Tackling the increasing complexity of therapeutic monoclonal antibodies with mass spectrometry

Trends

Trends in Analytical Chemistry, Vol. 48, 2013

Tackling the increasing complexity of therapeutic monoclonal antibodies with mass spectrometry Sara Rosati, Natalie J. Thompson, Albert J.R. Heck Mass spectrometry (MS) is emerging as an efficient method for the structural analysis of various new antibody-based therapeutic products. Here, we review the current trends and describe recent applications that highlight the use of MS to tackle the increasing complexity of monoclonal antibodies. ª 2013 Elsevier Ltd. All rights reserved. Keywords: Analytical method; Antibody (Ab); Antibody-drug conjugate (ADC); Biopharmaceutical; Biosimilar; Bispecific antibody; Glycosylation; Mass spectrometry; Native mass spectrometry; Sequence variant

1. Introduction Sara Rosati, Natalie J. Thompson, Albert J.R. Heck* Biomolecular Mass Spectrometry and Proteomics, Bijvoet Center for Biomolecular Research and Utrecht Institute for Pharmaceutical Sciences, Utrecht University, Padualaan 8, 3584 CH Utrecht, The Netherlands Sara Rosati, Natalie J. Thompson, Albert J.R. Heck Netherlands Proteomics Centre, Padualaan 8, 3584 CH Utrecht, The Netherlands

*

Corresponding author. Tel.: +31 (0)30 2536797; Fax: +31 (0)30 2536919.; E-mail: [email protected]

72

The use of macromolecular therapeutics has expanded very much in recent decades [1]. A great contribution to this growth comes from the introduction of therapeutic monoclonal antibodies (mAbs) [2]. Around 30 different Ab-based biopharmaceuticals have been approved by regulatory authorities for clinical use, predominantly of the IgG class, which is also naturally the most abundant class in humans. Mostly, they are used in cancer therapy, immunological diseases and infectious diseases [3]. A particular feature of these mAbs is their high specificity to their targets that generally leads to high patient tolerance, though many efforts are currently made to decrease immunogeneity issues [4]. IgG Abs consist of four polypeptide chains: two light and two heavy chains fused together to give rise to the characteristic Y shape (Fig. 1). Both intermolecular and intramolecular disulfide bridges contribute to the stability of the Ab. Complementarity-determining regions (CDRs) (i.e. the regions where the antigen binds) are situated at the extremity of each of the two antigen-binding fragments (Fabs). Since the introduction of mAbs in the clinic about 30 years ago, major advancements have been made to

optimize properties (e.g., immunogenicity, efficacy, and pharmacokinetics). From the early murine-based mAbs, through chimeric and humanized mAbs, we currently see mostly fully human mAbs entering the clinic. Furthermore, new entities are gaining popularity beside the conventional mAb products. Amongst them, Fab fragments, Ab-drug conjugates (ADCs), and bispecific Abs, have appeared and are already penetrating the market [1]. As the increasing molecular complexity of these new, improved therapeutic products puts a strain on their analysis, a number of different analytical techniques [e.g., separation techniques, nuclear magnetic resonance (NMR), circular dichroism (CD)] are currently used in the industry to cover multiple aspects of Ab analysis. [5,6] In this review, we focus on mass spectrometry (MS) as one of the most versatile technologies for the in-depth structural analysis of mAbs. MS can be used for the characterization of small and large alterations in the mAbs [e.g., sequence variants, diversity in glycosylation profiles, and other post-translational modifications (PTMs), including disulfide shuffling]. It may also be used to analyze highly complex products [e.g., mixtures of (bispecific) Abs or ADCs]. Based on these characteristics, MS has become an essential tool in the biopharmaceutical industry, for not only the analysis of new

0165-9936/$ - see front matter ª 2013 Elsevier Ltd. All rights reserved. doi:http://dx.doi.org/10.1016/j.trac.2013.02.013

Trends in Analytical Chemistry, Vol. 48, 2013

Trends

Figure 1. Structure of an IgG1 antibody. VH, Heavy chain variable domain; VL, Light chain variable domain; CH1, Heavy chain constant domain1; CH2, Heavy chain constant domain2; CH3, Heavy chain constant domain3; CL, Light chain constant domain; Fab, Antigen binding fragment; Fc, Crystallizable fragment.

