HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS		QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

J. Biol. Chem., Vol. 277, Issue 30, 26994-27005, July 26, 2002

This Article Abstract Full Text (PDF) All Versions of this Article: 277/30/26994    most recent M203189200v1 Alert me when this article is cited Alert me if a correction is posted Citation Map Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by van der Leij, F. R. Articles by Price, N. T. Search for Related Content PubMed PubMed Citation Articles by van der Leij, F. R. Articles by Price, N. T. Social Bookmarking                      What's this?

Structural and Functional Genomics of the CPT1B Gene for Muscle-type Carnitine Palmitoyltransferase I in Mammals^*

Feike R. van der Leij^§, Keith B. Cox^¶, Vicky N. Jackson, Nicolette C. A. Huijkman, Beatrijs Bartelds, Jaap R. G. Kuipers, Trijnie Dijkhuizen^**, Peter Terpstra, Philip A. Wood^¶, Victor A. Zammit, and Nigel T. Price

From the  Department of Pediatrics, Groningen University Institute for Drug Exploration, University of Groningen and Beatrix Children's Hospital, Groningen 9700RB, The Netherlands, the ^¶ Department of Genomics and Pathobiology, University of Alabama at Birmingham, Birmingham, Alabama 35294-0019, the  Department of Cellular Biochemistry, Hannah Research Institute, Ayr KA6 5HL, Scotland, United Kingdom, the ^** Department of Medical Genetics, University of Groningen, Groningen 9700RB, The Netherlands, and the  Department of Medical Biology, Molecular Biology Unit, University of Groningen, Groningen 9700RB, The Netherlands

Received for publication, April 3, 2002

	ABSTRACT

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS AND DISCUSSION REFERENCES

Muscle-type carnitine palmitoyltransferase I (M-CPT I) is a key enzyme in the control of -oxidation of long-chain fatty acids in the heart and skeletal muscle. Because knowledge of the mammalian genes encoding M-CPT I may aid in studies of disturbed energy metabolism, we obtained new genomic and cDNA data for M-CPT I for the human, mouse, rat, and sheep. The introns of these compact genes are 80% (mouse versus rat) and 60% (mouse versus human) identical. Sheep and goat, but not cow, pig, rodent, or human promoter sequences contain a short interspersed repeated sequence (SINE) upstream of highly conserved regulatory elements. These elements constitute two promoters in humans, sheep, and mice, and, contrary to previous reports, there is a second promoter in rats as well. Thus, the transcriptional organization of these genes is more uniform than previously supposed, with interspecies differences in the 5'-ends of the mRNAs reflecting differences in splicing; only in humans extensive splicing and splice variation is found in the 5'- and 3'-untranslated regions. In the mouse, intron retention was detected in heart, muscle, and testes and may indicate an additional mechanism of regulation of M-CPT I expression. Splice variation in the coding region was previously proposed to lead to expression of CPT I enzymes with altered malonyl-CoA sensitivity (Yu, G. S., Lu, Y. C., and Gulick, T. (1998) Biochem. J. 334, 225-231). However, when expressed in the yeast Pichia pastoris, none of three earlier described splice variants had CPT I activity. Therefore, the involvement of splice variation of M-CPT I in the modulation of malonyl-CoA inhibition of fatty acid oxidation may be less relevant than hitherto assumed.

	INTRODUCTION

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS AND DISCUSSION REFERENCES

Malonyl-CoA is a central metabolite in the regulation of fatty acid metabolism (1). Although it is an intermediate of fatty acid synthesis, it is also synthesized in tissues that are not lipogenic, e.g. skeletal and heart muscle. In these tissues malonyl-CoA may act as a sensor or hormone and fuel (2) through its ability to inhibit mitochondrial outer membrane carnitine palmitoyltransferase (CPT I).¹ This enzyme catalyzes the transesterification of long-chain fatty acyl-CoAs to long-chain acylcarnitines, which are carried into the mitochondrial matrix where acyl-CoA is regenerated for -oxidation (reviewed in Refs. 3-5). Detailed insight of CPT I genes may aid in the understanding and future treatment of major diseases that are related to disturbed energy metabolism such as cardiac hypertrophy, diabetes mellitus, insulin resistance, and obesity.

Two isoforms of CPT I are known: an L-isoform that is expressed in the liver, the neonatal heart, and a number of other tissues, and an M-isoform that is expressed in heart, skeletal muscle, and a limited number of other tissues (3). M-CPT I is encoded by CPT1B, which in humans is a gene of ~11 kb that contains 19 introns, the last of which remains unspliced when poly-adenylation at a site within intron 19 occurs (6). The human CPT1B gene is also known to contain two alternatively transcribed first exons (6-8), i.e. such that either exon 1A (7), also called U (8), or exon 1B, also called M, is transcribed. In the rat, only a single first exon has been found (9), equivalent to human exon 1B(M). The human CPT1B gene maps to the telomeric region of the long arm of chromosome 22 (6, 10). Because independent efforts to sequence CPT1B yielded genomic data (6, 7) and because chromosome 22 was the first human chromosome that was entirely sequenced (11), genomic analyses of human CPT1B are well documented (4, 6, 7, 12, 13). However, with the accumulation of expressed sequence tag information, new variants of the CPT1B gene in the human and other mammals become apparent.

Important animal models to study energy metabolism include mice, for which gene targeting provides unique tools to dissect the genetic and metabolic cross-talk at the molecular level (14), and chronically instrumented fetal and newborn lambs that allow physiologically relevant in vivo studies of energy metabolism in the heart around birth (15-17). The mouse gene for M-CPT I, Cpt1b, has been mapped to chromosome 15 (18). Indeed, human chromosome 22 and mouse chromosome 15 contain a region of orthologous genes, and this syntenic region includes CPT1B/Cpt1b (19). However, preliminary annotated bacterial artificial chromosome clones of the mouse that contain Cpt1b were assigned to the mouse X chromosome.

Here we report the genomic sequence of the mouse Cpt1b gene and compare it to its orthologues in human and rat. We made use of the presence of a gene (CHKL/Chetk- (6, 7, 11)) immediately upstream of CPT1B/Cpt1b to clone the CPT1B promoter from sheep to assign important promoter elements of CPT1B in mammals. Apart from occasional co-transcription of the upstream CHKL/Chetk- and CPT1B genes (12, 13), the human gene contains two alternatively used first exons and consequently is transcriptionally regulated by at least two promoters. Contrary to what was found earlier in rat cardiac myocytes (9), our observations indicate that the mouse, rat, and sheep genes utilize two promoters as well. However, other data provide evidence that the post-transcriptional processing of the human gene for M-CPT I is more complex than that in the other mammals studied thus far.

	EXPERIMENTAL PROCEDURES

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS AND DISCUSSION REFERENCES

Cloning and Sequencing of the Mouse Cpt1b Gene-- Mouse Cpt1b cDNA clones were obtained after RNA isolation from ICR mice and the derivative mice of substrain NOD/LtJ by reverse transcription-PCR through the use of random priming and oligo(dT) priming for first-strand generation (cDNA cycle kit from Invitrogen). The 5'-end of the mouse transcript was obtained by 5'-rapid amplification of cDNA ends (5' RACE kit from Invitrogen). PCR was applied to obtain a P1 clone (Genome Systems, St. Louis, MO) from a SV129 mouse genomic library. Subclones in pZERO (Invitrogen), pGEM (Promega), and pBluescript (Stratagene, La Jolla, CA) were sequenced by fluorescent double-stranded DNA dideoxy-sequencing on ALF and 96-capillary Megabace (both from Amersham Biosciences) and on ABI (Applied Biosystems/PerkinElmer Life Sciences) automated equipment.

