The Human Uncoupling Protein-3 Gene

Uncoupling protein-3 (UCP3) is a recently identified candidate mediator of adaptive thermogenesis in humans. Unlike UCP1 and UCP2, UCP3is expressed preferentially and at high levels in human skeletal muscle and exists as short and long form transcripts,UCP3 S and UCP3 L.UCP3 S is predicted to encode a protein which lacks the last 37 C-terminal residues of UCP3 L. In the present study, we have defined the intron-exon structure for the human UCP3 gene and determined thatUCP3 S is generated when a cleavage and polyadenylation signal (AATAAA) located in the last intron prematurely terminates message elongation. In addition we have mappedUCP3 to the distal segment of human chromosome 11q13 (between framework markers D11S916 and D11S911), adjacent toUCP2. Of note, UCP2 and UCP3 in both mice and humans colocalize in P1 and BAC genomic clones indicating that these two UCPs are located within 75–150 kilobases of each other and most likely resulted from a gene duplication event. Previous studies have noted that mouse UCP2 maps to a region of chromosome 7 which is coincident with three independently mapped quantitative trait loci for obesity. Our study shows thatUCP3 is also coincident with these quantitative trait loci raising the possibility that abnormalities in UCP3 are responsible for obesity in these models.

Uncoupling protein-3 (UCP3) is a recently identified candidate mediator of adaptive thermogenesis in humans. Unlike UCP1 and UCP2, UCP3 is expressed preferentially and at high levels in human skeletal muscle and exists as short and long form transcripts, UCP3 S and UCP3 L . UCP3 S is predicted to encode a protein which lacks the last 37 C-terminal residues of UCP3 L . In the present study, we have defined the intron-exon structure for the human UCP3 gene and determined that UCP3 S is generated when a cleavage and polyadenylation signal (AATAAA) located in the last intron prematurely terminates message elongation. In addition we have mapped UCP3 to the distal segment of human chromosome 11q13 (between framework markers D11S916 and D11S911), adjacent to UCP2. Of note, UCP2 and UCP3 in both mice and humans colocalize in P1 and BAC genomic clones indicating that these two UCPs are located within 75-150 kilobases of each other and most likely resulted from a gene duplication event. Previous studies have noted that mouse UCP2 maps to a region of chromosome 7 which is coincident with three independently mapped quantitative trait loci for obesity. Our study shows that UCP3 is also coincident with these quantitative trait loci raising the possibility that abnormalities in UCP3 are responsible for obesity in these models.
The control of body weight involves a regulated balance between energy intake and expenditure. Energy expenditure can be divided into three components (1): resting metabolic rate, physical activity, and adaptive thermogenesis, the latter being defined as the component of energy expenditure that changes in response to environmental stimuli such as cold exposure or chronic dietary excess. In rodents, an important site of adaptive thermogenesis is brown adipose tissue (reviewed in Ref. 2) where uncoupling protein-1 (UCP1), 1 expressed exclusively in brown adipocytes (3,4), promotes proton transport across the mitochondrial inner membrane. UCP1 decreases the proton electrochemical potential gradient, uncoupling fuel oxidation from ADP availability (reviewed in Refs. 5 and 6). Activation of UCP1, therefore, causes increased consumption of calories and generation of heat. UCP1-mediated effects on energy expenditure are regulated by changes in the level of sympathetic nervous system activity in brown fat. Cold exposure and overfeeding cause increased sympathetic stimulation of brown fat, simulating UCP1-mediated uncoupling and energy expenditure. The importance of this is demonstrated by the fact that mice lacking UCP1 are cold-intolerant (7). UCP1 is also regulated by purine di-and trinucleotides (ATP, ADP, GTP, and GDP) and free fatty acids, which inhibit and stimulate uncoupling activity, respectively (reviewed in Refs. 5 and 6).
UCP1 may be of lesser importance in humans in whom the mass of brown adipose tissue is limited. Instead, skeletal muscle is thought to be a major site of adpative thermogenesis (8 -12). UCP2 (13-15) is a recently described UCP1 homologue which, unlike UCP1, is expressed in most tissues. Because of its wide tissue distribution, UCP2 could have important effects on metabolic rate in humans. However, as UCP2 is expressed at high levels in many sites not thought to mediate adaptive thermogenesis, such as spleen, lymph node, thymus, and gastrointestinal tract (13)(14)(15)(16), its role in mediating regulated energy expenditure is unclear.
