DNA injections into Xenopus embryos: fate of injected DNA in relation to formation of embryonic nuclei

Microinjection experiments using cloned gene templates into fertilized eggs of Xenopus laevis provide an interesting experimental system to study factors involved in control of gene expression, as well as possible mechanisms of gene integration and rearrangements of injected DNA templates into the Xenopus genome. In addition, for many types of cloned genes it is possible to compare transcription characteristics obtained from an embryo injection experiment with results from gene-injected oocytes.

In the case of DNA injection experiments into Xenopus oocytes, systematic studies have been carried out on the stability and chromatin configuration of injected DNA following injection into the cytoplasm or into the nucleus (‘germinal vesicle’) of the large Xenopus oocyte. It was found that DNA can be injected into both cellular compartments, but that injected DNA is rapidly degraded after injection into the cytoplasm, whereas DNA injected into the nucleus of the oocyte is not degraded but assembled into chromatin (Wyllie, Laskey, Finch & Gurdon, 1978; Laskey, Gurdon & Trendelenburg, 1979).

By contrast, in a typical embryo injection experiment, the DNA sample is nearly always deposited somewhere in the cytoplasmic region of the animal half of a symmetrized Xenopus egg or an early 2-cell stage. From early investigations onwards, it was known that embryo-injected DNA is not only stable in the cytoplasm but is also replicated (Gurdon, 1974a; for a recent review on chromatin replication during embryogenesis see Laskey, 1985).

Whereas a rapidly increasing number of investigations is concerned with the biochemical and molecular analysis of the conformation and persistence of embryo-injected DNA (see below, and reviews by Gurdon & Melton, 1981; Etkin & DiBerardino, 1983; Etkin, Pearman, Roberts & Bektesh, 1984), relatively little is known of the ultrastructural localization of injected DNA samples. In a recent study, Forbes, Kirschner & Newport (1983) investigated structural aspects of injected DNA using microinjection of bacteriophage λ-DNA into unfertilized Xenopus eggs. It was shown that injected λ-DNA apparently triggers the formation of nuclear envelopes around small droplets of injected DNA. Microscopically such structures can be recognized as ‘pseudonuclei’ (Forbes et al. 1983).

These findings stimulated us to start a combined biochemical and structural analysis concerning the conformation of injected DNA and to characterize the ultrastructural aspects of embryo-injected DNA samples.

In most examples analysed so far, DNA was injected into fertilized eggs in the form of supercoiled circular DNA. The persistence and possible replication of injected circular DNA can thus be conveniently analysed by characterization of amounts of circular DNA at defined stages of early embryogenesis. In systematic studies of the major injection parameters it was first established that only limited amounts of DNA can be injected into embryos (e.g. when compared to amounts injectable into oocyte nuclei) in order to allow development of sufficient numbers of injected embryos (Gurdon & Brown, 1977; Bendig, 1981; Rusconi & Schaffner, 1981). Results from several embryo injection studies using different templates led to the conclusion that circular DNA persists and eventually may be amplified through early embryogenesis, but becomes degraded after the gastrula stage. However, it was also found that a small number of injected gene copies was integrated into the Xenopus genome and it was shown that such integrated copies can persist up to the adult stage (Rusconi & Schaffner, 1981; Andres, Muellener & Ryffel, 1984; Rusconi, 1985). Since most studies used heterologous gene templates for injection, several possibilities for the observed differences in replication and integration could be discussed (see Méchali & Kearsey (1984) for discussion on sequence requirements for DNA replication in Xenopus eggs). It was only recently that embryo-specific genes could be characterized and analysed for their persistence after injection into fertilized eggs. Interestingly it could be shown that a gastrula-specific clone is not amplified during early embryogenesis and persists as extrachromosomal DNA up to the tailbud stage. In this case, no integration of injected gene copies into the embryo genome could be detected (Krieg & Melton, 1985). To test our injection parameters in detail, we injected a chicken ovalbumin gene into fertilized Xenopus eggs and analysed the fate of injected DNA before and after gastrulation. As shown in Fig. 1A,B, presence of extrachromosomal circular DNA could be demonstrated at early cleavage stages. In contrast, DNA extraction from neurula stages showed that extrachromosomal gene copies were no longer present, but a small number of gene copies were found to be integrated into the Xenopus genome. Integrated copies persist up to the swimming tadpole stage. Preliminary results indicate that the integrated copies were complete and not rearranged. In addition, we observed a transient, low level transcription at cleavage stages 8 to 11 of injected embryos in contrast to results from oocyte injection experiments, where no significant amounts of ovalbumin RNA could be detected (Trendelenburg, Mathis & Oudet, 1980; Trendelenburg, 1983).

