Non-Random X-Chromosome Inactivation in Mouse X-Autosome Translocation Embryos—Location of the Inactivation Centre

J. Embryol. exp. Morph. 78, 1-22 (1983) Printed in Great Britain (E) The Company of Biologists Limited 1983

Non-random X-chromosome inactivation in mouse X-autosome translocation embryos—location of the inactivation centre

By SOHAILA RASTAN1 From the Division of Comparative Medicine, Clinical Research Centre, Harrow

SUMMARY X-chromosome inactivation was investigated cytologically using the modified Katida method which differentially stains inactive X-chromosome material at metaphase in balanced 13|-day female embryos heterozygous for four X-autosome rearrangements, reciprocal translocations T(X;4)37H, T(X;11)38H and T(X;16)16H (Searle's translocation) and the insertion translocation Is(7;X)lCt (Cattanach's translocation). In all cases non-random inactivation was found. In the reciprocal translocation heterozygotes only one translocation product ever showed Kanda staining. In addition in a proportion of cells from T(X;4)37H, T(X;11)38H &nd Is(7;X)lCt the Kanda staining revealed differential staining of X-chromosome material and attached autosomal material within the translocation product. In a study of 8£-day female embryos doubly heterozygous for Searle's translocation and Cattanach's translocation two unbalanced types of embryo were found. In one type of unbalanced female embryo of the karyotype 40(X(7)/X16;16/16) no inactivated X- chromosomal material is found. A second unbalanced type of female embryo, of the presumptive karyotype 40(X(7)/XN;16x/l6) was found in which two inactivated chromosomes were present in the majority of metaphase spreads. A simple model for the initiation of X- chromosome inactivation based on the presence of a single inactivation centre distal to the breakpoint in Searle's translocation explains these findings.

INTRODUCTION Chromosomal rearrangements between the X chromosome and the autosomes interrupt the physical continuity of the X chromosome and are thus of use in the study of both the randomness and mechanisms of X-inactivation. One property of X-autosome translocations is the spread of inactivation into autosomal loci attached to the X chromosome analogous to position-effect variegation in Drosophila (Baker, 1971; Cattanach, 1974; Russell & Mont- gomery, 1970). Female mice which are heterozygous for X-autosomal translocations and carry marker genes which are recessive in the relevant autosomes exhibit a variegated phenotype due to spreading of inactivation into attached autosomal material in cells where the translocated X chromosome is inactive.

1 Author's address: Division of Comparative Medicine, Clinical Research Centre, Watford Road, Harrow, Middlesex HA1 3UJ, U.K. 2 S. RASTAN To date 14 X-autosome translocations have been reported in the mouse (reviewed by Eicher, 1970 and Searle, 1981). X-autosome rearrangements are often characterized by non-random X-inactivation. This could be the result of initial random X-inactivation followed by selection for cells with the maximum genetic balance or the result of primary non-random X-inactivation caused by a disturbance in the process of X-inactivation due to rearranged control centres. There is ample evidence that cell selection does operate in the two cell populations generated by random X-inactivation, for example in the female mule and hinny (Giannelli & Hamerton, 1971; Hamerton et al. 1971; Hook & Brustman, 1971; Cohen & Rattazzi, 1972), in heterozygotes for certain X-autosome aberra- tions in mouse and man (Cattanach, 1975; Disteche, Eicher & Latt, 1979; Russell & Cacheiro, 1978) and in blood cell populations of women heterozygous for X-linked hypoxanthine phosphoribosyl transferase deficiency (Nyhan et al. 1970). The data on inactivation centres are less clear cut. Various models for the initiation of X-inactivation have been proposed based on either the concept of a single inactivation centre on the X chromosome (e.g. Russell & Cacheiro, 1978) or more than one inactivation centre (e.g. Eicher, 1970; Disteche, Eicher & Latt, 1981). Previous work on balanced carriers of various reciprocal X- autosome translocations in the mouse suggest that only one translocation product is ever inactivated in cells in which the normal X is active (Russell & Montgomery, 1965, 1970). In particular in the autoradiographic studies of Russell & Cacheiro (1978) in which late replication was used as evidence of inactivation in cells from adults and 18-day embryos heterozygous for six independent translocations T(X;7)2R1, T(X;7)3R1, T(X;7)5R1, T(X;7)6R1, T(X;4)1R1 and T(X;4)7R1 it was found that in each case the shorter translocation product was never late-labelling in any cell. These results support the concept of a single inactivation centre from which inactivation is able to spread in both directions. However the inactivation of both parts of an X chromosome divided by an autosomal insertion such as in Is(X;7)lCt (Cattanach's translocation) has been used to support the concept of at least two inactivation centres (Eicher, 1970). Searle's translocation (T(X;16)16H) is of particular interest in the study of the mechanism of X-inactivation as it is the only X-autosome translocation in the mouse described to date that has the breakpoint in the X in a more or less central position. It is also characterized by marked nonrrandom inactivation. Heterozygous females behave as if the normal X is completely inactive in all cells. All X-linked mutant genes located on either part of the translocated X behave as if dominant, without variegation in the heterozygote (Lyon, 1966; Lyon, Searle, Ford & Ohno, 1964). In a cytogenetic study using Budr (5-Bromo- 2'-deoxyuridine) labelling Takagi (1980) reported that in balanced Searle's translocation heterozygotes at 6-5 days most cells had the normal X chromosome inactivated. He also found an unbalanced carrier embryo of the karyotype 40(XN/X16;16/16) which did not show any asynchronously replicating Non-random X-chromosome inactivation in mouse embryos 3 chromosome in any cell, despite the fact that inactivation of the X16 translocation product (containing the centromeric portion of the X) would have restored genetic balance. He concluded that the X16 product was incapable of being inactivated due to lack of an -inactivation centre and also" suggested that the concurrence of at least two X-chromosome loci separated by the breakpoint in Searle's translocation was necessary for the homologous X to be inactivated. In conflict with these results is the report by Disteche et al. (1981) using Budr labelling in balanced female Searle's translocation heterozygotes that in adult bone marrow and in 9-day embryos respectively, 7 % and 1 % of cells had the X16 product late replicating. This has been used to support the concept of at least two inactivation centres on the X chromosome. The studies reported here were designed to obtain further data on this controversy using the modified Kanda method (1973), (Rastan, Kaufman, Handyside & Lyon, 1980; Rastan, 1981) for differential dark staining of the inactive X chromosome, on female embryos heterozygous for the reciprocal translocations T(X;4)37H, T(X;11)38H and T(X;16)16H (Searle's translocation) and the insertion Is(7;X)lCt (Cattanach's translocation).