Fluorescent in Situ Hybridization-- Mouse embryonic stem cells were cultured on gelatin and harvested, and chromosome spreads were made according to standard cytogenetic techniques. Chromosomal localization of the mouse Cpt1b gene was determined using a 10.4-kb genomic probe that extends from an SacI site in the Chetk-beta gene to an EcoRI site in intron 16 of Cpt1b. The probe was labeled with the Bio-Nick Labeling System (Invitrogen, Gaithersburg, MD). In situ hybridization was carried out essentially according to the manufacturer's protocol. To produce a computer-enhanced G-banding pattern, the slides were counterstained with 4,6-diamidino-2-phenylindole dihydrochloride. FISH analysis was done using CytoVision imaging software (Applied Imaging Corp.).

Cloning and Sequencing of the Sheep CPT1B Promoter-- A highly conserved sequence based on the sequences of the human, rat, and mouse choline/ethanolamine kinase genes upstream of CPT1B was used as the forward primer (see Table I, primer A). The reverse primer (B) was designed using the 5'-end of the sheep CPT1B cDNA sequence (GenBank^TM accession number AJ272435). This primer was also designed to work with the bovine CPT1B genomic sequence (exact match, not shown). High fidelity PCR (PfuTURBO, Stratagene) on genomic DNA from sheep (Finn-Dorset X-breed) yielded products that were cloned in pCR2.1 and pCR-TOPO vectors (Invitrogen) for sequencing. Four independent PCR reactions were used to obtain four clones that represented the two alleles in this breed (GenBank^TM accession numbers AJ288906 and AJ288907). The same primer pair was also used to amplify the promoter region of CPT1B from goat, pig, and cow.

Analysis of Murine and Ovine CPT1B Transcripts-- To determine whether murine, ovine, and rat CPT1B transcripts were derived from two promoters, giving rise to transcripts with alternate 5'-ends, as in humans, PCR was performed using cDNA prepared from heart total RNA. Upstream primers were designed based on regions of the genomic sequences corresponding to human exon 1A(U) or exon 1B(M). Suitable reverse primers were designed corresponding to the exon 2/3 boundary or to exons further downstream (see Table I). The primer pairs used were C and D, and E and D, for ovine exon 1A(U) and 1B(M) equivalents, respectively. Primer E corresponds to the extreme 5'-end of our cDNA sequence, which is upstream of the start site for human CPT1B exon 1B(M). For murine Cpt1b, primer pairs F and G, and H and I, were used to detect possible exon 1A(U) and 1B(M) transcripts, respectively. For rat Cpt1b, primer pairs F and J, and H and J, were used for the equivalent reactions (primers F and H match both mouse and rat sequences).

For detection of transcripts retaining intron 1, PCR was performed with cDNA from mouse heart, muscle, and testes, using primer K (Table I), which anneals to the region corresponding to the intron immediately upstream of the first Cpt1b coding exon, in combination with primer G.

5' and 3' RACE-- 5' RACE was performed using a technique designed to amplify only full-length capped mRNA (GeneRacer kit, Invitrogen). Total RNA was purified from mouse, rat, and sheep heart and mouse muscle and processed according to the manufacturer's recommendations. Reverse transcription was performed using an extended oligo(dT) primer and Superscript II reverse transcriptase (Invitrogen). High fidelity PCR was performed with a forward primer matching the RNA oligonucleotide ligated to the 5'-end of the de-capped mRNA, and Cpt1b-specific primer L or M (mouse), L or N (rat), or D (sheep) (see Table I). To characterize the 3'-ends of mouse Cpt1b, PCR was performed using the heart and muscle cDNAs, with a reverse primer corresponding to the 5'-extension on the oligo(dT) primer in combination with primer O (Table I). PCR products for 5' and 3' RACE were cloned and sequenced as for the ovine CPT1B promoter region PCR product. Four independent clones were characterized in each case.

Expression of Coding Region Splice Variants in Pichia pastoris-- The 2 and 3 splice variants of rat Cpt1b were generated from the cloned PCR products used as probes for ribonuclease protection assays, as described in a previous study (20). Briefly, the Csp45I-Acc65I fragment of 2 was used to replace the corresponding region of the normal (1) form of rat Cpt1b cloned into pGEM-5Zf(+). The complete 2 sequence was then subcloned into pGAPZ A as described (21). For 3, PCR was performed with primer P (Table I), which adds a silent AflII restriction site, and a T7 vector primer, using the cloned PCR product described above as template. The product was cleaved with AflII and EcoRI and used to replace the corresponding region of 1 in pGEM-5Zf(+), followed by subcloning into pGAPZ A as above. A silent AflII restriction site had previously been introduced into 1, as described previously (21). C-terminally myc- plus His₆-tagged versions of rat 1, 2, and 3 were generated by removing the stop codon and fusing the reading frame to the tag present in the pGAPZ A vector. The EcoRV (Cpt1b site)-ApaI (vector site) fragment of Cpt1b cloned into pGAPZ A was replaced by a pair of annealed oligonucleotides (Q and R, see Table I).

For human CPT1B, the 1 normal form was derived from the GFP-tagged construct described in Ref. 22, which was subcloned into pGAPZB as an XhoI-NotI fragment. A PCR product generated with primers S and T (Table I), where primer T replaced the six amino acids absent in the GFP construct, was cloned into pGEM-T. The SphI-AvrII fragments from 1 or the splice variant lacking the 5'-end of exon 14 (see Ref. 6; here termed 4) were separately added to this construct (using the vector-derived SphI site). Finally the two 3' regions of human 1 and 4 were separately cloned into the pGAPZB construct (replacing the fused GFP) as SphI-NotI fragments. Expression in P. pastoris, CPT assay, protein assay, and Western blotting were performed as described previously (22).

An antibody raised against rat muscle-type CPT I, which also cross-reacted with the human enzyme, was used to detect expression of CPT1B splice variants in P. pastoris. The 1507-bp NcoI-EcoRV fragment from rat Cpt1b, corresponding to amino acids 259-760 of the protein, was cloned into pET-20b. The Escherichia coli-expressed protein was purified using the C-terminal His₆ tag and used to raise antibodies in sheep.

Computation-- The DNAstar package from Lasergene was used for sequence analysis. For similarity measurements using the method of Hein (23), the Megalign module was applied using a gap penalty of 11, a gap length penalty of 3, and a K-tuple value of 6. Because of the equal lengths of the protein sequences, the coding sequence (CDS) and protein alignments were straightforward. BLAST (24) was applied to search the public domain data bases for expressed sequence tags (ESTs) to provide additional information. The Celera data base and DISCOVERY software were used for additional mouse sequence analyses. For transcription factor binding site searches the TFsearch program (www.cbrc.jp/research/db/TFSEARCH.html) and MATinspector (www.genomatix.de) were used.

