JOURNAL OF INTERFERON RESEARCH 8:539-547 (1988)Mary Ann Liebert, Inc., Publishers

The Long-Term Stability of Recombinant(Serine-17) Human Interferon-ß

JOHN GEIGERT,1 BARBARA M. PANSCHAR,1 STANLEY FONG,2HOLLY N. HUSTON,2 DANA E. WONG,2 DAVID Y. WONG,2

CAROL TAFORO,3 and MARK PEMBERTON"

ABSTRACT

Escherichia co/z'-derived (Serine 17) human interferon-/? (HuIFN-/?ser) was formulatedwith SDS and placed at multiple isothermal temperatures (-70°C to 37°C). Three stabil¬ity-indicating test methods (bioassay, SDS-PAGE, RP-HPLC) were used to evaluate thelong-term stability of this preparation. No change was observed when stored for nearly a

year at either -70°C or 4°C. At the elevated temperatures, proteolytic cleavage, nonco¬valent oligomer formation, and loss of antiviral activity were observed. The absence of acarrier protein makes this stable IFN-/3 frozen reference preparation useful as a standardin both biological assays and protein chemical methods of analysis.

INTRODUCTION

The long-term stability of recombinant human fibroblast interferon (HuIFN-/3), formulated·with human serum albumin, was shown to be stable for over 2 years when held under refrigera¬

tion."1 However, because of the presence of the protein carrier, only antiviral activity could beused as a stability-indicating test method. Therefore, it was not possible to determine if subtlechanges were occurring in the HuIFN-/3 protein itself.

There are a few nonprotein excipients that are suitable for solubilizing the hydrophobic HuIFN-ß molecule. One of these excipients is the anionic surfactant, sodium dodecyl sulfate (SDS). SDSdoes not decrease the antiviral activity of IFN-/3 and maintains nonglycosylated recombinantHuIFN-ß in solution at neutral pH.(2) SDS has also been shown to stabilize the bioactivity of nat¬ural HuIFN-/3 during physical mistreatment (e.g., heating at 100°C, freeze-thaw, and vortex agita¬tion).«3 "'

This interaction between SDS and HuIFN-/3 has been utilized effectively in the large-scale purifi¬cation of Escherichia co/z'-derived (Serine 17) HuIFN-/3 (HuIFN-/3SER).(2) Natural HuIFN-/3 con¬tains three cysteine amino acids in the protein, two of which are required to form the necessary di¬sulfide bond. The extra cysteine amino acid adds instability to the molecule by participating in di-

'Department of Product Evaluation, department of Quality Assurance, 'Department of Analytical Devel¬opment, and 'Department of Clinical Manufacturing, Cetus Corporation, 1400 Fifty-Third Street, Emeryville,CA 94608.

539

GEIGERT ET AL.

sulfide bond scrambling. To circumvent this problem, the HuIFN-/3 gene was engineered, by means

of site-specific mutagenesis, to code for serine in place of the extra cysteine at amino acid position17. The engineered gene, cloned into E. coli, produced HuIFN-/3SER.(s·6' Since only two cysteineamino acids remain in the protein, intramolecular disulfide bond scrambling is eliminated andintermolecular disulfide bond formation is minimized.

To provide a basis for comparison of various IFN-/3 materials used among different laboratories,it is important that reference preparations be made available. Reference preparations need to be es¬tablished not only for standardizing the bioassay methods used to assign potency but also for stan¬dardizing the physico-chemical assay methods used to assign protein purity.

A comprehensive stablity of purified recombinant HuIFN-/3sER, formulated with SDS, is now

being reported. At multiple isothermal conditions, both biological and protein chemical parameterswere monitored to determine the long-term stability of this preparation. A prediction of the stabil¬ity of HuIFN-/3sER under extended storage conditions is also made. This IFN-/? reference prepara¬tion, because it is formulated in the absence of a carrier protein, can be used to standardize bothbioassays and protein chemical test methods.

