Pharmaceutical Research, Vol. 15, No. 2, 1998
`
`Research Paper
`
`Effects of Non-Covalent Self-
`Association on the Subcutaneous
`Absorption of a Therapeutic Peptide
`
`Dean K.Clodfelter,! Allen H. Pekar,! Dawn M.
`Rebhun,' Kevin A. Destrampe,! Henry A. Havel,
`Sharon R. Myers,” and Mark L. Brader!*
`
`Received July 31, 1997; accepted November 7, 1997
`
`Purpose. To utilize an acylated peptide as a model system to investi-
`gate the relationships among solution peptide conformation, non-
`covalent self-association, subcutaneous absorption and bioavailability
`under pharmaceutically relevant solution formulation conditions.
`Methods. CD spectroscopy, FTIR spectroscopy, equilibrium sedimen-
`tation, dynamic light scattering, and size exclusion chromatography
`were employedto characterize the effects of octanoylation on confor-
`mation and self-association of the 31 amino acid peptide derivative
`des-amino-histidine(7) arginine(26) human glucagon-like peptide
`(7-37)-OH (IP(7)R(26)GLP-1). Hyperglycemic clamp studies were
`performed to comparethe bioavailability, pharmacokinetics, and phar-
`macodynamics of solution formulations of oct-IP(7)R(26)GLP-1
`administered subcutaneously to normal dogs.
`Results. Octanoylation of IP(7)R(26)GLP-1 was shown to confer the
`propensity for a major solvent-induced conformational transition with an
`accompanying solvent- and temperature-dependent self-association
`behavior. Formulations were characterized that give rise to remarkably dif-
`ferent pharmacodynamics and pharmacokinetics that correlate with dis-
`tinct peptide conformational and self-association states. These states
`correspondto:(i) a minimally associated a-helical form (apparent molec-
`ular weight = 14 kDa),(ii) a highly associated, predominantly §-sheet form
`(effective molecular diameter 20 nm), and (iii) an unusually large, micelle-
`like soluble B-sheet aggregate (effective molecular diameter 50 nm).
`Conclusions. Bioavailability and pharmacokinetics of a self-associating
`peptide can be influenced by aggregatesize and the ease of disruption of
`the non-covalent intermolecular interactions at the subcutaneoussite.
`
`Hydrophobic aggregation mediated by seemingly innocuous solution
`formulation conditions can have a dramatic effect on the subcutaneous
`
`bioavailability and pharmacokinetics of a therapeutic peptide and in the
`extreme, can totally preclude its absorption. A size exclusion chromato-
`graphic methodis identified that distinguishes subcutaneously bioavail-
`able aggregated oct-IP(7)R(26)GLP-1 from non-bioavailable aggregated
`oct-IP(7)R(26)GLP-1.
`
`KEY WORDS: peptide; aggregation; subcutaneous absorption;
`glucagon-like peptide-1.
`
` Biopharmaceutical Development, Lilly Research Laboratories, Eli
`Lilly and Company, Indianapolis, Indiana 46285.
`? Diabetes Research, Lilly Research Laboratories, Eli Lilly and
`Company, Indianapolis, Indiana 46285.
`> To whom correspondence should be addressed. (E-mail: brader_m_I
`@lilly.com)
`ABBREVIATIONS: Glucagon-like peptide-1(7-37)-OH (GLP-1);
`imidazopropionyl(7) arginine(26) lysine(34)-human glucagon-like
`peptide-1 (IP(7)R(26)GLP-1); imidazopropionyl(7) arginine(26) N°-
`octanoyl-lysine(34)-human glucagon-like peptide-1 (oct-IP(7)R(26)
`GLP-1); circular dichroism, CD; size exclusion chromatography,
`SEC; Fourier transform infrared, FTIR; phosphate buffered saline,
`PBS.
