Adipose-Derived Stem Cells Promote Intussusceptive Lymphangiogenesis by Restricting Dermal Fibrosis in Irradiated Tissue of Mice

Currently, there is no definitive treatment for lymphatic disorders. Adipose-derived stem cells (ADSCs) have been reported to promote lymphatic regeneration in lymphedema models, but the mechanisms underlying the therapeutic effects remain unclear. Here, we tested the therapeutic effects of ADSC transplantation on lymphedema using a secondary lymphedema mouse model. The model was established in C57BL/6J mice by x-irradiation and surgical removal of the lymphatic system in situ. The number of lymphatic vessels with anti-lymphatic vessel endothelial hyaluronan receptor 1 (LYVE-1) immunoreactivity increased significantly in mice subjected to transplantation of 7.5 × 105 ADSCs. X-irradiation suppressed lymphatic vessel dilation, which ADSC transplantation could mitigate. Proliferative cell nuclear antigen staining showed increased lymphatic endothelial cell (LEC) and extracellular matrix proliferation. Picrosirius red staining revealed normal collagen fiber orientation in the dermal tissue after ADSC transplantation. These therapeutic effects were not related to vascular endothelial growth factor (VEGF)-C expression. Scanning electron microscopy revealed structures similar to the intraluminal pillar during intussusceptive angiogenesis on the inside of dilated lymphatic vessels. We predicted that intussusceptive lymphangiogenesis occurred in lymphedema. Our findings indicate that ADSC transplantation contributes to lymphedema reduction by promoting LEC proliferation, improving fibrosis and dilation capacity of lymphatic vessels, and increasing the number of lymphatic vessels via intussusceptive lymphangiogenesis.


Introduction
The lymphatic system maintains tissue fluid homeostasis, carries proteins and large molecules, and plays a role in immune response and tumor metastasis [1,2]. Although the incidence of lymphoceles, lymphorrhea, and lymphedema after lymphatic dissection has been declining due to advances in surgical techniques, these iatrogenic disorders still lack definitive treatment and have a strong negative impact on patients' quality of life [3]. The secondary (acquired) type of lymphedema accounts for the majority of cases worldwide, occurring as a result of trauma, radiotherapy, surgery, infection, or a combination of these [2]. Lymphedema causes fibrosis in surrounding interstitial tissues, which, in turn, exacerbates lymphedema [4]. Therefore, cancer therapy with surgical lymph node dissection and radiotherapy may result in severe impairment of the lymphatic system.

Macroscopic Observation of Hind Limb Swelling
The schematic timeline of this study and representative images of the hind limb swelling are shown in Figure 1. All mice showed apparent swelling on the left hind limb within 24 h after surgery. Swelling healed in all mice in the X-ray/ADSC (−/−) group within 21-28 days. In the X-ray/ADSC (+/+) group, swelling improved in 4/6 mice by day 42. In the X-ray/ADSC (+/−) group, improvement time was sparse, and none of the mice improved faster than the mice in the X-ray/ADSC (+/+) group.
The mean lymphatic vessel area was significantly enlarged in all groups at days 2 and 14 ( Table  1). Vessel area in the X-ray/ADSC (−/−) group was further expanded at day 8, compared with that in the two X-ray (+) groups, and percentage of lymphatic vessel area was significantly increased in the X-ray/ADSC (−/−) and X-ray/ADSC (+/+) groups unlike in the X-ray/ADSC (+/−) group at day 8 ( Figure 2C).

Analysis of LEC Proliferative Activity
The effects of ADSC transplantation on lymphatic endothelial cell (LEC) proliferative activity were confirmed by immunofluorescence staining using anti-LYVE-1 and anti-proliferating cell nuclear antigen (PCNA) antibodies ( Figure 3A). When LYVE-1 positive cells formed a lumen, LYVE-1 and PCNA double-positive cells were considered as the proliferative lymphatic vessel.

