Anchoring interferon alpha to apolipoprotein A-I reduces hematological toxicity while enhancing immunostimulatory properties§

Authors

  • Jessica Fioravanti,

    1. Division of Hepatology and Gene Therapy, Center for Applied Medical Research, University of Navarra, Pamplona, Navarra, Spain
    Search for more papers by this author
  • Iranzu González,

    1. Division of Hepatology and Gene Therapy, Center for Applied Medical Research, University of Navarra, Pamplona, Navarra, Spain
    Search for more papers by this author
  • José Medina-Echeverz,

    1. Division of Hepatology and Gene Therapy, Center for Applied Medical Research, University of Navarra, Pamplona, Navarra, Spain
    Search for more papers by this author
  • Esther Larrea,

    1. Division of Hepatology and Gene Therapy, Center for Applied Medical Research, University of Navarra, Pamplona, Navarra, Spain
    Search for more papers by this author
  • Nuria Ardaiz,

    1. Division of Hepatology and Gene Therapy, Center for Applied Medical Research, University of Navarra, Pamplona, Navarra, Spain
    Search for more papers by this author
  • Gloria González-Aseguinolaza,

    1. Division of Hepatology and Gene Therapy, Center for Applied Medical Research, University of Navarra, Pamplona, Navarra, Spain
    Search for more papers by this author
  • Jesús Prieto,

    1. Division of Hepatology and Gene Therapy, Center for Applied Medical Research, University of Navarra, Pamplona, Navarra, Spain
    Search for more papers by this author
  • Pedro Berraondo

    Corresponding author
    1. Division of Hepatology and Gene Therapy, Center for Applied Medical Research, University of Navarra, Pamplona, Navarra, Spain
    • Prof. Jesús Prieto Division of Hepatology and Gene Therapy, Center for Applied Medical Research University of Navarra, Pamplona, Navarra, Spain
    Search for more papers by this author
    • These authors share senior authorship.

    • fax: +34948194717.


  • Potential conflict of interest: Nothing to report.

  • Supported by the agreement between FIMA and the “UTE project CIMA,” Fundación Pedro Barrié de la Maza and Condesa de Fenosa and Red de Inmunoterapia INMUNONET-SOE1/P1/E014. J.F. and J.M-E. were supported by a fellowship of Spanish Fondo de Investigación Sanitaria. P.B. was supported by a Juan de la Cierva contract from Ministerio de Educación y Ciencia and a Miguel Servet contract from Instituto de Salud Carlos III, FIS.

Abstract

Interferon alpha (IFNα) is widely used for the treatment of viral hepatitis but substantial toxicity hampers its clinical use. In this work, we aimed at improving the efficacy of IFNα therapy by increasing the IFNα half-life and providing liver tropism. We selected apolipoprotein A-I (ApoA-I) as the stabilizing and targeting moiety. We generated plasmids encoding IFNα, albumin bound to IFNα (ALF), or IFNα linked to ApoA-I (IA) and mice were treated either by hydrodynamic administration of the plasmids or by injection of the corresponding recombinant proteins or high-density lipoproteins containing IA. The plasma half-life of IA was intermediate between IFNα and ALF. IA was targeted to the liver and induced higher hepatic expression of interferon-stimulated genes than IFNα or even ALF. IA exhibits stronger in vivo antiviral activity than IFNα and the hematologic cytopenic effects of IA are milder than those observed when using IFNα or ALF. In contrast to IFNα, IA does not cause activation-dependent cell death of lymphocytes in vitro. Accordingly, in vivo studies showed that IA boosts T-cell immune responses more efficiently than IFNα or ALF. The difference in immunostimulatory activity between IFNα and IA disappears in scavenger receptor class B type I (SR-BI) knockout mice, suggesting that crosstalk between SR-BI and IFNα receptor is essential for enhanced induction of cytotoxic T cells by IA. Conclusion: Anchoring IFNα to ApoA-I prolongs the half-life of IFNα and promotes targeting to the liver. Importantly, the fusion protein shows increased immunostimulatory properties and lower hematological toxicity. (HEPATOLOGY 2011;)

