Anti-CD20-interferon-β fusion protein therapy of murine B cell lymphomas

Abstract

Type I interferons (IFNα/β) are cytokines with a broad spectrum of anti-tumor activities including anti-proliferative, pro-apoptotic, and immunostimulatory effects, and are potentially useful in the treatment of B cell malignancies and other cancers. To improve anti-tumor potency and diminish the systemic side effects of IFN, we recently developed anti-CD20-IFNα fusion proteins with in vitro and in vivo efficacy against both mouse and human lymphomas expressing CD20. Since IFNβ binds more tightly to the IFNα/β receptor (IFNAR) and has more potent anti-tumor activities, we have now constructed an anti-CD20 fusion protein with murine IFNβ (mIFNβ). Anti-CD20-mIFNβ was more potent than recombinant mIFNβ and anti-CD20-mIFNα in inhibiting the proliferation of a mouse B cell lymphoma expressing human CD20 (38C13-huCD20). Growth inhibition was accompanied by caspase-independent apoptosis and DNA fragmentation. The efficacy of anti-CD20-mIFNβ required the physical linkage of mIFNβ to anti-CD20 antibody (Ab). Importantly, anti-CD20-mIFNβ was active against tumor cells expressing low levels of IFNAR (38C13-huCD20 IFNAR^lo). In vivo, established 38C13-huCD20 tumors were largely insensitive to rituximab or a non-targeted mIFNβ fusion protein, yet treatment with anti-CD20-mIFNβ eradicated 83% of tumors. Anti-CD20-mIFNβ was also more potent in vivo against 38C13-huCD20 than anti-CD20-mIFNα, curing 75% versus 25% of tumors (p = 0.001). Importantly, while anti-CD20-mIFNα could not eradicate 38C13-huCD20 IFNAR^lo tumors, anti-CD20-mIFNβ treatment prolonged survival (p = 0.0003), and some animals remained tumor-free. Thus, Ab fusion proteins targeting mIFNβ to tumors show promise as therapeutic agents, especially for use against tumors resistant to the effects of mIFNα.

INTRODUCTION

Although first described for their ability to inhibit viral replication, type I IFNs have multiple actions exhibiting both anti-proliferative effects against and inducing apoptosis in tumor cells (1, 2). They also inhibit angiogenesis (3, 4) and exert multiple effects on the immune response (5, 6). In both humans and mice there is only one form of IFNβ, but multiple species of IFNα, with mice having 14 functional and 3 pseudo IFNα genes (7, 8).

Type I IFNs elicit their diverse responses by binding a shared cell surface receptor consisting of the two transmembrane proteins IFNα/β receptor (IFNAR) 1 and 2. Although binding the same receptor, IFNα and IFNβ differ significantly in their activity and affinity for these membrane proteins and in the downstream consequences of signaling. Data indicate that IFN first binds to IFNAR2 and subsequently recruits IFNAR1 in a transient fashion (9). Both receptors contribute in an additive manner to the total binding affinity of IFN and it is the stability of this ternary complex that determines both the anti-proliferative and anti-viral activity of the IFN (10). While the anti-proliferative potency was found to correlate with the binding affinity throughout the entire range of affinities examined, the anti-viral potency reached a maximum at binding affinities equivalent to that of wild-type human IFNα2 (10).

For both humans and mice the isoforms of IFNα differ in their relative tumor anti-proliferative and anti-viral activity. In man the 12 different subtypes of IFNα show a wide range of anti-proliferative activity with IFNα14 being the strongest. IFNα14 is nearly 10 times as potent against OVCAR cells as IFNα2, the IFNα subtype most commonly used in the clinic (11). In this assay human IFNβ is approximately 15 times more effective than IFNα2 (8). In the mouse, mIFNα11 and mIFNα12 have the most potent anti-proliferative activity and resemble murine IFNβ (mIFNβ) in their anti-proliferative activity (8). However, we used mIFNα1 in our initial fusion protein because the vast majority of the murine IFNαs resemble mIFNα1 in their anti-proliferative activity. mIFNβ is approximately 100-fold more effective than mIFNα1 in inhibiting proliferation and approximately 10-fold more effective in anti-viral activity (8).

IFNs were the first recombinant proteins to be used in the therapy of cancer (2, 12). Although IFN is effective in the treatment of a variety of malignancies, prolonged treatment with IFN can be problematic because of the associated side effects. The most common adverse events associated with IFNα therapy are flu-like symptoms, fatigue, anorexia, neutropenia, neuropsychiatric symptoms, and injection reactions; these side-effects may become dose-limiting (13). Frequent administration is also required because of the short in vivo half-life of IFN with IFNα2 having a half-life of only about 1 hour (14). PEGylation, in which the protein is covalently linked to linear or branched polyethylene glycols, can prolong the half-life and improve the efficacy of IFNα (4, 15) but the activity of the PEGylated IFN is often compromised (16). Both direct injection into the tumor and gene therapy have been used to deliver high levels of IFN to tumors. Intra-lesional injection of low-dose IFNα was effective for the treatment of patients with primary cutaneous marginal zone B cell lymphomas (17). In mouse models, IFNβ gene therapy caused the dramatic regression of human tumors apparently as a result of the direct anti-proliferative or cytotoxic activity of IFNβ (18). Interestingly, only a fraction of the cells had to be transduced to achieve tumor regression. These results have suggested that targeting type I IFN to tumor cells in vivo might substantially enhance its efficacy and tolerability.

CD20 is a nonglycosylated integral 33–37 kDa transmembrane phosphoprotein expressed on more than 95% of normal and neoplastic B cells. The elevated levels of the CD20 protein in B cell malignancies, its extracellular accessibility and the fact that it is not internalized, downregulated or shed, make it an excellent target for treating B cell cancers. Multiple mechanisms of action have been proposed for the efficacy of anti-CD20 monoclonal Abs (mAbs) in treating lymphoma including the induction of apoptosis, antibody-dependent cell-mediated cytotoxicity (ADCC), phagocytosis, complement-dependent cytotoxicity (CDC), cross-priming of CD8⁺ T cells (19) and down-regulation of Bcl-X_L with a concomitant increase in chemosensitization (20).

In initial studies we investigated the ability of anti-CD20-IFNα fusion proteins to treat B cell lymphomas in mice (21). Although the mIFNα1 and human IFNα2 fusion proteins had reduced IFNα bioactivity in vitro compared with native IFNα, CD20 targeting resulted in efficient anti-proliferative and pro-apoptotic effects against an aggressive rituximab-insensitive human CD20⁺ mouse lymphoma (38C13-huCD20) and a human B cell lymphoma (Daudi). Optimal tumor eradication required CD20 targeting. Importantly, potent anti-tumor efficacy was observed in the absence of toxicity. Gene knockdown studies demonstrated that tumor eradication required expression of IFNAR on the tumor cell surface and that anti-CD20 mAb fused with mIFNα had reduced efficacy both in vitro and in vivo against tumors with decreased expression of IFNAR.

