Centella Asiatica Ameliorates Radiation-induced Epithelial Barrier Dysfunction by Secreting Epidermal Growth Factor in Endothelial Cells

Seo Young Kwak Korea Institute of Radiological & Medical Sciences https://orcid.org/0000-0003-1252-3549 Sehwan Shim Korea Institute of Radiological & Medical Sciences Won Il Jang Korea Institute of Radiological & Medical Sciences Seung Bum Lee Korea Institute of Radiological & Medical Sciences Min-Jung Kim Korea Institute of Radiological & Medical Sciences Sunhoo Park Korea Institute of Radiological & Medical Sciences Sang Sik Cho Korea Institute of Radiological & Medical Sciences Hyewon Kim Korea Institute of Radiological & Medical Sciences Sun-Joo Lee Korea Institute of Radiological & Medical Sciences Hyosun Jang (  hsjang@kirams.re.kr ) Korea Institute of Radiological and Medical Sciences (KIRAMS) https://orcid.org/0000-0002-59105099


Results
CA treatment attenuated radiation-induced endothelial dysfunction in human umbilical vein endothelial cells and mitigated radiation-induced enteropathy in a mouse model. In particular, treatment of the conditioned media from CA-treated irradiated endothelial cells recovered loss of epithelial integrity by regulating zonula occludens 1 and desmoglein 2 in radiation exposure. We also determined that epidermal growth factor (EGF) is a critical factor secreted by CA-treated irradiated endothelial cells. Treatment with EGF, which can mimic the effect of CA-induced secretion in irradiated endothelial cells, effectively improved the radiation-induced epithelial barrier dysfunction. In addition, blockade of EGF in CA-induced endothelial secretome impeded epithelial barrier recovery. Finally, we identi ed the therapeutic effects of CA-induced endothelial secretome in a radiation-induced enteropathy mouse model with epithelial barrier restoration.

Conclusions
We have shown therapeutic effects of CA on radiation-induced enteropathy, with the recovery of endothelial and epithelial dysfunction, focusing on the crosstalk between endothelial cells and epithelial cells. Thus, our nding suggest that CA is an effective radio-mitigator against radiation-induced enteropathy.

Background
Radiation-induced intestinal injury is observed following clinical application of radiation for pelvic cancer or radiation exposure in a nuclear accident. Severe intestinal damage, with insu cient epithelial cell production and instability [1,2], leads to side effects like vomiting, weight loss, diarrhea, infections, and septic shock-induced death [3]. The endothelium has been described as an important component involved in gastrointestinal (GI) disease [4], and it has been proposed that the pathogenesis of radiationinduced enteropathy is associated with endothelial dysfunction and death [5][6][7]. Meanwhile, prevention of endothelial cell damage by growth factors, such as vascular endothelial cell growth factor or basic broblast growth factor, results in reduction of intestinal crypt cell damage, in ammation, organ failure, and death in radiation-induced GI toxicity [8].
The epithelium in the intestine is anatomically positioned in close proximity to a number of sub epithelial cell types, including endothelia. Crosstalk between epithelial cells and these sub epithelial cell populations contributes to epithelial function through paracrine signaling [9][10][11][12]. The secretome is de ned as a subset of a proteome that contains all proteins that are actively exported from the cell.
Typically, secreted protein plays a direct autocrine and/or paracrine role in a broad range of biological processes, including homeostasis, developmental regulation, immune defense, development of the extracellular matrix, and signal transduction [13][14][15].
The epithelial barrier is the rst line of defense in the GI tract that prevents the diffusion of pathogens into intestinal mucosa. Radiation-induced tissue damage and disrupted healing result in alteration of the tissue architecture and functions. Protein-protein networks connect epithelial cells, thereby forming intracellular junctions that include the adherens junction (AJ), tight junction (TJ), and desmosome [16]. Paracellular permeability across the epithelial cells is regulated by TJ, whereas AJ participates in cell-cell adhesive interactions [17]. Epithelial cells also form desmosomes that are involved in strength adhesive interactions [18,19]. Factors that prevent epithelial barrier dysfunction are important in developing therapeutic drugs for the radiation-induced GI damage [20][21][22].
Centella asiatica (CA), known as Asiatic pennywort, is widely used as a traditional herbal medicine in China and Indian. This tropical medicinal plant is enriched with bio avonoids, triterpenes, and selenium and has been reported to promote healing for ulceration, diarrhea, mental clarity, depression, and skin psoriasis [23][24][25][26]. In recent years, much attention has been paid to the potential of CA in the treatment of various types of disease and some putative mechanisms have been proposed, including antioxidant and lipid metabolism in the skin and neurons [27,28]. Importantly, it has also been reported that CA can protect endothelial cells, increase cell proliferation, inhibit apoptosis of endothelial cells, and block betaamyloid peptide aggregation [29]. Madecassoside, one of the triterpines isolated from CA, is known to preserve endothelial cells from oxidative injury by protection of mitochondria membrane potential and apoptosis [30]. Tumor necrosis factor-alpha from asiatic acid, which is the other triterpene in CA, attenuates endothelial barrier dysfunction, thereby resulting in prevention of atherosclerosis [31]. However, the effect of CA on radiation-induced endothelial cell damage has not yet been investigated.
The results of the present study show that CA ameliorated radiation-induced enteropathy with recovery of endothelial cell damage and epithelial barrier dysfunction. We hypothesized that the soluble factor secreted by CA-treated irradiated endothelial cells could repair the radiation-induced enteropathy by regulating the epithelial barrier. We found that the conditioned media (CM) of CA-treated irradiated endothelial cells reversed the radiation-induced epithelial barrier dysfunction in vitro as well as in the radiation-induced enteropathy in a mouse model. We also discovered that CA treatment of irradiated endothelial cells induced the secretion of epidermal growth factor (EGF), which is necessary for the repair of radiation-induced epithelial barrier dysfunction, including integrity and expression of junction proteins.
Of particular note, blocking EGF in CM using a neutralizing antibody failed to rescue the epithelial barrier dysfunction. Collectively, the results demonstrate that CA ameliorated radiation-induced enteropathy by regulating endothelial cell secretome.

