Here, we present an optimized protocol for imaging entire ovaries for quantitative and qualitative analyses using whole-mount immunostaining, multiphoton microscopy, and 3D visualization and analysis. This protocol accommodates high-throughput, reliable, and repeatable processing that is applicable for toxicology, clinical diagnostics, and genomic assays of ovarian function.
Female fertility and reproductive lifespan depend on the quality and quantity of the ovarian oocyte reserve. An estimated 80% of female germ cells entering meiotic prophase I are eliminated during Fetal Oocyte Attrition (FOA) and the first week of postnatal life. Three major mechanisms regulate the number of oocytes that survive during development and establish the ovarian reserve in females entering puberty. In the first wave of oocyte loss, 30-50% of the oocytes are eliminated during early FOA, a phenomenon that is attributed to high Long interspersed nuclear element-1 (LINE-1) expression. The second wave of oocyte loss is the elimination of oocytes with meiotic defects by a meiotic quality checkpoint. The third wave of oocyte loss occurs perinatally during primordial follicle formation when some oocytes fail to form follicles. It remains unclear what regulates each of these three waves of oocyte loss and how they shape the ovarian reserve in either mice or humans.
Immunofluorescence and 3D visualization have opened a new avenue to image and analyze oocyte development in the context of the whole ovary rather than in less informative 2D sections. This article provides a comprehensive protocol for whole ovary immunostaining and optical clearing, yielding preparations for imaging using multiphoton microscopy and 3D modeling using commercially available software. It shows how this method can be used to show the dynamics of oocyte attrition during ovarian development in C57BL/6J mice and quantify oocyte loss during the three waves of oocyte elimination. This protocol can be applied to prenatal and early postnatal ovaries for oocyte visualization and quantification, as well as other quantitative approaches. Importantly, the protocol was strategically developed to accommodate high-throughput, reliable, and repeatable processing that can meet the needs in toxicology, clinical diagnostics, and genomic assays of ovarian function.
Most mammalian females are born with a finite number of meiotically arrested oocytes stored within primordial follicles, constituting the ovarian reserve (OR)1,2. The OR determines the overall female reproductive lifespan and health3. The OR normally declines in size with aging and can be prematurely depleted upon exposure to certain genotoxic agents (radiation/chemotherapy) and environmental stresses (malnutrition), leading to infertility4,5,6. Idiopathic female infertility can often be attributed to the genetic and physiological quality of eggs developing from the OR and remains poorly understood7,8. Because female follicle endowment is largely predetermined by birth, it is essential to understand the regulatory mechanisms involved in the OR establishment and maintenance.
In mice, OR formation starts with the specification of primordial germ cells (PGCs) around Embryonic day (E) 7.52. The PGCs migrate to the genital ridges, where they will reside by approximately E10.59. The following extensive proliferation occurs with incomplete cytokinesis resulting in the formation of cysts that will be broken down later in development10,11. At approximately E12.5, gonadal sex is determined, and PGC proliferation halts in ovaries. In females, PGCs, now oocytes, enter meiotic prophase I (MPI) at approximately E13.512,13. Oocytes progress through extended MPI and arrest at the dictyate stage around the time of birth. During the first week after birth, each arrested oocyte is surrounded by granulosa cells, thereby forming a primordial follicle.
The number of primordial follicles in the OR of a female depends on how many oocytes survived the waves of oocyte elimination that occur before and during MPI arrest through apoptosis, autophagy, or necrosis14,15. The first wave occurs during fetal development and is known as FOA. FOA is an evolutionarily conserved process in females (mammalian and non-mammalian), whereby an estimated 50-80% of the oocytes are eliminated depending on the female species16,17,18,19. In mice, FOA occurs during E15.5 to E18.5 and has been attributed to the reactivation and expression of retrotransposon LINE-1 sequences causing oocyte death20,21. The second wave of oocyte elimination occurs through a meiotic checkpoint that eliminates oocytes with meiotic defects such as unrepaired DNA double-strand breaks (DSBs)22,23. The next wave of oocyte elimination occurs during cyst breakdown, culminating during the formation of primordial follicles, each of which contains a single oocyte10,11,24,25.
In mice, the primordial follicle reserve is largely established by puberty, after which it decreases as primordial follicles are activated for growth during regular reproductive cycles. The OR size varies among individual women and among different genetic strains of mice; yet, the genetic regulation of OR size is not well understood26,27,28,29. Genetic studies of OR regulation are hampered by the lack of standardized protocols to study the waves of oocyte elimination during prenatal and postnatal development. Several oocyte quantification methodologies have been developed in mice, with the most common and widely used being histomorphometric evaluation of histological sections30,31. In this technique, oocytes are identified on serial sections with histological stains, such as hematoxylin and eosin (H&E) and periodic acid-Schiff (PAS) or fluorescent markers. This technique is reliable if all conditions remain constant, including section thickness, efficient recovery of all sections throughout the ovary, and the counting schemes of individual laboratories. However, numbers reported by different laboratories often differ significantly and thus are not easily comparable.