Given the enhanced anti-proliferative activity of mIFNβ compared to mIFNα1, we have now extended these studies by evaluating the efficacy of anti-CD20 fused to mIFNβ (anti-CD20-mIFNβ) against mouse tumors expressing human CD20. Indeed we show that anti-CD20-mIFNβ has more potent anti-tumor activity than anti-CD20-mIFNα. Importantly, anti-CD20-mIFNβ is effective both in vitro and in vivo against a tumor with decreased expression of IFNAR, a tumor that was resistant to the effects of anti-CD20-mIFNα. Thus, fusion proteins targeting IFNβ to tumors show great promise as therapeutic agents, especially for use against tumors resistant to the effects of IFNα.

MATERIALS AND METHODS

Cell lines

38C13-huCD20 mouse B cell lymphoma cells, which express human CD20, were previously described (22). Both 38C13 (23) and 38C13-huCD20were cultured in IMDM (Invitrogen, Carlsbad, CA) supplemented with 5% calf serum (CS) (Atlanta Biologics, Lawrenceville, GA). The cell line 38C13-huCD20 IFNAR^lo was produced by transducing 38C13-huCD20 cells with a lentiviral vector encoding an shRNA targeting the IFNAR1 subunit of IFNAR with the sense sequence 5′ – GCGTCTACATTATAGATGACAA – 3′ as previously described (21). 38C13-huCD20 IFNAR^lo and Chinese Hamster Ovary (CHO) cells were cultured in IMDM supplemented with 5% CS.

Construction of expression vectors

The heavy (V_H) and light (V_L) chain variable regions of the anti-CD20 mAb 2B8(24), the murine parent of rituximab, were cloned into expression vectors for kappa light chain and human gamma heavy chain with the SGGGGS peptide linker at the end of C_H3, as previously described (21). mIFNβ was cloned using the primers 5′ – GGGATCCATCAACTATAAGCAGCTCCAGCTC –3′ and 5′ – GGTCTAGAATCAGTTTTGGAAGTTTCTGGTAAGTC – 3′ with cDNA obtained from 6.5.2, a mutant of the mouse myeloma cell line J558 (25). After sequence verification, the inserts were cloned as a BamH I/Xba I fragment into pcDNA3.1 in which the neo resistance marker had been changed to his (unpublished). The anti-CD20 V_H and IgG constant regions were then inserted as EcoR V/BamH I fragments. To produce a non-targeted fusion protein, the V_H of anti-CD20 was replaced with the V_H specific for the hapten dansyl (DNS) (26). Vectors were available for the production of the anti-DNS chimeric kappa light chain.

Protein production and purification

CHO cells were transfected with the heavy and light chain vectors using Lipofectamine Plus^TM according to the manufacture’s protocol (Invitrogen), and stable transfectants isolated. For protein production, cells were seeded into roller bottles, and once confluent, cells were expanded to 100–120 mL with IMDM + 1% Fetal Clone (Thermo Fisher, Waltham, MA). The supernatant was removed every 2–3 days and replaced with fresh medium. Cell-free culture supernatants were then passed through a protein A Sepharose 4B fast flow column (Sigma-Aldrich, St. Louis, MO) and the bound protein eluted with 0.1 M citric acid, pH 3.5. Eluted fractions were neutralized immediately with 2 M Tris-HCl pH 8.0. Fractions were run on SDS-PAGE gels and stained with Coomassie blue to verify protein purity and integrity. Concentrations of proteins were determined using the BCA assay (Pierce, Rockford, IL). Recombinant mIFNβ reference standard was obtained from the NIH (Bethesda, MD) or purchased from PBL InterferonSource (Piscataway, NJ). Rituximab was obtained from Genentech (South San Francisco, CA).

Determination of Binding to CD20

38C13-huCD20 cells were washed twice with PBS + 1% BSA + 0.1% sodium azide (FACS buffer), resuspended to 3x10⁶ cells/ml and 100 μl cells added to each tube. Rituximab and anti-CD20-mIFNβ fusion protein were serially diluted and added to the cells to achieve final concentrations of 0–106 nM. Cells were incubated on ice for 30 minutes, washed twice, then incubated with mouse anti-human kappa-PE (Invitrogen) for 30 minutes on ice. After washing twice, samples were run on a BD FACSCalibur flow cytometer using Cell Quest Pro software and data analyzed using FCS Express (De Novo Software, Los Angeles, CA).

Inhibition of proliferation assay

38C13, 38C13-huCD20, or 38C13-huCD20 IFNAR^lo cells (1x10⁴ per well) were seeded in 96-well plates and incubated with various treatments at 37°C for 48–72 h. Cell viability was quantified using MTS solution (Promega, Madison, WI) by measuring absorbance at 490 nm using a Synergy HT Multi-Detection Microplate Reader (BioTek Instruments, Inc., Winooski, VT). Data were analyzed by non-linear regression using Prism (GraphPad Software, Inc., La Jolla, CA) with the log [inhibitor] vs. the response with a variable slope with the IC₅₀ calculated. Data are expressed as the percentage of maximum metabolic activity of untreated cells.

Ki-67 assay

2x10⁵ 38C13-huCD20 cells were cultured with 20 pM of anti-CD20-mIFNβ for 0, 1, 6, 18, 24, 30, or 48 h at 37°C. At each time point cells were pelleted, washed twice with cold PBS, permeabilized with 1 ml cold methanol, washed twice with cold PBS + 3% BSA, and incubated for 1 h at 4°C with rabbit anti-Ki-67 (Abcam, Cambridge, MA) diluted 1:100 in PBS + 3% BSA. Cells were then washed twice, and incubated for 1 h at 4°C with 5 μl Fc Block (BD Pharmingen, San Jose, CA) and mouse anti-rabbit IgG-FITC (Sigma) diluted to 1:200 in PBS + 3% BSA. After incubating the cells with the secondary Ab, they were washed twice, resuspended in 2% paraformaldehyde (Electron Microscopy Sciences, Hatfield, PA), and stored at 4°C in the dark for later analysis by flow cytometry. Controls included unstained cells, cells incubated with Fc Block but stained only with mouse anti-rabbit IgG-FITC, or cells incubated with Fc Block plus rabbit anti-human IgE (MP Biomedicals, Irvine, CA) and mouse anti-rabbit IgG-FITC. Samples were analyzed by flow cytometry and data analyzed using FlowJo software (Tree Star, Inc., Ashland, OR).

Cell cycle and DNA fragmentation analysis

38C13-huCD20 cells (2x10⁵) were incubated with the indicated reagents for various times, harvested by centrifugation, supernatant removed by aspiration, and cells resuspended in 1 ml of hypotonic DNA staining buffer (1 mg/ml sodium citrate, 100 μg/ml PI, 20 μg/ml RNase A, and 0.3% Triton X-100 in dH₂0). Samples were kept on ice protected from light for 30 minutes or for a maximum of 1 h and analyzed by flow cytometry. Cell cycle analysis was performed using FlowJo with the Watson Model. For DNA fragmentation time course studies, 38C13-huCD20 cells were incubated with 100 pM of anti-CD20-mIFNβ for 1, 6, 24, or 48 h, washed and treated with 1X trypsin/EDTA (Invitrogen) for 10 minutes, washed once with IMDM + 5% CS, returned to culture for the remainder of the 48 h culture period before harvest and DNA staining as above.