MATERIALS AND METHODS X-autosomal translocations (for review see Searle, 1981) T(X;4)37H (Fig. 1A). A reciprocal translocation between the X and chromosome 4 to give long and short somatic marker chromosomes. The translocation will be hereafter abbreviated to T37H. T(X;11)38H (Fig. IB). A reciprocal translocation between the X and chromosome 11 which gives long and short somatic markers. The translocation will be hereafter abbreviated to T38H. Is(7;X)lCt (see Fig. 1C). This translocation involves an inverted piece of chromosome 7 inserted into the X to produce a long somatic marker. The translocation occurs as two types, Type I, balanced, with the inserted X and a deleted chromosome 7, and Type II with the inserted piece of chromosome 7 present as a duplication (Cattanach, 1974). Unlike carriers of the other X-autosome translocations males of both types can be fertile. In this study only Type II (i.e. unbalanced duplication form) females and males were used. The translocation will be hereafter abbreviated to IslCt. T(X;16)16H (see Fig. ID). A reciprocal translocation between the X and chromosome 16. The longer translocation product with the centromeric segment of the X (X16) corresponds roughly in length to the intact X, and the shorter translocation product with the centromeric segment of chromosome 16 (16X), to the intact chromosome 16 (Eicher, Nesbitt & Francke, 1972). The translocation will be hereafter abbreviated to T16H. S. RASTAN T37H T38H

ft spf wa-2 spf

wi vt b m

TICt D T16H

X16

spf spf

Bn Ta Gy Mo

c P ru-2

Fig. 1. Relative cytological lengths of the various translocations: (A) T(X;4)37H, (B) T(X;11)38H, (C) Is(7;X)lCt (Cattanach's translocation) and (D) T(X,16)16H (Searle's translocation).

Embryos Embryos heterozygous for the various X-autosome translocations were produced by mating spontaneously ovulating females heterozygous for T37H, Non-random X-chromosome inactivation in mouse embryos 5 T38H and IslCt to normal Fi males of the 3H1 strain (Fi between two inbred strains 101/H and C3H/HeH). The Xce status of the translocated Xes have not been characterized; however the X chromosome of the 3HI males is known to carry Xce?. For T16H/+ heterozygous embryos, as the translocation products are not sufficiently different in size from a normal X chromosome or a normal chromosome 16 to be morphologically distinguishable, embryos doubly heterozygous for T16H and IslCt were produced to facilitate distinguishing between X chromosomes. This was achieved by mating spontaneously ovulating T16H/+ females to fertile Type II IslCt males. The validity of this approach has been demonstrated by Takagi (1980) who produced T16H/+ and T16H/IslCt heterozygous embryos and confirmed cytologically that there was no behavioural difference between the XN and X^ in the presence of T16H. This was also demonstrated genetically by Cattanach (1974). The day of finding the copulation plug was designated day \ of pregnancy. Pregnant females were killed by cervical dislocation on day 13! of pregnancy, the embryos dissected out of the uterus and the extraembryonic membranes discarded. The embryos were sexed by dissection of the gonad, male embryos were discarded and female embryos were kept for chromosome preparations. A num- ber of pregnant T16H/+ females which had been mated to IslCt males were killed on day %\ of pregnancy in order to recover unbalanced embryos which would otherwise be resorbed later in pregnancy.

Chromosome preparations The 13^-day female embryos were treated individually by the modified Kanda (1973) method for revealing the inactive X chromosome at metaphase, as previously described (Rastan etal. 1980). Briefly the embryos were incubated in medium 199 containing colchicine (4/igml"1 final cone.) to accumulate metaphases, then placed for 15min in hypotonic 0-5 % KC1 (w/v) at 50 °C and fixed in 3:1 absolute alcohol/glacial acetic acid fixative. The embryos were then disaggregated in 60 % acetic acid for 5 min and slides made on a hotplate at 40 9C. After staining in 2% Giemsa (buffered at pH6-8) for 20 min the inactive X- chromosome material stains much darker than any of the other chromosomes. The 8£-day Searle's translocation embryos were dissected into two parts. One part was treated by the modified Kanda technique for revealing the inactive X chromosome and the second part was used for conventional chromosome preparation for G-banding for karyotype analysis (Wurster, 1972).

