Abstract Umbilical cord-derived mesenchymal stem cells (UC-MSC) are promising candidates for wound healing. However, the low amplification efficiency of MSC in vitro and their low survival rates after transplantation have limited their medical application. In this study, we fabricated a micronized amniotic membrane (mAM) as a microcarrier to amplify MSC in vitro and used mAM and MSC (mAM-MSC) complexes to repair burn wounds. Results showed that MSC could live and proliferate on mAM in a 3D culture system, exhibiting higher cell activity than in 2D culture. Transcriptome sequencing of MSC showed that the expression of growth factor-related, angiogenesis-related, and wound healing-related genes was significantly upregulated in mAM-MSC compared to traditional 2D-cultured MSC, which was verified via RT-qPCR. Gene ontology (GO) analysis of differentially expressed genes (DEGs) showed significant enrichment of terms related to cell proliferation, angiogenesis, cytokine activity, and wound healing in mAM-MSC. In a burn wound model of C57BL/6J mice, topical application of mAM-MSC significantly accelerated wound healing compared to MSC injection alone and was accompanied by longer survival of MSC and greater neovascularization in the wound. Keywords: Burn wound healing, Microcarrier, Amniotic membrane, Umbilical cord-derived mesenchymal stem cells Graphical abstract [41]Image 1 [42]Open in a new tab Highlights * • A micronized amniotic membrane was fabricated as a microcarrier to amplify MSC ​in vitro. * • Micronized amniotic membrane could improve the survival time and therapeutic functions of MSC ​in vivo. * • Micronized amniotic membrane seeded with MSC greatly accelerated burn wound healing. 1. Introduction In accordance with a report by the World Health Organization in 2018, an estimated 180 ​000 deaths every year are caused by burns, and the vast majority occur in low-and-middle-income countries ([43]https://www.who.int/news-room/fact-sheets/detail/burns). Burn wounds consist of three distinguished zones of coagulation, stasis, and hyperemia (Jackson's burn wound model), and have four specific characteristics: more inanimate tissues, more exudates, more susceptible to infection, and easier to leave scars [[44]1,[45]2]. The pathophysiological process of burn wound healing is intricate and dynamic, including three interrelated and overlapping phases of hemostasis/inflammation, proliferation and remodeling [[46][3], [47][4], [48][5]]. In clinical practice, early surgical debridement of the inanimate tissues followed by grafting are still the gold-standard for deep extensive burn wound care [[49]6]. However, for patients with large-scale burns, the contradiction between the “supply” and “demand” of autologous skin transplantation remains the main obstacle. With the development and innovations in stem cell therapy and skin substitutes, stem cell-based biomaterials have been widely studied to promote burn wound repair and skin tissue regeneration, showing good application prospects. Mesenchymal stem cells (MSC) derived from adipose, bone marrow, human umbilical cord or other tissues, and their exosomes or extracellular vesicles have been recommended as potential candidates for the wound repair because of their therapeutic properties, including anti-inflammatory and immunomodulatory potentials [[50]7,[51]8], regenerative capability [[52]9], re-epithelialization [[53]10] and neovascularization [[54]11]. MSC transplantation can accelerate wound healing in all stages and prevent wound contracture and scar formation [[55]12]. According to the current literature, MSCs have been widely employed in preclinical animal studies and clinical trials, and most of them suggest that MSCs are safe and promising therapeutic means [[56][13], [57][14], [58][15], [59][16]]. A recent randomized clinical study demonstrated that adipose derived MSC therapy enhanced healing rate of skin wounds by increasing the granulation tissue coverage rate and thickness of granulation tissue [[60]17]. Another open-label single-arm clinical trial published in 2022 showed that skin-derived ABCB5^+ MSCs accelerated wound healing of diabetic foot ulcers by increasing capillary proliferation and vascularization [[61]18]. These studies imply that MSC therapy will be a hopeful and essential approach for skin wound healing in the future. However, several