Editor’s note (10/9/2012): We are making the text of this article freely available for 30 days because the author, Sir John B. Gurdon, is one of the winners of the 2012 Nobel Prize in Physiology or Medicine. The full article with images, which was published in the December 1968 issue, is available for institutional users only at this time (pdf).
The means by which cells first come to differ from one another during animal development has interested humans for nearly 2,000 years, and it still constitutes one of the major unsolved problems of biology. Much of the experimental work designed to investigate the problem has been done with amphibians such as frogs and salamanders because their eggs and embryos are comparatively large and are remarkably resistant to microsurgery. As with most animal eggs, the early events of amphibian development are largely independent of the environment, and the processes leading to cell differentiation must involve a redistribution and interaction of constituents already present in the fertilized egg.
Several different kinds of experiment have revealed the dependence of cell differentiation on the activity of the genes in the cell's nucleus. This is clearly shown by the nonsurvival of hybrid embryos produced by fertilizing the egg of one species (after removal of the egg's nucleus) with the sperm of another species. Such hybrids typically die before they reach the gastrula stage, the point in embryonic development at which major cell differences first become obvious. Yet the hybrids differ from nonhybrid embryos only by the substitution of some of the nuclear genes. If gene activity were not required for gastrulation and further development, the hybrids should survive as well as nonhybrids. The importance of the egg's non-nuclear material—the cytoplasm—in early development is apparent in the consistent relation that is seen to exist between certain regions in the cytoplasm of a fertilized egg and certain kinds or directions of cell differentiation. It is also evident in the effect of egg cytoplasm on the behavior of chromosomes [see "How Cells Specialize," by Michail Fischberg and Antonie \V. Blackler; SCIENTIFIC AMERICAN, September, 1961]. Such facts have justified the belief that the early events in cell differentiation depend on an interaction between the nucleus and the cytoplasm.
Nuclear transplantation is a technique that has enormously facilitated the analysis of these interactions between nucleus and cytoplasm. It allows the nucleus from one of several different cell types to be combined with egg cytoplasm in such a way that normal embryonic development can take place. Until this technique was developed the only kind of nucleus that could be made to penetrate an egg was the nucleus of a sperm cell, and this was obviously of limited use for an analysis of those interactions between nucleus and cytoplasm that lead to the majority of cell differences in an individual.
The technique was first applied to the question primarily responsible for its development. The question is whether or not the progressive specialization of cells during development is accompanied by the loss of genes no longer required in each cell type. For example, does an intestine-cell nucleus retain the genes needed for the synthesis of hemoglobin, the protein characteristic of red blood cells, and a nerve-cell nucleus the genes needed for making myosin, a protein characteristic of muscle cells? If unwanted genes are lost, the possibility exists that it is the progressive loss of different genes that itself determines the specialization of cells, as August Weismann originally proposed in 1892. The clearest alternative is that all genes are retained in all cells and that the genes are inactive in those cells in which they are not required. Before describing the nuclear-transplant experiments that distinguish between these two possibilities, we must outline the methods used to transplant living cell nuclei into eggs.
The aim of a nuclear-transplant experiment is to insert the nucleus of a specialized cell into an unfertilized egg whose nucleus has been removed. Ingenious attempts in this direction were made many years ago by constricting an egg just after fertilization and then letting one of the early-division nuclei that appeared in the nucleated half of the egg enter the non-nucleated half. This method, however, is applicable only to the nuclei of early embryos whose cells are not normally regarded as being specialized. The first real success in transplanting living cell nuclei into animal eggs was achieved in 1952 by Robert W. Briggs and Thomas J. King, both of whom were working at the Institute for Cancer Research in Philadelphia. Their method, which has been generally adopted in subsequent work, involves three steps. Owing to the fortunate circumstance that the unfertilized egg of an amphibian has its nucleus (in the form of chromosomes) located just under the surface of the egg at a point visible through the microscope, it is not difficult to obtain an egg with no nucleus. This can be done by removing the region of the egg that contains chromosomes with a needle or by killing the nuclear material with ultraviolet radiation. The second step is to dissociate a tissue into separate cells, each of which can be used to provide a donor nucleus for transplantation. The cells separate from one another in a medium lacking calcium and magnesium ions, which are removed from the embryo more quickly by adding to the medium a chelating substance such as Versene.
