Bioconjugate Chem. 1992, 3, 402-407
402
Interferon Production of L929 and HeLa Cells Enhanced by Polyriboinosinic Acid-Polyribocytidylic Acid pH-Sensitive Liposomes Pierre G. Milhaud,+BBatrice Compagnon,* Alain Bienvenue,* and Jean R. Philippot*,* Universit6 Montpellier 11, Sciences et Techniques du Languedoc, DBpartement Biologie SantB, URA-CNRS 1191 Gbnbtique Molbculaire, and URA-CNRS 530 Interactions Membranaires, 34095 Montpellier Cedex 5, France. Received May 8, 1992
The double-stranded RNA polyinosinic acid-polycytidylic acid (PolyIC) is an inducer of interferons cy and p (IFN) genes. With L929 and HeLacells IFN pretreatment (priming) improves the IFN induction by PolyIC by several orders of magnitude. In the absence of the priming we demonstrate that PolyIC encapsulated into pH-sensitive liposomes (and not into pH-insensitive liposomes) enables L929 cells to secrete IFN efficiently and alow toxicity is observed; on primed cells pH-sensitive liposomes containing PolyIC trigger a high toxicity. With HeLa cells, the absence of the priming PolyIC encapsulated into pH-sensitive liposomes induces weak doses of IFN whereas free PolyIC was ineffective. Our experiments established that a pH drop (from 8 to 5.5) provoked a lipid mixing between pH-sensitive liposomes and cell membranes, likely by a fusion mechanism. Entrapment into pH-sensitive liposomes enhances the effect of PolyIC by several orders of magnitude, which might improve its therapeutic ability as an antitumor or anti-HIV agent.
INTRODUCTION
The double-stranded RNA polyriboinosinic acid-polyribocytidylicacid (PolyIC) is an inducer of interferon (IFN) cy and /3 genes (Whatelet et al., 1987). Pretreating the cells with IFN (priming) (Stewart et al., 1971) enhances the response of L929 or HeLa cells to PolyIC powerfully (De Clercq, 1981). On the other hand PolyIC induces a powerful toxicity against IFN-primed L929 cells (Stewart et al., 1972) the mechanism of which is not yet fully understood. However we have demonstrated that free PolyIC enters the cell through the acidic endocytic compartment before triggering that toxicity (Milhaud et al., 1987). Recently we demonstrated the induction of IFN by PolyIC loaded into liposomes which were targeted by antibodies to primed L929 cells (Milhaud et al., 1989). Surprisingly the IFNPolyIC toxicity which goes with the endocytosis of PolyIC is highly increased by liposome-entrapped PolyIC. These results suggest that either a higher amount of intracellular PolyIC and/or a more appropriate intracellular delivery potentiate PolyIC efficacy. Transfer of genetic material or synthetic nucleic acids interfering with gene expression (Leonetti et al., 1988) needs vectors to cross natural membranes, to escape degradative enzymes, and to reach their intracellular relevant targets. In several families of RNA viruses the delivery of the viral genome into the host cell cytoplasm is pHdependent (Marsh & Helenius, 1989). At low pH, as encountered in the endocytic compartment, a viral protein triggers the fusion of the viral and endosomal membranes, thus allowing delivery of the viral genome into the cytoplasm. pH-sensitive liposomes have been devised which simulate viral behavior although the mechanisms are different (Connor et al., 1984; Ellens et al., 1984; Nayar & Schroit, 1985). Dioleoylphosphatidylethanolamine
* Corresponding author: Dr. J. R. Philippot, Universit6 Montpellier 11, DBpartement Biologie Sant4, URA-CNRS 530 Interactions Membranaires, Case 107, Place E. Bataillon, 34095 Montpellier Cedex 5,France. Phone: (33)67143741.Fax: (33)67144286. t URA-CNRS 1191 GBn6tique MolBculaire. URA-CNRS 530 Interactions Membranaires.
