Fully recombinant IgG2a Fc multimers (stradomers) effectively treat collagen-induced arthritis and prevent idiopathic thrombocytopenic purpura in mice

Mice

We used 6- to 8-week-old, female C57BL/6 mice (Charles River Laboratories, Wilmington, MA, USA, Catalog 027), female F1(C57BL/6xDBA/2) (Jackson Laboratories, Bar Harbor, ME, USA, catalog 100006), FcγR2b^-/- (Jackson, Catalog 002848), and TLR4^-/- mice (Stefanie Vogel, University of Maryland) were used [11]. All experiments, except those contracted to outside vendors, were approved by the Animal Care and Use Committee of the University of Maryland School of Medicine. Two contract research organizations paid by Gliknik, Inc., performed CIA experiments in separate mouse lines. Washington Biotechnology (Baltimore, MD, USA) used DBA1/J mice for their experiments, and Bolder-BioPATH (Boulder, CO, USA) used DBA/1OlaHsd mice (Harlan Laboratories, Frederick, MD, USA).

Generation of stradomer and IgG2a Fc cDNA constructs

The Fc regions of the immunoglobulin sequence were fused to MDs either by splicing by overlap extension PCR or by using existing compatible restriction sites [12]. Polymerase chain reaction (PCR) primers were generated encoding the MD and an overlapping region of either the C or the N-terminus of the IgG Fc region and used for fusing the two sequences by PCR. The PCR products were then cloned into pcDNA3.3 by TA cloning (Invitrogen, Carlsbad, CA, USA) and confirmed by sequencing. Stradomers were transiently expressed in HEK293 cells after transfection.

Purification of stradomers

Purification was done by using automated chromatography (Aktaxpress, GE, Piscataway, NJ, USA) and Hitrap Mabselect, protein A-derived ligand columns (GE 28-4082-56), followed by desalting on a HiPrep 26/10 Column (GE 17-5087-01). Stradomers were separated into homodimeric and multimeric fractions by using the standard Aktaxpress purification protocol and a GE Hiload 16/60 Superdex 200-pg size exclusion column (GE 17-1069-01).

SDS-PAGE analysis

From 0.5 to 2 μg of nonreduced or reduced sample was loaded onto Nupage Novex 4-12% Bis-Tris Mini Gels (Invitrogen, Carlsbad, CA, USA, NPO322BOX) and were run according to the manufacturer's protocol (Invitrogen, IM-8042 Version H). Gels were washed with deionized water, stained for 30 minutes at room temperature with Bio-safe Coomassie Dye (Bio-Rad, Hercules, CA, USA, 161-0786), and destained with water.

Surface plasmon resonance

Surface Plasmon resonance (SPR) was performed on a GE Healthcare Biacore 3000 instrument or on a Biacore T100. Recombinant mouse FcγRI, FcγRIIb, FcγRIII, FcγRIV, and SIGN-R1 (RnD Systems; Minneapolis, MN, USA) were immobilized onto a CM4 (FcγRs) or CM5 (SIGN-R1) chip by amine coupling. Samples were serially diluted in HBS-EP (running buffer) from 1,000 nM to 0.5 nM and injected at 20 μl/min, followed by a dissociation phase. Flow cells were regenerated with 1 M MgCl at 100 μl/min and then washed with HBS-EP (FcγRI required 10 mM glycine pH 2.0 to regenerate). A blank flow cell was used as a reference. All samples were run in duplicate or triplicate. The KD was calculated by using a 1:1 Langmuir kinetic model. Unless otherwise specified, all reagents were from GE Healthcare (Piscataway, NJ, USA). Because many of the preparations used in our studies contained multimers of diverse molecular masses, we standardized our analysis by assigning a molecular mass of 50 kDa to homodimer isolates and 150 kDa to all other protein preparations.

