Investigation of the Associations between a Nanomaterial's Microrheology and Toxicology.

With the advent of Nanotechnology, the use of nanomaterials in consumer products is increasing on a daily basis, due to which a deep understanding and proper investigation regarding their safety and risk assessment should be a major priority. To date, there is no investigation regarding the microrheological properties of nanomaterials (NMs) in biological media. In our study, we utilized in silico models to select the suitable NMs based on their physicochemical properties such as solubility and lipophilicity. Then, we established a new method based on dynamic light scattering (DLS) microrheology to get the mean square displacement (MSD) and viscoelastic property of two model NMs that are dendrimers and cerium dioxide nanoparticles in Dulbecco's Modified Eagle Medium (DMEM) complete media at three different concentrations for both NMs. Subsequently, we established the cytotoxicological profiling using water-soluble tetrazolium salt-1 (WST-1) and a reactive oxygen species (ROS) assay. To take one step forward, we further looked into the tight junction properties of the cells using immunostaining with Zonula occluden-1 (ZO-1) antibodies and found that the tight junction function or transepithelial resistance (TEER) was affected in response to the microrheology and cytotoxicity. The quantitative polymerase chain reaction (q-PCR) results in the gene expression of ZO-1 after the 24 h treatment with NPs further validates the findings of immunostaining results. This new method that we established will be a reference point for other NM studies which are used in our day-to-day consumer products.

Introduction

The wide applicability of nanoparticles in consumer products necessitates their extensive study related to their safety and potential risk.[1] At the nanoscale, matter has fundamentally different properties from bulk materials.[2] As material is scaled down to nanoparticles, chemical properties, biological properties, optical properties, and electrical properties are different from their bulk counterpart.[3] Such unique properties could vastly alter the toxicity profile, requiring additional safety assessment considerations as compared to larger or bulk counterparts of the same materials. Although animal models are still used for risk assessment experiments, based on the 3R (refine, reduce, and replace) principle, the development of alternative testing methods is emphasizing in vitro model experiments.[4,5] Apart from acting as an alternative evaluation tool, in vitro assays also play a key role in understanding the mechanism of the biological activities of nanoparticles. There are many challenges associated with in vitro methodologies to ensure they are as robust and reliable as traditional in vivo approaches.[6] However, by overcoming such issues and adopting new testing strategies, we are able to improve safety assessments and reduce in vivo experiments. One of the challenges is the influence of physiochemical properties[7−9] as well as physical properties of nanosuspension[10] in the in vitro experiments. Poor characterization of nanoparticle suspension may lead to misinterpretation of nanotoxicity results. Therefore, proper characterization of the nanoparticle suspension is of the utmost importance. Studies have already been carried out regarding the major properties of nanoparticle suspension, such as absorptive properties,[11,12] release of metal ions,[13] and stability[14−16] of suspension, but the microrheological property of the nanoparticle suspension in the cell culture media has never been explored.

Rheological properties such as viscoelasticity have been found to be an essential characteristic of living tissues, structural proteins, and the extracellular matrix (ECM).[17,18] Viscoelastic materials exhibit a response between the two extremes of purely elastic where all of the input deformation energy can be “stored” and “recovered” during each cycle without any loss and purely viscous where all of the input deformation energy is dissipated or “lost” by internal friction in the system as it flows. In response to the application of an external stress, viscoelastic materials undergo deformation and stress relaxation in a time-dependent manner in response to a step deformation.[19] While traditional rheological studies are more based on concentrated proteins,[20,21] only little research has been carried out on dilute protein samples, due to the lack of adequate experimental techniques. Over the past decade, optical microrheological techniques like DLS microrheology have gained huge popularity for rheological characterization of various complex fluids[22,23] as well as dilute proteins.[24] In comparison to conventional mechanical rheometry, DLS microrheology can measure much less sample volume and can measure over a much wider frequency range.[25] Since DLS microrheology uses Brownian motion of tracer particles, less stress is applied on fluid, and this is of critical importance for many biological samples that can exhibit significant strain sensitivity.[26] DLS microrheology provides mean square displacement (MSD), which is the distance covered by nanoparticles with respect to movement of the tracer particle with time. By using MSD, linear viscoelastic parameters for the complex fluid matrix are extracted through a generalized Stokes–Einstein relationship.[27] Although different studies have been conducted on the microrheological properties of various proteins,[26] complex fluids,[22] and biological systems,[28] microrheological studies explicitly on the nanoparticle suspensions in the cell culture medium have never been addressed. The stability of any suspension under study greatly depends on its rheological property. Toxicological investigations on nanoparticles require a comprehensive physiochemical characterization including the rheological properties of the nanoparticle suspension. Hence, the main aim of this study is to investigate the microrheological properties of two model nanoparticles of cerium dioxide (CNP) and dendrimers in cell culture media and to know their subsequent toxicological profiles.