Studies on the ultrastructural localization of injected DNA require a detailed knowledge of the structural organization of embryonic nuclei and the surrounding cytoplasm during early cleavage stages. Most structural information up to now was obtained from the analysis of thick sections through paraffin-wax-embedded embryos. The published electron microscopic studies on early embryo nuclear structures are on rather specialized topics. Among these studies are (i) a detailed analysis on the dorsally located conspicuous yolk-free cytoplasmic zone which was found to be a characteristic of the symmetrized egg approximately 60 min post-fertilization. It was shown that in the still uncleaved egg a relatively large yolk-free zone exists in the dorsoanimal quadrant of the egg (Herkovits & Ubbels, 1979). The EM data showed that this zone is particularly rich in cytoplasmic vesicles and mitochondria. The area is characterized by the absence of large yolk particles (longitudinal diameter of this zone is 100–150μm). (ii) In another EM study, chromatin organization within embryonic nuclei of blastula and gastrula stages was compared (Csaba & Do, 1974).

To allow analysis of the nucleocytoplasmic architecture at high magnification, we decided to use sections through epon-embedded material. This approach has the advantage that well-fixed embryos can be sectioned at a precise thickness, normally 2–3μm, but thinner sections of large areas can also be obtained, if necessary, for optical as well as electron microscopic analysis. Sections of this kind allow light microscopic analysis using phase or interference contrast microscopy of unstained specimens at high magnification, in particular if a conventional light microscope is equipped with a system for video-enhanced contrast (see below). An overview of the typical nuclear structures seen in sections of early embryos is shown in Fig. 2. In most cases, typical interphase nuclei are seen and screening for the presence of interphase nuclei can be done rapidly at low magnification. If no interphase nuclei are seen at low magnification typical aspects of metaphase chromosomes can be recognized using phase contrast at high magnification (Fig. 2B,C; for direct comparison of dimensions, Fig. 2A-C is enlarged to the same scale). In addition, it was found that all nuclear structures (interphase nuclei as well as metaphase plates) are typically located in the central area of a large, predominantly yolk-free cytoplasmic zone. Such cytoplasmic regions are very large during the earliest cleavage cycles and become progressively smaller as cleavage proceeds. A comparison of such areas is shown in Fig. 3A (cleavage cycle 2) and Fig. 3B (cl. 7), again both micrographs are enlarged to the same scale. Comparison of Fig. 3A and B indicates in addition that, whereas the size of the cytoplasmic area decreases drastically, the mean diameter of embryonic nuclei does not change significantly during these early cleavages. This observation is substantiated by the dimensions and structural aspects of interphase nuclei of cleavages 2, 4 and 6, as shown in Fig. 4A-C.

As shown in Fig. 4A-C, nuclei are characterized by large diameters, e.g. up to 30μm, and the presence of intranuclear membrane folds. These observations are in line with earlier observations from in situ fixed embryonic nuclei (Hay & Gurdon, 1967; Bonnano, cited in Gerhart, 1980; and measurements on isolated embryonic nuclei, Farzaneh & Pearson, 1978). Details of the size distributions of mean diameters for interphase nuclei as well as for the corresponding yolk-free cytoplasmic zones where the nuclei were found to be located are shown in the diagram of Fig. 5. Whereas a drastic size reduction of mean diameters for cytoplasmic zones was observed from cleavage 2 to cleavage 10, a reduction of mean nuclear diameters was observed during cleavage cycles 11 to 14. Nuclear diameters are then around 10 μm and remain almost constant during the successive later stages of development (e.g. at stages 14 and 20 according to the staging system of Nieuwkoop & Faber, 1967; see Fig. 5). For comparison, first appearances of nuclear lamins L_Iand L_II are indicated in Fig. 5 (for details see Stick & Hausen, 1985; for localization studies on embryonic nuclei see Benavente, Krohne & Franke, 1985). The way in which this complex nuclear architecture is achieved is not yet established. Work is in progress to apply video-enhanced contrast light microscopy (for details see Weiss, 1986) to thick sections of embryonic nuclei. Using this methodology, different focus levels can be analysed for the presence and spatial orientation of intranuclear membranes (Fig. 6A,B), information which is difficult to obtain from thin-sectioned nuclei using electron microscopy (Fig. 6C).

