Thyroxine (T4) may promote re-epithelialisation and angiogenesis in wounded human skin ex vivo

There is a pressing need for improved preclinical model systems in which to study human skin wound healing. Here, we report the development and application of a serum-free full thickness human skin wound healing model. Not only can re-epithelialization (epidermal repair) and angiogenesis be studied in this simple and instructive model, but the model can also be used to identify clinically relevant wound-healing promoting agents, and to dissect underlying candidate mechanisms of action in the target tissue. We present preliminary ex vivo data to suggest that Thyroxine (T4), which reportedly promotes skin wound healing in rodents in vivo, may promote key features of human skin wound healing. Namely, T4 stimulates re-epithelialisation and angiogenesis, and modulates both wound healing-associated epidermal keratin expression and energy metabolism in experimentally wound human skin. Functionally, the wound healing-promoting effects of T4 are at least partially mediated via fibroblast growth factor/fibroblast growth factor receptor-mediated signalling, since they could be significantly antagonized by bFGF-neutralizing antibody. Thus, this pragmatic, easy-to-use full-thickness human skin wound healing model provides a useful preclinical research tool in the search for clinically relevant candidate wound healing-promoting agents. These ex vivo data encourage further pre-clinical testing of topical T4 as a cost-efficient, novel agent in the management of chronic human skin wounds.


Skin samples
Human scalp or corporal skin samples were obtained from patients undergoing plastic or reconstructive surgical procedures with informed consent and Institutional Research Ethics Committee permission (University of Luebeck: 06-109). The study was conducted according to The Helsinki Declaration of 1975 (revised 1983). Our study included samples from 6 patients aged from 26-67 years (average 52.6 years) (details, see Table 1).  Human skin wound healing organ culture model The human skin wound healing assay modified based on the previously published "punchin-a-punch" design [7,8,39] with the notable difference being that full-thickness (including subcutaneous fat) adult human skin was used and cultured in serum-free medium [38] (Fig  1). In the pilot study reported here, one to two "punches" were obtained for each experimental condition (control, T4 10, 100 or 1000nM) from each patient at the beginning of each culture and snap frozen at day 3 or 6 depending on the experimental group. Samples were frozen immediately for analysis (day 0) or transferred to six-well plates containing supplemented Williams E culture medium [38]. Each well contained 1-2 skin punches in 3 ml of medium. Skin samples were left untreated ("equilibration period") for the first 24 hours of the culture period. Then the medium was replaced for all samples; in the test conditions T4 was tested at concentrations ranging from 10 to 1000 nM, based on our previous human HF organ culture study [19]. Control and T4 supplemented culture medium were changed every 2 days and sample freezing was performed as per the culture protocol (Fig 1). Human skin fragments were embedded in Shandon Cryomatrix (Thermo Fisher Scientific; Waltham, MA, USA) before longitudinal cryosections (6μm) were obtained. Cryosections were stored at -80˚C until used.

Immunohistochemistry, immunofluorescence and quantitative immunohistomorphometry
The antibodies and corresponding detection methods which were used are described in Table 2. For detection of proliferating and apoptotic cells in this system, Ki67/TUNEL quantitative-immunohistomorphometry was performed as described previously [19,23,24,36,38,[56][57][58]. For the quantitative evaluation of the double-immunostaining results, DAPI-, Ki67-, or TUNEL-positive cells were counted in defined reference areas (see dotted line) in the newly regenerated epidermis (ETs), (Fig 1). The number of DAPI-positive cells served as "total number of cells", and the percentage of Ki-67-positive and/or TUNEL positive cells was calculated on this basis to enable comparison between control and test groups. Standard haematoxylin and eosin staining was used to determine the new ETs. Cytokeratin 6, CD31, bFGF, and fibroblast growth factor receptor 1 (FGFR1) immunofluorescence were detected using the previously described methods [7,19,44,59]. Mitochondrially encoded cytochrome c oxidase 1 (MTCO1) IR was detected by peroxidase-based avidin-biotin complex immunostaining, without counterstaining with Haematoxylin. A monoclonal antibody that , the IR in the ETs was analysed. Both outer and inner epithelial tongue were analysed, defined as the area from the edge of the stratum corneum where the punch had been placed to the corresponding point in the epidermal basal layer. For the angiogenesis parameters, CD31 IR, the number of CD31 positive nuclei and the number of vessel lumina were determined as previously described [44]. The Image J software (National Institute of Health, Bethesda, MD) was used for evaluation. All the samples were photographed for analysis with a Keyence Biozero-8000 Microscope (Keyence Corporation, Higashi-Nakajima, Osaka, Japan).

