CD40–CD40L interactions play a critical role in regulating immune responses. Blockade of CD40L by Abs, such as the anti-CD40L Ab 5c8, demonstrated positive clinical effects in patients with autoimmune diseases; however, incidents of thromboembolism (TE) precluded further development of these molecules. In this study, we examined the role of the Fc domain interaction with FcγRs in modulating platelet activation and potential for TE. Our results show that the interaction of the 5c8 wild-type IgG1 Fc domain with FcγRs is responsible for platelet activation, as measured by induction of PAC-1 and CD62P. A version of 5c8 with a mutated IgG1 tail was identified that showed minimal FcγR binding and platelet activation while maintaining full binding to CD40L. To address whether Fc effector function is required for immunosuppression, a potent Ab fragment, termed a “domain Ab” (dAb), against murine CD40L was identified and fused to a murine IgG1 Fc domain containing a D265A mutation that lacks Fc effector function. In vitro, this dAb–Fc demonstrated comparable potency to the benchmark mAb MR-1 in inhibiting B cell and dendritic cell activation. Furthermore, the anti-CD40L dAb–Fc exhibited a notable efficacy comparable to MR-1 in various preclinical models, such as keyhole limpet hemocyanin–induced Ab responses, alloantigen-induced T cell proliferation, “heart-to-ear” transplantation, and NZB × NZW F1 spontaneous lupus. Thus, our data show that immunosuppression and TE can be uncoupled and that a CD40L dAb with an inert Fc tail is expected to be efficacious for treating autoimmune diseases, with reduced risk for TE.
The CD40L and CD40 molecules are members of the TNF and TNFR superfamily, respectively, and mediate activation of innate and adaptive immune cells. CD40, which delivers activating intracellular signals, is prominently expressed on APCs, whereas CD40L, which has no signaling capacity, is notably induced on activated T cells and platelets. Postactivation, CD40L is cleaved rapidly from the cell surface, providing an important feedback mechanism that keeps CD40 activation in check. Indeed, soluble CD40L (sCD40L) levels are enhanced in various autoimmune diseases in both preclinical models and human patients (1–12). Moreover, the levels of sCD40L often track with disease severity or active versus quiescent stages of the disease; this is particularly evident in patients with systemic lupus erythematosus (5, 6). The engagement of CD40L with CD40 enhances the qualitative and quantitative nature of signals exchanged between a T cell and an APC. Some important outcomes of activating the CD40 pathway include upregulation of costimulatory molecules (such as CD80 and CD86) and adhesion molecules (such as ICAM-1); increased expression of both class I and II MHC molecules; production of proinflammatory cytokines (e.g., IL-1, TNF, and IL-6) and secretion of cytokines (e.g., IL-12 and IL-23) that aid with Th cell differentiation; and, in the case of B cells, the CD40 signals are obligatory for Ab class switch and affinity maturation. Human subjects expressing either a mutated CD40 or CD40L molecule suffer from an immunodeficient condition termed “hyper-IgM syndrome,” which is characterized by lack of IgG or IgA isotypes and leads to susceptibility to opportunistic infections (13, 14).
Because of its effect on multiple inflammatory cell types, the CD40–CD40L pathway has emerged as an attractive target for modulation of autoimmune diseases and transplant rejection. Targeted gene-knockout mice for CD40L exhibit impaired T and B cell responses and are protected from multiple experimental autoimmune diseases, including collagen-induced arthritis, experimental autoimmune encephalomyelitis, and type 1 diabetes (15). Furthermore, blockade of CD40L by mAbs results in protection from autoimmunity and graft rejection in various preclinical models (15). The substantial data in animal models of immune diseases propelled the clinical development of anti-CD40L mAbs. The clinical studies, primarily conducted with hu5c8 (anti-CD40L IgG1; Biogen) and IDEC-131 (anti-CD40L IgG1; Idec), gave meaningful results: efficacy with both molecules was noted in patients with immune thrombocytopenic purpura (16), and a positive clinical effect was noted with the Biogen molecule in lupus nephritis patients (17). However, these early clinical trials were marred with incidents of thromboembolism (TE), which precluded further development of these molecules (17). Subsequent to the clinical findings two major issues were raised: 1) what was the relationship between TE and anti-CD40L mAbs and 2) in addition to blocking CD40 binding, was the Fc-effector function a necessary component of the anti-CD40L mAb for efficacy? To this end, recent data suggested that binding of the Fc domain of anti-CD40L mAbs to the activating FcγRIIa (CD32a) receptor on platelets results in activation and aggregation of platelets (18). These data could explain the TE events with hu5c8 and IDEC-131, because both were of the IgG1 isotype that binds avidly to FcγRIIa. In contrast, the need for Fc-mediated effector function for therapeutic benefit, notably T cell depletion via Ab-dependent cell–mediated cytotoxicity, remains unresolved. The report by Ferrant et al. (19) concluded that Fc effector function was a necessary component for therapeutic benefit, whereas a detailed study by Waldmann’s group (20) suggested that an aglycosylated anti-CD40L IgG1 mAb (lacking Fc effector function) was equipotent to the wild-type IgG1 molecule in models of autoimmune diseases and transplantation.
