Monotreme glucagon-like peptide-1 in venom and gut: one gene – two very different functions

The importance of Glucagon like peptide 1 (GLP-1) for metabolic control and insulin release sparked the evolution of genes mimicking GLP-1 action in venomous species (e.g. Exendin-4 in Heloderma suspectum (gila monster)). We discovered that platypus and echidna express a single GLP-1 peptide in both intestine and venom. Specific changes in GLP-1 of monotreme mammals result in resistance to DPP-4 cleavage which is also observed in the GLP-1 like Exendin-4 expressed in Heloderma venom. Remarkably we discovered that monotremes evolved an alternative mechanism to degrade GLP-1. We also show that monotreme GLP-1 stimulates insulin release in cultured rodent islets, but surprisingly shows low receptor affinity and bias toward Erk signaling. We propose that these changes in monotreme GLP-1 are the result of conflicting function of this peptide in metabolic control and venom. This evolutionary path is fundamentally different from the generally accepted idea that conflicting functions in a single gene favour duplication and diversification, as is the case for Exendin-4 in gila monster. This provides novel insight into the remarkably different metabolic control mechanism and venom function in monotremes and an unique example of how different selective pressures act upon a single gene in the absence of gene duplication.

has not been fully resolved but it seems likely that exendins evolved by duplication from a glucagon-like peptide gene precursor 14 .
We investigated the incretin hormone system in monotremes because of their phylogenetic position as the most basal lineage amongst extant mammals. In addition monotremes feature extraordinary changes in their digestive system, which has led to loss of genes involved in protein degradation and metabolic control 18,19 , suggesting unique mechanisms of metabolic homeostasis. Also, monotremes produce venom during breeding season and in platypus there is an elaborate venom delivery system, apparently used in competition for mating 20,21 . These observations prompted us to investigate the expression and function of incretin hormones in monotremes. To our surprise we discovered that only a single GCG encodes a GLP-1 peptide with two functions, one in venom and the other in the gut. Detailed in vitro characterisation of monotreme GLP-1 peptides revealed evolutionary signatures that can be explained by selective conflict as a result of the recruitment of this molecule into venom.

Results
Sequence variations in monotreme Gcg genes are present in regions known to be important for the regulation of protein function. Gcg, Glp-1r, Gip and Dpp-4 genomic sequences were identified in the platypus genome database. At the beginning of this study there were no reported transcript sequences for these genes and some of the genomic sequences were incomplete. In this study for the first time cDNA sequences encoding platypus and echidna preproglucagon (encoded by the GCG gene) were identified (Supplementary Table S2). Phylogenetic analysis of the preproglucagon amino acid sequence in vertebrates revealed expected tree topology but also highlights sequence divergence in the monotreme lineage (Fig. 1b). The proglucagon peptide (Fig. 1a) includes a signal peptide (SP), glicentin-related polypeptide (GRPP), glucagon (GCG), intermediate peptides-1 and -2 (IP-1 and IP-2), glucagon like peptides 1 and 2 (GLP-1 and GLP-2), and is proteolytically processed into the mature glucagon-like peptides. While therian mammal GCG genes encode identical GLP-1 protein sequences, there are significant changes in the platypus Gcg ortholog. Importantly the inferred sequence of the platypus GLP-1 peptide (pGLP-1) differs in 11 of the 30 amino acids (37%) compared to human GLP-1 (hGLP-1, Fig. 1a, as reported previously 8 ). A smaller difference was seen between platypus and human GLP-2 (30%) or glucagon peptides (20%) (Supplementary Fig. S1). Notably we also discovered specific changes in the DPP-4 cleavage site in pGLP-1 (pGLP-1, Ala 8 to Ser) (Fig. 1a) and in the platypus GIP peptide ( Supplementary Fig. S1, Ala 2 to Ser), whereas there was no change to the cleavage sites in platypus glucagon or GLP-2 ( Supplementary Fig. S1). To investigate if this change is also present in the echidna GLP-1 (eGLP-1) we cloned the echidna Gcg (eGcg) transcript and found a different amino acid at residue eight (Ala 8 to Phe) in the eGLP-1 DPP-4 cleavage site (Fig. 1a), as well as a total of 17 differences from hGLP-1 of the 30 amino acids (57% changed). We also saw differences in echidna GLP-2 (pGLP-2) from hGLP-2 (45% changed), and in echidna glucagon from human glucagon (17% changed) ( Supplementary Fig. S2). However, there was no change to the DPP-4 cleavage sites in both peptides. Remarkably, GLP-1 peptide sequence comparisons revealed a total of 12 differences between the two monotreme sequences (i.e. 40% of the sequence), indicating not only divergence from other mammals but major divergence within the monotremes, which separated only 17-48 Million years ago 22 (Fig. 1b). These results raise fundamental questions about stability and potency of monotreme GLP-1.
