Genomic interval engineering of mice identifies a novel modulator of triglyceride production

A genomic DNA clone corresponding to Irf-1 was isolated from a 129SV genomic library (Stratagene) by using a probe generated by PCR with the following primers: 5′-AGAGCATAGCACTGGCCCTA-3′ and 5′-AGGCCTAGACTGGGGAGAAA- 3′. Vector ploxPNI was made by digesting ploxPneo-1 (11) with HindIII and religating (to remove a loxP site), and then digesting with BamHI and EcoRI, blunting the BamHI end with Klenow, and ligating with a 6.2-kb SmaI-EcoRI fragment from the 3′ end of Irf-1. PloxPNI was linearized with BamHI and was electroporated into R1 ES cells (12), and neo-resistant colonies were selected in the presence of 175 μg/ml of G418 (GIBCO/BRL). This construct deleted exon 10 of Irf-1, which is composed of the last 42 codons and the 3′UTR of the gene. A Csfgm genomic DNA clone was isolated from the 129SV genomic library by using a probe generated by PCR with the following primers: 5′-CCCTCTGAAAACGCTGACTC-3′ and 5′-CCATGACAGCATTGACTCTCA-3′. To create vector ploxPHTC, a 1.8-kb BamHI-SmaI fragment containing the PGK-hygromycin resistance gene was excised from pBS-PGKhyg/loxP/HPRT-D5′ (13) and was cloned adjacent to the HSV-tk gene in pPNT (14) digested with BamHI and EcoRI (blunted with Klenow). The 4.6-kb fragment containing PGK-hygromycin and HSV-tk in a head-to-tail arrangement was excised by BamHI and HindIII and was inserted into ploxPneo-1 digested with BglII and HindIII. The resulting vector, containing the cassette HSV-tk/PGK-hygromycin/loxP, was digested with HindIII and was ligated with a 7.6-kb HindIII genomic fragment located 2 kb downstream of Csfgm. PloxPHTC was linearized with XhoI, was electroporated in ES cells already targeted with the ploxPNI vector, and was grown under selection [180 μg/ml of hygromycin B (Sigma)]. Generation of deletion mice from the targeted ES cells was carried out as described (15). ES clones carrying the 450-kb deletion were injected into blastocysts of C57BL/6J mice. The males were bred to 129/SvEvTacfBR (Taconic Farms) to generate heterozygous mutant F1 mice.

Mice fed a regular mouse Chow diet were fasted overnight and were killed, and their livers and hearts were removed and homogenized. Lipids were extracted from the cell suspension as described (16) and were dissolved in PBS containing 5% Triton X-100. Quantitation of the triglyceride (without measuring free glycerol) and cholesterol levels were determined by using the commercially available enzymatic kits 337-B (Sigma GPO-Trinder kit) and 236691 (Boehringer Mannheim GmbH), respectively. For plasma lipid analysis, ≈100 μl of blood was collected by tail bleeding from mice fasted overnight. The concentrations of triglycerides (corrected for free glycerol) and total cholesterol in plasma were determined as described. For plasma-free fatty acids determination, 50 μl of blood was drawn in chilled paraoxonized capillary tubes (17). The tubes were placed in ice and immediately were centrifuged at 4°C. Free fatty acids were measured with the NEFA-C kit from Wako Pure Chemical (Osaka) GmbH, according to the manufacturer's recommendation. For fast protein liquid chromatography (FPLC) fractionation, 40 μl of pooled plasma was injected onto a Superose 6 column (3.2 × 30 mm, Smart-system, Amersham Pharmacia) and was eluted at a constant flow rate of 50 μl/min with PBS (pH 7.4, containing 1 mmol/liter EDTA). Fractions of 50 μl were collected and assayed for total cholesterol and triglycerides. Enzyme activities of lipoprotein lipase and hepatic lipase in postheparin plasma obtained from control (n = 5) and del¹¹/del¹¹ (n = 5) mice 5 min after bolus injection of sodium heparin (100 units/kg) into the tail vein were assayed as described (18).