MATERIALS AND METHODS

HuIFN-ßSER: Escherichia coli cells containing recombinant HuIFN-ßsER were harvested fromthe fermentation medium and disrupted to release the cytoplasmic inclusion bodies containing theIFN-/3. The particulate matter was suspended in SDS, reduced with dithiothreitol (DTT), and thenextracted with 2-butanol. IFN-/3 in the organic extract was precipitated by acid treatment and thenresolubilized in SDS. The solution was passed through a Sephacryl S-200 gel sizing column. Theprotein in the eluate was gently oxidized with iodosobenzoic acid (IBA) and then passed throughanother Sephacryl S-200 gel sizing column. Finally, the IFN-/3 solution was passed through a Seph-adex G-75 gel sizing column. The eluting buffer was 50 mM acetate buffer (pH 5.5) containing0.1% SDS and 2 mMEDTA. The pure HuIFN-/3SER that eluted was used in this study.(2·7» Its con¬centration was at 1.2 mg IFN-pVml as determined by the Lowry assay using bovine serum albuminstandards.

Multiple Isothermal Stability (MIS): The pure HuIFN-/3SER solution, described above, was dis¬pensed into borosilicate glass, Teflon-lined, screw-cap culture tubes (13 x 45 mm). The tubes were

placed in controlled-temperature units set at either -70°C, 4°C, 25°C, or 37°C. At set monthly in¬tervals, a tube was removed from each storage temperature and analyzed.

Antiviral Activity Measurement: The quantitative measurement of IFN-/3 antiviral activity wasmade by a virus yield reduction assay with fibroblast cells trisomie for chromosome 21 (GM2504),the Indiana strain of vesicular stomatitis virus (VSV), and baby hamster kidney cells (BHK-21).GM2504 cells were obtained from the NIGMS Human Genetic Mutant Cell Repository and main¬tained on minimum essential medium (Earles), supplemented with 10% fetal calf serum (FCS), and1% nonessential amino acids and vitamins. VSV was obtained from the American Type CultureCollection (ATCC) and maintained on the GM2504 cells. BHK-21 cells were obtained from theATCC and maintained on Dulbecco's modified Eagle's medium (DME) supplemented with 10%FCS.

IFN-/3-containing samples were added to confluent monolayers of GM2504, and the cells were in¬cubated for 24 h at 37°C (7% C02). The cells were infected with 106 pfu of VSV (a multiplicity ofinfection of 30) and then incubated for 50 min. After being rinsed with DME (2% FCS) to remove

unadsorbed VSV, the cells were incubated for 24 h at 37°C (7% C02). The reduction in virus pro¬duction due to the concentration of IFN-/3 present was measured by transferring supernatant fromthe virus-infected GM2504 cells to monolayers of BHK-21 cells, and then incubating the BHK-21cells for 60 min. After removing the supernatant, an agarose overlay was added, and the cells were

540

STABILITY OF HuIFN-/3SER

incubated for 18 h at 37°C (5% C02). The overlays were stained with Neutral Red dye, and thevisible plaques were counted. HuIFN-/3 standards were run in the assays. They were calibratedagainst the international HuIFN-/3 reference preparation (G-023-902-527) available from theNIAID (NIH).<8) By plotting log,0 virus infectivity reduction against log,0 International Units (IU),a dose-response standard curve was prepared. From the linear portion of this curve, a regressionline was drawn through the data and the activity for each sample was calculated. Because of the in¬herent interassay variability of bioassays ( ~ 50% RSD in the virus yield reduction assay), each sam¬

ple was assayed in duplicate on three successive days.

SDS-PAGE Analysis: The quantitative measurement of IFN-/3 protein purity was carried out bySDS-PAGE with the system of Laemmli (1970) using a 12-15% linear gradient polyacrylamide gel.Samples were heated for 3 min to approximately 95% in a glycerol/SDS solution either with (forreducing conditions) or without 2-mercaptoethanol (for nonreducing conditions). Fast Green dyewas used to stain the protein bands. Gels were scanned with a Camag Thin-Layer ChromatoScanner II and the peaks integrated with a Nelson 6000 data system using an exponential skimalgorithm.

RP-HPLC Analysis: The quantitative measurement of IFN-/3 protein purity was carried out byRP-HPLC using a SP 8700 pump, a WISP injector, and a Spectroflow 773 detector. The columnwas 5-/im 250 4.6-mm i.d. 214TP C4 (Vydac, Hesperia, CA) which was heated to 30°C in the sol¬vent column block heater. A precolumn, 30 2-mm i.d. packed with the same Vydac C, material,was used. The eluents, used at a flow-rate of 1.0 ml/min, were 10% acetonitrile with 0.1% trifluo-roacetic acid (TFA) and 100% acetonitrile with 0.1% TFA. The elution gradient was 10% aceto¬nitrile at 0 min, 47.8% acetonitrile at 22 min, 59.5% acetonitrile at 45 min, 100% acetonitrile at 50min, followed by a 5-min equilibration. Absorbance at 214 nm was used for the detection of pro¬teins. The peaks were integrated with a Nelson 6000 data system using an exponential skim algo¬rithm.

