Long-term live-cell imaging of mammalian preimplantation development and derivation process of pluripotent stem cells from the embryos


Author to whom all correspondence should be addressed.

Email: yamagata@biken.osaka-u.ac.jp


Mammalian fertilization is a process in which two highly specialized haploid gametes unite and endow totipotency to the resulting diploid zygote. This is followed by cell proliferation and the onset of differentiation during the brief period leading up to implantation. In these processes, a number of cellular components and structures are regulated spatially and temporally, as seen in repeated cell division, cell cycle progression, and epigenetic reprogramming. In mammals, the numbers of oocytes and embryos that can be collected are very limited. Therefore, analyses of molecular mechanisms are hampered because of difficulties in conducting biochemical analyses on such limited material. Furthermore, immunostaining methods require cell fixation and are insufficient for understanding ontogeny, because the processes observed in fertilization and early embryonic development progress in time-dependent manners and each phenomenon is connected with others by cause-and-effect relationships. Consequently, it is important to develop an experimental system that enables molecular imaging without affecting embryonic development. To achieve the above advantages, especially retrospective and prospective analyses, we have established a live-cell imaging system that enables observations under minimally invasive conditions. Using this approach, we have succeeded in visualizing and predicting the developmental potential of embryos after various perturbations. We also succeeded in imaging embryonic stem (ES) cell derivation in natural conditions. In this review, we describe a brief history of embryonic imaging and detailed protocols. We also discuss promising aspects of imaging in the fields of developmental and stem cell biology.

Importance of live-cell imaging in the study of preimplantation mammalian development

Mammalian fertilization is a process in which two highly specialized haploid gametes unite and endow totipotency to the resulting diploid zygote. This is followed by cell proliferation and the onset of differentiation during the brief period leading up to implantation. In these processes, a number of cellular components and structures are regulated spatially and temporally, as seen in repeated cell division, cell cycle progression, and epigenetic reprogramming. However, recent studies have indicated that these processes are not as secure as previously thought. Indeed, embryos from infertile couples were reported to show chromosome instability (Vanneste et al. 2009; Santos et al. 2010) and embryos reconstructed by assisted reproductive technologies are known to show epigenetic abnormalities, such as abnormal imprinting patterns (De Rycke et al. 2002; Kohda & Ishino 2013) in addition to genetic errors. Moreover, although it is an extreme example, embryos reconstructed by somatic cell nuclear transfer (SCNT) show global epigenetic errors and it has been thought that such abnormalities are the causes for the low success rate of animal cloning (Wakayama 2007; Ogura et al. 2013). Therefore, to determine the prerequisites to ensure full-term development, it is important to understand the events that take place in these abnormal embryos at the cellular and molecular levels.

In mammals, the numbers of oocytes and embryos that can be collected are very limited. Therefore, analyses of molecular mechanisms are hampered because of difficulties in conducting biochemical analyses on such limited material. Furthermore, immunostaining methods requiring cell fixation are insufficient for understanding ontogeny, because the processes observed in fertilization and early embryonic development progress in time-dependent manners and each phenomenon is connected with others by cause-and-effect relationships. Consequently, it is important to develop an experimental system that enables molecular imaging without affecting embryonic development. As for the advantages of live-cell imaging, several points can be raised. First, it allows observation of cells in a more natural state than conventional immunostaining techniques, which require invasive and lethal procedures such as fixation and permeabilization. Second, more information can be obtained by following development over time in seconds, minutes, hours, or even days. Third, this approach enables one to analyze the kinetics of molecules even in early embryos by studying intermolecular interactions using fluorescence resonance energy transfer (FRET), photomanipulation techniques such as fluorescence loss in photobleaching (FLIP), and fluorescence recovery after photobleaching (FRAP). Fourth, molecular-level phenomena observed in early embryos can be directly connected with their developmental potentials by culturing and transferring the embryos to host mothers after fluorescent observations (retrospective and prospective analyses). In particular, the retrospective and prospective analyses under the fourth point listed above have been less frequently described as an advantage of imaging techniques to date, but these are the most important and innovative aspects in the study of early embryonic development. For example, embryos constructed with particular reproductive technologies such as SCNT are considerably heterogeneous and their developmental rates vary between samples (Wakayama 2007). Furthermore, oocytes at an advanced stage of maturation are reported to be heterogeneous in terms of their developmental capacities (Wakayama et al. 2007). When such resulting embryos are immunostained, some abnormalities or differences with normal embryos might be observed, but it will be impossible to know their influences on subsequent development as the embryos are already dead. However, live-cell imaging allows us to link a specific phenomenon observed directly at a certain time to the embryo's developmental capacity by culturing and transplanting it to a recipient pseudopregnant mother. Therefore, to exploit the above advantages of live-cell imaging – especially for retrospective and prospective analyses – we aimed to establish a system that would enable observations under minimally invasive conditions.