Apoptosis and metabolic activity assays

2x10⁵ 38C13-huCD20 or 38C13-huCD20 IFNAR^lo cells were incubated with various treatments at 37°C for 48 h. Cells were then stained with Annexin V-FITC and propidium iodide (PI) to distinguish populations of early apoptotic (Annexin V⁺/PI^-) and dead (Annexin V⁺/PI⁺) cells using Vybrant Apoptosis Kit #2 (Molecular Probes, Carlsbad, CA), and analyzed by flow cytometry. Metabolic activity was quantitated using the LIVE/DEAD Cell Vitality Assay Kit -C₁₂ resazurin/SYTOX Green assay kit (Molecular Probes) per manufacturer’s instructions, substituting Annexin V-FITC for SYTOX Green counterstaining where indicated.

Caspase activation assay

38C13-huCD20 cells were seeded in 96-well black walled clear bottom plates at a concentration of 1x10⁴ cells/well in IMDM growth medium. Fusion protein or rituximab was added to a final concentration of 100 pM and samples incubated at 37°C. As a positive control, cells were treated with 10 μM of staurosporine and negative controls included no cells and no treatment. All treatments were performed in triplicate. At each time point, one plate was removed and 100 μl of caspase substrate (Promega), prepared according to the manufacturer’s protocol, was added to wells. Luminescence was read 45 minutes later on a Synergy HT plate reader. Z-DEVD-aminoluciferin was used as substrate for caspase 3/7 activity. Caspase 2 activity was assayed using Z-VDVAD-aminoluciferin as the substrate in the presence of Ac-DEVD-CHO and MG-132 to select for caspase 2 activity. Caspase 8 was assayed using Z-LETD-aminoluciferin as the substrate in the presence of MG-132 to select for caspase 8 activity. Caspase 9 activity was assayed using Z-LEHD-aminoluciferin as substrate in the presence of MG-132 to select for caspase 9 activity. All substrates were purchased from Promega. The pan-caspase inhibitor Z-VAD(OMe)-FMK (BD Pharmingen) was added at 20 μM and apoptosis analyzed at 24 or 48 h. When apoptosis was analyzed at 48 h a second treatment of 20 μM Z-VAD(OMe)-FMK was added at 24 h.

Determination of fusion protein half-life

Four mice were injected i.v. with 200 μg of anti-CD20-mIFNβ. Two mice were bled (from retro-orbital sinus) at 5 minutes, 6 h, 48 h, and 101 h; the other two mice were bled at 30 minutes, 24 h, and 101 h. Blood was allowed to clot on ice and the serum removed following centrifugation. Anti-CD20-mIFNβ was detected by ELISA using plates coated with anti-mIFNβAb (clone 7F-D3; Abcam). The standard curve was established using serial 1:2 dilutions into 1% BSA of anti-CD20-mIFNβ starting at 0.4 μg/ml. Controls included serum only, 1% BSA only and wells not coated with Ab. Serial dilutions of previously diluted serum were performed for the experimental serum samples. Plates were incubated at 4°C overnight, washed with PBS + 0.2% Tween 20, and bound fusion protein detected with biotinylated anti-kappa Ab (Immuno Biological Laboratories, Minneapolis, MN) followed by streptavidin-AP (Zymed Laboratories, San Francisco, CA). ELISA plates were developed with 1 mg/ml p-nitrophenyl phosphate (Sigma-Aldrich) in 1 M diethanolamine-0.25 mM MgCl, pH 9.8 and the absorbance was measured at 410 nm. The concentration of fusion protein remaining in circulation was extrapolated from the standard curve. Data were analyzed using Excel (Microsoft Corporation, Redmond, WA) and the T_1/2 calculated as 0.693/λ.

Determination of antibody-dependent cellular cytotoxicity (ADCC)

ADCC assays were performed using the lactate dehydrogenase (LDH) release assay method following the manufacturer’s protocol (Roche Applied Science, Indianapolis, IN) as previously described (22). Mouse effector cells were collected fromC3H spleens by mechanical disruption. After lysis of red blood cells using ammonium chloride buffer (0.15 M NH₄Cl, 10 mM KHCO₃, 0.1 mM EDTA, pH 7.4), splenocytes were washed and cultured for 48 h in RPMI 1640 supplemented with 100 U/ml penicillin/streptomycin, 2 mM L-glutamine, and 50 μM β-mercaptoethanol (all from Invitrogen; RPMI complete growth medium) containing 10% FCS (Omega Scientific, Tarzana, CA) with 1000 U/ml human rIL-2 (Chiron, Emeryville, CA). Cells werethen collected, washed, and counted. 38C13-huCD20 target cells were harvested from culture, washed, counted, and incubated for 30 minutes at 4°C in the presence or absence of 53 nM rituximab or anti-CD20-mIFNβ in RPMI complete growth medium containing 5% FCS. After incubation, target cells were seeded in 96-well U-bottom plates, in triplicate, at 1x10⁵ cells/well with increasing numbersof effector cells for 5 h at 37°C. Maximum and spontaneous release was measured by adding 0.1% Triton X-100 or medium, respectively, to target cells alone. After incubation, supernatant was collected, and LDH release was measured on a Spectramax 384 Plus microplate reader (Molecular Devices, Sunnyvale, CA). Percentage cytotoxicity was calculatedas 100 x [(effector:target cell mix release – effector cell control –spontaneous release) / (maximum release- spontaneous release)].

Determination of complement-dependent cytotoxicity (CDC)

38C13-huCD20 cells were harvested from culture, counted and resuspended to 1.2x10⁶ cells/ml in RPMI complete growth medium containing 10% FCS. 250 μl of cells were then added to tubes and rituximab or anti-CD20-mIFNβ added to achieve the indicated final concentration ranging from 0–53 nM. Triplicate samples were incubated for 15 minutes on ice to allow Ab binding. After incubation, rabbit serum (Sigma-Aldrich) or heat-inactivated rabbit serum was added as a source of complement to a final concentration of 10%, and samples incubated at 37°C for 3 h. After incubation, cells were washed twice in PBS, stained with PI, and analyzed by flow cytometry to assess viability.

In vivo anti-tumor activity against mouse B cell lymphoma

Six- to eight-week-old female C3Hf/Sed/Kam mice were bred and housedat the UCLA Defined Pathogen Colony according to institutionalguidelines. 38C13-huCD20 or 38C13-huCD20 IFNAR^lo cells were thawed from common dedicated frozen stocks three days before tumor challenge and split the daybefore use. For tumor challenge, cells were washed twice in HBSS (Invitrogen) and diluted to the appropriate concentration in HBSS. Challenge inocula consisted of 5x10³ 38C13-huCD20 or 38C13-huCD20 IFNAR^lo cells injected subcutaneously (s.c.) above the base of the tail. Experimental groups of 8 mice (except where indicated) were treated with 100 μg of anti-CD20-mIFNα, anti-CD20-mIFNβ, anti-DNS-mIFNβ, or rituximab intravenously (i.v.) on days 5, 6, and 7 or on days 5, 6, 7 and 14 post tumor challenge. Mice were followed for survival and sacrificed when tumors reached 1.4 cm in diameter as per institutional guidelines. Bidirectional tumor growth measurements were obtained throughout the experiment. All studies were performed in compliance with the U.S. Department of Health and Human Services Guide for the Care and Use of Laboratory Animals and were approved by the UCLA Animal Research Committee.