Analysis of LEC Proliferative Activity
The effects of ADSC transplantation on lymphatic endothelial cell (LEC) proliferative activity were confirmed by immunofluorescence staining using anti-LYVE-1 and anti-proliferating cell nuclear antigen (PCNA) antibodies ( Figure 3A). When LYVE-1 positive cells formed a lumen, LYVE-1 and PCNA double-positive cells were considered as the proliferative lymphatic vessel.
Rates of proliferative lymphatic vessels to all lymphatic vessels are shown in Figure 3B and Table 2. On day 0, the proliferative activity of the two X-ray (+) groups was significantly suppressed; lymphatic vessel dissection triggered the increased rates of proliferative lymphatic vessels. In the X-ray/ADSC (+/+) group, the rates significantly increased on day 8 and persisted until day 14, but in the X-ray/ADSC (−/−) group, the increase was not significant. Table 2. Ratio of proliferative lymphatic vessels with LYVE-1 and proliferating cell nuclear antigen (PCNA) immunoreactivity (mean ± SE).

Evaluation of Fibrosis Using Picro-Sirius Red Staining
Picrosirius red staining was performed to evaluate the severity of skin fibrosis on the left hind limb and the effects of ADSC transplantation ( Figure 4, Table 3).

Evaluation of Fibrosis Using Picro-Sirius Red Staining
Picrosirius red staining was performed to evaluate the severity of skin fibrosis on the left hind limb and the effects of ADSC transplantation ( Figure 4, Table 3).
The collagen structure in each group was as follows ( Figure 4): • X-ray/ADSC (−/−) group, day 0: Collagen fibers of type I randomly oriented and type III spread throughout the specimen. • X-ray/ADSC (−/−) group, day 14: Shorter and still randomly oriented type I collagen fibers; type III collagen fibers only in gaps of type I fibers. • X-ray (+) groups (X-ray/ADSC (+/−) and X-ray/ADSC (+/+)), day 0: Collagen fibers of type I densely deposited and type III rarely found. • X-ray/ADSC (+/−) group, day 14: Thicker and longer type I collagen fibers deposited in parallel. Type III collagen fibers only in places of a low density of type I collagen fibers. • X-ray/ADSC (+/+) group, day 14: Randomly oriented type I collagen fibers with lower density compared to the other two groups. Type III collagen fibers spread throughout the specimen, mainly in gaps of type I collagen fibers.   Red and green channel ratio was calculated following the formula (n = 3-6): [(area of red or green channel)/(area of whole specimen)]. Each channel area was measured using ImageJ software setting as described in "Evaluation of fibrosis using picrosirius red staining". The area of the whole specimen was measured using default values of the color threshold automatically set by ImageJ. * p < 0.05, ** p < 0.01 significantly different from day 0.
The collagen structure in each group was as follows ( Figure 4): • X-ray/ADSC (−/−) group, day 0: Collagen fibers of type I randomly oriented and type III spread throughout the specimen. • X-ray/ADSC (−/−) group, day 14: Shorter and still randomly oriented type I collagen fibers; type III collagen fibers only in gaps of type I fibers. • X-ray (+) groups (X-ray/ADSC (+/−) and X-ray/ADSC (+/+)), day 0: Collagen fibers of type I densely deposited and type III rarely found. • X-ray/ADSC (+/−) group, day 14: Thicker and longer type I collagen fibers deposited in parallel. Type III collagen fibers only in places of a low density of type I collagen fibers. • X-ray/ADSC (+/+) group, day 14: Randomly oriented type I collagen fibers with lower density compared to the other two groups. Type III collagen fibers spread throughout the specimen, mainly in gaps of type I collagen fibers.
In the X-ray (+) groups on day 0, the area ratios of red (type I collagen) and green (type III collagen) channels to the whole specimen significantly increased and decreased, respectively. Two weeks after skin and lymphatic vessel incision, both red and green channel ratios in tissues from the X-ray/ADSC (−/−) group significantly decreased. From the two X-ray (+) groups, both channel ratios were further decreased only in tissue from the X-ray/ADSC (+/−) group (Table 3).
Based on the above results, x-irradiation could cause fibrosis, such as changing the orientation of type I and decreasing type III collagen fibers. The therapeutic effect of ADSC transplantation on fibrosis was to recover the orientation of type I collagen fibers from parallel to random and increase the extracellular matrix (ECM), including type III collagen fibers.

