Characterization of Two Mst1-Deficient Mouse Models

Abstract

Mammalian sterile 20-like kinase 1 (Mst1) is a ubiquitously expressed serine/threonine kinase belonging to the family of Sterile 20-like kinases. MST1 has been inferred to play important roles in apoptosis and in the inhibition of proliferation in mammalian cells. Here, we describe the genetic characterization of Mst1-deficient mice produced by two distinct gene-trap insertions. Animals generated from clone RRT293 exhibit transmission ratio distortion favoring the mutated allele and is amplified with each generation. Inexplicably, while the mutated allele is favored for transmission, its homozygosity is embryonic lethal. By contrast, animals generated from the second Mst1 gene-trap clone, AJ0315, do not show any gross abnormalities. We find that the discrepancy in phenotype is most likely attributable to a second insertion in the RRT293 clone. Thus, a mutation in Mst1 alone does not affect survival. Our results set the stage for identification of the lethal second-site mutation that is paradoxically favored for transmission.

INTRODUCTION

Mammalian sterile 20-like kinase 1 (Mst1) is a serine/threonine kinase that belongs to the GC kinase subfamily of Sterile 20-like (Ste20-like) kinases (Creasy and Chernoff, 1995; Taylor et al., 1996). The Mst1 and the related Mst2 protein (78% identical) consist of an N-terminal Ste20-like catalytic domain, an auto-inhibitory domain, and the C-terminal regulatory region (Creasy and Chernoff, 1995; Dan et al., 2001). The C-terminal end contains both a dimerization domain and a region for the regulation of the kinase activity (Creasy et al., 1996). In cell lines, Mst1 can be activated by a variety of proapoptotic stimuli such as severe environmental stresses or stimulation of death receptors (Graves et al., 1998; Graves et al., 2001; Lee et al., 2001; De Souza et al., 2002; Yamamoto et al., 2003). Mst1 contains a caspase recognition motif between the catalytic and regulatory domains, and caspase cleavage releases a highly active catalytic fragment (Lee et al., 1998; Kakeya et al., 1999; Reszka et al., 1999). In addition to its activation by proapoptotic stimuli, Mst1 overexpression is sufficient to induce apoptosis (Graves et al., 1998; Ura et al., 2001b; Ura et al., 2001a; Glantschnig et al., 2002) and may be the major mediator of apoptosis in certain circumstances (49).

Overexpressed MST1 activates mitogen-activated protein kinase (MAPK), stress-response p38 MAPK and JNK (c-Jun N-terminal kinase) pathways (Graves et al., 1998; Ura et al., 2001b; Glantschnig et al., 2002). Mst1 is localized predominantly in the cytoplasm, but cycles through the nucleus (24) and the truncated catalytically active protein accumulates in the nucleus (Lee et al., 2001; Glantschnig et al., 2002; Lee and Yonehara, 2002). Mst1 has been shown to associate with Death-Associated Protein 4 in mammalian cells, which was suggested to promote apoptosis by co-localization and subsequent activation of p53 (Lin et al., 2002). The caspase-mediated cleavage of Mst1, functioning through activation of the JNK pathway, is reported to be essential and sufficient for inducing chromatin condensation during apoptosis (Ura et al., 2007).

The Drosophila Hippo (Hpo) is homologous to Mst1 and Mst2; Hpo together with Salvador and Warts, is proposed to comprise a tumor suppressor complex (Harvey et al., 2003; Jia et al., 2003; Pantalacci et al., 2003; Wu et al., 2003). Flies lacking Hpo exhibit increased cell proliferation and decreased apoptosis during eye development (Harvey et al., 2003; Udan et al., 2003; Wu et al., 2003). In mammalian cells, Mst1 associates with the tumor suppressor protein RASSF1A (Praskova et al., 2004; Avruch et al., 2005), which may serve to couple Mst1 to certain upstream inputs (28, 37). Evidence bearing on the putative role of Mst1 as a tumor suppressor in humans is scanty and conflicting (27, 41).

Two recent studies have identified likely physiologic functions of Mst1. In C. elegans, RNAi-mediated knockdown of the Mst1 homologue, Cst-1, was demonstrated to shorten life span and accelerate tissue aging, and over-expression of Cst-1 resulted in delayed aging and increased maximal life span (Lehtinen et al., 2006). Mst1, in complex with the RASSF5 polypeptide Nore1B/RAPL, also plays an important role in the activation of lymphocyte integrins in response to antigen receptor and chemokine stimulation (Katagiri et al., 2006). To gain further insight into the physiological functions of Mst1, we generated mice null for Mst1. We obtained two distinct null alleles of Mst1 through insertional mutagenesis. Herein, we attempted to breed the mice to homozygosity and, in one mouse model, found an unusual form of transmission ratio distortion when inactivation of Mst1 is accompanied by disruption of a second, as yet unidentified gene. However, in a second mouse model, inactivation of Mst1 alone yielded only minor phenotypes at the whole organism level. Implications of these findings for the function of Mst1 and explanations for the apparent TRD in one mouse model are discussed.

