Cell lineage labeling has turned out to be one of the most important techniques in developmental biology. During the past century, new methods of lineage labeling have been established prompted by the scientific issues and restrained by technical limitations at the respective time (e.g., application of vital dyes [Vogt, 1925; Keller, 1975], transplantation of labeled tissues [Spemann, 1931; Sadaghiani and Thiebaud, 1987], fluorescent dye injection [Bauer et al., 1994; Gross and Hanken, 2004], quantum dots [Rieger et al., 2005]). These methods are limited to accessible tissues at the surface of the embryo. Different approaches must be chosen for the staining of internal cells. One approach is the labeling of the tissue of interest at a developmental stage, when all embryonic cells can still be reached from the surface. Alternatively, labeling must be combined with microsurgery.
The injection of mRNAs encoding green fluorescent protein (GFP) or its derivatives has become an intriguing alternative to the labeling techniques mentioned above. The range of colors of fluorescent proteins has been greatly extended by the discovery of GFP-like proteins in various anthozoan species (for review, see Wiedenmann, 1997; Shaner et al., 2005). Photoactivatable fluorescent proteins like KFP1 (Chudakov et al., 2003) and green to red photoconvertible proteins like EosFP (Wiedenmann et al., 2004), Kaede (Ando et al., 2002), or Dendra (Gurskaya et al., 2006) have been recently described. The possibility to use these photoconvertible proteins for minimally invasive regional optical labeling down to subcellular levels offers tremendous potential for live cell imaging (Wiedenmann et al., 2004; Nienhaus et al., 2006).
Here, we prove the application of green to red photoconvertible proteins as a powerful tool to monitor early vertebrate development using EosFP and Xenopus laevis as a model organism. We demonstrate that regional optical labeling in different germ layers and at different developmental stages strongly facilitate the analysis of morphogenetic movements and formation of embryonic organs. Moreover, the method allows experiments that cannot be performed easily by conventional approaches as, for instance, specific labeling of cells from nonsurface layers.
Fluorescence, Photoconversion, and Histology
Capped EosFP-mRNA was injected into the vegetal pole of stage 3 Xenopus laevis embryos, resulting in labeling of all cells. A time-lapse movie was recorded beginning about 15 min after injection to analyze the appearance of green fluorescence. As shown for a typical embryo in Figure 1A, green fluorescence was detectable approximately 4 hr after injection (stage 8). A level at which photoconversion produced a clearly visible optical contrast was reached approximately 2.5 hr later (stage 9). Using the exposure time to measure levels of fluorescence, we found a 10-fold higher intensity than the cellular autofluorescence level at this stage (not shown). Green fluorescence did not decrease over the following 7 days (stage 48, compare Figs. 3 and 4). We conclude that the fluorescence levels are sufficient for conversion during the whole early development, except for ∼6.5 hr after injection of EosFP-mRNA.
Depending on the experiment, variable areas must be converted. The extension of the label can be easily controlled by the magnification of the microscopic lens and by positioning the iris diaphragm within the excitation light path as indicated for different microscopic lenses in Figure 1B. The smallest area (75 μm, conversion with the ×40 lens) corresponds to a diameter of four to five cells at the shown stage 22. Red fluorescence becomes visible immediately (i.e., during photoconversion). Under our experimental conditions, complete photoconversion of EosFP, could be achieved within 60 sec of illumination (not shown).
To illustrate morphogenetic movements, we used the photoconversion of EosFP at early gastrulation with the ×10 microscopic lens. After conversion of the Spemann organizer region at the beginning of gastrulation, a time-lapse movie was recorded (Supplementary Material S1, which can be viewed at http://www.interscience.wiley.com/jpages/1058-8388/suppmat). Single frames of this movie are shown in Figure 1C, illustrating mesodermal involution and convergent extension during gastrulation. This experiment demonstrates that the red fluorescence of the labeled mesodermal cells can be clearly detected even after internalization.
To demonstrate the utility of EosFP labeling in histology, EosFP mRNA was injected into the precursors of mesoderm and endoderm of stage 4 embryos. The ectoderm remained uninjected. At the beginning of gastrulation (stage 10.25) several mesodermal cells were photoconverted using the ×10 microscopic lens. Embryos were fixed at stage 12, when almost all mesodermal and endodermal cells were internalized. After embedding and sectioning embryos, the photolabeled cells can be detected in the internalized green fluorescent tissue by their red fluorescence. Ectodermal cells remained unlabeled, as expected (Fig. 1D). These results demonstrate the stability of both green and red fluorescence of EosFP during fixation, embedding, and sectioning.