	RESULTS AND DISCUSSION

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS AND DISCUSSION REFERENCES

Cloning and Localization of Mouse Cpt1b-- The sequence of a 12,242-bp genomic fragment (GenBank^TM accession number AJ278284), which contains the entire Cpt1b gene for M-CPT I from the mouse, was determined and compared with the mouse Cpt1b cDNA. In addition to a partial Cpt1b cDNA clone obtained earlier from NOD/Lt mice (18), several cDNA clones were obtained, sequenced, and assembled. This cDNA sequence represents a full-length transcript as it exists in ICR mice and is available in GenBank^TM (accession number AF017174). The sequence starts slightly further upstream than other mouse Cpt1b cDNA entries from the public data bases as well as that from the Celera data base. The conservation of the mouse, rat, and human Cpt1b/CPT1B cDNA sequences is as expected for transcripts of orthologous genes from these species (Table II). The mouse gene contains 19 exons and 18 introns, as does that from rat (Fig. 1), whereas the human orthologue shows some differences, as discussed in the next section.

View larger version (12K): [in this window] [in a new window]   Fig. 1.   Exon/intron structure of the human, rat, and mouse genes for muscle-type CPT I. Exons are shown as boxes, with splice variants indicated. Transcriptional and translational starts are represented by arrows with open and filled heads, respectively. Stops of translation are shown as blocked arrows, and poly-adenylation signals are represented by asterisks. The upstream gene for choline/ethanolamine kinase is shown as a dotted blocked arrow. A dashed line represents DNA that has not been sequenced.

Previous restriction fragment-length polymorphism analysis mapped mouse Cpt1b to chromosome 15, at about 50 centimorgans from the centromere, in a subtelomeric region that shares synteny with human chromosome 22 (18). Because of the appearance of data for a genome fragment encompassing mouse Cpt1b (GenBank^TM accession number AL135899), which was preliminarily assigned to the X chromosome, it was necessary to confirm the chromosomal location by an alternative method. Therefore, we performed FISH analysis on mouse chromosomes. After hybridization with a probe spanning exons 1-16 of the mouse Cpt1b gene as well as an upstream genomic region, signals were observed in metaphase cells as illustrated in Fig. 2. Using the computer-enhanced G-banding pattern of the 4',6-diamidino-2-phenyl-indole image; the labeled chromosomes were recognized as being the two chromosomes 15. Therefore, the mouse Cpt1b as been determined by this independent method is confirmed as residing on chromosome 15. Based on these results, the preliminary annotation of AL135899 will need to be revised.

View larger version (19K): [in this window] [in a new window]   Fig. 2.   Localization of mouse Cpt1b on chromosome 15. A, fluorescent in situ hybridization on a metaphase spread of cultured mouse embryonic stem cells. Arrows indicate the signal for the Cpt1b probe. B, computer-enhanced G banding pattern of the same metaphase spread. Arrows indicate both chromosomes 15.

Interspecies Genomic Conservation-- Structure analysis reveals that the mouse Cpt1b gene, like that for rat Cpt1b, contains 19 exons and 18 introns (Fig. 1), whereas the human orthologue contains an additional final intron that is either spliced (cDNAs of type I (6)) or poly-adenylated (cDNAs of type II). Human CPT1B also has two alternatively transcribed mutually exclusive first exons resulting from the use of two promoters. Here we show that the mouse and rat Cpt1b genes also contain two promoters but that the resultant leader exons are the same exon that start at different sites, due to non-conservation of a splice donor site (Fig. 1, see section on 5' RACE for details). Aside from these differences, comparison of the mouse, rat, and human CPT1B genes revealed complete conservation of all 18 splice junction positions at the nucleotide level (Table III). The majority of splice junction positions are in phase 0 (i.e. between the codons; 13 out of the 17 intron positions that are within the coding region), in one case in phase I (between the first and second nucleotide of a codon) and in three cases in phase II (between the second and third nucleotide of a codon). This preference for splice junction positions between codons may be biased by the fact that the third nucleotide of a codon (the wobble position) allows more flexibility to obey the consensus "G-gt" splice donor rule. Indeed the G-gt splice donor rule is obeyed in 14 of all the Cpt1b splice donor sites in the mouse and the rat and in 12 cases in the human gene. However, alternative explanations for splice phase 0 preference may also apply (25, 26), and there are genes for important enzymes in fatty acid metabolism known, e.g. the fatty acid synthase gene (27), that show a preference for phase I with a concomitant high number of glycine codons, the latter permitting compliance with G-gt.

Rat Cpt1b and human CPT1B exons 2-18 show a mean similarity of 84.9 ± 4.5% (9). This value is 84.8 ± 4.4% between mouse Cpt1b and human CPT1B and 92.2 ± 4.2% between mouse and rat Cpt1b. These values are not surprising, given the normal picture of gene conservation already known from the cDNA comparisons (Table II). However, the mean similarity of the introns of these genes is extraordinary high (e.g. 60% between human and mouse, Table III), given the fact that in a large scale comparison between human and mouse genes this parameter has been reported to be 35% (28). The high degree of similarity for CPT1B introns could be due to their short length (28). The similarity between mouse and rat Cpt1b introns is about 80% (Table III). Due to the relatively low number of genomic rat sequences known to date, no reference values from large scale comparisons of introns from orthologous murine genes are available yet.

Two poly(A) signals are present in human, rat, and mouse CPT1B/Cpt1b transcripts, but only in the human gene is the most upstream poly(A) signal contained within an intron. This intron is not functional in the rodent genes because of a lack of splice donor sites within the homologous sequence regions. It should be noted that for our comparisons we left the last human intron unspliced while poly-adenylation at the downstream poly(A) addition site was assumed (Tables II and III). However, such a situation is rarely seen in the EST data bases: only EST AI42100 uses the most downstream poly(A) signal with the last intron remaining unspliced.

Intraspecies Variants of Mouse Cpt1b-- The large amount of genomic sequence information available has allowed us to compare the mouse Cpt1b gene sequence (AJ278284), which is from strain 129/SvJ, with two other sources of Cpt1b sequence from strain 129.

The first source was GenBank^TM entry AL135899 from genomic library RPCI-21, which was derived from 129S6/SvEvTac (Sanger Centre, Cambridge, UK, see also www.chori.org/bacpac/21framefmouse.htm). Comparison shows one difference in the repeat length of a microsatellite that resides within intron 10; (TA)₁₀CTTT(TA)₁₇ in substrain 129S6/SvEvTac versus (TA)₁₀CTTT(TA)₁₆ in substrain 129/SvJ.

The second source is the genomic sequence within the Celera data bases (www.celera.com), which is compiled from four mouse strains; therefore, we used the primary fragment data from each strain for our comparisons. The sequences of 19 fragments from 129/SvJ and 7 fragments from 129/SvImJ were identical to our data. In terms of single-read overlaps, these fragments cover about two-thirds of AJ278284. Much more fragment information from the Celera data base is from strains A/J (42 fragments that cover the whole gene except for one gap) and DBA/2J (45 fragments with about four-fifths double-read coverage and five gaps). Apart from tentative single nucleotide polymorphisms (SNPs) that may be misinterpreted sequencing errors, the main difference between the 129/SvJ strain and the other two is again in the microsatellite in intron 10, with (TA)₁₂CTTT(TA)₁₄CTTT(TA)₁₄(CA)₅TACA(TA)₅(CATA)₃ (CATATA)₇TATATA in strains A/J and DBA/2J versus (TA)₁₀CTTT(TA)₁₆TG(TA)₇(CA)₅TACA(TA)₅(CATA)₂(CATATA)₁₄ in strain 129/SvJ (AJ278284 positions 5805-5992).

