In vivo Sealing of Fetoscopy-Induced Fetal Membrane Defects by Mussel Glue

Abstract

Introduction

The benefits of fetal surgery are impaired by the high incidence of iatrogenic preterm prelabor rupture of the fetal membranes (iPPROM), for which chorioamniotic separation has been suggested as a potential initiator. Despite the urgent need to prevent iPPROM by sealing the fetoscopic puncture site after intervention, no approach has been clinically translated.

Methods

A mussel-inspired biomimetic glue was tested in an ovine fetal membrane (FM) defect model. The gelation time of mussel glue (MG) was first optimized to make it technically compatible with fetal surgery. Then, the biomaterial was loaded in polytetrafluoroethylene-coated nitinol umbrella-shaped receptors and applied on ovine FM defects (N = 10) created with a 10 French trocar. Its sealing performance and tissue response were analyzed 10 days after implantation by amniotic fluid (AF) leakage and histological methods.

Results

All ewes and fetuses recovered well after the surgery, and 100% ewe survival and 91% fetal survival were observed at explantation. All implants were tight at explantation, and no AF leakage was observed in any of them. Histological analysis revealed a mild tissue response to the implanted glue.

Conclusion

MG showed promising properties for the sealing of FM defects and thereby the prevention of preterm birth. Studies to analyze the long-term tissue response to the sealant should be performed.

Mini-Summary

What does this study add to current knowledge?

• Current synthetic gluing materials for fetal membrane (FM) sealing suffer from poor cytocompatibility and very fast gelation times. In this study, we optimize a biomimetic adhesive (mussel glue [MG]) for the sealing of FM defects in vivo. We test the performance of this biocompatible glue in a newly developed ovine FM defect model that permits the placement of implants in a very controlled manner.

What are the main clinical implications?

• Despite the high morbidity and mortality associated with preterm birth, there is currently no sealing strategy to prevent the iatrogenic preterm prelabor rupture of the fetal membranes and consecutive preterm births after minimally invasive fetal interventions. This in vivo study shows the absence of amniotic fluid leakage and a high ovine fetal survival, which makes the use of MG a potential candidate for FM sealing. Furthermore, the umbrella-shaped receptor employed is compatible with the minimally invasive application of the material.

Introduction

Iatrogenic preterm prelabor rupture of the fetal membranes (iPPROM) remains the main complication after minimally invasive prenatal interventions. Such surgeries are performed for a variety of complications, such as twin-to-twin transfusion syndrome, lower urinary tract obstructions, congenital diaphragmatic hernia, and myelomeningocele [1]. IPPROM occurs in around 30% of fetal interventions [1, 2, 3, 4, 5], although rates as high as 60–100% have been reported [1], and about a 30% of babies have been reported to be born before 27 weeks of gestation [6]. Sequelae associated with preterm birth decrease the life quality of the affected preterm newborns and include blindness, deafness, intellectual impairment, chronic lung disease, and cerebral palsy [1, 5]. The exact mechanisms leading to iPPROM remain unknown; however, several risk factors such as postoperative chorioamniotic separation, fetal membrane (FM) damage, and apoptosis at the port site have been discussed [7]. Iatrogenic FM defects remain unhealed until the end of pregnancy [2, 8], and it is currently hypothesized that biomaterials that seal such defects could mitigate these risks and enable the prolongation of pregnancy.