`
`INTRODUCTION
`
`The efficacy of subcutaneously administered therapeutic
`proteins and peptides is critically dependent upon the absorp-
`tion and subsequent delivery of the biologically active form of
`the drugto thesite of action. The task of the formulation scien-
`tist includes conferring the appropriate shelf-life and in-use sta-
`bility as well as stabilizing the molecule in a form thatfacilitates
`optimal bioavailability and biopotency. The formulation design
`mayalso play the principal role in mediating drug absorption
`thus providing a means totailor appropriate pharmacokinetics
`for the drug. Currently, most biopharmaceuticals are adminis-
`tered via subcutaneousinjection or intravenous infusion. While
`the importance of chemical and physical stability are obvious,
`the potential pharmacological consequences of non-covalent
`solution structural phenomena suchas alternative molecular
`conformations and/or self-association may be a poorly appreci-
`ated but critical aspect of the overall therapeutic success of the
`drug. This is particularly relevant to discovery and early stage
`drug development where a large-scale screening strategy or
`an emphasis on speed-to-first-efficacy-dose may preclude a
`detailed solution characterization of the molecule prior to dos-
`ing (1). The non-covalent aggregation of protein pharmaceuti-
`cals is a well recognized problem that has been studied largely
`in the context of insoluble forms, including precipitates, fibrils,
`and gels (2,3). These insoluble products have been attributed to
`the formation of partially unfolded intermediates with an
`exposed hydrophobic region that drives the aggregation
`towards the pharmaceutically undesirable form (2-4). Previous
`studies of this mechanism have been performed by introducing
`a denaturant to generate the partially unfolded intermediate
`(4,5). By studying an acylated peptide we haveeffectively
`introducedan artificial hydrophobic regionto the surface of the
`peptide in the absence of a denaturant. This approach hasfacil-
`itated an investigation of the relationship between hydrophobic
`aggregation and subcutaneous bioavailability under pharma-
`ceutically relevant solution conditions.
`Glucagon-like peptide-1(7-37)-OH (GLP-1) is a 31 amino
`acid hormoneliberated by the proteolytic processing of the
`160 amino acid precursor protein, preproglucagon. GLP-1
`stimulates the secretion of insulin and thus has the ability to
`normalize blood glucose levels (6). Interest in GLP-1 and its
`analogshas intensified recently as the attractiveness of GLP-1
`as a potential therapeutic agent for the treatmentof type II dia-
`betes has been recognized (6-8). The structure of the analog we
`have studied is shown in Figure 1 and will be abbreviated as
`oct-IP(7)R(26)GLP-1. This molecule exhibits three changes
`from the native GLP-1(7-37)-OH structure: the amino group
`has been removed from the His(7) residue (becoming des-
`amino-histidine, or imidazopropionyl and is abbreviated herein
`as IP), Lys(26) has been replaced by Arg, and Lys(34) has been
`acylated with the straight chain fatty acid octanoic acid (abbre-
`viated as oct). The present study characterizes the effect of the
`octanoyl acylation on the structure, conformation and self-
`association of this GLP-1 analog and demonstrates that these
`properties are highly solvent dependent. This molecule thus
`affords an opportunity to investigate the relationships among
`peptide secondary structure, non-covalent molecular associa-
`tion, subcutaneous bioavailability and the pharmacodynamics
`
`0724-874 1/98/0200-0254$15.00/0 © 1998 Plenum Publishing Corporation
`
`254
`
`MPI EXHIBIT 1015 PAGE 1
`
`MPI EXHIBIT 1015 PAGE 1
`
`MPI EXHIBIT 1015 PAGE 1
`
`

`

`Effects of Peptide Aggregation on Subcutaneous Absorption
`
`of a single therapeutic peptide formulated under different, phar-
`maceutically typical solution conditions. Of additional signifi-
`cance to this study is the fact that the acylation of proteins in
`vivo is a recently recognized covalent modification involved in
`intracellular signaling pathways (9). The way in whichthe acy-
`lation mediates these processes is poorly understood. The
`IP(7)R(26) GLP-1 moiety thus represents a useful model system
`to investigate the effects of acylation on peptide conformational
`and associative behavior.
`
`MATERIALS AND METHODS
`
`Materials
`
`Oct-IP(7)R(26)GLP-1 and IP(7)R(26)GLP-1 were pro-
`vided by Eli Lilly and Companyas highly purified lyophilized
`powders. All other chemicals were of analytical reagent grade.
`Dulbecco’s phosphate buffered saline without Ca and Mg
`(PBS) was obtained from ICN Biomedical Inc. Solution con-
`centrations of oct-IP(7)R(26)GLP-1 and IP(7)R(26)GLP-1
`were calculated based on the respective extinction coefficients
`of 1.95 (mg/ml)! + cm! and 2.01 (mg/ml)! - cm7! at 279 nm.
`
`Size Exclusion Chromatography
`
`Based oncriteria of selectivity and resolution, a 25 cm x
`9.4 mm Zorbax GF-250 special column was chosenfor the pre-
`sent studies with a mobile phase comprising 14 mM sodium
`phosphate dibasic adjusted to pH 7.4 with 85% phosphoric
`acid. The flow rate was 1 ml/min andthe injection volume was
`50 yl of a 1 mg/mlsolution. The eluent was detected using UV
`absorbanceat 214 nm. (Detailed studies to investigate the chro-
`matographic properties of several SEC columns under various
`mobile phase conditions were undertaken and will be reported
`elsewhere (10)).