*
(DOPE) and oleic acid (OA)liposomes are stable at neutral pH. The tendency to revert the HI1 phase occurs under conditions of acidic pH in which the OA component becomes protonated (Liu & Huang, 1989a). The lipid bilayer of such liposomes is destabilized at low pH and the loaded material enters the cytoplasm by a still unknown mechanism. Whatever the vector, virus, or liposome, the material to be delivered enters the cell by the endocytic pathway and escapes lysosomal degradation at least partially. The efficiency of pH-sensitive liposomes has been demonstrated in vitro with the help of encapsulated drugs such as antitumor drugs (Collins et al., 1988; Connor & Huang, 1986) or entrapped toxin (Collins & Huang, 1987; Chu et al., 1990) or fluorescent markers of aqueous compartment (Connor & Huang, 1985) or membrane lipids (Duzgunes et al., 1987). Efficient gene transfers have been performed in vivo (Wang & Huang, 198713;Nayar & Schroit, 1989) and in vitro (Wang & Huang, 1987a; Wang & Huang, 1989). These previous observations prompted us to test the efficiency of pH-sensitive liposomes for PolyIC delivery. In the present study we show that PolyIC entrapped into pH-sensitive liposomes induced the secretion of IFN from L929 cells in the absence of priming and only triggered a weak toxicity. However in the absence or in the presence of priming, PolyIC entrapped into pH-sensitive liposomes only induced a weak antiviral activity in HeLa cells. EXPERIMENTAL PROCEDURES
Cell Lines and Viruses. Murine L929 cells (American Type Culture Collection, Rockville, MD; Ref. CCL 1)were grown in MEM medium (Gibco, Cergy Pontoise, France) supplemented with 5% (v/v) fetal bovine serum (FBS) and antibiotics. HeLa cells were obtained from G. Huez (Universitb libre de Bruxelles, Belgium); they were grown in RPMI 1640 medium enriched with glutamine and supplemented with 5% (v/v) FBS and antibiotics. The human lymphoblastic T CEM cells were cultured in RPMI 1640 supplemented with 10% FBS (v/v) and antibiotics. Vesicular stomatitis virus (VSV) and encephalomyocarditis virus (EMCV) were grown and titrated on L929 cells. Antibodies. Polyclonal rabbit antibodies to murine 0 1992 American Chemical Society
High Efficiency of PolyIC pH-Sensitive Llposomes
IFN-a-j3 were obtained from Lee Biomolecular Research (San Diego, CAI. The monoclonal murine antibodies (mAbs) to human IFN-a and to human IFN-j3 were obtained from Alpha Therapeutic Corp. (LosAngeles, CAI. The murine mAbs UM21.1 (IgG,2b), a generous gift from Dr. Kraaijeveld and Dr. Snippe (State University of Utrecht, the Netherlands), are neutralizing antibodies to EMCV (Vlaspolder et al., 1989). The goat peroxidase conjugated F(ab')z antibodies to whole mouse immunoglobulin G molecules were obtained from Immunotech (Marseille, France). Interferon and Double-Stranded RNA. Murine IFN-a-fl (specific activity 6 X lo4 units/mg), a generous gift from Dr. G. Rossi (Instituto Superiore di Sanita, Roma), was diluted to 4 X lo5 units/mL in phosphatebuffered saline (PBS: 137 mM NaC1, 2.7 mM KC1, 1.5 mM KHzP04, 8.1 mM NazHPO4, pH 7.4) containing 2 mg/mL bovine serum albumin and then aliquoted and frozen at -70 "C until use. Recombinant human interferon a2a (HuIFN-a2a) was similarly treated and stored. A high molecular weight complex of PolyIC (Pharmacia, Uppsala, Sweden) (2 mg/mL) was sonicated (Branson, Danbury, CT, 20 W, 10 X 30 s) until a mean length of 500-600 base pairs was obtained, as verified by 2% (w/v) agarose gel electrophoresis (buffer, 89 mM TrisHC1, 89 mM borate, 1 mM EDTA (pH 7.5), 0.5 pg/mL ethidium bromide). PolyIC was labeled by [ T - ~ ~ P I A using T P polynucleotide kinase, according to standard protocols, to a final specific activity of 5 X lo5 Bqlpmol. Preparation of pH-Sensitive Liposomes. In routine experimenta 20 pmol of dioleoylphosphatidylethanolamine (DOPE), cholesterol (Chol), and oleic acid (OA) in the molar ratio 2:2:1 were dried from solvents to a thin film under a stream of nitrogen. Materials to be encapsulated (0.7-1.4 mg of PolyIC) were introduced in a high ionic strength 150mM 4-(2-hydroxyethyl)-l-piperazineethanesulfonic acid (HEPES), 1mM ethylenebis(oxyethy1enenitri1o)tetraacetic acid (EGTA) and 150 mM NaC1, pH 8.01 and lipids were hydrated for 30 min at 37 "C. A 100-pmol portion of octyl glucoside in the corresponding buffer was added and the volume of the samples was adjusted to 0.50.65 mL with the buffer. After vigorous shaking the detergent was removed by dialysis against 50 mL of buffer containing 2 g of Amberlite XAD2 (Philippot et al., 1983). Liposome preparations were treated for 30 min at 37 "C with RNase A (EC 3.1.4.22) (25-100 pg/mL) to digest unencapsulated polynucleotides. Liposomes were collected on a 5%-20% (w/v) dextran (average M , = 162 000) gradient (Sigma) at 20 "C (45 min at 40 000 rpm). The yield of PolyIC encapsulated and the diameter of liposomes were measured for each sample. On a regular basis 2-4 pg of PolyIC/pmol of lipid was encapsulated, and the diameters were around 450-650 nm. The stability of the pH-sensitive liposomes was assessed as follows. pH-sensitive liposomes containing 32P-labeled PolyIC were prepared and separated from nonincorporated PolyIC. The day following their preparation the liposomes were incubated in different media for 30 min and then loaded on a Biogel 5-m column (Bio-Rad) equilibrated with the buffer used for their preparation. The leakage was expressed as the ratio of the radioactivity trailing as a second peak to the total radioactivity eluted from the gel. The liposomes were incubated in the preparation buffer or in MEM without serum at 37 "C: the leakage was 23 74 and 44 % ,respectively (mean of two determinations, standard deviation = f0.6) in the first experiment; in the second independent experiment 38 % and 63 5% were obtained. These experiments underscored the destabilizing effect of MEM on pH-sensitive liposomes.