Domain Abs (dAbs) are the smallest known Ag-binding fragments of Abs, ranging from 11 to 15 kDa (21). Each dAb contains three of six naturally occurring CDRs from an Ab, which are highly diversified loop regions that bind to target Ag. A dAb is either the variable domain of an Ab H chain (VH domain) or the variable domain of an Ab L chain (VL domain). dAbs are highly expressed in microbial cell culture; show favorable biophysical properties, including solubility and temperature stability; and are well suited to selection and affinity maturation by in vitro selection systems, such as phage display. dAbs are bioactive as monomers and, owing to their small size and inherent stability, can be formatted into larger molecules to create drugs with prolonged serum half-lives or other pharmacological activities. dAbs have demonstrated preclinical efficacy; currently, they are being evaluated in the clinic (22, 23).
In this study, we evaluated 5c8 and soluble isoleucine zipper (IZ)–CD40L immune complex (IC)–induced platelet activation using PAC-1 and CD62P markers on the platelets. Our results show that Fc–FcγRIIa interaction is a critical process; 5c8 with a modified IgG1* tail, which shows no binding to FcγRIIa, also demonstrated no activity in activating platelets. We also report the successful identification and generation of potent dAbs against murine CD40L (mCD40L). The mCD40L dAb, formatted as an Fc fusion with a modified murine IgG1 (mIgG1) Fc domain containing a D265A mutation that lacks Fc effector function, was equipotent to the benchmark MR-1 anti-mCD40L mAb in inhibiting immune responses in various animal models. Taken together, our data show that CD40L domain Abs with an inert Fc tail are expected to be efficacious for treating autoimmune diseases, with reduced risk for TE.
Materials and Methods
The genes encoding selected dAbs were cloned into pDOM13-Gas vector, as previously described (24). The genes for the coding regions of dAb–Fc and 5c8-IgG were synthesized from GeneArt in their entry clone vector pENTR221. Genes of interest were introduced into pTT22gate (National Research Council Canada) (25) or pUCOE-gate vectors using LR recombination. Productions were done using transient transfection or UCOE stable pool–expression systems. For transient expression, HEK293-6E cells (National Research Council Canada) (25) were transfected using a DNA/polyethylenimine ratio of 1:2 in F17 medium (Invitrogen, Carlsbad, CA) containing 4 mM l-glutamine and 0.1% pluronic F68 (Invitrogen). The day-5 conditioned medium was harvested for purification. For UCOE expression, CHO-S cells were used to generate a stable pool using DMRIE-C as transfection reagent. The production media contained a 1:1 mixture of CD CHO medium (Sigma-Aldrich, St. Louis, MO) and CHO medium (Invitrogen) with 1× TH (Invitrogen) and 8 mM final concentration of l-glutamine (Invitrogen). The day-7 conditioned medium was harvested for purification.
Purification of dAbs, Fc-fusion proteins, and Abs
V region genes for kappa L chains (Vk) dAbs were expressed in Escherichia coli as secreted proteins (24). The clarified supernatants were loaded on a column of immobilized protein L (Pierce, Rockford, IL), washed with PBS, and eluted with 0.1 M glycine (pH 2). The neutralized eluate was run on a Superdex 75 column in PBS running buffer to remove aggregate. The Vk dAb–Fc, 2m126-24–mIgG1 (D265A), contains a TVAAPS spacer between the Vk and mouse IgG1 (D265A) Fc domains. This molecule was expressed in HEK293-6E cells, the resulting day-5 conditioned medium was clarified, and 4 M NaCl was added. The dAb–Fc was purified on a Protein A Sepharose Fast Flow column (GE Healthcare, Piscataway, NJ), followed by a Superdex 200 column (GE Healthcare). All Abs and mouse CD40 (1–193) fused to a human IgG1 Fc were purified directly from the culture supernatant using a protein A column and eluted with 80 mM sodium acetate (pH 3). The neutralized eluates were run on a Superdex 200 column in PBS running buffer.
Purification of IZ-CD40L
CD40L was fused to a His-tagged trimerizing IZ domain at the N terminus (26). IZ-human CD40L (108–261) and IZ-mouse CD40L (107–260) were expressed in CHO cells. Clarified supernatants were purified using a three-step procedure involving Ni-NTA affinity chromatography (QIAGEN, Toronto, ON, Canada), followed by hydrophobic-interaction chromatography using a phenyl Sepharose column (GE Healthcare) and size-exclusion chromatography on a Superdex 200 column. His–IZ–CD40L was biotinylated using a 5-fold molar excess of Sulfo-NHS-LC-LC-biotin (Pierce), with unreacted reagent and reaction byproducts removed on a Sephadex G-25 column.
