Aglycosylated immunoglobulin G1variants
productively engage activating Fc receptors
Stephen L. Sazinskya,1, Rene ´ G. Ottb,1, Nathaniel W. Silverc, Bruce Tidorb,d, Jeffrey V. Ravetchb,2,
and K. Dane Wittrupa,e,f,2
Departments ofaBiological Engineering,cChemistry,dElectrical Engineering and Computer Science,eChemical Engineering, andfKoch Institute for
Integrative Cancer Research, Massachusetts Institute of Technology, 77 Massachusetts Avenue, Cambridge, MA 02139; andbLaboratory of Molecular
Genetics and Immunology, The Rockefeller University, 1230 York Avenue, New York, NY 10021
Contributed by Jeffrey V. Ravetch, September 16, 2008 (sent for review August 25, 2008)
Immunoglobulin G plays a vital role in adaptive immunity and
antibody-based therapy through engagement of its Fc region by
the Fc? receptors (Fc?Rs) on immune cells. In addition to specific
protein-protein contacts, N-linked glycosylation of the IgG Fc has
been thought to be essential for the recognition of Fc by Fc?R. This
requirement for the N-linked glycan has limited biomanufacture of
therapeutic antibodies by restricting it to mammalian expression
systems. We report here aglycosylated Fc domain variants that
maintain engagement to Fc?Rs, both in vitro and in vivo, demon-
strating that Fc glycosylation is not strictly required for the acti-
vation of immune cells by IgG. These variants provide insight into
how the N-linked glycan is used biologically in the recognition of
Fc by Fc?Rs, as well as represent a step toward the production in
alternative expression systems of antibody-based therapeutics
capable of eliciting immune effector functions.
surface plasmon resonance ? antibody engineering ? directed evolution ?
yeast display ? antibody biomanufacture
therapy (2, 3). IgGs act as the adaptor between a pathogen and
the immune response by simultaneously binding antigen through
their variable regions and activating an immune response
through interaction of conserved Fc regions with Fc?Rs on cells
of the activating receptors Fc?RI, Fc?RIIA, and Fc?RIIIA, and
the inhibitory receptor Fc?RIIB. While Fc?RI binds IgG with
high affinity (nanomolar binding constants), Fc?RIIA,
Fc?RIIB, and Fc?RIIIA bind IgG with micromolar affinity,
becoming activated only via avid multivalent interactions with
opsonized antigen (1). The binding of IgG to Fc?R is highly
sensitive to the presence of glycosylation at a single N-linked
glycosylation site at asparagine 297 (N297) in its CH2 domain (4,
5), with a loss of binding to the low-affinity Fc?Rs observed in
N297 point mutants (6, 7), enzymatic Fc deglycosylation (8),
recombinant IgG expression in the presence of the N-linked
glycosylation-inhibitor tunicamycin (9), or expression in bacteria
(10, 11). In addition, the nature of the carbohydrate attached to
sensitivity of Fc?R binding to specific glycoforms has limited
therapeutic antibody biomanufacture to mammalian expression
systems, and has led to the development of glycosylation-
engineered mammalian cell lines (13, 14) and microbial strains
with humanized glycosylation (15) as methods of enhancing
In crystal structures of the complex, Fc?R/Fc contact is
mediated not only by protein-protein contacts, but also by
specific interactions with the glycan on the Fc that are proposed
to contribute to binding affinity (16, 17). Additional intramo-
lecular contacts are made between the Fc-linked glycan and
residues on the IgG CH2 domain, and it is thought that these
interactions stabilize an open Fc conformation capable of being
engaged by Fc?R (4). Successive truncation of an IgG1glycan
c gamma receptor (Fc?R) engagement is essential to the
function of IgG in both immunity (1) and in antibody-based
results in an incremental loss of binding affinity (8) and con-
However, glycosylation is not strictly required for engagement of
all Ig receptors with their corresponding Fc ligands, notably in
the binding of IgE Fc to Ig?R (19). Interestingly, the IgE Fc
adopts a similar mode of binding to FcR as the IgG1Fc in the
IgG1 Fc:Fc?RIII complex and both receptors and Fcs share
structural similarity (20). In the IgG1 Fc:Fc?RIII complex,
extensive contacts are made by both chains of the IgG1hinge
region, with additional receptor contacts made by the B/C loop,
F/G loop, and both side chains and glycosylation of the C?/E loop
Fig. S1]. It is particularly striking that this loop plays a part in
receptor recognition through both direct side chain contacts as
well as in encoding information for a critical posttranslational
In the present study, we reasoned that by optimizing the
protein-protein interactions about the C?/E loop:Fc?R interface
at the expense of glycosylation, we could identify aglycosylated
IgG1 variants that maintain engagement to Fc?Rs. Here, we
T299 of the glycosylation motif lead to aglycosylated Fcs that
maintain engagement of Fc?Rs, and in a particular example are
active in vivo. Such aglycosylated antibodies would be facile
templates for further efforts to engineer Fc-effector functions
and potentially enable a far wider range of options for thera-
peutic antibody biomanufacture.
