Tissue-specific expression of unique mRNAs that encode proglucagon-derived peptides or exendin 4 in the lizard.

Glucagon-like peptide 1 stimulates insulin secretion and inhibits glucagon secretion, gastric emptying, and feeding, suggesting it may be biologically useful for the treatment of diabetes. A lizard glucagon-like peptide 1 (GLP-1)-related peptide, exendin 4, binds to the GLP-1 receptor and mimics the actions of GLP-1 in vivo To determine the genetic relationship between exendin 4 and GLP-1, we analyzed the structure and expression of pancreatic and intestinal proglucagon mRNAs in the reptile Heloderma suspectum Two different proglucagon cDNAs (lizard proglucagon I (LPI) and lizard proglucagon II (LPII)), with unique 3′-untranslated regions were identified. Two LPI mRNA transcripts, ∼1.6 and 2.1 kilobases, encoded glucagon and GLP-1 but not GLP-2 and were restricted in expression to the pancreas. In contrast, a 1.1-kilobase LPII mRNA transcript, encoding glucagon, GLP-1, and GLP-2 utilized a different 3′-untranslated region and was expressed in both pancreas and intestine. Lizard proglucagon mRNA transcripts were not detectable by reverse transcription-polymerase chain reaction or Northern blotting in salivary gland. A single class of lizard salivary gland proexendin cDNAs encoded the sequence of exendin 4 and a 45-amino acid exendin NH2-terminal peptide. Exendin mRNA transcripts were expressed in the salivary gland, but not pancreas or intestine. These data demonstrate that GLP-1 and exendin 4 represent related yet distinct peptides encoded by different genes in the lizard.

. In the pancreas, glucagon is the major bioactive PGDP, whereas in the intestine, processing of proglucagon to GLP-1, GLP-2, oxyntomodulin, and glicentin results in the liberation of a broader spectrum of bioactive peptides (8).
The PGDPs exhibit an increasing number of important biological activities. Glucagon, a 29-amino acid peptide hormone, is a key regulator of carbohydrate, protein, and lipid metabolism. Truncated forms of GLP-1, either the (7-37) or (7-36)amide molecules, display potent glucose-dependent insulinotropic actions both in vitro and in vivo (9) and mice with a targeted disruption of the GLP-1 receptor gene develop diabetes (10). GLP-1 infusion in human subjects lowered postprandial glycemia and decreased insulin requirements in patients with both insulin-dependent diabetes mellitus and non-insulindependent diabetes mellitus, raising the possibility that GLP-1 may be a useful adjunct for the treatment of patients with diabetes (11). A role for GLP-1 in the central control of feeding behavior and satiety was recently suggested by experiments showing ICV administration of GLP-1 potently inhibited feeding in rats (12). Taken together, these observations provide increasing evidence for the importance of GLP-1 in the physiological regulation of complex biological systems.
The isolation of a novel peptide from lizard venom that displays 52% identity to mammalian GLP-1 prompted studies examining the biological properties of this protein, designated exendin 4 (13). Experiments employing synthetic exendin 4 provided evidence that this peptide shares similar biological activities with mammalian GLP-1. Exendin 4 and truncated GLP-1-(7-36)-amide increased cAMP levels in guinea pig acinar cell preparations (14), and exendin 4 was subsequently shown to bind the GLP-1 receptor, stimulate glucose-dependent insulin secretion, and increase both cAMP accumulation and insulin gene expression in cultured islet cell lines in vitro (15). The insulinotropic GLP-1-like properties of exendin 4 suggest that this lizard peptide may also be therapeutically efficacious for the treatment of patients with diabetes.
