Chromosomal translocations are crucial events in the aetiology of many leukaemias, lymphomas and sarcomas, resulting in enforced oncogene expression or the creation of novel fusion genes. The study of the biological outcome of such events ideally requires recapitulation of the tissue specificity and timing of the chromosomal translocation itself. We have used the Cre‐loxP system of phage P1 to induce de novo Mll–Af9 chromosomal recombination during mouse development. loxP sites were introduced into the Mll and Af9 genes on chromosomes 9 and 4, respectively, and mice carrying these alleles were crossed with mice expressing Cre recombinase. A resulting Mll–Af9 fusion gene was detected whose transcription and splicing were verified. Thus, programmed inter‐chromosomal recombination can be achieved in mice. This approach should allow the design of mouse models of tumorigenesis with greater biological relevance than those available at present.
Chromosomal translocations are involved in the formation of many types of tumours (Rabbitts, 1994). There are two major outcomes involving either enforced expression (activation) of oncogenes near the translocation breakpoints, or fusion of genes on each chromosome resulting in the production of a fusion mRNA and protein. Animal models have been used to create a situation in which the consequences of the chromosomal translocation can be studied to identify biological pathways controlled by these translocation genes. The use of homologous recombination to create a gene fusion in one allele of a target gene, as described for the Mll gene (Corral et al., 1996), comes close to the reconstruction of the gene fusion resulting from chromosomal translocations. However, this approach has a number of drawbacks such as dominant lethality in some transgenics (Heisterkamp et al., 1991). In addition, a transgenic or knock‐in experiment only produces the equivalent of one reciprocal translocation product.
An ideal model would be the de novo creation of chromosomal translocations during mouse development and within specific cell types. A potential approach to this has been described using the Cre‐loxP system in embryonic stem cells, whereby loxP sites introduced on two different chromosomes were used to facilitate inter‐chromosomal translocations by transient expression of Cre recombinase (Smith et al., 1995; van Deursen et al., 1995). These results suggest that inter‐chromosomal events might be possible between loxP sites introduced in the mouse genome if Cre recombinase were expressed during mouse development. We have applied this strategy to produce inter‐chromosomal recombination between Mll and Af9 (Mllt3) genes in mice. Translocations involving the MLL gene on chromosome 11, band q23, are found in >15% of acute leukaemias in man. The MLL gene spans the breakpoint region and >40 different translocations have been identified (Look, 1997). Each MLL fusion tends to associate with a particular leukaemia phenotype, a frequent fusion partner being AF9 involved in the reciprocal translocation t(9;11)(p22;q23) (Iida et al., 1993; Nakamura et al., 1993). We have used homologous recombination to introduce loxP sites into the introns of mouse Mll and Af9 genes in ES cells, corresponding to the breakpoint regions of human leukaemias. These mutations have been carried into the germ‐line of mice and we have found that inter‐chromosomal recombination occurs during development under the control of Cre recombinase. Thus, the Cre‐loxP system is powerful for the induction of a site‐specific inter‐chromosomal recombination event in mice. This should allow further studies of the role of translocation genes in cancer.
Orientation of mouse Mll and Af9 genes permits chromosomal translocation
The aim of this study was to evaluate the creation of chromosomal translocations of Mll and Af9 genes in mouse using the Cre‐loxP system. Our strategy for Cre‐loxP‐mediated recombination between murine Mll and Af9 is schematically outlined in Figure 1. First, loxP sites were targeted into the Mll and Af9 genes by homologous recombination in mouse ES cells, into the introns corresponding to the breakpoint regions in human leukaemia patients with t(9;11)(p22;q23) (Figure 1A). Cre‐mediated recombination was subsequently attempted either by transient expression of Cre recombinase in ES cells carrying both the Mll–loxP (Mlox) and Af9–loxP (Alox) targeted alleles, or in mice with both targeted alleles and transgenic for Cre recombinase [deleter mice (Schwenk et al., 1995)]. Successful recombination was assayed by genomic PCR analysis using Mll‐ and Af9‐specific primers (Figure 1B), and expression of the fusion gene by reverse transcription–polymerase chain reaction (RT–PCR) with Mll‐ and Af9‐specific coding region primers (Figure 1C). The sequences of genomic and mRNA junctions are shown in Figure 1D and E.
