Antibodies Through Fcrn Binding Monoclonal Pharmacokinetics Of

IgG1 Allotypes Influence the Pharmacokinetics of Therapeutic Monoclonal Antibodies through FcRn Binding

This information is current as David Ternant, Christophe Arnoult, Martine Pugnière, of October 2, 2021. Christine Dhommée, Daniel Drocourt, Eric Perouzel, Christophe Passot, Nadine Baroukh, Denis Mulleman, Gérard Tiraby, Hervé Watier, Gilles Paintaud and Valérie Gouilleux-Gruart J Immunol published online 18 December 2015 http://www.jimmunol.org/content/early/2015/12/17/jimmun Downloaded from ol.1501780

Supplementary http://www.jimmunol.org/content/suppl/2015/12/17/jimmunol.150178 http://www.jimmunol.org/ Material 0.DCSupplemental

Why The JI? Submit online.

• Rapid Reviews! 30 days* from submission to initial decision

• No Triage! Every submission reviewed by practicing scientists by guest on October 2, 2021

• Fast Publication! 4 weeks from acceptance to publication

*average

Subscription Information about subscribing to The Journal of Immunology is online at: http://jimmunol.org/subscription Permissions Submit copyright permission requests at: http://www.aai.org/About/Publications/JI/copyright.html Email Alerts Receive free email-alerts when new articles cite this article. Sign up at: http://jimmunol.org/alerts

The Journal of Immunology is published twice each month by The American Association of Immunologists, Inc., 1451 Rockville Pike, Suite 650, Rockville, MD 20852 Copyright © 2015 by The American Association of Immunologists, Inc. All rights reserved. Print ISSN: 0022-1767 Online ISSN: 1550-6606. Published December 18, 2015, doi:10.4049/jimmunol.1501780 The Journal of Immunology

IgG1 Allotypes Inﬂuence the Pharmacokinetics of Therapeutic Monoclonal Antibodies through FcRn Binding

David Ternant,*,†,1 Christophe Arnoult,*,1 Martine Pugnie`re,‡ Christine Dhomme´e,* Daniel Drocourt,x Eric Perouzel,x Christophe Passot,*,{ Nadine Baroukh,* Denis Mulleman,*,‖ Ge´rard Tiraby,x Herve´ Watier,*,{ Gilles Paintaud,*,† and Vale´rie Gouilleux-Gruart*,{

Because IgG1 allotypes might have different half-lives, their influence on infliximab (G1m17,1 allotype) pharmacokinetics was investigated in a group of spondyloarthritis patients. Infliximab was found to have a shorter half-life in patients homozygous for the G1m17,1 allotypes than in those carrying the G1m3 with no G1m1 (G1m3,-1) allotype. Because the neonatal FcR (FcRn) is involved in the pharmacokinetics of mAbs, the interaction of different IgG1 allotypes with FcRn was examined using cellular

assays and surface plasmon resonance. G1m17,1 mAbs, such as infliximab and rituximab, were shown to bind more efficiently to Downloaded from FcRn and to be transcytosed better than the G1m3,-1 mAb cetuximab, which explains why infliximab is a better competitor for endogenous IgG1 in G1m3,-1 allotype–bearing patients. A set of four allotype variants of adalimumab (G1m17,1; G1m17,-1; G1m3,1; and G1m3,-1) was also tested for its binding to FcRn, revealing that the G1m3,1 variant, not present in commercial mAbs, binds more efficiently to FcRn and is transcytosed better than the other three variants, all of which are found in therapeutic mAbs. The Journal of Immunology, 2016, 196: 000–000. http://www.jimmunol.org/ he neonatal FcR (FcRn), described by Brambell in 1964, is bution throughout the body (6–10). More recently, FcRn was one of the receptors able to bind IgG via the Fc portion of found to be involved in phagocytosis, Ag presentation, and hu- T the latter (1, 2). However, it is functionally distinguish- moral and antitumoral immune responses, opening new functional able from other FcgRs by its binding to Fc at acidic pH 6 and by areas for this multifaceted molecule (11–15). its affinity for another substrate, albumin. IgG and albumin bind Composed of an a-chain and b2-microglobulin, FcRn is an to two opposite sites on FcRn and, therefore, do not behave as HLA class I homolog devoid of any peptide-binding groove be- competitors (3–5). Ligand transcytosis and recycling are well- tween domains 1 and 2 of the a-chain (2). The amino acids that documented FcRn properties that are responsible for the long are crucial for the interaction between FcRn and IgG or albumin half-life of IgG and albumin in blood, as well as for their distri- were identified by mutation experiments and crystallography by guest on October 2, 2021 studies (16, 17). This led to the isolation of IgG variants that bind better to FcRn while retaining pH dependency (17, 18). Such *Universite´ François Rabelais de Tours, CNRS UMR7292, Tours F-37032, France; variants have potentially high therapeutic value in the context †Laboratoire de Pharmacologie-Toxicologie, Centre Hospitalier Re´gional Universi- taire de Tours, Tours F-37032, France; ‡INSERM, U1194, Institut de Recherche en of the many therapeutic mAbs and other IgG Fc- or albumin- Cance´rologie de Montpellier, Universite´ de Montpellier; Montpellier F-34298, containing biologics that target a wide range of diseases, such France; xInvivoGen, Toulouse F-31400, France; {Laboratoire d’Immunologie, Centre Hospitalier Re´gional Universitaire de Tours, Tours F-37032, France; and ‖Service de as cancer, inflammation, infection, or genetic deficiencies (19–22). Rhumatologie, Centre Hospitalier Re´gional Universitaire de Tours, Tours F-37032, Indeed, the development of mAbs or biologics with prolonged France plasma half-lives would allow less frequent injections. Such 1D.T. and C.A. contributed equally to this work. agents could also compete with endogenous IgG for FcRn ORCIDs: 0000-0003-1783-7002 (N.B.); 0000-0002-2139-4171 (H.W.); 0000-0003- binding, accelerating the catabolism of, for example, patho- 0158-1356 (G.P.). genic autoantibodies. Received for publication August 7, 2015. Accepted for publication November 17, IgGs consist of four subclasses (IgG1 to IgG4), but IgG1 is the 2015. most highly represented among the Ab-based biologics approved This work was supported by a public grant overseen by the French National Research Agency as part of the “Investissements d’Avenir” program (reference: ANR-10- for sale around the world. Apart from the sequence diversity of the LABX-53-01) and by support from the European Regional Development Fund variable domains, Ab constant regions are identical among IgG1, (FEDER Grant Agreement: Presage 4940-37478; OutExFon). Measurement of with the exception of the natural genetic markers (Gms) or infliximab serum concentrations was carried out within the Centre Pilote de Suivi Biologique des Anticorps The´rapeutiques (CePiBAc) platform. CePiBAc is cofinanced polymorphisms called allotypes. The IgG1 H chain may express by the European Union. Europe is committed to supporting the Region Centre with FEDER. G1m1 [or G1m(a)], G1m2 [or G1m(x)], G1m3 [or G1m(f)], and Address correspondence and reprint requests to Dr. Vale´rie Gouilleux-Gruart, Uni- G1m17 [or Gm(z)] as the main allotypes. The allotypes are versite´ François Rabelais de Tours, CNRS UMR7292, 10 Boulevard Tonnelle´, 37032 inherited in a codominant Mendelian way, and various sets of Tours, France. E-mail address: [email protected] combinations are found in African, white, and Mongoloid pop- The online version of this article contains supplemental material. ulations (23, 24). Although the G1m3 and G1m17 allotypes are Abbreviations used in this article: FcRn, neonatal FcR; Gm, genetic marker; G1m-1, absence of G1m1; G1m3,-1, G1m3 with no G1m1; hFcRn, human FcRn; DhFcRn, located at the same position in the CH1 domain, the G1m1 truncated human FcRn; JurkatDhFcRn, DhFcRn-expressing Jurkat; PK, pharmacoki- counterpart corresponding to the absence of G1m1 [sometimes AF488 netics; rituximab , rituximab conjugated to Alexa Fluor 488 fluorescent dye; called non(a) or nG1m1 or G1m-1] in the CH3 domain cannot be SPR, surface plasmon resonance. detected serologically because its sequence is also present in other Copyright Ó 2015 by The American Association of Immunologists, Inc. 0022-1767/15/$30.00 IgG isotypes. Therefore, it is called an isoallotype (26). G1m-1 is