Ab-based products, but also the evaluation of their biosimilars.

2. Discussion 2.1. Mass-spectrometry instrumentation used for the structural analysis of antibodies The complex heterogeneity of mAbs is an analytically challenging system to characterize in depth. Traditionally, various techniques have been used in industry and academia alike to analyze mAbs. More recently, MS has been emerging as a versatile analytical technique that can address and elucidate many aspects of mAbs and be complementary to investigations by other approaches. MS can handle, with relative ease, single amino-acid substitutions and PTMs, but also binding interactions and global structural changes [7–11]. Until relatively recently, MS was relegated to small-molecule analysis because of the inability to transfer large biomolecules to the gas phase efficiently. However, in the late 1980s, the development of the ‘‘soft’’ ionization techniques, matrixassisted laser desorption ionization (MALDI) [9] and electrospray ionization (ESI) [12], opened MS to the world of macromolecules and thus also to intact protein analysis. Indeed, one of the first impressive demonstrations of the power of MALDI in 1988 was in the analysis of intact mAbs [13–15]. Initial ESI-based experiments analyzed mAbs under denaturing conditions in which acidic and organic solutions were used to transfer unfolded proteins to the gas phase. ESI was then coupled with time-of-flight (ToF) and triple-quadrupole mass analyzers, which possess a suitable mass range for studying

denatured proteins (i.e. up to m/z 4000) [7]. With this type of instrumentation, the precise molecular weight can be determined and, therefore, the purity and the heterogeneity of mAbs. However, under these denaturing conditions intramolecular and intermolecular interactions become disrupted, meaning that Ab tertiary structure and possible non-covalent complexes formed by these interactions cannot be studied. Around 2000, an alternative method emerged circumventing this issue, termed native MS, through which the non-covalent interactions that define protein tertiary and quaternary structure are largely preserved after ionization, thereby making MS compatible with the world of structural biology [16,17]. In native MS, a volatile buffer at neutral pH (e.g., aqueous ammonium acetate) is used with ‘‘gentle’’ nanoESI elution to transfer structurally intact proteins into the gas phase. Under these conditions, the proteins acquire fewer charges, thus pushing the detection window to a mass range over m/z 4000. To detect these high m/z ions successfully, ToF mass analyzers are typically used. The mass accuracy of ESIToF instruments for a protein the size of an intact Ab is typically 10–100 ppm, and the instrument mass resolution is 5000–10 000 (full width at half maximum resolution, FWHM) for state-of-the-art ToF and quadrupole-ToF (Q-ToF) instruments. With respect to intact Abs, the width of the isotopic distribution is approximately 25 Da making a resolution higher than 10 000 not particularly advantageous for mass measurement of mAbs [7]. Using native MS in conjunction with ToF mass analysis, possessing a theoretically infinite mass range, protein complexes up to megaDalton molecular weight (e.g., intact virus capsids) can now be analyzed

