`Vol. 93, PJ?· 5460-5465, May 1996
`Applied B10logical Sciences
`
`Irreversible inactivation of interleukin 2 in a pump-based
`delivery environment
`(interleukin 2 physical stability/protein delivery/protein-surface interactions)
`
`STELIOS T. TZANNIS*, WILLIAM J.M. HRUSHESKYt, PATRICIA A. WooDt, AND TODD M. PRZYBYCIEN**
`*The Howard P. lsermann Department of Chemical Engineering, Applied Protein Biophysics Laboratory, Rensselaer Polytechnic Institute, Troy, NY 12180-3590;
`and toepartment of Medical Oncology, Stratton Veterans Adminstration Medical Center, Albany, NY 12208
`
`Communicated by Robert Langer, Massachusetts Institute of Technology, Cambridge, MA, October 23, 1995
`
`The physical stability of pharmaceutical pro(cid:173)
`ABSTRACT
`teins in delivery environments is a critical determinant of
`biological potency and treatment efficacy, and yet it is often
`taken for granted. We studied both the bioactivity and physical
`stability of interleukin 2 upon delivery via continuous infu(cid:173)
`sion. We found that the biological activity of the delivered
`protein was dramatically reduced by =90% after a 24-hr
`infusion program. Only a portion of these losses could be
`attributed to direct protein deposition on the delivery sur(cid:173)
`faces. Analysis of delivered protein by size exclusion chroma(cid:173)
`tography gave no indication of insulin-like, surface-induced
`aggregation phenomena. Examination of the secondary and
`tertiary structure of both adsorbed and delivered protein via
`Fourier-transform infrared spectroscopy, circular dichroism,
`and fluorescence spectroscopy indicated that transient sur(cid:173)
`face association of interleukin 2 with the catheter tubing
`resulted in profound, irreversible structural changes that were
`responsible for the majority of the biological activity losses.
`
`The physical stability of formulated therapeutic proteins in
`delivery environments is a necessity for successful short- or
`long-term delivery via implantable or portable pumps. Protein
`delivery device surface interactions may induce conforma(cid:173)
`tional changes and aggregation, which may result in the
`inactivation of the delivered protein and therefore compro(cid:173)
`mise the intended therapeutic benefit.
`Protein-surface interaction phenomena have been studied
`extensively over the past 15 years. However, the focus has
`typically been to assess implantable materials for hemocom(cid:173)
`patibility (1). A notable exception is given by the extensive
`studies of the delivery of insulin via micropumps, which have
`shown surface-induced aggregation and consequent inactiva(cid:173)
`tion phenomena (2, 3). It has also been demonstrated that
`administration of low concentrations of interleukin (IL )-1 J3 via
`a syringe-pump system results in adsorptive losses of 20-80%
`(4). Our previous studies revealed =99% losses in the biolog(cid:173)
`ical activity of formulated IL-2 after 24-hr exposure to com(cid:173)
`mercial silicone rubber tubing in a recirculating system (S.T.T.,
`J.M. Hrushesky, P.A.W., & T.M.P., unpublished work). Sim(cid:173)
`ilarly, reductions of 80-90% of the biological activity of a Food
`and Drug Administration-approved IL-2 formulation have
`been reported during delivery via polyethylene catheter tubing
`(6). Addition of human serum albumin or surfactants to the
`reconstitution medium can significantly decrease adsorptive
`losses (2, 4, 7), although in the case of IL-2, the effectiveness
`of human serum albumin as a passivating agent is limited (6).
`In the present effort, we assessed the nature and extent of
`interactions of a commercial IL-2 formulation with a portable
`infusion pump/catheter tubing system. We also explored the
`impact of these interactions on the active delivery profile and
`established a structure-function relationship for both the
`
`adsorbed and delivered protein via a variety of spectroscopic
`techniques. IL-2 is a potent anti-tumor and immune modulator
`that has been administered using both high-dose bolus and
`continuous infusion strategies with promising clinical results in
`the treatment of cancer and acquired immune deficiency
`syndrome (AIDS). Moreover, IL-2 is representative of the
`hematopoietic growth factor family, with high sequence and
`structural homology with other therapeutic proteins such as
`granulocyte colony-stimulating factor, granulocyte macro(cid:173)
`phage colony-stimulating factor, erythropoietin, and growth
`hormone. The IL-2 case may serve as a paradigm for the
`optimal delivery of these therapeutic proteins.