Screening for Aglycosylated Fc Variants that Bind Fc?RIIA. To deter-
mine if glycosylation of the Fc was an absolute requirement for
Fc?R engagement by hIgG1, we constructed saturation mu-
tagenesis libraries at the Fc C?/E loop and screened them by
displaying the full-length IgG variants on the yeast cell surface
(Fig. S2). In this display system, the femtomolar affinity fluo-
rescein-binding 4m5.3 single-chain antibody (21) was reformat-
ted as a hIgG1, allowing 4m5.3 hIgG1 library variants to be
captured on fluorescein-labeled yeast from which they are
secreted, by using a cell-surface secretion capture assay (22) in
conjunction with an engineered leader sequence that allows for
the improved secretion of fully-assembled hIgG1from S. cerevi-
siae (Rakestraw JA, S.L.S., Piatesi A, Antipov E, K.D.W.,
unpublished data). Three saturation libraries centered about the
Author contributions: S.L.S., R.G.O., N.W.S., B.T., J.V.R., and K.D.W. designed research;
S.L.S., R.G.O., and N.W.S. performed research; S.L.S., R.G.O., N.W.S., B.T., J.V.R., and K.D.W.
analyzed data; and S.L.S., R.G.O., N.W.S., B.T., J.V.R., and K.D.W. wrote the paper.
The authors declare no conflict of interest.
1S.L.S. and R.G.O. contributed equally to this work.
2To whom correspondence should be addressed. E-mail: email@example.com or ravetch@
This article contains supporting information online at www.pnas.org/cgi/content/full/
© 2008 by The National Academy of Sciences of the USA
www.pnas.org?cgi?doi?10.1073?pnas.0809257105 PNAS ?
December 23, 2008 ?
vol. 105 ?
no. 51 ?
C?/E loop—theoretically encoding all amino acid combinations
at residues 296–299, 297–299, and 297–300—were pooled and
screened for binding to fluorophore-labeled tetramers of a
soluble form of the Fc?RIIA131Rallele by multiple rounds of
Using this screening strategy, in addition to glycosylated
variants and the wild type (WT) clone, mutants lacking the
canonical Asn-X-Ser/Thr N-linked glycosylation motif were en-
riched from the C?/E loop libraries for binding to Fc?RIIA (Fig.
1B). After two rounds of screening, three aglycosylated motifs
were identified: the double mutants S298G/T299A and S298G/
T299G, and the single mutant T299A. After a third round of
screening at increased stringency, the sublibrary was dominated
by the S298G/T299A variant, suggesting it as the highest-affinity
Fc?RIIA binding motif in the library (Fig. 1C).
To study the contributions of the S298G and T299A mutations
to receptor binding, a series of point mutants were constructed,
secreted from yeast, and assayed for their ability to bind to
Fc?RIIA131R(Fig. 1D). Both the S298G and T299A mutations
alone retain binding to Fc?RIIA at comparable or increased
levels to WT IgG1, and the S298G mutation in the aglycosylated
T299A background yields a variant capable of binding Fc?RIIA
to a much greater extent. The S298G/T299A double mutant is
incapable of rescuing binding in the N297Q, N297D, and N297A
backgrounds, suggesting there is a strong requirement for as-
paragine at position 297 for Fc?RIIA binding, even in the
absence of conjugated carbohydrate, a finding consistent with its
conservation in the initial library screen.
S298G/T299A Is Aglycosylated and Binds to Fc?Rs. To confirm that
the potential aglycosylated motifs identified from the yeast-
of engaging Fc?Rs, WT, S298G/T299A, and the nonreceptor
binding N297Q aglycosylated control were expressed and puri-
fied from HEK 293 cells. Both N297Q and S298G/T299A exhibit
increased mobility by reducing SDS/PAGE (Fig. 2A) and, crit-
ically, are not recognized by the mannose-specific lectin LCA
(Fig. 2B), consistent with the expected absence of N-linked
glycosylation. Surface plasmon resonance measurements show
that S298G/T299A binds to both Fc?RIIA alleles, Fc?RIIA131R
and Fc?RIIA131H, and to Fc?RIIB; however, this mutant does
not bind to Fc?RIIIA nor the complement component C1q, and
binding to Fc?RI was weakened by 10-fold. (Fig. 2C). S298G/
T299A binds Fc?RIIA131Rwith a dissociation constant (Kd) of
representation of the crystal structure of the hIgG1Fc complex with Fc?RIII
(PDB ID 1E4K). Fc?RIII is shown in green and both chains of the Fc in pale blue.
(red), and F/G loop (purple). Fc glycosylation is shown in yellow. The high-
lighted area shows an enlarged view of the C?/E loop interaction with Fc?RIII.