A brief history of embryo imaging

Following the discovery of green fluorescent protein (GFP) from the jellyfish Aequorea victoria (Shimomura et al. 1962) and the successful cloning of its cDNA (Prasher et al. 1992), this protein has been used as a vital cell marker in many organisms, including nematodes, fish, and mice (Chalfie et al. 1994; Amsterdam et al. 1995; Okabe et al. 1997). In addition, live-cell imaging – a technology in which cells and/or molecules in living tissues are labeled with GFP and observed by fluorescence microscopy – was developed and has become a powerful tool in various fields of biology. Recently, along with the development of novel fluorescent proteins and the improvement of microscopes, not only changes in the localization of molecules and cells, but also molecular kinetics and interactions can be elucidated using live-cell imaging.

To our knowledge, Ikawa et al. (1995) were the first to apply GFP technology in mammalian preimplantation embryos. They used GFP to select transgenic preimplantation embryos before embryo transplantation to recipient foster-mothers, based on the presence or absence of GFP fluorescence in the embryos. Zernicka-Goetz et al. (1997) applied the GFP method to trace the cell fate in living mouse embryos. In their experiments, mRNA encoding the MmGFP (slightly modified to be suitable for mammalian cells) was injected into one side of 2-cell embryos for cell fate tracing. In 1998, Brunet et al. (1998) expressed a fusion protein of GFP with β-tubulin in mouse oocytes by mRNA injection and succeeded to capture the first time-lapse recording of spindle formation. Another interesting application was demonstrated by Hadjantonakis et al. (1998), who used GFP for noninvasive sexing of mouse embryos before they were transplanted. They produced a transgenic male bearing a ubiquitously expressed GFP gene in the X chromosome. After mating with a normal female, the resulting female embryos exhibited GFP fluorescence, but the male embryos did not. Following their work, it became widely accepted that the live-cell imaging technique has advantages for the study of preimplantation development, so the technique was applied to a range of such analyses, including investigations of changes in the localization of certain molecules (Dard et al. 2001), the correlation between cell lineage and blastocyst axis (Plusa et al. 2005; Kurotaki et al. 2007), and genomic reprogramming in cloned embryos (Yamagata et al. 2007; Cavaleri et al. 2008; Mizutani et al. 2012).

Progress in microscopic technology has also boosted the attempts to reduce phototoxicity. To our knowledge, Squirrell's paper was the first report to describe low damage live-cell imaging (Squirrell et al. 1999). They succeeded in obtaining hamster pups after the fluorescent imaging of embryos using two-photon laser scanning microscopy. Unfortunately, in their report, the imaging period was relatively short (up to 24 h) and it was impossible to construct the three-dimensional structure of embryos because there were only a few images acquired in the vertical axis. Moreover, they failed to obtain any pups after the imaging probably because of the use of scanning-type confocal microscopy. Ross et al. (Ross et al. 2006) conducted a similar trial in mouse embryos using a spinning-disk microscope with a metal halide lamp as a light source. Although the pups were obtained after ‘snapshot’ imaging at one time point, they did not succeed in acquiring time-lapse images of preimplantation embryos without causing damage. In addition, both of these groups used only MitoTracker (Invitrogen, Carlsbad, CA, USA), a fluorescent agent used for mitochondrial imaging, as a marker, so the application was restricted.

At that time, we considered that these problems might be overcome by developing a new live-cell imaging technique for preimplantation embryos. If successful, it would become a powerful tool for the analysis of phenomena observed in early stages of embryo development and for determining their impact on developmental potential by both retrospective and prospective analyses. To this end, we developed a ‘low-invasive’ live-cell imaging technique by improving microscopy and optimizing the conditions for imaging (Yamagata et al. 2005, 2009b). Using this technique, we successfully obtained normal mouse pups from embryos injected with mRNAs encoding nucleus- and spindle-labeling proteins and time-lapse imaged these molecules throughout preimplantation development. We will describe the details of our imaging technology including the devices and fluorescence expression in the following sections.