`
`MATERIALS AND METHODS
`Materials. Recombinant human interleukin-2 (IL-2) ex(cid:173)
`pressed in E. coli (RU 49637, Batch no. BV 23621-066A;
`Roussel-Uclaf) was obtained as a monomeric, powdered clin(cid:173)
`ical formulation (1.2 ± 0.6 x 107 units/mg) with 0.01 mg of
`isopropanol per 0.5 mg of IL-2. The powder was stored at
`- 20°C before use and was reconstituted in 10 mM phosphate
`buffer (pH 7.4) with 150 mM NaCl at 25°C. The delivery
`system consisted of a programmable, portable, piston(cid:173)
`operated clinical infusion system (Panomat V-5; Disetronic
`Medical Systems, Plymouth MN) connected to a 60-cm length
`of 1.0-mm i.d. radiopaque medical grade silicone rubber
`catheter tubing (Dow-Corning) provided by Medtronic (Min(cid:173)
`neapolis). The tubing was cleaned, steam-sterilized, and hy(cid:173)
`drated with water for injection-grade water before use.
`Delivery Apparatus and Protocol. The borosilicate glass
`reservoir of the pump was loaded under sterile conditions with
`freshly reconstituted IL-2 solution at a concentration of either
`0.1 or 0.05 mg per ml and then connected to the catheter
`tubing. The entire experimental system was placed in a 37°C
`incubator. Following actual clinical protocols, the tubing was
`primed with the protein solution at the initiation of the
`experiment. The solution was pumped at 100 µ,L per hr and
`collected in 300-µ,l capacity sterile vials (flow system). Under
`these conditions the Reynolds number was 0.05, the wall shear
`rate was 0.28 sec- 1, and the Peclet number was 0.5. At 2, 6, and
`24 hr, a 10-cm portion of the tubing was removed from the
`collection vial end of the system and subjected to Fourier(cid:173)
`transform infrared (FTIR) analysis. In order to control for
`exposure at 37°C and adsorption on the pump reservoir walls,
`a similar reservoir loaded with IL-2 solution of the same
`volume and concentration was placed in the incubator (non(cid:173)
`flow system) and similarly sampled over 24 hr.
`Delivered Protein Characterization. Bioactivity. IL-2 bio(cid:173)
`logical activity was assessed via a proliferation assay based on
`the IL-2-dependent mouse spleen cell line, CTLL-2; this assay
`is described in detail elsewhere (Tzannis et al., unpublished
`work).
`
`The publication costs of this article were defrayed in part by page charge
`payment. This article must therefore be hereby marked "advertisement" in
`accordance with 18 U.S.C. §1734 solely to indicate this fact.
`
`Abbreviations: IL, interleukin; FTIR, Fourier-transformed infrared.
`tro whom reprint requests should be addressed.
`
`5460
`
`Downloaded by guest on July 11, 2021
`
`Novartis Exhibit 2177.001
`Regeneron v. Novartis, IPR2021-00816
`
`
`
`Applied Biological Sciences: Tzannis et al.
`
`Proc. Natl. Acad. Sci. USA 93 (1996)
`
`5461
`
`Protein concentration. The bicinchoninic acid protein deter(cid:173)
`mination method (Pierce) was used to quantitate the concen(cid:173)
`tration of delivered IL-2 over time.
`Secondary structure. Solution phase IL-2 secondary structure
`was characterized via far-UV CD, using a previously reported
`spectral acquisition and manipulation protocol (8) in conjunc(cid:173)
`tion with the structure fitting technique of Sreerama and
`Woody (9).
`Tertiary structure. Fluorescence spectroscopy was used to
`examine the environment of Trp121, the lone tryptophan of
`IL-2. Steady-state fluorescence measurements were per(cid:173)
`formed on a Perkin-Elmer LS50B Luminescence spectropho(cid:173)
`tometer equipped with a Neslab RTE-111 thermostat (Neslab
`Instruments, Portsmouth, NH) set at 25°C. The samples were
`excited at 295 nm (10), and five emission spectra, recorded
`between 300-450 nm with a 1-nm band width, were collected
`and averaged; the excitation and emission slits were set at 4.5
`nm. We also used iodide and acrylamide in static quenching
`experiments to probe the solvent accessibility of Trp121 (10-
`12). Iodide was added from a 5 M stock solution in the
`reconstitution buffer along with 1 mM Na2S2O3 to prevent lz
`formation, which could cause precipitation of the protein (13).