RESULTS AND DISCUSSIONIFN-ß preparation at the start

At the beginning of this study, the E. co/z'-derived, HuIFN-/3ser preparation was both highlyactive (3 108 IU/mg IFN-/3) and pure (>90% IFN-/3 monomer by SDS-PAGE and RP-HPLC).

Protein purity assignments are subject to interlaboratory variability. Depending upon themethod used to determine the start/stop of a protein (e.g., baseline drop, tangenital skim, ex¬

ponential skim), a SDS-PAGE or a RP-HPLC protein purity assignment value could be exag¬gerated. For this study, the exponential skim algorithm was used to integrate the protein peakswhich leads to the conservative estimate of protein purity for the IFN-/3 preparation.

At low temperatures (4°C or colder), SDS precipitates out of solution and so does the IFN-/3.However, upon warming to room temperature, the SDS returns to solution and so does the IFN-ß.

Stability by antiviral activityPotency stability is presented in Fig. 1. The IFN-/3 preparation showed no statistically significant

change in potency over the storage period when held at either -70°C or 4°C. A slight decrease inpotency did occur at 25°C and was clearly accelerated at the 37°C storage.

Stability by SDS-PAGEProtein stability by SDS-PAGE is presented in Fig. 2. The IFN-/3 preparation had no statistically

significant change in the protein purity over the storage period when held at either -70°C or 4°C.At elevated temperatures, proteolytic cleavage of IFN-/3 into lower-molecular-weight fragments was

541

GEIGERT ET AL.

9.00t

LOG

OF

SPEC

FIC

ACTIVITY

MONTHS IN STORAGE

FIG. 1. Potency stability of HuIFN-/3SER formulated with SDS, upon storage at -70°C ( ), 4°C(·), 25°C (A), and 37°C ( ). At each monthly time point, a sample from each temperature was

assayed up to six times. Specific activities are expressed as log,0 IU/mg IFN-/3. Linear least-squareregression lines are drawn through the data. The slope values are -0.001* (-70°C), 0 (4°C),-0.001 (25°C), and -0.004 (37°C). The asterisk (*) indicates that the -70°C slope value is not sta¬tistically significant from a value of 0.

readily occurring. Only trace levels of fragmentation were observed in the samples stored at 4°C or

colder. Oligomer formation was not a significant problem at any of the temperatures studied, .butwas observed at trace levels. Figure 3 illustrates the changes observed in the protein after 5 monthsin storage.

Stability by RP-HPLCProtein stability by RP-HPLC is presented in Fig. 4. There are two peaks on the chromatogram

referred to as peak and peak A. Both peaks migrate as IFN-/3 monomer on SDS-PAGE analysis.Peak A is a methionine sulfoxide IFN-/3 derivative. The IFN-/3 preparation had no statistically sig¬nificant change in the protein purity over the storage period when held at either -70°C or 4°C. Atelevated temperatures, noncovalent oligomers were forming. This was noted by the increased pro¬tein content eluting after the IFN-/3 monomer peak (i.e., peak B) in the chromatograms. Since thisdid not appear in the SDS-PAGE analyses, the noncovalent oligomers must be SDS-disassociable.No increase in peak A occurred upon storage at any of the temperatures studied. Figure 5 illustratesthe changes observed in the protein after 5 months in storage.

Comparison of the three stability-indicating test methodsFor storage at both -70°C and 4°C, the three stability-indicating test methods (potency, SDS-

PAGE, and RP-HPLC) did not detect any change in the HuIFN-/3SER material when held for

542

STABILITY OF HuIFN-/3SER100

MONTHS IN STORAGE

FIG. 2. Protein purity by SDS-PAGE of HuIFN-/3SER, formulated with SDS, upon storage at -70°C( ), 4°C (·), 25°C (A), and 37°C (T). At each monthly time point, a sample from each temperature was

analyzed. Percent IFN-/3 monomer is from SDS-PAGE reducing gel. Linear least-square regression linesare drawn through the data. The slope values are +0.15* (-70°C), -0.12* (4°C), -2.24 (25°C), and-10.0 (37°C). An asterisk (*) indicates that the slope value is not statistically significant from a value of 0.

