It would be extremely advantageous to the analysis of disease mechanisms in the spontaneous mouse model of type 1 diabetes, the nonobese diabetic (NOD) strain, if genes in this strain could be modified in vivo using embryonic stem (ES) cells and homologous recombination. However, a NOD ES cell line with adequate germline transmission has not yet been reported. We report the development of highly germline-competent ES cell lines from the F1 hybrid of NOD and 129 for use in NOD gene targeting. Consequently, we developed ES cell lines derived from (NOD × 129)F1 × 129 backcross 1 mice, which were intercrossed to select for homozygosity of particular regions of NOD genome known to contain disease loci.
The nonobese diabetic (NOD) mouse is a valuable model of human type 1 diabetes, particularly since the most important gene, the immune reponse gene IAb, and its human orthologue DQB1 share at least one of the same causal variants. However, allelic variation of this gene is not sufficient to explain disease and other genes are involved. Identification and functional analysis of candidate genes would be greatly facilitated if they could be altered in vivo by homologous recombination. Unfortunately, derivation of embryonic stem (ES) cells from the NOD mouse has proved extremely difficult. In addition, the modification of genes in ES cells from other strains such as 129 can lead to ambiguous results due to the unpredictable influence of linked 129-derived genes on the development of diabetes following the backcrossing of the chimeras to the NOD background. Thus far there is only one report of an NOD ES cell line that has demonstrated germline competence (1), but both its rate of growth and the level of germline transmission are too low to enable its use in gene targeting. We have had previous success in deriving ES cell lines from refractory strains of mice using microsurgery to explant epiblast of blastocysts that have been subjected to implantation delay (2). Here we present the results of using these same techniques, first with the NOD mouse and then with (NOD × 129)F1 hybrid embryos.
Our initial attempts to derive NOD ES cells involved ovariectomizing NOD mice on the third day of pregnancy to initiate implantation delay, and recovering the delayed blastocysts 7 days later. However, the number of blastocysts recovered was extremely low. Of 22 pregnant females, 18 yielded none and the remaining 4 gave only 17 living blastocysts. In contrast, of 38 nonovariectomized NOD females, 36 were pregnant with an average of nine blastocysts. This suggests that blastocyst viability is severely compromised by delaying implantation in the NOD mouse.
In subsequent experiments, NOD conceptuses were explanted on the first, third, or fourth day of pregnancy and transferred to non-NOD host dams in which implantation delay was then induced by ovariectomy. Epiblasts from a total of 372 delayed blastocysts were placed in culture, and 9 of these gave rise to colonies of cells that had a typical ES cell morphology. However, such colonies appeared only gradually, increased in size and number slowly, and began to differentiate into fibroblast-like cells once they reached a reasonable size. The cultures proved very difficult to propagate; most could not be maintained beyond passage 4. Two of the nine lines, which were more robust than the others, were derived from embryos removed from the NOD mice on day 1 of pregnancy. However, even these began to differentiate spontaneously into endoderm-like cells by passage 9–10. Five of the nine lines were karyotyped and three were trisomic for all or part of chromosome 11, an autosomal trisomy that has previously been associated with accelerated ES cell growth (3,4). The two remaining lines were both normal, one being XY and one XX.
There is evidence that maternal diabetes has a deleterious effect on preimplantation development in the mouse and rat (5,6). Although the majority of NOD mice in this study were too young to exhibit overt diabetes, they would have experienced the early immunological and developmental events that lead to diabetes, including insulitis, which might have a detrimental effect on the ability of blastocysts to give rise to stem cells. To test this, we attempted to derive ES cells from a congenic strain of mouse, NOD.H2h4. These mice never become diabetic, although they may develop some intraislet insulitis by 7–10 months of age (7). When used at 5 weeks of age, they therefore have a reduced frequency of the underlying defects. A total of 62 epiblasts were explanted into culture but no ES cells were obtained (Table 1). There is also some evidence that the NOD strain has variability in apoptotic pathways (8,9), which might interfere with the generation of viable ES cells. Hence, we attempted to isolate ES cells from a second congenic strain, NOD.B6 Idd3 B10 Idd5R8 (10), in which chromosome regions of the NOD type that appear to influence apoptosis have been replaced by introgression of C57BL/6 or C57BL/10 regions (8,9). This NOD congenic strain is also highly resistant to type 1 diabetes and has significantly reduced insulitis. No ES cells were obtained from the 28 cultured epiblasts. We conclude that the diabetic condition of NOD mice per se is unlikely to be responsible for failure to obtain stem cells.