Bloconlugate Chem., Voi. 3, No. 5, 1992 403
As a rule we used the liposomes the day following their preparation. Fluorescent pH-sensitive liposomes were prepared by replacing a few percent of DOPE by (4-nitrobenz-2-oxa1,3-diazolyl)phosphatidylethanolanine(NBD-PE): DOPE/ NBD-PE/Chol/OA, 1.75:0.25:2:1 (M/M). Lipid Mixing Assays. Vesicle membrane fusion results in the communication between two aqueous compartments initially separated by the two fusing membranes: it involves intermixing of aqueous contents and intermixing of membrane components. The last event can be followed using a fluorescencemethod. The high concentration (55% ) of NBD-PE (excitation, 460 nm; emission, 534 nm) present on the labeled pH-sensitive liposomes self-quenches its fluorescence. The fusion between these liposomes and cells dilutes the probe, resulting in an increase of NBDP E fluorescence (Wilschut & Hoekstra, 1986). Kinetics were carried out at room temperature in the cell of a spectrofluorimeter. Nonattached cells were used in this assay: 2.5 X 107 CEM cells in 1mL (10 mM Tris, 150 mM NaC1) at pH 8 were mixed to 3 pL of fluorescent liposomes and then the pH of the medium was shifted to pH 5.5 with 10% HC1 (v/v) (zero time). At the plateau of the reaction, 10 pL of Triton XlOO (10%) was added and the fluorescence measured. Fluorescence changes were graphically estimated and expressed as percent of fluorescence variation between zero time and Triton XlOO addition. In control experiments buffer was used instead of acidic solution. Induction of IFN and of IFN-PolyIC Toxicity. IFN production was obtained by modification of published protocols (Milhaud et al., 1989). L929 cells were seeded at lo5 cells/mL per well in 24-well tissue culture dishes and exposed to 800 U/mL IFN for 8 h. This step is defined as "priming" (Stewart et al., 1971). The cell monolayers (two wells per treatment) were washed and then various concentrations of free or liposome-encapsulated polynucleotides were added in serum-free medium and the dishes incubated for a further 2 h a t 37 "C. Cell monolayers were washed with MEM supplemented with 5% (v/v) FBS and incubated a t 37 "C for a further 18-20 h. The supernatant fluids were transferred to wells containing lo5L929 fresh cells, incubated for 20 h, and then challenged with lo5 IU/mL VSV. The virus was titrated by end-point dilutions 24 h later. Cell supernatants testing the antiviral activities induced by free or encapsulated PolyIC were incubated with antibodies to murine or human IFN-a-/3(lo00neutralizing units) for 2 h; they were then incubated with L929 or HeLa cells and the antiviral activities were estimated by challenge with EMCV and the enzyme immunoassay as indicated below. The antiviral activities were completely neutralized, indicating that IFN-a and/or IFN-/3 were secreted by the cells treated with either free or liposome encapsulated PolyIC (data not shown). Estimation of the PolyIC Toxicity. The toxicity was estimated on the primed and PolyIC-treated cells. As soon as a supernatant was transferred, 1mL of diluted trypsin was added onto the cell monolayers, and the cells were counted with a Coulter counter (Coultronics). The results were expressed as the ratio of the number of the surviving cells after treatment with free or encapsulated PolyIC to the number of control cells (i.e. untreated or treated with empty liposomes, respectively). Virus Titration by End-Point Dilution. Dishes containing L929 cells and virus were frozen and thawed twice, and the VSV titers measured by end-point dilutions on L929 cells seeded in 96-well tissue culture dishes. The number of infectious units (IU/mL) was estimated by the maximum likelihood method (Milhaud et al., 1983). The
404
Mllhaud et al.
