Ovary-derived Decellularized Extracellular Matrix-based Bioink for Fabricating 3D Primary Ovarian Cells-laden Structures for Mouse Ovarian Failure Correction

Fertility preservation is becoming a clinical duty in practice. Three-dimensional (3D) bioprinting technology is potentially realize ovarian morphological repair and reproductive endocrine function rebuild. There is no published work on 3D bioprinting ovary using a decellularized extracellular matrix (dECM)-based bioink, though dECM is the preferred matrix choice for an artificial ovary. The study aimed to explore swine ovarian dECM-based bioink to fabricate 3D primary ovarian cells (POCs)-laden structures for mouse ovarian failure correction. In this study, the ovarian dECM was converted to dECM-based bioink by dECM solution mixed with a seaweed gelatin blend solution of bioink that was characterized using scanning electron microscopy, circular dichroism, rheology, hematoxylin and eosin staining, and immunohistochemistry. The 3D scaffolds were, then, printed with or without POCs by the extrusion 3D bioprinter. The laden POCs viability was detected with the live/dead assay kit. A female castrated mouse model was established, and the mice were treated with five different methods. The results revealed that the 3D scaffold encapsulating POCs group had more positive signals in neoangiogenesis, cell proliferation and survival than the 3D scaffold group, and ensured sex hormone secretion. Meanwhile, the expression of germ cells in the 3D scaffold encapsulating POCs group was more intensely than the non-printed hydrogel encapsulating POCs group. The work shows that the 3D bioprinting ovary employing ovarian dECM-based bioink is a promising approach for ovarian failure correction.


Introduction
Fertility issues have become a crucial problem to an increasing number of women of reproductive age with malignancies [1] . While cryopreservation and transplantation of ovarian tissue is increasingly recognized as a method to restore fertility, it is not an infallible method for certain types of cancer cells due to the risk of implantation [2] . In recent years, tissue engineering techniques may provide an approach to solve the clinical problems by constructing bioengineered ovaries with an appropriate cell-and tissue-specific bioinks.
Bioengineered ovaries must mimic natural organs. Besides isolated follicles, it also requires autologous ovarian cells, which are required for follicle survival [3] . The follicles are separated from the stromal elements by encapsulating themselves in a basement membrane, and autologous ovarian cells are derived from cured ovarian samples [4] . Thus, it can be considered a safer means of restoring fertility in women with cancer.
Today, the materials used to construct bioengineered ovaries include both natural and synthetic polymers [5,6] . However, studies reported that the matrix of choice for artificial ovaries is the decellularized extracellular matrix (dECM) [7] , because synthetic polymeric materials do not possess every property of dECM [8] . Ovary dECM materials for bioengineering ovaries have produced some promising results. Laronda et al. [9] first successfully constructed scaffolds from acellular bovine and human ovarian tissues to support follicle growth and restoration of ovarian function in ovariectomized mice. Another study produced a decellularized porcine ovarian matrix that supported the survival of rat granulosa cells in vitro and improved estradiol hormone secretion [10] . Hassanpour et al. [11] seeded rat primary ovarian cells (POCs) on the decellularized human ovarian matrix and found follicle-like structures within the matrix 4 weeks after transplantation. Pors et al. [12] investigated human preantral follicles seeded on the human ovary dECM. In vivo assessment showed that the survival of follicle was higher in the decellularized human ovarian scaffolds after 3 weeks of xenografting in mice. Despite these promising results with ovarian dECMs, it is challenging to find a precise fit with follicles of different sizes in the pores of dECMs. An alternative approach is to convert ovarian dECMs into a temperature-sensitive hydrogel. This method can perfectly encapsulate isolated follicles or other ovarian cells while maintaining acellular ovarian tissue components. Chiti et al. [13] converted bovine ovarian dECMs into hydrogel and demonstrated that mouse preantral follicles were able to survive in the hydrogel.
The traditional tissue engineering techniques described above, such as seeding cells on dECMs or encapsulating cells in hydrogels, can mimic physiological ovarian tissue and improve ovarian function to some extent. However, precise control of the spatial distribution, oxygen diffusion, or cell structure between cells and matrix remains a challenge for conventional tissue engineering. A solution to these problems are three-dimensional (3D) bioprinting technology [14] . 3D bioprinting technology has the potential to achieve ovarian morphological repair and reproductive endocrine function rebuild. The material used for 3D bioprinting is called "bioink." Bioink consists of individual living cells or living cells with a supporting hydrogel component [15] . At present, there is no published work on 3D bioprinting of ovaries using bioink composed of dECMs. However, 3D bioprinting should not be confused with 3D printing of biomaterials. The former refers to the printing process of live cells, while the latter refers to the printing of biomaterials, which does not require live cell printing (the printed scaffolds can be seeded with live cells) [16] . The 3D-bioprinted cell-loaded scaffolds possess ideal spatial distribution. Thus, we hypothesize that 3D bioengineering ovarian constructs using ovarian dECM-based bioink for tissue infiltration and target tissue remodeling will facilitate cell distribution and survival.
In this study, we focused on 3D bioprinting porous cylindrical-shaped ovarian constructs employing swine ovarian dECM-based bioink encapsulating POCs to evaluate the efficacy of ovarian failure correction.

