Abstract
Some animals have the ability to generate large numbers of oocytes throughout life. This raises the question whether persistent adult germline stem cell populations drive continuous oogenesis and whether they are capable of mounting a regenerative response after injury. Here we demonstrate the presence of adult oogonial stem cells (OSCs) in the adult axolotl salamander ovary and show that ovarian injury induces OSC activation and functional regeneration of the ovaries to reproductive capability. Cells that have morphological similarities to germ cells were identified in the developing and adult ovaries via histological analysis. Genes involved in germ cell maintenance including Vasa, Oct4, Sox2, Nanog, Bmp15, Piwil1, Piwil2, Dazl, and Lhx8 were expressed in the presumptive OSCs. Colocalization of Vasa protein with H3 mitotic marker showed that both oogonial and spermatogonial adult stem cells were mitotically active. Providing evidence of stemness and viability of adult OSCs, enhanced green fluorescent protein (EGFP) adult OSCs grafted into white juvenile host gonads gave rise to EGFP OSCs, and oocytes. Last, the axolotl ovaries completely regenerated after partial ovariectomy injury. During regeneration, OSC activation resulted in rapid differentiation into new oocytes, which was demonstrated by Vasa+/BrdU+ coexpression. Furthermore, follicle cell proliferation promoted follicle maturation during ovarian regeneration. Overall, these results show that adult oogenesis occurs via proliferation of endogenous OSCs in a tetrapod and mediates ovarian regeneration. This study lays the foundations to elucidate mechanisms of ovarian regeneration that will assist regenerative medicine in treating premature ovarian failure and reduced fertility.
Stem cells that orchestrate ovarian regeneration have never been studied in the axolotl. This animal can regenerate many tissues perfectly, thus studying the underlying mechanisms of ovarian regeneration in this organism has great potential for translational outcomes. Many women suffer from premature ovarian failure and reduced fertility. With recent understanding of oogonial stem cells in vertebrates, inducing ovarian regeneration by therapeutic means is a desired application. Deciphering initiation and progression of ovarian regeneration via oogonial stem cells, in an animal that shows conservation of genes utilized in mammalian oogenesis, has great potential impact in the field of regenerative medicine.
Introduction
Ovarian health is critical for vertebrate reproduction, and unlocking regenerative potential in this organ would be transformative in treating premature ovarian failure and reduced fertility. Homeostatic ovarian regeneration, or the generation of new oocytes throughout life, varies considerably across animals [1]. Yet, recent studies provide evidence of mitotically active adult oogonial stem cells (OSCs) in Drosophila [2], teleost [3–5], mouse [6], and humans [7]. OSC transplant studies conducted on zebrafish [3, 8] and mouse [7, 9] also suggest that adult OSCs are the functional cell type of homeostatic oocyte regeneration. Reparative regeneration of functional oocytes has been demonstrated in teleost fish after partial ovariectomy [10] and chemical ablation of oocytes [11], but the evidence for recovery after ovarian injury is limited in other vertebrates. Furthermore, it is also unclear if recovery from ovarian injury occurs via an epimorphic or compensatory regenerative process.
Animals such as urodele amphibians and teleost fish retain the ability to regenerate complex tissues into adulthood, while other animals including mammals do not. Salamanders in particular regenerate a range of tissues including limbs, tails, spinal cord, brain, and heart, but it is unknown if salamanders regenerate sexual organs [12]. Furthermore, the cellular basis for regeneration in urodeles is just emerging in most organs [13] and it is unclear whether regeneration is driven by resident stem cells or through cellular dedifferentiation. Epimorphic regeneration, which entails the generation of a blastema at the injury site and most cellular proliferation is performed locally [14], drives urodele limb, mouse digit tip, teleost fin, and planarian head regeneration. Other tissues regenerate through compensatory regeneration, or the proliferation of cells found throughout the organ such as the mammalian liver [15] and lung regeneration [16, 17]. It is important to elucidate and compare how animals regenerate organs in order to replicate these strategies for therapeutic means.
The lab-raised salamander, the Mexican axolotl (Ambystoma mexicanum), is established as a model for understanding regeneration of many organs, but no evidence of ovarian regeneration nor a detailed characterization of the adult reproductive organs has been performed. The axolotl can lay hundreds of eggs per mating, mate multiple times in a year, and is capable of healthy reproductive function for years (life span ∼10-15 years). It is hypothesized that amphibians with high fecundity including Xenopus [18, 19] and urodeles [20] demonstrate adult oogenesis, but it has not been functionally demonstrated, and the process driving adult oogenesis is unclear. Elucidating the axolotl ovarian anatomy and gene expression profile in oogenesis are the first steps to take in order to test the comparable nature of this model organism to other vertebrates. Furthermore, studying specific cells and mechanisms that are activated in ovarian regeneration will deliver basis for translational outcomes.
In this study, we describe the anatomy of the axolotl ovary and testes and show that the molecular machinery utilized in other vertebrate ovaries are conserved in axolotl ovaries. Furthermore, we identify mitotically active OSCs in the adult axolotl ovary that are capable of oogenesis upon grafting into juvenile hosts. Last, we show regenerative capacity of the axolotl ovaries in vivo and reveal proliferation and differentiation patterns of OSCs and follicle cells (FCs).
