Biotechnology and
Applied Biochemistry Industry and regulatory experience of
the glycosylation of monoclonal
antibodies
Erik K.Read,Jun T.Park,and Kurt A.Brorson∗
Division of Monoclonal Antibodies,Office of Biotechnology Products,Center for Drug Evaluation and Rearch,
Food and Drug Administration,Silver Spring,MD,USA
Abstract.b to c
We surveyed23antibody-related marketing applications for glycoform analytical and functional information.Our databa analysis shows a clear trend of increasing sophistication of analytical methods ud to identify and quantify glycans. The have revealed a high degree of complexity and h
eterogeneity of glycans attached to antibody products.The nature of the complexity is influenced by product type and expression system,and may be associated with functional conquences in some but not all cas.
spread是什么意思Published2011.This article is a U.S.Government work and is in the public
domain of the USA.Volume58,Number4,July/August2011,Pages213–219•
E-mail:kurt.brorson@v Keywords:glycosylation,monoclonal antibodies
1.Introduction
The number of licend therapeutic monoclonal antibodies (mAbs)has been increasing over the past few years,with hun-dreds more already undergoing clinical study for indications for a variety of therapeutic applications,including cancer and inflammatory dias[1].Most of the products are pro-duced in conventional bioreactor-bad mammalian cell ,Chine hamster ovary(CHO)or murine myeloma transfectomas],although a few are produced by other expres-sion ,Escherichia coli)[2].Therapeutic antibodies must be demonstrated to meet applicable quality requirements to ensure continued safety,purity,and potency to convince regulators to allow m
arketing as a drug product.Part of the demonstration of product quality is an intensive biochemical characterization of the antibody itlf,which includes a thor-ough examination of glycan distribution and potential impacts of glycoform on function[3].This characterization is conducted in two major stages,(a)a complete glycan distribution char-acterization of reference standard or conformance lots of the antibody glycoprotein and(b)abbreviated testing of all sub-quent batches to establish manufacturing consistency and Abbreviations:α-gal,α-galactosyl residues;ADCC,antibody-dependent cellular cytotoxicity;BLAs,Biological Licen Applications;CE,capillary electrophoresis;CHO, Chine hamster ovary;CDC,complement-dependent cytotoxicity;exo,exoglycosida; MS,mass spectrometry;MoA,mechanism of action;mAbs,monoclonal antibodies;OP, oligosaccharide profiling;G0F,G1F and G2F,outer arm non,mono or bi-β-galactosylated variant of core fucosylated biantennary N-linked glycans;FDA,US Food and Drug Administration.
∗Address for correspondence:Kurt A.Brorson,PhD,Division of Monoclonal Antibodies, Office of Biotechnology Products,Center for Drug Evaluation and Rearch,Food and Drug Administration,Silver Spring,MD20903,USA.Tel.:1-301-796-2193;
Fax:1-301-827-0852;e-mail:kurt.brorson@v.
Received18April2011;accepted6May2011
DOI:10.1002/bab.35sharp是什么意思
Published online16August2011in Wiley Online Library
()comparability with the reference material.The tests ud in the analys span a wide range of analytical methodologies, which have grown more sophisticated over the years[4].
For the most part,glycans on commercial antibodies are attached at asparagine residues at or near position297 (N297)within the Fc portion of the protein[5].Mammalian cell culture-produced antibodies typically posss N-linked complex biantennary structures,with heterogeneous levels of terminal galactosylation and fucosylation of the core N-acetylglucosamine[6].To a lesr degree,terminal sialylation and bicting N-acetylglucosamine are also prent.Although the glycans do not directly impact the antigen-binding func-tion of the antibody protein,they can impact effector func-tions such as antibody-dependent cellular cytotoxicity(ADCC) or complement binding and activation(also known as CDC or complement-dependent cytotoxicity)[7].Examples of doc-umented impacts of glycosylation on antibody functionality in-clude,but are not limited to,(a)an inver correlation between ADCC activity on core fucosylation[8],(b)an increa in CDC ac-tivity with increa
d galactosylation[9],and(c)a positive corre-lation between anti-inflammatory activity and incread sialyla-tion[10].A subt of antibody-like products,Fc fusion proteins, posss more complex glycan distributions,including O-linked glycans.Thus,glycoform variation can impact the potency or in vivo distribution/clearance of therapeutic antibodies and needs to be characterized and controlled.As part of glycan char-acterization,the impact of glycan distribution on the product mechanism of action(,cancer cell destruction,down-modulation of inflammatory activity)is commonly evaluated byfirms wishing to market antibody-bad medicinal products.