Generation of 5c8 Ab fragments
The 5c8 IgG1 (27) was converted to a F(ab′)2 by incubation with 1% pepsin (Sigma-Aldrich) in 50 mM sodium citrate (pH 3.5) for 5 h at room temperature. The reaction was quenched by adding Tris to a final pH of 8. The F(ab′)2 was collected in the flow fraction of a Protein A column and run on a Superdex 200 column. LPS was lowered by passage through either Sartobind Q (Sartorius Stedim Biotech) or Mustang Q membranes in-line with an Acrodisc Unit (Pall). LPS was measured using an Endosafe-PTS unit (Charles River, Wilmington, MA), according to the manufacturer’s instructions.
Surface plasmon resonance
Surface plasmon resonance (SPR) studies were performed on a Biacore T100 instrument (GE Healthcare) at 25°C. The binding of 5c8-IgG1 or 5c8-IgG1* analytes was tested in 10 mM NaPO4, 130 mM NaCl, 0.05% p20 (PBS-T) (pH 7.1) on surfaces consisting of biot-IZ-human CD40L (hCD40L) that had been captured on a streptavidin sensor chip (GE Healthcare). The binding of monovalent dAbs, bivalent dAb–Fcs, or MR-1 Ab was evaluated in PBS-T (pH 7.1) on surfaces consisting of IZ-mCD40L that had been directly immobilized on a CM5 sensor chip (GE Healthcare) using standard ethyl(dimethylaminopropyl) carbodiimide/N-hydroxysuccinimide (NHS) chemistry, with ethanolamine blocking. The ability of dAbs or dAb–Fcs to block the binding of mCD40L-Fc to IZ-mCD40L surfaces was evaluated using the “dual-injection” function in T100 control software.
Isothermal titration calorimetry
Isothermal titration calorimetry (ITC) experiments were performed on a VP-ITC microcalorimeter (Microcal; GE Healthcare) in 10 mM sodium phosphate, 130 mM sodium chloride [pH 7.1]. Experiments consisted of 10-μl injections of 13.3 μM IZ-mCD40L trimer titrated into 3–4 μM 2m126-24 or 2m126-24–Fc (dAb molar concentration) or 10-μl injections of 40 μM 2m126-24 or 2m126-24–Fc titrated into 1.33 μM IZ-mCD40L trimer at either 25 or 37°C. Data were fitted using the “one set of sites” binding model supplied in the ITC for Origin 7.0 analysis software.
Static and dynamic light scattering
The oligomeric state of IZ-mCD40L:2m126-24–Fc complexes was studied by size-exclusion chromatography coupled to an in-line multiangle light scattering detector (SEC-MALS) or by dynamic light scattering (DLS). Samples were prepared by mixing IZ-mCD40L (3 mg/ml) and 2m126-24–Fc (5.3 mg/ml) in 20 mM NaPO4 and 150 mM NaCl (pH 7.2) at different molar ratios and then analyzed by SEC-MALS and DLS. Isocratic separations were performed on a Tosoh 3000 Super SW column connected to an Agilent 1100 series HPLC in buffer containing 200 mM K2HPO4, 150 mM NaCl (pH 6.9), and 0.02% Na azide (0.1 μm filtered) running at 0.5 ml/min. Samples were injected onto the column using an Agilent autosampler, and data were obtained from three online detectors connected in series: an Agilent diode array UV/vis spectrophotometer, followed by a Wyatt Technologies DAWN HELEOS II eighteen angle laser light scattering detector and then a Wyatt Optilab DSP interferometric refractometer. Data were collected and analyzed using Astra (Wyatt) and ChemStation (Agilent) software. DLS data were collected on a Wyatt DynaPro plate reader at 25°C, using 10 acquisitions of 5 s each, and measurements were recorded in triplicate and averaged to give the reported values. Intensity autocorrelation functions were fitted using the “Regularization” algorithm in the Dynamics software (Wyatt Technologies).
Platelet activation assay
Platelet activation was detected by flow cytometry using Abs against the well-established platelet activation markers P-selectin (CD62P) and PAC-1 (activated GPIIb/IIIa). Briefly, blood from healthy volunteer donors was drawn into glass sodium citrate tubes, mixed gently four times by inversion, and tested within 30 min. Blood was diluted 1:25 in modified Tyrode’s-HEPES buffer (10 mM HEPES, 0.9% NaCl, 3 mM KCl, 1 mM MgCl2, 0.1% BSA) containing 1 mM CaCl2 to which detection Abs (CD62P-PE and PAC1 FITC; BD Biosciences) and test reagents, such as sCD40L complexed to 5c8 constructs or ADP, a direct platelet agonist, were added, incubated for 30 min at room temperature, and diluted with PBS. In some groups, IV.3 (an FcγRIIa Ab) was incubated with blood samples for 30 min at 10 μM prior to addition of sCD40L/5c8 IC or ADP. Platelets were analyzed for activation by flow cytometry using a BD FACSCalibur. Initial experiments determined that sCD40L or 5c8-IgG1 alone did not activate platelets, but different IC ratios of 1:1–1:8 of 5c8:sCD40L significantly activated platelets. Subsequent experiments used 5c8:sCD40L at 1:3.5 molar ratio.