Before starting this analysis, it was necessary to determine the chromosomal location of mouse Af9(Mllt3) to ensure that inter‐chromosomal translocation between Mll and Af9 was feasible. The mouse homologue of MLL has previously been mapped to chromosome 9 (Ma et al., 1993). We mapped the murine Af9 gene to the central region of chromosome 4, by inter‐specific backcross analysis (E.C. Collins, N.G. Copeland, D.J. Gilbert, N.A. Jenkins, and T.H. Rabbitts, unpublished). Af9 co‐segregates with Ifna, located 42.6 cM from the centromere of chromosome 4, in a region of synteny with human chromosome 9, band p21–p22. Thus, insertion of loxP sites in the same relative orientation in the Mll and Af9 genes should allow a Cre‐dependent translocation between mouse chromosomes 4 and 9.
The targeted alleles of Mll and Af9 were made as shown in Figure 2. A loxP site was inserted between exons 8 and 9 of Mll (Figure 2A) and between the breakpoint exons, which we designated exon a and exon b of Af9 (Figure 2C). DNA from ES cell clones with targeted Mll was assessed by filter hybridization using a 3′ Mll probe and a hygromycin probe (Figure 2B). Similarly, Af9 gene targeting was assessed with a 3′ Af9 probe and with a puromycin probe (Figure 2D). Mice were derived from the Mll‐ and Af9‐targeted clones. In addition, a doubly transfected ES clone (MALX10) was used for transient expression of Cre recombinase as a preliminary assessment of inter‐chromosomal recombination between Mll and Af9 loxP‐targeted alleles. A pool of MALX10 ES cells was harvested after Cre transfection, and genomic DNA from 105 cells was tested for the presence of the Mll–Af9 translocation product by PCR (Figure 3A, left panel). A band of ∼364 bp was observed in the MALX10 cells expressing Cre (MALX‐Cre) but not in MALX10 cells or CCB control PCR reactions. The MALX‐Cre fragment was subcloned for sequence analysis and a nucleotide sequence was obtained corresponding to the genomic junction expected for a Cre‐mediated recombination between loxP sites in the Mll‐ and Af9‐targeted alleles (Figure 1D). The presence of the reciprocal translocation product was also confirmed by PCR using puromycin and hygromycin‐specific primers (data not shown) indicating that inter‐chromosomal recombination had indeed occurred. The transient Cre expression assay was repeated and in this case RNA was extracted for evaluation of the generation of in‐frame Mll–Af9 fusion mRNA by RT–PCR. A 301 bp fragment was obtained from the MALX‐Cre cDNA, but was absent from the MALX10 cDNA and CCB control cDNA (Figure 3A, right panel). The MALX‐Cre RT–PCR fragment was sequenced and the resulting data showed the Mll–Af9 mRNA junction expected from fusion of Mll exon 8 with Af9 exon b (Figure 1E).
Mll–Af9 gene fusion during mouse development mediated by Cre recombinase
The PCR results on the ES cells carrying both Mll and Af9 alleles with loxP sites show Cre‐mediated recombination is possible between chromosomes 4 and 9, consistent with chromosomal translocation having taken place. The ability of this recombination event to occur in mice was assessed using mouse strains with germ‐line transmission of the Mlox and Alox alleles (designated Mlox–Alox mice). These mice were crossed with transgenic mice expressing Cre recombinase under the control of a ubiquitous promoter [deleter mice (Schwenk et al., 1995)]. The resulting strains were inter‐bred to generate a pool of animals carrying both the Mlox and Alox alleles and the deleter transgene (designated Mlox–Alox–Cre mice). The possibility that inter‐chromosomal translocations had occurred was assessed in Mlox–Alox–Cre and Mlox–Alox mice between the ages of 6 and 8 months. Genomic DNA was extracted from a number of tissues and PCR amplifications were performed (Figure 3B). A PCR product (∼364 bp), corresponding to inter‐chromosomal recombination between Mll and Af9, was visible when using brain DNA as a template but not in any other tissue DNA. The sequence of this product was obtained to confirm its origin from the loxP recombination (data not shown). The occurrence of the recombination product in brain implies that the Mll and Af9 genes are accessible for inter‐chromosomal Cre–mediated recombination and, although RT–PCR results were only semi‐quantitative, the levels of Cre enzyme in brain appear compatible with these events (see below).
The presence of the fused chromosomal segment between Mll and Af9 genes in the Mlox–Alox–Cre mouse was confirmed by analysing expression of the in‐frame fusion mRNA by RT–PCR using cDNA templates prepared from the various tissues. This yielded similar results to the genomic PCR. An RT–PCR product from Mll–Af9 mRNA was produced from the brain cDNA template of the Mlox–Alox–Cre mouse and not from that of the Mlox–Alox mouse (Figure 4A). All cDNA templates were verified with an actin‐specific RT–PCR (Figure 4C). The expression of Cre recombinase was confirmed in all tissue RNA samples using RT–PCR (Figure 4B). Cre expression appeared highest in the brain.
