Human Epidermal Keratinocytes in Culture: A Story of Multiple Recipes for a Single Cell Type

In 2025, the skin biology community will celebrate 50 years of successful cultures of human keratinocytes. Since the pioneer work published by Rheinwald and Green in 1975 [1] to current methods in use around the globe, 5 decades have seen many ways to modify, improve, refine, sophisticate, simplify the culture procedures dedicated to this epidermal cell type. Thereby, cultures of keratinocytes have open, innovative approach to explore epidermal biology and pathology. We intend herein to present a brief historical path for these methods, in a context of diverse research groups, of growing knowledge and know-how, and of various socio-economical or even legal constraints. To date, data obtained by one or another of these methods are rarely compared to figure out if different experimental conditions can result in perfectly identical, same, similar, or conversely divergent conclusions. Indeed, each research team usually presents its methodology as the one-of-choice. Therefore, keeping in mind where those methods were coming from, for which purpose they were developed and eventually altered, and finally making an appropriate choice for future investigations have become requirements to exert valid studies of the human epidermis by mean of in vitro techniques.

To date, the number of papers listed in PubMed and retrieved with “human keratinocyte culture” has increased from 10 in 1975 to values above 600 each year since 2005 (565 in 2021 and 367 in 2022). Simultaneously, the number of those retrieved annually with “reconstructed human epidermis” has risen from 1 in 1975 to 179 in 2020 (176 in 2021 and 134 in 2022), revealing how important cultures of human keratinocytes have become for skin biology and experimental dermatology.

In the progress of cell culture, fibroblasts were easily cultured since the 60s, as were previously many cancerous cell lines. The crucial role of growth factors released by platelet degranulation into blood-derived serum for instance was then identified. Conversely, certain primary cell types were still difficult to expand enough, like vascular endothelial cells and epidermal keratinocytes. As a postdoc in Howard Green’s laboratory at MIT, James Rheinwald was studying a teratoma cell line in culture, searching to identify how precisely defined environmental culture conditions could determine the phenotype developed by this kind of abnormal cells [2]. During that study, they noticed that an epidermal phenotype could appear in certain cultures and then investigated whether these specific culture conditions favoring epidermal differentiation in teratoma cells would be adequate to grow primary epidermal keratinocytes, a hypothesis that turned out being perfectly correct. The successful procedure to grow human keratinocytes was published [1] side-by-side with the study performed on teratoma [2], but the bibliometric data regarding both articles reveal how much the publication of a technological breakthrough in the culture of human keratinocytes strongly impacted the field of skin cell biology and investigative dermatology. Indeed, whereas the teratoma study was referenced 424 times (according to SCOPUS on July 4, 2023), during the same period the paper on primary cells received a quite larger audience with 3,943 references to date.

Briefly, the 1975 procedure is based on easily available media (DMEM and Ham-F12) but requires lethally irradiated 3T3 fibroblasts as a feeder layer, 10% fetal calf serum, additional EGF, hydrocortisone, and an irreversible stimulation of cyclic-AMP production by the addition in culture medium of cholera toxin (Fig. 1). Overall, the growth conditions described by Rheinwald and Green were found to be very efficient, and soon after publication, they opened thrilling new perspectives for clinicians who expected such materials as a source of expanded autologous epidermal sheets, often critical to heal severely burned patients [3, 4]. During the late seventies and early eighties, the Rheinwald and Green procedure was explored by future outstanding and influential researchers, initially in Howard Green’s laboratory [5‒10], and then quickly spread around the globe. For an overview, it may be interesting to read Howard Green’s own story of this research, published in 1991 in Scientific American [11]. However, if the method is convenient to grow a lot of cells and produce cohesive stratified epidermal sheets, the composition of medium used in this procedure is poorly defined since the compounds found in serum may vary, like the secretory products released by the feeder cell layer. Consequently, there has been a rapidly growing interest in the cell culture community for the development of better defined culture conditions for human epidermal keratinocytes.

