A Cytoplasmic Structure Resembling Large Protein Aggregates Induced by Interferons

Abstract

IFP 35 is an interferon (IFN)-regulated leucine zipper protein, expression of which is observed in a variety of cell types including monocytes/macrophages, epithelial cells and fibroblasts. Using immunofluorescence studies, we demonstrate that IFP 35 is found in characteristic punctate cytoplasmic structures after IFN treatment. Co-localization experiments using double immunofluorescence and confocal laser scanning microscopy failed to show association of IFP 35 with known organelles (mitochondria, peroxisomes, endoplasmic reticulum, lysosomes, endosomes, Golgi complex), ribosomes, or actin filaments. Subcellular fractionation to separate membrane-associated from cytoplasmic proteins demonstrated that IFP 35 localizes to the cytoplasm. Separation of postnuclear supernatant from HeLa cells by gel filtration revealed that IFP 35 eluted at a molecular mass of 200-440 kD, suggesting that IFP 35 is part of protein complexes. Electron microscopic studies showed cytoplasmic clusters of a few aggregates of IFP 35 in IFN-treated cells which were neither associated with nor surrounded by a membrane. A combination of immunoprecipitation and immunofluorescence studies of cells transfected with a hemagglutinin epitopetagged IFP 35 expression construct demonstrated complex formation and co-localization of endogenous and transfected IFP 35. Taken together, our studies demonstrate that IFP 35 associates with unique cytoplasmic structures that are distinct from known organelles and resemble large protein aggregates.

INTERFERONS (IFNs) are a class of proteins with diverse biological functions. These pleiotropic effects include the ability to inhibit viral replication, to restrict growth of tumor cells in vitro and to regulate immunologically important functions of multiple cell types (Pestka et al. 1987; De Mayer and De Mayer-Guignard 1988). IFNs also affect intermediate filament assembly (Flohr et al. 1992; Jiang et al. 1994; Carroll et al. 1997) and cellular compartmentalization, e.g., by promoting phagolysosome fusion and acidification of vacuoles (Sibley et al. 1985, 1987; Kagaya et al. 1989). Interferons exert their effects mainly by altering gene expression (Boehm et al. 1997). Functions for a few of these IFN-induced proteins have been well characterized and contribute to our understanding of IFN-mediated effects. However, functions for the majority of these IFN-induced proteins are currently unidentified. The IFN-induced 35-kD protein termed IFP 35 belongs to this group of proteins (Bange et al. 1994). IFP 35 maps to chromosome 17q21 (Brown et al. 1995), forms homodimers in vitro, and is characterized by a leucine zipper motif at its amino terminus (Bange et al. 1994).

The leucine zipper motif consists of a periodic repetition of leucines or other hydrophobic residues forming a parallel, two-stranded, α-helical coil (O'Shea et al. 1991). Leucine zippers were first described in transcription factors able to form homo- or heterodimers (Landschulz et al. 1988; Jones 1990). Subsequently, it was found that leucine zipper motifs are also present in eukaryotic proteins distinct from transcription factors (White and Weber 1989; Brand and Castle 1993). At present, leucine zippers are considered to represent domains mediating protein-protein interactions, e.g., subunit interaction in voltage-gated K⁺ channels (Mc-Cormack et al. 1989), formation of tetramers as in the case of tyrosine hydroxylase (Vrana et al. 1994), and leucine zipper-dependent oligomerization, e.g., glucose transporter proteins (Pessino et al. 1991; Asano et al. 1992).

As an approach to understanding the function of IFP 35, we sought to characterize its subcellular localization. In the cytoplasm of eukaryotic cells, different compartments have been defined by morphological criteria. Prominent organelles are mitochondria, endoplasmic reticulum, Golgi apparatus, peroxisomes, endosomes, lysosomes, and various kinds of trafficking vesicles that are involved in endocytic and exocytic pathways. In addition to organelles, eukaryotic cells are characterized by an internal skeleton, the cytoskeleton, that gives the cell its shape, its capacity to move, and the ability to arrange and transport its organelles. Cytoplasmic proteins, which are not compartmentalized by association with vesicles, organelles, or the cytoskeleton, are in general distributed homogeneously within the cytoplasm. Notably, some rare examples of cytoplasmic proteins with a nondiffuse cytoplasmic distribution have been described, i.e., nonnuclear localization of the human p53 tumor suppressor protein appearing as punctate cytoplasmic structures (Ostermeyer et al. 1996), aggregates of ubiquitin and proteasomal antigens (Wojcik et al. 1996), and the human promyelocyte leukemia (PML) protein and its isoforms, which exhibit a variable expression pattern (Koken et al. 1994; Flenghi et al. 1995). With the exception of cytoplasmically localized p53 protein, little is known concerning the possible function and the structural basis of these unusual cytoplasmic structures. For p53 protein it has been suggested that altered cellular localization with cytoplasmic sequestration and concomitant nuclear exclusion is an alternate mechanism of inhibiting p53 function (Moll et al. 1992).