Photoconversion of Nonsurface Tissue Layers
Embryos were injected in the four vegetal cells of stage 4, that is, the precursors of endoderm and mesoderm. Ectodermal precursors remained uninjected and, therefore, nonfluorescent. During gastrulation, the EosFP expressing mesodermal and endodermal cells involute and are covered by unlabeled ectodermal cells. At stage 12.5 (late gastrula), fluorescence of these internalized cells can be observed across the overlaying nonfluorescent ectoderm (Fig. 2A). At this stage, we converted cells from a dorsolateral region using the ×10 microscopic lens (Fig. 2A). As expected from labeling of internalized cells, at later stages the converted cells were found exclusively in tissues derived from mesoderm (i.e., the somites, Fig. 2B).
The same photoconversion was performed using embryos expressing green fluorescence in all germ layers (Fig. 2C). Red fluorescence was later found in mesodermal cells, indicating the successful labeling of deeper cell layers, and also in epidermal cells, which originate from the ectodermal cells of the surface layer (Fig. 2D). We conclude that the light required for photoconversion of EosFP efficiently penetrates unlabeled and labeled layers, thereby enabling the marking of deeper cell layers.
Photoconversion of Different Mesodermal Tissues
Embryos were injected with EosFP mRNA in mesodermal and endodermal precursors as described above. Prospective mesodermal tissue of these embryos was converted at different gastrula stages using the ×10 microscopic lens and then analyzed at tadpole stages (35 to 48).
Cells of the Spemann organizer region were converted at the beginning of gastrulation. As expected, red fluorescence was found in the head and axial mesoderm of tadpoles (Fig. 3A). Embryos, which had been labeled in lateral regions at the end of gastrulation (stage 13), yielded tadpoles with red fluorescence in different elements of the pronephros (Fig. 3B,C; for details, see Brändli, 1999). Conversion of the heart precursors laterally to the Spemann organizer at early gastrula (stage 10) resulted in red fluorescence of heart tissues (Fig. 3D,E; for details, see Kolker et al., 2000). Labeling in a similar region approximately 1 hr later (stage 10.25) resulted in stained gills (Fig. 3F). In these and other experiments, strong red autofluorescence of the gall bladder was observed (Fig. 3D–F). Conversion of cells from the posterior portion of the Spemann organizer at early gastrulation resulted in labeling of somites (Fig. 3G) and posterior notochord. The latter is difficult to observe, because it is covered by somites. However, in the tail region, the notochord could be clearly distinguished from somites because of the different cell shape (Fig. 3H).
Most interestingly, the photoconversion of EosFP enables also the visualization of the circulatory system. For this purpose, one gill of a stage 45 embryo with a developed circulatory system was exposed for 2 min. Green fluorescent blood cells changed their emission to red during their passage through the exposed gill. Using exposure times of 15 sec for red and 0.3 sec for green fluorescence, photographs were taken far away from the place of conversion. The converted blood cells traced the blood vessels as shown for the tail (Fig. 3I; for details, see Levine et al., 2003).
These experiments clearly demonstrate that light-mediated labeling of mesodermal cells during and after gastrulation facilitates the monitoring of development and differentiation of a broad variety of mesodermal organs.
Photoconversion of the Neural Crest
To prove the fitness of this method for labeling of ectodermal cells, we chose the neural crest. EosFP mRNA was injected into all cells at stage 3. Conversion of small areas of the neural fold was performed at stage 15 using the ×10 microscopic lens as shown for one example (Fig. 4A). Positions for conversion along the neural fold were chosen according to known fate maps (Sadaghiani and Thiebaud, 1987). Embryos were photographed around stage 29 to identify migrating neural crest cells. Depending on the position of conversion different regions of neural and neural crest tissue were labeled, including forebrain (Fig. 4B), midbrain and mandibular crest (Fig. 4C), midbrain and mandibular crest/hyoid crest (Fig. 4D), hindbrain and branchial crests (Fig. 4E), and hindbrain/spinal chord and the most posterior portion of the branchial crests (Fig. 4F).
We further identified structures formed from these neural crest cells at stage 48. In accordance with the neural crest fate map (Sadaghiani and Thiebaud, 1987) different cartilaginous elements of the craniofacial and visceral skeleton show red fluorescence. Originating from a converted mandibular crest, Meckel's cartilage and the palatoquadrate are labeled (Fig. 4G). Photoconversion of the hyoid crest resulted in labeled ceratohyale (Fig. 4H). Photoconversion of the branchial crest resulted in labeling of the cartilage in the gills (Fig. 4I).