Animals
One hundred and twenty slaughterhouse-raised female swines (95 -100 kg, aged 6 months) were used to harvest the fresh ovary tissues. Twelve female Kunming mice (16 -20 g, aged 8 weeks) were used to determine the biocompatibility of the dECM-based bioink. One hundred and ninety female Kunming mice (13 -15 g, aged 4 weeks) were used to prepare the POCs and animal model. All procedures involving animals were conducted in compliance with the guidelines of the local animal ethics committee on animal care (No. 2019-P060).

Decellularization of ovarian tissues
Fresh swine ovarian tissues decellularization was conducted based on previous work [17] . First, the ovaries were cut into pieces (3 mm thick), cleaned with normal saline, and treated with phenylmethylsulfonyl fluoride (PMSF) solution (0.1 mM/L) in a shaker (130 rpm) for 48 h at 4°C, which could inhibit protease activity. Second, the tissues were placed into a hypotonic Tris buffer (pH 8.0) containing 0.1% sodium dodecyl sulfate (SDS) and 0.1 mM/L PMSF for 12 h at 4°C. Third, the tissues were submerged in Tris-buffered saline containing 0.1 mM/L PMSF and 1% Triton X-100 solution by continuous shaking (130 rpm) at 4°C for 7 days. Fourth, the tissues were soaked in nuclease solution (pH 7.5) containing 50 U/ml deoxyribonuclease I (Sigma, Poole, UK) and 1 U/ml ribonuclease A (Sigma, Poole, UK) by shaking (80 rpm) at 37°C for 12.5 h. Then, the tissues were immersed in 0.1% peroxyacetic acid and 20% ethanol at 4°C for 2 h, and finally freeze-dried. The ovarian dECM was then formed.

Decellularization evaluation
The residual DNA content of the dECMs was measured to evaluate the degree of decellularization. The total DNA was extracted using TIANamp genomic DNA kit (Tiangen, Beijing, China). The DNA concentration was quantified by NanoDrop 2000 spectrophotometry (Thermo Scientific, USA) and the size of the remaining DNA fragments was verified by gel electrophoresis.
Fresh ovarian tissues and the dECMs were embedded in paraffin. The cell residues in the tissues were detected using hematoxylin and eosin (H&E) staining and 4,6-diamidino-2-phenylindole (DAPI) staining. The dECMs components such as collagen and proteoglycan were assessed by Masson staining and toluidine blue (TB) staining. The presence of proteins or peptides was detected by SDS-polyacrylamide gel electrophoresis (SDS-PAGE). Routine electrophoresis, dyeing, and discoloration were performed in turn. The microstructure of the dECMs was observed using scanning electron microscopy (SEM; JSM-5600LV, JEOL).

Preparation of dECM-based bioink
The dECMs were pulverized using a small grinder with the help of liquid nitrogen. 100 mg dECM powder was taken, 3 ml hydrochloric acid solution (pH 2.0) and 60 mg pepsin were added to the dECM powder and digested at 37°C for 24 h. The dECM solution pH was adjusted using NaOH solution (10 M) (from 3.2 -3.5 to 7.0 -7.2) after solubilization. Then, 2 ml tri-distilled water was added into 15% (w/v) gelatin and 3% (w/v) sodium alginate (Sigma-Aldrich; Merck) and dissolved at 55°C for 30 min. Finally, dECM-based bioink working solution was produced by mixing 3 ml dECM solution (pH 7.0 -7.2) with the above 2 ml solution.