This report describes the results of a variety of techniques, including immunofluorescence microscopy, immunochemistry, subcellular fractionation, and electron microscopy, examining the intracellular distribution of IFP 35. Our results demonstrate that IFP 35 distributes in a unique IFN-induced punctate cytoplasmic pattern that is suggestive of large protein structures.

Materials and Methods

Cell Culture and Immunofluorescence Microscopy

HeLa S3 (Deutsche Sammlung von Mikroorganismen und Zellkulturen; Braunschweig, Germany), GM637 (Kucherlapati et al. 1978) (kindly provided by S.L. Gupta; Hipple Cancer Research Center, Dayton, OH), and WISH cells (Hayflick 1961) (kindly provided by P.v. Wussow; Medizinische Hochschule Hannover, Hannover, Germany), were cultured in Dulbecco's MEM containing 4500 mg/ml glucose (Life Technologies; Eggenstein, Germany). THP-1 cells (Tsuchiya et al. 1980; Auwerx 1991) (American Type Culture Collection; Rockville, MD) were grown in RPMI 1640 medium. Differentiation was induced by addition of 20 nM PMA (Sigma; Deisenhofen, Germany) for 24 hr before treatment with LFN-γ. All culture media contained 862 mg/liter L-alanyl-L-glutamine, 50 U/ml penicillin, 50 μg/ml streptomycin (Sigma), and 10% fetal calf serum (Life Technologies).

Cycloheximide (CHX, Sigma; 50 mg/ml stock solution in ethanol) was added to a final concentration of 50 μg/ml: this concentration inhibits protein synthesis in HeLa cells by greater than 95% as determined by [³⁵S]-cysteine incorporation. Nocodazole and cytochalasin D [stock solutions of 50 mM and 10 mM, respectively, prepared in dimethylsulfoxide (DMSO)] were diluted to the following final concentrations in culture medium: 50 μM, 25 μM, and 2.5 μM for nocodazole and 10 μM, 5 μM, and 2 μM for cytochalasin D. To control for unspecific effects of DMSO, cultures were treated with DMSO in the appropriate concentrations.

For immunofluorescence studies, cells were suspended at 1-4 × 10⁴ cells/ml in culture medium and added to clear glass coverslips (12-mm diameter; Assistent, Sondheim/Rhön, Germany) placed in 24-well tissue culture plates (1000 μl/well), THP-1 cells were incubated with 20 nM PMA for 24 hr before IFN induction. The following day, cells were treated with 1000 U/ml interferon-γ (Bioferon, 2 × 10⁷ U/mg protein; Rentschler, Laupheim, Germany) (kindly provided by P.v. Wussow).

According to the different antibodies used in the double immunofluorescence experiments, cells were fixed and permeabilized at room temperature (RT) as follows. For anti-mitochondrial antigen (BioGenex; San Ramon, CA), anti-KDEL peptide (kindly provided by J. Wehland; Gesellschaft fur Biotechnologische Forschung, Braunschweig, Germany), and anti-transferrin receptor antibodies (Oncogene; Uniondale, NY), cells were fixed for 10 min with 3% paraformaldehyde in PBS and permeabilized for 15 min with 0.2% Triton X-100. For the anti-S3a antibody (kindly provided by J. Stahl; Max-Delbrück-Centrum, Berlin, Germany), cells were fixed for 30 min with 3% paraformaldehyde and permeabilized for 15 min with 0.2% Triton X-100. For anti-cathepsin D (kindly provided by K. v. Figura; Universität Göttingen, Göttingen, Germany, and also obtained from Oncogene) and anti-mannose 6-phosphate receptor antibodies (kindly provided by K. v. Figura and by B. Hoflack, European Molecular Biology Laboratory, Heidelberg, Germany), cells were fixed for 45 min with 3% paraformaldehyde. For anti-catalase (Calbiochem; Bad Soden/Taunus, Germany), and anti-β COP antibodies (Sigma), cells were treated for 10 min with ice-cold methanol and 1 min with ice-cold acetone. For staining of IFP 35, two antibodies were used: a polyclonal rabbit antiserum (generated against an IFP 35 maltose binding fusion protein) (Bange et al. 1994), which can be used with cells fixed and permeabilized with each of the above-mentioned procedures and a monoclonal antibody (MAb) which can be used only on methanol/acetone-prepared cells. This MAb was generated by immunization of Balb/c mice with an IFP 35 fusion protein.