These experiments demonstrate that EosFP can be used to analyze the morphogenetic movements and the lineage of ectodermal cells. The red fluorescence can be followed for at least 7 days (i.e., stage 15–48).
Protein Injection and Photoconversion
To enable lineage labeling and conversion at early cleavage stages, that is, before sufficient fluorescence of EosFP after mRNA injection appeared, we decided to directly inject the recombinant EosFP: 3 ng of recombinant EosFP diluted in 10 nl Gurdon′s buffer were injected into each of the two dorsal blastomeres at stage 3 (Fig. 5A). These concentrations were well tolerated by the developing embryos. Ten-fold higher doses proved to be lethal. Regional photoconversion is possible right from the moment of injection. Here, we show photoconversion of a single blastomere at stage 6 (Fig. 5B). The descendants were found in the dorsal blastopore lip during gastrulation (Fig. 5C). At tadpole stages, labeling was observed, among others, in the posterior somites (Fig. 5D).
In the present study, we have introduced the application of green to red photoconvertible fluorescent proteins as a tool for the detailed analysis of early vertebrate development. The method requires only the equipment for microinjection and a fluorescence microscope with an appropriate camera system. As easy to perform as classic cell lineage labeling techniques, it provides several advantages. In contrast to rather broad labels after blastomere injection, local photoconversion allows the reduction of the labeled area to the single cell level. Subcellular resolution can be achieved by the use of focused laser beams (Wiedenmann et al., 2004; Nienhaus et al., 2006). Further spatial resolution of the optical labeling might be achieved by the combination of two-photon photoconversion and confocal imaging (Ivanchenko et al., 2005; Tsutsui et al., 2005).
In contrast to photoactivatable fluorescent proteins like KFP1, paGFP, or Dronpa (Patterson and Lippincott-Schwartz, 2002; Chudakov et al., 2003; Ando et al., 2004), photoconvertible fluorescent proteins like EosFP offer the convenience that the target cells or tissues are already visible due to the initial green fluorescence. Moreover, due to the large distance between the wavelengths efficient for photoconversion and the ones for detection of the label, with EosFP, the risk of undesired activation or deactivation is low in comparison to KFP1, Dendra, or Dronpa (Chudakov et al., 2003; Ando et al., 2004; Gurskaya et al., 2006).
With EosFP information is obtained from both the injected label and the converted label and can be easily combined afterward. Standard surface labeling techniques or transplantations allow labels of comparable size and also combinations of different labels. However, they remain restricted to the surface of an embryo or need time-consuming microsurgery. Problems such as wound healing and contamination with unwanted cells need to be taken in account. In contrast, photoconversion of EosFP or other photoconvertible fluorescent proteins allow minimally invasive labeling of internal tissues across surface layers. No negative response of the embryos to the photoconversion by light around 420 nm was observed in our experiments (Supplementary Material S2).
One limitation of the injection of mRNA for photoconvertible fluorescent proteins is that photoconversion cannot be performed earlier than 6.5 hr after injection. However, this “blind spot” at early cleavage stages can be easily closed by the injection of the recombinant EosFP, which can be converted almost immediately after injection.
The application of green to red photoconvertible fluorescent proteins in early development allows lineage labeling with high resolution, as it is shown here for EosFP and Xenopus laevis. The protein can be converted in almost any temporal and spatial dimension during early embryonic development. The red fluorescence can be traced for more than 1 week.
Because the detailed analysis of phenotypes after manipulation of gene expression, signal transduction, or morphogenetic events is an emerging field in developmental biology, labeling with photoconvertible fluorescent proteins promises to be a very powerful tool for developmental studies. The method appears also to be useful in studies of other model organisms as invertebrates, zebrafish, or mice. Transgenic animals will allow experiments even at later periods of their development as shown recently for zebrafish (Sato et al., 2006).
Plasmid and mRNA
The cDNA for d2EosFP was isolated after XhoI and XbaI digestion of pcDNA3-Flag1-d2EosFP (Wiedenmann et al., 2004) and inserted into pCS2+MT resulting in pCS2+MT-d2EosFP. Before in vitro transcription pCS2+MT-d2EosFP was linearized with NotI. Capped mRNA of d2EosFP was transcribed in vitro using the mMESSAGEmMACHINE kit (Ambion) and purified on RNeasy columns (Qiagen). Aliquots of a concentration of 20 pg/nl were stored at −20°C.