A smaller peptide derived from exendin 4 (following deletion of the first 8 amino acids), designated exendin-(9 -39), is an effective antagonist of the GLP-1 receptor, blocking both the binding and biological activities of GLP-1 and exendin 4 (13,15). Exendin-(9 -39) has proven to be a physiologically relevant antagonist for assessing the biological importance of GLP-1 by blocking the binding and activity of GLP-1 in vivo (16,17). The striking homology between lizard exendin 4 and mammalian GLP-1 raises the possibility that exendin 4 may simply represent the reptilian equivalent of GLP-1 that is also expressed in * The work was initially supported in part by a grant from the Canadian Diabetes Association and subsequently by the Medical Research Council of Canada. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBank TM  RT-PCR-Total cellular RNA from lizard intestine, pancreas, and salivary gland was isolated by the acid guanidinium thiocyanate method (18). For first strand cDNA synthesis, 5 g of total cellular RNA was reverse-transcribed using the SuperScript preamplificaton system (Life Technologies, Inc.). An aliquot of the first strand cDNA (0.1 volume) was subsequently amplified by PCR (35 cycles, annealing temperature of 50°C, extension temperature of 72°C). The degenerate oligonucleotides were designed according to the conserved NH 2 -terminal sequence (HSQGTF) of mammalian, bird, and amphibian sequence: Isolation of Pancreatic Proglucagon cDNA Clones-A H. suspectum cDNA library, made from poly(A) ϩ mRNAs isolated from the pancreas of adult lizard, consisting of 1 ϫ 10 6 independent recombinant phagemid, was constructed in the EcoRI sites of a pcDNA II vector (Invitrogen, San Diego, CA). The cDNA library was screened with a 209-bp EcoRI/BamHI [␣-32 P]dATP-labeled lizard cDNA generated by RT-PCR (see Fig. 1C). Nine positive clones were isolated and sequenced by using 35 S-dATP and a Sequenase version 2.0 kit (U. S. Biochemical Corp.) for direct dideoxynucleotide sequencing. In some cases, particular sequences were confirmed by subcloning DNA fragments into pBluescript II SK Ϫ for dideoxynucleotide sequencing (19).
Isolation of Exendin cDNA Clones-A H. suspectum cDNA library, made from poly(A) ϩ mRNAs isolated from the salivary glands of an adult lizard, consisting of 1 ϫ 10 6 independent recombinant phagemid, was constructed in the EcoRI sites of the pcDNA II vector (Invitrogen, San Diego, CA). The cDNA library was screened with a 95-bp EcoRI/ BamHI [␣-32 P]dATP-labeled lizard cDNA generated by RT-PCR (see Fig. 5C).
Southern Blot Analysis-DNA was isolated from H. suspectum liver tissue and digested with various restriction enzymes. The digested DNA was electrophoresed in a 0.7% agarose gel, vacuum-blotted, and UV-fixed using a UV Stratalinker (Stratagene, San Diego, CA). The membrane was prehybridized in a solution containing 10% formamide, 3 ϫ SSPE (1 ϫ SSPE ϭ 0.18 M NaCl, 0.01 M Na 2 HPO 4 , 1 mM EDTA), 1% SDS, 0.5% Blotto, and 0.5 mg/ml denatured herring sperm DNA for 4 h at 55°C. Hybridization was performed in a similar solution containing 10% dextran sulfate and probe A for proglucagon: a 201-bp [␣-32 P]dATP-labeled lizard proglucagon cDNA (nucleotides 243-443 in LPI and nucleotides 253-453 in LPII, see Figs. 2 and 3), or an exendin cDNA probe, for 12 h at 55°C. The final wash was done at 60°C. The membranes were exposed to Kodak XR-Omat RP-1 x-ray film with an intensifying screen at Ϫ70°C. DNA fragment sizes were estimated using 1-kb DNA markers.
Northern Blot Analysis-RNA from different tissues was separated in 1.0% agarose gels containing 2.6 M formaldehyde and vacuum-blotted to a Zeta-Probe membrane. Blots were hybridized with the following proglucagon cDNA probes: A, a 201-bp lizard proglucagon cDNA as described above; B, a 661-bp 3Ј-UTR fragment from LPI; or C, a 340-bp 3Ј-UTR fragment from the LPII. The exendin blot in Fig. 6 was hybridized with a full-length lizard salivary gland exendin cDNA probe. Hybridization and washing conditions were carried out as described previously (3). The approximate sizes of various mRNA species were determined from the mobility of a coelectrophoresed 0.24 -9.49-kb RNA ladder (Life Technologies, Inc.).

RESULTS
Isolation of a Lizard Proglucagon cDNA-As a first step in the elucidation of the potential relationship between exendin 4 and GLP-1, we generated a partial lizard proglucagon cDNA by RT-PCR. Two degenerate oligonucleotides (see "Experimental Procedures") were used to amplify first strand cDNAs prepared using RNA isolated from H. suspectum intestine, pancreas, and salivary gland (Fig. 1, A and B). Analysis of PCR products generated using cDNA from lizard pancreas revealed a ϳ215-bp partial proglucagon cDNA (Fig. 1B, lane 2). An identical albeit much less abundant band that comigrated with the pancreatic cDNA product was also obtained using the same primers and first strand cDNA from lizard intestine (Fig. 1B,  lane 3). In contrast, no PCR products were generated using the same primers and first strand cDNA prepared from either lizard liver or salivary gland (Fig. 1B, lanes 4 and 5). The size of the PCR product was larger than one would predict (for a partial mammalian proglucagon cDNA using the identical primers) and raised the possibility that one of the lizard PG-DPs, possibly IP-1, was larger than mammalian IP-1s, consistent with the results recently described for the chicken (20).
