Tween Protects Recombinant Human Growth Hormone against
`Agitation-Induced Damage via Hydrophobic Interactions
`
`NARENDRA B. BAM,† JEFFREY L. CLELAND,‡ JANET YANG,‡ MARK C. MANNING,§ JOHN F. CARPENTER,§
`ROBERT F. KELLEY,‡ AND THEODORE W. RANDOLPH*,|
`
`Contribution from SmithKline Beecham, King of Prussia, Pennsylvania 19406, Genentech, Inc.,
`S. San Francisco, California 94080, Center for Pharmaceutical Biotechnology, Department of Pharmaceutical
`Sciences, School of Pharmacy, University of Colorado Health Sciences Center, Denver, Colorado 80262, and
`Center for Pharmaceutical Biotechnology, Department of Chemical Engineering, University of ColoradosBoulder,
`Boulder, Colorado 80309.
`
`Received April 21, 1998. Accepted for publication September 21, 1998.
`
`Abstract 0 In the absence of surfactants, recombinant human growth
`hormone (rhGH) rapidly forms insoluble aggregates during agitation.
`The nonionic surfactant Tween 20, when present at Tween:protein
`molar ratios > 4, effectively inhibits this aggregation. Differential
`scanning calorimetry (DSC) of
`rhGH solutions showed melting
`transitions that decreased by ca. 2 (cid:176) C in the presence of Tween.
`Circular dichroism (CD) studies of the same thermal transition showed
`that the decrease is specific to the relatively high protein concentrations
`required for DSC. CD studies showed melting transitions that
`decreased with lower protein concentrations. Tween has an insig-
`nificant effect on the melting transition of rhGH at
`lower protein
`concentrations (0.18 mg/mL).
`Injection titration microcalorimetry
`showed that the interaction of Tween with rhGH is characterized by
`a weak enthalpy of binding. For comparison, interferon-g, another
`protein which has been shown to bind Tween, also shows weak
`enthalpy of binding. Fluorescent probe binding studies and infrared
`spectroscopic investigations of rhGH secondary structure support
`suggestions in the literature (Bam, N. B.; Cleland, J. L., Randolph, T.
`W. Molten globule intermediate of
`recombinant human growth
`hormone: stabilization with surfactants. Biotechnol. Prog. 1996. 12,
`801- 809) that Tween binding is driven by hydrophobic interactions,
`with little perturbation of protein secondary structure.
`
`Introduction
`
`Protective excipients are usually added to solutions of
`therapeutic proteins in order to maintain adequate shelf
`life and protect patients against possible adverse effects
`of protein aggregation and degradation. While the number
`of excipients added to a given protein formulation is
`typically small, a wide variety of such excipients has been
`used in protein products approved by the FDA and is
`available for inclusion in new formulations, providing the
`formulation developer with a huge number of possible
`excipient combinations. Achieving an optimal mixture of
`excipients is further complicated by the long-term nature
`of the formulation challenge. Desired shelf
`lives for
`therapeutic protein solutions are typically 18-24 months.
`Damage to a protein may be difficult to detect in acute
`stability studies, but appear at unacceptable levels over
`the time frame of months to years at the recommended
`storage conditions.1
`
`* Corresponding author. Tel (303) 492-4776, fax (303) 492-4341,
`email randolph@pressure3.colorado.edu.
`† SmithKline Beecham.
`‡ Genentech, Inc.
`§ University of Colorado Health Sciences Center.
`
`To facilitate the screening process for choosing an
`appropriate combination of excipients, accelerated degra-
`dation studies are often carried out.2-6 These studies aim
`to reveal instabilities of the protein in solution (and any
`protective effects of added excipients) by exposing the
`protein formulation to additional stresses, such as high
`temperatures, chaotropic agents, extreme pH’s, or agita-
`tion. For example, a common screening technique for
`effective excipients is to stress the protein solution ther-
`mally in a differential scanning calorimeter.
`In this
`technique, protein-unfolding points are measured as a
`function of excipient type and concentration. Additives
`that raise the melting point are presumed to be protein
`stabilizers, whereas those that lower the melting point are
`typically discarded from further consideration as potential
`excipients.
