Inactivation of the Peroxisomal Multifunctional Protein-2 in Mice Impedes the Degradation of Not Only 2-Methyl-branched Fatty Acids and Bile Acid Intermediates but Also of Very Long Chain Fatty Acids*

According to current views, peroxisomal β-oxidation is organized as two parallel pathways: the classical pathway that is responsible for the degradation of straight chain fatty acids and a more recently identified pathway that degrades branched chain fatty acids and bile acid intermediates. Multifunctional protein-2 (MFP-2), also called d-bifunctional protein, catalyzes the second (hydration) and third (dehydrogenation) reactions of the latter pathway. In order to further clarify the physiological role of this enzyme in the degradation of fatty carboxylates, MFP-2 knockout mice were generated. MFP-2 deficiency caused a severe growth retardation during the first weeks of life, resulting in the premature death of one-third of the MFP-2−/− mice. Furthermore, MFP-2-deficient mice accumulated VLCFA in brain and liver phospholipids, immature C27 bile acids in bile, and, after supplementation with phytol, pristanic and phytanic acid in liver triacylglycerols. These changes correlated with a severe impairment of peroxisomal β-oxidation of very long straight chain fatty acids (C24), 2-methyl-branched chain fatty acids, and the bile acid intermediate trihydroxycoprostanic acid in fibroblast cultures or liver homogenates derived from the MFP-2 knockout mice. In contrast, peroxisomal β-oxidation of long straight chain fatty acids (C16) was enhanced in liver tissue from MFP-2−/− mice, due to the up-regulation of the enzymes of the classical peroxisomal β-oxidation pathway. The present data indicate that MFP-2 is not only essential for the degradation of 2-methyl-branched fatty acids and the bile acid intermediates di- and trihydroxycoprostanic acid but also for the breakdown of very long chain fatty acids.

In mammals, ␤-oxidation is confined to two organelles, mitochondria and peroxisomes. Whereas mitochondrial ␤-oxidation is primarily involved in the catabolism of short, medium, and long chain fatty acids supplementing energy to the cell, peroxisomal ␤-oxidation seems to be responsible for the degra-dation of a number of less abundant carboxylates of different molecular structure (1,2). The substrates for peroxisomal ␤-oxidation known to date include very long straight chain fatty acids (containing more than 20 carbon atoms); 2-methylbranched chain fatty acids (e.g. pristanic acid that is formed after ␣-oxidation of the 3-methyl-branched fatty acid phytanic acid); prostanoids; dicarboxylic acids; and the C 27 bile acid intermediates di-and trihydroxycoprostanic acid, which are converted to the mature C 24 bile acids via peroxisomal ␤-oxidation.
Based on the substrate specificity and stereoselectivity of the newly discovered enzymes, different ␤-oxidation pathways have been proposed (2,13). The classical pathway catalyzed by palmitoyl-CoA oxidase, multifunctional protein-1 (also denoted as L-bifunctional protein because the hydrated species it generates has the L-configuration), and peroxisomal thiolase is generally accepted to be responsible for the oxidation of straight chain fatty acids. All enzymes of this pathway are found in different species and can be strongly induced in rodents by ligands of the peroxisome proliferator-activated receptor ␣ (PPAR␣). 1 The oxidation of 2-methyl-branched fatty acids and of the bile acid intermediates, di-and trihydroxycoprostanic acid, is believed to occur via a second pathway. In humans, this pathway consists of branched chain acyl-CoA oxidase (5), multifunctional protein-2 (the D-specific multifunctional protein, which is identical to 17␤-estradiol dehydrogenase type IV and which generates the D-hydroxy intermediate (9,14)), and sterol carrier protein-x (SCPx), the N-terminal part of which exerts thiolytic activity (12). In rats, the desaturation of bile acid intermediates is executed by trihydroxycoprostanoyl-CoA oxidase (3), which is the counterpart of the human branched chain acyl-CoA oxidase (15,16). A third acyl-CoA * These studies were supported by grants from Geconcerteerde Onderzoeksacties K.U. Leuven (GOA/99/09). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18  oxidase, pristanoyl-CoA oxidase, is very active in rat liver and is thought to be involved in pristanic acid breakdown (4,17). In mice, as in humans, pristanoyl-CoA oxidase is not detectable at the protein level (17,18). 2 Several in vitro and in vivo studies indicate that the separation between the two pathways may not be as strict as generally believed and that intermediates produced in one of these ␤-oxidation pathways can be shuttled to the other system. In particular, in vitro experiments have shown that the multifunctional proteins (MFP-1 and MFP-2) have a broad substrate spectrum; they are both capable of hydrating the enoyl-CoA esters of straight chain and 2-methyl-branched fatty acids and of the bile acid intermediates and they can both dehydrogenate straight chain and branched chain 3-hydroxyacyl-CoAs (6 -10, 19, 20). However, it was concluded from stereochemical studies that MFP-2 and not MFP-1 is responsible for the degradation of bile acid intermediates and pristanic acid. Indeed, only the hydroxylated intermediate produced by MFP-2 has the right configuration to be further dehydrogenated by the same enzyme (7,9,10,19).
