Toxicity studies are important in biochemical and medical research, and essential prior to the commercial use of newly developed chemicals and nanomaterials. The health and safety of researchers, production workers, end-users, and bystanders who may come into contact with new or innovative products, and any secondary products that may arise from the degradation of such products, is of great concern to government regulatory bodies and society as a whole [1
]. The benefits of the effective toxicity screening of chemicals and nanomaterials prior to their commercialization include better community health outlooks, reduced costs (healthcare and/or compensation payments), and faster paths to the market for new non-toxic products. Currently, biochemical and medical products are subjected to extensive testing before adoption or commercialization, but this is costly in terms of labour, time, and money. Many companies are specifically set up to assist with performing the biocompatibility (i.e., toxicology) screening for new formulations. Analysts predict that the in vitro toxicity testing market value will reach approximately 10 billion dollars in 2017 [4
]. Thus, new technologies for toxicity screening are attractive for their perceived economic, social, and environmental benefits.
Cytotoxicity analysis at a cellular level is concerned with how a given toxic chemical affects a given cell’s physical structure (e.g., membrane integrity) and its ability to viably replicate without damage to the daughter cell’s genetic code or normal functionality [5
]. Thorough cytotoxicity screening of a chemical requires the studied toxicant to be tested against different cell types, and is generally performed in static fluid in well plates. This approach requires laborious liquid handling, long hours of incubation, and large reagent volumes. The idea of an all-encompassing cytotoxicity test is a daunting task, as the adult human body contains trillions of eukaryotic cells with different phenotypes and functionality [6
] and, according to Vickaryous [7
], the number of unique cell types is 411, including 145 types of neuronal cells. The many different cell types in the human body makes the effective in vitro screening of potential toxic effects an enormous challenge, further complicated by the types of analysis required (e.g., viability, cell metabolism, and biochemistry). The growing number of novel chemicals and materials that are being suggested for various commercial applications multiplies the size of the challenge and calls for much faster and cheaper methods.
Nanotoxicity is an important subgroup of toxicity which considers the damaging effect of nanomaterials on cells. The first reports on the toxicity of nanomaterials on mammalian biology were reported in the 1990’s by Jani et al. [8
] and Penney et al. [9
]. Now, an awareness of the potential toxicity of those materials has reached beyond the scientific community to include regulators and consumers, who, in many cases, are not equipped with enough information to guide their decisions. All the while, more nanomaterials are being created and incorporated into consumer products, from sunscreens [10
] and cosmetics [2
], to antibacterial and antifouling coatings [13
]. A major apprehension pertaining to the exposure of the human body to nanoparticles is their physical size (typically 1–100 nm), which can allow them to enter cells via pathways that naturally transport biological and chemical species [5
]. Many nanoparticles can form free radicals and reactive oxygen species (ROS) [16
] from surrounding molecules, due to their increased reactivity and high surface area, which have the ability to cause damage to cellular membranes and proteins within cells, leading to inflammation and oxidative stress [17
]. Nanomaterials can also disassociate into ionic species upon reacting with biological tissue and fluids such as gastric juices, which can lead to the release of reactive ions that can damage the cellular environment and cause toxic effects [19
]. Genotoxicity can occur if a nanoparticle interferes with the delicate process of DNA transcription and replication, potentially knocking out one or more genes from the sequence and causing a range of negative effects, such as apoptosis (in which case the cells die off) [21
] or mutation, which can lead to the cells becoming cancerous [24
Meaningful nanotoxicity studies require high-throughput screening methods, as the toxic effects of a nanomaterial can be dependent on the core composition, size, shape, and surface modification that material possesses, suggesting that a high number of materials should be tested. The many different cell and tissue types of the human body can react in different ways to any given nanomaterial, meaning that an ideal toxicology screen should test every unique cell type. Thus, the desire to create a fast, stable screening process for the maximum number of combinations of nanoparticles and human cells possible is of great interest to the industry and health sectors. Where practical, these methods will enable the determination of safety exposure levels and maintain the health of both workers in the nanomaterial industries and the end-users of products containing nanomaterials. To this end, microfluidic approaches to toxicity screening have been investigated and are the topic of this review.