The reasons for including cerium dioxide nanoparticles and dendrimers as our model nanoparticles is due to the fact that these represent two physiochemically different NPs. CNPs are metallic oxide nanoparticles, whereas dendrimers are organic nanoparticles. Apart from that, their vast applicability in the consumer products makes them nanoparticles of interest. CNPs are used commercially in numerous industries, such as petroleum refining (as a catalyst for cracking),[29] coatings,[30] polishing agents (for glass mirrors, plate glass, television tubes, ophthalmic lenses, precision optics, and electronic wafers),[31] and fuel cells.[32] They are extensively used in consumer products such as semiconductors,[33] as diesel fuel additives,[34] and as additives in cigarettes.[35] It has been demonstrated that they can protect cell membranes from sources of oxidative stress (hydrogen peroxide, ultraviolet rays, and ionizing radiation), buffer reactive oxygen species, and thus decrease damage to cellular biomolecular structures, resulting from oxidative stress,[36] making them promising antioxidant agents for treating oxidative stress-related diseases.[37] Combined with their catalytic activity and electrochemical characteristics, their properties can be used to create highly sensitive, third-generation biosensors[38] as well as quenchers in fluorescent biosensors.[39] Dendrimers on the other hand show a wide range of medicinal and practical applications such as in photodynamic therapy,[40] as MRI contrast,[41] in tissue engineering,[42] in gene transfection,[43] and in drug delivery.[44]

Experimental Section

Cell Culture

Madin-Darby Canine Kidney (MDCK) cells (ATCC cat. no.: CCL-34) were cultured in Dulbecco’s Modified Eagle Medium (DMEM) supplemented with 10% fetal calf serum (FCS) (PAN-Biotech GmbH, Germany), 1% penicillin/streptomycin (PAN-Biotech GmbH, Germany), and 1% l-glutamine (PAN-Biotech GmbH, Germany). Cells were passaged two times per week.

Dynamic Light Scattering (DLS) and Zeta Potential

The hydrodynamic diameter and the zeta potential were determined using a Zetasizer Nano ZS from Malvern (Malvern Inc., UK) in DMEM cell culture media. The final concentrations, 10 μg/mL, 100 μg/mL, and 1000 μg/mL (10×, 100×, and 1000×), of CNP (European Commission, Luxembourg, Belgium) and 1%, 5%, and 10% (1×, 5×, and 10×) of the of PAMAM dendrimers 3.5 generation (10% weight in methanol) (Sigma-Aldrich, Munich, Germany) were prepared in DMEM media. Amounts of 1 mL of the above-mentioned concentrations of NPs with DMEM media were added to disposable cuvettes (Ratiolab GmbH, Hungary, Germany) to measure the hydrodynamic size of NPs. To calculate the zeta potential, a dip cell kit (Malvern Panalytical, Worcestershire, England) was used.

Data Collections, QSAR Model, and Statistical Analysis

There are many versatile in silico methods, and quantitative structure–activity relationship (QSAR) models have been developed to generate consensus predictions for the various physicochemical properties including water solubility (log S) and lipophilicity (log Po/w).[45,46] For data collection for the nanomaterials used in this study, we selected the third generation PAMAM dendrimer from the freely accessible web resource end point[47,48] and from the PubChem database[49] (accessed 10th October 2021). The cerium oxides were selected from the PubChem database and JRC Nanomaterial database[48] (accessed 10th October 2021). The quasi-SMILES for this study was derived from the PubChem database, and their symbols are analogues of the simplified molecular input line entry system (SMILES). The quasi-SMILES required for QSAR modeling of NMs represents the available eclectic information with similar lines of symbols. We calculated the mean of log S, log Po/w, and other physicochemical properties such as distribution coefficient (log D) and topological polar surface area (TPSA), using the SILICOS-IT program based on topological and molecular descriptors, as explained in previous studies.[50−53] We further performed statistical analysis to determine if a difference between the different groups for each category of NMs (dendrimer and CNP) existed.

DLS microrheology is a passive technique in which the dispersed probe or tracer particles are tracked in a complex fluid in order to determine its local and bulk rheological properties. Similar to mechanical rheometry, a strain is applied to the system through Brownian motion of the probe particle, and the change in position of the probe particle is used to measure deformation. Such a microrheological technique permits access to very high-frequency or short-time dynamics of even very dilute samples with less sample volume.[24]

Microrheology is a new measurement type available to users of the Zetasizer Nano ZS and ZSP. It allows the measurement of the viscoelastic modulus of samples within the linear viscoelastic region. Microrheology measurements require a software key to access the software features and functionality.

The protocol for DLS optical microrheology was followed as per the instructions of the manufacturer, Malvern Panalytical (Worcestershire, England).[54] Temperature was set to 25 °C and pH to 7.5 for the whole experiment. Polystyrene beads of 1.5 μm diameter (Polysciences, Inc., Warrington, United States) were used as tracer particles. According to the standard operating procedure of Malvern Panalytical for microrheology measurement, tracer compatibility with the sample and tracer concentration required were subsequently checked before measuring the microrheology. To check the tracer compatibility with NPs, the zeta potential of polystyrene beads was measured in 1 mL of continuous phase solution (here DMEM medium with 10% FBS, 1% penicillin/streptomycin, and 1% l-glutamine). Then, a small volume of NPs was added, and zeta potential was remeasured. If the zeta potential values differ significantly, it implies the interaction between tracer particles and NPs, resulting in adsorption and aggregation between them. The zeta potential in the presence of both tracer and sample particles should be within the set limit of 5 mV. After that, the tracer concentration was determined by following the instructions of the software. Initially 5 μL of tracer particles (polystyrene beads) is added to 1 mL of the complete DMEM media in a disposable cuvette, and then the scattering intensity is measured. After each reading, the software recommends how much of the tracer particle should be further added to get a relative scattering intensity of approximately 95% for tracer particles as compared to the NPs in the sample.

