0163-769X/99/$03.00/0
`Endocrine Reviews 20(6): 876–913
`Copyright © 1999 by The Endocrine Society
`Printed in U.S.A.
`
`The Glucagon-Like Peptides
`
`TIMOTHY JAMES KIEFFER* AND JOEL FRANCIS HABENER†
`Departments of Medicine and Physiology (T.J.K.), University of Alberta, Edmonton, Alberta, Canada
`T6G 2S2; and Laboratory of Molecular Endocrinology (J.F.H.), Massachusetts General Hospital,
`Howard Hughes Medical Institute, Harvard Medical School, Boston, Massachusetts 02114
`
`I. Introduction
`II. History of the Incretin Concept: Discovery of Gastric
`Inhibitory Polypeptide
`III. Discovery of GLP-1
`IV. Structures of GLPs and Family of Glucagon-Related
`Peptides
`V. Tissue Distribution of the Expression of GLPs
`A. Pancreatic a-cells
`B. Intestinal L cells
`C. Central nervous system
`VI. Proglucagon Biosynthesis
`A. Organization/structure of the proglucagon gene
`B. Regulation of glucagon gene expression
`C. Posttranslational processing of proglucagon
`VII. Regulation of GLP Secretion
`A. Overview
`B. Intracellular signals
`C. Carbohydrates
`D. Fats
`E. Proteins
`F. Endocrine
`G. Neural
`H. GLP-2
`VIII. Metabolism of GLPs
`A. GLP-1
`B. GLP-2
`IX. Physiological Actions of GLPs
`A. Overview
`B. Pancreatic islets
`C. Counterregulatory actions of GLP-1 and leptin on
`b-cells
`D. Stomach
`E. Lung
`F. Brain
`G. Liver, skeletal muscle, and fat
`H. Pituitary, hypothalamus, and thyroid
`I. Cardiovascular system
`J. GLP-2
`
`Address reprint requests to: Joel F. Habener, M.D., Laboratory of
`Molecular Endocrinology, Massachusetts General Hospital, 55 Fruit
`Street, WEL320, Boston, Massachusetts 02114 USA. E-mail: jhabener@
`partners.org
`* Received support from the Medical Research Council of Canada, the
`Alberta Heritage Foundation for Medical Research, and the Canadian
`Diabetes Association. E-mail: tim.kieffer@ualberta.ca
`† Investigator with the Howard Hughes Medical Institute and re-
`ceived support from US Public Health Service grants DK-30834, DK-
`25532, and DK-30457.
`
`X. GLP Receptors
`A. Structure
`B. Signaling
`C. Distribution
`D. Regulation
`E. GLP-2
`XI. Pathophysiology of GLP-1
`XII. GLP-1 as a Potential Treatment for Diabetes Mellitus
`XIII. Future Directions
`
`I. Introduction
`
`IT HAS been 15 yr since the initial discovery of the glu-
`
`cagon-like peptides (GLPs) as potential bioactive pep-
`tides encoded in the preproglucagon gene. The GLPs and
`glucagon are formed by alternative tissue-specific cleavages
`in the L cells of the intestine, the a-cells of the endocrine
`pancreas, and neurons in the brain. Glucagon-like peptide-1
`(GLP-1) is now known to be a potent glucose-dependent
`insulinotropic hormone, which has important actions on gas-
`tric motility, on the suppression of plasma glucagon levels,
`and possibly on the promotion of satiety and stimulation of
`glucose disposal in peripheral tissues independent of the
`actions of insulin. As a consequence of these properties,
`GLP-1 is under investigation as a potential treatment of di-
`abetes mellitus. GLP-2 was recognized only recently to have
`potent growth-promoting activities on intestinal epithelium.
`The interest in the GLPs has grown exponentially. By 1988
`there were 170 publications describing the properties of the
`GLPs. Five years later this number grew to 426 and currently
`(1999) more than 1,000 publications appear in the database
`of the National Library of Medicine (PubMed).
`Since the last comprehensive review of GLP-1 appeared in
`Endocrine Reviews in 1995 (1), many new developments have
`occurred and are described in this review. The purpose of
`this article is to emphasize the newer and what are perceived
`to be the more current and important aspects of the biology
`of the GLPs. For additional information and references, the
`reader is referred to several informative earlier reviews
`(1-12).
