Fate of mesoangioblasts in a vaginal birth injury model: influence of the route of administration

Currently cell therapy is considered as an experimental strategy to assist the healing process following simulated vaginal birth injury in rats, boosting the functional and morphologic recovery of pelvic floor muscles and nerves. However, the optimal administration route and dose still need to be determined. Mesangioblasts theoretically have the advantage that they can differentiate in skeletal and smooth muscle. We investigated the fate of mesoangioblasts transduced with luciferase and green fluorescent protein reporter genes (rMABseGFP/fLUC) using bioluminescence, immunofluorescence and RT-PCR in rats undergoing simulated birth injury. rMABseGFP/fLUC were injected locally, intravenously and intra-arterially (common iliacs and aorta). Intra-arterial delivery resulted in the highest amount of rMABseGFP/fLUC in the pelvic organs region and in a more homogeneous distribution over all relevant pelvic organs. Sham controls showed that the presence of the injury is important for recruitment of intra-arterially injected rMABseGFP/fLUC. Injection through the aorta or bilaterally in the common iliac arteries resulted in comparable numbers of rMABseGFP/fLUC in the pelvic organs, yet aortic injection was faster and gave less complications.

oxygen severely affected the survival of injected cells 18 . Since SC are known to be recruited to sites of inflammation 19 , systemic injection has also been employed. Intravenous administration is the least invasive intravascular route, yet it has the inherent disadvantage that the majority of cells becomes entrapped in the capillary beds of non-target organs, mainly the lungs 20 . An alternative is selective injection, e.g. in the arteries feeding the area of interest which may result in a more efficient engraftment and/or lower cell dose required and less off-target effects 21 . The intra-arterial route has been proven to be effective in many conditions, including renal ischemia-reperfusion injury, radiation injury and skeletal muscle injury [21][22][23] , however, it has not yet been investigated in the vaginal birth injury model.
In this study, we aimed to investigate the fate of stem cells when administered via local, intravenous or intra-arterial route in a rat model for simulated vaginal birth injury. Here we employ rat-derived mesoangioblasts (rMABs). MABs are vessel-derived SC with a high regenerative potential for muscular disorders 24 . Conceptually, MABs are good candidates for repairing birth injury because of their capacity to differentiate into skeletal and smooth muscle 25 . MABs have already been labeled with a double reporter viral vector encoding both enhance green fluorescent protein (eGFP) and firefly luciferase (fLuc) reporter to monitor their distribution in vivo by bioluminescence (BLI) and fluorescence 26 .

Results
Characterization of rMABs eGFP/fLUC . Rat MABs were isolated from 20-day-old skeletal muscle fetal biopsies and sorted for alkaline phosphatase after 10 days of culture. About 3% of the total cell population was alkaline phosphatase positive (Fig. 1A). Only fLuc expressing rMABs showed consistent bioluminescence when incubated with luciferin while WT MABs only showed background signal (Fig. 1A).
Next, we investigated the in vitro myogenic potential of rMABs eGFP/fLUC . To this end, we differentiated them to skeletal and smooth muscle lineages (Fig. 1B). When co-cultured with murine myoblasts (C2C12) they were able to form GFP+ chimeric myotubes, albeit with low efficiency. However, when differentiated towards smooth muscle, rMABs eGFP/fLUC efficiently expressed early and late smooth muscle markers alpha smooth muscle actin and calponin (Fig. 1B). Finally, we profiled rMABs eGFP/fLUC by flow cytometry analysis. rMABs eGFP/fLUC were highly positive for homing and adhesions marker HCAM (CD44) and MCAM (CD146), positive for pericytes marker PDGFR beta (CD140b), and negative for endothelial lineage marker PECAM-1 (CD31) and hematopoietic marker CD45 (Supplementary Figure 1).

Figure 1.
Isolation, manipulation and characterization of rat mesoangioblasts (rMABs). (A) rMABs were isolated from 20 days old rat fetuses from skeletal muscle of both hind limbs. After expansion, cells were stained for alkaline phosphatase and further isolated by FACs sorting. In order to track the cells further in vivo, rMABs were labelled with CHMWS-eGFP-T2A-fLuc viral vector and sorted by FACs for GFP expression. The engineered cell line of rMABs GPF+ Luc+ were further validated in vitro by BLI and immunofluorescence. (B) rMABs were differentiated towards skeletal and smooth muscle lineages (left to right). In co-cultures with C2C12 GFP+ rMABs formed few chimeric myotubes (arrow). After smooth muscle differentiation, rMABs expressed both calponin and alpha smooth muscle actin.
