Analysis of Molecular Heterogeneity in Therapeutic Ifn2b from Different

molecules

Article Analysis of Molecular Heterogeneity in Therapeutic IFNα2b from Diﬀerent Manufacturers by LC/Q-TOF

1, 1, 2 1 1, 1, Lei Yu † , Lei Tao †, Yinghua Zhao , Yonghong Li , Dening Pei * and Chunming Rao * 1 Division of Recombinant Drugs, National Institute for Food and Drug Control (NIFDC), Tiantan Xili No. 2, Dongcheng District, Beijing 100050, China; [email protected] (L.Y.); [email protected] (L.T.); [email protected] (Y.L.) 2 SCIEX China, Beijing 100015, China; [email protected] * Correspondence: [email protected] (D.P.); [email protected] (C.R.) These authors contributed equally to this work. † Academic Editor: Katarzyna Lech Received: 6 August 2020; Accepted: 29 August 2020; Published: 31 August 2020

Abstract: Recombinant human IFNα2b (rhIFNα2b), as an important immune-related protein, has been widely used in clinic for decades. It is also at the forefront of the recent emergence of biosimilar medicines, with numerous products now available worldwide. Although with the same amino acid sequence, recombinant proteins are generally heterogeneous due to post-translational modification and chemical reactions during expression, purification, and long-term storage, which could have significant impact on the final product quality. So therapeutic rhIFNα2b must be closely monitored to ensure consistency, safety, and efficacy. In this study, we compared seven rhIFNα2b preparations from six manufacturers in China and one in America, as well as four batches of rhIFNα2b preparations from the same manufacturer, measuring IFNα2b variants and site-specific modifications using a developed LC/Q-TOF approach. Three main forms of N-terminus, cysteine, methionine, and acetylated cysteine were detected in five rhIFNα2b preparations produced in E. coli (1E~5E) and one in Pseudomonas (6P), but only the native form with N-terminal cysteine was found in rhIFNα2b preparation produced in Saccharomyces cerevisiae (7Y). Two samples with the lowest purity (4E and 6P), showed the highest level of acetylation at N-terminal cysteine and oxidation at methionine. The level of oxidation and deamidation varied not only between samples from different manufacturers but also between different batches of the same manufacturer. Although variable between samples from different manufacturers, the constitution of N-terminus and disulfide bonds was relatively stable between different batches, which may be a potential indicator for batch consistency. These findings provide a valid reference for the stability evaluation of the production process and final products.

Keywords: interferon α2b variants; LC/Q-TOF; molecular heterogeneity; N-terminus; oxidation; acetylation; deamidation

1. Introduction Among the interferons (IFNs), IFN-α2 has been the most broadly evaluated clinically. IFN-α2 is an important immune-related protein, which has multiple functions in antiviral process, antiangiogenic, antiproliferative, and proapoptotic eﬀects in cancers. The U.S. Food and Drug Administration approved human IFN-α2a and IFN-α2b in 1986 for the treatment of hairy cell leukemia [1–3]. Recombinant human IFN α2b is a 165-amino acid single chain polypeptide with a molecular weight 19.26 kDa (Chinese Pharmacopoeia, 2020 version). Recently commercial IFNα2b can be produced in Escherichia coli (E. coli), Pseudomonas, yeast or mammalian cells [4,5]. Recombinant proteins are generally heterogeneous due to post-translational modiﬁcation and chemical reactions (such as oxidation, deamidation, aggregation, and so on) during expression,

Molecules 2020, 25, 3965; doi:10.3390/molecules25173965 www.mdpi.com/journal/molecules Molecules 2020, 25, 3965 2 of 11 purification, and long-term storage [6]. Heterogeneity is mainly reflected in the difference of charge, hydrophobicity, change of molecular weight, different type of disulfide bonds, etc. [7,8]. Endogenous proteins are structurally heterogeneous but still recognized as self by the immune system; however, recombinant proteins produced in heterologous systems may include structural variants that are non-self and potentially immunogenic [9]. Protein variants can be classified as either product-related impurities that have a negative impact on safety and efficacy or product-related substances that have no negative impact on safety and efficacy [7]. Downstream processing, which includes various chromatography and filtration steps, is optimized to remove process and product related impurities. However, these steps will also have a significant impact on the nature and levels of variants at drug substance levels. It is worth noting that targeting to remove all the heterogeneity through downstream processing is not practical. Therefore, modifications and degradations that inevitably occur during the manufacturing process have a significant impact on the final product quality [7]. The detection of product variant is pivotal for quality control. Besides, monitoring the type or level of modification is expected to reflect the stabilization of production process and consistency between different batches of products. Therefore, state-of-the-art physiochemical and analytical methods as well as biological assays are required to characterize the biological products in full for therapeutic applications. To monitor the purity and the biochemical and biological properties of rhIFNα2b, appropriate molecular and biological characterization methods should be developed [10]. Mass spectrometry (MS) has the advantages of high precision and high resolution in identifying structural and molecular information of micro-sample. MS can deliver the measurement of intact molecular weight, and in combination with peptide mapping it also can identify sequence variation and site modification. High resolution MS such as Quadrupole Time-of-Fight (Q-TOF) are excellent analytical tools for assessing protein heterogeneity and for structural characterization, and high-resolution techniques such as Ultra Performance Liquid Chromatography (UPLC) have been successfully coupled to high resolution MS analyzers in cases where the sample is complex and highly heterogeneous [6,11]. In this study, seven recombinant human IFNα2b preparations from six manufacturers in China and one in America were analyzed and compared by a developed LC/Q-TOF method. IFNα2b variants and site-specific modifications were identified and quantified both in intact protein level and peptide level.

