Mutations within a human IgG2 antibody form ... - Wiley Online Library

Mutations within a human IgG2 antibody form distinct and homogeneous disulfide isomers but do not affect Fc gamma receptor or C1q binding

Sandra Lightle, Serdar Aykent, Nathan Lacher, Vesselin Mitaksov, Kristine Wells, James Zobel, and Theodore Oliphant* Pfizer Global Research and Development, Chesterfield, Missouri Received 5 October 2009; Revised 7 December 2009; Accepted 30 December 2009 DOI: 10.1002/pro.352 Published online 29 January 2010 proteinscience.org

Abstract: Human IgG2 antibodies may exist in at least three distinct structural isomers due to disulfide shuffling within the upper hinge region. Antibody interactions with Fc gamma receptors and the complement component C1q contribute to immune effector functions. These interactions could be impacted by the accessibility and structure of the hinge region. To examine the role structural isomers may have on effector functions, a series of cysteine to serine mutations were made on a human IgG2 backbone. We observed structural homogeneity with these mutants and mapped the locations of their disulfide bonds. Importantly, there was no observed difference in binding to any of the Fc gamma receptors or C1q between the mutants and the wild-type IgG2. However, differences were seen in the apparent binding affinity of these antibodies that were dependent on the selection of the secondary detection antibody used. Keywords: disulfide isomers; disulfide mutants; anti-CD44 mAb; human IgG2; Fc gamma receptors; C1q; effector functions

Introduction Antibodies have become increasingly important as therapeutics with 22 approved in the United States. Although antibodies can be grouped into one of five isotypes, IgD, IgE, IgG, IgM, and IgA, therapeutic Abbreviations: BSA, bovine serum albumin; DPBS, Dulbecco’s phosphate buffered saline; DTT, dithiothreitol; EDTA, ethylenediaminetetraacetic acid; FAB, fragment antigen binding; Fc, fragment crystallizable; HCl, hydrochloric acid; HEPES, N-2hydroxylethylpiperazine-N-2 ethanesulfonic acid; HPLC, high performance liquid chromatography; HRP, horseradish peroxidase; IgG, immunoglubin gamma; LC/MS, liquid chromatography/mass spectroscopy; mAb, monoclonal antibody; MES, 2-(N-morpholino) ethanesulfonic acid; MW, molecular weight; NaCl, sodium chloride; PBS, phosphate buffered saline; SDSPAGE, sodium dodecyl sulfate polyacrylamide gel electrophoresis; SEC, size exclusion chromatography; TMB, 3,30 ,5,50 -tetramethyl benzidine. *Correspondence to: Theodore Oliphant, Pfizer Global Research and Development, 700 Chesterfield Parkway West, Chesterfield, MO 63017. E-mail: [email protected]

C 2010 The Protein Society Published by Wiley-Blackwell. V

human antibodies are most commonly of the IgG isotype. Human IgG antibodies can be further subdivided into four subclasses, IgG1, IgG2, IgG3, and IgG4, based on primary amino acid sequence as well as their binding affinities to immune effector proteins such as Fc gamma receptors, the FcRn receptor, and C1q. A majority of marketed antibodies are of the IgG1 subclass, although IgG2 and IgG4 antibodies have also been approved. Choices between subclasses during the design phase have been based on the desired level of effector functions such as antibody-dependent cell-mediated cytotoxicity (ADCC) initiated through interactions with FccRIIIa and complement-dependent cytotoxity (CDC) mediated through interactions with C1q. Previous studies have shown that antibodies of the IgG1 subclass can demonstrate higher levels of ADCC and CDC than IgG2 or IgG4 antibodies.1 This functional activity directly correlates with increased measured affinities of each IgG subclass to

PROTEIN SCIENCE 2010 VOL 19:753—762

753

Figure 1. Schematic 2D representation of structural variants of IgG2 molecules.

the effector proteins. Despite the large difference in observed effector function, IgG1 and IgG2 share significant sequence identity throughout the constant regions, including regions shown to interact with Fcc receptors and C1q.2,3 The four residues determined to make up the C1q binding core are conserved among all IgG subclasses.2 Thus, studies that attempt to explain this disparity have focused on the hinge region, where the sequence variation is highest.2,4–14 The unique lower hinge residues from an IgG2 have been placed into both an IgG1 and IgG3 antibody background4,11,12,15 and assayed for binding to Fc receptors. The IgG1 mutant showed reduced binding to all tested Fc gamma receptors, however, the reduction observed was well below that of the wild-type IgG2/Fc receptors affinities. Also, this mutant, that contained the IgG2 lower hinge residues, had reduced affinity for FcRn (the neonatal Fc receptor) which is involved in antibody recycling. The binding epitope for FcRn is within the lower CH2 region and does not involve the lower hinge region, indicating that these mutations are causing a structural disruption broader than simply recreating the IgG2/Fcc receptor binding epitope. Mutations outside of defined contact residues in the upper and middle hinge, as well as CH2, also have been shown to impact Fc receptor binding.6,12 This has been attributed to either the length and flexibility of the hinge or interactions of the Fab region with the Fc portion of the antibody.3,6 Although the crystal structure of a full-length human IgG1 antibody has been solved, a complete crystal structure of an IgG2 antibody does not yet exist.16 Models of IgG2 antibodies have been made using the IgG1 structure as a guide as well as the disulfide bond arrangements of IgG2 antibodies determined more than 30 years ago.17,18 In both IgG1 and IgG2 structures, the hinge region forms a flexible linker between the Fab portion of the antibody and the Fc, which allows for rotation around the hinge. However, it has recently been reported that IgG2 antibody preparations are not homogene-

