ETL, a Novel Seven-transmembrane Receptor That Is Developmentally Regulated in the Heart ETL IS A

,

A complex series of events takes place during growth and maturation of cardiac myocytes. The proliferative growth of cardiac myocytes is primarily limited to fetal and early neonatal periods of development (1). Postnatal maturation of cardiac myocytes is marked by cellular hypertrophy and is also accompanied by ventricular remodeling of the nonmyocyte compartment, such as extracellular matrix formation and coronary angiogenesis (2). The signals that coordinate these processes in cardiac muscle are not well understood, but several growth factors and hormones have been shown to influence heart development (3,4). It is becoming clear that developmentally regulated gene expression of specific extracellular factors and their cognate receptors contributes to cardiac muscle differentiation (4 -9).
Complex cellular responses, such as proliferation and differentiation, are frequently modulated by external stimuli. Intracellular signaling cascades, in turn, act as mediators to translate the stimulus into transcriptional activity. A large family of receptors involved in a broad spectrum of cell signaling is the G-protein-coupled seven-transmembrane (TM7) 1 receptor (GPCR) family. This family of molecules mediates signals from hormones, cytokines, light, and odorants (10). GPCRs, activated by humoral, endothelial, or platelet-derived factors, are also able to stimulate mitogen-activated protein kinase pathways (11,12), signaling intermediates involved in cellular mitogenesis and proliferation.
GPCRs have a common topology characterized by an extracellular N terminus, seven membrane-spanning helices flanked by a cytoplasmic tail. A group of receptors that shares homology in the heptahelical region and activated by peptide hormones is the secretin receptor family (13,14). Recently, a subfamily of secretins has emerged that exhibit cell-surface interaction and cell adhesion modules in unusually large extracellular domains (15)(16)(17)(18)(19). This novel subtype of GPCRs consists of a small number of EGF-TM7 receptors such as EMR1, a receptor of neuroectodermal origin (20), its mouse homolog, F4/80 (21), and CD97, a leukocyte-activating antigen (22). All three receptors contain EGF modules and mucin-like domains in the N terminus. The recently discovered Celsr1 gene also belongs to the EGF-TM7 group. In addition to EGF repeats, Celsr1 contains cadherin and laminin type repeats (16).
In an effort to pinpoint the stimuli and signal transduction machinery that regulate the transition of myocyte from hyperplasia to hypertrophy, coronary capillary formation, and extracellular matrix deposition in the mammalian heart, we conducted differential display analysis on mRNAs using fetal and adult cardiomyocytes. As a result, we have identified a new member of the EGF-TM7 receptor family named ETL (for EGF-TM7-latrophilin-related protein). The large extracellular domain of rat ETL consists of EGF modules, a Ser/Thr rich linker region, and a Cys-rich proteolysis domain. A seventransmembrane region is followed by a short cytoplasmic tail. Besides having structural homology to the EGF-TM7 family, ETL shares considerable similarity with closely related heptahelical receptors CL1 (calcium-independent receptor for latro-toxin (CIRL)/latrophilin 1), CL2, and CL3 (18,19). The similarity with latrophilins includes a Cys-rich domain that may direct endoproteolytic cleavage of the extracellular domain. The expression of the ETL mRNA is developmentally regulated in the heart, suggesting that ETL seven-transmembrane receptors may be important in cardiac switching from fetal to adult phenotypes.

EXPERIMENTAL PROCEDURES
Differential mRNA Display--Purified preparations of cardiomyocytes were generated as described (23,24) from 50 embryonic day 16, postnatal day 1, day 3, day 5, and day 12 rat hearts. Differential display was carried out using a RNAimage kit (GenHunter) according to the manufacturer's instructions. Differentially displayed bands were excised from the polyacrylamide gels and reamplified. The resultant PCR amplicons were tested for differential expression on Northern blots using whole heart poly(A ϩ ) RNA and then used as probes to screen cDNA libraries for full-length transcripts.
cDNA Library Construction and Screening-Double-selected poly(A ϩ ) mRNAs from rat heart and total RNAs from lung were obtained using Messagemaker (Life Technologies, Inc.). The cDNA libraries were constructed using a Superscript cDNA synthesis kit. The cDNAs were size-fractionated, adapted with linkers, ligated into ZipLox arms (Life Technologies, Inc.), and packaged using GigapackIII Gold (Stratagene). Approximately 30 clones were incorporated into the ETL cDNA contig. The GenBank TM accession number for rat ETL is AF192401, and the accession number for the type II isoform is AF192402.
Nucleotide Sequencing and Analysis-Clones obtained from cDNA library were sequenced and a contig was compiled using Sequencher 3.0 software. The consensus sequence was analyzed by Blast for homologies and ExPASy tools for protein motifs and patterns.