Moreover, given genetic differences, the use of different mouse strains can also influence oocyte counts. Additional computational approaches have been developed for histomorphometric evaluation and include the automated detection of oocytes using the fractionator approach, automatic counting using computational algorithms, and 3D reconstruction of histological images to prevent multiple counts of the same oocyte31,32,33,34,35,36. Even with these improvements added to histomorphometric evaluation, the technique is relatively labor-intensive, particularly for large-scale and high-throughput studies. The data collected may not be reproducible and comparable between studies due to differences in counting schemes, computer algorithms, and software used.
Recently, accelerated by the development of new medium-resolution multiphoton and light sheet microscopy and optical tissue clearing methods, 3D modeling and analysis techniques for intact ovaries are becoming the method of choice to efficiently quantify oocyte numbers and study protein localization and dynamics37,38. These 3D methods are typically advantageous compared to histological methods as tissues and organs are better preserved and kept intact. Moreover, 3D analysis and modeling provide additional insights into function and interactions within and between cell niches or substructures within the organ that may be missed in 2D analysis.
3D analysis of whole organs requires optimization of fixation, immunostaining, and optical clearing protocols for individual organs, such as ovaries, without tissue distortion or damage. Additional optimization of sample mounting for imaging is required for high-resolution microscopy and may depend on the imaging platform available. Finally, imaging of the whole intact ovary generates a large amount of data for subsequent computational analyses. Therefore, there is a need to develop standardized 3D methods for counting oocytes for comparative studies and across developmental stages.
This protocol uses standard immunostaining and previously reported clearing protocols, focusing on a simple, user-friendly, and high-throughput approach38,39,40,41. The protocol is optimized to analyze large numbers of prenatal and postnatal ovaries up to postnatal day 28 (P28) and varying sizes of ovaries from different mouse genetic backgrounds. The immunostaining steps are similar for all stages; however, the clearing protocols differ for pubertal ovaries due to their larger size, ScaleS4(0) and CUBIC for small and large ovaries, respectively40,41. Further, whole-body perfusion is performed in P28 mice before fixation to prevent autofluorescence from blood cells. A multiphoton microscope was built on the Leica DIVE/4Tune platform as an alternative to light sheet microscopy to acquire images, and IMARIS 3D Visualization and Analysis software with various analytical tools was chosen for this protocol. This protocol is simple to follow and less hands-on, hence time-saving. Moreover, oocyte quantification is relatively quick, depending on the size of the ovary and arrangement of oocytes.
All mice used were of the genetic strain C57BL/6J (see the Table of Materials). This strain has been fully sequenced and is standard for many studies on ovarian structure and function. Mice were housed according to NIH guidelines, and procedures performed were approved by the Institutional Animal Care and Use Committee of The Jackson Laboratory. Reagents and compositions used in this protocol are listed in Table of Materials and Table 1, respectively.
1. Preparation of reagents
2. Dissection and fixation of prenatal ovaries (Figure 1A)
3. Dissection and fixation of prepubertal ovaries (Figure 1B)
4. Perfusion, dissection, and fixation of pubertal ovaries (Figure 1C)
5. Whole-mount ovary immunostaining (Figure 2A)
NOTE: Practice sterile techniques during the immunostaining protocol, especially when changing buffers, to prevent contamination during extended incubation periods.
6. Clearing of immunostained whole-mount ovaries (Figure 2A).
NOTE: Perform all clearing steps in the dark by wrapping the plates in aluminum foil or placement in opaque containers. The clearing steps differ for prepubertal and pubertal ovaries.
7. Sample setup and imaging with a multiphoton microscope
NOTE: All steps described below were performed with a Leica DIVE/4TUNE/FALCON with two tunable mode-locked Ti:Sapphire multiphoton lasers with a pulse duration of 120 fs with a multi-immersion 16x/NA0.6 objective (immersion liquid = glycerol) with a maximal working distance of 2.2 mm. See the Table of Materials for details about the image acquisition software. Supplemental Table S1 and Supplemental Figure S1 show the settings used for this protocol. For other imaging platforms, consult with the microscopy core or follow the manufacturers' specifications/recommendations.
8. Image processing
NOTE: All steps described below were developed and performed using IMARIS 3D image visualization and analysis software.