The biochemical characterization and biocompatibility of the dECM-based bioink (1) Microarchitecture of the dECM-based bioink by SEM
SEM was performed to observe the microarchitecture of the bioink. First, the bioink was freeze-dried and then fixed in 2.5% glutaraldehyde at 4°C for 12 h. Second, the above samples were sequentially immersed in 70%, 80%, 90%, and 100% ethanol for 15 min. The samples were then coated with gold sputter and viewed under SEM.

(2) Circular dichroism (CD) spectra properties
To evaluate the protein structures and thermal stability of the bioink, we tested the bioink with CD, including room temperature CD and the variable temperature CD analysis. First, the baseline was measured, followed by the room temperature CD. The scanning speed was 60 nm, scanning band was 190 -260 nm. The average of three scan results after subtracting the baseline is the ellipticity. According to the results of CD at room temperature, the variable temperature CD analysis was performed. The detection wavelength was 199 nm, the starting temperature was 30°C, the ending temperature was 90°C, and the heating rate was 1°C/min. The thermal denaturation curve was, then, analyzed by a fitting equation model (a sigmoid curve) to obtain the thermal denaturation temperature (Tm) value.

(3) Rheological characterization
To assess the viscosity and the strain sweep of the bioink (before and after cross-linking of calcium chloride) and the dECM solution (pH 3.2 -3.5), we conducted rheological investigation on a rotation rheometer (Malvern Kinexus Ultra+) at 25°C. Amplitude sweep (0.01 -100 strain, 10% rad/s) was performed to assess the strain dependent storage modulus (G') and loss modulus (G'').

(4) Biocompatibility of the dECMs bioink in vivo
Twelve 8-week-old female Kunming mice were subcutaneously injected with 400 µl bioink on the back of each mouse. The injections were taken at the 1 st , 2 nd , 4 th , and 9 th week for H&E staining and immunohistochemistry (IHC) Staining (CD45, an inflammatory marker).

Isolation of the POCs
According to Hassanpour et al.'s protocol [11] , POCs were prepared from 4-week-old female Kunming mice and then encapsulated in the bioink. Briefly, each mouse was intraperitoneally injected with 10 IU Pregnant Mare Serum Gonadotropin (PMSG, Solarbio) followed by 16 U Chorionic Gonadotrophin for Injection (Harbin Sanma Animal Pharmaceutical Co. LTD) after 48 h, and then, the ovaries were isolated after 6 h. The ovaries were incubated in α-MEM medium (Gibco) with 1% penicillinstreptomycin (PS, Millipore) at 4°C. Afterward, the ovaries were, manually, dissected with fine needles into smaller fragments and incubated in a digestion solution consisting of Dispase II (Cat. No. D4693-1G, Sigma) and Collagenase I (Cat. No. 1904MG100, BioFROXX) at 37°C for 30 min. Then, the enzymatic process was terminated by equal volumes of α-MEM medium containing 10% FBS (BI). The suspension was filtered through a 100 µm cell strainer (Life Sciences, USA) and washed twice with the following culture medium: α-MEM medium with 10% FBS, 3 ng/ml Recombinant Murine Epidermal Growth Factor (Cat. no. 315-09, Pepro Tech), 100 mIU/ml Follicle Stimulating Hormone for Injection (Harbin Sanma Animal Pharmaceutical Co. LTD), and 1.5 U/ml chorionic gonadotrophin for injection and 1% International Journal of Bioprinting (2022)-Volume 8, Issue 3 PS. Cells were resuspended in 200 µl culture medium after centrifugation and then mixed well with the bioink for printing.

Printing of POCs-laden structure and culturing in vitro
A porous circular grid 3D scaffold (nozzle diameter was 340 µm, nozzle temperature was 20°C, and platform temperature was 4°C) was printed using a bioprinting system (Bio-Architect®-WS; Hangzhou Regenovo Biotechnology, Ltd.). During printing, the pneumatic pressure (0.18 -0.32 kPa) was adjusted to match the best nozzle scanning speed (6 mm). The bioink mixed with the POCs (1 × 10 6 cells/ml) was used to print. After printing, the POCs-laden 3D scaffolds (2 mm in height and 5 mm in diameter) were soaked in 5% calcium chloride solution (3 min) for cross-linking. Then, the POCs-laden 3D scaffolds were cultured in an incubator (37°C, 5% CO 2 ), and the POCs viability in the 3D scaffolds was tested with a live/dead assay kit (Molecular Probes, Inc., Cat. No. L3224) on the 1 st , 7 th , and 14 th day. Then, the cells were reviewed under a laser confocal microscope (Zeiss, Germany).

