Supplemental material to this article can be found at: http://jpet.aspetjournals.org/content/suppl/2016/07/19/jpet.116.236075.DC1

1521-0103/359/1/37–44$25.00 http://dx.doi.org/10.1124/jpet.116.236075 THE JOURNAL OF PHARMACOLOGY AND EXPERIMENTAL THERAPEUTICS J Pharmacol Exp Ther 359:37–44, October 2016 Copyright ª 2016 by The American Society for Pharmacology and Experimental Therapeutics

Unexpected Potency Differences between B-Cell–Activating Factor (BAFF) Antagonist Antibodies against Various Forms of BAFF: Trimer, 60-Mer, and Membrane-Bound s

Amy M. Nicoletti, Cynthia Hess Kenny, Ashraf M. Khalil, Qi Pan, Kerry L. M. Ralph, Julie Ritchie, Sathyadevi Venkataramani, David H. Presky, Scott M. DeWire, and Scott R. Brodeur Immune Modulation and Biotherapeutics Discovery, Boehringer Ingelheim Pharmaceuticals, Inc., Ridgefield, Connecticut

Received June 20, 2016; accepted July 18, 2016 Downloaded from

ABSTRACT Therapeutic agents antagonizing B-cell–activating factor/B- human B-cell proliferation assay and in nuclear factor kB reporter lymphocyte stimulator (BAFF/BLyS) are currently in clinical assay systems in Chinese hamster ovary cells expressing BAFF development for autoimmune diseases; belimumab is the first receptors and transmembrane activator and calcium-modulator Food and Drug Administration–approved drug in more than and cyclophilin ligand interactor (TACI). In contrast to the mouse jpet.aspetjournals.org 50 years for the treatment of lupus. As a member of the tumor system, we find that BAFF trimer activates the human TACI necrosis factor superfamily, BAFF promotes B-cell survival and receptor. Further, we profiled the activities of two clinically ad- homeostasis and is overexpressed in patients with systemic vanced BAFF antagonist antibodies, belimumab and tabalumab. lupus erythematosus and other autoimmune diseases. BAFF Unexpectedly, we revealed differences in inhibitory potencies exists in three recognized forms: membrane-bound and two against the various BAFF forms, in particular that belimumab secreted, soluble forms of either trimeric or 60-mer oligomeric does not potently inhibit BAFF 60-mer. Through this increased

states. To date, most in vitro pharmacology studies of BAFF understanding of the activity of BAFF antagonists against at ASPET Journals on September 26, 2021 neglect one or more of these forms. Here, we report a different forms of BAFF, we hope to influence the discovery of comprehensive in vitro cell-based analysis of BAFF in assay BAFF antagonist antibodies with distinct therapeutic mecha- systems that measure all forms of BAFF-mediated activation. We nisms for improvement in the treatment of lupus or other related demonstrate the effects of these BAFF forms in both a primary autoimmune pathologies.

Introduction factor (TNF) SF13B provided clinical benefit in SLE, resulting in Food and Drug Administration (FDA) approval of the selective Systemic lupus erythematosus (SLE) is a chronic multisys- BAFF antagonist, belimumab (Benlysta) (Furie et al., 2011; Liu tem autoimmune disease characterized by the generation of and Davidson, 2011; Navarra et al., 2011). autoantibodies directed against a variety of self-antigens BAFF is a highly conserved member of the TNF superfamily (Tsokos, 2011). These autoantibodies form pathogenic im- of cytokines, primarily expressed by innate immune cells mune complexes that deposit into organs, ultimately inducing including monocytes, macrophages, dendritic cells, and other inflammation and tissue damage (Cancro et al., 2009; Sang nonhematopoetic cells (Shu et al., 1999; Tribouley et al., 1999; et al., 2014). Many of the current therapeutic options for the Nardelli et al., 2001). Early reports on BAFF function in- treatment of SLE include antimalarial agents, nonsteroidal anti-inflammatory drugs, glucocorticoids, and immunosup- dicated a potential critical role for determining B-cell fate and pressive drugs, which all seek to reduce the autoimmune the establishment of tolerance (Moore et al., 1999). The inflammatory process but also can result in general immuno- physiologic activity of BAFF is reportedly mediated by suppression and toxicity (Tsokos, 2011). Opportunities for new signaling via three receptors: B-cell activating factor receptor therapies exist among patients with the most severe disease (BAFFR), transmembrane activator and calcium-modulator as well as for steroid-sparing treatments. In recent years, and cyclophilin ligand interactor (TACI), and, in some reports, inhibition of the cytokine B-cell–activating factor [BAFF, also B-cell maturation antigen (BCMA). These three receptors are known as B-lymphocyte stimulator (BLyS)], tumor necrosis primarily expressed on B cells at different developmental stages (Gross et al., 2000; Marsters et al., 2000; Wu et al.,

dx.doi.org/10.1124/jpet.116.236075. 2000; Thompson et al., 2001; Yan et al., 2001; Karpusas s This article has supplemental material available at jpet.aspetjournals.org. et al., 2002; Liu et al., 2003; Bossen and Schneider, 2006). In

ABBREVIATIONS: BAFF, B-cell–activating factor; BAFFR, B-cell–activating factor receptor; BCMA, B-cell maturation antigen; belim-comp, belimumab-comparator; BLyS, B-lymphocyte stimulator; CHO, Chinese hamster ovary; FDA, Food and Drug Administration; NF, nuclear factor; SLE, systemic lupus erythematosus; tabal-comp, tabalumab-comparator; TACI, transmembrane activator and calcium-modulator and cyclophilin ligand interactor; TEV, tobacco etch virus; TNF, tumor necrosis factor.