`Similarly, acrylamide was added from a 5 M stock solution also
`in the reconstitution buffer. The emission spectra of the
`protein solutions were corrected for background scatter, di(cid:173)
`lution and, in the quenching experiments, absorbance by KI
`and acrylamide (13, 14).
`Aggregation state. The extent of aggregation was determined
`via size exclusion HPLC analysis as follows: the stationary
`phase was a 30 cm x 7.8 mm i.d. Ultrahydrogel-250 (Waters)
`column; the mobile phase consisted of the IL-2 reconstitution
`buffer; the sample size was 50 µ,l; the mobile phase flow rate
`was 0.75 ml/min; and detection was performed at 280 nm by
`a Waters 486 UV /VIS detector. Relative IL-2 concentrations
`were determined by peak area integration using the Waters
`MILLENIUM software package.
`Adsorbed Protein Characterization. Attenuated total re(cid:173)
`flectance FTIR spectroscopy was used to characterize the
`adsorbed IL-2. Spectra were collected on a Perkin-Elmer 1800
`FT-IR spectrophotometer equipped with a DTGS detector
`that was continuously purged with dry nitrogen. Five hundred
`twelve double-sided interferograms were collected using a 45°
`germanium attenuated total reflectance crystal in the 4000-
`700 cm-1 range at 2 cm-1 resolution, coadded, triangularly
`apodized, and Fourier-transformed. IL-2 solution spectra were
`recorded at 0.1 mg/ml in 10 mM phosphate buffer saline (pH
`7.4). The observed spectra consist of contributions from the
`silicone rubber tubing, adsorbed protein, adherent residual
`buffer solution, and residual water vapor in the light path;
`solution spectra have contributions from buffer and water
`vapor. These contributions were subtracted based on a pro(cid:173)
`tocol detailed elsewhere (Tzannis et al., unpublished work).
`Surface concentration. The amide II band (1495-1590 cm-1)
`was used to determine the adsorbed protein concentration (5).
`The estimated catheter surface concentration for monolayer
`coverage of IL-2 is 0.41 µ,g/cm2, assuming an Euclidean
`surface (Tzannis et al., unpublished work). However, topog(cid:173)
`raphy information for the inner surface of the catheter tubing
`used in this work, obtained via scanning electron microscopy
`in the secondary electron mode, indicated that the actual
`surface area is 2 to 4 times greater than the Euclidean area.
`Secondary structure. The amide I (1600-1700 cm-1) band
`was used to assess the secondary structure content of adsorbed
`IL-2 as a function of the duration of surface exposure. The
`spectral processing and fitting routines and their method
`validation are described elsewhere (Tzannis et al., unpublished
`work).
`
`RESULTS
`Bioactivity of Delivered IL-2. As the results given in Table
`1 indicate, the biological activity of delivered IL-2 is severely
`reduced; after 24 hr of delivery, the 0.1 and 0.05 mg/ml
`solutions retain only =10% and =6% of their original activity,
`respectively. The activity of the nonflow control samples was
`essentially constant. The majority of these losses occur within
`the first 2 hr; thereafter the losses continue but at a diminished
`rate. The activity profile reflects the solution residence time in
`the catheter. The 0.05 mg/ml system suffered a greater activity
`loss on a percentage basis than the 0.10 mg/ml system.
`To what may these losses be attributed? The unchanged
`activity of the control samples allows us to rule out any
`significant role for the pump reservoir surfaces or mere
`incubation at 37°C. Exposure to the catheter at 37°C must be
`responsible. We dismiss denaturation due to flow effects in the
`tubing a priori, as Reynolds numbers in the creeping flow
`regime put wall shear rates at least three orders of magnitude
`smaller than the threshold reported for enzyme inactivation in
`Couette flow (15). Nor do we expect chemical degradation
`over the 24-hr period to be responsible. The known major
`degradation products of IL-2, oxidized Met104 and Cys125, have
`been shown to retain full activity (16-18); these residues are
`not crucial for the biological activity of IL-2. Disulfide scram(cid:173)
`bling is known to diminish IL-2 bioactivity (19), but at the
`temperature studied, it appears to be significant only at
`alkaline pH (20). Thus, we suspect that the catheter surface has
`mediated the physical inactivation of IL-2 by either direct
`adsorption, aggregation, or denaturation.
