NOTE
`
`Silicone Oil Induced Aggregation of Proteins
`
`LATOYA S. JONES, ALLYN KAUFMANN, C. RUSSELL MIDDAUGH
`
`Department of Pharmaceutical Chemistry, University of Kansas, Lawrence, Kansas 66047 3729
`
`Received 16 September 2004; revised 7 December 2004; accepted 10 January 2005
`
`Published online in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/jps.20321
`
`ABSTRACT: Prior to delivery to the patient, protein pharmaceuticals often come in
`contact with a variety of surfaces (e.g., syringes and stoppers), which are treated to
`facilitate processing or to inhibit protein binding. One such coating, silicone oil, has
`previously been implicated in the induction of protein aggregation. We have investigated
`the propensity of model proteins to aggregate when silicone oil is present in solution and
`find significant induction of aggregation in four proteins of various molecular weights
`and isoelectric points in the presence of 0.5% oil. The ability of silicone oil to induce
`conformational changes that might be responsible for this aggregation was also examined
`by a combination of circular dichroism (CD) and derivative UV spectroscopy. Neither
`method produces evidence of large conformational changes or alterations in thermal
`stability although in a limited number of cases some small changes suggest the possibility
`of minor structural alterations. The most probable explanation for silicone oil induced
`aggregation is that the oil has direct effects on intermolecular interactions responsible for
`protein association through interaction with protein surfaces or indirectly through
`effects on the solvent. ß 2005 Wiley Liss, Inc. and the American Pharmacists Association J
`Pharm Sci 94:918 927, 2005
`Keywords: protein aggregation; UV-vis spectroscopy; silicone oil; circular dichroism
`
`Silicone oil contamination has long been sus-
`pected of being responsible in some cases for the
`aggregation seen in certain protein pharmaceuti-
`cal preparations. Several publications in the 1980s
`implicated the release of silicone oil from dis-
`posable plastic syringes in the aggregation of
`insulin.1 – 5 The link between insulin aggregation
`and silicone oil was originally based on the
`observation that after multiple withdrawals from
`vials, patients using multi-dose preparations of
`insulin observed clouding of the solutions. In this
`regard, Chantelau et al. report a silicone oil
`contamination of up to 0.25 mg/mL in a 10 mL
`
`Correspondence to: C. Russell Middaugh (Telephone: 785
`864 5813; Fax: 785 864 5875; E mail: middaugh@ku.edu)
`
`Journal of Pharmaceutical Sciences, Vol. 94, 918–927 (2005)
`ß 2005 Wiley-Liss, Inc. and the American Pharmacists Association
`
`insulin vial when a standard procedure for filling
`1 mL siliconized syringes was performed three
`times each using 10 syringes.2 Referencing sili-
`cone oil contamination levels reported by a syringe
`manufacturer,6 Bernstein calculated that some
`of his patients who were prescribed low doses
`of insulin could have vials containing 4 mg of
`silicone oil when only 1/3 of the vial had been used.4
`The use of silicone oil is not limited to syringes. It is
`also used as a coating for porous glass vials to
`minimize protein adsorption and as a lubricant
`to prevent the conglomeration of rubber stoppers
`during filling procedures. In addition, it is the
`author’s experience that questions of silicone oil
`contamination and its potential role in protein
`aggregation arise frequently during the phar-
`maceutical development of proteins generally,
`although little information about this potential
`
`918
`
`JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 94, NO. 4, APRIL 2005
`
`Novartis Exhibit 2253.001
`Regeneron v. Novartis, IPR2021-00816
`
`