2. Nanomaterial Exposure Pathways in Biology
Nanomaterials can be taken up into a living body via the natural internalization pathways of ingestion, inhalation, or dermal uptake [25
], or through direct injection if used in nanomedicine [26
], and can pose a substantial risk to the viability of a cell, depending on the nature of their interaction with living cells. Once inside the body, nanomaterials can transmit from one tissue to another via the bloodstream and the surrounding tissues, potentially migrating into other organs such as the kidney or spleen. Once they enter the body, they come into contact with the body’s cellular structure and potentially gain entrance into the cells themselves [27
]. A diagram of nanoparticles and the bodily areas that they are able to access after uptake via inhalation is shown in Figure 1
. The uptake of nanoparticles to the body and different organs is particle size dependent. As an example of size-dependency on the uptake of nanoparticles, it was shown by Jani et al. [8
] that when a range of polystyrene particles ranging from 50 nm to 3 µm was introduced to a rat model via ingestion pathways, no particles above 100 nm reached the bone marrow and none larger than 300 nm were present in the bloodstream, whereas the 50 nm and 100 nm particles were absorbed at rates of 34% and 26%, respectively, into the liver, spleen, blood, and bone marrow tissues.
Nanoparticles can pass through the cellular membrane via the passive transport mechanics of diffusion and osmosis, requiring no activation energy [29
]. Alternatively, nanoparticles can be taken into the cell via active transport mechanisms, in which carrier proteins or ionic pumps within the cell membrane attach to the particle and use energy to move them across the cell membrane into the cytoplasm [27
]. Inorganic nanoparticles that require this mechanism to cross the membrane are often blocked from entering cells, unless they are coated with a biomolecule (such as transferrin) that facilitates their uptake by the carrier proteins, as was shown in Yang et al. [31
] in the case of transferrin-conjugated gold nanoparticles.
Nanoparticles that do not enter the cell via membrane diffusion or through membrane pores can still be transported into the cell via endocytosis [32
]—that is, the cytoplasm of the cell extends around a particle and engulfs it, forming an endocytic vesicle that retains them in the inner cytoplasm of the cell. From here, the particles can either: escape the vesicle and remain in the cytoplasm; persist in the vesicle and be consumed by a lysosome (an organelle full of enzymes that serve to digest foreign bodies that exist in the cytoplasm); or combine with other vesicles to form a multi-vesicle endosome contained in a secondary membrane, which stabilizes and contains the individual vesicles [30
Certain nanoparticles possess the ability to alter or bypass the membrane permeability, depending on their ionic potential or their shape. Nanoparticles that are shaped with sharp points or edges can mechanically damage the cell membrane, creating temporary nanochannels through which they can enter the cytoplasm [33
]. This can be exploited to create drug delivery mechanisms by coating nanoparticles such as carbon nanotubes with biocompatible molecules which attach and enter cells, or manufacturing nanoneedles from materials such as silicon or polymers that can mechanically puncture cells to deliver drugs directly into the cytoplasm [34
]; however, toxic nanomaterials could very easily enter cells by the same mechanism and induce cytotoxic effects.
The above pathways and cellular interactions are complex and very sensitive to the size, shape, chemistry, and surface charge of the nanomaterials, meaning that the importance of a high-throughput evaluation of nanotoxicity is growing commensurate with the rapid development of nanotechnology. In recent years, microfluidics technology has resulted in large impacts on the cytotoxicity screening of nanoparticles. This will be discussed in the following section, to underpin the later discussion of nanotoxicity screening using microfluidic chips.
4. Microfluidics for Nanotoxicity Screening
Microfluidic devices offer many advantages when it comes to cellular analysis with small sample volumes, reduced costs, controllability, and reproducibility. In addition, microfluidics offers the ability to introduce multiple biological conditions in a single device, and replicate in vivo conditions and dimensions. Thus, its usefulness in producing a high-throughput platform for toxicological experiments with nanomaterials is promising for screening applications. Small sample volumes can be very important when dealing with nanomaterial testing. Given that particulate matter may be produced from nanomaterials in extremely low concentrations, and that some nanomaterials are very expensive and produced in low quantities, there may be very limited amounts or diluted analytes to perform testing on. The behaviour of nanoparticles under flow is more difficult to quantify compared with that of macroscale particles. Therefore, the reproducible flow profiles and concentration gradients achieved in microfluidic nanotoxicity testing presents a distinct advantage.