`
`II. History of the Incretin Concept: Discovery of
`Gastric Inhibitory Polypeptide
`
`As a result of their discovery of secretin in 1902, Bayliss
`and Starling (13) speculated that signals arising from the gut
`after ingestion of nutrients might elicit pancreatic endocrine
`
`876
`
`CFAD Exhibit 1018
`
`The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 12 February 2015. at 12:28 For personal use only. No other uses without permission. . All rights reserved.
`
`1
`
`

`

`December, 1999
`
`THE GLUCAGON-LIKE PEPTIDES
`
`877
`
`responses and affect the disposal of carbohydrates. In 1906
`Moore et al. (14) postulated that the duodenum produced a
`‘chemical excitant‘ for pancreatic secretion and attempted to
`treat diabetes by injecting gut extracts. Zunz and Labarre (15,
`16) pursued this factor and prepared an intestinal extract free
`of secretin activity that was able to produce hypoglycemia in
`dogs. Labarre (16) introduced the term ‘incretin‘ to describe
`the humoral activity of the gut that might enhance the en-
`docrine secretion of the pancreas (16). Although other in-
`vestigators also reported the presence of hypoglycemic fac-
`tors in duodenal extracts (17-20), Loew and colleagues (21)
`were unable to lower blood glucose levels in dogs with
`extracts of dog or hog intestinal mucosa obtained by a num-
`ber of methods. Although these later extracts were tested
`only in fasting animals, after this report, interest in isolating
`an intestinal hypoglycemic factor declined.
`The development of a reliable RIA for insulin in the 1960s
`by Yalow and Berson (22), which allowed measurements of
`the circulating levels of this hormone, renewed interest in the
`search for incretins. It was demonstrated by both immuno-
`assay (23, 24) and bioassay (25, 26) that the action of glucose
`on the pancreas could not account completely for the insulin
`response observed in the blood. These reports demonstrated
`that iv glucose administration resulted in a lower plasma
`insulin response than when given by intrajejunal infusion,
`even though lower blood glucose levels were achieved by the
`later (Fig. 1). Perley and Kipnis (27) estimated the alimentary
`component to be close to 50% by subtracting from the insulin
`secretory response seen after oral glucose that insulin re-
`sponse obtained with the infusion of iv glucose, which du-
`plicated the oral blood glucose profile.
`In 1969, Unger and Eisentraut (28) named the connection
`between the gut and the pancreatic islets the ‘enteroinsular
`axis.‘ Creutzfeldt (29) suggested that this axis encompasses
`nutrient, neural, and hormonal signals from the gut to the
`islet cells secreting insulin, glucagon, somatostatin, or pan-
`
`creatic polypeptide (Fig. 2). Furthermore, Creutzfeldt (29)
`defined the criteria for fulfillment of the hormonal or incretin
`part of the enteroinsular axis as: 1) it must be released by
`nutrients, particularly carbohydrates, and 2) at physiological
`levels, it must stimulate insulin secretion in the presence of
`elevated blood glucose levels.
`One hormone that clearly fits the requirements to be an
`incretin is glucose-dependent insulinotropic polypeptide
`(GIP). GIP was originally isolated as an ‘enterogastrone,‘ or
`hormone secreted in response to fat or its digestive products
`in the intestinal lumen that inhibits gastric acid secretion (30).
`Brown and colleagues (31-33) isolated GIP from impure
`preparations of cholecystokinin (CCK) that contained acid-
`inhibitory activity using the canine Heidenhain pouch as a
`bioassay. GIP was shown to be a potent inhibitor of gastric
`acid and pepsin secretion and was thus originally named
`‘gastric inhibitory polypeptide‘ (34, 35). Earlier, Dupre´ and
`Beck (26) had demonstrated that a crude preparation of CCK
`also possessed insulinotropic activity. In 1972, Rabinovitch
`and Dupre´ (36) found that this insulinotropic action could be
`removed by further purification of the CCK. This observation
`resembled the loss of the acid-inhibitory activity reported
`previously by Brown and Pederson (33) during the purifi-
`cation of GIP from CCK and led Dupre´ et al. (37) to the
`hypothesis that GIP may also possess insulin-releasing ca-
`pabilities. In 1973, Dupre´ et al. (37) demonstrated that a
`purified preparation of porcine GIP infused intravenously in
`humans in concert with glucose stimulated the release of
`significantly greater quantities of immunoreactive insulin
`than when the same dose of glucose was administered alone.