`Nonionic Surfactants as Protein Formulation
`ExcipientssA class of common excipients is nonionic
`surfactants. There are a number of these surfactants that
`have been used in products approved for parenteral use
`by the FDA, and they have been found to be particularly
`effective at protecting proteins against aggregation (for a
`review of surfactants in protein formulations, see ref 7).
`The mechanism(s) by which these excipients act to protect
`proteins is not clear, although several possibilities exist.
`One such mechanism is competition with the protein for
`adsorption on various interfaces, such as the air/solution
`interface or vial/solution interface,8-11 thus protecting
`against denaturation and aggregation8 at these interfaces.
`Such a mechanism, if predominate, should cause the
`concentration dependence of any protective effect against
`protein damage to be correlated with the critical micelle
`concentration (cmc) of the surfactant. A second possible
`mechanism involves specific interaction with the protein’s
`surface, with the surfactant either acting to cover hydro-
`phobic sites where aggregation could potentially occur9 or
`acting as an “artificial chaperonin”10 to catalyze refolding
`of partially unfolded, aggregation-prone protein.15-18
`If
`such a specific interaction is responsible for protective
`action of a surfactant, the concentration-dependence of the
`surfactant effect should correlate with the molar ratio of
`surfactant to protein. A third mechanism by which some
`excipients (particularly sugars) protect protein is by the
`preferential exclusion mechanism proposed by Timasheff
`and co-workers.11 Surfactants, however, are generally
`operative at concentrations where volume exclusion effects
`can be neglected.
`We have recently shown that the commonly added
`nonionic surfactants in the Tween family of polysorbates
`interact in a specific fashion with recombinant human
`growth hormone, with Tween:protein binding stoichiom-
`etries in the range of 2.5-4:1.12 In this manuscript, we
`
`1554 / Journal of Pharmaceutical Sciences
`Vol. 87, No. 12, December 1998
`
`10.1021/js980175v CCC: $15.00
`Published on Web 10/31/1998
`
`© 1998, American Chemical Society and
`American Pharmaceutical Association
`
`Ex. 1009 - Page 1 of 7
`
`AMGEN INC.
`Exhibit 1009
`
`

`

`Tween Protects Recombinant Human Growth Hormone against
`Agitation-Induced Damage via Hydrophobic Interactions
`
`NARENDRA B. BAM,† JEFFREY L. CLELAND,‡ JANET YANG,‡ MARK C. MANNING,§ JOHN F. CARPENTER,§
`ROBERT F. KELLEY,‡ AND THEODORE W. RANDOLPH*,|
`
`Contribution from SmithKline Beecham, King of Prussia, Pennsylvania 19406, Genentech, Inc.,
`S. San Francisco, California 94080, Center for Pharmaceutical Biotechnology, Department of Pharmaceutical
`Sciences, School of Pharmacy, University of Colorado Health Sciences Center, Denver, Colorado 80262, and
`Center for Pharmaceutical Biotechnology, Department of Chemical Engineering, University of ColoradosBoulder,
`Boulder, Colorado 80309.
`
`Received April 21, 1998. Accepted for publication September 21, 1998.
`
`Abstract 0 In the absence of surfactants, recombinant human growth
`hormone (rhGH) rapidly forms insoluble aggregates during agitation.
`The nonionic surfactant Tween 20, when present at Tween:protein
`molar ratios > 4, effectively inhibits this aggregation. Differential
`scanning calorimetry (DSC) of
`rhGH solutions showed melting
`transitions that decreased by ca. 2 (cid:176) C in the presence of Tween.
`Circular dichroism (CD) studies of the same thermal transition showed
`that the decrease is specific to the relatively high protein concentrations
`required for DSC. CD studies showed melting transitions that
`decreased with lower protein concentrations. Tween has an insig-
`nificant effect on the melting transition of rhGH at
`lower protein
`concentrations (0.18 mg/mL).
`Injection titration microcalorimetry
`showed that the interaction of Tween with rhGH is characterized by
`a weak enthalpy of binding. For comparison, interferon-g, another
`protein which has been shown to bind Tween, also shows weak
`enthalpy of binding. Fluorescent probe binding studies and infrared
`spectroscopic investigations of rhGH secondary structure support
`suggestions in the literature (Bam, N. B.; Cleland, J. L., Randolph, T.