RESULTS AND DISCUSSION

Derivation of Mst1-null mice from ES clone RRT293

To generate a mouse lacking the Mst1 gene, we searched the BayGenomics website and identified two embryonic stem (ES) cell clones containing disruptions within the Mst1 gene. We first characterized clone RRT293, which contained the splice acceptor site from the gene trap vector (pGToLxf) between exons 4 and 5 of the Mst1 gene. Briefly, this gene trap vector is a promoter-less reporter construct consisting of a splice acceptor site followed by B-galactoscidase/neomycin selection marker and a poly A tail (Fig. 1A) (Gossler et al., 1989). Insertion of this vector within intron 4 of the Mst1 gene resulted in the production of a non-functional Mst1 transcript consisting only of exons 1, 2, 3, and 4 out of 10 exons. We injected RRT293 ES cells into C57BL/6J blastocysts and obtained chimeric male mice capable of transmitting the mutated Mst1 allele. Southern and PCR analysis confirmed the presence of heterozygous animals among the litters (Fig. 1A and 1C).

Figure 1.

Disruption of the Mst1 gene in two different ES cell clones used to generate mice. (A) Structure of the Mst1 gene and location of the gene-trap vector between exons 4 and 5 present in BayGenomics clone RRT293. The gene-trap vector contains a splice acceptor site (SA) within the En2 leader sequence, followed by a B-galactosidase reporter (B-gal), a Neomycin selection marker (Neo), and an IRES sequence containing a polyA sequence. The 5′ external probe and the internal probe are designated by horizontal bars. Arrows designate the primers used for genotyping pups via PCR. (B) Structure of the Mst1 gene and location of the gene trap vector for the second Mst1 clone (AJ0315). The gene-trap vector is present between exons 1 and 2 in this clone. (C) PCR gel (top) and the corresponding Southern blot analysis (bottom) for a typical intercross between second generation Mst1^+/- animals derived from the RRT293 clone. (D) PCR gel (top) and the corresponding Southern blot analysis (bottom) for a typical intercross between second generation Mst1^+/- animals derived from the AJ0315 clone.

To produce homozygous mice, we intercrossed Mst1^+/- animals but were unable to obtain any Mst1^-/- animals. To determine when embryonic lethality took place, we carried out timed matings between Mst1^+/- mice and genotyped embryos at E11.5 and E15.5. Only heterozygous and wildtype embryos were obtained (Table 1), indicating that Mst1^-/- lethality occurred before E11.5. Interestingly, heterozygous embryos were significantly over-represented.

Table 1.

Genotyping embryos from various stages of embryonic development resulting from the second generation of intercrosses (Mst1^+/- crossed to Mst1^+/-) for animals derived from the ES cell line RRT293. The genotype of absorbed embryos was not determined.

	E11.5	E12.5	E13.5	E15.5	Total
Mst1+/+	1	0	2	0	3 (9%)
Mst1+/-	8	8	9	4	29 (91%)
Mst1-/-	0	0	0	0	0
Total	9	8	11	4	32

P<0.005

Transmission ratio distortion results from intercrosses of RRT293-derived mice

Although the heterozygous intercrosses yielded no homozygous mice, we did notice an unusual transmission ratio distortion (TRD), defined as a significant departure from expected Mendelian inheritance ratios. Intriguingly, we noticed a TRD that favored Mst1^+/- animals over wildtype Mst1^+/+ which arose after two generations of intercrosses. The first generation of intercrosses between Mst1^+/- animals yielded 70% Mst1^+/- and 30% Mst1^+/+ animals (n = 90), a slight deviation from the expected frequency of 66% Mst1^+/- and 33% Mst1^+/+ in the presence of homozygous lethality (χ² = 0.067, P = 0.90) (Table 2). There were roughly the same number of males and females. Consistent with the loss of homozygous embryos, the average litter size was 5.9, slightly smaller than the average litter size of 7.0 pups/litter for C57BL/6J mice (Silver, 1995), though not statistically significant (χ² = 0.16, P = 0.5).

Table 2.

First generation of intercrosses for animals derived from the ES cell line RRT293. Statistical significance was determined using chi-squared analysis, and P-values were calculated for all genotypes (for a particular sex) comparing values to expected Mendelian frequencies in the presence of homozygous lethality (expected frequencies of 66% Mst1^+/- and 33% Mst1^+/+). Percentage values within parentheses indicate distribution of each genotype for a particular sex.

	Male *	Female #	Total **	Total litters	Average litter size
Mst1+/+	11 (28%)	17 (33%)	28 (31%)	15	5.9
Mst1+/-	28 (72%)	34 (67%)	62 (69%)
Mst1-/-	0	0	0
Total	39	51	90

^*P = 0.497

^#P = 1.0

^**P = 0.904

Curiously, heterozygous Mst1 animals increased in number relative to wildtype animals with each subsequent generation of intercrosses. In the second generation of intercrosses, 79% of all pups were Mst1^+/- (n = 29, χ² = 0.6954, P = 0.352) (Table 3). Again, there was no sex ratio distortion. There was a slight increase in average litter size for this generation (7.3 animals/litter) compared to the previous generation (5.9 animals/litter). Most surprising were the results of intercrosses in the third generation, which produced only heterozygous animals (n = 32, χ² = 5.33, P = 0.0003). Twice as many female animals were born as compared to male animals (Table 4). [Note: The sex-ratio distortion resulting from the third generation intercrosses may not be significant because only four litters were genotyped, and the average litter size for this generation was 6.4, slightly less than the previous generation.] The combined frequencies for all three generations indicate that there is a TRD favoring heterozygous over wildtype animals (78% Mst1^+/-; χ² = 2.65, P = 0.019) (Table 5).

