Diabetologia (1998) 41: 271±278
`
`Ó Springer-Verlag 1998
`
`Dipeptidyl peptidase IV resistant analogues of glucagon-like
`peptide-1 which have extended metabolic stability and improved
`biological activity
`
`C. F. Deacon1, L. B. Knudsen2, K. Madsen2, F. C. Wiberg2, O. Jacobsen1, J. J. Holst1
`
`1 Department of Medical Physiology, The Panum Institute, University of Copenhagen, Denmark
`2 Novo Nordisk A/S, MaÊ lùv, Denmark
`
`Summary Glucagon-like peptide 1 (GLP-1) has great
`potential in diabetes therapy due to its glucose-de-
`pendent stimulation of insulin secretion, but this is
`limited by its rapid degradation, primarily by dipepti-
`dyl peptidase IV. Four analogues, N-terminally sub-
`stituted with threonine, glycine, serine or a-ami-
`noisobutyric acid, were synthesised and tested for
`metabolic stability. All were more resistant to dipep-
`tidyl peptidase IV in porcine plasma in vitro, ranging
`from a t1/2 of 159 min (Gly8 analogue) to undetectable
`degradation after 6 h (Aib8 analogue; t1/2 for GLP-1
`(7±36) amide, 28 min). During i. v. infusion in anaes-
`thetised pigs, over 50 % of each analogue remained
`undegraded compared to 22.7 % for GLP-1 (7±
`36) amide. In vivo, analogues had longer N-terminal
`t1/2 (intact peptides: means, 3.3±3.9 min) than GLP-1
`(7±36) amide (0.9 min; p < 0.01), but these did not ex-
`ceed the C-terminal t1/2 (intact plus metabolite: ana-
`logues, 3.5±4.4 min; GLP-1 (7±36) amide, 4.1 min).
`Analogues were assessed for receptor binding using
`
`a cell line expressing the cloned receptor, and for
`ability to stimulate insulin or inhibit glucagon secre-
`tion from the isolated perfused porcine pancreas. All
`bound to the receptor, but only the Aib8 and Gly8 an-
`alogues had similar affinities to GLP-1 (7±36) amide
`(IC50; Aib8 = 0.45 nmol/l; Gly8 = 2.8 nmol/l; GLP-1
`(7±36) amide = 0.78 nmol/l). All analogues were ac-
`tive in the isolated pancreas, with the potency order
`(Aib8 > Gly8 >
`reflecting
`receptor
`affinities
`Ser8 > Thr8). N-terminal modification of GLP-1 con-
`fers resistance to dipeptidyl peptidase IV degrada-
`tion. Such analogues are biologically active and have
`prolonged metabolic stability in vivo, which, if associ-
`ated with greater potency and duration of action, may
`help to realise the potential of GLP-1 in diabetes
`therapy. [Diabetologia (1998) 41: 271±278]
`
`Keywords Glucagon-like peptide-1, analogue, dipep-
`tidyl peptidase IV, non-insulin-dependent diabetes
`mellitus, therapy.
`
`The insulinotropic hormone glucagon-like peptide-1
`(GLP-1) is the product of tissue-specific post-transla-
`tional processing of the glucagon precursor, progluca-
`
`Received: 9 September 1997 and in revised form: 20 October
`1997
`
`Corresponding author: Dr. C. F. Deacon, Department of Medi-
`cal Physiology, The Panum Institute, Blegdamsvej 3, DK-2200
`Copenhagen N, Denmark
`Abbreviations: GLP-1, Glucagon-like peptide-1; DPP IV,
`dipeptidyl peptidase IV; PG, proglucagon; GRF, growth hor-
`mone-releasing factor; TFA, trifluoroacetic acid; ANOVA,
`analysis of variance; NIDDM, non-insulin-dependent diabetes
`mellitus; Aib, a-aminoisobutyric acid; HSA, human serum al-
`bumin; BHK, baby hamster kidney.
`
`gon (PG) in the L-cells of the gastrointestinal mucosa
`[1, 2]. This results in the formation of GLP-1 (7±
`36) amide (corresponding to PG (78±107) amide),
`which is the predominant form in humans, although
`small amounts of non-amidated glycine-extended
`GLP-1 (7±37) are also produced [3]. GLP-1 is one of
`the most potent insulin secretagogues identified [4]
`and this, together with the glucose-dependency of its
`actions [5±7], has focussed interest on its role as a reg-
`ulator of blood glucose and its potential as a thera-
`peutic agent in the treatment of non-insulin-depen-
`dent diabetes mellitus (NIDDM) [6, 8±11].
