The C57BL/6J Mouse Strain Background Modifies the Effect of a Mutation in Bcl2l2

Bcl2l2 encodes BCL-W, an antiapoptotic member of the BCL-2 family of proteins. Intercross of Bcl2l2 +/− mice on a mixed C57BL/6J, 129S5 background produces Bcl2l2 −/− animals with the expected frequency. In contrast, intercross of Bcl2l2 +/− mice on a congenic C57BL/6J background produces relatively few live-born Bcl2l2 −/− animals. Genetic modifiers alter the effect of a mutation. C57BL/6J mice (Mus musculus) have a mutant allele of nicotinamide nucleotide transhydrogenase (Nnt) that can act as a modifier. Loss of NNT decreases the concentration of reduced nicotinamide adenine dinucleotide phosphate within the mitochondrial matrix. Nicotinamide adenine dinucleotide phosphate is a cofactor for glutathione reductase, which regenerates reduced glutathione, an important antioxidant. Thus, loss of NNT activity is associated with increased mitochondrial oxidative damage and cellular stress. To determine whether loss of Bcl2l2 −/− mice on the C57BL/6J background was mediated by the Nnt mutation, we outcrossed Bcl2l2 congenic C57BL/6J (Nnt −/−) mice with the closely related C57BL/6JEiJ (Nnt +/+) strain to produce Bcl2l2 +/− ; Nnt +/+ and Bcl2l2 +/− ; Nnt −/− animals. Intercross of Bcl2l2 +/− ; Nnt +/+ mice produced Bcl2l2 −/− with the expected frequency, whereas intercross of Bcl2l2 +/− ; Nnt −/− animals did not. This finding indicates the C57BL/6J strain background, and possibly the Nnt mutation, modifies the Bcl2l2 mutant phenotype. This and previous reports highlight the importance of knowing the genetic composition of mouse strains used in research studies as well as the accurate reporting of mouse strains in the scientific literature.

Inbred C57BL/6J mice are widely used in biological and biomedical research and for this reason the strain was selected to be the first mouse genome sequenced by a public consortium (Waterston et al. 2002). The use of inbred mouse strains that are presumed to be genetically homogenous at all loci reduces variability during analysis of a defined genetic modification, increases experimental reproducibility between different laboratories, and facilitates genetic mapping of strain-specific effects. Genetic drift is a concern in maintaining inbred species, and programs have recently been developed to monitor the genetic status of commonly used inbred mouse strains at commercial breeding facilities (Taft et al. 2006). Before such monitoring, C57BL/6J mice in some production facilities developed significant genetic alterations. For example, both the Snca gene, which encodes alpha-synuclein, and the adjacent multimerin1 locus were mutated by a 365-kb deletion that arose spontaneously in the C57BL/6JOlaHsd strain in England some time before 1999 (Specht and Schoepfer 2001). Similarly, the C57BL/6J substrain of C57BL/6 has a deletion of 17.8 kb of the Nnt gene, which encodes nicotinamide nucleotide transhydrogenase (Huang et al. 2006). The Nnt C57BL/6J mutant allele arose spontaneously at The Jackson Laboratory in Bar Harbor, Maine, between 1976 and1984. The allele has an in-frame deletion of exons 7211 and a missense (M35T) mutation in the mitochondrial leader peptide sequence that results in reduced expression of Nnt mRNA and no functional NNT protein (Huang et al. 2006).
NNT is located in the inner mitochondrial membrane, where it functions as a redox-dependent proton pump that uses the proton gradient across the inner mitochondrial membrane to catalyze interconversion of nicotinamide adenine dinucleotide phosphate (NADPH) and NAD + from NADP + and NADH in the mitochondrial matrix (Earle et al. 1978;Pedersen et al. 2008). NNT activity has been estimated to account for approximately 45% of total production of NADPH, with the remainder coming from the pentose phosphate pathway, mitochondrial NAD(P)-malic enzyme, and NADP-isocitrate dehydrogenase (Sauer et al. 2004;Vogel et al. 1999). In the mitochondrial matrix, NADPH is a cofactor for glutathione reductase, which catalyzes conversion of oxidized glutathione disulfide to glutathione (GSH) (Dalton et al. 2004;Vogel et al. 1999). Replenishment of this antioxidant (GSH) is important to control reactive oxygen species (ROS) and cellular redox status (Dalton et al. 2004). Loss of NNT activity is associated with decreased NADPH, which in turn reduces the ratio of GSH/glutathione disulfide, thereby making the mitochondrial environment more susceptible to ROS-induced damage (Arkblad et al. 2005;Sheeran et al. 2010).
Hence, a prediction is that the Nnt mutation in C57BL/6J would render mice more sensitive to genetic or environmental factors that influence cellular stress. Indeed, the Nnt mutation acts as a genetic modifier, causing mice lacking the mitochondrial matrix localized superoxide dismutase 2 to display a more severe phenotype in which they die during embryogenesis (Huang et al. 2006;Kim et al. 2010).
Here we report the effect of introducing a mutation of Nnt on the phenotype of Bcl2l2 2/2 mice that lack a death-protecting member of the BCL-2 family of proteins.
BCL-2 proteins play a central role in controlling apoptosis (Cory and Adams 2002;Danial and Korsmeyer 2004;Taylor et al. 2008). Previously, we generated mice mutant for Bcl2l2, which encodes BCL-W, an antiapoptotic member of the BCL-2 family of proteins (Ross et al. 1998). Intercross of Bcl2l2 +/2 mice on either mixed 129B6 (Ross et al. 1998) or 129, FVB strain background (Print et al. 1998) produced Bcl2l2 2/2 mice with the expected frequency. In contrast, we show here that most Bcl2l2 2/2 mice on a congenic C57BL/6J (Nnt mutant) background die before or at birth. We predicted that mutation of Nnt in C57BL/6J mice modifies the Bcl2l2 mutant phenotype. To test this prediction, we introduced a wild-type allele of Nnt by outcrossing Bcl2l2 mutant mice on a C57BL/6J (Nnt 2/2) congenic background with the closely related C57BL/6JEiJ (Nnt +/+) strain, followed by intercrossing and genetic analysis of the ratio of Bcl2l2 2/2 offspring recovered on either a Nnt 2/2 or Nnt +/+ background. The results indicate that Bcl2l2 2/2; Nnt +/+ mice are born alive at the expected frequency whereas Bcl2l2 2/2; Nnt 2/2 mice are not. Hence, the mutated Nnt allele, or a closely linked mutation, in C57BL/6J mice acts as a modifier of the mutant phenotype of loss of Bcl2l2.

