Content uploaded by Stefan Gerhardt
Author content
All content in this area was uploaded by Stefan Gerhardt on Nov 14, 2019
Content may be subject to copyright.
Structure of IL-17A in Complex with a Potent, Fully
Human Neutralizing Antibody
Stefan Gerhardt
1
, W. Mark Abbott
1
⁎, David Hargreaves
1
,
Richard A. Pauptit
1
, Rick A. Davies
1
, Maurice R. C. Needham
1
,
Caroline Langham
1
, Wendy Barker
1
, Azad Aziz
1
, Melanie J. Snow
1,2,3
,
Sarah Dawson
1
, Fraser Welsh
2
, Trevor Wilkinson
2
, Tris Vaugan
2
,
Gerald Beste
2
, Sarah Bishop
2
, Bojana Popovic
2
, Gareth Rees
2
,
Matthew Sleeman
2
, Steven J. Tuske
3
, Stephen J. Coales
3
,
Yoshitomo Hamuro
3
and Caroline Russell
2
1
AstraZeneca, Alderley Park,
Macclesfield, Cheshire SK10
4TG, UK
2
MedImmune, Milstein
Building, Granta Park,
Cambridge CB21 6GH, UK
3
ExSAR Corporation, 11 Deer
Park Drive, Suite 103,
Monmouth Junction, NJ 08852,
USA
Received 16 July 2009;
received in revised form
5 October 2009;
accepted 5 October 2009
Available online
14 October 2009
IL-17A is a pro-inflammatory cytokine produced by the newly identified
Th17 subset of T-cells. We have isolated a human monoclonal antibody to IL-
17A (CAT-2200) that can potently neutralize the effects of recombinant and
native human IL-17A. We determined the crystal structure of IL-17A in
complex with the CAT-2200 Fab at 2.6 Å resolution in order to provide a
definitive characterization of the epitope and paratope regions. Approxi-
mately a third of the IL-17A dimer is disordered in this crystal structure. The
disorder occurs in both independent copies of the complex in the asymmetric
unit and does not appear to be influenced by crystal packing. The complex
contains one IL-17A dimer sandwiched between two CAT-2200 Fab
fragments. The IL-17A is a disulfide-linked homodimer that is similar in
structure to IL-17F, adopting a cystine-knot fold. The structure is not
inconsistent with the previous prediction of a receptor binding cavity on IL-
17 family members. The epitope recognized by CAT-2200 is shown to
involve 12 amino acid residues from the quaternary structure of IL-17A, with
each Fab contacting both monomers in the dimer. All complementarity-
determining regions (CDRs) in the Fab contribute to a total of 16 amino acid
residues in the antibody paratope. In vitro affinity optimization was used to
generate CAT-2200 from a parental lead antibody using random muta-
genesis of CDR3 loops. This resulted in seven amino acid changes (three in
VL-CDR3 and four in VH-CDR3) and gave an approximate 30-fold increase
in potency in a cell-based neutralization assay. Two of the seven amino acids
form part of the CAT-2200 paratope. The observed interaction site between
CAT-2200 and IL-17A is consistent with data from hydrogen/deuterium
exchange mass spectrometry and mutagenesis approaches.
© 2009 Elsevier Ltd. All rights reserved.
Edited by I. Wilson Keywords: interleukin-17; cytokine; antibody; epitope; crystal structure
*Corresponding author. E-mail address: mark.abbott@astrazeneca.com.
Abbreviations used: CDR, complementarity-determining region; scFv, single-chain variable fragment; V
H
, variable
heavy; V
L
, variable light; SPR, surface plasmon resonance; HTRF, homogeneous time-resolved fluorescence; PDB, Protein
Data Bank; H/D, hydrogen/deuterium; PBS, phosphate-buffered saline; FRET, fluorescence resonance energy transfer;
DMEM, Dulbecco's modified Eagle's medium; MEM, minimal essential medium; NEAA, nonessential amino acids; MR,
molecular replacement.
doi:10.1016/j.jmb.2009.10.008 J. Mol. Biol. (2009) 394, 905–921
A
vailable online at www.sciencedirect.com
0022-2836/$ - see front matter © 2009 Elsevier Ltd. All rights reserved.
Introduction
IL-17A is one of six known members of the IL-17
cytokine family (IL-17A-F).
1
It is a secreted homo-
dimeric glycoprotein with a molecular mass of
∼35 kDa.
2
IL-17 family members play important
and distinct roles in adaptive immune responses.
They mediate their effects through the IL-17 receptor
family, of which there are five related members (IL-
17RA–IL-17RE; reviewed by Shen and Gaffen
3
and
Gaffen
4
). Both IL-17A and IL-17F can bind to either
IL-17RA or IL-17RC, and it has been proposed that
these colocalize at the cell surface and function as
heterodimeric receptors.
5
It has also been shown
that IL-17A and IL-17-F can form functional hetero-
dimers in human T-cells and can induce neutro-
philia in a murine adoptive transfer model.
6,7
Recent studies have identified Th17 cells as a
unique and distinct CD4
+
T-cell lineage that is
defined by the production of IL-17A, IL-17F, IL-6,
tumor necrosis factor, granulocyte–macrophage
colony-stimulating factor, IL-21, IL-22, and IL-26
(reviewed by Shen and Gaffen
3
and Bettelli et al.
8
).
Th17 cells are believed to have evolved as an arm of
the adaptive immune system and have a critical role
in maintaining inflammatory responses, a role
previously ascribed to Th1 cells. Th17 cells are
therefore emerging as strong candidates for drivers
of autoimmune disease.
9
IL-17A is not widely expressed in humans and is
only found at very low concentrations, specifically
in areas populated by Th17 cells. Interestingly, IL-
17A is expressed in disease compartments in a range
of autoimmune diseases (reviewed by Witowski
et al.
10
) such as rheumatoid arthritis,
11–13
multiple
sclerosis,
14,15
psoriasis,
16
and inflammatory bowel
disease.
17
In vivo studies have shown that IL-17A
has a distinct and critical role in driving both the
early initiation phase and the late progression phase
of disease in a number of preclinical models of
rheumatoid arthritis.
18
Given these recent findings, it is not surprising
that Th17 cells and members of the IL-17/IL-17
receptor family have become the focus of intense
investigation and have been viewed as potential
targets for therapeutic intervention. One group has
recently developed an anti-IL-17 antibody that is
currently in early clinical studies.
19
The reported crystal structure of IL-17F (which
has a 50% sequence identity to IL-17A) presents a
disulfide-linked homodimeric glycoprotein that
adopts a classical cystine-knot fold found in the
transforming growth factor β, bone morphogenetic
protein, and nerve growth factor superfamilies;
however, it lacks the classical disulfide bond
responsible for the canonical knot
20
and instead
has two serines replacing the cysteine residues. All
members of the IL-17 family lack the cysteine
residues required to form the knot, but instead
have conserved serines.
IL-17A is the most intensively studied member of
the IL-17 cytokine family, yet no experimentally
determined structure has been published to date.
Here, we describe the generation of CAT-2200, a
potent, fully human neutralizing monoclonal anti-
body to IL-17A, and reveal the crystal structure of
IL-17A in complex with a Fab fragment of this
antibody. This reveals the definitive epitope and
paratope of the antibody–antigen complex, fully
satisfying the experimental intention. It is interesting
to examine the structural context of the mutations
that result in the improved potency of the CAT-2200
antibody in relation to the parental clone and to
speculate which parts of the IL-17A structure might
be involved in receptor binding.
Results
Isolation of the anti-IL-17A antibody CAT-2200
IL-17A binding antibodies were isolated from a
large phage library displaying human single-chain
variable fragments (scFv)
21
by panning selections on
recombinant human IL-17A. A panel of scFv
isolated from these selections was identified by
their ability to neutralize the binding of recombinant
IL-17A to purified IL-17RA·Fc fusion (receptor–
ligand binding assay), with IC
50
values ranging
from 4 nM to N1000 nM (data not shown). These
scFv were reformatted as full-length IgG1 mole-
cules and tested for neutralization of human IL-17A
in a functional cell assay measuring the release of
IL-6 from HT1080 cells in response to IL-17A. The
most potent lead antibody identified from the
cytokine release assay, TINA12, neutralized the
activity of IL-17A with an IC
50
of 23 nM.
TINA12 was optimized for affinity by a random-
ized mutagenesis of the variable heavy (V
H
) and the
variable light (V
L
) complementarity-determining
region (CDR) 3. V
H
CDR3 and V
L
CDR3 were
mutated separately to generate a number of li-
braries. scFv phage libraries containing CDR3
variants of the lead antibody were subjected to
multiple rounds of affinity-based solution-phase
phage display selections. A panel of optimized
scFv was isolated from these selections through
their improved ability to neutralize the binding of
IL-17A to IL-17RA relative to the parental TINA12
antibody. The optimization process identified scFv
antibodies with IC
50
values of 0.6–12 nM in the
receptor–ligand binding assay (data not shown).
These optimized scFv were reformatted as IgG1 (λ
light chain) and tested for neutralization of human
IL-17A on HT1080/IL-6 release assay. The V
H
and
V
L
chains from several of the most potent antibodies
were recombined, and the most potent recombined
antibody was then reverted by mutagenesis to the
closest human germline sequence (genes VH3-23
and VL6-6a) in the VBASE database,
22
generating
the anti-IL-17A antibody CAT-2200. Any frame-
work residue that was reverted back to germline
was assayed to ensure that it did not affect antibody
affinity. CAT-2200 neutralized IL-17A with an IC
50
of 0.8 nM in the HT1080/IL-6 assay, representing a
906 IL-17A in Complex with a Neutralizing Antibody
∼30-fold improvement over TINA12. The seven
CDR3 amino acid changes (three in V
L
and four in
V
H
) that result in the increased potency of CAT-2200
compared that of the TINA12 parental clone are
shown in Table 1.
The binding affinity of both TINA12 and CAT-
2200 for IL-17A was measured using surface
plasmon resonance (SPR). The increase in apparent
affinity upon optimization of TINA12 to CAT-2200
was approximately 6-fold for the IgG1 molecule,
with almost all of the improvements caused by a
decrease in off-rate.