www.jimmunol.org/cgi/doi/10.4049/jimmunol.1501780 2 FcRn BINDING OF IgG1 ALLOTYPES typically found in conjunction with the presence of G1m3 in and natalizumab (Biogen). All were purchased from the respective phar- whites (26). Consequently, the main combinations of G1m allo- maceutical companies. Adalimumab variants were developed by InvivoGen types in this population are G1m17,1 and G1m3 with no G1m1 (Toulouse, France) using pFUSE-CHIg plasmids expressing the C region of the H chain of the human IgG1 allotypes. The four possible combinations of (G1m3,-1). The latter is more frequent (0.65–0.85) than G1m17,1 G1m17 and G1m3 allotypes were generated: G1m17,1; G1m17,-1; G1m3,1; (0.15–0.35) (23). and G1m3,-1. These variants bore no other allotype, such as G1m2 or the IgG3 allotypes [G3m15,16 (s,t)] are located within the FcRn G1m27 and G1m28 “supernumerary” allotypes, described in African binding site and are known to influence IgG3 binding to FcRn, populations (33). Consequently, these variants only differ at aa positions 120(K or R) of CH1, 12(D or E) and 14(L or M) of CH3. IgG3 half-life, and transport (27). In contrast, all of the classical G1m allotypes (G1m1, G1m17, and G1m3) are located outside of Cell lines the FcRn binding site. Nevertheless, previous studies using dif- Truncated human FcRn (DhFcRn)-expressing Jurkat (JurkatDFcRn) cells ferent human i.v. Ig preparations administered to immunodeficient were obtained by stable transfection of a Jurkat cell line with a plasmid patients and containing variable amounts of IgG1 allotypes sug- containing the human FcRn (hFcRn) sequence deleted from 99 nt corre- gest a differential IgG1 half-life according to the G1m patient sponding to the 33 aa in the C terminus of the protein (34). Plasmids b allotypes (28). Therefore, this study was undertaken to evaluate containing the hFcRn (full-length) and human 2-microglobulin–coding sequences were also transfected into the MDCKII cell line. The cells were the influence of the IgG1 allotypes exposed on patient IgG on the expanded for 24 h before G418 selection (1 mg/ml for Jurkat cells and pharmacokinetics (PK) of a mAb with a given allotype. In a cohort 0.4 mg/ml for MDCKII cells). The cell lines were maintained in RPMI of patients with spondyloarthritis that was treated with infliximab 1640 culture medium, supplemented with 10% heat-inactivated FCS, that bears the G1m17,1 allotypes, we found an increase in infliximab 2mML-glutamine, and G418. half-life, associated with a lower clearance, in patients expressing Transcytosis analysis the opposite G1m3,-1 allotype. We then demonstrated that this Downloaded from MDCKII cells expressing or not expressing hFcRn and human b2- phenomenon is linked to a difference in mAb FcRn binding, microglobulin were cultured in 24 Transwell plates (0.4 mm pore size; which, in turn, is related to allotypes. Costar; Corning) in RPMI 1640 buffer containing 10% FCS until conflu- ence. After 72 h and a cell wash in HBSS, the medium on the basal side of Materials and Methods each Transwell was replaced by HBSS adjusted to pH 6 with MES for Patients 20 min, whereas the upper compartments were incubated with HBSS (pH 7.4). Following this and after two washes of the basal and apical sides http://www.jimmunol.org/ This study was carried out in a cohort of 102 patients with spondyloarthritis of the wells with the corresponding HBSS, 300 ml HBSS (pH 6) containing that was treated in the Rheumatology Department of the University Hospital infliximab (3 mg/ml), in the presence or absence of the different mAbs to of Tours, Tours, France. The study protocol was in accordance with the be tested (200 mg/ml), was added to the lower compartment, whereas guidelines of the French Society of Rheumatology for the use of infliximab 300 ml HBSS (pH 7.4) was added to the upper side. After a 2-h incubation (29). All patients gave informed consent for IGHG1 genotyping. Patients period at 37˚C, the medium in the upper compartment was harvested to were treated with 5-mg/kg infusions of infliximab at wk 0, 2, 6, and 14; measure infliximab, using a validated ELISA assay (30). Each mAb was blood samples were collected per routine procedures prior to infusion at tested in duplicate. wk 2, 6, and 14. The infliximab concentration in these samples was measured using a validated ELISA technique (30). For a given patient, Flow cytometry binding analysis infliximab concentrations were excluded from the analyses if Abs to Rituximab conjugated to Alexa Fluor 488 fluorescent dye (rituximabAF488) infliximab were detected. was achieved using the Ab labeling kit (Invitrogen, Cergy-Pontoise, by guest on October 2, 2021 AF488 DNA extraction and IGHG1 CH1 359G/A polymorphism analysis France), according to the manufacturer’s instructions. Rituximab was used at 1 mg/ml in competition with either unlabeled rituximab, or DNA was extracted using the QIAamp DNA Blood Midi kit, according to cetuximab, or infliximab, or different variants of adalimumab at 1–100- the manufacturer’s instructions (QIAGEN). A genotyping method was fold rituximabAF488 concentrations. Jurkat and JurkatDFcRn cells were developed and validated in our laboratory to determine, at the IGHG1 gene mixed in a 1:1 ratio. The cells (1 3 105) were suspended in HBSS adjusted level, the antithetical (i.e., mutually exclusive) G1m3 and G1m17 allotypes to pH 6 with MES and incubated with rituximabAF488 and the different (aa 120 of CH1 according to IMGT, the international immunogenetics mAbs at various concentrations. After 30 min at 4˚C, the fluorescence database information system of unique numbering). Nucleotide 359G intensity was measured by flow cytometry (Coulter). Results were corresponds to the G1m3 allotype, and nucleotide 359A corresponds to the expressed as the percentage of rituximabAF488 binding inhibition on G1m17 allotype. This multiplex allele-specific PCR was published in JurkatDhFcRn cells. The point with rituximabAF488 alone was considered detail by Magdelaine-Beuzelin et al. (31). 100% binding for the determination of the percentage of inhibition binding. Untransfected Jurkat cells were used to evaluate Ab nonspecific PK analysis binding. Infliximab serum concentrations were analyzed by population PK modeling Recombinant hFcRn using Monolix 4.3.2 (Lixoft, Orsay, France). A one-compartment model with a first-order elimination rate constant best described the data. The The hFcRn coding sequence (cDNA clone MGC:1506 IMAGE:3163446), interindividual and residual variabilities of the PK parameters were de- deleted from its transmembrane and intracytoplasmic domain, was inserted scribed using exponential and proportional models, respectively. Because into the pCMV6 Kan/Neo plasmid to obtain pCMV6hFcRn. A histidine they are usually reported to influence the PK of therapeutic mAbs (32), body tag was introduced into the plasmid, resulting in pCMV6hFcRn-His. weight and sex were tested as covariates using power functions. Body The pCMV-SPORT6 plasmid encoding the human b2-microglobulin weight was centered on the median. The IGHG1 genotype was tested as a (NM_004048.2) was obtained from OriGene. The two plasmids dichotomous covariate and coded as A/A versus G carriers (heterozygotes (pCMV6hFcRn-His and pCMV-SPORT6 b2-microglobulin) were trans- and G/G homozygotes). Covariates were tested using likelihood ratio tests. fected into human HEK293 cells to produce soluble recombinant hFcRn. From pairs of nested models (i.e., models with versus without covariates), Large-scale transient transfection FreeStyle MAX Expression Systems the difference in their 22 ln-likelihood was tested using a x2 test. Co- (Invitrogen) were used, according to the manufacturer’s protocol, to drive variates were selected using forward and backward procedures. A covar- soluble hFcRn-His production. The cell culture supernatants containing the iate was considered significant if the corresponding p value was , 0.02. hFcRn-His secreted fusion protein were harvested, filtered, and stored at The goodness-of-fit was assessed by plotting population-predicted and 220˚C. Recombinant hFcRn was purified using a Nickel-IMAC resin individually predicted concentration versus observed concentration, indi- (HisPur Ni-NTA chromatography cartridge; Thermo Scientific) and an vidual weighted residuals versus individually predicted concentrations, and AKTA purifier instrument. normalized prediction distribution errors versus Gaussian distribution. Surface plasmon resonance Abs and reagents Real-time surface plasmon resonance (SPR) analyses were performed on a The following mAbs were used: infliximab (MSD), rituximab and trastuzumab BIA3000 apparatus (GE Healthcare) at 25˚C, at a flow rate of 50 ml/min in (Roche), cetuximab (Merck), adalimumab (AbbVie), panitumumab (Amgen), 10 mM PBS buffer (pH 6) and 0.005% P20 surfactant (GE Healthcare). The Journal of Immunology 3 hFcRn was covalently immobilized on a CM5 sensor chip (GE Healthcare) Table I. Patient characteristics at baseline (n = 102) at a relatively low density (500 RU) using EDC/NHS activation, according to the manufacturer’s instructions. The therapeutic mAbs were injected at Characteristics Value concentrations ranging from 1.6 to 100 nM for 180 s over an FcRn-coated flow cell and a control flow cell; the latter had the same chemical treatment Females (n [%]) 66 (65) but lacked FcRn. After a dissociation step of 400 s with running buffer, the Age (y, median [range]) 44.7 [18.2–75.7] sensor surfaces were regenerated using two pulses of PBS buffer (pH 7.4). Body weight (kg, median [range]) 73 [38–125] All curves were evaluated after double-referencing method data (35) using IGHG1 (359G/A polymorphism) a two-noninteracting binding sites model (BIAevaluation 4.2), according to AA - AG - GG (n [%]) 9 (9) - 29 (30) - 64 (65) previous work (36–39). Only the KD of the high-affinity site was consid- G1m17/17 - G1m17/3 - G1m3/3 ered for the correlation curve. To compare the binding kinetics of the different allotypes (50 nM) on immobilized FcRn, we measured responses (in RU) at three time points on the curves, 180, 200, and 580 s, designated as binding, stability 1, and stability 2, respectively. Data were read directly observed genotype frequencies, indicating that this population was on the curve to be independent of fitting. Kinetic and affinity constants of in Hardy–Weinberg equilibrium. Moreover, the genotype and al- allotypes G1m17,1 and G1m3,1 were evaluated at concentrations ranging lele frequencies were similar to our control population (31). from 6.25 to 200 nM and analyzed using a two-noninteracting binding sites A satisfactory description of infliximab concentrations was model (BIAevaluation 4.2). The stability of the coated surface was obtained using the PK model (Supplemental Table I). A total of checked by injection of rituximab (50 nM) as a positive control at the beginning and end of the series of experiments. Experiments were per- 102 patients was analyzed; however, because Abs to infliximab formed in triplicate and replicated on freshly immobilized FcRn each time. were detected at week 14 in eight patients, their corresponding infliximab concentrations were not accounted for in the model. Statistical analysis