`Recent studies have shown that GLP-1 itself is the
`subject of further enzyme cleavage. In particular,
`dipeptidyl peptidase IV (DPP IV; EC 3.4.14.5) is im-
`
`MPI EXHIBIT 1072 PAGE 1
`
`MPI EXHIBIT 1072 PAGE 1
`
`

`

`272
`
`C. F. Deacon et al.: Dipeptidyl peptidase IV resistant GLP-1 analogues
`
`portant, resulting in a metabolite which is N-termi-
`nally truncated by 2 amino acids [12±14]. The result-
`ing peptide, GLP-1 (9±36) amide, is an endogenous
`metabolite [13] which is an antagonist in vitro [15,
`16]. Moreover, exogenously administered GLP-1 is
`also rapidly degraded in both diabetic and non-dia-
`betic subjects, with GLP-1 (9±36) amide being the
`major metabolite [17]. In a strain of rats lacking
`DPP IV, this metabolite is not formed [14]. DPP IV
`is highly specific and has strict substrate requirements
`[18, 19], raising the possibility of developing ana-
`logues which the enzyme is unable to cleave. Studies
`with another peptide substrate of DPP IV, growth
`hormone-releasing factor (GRF), have shown that
`analogues with N-terminal amino acid substitutions
`have some resistance to the enzyme's action [20].
`The present study was undertaken to examine wheth-
`er small modifications of the N-terminus of GLP-1
`would also confer resistance to degradation by DPP
`IV, while retaining the peptide's biological activity.
`
`Materials and methods
`
`Peptide synthesis. Peptides were synthesised on an Applied
`Biosystems 431A peptide synthesiser (Foster City, Calif.,
`USA), according to the Fmoc strategy, using the following pro-
`tected amino acid derivatives: Fmoc-Arg(Pmc), Fmoc-
`Trp(Boc), Fmoc-Glu(OBut), Fmoc-Lys(Boc), Fmoc-Gln(Trt),
`Fmoc-Tyr(But), Fmoc-Ser(But), Fmoc-Thr(But), Fmoc-
`His(Trt)
`and
`Fmoc-Asp(OBut)
`(Calbiochem-Novabio-
`chem AG, LaÈ ufelingen, Switzerland). The peptides were de-
`protected and cleaved from the resin by treatment with trifluo-
`roacetic acid (TFA)/phenol/thioanisole/water/ethanedithiol
`(83.25:6.25:4.25:2.00) for 180 min. After evaporation of the
`TFA, the crude peptide was precipitated with diethyl ether
`and purified by semipreparative HPLC on a C18 reversed-
`phase column eluted with a gradient of acetonitrile in
`0.05 mol/l (NH4)2SO4, pH 2.5. Peptide-containing fractions
`were applied to a Sep Pak C18 cartridge (Waters-Millipore, Mil-
`ford, Mass., USA), eluted with 70 % acetonitrile/0.1 % TFA
`and lyophilised. The final products were characterised by ami-
`no acid analysis, analytical reversed-phase HPLC and by plas-
`ma desorption mass spectrometry. Purity was more than 95 %
`by HPLC with detection at 214 nm. Analogues, substituted at
`position 8 of GLP-1 with either threonine (Thr8-GLP-1 (7±
`37)), glycine (Gly8-GLP-1 (7±37)), serine (Ser8-GLP-1 (7±
`36) amide) or a-aminoisobutyric acid (Aib8-GLP-1 (7±
`36) amide and Aib8-GLP-1 (7±37)) were prepared.
`
`Peptide stability in porcine plasma in vitro. The stability of each
`peptide in porcine plasma was determined by incubation at
`37 °C with 300 pmol/l of GLP-1 (7±36) amide or each analogue
`for up to 6 h. This was followed by reversed-phase HPLC and
`RIA according to a previously published method [13], using
`antiserum 2135 as described below.
`
`Peptide pharmacokinetics in vivo. Danish LYY strain pigs (33±
`40 kg) were used. Food was withdrawn 24 h before surgery, but
`animals had free access to drinking water. After premedication
`with ketamine chloride (Ketaminol, 10 mg/kg; Veterinaria AG,
`Zurich, Switzerland), animals were anaesthetised with 1 % ha-
`lothane (Halocarbon Laboratories, River Edge, NJ, USA), and
`
`anaesthesia was maintained with intermittent positive pressure
`ventilation using an anaesthesia ventilator in a semi-open sys-
`tem. Catheters were placed in the right carotid artery for sam-
`pling of arterial blood, and into a vein of the left ear for peptide
`infusion. After surgical preparation, animals were heparinised,
`an infusion (0.9 % NaCl) was set up and given via the ear vein
`catheter (5 ml/min), and the animals were left undisturbed for
`30 min.