`

`problem is available in the scientific literature.
`Thus, the possibility that silicone oil
`induces
`the aggregation of proteins could have important
`implications for a wide variety of protein formula-
`tion and process development related phenomena.
`Therefore, the purpose of this investigation
`was to assess the ability of silicone oil to induce
`aggregation of a variety of proteins over a range of
`pH and to investigate whether several biophysical
`techniques that are sensitive to changes in protein
`secondary and tertiary structure can detect sili-
`cone oil
`induced conformational changes that
`might be responsible for aggregation. Four model
`proteins (ribonuclease A (RNase A), lysozyme,
`bovine serum albumin (BSA), and concanavalin A
`(ConA)) with a wide range of different physical
`characteristics were used (Table 1). The choice
`of buffer pH was based on both pharmaceu-
`tical relevance and the well-characterized pH-
`dependent oligomerization state of ConA.7 At the
`lowest pH examined (4.5), ConA is a dimer. At
`pH 6.5, it exists in dimeric and tetrameric forms.
`Above pH 7.0, ConA is primarily a tetramer.
`
`EXPERIMENTAL
`
`Materials
`
`Chicken egg white lysozyme (L7651), bovine
`serum albumin (A3294), ConA from Canavalia
`ensiformis (C7275), and ribonuclease A from
`bovine pancreas (R5125) were purchased from
`Sigma Chemical Company (St. Louis, MO). These
`proteins can be lyophilized without a high content
`lyoprotectant. Thus, all proteins were supplied
`as essentially salt-free lyophilized powders and
`were used without further purification. Silicone
`oil (S159–500) was purchased from Fisher Chemi-
`cal Company (Pittsburgh, PA). All buffer salts
`(sodium phosphate monobasic, sodium phosphate
`
`Table 1. Model Proteins
`
`SILICONE OIL INDUCED PROTEIN AGGREGATION
`
`919
`
`dibasic, sodium acetate, and sodium chloride)
`were ACS grade or higher. Solutions were pre-
`pared using distilled deionized water.
`
`Methods
`
`Preparation of Stock Solutions
`
`Three buffers (10 mM sodium phosphate, 130 mM
`NaCl, pH 6.5 and pH 7.2 and 10 mM sodium
`acetate, 130 mM NaCl, pH 4.5) were used. A stock
`solution (suspension) of 1% (w/v) silicone oil in
`buffer was prepared by combining silicone oil and
`buffer in a 50 mL polypropylene centrifuge tube
`and sonicating for 10 min in an FS30 (Fisher
`Scientific) ultrasonicating bath to create a disper-
`sion. All silicone oil suspensions were freshly pre-
`pared on the day they were used. Over the period
`of the experiments, the resulting dispersions were
`stable as judged by constant optical properties.
`Protein solutions were prepared in each buffer
`by adding buffer to an appropriate amount of lyo-
`philized protein to obtain a protein concentration
`between 1 and 2 mg/mL. The concentration of the
`protein was then determined based on its extinc-
`tion coefficient, and additional buffer was added
`to adjust the protein concentration to 1 mg/mL.
`The proteins were kept on ice or refrigerated until
`used.
`
`Turbidity
`
`Optical density measurements were used to
`monitor protein aggregation. Equal volumes of
`protein and silicone oil stock solutions were
`combined in 1.7 mL microcentrifuge tubes to
`create concentrations of protein and silicone oil
`of 0.5 mg/mL and 0.5% (w/v), respectively. The
`samples were mixed by gentle pipetting and in-
`spected visually to ensure a homogeneous appear-
`ance. Control protein samples in each buffer were
`
`No. of Various Types of Secondary Structure Units
`
`Protein
`
`Molecular Weight and (pI)12
`
`Helices
`
`b-Sheets
`
`Turns
`
`Ribonuclease A (RNase A)
`Lysozyme
`Bovine serum albumin (BSA)
`Concanavalin A (ConA)b
`
`13.7 kDa (8.8)
`14.4 kDa (11.0)
`66 kDa (4.9)
`102 kDa (tetramer) (4.5 5.5)
`
`4
`4
`60
`5
`
`12
`5
`
`26
`
`13
`11
`
`aSecondary structure content are based on the following structures deposited in the Protein Databank (RNase A: 4RAT; lysozyme:
`4LYZ; BSA: 1AO6; ConA: 1APN). http://www.rcsb.org/pdb/
`bThe secondary structure content of ConA is given for a monomer unit.
`
`JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 94, NO. 4, APRIL 2005
`
`Novartis Exhibit 2253.002
`Regeneron v. Novartis, IPR2021-00816
`
`