Interferon alpha (IFNα) is a key component of the innate immune system and plays an essential role in the defense against viral infections. In addition to direct antiviral effects,1 IFNα displays antiproliferative,2 proapoptotic,3 immunomodulatory,4 and antifibrogenic activities.5 Recombinant IFNα is widely employed for the treatment of chronic viral hepatitis and neoplastic diseases6, 7 but a number of side effects limit its use. Alterations in the hematopoietic system, mainly thrombocytopenia and leukopenia, are among the most important unwanted consequences of this therapy. As a result, IFNα is not indicated in patients with low platelet or leukocyte counts, which is the case with many subjects with advanced chronic viral infection.8 Another limitation of IFNα-based therapies is its short half-life in plasma. The therapeutic activity of IFNα in chronic viral hepatitis is enhanced by formulations that prolong its persistence in the circulation, including pegylation of the molecule (PEG-IFNα)9 or its fusion with stabilizing proteins such as albumin.10 However, these modifications of the IFNα molecule do not provide hepatic tropism, a property that would be desirable in the treatment of liver diseases such as chronic viral hepatitis. We therefore designed an IFNα fusion protein that combines both increased half-life and liver tropism. We selected apolipoprotein A-I (ApoA-I), the main protein component of high-density lipoproteins (HDLs), as the stabilizing and targeting moiety. HDLs are generated in the liver and remove cholesterol from peripheral tissues for delivery to hepatocytes.11 Consistent with the key role of HDLs in the reverse cholesterol transport, it has been shown that ApoA-I accumulates preferentially in the liver following its systemic administration.16 Scavenger receptor class B type I (SR-BI) plays a crucial role in HDL biology.13 After binding to SR-BI, HDLs mediate the uptake of cholesteryl esters and phospholipids from the cells and promote cytoprotective functions by imperfectly understood mechanisms involving multiple interactions between HDLs (lipid or protein moiety) and cell surface receptors.14 SR-BI is expressed at low levels in a wide variety of cells, and at high levels in the liver, adrenal glands, ovaries, testis, intestinal cells, phagocytes, and endothelial cells.11, 15

The present work evaluates the properties of a new fusion protein designed to increase the half-life of IFNα and to target the liver. A key finding of this work was the observation that by linking IFNα to ApoA-I, IFNα exhibits a marked reduction of hematological toxicity and a substantial increase in immunostimulatory activity.

Abbreviations

ALF, albumin bound to interferon alpha; ApoA-I, apolipoprotein A-I; EMCV, encephalomyocarditis virus; HDL, high density lipoprotein; IA, interferon alpha linked to apolipoprotein A-I; IFNα, interferon alpha; PLT, platelets; SR-BI, scavenger receptor class B type I.

Materials and Methods

Cell Lines.

The CT-26 cell line derived from BALB/c colorectal carcinoma, mouse-isolated splenocytes, and L929 cell line (mouse fibroblasts, American Type Culture Collection, LGC Promochem, Molsheim, France) were cultured as indicated in the Supporting Information Methods.

Animal Handling.

Female immunocompetent BALB/c or C57BL/6 mice between 5-7 weeks old were from Harlan; B6;129S2-Srb1tm1Kri (003379) were from the Jackson Laboratory. The mice were treated in accordance with the guidelines of the Center for Applied Medical Research (CIMA, Pamplona, Spain). Hydrodynamic administration of plasmids and infection with encephalomyocarditis virus (EMCV) were performed as mentioned in the Supporting Information Methods.

In Vivo Experiments with Recombinant IFN (rIFN),

Isolated HDL Containing IA Fractions, Recombinant Mouse IA, and Recombinant Human IA. Biodistribution and pharmacokinetic profiles were performed using recombinant IA (rIA) and rIFN with 6xHIS tag, a purification that allowed high recovery of IFN protein (both of them produced by GenScript, Piscataway, NJ). For bioactivity assays, we used mouse rIFN alpha (CHO derived mouse, Hycult Biotechnol, Uden, Holland), isolated HDL-IA, or rIA produced by GenScript with a tag that was excised by enterokinase digestion. The antiviral units of these preparations were measured by cytopathic effect (CPE) assay using rIFNα from PBL (Piscataway, NJ) as standard. Recombinant human IA was expressed and purified by GenScript.

Quantitative Polymerase Chain Reaction (qPCR) Analysis.

Primers for quantitative real-time reverse-transcription polymerase chain reaction (RT-PCR) are listed in Supporting Information Table 1. Total RNA from mice livers was isolated and processed as indicated in the Supporting Information Methods.

Cloning of Murine Apolipoprotein A-I (mApoA-I), Murine Interferon Alpha 1 (mIFNα1), and Murine Albumin Complementary DNA (cDNA).