Details of ‘low-invasive’ live-cell imaging technology

Device construction

Establishing a ‘low-invasive’ imaging platform is more critical for the three-dimensional imaging of embryos than of cultured cells, because the thickness of embryos (around 100 μm in diameter) requires exposure to multiple laser irradiations; it also takes a longer time to monitor development over multiple cell divisions. It is well accepted that repeated fluorescence microscopy observations severely impair cellular viability and functions through phototoxicity (Squirrell et al. 1999; Frigault et al. 2009; Yamagata et al. 2009b). Furthermore, because embryonic development and cellular differentiation take place three-dimensionally, conventional fluorescence microscopy using mercury or xenon lamps as a light source – the so-called wide-field microscope – is inadequate to capture whole images of specimens because of ‘out-of-focus blur’. Therefore, there were strong needs for establishing a confocal microscopic system optimized for long-term imaging in which the phototoxicity was minimized as much as possible to retain cellular viability and functions.

To satisfy these requirements, the use of two-photon laser confocal microscopy was one of the candidates because it uses long wavelength excitation (near-infrared) and only the confocal plane is excited (Squirrell et al. 1999). However, in our experience, this method of microscopy does not necessarily naturally cause low damage, probably because it needs a high-power laser to obtain sufficient fluorescence and much heat is produced by this beam. Therefore, we constructed another imaging system optimized for low-invasive, three-dimensional imaging (Fig. 1A). The key feature of low-invasive imaging is the use of a spinning-disk confocal unit (CSU series; Yokogawa Electric Corporation, Tokyo, Japan) and ultra-high sensitive electron-multiplying charge-coupled device (EM-CCD) camera (iXon series; Andor Technology, Belfast, UK) attached to a conventional inverted microscope. Of course, the major advantage of this confocal device is that three-dimensional cellular images can be taken by eliminating out-of-focus blur. As another advantage, the optical intensity projected from the tip of the lens is very low compared with those of mercury and xenon lamps (Fig. 1D; Table 1). Also, when compared with laser scanning-type confocal microscopy, the light power delivered per unit volume in specimen can be extremely low because the laser beam line is separated though the multiple pinholes of the Nipkow disk (Nakano 2002; Toomre & Pawly 2006). Indeed, the use of conventional laser scanning confocal microscopy did not result in any live pups being born after repeated exposure of hamster embryos (Squirrell et al. 1999). For longer and safer imaging, the samples should be exposed to as low an intensity of light as possible; instead, the fluorescence signals should be collected at the maximum level of efficiency using an ultrasensitive camera such as the EM-CCD camera we chose.

Table 1. Comparison of light energy emitted from the lens tip
MicroscopyLight sourcesMax power (mW, Catalog spec.)Light intensity from the lens tip (mW)*
488 nm561 nm
  1. *The intensities were measured using a power meter (Yokogawa Electric Corporation, TB200, see Fig. 1C,D. The values were determined when the light sources were set at maximum power. Total power of laser including all wavelengths. Because these values indicate rated input, actual output power would be lower. n.d., not detected.

Yokogawa's CSUSolid-state laser (488 nm)300.588n.d.
Solid-state laser (561 nm)25n.d.0.453
Argon-Krypton laser700.2810.225
Wide-field microscopyMercury100 00044.199.5
Xenon75 00016.816.7
Figure 1.

Devices used for ‘low-invasive’ live-cell imaging. (A) A photograph of the equipment. A conventional inverted microscope is attached to a Nipkow disk confocal unit, an electron-multiplying charge-coupled device (EM-CCD) camera, filter wheels, a z-axis motor and an automatic x–y axis stage. For excitation, solid-state lasers (405, 488 and 561 nm) are used as light sources. The filter wheel changes emission channels automatically. All are controlled using MetaMorph imaging software. Embryos are cultured in an incubator on the stage, which is set at 37°C and gassed with 5% CO2 in air at 160 mL/min. The company and model names of all the devices are listed in table below. (B) An incubator chamber on the stage. A glass-bottomed dish is put in the center of the chamber and surrounded by water bath. (C) A power meter (TB200, Yokogawa Electric Corporation). Arrow indicates the sensor that detects the light. (D) Measuring the power of 488 nm light at the lens tip. (E) A thermometer (TSU, Tokai hit) that has very narrow (φ0.25 mm) thermo censor (arrow). (F) Measuring the temperature of culture drop adjacent to the embryos by the tip of the thermometer censor (arrow).