Biolayer interferometry

Biolayer interferometry assay was performed on an Octet 96Red Instrument (ForteBio, Menlo Park, CA, USA) according to the manufacturer's instructions (Data Acquisition User Guide 6.4, ForteBio). For binding analysis, His-tagged mouse FcγRIIb (R&D Systems, Minneapolis, MN, USA, 1460-CD) and FcγRIII (R&D Systems 1960) were loaded on anti-penta-His Biosensors (ForteBio 18-5077) at 10 μg/ml in 1× kinetic analysis buffer (ForteBio,18-5032). After sensor-tip loading, protein association was measured by transfer of tips to preparations of either the monomeric (concentration 2,000 nM to 62.5 nM) or multimeric fractions (concentration, 100 nM to 3.13 nM) in 1× kinetics analysis buffer, and dissociation was measured by transfer of sensor tips to 1× kinetics buffer. Analysis was as described (Data Analysis User Guide 6.4ForteBio). Analysis was standardized, as described earlier by assigning a MW of 50 to homodimers and 150 to all other protein preparations.

Collagen-induced arthritis model

Collagen-induced arthritis experiments were performed by two contract research organizations (CROs; Washington Biotechnology, Inc., Columbia, MD, USA; Bolder BioPATH Inc., Boulder, CO, USA). All arthritis experiments, except for the dose-response data for 2A-2H_C, were generated by Washington Biotechnology by using the following protocol: On day 0, DBA1/J mice were immunized with 50 μl of a 4-mg/ml solution of Type II bovine collagen (Chondrex, Inc., Redmond, WA, USA, Cat. 20021) emulsified with an equal volume of complete Freund's Adjuvant (Sigma, St. Louis, MO, USA, Cat. 5506). A second vaccination was given on day 21 by using the same dose of Type II bovine collagen emulsified with an equal volume of incomplete Freund's Adjuvant. Mice were scored daily for signs of arthritis. Each paw was scored, and the sum of all four scores was recorded as the Arthritic Index (AI). The maximum possible AI was 16, as follows: 0, no visible effects of arthritis; 1, edema and/or erythema of one digit; 2, edema and/or erythema of two joints; 3, edema and/or erythema of more than two joints; and 4, severe arthritis of the entire paw and digits, including limb deformation and ankylosis of the joint. Starting on day 28 (treatment day 0), 10 collagen-immunized mice (all manifesting disease) were sorted into each of the treatment groups based on average AI (3.3). AI was measured for 14 treatment days, after which mice were euthanized. For the positive control group, mice were dosed orally with 10 mg/kg prednisolone daily. For the 2A-2H_C group, mice were dosed every fourth day (day 0, day 4, day 8, and day 12) with 400 μg 2A-2H_C (17.4 mg/kg). Fractionated 2A-2H_C was given at a lower dose (250 μg, 10.9 mg/kg).

A second contract research organization, Bolder BioPATH Inc. (Boulder, CO, USA), independently generated dose-response data to 2A-2H_C in a CIA model by using a slightly different protocol. In brief, male DBA/1OlaHsd (Harlan Laboratories, Frederick, MD, USA) mice were anesthetized with isoflurane and injected with 150 μl of Bovine Type II collagen (Elastin Products, Owensville, MO, USA), in Freund's complete adjuvant on day 0 and day 21 (Freund's complete adjuvant [with supplemental Mycobacterium. tuberculosis, 4 mg/ml], Difco). On study days 24 through 27, onset of arthritis occurred, and mice were randomized to treatment groups. Animals were randomized to a treatment group only after swelling was obviously established in at least one of four paws (score of 1 for the paw, minimum mean score of 0.25 for the animal). Attempts were made to assure approximately equal mean scores across all groups at the time of enrollment. Treatment was initiated after enrollment (treatment day 0). Mice were treated with intravenous 2A-2H_C given at varying doses (10 mg/kg, 20 mg/kg, and 40 mg/kg). The first dose was administered on day 0 and then every fourth day. IVIG was given at a dose of 2 g/kg by intraperitoneal injection on day 0 and every fourth day. Dexamethasone was given orally at a dose of 0.2 mg/kg daily beginning on day 0. During this time, clinical scores were given for each of the paws (right front, left front, right rear, left rear) as follows: 0, normal; 1, 1 hind- or forepaw joint affected or minimal diffuse erythema and swelling; 2, 2 hind- or forepaw joints affected or mild diffuse erythema and swelling; 3, 3 hind- or forepaw joints affected or moderate diffuse erythema and swelling; 4, marked diffuse erythema and swelling, or four digit joints affected; and 5, severe diffuse erythema and severe swelling of the entire paw, unable to flex digits, or ankylosis, if present. Under this protocol, the mean score of all four paw scores, rather than a sum, was used for the scoring index.