Fc variants isolated for Fc?RIIA binding after two rounds of FACS. Displayed
sequences represent the residues randomized in the saturation libraries,
denote number of times a particular mutant was isolated. In some cases,
identical protein sequences were isolated from multiple unique clones at the
DNA level. Sequences of glycosylated variants enriched from the screen have
been omitted. (C) Unique sequences of aglycosylated Fc variants isolated for
Fc?RIIA binding after a third round of FACS, using a more stringent screening
strategy. (D) Binding of yeast-produced 4m5.3 hIgG1 variants to 10 nM
Fc?RIIA131Rstreptavidin-Alexa 647 tetramers. IgG from yeast culture superna-
intensity (MFI) of receptor labeling was measured by flow cytometry. Protein
A Alexa 647 labeling of all samples was assessed to determine similar IgG
loading (data not shown). All data represent the average of two trials.
Aglycosylated C?/E loop variants with Fc?RIIA binding. (A) Cartoon
wt 5.5 5.09.813 4.60.3
N297Qn.b.n.b. n.b. n.b.n.b. n.b.
7.0 1.75.7 n.b.n.b. n.b.
Reducing SDS/PAGE of HEK-produced WT 4m5.3 hIgG1and the aglycosylated
variants S298G/T299A and N297Q. (B) Glycan blotting with the mannose-
specific lectin LCA (Upper). Coomassie staining of SDS/PAGE was assessed to
demonstrate similar protein loading (Lower). (C) Dissociation constants (Kd)
for binding of Fc?Rs and C1q to WT 4m5.3 hIgG1and aglycosylated variants;
Fc?Rs with 4m5.3 IgG immune complexes. Cells were incubated with 1 ?g of
ICs and analyzed by flow cytometry.
www.pnas.org?cgi?doi?10.1073?pnas.0809257105 Sazinsky et al.
Fc?RIIA131Hwith a Kdof 7.0 ?M, slightly weaker than WT. A
small increase in affinity compared to WT for Fc?RIIB was also
observed, suggesting a preferential binding of Fc?R with an
arginine at position 131 (see Fig. 2C and Fig. S3).
with Fc?RIIA131R, Fc?RIIA131H, and Fc?RIIB were labeled
Both WT and S298G/T299A IgG ICs label all receptor-
expressing CHO cells in a concentration-dependent manner,
demonstrating that S298G/T299A binds Fc?Rs in this context as
well. Consistent with the soluble binding measurements, S298G/
T299A IgG ICs label Fc?RIIA131R- and Fc?RIIB-expressing
only intermediate labeling of the Fc?RIIA131Hallele, compared
to WT and the aglycosylated control.
S298G/T299A Activates Fc?RIIA in Vivo. To determine whether this
aglycosylated IgG is functional in vivo, a murine platelet clear-
ance model was used to test the extent of S298G/T299A activity.
The platelet integrin antigen-binding antibody 6A6 was refor-
matted as a mouse-human IgG1chimera and the S298G/T299A
mutations subsequently introduced into the human Fc domain.
The antibody was tested in a transgenic mouse model, in which
the endogenous murine Fc?Rs have been deleted by gene
targeting and the human activation Fc?R, Fc?RIIA131R, is
expressed as a transgene, thus maintaining cell-type expression
appropriate for the human transgene (23). Mice with this
genotype were treated with WT, N297A, and S298G/T299A 6A6
hIgG1purified from HEK 293 cells and the extent of platelet
clearance measured over time (Fig. 3). After 4 hours, S298G/
T299A-6A6-treated mice (n ? 3) showed a statistically signifi-
cant drop in platelet count when compared to those treated with
N297A-6A6 or PBS, exhibiting a response that was comparable
to WT-6A6 and demonstrating the ability of S298G/T299A
to productively engage Fc?RIIA in vivo and result in platelet
Model of S298G/T299A-Fc?RIIA Interaction. To explore the struc-
tural basis for Fc?R binding of this aglycosylated Fc domain
variant, we constructed homology models of Fc:Fc?RIIA com-
plexes based on the previously solved structures of the IgG1Fc,
the Fc?RIIA structure (24), and the Fc:Fc?RIII complex (17)
(Fig. 4). Three features emerge from this modeling. First, in the
model of the WT interaction, there is only limited interaction
between the two N-linked glycans and Fc?RIIA (see Fig. 4A).
The asymmetric nature of the IgG1 Fc:Fc?RIIA interaction
predicts that the glycan attached to the B chain of the Fc dimer
may interact with residues K117, T119, F121, S126, and F129 of
the receptor, whereas the glycan attached to the other chain (the
A chain) does not make contact with Fc?RIIA. These
glycan:Fc?R contacts provide negligible calculated screened
electrostatic intermolecular interactions (approximately 0 kcal/
mol), compared to the much larger intramolecular ones between
glycan and Fc [roughly ?1.3 kcal/mol, with a dominant contri-
bution from N297/glycan(B)–D265(B)] and suggest that both
oligosaccharides are primarily interacting with their respective
Fc chains. Second, N297 is important for the Fc:Fc?RIIA
Platelet count (%)
0 10 20 3040 50 6070 80
mice treated with chimeric antiplatelet antibody 6A6 hIgG1or the aglycosy-
to platelet counts before injection, counts 4 h following injection of the 6A6
antibodies were 15.8 ? 13.9% for WT, 39.5 ? 5% for S298G/T299A, 71.0 ?