In vivo study: Female castrated mice model and grouping
The castrated mice model was performed as previously described [9] . Briefly, the mice with a regular oestrum cycle phase were operated on both sides of the back 1.0 cm incisions after being anesthetized successfully. The incisions were made over the ovaries location through the subcutaneous layers. Then, ovaries were removed from the top of the uterine horns and sutured the skin at the incision site. Two weeks after the operation, the vaginal orifice of ovariectomized mice was examined to determine whether the operation was successful or not. The incisions on both sides of the back of the mice were reopened and bluntly stripped to form a tunnel under the skin. The prepared grafts were implanted, and the incisions were sutured with absorbable sutures (Shanghai Pudong Jinhuan Medical Products Co., LTD).
Twenty mice were randomly divided into five groups (four mice in each group): (i) Sham-operated mice as the non-ovariectomized (OVX) group; (ii) ovariectomized mice without further treatment as the OVX-C group; (iii) ovariectomized mice receiving 3D scaffolds (8 × 8 × 3 mm 3 ) without cells as the 3D scaffold group; (iv) ovariectomized mice receiving 3D scaffolds with POCs (1 × 10 7 cells/ml) as the 3D scaffold encapsulating POCs group; and (v) ovariectomized mice receiving the bioink with POCs without printing as the hydrogel encapsulating POCs group.

Functional evaluation (1) Neoangiogenesis and cell proliferation
Three weeks after the surgery, vaginal smear was carried out with normal saline that was used to observe the change of each stage of the estrous cycle of the experimental mice. After 4 weeks of treating, the transplants were removed and routinely embedded them in paraffin and sectioned continuously. To evaluate the vascularization and cell proliferation of grafts in the above groups, IHC staining of a rabbit polyclonal antibody against the neovascular-specific cell surface markers CD31 (1:200; Servicebio Technology Co., Ltd., Cat. No. GB11063-2) and a rabbit polyclonal antibody Ki67 (1:100; Wanleibio Co.,Ltd., Cat. No. WL0280a) was performed. In addition, representative sections were stained with TUNEL following the manufacturer protocol (Servicebio™ Fluorescein Tunel Cell Apoptosis Detection Kit) to assess the survival of POCs in vivo.

(2) Sex hormones secretion
Blood samples were collected by eyeball on day 28 after transplantation. Then, the serum was separated overnight at 4°C. All blood samples were, then, immediately stored at −80°C until further analysis. The serum levels of estradiol (E 2 ), follicle stimulating hormone (FSH), and progesterone (P) were measured using a competitive enzyme-linked immunosorbent assay kit (CUSABIO). The sensitivity for E 2 , FSH, and P kits was 40 pg/ml, 2.5 mIU/ml, or 2.5 mIU/ml, respectively.

(3) Expression of POCs markers
To assess the specific markers of the POCs on the implanted grafts, immunofluorescence staining was done. The sections obtained from the above were dewaxed to water and rinsed in PBS (5 min

Statistical analysis
SPSS version 21.0 was used. Continuous variables are represented by mean ± standard deviation. The DNA content and IOD value were analyzed by a Student's t-test. For analyzing the hormone levels and the IOD value of CD31 of various groups, one-way analysis of variance following the LSD method between groups was performed. P <0.05 was considered as a significant difference.

Biochemical characterization of dECM
Decellularization of the ovary ECM is to minimize the loss and damage of ECM, while maximizing the removal of cellular materials. After physical and chemical treatment, the ovary color changed from red to white (Figure 1A). We analyzed the DNA content in native and decellularized tissues to assess the decellularization efficiency. As shown in Figure 1B, no DNA strip appeared in the dECMs, and DNA quantification showed that the DNA strips in the dECMs were significantly reduced (861.838 ± 18.06 vs. 48.48 ± 1.88 ng/mg, P < 0.0001). At the same time, H&E staining and DAPI staining confirmed that no cellular material remained after decellularization ( Figure 1C). Next, Masson staining and TB staining were performed to assess collagen and proteoglycan in the dECMs. As shown in Figure 1D, the compositions of the dECMs and native tissues share a very basic similarity. SDS-PAGE showed that there were various proteins in the dECMs ( Figure 1E). SEM revealed that the dECMs had no residual cellular components and the fiber orientation and structure were hardly affected ( Figure 1F).