Results
CA attenuates radiation-induced endothelial dysfunction.
To investigate the radio-mitigator effects of CA on irradiated endothelial cells, we performed several assays using human umbilical vein endothelial cells (HUVECs) in the presence or absence of CA. We used the CCK-8 assay in irradiated HUVECs to assess cell proliferation. Irradiation of HUVECs showed a signi cant downregulation of cell proliferation compared to the control, but CA treatment rescued the radiation-induced loss of cell proliferation (Fig. 1A, B). Because radiation induces cellular senescence [32], we tested the cellular senescence activity using a β-galactosidase (β-gal) assay. The β-gal activity was observed in irradiated HUVECs, but CA treatment of irradiated HUVECs displayed lower β-gal activity than irradiated HUVECs (Fig. 1C). A tube formation assay was performed to assess angiogenic capacity. Tubeforming activity of HUVECs was inhibited by radiation, but CA treatment restored the angiogenic activity of irradiated HUVECs (Fig. 1D). These results suggest that CA mitigated radiation-induced endothelial dysfunction, including proliferation, senescence, and angiogenic properties.
CA mitigates radiation-induced enteropathy in mouse model.
We evaluated the therapeutic effects of CA in mice using a radiation-induced enteropathy mouse model in which the abdomen of the mouse was irradiated. Mice were then either treated with CA or left untreated. Six days after irradiation, the effect of CA on radiation-induced enteropathy was determined using physiological and histological examinations. CA administration to the irradiated mouse attenuated loss of body weight compared to the irradiated group ( Fig. 2A). Histological analyses of irradiated intestine showed shorter villi length and crypt disruption, whereas CA treatment restored villi length and crypt numbers in irradiated mice (Fig. 2B, C). Histological scoring, accomplished by evaluating epithelial structural damage, vascular dilation, and in ammatory cell in ltration in the mucosa and submucosa, was lower in CA-treated irradiated mice than the irradiated group (Fig. 2D). Immunohistochemical activity for Ki-67, a proliferation marker, was also increased in the CA-treated irradiated mouse group compared to the irradiated group (Fig. 2E). As indicated by immunohistochemistry for the endothelial cell marker CD31, angiogenic activity was also higher in CA-treated irradiated mice than in the irradiated group (Fig. 2F). Taken together, these results suggest that CA alleviates radiation-induced enteropathy with restoration of endothelial dysfunction.
CA attenuates radiation-induced intestinal barrier dysfunction in mouse model.
The epithelial barrier requires a monolayer of epithelial cells to separate organs from the extracellular environment. An intact epithelium plays a pivotal role in defense against exogenous pathogens. Conversely, impaired intestinal epithelial barrier function is a hallmark of GI diseases, such as in ammatory bowel disease and celiac disease [33][34][35][36]. Based on the aforementioned knowledge, we investigated whether CA affects radiation-induced intestinal barrier dysfunction in a mouse model system. We evaluated bacterial translocation in mesenteric lymph nodes as a measure of the integrity of the intestinal barrier. The bacterial translocation in the mesenteric lymph nodes was increased in the irradiated mouse group compared to the control group, but it was decreased in the CA-treated mouse group compared to the irradiated group (Fig. 3A). Next, we assessed expressions of the several molecules regulating the barrier function. Immunohistochemistry analysis showed that cells positive for epithelial barrier-related molecules, such as villin, zonula occludens 1 (Zo1), Desmoglein 2 (Dsg2), and Claudin 3 (Cldn3), were decreased in the irradiated group compared to the control group. However, these expressions were restored in the CA-treated group (Fig. 3B). We also assessed mRNA levels of these molecules in intestinal tissue. The mRNA levels of epithelial barrier-related molecules in the irradiated mouse group showed a signi cantly lower expression compared to the control group, but CA treatment restored mRNA expression (Fig. 3C). Taken together, these results suggest that CA attenuated radiationinduced enteropathy thereby avoiding intestinal barrier dysfunction in a mouse model.