6.1% for N297A, and 85.7 ? 11.0% for those treated with PBS only. S298G/
T299A-treated mice displayed a statistically significant difference in platelet
reduction compared to N297A (P ? 0.017) and PBS (P ? 0.002), but not a
statistically significant difference compared to WT (P ? 0.068).
Platelet clearance in murine Fc?R knockout, Fc?RIIA131Rtransgenic
hFc?RIIA. (A) Structure of the WT B-chain Fc fragment bound to the con-
structed Fc?RIIA131Rhomology model. The portion of Fc?RIIA (chain C) high-
lighted as orange sticks shows those side chains that were conformationally
relaxed, through rotamerization, during model construction. (B) Structure of
the aglycosylated N297(B) interactions with a superposition of the possible
discrete side chain conformations (rotameric states) of N297(B), T119(C), and
S126(C) in the S298G/T299A mutant form of the Fc fragment. (C and D)
Structure of the K117(C)-D265(B) salt bridge in the WT structure (C) enclosed
salt bridge as seen in the aglycosylated S298G/T299A mutant (D). (E and F)
Residual potential after binding mapped onto the interaction face of the B
White regions indicate areas of ideal complementarity between the Fc frag-
ment and Fc?RIIA131R, while deep red or blue regions indicate areas of poor
complementarity because of ligand desolvation costs uncompensated by
interactions made upon binding. Red corresponds to negative residual po-
both panels, is indicated in units of kT/e.
Homology model of WT and S298G/T299A Fc interactions with
Sazinsky et al.
December 23, 2008 ?
vol. 105 ?
no. 51 ?
interaction. Aglycosylated N297 has the potential to make
hydrogen bond interactions across the interface with S126 of the
receptor (see Fig. 4B). These interactions may be mediated by a
bridging water molecule that can be observed nearby in an
unbound Fc?RIIA crystal structure (24). Replacement of N297
with glutamine or alanine disrupts this interaction (and fails to
make similar, stabilizing ones) and is consistent with the ob-
served absence of binding for such mutants (see Fig. 1D and Fig.
S5A). Interestingly, replacement with aspartic acid may be able
to make a similar interaction; however the greater desolvation
penalty of the charged side chain upon Fc?R binding likely
results in the reduced binding of this variant.
Finally, the intermolecular interaction between the aglycosy-
lated S298G/T299A mutant and Fc?RIIA includes a salt bridge
formed between D265 on the B chain of the Fc dimer and K117
on the Fc?R. In the WT structure, this interaction is shielded
from solvent by the oligosaccharide chain (see Fig. 4C). In the
aglycosylated S298G/T299A mutant, this salt bridge is exposed
to the solvent (see Fig. 4D), which nearly halves the screened
electrostatic interaction energy compared to WT (?5 kcal/mol
versus ?10 kcal/mol). However, this effect is more than com-
pensated in the S298G/T299A mutant by a reduced desolvation
penalty (25), a measure of the loss of electrostatic interactions
with solvent upon binding, resulting in an overall stabilized
structure. This effect is illustrated (see Fig. 4 E and F) by a
reduction in the residual electrostatic potential present on
D265(B) in the mutant compared to the WT; similarly, Fig. 4
also shows the S298G mutation results in a reduced desolvation
penalty that contributes to the stability of the mutant complex.
Thus, the predictions made by this homology model provide a
hypothetical mechanism for the stability of the aglycosylated
Fc:Fc?R complex, resulting from hydrogen bonding and elec-
trostatic interactions altered in the aglycosylated mutant.
Aglycosylated Fc Variants that Bind Fc?RIIIA. To evaluate the con-
tribution of individual sidechains in the C?/E loop to Fc?R engage-
ment, as well as the nature of the specificity between Fc?RIIA and
Fc?RIIIA seen in S298G/T299A, we constructed the full set of
single-point mutations at positions 297, 298, and 299, and assayed
yeast-secreted IgG variants for binding to both Fc?RIIA and
Fc?RIIIA (see Fig. S5). Side-chain scanning of 297 and 299
revealed additional mutations that remove the glycosylation motif
but retain residual weak receptor binding: T299H to Fc?RIIA, and
N297D and N297H to Fc?RIIIA176V(see Fig. S5 A and C). T299A
is the only aglycosylated mutant identified that displays dual
moderate binding to the Fc?RIIIA176Vallele. Interestingly, the
nature of the side chains at position 299, and not just glycosylation,
greatly impacts receptor binding, as the yeast-expressed glycosy-
lated T299S mutant binds all receptors to a much lesser extent than
the WT Fc.