Based on a broad experience in the development of culture media, the group of Richard Ham at the University of Colorado in Boulder, CO, further developed several media specifically designed for cell culture without serum, and named MCDB media after their Molecular, Cellular, and Developmental Biology department [12]. Steven T. Boyce developed during his doctoral research with Richard Ham the MCDB-153 medium to grow human keratinocytes in serum-free conditions [13]. One should notice that, even if addition of growth factors like insulin and EGF is perfectly defined in the 1983 published culture conditions, Boyce and Ham’s procedure requires a complex yet not fully defined bovine pituitary extract as an intrinsic component of the growth medium (Fig. 1). One essential characteristic of the Boyce and Ham serum-free culture conditions is the calcium ion concentration (0.3 mm), a concentration quite low in regard of the one (1–2 mm) usually found in most cell culture media and which is close to the concentration found in body fluids. The potential offered by a reduced calcium ion concentration had been initially observed in the group of Stuart H. Yuspa at the NIH, while studying cultures of mouse epidermal keratinocytes. Indeed, they had noticed that calcium ion concentration regulates growth and differentiation of mouse epidermal cells in culture [14]. In fact, a reduced calcium ion concentration prevents the formation of anchoring cell-cell junctions (adherens junctions and desmosomes) because transmembrane adhesion proteins of the cadherin family crucial for these junctions to develop require a minimal 1 mm calcium ion concentration to function and bind cells together. Thereby, a low calcium ion concentration impedes the normal stratification of keratinocytes, usually observed when the epidermal tissue develops or heals. After his thesis in Colorado, Steven T. Boyce was involved with surgeons in San Diego, CA, and then Cincinnati, OH, in developing epidermal grafting, which is crucial to treat burned patients. In this context, he thus continued to develop media dedicated to keratinocyte expansion, being a co-founder of Clonetics, the company that created KGM media based on MCDB-153, and distributed these days by Lonza.

Gary D. Shipley was another researcher who got doctoral training to serum-free cell cultures in Ham’s laboratory [15]. He then moved in the early eighties to the Mayo Clinic in Rochester, MN, working with Bob Scott’s team, including John J. Wille and Mark R. Pittelkow. Together, they quickly and successfully transferred the MCDB-153-based serum-free culture of keratinocytes [16]. This project was prolonged by Mark R. Pittelkow in the department of Dermatology at Mayo to explore keratinocyte cultures as a promising tool for investigative dermatology. In addition to keratinocyte cultures, Pittelkow at Mayo and Shipley who had moved to Oregon Health & Sciences University developed methods to culture another intriguing epidermal cell type, namely, melanocytes for which they published adequate growth conditions in 1989 [17]. Their method was also based on serum-free conditions. When Gary D. Shipley founded Cascade Biologics in 1992 and developed his own first evolution of MCDB-153 medium for keratinocyte culture, he named the new formula Medium 154. Cascade Biologics has been producing and marketing products concurrent to those of Clonetics and to the KSFM medium developed and distributed by GIBCO. In the late nineties, Shipley at Cascade Biologics created the EpiLife medium together with several dedicated supplements (e.g., HKGS, EDGS) with the intention to provide a keratinocyte growth culture medium that can allow significantly higher numbers of culture passages than previously released media when growing primary cells, either from newborn or adult skin. Since Thermo Fisher took over Cascade Biologics, more information on the performances of EpiLife can be read online [18].

Other media were since developed. Following supply shortage during the COVID-19 pandemic, an interesting comparison between several of them was recently published in 2022 by Moran and co-workers in Lisa Beck’s group in Rochester, NY [19].