UCP3 is a third member of the uncoupling protein family (15,16). It was identified by Boss et al. (15) using a homology-based screening method and by the present authors (16) as an expressed sequence tag (EST) deposited into the Washington University, St. Louis-Merck & Co. EST data base. UCP3 is distinguished from other UCPs by its relatively selective, high level expression in skeletal muscle (15,16) and the existence of two RNA transcripts (15), UCP3 L and UCP3 S , which are predicted to encode long (312 amino acids) and short (275 amino acids) UCP3 proteins, differing only by the presence or absence of C-terminal 37 residues. This difference could be significant because the region in question is homologous to a domain in UCP1 thought to mediate inhibition of uncoupling activity by purine nucleotides (17,18). The abundant and relatively selective expression of UCP3 in skeletal muscle suggests that it may be a mediator of adaptive thermogenesis in humans. Here we define the intron-exon structure of the human UCP3 gene, establish its chromosomal localization at 11q13, within 75-150 kb of UCP2, and define the genetic basis for the two UCP3 S and UCP3 L mRNA transcripts.

EXPERIMENTAL PROCEDURES
Intron-Exon Structure-Six sense and antisense PCR primer pairs corresponding to cDNA sequence were used to amplify genomic fragments from human genomic DNA (see Table I). The genomic PCR fragments were subcloned using the TA cloning system (pCR2.1 plasmid, Invitrogen, Carlsbad, CA) and were subjected to restriction enzyme digestion plus agarose gel electrophoresis and dideoxy sequencing using M13, T7, and internal UCP3 gene-specific primers. 3ЈRACE (rapid amplification of cDNA ends) was used to clone the 3Ј ends of UCP3 S and UCP3 L . 3ЈRACE was performed using the Marathon cDNA Amplification Kit, human skeletal muscle Marathon-Ready cDNA (both from CLONTECH) and a sense UCP3 primer (TCAGCCCCCTCGACTGTA) located in exon 6 (cDNA position relative to ATG ϭ ϩ761 to ϩ778).
Analysis of UCP3 S and UCP3 L mRNA Transcripts by RNase Protection Assay-RNase protection assays were performed as described previously (19) using two in vitro transcribed 32 P-labeled RNA antisense probes, one corresponding to UCP3 L , spanning exons 6 and 7 (ϩ631 to ϩ925 relative to ATG), and the other corresponding to UCP3 S , spanning exon 6 and the immediately adjacent UCP3 S 3ЈUTR (ϩ623 to ϩ900 relative to ATG).
UCP3 Chromosomal Localization-The Genebridge 4 Radiation Hybrid Panel (20 -22) was screened for the presence of hUCP3 (Research Genetics, Inc., Huntsville, AL) using the following PCR primer pair: sense ϭ GCGACAGAAAATACAGCGGGACTA (exon 4, cDNA position relative to ATG ϭ ϩ464 to ϩ487) and antisense ϭ GCAAAGGGCTG-GTAAAATGAACTG (intron 4, 192 to 169 bp downstream of the exon 4 splice donor). These primers amplified a 269-bp band from human genomic DNA and failed to amplify any signal from control, hamster genomic DNA.
Analysis of P1 and BAC Human and Mouse Genomic Clones for Colocalization of UCP2 and UCP3-P1 (human and mouse 129/ola) and BAC (mouse 129/SvJ) genomic libraries were screened (Genome Systems, St. Louis, MO) using gene-specific primers shown in Table II (P1  libraries) or a 32 P-labeled mUCP3 cDNA clone (BAC library). P1 and BAC DNA was isolated and analyzed for the presence of UCP2 and UCP3 using PCR (specific primer sets shown in Table II). Fig. 1, the human UCP3 coding sequence was found to be distributed over six exons (exons 2-7) spanning ϳ5.25 kb of genomic DNA. To obtain 5Ј upstream cDNA sequence of human UCP3, 5ЈRACE on human skeletal muscle Marathon cDNA was performed (16). Different clones were obtained and sequenced, and the longest ones were found to contain 183 bp 5Ј upstream of the ATG. Thus, at least one exon containing UCP3 5Ј-untranslated sequence was detected (exon 1). Sequence analysis indicated that the 3Ј-UTR of UCP3 S and the intron region between exon 6 and the AATAAA S in intron 6 were identical (see Fig. 1). The protein predicted to be generated by the UCP3 S transcript is truncated by an in-frame stop codon (tga) which follows a preserved glycine (G) codon (GGg) at residue position 275. This glycine codon in UCP3 L (GGA) is located at the splice junction between exons 6 and 7.