VLDL was isolated from pooled plasma of healthy human volunteers by ultracentrifugation at density 1.006 g/ml [40,000 rpm, SW40 rotor (Beckman Instruments, Palo Alto, CA) for 18 h at 4°C]. For VLDL turnover studies, VLDL was radiolabeled with ¹²⁵I by using IODO-BEADS (Pierce). The specific activity of ¹²⁵I-VLDL was 76 cpm/ng. Mice were fasted overnight and were injected in the tail vein with labeled VLDL (10 μg of tracer in 200 μl of 0.9% NaCl containing 2 mg/ml BSA). Blood samples of ≈50 μl were withdrawn from the retroorbital plexus into heparinized capillary tubes 2, 10, 20, 45, 90, 180, and 300 min after injection. The plasma content of ¹²⁵I-labeled apoB was determined after propan-2-ol precipitation and measurement of ¹²⁵I content of the pellet. The data were modeled by using a biexponential curve from which the fractional catabolic rate was calculated by using the reciprocal area under the curve.

Total RNA was extracted from various tissues by using Trizol (GIBCO/BRL) according to the manufacturer's protocol. Radiolabeled antisense riboprobes were generated by using the MAXIscript kit (Ambion) from DNA templates generated by PCR from genomic DNA. RNase protection assays were performed by using the RPAII kit (Ambion) according to the manufacturer's instructions. The protected fragments were separated on a 5% acrylamide denaturing gel, and the dried gel was exposed to x-ray film. A β-actin probe was used as a control.

To delete the 450-kb genomic interval between Irf-1 and Csfgm (Fig. 1), two loxP sites bracketing this region were introduced sequentially into embryonic stem (ES) cells. The first loxP site was introduced into the last exon of the Irf-1 gene by using the targeting vector ploxPNI, containing a neomycin resistance (neo^r) gene. After G418^r selection, correctly targeted ES cells were identified by Southern blot analysis. The second loxP site was introduced 2 kb downstream of Csfgm into the already ploxPNI-targeted ES cells via the targeting vector ploxPHTC, which contains tandemly linked bacterial hygromycin resistance (13) and herpes simplex virus thymidine kinase (HSV-tk) (14) genes. Southern blot analysis of hygromycin resistant clones was used to identify those that had undergone the correct targeting events. Twelve of the doubly targeted ES clones were expanded and electroporated with the Cre recombinase-expressing plasmid pBS185 (19). Because of the presence of loxP sites bracketing the hygromycin and HSV-tk markers, ES cells that had undergone Cre recombinase-mediated loxP recombination were identified by selecting for hygromycin sensitivity and 1-(2-deoxy-2-fluoro-β-d-arabinofuransyl)-5-iodouracil (FIAU) resistance. Cells surviving this selection were confirmed to contain the predicted 450-kb cis deletion (del¹¹) by Southern blot analysis (data not shown). The clones heterozygous for the deleted chromosome were injected into blastocysts to generate chimeric mice, and germline transmission of the deletion from two independently targeted ES cell lines was obtained. Analysis of 461 live births from the mating of heterozygous animals revealed the following genotypes: 25% (+/+), 49% (+/del¹¹), and 26% (del¹¹/del¹¹). Although this Mendelian ratio indicates the viability of the homozygous deletion mice, the majority of the del¹¹/del¹¹ mice died by 6 weeks of age, with the oldest surviving to 30 weeks (Fig. 2).

The del¹¹/del¹¹ mice appeared normal at birth but by 1 day of age developed abdominal swelling, the result of dramatic hepatomegaly. Gross examination revealed enlargement of livers and hearts of the del¹¹/del¹¹ mice (Table 1 and Fig. 3A). The body weight of these animals was significantly reduced compared with del¹¹/+ and wild-type littermates (Table 1). Microscopic analysis of liver sections stained with hematoxylin and eosin (Fig. 3B) and Oil Red O (data not shown) demonstrated cytosolic lipid accumulation. Triglyceride content was elevated dramatically in the livers and hearts of the del¹¹/del¹¹ mice whereas total cholesterol level in these organs was only minimally elevated (Table 1). Changes in the lipid content observed in the hearts and livers of the del¹¹/del¹¹ mice were also reflected in their plasma. Free fatty acid and triglyceride levels were markedly increased whereas minimal changes were observed in their plasma cholesterol levels (Table 1). The increased plasma triglycerides, as determined by fast protein liquid chromatography, were mainly confined to the VLDL-sized fractions (results not shown).

