In vitro tests of cell toxicity provide a rough assessment of the ability of cells relevant to a determined application to survive in the presence of specific materials or carriers, the latter being the subject of the present review. In general, two different protocols are observed, one providing a direct contact of the carrier with the cells, and another evaluating the effect of the contact with leachable materials (diffusible components, degradation products, etc.), thus considering an indirect cell contact [57,67].

A very wide number of assay techniques can be used to evaluate cytotoxicity. Table 1 describes those assays reported for the evaluation of chitosan carriers. Whatever the selected test, the generally desired outcome is that cell viability remains close to 100% after contacting with the tested material/system. Depending on which cellular characteristics are focused, the assays described to evaluate chitosan carriers are mainly divided into metabolic assays and membrane integrity assays. While the former assess occurrences of an early stage of cell death, the latter determine the occurrence of membrane disruption that is more frequent during the later stage of the process [68].

The cell membrane is a functional barrier, using several mechanisms to control the traffic of substances into and out of the cellular structure. A disrupted cell membrane implies cell death and, therefore, the assays providing the assessment of cell membrane integrity permit the discernment between live and dead cells. The trypan blue assay has been frequently used for this end, relying on the exclusion of the dye by viable cells, which possess intact cell membranes, while damaged cells include the dye because their membranes are no longer capable of controlling molecular permeation [64,69,70,71]. A viable cell thus exhibits a colorless cytoplasm, whereas a damaged one has a blue cytoplasm that is easily identified by microscopic observation, as is evidenced in Figure 3 [64,69,72]. Notwithstanding the easiness of the procedure, this assay has been referred as imprecise, leading to an overestimation of cell viability [69].

Propidium iodide, a red dye, is also used to indicate membrane disruption. It is membrane-impermeable owing to two positive charges existing in its structure [73], but it enters the cells when membrane damage occurs. It has a strong ability to intercalate DNA and registers a strongly enhanced fluorescence upon binding to nuclei acids, the fluorescence change being proportional to the number of damaged cells [64]. At an excitation of 488 nm, it gives a bright signal easily detected and quantified by flow cytometry. One of the problems with this assay is that the intercalation is reversible, so that the dye might leak out from cells that were dead before fixation and stain the previously viable cells [73].

Alternatively, an assay based on the determination of the amount of lactate dehydrogenase (LDH) that is leaking through a damaged cell membrane has also been used [70]. This enzyme is present in cell cytosol and catalyzes the reaction that transforms NADH + pyruvate into NAD⁺ + lactate. The increase of LDH activity in cell culture medium is proportional to the number of lysed cells. Two different protocols can be followed for the quantification of LDH. On one side, NADH can be directly quantified by spectrophotometry at 340 nm and its amount used to determine LDH level if the reaction starts with known levels of NADH and pyruvate [75]. On the other side, the most common approach uses commercial quantification kits based on a colorimetric assay. In these kits, the formation of NADH can be measured in a reaction where the tetrazolium salt INT is reduced to formazan, which is a colored substance. The amount of formazan is determined directly by spectrophotometry at 490 nm and proportionally correlates with the amount of LDH and, consequently, with the number of damaged cells [64]. The loss of intracellular LDH and its release into the culture medium is an indicator of irreversible cell death due to cell membrane damage [76,77].

The neutral red assay is one of the most frequently used, but it evaluates lysosomal membrane integrity rather than cell membrane integrity [71]. This assay is based on the principle that living cells are able to capture the dye, storing it in the lysosomes [64,67]. The efficiency of neutral red retention depends both on the pH of the lysosome and its membrane proton pump, which maintains the acid environment of the lysosomal compartment. Therefore, lysosomes in unstressed cells are able to retain the dye for longer periods and a destabilization of lysosomal membrane results in leaking of neutral red to the cytosol, decreasing the amount of dye retained in the cells. Neutral red is quantified by spectrophotometry at 540 nm [78].