The third and most difficult stage in the procedure involves the insertion of the donor-cell nucleus into the enucleated egg. Briggs and King found that this can be done by sucking an isolated cell into a micropipette that is small enough to break the cell wall but large enough to leave the nucleus still surrounded by cytoplasm. This compromise is required because the nucleus in an unbroken cell does not make the necessary response to egg cytoplasm, and conversely a bare nucleus without surrounding cytoplasm is readily damaged by exposure to any artificial medium. The broken cell with its cytoplasm-protected nucleus is injected into the recipient egg. The amount of donor-cell cytoplasm injected is very small and does not have any effect.
A useful extension of the basic nuc1ear-transplant technique is called serial nuclear transplantation. It involves the same procedure as the one just described except that instead of the donor nuclei being taken from the cells of an embryo or larva reared from a fertilized egg, they are taken from a young embryo that is itself the result of a nuclear transplant experiment. The effect is the same as in the vegetative propagation of plants, namely the production of a clone: a population consisting of many individuals all having an identical set of genes in their nuclei.
One other feature of nuclear-transplant experiments is of the greatest importance for their interpretation. It is the use of a nuclear marker whereby the division products of a transplanted nucleus can be distinguished from those of the host egg nucleus. A nuclear marker is virtually indispensable where attention is to be paid to the development of a very small percentage of eggs that have received transplanted nuclei, since one cannot otherwise be sure that an occasional error in enucleation by hand or by ultraviolet irradiation has not occurred. Only by the presence of a marker in the nuclei of a transplant embryo does one have proof of its origin.
A nuclear marker must be replicated and therefore be genetic. One of the most useful for nuclear transplantation is found in a mutant line of the South African clawed frog Xenopus laevis, discovered at the University of Oxford by Michail Fischberg. The nuclei of most normal frog cells contain two of the bodies called nucleoli; the nuclei of cells carrying the mutation never have more than one. This mutation is almost ideal as a nuclear marker because a sample of cells taken from any tissue at any develop mental stage beyond the blastula (the hollow sphere from which the gastrula arises) can be readily classified as being mutant or not.
We can now return to the question of whether or not genes are lost in the course of normal cell differentiation. Nuclear-transfer experiments are performed to answer this question on the assumption that if the combination of egg cytoplasm with a transplanted nucleus can develop into a normal embryo possessing all cell types, then the transplanted nucleus cannot have lost the genes essential for pathways of cell differentiation other than its own. For example, if a normal embryo containing a specialized cell type such as blood cells can be obtained by transplanting an intestine-cell nucleus into an enucleated egg, then the genes responsible for the synthesis of hemoglobin cannot have been lost from the intestine-cell nucleus in the course of cell differentiation. The only assumption here is that a gene, once lost, cannot be regained in the course of a few cell generations. It happens that the best evidence for the retention of genes in fully differentiated cells comes from two series of experiments carried out at Oxford on eggs of the frog Xenopus.
The fully differentiated cells used for these experiments were taken from the epithelial layer of the intestine of mutant tadpoles that had begun to feed. Intestine epithelium cells have a "brush border," a structure that is present only in cells specialized for absorption and that is assumed to have arisen as a result of the activity of certain intestine-cell genes. Not all the cells of the intestine are epithelial, but when the epithelial cells are dissociated, they can be distinguished from the other cell types by their large content of yolk, by the ease with which they dissociate in a medium that contains Versene and sometimes by their retention of the brush border.
The first experiments with intestinecell nuclei were designed to show that at least some of these nuclei possess all the genes necessary for the differentiation of all cell types, and therefore that some of the transplant embryos derived from intestine nuclei could be reared into normal adult frogs. Both male and female adult frogs, fertile and normal in every respect, have in fact been obtained from transplanted intestine nuclei. Although only about 1.5 percent of the eggs with transplanted intestine nuclei developed into adult frogs, all of these frogs carried the mutant nuclear marker in their cells; their existence therefore proves that at least some intestine cells possess as many different kinds of nuclear genes as are present in a fertilized egg.
Subsequent experiments with intestine nuclei were designed to show that many of these nuclei have retained genes required for the differentiation of at least some quite different cell types. In these experiments the criterion for gene retention was the differentiation of functional muscle and nerve cells by nuclei whose mitotic ancestors had already promoted the differentiation of intestine cells. Functional muscle and nerve cells are present in any nuclear-transplant embryo that shows the small twitching movements, or muscular responses, characteristic of developing tadpoles just before they swim. Out of several hundred intestine nuclear transfers, about 2.5 percent of the injected eggs developed as far as the muscular-response stage or further. The reason why the remainder did not reach this stage is not necessarily because that proportion of intestine nuclei lack the necessary genes. In some cases it is known to be the inability of certain recipient eggs to withstand injection; in others it is the incomplete replication of some of the transplanted nuclei or their daughter nuclei during cleavage. In either case a nuclear-transplant embryo should contain some cells with normal nuclei as well as some abnormal cells responsible for the early death of the embryo.