754

PROTEINSCIENCE.ORG

ous and instead consist of distinct isoforms.19–23 The isomeric structures are due to disulfide shuffling between cysteine residues within the light (LC) and heavy chains (HC) with the upper hinge sequence. Three IgG2 isomers (A, B, and A/B) have been characterized and shown to be structurally and, in at least some instances, functionally distinct.21 The canonical IgG2 structure, now referred to as IgG2-A, is present in low concentrations in purified antibody preparations and has been shown to convert to the A/B and B isomers in human serum.22 In the IgG2-A isomer, the cysteine residue at or near the end of each LC pairs with the cysteine residue within CH1 of the HC and the four cysteine residues within the hinge region form four interchain disulfide bonds. In isomer IgG2-B, the C-terminal cysteine residues in both of the LC pair with either one of the upper two cysteine residues in the hinge. In the IgG2-A/B isomer, the C-terminal cysteine residue of the LC is paired with the cysteine residue located in CH1. This pairing resembles that which is observed for the IgG2-A form. However, in the other Fab arm, the C-terminal cysteine from the LC is paired with one of the two upper cysteine residues in the hinge region of the HC, whereas the cysteine residing in CH1 is paired with the other residue in the hinge region from the opposite HC. To examine the role these disulfide isomers may have on effector function and antibody binding, a series of mutations have been designed on a common IgG2 antibody backbone. Characterization of these mutants indicates the generation of homogeneous isoforms that adopt either an IgG2-A-like structure or, in one instance, an IgG1-like structure. However, these mutations have little to no effect on Fc gamma receptor or C1q binding indicating that isomer composition has no impact on the accessibility of the binding sites. Also, differences determined by enzyme-linked immunosorbent assay (ELISA) binding studies between isomers could be entirely due to the secondary detection method used and not inherent in the antibody itself.

IgG2 Isomer Mutants and Their Effector Function

Table I. List of Anti-CD44 mAb Mutants Made by SiteDirected Mutagenesis to Evaluate the Role in Disulfide Isomer Patterns and Fc Receptor Binding. C127S is located in the CH1 Domain. Anti-CD44 mAb mutant evaluated

Analysis of mutants by nonreducing capillary electrophoresis

WT (IgG2) C127S C232S C233S WT (IgG1)

any significant differences between the mutants and the IgG2 wild-type antibody with only a 0.2 min retention time difference between all samples.

C232S/C233S (double mutant)

Results and Discussion The existence of disulfide-mediated isoforms in human IgG2 antibodies has been reported previously.19,22,23 The predicted structures of these disulfide isomers are shown in Figure 1 and labeled as IgG2-A, IgG2-B, or IgG2-A/B depending on the location of the disulfide bonds. This study investigated the effect of cysteine point mutants on disulfide isomer formation and the role each mutant plays in FccR, FcRn, and C1q binding. The following Cysteine (Cys) to Serine (Ser) HC mutants were made C127S, C232S, C233S, and C232/233S (see Table I). Residues are numbered according to the Kabat numbering system24 (Table II). The C127S mutation is located within the CH1 domain which traditionally pairs with C214 of the LC. The C232S and C233S mutations are located in the upper hinge region and have been shown previously to be involved in disulfide isomer formation.19,25

Initial characterization of anti-CD44 mAb mutants A human anti-CD44 IgG2 with a kappa LC was used for this study.26 The purification yield for each mutant was similar to wild type (data not shown) indicating that the mutations have no effect on expression. Nonreducing and reducing SDS-PAGE analysis was performed on all anti-CD44 mAbs listed in Table I with results shown in Figure 2(A,B). The data demonstrate the expected fully assembled antibody (HC2LC2) assembly on the nonreducing gel with only C233S giving an additional lower unidentified MW band which was seen previously.19 The reduced gel showed no bands other than the expected intact HC and LC, indicating that the extra band observed for C233S is not a contaminant. By analytical SEC [Fig. 2(C)], all mutated mAbs eluted as monomers. Analytical SEC did not detect

