Past, present and future of in vitro 3D reconstructed inflammatory skin models to study psoriasis

Psoriasis is a common chronic inflammatory skin disease with a significant socio‐economic impact that can greatly affect the patients' quality of life. The prevailing dogma in the aetiology and pathophysiology of this complex disease is that skin cells, immune cells and environmental factors contribute to psoriatic skin inflammation. For a better understanding of the disease pathogenesis, models are required that mimic the disease and which can be used to develop therapeutics. Over the last decades, in vitro human reconstructed skin models have been widely used in dermatological research and have also been developed to mimic psoriatic skin. This viewpoint summarizes the most commonly used in vitro models and the latest accomplishments for the combination of the dermal and epidermal compartments with other cell types and factors that are important players in the psoriatic skin environment. We aim to critically list the most complete and best‐validated models that include major psoriasis hallmarks with regard to gene and protein expression profile and epidermal morphology, but also discuss the shortcoming of the current models. This viewpoint intends to guide the development of in vitro 3D skin models that faithfully mimic all features of psoriatic skin. Such model will enable fundamental biological studies for a better understanding of the aetiology and pathophysiology of psoriasis and aid in novel therapeutic target identification and drug development studies.

For a better understanding of the disease pathogenesis, models are required that mimic the disease and which can be used to develop therapeutics. Over the last decades, in vitro human reconstructed skin models have been widely used in dermatological research and have also been developed to mimic psoriatic skin. This viewpoint summarizes the most commonly used in vitro models and the latest accomplishments for the combination of the dermal and epidermal compartments with other cell types and factors that are important players in the psoriatic skin environment. We aim to critically list the most complete and best-validated models that include major psoriasis hallmarks with regard to gene and protein expression profile and epidermal morphology, but also discuss the shortcoming of the current models. This viewpoint intends to guide the development of in vitro 3D skin models that faithfully mimic all features of psoriatic skin. Such model will enable fundamental biological studies for a better understanding of the aetiology and pathophysiology of psoriasis and aid in novel therapeutic target identification and drug development studies.

| Psoriasis hallmarks
Psoriasis skin has several typical histological features that should be taken into account when developing in vivo or in vitro models to study disease pathogenesis or therapeutics. First, acanthosis (epidermal thickening) is prominent and caused by a rapid keratinocyte turnover. [8] Normally, it takes about 30 days for a keratinocyte to migrate from the basal layer to the skin surface. In psoriasis, this process only takes about 6-8 days and this contributes to the excessive scaling of psoriatic skin. Moreover, premature differentiation and incomplete cornification of psoriatic keratinocytes result in the retention of nuclei in the cornified layer, better known as parakeratosis. The granular layer can be reduced or even be absent in lesional epidermis and the expression of late differentiation genes and proteins is affected by the inflammatory milieu. The pathogenesis of psoriasis is complex and a multitude of studies attempted to decipher the typical gene and protein expression signature of psoriatic skin and to distinguish it from other inflammatory diseases, like atopic dermatitis. Initially, studies appeared showing the contribution of one or two molecules to diseases pathology, whereas during the last years large-scale transcriptomics and proteomics studies have been performed comparing healthy, non-lesional and lesional skin and thereby identify disease markers (differentially expressed genes and proteins). [5,[9][10][11]

