Since the first publications on the artificial ovary, many advances have been made. At that time, no studies had yet proved the feasibility of the concept, so it might have appeared provocative to unable to propose this option as a future fertility restoration approach for women unable to undergo ovarian tissue transplantation. Today, different artificial ovary prototypes demonstrate that this technique is able to restore both endocrine and reproductive functions in a murine model, producing healthy pups. However, with a view to future clinical application, further studies with human ovarian tissue and human follicles are essential. The aim of this chapter is to highlight the latest developments and advances in the field, as well as future directions for prompt translation to a clinical setting.
Since the first publications on the artificial ovary, many advances have been made. At that time, no studies had yet proved the feasibility of the concept, so it might have appeared provocative to be unable to propose this option as a future fertility restoration approach for women unable to undergo ovarian tissue transplantation. Today, different artificial ovary prototypes demonstrate that this technique is able to restore both endocrine and reproductive functions in a murine model, producing healthy pups. However, with a view to future clinical application, further studies with human ovarian tissue and human follicles are essential. The aim of this chapter is to highlight the latest developments and advances in the field, as well as future directions for prompt translation to a clinical setting.
The Artificial Ovary as a Strategy to Reestablish Fertility
For cancer patients of childbearing age, fertility restoration often becomes an important concern after disease remission. In women, treatments such as chemo/radiotherapy can be very harmful to the ovaries, often causing loss of endocrine and reproductive functions, which results in premature menopause and emotional distress. An option for these patients is utilization of primordial follicles enclosed in ovarian cortex, since they represent 90% of the ovarian follicle reserve and there are no oncologic, legal or ethical restrictions to their use. Before a patient begins cancer treatment, a sample of her ovarian tissue containing primordial follicles can be removed and cryopreserved. Then, after disease remission, the tissue fragments can be grafted back to the patient. This technique has led to successful ovarian function restoration, as well as pregnancy .
Transplantation has so far been the only option to reestablish ovarian function from cryopreserved ovarian tissue in cancer survivors. Despite the promising results mentioned earlier, there is a legitimate concern about the possible presence of malignant cells in the frozen-thawed fragments, which could provoke a recurrence of the primary disease after reimplantation . Although many types of cancer never metastasize to the ovaries, leukemia is systemic in nature and poses a greater threat to the patient, while breast cancer is classed as moderate risk. Therefore, for patients diagnosed with these types of cancer, transplantation of ovarian tissue after disease remission is not advisable. Unfortunately, for these women, it is not yet possible to transplant their tissue after their health is restored.
A safe alternative for leukemia and breast cancer patients would be utilization of isolated follicles. Since the basement membrane encapsulating the ovarian follicle excludes capillaries, white blood cells and nerve processes from the granulosa compartment , grafting fully isolated follicles could be considered safer for these patients.
Although the main concern is being able to provide an alternative to restore fertility in patients who cannot benefit from ovarian tissue transplantation, the assembly and grafting of an artificial ovary would offer additional advantages compared to transplantation of ovarian tissue.
After removal, freezing, thawing and transplantation, ovarian tissue is subjected to hypoxia in the first days post-grafting, and this deprivation of oxygen and nutrients, as well as accumulation of metabolic waste, may lead to cellular damage. Indeed, it has been estimated that a significant percentage (50–95%) of primordial follicles may be lost due to ischemia [4–6], which would directly affect the lifespan of the graft. Therefore, the success of primordial follicle transplantation depends on the growth of new blood vessels in order to restore adequate perfusion. This may be improved by grafting endothelial cells together with isolated follicles [7, 8]. Moreover, using a matrix with an interconnected pore network could also enhance vascularization, while appropriate pore size and distribution would facilitate the diffusion of metabolites, oxygen and growth factors , which would have a positive effect on follicle survival and development. Angiogenic factors could also be added to the matrix, either chemically immobilized or physically entrapped. Sun et al.  reported that porous heparinized collagen/chitosan matrix combined with recombinant human granulocyte-macrophage colony-stimulating factor could improve cellular adhesion and migration, new vessel formation. Basic fibroblast growth factor (bFGF) was also shown to have a positive effect on vascularization in different studies. Jiang et al.  incorporated poly(lactic-co-glycolic acid) (PGLA) nanoparticles loaded with bFGF and vascular endothelial growth factor (VEGF) in bladder acellular matrix and reported an improvement in microvascular density and maturity in the grafts. Peters et al.  observed almost fourfold faster vascularization when PLGA microspheres were loaded with b-FBF. These authors reported that released b-FGF induced the formation of large and mature blood vessels in matrices implanted in the mesenteric membrane of rats. According to Bergmann and West , these factors could well influence the differentiation of mesenchymal stem cells from the bloodstream into endothelial cells and encourage microvascularization.