Fixed and permeabilized cells were rinsed three times with PBS and incubated with blocking buffer (0.1% saponin/10% FCS/PBS) for 10-30 min at RT, followed by incubation with the appropriate primary antibodies diluted in blocking buffer for 30-60 min at RT. After three washes with PBS and one wash with blocking buffer, cells were incubated with secondary antibodies [goat anti-rabbit, goat anti-mouse, donkey anti-rabbit conjugated to fluorescein isothiocyanate (FITC); goat anti-rabbit, goat anti-mouse, donkey anti-goat conjugated to tetramethylrhodamine isothiocyanate (TRITC); all obtained from Jackson ImmunoResearch, West Grove, PA]. Cells were washed three times with PBS, once with blocking buffer followed by three washes with PBS, and finally embedded using Mowiol (Calbiochem) in PBS, pH 7.4, containing 50 mg/ml 1,4-diazabicyclo[2.2.2.]octane (DABCO; Sigma) to stabilize fluorescence of FITC (for mannose 6-phospate receptor and cathepsin D, 0.5% saponin/PBS was used instead of 0.1% saponin/10% FCS/PBS throughout the procedure). Actin was visualized by staining with 5 μg/ml FITC-conjugated phalloidin (Sigma) for 60 min. Conventional epifluorescence was examined using a Zeiss Axioskop microscope equipped with a Plan-Neofluar X63 objective (Zeiss; Oberkochen, Germany) and appropriate filters. Laser scanning microscopy was performed with a BioRad MRC1024 system (BioRad; München, Germany) using an argon/krypton laser and an E2/UBHS filter set. Optical sections were obtained at 0.25-μm intervalls and the combined Z-series were printed on a Tektronix Phaser 440 (Tektronix; Köln, Germany).

Figure 1

Immunofluorescent localization of IFP 35. HeLa (A-C) and WISH (D) cells were plated on glass coverslips, grown for 24 hr, treated with 1000 U/ml IFN-γ for 24 hr (B-D) or left untreated (A). HeLa and WISH cells fixed with methanol/acetone were labeled with anti-IFP 35 MAb and developed with fluorescein-conjugated goat anti-mouse (A,B,D). For staining with the polyclonal anti-IFP 35 antiserum (C), HeLa cells were fixed for 10 min with 3% paraformaldehyde and permeabilized for 15 min with 0.2% Triton X-100 (developed with fluorescein-conjugated goat anti-rabbit). The polyclonal antibody (C) shows significantly higher nonspecific background staining compared to the MAb (B); preadsorption of the polyclonal antibody with antigen did not measurably influence the staining of the untreated control cells but completely eliminated the IFN-induced staining pattern (not shown). Cells were examined with a X63 Plan-Neofluar oil immersion objective on a Zeiss Axioscope microscope equipped with the appropriate filters.

