Detection of Pou5f1/Oct4, Emi2 and cyclin B1 mRNAs by whole-mount fluorescence in situ hybridization of mouse oocytes and embryos
We previously developed a high-sensitive and high-resolution section in situ hybridization method using in vitro-synthesized RNA probes and the TSA system [21, 23, 25]. In this method, organs and tissues such as zebrafish and mouse ovaries are fixed, dehydrated, and embedded in paraffin, followed by making thin sections, rehydration, and probe hybridization. This method enabled detection of the distribution of mRNA molecules in single cells of organs. However, it requires laborious efforts and large amounts of reagents to operate samples on the glass slides. In addition, small-size samples such as mammalian oocytes and embryos cannot be embedded in paraffin due to the inability to handle them in paraffin. Although an in situ hybridization method for mammalian oocytes and embryos without making paraffin sections would be useful, isolated mammalian oocytes and embryos are very fragile throughout all steps, particularly in the steps for hybridization and washing out non-hybridized RNA probes. To overcome these problems, we remarkably modified the procedure of our previous method and developed a whole-mount in situ hybridization method that is optimized for mammalian oocytes and embryos.
We first changed the fixation condition: the percentage of paraformaldehyde was increased up to 4%, which enables stable handling of mouse oocytes and embryos in all steps. Second, we modified the structure of glass pipettes by bending the bottom of the tip like an L-shape (Fig. 1A), which allows mild pipetting of the oocytes and embryos. Third, we added a detergent (0.1% Tween-20) to SSC wash buffers, which enables maintenance of the round shapes of oocytes and embryos without sticking to the walls of culture plate wells. All of these modifications enabled easy and safe handling of mouse oocytes and embryos. Since whole-mount samples tend to maintain endogenous peroxidase activities, which significantly increase the background of TSA signals, we treated the samples with 1% H2O2 in PBS containing 0.1% Tween-20 for 30 min to inactivate the endogenous peroxidase. Throughout all steps, we carefully placed the oocytes and embryos on the bottom of the plate well using the L-shaped glass pipettes (Fig. 1B).
After hybridization with DIG-labeled RNA probes and amplification of signals with the TSA system, we observed the mouse oocytes and embryos under a confocal microscope. No signal was detected in oocytes hybridized with the sense Pou5f1/Oct4 RNA probe (Fig. 2A). In contrast, bright signals were detected in the cytoplasm of mouse oocytes and 2-cell stage embryos hybridized with the antisense Pou5f1/Oct4 RNA probe (Fig. 2B and C). The signals of Pou5f1/Oct4 mRNA appeared to represent granular structures in the cytoplasm of oocytes and 2-cell stage embryos. In immature oocytes, many of the signals of Pou5f1/Oct4 mRNA appeared to congregate around the germinal vesicle (GV). In contrast, the signals appeared to be distributed all over the cytoplasm in 2-cell stage embryos.
The accumulation and distribution of Emi2 and cyclin B1 mRNAs were similarly examined. No signal was detected in oocytes hybridized with the sense Emi2 RNA probe, whereas bright signals were observed in the cytoplasm of oocytes hybridized with the antisense Emi2 RNA probe (Fig. 2D and E). The Emi2 mRNA was distributed throughout the oocyte cytoplasm (Fig. 2E). These signals disappeared in 2-cell stage embryos (Fig. 2F), consistent with the disappearance of Emi2/Fbxo43 mRNA after fertilization in RNA-seq analysis [6, 37]. No signal was detected in oocytes hybridized with the sense cyclin B1 RNA probe, whereas very bright signals were observed in oocytes hybridized with the antisense cyclin B1 RNA probe (Fig. 2G and H). The signals of cyclin B1 mRNA resembled granule structures and were also detected in the cytoplasm of 2-cell stage embryos, although the number and intensities of signals were reduced (Fig. 2H and I). Very bright signals were also detected in oocytes hybridized with the antisense cyclin B1 RNA probe labeled with Fluorescein but not in oocytes hybridized with the sense cyclin B1 Fluorescein-labeled RNA probe (Fig. S1A and B), confirming that DIG- and Fluorescein-labeled RNA probes were used in our method.
The differences in the signal intensities of Pou5f1/Oct4, Emi2 and cyclin B1 mRNAs are consistent with the differences in the amounts of mRNAs, i.e., the amount of cyclin B1 mRNA in mouse oocytes is significantly larger than the amounts of Emi2/Fbxo43 and Pou5f1/Oct4 mRNAs [6, 37]. Taken together, we conclude that this method enables detection of the accumulation and distribution of mRNAs in mouse oocytes and embryos in a highly specific manner with almost no background. Furthermore, the procedure of this method is easy and requires small amounts of reagents, even compared with in situ hybridization using paraffin-embedded sections [23], because all of the steps are performed using 24- and 96-well plates (Fig. 1B).
