REVIEW
`published: 30 June 2022
`doi: 10.3389/fimmu.2022.929895
`
`Role of Fc Core Fucosylation in the
`Effector Function of IgG1 Antibodies
`Jose´ e Golay 1*, Alain E. Andrea 2† and Irene Cattaneo 1†
`
`1 Center of Cellular Therapy "G. Lanzani", Division of Hematology, Azienda Socio Sanitaria Territoriale Papa Giovanni XXIII, Bergamo,
`Italy, 2 Laboratoire de Biochimie et The´ rapies Mole´ culaires, Faculte´ de Pharmacie, Universite´ Saint Joseph de Beyrouth, Beirut,
`Lebanon
`
`The presence of fucose on IgG1 Asn-297 N-linked glycan is the modification of the human
`IgG1 Fc structure with the most significant impact on FcɣRIII affinity. It also significantly
`enhances the efficacy of antibody dependent cellular cytotoxicity (ADCC) by natural killer
`(NK) cells in vitro, induced by IgG1 therapeutic monoclonal antibodies (mAbs). The effect
`of afucosylation on ADCC or antibody dependent phagocytosis (ADCP) mediated by
`macrophages or polymorphonuclear neutrophils (PMN)
`is less clear. Evidence for
`enhanced efficacy of afucosylated therapeutic mAbs in vivo has also been reported.
`This has led to the development of several therapeutic antibodies with low Fc core fucose
`to treat cancer and inflammatory diseases, seven of which have already been approved
`for clinical use. More recently, the regulation of IgG Fc core fucosylation has been shown
`to take place naturally during the B-cell immune response: A decrease in a-1,6 fucose has
`been observed in polyclonal, antigen-specific IgG1 antibodies which are generated during
`alloimmunization of pregnant women by fetal erythrocyte or platelet antigens and following
`infection by some enveloped viruses and parasites. Low IgG1 Fc core fucose on antigen-
`specific polyclonal IgG1 has been linked to disease severity in several cases, such as
`SARS-CoV 2 and Dengue virus infection and during alloimmunization, highlighting the in
`vivo significance of this phenomenon. This review aims to summarize the current
`knowledge about human IgG1 Fc core fucosylation and its regulation and function in
`vivo, in the context of both therapeutic antibodies and the natural immune response. The
`parallels in these two areas are informative about the mechanisms and in vivo effects of Fc
`core fucosylation, and may allow to further exploit the desired properties of this
`modification in different clinical contexts.
`
`Keywords: therapeutic antibodies, IgG, N-glycan, fucosylation, ADCC, NK cells, virus, humoral response
`
`INTRODUCTION
`
`IgGs are among the most abundant proteins in the circulation (700-1600 mg/dl in healthy adults), and
`specific IgGs are induced in response to infection, endogenous or allogeneic challenges, or by
`vaccination. Different IgG subclasses are found in man, which are very similar structurally but have
`distinct functions due to their differential binding to FcɣRs, complement components as well as other
`
`Edited by:
`Peter Boross,
`Genmab, Netherlands
`
`Reviewed by:
`Kutty Selva Nandakumar,
`Karolinska Institutet (KI), Sweden
`Gestur Vidarsson,
`Sanquin Research, Netherlands
`Mattias Collin,
`Lund University, Sweden
`
`*Correspondence:
`Jose´ e Golay
`jgolay59@gmail.com
`
`†These authors have contributed
`equally to this work
`
`Specialty section:
`This article was submitted to
`B Cell Biology,
`a section of the journal
`Frontiers in Immunology
`Received: 27 April 2022
`Accepted: 03 June 2022
`Published: 30 June 2022
`Citation:
`Golay J, Andrea AE and Cattaneo I
`(2022) Role of Fc Core Fucosylation
`in the Effector Function of
`IgG1 Antibodies.
`Front. Immunol. 13:929895.
`doi: 10.3389/fimmu.2022.929895
`
`Frontiers in Immunology | www.frontiersin.org
`
`1
`
`June 2022 | Volume 13 | Article 929895
`
`1 of 18
`
`OnCusp
`Ex. 1036
`
`

`

`Golay et al.
`
`IgG Fucosylation
`
`proteins. About 60% of plasma IgG is IgG1, 32% IgG2 and 4% each
`IgG3 and IgG4 in humans (1). IgGs are glycoproteins and their
`glycosylation pattern can change during time, due to age, diseases or
`environmental factors (2, 3).
`Therapeutic monoclonal antibodies (mAbs) have emerged as
`an important therapeutic option in cancer since the approval in
`1997 of the anti-CD20 antibody rituximab for the treatment of
`B-non Hodgkin's lymphoma (B-NHL). Since then, antibodies
`directed against different antigens expressed by cancer, immune
`cells or infectious agents have been developed to treat a variety of
`diseases. Indeed, so far, over 130 antibodies have been approved
`by the US and EU Drug Agencies, with 45% for oncological
`disorders, 27% for immune- or inflammation-related conditions
`and the rest for infectious or other diseases (4).