Materials and Methods
Animals
Leucistic (d/d) and GFP+ transgenic axolotls were purchased from Ambystoma Genetic Stock Center or from in house breeding at Northeastern University. Tissue collections were performed on animals anesthetized in 0.01% benzocaine. Animal care and use procedures were approved by the Northeastern University IACUC (#15-1138R).
Tissue Processing and Histology
Ovaries, fat-body, oviduct, and testes were fixed overnight at 4°C with Bouin's Fixative (Polysciences, Inc., Warrington, PA), washed with 70% ethanol, embedded in paraffin blocks and sectioned 8 µm. Slides were deparaffinized, rehydrated, and hemotoxylin/eosin (H&E) or Masson's trichrome stained (Thermo Scientific, Cambridge, MA) according to the manufacturer's protocol.
Semiquantitative Polymerase Chain Reaction (PCR) Analysis
RNA was extracted from ovaries, testes, and skin of adult animals using Qiagen RNEasy kits according to manufacturer's protocol. Total RNA obtained was reverse transcribed using iScript cDNA Synthesis Kit (Bio-Rad, Hercules, CA). Gene-specific primers were designed using sequences collected from the Ambystoma Gene Collection [21] as follows: Sox2_ F_ISH CATGACAAAGCACATGCACA, Sox2_R_ISH GCAAATGACAG AGCCGAACT, Vasa_F_ISH ATGGTGATCGGCTACAAAGG, Vasa_R_ ISH AGCTTTGCCAACATTTCCAC, Oct4_F_ISH GAAACATGCGCCAA GATGTA, Oct4_R_ISH GAAGCCAAAGAAAGCAAACG, Nanog_ F_ISH GGCATGTGAACCTCCGTAAG, Nanog_R_ISH TCAAAATGGA CGCTGAAATG, PiwiL1_F_ISH AACTGGCAGGGCATTATCAG, PiwiL1_R_ISH TCAACACCGCCATTTAACAA, PiwiL2_F_ISH GGCCA GGTACCATACGTGTT, PiwiL2_R_ISH AGCCTACATTGGTGGGAATG, 15_F_ISH CCCTCCTTTCCTTTTGCTTT, BMP15_R_ISH GTGGGATGC TCTGATCCACT, Lhx8_F_ISH TCCAACCTTCTAGGCTGTCG, Lhx8_R_ISH CAAAGGCTGGAGTCCAAGTG using Primer322, amplified via Frenche PCR Kit (Intron Biotechnology, Korea), and visualized on 1% agarose gels.
Immunohistochemistry
Tissues were processed as described for histology and antigen retrieval was performed pressure-cooking in 10% citrate buffer for 20 minutes (Cuisinart Electric Pressure Cooker CPC-600), washed with PBS, blocked with goat-serum, and incubated overnight with primary antibodies for Vasa (1:200, ab13840), Abgent custom Vasa antibody (1:200), Oct4 (1:250, ab19857), Histone H3 (1:200, ab14955), 5-bromo-2′-deoxyuridine (BrdU; 1:1,000, ab6326) (Abcam, Cambridge, MA), or GFP (1:500, sc-9996) (Santa Cruz, Dallas, TX). Slides were incubated 30 minutes with Alexa Fluor anti-rabbit 488 or anti-rat 594 secondary antibody at 1:400 and mounted with Slowfade Diamond Antifade Mountant with 4',6-Diamidino-2-Phenylindole (DAPI) (Thermo Fisher Scientific) and visualized under Leica DM2500 fluorescent microscope.
Cloning
PCR products for each gene were inserted into pGEM-t easy vector (Promega, Fitchburg,WI) and grown in DH5-alpha competent E.coli (New England Biolabs, Ipswich, MA). Colonies were sequenced verified and aligned to the mammalian gene sequences. Plasmids were isolated with Zyppy plasmid midi-prep kit (Zymo Research, Irvine, CA).
Riboprobe Synthesis and In Situ Hybridization
Sense and antisense riboprobes were generated using cloned PCR products using the following primers Oct4_F GAGGCTGC AGCTGGAATTAG, Oct4_R GAAGCCAAAGAAAGCAAACG, Sox2_F CCAACTTCACCAACGGACTT, Sox2_R GCAAATGACAGAGCCGAACT, Nanog_F TGCACCTTAGGAGGAGCTTT, Nanog_R TCAAAATGGACG CTGAAATG, Vasa_F TGATGCACCTGGTCAGAGAG, Vasa_R AGCTT TGCCAACATTTCCAC, DazL_F, BMP15, Lhx8_F TCCAACCTTCTA GGCTGTCG, Lhx8_R GCTAGCTCCAAAACCGTTCA, PiwiL1_F TTTCA GTCAGACGCGAAATG, PiwiL1_R TCAACACCGCCATTTAACAA, and PiwiL2_F CGCTTTGGGATGATTAAGGA, PiwiL2_R AGCCTACATTG GTGGGAATG [22]. DIG RNA Labeling mix (Roche Applied Science, Germany), SP6 and T7 RNA polymerases (Promega, Madison, WI) were used to generate probes. Ovaries were fixed for 3 hours in 4% paraformaldehyde (PFA) at 4°C, transferred to 10 and 30% sucrose solution, embedded in optimal cutting temperature (OCT) freezing medium. Tissues were sectioned at 20 µm using a Leica CM3050S cryostat, incubated at 55°C for an hour, washed with Rnase free water, phosphate-buffered saline/tween (PBST), and 2.5 µl Proteinase K, and fixed in 4% PFA to be incubated with hybridization solution for 2 hours at 55°C. Five hundred nanograms of sense or antisense riboprobe for each gene were used in overnight incubation at 55°C. Slides were washed and incubated at 4°C overnight with Anti-Digoxigenin-AP (1:5,000). Washes were performed with 1X Maleic acid buffer, signal was detected using nitro blue tetrazolium (NBT) and 5-bromo-4-chloro-3-indolyl-phosphate (BCIP) (Promega) at 37°C for 5 hours. Reaction was stopped with pH 5.5 PBS, fixed in 4% PFA, mounted with Permount mounting medium (Fisher Scientific, Hampton, NH) and visualized with Leica DM2500.