Gene Expression Analysis
mRNA extracted from the skin in edematous regions at days 0, 2, 8, and 14 was used for quantitative RT-PCR. Relative expression values were normalized by the mean value of the X-ray/ADSC (−/−) group at day 0 for each gene. Significant differences among groups are shown in Figure 5.

SEM Observation
To obtain three-dimensional structures of lymphatic regeneration in edematous regions, serial sections were prepared from the mouse left hind limb following surgical lymphatic vessel incision and ADSC transplantation. Lymphatic vessel positions were identified during scanning electron microscopy (SEM) overlaying images of the same sections stained with anti-LYVE-1 antibody ( Figure  6).

SEM Observation
To obtain three-dimensional structures of lymphatic regeneration in edematous regions, serial sections were prepared from the mouse left hind limb following surgical lymphatic vessel incision and ADSC transplantation. Lymphatic vessel positions were identified during scanning electron microscopy (SEM) overlaying images of the same sections stained with anti-LYVE-1 antibody ( Figure 6).

Discussion
The analysis of our results indicated that sustained hind limb lymphedema was caused by the inhibition of lymphatic vessel dilation and obstruction of afferent diffusion of interstitial fluid because of the fibrosis of surrounding interstitial tissues and delay of wound healing by x-irradiation. ADSC transplantation mitigated edema by promoting LEC proliferation, increasing the number of lymphatic vessels, accelerating wound healing, and improving dermal fibrosis and lymphatic vessel dilation capacity (Figure 7). In the X-ray/ADSC (+/+) groups, dilated lymphatic vessels and protruded LECs toward the lumen observed at days 8 and 14 (i.e., the time the number of lymphatic vessels increased) were considered structures analogous to an intraluminal pillar, the hallmark of intussusceptive angiogenesis. We anticipate that in the lymphedema region, intussusceptive lymphangiogenesis is occurring by restricting dermal fibrosis.