`The insulin response was sustained for the duration of the
`GIP infusion and was not observed during the euglycemic
`state (37). The glucose-dependent nature for the insulino-
`tropic activity of GIP was later demonstrated in vivo in dogs
`(38) and humans (39) and in the perfused rat pancreas (40).
`Furthermore, GIP released in response to the oral ingestion
`
`FIG. 1. Demonstration of the incretin
`concept. Blood glucose and insulin re-
`sponses after either intravenous or in-
`trajejunal glucose infusion in normal
`subjects. Although plasma glucose lev-
`els after intravenous glucose infusion
`were higher than those after intrajeju-
`nal glucose infusion, the latter gener-
`ated a larger insulin response. Based on
`these results, McIntrye et al. (23) sug-
`gested that a humoral substance was
`released from the jejunum during glu-
`cose absorption, acting in concert with
`glucose to stimulate insulin release
`from pancreatic b-cells. [Reproduced
`with permission from N. McIntyre et al.:
`Lancet 2:20-21, 1964 (23) © The Lancet
`Ltd.].
`
`The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 12 February 2015. at 12:28 For personal use only. No other uses without permission. . All rights reserved.
`
`2
`
`

`

`878
`
`KIEFFER AND HABENER
`
`Vol. 20, No. 6
`
`FIG. 2. The enteroinsular axis. After
`ingestion of nutrients, hormone secre-
`tion from different cell types of the pan-
`creatic islets [A (a), B (b), D (d), PP] may
`be modified by one or more modalities
`of: I, endocrine transmission; II, neuro-
`transmission; and III, direct substrate
`stimulation. [Reproduced with permis-
`sion from W. Creutzfeldt: Diabetologia
`16:75-85, 1979 (29) © Springer-Verlag].
`
`of fat yielded no increase in plasma insulin levels unless
`intravenous glucose was also administered (41-43). The glu-
`cose dependency of GIP-stimulated insulin secretion ap-
`peared to provide an important safeguard against inappro-
`priate stimulation of insulin release during a high-fat, low-
`carbohydrate meal. The recognition of
`this additional
`important physiological function of GIP led to the alternate
`designation glucose-dependent insulinotropic polypeptide
`(GIP) (44).
`In accordance with the roles of GIP as an enterogastrone and
`an incretin, immunoreactive GIP cells have been located in the
`upper small intestine of ruminants (45), humans, pigs, dogs
`(46), and rats (47). In the gastrointestinal tract of dog and man,
`immunoreactive GIP is present in cells predominantly in the
`midzone of the duodenal villi and, to a lesser extent, in the
`jejunum (48). Levels of GIP rise several fold shortly after in-
`gestion of a meal containing fat (41-43) or glucose (38, 49, 50).
`It appears glucose may act directly at the level of the GIP-
`secreting K cells to stimulate GIP release (51, 52).
`Studies employing GIP antisera to immunoneutralize en-
`dogenous GIP indicated that intestinal hormones other than
`GIP contribute substantially to the incretin effect (53, 54).
`These findings were supported by the observation that in-
`sulinotropic activity remained in intestinal extracts after re-
`moval of GIP by immunoadsorption (55). Finally, a major
`contribution to the incretin effect from the lower gastroin-
`testinal tract was shown in studies of patients after varying
`degrees of resection of the small intestine (56). The incretin
`effect of oral glucose correlated positively to the total length
`of residual small bowel rather than to an integrated release
`of GIP. Patients with preserved ileal residues had much
`larger incretin effects than patients with no ileal residues,
`despite equal integrated increases in plasma GIP, findings
`indicative of the presence of incretins other than GIP in the
`ileum (56).