In contrast to positions 297 and 299, where mutations largely
disrupt the N-linked glycosylation motif Asn-X-Ser/Thr, multi-
ple substitutions in a glycosylated Fc background are tolerated
at position 298 (see Fig. S5B). Fc?RIIA binding is much more
sensitive to substitution at position 298, with only glycine
(S298G) maintaining a level of binding that is comparable to
WT. In contrast, Fc?RIIIA tolerates an array of substitutions at
position 298, and the data highlight potential mutations for
engineering Fc?RIIIA versus Fc?RIIA/Fc?RIIB specificity,
such as the previously identified S298A and S298N mutations (6,
26). Only S298G maintained engagement to both Fc?RIIA and
Fc?RIIIA in our assay, a finding that, taken together with a
preference for threonine at 299 (T299), suggests an explanation
for the conservation of the motif N-S/G-T in IgG CH2 domains
across virtually all species (see Fig. S1).
While our initial efforts, which focused on Fc?RIIA, resulted
in specificity for Fc?RIIA and Fc?RIIB at the expense of
Fc?RIIIA binding (see Fig. 2C), the side-chain scanning data
suggested that aglycosylated Fcs that bind Fc?RIIIA with com-
parable affinity to WT could also be identified. Within the C?/E
loop, rational design of double mutants based upon the weakly
Fc?RIIIA-binding N297D and N297H substitutions yielded
variants that bound Fc?RIIIA176Vat levels 10 to 40% of WT and
with specificity for Fc?RIIIA (see Fig. 5A), a desired property
in engineering Fcs with enhanced immune effector functions (2).
In a separate strategy, the consensus mutations K326E, K290E,
and K290N—identified in a separate screen for improved
Fc?RIIA binding (data not shown) and through the efforts of
previous groups (26, 27)—were introduced into the T299A
background. Incorporation of the K326E mutation, located at
the base of the F/G loop, led to enhanced binding for Fc?RIIIA,
approaching WT levels for Fc?RIIIA176Vand weakly binding
Fc?RIIIA176F(see Fig. 5B). This result suggests that additional
second-site mutations at contact interfaces other than the C?/E
loop can lead to aglycosylated Fc?RIIIA- and Fc?RIIA-binding
Fcs with a range of affinities and specificities.
between IgG and Fc?Rs indicated a dependence on the N-linked
glycan attached to asparagine 297 on the IgG heavy chain. The
Fc variants described in this article clearly demonstrate that
glycosylation is not a strict requirement for Fc?R engagement,
either in vitro or in vivo. In an initial strategy, by generating
aglycosylated Fc variants that bind to Fc?RIIA and Fc?RIIB, we
could demonstrate that the set of mutations necessary to switch
from a WT glycosylated binder to a functionally aglycosylated
binder is fairly small; in our case it involved the introduction of
only two point mutations. In a second more directed screening
strategy, we could further demonstrate that by introducing
additional modifications into our aglycosylated mutants, we can
combine features from single mutants discovered from different
screenings, thereby modulating the overall affinity features of
the IgG variant. This combinatorial behavior of the contribution
of single mutations is of special interest for the engineering of
IgG variants with very well-defined binding properties.
In addition to the enhanced Fc?RIIA131Rbinding observed in
the aglycosylated S298G/T299A variant, we were able to restore
binding to Fc?RIIIA176Vto near WT levels, suggesting that
Relative binding (to wt)
secreted 4m5.3 hIgG1C?/E loop double point mutants (A) or T299A point
mutants with the ‘‘second-site’’ mutations K326E and K290E/N (B) were
loaded on fluorescein-conjugated yeast and assayed for binding to 10-nM
Fc?RIIIA176V, Fc?RIIIA176F, and Fc?RIIA131R(A only) streptavidin-Alexa 647 tet-
normalized to the WT signal.
Designed aglycosylated Fc mutants with Fc?RIIIA binding. Yeast-
www.pnas.org?cgi?doi?10.1073?pnas.0809257105Sazinsky et al.
further engineering can also lead to aglycosylated variants with
WT or improved binding to Fc?RIIIA. In particular, we antic-
ipate that introducing mutations into the T299A background,
which weakly binds both Fc?RIIA and Fc?RIIIA, will lead to
fully Fc?R competent aglycosylated antibody variants. Building
upon these aglycosylated Fc?RIIIA-binding variants will be
essential for their potential use as cytotoxic antibodies, which
have emerged as a promising class of therapeutics for treatment
of human cancer in recent years (28). Support for a critical role
for Fc?R engagement in the mechanism of antitumor activity,
and specifically for Fc?RIIIA, has come from three independent
studies, which found a strong, positive correlation between
patient response and the presence of specific alleles of the
activating Fc?R Fc?RIIIA that conferred enhanced binding for
the IgG1 Fc domain of the antibody (29–31). While the S298G/
T299A variant does not bind complement, the above studies, as
well as murine models that demonstrate a dominant role for
Fc?R engagement in therapeutic antibody activity (3), suggest
that restoration of complement binding would be unnecessary
for engineered Fc variants. In addition to their ability to bind
Fc?R, it will also be important to assess the stability of these
variants, as previously characterized Fc variants (32) and deg-
lycosylated WT Fc (33) have displayed reduced thermal stability.