Premature recruitment of primordial follicles has also been suggested after grafting of ovarian tissue, possibly caused by a deficiency in inhibitory mechanisms implicated in the quiescence of primordial follicles in intact ovaries . This is indeed very likely, since tissue collected for cryopreservation and transplantation comes from the ovarian cortex, where the vast majority of the follicular population is made up of primordial and primary follicles and there is a lack of larger follicles responsible for the production of inhibitory factors like anti-Müllerian hormone (AMH) and activin A. Using a matrix, follicular activation and development may be modulated through supplementation of inhibitory and growth factors implicated in the different stages of folliculogenesis. These factors could, for example, be added encapsulated in materials with different degradation rates, according to requirements after transplantation.
Apart from the interaction between granulosa cells (GCs) and oocytes, follicles require neighboring stromal cells to support their growth. These cells are recruited to differentiate into theca cells, which play an essential role in follicular development through secretion of androgens as well as improving perifollicular vascularity. Although freezing of ovarian tissue does not negatively affect the morphology or ultrastructural characteristics of primordial follicles , it is harmful to surrounding tissue, causing damage to the extracellular matrix (ECM) and stromal cell necrosis , resulting in large areas of fibrosis . The poor cellularity of tissue after freezing may influence the development of follicles, and could be involved in the lack of a structured thecal layer around secondary follicles and asynchrony between oocyte and follicular cell growth . Therefore, in order to improve follicular development in the matrix, a fragment of ovarian tissue could be removed before the matrix is grafted, with the aim of isolating fresh stromal cells. These cells would then be encapsulated with isolated follicles in the matrix before transplantation (Figure 33.1).
Due to the random distribution of primordial follicles in ovarian cortex, it is not possible to determine the number, or even the presence, of follicles in ovarian tissue to be transplanted to a patient , which could affect the lifespan of the graft. Grafting isolated follicles would allow not only the introduction of a high and known number of follicles into the host , but also assessment of follicular quality before grafting .
The main objectives of an artificial ovary are (1) to safely transplant isolated primordial and primary follicles, avoiding inadvertent contamination by malignant cells possibly present in ovarian tissue that could potentially lead to recurrence of the primary disease, and (2) to mimic as much as possible the human ovary in order to support follicle survival and development after transplantation, ensuring secretion of sex hormones and production of fertilizable mature oocytes.
Isolation of Ovarian Pre-antral Follicles
As mentioned earlier, the first essential step for creation of a bioengineered ovary is ovarian follicle isolation. This procedure allows dissociation of intact follicles from the surrounding ovarian stroma thanks to the presence of a thin layer of basal lamina which serves to delimit each follicle and thereby maintain its spherical architecture  (Figure 33.2). Technically, a number of mechanical and enzymatic techniques and combinations of both have been described for ovarian follicle isolation [23–26]. However, due to the fibrous nature of human ovarian cortex where the majority of primordial follicles reside, the most effective way of obtaining high follicle yield and quality is usually based on a combination of mechanical and enzymatic tissue digestion . Nevertheless, with a view to future clinical application, a further consideration needs to be addressed. No fragment of ovarian tissue is the same as another in terms of follicle density and follicle distribution, and this variability is even greater between different women and, especially, different ages [28, 29]. An optimal human ovarian follicle isolation protocol should therefore take into account all possible variations that may be encountered in the ovarian tissue of different patients. With this is mind, we recently adapted our previously established follicle isolation protocol . In our new study , we fractionated the timing of enzymatic tissue digestion into three different time points in order to protect the first fully isolated follicles from over-digestion, and then proceed with the protocol only for the remaining undigested tissue fragments. Thanks to this modified protocol, we were able to demonstrate that even after just a round of enzymatic digestion (30 minutes), 35% of follicles were completely isolated. Indeed, at the end of the procedure with the modified protocol, we obtained an overall greater number of pre-antral follicles (average follicle number/patient: X= +33 follicles compared to the established protocol), which also exhibited better follicle survival after grafting than in a previous study with the established protocol . These findings led us to conclude that a personalized ovarian follicle isolation approach is feasible and should be implemented on a case by case basis in order to make best use of precious ovarian follicles and hence maximize the chances of pregnancy after artificial ovary transplantation. Moreover, due to the very small size of their ovaries and the high follicle density in very young patients, some primordial and primary follicles may also be found inside the inner and softer ovarian medulla . For this reason, it should be considered an additional source of ovarian follicles for the artificial ovary, especially in very young girls.