Embryos are cultured in an incubator chamber (MI-IBC; TOKAI HIT, Shizuoka, Japan) placed on the microscope stage (Fig. 1A,B). In the chamber, a 35-mm glass-bottomed dish is placed in the center and surrounded by a water bath to maintain humidity inside. The temperature inside the chamber is regulated with four heaters: a stage heater, a water bath heater, a lens heater wound around the lens, and a top heater (for the cover of the chamber). Each heater is adjusted to keep the medium droplets in the dish at 37°C by measuring with a very narrow thermo sensor (TSU; TOKAI HIT, Fig. 1E,F). Furthermore, the whole imaging system is kept in a darkroom maintained at a constant 30°C. This is because almost all the heat of the chamber was lost from the system if it was kept at the usual room temperature even if the incubator was set at 37°C. Also, if too much stress is applied to each heater, they can malfunction. The air atmosphere was kept at 5% CO2 in 95% air by introducing a gas mix into the chamber at 160 mL/min. The system was equipped with an x–y-axis motored stage (Sigma Koki, Tokyo, Japan). This can record the precise position of the samples, so that multiple samples can be observed repeatedly.

Fluorescence expression

To express any exogenous protein in cultured cells, it is common to use transfection with plasmid DNA followed by transcription and translation of the protein. For example, when using GFP, it takes approximately 24–72 h to obtain sufficient fluorescence signals. In the case of mouse embryos, one has to wait until the 2-cell stage at the earliest to obtain measurable fluorescent signals, even if powerful promoters are used (Devgan et al. 2004). Hence, it is not possible to monitor phenomena just after fertilization using this strategy. Alternatively, it is possible to generate transgenic animals that express GFP-fused proteins in the oocyte (Plusa et al. 2005; Kurotaki et al. 2007), but this requires expensive facilities to maintain the mouse strains. Therefore, we decided to inject mRNA synthesized via in vitro transcription into the oocyte cytoplasm (Vassalli et al. 1989). Because translational efficiency depends on the length of the poly(A) chain of mRNA in oocytes (Richter 1999), our plasmid was designed to contain a longer poly(A) sequence at the 3′-end of the mRNA transcribed in vitro (Fig. 2B). Furthermore, for in vitro transcription, the addition of a cap structure at the 5′-end enhances the translation of exogenous mRNA in the oocyte. In our imaging system, in vitro synthesized mRNA was microinjected into unfertilized or fertilized oocytes, followed by insemination using in vitro fertilization (IVF) or other reproductive techniques as necessary (Fig. 2A). After that, the constructed embryos were observed by fluorescence microscopy while being cultured inside a CO2 incubator placed on top of the stage. Then, the observed embryos were transplanted into pseudopregnant females and their developmental capacities were determined (Fig. 2A).

Figure 2.

Procedures for imaging and plasmid map of pcDNA3.1 enhanced green fluorescent protein (EGFP)-poly(A). (A) After in vitro fertilization, anaphase/telophase stage oocytes (1–2 h after insemination) were injected with mRNA solutions, incubated until fluorescence become detectable (approximately 3 h) and placed under the imaging system. Metaphase II (unfertilized) oocytes were injected with mRNA when embryos produced by intracytoplasmic sperm injection (ICSI) or somatic cell nuclear transfer (SCNT) were used for imaging. The potential for further development of imaged embryos can be analyzed by transferring them to pseudopregnant recipient foster-mothers after image analysis. The mRNA with cap structure at 5′ terminal and long poly(A) tail in 3′ end, respectively was synthesized with a commercially available in vitro transcription kit using a pcDNA3.1-poly(A) vector (see below) as a template DNA. (B) We generated a plasmid (pcDNA3.1-poly(A)) to synthesize mRNA in vitro, which is efficiently translated in oocytes and zygotes. mRNA is transcribed under the T7 promoter with a stretched poly(A) sequence downstream of and a 3′-untranslated region (UTR) derived from the mouse TRF2 gene. As this plasmid also has a cytomegalovirus (CMV) promoter and putative poly(A) attachment signal (blue letters in sequence) in the 3′-UTR, encoded proteins can be expressed by transfection into various cell lines. The represented map is the plasmid inserted the EGFP open reading frame (ORF)―pcDNA3.1EGFP-poly(A)―and the sequence in below is for the boundaries of 3′ region of ORF.