`Most unconjugated therapeutic mAbs are IgG1 or in some
`cases IgG4 or IgG2. This is because the human IgG1 Fc moiety
`interacts efficiently with activating FcɣRs (FcɣRI, IIA, IIC, IIIA
`and IIIB), expressed on the surface of immune cells (1, 5). This
`interaction leads to antibody-dependent cellular cytotoxicity
`(ADCC) by NK cells (mostly via FcɣRIIIA, CD16A) (6, 7),
`antibody dependent phagocytosis (ADCP) by macrophages
`(mostly through FcɣRI, CD64 and to some extent FcɣRIIA,
`CD32A) (8–12) and ADCC/ADCP by polymorphonuclear
`neutrophils (PMN)(mostly via FcɣRIIA, CD32A) (13–15).
`IgG1 also interacts with FcɣRIIIB (CD16B), a GPI-linked
`molecules lacking activating domain, highly expressed by PMN
`and involved in PMN mediated ADCC and ADCP, but whose
`role may be either activating or inhibiting, perhaps depending on
`stimulus (13–16). Immune cell activation via FcɣRs also induces
`the release of cytokines and chemokines that may cooperate in
`eliminating the target cells but also induce unwanted side-effects
`(17). Finally the Fc region of human IgG1 can bind to the first
`component of the complement cascade C1q and activate the
`classical pathway of complement which may lead to cell lysis and
`death through complement dependent cytotoxicity (CDC), as
`well as phagocytosis by macrophages and PMN through
`complement receptors on these cells (18). Therefore, many
`therapeutic antibodies against cancer cells or other targets are
`of the IgG1 isotype to allow activation of a panoply of immune-
`mediated mechanisms, many of which rely on FcɣRs.
`When the activation of the immune system is not desired, for
`example when a therapeutic antibody is required only to
`neutralize the antigen, such as a growth factor or checkpoint
`inhibitor, then the human IgG4 or IgG2 subclasses are often
`chosen, because they do not interact efficiently with FcɣRs or
`with C1q. The more recent human IgG4 formats include a
`mutation in Fc (S228A) to avoid Fab arm exchange, a natural
`phenomenon that leads to IgG4 instability (19).
`Over the last 10-15 years, various modifications of antibody
`structures have been introduced to increase the efficacy of
`therapeutic mAbs in vitro and in vivo: these include extensively
`modified Abs with additional effector functions, such as
`bispecific antibodies (bsAbs), antibody-drug conjugates
`(ADCs) and fusion proteins carrying for example cytokines
`(17, 20). Less dramatic modifications of therapeutic mAbs
`include the introduction of point mutations in the Fc domain,
`
`as well as modification of Fc N-linked glycans that modulate IgG
`binding to FcɣRs and therefore enhance or abolish Fc mediated
`immune activation (ADCC, ADCP and/or CDC) (17, 20).
`In this paper, we will summarize the knowledge gained about
`the role of IgG1 N-glycan core fucosylation in the in vitro and in
`vivo functions of IgG1 antibodies. Ig isotypes or subclasses other
`than IgG1 bear N-glycans, but less is known about the role of
`core Fc fucosylation in their case and these will not be further
`discussed here. Interestingly, the studies on the role of Fc core
`fucose in therapeutic IgG1 mAbs has facilitated the detection and
`understanding of the significance of this modification, observed
`during the polyclonal IgG1 response to some infectious agents,
`alloimmunization and in some autoimmune conditions. The
`knowledge on these aspects will therefore also be summarized
`and discussed.
`
`THE IGG N-LINKED GLYCANS
`
`Human IgGs are glycosylated proteins with a complex and
`variable glycosylation pattern. An important and extensively
`studied N-glycosylation site is present at conserved Asparagine
`297 (Asn 297) in the CH2 domain, that interacts with FcɣRs. 20-
`30% of IgGs also bear N-glycans on Fab arms (21). Although
`there are reports of functional effects of different Fab N-
`glycosylation patterns in some antibodies (22), these are likely
`to be mostly antibody specific (23). Detailed structural studies of
`several commercial therapeutic mAbs has revealed that although
`Fab interacts with FcɣRIIIA and stabilizes the Fc-FcɣRIIIA
`binding, Fab fucosylation has a limited effect on the affinity of
`IgG for FcɣRIIIA (23, 24). The modulation of Fab fucosylation
`will therefore not be further discussed here.