Induction and assessment of ITP

Platelet counts were measured by serial tail-vein nicking on days 1 through 5. To measure platelet counts, 10 μl of blood was diluted in 15 μl citrate buffer (0.105 M,///3.2%). The samples were analyzed for absolute platelet count on a Drew Scientific HemaVet 950FS hemocytometer that was standardized daily. Control-group mice received no pretreatment and no platelet depletion. ITP mice received only platelet depletion with MWReg30 α-platelet antibody (BD Biosciences, San Jose, CA, USA). Mice in the experimental groups received pretreatment with stradomers, murine IgG2a Fc, or IVIG on day 1, administered 16 to 24 hours before platelet depletion with MWReg30 each evening on days 2 through 4 (2 μg diluted in 200 μl PBS, given intraperitoneally). Mice pretreated with murine IgG2a Fc or stradomers received only a single dose (400 μg intravenous) on day 1. Mice pretreated with IVIG received daily doses (2 g/kg by intraperitoneal injection every morning on days 1 through 5).

In some experiments, administration of 2A-2H_C on day 1 caused an initial decrease in platelet count before administration of MWReg30. Therefore, percentage platelet decrease is shown relative to the baseline measurement on day 2 by using the formula: % = measured platelet count/baseline platelet count on day 2. Experiments were done in replicate, as indicated, and all mice were included in pooled statistical analysis except for the following: for experiments testing the efficacy of the homodimeric versus multimeric fractions of 2A-2H_C in preventing ITP, three replicate experiments were done with six mice per group. One mouse in the ITP group was excluded because of spurious platelet values. Two of 18 mice in the 2A-2H_C group were excluded from day 5 analysis because of equipment failure that prevented measurement of platelet count.

GVHD model

The GVHD model was previously described [13]. On day 0, BDF1 mice were given 4 × 10⁷ C57BL/6 splenocytes and lymphocytes by tail-vein injection. Mice received no treatment, IgG2a Fc, or 2A-2H_C (both given at 400 μg IV) on day 0 and day 4. In the preventive study, 2A-2H_C was administered before adoptive transfer of B6 splenocytes and lymphocytes. On day 9, donor anti-host CTL activity was tested by ⁵¹Cr-release assay against H2d-expressing P815 cells (positive target) and EL4s that do not express H2d (negative control).

Statistics

Statisticians had access to all primary data. For the platelet data, Wilcoxon rank-sum tests were used to compare percentages between groups. For the Arthritis Index data, linear mixed-effects models with either one or two slopes were used to compare the Arthritis Index over time between groups. For the platelet and the RBC data, percentages of baseline values were calculated for each post-baseline time point and presented in the plots as means ± SEM. Wilcoxon rank-sum tests were used to compare percentages between groups at some later time points at which we expected differences to have developed between groups (days 4 and 5 for platelet data, and 60 minutes, 90 minutes, and 165 minutes after injection for RBC data). Unless otherwise indicated, all comparisons reported were made against the diseased with no-treatment group (ITP, CIA, or injected oRBCs with no pretreatment). Bonferroni's correction was used for significance where multiple comparisons were made, with the intended number of comparisons reported.