Statistical analysis
All the data are given as mean ± standard error of the mean (SEM). For the quantitative immunohistomorphometry analyses, the IR in the ETs was measured in up to 4 sections for each wounded skin fragment. One-Way ANOVA by appropriate post hoc comparisons was used at single time points, and if the data did not follow a Gaussian distribution, non-parametric tests were applied (i.e. Kruskal-Wallis test). Statistical analysis was carried out by Graphpad prism 5.01 (Graph Pad software, Inc., San Diego, CA, USA), and p<0.05 was regarded as significant.

Results
After wounding, a compact sheet of epidermal keratinocytes (the ETs) began to cover the wound edges in all groups (epiboly phenomenon [62]), as expected [21]. Histochemically, overall skin morphology was well-preserved until and including day 6 of organ culture, during which time no epidermal detachment from the basal membrane was seen, while the number of proliferating or apoptotic keratinocytes in the epidermis and HFs (Ki-67/TUNEL immunofluorescence microscopy) during the entire study window of 6 days was within the expected normal range [26,38].
Compared to the vehicle control, T4 administered to the culture medium significantly stimulated re-epithelialisation by day 3 after skin wounding in organ culture (Fig 2a-2d). Thereafter, higher concentrations of T4 rather slowed-down ET elongation (possibly due to an inhibition of keratinocyte migration). However, the total ET area (i.e. the mass of regenerated epithelium) was persistently larger in T4-treated compared to vehicle-treated wound skin fragments ex vivo (Fig 2a-2d).
Proliferation and apoptosis in the ETs were not significantly modulated by T4 (only highdose T4 treatment showed a [non-significant] proliferation-stimulatory trend in the ETs at day 3), as measured by quantitative immunohistomorphometry of Ki67+ or terminal deoxynucleotidyl transferase dUTP nick end labelling-positive (TUNEL+) cells in ETs (Fig 2e-2h). This may reflect the well-recognized complexity of T4's effects on the overall tissue modelling process, which represents a balance of keratinocyte proliferation, apoptosis, differentiation and migration effects [33]. This observation suggests that the re-epithelialisation-promoting effects of T4 primarily result from the stimulation of keratinocyte migration.
Furthermore, T4 treatment moderately increased CD31 IR, the number of CD31+ positive endothelial cells in T4-treated wounded skin fragments (Fig 3a-3e) and the microvessel density (Fig 3a, 3b and 3f), measured as described previously [44]. This suggests that T4 enhanced angiogenesis in wounded human skin fragments ex vivo, despite the fact that these cutaneous blood vessels are non-perfused and may collapse to some degree after surgical skin removal. Whilst the effect was modest, it was statistically significant.
Since bFGF is a key pro-angiogenic factor known to be up-regulated by T4 [45, 46], we further examined the role of this growth factor role in T4-stimulated wounded human skin. Indeed, T4 increased bFGF protein IR in the ETs (Fig 4a-4c). Next, FGFR1 IR was assessed, since the proangiogenic actions of T4 may be indirectly mediated at least in part via up regulating FGFR1 expression [64]. In fact, as shown in Fig 4d-4f, FGFR1 was significantly increased in ETs by T4 compared to control.
Most importantly, co-administration of inhibitory-bFGF antibody [45] counteracted the stimulatory effects of T4 on re-epithelialisation (Fig 5a and 5b), CD31 protein IR (Fig 5c), and number of CD31+ endothelial cells (Fig 5d), microvessel density (Fig 5e) and bFGF expression (Fig 5f). Taken together, this suggests that the wound healing-promoting effects of T4 in organ-cultured human skin are, at least in part, bFGF-dependent.