`The IgG Asn 297 N-glycans show a high degree of
`microheterogeneity, and they can be grouped in oligomannose,
`hybrid or complex type, the latter being the most abundant
`(about 90%) in IgG, either circulating or produced by cell lines in
`vitro (25) (Figure 1A). The presence of Fc N-glycan induces in
`general a more open structure compared to aglycosylated IgG,
`favors binding to activating FcɣRs and promotes antibody
`stability in vitro and in vivo (1, 26). The complex type N-
`glycosylation itself shows microheterogeneity: whereas it
`always contains a heptaglycan biantennary core structure (four
`GlcNAc and three mannose residues), the core can bear an
`additional bisecting GlcNAc (in about 10% of IgGs), and 1 or 2
`galactose residues (in 35% and 15% of IgG, respectively) and 1-2
`terminal N-acetylneuraminic acid (sialic acid, SA), on 10-15% of
`IgGs (Figure 1B). Finally, an a-1,6 fucose residue (core fucose) is
`present in 90% of complex type IgG N-glycans (Figure 1B).
`Interestingly the presence of bisecting GlcNAc inhibits a-1,6
`core fucosylation due to steric hindrance and therefore IgGs
`generally contain either a bisecting GlcNAc or a core fucose
`residue, although some IgGs may have both bisecting GlcNAc
`and fucose (2, 27–29). IgGs are composed of
`least 30
`glycovariants, to which specific abbreviations have been
`assigned: G0 (no Gal residue), G1 (1 Gal), G2 (2 Gals), F
`(fucose) etc (Figure 1B) (2, 27–29).
`
`Frontiers in Immunology | www.frontiersin.org
`
`2
`
`June 2022 | Volume 13 | Article 929895
`
`2 of 18
`
`OnCusp
`Ex. 1036
`
`

`

`Golay et al.
`
`IgG Fucosylation
`
`A
`
`B
`
`FIGURE 1 | The IgG Asn-297 N-glycan heterogeneity. Panel (A) major types of N-glycosylation observed in IgG. Panel (B) detail of complex type glycosylation with
`percentage of circulating IgG containing either fucose or bisecting N-Acetylglucosamine. The orange circle indicates the core structure. The blue broken lines and
`text indicate the heterogeneity of complex N-glycans, carrying either no Gal (G0), 1-2 Gal (G1/2), 1-2 Sialic acids (S1/2), with (F) or without core a-1,6-fucose.
`
`THE BIOSYNTHESIS OF HUMAN IGG N-
`LINKED GLYCANS
`
`N-glycosylation is a multi-step enzyme-mediated biochemical
`process. IgG N-glycan biosynthesis starts in the endoplasmic
`reticulum with the addition of a pyrophosphate-dolichol
`precursor (Dol-P, Glc3Man9GlcNAc2) to the Asn 297 N-
`glycosylation site of IgG. This structure is then trimmed by
`glucosidases and mannosidases, as the process moves to the
`Golgi, leading to the formation of high mannose, hybrid and the
`complex types N-glycans (26, 30–33). The main enzymatic
`reactions taking place in the Golgi are summarized in
`Figure 2. The stable overexpression or reduction/inhibition of
`some of these enzymes in antibody producing cell lines have been
`used to modify IgG glycan composition and perform structure-
`function studies of N-glycan microheterogeneity ( (31); and see
`below). The structure of N-linked and other IgG glycans and
`glycosylation pathways have already been extensively reviewed
`by others (30, 32, 34, 35) and we refer the readers to these more
`complete descriptions of the process.
`The enzymatic pathways for GDP-fucose biosynthesis and Fc
`core fucosylation are shown in Figure 3 (26, 30, 36). L-Fucose (6-
`deoxy-L-galactose) is a monosaccharide obtained by
`glycoprotein degradation or diet. In the salvage pathway, L-
`fucose is phosphorylated in the cytosol to fucose-1-phosphate by
`fucokinase (FUK), and then converted to GDP-fucose by GDP-
`fucose-pyrophosphorylase (GDPP), an essential substrate for the
`fucosylation of proteins (Figure 3). Alternatively, and most
`commonly, de novo GDP-fucose is synthesized from GDP-
`mannose by GDP-mannose 4,6 dehydratase (GMD) and then
`GDP-4-keto 6-deoxymannose 3,5-epimerase-4-reductase (FX)
`(Figure 3). GDP-fucose is transported to the ER via the
`
`SLC35C1 and SLC35C2 transporters and used by several
`fucosyltransferases (FUT) in the Golgi
`to fucosylate
`glycoproteins. There are 11 different FUTs, but only FUT 8
`catalyzes IgG Fc core fucosylation via an a-1,6 linkage
`(Figure 3). As mentioned above, the presence of bisected
`GlcNAc inhibits the addition of core a-1,6 fucose (26, 30, 36).
`These pathways show that there are multiple steps where a
`modulation of fucosylation can take place including the supply of
`diet fucose (36). The mechanisms of IgG Fc glycosylation
`regulation in B cells in healthy individuals and in disease are
`however still little understood.
