The human eye is a deeply specialized organ with a singular anatomy and physiology, comprehending several structures with specific physiological functions. Due to the complexity of the eye, ocular tissues are finely organized and orchestrated. As a result, optimal visual function is maintained while the passage of solutes, fluids, and also drugs is highly controlled . Briefly, the human eye is characterized by three main layers, which enclose many anatomical structures. The outermost layer is the fibrous tunic, composed of the cornea and sclera. The cornea and opaque sclera, its non-transparent extension, are inelastic structures that provide mechanical support to the eye globe, also protecting the eye from the external environment . Moreover, the cornea is covered by the tear film, whose composition ensures hydration, provides nutrients, and further limits the entering of toxins or particles into the eye . The middle layer (uvea or vascular tunic) includes the iris, pigmented epithelium, choroid, and ciliary body . Finally, the innermost layer of the eye is represented by the retina, which is a neurosensory structure fundamental for the vision process . According to its crucial role in regulating the vision process, many pathological conditions affecting the retina may progressively lead to an altered vision or blindness .
Based on these observations, along with the necessity to reduce tests on animals for evaluating the pharmacological profile of possible ocular drug candidates for given ophthalmic disorders (drug delivery/drug efficacy), including possible toxicity issues, the development of suitable and robust in vitro ocular models is a challenging task. These models allow to investigate the different aspects of the ocular pathophysiology of different diseases as well as the potential efficacy of possible therapeutic agents . Furthermore, the use of these in vitro tools can be relevant for studying cell surface biomarkers for drug delivery. In the last years, along the ocular in vitro models, isolated primary cultures are expected to reproduce in vivo cellular functions and morphology in a more accurate way; however, these kinds of cells are difficult to cultivate since they arrest their growth quickly. Moreover, considering the human primary cells, it is very problematic to obtain numerous isolates for the restricted availability of human donor eyes. In order to overcome this issue, several attempts aimed at exploiting immortalized cell lines have been described to be used for pharmacological and biological investigations . Unfortunately, the immortalized cell lines are characterized by altered gene expression patterns that often do not reflect the comportment of ocular cells in vivo, partially lacking the ability to mimic the complexity of the physiology of the human eye. However, the development of improved ocular cell-based models established also by reconstructing ocular tissues is fundamental for speeding up the discovery of safe ocular drugs with a relevant pharmacological profile. In this review, we report the most advance in vitro ocular models along with the computational approaches in the field of ophthalmic research. In fact, in the next sections, actual in vitro ocular models are discussed in detail, considering conventional two-dimensional (2D) models and advanced corneal three-dimensional (3D) models, with a particular focus on the application of the human cornea-like epithelium system and the potential models resembling human corneal diseases such as the zebrafish ocular surface. In addition, possible pharmacological application of 3D reconstructed human corneal tissues are reported as well as the most advanced in silico approaches in the field of ocular pharmacology and toxicology.
2. In Vitro Ocular Models
2.1. Opportunity and Application
Due to the complexity of the eye anatomy, a crucial issue in the development and realization of ophthalmic products and medical devices is to identify the specific mechanism of toxicity that could lead to severe adverse effects 
. For this reason, recognizing and classifying the potential risk of commercial products is highly recommended to clearly know possible side effects. Eye toxicity testing is therefore necessary and mandatory to ensure that risks associated with the use of specific ophthalmic products follow appropriate safety criteria. In vitro preclinical testing is nowadays a well-established and important experimental approach for evaluating the efficacy and safety of cosmetic, pharmaceutical, and nutraceutical products () 
. The realization and development of increasingly sophisticated experimental models, especially those based on reliable 3D cell cultures, can reduce the costs of experimental procedures, obtaining predictive information on the ocular tolerability and efficacy of a given product, severely limiting the in vivo experimentation on animals 
Figure 1. Relevant disease areas and the main model systems actually used for evaluating efficacy and/or safety of drugs and nutraceutical products.