`

`920
`
`JONES, KAUFMANN, AND MIDDAUGH
`
`prepared similarly by substituting the appro-
`priate buffer for the 1% silicone oil suspension.
`Silicone oil buffer blanks were prepared by
`combining equal volumes of silicone oil (1%) with
`the appropriate buffer. The protein samples and
`their corresponding buffer blanks (all in quad-
`ruplet) were then transferred to an untreated
`96-well microtiter plate. This plate was placed in a
`FLUOstar Galaxy (BMG Labtechnologies, Dur-
`ham, NC) microtiter plate reader that had been
`preheated to 458C. The instrument was pro-
`grammed to record the OD360 of the wells every
`5 min for 5 h. Immediately before the beginning
`of a cycle (one round of reading all of the wells in
`the plate), the plate was gently shaken using a
`4 mm orbital displacement for 5 s. The data was
`transferred to an Excel spreadsheet for analysis.
`The values recorded for all wells containing pro-
`tein were corrected for extraneous scattering by
`subtracting the average of the wells containing
`the corresponding buffer, with or without silicone
`oil. Changes in optical density were calculated by
`subtracting the buffer corrected OD360 of the
`silicone oil-free sample from the sample contain-
`ing 0.5% silicone oil.
`
`Secondary Structure Changes and
`Tm Determinations
`
`Circular dichroism (CD) spectroscopy was used to
`determine the effect of 0.5% silicone oil on the
`secondary structure and Tm of the model proteins.
`The samples and buffer blanks were prepared as
`described above, except a protein concentration of
`0.2 mg/mL was employed. A 0.1 cm pathlength
`cell was used for data collection. Far UV CD
`spectra of the samples and buffer at 208C were
`recorded using a Jasco J-720 CD spectropolari-
`meter (Easton, MD). The spectra of appropriate
`buffers were subtracted from the spectra of the
`protein samples prior to comparison. The Tm of
`each protein sample was determined by recording
`the signal at 222 nm as the temperature was
`increased from 20 to 908C at a rate of 158C per
`hour and analyzing the resulting trace using the
`Jasco thermal denaturation analysis algorithm.
`
`Second Derivative UV Spectroscopy
`
`The effect of silicone oil (0.5%) on the tertiary
`structure of the model proteins as a function of
`temperature was investigated using 2nd deriva-
`tive analysis of UV spectra. The samples were
`prepared as described for the OD360 studies. The
`absorbance spectra were collected from 20 to 908C
`
`JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 94, NO. 4, APRIL 2005
`
`in 2.58C intervals using an Agilent 8453 UV-
`Visible spectrophotometer (Palo Alto, CA). The
`temperature of the cell holder was controlled
`using a Peltier device and the sample was equi-
`librated at each temperature for 5 min prior to
`data collection. Data were collected at 1 nm in-
`tervals with a 25 s averaging time. Spectral anal-
`ysis was performed using a splining procedure as
`previously described using ChemStation software
`(Agilent).8
`
`RESULTS AND DISCUSSION
`
`Protein Aggregation
`
`The solution parameters for the aggregation
`study were selected to permit detection of protein
`aggregates due to the presence of silicone oil over
`a relatively short time. This primarily involved
`selection of an appropriate experimental tem-
`perature and silicone oil concentration. Thus, we
`selected conditions for all proteins such that only
`small changes in optical density occurred when
`silicone oil was not present. Although we have
`attributed the increases in turbidity to protein
`aggregation,
`it is possible that the observed
`increases are caused, at least in part, by the effect
`of the protein on the silicone oil dispersion itself.
`Unfortunately, there is no obvious experimental
`method to easily distinguish between this and
`turbidity increases due to protein aggregation.
`Our assumption that protein association is res-
`ponsible for turbidity is based on the fact that
`aggregated protein can be separated by centrifu-
`gation from the protein/silicone oil emulsions and
`directly identified in the pelleted material. It is
`also important to note that by monitoring the
`OD360 of the solutions as a convenient method
`for rapidly detecting protein aggregation, soluble
`aggregates may not be detected due to their simi-
`lar size and reduced refractive indices.
`At 458C, protein aggregation was minimal to
`undetectable at all three pH values in the absence
`of silicone oil (Figure 1). By including 0.25% or
`less silicone oil in the protein samples, there was
`little to no increase in the protein optical density
`under these same conditions (data not shown)
`at short times, indicating that insoluble aggre-
`gates were not formed (although the formation of
`soluble aggregates is not precluded). At a silicone
`oil concentration of 0.5% and over a period of 5 h,
`however, the protein solutions exhibited chan-
`ges in optical density indicative of aggregation
`
`Novartis Exhibit 2253.003
`Regeneron v. Novartis, IPR2021-00816
`
`