Natural materials such as fibrin glue (FG) showed promising FM sealing properties in vitro and ex vivo [9, 10]. However, FG-treated FM defects did not significantly improve FM integrity in vivo as compared to untreated controls, and fetal survival did not increase [11, 12]. Alternatives such as the treatment with the collagen-based sealants TissuePatch and DuraSeal failed as well to improve fetal survival [13]. While these materials have been reported to have good biocompatibility, the factor limiting their performance is likely their high susceptibility toward proteolytic degradation. On the other hand, commercially available synthetic glues which degrade slowly (in vivo stability for many months) showed FM cytotoxicity ex vivo [9, 14], and FG-treated defects failed to perform better than untreated sacks in vivo [13]. For these reasons, engineered nature-inspired bioadhesives have been developed for the sealing of FM defects [15, 16, 17, 18, 19, 20, 21]. Among these promising bioadhesives is a catechol-functionalized 4-arm poly(ethylene glycol) that cross-links in the presence of an oxidizing agent (commonly sodium periodate, NaIO₄). This mussel-mimetic sealant (mussel glue [MG]) showed good cytocompatibility, adhesion to the FMs in a wet environment, resistance to degradation in amniotic fluid (AF), and ability to seal punctured FMs ex vivo [9, 10, 22]. Given these promising results, it was tested in an in vivo rabbit FM defect model [23], where the survival rate of the fetuses was comparable to that of the untreated FM, but leakiness was reduced by more than half. To enable the extensibility of the MG, a copolymer of poly(propylene oxide) (PPO) and flexible poly(ethylene) oxide (PEO) was developed [24]. This extensible MG has shown similar sealing properties by ex vivo inflation testing to its less flexible counterpart [20]. Based on these previous successful results, the aim of this study was to evaluate the in vivo biocompatibility, stability, and FM sealing properties of the extensible MG in a preclinical sheep model.

Materials and Methods

Umbrella-Shaped Nitinol Molds

Umbrella-shaped nitinol molds were prepared as previously described [25]. Briefly, 200-µm thick nitinol molds (MeKo) were custom-made by laser cutting followed by temperature-based shape setting, further coated with a 0.004” thick polytetrafluoroethylene coating (ZEUS Industrial Products, Inc) and sterilized with ethylene oxide by Medistri SA. To facilitate handling and implantation, a Vicryl rapide 3-0 suture wire (Ethicon) was inserted through the umbrella.

MG Synthesis and Gelation Time Optimization

A MG consisting of a branched 4-arm copolymer of PPO and PEO polymers (Fig. 1) was produced and characterized as previously described [19] and stored as MG lyophilized precursor. For the adjustment of the gelation kinetics, the lyophilized MG precursor was dissolved in 2x phosphate-buffered saline (PBS) at pH 6, 6.5, 7.0, and 7.4. Equal volumes of MG precursor solution and sodium periodate (NaIO₄) were mixed, and the gelation time was measured with a timer (n = 3 gels per condition).

Fig. 1.

Elastic MG. a Chemical structure of the elastic MG, which is based on a star-shaped 4-arm PPO-PEO copolymer modified with catechol moieties. Reproduced from ref. [18] with permission. b pH dependency of MG gelation time. Data are depicted as individual data points ± SD. One-way ANOVA with Tukey's multiple comparisons test was performed. *p < 0.05, **p < 0.01 ***p < 0.001, ****p < 0.0001; SD, standard deviation.

Implant Preparation

For the preparation of implants, equal volumes of MG precursor solution (300 mg/mL in 2x PBS, pH 7) and sterilized NaIO₄ (12 mg/mL in Milli-Q water) were mixed, and 500 μL were immediately pipetted into an umbrella-shaped nitinol implant.

Animal Care

Animal care and housing were performed according to protocols approved by the Cantonal Veterinary Office of Zurich, Switzerland (license number ZH214/18), were in accordance with the Swiss Animal Protection law, and conformed with the European Directive 2010/63/EU of the European Parliament and of the Council of September 22, 2010, on the Protection of Animals Used for Scientific Purposes and to the Guide for the Care and Use of Laboratory Animals. For this study, 5 time-pregnant Swiss White Alpine ewes were used at gestational ages of 56–69 days (Table 1).

Table 1.

Study characteristics and outcomes

	Ewe 1		Ewe 2		Ewe 3	Ewe 4	Ewe 5
Gestational age at OP	69 days		64 days		57 days	64 days	67 days
Days of implant in sheep	11 days		11 days		10 days	12 days	12 days
Ewe survival			100%

Horn	Left horn	Right horn	Left horn	Right horn	Left horn	Left horn	Left horn	Right horn
Implants (total = 10)	1	1	1	1	1	1	2	2
Adhesions visible	Yes	Yes	No	No	No	Yes	No	Yes
AF leakage	No	No	No	No	No	No	No	No
Number of fetuses	1	1	2	2	1	2	1	1
Alive fetuses	1	1	2	2	0	2	1	1
Fetal survival					91%
Amnion bands	No	No	No	No	No	No	No	No
Fetal skin defects	No	No	No	No	No	No	No	No