Electron Microscopy

HeLa cells were seeded on 6-cm plastic dishes at a density of 5 × 10⁵ cells/dish and induced the following day with 1000 U/ml IFN-γ for 24 hr. Cells were then washed three times with PBS. Cells were fixed directly on Petri dishes with 0.2% glutaraldehyde and 0.5% formaldehyde (final concentration) in PBS for 1 hr on ice. After three washing steps using PBS containing 0.01% glycine for quenching of free aldehyde groups, cells were scraped from the Petri dishes and embedded in 1.5% agar in PBS. After solidification, small cubes were cut and samples embedded in LR White resin as follows. Sample cubes were dehydrated with 30, 50, 70, and 90% ethanol for 30 min each step on ice, then infiltrated with a 2:1 mixture of LR White resin [containing 1.5 μl accelerator (0.5% benzoylperoxide)/1 ml resin] and 90% ethanol for 1 hr on ice, followed by pure resin given two changes for 1 hr each. Samples were left in a third change overnight and subsequently placed in gelatin capsules with fresh resin and polymerized at 50C for 24 hr. Alternatively, samples were embedded according to the progressive lowering of the temperature (PLT) method by using Lowicryl K4M resin (Roth et al. 1981). Sample blocks were trimmed and ultrathin sections were cut with glass knives. Sections were collected on formvar-coated nickel 300-mesh grids and placed on PBS drops before immunolabeling. Labeling was done with protein G-purified rabbit IFP 35-specific immunoglobulin G (IgG) antibodies (250 μg or 125 μg of IgG protein/ml). Before labeling of the sections, the antibody solution was centrifuged at 20,000 × g for 10 min to sediment IgG aggregates. Sections were incubated at 4C overnight and washed with a mild spray of PBS for 30 sec. The bound antibodies were made visible with protein A-gold complexes (10 nm gold particle size). Before use, the protein A-gold suspension was centrifuged at 15,000 × g to sediment aggregated particles. Subsequently, the sections were washed with PBS containing 0.01% Tween 20, followed by a washing step with distilled water before air-drying. Sections were counterstained with 4% aqueous uranyl acetate, pH 4.5, for 10 min. Sections were examined in a Zeiss electron microscope EM910 at an accelerating voltage of 60 kV.

Figure 2

Laser scanning microscopy demonstrates lack of co-localization of IFP 35 with mitochondria (A), actin filaments (B), and peroxisomes (C). Subconfluent HeLa cells were treated with 1000 U/ml IFN-γ for 24 hr, followed by fixation with 3% paraformaldehyde for 10 min and permeabilization with 0.2% Triton X-100 for 15 min (A,B) or treatment with ice-cold methanol for 10 min followed by incubation in ice-cold acetone for 1 min (C). Cells were stained by indirect immunofluorescence with antibodies specific for IFP 35 and mitochondrial antigen (A) or with antibodies specific for IFP 35 and catalase (C). FITC-goat anti-rabbit (A) or goat anti-mouse (C) for labeling anti-IFP 35 antibodies and TRITC-goat anti-mouse (A) or goat anti-rabbit (C) antibodies were used as secondary antibodies. For double labeling of the actin cytoskeleton and IFP 35, cells were stained with rabbit anti-IFP 35 antibody, followed by labeling with FITC-conjugated phalloidin and TRITC-labeled goat anti-rabbit antibody (B). Optical sections were obtained at 0.25-μm intervalls; the combined Z-series are shown. Bars = 10 μm.

Subcellular Fractionation

Cells (7.5 × 10⁶ in 175-cm² flasks) were washed with PBS and detached using trypsin/EDTA (Sigma). Cells were sedimented and washed three times with PBS. Subsequent steps were carried out at 4C. The cell pellet was diluted to a density of 1.5 × 10⁷ cells/ml in 0.25 M sucrose supplemented with 0.1 mM PMSF and 10 μg/ml aprotinin and homogenized on ice in a Dounce homogenizer with a tight-fitting pestle. Cell disruption was evaluated by phase-contrast microscopy. Nuclei and unbroken cells were removed by centrifugation at 800 × g for 10 min. Postnuclear supernatant (750 μl) was layered on 10-ml Percoll prepared in 0.25 M sucrose and centrifuged for 1.5 hr at 27,000 × g (JA 20; Beckman Instruments, Palo Alto, CA). Fractions of 1 ml were collected starting at the top. Each fraction was run on SDS-PAGE and analyzed by immunoblotting.

Figure 3

Sensitivity of IFN-induced IFP 35 structures to CHX. Immunofluorescence analysis: HeLa cells were plated on glass coverslips, grown for 24 hr, and then treated with 1000 U/ml IFN-γ for 24 hr (B-D) or left untreated (A). After the 24-hr IFN treatment, 50 μg/ml CHX was added for 0.5 hr (B), 2 hr (C), or 3 hr (D). Immunofluorescence was carried out with a polyclonal antibody specific for IFP 35 and fluorescein-conjugated goat anti-rabbit antibody as secondary antibody. Cells were examined with a X63 Plan-Neofluar oil immersion objective on a Zeiss Axioscope microscope.