`
`Circular Dichroism Spectroscopy
`
`Circular dichroism spectra were recorded using an AVIV
`Model 62A DSspectrometer calibrated with (1S)-(+)-10-cam-
`phorsulfonic acid. Circular dichroism is reported as mean
`residueelipticity, [6],, having units of degrees + cm? + dmoi™!.
`Secondary structural analyses were performed using the pro-
`gram CONTIN (11-12).
`
`Analytical Ultracentrifugation
`
`Sedimentation equilibrium experiments were carried out
`at 4 °C in a Beckman XLAanalytical ultracentrifuge using
`absorbance optics. Cells with quartz windows and either 3 mm
`or 12 mm centerpieces were used. The apparent weight-average
`molecular weights, Myapp, were calculated from equation (1).
`
`Mwapp = (RT/((1- vp)@*rC)}-dC/dr
`
`(1)
`
`whereR is the gas constant, T is temperature, v is the partial spe-
`cific volume, p is the solvent density,r is radius, @ is the angular
`velocity and C is the total protein concentration in mg/ml. A par-
`tial specific volume of 0.726 ml/gram wascalculated from the
`amino acid composition (13). It was assumedthat the addition of
`the acyl group did not change the value of v. The solventdensity
`
`255
`
`was measured using a Paar DMA48densitometer. Data analyses
`to model the self-association and determine equilibrium con-
`stants (14) were carried out using the program NONLIN,avail-
`able through the National Analytical Ultracentrifuge Facility at
`the University of Connecticut. NONLINgives values of equilib-
`rium constants k,,, on (a mg/ml concentration scale) which refer
`to the reversible formation of an n-mer from n monomeric units,
`as described by equation (2). (Thus the tetramer constant k, 4 cor-
`respondsto the reaction of four monomersto give a tetramer; kj g
`corresponds to the reaction of eight monomers to give an
`octamer; k,s correspondsto the reaction of two tetramers to give
`an octamer). The corresponding molar equilibrium constant K; ,
`is given by equation (3).
`
`k,, =[n- mer]/[monomer]"
`
`Ky. = kj, ‘(monomer molecular weight)"/n
`
`(2)
`
`GB)
`
`Infrared Spectroscopy
`
`Infrared spectra were recorded on a Nicolet Magna 750
`Fourier transform infrared spectrophotometer equipped with a
`Nicolet Nic-Plan infrared microscope. Data were acquired for
`128 scans at 4 cm’. Data acquisition and second derivative
`analysis were performed with Omnic 3.1 software.
`
`Dynamic Light Scattering
`
`Measurements were performed using a Brookhaven
`Instruments 9000 autocorrelator and goniometer. All measure-
`ments were made with a 400 um pinhole at a 90° scattering
`angle using a Lexel Model 3500 argonion laser set at 488 nm
`as the scattering source. Sample temperature was maintained
`at 25°C or 5°C by a Neslab RTE-110 temperature bath.
`Brookhaven Instruments software was used to calculate the
`diffusion coefficients of the scattering species from the mea-
`sured autocorrelation function using the method of cumulants.
`Diffusion coefficients were converted to mean diameters using
`the Stokes-Einstein relationship. Values for solution viscosi-
`ties and refractive indices were assumedto be equal to those of
`pure water.
`
`In Vivo Testing
`
`Study Design and Animals
`
`GLP-1 is an incretin peptide hormonethat stimulates
`insulin secretion from the B-cell in a glucose dependent manner
`(6). Hyperglycemic clamp experiments (150 mg/dl) were con-
`ducted using chronically cannulated, overnight-fasted, conscious
`male and female beagle dogs weighing 8-15 kg. Pharmaco-
`dynamics were evaluated from the plasmainsulin data and phar-
`macokinetics were determined from blood drug levels. The
`insulin change and drug level areas under the curve were calcu-
`lated using the trapezoidal rule. Values are reported as the mean
`+ the standard error of the mean.Prior to initiation of the study
`the animals were judged to be healthy by physical examination
`and laboratory tests. Research adhered to the Principles of
`Laboratory Animal Care of the NIH.
`
`MPI EXHIBIT 1015 PAGE 2
`
`MPI EXHIBIT 1015 PAGE 2
`
`MPI EXHIBIT 1015 PAGE 2
`
`

`

`(deg.cm2.dmol"} 190
`
`Clodfelter et al.
`
`[9],
`
`300
`
`320
`
`210
`
`230
`
`250-260
`Wavelength (nm)
`
`280
`
`
`
`{@],deg.cm2.dmol"!
`
`256
`
`
`
`Fig. 1. The structure of imidazopropionyl(7) arginine(26) N®-octanoyl-
`lysine(34)-human glucagon-like peptide-1(7-37)-OH (oct-IP(7)R(26)
`GLP-1).