http://www.elsevier.com/locate/trac

73

Trends

Trends in Analytical Chemistry, Vol. 48, 2013

[18–21]. Until recently, ESI-ToF and ESI-Q-ToF represented the instruments of choice for the analysis of intact mAbs. These instruments are available from nearly all main MS vendors. 2.2. Analysis of sequence variants With the increased use of recombinant Abs as therapeutics, precise and detailed characterization of the primary structure is necessary as single mutations and/ or truncations can affect not only the efficacy but also the safety of the therapeutic. It has been demonstrated that unintended amino-acid substitutions, also known as sequence variants, have been increasingly detected in recombinant proteins produced using mammalian cell cultures [22]. These sequence variants can stem from multiple sources, including mutations at the DNA level, mis-incorporation of amino acids during protein assembly, or mis-cleavage during post-translational processing [23]. Relying on a robust basis in proteomics, MS has also been incorporated into mAb sequence-variant-analysis workflows due to the ability not only to detect but also to identify most types of sequence variants, as these produce indicative mass shifts. Sequence variants, often consisting of a few point mutations throughout the protein structure, are commonly small in mass relative to the intact mAb mass. Although a general procedure for the detection of sequence variants using MS in the intact mAb protein has been reported [24], the small mass differences and low populations of sequence variants, and current instrumentation limits in sensitivity and resolution, have resulted in the greater popularity of using bottomup approaches instead of top-down. For the analysis of the primary structure of proteins, the bottom-up MS-based method incorporates one or more proteases used to produce peptides of the mAbs. These subsequent peptides are separated using liquid chromatography (LC) and fragmented using tandem mass spectrometry (MS/MS). The MS/MS data are then searched against a database for identification. However, for those samples for which there is no sequence information at the gene level, this method must be adapted to a more de novo sequencing method. An example of such an adaptation has been shown by Perdivara et al. in the determination of the sequence of an Ab produced in hybridomas [25]. The MS/MS data were searched against a database containing Abs and common fragments, which assisted in the sequence positioning. This combined with overlapping peptides allowed the determination of each domain as well as the CDRs, adhering to the Kabat rules [26]. Detailed analysis of the MS/MS spectra revealed differences between the unknown Ab sequence and those previously recorded in the database and micro-heterogeneity in the CDR2 of the light chain and N-terminal truncation of the heavy chain. This example highlights the power of LC-MS/MS for not only 74

http://www.elsevier.com/locate/trac

protein sequencing but also in-depth characterization of variations across multiple samples. We wish to highlight two recent reports that illustrate the crucial incorporation of MS for the analysis of sequence variants. In 2010, Yang et al. published a method that combined both UV and MS analysis after LC for detection, identification, and quantification of lowlevel sequence variants [23]. This method utilized the error-tolerant search (ETS) mode developed by Creasy and Cottrell [27] and currently available through Mascot for the identification of sequence variants. To ensure the detection of low-abundance variants, the dynamic exclusion (DE) function was used during MS/MS to increase the data collected for low-intensity ions, which resulted in the detection of sequence variants present even below levels of 5%. With this MS-based method, they detected a sequence variant at 0.3% of the base protein level of a recombinant humanized Ab. However, it was observed that ETS was only appropriate for those sequence variants that resulted from a single mutation, as the software could not predict shifts due to multiple substitutions. Also, ETS produced several false positives, alternately assigned to certain effects (e.g., sodium adduction and S-carboxymethylation), making manual verification essential. With this knowledge, Zeck et al. implemented the use of a biological reference instead of ETS analysis using a theoretical reference [22]. In this method, LC-MS/MS runs of mAb digests are compared one-on-one using the SIEVE software in addition to the previous method of using the Mascot ETS analysis. Differences detected by SIEVE, which analyzes the data set based on retention time and m/z window for peak matching, are plotted as a ratio and allowed easy identification of those peptides that most probably contain a sequence variant. In this manner, those sequence variants that resulted from multiple amino-acid substitutions were detected, but confirmation of these substitutions still required manual verification. While all of these methods have successfully detected variations in the primary structure of mAbs, they require several sample-preparation steps and extensive data interpretation. As mentioned previously, limitations in MS instrumentation have so far somewhat precluded the use of intact mAbs for this analysis, but novel improvements in dynamic range and in mass resolution may bring intact protein analysis by MS back to the forefront of sequence variants analysis in the near future. 2.3. Evaluation of biosimilars As for small-molecule drugs, ‘‘copy versions’’ of biopharmaceuticals can be commercialized after expiry of the patent of the reference products. However, the size and the complexity of these molecules, partly caused by their ‘‘non-synthetic’’ production in biological cellular platforms, make it almost impossible to produce an exact copy of a biopharmaceutical reference product. Hence,