Gel Chromatography

Cells were centrifuged at 100 × g for 10 min at RT and then washed three times with PBS. The following steps were carried out at 4C. Cells were diluted in PBS or intracellular buffer (ICB: 3 mM NaCl, 140 mM KCl, 11 mM EGTA, 10 mM Hepes, pH 7.2, adjusted with KOH) to a density of 1 × 10⁶ cells/ml and homogenized using a Dounce homogenizer. Cell debris and nuclei were removed by centrifugation at 800 × g for 10 min. For ICB, MgCl₂ was added to final concentration of 1 mM and the postnuclear supernatant incubated on ice for 10 min. Postnuclear supernatant was filter-sterilized through a 0.22 μM filter and 100 μl of the filtrate was separated on a Superose 6 column (HR 10/30; Pharmacia, Uppsala, Sweden) using fast-performance liquid chromatography (FPLC, Pharmacia). Depending on the experiment, the column was equilibrated and eluted by using PBS or ICB/MgCl₂ at a flow rate of 0.25 ml/min. Fractions of 0.5 ml were collected and analyzed. The column was calibrated with molecular mass markers (Sigma) [thyroglobulin (669 kD), apoferritin (443 kD), β-amylase (200 kD), yeast alcohol dehydrogenase (150 kD), albumin (66 kD), bovine carbonic anhydrase (29 kD)].

Preparation of Crude Nuclear and Cytoplasmic Extracts

Cells (1 × 10⁷ in 175-cm² flasks) were washed with PBS and detached using trypsin/EDTA (Sigma). Cells were sedimented and washed twice with PBS. Subsequent steps were carried out at 4C. The cell pellet was diluted to a density of 1 × 10⁷ cells/500 μl in NP-40 buffer (10 mM Tris-HCl, pH 7.4, 10 mM NaCl, 3 mM MgCl₂, 0.5% NP-40), incubated for 10 min on ice, and centrifuged for 5 min at 800 × g. The supernatant was recentrifuged for 15 min at 15,000 × g and the resulting supernatant was used as crude cytoplasmic extract. The pellet from the low-speed centrifugation was diluted in Buffer C (20 mM HEPES, pH 7.8, 420 mM NaCl, 3 mM MgCl₂, 0.2 mM EDTA, 25% glycerin, 1 mM DTT, 0.1 mM PMSF, 10 μg/ml aprotinin) to a concentration of 1 × 10⁷ cells/75 μl, incubated for 45 min on ice with vortexing every 15 min, and centrifuged for 5 min at 15,000 × g. The resulting supernatant served as source for crude nuclear extract. Equal amounts of each fraction were separated by SDS-PAGE (Laemmli 1970).

Immunoblotting

Samples were diluted with SDS sample buffer, denatured, and separated by SDS-PAGE. After electrophoresis proteins were transferred on nitrocellulose membranes (Schleicher & Schuell; Dassel, Germany) by electroblotting (1 hr at 0.8 mA/cm²). Membranes were blocked with PBS containing 0.1% Tween 20 and 5% nonfat milk and then incubated for 1 hr with primary antibodies in PBS containing 0.1% Tween 20 and 1.5% nonfat milk. Membranes were washed three times with 0.1% Tween 20 in PBS and incubated with peroxidase-conjugated goat anti-mouse, goat anti-rabbit, or rabbit anti-goat antibodies (Jackson ImmunoResearch) for 1 hr, depending on the primary antibody used. Visualization was done by chemiluminescence (ECL; Pharmacia).

Immunoprecipitation

HeLa cells were plated in six-well plates at a density of 2 × 10⁵ cells/well, transfected as described below, and induced for 24 hr with 1000 U/ml IFN-γ. Cells were then washed with PBS, detached using trypsin/EDTA, sedimented, and washed twice with PBS. Cell lysis was performed using 1% NP-40 in PBS as lysis buffer. The cells were incubated for 45 min on ice at a concentration of 2 × 10⁷ cells/ml in lysis buffer and then centrifuged for 5 min at 15,000 × g at 4C. A total of 100 μl of the resulting supernatant was used for immunoprecipitation: 1 μg of anti-HA antibody clone 12CA5 (Boehringer; Mannheim, Germany) was added and incubated for 1 hr at 4C under mild shaking. Then 100 μl of 20% protein A-Sepharose (Sigma) in lysis buffer was added and incubated for 1 hr at 4C, followed by centrifugation at 350 × g for 5 min at 4C. The resulting pellet was washed four times with lysis buffer and resuspended in 20 μl of SDS sample buffer; 10 μl was separated by SDS-PAGE as described above.