`
`Formulation Preparation and Administration
`
`Formulations were injected subcutaneously into the dorsal
`aspect of the neck at a dose of 3 nmol/kg. Four formulations
`stabilizing distinct conformational and self-association states
`(vide infra) of oct-IP(7)R(26)GLP-1 were evaluated. These
`formulations were prepared by dissolving lyophilized oct-
`IP(7)R(26)GLP-1 under the following conditions: (i) 5 mM
`phosphate buffer, pH = 7.5 prepared at room temperature
`immediately prior to administration (Myapp approximately
`14 kDa by equilibrium sedimentation) (ii) PBS, pH = 7.5 pre-
`pared at room temperature immediately prior to administration
`(quadratic diameter approximately 10 nm by dynamic light scat-
`tering at time of dosing) (iii) PBS, pH = 7.5 prepared and stored
`at 5 °C for 24 hoursprior to dosing (quadratic diameter approx-
`imately 20 nm by dynamiclight scattering at time of dosing),
`and (iv) PBS, pH =7.5 prepared and stored at room temperature
`for 24 hours prior to dosing (quadratic diameter approximately
`50 nm by dynamiclight scattering at time of dosing).
`
`RESULTS
`
`Circular Dichroism Studies
`
`The far-UVcircular dichroism (CD) spectra of IP(7)R(26)
`GLP-1 and oct-IP(7)R(26)GLP-1 recorded under monomeric
`conditions in 5 mM phosphate buffer pH 7.5 are presented in
`Figure 2A (a) and (b) respectively. These spectra show that
`octanoylation causes a slight perturbation in the far-UV CD of
`the IP(7)R(26)GLP-1 peptide (Table I). Figure 2A (a) and
`(b) appear very similar to that reported for native GLP-1 (15)
`and were found to exhibit analogous concentration dependencies
`
`Fig. 2. Panel A: Far-UV CDspectra of IP(7)R(26)GLP-1 (a) and oct-
`IP(7)R(26)GLP-1 (b) prepared and recorded at 5 °C in 5 mM phos-
`phate buffer, pH 7.5. Oct-IP(7)R(26)GLP-1 prepared and recorded at
`5 °C (c) and 22 °C (d) in PBS, pH 7.5. Spectra (a) and (b) were recorded
`on 0.1 mg/ml and 0.02 mg/ml solutions, respectively. IP(7)R(26)GLP-
`1 and oct-IP(7)R(26)GLP-1 are monomeric under these conditions.
`Spectra (c) and (d) were recorded on 0.5 mg/ml solutions. Only minor
`differences in the CD spectrum ofoct-IP(7)R(26) GLP-1 in PBS were
`apparent over the concentration range 0.1-0.5 mg/ml. Panel B: Near-
`UV CDspectra of IP(7)R(26)GLP-1 (a) and oct-IP(7)R(26)GLP-1
`(b) prepared and recorded at 5 °C in 5 mM phosphate buffer, pH 7.5.
`Oct-IP(7)R(26)GLP-1 prepared and recorded at 5 °C (c) and 22 °C
`(d) in PBS, pH 7.5. Spectra were recorded on 0.5 mg/ml solutions.
`
`(an intensification with increasing concentration). A compari-
`son of the CD characteristics of each molecule in 5 mM phos-
`phate buffer pH 7.5 and in PBS showed that the IP(7)R(26)
`GLP-1 molecule possesses closely similar CD spectra in these
`two solvents, whereas the octanoylated peptide exhibits dra-
`matically different CD spectra. The CD spectrum of Figure 2A
`(c) corresponds to an oct-IP(7)R(26)GLP-1 solution prepared
`in PBS at 5 °C. Comparisonof this spectrum with Figure 2A (b)
`showsthat a major conformational rearrangementhas occurred.
`The CD spectrum of Figure 2A (d) corresponds to an oct-IP(7)R
`(26)GLP-1 solution in PBS prepared and recorded at 22 °C. The
`data of Table I show that under these solution conditions, an
`almost complete loss of o-helical structure has occurred.
`The corresponding near-UV CD spectra are presented in
`Figure 2B. These data show that under the higher ionic strength
`conditions (PBS), oct-IP(7)R(26)GLP-1 is characterized byrel-
`atively intense CD features with positive maximain the range
`265-295 nm. These results show that the changes in secondary
`structure evident from Figure 2A are accompanied by major
`changes in the chromophoric environments of the aromatic
`residues.