| Epidermal disease markers
In general, there are a few very robust differentially expressed genes that are additionally validated at the protein level using specific antibodies and have been replicated in many cohorts. In this paragraph, we will highlight widely used markers that have been used for the development of in vitro 3D psoriatic skin models.
The hyperproliferative epidermis is key for the development of the important disease hallmarks acanthosis, and excessive scaling of the skin. The efficacy of therapeutics for psoriasis patients is therefore also determined by the normalization of the proliferation index leading to a decrease in epidermal thickness. Positive staining with the Ki-67 antibody is widely used as a marker for proliferative cells.
In lesional psoriatic skin, the number of Ki-67-positive cells is dramatically increased indicating the rapid turnover of basal keratinocytes. The Ki-67 antigen is encoded by the MKI67 gene and was first identified in 1983. [12] Nowadays, accurate cell proliferation assays can directly quantitate newly synthesized DNA by following the incorporation of a deoxyribonucleoside analog that contains a detectable tag (eg EdU proliferation kits). [13] Cytokeratin-16 (CK-16) is related to hyperproliferation and abnormal epidermal differentiation and was found to be overexpressed in psoriatic plaques. [14,15] It is expressed in the suprabasal layers, and disappearance of CK-16 protein is related with clearance of disease. [16,17] CK-16 is absent in normal epidermis and the appearance of CK-16 in psoriatic skin is accompanied by the loss of CK-10, which is normally present in suprabasal keratinocytes of healthy skin. The expression of antimicrobial proteins (AMPs) is strongly induced in psoriatic epidermis, and this specific expression can therefore be valuable in distinguishing psoriasis from other inflammatory skin diseases, like atopic dermatitis. Robust markers for psoriatic disease are human beta-defensin 2 (hBD2 or DEFB4), skin-derived antileukoprotease (SKALP/elafin or PI3) and S100A7 (psoriasin). Defensins are small cationic proteins with antibacterial activity that can be subdivided into α-, β-and θ-defensins. hBD2 is absent in normal skin but has been found in psoriatic epidermis, where it is expressed and secreted from the keratinocytes in the granular layer. [18][19][20][21] SKALP/elafin is a protein that is localized in the suprabasal layer of the epidermis up to the stratum corneum. [22,23] S100A7 is one of more than twenty members of the S100 protein family and is barely expressed in healthy keratinocytes. Lesional psoriatic skin shows increased levels of this S100A7, hence its nickname "psoriasin". [24][25][26] It is, however, important to note that these markers may already be expressed in normal 3D skin models as they are generally upregulated in activated epithelia (eg upon wounding) and in early-stage 3D organotypic cultures. Therefore, it is vital to screen for multiple disease markers in the healthy control skin models to determine the best markers for monitoring the effect of disease-associated cytokines or cells in the respective models. Transcriptome analysis of the in vitro psoriasis models and comparison to metatranscriptomic data available from skin from psoriasis patients, and other inflammatory skin diseases could be valuable, yet costly, to extensively characterize and validate the psoriasis-like features of the 3D skin models. The use of not just one marker but a panel of disease-associated genes and proteins is instrumental for the validation of novel psoriasis models, both in vivo and in vitro. vail. [26] Current therapeutic strategies are designed to neutralize these cytokines or interfere in the signalling pathways that they activate. The so-called biologics that target TNFα (etanercept, infliximab, adalimumab), IL12/23 (ustekinumab) and IL-17 (secukinumab, ixikizumab) and small molecules targeting the downstream JAK-STAT pathway (tofacitinib) are effective treatments for moderate to severe psoriasis patients. Still, current research on therapeutics for psoriasis is greatly facilitated by animal models developed to mimic human psoriasis disease. [27][28][29] Although the field has advanced considerably, and the biologics changed the lives of many psoriasis patients, there is still an unmet need for newly developed drugs and identification of drug targets in the treatment of psoriasis. In-depth knowledge on the disease pathogenesis, in part facilitated by in vitro skin models, will facilitate future drug development studies.