`

`A
`
`ConA
`
`0
`
`E 0.2
`=
`""
`§
`
`<l
`
`SILICONE OIL INDUCED PROTEIN AGGREGATION
`
`921
`
`B
`0.4
`
`0
`SJ
`"'
`.2 3
`
`0
`
`E
`=
`0
`""
`6
`0
`<l
`
`0.2
`
`BSA
`
`0
`~ ' b o
`
`.4
`
`0
`~
`.2 3
`
`0
`
`0.0
`
`0
`
`60
`
`180
`120
`time (minutes)
`
`240
`
`.0
`300
`
`RNaseA
`
`C
`
`0.2
`
`0.0
`
`E
`=
`0
`""
`§
`
`<l
`
`0.4
`
`~ ~ 0.2
`""
`0 §
`=
`0.2 3
`
`0
`
`<l
`
`0.0
`
`0.0
`0
`
`D
`
`60
`
`180
`120
`time (minutes)
`
`240
`
`.0
`300
`
`lysozyme
`
`0.4
`
`0
`~
`=
`3
`
`0
`
`0
`
`60
`
`180
`120
`time (minutes)
`
`240
`
`0.0
`300
`
`0
`
`60
`
`180
`120
`time (minutes)
`
`240
`
`0.0
`300
`
`Figure 1. Effect of silicone oil (0.5%) on the optical density of model proteins at 458C.
`(A) Con A, (B) BSA, (C) RNase A, (D) lysozyme. The left y-axes refer to the white symbols
`and the right to the black. DOD360 is the optical density of the protein sample containing
`the silicone oil minus the optical density of the corresponding protein sample lacking
`silicone oil. White squares: DOD360 at pH 4.5; white triangles: DOD360 at pH 6.5; white
`circle: DOD360 at pH 7.2. The black symbols represent protein samples lacking silicone oil
`(control samples). Black squares: OD360 at pH 4.5; black triangles: OD360 at pH 6.5; black
`circles: OD360 at pH 7.2. The error bars represent the SEM of four replicates.
`
`(Figure 1). The extent and rate of the optical
`density changes are both protein and pH depen-
`dent. Although the 0.5% concentration of the oil
`employed is somewhat higher than that normally
`detected in protein pharmaceutical preparations,
`the amount employed can be considered as only
`a moderately accelerated condition, an approach
`usually deemed acceptable to replicate the longer
`times and lower silicone oil concentrations of more
`immediate interest.
`Silicone oil (0.5%) caused the OD360 of ConA at
`pH 4.5 to increase 0.19 U over the course of the first
`200 min (Figure 1A). The turbidity remained at
`this level for another hour, after which there was a
`slight decrease. At pH 6.5, the initial ConA
`aggregation occurred more rapidly and induced a
`greater change in OD360 (DOD 0.24 U in 100 min).
`At 200 min, the OD360 began a slight decline.
`Silicone oil had the least effect on ConA at pH 7.2,
`
`inducing an increase in OD360 of only 0.08 U within
`the first 50 min with the characteristic decline
`appearing at 240 min. ConA has previously
`been described as a hydrophobic protein due to
`its adsorption to hydrophobic surfaces. The de-
`crease in the effects of silicone oil on ConA at pH 7.2
`in comparison to the other pH values may be
`related to the fact that ConA is a tetramer at
`pH 7.2, potentially reducing exposure of a sub-
`unit interface that could interact with the oil.9,10
`The decline in OD360 after an initial increase
`observed for ConA at all pH values could reflect
`settling of aggregated material or the creation of
`particles with a reduced refractive index incre-
`ment (see below).
`The most dramatic silicone oil induced aggrega-
`tion occurred with BSA, as seen by the significant
`increase in optical density within the first 5 min at
`all pH values investigated (Figure 1B). The initial
`
`JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 94, NO. 4, APRIL 2005
`
`Novartis Exhibit 2253.004
`Regeneron v. Novartis, IPR2021-00816
`
`

