Participation of cultured teratocarcinoma cells in mouse embryogenesis

Mouse embryos of various developmental stages can form teratocarcinomas when grafted to ectopic sites in adult hosts (Stevens, 1968, 1970; Solter, Skreb & Damjanov, 1970). Stem cells of these tumours, known as embryonal carcinoma cells (EC cells), can be grown in culture and can produce a variety of differentiated cell types when injected into adult hosts (Kleinsmith & Pierce, 1964). To discover whether similar factors control both the differentiation of EC cells and those of the normal embryo, these cultured cells were restored to the environment of the early embryo by microsurgery (Papaioannou, McBurney, Gardner & Evans, 1975). Participation of EC cells in normal morphogenesis was observed. We now present the completed analysis of the animals from experiments with three cell lines and, in addition, the results of injection of a fourth in vitro cell line. Several more chimaeric animals have been detected, notably through the growth of tumours in adulthood. Cultured cell lines have been used with the intent of eventually isolating mutant EC cells and introducing these mutants into mouse stocks via the germ line.

The embryonal carcinoma cell lines C86 and Cl7 derived from terato-carcinomas of mouse strain C3H and the line SIKR-OSB derived from strain 129 have been described (Evans, 1972; McBurney, 1976; Papaioannou et al. 1975; Iles & Evans, 1977). In addition, a fourth EC cell line PCC3/A/1 (Jakob et al. 1973; Guénet, Jakob, Nicolas & Jacob, 1974) derived from the tumour OTT6050 from a 6-day 129-strain embryo (Stevens, 1970) was injected into blastocysts of the strain AG/Cam. PCC3/A/1 has 40 chromosomes and is XO (Nicolas et al. 1976). It has been in culture for over 300 generations and is genetically non-albino, white-bellied agouti, and homozygous for the a allele of the glucose phosphate isomerase locus (Gpi-1).

Injection of mechanically isolated clumps of PCC3/A/1 embryonal carcinoma into blastocysts was carried out as described for the other cell lines (Papaioannou et al. 1975). Additional experiments were also carried out in which 1-5 isolated PCC3/A/1 cells or C17 cells were injected directly into the inner cell masses of a number of blastocysts of the appropriate genotype. The PCC3/A/1 cells were mechanically dissociated; C17 cells were disaggregated in 0·5% trypsin. Injected embryos were transferred either to the oviduct or to the uterus of random bred CFLP females (Anglia Laboratory Animals Ltd) in the first or third day of pseudopregnancy respectively and allowed to develop to term.

The offspring were inspected frequently for ocular and coat colour chimaerism evidenced by black pigment in the albino CFLP hosts or agouti hairs in the extreme non-agouti AG/Cam hosts. Blood samples were taken at weaning for glucose phosphate isomerase (GPI) horizontal starch gel electrophoresis. All mice that reached breeding age were bred with mice from the same experiment or with mice from the strain of the host blastocyst to test for functional gametes of donor origin. This should yield pigmented offspring in experiments using C3H EC cells and white-bellied agouti offspring in experiments with 129 EC cells. However, since none of the chimaeras from cell line C86 were pigmented, it was thought possible that this cell line had lost the capacity to form pigment. Hence, test offspring from all animals injected with C86 were also typed for GPL Finally, at autopsy of the animals derived from blastocysts injected with EC cells, samples of tissues, internal organs, and any tumours detected were taken for GPI analysis and histology. Tumours were sectioned at 8-12 μm and alternate slides were stained with haematoxylin and eosin and Masson’s trichrome. A small proportion of the animals were lost, decomposed or were otherwise unavailable for this analysis.

The experiments are summarized in Table 1. While the rate of development of injected blastocysts is similar for all cell lines, Cl7 produced the highest rate of chimaerism. The results of injection of each cell line can be individually summarized as follows:

The six chimaeras resulting from injection of clumps of C86 stem cells have been described (Papaioannou et al. 1975). These six mice had no pigment in skin or eyes but they all developed poorly differentiated tumours (Fig. 1) at or shortly after birth and were also chimaeric in at least one normal tissue as judged by GPI. None reached breeding age. Of the other 68 offspring there is no information on three animals. The rest were scored negative for pigment cell and GPI chimaerism. Of the 57 mice that reached breeding age, 49 were fertile and were test bred. A total of 3376 progeny were scored, an average of 69 offspring per mouse (range 4–179), with no evidence of germ line chimaerism. The majority of experimental animals were killed at about one year of age; no further tumours or other abnormalities were detected.