2. Results

2.1. Intact Protein Analysis

2.1.1. Purity Analysis Seven IFNα2b preparations were analyzed by a developed LC/Q-TOF to reveal the molecular heterogeneity at intact protein level. The chromatograms of UV detector at 214 nm were shown in Figure1. As listed in Table1, the area percentages of the main peak at about 16.4 min based on UV signal were above 99.0% except for 4E (74.9%) and 6P (66.0%). Molecules 2020, 25, 3965 3 of 11 Molecules 2020, 25, x FOR PEER REVIEW 3 of 11

FigureFigure 1. UV 1. chromatogramsUV chromatograms of of seven seven IFN IFNαα2b2b preparations.preparations. 1E~5E 1E~5E were were expressed expressed in E. in coliE., coli6P was, 6P was expressedexpressed in Pseudomonas in Pseudomonasand and 7Y 7Y was was expressed expressed inin Saccharomyces cerevisiae cerevisiae. .

TableTable 1. 1.Area Area percentages percentages ofof peaks based based on on UV UV signal. signal.

SampleSample No. No. Peak Peak 1 (%) 1 (%) Peak Peak 2 2 (%) (%) Peak Peak 3 3(%) (%) Peak Peak 4 (%) 4 (%) Peak 5 Peak(%) 5 (%) 1E / / 99.31 ± 0.33 0.31 ± 0.003 0.38 ± 0.33 1E // 99.31 0.33 0.31 0.003 0.38 0.33 ± ± ± 2E 2E /// / 100.00100.00 / /// 3E3E 0.51 0.510.02 ± 0.02 / / 99.4999.49 ± 0.020.02 / // / ± ± 4E 4E / / 74.8974.89 ± 0.050.05 22.18 22.18 ± 0.020.02 1.89 ± 0.07 1.89 0.07 ± ± ± 5E5E 0.19 0.190.002 ± 0.002 / / 99.8199.81 ± 0.0020.002 / // / ± ± 6P 6P // / / 65.9765.97 ± 0.060.06 13.67 13.67 ± 0.020.02 20.36 ± 20.36 0.06 0.06 ± ± ± 7Y 7Y / / 0.74 0.74 ±0.01 0.01 99.27 99.27 ± 0.0040.004 / // / ± ±

2.1.2.2.1.2. Identification Identification of IFNof IFNα2bα2b Variants Variants IFNα2b variants were identified based on MS molecular weight (MW). The deconvoluted mass IFNα2b variants were identified based on MS molecular weight (MW). The deconvoluted spectra were shown in Figure 2 and non-convoluted mass spcetra refer to Supplementary Figure S1. massThe spectra native were IFN shownα2b inmolecule Figure 2 andcontains non-convoluted 165 amino mass acid spcetra residues refer to Supplementary(CDLPQTHSLGSRRTLMLLAQMRRISLFSCLKDRHDFGFPQEEFGNQFQKAETIPVLHEMIQQIFN Figure S1. The native IFNα2b molecule contains 165 amino acid residues (CDLPQTHSLGSRRTLMLLAQMRRISLFSCLKDRHDFGFPQEEFGNQFQKAETIPVLHEMIQQIFNLFLFSTKDSSAAWDETLLDKFYTELYQQLNDLEACVIQGVGVTETPLMKEDSILAVRKYFQRITLYLK STKDSSAAWDETLLDKFYTELYQQLNDLEACVIQGVGVTETPLMKEDSILAVRKYFQRITLYLKEKKYEKKYSPCAWEVVRAEIMRSFSLSTNLQESLRSKE) and two disulfide bonds, one between Cys 1 and SPCAWEVVRAEIMRSFSLSTNLQESLRSKE)Cys 98 and another between Cys 29 and Cys and138, twoaccording disulfide to Chinese bonds, Pharmacopeia, one between Cys2020 1version. and Cys 98 and anotherThe theoretical between molecular Cys 29 weight and Cys is 19,264.91 138, according Da. As shown to Chinese in Figure Pharmacopeia, 2C, the native 2020 form version. mainly The theoreticalappeared molecular in peak weight3 at about is 19,264.9116.4 min, and Da. a As “+Met” shown form, in Figurethe direct2C, translation the native product form mainly of the appearedgene retaining the N-terminal Met, was also found in peak 3, suggesting the addition of N-terminal Met in peak 3 at about 16.4 min, and a “+Met” form, the direct translation product of the gene retaining the has no obvious influence with the retention time of IFNα2b (Figure 2C). Except for “+Met”, other N-terminal Met, was also found in peak 3, suggesting the addition of N-terminal Met has no obvious modified forms of IFNα2b were also identified: N-terminal acetylated (increase of 42.04 Da), α influenceoxidized with (increase the retention of about time 16.00 of Da IFN at one2b site (Figure and 32.002C). Da Except at two for sites), “ + Met”,and deoxidized other modified (decrease forms of IFNofα 16.002b were Da), and also all identified: of the errors N-terminal were below acetylated 1 Da. The theoretical (increase and of 42.04 measured Da), MWs oxidized of each (increase form of aboutwere 16.00 listed Da atin oneTable site 2. andOnly 32.00 a trace Da amount at two sites),of deoxidized and deoxidized form was (decreasefound in sample of 16.00 7Y, Da), but and the all of the errorsoxidized were form below was found 1 Da. in The sample theoretical 4E and 6P, and and measured the proportion MWs was of each about form 24.2% were for 4E listed and 14.7% in Table 2. Onlyfor a trace 6P (Table amount 3), which of deoxidized may be a cause form for was the found poor purity. in sample 7Y, but the oxidized form was found in sample 4E and 6P, and the proportion was about 24.2% for 4E and 14.7% for 6P (Table3), which may be a cause for the poor purity.

