Targeted delivery of interferon-α to hepatitis B virus-infected cells using T-cell receptor-like antibodies

Authors


  • Potential conflict of interest: Nothing to report.

Abstract

During antiviral therapy, specific delivery of interferon-α (IFNα) to infected cells may increase its antiviral efficacy, trigger a localized immune reaction, and reduce the side effects caused by systemic administration. Two T-cell receptor-like antibodies (TCR-L) able to selectively bind hepatitis B virus (HBV)-infected hepatocytes of chronic hepatitis B patients and recognize core (HBc18-27) and surface (HBs183-91) HBV epitopes associated with different human leukocyte antigen (HLA)-A*02 alleles (A*02:01, A*02:02, A*02:07, A*02:11) were generated. Each antibody was genetically linked to two IFNα molecules to produce TCR-L/IFNα fusion proteins. We demonstrate that the fusion proteins triggered an IFNα response preferentially on the hepatocytes presenting the correct HBV-peptide HLA-complex and that the mechanism of the targeted IFNα response was dependent on the specific binding of the fusion proteins to the HLA/HBV peptide complexes through the TCR-like variable regions of the antibodies. Conclusion: TCR-L antibodies can be used to target cytokines to HBV-infected hepatocytes in vitro. Fusion of IFNα to TCR-L decreased the intrinsic biological activity of IFNα but preserved the overall specificity of the protein for the cognate HBV peptide/HLA complexes. This induction of an effective IFNα response selectively in HBV-infected cells might have a therapeutic advantage in comparison to the currently used native or pegylated IFNα. (HEPATOLOGY 2012;56:2027–2038)

Therapy for chronic hepatitis B (CHB) virus infection has made steady progress but several problems remain unsolved. Nucleoside analog therapeutics (e.g., lamivudine, adefovir, telbuvidine) directly suppress hepatitis B virus (HBV)-DNA synthesis, reduce viral replication, and improve histological signs of liver disease, but rarely achieve clearance of infection or sustained viral control.1, 2 A better durability profile of treatment response can be achieved with interferon-α (IFNα), a cytokine with known antiviral, immunomodulatory, and antiproliferative effects.3

Patients responsive to IFNα treatment have a lower rate of relapse and can achieve HBV surface antigen (HBsAg) clearance, but such responses are typically only seen in a minority of treated patients. In addition, the long-term tolerability of IFNα is low, with side effects such as flu-like illness, fatigue, fever, and bone marrow suppression being very common.4, 5

The reason why the majority of CHB patients do not respond to IFNα treatment has not been formally demonstrated, but in vitro experiments and animal models have shown that HBV can inhibit intracellular IFNα-mediated responses,6, 7 thereby potentially limiting the cytokine antiviral effect. Increasing the IFNα dose can be a strategy to counterbalance the HBV-mediated inhibitory effects, but current forms of IFNα are already at their maximum tolerated dose and duration. Targeting IFNα to the liver, while minimizing systemic effects, may be a strategy to increase its efficacy locally and may increase both efficacy and tolerability of IFNα-based therapy of HBV infection.

Strategies for selective delivery of cytokines to specific organs have already shown efficacy.8-10 Direct production of IFNα within the liver through different viral vectors experimentally improved induced liver cirrhosis in rats9 or inhibited HBV replication in a duck model of HBV infection.8 Here, we took advantage of recently developed antibodies that mimic the exquisite specificity of HBV-specific T cells (called T-cell receptor-like [TCR-L] antibodies)11 to produce TCR-L/IFNα fusion proteins targeting HBV-peptide human leukocyte antigen (HLA)-class I complexes expressed on HBV-infected hepatocytes. The ability of such fusion proteins to selectively exert biological activity mediated by IFNα on cells that present HBV antigens was determined.

Abbreviations

CHB, chronic hepatitis B; IFNα, interferon-α; ISRE, interferon stimulated response element; HBV, hepatitis B virus; HLA, human leukocyte antigen; TCR-L, T-cell receptor-like antibodies.

Materials and Methods

A detailed description is provided in the Supporting Materials.

Murine TCR-L Antibodies.

HLA-A2/peptide complexes (A201/HBs183-91 and A201/HBc18-27) were produced and used to immunize BALB/c mice. Splenocytes from immunized mice were fused using PEG1500 with NS1 myeloma cells.

Chimeric Mouse-Human TCR-L Parental Antibodies.

The gene segments encoding the mouse TCR-L kappa light (Vκ) and heavy chain variable regions (VH) were fused to gene segments encoding the human kappa light chain constant region (Cκ) or the human gamma-1 heavy chain constant region (CH1-Hinge-CH2-CH3), respectively.

Chimeric Mouse-Human TCR-L/IFNα Fusion Proteins.