视频英文Over the past25years,almost40antibody products have been approved for marketing by US Food and Drug Administration(FDA).The licensure decision is bad on infor-mation submitted in the marketing dossier including the above
213
described antibody reference characterization,batch analysis, analytical methodologies,and evaluation of MoA/glycoform de-pendencies.In this report,we survey information gathered from 23antibody and Fc fusion protein-marketing applications in or-der to provide an overview of industry experience with antibody glycan analytics,biochemistry,and function.
2.Materials and methods
Information related to glycan analysis,distribution,or function was extracted from the quality,and in some cas nonclinical (pharmacology and pharmacokinetics),ctions of23Biolog-ical Licen Applications(BLAs)that have been submitted to the FDA Division of Monoclonal Antibodies between1994and 2009.The databa was targeted to extract information from products containing antibody-derived protein quences,and the23were chon bad upon electronic availability of in-formation.Thus,the databa covered15traditional mAb,two mAb conjugates,three Fab fragments,and three Fc fusion pro-teins.To the extent available,the following data were gathered: (a)product ,submission date,cell source,mAb subtype,and species),(b)glycan ,methods,char-acterization testing,and lot relea specifications);(c)glycan profi,glycan identities,occupancy,and monosaccharide content);and(d)impact of glycan profi,Fc effector func-tions and treatment mode of action).
In many cas,veral batches of an antibody product were analyzed by more than one assay.Information for a par-ticular batch by each assay was captured to form a“record”in the FileMaker Pro9databa.Therefore,the databa con-tains a total of709test records of data extracted from char-acterization studies,comparability studies,validation studies, reference lots,and p
伞的英文单词roduction lots.
A wide variety of analytical methods were identified in our survey that could be sorted into three major categories: oligosaccharide profiling(OP),mass spectrometry(MS),and capillary electrophoresis(CE).Modalities were also combined, such as OP followed by MS or CE followed by MS.In many cas,assays were modified by treatment of the product with peptic digestion(peptic)or exoglycosida(exo),followed by fractionation and detection.Alternatively,the glycoform distri-bution of an entire antibody was assd(intact).The meth-ods posss varying levels of nsitivity and ranges of species detection.Some methods were highly quantitative for even low-level species,whereas others could detect,but not necessarily quantify,species prent at2%–5%or below.OP methods em-ployed rever pha,anion exchange,and hydrophilic inter-action liquid chromatography depending on whether charge or hydrophobicity was the basis of paration.CE parations were bad on molecular weight[CE-sodium dodecyl sulfate(SDS)] or charge[CE-isoelectric focusing(IEF)].Modalities of MS em-ployed a variety of matrix-assisted lar desorption/ionization and electrospray ionization-bad methods.Detection modes after the various paration methods included ultraviolet,fluo-rescence,amperometry,refractive index,MS,and gel staining.
In this manuscript,G0F,G1F,and G2F refer to outer armβ-galactosylated variants of core fucosylated biantennary N-linked glycans.The glycans can be terminally sialylated, thereby contributing to“%sialylation”in molecules with N-linked glycosylation(nonsialylated glycans are“neutral”).
3.Results and discussion
Certain effector functions of mAbs are influenced by glycan dis-tributions,and this may translate into altered potency[11].To generate a snapshot of industry experience of antibody glycosy-lation from various assays employed for analysis,distributions within mAb products,and possible impacts of glycoform hetero-geneity on MoA,we created a databa of information gathered from mAb licen applications(Fig.1).The datat was asm-bled from licen applications in electronic format.Defining the list this way included Fc fusion proteins,and antibody fragments regardless of expression ,mammalian li), but did somewhat bias the lection in favor of more recent products with more extensive glycan information.As can be en in Fig.1,the number of licend products has been grow-ing over the past two decades,dominated by tho produced in rodent cell line-bad expression ,CHO,NS0,and SP2/0).