Transepithelial Electrical Resistance (TEER)

The Millicell ERS-2 (electrical resistance system, STX2 electrode) (Merk, Darmstadt, Germany) was used to measure the transepithelial resistance (TEER) or barrier function of the MDCK cells. The cells were seeded at the seeding density of 240 000 cells onto transwell membranes of 12-well hanging cell culture inserts (Merck, Darmstadt, Germany) under submersed condition for 24 h. Twelve well plates were used to place the insets. Each transwell of the inset was filled with 500 μL of cells in DMEM media, and 12 wells were filled with 1 mL of DMEM media to provide a submerged condition for the cells. The next day, all three concentrations of both CNP and dendrimers as mentioned above were prepared in DMEM media, and the cells in the transwell were exposed to these concentrations for 24 h. After 24 h of exposure, the TEER measurements were performed. For background control, one insert without cells was used, which was subtracted from each measured data. Before measuring, the functionality checks of Millicell ERS-2m were done by connecting the STX04 test electrode to it through an inserting plug. The “R Adj” screw on the meter was adjusted with a flat head screw until the meter displayed 1000Ω. During the TEER measurement, the electrode was immersed in such a way that the shorter tip is dipped in the Millicell culture plate insert, and the longer tip is immersed in the outer well. It was made sure that the shorter tip did not contact cells growing on the membrane, and the longer tip only slightly touched the bottom of the outer well. Care was taken that the electrode was held steady and at a 90° angle to the plate insert. After the TEER measurement of 24 h exposure, the cells were incubated further in the incubator, and subsequent TEER measurements were performed on the third and seventh day of the NP exposure. To ensure the cell survival, the cells were supplemented with 200 μL of fresh DMEM media on both the transwell as well as the basolateral well every day until the seventh day of TEER measurement.

ZO-1 Staining

The cells were grown with the seeding density of 50 000 cells on the coverslips (Carl Roth, Karlsruhe, Germany). After reaching confluency, they were exposed to the NPs with the different concentrations as mentioned above for 24 h in the incubator (37 °C, 5% CO2). The cells were then fixed using 4% formaldehyde (Carl Roth GmbH, Karlsruhe, Germany) for 15 min at room temperature. They were then permeabilized with 0.5% triton X-100 (Merck, KGaA, Darmstadt, Germany) in TBS for 10 min and thereafter washed with 0.025% triton X-100 in TBS. Subsequently, a blocking step was performed with PBS containing 10% FCS. The cells were then treated with anti-ZO-1 antibody (Cat # A32728, Alexa Flour 647, Rockford, USA), which was diluted 1:200 in TBS and then incubated for 45 min in the dark at room temperature. After that, the cells were washed with PBS three times and counterstained with Hoechst (1:1000 times diluted in TBS). Samples were then analyzed by a confocal laser scanning microscope (LSM 700, Zeiss). Microscopic images of the fixed samples were acquired using a 63× objective. The images were processed in FIJI.

WST-Assay

The water-soluble tetrazolium salt-1 (WST-1) assay is the cell viability assay for analyzing the number of viable cells by the cleavage of tetrazolium salts added to the culture medium. The tetrazolium salts (slightly red) are cleaved to formazan (yellow) by cellular enzymes called mitochondrial dehydrogenase. The increase in the number of cells elevates the overall activity of mitochondrial dehydrogenase in the sample, as a result of which the amount of formazan dye formed is also increased. Since the enzyme is produced by metabolically active cells, this assay thus quantifies the metabolically active viable cells. The WST-assay was performed according to the protocol of Sigma-Aldrich (Cat # 5015944001 Roche, Basel, Switzerland). After the particle exposure as mentioned above, the cells in the 96-well plate were washed with PBS, and 100 μL of fresh DMEM medium without phenol red containing 10% WST reagent (Roche, Basel, Switzerland) was added into each well. Then the 96-well plate was incubated for half an hour in the incubator (37 °C, 5% CO2), and the absorbance was measured with a plate reader (BioTek Instruments, Inc., Vermont, USA) at 450 and 630 nm (reference wavelength). Eight technical replicates were taken.

ROS-Assay

The level of intracellular reactive oxygen species (ROS) generation was determined by using a CellROX deep red reagent (Cat # C10422, Invitrogen, Thermo fisher Scientific, Waltham, Massachusetts, USA) after treatment with different concentrations of NPs. According to the protocol provided by Invitrogen (Thermo fisher Scientific, Waltham, Massachusetts, USA), a final concentration of 5 μM of CellROX reagent was prepared. An amount of 20 μL of CellROX deep red reagent was mixed with 10 mL of complete DMEM medium without Phenol Red (PAN Biotech GmbH, Aidenbach, Germany). The old medium in a 96-well plate was removed, and the cells were washed three times with PBS. Then, 100 μL of the diluted CellROX deep red reagent was added to each well. Care was taken to avoid bubbles while pipetting, so negative pipetting was performed. The 96-well plate was kept in the Incubator for half an hour at 37 °C, with 5% CO2. The reactive oxygen species were measured by reading the fluorescence of the solutions kept in a 96-well plate using the plate reader (BioTek Instruments, Inc., Vermont, USA) at 640 nm (excitation) and 665 nm (emission) wavelengths. Calculations were performed using Microsoft Excel 2016 (Microsoft Corporation, Redmond, EA, USA).