obstacles limit their practical applications. First, cells are traditionally cultured in two-dimensional plates, and this culture method often leads to slow proliferation, low activity, easy differentiation, and even loss of the therapeutic potency of MSC because the plates cannot mimic the microenvironment in which stem cells live and proliferate in vivo [[62][19], [63][20], [64][21]]. Second, during the delivery of MSC, cells are generally injected intravenously for systemic distribution or are injected locally into target tissues, and shear stress during injection may damage the cells [[65]22]. Third, because of the more inanimate tissues and exudates in the burn wound microenvironment, it is not beneficial for the survival and proliferation of stem cells in the harsh wound environment [[66]23]. Consequently, stem cells have difficulty in performing their therapeutic functions. To overcome these obstacles, additional strategies must be employed to achieve optimal application of MSC. Microcarriers with 3D culture systems provide new ideas to address the aforementioned problems occurring in traditional cell culture, delivery, and therapeutic processes. This method has undergone several innovations to satisfy different demands [[67]24]. Microcarriers can provide larger and more surface matrices for MSC to adhere to and proliferate compared to 2D plates [[68]25]. Tan et al. verified that MSC seeded in CultiSpher S microcarriers proliferate quickly and with high cell viability [[69]26]. Moreover, this 3D cell culture strategy can contribute to the longer survival and more effective therapeutic applications of MSC [[70]19,[71]22,[72]27]. Kim et al. showed that the secretion and bioactivity of UC-MSC-derived extracellular vesicles could be enhanced by 3D culture [[73]28]. Zeng et al. discovered that preformed gelatin microcryogels, as injectable cell carriers, enhanced the survival of MSC in vivo and accelerated skin wound healing [[74]29]. Researchers have used synthetic polymers, such as polystyrene, plastic, or glass, as microcarriers to increase the expansion efficiency of stem cells; however, these microcarriers pose challenges in the downstream process [[75]30]. For example, these synthetic materials cannot degrade in vivo; therefore, stem cells need to be digested to separate them from the microcarriers before use, resulting in the loss of cell viability. To solve this problem, it is necessary to investigate new microcarriers that can not only have excellent biocompatibility to support the proliferation of MSC but can also be transported directly into wounds. The human amniotic membrane can meet these requirements and is expected to be a good microcarrier for MSC. The amniotic membrane contains type I-VII collagen, laminin, integrins, and other ECM components that can mimic the in vivo microenvironment, providing a “fertile soil” for stem cells to live and proliferate. In addition, the human amniotic membrane, a natural tissue from the placenta with excellent biocompatibility, can be applied directly in vivo, so that MSC need not be digested before use. Furthermore, the amniotic membrane plays a coordinated role with MSC to promote burn wound healing. The properties of the human amniotic membrane in promoting wound healing, skin tissue reconstruction, and regulating inflammation and microbial growth have been widely recognized by hundreds of clinical trials or pre-clinical papers, including the domains of burns and chronic ulcer treatment, peritoneal reconstruction, tendon repair, microvascular reconstruction, and corneal repair [[76][31], [77][32], [78][33], [79][34], [80][35], [81][36], [82][37], [83][38], [84][39], [85][40]]. In our previous studies, we made great progress in upgrading the preparation methods for human amniotic membranes and found that decellularized amniotic membranes could be used to construct natural microcarriers for stem cells. Zheng et al. found that cryopreserved living micronized amniotic tissue can modulate the local microenvironment of diabetic wounds, thereby accelerating healing in mice [[86]41]. Ji et al. prepared a cell niche for epidermal stem cells using an engineered human amniotic membrane, which acted as a dermal scaffold for healing and regeneration of full-thickness skin defects in mice [[87]42]. In this study, we proposed the use of a natural human