Besides this involvement in the degradation of branched compounds, there are indications that MFP-2 might also be physiologically important for the oxidation of straight chain fatty acids. During studies on linoleic acid (21) and palmitic acid 3 breakdown by purified peroxisomes, generation of both 3-hydroxyacyl-CoA stereoisomers was observed. In human fibroblasts, which contain more MFP-2 than MFP-1, MFP-2 was suggested to play a major role in the peroxisomal oxidation of long chain and very long chain fatty acids (VLCFA) (22). Furthermore, mice with an inactivated MFP-1 gene display normal levels of VLCFA (23), whereas a patient with a genetic defect in the MFP-2 gene was reported to accumulate, besides branched fatty acids and bile acid intermediates, also VLCFA (24). However, the latter observation could also be explained by an inhibitory effect of accumulating branched chain ␤-oxidation intermediates on MFP-1 (24).
The aim of the present study was to further investigate the in vivo role of MFP-2 by generating a mouse model with MFP-2 deficiency. Inactivation of the MFP-2 gene caused a severe growth retardation and death of one-third of the mice during the first postnatal weeks. Overall peroxisomal ␤-oxidation activity toward different substrates, individual enzyme activities, and accumulation of substrates for peroxisomal ␤-oxidation were examined. The results demonstrate that MFP-2 is of prime importance not only in the degradation of 2-methylbranched fatty acids and bile acid intermediates but also in the degradation of very long chain fatty acids.

Construction of the Targeting Vector and Generation of MFP-2 Ϫ/Ϫ
Mice-A P1 genomic clone encompassing 80 kilobase pairs of the mouse MFP-2 gene was obtained through Genome Systems (St. Louis, MO) by polymerase chain reaction screening using as primers AgCATgggAC-CATATgAAgAAACAg (forward) and CTCCgTCATCCgTgACCCAgCg-TTg (reverse). These oligonucleotides are located at base pairs 399 -423 and 582-606 of the published mouse cDNA (25) and encompass a 1.5-kilobase pair genomic DNA fragment. The P1 clone was partially mapped by restriction enzyme digestion and Southern hybridization and by polymerase chain reaction analysis to locate intron/exon boundaries. The targeting vector was constructed by subcloning a 3.5-kilobase pair BamHI/NheI fragment (located in the 5Ј-upstream region of the gene) as the 5Ј-flank and a 6.0-kilobase pair SspI/KpnI fragment (lo-cated in intron 3) as the 3Ј-flank in the pNT vector (26). Linearization of the targeting vector and electroporation in R1 ES cells were done as described previously (27). Southern analysis of 260 clones resistant to both G418 and ganciclovir revealed four clones that had undergone the correct homologous recombination (Fig. 1). Aggregation of these recombinant ES cell clones with morulae derived from Swiss mice yielded several chimeric mice, one of which transferred the MFP-2 ϩ/Ϫ genotype to the offspring.
Southern, Northern, and Western Analysis-Mice were genotyped by extraction of tail DNA, digestion with EcoRV, and hybridization of the Southern blot with the 3Ј external probe (Fig. 1). For Northern analysis, 20 g of total RNA extracted from adult mouse livers by the Chomczynski procedure (28) was loaded on a 1.1% (w/v) agarose gel containing 3% (w/v) formaldehyde. A 1071-base pair probe corresponding to the 1042-2113-base pair fragment of the mouse MFP-2 cDNA was generated by reverse transcription-polymerase chain reaction from mouse liver RNA and used for Northern hybridization in the presence of formamide.
␤-Oxidation of 14 (32), but dithiothreitol was omitted, and the specific activity was raised to 5 Ci/mol. Formation of CO 2 was negligible in homogenates, and oxidation rates were calculated from the generation of acid soluble products as described.