Metabolic assays are also of widespread application, generally focusing the mitochondrial metabolism. A reduction of cellular metabolic activity is generally accepted as an early indicator of cellular damage. The most common assay, among all those evaluating cytotoxicity, belongs to this category and is the MTT test (uses the compound 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide), which comprises the production of a dye by cells with active mitochondrial activity. This assay specifically evaluates an enzyme function, as the yellow tetrazolium salts are reduced to water-insoluble purple-blue formazan crystals by mitochondrial dehydrogenases. These crystals precipitate in the cytosol and are then solubilized after the induction of cellular lysis by a surfactant, enabling the absorbance to be read at 540 nm [67,76]. Variants of the assay include the MTS (3-(4,5-dimethylthiazol-2-yl)-5-(3-carboxymethoxyphenyl)-2-(4-sulfophenyl)-2H-tetrazolium) and XTT (2,3-bis-(2-methoxy-4-nitro-5-sulfophenyl)-2H-tetrazolium-5-carboxanilide) assay, which present the advantage of avoiding the solubilization step required in the MTT [64]. In all these assays, a dysfunction of mitochondrial activity is used as a sensor of disturbed cell function. A higher concentration of the dye relates to a higher amount of metabolically active cells, which is usually interpreted as higher cell viability. In some cases, this interpretation should not be as linear as it is taken generally, as the contact with a metabolic inhibitor would have the same effect without actually causing cell death.

Alamar blue assay also falls in the category of MTT, using the same oxidation-reduction principle. The blue coloured agent contains resazurin that is reduced to pink resorufin by metabolically active mitochondria, being then quantified by fluorimetry with excitation at 545 nm and emission at 590 nm [71,79]. In both MTT (or variants) and Alamar blue assays, the reduction has been believed to be mediated solely by mitochondrial enzymes. However, some works have been suggesting that other enzymes present in the cytosol and microsomes can have a contribution to the reduction reaction [80,81].

Chitosan carriers have been frequently tested by means of the above mentioned assays, with a predominance of the MTT assay, not only using the original polymer [30,82,83,84], but also chitosan derivatives [85,86]. In some cases, a comparison is performed between chitosan solutions and chitosan carriers. In many works, however, the isolated effect of materials composing the matrix of drug delivery systems or that of drug-loaded carriers is not tested, the evaluation being performed only on empty carriers. In these cases, it is not possible to determine rigorously the systems behavior, because a material may not cause cell injury but kill an animal due to its drug release pattern [57]

When analyzing the available studies reporting the assessment of chitosan carriers’ biocompatibility, one easily observes that a huge variety of conditions are applied. Even if focusing specifically on the works related with drug delivery, a large number of different cells are used, varied material concentrations are applied, considering different contact times. In addition, the intrinsic variations provided by chitosan as a polymer cannot be disregarded (chitosan as a base or salt, different salts, molecular weight, deacetylation degree, etc.), not to mention that very different carriers are also available. To make the subject even more difficult, different assays are used that evaluate distinct aspects of cellular toxicity. This obviously translates into a wide range of responses that are practically impossible to compare, due to unstandardized conditions of assessment.

Nevertheless, taking into account the overall information made available by the literature, it seems possible to say that the general trend indicates that in many cases chitosan solutions exhibit a certain degree of toxicity that is mostly dependent on the dose [60,87]. This confirms a fundamental principle of toxicology, first expressed by Paracelsus, saying that “The dose makes the poison”. Apart from depending on the dose, toxicity was also conditioned by the polymer characteristics, such as molecular weight and degree of deacetylation [60,88], as well as by the pH of the incubation medium and time of incubation [89,90]. In turn, it is also observed that when the polymer is used as matrix material of drug delivery systems, in most cases there is no overt toxicity in concentrations ranging up to approximately 1 mg/mL, although higher concentrations have been occasionally referred to not decrease cell viability as well [28,30,60,84,91,92,93]. However, the effect is still governed by the used dose [60,90,94]. Figure 4 provides examples of very low toxicity of different chitosan carriers determined using metabolic assays.

