Melanocortin 3 Receptor Has a 5′ Exon That Directs Translation of Apically Localized Protein From the Second In-Frame ATG

Abstract

Melanocortin-3 receptor (MC3R) is a canonical MSH receptor that plays an essential role in energy homeostasis. Variants in MC3R have been implicated in obesity in humans and mice. However, interpretation of the functional consequences of these variants is challenging because the translational start site of MC3R is unclear. Using 5′ rapid amplification of cDNA ends, we discovered a novel upstream exon that extends the length of the 5′ untranslated region (UTR) in MC3R without changing the open-reading frame. The full-length 5′ UTR directs utilization of an evolutionarily conserved second in-frame ATG as the primary translation start site. MC3R synthesized from the second ATG is localized to apical membranes of polarized Madin-Darby canine kidney cells, consistent with its function as a cell surface mediator of melanocortin signaling. Expression of MC3R causes relocalization of melanocortin receptor accessory protein 2, an accessory factor for melanocortin-2 receptor, to the apical membrane, coincident with the location of MC3R. In contrast, protein synthesized from MC3R cDNAs lacking the 5′ UTR displayed diffuse cytosolic distribution and has no effect on the distribution of melanocortin receptor accessory protein 2. Our findings demonstrate that a previously unannotated 5′ exon directs translation of MC3R protein that localizes to apical membranes of polarized cells. Together, our work provides insight on the structure of human MC3R and reveals a new pathway for regulation of energy metabolism.

The melanocortin receptor family, consisting of 5 G protein–coupled receptors (GPCRs), mediate key physiological functions, such as pigmentation, steroidogenesis, energy balance, and food intake (1). The melanocortin pathway is mediated by the products of the proopiomelanocortin gene: α-MSH, β-MSH, γ-MSH, and ACTH (2). Variations in the coding regions of melanocortin receptors have been associated with a number of medically important diseases and traits that result from dysregulated energy metabolism. For example, inactivating mutations of the melanocortin-2 receptor (MC2R) or melanocortin receptor accessory protein (MRAP) are known to cause familial glucocorticoid deficiency (3, 4), and mutations in melanocortin-4 receptor (MC4R) are responsible for 1% to 5% of early-onset human obesity (5–10). Like MC4R, melanocortin-3 receptor (MC3R) has been found to play a role in energy balance. Previous studies have reported several common and rare variants in MC3R that inconsistently associate with obesity (11–17). MC3R is thought to coordinate feeding-related behaviors and glucose metabolism through the action of the melanocortin signaling pathway in the brain (18–21). More recently, MC3R has been implicated as a modifier of cystic fibrosis (CF) (22). Because malnutrition is one of the major burdens in CF (23, 24), nutrition management can improve the survival rate in patients with CF. Although MC3R has been analyzed for association with various human traits that are caused by the failure to maintain energy homeostasis, the structure and the encoded product of MC3R have not been well characterized. In this study, we aimed to understand the molecular organization of MC3R by obtaining the full-length mRNA transcript, determining the native protein length, verifying its cellular distribution, and investigating the interaction with known accessory proteins.

The National Center for Biotechnology Information (NCBI) Reference Sequence Database predicts that MC3R is a single exon gene, and it is not clear whether translation begins at the first or second in-frame ATG. Although the second ATG is conserved across vertebrates, the first ATG is only conserved in nonhuman primates (www.genome.ucsc.edu). Prior studies indicate that both ATGs can function as translation initiation codons and that the sequence between the first and second ATG is not critical for the ligand binding of the receptor (25, 26). Recently, it has been shown that the second ATG is preferentially used as the translation initiation site in a truncated form of MC3R (27). Lack of information regarding the composition of the full-length mRNA transcript, as well as conflicting results on translation initiation, prompted us to identify native transcription and translation start sites to gain a more complete understanding of the MC3R protein.

Characterizing the localization of MC3R in a proper cell type is vital in deciphering its role in regulation of energy homeostasis. As a member of a family of proteins that signal through the GPCR pathway, MC3R should be localized to the plasma membrane to enable binding to its ligands. Several studies have reported cytoplasmic and membrane localization of MC3R in different cell types (11, 28, 29). One of the limitations of prior studies is that localization was assessed in nonpolarized cells. Such cell types do not represent the true physiological state of MC3R because it is known to be expressed in tissues, such as the brain, where polarized neurons have a distinct distribution of membrane proteins (30). Finally, all melanocortin receptors, including MC3R, have demonstrated interaction with accessory proteins, MRAP and its paralogue MRAP2 (3). In this study, we provide evidence that MC3R and MRAP2 colocalize in polarized Madin-Darby canine kidney (MDCK) cells, suggesting a potential role for MRAP2 in regulating the function of MC3R. The molecular characterization of MC3R provides information essential to resolving its role in energy metabolism.