Trends in Analytical Chemistry, Vol. 48, 2013

small differences compared to the reference product are at present still accepted by the regulatory authorities (e.g., FDA and EMA), who have termed these types of drugs ‘‘biosimilars’’ [28]. How similar does a biosimilar mAb have to be in order to enter the market and how may this be evaluated? Generally, qualitative and quantitative differences in PTMs are still accepted by the regulatory authorities, but primary sequence variations are not allowed [29]. Yet, as even minimal structural variations can affect efficacy and safety, preclinical and clinical trials along with biological and physicochemical characterization are generally required. The extent of molecular similarity may influence the level of preclinical and clinical trials requested by the regulatory authorities [30–32]. An indepth analytical investigation is therefore essential. As concisely described by Berkowitz et al., three main characteristics of biosimilars need to be assessed and may be used to determine similarity to the reference product: PTMs, three-dimensional structure, and protein-aggregation state. From their recent evaluation, it was apparent that MS can play an important role in each of these areas [5]. Here, we highlight a study carried out by Xie et al., which represents a good example of the employment of MS for the comparison of a biosimilar mAb and its reference product, with respect to sequence variants and PTMs [33]. Their work shows how a standard bottom-up approach yields identification and quantification of primary-sequence alterations and describes how glycosylation studies can be performed at the glycopeptide level and on enzymatically-released glycans using a mass spectrometer employing MALDI for ionization. It is worth noting that, prior to all bottom-up experiments, Xie et al. performed MS analysis at the intact Ab level, allowing detection of differences in primary sequence and glycosylation profiles at the intact protein level. In particular, qualitative and semi-quantitative assessment of the glycosylation profiles is possible at the intact protein level: mass differences of different glycans allow assessment of glycosylation heterogeneity and to some extent sugar-chain composition. However, due to the branched nature of the sugar chains, exact glycan structures cannot easily be solved by MS alone. An intrinsic disadvantage of intact protein analysis is that it is impossible to localize the exact site of modifications, so bottom-up experiments are necessary for this purpose. Insights into the region bearing the alterations were obtained by Xie et al., following reduction of the disulfide bridges in the intact Ab and subsequent analysis of the resulting single heavy and light chains. This type of information, although not very helpful for sequence-variant mapping, can be quite useful for glycosylation profiling, as it is well known that this modification occurs at Asn297 in the CH2 domain, though rarely found in the Fab region. Nevertheless, MS analysis of intact Abs is still valuable, generally requiring

Trends

minimal sample preparation, minimizing potential artificial in vitro alterations. MS analysis of intact Abs holds great potential to play a major role in not only assessment of similarities in biosimilars, but also quality-control assays for biopharmaceutical products. As structural differences occur in the attempt to ‘‘copy’’ a reference product, it is also true that structural alterations, although minimal, can occur among different batches of the same product. A fast assay that can reveal the presence of minor alterations can therefore be very useful for routine analysis. Finally, analysis of intact Abs can further benefit from improvements in the resolution of instruments, which we discuss in sub-section 2.8 below. 2.4. Analysis of monoclonal-antibody disulfide linkages A crucial aspect of the mAb structure is the correct localization of the existing disulfide bonds. It has become apparent that these disulfide bonds may get ‘‘scrambled’’ (i.e. breaking and making new disulfide bonds). Such processes not only change the tertiary structure, but also can affect the efficacy of the therapeutic mAb. The exact location of disulfide bonds and the tertiary structural integrity of the mAbs are therefore linked, and must be characterized in great detail. Recently, Wang et al. reported on an LC-MS/MS method to assess the location of disulfide interactions in mAbs accurately [34]. They used electron-transfer dissociation (ETD) in conjunction with collision-induced dissociation (CID) to locate disulfide bonds in peptides from an enzymatically-digested mAb. The combined use of ETD and CID on disulfide-bound peptides results in complementary fragmentation patterns, where CID typically produces backbone fragmentations and ETD preferentially cleaves the disulfide bonds (Fig. 2). A comparison of peptide masses with a list of all possible disulfide-bound tryptic peptides was used to determine which peptides were most likely to contain a disulfide bond, and selected peptides were subsequently dissociated by ETD then CID. With this method, the expected disulfide bonds and those resulting from scrambling under heat stress were readily detected. The effects of disulfide-bond scrambling on the mAb structure can also be evaluated by ion mobility spectrometry (IMS) [35]. IMS measures the drift time of ions, thereby providing a rotationally-averaged gas-phase collision cross-section, which can be correlated to the volume and shape of the analyte. Using IMS, different isomers of IgG2 were resolved and possessed altered disulfide linkages [36]. The various co-existing mAb structures could be separated by mobility measurements of IgG2 using also point mutations in the cysteine residues in the mAbs hinge region. In these two studies, IMS has been shown to be a powerful method to assess potential disulfide scrambling at both primary and tertiary structural levels. http://www.elsevier.com/locate/trac