`
`III. Discovery of GLP-1
`
`In the interim between the discovery in the 1970s of GIP
`as an important intestinal incretin hormone to the actual
`discovery of GLP-1, it was suspected that there must be a
`second incretin hormone in addition to GIP (54-56). The
`ushering in of the era of recombinant DNA technology in the
`late 1970s provided the means necessary for the identifica-
`tion of the ‘missing‘ incretin hormone. In the early 1980s, the
`cloning of cDNAs encoding the preproglucagons from pan-
`creata of the anglerfish was accomplished (57, 58). The an-
`glerfish was found to have two separate nonallelic prepro-
`glucagon genes, I and II, both encoding a glucagon and a
`glucagon-related peptide (GRP) (58). Notably, the glucagon-
`related peptide encoded in the anglerfish, preproglucagon-I,
`located carboxy proximal to the sequence of glucagon, bore
`a strong homology to the sequence of GIP, leading Lund et
`al. (57) to suggest that the anglerfish GRP-1 may be an in-
`testinal incretin hormone. In support of this supposition
`Lund et al. (59) showed that similar preproglucagon mRNAs
`were expressed in the anglerfish pancreas and intestine, a
`finding that strongly supported the prediction that GRP
`could be an incretin hormone. Subsequently, preprogluca-
`gon mRNAs were cloned from human (60) and rat (61) gut
`and shown to be identical in sequences to the mRNAs in
`pancreas.
`Shortly after the discovery of anglerfish GRP, the prepro-
`glucagon cDNAs of mammals were cloned (62-64) as well as
`the human gene (65). It became clear that the anglerfish
`GRP-I is a homolog of the GLP-1s encoded in the mammalian
`preproglucagons, which were subsequently proven to be
`potent insulinotropic incretins. There was, however, some
`uncertainty regarding the identification of the bioactive iso-
`form of GLP-1 that had true insulinotropic actions. Based on
`the amino acid sequence of the mammalian preprogluca-
`gons, the sites that would be predicted for posttranslational
`
`The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 12 February 2015. at 12:28 For personal use only. No other uses without permission. . All rights reserved.
`
`3
`
`

`

`December, 1999
`
`THE GLUCAGON-LIKE PEPTIDES
`
`879
`
`processing into peptide hormones were somewhat ambigu-
`ous. At the time it was generally believed that the yet-to-
`be-identified prohormone convertases (PCs) that enzymati-
`cally split prohormones into bioactive peptides required two
`adjacent basic amino acids, combinations of arginine, and
`lysine. The GLP-1 sequence begins with a histidine as the
`amino-terminal residue, as do most of the peptide hormones
`in the glucagon-related superfamily of hormones (Fig. 3). In
`the preproglucagon sequence, the first histidine is preceded
`by two basic amino acids, Lys-Arg, followed by four resi-
`dues, another single basic residue, arginine, and a second
`histidine. The thinking at that time was that the putative
`bioactive peptide that would theoretically be cleaved from
`the preproglucagon during posttranslational processing
`would be at the Lys-Arg yielding a peptide of 37 or 36 amino
`acids, depending on whether the C-terminal glycine was
`present or absent and whether the penultimate C-terminal
`arginine was amidated in the absence of the C-terminal gly-
`cine. Thus, the 1-37 and 1-36 GLP-1 peptide isoforms were
`the first to be synthesized and tested for biological activity.
`The results of the experimental testing were disappointing.
`One report questioned whether GLP-1 had any relevant ac-
`tivity: ‘How glucagon-like is glucagon-like peptide?‘ (66), as
`it had no effect on plasma glucose and insulin levels when
`administered to rabbits. Another report showed a weak stim-
`ulation of insulin secretion in cultured rat pancreatic islets at
`superpharmacological doses (25 nm) of GLP-1(1-36)amide
`and suggested that an N-terminally truncated peptide, GLP-
`1(7-36)amide, may be more active (67), as was suggested
`for GLP-1(7-37), an N-terminally truncated form of GLP-1(1-
`37) (67). These ideas were based upon alignment of the se-
`quence of GLP-1 with the other members of the glucagon
`superfamily of peptide hormones (see Fig. 3), which revealed
`that the best alignment was with the histidine at position 7,
`and not position 1 of GLP-1 (12, 63, 67). In 1986 it was
`discovered that GLP-1 was indeed further N-terminally trun-
`cated by posttranslational processing in the intestinal L cells
`(68, 69). In contrast to GLP-1(1-37), GLP-1(7-37) and (7-
`36)amide were found to be potent insulinotropic hormones
`in the isolated perfused pancreas of rats (70) and pigs (71),
`
`and in humans (72). Further, it was suggested that the weak
`insulinotropic actions of GLP-1(1-37) at micromolar concen-
`trations were probably artefactual due to a 0.1% level of
`nonspecific cleavage of GLP-1(1-37) to GLP-1(7-37) by non-
`specific cathepsins in the serum-implemented tissue culture
`media (73). At present it is well established that the GLP-1
`isoforms GLP-1(7-37) and GLP-1(7-36)amide are the bioac-
`tive insulinotropic peptides derived from preproglucagon in
`the intestine and the hind brain. The functions of the lesser
`GLP-1 isoforms GLP-1(1-37) and GLP-1(1-36)amide remain
`unknown.