`Four groups of four animals were used. Each group receiv-
`ed separate i. v. infusions of GLP-1 (7±36) amide and one ana-
`logue in a cross-over design with 80 min between each infu-
`sion. Synthetic GLP-1 (7±36) amide or analogues were dis-
`solved in saline containing 1 % human serum albumin (HSA;
`Behringwerke, Marburg, Germany), and infused at a rate of
`5 pmol × kg-1 × min-1 for 30 min using a syringe pump. Arterial
`blood samples (4 ml) were taken at 0, 5, 10, 15, 20, 22, 25, 27
`and 30 min from the start of the infusion. After 30 min, the in-
`fusion was stopped, and further blood samples were taken at
`1, 2, 4, 6, 10, 15, 20, 30, 40, 50 and 65 min. The amount of blood
`taken over the entire procedure was 160 ml, which, for a 40 kg
`pig, is equivalent to 4 % of the total blood volume.
`Blood samples were collected into chilled tubes containing
`EDTA (7.4 mmol/l final concentration; Merck, Darmstadt,
`Germany), aprotinin (500 kallikrein inhibitory equivalents/ml
`blood; Novo Nordisk, Bagsvñrd, Denmark) and diprotin A
`(0.1 mmol/l final concentration; Bachem, Bubendorf, Switzer-
`land), and kept on ice until centrifugation at 4 °C. Plasma was
`separated and stored at -20 °C until analysis with the RIAs de-
`scribed below. In addition, plasma collected during mins 22±30
`of each infusion was separately pooled for each group of four
`animals, extracted on Sep Pak C18 cartridges and analysed by
`reversed-phase HPLC and RIA as before [13], using antiserum
`2135, described below.
`
`Expression of the cloned human GLP-1 receptor. The human
`GLP-1 receptor cDNA was obtained from Dr. B. Thorens.
`The cDNA was subcloned into the pcDNA 1 vector (Invitro-
`gen Corporation, San Diego, Calif., USA) using the Hind III
`± EcoRI sites, and was then called pAH 260. Baby hamster kid-
`ney (BHK) cells were co-transfected with 20 mg pAH 260 and
`0.6 mg pSV 2 neo vector [21] using the methods described by
`Chen and Okayama [22], and grown in Dulbecco's modified
`Eagle's medium, 10 % fetal calf serum, 100 IU penicillin,
`100 mg/ml streptomycin and 1 mmol/l Na-pyruvate (all from
`Gibco, Life Technologies, Roskilde, Denmark). Stable clones
`were selected in medium containing 1 mg/ml Geneticin G-418
`(Gibco) and maintained at 37 °C, in an atmosphere containing
`5 % CO2. Stable clones were screened in a receptor binding as-
`say, and those expressing high levels of GLP-1 receptor were
`then recloned and screened in an adenylate cyclase assay to
`find clones with functional receptors. These clones were select-
`ed for further studies.
`
`Receptor binding. Receptor binding was carried out as previ-
`ously described [16], using BHK cells expressing the human
`pancreatic GLP-1 receptor. In brief, plasma membranes were
`prepared by homogenisation with two 10 s bursts using a Poly-
`tron PT 10±35 homogeniser (Kinematica, Lucerne, Switzer-
`land),
`in a buffer consisting of 10 mmol/l Tris-HCl with
`30 mmol/l NaCl, pH 7.4, containing in addition, 1 mmol/l dithi-
`othreitol, 5 mg/l leupeptin, 5 mg/l pepstatin, 100 mg/l bacitra-
`cin (all from Sigma, St. Louis, Mo, USA) and 16 mg/l aprotinin,
`and centrifuged on top of a layer of 41 % (wt/vol) sucrose at
`95 000 ´ g for 75 min. The white band between the two layers
`was diluted in buffer and centrifuged at 40 000 ´ g for 45 min.
`The precipitate containing the plasma membranes was sus-
`pended in buffer, and stored at -80 °C until use.
`
`MPI EXHIBIT 1072 PAGE 2
`
`MPI EXHIBIT 1072 PAGE 2
`
`

`

`C. F. Deacon et al.: Dipeptidyl peptidase IV resistant GLP-1 analogues
`
`273
`
`The binding assay was performed in 96-well filter microtitre
`plates. The buffer used was 50 mmol/l HEPES, pH 7.4, with the
`addition of 2.5 % (wt/vol) HSA grade V (Sigma). Peptide, trac-
`er and plasma membranes were incubated for 30 min at 30 °C.
`The tracer was prepared by iodination of GLP-1 (7±36) amide
`using the lactoperoxidase method [23]. Purification by HPLC
`as previously described [24] yielded 125I-Tyr19-GLP-1 (7±
`36) amide with a specific activity of 80 kBq/pmol.