Intron-Exon Structure-As shown in
Analysis of UCP3 S and UCP3 L mRNA Transcripts by RNase Protection Assay-An RNase protection assay probe corresponding to the UCP3 L transcript, spanning exons 6 and 7, was prepared. This probe contained 193 bp of exon 6 sequence and 100 bp of exon 7 sequence. As is shown in Fig. 2, two bands were protected, one of ϳ290 bp representing UCP3 L and another of ϳ190 bp representing UCP3 S . Additional RNase protection assays were performed using a probe corresponding to UCP3 S (data not shown). This probe contained 200 bp of exon 6 and 77 bp of adjacent 3Ј sequence corresponding to the UCP3 S 3Ј-untranslated region (3ЈUTR S , see Fig. 1). As would be predicted, two protected bands were obtained, one of ϳ280 bp representing UCP3 S and another of ϳ200 bp representing UCP3 L (data not shown). Quantitation of RNase protection assay results using in vitro transcribed sense UCP3 transcripts as a standard curve and total RNA extracted from five lean subjects (rectus abdominis muscle) revealed that there were ϳ15 amol (per g of total RNA) of UCP3 L transcripts and ϳ18 amol (per g of total RNA) of UCP3 S transcripts.
UCP3 Chromosomal Localization-A hUCP3 PCR primer set (see "Experimental Procedures") was applied to the Genebridge 4 Radiation Hybrid Panel (20 -22) generating the following data set for hybrid clones 1 through 93 (0 ϭ no amplification, 1 ϭ amplification and 2 ϭ ambiguous results): 1001012001 0000010101 0000010000 0200112000 1110000001 0000100001 0000000000 0100100000 001. These data were submitted to the MIT Center for Genome Research STS mapping server. 2 UCP3 was mapped to chromosome 11q13 (distal portion), 1.31 cR (lod Ͼ 3.0) below framework marker WI-6189. WI-6189 maps to 387.58 cR from the top of the Chr 11 linkage group on the Whitehead Institute Center for Genome Research radiation hybrid map, between framework markers D11S916 (384 cR) and D11S911 (391 cR). D11S916 and D11S911 have also been positioned on the Généthon human genetic linkage map, 85 and 89 cM from the top of the Chr 11 linkage group, respectively (23). It has previously been noted (13) that two ESTs representing UCP2, WI-13873 (accession number R49188) and WI-16720 (accession number T80845), have been independently mapped to this region (385.84 and 387.58 cR, respectively, Whitehead Institute Center for Genome Research). See Fig. 3 for the order of markers in this region and the position of framework markers on the genetic map.
Analysis of P1 and BAC Human and Mouse Genomic Clones for Colocalization of UCP2 and UCP3-Given the proximity of human UCP2 and UCP3 by radiation hybrid mapping, we investigated whether human UCP2 and UCP3 might be found together on P1 genomic clones. P1 genomic clones generally have genomic inserts of ϳ75-100 kb. In addition, we also investigated whether mouse UCP2 and UCP3 might be found together on P1 and BAC mouse genomic clones. BAC genomic 2 http://www-genome.wi.mit.edu/cgi-bin/contig/rhmapper.pl. clones generally have a genomic insert of ϳ150 kb. Analysis for mUPC3 was possible because we had recently cloned its corresponding cDNA. 3 Mouse UCP3 is 87% identical to human UCP3 at the amino acid level, 3 but is only 55% identical to mUCP1 and 72% identical to mUCP2. As is shown in Table II

DISCUSSION
In the present study we have analyzed the human UCP3 gene. It contains at least 7 exons spread over ϳ8.5 kb and is located on chromosome 11 (11q13), adjacent to UCP2. The UCP3 gene generates two mRNA transcripts, UCP3 L and UCP3 S , which are predicted to encode long and short UCP3 proteins, differing only by the presence or absence of 37 residues on the C terminus (15). These 37 residues are encoded by exon 7 which is missing from UCP3 S . Intron 6 contains a cleavage and polyadenylation signal (designated AATAAA S in Fig. 1). The AATAAA S signal terminates message elongation ϳ50% of the time, thus generating UCP3 S . When the AATA-AA S signal is bypassed, which seems to occur ϳ50% of the time, message elongation continues until the AATAAA L signal (located ϳ1.1 kb downstream of exon 7) is reached, thus generating UCP3 L .