We did not find any SNPs within the coding sequences of these genomic data (all single nucleotide differences observed were at the end of single read sequences). However, an additional comparison between three cDNA sequences, i.e. AF017174 from strain ICR, AB010826 from strain DDY, and the deduced transcript sequence from 129/SvJ, revealed the following SNPs that imply protein sequence differences (position numbering according to AF017174): 1666G(528E) in strains ICR and 129/SvJ is 1666A(528K) in strain DDY; 2365G(761V) in ICR is 2365A(761I) in 129/SvJ and DDY. Amino acid residue 528 is not conserved throughout the known M-CPT I protein sequences, although the equivalent position in L-CPT I is absolutely conserved as Glu. This residue lies within the region absent in the human CPT1B 4 splice variant (see below). The Ile residue at position 761 near the C terminus of the protein is absolutely conserved in all known Cpt1 sequences, however, Val represents a conservative change.

Insertion of a SINE Element in the Ovine CPT1B Promoter Region-- The human, rat, and mouse genomic sequences of this locus are highly similar and all show the conserved presence of an upstream gene for choline/ethanolamine kinase  (12) in close proximity (Fig. 1). The distance between the stop codon of this upstream gene and the start of the human exon 1A(U) or homologous regions in the murine genes is about 0.5 kb.

Analysis of the 5'-end of our cDNA for ovine CPT1B (GenBank^TM accession number AJ272435), which contains the equivalent of human exon M/rat exon 1 but extends a further 40 (compared to human)/73 (compared to rat) nt upstream, suggested a different promoter organization for the ovine gene. We cloned this genomic region by degenerate PCR. Using a forward primer based on the human, rat, and mouse sequences for the choline/ethanolamine kinase- gene and a reverse primer based on the ovine cDNA sequence itself, we obtained a 1.3-kb fragment that contains the ovine CPT1B promoter. The larger than expected size of this product is explained by the presence of the Alu-type repetitive element Art2 (29). This short interspersed repeated sequence (SINE) is specific for Bovidae (which include cattle, goats, and sheep) but is not found in Suina (which includes pigs) (30). To investigate the presence of this SINE within other mammalian CPT1B promoters, we performed a PCR reaction on genomic DNA from cow, goat, sheep, and pig (Fig. 3A). The fragment size of these PCR reactions was comparable between goat and sheep, but the PCR on bovine and porcine genomic DNA yielded a fragment of a size that indicates the absence of the SINE. Therefore, the presence of the SINE in the ovine promoter for CPT1B is probably the result of an insertion event that occurred relatively recently before sheep and goats diverged but after Bovinae (cattle) and Caprinae (goats and sheep) diverged. The absence of this SINE in pigs (Fig. 3A) is as expected (30).

View larger version (58K): [in this window] [in a new window]   Fig. 3.   Promoter comparisons and transcriptional start site positions. A, PCR analysis of sheep, goat, cow, and pig CPT1B promoters showing a 650-bp larger size in sheep and goats. B, schematic representation of the CPT1B promoter region in human, rat, mouse, and sheep. The legend is the same as for Fig. 1; the 650-bp SINE element is indicated. A dashed line represents DNA that has not been sequenced. C, promoter region alignment with residues that match the consensus (not shown) is shaded, conserved potential transcription factor binding sites are indicated by lines, and non-conserved but putative important binding regions are indicated by dashed lines. The question mark refers to a conserved region with unknown function. The transcriptional starts determined in this study are indicated by arrows for mouse (filled, pointing downward), rat (open, pointing downward), and sheep (filled, pointing upward). The human most upstream start identified previously (7) is indicated by an open arrow pointing upward. The human exons 1A(U) and 1B(M) are boxed based on the major transcriptional starts reported in a previous study (8).

Regulatory Sequence Conservation in Four Mammalian CPT1B Promoters-- CPT1B mRNA is highly expressed in heart, muscle, brown adipose tissue (31), and in testis (32). Its expression in the rat heart increases relative to Cpt1a during development (33), however, this increase is due to a decline in Cpt1a rather than to changes in Cpt1b (34). Electrical stimulation of cultured rat neonatal cardiac myocytes up-regulates Cpt1b expression, co-incident with a hypertrophic response (35). Expression of Cpt1b increases in brown adipose tissue from both rat (36) and mouse (37) following cold exposure. Cpt1b is not detected in immature testis but is highly expressed on maturation (32). CPT1B is expressed in white adipocytes from humans, rats, and hamsters (31) but not in those from mice, which express Cpt1a. Rat pre-adipocytes express only Cpt1a, whereas the Cpt1b isoform is expressed following differentiation. However, mouse 3T3-L1 cells express only Cpt1a throughout differentiation. It is not clear whether this differential expression is due to differences within the murine Cpt1b promoter or whether it occurs at the level of some regulatory factor(s), in which case species-specific expression of other genes would also be expected in white adipocytes.

The CPT1B promoter, like that for CPT1A, is TATA-less with basal expression driven by specificity protein 1 (Sp1). Recently, a region in the first exon/intron was shown to contain an enhancer of basal transcription in rat (38). Long-chain fatty acids have been shown to activate CPT1B transcription in rat primary cardiac myocytes, and the response element has been mapped for both the human and rat promoters (8, 39, 40). This fat-activated/fatty acid response element (FARE) (8, 41) is activated through peroxisome proliferator-activated receptor  (PPAR-). In PPAR- null mice, the CPT I inhibitor etomoxir fails to induce cardiac expression of the Cpt1b gene, whereas in wild type mice it does, suggesting that intracellular fatty acid derivatives comprise a signaling pathway in the heart mediated via PPAR- (39). Hypoxia reduces PPAR-/retinoic X receptor  DNA-binding activity and decreases Cpt1b expression (42), in line with a role for PPAR- in the regulation of the energy substrate switch from glucose to fatty acid oxidation.

The FARE overlaps with a COUP-TF binding site (8), and both elements are highly conserved throughout the four mammalian Cpt1b promoters (Fig. 3C). COUP-TF has been shown to modulate the responsiveness of the CPT1B promoter to fatty acids through transrepression (8). Several additional binding sites in the upstream cluster that were found in the rat Cpt1b promoter previously (40) are also conserved. These are the E-boxes, sites for myocyte enhancing factor-2, Sp1, and serum response factor (SRF). In addition, conserved putative sites for muscle-specific mitogen binding factor that overlap with the myocyte enhancing factor-2 site are present, with an additional second site present only in sheep. Sheep also differ in having additional upstream putative GC-box elements and a muscle initiator sequence (Fig. 3C).