The presence of Cre‐mediated recombination between Mll and Af9 chromosomes was investigated using brain RNA samples from 10 further Mlox–Alox–Cre mice compared with two Mlox–Alox control mice. All the brain cDNA templates from the Mlox–Alox–Cre mice generated the fusion mRNA product by RT–PCR amplification (Figure 5A, samples 1–10) but the Mlox–Alox controls did not (Figure 5A, lanes N1 and N2). The quality of cDNA templates was verified by amplification of an actin RT–PCR product (Figure 5B). Thus we conclude that inter‐chromosomal recombinations, presumably translocations, can be induced by the Cre‐loxP system in mice.
The Cre‐loxP system of phage P1 has been applied to induce site‐directed chromosomal translocations in mouse ES cells (Smith et al., 1995; van Deursen et al., 1995) resulting in the generation of ES cells constitutively carrying the induced translocation. We have now demonstrated that de novo Cre‐loxP‐mediated inter‐chromosomal recombination (presumably translocation) between non‐homologous chromosomes is possible in mice. This approach has potential applications in the field of chromosome engineering and should provide insights into the consequences of large‐scale DNA rearrangements in animals.
A limitation that remains is the apparent low efficiency of the inter‐chromosomal recombination. Presumably, the chance of synapsis between the two loxP sites, a prerequisite for Cre‐mediated strand exchange, is much lower for inter‐chromosomal than intra‐chromosomal events. Indeed, the deleter mice used in our experiments have previously shown efficient recombination between loxP sites in close proximity, in a wide variety of tissues (Schwenk et al., 1995). However, the cre gene appears to be only weakly transcribed in these mice. Our semi‐quantitative RT–PCR results suggest stronger cre expression in the brain than in other organs, which may account for the presence of the recombination signal in the brain samples only. Nonetheless, the occurrence of inter‐chromosomal recombination may also be influenced by the accessibility of the target genes, which in turn may depend on their level of transcription in a particular organ. While the expression of mouse Mll is ubiquitous (Yu et al., 1995), our results with Af9 gene expression show strong expression during brain development (E.C. Collins and T.H. Rabbitts, unpublished), which may also partly explain the propensity for rearrangements in brain rather than in other tissues.
An important potential use of in vivo Cre‐mediated chromosomal translocations is in the study of the specificity and biological role of translocation genes. In particular, the in vivo function of fusion genes may be studied by this approach. The use of tissue‐specific promoters to control the site and timing of Cre expression, either by homologous recombination knock‐in or transgenesis, is a potentially powerful strategy. For instance, to gain insight into the role of MLL fusion partners in determining the tumour phenotype after 11q23 translocations, it would be useful to compare the consequences of the Mll–Af9 fusion in myeloid and lymphoid lineages. Induction of Mll–Af9 translocation in lineage‐committed cells or in progenitors would remove the obligate myeloproliferative advantage conferred by the constitutive Mll–Af9 knock‐in fusion gene (Dobson et al., 1999). The subsequent incidence of leukaemia of lineage‐specific phenotype would provide important evidence about the timing of translocation and role of the fusion proteins.
Finally, the Mll–Af9 knock‐in mouse model showed that the fusion gene alone is sufficient for the development of leukaemia (Corral et al., 1996; Dobson et al., 1999), but the situation in human patients is almost certainly more complex. The chromosomal translocation may have consequences ranging beyond the creation of one fusion gene; for example it has been shown that both PML‐RARA and RARA‐PML products are important in the phenotype of acute promyelocytic leukaemia in mice (Pollock et al., 1999). [PML‐RARA and RARA‐PML products are made from the reciprocal chromosomal translocations der(15;17) and der(17;15) where PML = promyelocytic leukaemia gene; RARA = retinoic acid receptor α gene.] In addition, embryonic lethality resulting from the expression of fusion genes may require creation of de novo translocations for a full read‐out of the pathogenic consequences of chromosomal rearrangements. Other factors, such as changes in chromatin structure and gene regulation near the breakpoints of the chromosomal translocation, may also be important for leukaemogenesis. The strategy of Cre‐loxP‐mediated recombination between chromosomes in specific mouse tissues should provide insights into these features.