Measurement of Enzyme Activities-Acyl-CoA oxidase activities in liver homogenates from 6 -8-week-old mice were analyzed as described (35). For the measurements of palmitoyl-CoA oxidase, homogenates were pretreated with N-ethylmaleimide, and 100 M palmitoyl-CoA was used as substrate in the presence of 10 M bovine serum albumin. For the determination of branched chain acyl-CoA oxidase, the N-ethylmaleimide pretreatment was omitted, and 100 M 2-methylhexadecanoyl-CoA, in the presence of 40 M bovine serum albumin, was used as substrate. Due to the apparent absence of pristanoyl-CoA oxidase in mouse liver (17), this substrate is used by branched chain acyl-CoA oxidase and results in higher activities than those obtained with trihydroxycoprostanoyl-CoA. 4 Multifunctional proteins were measured by following the dehydration of 3-hydroxy-3-phenylpropionyl-CoA, a substrate introduced by Schulz and co-workers (36). For the CoA-ester synthesis, the hydroxysuccinimide esters (150 mol) (37) of the 3-hydroxy-3-phenylpropanoic acid isomers (Acros), crystallized from ethanol, were dissolved in 2.5 ml of tetrahydrofuran and mixed with an equal volume of 0.2 M NaHCO 3 containing 50 mol CoA and stirred for 3 h at room temperature. After removal of the organic solvent by a stream of nitrogen, pH was brought to 4 with acetic acid, 2 ml of methanol was added, and remaining ester was extracted with heptane/ethyl acetate (1:1, v/v). Insoluble material was removed by centrifugation, and the supernatant was adjusted to pH 7 with NaHCO 3 . For the assay, a 50-l aliquot of the homogenates, appropriately diluted, was added to cuvettes, followed by the addition of 400 l of reaction medium and 50 l of 1 mM CoA-ester (either the 3R-isomer for the measurement of MFP-1 or the 3S-isomer for the determination of MFP-2), and reactions were monitored at 25°C at 308 nm (⑀ ϭ 26, 500). Final optimized assay concentrations were 200 mM potassium phosphate buffer, pH 8.0, 1 mM EDTA, 0.02% (w/v) Thesit. The addition of the detergent Thesit caused a 2.5-4-fold increase in the activities for both substrates in fresh mouse liver homogenates, 4 probably by destroying permeability barriers. In mouse liver, approximately 90 -95% of the dehydratase activity measured with the 3R-isomer was sensitive to heating (50 -60°C; 2.5-10 min), 4 suggesting that contribution of the murine mitochondrial hydratase under our assay conditions is minor. In rat liver, 80% of the activity was heat-sensitive, in agreement with other reports (38).
GC Analysis of FA in Liver, Brain, and Plasma-A Bligh-Dyer extraction was performed on liver or brain from 6 -8-week-old mice (39). Lipids were further separated in neutral lipids, fatty acids, and phospholipids by ion exchange chromatography (Bond Elut NH 2 , 500 mg, Varian). The fractions containing phospholipids were subjected to acidic methanolysis in the presence of internal standards (heptadecanoic acid and heptacosanoic acid) followed by gas chromatography of the fatty acid methyl esters on a BPX70 column (40). Triglycerides were isolated from the fractions containing neutral lipids by TLC (solvent heptane/ diethylether/acetic acid, 70:30:1, v/v/v), eluted with chloroform/methanol (41), transmethylated, and analyzed by gas chromatography.
Analysis of Bile Acids-Bile, collected by puncturing the bladder, was pooled from different animals, and 20 -40 l were mixed with 50 l of 0.2 N NaOH and 600 l of ethanol. After boiling for 5 min, the mixture was centrifuged, and the precipitate was washed once with 600 l of ethanol. The combined supernatants, after adding 1 ml of water, were extracted twice with 2 ml of heptane and then acidified with 50 l of 1 N HCl, and bile acids were extracted with 2 ϫ 2 ml of diethylether. The ether phases were dried under nitrogen and dissolved in 20 l of 70% methanol, and 1 l was mixed with glycerol and analyzed by negative ion fast atom bombardment mass tandem mass spectrometry as described by Libert et al. (42).

