Results

Production of E2 gene knockout mice
To create E2 knockout mice, we used gene targeting in mouse ES cells to disrupt the E2 gene. The overall strategy for disrupting the E2 gene is illustrated in Fig. 1A. The gene targeting construct was designed to replace a 1.67 kb EcoRV/Smal genomic DNA fragment encompassing part of Exon 4 and all of Exon 5 with the PGKneo selectable marker cassette. Of 522 ES cell clones screened for targeting by Southern blot analysis, 29 (5%) displayed the predicted restriction fragment length polymorphisms indicative of gene targeting at the E2 locus. As illustrated in Fig. 1A and 1B, an E2 Exon 6 specific probe, which is external to the targeting construct, hybridizes to only a ~16 kb BglI restriction fragment from the wild type allele in the parental R1 ES cell line. In correctly targeted ES cells, this probe also hybridizes to a ~11 kb BglI restriction fragment. Targeting was confirmed with several additional restriction enzymes and probes (data not shown).
Correctly targeted ES cells were microinjected into blastocysts to produce chimeric mice. Chimerics were bred to C57BL/6J mice. Following germline transmission of the targeted allele, heterozygous mice were interbred to produce wild type (+/+), heterozygous (+/-) and homozygous (-/-) animals. Mice were genotyped by Southern blot analysis. The Exon 6 specific probe hybridized to only a ~16 kb BglI restriction fragment in +/+ mice, a ~11 kb BglI fragment in -/- mice, and to both of these fragments in +/- mice (Fig. 1B).
Mice homozygous for the E2 mutation were born at the expected frequency. Genotype analysis of pups derived from +/- by +/- matings revealed that +/+, +/-, and -/- mice were present at nearly the expected 1:2:1 frequency. Of the initial 60 animals genotyped, 19 (32%) were +/+, 27 (45%) were +/-, and 14 (23%) were -/-. Thus, the E2 gene was dispensable for normal embryonic development. However, as expected, nearly all homozygous mice died in the perinatal period. Immediately following birth, homozygous pups were indistinguishable from their +/+ and +/- littermates; they were vigorous, active and able to suckle. By mid to late day on postnatal day one, most -/- pups became moribund and were readily identifiable as they were lethargic, pale, and exhibited gasping respiratory movements. With few exceptions, -/- pups died within 72 hours of birth. We have observed one rare -/- pup that survived to postnatal day 13. The reason for the prolonged survival of this pup is unknown.

E2 deficient mice accurately model classic MSUD
To demonstrate that the gene targeting event created a true E2 null allele, we determined BCKDH activity in liver homogenates of postnatal day 1 mouse pups derived from +/- by +/- mating pairs. As shown in Figure 2A, the BCKDH activity in +/+ mice was readily detectable. In marked contrast, BCKDH activity was completely absent in -/- mouse livers. As expected, homogenates from +/- mice had approximately half the activity of their +/+ littermates (Figure 2A). The results from -/- mice are similar to that observed in humans with classic MSUD [1].
Figure 2  Biochemical characterization of the classic MSUD murine model. A, BCKDH enzyme activity in liver of newborn wild type control (+/+), heterozygous (+/-), and homozygous (-/-) knockout mice. Enzyme activity was significantly reduced in +/- liver compared to +/+, and was below the level of detection in -/- liver. B, Total BCAA concentrations in blood of mice. Total BCAA represent the sum of leucine, isoleucine, and valine. Total BCAA concentrations in blood from -/- mice were significantly elevated compared to +/+ and +/-. C, Ratio of total BCAA to alanine in blood of mice. This ratio was significantly elevated in -/- mice compared to +/+ and +/- mice. The numbers on the bars indicates the number of mice analyzed. *, Significantly different from +/+ (P < 0.001); **, significantly different from +/+ and +/- (P < 0.001). Immunohistochemistry with an E2 specific antibody was used to examine E2 protein in the mice. As shown in Figure 1C and 1D, immunoreactive E2 protein was abundant in liver and embryonic fibroblasts of+/+ mice. In marked contrast, immunoreactive E2 protein was absent in these same tissues of -/- mice.
Homozygous E2 knockout mice had a nearly 3-fold increase in blood (Figure 2B) and urine (data not shown) levels of BCAA (sum of leucine, isoleucine, and valine) as compared to their +/+ littermates. Because amino acids were analyzed by tandem mass spectrometry, the sum of BCAA shown in Figure 2B also may include alloisoleucine that may have been produced in -/- MSUD mice.
The metabolism of BCAA is linked with the synthesis of alanine, glutamate, and glutamine [33]. Because of impaired metabolism of BCAA in MSUD mice, and to further characterize the abnormal biochemistry in this model, we analyzed the blood levels of the alanine, glutamate, and glutamine. As shown in Table 1, the levels of all three amino acids in homozygous mice were markedly lower than the levels in +/+ or +/- mice. The levels of these amino acids in the +/- mice were comparable to those in +/+ mice (Table 1). Because of the abnormal decrease in the blood alanine level and marked rise in blood BCAA levels that are characteristic of MSUD, a recent report on MSUD patients has suggested that the ratio of BCAA/alanine provides a more sensitive measure of the abnormal biochemistry of MSUD than BCAA level alone [8]. Therefore, we also expressed the amino acid results as BCAA/alanine ratio. As shown in Figure 2C, this ratio in -/- pups was more than 6-fold higher than +/+ or +/- littermates. These blood amino acid results are consistent with the concentrations seen in patients with MSUD [1,8,15].