RNA preparation

After linearization of the template plasmid (Fig. 2B) at the Xba I, Xho I, or Apa I sites, mRNA was synthesized using RiboMAX Large Scale RNA Production Systems-T7 (Promega, Madison, WI, USA). For efficient translation of the fusion proteins in embryos, the 5′ end of each mRNA was capped using Ribo m7G Cap Analog (Promega), according to the manufacturer's protocol. To circumvent the integration of template DNA into the embryo genome, reaction mixtures of in vitro transcription were treated with RQ-1 RNase-free DNase I (Promega). Synthesized RNAs were treated with phenol–chloroform followed by ethanol precipitation. After dissolution in RNase-free water, mRNAs were subjected to gel filtration using a MicroSpin G-25 column (Amersham Biosciences, Piscataway, NJ, USA) to remove unreacted substrates and then stored at −80°C until use.

RNA injection

Each synthesized mRNA was diluted to an appropriate concentration using Milli-Q ultrapure water (Millipore, Madison, WI, USA) and an aliquot was placed on a micromanipulation chamber (Fig. 3A). Diethylpyrocarbonate (DEPC)-treated water should not be used for dilution, as this chemical is a broad inhibitor of any proteins including RNase by modifying their histidine residues and might affect embryonic viability if its activity persists in the solution. The concentration of mRNA is a critical factor for the imaging and embryo development. Indeed, although a higher concentration yields higher intensity, an excess amount of mRNA injected hampers the developmental capacity of embryos, probably because it interferes with the intrinsic translational machinery. Therefore, we usually inject 1–50 ng/μL mRNA solution, although the amount used depends on the RNA species.

Figure 3.

Procedures for the microinjection of mRNA. (A) A photograph of the injection chamber. A lid of 60 mm dish is used. Seven microliters of PVP solution (12% Polyvinylpyrrolidone diluted by Hepes-CZB medium) and Hepes-CZB medium for the injection, and >1 μL of mRNA solution were placed on the lid and covered by mineral oil. Before the injection, an inner wall of the glass pipet was soaked with PVP solution at once, washed by Hepes-CZB, and then, microinjection of mRNA was performed in injection drops as described in the text. (B) Scheme of mRNA injection using a piezo-activated micromanipulator. Once mRNA solution had been sucked into the glass pipette (2–3 μm diameter), piezo pulses were applied to oocytes placed in separate drops to break the zona (panel 1, 2) and plasma membrane (panels 3–4). A few picoliters of solution were introduced into the oocytes (panel 5) and the pipette was removed gently (panels 6). Panel 7 is a magnified image of the boxed area of panel 5 and shows the displacement of ooplasm by the emergence of the mRNA solution from the tip of the pipette (arrow). The volume of solution injected is controlled by seeing the extent of displacement. (C) Typical time-lapse images of spindle (Tubulin, green) and nuclear (H2B, red) dynamics during preimplantation development. The fertilized oocytes were injected with 5 ng/μL each of enhanced green fluorescent protein (EGFP)-α-tubulin and histone H2B-mRFP1 mRNA. The images were taken using UPlanApo ×20 oil lens. Selected images are shown at every 2 h from a series acquired every 7.5 min. The time after starting observation is indicated at the bottom of the panel. The images of 51 sections in the z-axis were computationally stacked into one picture.

In the case of IVF-generated embryos, at 1–2 h after insemination, the fertilized oocyte is recovered by dissolving the cumulus cell mass with hyaluronidase treatment. Anaphase II/telophase II oocytes are transferred to HEPES-buffered Chatot–Ziomek–Bavister (CZB) medium (Chatot et al. 1990) in the same injection chamber (Fig. 3A) and injected with mRNA using a piezo-activated manipulator (Prime Tech, Ibaraki, Japan) with a narrow glass pipette (1–3 μm diameter). Once the mRNA solution has been aspirated into the pipette, piezo pulses are applied to the oocyte to break the zona pellucida and oolemma. A few picoliters of solution are introduced into the oocyte and the pipette is removed quickly (Fig. 3B). The volume of mRNA solution introduced into the ooplasm is controlled by eye; the ooplasm at the tip of the pipette is pushed away a little by the emerging mRNA solution (Fig. 3B). Importantly, when we inject the fluorescent dye into the ooplasm, the variation in intensity in each embryo subsequently is very small (approximately 0.8–1.2 units where the average intensity is 1), suggesting that our mRNA injection technique is reproducible. The mRNA-injected embryos are incubated at 37°C under 5% CO2 in air for at least 3 h to translate the mRNA sufficiently for imaging. More than 200 oocytes can be injected in 1 h and the survival rate after mRNA injection is nearly 100%.