Before contemplating clinical application, experimental studies have to provide robust evidence of the safety of the procedures involved. Therefore, the artificial ovary has to be proven safe in its entirety, starting with the ovarian follicle isolation protocol. To this end, different studies were conducted in our laboratory [24, 30] in order to replace the use of collagenases, which are produced from Clostridium histolyticum and known to be a source of endotoxins, with a new generation of enzymes like Liberase. These enzymes are free of endotoxins and, most importantly, produced following guidelines for good manufacturing practice (GMP) and therefore ready for clinical application. This requirement was recently met when DNase I was added to our personalized ovarian follicle isolation protocol .
Another important concern regarding the safety of the follicle isolation procedure is possible contamination of the suspension of isolated follicles by malignant cells. Although all early-stage ovarian follicles are externally delimited by the basal lamina that physically separates them from the surrounding ovarian stroma, we cannot exclude potential contamination. Indeed, in experimental practice, when ovarian follicles are isolated from a suspension of digested ovarian tissue, some residual malignant cells may remain attached to the basal membrane or be inadvertently picked up along with the follicles and then grafted back to the patient. For this reason, studies were recently performed in an attempt to ascertain the safety of the procedure [33–35]. Soares et al.  demonstrated that despite follicle pick-up from an ovarian tissue suspension previously contaminated with a leukemia cell line (BV-173) being deemed unsafe, a simple follicle purging step based on three consecutive washes in saline solution was enough to remove any contaminating cells. However, since interaction between ovarian follicles and naturally resident leukemia cells may differ from that in artificially contaminated ovarian tissue suspensions, the authors initiated a similar study, but this time with ovarian tissue from leukemia patients . Using highly sensitive polymerase chain reaction (PCR) with a detection limit of two cells, the authors found that both ovarian tissue and unpurged isolated follicles were positive for leukemia contamination, but after follicle purging, no leukemic cells were found. This demonstrated the safety of the follicle isolation procedure, even with tissue from nine leukemia patients .
Another important study  evaluated the tumor-inducing ability of as few as 10–100 leukemia cells (BV-173) encapsulated inside an artificial ovary prototype and xenografted to immunodeficient mice for 20 weeks. Interestingly, only mice grafted with 3×106 cells (positive control) developed systemic disease within 4 weeks, while none of the animals grafted with 10 or 100 BV-173 cells developed tumors after 4 months of transplantation. These results suggest that even in case of inadvertent grafting of as many as 100 BV-173 cells inside an artificial ovary, no malignancy was induced, even in an immunodeficient mouse model. This provides further reassurance about the safety of the bioengineered ovary.
Mimicking the Human Ovary
As in the case of a natural ovary, the main goal of a bioengineered artificial ovary is to offer an environment that allows folliculogenesis to occur. Therefore, just like the natural organ, the matrix should (1) ensure proper communication between follicles and ovarian cells; (2) preserve their interaction with the ECM; (3) supply factors involved in follicular survival and development; and (4) maintain the original structure of follicles. In other words, the matrix should spatially and temporally mimic the ECM, the natural scaffold of the ovary. In order to do so, it should include some design parameters, such as interaction with cells, physical support for follicles, matrix porosity and supporting cell populations, which are all interconnected and influence each other. It must also be biocompatible and, from a practical point of view, capable of being handled.
Citing von der Mark et al. , “cells are surrounded by a wealth of information provided by the ECM, which presents adhesive and bioactive peptide epitopes located in matrix macromolecules and smaller glycoproteins, plus growth factors and cytokines trapped and sequestered by the matrix”. The ECM thus plays an essential role in cell fate: it regulates cell morphology, proliferation, migration, differentiation, orientation, production and secretion of molecules, and even death. For this reason, the artificial ovary matrix should modulate the interactions of cells and follicles, supporting cell adhesion, proliferation, migration and production of matrix proteins necessary to form a substrate for new cells required for follicular development. For follicles, the matrix should act as a supporting scaffold, preserving their original 3D structure and intercellular interactions between GCs and oocytes, which is essential to regulate follicular growth and development. The 3D arrangement of follicles is also influenced by cell migration and proliferation, induced by interaction of the matrix and cells. Moreover, for folliculogenesis to occur, it is necessary to maintain contact between GCs and oocytes because many aspects of oocyte growth and development are regulated by interactions with adjacent GCs [38, 39]. A rupture in the GC–oocyte connection would lead to uncoordinated growth and differentiation of somatic and germ cells .