Antibody-interferon fusion genes were assembled by cloning a chemically synthesized DNA fragment coding for mature human IFNα2a and a glycine-serine linker consisting of two Gly4Ser repeats (heavy chain…LSPG—GGGSGGGGS—IFNα) to the C-terminus of the TCR-L antibody heavy chain genes.

TCR-L and TCR-L/IFNα Binding to Target Cells.

Target cells were first incubated with cTCR-L or sTCR-L or mouse isotype control antibodies. After washing, antimouse IgG-APC-conjugated secondary antibodies were added.

Immunohistochemical Staining of Liver Biopsies.

Cryostat sections of liver biopsies were fixed in formalin-free tissue fixative and blocked with dual endogenous enzyme block. Sections were then incubated with sTCR-L or cTCR-L. Cellular cytoskeleton were visualized with anti-cytokeratin (CK3-6H5)-FITC.

Interferon-Stimulated Response Element (ISRE) Reporter Assay.

HepG2 cells were transfected with pISRE-Luc (Stratagene). Forty-eight hours posttransfection, cells were treated with HBV-TCR-L/IFNα ± 10 μg/mL HBV peptide, Roferon, or Pegasys in 10-fold serial dilutions for 24 hours. Cells were then incubated with SteadyGlo substrate for 1 hour followed by measurement of luciferase activity.

Measurement of Interferon Stimulated Gene (ISG) by Quantitative Reverse-Transcription Polymerase Chain Reaction (RT-PCR).

HepG2 cells were pulsed with 10 μg/mL HBc18-27 core or HBs183-91 surface peptides and incubated with cTCR-L/IFNα, sTCR-L/IFNα, sTCR-L, human IgG1 isotype control, or Roferon. Q-RT-PCR was performed to detect ISGs 6 hours posttreatment.

Quantification of S-Antigen Secretion from HBV-Transfected HepG2 cells.

HepG2 cells transfected with pEco63-1.3 (HBV 1.3x expression plasmid constructed using HBV sequence from pEco63) were treated with cTCR-L/IFNα ± 10 μg/mL HBc18-27 peptide, Roferon, or Peg-IFNα (Pegasys). After 72 hours the viral supernatant was collected and S-antigen was quantified using an HBsAg chemiluminescence immunoassay kit.

CD8 T-Cell Activation Assays.

HBV-specific CD8 T cells were cocultured with HBV peptide-pulsed or not pulsed HepG2 cells with TCR-L/IFNα, fixed, and stained for IFNγ-PE.

CXCL-9 and CXCL-10 Production.

HepG2 were incubated with HBV-specific CD8 T cells alone or with TCR-L/IFNα overnight. Supernatants were collected after 18 hours and concentrations of CXCL-9 and CXCL-10 were measured using the Cytometric Bead Array System (BD Biosciences, San Jose, CA). In selected experiments, intracellular cytokine staining using fluorescent-conjugated anti-CXCL-10 antibodies was used.

Results

Characterization of HBV-Specific TCR-L Antibodies.

We recently reported the production and characterization of a murine IgG1 antibody specific for the surface HBs183-91/A*02:01 complex (sTCR-L).11 A second antibody specific for core HBc18-27/A*02:01 complex, a dominant HLA-A201 HBV-epitope, was produced using the same method. Figure 1 shows the specificity data of both cTCR-L (specific for HBc18-27/A*02:01) and sTCR-L (specific for HBs183-91/A*02:01). Both TCR-Ls selectively recognize HLA-A*02:01+ targets pulsed with the respective specific peptides (Fig. 1A). In addition, both TCR-Ls bound to HBV-producing HepG2 cells, but did not bind to HepG2 cells that had not been transfected with HBV (Supporting Fig. 1) or cells pulsed with other A*02:01 binding peptides. The specific recognition of HBc18-27 pulsed cells or HBV-producing cells by cTCR-L antibodies was not influenced by the presence of serum from CHB patients (data not shown), as demonstrated for sTCR-L.11 The two antibodies were tested for their ability to recognize naturally infected cells by immunohistochemistry on frozen liver biopsies from patients with CHB (Fig. 1B) or by staining of isolated hepatocytes purified from CHB patients biopsies (Fig. 1C). Both antibodies specifically recognized, with variable frequencies, the hepatocytes of HLA-A*02:01+ patients with CHB, but they did not bind to hepatocytes purified from HLA-A*02:01-negative subjects (Fig. 1B,C).

Figure 1.