Monotreme Gcg and Dpp-4 exhibit similar tissue expression patterns to other mammals, but both genes are also expressed in venom. Expression analysis showed that platypus and echidna Gcg, Glp-1r and Dpp-4 tissue expression is similar to other mammals (Fig. 2) 23,24 , suggesting they play a similar role in monotremes. Surprisingly, both Gcg and Dpp-4 genes are also expressed in platypus and echidna venom (Fig. 2). In monotremes it is the same Gcg gene encoding GLP-1 that is expressed in gut and venom. This is in contrast to Heloderma suspectum, where the Gcg is expressed in gut and Exendin-4 in venom 15 .

Monotreme GLP-1 peptides are resistant to cleavage by human DPP-4 and human serum components but are cleaved in platypus and echidna sera, independently of DPP-4 activity.
Sequence alignments between human, platypus and echidna GLP-1 proteins (Fig. 1) revealed substitutions at the known DPP-4 cleavage site (Ala 8 in human) to Ser or Phe in platypus and echidna respectively, predicting that DPP-4 cleavage could be affected. To test if the specific amino acid changes at the DPP-4 cleavage site in pGLP-1 and eGLP-1 result in resistance to degradation, we compared their cleavage to that of human GLP-1 (hGLP-1) and the DPP-4 resistant exendin-4 (Ex-4) (Fig. 3). Incubation of the peptides with purified human DPP-4 resulted in rapid degradation of hGLP-1 (50% reduction of intact peptide within 1 hour) but not Ex-4. Significantly, both echidna and platypus GLP-1 were not degraded by human DPP-4 ( Fig. 3a), confirming that monotreme GLP-1 is resistant to DPP-4 cleavage. Next we investigated stability in human serum. In albumin-depleted human serum, platypus and echidna GLP-1 and Ex-4 remained stable whereas hGLP-1 was rapidly degraded (Fig. 3b). To investigate if monotremes employ a different way to break down GLP-1, we measured cleavage in platypus and echidna serum. Surprisingly, degradation of platypus and echidna GLP-1 was observed when incubated in platypus and echidna sera (Fig. 3c,d). Degradation was slower than for hGLP-1 but clearly measurable, with less than 50% uncleaved pGLP-1 and eGLP-1 remaining after 11 hours of incubation (Table 1). Interestingly, Ex-4 was cleaved slowly in echidna serum but remained intact in platypus and human sera. DPP-4 is not the only enzyme that can degrade GLP-1. Human neural endopeptidase (NEP24.11), for example also cleaves GLP-1 but utilizes different target sites within the peptide 25 . To further investigate whether monotremes evolved a DPP-4 independent pathway to degrade GLP-1, we firstly confirmed the presence of DPP-4 in platypus and echidna sera using a synthetic DPP-4 peptide as a substrate and a DPP-4 inhibitor to prevent this cleavage ( Supplementary Fig. S3), thus showing that cleavage was due to DPP-4. Despite the presence of DPP-4 inhibitor we were able to detect cleavage of both monotreme GLP-1 peptides ( Supplementary Fig. S4), showing that a different enzyme(s) must be responsible for their cleavage in monotreme sera. To gain further insight into the mechanism of degradation we used mass spectrometry to analyse the pGLP-1 and eGLP-1 cleavage products. We saw cleavage products that suggested trypsin or chymotrypsin-like activity with cleavage after basic and hydrophobic residues (Fig. 4). Together this supports the idea that in monotremes a DPP-4 independent system has evolved to regulate GLP-1 half-life and activity.