Hemi-Ovariectomy Surgery and BrdU Injections
A 0.5-cm posterolateral incision was made through the skin and muscle, providing entry to the abdominal cavity, on anesthetized animals. Oviduct was lifted and 2/3s of the right or left ovarian tissue was removed. Incision was sutured using 3-0 nylon suture (Johnson&Johnson, San Lorenzo, PR) and animals were kept out-moist for 3 hours in 18°C. Stitches were removed 5 days post-surgery. BrdU was injected intraperotoneally at 1 mg/g body weight using a 20 mg/ml BrdU stock in PBS (Fig. 6A, 6B).
Cell Grafting and Visualizing EGFP+ Cells
Adult ovarian cells were collected from a transgenic CAG:EGFP transgenic axolotl, dissociated into single cells in Hank's balanced salt solution (HBSS) (Thermo Fisher Scientific Inc.) and size sorted with Falcon 40 µm cell strainer. Cells were injected into the abdominal cavity of 39 white juvenile animals. After variable growth periods, the gonads were collected, fixed in 4% PFA for 3 hours 4°C at and mounted with OCT tissue freezing medium (Leica Biosystems). Blocks were cryosectioned to 15 µm. Slides were washed with di-water and mounted with DAPI or were put through immunohistochemistry (IHC) analysis with GFP and Vasa costaining with DAPI.
Statistical Analysis
Statistical analysis was performed using student's t test in the comparison of normal and regenerating tissues. Three biological replicates were analyzed per time point. For the uninjured ovary proliferation baseline for 24-hour normal proliferation was assessed based on five biological replicates.
Results
Oogonia Are Present in the Adult Ovary
Characterization by size and nuclear organization revealed that axolotl ovaries contain oocytes ranging from stage 0 (oogonia to diplotene) to stage VI (mature oocytes) (Fig. 1A). Stage 0 oocytes also showed a variety of developmental stages comparable to Xenopus ovaries [23]. Cells resembling oogonia, which had a small cytoplasm and ring of chromatin surrounding the nucleolus and threadlike nuclear chromatin (SN configuration) [24, 25], were observed as the earliest germ cell within stage 0 oocytes. Leptotene, synaptene, contraction, and pachytene phase germ cells with condensed chromatin and diplotene oocyte with diffused chromatin were seen as the oocyte matured into Stage I (Figs. 1A, 2B-a). Throughout the maturation of the stage III oocytes, the large cytoplasm was slowly replaced with yolk platelets (Fig. 1A). Given the overwhelming amount of yolk in 2,000-3,000 µm stage VI oocytes, they were not captured intact in an image. FCs were observed to encapsulate the oocytes from a minimum of 2 cells (diplotene) to over 500 cells (Stage V).
Oogonia are present in juvenile through adult ovaries and are abundant on the peripheral borders of the ovaries. (A): Anatomy of Oocytes: Stages of development. H&E staining analysis of six stages (0 to V) of oocytes out of seven. (B): Masson's trichrome staining analysis of four different sizes/sexual maturity groups of ovary. Arrows; oogonia. Scale bar 250 µm. (C): Digital sketch of the complete female reproductive system: the oviduct (o.d.), fatbody-cut (f.b.) and the ovaries (ov.). H&E analysis of 1, 2, 3, and 4 designated areas of caudal to rostral dissection (D) fat body and (E) oviduct. Scale bar 50 µm.
Mammalian germline stem cell markers are expressed in axolotl ovary and testes. (A): PCR analysis of Vasa, Oct4, Sox2, Nanog, Bmp15, PiwiL1, PiwiL2, DazL, and Lhx8 with EF1-alpha housekeeping gene. Testes (PC) and Skin (NC). Alignment to human genes represented by % amino acid similarity/match. (B): Histological analysis of (a) OG in the ovary and (b) SG in testes. (C-a): H&E analysis of testes; spermatogonia (Sg.), primary spermatocytes (Sc′), secondary spermatocytes (Sc′′), early (primary) spermatids (St′), and late (secondary) spermatids (St′′) (b) Immunohistochemistry analysis H3 (red), and Vasa (green) protein coexpression in adult testes, spermatogonia zoomed in. Scale bar 50 µm. Abbreviations: OG, oogonia; SG, spermatogonia; Sg, spermatogonia; Sc′, primary spermatocytes; Sc′′, secondary spermatocytes; St′, early spermatids; St′′, late spermatids.