Functional characterization of CAT-2200
The functional activity of CAT-2200 was assessed
against a variety of sources of IL-17A on a bio-
chemical assay and acting on different cell types.
The antibody potently inhibited the binding of IL-
17A to IL-17RA-Fc in homogeneous time-resolved
fluorescence (HTRF) format (Fig. 1a). To assess
antibody functional activity in a disease-relevant
cell system, we investigated the effects of CAT-2200
on recombinant IL-17A-induced IL-8 responses in
human primary chondrocytes. Escherichia-coli-
Fig. 1. IL-17A-induced responses and their inhibition by CAT-2200. (a) Inhibition of FLAG-tagged IL-17A binding to
IL-17RA-Fc using HTRF. Triangles, untagged E.-coli-derived IL-17A; circles, CAT-2200; squares, isotype control antibody.
(b) Recombinant E.-coli-derived IL-17A induced IL-8 release from primary human chondrocytes. Mean± SDfor one donor
(n=3). EC
50
=0.46 nM. (c) CAT-2200-mediated inhibition of E. coli IL-17A (2 nM) induced IL-8 release from primary
human chondrocytes. Mean± SEM for three donors. Mean IC
50
=1.56 nM. (d) CAT-2200-mediated inhibition of T-cell-
derived IL-17A induced IL-6 release from HT1080 cells compared to isotype control. Mean±SEM for experiments using
different T-cell supernatants.
Table 1. A comparison of lead antibody (TINA12) to optimized antibody (CAT-2200)
Light chain CDR3 Heavy chain CDR3 Affinity K
d
(nM) Potency IC
50
(nM)
89 90 91 92 93 94 95 96 97 95 96 97 98 99 100 100a 101 102 IgG IL-17A
(E. coli)IL-17A
(mammalian)
TINA12 Q S Y D D S S V V D L I W G V A G S 13.5 23 ND
CAT-2200 Q TYDPYSVVDLIHGV TRN 2.1 0.8 0.13
The numbering of amino acids is performed in accordance with Kabat et al.
23
Affinities were measured with BIAcore by immobilizing the
antibody and using standard procedures in accordance with the manufacturer's instruction. Potency was determined by the ability to
inhibit the production of IL-6 from HT1080 cells stimulated with recombinant IL-17A.
907IL-17A in Complex with a Neutralizing Antibody
derived IL-17A generated a dose-dependent increase
in IL-8 production from primary human chondro-
cytes with an EC
50
of 0.46 nM. CAT-2200 inhibited
this response with an IC
50
of 1.56 nM (Fig. 1b and c).
In a second system, recombinant IL-17A derived
from a mammalian cell line was shown to induce the
production of IL-6 from HT1080 cells with an EC
50
of approximately 0.3 nM. This effect could be
inhibited by CAT-2200 with an IC
50
of 8 nM when
using 1 nM IL-17A.
The neutralizing activity of CAT-2200 against
native IL-17A derived from human T-cells was
also analyzed (Fig. 1d). T-cells were cultured under
conditions enhancing IL-17A production.
24
Super-
natants contained IL-17A plus other mediators,
including IL-6 and tumor necrosis factor α, which
may synergize with IL-17A. T-cell supernatants
from three donors induced IL-6 release from
HT1080 cells, and the effect of CAT-2200 on IL-6
levels was assessed. The maximum inhibition with
CAT-2200 was 30%, which was maintained at con-
centrations of 8 nM and above. Isotype control IgG
showed no effect on IL-6 levels at this concentration.
Partial inhibition almost certainly reflects the pres-
ence of IL-6 in the T-cell-conditioned medium, as
well as other cytokines that would have been
produced under the conditioned medium of stimu-
lated T-cells. Thus, partial inhibition is likely to
represent that portion that is a result of the IL-17A
component. This suggestion is also supported by the
observation that the potency of CAT-2200 in this
assay is very similar to that in Fig. 1c when recom-
binant IL-17A is used. Thus, CAT-2200 is able to
neutralize the activity of a native T-cell-derived
source of IL-17A.
The cross-reactivity of CAT-2200 to different IL-17
family and species variants was assessed by the
ability of these proteins to inhibit the binding of
antibody to recombinant human IL-17A derived
from the mammalian HEK293/EBNA cell line. The
rank order of the binding of CAT-2200 to different
species variants was human NcynomolgusNcanine,
with no observed binding to murine IL-17A. In
addition, CAT-2200 showed no binding to human
IL-17 family members B–E. Some weak binding
(20% inhibition at 1 μM) to IL-17F was observed at
the highest concentration of IL-17F tested.
In summary, we have isolated an antibody that
can potently neutralize the effects of recombinant
and native human IL-17A on a number of cell
systems. Furthermore, the antibody does not cross-
react with the other IL-17 family members B–E;
however, it does recognize IL-17F, albeit with a low
potency.
Crystal structure of the IL-17A/CAT-2200 complex
Overall complex structure
The structure of IL-17A/CAT-2200 was refined to
2.6 Å resolution with R/R
free
of 21.2%/26.4%. The
asymmetric unit in the crystal contains two complex
molecules, each with two Fab fragments bound to the
IL-17A dimer. Hence, the final model of six molecules
present in the asymmetric unit comprises 2043 amino
acid residues. Of these, 1720 residues are located
within four molecules of the antibody Fab fragment
(heavy chains H, I, J, and K, and light chains L, M, N,
and O). The remaining 323 residues are found in the
two IL-17A homodimers (chains A/B and C/D).
More than 99% of all residues of the complex struc-
ture were found in the most favored and additionally
allowed regions of the Ramachandran plot.
25
Of the
remaining residues, 0.2% fall into the generously
allowed regions and 0.5% fall into the disallowed
regions.
The crystal structure shows that, in the antibody
complex, each IL-17A dimer is sandwiched between
two Fab fragments (Fig. 2), generating two equiva-
lent IL-17A/Fab interaction sites related by the IL-
17A dimer symmetry. The buried surface area per
interface is around 760 Å
2
. To our surprise, in the
complex structure, as illustrated in Fig. 2, the lower
portion of the IL-17A dimer is disordered, indicating
that it is flexible and adopts different orientations
throughout the crystal such that electron density is
averaged out and is not visible. Thus, it is not
possible to build a model for this part of the IL-17A
dimer with the current data. Two polypeptide
segments are affected: 34 or 35 amino acid residues
at the N-termini and 9 or 11 amino acids starting at
residue 100 or 101. The two independent copies
reveal the same disorder, differing in extent by just a
single residue. In the crystal, lattice interactions are
mediated through the Fab molecules only. There is
ample room in the lattice for the entire IL-17A
molecule to be present in a conformation equivalent
to that seen in IL-17F. Hymowitz et al. described IL-
17F as a ‘garment’with a ‘collar,’‘sleeves,’a‘body,’
and a ‘skirt’.
20
In the IL-17A structure presented
here, it is the skirt that is disordered. The epitope
interaction sites are at the collar and sleeves of the
structure.
An overlay of CAT-2200 Fab with Protein Data
Bank (PDB) entry 1AQK, demonstrating that the
CDRs are in canonical conformation (with the
exception of VH-CDR3), is shown in Fig. 3.
Structure of IL-17A and comparison to IL-17F
The structure of IL-17A and structural alignment
with IL-17F are shown in Fig. 4a and b. Within the
IL-17 cytokine family, IL-17A is the closest homo-
logue to IL-17F, with a 50% sequence identity. The
structure of IL-17F was solved by Hymowitz et al.
20
(PDB entry 1JPY), unexpectedly revealing a cystine-
knot fold.
26
The IL-17A dimer can be superposed
onto the IL-17F dimer with a root-mean-square
deviation (r.m.s.d.) of 1.1 Å for 132 C
α
positions. The
sequence identity for the overlaid portion of the
polypeptides is 64%, higher than the overall
sequence identity between the two molecules. This
is not surprising; apparently, the ordered part of the
molecule is the more conserved part. Each of the
protomers of IL-17A present in the asymmetric unit
of the crystal lattice can be superimposed onto each
908 IL-17A in Complex with a Neutralizing Antibody
other using 74–79 α-carbon atoms, giving an r.m.s.d.
of between 0.32 Å and 0.48 Å.
IL-17A has a homodimeric assembly. Each sub-
unit is formed by a set of two pairs of anti-parallel β-
strands (β1/β2 and β3/β4). A short helix from
Asp42 to Arg46 is the only helical feature. The IL-
17A monomer has the same cystine-knot architec-
ture identified in the crystal structure of IL-17F. The
classical cystine knot is formed by the unique
arrangement of six cysteine residues. In the structure
of IL-17A, Cys71 and Cys121, as well as Cys76 and
Cys123, connect β-strands 2 and 4 to form one part
oftheknot.Atruecystineknotrequiresan
additional disulfide to penetrate the ring formed,
but cysteine-to-serine replacements at positions 49
and 89 of the amino acid sequence of IL-17A
Fig. 3. Overlay of CAT-2200 Fab with PDB entry 1AQK. The constant domains are shown in white. PDB entry 1AQK is
shown with the light chain in black and with the heavy chain in red. CAT-2200 is shown with the light chain in blue and
with the heavy chain in green.
Fig. 2. Overall structure of the IL-17A/Fab complex. This and all other molecular illustrations in this work were
prepared using PyMOL (http://www.pymol.org). The IL-17A homodimer is shown with the two molecules of the dimer
in pale and dark yellow. The Fab fragments are shown with the light chain in blue and with the heavy chain in green. The
constant and V
H
and V
L
domains are labeled. The two interaction sites are equivalent, related by IL-17A dimer symmetry.
The lower portion of the IL-17A dimer is not visible on the electron density map and does not form part of this model. The
N-termini and C-termini of monomers A and B are indicated.