MATERIALS AND METHODS
Mice C57BL/6J (Nnt 2/2; cat. no. 000664) were purchased from The Jackson Laboratory (Bar Harbor, ME) in 1993 and have been maintained in our laboratory since then as a closed colony. The C57BL/ 6JEiJ (Nnt +/+, cat. no. 000924) substrain of C57BL/6J was established by Dr. Eva Eicher in 1976 at F129. Male C57BL/6JEiJ mice were purchased from The Jackson Laboratory in 2007 at F129+75. The ROSA41 strain has a mutant allele of Bcl2l2, encoding BCL-W (Ross et al. 1998). The mutation is generated by an insertion of the ROSA b-gal gene trap vector (Friedrich and Soriano 1991). The mutant allele is null for Bcl2l2 function (Ross et al. 1998). The ROSA41 mutation was initially produced on a mixed 129S5, C57BL/6J strain background (Friedrich and Soriano 1991;Ross et al. 1998), but has since (as mentioned in the section Crosses) been backcrossed to a congenic C57BL/6J background. Mice on this strain background are available from the Mutant Mouse Regional Resource Center (UC Davis). Mice were provided Purina 5020 and nonacidified tap water ad libitum. Mice were housed in Techniplast ventilated cages with Bed O'Cobs bedding (The Andersons, Inc, Maumee, OH) and cotton nestlets for enrichment, with a light cycle of 13-hr on, 11-hr off, with lights on at 0630. Experiments involving animals were approved by the UCI Institutional Animal Care and Use Committee.
Crosses B6129-Bcl2l2 GtROSA41Sor +/2 were backcrossed for 12 generations with C57BL/6J mice. To ensure that the Y chromosome and mitochondrial DNA were of C57BL/6J origin, at least one backcross was performed with a male or female C57BL/6J mouse respectively. N12Fx mice were intercrossed to generate the data in Table 1.

Statistical analysis
Differences between the expected and observed frequencies of animals of different genotypes were analyzed with a x 2 goodness-of-fit-test, and two-tailed Fisher's exact test via the method of summing small P values (Graph Pad Software, La Jolla, CA). Significance was defined as P # 0.05.