Discussion
The analysis of our results indicated that sustained hind limb lymphedema was caused by the inhibition of lymphatic vessel dilation and obstruction of afferent diffusion of interstitial fluid because of the fibrosis of surrounding interstitial tissues and delay of wound healing by x-irradiation. ADSC transplantation mitigated edema by promoting LEC proliferation, increasing the number of lymphatic vessels, accelerating wound healing, and improving dermal fibrosis and lymphatic vessel dilation capacity (Figure 7). In normal skin, type I collagen fibers are randomly oriented and contribute to appropriate interstitial tissue function. In edematous and recovering skin, type I collagen fibers are parallel deposited and interstitial tissue function is disrupted (e.g., lymphatic vessels cannot dilate and absorb the excessive fluids in the legion). ADSC transplantation improves the fibrosis by recovering the collagen fibers orientation and increasing the number of lymphatic vessels by promoting LEC proliferation and intussusceptive lymphangiogenesis. ADSC, adipose-derived stem cell; ECM, extracellular matrix; LEC, lymphatic endothelial cell.
Immunohistochemical staining with anti-LYVE-1 antibody showed that ADSC transplantation increased the number and area of lymphatic vessels compared with the X-ray/ADSC (+/−) group ( Figure 2B,C). In an acute lymphedema model, dilated lymphatic vessels were not functional, and interstitial fluid drainage was driven by interstitial forces [4,11,15,21,34]. However, these models do not reflect the severity of fibrosis in a clinical setting because they were prepared without xirradiation. X-irradiated lymphedema model is considered to cause TGF-β1-mediated tissue fibrosis Figure 7. Schematic images of adipose-derived stem cells (ADSC) transplantation effects and intussusceptive lymphangiogenesis. In normal skin, type I collagen fibers are randomly oriented and contribute to appropriate interstitial tissue function. In edematous and recovering skin, type I collagen fibers are parallel deposited and interstitial tissue function is disrupted (e.g., lymphatic vessels cannot dilate and absorb the excessive fluids in the legion). ADSC transplantation improves the fibrosis by recovering the collagen fibers orientation and increasing the number of lymphatic vessels by promoting LEC proliferation and intussusceptive lymphangiogenesis. ADSC, adipose-derived stem cell; ECM, extracellular matrix; LEC, lymphatic endothelial cell.
Immunohistochemical staining with anti-LYVE-1 antibody showed that ADSC transplantation increased the number and area of lymphatic vessels compared with the X-ray/ADSC (+/−) group ( Figure 2B,C). In an acute lymphedema model, dilated lymphatic vessels were not functional, and interstitial fluid drainage was driven by interstitial forces [4,11,15,21,34]. However, these models do not reflect the severity of fibrosis in a clinical setting because they were prepared without x-irradiation. X-irradiated lymphedema model is considered to cause TGF-β1-mediated tissue fibrosis and interferes with the long-term lymphatic function [35]. In the study, ADSCs improved the severity of interstitial fibrosis; thus, ADSCs may cause drainage of interstitial fluid by interstitial forces and its influx into lymphatic vessels through TGF-β1 pathway.
Immunofluorescence staining with anti-PCNA and anti-LYVE-1-antibodies showed that LEC proliferation was suppressed by 30 Gy x-irradiation a week earlier ( Figure 3B). Lymphatic vessel incision and ADSC transplantation stimulated LEC proliferation.
Picrosirius red staining revealed the organization of collagen fibers (Figure 4). The ADSC transplantation group showed a reduction in long and thick type I collagen fibers, a shift from parallel deposition to randomly oriented (normal disposition), and an increased ratio of the green channel, which represented a recovery of type III collagen fibers and ECMs [36,37]. These represent amelioration of fibrosis caused by x-irradiation and lymphatic incision. Fibrosis is one of the major factors that inhibit lymphatic regeneration [4]. ADSC transplantation appears effective against fibrosis because ADSCs and their secreted factors activate dermal fibroblasts and suppress inflammatory cells [26,38]. ADSC transplantation should also be effective against lymphorrhea, promoting lymphatic regeneration and surrounding tissue normalization.
Prox1 is well studied as a master switch determining the LEC fate in developing stages [39,40], but not sufficiently investigated in the adult stage; yet, its expression is considered to indicate switching of the LEC cell cycle from quiescent to proliferative [14]. The increased Prox1 mRNA expression in the X-ray/ADSC (−/−) group in our study indicate that Prox1 may relate to the lymphatic dilation.