`
`Table I. Secondary Structural Analyses of the CD Spectra of Figure 2AASSSSSSAS
`
`Peptide
`
`Solvent
`
`Temp/°C %a-helix
`
`%B-sheet
`
` %remainder
`
`43
`40
`16
`5
`5 mM PB
`IP(7)R(26)GLP-1
`42
`31
`28
`5
`5 mM PB
`oct-IP(7)R(26)GLP-1
`61
`16
`23
`5
`PBS
`oct-IP(7)R(26)GLP-1
`62
`36
`2
`22
`PBS
`oct-IP(7)R(26)GLP-1
`
`
`Note: Analyses were performed on 190-240 nm datausing the program CONTIN(11,12). PB refersto
`phosphate buffer pH 7.5 and PBS refers to 10 mM phosphatebuffered saline pH 7.5.
`
`MPI EXHIBIT 1015 PAGE 3
`
`MPI EXHIBIT 1015 PAGE 3
`
`MPI EXHIBIT 1015 PAGE 3
`
`

`

`257
`
`8.18 x 107” and Ky, = 1.97 x 10*, where K,4 and Ky corre-
`spond respectively to the formation of tetramers and octamers
`from monomers and Ky, corresponds to the formation of
`octamers from tetramers. For oct-IP(7)R(26)GLP-1 the values
`were K, 4=7.29 x 10!4, K,g = 4.00 x 10° and Kys = 7.53 x 10°.
`These models generated good fits to the experimental data
`(Figure 3), although we note that they may not necessarily rep-
`resent unique fits. The larger value of K,, for the acylated
`derivative is consistent with the greater concentration depen-
`dence of weight average molecular weight over the 0-0.5 mg/ml
`concentration range. However, the subsequent formation of
`octamers from tetramers is somewhat smaller for the acylated
`compoundthan for the non-acylated compound.
`
`Dynamic Light Scattering
`
`In view of the extreme degree of aggregation of oct-
`IP(7)R(26)GLP-1 in PBS, dynamic light scattering was
`selected as the most appropriate technique to characterizeself-
`association under these solution conditions. The time-depen-
`dence of the aggregation of oct-IP(7)R(26)GLP-1 in PBS at
`25 °C and 5 °C as‘monitored by dynamic light scattering is
`shown in Figure 4 (a) and (b) respectively. These data show
`that upon initial reconstitution of the lyophilized powder in
`PBS at 5 °C, the quadratic diameter is approximately 10 nm.
`This value increases to about 20 nm over a period of 24 hours.
`In contrast, for the reconstitution in PBS at 25 °C, theinitial
`quadratic diameter is approximately 20 nm. This value increases
`to about 50 nm overa period of 24 hours.
`
`Size Exclusion Chromatography
`
`The solvent-dependent CD spectral characteristics of oct-
`IP(7)R(26)GLP-1 were found to be accompaniedbya distinc-
`tive size exclusion chromatographic (SEC) signature. The SEC
`
`(nm)
`
`QuadraticDiameter
`
`°
`
`a (
`
`b)
`0,000e00 0 0900%o
`fogeoee?”?
`ee
`Mo90o%— 00070090
`
`0
`
`10
`
`20
`
`Time (hours)
`Fig. 4. Dynamic light scattering data for a 0.5 mg/mi solution of oct-
`IP(7)R(26)GLP-1 formulated in PBS at 25 °C (a) and 5 °C (b). These
`data show the time dependence of
`the aggregation of oct-
`IP(7)R(26)GLP-1 under these solution conditions. The aggregation
`states shownby dataset(a) at t= 0 and t = 24 hours correspondto those
`of the samples administered in the dog studies of Figure 7 A and B (b)
`and Figure 7 C and D (c), respectively. The aggregation state shown by
`dataset (b) at t = 24 hours corresponds to that of the PBS sample
`administered in the dog study of Figure 7 C and D (b).
`
`MPI EXHIBIT 1015 PAGE 4
`
`Effects of Peptide Aggregation on Subcutaneous Absorption
`
`Equilibrium Sedimentation Studies
`
`Detailed analytical ultracentrifugation experiments were
`performed on IP(7)R(26)GLP-1 and oct-IP(7)R(26)GLP-1
`under low ionic strength conditions. In 5 mM phosphate buffer
`pH 7.5 at 4 °C, the non-acylated compound reached equilib-
`rium in about 25 hours. For the acylated derivative, it was noted
`that a very slight loss of material occurred with time, possibly
`due to the slow formation of a small fraction of highly aggre-
`gated species. On the assumption that the acylated molecule
`wasclose to equilibrium, the apparent weight average molecu-
`lar weights were calculated and plotted against concentration
`(Figure 3). Each curve was constructed using data from three
`cells having different loading concentrations. Overlap of these
`data was good for each molecule, consistent with self-associa-
`tion. Since native GLP-1(7-37) has been reported to self-asso-
`ciate to tetramers (16), an ideal monomer-tetramer-octamer
`self-association model was chosen. A goodfit of concentration
`versus radius was achieved with this model for all three cells
`for IP(7)R(26)GLP-1 and is shownin Figure 3. For the acylated
`derivative, the monomer-tetramer-octamer model fit the data
`well in the lower concentration region butnot at higher concen-
`trations. However, by including a non-ideality term in the mod-
`eling, it was possible to achieve a good fit of these data to a
`monomer-tetramer-octamer association mechanism. The molar
`equilibrium constants obtained from the curve fit
`to the
`IP(7)R(26)GLP-1 data of Figure 3 were K, 4= 6.44 x 10"!, K, g=
`
`2
`
`1.5
`1
`[peptide] (mg/ml)
`Fig. 3. Equilibrium sedimentation data for IP(7)R(26)GLP-1 (open
`symbols) and oct-IP(7)R(26)GLP-1 (solid symbols) in 5 mM phos-
`phate buffer pH = 7.5, 4 °C. The solid curves were calculated for a
`monomer-tetramer-octamerself-association mechanism using the pro-
`gram NONLIN. M,/M, represents the weight average molecular
`weight obtained from ultracentrifugation divided by the molecular
`weight of monomer.