Monotreme GLP-1 peptides bind with lower affinity to the GLP-1 receptor than human GLP-1.
We then asked how changes in pGLP-1 and eGLP-1 affect binding and activation of the GLP-1 receptor (GLP-1R). All of the known key hGLP-1 residues (underlined in Fig. 1a) involved in binding to the human GLP-1R (hGLP-1R) core and four of the six C-terminal residues (excepting Ala 25 to Thr and Val 33 to Leu) involved in binding to the hGLP-1R N-terminal domain are conserved in pGLP-1 [26][27][28] . Interestingly, in echidna GLP-1 there is less conservation of the receptor binding residues with additional changes at the N-terminal receptor binding Phe 12 (conservatively substituted to Tyr) and Asp 15 (changed to Glu) residues. hGLP-1 Gly 22 , involved in kinking of the helix, is a Glu in the extended Ex-4 helix, leading to different modes of interaction with the GLP-1R 29 . The same substitution is seen in the monotreme GLP-1 sequences (Fig. 1a). The platypus GLP-1R amino acid sequence (deduced in this study from a sequenced cDNA transcript) is similar to the hGLP-1R (76% identity compared to hGLP-1R), including conservation of the residues important for ligand binding (Fig. 5). The pattern of pGLP-1R expression (Fig. 2b) is also similar to other mammals 30 . Receptor binding assays on hGLP-1R overexpressing cells showed that, compared to hGLP-1 and Ex-4 both platypus and echidna GLP-1 peptides have lower affinity for the human receptor (Fig. 6a, Table 2). hGLP-1 has an almost identical affinity for the platypus GLP-1R (pGLP-1R) and the human receptor, but unexpectedly both monotreme GLP-1 peptides had a significantly lower affinity than hGLP-1 for the pGLP-1R (Fig. 6b, Table 2). For both receptors monotreme GLP-1 peptides were equipotent with the GLP-1R agonist oxyntomodulin (OXM).

Monotreme GLP-1 peptides are less potent in their activation of the GLP-1 receptor.
We then investigated if this difference in affinities translates into a difference in activation of the GLP-1 receptor. As expected monotreme GLP-1 peptides showed significantly less potency than hGLP-1 in assays measuring cAMP accumulation, Ca 2+ mobilization and ERK1/2 phosphorylation acting through both human and platypus receptors (Fig. 6c-e). eGLP-1 showed a markedly lower potency at both human and platypus receptors that was even lower than OXM. Differences in the structure of monotreme GLP-1 peptides compared with hGLP-1 could account for the lower affinity for the receptor. Circular dichroism spectroscopy (CD) on Ex-4 and hGLP-1 yielded results similar to previously published data 26 and showed that all peptides utilized were folded correctly ( Supplementary Fig. S5). All peptides retained significant helical content, although pGLP-1 had more and eGLP-1 had slightly less than hGLP-1. As has been seen with Ex-4 26 , a difference in helical content can result in a different mode of interaction with GLP-1R.  Monotreme GLP-1 elicits a different signalling cascade compared to hGLP-1 when binding to the GLP-1 receptor. Closer examination of potencies in receptor activation revealed differential signalling bias for monotreme GLP-1 peptides in comparison to that elicited by hGLP-1. Distinct signalling bias arising through activation of the GLP-1R by different ligands (including OXM) has recently been established and may, at least in part, underlie differences in the physiological profile of naturally occurring ligands of the GLP-1R 12 . Indicators used to determine the signalling profile of peptides include cAMP and intracellular Ca 2+ mobilisation, which are involved in promotion of insulin release, and pERK1/2 that is part of the mitogenic signalling pathways activated via the GLP-1R 12 . Intriguingly, both the platypus and echidna GLP-1 peptides displayed a distinct pattern of signalling in comparison to hGLP-1 and the clinically approved mimetic Ex-4, which was apparent at both the human and platypus GLP-1 receptors (Fig. 6, Supplementary Figs S6 and S7, Table 3). The signalling profile of the monotreme GLP-1 peptides closely matched that of OXM with a bias towards pERK1/2, and to a lesser extent iCa 2+ , relative to cAMP (Fig. 6, Supplementary Figs S6 and S7, Table 3), although the bias towards calcium mobilisation was less apparent for the pGLP-1 at the human receptors (Fig. 6). These observations suggest that monotreme GLP-1 peptides may have gained new as yet undefined functions.