We next attempted to derive ES cells from ICR mice, from which the NOD strain was originally developed (11). As a positive control, we cultured delayed epiblasts from CBA mice concurrently, having successfully isolated CBA ES cells in a previous study (2). As is shown in Table 1, no ES cells were obtained from 45 ICR epiblasts, whereas three lines with the morphological appearance of ES cells were isolated from 17 CBA embryos. These results suggest that there are features of the genetic background of the ICR mouse and the related strain NOD that are inimical to the isolation and propagation of pluripotential ES cells. Similar findings for ICR were reported by Suzuki et al. (12).
An alternative route toward genetic targeting in NOD mice would be to use NOD hybrid ES cells. It would be necessary to screen clones and select only those in which the NOD chromosome had been targeted before injecting the cells into blastocysts. The chimeras would then be bred to NOD mice and offspring exhibiting donor coat color screened for the presence of the targeted construct. The 129 strain is the one that most readily gives rise to ES cells and might best be expected to overcome the refractory nature of the NOD strain.
Therefore, 129/Ola males were mated to NOD females and blastocysts were transferred to ovariectomized host dams on the fourth day of pregnancy. Of 16 epiblasts explanted, 14 gave ES cell lines, resulting in a success rate of 88%. All lines were karyotypically normal, with nine being male and five female. A marked difference in the size of the C-band of chromosome 8 between the NOD (large band) and 129 (small band) was evident in the F1 lines (Fig. 1), which could help to assess ES cell contributions in chimeras. The successful derivation of these lines also shows that residence of conceptuses in the NOD reproductive tract for 4 days clearly did not inhibit ES cell isolation.
Cells from each male line were injected into blastocysts from MF1 or ICR mice. Every line gave rise to chimeras, the proportion of which ranged from 31 to 100% (Table 2). The male chimeras were mated with MF1 females to test for germline transmission. All nine lines were transmitted through the germline, with between 50 and 100% of male chimeras giving rise to offspring of ES cell genotype. Germline transmission was also observed through a female chimera in two of the lines. The proportion of donor-type pups varied between 7 and 83%.
Since we have succeeded in deriving (NOD × 129)F1 hybrid ES cell lines that show a high level of chimerism and give a good rate of germline colonization, it will be possible to use these lines for genetic targeting of the NOD mouse genome. In most chromosome regions examined, sufficient sequence diversity exists between the 129/Ola and NOD genomes (data not shown) to allow the straightforward determination by DNA blotting and restriction enzyme mapping of which chromosome, NOD or 129, has been targeted with a vector. In addition, there is the potential to use the F1 hybrid lines to examine the loci that may control the ability of 129, as opposed to ICR, blastocysts to give rise to ES cells.
Our success at generating F1 ES cell lines indicated that it should be possible to generate ES cells from (129 × NOD)F1 × 129 backcross 1 mice subsequently intercrossed to fix to homozygosity particular regions of NOD genome (Table 3). We therefore selectively bred five such strains having NOD chromosome segments containing diabetes-susceptibility alleles on chromosomes 1, 3, 4, and 17 (13). To facilitate backcrossing after genetic manipulation of the ES cells, NOD homozygosity was maintained for at least 10 Mb flanking each selected Idd region. To date we have isolated five ES cell lines from the chromosome 17 congenic strain, two cell lines from the chromosome 4 strain, and one cell line from the chromosome 1 congenic strain. The latter, which is homozygous for NOD genome in the Idd5.2 region, is a karyotypically normal male line which, when injected into MF1 blastocysts, gave a rate of chimerism of 60%. Six of nine male chimeras demonstrated germline transmission, and the proportion of pups with ES cell genotype varied between 36 and 100%. We are currently testing for germline transmission of the remaining lines and are isolating new lines from congenic strains containing the Idd regions 5.1, 10, and 18 (Table 3).
The availability of ES cells that are homozygous for particular regions of the NOD genome will simplify the targeting procedure as compared with the (NOD × 129)F1 hybrid ES cells. Subsequent backcrossing to the NOD parental strain of the targeted genome region, with monitoring of flanking regions with genetic markers to ensure linked genes remain NOD derived, will ultimately produce NOD animals that possess a gene modified in the context of the NOD genome. With marker-assisted backcrossing of the chimeras to the NOD strain, determination of the effect that a given genetic modification has on disease development should be accomplished after four to five backcross generations and subsequent intercrossing. Apart from providing additional flexibility over and above the use of the conventional C57BL/6 ES cell line, the chromosome 17/major histocompatibility complex/Idd1 congenic ES cell line will be essential because there are a number of tightly linked disease loci in this region of chromosome 17. We have not demonstrated that these lines are germline competent after gene targeting, but we are keen to make the lines available for such experiments to be attempted.