To explore the mechanism underlying the hypertriglyceridemia, an investigation of triglyceride metabolism in mice of the various genotypes was performed. These studies analyzed the three major steps involved in VLDL triglyceride metabolism: lipolysis, hepatic uptake, and hepatic secretion. To determine whether the elevated levels of plasma triglycerides in the del¹¹/del¹¹ mice were attributable to impaired lipolysis, the activity of both hepatic lipase and lipoprotein lipase was measured in postheparin plasma. Similar activities were noted in plasma from del¹¹/del¹¹ and wild-type mice for hepatic lipase (10.6 ± 1.9 vs. 12.3 ± 0.7 μmol/ml/h, P = 0.076) and lipoprotein lipase (14.3 ± 4.8 vs. 18.1 ± 2.7 μmol/ml/h, P = 0.83). The hepatic uptake of human VLDL-apoB was assessed by injecting ¹²⁵I-labeled VLDL into the tail vein of animals of the different genotypes and determining the rate of VLDL removal. No significant differences were observed in clearance of VLDL particles from the circulation of del¹¹/del¹¹ and wild-type mice as reflected by the calculated VLDL fractional catabolic rates (0.33 ± 0.05 vs. 0.43 ± 0.08 pool apoB/h, respectively, P = 0.1573).

Alterations in hepatic VLDL-triglyceride secretion as a cause of the elevated plasma triglycerides levels were assessed by injecting fasted animals with Triton WR1339 coupled with timed triglyceride measurements (20). Triton WR 1339 inhibits lipolysis and hepatic uptake of VLDL triglycerides, thereby enabling assessment of changes in plasma VLDL triglycerides due to hepatic VLDL triglyceride secretion. After Triton WR 1339 injections, the hepatic VLDL-triglyceride secretion rates were ≈3- and 2-fold elevated in del¹¹/del¹¹ and del¹¹/+ mice, respectively, compared with control animals (Fig. 4). These data indicate increased hepatic triglyceride secretion as a major etiology of the hypertriglyceridemia in mice containing the deleted chromosome.

The abnormalities observed in the del¹¹/del¹¹ mice (decreased body weight, enlarged fatty livers and hearts, shortened life span, and increased hepatic triglyceride secretion) are, most likely, the consequence of loss of gene(s) function, and, therefore, transgene complementation was used to identify the individual gene(s) responsible for these traits. Transgenic mice were created by using mouse chromosome 11 BACs—219O10, 32I19, 33O24, and 54K15—and human YACs from the homologous region of human 5q31—A94G6 and 131F9 (Fig. 1). The four overlapping mouse BAC clones together encompass the entire deleted interval whereas the two human YAC clones span most of the deleted region, with the exception of a small gap (20 kb). Lines of mice homozygous for the deletion as well as hemizygous for each of the murine BAC and human YAC transgenes were generated through breeding and then were phenotypically assessed. The four murine BACs and the human YAC 131F9 failed to complement any of the abnormalities noted in the del¹¹/del¹¹ mice. In contrast, del¹¹/del¹¹ mice harboring the human YAC A94G6 were indistinguishable in body weight and liver and heart size (Table 1), life span (Fig. 2), plasma and hepatic triglyceride levels (Table 1), and hepatic VLDL-TG production (Fig. 4) from wild-type controls.

To identify the human gene(s) on YAC A94G6 responsible for complementing the deletion-associated abnormalities, we examined and compared the gene content of human YAC A94G6 and the mouse chromosome 11-deleted interval. A combination of mouse/human sequence comparisons and Southern and Northern blot analyses was used to compare the set of putative genes encoded by sequences on human YAC A94G6 to those present in deletion and wild-type mice. Of the nine putative genes encoded by human YAC A94G6, three, OCTN2, E2, and E3, failed to express because of their absence in the genome of mice homozygous for the chromosome 11 deletion. Of these genes, however, only OCTN2 appears to have a murine homologue, present in wild-type but absent in del¹¹/del¹¹ mice. OCTN2 has recently been identified as an organic cation transporter (21–23). Expression analysis of tissues from the YAC A94G6 transgenic animals, with probes specific for murine and human OCTN2, revealed a similar expression pattern, both genes being expressed at a high level in the kidney and at moderately high levels in the liver, heart, skeletal muscle, brain, and spleen (Fig. 5a). The mouse BACs 219O10 and 32I19 contained parts of, but not the entire, murine OCTN2 gene, and del¹¹/del¹¹ mice containing these BACs as transgenes did not produce a murine OCTN2 transcript (data not shown). The human transcription units designated E2 and E3 present within YAC A94G6 had been originally identified based on exact matches in the human expressed sequence tag database (9). Sequences homologous to these two putative human transcription units have not been reported in the mouse expressed sequence tag database nor detected within the murine genomic sequence that is available for the entire 450-kb deleted interval. Sensitive RNase protection analysis of the A94G6 YAC transgenics failed to detect E2 expression in the several tissues examined (Fig. 5b) whereas a human E3 transcript was easily detected in the liver and kidney (Fig. 5c). Because E3 comprised a small putative ORF (90 bp) and had an expression pattern similar to that of OCTN2, we explored whether it was an untranslated component of the OCTN2 transcript. Reverse transcription–PCR analysis of liver and kidney RNA from the YAC A94G6 transgenics with primers to E3 and the 3′ UTR of the OCTN2 cDNA generated a 0.9-kb product. This result indicates that E3 is not an independent transcription unit, but either a 3′ splice variant or a previously unrecognized 3′ end of the OCTN2 transcript. These reverse transcription–PCR results for E3 coupled with the absence of a murine homologue to E2 or detectable human E2 transcripts in the YAC A94G6 transgenics indicates OCTN2 as the one gene present on human YAC A94G responsible for complementing the deletion associated abnormalities.