Imaging conditions

To reduce phototoxicity, the conditions for image acquisition are also important, such as laser power, exposure time, and time intervals between the image acquisitions. We usually set the power of both the 488 and 561 nm laser projected from the lens tip at 0.1 mW. The power is measured by using optical power meter (TB200; Yokogawa Electric, Fig. 1C,D). In our experience, the developmental rate of the embryos excited by the 0.1 mW laser is similar to those of non-imaged controls, but when the excitation is set at 0.2 mW and more, the rate reduces and photobleaching occurs. When the laser power is set at the same value, a higher magnification objective lens yields higher toxicity, as the optical density per unit area becomes denser. Therefore, we usually use ×20 objective (UPlanApo ×20 oil, Olympus). If ×40 objective was used under the same conditions, the embryonic development was severely arrested.

In general, although longer exposure times yield brighter and higher signal-to-noise (S/N) ratio images, the phototoxicity becomes unacceptable. Because the embryo is thick, 51 images (2 μm intervals over 100 μm) are usually taken in the z-axis direction to cover it. This makes for a long total excitation time and causes phototoxicity. Therefore, we usually set the duration to ≤100 ms. In the case of CSU, when the exposure time is set at <50 ms, vertical stripes from the spinning disk appear in the acquired image. Time intervals between acquisitions are also effective to ensure optimal developmental capacity. Indeed, during a total of 70 h imaging (1-cell to blastocyst stage), when the interval between imaging was set at 7.5 and 15 min, the developmental capacity was similar to control embryos not subjected to imaging, but when it was reduced to 3.75 min, embryo development was reduced significantly (Yamagata et al. 2009b). Furthermore, excitations in the z-axis direction were not performed while keeping the shutter open – the ‘streaming acquisition’ mode – but the shutter was only opened during each acquisition. This is because continuous excitation by laser was extremely toxic to the embryos, suggesting that a recovery time between successive excitations is important to neutralize phototoxicity.

Practical aspects of imaging

In our experiments, embryos injected with mRNAs for histone H2B-mRFP1 and enhanced green fluorescent protein (EGFP)-α-tubulin were subjected to long-term imaging from the 1-cell to the blastocyst stages (Fig. 3C) and transferred to recipient mothers after imaging. Even though these embryos were exposed to laser light approximately 60 000 times during imaging, they produced healthy pups that grew to adulthood and were reproductively normal (Yamagata et al. 2009b). This proves that our system has remarkably low phototoxicity when using the Yokogawa Nipkow disk confocal unit (Nakano 2002; Toomre & Pawly 2006). The Yokogawa Electric Corporation has commercialized this novel microscopy system and packaged it into one box called the CV1000 (see the manufacturer's website at http://www.yokogawa.com/scanner/products/cv1000e.htm).

Fluorescent live-cell imaging of embryonic stem cell derivation

Pluripotent embryonic stem (ES) and induced pluripotent stem (iPS) cells are now recognized as promising resources for regenerative medicine and therapies (Amabile & Meissner 2009). However, how these pluripotent cells emerge in culture is still largely unknown. This is partly because of the long time required for their derivation and insufficient material being available for in-depth analyses (such as molecular biology experiments), both of which make them extremely difficult for detailed studies. For instance, the derivation of mouse ES cells from preimplantation embryos usually starts from a cell number of as much as <100 and requires 7–10 days before they can be cultured autonomously within the culture dish (Brook & Gardner 1997). There are a few researchers that have performed live-cell imaging of iPS cell derivation, but those studies were done at fairly low magnifications with relatively long time intervals with no z-axis (Chan et al. 2009; Smith et al. 2010). Thus, there is still a strong need for long-term and high-resolution live-cell imaging with low-invasive microscopy techniques in the emerging field of stem cell biology.