Analysis Translocation carriers were recognized by the presence of marker chromosome(s) in the metaphase cells. In the case of T37H and T38H both long and short marker chromosomes are present. For IslCt, because it is an insertion, a long marker chromosome only is observed. In the case of T16H as no long or 6 S. RASTAN short markers are present, it is not possible to screen directly for translocation heterozygosity. However to distinguish T16H heterozygotes at 13£ days a genetic approach was adopted based on the premise that T16H causes extreme non- random inactivation. Female embryos from the cross IslCt/Y males by T16H/+ females, will be of the type IslCt/+ or IslCt/T16H. Since it is known that T16H causes extreme preferential inactivation of the X not involved in the reciprocal translocation, embryos in which virtually all the cells showed inactivation of only the marker IslCt chromosome were deduced to be double heterozygotes of the type IslCt/T16H. Embryos which had two types of cells (i) with the long marker IslCt inactive or (ii) with a chromosome of approximately the size of a normal X inactive, were deduced to be heterozygous for IslCt only, i.e. of type IslCt/+, in which random X-inactivation had occurred. Metaphase cells from female embryos which had been established to be translocation heterozygotes were scored in the following way:- (1) for the presence of a dark staining normal X chromosome or a dark staining translocated X chromosome; (2) for whether both translocation products were dark staining or whether only one translocation product was dark staining in cells in which the normal X was the active chromosome, to distinguish between inactivation of one or both translocation products; and (3) for whether or not differential staining could be seen in an inactive translocation product(s). The 8i-day embryos were sexed cytologically by determining the presence or absence of a Y chromosome. This task was facilitated by the fact that the Y chromosome is the darkest staining element in over 70 % of cells from postimplantation male mouse embryos treated by the modified Kanda method (Ras- tan, 1981). Slides of embryos without a Y chromosome were assumed to be of female embryos and were scored for the presence or absence of a dark staining inactive X chromosome(s).

RESULTS T37H/+ and T38H/+ embryos Four adult T37H/+ heterozygous females produced a total of 19 embryos at 13| days of which 11 were female and 3 proved to be heterozygous for T37H on chromosomal analysis. All three embryos were balanced carriers of the translocation. Four adult T38H/+ heterozygous females produced a total of 21 embryos, of which 8 were female and 3 proved to be heterozygous on chromosomal analysis. Again, all three embryos were balanced carriers of the translocation. In both T37H heterozygotes and T38H heterozygotes the short translocation product was never dark staining in any cell. The dark X was either the normal X or the long translocated X (see Table 1 and Figs 2A and 2B). No cell in which both translocation products were dark staining was ever seen for either translocation. In both T38H and T37H heterozygotes the proportion of cells with one or the other X dark staining differed significantly from the 1:1 ratio expected for Non

3 3 Table 1. 8- No. of Metaphases

- chromosi s LTX dark Genotype of No. of Total XN A STX Both No dark embryo embryos Dark Diff (%) Unif. Total dark TX dark X (%±S.E.M.) (%±S.E.M.) T37H/+ 3 297 191 60 20 80 0 0 26 3 (70-5 ± 2-8) (75-0) (29-0 ± 2-8) nacti T38H/+ 3 196 113 59 6 65 0 0 18 (63-5 ± 3-6) (90-8) (36-5 ± 3-6) vation IslCt/+ 7 230 84 73 58 131 NA NA 15 (39-1 ±3-3) (55-7) (60-9 ±3-3)

Key TX = Translocated X LTX = Long translocated X STX = Short translocated X Diff. = Differential staining Unif. = Uniform staining 8 S. RASTAN random X-inactivation (f = 45-46 for T37H, P < 0-001 and / = 12-94 for T38H, P< 0-001). For both T37H and T38H heterozygotes when the long translocation product (4X or llx) was the dark-staining chromosome in the cell differential staining within the long translocated X could frequently be seen (Figs 2A and 2B). For both translocations dark staining is clearly seen in the distal part of the translocated chromosome, which corresponds roughly to the region of the X- chromosomal material, whereas the proximal part of the translocation, corresponding roughly to the autosomal part of the translocation is pale staining. Table 1 shows that for T37H on average 75 % of cells in which 4X was the inactive chromosome showed differential staining within the translocated chromosome, and for T38H on average 90-8 % of cells in which the llx was the inactive chromosome showed differential staining within the translocated chromosome.

IslCt/+ heterozygous embryos Four adult IslCt/+ females produced a total of 30 embryos at 13 days of which 16 were female and 7 proved to be heterozygous for IslCt on chromosomal analysis. All 7 embryos were unbalanced Type II carriers. The inserted X chromosome (X7) was the dark-staining chromosome in 60-9 % of these cells compared with the normal X as the dark-staining chromosome in 39-1 % of cells (Table 1). This represents a departure from the 1:1 ratio expected from random inactivation which is significant at P = 0-01 level. A proportion of cells (55 %) also showed differential staining within the X^ chromosome when inactive (Table 1). Fig. 2C shows a cell in which X(7) is the inactive X chromosome, where a pale-staining region corresponding to the region of the autosomal insertion can be seen.

T16H/IslCt doubly heterozygous embryos a) 131-day embryos Four adult T16H/+ females crossed to Type II IslCtY males produced a total of 19 embryos at 13| days, 8 of which were female. Five of these embryos were deduced to be of the type IslCt/+, as discussed in the Analysis (Materials and Methods). The three remaining embryos were deduced to be double heterozygotes of the type T16H/IslCt, with the karyotype 40(X(7>/X16;16x/16), as

Fig. 2. (A) Metaphase spread from a female embryo heterozygous for T37H with the 4X chromosome, the long translocation product (arrow), inactive and showing a differential pale-staining region corresponding to the autosomal part of the translocation. Arrowhead indicates short translocation product. (B) Metaphase spread from a female embryo heterozygous for T38H with the llx chromosome, the long translocation product (arrow), inactive and showing a differential pale-staining region corresponding to the autosomal part of the translocation. Arrowhead indicates short translocation product. (C) Metaphase spread from a female embryo heterozygous for IslCt with the translocated X (X^7)) inactive (arrow) showing a pale-staining region corresponding to the region of the autosomal insertion. Non-random X-chromosome inactivation in mouse embryos 9