In the early nineties, there was an enormous interest for the role of growth factors involved in controlling the proliferation of epidermal keratinocytes. Indeed, many research programs were exploring growth conditions that might trigger, support, and accelerate epidermal wound healing, a process so crucial to treat and hopefully save severely burned patients. As a cell biologist, one of us had been involved with the Brussels Military Hospital in this topic [20], being concerned by the need to detach epidermal cells from culture plastic before they could be used as epidermal sheets for grafting humans [21, 22], but also being puzzled by the lack of knowledge regarding the growth factors acting on keratinocytes. Indeed, whereas the question of detachment was crucial in regard to the emerging role attributed to integrins in cell adhesion and proliferation, the role of growth factors to promote keratinocyte proliferation on one hand, while delaying their differentiation on the other hand, was still poorly defined. With so many unanswered questions regarding growth control and epidermal healing, it is quite a pity that a paper published by the group of Shipley and Pittelkow, with Cook (1991) as first author [23], remained barely noticed by most researchers on keratinocytes. In that publication, the authors were demonstrating that keratinocytes growing in serum-free culture and attaining a high enough cell density may become auto-sufficient in terms of releasing amounts of growth factors necessary to trigger their own proliferation up to culture confluence (Fig. 2). This preliminary observation was soon followed by a second 1991 publication [24] in which Cook, Shipley, and colleagues identified amphiregulin as being the diffusible factor secreted by keratinocytes themselves at concentration high enough to bind to and trigger the downstream signaling from the EGF receptor required to keep proliferating keratinocytes in the cell cycle. Intrinsically, those publications were confirming that a closer control of medium components was required to understand all aspects of keratinocyte physiology. They were further proving that cultured cells alter the composition of the culture medium they are growing in. Of value for research on growth factors, those papers were also demonstrating that, in autocrine conditions, the culture medium can be perfectly defined and fully devoid of any added protein (except, of course the proteins secreted by the cultured keratinocytes themselves) [25].

Convinced by the potential offered by autocrine culture conditions to analyze growth factors and signaling working in keratinocytes, one of us joined the team of Mark R. Pittelkow at Mayo, Rochester, MN, as postdoc in 1993. The research goal was then to explore epidermal diffusible factors and their eventual consequences on keratinocytes through binding specific receptors. More precisely, the program was to analyze other receptors of the EGF receptor family in keratinocytes, namely, HER2, HER3, and HER4 encoded, respectively, by c-erbB2, c-erbB3, and c-erbB4 genes. A report on this research has been published many years later, in 2001 [26]. Simultaneous with the analysis of members of the EGF receptor family, that study also observed that keratinocytes are active producers of ligands for these receptors, i.e., neuregulins, also named heregulins or neu differentiation factor. To date, their roles in the epidermis remain unclear [27]. However, because neu differentiation factor was reported exerting specific induction of differentiation toward production of milk components in mammary cancer cells [28], culture conditions adequate to analyze epidermal differentiation in autocrine keratinocyte cultures had to be defined. Therefore, it was found that cell confluence, rather than elevated calcium ion concentration is an event committing keratinocytes toward terminal differentiation. It was further found that EGF, at concentrations usually chosen for keratinocyte cultures (1–100 ng/mL), acts as a strong inhibitor of the early differentiating phenotype in this cell type [29].

Today, elevating the calcium ion concentration in keratinocyte culture medium is frequently presented as the only way to induce differentiation in this cell type. Although elevated calcium ion concentration is surely required for the recruitment of members of the cadherin family to become involved in cell-cell anchoring junctions, and thereby in cell stratification, one must keep in mind that elevated calcium is not necessary to initiate differentiation when keratinocytes reach confluence and simultaneously lose their clonogenic potential. Increasing the calcium ion concentration can even be insufficient if EGF has been added and cell density is still low, far from culture confluence [29].

Fortunately, the role of cell density to favor epidermal differentiation has also been recognized in several recent studies on keratinocytes. However, what may be called “the calcium ion dogma” is still omnipresent and quite difficult to properly challenge in regard of the huge number of published papers that still rely on this principle alone to induce keratinocyte differentiation. In the next point, we will illustrate the critical role played by calcium ion concentration in the process of keratinocyte stratification, illustrating that proper organization of the epidermal tissue is not solely dependent on proliferation and differentiation but also on the crucial function played by anchoring cell-cell junctions.