Secretome of CA-treated endothelial cells repairs epithelial barrier dysfunction.
The epithelium, which is considered to be responsible for protection against exogenous pathogen, is constantly exposed to soluble factors produced by surrounding cells in the microenvironment [37][38][39]. Considering CA alleviates radiation-induced enteropathy with improvement of endothelial cell function in vitro, it was decided to evaluate the functional effect of a CA-treated endothelial cell secretome on epithelial cell damage repair. We used well-established in vitro models re ecting epithelial barriers [40,41] to evaluate the functional activity of endothelial cell secretome on a damaged epithelial barrier. The CM of HUVECs was collected after irradiation (IR) or irradiation followed by CA treatment (IR + CA) in serumfree medium. The CM of each sample was tested on a Caco-2 monolayer. As shown in Fig. 4A, the transepithelial electrical resistance (TEER) value of the CM from IR HUVECs-treated irradiated Caco-2 monolayers was decreased compared with that of IR HUVECs-treated non-irradiated Caco-2 monolayer. Otherwise, the CM of IR + CA HUVECs treatment increased the TEER value in irradiated Caco-2 monolayers (Fig. 4A). In addition, Fluorescein isothiocyanate (FITC) ux of the CM of IR HUVECs-treated irradiated Caco-2 monolayers was increased compared to the CM of IR HUVECs-treated non-irradiated Caco-2 monolayers. The CM of IR + CA HUVECs treatment on irradiated Caco-2 monolayers decreased FITC ux in FITC-dextran assay (Fig. 4B). The cell-cell contact strength of irradiated Caco-2 monolayers was also improved by the CM of IR + CA HUVECs treatment (Fig. 4C). We used immuno uorescence to evaluate the expression of barrier integrity-related molecules. In the mouse model, we identi ed that CA markedly upregulated Cldn3 expression in irradiated intestine. Unfortunately, loss of CLDN3 was not rescued by the CM of IR + CA HUVECs treatment in irradiated Caco-2 monolayers (data not shown). Although ZO1 and DSG2 were lost in the junctions of the CM of IR HUVECs-treated irradiated Caco-2 monolayers, the loss of junctional molecules was recovered by CM of IR + CA HUVECs treatment of irradiated Caco-2 monolayers (Fig. 4D). Consistent with these results, protein and mRNA levels of ZO1 and DSG2 were decreased in the CM of IR HUVECs-treated irradiated Caco-2 monolayers, whereas these expressions were restored by treatment of the CM of IR + CA HUVECs (Fig. 4E, F). Collectively, CA modulated the endothelial secretome to restore radiation-induced barrier dysfunction particularly that associated with ZO1 and DSG2.
EGF is identi ed as a key regulator of restoration of radiation-induced epithelial barrier dysfunction.
A number of studies have revealed that endothelial cells secrete a variety of biologically active growth factors, cytokine, extracellular matrix protein, and tissue remodeling enzymes [42]. The factors from endothelial cells may help restore the GI epithelium. To elucidate which secretory molecules in uence the repair of radiation-induced epithelial dysfunction, we performed cytokine array experiments to analyze secretome pro ling. Each CM of HUVECs [i.e., control (Con), irradiated HUVECs (IR), CA-treated HUVECs (CA), and CA-treated irradiated HUVECs (IR + CA)] was applied to the cytokine array. Cytokine analysis revealed changes in several factors, including EGF, interleukin (IL)-6, and IL-8 (Fig. 5A). Although IL-6 and IL-8 levels signi cantly decreased in the IR + CA group compared to the IR group, there was no response to irradiation. The EGF level decreased in the IR group compared with Con group and was signi cantly upregulated in the IR + CA group compared with the IR group. A well-known growth factor, EGF plays critical roles in cell proliferation and protects the GI mucosa from a variety of insults [43,44]. EGF has also shown to regulate a number of GI functions, de ning its physiologic role in the GI tract [45][46][47][48]. The quantity of EGF, determined using an enzyme-linked immunosorbent assay (ELISA), was signi cantly increased 2-fold in the IR + CA group compared to the IR group (Fig. 5B). EGF-positive HUVECs increased in immuno uorescence in the IR + CA group compared to the IR group (Fig. 5D). The mRNA level of EGF was also decreased in the IR group, but its expression was recovered in the IR + CA group (Fig. 5C). These results suggest that CA treatment of irradiated endothelial cells induced EGF production and secretion.
