To identify candidate materials which have sufficient potential to be taken forward for an in vivo tissue-engineering approach to restoring the tissue structure of the pelvic floor in women with stress urinary incontinence (SUI) or pelvic organ prolapse (POP).
Materials and Methods
Oral mucosal fibroblasts were seeded onto seven different scaffold materials, AlloDerm ( LifeCell Corp., Branchburg, NJ, USA), cadaveric dermis, porcine dermis, polypropylene, sheep forestomach, porcine small intestinal submucosa (SIS) and thermoannealed poly(L) lactic acid (PLA) under both free and restrained conditions.
The scaffolds were assessed for: cell attachment using AlamarBlue and 4,6-diamidino-2phenylindole (DAPI); contraction using serial photographs; and extracellular matrix production using Sirius red staining, immunostaining and scanning electron microscopy.
Finally the biomechanical properties of all the scaffolds were assessed.
Of the seven, there were two biodegradable scaffolds, synthetic PLA and natural SIS, which supported good cell attachment and proliferation.
Immunostaining confirmed the presence of collagen I, III and elastin which was highest in SIS and PLA.
The mechanical properties of PLA were closest to native tissue with an ultimate tensile strength of 0.72 ± 0.18 MPa, ultimate tensile strain 0.53 ± 0.16 and Young's modulus 4.5 ± 2.9 MPa.
Scaffold restraint did not have a significant impact on the above properties in the best scaffolds.
These data support both PLA and SIS as good candidate materials for use in making a tissue-engineered repair material for SUI or POP.
Extensive work has been carried out on materials for pelvic floor repair, spanning both synthetic and natural materials, in response to the very high incidence of stress urinary incontinence (SUI) and pelvic organ prolapse (POP) [1, 2]. There are many surgical procedures for both conditions, but the mid-urethral tension-free tape procedure is the treatment of choice in most women with SUI, and colporrhaphy is performed for POP.
The objective success rate of tension-free vaginal tape is 72–81% at 5 years [3, 4]. The reported risk of erosion is up to 4% at 5 years [5, 6]. POP procedures involve plication of weakened tissue and re-enforcement with mesh. The use of cell-free mesh reduces the recurrence rate by 15% [7, 8]; however, mesh causes erosion into adjacent tissues in 4–24% of patients . Considering this, the US Food and Drug Administration have recently published a public health notification of the serious complications associated with synthetic mesh use for POP procedures .
The most commonly used meshes are synthetic and non-biodegradable polypropylene (PPL), however, biological materials in the form of autografts, allografts and xenografts have also been used. Meta-analysis has shown that non-absorbable synthetic meshes have a lower objective prolapse recurrence rate than absorbable meshes, 8.8 vs. 23.1%, but the rate of erosion was higher, 10.2 vs 0.7%, respectively . Autografts have the disadvantage of donor site morbidity and reduced long-term success rates but do not lead to a foreign body reaction and rarely cause erosion [11, 12]. Cadaveric allografts have shown high medium-term failure rates [13, 14]. Histologically, these grafts show high degradation and moderate encapsulation . Similarly, the failure rates of xenografts are also quite high . In summary, when natural materials fail it is usually through their degradation and resulting poor mechanical properties. When synthetic materials fail it is usually through erosion of adjacent tissues or owing to further weakening of native tissue.
With the current acellular materials giving a high failure or complication rate, this is an area where it is appropriate to look at tissue engineering to introduce cells that can form a new tissue, integrate into the body and achieve biomechanical characteristics similar to healthy native tissue .
No tissue-engineered cellular meshes have yet reached clinical trials in patients with SUI/ POP. Thus far, studies have described attaching fibroblasts to a limited number of meshes and have shown cell survival on these materials [16-19]. Researchers have also reported in vivo data in a rat model, where muscle-derived stem cells and bone marrow-derived stem cells were attached to small intestinal submucosa (SIS) and silk matrix, respectively [20, 21]. The authors reported that the muscle-derived stem cells formed differentiated myotubes with spontaneous contractile activity and the bone marrow stem cells led to new collagen production. Both studies concluded that more in vitro work was required.