RESEARCH DESIGN AND METHODS
Natural matings were made between mice of the strains NOD (Taconic, Germantown, NY), NOD.B6 Idd3 B10 Idd5R8 (N12F2) (10), ICR, and CBA. Implantation was delayed by ovariectomizing females in the afternoon of the third day of pregnancy and injecting them subcutaneously with 1 mg of Depo-Provera (Pharmacia and Upjohn, Milton Keynes, UK). Mice of strains NOD and NOD.H2h4, aged 5 weeks, were superovulated and embryos were explanted on days 1, 3, or 4 (NOD) or day 4 (NOD.H2h4). Embryos were transferred to the uterus of strain PO (Pathology, Oxford, UK) mice that were subsequently ovariectomized as above. Females of the NOD strain were mated with 129/Ola males, and blastocysts were explanted on day 4 and delayed in PO mice as above. Mice were kept either on a 12-h light/dark cycle in which the dark period was from 1900–0700 h or on a 14-h light/10-h dark cycle in which the dark period was from 1300 to 2300 h. Delayed-implanting blastocysts were recovered 1 week after ovariectomy.
Recovery and microdissection of blastocysts was carried out as described in Brook and Gardner (2).
Isolation and culture of ES cell lines.
The culture medium was Dulbecco’s modified Eagle’s medium supplemented with 15% FCS, and with 2-mercaptoethanol, antibiotics, nonessential amino acids, and nucleosides as described in Robertson (14). Recombinant murine leukemia inhibitory factor (LIF) (ESGRO; Chemicon International, Temecula, CA) was added at 1,000 units/ml. Epiblasts microdissected from delayed implanting blastocysts were seeded onto feeder layers of irradiated mouse primary embryonic fibroblasts and cultured for 6 days. Colonies were then picked, trypsinized, and plated into fresh 4-well plates of feeder cells. After 3–4 more days of culture, when ES cell colonies were visible, each well was trypsinized and its entire contents were replated in a 35-mm dish of fresh feeders. Lines were then expanded as described previously (14).
ES cell lines in growth phase were briefly subjected to a mitotic arrestant and chromosome preparations were made by standard methods (15). Preparations were G-banded, and up to 10 complete metaphases analyzed for sex chromosomes and normality. In cases of mosaicism, further cells were examined until a representative analysis had been obtained.
Production of chimeras.
Blastocysts were explanted on day 4 after natural matings of strain MF1 or ICR mice and 10–15 ES cells were injected into each blastocyst, as described by Bradley (16), but with a pipette whose tip had been bevelled by machine (Narishige, Tokyo) at an angle of 45o, as described by Gardner (17), but with omission of the third needle. Injected blastocysts were transferred to the uterus of pseudopregnant PO mice, which were allowed to litter naturally. To test for germline transmission, mice with coat-color chimerism were mated with strain MF1, which enabled the donor coat color to be distinguished from that of the host.
Production of congenic mouse strains.
129 Ola (Harlan Olac) males were crossed with female NOD (Taconic) mice. These F1 mice were backcrossed once to 129 Ola, and appropriate heterozygous mice were then intercrossed to generate N2F2 mice that were selected for NOD homozygosity at Idd regions on chromosomes 1, 3, 4, and 17. N2F4 embryos were used to generate ES cells.
We thank Diabetes U.K., the Juvenile Diabetes Research Foundation, the Wellcome Trust, and the Royal Society for support. We thank Ann Yates for help in preparing the manuscript and Andy Forkner, Colin Hetherington, and their colleagues for their expert care of our mice.
Address correspondence and reprint requests to Dr. Frances A. Brook, Mammalian Development Laboratory, University of Oxford, Department of Zoology, South Parks Rd., Oxford, OX1 3PS U.K. E-mail:.
Received for publication 31 July 2002 and accepted in revised form 23 September 2002.
C.J.L. is currently affiliated with Breakthrough Breast Cancer Research Centre, Institute of Cancer Research, London, U.K.
ES, embryonic stem; LIF, leukemia inhibitory factor.