`
`IV. Structures of GLPs and Family of Glucagon-
`Related Peptides
`
`The GLPs belong to a larger family referred to as the
`glucagon superfamily of peptide hormones. These hormones
`are classified within this family based on their considerable
`sequence homology, having anywhere from 21% to 48%
`amino acid identity with glucagon. Included in this family
`are: glucagon, GLP-1(7-37) and -(7-36)amide, GIP, exendin-3
`and -4, secretin, peptide histidine-methionine amide (PHM),
`GLP-2, helospectin-1 and -2, helodermin, pituitary adenyl
`cyclase-activating polypeptides
`(PACAP)-38, and -27,
`PACAP-related peptide (PRP), GH-releasing factor (GRF),
`and vasoactive intestinal polypeptide (VIP) (Fig. 3). These
`peptide hormones are produced in the gut, pancreas, and the
`central and peripheral nervous systems and exhibit a wide
`variety of biological actions in which several act as neuro-
`transmitters. Notably, even peptide hormones that are co-
`encoded within the same precursor, such as the peptide
`hormones derived from the cleavages of preproglucagon,
`differ significantly in the physiological processes that they
`regulate. For example, the major function of glucagon is to
`maintain blood glucose levels during fasting, whereas GLP-1
`functions primarily during feeding to stimulate insulin re-
`lease and to lower blood glucose levels. On the other hand,
`GLP-2 appears to regulate the growth of intestinal epithelial
`cells.
`
`FIG. 3. Amino acid sequences of the members of the superfamily of glucagon-related peptides. Sequences include human glucagon, human GLPs,
`human GIP, exendins (Heloderma horridum), human secretin, human peptide histidine methionine (PHM), helospectins (Heloderma horridum),
`helodermin (Heloderma suspectum), human PACAP, human PACAP-related peptide (PRP), human GRF, and human VIP. Residues identical
`to those of glucagon in the same position are shaded. Standard single letter abbreviations are used for amino acids (IUPAC-IUB Commission
`on Biochemical Nomenclature): A, Ala; C, Cys; D, Asp; E, Glu; F, Phe; G, Gly; H, His; I, Ile; K, Lys; L, Leu; M, Met; P, Pro; Q, Gln; R, Arg; S,
`Ser; T, Thr; V, Val; W, Trp; and Y, Tyr.
`
`The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 12 February 2015. at 12:28 For personal use only. No other uses without permission. . All rights reserved.
`
`4
`
`

`

`880
`
`KIEFFER AND HABENER
`
`Vol. 20, No. 6
`
`Exendin-3, exendin-4, helospectin-1, helospectin-2, and
`helodermin were all isolated from lizard (Heloderma) venom.
`They are potent secretagogues of the exocrine pancreas (74).
`Helodermin shares 53% and 42% homology with human
`PACAP and VIP, respectively, and has high affinity for the
`VIP2 receptor (75). Exendin-4 is 53% homologous to mam-
`malian GLP-1 and acts as a high-affinity agonist on the GLP-1
`receptor (76, 77). Exendin-4 and GLP-1 may interact with
`specific receptors yet to be identified in guinea pig pancreatic
`acini tissue (78). In contrast, the amino-terminally truncated
`form of exendin-3(9-39) is a potent antagonist of GLP-1 ac-
`tions (76, 77). Lizard helodermin, exendin, VIP/PHI,
`PACAP, and glucagon/GLP-1 cDNAs have been cloned,
`revealing that separate genes exist for these peptides (79, 80).