REDUCEDNON

REDUCED

FN DIMER

IFN MONOMER

MWSTD

70° 4° 25° 37° -7<f 4° 25°37J

FIG. 3. SDS-PAGE analysis of HuIFN-/3SER, formulated with SDS, after 5 months in storage. Sevenmicrograms of IFN-/3 were loaded per lane. The gel was stained with silver prior to photographing.

543

GEIGERT ET AL.

4 5 6 7

MONTHS IN STORAGE

FIG. 4. Protein purity by RP-HPLC of HuIFN-/3SER, formulated with SDS, upon storage at-70°C ( ), 4°C (·), 25°C (A), and 37°C ( ). At each monthly time point, a sample from eachtemperature was analyzed. Percent IFN-/3 monomer is percent peak from RP-HPLC. Linearleast-square regression lines are drawn through the data. The slope values are +0.24* (-70°C),+0.21* (4°C), -2.49 (25°C), and -12.6 (37°C). An asterisk (*) indicates that the slope value is notstatistically significant from a value of 0.

nearly a year. When held at 25°C, both the SDS-PAGE (which detected proteolytic cleavage) andthe RP-HPLC (which detected noncovalent oligomer formation) test methods readily picked upchanges in the SDS formulated material, while the potency test method only gave a hint of changeoccurring. This clearly shows that a systematic evaluation of any reference perparation should in¬clude more than one stability-indicating test method. For storage at 37°C, the change in the proteinwas so significant that all three test methods readily observed it.

Overall, potency stability is not a sensitive measure of protein change compared to the physico-chemical methods of analysis. This is because of the inherent variability of the bioactivity measure¬

ments, which make it not possible to accurately assess subtle changes in the nature of the protein.

Arrhenius plots to predict long-term stabilityWhile it could take several years to complete the ongoing stability study of the IFN-/3 preparation

under frozen conditions, the multiple isothermal stability (MIS) study yields a prediction of whatthis extended stability might be. The MIS data was fit into the Arrhenius equation:

log = log A Ha2.303 RT

544

STABILITY OF HuIFN-/3SER

4°C

Peak

-70° C

Peak A

-—v~26 28 30 32 34 36

TIME (min)

FIG. 5. RP-HPLC analysis of HuIFN-/3SER, formulated with SDS, after 5 months in storage.Twenty micrograms of IFN-/3 injected per sample. The chromatograms are the 214-nm absorbancetracings. Peak A is the methionine-sulfoxide derivative of IFN-/3; peak is the IFN-/3 monomer.

where = specific rate of change (i.e., the slope values of the instability regression lines in Figs. I,2, and 4); A = frequency factor (a constant); Ha = activation energy (a constant); R = gas con¬

stant (1.99 cal deg"1 mol""1); and = absolute temperature (°K).By plotting log k versus 1/T, referred to as the Arrhenius plot, a linear regression line could be

drawn through the data. By extrapolating this line down to refrigeration (4°C) or freezing (-70°C)temperature conditions, a prediction of stability at these storage temperatures can be obtained. Fig¬ure 6 presents the Arrhenius plots for the three test methods used to monitor the stability of theHuIFN-/3SER material. Because only two temperatures can be plotted (i.e., only two temperaturesshowed any change in the protein), the result obtained by extrapolation will be somewhat weak.

545

GEIGERT ET AL.

LOG -

SDS- PAGERP-HPLC

£POTENCY

25°CT £32 3-3 3.« 3.5 3.6 3.7

1/T (x 1000)

FIG. 6. Arrhenius plot for HuIFN-/3SER, formulated with SDS, using potency (·), SDS-PAGE( ), and RP-HPLC (A). On the vertical axis, the K values correspond to the regression line slopesof Figs. 1, 2, and 4. On the horizontal axis, a 1/7"(1,000) value of 3.23 corresponds to 37°C and avalue of 3.36 corresponds to 25°C. The temperature of refrigeration (4°C) and freezing (-70°C)corresponds to l/T(1,000) values of 3.61 and 4.93, respectively.

When extrapolating down to refrigeration temperature (4°C; a / " 1,000 value of 3.61), the po¬tency results predict a 10% decrease in potency by 8 months in storage. Such a small decreasewould not be detected by the bioassay. The SDS-PAGE and RP-HPLC results predict a 10% de¬crease in purity only after 7 years of storage under refrigeration. Extrapolating even further, downto freezing temperature (-70°C; a l/T 1,000 value of 4.93), all three test method results predictthat less than 10% change will occur even after more than 50 years in storage.