RH Mapping-Primers RP29.2A-RP29.2B, that amplify a genomic STS from human BAC RP11-29e12 (Research Genetics), were used in PCR with DNAs from medium resolution (G3) and high resolution (TGN3) radiation hybrid panels (Research Genetics). Results were analyzed using the Stanford RH mapping data bases (available on the World Wide Web).

Human ESTs Incorporated into the Contig of hETL-Incorporated
Northern Blot Analyses-Equal amounts of RNA were run on formaldehyde gels, blotted on nylon membranes (GeneScreen) and UV-crosslinked. Membranes were stained with methylene blue for visualization of 28 and 18 S RNAs (25). Hybridizations were performed in Express Hyb solution (CLONTECH) followed by washes and autoradiography. For normalization of poly(A ϩ ) RNAs, blots were probed with GAPDH and quantified by PhosphorImager analysis (Molecular Dynamics, Inc., Sunnyvale, CA).
Construction of Expression Vectors-The cDNA sequence corresponding to the ETL open reading frame was amplified by RT-PCR from rat heart using primers 69F50XH and 69XBB, containing 5Ј restriction site overhangs. The amplified product was cloned into pBluescript, and 10 clones were sequenced. Clone 1 contained no mutations, while clone 3 harbored a T455A substitution. These were selected and subcloned in frame with Myc-His 3Ј tag sequences into the pcDNA 3.1 vector (Invitrogen). The extracellular region was amplified using primers 69F50BI and 69R45H3. The extracellular region up to the cleavage site, including Thr 455 , was amplified using 69F50BI and 69R46H3. PCR products were fused in frame with the Fc portion of human IgG and subcloned into the pcDNA 3.1 vector. The Fc portion of human IgG was also amplified by PCR using the human EST clone 809684 (GenBank TM accession number AA456339) as a template. The 1-455 aa extracellular region was also fused in frame with Myc-His 3Ј tag sequences in the pcDNA 3.1 vector.
Transfection into COS-7 Cells and Western Blotting-1.5 ϫ 10 5 COS-7 cells were seeded into six-well dishes and transfected with constructs using LipofectAMINE reagent (Life Technologies, Inc.) and 1 g of plasmid DNA purified with a Qiagen column. Proteins were harvested 48 h after transfection in 300 l of triple detergent solution (25). Ten l of total protein extracts and 10 l of a protein molecular weight ladder (New England Biolabs) were heated in the presence of 1% ␤-mercaptoethanol and resolved in Laemmli buffer (25) on 4 -20% SDSpolyacrylamide gel electrophoresis gradient gels (Bio-Rad) or 7.5% nongradient gels with 8 M urea. Rat ETL (rETL)-Myc was visualized using mouse monoclonal anti-Myc antibody (Oncogene Research Products) and the appropriate secondary antibody for ECL detection (Amersham Pharmacia Biotech). Peroxidase-labeled goat anti-human antibody (Kirkegaard & Perry Laboratories), followed by ECL detection, was used for ETL-Fc protein visualization. Membrane preparations were processed as described (17). Conditioned medium from COS-7 cells, transiently transfected with the 1-455-Myc extracellular domain of ETL was concentrated using a Centricon-3 spin column and loaded on SDS-polyacrylamide gels along with total cell lysates.
In Situ Hybridization-The transmembrane region of rat ETL was PCR-cloned into pBluescript using two primers with restriction site overhangs, TRE1A and TRH3B. Plasmid was linearized, and sense/ antisense probes were synthesized using a DIG RNA labeling kit (Roche) and used in hybridization at 1 ng/l. Paraffin-embedded 2-week-old rat lung and heart tissues were sectioned, 8 -9 m thick, deparaffinized in Hemo-DE, hydrated through a series of graded ethanol and water, and treated with proteinase K at 6 g/ml for 90 min. Sections were hybridized overnight at 60°C in a buffer containing 50% deionized formamide, 10% dextran sulfate, 1ϫ Denhardt's solution, 100 mM dithiothreitol, 20 mM Tris-HCl, pH 7.5, 5 mM EDTA, 300 mM NaCl, in diethylpyrocarbonate-treated water. Washes were as follows: 4ϫ SSC, 10 mM dithiothreitol for 1 h; 5% formamide, 2ϫ SSC, 20 mM dithiothreitol for 30 min at 50°C; 1ϫ NTE for 15 min at 37°C. Immunological detection was performed using a DIG nucleic acid detection kit (Roche Molecular Biochemicals). Washes consisted of 2ϫ SSC for 5 min, 0.1ϫ SSC for 15 min, buffer 1 (Roche) for 15 min, buffer 2 containing 20% sheep serum for 30 min. Washes were followed by incubation with anti-DIG-AP conjugate, 1:500 dilution, for 2 h. Postincubation washes were performed in buffer 1 (Roche Molecular Biochemicals) three times for 10 min and buffer 3 for 5 min, followed by incubation with the substrate, 1% nitro blue tetrazolium/5-bromo-4-chloro-3-indolyl phosphate in buffer 3 (Roche Molecular Biochemicals). Development was carried out for several hours and terminated for both antisense and sense slides at the same time. Slides were counterstained in 0.5% methylene blue, washed in water and 100% butanol, dehydrated in Hemo-DE, mounted, and photographed using a Zeiss microscope.
