XY follicle cells in ovaries of XX↔XY female mouse chimaeras

We have recently shown that the Sertoli cells of adult XX↔XY male mouse chimaeras are exclusively XY (Burgoyne et al. 1988). This finding provides support for the view expressed by Burgoyne et al. (1986) that the Y chromosome acts cell autonomously in bringing about Sertoli cell differentiation.

It is generally accepted that ovarian follicle cells and testicular Sertoli cells are derived from the same gonadal cell lineage (the ‘supporting cell’ lineage). If the presence of a Y chromosome in a cell of this lineage is both necessary and sufficient to ensure its development into a Sertoli cell, fetal XX↔XY gonads would be expected to contain patches of XY Sertoli cells together with patches of XX prefollicular cells, so that the gonads should manifest as fetal ovotestes. That they are indeed ovotestes has recently been claimed by Bradbury (1987). We have argued (Burgoyne et al. 1988) that in most XX↔XY chimaeras there is subsequent regression of the ovarian component because the Mullerian duct inhibitor (AMH) produced by the Sertoli cells also inhibits the formation and survival of oocytes (Vigier et al. 1987) and, without oocytes, differentiation of prefollicular cells into follicle cells does not occur. XX↔XY gonads should only continue as ovotestes if there are too few Sertoli cells, and consequently insufficient AMH, to bring about oocyte failure. In rare cases, chance exclusion of XY cells from the supporting cell lineage might allow XX↔XY gonads to develop as ovaries, but we would expect all the follicle cells to be XX.

However, Ford et al. (1974) reported finding XY mitotic cells (presumed to be follicle cells) in the ovaries of two XX↔XY female mouse chimaeras. Although other sources of abundant mitoses in the ovary are difficult to envisage, the lack of direct proof that these XY cells were follicle cells prompted us to reinvestigate the origin of follicle cells in XX↔XY chimaeric ovaries.

Aggregation chimaeras were produced by standard techniques. One component was BALB/c. The embryos giving rise to the other component were produced by mating males from a CBA strain that was homozygous for the T6 translocation marker (Ford et al. 1956) and in which the normal mouse Y had been replaced by a metacentric variant (Winking, 1978), with either (1) females from the same CBA stock, or (2) C57BL/6Mcl females, or (3) females from an F₁ cross between CBA/Ca and C57BL/6Mcl. BALB/c expresses the glucose phosphate isomerase electrophoretic variant GPI-1A while the CBA crosses express GPI-1B. 35 overt chimaeras were produced: 17♀, 1 and 17♂.

The female chimaeras were mated to BALB/c males and the genotype (deduced from coat colour) and sex of the progeny was recorded. The final objective was to obtain information on the sex chromosomal constitution of the follicle cells and oocytes of XX↔XY female chimaeras. Since offspring are rarely obtained from the XY component of XX↔XY females (Ford et al. 1975), only females with progeny predominantly or exclusively of one genotype were selected for analysis, but a strong bias in favour of GPI-1B progeny rendered this criterion of little value. In order to circumvent this problem some of the later females were karyotyped from spleen biopsies. The females selected for analysis were primed with 5i.u. PMSG (Folligon, Intervet Laboratories) and killed 44 h later. Samples of liver, kidney, spleen and adrenal were frozen in liquid nitrogen and stored at −70°C for subsequent GPI analysis; a sample of bone marrow was taken for subsequent chromosome analysis. The ovaries were removed and teased apart in M2 medium (Whittingham, 1971). The follicles released were ruptured and the oocytes with adhering follicle cells (‘cumulus’) were washed individually three times in M2, and then were pipetted up and down in a drawn pipette to detach the follicle cells. The oocytes and follicle cell samples were collected and frozen in less than 1 μl of M2 (Buehr & McLaren, 1985) and stored for less than 1 week at −20°C for subsequent GPI analysis.

The somatic tissue samples were homogenized in approximately three times their volume of phosphate-buffered saline (Dulbecco A: Oxoid), centrifuged briefly and the supernatant applied to the gel. Oocytes and granulosa cell samples were freeze-thawed three times and applied directly to the gel from the pipette in which they had been collected.

etc.

The number of cells collected in the cumulus samples varied, but we estimate that at least 50 cells were necessary to produce clear bands on the gel.

The use of the T6 translocation marker and the metacentric Y chromosome marker enabled both sex chromosome complements to be identified in all but one of the overt aggregation chimaeras. Metaphases from the two XX ↔XY females are illustrated in Fig. 1. The two XX ↔XY females and the XX ↔XY hermaphrodite identified in the present study, together with ten XX ↔XY males from the same series of chimaeras, identified during the study of Burgoyne et al. (1988), are listed in Table 1. As might be expected, the two females had a relatively minor XY component.

Table 2 gives the results of the GP1 typing of follicles and oocytes for the XX ↔XX and XX ↔XY mice we have analysed, together with all the breeding data. The following points should be noted. First, the oocyte and breeding data are in very good agreement. Second, there is a strong bias in favour of GPI-1B oocytes and GPI-lB-derived offspring. Of the two females with a majority of GPI-lA-derived offspring, ♀Q has a GPI-1B XY cell line which is expected to be at a selective disadvantage in the germ line; indeed, the presence of an XY-derived offspring from ♀Q and of XY-derived oocytes in L may in part be due to the selection in favour of GPI-1B in these chimaeras. Third, in contrast to the oocytes, the GPI-1A component is well represented in the follicles. Importantly, this holds true for XX ↔XY ♀D (Fig. 2) where the GPI-1A component is XY.

