Skin regeneration is accelerated by a lower dose of multipotent mesenchymal stromal/stem cells—a paradigm change

Multipotent mesenchymal stromal/stem cell (MSC) therapy is under investigation in promising (pre-)clinical trials for wound healing, which is crucial for survival; however, the optimal cell dosage remains unknown. The aim was to investigate the efficacy of different low-to-high MSC dosages incorporated in a biodegradable collagen-based dermal regeneration template (DRT) Integra®. We conducted a porcine study (N = 8 Yorkshire pigs) and seeded between 200 and 2,000,000 cells/cm2 of umbilical cord mesenchymal stromal/stem cells on the DRT and grafted it onto full-thickness burn excised wounds. On day 28, comparisons were made between the different low-to-high cell dose groups, the acellular control, a burn wound, and healthy skin. We found that the low dose range between 200 and 40,000 cells/cm2 regenerates the full-thickness burn excised wounds most efficaciously, followed by the middle dose range of 200,000–400,000 cells/cm2 and a high dose of 2,000,000 cells/cm2. The low dose of 40,000 cells/cm2 accelerated reepithelialization, reduced scarring, regenerated epidermal thickness superiorly, enhanced neovascularization, reduced fibrosis, and reduced type 1 and type 2 macrophages compared to other cell dosages and the acellular control. This regenerative cell therapy study using MSCs shows efficacy toward a low dose, which changes the paradigm that more cells lead to better wound healing outcome.


Introduction
After a skin injury, skin regeneration and wound healing of the epidermis and dermis are crucial to lowering the risk of infections associated with high mortality [1]. Therefore, in wound treatment, skin substitutes play an important role and provide temporary or permanent wound coverage [2] if autologous, allo-or xenografting therapy is unavailable. Cellularized skin substitutes aim to mimic skin and are being developed having great potential [3][4][5] once they are commercially available. Many acellular skin substitutes are widely used [6]. Integra® is one of the most recognized scaffolds worldwide and is approved for acute as well as chronic wounds [7]. It is a synthetic biodegradable bilayer consisting of a bottom acellular dermal matrix-a porous crosslink of bovine type I collagen and shark cartilage-and an upper-protecting silicon layer. This acellular dermal regeneration template (DRT) provides a scaffold for endogenous cell ingrowth and dermal stroma synthesis following healing.
However, we previously observed in an umbilical cord stem cell study using a bio-printer with direct cell depositioning onto burn wounds comparing to the cellularized DRT that even a lower dose regenerated the skin [10]. This raises the challenging question of which dose is optimal, as the general research hypothesis from these previous papers is that more cells lead to better wound healing outcomes.
The aim of this study was to determine the efficacy of low-to-high doses of MSCs incorporated into the DRT for wound healing and skin regeneration, applied once on full-thickness burn excised wounds.
MSCs were extracted from the stroma-Wharton's Jelly from umbilical cords [28,29], which we received from the Obstetrical and Gynecology Department at the Sunnybrook Hospital, cultured (Gibco™ DMEM, Thermo Fischer Scientific, enriched with 1% antibiotic-antimycotic solution, Gibco™, 1% L-Glutamine, Sigma Aldrich, and 10% fetal bovine serum, Gibco™ Life Technologies Corporation, USA), and expanded (until cell passage [3][4]. Further, stem cell differentiation assays were performed to confirm the differentiation potential into the mesenchymal lineages (adipose, cartilage, and bone) [10], as recently described and shown using our published protocols in a parallel project using the same cells seeded on Integra® [10], both followed after confirming the paracrine in vitro effects of the extracted cells for wound healing in our lab as previously shown [29,30].
As previously described [9], first, sorted UC-MSCs were resuspended and spun down. A cell count for viability was performed. Second, equal cell distributions for each wound treatment were transferred into 50 ml Falcon tubes containing + 25% of cells and 2 ml cell medium (Gibco™ DMEM, enriched with 1% antibioticantimycotic solution, 1% L-glutamine, and with 10% FBS). Third, the cells were resuspended and transferred into a petri-dish and homogenously pipetted with a multi-channel-pipette (VWR High Performance Signa-ture™) on the acellular Integra® on top of the bovine collagen, with the silicone side facing down on a sterile cell culture disk. The cells were seeded on DRT, which builds connections with the wound bed after surgical placement. Each DRT was prepared with 200-2,000, 000 cells/cm 2 according to our experimental protocol. The acellular control was prepared similarly with a mix of PBS and DMEM. Importantly, the DRTs absorbed the entire volume of the cells and PBS suspensions. Both groups were then placed in the incubator at 37°C at 5% CO2 until grafting on the pig. Shortly before surgery, the cellularized scaffolds were assessed under the microscope for floating cells indicating cell death and/or failure to integrate. No floating cells could be detected in either of the scaffolds, indicating full cell integration. From initial scaffold preparation until surgical grafting, less than 90 min of time had passed.
One Integra® scaffold with a cell density of 5000 cells/ cm 2 was assessed 12 h after cell incorporation and incubation at 37°C at 5% CO 2 using a confocal microscope. By imaging, cells were detected until a depth of 123 ± 21 μm, in the 1.3 mm thick scaffold, including the silicon bi-layer (Supplementary Figure 1A-D).

