2.1.

Animals, Skin Preparation, and Wounding

All animal procedures were performed under protocols approved by the Institutional Animal Care and Use Committee (IACUC) at the University of Illinois at Urbana-Champaign. Both male BKS.Cg-Dock7m $+/+$ Leprdb/J ($db/db$) diabetic mice and male C57BL/6 WT mice (Jackson Laboratory, Bar Harbor, Maine) were used in this study. Before imaging, skin preparation and wounding were performed under anesthesia (1.5% isofluorane gas mixed with 2% oxygen). Hair on the lower dorsal skin was first shaved with electric clippers (Peanut palm size clipper 8652, Wahl), and the remaining hair in the region was further removed under a surgical microscope using surgical tweezers. The shaved skin was then cleaned with rubbing alcohol, and full thickness excisional wounds were made using a sterile 1 mm biopsy punch (Miltex, Inc., Miami, Florida). The wounded skin was then gently cleaned with rubbing alcohol.

2.2.

Stem Cell Isolation and Fluorescence-Activated Cell Sorting

Stem cell isolation was performed under a laminar flow hood using sterile technique as previously reported.²⁹ Excised epididymal fat pads or gastrocnemius–soleus complexes were extensively minced in phosphate-buffered saline (PBS) and subjected to enzymatic digestions in 0.2% Type I collagenase (Worthington Biochemical Co., Lakewood, New Jersey) for 45 min with repeated trituration. After adding the inhibition solution [20% fetal bovine serum (FBS) in Hank’s balanced salt solution (HBSS)] the samples were spun for 5 min at $450\times g$ and filtered through a $70\text{-}\mu \mathrm{m}$ strainer. The cells were then blocked with antimouse CD16/CD32 (eBioscience, San Diego, California) antibody for 10 min on ice to prevent nonspecific Fc receptor-mediated binding. Following the blocking step, cells were incubated with a fluorescent-conjugated antibody cocktail (Anti Sca-1-PE, $600\mathrm{ng}/{10}^6\text{cells}$ and anti-CD45-APC, $300\mathrm{ng}/{10}^6\text{cells}$, eBioscience, San Diego, California) diluted in 2% FBS in PBS for 1 h on ice, followed by two washes in 2% FBS in PBS. Fluorescence-activated cell sorting (FACS) was performed on an iCyt Reflection System (iCyt, Urbana, Illinois) and Sca $1^{+}{\mathrm{CD}45}^{-}$ cells were collected in growth media [high glucose Dulbecco’s modified eagle’s medium (DMEM), 10% FBS, $5\mu g/\mathrm{mL}$ gentamycin].

2.3.

Stem Cell Labeling and Injection

FACS-isolated adipose- or muscle-derived Sca $1^{+}{\mathrm{CD}45}^{-}$ cells were seeded on 100 mm plastic culture dishes (${10}^6\text{cells}$) and the growth media was changed every 3 to 4 days. MDSCs were previously characterized as ${\mathrm{Pax}7}^{-}{\mathrm{CD}31}^{-}{\mathrm{CD}34}^{-}$ cells with multilineage potential.²⁸ After 8 days, cells were incubated for 10 min at 37°C with 5-mL StemPro Accutase Cell Dissociation Reagent (Life Technologies, Grand Island, New York). Equal volume of growth media was then added to stop the dissociation reagent and cells were subjected to low speed centrifugation (5 min at $450\times g$). The supernatant was discarded and the pellet was resuspended in 1-mL serum-free media (high glucose DMEM). Cells were then incubated for 20 min with $5\text{-}\mu \mathrm{L}$ Vybrant DiI cell-labeling solution (Life Technologies, Grand Island, New York) at 37°C in the dark, followed by three washes with serum-free media. Cells were then resuspended in HBSS at ${10}^6$ cells per 1 mL and were introduced into the wound perimeter using repeated subdermal injection ($100\mu \mathrm{L}$, $9{\times 10}^4$ to ${10}^5$ cells per wound). The skin around the wound site usually “bubbled up” after injection, which was an indication that cell suspension had been introduced.