`W. Molten globule intermediate of
`recombinant human growth
`hormone: stabilization with surfactants. Biotechnol. Prog. 1996. 12,
`801- 809) that Tween binding is driven by hydrophobic interactions,
`with little perturbation of protein secondary structure.
`
`Introduction
`
`Protective excipients are usually added to solutions of
`therapeutic proteins in order to maintain adequate shelf
`life and protect patients against possible adverse effects
`of protein aggregation and degradation. While the number
`of excipients added to a given protein formulation is
`typically small, a wide variety of such excipients has been
`used in protein products approved by the FDA and is
`available for inclusion in new formulations, providing the
`formulation developer with a huge number of possible
`excipient combinations. Achieving an optimal mixture of
`excipients is further complicated by the long-term nature
`of the formulation challenge. Desired shelf
`lives for
`therapeutic protein solutions are typically 18-24 months.
`Damage to a protein may be difficult to detect in acute
`stability studies, but appear at unacceptable levels over
`the time frame of months to years at the recommended
`storage conditions.1
`
`* Corresponding author. Tel (303) 492-4776, fax (303) 492-4341,
`email randolph@pressure3.colorado.edu.
`† SmithKline Beecham.
`‡ Genentech, Inc.
`§ University of Colorado Health Sciences Center.
`
`To facilitate the screening process for choosing an
`appropriate combination of excipients, accelerated degra-
`dation studies are often carried out.2-6 These studies aim
`to reveal instabilities of the protein in solution (and any
`protective effects of added excipients) by exposing the
`protein formulation to additional stresses, such as high
`temperatures, chaotropic agents, extreme pH’s, or agita-
`tion. For example, a common screening technique for
`effective excipients is to stress the protein solution ther-
`mally in a differential scanning calorimeter.
`In this
`technique, protein-unfolding points are measured as a
`function of excipient type and concentration. Additives
`that raise the melting point are presumed to be protein
`stabilizers, whereas those that lower the melting point are
`typically discarded from further consideration as potential
`excipients.
`Nonionic Surfactants as Protein Formulation
`ExcipientssA class of common excipients is nonionic
`surfactants. There are a number of these surfactants that
`have been used in products approved for parenteral use
`by the FDA, and they have been found to be particularly
`effective at protecting proteins against aggregation (for a
`review of surfactants in protein formulations, see ref 7).
`The mechanism(s) by which these excipients act to protect
`proteins is not clear, although several possibilities exist.
`One such mechanism is competition with the protein for
`adsorption on various interfaces, such as the air/solution
`interface or vial/solution interface,8-11 thus protecting
`against denaturation and aggregation8 at these interfaces.
`Such a mechanism, if predominate, should cause the
`concentration dependence of any protective effect against
`protein damage to be correlated with the critical micelle
`concentration (cmc) of the surfactant. A second possible
`mechanism involves specific interaction with the protein’s
`surface, with the surfactant either acting to cover hydro-
`phobic sites where aggregation could potentially occur9 or
`acting as an “artificial chaperonin”10 to catalyze refolding
`of partially unfolded, aggregation-prone protein.15-18
`If
`such a specific interaction is responsible for protective
`action of a surfactant, the concentration-dependence of the
`surfactant effect should correlate with the molar ratio of
`surfactant to protein. A third mechanism by which some
`excipients (particularly sugars) protect protein is by the
`preferential exclusion mechanism proposed by Timasheff
`and co-workers.11 Surfactants, however, are generally
`operative at concentrations where volume exclusion effects
`can be neglected.
`We have recently shown that the commonly added
`nonionic surfactants in the Tween family of polysorbates
`interact in a specific fashion with recombinant human
`growth hormone, with Tween:protein binding stoichiom-
`etries in the range of 2.5-4:1.12 In this manuscript, we
`
`1554 / Journal of Pharmaceutical Sciences
`Vol. 87, No. 12, December 1998
`
`10.1021/js980175v CCC: $15.00
`Published on Web 10/31/1998
`
`© 1998, American Chemical Society and
`American Pharmaceutical Association
`
`Ex. 1009 - Page 2 of 7
`
`

`

`explore the mechanisms of binding and the implications
`for such specific binding on the solution stability of
`recombinant human growth hormone.