Transfection of HeLa Cells

Cells were plated on six-well plates for preparing cell extracts or on 24-well plates covered with glass coverslips for immunofluorescence studies (2 × 10⁵ or 5 × 10⁴ cells/well, respectively). Cells were cultured overnight and transfected with SuperFect reagent (Qiagen; Hilden, Germany) according to the manufacturer's instructions. Briefly, plasmid DNA (4 μg) was diluted with Dulbecco's MEM containing 4500 mg/ml glucose to a final volume of 100 μl, and 10 μl Super-Fect reagent was added. The sample was vortexed for 10 sec, incubated for 10-30 min at RT and 600 μl culture medium was added. Cells of a six-well plate were washed twice with PBS, PBS was aspirated thoroughly, and the transfection reagent was added, followed by incubation for 10-18 hr at 37C (for transfection of cells grown in a 24-well plate 1 μg DNA was diluted to a volume of 60 μl with Dulbecco's MEM; 5 μl SuperFect reagent and 350 μl of culture medium were subsequently added). After transfection, cells were washed once with PBS and incubated in culture medium supplemented with 1000 U/ml IFN-γ for 24 hr. Cells were prepared for cell extract preparation, immunoprecipitation, or immunofluorescence studies as described above.

The following vectors, which have been described previously (Dorsey et al. 1995; Wang et al. 1996), were used: pDCR, pDCR-B-ATF and pDCR-IFP 35. pDCR is a eukaryotic expression vector expressing HA-tagged IFP 35 (pDCRIFP 35) or HA-tagged B-ATF (pDCR-B-ATF). For Western blot analysis, the HA-tagged protein was stained with MAb anti-HA 12CA5 (Boehringer) at a concentration of 5 μg/ml. For immunofluorescence studies, cells were fixed with 3% paraformaldehyde, permeabilized with 0.1% Triton-100 in PBS, and quenched with 0.05% NP-40 and 10% FCS in PBS. The anti-HA MAb was diluted to 0.8 μg/ml in 0.05% NP-40/10% FCS/PBS and cells were incubated with the antibody for 1 hr at RT and developed with TRITC-conjugated anti-mouse antibody.

Figure 4

Sensitivity of IFN-induced IFP 35 structures to cycloheximide. Western blot analysis: HeLa cells were grown in tissue culture flasks and treated for 24 hr with 1000 U/ml IFN-γ (Lanes 2-7) or left untreated (Lane 1). CHX (50 μg/ml) was added for various periods of time: 0.5 hr (Lane 3), 1 hr (Lane 4), 2 hr (Lane 5), 3 hr (Lane 6), or 6 hr (Lane 7). Cells treated simultaneously with IFN and CHX for 24 hr (Lane 8). Cells treated with CHX alone for 24 hr (Lane 9). Cells were harvested by using trypsin/EDTA solution, rinsed twice with PBS, and suspended in SDS-sample buffer (2 × 10⁷ cells/ml). Ten μl of each suspension was separated on 9.5% SDS-PAGE. Immunoblotting was carried out using polyclonal anti-IFP 35 antibody as primary and peroxidase-conjugated goat anti-rabbit antibody as secondary antibody.

Figure 5

Subcellular fractionation of IFP 35 molecules by Percoll gradient centrifugation. HeLa cells were treated for 24 hr with IFN-γ (1000 U/ml), harvested, washed, and further prepared as described in Materials and Methods. Postnuclear supernatant was layered on Percoll prepared in 0.25 M sucrose and centrifuged for 1.5 hr at 27,000 × g. Fractions of 1 ml were collected starting at the top and were analyzed for IFP 35 and marker protein distribution by immunoblotting. For labeling of marker proteins the following antibodies were used: anti-mitochondrial antigen (mitochondria), anti-transferrin receptor (early endosomes), anti-catalase (peroxisomes), anti-S26 (ribosomes; kindly provided by J. Stahl, Max-Delbrück-Centrum, Berlin, Germany) (Lutsch et al. 1991), and anti-cathepsin D (lysosomes). (A) A 15% Percoll gradient (starting density of 1.05 g/ml); (B) 6.5% Percoll gradient (starting density of 1.04 g/ml). Percoll density was determined using colored marker beads (Pharmacia).