`
`STRATEGIES TO GENERATE
`THERAPEUTIC IGG1 WITH LOW LEVELS
`OF FUCOSE LEVELS
`
`In the last 10-15 years, removal of IgGs Fc core fucose has been
`shown to be an important method to enhance ADCC by therapeutic
`IgG1 mAbs (37–39). Based on the knowledge of the pathways of N-
`glycan fucosylation described above, several strategies have been
`used to generate therapeutic IgGs with low or no fucose [reviewed
`by Pereira et al. (40)]. These are briefly summarized below:
`i. The selection of cell lines that naturally have low FUT8
`enzyme, such as the Y2B/0 rat cell line (38).
`ii. The creation of FUT8 knock out cell lines such as CHO
`FUT8-/- (41, 42). These produce antibodies completely lacking
`fucose, but may have a low growth rate unless specifically adapted.
`iii. The selection of natural cell lines with defects in other
`enzymes involved in fucosylation, such as Lec 13, which has
`defective GDP mannose 4,6 dehydratase (GMD) which converts
`
`Frontiers in Immunology | www.frontiersin.org
`
`3
`
`June 2022 | Volume 13 | Article 929895
`
`3 of 18
`
`OnCusp
`Ex. 1036
`
`

`

`Golay et al.
`
`IgG Fucosylation
`
`FIGURE 2 | Simplified scheme of enzymatic reactions involved in N-linked glycosylation of IgG. The major glycosylation steps and enzymes involved in N- linked
`glycosylation of IgGs taking place in ER and Golgi are shown. GnT, N-Acetyl glucosamine transferase; FUT, Fucosyl transferase; Man, Mannose; Gal, Galactose;
`GlcNAcm N-acetylglucosamine; SA, sialic acid (N-Acetylneuraminic Acid).
`
`FIGURE 3 | Major pathways of GDP-fucose biosynthesis and IgG Fc core fucosylation. FUK, Fucose Kinase; GDPP, GDP-fucose-pyrophosphorylase; GMD, GDP-
`mannose 4,6 dehydratase; FX, GDP-4-keto 6-deoxymannose 3,5-epimerase-4-reductase; FUT, Fucosyltransferase.
`
`Frontiers in Immunology | www.frontiersin.org
`
`4
`
`June 2022 | Volume 13 | Article 929895
`
`4 of 18
`
`OnCusp
`Ex. 1036
`
`

`

`Golay et al.
`
`IgG Fucosylation
`
`GDP mannose to GDP-fucose, the substrate for N-glycan core
`fucosylation (Figure 3) (37, 43).
`iv. The creation of engineered cell lines lacking GMD, GDP-
`L-fucose synthase (FX) and/or the GDP-fucose transporter
`SCL35C1 (Figure 3) (44, 45). For example, the FX-/- and
`®
`GMD-/- CHOZN
`cell clones produced IgG1 antibody with
`core fucosylation reduced to 6-8% or 1-3%, respectively. The
`FX-/- clones also showed some aberrant glycan forms
`®
`suggesting the GMD-/- CHOZN
`cell line may be the best
`choice (44).
`v. Another strategy is to engineer CHO clones with higher b-1,4-
`N-acetylglucoseaminyl-transferase III (GnTIII), best with a Golgi
`localization domain (2, 31, 46, 47). GnTIII catalyzes the transfer of
`GlcNAc to a core mannose residue in N-linked oligosaccharides via
`a b-1,4 linkage, which results in the formation of a bisected sugar
`chain. The bisection GlcNAc inhibits the transfer of fucose by
`FUT8, so that GnTIII overexpressing cells produce antibodies with
`lower levels of core fucose.
`vi. Another modality is to use non-mammalian cells such as
`plant or insect cells that are engineered to synthesize human-type
`N-glycans, to reduce potential immunogenicity, but lack core
`fucosylation capacity (48).
`vii. Cells can be treated with inhibitors of FUT8 such as 2F-
`Peracetyl-Fucose (49) or anti-FUT8 antibody (50).
`viii. Finally, antibodies can also be enzymatically modified in
`vitro by treatment with glycosidase and glycosynthase enzymes.
`This is somewhat more complex since it requires antibody
`deglycosylation followed by a controlled glycosylation (51, 52).
`The major strategies based on engineered mammalian cell
`lines are listed in Table 1.
`It is worth noting that, depending on the system used, the
`amount of fucose may vary from 0% for cell lines lacking FUT8
`enzyme to 10-30% for the those with reduced FUT8 or other
`enzymatic modifications (Table 1). Also, different production
`cell lines and systems may lead to antibodies with different N-
`glycan profiles, not only regarding fucose residues. These may in
`turn affect function of antibodies since carbohydrates other than
`fucose, for example galactose or sialic acid may affect CDC or
`inflammation, respectively, among others (33, 53). A more
`detailed description of the role of galactosylation and sialic
`acid is beyond the scope of this review.