Against this background, our aim was to look at a range of synthetic and natural candidate scaffolds, obtaining baseline conditions for the performance of these scaffolds when combined with cultured buccal fibroblasts. We have previously taken buccal fibroblasts and epithelial cells through to clinical evaluation to replace scar tissue in the urethra . Accordingly in this study we used oral fibroblasts as a candidate donor cell because of their minimal donor site morbidity and the fact that we have shown that they will produce new matrix when implanted in vivo.
In the present study, scaffolds were tested for their ability to support fibroblast attachment, the formation of new extracellular matrix (ECM) and to achieve biomechanical properties with cells present that were close to those of native tissue.
Materials and Methods
AlloDerm® (AL) and porcine dermis (PD) were obtained from LifeCell Corp. (Branchburg, NJ, USA), PPL from Gynecare Ethicon (Somerville, NJ, USA), sheep forestomach (SF) from Mesynthes Ltd, (Lower Hutt, New Zealand) and SIS from Cook Medical (West Lafayette, IN, USA). All the above samples were kind donations. Acellular cadaveric dermis (CD) was purchased from Euroskin (Beverwijk, Netherlands). It was washed in PBS, then incubated overnight with 1M NaCl. The epidermis was then removed leaving the de-epithelialized dermis. Sterile poly(L)-lactic acid scaffold was produced in the University of Sheffield by electrospinning . The material was heat annealed at 60 °C for 3 h to produce thermo-annealed poly(L)-lactic acid (PLA). The heat annealing process was investigated in another study (unpublished work) and was shown to lead to an increase in ultimate tensile (UT) strength from 0.2 to 0.55 MPa, ultimate tensile (UT) strain 0.42 to 0.74 and a reduction in Young's modulus (YM) from 3 to 1.2 MPa. This made the PLA easier to handle and brought its properties into the range for native tissue .
Fibroblast Isolation and Culture
Buccal mucosa was obtained from the Royal Hallamshire Hospital with informed consent. Samples were handled on an anonymous basis under a research tissue bank licence (number 08/H1308/39) under the Human Tissue Authority.
Buccal mucosa was cut into 0.5 cm2 pieces and incubated overnight (16 h) at 4°C in (0.4%) Difco-trypsin plus 0.1% w/v D-glucose in PBS, pH 7.45 (Difco Labs, Detroit, MI, USA). The epidermis was manually parted from the dermis. The dermis was minced with a scalpel in 5 mL 10% DMEM medium (444 mL DMEM + GlutaMaxTM (Gibco Invitrogen, Paisley, UK) supplemented with 50 mL fetal calf serum (Advanced Protein Products, Brierley Hill, UK) 5 mL penicillin (100 units/mL), streptomycin(100 μg/mL), and 2.5 mL fungizone (630 ng/mL; Gibco).
The minced tissue was incubated with 10 mL collagenase A (0.05% in 10%DMEM) at 37 °C in 5% CO2 overnight. The resulting suspension was centrifuged at 335.4 ×g for 10 min and the pellet resuspended in 10% DMEM. Thereafter, cells were seeded with a minimum of 5000 cells per T25 flask incubated at 37 °C in 5% CO2. Regular visual inspections were undertaken to observe cell morphology and exclude infection. Media were changed every 3 days and cells passaged before 80% confluency. Cells between passage 5–9 were used in the experiments.
Scaffold Preparation and Cell Seeding
All material preparation and seeding was undertaken in a sterile laminar flow culture hood reserved for human cell lines. The seven sterile prostheses were cut to 2 cm2 pieces and placed in six-well plates. Oral fibroblasts (800 000 per prosthesis), were seeded onto the scaffolds with 5 mL 10% DMEM solution. After cell attachment, a further 5 mL of 10% DMEM was added and the six-well plates incubated at 37 °C in 5% CO2. Scaffolds in media without cells and incubated dry without media were also included as controls. The constructs were cultured for 14 days. Each experiment was repeated nine times.