Table 3.

Second generation of intercrosses for animals derived from the ES cell line RRT293. Statistical significance was determined using chi-squared analysis, and P-values were calculated for all genotypes(for a particular sex) comparing values to expected Mendelian frequencies in the presence of homozygous lethality (expected frequencies of 66% Mst1^+/- and 33% Mst1^+/+). Percentage values within parentheses indicate distribution of each genotype for a particular sex.

	Male *	Female #	Total **	Total litters	Average litter size
Mst1+/+	3 (20%)	3 (21%)	6 (21%)	4	7.3
Mst1+/-	12 (80%)	11 (79%)	23 (79%)
Mst1-/-	0	0	0
Total	15	14	29

^*P = 0.273

^#P = 0.345

^**P = 0.836

Table 4.

Third generation of intercrosses for animals derived from the ES cell line RRT293. Statistical significance was determined using chi-squared analysis, and P-values were calculated for all genotypes (for a particular sex) comparing values to expected Mendelian frequencies in the presence of homozygous lethality (expected frequencies of 66% Mst1^+/- and 33% Mst1^+/+). Percentage values within parentheses indicate distribution of each genotype for a particular sex.

	Male *	Female #	Total **	Total litters	Average litter size
Mst1+/+	0	0	0	4	6.4
Mst1+/-	11 (100%)	21 (100%)	32 (100%)
Mst1-/-	0	0	0
Total	11	21	32

^*P = 0.019

^#P = 0.001

^**P = 0.00034

Table 5.

Total of all intercrosses for animals derived from the ES cell line RRT293. Statistical significance was determined using chi-squared analysis, and P-values were calculated for all genotypes (for a particular sex) comparing values to expected Mendelian frequencies in the presence of homozygous lethality (expected frequencies of 66% Mst1^+/- and 33% Mst1^+/+). Percentage values within parentheses indicate distribution of each genotype for a particular sex.

	Male *	Female #	Total **	Total litters	Average litter size
Mst1+/+	14 (22%)	20 (23%)	34 (22%)	23	6.5
Mst1+/-	51 (78%)	66 (77%)	117 (78%)
Mst1-/-	0	0	0
Total	65	86	151

^*P = 0.0437

^#P = 0.0474

^**P = 0.0188

Thus, intercrosses of animals derived from clone RRT293 exhibit an unusual form of TRD that favors animals containing one null Mst1 allele. This non-Mendelian pattern of inheritance apparently exhibits a generational effect, with the bias becoming stronger with each generation. To our knowledge, this is the first example of a TRD in a mammalian system with preferential transmission of a mutated allele that causes extinction or near-extinction of wildtype animals after three generations of intercrosses. A similar though distinct form of TRD is observed at the t-complex in mice (Schimenti, 2000; Lyon, 2003). Heterozygous t/+ male mice preferentially transmit the t chromosome to 99% of their offspring because t sperm (with a higher fertilization potential than wildtype sperm) out-compete + sperm (which have become functionally impaired) for oocyte fertilization. (Silver and Olds-Clarke, 1984; Olds-Clarke and Peitz, 1986; Lyon, 1987). The TRD results from various t-complex distorter genes acting in trans in all t/+ sperm, and only t sperm remain functional due to the expression of a t-complex responder gene (which is lacking in wildtype + sperm). Two Rho G protein signaling pathways have been implicated in TRD of the t-haplotype (Herrmann et al., 1999; Bauer et al., 2005; Bauer et al., 2007).

Given that Mst1 is an important intermediary for signaling in the apoptosis cascade, several underlying mechanisms could be envisioned for the generational TRD observed in the present study. For example, its haploinsufficiency in the oocyte might affect polar body survival or the rate of oocyte apoptosis in the female germline. During female meiosis, two consecutive divisions generates one oocyte — the true gamete -- and two polar bodies which are destined to degenerate (Pardo-Manuel de Villena and Sapienza, 2001b). The second polar body is believed to undergo apoptosis, in contrast to the first polar body which is thought to undergo necrosis (Bartholomeusz, 2003). Therefore, one attractive idea is that segregation of the Mst1 mutation to the polar body may interfere with its degradation and promote inheritance of the Mst1 mutation. Alternatively, Mst1 deficiency could affect gametic function, enhance fertilization, or improve embryo survival.