Characterization of cTCR-L and sTCR-L binding. (A) Histograms represent the different staining profiles of sTCR-L and cTCR-L. T2 cells were pulsed with 1 μM each of the indicated peptides and then incubated with 100 nM of TCR-L antibodies, washed, and stained with antimouse IgG1 APC-conjugated antibody. Cells were analyzed by flow cytometry. (Ba) Needle biopsies of HLA-A*02 (+) or (−) HBV-infected livers were cryosectioned, fixed with RCL-2, and stained with 2 μg/mL of TCR-L overnight at 4°C. Antimouse HRP polymer were used to detect bound TCR-L and Alexa Fluor 647-tyramide signal amplification kit was used to visualize TCR-L binding (red). The same slides were costained with anti-cytokeratin (green) to show cellular composition and liver architecture. Both cTCR-L and sTCR-L stained HLA-*02 (+) liver but were not detectable in HLA-*02 (−) liver. (Bb) Staining with s-TCR-L on HLA-*02 (+) liver biopsy with a larger field of view to include liver architecture (green). Envelope antigen expressing hepatocytes were detected by sTCR-L (red). sTCR-L bound exclusively to infected hepatocytes and not to endothelial or stromal cells of the hepatic triad, containing the hepatic artery, vein, and bile ducts (white arrow). Microscopy pictures have been normalized to the green channel. (C) The scatter diagram shows the frequency of cTCR-L/albumin and sTCR-L/albumin double-positive cells derived from the liver biopsies of indicated HLA-A2-positive or -negative patients. (D) Ability of cTCR-L and sTCR-L to recognize HBs183-91 or HBc18-27 peptides in association with the indicated HLA-*02 subtypes. HBV-B cells expressing the indicated HLA-A2 subtypes or HLA11 and HLA-A24 (negative controls) were incubated with 1 μM of the HBV peptides indicated above each graph. Bars represent the percentage of cells which bound the indicated TCR-L.

The possible broadness of applicability of both cTCR-L and sTCR-L in patients of different ethnicities infected by different HBV genotypes was studied by analyzing the TCR-Ls ability to recognize the peptides of the respective HBc18-27 and HBs183-91 epitopes of HBV genotypes A, B, C, D, E, and F presented by different HLA-A*02 allotypes. Amino acid sequences of the corresponding peptides are shown in Fig. 1D with a description of the HLA-A02* subtypes present in distinct human populations listed in Fig. 1D. HBV genotypes A, C, and D have a one amino acid polymorphism difference in the HBs183-91 peptide, as compared to genotypes B, E, and F. Similarly, genotypes A, D, E, and F have a one amino acid difference in the HBc18-27 peptide as compared to genotypes B and C (Fig. 1D).

The high specificity of both TCR-Ls for the cognate HLA-peptide complex was evident. The presence of a lysine (K) instead of an arginine (R) at position 187 of the HBs183-91 peptide (HBV genotypes B, E, F) abolished sTCR-L recognition. Note that, in contrast to sTCR-L antibodies, we have shown that HBs183-91-specific CD8T cells are able to recognize the 187K s183-91 peptide,11 demonstrating that the 187K s183-91 peptide is indeed bound to the HLA-A2* class I molecules on the surface of the cells. In addition, sTCR-L recognized the HBs183-91 HBV genotypes A, C, and D peptides only in the context of A0201, A0202, and A02011.

The cTCR-L showed a broader HBV genotype recognition profile, because it also recognized the HBc18-27 peptide sequence with isoleucine (I) instead of valine (V) at position 27 (characteristic of HBV genotypes B and C) when presented by A*02:01, A*02:02, A*02:11. Such promiscuity of recognition may be explained by the fact that the V27I polymorphism affects an epitope anchor residue which resulted in weaker MHC-binding ability, but does not effect TCR-recognition.12 Specific recognition was also detected when A*02:07 molecules presented either HBc18-27V or HBc18-27I peptides. However, this specific binding was weaker than the ones obtained with HLA- A*02:01, A*02:02, and A*02:11 as presenting molecules, and cTCR-L showed crossreactivity to nonpeptide pulsed target HLA- A*02:07 + cells. (Fig. 1D).

Design, Preparation, and Biochemical Characterization of TCR-L/IFNα.

The exquisite binding specificity of the two murine TCR-Ls suggests that they could be used to target IFNα to HBV-infected cells. For this purpose, we generated antibody fusion proteins in which mature human IFNα (without signal sequence) was linked to the C-termini of the two heavy chains of chimeric mouse-human TCR-L antibodies.

The fusion proteins called TCR-L/IFNα were constructed by joining the murine TCR-like antibody light and heavy chain variable domains (VL or VH) to the human kappa light chain constant region (CK) and the human γ-1 heavy chain constant region (CH1-Hinge-CH2-CH3), respectively. A flexible glycine-serine linker consisting of two Gly4Ser repeats was used for joining human mature IFNα2a to the penultimate C-terminal amino acid of the antibody heavy chain (heavy chain —LSPGGGGSGGGGS— IFNα2a). The last basic amino acid of the antibody heavy chain, a Lys, was removed in order to eliminate a potential cleavage site for endoproteinases with trypsin-like activity. This C-terminal Lys of an antibody heavy chain can be deleted without influencing antibody functions because this amino acid is cleaved off during secretion by cellular carboxypeptidases to a variable extent, as observed in many antibody preparations.13, 14 A schematic representation of the chimeric mouse-human TCR-L/IFNα and a three-dimensional model are displayed in Fig. 2A,B, respectively.