Thirteen chimaeric mice resulted from injection of Cl7 cells or cell clumps into blastocysts (Table 1). The 64 non-chimaeric mice were all scored negative for pigment and of the 57 that reached breeding age, 47 as well as 5 of the chimaeras, were fertile when test bred. In all 2889 offspring were scored (average 56/mouse, range 8–176) with no evidence of germ line chimaerism. Most of the non-chimaeric animals were killed at about one year of age and their internal organs were analysed for GPI. The number of progeny as well as the extent of chimaerism is detailed for the 13 chimaeras in Table 2. Chimaeras T6, T84, T85 and T89 were reported in the preliminary paper (Papaioannou el al. 1975) but the last three were still living at the time of publication. Of the 13 Cl7 chimaeras, five were normal tumour-free mice but with low levels of eye and/or coat pigmentation. Pigmented melanocytes were seen in both the retinal pigmented epithelium and in the choroid layer of the eye as illustrated in Fig. 2. T84 was the only one of these normal chimaeras to show internal chimaerism by GPI analysis. The other eight chimaeras all developed tumours. Mouse T6 was the only one to develop tumours shortly after birth; the others developed rapidly growing tumours when they were over 3 months of age and were killed within several weeks of the initial detection of the tumours. GPI analysis indicated that these tumours were largely composed of progeny of the injected embryonal carcinoma cells. Most of these mice showed little or no evidence of embryonal carcinoma-derived cells in normal tissues. T89 and T96 were negative for blood chimaerism at weaning but at death had trace amounts of EC type GPI in serum but not in packed red cells. This is perhaps due to the presence of the waste products from the large tumours. T134 also had EC type GPI in blood at death but not at weaning, in this case possibly attributable to post-mortem changes since the animal was found partially decomposed.

The single chimaera resulting from injection of clumps of this cell has been described (Papaioannou et al. 1975). Of the other 43 mice born seven were cannibalized at birth, the remaining 36 were negative for internal GPI chim-aerism. A total of 406 offspring were scored from 11 mice test bred (average 37/mouse, range 5–66). There were no germ line chimaeras.

Seventy-one of the 86 offspring which had received PCC3/A/1 cells were tested for chimaerism by GPI assay and examination of the coat (there is no information on the other 15 animals). Thirty-four were test bred giving a total of 653 progeny (average 19/mouse, range 3–47). There were four chimaeras. One, N85, had a small lump at the base of his tail that was primarily of the PCC3/A/1 GPI type. This animal died at 9 days of age (Table 3). Another chimaera, N27, had a patch of agouti hairs overlying a non-chimaeric lump of fatty tissue on her head. The other two chimaeras, both males, had small scattered patches of agouti hairs, but no other chimaerism was evident when these mice were assayed by GPI analysis. They had 16 and 40 progeny when test bred but with no evidence of germ line chimaerism.

Tumours composed largely of cell progeny of the injected cells developed following injection of three of the four cell lines. Results of GPI analysis of the Cl7 tumours are presented in Table 2 and results of examination of all tumours are summarized in Table 3 along with the locations. There were two main categories of tumour: those that were first detected at birth or shortly thereafter and were composed largely of undifferentiated embryonal carcinoma cells (Fig. 1) and those that became apparent later in the animals’ lives and tended to be mainly fibrosarcomas (Fig. 3). The former, with the exception of T6, were from cell line C86, the latter were all from cell line Cl7. It is interesting to note that the early tumours were mostly encapsulated subcutaneous tumours while those that developed later were more often located within skeletal muscle.

The fate of embryonal carcinoma cells from four in vitro cell lines has been examined following their injection into genetically dissimilar mouse blastocysts. These cells are capable of participation in normal embryogenesis (Papaioannou et al. 1975) as are teratocarcinoma stem cells from in vivo tumours (Brinster, 1974; Mintz & Illmensee, 1975; Illmensee & Mintz, 1976). However, the rate of chimaerism detected is low even after extensive GPI analysis and test breeding, being at most 24 % for isolated Cl7 cells. The levels of chimaerism were also very low, and chimaerism was sometimes detectable only by the presence of tumours developing in the adult animals. The rate of chimaerism in the studies using in vivo tumour cells (Mintz & Illmensee, 1975; Illmensee & Mintz, 1976) was in the same range (16–30 %) and levels of chimaerism were also generally low except in three cases. Germ line chimaerism was detected in one male mouse that was predominantly derived from the injected teratoma cells (Mintz & Illmensee, 1975). No germ line chimaeras were detected in our study. This may be related to the fact that the level of chimaerism was typically low. The cell lines used proved either XO or XX and would thus presumably only be capable of forming oocytes. Test breeding in females for detection of chimaerism is slow and in addition development of XO germ cells is likely to be impaired (Lyon, 1974). It is also relevant that the four embryonal carcinoma cell lines all have abnormal karyotypes as seen by G-banding in spite of normal or near normal chromosome numbers. This may put the cells at some disadvantage in the embryo or render them incapable of forming functional gametes.