Until recently, the only way to analyze ES cell derivation was to immunostain the outgrown colonies with various antibodies at multiple time points (Buehr et al. 2003). However, this procedure requires fixation of the samples, making it almost impossible to know the actual fate of the cells one is observing or interested in. Furthermore, as the ES cell derivation processes vary from embryo to embryo (or clone to clone), in terms of timing and mechanism, it is difficult to track this process with just a few snapshots from different embryos. To overcome these experimental limitations, we applied our low-invasive live-cell imaging technique to monitor this prolonged ES cell derivation process. We can now not only monitor the events that take place during this process, but also determine the outcomes of such events.

By combining transgenic mouse embryos expressing EGFP under the control of Oct3/4 (also known as Pou5f1) genomic locus (GOF18) and the low-invasive live-cell imaging technique described in the previous section, we have succeeded in visualizing the natural ES cell derivation processes for the first time (Yamagata et al. 2010). As shown in Figure 4A, we monitored the derivation process for 7–10 days at 1-h intervals and succeeded in deriving germline competent ES cells from the imaged samples. Through this study, we found that pluripotent Oct3/4-positive epiblast cells undergo extensive Caspase-dependent apoptosis before becoming authentic ES cells, when cultured in conventional ES cell derivation medium. In striking contrast to conventional methods, embryos cultured with chemical inhibitors against MEK and GSK3 in serum-free condition, known as the 2i or 3i methods (Ying et al. 2008; Nichols et al. 2009), showed marked suppression of apoptosis leading to a significant improvement in the ES cell derivation rate. From these retrospective analyses, we propose that suppression of apoptosis is one of the key events necessary for the successful establishment of ES cell lines. In the following sections, we will describe the details of imaging ES cell derivation.

Figure 4.

Experimental scheme for imaging embryonic stem (ES) cell derivation process from the embryos. (A) Schematic diagram of the experimental procedure to visualize the ES cell derivation process. Multiple embryos recovered from Oct3/4 enhanced green fluorescent protein (EGFP) transgenic mice were placed on feeder cells in a glass-bottomed dish and immediately imaged three-dimensionally using an auto x–y stage under the indicated conditions. After the imaging, outgrown colonies were picked and passaged to a new dish. Chimeric mice were produced from these ES cells and germline transmission ability was tested. (B) Stage incubator with glass-bottomed dish for incubation of embryos. Medium was replaced with fresh medium a few times during the course of the experiment. (C) Glass-bottomed dish after 10 days imaging. Embryos were placed at even intervals throughout the glass and some colonies are detectable by eye (yellow arrows). (D) The embryos were incubated in ES cell derivation medium in the presence of propidium iodide to detect cell death. Oct3/4-positive cells and dead cell containing PI are represented by green and red, respectively.


To visualize the ES derivation processes, we used embryos expressing the Oct3/4 EGFP reporter (GOF18) transgene. This reporter expresses EGFP under the control of an Oct3/4 18-kb genomic fragment, consisting of a minimal promoter and proximal and distal enhancers, and is well known to recapitulate the expression profile of endogenous Oct3/4 (Yoshimizu et al. 1999; Ohbo et al. 2003).

By mating DBA/2 and homozygous B6-Oct3/4 EGFP transgenic mice, 8-cell to morula stage embryos were collected from oviducts at 2.5–3.0 days post coitum (dpc). After brief treatment with acidic Tyrode's solution to remove the zona pellucida, embryos were placed on top of feeder cells (mouse embryonic fibroblasts) cultured in ES cell derivation medium. Feeder cells were seeded to the dish the day before embryo collection and these cells were used to ensure the attachment of embryos to the dish. We used 35 mm glass-bottomed dishes coated with gelatin and the embryos were placed at even intervals throughout the glass. In our case, we monitored about 15–20 embryos per imaging session (Fig. 4B,C). We started our imaging from these embryonic stages to determine the expression pattern of the Oct3/4 EGFP transgene during preimplantation development. As expected, EGFP signals were lost in the trophectodermal lineage, but were maintained in the inner cell mass of blastocysts.

Imaging conditions

The imaging device used in this experiment was basically the same as the one described in 'Fluorescence expression'. Briefly, a 20× oil immersion objective lens was used on the confocal microscope and images were taken at 1-h intervals with three channels (bright field, 488 and 561 nm), up to 10 days. For each embryo, 41 images were taken in the z-axis at 4-μm intervals for a total depth of 160 μm. As outgrown colonies moved around during the imaging, stage positions were adjusted approximately every 12 h. Furthermore, as our imaging was performed for a relatively long time, the old medium was replaced with fresh medium a few times during the course of the experiment.