A functional assay was performed by exposing recombinant EGF (rEGF) to determine whether CA-induced endothelial EGF secretion could ameliorate radiation-induced epithelium barrier dysfunction. The results indicated that the decreased TEER value in irradiated Caco-2 monolayers was increased by exposure to rEGF (Fig. 6A). An FITC-dextran assay indicated that FITC ux was elevated in media of irradiated Caco-2 monolayers, but it was diminished by rEGF treatment (Fig. 6B). Cell-cell contact strength was decreased in irradiated Caco-2 cell monolayers, but enhanced when rEGF was exposed to irradiated Caco-2 cell monolayers (Fig. 6C). Confocal staining revealed that immunohistochemical activities against ZO1 and DSG2 were diminished in the cellular junction of irradiated Caco-2 monolayers but were reinforced by rEGF treatment (Fig. 6D). The protein and RNA levels of ZO1 and DSG2 had the same pattern as the confocal staining result (Fig. 6E, F). EGF treatment of endothelial cells was also tested due to the possibility of an autocrine mode. The results indicate that rEGF treatment to HUVECs did not induce any mitigating effects like proliferation, anti-senescence, and angiogenic ability (supplement Fig. 1A-C). These results indicate that EGF, secreted by CA-treated irradiated HUVECs, reverts radiation-induced epithelial barrier dysfunction.
Endothelial-secreted EGF by CA treatment rescues radiation-induced barrier impairment with ZO1 and DSG2 regulation.
To examine whether EGF blockade in the CM of CA-treated irradiated HUVECs impeded the repair of epithelial barrier dysfunction, we abolished EGF in the CM of CA-treated HUVECs using a neutralizing antibody (anti-EGF). As shown in Fig. 7A, inhibition of EGF in CM of IR + CA HUVECs signi cantly reduced the TEER value compared with CM of IR + CA HUVECs treatment. Blocking of EGF also failed to decrease the FITC ux of CM of IR + CA HUVECs-treated irradiated Caco-2 monolayers (Fig. 7B). A dispase-based dissociation assay showed that reinforcement of cell-cell contact strength in the CM group of IR + CA HUVECs was abolished by neutralizing EGF (Fig. 7C). Similarly, expression of epithelial barrier-related molecules in the CM group of anti-EGF treatment did not increase as much as the CM group of IR + CA (Fig. 7D, E). Upregulated mRNA levels of ZO1 and DSG2 in the CM group of IR + CA HUVECs were also abolished by anti-EGF treatment (Fig. 7F). These results indicate that CA-induced endothelial EGF rescues radiation-induced epithelial barrier impairment with ZO1 and DSG2 regulation.
CM of CA-treated endothelial cells mitigates radiation-induced enteropathy with epithelial barrier restoration in mouse model.
To evaluate the therapeutic effect of CA-induced endothelial EGF on radiation-induced enteropathy, we administered the CM of IR + CA HUVECs to an irradiated mouse model. The mouse groups were as follows: control (Con), irradiated (IR), irradiated and injected with the CM of irradiated HUVECs (IR + CM), irradiated and injected with the CM of CA-treated irradiated HUVECs (IR + CA-CM), and irradiated and injected with rEGF (IR + rEGF). Histological examination revealed that villi shortening and crypt disruption by radiation were rescued in the IR + CA-CM groups. Elevated histological scoring in the IR group was signi cantly reduced in the IR + CA-CM and IR + rEGF groups (Fig. 8A, C). Otherwise, there were no signi cant differences in the IR and IR + CM groups (Fig. 8A, C). Immunoreactivity for Ki-67 as a proliferating marker was also increased in the IR + CA-CM and IR + rEGF groups than the IR + CM group (Fig. 8B). Physiological examination showed that the body weight of the IR + CA-CM group was higher than that of the IR group at days 5 and 6 following treatment (Fig. 8D). Of particular note, the immunohistochemical activity to Villin, Zo1, Dsg2, and Cldn3 was increased in the IR + CA-CM and IR + rEGF groups compared to the IR group (Fig. 8E). The mRNA levels, including Villin, Zo1, Dsg2, and Cldn3, in intestinal tissue were also increased in the IR + CA-CM and IR + rEGF groups compared to the IR group (Fig. 8F). These results suggest that CA-induced endothelial EGF e ciently alleviates radiation-induced enteropathy and rescues the barrier dysfunction.