Preoperative and Operative Handling

Intravenous (i.v.) anesthesia was administered to the ewes with 3 mg/kg BW ketamine hydrochloride (Ketasol®-100, Graeub AG) in combination with 0.3 mg/kg BW midazolam (Dormicum®, Roche) and 2–5 mg/kg BW propofol (Propofol-Lipuro®, B. Braun Medical AG). Positive pressure ventilation (fresh gas flow 1–1.5 L/min, 12–15 breaths/min, tidal volume 5–10 mL/kg, with a FiO₂ of 0.5) with isoflurane (2–3%) in oxygen/air mixture was used to maintain anesthesia after intubation. Preoperatively, all sheep received tetanus serum (Tetanus Serum Intervet, MSD Animal Health GmBH, 3 mL, s.c.) as well as 1.2 g of amoxicillin/clavulanic acid (Co-Amoxi-Mepha, Mepha Pharma AG). 5 mL of progesterone (Progesterone Stricker ad. us. vet., 25 mg/mL, Werner Stricker AG) was administered intramuscularly. Throughout the intervention, a constant rate infusion of 5 μg/kg BW/h i.v. sufentanil (Sufenta Forte®, Janssen-Cilag AG) and 8 mg/kg/h, i.v. ketamine (Ketasol®-100, Graeub AG) was maintained. In the surgical area, an 8–10 mL ropivacaine infiltration block (Ropivacaine Sintetica 1%, Sintetica SA) was placed. Ringer solution was administered intravenously at an infusion rate of about 5 mL/kg/h for fluid substitution throughout the intervention.

Operation

During the surgery, the ewe was in dorsal recumbence. The uterus was exposed by an infraumbilical midline laparotomy, excorporated, and a uterotomy was performed to have access to the uterine cavity (N = 8 horns). A 10 French trocar (Check-Flo Performer® Introducer, Cook Medical) inserted in a 10 Fr sheath introducer (Avanti®+ Introducer, 10 Fr, Cordis®) was used to puncture the uterine wall and the FMs additionally under view in order to simulate the external puncture during fetoscopy. Then, immediately after premixing of MG precursor and NaIO₄, the loaded umbrella was inserted through the uterotomy and placed close to the FM, and the suture wires were pulled through the sheath introducer to press the implant against the FMs on the site of the iatrogenic FM lesion. A total of 10 implants were sutured on 3–4 points to the FMs and the uterus to avoid displacement and to secure the umbrellas to the FMs and myometrium (Fig. 2). After placing the umbrellas, AF was replaced by warm saline solution (0.9%; B.Braun Medical AG), and the uterotomy was closed with a two-layered suture (PDS Plus 2-0; Ethicon). All ewes survived the operation and showed no signs of distress after the surgery until 10 days post implantation. All details of the operation are summarized in Table 1.

Fig. 2.

Sealing of ovine FM defects. a Exposure of FMs through a hysterotomy. b Inserted catheter containing a tourniquet to (c) place the MG against the amnion. d Sutures going through the myometrium and the umbrella-shaped receptor to immobilize the implant at the site of treatment.

AF Leakage Test

To test AF leakage before the euthanasia, milk was injected with a syringe in the vicinity of the implant, and leakage was monitored by visual inspection.

Euthanasia

Ten to twelve days post implantation, the ewes were euthanized in deep general anesthesia with 75 mg/kg BW Na-pentobarbital intravenously. Fetuses were euthanized by intracardial injection of Na-pentobarbital.

Histological Analysis

FMs from the implant area and areas far away from treatment sites were pinned on cork pieces and placed in 4% formalin for 24 h, after which they were kept in 0.02% (w/v) sodium azide in PBS until further processing. The nitinol frames of the implants were then carefully removed, the tissues were embedded in paraffin, and consecutive sections of 4 μm were prepared. Serial sectioning was performed until the puncture site was localized. Slides were then deparaffinized, rehydrated through an alcohol series, and stained with hematoxylin and eosin staining following standard protocols.