AlamarBlue (5 mL of 10% AlamarBlue in PBS, AbD Serotec, Kiddlington, UK) was added to each six-well plate 1 h after seeding 800 000 cells. This was then incubated for 60 min at 37 °C in 5% CO2. Absorbance at 570 nm was then read in a colorimetric plate reader (Bio-TEK, NorthStar Scientific Ltd, Leeds, UK) to obtain baseline values of colorimetric absorbance representing cell metabolic activity. On days 7 and 14, each sample was washed with PBS three times and placed in a new six-well plate with 5 mL AlamarBlue to assess metabolic activity at these time points.
After 14 days culture, cut portions of the tissue-engineered scaffolds were fixed in 3.7% formaldehyde in PBS. Then 0.8 mL of 1 ng/mL 4,6-diamidino-2phenylindole (DAPI; Gibco) was added to each well and incubated at 37 °C in 5% CO2 for 40 min. After three washes in PBS, the samples were observed through an Axon ImageXpress® fluorescent microscope (Molecular Devices Ltd, Union City, CA, USA) at λex385 nm/λem461 nm. The samples were imaged throughout their depth in all quadrants.
Restraint of Scaffolds
Scaffolds were assessed with and without restraint. Restraint involved suturing four corners of each material with a non-absorbable 5/0 monofilament suture (Ethilon®) to a metal grid for the 14 days of culture.
Contraction of Scaffolds
The extent of scaffold contraction with and without cells was assessed on days 7 and 14. Areas were measured relative to area at day 0 from digital photographs using ImageJ software (NIH, USA) to calculate areas. Cell-mediated contraction was calculated by subtracting the difference between scaffolds with and without cells.
Sirius Red Staining for Total Collagen Production
After formalin fixation (see above), 0.8 mL Sirius red stain (0.1% Direct Red 80 in saturated picric acid, Sigma-Aldrich, Dorset, UK) was added to each well of a 24-well plate. After 16 h, excess stain was washed off with distilled water. The specimens were dried, weighed and the stain from the tissue constructs eluted with 2 mL 0.2 M NaOH: methanol 1:1 to each well in the 24-well plate for 15 min. The absorbance was read at 490 nm in a spectrophotometer (Bio-TEK, NorthStar Scientific LTD, Leeds, UK). Scaffolds without cells acted as controls for calculating new collagen production.
After DAPI staining, 2 mL 1% BSA was added to each construct in the 24-well plates to reduce non-specific binding. Following three PBS washes, 100 μL primary antibody (AbD Serotec, MorphoSys UK Ltd, Oxford, diluted 1:50 in PBS) was added to each sample, and incubated at 37 °C in 5% CO2 for 30 min, followed by three PBS washes. Finally, 100 μL fluorescein isothiocyanate (FITC)-labelled secondary antibody (AbD Serotec, MorphoSys UK Ltd, diluted 1:50 in PBS) was added and the samples incubated at 37 °C in 5% CO2 for 30 min and washed three times with PBS. The constructs were imaged with an Axon ImageXpressTM fluorescent microscope (Molecular Devices Ltd). Excitation and emission wavelengths for FITC were λex495 nm/λem515 nm.
Assessment of the extent of immunostaining was done on a blind observer basis using a qualitative grading scale; absent = 0, mild presence = 1, good presence = 2, abundance = 3. Three blinded observers were presented with the images without being aware of their content and the median value from these scores was used.
Scanning Electron Microscopy
After formaldehyde fixation, the samples were incubated for 5 min with the addition of 2 mL 0.1 M cacodylate buffer. After removal of the cacodylate, 2 mL 2.5% gluteraldehyde in distilled water was added and incubated as above for 30 min. The gluteraldehyde was aspirated and again 2 mL of cacodylate buffer added to rinse any remaining gluteraldehyde. Then 500 μL of osmium tetroxide was added and incubated for 2 h. The osmium tetroxide was aspirated and 2 mL cacodylate buffer added and left for 15 min at room temperature. Subsequently the samples were incubated for 15 min with 75, 95 and 100% ethanol and finally freeze-dried for 16 h.