Gene Fusion. Primers and cloning procedures are given in Supporting Information Table 1 and the Supporting Information Methods. Gene fusion methodology is described in the Supporting Information Methods

Determination of mIFNα1.

mIFNα1 levels were measured by enzyme-linked immunosorbent assay (ELISA) as indicated in the Supporting Information Methods.

Isolation of HDLs,

Electrophoresis, and Immunoblotting Against mApoA-I. HDL isolation was performed by differential ultracentrifugation in sodium bromide gradient as described in the Supporting Information Methods. HDL+ or HDL− fraction samples were separated in 4%-20% TrisHEPES PAGE LongLife iGels (Nusep, Lane Cove, Australia) gradient gels, and transferred to a nitrocellulose membrane (Whatman, Kent, UK). mApoA-I was detected with goat polyclonal anti-apolipoprotein A1 (Santa Cruz Biotechnology, Santa Cruz, CA) and antigoat IgG (whole molecule) horseradish peroxidase (HRP)-conjugated (Sigma-Aldrich, St. Louis, MO) as a secondary antibody.

Determination of IFNα Activity and Cell Signaling.

IFNα bioactivity was determined by the ability of serial dilutions of the respective preparations to protect murine fibroblast L929 cells (3 × 104/microtiter well) against the cytopathic activity of the encephalomyocarditis virus (EMCV) as described in the Supporting Information Methods. For cell signaling experiments a total of 1.5 × 105 L929 cells were seeded onto six-well plates. After 24 hours the cells were left untreated or treated with rIFNα1 (500 U/mL), HDL (10 μg/mL), or HDL-IA (10 μg/mL containing 500 U/mL of IFNα activity). After 90 minutes the cells were washed with phosphate-buffered saline (PBS) 1× and collected in 150 μL of protein loading buffer and the material processed for western blotting using diverse antibodies as indicated in the Supporting Information Methods.

Tests for L929 Viability and Cytotoxicity.

ViaLight Plus Kit and ToxiLight BioAssay Kit (both from Lonza, Rockland, ME) were used as indicated in the Supporting Information Methods.

Hemogram.

Blood samples were analyzed in a Z1 Coulter Particle Counter with the settings recommended for each case by the manufacturer (all materials and reagents from Beckman Coulter, Fullerton, CA) as indicated in the Supporting Information Methods.

Proliferation of Hematopoietic Progenitor Cells and Percentage of Megakaryocytes in Bone Marrow.

BALB/c mice received hydrodynamic injection with plasmid encoding apolipoprotein A-I (pApo), pIFN, or pIA; 56 hours later, bromodeoxyuridine (BrdU) 7.2 mg/kg (Sigma) was administered to mice by intraperitoneal injection and 16 hours later mice were killed and bone marrow cells were isolated and stained for Lin (FITC− antimouse lineage antibody cocktail), c-Kit+ (PE antimouse CD117, both from Biolegend), and BrdU (BrdU Flow kit APC, BD Biosciences, San Jose, CA). To estimate the percent of megakaryocyte, CD41+ (PE antimouse CD41 from BD) cells were quantitated.

Flow Cytometry Analysis.

This was performed using FACSCalibur flow cytometer (Beckman Coulter) as described in the Supporting Information Methods.

In Vivo Killing Assays and Vaccination Against Tumor Peptide AH-1.

A detailed description of these procedures is given in the Supporting Information Methods.

Statistical Analysis of the Data. Statistical analysis of the data was performed using the Prism 5 computer program (GraphPad Software, San Diego, CA).

The tumor appearance data were represented in Kaplan-Meier graphs and analyzed by the log-rank test. Pharmacokinetics data were fitted to a one-phase decay for hydrodynamic injection using nonlinear regression analysis and compared by the extra sum-of-squares F test. The data studied at different times were analyzed by repeated-measures analysis of variance (ANOVA) followed by the Bonferroni test. The remaining parameters were analyzed by ANOVA and followed by Bonferroni's post-hoc analysis for carrying out multiple comparisons. P < 0.05 values were considered significant.

Results

Kinetics and Liver Targeting of IA.