:•. r 10 S. RASTAN previously discussed. In these three embryos the long marker IslCt chromosome was the dark-staining chromosome in 100 out of 102 cells that showed a dark- staining chromosome, i.e. 98-0 % which is in accord with the evidence previously discussed of the extreme non-random inactivation caused by T16H. The two cells that did not have the long IslCt marker dark staining both showed an exception- ally short dark-staining chromosome, deduced to be the 16X, the shorter translocation product of T16H with the centromeric segment of chromosome 16, on the basis of size. No cell with both translocation products dark staining was ever seen. b) 8j-day embryos 19 embryos were recovered from six pregnant T16H females. Fig. 3 shows the possible outcome of balanced and unbalanced zygotes produced from the cross T16H female and IslCt Y male. Five of these embryos were male, as ascertained by the presence of a dark-staining Y chromosome. As described in Analysis (Materials and Methods) for the 13|-day embryos two types of balanced female embryos were found (i) four embryos in which virtually all the cells showed inactivation of only the marker IslCt chromosome and (ii) seven embryos which had two types of cells, with the longer marker IslCt inactive or with a chromosome of approximately the size of a normal X-inactive. These proved on karyotype analysis to represent females of the type IslCt/T16H and IslCt/+ respectively, as deduced for the 13^-day embryos. Two female embryos which showed no dark-staining X-chromosomal material proved on karyotype analysis to be unbalanced Type (a) i.e. X(7)/X16;16/16 thus confirming Takagi's observation that no inactive X is present in unbalanced female embryos of this type (Type (a), Fig. 3) (Takagi, 1980). In addition one exceptional female embryo was recovered in which two dark- staining chromosomes, one longer and one shorter, could be seen in the majority of cells. The 42 cells (out of the 44 analysed) from this embryo with two dark- staining chromosomes could be divided into two types, (i) 18 cells in which the ratio of the length of the long dark-staining chromosome to the short dark- staining chromosome was about 2-5 (Figs 4A, B) and (ii) 24 cells in which the ratio of the long to the short dark-staining chromosome was about 1-75 (Figs 4C, D). One presumptive tetraploid cell was also found which showed four dark- staining chromosomes, two longer and two shorter (Fig. 4E). Of the two cells (out of the 44 analysed) which did not have two dark-staining chromosomes, one had a long dark-staining chromosome and the other had a short dark-staining chromosome. Unfortunately this embryo was small and retarded so no karyotype analysis was performed on it due to the small amount of material available. However, it was deduced to be a chromosomally unbalanced Type (b) (Fig. 3) with the karyotype 40(X(7)/XN;16x/16) for the following reasons: (1) The two classes of cells with the ratio of the longer to the shorter dark- staining chromosome of 2-5 and 1-75 are thought to represent cells in which FEMALE GAMETES

BALANCED UNBALANCED

16 < 4 16 1 X 116 X J X I X I" 16 (x 16]

4 ( < 16 X lY Tl6 Tl6 JY X X • \ y iv X [ I t Ix I h I 16; [X |16 ' 16

Sterile cf carrier Normal cf Cf nullisomic for Cf monosomic for W distal segment of X distal segment of 16 Hs

# # A o ( X |16 X X J1 6[l« {X J16 Tx X

/[ALE fx 1 j (x i j <16 16 j £7 16] 27 1 1% |x * V lx Jx xf Balanced $ carrier for T16H leterozygous for TlCt Unbalanced $ Unbalanced $ also carrying TlCt (Type a) (Type b)

Fig. 3. Diagram to show the possible balanced and unbalanced zygote s produced from the cross T16H/+ female x IslCt Y male. 12 S. RASTAN

*•*»

-* » 1+'

> / * A V

CQ Non-random X-chromosome inactivation in mouse embryos 13 plus 16X and XN plus 16X respectively are inactive. This surmise is supported by the fact that the long chromosome in cells with the ratio of long: short of 2-5 often showed a pale-staining region corresponding to the region of the autosomal insertion of chromosome 7 into the X in IslCt (see Fig. 4B) whereas the longer chromosome in cells with a ratio of long: short of 1-75 never showed such a pale- staining region; (2) The shorter, dark-staining chromosome is thought to be 16X, the shorter translocation product of T16H, as it appears to be shorter than a normal X chromosome appears after the heat/hypotonic treatment; (3) The Kanda method occasionally produces G-banding (Rastan, unpublished observations). In Fig. 4D where G bands are visible, the G band pattern is compatible with the dark-staining chromosome being 16X, the shorter translocation product; (4) In Fig. 4E, the tetraploid cell in which four dark-staining chromosomes can be seen, a pale-staining area near the centromere can be seen in both the shorter chromosomes, compatible with these chromosomes being the 16X translocation products and spread of heterochromatinization from the X into the autosome being incomplete.

DISCUSSION Three inferences may be drawn from the results presented in this paper; (l) non-random inactivation is seen in embryos heterozygous for the X-autosome translocations T37H and T38H as well as for T16H and IslCt; (2) when an X chromosome is divided into two parts by a reciprocal translocation, only one part is capable of Kanda staining (3) spread of genetic inactivation from the X chromosome into translocated autosomal material, which results in position- effect variegation, can be reflected by heterochromatinization at the chromosomal level, and the spread of heterochromatinization appears to be variable.

Inactivation of only one translocation product The fact that only one translocation product in the reciprocal translocations T37H and T38H was ever seen to be dark staining by the modified Kanda method