An additional feature of the Arrhenius equation is that the activation energy (Ha) for change canbe calculated from the slope value of the Arrhenius plot regression line:

Ha = -2.303 R (slope of the Arrhenius plot)The activation energies were calculated to be 16 kcal/mole for the loss of antiviral activity and 24kcal/mole for both the proteolytic cleavage and noncovalent oligomer formation in the IFN-/3 pro¬tein. These activation energy values are consistent with values previously obtained for HuIFN-/3.u" (It should be noted that the activation energy equation on page 208 of Ref. 1 is incorrect andthat the correct activation energy for lyophilized HuIFN-/3sER formulated with human serum al¬bumin is 25 kcal/mole).

ACKNOWLEDGMENT

This study was performed under a joint development program with Triton Biosciences Inc.,Alameda, CA 94501.

546

STABILITY OF HuIFN-/3SER

REFERENCES

1. GEIGERT, J., ZIEGLER, D.L., PANSCHAR, B.M., CREASEY, A.A., and VITT, C.R. (1987). Potencystability of recombinant (serine 17) human interferon-ß. J. Interferon Res. 7, 203-211.

2. LIN, L.S., YAMAMOTO, R., and DRUMMOND, R.J. (1986). Purification of recombinant human inter¬feron-/) expressed in Escherichia coli. Methods Enzymol. 119, 183-192.

3. SEDMARK, J.J., JAMESON, P., and GROSSBERG, S.E. (1977). Thermal and vortical stability of puri¬fied human fibroblast interferon, in: Human Interferon, Production and Clinical Use. W.R. Stinebring andP.J. Chappie (eds.). New York: Plenum Press, pp. 133-152.

4. HABEEB, A.F.S.A. (1977). Influence of conformation on immunochemical properties of proteins, in: Im-munochemistry of Proteins, Vol. 2. M.Z. Atassi (ed.). New York: Plenum Press, pp. 163-230.

5. MARK, D.F., LU, S.D., CREASEY, A.A., YAMAMOTO, R., and LIN, L.S. (1984). Site-specific muta-genesis of the human fibroblast interferon gene. Proc. Nati. Acad. Sci. USA 81, 5662-5666.

6. PESTKA, S. (1986). Interferon from 1981 to 1986. Methods Enzymol. 119, 3-13.7. SHAKED, Z., and WOLFE, S.N. (1985). Controlled oxidation of microbially produced cysteine-contain-

ing proteins. U.S. Patent 4,530,787.8. PESTKA, S. (1986). Interferon standards and general abbreviations. Methods Enzymol. 119, 14-23.9. COOK, A.W., NIDZGORSKI, F., CAME, P., MANN, G., CARTER, W.A., and ROANE, P.R. (1986).

Serum protein masking of the thermal sensitivity of the antiviral activity of purified human beta interferon:Implications for clinical studies. J. Biol. Response Modifiers 5, 499-503.

Address reprint requests to:Dr. John Geigert

Manager, IFN-ß ProjectCetus Corporation

1400 53rd StreetEmeryville, CA 94608

Received 16 November 1987/Accepted 15 March 1988

547

This article has been cited by:

1. Miranda M. C. Beers, Melody Sauerborn, Francesca Gilli, Vera Brinks, Huub Schellekens, Wim Jiskoot.2011. Oxidized and Aggregated Recombinant Human Interferon Beta is Immunogenic in HumanInterferon Beta Transgenic Mice. Pharmaceutical Research . [CrossRef]

2. Andrea Hawe, Wolfgang Friess. 2008. Development of HSA-free formulations for a hydrophobic cytokinewith improved stability. European Journal of Pharmaceutics and Biopharmaceutics 68:2, 169-182. [CrossRef]

3. J Bradley. 2002. A phase III comparison of radiation therapy with or without recombinant β-interferon forpoor-risk patients with locally advanced non-small-cell lung cancer (RTOG 93-04). International Journalof Radiation Oncology*Biology*Physics 52:5, 1173-1179. [CrossRef]

4. GÜNTHER R. ADOLF, ELISABETH TRAXLER, INGRID MAURER-FOGY. 1990. RecombinantEquine Interferon-β1: Purification and Preliminary Characterization. Journal of Interferon Research 10:3,255-267. [Abstract] [Full Text PDF] [Full Text PDF with Links]