Transfer of Primordial Germ-cells in Xenopus laevis

There have been many claims for the segregation of Anuran primordial germcells at an early embryonic stage. Most authors agree that these cells may be distinguished with ease in the most dorsal region of the larval endoderm and, somewhat later in development, at the base of the dorsal mesentery and in the undifferentiated gonad (see review by Johnston, 1951). Bounoure (1934) and Blackler (1958) claim to have traced the origin of the primordial germ-cells as early in development as the late blastula stage and to have recognized cell inclusions that become restricted to the germ line at all stages between the fertilized egg and the late blastula.

As pointed out by Everett (1945), some workers in this field of embryological study have firmly denied the existence of primordial germ-cells, while others have been cautious of accepting the principle that these cells give rise to any of the definitive sex-cells (gametes). The experimental studies reported in this paper present a proof that not only do primordial germ-cells actually exist in late neurula stages but that at least some of these cells are directly ancestral to some of the functional gametes.

The history of the primordial germ-cells in the South African Clawed Toad, Xenopus laevis, has already been documented by Nieuwkoop (1956 a, b) and Blackler (1958) and, in view of such other considerations as ease of rearing and breeding and the research programme of this laboratory, it was decided to employ embryonic material of this species.

The technique of transfer of the primordial germ-cells, described below, is a technique whereby germ-cells of embryo A can develop within embryo B. The application of the technique, as reported here, is concerned with aspects of the germ-cell problem, but it may be applied also to problems such as sex-reversal and the production and genetic analysis of abnormal embryos after nuclear transplantation (see Fischberg, 1960).

In the light of our knowledge of Xenopus germ-cells there appear to be three stages in development which offer the most likely chances of transfer success, as outlined by Blackler (1960). All transfers commented on in this present paper were effected using neurulae of stages 19–26 (Nieuwkoop & Faber, 1956).

The primordial germ-cells of the stage 23 neurula can be readily seen in sections stained by the Altmann-Volkonsky method. Their cytoplasm possesses the characteristic ‘cytoplasme germinal’ of Bounoure (1934), which is distributed adjacent to the nucleus and also apposed to the cell membrane. As may be seen in Plate 1, fig. A, the cells are quite closely aggregated—a point at variance with the situation in Rana temporaria—and situated deep in the posterior endoderm about 0·6 mm. anterior to the hind-end of the embryo (i.e. just anterior to the anus). They number between 20 and 30 cells.

After the decapsulation of both host and donor neurulae in full-strength Niu and Twitty solution, the endodermal region containing the primordial germcells of the host (together with the overlying ventral mesoderm and ectoderm) is excised and replaced by the same region taken from the donor embryo. Textfig. 1 indicates the scheme of the operation and the size of the piece removed or grafted.

The grafted material is kept in position by a glass bridge and may take 10 minutes or more to heal in place sufficiently to allow removal of the bridge. When the edges of the graft are almost obliterated by ectodermal overgrowth, the host is transferred to 0·1 Niu and Twitty solution. Here it must be stressed that ectodermal overgrowth is not to be taken as an indication that the graft has been harmoniously incorporated. Graft breakdown may occur as late in development as the stage 40 tadpole. The appearance of the host embryo after healing is shown in Plate 2, fig. D.

Only 40 per cent, of all transfers made were successful in that not only did the graft heal in place but the host continued to develop and began feeding normally. In later experimental series, not reported here, we have obtained 70–90 per cent, success in transfers, demonstrating that the skill of the operator is also involved as an important factor for success.

It will be appreciated that the removal of the host germ-cell region should theoretically result in complete host sterility. However, it is more reasonable to suppose that in practice some of the host’s germ-cells may not be removed during the operation: it is thus necessary to ensure that grafted germ-cells can be distinguished from any remaining host germ-cells. In this respect, the use of the nuclear marker discovered by Fischberg (see Elsdale, Fischberg,& Smith, 1958) has allowed such a distinction to be made.