75

Trends

76 http://www.elsevier.com/locate/trac

Trends in Analytical Chemistry, Vol. 48, 2013

Figure 2. Tandem mass spectra illustrating the use of ETD and CID for the localization of disulfide bonds in mAbs. Here, in particular, ETD was used to investigate and to prove the existence of disulfide bond ‘‘scrambling’’ in anti-HER2. Three MS/MS fragmentation modes are illustrated: (A) CID-MS2 for peptide-backbone fragmentation with the intact disulfide bond yielding some additional b and y ions; (B) ETD-MS2 for the preferential cleavage of the disulfide bond into the two peptides, P1 and P2, and minor backbone fragmentation resulting in c and z ions; (C) CID-MS3 of the P1 peptide released during ETD of the disulfide bond to confirm the identity of the peptides linked by the disulfide bond. (Reprinted from [34] with permission of ACS Publications, ª 2011).

Trends in Analytical Chemistry, Vol. 48, 2013

2.5. Analysis of Ab-drug conjugates Ab-drug conjugates (ADCs) are a rapidly emerging class of therapeutic mAbs, in which the mAb backbone is covalently linked to small cytotoxic drugs through a linker molecule. This type of biotherapeutic exploits the high selectivity of mAbs to selectively deliver drugs to target (tumor) cells, thus potentially limiting side effects [37–39]. The drug is generally linked to either the cysteine or lysine residues in the mAbs. Unfortunately, this conjugation process can often not be fully controlled with regard to the number of drug molecules linked to the Ab and to their distribution and exact locations. Therefore, the final product will be a mixture of Abs differing in load and location of the drug molecules, requiring thorough analysis to characterize them [40]. Because of differences in molecular mass caused by the different amount of drug loads, MS can be a valuable technique for the characterization of ADCs. To investigate the composition of the mixture, it is indispensable to perform the analysis at the intact protein level – in particular, when the drug is linked via cysteine residues, as such an attachment requires partial reduction of the disulfide bridges, potentially making it such that the Ab loses some of its interchain covalent bonds. As a result, with the denaturing conditions of conventional LC-MS analysis, sub-units could fall apart, thus losing any information about the composition of the intact functional Ab. In this case, native MS becomes the technique of choice because of its ability to preserve non-covalent interactions. A pioneering valuable example of the characterization of ADCs using native MS has been described by ValliereDouglass et al. (Fig. 3) [41].

Trends

2.6. Analysis of half-bodies, IgG4s and Fab-arm exchange An interesting sub-class of naturally-occurring Abs comprises IgG4 molecules. These Abs differ from the other classes, in that their inter-chain connectivity is less stable. The resulting non-covalent nature of some IgG4 Abs can give rise to a phenomenon called Fab-arm exchange (i.e. the exchange of heavy-light chain pairs among IgG4 molecules) [42]. This Fab-arm exchange phenomenon has been studied in detail in vitro by Rose et al. [43]. In their work, hingedeleted IgG4 constructs, therefore lacking important inter-chain disulfide bridges, were engineered to investigate the effect of non-covalent interactions between the CH3 domains of Abs on the formation of half-molecules (one light-heavy chain pair). By measuring the abundance of half (75 kDa) and full molecules (150 kDa) in the native MS spectra, Rose et al. were able to determine the solution-phase KD values of wild-type hingedeleted IgG4 and CH3-engineered IgG4 single and double amino-acid substituted mutants. The strength of the non-covalent interactions was found to correlate well with the in vivo observed Fab-arm exchange phenomenon [43]. Although, to date, approved mAbs are mostly IgG1 derivatives, a deep understanding of this exchange becomes necessary when IgG4 Abs are used as therapeutics, as they may undergo Fab-arm exchange in vivo with endogenous Abs [44]. 2.7. Analysis of bispecific antibodies and composite mixtures of antibodies With conventional therapeutic mAbs penetrating the market of biopharmaceuticals, the industry is now