Table 1  Summary of additional blood amino acid levels in the classic MSUD model and control littermates as determined by tandem mass spectrometry. All samples were collected on the day of birth. All values are mean +/- SEM.
MSUD Genotype  Alanine  Glutamate  Glutamine
(μmole/L)
+/+  196.1 ± 23.1 (6)  229.0 ± 12.0 (6)  76.8 ± 3.0 (6)
+/-  167.6 ± 9.9 (13)  251.0 ± 12.8 (13)  72.3 ± 1.6 (13)
-/-  94.6 ± 11.6* (5)  107.8 ± 4.8* (5)  50.4 ± 2.9* (5)
*: Significantly different from +/+ and +/- (P < 0.001). In summary, E2 knockout mice lack BCKDH enzymatic activity, E2 immunoreactivity, and have markedly elevated levels of BCAA in the blood and urine. These metabolic derangements ultimately result in neonatal lethality. These phenotypes are remarkably similar to that observed in humans with the classic form of MSUD. Thus, E2 knockout mice closely model classic MSUD.

Knockout mice expressing a human E2 transgene model a variant form of MSUD that mimics intermediate MSUD
To create a murine model of the intermediate variant form of MSUD, we used a transgenic strategy to rescue the severe elevation of BCAA and neonatal lethality that occurs in the classic MSUD mouse model. Our strategy was composed of a two part transgenic system to express human E2 in liver on the E2 knockout background. This bi-transgenic system consisted of a LAP-tTA transgene and a TRE-E2 transgene (Fig. 3A). The LAP-tTA transgene [34] directs high levels of liver specific expression of the tetracycline-controlled transactivator (tTA), a transcription factor that stimulates expression of promoters that harbor a transactivator response element (TRE). The TRE-E2 transgene was designed to express a human E2 cDNA from a TRE containing minimal promoter upon stimulation by tTA.
Figure 3  Transgenic mouse production and characterization. A, Transgenic strategy used to produce mice that express human E2. LAP-tTA transgenic mice have been previously described [34]. These mice express the tetracycline-controlled transactivator (tTA) from the liver specific LAP promoter. The TRE-E2 transgene contains the tetracycline response element (TRE) as part of the promoter, a synthetic intron (thin line), the human E2 cDNA, an alanine spacer, a c-myc epitope tag, and SV40 derived polyadenylation sequence. This construct was used to create several lines of transgenic mice. B, Western blot analysis of E2 protein in liver of control and intermediate MSUD mice. Note that the amount of human E2 protein (predicted MW ~54 Kd) in mice from Lines A and 525 A was variable but in many of the animals the amount was similar to the amount of mouse E2 (MW=~47 Kd) in control animals. Re-probing with a c-myc tag antibody confirmed the presence of the transgene derived, c-myc tagged, human E2 in transgenic mice but not in controls. Western blot analysis of brain (C), kidney (D), and muscle (E) revealed negligible amounts of transgene derived E2 in those tissues. All blots were re-probed with an actin antibody to allow amount of protein loaded in each lane to be compared. LAP-tTA mice were previously produced and characterized by the Bujard laboratory [34]. Pronuclear microinjection was used to produce transgenic mice that harbored the TRE-E2 transgene. From injections at the University of Pittsburgh, 2 transgene positive founders were identified that led to the generation of 3 different transgenic lines (Lines A, B, & D). From the injections at the University of Cincinnati, we obtained 8 transgene positive founders. Subsequent breeding of these founders revealed that many of the animals had multiple transgene insertion sites that segregated. We established a total of 15 different transgenic lines from these founders. Limited resources allowed us to only focus on a total of 9 of these lines. These lines differed substantially in transgene copy number as compared by Southern blot analysis of tail DNA (data not shown).
Transgenic TRE-E2 mice from each line (either founders or F1 offspring) were crossed to mice that were positive for the LAP-tTA transgene and were heterozygous for the E2 knockout. Ultimately, breeding pairs were used in which the mice were heterozygous for the LAP-tTA transgene, the TRE-E2 transgene, and the E2 knockout. To efficiently screen for the ability of the transgenes to rescue the knockout from neonatal lethality, we genotyped litters at weaning. This breeding strategy is expected to result in a theoretical maximum of animals with the rescue genotype (i.e., homozygous E2 knockout and positive for both transgenes) of 14%. Fig. 4A shows the percentage of mice alive at weaning with the rescue genotype from each transgenic line tested. Considerable variability was observed between lines. The line with the highest percentage of rescue mice at weaning was line 525 A. Approximately 10% of weaned, surviving pups from this line had the rescue genotype. Thus, the 525A transgene appears to be highly effective at rescuing the neonatal lethal phenotype of the E2 knockout. In contrast, line 520B completely failed to rescue the neonatal lethal phenotype. All other lines tested produced surviving rescue animals at a frequency between 1 and 5% of weaned pups.