`
`FUNCTIONAL CONSEQUENCES OF LOW
`CORE FUCOSE ON IGG1 ASN 297
`N-GLYCAN
`Binding to FcgRIIIA and FcgRIIIB
`Human, humanized or chimeric IgG1 antibodies lacking fucose
`on the Asn N-glycan bind with 10-100 fold higher affinity to
`human FcɣRIIIA and FcɣRIIIB (CD16B) (37, 54, 55). Structural
`studies have shown an increased carbohydrate-carbohydrate
`interaction between the N-glycans of FcgRIIIA and Fc,
`explaining the higher affinity of afucosylated IgG1 (56, 57).
`
`NK Cells and ADCC
`Since FcɣRIIIA is the major activating receptor on NK cells and
`mediates ADCC, the net result of increased binding to FcgRIIIA
`is a significant enhancement of ADCC by afucosylated IgG1
`antibodies with respect to their fully fucosylated counterpart (2-
`40 fold, also depending on galactosylation) (37–39) (Table 2). In
`addition, FcɣRIIIA has relatively low-medium affinity for IgG so
`that ADCC is inhibited by excess IgG in plasma. In contrast,
`ADCC induced by afucosylated IgG1, which has a significantly
`higher affinity for FcɣRIIIA, is not significantly inhibited by
`plasma IgG. Thus afucosylated anti-CD20 antibody may be more
`effective in inducing ADCC in whole blood by 2 mechanisms: 1)
`higher affinity for FcɣRIIIA and 2) significantly reduced
`inhibition by serum IgG (99). Afucosylated IgG1 has also been
`reported to induce greater FcɣRIIIA downmodulation from the
`NK cell surface, a phenomenon which takes place via shedding of
`the extracellular domain by the ADAM 17 metalloproteinase and
`may participate in the serial target cell killing by NK cells (100).
`Phagocytosis by Macrophages
`FcɣRIIIA is expressed by monocytes/macrophages, particularly
`M2 and red pulp macrophages as well as microglial cells.
`However, macrophages also express the activating FcɣRI and
`FcɣRIIA, and FcgRI is thought to be the major mediator of
`phagocytosis of IgG1 opsonized targets (10, 11, 101). Therefore,
`afucosylated therapeutic mAbs, despite binding with higher
`affinity to FcɣRIIIA, are not generally reported to significantly
`enhance phagocytosis, at least in vitro (10, 11, 102). There is
`some evidence that phagocytosis of targets opsonized with anti-
`
`TABLE 1 | Examples of cell lines and strategies developed for low fucose antibody production.
`
`Cell line or system
`
`Enzyme defect
`
`Approximate Fc core fucosylation level (normal is >90%)
`
`References
`
`Lec 13 (CHO mutant)
`YB2/0 (rat)
`CHO FUT8-/-
`(e.g. Potelligent®)
`CHO GMD-/-GFT-/-
`CHO FX-/-
`CHO GMD-/-
`CHO GnTIII+++
`(e.g. GlycomAbs®)
`
`Defective GMD
`Low FUT 8
`FUT8 KO
`
`GMD+SLC35C1 KO
`FX KO
`GMD KO
`GnTIII overexpression
`
`10%
`9-30%
`0%
`
`0%
`6-8%
`1-3%
`10-15%
`
`(37, 43)
`(38)
`(41, 42)
`
`(45)
`(44)
`(44)
`(46, 47)
`
`FUT, Fucosyltransferase; GMD, GDP-mannose 4,6 dehydratase; FX, GDP-4-keto 6-deoxymannose 3,5-epimerase-4-reductase; GnTIII, N-acetylglucosamine transferases III; SLC35C1,
`GDP-fucose transporter; KO, knock out.
`
`Frontiers in Immunology | www.frontiersin.org
`
`5
`
`June 2022 | Volume 13 | Article 929895
`
`5 of 18
`
`OnCusp
`Ex. 1036
`
`

`

`Golay et al.
`
`IgG Fucosylation
`
`CD20 by liver Kupfer cells in vivo is enhanced by the
`afucosylated mAb (103), but this point still needs to be further
`investigated and confirmed. In the context of polyclonal
`alloimmune anti-HPA1a IgG1 antibodies (see below),
`increased phagocytosis of target platelets in vitro by both
`monocytes (via FcɣRIIIA) and PMN (via FcɣRIIIB) has been
`reported (104).
`ADCC by ɣd T Cells
`FcɣRIIIA is also expressed by ɣd T cells and these can mediate
`ADCC in presence of IgG1 antibodies (105, 106). There are
`reports of increased ADCC by ɣd T cells in presence of low
`fucose anti-CD20 mAb obinutuzumab compared to rituximab
`
`(58, 107). ɣd T cells represent <5% of T cells in the circulation of
`healthy individuals; they are also localized in non-lymphoid
`tissues and constitute the majority of immune cells in some
`epithelia (108). The role of ɣd T cells in the response to IgG1
`antibody in vivo is not well understood.