`

`922
`
`JONES, KAUFMANN, AND MIDDAUGH
`
`change in OD360 for BSA at each pH was essen-
`tially identical (DOD360 0.13 U after 5 min). The
`change over the course of the experiment was
`pH dependent. At pH 4.5, the optical density of
`the BSA/silicone oil suspension increased slightly
`during the remainder of the incubation (<0.04 U
`increase in OD360 within the first 60 min). The
`largest increase in the turbidity of BSA was
`observed at pH 7.2, with a steep increase during
`the first 60 min followed by a slower rise for the
`duration of the incubation. The turbidity of the
`BSA/silicone oil solution at pH 6.5 increased
`steadily, albeit at a slower rate than the increase
`observed for the pH 7.2 system, following the
`initial jump at 5 min. At the end of 5 h, the OD of
`the pH 6.5 BSA/silicone oil solution was nearly
`equivalent to that at pH 7.2.
`Silicone oil induced aggregation of RNase A was
`greatest at pH 4.5 (Figure 1C). At this pH, the
`OD360 of the RNase A/silicone oil dispersion
`increased 0.2 U during the first 120 min. The
`pH 6.5 RNase A/silicone oil dispersion displayed a
`similar trend, but had a much smaller increase
`in the OD360. At pH 7.2, RNase A aggregation
`resulted in a decline in the OD360 during the first
`30 min followed by an increase that only begins
`to level off during the last 15 min of the 5 h
`incubation.
`The effect of silicone oil on the optical density of
`lysozyme differed dramatically from the other
`proteins (Figure 1D). At pH 4.5, silicone oil caused
`a steady increase in the turbidity over the 5 h,
`clearly reflecting aggregation. In contrast, the
`OD360 of the lysozyme/silicone oil mixtures at
`pH 6.5 and 7.2 decreased over extended periods
`(during 50–140 and 20–165 min, respectively).
`The pH 6.5 sample containing silicone oil ex-
`perienced a small increase in OD360, probably
`due to protein aggregation, prior to the decline.
`We attribute the decreasing OD to changes in the
`refractive index increment (dn/dc) of the protein
`particles as the protein undergoes further aggre-
`gation. Note that scattering intensity is pro-
`portional to the square of the refractive index
`increment. Thus, small decreases in the density
`of scattering particles relative to the monomeric
`protein itself can in principle lead to significant
`decreases in scattering.
`The results of this silicone oil-induced aggrega-
`tion study of several proteins reveal only limited
`information regarding general trends. The most
`obvious one is that the more hydrophobic proteins,
`BSA (classified as hydrophobic based on the well
`known presence of its apolar binding sites11) and
`
`JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 94, NO. 4, APRIL 2005
`
`ConA, had a greater tendency to aggregate than
`the relatively more hydrophilic ones (lysozyme
`and RNase A). This result was not unexpected
`and suggests that the interactions are at least
`in part apolar in nature. All proteins exhibited a
`pH-dependence in their tendency to aggregate in
`the presence of the oil. There was, however, no
`clear trend (e.g., a protein’s isoelectric point and
`the solution pH at which it experienced the largest
`(or smallest) change in optical density) to this de-
`pendence. For example, although BSA and ConA
`have similar isoelectric points (4.9 and 4.5–5.5,12
`respectively), ConA had the smallest silicone oil
`induced change in optical density at pH 7.2 while
`the pH 4.5 solution exhibited the least change for
`BSA. Although the increase in optical density was
`highest at pH 4.5 for both lysozyme and RNase A
`(two relatively small proteins with basic pIs),
`the kinetic profiles of the optical density data
`indicate that they are affected dissimilarly by the
`silicone oil.
`
`Effects of Silicone Oil on Secondary Structure
`and Thermal Stability
`
`Why do low concentrations of silicone oil cause
`proteins to aggregate? One possibility is that the
`oil structurally alters proteins resulting in aggre-
`gation competent states. To test this idea, we used
`CD to see if silicone oil induced changes in pro-
`tein secondary structure could be detected. Since
`absorption flattening is a potential problem in CD
`studies of proteins at high concentration and in
`turbid samples, it was necessary to lower the
`protein concentration to examine the effect of sili-
`cone oil on the secondary structure of the proteins.
`By using short pathlength cells and lower protein
`concentrations, undistorted CD spectra could be
`obtained in the presence of the oil. No unusual
`reduction in CD intensity or red shifts in peaks
`was observed, arguing that significant flatten-
`ing was not present. The far UV CD spectra of
`the four proteins at 208C in the presence and
`absence of silicone oil (0.5%) were indistinguish-
`able (data not shown). In most cases, the pre-
`sence of silicone oil (0.5%) also had no effect on
`the thermal unfolding temperature (Tm) of the
`proteins (Figure 2). BSA at pH 4.5 and 7.2 was
`the only protein for which silicone oil had even a
`modest statistically significant effect, with a 2–
`38C increase and decrease in Tm, respectively.
`This may at least partially reflect the existence
`of the well-known apolar binding sites on this
`protein.
`
`Novartis Exhibit 2253.005
`Regeneron v. Novartis, IPR2021-00816
`
`

`

`SILICONE OIL INDUCED PROTEIN AGGREGATION
`
`923
`
`ConA
`
`BSA
`
`70
`
`75
`
`65
`
`u
`
`0
`--; 60
`I-
`
`70
`
`u
`
`0
`--; 65
`I-
`
`55
`
`50
`
`80
`
`pH 4.5
`
`pH 6.5
`
`pH 7.2
`
`lysozyme
`
`60
`
`55
`
`70
`
`pH 4.5
`
`pH 6.5
`
`pH 7.2
`
`RNase A
`
`75
`
`u
`'";70
`I-
`
`0
`
`65
`
`u
`
`0
`--; 60
`I-
`
`65
`
`60
`
`pH 4.5
`
`pH 6.5
`
`pH 7.2
`
`55
`
`50
`
`pH 4.5
`
`pH 6.5
`
`pH 7.2
`
`Figure 2. Effect of silicone oil on circular dichroism (CD) Tm. Black bars: Samples
`lacking silicone oil. White bars: Samples contain 0.5% silicone oil. The error bars
`represent the error of the Tm determined from the nonlinear curve fit of CD ellipticity data
`as a function of temperature.
`
`Effects of Silicone Oil on Protein
`Tertiary Structure
`
`The effect of silicone oil on the tertiary structure
`of the proteins as a function of temperature and
`pH was investigated by 2nd derivative analysis
`of the proteins’ UV absorbance peaks. The use of
`derivative analysis permits peak positions in the
`presence of residual scattering to still be accu-
`rately determined since such contributions are
`manifested as only gradual background slope
`increases.13 Errors in peak position are increased,
`however, due to variations in intensity due to
`temporal effects as scattering becomes more
`intense. Note that intrinsic fluorescence spectro-
`scopy cannot be used for related analyses since
`the presence of the oil produces large spectral
`shifts due to changes in solution polarity. In con-
`
`trast, oil induced shifts in absorption spectra are
`much less (although see below). Peak assign-
`ments are based on previously established de-
`terminations and are as follows: phenylalanine
`(245–270 nm), tyrosine (265–285 nm), trypto-
`phan (265–295 nm).8,14
`The representative data presented in Figure 3
`are those for the peaks in each protein that had
`the most significant change. Although the correla-
`tion is imperfect, monitoring tertiary structure
`changes provided a better indication that the pro-
`teins were more aggregation prone in the presence
`of silicone oil. In general, ConA and BSA (Tyr/Trp
`peak shown) peak positions differed only in post
`transition regions when silicone oil was present
`(Figures 3 and 4). Interpretation of the data in
`this region is complicated by the increased error
`due to aggregation at higher temperatures with
`
`JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 94, NO. 4, APRIL 2005
`
`Novartis Exhibit 2253.006
`Regeneron v. Novartis, IPR2021-00816
`
`