Figure 3. Deconvoluted mass spectra of a deglycosylated antibody-drug-conjugate consisting of a mixture of six different species. Native MS was used to characterize the heterogeneity caused by differences in load and position of the conjugated drugs. The solid spectrum represents the ADC, whereas the dashed spectrum represents the original non-modified mAb. (Adapted from [41] with permission of ACS Publications, ª 2012).

http://www.elsevier.com/locate/trac

77

Trends

Trends in Analytical Chemistry, Vol. 48, 2013

Figure 4. Antibody glycoform profiling by native MS exploiting an Orbitrap mass analyzer. Illustrative native mass spectra of IgG with increasingly complex glycosylation profiles. (a) Full native mass spectrum of a deglycosylated IgG. (b) Native mass spectrum focusing in on an individual charge state of a glycosylated intact IgG. (c) Highly glycosylated IgG half-molecule. Individual glycoforms could be assigned based on the exact measured differences in mass between peaks, corresponding to 162 Da [hexose (galactose), H], 203 Da (GlcNAc, G), 146 Da (fucose, F) or 291 Da (sialic acid, S), as indicated by the lists on the right of each spectrum. (Reprinted from [53]).

looking forward to new improved products, particularly with regard to efficacy. Interesting classes of such molecules are the so-called bispecific mAbs, which are composed of fragments of two different mAbs that consequently can bind to two different types of antigens and can be exploited to enhance pharmacodynamic properties. As bispecific IgG1s do not occur naturally, molecular engineering is necessary for their production [45–47]. Even more complex than bispecific mAbs, composite mixtures of mAbs are also exploited as novel biophar-

78

http://www.elsevier.com/locate/trac

maceutical products. Despite not yet being available as therapeutics, mixtures of mAbs seem to be promising alternatives to single-mAb therapeutics. A therapeutic product consisting of a mixture of Abs can target multiple epitopes of the same antigens or different antigens involved in a particular disease, resulting in an improved efficacy [48]. Bispecific mAbs and composite mixtures of Abs enlarge the molecular complexity of mAbs dramatically, putting even greater demands on the analytical tools to characterize them. Recently, Rosati et al. described a

Trends in Analytical Chemistry, Vol. 48, 2013

Trends

new method enabling the simultaneous qualitative and quantitative analysis of bispecifics and mixtures of mAbs by using native MS [49]. Notably, these mixtures of Abs had been produced by co-expressing multiple forms of monospecific Abs in one cellular platform, an approach that also benefits from a significant reduction of costs [48,50]. Co-expression of Abs resulted in a mixture of variable amounts of monospecific and bispecific Abs. It was demonstrated that native MS represents a robust technique for characterization of Ab mixtures, at least comparable to other conventional methods [e.g., cation exchange chromatography (CEX)].

Especially in the field of biosimilars, but also for quality control or stability assays, the improved resolution and mass accuracy could notably increase the confidence of identification of PTMs and mass shifts caused by sequence variants. Moreover, the very high analysis speed (1–2 min per sample) makes this technique very suitable for routine analysis. Moreover, future developments may lead to the implementation of better fragmentation methods for native mAbs (e.g., ETD and photo-dissociation), opening options also to assess location directly (e.g., PTMs and drug-conjugates) [8].