A gold coater (Edwards Sputter Coater S150B, Crawley, UK) coated all samples. Three random images were taken for each sample at ×100, ×800 and ×5000 magnification with a Phillips XL-20 scanning electron microscope (Cambridge, UK). As above, the samples were then scored for ECM production by three blinded observers. The marking scale included no ECM = 0, little ECM = 1, good ECM = 2, abundant ECM = 3.
Assessment of Mechanical Properties of Scaffolds
Tensiometry was performed on day 14 and all samples containing cells were maintained in media until the time of testing. Samples were cut, measured and then the moist samples were clamped to the tensiometer (BOSE Electroforce Test Instruments, Eden Prairie, MN, USA) using a 22-N load cell and a ramp test at a rate of 0.05 mm/s. The first failure point or plateau was used to calculate the UT strength, and the displacement at this point for the UT strain. The linear gradient of the graph was taken as the YM.
Differences between scaffolds were statistically tested against a null hypothesis of no difference between samples using a two sample t-test with equal variance not assumed, A P value < 0.05 was considered to indicate statistical significance.
Figure 1 shows the electron microscopy appearance of the seven scaffolds after 14 days' culture with and without fibroblasts. AlamarBlue was used to assess the metabolic activity of cells on the seven scaffolds examined with and without restraint. The change in metabolic activity of seeded cells relative to day 0 is shown in Fig. 2. Cells on SIS and PLA showed the greatest increase in metabolic activity which was significantly greater than for the other scaffolds (P < 0.05). Restraint did not significantly increase cell metabolic activity with the exception of PPL. With unrestrained PPL there was very little cell attachment; however when scaffolds were under restraint then, surprisingly, cells were able to attach and increase in metabolic activity.
DAPI staining showed good cell distribution throughout the SIS and PLA scaffolds (images not shown). Unrestrained PPL did not show any cells, but restrained PPL had cells visible at mesh intersections.
Restrained scaffolds were unable to undergo any significant contraction either with or without cells (Fig. 3). Unrestrained SIS and PLA, however, contracted by 15% over 14 days and there was no significant difference between scaffolds with and without cells. CD with cells contracted by 17.6% compared to 7.7% without cells, which was significant (P = 0.017), suggesting that the extra contraction was cell-mediated. The remaining scaffolds contracted by <7%.
Extracellular Matrix Production
Three approaches were used to assess matrix production. Sirius red assessment of total collagen production after 14 days culture of cells on scaffolds is shown in Fig. 4. For the other two methodologies, assessments were done after culture of cells on scaffolds for 14 days using cell-free scaffolds as controls. Examples of these results are shown in Fig. 5, while Fig. 6 summarises these results.
The first thing to note is that the two best-performing scaffolds were SIS and PLA and for these scaffolds there was no significant effect of restraint. In the main, the amount of collagen IV production was negligible and will not be discussed further, except to note that the only detectable collagen IV production was on restrained CD. One interesting result, however, was that there was a suggestion that more ECM production could be detected when scaffolds were restrained. This was evident for total ECM assessed by electron microscopy for AL, CD, PPL and SIS. It was also apparent for production of collagen I for AL, CD, PPL, PD and SF, and evident for production of collagen III for CD and SF. Thus, in the scaffolds where ECM production was relatively poor, results were improved by maintaining scaffolds under restrained conditions.
Biomechanical Properties of Scaffolds
Figure 7 shows the UT strength, UT strain and YM of scaffolds grown with and without cells and under restrained and non-restrained conditions. The addition of cells to scaffolds did not change their mechanical properties significantly over 14 days of culture except for the UT strain of SF. Also the use of restraint did not significantly affect the mechanical properties of the scaffolds with the exception of CD, where restraint reduced the UT strength and UT strain (but did not significantly affect the YM). The dashed lines indicate the published biomechanical properties of tissue from women without pelvic organ dysfunction . For UT strength, AL, PD and PLA were closest to the native tissue, whereas for UT strain, PPL, PD, SF and PLA were closest to this range. SF, SIS and PLA had a YM within the range of native tissue. The only material to have all three parameters within the range of native tissue was PLA.
