A pilot study to establish an ovalbumin-induced atopic dermatitis minipig model

Skin is a physical barrier preventing water loss and protecting against penetration by external organisms (9). Atopic dermatitis (AD) is an inflammatory skin disorder that presents with pruritus, erythema, swelling, dryness, and fissures. It affects approximately 20% of all people at least once in their lifetime, more commonly in childhood (24, 33). A defect of epidermal integrity due to AD facilitates allergen or hapten entry into the skin and triggers an immune response (20). As AD has complex aetiology involving immune system disorders and/or overactivity, the treatment approach is mainly focused on symptom relief. Recently, a monoclonal antibody against a specific target protein has emerged as another therapeutic option along with the use of conventional drugs (7). Single treatment with lokivetmab, the anti-interleukin 31 (IL-31) antibody is an effective treatment to attenuate pruritus in dogs (30, 32, 39, 41, 42, 43). Precise methods to test the safety and efficacy of new drugs are essential and among them animal models play a significant role in developing novel diagnostic tools and drugs. They are also useful in understanding disease pathogenesis and in the specific case of AD an animal model has both utilities.

The Nc/Nga mouse has been identified as a spontaneous AD animal model (27), which have established by several researchers.

Epicutaneous sensitisation with allergens (20, 40) or haptens (26, 28) has been widely used to establish an AD or allergic contact dermatitis (ACD) model. House dust mites and allergens such as ovalbumin evoke a Th2-dominated response (15, 16). However, haptens such as oxazolone, 2,4-dinitrofluorobenzene (DNFB), or 2,4,6‐trinitrochlorobenzene (TNCB), mediates the Th1 response (28). Another AD animal model can be established through overexpression of cytokine genes e.g. interleukin 4 (IL-4) and IL-31 (6, 8). Considering the cost-effectiveness and the lesser labour demand with animal models exploiting rodents, researchers favour such. However, rodent skin is different to human skin in terms of anatomy and physiology (22, 23), and differences in drug permeability and local immune reactions between the two are unavoidable. Therefore, a more reliable AD animal model similar to human is required.

Porcine skin is composed of an epidermis, dermis, and tightly connected subcutaneous layer. The thickness of the epidermis is 30–140 μm in pigs, while it is 50–120 μm in humans (13, 25). In addition, the thickness ratio of epidermis-dermis is approximately 1:10 to 1:13 in pigs, which is similar to that in humans (29). Moreover, the blood vessel and nerve distribution in the dermis are comparable with those in humans (45). These numerous similarities make the porcine a superior model for studies of skin wound healing (2, 17), burns (1), transdermal drug delivery (3, 11), and ACD (44). Thus, we selected the Yucatan minipig to establish the required AD animal model. In this study, we compared the known DNFB-induced AD with a novel ovalbumin-induced AD minipig model. Gross observation, histopathology, and cytokine analysis indicated that ovalbumin is more reliable AD model.

Animals, husbandry, and feeding. Seven specific pathogen–free Yucatan minipigs, (Sus Scrofa) aged 8–10 months and weighing 15.05–21.17 kg (median weight 18.97 kg) were supplied from Optipharm (Osong, South Korea). They were transported in filter boxes and acclimatised for 7 days in the minipig facility at the Korea Institute of Toxicology. The experimental and control animals were housed individually in a perforated-bottom cage (850mm × 895mm × 845mm) without bedding. Room temperature and humidity were regulated in respective 19–27°C and 30–70% ranges. Fluorescent lighting of 300–700 Lux and air changes 10–20 times/h were maintained. Water was provided ad libitum and feed (PurinaMills, Gray Summit, MO, USA) was provided at the rate of 2% of the body weight per day. All the animal experiments were conducted under the Institutional Animal Care and Use Committee guideline of the Korea Institute of Toxicology (IACUC approval nos 20-1-0071 and 20-1-0158).