Platypus GLP-1 stimulates insulin release in mouse islet cells. Ultimately, the signal cascade triggered by incretins results in the release of insulin from pancreatic islet cells. We investigated the ability of pGLP-1 to stimulate insulin release from isolated mouse islets. Results showed that 100 nM pGLP-1 can stimulate insulin release in vitro similar to hGLP-1 (Fig. 6f). It appears at least at mouse islet GLP-1R that pGLP-1 would act with classical incretin function to promote insulin release, although whether this is the primary function in the platypus remains to be proven.

Discussion
Incretin hormones play a key role in regulation of mammalian metabolism through regulation of insulin secretion as well as by decreasing gastric emptying, food intake, body weight and increasing satiety. The importance of insulin regulation and metabolic control has led to recruitment of genes in this pathway in venom function in various species. Of medical importance, the discovery of the DPP-4 protease resistant exendin-4 from venom of Heloderma suspectum paved the way for the successful development of GLP-1 analogues as important treatments for insulin resistant type 2 diabetics.
Monotremes are a fascinating species to investigate this as they represent the third and most basal lineage of extant mammals. Importantly they have undergone remarkable changes of their digestive system and feature production of potent venom for intraspecific conflict during the breeding season. This prompted us to investigate the incretin hormone GLP-1, its receptor (GLP-1R) and the GLP-1 regulatory enzyme DPP-4 in monotremes.
Overall the key genes in this pathway are conserved in monotremes showing that the insulin release pathway Degradation of hGLP-1, pGLP-1 and eGLP-1 in echidna serum for seven hours was monitored by RP-HPLC, fragments were collected and analysed by mass spectometry. Known cleavage sites of DPP-4 and NEP24.11 are shown above the hGLP-1 sequence by red and green arrows, respectively. Identified cleavage sites in pGLP-1 and eGLP-1 are highlighted by blue arrows.
has been maintained despite the extensive changes in the digestive system. However, we discovered remarkable changes in GLP-1 sequence and function, which we propose are the result of an unusual evolutionary trajectory of this important gene. When we compared platypus and echidna GCG we discovered more changes than would be expected given that monotremes diverged only 17-48 Million years ago. This accelerated rate in evolutionary change maybe driven by new roles for this gene.
Indeed we discovered that the same monotreme Gcg gene encoding GLP-1 is expressed in monotreme gut and venom. In contrast, Ex-4 is only found in the venom of the lizard Heloderma suspectum, with the endogenous GLP-1 being DDP-4 sensitive and 84% identical at the amino acids level to human GLP-1 14,15 . Further changes involving receptor affinity and activation occurred specifically in monotremes. The GLP-1 peptide found in all other mammalian species binds the human GLP-1R with an IC50 of 7.9 × 10 −10 M (Table 2), whereas the monotreme GLP-1 peptides bind the human and platypus GLP-1R with 50 fold lower affinity. As human GLP-1 is able to bind with high affinity and potently activate the platypus GLP-1R ( Fig. 6 and (Supplementary Fig. S6), we conclude that the platypus receptor functions in a similar manner to hGLP-1R. The significance of the low pGLP-1R binding affinity for monotreme GLP-1 peptides to biological activity is as yet unclear.
