3D Approaches to Culturing Bovine Skin: Explant Culture versus Organotypic Skin Model

In bovine medicine, claw lesions have been identified as the main cause of lameness in dairy cattle that impacts on all key performance indicators of production [1‒3]. Moreover, it is one of the three most important reasons for premature culling [4].

Claw lesions may arise from noninfectious diseases, which are summarized under the term claw horn disruption lesions [5], or from infections, most commonly digital dermatitis (DD). DD has developed into a global animal health- and welfare issue due to the lack of effective preventive and therapeutic measures [6]. Despite substantial efforts to gain insight into its pathogenesis for more than 40 years, many underlying mechanisms of this polymicrobial disease remain insufficiently explored. Furthermore, due to its multifactorial nature, disease onset and outcome are even more difficult to track down in vivo and suitable laboratory model systems are urgently needed [7].

Since the 1950s, human skin models displaying various degrees of complexity have been developed for purposes of both basic science and medical applications. Two-dimensional (2D)-single-cell cultures gave rise to insights into cellular morphology and signaling [8]. Three-dimensional (3D) models promoted functional studies and drug development [9]. Furthermore, explants as well as reconstructed skin models (also termed multilayered skin equivalents or organotypic skin models) were shown to be valuable tools to study the interaction of skin cells and pathogens, may they be of bacterial [10, 11], fungal [12], or viral [13] nature. The latter have greatly advanced as applications in human medicine [14]: novel skin equivalents such as EpiGARD^® (Biovision GmbH, Ilmenau, Germany), INTEGRA^™ (Integra LifeSciences Corporation, Princeton, NJ, USA), and ApliGraf^® (OrganoGenesis Inc., Canton, MA, USA) [15] have been market-approved as therapeutic products and are commercially available.

In veterinary dermatological research, however, availability of such models has remained scarce to this day. In canids, equids, and ovids, 3D-reconstructed epidermis models derived from primary cell culture, stem cells, or progenitor cells, respectively, have been described [16‒20]. Skin explants clearly yielded more interest, as comprehensively reviewed by Souci and Denesvre [21]. Reports exist for pigs, dogs, chickens, sheep, rabbits, guinea pigs, horses, and mice [21‒23]. A recent skin explant infection model on foot rot, a disease of small ruminants related to DD in cattle, has been published [24]. In that context, studies on 2D cultures and bovine fibroblast and keratinocyte cultures being exposed to infectious agents are available [6, 25, 26]. Surprisingly, the only account in literature of bovine claw tissue culture dates back to 1995, focusing on keratin synthesis and deposition [27]. Since the literature on bovine skin models is scarce, our primary goal was to establish new tools for the study of bovine skin. Therefore, this manuscript provides detailed experimental setups for two different methods, their assessment, and a comparison of their advantages and disadvantages in terms of practicability, feasibility, and scientific utility. Both herein described skin models might enable new insights into the study of bovine skin physiology as well as infectious skin diseases such as DD.

For both, i.e., explant and organotypic model, bovine distal limbs were obtained from a local abattoir after routine slaughtering of the animals. Therefore, no ethical approval of a regulatory board was required. Limbs were thoroughly washed with running tap water, and dirt was removed with a soft-bristle toothbrush, and disinfected using 70% ethanol. All skin samples were prepared for either culture within less than 10 h postmortem.

Figure 1 provides an overview of the establishment and setup of both models starting with a schematic drawing of mammalian skin (Fig. 1a). Composition and manufacturers of all solutions and cell culture reagents used in this study are listed in Table 1.

For explant cultures, skin samples were extracted from the area between heel bulbs and dewclaws (Fig. 1b). A representative micromorphological hematoxylin and eosin (H&E) stain of such day 0 skin samples is shown in Figure 1c. In total, n = 20 distal limbs were sampled using 8 mm diameter biopsy punches, and the specimens were transferred to the laboratory in phosphate-buffered saline (PBS). Subcutaneous fat tissue was removed by scalpel, leaving a 2–3-mm-thick skin layer intact. Each prepared explant was placed into one well of a 12-well plate (3.5 cm² growth area) individually. Afterwards, the cavities were filled with the explant culture medium w/AB* to achieve an air-liquid-interface culture (Fig. 1d). The well plates were incubated in normoxic culture at 37°C, 5% CO₂, and 95% relative humidity (RH). Medium was changed every 2–3 days and sampling took place on days 0, 2, 5, and 7 to enable close-knit monitoring of explant progression, while taking into consideration previous results, i.e., that few changes are expected in the early onset of culture. Explants showing signs of bacterial or fungal contamination were discarded.