`In addition, we
`evaluate the capability of accelerated testing via dif-
`ferential scanning calorimetry to predict the protective
`ability of Tween against agitation-induced aggregation
`processes.
`
`Materials and Methods
`rhGH Stability against Aggregation-Induced Damages
`A 1 mL amount of a 5 mg/mL solution of rhGH (Genentech, Inc.)
`in 10 mM sodium citrate, pH 6.0, and various concentrations of
`Tween 20, Tween 40, or Tween 80 (Pierce) were sealed and capped
`in 3 mL glass vials. The vials were shaken on a Glass-Col shaker
`at 120 shakes/min. Samples were taken every few hours and
`analyzed for aggregated protein by size exclusion chromatography.
`Insoluble aggregates were removed by filtering through a 0.22 (cid:237)M
`filter.
`rhGH in the filtrate was analyzed by size exclusion
`chromatography, using a Tosoh TSK 2000SW column (30 cm long)
`with a running buffer of 10 mM sodium citrate, pH 6.0, a flow
`rate of 1 mL/min, and protein detection by absorbance at 280 nm.
`Surface TensiometrysThe critical micelle concentration (cmc)
`for Tween 20 in 10 mM sodium citrate, pH 6.0, and also in 10mM
`sodium citrate, pH 6.0, with 45 mg/ml mannitol, and 0.2% phenol
`was determined using a Rame-Hart pendant drop apparatus.
`Pendant drops were analyzed using a Panasonic CCD camera
`fitted with a Nikon micro NIKKOR 55 mm lens and a 27.5 mm
`extension tube. Digital images of pendent drops were recorded
`using an image capture card and associated software (Video Image
`100, Scion Corp.) on a Macintosh IIfx computer.
`Images of
`pendant drops were recorded after 15-20 min of equilibration.
`Surface tensions were calculated by fitting calculated profiles to
`images as described previously.13
`Differential Scanning CalorimetrysExperiments to mea-
`sure the thermally induced unfolding transition of rhGH were
`conducted on a Hart microcalorimeter. Solutions containing 3 mg/
`mL rhGH in 10 mM sodium citrate, pH 6.0, were loaded into a 1
`mL ampule. A second ampule was filled with buffer alone. Scans
`were recorded from 25 to 95 °C, at a scan rate of 60 °C/h. The
`thermogram for the buffer was subtracted from that for the
`protein-containing solution. Experiments were also conducted
`using buffer solutions containing Tween. In all cases, the melting
`temperature of rhGH was identified as the temperature at the
`peak of the unfolding endotherm. It should be noted that the
`unfolding was not completely reversible. Rapidly cooled samples
`exhibited an endotherm that was reduced by about 20% upon
`rescanning.
`Circular DichroismsCircular dichroism experiments were
`performed on an AVIV 62DS CD spectrometer equipped with
`variable temperature cell holder. rhGH in 10 mM sodium citrate,
`pH 6.0, was filtered through a 0.22 (cid:237)m filter to remove any
`particulates, and the final rhGH concentration was determined
`by UV absorbance at 278 nm ((cid:15) ) 18 890 L/mol cm). Thermally
`induced unfolding of rhGH was monitored by scanning from 45 to
`95 °C at a scan rate of 60 °C/h. Loss in the R-helix content was
`monitored at 222 nm.14 Thermal melting curves for rhGH were
`obtained by converting the molar residual ellipticity values to
`fraction of rhGH unfolded using the following expression:
`
`f
`D )
`
`(yi - y0)
`(yD - y0)
`
`where fD is the fraction of protein in the unfolded state, yi is the
`molar residual ellipticity at a given temperature, and y0 and yD
`are the molar residual ellipticity of the folded state and unfolded
`state, respectively.