Materials and Methods

Cell culture

Human bronchial epithelial (HBE) cells and cystic fibrosis bronchial epithelial (CFBE41o-) cells carrying a homozygous F508del cystic fibrosis transmembrane conductance regulator (CFTR) mutation were maintained in MEM (Gibco). The Flp-In Chinese hamster ovary (CHO) cells were maintained in Ham's F-12 medium (Gibco) with 100 μg/mL Zeocin (Invitrogen). MDCK cells were maintained in DMEM (Cellgro). MDCK cells stably expressing the CFTR tagged with green fluorescent protein (GFP) at the N terminus (GFP-WT-CFTR) were maintained in DMEM with 100 μg/mL hygromycin (Invitrogen). All media were supplemented with 10% fetal bovine serum (Cellgro) and 1% penicillin-streptomycin (Gibco) at 37°C in an atmosphere of 5% CO₂.

RNA isolation

Total RNA was isolated from human brain tissue, HBE cells, and CFBE41o- cells by using an RNeasy Mini Kit (QIAGEN) as described by manufacturer's protocol. It was then treated with DNA-free DNase (Ambion) to remove contaminating DNA as described by manufacturer's protocol.

5′ and 3′ rapid amplification of cDNA ends (RACE)

5′ and 3′ RACE were performed using a SMARTer RACE cDNA Amplification Kit (Clontech), in accordance with the manufacturer's protocol. The mRNA extracted from CFBE41o- cells was used for RACE. The gene-specific primers were designed on the plus and minus strands of the reference sequence of MC3R (NM_019888). MC3R cDNA was amplified with a universal primer and an MC3R-specific primer. Nested PCR amplification was subsequently performed with the nested universal primer and the nested MC3R-specific primer to confirm the specificity of the primary PCR products. The nested PCR products were gel purified and sequenced with an 3730xl DNA analyzer (Applied Biosystems).

RT-PCR

DNase-treated RNA was reverse-transcribed using an iScript cDNA Synthesis Kit (Bio-Rad Laboratories). PCR was performed using Taq DNA polymerase (Invitrogen). After a 6-minute denaturation at 95°C, 40 cycles of 95°C for 45 seconds, 59°C for 45 seconds, and 72°C for 1 minute were performed, followed by a 10-minute extension at 72°C. RT-PCR products were visualized by agarose gel electrophoresis with ethidium bromide staining.

Plasmids

The full-length MC3R plasmid was generated by amplifying exon 1 and exon 2 of MC3R from genomic DNA (gDNA). The 260-bp exon 1 band and 1131-bp exon 2 band were extracted from the gel and reamplified via fusion PCR to connect the 2 overhanging products. The 1391-bp fusion amplicon was then subcloned into a StrataClone mammalian expression vector, pCMV-SC-CM, that contains a C-terminal c-Myc epitope tag vector (Agilent Technologies). The parMC3R-MYC plasmid was created by subcloning the known reference sequence of MC3R (NM_019888) into a StrataClone mammalian expression vector with a C-terminal c-Myc tag, pCMV-SC-CM (Agilent Technologies). The sequences of each construct were verified by Sanger sequencing. The GFP-WT-CFTR plasmid was generated by removing full-length cDNA from the existing peGFP-CFTR plasmid by enzyme digestion and ligating into the multiple cloning sequence of the pcDNA5/FRT vector (Invitrogen). The construction of the pRK5-GFP plasmid was described previously (31). The pcDNA3.1 plasmid encoding partial MC3R with a 3× hemagglutinin (HA) tag on the N terminus (HA-parMC3R) was purchased from Missouri S&T cDNA Resource Center (www.cdna.org). MRAP-FLAG and MRAP2-FLAG plasmids were generous gifts from Dr. Li Chan (Centre for Endocrinology, William Harvey Research Institute, London, UK).

Site-directed mutagenesis

The first and/or second in-frame ATG was changed to GCG using a QuikChange II XL Site-Directed Mutagenesis Kit (Agilent Technologies), according to the manufacturer's protocol. To prevent the c-Myc tag of parMC3R-MYC plasmid from being expressed, nonsense mutations were introduced at the end of the partial MC3R sequence.

Transfection

The cells were plated on the glass coverslips inside 12-well plates. When the cells were nearly 90% confluent, they were transiently transfected with 3.2 μg of a single plasmid or cotransfected with 1.6 μg of each of the 2 plasmids along with 4.8 μL of Lipofectamine 2000 (Invitrogen), according to the manufacturer's protocol. Cells were fixed 2 days after transfection.