Discussion
Radiation is currently used as a component of therapy for a wide range of malignant conditions. Although the threat of nuclear terrorism is rare, it can happen, so it is necessary to prepare a countermeasure for radiation-induced damage. Radiation-induced intestinal injury, due to sensitive organ to radiation, leads to severe side effects, including vomiting, diarrhea, bacterial infection, and septic shock-induced death. Radiation-induced enteropathy has increasing morbidity and mortality, and such conditions require development of therapeutic reagents, such as radiation-protector or radio-mitigator.
Despite advances in radio-protectors (e.g., amifostine for acute radiation syndrome), there are no promising agents for an effective radio-mitigator.
The potential medicinal plant CA is widely used in traditional medicine in the Orient and has been applied to skin lesions, ulcerations, and diarrhea [25,26]. In addition, its active constituents, primarily the main chemical components of pentacyclic triterpene derivatives (e.g., asiaticoside, asiatic acid, madecassoside, and madecassic acid have been reported to recover the damaged tissue [49]. Madecassoside has been reported to protect endothelial cells against oxidative stress [30], and asiaticoside has been reported to heal the incision through the formation of epithelial layer [50]. These reports indicate that CA is a promising reagent for the rescue of damaged tissues.
The effect of CA on survival rate against irradiation during clinical radiotherapy has been reported [51].
Treatment of Swiss albino with CA as a radioprotector at a sublethal dose of Co-60 gamma irradiation has been shown to prolong the survival rate [52]. Administration of CA has a dramatic radioprotection effect on radiation-induced body weight loss and conditioned taste aversion [53]. However, no studies have been reported on the effect of CA on radiation-induced enteropathy. In present study, we investigated the radio-mitigating effect of CA, focusing on crosstalk between endothelial and epithelial cells in vitro and in a mouse model. We found that CA ameliorates radiation-induced enteropathy through modulation of radiation-induced endothelial cell secretome. We also identi ed EGF as an endothelial cell driven-key regulator to repair radiation-induced epithelium disruption. Our ndings also demonstrate that endothelial-derived EGF by CA treatment improved the epithelial barrier damage on radiation-induced enteropathy.
Interactions between intestinal epithelial cells and the subepithelial cellular components play important roles in controlling intestinal barrier function under pathological conditions [10,54]. Studies have shown that crosstalk between endothelia and epithelial barrier is critical for regulation of tissue homeostasis and protection against pathogens or tissue damaging agent in human airways [55]. The endothelial-epithelial biochemical crosstalk pathway was studied using a human intestinal crypt cell line grown in noncontact co-culture with HUVEC. Endothelial cells secreted the 6-keto-prostaglandin F 1 alpha, a stable hydrolysis product of prostacyclin, resulting in epithelial cell activation through paracrine action [10]. In this study, we investigated the interactions between intestinal epithelial cells and endothelial cells in radiation exposure conditions. By applying the CM of CA-treated endothelial cells to irradiated epithelial cells, we demonstrated that the secretome of CA-treated endothelial cells could rescue the radiation-induced epithelial dysfunction. EGF, a well-known monomeric peptide present in the GI lumen, plays important roles in mitogenesis in tissue [56][57][58]. EGF and its related peptides have been implicated in the promotion of cell proliferation in wound healing, such as in re-epithelialization [59,60]. Furthermore, secreted EGF from bone marrow endothelial cells accelerates hematopoietic stem cell recovery [61]. It is well known that rEGF treatment promotes survival after radiation exposure [62] and protects radiation-induced enteropathy [63].
Otherwise, there is little information about the effects of EGF on radiation-induced epithelial barrier damage. In our recent study, CA-induced EGF secretion rescued the impaired epithelial barrier in irradiated Caco-2 monolayers and in enteropathy mouse model. The use of EGF neutralizing antibody reversed the relieving effect and failed to rescue epithelial barrier dysfunction. Taken together, these ndings indicated that CA-induced EGF was a modulator that contributed to recover the radiation-induced epithelial dysfunction. This is the rst evidence for a functional cellular response of CA on damaged tissue.
Breakage of the epithelium barrier integrity is one of the important characteristics of radiation-induced enteropathy. Gut epithelial barrier is the rst defense to protect the extra insults. It has been reported that the epithelial barrier damaged by radiation or in ammatory stimuli leads to downregulation of TEER and integrity, and fragmentation of cell-cell interactions [64,65]. Complexes of intercellular junctions, including TJs (e.g., ZO1, CLDN3), AJs, and desmosomes (e.g., DSG2), are the principal components of the intestinal barrier. In particular, ZO1 alteration contributes to disturbance of epithelial barrier. Loss of ZO1 with barrier dysfunction has been shown in dextran sulfate sodium (DSS)-induced colitis and sepsis in a pseudomonas aeruginosa infection mouse model [66,67]. Epithelial ZO1-de cient mice display severe mucosal damage with increased permeability following DSS application [68]. Also, DSG2 is required for the integrity of the intestinal epithelial barrier in vitro and in vivo [21,22]. Intestinal epithelial DSG2 knockout mice exhibit severe colitis from DSS treatment with increasing intestinal permeability [22]. In this study, CA-induced EGF increased expression of ZO1 and DSG2 in irradiated Caco-2 monolayers and intestinal epithelium of radiation-induced enteropathy. Taken together, upregulation of ZO1 and DSG2 by CA-induced EGF contributes to the recovery of epithelial barrier damage in irradiation.