`Fluorescent Probe Binding StudiessFluorescence measure-
`ments were performed on a SLM 48000S spectrofluorimeter, at
`an excitation wavelength of 410 nm and a detection wavelength
`of 500 nm. Three hydrophobic probe molecules were used:
`1-anilinonaphthalene-8-sulfonic acid (1,8-ANS), 1,1¢ -bis(4-anilino)-
`naphthalene-6-sulfonic acid (2,6-ANS), and 2-(p-toluidinyl)naph-
`thalene-6-sulfonic acid (2,6-TNS), all from Sigma Chemical, and
`used as received. Stock solutions of both probe (0.3 M) and protein
`(300 (cid:237)M) were prepared in buffer (for rhGH: 45 mg/mL mannitol,
`
`0.25% phenol in 10 mM sodium citrate, pH 6.0; for recombinant
`human interferon-(cid:231) (rhIFN-(cid:231), Genentech, Inc.): 40 mg/mL man-
`nitol in 5 mM sodium succinate, pH 5.0). The concentrated probe
`was titrated into 2 mL of buffer and intensity plotted against probe
`concentration to verify that the probe was soluble under all
`experimental conditions. Probe and protein solutions were mixed
`(with additional buffer as necessary) to form dilute protein
`solutions (1-4 (cid:237)M) with excess of probe (10-100 (cid:237)M), or vice versa,
`with dilute probe solutions (1-4 (cid:237)M) and an excess of protein (20-
`40 (cid:237)M). Fluorescence for each solution was measured at 25 °C.
`For studies with [protein] . [fluorescent probe], a plot of the
`inverse fluorescence intensity 1/I versus the inverse of the molar
`fluorescent probe concentration at constant protein concentrations
`gives a linear plot whose x-intercept is the negative inverse
`dissociation constant, -1/Kd. Three different protein concentra-
`tions were used, and the average value of the three intercepts was
`computed. A similar plot of 1/I versus 1/[protein] at constant
`fluorescent probe concentration gives a linear plot with an
`x-intercept of the negative binding stoichiometry divided by the
`dissociation constant, -n/Kd.15
`Injection Titration MicrocalorimetrysMicrocalorimetry
`experiments were performed in duplicate on a Microcal, Inc.
`Omega titration calorimeter. The injection syringe was filled with
`a concentrated stock containing 12.8 mM Tween 20 or Tween 80.
`Forty injections of 5 (cid:237)L each were made into the 1.4 mL
`calorimeter cell, which contained either 100 mM phosphate buffer,
`pH 6.5, or the same buffer containing 0.175 mM protein. The cell
`was stirred at 100 rpm and was thermally equilibrated at 25 °C
`for 30 min until a smooth baseline was obtained, and injections
`were initiated. The thermogram was allowed to return to baseline
`between injections (ca. 5 min between injections). Peaks were
`integrated using software from Microcal to yield the enthalpy
`generated for each injection.
`Infrared SpectroscopysSpectra in the amide I region (1600
`to 1700 cm-1) were recorded on a Nicolet Magna 550 FTIR
`spectrometer equipped with a DTGS detector, 256 scans being
`recorded for each spectrum. The interferogram was collected in
`single beam mode, with a 4 cm-1 resolution. A 10 mg/mL solution
`of rhGH in sodium citrate buffer, pH 6.0, was filtered through a
`0.22 (cid:237)m filter and placed in a cell with CaF2 windows and a 6 (cid:237)m
`spacer. The system was purged with N2 before recording the
`spectrum. A buffer blank was recorded under identical conditions,
`but without protein, and subtracted. Water vapor corrections were
`made as previously described.16 The experiments were repeated
`after equilibrating rhGH with a 10:1 molar ratio of Tween 40 for
`2 h; buffer containing the same concentration of Tween 40 was
`used as a blank. Second derivative spectra were calculated using
`the derivative function of Nicolet Omnic software, after first
`smoothing with a 7-point smoothing routine to remove white noise.