Characterization and biocompatibility of the bioink
The generation of 3D ovarian tissue construction includes ovarian tissue decellularization, bioink preparation, and 3D bioprinting (Figure 2). Among them, the bioink is particularly important. SEM showed the bioink with a porous reticular microarchitecture ( Figure 3A). Porous diameters were measured to be 75.58 ± 35.64 µm. The results of CD spectrum ( Figure 3B) indicated that the bioink had a typical collagen triple helical conformation, showing an inverted S-shaped curve (the maximum positive and negative absorption peaks at 225 nm and 199 nm, respectively). From the thermal denaturation curve, it was known that the unwinding of the triple helix structure mainly occurred between 45°C and 70°C, and Tm was 59°C.
The rheological properties of the dECM solution and the bioink are shown in Figure 3C. Both the G' value of the dECM solution and the bioink were significantly higher than G", showing solid-like properties. G' value of the Biocompatibility of bioinks in vivo was evaluated by subcutaneous xenotransplantation. All mice were free of any complications within 9 weeks of implantation. At 1 st week, a potent angiogenesis response could be seen on the bioink tissues surface (Figure 3D). H&E staining showed that there were numerous inflammatory cells 1 week after implantation, principally aggregated in the  junction of the skin and the bioink. During the 2 nd week, inflammatory cells increased in the bioink. However, from the 2 nd to 9 th week, the total amount of inflammatory cells decreased, and the volume of bioink injected in the 9 th week was significantly lower than that in the 1 st week. Furthermore, the expression of CD45 (identify the inflammatory cells) was similar to the H&E staining results (Figure 3E).

Viability of POCs in the 3D scaffold in vitro
As shown in Figure 4A, the isolated POCs included a large number of stromal cells and follicles at different developmental stages. After culturing 2 weeks, the 3D scaffolds maintained their original shape, but appeared softer than they were in the beginning. On the other hand, the shear force generated during printing and residual chemicals in the dECM-based bioink may affect cellular viability. Therefore, the live/dead assay was performed to assess the survival status of the laden POCs. The assay showed that the POCs in the 3D scaffolds showed high cell viability on the 1 st day (>95%) and remained around 90% on the 14 th day ( Figure 4B).

In vivo evaluation of the printed ovarian tissue constructs
Four weeks after implantation, macroscopic observations showed a slower degradation of scaffolds compared with hydrogel, the above constructs maintained their shape, and functional blood vessels were developed ( Figure 5). IHC results demonstrated the transplanted tissues had neovascularization ( Figure 6A-C, red arrows). The positive signal of CD31 increased significantly in the group of 3D scaffold encapsulating POCs (0.0063 ± 0.0011) than in the other groups (0.0016±0.0009 in the hydrogel encapsulating POCs group and 0.0032 ± 0.0008 in the 3D scaffold group (P = 0.000 and 0.001). The expression of CD31 in the hydrogel encapsulating POCs group was lower than the 3D scaffold group (P = 0.042) (Figure 6G). POCs were immune stained with Ki67 to evaluate the cell proliferation (Figure 6D-F). More positive Ki67 signals (cells in brown) were detected in the 3D scaffold encapsulating POCs group (0.0074 ± 0.0017) than in the hydrogel encapsulating POCs group (0.0036 ± 0.0010 (P = 0.001) (Figure 6G). TUNEL staining was used to evaluate the cell apoptosis in the grafted tissues. The numbers of TUNEL-positive cells in both groups were similar ( Figure 6H).
Hormone evaluation showed that the levels of serum E 2 increased significantly in the 3D scaffold encapsulating POCs group (465.91 ± 24.77 pg/ml) than those in the OVX-C group (332.28 ± 26.17 pg/ml, P = 0.000) and in the 3D scaffold group (390.06 ± 41.47 pg/ml, P = 0.014) and were close to those in the non-OVX group (494.31 ± 35.96 pg/ml, P = 0.292) and in the hydrogel encapsulating POCs group (424.69 ± 24.26 pg/ml, P =0.138). In the 3D scaffold encapsulating POCs group, implantation led to the restoration of FSH (44.69 ± 24.17 mIU/ml) and P (2.55 ± 1.34 ng/ml) to the physiological level (FSH: 50.34 ± 2.73 mIU/ml; P: 3.35 ± 2.56 ng/ml). Serum P levels in the 3D scaffold group (12.75 ± 2.59 ng/ml) and the hydrogel encapsulating POCs group (14.52 ± 3.02 ng/ml) were similar to those of the OVX-C mice (12.75 ± 2.59 ng/ml). Serum E 2 and FSH levels in the group of 3D scaffold and hydrogel encapsulating POCs did not restore up to the physiological level compared to non-OVX group, although their serum E 2 and FSH levels were improved (Figure 7A-C).