Soluble IZ-mCD40L–driven murine primary B cell–proliferation assay
A total of 1 × 105 mouse splenic B cells was incubated in complete RPMI 1640 medium containing 100 μg/ml gentamicin (Life Technologies, Grand Island, NY), 10% v/v FCS (Summit, Fort Collins, CO), and 5 × 10−5 M 2-ME (Sigma-Aldrich) with 1 μg/ml IZ-mCD40L (generated in-house), along with serial dilutions of dAb, dAb–Fc, or mAb in a final volume of 200 μl/well in a 96-well round-bottom plate (BD Falcon, Bedford, MA). The plates were incubated at 37°C, 5% CO2 for 72 h, following which 0.5 μCi/well [3H]thymidine (PerkinElmer, Waltham, MA) was added for 6 h. B cell proliferation was quantitated based on [3H]thymidine incorporation.
CHO-mCD40L–driven murine primary B cell–proliferation assay
CHO cells were transfected with mCD40L (CHO-mCD40L) to generate a stable cell line expressing high levels of CD40L on the cell surface. Preirradiated CHO-mCD40L cells (1 × 103; 10,000 rad) were cultured with mouse splenic B cells (1 × 105) in complete RPMI 1640 medium along with serial dilutions of dAb, dAb–Fc, or mAb in a final volume of 200 μl/well in a 96-well round-bottom plate (BD Falcon). The plates were incubated at 37°C, 5% CO2 for 72 h, following which 0.5 μCi/well [3H]thymidine was added for 6 h. B cell proliferation was quantitated based on [3H]thymidine incorporation.
Murine dendritic cell cytokine assay
Murine dendritic cells (DCs) were generated by culturing bone marrow cells from femurs of BALB/c mice in complete media with 20 ng/ml recombinant murine GM-CSF (PeproTech, Rocky Hill, NJ) in six-well plates (BD Falcon) for 6 d at 37°C, 5% CO2. CHO-mCD40L cells were irradiated (10,000 rad) and cultured at 5 × 103 cells/well in 96-well flat-bottom plates (BD Falcon) in complete media in the presence or absence of serial dilutions of dAb–Fc or mAb for 1 h at 37°C, 5% CO2. DCs were washed extensively and added to the pretreated CHO-mCD40L cells at a 20:1 ratio (1 × 105 cells/well), followed by the addition of 1 μg/ml LPS (Sigma-Aldrich). Plates containing the cell mixtures were incubated for 48 h at 37°C, 5% CO2.
DBA/2 mice were obtained from The Jackson Laboratory (Bar Harbor, ME). BALB/c and pregnant C57BL/6 mice were from Harlan Laboratories (Indianapolis, IN). NZB × NZW F1 mice were obtained from Charles River. All procedures involving animals were reviewed and approved by the Bristol Myers Squibb Institutional Animal Care and Use Committee.
Keyhole limpet hemocyanin–induced Ab response in mice
Female BALB/c mice (8–10 wk old) were injected i.p. with 250 μg keyhole limpet hemocyanin (KLH) on day 0. Mice were dosed s.c. with MR-1 (hamster anti-mouse CD40L Ab) or anti-CD40L dAb (2m126-24-Fc) on days −1 and +6. Blood was collected (retro-orbital), and the serum was analyzed for anti-KLH IgM on day 7 and IgG on day 14 by ELISA. Serum from BALB/c mice collected on day 14 after immunization with KLH was pooled and used as a positive comparator; the data are expressed as a ratio of the titer of the test serum/pooled BALB/c serum.
In vivo alloantigen-induced T cell response
Male BALB/c mice (8–10 wk old) were given footpad injections of 1 × 107 splenocytes from male DBA-2 mice (8–10 wk old). MR-1 or mCD40L dAb–Fc (2m126-24–Fc) was administered s.c. at the indicated doses the day before the cell injections. The draining popliteal nodes (DLNs) were collected 3 d after alloantigen challenge, and DLN cells were counted on a Vi-Cell XR Cell Viability Analyzer (Beckman Coulter, Brea, CA).
Mouse “heart-to-ear” transplantation model
Heart grafts from neonatal (48–72 h) C57BL/6 mice were implanted into a s.c. pocket created in the ear pinnae of female BALB/c mice. Mice were treated with CTLA-4 Ig (i.p. twice a week), mCD40L dAb–Fc (2m126-24–Fc, s.c. once a week), or a combination of both, with the dosing initiated the day prior to transplantation. Time to rejection was defined by the absence of cardiac contractility for three consecutive days, as assessed daily by electrocardiography.
