Multipotent Adult Progenitor Cells Support Lymphatic Regeneration at Multiple Anatomical Levels during Wound Healing and Lymphedema

Lymphatic capillary growth is an integral part of wound healing, yet, the combined effectiveness of stem/progenitor cells on lymphatic and blood vascular regeneration in wounds needs further exploration. Stem/progenitor cell transplantation also emerged as an approach to cure lymphedema, a condition caused by lymphatic system deficiency. While lymphedema treatment requires lymphatic system restoration from the capillary to the collector level, it remains undetermined whether stem/progenitor cells support a complex regenerative response across the entire anatomical spectrum of the system. Here, we demonstrate that, although multipotent adult progenitor cells (MAPCs) showed potential to differentiate down the lymphatic endothelial lineage, they mainly trophically supported lymphatic endothelial cell behaviour in vitro. In vivo, MAPC transplantation supported blood vessel and lymphatic capillary growth in wounds and restored lymph drainage across skin flaps by stimulating capillary and pre-collector vessel regeneration. Finally, human MAPCs mediated survival and functional reconnection of transplanted lymph nodes to the host lymphatic network by improving their (lymph)vascular supply and restoring collector vessels. Thus, MAPC transplantation represents a promising remedy for lymphatic system restoration at different anatomical levels and hence an appealing treatment for lymphedema. Furthermore, its combined efficacy on lymphatic and blood vascular growth is an important asset for wound healing.

media as reference condition, both supplemented with 1% FBS. After 48 hours, cells were fixed with 100% methanol, permeabilised with Triton-X-100 and stained overnight with anti-Ki67 primary antibody (Supplementary Table S2). The day after, cells were washed, incubated with Alexa488conjugated secondary antibody and Hoechst for nuclear staining. Proliferation index was determined by counting the fraction of Ki67 + cells of the total (Hoechst + ) cells in 5 randomly chosen fields taken per replicate with a Zeiss MRm camera mounted onto an Axiovert200M microscope and equipped with Axiovision 4.8 software, at 20x magnification (using an LD Plan-Neofluar objective lens, NA 0.4) by a blinded observer.
Lymphatic endothelial cell migration. To estimate the effect of MAPC-conditioned media on lymphatic endothelial cell migration, a Boyden chamber assay was performed. Briefly, transwell inserts (containing polycarbonate filters with 8 μm pore size; Costar, Corning) were coated overnight with 0.2% gelatin. The bottom compartment of a 24-well plate was filled with 0.3 ml non-conditioned media or with 0.3 ml of mMAPC-or hMAPC-conditioned media. Following rehydration for 1 hour with deionised water, inserts were placed into the 24-well plate and each was loaded with 0.3 ml EGM-2-MV/0.5% FBS containing 5x10 4 lymphatic endothelial cells. Following incubation for 24 hours at 37°C/5% CO 2 , cells were fixed in methanol for 30 minutes at -20°C. Next, cells were stained with Wright-Giemsa's staining solution (Sigma WG32) for 7 minutes and rinsed with deionised water for 10 minutes. Inserts were lifted and cells on the upper side of the membranes were removed by gentle rubbing using a cotton swab. Pictures of the inserts were taken with a Zeiss MRc5 camera mounted onto an Axiovert200M microscope and equipped with Axiovision 4.8 software, and transmigrated cells were manually counted in 3 random fields per insert at 20x magnification (using an LD Plan-Neofluar objective lens, NA 0.4) by a blinded observer.