RT-qPCR

The cells were seeded at the seeding density of 300 000 in 6-well plates. When the cells were confluent, they were treated with all three concentrations of CNP and dendrimers and incubated for 24 h. The cells were detached from wells using trypsin (500 μL), and 1 mL of DMEM was used to stop the reaction. The cell pellet was generated by centrifuging the cells containing 1.5 mL of Eppendorf for 6 min at 4 °C and 9000g (rcf). Then the RNA extraction was performed using the protocol as described by the manufacturer (Qiagen GmbH, Hilden, Germany). The concentration of RNA was measured using the NanoDrop spectrophotometer with the software NanoDrop 1000 3.7.1 (peQLab, Biotechnology GmbH, Erlangen, Germany). After that, cDNA was generated using the protocol of a reverse transcriptase (RT) kit (Applied Biosystem, Thermo Fisher Scientific, USA). There were two technical replicates, so for three different concentrations of each nanoparticle and two repeats of control, a master mix of 14 reactions was prepared (1 extra for having extra volume, so 15 total reactions). An amount of 500 ng of RNA mixed with 5 μL of double distilled (dd) water and 5 μL of the master mix was added in each flat cap strip (Nerbe plus GmbH and Co., Winsen, Germany). The strips were vortexed in a microcentrifuge (VWR-mini Star, Korea) shortly before starting the PCR. The cDNA was synthesized using PCR (PeQLab, Biotechnology GmbH). An amount of 40 μL of RNase-free water was added in cDNA and stored at −20 °C. Finally, for RT-qPCR sample preparation, a 96-well plate from the PCR cool block was taken. HPRT was used as a reference gene. A fast SYBR green master mix was made for each gene ZO-1 and HPRT using the protocol in the RT-PCR kit (Applied Biosystems, Thermo Fischer Scientific, USA). Amounts of 1 μL of cDNA and 9 μL of the master mix were added in each well. The plate was sealed with an RT-PCR seal which is an optical clear film (Th. Geyer GmbH, Renningen, Germany) and shortly spinned in a miniplate spinner (mps 1000, Labnet). The gene expression of ZO-1 was then measured using RT-PCR (Quant Studio 3, Applied Biosystems, Thermo Fischer Scientific, USA). The results were exported and processed using Microsoft Excel 2016 (Microsoft Corporation, Redmond, EA, USA).

Statistical Analysis

All data are shown as mean ± standard deviation. If not stated otherwise, the data represent three independent experiments. For statistical analysis, a two-sample t-test was performed using Origin Pro 9.1 G64 Bit software. *P ≤ 0.05 was considered as significant; **P ≤ 0.0001; and ***P ≤ 10E–10.

Results and Discussion

QSAR Analysis to Select the Optimum Physicochemical Parameters for the Target Formulations

We connect the in silico methods, optical microrheology, and toxicological profiling of two classes of NMs demonstrated in Figure by showing the different steps of the sequential workflow.

Figure 1

Schematic showing the different steps of the characterization: (A) in silico analysis, (B) DLS microrheology-based viscoelasticity quantification, and (C) in vitro assays for the subsequent analysis of relevant toxicology parameters in detail.

We used a very different class of nanomaterials to correlate the microrheological changes when they are mixed into the biological medium. Since we rely on viscoelastic measurement using dynamic light scattering (DLS) in aqueous media and a subsequent cytotoxicity investigation, it is useful to have the solubility and lipophilicity of the two types of materials in priori. Further optimizing the physicochemical parameters in silico additionally facilitates the ease of handling and the preparation formulation for the toxicological analysis, as recently machine learning advances boosted the toxicology analysis in vitro.[55−57] After analyses, we compared the predicted values with optimum solubility and lipophilicity, as shown in Table (log S as the first priority and then log P, TPSA, and log D). For the selection, we considered the optimum solubility value with moderate lipophilicity, TPSA, and log D to be suitable for the formulation of the microrheology and toxicological profiling. We present in Table only eight representative compounds with minimum, moderate, optimal, and suboptimal log S values after analysis of the list of dendrimers and CNP quasi-SMILES. Once the best compounds for each NM were selected, which were Den S1 and CNP2 (Table ), we looked at the Sigma-Aldrich chemical information database based on their CAS number or quasi-SMILE to purchase and began with experimental analysis for viscoelastic and toxicology measurements[58] (accessed 5 November 2021).