`Aggregation Status of Delivered IL-2. As IL-2 aggregates
`often lack biological activity [refs. 17 and 21; although one
`commercial IL-2 preparation is formulated as a reversible
`aggregate (refs. 19 and 22)], we examined delivered IL-2 for
`evidence of increased levels of aggregates. The results in Table
`1 show that the delivered IL-2 is at all times largely monomeric.
`Both the flow and nonflow samples exhibited similar aggre(cid:173)
`gation patterns over the time examined. At all times the
`chromatograms are characterized by four distinct peaks with
`the monomer (15.1 ± 0.4 kDa) forming the predominant
`species ranging from 73 wt % initially to 93 wt % after 24 hr
`in the flow system. Two higher molecular weight species, which
`probably reflect dimers/trimers (35.2 ± 0.5 kDa) and pen(cid:173)
`tamers/hexamers (79.4 ± 1.0 kDa), represent only a small
`portion of the overall delivered protein and decrease with time.
`These high molecular weight species likely represent reversible
`aggregates formed upon reconstitution. Further, these aggre(cid:173)
`gates are likely biologically active, as the initial solution activity
`is within experimental error of the nominal bioactivity ex(cid:173)
`pected upon reconsitution per the manufacturer's product
`literature. The decline and disappearance of high molecular
`weight species also suggests that the surface-induced inactiva(cid:173)
`tion of IL-2 does not proceed through disulfide reduction,
`because that would probably lead to enhanced self-association
`in solution. Based on these findings, we conclude that aggre(cid:173)
`gation phenomena cannot explain the observed biological
`activity losses.
`In the course of this analysis of aggregation state, a low
`molecular weight species (8.0 ± 0.3 kDa) was detected at low
`levels throughout the 0.1 mg/ml experiments. This apparent
`cleavage fragment may account for a small percentage of the
`observed activity losses for these experiments. It is unlikely
`that this species is the result of a catheter surface-induced
`cleavage process, because a similar peak (8.4 ± 0.3 kDa)
`appears at comparable concentrations in the corresponding
`nonflow control. No such species was observed in the 0.05
`mg/ml experiments; if it is present at all, it may be at levels
`below the detectable limit.
`Delivered IL-2 Concentration Profile. Another possibility is
`that significant amounts of IL-2 have been physically removed
`
`Downloaded by guest on July 11, 2021
`
`Novartis Exhibit 2177.002
`Regeneron v. Novartis, IPR2021-00816
`
`
`
`5462
`
`Applied Biological Sciences: Tzannis et al.
`
`Proc. Natl. Acad. Sci. USA 93 (1996)
`
`Table 1. Delivered and adsorbed IL-2 assay results
`
`Bioactivity (x105 units/ml)*
`Flow
`Nonflow
`Concentration (µg/ml)
`Flow
`Nonflow
`Mass delivered (% of initiaI)t
`Size distribution (wt % )
`Flow
`Low molecular weight species
`Monomer
`Dimer/trimer
`Pentamer/hexamer
`Nonflow
`Low molecular weight species
`Monomer
`Dimer/trimer
`Pentamer/hexamer
`
`Surface exposure, 0.1 mg/ml IL-2 solution
`
`Surface exposure, 0.05 mg/ml IL-2 solution
`
`Initial
`solution
`
`2 hr
`
`10 hr
`
`24 hr
`
`Delivered IL-2
`
`Initial
`solution
`
`2 hr
`
`10 hr
`
`21 hr
`
`3.2 ± 1.2 0.75 ± 0.16
`6.8 ± 2.2
`ND
`
`0.52 ± 0.03 0.30 ± 0.09 5.2 ± 1.5 0.53 ± 0.21
`ND
`6.1 ± 1.9
`5.2 ± 1.5
`ND
`
`0.52 ± 0.22 0.35 ± 0.09
`ND
`5.35 ± 0.49
`
`103 ± 8
`108 ± 8
`100.0
`
`77 ± 5:f:
`106 ± 8:f:
`78.4 ± 6.6:f:
`
`79 ± 6
`106 ± 8
`77.8 ± 7.2
`
`82 ± 7
`100 ± 7
`87.0 ± 9.2
`
`54 ± 4
`57 ± 4
`100.0
`
`34 ± 2:f:
`55 ± 4:f:
`69.1 ± 7.2:f:
`
`34 ± 2
`55 ± 4
`68.8 ± 6.6
`
`39 ± 3
`59 ± 4
`69.1 ± 5.3
`
`0.0
`61.9
`27.1
`11.0
`
`0.0
`61.9
`27.1
`11.0
`
`6.1
`67.9
`19.1
`6.9
`
`0.0
`100§
`0.0
`0.0
`
`5.3
`92.2
`1.7
`0.8
`
`11.6
`88.4
`0.0
`0.0
`Adsorbed IL-2
`
`4.8
`93.0
`0.9
`1.3
`
`5.0
`95.0
`0.0
`0.0
`
`0.0
`70.5
`25.1
`4.4
`
`0.0
`70.5
`25.1
`4.4
`
`0.0
`61.0
`19.0
`20.0
`
`0.0
`50.2
`26.2
`23.6
`
`0.0
`82.2
`10.3
`7.5
`
`0.0
`95.5
`2.2
`2.3
`
`0.0
`96.9
`1.1
`2.0
`
`0.0
`83.0
`7.8
`9.2
`
`Surface Concentration (µg/cm 2)
`0.12,r
`0.44
`FTIR
`0.05
`0.0
`1.1
`0.31
`0.0
`0.05'
`0.35 ± 0.1
`0.08 ± 0.03
`0.43 ± 0.11
`0.12 ± 0.07:f:
`0.92 ± 0.2
`0.5 ± 0.1
`0.0
`0.0
`Mass balance from solution assayt
`*The Roussel-Uclaf batch records give activities for their bioassay equivalent to 12 ± 6 X 105 and 6.0 ± 3.0 X 105 units/ml for the 0.1 and 0.05
`mg/ml solutions, respectively; errors represent standard deviations of octuplicate assays.
`tconcentrations corrected for adsorption in nonflow system; error propagated from standard deviations of the protein assay standardization curve.
`:!:Samples determined at 4 hr.
`§A tail at low retention times signifies the presence of higher molecular weight species; their actual size is indeterminant.
`'liSamples determined at 6 hr.
`
`from solution by adsorption to the catheter surface, thereby
`reducing the total activity. The delivered IL-2 concentration
`was measured at 2-hr intervals; only the 2-, 10-, and 24-hr
`measurements are shown in Table 1. At 24 hr, the delivered
`IL-2 concentration was 20-30% less than that charged to the
`device, after correction for adsorption to the reservoir. The
`adsorptive losses are higher on a percentage basis for the lower
`concentration solution, reflecting the adsorptive capacity of
`the tubing. Thus, adsorptive losses can only account for a
`fraction of the observed bioactivity reduction.
`For both initial IL-2 concentrations, the adsorptive losses
`occur rapidly during the first 4 hr. After this point, the
`delivered concentrations appear to level off. This pseudo(cid:173)
`steady state may reflect a shift from kinetic to thermodynamic
`control of protein adsorption as the competition for adsorption
`sites stiffens and as adsorbed species optimize their interac(cid:173)
`tions with the surface (23).
`Adsorbed IL-2 Concentration Profile. To show that protein
`was indeed adsorbed to the catheter, we examined the exposed
`tubing via FTIR spectroscopy. The concentration of IL-2
`adsorbed on the catheter as a function of exposure time,
`calculated from the FTIR amide II band spectra as well as that
`obtained from a mass balance from the bicinchoninic acid
`assay of the solution concentration, is given in Table 1. The
`results from these techniques are in good agreement, indicat(cid:173)
`ing that our spectral processing is conservative. As expected,
`the surface concentration is higher when the solution concen(cid:173)
`tration is higher. The surface concentrations in both the 0.05
`and 0.1 mg/ml solution experiments track upwards, indicating
`that the adsorptive capacity of the tubing has not yet been
`reached. Our calculated value for monolayer coverage (5) and
`the Euclidean area of the tubing imply that there are multiple
`layers of IL-2 on the surface after 24 hr of exposure, in
`agreement with the reported behavior of silicone rubber
`surfaces in similar adsorption studies (24). However, by ac-
`
`counting for the roughness of the tubing surface, we estimate
`that the actual surface coverage is closer to a single IL-2 layer
`after 24 hr of exposure.