The culture of human keratinocytes performed in accordance with the original Rheinwald and Green protocol produces stratified epidermal sheets. The produced tissue is however quite simplified and exhibits early expression of involucrin, a late differentiation marker, as soon as in the first suprabasal layer [21]. Conversely, the expression of the early differentiation marker keratin 10, normally expressed as soon as keratinocytes leave the basal cell layer, is much delayed. Thus, epidermal cultures produced this way do not reflect a normal program of epidermal differentiation. However, their abnormal phenotype can be explained for instance by the high level of EGF in the culture medium. Indeed, we surprisingly observed in unpublished results that EGF must be present in culture medium to trigger involucrin expression when keratinocytes are cultured suspended in methylcellulose, a protocol used in many papers intending to characterize the control of epidermal differentiation [8, 30].

In other words, a commitment of keratinocytes toward terminal differentiation certainly happens when cultures stratify. However, the keratinization process still requires exposure of differentiating keratinocytes to the air-liquid interface to initiate the appearance of keratohyalin granules, typical of the granular layer, and then to continue with loss of nuclei and of most organelles in the cornified layer, simultaneously with the formation of the cornified envelope and secretion of lamellar bodies toward intercellular spaces. This critical role of the air-liquid interface to keratinization was originally reported by Pruniéras and collaborators in 1983 [31], in the same supplementary issue of the Journal of Investigative Dermatology that published the first description of serum-free keratinocyte cultures. Of course, although the exposure of cultured keratinocytes to air is necessary when producing completely keratinized in vitro models of epidermis, keratinization finally happens when epidermal sheets are grafted on cutaneous wound, or simply when reepithelialization occurs during cutaneous wound healing from margins. Thus, when cultured sheets of epidermal keratinocytes are used to heal wounded skin, cohesive cultured keratinocytes do not need full keratinization prior to being grafted [4]. Indeed, keratinization and tissue reorganization naturally happen in healed epidermis when grafting is successful [32].

For 3D reconstruction of the epidermis, the use of serum-free cultures of keratinocytes is problematic since the low calcium ion concentration impedes the formation of epidermal sheets made of cohesive keratinocytes like those created thanks to the Rheinwald and Green procedure. As it is explained above, the advantage sought when keeping keratinocytes in medium containing low calcium ion concentration is precisely the impaired stratification and thereby the preservation of basal-type keratinocytes. Indeed, the cell phenotype of basal keratinocytes is the only one able to proliferate, thanks notably to the function of integrins engaged with extracellular matrix components to maintain this phenotype. Wishing to harness serum-free cultures of keratinocytes for grafting severely burned patients, Pittelkow and Scoot published a two-steps procedure in 1986 [33]. In the first step, one can take advantage of the high potential for keratinocyte expansion provided by serum-free culture conditions. In a second step required to create cohesive epidermal sheets useful for grafting, one can increase the calcium ion concentration of confluent cultures up to 1.5 mm, thereby allowing keratinocytes to attach to each other before being harvested with dispase (Fig. 3).

Dispase was found by Howard Green [3] to precisely digest the basal anchorage of keratinocytes without any alteration of desmosomal cell-cell junctions. As an immediate consequence, keratinocytes basally detached by dispase exhibit a strong actin-related cell contraction accompanied by internalization of hemidesmosomes with their components, followed by integrin recycling [21, 22, 34]. If kept detached, epidermal keratinocytes embedded in stratified tissues quickly reorganize, protecting undifferentiated cells by surrounding them with differentiated ones [8, 21, 35, 36].