The development of novel in vitro approaches is firstly linked to the campaigns carried out in this decade by few associations, which strongly ask for a significant reduction in animal testing 
, leading to finding alternative methods and solutions to animal testing in the cosmetic, pharmaceutical, and nutraceutical fields. In particular, the principle of the 3Rs (Refine, Reduce, Replace) has been considered a stimulating opportunity to improve in vitro methods, even if nowadays it is not possible to completely abolish animal experimentation 
. The scientific world gave the introduction of experimental in vitro models a strong impulse, implementing European Union (EU)-validated alternative methods in compliance with Good Laboratory Practice (GLP). In addition, the new Medical Devices Regulation (MDR 745/2017) is a further opportunity for companies operating in the preclinical sector 
stimulating the exponential evolution of in vitro technologies. In particular, the potential of 3D systems (3D human tissues reconstructed in vitro) often proved to be more relevant and predictive than monolayer cell models. Many 3D models, at first, were quickly developed under the regulatory push, in order to replace animal models. They have been included in numerous OECD (Organisation for Economic Co-operation and Development) validation studies. In a short time, they became increasingly predictive with respect to the evaluation of a possible drug candidates, and were rapidly adopted in preclinical research. The advantages of using experimental in vitro 3D cell cultures models is due to their complex organization and structure, which is very similar to in vivo tissue 
, showing reproducible results in pharmacological and toxicological responses to reference substances.
The sections below aim at highlighting the substantial differences between conventional 2D models and advanced corneal 3D models, describing specific test guidelines already adopted for the evaluation of important toxicological and pharmacological responses.
2.2. Conventional 2D Models
The investigation of basic developmental or differentiation processes can be studied using primary or immortalized human cells deriving from the cornea, retina, and conjunctiva to understand and clarify pathophysiological conditions or to set up models in order to reproduce specific disease models and to perform toxicological and pharmacological studies 
. Epithelial cells, keratocytes, fibroblasts, and trabecular meshwork cells are critical components required for the normal function of the ocular cell system. Atypical cell proliferation and regulation within the ocular cell system contributes to the development of disorders such as corneal inflammation, proliferative retinopathy, macular degeneration, glaucoma, and retinoblastoma. Cell culture models allow to evaluate the physiology of the different ocular cell types outside the living organism in reproduced conditions that mimic, as closely as possible, the environment of the tissue or organ from which they derive 
. Among the possible applications, we can mention: (a) the investigation of the physiological processes of the cell life and of the response to exogenous treatments in a controlled environment 
; (b) the evaluation of the effect of various molecules and drugs on specific cell types; and (c) the study aimed at generating reconstructed tissues (e.g., artificial corneal tissues). In the living organisms, cells are kept vital thanks to the supply of nutrients, supported by the vascular system which, through the capillary vessels, nourishes the tissue and abolishes those harmful molecules deriving from the cell metabolism. In vitro, the role of the vascular system is substituted by the culture medium, a highly nutritious liquid medium. It contains fundamental substances, such as glucose, amino acids, vitamins, and minerals, absolutely necessary for the physiological processes of cells, and animal serum, which supports cell growth and proliferation. Thanks to these culture conditions, ocular cell-culture models offer several advantages over animal experimental models, including a higher reproducibility, easier handling, and reduced costs, but still giving the possibility to study mechanistic processes of physiological or pathological altered pathways. Corneal cells can be directly exposed to test samples (chemicals or environmental matrix samples) at low and relatively defined concentrations 
. In this regard, although distribution and excretion phenomena (which occur in in vivo exposure) do not occur, the bioavailable concentration of the test sample must be taken into account even in the in vitro models. The interaction of the sample with the cells allows a very rapid evaluation (even by hours) of the effect on cell activities and also allows to verify the reversibility of the response. Animal cell cultures can be used as a low-cost, rapid screening tool for toxicological and pharmacological evaluation of chemicals. Moreover, the problems deriving from interspecies variability are avoided if cells of human origin are used 
. However, primary cells usually can only be used for limited passages before starting to lose their normal physiology and structural characteristics. Immortalized cell lines can be used for several passages, but they show the likelihood of developing chromosomal abnormalities, reduced expression of key markers, or abnormal growth 
. However, there are also some limitations in the use of cell cultures. The in vivo–in vitro translation causes the loss of specific cell–cell interactions, histological characteristics of the tissue of origin and the components involved in homeostatic regulation (especially those of the nervous and endocrine systems). There are also metabolic alterations with a drop in some enzymatic levels (e.g., cytochrome P450) or changes in metabolic cycles, so that the energy metabolism of cells is largely based on glycolysis. Due to the strong selection in favor of the most actively proliferating cells, the culture also suffers a loss of differentiated properties.
2.3. Advanced Corneal 3D Models
To date, regarding ocular studies, there are several in vitro methods that have been developed. Some of them suffer from diverse drawbacks, as in the case of organotypic and cell-based testing methods, that are scarcely compatible with real human eyes. Moreover, differences between species caused by the use of animals’ eyes may lead to an excessive and insufficient prediction of the eye irritation. The monolayer cell cultures employed in in vitro testing do not realistically reproduce the complicated 3D environment of real ocular tissues. Artificially rigid and flat surfaces of culture plates may alter cell metabolism and intrinsic functionality. To overcome these inaccuracies, 3D models equivalent to the human cornea have been developed based on normal human cells which are grown on an inert polycarbonate insert ().