Recently published reports have described the use of Capillary Gel Electrophoresis (CGE) as an analytical tool for studying disulfide isomer heterogeneity.27 When wild-type IgG2 antibodies were analyzed by CGE, two peaks were found, peak #1 was identified as the IgG2-A and IgG2-A/B isomer and peak #2 was identified as the IgG2-B isomer, which was supported by Lys-C peptide mapping and historical data collected from orthogonal techniques including reverse-phase chromatography and cation exchange chromatography (data not shown). Wild-type anti-CD44 IgG2, anti-CD44 IgG1, and the mutants were analyzed using CGE with the results shown [Fig. 3(A)]. The electropherogram of the wild-type anti-CD44 IgG2 antibody gave two distinct peaks indicating that the previously reported isoform heterogeneity was present. In contrast, the C232S, C233S, and C232S/C233S mutants gave only a single peak by CGE that had the same migration time as the first peak in the wild-type mAb. These results suggest that disulfide-mediated isoforms around the hinge reason are absent and that these mutants appear to be isoform Ig2-A-like by CGE. Likewise, the C127S mutant and the IgG1 antibody gave one peak by CGE of the same migration time, thereby indicating that the C127S IgG2 mutant behaves similarly to an IgG1 antibody by CGE. These conclusions were corroborated by peptide mapping. Several different mixtures of individual mutants were made to further confirm the effect of the mutations on the profile by CGE as shown [Fig. 3(B)]. A mixture of the C232S and C233S mutants gave a single peak, thereby making these two variants indistinguishable by CGE. If the C127S mutant was then added to this mixture, two peaks were visible. This indicates that the C127S mutant was distinct from the C232S and C233S mutants. However, mixing the C127S mutant with wild-type IgG1 gave a single peak while in contrast, the mixing of wildtype IgG1 and IgG2 mAbs gave a peak with a distinct shoulder [Fig. 3(B)]. These results suggest that the isomeric structure generated by the C127S mutation is different from those of the C232S and C233S mutants and the C127S mutant closely resembles an IgG1-like structure.

Table II. The Hinge Sequence of the Human IgG2 Subclass with the Corresponding Kabat Residue Number IgG2 hinge sequence

E

R

K

C

C

V

E

C

P

P

C

Kabat residue numbering

226

227

228

232

233

235

237

239

240

241

242

Lightle et al.

PROTEIN SCIENCE VOL 19:753—762

755

Figure 2. Characterization of wild-type IgG1 and IgG2 and IgG2 Cys–Ser anti-CD44 mAb mutants by SDS-PAGE and analytical SEC. A: Proteins analyzed by SDS-PAGE under nonreducing conditions or (B) reducing conditions. C: Analytical SEC of WT IgG2 (red), IgG2 C127S (black), IgG2 C232S (green), IgG2 C233S (blue), and IgG2 C232S/C233S double mutant (purple).

Lys-C mapping of mutants The endoproteinase Lys-C mapping technique can be used to provide qualitative evidence for the existence of the three isoforms as previously reported.22,23,25 The key fragments used to assign identities to the isomeric structures observed in the fingerprints of anti-CD44 IgG2 wild-type antibody and three Cys– Ser variants (C232S, C233S, and C127S) are shown in Figure 4. The masses associated with the relevant

fragments are listed in Table III. For the wild-type anti-CD44 IgG2 mAb, the presence of four key fragments in the mass fingerprint confirms the existence of IgG2-A, IgG2-A/B, and IgG2-B isoforms. In the wild-type IgG2 mAb, the hinge dimer was detectable at very low levels suggesting that the IgG2-A isoform is the least abundant of the isomers. In sharp contrast, the fingerprints of the C232S and C233S mutants reveal a total absence of B and A/B

Figure 3. Nonreducing CGE-SDS electropherograms of (A) wild type, C232S, C233S, C127S, C232S/C233S, and IgG1. B: Analyses of anti-CD44 IgG2 wild type and mutant mixtures by CGE wild type, C232S and C233S mixture, C232S, C233S, and C127S mixture C127S and C232S/C233S mixture, C127S and IgG1 mixture, C232S, C233S, C127S, and C232S/C233S.

756

PROTEINSCIENCE.ORG


Figure 4. Endoproteinase Lys-C mapping of nonreduced anti-CD44 IgG2 WT and C232S, C233S, and C127S mutants.

fragments, thereby suggesting a totally different distribution of isoforms than the wild-type antibody. Furthermore, the classical HC–LC peptide, shown to be of moderate height in the fingerprint of the wildtype antibody, has more than doubled in size in the fingerprints of the C232S and C233S mutants. This observation gains further significance as it is also accompanied by the presence of a substantially large new peak (MW ¼ 5324 Da), which is consistent in mass with a hinge dimer that contains a Cys–Ser substitution at either position 232 or 233. The combination of (1) a classical HC–LC peak of maximal height, (2) a prominent hinge dimer peak containing three disulfides instead of four, (3) the absence of B fragments, and (4) the absence of A/B fragments suggests that the C232S and C233S variants are solely in an IgG2-A-like structure, which is consistent with the data from CGE. The only apparent difference between the proposed IgG2-A-like structures

observed with the C232/233S mutants and the classical IgG2-A structure is the elimination of one disulfide bond within the hinge region. The Lys-C fingerprint of the C127S variant demonstrates an absence of B fragments, A/B fragments, the A (HC–LC) fragment (10,096 Da), as well as a hinge dimer peak (Fig. 4). These observations suggest that the C127S mutant does not exist as any previously identified isomer. There are, however, a number of unique fragments within the Lys-C fingerprint which indicate that the C127S variant may exist as an alternate structure. The prominent peak with a mass of 9266 Da is consistent with the HC component of the HC–LC fragment (10,096 Da) that is missing its traditional disulfide-linked LC peptide partner (810 Da). This heptamer (SFNRGEC), which was not observed as a lone entity within the fingerprint, appears instead to be associated in disulfide linkage with the hinge dimer. As evidenced by the

Table III. Mass Measurement of Disulfide-Linked Peptides in Anti-CD44 IgG2 Monoclonal Antibodies mAb

Hinge dimer fragment

Isomer A (HC–LC) fragment

Isomer A/B fragment (s)

Isomer B fragment (s)

Unique HC fragment

Wild type

5354 Da

10,096 Da

5324 Da 5324 Da 6973 Da

10,096 Da 10,096 Da None

26,161 Da 26,771 Da None None None

None

C232S C233S C127S

15,451 Da 16,064 Da None None None

Lightle et al.