As future infection experiments with anaerobic Treponema spp. might require hypoxic incubation, a second short-term culture approach (24 h) involved culturing a subset of skin samples in a hypoxic incubator set at 1% O₂. For (immuno)histological sample analysis, day 0 skin and cultured explants were fixed in 4% neutral buffered formaldehyde at 4°C.

For the standardized organotypic models, single-cell cultures of bovine keratinocytes (Fig. 1e) and fibroblasts (Fig. 1f) were generated according to Blanchard et al. [28] with minor modifications to accommodate bovine skin cells. The area between the heel bulbs and dewclaws was cautiously shaved with a disposable razor (in total n = 50 limbs in 22 isolations, Fig. 1b). Skin samples (full thickness skin, approximately 3 cm × 5 cm edge length) were excised using a scalpel and hemostats, transferred to transport medium w/AB and incubated for 1 h at room temperature (RT). Dermal tissue and subcutaneous fat were dissected and discarded; the remaining tissue was cut into cubes of 2 × 2 × 2 mm edge length. After washing with PBS, the tissue cubes were incubated at 4°C in Trypsin-EDTA (T-EDTA) w/AB overnight. The resulting cell suspensions were further single-cell-suspended using cell strainers with a pore diameter of 70 μm, followed by centrifugation at 1,500 rpm for 4 min. The cell pellet was resuspended in 12 mL of keratinocyte culture medium K-SFM w/AB. The cells were seeded into 6-well plates and incubated under normoxic conditions (37°C, 5% CO₂, 95% RH) with media changes at 2- to 3-day intervals.

Primary cultures were differentially trypsinized in order to establish pure cultures of both cell types as previously reported [28]. After 3 min of T-EDTA exposure (0.25%), the detached cells were mainly fibroblasts. Those cells were centrifuged, plated in fibroblast growth medium (Fb-DMEM), and incubated as outlined above. The remaining cells in the first plate were predominantly keratinocytes. They were provided with fresh K-SFM w/AB and also further incubated. This procedure was repeated every 3 days until keratinocyte cultures were free of fibroblasts, which was controlled by morphological examination using an inverted microscope (Olympus IX71, Olympus Europe SE&Co.KG, Hamburg, Germany).

Confluent keratinocyte cultures were subcultured using 0.25% T-EDTA for 3 min and subsequently 0.05% T-EDTA for 10 min at 37°C. Keratinocytes were reseeded in a ratio of 1:2 and used for organotypic models in passages 3 and 4. Fibroblasts were subcultured similarly using 0.25% T-EDTA for 3 min at 37°C, reseeded in ratios up to 1:6, and used for organotypic models from passage 5 on.

To prevent fibroblasts from further division once seeded into collagen pads (see section on the organotypic model), they were mitotically inactivated using Mitomycin C (MMC) prior to seeding. Fibroblasts were grown to 90% confluency in T175 flasks and washed thrice with PBS. Afterward, 10 mL of MMC medium were added for 3–5 h at 37°C. The MMC medium was removed, and the cells were washed three times with PBS, supplemented with Fb-DMEM, and further cultivated for at least 2 days before further manipulation.

The organotypic models were composed of two parts, i.e., the lower dermal equivalent consisting of a collagen pad with interspersed mitotically inactive fibroblasts and the upper epidermal part eventually consisting of stratified, differentiated keratinocytes (Fig. 1g). The collagen pads consisted of eight parts of chilled collagen type I mixed with one part of chilled 10× DMEM. Inactivated fibroblasts were detached as described above, resuspended at a density of 1 × 10⁴ cells/collagen pad in one-part fetal bovine serum, and mixed with the collagen-DMEM solution. 1 mL of this solution was pipetted into transwells (12-well; Greiner Bio-One, Frickenhausen, Germany) and incubated at 37°C for 1 h. After collagen polymerization, 0.5 mL of Fb-DMEM was pipetted into the lower cavity and on top of the collagen pad; they were kept at 37°C, 5% CO₂, and 95% RH. The inactive fibroblasts stretched out and returned to their typical spindle morphology by the next day. Dermal equivalents were prepared the day prior to keratinocyte seeding but could be stored for up to 5 days in the incubator, with medium changes every 2 days.