Immunofluorescence and Confocal Microscopy-COS-7 cells were seeded at a density of 4 ϫ 10 4 on chamber slides, transiently transfected with an ETL-Myc expression construct, allowed to recover for 24 -48 h, serum-starved for 4 h where indicated, and stained for immunofluorescence. Washed cells were fixed in Zamboni (Vector Laboratories), permeabilized in 0.2% Triton X-100, and blocked in 1% nonfat milk. Cells were then incubated with anti-Myc antibody (Oncogene Research Products) followed by appropriate secondary IgG conjugated to biotin and Alexa-594 conjugated to streptavidin (Molecular Probes, Inc., Eugene, OR). Cells were examined by fluorescent microscopy using a Texas Red filter or by confocal microscopy at 594-nm wavelength. Primers

RESULTS
ETL Is Up-regulated in the Heart after Birth-To identify genes involved in heart development, we conducted a differential display of fetal and postnatal mRNAs isolated from purified cardiac myocytes using fetal, day 1, day 3, day 5, and day 12 rat hearts. Several genes were identified by RT-PCR as up-regulated or down-regulated during heart development. Here, we describe clone 69, subsequently named ETL, that is up-regulated postnatally during cardiomyocyte development (Fig. 1A). The differentially expressed PCR amplicon was sub-sequently excised from the polyacrylamide gel and used for Northern blot analyses.
The developmentally regulated expression of rETL was confirmed on poly(A ϩ ) mRNA Northern blots from rat whole hearts, where RNAs were extracted at several developmental time points: fetal, day 1, day 3, and day 12 (Fig. 1B). The experiments identified an increasing abundance of rETL mRNA that correlated with developmental age. The GAPDHnormalized images of rETL indicated a 4.6-fold increase in expression by day 12 and 1.9-fold increase by birth in comparison with embryonic day 16 mRNA levels. Using rat poly(A ϩ ) RNA Northern blots, we examined the expression of rETL in adult tissues. Rat ETL is abundantly expressed in heart, lung, and kidney; less evident expression is observed in brain, skeletal muscle, liver, and spleen, with the exception of testis, where no expression is detected (Fig. 1D). Similarly, differential expression of human ETL (hETL) was evident by probing total RNA Northern blots from 3-month fetal and adult human hearts ( Fig. 1C) with PCR-cloned human cDNA. Unlike rat, two messages were detected in human heart and may represent distinct isoforms of hETL.
Two Alternatively Spliced Isoforms of rETL-The PCR products from differential display were used to screen an adult rat heart library to generate a full-length cDNA contig. Several screens of our rat heart cDNA library yielded 30 overlapping clones. Sequence analyses revealed a cDNA sequence of 4274 bp that approximately corresponds to the size of the 4.4-kilobase mRNA observed on rat heart Northern blots (Fig. 1B). The longest open reading frame is 2214 bp, starting with the putative initiation Met and ending with a TAA termination codon. The 3Ј-untranslated region is 1821 bp and contains a putative polyadenylation signal. The 5Ј-untranslated sequence is 235 bp and contains a stop codon preceding the open reading frame.
Further analyses of cDNA clones revealed a second isoform of rETL. Several clones encompassed an in-frame insertion sequence between nucleotides 2129 and 2130. The alternatively spliced exon of 234 bp contains a TAA termination codon and leads to premature termination with a predicted peptide length of 660 aa, thereby deleting amino acids C-terminal to the fifth transmembrane domain.
The Human ETL Homolog, hETL, Is Highly Conserved with rETL in the Transmembrane Domain-Homology searches with rat ETL cDNA sequences revealed several highly homologous human ESTs, mostly residing in 3Ј-coding and -untranslated regions. The most striking homology was observed to unordered contigs of genomic sequences of human BAC RP11-29e12 (accession number AC024326.2). The homology extends through the majority of the open reading frame, including sequences of the 3Ј-untranslated region. To obtain physical clones, we designed human-specific pairs of PCR primers in the regions of indicated homology and employed RT-PCR on cDNAs from human heart. Sequences of RT-PCR products and several human ESTs allowed us to identify a total of 13 exons (Table I) spanning 2689 bp. Exon 1 contains the putative translation initiation Met; however, it does not have an in-frame stop codon upstream of the Met. Exon 1a was revealed by a high homology to the rat protein, yet is it spliced out in several tissues we examined and in EST 112416. Exon 13 carries a conserved translational stop codon and 3Ј-untranslated region sequences. Pairwise comparisons of human ETL cDNA sequence and predicted peptide sequence against the rat homolog indicated 80% identity at the nucleotide and 87% similarity at the amino acid level. In obtained 3Ј-untranslated region sequences, 234 bp show striking interspecies identity (87%). This untranslated region is located ϳ200 bp from the conserved stop codon and potentially represents a regulatory element for mRNA expression or stabililty.