909IL-17A in Complex with a Neutralizing Antibody
Fig. 4 (legend on next page)
910 IL-17A in Complex with a Neutralizing Antibody
preclude final knot formation. This is exactly the
same situation seen in IL-17F: Ser49 (IL-17A) adopts
the same rotamer conformation found at Ser50 in the
IL-17F dimer, similarly for Ser89 (IL-17A) and Ser90
(IL-17F).
An additional intermolecular disulfide bond,
corresponding to that between Cys17 and Cys107
in IL-17F, must also be present in the structure of IL-
17A (between Cys10 and Cys106), as the recombi-
nant protein migrated with a molecular mass of
17 kDa and 35 kDa when analyzed by SDS-PAGE
under reducing and nonreducing conditions, respec-
tively. However, this disulfide is not visible, as it
occurs in the disordered part of the IL-17A structure.
The IL-17A/CAT-2200 interface
The crystal structure allows epitope interactions
between IL-17A and CAT-2200 Fab to be examined
in atomic detail. These are shown in Fig. 5 and
captured in Table 2. It is only necessary to describe
one of the two interaction sites, since they are
equivalent. Epitope–paratope interactions involve
all CDRs from both heavy and light chains, and
amino acid residues from both monomers of the IL-
17A dimer. The heavy chain interacts with both
chains A and B in the IL-17A dimer, while the light
chain interacts only with monomer B.
Twelve amino acids from IL17A form the epitope
that interacts with 16 amino acid residues in the
antibody paratope. The interactions include nine
hydrogen bonds and nonpolar van der Waals inter-
actions. The amino acid residues in IL-17A that form
part of the epitope are Ser40-Tyr43 (inclusive);
Arg46 in chain A; and Leu74, Pro91, Tyr85-Asn88,
and Pro126-Ile127 in chain B of the dimer. Residues
contributed from the light chain are Ala29-Tyr32
(inclusive; from CDR1), Phe49 (FW2), Gln53
(CDR2), and Tyr91 and Pro93 (CDR3). Residues
from the heavy chain are Thr28 (FW1), Ser30 (FW1)
to Tyr32 (CDR1), Tyr58 (CDR2), and Leu96-His98
(CDR3).
CAT-2200 binds to IL-17F with an affinity that is
approximately 3 orders of magnitude weaker than
that for IL-17A. Of the 12 residues in IL-17A that
form the epitope, five are different in IL-17F. These
differences are shown in red and in brackets in
Fig. 5. Several of the changes are nonconservative
Fig. 5. Wall-eyed stereo representation of the IL-17A epitope recognized by CAT-2200. Residues from the IL-17A
homodimer are shown in pale or dark yellow. Residues from the Fab fragments are shown with the light chain in blue and
with the heavy chain in green. Labelled residues allow the interactions listed in Table 2 to be readily located. Residues in
brackets are the equivalent residue in IL-17F. When no residue in brackets is indicated, then it is identical between IL-17A
and IL-17F.
Fig. 4. (a–g) have the same orientation. (a) Wall-eyed stereo figure of the IL-17A homodimer, with disulfide bonds
shown in cyan. The β-strands are labeled for one of the monomers. The N-termini and C-termini of the ordered portion of
the IL-17A model are indicated for the other monomer. (b) Structural overlay with IL-17F (shown in green; IL-17A is shown
in two shades of yellow, as described previously). This clarifies the extent of the missing disordered portion of the IL-17A
dimer. The N-termini and C-termini here are the termini of IL-17F. (c) A surface representation of the IL-17F homodimer
structure showing residues Asn35-Met40 in red. This peptide has been described as the “right-hand wall”of the cavity that
is visible just to the left of Arg37, and the cavity is suggested to be a receptor binding pocket.
20
(d) Surface representation of
the IL-17A homodimer. The peptide Asn36-Tyr44 is shown in red: unlike IL-17F, no cavity is seen in IL-17A, may be
because the antibody induces a conformation change in the peptide, which is also shown in red in (a) and (b); it occurs at the
N-terminus of the ordered portion of the molecule, suggesting perhaps that the disorder might even be a consequence of
antibody binding. (e) Modeling of the IL-17F conformation of Asn36-Tyr44 (highlighted in red) into the IL-17A structure,
demonstrating that a cavity appears (although there are steric clashes with the antibody). (f) A superposition of IL-17A as
ball-and-stick with a surface representation of IL-17F. Epitope residues are shown in red. This shows that the epitope
region would be immediately adjacent to the IL-17F cavity if it were preserved in IL-17A. (g) A close-up of the cavity region
from (f) demonstrating how equivalent residues fill the cavity.
911IL-17A in Complex with a Neutralizing Antibody
and therefore provide a potential rationale for the
much weaker binding to IL-17F.
The crystal structure contains the unglycosylated
cytokine, raising the question of whether glyco-
sylation might affect the antibody binding mode
revealed. There is a single N-linked glycosylation
site in IL-17A at Asn45. Although this is adjacent to
Arg46 (which is part of the epitope) and hence close
to the antibody binding site, the side chain of Asn45
is oriented towards the solvent and away from the
bound antibody, and glycosylation is unlikely to
prevent antibody binding in the manner shown or to
contribute to antibody binding. In addition, there is
no significant difference in antibody recognition
between E.-coli-derived IL-17A (unglycosylated)
and mammalian-cell-derived IL-17A (partially gly-
cosylated; data not shown).
Structural context of sequence differences
between TINA12 and CAT-2200
A total of seven amino acid changes were identi-
fied in CDR3 between the optimized antibody CAT-
2200 and the parent TINA12. The crystal structure of
the IL-17A/Fab complex allows us to examine the
structural context of these changes and to speculate
on how they might have improved affinity. Two of
the changes, D93P (in VL-CDR3) and W98H (in VH-
CDR3), are part of the paratope (Fig. 6). Pro93 in VL-
CDR3 forms a stacking interaction with Tyr85 of IL-
17A, probably providing a significant improvement
in interface surface complementarity. Interestingly,
the D93H mutation also appeared in other opti-
mized constructs, consistent with the notion that
side-chain stacking may be beneficial. The D93P
mutation would restrict the psi main-chain torsion
angle for the proline residue to ∼− 60°, which would
help rigidify CDR3, although comparison is difficult
without the structural details of TINA12. The second
mutation in the paratope is W98H. In the structure,
this residue has a main-chain amide hydrogen bond
to the side chain of Asn88 in IL-17A. Since it is a
main-chain hydrogen bond, it is also presumably
present in TINA12.
Fig. 6. Wall-eyed stereo representation of the CAT-2200 paratope. Residues from the IL-17A homodimer are shown in
pale or dark yellow. Residues from the Fab fragments are shown with the light chain in blue and with the heavy chain in
green. The seven amino acids that are different between CAT-2200 and the lead antibody TINA-12 have been labeled.
Table 2. IL-17A/Fab direct interactions
IL-17A Fab Distance (Å)
Hydrogen bonds
Ser A40 O Ser H30 OG 2.9
Asp A42 N Ser H31 OG 3.2
Asp A42 OD1 Thr H28 OG 2.9
Asp A42 OD1 Tyr H32 OH 2.7
Arg A46 NH2 Leu H96 O 2.6
Arg A46 NH1 Ser H31 O 2.8
Tyr B85 O Tyr L91 OH 2.5
His B86 NE2 Ala L29 O 3.0
Asn B88 OD1 His H98 N 2.8
Nonpolar (distance corresponds to the closest atom pair)
Ser A40 Thr H28 3.5
Ser A41 Thr H28 3.7
Tyr A43 Ser H31 3.6
Ser A41 Ser H31 3.4
Arg A46 Tyr H32 3.4
His B86 Asn L30 3.5
His B86 Tyr L31 3.8
Tyr B85 Tyr L31 3.7
His B86 Tyr L32 3.4
Pro B126 Tyr L32 3.3
Pro B126 Phe L49 3.4
Ile B127 Gln L53 3.2
Tyr B85 Pro L93 3.7
Leu B74 Pro L93 3.6
Leu B74 Tyr H58 3.6
Pro B126 Leu H96 3.6
Asn B88 Ile H97 3.4
Met B87 Ile H97 3.8
His B86 Ile H97 3.8
Leu B74 His H98 3.9
The residue number contains a chain indicator (H: Fab heavy
chain; L: Fab light chain; A: monomer A in IL-17A; B: monomer B
in IL-17A). The distance cutoff used for hydrogen bonds is 3.2 Å,
and that for nonpolar interactions is 4.0 Å.
912 IL-17A in Complex with a Neutralizing Antibody
The other five substitutions occur away from the
interface, and any effect on the paratope would have
to be indirect via either a stabilizing effect on V
H
–V
L
interactions or intrachain contacts. Two of these
mutations (S90T and S94Y) form a hydrogen bond
between CDR3 and the backbone amide of Tyr32
(VL-CDR1) and Tyr59 (VH-CDR2), respectively,
which may help stabilize loop conformations that
affect the shape of the paratope. The effects of these
distant changes are unlikely to improve binding
directly, although they may improve the stability of
the antibody fragment, thereby lowering binding
energy and indirectly promoting antigen binding.
The three remaining amino acid substitutions in
the C-terminal part of VH-CDR3 are A100aT,
G101R, and S102N. They are located away from
the paratope, but close to V
L
residues Tyr36 and
Ile46. Thr100a forms a hydrogen bond with V
L
Tyr36 that is not possible in TINA12. Perhaps these
mutations help stabilize V
H
–V
L
contacts in the
antibody, leading to improved binding.
Analysis of IL-17A/CAT-2200 interaction via
hydrogen/deuterium exchange and mutagenesis
Two additional approaches were taken in order to
analyze the interaction between IL-17A and CAT-
2200: hydrogen/deuterium (H/D) exchange coupled
to mass spectrometry and mutagenesis.
The principles and methods behind epitope
mapping by H/D exchange mass spectrometry are
that the rate of exchange of the antigen is measured
in the presence and in the absence of antibody and
the two are compared.
27
Following exchange, the
protein is digested with pepsin, and those peptides
that exchange more slowly in the presence of
antibody are the H/D-exchange-defined epitope.