Remarkably, a region between the upstream and downstream clusters of (potential) binding elements, which was shown to be of minor importance for basal transcription in neonatal cardiac myocytes (40), contains a highly conserved motif (indicated by a question mark in Fig. 3C). For this motif we found overlaps with matrices for LyF-1 and Ikaros2; these have implied roles in lymphocyte differentiation (43). Although to our knowledge a role of CPT1B in lymphocyte differentiation has not been studied, we know that fatty acids are important energy substrates for lymphocytes and human lymphoid cells are known to express CPT1B, because we previously cloned CPT1B cDNAs from lymphoblasts.² However, the possibility should not be excluded that one or more other unknown factors could be of importance for binding to this conserved motif. In view of its limited importance for expression in neonatal cardiac myocytes, this conserved region could well be a module within the CPT1B promoter that functions in non-cardiac tissues. It should be noted that this region might represent a second or degenerate COUP/FARE site, because it contains the core sequences GACC (+ strand) and AAAG ( strand).

The downstream cluster contains two Sp1 sites, the first of which is not conserved in terms of position and orientation. The second Sp1 site is conserved but flanked by gaps in the alignment for the human and sheep sequences; therefore, this site is within different immediate context in these species compared with the rodents (Fig. 3C). There is less conservation at the nucleotide level of the binding site for GATA (Fig. 3C), which is known to regulate rat Cpt1b expression in conjunction with SRF (40). The deviations in this region do not interfere with the correct prediction of the potential GATA binding site in all four species, but the human sequence lacks the core TATC. This is in line with the fact that human CPT1B shows a robust response to long-chain fatty acids, whereas the response in rat is more modest, with the rat gene being shown to be strongly GATA-4-dependent with co-activation by SRF (40).

A second GATA site was experimentally determined in the rat (40). In this case (indicated with a dashed line in Fig. 3C) the core TATC sequence and also the prediction of GATA binding propensity is limited to the rat Cpt1b promoter. Therefore, the other species may behave differently in GATA complex formation at or around this site. The CA-box sequence CCACCC, of which deletion or mutation in the rat promoter causes a drastic decrease in reporter gene expression (40), is also not fully conserved. However, the Nkx binding site is conserved in all four promoters and indeed Nkx2.5 was shown to interact synergistically with GATA-4 and SRF. Another likely candidate for interactions with the serum response factor is Elk-1, for which a strongly conserved binding sequence exists downstream of the second Sp1 site (Fig. 3C). This site is partially overlapped by a CCAAT/enhancer binding protein recognition sequence, which strengthens its potential importance (44).

Thus, in addition to the experimental evidence provided by others (8, 34, 40, 41), the conserved presence of important binding sites suggests a conserved gross architecture of transcription protein complex formation, whereas the differences that clearly exist both in the sequences (Fig. 3) and promoter specificity of CPT1B in different species (see e.g. Ref. 31) point to differences in modular promoter function. It should be noted in this respect that we also identified putative estrogen and cyclic AMP response elements and steroid regulatory element binding protein recognition sites, but these were not conserved either as sequence per se or in position/orientation (not shown). The availability of sequences for this compact regulatory region from four different species should allow further investigation of the similarities and differences of transcriptional regulation of this important gene and, in particular, will facilitate further investigation of Cpt1b expression in mouse white adipocytes.

Analysis of Murine and Ovine CPT1B Transcripts-- To determine whether murine and ovine CPT1B transcripts were derived from two promoters, giving rise to transcripts with alternate 5'-ends as in humans, PCR was performed using cDNA prepared from heart total RNA. Upstream primers were designed based on regions of the genomic sequences corresponding to human exon 1A(U) or exon 1B(M). Suitable reverse primers were designed corresponding to the exon 2/3 boundary or to exons further downstream. For both mice and sheep, the sizes of the resultant PCR products were not compatible with mutually exclusive leader exons, as found in humans. Thus, although the product for 1B(M) was of the predicted size, that for exon 1A(U) was in each case ~170 bp longer than expected. Cloning and sequencing of the products showed that the 1B product contained the 3'-end of exon 1B(M) spliced to exon 2 as expected. However, all 1A products contained the end of exon 1A(U) but also contained all the genomic sequence downstream as far as the 3'-end of exon 1B(M), again spliced to exon 2. Thus the 1B(M) product was nested within the 1A product. Our results thus show that murine transcripts exist where exon 1 is at least 185 nt long.

5' Rapid Amplification of cDNA Ends-- The results described above are compatible with two possibilities, which would not be distinguished using the method of PCR with internal primers: (i) there is only a single promoter corresponding to the first human promoter, such that exon 1 contains at least most of the regions corresponding to human exon 1A(U), together with intron 1 and exon 1B(M); or (ii) two (or more) promoters exist as in humans, and an equivalent of exon 1B(M) is not only part of a bigger first exon but also an independent first exon on its own. To distinguish between these two possibilities and to find the transcription start sites, 5' RACE was performed using a method that amplifies only capped full-length mRNA.

5' RACE was performed on mouse muscle and heart cDNA using a Cpt1b-specific reverse primer (corresponding to a region of exon 6) in combination with a primer that anneals to a sequence derived from an RNA oligonucleotide ligated to the 5'-end of the mRNA. For both tissues, only a single product was observed, corresponding to a mature transcript with ~50 nt upstream of the first coding exon (exon 2). Thus, although our previous PCR analysis showed that longer transcripts (with a leader exon of at least 185 nt long) were present within the cDNA pool, the longer PCR product was not generated using the forward RACE primer. This was attributed to competition with the shorter template for both primers. Also (as for intron 1) the sequence between the regions corresponding to human exons 1A(U) and 1B(M) is highly GC-rich (61% overall, with a region of 46 nt being over 76% G or C), with the consequent high degree of secondary structure probably impairing reverse transcription and efficiency of PCR amplification. To overcome this, an alternative reverse primer with a binding site upstream of the start of exon 1B(M) was used, such that the shorter transcript could no longer be amplified by the PCR. Using this strategy we successfully amplified 5'-ends for the longer mRNA. Cloning and sequencing of four independent clones in each case showed the major transcription sites to be as indicated in Fig. 3C (starting 234 and 232 bp upstream of intron 1).

5' RACE was also performed using ovine heart cDNA. Here, products of two different sizes were amplified using a single primer pair, suggesting exon 1 lengths of ~50 and 240 nt, with the longer product being more abundant. Sequencing showed that exon 1B(M) begins 42 (two clones) or 43 nucleotides (two clones) upstream of exon 2, whereas exon 1A(U) starts 240 (one clone), 233 (one clone), or 232 (two clones). Our original PCR with a primer based on the 5'-end of our cDNA, which was designed to detect the downstream promoter activity, was thus in fact detecting transcripts from the upstream promoter. The transcriptional start sites for the downstream ovine Cpt1b promoter identified here, which are separated by only one nucleotide, are exactly the same as sites determined for mouse.