`
`Isolated perfused pancreas. Danish LYY strain pigs (14±
`16 kg) were fasted overnight and anaesthetised with chloral-
`ose (100 mg/kg; Merck). The pancreases were isolated as pre-
`viously described [25] and perfused in a single-pass system,
`using a gassed (5 % CO2 in O2) Krebs-Ringer-bicarbonate
`perfusion medium containing, in addition, 0.1 % HSA, 5 %
`dextran T 70 (Pharmacia, Uppsala, Sweden) and 5 mmol/l
`glucose. The venous effluent was collected for 1-min intervals,
`centrifuged at 4 °C, and stored at -20 °C until analysis. Syn-
`thetic GLP-1 (7±36) amide or peptide analogues were dis-
`solved in 0.04 mol/l phosphate buffer, pH 7.4, containing 1 %
`HSA, and infused into the arterial line using a syringe pump,
`to give final concentrations of 1 nmol/l in the perfusate. Pep-
`tides were infused for 10 min periods, separated by 10±
`15 min rest periods, during which time endocrine secretion re-
`turned to basal levels. Insulin and glucagon concentrations in
`the venous effluent were measured by RIA as described be-
`low.
`
`Hormonal analysis. HPLC fractions were analysed using anti-
`serum 2135 [26, 27], which is ªside-viewingº, and which rec-
`ognises all molecules containing the central sequence of
`GLP-1 regardless of C- or N-terminal truncations or exten-
`sions. It cross-reacts fully with GLP-1 (7±37), 79 % with
`GLP-1 (9±36) amide, and has a detection limit of 5 pmol/l.
`Plasma samples were assayed for GLP-1 immunoreactivity
`using RIAs which are specific for each terminus of the mole-
`cule. N-terminal immunoreactivity was measured using the
`newly described antiserum 93242 [28], which has a cross-reac-
`tivity of approximately 10 % with GLP-1 (1±36) amide, and
`less than 0.1 % with GLP-1 (8±36) amide and GLP-1 (9±
`36) amide, and has a detection limit of 5 pmol/l. C-terminal
`immunoreactivity was measured using antiserum 89390 or
`92071 as appropriate. Antiserum 89390 [3, 29] has an abso-
`lute requirement
`for the intact amidated C-terminus of
`GLP-1 (7±36) amide, and cross-reacts less than 0.01 % with
`C-terminally truncated fragments, and 83 % with GLP-1 (9±
`36) amide. The detection limit is 1 pmol/l. Antiserum 92071
`[3] is specific for the C-terminus of GLP-1 (7±37) and cross-
`reacts less than 0.1 % with GLP-1 (7±36) amide. It has a de-
`tection limit of 4 pmol/l. For all assays, the intra-assay coeffi-
`cient of variation was less than 6 %. GLP-1 (7±36) amide,
`GLP-1 (7±37) or appropriate analogue were used as standard,
`and 125I-labelled GLP-1 (7±36) amide or GLP-1 (7±37) were
`used as tracer. Separation of antibody-bound from free pep-
`tide was achieved using plasma-coated charcoal [26, 29]. Plas-
`ma samples were extracted with 70 % ethanol (vol/vol, final
`concentration) before assay, giving recoveries of 75 % [27].
`The cross-reactivity of each analogue was determined for
`each antiserum.
`Venous effluent from the perfused pancreas was assayed
`using antiserum 2004 for insulin [27], and antiserum 4305 for
`glucagon [30].
`
`Calculations. During the peptide infusions, stable arterial pep-
`tide levels were achieved after 20 min, so the plateau concen-
`tration was defined as the mean of the last four measurements
`during the infusion. The plasma t1/2 was calculated by ln-linear
`
`regression analysis of peptide concentrations in samples col-
`lected after termination of the infusion, after subtraction of en-
`dogenous arterial GLP-1 concentrations. Insulin and glucagon
`output from the perfused pancreas are expressed as percentag-
`es of basal secretion, which is defined as 100 %.
`
`Statistical analysis. Data are expressed as means ± SEM, and
`were analysed using GraphPAD InStat software, version 1.13
`(San Diego, Calif, USA). In vitro data were analysed using
`one-sample t-tests, analysis of variance (ANOVA) and 2-tailed
`t-tests with correction for multiple comparisons as appropriate.
`In vivo data were analysed using repeated measures ANOVA
`and 2-tailed t-tests for paired or unpaired data as appropriate.
`Values of p less than 0.05 were considered to be significant.
`
`Results
`
`Analogue cross-reactivity. All analogues cross-react-
`ed more than 80 % with the appropriate C-terminal
`RIA; however, the cross-reactivity with the N-termi-
`nally directed 93242 assay varied considerably (Aib8-
`GLP-1, < 5 %; Gly8-GLP, 17 %; Ser8-GLP-1, 21 %;
`Thr8-GLP-1, 34 %). In practice, this meant that C-ter-
`minal assays (with appropriate standard) could be
`used for each analogue (detection limit, 5 pmol/l),
`but N-terminal immunoreactivity, using the appropri-
`ate analogue as standard, could be measured for only
`three analogues (Gly8-GLP, Ser8-GLP-1 and Thr8-
`GLP-1; detection limit, 20 pmol/l).
`
`Peptide stability in porcine plasma in vitro. GLP-1 (7±
`36) amide was degraded by porcine plasma in vitro at
`37°C, with a t1/2 of 28.1 ± 1.2 min (n = 12). HPLC
`analysis revealed the time-dependent generation of
`a second peak corresponding to GLP-1 (9±36) amide,
`and no other immunoreactive peaks were detected.