The next day, the medium on top and below the collagen pads was aspirated and 0.5 mL K-SFM w/o AB was added to the lower cavity. Keratinocytes were detached from the culture flask as described above and resuspended in K-SFM w/o AB at a density of 1 × 10⁶ cells/mL; 0.65 mL of this suspension was pipetted onto the collagen pad. Medium in the lower cavity was changed twice a week. Medium in the upper cavity, i.e., on top of the pad, was aspirated 2 days after keratinocyte seeding in order to create an air-liquid interface culture. In some samples (n = 6), the medium was changed to a modified K-SFM medium containing additional Ca²⁺ with a final concentration of 1.2 mm.

Routinely, at 7, 14, and 21 days, the organotypic models were fixed in 4% paraformaldehyde for 1 h and subsequently embedded in Histogel^™ (HG-4000-012, Thermo Fisher Scientific, Waltham, MA, USA) according to the manufacturer’s instructions. Some organotypic models were incubated further (up to 35 days) and then processed for morphological analysis.

In total, 31 formalin-fixed paraffin-embedded blocks of day 0 bovine skin (n = 9) and cultured skin explants were used for analysis (kinetic experiments: n = 4 at days 2, 5, and 7, respectively; n = 5 after 24 h of normoxic and hypoxic cultivation, respectively). Formalin-fixed paraffin-embedded sections of skin explants were cut into 1–2 μm thin sections using a sliding microtome (SM2010 R, Leica, Wetzlar, Germany) and the sections were mounted on glass slides. Deparaffinization and rehydration according to standard protocols were carried out. Routine H&E staining was performed to assess the microscopic architecture of the skin explants. To verify the cross-reactivity of the antibody panel to bovine tissue, the procedures for optimal staining described in Table 2 were established in day 0 samples.

Chromogenic immunohistochemistry was performed using the avidin-biotin-peroxidase complex method (Vector Laboratories, Newark, CA, USA) to visualize Ki-67⁺ and cleaved Caspase-3⁺ cells in skin explants. The slides were dewaxed for 15 min at 65°C and then rehydrated using xylenes followed by a standard ethanol gradient. Endogenous peroxidase was blocked by applying 0.5% hydrogen peroxide (Merck KGaA, Darmstadt, Germany) dissolved in methanol for 30 min at RT followed by washing in tris-buffered saline. For antigen retrieval, slides were incubated at 96°C for 25 min in citrate buffer at pH 6.0. Blocking of unspecific epitopes was carried out using 5% goat serum for 20 min at RT; the incubation with primary antibodies raised against Ki-67 and cleaved Caspase-3 ensued overnight in a humid chamber at 4°C. The slides were washed two times with tris-buffered saline and were subsequently incubated with biotinylated secondary antibodies (Vector Laboratories) for 30 min at RT, followed by the avidin-biotin-peroxidase complex method using 3-amino-9-ethylcarbazole (Medac Diagnostika, Wedel, Germany) as a chromogen and hematoxylin as counterstain. The stained slides were mounted with an aqueous mounting medium (VWR now Avantor, Darmstadt, Germany). To obtain chromogenic bright field images, a research microscope (Olympus BX 53; Olympus Life Sciences, Japan) equipped with a 5-megapixel digital camera (DP26; Olympus) and the manufacturer’s operating software (cellSens) were used. Photomicrographs were normalized in white balance and detail sharpening was applied with the unmasking filter in Photoshop 2023 (Adobe Inc., San Jose, CA, USA).