The different behaviors observed between chitosan molecules and carriers might be explained by the distinct conformation that molecular chains adopt when formulated as carriers, which does not allow the exposure of many of the groups responsible for interaction when in solution. Nevertheless, there are also works reporting a similar cytotoxic pattern for both chitosan solutions and particles [60].

This discrepancy of results is unfortunately observed in the literature and is attributed to the previously referred multitude of assay conditions that are used, not only focusing different structural materials and cells, but also in what concerns the general specificities of the assays. In the cases where chitosan is combined with other polymers to compose the systems’ matrix, the selected secondary polymer naturally affects the overall cytotoxicity of the carrier [95,96].

Many works also report the use of chitosan as a coating material [28,87,97,98]. These approaches can endeavor to achieve different outcomes, from an increased mucoadhesion, an enhancement of the biocompatibility profile by surface modification, or a general improvement in the formulation performance. Whatever the objective, the final biocompatibility of the system will obviously depend on its total composition. In this regard, it has been generally reported that no important toxicity is observed when chitosan is coating a specific core [87,97,99,100] or an improvement of the overall toxicity is obtained due to the presence of the polysaccharide [28,96,98].

One important question that is worth mentioning is that, whatever the selected assay, the samples incubated with the cells should be sterile, to avoid cell contamination and a misjudgment of cytotoxicity. It is further considered that conducting multiple tests, addressing distinct aspects of cellular function, is advantageous to ensure that valid conclusions are drawn, permitting a higher confidence on the observations [64]. An example of this strategy is evidenced in Figure 5, in this case with an observation of some differences in the quantification of cell viability, which are attributed to different sensitivities of the used tests.

Although in some cases a toxic effect can result in cell death, there are many other subtler outcomes that do not cause sufficient harm to induce cell death, instead interfering in the normal functions or generating cellular irritancy or stress. In this context, several assays permit the monitorization of cell function. The assays performed in the ambit of the evaluation of chitosan carriers are described in Table 2.

The adenosine triphosphate (ATP) assay assesses the functional integrity of cells, as cell injury results in a decrease of cytoplasmic ATP. ATP plays an important role in energy exchange in biological systems, serving as the principal immediate donor of energy and being present in all metabolically active cells, as any specialized function will demand the use of energy. ATP level can be determined by a luminescent assay in which an enzyme, luciferase, is used to catalyze the formation of light from ATP and luciferin. The emitted light intensity is determined and linearly related with the ATP concentration [101,102]. This assay is much more recent than others and, therefore, its use is not widespread. Works reporting its application indicate no alterations in ATP intracellular levels, both upon incubation with chitosan solution [103] and chitosan-based (polylactide-co-glycolide/chitosan) carriers [101]. This latter example is depicted in Figure 5.