`Spectra were normalized as reported previously.17
`
`Results and Discussion
`Surfactant Effects on rhGH StabilitysIndustrial
`protein formulations may be exposed to stresses due to
`agitation for a number of reasons. During filling of vials
`or during shipping, for example, the protein may be
`exposed to stresses at the air-solution interface. To test
`the effect of surfactant on the tendency of rhGH to
`aggregate during agitation, we conducted shaker vial
`studies at various concentrations of Tween. Results from
`these shaker vial studies showed that Tween 20 is the most
`effective for reducing aggregation of rhGH. As shown in
`Figure 1a, rhGH shaken without the addition of Tween
`aggregates rapidly resulting in nearly complete loss of
`monomeric protein within 10 h. Similar results are seen
`at Tween 20:rhGH molar ratios of 1:1. However, at a
`stoichiometric ratio of 2:1 Tween 20:rhGH, the half-life of
`monomeric rhGH is increased approximately 6-fold. In-
`creasing the stoichiometric ratio still further to 4:1 or 7.2:1
`confers essentially complete protection against agitation-
`induced aggregation. In contrast, Tween 40 and Tween
`80 both protect rhGH against aggregation, but to a lesser
`degree than does Tween 20. As shown in Figure 1, addition
`of Tween 40 or Tween 80 at surfactant:protein ratios of up
`
`Journal of Pharmaceutical Sciences / 1555
`Vol. 87, No. 12, December 1998
`
`Ex. 1009 - Page 3 of 7
`
`

`

`Table 1sEffect of GuHCl and Tween on RhGH Melting Temperatures
`Obtained by Various Techniques. Representative Experiments
`Carried out in Triplicate Suggest That Experimental Error on Both
`DSC and CD of Approximately – 0.2 (cid:176) C at a 95% Confidence Interval
`
`DSC, 3 mg/mL
`rhGH, (cid:176) C
`
`CD, 3 mg/mL
`rhGH, (cid:176) C
`
`CD, 1 mg/mL
`rhGH, (cid:176) C
`
`CD, 0.18 mg/mL
`rhGH, (cid:176) C
`
`condition
`buffer
`2 M GuHCl
`10:1 Tween 20
`10:1 Tween 40
`10:1 Tween 80
`
`79.2
`80.0
`77.0
`77.0
`77.1
`
`79.1
`79.0
`78.0
`77.8
`77.9
`
`80.5
`NS
`80.5
`NS
`NS
`
`88.8
`86.1
`89.4
`89.4
`89.3
`
`apparent cmc’s in the presence of rhGH are shifted to lower
`Tween concentrations.12 The concentration of Tween 20
`required to protect the protein against aggregation (about
`1 mM) is much higher than cmc’s measured in the absence
`of protein or apparent cmc’s measured in the presence of
`rhGH. Thus, the concentrations of surfactant required to
`protect the protein from agitation-induced aggregation do
`not appear to correlate with the cmc of the surfactant,19
`but rather with a specific stoichiometry that is in reason-
`able agreement with previously reported Tween:rhGH
`binding stoichiometries.12
`Characterization of the Interactions of Tween with
`the Surface of rhGHsBased on the aggregation study
`reported above and our previously reported Tween:rhGH
`binding measurements,12 a protective mechanism that
`involves specific binding of Tween to rhGH seems likely.
`As mentioned above, there are at least two possible
`mechanisms by which specific binding may protect proteins
`against aggregation. The native state may be thermody-
`namically stabilized upon binding, resulting in a lesser
`tendency of the protein to partially expose interior and
`presumably “stickier”, more aggregation-prone hydrophobic
`residues. Alternatively, surfactant binding may inhibit
`aggregation sterically by blocking hydrophobic contact sites
`on the surface of either the native or partially unfolded
`protein. To evaluate the first possibility, we conducted
`thermal melting studies by differential scanning calorim-
`etry and circular dichroism spectroscopy.
`Results of the thermal melting studies are presented in
`Table 1. At rhGH concentrations of 3 mg/mL, DSC scans
`show an apparent melting endotherm with a maximum
`occurring at 79.1 °C. The addition of Tween 20, Tween 40,
`or Tween 80 at a surfactant:protein molar ratio of 10:1
`causes the apparent melting temperature to be depressed
`by approximately 2 °C. This result, taken by itself, might
`suggest that Tweens destabilize the native structure of
`rhGH. However, additional experiments gave the surpris-
`ing result that addition of 2 M guanidinium hydrochloride
`(GuHCl), a known protein destabilizer, increases the melt-
`ing temperature by nearly a degree (Table 1).
`Melting temperature results obtained by DSC were
`verified by monitoring the unfolding transition of rhGH by
`CD spectroscopy. Melting curves obtained at rhGH con-
`centrations of 3 mg/mL matched those obtained by DSC.
`As shown in Table 1, addition of 10:1 Tween 40 causes a
`slight depression in the melting temperature of rhGH.