A puromycin‐loxP cassette was obtained by inserting a 115 bp EcoRI–SalI fragment from PL2XR into pPGK‐puro. A loxP fragment from pPGK‐puro‐loxP was cloned into EcoRV of pPGK‐hygro to generate a loxP‐hygromycin cassette. Targeting vector pMll‐loxP‐hygro has a 5.5 kb EcoRI genomic fragment, containing the region of the Mll gene corresponding to the human breakpoint region, in the EcoRV site of pBluescript. The loxP‐hygro cassette was cloned into a genomic BglII site in Mll intron 8. HSV thymidine kinase (tk), from pMC1‐tk (Thomas and Capecchi, 1987), was cloned into the XbaI site. Targeting vector pAf9‐puro‐loxP has an 8.5 kb genomic SphI–EcoRV fragment, containing the breakpoint region of 129 mouse Af9, in the XbaI site of pBluescript. The puro‐loxP cassette was cloned into an XbaI site in the Af9 breakpoint intron (designated intron a). The tk fragment was introduced into the NotI site.
ES cells were transfected as described (Warren et al., 1994). CCB ES cells were grown on drug‐resistant mouse embryonic feeders during selection. pMll‐loxP‐hygro was introduced into ES cells grown on hygromycin‐resistant feeders (Johnson et al., 1995) to make Mlox cells and pAf9‐puro‐loxP into ES cells grown on puromycin‐resistant feeders [C57BL/6J‐TgN(pPWL88puro) 2Ems JR2355, E.M. Simpson, unpublished data] to make Alox cells. Finally, Mlox cells grown on puromycin‐resistant feeders were re‐transfected with pAf9‐puro‐loxP to make doubly targeted cells, MALX10. These were transiently transfected with Cre recombinase expression vector, pPGKCrebpA. Cells were grown for 72 h without selection and harvested for DNA and/or RNA extraction (MALX‐Cre).
DNA preparation and filter hybridization analysis.
DNA from ES cells and from mouse tissues was made using Puregene (Gentra Systems). Ten micrograms of DNA were digested with restriction enzyme and resolved on 0.8% agarose gels. After electrophoresis, the DNA was transferred to nylon membranes and hybridized to radioactive probes (LeFranc et al., 1986).
RNA extraction and cDNA preparation.
Extraction of total RNA was carried out with Trizol (Gibco‐BRL). cDNA was generated using SuperRT reverse transcriptase (HT Biotechnology). cDNA from 5 μg of total RNA was diluted to 100 μl.
Detection of Cre‐mediated translocations by PCR.
Genomic DNA (0.5 μg) was subjected to 35 cycles of PCR amplification with primers 1a (Mll intron primer 5′‐gtccccataacacccagagtagtg‐3′) and 1b (Af9 intron primer 5′‐cctcattctgacagaccagagcca‐3′) (Figure 1B) at 95°C for 1 min, 65°C for 1 min and 72°C for 1 min. Each PCR sample (1 μl) was used as a substrate for 35 further cycles, with nested primers 2a (Mll intron primer 5′‐gggcatgtagaggtaagacgcctg‐3′) and 2b (Af9 intron primer 5′‐atctccagggactgaatctagggc‐3′) (Figure 1B). PCR fragments were resolved on 2% agarose gels and cloned for sequence analysis.
RT–PCR was carried out as above, with 1 μl of cDNA as template, and primers 3a (Mll exon primer 5′‐ggagtccacaggatcagagtggac‐3′) and 3b (Af9 exon primer 5′‐tcaggatgttccagatgtttccag‐3′) (Figure 1B), and then with nested primers 4a (Mll exon primer 5′‐cctgacctctgtccccataacacc‐3′) and 4b (Af9 exon primer 5′‐gactgtggttttgtccagcgagca‐3′). RT–PCR with actin‐specific primers (5′‐atggccactgccgcatcctcttcc‐3′ and 5′‐cacgatggaggggccggactcatc‐3′) was carried out using 20 cycles as above. The touch‐down method was used from 63 to 55°C (Don et al., 1991) for RT–PCR with Cre‐specific primers (5′‐cgtatagccgaaattgc‐3′ and 5′‐caatcgatgagttgcttc‐3′).
We thank Andrew McKenzie for loxP and Cre plasmids, Anton Berns for pPGK‐hygro and Peter Laird for pPGK‐puromycin and pPWL88puro. We also thank Kevin A. Johnson, Charles P. Lerner, Thomas R. Ulrich, Craig A. Leach, Elizabeth R. Linnell and William C. Rafferty for development of mice for puromycin‐resistant feeders. E.C.C. was a recipient of an LRF Gordon Pillar Studentship. E.M.S. was supported by NIH grant 1R01MH/HD57465.
- Copyright © 2000 European Molecular Biology Organization