Results

Localization of IFP 35 by Immunofluorescence

HeLa cells were induced with IFN and processed for immunostaining using monoclonal anti-IFP 35 as primary antibody. IFN treatment led to an enlargement in both size and number of cytoplasmic structures stained by anti-IFP 35 antibody. Whereas control cells incubated in medium rarely showed punctate structures, IFN-treated cells showed a speckled staining pattern with 10-20 larger, irregularly shaped elements distributed throughout the cell. In addition to these structures, a relatively diffuse cytoplasmic staining was observed (Figures 1A and 1B). Sometimes tinier structures of more uniform size were also observed. Treatment with IFN-α resulted in the same speckled pattern (data not shown). A polyclonal anti-IFP 35 antiserum as primary antibody gave identical immunofluorescence results with somewhat higher background staining (Figure 1C). Similar observations were made in the human amnion tissue-derived epithelial cell line WISH (Figure 1D), the human fibroblast cell-line GM637, and in PMA-differentiated monocyte/macrophage THP-1 cells (data not shown).

Time course experiments revealed that the characteristic speckled distribution of IFP 35 was observed as early as 6-12 hr after IFN treatment and lasted for over 96 hr. Washing the cells after a 24-hr IFN treatment had no measurable effect on IFP 35 expression during the time period observed (96 hr; data not shown).

The subcellular distribution of IFP 35 was suggestive of vesicle-like structures. To identify the vesicular structure, IFN-treated HeLa cells were co-stained with antibodies specific for IFP 35 and antibodies targeting specific cell organelles: anti-mitochondrial antigen for mitochondria (Herzberg et al. 1993), anti-cathepsin D for lysosomes (Hasilik and Neufeld 1980), anti-mannose 6-phosphate receptor for late endosomes and trans-Golgi network (Schulze-Garg et al. 1993), anti-βCOP for Golgi complex (Allan and Kreis 1986), anti-KDEL peptide for endoplasmic reticulum (Vaux and Fuller 1991), anti-transferrin receptor for early endosomes (Trowbridge and Omary 1981), anti-catalase for peroxisomes (Heikoop et al. 1991), and anti-S3a for ribosomes (Lutsch et al. 1991). No known subcellular structure appeared to have accumulated IFP 35 because no co-localization of IFP 35 with any of these structures was found. The results of the co-localization experiments are shown for mitochondria and peroxisomes (Figure 2A and 2C).

A set of experiments was performed to investigate whether IFP 35 is associated with the cytoskeleton or the microtubule system. Staining with phalloidin to label actin filaments (Wulf et al. 1979) did not show colocalization with IFP 35 (Figure 2B). Treatment with cytochalasin D to disrupt the actin cytoskeleton (Pomidor et al. 1995), although grossly affecting cell morphology, did not inhibit the IFN-induced speckled pattern of IFP 35 expression (not shown). The effect of microtubule disruption on the distribution of IFP 35 was examined in cells treated with nocodazole (Zieve et al. 1980; Rennison et al. 1992) during IFN addition. Depolymerization of microtubules had no effect on the distribution of IFP 35 (not shown). Addition of CHX to prevent protein synthesis had a marked effect on IFP 35 expression. Blocking of protein synthesis with 50 μg/ml CHX 24 hr after IFN stimulation completely eliminated the speckled IFP 35 structures within 3 hr after CHX addition (see Figure 3). CHX addition did not affect the more diffuse cytoplasmic staining of IFP 35. Preabsorption of the polyclonal antibody with IFP 35 antigen eliminated the speckled staining pattern of IFN-treated cells and the more diffuse cytoplasmic staining of IFN-treated cells exposed to CHX (not shown). Immunoblot analysis carried out in parallel revealed that CHX did not markedly affect the expression of IFP 35 at the protein level within 6 hr after CHX treatment in IFN-induced cells (Figure 4). These data suggest that a protein which is distinct from IFP 35 and is CHX-sensitive is needed in addition to IFP 35 to produce the IFN-induced structures observed by immunofluorescence microscopy.