After hydrodynamic injection of plasmids encoding IA (pIA), IFNα (pIFN) or albumin bound to IFNα (pALF), we found that plasma half-life of IA was 2.6-fold that of IFNα (Supporting Information Fig. 1A,B). Hepatic levels of transgenic mRNA and its kinetics were similar after administering pIA or pIFN, indicating that the extended presence of IA in plasma is due to increased stability of the protein (Supporting Information Fig. 2). On the other hand, the half-life of IA was only 1.6 times less than that of ALF, the formulation with the longest half-life reported to date for an IFNα conjugate10 (Supporting Information Fig. 1C). Pharmacokinetic analysis after the administration of rIA confirmed these data, as recombinant IFNα (rIFNα) presented a sharp decay in mouse plasma levels while the concentration of rIA decreased slowly (Supporting Information Fig. 1D).

Interestingly, we found that after hydrodynamic administration of pIA, all circulating IA produced by the liver was incorporated into HDL particles (Supporting Information Fig. 3A,B) and that, as a consequence, the HDL fraction of plasma displayed antiviral activity. In contrast, in mice treated with pIFN, antiviral activity was only found in lipoprotein-depleted serum (Supporting Information Fig. 3C). After intravenous injection of rIA, only a minor fraction (10%) of this protein was detected in isolated HDLs (Supporting Information Fig. 3D,E)

Native ApoA-I has strong liver tropism.16 Thus, we reasoned that linkage of IFNα to ApoA-I might result in targeting IFNα to the liver. To test this hypothesis, we analyzed the distribution of IFNα by ELISA in different organs (liver, brain, lung, heart, kidney, and spleen) at 5 and 150 minutes following intravenous (IV) administration of 1.6 μg of rIA or rIFNα. At 5 minutes, IFNα immunoreactivity was mostly detected in kidney and spleen, whereas IA was predominantly accumulated in the liver at 150 minutes postinjection. In the case of rIFNα, the cytokine was barely detectable at this timepoint in all organs examined (Fig. 1A,B).

Figure 1.

Hepatic expression of the fusion protein IA induces ISGs in the liver more efficiently than the expression of IFNα or the fusion protein albumin-IFNα due to liver targeting. (A,B) IFNα concentrations measured by ELISA in selected organs (liver, brain, lung, heart, kidney, and spleen) 5 or 150 minutes after IV injection of 1.6 μg of rIFNα or rIA (ND: not detectable). (C) Expression of (2′, 5′ OAS), USP18, and ISG15 in the liver 24 hours after the IV injection of PBS, rIFNα (10,000 U/mouse), or HDL containing IA (HDL-IA, 10,000 U/mouse) or (D) rIFNα or rIA (70000 U/mouse) or (E) three days after hydrodynamic injection of plasmids encoding ApoA-I (pApo), IFNα (pIFN), or IA (pIA); or (F) 3 days after hydrodynamic injection of plasmids encoding IA (pIA), or albumin-IFNα (pALF) or empty plasmid (pMock). Data are expressed as mean ± standard error of the mean (SEM) of two independent experiments (n = 4-6 mice per group). *P < 0.05; **P < 0.01; ***P < 0.001 (one-way ANOVA followed by the Bonferroni Multiple Comparison test).

We then quantitated hepatic interferon-stimulated genes (ISGs) messenger RNA (mRNA) levels 24 hours after IV injection of 10,000 U of rIFNα or the same antiviral units of purified HDLs containing IA (HDL-IA) or 24 hours after administration of 70,000 U of rIFN or rIA. ISGs activation was significantly greater when using HDL-IA (Fig. 1C) or rIA (Fig. 1D), suggesting preferential signaling to the liver when IFNα was linked to ApoA-I. Confirming these data, hepatic expression of ISGs at day 3 following injection of pIA or pIFN was higher with the former treatment (Fig. 1E). We also found that at day 3 after hydrodynamic injection of pIA or pALF, the expression of ISGs in the liver tended to be higher following pIA administration (for ubiquitin-specific peptidase 18 [USP18] differences reached statistical significance) (Fig. 1F) despite the fact that the serum concentration of IA was half that of ALF at this timepoint (Supporting Information Fig. 1C).

Antiviral Activity and Cytotoxic Effects of IFNα and IA.

Studies using L929 mouse fibroblasts incubated with rIFNα or the same antiviral units of HDL-IA or of rIA showed that the phosphorylation of STAT-1 and -2 was similar in both cases (Fig. 2A). However, the administration of 70,000 IU of rIA was able to protect 50% of the mice against a lethal challenge with EMCV, whereas 100% of mice treated with the same antiviral units of rIFNα succumbed (Fig. 2B). Therefore, the enhanced stability of IA in circulation allows a greater in vivo antiviral activity.