37 38 Nicoletti et al. humans, BAFFR is widely expressed by all B cells, with the distinct oligomeric states remain unclear, and in vitro phar- exception of bone marrow plasma cells, and BAFF-BAFFR macology assays often neglect one of these forms. interactions have been proven critical to maintenance of We sought to characterize and compare the activities of primary B cells (Oren et al., 2002; Ng et al., 2004). BAFFR membrane-bound and secreted, soluble BAFF forms with selectively recruits TNF receptor–associated factor 3, leading trimer and 60-mer oligomeric states in a set of in vitro, cell- to downstream signaling cascades that can modulate B-cell based assay systems. Further,weprofiletwomonoclonal survival and differentiation, including activation of the nu- antibodies generated based on the published sequences of clear factor kB (NFkB) pathway (Mukhopadhyay et al., 1999; belimumab and tabalumab for their inhibitory capacity Claudio et al., 2002; Hatada et al., 2003; Bossen and against these various BAFF forms. Interestingly, we find Schneider, 2006; Mackay and Schneider, 2009). Transgenic significantly differing potencies at inhibiting the various mice that overexpress BAFF display phenotypic characteris- membrane and soluble forms of BAFF. Results reported here tics that closely mimic those observed in SLE and Sjögren’s provide insight on the biologic activity of membrane-bound syndrome, including increased numbers of anti-DNA anti- and soluble, secreted BAFF as well as the pharmacologic bodies, B-cell expansion, proteinuria, and glomerulonephritis mechanisms of BAFF antagonists. A better pharmacologic (Mackay et al., 1999; Yan et al., 2001; Groom et al., 2002). In understanding of the inhibitory capability of BAFF antago- humans, serum levels of BAFF are significantly elevated in nists against all forms of BAFF has the potential to impact patients with SLE, Sjögren’s syndrome, and rheumatoid the therapeutic success of these antibodies in the treatment arthritis compared with healthy controls (Cheema et al., of lupus and related pathologies. Downloaded from 2001; Mariette et al., 2003; Tan et al., 2003). In humans, BAFF exists in both a membrane-bound form (of 285 amino acids) and two soluble forms (Moore et al., 1999; Materials and Methods Schneider et al., 1999; Thompson et al., 2001). Soluble BAFF Antibodies. Amino acid sequence for the belimumab-comparator exists predominantly as a trimer or 60-mer (Liu et al., (belim-comp) antibody was derived from patent US20110293610 2002; Cachero et al., 2006); however, the oligomeric state of (Ruben et al., 2011). Belim-comp was produced on a human IgG1 jpet.aspetjournals.org soluble BAFF in vivo has been the subject of debate. Both wild-type framework with a l light chain to conform to the FDA- soluble and membrane-bound forms of BAFF are biologically approved molecule (Baker et al., 2003). Belimumab was also pur- active (Schneider et al., 1999; Liu et al., 2002; Liu et al., chased as the marketed product (GlaxoSmithKline, London, UK) and 2003; Zhukovsky et al., 2004), and data from transgenic mice showed identical inhibitory potency to the inhouse produced material indicate a role for soluble BAFF in homeostasis; membrane- in the Chinese hamster ovary (CHO)-BAFFR assay (data not shown). Amino acid sequence for the tabalumab-comparator (tabal-comp) bound BAFF plays an accessory role (Bossen et al., 2011). antibody was derived from World Health Organization INN publi- at ASPET Journals on September 26, 2021 Binding affinities of BAFF for BAFFR, TACI, and BCMA have cation vol. 25 (WHO, 2011). Tabal-comp was produced on a human been previously described (Wu et al., 2000; Yu et al., 2000; IgG4-Pro (S228P) framework with a k light chain to conform to the Bossen and Schneider, 2006), although the subsequent down- clinical-stage molecule (Manetta et al., 2014). Each antibody was stream signaling properties of the various BAFF forms have cloned in pTT5-Neo, expressed in CHO-E cells, and purified by not been well studied. The clinical consequences of blocking standard techniques.

Fig. 1. Various BAFF form reagents, an- tibodies used, and functional assays. (A) The forms of human BAFF generated as reagents for this study. (B) Antibodies with sequences identical to belimumab and tabalumab were generated. (C) Left and center: Assays used in this study: CHO cells expressing either human BAFFR or TACI, also stably expressing an NFkB luciferase reporter. Right: Primary CD19+ B cell proliferation, as measured by 3[H] thymidine incorporation at 72 hours. BAFF Antibodies Differ in Inhibitory Potency against BAFF Forms 39

Production and Purification of BAFF Trimer and 60-Mer. BAFF trimer (amino acid 72–285) and BAFF 60-mer (amino acids 134–285) constructs were designed as annotated by UniProt with a 6XHis tag and a tobacco etch virus (TEV) cleavage site added to the N terminus. Both expression vectors (pTT_6xHis-TEV-hBAFF trimer and pTT_6xHis-TEV-hBAFF 60-mer that was subcloned into pLVX-IRES-ZsGreen lenti vector, Clontech Laboratories, Inc., Moun- tain View, CA) were transfected in human embryonic kidney (HEK293-6E) cells (American Type Culture Collection, Manassas,

VA) and grown for 72 hours at 37°C, 5% CO2, 130 rpm. Culture supernatant was harvested and prepared for purification by adding protease inhibitor, 1 mM TCEP, 0.01% CHAPS, pH adjusted to 7.4. Recombinant BAFF trimer was purified by Ni-NTA affinity chromatography. Briefly, the culture supernatant was loaded onto Ni-Sepharose column (QIAGEN Inc., Valencia, CA) using an Akta purifier system at a flow rate of 10 ml/min at 4°C. Column was washed

with buffer A (50 mM NaPO4, 300 mM NaCl, 0.01% CHAPS, 1.0 mM TCEP, pH7.0 supplemented with 10 mM imidazole) and eluted with

250 mM imidazole in buffer A. Eluted protein was dialyzed overnight Downloaded from

against 100 mM HEPES, 5.0 mM CaCl2, 1.0 mM TCEP, pH 7.5, and incubated with His-tagged Furin at 25°C for 16 hours. A Ni-NTA drip column was used to remove furin and cleaved N-terminal fragment. Protein was further purified by size exclusion chromatography (Sephacryl S-200HR resin with a column volume of 1178 ml and dimensions of 5 cm  60 cm) in 1  phosphate-buffered saline, 1.0 mM