B A
However, vaginal smear observations did not perfectly coincide with these findings. The estrous cycle of mice in the OVX-C group rested on anestrus for a long time that were called estrous cycle inhibition. The results of the 3D scaffold and the hydrogel encapsulating POCs groups were similar (a large number of leukocyte and a fraction of epithelial cells). High level of anestrus was observed in the 3D scaffold encapsulating POCs group,   (Figure 7D-G).

Discussion
This study shows that the 3D POCs-laden structures can support the long-term survival of POCs in vitro and in vivo. The female castrated mouse model demonstrated that the 3D scaffolds encapsulating POCs were beneficial to improve ovarian function. Although estrus was observed in only one mouse by vaginal smears, the 3D scaffold encapsulating POCs construct initiated puberty in the ovariectomized mice.
ECM supports tissue formation specific to the implant site for structural remodeling rather than dysfunctional scar tissue formation [18] . Early studies have proven that tissue-specific ECM bioscaffolds can aid in the recovery of injured tissue function [8] and tissue formation [19,20] , which underscores the advantages of tissue specificity. It is known that once the cellular material remaining in the ECM is transplanted into the recipient, an immune  Figure 8. Detection of the expression of ER-α, PR, inhibin-α and FSHR within the 3D scaffold encapsulating POCs and hydrogel encapsulating POCs constructs after 4-week implantation. DAPI located in the nucleus (blue), the specific primary antibody located in the cell membrane and cytoplasm (green). Green and blue superimposed to appear white, representing the expression of steroid hormone receptors in the groups (white arrows). The expression of ER-α, Inhibin-α and FSHR were more in the 3D scaffold encapsulating POCs group than in the hydrogel encapsulating POCs group. However, there was no significant difference in the expression of PR (P = 0.12). *P < 0.05.