Figure 2.

HDL containing IA induces typical IFNα signaling, displays antiviral activity in vivo, but does not cause cell toxicity. (A) Mouse fibroblast L929 cells were incubated for 90 minutes at 37°C with mock, rIFNα (500 U/mL), HDL (10 μg/mL), HDL with incorporated IA (10 μg/mL of HDL containing 500 U/mL of IFNα activity), or rIA (500 U/mL), and signaling pathways activated by IFNα were analyzed by western blot against phosphorylated and total protein levels of STAT1, and STAT2, actin levels are shown as loading control. (B) In vivo antiviral activity. A lethal dose of EMCV was intraperitoneally injected 3 hours following the intravenous injection of either PBS, 70,000 U rIFNα or rIA (n = 7 mice per treatment) (log-rank [Mantel Cox] test *P < 0.05, **P < 0.01). (C) Cell viability assay and (D) cytotoxicity assay in mouse L929 cells incubated for 3 days in the presence of PBS, HDLs (10 μg/mL), rIFNα (500 U/mL), HDL-IA (10 μg/mL of HDLs containing 500 U/mL of IFNα activity), or rIFNα (500 U/mL) plus native HDLs (10 μg/mL), or rIA (500 U/mL). Cell proliferation was quantified by ATP bioluminescence measurement and the cytotoxicity by the release of adenylate kinase determined by bioluminescence (RLUs, relative luminescence units). Data are expressed as mean ± SEM of two independent experiments. ***P < 0.001 (one-way ANOVA followed by the Bonferroni Multiple Comparison test).

To determine whether recombinant human IA (rhIA) could display antiviral activity in a clinically relevant viral infection we performed in vitro antiviral assays of rhIA in Huh7 cells infected with a hepatitis C virus (HCV) full-length replicon. We found that rhIA vigorously inhibited HCV replication and HCV core protein expression (Supporting Information Fig. 4).

An important difference in the biological effects of IFNα and IA emerged when cell viability and cytotoxicity were analyzed in L929 cells exposed to either IFNα or rIA or HDL-IA. We found that, whereas IFNα (at the dose used for signaling experiments) caused an increase in cell death, cells treated with the same antiviral units of HDL-IA or rIA behaved like untreated control cells (Fig. 2C,D).

Differential Effects of IA and IFNα on Hematological Toxicity.

Lack of Toxicity in Mice with Long-Term Exposure to IA. In keeping with the above findings, we observed that IFNα and IA were not comparable with regard to their effects on the hematopoietic system. Three days after plasmid injection, platelets and leukocytes were thus significantly higher in pIA-treated mice than in pIFN- or pALF-treated mice (Fig. 3A,B). Although the white blood cell (WBC) count decreased the first day after therapy with pIFN or pIA (possibly involving shifts between circulating and marginal pools17), the leukocyte number returned to normal at day 3 in mice given IA, but not in those that received IFNα.

Figure 3.

Mice treated with IA do not manifest hematological toxicity. (A) Peripheral blood leukocyte (WBC) and platelet (PLT) counts 1, 3, and 6 days after hydrodynamic injection of plasmids encoding ApoA-I (pApo), IFNα (pIFN), or IA (pIA); (B) 3 days after hydrodynamic injection of plasmids encoding albumin-IFNα (pALF), IA (pIA), or empty plasmid (pMock); (C) proliferation of hematopoietic progenitor cells and percentage of CD41+ megakaryocytes in bone marrow. BALB/c mice received a hydrodynamic injection of pApo, pIFN, and pIA. Three days later, bone marrow cells were isolated and proliferation of hematopoietic progenitor cells (LincKit+) was determined by BrdU incorporation and percentage of CD41+ megakaryocytes was assessed by flow cytometry. (D) WBC and PLT counts 3 days after IV injection of PBS, rIFNα (10,000 U/mouse), or HDL containing IA (HDL-IA) (10,000 U/mouse) and (E) number of T cells (CD3+) and of myeloid cells (CD11b+) analyzed by flow cytometry. Data are expressed as mean ± SEM; n = 3-4 mice per group (two-way ANOVA (A) or one-way ANOVA (B-E), followed by the Bonferroni Multiple Comparison test. *P < 0.05. **P < 0.01. ***P < 0.001).