Human ETL Maps to Chromosome 1-Using 29.2A-2B primers, we amplified a genomic STS from BAC RP11-29e12 and DNAs from medium and high resolution Radiation Hybrid panels (Research Genetics). We linked hETL to two chromosome 1 markers, SHGC-21318 within 8 centirays and to SHGC-57820 within 15 centirays with LOD scores of 8 and 11, respectively (data not shown). These markers, although not ordered on the chromosome 1 map, are tightly linked to the D1S500, a GDB locus located on the 1p32-p33 band of chromosome 1 (26). A genomic clone containing mouse ETL, BAC 322E17, was mapped by fluorescent in situ hybridization analysis on mouse chromosome 3, H3-H4. 2 This mouse genomic region is syntenic with human chromosome 1p32-p33.

FIG. 1. ETL expression is up-regulated in the adult heart.
A, to characterize differences in gene expression in adult and fetal hearts, we conducted a differential display analysis of mRNAs from rat cardiac myocytes from embryonic day 16, postnatal day 1, day 3, day 5, and day 12 rat hearts. The arrow indicates the differentially displayed RT-PCR product, clone 69, subsequently named rETL. B, to confirm the regulated expression of rETL, we performed Northern analysis using the PCR amplicon from the differential display as a probe on whole rat heart poly(A ϩ ) RNA blots. Lanes represent rat heart RNA from the fetal stage embryonic day 16 through postnatal day 12, as designated. A probe for GAPDH was used as a normalization control. The data on differential expression of rETL were reproduced using three different batches of mRNA. C, Northern analyses of the hETL transcript using total RNA from human fetal (3 months) and adult heart, as designated. The blot was probed with human ETL cDNA. D, Northern blot of poly(A ϩ ) RNA from rat adult tissues probed with rat ETL cDNA. The lane marked muscle represents skeletal muscle.
2). The N-terminal region begins with a 19-aa signal peptide, followed by a short domain (ϳ26 aa) related to a lectin-type domain (27), one EGF-like domain, and two identical Ca 2ϩbinding EGF domains (ϳ91 aa). A Ser/Thr-rich linker region of ϳ297 aa follows the EGF domains and precedes a conserved Cys-rich proteolysis domain (ϳ50 aa). The latter has recently been described in a small number of transmembrane proteins and is also referred to as the GPCR proteolysis site (GPS) (19,28,29). This region, together with the short strech of the Ser/Thr linker and the adjacent transmembrane and cytoplasmic domain, shows the highest conservation between rat and human proteins (Fig. 3B).
Analyses of the putative hETL protein revealed only one Ca 2ϩ -EGF binding domain. This domain is potentially encoded by exon 1a. However, in human EST 112416, encompassing exons 1-3, as well as in our RT-PCR experiments using human heart, placenta, lung, and kidney RNAs, exon 1a is spliced out. Since this exon shows a high degree of identity in humans and rats (80% of 150 bp) and codes for a potentially important functional EGF domain, we cannot rule out the existence of isoforms carrying exon 1a.
The extracellular domain of rETL also contains a potential Asn hydroxylation site within each Ca 2ϩ binding EGF-like domain. One O-linked putative glycosylation site is found at Ser 388 in the Ser/Thr rich domain. The extracellular domain also carries nine potential N-linked glycosylation sites, most of them within the Ser/Thr-rich linker and GPS domain, and all but one conserved in both species. Rat and human ETLs also possess cAMP-and cGMP-dependent protein kinase phosphorylation sites and several common potential protein kinase C and casein-kinase II phosphorylation sites.
The hydropathy and homology analyses of the rETL transmembrane segment predicted seven helices and a class II/ secretin G-protein-coupled receptor signature sequence. Fig.  3A displays the alignment of rat ETL and a human homolog in the heptahelical domain along with several members of the secretin family, EGF-TM7 subfamily members, and several receptors with large extracellular domains. The overall structure is most similar to the EGF-TM7 family, with 30% identity to EMR1 and CD97 in pairwise comparisons (data not shown). The ETL TM7 segment together with the adjacent Cys-rich proteolysis domain and the Ser/Thr linker, displays significant homology to the three related heptahelical receptors, CL1, CL2, and CL3, with 40% identity. Based on these homologies, we designated our protein ETL (for EGF-TM7-latrophilin-related protein). A comparison of rETL and rat CL1-3, revealed conservation of the proteolysis domain, with 60% identity in the region (Fig. 3B). CL1 is endoproteolytically cleaved at the end of this domain (18), and cleavage of CL2 and CL3 has been referenced as well (19). The proteolysis domain is characterized by several invariant amino acids; most numerous among them are Cys residues. Several recently discovered TM7 receptors with large extracellular domains also exhibit this motif (Fig. 3B and Ref. 19), but cleavage has not been documented.