Upon confirming the comparable morphology of oogonia, meiotic germ cells, and later stage oocytes in the adult ovary, we next determined where oogonia were present by ontogeny. Ovaries were examined by Masson's trichrome staining from different sexual maturity groups, ranging from 4 months (sexually immature juvenile) to 1.5 years old (sexually mature) (Fig. 1B). In the juvenile ovary, along with early germ cells, only stage 0 and I oocytes were observed. As the ovaries matured, stage II and III oocytes were seen in addition to the earlier stages. In the adult ovaries, all stages of oocytes were present from oogonia to stage VI oocytes. Next, we determined if oogonia were present in specific anatomical locations/stem cell niche within the ovary.
Cross-sections were examined along the rostro/caudal axis of the ovary to determine a specific niche of oogonia. Ovaries showed no distinct anatomical location of a specific type of oocytes (Fig. 1C). A variety of oocytes from different developmental stages were seen throughout the ovarian tissue. However, similar to the teleost Medaka, most germ cells resided in the ovarian cords in the peripheral borders, and clusters of germ cells resembled germinal cradles described [4]. Furthermore, some germ cells resided further away from the peripheral cords beside the FCs which, hereafter, we refer to as the follicular zone. Oocytes that had matured passed stage I had moved toward the center of the ovary, making it possible for the germ cells and diplotene oocytes to remain abundantly in the peripheral borders. Interestingly, fat body also was found to be in direct contact with the outside surface of the ovarian tissue (Fig. 1C, 1D).
Pluripotency and Germline-Specific Markers Are Expressed in OSCs
In order to investigate the molecular mechanisms of OSCs, we characterized transcription of a panel of genes expressed in the germline during vertebrate development [26–28]. Some of these genes, which have a role in germline production in mammals [29], were also previously shown in the axolotl [30]. Semiquantitative PCR showed that Vasa, Sox2, DazL, PiwiL1 and PiwiL2 genes were expressed in both ovary and testes, while Oct4, Bmp15, Lhx8 and Nanog were ovarian specific (Fig. 2A). Amino acid sequences of OSC genes were also highly conserved with mammalian homologs (Fig. 2A).
In addition, immunohistochemistry (ISH) visualization determined the localization of mRNAs coding for pluripotency and germline markers. Annealing of anti-sense probes detected Bmp15, DazL, PiwiL1, PiwiL2, Nanog and Vasa mRNA in the cytoplasm of germ cells located in adult ovaries, whereas sense probes gave no signal (Fig. 3A). As expected, not all stage 0 oocytes showed staining when challenged with the antisense probes for these six genes, suggesting limitation in stemness as they may be progressing towards meiotic differentiation. ISH results were supported by IHC analysis upon detection of germline-specific marker Vasa and pluripotency marker Oct4 in the cytoplasm of a portion of stage 0 oocytes (Fig. 3B). All later stages of oocytes including diplotene stage did not show any protein expression of Vasa or Oct4. In addition, following digest of the ovary, smear analysis also confirmed that Vasa expression is localized to the cytoplasm of early germ cells (Fig. 3C).
Stage 0 oocytes show germline and pluripotency marker localization. (A): In situ hybridization analysis of BMP15, DazL, PiwiL1, Nanog, PiwiL2, and Vasa mRNA localization in adult ovaries. Asterisk: control stage 0 oocytes with no staining. Arrows: stage 0 oocytes with staining. (B): Immunohistochemistry analysis of cytoplasmic Vasa (green or red) and Oct4 (green) protein expression in adult ovaries. (C): Juvenile ovary smear analyzed for Vasa (red) expression. Scale bar 50 µm.
Furthermore, in order to understand the cellular origins of gametogenesis, we performed histological and immunohistochemical comparisons using spermatogonial stem cells (SSCs) as a positive control. Spermatogonia in the axolotl followed the stereotypical differentiation pattern seen in mammals [31]: first by differentiating into primary and secondary spermatocytes, then to early and late spermatids, and finally to sperm (Fig. 2B-a, 2C). Histological comparisons between oogonia and spermatagonia revealed striking morphological resemblances such as diffused chromatin and small cytoplasm (Fig. 2B). Upon IHC analysis of the testes, both the H3 mitotic marker (nuclear- phosphorylated S10 on histone 3), and Vasa were coexpressed in SSCs likely responsible for continuous sperm formation (Fig. 2C-b). Additionally, synchronized meiotic activity was observed via H3 expression in pockets of Vasa− spermatocytes further differentiating into spermatids. Overall, the morphology and gene expression patterns we detected in the OSCs of the axolotl show a high level of conservation with developing germ cells and adult OSCs from other vertebrates.