Ki-67⁺ and cleaved Caspase-3⁺-labeled slides were scanned as whole slide images (WSIs) with a slide scanner (Axioscan 7; Carl Zeiss Microscopy, Jena, Germany) equipped with a ×40 magnifying objective (Plan-Apochromat ×40/0.65, Carl Zeiss Microscopy). WSIs were analyzed using QuPath Version 0.4.3 [29]. For all WSIs, regions of interest were defined by manual annotation of the epidermis. To quantitate all cells and also immune-positive cells, the positive cell detection mode was chosen by applying the following criteria: optical density sum, pixel size: 0.5 μm; nucleus parameters: background radius: 3 μm, sigma: 2 μm, minimum area: 10 μm, maximum area: 400 μm, cell expansion: 5 μm, singular intensity threshold parameter for mean nuclear diaminobenzidine: 0.2. From these data, Ki-67⁺ and cleaved Caspase-3⁺ density were expressed as a number of positive cells per mm². Immuno-positive cells per mm² were imported in a GraphPad Prism file (Version 10.0, GraphPad Software Inc., Boston, MA, USA) for statistical analysis and graph plotting. Data were not normally distributed due to the small sample size. Therefore, the nonparametric Kruskal-Wallis test with Dunn’s post hoc test for multiple comparisons between groups was performed. The plotted data are expressed as mean with the standard error of the mean. A p value of ≤0.05 was considered statistically significant.

For immunostainings of cultured cells, cells were grown to subconfluence and fixed with 4% buffered paraformaldehyde in 48-well-plates, washed with PBS, and permeabilized using 0.1% Triton X100 (Sigma now Merck KGaA) for 10 min at RT. Blocking ensued using 5% normal serum of the host species of the secondary antibody (goat or donkey normal serum) in PBS for 30 min at RT. Subsequently, the primary and secondary antibodies were applied as specified in Table 2. Omission of the primary antibody in one sample served as control of unspecific staining of the secondary antibody. All immunofluorescence procedures included a nuclear counterstain using Hoechst 33342.

The fixed organotypic models were also embedded in paraffin blocks, sectioned at a thickness of 1–2 μm, and stained according to the procedures described in Table 2. Photomicrographs of the stained cells and organotypic models were taken with a Nikon TE2000S inverted microscope (Nikon Europe B. V., Amsterdam, The Netherlands).

Skin explant culture could be established reliably when bacterial or fungal contamination was excluded. After initial pilot testing, air-liquid interface culture could be maintained successfully in such a way that neither desiccation nor submersion of the explants occurred.

Day 0 skin samples were used to verify antibody cross-reactivity in bovine tissue. The epidermal and dermal lineage markers displayed the following patterns: Keratin (K) 10 labeled all skin layers from suprabasal keratinocytes up to the stratum corneum (Fig. 2a). Basal and suprabasal keratinocytes exhibited abundant intracytoplasmic staining for K14 (Fig. 2b). Involucrin (IVL) showed diffuse cytoplasmic immunolabeling (Fig. 2c), while Loricrin (LOR) showed fine membranous staining. Both markers showed a suprabasal expression (Fig. 2d). Desmoglein-1 (DSG1) expression marking keratinocyte cell-cell junctions was mainly detectable in the upper layers (Fig. 2e). The dermal stroma of the pronounced papillary body was clearly demarcated as shown by the positive Vimentin (VIM) immunostaining (Fig. 2f).

H&E-stained skin explants were investigated by routine light microscopy to evaluate the course of culture. In day 2 samples, no significant differences in morphology were seen compared to day 0. Starting at day 5, two out of four skin explants showed a mild multifocal intracellular edema predominantly at the tissue edges and in the stratum spinosum, with single keratinocytes in a state of ballooning degeneration and small foci of necrosis with cleft formation between the epidermis and the adjacent superficial dermis (Fig. 3a–d). Multiple basal cells were mitotically active at day 0 as confirmed immunohistochemically. A significant decline of Ki-67⁺ cells was seen over time and at day 7 no Ki-67⁺ cells were detected (day 0: 31.56 Ki-67⁺ cells/mm², day 7: 0.00 Ki-67⁺ cells/mm², p = 0.0287). At day 7, all four skin explants displayed a ballooning degeneration of keratinocytes in the stratum spinosum and cells in the stratum basale showed clusters of segmental apoptosis. Reciprocal to the Ki-67 results, the number of cleaved Caspase-3⁺ cells increased toward day 7 of culture (day 0: 15.57 Caspase-3⁺ cells/mm², day 7: 82.48 cleaved Caspase-3⁺ cells/mm², p = 0.0360; Fig. 3e–h).