Chitosan carriers are often proposed in the ambit of systemic mucosal delivery of drugs, which means that the transported drug is expected to permeate an epithelial barrier and enter the blood circulation to be systemically distributed. Drug permeation through the transcellular pathway is dependent on several physicochemical properties of the drugs, namely on their lipophilicity. In turn, mucosal epithelia are generally characterized by the existence of a tight cell barrier, which evidences tight junctions between adjacent cells, so that the free diffusion of molecules by the paracellular route is prevented [104,105,106]. The intensity of the tightness of these junctions can be determined in cell cultures by a parameter called transepithelial electrical resistance (TER), which provides a measurement of the epithelial barrier properties and integrity [107,108,109]. TER has been pointed frequently as a sensitive marker for cellular damage, indicating sub-lethal toxicity [67,101,110] and any alteration in TER suggests a change in the epithelial barrier function [110]. Chitosan is known to have the ability to transiently open epithelial tight junctions, permitting an increase of drug permeation by acting as an absorption promoter. The mechanism underlying chitosan effect relies on the interaction of its protonated amino groups with cell membranes. This interaction produces a reversible structural reorganization of junction proteins, encompassing a specific redistribution of cytoskeletal F-actin and the tight junction protein ZO-1, which results in the junction opening [4,111]. With such an ability, chitosan and chitosan carriers have been frequently reported to improve the mucosal absorption of drugs through roughly all the routes of administration, contributing to increased bioavailability [21,22,87,112,113]. However, by the proper concept of permeation improvement by the mentioned mechanism, chitosan interferes with the normal barrier function of the epithelia. What makes this alteration acceptable is its non-permanent character, as the opening of tight junctions is known to be reversible when the contact with the polymer or the carriers is ceased [4,114,115,116]. An example of this effect is depicted in Figure 6, using chitosan/cyclodextrin nanoparticles.

In the majority of works it is reported that the decrease of TER induced by chitosan is more pronounced for chitosan solutions than for chitosan carriers, which has been attributed to the fact that in the carriers there are less chemical groups exposed that are available for an interaction with cell surfaces, as compared with the polymer chains available in solution [117,118]. However, in some studies it has been observed that nanoparticles produced a more accentuated decrease of TER than solutions [116]. One possible explanation for these different observations is the fact that the carriers/solutions are in contact with the cells in media of different pH. As chitosan pK_a is around 6.5, the polymeric chains in solution or the carriers will form (or not) aggregates in dependence of the final pH of the medium they are incubated in. These aggregates will have different sizes, which promote different patterns of contact with the cells.

If considered that TER can be used to indicate alterations of cell function, as already explained, it is obvious that chitosan carriers evidence that effect. However, the demonstration that the effect is reversible, permitting a complete recovery of the cellular function upon removal of the stressing agent, is also a remarkable outcome. Although these observations cannot be used to say that chitosan carriers are biocompatible, the reversibility of the effect on the barrier integrity makes it being acceptable.

Genotoxicity assays provide an examination of the ability of chitosan carriers to damage cellular DNA upon contact with the cells. The most used test is called the comet assay (Table 2), which assays the DNA damage in individual cells using gel electrophoresis [64]. The cells are exposed to an electric field to attract the broken and negatively charged DNA to the anode. After that separation, a fluorescent dye is used to stain DNA, like propidium iodide. Afterwards, the gel is read and cells appear distributed as comets, with intact DNA residing in the head and broken DNA migrating to form the tail. The extent of the tail gives an indication on the number of DNA strand breaks [64,119].

Genotoxic evaluation of chitosan carriers is not frequent and a very scarce number of works are available. Chitosan oligosaccharides and a low molecular weight chitosan indicated an absence of genotoxicity in lymphocytes [120]. In a work with chitosan-coated silver nanoparticles proposed as an alternative to conventional antibiotics, the comet assay revealed that a concentration of nanoparticles of 3 ppm does not elicit a genotoxic effect on mouse macrophages (RAW264.7 cells). In contrast, 20 ppm of the nanoparticles induce the formation of a comet with a considerable tail [82].

The development of an inflammatory response as a result of cell contact with carriers has also been used as an indicator of biocompatibility. Testing this hypothesis encompasses the quantification of the release of various inflammatory markers, namely the cytokines IL-6, IL-8 and the tumor necrosis factor (TNF)-α. While IL-8 is a chemotatic agent for inflammatory cells, IL-6 is responsible for neutrophil activation and TNF-α is an acute inflammatory response cytokine [121]. A limited number of works dealing with chitosan carriers for drug delivery applications refer to this evaluation. PLGA-CS nanoparticles demonstrated to not induce an increase in any of these inflammation markers upon incubation with Calu-3 cells in nanoparticle concentrations up to 0.2 mg/mL [96]. Chitosan microspheres encapsulating PLGA nanoparticles generally evidenced no ability to increase TNF-α secretion by RAW264.7 cells (macrophages) in concentrations up to 1.6 mg/mL [28]. The same cell line evidenced an absence of effect on the levels of IL-6, IL-8 and TNF-α mediated by the presence of chitosan-DNA nanoparticles up to 24 h, even at the higher nanoparticle concentration tested (60 µg/mL) [122].