2.4.

Multimodal Microscope

Imaging of the skin was performed using a custom-built integrated MPM system described previously.^25,26 The system utilizes a tunable titanium:sapphire laser (Mai Tai HP, Spectra Physics) as the laser source. The center wavelength of excitation for both SHG and TPF imaging was 920 nm to achieve optical excitation of both the DiI-labeled stem cells and the endogenous collagen, and to remove unwanted background signal such as autofluorescence signals from other skin components.³⁰ The laser was directed through two scanning mirrors and the sample arm, and was focused onto the skin using a 0.95-NA objective lens (XLUMP20X, Olympus). The TPF and SHG signals were separated by a dichroic mirror and were simultaneously detected by two separated photomultiplier tubes (Fig. 1). With this system setup, spatial coregistration of TPF and SHG is achieved since the two types of signals are detected at the same time. The spatially coregistered MPM images can yield better insight into the relationship between the localization and cellular activity of stem cells and collagen structure during wound healing.^26,31

Fig. 1

Schematic of the MPM microscope system. M, mirror; SM, scanning mirror; L, lens; DM, dichroic mirror; O, objective; PMT, photomultiplier tube.

2.5.

Study Design

Both diabetic mice ($n=18$) and the WT mice ($n=18$) were examined in this study. Each type of mouse was divided to three subgroups (six mice/group): no treatment (saline injection), ADSCs treatment, and MDSCs treatment. All six animals in each group were wounded on Day 0. Two of the six mice in each group were randomly chosen to be imaged using the multimodal microscope on Days 0 (after wounding), 1, 2, 4, and 8. The saline/stem cell injections were performed on Day 0 after imaging, and the wounds were covered with Tegaderm (3M, Indianapolis, Indiana) and surgical gauze to minimize animal scratching and photobleaching of the DiI-labeled cells. In addition to the microscope imaging, standard photographs of the wounds on all animals were acquired on each imaging day for comparison.

2.6.

Image Acquisition and Processing

TPF and SHG images of the wounded skin were acquired concurrently with image dimensions of $\sim 2\times 2{\mathrm{mm}}^2$ using a motorized stage to scan in the lateral dimensions. The position of the laser beam relative to the skin wound was used as a reference to ensure that the wound site and the same area of skin were captured within the field-of-view during scanning each day. Large-area images were assembled as mosaics of $10\times 10$ individual images that were stitched together in postprocessing using MATLAB^® (Mathworks) and ImageJ (National Institutes of Health). TPF and SHG images of the wounds were overlaid using ImageJ.

2.7.

Wound Size and Collagen Remodeling Measurements

The size of wounds was measured in ImageJ from the photographs captured with a surgical microscope on each imaging day. The percent of wound closure was calculated by dividing the change in wound size compared to the initial wound size (Day 0) by the initial wound size.

To calculate the percent of collagen remodeling, segmentation of the SHG collagen images was performed using MATLAB^®. Each SHG image was first converted to a binary image utilizing Otsu’s method, and the wound bed and its area in each binary image can then be identified and measured. The percent of collagen remodeling was calculated using the same method as the percent of wound closure described above.

2.8.

Statistical Analysis

All data are presented as $\text{means}\pm \text{standard}$ error (SE). Wound healing dynamics (changes in wound size based on photographs) were first evaluated for WT mice, and then repeated for diabetic mice. Two-way repeated measures ANOVA ($5\times 3$) were used to investigate main and combined effects of time (Days 0, 1, 2, 4, and 8) and treatment (saline control versus ADSCs versus MDSCs) on wound size, followed by Fisher’s least significant difference post hoc analysis. Statistical analysis was performed using SPSS Ver. 20 (IBM, Chicago, Illinois). Differences were considered significant at $p\le 0.05$. In addition, a nonparametric test was performed using Friedman’s test coupled with multiple comparison analysis, and the result was comparable to that from two-way repeated ANOVA.

3.1.