A Portion of Germ Cells Are Mitotically Active and Express Vasa
Locating germline and pluripotency markers in stage 0 oocytes focused our search for OSCs to stages earlier than diplotene. Thus, mitotic activity in the adult ovary was tested as a next logical step. As seen above in testes, SSCs are the only cell type to exhibit both germline marker Vasa and mitotic activity marker H3 (Fig. 2C-b). Therefore, 15 different sections from three adult females were investigated through IHC analysis for costaining with Vasa and H3. A total of 1,359 germ cells were counted in the peripheral borders, and a total of 443 were counted in the follicular zones (Fig. 4A). In order to clarify the distinction between germ cells and FCs, which are also capable of mitotic division, morphological differences were considered and FCs were disregarded in the counts (Fig. 4A). H3 expression upon mitotic activity in the FCs also was utilized as a positive control. IHC analysis was quantified to show that 2.46 ± 0.33% of germ cells expressed only the Vasa marker and 0.11 ± 0.01% of germ cells in the periphery were positive for both Vasa expression in the cytoplasm and H3 expression in the nucleus simultaneously (Fig. 4B). In the follicular zone, 2.04 ± 0.31% of germ cells expressed cytoplasmic Vasa marker, while only 0.08 ± 0.01% of these were both Vasa and H3 positive (Fig. 4B). Overall, 4.32 ± 0.27% of the Vasa expressing germ cells were observed to be mitotically active OSCs in the peripheral borders of the adult axolotl ovary. Furthermore, 3.84 ± 0.98% of Vasa expressing germ cells were mitotically active OSCs in the follicular zone (Fig. 5C).
Mitotically active, Vasa expressing oogonia are present in the adult ovary. (A): Immunohistochemistry analysis of Vasa (Green) and H3 (Red) protein coexpression in oogonia of the adult ovary located in the peripheral borders (in clusters or not) and at the follicular zone. Dividing follicle cells H3 staining (PC). (B) Graphical representation of germ cells (100%), % of cells expressing Vasa and % of cells that are dividing and Vasa expressing OSCs. Scale bar 50 µm. Abbreviation: OSCs, oogonial stem cells.
GFP Adult OSCs inhabit the white host ovaries and give rise to GFP oocytes. (A): Immunohistochemistry (IHC) analysis of a developing ovaries via H3 (green) and VASA (red) coexpression. (B): H&E and IHC analysis of 4 month old ovary. G: Gonads, N: Notochord, K: Kidney, B: Blood Cells. (C): Grafting schematics. 4 months post grafting (mpg) white hosts with native GFP (green) (D) oocyte in the ovary (E) testicular cell in testes. IHC analysis of GFP and Vasa coexpressing OSCs and only GFP expressing new oocytes in white host ovary (F) 2-3 mpg and (G) 7 months mpg. Arrows: OSCs, Asterisks: progeny GFP oocyte, B.C: Blood Cells. Scale bar 50 µm. Abbreviations: BCIP, 5-bromo-4-chloro-3-indolyl-phosphate; DAPI, 4',6-diamidino-2-phenylindole; dpf, days post fertilization; EGFP, enhanced green fluorescent protein; GFP, Green Fluorescence Protein; HBSS, Hank's balanced salt solution; NBT, nitro blue tetrazolium; OCT, optimal cutting temperature; OSCs, oogonial stem cells; PBST, phosphate-buffered saline/tween; PCR, polymerase chain reaction; PFA, paraformaldehyde; SSC, spermatogonial stem cells.
GFP Oocyte Stem Cells Give Rise to GFP Oocytes in White Host Ovary
In order to study the formation of OSCs, we characterized primordial germ cell (PGC) differentiation into OSCs during development. Many H3+/Vasa+ OSCs were detected in the 30 days post fertilization (dpf) (data not shown), 60 dpf, 90 dpf, and 120 dpf animals (Fig. 5A, 5B). Histological analysis of 120 dpf animals were also completed to show the location of developing ovaries in the juvenile animal with respect to the notochord, kidneys, and fat-body (Fig. 5B).
We next sought functional confirmation for the identification of OSCs in the adult ovary. Grafting experiments were conducted to test if EGFP adult OSCs could inhabit white host ovaries and give rise to ovarian or testes cells. EGFP ovaries were selected through a 40-µm cell strainer to eliminate diplotene and mature stage oocytes. Size-sorted EGFP cells were grafted into white sexually immature host animals 90 to 120 dpf, thus with no preidentification of male or female sex (Fig. 5C). Five animals were sacrificed 60 days post injection (dpi) and four at 90 dpi. Limited engraftment was observed, revealing 2 EGFP+ cells at 60 dpi: one implanted in the testes (Fig. 5D) and the other in the ovary (Fig. 5E). To address whether younger hosts accept grafting more efficiently, grafts were repeated into 10 animals of 30 dpf (Fig. 5C). Four ovaries showed 12 EGFP+/Vasa+ coexpressing OSCs as well as one EGFP expressing ovarian cell (Fig. 5F). Two more animals were sacrificed 7 months post grafting, which contained 27 EGFP+/Vasa+ and 8 EGFP+ oocytes in one animal and 26 EGFP+/Vasa+ and 10 EGFP+ oocytes in the other animal (Fig. 5G). Altogether, grafting occurred at a low frequency, although it was clear that grafted adult OSCs were capable of differentiating into both oocytes and testicular tissue. Yet, in our study it was not confirmed that female OSCs were capable of transdifferentiating into spermatocytes.