`

`924
`
`JONES, KAUFMANN, AND MIDDAUGH
`
`ConApH6.5
`
`BSApH4.5
`
`-
`
`_ 268.4
`E
`:::
`:; 268.0
`
`Q,
`
`OJ)
`::: 267.6
`-9:!
`>
`e,: 267.2
`~
`
`266.8
`
`I ~
`
`,.
`
`lZ
`
`I I
`
`1
`
`ll'
`
`'I'
`
`'I'
`
`lj!
`
`\I
`
`'I'
`
`f
`
`-
`
`-280
`E
`:::
`,S 279
`oJ)
`:::
`.S:!
`~ 278
`~
`~
`
`277
`
`j
`~ i ! ! i i i i H
`
`!
`
`□
`
`□
`
`20
`
`40
`
`60
`
`80
`
`20
`
`40
`
`60
`
`80
`
`Temp. (0 C)
`
`Temp (0 C)
`
`RNase A pH 7.2
`
`lysozyme pH 7 .2
`
`-
`-
`-
`
`~ 253.8
`.:::
`OJ)
`:::
`-9:!
`Q,
`>
`e,:
`~
`
`253.4
`
`253.0
`20
`
`I I
`
`□
`
`0
`
`292.2
`
`: '. ! : ; 1 I 1 l ! I !
`
`8
`:::
`'-'
`.:::
`"S'Jl 291.8
`:::
`-9:!
`Q,
`>
`e,:
`~
`
`291.4
`
`~ ~ ~ ~ ~ ~
`Q 2 2 2
`! ! I I I •
`! ! ! !
`
`0
`
`2
`
`~ ~
`•
`I f I
`
`40
`
`60
`
`80
`
`20
`
`40
`
`60
`
`80
`
`Temp (0 C)
`
`Temp (0 C)
`
`Figure 3. Representative second derivative UV peak positions for each of the model
`proteins. Black symbols: No silicone oil; white symbols: samples contain 0.5% silicone oil.
`The error bars represent the SEM of two or three replicates.
`
`the silicone oil samples having larger error bars,
`presumably due to more intensive aggregation. As
`expected, peak positions are slightly shifted in the
`presence of the oil. In the lysozyme system, the
`sample containing silicone oil is red shifted with
`respect to the silicone oil-free sample. While this
`could, in principle, be indicative of a conforma-
`tional change that resulted in aromatic side chains
`moving to a less polar environment, it is much
`more likely that this simply reflects a decrease in
`polarity of the solvent due to the presence of the
`silicon oil. In contrast, the ca. 253 (Figure 3) and
`259 nm (data not shown) phenylalanine peaks of
`RNase A in the silicone oil-free buffers (pH 7.2 and
`
`6.5, respectively) were red shifted with respect to
`the samples containing silicone oil. This suggests
`that the environments of the phenylalanine resi-
`dues are shifted to a more polar environment in the
`presence of the silicone oil. Since the phenylala-
`nine residues in RNase A are deeply buried in the
`interior of the protein,15 this shift suggests that
`the structure has been slightly altered in the sili-
`cone oil sample, increasing the exposure of one or
`more of the phenylalanine residues to the solvent.
`Plots of peak position as function of temperature
`provide a measure of tertiary structure thermal
`stability. Slight differences in transition tempera-
`tures can be seen with certain peaks in the case of
`
`JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 94, NO. 4, APRIL 2005
`
`Novartis Exhibit 2253.007
`Regeneron v. Novartis, IPR2021-00816
`
`