Handling and Treating of Embryos
Embryos were staged according to the normal table (Nieuwkoop and Faber, 1956). In general, embryos were kept at 21°C. If experimental settings required (e.g., staging for injection) embryos were kept at 14°C. In vitro fertilization, microinjection, and culture of embryos were carried out as previously described (Winklbauer, 1990; Wacker et al., 2000). All experiments were performed using pigmented embryos. For labeling, embryos were injected with 1 to 4 times 10 nl of EosFP mRNA (20 pg/nl) at stage 2, stage 3, or stage 4. Doses higher than 1 ng per embryo resulted in malformations (i.e., arrest of gastrulation, deformed axial structures) and were, therefore, not used. After injection, embryos were kept in the dark to prevent unspecific conversion.
Purification of Recombinant EosFP
The cDNA of EosFP was cloned in pQE32 (Qiagen), which introduces a 6 times His-tag to the N-terminus of the protein. Escherichia coli (M15 pREP4) were transformed and grown at 37°C to an OD600 of 0.6. Upon induction of protein expression with isopropyl-β-D-thiogalactopyranoside (IPTG), the culture was swirled overnight at 22°C. The harvested cells were disrupted by sonication, and the cleared lysate was subjected to metal affinity chromatography using Talon Matrix (Clontech). The eluate was concentrated and used for size exclusion chromatography. Peak fractions were collected, sterile filtered, and concentrated. The phosphate buffered saline used for chromatography was replaced with Gurdon's injection buffer (15 mM Tris pH7.5, 88 mM NaCl, 1 mM KCl) by repeated concentration steps. The protein concentration was adjusted to 30 mg/ml. Stocks were stored at 4°C. Three ng of EosFP diluted in 10 nl Gurdon's buffer were used in up to four injections per embryo.
Photoconversion and Imaging
Photoconversion and imaging was performed using an inverse fluorescence microscope (DM IRB, Leica) equipped with a 100 W Hg light source (HBO 103W/2, Osram). The action spectrum of photoconversion of EosFP peaks around 395 nm (Wiedenmann et al., 2004). In the present experiments, photoconversion was achieved by irradiating the cells with 405–425 nm light using a BP430/50 excitation filter and a 425 DCLP beam splitter. The size of the converted region was defined by the magnification of the microscopic lens and by positioning the iris diaphragm within the excitation light path. Conversion time was generally 60 sec, unless otherwise noted. During photoconversion, the embryos were kept in the desired position by acrylglass barriers (1 mm × 2 mm × 10 mm), which were fixed in the Petri dishes with silicone grease. Green and red fluorescence was imaged using the following filter sets: Green, HQ480/40 (ex), Q505LP (Beamsplitter), HQ527/30 (em); Red, HQ545/30 (ex), Q565LP (Beamsplitter), 610/75 (em). Images were taken with a digital camera (C4742, Hamamatsu) using the Openlab software (Improvision). During imaging, tadpoles were anesthetized using 500 μg/ml MS 222 (Sigma). Depending on the fluorescence levels exposure times varied from 50 to 1,000 msec for early developmental stages and up to 6,000 msec for late developmental stages. One photograph of green fluorescence and one of red fluorescence were, respectively, taken and then overlaid using Adobe Photoshop. For Figure 3A and B, several photographs were joined together to show the whole embryo.
For the time-lapse movie frames were grabbed every 4 min, beginning at stage 10.5 and ending at stage 14. The exposure time was 0.8 sec. The filter combination for red fluorescence was used. The movie was generated using the ImageJ software (http://rsb.info.nih.gov/ij/).
Embryos were fixed with MEMPFA (0.1 M MOPS pH 7.4; 2 mM EGTA; 1 mM MgSO4, 4% paraformaldehyde) at room temperature for at least 4 hr or preferentially at 6°C overnight. For Vibratome sectioning, embryos were embedded in gelatin/albumin or agarose (2%) following standard protocols. Gelatin/albumin shows red autofluorescence, which was not observed for agarose. Forty-micrometer sections were placed on slides in Aquatex (Merck) and covered with a coverslip.
We thank R. Winklbauer, H. Jansen, and M. Schuff for helpful comments on the manuscript. We also thank S. Schirmer and A. Röβner for excellent technical assistance. W.K. and F.O. were funded by grants of the Deutsche Forschungsgemeinschaft and J.W. was funded by the State of Baden Württemberg.