9. Oocyte Quantification
NOTE: Whole ovary immunofluorescence and 3D image visualization and analysis can be used for the estimation of oocyte numbers in whole ovaries (Figure 3 and Figure 4) using the Spot feature. The GCNA signal can be used to quantify oocytes in prenatal and prepubertal ovaries, as shown in Figure 4 (P5). In pubertal ovaries, use the DDX4 signal to quantify two oocyte populations in non-growing follicles ("ring-like" structure, closed arrow) and growing follicles (large structures, open arrow, Figure 4, P28).
10. Quantifying protein expression in ovaries
NOTE: There are several ways to quantify oocyte expression of specific markers using both the Spots feature (section 9 and step 10.1) and Surfaces feature (step 10.2). The Spots feature can be used for proteins with distinct localization patterns such as nuclear markers (GCNA), and the Surfaces feature can be used for proteins with non-uniform localization patterns as shown in Figure 5A where LINE-1 ORF1p intensities in E15.5 and E18.5 ovaries were measured. To calculate and compare the intensity of the protein of interest between two samples (e.g., timepoints, treatments, or genotypes), collect images with the same properties. Use samples with a more intense signal to determine the parameters that can be stored and used for the other samples.
11. Estimation of total oocyte numbers in damaged ovary with computational correction
NOTE: If a minor ovary damage occurs during dissection, it may be possible to computationally estimate total oocyte count. It is recommended to use intact ovaries from the same strain and developmental stage for oocyte number estimation as shown in Figure 6. Simulations performed with ovaries at E15.5 indicate that correcting for a ≥30% loss results in a significant deviation from actual numbers (Figure 6C).
Immunostaining and imaging of the whole ovary enables the visualization and quantification of oocytes or protein expression in ovaries at different developmental stages using the same technique and markers (Figure 3). This protocol was developed for a large-scale project in which analysis of ovaries at multiple stages and from multiple mouse strains was required. Here, we present data gathered for the C57BL6/J strain, a standard strain for genetic analysis. The technique presented here is straightforward, results can be obtained within 14-19 days (Figure 2D) and can be used for ovaries from prenatal, prepubertal, and pubertal females (Figure 1 and Figure 2A,B).
This approach was used to study the dynamics of the oocyte loss that occurs naturally during the formation of the ovarian oocyte reserve. In many organisms including the mouse, oocyte numbers are thought to peak during fetal life around the time that oocytes enter meiosis ~E13.5. Oocyte numbers decrease due to the still-not-fully-understood process of fetal oocyte attrition (FOA) (from ~E15.5 to E18.5), which has been mechanistically linked to the expression of the retrotransposon LINE-120,21. More oocytes are eliminated after E18.5 to P0 due to the elimination of abnormal oocytes by a meiotic quality checkpoint22,23. To investigate these processes using this protocol, we immunostained and imaged ovaries at different developmental stages using DDX4 and GCNA as oocyte markers as shown in Figure 3.
Small oocytes expressing GCNA and DDX4 are seen from E15.5 onwards and they represent oocytes during MPI or arrested at dictyate within primordial follicles. Larger growing oocytes with strong DDX4 expression are detected already in P2 ovaries where they most likely represent the first wave of follicles44. Increasing numbers of larger oocytes are seen in P5 and P28 ovaries. Images obtained by multiphoton microscopy (Figure 4A) were processed using IMARIS software and 3D rendering was performed to identify small and larger growing oocytes based on the nuclear GCNA signal and the size delineated by the DDX4 signal (Figure 4A and Video 1). Oocytes were counted as described in the protocol and results are summarized in Figure 4B. In agreement with previous studies, we observed a significant oocyte loss from E15.5 to E18.5 (~32%) and E18.5 to P2 (~24%). By the time females reach puberty (P28), only ~30% of oocytes present at E15.5 have survived.
This method can be also used to observe the consequences of genotoxic treatments such as irradiation, which has been shown to completely eliminate the primordial follicle reserve within one week23,38,45. A significant visual difference is evident between the whole ovary treated with radiation and the untreated control in Figure 3 (comparing P28 ovaries from treated and untreated females). In the P28 ovary without radiation exposure, we observed two oocyte populations labeled with DDX4; abundant small oocytes in primordial follicles (closed arrow); and larger oocytes of various sizes typically found in growing follicles (e.g., primary, secondary, and preantral) (Figure 3). In contrast, the ovary from a female exposed to 0.5Gy of γ-radiation at P7 is completely devoid of small oocytes in primordial follicles as previously observed in 2D sections. Interestingly, only larger oocytes of similar size survived radiation; these may be the first wave follicles already growing in the P7 ovary, as they are known to be resistant to radiation46 (Figure 3, P28).