Lymphatic endothelial cell sprouting. To test the effect of mMAPC-conditioned media on lymphatic endothelial cell sprouting, lymphatic endothelial cell spheroids were allowed to form by applying 25 μl droplets (containing 1,000 lymphatic endothelial cells in a 20% methylcellulose/EGM-2-MV mixture) onto non-attachment plates and incubating them upside down at 37°C/5%CO 2 . The next day, spheroids were carefully washed in PBS/2%FBS, collected by gentle centrifugation, carefully resuspended in methylcellulose/FBS/collagen (Purecol Advanced Biomatrix) and seeded into 24-well plates (0.5 ml/well). Following incubation of 30 minutes at 37°C/5% CO 2 , 0.5 ml mMAPC-conditioned media (1:1 mix with serum-free lymphatic endothelial cell media) or 100% serum-free lymphatic endothelial cell media as reference condition was added on top of the collagen/spheroid gel. Pictures were taken 24 hours later at 20x magnification (using an LD Plan-Neofluar objective lens, NA 0.4) with a Zeiss MRm camera mounted on a Zeiss Axiovert200M microscope and the number of sprouts per spheroid was determined by manual counting by a blinded observer.
Antibody array, RNA isolation, cDNA preparation, quantitative (q)RT-PCR and flow cytometry Antibody arrays were purchased from R&D Systems (ARY015 for mouse; ARY007 for human; cytokine/growth factors represented in the arrays are listed in Supplementary Table S1 online) and run according to the manufacturer's instructions. Briefly, protein content in the 72 hour (non-)conditioned media was determined by BCA assay and equal amounts of protein were used for all conditions. Following overnight incubation, the signals of the retained proteins were revealed by a luminol-based detection reaction using a ChemiDoc XRS+ molecular imager (Bio-Rad) and quantified using Image Lab software (version 4.0, Bio-Rad Labs). For quantification, arrays of the conditioned media (n=3) and the corresponding non-conditioned media (n=1) were developed together and mean pixel density in every spot of the array was determined. The D mean pixel density was calculated by subtracting the value of the non-conditioned media from the corresponding values of the conditioned media, averaged and expressed in arbitrary units. To determine % overlap between species, we determined for those proteins that were common analytes on both arrays whether or not there was a signal detected above that of the non-conditioned media.
Total RNA from cell lysates was extracted using TRIzol reagent (Invitrogen). mRNA was reverse transcribed using Superscript III Reverse Transcriptase (Invitrogen) and cDNA underwent 40 amplification rounds on an ABI PRISM7700 cycler PerkinElmer/Applied Biosystems) for SYBR-Green-based qRT-PCR, as described 5 . Primer sequences for qRT-PCR are listed in Supplementary   Table S3. mRNA levels were normalised using GAPDH as house-keeping gene. To analyse LYVE1 expression on the surface of differentiated mMAPCs, cells were harvested by gentle trypsinisation, washed with FACS staining buffer (PBS + 1 mmol/L EDTA + 25 mmol/L Hepes + 1% BSA) and incubated with primary antibody (Upstate, 07-538) or the corresponding rabbit IgG isotype for 20 minutes at room temperature in the dark. After washing with FACS buffer, cells were incubated with biotinylated goat-anti-rabbit secondary antibodies for 20 minutes at room temperature in the dark.
Next, samples were washed and incubated in the dark for 20 minutes with allophycocyanin (APC)labelled streptavidin. To select for viable cells, 7-AAD was added 10 minutes before running the samples on a FACS Aria I (Beckton Dickinson) for analysis.

Mouse models
As MAPCs do not express Major Histocompatibility Complex-I and -consequently -are sensitive to natural killer cell-mediated clearance, all mice (including PBS controls) were injected intraperitoneally with 20 µl anti-asialo GM1 antibodies (Wako Chemicals, Osaka, Japan; 20x diluted in PBS) 1-2 hours before transplantation and every 10 days thereafter. These antibodies selectively eliminate natural killer cells without affecting macrophage or lymphocyte function 6 .
Linear wound model: At day 0, a 12-mm linear skin incision was inflicted with a scalpel on the back of 12 week-old C57Bl/6 male mice after they were anaesthetised with a mixture of 100 mg/kg ketamine and 10 mg/kg xylazine. Immediately after wounding, mice were injected in the muscle fascia underneath the skin wound with 1x10 6 mMAPCs (resuspended in PBS) or PBS alone divided over three equally spaced injection spots. To avoid wound infection, mice were housed individually in cages without bedding. Wound dimensions were measured daily under isoflurane anaesthesia using digital calipers (VWRI819-0012, VWR) and pictures were taken using a NikonD1 camera and Camera-Control-Pro software. At day 4, bright field and fluorescence pictures of the wound area were taken with a Zeiss MRc5 camera mounted on a Zeiss Lumar microscope (using a Neolumar S lens, 0.8x magnification, FWD 80 mm). At day 10, mice were euthanised, the residual skin wound and underlying muscle tissue were dissected out, fixed in zinc-paraformaldehyde and prepared for embedding in paraffin or optimal cutting temperature (OCT) and sectioning.