In Table 3, the GPI-1 A contribution to the follicles and oocytes is compared with the proportion of GPI-1A in non-ovarian tissues. In addition to the rarity of GPI-1A oocytes, the GPI-1A component is markedly under-represented in the coat of the cross-bred ↔inbred mice, and to a lesser extent in the inbred♀inbred combination. Comparing the GPI-1A contribution to the follicles in XX ↔XY ♀D with that in the XX ↔XX females does not indicate any marked deficiency of XY cells. Apparently the presence of a Y chromosome interfered very little with the formation of follicle cells in this chimaera. In L, the only follicle analysed was GPI-1A (XX).

The present results support the conclusion of Ford et al. (1974) that XY cells can contribute to the follicles of XX ↔XY female mouse chimaeras.

How can the formation of XY follicle cells be reconciled with our earlier observation that Sertoli cells are exclusively XY in XX ↔XY male mouse chimaeras? The dilemma can be stated as follows: if follicle cells and Sertoli cells are derived from the same cell lineage, and the Y acts cell autonomously within this lineage to bring about Sertoli cell differentiation, how can XY supporting cells avoid commitment to the Sertoli cell pathway and become follicle cells? It is possible to escape this dilemma either by rejecting the assumption of a common cell lineage, or by rejecting our claim that the Y acts cell autonomously in this lineage. However, we feel that our conclusion that the Y acts cell autonomously is soundly based, and evidence from Taketo-Hosotani et al. (1985) that follicle cells can ‘transdifferentiate’ (see Burgoyne, 1988) into Sertoli cells confirms impressions from the study of T16/X Sxr ovotestes (Ward et al. 1988) and XO ↔XY mosaic ovotestes (P.S.B., unpublished data) that Sertoli cells and follicle cells are derived from the same lineage.

McLaren (1987) has suggested an alternative solution whereby a product of the supporting cell lineage, produced as a consequence of Y chromosome activity, has to reach a certain threshold concentration in order to complete commitment of the supporting cells to the Sertoli cell pathway. With too few XY supporting cells this threshold is not reached, thus allowing the cells to differentiate as follicle cells. This explanation is compatible with our present result, since the XY component was smaller in female D, in which we identified XY follicle cells, than in any of the XX ↔XY male chimaeras, in which the XY supporting cell lineage evidently gave rise to Sertoli cells. It cannot however account for the female chimaera described by Ford et al. (1974) in which 97 % of the presumed follicle cells were XY, and the XY contribution to other somatic tissues was correspondingly large.

It may be significant that both female XX ↔XY chimaeras reported by Ford et al. (1974) had an XY component derived from the AKR strain. Washburn & Eicher (1983) showed that the AKR Y chromosome on a C57BL/6 background, in association with the Thp mutation, leads to the development of ovaries and ovotestes in which all the follicle cells must be XY. They suggest (Eicher & Washbum, 1986) that the AKR Y chromosome carries a relatively late-acting allele at the testis-determining locus, Tdy, which fails partially or completely to pre-empt the early-acting C57BL/6 programme for ovary determination. The same explanation has been advanced to account for the occurrence of XY females when a Mus domesticus Y?08 chromosome is introduced into a C57BL/6 background (Eicher & Washburn, 1983, 1986). We have observed that the YP°S chromosome is indeed later acting than the C57BL Y, as judged by the timing of testis cord formation in C3H ♀ × C57BL/6 ♂ and C3H ♀ x C57BL/6-YPos ♂crosses (S. J. Palmer and P.S.B., unpublished data).

Burgoyne (1988) suggested that a similar mismatch between the timing of testis determination and ovary determination could explain the presence of XY follicle cells in XX ↔XY female chimaeras, if the XY component acts too late to pre-empt the process of ovary determination initiated by the XX component. Unlike the McLaren (1987) ‘threshold’ concept, the mismatch model can accommodate the female XX ↔XY chimaeras of Ford et al. (1974), in which the XY supporting cell population would have been recruited to form follicle cells before the late-acting AKR Y chromosome could initiate Sertoli cell development. In the present study, the XY follicle cells were found in an XX ↔XY female in which the XY component was inbred and the XX component was crossbred, a situation which would be expected to lead to the XY component being retarded relative to the XX component.

If this ‘mismatch’ explanation for the formation of XY follicle cells in XX ↔XY chimaeras is correct, then it requires that follicle cell development (unlike Sertoli cell development) is not cell autonomous, but involves cell-cell interaction. The inducing signal could derive from oocytes, since in the absence of oocytes no differentiation of follicle cells occurs. Macintyre, Baker & Wykoff (1959) reported the inhibition of Sertoli cell differentiation in XY embryonic rat gonads cografted with older ovaries: the time frame within which the ovaries were effective coincided with the presence of early meiotic oocytes.

In conclusion, we now envisage two routes by which XX ↔XY chimaeric mouse gonads can develop as ovaries. First, in XX ↔XY chimaeras with a very minor XY component, change exclusion of XY cells from the supporting cell lineage will allow ovarian development. Second, genetic differences between the two components of the chimaera can result in a developmental mismatch, with the testis-determination process being pre-empted by the process of ovary determination initiated by the XX component. Clearly, we now need to find out whether XY follicle cells can also form in sex chromosome chimaeras where there is no reason to expect a developmental mismatch between the two components. Because the gonads of such mice may only rarely develop as ovaries, it will be important to include ovotestes in any future studies.

We thank Robin Lovell-Badge for carrying out some embryo transfers and, together with Peter Koopman, for suggested amendments to the manuscript.