`Adsorbed IL-2 Secondary Structure. Numerous studies
`have shown that protein-surface interactions can be denatur(cid:173)
`ing, particularly when the surface is hydrophobic (25). Amide
`I band IR spectra for adsorbed IL-2 and the underlying
`component bands obtained by our deconvolution procedure
`are shown as a function of exposure time in Fig. 1; the
`corresponding secondary structure estimates are listed in
`Table 2. Native IL-2 in solution is a highly helical protein, as
`shown by the intense band located at 1657 cm- 1; a-helix
`content comprises almost 63% of the structure. This is in good
`agreement with published IL-2 structural analyses that indi(cid:173)
`cate a total of 42-65% helix and little or no /3-sheet content
`(16, 26). The band appearing at 1630 cm- 1 can be assigned to
`extended loop structures that connect the helical segments of
`the four-helix bundle.
`Upon adsorption, IL-2 molecules lose an important part of
`their native structure; a-helices and extended loop bands
`disappear while /3-sheet signals appear. After 2 hr of surface
`exposure, the prominent a-helix signal is replaced by intense
`/3-sheet bands located at 1625, 1633, and 1639 cm-1 and a
`random structure band at 1650 cm- 1; at this point, /3-sheet and
`random structures compose =50% and 25% of the secondary
`structure content, respectively. Over time, the patterns of
`increased /3-sheet and random structures persist ( =43% and
`13% of the structure at 24 hr, respectively), while there is
`partial recovery of a-helix ( =20%) and diminution of /3-tum
`( = 12%) bands. Either the sampled surface population relaxes
`to a more native-like state, or the subsequently deposited IL-2
`molecules fail to undergo the initial dramatic secondary
`structure changes. During this time, the dominant /3-sheet
`band position shifts from 1625 (2 and 6 hr) to 1631 cm- 1 (24
`hr). This high wavenumber /3-sheet shift is accompanied by a
`
`Downloaded by guest on July 11, 2021
`
`Novartis Exhibit 2177.003
`Regeneron v. Novartis, IPR2021-00816
`
`
`
`Applied Biological Sciences: Tzannis et al.
`
`Proc. Natl. Acad. Sci. USA 93 (1996)
`
`5463
`
`1600
`
`1620
`1640
`1660
`1680
`Frequency, cm·1
`
`1700
`
`1600
`
`1620
`1640
`1680
`1680
`Frequency, cm·1
`
`1700
`
`1600
`
`1820
`1840
`1660
`1660
`Frequency, cm·1
`
`1700
`
`1600
`
`1820
`1840
`1660
`1680
`Frequency, cm·1
`
`1700
`
`FIG. 1. Deconvoluted and curve-fitted infrared amide I band envelopes of IL-2 (A) in solution and adsorbed on the catheter after 2 hr (B),
`6 hr (C), and 24 hr (D). The IL-2 concentration charged to the device was 0.1 mg/ml. Spectra are corrected for buffer, silicone rubber tubing,
`and water vapor contributions where appropriate (5).
`
`similar rearrangement of the /:I-turns, as their major band shifts
`from 1667 to 1679 cm-1 after 24 hr, indicating stronger hydrogen
`bonding patterns of the surface-stabilized molecules. Thus, IL-2
`is denatured upon adsorption to silicone rubber tubing.
`
`Table 2. Secondary structure of adsorbed and delivered IL-2 as a
`function of exposure time
`
`Surface exposure
`time
`
`2 hr
`
`6 hr
`
`24 hr
`
`Initial
`solution
`
`Delivered IL-2 Secondary Structure. To probe the hypoth(cid:173)
`esis that surface-denatured protein may desorb and persist in
`solution, we analyzed the delivered IL-2 samples via CD
`spectroscopy. Far-UV CD spectra of the delivered IL-2 as a
`function of exposure time are shown in Fig. 2; the correspond(cid:173)
`ing secondary structure estimates are given in Table 2. The
`initial solution sample estimates are in good agreement with
`the FTIR results. Deconvolution results for the flow system
`spectra show a rapid =75% decrease in a-helix content and an
`
`40000r-T""r-T""r-T""r-T""r-T""r-T""T"T""T"T""T"T""T"T""T"'T"T"'T"T"'T"T"'T",......,
`
`20000
`
`0
`
`-20000
`
`I
`
`I
`
`I
`
`190
`
`200
`
`210
`
`220
`
`230
`
`240
`
`250
`
`Wavelength, nm
`
`Flo. 2. Far-UV CD spectra of delivered IL-2 as a function of
`exposure time are shown: native IL-2 in solution (--) and after 2 hr
`(· • • • • ·), 6 hr(--), 12 hr (-•-),and 24 hr (- .. -)of exposure to the
`catheter. All spectra refer to 0.1 mg/ml delivery experiments.