In a normal embryo of X. laevis all of the constituent cells potentially bear two nucleoli per nucleus, although some may contain only one nucleolus, probably the result of fusion. Such embryos can be termed 2N. In embryos heterozygous for the nuclear marker, every cell has a nucleus with only one nucleolus. These marked embryos can be termed 1N.

Plate 1, figs. B, C, show cells of a normal (2N) and a marked (1N) tadpole. The mutation behaves in a simple Mendelian manner so that, when a 1N frog (of either sex) is mated with a 2N frog, half the offspring are 1-nucleolate and half are 2-nucleolate (actually, 48·5 per cent. 1N : 51·5 per cent. 2N in a multisample analysis of over 550 offspring (Fischberg & Wallace, 1960)).

When germ-cells from a normal (2N) neurula are grafted into a marked (1N) neurula and the host gonads later examined microscopically, any 2N cell observed must be necessarily of graft origin. It is, however, not possible to determine whether cells carrying a single nucleolus are of host or graft origin.

For the rest, the technique of germ-cell transfer must ensure that the grafted cells are capable of developing into functional gametes. Some experimental animals were raised to sexual maturity, and in these both possible combinations of the nuclear marker have been employed. The 1N-in-2Ncombination is superior since it is only by the mating of adult frogs of this character that certain offspring may be ascribed unreservedly to the grafted material. Text-fig. 2 shows how the percentage of graft success may be calculated from inspection of the offspring derived from the mating of normal (2N) frogs with experimental frogs of either combination.

Three marked (1N) animals into which an unmarked (2N) graft had been made at the neurula stage were killed at completion of their metamorphosis and their kidneys and attached gonads fixed and sectioned. All 3 pairs of gonads possessed some gonocytes of 2N type (see Plate 2, figs. E-H). In the females, many oocytes were beginning their growth phase, making the nucleolar count impossible, but it was easy to assess the number of nucleoli in the less-developed oogonia.

Twelve experimental frogs were raised to sexual maturity and mated to unmarked normal frogs according to the scheme of Text-fig. 2.

Three of these frogs were of the 2N-in-1N combination and the results of their matings with normal (2N) frogs are summarized in Table 1.

The remaining 9 frogs were of the 1N-in-2N combination and have been mated with 2N frogs. All the 1Noffspring can be ascribed directly to the grafted material, and an equivalent number of the 2N offspring are of like source although they cannot be distinguished from those 2N offspring derived from host germ-cells remaining after the operation. The results of these matings are summarized in Table 2.

The experiments described above demonstrate that grafts of posterior endoderm containing the so-called primordial germ-cells can be successfully carried out between embryos of Xenopus at the neurula stage. The healing process is facilitated if the ecto-mesodermal coat of the ventral endoderm is transferred as well, but its presence is not essential for success. It is clear that some cells of the grafted material, in successful operations, make their way into the host gonadal ridges and here are able to perform the differentiation necessary for the production of functional eggs and sperm. However, it must be readily admitted that it is not possible to state if the graft gametes are derived from some of the transplanted endodermal cells, from the so-called primordial germ-cells described by Bounoure and Blackler, or from both sources. Nevertheless, there are reasons, which derive their force from being taken together, for supposing that only the primordial germ-cells are involved. In the first place, the technique was specifically designed to ensure, as far as possible, that cells containing the ‘cytoplasme germinal’ were transferred. Secondly, if the grafted endoderm (whose cells are much more numerous than the primordial germ-cells) provided the functional sex-cells, it is difficult to interpret those experimental cases in which the graft healed in perfectly but none of its cells gave rise to gametes. Moreover, there is no reason to suppose that if ‘germinal’ endoderm cells exist they have the same distribution and aggregation as the primordial germcells.