Fig. 4. Metaphase chromosomes from an exceptional unbalanced female embryo carrier of Searle's translocation in which there are two dark-staining inactive chromosomes per cell, one longer and one shorter (arrows). (A) and (B) cells in which the ratio of the long: short dark-staining chromosome is 2-5. Note pale- staining region corresponding to the autosomal insertion in IslCt in longer dark- staining chromosome in (B). (C) and (D) cells in which the ratio of the long: short dark-staining chromosome is 1-75. (E) A tetraploid cell with four inactive dark- staining chromosomes, two longer and two shorter (arrows). Note pale-staining region on the proximal part of the two shorter dark-staining chromosomes. 14 S. RASTAN in 13|-day embryos supports the concept of a single inactivation centre located distal to the breakpoints on the X in T37H and T38H, resulting in inability of the short translocation product in each case to become inactivated. An alternative explanation is that lack of dark staining of the small translocation product could be an artefact. As discussed elsewhere (Rastan, 1981) the modified Kanda method does not reveal an inactive X in 100 % of cells, and it could be that the small translocation products in these cases are more prone to failure of the differential staining perhaps by virtue of their small size. This explanation is considered unlikely, however, as the Kanda method reveals a dark-staining Y chromosome in over 70% of cells from 13^-day male embryos (Rastan, 1981) and therefore small size per se is unlikely to preclude or interfere with the differential staining produced by the modified Kanda method. In addition, the fact that for T37H and T38H not a single metaphase was found with the short translocation product dark staining constitutes compelling evidence that the observed results represent failure of the short translocation products to become inactivated, and are not artefactual. Unfortunately genetic evidence is somewhat lacking due to a shortage of suitable markers near the centromere of the X chromosome. The data presented here supporting the concept of a single inactivation centre on the X chromosome are in agreement with the autoradiographic studies of Russell & Cacheiro (1978) on six of Russell's X-autosome translocations in which the short translocation product was never late-labelling in any cell. There is also some genetic evidence from Russell's translocations which suggests that only one part of an X divided by a reciprocal translocation is capable of inactivating attached autosomal loci (Russell & Montgomery, 1965, 1970). In particular in the R2 translocation (T(X;7)2R1), a reciprocal translocation between the X chromosome and chromosome 7, in which the breakpoint on chromosome 7 is between c and p (c being distal to p) and the breakpoint in the X is 23 units on the proximal side of Ta, the c locus is never inactivated, even though this locus lies close to the breakpoint and is readily inactivated in the three other reciprocal translocations involving chromosome 7, namely, T(X;7)3R1 T(X;7)5R1 and T(X;7)6R1. The inactivation of both parts of an X chromosome divided by an autosomal insertion as in IslCt has been used to support the concept of at least two inactivation centres. However, there is no a priori reason why one inactivation centre should not be capable of inactivating both parts of the X separated by an autosomal insertion if the spread of inactivation were not interrupted in the autosomal material. At first sight the differential pale staining of a region corresponding to the autosomal insertion found in IslCt when treated by the modified Kanda method may appear incompatible with this last stipulation. However, Cattanach (1974) has shown that subsequent reactivation of initially inactivated autosomal loci is possible in IslCt. This could explain the pale-staining region seen in about 55 % of cells in which the X^ was the inactivated dark-staining Non-random X-chromosome inactivation in mouse embryos 15 chromosome in 13^-day embryos, and also the observation that the autosomal insertion was early replicating in about 50 % of cells from 9-day and 13-day IslCt/+ embryos labelled with Budr in which the X(7) was the late-replicating chromosome(Disteche et al. 1979).

Differential staining within translocated chromosomes The differential staining seen in a proportion of cells within the translocation products when inactive could represent variable and limited spread of heterochromatinization from the inactive X portion into the attached autosomal material, or it could be the result of initial complete heterochromatinization of attached autosomal material followed by subsequent retreat of heterochromatinization in some cells during development of the embryo (cf. Cattanach, 1974). In spite of differential pale staining of what appears to be all, or most of, the autosomal region in the 4X chromosome in the majority of T37H/+ cells which have the long-translocation product inactive, it is known that X inactivation can spread at least 25 map units, the distance between b and the X- chromosomal breakpoint in T37H (Beechey & Searle, 1977; Searle & Beechey, 1977). Unfortunately, no inferences may be drawn from this as the precise relationship between map distance and cytological distance remains unknown. In addition, measurement of the relative lengths of the dark- and pale-staining regions in the translocated chromosome in different cells from an embryo, while giving an estimate of variability of extent of spread of heterochromatinization, is of limited value as length relationship will be complicated by the extension of the chromatin of the non-heterochromatinized portion due to heat denaturation.

Non-randomness of X-inactivation In all the translocation heterozygotes used in the present study a departure from the 1:1 ratio expected from random inactivation is seen. Extreme non- random inactivation of the X chromosome not involved in the translocation has already been established for T16H/+ heterozygotes (Ohno & Lyon, 1965), and for IslCt/+ heterozygous embryos Disteche et al. (1979) have shown that cell selection causes a departure from the 1:1 ratio expected from random inactivation, the direction of which depends on whether the embryo is the balanced Type I carrier or the unbalanced Type II carrier. This rationale explains the non- random inactivation in Type II unbalanced IslCt/+ heterozygous embryos found in this study. Cell selection arguments can be convincingly used to explain the non-random inactivation seen in both T37H and T38H heterozygotes as well as in IslCt heterozygotes. Spread of inactivation into attached autosomal loci would result in partial monosomy for the loci in question in cells in which the translocated X was inactivated, and would thus be selected against to a greater or lesser degree and might even result in cell death. This is in accord with the results presented here for T37H and T38H. The fact that size and viability are reduced in females 16 S. RASTAN heterozygous for T37H supports the idea of selective death of some cells with the translocated chromosome inactive in the developing embryo. However, for the reciprocal translocations another possibility must be considered. Since the results presented here and by Russell & Cacheiro (1978) indicate that only one translocation product is capable of being inactivated in the case of reciprocal translocations, one might also expect selection against cells in which the translocated X is inactive due to the remaining active segment of the X resulting in lack of dosage compensation ('functional disomy') for some X- linked loci. The larger the non-inactivated piece of X, the greater one would expect selection against such cells to be. T37H and T38H both have very small non-inactivated segments of X, but T16H, the only X-autosome translocation in the mouse in which the X is divided into two more or less equal-sized parts, is characterized by extreme non-random inactivation.