Full-thickness burn porcine model
Yorkshire pigs [36] were used (N = 8) which possess similar anatomic and physiologic skin characteristics and comparable pigmentation to humans [16,37,38]. Large wound sizes did not allow spontaneous healing via contracture [39]. The model has been validated from other authors as a sufficient full-thickness burn excised wound model [16,37,38].
One week after being acclimatized and treated with preventive antibiotic for 5 days (ceftiofur injection daily), all eight 4-month-old male Yorkshire pigs, with a minimal weight 25 kg and length of 60 cm, were exposed to fullthickness burn injuries until the muscle fascia of multiple 5 × 5 cm wounds (TBSA of 25%) on the dorsal back after a standardized protocol under general anesthesia and analgesia (Buprenorphine 0.05 mg kg − 1 subcutaneous, ketamine 0.2 mg kg − 1 subcutaneous combined with atropine 0.5-1.0 mg depending on the heart rate, as well as isoflurane 5%/l/O 2 intubation).

Wound treatment
Full-thickness burn tissue excision and hemostasis were performed 48-h post-burn until the muscle fascia on the surgery day (day 0), and wounds were treated with the prepared cellularized DRT and the acellular control (Integra® alone). The scaffolds were additionally fixed via skin stapler on the wound edges. Regular wound dressing changes (2-3 times/week), as well as 4 mm tissue punch biopsies, were performed at determined timepoints. Wound dressing was applied using a layer of topical antibiotics (Polysporin®), fat-gauze (Jelonet®), multiple layers of gauze, as well as adhesive dressing (Tegaderm®), and a costume-made animal compression jacket (Fig. 2a).

Presence of labeled cells on the wounds
Sorted UC-MSCs (1,000,000) were labeled with 6 μl of a lipid cell surface dye (DiO; Vybrant Cell Labeling Kit, eligible for flow cytometry, DiO yellow channel (V-22886) Abs 484(nm)/Em 501 (nm), FITC) [40]. Additionally, cell viability after labeling was performed according to the manufacturer's protocol and assessed 12 h using Live/Dead® Viability/Cytotoxicity Kit, Invitrogen (Calcein 494/517 nm, Ethidium homodimer-1/DNA 528/ 617 nm). The labeled cells were incorporated with a density of 40,000 cells/cm 2 into equally cut 5 × 5 cm meshed acellular DRT, and were grafted on fullthickness burn excised wounds on day 0. Full-thickness tissue biopsies were taken on days 2, 4, 7, and 9 at every dressing change from rotational quadrants of the wounds. The tissue biopsies were collagenased and analyzed via flow cytometry for detection of a double positive signal with DiO on CD90+ cells (BV510) (eBioscience). Labeled cells (CD90+, DiO) were present in the wound biopsy on the pigs until day 7 in a repeated experiment (Supplementary Figure 1E).

Wound healing assessment
On day 28, photography and biopsies were taken from each wound center and fixed in formalin, followed by 70% EtOH. Paraffin-embedded slides were stained after protocols for Masson's trichrome and immunohistochemistry. Antibodies used were CD11b (ab133357, rabbit monoclonal, Abcam), CD163 (ab87099, rabbit polyclonal, Abcam), CD31 (ab28364, rabbit polyclonal, Abcam), and aSMA (ab18415, monoclonal, Abcam), which were visualized via HRP polymer detection, followed by betazoid DAB chromogen kits (Biocare), before mounting and evaluation by light microscopy (Lei-caDM 2000 LED). All histology samples were assessed on three different points on the epidermis and in the dermis, measuring in the same depth, from the epidermis 2000 μm into the dermis. Two blinded independent researchers evaluated each sample, and two blinded plastic surgeon clinicians evaluated the photography, being familiar with the chosen approach.
The outcome of each wound healing parameter was analyzed descriptively (median, IQR) in the nonparametric data-set using Microsoft Excel. Graphical presentation was performed using GraphPad Prism Version 8.0. For the Supplementary Material for the dose-curves, the statistical program Python was used. Graphical illustrates is shown with a regression (of order 2) with the line-of-best-fit, and with a 95-confidence interval (Supplementary Material Figure 2).