SCIEnTIFIC REPORtS | (2018) 8:10604 | DOI:10.1038/s41598-018-28967-w Tracking of carbon particles following local and intra-arterial injections. When Chinese ink was locally injected, black (carbon) particles could be observed on histological sections in the connective tissue around the vagina, rectum and levator ani (Supplementary Figure 2A). There were no carbon particles visible within the interstitial and vascular spaces of these pelvic organs (Supplementary Figure 2C,E,G,I). Conversely, carbon particles were homogeneously distributed in the interstitial and vascular space of pelvic organs when the injection was performed by arterial route (Supplementary Figure 2B,D,F,H,J), indicating that the initial distribution into the target area was effective using these routes of administration.
Fate of the rMABs eGFP/fLUC after injection through different routes. One hour after injection of rMABs eGFP/fLUC , IA-Ao administration route was associated with the highest amount of rMABs eGFP/fLUC in the pelvic organs region, as evidenced by BLI ( Fig. 2A, left). Following IA injection, the amount of rMABs eGFP/fLUC increased significantly at 1d (Fig. 2B), with a significantly higher amount of rMABs eGFP/fLUC in the pelvic organs region compared to IV and local administration ( Fig. 2A, middle). Following IV injection, most of the BLI signal was found in the lungs and tail at 1 h and 1 day after injection. Animals of all groups showed a significant decrease of rMABs eGFP/fLUC at 3d, yet with the highest remaining cell number in the IA-Ao group ( Fig. 2A, right). Individual data are shown in Supplementary Figure 3.
When biodistribution was investigated ex-vivo, rMABs eGFP/fLUC were more homogeneously distributed in the pelvic organs at 3d following IA administrations (Fig. 3A). A significant drop in the BLI signal intensity was observed at 7 days in all groups. Further, we detected GFP gene expression in the pelvic organs at 3 days and 7 days. Both IA injected animals had a comparable amount of rMABs eGFP/fLUC in the pelvic floor and a significantly higher amount of rMABs eGFP/fLUC in the vagina, urethra and bladder compared to those locally injected at 3 days  IV-intravenous; IA-CIa-intra-arterial (common iliacs); IA-Ao (intra-arterial (aorta). *p < 0.05; **p < 0.001; ***p < 0.0001. at 7 days in all groups. Further, GFP + cells were tracked in the IA-Ao group by immunofluorescence. GFP + MABs could be found in the vagina, bladder, urethra, rectum and levator ani 2 days after injection (Fig. 4).
Sixty percent of IA-CIa rats died due to thromboembolism in the hind limbs. These rats were not included in this study. When we started administrating heparin the rate of thromboembolism dropped to 5% of the rats. Overall, we had a 15% mortality immediately after MABs were injected systemically.
Influence of the injury on the fate of the rMABs. rMABs were injected in a non-injured group by the IV, local and IA-Cia route in order to investigate the influence of simulated birth injury on the fate of the rMABs eGFP/fLUC (Fig. 5). The injury did not affect the fate of the rMABs eGFP/fLUC in the local or IV groups. However, the non-injured rats showed a significantly lower amount of rMABs eGFP/fLUC compared to the injury cohorts when injected by the intra-arterial route (Fig. 5).

Discussion
We studied the short-term effect of different routes for rMABs eGFP/fLUC administration in a standardized rat model of birth injury, combining nerve crush and vaginal distension. The most important findings are that intra-arterial delivery resulted in (1) the highest amount of rMABs eGFP/fLUC in the pelvic organs region at 1 and 3 days after injection and in (2) a more homogeneous distribution of rMABs eGFP/fLUC over all relevant pelvic organs early after injection. Moreover, (3) the presence of the simulated birth injury is key for recruitment of intra-arterial injected rMABs eGFP/fLUC .