TCEP, 0.01% CHAPS, pH 7.4. jpet.aspetjournals.org Recombinant BAFF 60-mer was initially loaded onto a Ni-NTA affinity column and washed in buffer A (20 mM Tris, 100 mM NaCl, pH 8.2). Bound protein was eluted with 250 mM imidazole in buffer A. Purified protein was further purified by size-exclusion chromatogra- phy (HiPrep 16/60 Sephacryl S-400 column). Samples were analyzed by SDS-PAGE and analytical ultracentrifugation. Sedimentation velocity experiments were conducted in a Beckman Optima XL

ultracentrifuge (Beckman, Palo Alto, CA). The qualitative purity at ASPET Journals on September 26, 2021 apparent size distribution of the trimer and the 60-mer sample solutions were compared using the g(s) wide distribution analysis of SEDANAL. The percentage of various species in the 60-mer prepara- tion was fit by SEDFIT version 14.1 using the continuous c(s) distribution model (Schuck, 2000) (Supplemental Fig. 1). Membrane-Bound BAFF Cells. CHO-K1 cells from ATCC (Manassas, VA) were grown in F-12K medium supplemented with 10% fetal bovine serum and transfected with mammalian expression vector pcDNA3.1 encoding human BAFF (NM_006573) using Lipofectamine (Life Technologies Corporation, Grand Island, NY) per the manufacturer’s instructions. A stably transfected single cell clone was generated by growth in 850ug/ml G418 followed by fluorescence-activated cell sorting using anti-human CD257 mAb clone 1D6 (eBioscience, Inc., San Diego, CA) (Supplemental Fig. 2). Cells were treated with 0.125% paraformaldehyde (Electron Micros- copy Sciences, Hatfield, PA) and incubated at room temperature for 1 hour. The fixed membrane-bound human BAFF cells were then washed and resuspended in complete media overnight before use in membrane-bound assay. NFkB Luciferase Reporter Assay. CHO-K1 cells were trans- fected with a pcDNA3.1mammalian expression vector encoding either untagged human BAFFR (NM_052945) or TACI (NM_012452) using Lipofectamine as previously described. Cells were also transduced with NFkB Luciferase lentiviral particles (QIAGEN Inc., Valencia, CA) according to the manufacturer’s instruction and grown in the presence of 10 mg/ml puromycin. Determination of the number of BAFF sites per cell was performed using the QuantiBrite system (BD Fig. 2. Activity of soluble and membrane-bound BAFF forms in BAFFR Biosciences, San Jose, CA) according to the manufacturer’s protocol. and TACI activity reporter assays. (A) Human BAFFR- and (B) TACI- expressing CHO NFkB reporter cells were stimulated with serial titrations of soluble BAFF trimer (red) or 60-mer (blue). All NFkB-mediated luciferase signals were measured at 24 hours. (C) CD19+ human primary B cells were stimulated with 2 mg/ml anti-IgM and various concentrations of either expressing membrane-bound human BAFF were serially titrated in the BAFF trimer (red) or 60-mer (blue). Cell proliferation was measured at indicated cell numbers with a fixed number (80,000) of CHO-BAFFR- 72 hours by 3[H] thymidine incorporation. (D) Fixed, washed CHO cells NFkB reporter cells; luciferase activity was measured at 24 hours. 40 Nicoletti et al.

Human BAFFR or TACI CHO NFkB luciferase reporter cells were acids 72-285 of human BAFF, purified at pH 7.4, and cleaved plated in 96-well flat bottom white plates in X-VIVO media (Lonza, between Arg133 and Ala134, no 60-mers are formed, resulting Basel, Switzerland). Serial dilutions of recombinant soluble human in only the trimic form (blue line, Supplemental Fig. S1). BAFF forms or fixed membrane-bound BAFF cells, in the absence BAFF is also expressed as a membrane-bound protein, or presence of BAFF antagonist antibodies at the indicated concen- which is continuously shed by furin proteolytic cleavage. To trations, were added into assay wells containing reporter cells and generate a reagent for assessing the activity of membrane- incubated for 24 hours at 37°C. Plates were then acclimated to room temperature, and luciferase activity was measured. bound BAFF, we engineered a CHO cell line that stably Primary Human B-Cell Proliferation Assay. Negatively se- overexpressed human BAFF (Supplemental Fig. 2). This cell lected B cells were obtained from frozen human peripheral blood line was then paraformaldehyde-fixed and washed to prevent (AllCells, LLC, Alameda, CA). Sexes of cell donors were as follows: further soluble BAFF shedding. Quantitation of the number of Fig. 2C 5 60-mer, two males, one female; 3mer, one male; Fig. 4A, two molecules per cell using flow cytometry demonstrated approx- males, one female; Fig. 4B, two males, one female; and Fig. 4C, one imately 20,000 BAFF molecules per cell. In supporting male, one female. Primary B cells were thawed in RPMI complete experiments, we confirmed that the fixed CHO-BAFF cells media (Life Technologies Corporation, Grand Island, NY) supple- do not continue to shed soluble or cleaved BAFF from the cell mented with 10% fetal bovine serum, 1% penicillin/streptomycin, 1% surface during the course of the assay by performing enzyme- sodium pyruvate, 1% HEPES, 1% nonessential amino acids, 1% (v/v) linked immunosorbent assay on the assay media after com- L-glutamine, and 0.1% (v/v) 2-mercaptoethanol . In a 96-well flat- pletion of the assay (data not shown). These reagents were