amniotic membrane to fabricate microcarriers called micronized amniotic membrane (mAM) to amplify UC-MSC in a 3D culture system. The mAM microcarrier was used to investigate the amplification efficiency and maintenance of MSC function in vitro. Finally, after applying mAM-MSC to burn wounds, their functions and mechanisms in accelerating wound repair were explored. mAM not only has the general characteristics of microcarriers, such as a large specific surface area, but also provides the natural cell niche to simulate the in vivo cell growth microenvironment, thus promoting the adhesion, growth, and the biological functions of MSC. In addition, the mechanical properties of mAM allow MSC to be conveniently delivered and glued onto wounds by just smearing mAM-MSC on the wound surface without injection, reducing the harm caused by shear stress from injection. The natural cell niche provided by mAM could protect MSC from tolerating ischemic and hypoxic conditions in the burn wound after transplantation, thus promoting the survival of MSC. 2. Materials and methods 2.1. Study design and ethical considerations The objectives of this study were to analyze the impact of mAM as a microcarrier to amplify MSC in vitro and used mAM and MSC complexes to repair burn wounds. All in vitro experiments were divided into 2D-cultured MSC (MSC) group and 3D-cultured MSC group (mAM-MSC) to compare the differences of cell proliferation, growth and paracrine function. For in vivo experiments, 6–8 weeks C57BL/6 mice were used with 12 mice per group to ensure statistical power and deep second-degree burn wounds were created on each side of the dorsal skin. Number of repeats was specified in each figure legend. This study was approved by the Ethics Committee of Changhai Hospital, Shanghai, China, and conducted according to the animal use and care committee of Changhai Hospital. 2.2. Fabrication of mAM All procedures were approved by the Ethics Committee of Changhai Hospital (Shanghai, China), and informed consent was obtained from all donors. Placental tissue from healthy donors was obtained within 24 ​h of cesarean section. First, the amniotic and chorionic membranes were separated using blunt separation. Then the amniotic membrane was washed with phosphate-buffered saline (PBS) and stored in PBS containing 10% dimethyl sulfoxide (DMSO) at −80 ​°C. The cellular and DNA components of the amniotic membrane were eliminated by three repetitive freeze-thaw cycles and digestion with 1 ​mg/mL DNase (Sigma, USA) using our previously described methods [[88]42]. Decellularized amniotic membrane was homogenized into microparticles about 400 ​μm in size by a peeler and freeze-dried to produce mAM. 2.3. Isolation and culture of UC-MSC The obtained umbilical cords were washed with PBS and cut into 2 ​cm long pieces. The umbilical vein was peeled off from each piece, together with the two arteries. The remainder is mainly Wharton's jelly. Wharton's jelly was cut into blocks approximately 1 ​mm in size and placed in a culture dish. The dish was placed in an incubator for 2 ​h, and then complete medium was dropped slowly. The tissue was left undisturbed for 7 days in a 37 ​°C incubator with 5% CO[2]. Subsequently, cells could be seen migrated out of the tissue blocks. When cells reached 80–90% confluence, the primary cells were collected and passaged. Passage 3–5 ​cells were used in subsequent experiments. 2.4. Construction of mAM-MSC A 3D culture system was used to construct mAM-MSC. 30 ​mL of MSC suspension (5 ​× ​10^4 ​cells/mL) together with 300 ​mg mAM were added to the 3D culture system, and the rotation mode was set at 35 ​rpm for 5 ​min, 0 ​rpm for 1 ​h, for 24 cycles on day 0 and changed to a constant speed mode of 40 ​rpm after day 0. Subsequently, the samples were harvested for scanning electron microscopy (SEM), cytoskeleton staining, Cell Counting Kit-8 (CCK8) assay, live/dead staining, RT-qPCR, and RNA sequencing (RNA-seq). 2.5. Scanning electron microscopy (SEM) The mAM-MSC in 3D culture system on day 3 post seeding was collected, rinsed three times in PBS, fixed in 2.5% glutaraldehyde for 2 ​h, and then cleaned once more with PBS. The mAM-MSC and mAM specimens were sprayed with gold and observed by SEM after dehydration with ethanol and tertiary butyl alcohol. Images of the two groups were captured using a scanning electron microscope (TM4000PLUS II, Hitachi, Japan). At least three replicates were used for each experiment. 