`To date, no evidence has been uncovered for the existence of
`mammalian homologues of the lizard helodermin or exendin
`(79, 80). It appears that helodermin and exendin-4 are not the
`evolutionary precursors to mammalian PACAP/VIP or
`GLP-1 but represent a distinct family of peptides. It seems
`likely that the high-affinity and biological activities of helo-
`dermin and exendin-4 on the mammalian VIP2 and GLP-1
`receptors, respectively, are a result of convergent evolution
`(80).
`It is proposed that the proglucagon gene arose by the
`duplication of an ancestral gene approximately 800-1,000
`million years ago (81). The structural organization of the
`genes of the glucagon superfamily of peptide hormones sug-
`gests that the ancestral gene consisted of four exons, which
`encoded the 59-untranslated region of the mRNA, the signal
`peptide, the hormone and the 39-untranslated region of the
`mRNA, respectively (82). The glucagon superfamily of hor-
`mones may have arisen by duplication and amplification of
`this basic gene, followed by a further duplication and am-
`plification of the exon encoding the glucagon hormone do-
`main to generate the multiple GLPs observed in preproglu-
`cagon (82). Based on statistical analysis of DNA sequences of
`the preproglucagon genes from bovine, human, hamster, and
`anglerfish, Lopez et al. (83) postulated that the two anglerfish
`genes arose from gene duplication approximately 160 million
`years ago (83). Furthermore, this analysis suggested that the
`GLP-2 sequence originated by duplication of the glucagon or
`GLP-1 sequence before the earliest divergence of fish (83).
`However, until recently, it was believed that GLP-2 was not
`expressed in either fish or birds (11, 84). Irwin and Wong (85)
`discovered that, unlike pancreatic proglucagon of fish and
`birds, the intestinal proglucagon does contain the sequence
`of GLP-2. Therefore, fish and bird proglucagon mRNAs from
`pancreas and intestine have different 39-ends that are due to
`alternative mRNA splicing (85). The recent cloning of the
`frog (Xenopus) proglucagon cDNAs revealed the presence of
`three distinct GLP-1 peptides in addition to glucagon and
`GLP-2 (86). It has been postulated that the first exon dupli-
`cation event resulting in the appearance of glucagon and GLP
`occurred at least 405-800 million years ago (81, 83). A du-
`plication of the GLP-containing exon, giving rise to GLP-1
`and GLP-2, may have occurred between 365 (divergence of
`mammals and amphibians) and 405 (divergence of cartilag-
`inous fish and tetrapods) million years ago (81). The amino
`acid sequences of the preproglucagon genes are highly con-
`served among mammals (Fig. 4), and the products derived
`
`from proglucagon, glucagon (Fig. 5), and GLP-1 (Fig. 6) are
`highly conserved throughout the evolution of animal spe-
`cies. The amino acid sequence of glucagon is highly con-
`served during the evolution of tetrapods (3 substitutions
`between salamander and human), even more than the se-
`quences of either GLP-1 (7 substitutions) or GLP-2 (15 sub-
`stitutions). The high degree of conservation of the glucagon
`and GLP sequences during evolution indicates the impor-
`tance of the physiological processes regulated by these hor-
`mones.
`The conservation of GLP-1 also reflects the fact that es-
`sentially the entire amino acid sequence of GLP-1 is required
`for full biological activity. Removal of the N-terminal histi-
`dine (5 GLP-1 8-37) results in drastic loss of receptor binding
`and insulinotropic activity (87-90). The positive charge of the
`imidazole side chain of the histidine residue appears to be
`crucial for GLP-1 actions (91). Likewise, N-terminal trunca-
`tion of this histidine from the related insulinotropic peptide
`exendin-4(1-39) reduces agonist activity by approximately
`10-fold (92). Notably, N-terminal truncation of exendin-4 by
`two residues yields a peptide that binds with the same af-
`finity as full-length exendin but antagonizes GLP-1 action
`(92). In contrast, an N-terminal truncation of GLP-1 by two
`residues reduces binding affinity to approximately 1% that
`of the intact molecule (92, 93). Also, addition of an amino acid
`to the N terminus of GLP-1(6-37) also reduces biological
`activity (87, 89). Truncation at the C terminus also reduces the
`biological activity of GLP-1 considerably (87, 88, 90, 93).