Effect of Treatment in Wild-Type Mice

After wounding, WT mice were randomly divided into three treatment groups (six per group): saline treatment (control), ADSC treatment, and MDSC treatment. Figure 2(a) shows the representative photographs of the wounded skin on Days 0, 2, 4, and 8, obtained using the surgical microscope. The area of skin that was considered part of the wound was identified with dashed lines in the photographs [Fig. 2(a)]. Good healing was observed in all three groups. However, a better healing trend was observed in cell-treated groups compared to the saline treatment group [Fig. 2(b)], and the percent wound closure of both cell-treated groups was significantly higher than the saline group on Day 2 [Fig. 2(b), ADSCs: $p\le 0.01$; MDSCs: $p\le 0.05$].

Fig. 2

Wound size comparison across all treatment groups in WT mice. (a) Representative photographs of the wounded skin in each treatment group on Days 0, 2, 4, and 8. Area of the skin that is considered part of the wound was identified by dashed line. Scale bar: 1 mm applies to all photos. (b) Comparison of the percent of wound closure measured from the photographs ($n=6$, $\text{mean}\pm \mathrm{SE}$). *On Day 2, the percentage wound closure of nontreated group was significantly lower than both ADSCs treatment group ($p\le 0.01$) and MDSCs treatment group ($p\le 0.05$).

Two of the six animals in each group were randomly chosen for imaging utilizing the MPM system. Composite images of SHG and TPF illustrate the progression of collagen remodeling (SHG) and the localization of ADSCs and MDSCs (TPF) during the course of study (Fig. 3). Collagen matrix remodeling was observed in all three treatment groups. However, a more organized collagen network, consisting of stronger SHG signals as well as a "basket-weave" structural organization associated with normal skin,⁶ was readily apparent in the stem-cell treated groups compared to the saline control groups [Fig. 3(a)]. A collagen network with the “basket-weave” structure would be mechanically stronger and provide better structural support than a collagen network without this type of structure.⁶ This observation was most evident on Days 4 and 8, as both cell-treated groups demonstrated visible collagen contraction around the wound bed. In addition, the TPF fluorescence signal from stem cells was visible in both cell-treated groups, and ADSCs and MDSCs appeared to remain close to the edge and center of the wound during most of the healing. Finally, the rate of healing was assessed by quantifying the percent of collagen remodeling (see Sec. 2.7). The rate of collagen remodeling in the wound was slightly faster in both cell-treated groups compared to the saline controls, shown by the faster rise of dynamic trends [Fig. 3(b)]. The MPM results reflect a similar trend as wound closure measurements obtained by traditional photography [Fig. 2(b)], but offering additional new high-resolution subsurface visualization of the collagen network.

Fig. 3

Comparison of collagen remodeling and dynamic movement of stem cells in all three groups of WT mice. DiI-labeled stem cells were administered after imaging on Day 0, and two out of six mice from each treatment group were selected for MPM imaging. (a) Representative composite MPM images of the collagen matrix (SHG, blue), and the DiI-labeled stem cells (TPF, red) of the wounded skin in each treatment group on Days 0, 2, 4, and 8. Scale bar: $200\mu \mathrm{m}$ applies to all images. (b) Comparison of the percentage collagen remodeling calculated using the SHG images ($n=2$, solid dot: animal #1; hollow dot: animal #2). Both treated groups showed slightly faster rate of collagen remodeling compared to nontreated group.

3.2.

Effect of Treatment in Diabetic ($db/db$) Mice

$Db/db$ mice were also randomly divided into three groups and received the same treatments as WT mice. As observed in the photographs of the wounded skin, the wound size increased in $db/db$ mice without treatment (control) on Days 2 and 4, and the skin around the wound bed appeared to deteriorate with skin reddening [Fig. 4(a)]. In contrast, both stem cell-treated groups showed a progressive decrease in the wound size [Fig. 4(a)]. The percent of wound closure was also calculated. One day following treatment (Day 1), the average wound size was significantly smaller in the group treated with MDSCs compared to the saline group [Fig. 4(b), $p\le 0.05$]. Wound size was significantly smaller in both cell-treated groups two days post-treatment [Fig. 4(b), $p\le 0.05$ for ADSCs and $p\le 0.01$ for MDSCs], and remained smaller 4 and 8 days post-treatment [Fig. 4(b), $p\le 0.01$ for both cell types].