MoleculesMolecules 20202020, ,2525, ,x 3965 FOR PEER REVIEW 4 4of of 11 11

FigureFigure 2. 2. IdentificationIdentiﬁcation ofof IFN IFNα2bα2b variants variants based based on deconvolutedon deconvoluted mass mass spectra. spectra. (A). Oxidized (A). Oxidized IFNα2b IFNin peakα2b in 1. peak Spectrum 1. Spectrum from from 14.095 14. to095 14.420 to 14.420 min min of sampleof sample 6P. 6P. (B). (B Deoxidized). Deoxidized IFN IFNα2bα2b in in peak peak 2. 2.Spectrum Spectrum from from 15.927 15.927 to 16.098to 16.098 min min of sample of sample 7Y. (C 7Y.). Native (C). Native and +Met and IFN +Metα2b IFN in peakα2b 3.in Spectrumpeak 3. Spectrumfrom 16.509 from to 16.509 16.920 to min 16.920 of sample min of 6P. sample (D). Acetylated 6P. (D). Acetylated IFNα2bin IFN peakα2b 4. in Spectrum peak 4. Spectrum from 17.246 from to 17.24617.588 to min 17.588 of sample min of 6P. sample Input 6P. spectrum Input spectrum isotope resolution: isotope resolution: 3,000. 3,000.

Table 2. Identiﬁcation of IFNα2b variants. Table 2. Identification of IFNα2b variants. Retention IFNα2b RetentionShift of AverageShift of TheoreticalTheoretical Average MeasuredMeasured Average IFNαTime2b Error Error (Da) Variants Time MW (Da)Average MWAverage (Da) MW AverageMW (Da) Variants(min) (Da) ~14.2 (min) +16.00MW (Da) 19,280.91(Da) MW 19,280.68 (Da) 0.23 Oxi − ~14.2 +32.00+16.00 19,296.91 19,280.91 19,280.68 19,296.50 −0.23 0.41 Oxi − dOx ~16.0 16.00+32.00 19,248.1619,296.91 19,296.50 19,248.66 −0.41 0.5 − Native * ~16.4 / 19,264.91 19,264.19 0.72 dOx ~16.0 −16.00 19,248.16 19,248.66 0.5 − +Met +131.20 19,396.11 19,395.33 0.78 Native * ~16.4 / 19,264.91 19,264.19 −0.72 − ACE ~17.2 +42.04 19,306.95 19,306.49 0.46 +Met +131.20 19,396.11 19,395.33 −0.78 − * The amino acid sequence of native IFNα2b is from Chinese Pharmacopoeia, 2020 version. ACE ~17.2 +42.04 19,306.95 19,306.49 −0.46 * The amino acid sequence of native IFNα2b is from Chinese Pharmacopoeia, 2020 version. Table 3. Deconvoluted area of oxidized IFNα2b in sample 4E and 6P. Table 3. Deconvoluted area of oxidized IFNα2b in sample 4E and 6P. Sample 4E Sample 6P IFNIFNαα2b2b Variant Sample 4E Sample 6P 1 2 3 Average 1 2 3 Average Variant 1 2 3 Average 1 2 3 Average IFNα2b + M, 8690.8 8628.6 8580.7 8580.7 3604.8 3567.6 3609.5 3594.0 IFNOxidation-1α2b + IFNM,α 2b + M8690.8 24,271.3 8628.6 22,641.7 8580.7 22,905.7 8580.7 22,905.7 3604.8 23,047.2 3567.6 21,887.2 3609.5 23,409.9 3594.0 22,781.4 IFNα2b, Oxidation-1 0.0 0.0 0.0 0.0 897.2 864.6 888.4 883.4 Oxidation-2 IFNα2bIFN +α M2b, 24,271.3 22,641.7 22,905.7 22,905.7 23,047.2 21,887.2 23,409.9 22,781.4 4722.9 4960.8 4913.6 4913.6 1599.2 1727.5 1717.8 1681.5 IFNOxidation-1α2b, IFNα2b0.0 18,739.6 0.0 19,652.7 0.0 19,395.4 0.0 19,395.4 897.2 12,413.0 864.6 12,756.8 888.4 13,133.5 12,767.8883.4 Oxidation-2 IFNα2b, 4722.9 4960.8 4913.6 4913.6 1599.2 1727.5 1717.8 1681.5 Oxidation-1 IFNα2b 18,739.6 19,652.7 19,395.4 19,395.4 12,413.0 12,756.8 13,133.5 12,767.8

Molecules 2020, 25, 3965 5 of 11 Molecules 2020, 25, x FOR PEER REVIEW 5 of 11

Oxidation in Table 3. Cont.

IFNα2b +M 26.4 27.6 Sample27.9 4E 27.3 14.0 13.4 Sample 6P13.6 13.5 IFNα2b Variant (%) 1 2 3 Average 1 2 3 Average Oxidation in Oxidation in 20.126.4 20.2 27.6 20.3 27.9 20.2 27.3 20.3 14.0 13.419.8 13.620.1 13.5 20.1 IFNIFNαα2b2b+ M(%) (%) Oxidation in Oxidation in 20.1 20.2 20.3 20.2 20.3 19.8 20.1 20.1 IFNα2b (%) Oxidationtotal in total 23.8 24.3 24.5 24.2 15.1 14.5 14.8 14.7 23.8 24.3 24.5 24.2 15.1 14.5 14.8 14.7 IFNIFNαα2b2b (%) (%)

TheThe variants variants with with an an increase increase of of about about 1 1 Da Da and and 2 2 Da Da were were found, found, suggesting suggesting deamidation deamidation and and freefree thiols thiols might might exist. exist. But But the the data data accuracy accuracy was was insu insufficientfficient to to properly properly distinguish distinguish these these variants variants withwith small small di ffdifferenceerence at MW.at MW. Besides, Besides, although although a small a small quantity quantity of dimer of formsdimer wereforms found were in found sample in 6P,sample the measured 6P, the measured molecular molecular weight (about weight 38.5 (about KDa) was38.5 highlyKDa) was variable, highly which variable, could which be caused could by be multiplecaused by modifications. multiple modifications.