RESULTS

Ab fusions with murine IFNβ are more effective than fusions with murine IFNα1 in inhibiting proliferation of lymphoma cells

Anti-CD20 and anti-DNS fusion proteins were produced with mIFNβ joined to the C-terminus of the C_H3 domain of human IgG1 by a SGGGGS linker (Fig. 1A). Use of IgG1 fusion proteins permits direct comparison to the IgG1 anti-human CD20 mAb rituximab. The anti-CD20-mIFNβ fusion protein bound avidly to cells expressing huCD20 (Fig. 1B), similar to rituximab. Slightly greater binding was observed with the fusion protein, possibly reflecting simultaneous binding of the fusion protein to both CD20 and IFNAR. To assess the ability of the mIFNβ moiety within the antibody-mIFNβ fusion proteins to inhibit cellular proliferation, 38C13 lymphoma cells were incubated with various concentrations of anti-CD20-mIFNβ, anti-DNS-mIFNβ, or recombinant mIFNβ for 48 h and the metabolic activity of the cells measured using the MTS assay (Fig. 1C). Under these conditions there is no targeting of the fusion proteins and the assay is measuring only the intrinsic anti-proliferative IFN activity of the fusion protein. The fusion proteins containing mIFNβ were very effective in inhibiting the proliferation of 38C13, with both fusion proteins exhibiting similar activity and being somewhat more effective on a molar basis than recombinant mIFNβ. This contrasts with our previous observations using fusions with mIFNα in which the non-targeted fusion proteins were less effective than recombinant mIFNα in inhibiting cellular proliferation (21).

Figure 1. Construction and characterization of Ab fusion proteins containing mouse IFNβ.

A. Diagram of the fusion proteins in which mIFNβ was joined to the C-terminus of human γ1 heavy chains specific for either CD20 or dansyl (DNS). The heavy chains were expressed with the appropriate light chain to generate the antigen-specific fusion proteins. B. The anti-CD20-mIFNβ fusion protein binds to CD20-expressing 38C13-huCD20 cells with affinity comparable to rituximab. Cells were incubated on ice with varying equimolar concentrations of rituximab or anti-CD20-mIFNβ, bound Ab detected with anti-human kappa-PE, and analyzed by flow cytometry. Data are expressed as mean fluorescence intensity (MFI) vs. protein concentration. C. The fusion proteins are more effective than mIFNβ in inhibiting the growth of 38C13 lymphoma cells. 38C13 cells were incubated with differing concentrations of mIFNβ, anti-CD20-mIFNβ, or anti-DNS-mIFNβ for 48 h. The metabolic activity of the remaining cells was then measured using the MTS assay. Data are expressed as the percent metabolic activity in the treated cells compared with untreated cells. Data shown are the mean of quadruplicate samples ± SD. The IFN concentration assumes 2 moles of IFN for each mole of fusion protein. D. IFNβ fusion proteins are more effective than IFNα fusion proteins in inhibiting the growth of 38C13-huCD20 (left panel) and 38C13-huCD20 IFNAR^lo (right panel) with CD20-targeting making them even more effective. 38C13-huCD20 and 38C13-huCD20 IFNAR^lo cells were incubated with differing concentrations of the indicated proteins for 72 h and cellular metabolic activity measured and displayed as described above.

To determine the contribution of antigen targeting to the ability of the antibody-mIFNβ fusion proteins to inhibit cellular proliferation and to compare their efficacy with mIFNα fusion proteins, 38C13-huCD20 lymphoma cells were incubated with various concentrations of anti-CD20-mIFNβ, anti-DNS-mIFNβ, recombinant mIFNβ, anti-CD20-mIFNα, or recombinant mIFNα for 72 h and the metabolic activity of the cells measured using the MTS assay (Fig. 1D). Several things are noteworthy. As has been previously observed (21) anti-CD20-mIFNα was very effective in inhibiting the proliferation of 38C13-huCD20, with anti-CD20-mIFNα (IC₅₀ = 10.3 pM) more effective than recombinant mIFNα (IC₅₀ = 750 pM). Although recombinant mIFNβ (IC₅₀ = 70.7 pM) is more potent than recombinant mIFNα, targeted anti-CD20-mIFNα is more potent than non-targeted recombinant mIFNβ. Interestingly, the non-targeted anti-DNS-mIFNβ (IC₅₀ = 12.7 pM) was approximately as active as the targeted anti-CD20-mIFNα and was more effective than recombinant mIFNβ. Targeting the mIFNβ fusion protein to CD20 expressed on the cell surface made it even more potent (IC₅₀ = 3.3 pM). The fact that fusion protein was effective in the pM range while rituximab has been reported to have an affinity of 8 nM (http://www.rxlist.com/rituxan-drug.htm) probably reflects the increased avidity of the fusion protein, which would bind to cells through both the two Ab binding sites and the two attached IFNβ molecules. Overall, anti-CD20-mIFNβ was approximately 3-times more potent in vitro than anti-CD20-mIFNα, a protein that has been demonstrated to have potent anti-tumor activity both in vitro and in vivo (21).

Cell killing by IFNα fusion proteins has been shown to require engaging the IFNAR and decreased expression of IFNAR1 has been shown to make 38C13-huCD20 cells relatively insensitive to treatment with anti-CD20-mIFNα (21). In contrast, cells with decreased expression of IFNAR1 (38C13-huCD20 IFNAR^lo) remain sensitive to inhibition by anti-CD20-mIFNβ (IC₅₀ = 10.7 pM) although they are not as sensitive as the parental cell line with normal IFNAR expression (Fig. 1D). Recombinant mIFNβ (IC₅₀ = 420 pM) is also effective against 38C13-huCD20 IFNAR^lo but targeted anti-CD20-mIFNβ is approximately 40-fold more effective.

Treatment of 38C13-huCD20 with anti-CD20-mIFNβ causes DNA fragmentation and alterations in cell cycle

To determine if inhibition of proliferation is associated with changes in the cell cycle, 38C13-huCD20 was treated with varying concentrations of anti-CD20-mIFNβ or anti-DNS-mIFNβ for 72 h and DNA content analyzed by flow cytometry following staining with PI. Treatment with either anti-CD20-mIFNβ or anti-DNS-mIFNβ results in dose-dependent loss of cell viability and alterations in the cell cycle with an accumulation of cells in G0/1 and a decrease in cells in S phase (Fig. 2A). However, anti-CD20-mIFNβ is far more potent than anti-DNS-mIFNβ, with treatment of only 20 pM anti-CD20-mIFNβ resulting in a near maximal affect. In contrast, anti-DNS-mIFNβ, even at concentrations of 2500 pM, never achieves this level of cell G0/1 accumulation and S phase reduction, clearly demonstrating the importance of targeting for fusion protein efficacy.