Conclusions
We found that CA attenuated radiation-induced endothelial dysfunction in vitro, including proliferation, senescence, and tube formation activity. We have also shown therapeutic effects of CA on radiationinduced enteropathy, with the recovery of endothelial and epithelial dysfunction, focusing on the crosstalk between endothelial cells and epithelial cells. In particular, we identi ed EGF, a key factor secreted by endothelial cells to repair radiation-induced epithelial barrier dysfunction. Furthermore, by using a neutralizing anti-EGF antibody, we have shown the failure of the restoration of the radiation-induced epithelial barrier dysfunction and the related molecules expression in Caco-2 monolayers. The CM of CAtreated HUVECs or rEGF was administrated to a mouse model, and the results show recovery of radiationinduced epithelial dysfunction, including increased expression of epithelial barrier-related molecules. Thus, our study results suggest the use of CA as an effective radio-mitigator against radiation-induced enteropathy.

Materials And Methods
Cell culture and reagents HUVECs (Lonza, Basel, Switzerland) were cultured in EGM-2 medium supplemented with endothelial growth kit components (Lonza). Passage number of HUVECs used in experiments was between 4 and 7.
Human Caco-2 cells were maintained in Dulbecco's Modi ed Eagle Medium (DMEM, Gibco, Grand Island, NY, USA) supplemented with 10% fetal bovine serum (FBS, Gibco) and 1% antibiotics. All cells were grown in a humidi ed incubator at 37ºC with 5% humidity. Based on previous studies [41, 69], Caco-2 cells were grown into a con uent monolayer for in vitro experiments as a barrier function model. To obtain CM from HUVECs, irradiated HUVECs were either treated or not with CA (USP, MD, USA) for 24 h. After incubation, the media was exchanged with fresh serum-free EBM-2 media. Caco2 used in this study was between passage 18 and 33.

Animals
Speci c pathogen-free male C57BL/6 mice were obtained from Harlan Laboratories (Indianapolis, IN, USA) and maintained in speci c pathogen-free conditions at the KIRAMS animal facility. All mice were housed in a temperature-controlled room with a 12-h light/dark cycle. Food and water were provided ad libitum. The mice were acclimated for 1 week before commencement of the experiments and were grouped as follows: control (Con), irradiation (IR), irradiation with CA treatment (

CCK-8 assay
HUVECs were seeded in a 96-well plate. On the next day, cells were irradiated at 10 Gy and treated with varying concentrations of CA. After a 48-h incubation, the CCK-8 reagent was added and measured using a microplate reader at a wavelength of 450 nm. The experiments were carried out at least in triplicate.

β-Galactosidase assay
HUVECs were irradiated at 10 Gy using a 137 Cs γ-ray source (Atomic Energy of Canada, Ltd, Canada) with a dose rate of 3.25 Gy/min. Irradiated HUVECs were subsequently treated with CA for 48 h. Prepared cells were xed with 4% paraformaldehyde and subsequently stained using a β-galactosidase kit (Cell Signaling Technology, Danvers, MA, USA) according to the manufacturer's instructions.
Tube formation assay Irradiated HUVECs were re-seeded onto Matrigel-coated transwell (Corning, NY, USA) followed by treatment with or without CA for 6 h. Angiogenic ability was observed under a light microscope and plotted using Image J.