| MODEL S TO S TUDY PSORIA S IS
As human in vivo studies have strong limitations due to practical and ethical reasons, experimental models for psoriasis are needed that faithfully mimic the disease. In general, there are two types of experimental models, in vivo animal models and in vitro cell/ tissue culture models. Murine models showing psoriasiform inflammation have been extensively used and they vary from genetic (spontaneous mutation, transgenic or knockout models) and "ligand" induced models, like the widely used imiquimod model.
Multiple reviews have summarized the animals models used to study psoriasis research, [31][32][33][34][35] and to date, many of these models are still used for preclinical studies on newly developed therapies.
However, besides the strong societal urge to reduce the use of experimental animals, there are ample scientific reasons to refrain from animal models in psoriatic research. We have come to learn that the skin, immune system and microbiome of rodents is significantly different from humans. Over the last decade, we and others have strongly invested in the development of in vitro 3D human skin models to faithfully mimic in vivo healthy and diseased skin. Nowadays, in vitro 3D skin models are widely used to study skin biology, disease pathogenesis and therapeutics and should be considered as the gold standard in experimental dermatology research and superior to the conventional monolayer cultures.
The aim of this review is to provide a concise overview on the different types of in vitro psoriatic models present to date, and a viewpoint on the quality of the models and presence or absence of psoriasis-associated cells, phenotypical hallmarks and gene/ protein expression signatures therein. The field should strive to use 3D skin models with a high-quality epidermis that faithfully mimics native human skin. We therefore made a selection of papers that present a histological assessment of the reconstructed epidermis and we evaluated which features of psoriasis skin are present in the respective models. For a complete overview of all in vitro reconstructed skin models that are used to study psoriasis pathophysiology, we highly recommend a very recently published review by Desmet et al [36]

| Conventional in vitro skin models
For the development of in vitro psoriatic skin models, one needs human keratinocytes to generate in vitro skin models. The development of human skin models rapidly evolved after the Rheinwald and Green laboratory described the isolation and clonal expansion of primary human keratinocytes using a mouse fibroblast feeder layer and their studies on keratinocyte differentiation in monolayer cultures. [37,38] To study psoriasis pathogenesis, the addition of inflammatory mediators to the culture medium [39][40][41][42][43][44][45][46][47] and co-culture with immune cells and/or fibroblasts has been reported. [48][49][50][51] Drug screening models have also been developed by the addition of feta calf serum or psoriasis-associated cytokines to keratinocytes and analysed for psoriasis-related gene and protein signatures. [52][53][54] Simplicity, high throughput and reproducibility are major advantages of monolayer keratinocyte cultures. Yet, epidermal stratification is lacking in these models and important morphological features cannot be studied. To study psoriasis pathophysiology and for drug screening purposes, we therefore need models that better resemble the natural architecture and functions of human skin.

| Organotypic psoriasis skin models
Psoriatic 3D skin models can be constructed using several cell types or matrices depending on the purpose of the model and research question. The main building blocks to construct a psoriatic in vitro model are listed in Table 1, and herein, we have cited the studies that have developed and validated the models. In Table 2, the presence or absence of the most prominent psoriasis hallmarks and the gene/ protein expression profile in each of these models are listed. The advantages, disadvantages and the main conclusion of the papers will be discussed in the paragraphs below.

| Patient-derived skin models
Mimicking a disease-specific phenotype in 3D organotypic models is a challenging task. Psoriasis is a multifactorial polygenic disease, and it is difficult to mimic all features of the disease in the in vitro environment. The study by Barker et al [47] was the first to describe an in vitro skin equivalent model for psoriasis. Keratinocytes and fibroblasts from healthy volunteers and psoriatic patients were used to generate full-thickness skin equivalents. Although the morphology of the skin equivalents that time was quite poor, some striking differences were found between the psoriatic and normal skin equivalents. In the skin equivalent generated from psoriasis-derived cells, pro-inflammatory cytokine production and chemokine receptor expression were strongly induced and more proliferative keratinocytes were found. [55] Based on this model, others have also used keratinocytes and fibroblasts from psoriasis patients in combination with cells from healthy volunteers to generate full-thickness skin equivalents. Herein, the presence of lesional keratinocytes and lesional fibroblasts induced a psoriatic phenotype characterized by epidermal thickening, increased involucrin expression and less filaggrin and loricrin expression. [56] The commercially available psoriasis skin model