Figure 2.

Design of the TCR-L/IFNα fusion molecules. (A) Schematic diagram of the TCR-L/IFNα fusion proteins. The TCR-L/IFNα fusion proteins are composed of the chimeric mouse-human TCR-L antibody and the mature human IFNα2a, which is fused to the C-terminal ends of the human gamma-1 antibody heavy chains by way of a glycine-serine linker. The peptide sequence for the linker consisting of two Gly4Ser repeats is shown. (B) 3D-model of TCR-L/IFNα. The 3D-model of the TCR-L/IFNα molecule was constructed using a known human IgG1 antibody structure, the published human IFNα2a structure, and a modeled glycine-serine linker. (C) SDS-PAGE analysis of cTCR-L and sTCR-L molecules. The various recombinantly produced proteins were analyzed by SDS-PAGE after purification to homogeneity under nonreducing (left panel) and reducing (right panel) conditions.

The TCR-L/IFNα fusion proteins were expressed transiently in HEK293 human embryonic kidney cells and purified by two chromatographic steps to greater than 95% purity and an aggregate content below 1%. Purity, absence of aggregates, and correct composition of the tetrameric cTCR-L/IFNα were analyzed by reducing and nonreducing sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) (Fig. 2C,D) and analytical size-exclusion chromatography (data not shown). The migration of the antibodies under nonreducing conditions in SDS-PAGE was consistent with the molecular masses expected from the assembly of the four protein chains that together form the TCR-L/IFNα fusion protein (Fig. 2C). The molecular masses of the cTCR-L/IFNα and sTCR-L/IFNα as predicted from the amino acid sequences are 184998 Da and 184242 Da, respectively. SDS-PAGE analysis of the TCR-L/IFNα under reducing conditions further revealed a band for the TCR-L/IFNα heavy chain with an apparent molecular mass of ≈74 kDa in comparison to the parental antibody heavy chain with an apparent molecular mass of ≈55 kDa, reflecting the increased mass from IFNα addition and expected glycosylation heterogeneity caused by N-linked antibody Fc-glycosylation and potential O-linked glycosylation of the IFNα moiety. The identity and integrity of the protein sequences of the antibody fusion proteins were verified by mass spectrometry after removal of N- and O-glycans by enzymatic treatment with peptide-N-glycosidase (50 mU), neuraminidase (50 mU), and O-glycosidase (2.5 mU) as described for HEK293-derived CrossMab antibodies15 (data not shown).

IFNα Biological Activity of TCR-L/IFNα.

Fusion of IFNα to large molecules, like polyethylene glycol (Peg) or albumin, increases its plasma half-life dramatically16-18 but the effect of such modifications on the IFNα biological activity remains difficult to predict. Different Peg-IFNα or albumin/IFNα conjugates showed 1%-30% of the biological activity of the native IFNα,18, 19 whereas IL-2 fused to an antitumor antibody retained its full biological activity.20

The biological in vitro activity of our two TCR-L/IFNα fusion proteins in comparison to native unconjugated IFNα was initially tested using Mardin-darby bovine kidney (MDCK) cells infected with vesicular stomatitis virus (VSV). MDCK cells do not express HLA-A*02 and, therefore, inhibition of VSV replication mediated by IFNα conjugates reflects the intrinsic IFNα activity of the protein conjugates in the absence of targeting. TCR-L/IFNα were much less active in suppressing VSV replication than IFNα2a (Roferon A) and retained only about 3% of IFNα activity on a molar basis (data not shown).

We then tested the IFNα biological activity of TCR-L/IFNα on HepG2 cells by analyzing the expression of selected ISGs (MX1, OAS1) by q-PCR or by using an ISRE luciferase reporter system transiently expressed in HepG2 cells. Figure 3A displays the initial results obtained using the sTCR-L/IFNα fusion protein. Although Roferon activity was independent of any treatment of HepG2 cells, sTCR-L/IFNα-activated ISG genes exclusively in HepG2 cells pulsed with the relevant HBs183-91 peptide but not in HepG2 cells left untreated or pulsed with an irrelevant peptide (Fig. 3A). The exclusive activity of TCR-L/IFNα on target cells expressing the cognate HBV/peptide was confirmed by the stimulation of ISRE luciferase reporter gene expression (Supporting Fig. 2).

Figure 3.