Figure 4  Characterization of the intermediate MSUD murine model. A, Survival analysis of transgenic rescue lines. Presented are percentages of mice alive at weaning from each transgenic line tested that had the rescue genotype (i.e., homozygous for knockout of endogenous E2 and positive for both the LAP-tTA and TRE-E2 transgenes). Also plotted is the theoretical maximum frequency at which this genotype is expected in this population of animals. The numbers on the bars indicates the numbers of observations for each line. B, Ratio of total BCAA to alanine in blood of mice from controls and Lines A and 525A. Total BCAA represent the sum of leucine, isoleucine, and valine. BCAA/alanine values were significantly greater for Lines A and 525A compared to controls at all ages tested. The numbers on the bars indicates the numbers of samples analyzed. *, P ≤ 0.01; **, P < 0.005. C, BCKDH enzyme activity in liver of control and Lines A and 525A mice. *, P ≤ 0.001; **, P ≤ 0.0001. D, Survival curves for mice from Lines A and 525A. Rescue mice alive at weaning were monitored until they were moribund and subsequently sacrificed or were found dead in their cages. Lines A and 525A were selected for detailed characterization. Fig. 4B shows the results of blood amino acid analysis presented as a ratio of total BCAA to alanine. The BCAA/alanine values for mice from both lines were significantly elevated compared to controls for all of the time points analyzed. Note that the BCAA/alanine values for the transgenic mice are intermediate between controls and knockouts (see Figure 2C). Blood levels of alanine, glutamate, and glutamine were reduced in Line A mice, as was glutamate in Line 525A mice, compared to controls (Table 2).
Table 2  Summary of additional blood amino acid levels in the intermediate MSUD model and littermate controls as determined by tandem mass spectrometry. Samples were collected from mice that were 4–6 weeks of age. All values are mean +/- SEM.
MSUD Genotype  Alanine  Glutamate  Glutamine
(μmole/L)
control  225.2 ± 36.6 (11)  106.5 ± 10.5 (11)  67.8 ± 7.6 (11)
525A  175.0 ± 28.7 (10)  68.4 ± 7.7** (10)  58.0 ± 7.0 (10)
A  144.1 ± 16.9* (10)  65.9 ± 8.9** (10)  37.7 ± 6.5**,*** (10)
*: Significantly different from control (P < 0.05).
**: Significantly different from control (P < 0.01).
***: Significantly different from Line 525A (P < 0.05). Because the LAP-tTA mice that were used to drive expression of transgenic human E2 have been demonstrated to produce liver specific expression [34], BCKDH enzyme activity and production of human E2 protein in liver of Lines A and 525A was examined. BCKDH enzymatic activity in liver from Lines A and 525A was ~6 and 5%, respectively, of the enzymatic activity present in control liver (Fig. 4C). As shown in Fig. 3B, the amount of human E2 (predicted MW~54 Kd) in these transgenic mice was quite variable between mice. In some of these transgenic mice, the level of human E2 was approximately equal to the amount of mouse E2 (~47 Kd) produced in liver of nontransgenic control mice. The observations that these near normal amounts of E2 protein result in only ~5–6% of normal BCKDH enzyme activity suggest that transgene derived E2 is functioning at a suboptimal level. It is probable that the c-myc tag at the carboxy terminus of the transgene derived human E2 interfered with enzymatic activity. This interpretation is consistent with previous studies which have revealed that the carboxy terminus of an analogous subunit of the pyruvate dehydrogenase complex is essential for subunit interactions [35] and insertion of a Hisx6 tag on the carboxy terminus of an analogous E. coli subunit interfered with normal subunit assembly [36]. It is also possible that human E2 was not fully functional when complexed with mouse E1 and E3 subunits. Human E2 shares ~88% identity to mouse E2 at the amino acid level. We also used western blot analysis to analyze expression of the E2 transgene in brain, kidney and muscle. As shown in Figure 3C, D and 3E respectively, expression of E2 is negligible in those tissues.
Long-term survival of the mice from these two transgenic lines that survived beyond weaning is plotted in Fig. 4D. Although survival data for control animals was not collected, it is readily apparent that survival of the rescue mice was compromised. By 16 weeks, all mice of Line A were moribund and humanely sacrificed, or were found dead in their cage. Survival of Line 525A appeared to be somewhat better. At 20 weeks of age, ~12% of mice (2 of 16) were still alive. These two rare survivors died at 40 and 60 weeks of age.
In summary, these surviving mice had BCAA/alanine ratios that were intermediate between controls and knockouts and they expressed ~5–6% of normal BCKDH enzyme activity in the liver. These phenotypic observations are remarkably similar to the clinical phenotype observed in MSUD patients with the intermediate form of the disease [1]. Thus, these rescue mice represent a very useful model of the intermediate MSUD phenotype.