Generation and characterization of Mst1-null mice derived from AJ0315 ES cells

Because of the unusual nature of this TRD, we sought to confirm the phenotype by analyzing a second gene-trap insertion of Mst1, clone AJ0315, obtained from the Sanger Institute Gene Trap Resource. The Mst1 gene in this clone is interrupted between exons 1 and 2, resulting in a severely truncated (only 70 bp) and non-functional Mst1 transcript. Western blot analysis indicated reduced levels of the Mst1 protein (Fig. 2A,B), and Northern analysis also confirmed the production of the truncated transcript caused by the gene traps in RRT293 and AJ0315 ES cells (Fig. 2C and data not shown). We then derived chimeric mice from AJ0315 ES cells and successfully transmitted the mutated Mst1 allele through the germline (Fig.1B and 1D). Heterozygous pups sired by the chimeric males were then outcrossed to C57BL/6J with no obvious anomalies in segregation ratios or litter sizes (Fig. 3).

Figure 2.

Mst1 protein and RNA levels in gene-trap ES cell clones and mouse tissues. (A) Western blot analysis of protein lysates from wildtype ES cells and both Mst1 disrupted gene-trap ES clones confirms that Mst1 protein levels are reduced in the mutants by approximately 50%. The Mst1 protein (60 kDa) is indicated by the arrow. (B) Western blot analysis of testicular and ovarian tissue from wildtype, heterozygous, and homozygous null Mst1 animals (derived from clone AJ0315) also indicates a reduction of Mst1 protein levels, relative to wildtype, in these tissues. (C) Northern analysis of Mst1 transcripts from RRT293 ES cells and liver tissue from AJ0315 animals shows reduction of full-length Mst1 transcripts (7 kb) in mutant mice. (D) Northern analysis using an exon 1-2 oligo probe shows truncated transcript of 395 nt for RRT293. Data not shown for AJ0315.

Figure 3.

First outcrosses of heterozygous Mst1 animals (derived from clone AJ0315) to C57BL/6. (A) Outcross matings between heterozygous Mst1 females and C57BL/6 male mice. (B) Outcross matings between heterozygous Mst1 males and C57BL/6 female mice.

AJ0315-derived Mst1-mutant mice are viable

Next, we performed intercross matings to determine if TRD also exists for animals derived from clone AJ0315. We found several unexpected results. First, among multiple litters, we were able to obtain Mst1^-/- pups of both sexes (Table 6), indicating that absence of the Mst1 kinase is not lethal in mice. This was surprising, considering that Mst1^-/- mice were never obtained from clone RRT293.

Table 6.

Intercrosses between the first generation of animals derived from ES cell line AJ0315. Statistical significance was determined using chi-squared analysis, and P-values were calculated for all genotypes (for a particular sex) comparing values to expected Mendelian frequencies of 50% Mst1^+/-, 25% Mst1^+/+ and 25% Mst1^-/- . Percentage values within parentheses indicate distribution of each genotype for a particular sex.

	Male	Female*	Total	Total litters	Average litter size
Mst1+/+	39 (25%)	26 (18%)	65 (21%)	46	6.7
Mst1+/-	81 (51%)	73 (50%)	154 (51%)
Mst1-/-	38 (24%)	46 (32%)	84 (28%)
Total	158	145	303

^*P = 0.06

Second, aberrations in allele transmission in AJ0315-derived mice was much less severe — if present to any significant degree -- than for RRT293-derived mice. For the AJ0315-derived strain, male animals exhibited Mendelian frequencies of inheritance (assuming expected ratios to be 25% Mst1^+/+, 50% Mst1^+/-, and 25% Mst1^-/-), although there may be a slight bias favoring Mst1^-/- female animals (32% Mst1^-/-; n = 303, χ² = 2.6, P = 0.1) (Table 6). Closer examination of intercross matings revealed, strangely, that first litters from first-time mothers showed TRD in which there were higher numbers of Mst1^-/- female than expected (41% Mst1^-/-; n = 132, χ² = 8.0, P < 0.005) (Table 7A). However, the effect disappeared in the second and third litters (Tables 7B,C), although there were slightly higher numbers of male Mst1^+/- (63%) than expected. The average litter size for both first litters and second litters was similar (6.8 animals/litter). Thus, for the first-generation intercrosses, a TRD is significant only in the first litters.

Table 7a.

The first litters (from first-time mothers) from the first generation of intercrosses. Statistical significance was determined using chi-squared analysis, and P-values were calculated for all genotypes (for a particular sex) comparing values to expected Mendelian frequencies (50% Mst1^+/-, 25% Mst1^+/+ and 25% Mst1^-/-). Percentage values within parentheses indicate distribution of each genotype for a particular sex.

	Male	Female *	Total #	Total litters	Average litter size
Mst1+/+	20 (35%)	10 (13%)	30 (23%)	21	6.3
Mst1+/-	21 (37%)	34 (45%)	55 (42%)
Mst1-/-	16 (28%)	31 (41%)	47 (36%)
Total	57	75	132

^*P = 0.002

^#P = 0.004

Table 7b.

The second litters (from the same mothers) from the first generation of intercrosses.