In order to avoid breakdown of the metabolic link between GCs and oocytes, follicles need to maintain their 3D structure. Three-dimensional matrices would be able to effectively mimic physiological conditions, since many cellular processes in organogenesis occur exclusively in 3D . Previous studies on in vitro culture of isolated human follicles have shown that preservation of their 3D structure positively affects their survival and growth [19, 42].
While the matrix needs to be able to support the 3D structure of follicles, it should not be excessively stiff to prevent their exponential growth. An extremely rigid scaffold may decrease the proliferation rate of GCs and oocyte growth, and affect actin organization in growing follicles , which could lead to diminished follicular growth and even apoptosis.
It is also important to stress that ovarian follicles are exceptional in that they can grow to around 600x their size during folliculogenesis (the human follicle grows from 30 μm in its primordial stage, to 18 – 24 mm when it is ready to ovulate). In addition, they recruit cells and vessels to support their development. An ideal scaffold would need to degrade in order to allow exponential growth of follicles, formation of vessels, and proliferation of stromal cells. Ideally, the artificial ovary should offer an appropriate initial environment for follicles that would be replaced by a new “ovarian-like” structure after a few weeks of grafting. The degradation rate of the scaffold is thus an essential parameter in the success of grafting.
Although biodegradation should be conceived with follicular development in mind, other factors should also be taken into account. The biodegradation rate cannot be faster than cell migration and proliferation or ECM synthesis and stabilization in the scaffold, because cells would lose physiochemical factors for tissue regeneration  and isolated follicles would lose their 3D support. On the other hand, slower degradation would inhibit cell penetration and consequently ECM formation in the matrix  and negatively affect follicular growth. Therefore, the material should be carefully selected in order to control the desired degradation rate.
Since the dawn of the artificial organ age, one of the main questions that researchers have asked themselves is “what is the most appropriate material to replicate the architecture and functionality of the natural organ?” Numerous experiments using different polymers have been performed in recent decades, and suggest that each organ is a unique entity in terms of its microarchitecture and functionality, so the requirements for its assembly cannot be generalized . A wide range of different synthetic and natural matrices [45–52] have therefore been tested for development of an artificial ovary prototype.
The only synthetic polymer that has been utilized to graft pre-antral follicles is poly(ethylene glycol) (PEG) [45–47]. PEG is a linear-chained polymer consisting of a simple oxygen-carbon-carbon repeating unit. Shikanov’s group has successfully used PEG gels modified with Arg-Gly-Asp and the protease-sensitive peptide Leu-Gly-Pro-Ala to graft mouse pre-antral follicles, showing that this polymer allows follicle survival and development. However, preliminary experiments with different types of PEG [45, 48] have revealed that this polymer does not seem to support survival of isolated human pre-antral follicles (Chiti and Amorim, unpublished results).
The majority of studies on bioengineered ovaries in the literature have been used as natural materials (Table 33.1), probably because of their superior biocompatibility . Nevertheless, many other features need to be taken into account when selecting the optimal candidate for a bioengineered organ. At the start, very promising results were obtained when isolated mouse follicles were encapsulated in collagen  or plasma clot [50, 55]. The pioneering study of Telfer et al.  using collagen to in vitro culture and then transplant isolated preantral follicles resulted in embryos that were obtained after in vitro-fertilized oocytes recovered from large follicles 10 days post-grafting. Gosden’s [50, 55] studies went even further, reporting the birth of pups after transplantation of isolated follicles in plasma clots.
|Types of matrix|
Based on the successful results reported with mouse follicles encapsulated in plasma clot [50, 55], Dolmans et al. [56, 57] encapsulated isolated human pre-antral follicles in autologous plasma clots and xenografted them to immunodeficient mice for one week or five months. After short-term grafting , the follicles were able to develop to the secondary stage, while after long-term transplantation , antral follicles were also found in the grafts. In spite of these promising results, due to the great variability in plasma composition and unpredictable degradation rate after grafting, the reproducibility of results, which is essential for future clinical application, might not have been guaranteed, so other materials had to be explored.