In both normoxic and hypoxic culture conditions, the mitotic activity of basal cells significantly decreased (normoxic: day 0: 45.68 Ki-67⁺ cells/mm², day 1: 11.84 Ki-67⁺ cells/mm², p = 0.0140), while an increase was noted for Caspase-3⁺ cells (normoxic: day 0: 7.16 Caspase-3⁺ cells/mm², day 1: 29.70 Caspase-3⁺ cells/mm², p = 0.0327), respectively (Fig. 3i, j). A mild perivascular, lymphohystiocytic inflammation of the superficial dermis (Fig. 3b) was seen in all samples of the kinetic experiments, and in 9 out of 15 skin explants in the incubation experiments comparing normoxia and hypoxia.

The cultured keratinocytes and fibroblasts were stable in the routine cell culture after differential trypsinization, displaying typical cell morphology, good adherence, and steady division. The combination of the mechanic removal of dermal tissue and the enzymatic overnight digestion gave rise to a high number of bovine skin cells in the initially heterogeneous culture. K-SFM supported the growth of more than 1.5 × 10⁶ keratinocytes in 18 out of 22 isolations. Since fibroblast growth was also supported by this medium, a separate isolation protocol was not necessary.

Cultured keratinocytes showed a typical cobblestone-like morphology and grew preferentially from small cell clusters to confluency within 1–3 weeks (Fig. 4a). Generally, the presence of fibroblasts in the cell culture vessel accelerated keratinocyte growth and proliferation. Keratinocyte identity and their basal differentiation state were verified by positive K14 stainings (Fig. 4c) while being K10^-. Cell viability and the ability to proliferate were indicated by mitotically active cells expressing Ki-67 (Fig. 4e).

Fibroblasts exhibited a thin, elongated, spindle shape morphology and adhered quickly to cell culture vessels (Fig. 4b). VIM as a marker for mesenchymal cells stained positive in fibroblasts (Fig. 4d) as well as α-smooth muscle actin (α-sma, Fig. 4f). In contrast, cytokeratins were not detected in fibroblasts, hereby additionally confirming their identity. As with the keratinocytes, their proliferative ability was derived from Ki-67⁺ stainings. The MMC treatment decreased the number of proliferating fibroblasts in the cell culture vessel by about one-third (Ki-67⁺ nuclei vs. total amount of nuclei, Fig. 4g, h). The above-described cell morphological and adherence characteristics did not change after MMC treatment.

In organotypic bovine skin models, epidermis and dermis were mimicked by stratified keratinocytes and mitotically inactive fibroblasts embedded in bovine collagen pads, respectively (Fig. 5). The epidermal part consisted of several layers of gradually flattened keratinocytes. However, the characteristic morphological transformation from cuboidal basal keratinocytes to flattened, anucleated corneocytes was found only in some regions of the epidermal compartment. In some models, malformed cell layers on top of the collagen pad were observed (Fig. 5e). Straighter stratification and a more pronounced morphological change were observed in samples cultured in K-SFM w/o AB with an increased Ca²⁺ concentration.

The state of keratinocyte differentiation was investigated using the antibody panel validated in the aforementioned day 0 skin explants. At day 7 of incubation, K14⁺ keratinocytes built three (Fig. 5a) to five layers (Fig. 5b); VIM⁺-MMC-treated fibroblasts in the subepidermal collagen matrix displayed an elongated morphology (representatively highlighted by short white arrows in Fig. 5a,c). By day 14, keratinocytes formed a multilayered epidermal structure (Fig. 5c). By day 21, DSG1 expression was most intense in suprabasal cell layers (Fig. 5d). LOR and IVL showed a prominent suprabasal intracytoplasmic staining (Fig. 5e-f). Few Ki-67⁺ cells were observed (white arrows in Fig. 5g), and the K14-distribution pattern (Fig. 5h) resembled the d14 state. The anti-K10 antibody gave no reproducible staining results and was therefore excluded from the analysis (data not shown). At day 35, multiple epithelial cells showed signs of keratinization (black arrowheads in Fig. 5i).