In addition to the quantification of oocyte numbers, this protocol can be used for immunostaining and quantitative analyses of other proteins involved in oocyte development. For example, we used antibodies against LINE-1ORF1p protein, an RNA-binding chaperone protein, produced by LINE-1 retrotransposons (Figure 5). Increase in the abundance of LINE-1 elements from E15.5 to E18.5 has been proposed to cause oocyte elimination20,21. Indeed, using multiphoton-captured images and signal intensity analysis in IMARIS, LINE-1 ORF1p levels were observed to increase significantly in oocytes during this time, which correlates with a significant drop in oocyte numbers as shown in Figure 4B.
In circumstances where a part of an ovary is damaged or lost, ovaries from the same strain and developmental stage can be used to computationally estimate the number of oocytes within the missing region as shown in Figure 6. Data from E15.5 ovaries were used for simulations to test the accuracy of computational corrections. Oocytes in ovaries with up to 30% tissue damage can be computationally estimated to have ≤10% deviation compared to oocytes in an intact ovary, suggesting that 3D ovary staining and modeling can be effectively used to salvage data from precious tissues. Simulations performed with ovaries at E15.5 indicate that correcting for a ≥30% loss results in a significant deviation from actual numbers (Figure 6C).
Figure 1: Ovary dissection and perfusion from females from prenatal and postnatal stages. (A–C) Ovary dissection from different stages requires different techniques, which are depicted schematically to complement the descriptions in the video. (D) Ovaries at postnatal day 5 (P5) are much smaller than at P28, which necessitates a different clearing protocol as described in the protocol. (E,F) Proper perfusion of the ovaries is important to eliminate background staining from red blood cells. Non-perfused reproductive tracts and ovaries have pink hue while perfused organs will turn white. Abbreviation: RI = non-perfused reproductive tracts. Images A–C were created with BioRender.com. Please click here to view a larger version of this figure.
Figure 2: Immunostaining, clearing, and imaging of whole mouse ovaries. (A) Flow chart depicts shared and specific steps for immunostaining and clearing protocol for prenatal/prepubertal and pubertal ovaries. (B) Cleared ovaries are mounted in a drop of clearing solution in the middle of an adhesive well sandwiched between two cover slips. For imaging, the mounted samples are placed in a 3D printed adaptor slide. (C) Greyscale multiphoton images of non-perfused vs. perfused ovaries, immunostained with the oocyte marker DDX4, showing improved image quality and lower nonspecific staining after perfusion. The closed arrow indicates an oocyte with a thin layer of cytoplasmic DDX4 staining typical for primordial follicles, and the open arrows show larger oocytes within growing follicles. Asterisk indicates autofluorescence from blood vessels. (D) 3D renders from confocal vs. multiphoton images of P5 ovaries, immunostained with oocyte markers GCNA (green) and DDX4 (magenta), showing a significant spherical aberration from confocal microscopy and virtually none from multiphoton. Scale bars = 100 µm (C, D) and 50 µm (insets in D). Abbreviations: GCNA = germ cell nuclear acidic peptidase; DDX4 = DEAD-box helicase 4. Image A was created with BioRender.com. Please click here to view a larger version of this figure.
Figure 3: Representative 3D-rendered images of ovaries of different developmental stages. 3D renderings from multiphoton images of whole mount ovaries immunostained with GCNA (green) and LINE-1 ORF1p (blue) or DDX4 (magenta), in prenatal and postnatal ovaries, respectively. Individual channels are presented in grayscale to show nuclear and cytoplasmic signals. The white boxes outline regions magnified in the left bottom insets. Two different ovaries are shown for P28. The ovary from the control non-irradiated female (top) contains a large population of small oocytes in primordial follicles (see inset). In contrast, the ovary from the female irradiated at P7 with 0.5 Gy of γ-radiation (IR) is completely devoid of small oocytes in primordial follicles (see inset). The closed arrow indicates an oocyte with a thin layer of cytoplasmic DDX4 staining typical for primordial follicles, and open arrows show larger oocytes within growing follicles. Scale bars = 100 µm; 30 µm for P28 insets; 10 µm for all other insets. Insets contain magnified views of 3D-rendered images generated in IMARIS and may differ slightly from low-magnification images due to perspective. Abbreviations: Non-IR = control non-irradiated; IR = irradiated at P7 with 0.5 Gy of γ-radiation; GCNA = germ cell nuclear acidic peptidase; DDX4 = DEAD-box helicase 4; LINE-1 = long interspersed nuclear element-1. Please click here to view a larger version of this figure.