`
`ADCC and Trogocytosis by PMN
`IgG1 with low Fc core fucose also binds more strongly than
`fucosylated antibody to FcɣRIIIB, which has >97% sequence
`identity with FcɣRIIIA in its extracellular IgG binding domain
`(14, 15, 55). FcɣRIIIB is a GPI-linked receptor lacking activating
`module (ITAM), is expressed only by PMN and at high levels on
`these cells (5, 55, 109). Low fucose anti-CD20 therapeutic
`
`TABLE 2 | Selected therapeutic antibodies with low or no fucose, approved by FDA/EU or in clinical development.
`
`Diseases
`
`Major findings
`
`references
`
`% f
`
`ucose
`
`Method of
`defucosylation
`
`Antibody
`isotype
`
`Antibody name
`(code)
`
`Antigen
`
`1.1. Approved antibodies (FDA/EU)
`Obinutuzumab
`CD20
`Humanized
`(GA101)
`IgG1
`
`CHO overexpressing
`GnTIII (GlycoMAb)
`
`About
`15%
`
`B-NHL
`
`Higher ADCC by NK and ɣd Tells, more effective than
`RTX in vivo in some mice models or in primates
`
`(47, 55,
`58–62)
`(63)
`(64)
`
`(65)
`
`(66)
`
`(67, 68)
`
`(69–72)
`
`(73–75)
`
`(76, 77)
`
`(78)
`
`(79)
`
`(80–82)
`
`(83)
`
`(84)
`
`(85)
`(86)
`
`(Continued)
`
`Phase III in CLL compared to RTX in combination with
`CLb
`Phase III studies with different chemotherapy regimen
`in CLL and compared to RTX
`Phase III studies in diff chemo combinations compared
`to RTX in untreated FL.
`Equivalent ADCC by afucosylated mAb, but with 10-fold
`lower antigen expression on target compared to
`fucosylated mAb
`Increased ADCC in vitro. Depletes B cells more
`effectively than fucosylated antibody in hCD19
`transgenic mice (PB, spleen and BM)
`Increased ADCC in vitro. Efficacy in depleting
`eosinophil in non-human primates and in clinical
`Phase III studies
`Increased ADCC.
`In vivo increased activity in hFcɣRIII+ mice. Phase III
`trial in breast cancer compared to trastuzumab
`
`Increased ADCC of naked mAbs. Phase II ORR 31%.
`72% survival at 6 months
`
`0%
`
`0%
`
`0%
`
`0%
`
`0%
`
`Cutaneous T
`cell lymphoma
`
`Neuromyelitis
`optica
`
`Severe
`asthma with
`eosinophilia
`Advanced
`metastatic
`HER2+++
`breast cancer
`Multiple
`myeloma
`
`Mogamulizumab
`(KW-7061)
`
`CCR4
`
`Humanized
`IgG1
`
`CHO FUT8 -/-
`(Potelligent)
`
`Inebilizumab
`(MEDI 551)
`
`Benralizumab
`(MEDI-563)
`
`CD19
`
`IL-5Ra
`
`Humanized
`IgG1
`
`CHO FUT8 -/-
`(Potelligent)
`
`Humanized
`IgG1
`
`CHO FUT8 -/-
`(Potelligent)
`
`Margetuximab
`(MGAH22)
`
`HER2
`
`Chimeric
`IgG1
`
`CHO FUT8 -/-
`Also mutation in Fc to
`decreased CD32B
`binding
`CHO FUT8 -/-
`
`BCMA
`
`IgG1-MMAF
`ADC
`
`Belantamab
`vedotin
`(GSK2857916)
`Amivantamab
`(JNJ-61186372)
`
`EGFRxMET Humanized
`bispecificIgG1
`
`Low fucose producing
`cell line
`
`<10% Non-small cell
`lung cancer
`(NSCLC)
`
`Increased ADCC, not ADCP compared to high fucose
`variant. Phase III NSCLC
`
`1.2. Selected antibodies in clinical studies
`Ublituximab
`CD20
`Chimeric
`(Emab-6)
`IgG1
`
`YB2/0
`
`24%
`
`Tomuzotuximab
`(cetuGEX)
`
`EGFR
`
`EGFR
`
`Imgatuzumab
`GA201
`(RG7160)
`
`Humanized
`IgG1
`(cetuximab
`seq)
`Humanized
`rat IgG1
`(ICR62)
`
`Glyco Express
`System®
`
`0%
`
`CLL, B-NHL,
`multiple
`sclerosis,
`neuromyelitis
`optica
`Advanced
`carcinoma
`
`High ADCC and ADCP (not compared with fully
`fucosylated antibodies). Phase I and II trials in B-NHL,
`CLL and autoimmune diseases + neuromyelitis optica.