`Circular dichroism is a more sensitive technique than DSC,
`enabling spectra to be recorded at protein concentrations
`an order of magnitude lower than is practical with con-
`ventional DSC.
`It is interesting to note that at lower
`concentrations (0.18 mg/mL) of rhGH, the results differ
`strikingly from results at high concentrations. At a rhGH
`concentration of 0.18 mg/mL, the melting temperature is
`nearly 10 °C higher than that observed at a concentration
`of 3 mg/mL. Addition of 10:1 Tween 40:rhGH at this lower
`protein concentration slightly increases the melting tem-
`
`Figure 1sPercent monomeric rhGH retained in solution as a function of time
`in shaker vial studies. Monomer values obtained by size exclusion HPLC.
`Filled circles, no Tween; squares, Tween 20:rhGH 1:1; diamonds, Tween 20:
`rhGH 2:1; triangles, Tween 20:rhGH 4:1; open circles, Tween 20:rhGH 7.2:1.
`Curves are included simply to guide the eye; no mechanistic implications are
`suggested.
`
`Figure 2sCritical micelle concentration (cmc) for Tween 20 in 10 mM sodium
`citrate, pH 6.0, 45 mg/mL mannitol, as determined by pendant drop surface
`tensiometry. The cmc occurs at
`the break in slope of
`the curve, or
`approximately 100 mM Tween 20.
`
`to 5:1 increases the half-life of monomeric rhGH in solution
`by about 4-fold over the half-life in the absence of surfac-
`tant. At similar molar ratios, addition of PEG 3350 confers
`no protection against aggregation (data not shown).
`It
`should be noted that the concentrations of Tween used in
`this experiment (even at a 1:1 Tween:rhGH molar ratio,
`where the Tween concentration was 225 (cid:237)M) are above the
`reported cmc values for Tweens, which are reported as 59
`(cid:237)M and 12 (cid:237)M for Tween 20 and Tween 80, respectively.18
`Because cmc’s can be affected by buffer salts and other
`formulation ingredients, we measured the operative cmc
`in this formulation by surface tensiometry. Surface tension
`measurements presented in Figure 2 showed that the cmc
`for Tween 20 in the formulation buffer that we used for
`rhGH is about 100 (cid:237)M, or about twice the reported value
`in distilled water, 59 (cid:237)M. Comparison of measured cmc’s
`in 10 mM sodium citrate, pH 6.0, in the presence and
`absence of 45 mg/mL mannitol, 0.2% phenol, were not
`significantly different at the 95% confidence level (N ) 3,
`data not shown). Cmc’s in the presence of protein are
`difficult to measure, but previous work suggests that
`
`1556 / Journal of Pharmaceutical Sciences
`Vol. 87, No. 12, December 1998
`
`Ex. 1009 - Page 4 of 7
`
`

`

`perature. At 1 mg/mL rhGH, the melting temperature is
`intermediate between the temperature observed at con-
`centrations of 3 mg/mL or 0.18 mg/mL, and Tween 40
`addition at a molar ratio of 10:1 Tween 40:rhGH causes
`no discernible change in the melting temperature.
`It
`should also be noted that CD scans in the far UV showed
`that the melting of monomer (at protein concentrations
`<0.2 mg/mL) is fully reversible. On the basis of the CD
`results, it appears that the DSC results taken at higher
`protein concentration (where unfolding was not reversible)
`do not reflect true melting points of the monomeric protein,
`but perhaps rather the effect of aggregation on the appar-
`ent unfolding temperature of the protein. Thus, addition
`of GuHCl, which presumably would disfavor such ag-
`gregates, causes an apparent increase in the melting point.