As described for other tests, the highest limitation in performing a comparison of results lies in the variety of conditions, in this particular case most related in the wide range of concentrations tested. Interestingly, from concentrations as low as 0.06 mg/mL to much higher values (1.6 mg/mL) the observation remained very similar and reported no ability of the tested carriers to induce inflammatory responses.

In vivo studies are essential for the final establishment of biocompatibility. Even when assuming that in vitro cell studies reflect the in vivo observations considering the same cells, it cannot be forgotten that in cell cultures only a few of the in vivo variables are accounted for, thereby not permitting a real simulation of the in vivo environment. In addition, cell cultures are very sensitive to environmental changes, such as temperature, pH and nutrient concentrations, demanding a confirmation of observations in an in vivo setting [64]. As critically pointed out in a recent review addressing the biocompatibility of drug delivery systems, a carrier may not cause any tissue injury at all, but kill an animal either from drug release or from problems related with intravascular coagulation, embolic events, chelation of ions vital to homeostasis, etc. [57]. This actually means that, although not directly cytotoxic, the carrier (as any material) can induce a destructive reaction thereafter [57], which makes in vivo assays mandatory and determinant after a certain moment and, definitely, well before any attempt of claiming for a human application of the developed carrier. This ambitious step needs to be accounted for regarding both the carrier effectiveness and safety.

Chitosan biodegradability represents a step forward concerning safety requirements, as it prevents its accumulation in the organism. As mentioned in Section 2, chitosan of suitable molecular weight is said to be directly cleared by the kidney, while that of excessive molecular weight is first degraded into fragments that undergo renal clearance afterwards [11,21]. As previously observed in some in vitro tests, chitosan carriers present a decreased toxicity profile as compared with the corresponding polymeric solutions, because the interaction pattern with cellular structures is different between free polymeric chains and the carriers. The same observation applies, in some cases, to the comparison between loaded and unloaded chitosan carriers. In fact, the loading of a carrier with a drug might cause surface alterations on the carrier, for instance at the level of its surface charge. Some works report that drug loading produces a charge decrease, thereby modifying the interactions with cells and the microenvironment, often leading to decreased toxicity [16]. However, the previous observation that drug release by itself can cause acute toxicity, depending on the release pattern, cannot be disregarded.

There are mainly two different in vivo studies that have been used to assess the biocompatibility of chitosan carriers. By one side it is important to observe histopathological effects resulting from the contact with the carrier. In turn, it is also relevant to determine the inflammatory response induced by the carrier, which is monitored by the determination of several pro-inflammatory cytokines. Additionally, the appearance of relevant alterations in normal clinical signs (diarrhea, fever or other systemic symptoms) is frequently monitored. In what concerns the evaluation of chitosan carriers, there is a high incidence in the assessment of both the oral and the pulmonary routes, but others have also been approached.

The LD₅₀ of paclitaxel-loaded chitosan micelles administered intravenously to mice was 72.16 mg/kg. Moreover, intravenous administration to rabbits did not induce histophatological effects at a dose of 6 mg/kg/day during 3 days [123].