`Phase III in CLL with or w/o ibrutinib (ORR 85% vs
`65%)
`Increased ADCC in vitro, Phase I study
`Phase II study comparing CetuGEX with cetuximab
`combined with chemo: no difference observed
`
`CHO stably expressing
`GnTIII (GlycomAb)
`
`15%
`
`Carcinoma
`
`Increased ADCC in vitro. Higher efficacy in mouse
`models (SCID beige or SCID hFcɣRIIIA tg) also in
`combination with chemotherapy
`Phase I study in EGF+++ solid tumors
`
`Frontiers in Immunology | www.frontiersin.org
`
`6
`
`June 2022 | Volume 13 | Article 929895
`
`6 of 18
`
`OnCusp
`Ex. 1036
`
`

`

`Golay et al.
`
`TABLE 2 | Continued
`
`IgG Fucosylation
`
`Diseases
`
`Major findings
`
`references
`
`% f
`
`ucose
`
`Method of
`defucosylation
`
`Antibody
`isotype
`
`Antibody name
`(code)
`
`Antigen
`
`Open label study in advanced CRC. Decrease NK
`post treatment in PB
`Enhanced ADCC in vitro
`Favorable combination of GA201 and chemo in vitro
`and in carcinoma models in SCID mice
`Open label study of GA201 vs cetuximab in head &
`neck squamous carcinoma (N=44). Greater decrease
`in NK in PB and greater cytokine release with GA201
`vs CTX. No difference in clinical response.
`Phase I
`
`Increased ADCC in vitro.
`Enhanced activity in vivo in hFcɣRIIIA tg mice.
`Phase I in HER2+++ solid tumors
`Phase I study
`
`Increased ADCC in vitro. Efficacy in vivo in mouse sc
`SCID model. Recruitment of NK and T cells into
`tumor.
`
`(87)
`(88)
`
`(89)
`
`(90)
`
`(37, 91–94)
`
`(95, 96)
`
`(97, 98)
`
`Ulcerative
`colitis
`HER2+++
`tumors
`
`0%
`
`0%
`
`Hematological
`and solid
`cancers
`Gastric cancer
`FGFR2b+++
`
`Human IgG1
`
`FUT8 -/-Potelligent
`
`0%
`
`KHK4083
`
`Tragex
`
`Cusatuzumab
`(JNJ-74494550,
`ARGX-110)
`Bemarituzumab
`(AMG 522)
`
`OX40
`
`HER2
`
`CD70
`
`FGFR2b
`
`Humanized
`IgG1
`
`Humanized
`IgG1
`
`Humanized
`IgG1
`
`Glyco Express
`system®
`(FUT8-/-)
`CHO FUT8 ko
`(Potelligent)
`
`CHO FUT8
`
`ADCC, Antibody dependent cellular cytotoxicity; ADCP, Antibody dependent cellular phagocytosis; BM, Bone marrow; B-NHL, B-Non Hodgkin’s lymphoma; CLb, chlorambucil; CLL,
`Chronic lymphocytic leukemia; CR, Complete response; EGFR, Epidermal growth factor receptor; FL, follicular lymphoma; PB, Peripheral blood; NSCLC, Non-small cell lung cancer; ORR,
`Overall response rate; RTX, Rituximab; SCID, Severe combined immunodeficient.
`
`antibody obinutuzumab was shown to activate PMN more
`effectively than rituximab, which is ≥90% fucosylated, initially
`suggesting that enhanced FcɣRIIIB binding was responsible for
`this effect (55). However, subsequent experiments performed
`with PMN isolated from a rare FcɣRIIIB null donor, as well as
`the observation that PMN may express very low levels of
`FcɣRIIIA, indicated that PMN activation by afucosylated anti-
`CD20 antibodies may be mediated by FcɣRIIIA, FcɣRIIIB, or
`both, depending on conditions (110). The presence of low levels
`of FcgRIIIA on resting or activated PMN however still needs to
`be confirmed. The activation of PMN by anti-CD20 antibodies
`does not induce ADCC or phagocytosis, but only trogocytosis
`and cytokine production (55, 110). Interestingly other
`antibodies, such as anti-EGFR antibodies do mediate ADCC of
`tumor targets by PMN, but this function strictly requires
`FcɣRIIA (15). Furthermore, PMN and FcɣRIIA mediated
`ADCC is inhibited by FcɣRIIIB (111, 112). Indeed, ADCC by
`PMN is diminished in the presence of afucosylated anti-EGFR,
`because the latter binds more strongly to FcɣRIIIB, which is
`highly abundant on PMN and is thought to compete with
`FcɣRIIA. Therefore, at least in this context, FcɣRIIIB may act
`as a decoy receptor (111, 112). It remains to be established
`whether and how the level of fucosylation of other antibodies, for
`example those directed against microbes, affect PMN functions.