`Incubation of the GLP-1 analogues revealed a signif-
`icantly (p < 0.0001) prolonged t1/2 compared to GLP-
`(Gly8-GLP-1, 159 ± 12 min,
`1 (7±36) amide itself
`n = 3; Ser8-GLP-1, 174 ± 12 min, n = 9; Thr8-GLP-1,
`197 ± 14 min, n = 3), and again, HPLC analysis
`showed the formation of only one other peak, in addi-
`tion to the intact peptide. Degradation of Aib8-GLP-
`1 (n = 9) was undetectable after 6 h.
`
`Peptide pharmacokinetics in vivo. Stable arterial pep-
`tide concentrations were reached after 20 min of infu-
`sion. In all groups, concentrations determined by the
`C-terminal RIA (which measures both intact and N-
`terminally degraded peptide) exceeded those deter-
`mined by the N-terminal assay (illustrated for Thr8
`analogue group in Figure 1), while the ratio of N-ter-
`minal to C-terminal immunoreactivity was greater
`for each analogue than for GLP-1 (7±36) amide (Ta-
`ble 1). HPLC analysis of plasma pooled from each
`group during each infusion revealed two immunore-
`active peaks, corresponding to the intact peptide and
`the N-terminally truncated metabolite.
`In each
`
`MPI EXHIBIT 1072 PAGE 3
`
`MPI EXHIBIT 1072 PAGE 3
`
`

`

`274
`
`C. F. Deacon et al.: Dipeptidyl peptidase IV resistant GLP-1 analogues
`
`Fig. 1. Plasma GLP-1 immunoreactivity in the carotid artery,
`measured with C-terminally ( T ) and N-terminally ( X ) di-
`rected RIAs. Anaesthetised pigs received separate i. v. infu-
`sions (5 pmol × kg-1 × min-1) of GLP-1 (7±36) amide and Thr8-
`GLP-1 (7±37) in a cross-over design with 80 min between
`each infusion. Data are mean ± SEM; n = 4. The horizontal ar-
`row indicates the period of the infusion
`
`group, a greater percentage of the analogue remained
`undegraded compared to GLP-1 (7±36) amide (Ta-
`ble 1).
`For all four groups, there was no significant differ-
`ence between the C-terminal t1/2 for GLP-1 (7±
`36) amide or the analogues, but the N-terminal t1/2
`for each analogue was significantly prolonged (Ta-
`ble 2). For all four groups, the N-terminal t1/2 was
`t1/2 for GLP-1 (7±
`shorter than the C-terminal
`36) amide, but there was no significant difference be-
`
`Fig. 2. Binding affinity of GLP-1 analogues using the cloned
`human GLP-1 receptor. Individual curves are from one repre-
`sentative experiment where the data points are expressed as
`the mean of duplicate samples
`
`tween N- and C-terminal t1/2 for the analogues (Ta-
`ble 2).
`
`Receptor binding. All the analogues bound to the
`cloned human pancreatic GLP-1 receptor (Fig. 2),
`but with widely differing binding affinities (IC50; Ta-
`ble 3). The Aib8 and Gly8 analogues had similar affin-
`ities compared to GLP-1 (7±36) amide, while the oth-
`er two analogues had lower receptor affinities than
`GLP-1 (7±36) amide.
`
`Isolated perfused pancreas. Perfusion of the pancreas
`with 1 nmol/l GLP-1 (7±36) amide or the analogues
`increased insulin and decreased glucagon release rel-
`ative to basal secretion (Fig. 3). Of the analogues,
`Aib8-GLP-1 (7±36) amide was at least as potent as
`GLP-1 (7±36) amide in stimulating insulin and inhib-
`iting glucagon secretion, and was
`significantly
`(p < 0.05) more potent than the Ser8 and Thr8 ana-
`
`Table 1. Plasma concentrations of GLP-1 peptides attained during infusion of GLP-1 (7±36) amide and N-terminally modified an-
`alogues calculated using N- and C-terminally directed RIAs
`
`Infusion
`
`Plateau concentrations (pmol/l)
`
`N-terminal
`76.0 ± 12.2a
`GLP-1 (7±36) amide
`437.8 ± 49.1a
`Thr8-GLP-1 (7±37)
`108.0 ± 8.6a
`GLP-1 (7±36) amide
`336.0 ± 10.4a
`Gly8-GLP-1 (7±37)
`104.7 ± 22.8a
`GLP-1 (7±36) amide
`331.0 ± 59.7b
`Ser8-GLP-1 (7±36) amide
`80.3 ± 7.4a
`GLP-1 (7±36) amide
`Aib8-GLP-1 (7±36) amide
`NDL
`Values are mean ± SEM; n = 4, except for the % intact peptide
`after HPLC, which was calculated after HPLC analysis of a
`single sample of plasma pooled from four animals.