| Cytokine-mediated psoriasiform models
To overcome the limited availability of patient cells, in vitro psoriasis models based on using keratinocytes from healthy volunteers who underwent plastic surgery have been the mainstay. Our group showed that after the addition of psoriasis-associated cytokines, TNFα, IL-1α, IL-6 and IL-22, healthy keratinocytes behaved like psoriasis keratinocytes at the molecular level. Cytokine stimulation resulted in an increased expression of many psoriatic markers like CK-16, hBD-2 and SKALP/elafin and this induction could be inhibited by anti-inflammatory drugs. [59] Other models including the addition of Th17 cytokines IL-17 and IL-22 revealed induction of morphological psoriasiform signs of epidermal acanthosis and parakeratosis. Moreover, these studies also showed cytokine-induced gene and protein expression patterns that resemble lesional psoriatic plaques, like the induction of S100 family members, hBD2, CK-16 or IL-8. [40,60,61] However, in all models, acanthosis (epidermal thickening) was not due to keratinocyte hyperproliferation but caused by hypertrophy (cell enlargement).
primarily at the level of the keratinocyte, restored the expression and morphology to that of normal skin. Therefore, the N/TERT cell lines are considered a valuable tool to study the biology as well as treatment options of psoriatic skin. [58]

| Co-culture psoriatic models with immune cells
As the interaction between keratinocytes and immune cells is vital in the pathophysiological process and development of psoriasis, researchers attempted to study skin and immune cell interaction in the context of psoriasis using conventional monolayer cell co-cultures. [49][50][51] In these conventional models, there is a direct interaction possible between immune cells and keratinocytes or in case of transwell cultures, the cells are physically separated by a plastic membrane.

| 3D skin models including endothelial cells
In 2014, Ayata et al [66] included endothelial cells within the dermal compartment of their self-assembly psoriasis skin model to generate branch-like capillary structures. Although epidermal psoriasis features were not studied herein and differences with regard to the formation of the capillary structures were minor, "vascularized" models may enable the study on the contribution of pro-angiogenic factors in lesional psoriatic skin and may facilitate anti-angiogenic drug development.

| THE PERFEC T IN VITRO PSORIA S IS S K IN MODEL : ARE WE THERE YE T ?
For more than three decades both academic and industry-related on all in vitro psoriasis skin models present to date, [36] guides researchers in finding the best model present to date for their research project. However, we also must admit that the psoriatic phenotype of the available models still needs improvement to completely replace in vivo experimental models of psoriasis.
We would advocate that efforts should be directed to identify relevant keratinocyte mitogens and to incorporate these in the 3D skin models. This could be achieved by addition of these mitogens (or combinations thereof), or by application of other cells (immunocytes, endothelial cells, fibroblasts) that are a source of these elusive mitogens. Yet, the later may be more difficult to achieve.
One important aspect that requires attention is whether we should continue to strive for complexity by generating models including keratinocytes, fibroblasts, immune cell subsets, nerves, microbiota or vasculature. As every cell type has its own culture requirements, and immune cells for example typically do not function well in a culture medium designed for keratinocytes and vice versa, perhaps should we honour the beauty of simplicity and design a research pipeline by combining simple mono-cultures in a multi-step approach? For such purpose, the skin-on-a-chip technology could provide a standardized, dynamic, and medium to high-throughput model. This technology enables a dynamic cell culture on a microfluidic device that can be constructed of several layers separated by transparent, porous membranes to allow cellular communication and interaction to study inflammation and drug treatment. [67,68] Ultimately, in vitro models for psoriasis will certainly refine and reduce experimental animal models but perhaps may never fully replace them.

ACK N OWLED G EM ENTS
HN and EB are funded by The Netherlands Organization for Health Research and Development. HN is funded by a ZonMw TOP grant 91211052, EB by a VENI grant 91616054 from ZonMw.

CO N FLI C T O F I NTE R E S T S
The authors have declared no conflicting interests.

AUTH O R CO NTR I B UTI O N S
EB initiated and supervised the study. HN and EB wrote and edited the manuscript.