`
`Cytokine Release by Immune Cells
`Several studies indicate that afucosylated IgG1 antibodies induce
`a more rapid and intense cytokine release by NK cells,
`monocytes/macrophages, PMN and/or ɣd T cells compared to
`their fully fucosylated counterparts. Induced cytokines include
`IFN-ɣ, MCP1, IL-6, TNF, MIP1ab, Rantes, IL-8 (55, 100, 107,
`113–115). The level of cytokines induced in vitro is however
`generally quite low and the significance of such release in vivo is
`
`not fully clear. Nonetheless, the more frequent or severe
`immediate reaction syndrome observed in patients treated with
`obinutuzumab compared to rituximab indicates that increased
`cytokine release, particularly IL-6 and IFN-ɣ by afucosylated
`antibodies may be relevant in vivo (64, 116, 117). Cytokine
`release should therefore be carefully studied during the pre-
`clinical and clinical development of afucosylated antibodies.
`The functional effects of IgG1 Fc core afucosylation are
`summarized in Figure 4.
`
`EVIDENCE THAT AFUCOSYLATION
`ENHANCES THE EFFICACY OF
`THERAPEUTIC MABS IN VIVO IN
`ANIMAL MODELS
`
`Studying the role of Fc core fucosylation using small animal
`models is complicated by the fact that mice express different set
`of FcɣRs compared to humans (5, 118–120). The more recently
`discovered mouse FcɣRIV molecule has been shown several
`years ago to be the murine ortholog of human FcɣRIIIA and
`to be an important mediator of human IgG1 efficacy in mice.
`Afucosylated human IgG1 binds with higher affinity also murine
`FcɣRIV (121). This suggests that, despite species differences,
`human IgG1 therapeutic antibodies can be tested in mice, at least
`for some aspects of their functions. However, because of
`differences in FcɣR expression pattern in mice, humanized
`mice or human FcɣRIIIA transgenic mice may be used more
`appropriately. Nonetheless one should bear in mind that mice do
`not have the equivalent of FcɣRIIIB, which therefore makes this
`species not completely adequate to test the role of PMN in
`antibody efficacy in vivo, unless fully humanized models are used
`
`Frontiers in Immunology | www.frontiersin.org
`
`7
`
`June 2022 | Volume 13 | Article 929895
`
`7 of 18
`
`OnCusp
`Ex. 1036
`
`

`

`Golay et al.
`
`IgG Fucosylation
`
`FIGURE 4 | Major mechanisms of action of Fc core afucosylated IgG1 antibodies. IgG1 antibodies lacking Fc core a-1,6-fucose show a 10-100 fold increased
`binding to FcɣRIIIA and FcɣRIIIB on the indicated immune cells (NK, monocytes/macrophages and PMN), which results in increased ADCC by NK cells, enhanced
`competition with plasma IgGs, increased PMN activation, increased inhibition of FcgRIIA-mediated ADCC by PMN, induced by some antibodies (e.g. anti-EGFR
`mAbs). The effect of monocyte/macrophage induced ADCP is less clear. Low Fc core fucose can also increase release of cytokines, such as IL-6, TNF-a and IL-8,
`both in vitro and in vivo.
`
`(119). Despite these caveats, it is worth noting that afucosylated
`or low fucose anti-cancer mAbs have consistently been shown to
`control tumor growth more efficiently than their fully
`fucosylated parent molecules in mice (69, 84, 92, 122), as well
`as macaque, the latter having a more similar pattern of FcɣRs to
`humans (123). These results have led to an increased attention to
`the N-glycan profile of therapeutic mAbs (124), as well as the
`development of new afucosylated therapeutic antibodies in a
`variety of clinical contexts. These are discussed below.
`
`ANTI-CANCER MABS WITH LOW CORE
`FUCOSE IN THE CLINIC
`
`Many antibodies with no or low Fc core fucose have entered pre-
`clinical or clinical development to treat different diseases. Seven
`such antibodies have been approved so far by FDA/EMA
`(Table 2). They include one bispecific and one ADC, the
`others being unconjugated IgG1 antibodies. Five are approved
`for cancer therapy and two for immune/inflammatory diseases.
`Many other antibodies are in clinical trial (Table 2, part 2) or in
`pre-clinical development (47, 55, 58–79). In all cases, a
`significantly enhanced ADCC activity by NK cells has been
`shown in vitro and, in some, more effective target depleting
`
`activity in vivo in mice or non-human primates, as discussed
`above (Table 2).
`It is obviously difficult to evaluate whether removal of Fc core
`fucose significantly increases the efficacy of therapeutic anti-
`cancer mAbs in patients without performing phase III
`comparative studies of parent and afucosylated counterpart,
`studies that are rarely performed. Glycoengineered anti-CD20
`antibody obinutuzumab was the first therapeutic antibody with
`low fucose to be approved in 2013 for the treatment of B-cell
`neoplasia. It has been extensively compared with rituximab in
`vitro and in clinical trials in Phase III studies in chronic
`lymphocytic leukemia and B-NHL, usually in combination
`with chemotherapy. Obinutuzumab has shown either enhanced
`activity or, in some cases, non-inferior activity to rituximab
`depending on disease contexts and treatment pr