Increased Fgf2 and Tgfb expression was reported in in vitro studies using ADSCs for wound healing [23,41]. Enhancement of type I and type III collagen expression was reported in the ADSC treated group [42], consistent with our picrosirius red staining results. From the viewpoint of lymphatic regeneration and wound healing, we considered that ADSC transplantation was effective in reducing lymphedema by affecting these cytokine profiles.
In contrast, Vegfc expression was significantly suppressed in the ADSC transplantation group despite an increase in the number of lymphatic vessels. Although several reports indicate the importance of VEGF-C/VEGFR-3 signaling for lymphangiogenesis [16,[43][44][45], prolonged inhibition using the soluble VEGFR-3 decoy receptor did not affect adult lymphatic vessels [46]. Saijo et al. reported VEGF-C and HGF secretion from ADSCs even after γ-irradiation [23]. The mechanisms of lymphatic vasculature network formation, remodeling, and adaptation to physiological and pathological challenges are controlled by an intricate balance of growth factors and biomechanical cues, which transduce signals for the readjustment of gene expression and lymphatic endothelial migration, proliferation, and differentiation [47,48]. ADSCs may relate to the composure of certain ECM proteins [e.g., the collagen-and calcium-binding EGF domains 1 (CCBE1) protein, a disintegrin, and a metalloproteinase with thrombospondin motif 3 (ADAMTS3)], which activate VEGF-C as needed. Further investigations on the efficacy of ADSC treatment for VEGF-C protein quantity and activity are required.
SEM imaging revealed LECs extended toward the inner part of dilated lymphatic vessels in the lymphedema region ( Figure 6). This structure resembled the structure of the intraluminal pillar in intussusceptive angiogenesis [49]. Thus, we predicted that intussusceptive lymphangiogenesis by ADSC transplantation occurred in a secondary lymphedema mouse model. Angiogenesis studies described two major modes, sprouting and intussusception [50,51]. The molecular mechanisms underlying intussusception are not fully understood, but several cytokines other than VEGF have been reported. Stromal-derived factor-1 (SDF-1) and its receptor CXC chemokine receptor-4 (CXCR-4), FGF2, inhibition of Notch signaling, shear stress, and nitric oxide are reported to promote intussusceptive angiogenesis [52][53][54][55][56]. Intussusceptive angiogenesis has several advantages: (1) does not require cell proliferation, (2) can rapidly expand an existing capillary network, and (3) can maintain organ function during replication [53]. These advantages are reasonable if intussusceptive lymphangiogenesis occurs in the lymphedema region. Intussusceptive lymphangiogenesis, as a mechanism, is reported in lymphatic malformation/lymphangiomas and sinuses of developing human fetal lymph nodes [57,58]. Díaz-Flores et al. suggested a possible molecular mechanism of intussusceptive lymphangiogenesis in the developing lymph node by which a high abundance of VEGF-C whole lymph node cells, without VEGF-C gradient, results in the nonsprouting engulfment of the lymph node anlage by LECs [57]. Further studies are required to understand the molecular mechanisms and morphology of the lymphatic regeneration at lymphedema status.
In conclusion, ADSC transplantation accelerated LEC proliferation, increased lymphatic vessel numbers, and mitigated fibrosis of the surrounding interstitial tissue. ADSC transplantation efficacy on wound healing is established [24,59,60]. For clinical secondary lymphedema and lymphorrhea, which are thought to be caused by surgical lymph node dissection and radiotherapy, ADSC transplantation is considered to be effective for both lymphatic function recovery and wound healing.
Growth factors secreted by ADSC transplantation (i.e., VEGF, FGF2, TGF-β, HGF, platelet-derived growth factor (PDGF), keratinocyte growth factor (KGF), fibronectin, and collagen I and increased expression of VEGFR-3) promote lymphatic regeneration [24,26,59,61]. Factors associated with lymphangiogenesis and fibrosis include: Tgfb1, suppression of which improves both fibrosis and lymphangiogenesis [18,62]; tumor necrosis factor (TNF)-α and interleukin (IL)-1 [48]; and neutrophils that increase lymphatic vessel density by increasing the number of VEGF-A/VEGFR-2 complexes and releasing VEGF-D [14]. Moreover, factors in the ECM (e.g., CCBE1 and ADAMTS3) required the activation of VEGF-C [47,48] and dipeptidyl peptidase (DPP)-IV expressed on the surface of LEC, interact with the ECM and are involved in LEC migration and tube formation [63]. Thus, the effects of ADSC on lymphedema could be clarified by assessing not only the mRNA level but also the state of these cytokines and complex formations at the protein level.
ADSC transplantation is useful for the prevention and treatment of lymphedema subsequent to irradiation and surgery, improving the fibrosis and recovery of lymphatic function. Further studies are required on the factors that contribute to accelerating lymphatic regeneration effectively.