Biological activity of TCR-L/IFNα fusion proteins. (A) Bars represent the fold changes of the expression of indicated IFNα responsive genes after 6 hours. Different bars represent the different concentrations of IFNα (Roferon), sTCR-L/IFNα, sTCR-L, and IgG with or without 10 μg/mL of the indicated HBV peptides. q-RT-PCR was performed to quantify expression of the OAS1 and MX1 genes. Values were normalized for 18S rRNA expression and represent three independent experiments. (B) Same as (A) except cTCR-L/IFNα ISG induction data are shown. (C) Bars represent ISG MX1 and OAS1 induction by indicated concentrations of cTCR-L/IFNα and sTCR-L/IFNα expressed as percent of the maximal corresponding induction by IFNα at the highest concentration (220 pM). (D) HepG2 cells were transfected with pISRE-luc reporter plasmid, treated with serially diluted IFNα proteins, IFNα, Pegasys, cTCR-L/IFNα, with or without peptide pulsing. Preincubation of HBc18-27 peptide pulsed HepG2 cells with 10 μg/mL cTCR-L for 30 minutes inhibited the enhanced activity of the IFNα reporter system by cTCR-L/IFNα. Luciferase activity was measured 48 hours posttreatment.

Note that, although it has been reported that unconjugated TCR-L can have an effect on target cells (i.e., induction of apoptosis21), sTCR-L did not induce any activation of ISG on the targets (Fig. 3A).

A comparative analysis of the ability of the two TCR-L/IFNα fusion proteins to activate IFNα-stimulated genes showed that cTCR-L/IFNα induced an IFNα response in specific target cells of identical magnitude to Roferon (Fig. 3B), whereas sTCR-L/IFNα was less potent (Fig. 3A). The difference in the relative ISG induction activity of the two TCR-L/IFNα molecules in peptide pulsed HEPG2 cells in comparison to IFNα was calculated (Fig. 3C). Induction of both MX1 and OAS1 expression by cTCR-L/IFNα were clearly better than the induction by sTCR-L/IFNα.

The HBV peptide-dependent IFNα activity of the TCR-L/IFNα suggests that this activity requires binding to HBV-peptide/HLA-complexes. We thus tested whether blocking the binding of TCR-L/IFNα to HBV-HLA-complexes could abolish the IFNα activity on the targets. Figure 3D shows the results obtained with cTCR-L/IFNα in which the HBc18-27 peptide-pulsed HepG2 cells were preincubated with an excess of unconjugated cTCR-L prior to the addition of cTCR-L/IFNα. The preincubation of cTCR-L suppressed the activity of cTCR-L/IFNα on specific targets down to the levels achieved on unpulsed HepG2 cells (Fig. 3D).

Activity of TCR-L/IFNα on HBV-Infected Hepatocytes.

The specific biological activity of different concentrations of TCR-L/IFNα in comparison to IFNα was further tested on hepatocytes isolated from two HLA-A*02:01-positive and two HLA-A*02:01-negative donors and infected in vitro by HBV. Because the availability of HLA-A*02:01-positive hepatocytes was limited, these experiments were performed using only the more potent cTCR-L/IFNα fusion protein. Induction of ISGs was determined in the presence or absence of HBc18-27 peptide pulsation and the biological activity of TCR-L/IFNα was compared with that of IFNα. IFNα induced expression of ISGs (OAS1 and IFI6) in a dose-dependent manner in HLA-A*02-positive and -negative HBV-infected hepatocytes (Fig. 4A). However, cTCR-L/IFNα significantly induced ISG gene expression in HLA-A*02-positive HBV-infected hepatocytes but only weakly at high concentrations in HLA-A*02-negative ones. Addition of HBc18-27 peptide to infected HLA-A*02-positive hepatocytes slightly increased ISG induction, whereas it did not have any effect on HLA-A*02-negative hepatocytes (Fig. 4A). The difference between the relative activity of cTCR-L/IFNα in comparison to IFNα in HLA-A*02-positive and -negative HBV-infected hepatocytes was calculated (Fig. 4B). ISGs induction by cTCR-L/IFNα was similar to the one of IFNα in HLA-A*02-positive hepatocytes (1.9- and 1.6-fold lower) but was much lower (6.8- and 7.8-fold lower) in HLA-A*02-negative HBV-infected hepatocytes. The addition of peptides did not have any significant effect (Fig. 4B). These results demonstrate the specific activity of TCR-L/IFNα on HBV-infected primary hepatocytes, confirming that expression of the correct HLA-HBV peptide complex is required for efficient induction of IFNα signaling and that this induction can be mediated by HLA-HBV peptide complexes generated in the context of natural infection.

Figure 4.