Immunocytochemistry and confocal microscopy

Two days after transient transfection, cells were fixed in 3.7% formaldehyde for 20 minutes. Staining with wheat germ agglutinin (WGA) (W11262; Molecular Probes) was performed before permeabilization by incubating the cells in diluted WGA (1:100) for 5 minutes. Staining with antibodies was performed after the cells were permeabilized with 0.5% Triton X-100 for 5 minutes and blocked with 2.5% goat serum for 30 minutes. The following primary antibodies were used for various proteins: HA (Sigma-Aldrich), c-Myc (Sigma-Aldrich), FLAG (Sigma-Aldrich), MC3R (sc-8990; Santa Cruz Biotechnology), GFP (Molecular Probes), ZO-1 (Invitrogen), and Na⁺/K⁺-ATPase (Millipore). The secondary antibodies were conjugated, respectively, to anti-mouse or anti-rabbit Alexa Flour 488 (Invitrogen) or Cy3 (Sigma-Aldrich). The primary and secondary antibodies were diluted to 1:200 and 1:50, respectively, in blocking buffer. The cells were incubated with primary antibodies for 1 hour, washed 3 times for 5 minutes each, incubated with secondary antibodies for 1 hour, and washed 4 times for 15 minutes each. ProLong Gold Antifade Reagent with DAPI (Molecular Probes) was used to stain the nuclei and mount the cells. Fluorescence was imaged using a Zeiss LSM510 confocal microscope at the Johns Hopkins Microscope Facility.

Results

Human MC3R gene has 2 exons

The NCBI Reference Sequence Database predicts MC3R to be a single exon gene and its transcript as intronless (NM_019888). The reference sequence of MC3R places the first in-frame ATG at bp 2 to 4, whereas the second and the third in-frame ATGs are located 37 and 109 codons downstream, respectively (Figure 1A). Because none of these ATGs has the characteristics of a typical translation start site (ie, Kozak consensus sequence), we performed 5′ and 3′ RACE to determine the full-length of the MC3R mRNA transcript (Supplemental Table 1). Because we are exploring the role of MC3R in human lung diseases, we performed RACE upon mRNA extracted from bronchial epithelial cells from a CF patient (CFBE41o- cells).

Figure 1.

Mapping of the 5′ transcription start site and the gene structure of MC3R. A, Nucleotide locations of the first 3 in-frame start codons in the reference sequence of MC3R (NM_019888) are annotated using the February 2009 (hg19) assembly of the human genome. The underlined sequence around each start codon demarcates the region inspected for correspondence to the Kozak consensus sequence (GCCGCCATGG). B, This diagram summarizes the gene structure of MC3R deduced from 5′ and 3′ RACE experiments. 5′ RACE found a novel exon upstream of the reference sequence of MC3R. A start codon was not detected in-frame. The original exon is 27 bp longer than previously predicted. The 252+-bp upstream exon (labeled now as exon 1), 2480-bp intron, and 1111-bp original exon (relabeled as exon 2) all contained splicing consensus sequences. The diagram also illustrates the positions of the 3 pairs of MC3R primers used in RT-PCR. C, Image shows agarose gel electrophoresis of ethidium bromide–stained gDNA and cDNA fragments amplified by PCR. RT-PCR demonstrated that MC3R mRNAs from human brain tissue, HBE cells, and CFBE41o- cells include sequences from exon 1 and exon 2. The label at the top of each lane of the gel represents the different primer that was used to amplify the product. Before RT, RNA was treated with DNase to eliminate the contaminating gDNA. The TBP primers were designed from 2 different exons with an intervening intron that is short enough to be amplified during elongation. The size of the TBP band indicated that there is no gDNA contamination in the cDNA samples (797 bp in gDNA and 116 bp in cDNA).

The 5′ RACE mapped the start site of transcription approximately 527 bp 5′ of the reference sequence of MC3R. The full-length transcript contains an additional 5′ exon of ∼252 bp, labeled now as exon 1, and an intron of 248 bp (Figure 1B). The original exon, relabeled as exon 2, is 27 bp longer than previously predicted. The 3′ RACE verified that MC3R does not have any downstream exon and that a poly(A) tail is added 116 bp downstream of the stop codon. The highly conserved canonical poly(A) signal AAUAAA was not found in the 3′ UTR. However, 2 potential poly(A) signals, AAGAAA and UAUAAA, were found 46 and 21 bp from the poly(A) tail, respectively.