For immunohistochemistry, slides were processed using an Autostainer Link 48 (Agilent Technologies Inc.). Online supplementary Table 1 (for all online suppl. material, see www.karger.com/doi/10.1159/000528473) summarizes the antibodies used and the specific details of each staining. Briefly, after deparaffinization, the adequate antigen retrieval was performed at 98°C for 20 min either with Target Retrieval Solution pH 6 (Agilent) or Target Retrieval Solution pH 9 (Agilent). Hydrogen peroxide (Merck) incubation for 10 min at RT was used to block endogenous peroxidase. Slides were washed 5 min with ddH₂O and 5 min with wash buffer (Agilent) and incubated with 2% normal goat serum (Vector). Then, primary antibodies were incubated for 60 min at RT. Slides were then washed with wash buffer (Agilent) and incubated with the corresponding detection systems conjugated to HRP (EnVision HRP, Agilent) for 20 min at RT. Sections were washed again with wash buffer for 5 min and incubated with 3,3'-diaminobenzidine substrate buffer (Agilent), which was used as a detection chromogen, for 10 min. Samples were counterstained with FLEX hematoxylin (Agilent) for 10 min, dehydrated, and mounted with Pertex (VWR Chemicals).

Semiquantitative Scoring of Immunohistological Sections

For the semiquantitative analysis of cell proliferation and immune cell recruitment in the vicinity of the implants, 3 samples treated with MG were compared to 3 control areas from the same uterine horn. The semiquantitative analysis was performed by scoring the samples 0 (none) to 3 (strong) response.

Imaging

Slides were scanned with a NanoZoomer-XR C12000 (Hamamatsu) and scans were processed in NPDI.Viewer2 (Hamamatsu).

Statistical Analysis

All data are reported as mean ± standard deviation. To compare the gelation time of MG, a one-way ANOVA with Tukey's multiple comparisons test was performed. For the statistical analysis of the semiquantitative scoring of IHC, unpaired t tests were performed. Statistical analysis was performed with GraphPad Prism (version 8.0.0).

Results

Optimization of MG Gelation Time

In this project, we used a previously developed elastic MG for the treatment of FM defects in a sheep model. This sealant consists of a 4-arm star-shaped amphiphilic PPO-PEO block copolymer that terminates with catechol moieties [18] (Fig. 1a). The cross-linking of this sealant in the presence of the oxidizing agent sodium periodate (NaIO₄) occurs within seconds, resulting in a biocompatible hydrogel with strong tissue adhesion [18]. To give sufficient time for handling and in vivo application, we tailored the gelation kinetics of the sealant by modifying the pH of the cross-linking solution. When reducing the pH from 7.4 to 6.0, gelation time increased from 15 ± 1 to 113 ± 3 s (Fig. 1b). In a compromise between a technically compatible gelation time and a pH in the physiological range, a pH of 7 (gelation time 23 ± 2 s) was selected as optimal.

Ewe and Fetal Performance

Next, to evaluate the in vivo biocompatibility, stability, and FM sealing capacity of the elastic MG, we adapted our earlier described sheep model for the closure of FM defects with sealing patches [26]. Here, an open surgery approach was chosen to enable the induction of the FM defect and the application of the umbrellas containing the gelling MG under sight and therefore in a tightly controlled manner. Additionally, to stably position the MG on the FM defect, the umbrella-shaped receptors were immobilized to the FMs by sutures through the myometrium (Fig. 2).

A total of 10 implants were placed in 8 horns from 5 ewes with a mean gestation age of 64 ± 4 days (Table 1). All ewes survived the operation without noteworthy complications, and none of them miscarried. At explantation, operation and suture-induced adhesions were observed on 4 out of 8 horns. Overall, fetal survival was 91% (10/11). Of note, in the uterine cavity of the deceased fetus, very small and poorly vascularized placentomes were observed. AF leakage tests revealed that all implant sites were tightly sealed, and no amnion bands or skin defects were observed in any of the fetuses.