Patients requiring pelvic floor reconstruction have received a variety of acellular synthetic and acellular biomaterials with varying results. Essentially, the pelvic tissues are weakened, reducing the options for surgeons to perform effective repair using native tissues and the use of relatively strong scaffolds, such as PPL, has resulted in high levels of erosion. In addition, natural cell-free materials have also failed historically, often because these become broken down, degrade and fail to offer sufficient mechanical support.
Several groups are now therefore engaged in developing tissue-engineered approaches for introducing cell-based scaffolds for repair and regeneration of the pelvic floor. One of the challenges is in knowing which mechanical properties to aim for. A recent systematic review of materials that have been used failed clinically and experimentally to find any simple correlation between the strength of materials implanted and clinical outcome . This is not surprising given that the mechanical properties of the implanted materials are only one part of the story; there is also the host response to these materials and the matter of whether they can be successfully remodelled. Tissue incorporation that induces fibrosis may succeed if the extent of the fibrosis is not too great and this contributes to long-term survival and support of the pelvic organs, they may fail if the fibrosis continues, resulting in tissue contraction and this is likely to lead to erosion of the tissues to which they are sutured.
From our assessment of the seven candidate materials under free and restrained conditions, a natural (SIS) and synthetic (PLA) scaffold were the clear frontrunners in terms of promoting good cell attachment, ECM production and achieving mechanical properties in the desirable range for clinical use. Restraint was interesting in that it did not affect metabolic activity or matrix production for the best performing scaffolds (SIS and PLA); however, results for ECM production on some of the scaffolds (AL, CD, PPL and SF) were improved by using this method.
Previous studies on SIS, PD, PPL and PLA/polyglycolic acid have been shown to support cells [16-19, 25, 26]; however, this is the first study to provide a comparative assessment of these materials and we were able to show, in addition, that the metabolic activity of cells increased more than threefold on SIS and PLA over 14 days of culture. SIS has long been known to provide good cytocompatibility [27, 28], possibly owing to the scaffold configuration and the availability of growth factors in the scaffold .
When it comes to scaffold integration into the body, the host immune response is extremely important for tissue survival and possibly rapid neovascularisation. Thus, a current popular hypothesis is that cross-linked natural collagenous materials which are excessively cross-linked cannot be remodelled by macrophages and may lead to a permanent state of inflammation and a macrophage M1 response, while scaffolds that can be remodelled by macrophages (an M2 response) may achieve better long-term integration into the body. Consistent with this, as collagen is broken down by enzymes in the body, the resulting fragments induce rapid neovascularisation .
While synthetic non-absorbable scaffolds can result in erosion, it is also true that they provide a successful solution for this problem for many women. These are multiple situations where the patient's response to the implanted material is a major determinant of clinical outcome. Against this background the development of scaffolds that are ultimately designed to degrade, leaving new tissue in place, seems a sensible solution.
Recently the abluminal surface of SF was reported to be superior for cell attachment than the luminal side, which we used . Also fibroblasts were found to proliferate only around the periphery of AL grafts with limited penetration into the matrix , which may explain the results with these materials. It has been proposed that the pore size, random orientation and fibre diameter of PLA make it a good scaffold for cell attachment . This is certainly consistent with previous work showing fibroblast culture throughout PLA scaffolds implanted in rats , which may explain its superiority over the other scaffolds.
We found that PPL failed to support cells unless it was restrained. DAPI imaging and electron microscopy showed that cells and ECM were concentrated at mesh intersections if restrained, but fibroblasts were unable to bridge the macroporous pores (2.25 mm in diameter) in the unrestrained material. With restraint, our explanation is that the pores were collapsed mechanically, so allowing better cell attachment. This is consistent with the finding of Skala et al.  who reported better fibroblast attachment to meshes with smaller pores.
After implantation, all materials will be subjected to cell interactions with the scaffolds and it has been suggested that PPL contracts by ∼15% after implantation [32, 33]. Accordingly, we examined the extent of intrinsic contraction of the scaffolds both without and with cells and found that PLA and SIS both contracted by ∼15%. Contraction could be prevented in vitro by the use of restraint (which we showed also leads to a modest increase in ECM production for some of the scaffolds). This is an important finding as it shows that fixing these meshes in place (as is done clinically) does not lead to reduction of collagen production and does not affect the mechanical properties of the mesh.