Given the small number of mutations required to achieve
N-linked glycosylation-independent Fc?R binding, it is striking
that all naturally occurring IgGs nevertheless use this posttrans-
lational modification. Among different antibodies there is vari-
ation in the fucose and galactose-sialic acid attached to the core
glycan structure (Fig. S6), and it has been reported that these
variations dramatically influence the antibody activity. The
absence of fucose in the glycan was reported to enhance the
affinity of Fc?RIIIA for IgG up to 50-fold (13), thereby switch-
example, for cytotoxic antibodies, but also occurs when autoan-
tibodies generate pathogenic immune complexes and activate
autoimmune cascades. In contrast to fucose, the presence of
terminal sialic acid was demonstrated to be the critical factor for
the anti-inflammatory action of high-dose intravenous Ig (12,
34). Sialic acid reduces the affinity of Fc?Rs to IgG by 5- to
10-fold (12) and, in addition, marks IgGs and subsequently
allows them to bind to non-FcR lectins (34) and mediate
downstream actions through these unique interactions, resulting
in anti-inflammatory responses, including the up-regulation of
Fc?RIIB on effector macrophages (35). The conservation of the
N-S/G-T glycosylation motif among different species (Fig. S1) at
the expense of this posttranslational variability supports the view
that the glycan, although not necessarily required for Fc?R
binding, serves as a platform for further modulation of the IgG’s
function between an anti-inflammatory or inflammatory mode.
Finally, our demonstration that IgG variants that have uncou-
pled Fc?R binding from N-linked glycosylation can be generated
opens up unique possibilities for protein engineering and bi-
omanufacture. Our results suggest that receptor binding affinity
and specificity can be engineered on the simpler template of an
unmodified polypeptide chain, and these properties can be
selected by yeast surface display of aglycosylated Fc mutant
libraries. Such mutants could then be produced in essentially any
recombinant expression system without loss of the desired
altered effector functions.
Materials and Methods
Library Construction. Libraries were constructed by homologous recombina-
tion of a mutated heavy-chain constant-region insert into the 4m5.3 heavy-
ods (36). The 4m5.3 heavy-chain secretion vector was previously constructed
of the hIgG1CH1 to CH3 constant domains (Rakestraw JA, S.L.S., Piatesi A,
Antipov E, K.D.W., unpublished data). Preparation of mutagenic C?/E loop
gene inserts and digested template vector are detailed in SI Materials and
Methods. Gene inserts were transformed with digested template vector by
YVH10, containing a chromosomally integrated copy of the 4m5.3 light-chain
yeast-secretion vector. The 296–299 and 297–300 saturation libraries had
DNA level (324? 1.0 ? 106); the 297–299 library had ?4 ? 107transformants.
Oligonucleotides. Oligonucleotides are detailed in SI Materials and Methods.
Library Screening. Library screening was performed using the cell-surface
secretion assay (CeSSA) (22). Briefly, libraries were grown in SD-CAA (2%
glucose, 0.67% yeast nitrogen base, 0.54% Na2HPO4, 0.86% NaH2PO4?H2O,
0.5% casein amino acids) at 30 °C to an OD600of ?5, and then induced in YPG
(2% galactose, 2% peptone, 1% yeast extract, 0.54% Na2HPO4, 0.86%
labeled with fluorescein-PEG-NHS (Nektar) and reinduced in YPG containing
15% PEG (wt/vol) at 20 °C for 36 h. Cells were washed with PBS containing
0.1% (wt/vol) BSA (PBS/BSA) and labeled with biotinylated hFc?RIIA131Rpre-
loaded onto streptavidin-Alexa 647 (Invitrogen). The library was sorted on a
BD FACSAria (Becton Dickinson) and collected cells grown in SD-CAA supple-
mented with penicillin/streptomycin (Invitrogen), for a total of three rounds
of screening. Library populations were labeled at increasingly stringent con-
centrations of Fc?RIIA tetramer as follows: round one (50-nM Fc?RIIA tet-
ramer), round two (2-nM Fc?RIIA tetramer), and round three (80-pM Fc?RIIA
tetramer). All clones isolated from screening were retransformed into
YVH10/LC and individually assayed for Fc?RIIA binding.
Characterization of Yeast-Secreted Fc Mutants. Fc mutants freshly transformed
into YVH10/LC were grown in 5-ml SD-CAA at 30 °C until an OD600?5, then
induced in 5-ml YPG at 20 °C for 72 h. Cell-culture supernatants were loaded
onto fluorescein-conjugated yeast overnight at 4 °C; yeast were then washed
with PBS/BSA, labeled with 10 nM of biotinylated Fc?R preloaded onto
streptavidin-Alexa 647 at 4 °C for ?2 h, and analyzed by flow cytometry.
Labeling with 10-?g/ml Protein A-Alexa 647 (Invitrogen) was performed as a
separate IgG loading control for all samples.