In the global effort to move basic and applicative research designs away from in vivo animal experiments, advancing 3R principles are implemented more and more. This study describes, evaluates, and compares two different skin models to be utilized in DD research to further dissect its still incompletely understood pathology. Our models represent relevant ex vivo/in vitro substitutes for the use of live cattle in this context due to several reasons: DD had been successfully induced in vivo in cattle [30] and calves [31], but the pathophysiology was not reflected to the full extent in the mentioned studies. Besides ethical issues and highly demanding regulatory efforts, exploring DD in live animals faces further limitations, such as that only gross disease symptoms and the final inflammation progression may be observed due to a limited number of samples taken. We therefore believe that in vitro models represent the method of choice for analyzing disease progression and the impact of each contributing factor making especially use of their reduced complexity and versatility. Additionally, easily acquirable source material is ethically more sustainable, as bovine limbs are usually considered abattoir waste. Both models are homologous allogenic approaches as the used serum was (fetal) bovine serum. Here, an otherwise constant criticism towards serum supplementation (foreign protein intake into, e.g., human cultures) does not pose a problem. Furthermore, without significant limitations in material acquisition, multiwell-plate approaches enable many different in vitro protocols and conditions to be tested in parallel. A combination of deciding external factors known to impact on skin integrity and physiology can be simulated, e.g., anaerobic conditions or maceration.

Bovine skin explants are relatively simple to set up and culture, without the necessity of expensive or sophisticated materials and tools. Although only healthy-looking limbs without any macromorphological alterations were selected, a mild perivascular lymphohistiocytic dermatitis was seen in some of the samples taken. This may reflect the housing conditions of the donor animals (i.e., heightened exposure to manure and moisture around feet), which impact on their skin physiology [32]. In our view and in the given context of DD, this does not necessarily present a disadvantage since it depicts probably ubiquitous subclinical findings in the field. Several shelf-life factors of our skin explant model deserve further attention. In human skin explants, the initial surgical excision resulted in an increased presence of apoptotic and necrotic cells, while the longer cultivation did not significantly affect cell viability and death [33]. In contrast, our observations using established proliferation and apoptosis assays (Ki-67 and cleaved Caspase-3, respectively) revealed similar changes but only during the progression of culture instead of being an initial effect. The decreasing viability found in our study is in agreement with studies conducted in human and ovine explant cultures [24, 33, 34]. Additionally, focal necrosis at the edges of the explants became apparent on day 5 and progressed during further cultivation. We ascribe the latter to the restricted nutrient supply and metabolite removal due to the discontinued physiological dermal microcirculation in the isolated sample. However, as we have previously reported for guinea pig skin explants, cultivation up to 14 days is achievable with the mentioned culture conditions, of course with interindividual differences [22]. Human skin explants serving as a reliable ex vivo/in vitro wound healing model were also described to maintain structure and barrier function for 14 days [35]. Additionally, Neil and colleagues found correlations between metabolic gene expression changes and gross morphological changes also in human skin explants [36]. It becomes clear that from day 7 on, those morphological alterations resemble the ones observed in our bovine explants. For longer study durations, it might be necessary to consider supplementary methods, e.g., a perfusion model, to mitigate the influence of missing circulation [37].

Submersion reduces exposure to the gas mixture in culture and desiccation can completely terminate the sample. Therefore, we decided to use an air-liquid interface that has proven to successfully work in different species and indications [38]. Besides being closer to the physiological state, in organotypic models, this is the only way to induce stratification and differentiation of keratinocytes [39]. Moreover, observed swelling effects in submerged skin explants and the risks of malnutrition and desiccation of the culture were reduced. The explants still need to be monitored on a daily basis for the amount and level of medium.

To the best of our knowledge, there are no data available on animal skin explants cultured under hypoxic conditions. Yet in human medicine, it has been found that the physiological oxygen level in skin ranges from 1 to 8% and is therefore a lot lower than in the atmosphere [40]. From the studies analyzing 2D skin cell culture and reconstructed human skin equivalents, it can be concluded that “physoxic conditions” give rise to a more physiological behavior of skin models [41, 42]. Summarizing our results, this is also seen in our short-term incubation experiments.

A major advantage of organotypic models is their excellent controllability in terms of composition and reproducibility. As with the explants, the multiwell-plate approach enables many different conditions to be tested simultaneously, e.g., different media or anaerobic cultures. Critical verification of cell identity and differentiation state throughout the culture is essential. The H&E stains of our organotypic models clearly showed keratinization and incipient cornification in the upper layers. K14⁺ keratinocytes in suprabasal layers are in agreement with the immunostaining analysis of day 0 skin sections: due to the pronounced papillary body, suprabasal keratinocytes display abundant K14 expression as well. Nevertheless, the differentiation process could and should be further assessed by the analysis of late differentiation markers, such as DSG1, LOR, and IVL. Keratinocytes in the organotypic models did express the aforementioned proteins, which verifies them as reproducible markers in the model. However, due to the fact that these models are cultured at high humidity and are not exposed to external abrasion or shedding, the cornification is altered and the cells retain their nuclei longer than in vivo. Overall, the differentiation of the cells does not resemble physiological stratification as in native epidermis, but it has to be taken into account that the epidermal turnover rate in cattle is probably only just reached (if at all). To our best knowledge, there is no definite information in the literature for cattle regarding the keratinocyte turnover time, but it is probably longer than 21 days (mouse: 8–9.5 days, guinea pig: 13.5 days, human: 40–56 days [21]) and also varies profoundly with regard to body site. Therefore, a longer culture duration might lead to further stratification and more realistic compartmentalization [43].