TCR-L/IFNα biological activity on human HBV-infected primary hepatocytes. (A) Primary human hepatocytes from HLA-A2-positive (+) and HLA-A2 (−)-negative donors were infected with HBV and treated with serial dilutions of IFNα, cTCR-L, or cTCR-L/IFNα with or without 10 μg/mL HBc18-27 peptide. q-RT-PCR was performed to detect expression of OAS1 and IFI6. Bars represent the fold changes of the expression of indicated IFNα stimulated genes 14 hours posttreatment/infection. Values were normalized for 18S rRNA expression. (B) Mean and SE of fold differences in OAS1 and IFI6 gene expression induction by cTCR-L/IFNα with and without pulsing in comparison to IFNα at all corresponding doses are shown. Statistical differences of relative ISG inductions of cTCR-L/IFNα with or without 10 μg/mL HBc18-27 peptide (comparing to IFNα) in HLA-A2 (+) and HLA-A2 (−) donor hepatoyctes were analyzed by using one-pair t test and the P values are indicated.

Specific Delivery of TCR-L/IFNα to HBV-Expressing Targets.

To further explore the mechanism of the targeted activation of IFNα-induced genes on cells expressing the cognate HBV peptide/HLA complexes, we compared the binding of cTCR-L/IFNα and sTCR-L/IFNα, and of the native, unconjugated TCR-L antibodies, to cells expressing the respective HBV-peptide/HLA-complexes. It was initially considered that the overall binding affinity of TCR-L/IFNα might be dominated by the IFNα part of the molecule because the affinities [as determined by surface plasmon resonance (SPR)] of the two TCR-Ls for their specific HBV-peptide/HLA complexes (cTCR-L Kd = 22 nM; sTCR-L Kd = 32 nM) were lower than the published affinity of the native unconjugated IFNα2a for its own high-affinity receptor (Kd = 5 nM).22 However, because it has been described that conjugation of IFNα to other molecules can substantially alter the affinity to its receptor,19 the binding of the conjugated molecules cTCR-L/IFNα and sTCR-L/IFNα to HepG2 cells stably transfected with HBV genotype D (HepG2-13) or untransfected HepG2 control cells was studied. Figure 5A shows that both cTCR-L/IFNα and sTCR-L/IFNα fusion proteins bound specifically to the HBV-transfected cell line similarly or better than the corresponding unconjugated TCR-L antibodies. Similar specific binding was also seen using HBV peptide-pulsed HepG2 cells or other HLA-A*02+ cell lines, whereas there was no staining observed using a control IgG1/IFNα fusion protein or using cell lines not expressing HLA-A*02 (data not shown). To further assess the binding specificity of the cTCR-L/IFNα and sTCR-L/IFNα fusion proteins, we analyzed their ability to bind specifically to target cells expressing HBV epitopes in a mixed HBV-positive and HBV-negative cell population. For this purpose, HBV-producing HepG2-13 and the parental HepG2 cells were labeled with two different concentrations of CSFE, mixed in a 1:1 ratio, and then stained with cTCR-L/IFNα fusion proteins. Binding of fusion protein to the target was measured by flow cytometry using a fluorescent antihuman IgG1 secondary antibody. Indeed, cTCR-L/IFNα fusion proteins bound preferentially to HBV-producing target cells (HepG2-13), whereas binding to parental cells was negligible (Fig. 5B). These data also demonstrate that TCR-L/IFNα fusion proteins maintain the binding properties of the respective TCR-L antibodies. The selective detection of IFNα biological activity on cells expressing the cognate HBV peptide/HLA-complex is thus the direct consequence of the exclusive targeting of IFNα to cells expressing the cognate HBV-peptide/HLA-complexes, whereas IFNα biological activity on cells not expressing the cognate complexes is dramatically reduced by the covalent attachment of IFNα to the antibodies.

Figure 5.

Characterization of TCR-L/IFNα binding ability. (A) Indicated concentrations (5/50/500 pM) of TCR-L/IFNα and TCR-L were used to stain HepG2 and HepG2-13 (hepatic cell line producing HBV). Specific binding is indicated as frequency of cells stained and mean fluorescence intensity (MFI) of the respective target. (B) Dotplot diagrams show the ability of c-TCR-L/IFNα to bind to HepG2 and HepG2-13 present in a single well and labeled with (HepG2) or without (HepG2-13) CSFE. The data shown are representative of three independent experiments.

Antiviral and Immunomodulatory Effects of TCR-L/IFNα.

The therapeutic efficacy of IFNα treatment is linked to its antiviral and immunomodulatory effects. We tested whether our TCR-L/IFNα could specifically induce antiviral and immunomodulatory effects. The antiviral effector function of TCR-L/IFNα was measured using a system in which HepG2 cells were transfected with the entire HBV genome. These HBV-transfected HepG2 cells secrete HBsAg that is quantifiable in the supernatants of the cells 4-5 days after transfection. In the experiment shown in Fig. 6A, HepG2 cells were treated with the indicated concentrations of either cTCR-L/IFNα ± HBc18-27 peptide, or IFNα 6 hours after transfection with the HBV construct. On day 4 supernatants were collected and the respective concentrations of HBsAg were quantified by enzyme-linked immunosorbent assay (ELISA). Although IFNα significantly reduced the amount of HBsAg secreted into the cell culture medium, the addition of cTCR-L/IFNα fusion protein did not have any effect on HBsAg production by the transfected cells. However, if the same cells were pulsed with HBc18-27 peptide at the time of cTCR-L/IFNα treatment, significant inhibition of HBsAg secretion was observed, similar to the levels detected with the control IFNα molecule (Fig. 6A). Under these experimental conditions (HBV transfection) therefore, the expression of endogenous HBV peptide may have not been sufficient to allow adequate levels of TCR-L binding. Nevertheless, these results show that the cTCR-L/IFNα fusion molecule possesses an HBV-specific antiviral effect, dependent on the level of HBV peptide/HLA-complexes on the surface of target cells.