To ascertain the identity of the putative partial proglucagon cDNAs generated from pancreas and intestine, the PCR products were subcloned and subjected to DNA sequence analysis. Lizard pancreatic and intestinal (partial) cDNA clones contained an open reading frame encoding identical nucleotide sequences corresponding to: lizard glucagon, a larger than expected peptide immediately carboxyl-terminal to glucagon des- ignated lizard intervening peptide I, and the NH 2 -terminal region of lizard GLP-1 (see Fig. 2 for sequence). The amino acid sequence of lizard glucagon shares 92 and 97% identity with chicken and human glucagons, respectively (Table I). As predicted, lizard IP-1 was similar in size to the chicken (20) sequence (24 amino acids), but considerably larger than mammalian IP-1s, that are generally 6 amino acids long with two pairs of dibasic amino acid residues located at both the amino-and carboxyl termini.
Isolation and Characterization of cDNAs Encoding Lizard Proglucagon-To more completely elucidate the structure of the lizard proglucagon mRNA(s), the 209-bp lizard pancreatic proglucagon cDNA generated by RT-PCR was used as a probe for screening a H. suspectum adult pancreatic cDNA library. A total of nine positive cDNA clones were isolated (from approximately 500,000 recombinant cDNA clones screened). Sequence analyses revealed two different classes of lizard proglucagon cDNAs, arbitrarily designated lizard proglucagon I (LPI) and lizard proglucagon II (LPII). LPI cDNAs were comparatively more abundant (than LPII), representing seven of nine clones isolated. The longest LPI cDNA (which was completely sequenced on both strands), was 1,196 bp and contained 85 bp of 5Ј-UTR, a 450-bp open reading frame (ORF) and a 661-bp 3Ј-UTR (Fig. 2B). The structure of the LPI open reading frame (which terminates immediately 3Ј-to the GLP-1 sequence and lacks IP-2 and GLP-2) resembles the pancreatic proglucagon cDNAs previously isolated from fish and chicken (20,21). In contrast, the longest LPII cDNA isolated (completely se- quenced on both strands), was 1,050 bp and contained 95 bp of 5Ј-UTR, a 615-bp ORF, and 340 bp of 3Ј-UTR (Fig. 3B). The structure of the LPII cDNA corresponds to the mammalian proglucagon cDNAs (and the fish and chicken intestinal proglucagon cDNAs) that also contain IP-2 and GLP-2 (2,5,22,23).
Comparison of the two classes of lizard pancreatic proglucagon cDNAs reveals that the nucleotide sequences from 1 to 531 in LPI are identical to the nucleotide sequences from positions 11 to 541 in the LPII cDNAs, resulting in the first 149 amino acids of the LPI and LPII proglucagons being identical, to end of the GLP-1 sequence (Figs. 2 and 3). Immediately 3Ј to the GLP-1 sequence, the sequences of LPI and LPII diverge, with LPI containing a single glutamic acid residue followed by a stop codon (Fig. 2B), whereas the open reading frame of LPII encodes a second intervening peptide (IP-2) and GLP-2 (Fig. 3B). Both LPI and LPII cDNAs contain a similar Kozak sequence at the beginning of the translation start sites (24). Two putative polyadenylation signals are found in the LPII cDNA (at nucleotides 719 -724 and 825-830; Fig. 3B).
As neither LPI or LPII cDNAs contained an exendin 4-like sequence, we hypothesized that exendin 4 may be encoded within a unique lizard intestinal proglucagon cDNA that differed from LPI and LPII, possibly generated by alternative RNA splicing. A series of RT-PCR experiments were carried out using lizard intestinal RNA and two different combinations of lizard proglucagon primers; both primer sets utilized a single common 5Ј-primer (see "Experimental Procedures") and one of two primers corresponding to the unique 3Ј-UTRs (see "Experimental Procedures") from either LPI or LPII (Fig. 4). A single ϳ700-bp band was amplified using the LPII-specific primers. Sequence analysis of the cloned intestinal proglucagon cDNA products determined that the open reading frame encoded by the intestinal cDNA was identical to the sequence of pancreatic LPII (that also encodes IP-2 and GLP-2). In contrast, no intestinal PCR product was generated using LP-1-specific primers (data not shown). These results demonstrate that proglucagon gene expression in lizard intestine differs from pancreas and gives rise to a single class of proglucagon mRNA transcript that includes PGDPs carboxyl-terminal to GLP-1, consistent with previous studies in fish and chicken (20,21).