Histological Examination of Implants

At explantation, a normal histopathological appearance was observed in the control tissue (Fig. 3a, top row). All umbrella-shaped implants were found at the site of the puncture. Hematoxylin and eosin-stained tissue sections revealed that elastic MG was present on all samples and could be observed on the FMs and throughout the created uterus defects (Fig. 3a, bottom row). There were no signs of degradation, indicating that the glue was resistant to spontaneous or proteolytic degradation during the observed 10-day exposure to AF, FMs, and uterine tissues. The MG was firmly adhering to both the myometrium and the FMs, and no cellular infiltration was observed in neither of them (Fig. 3a, bottom row). Semiquantitative scoring of Iba1 staining revealed similar numbers of macrophages in the FMs both for the control and the implanted sites (Fig. 3b, 4), but clusters of macrophages were observed surrounding the MG in the myometrial area (Fig. 3c). Furthermore, similar numbers of proliferating cells (assessed by Ki67 staining) were observed in the vicinity of the implant in the FMs (Fig. 3b, 4) and in higher numbers in the myometrial area than in the control tissue, as scored semiquantitatively (Fig. 3c, 4). Finally, smooth muscle actin (SMA) indicates the presence of activated fibroblasts and myofibroblasts in the FMs and of these and normal smooth muscle cells in the myometrium, which were similar in numbers both for the control and the tissue with implants (Fig. 3b, c, 4). Overall, the observed mild changes in tissue composition are compatible with a mild foreign body reaction to the implanted biomaterial.

Fig. 3.

Histological analysis of MG-loaded implants. a Representative hematoxylin and eosin staining of control and implantation sites. Left column scale bars, 1 mm; middle and right column scale bars, 250 μm. Black squares are representative of the FM areas depicted in (b); white squares are representative of the myometrium areas depicted in (c). b, c Representative immunohistochemistry images of the (b) FM area and c) myometrium in control and implantation sites for macrophages (Iba1), proliferating cells (Ki67), smooth muscle cells, activated fibroblasts, and myofibroblasts (smooth muscle actin [SMA]). Dotted lines mark the border between the implanted mussel glue (MG) and the native tissue. Scale bars: 50 μm.

Fig. 4.

Semiquantitative scoring of the tissue reaction to MG implants. Samples (n = 3 per condition) were scored from 0 (absence) to 3 (strong presence) of macrophages, proliferating cells, and activated cells in control samples and in the vicinity of MG implants. Data are depicted as individual data points ± SD. Unpaired t tests were performed. *p < 0.05, **p < 0.01 ***p < 0.001, ****p < 0.0001, n.s., nonsignificant; SD, standard deviation.

Discussion

Here, we describe the sealing of sheep FM defects with an elastic MG. To test its in vivo performance, we used a pregnant sheep model in combination with previously described polytetrafluoroethylene-coated nitinol implants [25]. Leakage and histological evaluations revealed that the elastic MG forms tight interactions with the amnion and closes uterine defects. Additionally, treatment with MG did not induce adverse effects on treated tissues, and the tissue response was compatible with a mild immune cell infiltration and cell activation, as observed by Ki67, Iba1, and smooth muscle actin stainings.

To seal defects in the sheep model, we used the elastic variant of the MG which has been previously shown to seal punctured FMs ex vivo with a similar efficiency as the more rigid MG [9]. Indeed, both glues were able to withstand contractions comparable to those experimented during labor when tested ex vivo [10]. While the more rigid MG in combination with decellularized human amnion plugs has shown beneficial sealing in the rabbit FM defect model [23], concerns regarding its long-term stability, especially in large animal models and humans, remain. The risk of failure due to FM growth and deformation is expected to be significantly reduced for the elastic MG since it is more deformable.

In this study, the very short gelation time of the elastic MG was prolonged by lowering the pH of the precursor solution. The use of precursor solutions at pH 7.0, a level which is still compatible with tissue physiology, enabled the deployment of the liquid biomaterial onto the FM defect. While histological examination revealed that MG tightly adhered to the surrounding tissue, this finding has to be confirmed with implants that are not secured to the myometrium. Not surprisingly, being a densely cross-linked material without cell adhesion or proteolytic degradation sites, the elastic MG did not show any signs of spontaneous degradation and was not infiltrated by cells, which confirms its previously described ex vivo [10] and in vivo [23] stability in AF.

In this study, a 100% ewe survival is reported, with no apparent consequences of the surgery to the ewes' health. Nonetheless, we noted the decease of one fetus, which was not attributed to the implanted biomaterial but rather the underdeveloped placentomes observed at explantation which could relate to the small size of the ewe. Importantly, we observed tight sealing of all implants at explantation, with no amniotic bands in any of the fetuses 10 days after implantation.