Oogonial Stem Cells are Activated in Ovarian Regeneration Giving Rise to Many New Oocytes
In order to trace when and where cells participate in a regenerative response after hemi-ovariectomy, proliferating cells were tracked by BrdU incorporation analysis on the 1st, 7th, 14th and 21st days post-surgery (dps) (Fig. 6A). 2171 germ cells in 26 animals were analyzed. Proliferation was observed to be initiated by 24 hours after injury (Fig. 6E) and was not confined to a specific region directly adjacent to the injury. Over the first 3 weeks after surgery, at each 24 hour period investigated, a significantly higher number of OSCs were proliferating compared to uninjured ovaries (Fig. 6E). Also, the number of stage 0 oocytes generated increased after injury at all time points (Fig. 6F). OSCs were identified via BrdU+/Vasa+ whereas newly generated oocytes were detected by BrdU+/Vasa- (Fig. 6C).
Oogonial stem cells (OSCs) are activated during ovarian regeneration and regeneration triggers follicle cell (FC) proliferation. Schematics of experimental design to analyze proliferation of cells in surgically induced regeneration of the ovary (A) via BrdU pulse chase tracking for 35 days (B) via BrdU assessment for each 24 hours of exact time points. (C): Representative IHC images. Arrows: Vasa (Red) and BrdU (Green) coexpressing OSCs quantified in regenerating and uninjured ovary, Arrowheads: only BrdU expressing new oocytes. (D): Arrows: representative images of BrdU expressing FCs of stage 1&2, 3, and 4 oocytes quantified. Scale bar 50 µm. Graphical comparison of regenerating and noninjured normal ovary for (E) OSC proliferation (F) New oocytes generated (G) total FC proliferation (H) FC proliferation based on stages of oocyte development on days 1, 7, 14, and 24 for 24 hours. Graphical comparison of regenerating and noninjured ovary for (I) OSC proliferation (J) new oocytes generated (K) total FC proliferation (L) FC proliferation based on stages of oocyte development accumulating on days 1, 7, 14, 21, 28, and 35. Scale bar 50 µm. *, p < .05; **, p < .005; ***, p < .001; ****, p < .0001. Abbreviation: BrdU, 5-bromo-2′-deoxyuridine.
Furthermore, a BrdU pulse/chase experiment was performed for 35 days to analyze the accumulation of OSCs, OSC progeny, and the proliferating FCs in uninjured ovaries and during regeneration (Fig. 6B). Overall, 2907 germ cells in 36 animals were evaluated. Control ovaries showed significant increase in BrdU+ resident OSCs 28 and 35 days after BrdU injection (Fig. 6I, 6C). On the contrary, in the regenerating ovaries, the OSC activity was increased on day 1 and was followed by an immediate reduction in the BrdU signal (Fig. 6I). However, the number of new oocytes accumulated was higher in the regenerating ovary in comparison to the uninjured ovaries over 35 days (days 1, 14, and 28 being the most significant) (Fig. 6J). Three animals that underwent the surgery and were not sacrificed were able to resume reproductive ability after a healing period of 5 months. Interestingly, throughout the regeneration experiments, a localized response near the injury site was not observed, thus suggesting a compensatory mechanism of organ regeneration rather than an epimorphic regenerative process.
An important aspect of oocyte growth and differentiation is the production of FCs. Therefore, proliferating FCs were evaluated and quantified in regenerating ovaries based on maturity stages of oocytes; diplotene, I & II, III, and ≥ IV (Fig. 6D). Of the 27,025 cells counted, BrdU+ cells were significantly higher on days 1, 7, and 21 dps when compared to uninjured/normal ovaries (Fig. 6G). When each oocyte stage was taken into consideration, the FCs of diplotene and stage III oocytes were the most responsive to regenerative signals (Fig. 6G). A significant increase in the FCs of the stage III oocytes and a significant decrease in the FCs of diplotene oocytes were observed on day 1, 7, and 14 (Fig. 6H). Of the 56,904 FCs counted in the pulse/chase experiment, total BrdU accumulation was overall higher in the regenerating ovaries yet only met significance at days 1 and 35 dps, likely due to high variation across samples (Fig. 6K). FCs surrounding diplotene, stage I, and stage II oocytes overall showed less proliferation compared to normal ovaries (Fig. 6L). On the contrary, stages greater than III showed more proliferation upon regeneration signals (Fig. 6L).
Discussion
The increasing availability of molecular tools applicable to the axolotl is allowing us to elucidate the mechanisms of regeneration for many organs, thus ensuring the potential of tremendous impact on translational regeneration studies. Given that females can produce thousands of oocytes every year, the axolotl provides a resourceful model to study ovarian regeneration as well as adult oocyte production. As we were able to show that the vertebrate molecular mechanisms of oogenesis are conserved in the axolotl, we established it as a clinically relevant model for ovarian regeneration.
Our results showed that, in contrast to the SSCs in testes, OSCs were not confined to specific anatomical regions of the ovary. Instead, they were abundant in the peripheral lining of the ovaries (some in clusters) directly in contact with the fat-body. The proximal positioning of the fat-body to OSCs may be a result of the secretory functions of this endocrine organ in order to deliver developmental signals or hormones [32]. It will be important to test whether signals arising from the fat-body support or even initiate ovarian regeneration.