None None 9266 Da


757

presence of two separate peaks with masses of 6973 Da, the classical hinge dimer (5354 Da) appears to contain two copies of the LC peptide (810 Da). This observation suggests that endoproteinase Lys-C treatment of the C127S mutant may have produced a hinge dimer that also contains two SFNRGEC peptides that are disulfide linked to either C232 or to C233 of each HC. These Cys residues were judged to be the most likely candidates for disulfide-linked peptide attachment onto the hinge region as the lower two cysteine residues in the hinge have not been implicated in any type of isomerization event. If this prediction is correct, the C127S variant could exist as one or both of the structures shown (Fig. 4). The proposed disulfide bond pattern of the C127S variant appears to greatly resemble that which is seen in a conventional IgG1 antibody.

Analysis of Fcc receptor and FcRn binding by surface plasmon resonance The effect of disulfide mutants on their ability to bind Fc receptors was investigated using SPR. The three classes of FccRs tested were: (1) the high affinity receptor FccRI, (2) the two polymophic variants of the low affinity receptor FccRIIa/131H and FccRIIa/131R, and (3) the two polymorphic variants of the medium affinity receptor, FccRIIIa/158F and FccRIIIa/158V. The anti-CD44 IgG2 mutants listed in Table I were compared to wild-type IgG2 with results shown in Figure 5. Anti-CD44 IgG1 was used for comparison. Binding affinities for wild-type IgG1 and IgG2 were consistent with previous reported values.28 Importantly, none of the mutants, including C127S that is IgG1 like by CGE and Lys-C mapping, showed altered binding to any of the receptors. The IgG subclass-specific Fc binding differences between IgG1 and IgG2 were maintained in the IgG2 mutants for all of the receptors tested. This suggests that the lack of effector function for IgG2 antibodies is not due to their structural isomer state. The anti-CD44 IgG2 mutants were also evaluated for their ability to bind FcRn receptors

Figure 5. Characterization of anti-CD44 mutations binding to Fc gamma receptors and FcRn by SPR.

758

PROTEINSCIENCE.ORG

Figure 6. Analysis of anti-CD44 mAb mutants effect of C1q binding by ELISA.

through comparisons with the IgG1 and IgG2 wildtype molecules at pH 6.0. All of the IgG2 mutants bound with similar affinity as the IgG2 wild-type mAb, whereas the IgG1 bound to the FcRn receptor with a slightly higher affinity (1.4 nM vs. 3–4 nM). These values are similar to those values obtained using a biosensor-based assay in which the FcRn is immobilized on the biosensor surface.29

Analysis of C1q binding by ELISA to anti-CD44 mAbs A binding ELISA was performed to evaluate the effect of anti-CD44 mAb Cys–Ser mutants listed in Table I on complement component C1q binding with results shown in Figure 6. There was no apparent effect on the Clq binding of the four mutants as compared with the wild-type IgG2 mAb. In fact, the wild-type mAbs and mutant mAbs were indistinguishable from each other indicating that there was no effect observed on C1q binding as a consequence of the hinge and CH1 Cys–Ser mutations. The IgG1 isotype of the anti-CD44 mAb was used as a positive control in this assay, and it gave a 7-fold increase in signal at the highest concentration of mAb tested compared with all the IgG2 molecules.

Figure 7. Apparent binding of anti-CD44 mAbs mutants using two different secondary antibodies by ELISA.


Table IV. Affinity Binding Data by SPR for Anti-CD44 IgG2 WT and Cysteine to Serine Mutants Sample

ka (M1 s1)

SD ka

kd (s1)

SD kd

KD (M)