Discussion
Here we report how organ-cultured, experimentally wounded human skin can be used to identify clinically relevant wound-healing promoting drugs, and to dissect underlying candidate mechanisms of action. The main advantage of the current model is its use of full-thickness human skin (which is readily available as excess tissue from plastic surgery) in a serum-free setting. Thus, not only the interaction between the epidermis, dermis and subcutaneous tissue is preserved, but the skin appendages, including intact HFs and sweat glands, all of which may play a substantial role in wound healing [37, 65,66] are also present. In contrast to 3D skin culture models, it is therefore potentially possible to study the influence of the full range of resident skin cell populations on cutaneous wound healing. Future work may utilise the model to carefully study the influence of the HF on wound healing in human skin ex vivo. Notably, we show ex vivo data to suggest that T4, a widely used, cost efficient drug with well-known toxicology, may promote key features of human skin wound healing.
The model also enables selected skin signalling pathways to be studied, just as in serum-free human HF organ culture [67][68][69]. Namely, provide preliminary preclinical evidence that the re-epithelialisation-and angiogenesis-promoting effects of T4 in experimentally wounded human skin ex vivo primarily are due to the up-regulation of bFGF/FGFR1-mediated signalling in the wounded epithelium.
Of course, it is important to also recognize the limitations of the ex vivo model used here. These limitations include, besides the obviously missing functional skin innervation and perfusion, the relatively short time-frame in which wound healing can be studied here. After approximately 6 days, epidermal detachment becomes a prominent feature, preventing longterm application of the model. For this reason we limited the culture time to avoid confounding the data by increasing tissue degeneration effects, and thus focused on early to mediumterm interventions to promote wound healing. Another important limitation is that we used skin explants from skin containing either terminal or vellus HFs. Recently studies have confirmed the superior wound healing promoting properties of hair-bearing skin when compared to skin from "hair-free" sites [65]. Whether T4 would exert superior wound healing promoting properties in terminal hair-bearing skin compared to skin dominated by vellus HFs or hairfree palmoplantar skin remains to be clarified. It must also be borne in mind that the thyroidstatus of the subjects at the time of surgery was not known.
To consolidate the pilot results reported here, it is important that the current study is repeated both, with skin from an increased number of subjects and in additional wound healing models, including. for example humanized mouse in vivo models of wound healing [70] to confirm the positive effect of T4 on wound healing. Such pre-clinical testing is essential given the possible side-effects of topical T4 application in human subjects, included systemic absorption which might lead to a hyperthyroid state, a key consideration as increased T4 serum levels are strictly to be avoided. Reassuringly though, topical T4 may fail to influence circulating T4 levels significantly [71]. Whether the absence of perfusion-derived endocrine signalling ex vivo may render organ-cultured skin "functionally" hypothyroid and whether this is compensated for by increased intracutaneous conversion of residual T4 to tri-iodothyronine [19] is unknown. Caution must be exercised with extrapolating the effects of T4 on human skin healing ex vivo to the more complex in vivo situation, as the model does not satisfactorily permit the study of the important contributions of platelets, neutrophils and circulating T cells to wound healing [72,73]; while resident immunocytes like mast cells and macrophages are abundantly present and can be instructively studied ex vivo [74][75][76][77] Yet, the ex vivo effects of T4 reported here are well in line with its in vivo effects reported in rodent wound healing models [22,31,32] Given the overall medical importance of wound healing disorders and the urgency to develop more effective, affordable, and safe wound healing-promoting agents for the treatment of chronic skin ulcers [7,78]. Our pilot data encourage the preclinical systematic exploration of T4 as a potential wound healing promoter in experimentally wounded human skin, using additional wound healing models [8] as the next step before determining whether T4 is a plausible future skin ulcer therapy.
Supporting information S1

Acknowledgments
Most of the work reported here was performed while GYZ, EAL, NM and RP worked jointly in the Department of Dermatology, University of Lübeck. The experiments were formed/data were generated by GYZ, with help from NM. EAL and RP contributed to data analysis and manuscript writing. FS and WF provided human tissue samples and helped with manuscript editing. The excellent technical assistance of Gabriele Scheel and Claudia Kremling and the help of Dr. Tian Liao with evaluations are gratefully acknowledged as is the contribution of Dr. M. Geissen (UKE, Hamburg). This work was supported in part by a grant from the Federal Ministry of Economics, Germany (AiF/ZIM Programme) to R.P. G-Y Zhang was supported by a Chinese government award for outstanding self-financed students abroad (China Scholarship Council) and the National Natural Science Foundation of China (81101430).