`
`Mw/Mi
`
`0
`
`0.5
`
`MPI EXHIBIT 1015 PAGE 4
`
`MPI EXHIBIT 1015 PAGE 4
`
`

`

`258
`
`Clodfelter et al.
`
`profiles of IP(7)R(26)GLP-1 and oct-IP(7)R(26)GLP-1 pre-
`pared in 5 mM phosphate buffer pH 7.5 are presentedin Figure 5
`(a) and (b) respectively. SEC profiles (c) and (d) correspondto
`oct-IP(7)R(26)GLP-1 solutions prepared in PBS at 5 °C and
`22 °C respectively, and aged at these temperatures for 24 hours
`prior to analysis. These data show that the SEC retention times
`of oct-IP(7)R(26)GLP-1 solutions prepared in 5 mM phosphate
`buffer at 5 °C and in PBS at 5 °C are equivalent. However,in
`PBSat 22 °C the SECretention time of oct-IP(7)R(26)GLP-1 is
`dramatically different and indicative of a much higher apparent
`molecular weight. It was determined that this aggregated
`species could be disaggregated in solution by incorporating
`30% acetonitrile into the PBS solvent. SEC analysis of this
`sample (Figure 5 (e)) produced an SECretention time equalto
`that of Figure 5 (b). This result indicates that the species corre-
`sponding to the SEC peak in Figure 5 (d) is a hydrophobically
`associated soluble aggregate. The broadnessof the oct-IP(7)R
`(26)GLP-1 peak in (b)}-(e) is probably due to enhanced hydro-
`phobic interactions between the peptide and the column pack-
`ing in addition to on-column equilibria involving a range of
`aggregationstates (17). The chromatogramsof Figure 5 (b) and
`(c) exhibit the sameretention time, however, the dynamiclight
`scattering and equilibrium sedimentation data presented herein
`show that these samples correspond to different states of self-
`association. The difference between these two aggregation
`States is not evident by this SEC method, thus it is inferred
`that the SEC column conditions disrupt the non covalent self-
`association that occurs in PBS at 5 °C. In contrast, the SEC
`characteristics suggest a different entity in PBS at 22 °C, an
`interpretation in accord with the dynamic light scattering data
`of Figure 4 and the distinctive CD spectroscopic characteris-
`tics. Evidently, the peptide molecules are more strongly self-
`associated in PBS at 22 °C than underthe solution conditions of
`Figure 5 (a)-(c). It is speculated that this aggregate corresponds
`to a micellar species for which the different CD and SEC char-
`acteristics in PBS at 5 °C versus 22 °C, correspond to tempera-
`tures that bracketthecritical micelle temperature.
`
`Infrared Spectroscopy
`
`Infrared spectroscopy provides a convenient method to
`probe protein conformation andstructure in the solid state. The
`infrared spectral profiles of the amide region of oct-IP(7)R
`(26)GLP-1 are presented in Figure 6. Spectrum (A) corre-
`spondsto oct-IP(7)R(@6)GLP-1, lyophilized from 5 mM phos-
`phate buffer pH 7.5. This spectrum exhibits amide I and amide
`Ii bands at 1659 cm™! and 1542 cm" respectively, values con-
`sistent with an appreciable a-helical secondary structure con-
`tent (18). Spectra (B) and (C) correspond to samples of
`oct-IP(7)R (26)GLP-1 each lyophilized from PBS, pH 7.5, that
`had been aged 24 hours at 5 °C and 22 °C respectively. Spectra
`(B) and (C) both exhibit features at 1695 cm™! (shoulder) 1660
`cm7! (main) and 1625 cm“! (shoulder), resolved in the second
`derivative spectra (not shown). By comparison to spectrum (A)
`these spectra show that a major conformational change has
`taken place as a result of dissolving oct-IP(7)R(26)GLP-1 in
`PBS prior to lyophilization. The IR spectral profiles of (B) and
`(C) are consistent with a significant increase in B-sheet content
`(18). The data of Figure 6 establish that in the solid state the
`oct-IP(7)R(26)GLP-1 species can adopt a secondary structure
`that is either predominantly a-helical or predominantly B-sheet
`depending upon the solution conditions from which it was
`lyophilized. These infrared results on the solid state are in con-
`formity with the CD data for oct-IP(7)R(26)GLP-1 in solution.