Compared to other animal models, the sheep pregnancy model has several advantages, including the comparable size of its uterus to the human, a gestation period which allows long-term observations, and the absence of spontaneous healing of FM defects [27, 28]. However, the occurrence of dispersed placentomes and FM foldings makes the selection of treatment sites by minimal invasive procedures difficult. By choosing the open-surgery approach employed here, these limitations were circumvented by selecting suitable treatment sites under view. Furthermore, the open surgery allowed to apply cross-linking-initiated MG accurately at the FM defect and to secure it in this position by sutures.

Taken together, the advantages of the MG material, the increased stability when compared to natural occurring materials such as amniopatch [29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39], maternal blood cloth patches [40], collagen plugs [41] and slurry [42], absorbable gelatin plugs [43, 44], and FG [45] − all of which have been tested in patients − make this biomimetic glue a promising biomaterial for FM sealing. Long-term studies, ideally lasting until the time of delivery, would be the next logical step and should be performed to assess if MG can maintain a tight fetal cavity until the end of pregnancy and is hence a strong candidate for the prevention of iPPROM. Furthermore, further optimization of the material is needed for treatments in a minimally invasive manner using recently developed devices [25, 46]. For instance, the biomaterial's viscosity and gelation time should be optimized to allow the injection of the premixed components necessary for MG distribution and gelation. After that, studies with the haplorhini suborder of primates would pose translatable models since they are big in size, have long gestational periods, and the same placentation system as humans [47].

Conclusion

In conclusion, the feasibility and effectiveness of employing an engineered MG for FM sealing have been demonstrated in a sheep model. Further optimization of the biomaterial and experiments that examine the long-term performance of the MG employed here are needed.

Statement of Ethics

Animal care and housing was performed according to protocols approved by the Cantonal Veterinary Office of Zurich, Switzerland (license number ZH214/18), were in accordance with the Swiss Animal Protection law, and conformed with the European Directive 2010/63/EU of the European Parliament and of the Council of September 22, 2010, on the Protection of Animals Used for Scientific Purposes and to the Guide for the Care and Use of Laboratory Animals. For this study, 5 Swiss White Alpine (SWA) ewes were used at gestational ages of 56–69 days (Table 1).

Conflict of Interest Statement

Yannick R. Devaud is a founder and shareholder of KOVE medical AG, which developed the umbrella receptor technology. Yannick R. Devaud, Martin Ehrbar, and Nicole Ochsenbein-Kölble are co-inventors of the umbrella receptor technology described in patent application EP17154743.3A filed in 2017. All other authors declare to have no conflict of interest.

Funding Sources

This research was supported by Swiss National Science Foundation grant 31003A_141051/2, the Innosuisse grant 55482.1 IP-LS, and the National Institutes of Health (USA) grant R01 EB022031.

Author Contributions

Eva Avilla-Royo designed, performed, and analyzed the in vitro experiments and was responsible for the design of the in vivo experiments together with Martin Ehrbar and Nicole Ochsenbein-Kölble. Eva Avilla-Royo performed the optimization and preparation of the materials for in vivo experiments and processed all the implants after explantation. Yannick R. Devaud developed the umbrellas and supported the preparation of materials for in vivo experiments. Frauke Seehusen and Josep M. Monné Rodriguez interpreted the histological specimens, and Frauke Seehusen scored the immunohistochemistry samples. Miriam Weisskopf and team supported the animal surgeries and were in charge of animal caretaking. Nele Strübing assisted during surgeries. Phillip B. Messersmith produced the mussel glue precursor. Ueli Moehrlen, Ladina Vonzun, Martin Ehrbar, and Nicole Ochsenbein-Kölble performed the animal surgeries. The project was supervised by Ladina Vonzun, Martin Ehrbar, and Nicole Ochsenbein-Kölble. Martin Ehrbar and Nicole Ochsenbein-Kölble acquired funding. Eva Avilla-Royo wrote the initial draft, and all the authors contributed to writing and manuscript revision.

Data Availability Statement

All data generated during this study is included in this article and its online supplementary material. Further questions can be addressed to the corresponding authors.

Supplementary Material

Supplementary data

Funding Statement

This research was supported by Swiss National Science Foundation grant 31003A_141051/2, the Innosuisse grant 55482.1 IP-LS, and the National Institutes of Health (USA) grant R01 EB022031.