Atropine (0.01 mg/kg) was administered intraperitoneally as premedication. Under sterile and temperature-controlled (37°C) conditions, standardised full-thickness wounds were made with a 0.5 cm biopsy puncher (Stiefel Laboratories, Offenbach am Main, Germany) on the back of the mouse in the mid-dorsal region. A silicone ring was fixed (using Histoacryl tissue adhesive, Braun, Diegem, Belgium) and sutured around the wound and wounds were treated with PBS or 5x10 5 hMAPCs. In a separate subset of mice, hMAPCs were transduced with an eGFP-coding lentivirus prior to transplantation. An occlusive dressing (Tegaderm, 3M, Diegem, Belgium) was used to keep the wound moist. All wounded mice were housed individually to avoid fighting and to prevent removal of the occlusive wound dressing. Every other day, the occlusive dressing was renewed under isoflurane anaesthesia and pictures were taken using a NikonD1 camera and Camera-Control-Pro software. Wound size was measured using Image J software and was expressed as the % versus the size at day 0 for each individual mouse. At 5 or 10 days after wounding, mice were euthanised and square skin fragments including the circular wound area and a rim of normal skin were dissected out, rinsed in PBS and post-fixed overnight at 4°C using zinc-paraformaldehyde. Following fixation, skin fragments were separated in two equal pieces at the midline of the wound and processed for paraffin or OCT embedding and sectioning.
Skin flap model: At day 0, 12 week-old athymic nude Foxn1 male mice (Harlan) were anaesthetised with an intraperitoneal injection of ketamine (100 mg/kg) and xylazine (10 mg/kg). The lymphatic network in the abdominal skin was severed by elevating an epigastric skin flap and suturing it back to its original position, as previously described 7 . Continuous blood supply to the flap was ensured by retaining a vascular pedicle including the right inferior epigastric artery and vein (Fig.  3a). One day after resuturing the skin flap, 1x10 6 mMAPCs, 1x10 6 hMAPCs or PBS (divided over 4 injection spots; Fig. 3a) were injected around the wound edges. Two or 4 weeks later, the axillary regions were exposed and axillary lymph node drainage was monitored by microlymphangiography for 15 minutes after intradermal injection of 10 μl FITC-dextran (mol wt 2,000 kDA, Sigma-Aldrich; hMAPCs) or 10 μl Rhodamin-B-isothiocyanate-dextran (mol wt 70 kDA, Sigma-Aldrich; mMAPCs) under the wound border (Fig. 3a). Bright field and fluorescence pictures were taken at 15 minutes with a Zeiss MRc5 camera mounted onto a Zeiss Lumar microscope (using a Neolumar S lens, 0.8x magnification, FWD 80 mm). Mice were subsequently euthanised, the skin wound area around the cell engraftment/microlymphangiography areas excised, fixed and processed for paraffin or OCT embedding and sectioning.