All those progresses made in culture of epidermal keratinocytes have further opened up the development of 3D skin models, initially fueled by ethically driven, then legal, restrictions in the use of laboratory animals, especially in cutaneous toxicology. By combining cell proliferation with differentiation of cultured human keratinocytes in a stratified tissue exposed to the air-liquid interface, it has become feasible to model the epidermis quite close to its histological appearance in vivo. However, let us immediately notice that, even if tissues reconstructed with cultured cells resemble the normal human epidermis, they are most often made of 1 cell type, keratinocytes exhibiting phenotypes from proliferative to fully keratinized (unless other epidermal cells are eventually inserted, e.g., melanocytes, or neuronal terminations).

Rosdy and Clauss [37] described in 1990 a way to reconstruct the human epidermis without any requirement for dermal extracellular matrix (Fig. 3). The principle developed was to take advantage of permeable membranes to feed from the bottom the epidermal tissue made of cultured keratinocytes seeded at high cell density, while exposing them to air on the upper side. Beside other skin models developed on collagen, like Skin2 or EpiSkin models, Martin Rosdy produced and distributed reconstructed human epidermis (RHE) during the 90s through a company he had created and named SkinEthic.

In the nineties, working with Alain Coquette of SGS-Biopharma company in Wavre, Belgium, we developed assays for in vitro toxicological evaluation of chemicals on skin models. The first model chosen by our colleague was the Skin2 model produced by Advanced Tissue Sciences. However, we were rapidly disappointed in 1996 when owners of Advanced Tissue Sciences interrupted the production and distribution of their model. Nevertheless, turning onto SkinEthic RHE to evaluate chemical toxicity revealed very promising results [38, 39]. Some assays trying to distinguish irritating chemicals from sensitizing ones were found to be potentially predictive in reference to characterized chemicals. Procedures using the SkinEthic RHE model were then validated by other groups [40].

Soon after the beginning of this century, the L’Oréal company, already owning EpiSkin, took over SkinEthic. This suddenly generated strong expectations for autonomous production of the epidermal models. Indeed, relying on a company for tissue production impedes researchers from controlling experimental conditions like the culture environment during tissue reconstruction. The control of SkinEthic by L’Oréal was further placing several cosmetic companies at risk to depend on a concurrent one and on potential interruption of tissue availability for their own usual analyses. For academic laboratory like ours, relying on tissues built by others was further considered too costly. For all these reasons and in regard to our expertise in keratinocyte cultures, we developed and published our own simple procedure to produce RHE with characteristics close to the normal human epidermis (Fig. 4) [41]. Our model was found adequate to analyze the effects of chemicals or drugs on the tissue, or to identify mechanisms involved in tissue responses [41, 42]. Accordingly, the principles of an open-source model of RHE were proposed, together with those of other colleagues, and are still ongoing [40, 43, 44].

With the same purpose to provide autonomy to researchers in their own keratinocyte cultures and RHE production, the CELLnTEC company has been one of the first to distribute media and procedures designed to allow in-house RHE production. Indeed, for instance, in the field of genodermatoses, being autonomous in tissue production renders feasible tissue reconstruction made with keratinocytes isolated from patients holding mutations [45‒47].

Uncontrolled dependency on research materials is too much often frustrating for researchers and may sometimes become a real nightmare. For instance, in relation with geographical and cultural sanitary habits, collecting newborn foreskins is rather easy in Northern America, where circumcision immediately after birth is common. Conversely, getting such material in Europe on a regular basis is almost impossible. Therefore, only plastic surgery can be an easy and regular source of large number of normal human skin samples in Europe, but in most cases, the skin comes from adult donors. Another consequence is interesting to notice. Whereas newborn keratinocytes are collected from male donors, adult cells are most often isolated from female donors (breast reduction, abdominoplasty), creating some gender bias between available newborn and adult human keratinocytes.