The phenomenon of ligand-directed signalling bias for GLP-1R has recently been described using the endpoints of cAMP production, pERK1/2 and intracellular Ca 2+ . For example, hGLP-1 and OXM exhibit bias for cAMP over pERK1/2, whereas Ex-4 and GLP-1  are not biased. In platypus and echidna the low pGLP-1R binding promote a bias towards pERK1/2 signalling compared to human GLP-1. This indicates that the differences in monotreme sequences promote a change in the way the peptides interact with the GLP-1R and engendering the receptor with a unique conformation to drive distinct receptor function. It also suggests that monotreme GLP-1 peptides might promote unique, as yet unknown functions through the montreme GLP-1Rs. It will be interesting to understand the functional consequences of these changes, which are relevant for possible application in the development of GLP-1R agonists for medical application in diabetes treatment and for better understanding of the function of these changes in monotreme GLP-1. Lineage specific changes in the sequence and function have been described in hystricomorph rodents for the key metabolic gene insulin 31 . However, monotremes present the first example were such a gene, GCG in this case, appears to maintain its role in gut as well as adopting a new role in venom. The fact that the venom is used towards members of the same species during the mating season may have favored the use of the endogenous gene as the most effective venom component. This may have set the scene for an adaptive conflict where on one side selection would favour tolerance to a spike in GLP-1 in blood as a result of envenomation and on the other increased potency as a venom component. At the same time insulin releasing function of GLP-1 has to be maintained. Insulinotropic effects associated with GLP-1-like peptides Residues different between hGLP-1R and pGLP-1R are highlighted in red. Where a residue is marked blue or green but is different to hGLP-1 it is highlighted by a red circle. The putative signal peptide cleavage site is depicted with a black arrow.
Scientific RepoRts | 6:37744 | DOI: 10.1038/srep37744 have been reported in a range of venomous species including arthropods, reptiles [32][33][34][35][36] , and recently in cone snail venom, which induces severe hypoglycemic shock in its fish prey 16 . In contrast to the evolution of Gcg-like genes and insulin mimetics in other species, monotremes are the first examples of species that have recruited the endogenous GLP-1 system into venom. The fact that monotreme GLP-1 is also expressed in monotreme venom raises the possibility that DPP-4 resistance is selected for when GLP-1 like molecules are recruited into venom function. However, monotremes use venom during intraspecific conflict rather than for envenomation of prey, as is the case for most other venomous species. While the DPP-4 resistance is likely a change to enhance the insulinotropic effect, the changes in the degradation system may have evolved as a countermeasure to the venom's effect  mediated by GLP-1 on blood glucose. Further more, it may be that the decreased affinity for the receptor has also evolved as a protective mechanism in response to the use of GLP-1 in venom. Clearly the insulin releasing property of GLP-1 is common in tetrapods but its function in venom is novel and monotreme specific. Such acquisition of new function, termed moonlighting, is observed in many genes. However, if a newly acquired function results in selective conflict it has been postulated that gene duplication is favoured 37 . Alternatively functional diversification can be the result of duplication events, which maybe the case for the many venom genes, including exendins that are likely to be the result of gene duplication involving GCG and GIP 9,14 . In monotremes the changes observed indicate that there is selective conflict between GLP-1's function in venom and its traditional function. It is therefore surprising that GCG gene duplication has not been selected for. However, gene duplication would mean an increase in gene dosage, which would likely be deleterious as insulin release and glucose tolerance is sensitive to GLP-1 dosage 38,39 .
In summary, we propose that in monotremes an evolutionary arms race between the function of GLP-1 in gut and in venom can explain the changes observed. Evolution of DPP-4 independent GLP-1 degradation and decreased receptor activation may have evolved in response to GLP-1 in venom. The independent evolution of these components affecting glucose homeostasis and insulin release also highlights the importance of metabolic control as a target for venomous species. This maybe the first example of a gene where selective conflict has not favoured the evolution of gene duplication. RNA extraction. Platypus and echidna tissues were obtained from adult animals in accordance with ethics guidelines (approved by Adelaide University Animal Ethics Committee permit AEC S-49-200 to F.G). Total RNA was extracted from snap frozen platypus tissues (frontal cortex, pancreas, liver, lung, small intestine, stomach, heart, venom gland, testis, muscle, lymph and kidney) and echidna tissue (small intestine, pancreas, liver, venom gland, heart and brain) using TRIzol (Invitrogen, USA) according to the manufacture's instructions. RNA was resuspended in nuclease free water and stored at −80 °C.