bottom assay plate, human BAFF (in soluble or membrane-bound Downloaded from forms) and 2.0 mg/ml anti-human IgM (Jackson ImmunoResearch subsequently used in comparisons of BAFFR and TACI Laboratories, Inc., West Grove, PA) were preincubated with BAFF activity in cellular assays as described below. antagonist antibodies for 30 minutes at 37°C. Then 100,000 purified Functional Activity of Various BAFF Forms in Cell- human B cells were added the wells in triplicate and further incubated Based Assays for BAFFR and TACI. Soluble, secreted for 72 hours at 37°C, 5% CO2; 16 hours before harvest, each well was trimer or 60-mer forms of BAFF were first characterized in 3 pulsed with 1 mCi/well [H] thymidine. Cells were harvested onto filter NFkB luciferase reporter assays in CHO cells expressing mats using a Molecular Devices Micro96 Harvester (Molecular De- either the BAFF receptor (BAFFR) or TACI (Fig. 1B). Activity jpet.aspetjournals.org vices, LLC, Sunnyvale CA). Filter mats were dried at 37°C for 1 hour, in these assays both confirms functionality of the proteins we and 3[H] thymidine incorporation was measured using a MicroBeta2 produced and allowed us to set EC concentrations to use for 2450 Microplate Counter (PerkinElmer, Waltham, MA). 80 Statistics. Statistical analysis was performed using GraphPad follow-up BAFF antagonist studies carried out later in this Prism, version 6 (GraphPad Software, Inc., La Jolla, CA) or XLfit, report. Figure 2A illustrates that soluble 60-mer BAFF was version 4 (ID Business Solutions Limited, Guildford, Surrey, UK). more potent than the trimeric form at stimulating BAFFR- k BAFF antagonist IC50 values were calculated using four parameter mediated NF Bactivity,withEC50sof1.3pMand19pMfor nonlinear regression. 60-mer and trimer, respectively (Fig. 2A; Table 1). The same at ASPET Journals on September 26, 2021 pattern was evident when TACI activity was measured, Results where soluble 60-mer displayed an EC50 of 51 pM and trimer was 156 pM (Fig. 2B; Table 1). In these assay systems, Generation of BAFF Reagents: Soluble and Membrane- soluble 60-mer BAFF was a more potent stimulus of BAFFR Bound. As a first step in a systematic approach to function- than TACI. ally comparing various BAFF forms and how antagonistic Human B lymphocytes express all three receptors for antibodies might differently modulate these activities, it was BAFF (BAFFR, TACI, and BCMA) and were used to further necessary to generate and fully characterize BAFF trimer, explore the physiologic relevance of data reported using the 60-mer, and membrane-bound forms (Fig. 1A). Previously, BAFFR and TACI CHO NFkB luciferase reporter assays there has been debate in the literature regarding the physi- (Fig. 2C). In these primary human B cells, soluble 60-mer ologic importance of BAFF 60-mer form. Careful analysis, BAFF again demonstrated increased potency compared with however, reveals that BAFF 60-mer formation is independent the trimeric form for induction of cell proliferation in the of added epitope tags and represents a bona fide oligomer of presence of anti-IgM (Fig. 2C), paralleling the effects ob- the cytokine (Cachero et al., 2006). Specifically, for our pro- served in the CHO cells. Here, the 60-mer’sEC50 was 4.5 pM duction of the 60-mer, amino acids 134–285 were used from versus 46 pM for the trimer. the human BAFF sequence and produced a trimer/60-mer Lastly, to characterize the membrane-bound BAFF re- equilibrium, which at pH 8.2 and analytical concentrations is agent and determine an optimal challenge concentration for approximately 77% 60-mer (see Supplemental Fig. 1 and antagonist studies, we used the cell:cell ratio as a proxy Supplemental Table 2). Conversely, when we expressed amino for membrane-bound BAFF concentration and carried out

TABLE 1

EC50 and EC80 values of BAFF forms

Receptor/Assay BAFF Form EC50 (pM), Mean (S.D.) EC80 (pM), Mean (S.D.) CHO-BAFFR Trimer 19 (4.2), n = 7 70 (25.3), n =7 60-Mer 1.28 (0.16), n = 3 4.8 (0.6), n =3 CHO-TACI Trimer 156 (30.7), n = 5 222 (26.7), n =5 60-Mer 51 (4.0), n = 2 86 (1.3), n =2 B-cell proliferation Trimer 46 (3.7), n = 2 328 (155.2), n =2 60-Mer 4.47 (1.4), n = 3 14.40 (4.1), n =3

BAFF, B-cell–activating factor; BAFFR, B-cell–activating factor receptor; belim-comp, belimumab-comparator; CHO, Chinese hamster ovary; TACI, transmembrane activator and calcium-modulator and cyclophilin ligand interactor. BAFF Antibodies Differ in Inhibitory Potency against BAFF Forms 41 Downloaded from

Fig. 3. Activity of BAFF antagonist antibodies against EC80 of BAFF trimer, 60-mer, and membrane-bound in BAFFR and TACI activity assays. (A) CHO-hBAFFR-NFkB luciferase reporter cells were stimulated with 52 pM soluble trimeric BAFF in the presence and absence of BAFF antagonists for 24 hours. (B) CHO-hBAFFR-NFkB cells were stimulated with 4.2 pM soluble 60-mer BAFF in the presence and absence of BAFF antagonist for 24 hours. (C) CHO hBAFFR-NFkB cells were stimulated with fixed cells expressing membrane-bound (m.b.) BAFF at a 1:1 ratio in the presence and jpet.aspetjournals.org absence of BAFF antagonist for 24 hours. (D) CHO-hTACI-NFkB luciferase reporter cells were stimulated with 288 pM soluble trimeric BAFF in the presence and absence of BAFF antagonists for 24 hours. (B) CHO-hTACI-NFkB cells were stimulated with 67.4 pM soluble 60-mer BAFF in the presence and absence of BAFF antagonist for 24 hours. (C) CHO-hTACI-NFkB cells were stimulated with fixed cells expressing membrane-bound BAFF at a 1:1 ratio in the presence and absence of BAFF antagonist for 24 hours. titration curves in the BAFFR CHO NFkB activity assay (Fig. writing. These antagonistic antibodies were evaluated for

2D). Based on these data, we conclude that the membrane- their ability to neutralize soluble and membrane-bound at ASPET Journals on September 26, 2021 bound BAFF stimulates both BAFFR and TACI and that a 1:1 BAFF in the BAFFR and TACI CHO NFkB luciferase assays cell ratio resulted in optimal stimulation for follow-up studies (Fig. 3; Table 2). in CHO NFkB reporter assays for BAFFR and TACI and a 4:1 At the BAFFR, the belim-comp and tabal-comp anti- ratio (B cell to CHO-BAFF) for studies involving human B-cell bodies both neutralized soluble, trimeric BAFF (Fig. 3A), proliferation (Supplemental Fig. 3). and similar IC50 values were obtained (Table 2). In con- BAFF Antagonists Exhibit Differential Inhibition trast, inhibitory potencies versus soluble BAFF 60-mer were of Membrane-Bound and Soluble Oligomeric BAFF strikingly different (Fig. 3B). Whereas the tabal-comp anti- Forms. Given that the various forms of BAFF could expose body demonstrated remarkable potency of 15 pM, the belim- or hide certain antibody-binding epitopes in the context of an comp antibody demonstrated complete inhibition only at the oligomeric structure, we investigated if BAFF antagonist highest concentration tested (67 nM), and a conclusive curve antibodies display any functional differences in their ability fit for IC50 was unable to be performed (Table 2). Neutrali- to inhibit these various BAFF forms. We generated anti- zation of membrane BAFF-induced NFkB activity also bodies from available patent sequences that are identical in revealed differences in inhibitory potencies between the belim- amino acid sequence to belimumab and tabalumab (here comp and tabal-comp antibodies, demonstrating greater simi- referred to as belim-comp and tabal-comp), the two most larity to the pattern observed with BAFF trimer rather than clinically advanced BAFF antagonists at the time of this with 60-mer. (Fig. 3C).