Similar to cytotoxicity tests with other chemicals, the toxicity testing of nanomaterials is generally performed in bulk by seeding cells suspended in growth media into well plates using pipettes [17
]. The cells are then exposed to nanomaterials in static conditions, which may cause the nanomaterials to adsorb or sediment onto the exposed surface of the cells under gravity [60
]. Once these cells have been exposed to the nanomaterials for a certain time period, the cells can be assayed for their viability in a number of ways. Particular cell types may exhibit changed membrane permeability values and nanomaterial uptake properties when their morphology changes under flow, as compared to their sedentary morphologies, and thus, a static nanotoxicity test may provide inaccurate data.
A review by Mahto et al. [61
] goes into exceptional detail on the subject of nanomaterials in microfluidic environments, and brings up a number of important details regarding nanotoxicity. First, it refers to a number of studies that show that nanoparticles, depending on their size and shape, are passively taken up into nearly all cell types via endocytotic pathways [27
]. Secondly, it refers to the potential pitfalls of current in vitro nanotoxicity testing methods. Notably, nanoparticles often react with the organic dyes commonly used in cell-based assays, meaning that they cannot be properly assayed [64
]. Cell exposure to nanomaterials is often improperly controlled, with aggregation and sedimentation leading to very different exposure profiles (as seen in Figure 5
). Current nanotoxicity testing also utilizes immortalized cell lines which differ significantly from primary human tissue [65
]. The review article mentions platforms that look to circumvent the current issues with nanotoxicity testing, such as the device produced by Richter et al. [66
], which used non-invasive electrodes to electrochemically measure the amount of collagen production as a label-free marker of cell viability. This device could detect the nanotoxicity of silver nanoparticles after 2 h of exposure, as compared to the lack of nanotoxicity seen for gold nanoparticles over a period of 24 h. It could also detect reductions in collagen production, given a sub-lethal concentration of silver nanoparticles. For further discussion, the authors recommend Mahto’s review paper to the reader.
The well-plate methodology for nanotoxicity screening has been debated for its suitability in replicating in vivo conditions [27
]. This is investigated in detail in a paper by Mahto et al. [60
], where nanoparticles influenced by gravity in a static system formed a concentration gradient within a cell culture plate. The static conditions involved in well-plate analysis, i.e., pipetting nanomaterial solutions on top of a cell culture and allowing them to sediment on top of the cellular layer, are thought to have limitations in providing accurate nanotoxicology data for cells that are under shear stress from flowing biological fluids such as arterial, lymphatic, and renal cells in vivo. To compare the differences between static and flowing nanoparticles, they tested a sample of core/shell CdSe/ZnSe quantum dots in static tissue culture plates for 12 h at 8–80 pM to discover the optimal cytotoxic range, and then exposed the same quantum dots through a microfluidic concentration gradient generator in cell culture medium to murine embryonic fibroblast cells. When exposed to the quantum dots under flow, the cells exhibited apoptosis effects, namely detachment and dose-dependent morphological changes. However, the difference between the two exposure conditions at 40 pM was significant, in that the static conditions showed higher percentages of cell death and increased cell deformities, suggested to be due to the physicochemical stress of the sedimentation of quantum dots onto the cell membranes.
The effect of the shear-stress effect of flowing media over cells has been investigated by groups such as Kim et al. [71
], utilizing bound endothelial cells in a single microchannel and exposing them to mesoporous silica nanoparticles. The shear-stress forces were tuned to mimic those expected in the arterial and capillary system of a healthy human (5–6 N/m2
), in order to observe any differences between these values and those of a static system. The nanoparticle concentration was also tuned to eliminate the effect of higher dosages during periods of a higher flow rate/higher shear-stress, so that the shear-stress forces were the primary variable. The unmodified silica nanoparticles used in this paper were found to increase in toxicity as the shear-stress increased, indicating an increase of cell membrane morphology and/or permeability under normal bodily shear-stress conditions, whereas the same dosages showed a reduced toxicity when applied in a static environment. This result suggests that static nanotoxicity tests may not be representative of the actual toxicity in a human body. If this is indeed the case, future toxicity tests would benefit from reproducing human vascular conditions in their flow and shear-stress properties.