SD KD

Wild type C127S C232S C233S C232S/C233S

4.07Eþ06 3.87Eþ06 4.04Eþ06 3.93Eþ06 4.21Eþ06

8.50Eþ05 8.66Eþ05 8.47Eþ05 8.42Eþ05 8.86Eþ05

3.29E05 3.36E05 3.27E05 3.81E05 4.07E05

6.86E06 7.31E06 7.47E06 7.25E06 7.24E06

8.17E12 8.81E12 8.18E12 9.85E12 9.81E12

1.13E12 1.33E12 1.25E12 1.44E12 1.41E12

Binding potency analysis of anti-CD44 mAbs The binding potencies of anti-CD44 IgG2 Cys–Ser mutants were compared to wild type using an indirect ELISA assay (Figure 7). Two secondary detection antibodies that are known to specifically bind IgG2 antibodies at different regions were used.30 HP6014 binds to the hinge region, whereas HP6002 binds to an epitope within the Fc portion of the antibody. EC50 values were determined and compared to the IgG2 wild-type antibody using both secondary detection mAbs. No significant difference in the apparent binding of each of the mutant antibodies to CD44 extracellular domain (ECD) was observed with HP6002. In terms of the HP6014 antibody, there was no difference in the binding of the C127S mutant. However, significant differences were seen with the two upper-hinge mutants (C232S and C233S). The C232S mutant had an apparent binding affinity that was 1.5-fold higher than wild-type IgG2. The C233S mutant had an even larger increase in apparent affinity of roughly 3-fold. When the double mutant, C232S/C233S, was analyzed, the largest difference, 6.5-fold, was seen. Differences between the C232S and C233S mutants were not observed in any of the other methods used to analyze these mutants with the exception of reduced SDS-PAGE analysis. To confirm that the difference in apparent binding affinities of the mutants were entirely due to the secondary antibody used and not because of inherent affinity differences caused by the mutants, a biophysical assay was developed to measure the kinetics of association and dissociation of each antibody Cys to Ser variant using SPR technology (Table IV). The data suggest that there are no significant changes in the kinetics of binding to CD44 ECD as a result of the Cys to Ser substitutions. This indicates that CD44 binding appears to be insensitive to disulfide isomer status in this particular example of antibody–target interaction.

Materials and Methods Cloning and expression of anti-CD44 monoclonal antibodies Mutants were made using the Quickchange sitedirected mutagenesis kit (Stratagene, La Jolla, CA). Mutations were confirmed by sequencing and

Lightle et al.

expressed transiently in HEK293F cells using the Freestyle 293 expression system (Invitrogen, Carlsbad, CA).

Purification and characterization of anti-CD44 monoclonal antibodies Secreted mAbs from transiently transfected HEK293 were purified using HiTrap protein A columns (GE Healthcare, Piscataway, NJ), dialyzed into 20 mM sodium acetate, 120 mM NaCl, pH 5.5, and stored at 4 C. The concentration of each mAb was determined using absorbance at 280 k. Both nonreducing and reducing SDS-PAGE were performed using 4–12% NuPAGE gels (Invitrogen, Carlsbad, CA) in MES buffer. Briefly, the nonreducing samples had 25 mM iodoacetamide added and reducing samples had 10 mM DTT added. Samples were then denatured at 70 C for 3 min with 4–8 lg mAb loaded per lane. Analytical SEC analyses were performed on a Waters 2695 HPLC (Milford, MA) using a TSK gel Super SW3000 (7.8 mm 300 mm, Tosoh Bioscience, Montgomeryville, PA) column. The column was operated at a flow rate of 0.5 mL min1 with a mobile phase of 0.2M disodium phosphate pH 7.2 with 25 lg antibody injected for each sample.

Nonreduced CGE CGE sample preparations consisted of: 1 mg mL1 protein in 50% 100 mM Tris-HCL, 1% SDS, pH 9 (ProteomeLab SDS-MW sample buffer, Beckman Coulter, Fullerton, CA), 5% iodoacetamide (250 mM), and 2 lL 10 kDa reference standard (Beckman Coulter) in water. A 10 kDa reference standard was included in the sample preparation which served as a migration time reference marker allowing the samples to be compared based on migration time. Sample preparations were heated at 65 C in a water bath for 10 min and analyzed. Analyses were preformed with a ProteomeLab PA800 (Beckman Coulter) instrument equipped with a photo diode array detector (PDA) using methods based on manufacturer’s recommendations with some modifications. Separations were generated using a bare-fused silica capillary (50 lm I.D., Polymicro Technologies, Phoenix, AZ) with a total length of 30.2 cm and an effective length of 20.2 cm. The PDA detector was set to a wavelength of 220 nm. The capillary was preconditioned and then filled with CE-SDS buffer (Bio-Rad,


759

Hercules, CA). Samples were injected at 5 kV for 30–40 s and separated at 15 C with 15 kV applied for 30 min with 20 psi pressure applied to the capillary inlet and outlet.

Endoproteinase Lys-C peptide mapping Approximately 80 lg of protein was incubated at 37 C in the presence of 8M guanidine HCl and 15 mM iodoacetamide for about 6 h. The digestion reaction was prepared by adding 100 lL of the denatured antibody into a solution consisting of 4M urea in 100 mM Tris, pH 7.5. Endoproteinase Lys-C was then added [enzyme:substrate ¼ 1:15 (w:w)] and incubated at 37 C for a minimum of 3 h. The digested samples were applied to a Vydac 214TP52 ˚ pore size, 2.1 C4 column (5 lm particle size, 300 A 250 mm) (Hesperia, CA) set at 60 C and analyzed using an Agilent 1100 HPLC (Santa Clara, CA). A linear gradient from 0 to 45% B in 120 min at a flow rate of 0.2 mL min1 was used to elute the peptides. Mobile phase A consisted of 0.1% trifluoracetic acid in water, whereas mobile phase B consisted of 0.1% trifluoracetic acid in acetonitrile. Peptides were monitored at 214 nm. To identify the peaks in the chromatogram, LC/MS was performed using the same separation as described earlier, but the eluent from the HPLC was directed into a Q-TOF Micro (Waters) electrospray time-of-flight mass spectrometer.