TABLE 2

IC50 values of BAFF antibodies

Receptor/Assay BAFF Form Tabal-Comp IC50 (pM), Mean (S.D.) Belim-Comp IC50 (pM), Mean (S.D.) CHO-BAFFR Trimer 70 (14.1), n = 4 264 (36.6), n =4 60-Mer 15 (0.2), n = 3 n.d., max = 95%, n =3 m.b. 240 (80.1), n = 4 n.d., max = 79%, n =4 CHO-TACI Trimer 233 (96.8), n = 3 435 (25.1), n =3 60-Mer 73 (36.0), n = 2 1969 (2542), n =2 m.b. 1102, n = 1 2103, n =1 B-cell proliferation Trimer 842 (21.2), n = 3 958 (83.8), n =3 60-Mer 98 (2.4), n = 2 n.d., max = 74%, n =2 m.b. 160 (27.7), n = 2 303 (37.0), n =2

BAFF, B-cell–activating factor; BAFFR, B-cell–activating factor receptor; belim, belimumab; CHO, Chinese hamster ovary; comp, comparator; m.b., membrane bound; n.d., not done; TACI, transmembrane activator and calcium-modulator and cyclophilin ligand interactor. 42 Nicoletti et al. At the TACI receptor, both belim-comp and tabal-comp A antibodies were again similar in their potency at blocking BAFF trimer (435 pM and 233 pM, respectively) (Fig. 3D; Table 2); however, as with the BAFFR assay, we also observed a significantly different inhibitory potency in the TACI assay with 60-mer BAFF (Fig. 3E). Here the tabal-comp antibody was a highly potent inhibitor of 60-mer with an IC50 of 73 pM, and the belim-comp antibody was nearly 30-fold less potent with an IC50 of 1.97nM. The observed inhibitory potencies of the two antibodies against membrane-bound BAFF were similar in the TACI activation assay, with IC50s of 2.1nM and 1.1 nM for belim-comp and tabal-comp, respectively (Fig. 3F; Table 2). Taken together, these results demonstrated the ability of tabal-comp to potently neutralize soluble and membrane-bound forms of BAFF activity mediated by TACI and BAFFR, whereas belim-comp potently neutralized only B the soluble, trimeric, and membrane-bound forms. Potency Differences in BAFF Antibodies Reflected in Downloaded from Primary B-Cell Proliferation. To understand whether the data obtained in the CHO assays reflects physiologically relevant activity in primary cells with endogenously expressed receptors for BAFF, we profiled these antibodies in human CD19-positive B-cell proliferation assays. The belim-comp and tabal-comp antibodies were evaluated for their ability to jpet.aspetjournals.org neutralize BAFF-induced B cell proliferation in the presence of anti-IgM. Figure 4A illustrates that the belim-comp and tabal-comp antibodies are nearly equivalent in their capacity to neutralize soluble trimeric BAFF, with IC50 values of 0.96 nM and 0.84 nM, respectively. Confirming our surprising results with soluble 60-mer BAFF in the CHO assays, differences in C inhibitory potencies were again observed for these two at ASPET Journals on September 26, 2021 antibodies with respect to B-cell proliferation (Fig. 4B). Here, the tabal-comp antibody demonstrated superior potency (98 pM IC50) compared with belim-comp (IC50 could not be determined owing to maximal observed efficacy of only 74% at the highest tested concentration of 67 nM). Results for neutralization of membrane BAFF using primary B cells were similar to those observed with trimer, and consistent with findings in the CHO assays (Fig. 4C). Here, the tabal- comp antibody neutralized membrane-bound BAFF with slightly increased potency (160 pM IC50)comparedwith belim-comp (303 pM IC50)(Table2).