	Male	Female	Total	Total litters	Average litter size
Mst1+/+	6 (14%)	6 (24%)	12 (18%)	10	6.8
Mst1+/-	27 (63%)	12 (48%)	39 (57%)
Mst1-/-	10 (23%)	7 (28%)	17 (25%)
Total	43	25	68

Table 7c.

The third litters (from the same mothers) from the first generation of intercrosses.

	Male	Female	Total	Total litters	Average litter size
Mst1+/+	8 (27%)	8 (31%)	16 (29%)	7	8.0
Mst1+/-	17 (57%)	13 (50%)	30 (54%)
Mst1-/-	5 (17%)	5 (19%)	10 (18%)
Total	30	26	56

We next examined what happened in subsequent generations. Intercrosses of second-generation Mst1^+/- mice exhibited no unusual patterns of inheritance (Table 8). However, in the third generation intercross, fewer Mst1^-/- males were obtained (8.5%; n = 151, χ² = 7.8, P = 0.005) and slightly more Mst1^+/- animals were obtained overall (62% ; χ² = 2.0, P = 0.1) (Table 9). To determine whether Mst1 protein expression levels had been altered in the gametes of animals from this generation, we performed Western blot analysis on testes and ovaries from wildtype, heterozygous, and homozygous littermates. Expected levels were found for all (Fig. 2B). In the fourth-generation intercross, inheritance patterns reverted to that expected (Table 10). When averaged across all four generations, no significant TRD was evident at all (Table 11). Next, we asked whether matings between heterozygous and homozygous mice might unmask more severe non-Mendelian patterns of inheritance. None was evident (Figure 4A, 4B). Combined, these results showed that Mst1 deficiency in the AJ0315-derived mice does not yield the generational TRD favoring heterozygotes that was readily observed in the RRT293-derived strain.

Table 8.

Intercrosses between the second generation of animals derived from ES cell line AJ0315. Statistical significance was determined using chi-squared analysis, and P-values were calculated for all genotypes (for a particular sex) comparing values to expected Mendelian frequencies (50% Mst1^+/-, 25% Mst1^+/+ and 25% Mst1^-/-). Percentage values within parentheses indicate distribution of each genotype for a particular sex.

	Male	Female	Total	Total litters	Average litter size
Mst1+/+	24 (27%)	14 (17%)	38 (23%)	24	7.0
Mst1+/-	46 (52%)	43 (53%)	89 (53%)
Mst1-/-	18 (21%)	24 (30%)	42 (25%)
Total	88	81	168

Table 9.

Intercrosses between the third generation of animals derived from ES cell line AJ0315. Statistical significance was determined using chi-squared analysis, and P-values were calculated for all genotypes (for a particular sex) comparing values to expected Mendelian frequencies (50% Mst1^+/-, 25% Mst1^+/+ and 25% Mst1^-/-). Percentage values within parentheses indicate distribution of each genotype for a particular sex.

	Male *	Female	Total #	Total litters	Average litter size
Mst1+/+	21 (29.6%)	15 (19%)	36 (23%)	21	7.2
Mst1+/-	44 (62%)	48 (61%)	92 (61%)
Mst1-/-	6 (8.5%)	16 (20%)	22 (15%)
Total	71	79	151

^*P = 0.005

^#P = 0.006

Table 10.

Intercrosses between the fourth generation of animals derived from ES cell line AJ0315. Statistical significance was determined using chi-squared analysis, and P-values were calculated for all genotypes (for a particular sex) comparing values to expected Mendelian frequencies (50% Mst1^+/-, 25% Mst1^+/+ and 25% Mst1^-/-). Percentage values within parentheses indicate distribution of each genotype for a particular sex.

	Male	Female	Total	Total litters	Average litter size
Mst1+/+	18 (27%)	14 (24%)	32 (26%)	15	8.3
Mst1+/-	33 (49%)	33 (57%)	66 (53%)
Mst1-/-	16 (24%)	11 (19%)	27 (23%)
Total	67	58	125

Table 11.

Total intercrosses between animals derived from ES cell line AJ0315. Statistical significance was determined using chi-squared analysis, and P-values were calculated for all genotypes (for a particular sex) comparing values to expected Mendelian frequencies (50% Mst1^+/-, 25% Mst1^+/+ and 25% Mst1^-/-). Percentage values within parentheses indicate distribution of each genotype for a particular sex.

	Male	Female	Total	Total litters	Average litter size
Mst1+/+	102 (27%)	69 (19%)	171	106	7.3
Mst1+/-	204 (53%)	197 (54%)	401
Mst1-/-	78 (20%)	97 (27%)	175
Total	384	363	747

Figure 4.

Backcrosses, outcrosses, and incrosses of Mst1 animals derived from clone AJ0315. The average litter sizes and total number of litters analyzed is shown for each mating. (A) Backcross matings between homozygous null female Mst1 animals and heterozygous male Mst1 mice. (B) Backcross matings between heterozygous female Mst1 animals and homozygous null male Mst1 mice. (C) Outcross matings between homozygous null Mst1 female mice and C57BL/6 male mice. (D) Outcross matings between homozygous null Mst1 male mice and C57BL/6 male mice. (E) Outcross matings between homozygous null Mst1 female mice and Mus Castaneus male mice. (F) Outcross matings between homozygous null Mst1 male mice and Mus Castaneus female mice. (G) Incross matings between homozygous null Mst1 animals.