Table 1

In Silico Determination of the Physiochemical Properties Like Solubility (log S), Lipophilicity (log P), Diffusion Coefficient (log D), and Topological Polar Surface Area (TPSA) of Different Dendrimers and CNPsa

sample	log S (log mol/L)	log P (log mol/L)	log D (log L/kg)	TPSA
Den S1	–0.11	–4.67	–4.097	167
Den S2	0.394	–8.763	–6.578	576.8
Den S3	–0.71	–3.899	–3.683	226.9
Den S4	4.242	–20.673	–14.521	1036.6
CNP1	–4.783	–0.3636	–0.424	704.6
CNP2	0.688	–0.206	–0.36	172.1
CNP3	–0.806	–1.6494	–0.247	316.2
CNP4	0.033	–0.2401	–0.254	79.8
optimal range	–4–0.5	0–3	1–3	0–140

Den S(1–4): dendrimer samples (1–4). CNP (1–4): cerium dioxide (1–4).

Nanoparticle Size Distribution and Zeta Potential

The hydrodynamic size and aggregation behavior through zeta potential of CNP and dendrimers at different concentrations were characterized using DLS. With the increase in concentration, there is a gradual increase in the hydrodynamic size denoted by the Z-average (d/nm) as well as polydispersity index (PI) and a decrease in the diffusion coefficient for both CNP and dendrimers, as shown in Table . The diffusion coefficient is directly proportional to the concentration gradient between the particle along with its corona and the solvent. This concentration gradient is high in the case of low concentrated suspension, due to which a consequent decrease in the diffusion coefficient is seen with the increase in concentration (Table ). On the other hand, the increase in the concentration increases the PI, which indicates the polydispersity of the suspension and broadness of the size distribution. This trend is clearly seen in Figure , where 10× and 100× concentrations of CNP have well-defined narrow peaks, whereas 1000× concentration shows a broadness in peak along with an additional peak. This depicts the aggregation behavior of CNPs, which increases with increasing concentration. However, in the case of dendrimers, there is a shift of the Gaussian distribution peak to higher size with the increase in concentration, possibly due to the interaction with the organic particles in DMEM since dendrimers are also organic nanoparticles. Overall, the PI is well below 0.7, which is regarded as highly polydisperse, and since our PI values are around 0.5 or below, they meet the standards for DLS measurements. Apart from that the zeta potential values for CNP at different concentrations are recorded between 0 and −9.4, which show more aggregation and less stability. Comparatively, the dendrimers show zeta potential between −3 and −14.2 at three different concentrations, which is relatively stable compared to CNPs. The XRD data show the crystalline nature of cerium oxide, and a mixed redox population of cerium dioxide with high Ce3+ concentration was observed (Supporting Information Figure 1).

Table 2

NP Characterization of Different Concentrations of CNPs and Dendrimers in DMEM Mediaa

SN.	sample name	Z-average (d/nm)	polydispersity index (PDI)	diffusion coefficient (μ2/s)	zeta potential (mV)
1	CNP 10×	316.7	0.261	1.35	0.409
2	CNP 100×	411	0.284	1.04	–9.39
3	CNP 1000×	451.1	0.475	0.945	–0.521
4	dendrimers 1×	25.18	0.434	28.1	–13.3
5	dendrimers 5×	34.37	0.503	12.4	–3.82
6	dendrimers 10×	44.42	0.544	9.6	–14.2

For CNP 10×, 100×, and 1000× represent 10 μg/mL, 100 μg/mL, and 1000 μg/mL, respectively. For dendrimers, 1×, 5×, and 10× represent 1%, 5%, and 10% of the stock solution (10% weight in methanol) of dendrimers, respectively.

Figure 2

Size distribution of different concentrations of cerium dioxide nanoparticles (A, B, C) and dendrimers (D, E, F) in DMEM media. A, B, and C represent 10 μg/mL, 100 μg/mL, and 1000 μg/mL concentrations of CNP, respectively, and D, E, and F represent 1%, 5%, and 10% (1×, 5×, and 10×) of the stock solution (10% weight in methanol) of dendrimers, respectively.

Microrheological Characterization

A wide range of microrheological studies have been carried out, such as high-throughput viscosity measurement of proteins,[26] viscoelasticity of other complex fluids,[59] and biological samples.[60] However, the influence of the nanoparticle concentration on the microrheology of the cell culture media like DMEM has never been studied. In order to effectively use optical microrheological techniques, one of the main challenges is to determine which tracer particles to use to probe the rheological response. Since the hydrodynamic size of the CNP is in the range of 400 nm and that of dendrimers is in the range of 40 nm, a relatively larger tracer particle is required, which could dominate the light scattering of the NPs. Therefore, polystyrene beads with a diameter of 1.5 μm (Polysciences, Inc., Warrington, United States) were used to employing stress on the NPs. Before the measurement of the microrheology of NPs, the compatibility of the nanoparticles and the tracer particles was checked for each concentration by determining the zeta potential of the tracer particles in DMEM media with and without the NPs. It was observed that the zeta potential in the presence of both a tracer and NPs at different concentrations was within the set limit of 5 mV, which could confirm the compatibility of polystyrene beads and NPs used for the experiment. Apart from that, the concentration of polystyrene beads was used in such a way that their particle size distribution (PSD) accounted for at least 95% of the total area.