Lymph node transplantation model: At day 0, 12 week-old athymic nude Foxn1 female recipient mice (Harlan) were anaesthetised with an intraperitoneal injection of ketamine (100 mg/kg) and xylazine (10 mg/kg). To visualise the lymph nodes, the right axilla region was exposed and mice were injected with a 3% Evans Blue solution in the palm of the right paw after which lymph nodes were removed along with the surrounding lymphatic (collector) vessels. A pocket just caudal of the axillary vessels, aligned by the lateral axillary fat pad, the M. pectoralis and the M. latissimus dorsi was prepared. Donor lymph nodes were dissected from mice ubiquitously expressing DsRed (B6.Cg-Tg(CAG-DsRed*MST)1Nagy/J; for mice receiving hMAPCs or PBS and followed up for 4 or 8 weeks) or eGFP (C57Bl/6-Tg(CAG-EGFP)1Osb/J; for mice receiving hMAPCs or PBS and followed up for 4 or 16 weeks) and cut in two halves through the hilus. The cut lymph node was subsequently implanted into the recipient pocket (hilus oriented medially and cut surface facing upwards) and fixed in place with two permanent sutures (using 9-0 nylon non-absorbable suture, Monosof). Cold growth factor-reduced Matrigel (100 μl; Beckton Dickinson) mixed with 0.5x10 6 hMAPCs or PBS was applied into the pocket and allowed to solidify for 10 minutes. The skin was subsequently closed and the wound covered with Tegaderm dressing. Four, 8 or 16 weeks later, mice were anaesthetised with a ketamine/xylazine mixture and subjected to microlymphangiography following injection of 10 μl FITC-conjugated L. esculentum lectin (Vector Laboratories; in recipients of DsRed + donor lymph nodes) or 10 μl Texas Red-conjugated L. esculentum lectin (in recipients of eGFP + lymph nodes) in the palm of the right paw. Drainage of the implanted lymph node was monitored for 15 minutes and bright field and fluorescence pictures were taken at the end with a Zeiss MRc5 camera mounted onto a Zeiss Lumar microscope (using a Neolumar S lens, 0.8x magnification, FWD 80 mm). Mice were subsequently euthanised, the axilla regions containing the transplanted lymph node excised, fixed and processed for paraffin or OCT embedding and sectioning. Two additional sets of mice were subjected to in vivo magnetic resonance imaging (MRI), as described 8 at 4 or 16 weeks after lymph node transplantation. Briefly, mice were anaesthetised with isoflurane and mustard oil (diluted 1/5 in mineral oil) was applied with a cotton stick on both fore limbs for 2 x 15 minutes to elicit vascular hyperpermeability and aggravate edema. Mice were allowed to recover for another 30 minutes before MRI recording. Temperature and respiration were monitored throughout the experiment and maintained at 37°C and 100 -120 breaths per minute. MR images were acquired with a 9.4 T Biospec Bremen, Germany) reported as ratios between the lymph node implanted side versus the control side.
Calculation of the T 2 parameter maps of the manually delineated edema of the paws (or an area of the same size and located in the same region in the absence of edema) was done using Paravison 5.1 (Bruker Biospin).

Histology and morphometry
Morphometric analyses were performed on 7 μm paraffin sections, 10 μm cryosections or bright field pictures of exposed skin regions by blinded observers. Lymphatic (determined on LYVE1-, podoplanin-, Flt4-or Prox1/αSMA-stained sections) or blood (determined on CD31- Haematoxylin&Eosin and Sirius red staining were performed as previously described 1 . For CD31, podoplanin, Flt4 or PCK immunohistochemical staining, antigen retrieval was performed by boiling in target retrieval solution s1699 (Sigma). After cooling down in TBS, endogenous peroxidase activity was quenched in 0.3% H 2 O 2 in methanol. Slides were incubated with primary antibody overnight. A list of primary Ab's is provided in Supplementary Table S2. After washing in TBS, slides were incubated for 2 hours with biotinylated rabbit-anti-rat (CD31 and Flt4), goat anti-mouse (PCK) or goat-anti-hamster (podoplanin) antibodies and the detection signal was amplified with a tyramide signal amplification system (Perkin Elmer, NEL700A). Nuclei were revealed by haematoxylin counterstaining and slides were mounted with DPX mountant (Sigma). For LYVE1 immunofluorescence