`

`SILICONE OIL INDUCED PROTEIN AGGREGATION
`
`925
`
`ConA (Figure 4). Melting curves for ConA (pH 4.5)
`display transitions between 60 and 658C. In con-
`trast, ConA at pH 6.5 and 7.2 both manifest
`transitions between 65 and 708C. These pH de-
`pendent effects are in agreement with the trend
`observed in Tm calculations. That is, ConA at pH
`4.5 has a lower transition temperature than ConA
`at pH 6.5 or 7.2. The transition temperatures for
`
`ConA at the various pH values were, however,
`essentially independent of the presence of silicone
`oil, which was also true for the Tm’s calculated from
`CD measurements. Furthermore, the transition
`temperatures of the peak position curves for all of
`the proteins investigated were essentially inde-
`pendent on the presence of (0.5%) silicone oil (e.g.,
`Figure 3).
`
`peak I pH 4.5
`
`peak 2 pH 4.5
`
`peak 3 pH 4.5
`
`• • • ■ • • ■ ~ t
`
`255
`
`]: 254
`°t. 253
`=
`..!l
`"
`;. 252
`~
`
`251
`
`f
`
`~
`
`I S 261
`
`262
`
`5
`'t,, 260
`=
`..!l
`~ 259
`"
`~ 258
`
`■ ■ ■ • ■ ~ ■ ~ ~
`
`I
`
`20
`
`40
`
`60
`
`80
`
`Temp. (°C)
`
`257
`20
`
`40
`
`60
`
`80
`
`Temp. ( 0 C)
`
`S 271
`5
`-=
`jf 269
`..!l
`"
`;,
`" 267
`~
`
`265
`20
`
`I 11
`
`~ ■ ■ ■ ■ ~ ■ i
`
`~
`
`■
`
`40
`
`60
`
`80
`
`Temp. (°C)
`
`peak 1 pH 6.5
`
`peak 2 pH 6.5
`
`peak 3 pH 6.5
`
`. .
`
`V
`
`.- V V
`
`i V V f l ! I
`
`...
`
`y
`
`"
`
`?
`
`•
`
`"
`
`"
`
`....
`
`i
`
`'
`
`~ i ;
`
`i I
`y
`
`"
`
`'t'
`
`"
`
`•
`
`I
`
`255
`s
`5 254
`t
`,i 253
`"
`;,
`"
`~ 252
`
`251
`20
`
`255
`
`]: 254
`
`t
`ii 253
`"il
`;,
`"
`~ 252
`
`251
`20
`
`262
`S 261
`s
`'t,, 260
`=
`..!l
`" 259
`;,
`"
`~
`258
`
`257
`20
`
`y
`
`40
`
`60
`
`80
`
`Temp.(°C)
`
`! I ! 269
`I OJ) =
`
`-s
`..
`..!l
`"
`"
`~ 267
`
`~
`
`•
`
`'
`
`....
`
`40
`
`60
`
`80
`
`20
`
`40
`
`60
`
`80
`
`Temp. ( 0 C)
`
`Temp. (°C)
`
`peak 1 pH 7.2
`
`peak 2 pH 7.2
`
`peak 3 pH 7.2
`
`!
`i ~ i i ~ • i · Q
`
`q2
`l ;i 260
`I
`
`261
`s
`=
`OJ)
`=
`..!l
`"
`;, 259
`"
`~
`
`~ i
`
`••
`
`i
`
`•
`
`•
`
`•
`
`i I ! PI
`
`269
`
`..... 11 l
`ii 267 . ~ .
`..
`
`]: 268
`-s
`
`"il
`;,
`"
`~ 266
`
`40
`
`60
`
`80
`
`Temp. (°C)
`
`258
`20
`
`40
`
`60
`
`80
`
`Temp. (°C)
`
`265
`20
`
`40
`
`60
`
`80
`
`Temp. ( 0 C)
`
`Figure 4. Second derivative UV peak position of concanavalin A (ConA) at pH 4.5, 6.5,
`7.2. Black symbols: no silicone oil; white symbols: samples contain 0.5% silicone oil. The
`error bars represent the SEM of three replicates. Aromatic amino acids represented by
`the various peaks are as follows: Phe (peaks 1 3); Tyr (peak 4); Tyr/Trp (peak 5); and Trp
`(peak 6).
`
`JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 94, NO. 4, APRIL 2005
`
`Novartis Exhibit 2253.008
`Regeneron v. Novartis, IPR2021-00816
`
`