After implantation the biomechanical properties of materials can change. Current repair materials have shown a reduction in UT strength but there are variable results reported with respect to YM . Current thinking is that materials should be produced to replicate the biomechanical properties of the native tissues in the hope that they will have sufficient strength to provide support to the pelvic floor on implantation but not be too stiff to erode native tissue. In addition to this, the YM of materials has been shown to influence cellular outgrowth , and this may explain why materials with YM in the physiological range had good cytocompatability in our study. The most promising of the materials tested with respect to biomechanical properties was PLA, with SIS a close second.
Previously, SIS has been used clinically in POP and SUI procedures and cure rates have been shown to be improved over traditional colporrhaphy over 1 year , but long-term success rates for POP procedures are lacking. With SUI procedures, SIS has not been found to provide a durable repair over 5 years . This may be attributable to the use of acellular SIS which does not leave any neo-fascia behind as it degrades. The key difference with our proposed tissue-engineered mesh is that it is cellularized and by the time the scaffold degrades, the implanted cells will have produced a neo-tissue which survives in the long term.
The production of new ECM consisting of collagen I, III and elastin was greatest on SIS and PLA. CD and AL already contain these, making interpretation of new matrix production difficult. A key question is whether this ECM is good enough to serve its purpose in strengthening the pelvic floor repair. There are limited data on this and no long-term in vivo studies assessing the quality of tissue produced after transplantation of fibroblasts with a scaffold in to the pelvic floor. In the short term, Zou et al.  have shown good collagen production by mesenchymal stem cells on a silk scaffold which led to increases in UT strength and YM of the tissues at 4 and 12 weeks, suggesting that beneficial ECM was produced by these cells. In a rat incontinence model (sciatic nerve transection), the authors also showed increases in leak point pressure with the use of a silk sling with mesenchymal stem cells . Similarly, Cannon et al.  have shown improvements in leak point pressure after the implantation of acellular and cellular SIS with muscle-derived cells.
Another study has reported the use of minced muscle tissue at the time of implantation . The authors reported that introduction of immediately excised and minced muscle tissue led to survival and organisation of muscle cells in the implanted graft. Although there were limitations to this study, such as the use of an abdominal wall model, the use of small graft (1–2 cm) and the use of striated cells (which are not present in endopelvic fascia) and the lack of a control, i.e. a graft without cells, the concept of using immediately minced tissue for implantation is attractive and merits further investigation. Achieving tissue outgrowth over a large area as will be required for a gynaecological mesh is a challenge which still needs to be overcome.
Although other studies have reported improvements in collagen and elastin production with fixed uni-axial strain [38, 39], we did not see any with our scaffolds using bi-axial strain except for PPL. In the present study we used buccal fibroblasts which can be harvested readily under local anaesthetic to compare critically the behaviour of a range of natural and synthetic scaffolds for their potential to form tissue-engineered implants. It is proposed that when these autologous cells are implanted on the scaffolds they will continue to produce a neo-fascia which will continue to strengthen the weakened pelvic floor. We recognise there is diversity in fibroblasts from different sites  and that fibroblasts have been reported to retain positional memory of their previous functional role over a number of generations . We found buccal fibroblasts transplanted ectopically produced clinically useful tissue when used to replace scar tissue in the urethra . Hence, we suggest these as a good candidate cell source for this purpose. Whether these cells continue to provide a suitable neo-tissue after scaffold degradation will now need to be investigated in in vivo studies.
On the basis of this work we suggest that two scaffolds are worth taking into further studies, PLA and SIS. Key questions to now be asked of these scaffolds are what the host response will be to these scaffolds implanted with cells, how these materials might promote neovascularisation and how they will cope with the dynamic stresses placed on them in the pelvis in vivo.
We thank The Urology Foundation and Robert Luff Foundation for funding.