Experimental procedure. The experiment was performed from March to June 2020. To establish the AD minipig model, the experimental animals were administered xylazine (0.7 mg/kg, IM; Bayer, Leverkusen, Germany) and ketamine (20 mg/kg, IM; Yuhan Corporation, Seoul, Korea) mixture as an anaesthetic, then the back skin was shaved with clippers, sterilized with 70% isopropyl alcohol, and DNFB or ovalbumin was applied using Tegaderm™ (cat no. 3584; 3M, St. Paul, MN, USA). For DNFB treatment, the minipigs were sensitised with 1 mL of 10% DNFB dissolved in acetone:dimethyl sulphoxide (DMSO):olive oil (5:1:3, v/v/v) for 24 h (one individual) or 30 min (two individuals) on day 1, and next were challenged with 1 mL of 1% DNFB dissolved in acetone:olive oil (8:1.9, v/v) for 2 h on day 15. For ovalbumin treatment, two minipigs were sensitised with 1 mg ovalbumin dissolved in normal saline for 16 days. The remaining two minipigs served as controls. For gross observation, macroscopic images were acquired on days 3, 8, 10, 15, and 17. On day 18, minipigs were euthanised by pentobarbital sodium (100 mg/kg, IV; JW Pharmaceutical, Seoul, South Korea) and skin tissue samples from all four quadrants of the treated area were obtained using a 5 mm biopsy punch for histology and molecular analyses and blood samples (10 mL) were obtained for serum cytokine analysis.

Masson’s trichrome staining. Masson’s trichrome staining was performed following the manufacturer’s protocol (cat no. IFU-2; ScyTek, Logan, UT, USA). Deparaffinised slides were incubated with Weigert’s iron haematoxylin, solutions of Biebrich scarlet-acid fuchsin, phosphomolybdic-phosphotungstic acid, and finally aniline blue. Then, the slides were rinsed with 1% acetic acid solution. The collagen connective tissues were stained blue, nuclei were stained dark red/purple, and cytoplasm was stained red/pink.

Immunohistochemistry (IHC). The skin tissues were fixed in 10% neutral buffered formalin overnight and then embedded in paraffin. Then, the tissue samples were sectioned at 5 μm, deparaffinised, processed for antigen retrieval, blocked, and finally incubated with the primary and peroxidase-conjugated secondary antibodies. For the peroxidase-conjugated secondary antibody, 3,3′-diaminobenzidine (DAB) substrate was used followed by haematoxylin for nuclear counterstaining. Primary antibodies against CD4 (cat. no. MA5-12259; 1:5 dilution; Invitrogen, part of Thermo Fisher Scientific, Waltham, MA, USA), major basic protein (MBP; cat. no. NBP 1-42104; 1:10 dilution; Novus Biologicals, Centennial, CO, USA), and CD11b (cat. no. ab34216; 1:10 dilution; Abcam, Cambridge, UK) were used. The samples were mounted on slides and photographed with an AxioCam microscope camera, and the images were analysed with AxioVision software (both Zeiss, Oberkochen, Germany).

Quantitative reverse transcription-polymerase chain reaction (qRT-PCR). For gene expression analysis, the skin tissues were processed for RNA extraction with an RNeasy Mini Kit (Qiagen, Hilden, Germany) and reverse transcription with iScript RT Supermix for RT-qPCR (Biorad, Hercules, CA, USA). Glyceraldehyde-3-phosphate dehydrogenase was used as an endogenous control for normalisation. The quantitative chain reaction was performed with intron-spanning primers, the sequences of which are listed in Table 1. Fold induction was quantified using the 2^−ΔΔCT method.

Enzyme-linked immunosorbent assay (ELISA). To analyse the cytokines in the skin, samples were homogenised in 200 μL ice-cold phosphate-buffered saline and centrifuged at 13,000 rpm for 10 min at 4°C. Then, the supernatant was obtained. To analyse the cytokines in the blood, whole blood was collected into a conical tube and allowed to clot for 30 min at room temperature. Then, it was centrifuged for 10 min at 3,000 rpm, and the supernatant was collected as serum. Porcine interleukin 4 (IL-4), interferon gamma (IFNγ) and interleukin 13 (IL-13) were measured by ELISAs (cat. nos DY654 and 985; DuoSet ELISA, R&D Systems, Minneapolis, MN, USA for IL-4 and IFNγ; and cat. no. ESIL13 for IL-13; Invitrogen) according to the manufacturer’s protocols.

Statistical analyses. Student’s t-test was used for comparisons of two samples. P values < 0.05 were considered significant. Two biological replicates and three experimental replicates were carried out.