Mst1^-/- AJ0315-derived mice are grossly normal but show reduced fertility

Given that Mst1^-/- pups were born at expected frequencies overall, we examined whether the AJ0315-derived mice showed any gross phenotype. To test their fertility, we crossed homozygous animals and found that both male and female Mst1^-/- animals are fertile but seem to have smaller litter sizes (average 4.8 animals/litter) (Fig. 4G). By contrast, control matings between Mst1^-/- animals and wildtype C57BL/6J mice showed normal litter sizes (7.3-8.3 animals/litter), which were similar to average sizes for inbred C57BL/6J crosses and to the four-generational average for the Mst1^+/- animal matings (Fig.4C, 4D). Nor did outcrossing Mst1^-/- mice to Mus castaneus (CAST) show any effect on fertility (7.3 animals/litter for Mst1^-/- female x CAST male; 5.1 animals/litters for CAST female x Mst1^-/- male), with similar numbers of male and female pups (Fig.4E, 4F) [Note: CAST females have smaller litter sizes on average; therefore, the average of 5.1 pups/litter is normal for the CAST female x Mst1^-/- male cross). Thus, only crosses between Mst1^-/- mice resulted in smaller litter sizes than expected. Mst1^+/- and Mst1^-/- animals also exhibited no evidence of premature mortality, at least up to 15 months of age. A more detailed characterization has revealed significant defects in leukocyte function, however (Zhou, et al., in preparation). We conclude that not only are Mst1^-/- animals viable but they are also fertile and show no gross abnormalities.

Differences between the two Mst1^-/- ES clones

The dramatically different phenotypes observed for RRT293 and AJ0315 mice prompted us to address underlying differences in the insertional mutagenesis. One possibility is that the diverse phenotypes are caused by differences between the two truncated Mst1 transcripts, with clone RRT293 producing a transcript containing exons 1-4 and clone AJ0315 producing a non-function transcript consisting of exon 1. A second hypothesis is that one of the Mst1 clones may have a second genetic disruption, which together with the Mst1 disruption could present a different phenotype. To test this idea, we performed DNA fluorescence in situ hybridization (DNA-FISH) to determine the number of integrations using the pGToLxf gene-trap vector as a probe. Indeed, two integration sites could be observed in RRT293 (Fig. 5A). This is confirmed by Southern blot analysis using SphI to digest the genomic DNA and an internal probe specific to the B-galactosidase region of the pGToLxf vector (probe designed by BayGenomics). Two bands were detected in RRT293, whereas only one band was observed in AJ0315 (Fig. 5B). We conclude that the embryonic lethality and the unusual TRD observed for RRT293-derived mice most likely resulted from interruption of two genetic loci, Mst1 and a second as yet unidentified locus.

Figure 5.

RRT293 mice carry two interrupted genes. (A) DNA-FISH analysis of RRT293 ES cells using Cy3-labeled gene-trap vector as a probe (shown in red). A probe for Xist DNA (shown in green) was used as a positive control. The white arrows indicate the pinpoints (red) corresponding to the location of the gene-trap vector within nuclei. A total of 51 nuclei were counted, and 29 of these nuclei contained 2 red spots (pGToLXF) together with 1 green spot (X-chromosome). (B) Southern blot analyses of SphI-digested genomic DNA from RRT293 ES, AJ0315 ES, and wild-type mice using an internal probe for B-galactosidase. Note that the Mst1 gene trapped band is present in AJ0315 and RRT293 but not in wild-type (negative control). The second-site insertion is evident in RRT293.

Conclusions

Here, we have generated and characterized two different Mst1-deficient mouse models with very different phenotypes. Animals derived from the RRT293 clone carried two genetic disruptions, one in Mst1 and the other in an unknown genetic element. This combination resulted in embryonic lethality for homozygous individuals and a form of TRD exhibiting three unusual characteristics: (i) It favors heterozygotes; (ii) it shows a generational effect resulting in stronger representation of the mutated allele in each subsequent generation; and (iii) the wildtype allele is extinguished after three generations. Paradoxically, the mutant allele favored for transmission is lethal in the homozygous state.

By contrast, AJ0315 mice are viable and fertile, displaying only mildly reduced fertility in the homozygotes. The mice also show very mild TRD present only in the first generation and apparently only among the first-time moms. The mild phenotype suggests functional redundancy with other members of the Mst family. Because the AJ0315 Mst1-deficient mouse model did not recapitulate the RRT293 phenotype, we conclude that Mst1 alone cannot be responsible for the observed TRD or inviability.