RESULTS AND DISCUSSION
Intercross of Bcl2l2 GtROSA41Sor +/2 on a mixed C57BL/6J, 129S5 strain background produced Bcl2l2 2/2 mice with the expected frequency (Ross et al. 1998). An independent laboratory reported the same result from intercross of Bcl2l2 +/2 mice on a mixed 129S1, C57BL/6J, FVB strain background (Print et al. 1998). In contrast, after backcross to a congenic C57BL/6J background, homozygous mutant Bcl2l2 GtROSA41Sor mice were recovered at a significantly reduced frequency (Table 1). Genotyping of dead animals found among newborn offspring revealed a significant increase in homozygous mutant animals (Table  1). Visual inspection, necropsy, and histology of dead animals provided no obvious insight into the cause of death of Bcl2l2 2/2 animals (data not shown).
Our data implicate the Nnt mutation in C57BL/6J mice as the genetic modifier responsible for loss of a proportion of Bcl2l2 2/2 animals. Mutation of Nnt increases the sensitivity of an animal to oxidative stress (Arkblad et al. 2005). Loss of BCL-W increases the likelihood that a cell (e.g. neurons, intestinal epithelial cells) will undergo apoptosis in response to cellular damage or oxidative stress (Courchesne et al. 2011;Pazyra-Murphy et al. 2009;Pritchard et al. 2000). Taken together, this suggests a model in which loss of NNT activity reduces the survival of BCL-W-deficient mice by increasing oxidative stress within cells in which BCL-W is normally expressed and that are essential for embryogenesis or early postnatal survival. However, whether the death of Bcl2l2 2/2, Nnt 2/2 mice is caused by increased apoptosis of a specific cell type remains unknown. At present we cannot exclude the possibility that other loci on mouse chromosome 13 linked to Nnt contribute to this effect. However, on the basis of the aforementioned model, it seems plausible that the Nnt mutation in C57BL/6J mice contributes at least some, if not all, of the modifier effect that enhances the impact of loss of Bcl2l2.
The C57BL/6J strain background also modifies the severity of the phenotype of mice lacking manganese super-oxide dismutase 2 (sod2) (Huang et al. 2006;Kim et al. 2010). Sod2 2/2 mice on a DBA/2J strain background are born at normal frequency but die at approximately postnatal day 8 as the result of severe acidosis. In contrast, most Sod2 2/2 on a C57BL/6J background die at approximately embryonic day 15 as the result of dilated cardiomyopathy. A QTL analysis of long-lived Sod2 2/2 mice generated from DBA/2J-Sod2 +/2 mice back-crossed to C57BL/6J (N7) mapped the modifier locus to a 10-Mb region on distal chromosome 13, which contains Nnt (Huang et al. 2006). Mutation of the Nnt locus in C57BL/6J mice has also been shown to be responsible for glucose intolerance and defective insulin secretion in C57BL/6J mice (Freeman et al. 2006).
These results reemphasize the importance of knowing the genetic composition of mouse strains used in studies, as well as accurately reporting mouse strains in the scientific literature (Stevens et al. 2007;Taft et al. 2006;Wotjak 2003). They also emphasize the particular importance of knowing the substrain of C57BL/6 mice used in studies involving analyzing of cellular stress and cell death. The latter point is illustrated by a recent study (Ni et al. 2008), which confirmed a report (Bouillet et al. 2001) of increased severity of a mutant phenotype of Bcl2 2/2 mice on a C57BL/6 background. Ni et al. 2008 described the C57BL/6 mice used as being from The Jackson Laboratory, although the authors did not specify which of the six sublines of C57BL/6 mice available at The Jackson Laboratory they used. The absence of such information hinders the present and future scientific community from fully interpreting a published report.
The advent of high-density single nucleotide polymorphism (SNP) genotyping platforms offers a powerful tool to monitor genetic stability within an individual sub-strain in a breeding facility (Petkov et al. 2004;Yang et al. 2009). Recently SNP mapping was used to identify a copy-number variation allele on chromosome 19 of C57BL/ 6J mice that affects expression of the Ide and Fgfbp3 genes and that is segregating in what were presumed to be inbred C57BL/6J mice (Watkins-Chow and Pavan 2008). Continued improvements in methods and reduced cost of SNP genotyping should ultimately facilitate routine genotyping of mouse strains used in research studies.

ACKNOWLEDGMENTS
We thank L. B. Rowe (The Jackson Laboratory) for communicating results before publication. Supported in part by grants HD-36437 and HD-45913 from the National Institutes of Health to G.R.M.