Conceptually yet also clinically, the ideal delivery route is topical so that the transplanted cells can directly and efficiently home to the (therapeutic) target tissues without leakage to other organs, yet avoiding extra damage to the surrounded tissue during the injection. Despite the fact that local injection allowed efficient local delivery of rMABs eGFP/fLUC 12,17 , the cells tended to gather only at the injection sites without being diffused through the pelvic organs. The IV route is usually considered the least invasive technique of injection, however, when injected by this route we could detect relevant numbers of rMABs eGFP/fLUC only in the lungs, as described previously 20 . In light of the limitations of the previous methods for cell delivery in birth injury, we proposed two alternative routes to inject MABs through the aorta and the common iliacs arteries. In this study, we show that rMABs eGFP/fLUC injected intra-arterially resulted in the most efficient homing and homogeneous distribution in the pelvicorgans in the birth injury model. Other researchers have also observed a more homogenous distribution and longer survival of the SC when they were injected IA 23,27 . Moreover, engraftment and regeneration were observed when SC were injected IA in a muscle injury model in a non-primate 28 .
IA-Ao rats showed higher rMABs eGFP/fLUC survival at 1 and 3 days compared to IA-CIa rats. IA-Ao injection had other advantages, such as the need of only one puncture instead of two, leading to a faster procedure with less bleeding and no thromboembolism. Besides, less SC leakage was observed in IA-Ao rats due to technical reasons, given the wider diameter of the aorta compared to the common iliac, which facilitates arterial puncture. Conversely, the occurrence of thromboembolism observed in the IA-CIa group is probably due to the longer compression of the arteries after puncture, required for hemostasis.  It is controversial if the recruitment of the SC is produced by margination, rolling, specific attachment, and extravasation in postcapillaries and venules 29,30 or whether they are mechanically trapped in capillaries 28 . Our data supports the idea that the recruitment of the rMABs eGFP/fLUC in the IA cohort was due to the inflammatory response since higher amount of rMABs eGFP/fLUC were observed in the injury groups compared to sham controls irrespective of the time point. SC are known for their chemotactic properties due to specifics cytokines, such as monocyte chemotactic protein-3 (MCP-3) 31 . Moreover, MCP-3 is known to play a role in the postischemic recovery and it is upregulated in the pelvic organs immediately after vaginal distension injury 32,33 . One other explanation can be that the inflammatory response increases vasculature permeability, which may lead to higher entrapment of cells in injured organs, similar to the EPR-effect 34 .
We used a vasodilator in order to increase homing of the rMABs eGFP/fLUC injected systemically; as recommended previously 35 . However, we observed an improvement only in the IA group (data not shown). Two possible hypotheses arise: vasodilatation made arterial puncture easier, leading to a lower leakage of rMABs eGFP/fLUC ; or the vasodilator alleviated the vasoconstriction due to the ischemia and reperfusion injury in the pelvic organs 3,36 , leading to a higher amount of rMABs eGFP/fLUC reaching the target area.
The thromboembolism observed in the IA-CIa group was nearly completely solved by the administration of heparin. We think that the thromboembolism was in part due to the injection of large amount of cells, associated to the arterial injury and the compression after injection, as described previously 37 . Moreover, mesenchymal SC have been shown to have a procoagulant activity inducing thrombogenesis in vivo 38 . This effect was avoided by heparin treatment before MSC injection in vivo.
Here, we observed a significant increase of rMABs eGFP/fLUC in the pelvic organs region one day after injection, followed by a significant drop at 3 days. Our results are compatible with those of Dai et al. 17 , who found an increase of MSC viability at 1 day, followed by a drop after 2 days and completely elimination by 4 days in a birth injury rat model. Interestingly, MSC persisted for 8 weeks in the birth injury BALB nude mice model 12 . Therefore, we hypothesize that this drastic reduction of rMABs eGFP/fLUC viability is due to the adaptive immune system. Currently, cell rejection by the host is a concern in the cell therapy field 26 . Even though, it has been described previously that MABs were inert to the immune-system 39 , isolated T-cells from mMAB-injected muscle were clearly reactive against re-exposure to mMABs 40 . Moreover, it has been proved that the use of immunosuppressors increase the viability of MABs in the long-term 26 . Route of administration and dosage are two critical factors determining the efficiency of cell therapy. Since it has been suggested that a lower dose of SC is necessary for IA injection 21 , the ideal dose for cell therapy in the birth injury model still has to be investigated. Here, we investigated the most efficient homing and biodistribution using several administration routes in the birth injury model. Yet, for investigating their engraftment, it would be wise to use autologous injection or an efficient immunosuppressive therapy 41 . Another limitation of this study is that the functional outcome of the administration routes was not tested. Therefore, a correlation of the fate of the cells in the pelvic area with the effect the MABs may execute has not been established yet. Although most studies have shown that cell therapy was more efficient when administered cells are indeed homing to the target organ 14,15 , further studies will be necessary to elucidate the functional effects of MABs or their secretome in this model.