Figure 6.

Targeted antiviral and immunomodulatory effect of TCR-L/IFNα. (A) TCR-L/IFNα inhibition of HBsAg production. HepG2 cells were transfected with pEco63-1.3 (HBV 1.3x plasmid DNA) for 6 hours followed by treatment with 24 pM or 8 pM of cTCR-L/IFN ± HBc18-27 peptide or Roferon for 3 days. Percent inhibition of HBsAg secretion was calculated based on transfected, untreated control cells. SEM was calculated from three independent experiments. (B) TCR-L/IFNα mediates up-regulation of HLA-class I expression. Surface expression of HLA-class I, MIC-A, and MIC-B was measured by staining with the corresponding specific antibodies. Mean fluorescence intensity (MFI) of the target cells treated with sTCR-L/IFNα or sTCR-L are displayed in the histograms. Original PLC cells (PLC), a hepatocellular carcinoma cell line known to express HBsAg, were HLA-A2-transfected (PLC-A2), HLA-A2-transfected, and pulsed with HBs183-91 peptide (PLC-A2 +HBV pep), or not transfected with HLA-A2 molecules. These different cell lines were treated with 50 pM sTCR-L/IFNα or sTCR-L and stained after 12 hours. (C) Impact of TCR-L/IFNα on HBV-specific CD8 T-cell recognition and CXCL-9 and CXCL-10 production. HepG2 were pulsed or not pulsed with HBc18-27 and HBs183-91 peptides (1 μM). cTCR-L/IFNα (50 pM) and s183-91-specific T cells (E/T ratio; 1/10) were added to the target cells. The frequency of IFNγ-producing HBV-specific CD8 T cells out of total T cells was measured after 6 hours of incubation by intracellular cytokine staining. Supernatants of different experimental conditions were collected after 18 hours of incubation and analyzed for the content of CXCL-9 and CXCL-10 with a microbeads assay. The experiment was repeated twice. Data of one of representative experiment are shown.

We also analyzed whether immunomodulatory functions directly mediated by IFNα could be directed toward the specific target cells. First, we tested whether TCR-L/IFNα was able to up-regulate HLA-class I molecules or NK-cell costimulatory molecules MICA and MICB.3 Experiments using HBV-transfected or parental HepG2 cell lines showed only a minimal increase of HLA-class I, MICA and MICB expression, and only in the presence of very high concentrations of TCR-L/IFNα (1 nM, data not shown). Using a different cell line of hepatic origin (PLC), an HBsAg+ hepatocellular carcinoma line that can be recognized by HBV-specific CD8T cells,23 we observed the ability of TCR-L/IFNα to slightly up-regulate HLA-class I expression in an antigen-specific fashion and at low concentrations (50 pM). No changes of MICA and MICB expression were observed in these cell lines at all concentrations tested (Fig. 6B). These results suggest that TCR-L/IFNα may facilitate CD8T cell recognition of HBV expressing cells.

We thus analyzed whether TCR-L/IFNα could increase the effect of HBV-specific CD8T cell recognition. We utilized HepG2 cells as target cells and HBV-specific CD8T cells as effectors and tested the effect of TCR-L/IFNα on CD8T activation (IFNγ production) as well as the effect on target cells (secretion IFNγ inducible chemokines CXCL-9 and CXCL-10). To avoid competition between TCR-L/IFNα and HBV-specific CD8T cells for the identical HLA-class I/HBV peptide complexes, we tested the effect of cTCR-L/IFNα (specific for HBc18-27/A*02:01) on HBs183-91-specific CD8T cells by using HepG2 cells pulsed with both HBs183-91 and HBc18-27 peptides.

Figure 6C shows that the CD8T cell function was neither affected by the presence of cTCR-L/IFNα nor by an IgG1/IFNα control (Fig. 6C, CD8) This is consistent with the minor effect of cTCR-L/IFNα on HLA-class I expression in HepG2. However, by measuring the concentration of chemokines present in the supernatants under different experimental conditions, we could demonstrate that TCR-L/IFNα induces specific alteration of target cell responsiveness. Despite identical HBV-specific CD8T activation, chemokine production by target was increased specifically by cTCR-L/IFNα but not by control IgG1/IFNα (Fig. 6C). Importantly, the fusion proteins did not activate chemokine production without concomitant CD8T cell activation. The ability of TCR-L/IFNα to increase chemokine production on specific target cells was further investigated by incubating IFNγ-treated HepG2 cells with sTCR-L/IFNα and analyzing their CXCL-10 production. Only HBs183-91 pulsed cells incubated with sTCR-L/IFNα displayed an increase in CXCL-10 production (Supporting Fig. 3).