Figure 7 shows a map of the rate of exchange across
IL-17A in the absence of antibody. Of particular note
is that the first 42 amino acids of IL-17A exchange
almost completely even at the shortest time of 15 s,
suggesting that this part of the protein is extremely
dynamic. The averages of the differences in the rates
of exchange in the absence and in the presence of
antibody are shown in Table 3. Two peptides whose
exchange properties altered significantly upon
binding to CAT-2200 (residues 45–53 and 71–87)
were identified. A third peptide encompassing
residues 119–132, whose exchange properties were
slightly altered, was identified. When comparing
these regions to the crystal structure, it is clear why
the exchange is altered when the IL-17A is bound to
CAT-2200. Peptide 45–53 contains Arg46, which
makes several interactions with the heavy chain.
Peptide 71–87 contains Tyr85 and His86, both of
which form multiple interactions with the light
Table 3. Difference in the H/D exchange rates of IL-17A
in the presence and in the absence of CAT-2200
Start End Difference (%)
−19 −15 2
−12 −52
−223 3
26 42 1
45 53 11
56 68 1
71 87 18
90 97 2
100 110 3
113 116 1
119 132 6
IL-17A was deuterated and exchanged back to hydrogen in the
presence and in the absence of antibody, as described in Materials
and Methods. The average difference in the deuteration levels of
different peptic fragments of the protein was determined by
digestion with pepsin and analysis by mass spectrometry.
Residue 1 is Gly24, as described in SwissProt accession number
Q16552. Residues with negative numbers are in Avi tag.
Fig. 7. H/D exchange pattern of IL-17A. Each horizontal color block represents an analyzed peptic peptide, and each
block contains a number of time points. The N-terminal 21 residues constitute an Avi tag sequence. Numbering starts at
the first IL-17A residue. Deuterium build-up patterns of IL-17A in solution (from top: 15 s, 50 s, 150 s, 500 s, and 1500 s).
The deuteration level of each peptide at each time point is color coded (see bottom right).
913IL-17A in Complex with a Neutralizing Antibody
chain, and also Leu74 and Met87, which are
involved in the binding interface. Peptide 119–132
contains Pro126 and Ile127, which interact with the
light chain. In summary, the data from H/D
exchange mass spectrometry are consistent with
the crystal structure and demonstrate the value of
this approach.
A mutant of IL-17A was made in order to further
understand the interaction with CAT-2200. This
mutant took advantage of the observation that CAT-
2200 binds to human IL-17A with high affinity, but
does not bind to murine IL-17A (Table 1). The
mutant IL-17A was designed from the H/D
exchange experiments covering the region 71–89,
where perturbation in H/D exchange was most
pronounced. Every amino acid in that region that
varied between the human sequence and the mouse
sequence was changed from human residue to
murine residue (L74Q, G75R, I77V, D80E, N82K,
V83L, and Y85H).
The mutant and wild-type proteins were ex-
pressed in mammalian cells (HEK293/EBNA), puri-
fied, and analyzed for their binding to CAT-2200 by
SPR. The functional ability of the mutants to induce
IL-6 production from HT1080 cells, in addition to the
effects of CAT-2200 on neutralizing this response,
was also assessed. The mutant protein was bio-
logically active in the HT1080 assay, with a slight
drop in potency (4 nM versus 0.3 nM). CAT-2200 was
unable to inhibit the activity of the mutant IL-17A at
concentrations up to 50 nM. Furthermore, this
mutant IL-17A showed no binding to CAT-2200 by
SPR. It can be concluded that the region between 71
and 89 is crucial for the interaction with CAT-2200.
This is consistent with the data from X-ray crystallo-
graphy and H/D exchange.
Discussion
IL-17A is the most studied member of the IL-17
family, but its structure has remained elusive
despite the publication of the structure of its closest
relative, IL-17F.
20
We report here the partial struc-
ture of IL-17A in complex with a neutralizing anti-
body. The crystal structure reveals approximately
two-thirds of the cytokine that has a structure very
similar to that of IL-17F. The remaining third of the
molecule (the ‘skirt’) is disordered (Fig. 4b). Our
observation is in contrast to IL-17F, where the whole
structure is resolved. However, the B-factors of the
“skirt”are high in IL-17F, suggesting that it may
also have inherent flexibility. Furthermore, in the IL-
17F structure, some crystal lattice contacts are made
through the skirt region, and it is possible that these
would stabilize what might otherwise be a flexible
portion of IL-17F. In the IL-17A crystal, lattice
contacts are mediated solely through the Fab
fragments. Although a key cysteine involved in
intermolecular disulfide bonding is conserved, the
overall level of homology across the N-terminal 34
residues that are disordered in our structure is quite
low at approximately 18%, and it would not be
unusual to see structural differences between IL-17A
and IL-17F in this region. In our crystal lattice, there
is space for IL-17A to adopt a conformation
equivalent to that adopted by IL-17F without any
clashes, but this is not observed. Thus, it seems
reasonable to suggest that IL-17A has an inherent
flexibility in a part of its structure. This could make
crystallization difficult in the absence of antibody
and might be an explanation for why no structure
has been previously published on such a well-
studied cytokine. The hypothesis of inherent flexi-
bility is supported by H/D exchange data in the
absence of antibody. In short, it is clear that IL-17A
and IL-17F share similar structures across approxi-
mately two-thirds of the protein. We do not know
whether the other third that is flexible in our IL-
17A/Fab complex is a distortion of the IL-17F
conformation as a result of antibody-induced
conformational change, or whether it is representa-
tive also of the unbound cytokine. Similarly, in the
absence of a complete structural homolog, we are
unable to suggest with certainty that the IL17F
structure is representative of the family across this
third of the molecule.
In the central region of the molecule, the IL-17F
structure reveals a cavity that Hymowitz et al.
speculated to be a receptor interaction site.
20
All the
residues speculated to form the cavity in IL-17F are
observed in our partial IL-17A structure, yet no such
cavity is seen in the IL-17A/Fab complex structure
presented here (Fig. 4d). Despite this observation, it is
quite feasible that IL-17A has a receptor-interacting
cavity similar to that proposed for IL-17F, and that
CAT-2200 binding distorts the cytokine structure to
occupy the cavity. The epitope residues of IL-17A
that are close to the cavity are shown in red in Fig. 4f,
superimposed on a surface representation of IL-17F.
This shows that the antibody binding site is
immediately adjacent to the cavity site in IL-17F; for
clarification, there is no cavity in the IL-17A/Fab
complex structure. The conformation of residues
Asn36-Tyr44 in this central region (the “right wall”of
the cavity in Hymowitz et al.
20
) is different between
IL-17A and IL-17F (Fig. 4c and d): modeling of the IL-
17F conformation into the IL-17A structure recreates
the cavity (Fig. 4e), but creates a steric clash with the
antibody. Thus, it is possible that a cavity exists in the
unliganded IL-17A structure similarly to IL-17F, and
that the antibody binds on the edge of this extant
cavity and distorts the “right wall”such that it
repositions to fill the cavity (Fig. 4g). This would
imply that antibody binding inhibits receptor inter-
action by inducing a conformational change that
eliminates the receptor binding pocket, possibly in
addition to steric factors limiting receptor access due
to the size of the antibody. The induced conforma-
tional change may well be the origin of the observed
disorder, and the IL-17A may resemble IL-17F in
conformation in the absence of antibody. Equally,we
must consider the alternative possibility that no
equivalent cavity exists in IL-17A and that the
structure shows the antibody binding to the native
conformation (which is, in part, inherently flexible
914 IL-17A in Complex with a Neutralizing Antibody
and disordered), blocking binding to the receptor at a
different part of the cytokine. Either way, the effect is
unlikely to be a crystal artifact, since it is observed in
both independent copies of the complex in the
asymmetric unit. It would be of interest to examine
a crystal structure of IL-17A in isolation to see
whether the cavity is present in the absence of
antibody, although this experiment might be difficult
if IL-17A is truly flexible.
There is currently much discussion about the exact
nature of the receptor complex for IL-17A and IL-17F
as homodimers or heterodimers (reviewed recently
by Gaffen
4
). In humans, IL-17RA binds IL-17A much
more tightly than IL-17F, whereas IL-17RC binds
both IL-17A and IL-17F with high affinity, although
the situation is further complicated by the presence
of different splice variants of IL-17 RC with differing
affinities.
28
The most likely scenario is that the
receptor is an obligatory multimeric complex of IL-
17RA, IL-17RC, and IL-17A, implying that the
mechanism by which the two receptors interact
with IL-17A is different. Whether there is a receptor
binding cavity on IL-17A and, if so, how binding to
the two receptor subunits is achieved will require
further studies and, ultimately, a crystal structure of
the receptors in complex with the ligand.
The in vitro affinity maturation process used in
this study led to a modest increase in affinity and
potency, and was the result of seven amino acid
changes in the CDR3 loops. The contribution of each
specific change was not determined experimentally.
Previous studies have shown that residues remote
from the binding site, as well as those at the
periphery of the binding pocket and in the middle
of the VH-CDR3 loop, can all influence affinity.
29–32
Two of the seven changes in this study involved
amino acids at the interface (D93P and W98H);
however, among these, only D93P has a clear shape-
matching consequence. Pro93 may also help rigidify
the VL-CDR3 loop conformation. The other five
changes can only affect the paratope indirectly,
perhaps by improving antibody stability and loop
conformations, providing a more suitable scaffold
for IL17A binding.
The in vitro optimization of the anti-IL-17A
antibody TINA12 described here focused solely on
saturation mutagenesis of the CDR3 loops. This
strategy was chosen as CDR3 loops were thought to
be at the center of the antigen binding pocket and to
contribute most to antibody paratope structures.
The structure of the IL-17A/CAT-2200 complex
revealed that all six CDR loops are involved in the
binding of IL-17A, with most binding contacts being
contributed by VH-CDR1 and VH-CDR3. It could
therefore be possible to achieve even greater affinity
gains by targeting more than just the CDR3 loops. It
remains difficult to predict the most appropriate
target residues of a given antibody for mutagenesis
that will result in higher affinity and/or potency,
although improved computational methods lead to
more rational approaches to affinity maturation.