Analysis of Lizard Proglucagon Gene Expression-To ascertain whether mRNA transcripts related to proglucagon could be detected in salivary gland, the likely site of exendin 4 expression, RNA was isolated from H. suspectum intestine, liver, pancreas, and salivary gland and analyzed by Northern blot-ting with different lizard proglucagon cDNA probes. Three mRNA transcripts of ϳ1.1, 1.6, and 2.1 kb were detected in RNA from lizard pancreas (Fig. 4A). In contrast, only a single 1.1-kb mRNA transcript (that comigrated with the pancreatic transcript) was detected in RNA from lizard intestine with a cDNA probe containing sequences common to both LPI and LPII cDNAs (Fig. 4A). Furthermore, using a cDNA probe specific to the LPI 3Ј-UTR sequences, only the 1.6-and 2.1-kb mRNA transcripts were detected in RNA from lizard pancreas; however, no hybridization was observed using the LPI 3Ј-UTR probe and RNA from lizard intestine (Fig. 4B) Consistent with the results predicted from cDNA cloning, the 1.1-kb mRNA transcript (corresponding to LPII) was observed in RNA from both pancreas and intestine using a cDNA probe specific for LPII (Fig. 4C). In contrast, no hybridizing transcripts were detected in RNA isolated from either lizard salivary gland or liver with any of the three proglucagon cDNA probes (Fig. 4, A-C). The lack of hybridzation using liver and salivary gland RNA was not attributable to any noticeable differences in the quality of the various RNA preparations (Fig. 4D).
Isolation of Exendin Salivary Gland cDNAs-To isolate exendin cDNA sequences from lizard salivary gland RNA, we prepared degenerate oligonucleotide primers against the lizard exendin 4 amino acid sequence for use in RT-PCR experiments (Fig. 5A). RNA from lizard salivary gland generated a PCR product of the predicted size (Fig. 5B), and sequencing of this partial cDNA demonstrated that it encoded the sequence of exendin 4 (Fig. 5C). To isolate a full-length exendin cDNA, we prepared and screened a lizard salivary gland cDNA library with the partial exendin cDNA probe. A single class of exendin cDNA clones was isolated from the library that encoded exendin and a 45-amino acid peptide NH 2 -terminal to the exendin sequence, designated exendin NH 2 -terminal peptide (ENTP). The ENTP sequence was followed by a dibasic pair of amino acids, lysine and arginine, characteristic of prohormone convertase cleavage sites, and the 40-amino acid exendin sequence (Fig. 5D). No peptide sequences exhibiting homology to ENTP were found following search of the latest GenBank™ release.
Northern blot analysis carried out using lizard RNA demonstrated the presence of a single exendin mRNA transcript in RNA from salivary gland (Fig. 6A). In contrast, no exendin mRNA transcripts were detected in RNA from lizard liver, pancreas, or intestine. The complexity of organization of the lizard exendin and proglucagon gene(s) was examined by Southern blot analysis   (Fig. 6, B and C). A simple pattern of hybridization was obtained using the exendin probe, clearly distinct from the bands generated with a lizard proglucagon cDNA probe common to the open reading frames of LPI and LPII. For example, a doublet of less than ϳ3.0 kb and a second fragment of ϳ1.8 kb was detected following EcoRI digestion (and hybridization) with the exendin probe. In contrast, hybridizing fragments of ϳ5.7 and 4.0 kb and a doublet of ϳ3.2 kb was detected with the proglucagon probe ( Fig. 6, B and C). These results are consistent with unique exendin and proglucagon genes and the presence of one, but most likely two H. suspectum proglucagon genes, as described previously for fish (21,25,26). DISCUSSION The biological diversity of the proglucagon-derived peptides in mammals derives from the cell-specific expression of prohor- mRNA transcript (and ORF) is present in pancreas and intestine from chicken or fish.
In contrast to proglucagon gene expression in mammals, chicken, or fish, the characterization of proglucagon gene expression in H. suspectum described here is clearly unique for several reasons. First, two structurally different classes of proglucagon mRNA transcripts, encoding different proglucagons and unique 3Ј-UTRs were isolated from lizard pancreas, but only a single type of proglucagon cDNA was obtained from lizard intestine. Furthermore, the 1.1-kb (LPII) transcript (that contains the largest ORF with proglucagon sequences COOH-terminal to GLP-1) is expressed in both pancreas and intestine, likely as a result of alternative RNA splicing. In contrast, the pancreas-specific unique 1.6-and 2.1-kb transcripts represent LPI cDNAs that encode glucagon and GLP-1, but not GLP-2 (Fig. 4). The likely explanation for the generation of three different pancreatic proglucagon mRNAs is alternative splicing of a single (or possibly two highly related) lizard proglucagon RNA(s) to generate mRNAs with different open reading frames and unique 3Ј-untranslated regions.