Stradomers exist as highly ordered multimers that can be fractionated by size

To evaluate the potential of the IgG2 hinge as an MD, we created cDNA constructs in which the hinge region of human IgG2 was linked to either the carboxy (2A-2H_C) or amino (2A-2H_N) termini of murine IgG2a Fc. In parallel, we developed compounds with the previously described ILZ at the carboxy (2A-ILZ_C) or amino termini (2A-ILZ_N) (Figure 1A) [15]. None of these fusion proteins contains an Fab region. We named the resultant proteins "stradomers" to distinguish them from naturally occurring homodimeric and multimeric immunoglobulins.

On purification, analysis of these compounds with SDS-PAGE revealed that both of the stradomers containing the isolated IgG2 hinge region exist as homodimers (50 kDa) and as highly ordered multimers, which occur at a far higher percentage than found with recombinant IgG2a Fc (Figure 1B). The fact that these multimers are present under denaturing, but not reducing conditions, suggests that they are covalently linked. Although both ILZ compounds were also composed of homodimers and multimers, the presence of higher-order multimers was less than that observed for the 2H stradomers. This lack of higher-order multimers was most pronounced for the 2A-ILZ_C protein. As expected, recombinant IgG2a Fc was present at the appropriate size for a homodimer (45 kDa). In addition, a less-intense band, which likely represents a multimer, was present at approximately 90 kDa (Figure 1B).

As a first step toward characterizing the function of these stradomers, we isolated the homodimeric and multimeric forms of one representative stradomer, 2A-2H_C. SDS-PAGE analysis of the resultant fractions revealed that the higher-order multimers could be separated from the homodimeric fractions with a relatively high degree of purity (Figure 1C, D). These data demonstrate that incorporation of a novel human IgG2H MD onto either the amino or carboxy terminus of mouse IgG2a Fc results in a fusion protein, which forms highly ordered Fc multimers that can be fractionated according to Fc valency.

Stradomer binding to low-/intermediate-affinity canonic FcγRs and SIGN-R1 directly correlates with the degree of multimerization

Low-/intermediate-affinity FcγRs, including FcγRIIb, FcγRIII, and FcγRIV, are recognized to form more-stable associations with Fc-bearing immune complexes than with nonmultimerized Fc of the appropriate isotype [16]. Furthermore, the mechanisms by which both IVIG and Fc-bearing immune complexes mediate tolerance are reported to depend on engagement of selected low/intermediate receptors [2]. Based on these data, we sought to determine whether these fully recombinant stradomers could effectively engage the canonic FcγRs and to correlate the stability of these interactions with Fc valency.

Evaluation of stradomer interactions with murine FcγRs by surface plasmon resonance (SPR) revealed that, in comparison to the transient associations observed with recombinant IgG2a Fc, the 2A-2H_C, 2A-2H_N, and 2A-ILZ_N proteins formed stable interactions with FcγRI, FcγRIIb, FcγRIII, and FcγRIV. The interactions of 2A-ILZ_C with these receptors were similar to those of IgG2a Fc, consistent with the lack of highly ordered multimer formation. Additionally, based on reports that suggest that the C-type lectin, SIGN-R1, regulates the antiinflammatory effects of IVIG through specific interactions with α-2,6-sialylated Fcs [17], we also evaluated the ability of the stradomers to interact with SIGN-R1. The ability of unfractioned stradomers to bind SIGN-R1 was superior to that of IgG2a Fc but more limited in comparison with the canonic receptors, and the stability of these associations was generally less than that observed for interactions with the canonic FcγRs (Figure 2A, Table 1).

To place our findings in context with previous reports demonstrating that the valency of Fc-containing compounds affects interactions with their cognate receptors [16], we evaluated binding of the individual fractions of 2A-2H_C with FcγRIIb, FcγRIII, and SIGN-R1 by SPR (Figure 2B), and with FcγRIIb and FcγRIII by Biolayer Interferometry (see Additional File 1, Figure S1). Consistent with the findings of others, higher-order multimers of 2A-2H_C formed more-stable interactions with both FcγRIIb and FcγRIII than did the smaller isolates. Unexpectedly, the higher-order isolates also bound with greater avidity to SIGN-R1, suggesting that this may be a low- to moderate-affinity receptor. Taken in concert, these data demonstrate that fully recombinant stradomers can recapitulate the binding avidity of Fc-bearing immune complexes.