The short cytoplasmic tail carries a tyrosine kinase phosphorylation site that could play a role in desensitization of the receptor (30) or coupling to a tyrosine kinase signaling pathway (11). A putative tyrosine phosphorylation site, preserved in both rat and human ETLs, could be involved in cross-talk  between signal transduction modules employing tyrosine kinases such as a MAP kinase pathways (11) and could represent a scaffold for the assembly of a phosphotyrosine-dependent complex. The overall structure of rETL suggests that the protein might participate in both cell surface events such as cellcell recognition and adhesion and in signal transduction cascades. rETL Forms a Stable Receptor-Dimer in COS-7 Cells-We expressed rETL protein tagged with a Myc epitope on the C terminus in COS-7 cells and identified an ϳ85-kDa protein using anti-Myc antibody on total protein lysates by Western blot analyses. This 85-kDa band was not present in vector alone transfections (Fig. 4A). The observed mass of 85 kDa closely correlated with the expected mass of the mature rETL-Myc-His protein (83.7 kDa). We also observed a broad intense band at ϳ175 kDa. This band probably represents an rETL dimer with additional post-translational modifications that is stable in the presence of most reducing agents and boiling. Extraction of total proteins with 6 M guanidine hydrochloride led to nearly complete disappearance of the 175-kDa band (Fig. 4B, ETL lane).
rETL Is Cleaved within the Putative Extracellular Domain-Conservation of the rETL Cys-rich domain, a domain known to undergo cleavage in CL1, CL2, and CL3 receptors (18,19), prompted us to test whether rETL also undergoes proteolytic processing. Previously, the cleavage of CL1 was only demonstrated in the presence of 8 M urea, both in gel and sample buffer (18). When total protein extracts from COS-7 cells transfected with the rETL construct were subjected to these conditions, a band of ϳ35 kDa appeared, and the 85-kDa band became less intense (Fig. 4D, ETL lane). This observation correlated with the predicted products of cleavage between Leu 454 and Thr 455 of a 48-kDa peptide (untagged) and a C-terminal 35.7-kDa peptide (tagged with Myc) (Fig. 4D, ETL lane). The 85-kDa doublet in rETL probably represents the mature and precursor rETL proteins (Fig. 4D, ETL lane). The fact that only certain chaotropic agents allowed us to observe the cleaved products suggests that these cleaved peptides stay bound as it has been shown for CL1 receptor (18). This proteolytic processing may play an important role in the formation of a functional receptor.
To further analyze the processing of rETL, we used a mutant clone, rETL*T455A. This clone, generated as a PCR cloning artifact, carries a mutation at the conserved Thr residue previously determined to be at the processing site in latrophilin (18). The mutated protein shows resistance to cleavage as assessed by the failure to detect a 35-kDa band in the presence of 8 M urea (Fig. 4D, ETL*T455A lane). This suggests that Thr 455 is required for proteolytic processing. Interestingly, 8 M urea did not lead to the disappearance of the 175-kDa band, implying strong modifications after receptor-dimer formation. In 6 M guanidine HCl preparations, rETL*T455A displays a od) amino acid changes are shown in light gray columns. Alignments were built using ClustalW. B, amino acids in the Cys-rich proteolysis domain of rat and human ETLs were aligned with several transmembrane molecules carrying this domain. Aligned proteins included seventransmembrane receptors CL1, CL2, CL3, hCD97, lectomedin-1 ␣ (hLec1), Celsr1 receptor, (hCelsr1), Flamingo seven-pass transmembrane cadherin (mFlamingo), brain-specific angiogenesis inhibitor 3 (hBAI3), and serpentine receptor (mCYT28). Other molecules, such as KIAA0279, h287, hR29368_2, hMEGF2, and hTM7XN1, are putative TM7 receptors. CeB0286 is a Caenorhabditis elegans putative G-protein-coupled receptor; SuREJ is a Strongylocentrotus purpuratus sperm receptor for egg jelly. GenBank TM accession numbers are given in parentheses. The putative proteolytic processing site is shown by an arrow, and the location of the mutation in rETL*T455A is indicated by a white letter above the arrow.

FIG. 3. ETL is related to the secretin peptide hormone receptor family and has a conserved G-protein-coupled receptor proteolysis domain.