`
`N-terminal/C-terminal
`immunoreactivity
`
`% intact peptide
`after HPLC
`
`C-terminal
`16.4 ± 1.6c
`484.3 ± 103.1
`24.9
`62.2 ± 0.6
`689.3 ± 84.4
`62.5
`22.4 ± 2.1c
`500.3 ± 67.7
`23.8
`53.8 ± 2.4
`627.0 ± 24.0
`ND
`26.4 ± 2.1c
`387.0 ± 59.5
`24.8
`55.7 ± 3.9
`629.7 ± 137.1
`52.2
`31.7 ± 2.1
`267.8 ± 22.6
`17.2
`530.3 ± 16.2
`67.1
`NDL
`a p < 0.05 vs C-terminal concentration; b NS, p > 0.05 vs C-ter-
`minal concentration; c p < 0.005 vs analogue; ND, not deter-
`mined; NDL, not determined due to lack of cross-reactivity
`
`MPI EXHIBIT 1072 PAGE 4
`
`MPI EXHIBIT 1072 PAGE 4
`
`

`

`C. F. Deacon et al.: Dipeptidyl peptidase IV resistant GLP-1 analogues
`
`275
`
`Table 2. In vivo plasma t1/2 for GLP-1 (7±36) amide and N-terminally modified analogues calculated using N- and C-terminally di-
`rected RIAs
`
`Group
`
`Thr8-GLP-1 (7±37)
`Gly8-GLP-1 (7±37)
`Ser8-GLP-1 (7±36) amide
`Aib8-GLP-1 (7±36) amide
`Values are mean ± SEM; n = 4
`a NS, p > 0.05 vs C-terminal t1/2 for analogue; b p < 0.01 vs N-
`terminal t1/2 for GLP-1 (7±36) amide; c NS, p > 0.05 vs C-termi-
`
`GLP-1 (7±36) amide t1/2 (min)
`Analogue t1/2 (min)
`N-terminal
`C-terminal
`N-terminal
`C-terminal
`0.7 ± 0.05d
`4.0 ± 0.1
`3.9 ± 0.2ab
`4.2 ± 0.4c
`0.9 ± 0.03d
`4.3 ± 0.5
`3.3 ± 0.4ab
`3.5 ± 0.7c
`0.9 ± 0.06d
`4.1 ± 0.2
`3.7 ± 0.4ab
`4.2 ± 0.2c
`1.1 ± 0.07d
`4.3 ± 0.3
`4.4 ± 0.2c
`NDL
`nal t1/2 for GLP-1 (7±36) amide; d p < 0.01 vs C-terminal t1/2 for
`GLP-1 (7±36) amide; NDL, not determined due to lack of
`cross-reactivity
`
`Table 3. Receptor binding affinities of GLP-1 (7±36) amide
`and N-terminally modified analogues in baby hamster kidney
`cells expressing the human pancreatic GLP-1 receptor
`
`Peptide
`
`Binding affinity
`(IC50, nmol/l)
`0.78 ± 0.29
`GLP-1 (7±36) amide
`49 ± 3.7a
`Thr8-GLP-1 (7±37)
`2.8 ± 0.42b
`Gly8-GLP-1 (7±37)
`9.0 ± 1.9a
`Ser8-GLP-1 (7±36) amide
`0.45 ± 0.05b
`Aib8-GLP-1 (7±37)
`Values are mean ± SD of separate triplicate experiments.
`a p < 0.001 vs GLP-1 (7±36) amide; b NS, p > 0.05 vs GLP-1 (7±
`36) amide
`
`n
`
`8
`4
`4
`3
`3
`
`logues in raising insulin output. It was also the most
`potent (p < 0.05) of all the analogues in reducing glu-
`cagon output. The Gly8 analogue was not significantly
`different to GLP-1 (7±36) amide in stimulating insu-
`lin or inhibiting glucagon secretion, but was more po-
`tent (p < 0.05) than the Ser8 and Thr8 analogues in in-
`hibiting glucagon release, while the Thr8 analogue
`was the least potent of the analogues tested.
`
`Discussion
`
`This study has demonstrated that small alterations in
`the N-terminus of GLP-1 confer resistance to the ac-
`tion of the enzyme DPP IV. Such analogues retain bi-
`ological activity, and have an improved metabolic sta-
`bility.
`In incubations with human plasma in vitro, DPP
`IV is the main enzyme responsible for GLP-1 degra-
`dation [12, 13], and a thorough study by Pauly et al.