We found that OSCs express many of the genes that regulate vertebrate oogenesis. Germline expression of the RNA helicase Vasa (Ddx4) is highly conserved across vertebrates [33, 34], which was supported by our results. We found that transcription factors Oct4, Sox2, and Nanog, which are involved in the maintenance of pluripotent stem cells [35, 36], were also expressed, suggesting that there is maintenance of germ cell competency in the adult ovary. This expression in the OSCs also coincides with Sox2 expression seen in SSCs [37]. The ovarian-specific expression of Bmp15 supports the presumptive role of BMP15 in oocyte and FC maturation [38], suggesting that an ongoing folliculogenesis is occurring in adults. Furthermore, PiwiL1 and PiwiL2, which maintain the DNA integrity in gametogenesis by suppressing transposable elements [28], were both expressed in ovaries and testes. We also observed expression of the RNA-binding molecule Dazl in adult ovary and testes, which is a highly conserved germ cell marker [30, 39] involved in spermatogenesis and oogenesis possibly by regulating Vasa. Last, we found that Lhx8 was highly expressed only in the ovaries, which supports our current understanding that it regulates primordial follicles and folliculogenesis [40]. The location and the gene expression profile of OSCs presented here may further be utilized to isolate and culture them for future experimentation such as nuclear manipulation of germline cells in order to develop new transgenic tools in the axolotl. In addition, vertebrate germline markers expressed in testes were as expected due to common PGC lineage with ovarian tissue. Thus, the gene expression data can further be interpreted to suggest that vertebrate (including mammalian) mechanisms of spermatogenesis are also conserved in the adult axolotl.
Furthermore, we observed that oogonial and SSCs possessed morphological similarities with one another, showed mitotic activity, and expressed pluripotency and germline markers. Vasa+/H3+ cells clearly identified SSCs in testes, which had similar morphology to other urodeles [41]. We also observed spermatocyte pockets undergoing synchronized meiosis. When we repeated the same experiment with the ovarian cells and identified the OSCs based on these criteria, we quantified their abundance during homeostasis in a normal adult ovary. Only stage 0 oocytes that were smaller than the diplotene (germ cells) were counted, since morphologically diplotene is easily differentiated from the remaining stage 0 oocytes, and is too late in development to be suspected as a potential OSC. Within germ cells, we found oocytes that were simultaneously expressing Vasa and actively dividing at the moment of analysis in the peripheral borders. Similar results were obtained in the follicular zone, even though three times as many stage 0 oocytes were found at the peripheral borders. It is important to note that not all OSCs will be actively dividing at a given moment, as the oogenesis demand will vary. Thus, it can be suggested that the adult ovaries may have a larger OSC reserve, yet at any given moment only 0.11% can be definitely identified as actively dividing OSCs. We also wanted to establish the role of OSCs in development, thus we investigated the mitotic activity along with Vasa expression in juvenile animals. Our findings also confirmed the presence of OSCs giving rise to many new oocytes in rapidly developing ovaries, further supporting our method of OSC detection. Overall, our findings prove that, even though there is a small percentage, our data coincides with the occurrence of OSCs and SSCs found in other organisms [3, 11, 37]. This indicates that adult OSCs maintain mitotic activity to continuously be utilized in oogenesis, comparable to the SSC activity in the testes giving rise to new sperm throughout the life span.
Validation of OSC function was demonstrated by grafting EGFP+/Vasa+ OSCs into white hosts. Younger host animals yielded more EGFP+/Vasa+ OSCs inhabiting the ovaries, possibly due to ongoing developmental signals that can guide germ cells to the developing gonads. Moreover, we also detected EGFP+/Vasa+ cells in testes, suggesting transdifferentiation of OSCs into spermatogonia. This suggests that OSCs may have transdifferentiation potential when in the correct environment, which was also seen in studies in other species [3, 42] including other urodeles [43]. Yet, without further experimentation we cannot conclude that this is the case in our findings. In addition, considering EGFP is silenced after the early stages of oocyte development, the EGFP oocytes we were able to detect after long-term engraftment likely underestimates the total amount of donor-derived oocytes.
Upon identification of OSCs in the axolotl ovary, we next showed that the axolotl ovary can regenerate, and this process is driven by OSC activity. Our novel hemi-ovariectomy surgery allowed us to injure ovarian tissue and track axolotl ovarian regeneration in vivo. Surprisingly, within 24 hours after surgery, a regenerative response including proliferation and differentiation was observed to be widespread across the ovary. This is in contrast to other modes of epimorphic regeneration such as limb regeneration (in similarly sized animals), where proliferation is limited in the first week after injury and is spatially restricted to the injury site [44]. Overall, we attribute this quick compensatory response upon injury to the readily available pool of resident OSCs that normally participate in homeostatic state.
Indeed, the regenerative response that mounted in the first 24 hours was substantial. Yet, BrdU can only be incorporated into OSCs that are in the process of DNA replication during asymmetric or symmetric cell division, thus this can only be accomplished only by premeiotic or premitotic OSCs. Another division to consider in the gonads is the second division of meiosis, seen by the pockets of dividing spermatocytes in testes with the H3 marker. Yet, BrdU cannot be incorporated by germ cells going through meiosis II, since they do not go through an S-phase. We also considered DNA repair as a contributor to such high amounts of BrdU signal. BrdU can be picked up during DNA repair, yet in comparison to the DNA replication it is in trace amounts [45]. Thus, we are confident that the BrdU+ cells truly reflect the progeny and self-renewal in ovarian regeneration.