33
In this study, the three approaches taken to map
the epitope of CAT-2200 gave consistent results.
Crystallography would generally be the method of
choice, but it is not always possible to crystallize a
complex of antigen with a Fab fragment, thus
necessitating other approaches. The discontinuous
nature of the epitope reported in this study made
approaches based on the binding of the antibody to
short constrained peptides unfeasible. We did
attempt to map the epitope by these peptide-based
approaches but, unsurprisingly, did not identify any
of the regions shown by other methods to interact
with CAT-2200 (data not shown). Methods that rely
on an analysis of the interaction with a fully folded
intact tertiary—and, if appropriate, quaternary—
structure are clearly preferable. H/D exchange,
coupled to targeted mutagenesis, is shown to be a
useful approach to doing this, although clearly
without the resolution of a crystal structure. The
data on mutant IL-17A generated in this study are
also strongly suggestive that although there are
interactions between multiple amino acids in IL-17A
and multiple amino acids in all CDRs of the
antibody, there are likely to be a much smaller
number of residues that are key to high affinity.
Tyr85 in IL-17A seems to be a likely candidate as one
of these residues.
In summary, we have described here the structure
of IL-17A in complex with an inhibitory antibody.
Our data show a β-sheet structure in a cystine-knot
fold (similar to IL-17F) and are suggestive that IL-
17A may be partially flexible. It is possible that the
antibody inhibits the cytokine by distorting a
receptor binding cavity, although this hypothesis
and the molecular details on how the cytokine
interacts with IL-17 receptors require further signifi-
cant work. The optimization of TINA12 to the lead
antibody CAT-2200 not only affected epitope–
paratope interactions but also selected for distant
amino acid residues that may improve antibody and
loop stability and conformations that encourage IL-
17A/CAT-2200 interactions. This antibody is potent
in relevant cellular neutralization assays and is
suitable for further clinical development for a
variety of inflammatory conditions.
Materials and Methods
Expression and purification of IL-17A
Human IL-17A without the signal sequence and starting
at Gly24 (SwissProt accession number Q16552) was cloned
into the vector pT7#3.3
34
for expression in E. coli.
Numbering throughout the article is performed with this
glycine as residue 1, as this is the first residue after the
removal of the signal sequence. Recombinant human IL-
17A for crystallization and H/D exchange studies was
expressed in E. coli, and inclusion bodies were isolated,
using well-described methods.
35
The protein used for
crystallization had no added tags, whereas the protein
used in the H/D exchange experiments had an N-terminal
Avi tag. Inclusion body protein was solubilized by
homogenization into a solubilization buffer [50 mM Tris
(pH 8.5), 6 M guanidine HCl, and 10 mM DTT] at room
temperature for 1 h. Solubilized protein was refolded at a
915IL-17A in Complex with a Neutralizing Antibody
final concentration of 0.5 mg/ml by rapid dilution into a
refold buffer with vigorous stirring [0.1 M CAPSO
(pH 9.5), 0.9 M arginine, and 0.3:0.03 mM reduced/
oxidized glutathione] at room temperature and left to
stand overnight. Refolded protein was concentrated 5-fold
using a 10 kDa molecular weight cutoff spiral cartridge
(Amicon proflux™M12 Tangential flow system). Con-
centrated protein was purified by size-exclusion chroma-
tography using S75 (GE Healthcare Biosciences) columns
equilibrated and developed with 50 mM Tris (pH 7.4) and
0.15 M NaCl.
Mutant IL-17A protein was cloned with a C-terminal
histidine tag using Gateway© technology (Invitrogen) and
conventional mutagenesis methods (using Stratagene
multisite mutagenesis kit) or polymerase chain reaction.
The cloned gene was inserted into the expression vector
pCEP4 using the LR Clonase reaction.
Wild-type and mutant IL-17A with C-terminal His tags
were expressed in HEK293/EBNA cells. The cells were
transfected with 30 μg of DNA using polyethyleneimine
transfection reagent (Sigma), and the cells were incubated
overnight. The media were removed, replaced with fresh
serum-free media, and incubated for a further 72 h. The
supernatants from these cells were collected and used for
the purification of each recombinant protein.
Supernatants from the transfected HEK293/EBNA cells
were collected, and the pH was adjusted to 7.4 prior to
purification by affinity chromatography on Ni-Sepharose
Fast Flow (GE Healthcare). The purified protein was
dialysed into phosphate-buffered saline (PBS) and ana-
lyzed by SDS-PAGE, Edman N-terminal sequencing, and
Western blot analysis using anti-histidine antibodies as
detection tools. The concentration was determined by
absorbance at 280 nm.
Isolation of anti-IL-17A antibody
Large scFv human antibody libraries cloned into a
phagemid vector, based on filamentous phage M13, were
used for selections.
21,36,37
Anti-IL-17A-specific scFv anti-
bodies were isolated from phage display libraries using a
series of selection cycles on recombinant E.-coli-expressed
human IL-17A, essentially as previously described.
36
IL-17
neutralizing antibodies were identified from selections by
screening individual scFv expressed from E. coli for
inhibition of recombinant IL-17A binding to purified IL-
17RFc. Neutralizing scFv with unique sequences were
then expressed in E. coli and purified by affinity chroma-
tography.
38
The potency of the purified scFv was then
determined with the receptor–ligand binding assay and
the HT1080/IL-6 cytokine release assay in response to
human IL-17A, as described below.
The most potent lead antibody identified from the
cytokine release assay was then optimized using a
targeted mutagenesis approach and affinity-based phage
display selections. Large scFv phage libraries derived
from the lead clones were created by oligonucleotide-
directed mutagenesis of the V
H
/V
L
chain CDR3 using
standard molecular biology techniques.
39
scFv phage libraries containing variants of the lead
antibody were subjected to multiple rounds of affinity-
based solution-phase phage display selections, in the
presence of decreasing concentrations of human IL-17A at
each selection round (50 nM–50 pM), in order to select
variants with a higher affinity for IL-17A. The selections
were performed essentially as described previously.
40
We identified a panel of optimized scFv isolated from
these selections, which had improved ability to neutralize
the binding of IL-17A to IL-17RA relative to the un-
optimized lead antibody. These optimized scFv were
converted into IgG1 format, as described below, and then
tested for neutralization of human IL-17A in the HT1080/
IL-6 release assay. The V
H
and V
L
chains from several of
the most potent antibodies were recombined, and the most
potent recombined antibody was reverted by mutagenesis
to the closest human germline sequence in the VBASE
database,
22
generating the anti-IL-17A antibody CAT-
2200. The antibody has been numbered in accordance with
Kabat et al.
23
Clones were converted from scFv into IgG format by
subcloning the V
H
and V
L
domains into vectors expressing
whole antibody heavy and light chains, respectively. The
V
H
domain was cloned into a vector (pEU15.1) containing
the human heavy chain constant domains and regulatory
elements to express whole IgG heavy chain in mammalian
cells. Similarly, the V
L
domain was cloned into a vector
(pEU4.4) for the expression of the human light chain (λ)
constant domains and into regulatory elements to express
whole IgG light chain in mammalian cells. Vectors for the
expression of heavy and light chains were originally
described by Persic et al.
41
Cambridge Antibody Tech-
nology vectors have been engineered simply by intro-
ducing an OriP element. To obtain IgGs, we transfected
the heavy-chain and light-chain IgG-expressing vectors
into HEK293/EBNA mammalian cells. IgGs were
expressed and secreted into the medium. Harvests were
pooled and filtered prior to purification. The IgG was
purified using Protein A chromatography.
The Fab fragment of CAT-2200 was generated using
papain, then purified, buffer exchanged into PBS (pH 7.2),
and concentrated to approximately 10 mg/ml for use in
crystallization experiments.
Receptor–ligand binding assay
IL-17A neutralizing antibodies were identified from
selection outputs by screening E.-coli-expressed scFv in a
single-point HTRF® fluorescence resonance energy trans-
fer (FRET) assay (CIS Bio International) in which the
binding of recombinant human IL-17A to purified IL-
17RA Fc fusion (R&D Systems) was measured.
scFv were incubated with 1.5 nM biotinylated recom-
binant human IL-17A and 3 nM IL-17RA Fc fusion protein
for 5 h at room temperature. Receptor–ligand binding was
detected using 0.6 nM streptavidin cryptate donor
fluorophore and 10 nM anti-human Fc XL665 conjugated
acceptor fluorophore. FRET signal was measured by
reading time-resolved fluorescence emissions at wave-
lengths of 620 nm and 665 nm using an EnVision plate
reader (Perkin-Elmer). Data were analyzed by calculating
the percent ΔFvalue for each sample, which was then
used to determine percent specific binding.
For IC
50
value determination, purified scFv were
titrated into the assay 3-fold (in duplicate) over 11 points,
and data were curve fitted by nonlinear regression
analysis using GraphPad Prism software.
CAT-2200 cross-reactivity assay
The cross-reactivity of CAT-2200 to different IL-17
species or family members was assessed using an HTRF®
FRET assay (CIS Bio International), in which the binding of
CAT-2200 to tagged human IL-17A was measured.
Different IL-17 species or family variants that could
bind CAT-2200, and therefore compete with the tagged
human IL-17A for binding the antibody, were identified as
916 IL-17A in Complex with a Neutralizing Antibody
giving concentration-dependent inhibition in the assay
from which IC
50
values were calculated.
A titration series of each IL-17 species or family member
(either generated in-house or purchased from R&D
Systems and Peprotech) was incubated with 4 nM
mammalian-expressed human IL-17A C-terminally
tagged with FLAG and histidine, and 0.6 nM CAT-2200
antibody for 2 h at room temperature. Binding of the
tagged human IL-17A to CAT-2200 was detected using
1.6 nM anti-FLAG tag cryptate-labeled antibody (CIS Bio
International) and 20 nM anti-human Fc antibody
conjugated to the acceptor fluorophore XL665 (CIS Bio
International).