T16H/+ heterozygotes Takagi (1980) suggested that the extreme non-random inactivation seen in T16H heterozygotes was the result of (1) inability of the X16 product to become inactive, (2) inactivation in favour of XN and (3) rapid elimination of 16X inactive cells by cell selection. He further suggested that there was no X-inactivation at all in cells of embryos of the type (XN/X16;16/16) or (X^/X16; 16/16) due to inability of the X16 product to be inactivated and the necessity for the concurrence of at least two chromosomal loci separated by the T16H breakpoint for the homologous X chromosome to become inactivated. An alternative inter- pretation of the results of Takagi and this paper is presented here in a simple model (Fig. 5) for the initiation of X-inactivation which explains the following facts: (1) inactivation of only one part of an X chromosome separated by a reciprocal translocation (Russell & Cacheiro, 1978; present data); (2) non-inactivation of the X16 product as found by Takagi (1980); (3) lack of inactivation of any kind in unbalanced embryos of the type 40(XN/X16;16/16) or 40(X(?VX16;16/16); (4) inactivation of two chromosomes in the exceptional unbalanced embryo deduced to be of the karyotype 40(X(7)/XN;16x/16).

Details of the model (Fig. 5) This model is an extension of one of the hypotheses suggested by Russell & Cacheiro (1979). It deals with only the initiation of X-inactivation, not the spread

Fig. 5. Model for initiation of X-chromosome inactivation based on a single inactivation centre on the X chromosome distal to the breakpoint in T16H. Only one inactivation centre per cell may be blocked; physical linkage to an empty inactivation centre guarantees inactivation. Any other X-chromosomal material remains active. (For further explanation see text.) Non-random X-chromosome inactivation in mouse embryos 17 BALANCED T16H/+HETEROZYGOTE Xn/X16 . 16X/16 Alternative 1 Alternative 2

Xn X16 16X 16 xn X16 lgX lfi

XnON XnOFF X16ON X16ON 16X OFF 16XON UNBALANCED 9 TYPE a X(7)/X16 . 16/16 Alternative 1 Alternative 2 -16 16 16 same as alternative 1

X16ON X(7) ON UNBALANCED $ TYPE b X(7)/XN . 16X/16 Alternative 1 Alternative 2 Alternative 3 16X X™ Xn 16 16X X(7) Xn 16 16X X(7) Xn 16