Bloconjugate Chem., Vol. 3, No. 5, 1992
99% confidence limits for the number M or IU/mL, obtained with a dilution factor 10 are M X 0.3 < M < M X 3.38. Virus Titration by Enzyme Immunoassay. Relevant cells (L929 or HeLa) in 24-well Linbro dishes were pretreated by serial dilutions of the supernatants containing the secreted IFN. A dose effect of murine IFNa-8 (for L929 cells) or HuIFN-a2a (for HeLa cells) was included in the assay. After a 20-h incubation the dishes were infected and reincubated for 5-7 h. The virus yield was estimated by a direct enzyme immunoassay of EMCV according to a modification of a previously published protocol (Vlaspolder et al., 1989). The immunoassay measured virus yields which in turn measured IFN concentrations. Briefly the cells were fixed by ethanol/ acetone (1/1) and washed with PBS. mAbs UM 21.1 were added, incubated at 37 "C for 1h, and then washed with PBS. Polyclonal antibody to whole mouse immunoglobulin, conjugated peroxydase, was added and incubated for 1 h at 37 "C and washed with PBS. The conjugated peroxidase was revealed with the help of 0.1 mM 2,2'azinobis(3-ethylbenzthiazoline-6-sulfonate) (Boehringer) or 4 mM 1,2-phenylenediamine (Fluka) as chromogen in the presence of 0.003% H 2 0 2 as substrate. The colored reaction was read at 410 or 492 nm, respectively. Comparison between the results at 0.5 absorbance allowed an accurate titration of IFN. RESULTS
Liposome Encapsulated PolyIC Resists RNase Treatment. Incubation of free PolyIC or that encapsulated into pH-sensitive liposomes with IFN-pretreated L929 cells induced a transferable antiviral activity to mouse cells. To ascertain that the secreted antiviral activity could be only ascribed to the encapsulated PolyIC, a RNase treatment (50-100 pg/mL) of the liposomes was carried out as indicated in the methods section. We checked for the efficacy of this step. PolyIC was encapsulated into pH-sensitive liposomes whereas pH-sensitive empty liposomes were incubated with labeled PolyIC; both liposome preparations were RNase treated as indicated in the methods. The induced antiviral capabilities of these liposomes were tested. PolyIC encapsulated into liposomes induced an antiviral activity while empty liposomes whose bound PolyIC had been chopped up by RNase were devoid of any activity. Independent experiments (n = 2) showed that pHsensitive liposomes, which had been pretreated with RNase as indicated above in the methods, were destroyed by a second harsh RNase treatment (50-100 pg/mL). Table I shows that the activity of the encapsulated PolyIC was still shielded from a second weak RNase treatment (2 pg/ mL). These experiments emphasized that pH-sensitive liposomes provided PolyIC with protection. PolyIC Encapsulated into pH-Sensitive Liposomes Induces IFN and a Weak Toxicity against Nonprimed L929 Cells. It has been reported that DEAE dextran is instrumental in enhancing the penetration of PolyIC into L929 cells (De Clercq, 1981) and triggers the induction of IFN in the absence of priming. Similarly pH-sensitive liposomes made it possible for the A chain of diphtheria toxin to penetrate and kill diphtheria-toxin-resistant L929 cells (Collins & Huang, 1987). We examined whether pHsensitive liposomes could induce the production of IFN in the absence of priming. As a preliminary experiment we titrated by ELISA the antiviral activity induced by free PolyIC from nonprimed and IFN-primed cells. As little as 0.02 pg/mL of free
Table I. RNase Resistance of PolyIC-Encapsulated Preparationse
liposome preparationb untreated RNase treated free PolyIC 100 0 lip. pHS-IC 99.4 98.5 lip. pHS-0 0 0 lip. pHNS-IC 97.6 93.8 lip. pHNS-0 0 0 Free and encapsulated PolyIC (2 pg/mL) were incubated in MEM in the absence or in the presence of RNase A (2 gg/mL) at 37 "C. This was a second RNase treatment in addition to the initial one performed during liposome preparation. After 1-h incubation mixtures were checked for antiviral activity: virus growth was titrated by endpoint dilution. The control viral growth was 1.3 X lo7 IU/mL and in the presence of PolyIC was 1.3 X lo4 IU/mL. The results are expressed as the percents of viral reductions (1.3 X lo7- Yi1.3 X l o 7 - 1.3 X lo4 IUlmL). The influence of RNase on virus growth was corrected accordingly. lip. pHS and lip. pHNS stand for pH-sensitive liposome and pH-insensitive liposome, respectively. -IC indicates that PolyIC was encapsulated, -0 indicates that the liposomes were empty. Table 11. Induction of IFN and Toxicity to L929 Cells of Free PolyIC and PolyIC Encapsulated into pH-Sensitive Liposomese values at PolyIC concn, pglmL
free
encapsulated
cell treatment
0.2
2
0.2
2
untreated cells