In order to improve the cellular seeding and viability of hepatocytes in a microchannel, the group of Liu et al. [72
] produced an electrospun biocompatible scaffold inside of a microfluidic device. By creating a 3D micro-environment of fibres for the liver cells, they were able to form a micro-perfusion environment which overcame the previous limitations of the lack of scaffold stiffness and the permeability to large molecules/cells. Using this platform, they were able to culture hepatocytes on the scaffold without microfluidic flow, then washed through and assayed for viability by measuring the albumin/urea secretion of the cells. The viability was determined to be higher under microfluidic perfusion than without perfusion flow or in static conditions. Upon the addition of silver nanoparticles, they could measure the amount of cell membrane damage with a commercial lactate dehydrogenase assay kit and found that the biomimetic 3D hepatocyte spheroids were more sensitive to silver nanoparticle damage than on a 2D tissue culture plate.
The use of cytometric methods can be integrated into microfluidic platforms to provide rapid and low cost nanotoxicity data, as was shown by Park et al. [73
]. Their group cultured adherent cells (HeLa) directly into channels in a PDMS-glass microfluidic device and incubated the entire chip for 48 h to allow the cells to spread and grow over the analysis areas. They then introduced silver nanoparticles using a syringe pump via a concentration gradient generator. The silver nanoparticles induced both morphological changes in the cells and a colorimetric response to the MTT assay, which was also investigated in its conventional use in well-plates to compare against the microfluidic response. Optical brightfield images of the cell culture areas inside the channels were acquired post-exposure, to determine cell viability from the morphology and absorbance data. Dose-dependency was clearly observed for the toxicity of the silver nanoparticles and the half-lethal concentration (LC50) of the nanoparticles could be calculated. The LC50 from the microfluidic experiments was comparable to the estimated value from the conventional 96 well-plate method. The benefits of this technique using microfluidic approaches may include lower costs and the ease of use.
When single-cell responses to nanoparticle solutions are studied, using cell traps allows for very specific analyses to be performed on individual cells. Cunha-Matos et al. [67
] formed cell traps designed to accept single cells inside a microfluidic channel using soft lithographic techniques, and then seeded the traps with functionalized gold nanorods, followed by Raman-active molecules and a coating of polyelectrolytes and proteins, that allowed them to bind primary bone marrow dendritic cells. The nanorods were then visible under surface-enhanced Raman spectroscopy (SERS), which allowed for the real-time visualization of nanoparticle concentration gradients as they were applied to the cells under flow. They used a live-cell incubator on a microscope stage to keep the cells in a biologically compatible environment for a duration of 24 h, over which they were able to assess each individual trapped cell for its response to nanoparticles. They were also able to add viability dyes to detect apoptosis and necrosis responses to the nanoparticles.
Adding electrodes to cell-trapping microwells via metallic deposition followed by chemical etching allows for electrophoretic measurements of cell viability. A thesis by Pratikkumar [74
] detailed a microfluidic device that incorporates a combination of dielectrophoresis (DEP) and microwell methods of trapping single cells in wells aligned with gold microelectrodes, which allowed for the analysis of cells using electrochemical methods. The DEP forces could be switched on and off during the cell capture step, allowing the targeted capture of specific cells on the individual electrode/well features. Once the cells were captured, copper oxide nanoparticles were introduced into the microfluidic channel and flowed over the cell membranes to study the morphological response in each microwell. The author states that the impedance-based cell analysis was rapid, simple, label-free, and non-invasive. Measuring impedance vs. time revealed a significant drop in impedance after exposure to CuO nanoparticles. This correlated to a reduction in cell size and detachment from the electrode surface, which indicated a loss of viability due to toxic effects.
Similarly, a device containing impedance electrodes was also produced by the group of Rothbauer et al. [70
], but with larger cell culture chambers instead of microwells. Human lung adenocarcinoma cells were cultured by on-chip perfusion in serum-containing media, until a confluent layer was formed across the electrode surface. Silica nanoparticles were administered under flow in serum-free media (in order to prevent contamination and a change in bioactivity of the nanoparticles). Serum-containing media was perfused again, in order to regenerate the tumour cells from their previous treatment. A metabolic assay was performed in parallel to the electrical impedance assay, and found that the AmSil30 silica nanoparticles caused a reduction in tumour regeneration and re-attachment to the electrode surface. Additionally, the presence of microfluidic flow in the device caused a reduction in regenerative capacity dependant on the flow velocity, indicating that the shear stress exerted on the cells played an important role in increasing the extent of nanoparticle uptake and thus the toxicity.