To confirm the RACE results, RT-PCR was performed using primers positioned in exons 1 and 2 of MC3R (Supplemental Table 2). DNA fragments of the expected size were obtained upon amplification of gDNA from human lymphocytes and cDNA synthesized from primary human brain tissue (temporal lobe) mRNA (Figure 1C). In gDNA and brain cDNA, the MC3R primer pair 1 amplified a fragment of 1084 bp that corresponded to the length expected from exon 2. Primer pair 2 amplified a 921-bp gDNA fragment and a 673-bp brain cDNA fragment that were extended from exon 1 to exon 2, consistent with the presence or excision, respectively, of the 248-bp intron. Primer pair 3 amplified a 476-bp brain cDNA fragment that extended from the exon 1 to exon 2 junction to exon 2, indicative of intron splicing. Primer pair 3 generated only nonspecific products upon amplification of gDNA, as expected. These results show that MC3R contains an upstream intron that is excised in brain mRNA. The cDNAs synthesized from HBE and CFBE41o- cell RNA were also studied. Each cell line generated amplification products identical to those obtained from amplification of brain cDNA (Figure 1C). Thus, MC3R is composed of 2 exons that are spliced to form an mRNA of approximately 1363 bp in primary brain tissue and in pulmonary airway epithelial cells.

Translation is initiated at the second in-frame ATG of MC3R

To determine whether translation initiated at the first or second in-frame ATG (or elsewhere), we created a plasmid containing a full-length MC3R cDNA and then mutated the first 2 ATG codons to GCG (Supplemental Table 3). The full-length MC3R cDNA was created by joining exon 1 and exon 2 of MC3R via fusion PCR and was fused in-frame to a c-Myc epitope at the C terminus to facilitate detection of translated protein. After the introduction of mutations to the first and/or second ATG and sequence verification, the Flp-In CHO cells were transiently transfected with each plasmid, and expressed protein was detected with either an anti-MC3R polyclonal antibody directed against an 88-amino acid peptide from the N terminus of MC3R (Figure 2, B–E) or an anti-c-Myc antibody (Figure 2, F–I) or both (Figure 2A). As a control for transfection, a plasmid encoding GFP was cotransfected with each MC3R construct (Figure 2, B–I).

Figure 2.

MC3R translation begins at the second methionine. Fluorescent photomicrographs of CHO cells in the x-y plane. Cells were transiently transfected with the MC3R wild-type plasmid or a plasmid that has the first and/or second ATG mutated to GCG and cotransfected with GFP as a transfection control (B–I). Cells were stained with both anti-MC3R antibody and anti-c-Myc antibody to confirm the expression of a fusion protein, MC3R-myc (A). Cells were stained with anti-MC3R antibody (B–E) or anti-c-Myc antibody (F–I) with a secondary antibody linked to Cy3 (red) to detect MC3R-myc. Cells were also stained with anti-GFP antibody and a secondary antibody linked to Alexa Fluor 488 (green) to detect GFP (B–I). The green fluorescence indicated that GFP was expressed in approximately 15% of cells in each transfection. In the merge row, red and green fluorescence were combined and 4′,6-diamidino-2-phenylindole was used to stain the cell nuclei. A, Fluorescence from the anti-MC3R and anti-c-Myc staining overlaps in every transfected cell. B, Wild-type MC3R is detected in most cells that express GFP. C, When the first ATG is mutated, MC3R is detected in about 90% of cells expressing GFP. D, MC3R is not detected when the second ATG is mutated. E, When both the first and second ATG are mutated, MC3R is not detected. F–I, The same staining pattern is observed when anti-c-Myc antibody is used instead of anti-MC3R antibody. Scale bars correspond to 50 μm.

To validate the specificity of the anti-MC3R antibody, it was used in conjunction with the anti-c-Myc antibody to detect an MC3R-myc fusion protein (Figure 2A). The fluorescence from both antibodies overlapped in every transfected cell, which verified the specificity of the anti-MC3R antibody. Confocal microscopy revealed that MC3R protein is detectable in similar numbers of cells when the first ATG is intact or mutated (Figure 2, B and C). Specifically, wild-type MC3R was expressed in 59 of 65 GFP-expressing cells and MC3R with the first ATG mutated was detected in 58 of 60 GFP-expressing cells. However, protein was not detected in 40 GFP-expressing cells when the second ATG was mutated (Figure 2D) or in 45 GFP-expressing cells when both ATGs were mutated (Figure 2E). Transfection efficiency based on GFP staining ranged from 12.2% to 15.0%, indicating that failure to detect MC3R protein when the second ATG was mutated was not due to transfection issues. To verify that the anti-MC3R antibody was detecting MC3R-myc protein, the experiment was repeated using the anti-c-Myc antibody (Figure 2, F–I). Protein was detected by the anti-c-Myc antibody in cells transfected with the wild-type MC3R and MC3R with first ATG mutated but not when the second ATG or both ATGs of MC3R were mutated (Figure 2, F–I). Taken together, these results indicate that translation of MC3R is initiated at the second ATG and the full-length protein encompasses 323 amino acids. This conclusion is consistent with cross-species conservation of the second but not the first ATG of MC3R and with studies of truncated 5′ MC3R fused to reporter protein that demonstrated preferential initiation from the second ATG (27).