Mean infliximab half-life (t1/2) was 15.5 d; its volume of distri- Downloaded from For flow cytometry and SPR, data are the mean and SD of at least three bution increased significantly with increasing body weight and independent experiments. The Mann–Whitney U test was used to deter- was higher in males than in females. In addition, the IGHG1 CH1 mine significant differences. For SPR, binding, stability 1, and stability 359G/A genotype was significantly associated with infliximab- 2 values of the G1m3,-1 or G1m17,-1 adalimumab variant were compared with the G1m3,1 variant, using the G1m17,1 variant as the reference estimated t1/2, with its value less in homozygous A/A (G1m17 (100%). Statistical analysis was performed using GraphPad Prism 5. All allotype) patients than in G carriers (t1/2 = 11.7 d and elimination tests were bilateral, and the a risk was set at 5%. rate constant = 0.059/d versus 16.1 d and 0.043/d, respectively, p = http://www.jimmunol.org/ 0.0082, Fig. 2). Results Considering the fact that infliximab bears the G1m17,1 allo- Infliximab PK parameters are influenced by G1m patient types, we hypothesized that if G1m3,-1 and G1m17,1 combinations allotypes of allotypes have different affinities for FcRn, competition for FcRn binding between endogenous IgG and the therapeutic mAb could To evaluate the influence of the G1m patient allotypes on infliximab occur. This would explain the differences observed in PK param- PK parameters, we investigated a cohort of patients being treated eters. If G1m17,1 IgG1 has a higher affinity for FcRn, infliximab for spondyloarthritis. Because the G1m17 and G1m1 allotypes are (G1m17,1) also could compete better with the endogenous G1m3,-1 in perfect linkage disequilibrium in Europeans (26), the determi-