In conclusion, interesting changes in nucleocytoplasmic compartmentalization can be observed from thick-sectioned material. It is clear, however, that additional studies are required for a more close correlation of the observed structural changes with the cell cycle data (Newport & Kirschner, 1982; for recent review see Satoh, 1985).

For structural studies on embryo-injected DNA we chose injection experiments of bacteriophage λ-DNA for the following reasons.

(i) In a detailed investigation Forbes et al. 1983 showed that λ-DNA is a good template for induction of membrane structures around injected DNA, if the DNA was injected into unfertilized eggs. It was observed that, 1·5 to 2h following injection into the unfertilized egg, A-DNA is predominantly contained in nuclear structures approximately 6μm in diameter (range 1–20 μm).

(ii) Using electron microscopy it was shown that the DNA samples were surrounded by a typical nuclear membrane structure (Forbes et al. 1983).

For our structural investigations we injected 2–5 ng of A-DNA into fertilized eggs. Special care was taken to perform the experiment exclusively on batches of fertilized eggs that exhibited a particularly high synchrony of cleavages. In agreement with results from earlier investigations using injection of relatively high amounts of DNA (Gurdon & Brown, 1977; Rusconi & Schaffner, 1981) no significant percentage of embryos was noted that were arrested during early cleavage. In fact, in good batches of fertilized injected eggs, as many as 60–80 % of injected embryos reached cleavage cycle 10 as compared to controls. Embryos were fixed during early cleavage cycles, and thick sections analysed for the presence of small ‘pseudonuclei’ in the large, yolk-free cytoplasmic zones of early embryonic cells (see preceding section for details). Under the light microscopic screening conditions used (phase contrast microscopy of 3 μm sections) nuclear structures with diameters of 3/zm and more should be visible in the yolk-free cytoplasmic zones. Yolk-free cytoplasmic zones were analysed using serial sections from 30 embryos. Morphological evidence for the presence of bacteriophage DNA-derived nuclear structures within these cytoplasmic areas was obtained for 30% of injected embryos. In most cases ‘pseudonuclei’ were observed to occur as clusters of small nuclear structures, approximately 4–8 μm in diameter, which were found to be typically located at the periphery of individual yolk-free zones (Fig. 7A, cleavage 5 embryo). From their ultrastructure (Fig. 7C) these structures resemble very closely the nuclear structures seen in bacteriophage DNA-injected unfertilized eggs (cf. figs 2, 3, 5 in Forbes et al. 1983). An interesting result of our investigation was the observation that, in all cytoplasmic zones where we observed typical ‘pseudonuclei’, no coexistence of ‘pseudonuclei’ next to a typical large interphase nucleus was seen. In this regard, two types of results were obtained.

(i) Embryos fixed at cleavage cycles 2, 3 and 4 following DNA injection showed the presence of up to 10 closely associated ‘pseudonuclei’ in up to two yolk-free cytoplasmic zones per individual embryo. In most cases observed, ‘pseudonuclei’ were arranged as a straight line at the periphery of the cytoplasmic zone. Where no ‘pseudonuclei’ were found in this class of embryos, the cytoplasmic zones showed the presence of highly reticulated embryonic nuclei.

(ii) The second group of embryos was characterized as follows. Apparently, these embryos were fixed at metaphase, since in almost all of their cytoplasmic zones typical metaphase plates could be observed (cleavages 3/4, 4/5, 6/7; for typical aspect of metaphase plates in noninjected embryos see Fig. 2B,C). In the cytoplasmic zones of these embryos, where ‘pseudonuclei’ were detected, metaphase plates of embryonic chromosomes were also seen (Fig. 7B). Work is in progress to use this approach in combination with autoradiography (Gurdon & Brown, 1977), DNA-staining techniques (Forbes et al. 1983) and scanning and transmission electron microscopy (Tröster et al. 1985) to analyse the interaction of injected λ-DNA with the embryonic nuclei during interphase and metaphase in more detail.

This study would not have been possible without the excellent assistance of Andrea Laier, Roger Fischer and Gisela Weise. Karsten Richter and Hans Jürgen Boxberger are thanked for their help in preparing and analysis of thick sections. We also thank E. del Pino, A. Hofmann, H. Steinbeisser, H. Trôster and A. Wild for helpful discussions. The study was financially supported by grants of the DFG (given to M.F.T.) and INSERM (given to P.O.).