MC3R displays discrete apical and subapical localization in polarized MDCK cells

Expression of MC3R in nonpolarized cells revealed membrane localization, as expected for a transmembrane signaling protein (4, 11, 28). MC3R is abundantly expressed in the neurons of adult rat brain (32), which are polarized cells that rely on discrete localization of specific membrane proteins (ie, axonal vs dendritic). Polarized MDCK cells distribute proteins in a pattern consistent with that of polarized neuronal cells because both cell types share many common mechanisms of protein targeting (ie, apical/axonal and basolateral/dendritic) (33–35). To assess whether MC3R localizes to specific membrane compartments, the location of MC3R was investigated by confocal microscopy of polarized MDCK cells.

Polarization of MDCK cells was verified by staining with anti-ZO-1 antibody, a marker for tight junctions, or anti-Na⁺/K⁺-ATPase antibody, a marker for the basolateral membrane surface. To mark the location of the apical membrane, the cells were cotransfected with a plasmid expressing GFP-WT-CFTR, a protein known to localize to the apical surface of MDCK cells (36–38). Confocal imaging of these cells revealed that the proteins were present in their expected locations (Figure 3, A and B). When the cells were not transfected with MC3R or with the GFP-WT-CFTR plasmid, there was an absence of staining using either anti-MC3R antibody or anti-GFP antibody (Figure 3C). The cDNA encoding native MC3R was used in Figure 3, D and E, to avoid aberrant localization due to the presence of an epitope tag. In 11 of 16 cells examined, MC3R displayed localization that was coincident with CFTR. The overlap of 2 proteins resulted in yellow fluorescence in the merge, demonstrating that MC3R is localized at the apical surface (Figure 3D). In the remaining 5 cells, MC3R demonstrated localization to regions near the apical membrane; however, the distribution did not overlap with CFTR (Figure 3E). Thus, in polarized cells, MC3R primarily appears to be located either in the same region as CFTR or, less often, in a subapical compartment.

Figure 3.

MC3R exhibits apical localization in polarized MDCK cells. Confocal microscopy images of MDCK cells taken in the x-z plane with the bottom row in each panel representing the composite scan and the panels above showing red or green fluorescence. In the merge row, cells were counterstained with 4′,6-diamidino-2-phenylindole to detect the nuclei. A, Cells were immunostained with anti-ZO-1 antibody (red), a tight junction marker, to demonstrate that apical-basolateral polarity is present. B, Immunostaining of the MDCK cells for Na⁺/K⁺-ATPase showed staining consistent with localization to basolateral membranes. C, When the cells are not transfected with MC3R or GFP-WT-CFTR plasmids, there is a lack of staining with the anti-MC3R antibody and anti-GFP antibody. D, Localization of MC3R overlapped with that of CFTR in most cells examined. E, Localization of MC3R near the apical membrane did not overlap with CFTR in the minority of cells examined. Scale bars correspond to 10 μm. AP, apical; BL, basolateral.

Absence of the 5′ UTR results in translation initiation at the first in-frame ATG and synthesis of protein that is aberrantly localized in polarized MDCK cells

To investigate whether the newly discovered 5′ UTR was responsible for the preferential use of the second in-frame ATG for translation initiation, exon 1 and 27 bp from the 5′ end of exon 2 were removed to create a partial version of MC3R (parMC3R). The partial version corresponds precisely to the 1084-bp single exon of MC3R that is annotated by the NCBI RefSeq Database (NM_019888). The parMC3R cDNA was subcloned into a vector with a C-terminal c-Myc tag to enable detection in transfected cells. The protein translated from the parMC3R cDNA displayed a cytoplasmic distribution in polarized MDCK cells (Figure 4A) that was distinctly different from protein translated from the full-length cDNA (Figure 3, D and E). The detection of protein synthesized from the parMC3R cDNA by both the N-terminal antisera and the C-terminal epitope tag indicated that translation initiated somewhere within the first 88 amino acids and continued to the native C terminus. As only the first and second in-frame ATGs lie within the region recognized by the N-terminal antisera, we mutated each ATG to assess where translation was initiated in the parMC3R cDNA. When the first ATG of this partial construct was mutated to GCG, the translated protein localized to the apical regions of the polarized MDCK cells as inferred by the location of ZO-1, a protein that marks tight junctions and plays an integral role in the maintenance of apicobasal polarity (Figure 4B). The same pattern was noted for protein translated from the second ATG of the full-length MC3R cDNA (Figure 3, D and E). When the second ATG of the partial construct was mutated to GCG, confocal microscopy revealed MC3R fluorescence above as well as below the location of ZO-1, which suggests broad cytoplasmic distribution (Figure 4C). From these results, we conclude that translation of parMC3R cDNA is initiated from the first ATG, generating a protein that does not localize to the cell membrane.