`If the mechanism by which Tween protects growth
`hormone against agitation-induced aggregation is by sta-
`bilizing the protein’s native state, a significant increase in
`the melting temperature in the presence of Tween might
`be expected. Even at low protein concentrations, the
`increase in melting temperature in the presence of Tween
`is minor. Thus, taken together, the DSC and CD results
`provide little evidence that the mechanism by which Tween
`inhibits agitation-induced aggregation is by native state
`stabilization. In contrast, previously published free ener-
`gies of unfolding measured at room temperature in chao-
`trope-induced denaturation studies using low protein
`concentrations (0.1 mg/mL rhGH) showed that Tween does
`stabilize the native state of rhGH by 1-4 kcal/mol.20 A
`possible explanation for the discrepancy between the effects
`of Tween on thermal melting which indicate little or no
`structural stabilization, and the results from chaotrope-
`induced denaturation studies which showed significant
`structural stabilization, is that high temperature alters the
`relative interactions of Tween with the native and unfolded
`states of rhGH. Thus, neither the CD nor the DSC studies,
`which measure a thermal transition at 80-90 °C, may be
`of particular relevance to the formulation problem if the
`mechanism involves a hydrophobic interaction between
`excipient and protein that would be altered at elevated
`temperatures.21
`The apparently anomalous results from the DSC study
`which requires relatively high protein concentrations (3 mg
`/ml) warrant a warning note for the use of DSC for
`accelerated testing of protein formulations. Had DSC been
`used as the primary screening tool in design of these
`formulations, the observed decrease in the melting point
`of rhGH upon addition of Tween might have caused Tween
`to be discarded from further consideration, despite its
`efficacy in preventing aggregation. Further, the concentra-
`tion-dependency of DSC techniques for determining ther-
`mal melting transitions for rhGH has been overlooked in
`some studies.27
`An alternative mechanism for rhGH stabilization against
`agitation-induced aggregation is that the binding of Tween
`acts to block “sticky” hydrophobic patches on the native
`protein’s surface. We conducted fluorescent probe binding
`studies to characterize the hydrophobicity of rhGH and to
`check for possible sites that could be favorable for protein
`aggregation and/or surfactant binding. For comparison, we
`also conducted fluorescent probe binding studies on rhIFN-
`(cid:231), a protein with a more hydrophobic surface than rhGH
`as determined by both the hydrophobicity scale of Von
`Heine22 and that of Manavalan and Ponnuswamy.23 Bind-
`ing stoichiometries for various fluorescent probes to rhGH
`and rhIFN-(cid:231) are shown in Table 2. Solubility limitations
`for the ANS-protein complex prevented analysis of the
`binding of 1,8-ANS to rhIFN-(cid:231) and generally limited the
`accuracy of the technique for all of the other rhIFN-(cid:231):probe
`pairs.
`It is interesting to note that the dissociation
`
`Table 2sBinding Stoichiometries and Dissociation Constants for
`rhGH and rhIFN-(cid:231) with Various Fluorescent ANS Derivatives
`
`protein
`
`RhGH
`
`rhIFN-(cid:231)
`
`probe
`bis- ANS
`1,8- ANS
`2,6- ANS
`2,6- TNS
`bis- ANS
`1,8- ANS
`
`binding
`stoichiometry
`
`2.3
`4.0
`0.6
`2.2
`9
`NS
`
`Kd ((cid:237)M)
`100
`590
`310
`510
`40
`9.1
`
`constants for ANS probes to rhGH are about 1 order of
`magnitude higher than the dissociation constants for
`rhIFN-(cid:231) binding to bis-ANS probes. While ANS:protein
`dissociation constants are weak for both proteins, consis-
`tent with the binding being driven by hydrophobic interac-
`tions, ANS associates more strongly with rhIFN-(cid:231), which
`is the more hydrophobic of the two proteins. In addition,
`the binding stoichiometry for bis-ANS is greater for rhIFN-(cid:231)
`than for rhGH, consistent with previous EPR studies12 of
`Tween 40 binding to rhIFN-(cid:231) and rhGH, which showed
`that approximately twice as many moles of Tween 40 bind
`to rhIFN-(cid:231) as to rhGH (6 versus 2.5-3).
`Binding Enthalpies for Tween 20 to rhGHsThe
`weak binding of ANS to the surface of rhGH suggests that
`the interaction of hydrophobic probes such as ANS is driven
`by hydrophobic interactions. To test the hypothesis that
`Tween-rhGH interactions are also the result of hydrophobic
`interactions, we turned to injection titration microcalorim-
`etry. The calorimetric detection of surfactant binding to
`proteins is complicated by thermal events associated with
`surfactant micelles. Heats of dilution for monomer and
`micelle, as well as the heat of micellization, must be taken
`into account. To measure the heat of binding to protein,
`concentrated surfactant (ca. 20 times the crit