`

`926
`
`JONES, KAUFMANN, AND MIDDAUGH
`
`peak 4 pH 4.5
`
`' ' ' ' ' . . I I i I I
`
`40
`
`60
`
`80
`
`Temp. (° C)
`
`peak4 pH 6.5
`
`• i 1
`.
`
`•
`
`•
`
`y V V ~ I
`
`9
`
`9
`
`280
`
`e
`
`C
`:; 278
`'c,
`C
`a,
`ai
`> 276
`"' ;i:
`
`274
`20
`
`e 277
`-=-
`.c
`'c,
`C
`a,
`ai 275
`>
`"' ;i:
`
`286
`
`e
`
`C
`:; 285
`'c,
`C a,
`ai
`~ 284
`;i:
`
`283
`20
`
`285
`
`e
`-=-
`.c
`'c,
`C 284
`a,
`ai
`> "' ;i:
`
`peak 5 pH4.5
`
`peak 6 pH 4.5
`
`t ~
`• • ■ ■ 1 ■ ■ • ~ i ! I
`
`40
`
`60
`
`80
`
`Temp. (OC)
`
`e 293 .0
`-=-.c
`'c,
`C a,
`ai
`> "' ;i: 291.0
`
`29 2.0
`
`290.0
`20
`
`■ • ■ ■ ■ • ■ ~ i ' I !
`
`40
`
`60
`
`80
`
`Temp. (° C)
`
`peak6 pH 6.5
`
`peak5 pH 6.5
`
`• • • • • • • • • I I [ .
`
`9
`
`e 292
`.s
`.c
`'c,
`C
`a,
`ai 290
`> ;
`
`273+--- -~ - - -~ - - -~ - -
`20
`40
`60
`80
`
`2 8 3+ - - - -~ - - -~ - - -~ -
`20
`40
`60
`80
`
`288+ - - - -~ - - -~ - - -~ -
`40
`20
`60
`80
`
`.:
`
`Temp. (°C)
`
`Temp. (°C)
`
`peak4 pH 7.2
`
`" I
`! I ! I
`
`~ ij
`
`•
`
`i
`
`~ •
`
`; ~
`
`40
`
`60
`
`80
`
`Temp. (°C)
`
`e 277
`-=-.c
`'c,
`C
`a,
`ai 275
`
`> "' ;i:
`
`273
`20
`
`285.0
`
`e
`-=-.c 284.6
`'c,
`C a,
`ai
`> 284.2
`"' ;i:
`
`283.8
`20
`
`peak 5 pH 7.2
`
`~
`
`•
`
`•
`
`i
`
`i
`
`•
`
`•
`
`i
`
`I I "
`f I
`
`40
`
`60
`
`80
`
`Temp. (°C)
`
`Figure 4.
`
`(Continued)
`
`Temp. (°C)
`
`peak 6 pH 7.2
`
`293.0
`
`e 292.6
`-=-.c
`'c, 292.2
`C a,
`ai 291 .8
`> "' ;i:
`
`291.4
`
`291 .0
`20
`
`I 2 I ~
`!
`i
`
`!
`
`! .
`• •••
`
`i
`
`i
`
`•
`
`40
`
`60
`
`80
`
`Temp. (°C)
`
`In the discussion of the data from both the
`secondary and tertiary structure studies, no dis-
`tinction was made regarding the location of
`the protein responsible for the aggregation. The
`analytical techniques that we have employed are
`capable of reporting the biophysical properties
`of proteins in the bulk water phase, adsorbed at
`the oil/water interface, or partitioned into a bulk
`oil phase. Unfortunately, they do not distinguish
`between these environments. The data represent
`the average environment of the soluble protein.
`Thus, it is plausible that the aggregated pro-
`
`tein has undergone gross conformational changes
`that we have simply been unable to detect. Such a
`scenario would be possible for a small population of
`structurally altered, aggregation competent pro-
`tein. For example, if only 1% of the protein is
`adsorbed at the oil/water interface at any given
`time and the remaining soluble protein is structu-
`rally unaltered, relatively large perturbations in
`the structure of adsorbed protein may be experi-
`mentally reflected as only a minor change in the
`overall structure. This would be indistinguish-
`able from the situation in which a large population
`
`JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 94, NO. 4, APRIL 2005
`
`Novartis Exhibit 2253.009
`Regeneron v. Novartis, IPR2021-00816
`
`

`

`of the protein (e.g., 99%) is adsorbed at the oil/
`water interface but has only a slightly altered
`structure. Since the purpose of this study was
`to determine whether common analytical tech-
`niques could detect biophysical changes leading to
`silicone oil induced protein aggregation and not
`necessarily to elucidate the specific location of the
`altered protein, the task of distinguishing between
`proteins in the various environments was not
`pursued.
`
`CONCLUSIONS
`
`In general, methods that are commonly used to
`monitor changes in protein structure do not pro-
`vide consistent evidence that silicone oil (0.5%)
`induces major structural changes that might be
`responsible for inducing aggregation. Most com-
`monly, protein aggregation is thought to arise
`from molten globule like states of proteins. Such
`states are usually detected by tertiary (but not
`secondary) structure alterations. That is clearly
`not the case here. Thus, it is possible that more
`subtle structural changes may be involved in
`the aggregation processes as hinted by some of
`the above observations. For example, changes
`in the dynamic states of the protein, i.e., shifts
`in the distribution of their native states may
`mediate aggregation. Most importantly, a direct
`effect of the oil on the interactions that mediate
`protein/protein interactions responsible for aggre-
`gation seems likely. Whatever the molecular
`basis for the observed aggregation behavior, how-
`ever, it can clearly be minimized by reducing the
`content of silicone oil in protein pharmaceutical
`formulations.
`
`ACKNOWLEDGMENTS
`
`We would like to thank Aaron Mohs for his con-
`tributions toward the dev