Organ-on-a-chip devices have only recently been used in nanotoxicity, and most often they utilize a single organ type. As their suitability for multi-organ toxicity assays becomes more fully realized, they will likely become more widely used. The group of Huh et al. [75
] produced a lung-on-a-chip by seeding human alveolar epithelial cells and microvascular endothelial cells onto opposite sides of a porous PDMS membrane coated with an extracellular matrix protein. This membrane was sandwiched in between PDMS layers with large adjacent side channels, which were deformed by the application of a vacuum. This meant that the membrane could be subjected to mechanical stretching to simulate the action of breathing. The epithelial cells were exposed to air after their initial seeding was successful, while the endothelial cells remained exposed to culture medium with added blood-borne immune cells, providing an air-liquid interface to mimic the natural lung environment. In order to determine the device’s response to nanomaterials, a solution of 12 nm silica nanoparticles in fluid was injected over the epithelial layer and aspirated to leave a thin layer, mimicking an aerosol uptake of the solution. The silica nanoparticles were found to promote the inflammation of the underlying endothelial layer, seen by an increase in the expression of Intercellular Adhesion Molecule 1 (ICAM-1) and the increased capture of neutrophils, a type of white blood cell. This device was considered to have increased the efficacy due to the mechanical breathing motions, as the motion only promoted ICAM-1 production upon exposure to the silica nanoparticles. This finding suggests that a lung-on-a-chip device with a breathing motion may give more accurate results on nanotoxicity than a static culture.
Microfluidics nanotoxicity screening offers a range of potential advantages over traditional screening methods. The ability to integrate parallel streams on the same chip allows for high-throughput screening in a small form factor, as well as the reduced use of reagents/analytes and a reduction in the overall testing time. However, microfluidic screening is not yet widely employed in nanotoxicity testing. This may be due to the many parameters that require optimization if an agreed standard operating procedure is to be broadly accepted.
Indeed, the standardization of testing is the most critical roadblock against the adoption of microfluidic nanotoxicity screening. In order to achieve this task, many fundamental studies on the interactions between nanomaterials and channel-bound cells must be performed. Parameters that must be defined include: the effects of channel dimension and flow rate on the amount of nanomaterial exposure; differences in exposure along the length of the channel; and potential run-on effects of affected cells upon downstream cells. The viability assay methods must also be normalized, whereas groups are currently researching multiple variations on dye assays, microscopy, and flow cytometry.
Currently, microfluidics for nanotoxicity screening faces a significant challenge, in that the adoption of a standard device design is perhaps the first necessary step to produce an accepted industry method. However, most research groups involved in the study of this field have their own unique ideas and designs for their devices. It is rare for a research group to precisely follow the designs of another in their own experiments, and therefore, most fundamental studies are performed on different platforms. It would require a concerted effort between multiple groups to agree on a design and perform the required standardization testing to make it suitable for commercialization, which has not yet occurred.
The fact that many different methodologies are still being researched indicates that each method may have its own merits. While droplet microfluidic methods do offer a substantial increase in the number of tests able to be performed in a single device, they tend to be more suited to single-cell or single-cluster analyses. The latter is of interest to fundamental studies of particle or drug uptake in cells that are often found alone or in small clusters. However, the segregation of cellular analysis is less favourable for toxicity screening. Adherent cells, for example, will not adopt the same layer configuration found in normal biology and would therefore give less meaningful data. To address this, researchers are turning to multi-organ labs-on-a-chip platforms. These 3D cellular environments are likely to achieve more biologically relevant discoveries, including any run-on effects of toxic species between cell types. Coupling these 3D cellular environments with microfluidic flow will facilitate the collection of uptake profiles of toxicants under conditions that mimic arterial shear stresses on cell types, i.e., better models of toxicity in the complex environment of the human body.
To summarize, it is likely that microfluidic technologies that are taken-up by the nanotoxicity screening industry will provide data that is extremely difficult to obtain through current well-plate methods. The most likely candidate for this is the organ-on-a-chip style of device, due to the close mimicry of the human body, as discussed above. If these platforms prove to be successful (biologically relevant), fast, accurate, and inexpensive, microfluidics-enabled nanotoxicity screening may become a widely accepted testing platform for industry and regulators alike.