`
`28
`42
`13
`17
`0
`
`39
`38
`9
`7
`
`52
`21
`13
`13
`
`63
`0
`17
`0
`20
`
`65
`10
`16
`10
`
`8
`48
`20
`25
`0
`
`19
`50
`23
`11
`
`19
`48
`21
`12
`0
`
`31
`43
`17
`8
`
`Adsorbed IL-2:
`FTIR estimates (structure % )
`a-Helix
`/3-Sheet
`/3-Turns
`Random
`Extended
`Delivered IL-2:
`CD estimates (structure % )*
`a-Helix
`{3-Sheet
`/3-Turns
`Random
`Nonflow control:
`CD estimates (structure % )*
`a-Helix
`65
`54
`56
`/3-Sheet
`10
`24
`20
`/3-Turns
`16
`16
`10
`Random
`10
`10
`9
`All data correspond to 0.1 mg/ml delivery experiment.
`*Standard errors associated with CD estimates were determined to be
`on the order of ±6%via the use of the SELCONprogram (9) on a series
`of native IL-2 solution spectra.
`
`Downloaded by guest on July 11, 2021
`
`Novartis Exhibit 2177.004
`Regeneron v. Novartis, IPR2021-00816
`
`
`
`5464
`
`Applied Biological Sciences: Tzannis et al.
`
`Proc. Natl. Acad. Sci. USA 93 (1996)
`
`increase in /3-sheet content relative to the nonflow system over
`the first 2 hr of delivery. After longer exposure times, the
`secondary structure contents appear to relax back toward the
`initial native contents. This apparent partial recovery of helix
`content likely reflects changing protein populations in solution
`as time progresses. At all exposure times, there is good
`agreement between the CD structural estimates for delivered
`IL-2 and the FfIR structure estimates for adsorbed IL-2. This
`agreement is particularly striking with respect to the rapid loss
`and partial recovery of helix content as well as the increase in
`/3-sheet content on surface exposure. These results are con(cid:173)
`sistent with the hypothesis above.
`Delivered IL-2 Tertiary Structure. To further probe the
`conformational status of the delivered protein, fluorescence
`spectroscopy was employed. Steady-state fluorescence spectra
`of IL-2 during delivery, which are indicative of the polarity of
`the environment of the lone Trp residue at position 121, are
`shown in Fig. 3. The maximum intensity of the initial native
`IL-2 emission spectrum is located at 328 nm, indicating that
`Trp121 is buried in the interior of the helix bundle (11, 12). For
`IL-2 denatured in 6 M guanidinium chloride, there is a 30-nm
`red shift of the emission maximum, indicating that this residue
`becomes solvent-exposed. Also, a signal intensity comparison
`of these spectra reveals substantial fluorescence quenching in
`the native state. This quenching is thought to arise via inter(cid:173)
`
`action with the neighboring Cys125 residue (11, 12r When the
`
`C-terminal helix on which both Trp121 and Cys 25 reside is
`disrupted, this quenching is reduced; this helix is known to be
`crucial for biological activity (27). After 2 and 6 hr of surface
`exposure, the IL-2 samples exhibit a 12- to 15-nm red shift as
`well as a dequenching, which are intermediate to the native and
`guanidinium chloride-denatured spectra, indicative of a large
`population of nonnative molecules in the solution. The static
`quenching experiments yielded similar results as shown by the
`Stern-Volmer plots in Fig. 4. In native IL-2, Trp121 is buried
`in the core of the four-helix bundle and is inaccessible to iodide
`and partially accessible to acrylamide, in good agreement with
`literature studies on IL-2 (11, 12) and human growth hormone
`(10). In the guanidinium chloride-denatured state, Trp121
`becomes fully accessible to both quenchers. The surface(cid:173)
`exposed samples (2- and 6-hr) again give results intermediate
`to the native and guanidinium chloride-denatured states.
`Trp121 has become partially accessible to both quenchers after
`incubation with the tubing, indicating a swelling or partial
`unfolding of the delivered IL-2 molecules.