In the experiments, care was taken at the time of operation to graft pieces of endoderm very similar in size. In view of the argument at the end of the last paragraph, one is naturally inclined to ask why the graft is successful in some cases and not in others. We would draw attention to the significant result that graft success is either very high or, with the exception of one case, very low indeed. This result is what one might expect if the primordial germ-cells are so closely aggregated that small differences in the size of the piece grafted would determine whether or not they were included. A comparison of Text-fig. 1 and Plate 1, fig. A shows that the primordial germ-cells lie very near the dorsal periphery of the grafted region. It is impossible, of course, to cut pieces of exactly the same size, and thus justifiable to expect the clump of germ-cells to be wholly included in the graft or to be wholly excluded from it.

Moreover, we feel that the embryonic stage of the donor must be considered. As shown in Plate 1, fig. A, the primordial germ-cells are already quite dorsal in the posterior endoderm of neurulae at stage 23. Some of the donor neurulae used in the experiments were already at stage 26, and it is quite possible that when grafting from such late neurulae the primordial germ-cells have moved too far dorsally to be included in the piece excised. Unfortunately, the recipient embryos were not maintained in isolation, through considerations of space, from the time of operation to sexual maturity, and the records have become, in consequence, somewhat less detailed. An inspection of the protocols suggests a rough concordance between the number of neurulae, later than stage 23, used as donors and the number of frogs showing no graft success, but this should be regarded only as an indication, and no proof, of the interpretation advanced earlier in this paragraph. Anyone who cares to repeat, or utilize, the technique for neurula stages is advised not to use as donor material any embryo later than stage 23 of the normal table.

Whatever view one likes to take in this matter, it is certain that at least some of the definitive gametes are the descendants of cells located in the posterior endoderm of neurulae. While the ‘cytoplasme germinal’ is a very useful cellular marker for recognizing the primordial germ-cells in sections of fixed embryonic material, the experiments reported here do not in any way support the idea that the ‘cytoplasme germinal’ acts as a germ-cell determinant. This is a different problem, requiring an extensive experimental analysis not so far performed.

No graft success has been obtained with mature female animals in these studies (see Tables 1 and 2), although graft oocytes have been detected in the sectioned material. This failure is of no significance, since in further work, using the technique but for a differently designed experiment, we found that females were quite capable of laying fertilizable eggs derived from grafted primordial germcells.

It is noticeable that in those experimental cases where graft success has been nil, the frogs, nevertheless, produce numerous gametes of host origin. Sterilization of the host during the operation is thus not always achieved, and the gametes produced subsequently may have been derived from an unreduced number of host primordial germ-cells or from a reduced number of these cells that have subsequently regulated by an increase in their mitotic rate. This latter situation may be compared with that described by Bounoure, Aubry, & Huck (1954) in which the primordial germ-cells, reduced in number by ultra-violet irradiation, nevertheless, completely repopulate the gonad. We feel the character of our experiments was not such as to justify raising again the issue of metaplasia to explain the failures.

In conclusion we believe that (1) primordial germ-cells definitely exist in neurula stages of X. laevis, and (2) these cells are directly ancestral to at least some of the definitive gametes. The possibility remains that some of the gametes are derived from some secondary source, though this seems unlikely in view of the high percentage of graft success in successful ‘takes’.

1. Une technique est décrite dans laquelle les prétendues cellules germinales primordiales de Xenopus laevis, ainsi que quelques cellules somatiques qui les accompagnent, peuvent être greffées d’un embryon à l’autre au stade neurula. Un marqueur nucléaire permet de reconnaître les cellules greffées.

2. Quelques cellules du greffon migrent dans les gonades de l’hôte et y donnent naissance à des gamètes fonctionnels. La conclusion indique que les cellules germinales primordiales existent au stade neurula, que ces cellules sont probablement identiques aux prétendues cellules germinales primordiales décrites dans les études histologiques, et qu’elles sont directement à l’origine d’une partie au moins des ovules et des spermatozoïdes définitifs.

The authors wish to thank the British Empire Cancer Campaign for a grant to support work of which these studies form a part.