Microrheological properties like mean square displacement and viscoelasticity have important contributions to the aggregation properties.[24] They measure the spatial extent covered by random motion. Insights into the motion of the nanoparticles can be quantitatively gained through the MSD, which can be obtained from the electric field autocorrelation function.[61] Since the temperature (25 °C) as well as the pH (7.4) of DMEM media were kept constant, the NP suspension in DMEM acted as a nonviscous system, although DMEM has 10% fetal bovine serum which can act as a cocktail of proteins. The protein is not denatured at 25 °C and therefore does not have large length-scale structures. As depicted by Figure (A,B) and the Supporting Information Figure 2, both the MSD and correlation coefficient comply with each other. Figure shows that the MSD is inversely proportional to the viscoelastic modulus, which is in agreement with the generalized Stokes–Einstein relation.[62] It is evident from Figure that for CNP an increase in concentration decreases the viscoelastic modulus, and the MSD is captured for a relatively shorter time window, which can be due to the aggregating behavior of CNPs with the increasing concentration. However, in the case of dendrimers, an increase in concentration tends to increase the viscoelasticity, and the MSD is recorded for a relatively longer time frame as compared to CNPs. Also, the MSDs for dendrimers run close to each other and even overlap with each other toward the end. This is because the dendrimers and DMEM media are both in the organic phase and increase in concentration, which increases the total viscoelasticity of the whole system. However, at the highest concentration (10×), the dendrimers show a sharp decline after reaching the peak. This could be due to the interaction of dendrimers with the other organic molecules of the DMEM media.[63] Nevertheless, in the case of CNPs the lowest concentration seems to contribute to the overall viscoelasticity due to the fact that there is less aggregation of 10× CNPs and that there is a relatively stable suspension. However, as the concentration increases, there is aggregation of the CNPs, and this results in the decrease in the overall viscoelasticity. In the case of 1000×, there is a sharp decline since the CNPs aggregate quickly at this concentration.

Figure 3

Mean square displacement (A, B) and viscoelastic moduli (C, D) of different concentrations of (A, C) cerium dioxide and (B, D) dendrimers in DMEM media. For CNP, 10×, 100×, and 1000× represent 10 μg/mL, 100 μg/mL, and 1000 μg/mL, respectively. For dendrimers 1×, 5×, and 10× represent 1%, 5%, and 10% of the stock solution (10% weight in methanol) of dendrimers, respectively.

Metabolic Activity Analysis

The epithelial kidney cells MDCK were chosen for the experiment as the model cell lines as the kidney plays an important role in renal clearance, which makes it a favorable place for accumulation and toxicity of foreign bodies including nanoparticles. Due to easy cultivation, maintenance of the functional membrane transporters and cytochromes makes them the mostly used cell line for in vitro studies.[64] The MDCK cells in the submerged culture were exposed to three different concentrations of both CNPs and dendrimers for 24 h, after which the influence on the metabolic activity (WST-1) and generation of ROS was investigated. As shown in Figure A, the highest concentration (1000×) of CNP shows the maximum viability as compared to the positive control and other concentrations of CNPs and dendrimers. This could be because CNPs are known as nanoenzymes (catalysts), and they have effect on cell proliferation. In a study carried out by Wang and his research group, the effect and mechanisms of CNP on osteoblasts, the cell proliferation, cellular uptake, endocytosis mechanism, cell cycle, and cell adhesion forces were analyzed. The results showed that CNP promoted the proliferation of a primary osteoblast as well as increased the cell adhesion force,[65] whereas another study carried out by Zhang et al. suggests that the concentration of Ce3+ and the time of culture had an effect on the proliferation, differentiation, adipocyte transdifferentiation, and mineralization function of primary osteoblasts.[66] Apart from that, according to the studies done by Dai and his co-workers, the aggregation potential increases with increasing concentration and changing variables.[67] This could act as a factor influencing the availability of CNPs to the cells in submerged culture with the increase in concentration. However, our results contradict the study conducted by Sauer et al. in which they reported the cytotoxic effect of the CNP on the rat lung cells at the concentration of 1000 μg/mL and no toxicity below this concentration.[68] Since the CNP showed the absorbance at the wavelength of the WST-1 measurement (Supporting Information Figure 3), the cells exposed to CNPs were washed twice before the exposure of the WST-1 reagent, but the complete removal of the CNP was not possible for 100 μg/mL and 1000 μg/mL in order to prevent the washaway of cells. Some part of the absorbance for 100 and 1000 μg/mL may be contributed by the leftover CNP. In the case of dendrimers, as depicted in Figure A, the percentage viability or metabolic activity decrease with the increase in concentration. Poly(amido amine) (PAMAM) dendrimers are characterized by concentration- and generation-dependent toxicity.[69,70] Molecular interactions between negatively charged cell membranes and positively charged dendrimers explain the cytotoxicity of cationic dendrimers. As a result of such interactions, nanopores are formed in the cell membrane, which cause damage, leakage of cellular content, and subsequently cell death.[71] However, studies show that the dendrimers terminated with neutral or anionic groups seem to be much less toxic than cation dendrimers.[72] A new insight was shed by a study carried out by Mukherjee et al. in which they disclosed an indirect impact of PAMAM dendrimers on cell viability. They suggested that the indirect mechanism of generation-dependent PAMAM cytotoxicity results from the depletion of medium components due to the absorption of proteins from media by dendrimers.[63] This could also explain the decrease in cell viability with an increase in dendrimer concentration, as shown in Figure A. Our dendrimer results comply with the study conducted by Malik et al. in which carboxylate PAMAM dendrimers showed a hemolytic effect after 24 h.[73] Although a significant decrease in cell viability should have corresponded to severe changes in cellular processes, such as massive ROS generation or vice versa, the WST-1 result does not exactly correspond to the ROS results for both CNP and dendrimers. Nevertheless, as compared to the control, the reactive oxygen species (ROS) is shown to be high in all the treated samples (Figure B). Correlating viscoelasticity with the metabolic changes of the cell, we see a clear trend for both categories of NMs with subsequent changes in metabolic activity. Comparing Figure C vs Figure A, we see that cerium dioxide with the highest viscoelasticity exhibits minimum metabolic activity and vice versa for the dendrimer (compare Figure D vs Figure A). The possible explanation for the different behavior of two classes of NMs could be explained based on the differences of versatile surface properties of inorganic cerium dioxide versus the stable behavior of the organic dendrimer. As aforementioned, the metabolic activity pf CNPs increases with the increasing concentration due to the nanozyme effects of CNPs, while the viscoelasticity decreases because of the sticky and aggregating nature of CNPs. We observe the opposite effect and correlation in surface-stable dendrimers.[65]