For detecting cell death, 0.5 μg/mL propidium iodide was added to the medium (Fig. 4D). To suppress apoptosis, medium containing pan-Caspase inhibitor Z-VAD-FMK (Peptide Institute, Osaka, Japan) was replaced every day because of the short half-life of this inhibitor.

Chimeric mouse production after imaging

After the imaging, outgrown colonies were picked up one at a time under a stereomicroscope and the clumps were trypsinized and plated onto 96-well plates containing feeder cells. By passaging several times, the ES cells expanded; their normality was tested by karyotyping and immunostaining for pluripotency proteins. ES cell lines, which showed normal karyotypes were then chosen to generate chimeric mice. Approximately 20 cells were introduced into the perivitelline space of 4- or 8-cell stage ICR strain embryos by piezo-assisted microinjection. After culture, the chimeric blastocysts were transferred into the uterus of pseudopregnant ICR strain surrogate mothers at 3 dpc. Finally, the male chimeric mice obtained were crossed with ICR females to confirm the germline transmission of ES cells.

Problems to be solved

A recent study using genetic lineage tracing has proposed that epiblast cells enter a germ cell-like state before becoming ES cells (Chu et al. 2011). As we used a single reporter system to monitor the ES cell derivation process, we could not determine whether the epiblast cells had undergone epigenetic reprogramming or transition to a different cell state prior to their establishment as ES cells. Therefore, it will be necessary for future studies to monitor this derivation process with multiple reporter systems, especially with those that mark primordial germ cells (Ohinata et al. 2008).

Future perspectives

The most notable advantage of our imaging technique is that it only causes minimal damage to cells. Our approach not only enables long-term observation, but also makes it possible to predict the fate of embryonic cells after imaging. For example, we demonstrated previously that the abnormal pattern of methylated DNA in the pronuclei of embryos constructed by round spermatid injection had adverse effects on blastocyst formation by continuously monitoring each embryo after the imaging (Yamazaki et al. 2007). On the other hand, we reported that abnormal chromosome segregation during first mitosis occurred frequently in mouse embryos produced by intracytoplasmic sperm injection (ICSI) and that is the major cause for the early pregnancy loss of ICSI-derived embryos (Yamagata et al. 2009a). Furthermore, the incidence of abnormal chromosome segregation during early cleavage is much higher in SCNT-derived embryos and this appears to be the main cause of the low success rate of animal cloning (Mizutani et al. 2012).

In conclusion, it is becoming increasingly important not only to trace embryonic development and stem cell lineages by live-cell imaging, but also to assess the potential and the capacity of such embryos and cells after imaging. This is particularly important for pluripotent stem cells, as not all ES or iPS cells possess the full capacity for pluripotency (such as germline transmission ability). Through the rapid advancement of sequencing technologies, RNA-Seq transcriptome analysis was performed to track mouse ES cell derivation (Tang et al. 2010). Although this kind of study is extremely informative, we cannot correlate the gene expression profiles with the actual derivation process at present. Accordingly, as in our studies and those of other groups, retrospective studies using live-cell imaging in combination with various reporter systems will be essential to address the key events that discriminate full and (or) partial acquisition of pluripotency during the ES and iPS derivation processes. In this light, the various reporter mice available to date (discussed in the article by Abe & Fujimori in this issue) will also become powerful tools to analyze the fate not only of pluripotent stem cells, but also multiple types of somatic stem cells (Abe et al. 2011). In particular, those markers that label the cell nucleus (e.g. H2B-EGFP) will be essential for lineage tracing, as those that label the whole cell compartment cannot be used for cell counting and cell tracking (Abe et al. 2011). Furthermore, although the live-cell imaging of cell-cycle and/or epigenetic states are still evolving (Yamagata et al. 2007; Sakaue-Sawano et al. 2008; Hayashi-Takanaka et al. 2009, 2011; Abe et al. 2012), these different types of reporters in combination with lineage tracing analysis will become important for future research in the fields of developmental and stem cell biology.


We apologize to those whose primary work could not be cited because of space constraints. We thank Drs Teruhiko Wakayama and Masaru Okabe for their comments and help, and the engineers of microscope companies for their assistance to make the imaging system. Our research was supported in part by a Grant-in-Aid from the Ministry of Education, Culture, Sports, Science and Technology, Japan, and the Japan Society of the Promotion of Science.