FRET assay signal measurement and data analysis were
carried out as described for the receptor–ligand binding
assay.
Biological assays
Assessment of recombinant IL-17A and CAT-2200 in
primary chondrocytes
Cartilage was obtained (with full ethical consent) after
total knee replacement in osteoarthritis patients. Cartilage
was digested overnight in 2 mg/ml collagenase (Sigma) in
chondrocyte media [Dulbecco's modified Eagle's medium
(DMEM), L-glutamine, 10% fetal bovine serum, and 1%
minimal essential medium (MEM) nonessential amino
acids (NEAA)] at 37 °C. Cell digest was strained to remove
undigested cartilage, and chondrocytes were isolated
through centrifugation. Chondrocytes were expanded in
culture flasks containing chondrocyte media supple-
mented with amphotericin B (2.5 μg/ml; Sigma).
Chondrocytes were used at passage 1. One donor was
used for generation of EC
50
(data points in triplicate), and
the mean of three donors (data points in triplicate) was
used to generate an IC
50
value. Chondrocytes were
harvested and seeded at 1.5× 10
4
cells/well overnight in
chondrocyte media. For EC
50
determination, recombinant
IL-17A was serially diluted to a final concentration, and
media were removed from cells and replaced by 100 μlof
stimulation media prior to incubation overnight (37 °C).
For IC
50
determination, CAT-2200 was serially diluted in
chondrocyte media to 2× final concentration, and E.-coli-
derived human IL-17A was diluted in chondrocyte media
to 2× final concentration (final concentration, 2 nM).
Antibody/IL-17A solutions were incubated for 1 h (37 °C).
Media were removed from chondrocytes and replaced by
100 μl of antibody/recombinant IL-17A mix prior to
incubation overnight (37 °C).
Cell culture supernatants were collected and analyzed
for IL-8 by ELISA (Cytoset; Biosource), in accordance with
the manufacturer's instructions, using 3,3′-5,5′-terta-
methylbenzidine detection. Data were generated from
standard curves to give total IL-8 per well. Data were
plotted in Originv7.5 software (OriginLab Corp.) using
picograms per milliliter of IL-8 response against log
10
(IL-
17A) or the percentage of maximum IL-8 response against
log
10
(antibody), generating EC
50
and IC
50
values,
respectively.
Assessment of recombinant IL-17A and CAT-2200 in
HT1080 cells
For characterization of recombinant ligands, HT1080
cells (no. 85111505; European Collection of Cell Cultures)
were seeded overnight in 96-well flat-bottomed tissue
culture assay plates at 5× 10
4
cells/well in media (MEM
with Earle's salts, L-glutamine, 10% fetal bovine serum,
and 1% MEM-NEAA). A titration of antibody was per-
formed in PBS and preincubated with relevant IL-17A
(1 nM). Media were removed, and cells were incubated
overnight with the antibody/IL-17A solution. Super-
natants were harvested for ELISA analysis. IL-6 was
assessed using IL-6 Duoset antibody pairs (R&D Systems)
in accordance with the manufacturer's instructions, with
the exception that europium-labeled streptavidin DEL-
FIA® detection (Perkin-Elmer) was used instead of
streptavidin horseradish peroxidase detection. Data were
generated from standard curves in Excel to give the total
IL-6 per well. IC
50
values were determined by plotting the
percent control response against log antibody concentra-
tion using Prism software (GraphPad) and by curve fitting
the data by nonlinear regression analysis.
Assessment of native T-cell-derived IL-17A and
CAT-2200 in HT1080 cells
Peripheral blood mononuclear cells were isolated from
blood obtained from three healthy volunteers using
Lymphoprep (Axis Shield) density gradient separation.
Peripheral blood mononuclear cells were incubated on a
nylon wool column (Polysciences) in media (RPMI
supplemented with 10% fetal calf serum, glutamine,
penicillin, and streptomycin, and 1% MEM-NEAA), and
the effluent, containing T-cells, was collected. T-cells were
stimulated at 5× 10
6
cells/ml (final concentration) in flasks
in the presence of anti-CD3/anti-CD28 (both at a final
concentration of 1 μg/ml; Sigma), transforming growth
factor β(5 ng/ml; R&D Systems), IL-23 (50 ng/ml; R&D
Systems), anti-IFNγ(10 μg/ml; R&D Systems), and goat
anti-mouse IgG cross-linker (4 μg/ml, added 15 min after
other reagents; Sigma). Supernatants were harvested by
centrifugation after 48 h.
HT1080 cells were seeded overnight at 1.5×10
4
cells/
well in media (DMEM supplemented with 10% fetal calf
serum, glutamine, penicillin, and streptomycin, and 1%
MEM-NEAA). Serial dilutions of antibodies were pre-
pared in media at 2× final concentration and mixed in
equal volume with conditioned T-cell media and incu-
bated (at 37 °C for 1 h). Media were removed from the cells
and replaced with antibody solutions, and the cells were
incubated overnight. Supernatants were removed and
analyzed with IL-6 ELISA (Cytoset; Biosource), in
accordance with the manufacturer's instructions, using
3,3′-5,5′-tertamethylbenzidine detection. Data were gen-
erated from standard curves using Origin v7.5 to give the
IL-6 concentration per well. The response for each T-cell
was expressed as the percentage of the maximum level of
IL-6 generated by T-cell supernatant from each donor.
This response represents the contributory effects of all
mediators present in the T-cell supernatant plus any
endogenous T-cell-derived IL-6 present.
Assessment of mutant and wild-type IL-17A in HT1080 cells
HT1080 cells were seeded overnight in 96-well flat-
bottomed tissue culture assay plates at 5 ×10
4
cells/well in
media (DMEM with L-glutamine, 1% penicillin and
streptomycin, 10% heat-inactivated fetal bovine serum,
and 1% MEM-NEAA).
Titrations of antibodies were prepared in assay media
and preincubated with IL-17A or mutant forms of IL-17A
(equating to the previously determined EC
50
; data not
shown) for 1 h at 37 °C. Media were aspirated from the
cells, and antibody/IL-17A solution was incubated over-
night. Supernatants were harvested and assayed for IL-6
using ELISA (Cytoset; Biosource), as described above.
917IL-17A in Complex with a Neutralizing Antibody
H/D exchange analysis
Methods for performing H/D exchange analysis have
been described in detail.
27
Prior to H/D exchange
experiments, the digestion and separation conditions of
IL-17A were optimized to generate and follow a set of
peptides that covers the entire sequence of IL-17A. Two
parts of an exchanged protein sample were diluted with
one part of a cold acidified buffer [2 M urea and 1 M tris(2-
carboxyethyl)phosphine (pH 3.0)] and incubated for 1 min
at 1 °C. The exchanged and quenched sample was then
passed over an immobilized pepsin column (bed volume,
104 μl) at a flow rate of 200 μl/min at 0 °C, followed by
desalting on a reverse-phase trap column and separation
by a C18 column. In the process, the protease-generated
peptides from IL-17A were identified by the combination
of data-dependent MS/MS mode with dynamic exclusion
and SEQUEST software program (Thermo Electron Cor-
poration, San Jose, CA), as described previously.
42
Mass
spectrometric analyses were carried out with a Thermo
Finnigan LCQ™mass spectrometer (Thermo Electron
Corporation) with a capillary temperature of 200 °C.
Epitope mapping by H/D exchange
On-solution/off-column exchange experiments
CAT-2200 was immobilized on Poros 20 AL media, in
accordance with the manufacturer's instructions and as
described previously,
42
and packed in a column (bed
volume, 104 μl). On-exchange reaction was initiated by
diluting 20 μl of 0.5 mg/ml IL-17A with 60 μlof
deuterated buffer [50 mM Tris and 150 mM NaCl
(pH 7.5)]. The reaction mixture was incubated at 1 °C for
varying times (15–5000 s). The on-exchanged solution was
loaded onto the antibody column, which was preequili-
brated with 75% deuterated buffer [50 mM Tris and
150 mM NaCl (pH 7.5)]. The off-exchange reaction was
initiated by washing the column with aqueous buffer
[50 mM Tris and 150 mM NaCl (pH 7.5)]. Off-exchange
reaction continued for the same duration as the preceding
on-exchange reaction. The introduction of 120 μl of 0.8%
formic acid quenched the exchange reactions and eluted
out the antigen from the antibody column. The final 40 μl
of acid elution was subjected to digestion and analysis, as
described above.
On-column/off-column exchange experiments
Twenty microliters of 0.5 mg/ml IL-17 with 60 μlof
50 mM Tris and 150 mM NaCl (pH 7.5) was loaded onto
the antibody column. On-exchange reaction was initiated
by passing 200 μl of 50 mM Tris and 150 mM NaCl
(pH 7.5) in 75% D
2
O over the antibody column. The
column was incubated at 1 °C for varying times (150–
5000 s). Off-exchange reaction was initiated by washing
the column with aqueous buffer [50 mM Tris and 150 mM
NaCl (pH 7.5)]. Off-exchange reaction continued for the
same duration as the preceding on-exchange reaction.
The introduction of 120 μl of 0.8% formic acid quenched
the exchange reactions and eluted out the antigen from the
antibody column. The final 40 μl of acid elution was
subjected to digestion and analysis, as described above.
Fully deuterated experiment
The fully deuterated sample was prepared by incubating
the mixture of 80 μl of 0.5 mg/ml IL-17 and 240 μlof50mM
Tris with 150 mM NaCl (pH 7.5) in D
2
O at 60 °C for 4 h. The
fully deuterated sample was loaded onto the antibody
column preequilibrated with 75% deuterated buffer. The
antigen was then eluted out with 120 μl of 0.8% formic acid,
and the final 40 μl was subjected to the aforementioned
process.
Determination of the deuteration level of each peptide
after on–off exchange reaction
The percent deuteration of each peptide was calculated
using the following formula, and corrections for back-
exchange during the protein processing step were made
by employing methods described previously:
43
Deuteration Level %ðÞ=mP
ðÞ
mN
ðÞ
mF
ðÞ
mN
ðÞ
MaxD 100
where m(P), m(N), and m(F) are the centroid values of
partially deuterated (on–off exchanged) peptide, nondeu-
terated peptide, and fully deuterated peptide, respectively.