2.1.3.2.1.3. N-Terminal N-Terminal Heterogeneity Heterogeneity FourFour forms forms of of N-terminus N-terminus were were found: found: Met, Met, Cys, Cys, acetylated acetylated Cys, Cys, and and acetylated acetylated Met, Met, and and the the percentagepercentage of of each each form form were were shown shown in Figurein Figure3. For 3. samplesFor samples 1E, 3E, 1E, 5E, 3E, and 5E, 7Y, and only 7Y, native only native form with form N-terminalwith N-terminal Cys was Cys found. was found. For sample For sample 2E, about 2E, 10% about has 10% an addition has an addition of N-terminal of N-terminal Met. For Met. sample For 4Esample and 6P, 4E a and large 6P, proportion a large proportion of N-terminal of N-terminal Met and Met acetylated and acetylated Cys existed, Cys evenexisted, more even than more native than form,native which form, may which be another may causebe another of poor cause purity. of Moreover, poor purity. a small Moreover, amount of a acetylated-Met small amount was of foundacetylated-Met in sample was 4E (about found 1.3%). in sample 4E (about 1.3%).

α FigureFigure 3. 3.Percentage Percentage of of IFN IFNα2b2b variants variants with with di differentfferent N-terminus. N-terminus. 2.2. Evaluation of Modifications in IFNα2b by Peptide Mapping 2.2. Evaluation of Modifications in IFNα2b by Peptide Mapping 2.2.1. Peptide Map 2.2.1. Peptide Map To further study the molecular heterogeneity, the IFNα2b samples were digested by trypsin and analyzedTo atfurther peptide study level. the The molecular distribution heterogeneity, and intensity the of peaksIFNα2b in thesamples peptide were maps digested of seven by samples trypsin wereand basicallyanalyzed consistent,at peptide exceptlevel. The for distribution partial regions, and and intensity amino of acid peaks sequences in the peptide coverage maps was of above seven 99%.samples Taking were sample basically 4E for consistent, example, except the peaks for betweenpartial regions, 26 min andand 27.5amino min acid were sequences notably dicoveragefferent fromwas sampleabove 99%. 1E, for Taking substantial sample “+ 4EMet” for andexample, acetylated the peaks IFNα between2b variants 26 existedmin and in 27.5 sample min 4Ewere (Figure notably4). different from sample 1E, for substantial “+Met” and acetylated IFNα2b variants existed in sample 4E (Figure 4).

Molecules 2020, 25, 3965 6 of 11 Molecules 2020, 25, x FOR PEER REVIEW 6 of 11

Figure 4. PeptidePeptide maps of sample 1E and 4E.