Fibroblasts naturally produce, among other enzymes, collagenases, which lead to a contraction and reduction in the size of the prepared collagen pads during culture [25]. To keep the number of collagenase-secreting fibroblasts constant, as it is the case with fibrocytes residing in the dermis, proliferation was inhibited by MMC treatment. Thereafter, the inactivated fibroblasts still exhibited the typical elongated cell morphology, intact nuclei, and VIM positivity. Consequently, the severe contraction of the collagen pads we faced during our pilot testings was reduced to an appropriate level. Regarding the complexity of the organotypic model, in the herein described version displaying primary keratinocytes and fibroblasts, the physiological interactions of these 2 cell types may be reproduced up to a certain point. Nevertheless, the microanatomy of skin cannot be fully mimicked relying only on those 2 cell types. Models with higher complexity, i.e., incorporating other cell types [44], and with a higher anatomic fidelity, are subject to future studies. In human medicine, research on 3D skin models has developed much further, as reported with integrated immune cells [45] and neural components [46].

The already complex setup of the organotypic models is time-consuming and material-demanding, which inevitably makes it cost-intensive and creates a conditional disadvantage. However, in cases of multifactorial diseases such as DD, it is of crucial importance to establish reliable models that allow for controlled, realistic conditions – therefore, the organotypic skin models also reach their trade-off.

With both models having strengths and weaknesses, careful considerations on which of their features may be taken advantage of should be made during the initial experimental design phase. If the research focus lies on cell-cell interactions or signaling pathways, utilizing the explant models may be beneficial since they resemble a closer reflection of the in vivo situation. On the other hand, the organotypic culture allows a more standardized setup and a higher reproducibility, which may prove of use when testing compound effects or toxicity. Both models can be understood as complementing each other, and with further optimization on, e.g., culture duration, differentiation state, and cocultivation with infectious agents, their full potential can be maximized.

Hilke Gräfe and Elfi Quente from the Institute of Veterinary Pathology, Leipzig University, are thanked for their valuable technical support. Dörte Döpfer and Megan Kulow at the School of Veterinary Medicine, University of Wisconsin-Madison, are also thanked for technical support and methodological advice. Parts of the organotypic model were established in C.-M.B.’s master’s thesis.

No ethics approval was needed for this study as all animal source material was acquired from local abattoirs.

The authors have no conflicts of interest to declare.

The study was mostly self-funded, and some of the costs for laboratory consumables were paid for by the “Ellenberger-Baum Förderkreis e.V.” (Institute of Veterinary Anatomy, Leipzig University).

Conceptualization: Jule Kristin Michler, Christoph K.W. Mülling, and Vuk Savkovic. Methodology and Validation: Jan Schinköthe, Jule Kristin Michler, Nadia Ayurini Anantama, and Christina-Marie Baumbach. Formal analysis and investigation were performed by Jan Schinköthe, Christina-Marie Baumbach, and Jule Kristin Michler. Writing and original draft preparation: Christina-Marie Baumbach, Nadia Ayurini Anantama, and Jule Kristin Michler. Review: Vuk Savkovic, Jan Schinköthe, and Christoph K.W. Mülling. Editing: Christina-Marie Baumbach, Nadia Ayurini Anantama, and Jule Kristin Michler. Visualization: Jan Schinköthe and Jule Kristin Michler. All authors have read and agreed to the submitted version of the manuscript.

Christina-Marie Baumbach and Nadia Ayurini Anantama contributed equally to this work as first authors.Jan Schinköthe and Jule Kristin Michler contributed equally to this work as last authors.

All data are provided within the article. Further inquiries can be directed to the corresponding author.