NZB × NZW F1 spontaneous lupus model
Female NZB × NZW F1 mice (22–24 wk old) were treated with PBS, MR-1 (20 mg/kg), or mCD40L dAb–Fc (2m126-24–Fc; 5 or 20 mg/kg) s.c. once a week (n = 13/group). Mice were monitored weekly for body weight, survival, and proteinuria by dipstick (Fisher; 0: negative; 0.5: trace; 1: 30 mg/dl; 2: 100 mg/dl, 3: 300 mg/dl; 4: >2000 mg/dl). Mice were bled every other week for anti-dsDNA Abs measured by ELISA. The data are expressed in arbitrary units as a ratio of the titer of the test serum/pooled serum from MRL/lpr mice. At the end of the study, one kidney from each mouse was collected in RNA Later and homogenized in mRNA Catcher Lysis Buffer with a tissue lyser. mRNA was purified using mRNA Catcher PLUS, according to the manufacturer’s protocol (Invitrogen). cDNA was synthesized using SuperScript II with Random Hexamers primer. PCR was performed with SYBR Green Master Mix (Invitrogen). Relative quantification analysis was determined with the 2−ΔΔCT method, using cyclophilin (PPIA) as the housekeeping gene. Expression of inflammatory cytokines and IFN-inducible genes were analyzed. The following primer sequences were used: Irf1: 5′-CCTTCCACCTCCGAAGCCGC-3′ and 5′-TGTGTCGGCTGCCACTCAGAC-3′; Irf7: 5′-GAGTCTGGGGCAGACCCCGT-3′ and 5′-CTGCGCTCGGTGAGAGCTGG-3′; Irf9: 5′-GGCTCAGGCCCTGCCCATTTCTT-3′ and 5′-CTCCACGATCCAGCTCCGCA-3′; Ifit1: 5′-AGAGCAGAGAGTCAAGGCAGGT-3′ and 5′-TGGTCACCATCAGCATTCTCTCCCA-3′; Ifit3: 5′-GCTCAGCCCACACCCAGCTTT-3′ and 5′-AGATTCCCGGTTGACCTCACTCAT-3′; Ligp1: 5′-GGACACAGGAGTTTCTGTGCCTTT-3′ and 5′-AGGTGAAGAGAACAGCTGACCCA-3′; Stat1: 5′-GCTGAGTTCCGACACCTGCAAC-3′ and 5′-AGCCTGGCTGGCACAACTGG-3′; Stat2: 5′-CCCGGAGGAGGGCCAGAGAC-3′ and 5′-GCGGCCCCTTGTTGCCCTTT-3′; Gbp2: 5′-AGCTGCTAAACTTCGGGAACAGGA-3′ and 5′-AGAGGTTTGGGCCTTGGGCCT-3′; Cxcl9: 5′-GCAGTGTGGAGTTCGAGGAACCCT-3′ and 5′-GGATCGTGCCTCGGCTGGTG-3′; and Fcgr4: 5′-AGCCTAGGCGATCCAGGGTCT-3′ and 5′-TGAAGACCCCTCCGCACAGA-3′.
The remaining kidney was divided for collection into 10% neutral buffered formalin and ZincTris fixatives. Fixed tissues were routinely processed into paraffin blocks, sectioned at 3 or 5 μm, and stained with H&E, period acid–Schiff, and Masson’s trichrome stains (Histology Consultation Services, Everson, WA). Kidney sections were graded using previously described criteria (28). Glomerular damage was assessed based on mesangial matrix thickening, cell proliferation; crescent formation, cellular deposits/casts in Bowman’s space; cellular infiltration, mononuclear cells in glomerular tufts; and fibrosis of Bowman’s capsule. Tubular damage was assessed based on infiltration of mononuclear cells, severity of tubular damage, and protein casts. Tubulointerstitial damage was assessed based on fibrosis and infiltration of mononuclear cells. Each subcategory was assigned a score from 0 to 4, with the scores for glomerular indices representing the mean from 20 glomeruli/kidney. The total score for each mouse was the sum of the above nine subcategories, with a maximum possible score of 36. Kidney sections also were stained immunohistochemically to evaluate IgM (biotinylated goat anti-mouse IgM) and IgG (goat anti-mouse IgG; both from Vector Laboratories, Burlingame, CA) deposition in the kidney.