Figure 4: Image processing, 3D display of oocytes, and quantification results. (A) 3D renders from multiphoton images (left) were processed in IMARIS using Gaussian filter (middle) and oocytes of small and large sizes were identified using the spot feature (right). P5 and P28 ovaries shown as example. Scale bars = 100 µm (P5); 10 µm (P5 insets); 300 µm (P28); 50 µm (P28 insets). (B) Small oocytes positive for GCNA were quantified in ovaries from different stages using spot features (top). To illustrate decreasing numbers of oocytes during development, the average % of oocytes were calculated at each stage as compared to the average number present in the earliest stage counted at E15.5 (bottom). Note the large drop in oocyte numbers from E15.5 to E18.5. Data are presented as means ± SD. Statistical analyses were performed using GraphPad Prism software and analyzed by one-way ANOVA, and the significance was determined by Bonferroni's post hoc multiple comparison test. * P ≤ 0.05; **** P ≤ 0.0001; ns >0.05. Abbreviation: GCNA = germ cell nuclear acidic peptidase; ns = not significant. Please click here to view a larger version of this figure.
Figure 5: Detection and quantification of theLINE-1 ORF1p expression in fetal oocytes. (A) 3D renderings from multiphoton images show LINE-1 ORF1p (blue) expression in E15.5 and E18.5 fetal oocytes (marked by green GCNA). (B) 3D surfaces generated in IMARIS using top row images in panel A. (C) LINE-1 ORF1p intensity analysis shows higher expression of LINE-1 ORF1p per oocyte at E18.5 than E15.5. Scale bars = 100 µm; 10 µm (panel A insets); 30 µm (panel B insets). Data are presented as means ± SD. Statistical analyses were performed using GraphPad Prism software and analyzed by Student's t-test, and the significance was determined by Mann-Whitney U test. P ≤ 0.0001; ns >0.05. Abbreviations: GCNA = germ cell nuclear acidic peptidase; LINE-1 = long interspersed nuclear element-1. Please click here to view a larger version of this figure.
Figure 6: Method for estimating total oocyte numbers in damaged ovarian samples with computational correction. (A) Model of GCNA-positive (green) E15.5 oocytes with five 10% regions highlighted in red, blue, gray, magenta, and brown in an intact ovary. Each succeeding image, after the intact ovary, represents a simulated ovary with 10% incremental regions missing up to 50%. (B) Schematic of computational method to estimate oocyte number in damaged sample. (C) Total oocyte numbers in simulated ovaries were compared to numbers in the original intact ovaries (considered 100%) and the difference is presented as % deviation. Simulation from six individual ovaries with 10-50% volume missing. Scale bars = 80 µm. Abbreviation: GCNA = germ cell nuclear acidic peptidase. Please click here to view a larger version of this figure.
Video 1: 3D rendering and 3D modeling of oocytes in P5 ovaries. Please click here to download this Video.
Solutions and buffers | Reagent(s) | Composition |
Fixative | Paraformaldehyde | 4% (v/v) |
Permeabilization buffer | Polyvinyl alcohol (PVA) | 0.2% (w/v) |
Sodium borohydride | 0.1% (w/v) | |
Triton X-100 | 1.5% (v/v) | |
Blocking buffer | Bovine Serum Albumin (BSA) | 3% (w/v) |
1 M Glycine (pH 7.4) | 2% (w/v) | |
Triton X-100 | 0.1% (v/v) | |
200x Penicillin-Streptomycin | 1% (v/v) | |
10% Sodium azide | 0.2% (v/v) | |
goat serum | 10% (v/v) | |
Washing buffer | PVA | 0.2% (w/v) |
Triton X-100 | 0.15% (v/v) | |
10% Sodium azide | 0.1% (v/v) | |
ScaleS4(0) solution (pH 8.1) | D-Sorbitol | 40% (w/v) |
Urea | 24% (w/v) | |
Glycerol | 10% (v/v) | |
DMSO | 20% (v/v) | |
ScaleCUBIC-1 solution | Urea | 25% (w/v) |
N,N,N′,N′-Tetrakis(2-Hydroxypropyl)ethylenediamine | 25% (v/v) | |
Triton X-100 | 15% (v/v) | |
Sucrose Solution | Sucrose | 20%(w/v) |
ScaleCUBIC-2 solution | Sucrose | 50% (w/v) |
Urea | 25% (w/v) | |
Triethanolamine | 10% (w/v) | |
Triton X-100 | 0.1% (v/v) |
Table 1: Solutions and buffers.
Supplemental Figure S1: Image acquisition preferences. Please click here to download this File.
Supplemental Table S1: Image acquisition settings. Please click here to download this Table.