Mice. ??/?Fc?RIIB?/?mice were generated in the Ravetch laboratory, back-
crossed for 12 generations to the C57BL/6 background, and crossed to
were used for the experiments and maintained at the Rockefeller University
animal facility. All experiments were performed in compliance with federal
laws and institutional guidelines and have been approved by the Rockefeller
Cell Culture. CHO cells were cultured according to the American Type Culture
Collection guidelines. CHO-hFc?RIIA131H, CHO-hFc?RIIA131R, and hFc?RIIB
were obtained by transfection of the pCMV-Script-hFc?RIIA131H, CHO-
hFc?RIIA131R, and hFc?RIIB plasmids and subsequent selection with 1-mg/ml
Antibodies and Recombinant Proteins. The 6A6-human Fc chimeric variants and
were cultured in DMEM supplemented with 1% Nutridoma SP (Roche). Cell-
culture supernatants were harvested 6 days after transfection, and protein was
precipitated by ammonium sulfate precipitation. The 4m5.3-human Fc chimeric
variants were produced by transient transfection of 293F cells (Invitrogen) and
subsequent purification from cell-culture supernatants. For protein production,
cells were cultured in Freestyle 293F Expression Medium (Invitrogen). Recombi-
were purified with protein G Sepharose (GE Healthcare) or immobilized protein
A (Pierce) by affinity chromatography. All proteins were dialyzed against PBS.
Purity was assessed by SDS/PAGE followed by Coomassie Blue staining.
Immune Complex Binding Assay. For studying immune complex binding to
(anti-FITC) chimera with 10 ?g of BSA-FITC (Sigma) in 1-ml PBS for 2 h at 37 °C
Sazinsky et al.
December 23, 2008 ?
vol. 105 ?
no. 51 ?
while shaking gently. CHO cells were stained for 2 h at 4 °C with 1 ?g, 0.5 ?g, Download full-text
0.2 ?g, or 0.1 ?g of ICs, washed with PBS and analyzed by FACS analysis.
Surface Plasmon Resonance Analysis. To determine the interaction between
soluble hFc?-receptors RIa (R&D Systems), Fc?RIIA131H, Fc?RIIA131R, Fc?RIIB,
Fc?RIIIA, CIq (Calbiochem), and 4m5.3 antibody chimera, steady-state affinity
coupling. Soluble hFc?-receptors were injected in five different concentra-
tions through flow cells at room temperature in HBS-EP running buffer
(Biacore) for 3 min at a flow rate of 30 ?l/min and dissociation was observed
to a control flow cell using Biacore T100 Evaluation software.
Lectin Blot. Lectin blotting is detailed in SI Materials and Methods.
In Vivo Model Systems. Mice were injected intravenously with 50-?g 6A6-hFc1
WT, N297A, or S298G/T299A in 100-?l PBS. Platelet counts were determined
?l from the retro-orbital plexus and measuring platelet counts of a 1:10
dilution in PBS/5% BSA in an Advia 120 hematology system (Bayer). Platelet
standard deviation of three mice per group.
Computational Modeling. Computational modeling is detailed in SI Materials
ACKNOWLEDGMENTS. We thank V. Voynov for assistance with HEK cell
culture; A. Rakestraw for helpful discussions and suggestions; the Massachu-
setts Institute of Technology Flow Cytometry and Biopolymers core facilities;
E. Clowney, A. Kim, T. Shabaneh and J. Pagan for technical assistance; and V.
Bryant, S. Howland, and M. Schmidt for critical review of the manuscript. This
work was supported by grants from the National Cancer Institute at the
Fellow at the Rockefeller University.
1. Nimmerjahn F, Ravetch JV (2008) Fcgamma receptors as regulators of immune re-
sponses. Nat Rev Immunol 8:34–47.
2. Desjarlais JR, Lazar GA, Zhukovsky EA, ChuSY (2007) Optimizing engagement of the
immune system by anti-tumor antibodies: an engineer’s perspective. Drug Discov
3. Nimmerjahn F, Ravetch JV (2007) Antibodies, Fc receptors and cancer. Curr Opin
4. Jefferis R, Lund J (2002) Interaction sites on human IgG-Fc for FcgammaR: current
models. Immunol Lett 82:57–65.
on the biological function and structure of human immunoglobulins. Annu Rev
Fc gamma RI, Fc gamma RII, Fc gamma RIII, and FcRn and design of IgG1 variants with
improved binding to the Fc gamma R. J Biol Chem 276:6591–6604.
7. Tao MH, Morrison SL (1989) Studies of aglycosylated chimeric mouse-human IgG. Role
of carbohydrate in the structure and effector functions mediated by the human IgG
constant region. J Immunol 143:2595–2601.
8. Mimura Y, et al. (2001) Role of oligosaccharide residues of IgG1-Fc in Fc gamma RIIb
binding. J Biol Chem 276:45539–45547.
IgG3 monoclonal antibodies can eliminate recognition by human cells expressing Fc
gamma RI and/or Fc gamma RII receptors. Biochem J 259:347–353.