DNFB-induced ACD minipig model. DNFB, a hapten, was used to characterise the ACD in minipigs (44). To determine whether AD had been established under DNFB-treatment conditions, we performed macroscopic and microscopic analyses. The test sites demonstrated severe skin damage on gross observation. Skin erosion and ulceration with mild exudate were observed and remained scabbing was found (Fig. 1A). In histopathological analysis, the epidermis showed diffuse hyperplasia (acanthosis) and intracellular oedema of the epithelial cells (spongiosis). Moreover, perivascular immune cell infiltration was observed at the epidermis–dermis junction (Fig. 1B). To clarify which immune cells were deposited, immunohistochemistry was performed for identification of CD4, MBP, and CD11b. Notably, all three types of immune cells, i.e. CD4+ T lymphocytes, MBP+ eosinophils, and CD11b+ macrophages were observed at the epidermis–dermis junction (Fig. 1C). We observed that 10% DNFB treatment for 24 h disrupted of the skin barrier, damaged cells in the epidermis, and facilitated immune cell infiltration into the dermis. The skin damage was extremely severe, unlike that in human AD. Thus, we modified the experimental protocol and conducted the experiments again.

Fig. 1

After the shorter, 30 min sensitisation, the outer layer of the skin was found to be swollen and red and a crust had formed over it (days 3–10). However, the severity was less than that exhibited by the minipigs which had been sensitised for 24 h (Fig. 1A). From the challenge, scabbing remained constant with no phenotypic difference from day 15 to day 17 (Fig. 2A). In Masson’s trichrome staining, an increase in the epidermis thickness (acanthosis) when compared with that of the untreated skin, spongiosis of the epidermis, and moderate immune cell infiltration in the dermis were observed (Fig. 2B and 2C). In immunohistochemistry, the counts of CD4+ lymphocytes and MBP+ eosinophils were massively increased, while CD11b+ macrophages were not detected (Fig. 2D–2F). These data suggest that 30 min treatment with 10% DNFB inflicts more similar AD to the human variety than 24 h treatment in terms of gross observation and histological findings. However, it may not be a suitable AD model due to skin layer damage. Another ligand was applied to develop an AD minipig model without this problematic aspect.

Fig. 2

Ovalbumin-induced AD minipig model. Ovalbumin-induced AD provoked a dominant Th2 response like human AD (16). It was noted that ovalbumin-treated sites gradually developed redness and hyperkeratosis. Scabs or exudate were not detected (Fig. 3A). However, ovalbumin also mediated acanthosis, spongiosis, and mild immune cell infiltration (Fig. 3B and 3C). Infiltrated immune cells in the skin were identified as CD4+ lymphocytes or eosinophils (Fig. 3D and 3F). These results suggested that ovalbumin induces AD, indicating similarity to human AD without any skin barrier abnormality.

Fig. 3

Analysis of cytokines in the skin and serum of subjects with induced AD. To determine which cytokines were regulated by DNFB or ovalbumin, we analysed the cytokines in the skin and serum samples. We detected an increase in the counts of CD4+ lymphocytes and eosinophils under DNFB or ovalbumin treatment, which suggests the upregulation of the cytokines (Figs 2–3). The cytokine mRNA expression levels of IL-4, IL-13, and IFNγ as well as the absolute levels of IL-4, IL-13, and IFNγ were found to be enhanced in both DNFB- and ovalbumin-induced dermatitic minipig skin (Figs 4A–4C and 5A–5C). Interestingly, we also noted that the cytokine levels had increased in the serum sample, indicating a strong relationship between local inflammation and systemic inflammation (Fig. 5A–5C). Although these cytokines were related to the Th1 or Th2 immune response, we found that DNFB and ovalbumin-induced AD upregulated all three cytokine levels in the skin and serum samples.