Figure 4.

In absence of the 5′ UTR, MC3R is mislocalized to the cytoplasm of polarized cells. Confocal microscopy images of MDCK cells taken in the x-z plane. A, MDCK cells were immunostained with anti-MC3R antisera (red) and anti-c-Myc antibodies (green) to detect the expression of a fusion protein, parMC3R-MYC. Protein synthesized from the partial exon 2 showed diffuse cytoplasmic distribution in polarized MDCK cells. Fluorescence from the anti-MC3R and anti-c-Myc staining overlapped almost completely (see merge panel). B, When the first ATG of the partial construct was mutated to GCG, this led to translation initiation from the second ATG. Immunostaining with anti-ZO-1 antibody illustrated that this form of MC3R protein has discrete apical localization. C, A form of MC3R is detected when the second ATG of the partial construct is mutated to GCG, but this protein displays diffuse cytoplasmic localization. D, Staining with WGA outlined the plasma membrane. Immunostaining with anti-MC3R antisera revealed that the untagged parMC3R has cytoplasmic localization under the cell membrane. E, MDCK cells stably expressing GFP-WT-CFTR were immunostained with anti-GFP antibody to detect CFTR and with anti-HA antibody to detect HA-parMC3R. GFP-WT-CFTR protein was used as a marker for apical surface expression. The commercially available HA-parMC3R plasmid is missing the 5′ UTR and has the partial exon 2 of MC3R fused in-frame to the HA tag at the N terminus, thus forcing the inclusion of the amino acids from the first to the second ATG. A diffuse cytoplasmic stain was observed for HA-parMC3R. These experiments reveal that the partial exon 2 of MC3R encodes aberrant forms of the protein that are distributed in the cytoplasm of polarized cells, which is distinctly different from the localization of the protein encoded from the full-length MC3R. Scale bars correspond to 10 μm. AP, apical; BL, basolateral.

Further studies were performed to exclude alternative explanations for the cytoplasmic location of MC3R protein translated from the first ATG. To test whether the C-terminal tag interfered with protein biogenesis, the tag was removed (Supplemental Table 3). MDCK cells transfected with the untagged parMC3R plasmid were stained with WGA, a plasma membrane marker, and with the N-terminal anti-MC3R antisera. Protein encoded by the untagged parMC3R cDNA was cytoplasmic and did not colocalize with the cell membrane marker (Figure 4D). We then transfected MDCK cells with a commercially available construct, labeled HA-parMC3R, that contains the partial exon 2 of MC3R fused in-frame to the HA epitope at the N terminus (www.cdna.org). This construct was selected because it has been used in a number of functional studies of MC3R (3, 28, 39–41). Translated products from this vector include amino acids from the first ATG to the second ATG that we predict are not present in native MC3R. Staining of transfected cells with the anti-HA antibody revealed a diffuse cytoplasmic distribution in polarized MDCK cells that was distinct from apically located GFP-WT-CFTR (Figure 4E). Together, these studies indicate that the 5′ UTR from exon 1 and the 5′ of exon 2 are essential for directing translation of a membrane localizing form of MC3R that begins at the second in-frame ATG.

MRAP2 colocalizes with MC3R in polarized cells

It has been shown that accessory proteins, MRAP and MRAP2, support cell surface expression of MC2R in nonpolarized cells and interact with melanocortin receptors 1 to 5 to modulate signaling to intracellular second messengers (3). These single-pass transmembrane accessory proteins have been shown to localize at the cell surface as well as to the endoplasmic reticulum (42, 43). However, the location of MRAP and MRAP2 in polarized cells has not been established. In addition, it is not known whether the accessory proteins colocalize with melanocortin receptors in polarized cells. To address this issue, MRAP and MRAP2 were expressed in polarized MDCK cells by themselves and in the presence of MC3R and HA-parMC3R. The HA-tagged version of MC3R was used because it localizes to a cellular compartment different from that for native MC3R.

MRAP and MRAP2 displayed diffuse cytosolic distribution in polarized MDCK cells (Figure 5, A and B). The localization of MC3R remained the same in the presence of the accessory proteins, as it was observed at or near the apical surface of the cell (Figure 5, C–E). Interestingly, the distribution of MRAP2 was substantially different when it was coexpressed with MC3R, whereas MRAP distribution was unchanged (Figure 5, D and E). MRAP2 was found at discrete locations that partially overlapped with MC3R in the apical membrane (see merged x-z image in Figure 5E). On the other hand, HA-parMC3R did not affect the localization of MRAP2. MRAP and MRAP2 remained cytoplasmic when they were coexpressed with HA-parMC3R (Figure 5, G–H). These results suggest that MC3R translated from the second ATG is in the same cellular compartment as MRAP2.