Fig. 4. Relevance of BAFF antagonist antibodies for neutralizing primary Discussion human B-cell proliferation. (A) Primary human CD19+ B cells were The oligomeric state of soluble BAFF under physiologic stimulated with 196 pM (EC80) soluble trimeric BAFF in the presence and absence of BAFF antagonists for 72 hours. (B) Primary human CD19+ B conditions has been the subject of much debate (Liu et al., cells were stimulated with 14 pM (EC80) soluble 60-mer BAFF in the 2002, 2003; Zhukovsky et al., 2004; Cachero et al., 2006; Liu presence and absence of BAFF antagonist for 72 hours. (C) Primary and Davidson, 2011), in particular the existence of the 60-mer human CD19+ B cells were stimulated with fixed cells expressing membrane-bound BAFF at a 4:1 ratio in the presence and absence of form. Liu et al. (2003) first reported that the oligomeric state of BAFF antagonist for 72 hours. BAFF in solution was pH-dependent and that 60-mer forma- tion was structurally mediated by the longer DE loop connect- ing anti-parallel b strands (Liu et al., 2003). The biologic 60-mer in cell culture supernatants of endogenously express- relevance of 60-mer BAFF was later challenged by others, who ing U937 cells or after transient transfection of 293T cells, concluded that soluble BAFF is solely trimeric under normal revealing that BAFF 60-mer is a form produced by cells under physiologic conditions and that 60-mer formation is an artifact normal protein processing and secretion pathways. Further, of an amino-terminal histidine tag (Zhukovsky et al., 2004). BAFF transgenic mice have 60-mer present in their serum These results were later refuted by others with conclusive, (Bossen et al., 2008), adding to the weight of evidence that comparative studies supporting that 60-mer formation is 60-mer is a relevant in vivo form, at least in the mouse. BAFF indeed an intrinsic property of the BAFF protein and not a 60-mer has not been identified in human serum to date; consequence of epitope tagging (Cachero et al., 2006). Addi- however, the cell line experiments cited above support that tionally, these authors showed significant detection of BAFF this form is endogenously produced by human cells. For the BAFF Antibodies Differ in Inhibitory Potency against BAFF Forms 43 purposes of this in vitro pharmacologic report, the compre- Clinical investigation into the efficacy of BAFF antagonists hensive protein characterization performed here, combined in human SLE was initially supported by observations such as with the results from Cachero et al. (2006), clearly indicate elevated serum levels of BAFF in SLE patients (Zhang et al., that achieving the oligomeric state of human BAFF in a 2001) and transgenic mice overexpressing BAFF that dis- recombinant form is achievable and depends on sequence and play manifestations of SLE, including elevated levels of purification methods. immunoglobulin, anti-double-stranded DNA antibodies, Binding affinities of BAFF for BAFFR, TACI, and BCMA and circulating immune complexes (Mackay et al., 1999; have been previously described (Wu et al., 2000; Yu et al., Yan et al., 2001). The FDA approval of belimumab (Benlysta) 2000; Bossen and Schneider, 2006), although the subsequent in 2011 may represent validation of the involvement of downstream signaling properties of the various BAFF forms BAFF in SLE pathogenesis; however, the efficacy of have not been well studied. In the current study, we chose to belimumab in SLE has been described by some experts as focus our characterization on BAFFR and TACI, where BAFF marginal and potentially not clinically relevant (Chiche affinity is highest (Schneider, 2005). Additionally, BAFF- et al., 2012). Despite this, additional BAFF antagonists con- mediated B-cell survival is believed to be predominantly tinue to be evaluated in the clinic (Mackay and Schneider, 2009; carried out by BAFF-R (Claudio et al., 2002; Bossen et al., Furie et al., 2011; Navarra et al., 2011; Boneparth and 2008). Based on the results presented here, if BAFF 60-mer is Davidson, 2012). Tabalumab was recently evaluated in two shown to activate BCMA signaling, studies can be performed phase 3 clinical trials of SLE (Lilly to discontinue development to determine whether a similar pattern of inhibitory potencies of tabalumab based on efficacy results in phase 3 lupus studies, Downloaded from holds true for signaling at BCMA as well. Eli Lilly and Co., https://investor.lilly.com/releasedetail.cfm? We show here that membrane-bound and soluble forms ReleaseID5874281), and subsequently further clinical devel- of BAFF were functionally active in the induction of NFkB- opment was halted; blisibimod, a third BAFF antagonist of driven luciferase assays, triggered either through BAFFR or different structure, is the subject of an ongoing phase TACI receptors. Consistent with previous literature (Bossen 3 clinical trial in SLE (Scheinberg et al., 2016). Although et al., 2008), we find that, compared with trimeric BAFF, not the subject of our current report, we generated a protein jpet.aspetjournals.org 60-mer was approximately 100-fold more potent in stimulat- construct equivalent to blisibimod and observed that its ing BAFFR-mediated NFkB driven luciferase; however, po- activity was approximately equivalent to the tabal-comp tencies for BAFF trimer and 60-mer were more comparable in antibody at blocking BAFF 60-mer in the BAFFR CHO assay the TACI NFkB assay. This result, specifically that trimer is (data not shown). Conclusive head-to-head clinical compar- capable of signaling via the TACI receptor, contradicts a isons of these BAFF antagonists have not been performed. previous report (Bossen et al., 2008) in which the authors Given the large unmet medical need in SLE and the continued showed that TACI has specificity for higher-order BAFF pharmaceutical interest in BAFF antagonists, understanding at ASPET Journals on September 26, 2021 oligomers exclusively, with no signaling by trimeric forms. the distinct molecular mechanisms of these BAFF antagonists One potential explanation for the difference observed here is may be crucial to achieving clinical efficacy beyond that that the previous report was done in mouse cells, and we have observed with belimumab. used only human BAFF, BAFFR, and TACI sequences, reflecting potential species differences. In agreement with Acknowledgments this novel finding that human BAFF trimer can bind and The authors thank Daniela Cazacu for cell line generation; Nichole signal via TACI receptor, we also find that membrane-bound Dudek, Volga Pasupuleti, and Helen Wu for reagents provided; BAFF, believed to be primarily trimeric (Mackay and Schneider, Anthony Kronkaitis, Zhong-Fu Huang and Chi-Ming Yu for antibody 2009), can also signal in our TACI assay system. expression and purification; John Miglietta for construct design; David Hayes for biophysical analysis; and Donald Van Leeuwen for No systematic, head-to-head comparative study of BAFF experimental contribution. antagonist antibodies has been published to date. Thus, we aimed to profile two of these antibodies for their activity Authorship Contributions against the three forms of BAFF. Most interesting, whereas Participated in research design: Nicoletti, Khalil, Ralph, Presky, our two comparator antibodies were nearly identical in their DeWire, Brodeur. activity against BAFF trimer and membrane-bound forms in Conducted experiments: Nicoletti, Kenny, Khalil, Ralph. both CHO-BAFFR and primary human B cell systems, we Contributed new reagents or analytic tools: Nicoletti, Kenny, revealed an approximately three orders of magnitude differ- Khalil, Pan, Ritchie. ence in their inhibitory potency at blocking the signaling Performed data analysis: Nicoletti, Kenny, Khalil, Ralph. effects of BAFF 60-mer. One likely explanation for this result Wrote or contributed to the writing of the manuscript: Nicoletti, is that the binding epitope for the belim-comp antibody is Venkataramani, Presky, DeWire, Brodeur exposed on the trimer and membrane-bound forms, yet is References hidden within the macromolecular structure of the 60-mer Baker KP, Edwards BM, Main SH, Choi GH, Wager RE, Halpern WG, Lappin PB, form. Conversely, the epitope bound by the tabal-comp Riccobene T, Abramian D, Sekut L, et al. (2003) Generation and characterization of LymphoStat-B, a human monoclonal antibody that antagonizes the bioactivities of antibody may be exposed in all three BAFF forms. Our own B lymphocyte stimulator. Arthritis Rheum 48:3253–3265. preliminary efforts to determine the binding region for the Boneparth A and Davidson A (2012) B-cell activating factor targeted therapy and “ ” lupus. Arthritis Res Ther 14(Suppl 4):S2. belim-comp antibody revealed protection in the D/E loop Bossen C, Cachero TG, Tardivel A, Ingold K, Willen L, Dobles M, Scott ML, Maquelin region of BAFF trimer (data not shown), known to be critical A, Belnoue E, Siegrist CA, et al. (2008) TACI, unlike BAFF-R, is solely activated by oligomeric BAFF and APRIL to support survival of activated B cells and plasma- for 60-mer formation (Liu et al., 2002; Cachero et al., 2006). blasts. Blood 111:1004–1012. Further epitope mapping with both antibodies is needed to Bossen C and Schneider P (2006) BAFF, APRIL and their receptors: structure, function and signaling. Semin Immunol 18:263–275. clarify whether this is the underlying mechanism explaining Bossen C, Tardivel A, Willen L, Fletcher CA, Perroud M, Beermann F, Rolink AG, the functional results presented here. Scott ML, Mackay F, and Schneider P (2011) Mutation of the BAFF furin cleavage 44 Nicoletti et al.