Fc gamma and FcRn binding by SPR Fc gamma receptor binding and FcRn receptor binding were measured by SPR using a Biacore 3000 instrument (GE Healthcare, Piscataway, NJ). The human Fc gamma receptor I was purchased from R&D Systems (Minneapolis, MN). The two FccRIIa polymorphic variants (131H and 131R), the two polymorphic variants FcgRIIIa (158V and 158F), and human FcRn were produced in-house by transient expression in HEK293F cells. For Fc gamma receptor binding studies, the anti-CD44 mAbs listed in Table I were captured on a CM5 sensor chip (GE Healthcare) with an immobilized protein A surface. The protein A (Thermo Scientific, Rockford, IL) was immobilized on a sensor chip using the standard primary amine coupling protocol according to manufacturer’s instructions. All measurements were performed in 10 mM HEPES, 150 mM NaCl, 3 mM EDTA, 0.005% polysorbate 20, and pH 7.4. Fc receptors were injected over the protein A captured antibody surface for 2 min at 50 lL min1 followed by a 2–3 min dissociation phase. One protein A flow cell surface without antibody capture was used for reference subtraction. If necessary, the chip was regenerated with 10 mM phosphoric acid to return the signal to baseline before subsequent injections.

760

PROTEINSCIENCE.ORG

For FcRn binding studies, the FcRn receptor was immobilized on a CM5 chip using the standard primary amine coupling protocol. All measurements were performed in 50 mM MES, 120 mM NaCl, 0.005% polysorbate 20, pH 6 with 0.5 mg mL1 BSA added. The anti-CD44 mAbs listed in Table I were injected over the immobilized FcRn receptor for 3 min at a flow rate of 50 lL min1 followed by a 3 min dissociation phase. One flow cell was mock immobilized with ethanolamine for reference subtraction. The surface was regenerated with two 6 s pulses of HEPES buffered saline (HBS) at pH 7.4 to return signal to baseline before subsequent injections. Data were fit to a 1:1 binding model (Langmuir) using Scrubber2 data analysis software (BioLogic software, Campbell, Australia). Kinetic variables were used to calculate the equilibrium dissociation constant (KD).

C1q binding by ELISA to anti-CD44 mAbs To examine if the anti-CD44 IgG2 cysteine mutations had any effect on Clq binding, an ELISA was performed, similar to previously published methods.31 The anti-CD44 mAbs were twice diluted to a concentration of 100 lg mL1 in PBS then serially diluted to an assay range of 1.56–100 lg mL1. Duplicates of each mAb concentration were bound to individual wells of a 96-well Maxisorp Immunoplate (Nunc, Rochester, NY) overnight at 4 C. The wells were blocked with 6% albumin (Sigma, St. Louis, MO) in PBS for 2 h at 37 C. The plate was washed 5 with PBS/0.05% Tween-20 after each incubation. The plate was incubated with 4 lg mL1 human Clq protein (Quidel, San Diego, CA) for 1 h at 37 C, washed, and then incubated with goat anti-C1q antibody (Quidel) for 1 h at 37 C. The plate was washed and incubated with donkey anti-goat IgG HRP (Promega, Madison, WI) for 1 h at room temperature. The plate was developed with TMB, stopped with STOP solution (Cell signaling Technology, Danvers, MA), and read at 450 nm.

Binding potency of anti-CD44 mAbs to their target by ELISA To examine if the anti-CD44 IgG2 cysteine mutations had any effect on binding potency, an indirect binding ELISA was performed using two different secondary antibodies. HP6014 was obtained from Zymed (San Francisco, CA) and HP6002 from Southern Biotech (Birmingham, AL). An indirect ELISA was performed by diluting the Fc-linked CD44 ECD that was made in-house into DPBS and coating onto the surface of a 96-well polystyrene microtiter plate (0.25 lg mL1 for HP6014 and 1 lg mL1 for HP6002). After incubation, the plate was washed three times with DPBS containing 0.05% Tween-20. The plate was blocked with DPBS containing 1% BSA (secondary HP6014) or 1% BSA þ 0.2% Tween-


20 (secondary HP6002) and then washed as before. The anti-CD44 IgG2 mAbs were prepared in triplicate by diluting to an initial concentration of 600 ng mL1 in the BSA solution then serially diluting to 0.15 ng mL1 (secondary HP6014), or an initial concentration of 4.8 lg mL1 in the BSA/Tween solution then serially diluting to 1.2 ng mL1 (secondary HP6002) to obtain a dose-response curve and applied to the plate. The plate was incubated with HRP-conjugated mouse anti-human IgG2 (HP6014), or HRPconjugated mouse anti-human IgG2 (HP6002). The plate was developed with TMB substrate (Cygnus, Southpoint, NC) and stopped using 1M sulfuric acid and read at 450 nm. The data were analyzed using a 4-parameter fit to obtain an EC50 value. Because of lack of equivalency of some curves, a reduced 4-parameter fit was not used. In these cases, the EC50 value was compared to the EC50 result for the wildtype sample (run on each plate) to obtain a ratio. The ratio of wild-type EC50 to sample EC50 was graphed to show apparent binding.