2.8. High-resolution MS of intact antibodies The development of high-resolution MS (HRMS) for the analysis of intact proteins is reintroducing the idea of Ab analysis without the exhaustive sample-preparation steps required for bottom-up analyses. Two papers recently pushed the limits of HRMS with respect to the analysis of intact Abs. In 2011, Valeja et al. demonstrated unit mass baseline resolution for a 150 kDa mAb using a Fourier Transform Ion Cyclotron Resonance (FTICR) mass spectrometer [51]. This achievement was reached by increasing the desolvation of the ions prior to injection into the ICR as well as automated phase correction. This unit mass baseline resolution is a great instrumental achievement, and may facilitate the identification of very small structural variations in mass (e.g., deamidation). Recently, we explored the potential of an Orbitrap mass analyzer (Exactive Plus, ThermoFisher Scientific, Germany) for the analysis of native intact Abs [52,53]. In our preliminary studies, we observed high resolution and sensitivity in the Orbitrap instrument using this new native MS mode. High resolution, together with high mass accuracy, allowed not only confident identification of a single mAb in a mixture, but also qualitative characterization of glycosylation profiles. As already anticipated in sub-section 2.4, characterization of Abglycosylation profiles could benefit from improved resolution. For instance, Ab Fc regions are often engineered to alter glycosylation patterns that may improve pharmacokinetics and decrease immunogenicity [54–56]. This could result in unusual highly complex mixtures of Abs with different glycosylation patterns [57]. A comprehensive characterization of such mixtures could represent a real challenge if the instrument resolution is not sufficient. The glycosylation profile of a highly glycosylated IgG half-molecule was characterized using a modified Orbitrap mass spectrometer, providing spectra with unprecedented experimental mass-resolving power (Fig. 4) [53]. With the improved mass resolution provided by this Orbitrap-based instrument, the analysis of Abs at the intact level could gain wider use, even in the qualitycontrol laboratories of biopharmaceutical companies.

3. Conclusions mAbs represent a fast emerging class of therapeutics in biopharmaceuticals. As their development and expansion in the clinical arena have been very rapid in recent years, there is even further great expectation for the future. Along with the increasing complexity of new Ab entities, analytical techniques need to be optimized to characterize these products thoroughly. In this review, we show how MS already plays an important role in the in-depth characterization of multiple features of therapeutic mAbs. However, new challenges from the development of new therapeutic entities have to be expected for the future. We therefore strongly encourage the biopharmaceutical analytical world, and in particular the field of MS, to optimize further, to improve and to implement new methods for the analysis of biopharmaceuticals. Acknowledgements This work was supported in part by STW (Project 10805). The Netherlands Proteomics Centre, embedded in The Netherlands Genomics Initiative, is acknowledged for funding. References [1] I. Strickland, World Preview 2018: Embracing the Patent Cliff, EvaluatePharma, London, UK, 2012. [2] G. Walsh, Nat. Biotechnol. 28 (2010) 917. [3] J.M. Reichert, mAbs 4 (2012) 413. [4] A.C. Chan, P.J. Carter, Immunology 10 (2010) 301. [5] S.A. Berkowitz, Nat. Rev. Drug Discov. 11 (2012) 527. [6] H.S. Samra, F. He, Mol. Pharmaceut. 9 (2012) 696. [7] Z. Zhang, H. Pan, X. Chen, Mass Spectrom. Rev. 28 (2009) 147. [8] N.J. Thompson et al., Chem. Commun. 49 (2013) 538. [9] M. Karas et al., Int. J. Mass Spectrom. Ion Process. 78 (1987) 53. [10] A. Beck et al., Anal. Chem. 85 (2012) 715. [11] A. Beck, S. Cianferani, A. Van Dorsselaer, Anal. Chem. 84 (2012) 4703. [12] J.B. Fenn et al., Science (Washington, DC) 246 (1989) 64. [13] R.C. Beavis, B.T. Chait, Rapid Commun. Mass Spectrom. 3 (1989) 233. [14] R.C. Beavis, B.T. Chait, Rapid Commun. Mass Spectrom. 3 (1989) 432.