2.6. Cytoskeleton staining Cytoskeleton staining was used to visualize the cell morphology and structure. The mAM-MSC cultured for 3 and 7 days were collected and stained with phalloidin (Phalloidin-iFluor 594, Abcam, USA). First, mAM-MSC were fixed with paraformaldehyde (4%, PFA) for 10 ​min and then permeabilized with Triton X-100 (0.5%) for 5 ​min. After cleaning mAM-MSC with PBS, they were incubated with phalloidin for 30 ​min at room temperature in the dark. Finally, a DAPI solution (Wako Pure Chemical Industries, Japan) was used to stain the cell nuclei. A confocal microscope (Olympus, Tokyo, Japan) was used to observe fluorescence. 2.7. Viability and proliferation detection of mAM-MSC To verify the viability and proliferation of mAM-MSC in the 3D culture system, CCK-8 assay and live/dead staining were conducted. A LIVE/DEAD Viability/Cytotoxicity Kit (Invitrogen, USA) was used to detect the cytoviability of MSC on mAM using fluorescence microscopy (Leica, Germany). mAM-MSC samples were collected from the 3D culture system on days 1, 3, and 7 for the cell proliferation assay using a CCK-8 kit (Dojindo, Japan) according to the manufacturer's instructions. The control group consisted of MSC cultured in 12-well plates, with 1 ​mL MSC suspension (5 ​× ​10^4 ​cells/mL) added to each well. For the mAM-MSC group, 30 ​mL of MSC suspension (5 ​× ​10^4 ​cells/mL) and 300 ​mg of mAM were co-cultured in a 3D culture system. 1 ​mL of suspension was drawn from the 3D culture system and transferred into one well of a 12-well plate. One hundred microliters of CCK8 were added to each well of both groups for incubation of up to 2 ​h. Subsequently, 100 ​μL of supernatant was aspirated from each well and put in wells of a 96-well plate. Absorbance was measured at 450 ​nm using a microplate reader (Biotek, USA). Three replicates were analyzed for each treatment group at each time point. 2.8. Transcriptomic analysis The mAM-MSC samples cultured for 3 days in the 3D culture system were collected and marked as the mAM-MSC group (n ​= ​3), whereas MSC cultured in 12-well plates were used as the control and marked as the MSC group (n ​= ​3). Total RNA was extracted from each sample using TRIzol reagent (Invitrogen, USA). RNA libraries were constructed, and transcriptomic sequencing was performed by OE Biotech Co., Ltd. (China). For data analysis, DEGs were identified using DESeq. P value ​< ​0.05 and fold change >1.5 or <0.5 were considered significant. Gene ontology (GO) enrichment, Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analysis, and Gene Set Enrichment Analysis (GSEA) were performed for in-depth bioinformatics analysis. 2.9. Quantitative real-time PCR TRIzol reagent (Invitrogen, USA) was used to extract RNA from 2D-cultured MSC, mAM-MSC, and mouse burn wound tissues according to the manufacturer's instructions. M-MLV reverse transcription reagents (Takara, Japan) were used to reverse transcribe the cDNA. Real-time PCR was performed on a CFX96 (Bio-Rad, USA) using an SYBR premix qPCR kit (Takara, Japan). The primer sequences are presented in [89]Table 1. Table 1. Primer sequences. Gene Forward (5′-3′) Reverse (5′-3′) Human VEGF CAGGGAAGAGGAGGAGATG CTGGGTTTGTCGGTGTTC Human TGFB1 ACCTCGGCTGGAAGTGG CCGGGTTATGCTGGTTGT Human IGF1 CCCTGGACAATCAGACGAA GTCGCAAACCGAACAGC Human ANGPT1 CAATGGTCTCACGTTCTCAC CCCGTAAGTCAGATGTTGTTT Human GAPDH GGTCACCAGGGCTGCTTTTA GGATCTCGCTCCTGGAAGATG Mice VEGF AGCTTCCTACAGCACAGCAG CACAGTGAACGCTCCAGGAT Mice TGFB1 TTGGCCATGAATTTGACCTC ACATCAGTCTCATTCACAGC Mice ANGPT1 GGAGTCGGCATGAATCGCT GAATGGGATCCCCCTCGG Mice HGF TACTCAGCGTCACAGCATGG GCAGGACCGGCACAAGTC Mice IGF1 GCTGCTGGCTTCTAAGTGTG ACTGCCCAGTTCGTTTCAG Mice MMP8 CAACCCCATCCAACCTACT GAAGGGCCAGAACAGAGAT Mice GAPDH GGTCGGTGTGAACGGATTTG TGTAGACCATGTAGTTGAGGTCA [90]Open in a new tab VEGF: vascular endothelial growth factor; TGFB1: transforming growth factor-b1; IGF1: insulin-like growth factor-1; ANGPT1: angiopoietin-1; HGF: hepatocyte growth factor; MMP8: matrix metallopeptidase 8. 