site impairs B-cell homeostasis and antibody responses. Eur J Immunol 41: Nardelli B, Belvedere O, Roschke V, Moore PA, Olsen HS, Migone TS, Sosnovtseva S, 787–797. Carrell JA, Feng P, Giri JG, et al. (2001) Synthesis and release of B-lymphocyte Cachero TG, Schwartz IM, Qian F, Day ES, Bossen C, Ingold K, Tardivel A, stimulator from myeloid cells. Blood 97:198–204. Krushinskie D, Eldredge J, Silvian L, et al. (2006) Formation of virus-like clusters Navarra SV, Guzmán RM, Gallacher AE, Hall S, Levy RA, Jimenez RE, Li EK, is an intrinsic property of the tumor necrosis factor family member BAFF (B cell Thomas M, Kim HY, León MG, et al.; BLISS-52 Study Group (2011) Efficacy and activating factor). Biochemistry 45:2006–2013. safety of belimumab in patients with active systemic lupus erythematosus: a Cancro MP, D’Cruz DP, and Khamashta MA (2009) The role of B lymphocyte stim- randomised, placebo-controlled, phase 3 trial. Lancet 377:721–731. ulator (BLyS) in systemic lupus erythematosus. J Clin Invest 119:1066–1073. Ng LG, Sutherland AP, Newton R, Qian F, Cachero TG, Scott ML, Thompson JS, Cheema GS, Roschke V, Hilbert DM, and Stohl W (2001) Elevated serum Wheway J, Chtanova T, Groom J, Sutton IJ, Xin C, Tangye SG, Kalled SL, Mackay B lymphocyte stimulator levels in patients with systemic immune-based rheumatic F, and Mackay CR (2004) B cell-activating factor belonging to the TNF family diseases. Arthritis Rheum 44:1313–1319. (BAFF)-R is the principal BAFF receptor facilitating BAFF costimulation of cir- Chiche L, Jourde N, Thomas G, Bardin N, Bornet C, Darque A, and Mancini J (2012) culating T and B cells. Journal of immunology (Baltimore, Md: 1950) 173:807–817. New treatment options for lupus: a focus on belimumab. Ther Clin Risk Manag 8: Oren DA, Li Y, Volovik Y, Morris TS, Dharia C, Das K, Galperina O, Gentz R, 33–43. and Arnold E (2002) Structural basis of BLyS receptor recognition. Nat Struct Biol Claudio E, Brown K, Park S, Wang H, and Siebenlist U (2002) BAFF-induced NEMO- 9:288–292. independent processing of NF-kappa B2 in maturing B cells. Nat Immunol 3: Ruben SM, Barash SC, Choi GH, Vaughan T, and Hilbert D (2015) inventors, Human 958–965. Genome Sciences, Inc., assignee. Antibodies that immunospecifically bind to B Furie R, Petri M, Zamani O, Cervera R, Wallace DJ, Tegzová D, Sanchez-Guerrero J, lymphocyte stimulator protein. U.S. patent 9187548B2. 2015 Nov 15. Schwarting A, Merrill JT, Chatham WW, et al.; BLISS-76 Study Group (2011) A Sang A, Zheng YY, and Morel L (2014) Contributions of B cells to lupus pathogenesis. phase III, randomized, placebo-controlled study of belimumab, a monoclonal an- Mol Immunol 62:329–338. tibody that inhibits B lymphocyte stimulator, in patients with systemic lupus Scheinberg MA, Hislop CM, and Martin RS (2016) Blisibimod for treatment of sys- erythematosus. Arthritis Rheum 63:3918–3930. temic lupus erythematosus: with trials you become wiser. Expert Opin Biol Ther Groom J, Kalled SL, Cutler AH, Olson C, Woodcock SA, Schneider P, Tschopp J, 16:723–733. Cachero TG, Batten M, Wheway J, et al. (2002) Association of BAFF/BLyS over- Schneider P (2005) The role of APRIL and BAFF in lymphocyte activation. Curr Opin expression and altered B cell differentiation with Sjögren’s syndrome. J Clin Invest Immunol 17:282–289. 109:59–68. Schneider P, MacKay F, Steiner V, Hofmann K, Bodmer JL, Holler N, Ambrose C, Downloaded from Gross JA, Johnston J, Mudri S, Enselman R, Dillon SR, Madden K, Xu W, Parrish- Lawton P, Bixler S, Acha-Orbea H, et al. (1999) BAFF, a novel ligand of the tumor Novak J, Foster D, Lofton-Day C, et al. (2000) TACI and BCMA are receptors for a necrosis factor family, stimulates B cell growth. J Exp Med 189:1747–1756. TNF homologue implicated in B-cell autoimmune disease. Nature 404:995–999. Schuck P (2000) Size-distribution analysis of macromolecules by sedimentation ve- Hatada EN, Do RK, Orlofsky A, Liou HC, Prystowsky M, MacLennan IC, Caamano J, locity ultracentrifugation and lamm equation modeling. Biophys J 78:1606–1619. and Chen-Kiang S (2003) NF-kappa B1 p50 is required for BLyS attenuation of Shu HB, Hu WH, and Johnson H (1999) TALL-1 is a novel member of the TNF family apoptosis but dispensable for processing of NF-kappa B2 p100 to p52 in quiescent that is down-regulated by mitogens. J Leukoc Biol 65:680–683. mature B cells. J Immunol 171:761–768. Tan SM, Xu D, Roschke V, Perry JW, Arkfeld DG, Ehresmann GR, Migone TS, Karpusas M, Cachero TG, Qian F, Boriack-Sjodin A, Mullen C, Strauch K, Hsu YM, Hilbert DM, and Stohl W (2003) Local production of B lymphocyte stimulator and Kalled SL (2002) Crystal structure of extracellular human BAFF, a TNF protein and APRIL in arthritic joints of patients with inflammatory arthritis. jpet.aspetjournals.org family member that stimulates B lymphocytes. J Mol Biol 315:1145–1154. Arthritis Rheum 48:982–992. Liu Y, Hong X, Kappler J, Jiang L, Zhang R, Xu L, Pan CH, Martin WE, Murphy RC, Thompson JS, Bixler SA, Qian F, Vora K, Scott ML, Cachero TG, Hession C, Shu HB, et al. (2003) Ligand-receptor binding revealed by the TNF family member Schneider P, Sizing ID, Mullen C, et al. (2001) BAFF-R, a newly identified TNF TALL-1. Nature 423:49–56. receptor that specifically interacts with BAFF. Science 293:2108–2111. Liu Y, Xu L, Opalka N, Kappler J, Shu HB, and Zhang G (2002) Crystal structure Tribouley C, Wallroth M, Chan V, Paliard X, Fang E, Lamson G, Pot D, Escobedo J, of sTALL-1 reveals a virus-like assembly of TNF family ligands. Cell 108: and Williams LT (1999) Characterization of a new member of the TNF family 383–394. expressed on antigen presenting cells. Biol Chem 380:1443–1447. Liu Z and Davidson A (2011) BAFF inhibition: a new class of drugs for the treat- Tsokos GC (2011) Systemic lupus erythematosus. N Engl J Med 365:2110–2121. ment of autoimmunity. Exp Cell Res 317:1270–1277. World Health Organization (WHO) (2011) International nonproprietary names for