`Substitution in the N-terminal part of the GLP-1 molecule
`with the corresponding glucagon residues impaired the af-
`finity for the GLP-1 receptor only moderately whereas ex-
`changes in the C-terminal portion of GLP-1 decreased the
`affinity for the GLP-1 receptor more than 100-fold (94). In
`contrast, the binding affinity of GLP-1 to its receptor is more
`sensitive to GIP-like changes in the N-terminal region than
`to changes in the C-terminal region (95).
`Another approach to understanding the structure-activity
`relationships of GLP-1 has been obtained from studies of
`peptide analogs in which individual amino acids are sub-
`stituted. These studies revealed that the residues in positions
`1 (His), 4 (Gly), 6 (Phe), 7 (Thr), 9 (Asp) 22 (Phe), and 23 (Ile)
`are important for the binding affinity and biological activity
`of GLP-1 (96-98). Two-dimensional nuclear magnetic reso-
`nance of GLP-1 in a membrane-like environment (a dode-
`cylphosphocholine micelle) revealed that GLP-1 consists of
`an N-terminal random coil segment (residues 1-7), two he-
`lical segments (7-14 and 18-29), and a linker region (15-17) –
`a structure similar to that observed for glucagon (99). Thus
`far, attempts to generate smaller active fragments of GLP-1
`that retain potent insulinotropic activity have failed (89, 98,
`100).
`
`V. Tissue Distribution of the Expression of GLPs
`
`A. Pancreatic a-cells
`
`Pancreatic a-cells were discovered in 1907 as histolog-
`ically distinct cells from the b-cells of the islets of Lang-
`erhans (101). It was not until 1962 that a-cells were shown
`by immunofluorescence staining studies to be the source
`
`The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 12 February 2015. at 12:28 For personal use only. No other uses without permission. . All rights reserved.
`
`5
`
`

`

`December, 1999
`
`THE GLUCAGON-LIKE PEPTIDES
`
`881
`
`FIG. 4. Amino acid sequences of proglucagon from seven mammalian species. GenBank accession numbers are given in parentheses. Major
`proglucagon products are indicated by bars; GRPP, glicentin-related pancreatic peptide; IP-1 and IP-2, intervening peptides; GLP-1 and GLP-2,
`GLPs. Shaded residues are completely conserved between the seven species. Standard single letter abbreviations are used for amino acids
`(IUPAC-IUB Commission on Biochemical Nomenclature): A, Ala; C, Cys; D, Asp; E, Glu; F, Phe; G, Gly; H, His; I, Ile; K, Lys; L, Leu; M, Met;
`P, Pro; Q, Gln; R, Arg; S, Ser; T, Thr; V, Val; W, Trp; and Y, Tyr.
`
`of glucagon (102). The a-cells are one of four distinct
`polypeptide-secreting islet cell types: glucagon-secreting
`a-cells, insulin- and amylin-secreting b-cells, somatosta-
`tin-secreting d-cells, and pancreatic-polypeptide-secreting
`F cells. These cells are arranged in highly organized pat-
`terns within the islets. In rodents, a-cells and d-cells exist
`on the surface or mantle of the islet surrounding the core
`of b-cells, although the patterns of distribution of the a-,
`d-, and b-cells differ among animal species (103-105). Un-
`certainties in the islet vascular architecture and the direc-
`tion of blood flow within the islets (106) still cloud the
`functional significance of the anatomic arrangement of
`hormone-producing islet cells.