Stradomers can effectively treat collagen-induced arthritis

To begin to understand the function of the stradomers in vivo, we used a CIA model thought to recapitulate closely many of the essential features of rheumatoid arthritis. CIA studies were performed at two different contract research organizations, with selected experiments blinded to the test articles. Animals receiving the drug experienced only limited toxicities, the major manifestations of which included transient piloerection and a brief period of lethargy. In the CIA model, all stradomers showed pronounced therapeutic effects in treating CIA (P < 0.0001 for 2A-2H_C, 2A-2H_N, and 2A-ILZ_C; P = 0.0021 for 2A-ILZ_N) (Figure 3A). Interestingly, the 2A-ILZ_C fusion protein, which forms limited higher-order multimers in solution, was able to alleviate CIA significantly (P < 0.0001). Although IVIG was also therapeutic, the kinetics and potency of this effect were delayed and less robust than those observed with 2A-2H_C (P < 0.0001 for IVIG) (Figure 3B).

Based on the finding that 2A-ILZ_C, which had a low percentage of multimers, can alleviate CIA, we isolated the homodimeric and multimeric fractions of 2A-2H_C to determine the importance of Fc multimers in this model. Whereas the homodimeric fraction of 2A-2H_C demonstrated some therapeutic benefit in alleviating CIA, this treatment effect was less robust than that observed with the multimeric fraction (P < 0.0001 for homodimeric and multimeric fractions of 2A-2H_C) (Figure 3C). These findings are consistent with the idea that Fc homodimers may have some therapeutic benefit in selected models based on their ability to block transiently FcγRs and to decrease the half-life of pathogenic autoantibodies by competing for binding with the neonatal receptor (FcRn) [18]. It is also the case that murine IgG2a Fc, unlike human IgG1 Fc, autodimerizes (Figure 1B), which may contribute to the activity of IgG2a Fc "homodimers" in murine models.

Finally, to understand whether our findings were dose dependent, we had a third-party CRO blinded to dose and control evaluate doses of 10, 20, and 40 mg/kg in the CIA model. Our data indicate that a dose-dependent difference exists in potency (Figure 3D). These findings suggest that although a dose-response exists to the stradomers in CIA, a therapeutic threshold may exist.

Stradomers can effectively prevent idiopathic thrombocytopenic purpura

To extend our findings, we evaluated the therapeutic utility of stradomers in murine models of ITP and graft-versus-host disease (GVHD). The ITP model differs from CIA in that anti-platelet antibody is directly administered to the host, and thus is not dependent on secondary antibody development as a result of antigen priming/boosting. In the ITP model, a single intravenous dose of 400 μg of 2A-2H_C protected mice against platelet loss (P < 0.0001, days 4, 5), whereas recombinant murine IgG2a Fc was less effective (P = 0.0145; P = 0.178 days 4, 5, neither significant with the Bonferroni correction) (Figure 4A). Fractionation of 2A-2H_C revealed that the observed protective effects were primarily attributable to the multimeric (P < 0.0001, days 4, 5) rather than the homodimeric fraction (P = 0.0093; P = 0.140 days 4, 5, neither significant with the Bonferroni correction) (Figure 4B). Consistent with previous mouse ITP prevention studies, daily high-dose administration of IVIG provided nearly complete protection against platelet loss. Functionally, although 2A-ILZ_N conferred significant protection from MWReg30-induced platelet depletion (P = 0.0056; P = 0.0001; days 4, 5), unlike that in the CIA model, the 2A-ILZ_C protein at the same dose was not effective (P = 0.80; P = 0.76 days 4, 5) (Figure 4C).