Altogether, intra-arterial injections of rMABs eGFP/fLUC resulted in a more efficient homing and distribution of mesoangioblasts in the pelvicorgans in rats after birth injury model. Aorta and bilateral common iliac administration resulted in comparable number of MABs in the pelvic organs. From a technical point of view, aortic injection is faster and results in less complication compared to iliac injection.

Material and Methods
Generation of reporter rMABs and Cell Culture. In order to investigate the most efficient delivery route for SC in birth injury model, we isolated rMABs from 20-day-old rat fetuses (Fig. 1A). Skeletal muscle from both hind limbs was harvested and processed as previously described 42 . Briefly, tissue biopsies were minced in ~2 mm size pieces and plated on collagen coated 6 cm dishes. After 10 to 14 days alkaline phosphatase positive cells were sorted. rMabs were cultured at 37 °C in a 5% CO 2 , 5% O 2 humidified incubator in DMEM supplemented with 20% FBS, 1% Pen-Strep, 1% L-glutamine, 1% sodium pyruvate, 1% non-essential amino acids, and 0.2% b-mercaptoethanol, (all reagents from GIBCO,USA). To enable tracking rMABs after injection in vivo and in real time, and to prove cell viability, rMABs were transduced with a HIV-derived lentiviral vector LV_CMV-eGFP-T2A-fLuc constructed by the Leuven Viral Vector Core at 1:100 concentration for 48 hours (virus titer 2.34e + 08 TU/ml), and subsequently sorted as GFP + fraction.
Smooth muscle differentiation was induced the day after with DMEM high glucose, supplemented with 2% of heat-inactivated horse serum (HS), 1% penicillin/streptomycin solution, 2 mM glutamine, 1 mM sodium pyruvate, and 10 ng/mL TGFβ (Peprotech, Rocky Hill, USA) for 7 days. Myogenic differentiation was carried out by seeding C2C12 and rMABs eGFP/fLUC in a 1:3 ratio on collagen coated dishes. When cultures reached 80% confluence, myogenic differentiation was induced by incubating the cells with DMEM high glucose, supplemented with 2% of HS, 1% penicillin/streptomycin solution, 2 mM glutamine, and 1 mM sodium pyruvate (all reagents from GIBCO) for 5 days. At the end of differentiations, cultures were fixed with 4% paraformaldehyde (PFA, Polysciences Europe GmbH, Germany) and stained. AlexaFluor-conjugated donkey secondary antibodies (Thermo Fisher Scientific, Ghent, Belgium). Nuclei were counterstained with Hoechst. The details of primary antibodies and respective dilutions are described on Table 1.

Animals and simulated vaginal delivery injury model. The animal experiments were evaluated and
approved by the Animal Ethics Committee of the KU Leuven (P271-2015) and was performed according to international guidelines. Sixty-three female virgin Sprague-Dawley rats of 12-week-old (250-300 g) were used. Rats underwent either a simulated childbirth injury by pudendal nerve crush and vaginal distension (PNC + VD; n = 42) or sham (n = 15), as described previously 43 (Supplementary Figure 4). We chose the combination of pudendal nerve crush and vaginal distention since it mimics better birth injury observed in humans 43 . Rats were anesthetized by intraperitoneal injection of ketamine (70 mg/kg), xylazine (7.5 mg/kg) and buprenorphine (0.05 mg/kg). To induce PNC injury, an incision was made in the dorsolumbar area; the pudendal nerve was identified in the ischiorectal fossa and crushed twice with a needle holder for 30 sec. For simulated VD, a modified 10Fr Foley catheter was inserted into the vagina and the balloon was inflated to 3 mL for 4 h. Sham operations consisted of pudendal nerve dissection without crushing and catheter insertion for 4 hours without balloon inflation. All animals were kept on a heating pad during surgery and recovery. For post-operative pain-relief, buprenorphine was administered IP for 2 days (0.1 mg/kg, BID).