IgG1 of the patient, leading to a better salvage of infliximab and by guest on October 2, 2021 nation of the mutually exclusive G1m3 and G1m17 allotypes al- an increase in its half-life in these particular patients. Conversely, lows the two main allotype combinations (G1m17,1 and G1m3,-1) a shorter infliximab half-life should be observed in patients ho- to be distinguished (Fig. 1). This determination was done at the mozygous for G1m17,1 allotypes, as was measured in our cohort IGHG1 gene level by genotyping the 359G/A nucleotide poly- of patients. morphism (31): G1m3 corresponding to 359G and G1m17 corresponding to 359A. Patient characteristics and genotypes are G1m17 IgG1 is transcytosed more efficiently than G1m3,-1 detailed in Table I. The IGHG1 CH1 359G/A genotype frequen- IgG1 by FcRn cies deduced from allele frequencies were very similar to the To extend our observations, we tested our hypothesis in a transcytosis model by evaluating the transport of mAbs through an adherent MDCKII cell monolayer expressing hFcRn. Various IgG1 therapeutic mAbs bearing different allotypes and mAbs from other

FIGURE 1. G1m allotype residues (following IMGT and EU numbering) positioned on a model of human IgG1 (25). The cysteines involved in the two interchain disulfide bridges are shown in black. The allotype no- menclature used in this article is in bold type. The polymorphic residues corresponding to the studied G1m allotypes are shown in red. Those in pink represent the G1m2 allotype (i.e., replacement of the alanine in position 110 [IMGT]/431 [EU] by a glycine); when present, G1m2 is asso- FIGURE 2. Estimated infliximab half-life according to IGHG1 geno- ciated with the G1m17,1 allotype combination. Residues constituting the type. For each genotype, the interindividual distribution of half-life is FcRn binding site are shown as light green (loss of binding) or dark green shown as box-and-whisker plots that represent, from bottom to top, the 5th, (binding improvement) (8); they do not overlap allotypic variations. All of 25th, 50th, 75th, and 95th percentiles. The half-life was significantly lower these residues were superimposed on the IgG1 model (25) using PyMOL in homozygous A/A patients than in G carriers (p = 0.0082), but it was not Molecular Graphics System, version 1.7.4 (Schro¨dinger). significantly different between G/A and G/G patients. 4 FcRn BINDING OF IgG1 ALLOTYPES

Table II. Therapeutic mAbs with the transcytosis results, improved inhibition of rituximabAF488 binding was observed in the presence of G1m17,1 mAbs, such as G1m G1m infliximab and rituximab, compared with the presence of the mAb IgG Subclass Allotype (CH1) Allotype (CH3) G1m3,-1 mAb cetuximab. Rituximab IgG1 17 1 Infliximab IgG1 17 1 G1m3,-1 and G1m17 IgG1 bind differently to FcRn a Trastuzumab IgG1 17 21 Finally, to confirm our results, we used SPR to measure the FcRn Cetuximab IgG1 3 21 Panitumumab IgG2 binding of therapeutic IgG mAbs. The binding kinetics of different Natalizumab IgG4 IgG1 were measured on immobilized FcRn, in parallel with mAbs from other IgG subclasses, which acted as controls, in a two- aG1m1 in trastuzumab has been engineered (46). noninteracting binding sites model (Supplemental Fig. 2). Higher KD (high-affinity binding site) values were found for G1m3,-1 IgG subclasses (Table II) were evaluated in this model to compare IgG1 (cetuximab, 27.6 nM) than for G1m17 IgG1 mAb their ability to be transported through FcRn-expressing cells, an (G1m17,-1 trastuzumab, 8.0 nM and G1m17,1 rituximab, 2.8 nM), essential step for IgG biodistribution. The evaluation was con- with the values for the former being close to those for IgG2 ducted as a competitive experiment, measuring infliximab trans- (panitumumab, 27.6 nM) and IgG4 (natalizumab, 13.3 nM). In cytosis in the presence of an excess of other selected mAbs Fig. 5, the KD values were plotted against the percentage of (Fig. 3). IgG2 (panitumumab) and IgG4 (natalizumab) inhibited transcytosis inhibition evaluated from Fig. 3. A correlation was infliximab transcytosis by 0.3 and 13.7%, respectively, whereas a found between FcRn binding and the percentage inhibition of IgG difference in the competition of infliximab was observed with transcytosis, showing that the mAbs that demonstrate the lowest Downloaded from IgG1, G1m3,-1 IgG1 being less efficient than G1m17 bearing affinities for FcRn were those presenting the lowest levels of IgG1 to compete with the G1m17,1 infliximab (18.5% for transcytosis inhibition. This group included the G1m3,-1 mAb, as cetuximab versus 45.8 and 49.3% for trastuzumab and rituximab, well as IgG2 and IgG4. respectively, Fig. 2). Associated with the G1m17 allotype in the G1m3,1 IgG1: the best allotype combination for FcRn binding CH1, the G1m1 allotype did not seem to influence the binding because rituximab (G1m17,1) and trastuzumab (G1m17,-1), To avoid any participation of the variable domain of therapeutic http://www.jimmunol.org/ which differ in allotype at this position in the CH3, inhibited mAbs or the influence of other parameters, such as the mAb buffer infliximab transcytosis to a similar degree (Fig. 3). Importantly, no or the cell system/production method used for commercialized infliximab transcytosis was observed with untransfected MDCK mAbs, we tested IgG1 variants carrying the four possible com- cells (data not shown), underlining the hFcRn dependence of the binations of G1m3 and G1m17 allotypes but with the same VH and phenomenon. VL domain. InvivoGen generated G1m3,-1, G1m3,1, G1m17,-1, and G1m17,1 variants of adalimumab, an anti-TNF therapeutic G1m17,1 IgG1 binds better to cell-expressed FcRn than mAb. These mAbs were first analyzed by SPR. The kinetics ob- G1m3,-1 IgG1 tained with the different allotype variants were very similar To evaluate the binding of mAb to FcRn more precisely, we used a (Fig. 6A). Therefore, to assure independence from the quality of by guest on October 2, 2021 cellular assay consisting of Jurkat cells transfected with hFcRn fitting, we compared three significant points (RU) on the response truncated at its C terminus, leading to FcRn expression on the cell curve that corresponded to the binding at the end of mAb injection membrane. Analyses were performed using flow cytometry at pH 6. and the mAb–FcRn complex stability after a short and then a long The results are presented in Fig. 4 as the percentage inhibition of dissociation time with running buffer. The values obtained with rituximabAF488 binding. This assay was validated using rituximab G1m17,1 were used as reference (100%). G1m3,-1 showed a with His at aa position 115 in the CH3 mutated in Ala (H115A) in decrease in binding and stability levels of 15% (Fig. 6B). These the Fc portion, leading to partial abolishment of FcRn binding (40). data indicate that the results obtained with commercial G1m17,1 This variant was unable to compete efficiently with rituximabAF488, (infliximab and rituximab) and G1m3,-1 (cetuximab) mAbs were as assessed by flow cytometry (Supplemental Fig. 1). In accordance similar to our adalimumab variants. Unexpectedly, the G1m3,1 adalimumab variant was the most effective in terms of FcRn binding (+10%) and complex stability (+40%). To reinforce this 24 result, binding kinetics of the G1m17,1 and G1m3,1 variants were