To further characterize the different impact of IFNα and IA on hematopoiesis, we analyzed the number of proliferating bone marrow hematopoietic precursor cells (Lin c-Kit+) and the percentage of megakaryocytes in the bone marrow in mice subjected to these treatments. In both cases the administration of plasmids encoding IFNα or IA induced a significant elevation in the number of BrdU-positive hematopoietic precursors, and in the percentage of megakaryocytes, but these increases were significantly higher in the group treated with IFNα (Fig. 3C). This cytokine has been shown to activate bone marrow hematopoietic precursor cells18 and, in addition, it may elevate megakaryocyte counts in bone marrow in response to thrombocytopenia. IA also increases the number of megakaryocytes in bone marrow but this occurs in the absence of significant thrombocytopenia. This might suggest a direct stimulation of the hematopoietic precursors in mice treated with pIA. In agreement with this notion, we observed that the administration of a low dose of IFNα (10,000 U) or the same antiviral dose of HDL-IA had a different influence on blood cells. While, at a low dose, IFNα did not cause changes in blood cell counts, the same HDL-IA dose induced a marked rise in leukocytes (neutrophils, lymphocytes, and monocytes) and platelets, reaching numbers significantly above normal values (Fig. 3D,E).

To assess the safety of long-term exposure to IA we transduced the liver of C57BL/6 mice with 2.5 × 1010 gc/mouse of AAV 2/8 vector encoding IA (AAV-IA) or luciferase (AAV-Luc) and we examined the animals 30 days after vector administration. No apparent toxicities and no relevant changes in blood counts, serum biochemistry, or liver histology were observed in animals treated with AAV-IA despite the fact that serum IFNα was present at high levels at the time of sacrifice (Supporting Information Fig.5 and Supporting Information Table 2).

Immunostimulatory Properties of IA.

Six days after injecting pIFN, pIA, or pApo (a plasmid encoding apolipoprotein A-I used as a control), we analyzed the number and activation status (as estimated by the percentage of CD69+ cells) of immunocytes in the spleen. The percentage of dendritic cells, macrophages, B cells, and natural killer (NK) cells positive for CD69 were similar in mice treated with pIFN and pIA (Supporting Information Fig. 5). However, administration of pIA caused a greater increase in the total number of splenocytes and in the percentage of CD8+ and CD4+ T cells expressing CD69 than when injecting pIFN (Fig. 4A). In vivo killing assays against a BALB/c immunodominant β-galactosidase epitope were performed in mice which, 7 days previously, received a hydrodynamic coinjection of a plasmid encoding LacZ together with pIFN or pIA or pApo. These studies showed that the group treated with pIA exhibited a cytolytic activity significantly greater than the other two groups (Fig. 4A). We also observed differences between pIFN and pIA in the induction of protective immunity in a murine model of vaccination against CT-26 colon cancer. Vaccination was performed by subcutaneous administration of the AH-1 peptide (which corresponds to the tumor immunodominant epitope) 24 hours after hydrodynamic injection of either pApo, pIFN, or pIA. Ten days later, the animals received a subcutaneous injection of 5 × 106 CT-26 cells in the right flank. We observed that adjuvant treatment with pIA, but not with pIFN, was associated with significant protection against tumor growth as compared to the control group given pApo (Fig. 4B).

Figure 4.

Enhanced immunostimulatory activity of the fusion protein IA. (A) BALB/c mice received a hydrodynamic injection of plasmids encoding ApoA-I (pApo), IFNα (pIFN), or IA (pIA). Six days later, the number of splenocytes was analyzed (n = 4 mice per group), and the expression of the early activation marker CD69 in CD4+ T cells and CD8+ T cells in splenocytes (n = 12 mice per group). In vivo killing assay against a BALB/c immunodominant β-galactosidase cytotoxic epitope 7 days after a hydrodynamic coinjection introducing a β-galactosidase coding plasmid and either pApo, pIFN, or pIA (n = 13 mice per group) (one-way ANOVA followed by the Bonferroni Multiple Comparison test. **P < 0.01. ***P < 0.001). (B) Prevention of CT-26 colon carcinoma appearance in mice receiving a hydrodynamic injection of pApo, pIFN, or pIA and 24 hours later vaccinated with an immunodominant CT-26 epitope (50 μg/mouse of AH-1 peptide) (n = 13 mice per group). Data were compared by a log-rank test. **P < 0.01. (C) Number of splenocytes, T-cell activation, and in vivo killing assay after hydrodynamic injection of plasmids encoding albumin-IFNα (pALF), IA (pIA), or empty plasmid (pMock) (n = 4-5 mice per group). (D) In vivo killing assay against a C57Bl/6 immunodominant β-galactosidase cytotoxic epitope in wildtype (SR-BI+/+), heterozygous (SR-BI+/−), and null SR-BI (SR-BI−/−) mice (n = 3-7 mice per group). The fold change ratio of specific lysis between pIA and pIFN in each genotype is represented. (E) In vivo killing assay against a C57BL/6 immunodominant ovalbumin cytotoxic epitope 7 days after IV coinjection of 100 μg of ovalbumin and PBS or 70,000 IU of rIFN or rIA (n = 4 mice per group). Data are expressed as mean ± SEM (one-way ANOVA followed by the Bonferroni Multiple Comparison test, *P < 0.05, **P < 0.01, ***P < 0.001).