`
`In Vivo Testing
`
`The data of Figure 7 compare the plasmainsulin responses
`(A and C) and the plasma drug levels (B and D) for oct-
`
`arn
`
`Abs
`214nm
`
`Absorbance
`
`ioe
`
`0
`
`240
`
`480
`
`720
`
`960
`
`1200
`
`Time (seconds)
`Fig. 5. Size exclusion chromatograms recorded on IP(7)R(26)GLP- 1
`(a) and oct-IP(7)R(26)GLP-1 (b) solutions prepared at 5 °C in 5 mM
`phosphate buffer, pH = 7.5. Chromatograms(c) and (d) were recorded
`on oct-IP(7)R(26)GLP-1 solutions prepared in PBS at 5 °C and 22 °C
`respectively, and aged at these temperatures for 24 hoursprior to
`analysis. Chromatogram (e) corresponds to a solution of oct-
`IP(7)R(26)GLP-1 prepared in an identical mannerto (d) that had 30%
`acetonitrile incorporated into the sample solvent immediately prior to
`SEC analysis. All samples were prepared at a peptide concentration of
`1 mg/ml.
`
`1800
`
`1700
`
`1400
`1500
`1600
`Wavenumber (cm-1)
`Fig. 6. FTIR spectra recorded on solid samples of oct-IP(7)R
`(26)GLP-1 prepared by lyophilization from the following solutions;
`(A) 5 mM phosphate buffer, pH 7.5, T = 5 °C (B) PBS, pH 7.5, T=5
`°C (C) PBS, pH 7.5, T = 22°C.
`
`1300
`
`MPI EXHIBIT 1015 PAGE 5
`
`MPI EXHIBIT 1015 PAGE 5
`
`MPI EXHIBIT 1015 PAGE 5
`
`

`

`Effects of Peptide Aggregation on Subcutaneous Absorption
`
`anQo2om)
`
`nN>2Oo Plasma[oct-IP(7)R(26)GLP-1](pM) SSs
`Plasma[Insulin](ng/ml)
`
`
`
`
`2000 Plasma[Oct-IP(7)R(26)GLP-1](pM)
`
`
`Plasma[Insulin](ng/ml)
`
`
`
`259
`
`IP(7)R(26)GLP-1 administered underfour different formulation
`conditions. The data labeled (a) correspondto a vehicle control in
`which PBS was administered. In Figure 7 A and B, the curves
`labeled (c) correspond to oct-IP(7)R(26)GLP-1 formulated in
`5 mM phosphate buffer, and the curves labeled (b) correspondto
`oct-IP(7)R(26)GLP-1 formulated in PBS at room temperature
`and administered immediately upon dissolution of
`the
`lyophilized oct-IP(7)R(26)GLP-1 in these solvents. In this exper-
`iment, the quadratic diameter of the peptide in the formulation
`corresponding to (b) was determined by dynamic lightscattering
`to be 20 nm (Figure 4). The apparent molecular weight ofthe for-
`mulation corresponding to (c) was determined by equilibrium
`sedimentation to be about 14 kDa. The data of Figure 7 A and B
`show that oct-IP(7)R(26)GLP-1 formulated in 5 mM phosphate
`buffer, pH 7.5 elicits a more rapid onset of action and is more
`rapidly absorbed from the subcutaneoussite than the PBS for-
`mulation. The areas under the curves A(b) and A(c) are 190+ 50
`and 340 + 140 ng/ml: min respectively. The areas under the
`curves B(b) and B(c) are 240,000 + 20,000 and 400,000 + 90,000
`pM-minrespectively. This comparison suggeststhat the bioavail-
`ability of oct-IP(7)R(26)GLP-1 is greater for the 5 mM phos-
`phate buffer formulation than for the PBS formulation.