16XON 16X OFF 16X OFF X(7) OFF X<7) ON X(7) OFF XnOFF XnOFF X"ON • FULL INACTIVATION CENTRE D EMPTY INACTIVATION CENTRE

Fig. 5 18 S. RASTAN of inactivation along the X chromosome nor the maintenance of inactivation. The following conditions are stipulated: (1) the presence of a single inactivation centre on the X chromosome located distal to the breakpoint in T37H, T38H and T16H; (2) the inactivation centre may be in one of two states, 'free' or 'blocked'. The possible ways in which the inactivation centre could be blocked include by attachment of an episome (Grumbach, Horishima & Taylor, 1963), attachment of protein products of a particular pair of autosomes or one set of autosomes etc. The details are not important to the general scheme of the model; (3) only one inactivation centre may be blocked in any cell, i.e. there is only one episome, or sufficient autosomal product etc., to block one inactivation centre; (4) if X-chromosomal material is physically linked to a free inactivation centre, inactivation is guaranteed. If, on the other hand X-chromosomal material is physically linked to a blocked inactivation centre, or no inactivation centre is present, it remains active; (5) initially any inactivation centre present in a cell may be blocked at random. For balanced carriers of T16H there are two alternative patterns of inactivation as two inactivation centres are present in each cell. Inactivation pattern alternative 1 produces cells which are not dosage compensated ('functionally disomic') for the proximal part of the X chromosome and partially monosomic for the proximal part of chromosome 16 to a greater or lesser extent, depending on the extent of spread of inactivation from the X into chromosome 16. Inactiva- tion pattern alternative 2, on the other hand, results in a cell which is balanced for both the X chromosome and chromosome 16. Although Takagi (1980) reported that in balanced Searle's translocation heterozygotes at 6-5 days most cells inactivated the normal X chromosome, McMahon & Monk (1982) showed that in T16H/+ female embryos heterozygous for isozymes of the X-linked phosphoglycerate kinase (PGK-1) X-chromosome inactivation (thought to be complete in the embryo by 5-5 days (Rastan, 1982a)) is initially random (with respect to the Pgk-1 locus) and followed by rapid selection against those cells having inactivated the Pgk-1 locus carried on T16H. Even if this meant losing half the cells of the early embryo full compensatory growth is known to be able to occur (Snow & Tarn, 1979). In this paper non-random inactivation in T16H heterozygotes is postulated to be the result of a combination of primary disturbance of inactivation centres, resulting in inactivation of only one translocation product, and secondary cell selection. For unbalanced carriers of Type (a) i.e. X(7)/X16;16/16, as shown in Fig. 5, only one inactivation centre is present, so there is only one possible alternative. X16 can never be inactivated because there is no inactivation centre present on this chromosome. In every cell the single inactivation centre present on X^ will be necessarily blocked and thus incapable of being inactivated and therefore there will be no X-inactivation at all in such embryos. This model explains the Non-random X-chromosome inactivation in mouse embryos 19 result found by Takagi (1980) and confirmed in the present study for such unbalanced embryos. For unbalanced carriers of Type (b), i.e. X(7)/XN;16X/16, Fig. 5 shows that tljere are three alternative patterns of inactivation, as three inactivation centres are present in each cell, each resulting in two inactivated chromosomes per cell. It must be remembered that even before inactivation all cells of such an embryo are monosomic for the distal half of chromosome 16. Cells with inactivation pattern alternative 1 will be, in addition, functionally nullisomic for the proximal part of the X chromosome and would thus be rapidly eliminated by cell selection. Disregarding for the time being the possible selective effects of having the X^ chromosome switched on or off in a cell, cells with inactivation pattern alter- natives 2 or 3 will both be partially monosomic for part of the proximal part of chromosome 16 to the same degree, which will depend on the extent of spread of inactivation from the X chromosome into chromosome 16. The two cell types, with either XN and 16X inactivated or X(7) and 16* inactivated, could thus be expected to be represented in equal proportions. However, as the embryo is also an unbalanced Type II carrier for IslCt, cell selection considerations for maximum genetic balance predict that inactivation pattern alternative 3, with X^ and 16X inactivated, would be the one most likely to restore maximum genetic balance. The single unbalanced embryo found of this putative type had, in fact, a slight preponderance of cells with XN plus 16X inactive, contrary to expectations. This difference is not statistically significant, however, and may be an artefact caused by low cell numbers. The finding of such an unbalanced embryo at 8£ days is probably a very rare event. Monosomies are known generally to be more deleterious to embryonic development than trisomies (reviewed by Searle, 1981) and are likely to be eliminated early in development. Takagi (1980) found no embryos of this type in a total of 88 embryos at 6£ to 8£ days from T16H/+ mothers. It is possible that the presumed karyotype of the unbalanced embryo is incorrect. In Fig. 3 showing the balanced and unbalanced gametes from T16H/+ heterozygous females the possibility of so-called 'adjacent-2' disjunction, in which homologous centromeres proceed to the same pole (McClintock, 1945) is not included. If adjacent-2 disjunction had occurred in this case an alternative karyotype for the unbalanced female embryo could be 40(X(7)/XN/X16;16). Such an embryo would still be monosomic, but for the proximal part of chromosome 16 rather than the distal part. If the exceptional unbalanced embryo were of this karyotype it would place the postulated inactivation centre in the model proximal to the breakpoint in T16H rather than distal. However, as an inactivation centre proximal to the breakpoint in T16H is incompatible with the evidence that the X16 product is incapable of being inactivated, the possibility of the unbalanced embryo having such a karyotype is considered to be unlikely. The report by Disteche et al. (1981) that the X16 product may be late replicating in a small proportion of cells has been used to support the concept of at least 20 S. RASTAN two inactivation centres on the X chromosome. However, according to Takagi (1980) and as reported here, no cell in which the X16 product was inactive was ever seen, even in unbalanced embryos where inactivation of the X16 product would have restored genetic balance. Non-inactivation of the X16 product due to absence of an inactivation centre is thus a basic tenet of the model proposed here. It is also known that the inactive X, although allocyclic, is not necessarily always late replicating. There is evidence that the allocyclic X chromosome may be either late replicating or early replicating in extraembryonic tissues of the early postimplantation mouse embryo (Takagi, Sugawara & Sasaki, 1982). In addition, Takagi (personal communication) has recently analysed replication patterns in bone marrow cells from adult female T16H/+ heterozygotes using Budr labelling and has found that the allocyclic X chromosome may be early replicating in a considerable proportion of cells, ranging from 30 % to 67 % of cells depending on the individual. In view of this variability in replication pattern of the allocyclic X it is possible therefore that the supposed presence of the X16 chromosome late replicating in a small minority of cells in the 1981 study of Disteche et al. was due to a misinterpretation. The model of initiation of X-inactivation presented here is compatible with the data from all Russell's X-autosome translocations especially the genetic evidence of non-inactivation of the c locus in the R2 translocation (Russell & Montgomery, 1965, 1970). Russell's data have placed the postulated inactivation centre between the breakpoints in R2 and R6. The model presented in this paper considerably further localizes the postulated inactivation centre to between the breakpoints of T16H and R6. It is of interest to note that this location of the inactivation centre, distal to the breakpoint in T16H and proximal to the breakpoint in R6 is compatible with it mapping in the same region as the Xce locus (Cattanach & Papworth, 1981). Although there is no direct proof that the Xce locus is the inactivation centre there is both biochemical evidence (John- ston & Cattanach, 1981) and cytogenetic evidence (Rastan, 1982b) that alleles at the Xce locus do, indeed, cause primary non-randomness of X-inactivation. Different alleles of the Xce locus could specify differences at the inactivation centre and thus affect the probability of an X chromosome carrying them remaining active or becoming inactivated. As only one inactivation centre per cell may be blocked off the model also explains the inactivation of all but one X in individuals with supernumerary X chromosomes. The model is at present being tested using different pluripotential stem cell lines derived from parthenogenetic embryos (E.K. lines) (Evans & Kaufman, 1981) containing various deletions of the second X chromosome (Robertson, Evans & Kaufman, 1983).

I would like to thank Sheila Brown for technical assistance and Mary Lyon for critically reading the manuscript and for much valuable discussion. Part of this work was supported by an MRC Studentship. Non-random X-chromosome inactivation in mouse embryos 21