MaxD is the maximum deuterium incorporation calculated
by subtracting the number of prolines in the third or later
amino acid and two from the number of amino acids in the
peptide of interest (assuming the first two amino acids
cannot retain deuterons).
44
For IL-17, the deuterium
recovery of a fully deuterated sample [(m(F)−m(N))/
MaxD] was, on average, 68%.
Crystallization and X-ray diffraction data collection
Analytical gel filtration using an Ettan LC (GE Health-
care) fitted with a Superdex 200 PC 3.2/30 column (GE
Healthcare) identified that an IL-17A/Fab ratio of 3.6:1
produced the maximum yield of the complex. IL-17A and
Fab were then mixed together in this ratio and concen-
trated using 500 μl of a 30-kDa molecular mass cutoff
centrifugal concentrator (Viva Spin). The protein was
concentrated to 6.5 mg/ml, whereupon a 50-μl aliquot
was saved. The remainder was further concentrated up to
14.1 mg/ml. Hanging-drop vapor-diffusion crystalli-
zation was carried out at 20 °C using Nextal crystalli-
zation trays at the two concentrations of protein. Drops
were formed by mixing 2 μl of protein with 2 μl of the
reservoir solution.
Diffraction-quality crystals grew after 2–3 weeks from
the high-concentration protein hanging-drop experiments
with a reservoir solution containing 90.9 mM sodium
propionate, sodium cacodylate, bis-Tris propane buffer
(pH 4.0), 9.1 mM bis-Tris propane buffer (pH 9.5), 100 mM
ammonium sulfate, and 14% wt/vol polyethylene glycol
3350. Crystals were harvested, cryoprotected using (−)(−)
butane-2,3-diol (20% vol/vol in the corresponding well
solution), and flash-cooled at 100 K using an Oxford
Cryostream. The crystals were screened for diffraction
quality using a Rigaku FRe rotation anode generator
equipped with a Saturn 944 CCD detector, ACTOR
sample-changing robotics, and an XStream cryohead.
Those crystals that showed ordered diffraction were
stored in liquid nitrogen in readiness for data collection.
Diffraction data were collected from two single crystals
at 100 K in beamline ID29 at the European Synchrotron
Radiation Facility by employing an ADSC Quantum
Q315r detector. Two data sets from two crystals were
recorded at a wavelength of 0.9787 Å in rotation frames of
0.75° and 0.2° over an angular range of 225° and 155°,
respectively. Intensity data from each data set were
integrated with MOSFLM,
45
and merged and scaled
918 IL-17A in Complex with a Neutralizing Antibody
individually with programs from the CCP4 suite.
46,47
Images with an internal R
merge
of N20% were excluded
from the final scaling. Data collection statistics are shown
in Table 4. The crystals belong to the monoclinic space
group P2
1
, with typical cell dimensions of a=98.5 Å,
b=66.7 Å, c= 203.8 Å, and β= 91.7°. The crystallographic
asymmetric unit contains two complex assemblies (i.e.,
four molecules of Fab and two dimers of IL-17A), resulting
in a solvent content of 54% corresponding to a Matthews
coefficient of 2.68 Å
3
/Da.
Structure determination
The complex structure of IL-17A/Fab was solved at
3.0 Å by molecular replacement (MR) using the program
MOLREP. In the first trial, three molecules of the variable
domain (residues 2–110 of the light chain and residues 2–
124 of the heavy chain) of Fab (generated from PDB entry
1AQK
48
) were positioned and oriented, giving an R-factor
of 57% and a correlation coefficient of 16.7%. In a second
MR experiment, the positions of the three located Fab
variable domains were fixed, allowing three molecules of
the constant domain of Fab (residues 111–216 of the light
chain and residues 125–226 of the heavy chain) to be
located, giving an R-factor of 53% and a correlation
coefficient of 29.6%. The correctly generated Fab molecule
indicated a correct MR solution, which was used to place
the fourth Fab molecule in the asymmetric unit, now
giving an R-factor of 48% and a correlation coefficient of
41.5%. Restrained maximum likelihood refinement with
isotropic temperature factors using REFMAC resulted in
R
work
/R
free
of 33.6%/41.5%. Introduction of the correct
amino acid sequence into the model and manual
rebuilding were performed using the molecular graphics
program Coot,
47
lowering R
work
/R
free
to 29.5%/35.8%. At
this stage, MR failed to reveal the IL-17A dimer
molecules, while the density for these molecules was
visible on the F
o
−F
c
electron density difference map.
Polyalanine strands of various lengths could be built into
2F
o
−F
c
electron density, enabling the conserved amino
acid motif of IL-17A (50-RSTSPW-57) to be modeled un-
ambiguously through the tryptophan marker. This finally
allowed an overlay of the model of IL-17F
20
to complete
the building for the first IL-17A homodimer. After a
round of refinement (R
work
/R
free
=27.2%/33.4%), manual
rebuilding, and introduction of the correct IL-17A amino
acid sequence, we realized that a third of the homodimer
was disordered, and MR with MOLREP could now be
used successfully to place the second IL-17A homodimer,
using the ordered portion of the refined IL-17A model as
trial structure. The IL-17A/Fab complex structure, con-
taining two assemblies with a total of four Fab molecules
and two IL-17A dimers in the asymmetric unit, was
refined to convergence (R
work
/R
free
=21.2%/26.4%) with
REFMAC before water molecules and sulfate ions were
added manually using Coot. The distances were obtained
using the CCP4 program CONTACT.
49
Accession number
Structure coordinates have been deposited in the PDB
with accession number 2VXS.
Acknowledgements
We are grateful to David Lowe for valuable
discussions, Kevin Maggott and Patrick Dufner for
technical assistance with LI/LO, Lekan Daramola
and Richard Turner for generation of Fab, and
Markus Ganzlin and Judith Stanway for fermenta-
tion of E.-coli-expressing recombinant IL-17A.
References
1. Rouvier, E., Luciani, M. F., Mattei, M. G., Denizot, F. &
Golstein, P. (1993). CTLA-8, cloned from an activated
T cell, bearing AU-rich messenger RNA instability
sequences, and homologous to a herpesvirus saimiri
gene. J. Immunol. 150, 5445–5456.
2. Fossiez, F., Djossou, O., Chomarat, P., Flores-Romo,
L., Ait-Yahia, S., Maat, C. et al. (1996). T cell
interleukin-17 induces stromal cells to produce pro-
inflammatory and hematopoietic cytokines. J. Exp.
Med. 183, 2593–2603.
Table 4. Crystal parameters, X-ray data processing, and
refinement statistics
Space group P2
1
Wavelength (Å) 0.9787
Cell constants
a(Å) 98.5
b(Å) 66.7
c(Å) 203.8
β(°) 91.7
Resolution range (Å) 102.1–2.63
Completeness overall (%) 100 (100)
Unique reflections 79,235
Multiplicity 7.0 (6.9)
R
merge overall
(%)
a
12.8 (46.0)
R
cryst
(%)
b
21.2 (25.3)
R
free
(%) 26.4 (33.2)
Nonhydrogen protein atoms 15,396
Nonhydrogen sulfate atoms 60
Water molecules 214
r.m.s.d. from ideal values
Bond lengths (Å) 0.01
Bond angles (°) 1.54
Average Bvalues (Å
2
)
IL-17A main-chain atoms 49.6
IL-17A all atoms 50.0
Fab
heavy chain
main-chain atoms 39.1
Fab
heavy chain
all atoms 39.2
Fab
light chain
main-chain atoms 45.3
Fab
light chain
all atoms 45.4
Sulfate ions 74.3
Water molecules 38.0
R
free
is the cross-validation R-factor computed for the test set of
5% of unique reflections.
Values in parentheses correspond to the highest-resolution shell
(2.77–2.63 Å).
TLS refinement was used, and the Bvalues include the TLS
contribution.
99.3 % of all residues of the complex structure were found in the
most favored and additionally allowed regions of the Ramachan-
dran plot.
25
Of the remaining residues, 0.2% fall into the
generously allowed regions and 0.5% fall into the disallowed
regions.
a
R
merge
=∑
hkl
[(∑
i
|I
i
−〈I〉
|)/∑
i
I
i
].
b
R=∑
hkl
‖F
obs
|−|F
calc
‖/∑
hkl
|F
obs
|.
919IL-17A in Complex with a Neutralizing Antibody
3. Shen, F. & Gaffen, S. L. (2008). Structure–function
relationships in the IL-17 receptor: implications for
signal transduction and therapy. Cytokine,41,92–104.
4. Gaffen, S. L. (2009). Structure and signalling in the
IL-17 receptor family. Nat. Rev. Immunol. 9, 556–567.
5. Miossec, P. (2007). Interleukin-17 in fashion, at last: ten
years after its description, its cellular source has been
identified. Arthritis Rheum. 56, 2111–2115.
6. Wright, J. F., Guo, Y., Quazi, A., Luxenberg, D. P.,
Bennett, F., Ross, J. F. et al. (2007). Identification of an
interleukin 17F/17A heterodimer in activated human
CD4
+
T cells. J. Biol. Chem. 282, 13447–13455.
7. Liang, S. C., Long, A. J., Bennett, F., Whitters, M. J.,
Karim, R., Collins, M. et al. (2007). An IL-17F/A
heterodimer protein is produced by mouse Th17 cells
and induces airway neutrophil recruitment. J. Immunol.
179, 7791–7799.
8. Bettelli, E., Korn, T., Oukka, M. & Kuchroo, V. J.
(2008). Induction and effector functions of TH17 cells.
Nature,453, 1051–1057.
9. Weaver, C. T., Hatton, R. D., Mangan, P. R. &
Harrington, L. E. (2007). IL-17 family cytokines and
the expanding diversity of effector T cell lineages.
Annu. Rev. Immunol. 25, 821–852.