Another factor of possible importance for the incorporation of cells into the embryo is the difference of cell cycle times between the embryo and the injected cells. In the embryo average cell cycle time at 6·5 days is 5 h and cycle times as little as h are seen in a certain highly proliferative zone (Snow, 1976 and personal communication). In contrast, the cell cycle times of the cultured embryonal carcinoma cells range from 12 to 18 h. Unless the injected cells shorten their cycles fairly rapidly, they are unlikely to contribute in a major way to the developing embryo. An asynchrony of cell cycle times could also help explain the unusual distribution of embryonal carcinoma-derived cells in the chimaeras, most strikingly seen in the coat colour chimaeras (Papaioannou et al. 1975). This is quite unlike the extensive contribution of the two cell types to most tissues seen when chimaeras are formed by injection of synchronous embryonic inner cell mass cells (Ford, Evans & Gardner, 1975). Illmensee & Mintz (1976) have also pointed out that differences in cell adhesiveness between ICM and injected cells may result in delayed teratoma cell integration and account for the sporadic distribution of teratoma-derived cells in their chimaeras.

There are several possibilities to explain the high incidence of tumours in our chimaeras. The transition that occurs when normal embryonic cells become teratocarcinomas can be reversed as shown by the normal tissues in the chimaeras and in particular by the formation of functional sperm in one chimaera obtained by Mintz & Illmensee (1975). This may depend on a variety of factors such as incorporation of the EC cells into the embryo. Our experiments always involved injection of more than one cell and it could be that the normal tissues and the tumours were derived from different injected cells. Single EC cell injections would clarify this point (Illmensee & Mintz, 1976).

Mintz, Illmensee & Gearhart (1975) found that the majority of dividing core cells of OTT6050 embryoid bodies they used as an in vivo source of EC cells for injection into blastocysts contained 40 chromosomes. However, they did not provide a detailed karyotypic analysis of these cells. Karyotypic abnormalities have been demonstrated for all the cell lines used in the present studies, in spite of the fact that two have a modal chromosome number of 40 (Iles & Evans, 1977; McBurney, 1976; Nicolas et al. 1976). These abnormalities could be an important factor affecting the capacity of the cells to participate in embryogenesis. To investigate this, injection of cultured EC cells with normal karyotypes is now underway in our laboratory. Since all of the cell lines used have been in culture for extended periods and are thus well adapted to culture conditions, there may be other culture related changes, for example in cell surface properties, genetic load or cell cycle time, that cause these cells to retain tumour forming capacity. One explanation for the apparent rapid growth of tumours shortly after birth would be that some injected cells behave autonomously and grow exponentially without being incorporated into the developing embryo. The late developing tumours must be otherwise explained. Whether they developed from injected cells which somehow remained quiescent for several months or from their more differentiated progeny cannot be determined from our limited information. The formation of fibrosarcomas has also been observed following the injection into adult animals of cell lines isolated from in vitro differentiated EC cells (J.-F. Nicolas, personal communication).

The occurrence of rapidly growing tumours may indicate a serious limitation to the use of in vitro EC cell lines if it proves to be general. By contrast, only one teratoma-derived tumour has been reported from the in vivo cells (Illmensee & Mintz, 1976). However, the potential advantages of using cultured EC cells for introduction of selected mutants into mouse strains by blastocyst injection makes it all the more important to attempt to discover the cause of the tumours, i.e. the reason for the incomplete recovery of normal function when these cells are returned to an embryonic environment. It is also important in order to further define the validity of using cultured EC cells as a model of embryogenesis.

We wish to thank H. Mardon, L. Ofer, E. P. Evans, Dr C. F. Graham, Mrs M. White and also Dr E. Ilgren who helped us with the classification of the tumours.

The work was supported by the Cancer Research Campaign, the Medical Research Council, the Centre National de la Recherche Scientifique, the Institute National de la Santé et de la Recherche Médicale (No. 76.4.311 AU), the National Institute of Health (CA 16355), the Fondation pour la Recherche Médicale Française, and the Fondation André Meyer. C.B. was supported by a short term EMBO Fellowship.