Figure 2. Schematic representation of a 3D in vitro corneal model: (A) Corneal endothelial cells grow on a permeable support up to confluence. (B) A 3D matrix containing stromal cells grows on top of the endothelial layer. (C) Epithelial cells are seeded on the stromal layer; then, exposure to air–liquid interface results in a stratified epithelium.
These tissues are validated and standardized, and each batch is derived from a single donor, giving a huge advantage in terms of accuracy and reproducibility. The human cornea is formed by the epithelium, stroma, and endothelium. Although ideal 3D models equivalent to human cornea should have all the three components of the cornea, only the human cornea-like epithelium (RhCE) has been presently developed, due to technical limitations. However, the corneal epithelium is the most important part to assess eye irritation, because it represents the outermost layer of the cornea, which protects the underlying tissue by excluding foreign material. There are a large variety of corneal models used to predict eye irritation including EpiOcular™, SkinEthic HCE, the Labcyte Cornea model, and MCTT HCE™. In particular, with regard to the reconstructed corneal tissue, the cells form a stratified and well-organized epithelium that is structurally, morphologically, and functionally similar to the human cornea presenting basal, wing, and mucosal cells 
. These models are used to study drug delivery, as they represent a metabolically active tissue with the presence of tight junctions, characteristics of the human corneal epithelium. In addition, it has been shown that this type of tissue can be stimulated for the release of cytokines characteristic of an inflammatory state.
2.4. Pharmacological Application of 3D Reconstructed Human Corneal Tissues: The Dry Eye Model
Dry eye syndrome is caused by chronic dehydration of the conjunctiva and cornea, which induces irritation . It is mainly due to a quantitative reduction or qualitative alteration of the tear film, which physiologically covers, lubricates, and protects corneal tissue . Poor production or excessive evaporation of tears can be a complication of blepharitis, conjunctivitis (including allergic forms), and other inflammatory eye diseases . Dry eye syndrome can also result from systemic diseases such as systemic lupus erythematosus and rheumatoid arthritis. Moreover, the disorder is typical in elderly patients (for the atrophy of the tear glands), in menopausal women (for the new hormonal balance), and in those who wear contact lenses . Dry eye can be also related to iatrogenic causes for the use of several systemic drugs (antihypertensives, anxiolytics, sleeping pills, antihistamines). The most common symptoms due to dry eye syndrome are itching, burning, irritation, and annoyance to light (photophobia). In addition, a sensation of a foreign body pulling and scratching inside the eye, blurring of vision, difficulty opening the eyelid when waking up, eye pain, and hyperemia (red eyes) may also occur . Tiredness or fatigue of the eyes may also appear and, in some patients, the appearance of mucus inside or around the eye is observed. All these disorders increase as a result of prolonged visual strain or under particular environmental conditions, such as exposure to wind or heat or staying in dusty, smoky, air-conditioned, or heated environments. In the most serious cases, the eye is exposed to increased friction due to eyelid movement and an increased risk of infections. In addition, it can degenerate to the appearance of lesions to the external structures of the eye: scarring, neovascularization, infections, and ulceration. Treatment for dry eye syndrome includes therapies that may vary depending on the cause and type of the disorder. Generally, medication is prescribed with eye drops or lubricating gels to help the eye to stay moist and clean . When the patient’s eye allows it, it is also possible to prescribe protective contact lenses to protect the organ from rubbing with the eyelid. Developing novel medical devices for the treatment of dry eye syndrome is nowadays a challenging issue and 3D human corneal tissues have been recently used for the setup of in vitro dry eye model . In particular, several research papers describe the realization of in vitro dry eye condition by exposing the tissues to specific conditions. Reconstructed corneal tissues are first treated with 0.6 M sorbitol in order to create a hyperosmolar environment mimicking the qualitative alteration of tear film. Furthermore, the tissues are exposed to 40 °C and 40% humidity for simulating the dryness . After 24 h, tissues can be treated with the tested medical devices, and several biomarkers can be investigated. In particular, besides the tissue viability, also pro-inflammatory biomarkers (e.g., TNF, interleukins) and specific metalloproteinases which are responsible for corneal remodeling processes can be measured in order to characterize the efficacy of medical devices or drugs .