A, rat and human ETL transmembrane domains were aligned with several members of the secretin receptor family. GenBank TM accession numbers are given in parentheses. h, r, or m, sequence from human, rat, or mouse species. The aligned proteins are as follows: rCL1/latrophilin, receptor for latrotoxin; rCL2; rCL3; epidermal growth factor module-containing mucin-like receptor 1 (hEMR1); leukocyte antigen (hCD97); vasoactive intestinal peptide receptor (hVIPR); secretin receptor (hSECR); glucagon receptor (hGLUCR); and corticotropin-releasing factor receptor (hCTRFR). Transmembrane domains are shown in blocks and designated as TM1 to TM7; identical amino acids are shown in white on dark gray background and designated by an asterisk. Conservative (colon) and semiconservative (peri-higher ratio of dimer to monomer conformation than wild type (Fig. 4B). Stable dimer formation for rETL*T455A may also contribute to the overall higher stability and, therefore, higher quantity of the protein.
rETL Transmembrane Domain Is Required for Dimerization but Not Cleavage-To determine whether the transmembrane domain is required for endoproteolytic processing, we constructed a fusion protein consisting of the rETL exodomain (aa 1-483) and the 226-aa Fc portion of human immunoglobulin (ETL483-Fc). The predicted molecular mass of the resultant peptide is 79.6 and 77.1 kDa, for the precursor and mature proteins, respectively. Under regular Laemmli conditions, we observed two peptides, ϳ30 and 80 kDa (Fig. 4E, ETL483-Fc  lane), indicating that the truncated protein can be cleaved in the absence of the heptahelical portion. The predicted cleavage between Leu 454 and Thr 455 was expected to produce a mature N-terminal peptide of 48 kDa (untagged) and 29.6 kDa (tagged with Fc) C-terminal peptide. The 80-kDa band most likely represents the unprocessed protein. Interestingly, no dimer formation was observed, suggesting that although the transmembrane domain is not required for cleavage, it is required for dimer formation. We also constructed a fusion protein consisting of the rETL exodomain up to the Thr 455 residue, including ETL455-Fc, and observed no 30-kDa species (Fig. 4E,  ETL455-Fc lanes). These observations demonstrate that the amino acids immediately following the processing site play an important, most likely conformational, role in cleavage.
rETL Detected in Membrane Preparations-Both rETL and rETL*T455A were detected in membrane preparations of COS-7 cells transfected with these constructs (Fig. 4C). Interestingly, no 85-kDa protein species were observed in the Triton-extracted membrane proteins. When proteins were resolved under regular Laemmli conditions, wild type rETL appeared as a 68-kDa doublet, approximately corresponding to the predicted mass of a cleaved dimer (Fig. 4C, ETL lane), 71.4 kDa. In these experiments, rETL was not detected as an 85-or 175-kDa protein species, corresponding to the rETL uncleaved monomer or uncleaved dimer, respectively. The mutation T455A did not affect the membrane association of the rETL*T455A protein. The cleavage-resistant ETL*T455A protein was also detected in dimer-only conformation, as revealed by the presence of a 175-kDa band on the Western blot (Fig. 4C, ETL*T455A lane). In addition to the rETL membrane association, these results suggest that rETL is cleaved during intracellular processing because the membrane preparations included both the ER and plasma membranes.
rETL Is a Plasma Membrane-associated Protein-The predicted amino acid sequence of rETL and the substantial structural homology to known receptor families suggested that rETL localized to the plasma membrane. Further data supporting a plasma membrane localization for rETL were obtained using Western analyses. As noted above and in Fig. 4C, rETL was detected in membrane preparations. We also generated a construct encoding only the extracellular domain of rETL. When the exodomain of rETL, aa 1-455, tagged C-terminally with Myc epitope, was transiently expressed in COS-7 cells, it was detected as a soluble protein in both conditioned medium (Fig.  4F, ETL455 sup) and whole cell lysates (Fig. 4F, ETL455 cell). Taken together, these data are highly suggestive that rETL is a plasma membrane protein.
We used confocal microscopy to determine a subcellular localization of C-terminally Myc-tagged rETL in COS-7 cells. Rat ETL was observed in the proximity of plasma membrane and intracellular vesicles in permeabilized cells only (Fig. 5A). Transient transfection allowed us to control the specificity of indirect immunofluorescence, since only a percentage of cells receive the plasmid and fluoresce (data not shown). The vesicles most likely represent endoplasmic reticulum, Golgi apparatus, and cytoplasmic transport vesicles involved in process-

FIG. 4. rETL forms a dimer and undergoes endoproteolytic cleavage in transfected COS-7 cells.