The domain encoded by exon 7 is highly homologous to Cterminal residues found in UCP1 and UCP2, thus UCP3 S is unique in lacking these residues. Since this region is believed to participate in purine nucleotide-mediated inhibition of UCP1 uncoupling activity (17,18), UCP3 S may have increased uncoupling activity. Alternatively, UCP3 S could have reduced activity or no activity due to the possible absence of critical residues. The biological significance of UCP3 S will need to be the focus of future investigations.
The human UCP3 gene maps to the distal segment of 11q13, adjacent to UCP2. Human UCP1, on the other hand, is located on chromosome 4 (24). In this context, it is noteworthy that UCP2 and UCP3 are more similar to each other than to UCP1.  1. Human UCP3 gene structure. Human UCP3 gene with start codon (ATG), stop codons (TGA S for UCP3 S and TGA L for UCP3 L ), and cleavage and poly(A) adenylation signals (AATAAA S for UCP3 S and AATAAA L for UCP3 L ) are shown above. Exons are coded from 1 through 7. 3Ј-Untranslated regions for UCP3 S and UCP3 L are shown as UTR S and UTR L , respectively. The GenBank TM accession numbers for each exon and flanking intronic sequences are consecutive from exon 1 to 7: AF012196, AF012197, AF012198, AF012199, AF012200, AF012201, and AF012202. Schematic cDNAs are shown below the gene structure. On the bottom is the exact location of the splice donors and splice acceptors (uppercase letters refer to exon sequence, lowercase letters refer to intron sequence). Amino acids adjacent to the splice sites are shown below the nucleotide sequence.

FIG. 2. UCP3 RNase protection assay.
A probe spanning exons 6 and 7 (see "Experimental Procedures" for details) was used to assess UCP3 S and UCP3 L mRNA expression in human skeletal muscle total RNA (isolated from quadriceps muscle, male subject, age 32, RNA purchased from CLONTECH, catalog number 64033-1). Total RNA ranging in amounts from 0 -10 g were assessed. A cyclophilin probe was used to control for quality of RNA. Expected size of each signal is shown.
Both mouse and human UCP2 and UCP3 genes colocalize on P1 and BAC genomic clones, indicating that these two UCPs are within 75-150 kb of each other. Given this and the high degree of similarity between UCP2 and UCP3 (ϳ70% at the nucleotide level), it is likely that one UCP gene arose from the other via a duplication event. However, despite their common origin and similar sequence, the two UCPs are unique, being distinguished by their different patterns of expression and the existence of a short form for UCP3, but not UCP2. 3 The close proximity of UCP2 and UCP3 and the similarity in nucleotide sequence have additional implications. Unequal crossovers during meiosis could generate alleles with deletions, duplications, or gene conversions of UCP2 and/or UCP3, as observed with ␣-globin (25), 21-hydroxylase (26), and 11␤-hydroxylase (27) genes. Also, the close proximity of UCP2 and UCP3 will prevent genetic linkage studies from discriminating between UCP2 and UCP3. Of interest, a prior study mapped mUCP2 to chromosome 7 (13), tightly linked to the tubby mutation. This region is coincident with a quantitative trait locus for obesity in three mouse models (28 -30) and one congenic strain (31). Since mUCP2 and mUCP3 are adjacent, it is possible that an abnormality in one or both of these genes is responsible for obesity. In humans, the Bardet-Beidl syndrome (BBS1, MIM#209901) consisting of retinal degeneration, polydactyly, hypogonadism, mental retardation, and obesity has been linked to 11q13 (32-34) (significant lod scores with markers D11S1883 and D11S913, see Fig. 3). However, in this case UCP2 and/or UCP3 are unlikely candidate genes given that they are positioned at least 12 cM distal to BBS1 (no significant linkage between BBS1 and marker D11S916).
In summary, genes for UCP2 and UCP3 are highly homologous and are located in close proximity on chromosome 11q13. UCP3 is distinguished from UCP2 and UCP1, however, by its selective and high level expression in skeletal muscle and the expression of a short form transcript, UCP3 S , generated by a cleavage and polyadenylation signal (AATAAA) located in the last intron. Given its proximity to the UCP2 gene, the mouse UCP3 gene is also coincident with 3 independently mapped quantitative trait loci for obesity (28 -30), raising the possibility that abnormalities in UCP3 are responsible for obesity in these models. Thus, human linkage studies for the UCP2/ UCP3 locus along with mutational analyses of mouse and human UCP2 and UCP3 genes should be the focus of future investigations.