Mackay F and Schneider P (2009) Cracking the BAFF code. Nat Rev Immunol 9: pharmaceutical substances (INN). WHO Drug Information 25:195–196. at ASPET Journals on September 26, 2021 491–502. Wu Y, Bressette D, Carrell JA, Kaufman T, Feng P, Taylor K, Gan Y, Cho YH, Garcia Mackay F, Woodcock SA, Lawton P, Ambrose C, Baetscher M, Schneider P, AD, Gollatz E, et al. (2000) Tumor necrosis factor (TNF) receptor superfamily Tschopp J, and Browning JL (1999) Mice transgenic for BAFF develop lym- member TACI is a high affinity receptor for TNF family members APRIL and phocytic disorders along with autoimmune manifestations. JExpMed190: BLyS. J Biol Chem 275:35478–35485. 1697–1710. Yan M, Brady JR, Chan B, Lee WP, Hsu B, Harless S, Cancro M, Grewal IS, and Dixit Manetta J, Bina H, Ryan P, Fox N, Witcher DR, and Kikly K (2014) Generation and VM (2001) Identification of a novel receptor for B lymphocyte stimulator that is characterization of tabalumab, a human monoclonal antibody that neutralizes mutated in a mouse strain with severe B cell deficiency. Curr Biol 11:1547–1552. both soluble and membrane-bound B-cell activating factor. J Inflamm Res 7: Yu G, Boone T, Delaney J, Hawkins N, Kelley M, Ramakrishnan M, McCabe S, Qiu 121–131. WR, Kornuc M, Xia XZ, et al. (2000) APRIL and TALL-I and receptors BCMA and Mariette X, Roux S, Zhang J, Bengoufa D, Lavie F, Zhou T, and Kimberly R (2003) TACI: system for regulating humoral immunity. Nat Immunol 1:252–256. The level of BLyS (BAFF) correlates with the titre of autoantibodies in human Zhang J, Roschke V, Baker KP, Wang Z, Alarcon GS, Fessler BJ, Bastian H, Sjögren’s syndrome. Ann Rheum Dis 62:168–171. Kimberly RP, and Zhou T (2001) Cutting edge: a role for B lymphocyte stimulator Marsters SA, Yan M, Pitti RM, Haas PE, Dixit VM, and Ashkenazi A (2000) In- in systemic lupus erythematosus. Journal of immunology (Baltimore, Md: 1950) teraction of the TNF homologues BLyS and APRIL with the TNF receptor homo- 166:6–10. logues BCMA and TACI. Curr Biol 10:785–788. Zhukovsky EA, Lee JO, Villegas M, Chan C, Chu S, and Mroske C (2004) TNF Moore PA, Belvedere O, Orr A, Pieri K, LaFleur DW, Feng P, Soppet D, Charters M, ligands: is TALL-1 a trimer or a virus-like cluster? Nature 427:413–414, discussion Gentz R, Parmelee D, et al. (1999) BLyS: member of the tumor necrosis factor 414. family and B lymphocyte stimulator. Science 285:260–263. Mukhopadhyay A, Ni J, Zhai Y, Yu GL, and Aggarwal BB (1999) Identification and characterization of a novel cytokine, THANK, a TNF homologue that activates Address correspondence to: Scott DeWire, 900 Ridgebury Road, Ridgefield, apoptosis, nuclear factor-kappaB, and c-Jun NH2-terminal kinase. J Biol Chem CT 06877. E-mail: [email protected] 274:15978–15981. Supplemental data package for:

Unexpected potency differences between BAFF antagonist antibodies against various forms of BAFF: trimer, 60-mer, and membrane-bound.

Amy M. Nicoletti, Cynthia Hess Kenny, Ashraf M. Khalil, Qi Pan, Kerry L. M. Ralph, Julie Ritchie,

Sathyadevi Venkataramani, David H. Presky, Scott M. DeWire, and Scott R. Brodeur

Journal of Pharmacology and Experimental Therapeutics, 2016.

JPET # 236075 SUPPLEMENTAL FIGURE 1: Characterization of recombinant human membrane-bound BAFF and soluble BAFF with trimer and 60-mer oligomeric states

(A) Wide distribution analysis g(s) apparent sedimentation coefficient distribution functions generated by SEDANAL compared to the apparent distribution function of an ideal, simulated

60mer (green fit line). (B) c(s) curve fit of the predominantly 60mer sample (particle percentage calculated by subtracting total c(s) concentration from starting concentration measured at 3000

RPM). (C) Particle percentage calculated by subtracting total c(s) concentration from starting concentration measured at 3000 RPM.

SUPPLEMENTAL FIGURE 2: Expression of membrane-bound BAFF on stably transfected

CHO cells

CHO cells were transfected with pcDNA3.1 encoding full-length human BAFF and expression of membrane-bound human BAFF was detected following immunostaining using phycoerythrin

(PE)-conjugated anti-BAFF clone 1D6 or isotype control antibody for 30 mins prior to flow cytometric analysis. Dark grey histogram represents cells with buffer only; dotted line represents cells plus isotype control; and histogram with solid line represents cells plus anti-BAFF mAb clone 1D6.

SUPPLEMENTAL FIGURE 3: Cell titration for mbBAFF in B cell proliferation assay

Fixed, washed CHO cells expressing membrane-bound human BAFF were serially titrated in at the indicated cell numbers with a fixed number of CD19+ human primary B cells (100,000 per well of 96 well plate), proliferation was measured by 3[H] thymidine at 72 hours.

A

B C Fragments Trimer Oligomers 60mer Larger Particles

0.0% 6.5% oligomers 76.6% 6.0% 6.9%

Supplemental Figure 1 CHO‐K1

CHO‐BAFF

BAFF binding

Supplemental Figure 2 Supplemental Figure 3