Fig. 4

Fig. 5

In this study, we developed an AD minipig model by applying DNFB and ovalbumin ligands with 24 h or 30 min treatment times. As AD is diagnosed based on the symptoms of skin inflammation (such as pruritus, erythema, and hyperkeratosis), the potential for similar gross observations in an animal model as in human AD is essential. The minipig is the only experimental animal that has tight subcutaneous connective tissue and similar thickness of the skin layer to humans. Thus, this species were chosen to establish the AD animal model. Sensitisation only with DNFB had been used to develop an AD minipig model prior to this research (44). DNFB triggers the formation of Langerhans or dendritic cells in the dermis when it induces AD, and these cells migrate to the lymph node and then prime naïve T cells. When the skin is re-exposed to DNFB, allergen-specific CD8+ T cell-mediated skin inflammation occurs (10). However, DNFB in a solvent (acetone:DMSO:olive oil mixture) seems to damage the skin severely. Regardless of the DNFB treatment duration being short or long, DNFB-applied skin was affected by physical disruption of the skin barrier (Figs 1A and 2A). Ovalbumin-induced AD involves different pathological mechanisms to the DNFB-induced condition. Ovalbumin-sensitised skin was infiltrated by more CD4+ T cells and eosinophils, and responded in the dominant-Th2 manner producing more IL-4, IL-5, and IL-13 (26, 27). As ovalbumin mainly mediates the Th2 response, we assumed that ovalbumin treatment may induce typical AD. In fact, the skin did not reveal any epithelial defect, but redness and hyperkeratosis were detected after ovalbumin treatment (Fig 3A). Furthermore, DNFB and ovalbumin increased the counts of CD4+ T lymphocytes and MBP+ eosinophils in the dermis (Figs 2D, 2E, 3D and 3E). Ovalbumin-induced AD in mice showed tissue infiltration by CD4+ T lymphocytes and CD11b+ macrophages (18), while eosinophil infiltration was found in DNFB-induced AD in the same experimental model (19). The difference in the infiltrating immune cells might be related to different species (rodent vs. porcine) or tissue collection time points. Although both DNFB and ovalbumin increase the immune cell infiltration, the skin damage induced by DNFB suggests that the use of ovalbumin is a method equally to be recommended for developing an AD minipig model.

It is known that DNFB or ovalbumin induce different immune responses (Th1 vs. Th2). However, some other studies have reported that T cell polarisation is regulated by the progression of AD. The Th2 response is dominant in acute AD, while the Th1 response is

dominant in the chronic form (5, 34). Moreover, cutaneous ovalbumin sensitisation mediates the combination of Th1, Th2, and Th17 immune responses (35). In fact, we noted that DNFB or ovalbumin upregulate both Th2-related IL-4 and IL-13 and Th1-related IFNγ with statistical significance (Fig. 4A–4C). Moreover, the absolute cytokine levels of IFNγ were upregulated in DNFB- or ovalbumin-treated skin also with statistical significance (Fig. 5C). Thus, it seems that the present AD model induced by DNFB or ovalbumin elicited combined Th2 and Th1 immune responses. Interleukins 4 and 13 upregulated collagen synthesis via the extracellular signal-related kinase pathway in human dermal fibroblasts (4). In addition, the skin barrier protein filaggrin was decreased by IL-4 and IL-13 (14). Interferon gamma regulates the immunological functions of epidermal keratinocytes such as leukocyte migration, immune cell-related surface marker expression, and pro-inflammatory cytokine production (12, 31, 37). Thus, our findings of increased expressions of IL-4, IL-13, and IFNγ are strongly related to epidermal thickness and immune cell infiltration.

There is evidence that AD initiates local skin inflammation, elevates cytokines levels and activates T cell, leading to a systemic inflammatory response (40). Airway inflammation (21) as well as cardiovascular and neuropsychiatric disorders (36, 38) accompanied AD. Fig. 5 demonstrates a similar cytokine expression pattern in the serum to that in the skin, indicating the possibility of systemic inflammation under DNFB or ovalbumin treatment. In the present study, the outcomes of induction of AD in minipig models by DNFB and ovalbumin were compared. Both DNFB and ovalbumin mediated epidermal hyperplasia, epidermal oedema formation, and CD4+ T lymphocyte and eosinophil infiltration into the dermis, and upregulated inflammatory cytokine expression. Interestingly, DNFB induced severe skin damage, while ovalbumin showed a similar macroscopic phenotype to that of human AD. Based on these results, we concluded that the ovalbumin-treated AD minipig is the more reliable and representative animal model. To the best of our knowledge, this is the first study utilising ovalbumin to induce AD in a minipig model. We hope that the ovalbumin-induced AD minipig model becomes a valuable tool for development of drugs to treat AD. Proven as it is by histopathological findings for phenotype and immune cell infiltration and molecular findings for cytokines, the ovalbumin-induced AD minipig model is a sufficiently representative model for this purpose in our contention.