The aim of this study was to compare two methods for transvaginal delivery of autologous endometrial mesenchymal stem cells (eMSC) on a novel polyamide (PA) synthetic mesh with biomechanical properties designed to match human vaginal tissue using an ovine model of vaginal wall weakness. The second aim was to determine the retention of implanted autologous Iodex-labelled eMSC and to assess their effect on the extracellular matrix and inflammatory response to the implanted gelatin-coated (PA/G) and non-coated PA mesh.

Multiparous sheep (n=30) which underwent quantification of pelvic organ prolapse using a modified POP-Q (Pelvic Organ Prolapse Quantification) were divided into 4 experimental and 1 incisional control group. Subtotal hysterectomy was performed via ventral midline laparotomy on 12 sheep to collect endometrial tissue. Ovine autologous eMSC were isolated from hysterectomy tissue utilizing cell sorting by flow cytometry of CD271+ ovine eMSCs, cell culture and eMSC labeling with FITC-labelled IODEX paramagnetic nanoparticles. Polyamide synthetic grafts were fabricated and used alone or with eMSC or were dip-coated in 12% porcine gelatin and stabilized with 0.5% glutaraldehyde for PA/G constructs; autologous labelled ovine eMSC-seeded PA/G constructs were also prepared. 
Surgery consisted of a 40 mm, full-thickness midline incision on the posterior vaginal wall and eMSC/PA/G (n=6), PA/G (n=6), PA (n=12) were surgically implanted.  A 2 step procedure occurred in the fourth group with eMSC in a gelatin hydrogel (eMSC/G) were applied to the 6 already implanted PA meshes (PA + eMSC/G) and the gelatin crosslinked in situ with blue light. No graft was placed for the incisional controls.  After suturing the ewes were allowed to recover and were harvested 30 days post implantation. The analyses of explants included histology, collagen analysis using sirius red birefringence, image analysis of immunofluorescence and  immunohistochemistry stained tissues for macrophage markers. Biomechanical testing used the ball-burst test method.
 All data were analysed using GraphPad Prism 7.02; Image analysis of histological data were assessed using a 1-Way ANOVA with Tukey’s multiple comparison post hoc test and biomechanical data was assessed using standard t-test. Kruskal-Wallis with Dunn’s multiple comparisons tests was used for non-parametric data.

POP-Q measurements showed no significant change after one month of implantation for any group. Almost half (42%) of the 12 PA/G implants had vaginal exposure by 30 days with 4 exposures in PA/G explants and 1 exposure in an eMSC/PA/G explant. No graft exposure was observed in the PA construct explants in any of these 12 samples with or without eMSC. (p=0.012)
PA/G constructs were characterised by poor mesh/tissue integration and substantial folding while PA constructs exhibited superior mesh/tissue integration.
Histologically, the muscularis layer appeared disrupted especially within the PA/G constructs without cells (p<0.05). The eMSC/PA/G graft was associated with increased numbers of myofibroblasts compared with the PA + eMSC/G construct. A similar percentage of mature and immature collagen stained area was noted for both PA/G and PA irrespective of the presence of eMSC after 30 days. 
In PA constructs alone or with eMSC, the elastin fibre content was retained compared to PA/G  (p<0.001) and eMSC/PA/G  (p<0.01) explants. By confocal microscopy, Iodex-FITC-labelled eMSC on PA/G scaffolds were observed in vivo in frozen sections of the vaginal tissue implanted with eMSC/PA/G and PA + eMSC/G at 30 days. Dual colour immunofluorescence  showed that eMSC cells did not co-localise with smooth muscle fibres or leukocytes. PA/G constructs provoked a greater pro-inflammatory M1 macrophage response than did PA constructs, irrespective of the seeding of autologous eMSCs. 
Biomechanical testing showed that the average load (N) vs elongation (mm) curves and the linear region stiffness (gradient of load/extension) did not differ between any group.

This study demonstrated that the composition and design of tissue engineering constructs for delivering a cell-based therapy is critical for transvaginal repair surgery in a novel multiparous ovine model of POP. Unlike previous studies using multiparous ewes, we pre-selected sheep and experimental groups were matched on their POP-Q scores. In a 2-step procedure, PA mesh with excellent drapeability, implanted first with eMSC delivered subsequently in situ, integrates in a far superior manner than a composite mesh of PA/G. This approach is also translation friendly for surgeons and more practical for technicians preparing the cells. Implanted autologous eMSC persisted for 30 days in the vaginal wall near mesh filaments, influencing both the host immune response, ECM remodelling and fibrotic response.

In a time when commercial mesh has been, in many cases, withdrawn from market, new bioengineering approaches tested in an authentic POP preclinical model offers potential therapies for women with POP who currently have few effective treatment options. Such an approach has potential to advance towards clinical studies.