Serial nuclear transplantation offered a means of overcoming both difficulties. A sample of nuclear-transplant embryos whose development was so abnormal they would have died before reaching the muscular-response stage provided nuclei for serial-transplant clones. In 70 percent of the serial-transplant clones some of the embryos developed as far as the muscular-response stage or beyond it. By adding the proportion of nuclei shown by first transplants to be able to support muscular-response differentiation to the proportion shown by serial transplantation to possess this capacity, we can conclude that at least 20 percent of the intestine epithelium cells must have retained the genes necessary for muscle-cell and nerve-cell differentiation.
There is no reason to believe that muscle-cell or nerve-cell genes have been lost or permanently inactivated in the remaining 80 percent of transplanted intestine nuclei. There are many reasons why it might not have been possible to demonstrate their presence. For example, about 50 percent of all the eggs that received intestine nuclei failed to divide even once. When a sample of these eggs was sectioned, they were found to contain either no nucleus at all or else a nucleus that was still inside an intact intestine cell. In the first instance the nucleus presumably stuck to the injection pipette and was never deposited in the egg; in the second, the donor cell was never broken so as to liberate its nucleus, a technical error that is easy to make with very small cells. In both cases the developmental capacity of the intestine nuclei was not tested, and the recipient eggs that failed to divide should not be counted in the results.
It is clear from these experiments that the loss or permanent inactivation of genes does not necessarily accompany the normal differentiation of animal cells. This conclusion is not inconsistent with a recent finding: the "amplification" of the genes responsible for synthesizing the ribonucleic acid (RNA) of the sub cellular particles called ribosomes. This phenomenon was demonstrated in amphibian oocytes, the cells that give rise to mature eggs. The nuclear-transfer experiments just described do not exclude such amplification, which simply alters the number of copies of one kind of gene in a nucleus. Instead they show that specialized cells always have at least one copy of every different gene.
The inability of some transplanted nuclei to support normal development has attracted considerable interest be cause it is always found that the proportion of nuclei showing a restricted developmental capacity increases as the cells from which they are taken become differentiated. Furthermore, serial nuclear-transplant experiments conducted by Briggs and King (and subsequently by others) have shown that all the embryos in a clone derived from one original nuclear transplant often suffer from the same abnormality, whereas the embryos in a clone derived from another original transplant may suffer from a different abnormality. Some of the abnormalities of nuclear-transplant embryos can therefore be attributed to nuclear changes that can be inherited.
The discovery that these changes arise as a result of nuclear transplantation, and not in the course of normal cell differentiation, was an important one. This was first established by Marie A. DiBerardino of the Institute for Cancer Research, who made a detailed analysis of the number and shape of chromosomes in nuclear-transplant embryos. Abnormal embryos were usually found to suffer from chromosome abnormalities that were not present in the donor embryos, a finding that at once explains why the factors causing many of the developmental abnormalities of nuclear-transplant embryos are inherited. The fact that chromosome abnormalities arise after nuclear transplantation does not necessarily mean that they are of no interest; there could be a connection between the kind of chromosome abnormality encountered and the cell type of the donor nucleus concerned. In spite of an intensive search, however, no such relationship has yet been found.
The origin of these chromosome abnormalities is probably to be understood as an incompatibility between the very slow rate of division of differentiating cells—only one division every two days or more—and the rapid rate of division in an egg, which starts to divide (and causes any injected nucleus to try to divide) about an hour after injection. Unless an injected nucleus can complete the replication of its chromosomes within this brief period, they will be torn apart and broken at division. This concept is supported by the observation, made at Oxford in collaboration with my colleagues C. F. Graham and K. Arms, that many transplanted nuclei continue to synthesize the genetic material DNA right up to the time of the first nuclear division, whereas sperm and egg nuclei always complete this synthesis well before division. Presumably molecules associated with the DNA of specialized cells prevent the chromosomes of such cells from undergoing replication as rap idly as those of sperm nuclei, thereby leading to the chromosome abnormalities commonly observed in nuclear-trans plant embryos.