2.2.2. Modifications Modifications in IFN α2b from Different Different Manufacturers ModificationsModifications atat potential potential sites sites were were next analyzednext analyzed at peptide at peptide level, including level, including remain of N-terminalremain of N-terminalMet, N-terminal Met, acetylation,N-terminal fracture acetylation, or mismatch fracture of or disulfide mismatch bond, of disulfide oxidation bond, at Met, oxidation and deamidation at Met, andat asparagine deamidation (Asp), at whichasparagine are incidental (Asp), which in recombination are incidental proteins. in recombination The results showedproteins. that The the results level showedof each modificationthat the level significantly of each modification varied among significantly different varied samples. among different samples. Four forms of N-terminus were screened according to the results of intact protein analysis: Met, Met, Cys, acetylated Cys Cys and and acetylated acetylated Met. Met. As As shown shown in in Figure Figure 5A,5A, no no Met Met was was found found in in N-terminal N-terminal of IFNof IFNα2bα 2bpreparation preparation produced produced in inSaccharomycesSaccharomyces cerevisiae cerevisiae (sample(sample 7Y), 7Y), and and the the highest highest level level of N-terminal MetMet (about(about 37%)37%) was was found found in in IFN IFNα2bα2b preparation preparation produced produced in Pseudomonasin Pseudomonas(sample (sample 6P). 6P).The The proportion proportion of N-terminal of N-terminal Met Met varied varied between between diff erentdifferent IFN αIFN2bα preparations2b preparations produced produced in E. in coli E. coliexpression expression system: system: sample sample 4E was4E was high high (35%), (35%), sample sample 2Ewas 2E was moderate moderate (16%), (16%), sample sample 3E was 3E was low low(3%) (3%) and sampleand sample 1E and 1E 5E and were 5E extremely were extremely low (below low 1%).(below Acetylated 1%). Acetylated Cys was foundCys was in sample found 4Ein sampleand 6P, and4E and only 6P, a littleand only amount a little of acetylated amount of Met acet wasylated found Met in was sample found 4E in (about sample 0.16%). 4E (about These 0.16%). results wereThese basically results were consistent basically with consistent the results with of intactthe results protein of intact analysis. protein analysis. Two disulfidedisulfide bondsbonds werewere formed formed in in native native IFN IFNα2b,α2b, with with one one between between T1 T1 (Cys1) (Cys1) and and T10 T10 (Cys (Cys 98) 98)and and another another between between T5 (Cys T5 29) (Cys and 29) T17 and (Cys T17 138). (Cys To evaluate 138). To the evaluate disulfide the bonds, disulfide we analyzed bonds, fourwe analyzedpeptides containingfour peptides Cys, containing searched “true”Cys, searched forms with “true” correct forms disulfide with correct bonds disulfide (T1 = T10 bonds and T5(T1= =T17), T10 “free”and T5 forms = T17), with “free” free thiols,forms andwith “mismatch” free thiols, formsand “mismatch” with incorrect forms disulfide with incorrect bonds (T1 disulfide= T5, T1 bonds= T17, (T1T5 = = T10T5, T1 and = T10T17,= T5T17) = T10 and and calculated T10 = T17) the and percentages calculated of the each percentages form of peptides. of each form The resultsof peptides. were Theshown results in Figure were5 B.shown The percentagein Figure 5B. of correctThe percenta disulfidege bondsof correct was disulfide the lowest bonds in sample was the 3E. lowest in sampleOxidation 3E. always happens in Met residue, so the level of oxidation in four peptides containing Met residue,Oxidation T3 always (Met 16 happens & Met 21), in Met T8 (Met residue, 59), T10so the (Met level 111), of andoxidation T18 (Met in four 148) peptides were analyzed containing and Metcalculated. residue, As T3 shown(Met 16 in & FigureMet 21),5C, T8 the (Met oxidation 59), T10 level (Met of 111), sample and T18 4E was(Met the 148) highest were analyzed followed and by calculated.sample 6P, whichAs shown is consistent in Figure with 5C, thethe resultsoxidation of intactlevel proteinof sample analysis. 4E was T3 the was highest the most followed oxidized by samplepeptide 6P, for which it has two is consistent Met residues. with the results of intact protein analysis. T3 was the most oxidized peptideAnother for it has highly two frequent Met residues. chemical modifications in proteins is the deamidation of Asp residue. FourAnother peptides highly containing frequent Asp residue,chemical T7 modifications (Asp 45), T8 in (Asp proteins 65), T10 is (Aspthe deamidation 93), and T19 of (Asp Asp 156) residue. were Fourassessed. peptides As shown containing in Figure Asp5 D,residue, Asp 156 T7 was(Asp the 45), most T8 (Asp deamidated 65), T10 site, (Asp and 93), Asp and 65 T19 was (Asp the least 156) weredeamidated assessed. site. As shown in Figure 5D, Asp 156 was the most deamidated site, and Asp 65 was the least deamidated site.

Molecules 2020, 25, 3965 7 of 11 Molecules 2020, 25, x FOR PEER REVIEW 7 of 11 Area percentage (%)

Figure 5. Modifications in IFNα2b from different manufacturers. (A). N-terminal amino acid residue; (BFigure). Disulfide 5. Modifications bonds; (C). in Oxidation IFNα2b infrom peptides different containing manufacturers. Met residue (A). N-terminal including T3, amino T8, T10 acid and residue; T18; (D(B).). Deamidation Disulfide bonds; in peptides (C). Oxidation containing in Asppeptides residue containing including Met T7, residue T8, T10 andincluding T19. T3, T8, T10 and T18; (D). Deamidation in peptides containing Asp residue including T7, T8, T10 and T19. 2.2.3. Modifications in IFNα2b of Different Batches 2.2.3.Four Modifications batches of IFNin IFNα2bα preparations2b of Different from Batches the same manufacturer were analyzed to evaluate the batch-to-batchFour batches consistency of IFNα in2b modification. preparations The from constitution the same ofmanufa N-terminus,cturer disulfidewere analyzed bonds, to oxidation evaluate andthe deamidationbatch-to-batch were consistency shown in Figurein modification.6. No significant The constitution di fference wasof N-terminus, found in the disulfide constitution bonds, of Noxidation terminus and among deamidation different batches were shown (Figure in6A). Figure The proportion6. No significant of “mismatch” difference and was “free” found disulfide in the bondsconstitution in S01 andof N S02terminus was a among little higher different than batche S03 ands (Figure S04 (Figure 6A). The6B). propor As to thetion oxidationof “mismatch” rate, T3and and“free” T10 disulfide in S01 and bonds S02 werein S01 higher and thanS02 was S03 a and little S04, higher while than the imageS03 and is theS04 reverse (Figure for 6B). T8 As and to T18 the (Figureoxidation6C). rate, The overallT3 and deamidation T10 in S01 and rate inS02 S01 were and higher S02 were than lower S03 thanand S03S04, and while S04 the (Figure image6D). is the reverse for T8 and T18 (Figure 6C). The overall deamidation rate in S01 and S02 were lower than S03 and S04 (Figure 6D).