`[31] concluded that only minor secondary degrada-
`tion could be attributed to other serum proteases. Ac-
`cordingly, the in vitro t1/2 of the GLP-1 analogues
`were considerably extended relative to GLP-1 (7±
`36) amide. There was no detectable degradation of
`the Aib8 analogue, and the slow degradation of the
`other three analogues could reflect the substrate
`specificity of DPP IV, as was found for GRF; substitu-
`tion of the alanine in position 2 of modified GRF an-
`alogues with either serine or threonine reduced the
`
`Fig. 3. Insulin and glucagon secretion by the isolated perfused
`porcine pancreas during infusion of 1 nmol/l GLP-1 (7±
`36) amide or N-terminally modified analogues. Data are ex-
`pressed as a percentage of basal output (defined as 100 %),
`and are mean ± SEM of 4 experiments, except for the glucagon
`output during GLP-1 (7±36) amide infusion where n = 3.
`* p < 0.05; ** p < 0.01 compared to basal output
`
`cleavage rates to less than 5 %, of that of the Ala2 an-
`alogue [20]. That DPP IV is also the main enzyme re-
`sponsible for in vivo N-terminal degradation of GLP-
`1 is illustrated by the increased proportions of N-ter-
`minal immunoreactivity (reflecting intact peptide)
`seen during i. v. infusion of each analogue relative to
`GLP-1 itself, and confirmed by HPLC analysis which
`revealed only two immunoreactive peaks corre-
`sponding to the intact peptide and the N-terminally
`truncated metabolite. In these in vivo studies, ana-
`logue N-terminal t1/2 were prolonged and equalled
`the C-terminal t1/2, but it is noteworthy that they
`could not be extended beyond those determined by
`
`MPI EXHIBIT 1072 PAGE 5
`
`MPI EXHIBIT 1072 PAGE 5
`
`

`

`276
`
`C. F. Deacon et al.: Dipeptidyl peptidase IV resistant GLP-1 analogues
`
`the C-terminal assay, presumably due to the presence
`of other, non-DPP IV-mediated degradation path-
`ways which become relevant in the intact animal. In
`studies of GLP-1 metabolism in the rat [14], dog [16,
`32] and pig [33], N-terminal degradation was shown
`to be particularly important, but the involvement of
`other enzymes was also indicated. Renal metabolism
`leading to substantial degradation to small, undetect-
`able fragments or complete cleavage was indicated to
`be a major route of GLP-1 elimination [33], and may
`explain why no other metabolites were detected after
`HPLC analysis in the present study.
`The binding affinity (IC50) for GLP-1 in the pre-
`sent study correlates well with the previously pub-
`lished Kd of 0.5 nmol/l [34]. All four analogues also
`bound to the cloned GLP-1 receptor, with the Aib8
`and Gly8 analogues having similar affinities to GLP-
`1 itself. It may be that, for the other analogues, steric
`hindrances cause the reduction in binding affinity,
`with the polar hydroxyl group in the serine and threo-
`nine residues impairing binding due to spatial con-
`straints.
`In the isolated perfused pancreas, all the GLP-1
`analogues were, to varying extents, capable of releas-
`ing insulin and/or inhibiting glucagon secretion, with
`potency orders reflecting receptor affinities. Similar
`findings were seen for GRF; serine, threonine and
`glycine substituted analogues were stabilised against
`proteolysis in plasma, but had low inherent growth
`hormone releasing activity when tested in vitro [35].
`However, the potency order in vivo changed, with
`the Thr2 analogue becoming more potent than native
`GRF, which may be explained by the improved meta-
`bolic stability of the analogue in vivo compensating
`for its lower in vitro potency.
`GLP-1 has been suggested to be a useful new ther-
`apy in the treatment of NIDDM [6, 8±11]. However,
`the therapeutic potential of native GLP-1 is limited
`by its susceptibility to degradation by DPP IV [12±
`14], with the concomitant formation of a metabolite
`which may act as an antagonist in its own right [15,
`16]. The short metabolic stability of the intact pep-
`tide can be offset by sustained i. v. infusion or by de-
`velopment of a protracted formulation, but this does
`not overcome the effect of DPP IV action. In recent
`studies in patients with NIDDM in whom GLP-1
`was infused overnight, Rachman et al. [36, 37] ob-
`served that, despite an initial fall, blood glucose con-
`centrations gradually increased during the night, al-
`though they remained lower than in the saline-treat-
`ed group. These authors concluded that ªGLP-1
`may have been losing its efficacy overnightº. Anoth-
`er explanation could be that concentrations of GLP-
`1 (9±36) amide increase during the overnight GLP-1
`infusion, so that the gradual rise in blood glucose
`may not be due to a post-receptor down-regulation,
`but rather to a local accumulation of metabolite an-
`tagonising the action of the residual intact peptide.