In some, but not all, individual experiments, we observed a significant platelet decrease 16 to 24 hours after infusion of 2A-2H_C, before the administration of MWReg30 (see Additional File 2, Figure S2A). Detailed studies in TLR-4^-/- mice, which lack the endotoxin receptor [11], revealed that this effect is not due to contaminating endotoxin. However, flow cytometry-based platelet-binding studies suggest that these observations may be secondary to the ability of 2A-2H_C to bind to platelets, resulting in their transient sequestration (see Additional File 2, Figure S2B). Taken in concert, our findings confirm that the protection conferred by recombinant IgG2a Fc fusion compounds against the development of ITP is not restricted to an individual MD or MD location, and suggest that protection may be related to the percentage and/or size of multimers in the preparation, as is the case with IVIG aggregates [4, 16].

Importantly, unlike ITP and CIA, in which antibody responses are recognized to play an important role in disease pathogenesis, 2A-2H_C was not effective in either the prevention or the treatment of GVHD (see Additional File 3, Figure S3). These differences in therapeutic utility between model systems might be explained by the fact that this GVHD model is primarily T-cell mediated, a fact that may be exploited in future studies as a means to determine the mechanisms by which stradomers mediate their biologic effects [13].

2A2H_C inhibits depletion of circulating opsonized RBCs

Our data indicate that stradomers alleviate CIA and prevent ITP, suggesting that they may represent an advance toward a recombinant analogue of IVIG. The primary hurdle to creating a fully recombinant IVIG is the absence of a clear mechanism of action. Specifically, IVIG is postulated to mediate its antiinflammatory effects through multiple mechanisms, including, but not limited to, (a) saturating the neonatal receptor, FcRn, so that pathogenic autoantibodies are more rapidly cleared from the circulation; (b) binding to SIGN-R1/DC-SIGN and/or the activating FcγIIIa receptor, with subsequent upregulation of the FcγRIIb inhibitory receptor; and (c) competitive inhibition of Fcγ receptor signaling [19–21]. Further complicating matters are the facts that these mechanisms are not mutually exclusive and that the therapeutic effects of IVIG may prove to be disease specific.

It has been postulated that IVIG-mediated protection against ITP is partly due to its capacity to saturate the reticuloendothelial system (RES), preventing phagocytosis of autoantibody opsonized platelets. To evaluate this possibility further, we tested the ability of 2A-2H_C to block the RES, by using an opsonized RBC (oRBC) depletion model [22]. Our findings reveal that, like IVIG, 2A-2Hc inhibits the removal of oRBCs from the circulation, indicating that 2A-2H_C likely interferes with the RES function (Figure 5A). Furthermore, because these RBCs were opsonized ex vivo, our data suggest that the function of 2A-2H_C in ITP cannot solely be attributed to blockade of the FcRn, which has the potential to decrease the half-life of the MWReg30 anti-platelet antibody used in our ITP model [23].

2A-2H_C protection requires FcγRIIb

Next, we sought to define the role of the inhibitory FcγRIIb receptor in 2A-2H_C-mediated protection against ITP. The decision to interrogate FcγRIIb was based on previous studies that have clearly documented that this receptor is necessary for the protective effects of IVIG in ITP models [24, 25]. By using the same strain of FcγRIIb^-/- mice studied by Siragam and colleagues [25], we observed a baseline increase in the hematocrit and a decrease in the platelet counts of FcγRIIb^-/- mice, which limits the dynamic range of the ITP assay (see Additional File 4, Figure S4). Consistent with the findings of other investigators, our data convincingly demonstrate that FcγRIIb is required for the protective effects of IVIG in a model of ITP (Figure 5B) [24, 25]. Our data also suggest that 2A-2H_C- and 2A-ILZ_N-mediated protection against ITP is likely dependent on FcγRIIb binding, as neither agent conferred statistically significant protection against platelet loss in FcγRIIb^-/- mice (Figure 5B). Importantly, the reduced range of this assay introduces a potential source of bias, which mandates caution when interpreting these data.