Molecules 2020, 25, 3965 8 of 11 Molecules 2020, 25, x FOR PEER REVIEW 8 of 11

Figure 6. Modification in IFNα2b of different batches from the same manufacturer. (A). N-terminal aminoFigure acid 6. Modification residue; (B). Disulfidein IFNα2b bonds; of different (C). Oxidation batches from in peptides the same containing manufacturer. Met residue (A). N-terminal including T3,amino T8, T10acid and residue; T18; (D(B).). Deamidation Disulfide bonds; in peptides (C). Oxidation containing in Asp peptides residue containing including T7,Met T8,residue T10 andincluding T19. T3, T8, T10 and T18; (D). Deamidation in peptides containing Asp residue including T7, T8, T10 and T19. 3. Discussion 3. DiscussionProtein synthesis is always initiated with either Met or N-formyl Met. For most proteins, the N-terminalProtein Met synthesis residue is is always removed initiated enzymatically with either after Met the or initiation N-formyl of Met. translation, For most which proteins, is often the crucialN-terminal for the Met function residue and is removed stability enzymatically of proteins [12 afte–14r]. the In initiationE. coli, this of translation, step is controlled which byis often the methionylcrucial for aminopeptidase the function and (MetAP) stability [13]. of If theproteins next residue[12–14]. in In the E. sequence coli, this has step a small is controlled side chain by such the asmethionyl Cys, just aminopeptidase like IFNα2b, the (MetAP) removal [13]. of N-terminal If the next Met residue is more in the likely sequence to happen has [a15 small,16]. However,side chain thesuch N-terminal as Cys, just Met like can IFN neverα2b, be the completely removal removed. of N-terminal Consequently, Met is more this modificationlikely to happen can still[15,16]. be observedHowever, in the substantial N-terminal amounts Met can in never mature be IFN compleα2b obtainedtely removed. from Consequently,E. coli expression this systems modification [17], whichcan still was be further observed confirmed in substantial by this study.amounts N-terminal in mature Met IFN residueα2b obtained was detected from inE. samples coli expression 1E~5E expressedsystems [17], in E. which coli and was 6P expressedfurther confirmed in Pseudomonas by thisat study. peptide N-terminal level, although Met residue not detected was detected in sample in 1E,samples 3E, and 1E~5E 5E at expressed intact protein in E. coli level. and However, 6P expressed no N-terminal in Pseudomonas Met couldat peptide be detected level, although in sample not 7Ydetected expressed in sample in Saccharomyces 1E, 3E, and cerevisiae 5E at ,intact suggesting protein that level. a more However, rigorous no mechanism N-terminal to Met remove could the be N-terminaldetected in Met sample existed 7Y in eukaryoticexpressed cells.in Saccharomyces The removal cerevisiae of Met had, suggesting no effect on that the retentiona more time,rigorous so twomechanism forms co-existed to remove in peakthe N-terminal 3(Figure1). Met Current existed methods in eukaryotic for purity cells. analysis The removal such as RP-HPLC of Met had and no SEC-HPLCeffect on the hardly retention separate time, “+ soMet” two from forms native co-existed IFNα2b, in peak which 3(Figure necessitates 1). Current the use methods of high-resolution for purity MSanalysis methods. such as RP-HPLC and SEC-HPLC hardly separate “+Met” from native IFNα2b, which necessitatesIn eukaryotic the use cells, of hi upgh-resolution to 98% proteins MS are methods. N-terminally acetylated, and acetylation could stabilize the proteinIn eukaryotic by protecting cells, itup from to 98% degradation proteins via are N-terminal N-terminally proteases acetylated, [18]. Althoughand acetylationN-terminal could acetylationstabilize the was protein considered by protecting to be infrequent it from in prokaryotic degradation cell via [18 ,N-terminal19], in this case proteases acetylation [18]. was Although found inN samples-terminal 4E acetylation and 6P, which was wasconsidered consistent to withbe infrequent a previous in report prokaryotic [17], suggesting cell [18,19] that, N-terminalin this case acetylationacetylation may was occur found in recombinantin samples 4E proteins and 6P, from which bacteria was expressionconsistent systems.with a previous It was reported report that[17], acetylationsuggesting could that N-terminal occur at N-terminal acetylation Cys may residue occur in in IFN recombinantα2b, and the proteins acetylated from formbacteria had expression only 10% ofsystems. the activity It was of reported native molecule that acetylation [17]. To overcomecould occur this at N-terminalN-terminal heterogeneity,Cys residue in manyIFNα2b, strategies and the acetylated form had only 10% of the activity of native molecule [17]. To overcome this N-terminal heterogeneity, many strategies have been developed, such as engineered interferon derivatives with

Molecules 2020, 25, 3965 9 of 11 have been developed, such as engineered interferon derivatives with phenylalanine residue directly downstream of the N-terminal Met [20], and co-expression of engineered MetAP in E. coli [14]. The formation of disulfide bonds is necessary to the spatial structure of proteins. We observed over 10% of “free” and “mismatch” forms in all samples including five produced in E.coli, one in Pseudomonas and one in Saccharomyces cerevisiae. The percentages of correct forms varied between different manufacturers, and relatively stable between different batches, demonstrating the status of disulfide bonds were related to expression procedure. Oxidation of Met residues and deamidation of Asp residues are two of the most common forms of chemical modifications that compromise the efficacy of therapeutic proteins [21]. In intact protein analysis, oxidized IFNα2b was only found in sample 4E (24.2%) and 6P (14.7%), but in peptide mapping, oxidized peptides were found in all seven samples, suggesting that oxidation could happen during sample preparation. Moreover, lesser sensitivity and multisite modifications could also be reasons for fewer variants were found in intact protein analysis. These data indicate that although inevitable, these chemical modifications could be restricted by optimizing the manufacturing process and storage conditions.

4. Materials and Methods

4.1. Materials

Iodoacetamide (IAA), guanidine hydrochloride, ammonium bicarbonate (NH4HCO3), and trifluoroacetic acid (TFA) were obtained from Sigma-Aldrich (St. Louis, MO, USA), acetonitrile (ACN) and water were from Fisher Scientific (Ottawa, Canada). All reagents were of analytical grade. Trypsin was purchased from Promega (Madison, WI, USA), and 3kD ultrafiltration device was from Millipore (Tullagreen, Ireland). Recombinant human IFNα2b preparations 1E~5E, 6P, and 7Y were archived samples that had been preserved at 80 C in National Institute for Food and Drug Control (NIFDC), China, provided − ◦ by six manufacturers from China and one manufacturer from America. Samples were anonymised and identified via randomly assigned codes 1–7. Samples 1E~5E were manufactured in E. coli expression system, 6P was manufactured in Pseudomonas expression system and 7Y was manufactured in Saccharomyces cerevisiae expression system. Sample 2E_1~4 were four batches of 2E from the same manufacturer. The production dates of these samples ranged between 2017 and 2018, and the analyses were conducted in 2019.