Fc mutations abolish platelet activation through an FcγRIIa-dependent mechanism
We established an experimental condition in which platelet activation is induced by the IC of soluble IZ-tagged IZ-hCD40L trimer and the anti-CD40L Ab, 5c8-IgG1, as measured by the induction of PAC-1 and CD62P markers on the platelets by flow cytometry. These markers were chosen specifically because they are sensitive and established biomarkers that represent the earliest stages of platelet activation. FcγRIIa plays a critical role in the process, because the blocking Ab against FcγRIIa, IV.3, specifically inhibited IC-induced platelet activation while showing no effect on the platelet activation induced by ADP, a direct platelet agonist (Fig. 1A). We then generated a series of 5c8 Abs with Fc domains containing mutations that were designed to disrupt FcγR binding, in particular, FcγRIIa present on the surface of platelet cells. Using SPR, these 5c8 molecules were tested for binding to human IZ-CD40L, as well as to the “high-affinity” human FcγRI and the “low-affinity” human FcγRIIa, FcγRIIb, FcγRIIIa, and FcγRIIIb. The data for one of these molecules, containing a modified IgG1 Fc domain denoted as IgG1* [the same IgG1 Fc used in abatacept (29, 30)], were compared with 5c8-IgG1 (Fig. 2). Our data showed that the Fc mutations had no impact on 5c8 binding to the CD40L target protein (Fig. 2A, 2B), but they significantly reduced the binding to FcγRI, as seen by the faster dissociation rate, and resulted in undetectable binding to the low-affinity FcγRs, including FcγRIIa. Subsequently, 5c8-IgG1 and 5c8-IgG1* were tested in platelet-activation assays. As shown in Fig. 1B, 5c8-IgG1 showed platelet activation, as expected. On the contrary, 5c8-IgG1* showed no or extremely low activity in the platelet assays, comparable to what is observed with 5c8-F(ab)′2 that lacks the Fc tail.
Generation of anti-mouse CD40L dAb molecules
The technology detailing the in vitro selection and maturation of dAbs was described previously (31). Briefly, to generate anti-mCD40L dAb molecules for use in preclinical mouse models, we performed in vitro dAb selections using the mouse CD40L to identify initial hits that were then subjected to successive rounds of affinity maturation to select for clones that had increased target-binding affinity and potency. Table I lists examples of three dAb lineages wherein gain in potency in the primary B cell assay was noted between the parent clone and their matured progeny. In all cases the dAbs were substantially more potent at inhibiting B cell proliferation induced by the soluble IZ-mCD40L protein and less potent at inhibiting B cell proliferation driven by CHO cells expressing mCD40L. Although maturation resulted in gain of potency, the range of dAb potency, even for the matured molecules, was weaker compared with the benchmark MR-1 mAb, especially in the CHO-mCD40L–driven assay (Table I).
Fc formatting significantly improves dAb potency
As a strategy to improve the pharmacokinetic properties, as well as potentially improve potency, several lead dAbs were formatted as Fc fusions. The potency of the bivalent dAb–Fc molecules was tested in primary B cell assays (soluble IZ-mCD40L– or CHO-mCD40L–driven B cell proliferation), as well is in a DC assay (DC IL-6 production driven by CHO-mCD40L), in which a notable gain in potency was evident for dAbs from each lineage upon Fc formatting (Table I). The gain in potency was considerably more pronounced in the CHO-mCD40L–driven assays (120 to >6000-fold) compared with the soluble IZ-mCD40L–driven assay (5–13 fold), suggesting a significant contribution of avidity in dAb–Fc binding to mCD40L on the CHO cell surface, which is not observed with the soluble trimeric ligand.
Mechanism of action of mCD40L dAb–Fcs
To characterize the mechanism of action for the dAb–Fc molecules in more detail, we investigated the interactions between dAb or dAb–Fc molecules and IZ-mCD40L trimer using several biophysical techniques, including SPR to represent binding to surface-presented mCD40L, ITC to represent binding to soluble mCD40L, as well as static light scattering and DLS to investigate the oligomeric state of the complexes.