2.10. Evaluation of the biological effects of mAM-MSC on HUVEC The biological effects of mAM-MSC on HUVEC, including migration and angiogenesis, were evaluated using transwell and tube formation assays. The mAM-MSC was cultured in 3D culture system firstly with complete medium (DMEM containing 10% FBS) for 72 ​h and then switched to serum-free DMEM. After 48 ​h, the supernatant was collected, centrifuged, and filtered for subsequent experiments. Supernatants of the 2D-cultured MSC were collected in the same manner. DMEM was used as the negative control. 2.10.1. Cell migration assay A 24-well transwell chamber (8 ​mm pore size, Corning, USA) was used to conduct the cell migration assay. The bottom chamber was filled with 600 ​μL supernatant of 2D-cultured MSC, supernatant of mAM-MSC, and DMEM (negative control group, NC group), in that order. HUVECs (1 ​× ​10^5 HUVECs diluted in serum-free high-glucose DMEM) were seeded in the upper chamber. After 24 ​h, cells in the upper layer of the membrane were wiped with swabs, and the migrated cells were stained with 0.1% crystal violet (Sigma, USA) after fixation with 4% PFA. Stained cells were observed under a light microscope (Leica, Germany), and cells from at least three wells were counted using ImageJ software (NIH, USA) for each group. 2.10.2. Tube formation assay Fifty microliters of Matrigel (5 ​mg/mL, BD Biosciences, USA) were added to each well of a 96-well plate and allowed to solidify at 37 ​°C. Then, 4 ​× ​10^4 HUVECs resuspended in 100 ​μL supernatant of 2D-cultured MSC, supernatant of mAM-MSC, and DMEM, in order, were seeded on the Matrigel of each well. After 4 ​h, cells were observed under a light microscope (Leica, Germany) and photographed. Vascularization was reflected by counting the number of nodes and the total length of the tubes. 2.11. Preparation of Luc ​^+ ​MSC Lentiviruses expressing luciferase (Luc) from Hanbio Tech (China) were used to infect MSC when the cells reached 30–50% confluency in a 6-well plate. At the same time, polybrene (Hanbio Tech, China) was added into each well to reach an ultimate concentration of 6 ​μg/mL, and then, the plate was shaken and incubated overnight at 37 ​°C. After incubation, the medium containing lentiviruses was removed, and complete medium was added. After amplification for one passage, positive cells (Luc^+) were collected and purified using puromycin (Sigma, USA). Luc^+ MSC from passage 4–6 were utilized for bioluminescence imaging experiments. 2.12. Burn wound model and treatment Male C57BL/6 mice (aged 6–8 weeks) were obtained from SLAC Laboratory Animal Co., Ltd. (Shanghai, China), and animal experiments were approved by the Institutional Animal Care and Use Committee of the Naval Medical University, Shanghai, China. Mice were randomly and equally divided into four groups: mAM-MSC, mAM, MSC, and sham. After isoflurane inhalation anesthesia and hair removal, a deep second-degree burn wound was created on each side of the dorsal skin. In brief, a copper rod (1 ​cm in diameter, heated to 100 ​°C) was applied to burn the dorsal skin of the mice for 5 ​s [[91]43]. After 48 ​h, a wound 1 ​cm in diameter was formed after the inanimate tissue was surgically removed from healthy skin, and this time point was marked as day 0. A donut-shaped silicone sheet was interrupted to the wound edge by 4–0 sutures to prevent contraction of the skin surrounding the wound, so that the wound healing rate could be monitored and measured objectively. The mAM-MSC and mAM were spread on the wound surface in the mAM-MSC and mAM groups, respectively. The number of cells in the mAM-MSC group was calculated using a NucleoConuter NC-200 (ChemoMetec, Denmark) [[92]44,[93]45], and the cell number in mAM-MSC was adjusted to 5 ​× ​10^5. In the MSC group, cells resuspended in 100 ​μL PBS were subcutaneously injected around each wound at four sites. The number of cells applied to each wound site was 5 ​× ​10^5. All wounds were covered with a transparent silicone film to prevent drying. The bandage integrity was assessed daily. Photographs of each wound in the four groups were taken at every time point, and the remaining wound area was calculated using ImageJ software (NIH, USA). The remaining wound area (%) ​= ​S[r]/S[i] ​× ​100%, where S[i] is the initial wound area on day 0, and Sr is the remaining wound area on days 3, 7, and 11. On day 7 and day 11 after treatment, five mice in each group were sacrificed, and the wound area was cut for further RT-qPCR, H&E staining, and immunohistochemical staining of CD31. 