RESULTS
Targeting of the Mouse MFP-2 Gene-With the aim of abolishing both enzymatic activities (hydratase and dehydrogenase) of the MFP-2 gene product, a targeting vector was constructed such that after homologous recombination the first three exons were deleted (Fig. 1). The successful inactivation of the MFP-2 gene in MFP-2 Ϫ/Ϫ mice was demonstrated by Northern and Western blotting showing the total absence of MFP-2 transcripts and protein (Fig. 1). This was further confirmed by analyzing the dehydration of 3S-hydroxy-3-phenylpropionyl-CoA, catalyzed by the D-specific hydratase domain of MFP-2. The dehydration activity was easily measurable in liver homogenates from wild type and MFP-2 ϩ/Ϫ mice but was negligible in liver homogenates derived from MFP-2-deficient mice (Table I).
Growth Retardation and Survival of MFP-2 Knockout Mice-Genotyping 158 pups from 15 litters demonstrated a normal Mendelian inheritance of the MFP-2 Ϫ/Ϫ allele (22% MFP-2 ϩ/ϩ ; 52% MFP-2 ϩ/Ϫ ; 26% MFP-2 Ϫ/Ϫ ). At birth, the MFP-2 Ϫ/Ϫ mice were indistinguishable from their wild type and heterozygous littermates. However, from postnatal day 2 on, a marked growth retardation of the MFP-2-deficient mice was observed (Fig. 2, A and B). Weight gain was particularly low during the lactation period resulting in an average 50% reduction of body weight in comparison with wild type littermates, at the time of weaning. Approximately 30% of MFP-2 Ϫ/Ϫ mice were more severely affected, grew only marginally, and died between postnatal days 2 and 12 (Fig. 2C). After weaning at 3 weeks of age, weight gain of MFP-2 Ϫ/Ϫ mice resumed, and their body weight was on average 30% lower than controls in the adult stage. The growth curves of a representative litter in which all pups survived into adulthood is shown in Fig. 2B. Macroscopic and light microscopic inspection of the major MFP-2 Ϫ/Ϫ organs did not reveal any changes with the exception of the colon content of suckling mice. These feces had a more yellow fluid appearance and contained more undigested residues as compared with those from age-matched controls. No fat droplets were identifiable by Sudan red staining. Female MFP-2-deficient mice were fertile, but males had a strongly reduced fertility.
Peroxisomal and Mitochondrial ␤-Oxidation of 14 C-Labeled Substrates-The consequences of inactivation of the MFP-2 gene on the catabolism of different substrates for peroxisomal ␤-oxidation were tested by incubating fibroblast cultures and liver homogenates with 14 C labeled substrates (Tables II, III). Palmitic acid was degraded to the same extent in fibroblast cultures derived from MFP-2 ϩ/ϩ and MFP-2 Ϫ/Ϫ mice. However, in intact cells a major mitochondrial contribution in the breakdown of this long chain fatty acid can be expected. When oxidation of palmitoyl-CoA was studied in liver homogenates under conditions favoring the peroxisomal pathway (presence of cyanide; absence of carnitine and albumin) (31), activities in homogenates of MFP-2 Ϫ/Ϫ mice were enhanced 2.5-fold. In contrast, the mitochondrial ␤-oxidation of palmitate was not affected by the absence of MFP-2. When fibroblasts were incubated with lignoceric acid, a specific substrate for peroxisomal ␤-oxidation, degradation was 10 times lower in MFP-2-deficient fibroblasts than in controls, strongly suggesting that MFP-2 is more important than MFP-1 for VLCFA breakdown. As expected, the oxidation of a 2-methyl-branched fatty acid, 2-methylhexadecanoic acid, was markedly reduced in MFP-2deficient fibroblasts (80% reduction as compared with wild type fibroblasts). Trihydroxycoprostanic acid was not ␤-oxidized by fibroblasts, 4 presumably because of lack of uptake or activation. In liver homogenates prepared from MFP-2-deficient mice, the degradation rate of trihydroxycoprostanic acid was 5 times lower than in homogenates from wild type mice. ␣-Oxi- dation rates of 3-methylhexadecanoic acid were rather low in mouse fibroblasts, but no differences were apparent. 4 These analyses indicate that in the absence of MFP-2, the peroxisomal degradation of VLCFA, 2-methyl-branched fatty acids, and bile acid intermediates is severely reduced but that the mitochondrial and peroxisomal oxidation of long straight chain fatty acids (C16) is unaltered or even increased.