4.2. Sample Preparation For intact protein analysis, all samples were directly diluted to 5 mg/mL in solvent A (98% water, 2% ACN, 0.1% TFA) for LC/MS analysis. For peptide mapping, 100 µg of proteins in each sample were denatured in 6 M guanidine hydrochloride solution, and then alkylated with 100 mM IAA in the dark for 30 min. The solution was centrifuged at 14,000 g through 3 kD filters to remove the guanidine hydrochloride, and the × intercepted proteins were rinsed twice with 50 mM NH4HCO3. Protein digestion was done by adding 2 µg of trypsin to each sample and followed by incubation for 15 h at 37 ◦C. Finally, the peptides were centrifuged at 14,000 g for 15 min through 3 kD filters and the filtrate was adjusted to a concentration × of 1 µg/µL with solvent A.

4.3. LC-MS/MS The intact protein and peptide mapping samples were all analyzed with an UPLC, ExionLC coupled with a Q-TOF MS system, X500B (SCIEX, Framingham, MA, USA). The Photo-Diode Array (PDA) detector was integrated into the UPLC system, the wave length was set to 214 nm. The mobile phase consisted of solvent A (98% water, 2% ACN, 0.1% TFA) and solvent B (98% ACN, 2% water, 0.1% TFA). The wavelength of UV detector was set to 214 nm. SCIEX OS 1.4 software (SCIEX, Framingham, MA, USA) was used for control of data acquisition. Molecules 2020, 25, 3965 10 of 11

For intact protein analysis, the proteins were loaded on a BEH C18 column (1.7 µm, 2.1 mm × 100 mm, Waters, Milford, MA, USA) and separated with a 35 min gradient from 35–60% solvent B over 20 min. The ﬂow rate was set to 0.2 mL/min. For MS acquisition, positive ion mode was applied. The spray voltage was set to 4500 V, the declustering potential (DP) was 145 V, the curtain gas was 30 psi, and the heated interface was set to 500 ◦C. The mass range m/z was set to 600~5000 with a 1 s accumulation time. For peptide mapping analysis, the peptides were separated with a BEH C18 column (1.7 µm, 2.1 mm 100 mm, Waters, Milford, MA, USA) and a 45 min gradient from 5–55% solvent B. The ﬂow rate × was 0.2 mL/min. For MS acquisition, the spray voltage was set to 5500 V, DP was 100 V, the curtain gas was 30 psi, and the heated interface was set to 550 ◦C. The information dependent acquisition (IDA) mode was applied. For TOF MS scan, positive ion mode was applied, the mass range m/z was set to 200~2,000 with a 250 ms accumulation time. For TOF MS/MS scan, 15 candidate ions per TOF MS scan were selected for analysis, dynamic collision energy was applied. The mass range m/z was set to 100~2,000 with a 50 ms accumulation time. SCIEX OS 1.4 software (SCIEX, Framingham, MA, USA) was applied in viewing and processing raw data. BioPharmaView 3.0 (SCIEX, Framingham, MA, USA) was used in the multiple attributes monitoring (MAM) analysis of intact protein and peptide mapping.

5. Conclusions Molecular heterogeneity was ubiquitous in recombinant proteins for post-translational modification and chemical reactions during expression, purification, and long-term storage. Identifying these modifications is an essential element of quality control of therapeutic recombinant proteins and offers critical quality attribute data to document and control process stability. Our study detected IFNα2b variants and site-specific modifications in seven therapeutic IFNα2b preparations from different manufacturers, suggesting that the level of oxidation and deamidation were vulnerable to many factors, which were hard to control, and the constitution of N-terminus and disulfide bonds was relatively stable between different batches, which may be a valid indicator for batch consistency. These findings provide a valid reference for the stability evaluation of the production process and final products.

Supplementary Materials: The following are available online at http://www.mdpi.com/1420-3049/25/17/3965/s1, Figure S1: Non-deconvoluted mass spectra of IFNα2b variants. Author Contributions: Conceptualization, D.P.; methodology, Y.Z.; software, Y.Z.; validation, L.T. and Y.L.; formal analysis, L.Y., L.T. and Y.L.; investigation, D.P.; resources, C.R.; data curation, Y.Z. and L.Y.; writing—original draft preparation, L.Y. and L.T.; writing—review and editing, D.P. and C.R.; visualization, L.Y. and L.T.; supervision, C.R.; project administration, C.R.; funding acquisition, C.R. All authors have read and agreed to the published version of the manuscript. Funding: This work was financially supported by grants from the National Science and Technology Major Projects of China [grant number 2018ZX09101001-003]; and the National drug standard improvement project of China [grant number 2019S04]. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. Conflicts of Interest: The authors declare no conflict of interest.