staining, antigen retrieval was performed by boiling in target retrieval solution s1699 (Sigma). After cooling down in TBS, endogenous peroxidase activity was quenched in 0.3% H 2 O 2 in methanol. Slides were incubated with primary antibody overnight. After washing in TBS, slides were incubated for 2 hours with biotinylated goat-anti-rabbit antibody and the detection signal was amplified with a tyramide-Cy3 or tyramide-fluorescein signal amplification system (Perkin Elmer, NEL704A or NEL701A). When combined with CD45 immunofluorescence staining, slides were subsequently incubated with primary anti-CD45 antibody overnight, followed by a 2 hour incubation with goat-anti-rat-Alexa488. For eGFP or vimentin immunofluorescence staining, antigen retrieval was performed by boiling in citrate buffer (pH=6). After overnight incubation with primary antibody, slides were incubated for 1 hour with Alexa-conjugated donkey-anti-chicken (eGFP) or goat-anti-mouse (vimentin) antibodies. For combined LYVE1/vimentin immunofluorescence staining, antigen retrieval was performed by boiling in citrate buffer (pH=6) and tissues were permeabilised by incubation in Triton 0.1% in PBS. After overnight incubation with primary antibodies, slides were incubated for 1 hour with goat-anti-mouse-Alexa488 and goat-anti-rabbit-Alexa568. For combined Prox1/αSMA immunofluorescence staining, antigen retrieval was performed by boiling in citrate buffer (pH=6) and tissues were permeabilised by incubation in Triton 0.1% in PBS. After overnight incubation with Prox1 primary antibody, slides were incubated for 1 hour with biotin-conjugated goat-anti-rabbit Ab and the detection signal was amplified with a tyramide-Cy3 or tyramide-fluorescein signal amplification system (Perkin Elmer, NEL704A or NEL701A). Slides were subsequently stained with Cy3-conjugated αSMA for 2 hours or with unconjugated SMA followed by goat-anti-mouse-Alexa660. For combined Prox1/eGFP immunofluorescence staining, antigen retrieval was performed by boiling in citrate buffer (pH=6) and tissues were permeabilised by incubation in Triton 0.1% in PBS. After overnight incubation with Prox1 and eGFP primary antibodies, slides were incubated for 1 hour with biotin-conjugated goatanti-rabbit and Alexa488-conjugated donkey-anti-chicken antibodies and the Prox1 detection signal was amplified with a tyramide-Cy3 signal amplification system (Perkin Elmer). Combined mCD31 and hCD31 staining was performed as previously described 9 . Immunofluorescence-stained slides were sealed with ProLong Gold Antifade Reagent with DAPI (Life Technologies; P36931). Images were recorded on a Zeiss Axiovert 200M microscope (at 20x or 40x magnification with LD Plan-NeoFluar lenses with NA 0.4 and NA 0.6, respectively), a Zeiss Axio Imager Z1 microscope equipped with a Zeiss MRc5 camera (at 10x, 20x or 40x magnification with EC Plan-NeoFluar lenses with NA 0.3, NA 0.5 and NA 0.75, respectively) or a Leica Leitz DMRBE equipped with a Zeiss MRc5 camera (at 5x or 20x magnification with a N Plan lens, NA 0.11 and a PL Fluotar lens, NA 0.5, respectively) and Axiovision 4.8 software. Images were cropped, pseudo-coloured and contrast adjusted using Photoshop (Adobe).

Statistics
Quantitative data represent mean ± s.e.m. 'N' represents the number of independent biological replicates on which statistical tests were performed. For qRT-PCR, measurements were performed in technical duplicate and averaged for each biological replicate. Tests used for statistical analyses are mentioned in the results and figure legends text. Normality of the data was tested by the Shapiro-Wilk test. Comparisons among two groups were performed by unpaired two-tailed Student's t-test in case of normal distribution or by Mann-Whitney-U test in cases where data were not normally distributed or normality could not be assumed. Multiple-group comparisons were done by 1-way ANOVA with Tuckey's post-hoc test (normal distribution) or Kruskal-Wallis test followed by Dunn's post-hoc test (no normality assumption). Wound size, length and width were evaluated by repeated measures ANOVA, followed by Fisher least-significant-difference test. Data were considered significant if the P-value was less than 0.05. All analyses were performed with Graphpad Prism