Binding potency of anti-CD44 mAbs to their target by SPR The binding and dissociation kinetics of the antiCD44 IgG2 cysteine mutants (Table I) to CD44 were measured by SPR and compared with those of the anti-CD44 IgG2 wild type using a Biacore T100 instrument (GE Healthcare). The Fc-linked CD44 ECD was immobilized on the CM5 sensor chip with an immobilized protein A surface (as previously described). Each of the IgG2 mutants was diluted twice to a final starting concentration of 2 nM in buffer containing 10 mM HEPES, 150 mM NaCl, 3 mM EDTA, and 0.05% polysorbate 20. The antiCD44 mAbs were injected over the immobilized CD44 chip in order of increasing concentration for 200 s at a flow rate of 65 lL min1 followed by a 300 s dissociation phase. The surface was regenerated with of two brief pulses of 8.5% phosphoric acid spiked in the running buffer, followed by a brief equilibration time of 3 min before the beginning of the next injection cycle. The average response from at least three blank buffer injections was subtracted from the rest of the responses for each injection cycle and for each flow cell. The data were fit to a 1:1 binding model (Langmuir). The final reported parameters are the average of all six determinations.

Conclusions The results herein confirm earlier reports that show cysteine mutants in the upper-hinge region or within CH1 of an IgG2 antibody can reduce heterogeneity caused by disulfide shuffling.19 The C127S, C232S, C233S, and C232S/C233S mutants were shown to be homogeneous when analyzed by CGE. However, differences between the mutants have been established by CGE analysis. The C127S

Lightle et al.

mutant elutes as a separate peak from either the C232S or C233S mutant. It is also indistinguishable from the IgG1 antibody in this analysis. On the basis of tryptic mapping, the C127S mutant contains peptides that are consistent with a disulfide bond between C214 of the HC and either C232 or C233 in the upper hinge. Of all four canonical human IgG subclasses, only IgG1 has a disulfide bond between the HC and the hinge region. Although this indicates that the C127S mutant is IgG1 like in terms of the hinge structure, this mutant retains an IgG2’s affinity for C1q, FcRn, FccRI, FccRIIa, and FccRIIIa. Although the ADCC and CDC activities of these mutants were not directly assessed in this report, based on the lack of change in binding affinities to C1q or Fc gamma receptors for any of the mutants, it is unlikely that measured effector functions would differ from wild-type IgG2. The C232S and C233S mutants were indistinguishable from each other based on CGE analysis and tryptic mapping and appear to possess an entirely IgG2-A-like structure. However, the potency assay showed significantly different results when the HP6014 secondary antibody was used. The reasons for this are unclear but indicate that there are subtle conformational differences between these two mutants. Because of this, care must be taken when assessing potency of IgG2 antibodies using indirect methods. For use as therapeutic agents, each of these mutants may offer some advantage in terms of homogeneity over a wild-type IgG2 antibody. Nevertheless, subtle differences exist between the variants that could complicate selection. Further study, including thermostability, long-term stability, and formulation analyses should be done to aid in therapeutic antibody design.

Acknowledgments The authors thank the following people for their help and support on this study: Sandeep Kumar, James Carroll, Bilikallahalli Muaralidhara, Qian Wang, Bernie Violand, Graeme Bainbridge, and Jerry Casperson. They also thank Kristen Shannon, Jing Ming, and James Duerr for supplying the Fc receptors.

References 1. Bruggemann M, Williams GT, Bindon CI, Clark MR, Walker MR, Jefferis R, Waldmann H, Neuberger MS (1987) Comparison of the effector functions of human immunoglobulins using a matched set of chimeric antibodies. J Exp Med 166:1351–1361. 2. Idusogie EE, Presta LG, Gazzano-Santoro H, Totpal K, Wong PY, Ultsch M, Meng YG, Mulkerrin MG (2000) Mapping of the C1q binding site on rituxan, a chimeric antibody with a human IgG1 Fc. J Immunol 164: 4178–4184. 3. Sondermann P, Huber R, Oosthuizen V, Jacob U (2000) The 3.2-A crystal structure of the human IgG1 Fc fragment-Fc gammaRIII complex. Nature 406:267–273.