Figure 4

(A) WST-1 and (B) ROS-assay after 24 h exposure of CNPs and dendrimers at different concentrations. Control represents untreated cells. The bars show a mean of 3 biological replicates (3n). CNPs 10×, 100×, and 1000× represent 10 μg/mL, 100 μg/mL, and 1000 μg/mL, respectively. Dendrimers 1×, 5×, and 10× represent 1%, 5%, and 10% of the stock solution (10% weight in methanol) of dendrimers, respectively. Data are shown as mean ± SD: *p ≤ 0.05, **P ≤ 0.0001, and ***P ≤ 10E–10 is compared to the respective control group.

Characterization of Barrier Function with Transepithelial Resistance (TEER) Measurement and Tight Junction Staining and ZO-1 Gene Expression through q-PCR

Unlike endothelial cells, epithelial cells like MDCK cells are able to produce tight junction proteins and form a monolayer.[74] The barrier function of the in vitro MDCK cell monolayer was characterized using TEER measurements and immunostaining analysis of the tight junction protein zonula occludens-1 (ZO-1). An electrical resistance measurement across a cellular monolayer using TEER ensures the integrity and permeability of a monolayer. For this, MDCK cells were cultured in the 12-well hanging inserts in the submerged culture, which was exposed to various concentrations of CNPs and dendrimers, and the subsequent effect on barrier function was measured using TEER after the first, third, and seventh days of exposure. A total of two biological replicate and three technical replicates were considered during the measurement. As shown in the figure, the maximum TEER measurements were recorded at the third day after the exposure of the MDCK cells with various CNP and dendrimer concentrations. According to Lars et al. the resistance values as well as time required to achieve the high TEER value differ from cell line to cell line along with the nutrient content and the cell culture conditions.[75] The TEER values show the maximum at the third day for each sample, including in the control without treatment with NPs. In the case of CNPs, the resistance shows a dose-dependent increase which could be due to the aggregating nature of CNPs.[76,77] Apart from that, CNPs also show the cell proliferating effect which may have caused more proliferation and more tight junctions with the increase in the concentration.[65] In the case of the sample treated with a 10× dendrimer, a low resistance of 40.52 Ω*cm2 is measured, rendering the high permeability after 24 h exposure. On the third day, the average TEER measurement was observed to be 129.43 Ω*cm2. Since the measurements are taken at only three different days, the exact days and exact highest TEER values were not recorded, but the trend is compliant with the findings of Cho et al., which stated the highest TEER value for the MDCK cell line to be at the fifth day.[78] The literature describes two different strains of MDCK: strain I generates an epithelium with transepithelial resistance (TEER) above 1000 Ω*cm2,[79,80] and strain II generates an epithelium with TEER of 100 Ω*cm2 or less.[79,81] Since all the values of TEER generated were in the range of around 100 Ω*cm2, the strain of MDCK we have might be strain II. For TEER measurements performed manually with a Millicell ERS-2, the electrode position has a significant impact on the resistance value. One explanation for the large standard deviation obtained in our experiments could be this factor. Such problems can be mitigated by the use of integrated electrodes in organ on chip systems for the TEER measurements.[82] Incorporating immobilized TEER electrodes directly within the chip model and in close proximity to the cellular monolayer will not only reduce the contribution of electrical resistance from the cell culture medium but also reduce any motion-related signal noise.[83]

ZO-1 proteins are well expressed by MDCK cell lines. The molecular characterization of ZO-1 proteins[70] as well as their function in the assembly of a tight junction in the MDCK cell line are well studied.[84] The immunostaining of the tight junction protein was performed after 24 h of exposure of the MDCK cells with the different concentrations of CNP and dendrimers. As per the TEER values, the cells exposed with 10× dendrimers showed the least TEER measurement after 24 h. This is in agreement with the ZO-1 staining which shows the distorted ZO-1 formation with the cells in the patch rather than the confluent (Figure ) and also the reduced confocal microscopy signal intensities for both ZO-1 staining and DAPI staining as shown in Figure B. This shows the cytotoxic nature and loss of viability caused by the dendrimers at higher concentrations. The measured ZO-1 mRNA by using the qPCR method as shown in Figure C further validates the immunostaining results. As seen in Figure C, the exposure of MDCK cells with various concentrations of CNP and dendrimers tends to influence the expression of ZO-1 genes differently (Figure ). 10× CNPs and all three concentrations of dendrimers show a slightly more than 1-fold and 100× CNP by a 2.7-fold increase of ZO-1 expression, whereas 1000× CNP exposure tends to downregulate the ZO-1 expression. The overexpression of the ZO-1 gene as compared to control may be explained as a defense mechanism of the cells to the external stimuli.