4.Glycan assays
Our survey identified three major class of glycan assays:OP, CE,and mass spectrometry.During OP,glycans are enzymati-cally relead from the protein,and then subjected to a variety of modifications that enhance detection and alter their chemi-cal properties.Later,the modified glycans are parated via liquid chromatography.In the studies within this databa,CE-SDS and CE-IEF have been ud to parate variants of glycopep-tides or enzymatically relead glycans.With both OP and CE, the glycan species within the elution peaks may be identified and quantified by relative mobility and the intensity,respec-tively.The glycans relead from or within glycopeptides were also analyzed by MS where the fragmentation pattern is ud to determine the molecular mass of the analyte and deduce the attached glycan species.Rarely,the ion peak intensities were reported as a miquantitative measure of the glycan species distribution.All three major class of glycan identification can be combined with predigestion with peptida to parate dis-tinct N-glycan attachment sites and/or enhance assay resolu-tion.Alternatively,exo treatments may also be added to reduce sample complexity by removing specific glycan modifications or to quantitatively“quence”the glycan pool of variants.Fre-quently,OP and CE methods were combined with MS to more thoroughly identify the constituents of the eluted peaks.
There are also methods that quantify the total mass of spe-cific sugar modifications within the glycan
pool.In monosaccha-ride analysis,the glycan pools are acid hydrolyzed and then p-arated by chromatography to quantify the absolute or relative mass of fuco,manno,galactos,N-acetylglucosamine, and occasionally N-acetylgalactosamine.Sialic acid levels were quantified by weak acid or enzymatic relea,followed by de-rivitization and chromatographic paration.
214Biotechnology and Applied Biochemistry
Fig.1.Databa composition(a)by product type,(b)year of product relea,(c)production cell type,and(d)production cell type by year of relea.Fusions are globular protein domains connected to an IgG Fc domain,whereas fragments are the antigen binding Fab domain.Protein production cell types include Chine hamster ovary(CHO),murine myeloma(Sp2/0, NS0),or Escherichia coli li).As expected,Fabs li produced products possd no glycosylation.
Becau of the large number of formats and redundant data produced,not all assay types are ud for each product. Instead,subts are employed,as judged sufficient for charac-terization of individual products.美国大法官金斯伯格去世
4.1.Lot relea versus characterization
Antibody biochemical testing occurs primarily at two major lev-els,lot relea and characterization.Lot relea tests are per-formed on a100%basis on each batch of drug substance(or product),with the goal of monitoring production consistency and screening for outlier batches.Characterization tests are performed on one or a few of the consistency batches immedi-ately prior to licensure or after major process changes.Antibody characterization generally involves a comprehensive analytical testing program to gain a detailed understanding of the bio-chemistry of the molecule.
From the survey,it is clear that for most products,glycan characterization is quite extensive as nearly half are analyzed by three or more assays(Fig.2).Only ven had no glycan-specific characterization,and of the,three were Fabs that lacked the N297site of N-linked glycosylation site.The remaining four were older products(from the1990s).In contrast,most antibody products are not tested for glycan distribution at lot relea.
A comprehensive understanding of the biomolecule’s glycans and material attributes and process parameters that can impact glycans may obviate the need for extensive ongoing monitoring. It should be emphasized that as per regulatory guidance,
lot Fig.2.Distribution of glycan analys.A summary of the number of distinct analytical tests performed(a)for product characterization and(b)as a lot relea specification for all of the BLAs within the databa.The ven products that had no glycan-specific characterization were Fabs or older products.
Experience with glycosylation of mAbs215
Fig.3.General trends in adopted glycan assays over time.(a)Increasing adoption and complexity of assay type.Each bar reprents one(thin)or more assays adopted in a particular year.Predominant assay ud to detect O-glycans.(b)Increasing glycan assay information density over time;numbers are averages of glycans identified or quantified.(c)Increasing number of distinct glycan assays over time.Numbers are averages of the number of glycan assays per product approved in a particular time bracket.Although most of the O-glycan testing is performed at the characterization stage,Fc fusion proteins also can have relea tests and specifications in some cas.
relea testing and specifications are only part of an overall control strategy[3]and that tight process control during cell culture can be sufficient to limit glycan heterogeneity within a range determined by an initial extensive characterization.