Although we did not isolate proglucagon cDNAs approximating 1.6 or 2.1 kb in size after screening the pancreatic cDNA library, the size of these two RNA transcripts may be attributable in part to either a long poly(A) tail, differential utilization of polyadenylation sites in the 3Ј-UTR, alternative RNA splicing (perhaps to include additional 3Ј-UTR sequence) of a mature transcript and/or incomplete processing of a nuclear RNA precursor (for the 2.1-kb RNA). Precedents for several of these possibilities derives from previous studies of proglucagon gene expression in various species. A second smaller pancreatic proglucagon mRNA transcript was identified and characterized in a hamster islet cell line (31) that arises as a consequence of the utilization of a more proximal poly(A) consensus sequence in the hamster proglucagon 3Ј-UTR. Furthermore, different classes of pancreatic proglucagon mRNA transcripts (with identical open reading frames (but longer poly(A) tails compared with wild-type pancreatic proglucagon RNAs) have been detected in RNA isolated from human and rat pancreatic endocrine tumors and fetal pancreas (32,33), and alternative RNA processing likely accounts for tissue-specific expression of proglucagon mRNAs in fish and chicken (20,21). Although the precise structural basis for the difference in size of the 1.6-and 2.1-kb lizard pancreatic proglucagon mRNAs remains to be determined, the results of our RT-PCR, Northern blotting and cDNA cloning experiments strongly suggest that these two LPI mRNAs likely share common open reading frames and at least some 3Ј-UTR sequence, in contrast to the different ORF and 3Ј-UTR of the LPII mRNA. Furthermore, the nucleotide sequences of the LP1 and LPII cDNAs extending up to GLP-1 are identical. Hence LPI and LPII may be derived from the same gene and alternative splicing accounts for the carboxyl-terminal differences, or alternatively, they may represent the products of two highly related genes.
An initial goal of these studies was the clarification of the possible relationship between exendin 4 and lizard GLP-1. Although exendin 4, as isolated and sequenced from H. suspectum venom, is 52% identical to mammalian GLP-1 (13), the cloning of lizard proglucagon cDNAs demonstrated that the sequence of exendin 4 is not encoded by lizard preproglucagon, and lizard exendin 4 is now shown to be only 45% identical to lizard GLP-1. The isolation of exendin 4 cDNA clones from lizard salivary gland, taken together with the results of Northern and Southern blot analyses, clearly establish for the first time that exendin 4 and lizard GLP-1 are unique peptides encoded by different genes.
Lizard GLP-1 shares 90 and 84% identity to chicken and human GLP-1, respectively, and lizard glucagon shares 92 and 97% identity to chicken and human glucagon, respectively (Table I). In contrast, although lizard and chicken GLP-2 are reasonably similar in sequence (76% identity), lizard GLP-2 is much less similar than the rat or human (46%) peptides. Furthermore, lizard and chicken IP-1s are both much larger than all other mammalian and fish IP-1 sequences. This also implies that lizard and chicken oxyntomodulin and glicentin, that both contain the IP-1 sequence at their carboxyl terminus, are considerably larger than their mammalian counterparts. Taken together, the differential conservation of peptide homologies raises the possibility that some of the lizard PGDPs may have diverged not only in sequence, but also in biological function, from their mammalian counterparts. Interestingly, although rat and human GLP-1 are separated from IP-1 by a pair of basic amino acid residues (KR), the cognate lizard sequence is RH. The absence of the dibasic amino acid pair at the NH 2 terminus of lizard GLP-1 raises the possibility that the amino terminus of lizard GLP-1 may be recognized and processed in a manner that differs from the processing observed for vertebrate GLP-1s. In contrast, the first 2 amino acids of truncated GLP-1, HA, are identical in the human and lizard. The functional implications of the NH 2 -terminal differences in lizard GLP-1 sequence remain to be established.
Taken together, the demonstration that exendin 4 is encoded by a lizard gene distinct from proglucagon in the lizard raises the possibility that other species, perhaps mammals, may also contain distinct exendin genes. The observation that lizard exendin 4 binds to the mammalian GLP-1 receptor and stimulates insulin secretion, whereas a truncated exendin 4-(9 -39) peptide functions as an antagonist of the mammalian GLP-1 receptor both in vitro and in vivo (34) provides indirect evidence supporting the existence of a mammalian exendin 4 gene with potential physiological implications for control of insulin secretion in vivo.