Having concluded that the specialization of cells involves the differential activity of genes present in all cells, rather than the selective elimination of unwanted genes, we can now consider how genes are activated or repressed during early embryonic development. Nuclear transplantation has been used to demonstrate that the signals to which genes or chromosomes respond are normal constituents of cell cytoplasm. This information has come from experiments in which the nucleus of a cell carrying out one kind of activity is combined with the enucleated cytoplasm of a cell whose nucleus would normally be active in quite another way. One of two results is to be expected: either the transplanted nucleus should continue its previous activity or it should change function so as to conform to that of the host cell to whose cytoplasm it has been exposed. For the purposes of these experiments changes in nuclear activity have to be recognized by the appearance of direct gene products and not by the much less direct criterion of the normality of nuclear-transplant embryo development. Many of these experiments have been carried out in collaboration with Donald D. Brown of the Carnegie Institution of Washing ton or with another of my Oxford colleagues, H. R. Woodland.
The first experiments were designed to find out if the different functions performed by any one gene—the synthesis of DNA, the synthesis of RNA and chromosome condensation in preparation for cell division—are determined by cytoplasmic constituents. Three kinds of host cell were used: unfertilized but activated eggs whose nucleus would normally synthesize DNA but no RNA; growing oocytes in which the nucleus synthesized RNA but not DNA, and oocytes maturing into eggs, in which situation the nucleus consists of condensed chromosomes arranged in the "spindle" of cell division, and synthesizes neither RNA nor DNA. Two kinds of test nuclei were used: nuclei from adult brain tissue, which synthesize RNA but almost never synthesize DNA or divide, and nuclei from embryonic tissue at the mid blastula stage of development; mid-blastula nuclei do not synthesize RNA but synthesize DNA and divide about every 20 minutes. For technical reasons it was desirable to inject each host cell with many nuclei, even though this can prevent the subsequent division of the injected cell. The results were clear: In all respects tested the transplanted nuclei changed their function within one or two hours so as to conform to the function characteristic of the normal host-cell nucleus. Mid-blastula nuclei injected into growing oocytes stopped synthesizing DNA and dividing and entered a continuous phase of RNA synthesis that lasted for as long as the injected oocytes survived in culture (about three days). Adult brain nuclei injected into eggs stopped RNA synthesis and began DNA synthesis. When the same nuclei were injected into maturing oocytes, they synthesized neither RNA nor DNA but were rapidly converted into groups of chromosomes on spindles.
The next set of experiments was designed to find out if cytoplasmic components can repress or activate genes, that is, if they can select which genes in a nucleus will be active at any one time. Advantage was taken of the natural dissociation that exists in the time of synthesis of different classes of RNA during the early embryonic development of Xenopus. The work of several investigators has established the following sequence of events in Xenopus embryos. For the first 10 divisions after fertilization no nuclear RNA synthesis can be detected. Just after this—at the mid-late blastula stage—the cells synthesize large RNA molecules, which are believed not to include ribosomal RNA but which are likely to include "messenger" RNA. Toward the end of the blastula stage "transfer" RNA synthesis is first detected; this is followed a few hours later, during the formation of the gastrula, by the synthesis of ribosomal RNA.
The extent to which these events are under cytoplasmic control has been investigated by transplanting into enucleated eggs single nuclei from embryonic tissue at the neurula stage of development, the one that follows the gastrula stage. As the nuclear-transplant embryos develop, RNA precursor substances that have been labeled with radioactive atoms (for example uridine triphosphate labeled with tritium, the radioactive form of hydrogen) are used to determine the classes of RNA being synthesized at each stage. Autoradiography has shown that a neurula nucleus, which synthesizes each main kind of RNA, stops all detectable RNA syntheis, that is, it no longer incorporates labeled RNA precursors, within an hour of transplantation into egg cytoplasm. Furthermore, chromatography and other kinds of analysis show that, when the transplant embryos are reared through the blastula and gastrula stages, they synthesize heterogeneous RNA, transfer RNA and ribosomal RNA in turn and in the same sequence as do embryos reared from fertilized eggs.
Taken together, these experiments have shown that changes in the type of gene product (for example the synthesis of RNA or DNA), as well as changes in the selection of genes that are active (for example the synthesis of different types of RNA), can be experimentally induced. Since a high proportion of transplanted neurula nuclei support entirely normal development, the results show that egg cytoplasm must contain constituents responsible for independently controlling the activity of different classes of genes in normal living nuclei.
We can now consider what is perhaps the most interesting question of all: What is the mechanism by which cytoplasmic components bring about changes in gene activity? Of the various changes in chromosome and gene activity that can be experimentally induced in transplanted nuclei, special attention has been devoted to the induction of DNA synthesis by egg cytoplasm. It is easier to analyze than other changes, and it seems likely to exemplify certain general principles of cytoplasmic regulation in early embryonic development.