Induction of Peroxisomal ␤-Oxidation Enzymes-As already mentioned, no MFP-2 dehydration activity could be measured in liver homogenates of MFP-2 Ϫ/Ϫ mice. In contrast, the activ-ity of MFP-1, palmitoyl-CoA oxidase, and urate oxidase appeared to be induced 2-4-fold as compared with wild type controls (Table I). Western blot analysis confirmed the induction of palmitoyl-CoA oxidase, MFP-1, and peroxisomal thiolase (Fig. 3). The peroxisomal membrane protein PMP70 and the microsomal CYP4A -hydroxylase were induced to a lesser extent, whereas the expression of PEX5p, the import receptor of most peroxisomal matrix proteins, was unaltered as compared with controls (data not shown). These protein inductions were continuously present from the first postnatal week into adulthood.
Curiously, also branched chain acyl-CoA oxidase, an enzyme that is hardly influenced in rat by fibrate treatment (3), was induced. The activity of this enzyme was hardly measurable in wild type mice, but it could easily be detected in liver homogenates of the MFP-2 Ϫ/Ϫ mice (Table I).
Fatty Acid Analysis in Tissues-In view of the reduced oxidation rates of VLCFA, accumulations of these carboxylates could be expected in tissues of MFP-2 knockout mice. C 26 levels were quantified in the phospholipid fraction of brain and liver using GC analysis of the fatty acid methyl esters. In livers and brains of adult MFP-2 Ϫ/Ϫ mice, C 26 levels were increased 3-6fold as compared with age-matched controls (Fig. 4A). Up until 6 weeks of age, no accumulation of the 2-methyl-branched fatty acid pristanic acid was found in the liver phospholipid or triglyceride fraction. However, supplementation of mouse chow with phytol (5 mg/g), as was previously done for SCPx knockout mice (43), led to increased levels of pristanic acid and to a greater extent of its precursor phytanic acid in the phospholipid and triglyceride fraction of MFP-2 Ϫ/Ϫ livers (Fig. 4B). The increased levels of branched chain fatty acids in MFP-2-deficient mice caused weight loss, cataract, and ataxia very similar to the observations in SCPx knockout mice treated with this dose of phytol (43).
Bile Acid Analysis-The degradation of cholesterol into C 24 bile acids involves two or three hydroxylations of the ring structure and shortening of the side chain by -oxidation followed by peroxisomal ␤-oxidation. As revealed by mass spectrometric analysis of the bile acids extracted from the bile, the taurine conjugate of a trihydroxylated C 24 bile acid (m/z 514) is the most abundant species in control mice, followed by a dihydroxylated C 24 compound (Fig. 5, A and C). In mice, the trihydroxylated C 24 bile acids consist predominantly of ␣and ␤-muricholic acid and to a lesser extent of cholic acid. In bile of the MFP-2 knockout mice, these C 24 bile acids were still present, but various other bile acids with larger masses are seen (Fig. 5, B and D). By daughter ion analysis, the major species with m/z 554 was shown to consist of taurine conjugates of trihydroxylated C 27 bile acids possessing a double bond in the side chain (Fig. 5E). To further evaluate bile acid levels, the expression of cholesterol 7␣-hydroxylase, which is under a negative feedback inhibition by bile acids (44), was analyzed by Western blotting. In microsomes of MFP-2 knockout mice, cholesterol 7␣-hydroxylase was only slightly up-regulated as compared with wild type mice (data not shown), probably reflecting the presence of considerable levels of normal C 24 bile acids in adult MFP-2 knockout mice.  (4) Values represent the mean Ϯ S.E. of the activities expressed as nmol ⅐ min ⅐ mg protein (n ϭ number of samples).

DISCUSSION
Based on detailed substrate specificity studies of the different enzymes involved, peroxisomal ␤-oxidation was proposed to be organized in two parallel pathways: one pathway responsible for the degradation of straight chain fatty acids and a second pathway, whose enzymes were more recently identified, responsible for the breakdown of pristanic acid and bile acid intermediates. However, the present study demonstrates that straight VLCFAs accumulate, besides branched chain fatty acids and bile acid intermediates, in mice lacking MFP-2, an enzyme belonging to the second pathway. These findings are in accordance with the increased levels of VLCFA recently observed in MFP-2-deficient patients (24). It can be argued that the impaired degradation of VLCFA under conditions of MFP-2 deficiency is due to the accumulating branched chain enoyl-CoAs that competitively inhibit MFP-1. However, in the present study it was unequivocally demonstrated that the degradation of VLCFA was also impaired in fibroblast cultures of the MFP-2 knockout mice, which are not exposed to pristanic acid or bile acid intermediates. In addition, the oxidation of long chain fatty acids was even enhanced in liver homogenates of MFP-2-deficient mice as compared with wild type, rather supporting increased activity of MFP-1. Taking further into account the fact that VLCFA levels are not increased in MFP-1deficient mice (23), it can be concluded that MFP-2 and not MFP-1 is the principal enzyme responsible for the degradation of VLCFA. Consequently, the function of MFP-1 seems to be restricted to the oxidation of long chain fatty acids and probably of eicosanoids and dicarboxylic acids.