Histological analysis of the intestine
Mouse small intestinal tissue samples were xed with a 10% neutral buffered formalin solution, embedded in para n wax, and sectioned transversely at a thickness of 4 µm. The sections were then stained with hematoxylin and eosin (H&E). Evidence of intestinal mucosal injury was quanti ed (0 = none, 1 = mild, 2 = moderate, 3 = high) in H&E-stained sections of the ileum as a reference. The severity of radiation-induced enteritis was assessed by the degree of maintenance of the epithelial architecture, crypt damage, vascular enlargement, and in ltration of in ammatory cells in the lamina propria. This assessment is a modi cation of the histological score parameter used by Sung et al. [70]. To perform immunohistochemical analysis, slides were subjected to antigen retrieval and then treated with 0.3%

Immunocytochemical staining
Caco-2 monolayers on coverslips were harvested and immuno uorescence analysis was performed. Cells were xed with paraformaldehyde, blocked and permeabilized with 1% BSA and triton-X100 for 30 min at room temperature, and incubated with the primary antibodies speci c for ZO1 and DSG2. Samples were incubated for 1 h at room temperature with the Alexa Fluor 488 (green)-conjugated anti-rabbit IgG and Alexa Fluor 592 (red)-conjugated anti-mouse IgG (Thermo Fisher Scienti c) as secondary antibodies. After washing with DPBS, cells were count-stained with DAPI and mounted using Vectashield HardSet mounting medium. Fluorescence was examined using a confocal laser scanning microscope (LSM410; Carl Zeiss, Germany).

Bacterial translocation
To evaluate barrier function, treated mice were sacri ced, and the mesenteric lymph nodes were harvested under sterile conditions. The mesenteric lymph nodes were homogenized with sterile PBS and beads. The homogenized mixtures were centrifuged to remove cell debris and subsequently spread onto MacConkey agar (BD Biosciences). After incubation overnight, the colony-positive plates were counted.
Data were graphed as the percentage of individual mice exhibiting colonies compared to individual control mice.

Western blot
Cell lysates were washed with PBS and lysed in cold RIPA supplemented with a cocktail of protease and phosphatase inhibitor (Roche) on ice. Protein concentrations were determined by a bicinchoninic acid (BCA) method using Pierce BCA protein Assay (Thermo Fisher Scienti c). Equal quantity of samples mixed with sodium dodecyl sulfate (SDS)-containing sample buffer were boiled at 95°C for 5 min and separated by SDS-poly acrylamide gel electrophoresis. Proteins were transferred to polyvinylidene uoride for immunoblotting (Bio-rad). The membrane was blocked with 5% skim milk in C. Primary antibodies diluted in tris-buffered saline and Tween 20 (TBS-T) were incubated overnight at 4°C. The following antibodies were used: anti-ZO1 (Thermo Fisher Scienti c), anti-DSG2 (Abcam), and anti-β-Actin (Santa Cruz, CA, USA). Following overnight incubation, the membrane was washed and incubated with horseradish peroxidase-conjugated secondary antibodies for 1 h (Santa Cruz.) diluted in TBS-T. The membrane was washed, and proteins were detected using an enhanced chemiluminescence reagent (Pierce, Thermo Fisher Scienti c).

TEER measurement
Caco-2 cells were seeded into the upper chamber of transwell (0.4-µm pore size, Corning) and cultured for 21 days to form epithelial monolayers. Caco-2 monolayers were exposed to radiation and followed by treatment with various experimental conditions. The EVOM system (WPI, Sarasota, FL, USA) was used to measure TEER values.

FITC-dextran ux measurement
Caco-2 cells were seeded into the upper chamber of transwell inserts (0.4-µm pore size, Corning) and cultured for 21 days. Caco-2 monolayers in the transwell were irradiated and incubated under various experimental conditions with 500 µg/ml of FITC-dextran (Sigma-Aldrich, St. Louis, MO, USA). Media in the lower-chamber were taken after 48 h and uorescence was subsequently measured using a microplate uorescence reader (excitation at 450 nm and emission at 520 nm). The ux of FITC into the lowerchamber was calculated as a percentage corresponding to control sample.