`mRNA encoding proglucagon can be detected by PCR
`early in the wall of the embryonic foregut at the 20-somite
`stage and is restricted to the area of the duodenum from
`which the pancreas will develop 10-12 h later (107). The
`pancreas arises as two outbuddings of the gut tube shortly
`after it is formed early in development at embryonic day 9
`(ED 9) in the mouse (for review see Ref. 108). By ED 10, the
`budding anlaga fuse to become the dorsal and ventral pan-
`creas with their respective ducts. By immunofluorescence,
`glucagon-positive cells are identifiable at ED 10.5 in the dor-
`sal bud and at ED 11.5 in the ventral bud (109). Individual
`hormone-containing cells are located within the epithelium
`of pancreatic ducts, and clusters of endocrine cells are found
`in the pancreatic interstitium. Starting on ED 16.5, islets begin
`to form, and by day 18.5 the islets consist of centrally located
`b-cells with the adult ‘one cell-one hormone‘ phenotype
`(109). The expression of specific transcription factors is in-
`volved in the determination of cell lineages that determine
`the development of islet-specific cells of the endocrine pan-
`creas. For example, recent findings indicate that disruption
`
`of the gene for the transcription factor Pax6 in mice results
`in a near-complete failure in the development of a-cells (110),
`whereas disruption of Pax4 results in the absence of mature
`b- and d-cells (111) (Fig. 7). Mice lacking the transcription
`factor Nkx2.2 have diabetes due to arrested differentiation of
`pancreatic b-cells (112).
`As illustrated in Fig. 8, cell-specific processing of proglu-
`cagon in pancreatic a-cells leads primarily to the production
`of glucagon. However, immunoreactive GLP-1 is detectable
`in rat pancreatic a-cells by immunocytochemistry (113).
`Fully processed GLP-1 (7-36 amide and 7-37) is also visual-
`ized in pancreatic rat extracts by using chromatographic
`techniques and RIAs (114, 115). A recent investigation de-
`tected predominantly GLP-1 (1-36) amide in extracts of rat
`pancreas (116). Using similar techniques, small amounts of
`N-terminally extended GLP-1 (1-36 amide and 1-37) are also
`found in extracts from porcine and human pancreas (117,
`118). In addition, immunoreactive GLP-1 is secreted from the
`arginine-perfused rat pancreas and glucose-stimulated iso-
`lated rat islets, as detected by RIA (113, 114). The relatively
`small quantity of GLP-1 produced by the pancreas might
`have important local actions within the islets.
`
`B. Intestinal L cells
`
`Intestinal cells are reported to react with glucagon-specific
`C-terminal antisera, although the intestinal immunodeter-
`minant responsible for the immune reaction appears to differ
`chemically from pancreatic glucagon (152, 153). Antibodies
`directed against the midpart of glucagon and antibodies
`against the nonglucagon part of the glicentin molecule reveal
`a large population of endocrine cells in the small and large
`bowel that express proglucagon and its fragments (154-156).
`
`The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 12 February 2015. at 12:28 For personal use only. No other uses without permission. . All rights reserved.
`
`6
`
`

`

`882
`
`KIEFFER AND HABENER
`
`Vol. 20, No. 6
`
`FIG. 5. Amino acid sequences of vertebrate glucagons. Classes are as indicated and residues identical to those of human glucagon in the same
`position are shaded. Standard single letter abbreviations are used for amino acids (IUPAC-IUB Commission on Biochemical Nomenclature):
`A, Ala; C, Cys; D, Asp; E, Glu; F, Phe; G, Gly; H, His; I, Ile; K, Lys; L, Leu; M, Met; P, Pro; Q, Gln, R, Arg; S, Ser; T, Thr; V, Val; W, Trp; and
`Y, Tyr. Reference numbers indicate the source of the corresponding sequence.
`
`In contrast to the pancreas where GLPs represent minor
`products, GLPs are fully processed in abundance in the in-
`testine, representing the major source of circulating GLPs
`(115-117). By virtue of their ultrastructure as assessed by
`electron microscopy, these cells are designated as L cells that
`clearly differ from pancreatic a-cells in the morphology of the
`granules (154, 156-158). The intestinal L cells are flask shaped
`and open-type, the microvilli reach the intestinal lumen, and
`a domain rich in endocrine granules exists near the basal
`lamina (159, 160) (Fig. 9). The shape of the L cells suggests
`that the cells can respond to changes in the environment
`within the intestinal lumen, resulting in a basal discharge of
`their granular contents.
`The L cells are the second most abundant population of
`endocrine cells in the human intestine, exceeded only by the
`population of enterochromaffin cells. A high abundance of L
`cells is present in the distal jejunum and ileum, and an in-
`creasing abundance of L cells is demonstrable along the
`colon, with the highest concentration in the rectum (160-163).
`L cells first appear in human fetuses at the 8th week of
`gestation in the ileum, the 10th