http://www.elsevier.com/locate/trac

79

Trends

Trends in Analytical Chemistry, Vol. 48, 2013

[15] R.C. Beavis, B.T. Chait, Rapid Commun. Mass Spectrom. 3 (1989) 436. [16] J.A. Loo, Int. J. Mass Spectrom. 200 (2000) 175. [17] A.J. Heck, Nat. Methods 5 (2008) 927. [18] C. Uetrecht et al., Proc. Natl. Acad. Sci. USA 105 (2008) 9216. [19] C. Uetrecht, A.J. Heck, Angew. Chem., Int. Ed. Engl. 50 (2011) 8248. [20] G.K. Shoemaker et al., Mol. Cell. Proteomics 9 (2010) 1742. [21] A. Schreiber et al., Nature (London) 470 (2011) 227. [22] A. Zeck et al., PLoS One 7 (2012) e40328. [23] Y. Yang et al., mAbs 2 (2010) 285. [24] Y. Wade, J. Chapman, J. Struct. Anal. Prot. Var. Mol. Biol. 16 (1996) 101. [25] I. Perdivara et al., Anal. Bioanal. Chem. 391 (2008) 325. [26] E.A. Kabat, T.T. Wu, J. Immunol. 147 (1991) 1709. [27] D.M. Creasy, J.S. Cottrell, Proteomics 2 (2002) 1426. [28] M. Weise et al., Nat. Biotechnol. 29 (2011) 690. [29] J.M. Reichert, A. Beck, H. Iyer, mAbs 1 (2009) 394. [30] M. McCamish, G. Woollett, mAbs 3 (2011) 209. [31] US Food and Drug Administration, Guidance for Industry on Biosimilars, 2012 . [32] European Medicines Agency, Biological guidelines . [33] H. Xie et al., mAbs 2 (2010) 379.

80

http://www.elsevier.com/locate/trac

[34] [35] [36] [37] [38] [39] [40] [41] [42] [43] [44] [45] [46] [47] [48] [49] [50] [51] [52] [53] [54] [55] [56] [57]

Y. Wang et al., Anal. Chem. 83 (2011) 3133. C. Uetrecht et al., Chem. Soc. Rev. 39 (2010) 1633. D. Bagal et al., Anal. Chem. 82 (2010) 6751. C. May, P. Sapra, H.P. Gerber, Biochem. Pharmacol. 84 (2012) 1105. E.L. Sievers, P.D. Senter, Annu. Rev. Med. 64 (2013) 15. S.C. Alley, N.M. Okeley, P.D. Senter, Curr. Opin. Chem. Biol. 14 (2010) 529. P.D. Senter, Curr. Opin. Chem. Biol. 13 (2009) 235. J.F. Valliere-Douglass, W.A. McFee, O. Salas-Solano, Anal. Chem. 84 (2012) 2843. A.F. Labrijn et al., J. Immunol. 187 (2011) 3238. R.J. Rose et al., Structure 19 (2011) 1274. A.F. Labrijn et al., Nat. Biotechnol. 27 (2009) 767. D. Chelius et al., mAbs 2 (2010) 309. K. Gunasekaran et al., J. Biol. Chem. 285 (2010) 19637. Z. Xie et al., J. Immunol. Methods 296 (2005) 95. T. Logtenberg, Trends Biotechnol. 25 (2007) 390. S. Rosati et al., Anal. Chem. 84 (2012) 7227. J. de Kruif et al., Biotechnol. Bioeng. 106 (2010) 741. S.G. Valeja et al., Anal. Chem. 83 (2011) 8391. R.J. Rose et al., Nat. Methods 9 (2012) 1084. S. Rosati et al., Angew. Chem., Int. Ed. Engl. 51 (2012) 12992. R. Jefferis, Nat. Rev. Drug Discov. 8 (2009) 226. D. Ghaderi et al., Nat. Biotechnol. 28 (2010) 863. A. Natsume et al., Cancer Res. 68 (2008) 3863. R.J. Rose et al., mAbs 5 (2013) 219.