For the administration of carriers through the lung route, several different observations were performed using different particles. Chitosan-graft-spermine/pDNA nanoparticles administered to mice using a nose-only device revealed an absence of detectable damage. The histopathological evaluation of the lungs evidenced absence of necrosis, metaplasia, anaplasia in pneumocytes, atelectasis or emphysema. Capillary vessels within the alveolar wall were not enlarged and damaged endothelial cells were rarely observed. Neither congestion nor hemorrhage was noticeable, along with an absence of infiltration of inflammatory cells. In addition, abnormal features were not detected in other major organs (brain, heart, lung, liver, kidney and spleen) [124]. In contrast, the intratracheal administration of glycol chitosan/cholanic acid nanoparticles to mice (2 mg/kg) induced mild inflammation. Transient neutrophilic pulmonary inflammation was observed from 6 h to 3 days after administration and the lung expression of pro-inflammatory cytokines (IL-1β, IL-6, and TNF-α), as well as that of the chemokine MIP-1α, increased during the first 24 h, recovering to normal levels thereafter [125]. A more pronounced inflammatory effect was detected upon intratracheal administration to rats of unloaded chitosan microparticles (2–10 mg/kg of particles). A dose-dependent pro-inflammatory effect was manifested by increased levels of bronchoalveolar lavage fluid protein, lactate dehydrogenease activity and increases in lung tissue myeloperoxidase activity and leukocyte migration. A cytological examination of bronchoalveolar fluid further evidenced a large infiltration of neutrophils, which are inflammatory cells [126]. Unfortunately, the authors did not include any group of animals examined sometime after the inhalations, to verify whether a recovery from the inflammation occurred.

It is also noticeable that in the latter work the used dose varied from 2 to 10 mg/kg, while the previous one only assessed a dose of 2 mg/kg. Obviously the different composition of the particles in both works has a great impact on the observed response, apart from the fact that the former work assesses nanoparticles, while the latter tests microparticles. However, it is important to mention that chitosan microparticles administered at the dose of 2 mg/kg induced a response similar to that of the control in practically all the assayed parameters [126], thus being considered to not induce an inflammatory response. When comparing both systems assessed at the same concentration (2 mg/kg), a similar absence of inflammatory effect was determined.

For the oral administration varied responses were also reported. An evaluation of the hepatotoxicity of 3 doses (125 mg/kg each dose) of unloaded alginate/chitosan microparticles administered to guinea pigs spaced by 10 days, revealed no toxic effect by measurement of the indicators of liver function serum bilirubin, alanine aminotransferase and alkaline phosphatase [127]. Unloaded chitosan/poly-γ-glutamic acid nanoparticles were reported to not induce significant toxicity upon oral administration of a daily dose of 100 mg/kg for 14 days. As compared to untreated animals, there was no evidence of different clinical signs. In addition, no pathological changes in liver, kidney and intestinal segments were observed [128]. The ability of these nanoparticles to induce hepatotoxicity was also verified by monitoring alanine aminotransferase and aspartate aminotransferase. No alteration in the parameters was obtained after the administration, nor was a histological effect observed [129]. Trimethyl chitosan oligomer/DNA nanoparticles administered orally twice a week during 4 weeks, caused slight diarrhea at 3 weeks in animals treated with 0.5 mg/mL of nanoparticles (containing 0.1 mg/mL DNA), which was relieved upon stopping the administration [130]. This was considered very mild toxicity.

A very small increase of serum cytokine levels (IL-2, IL-6, IL-12, TNF-α) was observed upon administration of chitosan and polyethylenelglycol-coated PLGA nanoparticles (20 mg/mL) to mice, which completely recovered to normality at 6 h [131]. Chitosan reduced gold nanoparticles demonstrated a very high LD₅₀ (greater than 2000 mg/kg) after administration to rats for a 28 day period. No significant changes were detected in clinical signs, body weight, food consumption, hematological parameters, organ weights or histopathological aspects [132].

Although occasionally some augmented inflammatory response has been reported, the general trend seems to be that no important toxicity or only minimal toxic effects are generated by the administration of chitosan-based carriers. This might be a significant indicator of chitosan safety, but ascertaining the specific conditions to be established in each administration modality and, somehow, provide a standardization of the procedures, remains the most important issue.