Therefore, as in humans, mouse Cpt1b and sheep CPT1B are transcribed from two different promoters, giving rise to equivalent primary transcripts. However, the resultant mature mRNAs differ in the two cases, as shown schematically for mouse and human in Fig. 1. Thus, human CPT1B transcripts resulting from the U or the M promoter contain completely different mutually exclusive leader exons; either exon 1A(U) or exon 1B(M). Usage of promoter M in sheep or mice, which lies within exon 1, generates message with a leader exon equivalent to that produced in humans. However, in sheep and mice, U promoter activity produces a much longer leader exon than in the case of humans. This exon contains the equivalent of both human exons 1A(U) and 1B(M), together with the genomic sequence separating these two regions. This is due to the non-conservation in sheep and mice of a donor splice site ~100 nt downstream of exon 1A(U) (Table III). The U promoter transcript thus contains a 5'-lengthened version of the leader exon present in promoter M transcripts.

Analysis of Rat Cpt1b Transcripts-- Although 5' RACE suggested that rat Cpt1b only has a single promoter with a short (20 nt) leader exon equivalent to human exon 1B(M) (9) and there are no ESTs or cDNAs in the data base to contradict this, it would be predicted from the high degree of conservation of this region, that a second promoter also exists in rat. In particular, the striking clustering of transcription factor binding sites into two groups within the regulatory region of the human, sheep, and mouse genes is also conserved in rat (40) (Fig. 3C). Because the splice site downstream of exon 1A(U) is also absent in the rat gene (Table III), it might be predicted that rat transcripts from this upstream promoter would have a similar structure to those found for mouse and sheep. In particular the GC-rich region that led to our initial failure to identify mouse promoter 1A(U) transcripts could also explain the failure to detect these transcripts in rat (9).

To investigate this, PCR was performed as for mouse Cpt1b, using the same forward primers, which match to the regions corresponding to human exons 1A(U) and 1B(M). We achieved exactly the same results as for mouse; i.e. transcripts resulting from two promoters were detected.

5' RACE was also performed as for the mouse gene. Products for the downstream promoter were detected using a reverse primer downstream of exon 1. Sequencing of four different clones revealed leader exon lengths of 14, 33, 52, and 63 nt, compared with the previously mapped size of 20 nt. The 63- and 33-nt exons exactly match sites identified for the mouse Cpt1b gene (Fig. 3C). As for 5' RACE for the mouse, to isolate products for the upstream promoter, it was necessary to use a reverse primer binding upstream of the M promoter start sites. Using this strategy we successfully isolated clones that start at 243 (four clones) and 241 nt (one clone) upstream of intron 1. These sites are at adjacent positions to those in mouse (Fig. 3C), again confirming the existence of a second upstream promoter in rat.

Several studies have been performed using reporter constructs derived from the rat Cpt1b gene, and these have thus been based on the incorrect assumption that there is only a single promoter (34, 40). Our finding does not invalidate these studies: elements that specifically regulate the downstream promoter are not necessarily located only in the region between the two transcriptional start sites. Indeed the FARE co-regulates both promoters for the human gene (8). In a previous study (34), a reporter construct with 391 nt upstream of the mapped rat transcriptional start site (391 to +80) behaved similarly to a longer construct (1188 to +80 nt) in primary cultured rat neonatal cardiac myocytes. Similar constructs (391 to +80 and 1108 to +80) showed comparable behavior when expressed in L6 myoblasts (40). However, in H4IIE hepatoma cells, the longer construct was 4-fold less active than the shorter form, suggesting the presence of negative elements in the 1108 to 392 region that repress expression in liver-type cells. This region mostly contains the 3'-end from the upstream choline kinase gene, with the remaining ~160 nt being from the Cpt1b gene.

The existence of two promoters in rat, rather than the previously supposed single promoter, does, however, have important implications for reporter assays, which measure the combined effects of transcription and translation. As discussed below, transcripts from the U promoter, which have an extended 5'-untranslated repeat, are likely to be translated less efficiently than the M transcripts. Thus, this type of reporter assay (which retains the rat Cpt1b leader exon) will underestimate changes in transcription from the upstream promoter. Reporter assays on the human CPT1B gene that use constructs containing exons 1A(U) and 1B(M) and thus generate transcripts retaining either exon 1A(U) or 1B(M) may also be confounded by the relative influence of the alternate leader exons on translation although possibly to a lesser than is the case for the rat gene.

The differences between human versus ovine and murine CPT1B transcripts may have important consequences for the stability and translatability of the mRNA. Human exon 1A(U) does not contain an AUG codon, whereas human exon 1B(M) contains an upstream open reading frame (uORF), which terminates in exon 2 downstream of the start codon for the CPT1B ORF. However, the start codon for this uORF is less than 10 nt away from the 5'-end of the transcript and is thus unlikely to support translational initiation. Mouse exon 1B(M) contains a similar uORF that terminates 7 nt upstream of the Cpt1b ORF. Again the proximity to the 5'-end makes it unlikely that this uORF is translated. However, mouse exon 1B(M) transcripts contain exactly the same uORF but with much more sequence upstream. In this case the uORF, whose AUG codon lies in good context according to a previous study (45), is likely to be translated. This is likely to reduce significantly the translation of murine Cpt1b from exon 1A(U) transcripts, which would rely on leaky scanning or re-initiation. Both forms of the ovine CPT1B transcript lack uORFs.

3' Rapid Amplification of cDNA Ends-- 3' RACE was performed using mouse heart and muscle cDNAs synthesized with a modified oligo(dT) primer, as described. For both tissues two sizes of PCR products were generated, and sequencing of four cloned products indicated addition of poly(A) at two different sites, i.e. AAAAATAAGGCTGTATAATTCTT(A)_n (upstream site) and GGAATAAATTCTTACTTTAGAACCTT(A)_n (downstream site).

Therefore, mouse exon 19 contains two poly-adenylation signals, separated by 219 nt (Fig. 1). Both regions show a partial match to the consensus GU-rich (YGUGUUYY) rather than a U-rich type downstream element found within 30 nucleotides of the poly(A) addition site (46). The upstream and downstream sequences (UAUGUUUU and UCUGUUUG, respectively) differ from the consensus at one or two positions, as indicated. Ignoring the size of the poly(A)-tail, the two mRNAs resulting from usage of the alternative poly-adenylation sites differ in length by 217 nt. Sequences of cloned cDNAs within the data bases, together with EST sequences, also confirm usage of both of these signals. Thus sequences with accession numbers AF017174 and BC018270, and with AB010826, correspond to use of the upstream and downstream signals, respectively. Within the EST data base, no obvious relationship between tissue source and transcript type is discernible. The shorter transcripts are ~1.5-fold more abundant than the longer form, suggesting a slight preference for use of the upstream poly-adenylation signal, as might be predicted.

The rat Cpt1b gene also contains the same two poly-adenylation signals in highly similar context within the final exon and thus might be expected to have alternate 3'-ends. Indeed, of the 21 currently available rat Cpt1b 3'-end ESTs there are two (AW434701 and BI287369) that use the upstream poly(A) signal. Analysis of human EST sequences revealed at least four different 3'-end variants in addition to the two previously described (6). These utilize 1) either of two alternate cryptic splice sites within intron 19 (e.g. accession numbers AI274496 and AB051457), thus giving an extended exon 19 spliced to exon 20, or 2) a cryptic splice acceptor site within exon 20 (e.g. accession number BE552079), thus giving exon 19 spliced to a 5'-truncated exon 20. In addition EST clone accession number AI142100 retains intron 19, whereas upstream introns are spliced conventionally (results not shown).