A-F, Western blot analysis of total proteins obtained from transiently transfected COS-7 cells, lysed as indicated below. Proteins were resolved on gradient gels, stained with monoclonal anti-Myc antibody followed by ECL detection. A, cells were transfected with pcDNA-ETL-Myc constructs or control vector as indicated and lysed in triple detergent. Note that both rETL and rETL*T455A display 175-and 85-kDa proteins, suggesting dimer formation. B, cells were transfected as in A and lysed with 6 M guanidine hydrochloride, precipitated, and dissolved in 8 M urea. Note that under highly denaturing conditions, the ratio of 175-to 85-kDa protein is much higher for the mutant clone than for wild type. C, membrane proteins from COS-7 cells, transfected as in A, were lysed in Triton detergent. Note that sizes of the detected rETL and rETL*T455A peptides correspond to the cleaved and uncleaved dimers, respectively. The 85-kDa species were not detected in these experiments. The faint band between the two arrows is ϳ120 kDa and may correspond to a dimer of cleaved and uncleaved rETL. D, cells were transfected and lysed as in A and resolved on 7.5% SDS-polyacrylamide gel containing 8 M urea in the gel and sample buffer. Note that these denaturing conditions revealed a proteolytic cleavage of rETL. As predicted, mutant rETL resisted cleavage. E, cells on the left were transfected with the construct of extracellular domain of rETL (aa 1-483), fused to Fc. On the right, cells were transfected with the construct of extracellular domain of rETL N-terminal to the cleavage site (aa 1-455, including Thr 455 at the cleavage site), fused to Fc. Anti-Fc antibody and subsequent ECL were used in detection. Note that amino acids 456 -483 are required for proteolytic cleavage. F, COS-7 cells were transfected with rETL (aa 1-455) tagged with Myc. Conditioned medium (ETL455sup) was collected, and cell lysates (ETL455cell) were prepared as in A. Note that the extracellular domain is a soluble extracellular peptide. Two bands below the 62-kDa arrow in the ETL455cell lane represent the nonspecific antibody binding, as determined by mock transfections as well. ing and trafficking of the receptor. After 4 h of serum starvation, we observed rETL in perinuclear vesicles only (Fig.  5B). These data may be explained by a rapid turnover or internalization rate of rETL in COS-7 cells in the response to serum starvation.
rETL Is Expressed in Cardiac Myocytes, Bronchiolar and Vascular Smooth Muscle-To more precisely ascertain the localization of rETL expression in adult rat heart and lung tissues, and shed light on the potential sources of ligand, we performed in situ mRNA hybridization on paraffin sections of 12-day-old rat hearts and lungs. Using an antisense probe from the transmembrane region (Fig. 6, left images), we show expression of rETL in cardiomyocytes (Fig. 6A) and vascular smooth muscle cells in coronary vessels (Fig. 6C) in the epicardial layer of the heart. In lung, we observed staining of vascular smooth muscle cells in blood vessels (Fig. 6E) as well as smooth muscle cells in bronchioles (Fig. 6G). Hybridization of a sense probe to similar sections served as a control (Fig. 6, right  images). The expression of rETL in smooth muscle was confirmed by RT-PCR on rat smooth muscle poly(A ϩ ) RNA (data not shown). The expression of rETL was also detected in vessels of the kidney. 2

DISCUSSION
In this study, we have isolated and characterized a novel cDNA clone, ETL, encoding a new member of the EGF-TM7 subfamily of receptors. Similar to all members of the EGF-TM7 receptor family, ETL has a tripartite domain structure consisting of a large extracellular domain, a seven-membrane-span-ning domain, and a short cytoplasmic tail. A characteristic feature of the EGF-TM7 protein group, an unusually large exodomain, incorporates cell surface interaction modules, such as EGF-like motifs, lectin-like motifs, and a Ser/Thr rich domain, with numerous sites for N-and O-linked glycosylation. Rat ETL has several EGF domains, which are often found in extracellular portions of a large number of proteins, including fibrillin, fibulin, entactin, tenascin, and thrombospondins (31), and are functionally associated with protein-protein interactions. The evolutionary implications of the addition of cell surface modules, a feature previously found only in single membrane-spanning molecules, to the GCPR's signal transduction domain remains obscure. One possibility is that these modules are essential for ligand specificity or presentation.
Ca 2ϩ , potentially bound to an Asn ␤-hydroxylation site inside EGF domain, is likely to stabilize protein-protein interactions of rETL (32). Most of the members of the EGF-TM7 family display this Ca 2ϩ -binding feature, suggesting a common mechanism for ligand binding. To date, the only known ligand for the EGF-TM7 receptor family is CD55, or decay-accelerating factor, a complement component. CD55 interacts with one of the isoforms of the CD97 receptor that is expressed on the cell surface of leukocytes (33). CD97 carries several tandem EGF repeats, and deletion analyses showed that binding requires both Ca 2ϩ -EGF domains and Ca 2ϩ for ligand binding (34). In several human tissues examined, the Ca 2ϩ -binding EGF domain of hETL appears to be spliced out. However, protein motif analyses detected another potential Ca 2ϩ -binding domain in hETL, an EF-hand signature sequence. This human-specific domain is present in a large family of calcium-binding proteins (35).