This article presents a detailed 3D immunostaining and imaging protocol for prenatal and postnatal ovaries for high-throughput and comparative studies for germ cell quantification and protein localization. We developed this protocol to analyze oocyte numbers in ovaries (N=6-12) at six developmental time points in 10-16 different strains, where 2-4 24-well plates are typically processed at one time. This method can be adapted for other organs or cellular markers. For example, this protocol can be used to label and visualize somatic cells, such as granulosa cells in the ovary, using appropriate antibodies, thus facilitating studies of somatic-germ cell interactions or development of other ovarian cell types.
One limitation of this protocol and antibody combination is the definitive identification of different follicular stages. DNA stains used in immunofluorescence and 2D imaging to identify follicular stages by layers of granulosa cell nuclei are insufficient for 3D approaches. The 4′,6-diamidino-2-Phenylindole (DAPI) or Hoechst are rarely used for whole-organ staining due to limited light penetration and decaying signal in the middle of large tissue. Propidium iodide (PI) is a small molecule able to penetrate deep into the tissue but is difficult to capture on the multiphoton imaging system due to spectral overlap. Better resolution of follicular stages can be achieved by additional markers specific to granulosa cells such anti-Mullerian hormone (AMH) or FOXL247,48. However, although these markers are helpful in differentiating larger growing follicles, they will not distinguish primordial from primary follicles. Until specific markers for these early stages become available, oocyte markers such as DDX4 or GCNA offer the best indicator of follicle development. Further, this protocol works for prenatal male gonads but has not yet been tested in the postnatal testis where its size may be the limiting factor. Critical issues are listed below for better visualization and quantification of germ cells in ovaries of varying sizes and the techniques and steps taken to ensure good quality immunostaining for downstream analysis are highlighted.
The first critical step is to prevent damage to the ovaries before and after fixation. Damaged ovaries may lack parts of the organ structure, affecting the oocyte numbers and skewing the experimental results. To prevent damage, extra somatic tissue is left attached to the larger ovary after dissection to grab with forceps for transferring the ovaries. Further, for smaller ovaries, transferring them with a pipette tip cut wide enough for the ovaries to easily fit within the tip avoids damage. If minor tissue damage results in a part of the ovary missing, it can be mitigated by a computational method where total oocyte numbers in the damaged sample can be extrapolated using intact samples of the same type (Figure 6).
However, computational simulations revealed that the accuracy of prediction diminishes with increasing size of the damaged area (100.8% ± 0.2, 97.2% ± 1.5, and 90.3% ± 1 for 10, 20, and 30% damage, respectively Figure 6C). Based on simulations, we recommend that samples with >30% loss be excluded from analysis. Multiple intact ovaries from the same sample type were used to calculate the average number of oocytes in the area similar to the one missing in damaged sample. This number was then used to predict the total number of oocytes in the ovary with damage. Using similar size area in the same ovary abutting the missing region may be used as an alternative but was not tested.
Another issue is to avoid contaminating cells that might cause autofluorescence. For pubertal ovaries, perfusion, as described above, with both 1x PBS and 1% PFA must be performed to eliminate blood cells that will cause autofluorescence (Figure 2C). Perfusion with PBS should continue until organs, especially the ovaries and kidneys, are cleared of blood and become white before switching to 1% PFA (Figure 1E,F). Inefficient perfusion with PBS will result in a high background that may mask and make the visualization of small oocytes difficult. Perfusion with lower (1%) or higher (4%) percentages of PFA results in better image quality than perfusion with PBS alone; hence, the lower percentage of PFA was used.
Optimization of antibody staining requires attention to several variables. For the permeabilization step, 4 h was found to be ideal for this assay, with periods longer than 4 h (6-8 h tested) resulting in no staining at all. Because the amount and cost of antibodies used for large-scale immunostaining experiments can be high, this protocol was tested and optimized for reusing antibody mixes without any loss of immunostaining quality. For reuse, the antibody mixture must be supplemented with fresh antibody as described in the protocol. Washing steps are also critical for the quality of immunostaining and should be performed longer to achieve a better signal-to-noise ratio during imaging and should be determined for each antibody of choice. For best results, we recommend preparing the ScaleS4(0) clearing solution freshly for each use, while ScaleCUBIC-1 and ScaleCUBIC-2 can be stored at room temperature in the dark for ~1-2 months.