SPR experiments on immobilized IZ-mCD40L surfaces showed that the dAb–Fcs, like MR-1, had slower dissociation rates than the monovalent dAbs, consistent with the expected impact of avidity due to increased valency and binding to a surface-immobilized target (Fig. 3A, 3B). dAbs, dAb–Fcs, and MR-1 also efficiently blocked the binding of mCD40L-Fc to IZ-mCD40L, with dAb–Fcs and MR-1 being more effective blockers than the monovalent dAbs, likely due to the slower dissociation kinetics (Fig. 3C). In contrast to the surface-based SPR studies, solution-based ITC data showed similar thermodynamics for IZ-mCD40L binding to the monovalent 2m126-24 (Kd = 7 nM) compared with the bivalent 2m126-24–Fc (Kd = 5 nM) (Fig. 3D). This implies that each dAb domain of the dAb–Fc may bind to a separate IZ-mCD40L trimer, rather than cooperatively binding to the same IZ-mCD40L trimer molecule. Solution-based light scattering data supported this hypothesis, showing that mixtures of IZ-mCD40L with 2m126-24–Fc in different ratios form higher-order complexes in solution rather than forming a defined 1:1 complex (Fig. 3E, 3F). In particular, SEC-MALS data identified the mass for the dominant, most thermodynamically stable state, as ∼410–530 kDa, which is consistent with a valency-satisfied complex of three IZ-mCD40L trimers (71.4 kDa each) cross-linked by approximately three or four dAb–Fc molecules (∼75.5 kDa each). A schematic model is shown in Fig. 3G. Similarly sized higher-order structures were observed for Abs in complex with other TNF family proteins, including 5c8 and other human CD40L-binding mAbs, as well as anti-TNF mAbs (27, 32–34). Confocal microscopy showed that these small soluble complexes can translate into even larger ordered structures on the cell surface, presumably driven by avidity (32). Therefore, the higher-order complexes observed for 2m126-24–Fc bound to IZ-mCD40L suggest that dAb–Fcs have the potential to form similar-ordered structures on the cell surface, which may explain the dramatically improved potency in the CHO-mCD40L assay. In contrast, the less potent monovalent dAb, 2m126-24, which should not be capable of forming these ordered complexes, was found to form an ∼106-kDa complex with IZ-mCD40L in solution, as measured by SEC-MALS; this is consistent with the expected complex of three dAbs (12 kDa each) bound to one IZ-mCD40L trimer (data not shown). Collectively, these data suggest that the dramatically increased potency for dAbs upon Fc formatting may be related to the ability of dAb–Fcs to bring together multiple CD40L molecules to form high-affinity ordered and cross-linked structures on the cell surface, similar to 5c8 or other bivalent TNF family antagonists.
In vivo efficacy of a mouse CD40L dAb–Fc
The Fc-formatted dAbs exhibited comparable potencies to the benchmark MR-1 mAb in the various primary cell assays (Table I), which enabled us to compare them in various in vivo mouse models. In particular, 2m126-24–Fc was chosen for in vivo studies because it was equipotent to MR-1 in both the B cell and DC assays. This molecule was engineered to contain a murine IgG1 Fc domain with a D265A point mutation (2m126-24–mIgG1–D265A), which is known to disrupt FcR binding (35). The lack of binding of 2m126-24–mIgG1–D265A to murine FcRs mCD64, mCD32b/c, mCD16, or mCD16-2 was confirmed using SPR (36). In the same study, efficient binding to the murine FcRs was observed with 2m126-24 fused to wild-type murine IgG2A (36). To address whether Fc effector function is a necessary component for immunosuppression, this anti-CD40L dAb molecule containing an “inert” Fc domain was evaluated in various animal models along with the benchmark MR-1 mAb.
KLH injection in mice induces T cell–dependent Ab responses, as measured by serum anti-KLH IgM and IgG titers. The control MR-1 Ab, dosed once a week s.c., resulted in a dose-dependent reduction in IgG titers, with ED50 calculated to be 0.25 mg/kg. 2m126-24–Fc, also dosed once a week s.c., resulted in a dose-dependent suppression of IgG titers with ED50 of 0.26 mg/kg, comparable to the MR-1 control (Fig. 4).
Local administration of alloantigen induces T cell expansion in the DLNs. The DBA-2 splenocytes were injected into BALB/c recipients through the footpad, which led to a 7-fold increase in the numbers of DLN cells (Fig. 5). This expansion of DLN cells was inhibited by either MR-1 or 2m126-24–Fc (both s.c. once a week) in a dose-dependent fashion with similar ED50 (∼2 mg/kg).
In another permutation, the efficacy of blocking CD40–CD40L was evaluated in a transplantation model. Implantation of neonatal heart grafts into the ear of MHC-mismatched recipients results in graft rejection. The typical median survival time is 12 d when C57BL/6 neonatal hearts are used as donors and BALB/c mice are used as recipients. The monotherapies with 3 or 20 mg/kg of 2m126-24–Fc or 25 mg/kg of CTLA-4 Ig, given once a week s.c., had little or no impact on graft survival. However, in the groups treated with a combination of 20 mg/kg 2m126-24–Fc and 25 mg/kg CTLA-4 Ig, the graft survival was significantly prolonged, with a median survival time of 35 d (Fig. 6).