Materials
Phylogeny. An evolutionary comparison was made of monotreme GCG genes with orthologues in other vertebrate species (encoded by GCG genes listed in Supplementary Table S2). The phylogenetic tree was constructed based on the preproglucagon multiple amino acid sequence alignment using ClustalW2 41 . Neighbor Joining algorithm with bootstrap analysis using 1000 replicates was conducted in MEGA4 software using standard settings 42 . The evolutionary distances were computed using the Poisson correction method and are in the units of the number of amino acid substitutions per site 43 . The analysis involved 9 amino acid sequences. All positions containing gaps and missing data were eliminated. cDNA synthesis. cDNA was synthesized from 3 μg RNA with Superscript III Reverse Transcriptase  Circular dichroism spectroscopy (CD). Far-UV CD spectra were measured on a Jasco J-815 spectropolarimeter (Jasco Inc., Easton, MD) at 20 °C in a 1 mm quartz cuvette. The scanning range was 185-300 nm at a speed of 20 nm/min, the bandwidth was 1 nm, and the spectra were accumulated 5 times. The concentration of hGLP-1, pGLP-1, eGLP-1 and Ex-4 was 75 μM in 10 mM sodium phosphate buffer (pH 7.0). The secondary structure of each peptide was estimated using the CONTIN algorithm 44,45 . Transfections and cell culture. Human and pGLP-1R cDNAs were isogenically integrated into FlpIn-Chinese hamster ovary (FlpInCHO) cells (Invitrogen) and selection of receptor-expressing cells accomplished by treatment with 600 μg/ml hygromycin B as described previously 46 . Transfected and parental FlpInCHO cells were maintained in DMEM supplemented with 10% heat-inactivated FBS and incubated in a humidified environment at 37 °C in 5% CO 2 . INS-1(832/13) cells were cultured in RPMI-1640 medium supplemented with 10 mM HEPES, 10% heat-inactivated FBS, 2 mM L-glutamine, 1 mM sodium pyruvate, 0.05 mM β-mercaptoethanol, 100 iU/ml penicillin, and 100 g/ml streptomycin at 37 °C in a humidified 5% CO 2 atmosphere.
Radioligand binding assay. FlpInCHO-hGLP-1R and FlpInCHO-pGLP-1R cells were seeded at a density of 3 × 10 4 cells/well and INS-1(832/13) cells at 10 5 cells/well into 96-well culture plates and incubated overnight at 37 °C in 5% CO 2 . Growth media was replaced with binding buffer [phenol-free DMEM containing 25 mM HEPES and 0.1% (w/v) BSA] containing 0.7 nM 125 I-Ex(9-39) or 0.15 nM 125 I-hGLP-1 and increasing concentrations of unlabelled ligand. Cells were then incubated overnight at 4 °C, followed by three washes with ice-cold 1 × PBS to remove unbound radioligand. Cells were then lysed in 0.1 M NaOH, and radioactivity determined by γ-counting as described previously 47 . cAMP accumulation assay. FlpInCHO containing 5 mM HEPES, 0.5 mM 3-isobutyl-1-methylxanthine (IBMX) and 0.1% (w/v) BSA], and cAMP detection was carried out as described previously 48 . All values were converted to concentration of cAMP using a cAMP standard curve performed in parallel, and data were subsequently normalized to the response of 100 nM forskolin.

Measurement of insulin.
Insulin concentrations were determined by a commercially available radioimmunoassay specific for rodent insulin (Linco Research Immunoassay, St. Charles, MO) as previously described 50 .
Data analysis. All data were analysed in Prism 6.0c (GraphPad Software Inc., San Diego, CA). Concentration response signalling data were analysed using a three-parameter logistic equation as described previously 46,47 .
Signalling bias was analysed as described 51 . Briefly, quantification of signal bias was performed using pharmacologically derived parameters of agonist affinity (Ka) and efficacy (tau) for each ligand in each of the three signalling pathways (cAMP accumulation, ERK1/2 phosphorylation, Intracellular Ca 2+ mobilisation). The transduction ratio (tau/Ka) was extracted from standard concentration-response data that was analysed with the operational model of agonism (Kenakin & Christopoulos 2012). This value was used to calculate ΔΔ(tau/Ka) values through normalization of the transduction coefficient (tau/Ka) for each ligand in each signalling pathway to the reference ligand (hGLP1 in black) and the reference signalling pathway (cAMP). Data are presented on a log scale. Statistical analysis was by One-way ANOVA (nonparametric) with Dunnett's post test unless otherwise stated in the figure legends.