In order to overcome the encountered drawbacks, Vanacker et al. [58, 59] investigated alginate, a natural polysaccharide obtained from brown algae. The authors conducted two different autografting studies, first with isolated mouse ovarian cells  and then with isolated mouse follicles . In the second study, despite demonstrating that a GMP-compliant alginate-based matrix was able to support mouse follicle survival (21%) and growth after 7 days of transplantation, with 100% recovery of their alginate beads, matrix degradation and particularly revascularization appeared to be limited after grafting. These latter observations presented new obstacles to possible future clinical translation. Indeed, it was already well known that the longer the hypoxic period after grafting, the greater the follicle loss seen during the first few days post-transplantation . The biomaterial of choice must therefore have a controlled degradation rate, but at the same time allow cell infiltration and revascularization after transplantation.
Another natural material, namely fibrin, was also emerging in the field of tissue engineering thanks to its already numerous clinical applications , encouraging different teams to investigate this new scaffold for construction of the artificial ovary prototype. Luyckx et al. [61, 62] were the first to demonstrate that use of a fibrin-based matrix with low concentrations of fibrinogen and thrombin (12.5 mg /ml fibrinogen and 1 IU/ml thrombin) resulted in greater mouse follicle survival (31%) and better vascularization after short-term autografting than did an earlier alginate-based matrix . A few years later, another study  with mouse follicles encapsulated in fibrin demonstrated that addition of collagen or alginate to the fibrin-based matrix did not improve follicle survival after 2 days of in vitro culture, while association with VEGF led to faster restoration of endocrine and even reproductive functions after grafting to previously ovariectomized mice. Hence, according to these very promising findings, fibrin appeared to be the optimal candidate for creation of a bioengineered ovary prototype. Further studies with human follicles were nevertheless required to confirm its efficacy. Unfortunately, when human follicles were encapsulated in a fibrin-based scaffold with low concentrations of fibrinogen and thrombin  and xenografted to immunodeficient mice, the results were disappointing (Amorim, unpublished results). In order to elucidate the cause of this discrepancy, new investigations with mouse [64, 65] and human  follicles were initiated by the same team. Importantly, Paulini et al.  hypothesized that due to the denser texture of human ovarian cortex compared to mouse ovarian tissue, the consistency of the fibrin-based matrix may also play a role in human follicle survival. With this in mind, the authors tested a new, more rigid fibrin-based matrix (50 mg /ml fibrinogen and 10 IU/ml thrombin) for human ovarian follicle encapsulation and short-term xenografting. After one week of transplantation, they found a higher human follicle recovery rate (21%) than with the previous fibrin combination (12.5 mg /ml fibrinogen and 1 IU/ml thrombin), suggesting that the requirements for human follicle encapsulation and grafting may well be different from those of mouse follicles in an artificial ovary prototype . This hypothesis was recently confirmed by a new study , where increasing combinations of fibrinogen (12, 30, 50 and 75 mg/ml) and thrombin (1, 50 and 75 IU/ml) were comprehensively investigated by scanning electron microscopy and rheology and compared with human ovarian cortex. The authors demonstrated that the fibrin matrix with the lowest concentrations of fibrinogen and thrombin (12.5 mg /ml fibrinogen and 1 IU/ml thrombin)  significantly differed from human ovarian cortex in terms of fiber thickness and also rigidity, which is why a new combination (50 mg /ml fibrinogen and 50 IU/ml thrombin) more closely resembling human ovarian tissue microarchitecture was chosen for further investigation in longer-term xenografting studies .
Another emerging frontier in tissue engineering is decellularization of natural tissues and organs using a combination of chemical, enzymatic and physical processes, the final goal of removing all naturally resident cells to obtain an allogeneic or xenogeneic cell-free ECM. These natural scaffolds, unlike the others, should retain the physiological architecture and biochemical composition of the original tissue, and therefore constitute a better environment for ovarian follicles and cells. Only a few very recent studies [67–70] on this novel strategy were found in the literature. Current data already show that whole ovaries or ovarian tissue fragments from different species, including humans, can be successfully decellularized and used directly as scaffolds [68–70] or thermosensitive hydrogels  to allow follicle survival in different experimental models. All this is promising, but further optimization of decellularization techniques is required to limit the harshness of decellularization protocols in order to retain as many recently identified proteins as possible in the human ovarian ECM . Although it is too early to draw any definitive conclusions on the most suitable scaffold for creation of a bioengineered ovary, great advances have been made, in just a few years. This will allow researchers to reach fundamental goals and acquire essential knowledge on the requisites of human ovarian follicles and conditions for their survival and development in an artificial ovary prototype.