20 Infliximab alone 16 + rituximab + trastuzumab 12 + cetuximab 8 + panitumumab + natalizumab Transcytosed

Infliximab (ng/mL) 4 0 Basal side FIGURE 3. Transcytosis of therapeutic mAbs in MDCKII cells expressing hFcRn and b2-microglobulin. Infliximab (3 mg/ml) was added to FIGURE 4. Flow cytometry analysis of mAb binding to membrane- the basal side of the Transwells, with or without 200 mg of rituximab (IgG1), expressed FcRn. Jurkat and JurkatDhFcRn cells (v/v mix) were incubated trastuzumab (IgG1), cetuximab (IgG1), panitumumab (IgG2), or natalizumab at pH 6 with 1 mg/ml rituximabAF488 in the presence of 0, 1, 10, 25, 50, or (IgG4). This was carried out in duplicate for each condition. After a 2-h 100 mg/ml therapeutic mAbs. Fluorescence intensities were measured by flow incubation period, infliximab concentrations in the apical-side medium cytometry at 4˚C after 30 min. RituximabAF488 alone was considered as 100% were measured by ELISA. These results are representative of those of four binding. Results are presented as percentage inhibition of rituximabAF488 independent experiments, in which each mAb was tested at least twice. binding (n = 4). **p # 0.01, ***p # 0.001 versus cetuximab. The Journal of Immunology 5

infliximab increase its elimination rate. Therefore, the influence of IGHG1 genotype on infliximab elimination that was observed is clearly independent from immunization against infliximab. We then hypothesized that this phenomenon was due to variable affinity of IgG1 allotypes for FcRn, an essential receptor in mAb PK. By using cellular assays to analyze FcRn binding, we found that the binding of G1m17,1 mAbs to FcRn was greater than that of G1m3,-1 IgG1. More precisely, mAbs bearing the G1m17 allotype are better competitors than the G1m3,-1 mAb cetuximab in FcRn-dependent transcytosis experiments and JurkatDFcRn FIGURE 5. Correlation between mAb affinities for FcRn measured by binding assays, providing a straightforward explanation for our SPR and mAb transcytosis. The binding of rituximab (1), trastuzumab (2), clinical findings. Indeed, in G1m17,1-homozygous patients, all of cetuximab (3), natalizumab (4), and panitumumab (5) to FcRn was eval- the endogenous IgG1 act as strong competitors for infliximab, uated using SPR. KD values are plotted on the y-axis, and percentages of infliximab transcytosis inhibition are on the x-axis. which shares the same G1m allotype. In G1m3,-1–homozygous patients, infliximab outmatches all of the endogenous IgG1 and, thus, benefits more from FcRn recycling, resulting in its longer compared using SPR. The G1m3,1 variant had a greater affinity half-life. The situation is more complex for heterozygous patients, for FcRn than did the G1m17,1 variant (KD = 1.6 and 5.3 nM, in whom infliximab PK is not intermediate (as would be ex- respectively) (Supplemental Fig. 3). These results were strength- pected in the case of an allele dose effect) but similar to what is ened by experiments using JurkatDhFcRn cells, which showed observed in G1m3,-1–homozygous patients. Both alleles are Downloaded from AF488 greater inhibition of rituximab binding to FcRn at pH 6 by presumably expressed in a codominant manner: half of the en- G1m3,1, G1m17,1, and G1m17,-1 adalimumab than by G1m3,-1 dogenous IgG1 produced is G1m17,1, whereas the other half is adalimumab, with G1m3,1 adalimumab being the most efficient G1m3,-1. Therefore, endogenous G1m17,1 IgG1 might be present (Fig. 7). in an insufficient quantity to saturate the FcRn pool, leaving it to recycle infliximab without any constraint, exactly like it does in Discussion G1m3,-1 homozygotes. http://www.jimmunol.org/ In this study, we showed that the IGHG1 gene polymorphism, Because the competitive experiments that we performed using which leads to the expression of different allotypes on patient cell-based assays measured Ab binding to FcRn in an indirect IgG1, influences the PK of therapeutic mAbs. We found a higher manner, we used SPR to measure the direct binding on immobilized elimination rate, and consequently a shorter half-life, of G1m17,1- FcRn. The findings allowed two groups of mAbs to be identified, expressing infliximab in G1m17,1-homozygous patients. Our each one characterized by mAbs with closely related KD values and analysis excluded infliximab concentrations in patients immunized behaving identically in transcytosis. The results also showed that against infliximab to avoid a confounding factor, because Abs to the group of mAbs with the lower binding to FcRn, as assessed by by guest on October 2, 2021 A Binding (175s) G1m3,1 80 G1m3,-1 Stability 1 (200s) G1m17,1 G1m17,-1 60