`The data of Figure 7 C and D compare the plasmainsulin
`response and the plasma drug levels for oct-IP(7)R(26)GLP-1
`formulated in PBS and aged 24 hours at 5 °C (b) and at room
`temperature (c) prior to administration. The dynamiclight scat-
`tering data of Figure 4 showsthat these formulation conditions
`correspond to quadratic diameters of 20 nm and 50 nm forthe
`5 °C and room temperature formulations respectively. The data
`of Figure 7 C and D showthat the plasma insulin response of
`the PBS formulation aged at room temperature is negligible and
`that the plasmalevels of oct-IP(7)R(26)GLP-1 are dramatically
`reduced. The areas under the curves C(b) and C(c) are 330 + 90
`and 10 + 100 ng/ml- min respectively.
`The areas under the curves D(b) and D(c) are 250,000 +
`40,000 and 50,000 + 10,000 pM min respectively. These data
`establish that the effect of aging the oct-IP(7)R(26)GLP-1 PBS
`formulation at room temperature has been to diminish its
`absorption from the subcutaneoussite so severely that its phar-
`macodynamiceffect has been totally abolished.
`
`DISCUSSION
`
`The biophysical studies herein identify three distinct con-
`formational and self-association states of oct-IP(7)R(26)GLP-1
`
`MPI EXHIBIT 1015 PAGE 6
`
`200
`
`6000_—COO.v’lOD™DUU”
`
`Fig. 7. Comparison of the pharmacodynamics and pharmacokinetics
`of oct-IP(7)R(26)GLP-1 formulated under four different solution con-
`ditions. Plasma insulin response and oct-IP(7)R(26)GLP-1 concentra-
`tions were measured in normal, overnight-fasted dogs after
`subcutaneousinjection of 3 nmol/kg of oct-IP(7)R(26)GLP-1. Vehicle
`controls (n = 5) correspond to a subcutaneous injection of PBS (a).
`Panels A and B: Pharmacodynamics (A) and pharmacokinetics (B) of
`oct-IP(7)R(26)GLP-1 in PBS (n = 6) (b) versus 5 mM phosphate
`buffer pH 7.5 (n = 4) (c). The formulations were prepared at room tem-
`perature and administered immediately upon reconstitution of
`lyophilized peptide. Panels C and D: Pharmacodynamics (C) and
`pharmacokinetics (D) of a solution of oct-IP(7)R(26)GLP-1 prepared
`in PBS andstored at 5 °C, (n = 2) (b) and ambient room temperature
`(n = 2) (c) for 24 hoursprior to administration.
`
`4000
`
`0
`
`-50
`
`50
`
`0
`.
`Time from Injection (min)
`
`100
`
`150
`
`MPI EXHIBIT 1015 PAGE 6
`
`MPI EXHIBIT 1015 PAGE 6
`
`

`

`260
`
`Clodfelteret al.
`
`that are each stabilized by different, but pharmaceutically typical
`solution conditions. In 5 mM phosphate buffer pH = 7.5, oct-
`IP(7)R(26)GLP-1 possesses significant o-helical structure and
`exhibits an apparent molecular weight consistent with a predom-
`inantly tetrameric self-association state. In PBS, the secondary
`structure is predominantly B-sheet, andthe self-association state
`is temperature dependent. At 5 °C in PBS, oct-IP(7)R(26)GLP-1
`possesses a quadratic diameter of about 20 nm, whereasat 22 °C
`in PBS it is 50 nm andthis peptide exhibits a unique SEC reten-
`tion time. The biological data herein show that, in comparison to
`the o-helical formulation, the highly associated B-sheet form of
`oct-IP(7)R(26)GLP-1 retains a high level of subcutaneous
`bioavailability when formulated and stored at 5 °C. Evidently,
`under these formulation conditions the non-covalent B-sheet-B-
`sheet interactions are weak enough to be disrupted by dilution
`within the subcutaneousspace,thusfacilitating drug transporta-
`tion and absorption. However, these additional non-covalent
`interactionsgive rise to perturbed pharmacokinetics, presumably
`due to altered dissociation and absorption properties at the sub-
`cutaneoussite. In addition, differences in the relative transport
`properties via the lymphatic and capillary vessels of the two
`physicochemical states may also contribute to the different phar-
`macokinetics of these formulations (19). In contrast, the PBS for-
`mulation of oct-IP(7)R(26)GLP-1i at room temperature is
`biologically inactive. Our experiments suggest that under these
`solution conditions, the formation of this micelle-like form is
`essentially irreversible and dilution within the subcutaneous
`space is not sufficient to dissociate this aggregate into readily
`absorbed units. The conformation(s) of GLP-1 relevant to recep-
`tor binding is not known; however, it is noteworthy that oct-
`IP(7)R(26)GLP-1 represents an exampleofa peptide that may be
`administered in two distinctly different secondary structural
`states, both of which elicit biological responses. Evidently this
`peptide adopts the biologically active conformation upon post-
`administration equilibration or alternatively, the population of a
`unique se