Figure 6

Confocal images of cells after 24 h exposure with different concentrations of dendrimers. The cells were immunostained with the ZO-1 antibody conjugated with Alexa fluor 647 and finally stained with DAPI. The first row depicts the overlay of both ZO-1 staining and DAPI staining. The second row represents ZO-1 staining, and the third row represents DAPI staining. The first column represents the controls which are not exposed to any NPs. The concentrations 1×, 5×, and 10× represent 1%, 5%, and 10% of the stock solution (10% weight in methanol) of dendrimers, respectively.

Figure 5

(A) TEER measurement of cells treated with different concentrations of cerium dioxide nanoparticles (CNPs) and dendrimers (Den) after the first, third, and seventh day of exposure. (B) Mean signal intensity of ZO-1 and DAPI in cells treated with different concentrations of cerium dioxide nanoparticles (CNPs) and dendrimers (Den) obtained by using a confocal microscope. (C) Relative gene expression of the ZO-1 tight junction protein after 24 h exposure of CNPs and dendrimers at different concentrations. Control represents untreated cells. The bars show a mean of three biological replicates (3n). For CNPs 10×, 100×, and 1000× represent 10 μg/mL, 100 μg/mL, and 1000 μg/mL, respectively. For dendrimers 1×, 5×, and 10× represent 1%, 5%, and 10% of the stock solution (10% weight in methanol) of dendrimers, respectively. Data are shown as mean ± SD: *p ≤ 0.05, **P ≤ 0.0001, and ***P ≤ 10E–10 are compared to the respective control group.

Figure 7

Confocal images of cells after 24 h exposure with different concentrations of CNPs. The cells were immunostained with the ZO-1 antibody conjugated with Alexa fluor 647 and then stained with DAPI. The first row depicts the overlay of both ZO-1 staining and DAPI staining. The second row represents ZO-1 staining, and the third row represents DAPI staining. The first column represents the control sample which is not exposed to any NPs. The concentrations 10×, 100×, and 1000× represent 10 μg/mL, 100 μg/mL, and 1000 μg/mL of CNPs, respectively.

In our previous study,[85] we established a machine-learning-based graph modeling and correlation approach using a tight junction protein ZO-1-mediated alteration in the cell phenotype to quantify and propose it as indices of cell–NM interactions. We found that the phenotypic variation such as cell shape and nucleus area in the epithelial cell is determined by the physicochemical properties (e.g., shape, size, zeta potential, concentration, diffusion coefficients, polydispersity, and so on) of the different classes of nanomaterials, which critically regulate the intracellular uptake or cell membrane interactions when exposed to the epithelial cells at sublethal concentrations. By analyzing the intrinsic and extrinsic properties of the representative nanomaterials (NMs) using optical (dynamic light scattering, NP tracking analysis) methods, a set of nanodescriptors related to cell–NM interactions via phenotype adjustments were created. In relation with toxicology, we established a machine-learning algorithm fitting correlation function, which was used to successfully predict cell and nuclei shapes and polarity functions as phenotypic markers for five different classes of organic and inorganic nanoparticles.

Conclusion

By taking the growing applicability of NPs in consumer products into consideration, in this study we established a new method based on DLS microrheology to analyze the MSD and viscoelasticity of two model NPs that are dendrimers and CNPs at three different concentrations. MSD results showed that they were inversely proportional to the viscoelastic modulus, which infers that our results are in agreement with the generalized Stokes–Einstein relation. The decrease in viscoelastic modulus with the increase in concentration suggests the aggregating potential of CNP, whereas the increasing viscoelasticity with the increasing concentration of dendrimers reflects on the organic nature of the dendrimers. Further investigation into the toxicological profile showed that dendrimers show cytotoxic effects with the increasing concentrations and are in agreement with TEER, ZO-1 immunostaining, and ZO-1 expression experiments. However, in the case of CNPs due to its high aggregating nature, WST and TEER results at higher concentrations seem to interfere, and further investigation into this matter is required. The inverse correlation in viscoelasticity and the metabolic nature of cerium dioxide versus dendrimers might arise due to their different category of material property (inorganic vs organic). This study emphasizes the importance of microrheology for the NP characterization and its influence on the toxicological profile. Moreover, our approach is based on the difference in the concentration of two different model nanoparticles. The findings in this study can support further microrheological studies by changing several parameters like temperature, pH, different biological media, and more. Further research in this field can help in the development of in vitro methods for the safety and risk assessment of nanoparticles and additionally narrow the gap between in vivo and in vitro models.