The RRT293 mouse provides an excellent opportunity to investigate the basis of TRD and “meiotic drive”, the phenomenon in which specific alleles are unequally or ‘selfishly’ passed on to subsequent generations. We do not presently know whether this form of TRD is an example of true meiotic drive (Ruvinsky, 1995; Pardo-Manuel de Villena and Sapienza, 2001b; Pardo-Manuel de Villena and Sapienza, 2001c; Pardo-Manuel de Villena and Sapienza, 2001a; de La Casa-Esperon et al., 2002), as we do not yet know whether the segregation distortion is present from the time of conception (as observed for the Om locus (Pardo-Manuel de Villena et al., 2000; Pardo-Manuel De Villena et al., 2000; Wu et al., 2005; Bell et al., 2006) or occurs as a result of unequal embryo viability. Nor do we know whether the TRD results from a non-Mendelian segregation of the chromosomes during meiosis or whether, instead, gametic dysfunction causes the unequal recovery of the disrupted alleles. TRD systems often involve driver genes, which exhibit biased expression, and suppressor genes, which function to keep driver genes in check and prevent biased inheritance (Schimenti, 2000; Pennisi, 2003). At the molecular level, deletion of Mst1 and the second genetic element may have disrupted expression of a suppressor gene, unleashing a selfish driver gene resulting in the observed TRD favoring the mutated allele(s).

The favored inheritance of the mutated allele(s) is extraordinary for another reason: How can an allele be favored when its homozygosity causes embryonic lethality? Experiments are underway to rederive animals from the RRT293 clone, in an effort to recover from an unfortunate total-colony loss. Its re-derivation will enable the identification of the second-site insertion and improve our understanding of the various causes of TRD -- in particular, one that favors the loss of function of an apparently essential locus.

EXPERIMENTAL PROCEDURES

Characterization of ES cell lines and generation of mice lacking Mst1

E14Tga ES cell lines (derived from129P2/OlaHsd mice) containing disruptions in Mst1 were obtained from BayGenomics (clone RRT293) and from the Sanger Institute Gene Trap Resource (clone AJ0315). The position of splice acceptor site within the Mst1 region for both clones was determined using PCR. For the BayGenomics RRT293 ES cell line, the gene trap vector inserted between exons 4 and 5 of Mst1. The position was verified using genomic DNA and Mst1primers MST1_2L (5′-GCCTTCTGTGTTTAGCAA) and MSTExon 4D (5′-AGAGCCTCAGGTAGATAG) along with a primer from the En2 intron region of the gene trap vector, MST1_V1L (5′-CTGTCCCTCTCACCTTCTAC). The position of the gene trap insertion was also verified by Southern blot using BamHI-digested DNA and an external 5′ (relative to the insertion site) probe. The 5′ probe was PCR-labeled using primers MST1_894F (5′-CTCATGATAGAACCAATCTTC) and MST1_1710R (5′-CCTGGCTTTTATTGTTTACTC), which detects a wildtype band at 4.7 kB and a mutant band at 3.6 kB.

The position of the gene trap vector for ES cell line AJ0315 was also determined using both PCR and Southern blots. The PCR primers Mst1-3p-9942U (5′AAAAAGAGCAAATGAATGA), Mst1-3p-9922U (5′-ATTTCTGGTGTGGCTTTGT), and Mst1-3p-10326L (5′-GCATATGGTCAAGTCAGTC) were used to map the location of the splice acceptor site using PCR. This location was verified by Southern blot using BamHI-digested genomic DNA and an external 5′ probe, generated using primers Mst1-5p-2878U (5′-GGGAAGGCTTGTGTTGAAT) and Mst1-5p-3340L (5′-TAGGGCGCAGTCAAGAAAC). The probe was labeled using PCR, and detects the wildtype allele at 10.6 kB and the mutated allele at 7.8 kB. Both ES cell lines were injected into C57Bl/6J blastocysts to obtain chimeras. Two male chimeras were obtained for each clone injected, and they were mated to C57BL/6J females for germline transmission. Animals were outcrossed twice to C57BL/6J mice. Analysis of the targeted mutation was carried out on animals of a mixed 129/Sv-Ola-C57BL/6J background.

Genotyping mice

Mice were weaned between 3-4 weeks of age, and genotyped by genomic DNA Southern hybridization using an external 5′ probe and also by PCR on genomic DNA prepared from tail clips. PCR genotyping of animals derived from the RRT293 cell line was carried out using primers Mst1_V1L, Mst1_2L, and MSTExon 4D. Southern blot analysis was performed as described above using primers MST1_894F and MST1_1710R to label the 5′ external probe. Animals derived from the AJ0315 cell line were also genotyped both by Southern blot using an external 5′ probe and by PCR using primers Mst1_2F (5′-GGTGCTGCCTTATTCGTTGT), Mst1_2R (5′-GCCTTCCACCACACCATTAG), and GVR1 (5′-CACTCCAACCTCCGCAAACTC). The primers used for construction of the 5′ external probe were Mst1-5p-2878U and Mst1-5p-3340L. Genomic DNA was digested with BamHI, which generates a wildtype band at 10kB and a mutant band at 8 kB. Primers pGToLxf 2444F (5′-TTATCGATGAGCGTGGTGGTTATGC) and pGToLxf 3124R (5′-GCGCGTACATCGGGCAAATAATATC) were used to generate an internal probe for B-galactosidase for hybridization to the gene-targeting vector pGToLxF. For Southern analysis using an internal probe to the B-galactosidase region, genomic DNA was digested with SphI.