REFERENCES BAKER, W. K. (1971). Evidence for position effect suppression of the ribosomal RNA cistrone in Drosophila melanogaster. Proc. natn. Acad. Sci. U.S.A. 68, 2472-2476. BEECHEY, C. V. & SEARLE, A. G. (1977). Mouse News Lett. 56, 39. CATTANACH, B. M. (1974). Position effect variegation in the mouse. Genet. Res. Camb. 23, 291-306. CATTANACH, B. M. (1975). Control of chromosome inactivation. Ann. Rev. Genet. 9, 1-18. CATTANACH, B. M. & PAPWORTH, D. (1981). Controlling elements in the mouse. V. Linkage tests with X-linked genes. Genet. Res. Camb. 38, 57-70. COHEN, M. M. & RATTAZZI, M. C. (1971). Cytological and biochemical correlation of the late X-autosome replication in the mule. Proc. natn. Acad. Sci., U.S.A. 68, 544-548. DISTECHE, C. M., EICHER, E. M. & LATT, S. A. (1979). Late replication in an X-autosome translocation in the mouse. Correlation with genetic inactivation and evidence for selective effects during embryogenesis. Proc. natn. Acad. Sci., U.S.A. 76, 5234-5238. DISTECHE, C. M., EICHER, E. M. & LATT, S. A. (1981). Late replication patterns in adult and embryonic mice carrying Searle's X-autosome translocation. Expl Cell Res. 133, 357-362. EICHER, E. M. (1970). X-autosome translocations in the mouse: total inactivation versus partial inactivation of the X-chromosome. Adv. in Genet. 15, 175-259. EICHER, E. M., NESBITT, M. N. & FRANCKE, U. (1972). Cytological identification of the chromosomes involved in Searle's translocation and the location of the centromere in the X-chromosome of the mouse. Genetics 71, 643-648. EVANS, M. J. & KAUFMAN, M. H. (1981). Establishment in culture of pluri-potential cells from mouse embryos. Nature 292, 154-156. GIANNELLI, F. & HAMERTON, J. L. (1971). Non-random late replication of X-autosomes in mules and hinnies. Nature 232, 315-319. GRUMBACH, M. M., HORISHIMA, A. & TAYLOR, J. H. (1963). Human sex chromosome abnor- malities in relation to DNA replication and heterochromatinization. Proc. natn. Acad. Sci., U.S.A. 49, 581-589. HAMERTON, J. L., RICHARDSON, B. J., GEE, P. A., ALLEN, W. R. & SHORT, R. V. (1971). Non- random X-autosome expression in female mules and hinnies. Nature 232, 312-315. HOOK, E. B. & BRUSTMAN, L. D. (1971). Evidence for selective differences between cells with an active horse X-chromosome and cells with an active donkey X-chromosome in the female mule. Nature 232, 349-350. JOHNSTON, P. G. & CATTANACH, B. M. (1981). Controlling elements in the mouse. IV. Evidence of non-random X-inactivation. Genet. Res. Camb. 37, 151-160. KANDA, N. (1973). A new differential technique for staining the heteropycnotic X chromosome in female mice. Expl Cell Res. 80, 463-467. LYON, M. F. (1966). Lack of evidence that inactivation of the mouse X-chromosome is incomplete. Genet. Res. Camb. 8, 197-203. LYON, M. F., SEARLE, A. G., FORD, C. E. & OHNO, S. (1964). A mouse translocation suppres- sing sex-linked variegation. Cytogenetics 3, 306-323. MCCLINTOCK, B. (1945). Neurospora I. Preliminary observations of the chromosome of Neurospora crassa. Am. J. Bot. 32, 671-678. MCMAHON, A. & MONK, M. (1982). X-chromosome activity in female mouse embryos heterozygous for PGK-1 and Searle's translocation, T(X;16)16H. Genet. Res. Camb. 41, 69-83. NYHAN, W. L., BAKAY, B., CONNOR, J. D., MARKS, J. K. & KEELE, D. K. (1970). Hemizygous expression of glucose-6-phosphate dehydrogenase in erythrocytes of heterozygotes for the Lesch-Nyhan syndrome. Proc. natn. Acad. Sci., U.S.A. 65, 214-218. OHNO, S. & LYON, M. F. (1965). Cytological study of Searle's X-autosome translocation in Mus musculus. Chromosoma (Berl.) 16, 90-100. RASTAN, S. (1981). Aspects of X-chromosome inactivation in mouse embryogenesis. (D. Phil. Thesis, University of Oxford). RASTAN, S. (1982a). Timing of X-chromosome inactivation in post implantation mouse embryos. /. Embryol. exp. Morph. 71, 11-24. 22 S. RASTAN RASTAN, S. (1982ft). Primary non-random X-inactivation caused by controlling elements in the mouse demonstrated at the cellular level. Genet. Res. Camb. 40, 139-147. RASTAN, S., KAUFMAN, M. H., HANDYSIDE, A. & LYON, M. F. (1980). X-chromosome inactivation in the extraembryonic membranes of diploid parthenogenetic mouse embryos demonstrated by differential: staining: Nature288, 172-173. ROBERTSON, E. J., EVANS, M. J. & KAUFMAN, M. H. (1983). X-chromosome instability in pluripotential stem cell lines derived from parthenogenetic embryos. /. Embryol. exp. Morph. 74, 297-309. RUSSELL, L. B. & CACHEIRO, N. L. A. (1978). The use of mouse X-autosome translocations in the study of X-inactivation pathways and non-randomness. In: "Genetic Mosaics and Chimaeras in Mammals", (ed. L. B. Russell). New York and London: Plenum Press. RUSSELL, L. B. & MONTGOMERY, C. S. (1965). The use of X-autosome translocations in locating the X-chromosome inactivation centre. Genetics 52, 470-471. RUSSELL, L. B. & MONTGOMERY, C. S. (1970). Comparative studies on X-autosome translocations in the mouse. II. Inactivation of autosomal loci. Segregation and mapping of autosomal breakpoints in five T(X;l)'s. Genetics 64, 281-312. SEARLE, A. G. (1972). Is sex-linked Tabby really recessive in the mouse? Heredity 17, 297. SEARLE, A. G. (1981). Numerical variants and structural rearrangements. In: "Genetic Variants and Strains of the Laboratory Mouse", (ed. M. C. Green). Stuttgart: Gustav Fisher. SEARLE, A. G. & BEECHEY, C. V. (1977). Mouse News Lett. 57, 18. SNOW, M. H. L. & TAM, P. P. L. (1979). Is compensatory growth a complicating factor in mouse teratology? Nature 279, 555-557. TAKAGI, N. (1980). Primary and secondary non-random X-chromosome inactivation in early female mouse embryos carrying Searle's translocation T(X;16)16H. Chromosoma 81, 439-459. TAKAGI, N., SUGAWARA, O. & SASAKI, M. (1982). Regional and temporal changes in the pattern of X-chromosome replication during the early postimplantation development of the female mouse. Chromosoma 85, 275-286. WURSTER, D. H. (1972). Mouse chromosomes identified by trypsin-Giemsa banding. Cytogenetics 11, 379-387.

(Accepted 10 August 1983)