Regarding the continuous availability of models, we have reported above how the Skin2 models of reconstructed skin were once available in the 90s but then suddenly became unavailable when their interrupted production was decided as a business-related choice made by its providers. Realistically, the same risk exists for each commercially available model or for any product critical for their production. Indeed, a similar problematic situation might happen if, for instance, some company providing a specific medium used in RHE production decides to interrupt the product.

Maybe worse than interruption of availability can be an alteration in the quality of the materials necessary for tissue cultures. We experienced such a situation in 2018 together with many other laboratories and finally received no explanation at all from involved providers when the situation returned to normal, 1 year later, as if nothing had ever happened. Today, with networking in place between people involved in skin models (see website of NETSKINMODELS COST Action: https://netskinmodels-cost.com), we hope that any problem observed in one laboratory can quickly be shared with others, avoiding as much as possible some waste of time and money.

Still unpredicted was the COVID-19 pandemic and the related sudden scarcity of certain plastic-related products, and some temporary unavailability of media or growth supplements when the whole world business and transportation became down for a quite long period of time. About this question and the requirement to evaluate new products for experiments performed on keratinocytes, Moran published in 2022 an interesting analysis when they compared alternative cell culture media [19]. Recently, we have compared performances of our usual growth supplement containing animal-derived products (HKGS from Thermo Fisher) with several animal-free supplements provided by AvantBio Inc. in trials of epidermal reconstruction using primary, newborn, or adult, or immortalized human keratinocytes (Bajsert J, Cook PW, Poumay Y, unpublished).

Finally, especially with the emergence of DNA edition by CRISPR-Cas9 principles and their use in keratinocytes [48‒50], the utilization of immortalized keratinocytes for tissue reconstruction has received a lot of interest. For instance, it has been clearly demonstrated that cell lines like N/TERT keratinocytes, originally created by Dixon and co-workers [51] in Jim Rheinwald laboratory, can be used to produce RHE as models of epidermis, even in pathological conditions [52]. Conversely, HaCaT keratinocytes [53] are found to be totally unable to keratinize into a decent epidermis, although cultured in similar conditions [54].

Cultures of human keratinocytes have been progressively developed under various circumstances during the last half-century. The different improvements and sophistication of methods to culture epidermal cells have culminated when tissue reconstruction of the epidermis emerged as a flexible tool. Today, the reconstruction of human epidermis has become a generalized procedure in cutaneous toxicology, driven by legal restrictions in the use of laboratory animals. As an additional benefit of refinements in production and certainly thanks to the spreading of know-how, the reconstruction of human epidermis has become commonly used in studies of cutaneous tissue and cell biology. It is nowadays explored to also mimic skin pathologies, or to evaluate treatments when the epidermis is involved in disease [55, 56]. RHE availability further allows studies of interactions between microorganisms and keratinocytes through the reconstructed barrier [57]. Still, models incorporating different cell types have been described and are certainly gaining increasing popularity in the field of investigative dermatology. Similarly, for questions regarding epidermal-dermal interactions [58], models that combine both tissues remain of interest.

To date, while successful methods for keratinocyte cultures have been described and commonly used for 50 years, there is still a lot of space for improvement, sophistication, and better understanding of in vitro epidermal models through powerful analytical tools [59, 60]. The demand and expectations remain high to understand the epidermis and investigate what is going wrong up there when the barrier is challenged and eventually skin disease follows. It means that novelty on one hand should focus on robustness of simple models, most likely through transparent fully open procedures, using globally available defined media and supplements. On the other hand, since simple models like RHE do not contain all skin components, a requirement for enhanced complexity in models remains. Experience from the past suggests that networking should be pursued to keep skin models accessible affordably to the broad community of researchers in dermatology, cosmetology, and toxicology.

The authors have no conflicts of interest to declare.

E.F. is a research fellow funded by the Région Wallonne, MYCEPI Grant No. 1910074.

Y.P. wrote the first draft of the manuscript. E.F. contributed to information regarding 3D cell cultures and edited the concluding remarks.