Western blot analysis

ES cells and tissues from wildtype, heterozygous, and homozygous littermates were sonicated three times (10 seconds each) in a lysis buffer containing 10 mM Trish pH 7.5, 150 mM NaCl, 5 mM EDTA, 1% Triton X-100, 20 mM 2-mercaptoethanol, and 1 mM PMSF. Protein lysates (15 μg/lane) were run on a 10% SDS-PAGE gel, then transferred to a PVDF membrane. The membrane was blocked with 5% nonfat dry milk in PBS containing 0.1% Tween-20, then incubated with Mst1 antibody (diluted 1:2500; Ref. (Praskova et al., 2004)) at 4°C overnight. The next day the membrane was washed four times with PBS/Tween, then incubated with anti-rabbit horseradish peroxidase-labeld secondary antibody (diluted 1:20,000) for 1 hour at room temperature. The membrane was washed again with PBS/Tween, then visualized using the West Pico Chemilluminescence system (Pierce).

Northern analysis of Mst1 transcripts

RNA (20 ug/ lane) was isolated from RRT293 cells and the liver of wildtype, heterozygous, and homozygous Mst1 animals (from AJ0315 ES cells). The RNA samples were separated on denaturing 1% agarose-formaldehyde gels for 3 hours at 100V, then transferred to a Hybond-XL membrane overnight. The membrane was pre-hybridized using UltraHyb or UltraHyb oligo buffer (Ambion) for 1 hour at 42°C, followed by overnight hybridization of the ³²P-radiolabeled probe. Two different probes were used in order to detect either the full-length transcript (a random-labeled probe specific to the 3′-end region of Mst1) or the truncated transcripts (two oligo probes (5′-CCCTCTCCAAGCTTCTCTAA, 5′-CTGCGCGGTGGGTCCTCAG) that were end-labeled with PNK). The following morning, the membranes were washed twice with low-stringency buffers (2X SSC/0.1% SDS) followed by two washes with high-stringency buffers (0.1X SSC/0.1% SDS), at 42°C. The membranes were visualized by exposing to a phosphoimaging screen for 1 day.

DNA Fluorescence In situ Hybridization

The RRT293 ES cells were cultured to 70% confluency, then harvested by trypsinization and washed with 1X PBS. The cells were cytospun onto clean glass slides for 10 min at 2000 rpm, then immersed in 1X PBS for 5 min, CSK buffer for 30 sec, CSK + 0.5% Trition for 1 min, and CSK buffer for 1 min (all buffers were ice-cold). The slides were fixed with 4% paraformaldehyde for 10 min at room temperature, then stored in 70% ethanol at 4°C. In preparation for DNA-FISH hybridization, the slides were incubated 2 min each in 70%, 80%, 90%, 100% ethanol, then treated with RNase (200 ug) for 40 min at 37°C. The enzyme was inactivated via incubation with 0.2N HCl + 0.5% Tween-20 for 10 min, then the slides were denatured using 70% formamide + 2X SSC at 80°C for 10 min, followed by quenching in the ethanol series. The gene-trap vector (pGToLxf) was labeled using dUTP-Cy3 (GE Healthcare) and the positive control probe (the Xist gene, for detection of the X-chromosome) was labeled using dUTP-FITC (GE Healthcare), both using a Nick Translation kit (Roche). For hybridization, 150 ug of labeled pGToLxf and 100 ug of Xist probe were added to each slide, then hybridized overnight at 37°C. The following day the slides were washed two times (5 min each) using 2X SSC + 50% formamide at 45°C, followed by two washes of 2X SSC for 5 min each. The slides were mounted using DAPI-Vectashield, and images were taken using a Nikon Eclipse 90i microscrope and processed using Volocity 4.2 software.

Animal matings and statistical analysis

After heterozygous animals (sired by the chimeric males) were outcrossed to C57BL/6 animals, the resulting heterozygous animals were mated to each other, and pups were designated as the second generation. The second generation, non-littermate, heterozygous animals were mated, resulting in the third generation. These intercrosses were extended to the fourth generation for animals derived from the RRT293 cell line and to the fifth generation for animals derived from the AJ0315 cell line. Backcrosses (mating between homozygous and heterozygous animals) were carried out using second generation animals. Outcrosses between homozygous animals from clone AJ0315 and Mus castaneus and C57BL/6J animals were performed using third generation homozygous null animals. Statistical significance for all matings was determined using the chi-squared test, comparing the total observed numbers (for each sex) of wildtype, heterozygous, and homozygous deleted Mst1 animals. The expected values for animals derived from the AJ0315 clone were determined by calculating the expected Mendelian ratios for each genotype (50% Mst1^+/- , 25% Mst1^+/+ and 25% Mst1^-/-) given the total number of pups. Chi-squared analysis for RRT293-derived animals took into account homozygous lethality, thus the expected Mendelian ratios utilized for the test were 66% Mst1^+/- and 33% Mst1^+/+ for the total number of pups.