`
`In these studies [36, 37], the relative concentrations
`of intact GLP-1 and metabolite were not assessed,
`but this was addressed in another study, where GLP-
`1 (9±36) amide was found to reach 80 % of the total
`plasma immunoreactivity during an i. v. infusion of
`GLP-1 in NIDDM and control subjects [17]. Howev-
`er, marked desensitisation appears not to occur, since
`when GLP-1 was infused in NIDDM patients contin-
`uously i. v. over a 7-day period, both fasting and post-
`prandial glucose levels were significantly lower on
`the 7th day of treatment compared to pre-treatment
`values [38]. This study [38] was the first to show that
`continuous GLP-1 infusion is capable of improving
`glycaemic control in NIDDM patients for a pro-
`longed period, although it should be borne in mind
`that neither this [38] nor any other study has un-
`equivocally excluded the possibility that some loss
`of efficacy occurs over the course of the treatment
`period.
`It appears that in order to maintain its effect,
`GLP-1 must be continuously present. Thus, in NID-
`DM patients, when the peptide was infused for only
`16 h instead of continuously over 24 h, the beneficial
`effect on fasting blood glucose was reduced [39].
`Similar results are seen when GLP-1 is given to
`NIDDM patients as repeated s. c. injections. When
`administered 3 times daily before meals, it retained
`its ability to reduce postprandial increases in blood
`glucose throughout the 1 week course of the study
`[40], but the effect was lost between meals, suggest-
`ing that GLP-1 concentrations fell below the thresh-
`old for activity between successive injections. An
`earlier study be the same group [41] followed the ki-
`netics, and showed that total GLP-1 immunoreactiv-
`ity returned to basal levels by 215 min after the pep-
`tide injection. Similar findings were reported when
`immunoreactive GLP-1 was characterised by HPLC
`after s. c. administration to diabetic and non-diabetic
`subjects [17]. By 30 min after administration, the
`metabolite accounted for around 80 % of the plasma
`immunoreactivity, and concentrations of both GLP-
`1 (7±36) amide and GLP-1 (9±36) amide returned to
`basal values within 4 h. The effect of repeated s. c.
`injections of GLP-1 given before meals was con-
`firmed in another study using poorly controlled
`NIDDM patients on sulphonylurea therapy [42].
`Here, the peptide maintained its beneficial effect
`on postprandial glucose levels over the 3-week study
`period, although, again, it had no effect on fasting
`glucose levels. Thus, it appears that when GLP-1 is
`given as repeated injections before meals, full 24 h
`control of blood glucose is not attained. Tachyphy-
`laxis appears not to occur, possibly because both in-
`tact peptide and the metabolite are eliminated be-
`tween successive injections, preventing accumulation
`of a potential antagonist, however, leaving a period
`when there is insufficient intact GLP-1 left to main-
`tain an effect.
`
`MPI EXHIBIT 1072 PAGE 6
`
`MPI EXHIBIT 1072 PAGE 6
`
`

`

`C. F. Deacon et al.: Dipeptidyl peptidase IV resistant GLP-1 analogues
`
`277
`
`The kinetic data reported here for the GLP-1 ana-
`logues indicate that resistance to degradation by
`DPP IV is associated with prolonged metabolic sta-
`bility in vivo, raising the possibility that such ana-
`logues may have greater potency and duration of ac-
`tion. Clearly, further dynamic studies in vivo are re-
`quired to test whether this hypothesis is valid, and to
`see whether the 3-fold improvement in plasma t1/2 of
`the analogues in itself is sufficient to maintain an ef-
`fect or requires development of a protracted formula-
`tion. Nonetheless, DPP IV-resistant analogues may
`be one method of realising GLP-1's potential in dia-
`betes therapy, by extending its duration of action
`while, at the same time, minimising the build up of
`undesirable metabolites.
`
`Acknowledgements. This work was supported by grants from
`the Danish Medical Research Council and the Danish Biotech-
`nology Programme. The technical assistance of Letty Klarskov,
`Mette Olesen and Mette Frost is gratefully acknowledged. Siv
`Hjort is thanked for kindly providing the pAH 260 vector.
`
`References
`
`1. Holst JJ (1983) Gut glucagon, enteroglucagon, gut gluca-
`gonlike immunoreactivity, glicentin: current status. Gastro-
`enterology 84: 1602±1613
`2. érskov C (1992) Glucagon-like peptide-1, a new hormone
`of the entero-insular axis. Diabetologia 35: 701±711
`3. érskov C, Rabenhùj L, Wettergren A, Kofod H, Holst JJ
`(1994) Tissue and plasma concentrations of amidated and
`glycine-extended glucagon-like peptide-1 in humans. Dia-
`betes 43: 535±539
`4. érskov C, Holst JJ, Nielsen OV (1988) Effect of truncated
`glucagon-like peptide-1 [proglucagon-(78±107) amide] on
`endocrine secretion from pig pancreas, antrum and non-an-
`tral stomach. Endocrinology 123: 2009±2013
`5. Kreymann B, Williams G, Ghatei MA, Bloom SR (1987)
`Glucagon-like peptide 17±36: a physiological incretin in
`man. Lancet ii: 1300±1304
`6. Nathan DM, Schreiber E, Fogel H, Mojso