Figure 2. Treatment of 38C13-huCD20 with anti-CD20-mIFNβ or anti-DNS-mIFNβ causes DNA fragmentation and alterations in cell cycle.

A. 38C13-huCD20 cells treated with varying concentrations of anti-CD20-mIFNβ or anti-DNS-mIFNβ for 72 h were stained with PI and analyzed by flow cytometry to determine the proportion in G0/1, S, or G2 phases of the cell cycle. B. 38C13-huCD20 cells treated with 20 pM anti-CD20-mIFNβ for varying lengths of time were stained with rabbit anti-Ki-67 followed by mouse anti-rabbit IgG-FITC and analyzed by flow cytometry (solid line). Controls include unstained cells (faint dotted line), untreated cells stained only with mouse anti-rabbit IgG-FITC (shaded peak), and untreated cells stained with rabbit anti-Ki-67 and mouse anti-rabbit IgG-FITC (dashed line).

The Ki-67 protein is present during all active phases of the cell cycle, but is absent from cells in G0 (27). To determine if cells treated with anti-CD20-mIFNβ have exited the cell cycle and ceased dividing, 38C13-huCD20 cells were treated with 20 pM of anti-CD20-mIFNβ for varying lengths of time and their expression of Ki-67 determined by flow cytometry (Fig. 2B). Following 6 h of treatment there was no change in Ki-67 expression (data not shown). However, by 24 h decreased Ki-67 expression is clearly evident, and by 48 h virtually all cells had decreased Ki-67 expression, with the majority of the cells expressing none. Thus, treatment with anti-CD20-mIFNβ causes tumor cells to exit the cell cycle thereby contributing to inhibition of proliferation and the observed decrease in metabolic activity.

Treatment with 20 pM of anti-CD20-mIFNβ causes maximal apoptosis and loss of metabolic activity

To determine if the observed inhibition of proliferation following treatment with anti-CD20-mIFNβ is also associated with apoptosis and inhibition of metabolic activity, 38C13-huCD20 cells were treated with 20 pM of anti-CD20-mIFNβ for 48 h and examined by flow cytometry (Fig. 3). Following staining with Annexin V-FITC and PI, over 50% were found to have undergone apoptosis as defined by being Annexin V-positive (Fig. 3A). Further increasing the treatment dose to 2500 pM did not increase the percentage of Annexin V-positive cells. However, decreasing the treatment dose below 20 pM did result in a decrease in the percentage of Annexin V-positive cells (data not shown). Non-targeted anti-DNS-mIFNβ was much less effective in causing apoptosis, with a concentration of 2500 pM of anti-DNS-mIFNβ failing to achieve the same degree of apoptosis as 20 pM of anti-CD20-mIFNβ with only 41.9% of the cells Annexin V-positive.

Figure 3. Anti-CD20-mIFNβ is effective in causing apoptosis and alterations in metabolic activity in 38C13-huCD20 cells.

A. 38C13-huCD20 cells were incubated with 20 or 2500 pM anti-CD20-mIFNβ or anti-DNS-mIFNβ or left untreated for 48 h and then examined by flow cytometry following staining with Annexin V-FITC and PI. B. Cells treated as indicated for 48 h were analyzed by flow cytometry for membrane alterations and metabolic activity following staining with Annexin V-FITC and resazurin. The percentage of cells in each quadrant is indicated.

To examine the relationship between alterations in membrane structure and metabolic activity, cells were treated with anti-CD20-mIFNβ for 48 h and examined following staining with Annexin V-FITC and resazurin. Annexin V staining of exposed phosphatidylserine indicates change in membrane structure and C₁₂-resazurin is reduced to the red-fluorescent C₁₂-resorufin by metabolically active cells. Staining with Annexin V correlated with loss of metabolic activity as Annexin V-positive cells failed to reduce C₁₂-resazurin (Fig. 3B). However, a population of metabolically active cells remained even when the treatment dose was increased to 2500 pM of anti-CD20-mIFNβ. Similar results were observed when cells were stained with SYTOX green, a nucleic acid stain that crosses compromised plasma membranes and C₁₂-resazurin, with cells with compromised membranes being metabolically inactive (data not shown). Thus, anti-CD20-mIFNβ treatment causes alterations in the plasma membrane as evidenced by both Annexin V and SYTOX green staining and these alterations correlated with a reduction in metabolic activity, and maximum effects seen following treatment with only 20 pM of anti-CD20-mIFNβ. Similar results were obtained after treatment with anti-DNS-mIFNβ except that much larger doses were required to achieve the maximum effect (data not shown).

Covalent linkage of anti-CD20 mAb with mIFNβ is required for DNA fragmentation, alterations in cell cycle and apoptosis

To determine if mIFNβ must be physically linked to anti-CD20 mAb for efficacy, 38C13-huCD20 cells were incubated for 48 h with 20 pM of rituximab, 40 pM mIFNβ, 20 pM anti-DNS-mIFNβ, 20 pM anti-CD20-mIFNβ, or 20 pM rituximab + 40 pM mIFNβ and then DNA content analyzed by flow cytometry (Fig. 4A). Alterations in cell cycle distribution were observed only following treatment with 20 pM anti-CD20-mIFNβ and not following treatment with the same of amount of mIFNβ or mIFNβ + rituximab (Fig. 4B). Anti-DNS-mIFNβ also was ineffective at this dose, again demonstrating the importance of targeting mIFNβ to the CD20 antigen expressed on the cell surface for anti-tumor efficacy.

Figure 4. Physical linkage of mIFNβ with anti-CD20 is required for effective DNA fragmentation, alterations in cell cycle and apoptosis.

A. 38C13-huCD20 cells were incubated with medium alone, 20 pM rituximab, 20 pM anti-DNS-mIFNβ, 20 pM anti-CD20-mIFNβ, 40 pM mIFNβ, or 40 pM mIFNβ + 20 pM rituximab for 48 h, stained with PI and analyzed by flow cytometry. B. Cell cycle distribution of treated cells from A. C. 38C13-huCD20 cells were incubated as above for 48 h, stained with Annexin V-FITC and PI and analyzed by flow cytometry. The percentage of cells in each quadrant is indicated.

To determine if mIFNβ fused to anti-CD20 mAb was more effective in inducing apoptosis than mIFNβ present at the same time as anti-CD20 mAb but not physically linked to it, 38C13-huCD20 cells were incubated with 20 pM of rituximab, 40 pM mIFNβ, 20 pM anti-CD20-mIFNβ, 20 pM anti-DNS-mIFNβ, or 20 pM rituximab + 40 pM mIFNβ for 48 h, and apoptosis measured by flow cytometry following staining with Annexin V-FITC and PI (Fig. 4C). Effective apoptosis of 38C13-huCD20 cells was only seen following treatment with anti-CD20-mIFNβ. Approximately 34% of the cells were Annexin V⁺ following treatment with anti-CD20-mIFNβ while only 10–13 % were Annexin V⁺ following all of the other treatments. Thus anti-CD20-mIFNβ was effective in inducing apoptosis in 38C13-huCD20 with physical linkage of mIFNβ to anti-CD20 mAb necessary for efficacy. Untargeted anti-DNS-mIFNβ was not effective in causing apoptosis, demonstrating that it is cell surface targeting, not just linkage of mIFNβ in the fusion protein which results in its efficacy.