10. Witowski, J., Ksiazek, K. & Jorres, A. (2004).
Interleukin-17: a mediator of inflammatory responses.
Cell. Mol. Life Sci. 61, 567–579.
11. Bessis, N. & Boissier, M. C. (2001). Novel pro-
inflammatory interleukins: potential therapeutic
targets in rheumatoid arthritis. Jt. Bone Spine,68,
477–481.
12. Nakae, S., Nambu, A., Sudo, K. & Iwakura, Y. (2003).
Suppression of immune induction of collagen-induced
arthritis in IL-17-deficient mice. J. Immunol. 171,
6173–6177.
13. Lubberts, E., Koenders, M. I., Oppers-Walgreen, B.,
van den Bersselaar, L., Coenen-de Roo, C. J., Joosten,
L. A. & van den Berg, W. B. (2004). Treatment with a
neutralizing anti-murine interleukin-17 antibody after
the onset of collagen-induced arthritis reduces joint
inflammation, cartilage destruction, and bone erosion.
Arthritis Rheum. 50, 650–659.
14. Park, H., Li, Z., Yang, X. O., Chang, S. H., Nurieva, R.,
Wang, Y. H. et al. (2005). A distinct lineage of CD4 T
cells regulates tissue inflammation by producing
interleukin 17. Nat. Immunol. 6, 1133–1141.
15. Langrish, C. L., Chen, Y., Blumenschein, W. M.,
Mattson, J., Basham, B., Sedgwick, J. D. et al. (2005).
IL-23 drives a pathogenic T cell population that
induces autoimmune inflammation. J. Exp. Med. 201,
233–240.
16. Arican, O., Aral, M., Sasmaz, S. & Ciragil, P. (2005).
Serum levels of TNF-alpha, IFN-gamma, IL-6, IL-8,
IL-12, IL-17, and IL-18 in patients with active psoriasis
and correlation with disease severity. Mediat. Inflamm.
2005, 273–279.
17. Fujino, S., Andoh, A., Bamba, S., Ogawa, A., Hata, K.,
Araki, Y. et al. (2003). Increased expression of
interleukin 17 in inflammatory bowel disease. Gut,
52,65–70.
18. Koenders, M. I., Joosten, L. A. & van den Berg, W. B.
(2006). Potential new targets in arthritis therapy:
interleukin (IL)-17 and its relation to tumour necrosis
factor and IL-1 in experimental arthritis. Ann. Rheum.
Dis. 65, iii29–iii33.
19. Di Padova, F. E. (2008) IL-17 antagonistic antibodies.
Patent number WO2006013107.
20. Hymowitz, S. G., Filvaroff, E. H., Yin, J. P., Lee, J., Cai,
L., Risser, P. et al. (2001). IL-17s adopt a cystine knot
fold: structure and activity of a novel cytokine, IL-17F,
and implications for receptor binding. EMBO J. 20,
5332–5341.
21. Lloyd, C., Lowe, D., Edwards, B., Welsh, F., Dilks, T.,
Hardman, C. & Vaughan, T. (2009). Modelling the
human immune response: performance of a 1011
human antibody repertoire against a broad panel of
therapeutically relevant antigens. Protein Eng. Des. Sel.
22, 159–168.
22. Tomlinson, I. M., Williams, S. C., Corbett, S. J., Cox, J.
P. L., & Winter, G. (1997). V Base, http://www.mrc-
cpe.cam.ac.uk/imt-doc/public/INTRO.html.MRC
Center for Protein Engineering, University of Cam-
bridge, Cambridge, UK.
23. Kabat, E. A., Wu, T. T., Perry, H. M., Gottesmann, K. S.
& Foeller, C. (2008). Sequences of Proteins of
Immunological Interest Public Health Service, NIH,
Washington, DC.
24. Mangan, P. R., Harrington, L. E., O'Quinn, D. B.,
Helms, W. S., Bullard, D. C., Elson, C. O. et al. (2006).
Transforming growth factor-beta induces develop-
ment of the T(H)17 lineage. Nature,441, 231–234.
25. Laskowski, R. A., MacArthur, M. W., Moss, D. S. &
Thornton, J. M. (1993). PROCHECK: a program to
check the stereochemical quality of protein structures.
J. Appl. Crystallogr. 26, 283–291.
26. McDonald, N. Q. & Hendrickson, W. A. (1993). A
structural superfamily of growth factors containing a
cystine knot motif. Cell,73, 421–424.
27. Coales, S. J., Tuske, S. J., Tomasso, J. C. & Hamuro,
Y. (2009). Epitope mapping by amide hydrogen/
deuterium exchange coupled with immobilization of
antibody, on-line proteolysis, liquid chromatography
and mass spectrometry. Rapid Commun. Mass
Spectrom. 23, 639–647.
28. Kuestner, R. E., Taft, D. W., Haran, A., Brandt, C. S.,
Brender, T., Lum, K. et al. (2007). Identification of the
IL-17 receptor related molecule IL-17RC as the
receptor for IL-17F. J. Immunol. 179, 5462–5473.
29. Khalifa, M. B., Weidenhaupt, M., Choulier, L.,
Chatellier, J., Rauffer-Bruyere, N., Altschuh, D. &
Vernet, T. (2000). Effects on interaction kinetics of
mutations at the V
H
–V
L
interface of Fabs depend on
the structural context. J. Mol. Recognit. 13, 127–139.
30. Cauerhff, A., Goldbaum, F. A. & Braden, B. C. (2004).
Structural mechanism for affinity maturation of an
anti-lysozyme antibody. Proc. Natl Acad. Sci. USA,
101, 3539–3544.
31. Thom, G., Cockroft, A. C., Buchanan, A. G., Candotti,
C. J., Cohen, E. S., Lowne, D. et al. (2006). Probing a
protein–protein interaction by in vitro evolution. Proc.
Natl Acad. Sci. USA,103, 7619–7624.
32. Chen, Y., Wiesmann, C., Fuh, G., Li, B., Christinger,
H. W., McKay, P. et al. (1999). Selection and analysis
of an optimized anti-VEGF antibody: crystal structure
of an affinity-matured Fab in complex with antigen.
J. Mol. Biol. 293, 865–881.
33. Lippow, S. M., Wittrup, K. D. & Tidor, B. (2007).
Computational design of antibody-affinity improve-
ment beyond in vivo maturation. Nat. Biotechnol. 25,
1171–1176.
34. Tobbell, D. A., Middleton, B. J., Raines, S., Needham,
M. R., Taylor, I. W., Beveridge, J. Y. & Abbott, W. M.
(2002). Identification of in vitro folding conditions for
procathepsin S and cathepsin S using fractional
factorial screens. Protein Expression Purif. 24,
242–254.
35. Rudolph, R. & Lilie, H. (1996). In vitro folding of
inclusion body proteins. FASEB J. 10,49–56.
920 IL-17A in Complex with a Neutralizing Antibody
36. Vaughan, T. J., Williams, A. J., Pritchard, K., Osbourn,
J. K., Pope, A. R., Earnshaw, J. C. et al. (1996). Human
antibodies with sub-nanomolar affinities isolated
from a large non-immunized phage display library.
Nat. Biotechnol. 14, 309–314.
37. Hutchings, C. (2001). Generation of naïve human anti-
body libraries. pp. 93, Springer Laboratory Manuals,
Berlin.
38. Bannister, D., Wilson, A., Prowse, L., Walsh, M.,
Holgate, R., Jermutus, L. & Wilkinson, T. (2006).
Parallel, high-throughput purification of recombinant
antibodies for in vivo cell assays. Biotechnol. Bioeng. 94,
931–937.
39. Clackson, T. & Lowman, H. B. (2004). Phage Display—
A Practical Approach Oxford University Press, Oxford,
UK.
40. Thompson, J., Pope, T., Tung, J.S., Chan, C., Hollis, G.,
Mark, G. & Johnson, K. S. (1996). Affinity maturation of
a high-affinity human monoclonal antibody against the
third hypervariable loop of human immunodeficiency
virus: use of phage display to improve affinity and
broaden strain reactivity. J. Mol. Biol. 256,77–88.
41.Persic,L.,Roberts,A.,Wilton,J.,Cattaneo,A.,
Bradbury, A. & Hoogenboom, H. R. (1997). An
integrated vector system for the eukaryotic expression
of antibodies or their fragments after selection from
phage display libraries. Gene,187,9–18.
42. Hamuro, Y., Burns, L., Canaves, J., Hoffman, R., Taylor,
S. & Woods, V. (2002). Domain organization of D-
AKAP2 revealed by enhanced deuterium exchange–
mass spectrometry (DXMS). J. Mol. Biol. 321,703–714.
43. Zhang, Z. & Smith, D. L. (1993). Determination of
amide hydrogen exchange by mass spectrometry: a
new tool for protein structure elucidation. Protein Sci.
2, 522–531.
44. Bai, Y., Milne, J. S., Mayne, L. & Englander, S. W.
(1993). Primary structure effects on peptide group
hydrogen exchange. Proteins,17,75–86.
45. Leslie, A. (1991). Macromolecular data processing. In
Crystallographic Computing (Moras, D., Podjarny, A. D.
& Thierry, J. C., eds), pp. 27–38, Oxford University
Press, Oxford.
46. Potterton, E., Briggs, P., Turkenburg, M. & Dodson, E.
(2003). A graphical user interface to the CCP4
program suite. Acta Crystallogr. Sect. D,59, 1131–1137.
47. Emsley, P. & Cowtan, K. (2004). Coot: model-building
tools for molecular graphics. Acta Crystallogr. Sect. D,
60, 2126–2132.
48. Faber, C., Shan, L., Fan, Z., Guddat, L. W., Furebring,
C., Ohlin, M. et al. (1998). Three-dimensional structure
of a human Fab with high affinity for tetanus toxoid.
Immunotechnology,3, 253–270.
49. The CCP4 suite: programs for protein crystallography.
Acta Crystallogr. Sect. D,50. (1994)., 760–763.
921IL-17A in Complex with a Neutralizing Antibody