In additional experiments, the immunological changes induced by pIA and pALF were compared. The hydrodynamic administration of pIA or pALF caused a similar rise in the number of splenocytes and in the percentage of splenic CD69+CD4+ and CD69+CD8+ cells (Fig. 4C; Supporting Information Fig. 6). Because pIA overrides pIFN but is equal to pALF in increasing the number and activation of splenocytes, it seems possible that these phenomena might be related to the prolonged persistence of both IA and albumin-IFNα in the circulation. However, IA largely surpassed albumin-IFNα in its ability to stimulate cytotoxic T cell responses, as demonstrated by in vivo killing assay against the immunodominant β-galactosidase epitope (Fig. 4C).

As SR-BI is a potential receptor for the ApoA-I moiety of IA, we wished to investigate whether the interaction of this molecule with SR-BI could mediate the potent immunostimulatory effects exhibited by IA. With this in mind, we analyzed in vivo killing against β-galactosidase epitope following vaccination with hydrodynamic coinjection of a plasmid encoding LacZ together with pIFN or pIA or pApo in wildtype mice and in SR-BI+/− or SR-BI−/− animals. In agreement with our previous data, pIA showed an 8-fold increase in the specific lysis compared to that induced by pIFN in wildtype mice. However, this difference was reduced to 2-fold in SR-BI+/− animals and disappeared in SR-BI null mice. Thus, IA requires interaction with SR-BI to display maximal adjuvant activity (Fig. 4D).

We also analyzed whether rIA displays enhanced adjuvant activity. To this end, we administered intravenously recombinant ovalbumin as an antigen together with a single administration of 70,000 IU of rIFNα or rIA. Although the former treatment was unable to increase the in vivo killing, rIA significantly enhanced CTL-mediated lysis (Fig. 4E).

Effects of IA and IFNα on Activation-Dependent T-Cell Death.

IFNα has been shown to induce activation-dependent cell death in lymphocytes.19, 20 Because our data suggested that IA might be more effective than IFNα in promoting the expansion of stimulated T cells, we analyzed lymphocyte proliferation and viability following T-cell stimulation with α-CD3 and α-CD28 in the presence of IFNα or the same antiviral units of HDL-IA. Flow cytometry analysis of stimulated lymphocytes showed that the number of large blast cells was markedly reduced in the presence of IFNα, whereas values were similar to controls when HDL-IA was added to the culture (Fig. 5A). In these experiments, we assessed lymphocyte proliferation by carboxyfluorescein diacetate succinimidyl ester (CFSE dilution) (Fig. 5B) and lymphocyte death by 7-AAD incorporation (Fig. 5C). We found that both cell proliferation and cell death were similar in control wells and in wells containing HDL-IA. In contrast, in the presence of IFNα, lymphocyte proliferation was reduced and the number of nonviable lymphocytes was greatly increased. Similarly, we observed that the proliferation and viability of activated lymphocytes was higher when rIA was present in the medium than in the presence of IFNα (Fig. 5A-C). These data are consistent with the notion that preservation of viability of activated T cells may underlie the higher effectiveness of IA in boosting T-cell-mediated immunity.

Figure 5.