Time course of commitment to undergo DNA fragmentation

An important question is how long must the fusion protein be bound to IFNAR for the anti-tumor effects to be irreversible. When 100 pM of anti-CD20-mIFNβ was added to cells, excess non-bound fusion protein washed away, and cells examined for DNA content 48 h later, extensive DNA fragmentation was seen (data not shown). However, when bound protein was removed by treating the cells with trypsin, DNA fragmentation was observed if the fusion protein had been present for 24 h but not if it had been present for only 6 h (Fig. 5A). Having the fusion protein present for 48 h did not result in a further increase in the extent of DNA fragmentation observed. Therefore, sometime between 6 and 24 h the cells become committed to undergo DNA fragmentation and cell death.

Figure 5. Time course of commitment to DNA fragmentation and caspase activation following treatment with anti-CD20-mIFNβ.

A. 38C13-huCD20 cells were treated with 100 pM of anti-CD20-mIFNβ for the indicated time and then bound fusion protein removed by trypsin treatment. DNA content was analyzed by flow cytometry at 48 h following cell permeabilization and staining with PI. Untreated cells are indicated by the filled peak and treated cells are indicated by the solid line. B. Caspase activity was analyzed at the indicated times following treatment with 100 pM of anti-CD20-mIFNβ, anti-DNS-mIFNβ, or rituximab. Treatment with 10 μM of staurosporine served as a positive control. Data are shown as the mean + SD of the relative caspase activity (RLU) of triplicate samples.

Caspase activation and the anti-tumor activity of anti-CD20-mIFNβ

To determine if caspase activation was associated with the observed DNA fragmentation, 38C13-huCD20 cells were incubated with 100 pM of anti-CD20-mIFNβ, 100 pM anti-DNS-mIFNβ, or 100 pM rituximab for varying lengths of times and caspase activation measured. Treatment with 10 μM of staurosporine served as a positive control. Little activation of caspase 2, 8, or 9 was observed following treatment with anti-CD20-mIFNβ (data not shown). Modest activation of the executioner caspases 3 and 7 was seen after 24 h of treatment but not after 6 h, consistent with the time course for the observed DNA fragmentation (Fig. 5B).

Caspase activation does not play a significant role in anti-CD20-mIFNβ-induced DNA fragmentation or apoptosis

Although only modest caspase activation was observed, it was important to determine if this played a significant role in the activity of anti-CD20-mIFNβ. However, inhibition of caspase activity using 50 μM of the pan-caspase inhibitor Z-VAD(OMe)-FMK had little to no effect on the amount of DNA fragmention and cell cycle alterations observed following treatment with 20 pM of anti-CD20-mIFNβ for 24 (data not shown) or 48 h (Fig. 6A,B). Similar results were obtained when the efficacy of Z-VAD(OMe)-FMK on anti-CD20-mIFNβ induced apoptosis was assessed by Annexin V-FITC and PI staining (Fig. 6C). Complete inhibition of caspase activity by Z-VAD(OMe)-FMK on staurosporine- and fusion protein-treated cells was confirmed using the luminescence caspase activity assays (data not shown). Thus, it appears that most of the cell cycle alterations and apoptosis observed following treatment with anti-CD20-mIFNβ took place through caspase-independent pathways.

Figure 6. Caspase inhibition does not influence anti-CD20-mIFNβ induced changes in cell cycle or apoptosis.

A. 38C13-huCD20 cells were treated with medium alone, 20 pM of anti-CD20-mIFNβ, 50 μM of the pan-caspase inhibitor Z-VAD(OMe)-FMK, or 50 μM of Z-VAD(OMe)-FMK + 20 pM of anti-CD20-mIFNβ for 48 h, permeabilized, stained with PI and analyzed by flow cytometry. B. Cell cycle distribution of treated cells from A. C. 38C13-huCD20 cells were treated with medium alone, 20 μM of Z-VAD(OMe)-FMK, 20 pM anti-CD20-mIFNβ, or 20 μM of Z-VAD(OMe)-FMK + 20 pM anti-CD20-mIFNβ for 48 h and analyzed by flow cytometry following staining with Annexin V-FITC and PI. The percentage of cells in each quadrant is indicated.

Effector functions of anti-CD20-mIFNβ

In vivo efficacy of fusion protein therapy might reflect many different mechanisms including the immunologic effector functions of the Ab within the fusion protein. Therefore, the ability of the fusion protein to carry out ADCC and CDC was determined. Against 38C13-huCD20 cells, anti-CD20-mIFNβ was even more effective than rituximab in carrying out ADCC (Fig. 7A) (p = 0.01 at all concentrations). This improved efficacy may reflect the fact that anti-CD20-mIFNβ appears to bind more strongly to 38C13-huCD20 than does rituximab (see Fig. 1B). In contrast to what was observed with ADCC, anti-CD20-mIFNβ was not able to mediate CDC, a property of the unmodified mAb (rituximab) (Fig. 7B).

Figure 7. Anti-CD20-mIFNβ is effective in carrying out ADCC but not CDC.

A. ADCC was assessed by LDH release following incubation of 38C13-huCD20 target cells for 5 h at 37°C with 53 nM of anti-CD20-mIFNβ or rituximab and IL-2-stimulated splenocytes from C3H mice at different effector:target ratios. Percentage cytotoxicity was calculatedas 100 x [(effector:target cell release – effector cell control – spontaneous release) / (maximum release-spontaneous release)]. Data are expressed as mean ± SD for triplicate samples. B. To determine CDC, 38C13-huCD20 cells were incubated in triplicate with rituximab or anti-CD20-mIFNβ at concentrations ranging from 0–53 nM for 15 minutes on ice. Rabbit serum, as a source of complement, was added to a final concentration of 10% and samples incubated at 37°C for 3 h. Following washing, cells were stained with PI and analyzed by flow cytometry to assess viability. Data are expressed as the mean ± SD of the percentage of PI-positive events.

Anti-CD20-mIFNβ has more potent in vivo efficacy than anti-CD20-mIFNα against huCD20-expressing lymphomas, including those with decreased IFNAR expression

We chose to determine the in vivo efficacy of anti-CD20-mIFNβ using 38C13-huCD20, which was previously used to evaluate the efficacy of anti-CD20-mIFNα (21). 38C13-huCD20 expresses physiologic levels of human CD20 and is a model of rituximab-insensitive lymphoma when injected s.c. into syngeneic C3H mice (22). Free recombinant IFNβ has a very short in vivo half-life of 3–5 h (28). However, similar to what had been observed with mIFNα fusion proteins (29), following i.v. injection and distribution from the circulation into the tissues, anti-CD20-mIFNβ cleared from the serum with a half-life of approximately 8 h (data not shown). To rule out any adverse effects of fusion protein on normal host leukocyte populations, mice were treated daily for 3 days with PBS or 100 μg of anti-CD20-mIFNβ, and spleens and lymph nodes harvested 24 hours later for quantitation of B cells, T cells, and myeloid cells. No toxic effects against these leukocyte populations were observed (data not shown).