OCTN2 has been reported as having the capability of transporting carnitine across the cellular membrane with high specificity (24). Recently, defects in this gene have been implicated as being responsible for primary carnitine deficiency in humans (22, 23) and in the murine model for this condition, the juvenile visceral steatosis (jvs) mouse (25). Based on this information, we investigated whether mice containing the del¹¹ allele suffered from general carnitine deficiency. Total carnitine levels were markedly decreased in plasma of del¹¹/del¹¹ mice compared with wild-type mice (13.1 ± 2.1 vs. 56.0 ± 3.1 nmol/ml, P < 0.021, respectively). In addition, a significant decrease in total carnitine levels was noted in plasma of the del¹¹/+ mice (44.4 ± 1.9 nmol/ml, P < 0.021). Because the phenotypic abnormalities in humans with primary carnitine deficiency and in jvs mice containing an OCTN2 missense mutation are corrected by dietary carnitine supplementation, we investigated whether massive systemic carnitine administration could correct any of the abnormalities in the del¹¹/del¹¹. l-carnitine (1 mg) was injected into del¹¹/del¹¹ mice daily from postnatal days 10–28, which is the dose regimen reported to correct some of the abnormalities in jvs mice (26). This administration of carnitine failed to rescue any of the phenotypic abnormalities noted in the del¹¹/del¹¹ mice.

An unexpected finding of this study was that, despite the complete absence of a contiguous, 450-kb, gene-rich interval, the del¹¹/del¹¹ mice were viable. This represents the largest reported chromosomal deletion in mice or humans associated with viability in the homozygous state (27). Although many of the deletion-associated abnormalities in the del¹¹/del¹¹ mice were corrected by the expression of the human OCTN2-containing transgene, the differences between the jvs and del¹¹/del¹¹ mice, including the failure of carnitine rescue and viability differences, suggest that genes in the deleted interval other than OCTN2 contribute to their phenotype. These mice, with and without the various transgenes, coupled to careful phenotyping, will serve as useful reagents for further deciphering the function of the various genes in the deleted interval. Although the approach used in this study is still labor-intensive, it illustrates a general strategy for screening, in either heterozygous and/or homozygous deletion mice, the properties of several individual genes in parallel contained within a genomic interval of particular interest.

Abnormalities in fatty acid metabolism have long been noted in patients with primary carnitine deficiency [systemic carnitine deficiency; online Mendelian inheritance in man (OMIM) no. 212140] and the jvs mice (28) whereas alterations specifically in the metabolism of triglycerides have not previously been explored. The phenotypic screens and follow-up analyses of the del¹¹/del¹¹ mice, all carried out before identification of OCTN2's physiological role, were unbiased by assumptions of the functions of individual genes within the deleted interval. This facilitated the discovery of a novel and previously unappreciated aspect of the carnitine-deficient phenotype: the relationship between carnitine, hepatic triglyceride production, and OCTN2. The findings of the present study, coupled with reports of carnitine administration significantly reducing plasma triglyceride concentrations in hypertriglyceridemic hemodialysis patients (29, 30) as well as in patients with type IV hyperlipoproteinemia (31), provides genetic evidence in support of a critical role for carnitine transport in the regulation of plasma triglyceride metabolism. The analysis of the del¹¹/del¹¹ as well as the del¹¹/+ mice suggest that reductions in carnitine, whose intracellular concentration appears to be affected by modest decreases in OCTN2 activity, leads to the shunting of fatty acids away from mitochondrial transport into triglyceride storage and secretory pools. In the discovery of OCTN2's ability to modulate hepatic triglyceride secretion and plasma levels, we have identified OCTN2 as a potential intervention target for the management of this cardiovascular risk factor.