Dispase-based dissociation assay
To evaluate cell-cell adhesive strength, Caco-2 monolayers were washed and incubated in dispase II (2.4 U/ml, Roche) and collagenase type I (Gibco) for 30 min. To apply a mechanical stress, the Caco-2 monolayers were carefully subjected to pipetting with an automatic pipet. Released single cells were observed a digital camera.
Human protein cytokine array HUVECs were irradiated and followed by CA or not in complete media. After 24 h, cells were washed once with PBS and exchanged to fresh serum-free medium. The CM of HUVECs were collected and spun down for removing cell debris. The CM was analyzed using the proteome pro ler™ Human Cytokine Array Kit (R&D systems, Minneapolis, MN, USA) according to the manufacturer's instructions. Densitometry was performed with Image J (National Institute Health) to determine the relative abundance of cytokines in the CM.

ELISA
To quantify EGF, the CM was collected and spun down to remove cell debris. The CM was subjected to ELISA (R&D Systems) according to the manufacturer's instructions.

Neutralization of EGF
In the neutralizing experiment, each CM sample was prepared as described above and incubated with 100 ng/ml of anti-EGF (R&D systems) for 1 h to bind the antibody. Caco-2 monolayers were washed with PBS and pre-incubated medium was added. Cells were incubated for 48 h and subsequently analyzed by additional assays.

Statistical analysis
The in vitro data was plotted as mean ± standard deviation of the mean, and animal data are plotted as the mean ± standard error of the mean. Statistical analyses were performed using one-way analysis of variance (ANOVA) with the Tukey's multiple comparison test. Values of P < 0.05 were considered statistically signi cant.

Consent for publication
All authors reached an agreement to publish this study in this journal.

Availability of data and materials
Data will be provided upon request.

Competing interests
There are no con icts of interest to declare.

38.
Tessner TG, Muhale F, Riehl TE, Anant S, Stenson WF: Prostaglandin E2 reduces radiationinduced epithelial apoptosis through a mechanism involving AKT activation and bax translocation.    demonstrating the mRNA levels of Villin, Tjp1, Dsg2, and Cldn3 in small intestine of each group. Data are presented as the mean ± standard error of the mean; n = 5 mice per group. *P < 0.05 compared to the Con group; #P < 0.05 compared to the IR group. Scale bars represent 100 μm.

Figure 4
The conditioned medium of Centella asiatica-treated HUVECs exerts recovery effect of radiation-induced epithelium dysfunction. (A) The transepithelial electronical resistance (TEER) value of Caco-2 monolayers on Transwell inserts was determined using the EVOM system. The TEER value of each group is shown in a bar graph. The groups are as follows; Conditioned medium (CM) of irradiated (IR)-HUVECs treating nonirradiated Caco-2 monolayers, CM of CA (Centella asiatica)-treated irradiated (IR+CA)-HUVECS treating non-irradiated Caco-2 monolayers, CM of IR treating irradiated Caco-2 monolayers, and CM of IR+CA treating irradiated Caco-2 monolayers. (B) The ux of FITC-dextran (4 kDa) in the lower-chamber was measured using a microplate uorescence reader (excitation at 450 nm and emission at 520 nm). The bar graph is shown as a fold of ux of uorescence normalized to CM of IR treating non-irradiated Caco-2 monolayers. (C) Dispase-based dissociation activity of each Caco-2 monolayer was determined. Treated Caco-2 monolayers were incubated in dispase II (2.4 U/ml) and collagenase type I for 30 min. After applying mechanical stress, the fragmentation of Caco-2 monolayers was observed using a digital camera. (D) The intensity of zonula occludens 1 (ZO1) and desmoglein 2 (DSG2) on intercellular junction of Caco-2 monolayers was assessed by confocal staining. Caco-2 monolayers on coverslips were stained with primary antibody against to ZO1 and DSG2. After mounting the samples, uorescence was examined using a confocal laser scanning microscope (Carl Zeiss). (E) Protein levels of ZO1 and DSG2 were assessed by western blot analysis. (F) mRNA levels of ZO1 and DSG2 were assessed by qRT-PCR.
Data are presented as the mean ± standard deviation of the mean; n = 3 per group. *P < 0.05 compared to CM of IR treating non-irradiated Caco-2 monolayers; #P < 0.05 compared to the CM of IR treating irradiated Caco-2 monolayers. Scale bars represent 10 μm. EGF-positive cells were observed by confocal laser scanning microscope. Data are presented as the mean ± standard deviation of the mean; n = 3, *P < 0.05 compared to the Con group; #P < 0.05 compared to the IR group. Scale bars represent 50 μm.