761

4. Armour KL, van de Winkel JG, Williamson LM, Clark MR (2003) Differential binding to human FcgammaRIIa and FcgammaRIIb receptors by human IgG wildtype and mutant antibodies. Mol Immunol 40:585–593. 5. Brekke OH, Michaelsen TE, Aase A, Sandin RH, Sandlie I (1994) Human IgG isotype-specific amino acid residues affecting complement-mediated cell lysis and phagocytosis. Eur J Immunol 24:2542–2547. 6. Dall’Acqua WF, Cook KE, Damschroder MM, Woods RM, Wu H (2006) Modulation of the effector functions of a human IgG1 through engineering of its hinge region. J Immunol 177:1129–1138. 7. Duncan AR, Winter G (1988) The binding site for C1q on IgG. Nature 332:738–740. 8. Duncan AR, Woof JM, Partridge LJ, Burton DR, Winter G (1988) Localization of the binding site for the human high-affinity Fc receptor on IgG. Nature 332: 563–564. 9. Michaelsen TE, Brekke OH, Aase A, Sandin RH, Bremnes B, Sandlie I (1994) One disulfide bond in front of the second heavy chain constant region is necessary and sufficient for effector functions of human IgG3 without a genetic hinge. Proc Natl Acad Sci USA 91:9243–9247. 10. Redpath S, Michaelsen TE, Sandlie I, Clark MR (1998) The influence of the hinge region length in binding of human IgG to human Fcgamma receptors. Hum Immunol 59:720–727. 11. Sensel MG, Kane LM, Morrison SL (1997) Amino acid differences in the N-terminus of C(H)2 influence the relative abilities of IgG2 and IgG3 to activate complement. Mol Immunol 34:1019–1029. 12. Shields RL, Namenuk AK, Hong K, Meng YG, Rae J, Briggs J, Xie D, Lai J, Stadlen A, Li B, Fox JA, Presta LG (2001) High resolution mapping of the binding site on human IgG1 for 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. 13. Tao MH, Smith RI, Morrison SL (1993) Structural features of human immunoglobulin G that determine isotype-specific differences in complement activation. J Exp Med 178:661–667. 14. Xu Y, Oomen R, Klein MH (1994) Residue at position 331 in the IgG1 and IgG4 CH2 domains contributes to their differential ability to bind and activate complement. J Biol Chem 269:3469–3474. 15. Armour KL, Clark MR, Hadley AG, Williamson LM (1999) Recombinant human IgG molecules lacking Fcgamma receptor I binding and monocyte triggering activities. Eur J Immunol 29:2613–2624. 16. Saphire EO, Parren PW, Pantophlet R, Zwick MB, Morris GM, Rudd PM, Dwek RA, Stanfield RL, Burton DR, Wilson IA (2001) Crystal structure of a neutralizing human IGG against HIV-1: a template for vaccine design. Science 293:1155–1159. 17. Frangione B, Milstein C (1968) Variations in the S-S bridges of immunoglobins G: interchain disulfide bridges of gamma G3 myeloma proteins. J Mol Biol 33: 893–906.

762

PROTEINSCIENCE.ORG

18. Milstein C, Frangione B (1971) Disulphide bridges of the heavy chain of human immunoglobulin G2. Biochem J 121:217–225. 19. Allen M, Guo A, Martinez T, Han M, Flynn G, Wypych J, Liu YD, Shen W, Dillon T, Vezina C, Balland A (2009) Interchain disulfide bonding in human IgG2 antibodies probed by site-directed mutagenesis. Biochemistry 48:3755–3766. 20. Dillon TM, Bondarenko PV, Rehder DS, Pipes GD, Kleemann GR, Ricci MS (2006) Optimization of a reversed-phase high-performance liquid chromatography/mass spectrometry method for characterizing recombinant antibody heterogeneity and stability. J Chromatogr A 1120:112–120. 21. Dillon TM, Ricci MS, Vezina C, Flynn GC, Liu YD, Rehder DS, Plant M, Henkle B, Li Y, Deechongkit S, Varnum B, Wypych J, Balland A, Bondarenko PV (2008) Structural and functional characterization of disulfide isoforms of the human IgG2 subclass. J Biol Chem 283:16206–16215. 22. Liu YD, Chen X, Enk JZ, Plant M, Dillon TM, Flynn GC (2008) Human IgG2 antibody disulfide rearrangement in vivo. J Biol Chem 283:29266–29272. 23. Wypych J, Li M, Guo A, Zhang Z, Martinez T, Allen MJ, Fodor S, Kelner DN, Flynn GC, Liu YD, Bondarenko PV, Ricci MS, Dillon TM, Balland A (2008) Human IgG2 antibodies display disulfide-mediated structural isoforms. J Biol Chem 283:16194–16205. 24. Kabat EA, Perry HM, Gottesman KS, Foeller C, editors (2008). Sequence of proteins of immunologic interest, 5th ed. Bethesda, MD: National Institutes of Health, 1991. 25. Martinez T, Guo A, Allen MJ, Han M, Pace D, Jones J, Gillespie R, Ketchem RR, Zhang Y, Balland A (2008) Disulfide connectivity of human immunoglobulin G2 structural isoforms. Biochemistry 47:7496–7508. 26. Xu XB, Vahe B, Meaddough E, Huang H, Yang L, Toy K, Srinivasan M, Badkar AV (filed 2007) Recombinant monoclonal human anti-human CD44 antigen antibodies and fragments for diagnosis and treatment of inflammation and autoimmune disease. WO2008079246. 27. Guo A, Han M, Martinez T, Ketchem RR, Novick S, Jochheim C, Balland A (2008) Electrophoretic evidence for the presence of structural isoforms specific for the IgG2 isotype. Electrophoresis 29:2550–2556. 28. van de Winkel JG, Capel PJ (1993) Human IgG Fc receptor heterogeneity: molecular aspects and clinical implications. Immunol Today 14:215–221. 29. Vaughn DE, Bjorkman PJ (1997) High-affinity binding of the neonatal Fc receptor to its IgG ligand requires receptor immobilization. Biochemistry 36:9374–9380. 30. Harada S, Hata S, Kosada Y, Kondo E (1991) Identification of epitopes recognized by a panel of six antihuman IgG2 monoclonal antibodies. J Immunol Meth 141:89–96. 31. Oganesyan V, Gao C, Shirinian L, Wu H, Dall’Acqua WF (2008) Structural characterization of a human Fc fragment engineered for lack of effector functions. Acta Crystallogr D 64:700–704.