These results raise the question of which oxidase and which thiolase are involved in the degradation of VLCFA. In vitro, palmitoyl-CoA oxidase, branched chain acyl-CoA oxidase, and pristanoyl-CoA oxidase can desaturate lignoceroyl-CoA (5,45). The importance of palmitoyl-CoA oxidase in in vivo conditions is supported by the fact that VLCFA accumulate in patients and in mice with a deficiency of this enzyme (46,47). No changes in VLCFA levels were observed in SCPx knockout mice (43), suggesting that peroxisomal thiolase and not SCPx is responsible for cleavage of the 3-keto derivatives of the VLCFA-CoAs. These results call for a reconsideration of the organization of peroxisomal ␤-oxidation; instead of a strict separation of the two pathways, substrates desaturated by acyl-CoA oxidase can be further converted by either MFP-1 (long chain fatty acids) or MFP-2 (VLCFA), and the reaction products of the latter enzymes can converge again for the thiolytic cleavage by thiolase.
The occurrence of immature C 27 bile acids, carrying an unsaturated side chain, in the bile of MFP-2-deficient mice confirms the importance of MFP-2 (6,7,10) in the formation of bile acids. However, normal C 24 bile acids were also present, similar to the observation in a patient with complete MFP-2 deficiency (48,49). This suggests that alternative pathways, which bypass MFP-2, convert cholesterol into mature bile acids. The sterol 25-hydroxylase pathway involving cleavage of acetone from the side chain could account for some of the bile acid synthesis, although it was demonstrated that the activity of the microsomal 25-hydroxylase is relatively low in mouse liver (50). It has been recently reported that in brain, excess cholesterol can be converted to 24S-hydroxycholesterol, which is readily secreted across the blood-brain barrier into the circulation and eliminated via the liver (51,52). It was suggested that this intermediate is metabolized to bile acids in the liver, which implies ring hydroxylations and oxidation of the side chain. This would lead to the formation of 24S,25R-3,7,12,24-tetrahydroxycoprostanoyl-CoA, which has the correct stereochemical configuration for further dehydrogenation by MFP-1. As an additional explanation for the occurrence of normal bile acids, it can be proposed that trihydroxycoprostenoyl-CoA is hydrated by MFP-1 with the formation of 24S,25S-3,7,12,24-tetrahydroxycoprostanoyl-CoA, which after conversion by a racemase to the 24S,25R-stereoisomer (53) can again be dehydrogenated by MFP-1.  The essential role of MFP-2 in the degradation of branched chain fatty acids was demonstrated by supplementing phytol, a tetramethyl-branched fatty alcohol, to the diet of MFP-2-deficient mice. Standard mouse chow indeed contains very few precursors of branched chain fatty acids, which is in contrast to the daily dietary intake by humans of phytol and phytanic acid present in dairy products. Phytol is converted to phytanic acid, which is subsequently shortened by ␣-oxidation with the formation of pristanic acid. The latter 2-methyl-branched fatty acid is further degraded by peroxisomal ␤-oxidation. After a few weeks of phytol treatment, accumulations of pristanic acid and its precursor phytanic acid were found in liver triglycerides and phospholipids of the MFP-2 Ϫ/Ϫ mice, causing severe weight loss and neurological symptoms. Since wild type and MFP-2 ϩ/Ϫ mice tolerated this diet well, it is evident that MFP-2 is necessary for the degradation of branched chain fatty acids.