Our results show that ovarian regeneration occurs through OSC and FC proliferation and OSC differentiation. Upon injury, we observed a significant increase in the rate of both OSC proliferation and differentiation on days 1, 7, and 14. However, on day 21, OSC activity began to slow, yet still remained elevated compared to the uninjured ovaries. FC proliferation was also significantly high at days 1, 7, and 21. Further confirming these findings, through cumulative BrdU pulse/chase, we detected increase in the total amount of cells that were newly generated from resident OSCs and FCs over 35 days in the injured ovary. The BrdU signal in the OSCs slightly increased within the 7 days of injury then diminished quickly due to a high rate of proliferation, as supported by the previous experiment. Concurrently, new oocytes/progeny continued accumulating throughout regeneration.
In the control ovaries, the number of OSCs remained at a baseline until day 21, since there were no injury-induced signals. We were able to estimate the normal OSC self-renewal cycle to be between every 21st and 28th days since the uninjured ovaries showed their first significant OSC accumulation signal at day 28. Furthermore, accumulation of new OSCs was seen by day 35 in the normal ovary as expected.
We also demonstrated that upon regeneration signals, ovarian maturation, and thus the expansion of the follicles are accelerated. This is due to the induction of FC proliferation in the injured ovary. It will be important in the future to determine if new FCs arise from a specific somatic stem cell population, such as the Lgr5+ somatic stem cells that drive epithelial regeneration in the adult mammalian ovary [46]. Interestingly, FCs of mature oocytes (stages III to V) showed the most proliferation. The large surface area of these oocytes may have resulted in larger amounts of regenerative signals received, or their central position may have also been a factor. These results also support the conclusion that oocytes in the stage of rapid yolk accumulation require more FC generation. On the contrary, the FCs of smaller oocytes (diplotene to stage II) showed decreased proliferation in the regenerating ovary. We attribute this result to disruption of the normal order of maturation by injury, and to the increasing number of OSCs and new oocytes in the regenerating ovary. We suggest that loss of ovarian tissue triggers accelerated maturation of later stage oocytes while facilitating the creation of new oocytes in order to sustain the homeostasis of the tissue composition in the axolotl. Further transcriptomic analysis on these time points of ovarian regeneration may reveal specific genes upregulated in order to orchestrate OSC and FC proliferation.
To our knowledge, we have provided the first evidence of ovarian regeneration in an amphibian and definitive evidence of in vivo ovarian regeneration in tetrapods. The clinical importance of these findings is promising in resolving reproductive dysfunction in mammals. Since we provide the evidence of OSC activation upon regeneration signals in the axolotl, our data suggest that once regenerative factors are elucidated, ovarian tissue can be triggered to produce more oocytes. The induction of new oocytes and ultimate ovarian regeneration would be invaluable to those patients who suffer from infertility or decreased ovarian function due to a secondary complication of a disease, cancer treatments, or surgical removal of partial ovarian tissue [47]. Once the OSCs are activated to repopulate the ovaries with oocytes, full or partial recovery of fertility may be possible. In addition, stimulation of OSCs to produce more oocytes may provide insight to cure early menopause, which results in osteopenia, osteoporosis, decreased metabolic rate, among other maladies. Regenerative signals do not only affect the OSCs, but they also trigger FCs to proliferate. FC proliferation results in the expansion of the follicle that encapsulates the oocyte, and when stimulated with such signals, it may be utilized in oocyte maturation or estrogen production. Further studies to determine which regeneration signals/factors molecularly enable stimulation of these stem cells and FCs are needed in order to suggest possible treatments to reproductive disease and contribute to regenerative medicine.
Conclusion
Overall, our results show that the oogonia are present in juvenile to adult axolotl ovaries and are abundant in the peripheral borders. Nuclear and cytoplasmic morphology of the oogonia is similar to spermatogonia which is capable of giving rise to sperm throughout life. The axolotl is a translational model since it utilizes the same genes involved in mammalian oogenesis. Futhermore, OSCs identified in the axolotl express germline & pluripotency markers and are capable of self-renewal through mitotic division, similar to spermatogonia. These OSCs can also differentiate into oocytes, and they are viable when grafted to resume both symmetric and asymmetric division. There is baseline OSC activity in ovaries in homeostasis and OSC self-renewal cycle occurs between 21 to 28 days. When the ovaries are injured through surgery, a compensatory regeneration response is generated. OSCs increase both their proliferation and differentiation in order to compensate for the loss of tissue. Follicular cells also increase their proliferation resulting in follicular maturation to speed up. With this study, we present the first evidence of ovarian regeneration in an amphibian and definitive evidence of in vivo ovarian regeneration in tetrapods.
Acknowledgments
This Research was supported by Northeastern University to J.R.M. and P.E. We thank Jordan Bernard and Kelsha Sanchez for help with animal care and imaging. We also thank Dr. Jonathan Tilly and Dr. Dori Woods for their support. FJD III. A.S. is currently affiliated with School of Life Sciences, University of Nottingham, Nottingham, U.K.
Author Contributions
P.E.: Conception and design, collection and/or assembly of data, data analysis and interpretation, manuscript writing, final approval of manuscript; A.S.: Conception and design, collection and/or assembly of data, data analysis and interpretation, final approval of manuscript; J.R.M.: Financial support, conception and design, collection and/or assembly of data, data analysis and interpretation, manuscript writing, final approval of manuscript.
Conflicts of Interest
The authors indicate no potential conflicts of interest. .