Figure 5.

MRAP2 is redistributed in the presence of MC3R. Confocal microscopy images of MDCK cells taken in the x-y and x-z planes. Each photomicrograph shows immunostained cells after transfection with the plasmid listed on the top (none, MRAP-FLAG, or MRAP2-FLAG) and the plasmid listed across the left (none, MC3R or HA-parMC3R). The epitopes (HA and FLAG), Na⁺/K⁺-ATPase, ZO-1, and MC3R along the top row and left column are labeled according to the color of the fluorescence. Nuclei were counterstained with 4′,6-diamidino-2-phenylindole in blue. For detection of proteins in the apical compartment, x-y sections were above the nucleus. For detection of cytoplasmic or basolateral proteins, x-y sections transected through the nucleus. A, MRAP alone displays a diffuse cytoplasmic distribution. Red staining in a “chicken wire” pattern corresponds to the staining of ZO-1 in tight junctions formed by polarized cells. B, MRAP2 alone displays diffuse cytoplasmic staining in polarized MDCK cells. C, MC3R is localized near the apical membrane and is distinct from the localization of the basolateral protein Na⁺/K⁺-ATPase. D, MRAP displays diffuse cytoplasmic distribution, whereas coexpressed MC3R is localized near the apical membrane. E, MRAP2 is localized to the plasma membrane in the presence of MC3R. F, HA-parMC3R is expressed in the cytoplasm of polarized MDCK cells. G, MRAP displays diffuse cytoplasmic distribution when coexpressed with HA-parMC3R. H, MRAP2 localization is unchanged in the presence of HA-parMC3R. Scale bars correspond to 10 μm. AP, apical; BL, basolateral.

Discussion

In this study, we characterized the transcription and translation start sites of human MC3R and determined its localization in the presence or absence of MRAPs in polarized cells. Our results indicate that MC3R is a 2-exon gene that requires a 5′ UTR for translation initiation from the second in-frame ATG. Furthermore, MC3R translated from the second ATG is apically located, whereas protein that initiates at the first ATG remains cytoplasmic in polarized cells. Colocalization of MC3R and MRAP2 at the cell membrane raises the possibility that MRAP2, in addition to being an accessory factor for MC2R, might affect the cellular function of MC3R. Because several variants in MC3R have already been implicated in obesity (11–14, 18–21, 29), elucidation of the MC3R gene structure and protein localization is a critical step in interpreting its role in energy homeostasis.

Incomplete annotation of the 5′ regions of human as well as mouse genomes illustrates an urgent need to experimentally validate the gene architecture (44–48). Experimental examination of transcriptional start sites of 106 mouse genes by 5′ RACE revealed that more than half of the genes produced sequences that were longer than the predicted annotation (44). In the case of MC3R, we identified a novel upstream exon that extends the length of the 5′ UTR without changing the open-reading frame (ORF). The discovery of a noncoding first exon of MC3R should not be a surprise because approximately 40% of the known human genes have noncoding first exons (49). Determining the transcription start site of MC3R and therefore its promoter region is of great interest because MC3R is involved in the regulation of numerous physiological processes. Although it is known that both enhancer and suppressor elements can be found tens of thousands of bases upstream or downstream from the transcription start site (50), most essential control elements are usually present within the proximal promoter (51). Moreover, UTRs have been known to regulate the synthesis of a protein by affecting mRNA stability (52), translation efficiency (53), and protein trafficking (54). Changes in the length or the sequence of the 5′ UTR have been implicated in various human diseases, such as hereditary thrombocythemia (55) and X-linked Charcot-Marie-Tooth disease (56). Identification of the native transcription start site is essential in defining the location of the promoter and exploring cis-regulatory elements that control gene expression.

Before the discovery of this upstream exon, the first ATG was an unlikely translation start site because the 5′ UTR, consisting of one nucleotide, would not provide enough space for binding of various trans-acting assembly factors. Although evolutionarily conserved across vertebrates, the second ATG is followed by a poor Kozak sequence and lacks an optimal context for ribosome recognition. In February 2011, Ensembl listed 2 transcripts under MC3R (http://feb2011.archive.ensembl.org/), but the second transcript was removed in April 2011 (http://apr2011.archive.ensembl.org/). Interestingly, both MC3R transcripts were predicted to contain only one exon but differed in their translation start sites. The MC3R transcript that is currently listed on Ensembl labels the second in-frame ATG as the translation start site based on cross-species conservation. Despite the current annotation, most of the previous literature on MC3R assumes that translation begins at the first ATG, and, therefore, the single nucleotide polymorphisms are named by counting from the first methionine. For example, 2 common MC3R variants, rs3746619 and rs3827103, that have been frequently investigated are referred to as the T6K or V81I variant, respectively (16, 17, 57–60). Because our results confirm that translation of MC3R begins at the second ATG, located 37 amino acids downstream of the first ATG, rs3746619 (ie, T6K) should be considered a noncoding variant in the 5′ UTR rather than a missense variant.