Fig. 4

Wound size comparison across all treatment groups in $db/db$ mice. (a) Representative photographs of the wounded skin in each group on Days 0, 2, 4, and 8. Area of the skin considered as part of the wound was identified by dashed lines. Scale bar: 1 mm applies to all. (b) Comparison of the percent of wound closure from all three groups of mice measured from the photographs ($n=6$, $\text{mean}\cong \mathrm{SE}$). * On Day 1, the percent wound closure of nontreated group was significantly lower than MDSCs treatment group ($p\le 0.05$). ** On Day 2, the percent wound closure of nontreated group was significantly lower than both ADSCs treatment group ($p\le 0.05$) and MDSCs treatment group ($p\le 0.01$). *** On Days 4 and 8, the percent wound closure of nontreated group was significantly lower than both ADSCs and MDSCs treatment groups ($p\le 0.01$).

Two of the six animals from each group were randomly selected for imaging with the MPM system. While the nontreated diabetic mice exhibited increased wound size during the first 4 days, immediate and progressive healing was observed in the stem cell-treated groups (Fig. 5). TPF signal was recorded inside the wound bed in both cell-treated groups [Fig. 5(a)]. Stem cells appeared to be localized to the center of the wound bed. Quantitative data obtained from the MPM images [Fig. 5(b)] reflected comparative results from the photographs [Fig. 4(b)]. However, it appeared that the collagen matrix remodeled at a slower rate than the actual wound closure [Figs. 4(b) and 5(b)]. In addition, by further comparing the photographs [Fig. 4(a)] and the SHG images [Fig. 5(a)], it was observed that the darker and reddened skin around the edge of wounds, seen in the photographs, corresponds to areas with minimal or no collagen signal. This observation is shown in Fig. 6, which again emphasizes the importance of having subsurface information to fully understand the condition of the healing skin.

Fig. 5

Comparison of collagen remodeling and dynamic movement of stem cells in all three groups of $db/db$ mice. DiI-labeled stem cells were administered after imaging on Day 0, and two out of six mice from each treatment group were selected for MPM imaging. (a) Representative composite MPM images of the collagen matrix (SHG, blue) and the DiI-labeled stem cells (TPF, red) of the wounded skin in each treatment group on Days 0, 2, 4, and 8. Scale bar: $200\mu \mathrm{m}$ applies to all images. (b) Comparison of the percentage collagen remodeling calculated using the SHG images ($n=2$, solid dot: animal #1; hollow dot: animal #2). Both treated groups showed noticeably faster rate of collagen remodeling compared to nontreated group. The nontreated group showed an increase in wound size during the study, and there were minimal changes in wound size on Day 8 compared to Day 0.

Fig. 6

Overlay of the standard photograph and SHG image from the nontreated mouse on Day 4, which shows the lack of SHG signals in the reddened skin seen in the photograph (blue arrow). Scale bar: $500\mu \mathrm{m}$.

Closer examination of the wound bed revealed colocalization of fluorescently labeled stem cells with collagen fibril network in both WT and $db/db$ mice (representative image for $db/db$ mouse shown in Fig. 7, indicated with arrows). The observation suggests the capacity for stem cells to attach to and/or remodel the ECM. In addition, Fig. 7 shows that MPM imaging provides the level of in vivo subsurface details of the skin microenvironment that cannot be observed in standard photographs.

Fig. 7

Microscopic visualization of collagen network remodeling and cellular dynamics in treated $db/db$ mice. In the SHG and TPF images acquired at the area indicated in the photograph (black box), locations where visible colocalization of collagen fibers and stem cells were identified (green arrows in MPM images). Scale bars: photograph $250\mu \mathrm{m}$; MPM images: $50\mu \mathrm{m}$.