References

1. Pestka, S. The Interferons: 50 Years after Their Discovery, There Is Much More to Learn. J. Biol. Chem. 2007, 282, 20047–20051. [CrossRef][PubMed] 2. Tarhini, A.A.; Gogas, H.; Kirkwood, J.M. IFN-alpha in the treatment of melanoma. J. Immunol. 2012, 189, 3789. [CrossRef] [PubMed] 3. Ghosh, D.; Ghosh, D.; Parida, P. Physiological Proteins in Therapeutics: A Current Review on Interferons. Mini Rev. Med. Chem. 2016, 16, 947. [CrossRef][PubMed] 4. Wang, Y.; Liu, D.; Crowell, L.E.; Love, K.R.; Hancock, W.S. The application of HPLC/MS analysis with a multi-enzyme digest strategy to characterize diﬀerent interferon product variants produced from Pichia pastoris. Amino Acids 2019, 51, 1353. [CrossRef] Molecules 2020, 25, 3965 11 of 11

5. Parker, B.S.; Rautela, J.; Hertzog, P. Antitumour actions of interferons: Implications for cancer therapy. Nat. Rev. Cancer 2016, 16, 131–144. [CrossRef] 6. Carini, M.; Regazzoni, L.; Aldini, G. Mass spectrometric strategies and their applications for molecular mass determination of recombinant therapeutic proteins. Curr. Pharm. Biotechnol. 2011, 12, 1548. [CrossRef] 7. Liu, H.; Nowak, C.; Shao, M.; Ponniah, G.; Neill, A. Impact of cell culture on recombinant monoclonal antibody product heterogeneity. Biotechnol. Prog. 2016, 32, 1103. [CrossRef] 8. Hsiao, K.; Alves, J.; Patel, R.; Adams, M.; Nashine, V.; Goueli, S. A High-Throughput Bioluminescent Assay to Monitor the Deamidation of Asparagine and Isomerization of Aspartate Residues in Therapeutic Proteins and Antibodies. J. Pharm. Sci. 2017, 106, 1528. [CrossRef] 9. Jefferis, R. Biologics: Structural heterogeneity and immunogenicity. Br. J. Hosp. Med. 2017, 78, 443. [CrossRef] 10. Liu, Y.H.; Wylie, D.; Zhao, J.; Cure, R.; Cutler, C.; Cannon-Carlson, S.; Yang, X. Mass spectrometric characterization of the isoforms in Escherichia coli recombinant DNA-derived interferon alpha-2b. Anal. Biochem. 2011, 408, 105. [CrossRef] 11. Tao, L.; Pei, D.N.; Han, C.M.; Chen, W.; Rao, C.M.; Wang, J.Z. Characterization and comparison of interferon reference standards using UPLC-MS. Yao Xue Xue Bao Acta Pharm. Sin. 2015, 50, 75. 12. Sherman, F.; Stewart, J.W.; Tsunasawa, S. Methionine or not methionine at the beginning of a protein. Bioessays 1985, 3, 27. [CrossRef][PubMed] 13. Ben-Bassat, A.; Bauer, K.; Chang, S.Y.; Myambo, K.; Boosman, A.; Chang, S. Processing of the initiation methionine from proteins: Properties of the Escherichia coli methionine aminopeptidase and its gene structure. J. Bacteriol. 1987, 169, 751–757. [CrossRef][PubMed] 14. Liao, Y.D.; Jeng, J.C.; Wang, C.F.; Wang, S.C.; Chang, S.T. Removal of N-terminal methionine from recombinant proteins by engineered E. coli methionine aminopeptidase. Protein Sci. 2004, 13, 1802. [CrossRef][PubMed] 15. Hirel, P.H.; Schmitter, M.J.; Dessen, P.; Fayat, G.; Blanquet, S. Extent of N-terminal methionine excision from Escherichia coli proteins is governed by the side-chain length of the penultimate amino acid. Proc. Natl. Acad. Sci. USA 1989, 86, 8247–8251. [CrossRef][PubMed] 16. Bradshaw, R.A.; Brickey, W.W.; Walker, K.W. N-terminal processing: The methionine aminopeptidase and N alpha-acetyl transferase families. Trends Biochem. Sci. 1998, 23, 263. [CrossRef] 17. Ahsan, F.; Arif, A.; Mahmood, N.; Afza Gardner, Q.-T.-A.; Rashid, N.; Akhtar, M. Characterization and bioassay of post-translationally modified interferon alpha-2b expressed in Escherichia coli. J. Biotechnol. 2014, 184, 11. [CrossRef] 18. Johnson, M.; Coulton, A.T.; Geeves, M.A.; Mulvihill, D.P. Targeted Amino-Terminal Acetylation of Recombinant Proteins in E. coli. PLoS ONE 2010, 5, e15801. [CrossRef] 19. Arnesen, T.; Van Damme, P.; Polevoda, B.; Helsens, K.; Evjenth, R.; Colaert, N.; Varhaug, J.E.; Vandekerckhove, J.; Lillehaug, J.R.; Sherman, F.; et al. Proteomics analyses reveal the evolutionary conservation and divergence of N-terminal acetyltransferases from yeast and humans. Proc. Natl. Acad. Sci. USA 2009, 106, 8157. [CrossRef] 20. Ahsan, F.; Gardner, Q.A.; Rashid, N.; Towers, G.J.; Akhtar, M. Preventing the N-terminal processing of human interferon α-2b and its chimeric derivatives expressed in Escherichia coli. Bioorg. Chem. 2018, 76, 294–302. [CrossRef] 21. Bandi, S.; Singh, S.M.; Shah, D.D.; Upadhyay, V.; Mallela, K.M. 2D NMR Analysis of the Effect of Asparagine Deamidation Versus Methionine Oxidation on the Structure, Stability, Aggregation, and Function of a Therapeutic Protein. Mol. Pharm. 2019, 16, 4621–4635. [CrossRef][PubMed]

Sample Availability: Samples of rhIFNα2b are available from the authors.

© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).