Retention of Intron 1 in Mature Transcripts-- Analysis of sequences for the 5'-ends of human EST clones showed the existence of several clones with apparent retention of intron 1, with normal splicing of downstream exons (not shown). These clones were obtained predominantly from testes. Interestingly, the CPT1B clone (accession number U62733) obtained by the group of McGarry (10) from a human heart cDNA library also contains 27 nt derived from intron 1 at the extreme 5'-end. Thus PCR was used to determine whether such transcripts exist in mouse testes, skeletal muscle, and heart.

Transcripts retaining the 5' intron were readily detected in cDNA from all three tissues (not shown). Intron 1 is 452 nt long and has a high GC (67.7%) content, which includes regions of very high GC content, including a run of 21 G and Cs. Indeed this region proved very hard to sequence, and we needed to use various genomic sub-clones of different length to obtain reliable data. Thus, it is not surprising that such transcripts were not detected by 5' RACE or during PCR with forward primers based on exon 1A(U) or 1B(M). The GC-rich regions are likely to have impaired reverse transcription and/or PCR.

All of the human ESTs that show apparent retention of intron 1 are truncated at the 5'-end, presumably due to impaired reverse transcription. The longest (accession number BI560870) contains 142 nt out of the 450 nt of human intron 1. Thus it is not possible to associate intron 1 retention with usage of either promoter, although it might be expected that in testes transcription would be from the U (universal) rather than the M (muscle) promoter. Likewise, in mouse further experimentation is needed to investigate this phenomenon. It should be noted that the unusual transcripts found to contain regions coding for choline/ethanolamine kinase as well as CPT1B (13, 47) did not contain this intron, and thus our PCR could not have been detecting these transcripts.

Retention of 5'-introns has been argued to be a possible additional means of regulating mammalian gene expression (48), especially because the first intron is excised in a different way to internal introns (49). In extreme cases, for example expression of gonadotropin-releasing hormone in extra-hypothalamic tissue (50), expression of a gene is allowed in an ectopic tissue, but its translation is prevented by non-removal of the 5'-intron. Intron 1 of murine Cpt1b does not contain an additional AUG codon, however, the high GC-content noted above might result in mRNA with a high degree of secondary structure that could impair translation.

Expression of Coding Region Splice Variants in P. pastoris-- In addition to the alternate splicing described above, three splice variants that alter the protein-coding region of CPT1B have previously been described (Fig. 1). The 2 and 3 variants (where 1 refers to the pattern of splicing normally observed) have been described for both humans (51) and rats (20). 2 lacks part of exon 3 and all of exon 4, due to use of a cryptic splice donor site within exon 3. The resultant protein thus lacks the second transmembrane domain and a region shown to target liver-type CPT I (L-CPT I) to mitochondria (52) (Fig. 4A). This mitochondrial targeting sequence also contains a histidine residue (His¹⁴⁰) that is proposed to be involved in CPT I-malonyl-CoA interactions (53). In the case of 3, exon 5 is skipped (see Fig. 4A). These variants have been suggested to result in forms of M-CPT I with altered interaction with their inhibitor, malonyl-CoA (20, 51). A third variant of human CPT1B, here termed 4, lacks 30 nt from the 5'-end of exon 14 due to use of a cryptic splice acceptor site within this exon, thus resulting in in-frame deletion of 10 amino acids (⁵²⁶CQAVIESSYQ⁵³⁵) (6). Three of these residues (Cys, Ile, and Ser⁵³³) are absolutely conserved in all known mammalian CPT I sequences.

View larger version (34K): [in this window] [in a new window]   Fig. 4.   Expression of splice variants of rat and human CPT1B in P. pastoris. A, schematic representation of the regions deleted within the N-terminal region in the 2 and 3 splice variants. The transmembrane (TM) domains, the domain that loops into the intermembrane space (IMS Loop), and the mitochondrial targeting sequence (MTS) are indicated. B, expression of C-terminally His6-tagged rat Cpt1b 1, 2, and 3 in P. pastoris. Whole cell-free extracts were resolved by SDS-PAGE using 4-12% NuPage gels with MOPS buffer (Invitrogen). After blotting, expression was detected with an antibody against the His6 tag. Only one of the 1 clones and four out of five 2 and 3 clones are shown. C, expression of (non-tagged) human CPT1B 1 and 4 in P. pastoris. Expression was detected with an antibody to recombinant rat muscle-type CPT I protein. Lane 1 contains rat heart mitochondria, and lane 2 contains a non-transformed Pichia control extract. Only two of the four 1 clones assayed are shown. Arrows show the position of migration of full-length rat and human M-CPT I.

Assays of CPT activity for five independent P. pastoris clones, each for "normal" human and rat M-CPT I (1), the rat 2 and 3 splice variants, and the human 4 variant, were performed using 35 and 135 µM palmitoyl-CoA. The rat and human 1 clones had activities of 1.38 ± 0.20 and 4.31 ± 0.17 nmol/min/mg of protein, respectively, at the higher palmitoyl-CoA concentration. No activity above background was detected for the 2, 3, or 4 variants.

Because the constructs used to generate the splice variants were identical to their respective rat and human CPT1B 1 forms apart from the missing exon(s) or parts of exons, incorporation into the Pichia genome and efficiency of translation are likely to be similar. Thus clones selected with the same concentration of zeocin antibiotic would be expected to express similar amounts of CPT1B protein. Western blotting with M-CPT I-specific antibodies was used to examine expression of the various forms of the protein. Levels of expression of rat M-CPT I in Pichia were too low to detect unequivocally using our antibody, even though CPT activity could be detected for 1. Therefore, C-terminally tagged versions of 1, 2, and 3 were generated. The activity of the Pichia-expressed 1 form was similar to that of its untagged parent (1.02 ± 0.09 nmol/min/mg of protein; n = 4), which was also selected for similar copy number with 0.5 mg/ml zeocin, showing that the tag does not interfere with CPT activity, as has been shown for rat L-CPT I (54).³ As for the untagged versions, no CPT activity was detected for the Pichia-expressed and tagged 2 and 3 isoforms, even when using selection for a higher copy number with 2 mg/ml zeocin.

Western blotting verified that these variant proteins were indeed expressed (Fig. 4B). Despite a higher copy number, tagged 2 was detected at similar levels to tagged 1, suggesting an increased susceptibility to proteolysis of the former in vivo or during preparation of extracts. Conversely, tagged 3 protein was detected at a higher level, as expected from its higher copy number in the Pichia genome. These forms of rat M-CPT I were only detected as single bands, suggesting that any proteolytic degradation that occurs involves rapid removal of the C-terminal tag, such that these additional species are not detected.

Previous studies have shown that a yeast-expressed mutant form of L-CPT I with deletion of a region corresponding to a mitochondrial targeting sequence showed increased susceptibility to proteolysis by trypsin in intact mitochondria (52). This mitochondrial targeting sequence (contained within residues 123-147 of rat L-CPT I, which is equivalent to residues 125-149 of rat M-CPT I) is within the region absent in rat 2 (residues 84-153), in accord with the likelihood that lower protein detection levels are due to increased proteolysis. Rat L-CPT I mutants lacking residues 127-142, like rat Cpt1b 2, are catalytically inactive when expressed in Pichia.³

In the case of human CPT1B, expression levels were sufficiently high to allow detection with M-CPT I-specific antibody (Fig. 4C). Full-length human 4 was detected at much lower levels than the 1 form. Despite this, sufficient full-length protein was present for some clones to allow measurement of CPT activity if 4 were to have had similar specific activity to the parental 1 form. However, we were unable to detect any activity for expressed 4.