It is unclear at this point whether the severe growth retardation during the first postnatal weeks is caused by a low food intake or by malabsorption. In addition, it has not been resolved whether this deficit in weight gain is related to the accumulation of VLCFA, reduced levels of normal bile acids, or to the presence of immature bile acids. Since the low growth rate of MFP-2-deficient mice occurs in particular during the period of high fat intake, it seems likely that it is related to the abnormalities of the bile acid composition in the intestines. In adult MFP-2-deficient mice, high levels of normal bile acids seem to be formed by alternative pathways, but whether these pathways are also operative in the early postnatal period needs to be evaluated. Possibly, the C 27 bile acids could be toxic for the intestines, which might be alleviated when solid food is ingested. Further analysis of fat and bile acid content in the intestines during the lactation period and treatment of adult MFP-2-deficient mice with high fat diets might clarify this issue. Interestingly, two other mouse models with inactivation of enzymes involved in the side chain degradation of cholesterol have been generated. In mice lacking sterol 27-hydroxylase, the first enzyme in the pathway, the bile acid pool in intestines and bile was strongly reduced, but only traces of C 27 bile alcohols (50) were found. SCPx-deficient mice were shown to accumulate C 23 bile acids (including norcholic acid) in bile and in serum and to a much lesser extent an unusual C 26 bile alcohol carrying a keto function at C-24 (54). Remarkably, no growth abnormalities were described for these two mouse models (43,50), whereas mice with inactivation of cholesterol 7␣hydroxylase (55) displayed a severe growth retardation and postnatal death. These observations suggest that not only the reduced levels of regular bile acids but also the type and levels of abnormal bile acids might be responsible for influencing growth. Strikingly, palmitoyl-CoA oxidase knockout mice, which presumably have normal bile acids but increased VLCFA, displayed a growth retardation starting from the second postnatal week (47). A mechanism for this impaired growth was not reported.
The absence of MFP-2 causes an up-regulation of the peroxisomal ␤-oxidation enzymes of the classical pathway, palmitoyl-CoA oxidase, MFP-1, and thiolase, which is already apparent 1 week postnatally. Since the expression of the microsomal enzyme CYP4A, the peroxisomal membrane protein PMP70, and urate oxidase was also increased, this pattern of protein induction seems to coincide with the gene induction pattern of the nuclear receptor PPAR␣. In palmitoyl-CoA oxidase knockout mice, similar enzyme inductions were reported (47). It was further demonstrated that PPAR␣ activation underlay the enzyme up-regulation, since double knockout mice lacking both palmitoyl-CoA oxidase and PPAR␣ failed to increase expression of these genes (56). Also in SCPx knockout mice, PPAR␣regulated genes were induced, and it was suggested that the accumulating phytanic acid acts as the activating PPAR␣ ligand under these circumstances (57). It can be expected that also the inactivation of MFP-2 impairs the degradation of a PPAR␣ activator, but the nature of this ligand needs to be investigated. Unexpectedly, also the protein levels and activity FIG. 4. Accumulation of substrates for peroxisomal ␤-oxidation. A, levels of C 26 fatty acid in the phospholipid fraction of liver and brain of heterozygous MFP-2 ϩ/Ϫ mice and MFP-2 Ϫ/Ϫ mice. B, GC profiles of fatty acid methyl esters derived from the triglyceride fraction of a MFP-2 ϩ/Ϫ mouse fed a normal diet (left), a MFP-2 ϩ/Ϫ mouse supplemented with phytol (middle), and a MFP-2 Ϫ/Ϫ mouse supplemented with phytol (right). The following peaks are marked: C16:0 (1); C18:0 (2); pristanic acid (3); phytanic acid (4). of branched chain acyl-CoA oxidase were induced in MFP-2deficient mice. In rats, a lack of responsiveness of this enzyme to PPAR␣ activators has been reported (3,16). Possibly, the mouse gene contains PPAR-responsive elements, since clofibrate treatment also up-regulated this enzyme in this species. 4 In conclusion, the present study has unequivocally demonstrated that in mice MFP-2 is essential for the degradation of saturated very long chain and branched chain fatty acids and for the formation of mature bile acids. shown. This peak probably consists of a mixture of taurine-conjugated ⌬24-trihydroxylated cholestanoic acids (fragmentation patterns in F) with hydroxyl groups in different positions of the ring structure, which are the precursors of ␣/␤-isomers of muricholic acids (specific fragments m/z 413) or of cholic acid (specific fragment m/z 399). Also, a dihydroxylated ⌬24-analogue (m/z 538) and the saturated trihydroxylated C 27 bile acid (m/z 556) are present in bile of MFP-2 Ϫ/Ϫ mice.