The ScaleCUBIC-1 clearing is performed at 37 °C; importantly, higher temperatures should be avoided as they will result in reagent precipitation, which can affect the quality of the immunostaining. Mounting multiple samples for imaging can be time-consuming. Agar or other embedding methods used in other protocols are labor-intensive and may need an additional re-clearing of samples47,49,50. Reusable silicon gaskets can be used as adhesive wells that are easy to use with coverslips and are offered in different sizes and depths to accommodate other sample sizes. To achieve high-quality images, a sufficient volume of clearing solution must be used; too little will result in poor image quality, and too much solution can leak onto the microscope. This step should be optimized for specific samples depending on their sizes. Glass-bottom dishes may be used but samples must be immobilized for imaging as even the slightest movements will distort the image. Silicon gaskets may be used in combination with a glass-bottom dish and top coverslip to immobilize the samples for imaging. However, this will limit the ability to flip the sample if imaging from both sides is needed due to the size of ovary.
Another critical choice for this large-scale, high-throughput approach was imaging on the Leica DIVE/4TUNE/FALCON platform with two tunable Spectra-Physics multiphoton (MP) lasers. The benefits of using the DIVE platform include ease of sample mounting (described above), minimization of optical distortion (compared to confocal platforms), and most importantly, imaging speed and management of acquired image data. Imaging samples with the multiphoton lasers at higher magnification (16x) is much faster and causes less photodamage than the confocal lasers on the Leica DIVE or SP8 using similar settings for image acquisition with LAS X software.The MP beam is low, its fluorophore excitation is highly localized to the focal point and achieves more depth at low magnification without inflicting photodamage despite a higher number of spatial/axial imaging steps (confocal: ~300-400 sections maximally, MP: >800 sections).
The 16x/NA0.6 objective allows correction for slight refractive index mismatches due to sample and mountant distribution along the optical axis via a mechanical correction collar, which is set optimally for maximal signal, half-way into the sample. Moreover, spherical structures, such as oocytes, can be distorted by confocal image acquisition as depth-dependent localization of the confocal plane (signal) due to mountant and sample refractive index mismatches at a given pinhole setting can occur (Figure 2D). As the fluorescence signal is being captured via the instrument's internal pinhole to remove out-of-focus signal, the intrinsic optical distortion (spherical aberration) may get worse deeper into the sample, which is problematic for larger ovaries (i.e., ~400-600 micrometer thickness).
With the same matching lens-correction settings, virtually no spherical aberration can be detected on the DIVE MP system with varying emission wavelengths, thus simplifying quantification and co-localization studies (Figure 2D). Last, the high-power pulsed laser allows for exciting lower fluorophore concentration deep inside the sample, in particular, when it is used with the Z-depth excitation correction. Moreover, the capture of weaker signals is also enhanced by the MP hybrid detector (HyD-RLDs), which is a combination of regular photomultiplier tubes and avalanche photon-counting detectors, allowing for more than 30 times higher signal detection over a wide range of emission wavelengths.
Light sheet (LS) multiphoton microscopy is another emerging platform for 3D tissue imaging47. However, LS demands more setup time for macroscopic samples, including agar embedding and set up for imaging. Moreover, images acquired at higher magnification with sample rotation can generate large >100 GB files, which may be necessary to remove light sheet shadowing artifacts (due to refractive index mismatches of the outside medium and the interior of the sample, or sample-surface distortions). The presented method of mounting samples in solution in adhesive wells is simple and fast.
Moreover, this approach allows the imaging of multiple ovaries in one well with one setup and without flipping or rotation. However, flipping the coverslip "sandwich" may be needed for larger ovaries. The DIVE system imaging is faster than confocal and LS; at 16x magnification, the prepubertal ovary is imaged in 3-5 min (Z-step 2 µm) and the pubertal ovary in 1 h (Z-step 5 µm). Moreover, files generated in LAS X are of manageable size (2-20 GB), which is also critical for a large number of samples for downstream analysis. Although this immunostaining protocol was optimized for use in combination with the DIVE platform, immunostained samples can be imaged using other MP/LS platforms with some modifications.
The advantage of the whole ovary staining with 3D modeling protocol is the efficient and relatively easy data analysis compared to manual data acquisition in 2D analysis. The IMARIS 3D visualization and analysis software was chosen because it is commercially available and offered by many microscopy cores, does not require programming skills, and is compatible with many image acquisition software platforms such as LAS X from Leica or ZEN from ZEISS. With IMARIS, the standard parameters are set up as highlighted in the protocol section, and these parameters can be used for other images and samples with minor changes, thus improving reproducibility and efficiency.
Image analysis and quantification are optimized for IMARIS software; however, similar analyses can be done on other commercial or open-source data visualization software following similar principles. In conclusion, an optimized protocol has been presented here for imaging entire ovaries for quantitative and qualitative analyses. This was purposefully adapted for the demands of the high-throughput processing that will be increasingly required for toxicology testing, clinical and diagnostic purposes, and genome-wide analyses of regulators of ovarian sufficiency and function.