Ethical approval
This study was approved and performed in accordance with the guidelines and regulations of the Research Ethics Board (REB), Sunnybrook Health Science Centre (REB # 017-2011) (AUP # 16-600). It was executed accordingly in agreement with the Animal Policy and Welfare Committee of the University of Toronto, where veterinarian technicians monitored the procedure and wellbeing in routine safety and health checks. During the trials, no adverse events occurred. The investigated animals maintained their health during the entire experiment.

Macroscopical wound healing
Wound healing was assessed via photography after 4 weeks after treatment, as per the definition in the remodeling phase [41]. The epithelialization area per wound was calculated [(area without epithelialization in cm 2 on day 28 × 100)/initial wound size in cm 2 on day 0)]. The MSC-treated groups showed a median between 96 and 81% epithelialization compared to the acellular control with a median of 92% (IQR 89-95). The low dose group with 5000 cells/cm 2 showed the fastest epithelialization with 96% epithelialization (IQR 91-97), followed by 40,000 cells/cm 2 with 95% epithelialization (IQR 89-96). The lowest dose of 200 cells/cm 2 and high doses of 200,000-2,000,000 cells/cm 2 showed inferior wound healing compared to the acellular control with epithelialization between 81 and 91% (IQR 69-92) (Figs. 1c and 2b).
Scarring was assessed using the Vancouver Scar Scale (VSS, vascularity, pigmentation, pliability, and height), which is the most recognized and validated [42] scar scale [43,44], and has been used previously for skin graft assessment [10,44]. The MSC-treated group of 40, 000 cells/cm 2 showed the lowest scarring with a median VSS of 6 with the narrowest interquartile range (IQR 6-7). The highest dose of 2,000,000 cells/cm 2 (IQR 4-9) and the lowest dose of 200 cells/cm 2 (IQR 5-9) both had the same median VSS of 6. The other MSC-treated groups of 5000, 200,000, and 400,000 cells/cm 2 showed a median VSS of 8 (all IQR 7-8), compared to the acellular control with the same median VSS of 8 (IQR 7-10). Overall the MSC-treated groups appeared less inflamed, with a more homogenous scar texture. The lowest and the highest dose had a sample size of N = 3, while the other dose groups had N = 6 ( Figs. 1c and 2b).