To compare the in vivo anti-tumor efficacy of anti-CD20-mIFNβ with rituximab and to assess the importance of fusion protein targeting, mice injected s.c. with 5x10³ 38C13-huCD20 tumor cells were treated i.v. on days 5, 6, 7 and 14 post tumor challenge with 100 μg of rituximab, anti-DNS-mIFNβ, or anti-CD20-mIFNβ fusion proteins, followed for tumor growth, and sacrificed when tumors reached 1.4 cm in diameter. In this stringent setting, rituximab increased median survival by only 2 days with no animals cured (p = 0.0035 vs. PBS). Similarly, the untargeted fusion protein anti-DNS-mIFNβ prolonged median survival by only 4 days but did not result in tumor eradication (p = 0.0003 vs. PBS) (Fig. 8A). Only anti-CD20-mIFNβ achieved tumor eradication with 83% of the treated animals cured (p = 0.0003 vs. rituximab and anti-DNS-mIFNβ).

Figure 8. Targeted mIFNβ is effective against syngeneic huCD20-expressing lymphomas in vivo.

A. Anti-CD20-mIFNβ is more effective than anti-DNS-mIFNβ and rituximab in treating established 38C13-huCD20 tumors. Survival and tumor growth curves for mice injected s.c. with 5x10³ tumor cells on day 0 and treated i.v. with 100 μg of anti-CD20-mIFNβ, anti-DNS-mIFNβ, or rituximab on days 5, 6, 7 and 14 post tumor challenge. Groups included 8 mice except for anti-CD20-mIFNβ, where n=6. B. Anti-CD20-mIFNβ is more effective than anti-CD20-mIFNα in treating mice bearing 38C13-huCD20 tumors. Survival and tumor growth curves for mice injected s.c. with 5x10³ tumor cells on day 0 and treated i.v. with 100 μg of anti-CD20-mIFNα or anti-CD20-mIFNβ on days 5, 6, and 7 post tumor challenge. C. Anti-CD20-mIFNβ is more effective than anti-CD20-mIFNα in treating mice bearing 38C13-huCD20 IFNAR^lo tumors. Survival and tumor growth curves for mice injected s.c. with 5x10³ tumor cells on day 0 and treated i.v. with 100 μg of anti-CD20-mIFNα or anti-CD20-mIFNβ on days 5, 6, and 7 post tumor challenge. Arrows indicate days of treatment. Tumor growth curves for the individual mice in each group are included to the right of each survival curve. The numbers at the bottom of the tumor growth curves indicate the number of mice without tumor at day 60.

To directly compare the relative in vivo efficacy of anti-CD20-mIFNα and anti-CD20-mIFNβ fusion proteins, mice injected s.c. with 5x10³ tumor cells were treated with 100 μg of anti-CD20-mIFNα or anti-CD20-mIFNβ fusion protein i.v. on days 5, 6, and 7 post tumor challenge, followed for tumor growth and sacrificed when tumors reached 1.4 cm in diameter. All mice treated with PBS succumbed to tumor by day 14 (Fig. 8B). As observed earlier, anti-CD20-mIFNα was partially effective against the tumor with the death of all of the mice delayed compared to the PBS control and 25% of the mice surviving tumor-free (p < 0.0001). However, anti-CD20-mIFNβ was even more effective, achieving 75% tumor eradication (p = 0.023 compared to anti-CD20-mIFNα) (Fig. 8B). Importantly, using these doses, animals did not display any signs of acute IFN-mediated systemic toxicity such as lethargy, ruffled fur, or weight loss.

In previous studies anti-CD20-mIFNα failed to provide protection against the growth of a tumor (38C13-huCD20 IFNAR^lo) with decreased cell surface expression of the IFNAR resulting from the use of shRNA to target IFNAR1 (21). 38C13-huCD20 IFNAR^lo remains a very aggressive tumor with 100% of PBS treated mice sacrificed by 15 days (Fig. 8C). While treatment with 100 μg of anti-CD20-mIFNα on days 5, 6, and 7 was able to delay tumor growth by one week (p < 0.0001 vs. PBS), all animals succumbed to tumor by day 22. However, anti-CD20-mIFNβ was significantly more effective against this resistant tumor, with the survival of all mice prolonged compared to mice treated with anti-CD20-IFNα, and 25% of the mice remaining tumor-free (p = 0.0003 compared to anti-CD20-mIFNα). Thus, fusion proteins targeting IFNβ to tumor cells may be efficacious therapeutics for the treatment of tumors that are not effectively treated with IFNα.

DISCUSSION

Anti-proliferative effects of IFN have been associated with enhanced stability of the IFN/IFNAR complex. mIFNβ has higher affinity for IFNAR than does mIFNα1. Therefore one goal of the current study was to determine if antibody-targeted mIFNβ would be a more effective anti-cancer therapeutic than antibody-targeted mIFNα1. Indeed we found that fusions with mIFNβ had very potent IFN bioactivity. In contrast to what we had observed with mIFNα1 (21), fusion of mIFNβ to the carboxy-terminus of the heavy chain does not impede its anti-proliferative activity (Fig. 1). In fact, when its ability to inhibit the proliferation of 38C13 cells was analyzed, anti-DNS-mIFNβ and anti-CD20-mIFNβ were both more effective than recombinant mIFNβ. Although it is generally accepted that the type I IFNs function as monomers, early studies had indicated that the functional unit for IFNβ is a dimer (30). In the fusion protein two mIFNβ molecules are present at the end of C_H3 and this close proximity may enhance their function. The structure of mIFNβ may also influence its improved activity in the fusion proteins. mIFNβ contains three predicted N-glycosylation sites, all of which are used (31); mIFNα1 contains only one. mIFNβ also has two deletions in the AB loop proposed to interact with the IFNAR2 subunit of the receptor and lacks the disulfide bridge between the AB loop and helix E which is conserved among all other type I IFNs (32). Thus, it is likely that the interaction of mIFNβ with the IFNAR differs locally from that of mouse and human IFNα and human IFNβ.

Different type I IFNs react differently with the two receptor subunits resulting in variable downstream effects (reviewed in (33)). IFNAR1 is associated with TYK2 and IFNAR2 with JAK1. Interestingly, mIFNβ was found to stimulate apoptosis in pro-B cells from STAT1 deficient mice, but not in pro-B cells from TYK2 deficient mice that had normal activation of STAT1 and STAT2 but were deficient in STAT3 phosphorylation. Inhibition of STAT3 activation in wild-type pro-B cells also blocked apoptosis. mIFNβ but not mIFNα induced programmed cell death in primary murine IL-7-dependent bone marrow derived pro-B cells with TYK2-mediated tyrosine phosphorylation of STAT3 required for mIFNβ to stimulate apoptosis (34).