HDL containing the fusion protein IA or rIA preserve the viability of activated T cells. (A) Representative forward scatter versus side scatter plot and (B,C) representative histogram of in vitro stimulated splenocytes CD4+ (right) or CD8+ (left) proliferation assessed by CFSE dilution (B) and cell death analyzed by 7-AAD incorporation (C) after 3 days treatment in a α-CD28 and α-CD3 coated plate with IL-2 (75 U/mL) and either PBS (red), rIFNα (500 U/mL) (blue), HDL containing IA (HDL-IA) (500 U/mL) (green), or rIA (500 U/mL) (orange).

Fusion of IFNα to Other Apoliproteins Present in HDLs (ApoE and F).

We studied whether fusion of IFNα to other apolipoproteins found in HDLs such as ApoE or ApoF could confer IFNα properties similar to IA. We prepared plasmids encoding these fusion proteins which were administered to mice by hydrodynamic injection. IFN-ApoF was unstable and was not expressed (data not shown). IFN-ApoE was expressed and, similar to IA, manifested liver targeting and little hematological toxicity, but at variance with IA exhibited no differences with respect to IFNα in half-life or immunostimulatory properties (Supporting Information Fig. 8).

Discussion

In order to increase the half-life and to target IFNα to the liver, we fused IFNα to ApoA-I. We used this strategy because the half-life of ApoA-I is 2 to 3-fold that of IFNα, being comparable to PEG-IFNα.12, 14, 15, 21 On the other hand, it has been demonstrated that after its systemic administration ApoA-I accumulates in the liver.16 Indeed, we found that IA had a longer half-life than IFNα and exhibited hepatic tropism. Liver targeting was ascertained by pharmacokinetic and biodistribution studies and also by the more intense activation of hepatic ISGs when using pIA for liver transduction than when employing pIFN or pALF, despite the fact that albumin-IFNα has a longer half-life than IA.

Unexpectedly, fusion of IFNα to ApoA-I markedly changed the biological properties of IFNα. Thus, experiments in murine fibroblasts showed that the cytotoxic effects of either HDL-IA or rIA were significantly lower than those of IFNα. Moreover, although IFNα induces activation-dependent cell death in T lymphocytes,19, 20 this effect is much lower with IA. Thus, IA facilitates lymphocyte proliferation in response to mitogens. These in vitro results are in accordance with data from in vivo killing assays where we found that the adjuvant efficacy of IA was far superior to that of IFNα. These findings indicate that the immunostimulatory properties of IA are not merely related to its persistence in circulation but rather to the fact that IA does not induce cell death in activated lymphocytes as does IFNα.

In order to explore whether the interaction of IA with SR-BI (the main ApoA-I receptor11) is required for its enhanced adjuvant activity, we performed in vivo killing assays after LacZ vaccination using as adjuvants pIA or pIFN in SR-BI+/−, SR-BI−/−, and wildtype mice. Our data showing that the adjuvant effect of IA was superior in wildtype animals but not in SR-BI−/− animals indicates that ligation of IA to SR-BI is needed for the fusion protein to display its potent immunostimulatory properties.

The lower cytotoxic activity of IA compared to IFNα is also reflected by their different influence on hematopoiesis. One of the side effects of IFNα treatment is the development of leukopenia and thrombocytopenia, which limits its use in patients who already have diminished blood counts.8 In contrast to IFNα, the administration of IA did not significantly reduce the number of platelets and only caused a transient drop in leukocytes, which rapidly recovered. Moreover, the bone marrow progenitor compartment was induced to proliferate in the absence of significant modifications in the number of circulating leukocytes and platelets, suggesting that IA might stimulate myelo and thrombopoiesis. The safety of IA was also demonstrated in mice whose liver was transduced with an AAV vector encoding IA. No hematological or biochemical toxicity was found in these animals at day 30 after vector administration.

The improvement of the pharmacological profile of IFNα resulting from its linkage to ApoA-I was not reproduced by fusion to other apolipoproteins present in HDLs. Anchoring IFNα to ApoF generated an unstable molecule which was not expressed. On the other hand, the fusion protein IFNα-ApoE exhibited liver targeting and reduced cytopenic effects, but its half-life and immunostimulatory properties were similar to those of IFNα.

In conclusion, liver targeting, prolonged half-life, enhanced immunostimulatory functions, and reduced hematological toxicity are properties of IA which make this molecule a promising therapy for patients with viral and/or neoplastic diseases affecting the liver.

Acknowledgements

We thank C. Gomar, I. Echeverría, N. Casares, J. Lasarte, and P. Sarobe for advice and technical support.

Ancillary

Advertisement