40 Stability 2 (580s)

20 FIGURE 6. SPR analysis of adalimumab variant binding (RU) Response to FcRn. (A) Sensorgram of allotype variants (G1m17,1; G1m3,-1; G1m17,-1; and G1m3,1) of adalimumab. To 0 compare binding kinetics, responses (in RU) were measured at three time points on the curves: 180, 200, and 580 s 0 200 400 600 designated binding, stability 1, and stability 2, respectively. B Time (s) (B) Comparison of binding, stability 1, and stability 2 values of the four variants considering G1m17,1 adalimumab % as the reference (100%). Replicate experiments were per- 180 formed on freshly immobilized FcRn (n =4).*p # 0.05, 160 **p # 0.01 versus G1m3,1 variant. 140 120 100 * ** ** * ** * 80 % binding 60 % stability 1 40 20 % stability 2 0 G1m17,1 G1m3,-1 G1m17,-1 G1m3,1 Allotype variants of adalimumab 6 FcRn BINDING OF IgG1 ALLOTYPES

proteins, such as etanercept, to FcRn can also be explained by the absence of Fab and/or the disturbed hinge region (48). The k or l L chains also were shown to influence binding kinetics to murine FcRn (43). The L chains were not involved in the current study because all of the IgG1 tested displayed the same Km3 L chain (26). G1m3 and G1m17 allotypes are located in the Fab, far away from FcRn binding. In light of the data published by Schoch et al. (44), it cannot be excluded that changes, such as the positively charged side chain R/K aa at position 120 in the CH1, may modify the charge of the Fab and affect electrostatic interaction with FIGURE 7. FcRn binding analysis of adalimumab variants by flow FcRn/negatively charged plasma membrane (44). This phenome- cytometry. Jurkat and JurkatDhFcRn cells (v/v mix) were incubated (pH 6) non could also explain why the difference between allotypes was AF488 with 1 mg/ml rituximab in the presence of the four allotypes (G1m17,1; greater in cellular assays than in SPR analyses. This hypothesis is G1m3,-1; G1m17,-1; and G1m3,1) and IgA variants of adalimumab at also in line with the work of Warwicker et al. (49), who suggested m 0, 1, 10, 25, 50, and 100 g/ml. Fluorescence intensities were measured that, in terms of positive charge, Lys (K) is more effective than by flow cytometry at 4˚C after 30 min. RituximabAF488 alone was con- Arg (R) in promoting solubility by supercharging. More surprising sidered as 100% binding. Results are presented as percentage inhibition of rituximabAF488 binding (n =4).*p # 0.05 versus G1m17,1 variant. is the influence of the G1m1 allotype (D/E12 and L/M14 in the CH3 for presence/absence of G1m1) on FcRn binding. This allotype, located close to the FcRn binding site (Fig. 1), may affect SPR, competes less with the transcytosis of G1m17,1 infliximab. FcRn binding but unexpectedly impact FcRn binding, depending on Downloaded from In this group, we identified the G1m3,-1 mAb cetuximab, as well the G1m3 or G1m17 allotype background. There is no straight- as IgG2 and IgG4 mAbs. For the SPR analyses, we immobilized forward explanation for this, although modifications in charge FcRn and used a two-noninteracting binding sites fitting model, distribution acting on conformational flexibility of the molecule or according to previous work (36–39). A recent study showed that on protein solubility could be evoked (44, 49). These data show, for affinity values could be influenced by the immobilization protocol the first time to our knowledge, that a residue in the Fab portion http://www.jimmunol.org/ and the nature of the chip (41). Therefore, all KD values obtained (CH1) can modulate FcRn binding in association with residues under the same conditions should be considered relative values present in the CH3 domain, although neither has an effect by itself. and should be interpreted only in relation to each other. However, The G1m3,1 combination of allotypes is primarily encountered as a result of the debate surrounding its evaluation, SPR was not in Mongoloids (24, 26), although no genomic sequence has been the only approach used to study FcRn binding in the present work. described (see IMGT). If we apply the same reasoning to the Indeed, membrane-expressed FcRn, possibly bridged by Fc mol- competition for FcRn binding between endogenous IgG1 allotypes ecules to form an oligomeric ribbon, could differ from the and therapeutics mAbs, we might assume that the therapeutic half- recombinant hFcRn coated on the sensorchips (42, 43). This might life of mAb in patients with G1m3,1 IgG1 would be low compared explain why the difference in FcRn binding between the G1m17,1 with patients with other combinations of allotypes. This would by guest on October 2, 2021 and G1m3,-1 allotypes was not as obvious as that observed with hold true even for therapeutic IgG2 or IgG4. Conversely, a ther- FcRn-expressing Jurkat cells, even though the results all point in apeutic mAb designed with a G1m3,1 format would have a longer the same direction. This highlights once again that some in vivo half-life compared with other IgG1 allotypes, regardless of the properties of mAbs may be missed by SPR-based analyses. patient IGHG1 genotype. To avoid any possible participation of the VH and VL sequences The influence of IgG1 allotypes on FcRn binding underlines or the variable mAb glycosylation (44, 45), we designed and once again the need to take into account endogenous IgG1 allotypes produced IgG variants bearing the four possible allotype combi- when evaluating new molecules and comparing their efficacy. nations (G1m17 or G1m3 in the CH1 mixed with G1m1 or G1m-1 Moreover, careful attention must be paid to G1m allotypes when in the CH3). Results similar to those described with therapeutic developing new IgG1-based therapeutics. mAbs were obtained when using adalimumab variants, indicating that the variable domain and other confounding factors, such as Acknowledgments mAb pharmaceutical formulation, did not contribute to the dif- We thank Marc Ohresser and Jeremy Pottier for advice and analysis of the ferences in the binding of G1m variants to FcRn. More surpris- results and Marie-Noe¨lle Marson, Anne-Claire Duveau, Caroline Brochon, ingly, we identified the G1m3,1 variant as the best combination in Celine Desvignes, and Corinne Henriquet for technical assistance. We are terms of FcRn binding; the presence of G1m1 plus G1m3 greatly also grateful to all of the nurses and the medical staff from the Rheuma- increased binding to FcRn. Because of the limits of the SPR ap- tology Department. We thank Alexandra Farrell for linguistic revision and proach, as discussed previously, the four variants were also tested editing services. on JurkatDFcRn cells using a rituximabAF488 inhibition assay. Again, the G1m3,1 variant was by far the best FcRn binder. No Disclosures specific difference was observed for the G1m17,-1 variant of The authors have no financial conflicts of interest. adalimumab, in accordance with the data obtained for trastuzumab that were generated by engineering (46). References The superior binding of G1m3,1 to FcRn, as well as the lower 1. Brambell, F. W., W. A. Hemmings, and I. G. Morris. 1964. A theoretical model binding of G1m3,-1 compared with G1m17,1 or G1m17,-1, reveals of gamma-globulin catabolism. Nature 203: 1352–1354. that amino acid modifications remote from the FcRn binding site 2. Simister, N. E., and K. E. Mostov. 1989. An Fc receptor structurally related to can modulate IgG binding to FcRn. These data reinforce recent MHC class I antigens. Nature 337: 184–187. 3.Raghavan,M.,V.R.Bonagura,S.L.Morrison,andP.J.Bjorkman.1995. studies by Monnet et al. (47) and Schoch et al. (44), who showed Analysis of the pH dependence of the neonatal Fc receptor/immunoglobulin G that mutations located outside of the FcRn binding site (e.g., in the interaction using antibody and receptor variants. Biochemistry 34: 14649– 14657. Fc hinge region or in the variable domain, respectively) may en- 4. Chaudhury, C., S. Mehnaz, J. M. Robinson, W. L. Hayton, D. K. Pearl, hance FcRn binding. The lower binding of Fc-containing fusion D. C. Roopenian, and C. L. Anderson. 2003. The major histocompatibility The Journal of Immunology 7