NZB × NZW F1 is a classical model of spontaneous lupus that develops severe lupus-like phenotypes comparable to that of lupus patients. These lupus-like phenotypes include lymphadenopathy, elevated serum anti-nuclear autoantibodies, and IC-mediated glomerulonephritis. NZB × NZW F1 female mice, aged 22–24 wk, were used for the study. 2m126-24–Fc at both doses (5 and 20 mg/kg) and MR-1 (20 mg/kg) given s.c. once a week markedly reduced the serum dsDNA titers (Fig. 7A) and the levels of proteinuria (Fig. 7B). The body weight loss was prevented (Fig. 7C), and the survival rate was increased (Fig. 7D) in all treatment groups. Furthermore, a significant reduction in gene expression of IFN-inducible genes (Irf1, Irf7, Irf9, Ifit1, Ifit3, Iipg1, Stat1, Stat2, Gbp2, Cxcl9, and Fcgr4) was observed along with inflammatory cytokines/chemokines (Fig. 7E, data not shown). These data are in agreement with a recent report by Duffau et al. (4) that identified a key role for CD40 activation in type 1 IFN production by plasmacytoid DCs in lupus patients. Microscopic evaluation revealed significant glomerulonephritis in the kidneys of the PBS-treated control mice. This was characterized by varying degrees of hypercellularity and matrix deposition in the glomeruli, protein casts in tubular lumens, and inflammatory cell infiltrates, primarily composed of lymphocytes and plasma cells, in the interstitial and perivascular tissues. Immunohistochemistry revealed deposition of IgM and IgG in the glomeruli, tubules, and interstitium. Treatment with 2m126-24–Fc or MR-1 significantly reduced histological changes (Fig. 8). Taken together, these findings show that a mouse CD40L dAb formatted with a modified mIgG1 that lacks Fc effector function is equipotent to anti-CD40L mAb MR-1 in inhibiting T cell proliferation, Ab production, and lupus nephritis in animal models.
Several key findings emerged from our current study. First, we confirmed that Fc–FcR interactions contribute toward activation of platelets and that this is likely the molecular explanation for anti-CD40L mAb-induced thromboembolic events observed in previous human clinical studies. Second, we demonstrate that anti-CD40L dAbs with inert Fc molecules can be generated to avoid the potential toxicities, including TE, related to platelet activation. And, third, we provide data in various in vivo models that support the conclusion that Fc effector function is not an obligatory requirement for the biological efficacy of anti-CD40L molecules.
The previous experience with three anti-CD40L mAbs inducing platelet activation in human subjects and nonhuman primates highlighted the potential high-risk liability of targeting CD40L. However, all of the previous anti-CD40L mAbs were of the IgG1 isotype, which suggested a shared functional phenotype that may have contributed to platelet activation. Understanding the relationship between Fc and platelet activation is perhaps most directly addressed via controlled in vitro assessments of platelet biology. To this end, our results and those from Robles-Carillo et al. (18) confirm that IgG1 binding to platelet CD32a (FcγRIIA) is likely the link to susceptibility to TE. Interestingly, other IgG1 Abs against TNF superfamily members (including numerous anti–TNF-α mAbs) have been in the clinic with no apparent incidents related to TE; hence, why the predisposition with targeting CD40L? We believe that this may be due to the fact that many inflammatory diseases (such as lupus, inflammatory bowel disease, and others) exhibit a prothrombotic state characterized by a significant increase in the numbers of activated platelets expressing high levels of cell surface CD40L. These activated platelets can be bound by anti-CD40L mAbs that further perpetuate this state by cross-linking other platelets via the IgG1 Fc binding to the CD32a receptor on platelets. Such a reaction may lead to platelet aggregation and, ultimately, thromboembolic events. In addition, another component may involve complexes of sCD40L+anti-CD40L mAbs that may engage platelets directly via CD32a, leading to activation and potential aggregation. Of note, the mouse platelets do not express CD32a, which was a major reason why this aspect of CD40L targeting and biology was not evident in preclinical mouse models.
The current study also highlights the unique advantages of using Ab-engineering technology, such as dAbs, which can be used effectively to generate high-affinity molecules against biological targets of interest (31). Because dAbs are derived from a single variable domain of either the H or L chain, this may allow access to unique targeting domains based on the small dAb footprint. In this study, we were able to identify and mature potent anti-CD40L dAb molecules, many of which exhibited a notable increase in potency upon targeted affinity maturation. The fusion of a bivalent dAb with Fc improved pharmacokinetics properties, as well as potency, with a significant contribution of avidity in dAb–Fc binding to CD40L-expressing cells.
In conclusion, our data support revisiting a CD40L-blockade approach in the clinical setting. A potent CD40L dAb fused with an inert Fc tail is expected to be efficacious for treating autoimmune diseases with reduced risk for TE.
The authors have no financial conflicts of interest.
We thank Mike Doyle for discussions on the biophysical binding molecular mechanisms, Mian Gao for discussions on cloning and expression strategy, and Preeti Sejwal for assistance with SEC-MALS experiments.
Abbreviations used in this article:
- domain Ab
- dendritic cell
- draining lymph node
- dynamic light scattering
- human CD40L
- immune complex
- isothermal titration calorimetry
- isoleucine zipper
- keyhole limpet hemocyanin
- murine CD40L
- murine IgG1
- soluble CD40L
- size-exclusion chromatography coupled to an in-line multiangle light-scattering detector
- surface plasmon resonance
- V region genes for kappa L chains.
- Received December 5, 2013.
- Accepted February 19, 2014.
- Copyright © 2014 by The American Association of Immunologists, Inc.