng) while the sensitivity of colloidal coomassie and the fluorescent dyes are in between. However, it is important to bear in mind that the different stains interact differently with the proteins and therefore one protein may stain very well with one staining method but not with the other (Fig. 9). For example, silver ions react with negatively charged groups and therefore stain glycoproteins containing negative sialic acid very well.

HDL fraction.

**4.2 Protein detection** 

Spectrometry in Studies of Nanoparticle-Protein Interactions 345

nanoparticles is apo A-I, the major constituent of HDL. This implicates the need of more investigations of HDL as a possible target of nanoparticles that may influence the cholesterol metabolism and increase the risk of cardiovascular disease. HDL can be isolated based on density, size or protein content (e.g. apo A-I) using ultracentrifugation, size-exclusion chromatography or immune-affinity chromatography, respectively, each technique with its own merits and drawbacks. Thus, the rather harsh conditions during ultracentrifugation in high salt gradients may remove weakly associated proteins while the rather mild conditions during chromatography may favor unspecific co-purification of proteins with the cholesterol particle. We have previously mapped the protein content of HDL isolated by two-step density gradient ultracentrifugation (Karlsson et al., 2005). In this study we have compared ultracentrifugation and anti-apo A-I affinity chromatography to isolate HDL from the same plasma sample. As shown in figure 8, more proteins were obviously identified in immune-affinity purified HDL. However, some of the proteins must be considered as possible plasma contaminants as they were not, as e.g. the apolipoproteins, enriched in the