适用于英文4.2.Historical analysis
We next enumerated the number of glycan-specific tests in each BLA,binning them by4-year periods and by purpo(lot re-lea vs.characterization).There is a trend toward incread reliance on early characterization in terms of the number of tests per BLA(Fig.3c).For example,when thefirst few anti-bodies were being licend in the mid-1990s,on average fewer than two characterization tests were performed on each product. More recently,almost four tests per product are performed on average.
如何学习美容美发Our analysis also revealed that test complexity is increas-ing(Fig.3a).For example,combinations of backbone methods with ancillary treatment and/or detection are increasingly ud (e.g.,addition of exo treatment combined with OP and MS, CE and MS).By combining methodologies,additional informa-tion is gathered about glycan structure and heterogeneity;for example,the overall number of identified glycans(Fig.3b) is being quantified,especially after2006.It should be noted that although up to10–11glycans are identified in recent BLAs, many are very low-level variants where quantification was not achieved.
Overall,increasing reliance on enhanced analytical char-acterization is a trend that is growing.Thus,a full understanding of glycan diversity early in development appears to be a key fac-tor in successfully developing safe and effective products[3].
5.Glycan levels and types
Generally,mAb products have been produced in CHO cells(from Chine Hamsters)or mou myeloma transfectomas such as SP2/0or NS0(Figs.1c and1d).Antibody fragment products,in-cluded in this analysis,are largely produced by bacteria,where glycosylation pathways do not exist.In mammalian products, glycans are attached at or near N297within the Fc portion of the antibody and are typically N-linked complex biantennary
元宵节 英文216Biotechnology and Applied Biochemistry
Fig.4.Utilization of specific analys of glycan distributions.(a)Number of BLAs(out of23)testing for each glycan property.
(b)Number of specific tests per BLA for each glycan property.(c)Cell substrate type and glycan properties.Data are averaged from50to100batches of two to four murine and10CHO products.(d)Number of BLAs testing Fc effector functions and the conclusion drawn by the analysis.For at leastfive products,each of the four parameters was tested.
structures,with heterogeneous levels of terminal galactosyla-tion,bicting N-acetylglucosamine,terminal sialylation,and fucosylation of the core N-acetylglucosamine.To a lesr de-gree,high-manno-containing structures are usually prent. As stated above,the level of testing capable of elucidating the existence of the specific glycan modifications is extensive (Figs.4a and4b).Certain modifications em to be of particular ,outer arm galactosylation,sialylation,core fucosy-lation),but other modifications are also evaluated on a ca-dependent ,high-manno structures).At lot relea, assays captured detailed information about glycans but accep-tance criteria were in many instances not specific for individual species or amounts.
To evaluate the glycosylation of typical commercial anti-bodies,we averaged values of%neutral,%sialylated,%,glycans as a percentage of total mass of molecule), and%,percentage of heavy chains with an at-tached glycan)levels of batches in our da
taba.This infor-mation was further analyzed by cell substrate(u myeloma;Fig.4c)and product type(intact mAb vs.Fc fusion protein;Table1).
5.1.CHO versus murine myeloma substrates
Our databa confirms that commercial mAbs largely posss typical mammalian glycan structures.The N-linked sites at
Table1
Glycans of antibodies and Fc fusion proteins
N-linked glycosylation O-linked glycosylation Product Number of sites Occupancy(%)Number of sites Occupancy(%) Antibody N=10a,b2>970N.A.
Fc fusion N=3a3–14Variable0–10Variable
a Total number of BLAs(products)in this category.
b Only CHO-derived antibody products are included in this comparison becau all F
c fusion products are made in CHO cells.
SP2/0and NS0antibody products have the same N-linked sites with similar levels of occupancy.
三级考试时间Experience with glycosylation of mAbs217