Analysis of the properties of Pou5f1/Oct4 RNA granules
Our previous study demonstrated that translationally repressed pou5f3 mRNA, which encodes Pou5f3, a homolog of Pou5f1/Oct4, was stored in zebrafish oocytes as RNA granules in a solid-like state [26]. These RNA granules changed into liquid-like droplets shortly after fertilization. We then analyzed the properties of Pou5f1/Oct4 RNA granules by treating mouse oocytes and embryos with hexanediol, which dissolves assemblies in a liquid-like state but does not affect assemblies in a solid-like state [38]. As a control, oocytes and embryos were treated with hexanetriol, a less effective chemical [38]. Whole-mount in situ hybridization of oocytes showed that Pou5f1/Oct4 RNA granules were not altered by treatment with hexanediol or hexanetriol (Fig. 3A and B). In contrast, Pou5f1/Oct4 RNA granules were dissolved by treatment with hexanediol, but not by treatment with hexanetriol, in 2-cell stage embryos (Fig. 3A and C). These results suggest that Pou5f1/Oct4 mRNA assembles into solid-like granules in oocytes and that the property is changed into a liquid-like state in embryos as in the case of zebrafish pou5f3 mRNA.
High-resolution analysis of cyclin B1 and Emi2 mRNAs
We then tested whether this whole-mount in situ hybridization method was applicable to the detection of different mRNAs simultaneously in the same sample. We first performed double in situ hybridization of cyclin B1 and Emi2 mRNAs by hybridizing mouse oocytes with the Fluorescein-labeled cyclin B1 RNA probe and the DIG-labeled Emi2 RNA probe. As observed in our previous study [13], section in situ hybridization showed that both mRNAs were distributed by forming different RNA granules (Fig. 4A). The whole-mount in situ hybridization method detected both mRNAs simultaneously in the same oocyte (Fig. 4B) and, remarkably, the intensities of the signals were significantly high compared with those in section in situ hybridization. The granular structures of cyclin B1 mRNA and Emi2 mRNA were observed in enlarged views of images (Fig. 4C). We found that the sizes of cyclin B1 RNA granules were larger than those of Emi2 RNA granules and that Emi2 RNA granules were localized in the spaces between cyclin B1 RNA granules. Consistent with observations in our previous study [13], granules of Emi2 mRNA and cyclin B1 mRNA were rarely co-localized in the cytoplasm of immature oocyte (Fig. 4B and C). Taken together, the results indicate that the whole-mount in situ hybridization method can simultaneously detect different mRNAs with high sensitivity and high resolution.
Super-resolution analysis of Pou5f1/Oct4 and cyclin B1 mRNAs
To investigate the localization of Pou5f1/Oct4 and cyclin B1 mRNAs in mouse oocytes, we then performed double in situ hybridization of Pou5f1/Oct4 and cyclin B1 mRNAs in immature oocytes. Consistent with the observations in single in situ hybridization (Fig. 2B), Pou5f1/Oct4 RNA granules were detected in the oocyte cytoplasm and were mainly localized in the area around the GV (Fig. 5A). cyclin B1 RNA granules were distributed throughout the oocyte cytoplasm (Fig. 5A). Enlarged views showed that Pou5f1/Oct4 and cyclin B1 RNA granules were distributed as different granules and that their sizes seemed to be varied (Fig. 5B). As in the case of cyclin B1 and Emi2 mRNAs, distinct RNA granules were rarely co-localized.
To examine the relationship between Pou5f1/Oct4 and cyclin B1 mRNAs in more detail, mouse oocytes were observed under an N-SIM super-resolution microscope, which allows a better resolution (< 120 nm) than that of conventional confocal microscopy. We constructed 3-dimensional (3D) distributions of the mRNAs in almost whole oocytes by taking 30-µm-thick z-stacks at 0.3-µm intervals (Fig. 5C). We first analyzed a 25 µm2 square region of z-stacks (boxed region in Fig. 5C). Although a single projection image of this region showed that some Pou5f1/Oct4 and cyclin B1 RNA granules were overlapped (Fig. 5D), analysis of the 3D reconstruction of the same image indicated that these RNA granules were not co-localized (Fig. 5E). We expanded this analysis to the whole oocyte. In analysis of the x-y axis of 1-µm-thick z-stacks, co-localization of Pou5f1/Oct4 and cyclin B1 RNA granules was not observed (Fig. 6). Similar results were obtained in the analysis of reconstruction in x-z and y-z axes (Fig. 7). Moreover, this super-resolution analysis demonstrated that Pou5f1/Oct4 and cyclin B1 mRNAs were distributed as granules of similar sizes (Fig. 5C-E, Figs. 6 and 7). RNA granules consisting of Pou5f1/Oct4 mRNA were frequently localized close to each other. A similar localization pattern was observed in RNA granules consisting of cyclin B1 mRNA.