10. Mazor Y, Van Blarcom T, Mabry R, Iverson BL, Georgiou G (2007) Isolation of engi-
rapid and efficient production of aglycosylated antibodies. J Immunol Methods
12. Kaneko Y, Nimmerjahn F, Ravetch JV (2006) Anti-inflammatory activity of immuno-
globulin G resulting from Fc sialylation. Science 313:670–673.
13. Shields RL, et al. (2002) Lack of fucose on human IgG1 N-linked oligosaccharide
improves binding to human Fcgamma RIII and antibody-dependent cellular toxicity. J
Biol Chem 277:26733–26740.
14. Yamane-Ohnuki N, et al. (2004) Establishment of FUT8 knockout Chinese hamster
ovary cells: an ideal host cell line for producing completely defucosylated antibodies
with enhanced antibody-dependent cellular cytotoxicity. Biotechnol Bioeng 87:614–
15. Li H, et al. (2006) Optimization of humanized IgGs in glycoengineered Pichia pastoris.
Nat Biotechnol 24:210–215.
16. Radaev S, Motyka S, Fridman WH, Sautes-Fridman C, Sun PD (2001) The structure of a
human type III Fcgamma receptor in complex with Fc. J Biol Chem 276:16469–16477.
human IgG1 Fc fragment-Fc gammaRIII complex. Nature 406:267–273.
18. Krapp S, Mimura Y, Jefferis R, Huber R, Sondermann P (2003) Structural analysis of
human IgG-Fc glycoforms reveals a correlation between glycosylation and structural
integrity. J Mol Biol 325:979–989.
19. Basu M, et al. (1993) Purification and characterization of human recombinant IgE-Fc
fragments that bind to the human high affinity IgE receptor. J Biol Chem 268:13118–
20. Garman SC, Wurzburg BA, Tarchevskaya SS, Kinet JP, Jardetzky TS (2000) Structure of
the Fc fragment of human IgE bound to its high-affinity receptor Fc epsilonRI alpha.
21. Boder ET, Midelfort KS, Wittrup KD (2000) Directed evolution of antibody fragments
with monovalent femtomolar antigen-binding affinity. Proc Natl Acad Sci USA
22. Rakestraw JA, Baskaran AR, Wittrup KD (2006) A flow cytometric assay for screening
improved heterologous protein secretion in yeast. Biotechnol Prog 22:1200–1208.
23. McKenzie SE, et al. (1999) The role of the human Fc receptor Fc gamma RIIA in the
immune clearance of platelets: a transgenic mouse model. J Immunol 162:4311–4318.
24. Maxwell KF, et al. (1999) Crystal structure of the human leukocyte Fc receptor, Fc
gammaRIIa. Nat Struct Biol 6:437–442.
25. Lee LP, Tidor B (2001) Optimization of binding electrostatics: charge complementarity
in the barnase-barstar protein complex. Protein Sci 10:362–377.
26. StavenhagenJB, etal.(2007)Fcoptimizationoftherapeuticantibodiesenhancestheir
ability to kill tumor cells in vitro and controls tumor expansion in vivo via low-affinity
activating Fcgamma receptors. Cancer Res 67:8882–8890.
27. Idusogie EE, et al. (2001) Engineered antibodies with increased activity to recruit
complement. J Immunol 166:2571–2575.
28. Waldmann TA (2003) Immunotherapy: past, present and future. Nat Med 9:269–
29. Cartron G, et al. (2002) Therapeutic activity of humanized anti-CD20 monoclonal
antibody and polymorphism in IgG Fc receptor FcgammaRIIIa gene. Blood 99:754–
30. Weng WK, Czerwinski D, Timmerman J, Hsu FJ, Levy R (2004) Clinical outcome of
lymphoma patients after idiotype vaccination is correlated with humoral immune
response and immunoglobulin G Fc receptor genotype. J Clin Oncol 22:4717–4724.
31. Weng WK, Levy R (2003) Two immunoglobulin G fragment C receptor polymorphisms
independently predict response to rituximab in patients with follicular lymphoma. J
Clin Oncol 21:3940–3947.
32. Oganesyan V, Damschroder MM, Leach W, Wu H, Dall’Acqua WF (2008) Structural
characterization of a mutated, ADCC-enhanced human Fc fragment. Mol Immunol
33. Mimura Y, et al. (2000) The influence of glycosylation on the thermal stability and
effector function expression of human IgG1-Fc: Properties of a series of truncated
glycoforms. Mol Immunol 37:697–706.
34. Anthony RM, et al. (2008) Recapitulation of IVIG anti-inflammatory activity with a
recombinant IgG Fc. Science 320:373–376.
35. Nandakumar KS, et al. (2007) Endoglycosidase treatment abrogates IgG arthritoge-
nicity: importance of IgG glycosylation in arthritis. Eur J Immunol 37:2973–2982.
36. Chao G, et al. (2006) Isolating and engineering human antibodies using yeast surface
display. Nat Protoc 1:755–768.
www.pnas.org?cgi?doi?10.1073?pnas.0809257105Sazinsky et al.