Animals
C57BL/6J male mice, 5-6-week-old, were purchased from Japan SLC, Inc. (Shizuoka, Japan). The mice were fed a standard laboratory diet and water ad libitum for more than 1 week before starting experiments. Experiments involving animals were approved by the Animal Research Committee of Shimane University (Approval No. IZ30-127 on January 23, 2019) and conducted according to the Regulations for Animal Experimentation at Shimane University.

Mouse Secondary Lymphedema Model and Surgical Preparation
A mouse secondary lymphedema model was prepared as previously described [12,32] with minor modification. Briefly, 8-week-old mice (body weight 22.2-26.5 g) were intraperitoneally anesthetized with medetomidine, midazolam, and butorphanol mixed anesthesia and subjected to x-irradiation using MBR-1520R (Hitachi, Ltd., Tokyo, Japan) to the left hind limb at 30 Gy in a single dose 7 days before surgery. After irradiation, mice were intraperitoneally anesthetized and subjected to a circumferential incision in the inguinal region to the muscle layer. Under a microscope, the superficial collecting lymph vessels were cut and cauterized, and the 5-mm wide gap was left open.

Grouping of the Experimental Animals
Mice prepared for lymphedema were divided into three groups. In the ADSC transplantation groups, 7.5 × 10 5 ADSCs per mouse were injected with 0.3 mL cell preservative solution (CELLBANKER 2, Takara Bio Inc., Shiga, Japan). Each solution was injected subcutaneously into the distal (to the incised part) limbs and proximal limbs, 24 h after surgery, with a 27-gauge needle. In the no-ADSC transplantation groups, a mock injection of normal saline solution was similarly administered. The three groups consisted of the following: 1.
Each stained section was scanned at low magnification (40×) to select areas containing the most lymphatic vessels (high power field: HPF). Four HPFs per section were measured at high magnification (200×), and the lymphatic vessel number, mean area, and exclusive area of lymphatic vessels were calculated.
Stained sections were visualized using a fluorescent microscope (ECLIPSE 80i, Nikon Corporation, Tokyo, Japan), and scanned at 200× magnification to select the HPF for lymphatic vessels. To identify proliferative LECs, merged images of each channel were created using ImageJ software (version 1.52p, National Institutes of Health, Bethesda, MD, USA). For statistical analysis, four merged images of double-stained regions within each section were selected.

Evaluation of Fibrosis Using Picro-Sirius Red Staining
Collagen fibers in the specimens were visualized using Picrosirius Red Stain Kit (For Collagen) (ScyTek Laboratories, Inc., West Logan, UT, USA). Serial sections were deparaffinized and washed as mentioned in "LYVE-1 immunoreactivity", covered by picrosirius red solution and incubated for 1 h at room temperature. Specimens were rinsed twice with 0.5% acetic acid and immersed in 70%, 90%, 95%, 99%, and 100% ethanol and xylene.
After mounting as above, specimens were examined using a polarizing microscope (BX-51 with U-POT (polarizer) and U-ANT (analyzer), Olympus Corporation, Tokyo, Japan). To determine the proportions of differently colored collagen fibers, digital images of picrosirius red stained specimens were segmented into two-color threshold bands according to Zerbinati and Calligaro [36] with the slight modification that values were decided as follows (for ImageJ): red channel: hue 0-34, saturation 106-255, brightness 106-186; green channel: hue 35-118, saturation 106-255, brightness 26-186. The entire area of the specimen, which color threshold values were set as default for ImageJ software, and areas of each channel were measured.

Observation Using SEM
For SEM, 15-µm-thick serial sections of paraffin-embedded specimens were prepared and mounted on a MAS-GP type A-coated slide grass (Matsunami Glass Ind., Ltd., Osaka, Japan). In parallel, 5-µm-thick serial sections for LYVE-1 staining were prepared to identify the lymphatic vessel position by SEM. Sections were sequentially immersed in xylene and 99% ethanol for deparaffinization, then transferred in 100% ethanol and tert-butyl alcohol. Slides in frozen tert-butyl alcohol were lyophilized with JFD-320 (JEOL Ltd., Tokyo, Japan), gold ion coated by VX-10A (EIKO ENGINEERING, LTD., Ibaraki, Japan), mounted on an aluminum sample stage using double-sided tape and observed by JSM-6510 (JEOL Ltd.) in secondary electron image mode. laboratory instrument. We would like to thank Editage (www.editage.com) for English language editing.

Conflicts of Interest:
The authors declare no conflict of interest.