Epidermal regeneration
Histological assessment was also performed 4 weeks after surgery, where tissue biopsies from the wound centers were taken and stained after Masson's trichrome protocol. For references, healthy porcine skin representing the physiological condition had a median of 165 μm (IQR 159-182 μm), and burn wounds, without any treatment, had a median of 63 μm (IQR 49-75 μm). Hypoand hyperplasia were defined as inferior or superior epidermal thickness from the interquartile range of the healthy skin. The best regenerated epidermal thickness was achieved from the dose of 200,000 cells/cm 2 with a median of 157 μm (IQR 99-198), followed by the dose of 40,000 cells/cm 2 with a median of 189 μm (IQR 132-262), and the dose of 400,000 cells/cm 2 with a median of 131 μm (IQR 116-149). The acellular control showed a median of 177 μm (IQR 64-383 μm), although it lagged in epidermal regeneration and demonstrated a high range of hypo-and hyperplastic epidermal thickness, where the Integra® scaffold was incompletely degraded by day 28. The DRT was visible in none of the MSCtreated groups. The dose of 5000 cells/cm 2 showed a median of 284 μm (IQR 205-286 μm) and, according to the reference, was defined as hyperplasia, although the histology showed a very homogenous epidermal regenerated architecture with rete ridges comparable to the other MSC-treated groups (Figs. 1c and 3a, Supplementary Figure 1F).
The tissue was also stained for positive inflammatory markers. Due to the high cross-reactivity of the antibodies in the pig tissue, it was challenging to find reliable markers for macrophages. A clear signal was found for CD11b and CD163. CD11b is a pan-macrophage marker, which is expressed on a variety of leukocytes and is upregulated on activated cells, including type 1 macrophages [49,50]. Due to the observation that the tissue showed different states of present inflammatory cells in the remodeling phase on day 28 depending on . K indicates 1000. Heat color map: dark blue indicates healthy skin as the physiologic and best condition. Lighter blue shades are first, second, and third best, respectively. Light yellow shaded color indicates the acellular control, which is the current treatment standard used in clinic. Everything from yellow to dark orange indicates a worst outcome compared to the acellular control. Orange indicates burn alone, the worst condition wound location, the epidermal border region and the dermis was assessed separately to quantify differences. The lowest CD11b-positive cell counts were found in the epidermal border region in the wounds with 40, 000 cells/cm 2 with a median of 8 (IQR 6-11), followed by 200 cells/cm 2 with a median of 12 (IQR 5-15) and 5000 cells/cm 2 with a median of 12 (IQR 7-18). The wounds with 400,000 cells/cm 2 showed a median of 22  and the wounds with 2,000,000 cells/cm 2 showed a median of 37 (IQR 31-44), which was more compared to the acellular control with a median of 17 (IQR 10-30). Evaluating the dermal region, the dose of 40,000 cells/cm 2 showed the lowest positive cell count of 7 (IQR 6-11), followed by the dose of 5000 cells/cm 2 with a median of 9 (IQR 8-10), and 200 cells/cm 2 with also a median of 9 (IQR 6-16). The acellular control showed a lower median of positive counted cells of 32 (IQR 15-49) compared to the highest dose group with 2, 000,000 cells/cm 2 with a median of 36 (IQR 8-50) (Figs. 1c and 3e).
Along with the pro-inflammatory marker CD11b, the tissue was stained for CD163, which is a marker expressed on anti-inflammatory and pro-repair cells such as type 2 macrophages [51,52]. In the epidermal border region, all MSC-treated groups showed a lower positive cell count of CD163 positive cells, compared to the acellular control. The wounds with 5000 cells/cm 2 showed the lowest median of 10 (IQR 9-20), followed by 40,000 cells/cm 2 with a median of 12 (IQR 8-24), and 200 cells/cm 2 with a median of 25 (IQR , than the acellular control with a median of 68 (IQR 48-72). In the dermal part, all MSC-treated groups showed fewer positive cells than the acellular control with a median of 68 (IQR 34-72). Within the different dose groups, we found the lowest positive cell count when treating wounds with 5000 cells/cm 2 with a median of 8 (IQR 6-13), followed by 40,000 cells/cm 2 with a median of 12 , and 200,000 cells/cm 2 with a median of 31 (IQR 25-34) (Fig. 1c, 3f).

Discussion
In our low-to-high MSCs-dose treatment model, where we evaluated 8 wound healing parameters, we show that the low dose of 40,000 cells/cm 2 regenerates the fullthickness burn excised wounds most efficaciously, followed by an even lower dose of 5000 cells/cm 2 . Third was equally effective at 200 and 200,000 cells/cm 2 compared to higher dosages up to 2,000,000 cells/cm 2 . This is an important finding given that previous studies have hypothesized that more cells lead to a better Fig. 2 Overview experiment, macroscopical wound healing. a Overview of the experiments. b Photography of macroscopical wounds on day 28 from the initial 5 × 5 cm full-thickness burn excised wounds outcome in skin healing [9,[12][13][14][15][16]32]. MSC cell therapy is a potentially powerful treatment and (autologous) sources are readily and cost-effective available. Therefore, determining cell dosage for clinical trials is essential to preventing therapy failure. Our study with a wide dose range fills a gap with respect to dosage and discusses the effects of future cell-based therapy.
We confirm with our pre-clinical results' previous findings stating that mesenchymal stromal/stem cell therapy improved macroscopical wound healing with faster epithelialization, reduced scarring, and reduced inflammation. Furthermore, we proved that this beneficial cell therapy with pro-angiogenic and fibroproliferative effects increased collagen formation, increased neovascularization, and reduced fibrosis [11,30,[53][54][55]. Additionally, we demonstrate that the newly cellularized MSCs treatment is safe and accelerates wound healing more effectively compared to the acellular control used in clinic.
We hypothesize that the better outcome in the low dose range is explainable due to the very simple adage "the dose makes the poison" and with three underlying mechanisms (based on a publication in Cell, of a mathematical model of cell circuits of cell proliferation and death [56]). First, an excessive amount of grafted stem cells, such as 2,000,000 cells/cm 2 , may be proliferating to a maximum consuming space and use all available growth resources. This generates a lack of nutrients and possible hypoxia in the wound environment which would lead to cell death. Massive signaling occurs which needs to be regulated and may take longer until tissue regeneration occurs compared to other cell dosages. It has been shown that MSCs reduce hypoxia-induced apoptosis [57] and additionally showed a beneficial initial inflammatory upregulation in MSCs that prevents hypertrophic scar formation [54,[58][59][60], which would be in line with our findings. For very low initial cell concentrations, the cell numbers may be declining since the critical threshold of hemostasis is not reached, but the paracrine signals may provide the neighboring cells in the wound bed a very early proliferative healing boost as shown in the results. Given the initial appropriate range of dose, hemostasis can be achieved faster, leading to the most optimal accelerated healing. However, the explanation why different cell-dosages have varying efficacies might be more complex. The extracellular microenvironment (and the biomaterial as cell carrier itself) is taken into account. For instance, recent studies have shown that the collagen scaffold as MSC carrier leads to inferior wound healing compared to xenografts [61], but the DRT also demonstrated superior healing compared to a soft, fast biodegradable biomaterial [10]. This highlights that the extracellular components also play a detrimental key role in guiding the cells.
This analysis presented here evolved after an unexpected observed low-dose healing phenomenon in an ongoing trial, where we retrospectively analyzed our collected dataset. We therefore recommend for future research to create a dose-model that is translatable and to implement objective scientific methods to determine healing or any outcome measures of interest (Supplementary Material, Table 1).