The origin of the cytoplasmic condition that induces DNA synthesis has been investigated by injecting adult brain nuclei, together with a radioactive labeling substance, into growing and maturing oocytes. The inducing factor appears just after an increase in the level of pituitary hormone has caused an oocyte to mature into an egg, an event that is accompanied by intensive RNA and protein synthesis.
Concerning the identity of the inducing factor, the first candidate to be considered was simply the presence of an adequate supply of DNA precursor substances. Woodland, however, has injected growing oocytes with 10 times the amount of all four common DNA precursors believed to be present in the mature egg. One of the precursors, thymidine triphosphate, had been labeled with tritium. In spite of the availability of these precursors, the brain nuclei did not incorporate the labeled thymidine into DNA. Although this experiment requires further analysis before DNA precursors can be excluded as inducers of DNA synthesis, it encourages a search in other directions.
The next candidate to be considered was DNA polymerase, an enzyme that promotes the incorporation of precursor substances into new DNA in a way that is specified by the composition of the preexisting "template" DNA. DNA polymerase activity in living cells has been tested by introducing purified DNA and tritium-labeled thymidine into eggs. In collaboration with Max Birnstiel of the University of Edinburgh we have established that the injected DNA serves as a template for synthesis of the same kind of DNA. When DNA and labeled thymidine are introduced into oocytes, no DNA replication can be detected. This means that the cytoplasmic factor in ducing DNA synthesis in eggs includes DNA polymerase or something that ac tivates this enzyme. It is doubtful, how ever, that this is the only constituent of the inducer. If it were, the injection of egg cytoplasm (which contains DNA polymerase) into oocytes might be expected to induce DNA synthesis, a result that is not in fact obtained. This experiment, in which purified DNA is replicated in the cytoplasm of unfertilized eggs, also serves to demonstrate that constituents of injected brain nuclei other than their DNA are not required in order to initiate the particular reaction being discussed here.
The last aspect of this reaction on which some information is available concerns the mechanism by which the inducing factors in the cytoplasm interact with the DNA in the nucleus. It was noticed several years ago by Stephen Subtelny, now at Rice University (and subsequently by others), that transplanted nuclei increase in volume soon after they have been injected into eggs. A pronounced swelling is also observed in nuclei injected into oocytes; the swelling is therefore not directly related to a particular type of nuclear response. During this nuclear enlargement chromatin (which contains the genetic material in the nucleus) becomes dispersed and, as Arms has demonstrated, cytoplasmic protein also enters the swelling nuclei. While working at Oxford, Robert W. Merriam of the State University of New York at Stony Brook found a close temporal relation between the passage of cytoplasmic protein into enlarging nuclei and the initiation of DNA synthesis. The interpretation of these events currently favored by those of us involved in the experiments is that the nuclear swelling and chromatin dispersion facilitate the association of cytoplasmic regulatory molecules with chromosomal genes, thereby leading to a change in gene activity of a kind determined by the nature of the molecules that enter the nucleus.
The experiments described here have established two general conclusions. First, nuclear genes are not necessarily lost or permanently inactivated in the course of cell differentiation. Second, major changes in chromosome function as well as in different kinds of gene activity can be experimentally induced by normal constituents of living cell cytoplasm. The same type of experiment is now proving useful in attempts to determine the identity of the cytoplasmic components and their mode of action.
We have had to restrict our attention to what can be described as sequential changes in gene activity, that is, differences between one developmental stage and the next. These may be compared with regional variations in nuclear activity, that is, differences between one part of an embryo and another at the same developmental stage. The latter are hard to study biochemically because of the difficulty in obtaining enough material. There is no obvious reason, however, why the processes leading to the two types of differentiation should be fundamentally different.
Experiments analogous to those described here have been conducted with bacteria infected with viruses, with nuclear transplantation in protozoans and with fusion in mammalian cells. Each kind of material is well suited for certain problems; nuclear transplantation utilizing amphibian eggs and cell nuclei is especially suited to the analysis of processes that lead to the first major differences between cells. Only after these differences have been established by constituents of egg cytoplasm are cells able to respond differentially to the important agents that guide development, such as inducer substances and hormones. Finally, the technique of nuclear transplantation may be used to introduce cell components other than the nucleus into the cytoplasm of different living cells; this is likely to be of great value for the more detailed analysis of early development and cell differentiation.