The low expression level for 4, despite similar copy number to 1, is probably due to degradation of the protein within the yeast, or an extreme susceptibility to proteolysis during preparation of cell-free extracts, despite the presence of a mixture of protease inhibitors. Several faster migrating immunoreactive species were observed, with one of ~60 kDa being prominent (Fig. 4C), which may represent a highly folded core derived from the catalytic C-terminal segment (52). Part of the structure surrounding the catalytic site in rat L-CPT I (residues 368-568) was recently modeled (55). The missing residues in M-CPT I 4 are equivalent to residues 526-535 of rat L-CPT I and are thus contained within this important region. These residues are proposed to lie within an -helical region away from the catalytic channel.

It is unlikely that the lack of activity associated with the expressed 2, 3, and 4 was due to an inability of the translated proteins to insert in membranes. Low speed centrifugation of extracts from Pichia-expressing rat (His₆-tagged) and human CPT1B and the three splice variants, followed by Western blotting, showed that in each case the CPT I protein was associated with a particulate rather than a soluble fraction (not shown). The fractions of L-CPT I present within rat liver peroxisomes and microsomes are catalytically active, as is L-CPT I expressed in these organelles in Pichia (56).³ Thus even if 2, 3, and 4 were targeted to non-mitochondrial membranes, we would expect to detect any activity present. It is possible that the splice variants adopt a different membrane topology to that of parental M-CPT I, particularly in the case of 2, which lacks one of the two transmembrane segments. However, even this possibility is unlikely to have accounted for the lack of catalytic activity, because freeze-thawing of whole-cell extracts or mild digitonin treatment failed to unmask any putatively latent activity.

Examination of human and mouse EST sequences shows that several other splice variants exist. For example a mouse EST clone (accession number BG865012) lacks 46 nt from the 5'-end of exon 13 due to the use of a cryptic splice acceptor site within this exon, in a similar fashion to the human 4 variant. However, the translational frameshift present in this case almost certainly results in a truncated non-functional protein. Human clones (accession numbers BF739380 and AW606583) exist where the cryptic splice donor site within exon 3 is used, as in 2, but, unlike the 2 isoform, exon 4 is retained. Again this results in introduction of a frameshift and premature stop codon. The former of these clones also lacks 44 nt from the 5'-end of exon 5 due to use of a cryptic splice site within exon 5. CPT1B ESTs retaining internal introns can also be identified. It is suggested that the 2, 3, and 4 CPT1B splice variants, together with the additional variants described above, represent failures in the fidelity of the splicing machinery. This does not, however, preclude a functional role for the resultant proteins, especially where deleted/inserted regions do not introduce a translational frameshift. However, in the case of 2, 3, and 4, these proteins lack CPT activity.

Conclusions-- We analyzed genomic and cDNA sequences of CPT1B and showed that this compact gene is moderately conserved in mammals, with a relatively high degree of conservation of the intron sequences. Independent of the fact that co-transcription of the upstream CHKL/Chetk- gene and CPT1B may occur, conserved elements constitute two functional promoters in human, mouse, and sheep. Significantly, we show that rat, like the other species, has a second promoter, and suggest that the previous failure to detect transcripts from the upstream promoter is a result of the high degree of secondary structure downstream of the transcriptional start site. Thus the transcriptional organization of CPT1B genes is more uniform than previously supposed, with interspecies differences in the 5'-ends of mature mRNAs reflecting differences in splicing. Sheep and goat, but not cow, pig, rodent, or human promoter sequences, contain a SINE element upstream of the conserved regulatory elements. This may have important consequences for CPT1B expression in the former species, and this is currently being investigated. Only in the human is extensive splicing and splice variation found in the 5'- and 3'-untranslated regions, and we identified additional 3'-end variants. Thus, the post-transcriptional processing of the human gene for M-CPT I is more complex than its expression in the other mammals studied thus far.

Together several variants and their combinations for the 5'- and 3'-ends of CPT1B mRNA have been described: alternate promoter usage, intron 1 retention, alternate splicing of the final exon in humans, and use of alternate poly-adenylation sites. These different mRNAs may differ significantly in their translatability, stability, and possibly subcellular targeting.

Although splice variation in the coding region was proposed to play a major role in the expression of CPT I enzymes with altered malonyl-CoA sensitivity, none of three earlier described splice variants showed CPT activity when expressed in recombinant Pichia. Therefore, splice variation is probably not involved in the modulation of malonyl-CoA sensitivity of the inhibition of fatty acid oxidation.

	ACKNOWLEDGEMENTS

Partial PCR fragments of the rat splice variants were kindly provided by Dr. Tod Gulick, Massachusetts General Hospital, Boston, MA, and genomic DNA from domestic mammals was kindly provided by Dr. M. Barber, Ayr, Scotland. We thank Drs. L. Wilming and L. Smink at the Sanger Centre, UK for communicating additional information about AL135899.

	FOOTNOTES

^* This work was supported by the Netherlands Heart Foundation (NHS 97.093 and NHS 2001.081), National Institutes of Health Grant RO1-RR-02599, the Scottish Executive Environment and Rural Affairs Department, and the British Heart Foundation.The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

The nucleotide sequence(s) reported in this paper has been submitted to the GenBank^TM/EBI Data Bank with accession number(s) AJ278284, AJ288906, and AJ288907.

^§ To whom correspondence should be addressed: University of Groningen, Beatrix Children's Hospital and Groningen University Institute for Drug Exploration, Department of Pediatrics, Research Laboratory CMCV-2, P. O. Box 30001, Groningen 9700RB, The Netherlands. Tel.: 31-50-361-1542 or 31-50-361-3251; Fax: 31-50-361-1746; E-mail: f.r.van.der.ley@med.rug.nl.

Published, JBC Papers in Press, May 15, 2002, DOI 10.1074/jbc.M203189200

² F. R. van der Leij, N. C. A. Huijkman, and B. Bartelds, unpublished results.

³ V. N. Jackson, N. T. Price, and V. A. Zammit, unpublished observation.

	ABBREVIATIONS

The abbreviations used are: CPT I, carnitine palmitoyltransferase I; L-CPT I, liver-type CPT I; M-CPT I, muscle-type CPT I; EST, expressed sequence tag; RACE, rapid amplification of cDNA ends; FISH, fluorescence in situ hybridization; GFP, green fluorescent protein; SNP, single nucleotide polymorphism; nt, nucleotide(s); Sp1, specificity protein 1; FARE, fat-activated/fatty acid response element; PPAR, peroxisome proliferator-activated receptor; SRF, serum response factor; uORF, upstream open reading frame; MOPS, 4-morpholinepropanesulfonic acid; SINE, short interspersed repeated sequence; COUP-TF, chicken ovalbumin upstream promoter transcription factor; CDS, coding sequence.

	REFERENCES

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS AND DISCUSSION REFERENCES