Limitations
We did not determine the cell viability, the state of differentiation, or the potential harm of the delivered cells on the in vivo wounds after grafting (of the > 153.9 billion cells). This would have been interesting but not feasible in such large inflicted injuries primarily investigating wound healing (and therefore avoiding wound biopsies for proofing). Each cell manipulation can potentially affect the transplanted cells by inducing down-stream changes [62]. We performed cell sorting for the MSCs surface markers 1 day before surgery and found similar quantities as other researchers found after large scale expansion using UC-MSCs [63]. Before incorporating into the DRT, the cells had a homogenous morphology by adhering on the plastic culture flask before preparation and same-day-surgery.
In comparable cell tracing experiments in Integra®, it was shown that the cells were also no longer detectable [33,34] after 1 week, using the same cell surface dye [40]. These wounds were excluded in our calculation due to the multiple biopsies needed for analysis. Exact cell tracing using eventually methods such as GFP + -tracing to determine cell fate would have been interesting, but this was not our primary focus.
Furthermore, it would have been interesting to take biopsies and investigate the molecular cytokine profile [64] and perform quantitative analysis of paracrine effects [65]. Moreover, we could have investigated the survival in each dose wound and measured hypoxia. However, due to the constraints of the study and the costly time-consuming nature of porcine research, we did not perform these analyses herein. Fine-tuning and optimizing the cell dosage as well as measuring alterations in the cytokine/chemokine profile from various cell concentrations and cell sources might be done in the future in a prospective setting.
For statistical analysis, we tried to cluster MSC doses in our non-parametric dataset to a low-middlehigh dose, using generalized estimating equations (GEEs) models. This estimation model accounted for dependencies among the data introduced by multiple wounds on the same pig ("healing capacity of each individual"), including the treatments (7 treatments, 2 references), which were performed on 3-7 different pigs, between 3 and 12 times. However, due to the low N the determination between the clusters would have neglected the third best results of 200,000 cells/ cm 2 . Additionally, the (extreme) lowest dose of 200 cells/cm 2 and (extreme) highest dose of 2,000, 000 cells/cm 2 would have not been included in the model, due to the sample size. Therefore, we decided to show the descriptive data rather than the misleading GEE significance.

Future directions
Our results in this pre-clinical study highlight two directions for future research. First, the dose-model can be translated to humans for potential autologous MSCs treatment trials, due to the principal similar skin structure from pigs to humans. MSCs treatment works at any dosing-from low to high; however, it is crucial to determine the most optimal cell therapy for patients, for partial and full-thickness regeneration, for acute and chronic wound healing.
Second, an off-the-shelf therapy using the released paracrine products will be successful, if the optimized "cell dose cocktail" is quantitatively determined. This therapy can then act as a frontier in regenerative medicine.

Conclusion
This study gives new insights based on a cell-dosedependent wound healing model in full-thickness skin regeneration and shows most efficacy in low doses compared to higher dosages. This is a decisive finding for future investigations using stromal/stem cells. To the best of our knowledge, there are no comparisons available that including such a wide range of doses in such a large pig animal trial. This cell-dose model can be translated and implemented to innovative, regenerative stem cell therapy.

Supplementary Information
The online version contains supplementary material available at https://doi. org/10.1186/s13287-020-02131-6.  Figure S2. Line of best fit for cell dose concentration per parameter. Each graph illustrates a regression (of order 2) with the line-of-best-fit; xaxis is cell dose concentration shown on a logarithmic scale and on the y-axis are the parameters. The 95-confidence-interval is the area in light blue. (Not normalized, raw data set.). Table S1. Future directions, outlook and other research questions. Potential associated explanations for different outcome in the presented data set.