Discussion

In this work we demonstrate that TCR-L antibodies can be used to deliver a cytokine selectively to HBV-infected cells. IFNα was chosen as a proof-of-concept therapeutic molecule for a number of reasons. IFNα has been used for many years for the treatment of patients with various cancers or viral diseases. In addition, IFNα has demonstrated efficacy in clearing HBV infection with evidence for both direct antiviral and immunomodulatory effects. We found that genetic fusion of IFNα to TCR-L altered the biological activity of IFNα, resulting in a molecule that maintains its full IFNα activity only on cells expressing the correct HBV-peptide HLA-complex. Linking IFNα to other molecules (like Peg or albumin) has been previously described to reduce its biological activity substantially, which might be due to steric hindrance that prevents the binding of the cytokine to its receptor.18, 19 Our data are consistent with previous reports of conjugation impact on intrinsic IFNα activity but also show that specific binding of TCR-L/IFNα to target cells through recognition of the cognate HBV peptide/HLA complex can unmask the full biological activity of the IFNα on the target cells.

Although the mechanism for this targeted IFNα activity is not completely understood, our data support the hypothesis that directing the fusion protein through the antibody recognition site to the cell surface allows the linked IFNα molecules to interact with its receptors more effectively. Interestingly, of the two targeted IFNα constructs tested in this report, cTCR-L/IFNα, which has higher affinity for the HLA/HBV peptide complex, also demonstrated greater ISG induction activity in target cells. We have previously described the effectiveness of targeting and increasing local concentration of an effector molecule with BFFI, a recombinant fusion protein consisting of an human immunodeficiency virus (HIV) fusion inhibitor peptide and an antibody against HIV receptors, which showed much greater antiviral potency than the antibodies or fusion inhibitors alone.24, 25 The practical consequences of these findings are that therapy with TCR-L/IFNα fusion molecules may activate IFNα-mediated responses exclusively in the liver, at the site of HBV infection, without inducing a systemic IFNα response. Such characteristics could provide a significant therapeutic advantage in comparison to currently used native or pegylated IFNα molecules that are dose limited by their safety and tolerability profiles.26 Increased liver concentrations of TCR-L/IFNα fusion molecules may also permit reduced dosing requirements, further decreasing potential systemic toxicity. On the other hand, it might be argued that the specific delivery of IFNα only to infected cells might limit its efficacy, because it would not alter the antiviral state of noninfected neighboring hepatocytes. In addition, the need for the expression of HBV peptides/HLA-complexes on the target might impair the interferon delivery to hepatocytes with a low HBV replication level. Studies are ongoing to determine if local delivery to the liver and increased efficacy can be realized in vivo, although there are limitations to evaluating this mechanism in animals because of the HLA-class I specificity of our described TCR-L/IFNα molecules. HBV infection in humanized mouse models such as the HBV-infected uPA SCID human liver chimeric mouse27 could be used to evaluate the ability of TCR-L/IFNα targeting on human hepatocytes grafted in mouse liver. Although liver-specific induction of IFNα cannot be evaluated because of lack of crossreactivity of human IFNα with mouse receptors, this animal model may help address the question whether HBV-infected cell specific delivery and/or locally increased concentration of IFNα in the liver could enhance the intrinsic antiviral efficacy

It was recently reported that an anti-CD20 antibody fused with IFNα could increase the therapeutic IFNα effect against B-cell lymphoma without concomitant systemic toxicity,28 supporting the use of antibodies to deliver cytokines to specific target cells. If the concept of TCR-L antibody delivery of IFNα is proven clinically, this strategy could in principle be extended to also target additional HLA classes, and thus potentially providing benefit to increasing proportions of the HBV-infected population. In addition, other therapeutic conjugates could be explored individually or in combination with IFNα to further optimize efficacy. TCR-L based approaches of personalized medicine may offer distinct selectivity, efficacy, and/or safety advantages over broadly applied therapies in treating viral diseases even though further studies are needed to delineate these aspects clinically.

Acknowledgements

The authors thank Wolfgang Schäfer for preparing the 3D-model of the TCR-like antibody interferon-α2 fusions, Sebastian Scholz for antiviral activity testing, Stefan Lorenz for purification of the proteins, Michael Molhoj for mass spectrometry of the TCR-L/IFNα fusion proteins, and Uri Lopatin for intellectual support of this project.

Ancillary