Although it is thought that translation initiates at the first ATG in most eukaryotic genes, upstream ATGs are found in 15% to 50% of the 5′ UTRs, depending on the organism (61). This observation reveals that deviations from the “first-ATG rule” may be more common than we may have recognized. There are several possible reasons that translation initiates at the second ATG in MC3R. The sequence context around the first ATG may be far from optimal for initiating translation, resulting in leaky scanning of the ribosome (62). A hairpin secondary structure downstream of the first ATG may stall the ribosome and increase the likelihood of recognizing the second ATG (63). The upstream ATG found in the 5′ UTR of MC3R may be involved in ensuring a low basal translational level, as it has been demonstrated that cDNAs with long 5′ UTRs with several upstream ATGs have a weak start context (64, 65). Moreover, upstream ORFs have been correlated with significantly reduced protein expression of the downstream ORF (66). Given that variants that create a polymorphic upstream ORF, which alters cellular expression of the downstream protein, have already been identified in MC2R (66), discovery of such variants in MC3R can potentially explain the mechanism that affects protein expression and perhaps phenotypic variation.

The significance of the 5′ UTR in MC3R goes beyond regulating its translation initiation because our data indicate that protein localization is also affected by the methionine that is used to initiate translation. To investigate the localization of MC3R, we used polarized MDCK cells that have been extensively utilized as a model for studying the sorting of membrane proteins (33, 67, 68). Our study demonstrates that MC3R is localized at or near the apical surface in polarized MDCK cells. In contrast, MC3R produced from partial constructs that lack the 5′ UTR displays cytoplasmic distribution, indicating that the protein is unable to efficiently localize to the cell surface. Previous studies with the parMC3R constructs and the commercially available HA-parMC3R plasmid to study localization in nonpolarized cell types concluded that this form of MC3R displays cell surface as well as intracellular expression (11, 28, 29). We have demonstrated that the absence of the 5′ UTR causes incorrect use of the first ATG for translation initiation. Therefore, we conclude that the partial constructs and the HA-parMC3R plasmid synthesize a protein that incorrectly includes the sequence between the 2 start codons, resulting in a distinctively different localization pattern compared with that of native MC3R. Whereas epitope tagging can be powerful and convenient in rapid analysis of protein function, it can interfere with the function or cellular processing of the tagged protein. For example, it has been shown that GFP tagging of the human acetylcholine receptor alters the channel property in Xenopus oocytes (69), and glycosylation of angiotensin type II receptors was affected by epitope tagging in HEK293 cells (70). This observation emphasizes the importance of understanding the potential impact of epitope tagging on protein expression and use of constructs that represent the true context of the native protein.

The signaling and trafficking properties of GPCRs are regulated by the receptor-interacting proteins that are differentially expressed in distinct cell types (71). Our study reveals colocalization of MC3R and MRAP2, suggesting that MRAP2 may have additional roles besides being an accessory protein for MC2R. This result is consistent with that in a previous study, which demonstrated that MRAPs can down-regulate the expression and signaling of MC3R (3). Given these results, it is possible that apical localization of MC3R is due to the presence of endogenous MRAP2 in MDCK cells. However, MRAP2 expression is confined to the adrenal gland and brain and is absent in the kidney, the organ type from which MDCK cells are derived (3). Although colocalization of MC3R and MRAP2 does not imply physical interactions between the 2 proteins, it is possible that MC3R and MRAP2 indirectly interact with each other by binding to an intermediate protein or by acting in the same signaling pathway. These studies raise the questions of how and why MRAP2 interacts with other melanocortin receptors. It is possible that this interaction may vary in a cell type-specific manner and in response to changes in the environment. In contrast to MRAP2, MRAP does not colocalize with MC3R at the plasma membrane in MDCK cells. Additional factors, such as unidentified accessory proteins, may be required to facilitate trafficking of MC3R and MRAP to the cell surface in this specific polarized cell system. Because our data indicate that MRAP2 could be an accessory protein that influences functional expression of MC3R, provided that these proteins are coexpressed in vivo, further work will be needed to investigate the mechanism by which MRAP2 modulates the function of MC3R.

In summary, our results indicate that MC3R is a 2-exon gene that requires a 5′ UTR for translation, localization, and potential interaction with MRAP2. Our study defines the native gene structure and the protein expression of MC3R, an important member of the melanocortin family that plays an integral role in energy homeostasis and is implicated as a risk factor for obesity.

Additional material

Supplementary data supplied by authors.