`
`DISCUSSION
`To elucidate the role of protein-surface interactions in a
`medical-grade silicone standard central venous catheter, com-
`
`mon to virtually all implantable and portable delivery systems,
`we examined the biological activity and structural stability of
`IL-2 during delivery at 37°C. In agreement with previously
`published clinical studies that ascribed the bioactivity losses of
`IL-2 during continuous infusion to deposition on the polyeth(cid:173)
`ylene catheter tubing (6), we found order of magnitude
`bioactivity losses resulting after exposure of the protein solu(cid:173)
`tion to silicone rubber catheter tubing. Losses due to deposi(cid:173)
`tion were significant but small; losses due to incubation at
`elevated temperature, shear-induced denaturation, and sur(cid:173)
`face-mediated aggregation were insignificant. The majority of
`these losses were attributed to transient, denaturing interac(cid:173)
`tions between IL-2 and the surface of the catheter. The
`similarities between the secondary structure analyses of ad(cid:173)
`sorbed IL-2 by FfIR spectroscopy and delivered IL-2 by CD
`spectroscopy, both in extent of perturbation of the native
`helical structure and in temporal pattern, are consistent with
`this interpretation. The Trp121 emission spectra and static
`quenching results for delivered IL-2 provide further insight
`into the nature of the surface-induced perturbation and the
`observed loss of activity: the catheter surface induces a
`transition to a state with expanded hydrophobic core (10)
`accompanied by alterations in helix D, which is critical for
`biological activity (27). The surface catalyzes structural
`changes that are not readily reversed upon desorption.
`Protein denaturation upon surface interaction is a well(cid:173)
`documented phenomenon (28-32). Kondo et al. (29) have
`indicated that a relationship exists between the extent of
`surface-induced conformational changes and the value of the
`adiabatic compressibility, /3s, of the protein. The value of /3s for
`IL-2, as calculated from its partial specific volume (32), is 9.77
`x 1012 cm2dyne- 1 (1 dyne= 10 µ,N). This is close to the value
`reported for bovine serum albumin, which is known to undergo
`dramatic conformational changes upon surface exposure (28,
`29). It has also been shown by Gavish et al. (33) and Gekko et
`al. (32) that a-helices, in contrast to /3-sheets, form dynamic
`domains of the protein structure and are responsible for
`volume and energetic fluctuations of proteins in the solution
`state; this may predispose helical proteins such as IL-2 to
`alterations upon adsorption. Further, in lattice-based simula(cid:173)
`tions of homopolymer chains adsorbing to surfaces, the chains
`were observed to "dock" initially and then flatten out, under(cid:173)
`going conformational changes as the polymer-surface contacts
`are maximized (34). The same studies also indicated that
`certain types of organized secondary structures become prev(cid:173)
`alent as a surface is approached. These altered, surface(cid:173)
`adapted species may desorb due to unfavorable lateral inter(cid:173)
`actions on the surface or to the attainment of an unstable,
`high-energy intermediate state (35, 36).
`Structure analyses of adsorbed species indicate that the
`extended loop structures of IL-2 disappear upon adsorption,
`tempting speculation as to the mechanism by which the protein
`and surface interact. The loops are on the exterior of IL-2 and
`are rich in hydrophobic residues and are therefore likely to be
`the part of the molecule that is first presented to the surface .
`Chou and Scheraga (5) have proposed that the loop-helix
`interactions are crucial to the structural integrity of four-helix
`bundle proteins like IL-2. We hypothesize that once these
`segments are involved in surface interactions, their stabilizing
`effect on the four-helix bundle is dissipated. As a consequence,
`the helices of the molecule are able to engage the surface,
`initiating a surface-contact optimization process, resulting in
`further deviations from the native structure.
`IL-2 is one of a number of pharmaceutical proteins that are
`currently under clinical investigation for intermittent and/or
`continuous infusion, pump-based delivery strategies. However,
`with the currently available devices and catheters, such treat(cid:173)
`ment programs do not guarantee complete or even predictable
`delivery. This is of particular concern when the dose-response
`relationship is highly uncertain.
`
`,,,,..
`
`....
`
`'
`
`/
`
`I
`_.--/.. . ..._
`I
`.. -·f
`I
`
`··---
`
`/
`
`/
`
`I
`
`Q)
`
`120
`.... 100
`~
`=
`rn
`j
`5
`~
`~
`-
`::I
`ll.
`
`80
`
`60
`
`40
`
`20
`
`'-
`
`"
`
`'-
`
`'-
`
`'
`··, ..
`-.:\_,
`
`·;,;
`
`'-
`
`'-
`
`'-
`
`'
`
`'
`
`'-
`
`"
`'
`'
`' ' ....
`400
`
`'-
`
`'- -...;:
`
`420
`
`320
`
`340
`
`360
`
`380
`
`Wavelength, nm
`
`FIG. 3. Trypt