As noted in Fig. 3B, we found a significant conservation between TM7 regions of secretin receptors and ETL. The importance of this homology is unknown but may indicate a conservation of critical residues for G-protein coupling to intracellular loops and/or ligand-binding determinants within the extracellular loops of GCPR transmembrane domains (36 -38).
We show here that ETL carries a Cys-rich proteolysis domain with a high degree of identity to CL1, CL2, and CL3 heptahelical receptors. CL1, also known as latrophilin, is shown to interact with ␣-latrotoxin, a toxin from the black widow spider (39). Recent studies showed that latrophilin tethers ␣-latrotoxin to the membrane, initiating Ca 2ϩ channel formation (39). The proteolysis domain shared with CL1-3 receptors is present in all EGF-TM7 receptor family members. Recently, this motif has been identified in other transmembrane molecules, such as polycystin-1 (PKD1) (28), the protein defective in ADPKD polycystic kidney disease (40), and two other PKD1-related proteins, sea urchin egg jelly receptor, REJ, and its human homolog, PKDREJ (41,42). The latter three proteins exhibit a different number of membrane helices, from 1 to 11. However, similar to CL1, these proteins support extracellular Ca 2ϩ influx after activation, functional characteristics thought to be due to the common proteolysis domain and first transmembrane domain (28). This commonality supports the hypothesis that ETL, upon activation, is involved in cation influx.
Cleavage within the proteolysis domain of CL1 (18,19) and rETL, shown here, suggests that other proteins carrying the conserved domain may also be cleaved. The detection of rETL in cleaved dimer conformation in membrane preparations suggests that the proteolytic activity is an intracellular event. The 68-kDa doublet may imply that cleavage of the extracellular domain precedes the cleavage of N-terminal signaling peptide. The detection of rETL cleavage with a limited number of chaotropic agents suggests that the cleaved N-terminal domain of FIG. 6. ETL is expressed in cardiac myocytes and vascular and bronchiolar smooth muscle. Paired paraffin sections of 2-week-old rat hearts and lungs were hybridized with DIG-labeled antisense probes on the left and sense probes on the right. The in situ hybridization signal was detected as a purple product of the bound nitro blue tetrazolium/5-bromo-4-chloro-3-indolyl phosphate substrate (Roche Molecular Biochemicals). Sections were counterstained in 0.5% methyl green. Magnifications of ϫ 25-125 were used to accommodate the sizes of stained structures. A and B, cross-section of heart, at the level of the ventricle, exhibiting a dense cardiomyocyte population (magnification ϫ 125). C and D, cross-section of a coronary vessel in the epicardial layer of the heart (magnification ϫ 125). E and F, cross-section of a pulmonary arteriole (magnification ϫ 100). G and H, lung bronchiole, cross-section (ϫ 25 magnification). Sections probed with sense and antisense probes were hybridized and developed in parallel. rETL remains tightly tethered to the transmembrane domain, as has been shown for CL1 receptor cleavage product (18). Interestingly, the mutation T455A in rETL that we have shown completely abolishes cleavage is present in the native EMR1 receptor, but there is no biochemical data available on EMR1 processing. Based on the high degree of homology in the proteolysis domain of rETL and CL1 and on the size of rETL cleaved product that we observed in our experiments, it is likely that rETL utilizes the same site for proteolytic processing. Therefore, the T455A mutation most likely resides directly at the cleavage site, as is the case for CL1 (18). The ETL*T455A clone will be useful for determining if post-translational processing is important for receptor function when an ETL ligand is discovered. Prior to isolation of ETL ligand, this cleavage site mutation could be tested in a known exogenous ligand-receptor function system such as latrotoxin-latrophilin/CL1, given the high degree of conservation of these proteolysis domains.
Unlike other members of the EGF-TM7 family, which are often described as glycosylated proteins on the surface of leukocytes (15), ETL is expressed in cells of mesenchymal origin such as cardiac myocytes and smooth muscle. ETL is the first TM7 receptor containing EGF-like motifs that is developmentally regulated in the heart. Developmental regulation of ETL expression in rat and human heart coincides with the terminal differentiation of cardiac muscle and the switch from hyperplastic to hypertrophic growth phases in mature cardiac muscle. ETL may be an important effector in these processes. The expression of ETL in coronary vessels may also suggest involvement in coronary angiogenesis. The cell adhesion and cell-cell interaction modules of ETL may facilitate interaction with the extracellular matrix that is deposited in the heart after birth and be important in cell-cell communication. Although the function of the ETL receptor remains unknown at this time, regulated ETL expression implies an involvement in cardiac development.