Injection routes. Animals were kept under general anesthesia during administration of cells with 1.5% isoflurane in 100% oxygen at 1.5 L/min. Schematic drawings of all injection routes are displayed in Fig. 6. Local administration was done laterally at both sides of the vagina, using a 26 G vascular catheter needle (BD Neoflon Cannula, Becton Dickinson and company, Aalst, Belgium). IV injection was performed in the distal part of the tail vein using a 24 G vascular catheter (BD Insyte, Becton Dickinson and company). IA administrations were performed by either injecting in both the left and right common iliac arteries (IA-CIa), or by one injection in the aorta (IA-Ao). First a ventral midline laparotomy was performed. Next, for IA-CIa the right common iliac was dissected and a loose ligature was placed close to the aorta bifurcation to facilitate the insertion of the needle later on. In order to direct the flow of the injected cells towards the pelvicorgans, the right external iliac was occluded with a vascular clamp for 10-20 s. Antegrade catheterization at a 45°angle was performed using a 33 G needle (Acu-Needle, Acuderm, Fort Lauderdale-FL, USA) directly into the right common iliac artery. After injection, the needle was removed, and the injection site was compressed with a resorbable collagen membrane (Lyostypt ® , B. Braun, Aesculap, Tuttlingen, Germany). The same procedure was performed on the contralateral side.
For the IA-Ao injection, the procedure was performed in a similar way, except that a single injection was administered after placing the loose ligature around the aorta 1.5 cm above the iliac bifurcation. After injection(s), the abdominal wall was closed with a 3-0 monofilament polypropylene suture in two layers (Prolene, Ethicon, Zaventem, Belgium). Prior to the formal experiments, we tested these routes with injections of Chinese ink in non-injured rats and tracked the carbon particles (n = 3 for each route). This was perfomed to determine the initial track of the injection. Immediately after injection, those animals were euthanized and pelvic organs were fixed in 4% of PFE for H&E staining. rMABs eGFP/fLUC transplantation. One hour after simulated VD injury (n = 10/group) or sham (n = 5/ group), rats were randomly assigned to receive 2 × 10 6 rMABs eGFP/fLUC either locally (perivaginal-bilateral), IV (tail vein), IA-CIa or IA-Ao. rMABs eGFP/fLUC were resuspended in 30 µL (vagina) or 800 µL (IV or IA) of physiological solution. All rats received heparin (400 UI/kg; SC) 1 h before injection and the vasodilator isosorbide dinitrate (1 mg/kg; IV) 1 minute before injection.

Distribution and viability of rMABs eGFP/fLUC in vivo and ex vivo.
For in vivo bioluminescence (BLI), rats were first anesthetized with 1.5% of isoflurane in 100% of oxygen and then given a single injection containing d-luciferin potassium salt dissolved in phosphate-buffered saline (PBS) (126 mg/kg). None of the rats evaluated had the pelvic organ region shaved. Ten minutes after luciferin injection, the rats were placed in the imaging chamber (IVIS Spectrum, Perkin Elmer, X). Next, consecutive frames were acquired for 5 min until the maximum signal intensity was reached. A region of interest was drawn around the pelvic organs. The maximal radiance (p/s/ cm2/sr) was measured within this region. Images were analyzed using Living Image version 4.5. BLI data was obtained 1 h, 1 day and 3 days following rMABseGFP/fLUC injection. Ex vivo BLI measurements were done at 3 or 7 days after injection. Rats were euthanized immediately after in vivo BLI and the urinary bladder, urethra, vagina, rectum, levator ani muscles, lungs and spleen were harvested and imaged similarly ex vivo. After BLI analysis, samples were snap frozen and stored in OCT at −80 °C until further analysis.
Immunofluorescense. In order to confirm the presence of rMABs eGFP/fLUC in the pelvicorgans, serial sections from the IA-Ao group (n = 2) were fixed with 4% PFA, permeabilized, and immunostained for eGFP (ab5450, Abcam,) and laminin (Ab11575; Abcam). Alexa 488 (Thermo Fisher Scientific) and Alexa 647 (Thermo Fisher Scientific) were used as secondary staining Ig G and Hoechst for nuclear staining. Isotype control antibodies were used as negative staining. Images were taken with a fluorescence microscope (ECLIPSE Ti, Nikon) using NIS elements software (Nikon). RT-PCR for GFP. RNA was extracted from frozen tissues following TRIzol extraction protocol (Life Technologies). 500 ng of RNA were retrotranscribed into cDNA (SSIII cDNA production kit, Thermo Fisher Scientific). qPCR was performed on 1/10 diluted cDNA, using ViiA7 384-plate reader (Thermo Fisher Scientific). Final primers concentration 100 nM, final reaction volume 10 µl, PGK and GAPDH as internal reference; thermal profile 95 °C 15 seconds, 60 °C 60 seconds, 40x. Primers are listed in Table 2. Data are shown as relative expression normalized on respective organ of local injections.
Statistics. Data