2.13. Bioluminescence imaging The survival time of transplanted mAM-MSC (Luc^+) in mouse burn wounds was assessed using bioluminescence imaging. Luc^+ MSC resuspended in PBS was used as the control. After anesthesia, mice were injected intraperitoneally with 10 ​μL D-luciferin (150 ​mg/mL in PBS, Yeasen Biotechnology, China) per gram of mouse weight. Images were obtained with a 3-min exposure to a small animal in vivo imaging system (IVScope 8500, Clinx Science Instruments Co., Ltd, China). Images were acquired on days 0, 3, 7, and 9. Bioluminescent photons were measured in each wound area using the Clinx IVScopeEQ Capture software. 2.14. Histological analysis Hematoxylin and eosin (H&E) staining and CD31 immunohistochemical staining were performed to observe re-epithelialization and vascularization of the burn wound. The wounds together with 2 ​mm of the surrounding healthy skin edge, were excised and fixed with 10% PFA overnight at 4 ​°C. After embedding in paraffin, the tissues were cut into sections and stained with an H&E staining kit (Beyotime, China) and a primary antibody against CD31 (1:50, Abcam, USA) to visualize angiogenesis. Re-epithelialization was measured at day 11 by H&E staining. Digital images were obtained and analyzed using Image-Pro Plus Software. Wound edge was calculated by tracing the distance between the leading edges of epithelium within the wound. 2.15. Statistical analysis All data were statistically analyzed with SPSS 20.0 and presented as mean ​± ​standard deviation (SD). GraphPad Prism 8 software was used for statistical analyses and generation of figures. Differences between two groups were analyzed by two-tailed Student's t-test. For multiple comparisons, one-way analysis of variance (ANOVA) and Tukeys multiple comparisons test were used. P ​< ​0.05 was considered as statistically significant. 3. Results 3.1. Fabrication of mAM The mAM was fabricated using our previously described protocol [[94]42]. The general view of the cleaned amniotic membrane was presented in [95]Fig. 1A. The fluorescent images of DAPI indicated that the residual DNA content of the amniotic membrane was eliminated completely after three repetitive freeze-thaw cycles and DNA enzymatic digestion ([96]Fig. 1B). After being decellularized and cut into microparticles, the mAM appeared as white translucent quadrilaterals, with the size of about 400 ​μm, as measured by bright field image ([97]Fig. 1C). Fig. 1. [98]Fig. 1 [99]Open in a new tab Fabrication of mAM. (A) General view of the cleaned amniotic membrane. (B) DAPI images of amniotic membrane before and after decellularization. (C) Bright field images of mAM at different magnifications. 3.2. Characteristics of mAM-MSC The SEM images of mAM and mAM-MSC showed obvious differences in surface topography ([100]Fig. 2A). Before co-culturing with MSC, the surface of the mAM was smooth and flat. When co-cultured with MSC for 3 days, the surface of the mAM was attached to the MSC, which maintained a spindle-like stretched shape. Cytoskeleton staining was conducted when mAM-MSC was cultured for 3 or 7 days. [101]Fig. 2B shows that the cytoskeleton was dyed red and had a long spindle shape, extending through the whole cell; however, there were no obvious morphological differences between days 3 and 7. In the CCK8 assay, compared with the control group (traditionally 2D-cultured MSC), OD values in the mAM-MSC group were higher from day 1 to day 7, with an ascending trend (all P ​< ​0.05, [102]Fig. 2C). As shown in [103]Fig. 2D, the live/dead staining results showed that MSC adhered to the surface of mAM in the 3D culture system on days 1, 3, and 7, and more cells were detected on mAM after prolonged culture times, especially on days 3 and 7. More importantly, nearly no dead cells, which stained red, were found on mAM during the 7-day culture period. Fig. 2. [104]Fig. 2 [105]Open in a new tab Characteristics of mAM-MSC. (A) SEM images showing the morphology of mAM and mAM-MSC at different magnifications. Red arrows indicate microscopic MSC. (B) Immunofluorescence images of phalloidin indicating the cytoskeleton of MSC on mAM cultured for 3 and 7 days. (C) Comparing cell proliferation rate of MSC in traditional 2D plate and on mAM by CCK8 assay. (D) Live/dead staining of MSC seeded on mAM on day 1, 3 and 7 using fluorescence microscopy. ∗P ​< ​0.05 and ∗∗P ​< ​0.01. (For interpretation of the references to colour in this figure legend, the