The authors have nothing to disclose.
This work was supported by the National Institutes of Health grants (R01 HD093778 to E.B-F and T32 HD007065 to R.B). We thank Zachary Boucher for his assistance with radiation experiment. We thank Mary Ann Handel for critical reading of the manuscript. We gratefully acknowledge the contribution of Sonia Erattupuzha and the Microscopy Core Service at The Jackson Laboratory for expert assistance with the microscopy work described in this publication and Jarek Trapszo from the Scientific Instrument Services at The Jackson Laboratory for designing the 3D-printed adaptor slide.
Benchtop Incubator | Benchmark Scientific | H2200-H | 37 °C incubator |
Bovine Serum Albumin (BSA) | VWR | 97061-416 | |
C57BL/6J | The Jackson Laboratory | 000664 | mouse inbred strain |
Dimethyl sulfoxide (DMSO) | Sigma-Aldrich | D1435 | Hazardous material |
D-Sorbitol | Sigma-Aldrich | S6021 | |
Dumont #5 Forceps | FST | 91150-20 | |
FastWells Reagent Barriers | GraceBio | 664113 | Sticky and flexible silicone gasket (adhesive well) |
Fine Scissors | FST | 91460-11 | |
Glycerol | Sigma-Aldrich | G2025 | |
Glycine | ThermoFisher Scientific | BP381-500 | |
Goat anti-Rabbit IgG (H+L) Cross-Adsorbed Secondary Antibody, Alexa Fluor 647 | Invitrogen | A-21246 | Dilution 1:1000 |
Goat anti-Rat IgG (H+L) Cross-Adsorbed Secondary Antibody, Alexa Fluor 555 | Invitrogen | A-21434 | Dilution 1:1000 |
Goat serum | Sigma-Aldrich | G9023 | |
IMARIS Software | Oxford Instruments | Version 9.7.0 | Image visualization and analysis software |
Insight X3 | Spectra-Physics | InSight X3 Tunable Ultrafast Laser | Laser for Multiphoton Imaging |
LASX software | Leica | Version 3.5.6 | Image acquisition software |
Leica DIVE/4TUNE/FALCON | Leica | Leica Dmi8, 2P-M-ready: # 158005406 | Multiphoton Microscope |
MaiTai HP | Spectra-Physics | Mai Tai DeepSee One Box Ultrafast Laser | Laser for Multiphoton Imaging |
Masterflex Pump Controller | SPW Industrial | Model: 7553-50 | Peristaltic pump for perfusion |
Mayo Scissors | FST | 14010-17 | 5” –7” blunt/blunt scissors for decapitation |
Micro Cover Glasses, Square, No. 1.5 25x25mm | VWR | 48366-249 | |
Mini BioMixer | Benchmark Scientific | B3D1020 | shaker/nutator for 37 °C incubator |
Nikon Ergonomic SMZ1270 | Leica | SMZ1270 | stereomicroscope |
Paraformaldehyde 16% (formaldehyde aqueous solution) | Electron Microscopy Sciences | 15710 | Hazardous material |
PBS Tablets, Phosphate-buffered Saline | ThermoFisher Scientific | BP2944100 | Dissolve in Milli-Q water |
Penicillin-Streptomycin, 200x, Dual Antibiotic Solution | ThermoFisher Scientific | ICN1670249 | |
Polyvinyl alcohol (PVA) | Sigma-Aldrich | P8136 | |
Quadrol (N,N,N′,N′-Tetrakis(2-Hydroxypropylethylenediamine) | Sigma-Aldrich | 122262 | |
Rabbit anti-DDX4/MVH | Abcam | ab27591 | Dilution 1:500 |
Rabbit anti-LINE-1 ORF1p | Abcam | ab216324 | Dilution 1:500 |
Rat anti-TRA98/GCNA | Abcam | ab82527 | Dilution 1:500 |
Sodium azide | Sigma-Aldrich | S2002 | Hazardous material |
Sodium borohydride | Sigma-Aldrich | 452882 | Hazardous material |
Sucrose | ThermoFisher Scientific | S0389 | |
Tekmar Orbital Shaker | Bimedis | VXR-S10 | shaker for room temperature |
Triethanolamine | Sigma-Aldrich | 90279 | |
Triton X-100 | Sigma-Aldrich | X100 | |
UNOLOK Infusion Set | MYCO Medical | 7001-23 | needles for perfusion |
Urea | Amresco | 97061-920 | |
X-Cite 120LED | Excelitas | S/N XT640-W-0147 | low-power LED fluorescence lamp |