We found that antibody-mIFNβ fusion protein treatment causes caspase-independent cell death, as evidenced by DNA fragmentation, exit from the cell cycle, and loss of metabolic activity correlating with compromised membrane integrity. DNA fragmentation is seen in 38C13-huCD20 cells undergoing apoptosis with a sub-G1 apoptotic peak characteristic of DNA fragmentation observed (See Figs. 5 and 6). DNA fragmentation during apoptosis is carried out by CAD, a caspase-activated DNase. CAD cannot be synthesized by itself but its inhibitor, ICAD, must be present in the cytoplasm to assist in the correct and productive folding of the CAD polypeptides. When cells are induced to undergo apoptosis, caspase 3 cleaves ICAD to dissociate the CAD:ICAD complex allowing CAD to cleave chromosomal DNA (35). However, cells can be killed by apoptotic stimuli without DNA fragmentation, as MCF-7 breast carcinoma cells that lack active caspase 3 remain sensitive to TNF- or staurosporine-induced apoptosis, although no DNA fragmentation was observed (36). In our studies active caspase 3 is observed at 24 h, a time-point at which DNA fragmentation is observed (Fig. 5). However, treatment with the pan-caspase inhibitor Z-VAD(OMe)-FMK had little to no effect on apoptosis or DNA fragmentation following treatment with anti-CD20-mIFNβ, suggesting that their occurrence is caspase-independent (Fig. 6).

The observed efficacy of antibody-targeted IFN may also relate to the requirement for extended exposure to anti-CD20-mIFNβ for cells to commit to apoptosis. Extended signaling through mIFNβ bound to the receptor is necessary for the cells to commit to DNA fragmentation and apoptosis, as removal of IFN bound for up to 6 h resulted in no DNA fragmentation (Fig. 5A). IFNβ-induced apoptosis in melanoma cells was also found to be delayed and became apparent by Annexin V staining only after ~40 h of treatment (37). IFNα-induced apoptosis of Daudi cells was also seen only 48 h after stimulation with a further increase in proportion up to 72 h (38). This contrasts with what is seen with other apoptotic agents such as campothecin, staurosporine or Fas Ab that induce apoptosis within a few hours of treatment. There was also a delay in cells exiting the cell cycle following treatment with anti-CD20-mIFNβ as evidenced by a decrease in expression of Ki-67 (Fig. 2B).

It should be appreciated that the anti-CD20-mIFNβ fusion protein is more than just the sum of its parts. Anti-CD20-mIFNβ is more effective than anti-CD20 mAb plus mIFNβ in causing both DNA fragmentation and apoptosis in treated cells (Fig. 4). It is not just the fusion per se that is responsible for the enhanced activity since anti-DNS-mIFNβ is less active than anti-CD20-mIFNβ in both the in vitro assays and in treating tumors in vivo. Targeting of IFN by binding to CD20 on the cell surface plays a critical role in the activity and efficacy of anti-CD20-mIFNβ. Targeted mIFNβ is not identical to recombinant mIFNβ in its activity and efficacy.

In addition to directly causing tumor cell death, antibody-IFN fusion proteins may also interact with the host immune effector system. Interestingly, anti-CD20-mIFNβ has unimpaired and possibly enhanced ability to carry out ADCC (Fig. 7A). The binding sites for the FcγRs used for ADCC are in the hinge and the hinge proximal region of C_H2 (39, 40). One FcγR binds to one Fc interacting with the two C_H2 domains in an asymmetric manner. When IgG is bound to a FcγR the two Fab arms are modeled to be free to interact with bulky antigens. Fusion with mIFNβ does not appear to affect this binding site.

Unexpectedly, we found that anti-CD20-mIFNβ was unable to carry out CDC (Fig. 7B). This contrasts with what we have observed with anti-CD20-human IFNα, which we have found to have enhanced ability to carry out CDC against human lymphoma cells (41). Although a crystal structure of C1q, the first component of the classical complement cascade, bound to Fc has not been published, site directed mutagenesis has implicated amino acids at positions 234, 235, 270, 322, 326, 329, 331, and 333 on an exposed face within C_H2 as playing an important role in complement activation (42–48). The presence of mIFNβ in the fusion protein may either alter the conformation of the binding site or impede access to the site, thus limiting CDC. Although the anti-CD20-mIFNβ fusion protein has a half-life of approximately 8 h, that is still far shorter than the extended half-life in mice of approximately 200 h reported for human IgG1 (49). One important determinant of the half-life of IgG is its binding to FcRn at a site located at the C_H2/C_H3 interface in proximity to the C1q binding site and the ability of the fusion protein to bind FcRn may also be impaired. An additional possibility is that the attached mIFNβ binds to IFN receptors and that this binding results in the rapid clearance of the fusion protein.

Antibody-mIFNβ fusion proteins are more effective than antibody-mIFNα fusion proteins in treating tumors in a syngeneic mouse model of B cell lymphoma. Comparison of the in vivo efficacy of anti-CD20-mIFNβ to identical doses of anti-CD20-mIFNα using the 38C13-huCD20 tumor model showed that the mIFNβ fusion protein was more effective, with 75% of the mice remaining tumor-free compared to only 25% of the mice treated with anti-CD20-mIFNα (Fig. 8B). Especially important is that anti-CD20-mIFNβ was also effective in preventing the growth of 38C13-huCD20 IFNAR^lo tumors with 25% of the mice remaining tumor-free while none of the mice treated with anti-CD20-mIFNα survived (Fig. 8C). As discussed above, IFNβ has a higher affinity than IFNα for the IFNAR and its greater in vivo efficacy likely reflects this improved ability to activate the IFN signaling pathway especially in cells with decreased expression of IFNAR. Thus, fusion proteins targeting IFNβ to tumor cells may provide effective therapeutics for the treatment of cancers that are not effectively treated with IFNα.

Numerous studies have documented the potent biologic activities of type I IFNs against B cell malignancies. However, to date their utility has been restricted by dose-limiting toxicities as more effective treatment was found at higher IFN doses (50). Targeting IFN directly to tumors by fusing it to an Ab recognizing a tumor associated antigen appears to be an effective way to overcome these limitations. Our earlier studies have shown that anti-CD20-IFNα fusion proteins possess a markedly improved therapeutic index compared to recombinant IFNα and can eradicate both mouse and human lymphomas in vivo with a small fraction of the IFNα required for an effective systemic dose (21). An alternative approach of targeting IFNα to CD20 expressing cells using a hexavalent anti-CD20 antibody also found strong anti-lymphoma effects in pre-clinical mouse models (51). Our current studies suggest that targeting of a more potent type I IFN such as IFNβ using an Ab recognizing a tumor associated antigen may produce an even more effective cancer therapeutic that is active against tumors resistant to the effects of IFNα. Therefore, we propose that IFNβ fusion proteins warrant further evaluation as cancer therapeutics.