complex-related Fc receptor for IgG (FcRn) binds albumin and prolongs its 28. Oxelius, V. A., and M. M. Eibl. 1996. Different Gm allotype amounts in human lifespan. J. Exp. Med. 197: 315–322. intravenous immunoglobulin (IVIG) preparations; survival of foreign Gm allo- 5. Andersen, J. T., B. Dalhus, J. Cameron, M. B. Daba, A. Plumridge, L. Evans, types in immunodeficient patients. Clin. Exp. Immunol. 106: 203–207. S. O. Brennan, K. S. Gunnarsen, M. Bjøra˚s, D. Sleep, and I. Sandlie. 2012. 29. Pham, T., B. Fautrel, E. Dernis, P. Goupille, F. Guillemin, X. Le Loe¨t, P. Ravaud, Structure-based mutagenesis reveals the albumin-binding site of the neonatal Fc P. Claudepierre, C. Miceli-Richard, M. de Bandt, et al; Club Rhumatismes et receptor. Nat. Commun. 3: 610 doi:10.1038/ncomms1607. Inflammation (CRI/SFR). 2007. Recommendations of the French Society for 6. Story, C. M., J. E. Mikulska, and N. E. Simister. 1994. A major histocompati- Rheumatology regarding TNFalpha antagonist therapy in patients with ankylosing bility complex class I-like Fc receptor cloned from human placenta: possible role spondylitis or psoriatic arthritis: 2007 update. Joint Bone Spine 74: 638–646. in transfer of immunoglobulin G from mother to fetus. J. Exp. Med. 180: 2377– 30. Ternant, D., D. Mulleman, D. Degenne, S. Willot, J.-M. Guillaumin, H. Watier, 2381. P. Goupille, and G. Paintaud. 2006. An enzyme-linked immunosorbent assay for 7. Junghans, R. P., and C. L. Anderson. 1996. The protection receptor for IgG therapeutic drug monitoring of infliximab. Ther. Drug Monit. 28: 169–174. catabolism is the beta2-microglobulin-containing neonatal intestinal transport 31. Magdelaine-Beuzelin, C., S. Vermeire, M. Goodall, F. Baert, M. Noman, receptor. Proc. Natl. Acad. Sci. USA 93: 5512–5516. G. V. Assche, M. Ohresser, D. Degenne, J.-M. Dugoujon, R. Jefferis, et al. 2009. 8. Roopenian, D. C., and S. Akilesh. 2007. FcRn: the neonatal Fc receptor comes of IgG1 heavy chain-coding gene polymorphism (G1m allotypes) and development age. Nat. Rev. Immunol. 7: 715–725. of antibodies-to-infliximab. Pharmacogenet. Genomics 19: 383–387. 9. Oganesyan, V., M. M. Damschroder, K. E. Cook, Q. Li, C. Gao, H. Wu, and 32. Dirks, N. L., and B. Meibohm. 2010. Population pharmacokinetics of therapeutic W. F. Dall’Acqua. 2014. Structural insights into neonatal Fc receptor-based monoclonal antibodies. Clin. Pharmacokinet. 49: 633–659. recycling mechanisms. J. Biol. Chem. 289: 7812–7824. 33. Lefranc, M.-P., and G. Lefranc. 2012. Human Gm, Km, and Am allotypes and 10. Ward, E. S., and R. J. Ober. 2009. Chapter 4: Multitasking by exploitation of their molecular characterization: a remarkable demonstration of polymorphism. intracellular transport functions the many faces of FcRn. Adv. Immunol. 103: 77– Methods Mol. Biol. 882: 635–680. 115. 34. Passot, C., N. Azzopardi, S. Renault, N. Baroukh, C. Arnoult, M. Ohresser, 11. Vidarsson, G., A. M. Stemerding, N. M. Stapleton, S. E. Spliethoff, H. Janssen, M. Boisdron-Celle, E. Gamelin, H. Watier, G. Paintaud, and V. Gouilleux- F. E. Rebers, M. de Haas, and J. G. van de Winkel. 2006. FcRn: an IgG receptor Gruart. 2013. Influence of FCGRT gene polymorphisms on pharmacokinetics of on phagocytes with a novel role in phagocytosis. Blood 108: 3573–3579. therapeutic antibodies. MAbs 5: 614–619. 12. Baker, K., T. Rath, M. Pyzik, and R. S. Blumberg. 2014. The Role of FcRn in 35. Morton, T. A., and D. G. Myszka. 1998. Kinetic analysis of macromolecular Antigen Presentation. Front. Immunol. 5: 408 doi:10.3389/fimmu.2014.00408. interactions using surface plasmon resonance biosensors. Methods Enzymol. 295: 13. Cervenak, J., B. Bender, Z. Schneider, M. Magna, B. V. Carstea, K. Liliom, 268–294. Downloaded from A. Erdei, Z. Bosze, and I. Kacskovics. 2011. Neonatal FcR overexpression 36. Vaughn, D. E., and P. J. Bjorkman. 1997. High-affinity binding of the neonatal Fc boosts humoral immune response in transgenic mice. J. Immunol. 186: 959–968. receptor to its IgG ligand requires receptor immobilization. Biochemistry 36: 14. Baker, K., T. Rath, M. B. Flak, J. C. Arthur, Z. Chen, J. N. Glickman, I. Zlobec, 9374–9380. E. Karamitopoulou, M. D. Stachler, R. D. Odze, et al. 2013a. Neonatal Fc re- 37. Martin, W. L., and P. J. Bjorkman. 1999. Characterization of the 2:1 complex ceptor expression in dendritic cells mediates protective immunity against colo- between the class I MHC-related Fc receptor and its Fc ligand in solution. rectal cancer. Immunity 39: 1095–1107. Biochemistry 38: 12639–12647. 15. Pyzik, M., T. Rath, W. I. Lencer, K. Baker, and R. S. Blumberg. 2015. FcRn: The 38. West, A. P., Jr., and P. J. Bjorkman. 2000. Crystal structure and immunoglobulin