D C B
A E response will be triggered [21] . It has been shown that if the DNA content in the ECM is less than 50 ng/mg [22] it does not trigger an immune response, but promotes tissue remodeling [23] . Our results meet this criterion and indicate the suitability for implantation. Another challenge in decellularization is to retain as much of the ECM components as possible while effectively removing the cells. Collagen fibers and proteoglycans are the major components of the basement membrane [24] , and their presence, further, demonstrates that the decellularization process preserves the major components of the ECM. An intact basement membrane is important for tissue growth and differentiation [25] . Several studies have identified more than 200 unique protein molecules present in the decellularized human vocal fold scaffolds [26] . Our study also confirmed the presence of proteins in the decellularized ovarian tissue, which would guarantee the advantages of bioink 3D scaffolds based on dECMs. Indeed, decellularized ovarian tissue has been shown to preserve 3D follicular structures and support follicle development in vivo [9][10][11][12] .
However, the dECMs often lack tailored microgeometry [27] , resulting in cell distribution confined to the surface of the material, with only a fraction of cells infiltrating the internal regions [28] . Furthermore, once implant the dECMs, the cells infiltrated, or seeded on the dECMs are mainly sustained by the diffusion of oxygen and nutrients before forming a vascular network [29] . The above problems can be resolved by 3D bioprinting technology. The 3D-bioprinted cell-loaded scaffolds have an ideal spatial distribution. Thus, the deposition of bioink can be precisely controlled (down to the micron) to form tissue ultrastructure [30] . The 3D scaffold retains its 3D ultrastructure before degradation. Besides provides the location of the substrate, it also determines the spatial distribution of the loaded cells [31] , and then creates a suitable living environment for the cells. After the 3D-printed scaffold loses its physical support, the cells in the scaffold can build their own ECM locally by secreting matrix proteins, enabling finer tissue remodeling. Furthermore, several studies have shown that 3D scaffolds with 50-200 µm pores could not only promote cell proliferation, migration, and infiltration, but also promote tissue regeneration and repair through loading more cells [16,32] . The higher survival rate of POCs within the 3D scaffolds during first 2 weeks of in vitro culture suggests that the use of porous 3D scaffolds with dECM-based bioink allows sufficient nutrient and oxygen delivery to the cells within the grafts, making long-term in vitro culture become possible.
As we all known, dECM has poor mechanical properties. The previous studies have used polycaprolactone (PCL) to assist adipose tissue in dECM to print 3D scaffolds [33] . However, if the mechanical properties of 3D-bioprinted structures do not match with the natural tissue (PCL is stiffer than adipose tissue), this can lead to complication. Therefore, using a lower stiffness material or an appropriate strategy to match, the mechanical properties seem to be more suitable than using PCL in the 3D-bioprinted structural design. Therefore, we mixed the dECM solution with the seaweed gelatin blend solution to match the mechanical properties of natural tissues. The hardness of the bioink is increased by chemical cross-linking of the seaweed with divalent cations such as Ca 2+ or Sr 2+ . Gelatin has high viscosity and easy freezing properties, and it is homologous to collagen [34] . The dECM-based bioink conformed to the typical collagen triple helix conformation and remained stable at physiological temperature. After printing, the mechanical properties of the 3D scaffolds were enhanced by calcium ion cross-linking and maintained an open porous 3D structure, allowing nutrients transfer for tissue remodeling.
The biocompatibility of the dECM-based bioink must first be evaluated before assessing the role of 3D bioprinting structures in vivo, which is one of the great concerns in regenerative medicine [35] . We observed that the inflammation elicited by the injected bioink gradually decreased with the passage of time. It shows the bioink with an ability that performs an appropriate host response in the specific application. Meanwhile, the live/dead assay also shows that the bioink is non-cytotoxic. From these, we can conclude that the dECM-based bioink has good biocompatibility.
In vivo testing of 3D scaffolds, 3D scaffolds encapsulating POCs, and hydrogels encapsulating POCs in the female castrated mouse model yielded some interesting results. Bioink in a normal subcutaneous mice induced a strong angiogenic response at 1 week after implantation and promoted the migration of inflammatory cells at 2 weeks, possibly due to proteolytic stimulation of cell migration by the bioink or increased availability of major angiogenic growth factors during bioink degradation [36] . Neovascularization of implanted structures is an important indicator for in vivo studies. There is considerable synergistic effect of POCs and decellularized ovarian tissue on angiogenesis of implanted structures. As reported elsewhere, stromal cells in POCs can promote angiogenesis [3,10,37] . However, neovascularization was also observed in the 3D scaffold group. It has been reported that dECM has the potential capacity for angiogenesis [38] . Thus, decellularized ovarian tissues may induce angiogenesis from the peri-host tissue to the grafts. Macroscopically, in the 4-week study, the volume of the hydrogels encapsulating POCs decreased more rapidly than those of the 3D scaffolds and the 3D scaffolds encapsulating POCs in the 4-week study. This suggests that the 3D-printed scaffolds have a relatively slow degradation rate, thus maintaining the volume of the graft structure and allowing tissue regeneration.
Notably, the artificial ovaries are designed to mimic the two representative functions of the native ovaries: Egg production and sex hormones secretion [39] . Therefore, we assessed the survival and proliferation of POCs within the bioink at 4 weeks after implantation. Only a few cells were TUNEL positive, indicating a high survival rate of POCs in the artificial ovaries. A high rate of Ki67-positive cells was also found, indicating that 3D POCs-laden structures can support the proliferation of POCs. Meanwhile, serum levels of sex hormones were significantly increased and germ cell receptors were expressed, with estrus observed by vaginal smear in one ovariectomized mouse treated with 3D POCs-laden scaffold. Moreover, the printed constructs were more strongly expressed (neovascularization, cell proliferation, and germ cell receptors) than the non-printed constructs, confirming the advantages of 3D bioprinting.

Conclusion
This work shows that dECM-based bioink offers a new approach to fabricate bionic 3D structures. 3D POCs-laden structures play an important role in repairing damaged ovarian function, and it is a promising method for fertility preservation. Conceivably, with the development of 3D bioprinting technology and the complementarity between multiple disciplines, 3D-bioprinted cell structures may become an effective treatment for patients with ovarian insufficiency.