Architect Behind the Immune and Nonimmune Functions of IgG and Albumin. G binding properties of the human major histocompatibility complex-related Fc http://www.jimmunol.org/ J. Immunol. 194: 4595–4603. receptor(,). Biochemistry 39: 9698–9708. 16. Shields, R. L., A. K. Namenuk, K. Hong, Y. G. Meng, J. Rae, J. Briggs, D. Xie, 39. Datta-Mannan, A., D. R. Witcher, Y. Tang, J. Watkins, W. Jiang, and J. Lai, A. Stadlen, B. Li, et al. 2001. High resolution mapping of the binding site V. J. Wroblewski. 2007. Humanized IgG1 variants with differential binding on human IgG1 for Fc gamma RI, Fc gamma RII, Fc gamma RIII, and FcRn and properties to the neonatal Fc receptor: relationship to pharmacokinetics in mice design of IgG1 variants with improved binding to the Fc gamma R. J. Biol. and primates. Drug Metab. Dispos. 35: 86–94. Chem. 276: 6591–6604. 40. Firan, M., R. Bawdon, C. Radu, R. J. Ober, D. Eaken, F. Antohe, V. Ghetie, and 17. Schmidt, M. M., S. A. Townson, A. J. Andreucci, B. M. King, E. B. Schirmer, E. S. Ward. 2001. The MHC class I-related receptor, FcRn, plays an essential role in A. J. Murillo, C. Dombrowski, A. W. Tisdale, P. A. Lowden, A. L. Masci, et al. the maternofetal transfer of gamma-globulin in humans. Int. Immunol. 13: 993–1002. 2013. Crystal structure of an HSA/FcRn complex reveals recycling by com- 41. Abdiche, Y. N., Y. A. Yeung, J. Chaparro-Riggers, I. Barman, P. Strop, petitive mimicry of HSA ligands at a pH-dependent hydrophobic interface. S. M. Chin, A. Pham, G. Bolton, D. McDonough, K. Lindquist, et al. 2015. The Structure 21: 1966–1978. neonatal Fc receptor (FcRn) binds independently to both sites of the IgG

18. Kuo, T. T., and V. G. Aveson. 2011. Neonatal Fc receptor and IgG-based ther- homodimer with identical affinity. MAbs 7: 331–343. by guest on October 2, 2021 apeutics. MAbs 3: 422–430. 42. Sańchez, L. M., D. M. Penny, and P. J. Bjorkman. 1999. Stoichiometry of the 19. Santagostino, E., C. Negrier, R. Klamroth, A. Tiede, I. Pabinger-Fasching, interaction between the major histocompatibility complex-related Fc receptor C. Voigt, I. Jacobs, and M. Morfini. 2012. Safety and pharmacokinetics of a and its Fc ligand. Biochemistry 38: 9471–9476. novel recombinant fusion protein linking coagulation factor IX with albumin 43. Gurbaxani, B., L. L. Dela Cruz, K. Chintalacharuvu, and S. L. Morrison. 2006. (rIX-FP) in hemophilia B patients. Blood 120: 2405–2411. Analysis of a family of antibodies with different half-lives in mice fails to find a 20. Zalevsky, J., A. K. Chamberlain, H. M. Horton, S. Karki, I. W. L. Leung, correlation between affinity for FcRn and serum half-life. Mol. Immunol. 43: T. J. Sproule, G. A. Lazar, D. C. Roopenian, and J. R. Desjarlais. 2010. Enhanced 1462–1473. antibody half-life improves in vivo activity. Nat. Biotechnol. 28: 157–159. 44. Schoch, A., H. Kettenberger, O. Mundigl, G. Winter, J. Engert, J. Heinrich, and 21. Ko, S.-Y., A. Pegu, R. S. Rudicell, Z. Y. Yang, M. G. Joyce, X. Chen, K. Wang, T. Emrich. 2015. Charge-mediated influence of the antibody variable domain on S. Bao, T. D. Kraemer, T. Rath, et al. 2014. Enhanced neonatal Fc receptor function FcRn-dependent pharmacokinetics. Proc. Natl. Acad. Sci. USA 112: 5997–6002. improves protection against primate SHIV infection. Nature 514: 642–645. 45. Souders, C. A., S. C. Nelson, Y. Wang, A. R. Crowley, M. S. Klempner, and 22. Ward, E. S., S. C. Devanaboyina, and R. J. Ober. 2015. Targeting FcRn for the W. Thomas, Jr. 2015. A novel in vitro assay to predict neonatal Fc receptor- modulation of antibody dynamics. Mol. Immunol. 67(2 Pt. A): 131–141 . mediated human IgG half-life. MAbs 7: 912–921. 23. Lefranc, G., and M.-P. Lefranc. Gm allotypes and Gm haplotypes. Allotypes. In 46. Carter, P., L. Presta, C. M. Gorman, J. B. Ridgway, D. Henner, W. L. Wong, IMGT Repertoire (IG and TR). IMGT, the international ImMunoGeneTics infor- A. M. Rowland, C. Kotts, M. E. Carver, and H. M. Shepard. 1992. Humanization mation system. Available at: http://www.imgt.org/IMGTrepertoire/Proteins/allotypes/ of an anti-p185HER2 antibody for human cancer therapy. Proc. Natl. Acad. Sci. human/IGH/IGHC/Hu_IGHCallotypes1.html Accessed: December 8, 2015. USA 89: 4285–4289. 24. Dugoujon, J.-M., S. Hazout, F. Loirat, B. Mourrieras, B. Crouau-Roy, and 47. Monnet, C., S. Jorieux, R. Urbain, N. Fournier, K. Bouayadi, C. De Romeuf, A. Sanchez-Mazas. 2004. GM haplotype diversity of 82 populations over the C. K. Behrens, A. Fontayne, and P. Mondon. 2015. Selection of IgG Variants world suggests a centrifugal model of human migrations. Am. J. Phys. Anthropol. with Increased FcRn Binding Using Random and Directed Mutagenesis: Impact 125: 175–192. on Effector Functions. Front. Immunol. 6: 39 doi:10.3389/fimmu.2015.00039. 25. Clark, M. R. 1997. IgG effector mechanisms. Chem. Immunol. 65: 88–110. 48. Suzuki, T., A. Ishii-Watabe, M. Tada, T. Kobayashi, T. Kanayasu-Toyoda, 26. Jefferis, R., and M.-P. Lefranc. 2009. Human immunoglobulin allotypes: pos- T. Kawanishi, and T. Yamaguchi. 2010. Importance of neonatal FcR in regu- sible implications for immunogenicity. MAbs 1: 332–338. lating the serum half-life of therapeutic proteins containing the Fc domain of 27. Stapleton, N. M., J. T. Andersen, A. M. Stemerding, S. P. Bjarnarson, human IgG1: a comparative study of the affinity of monoclonal antibodies and R. C. Verheul, J. Gerritsen, Y. Zhao, M. Kleijer, I. Sandlie, M. de Haas, et al. Fc-fusion proteins to human neonatal FcR. J. Immunol. 184: 1968–1976. 2011. Competition for FcRn-mediated transport gives rise to short half-life of 49. Warwicker, J., S. Charonis, and R. A. Curtis. 2014. Lysine and arginine content human IgG3 and offers therapeutic potential. Nat. Commun. 2: 599 doi:10.1038/ of proteins: computational analysis suggests a new tool for solubility design. ncomms1608. Mol. Pharm. 11: 294–303.