Figure 6

Native molecular mass determination of IFP 35 by gel filtration chromatography. HeLa cells were treated with 1000 U/ml IFN-γ for 24 hr, harvested, and washed. Cells were diluted in PBS to a density of 1 × 10⁶ cells/ml, homogenized, and cell debris and nuclei were removed by low-speed centrifugation. Postnuclear supernatant was passed through a sterile filter and separated by FPLC on a Superose 6 column. Elution was performed with PBS at a flow rate of 0.25 ml/min. Fractions of 0.5 ml were collected and analyzed by immunoblotting. The column was calibrated with molecular mass markers (Sigma): thyroglobulin (669 kD), apoferritin (443 kD), β-amylase (200 kD), yeast alcohol dehydrogenase (150 kD), albumin (66 kD), bovine carbonic anhydrase (29 kD) (using ICB instead of PBS gave the same results; data not shown).

Subcellular Localization

To determine the cellular compartment containing IFP 35, postnuclear supernatants from IFN-treated HeLa cells were fractionated on Percoll gradients. HeLa cells were treated with IFN-γ for 24 hr, homogenized, and cell debris and nuclei removed by centrifugation. The postnuclear supernatant was gradient-centrifuged in 15% Percoll (starting density 1.05 g/ml). To control for the integrity of cell organelles, Western blot analysis with antibodies against different compartments was carried out and demonstrated that the organelles were mostly intact (owing to the fragility of mitochondria, these organelles were partly disrupted by homogenization). The marker for early endosomes and plasma membrane (transferrin receptor) migrated in the low-density fractions (peak fraction 2), as did peroxisomes (peak fraction 2). The lysosomal marker cathepsin D and the mitochondrial marker peaked at densities higher than 1.065 g/ml. IFP 35 molecules were found in the least dense fractions (peak fraction 1 with trailing in the subsequent fractions), distinct from the endosome fractions and the peak of peroxisomes (see Figure 5A). Gradient centrifugation in 6.5% Percoll with a starting density of 1.04 g/ml for higher resolution of the low-density fractions separated IFP 35 from cellular organelles (see Figure 5B). IFP 35 peaked in the same fractions as ribosomes, demonstrating that this protein is confined to the cytoplasm. The results of the cell fractionation experiments are consistent with the apparent lack of co-localization of IFP 35 with organelles in the immunofluorescence studies, although they do not define the structures involved.

To further characterize the cytoplasmic localization of IFP 35, gel filtration experiments to determine its native molecular mass were performed. Postnuclear supernatant from IFN-treated HeLa cells was separated by gel filtration using FPLC and a Superose 6 column (Figure 6). IFP 35, which by itself is a 35-kD protein (Bange et al. 1994), eluted as a high molecular mass protein of 200-440 kD, suggesting that IFP 35 is found in complexes.

Figure 7

Ultrastructural localization of IFP 35 using the postembedding labeling technique. (A) An overview (arrow points to a cluster of gold particles). (B,C) Representative clusters (125 μg IgG/ml of the first antibody was used). A statistical evaluation revealed that 8.8 ± 1.8 gold particles were detected per cluster (50 clusters counted). (D) A representative cluster applying 250 μg IgG/ml of the first antibody; the grayish appearance of the cluster results from aggregated antibodies bound to IFP 35 that have been stained with uranyl acetate. Cells treated with a control polyclonal antibody against maltose binding protein followed by protein A-gold complexes were devoid of any label. G, gold particle; M, mitochondrion; N, nucleus. Bars = 200 nm.

Electron microscopic studies were carried out to examine the structural basis for the immunofluorescence data. Electron microscopy applying the postembedding labeling technique using polyclonal anti-IFP 35 antibody and protein A-gold complexes (10-nm gold particle size) demonstrated that IPF 35 did not localize to any known cellular compartment. Gold particles were mainly observed in clusters distributed throughout the cytoplasm. Clusters typically contained 6-12 gold particles, indicating that IFP 35 forms aggregates in the cytoplasm of IFN-treated cells (see Figure 7). Using Lowicryl K4M as embedding resin gave identical results (not shown). Counterstaining of the sections with uranyl acetate revealed that the clusters of IFP 35 were not associated with intracellular vesicles, the cell membrane, or the nuclear surface. In the statistical evaluation of the number of gold particles per cluster (over 50 clusters counted), only one cluster located at the cell membrane and two clusters in the vicinity of the nuclear surface were found. It is yet to be determined whether other proteins are involved in the formation of these structures, as suggested by the CHX experiments. Control cells not treated with interferon were almost devoid of any label; if at all, only single gold particles instead of clusters were found.

Co-localization of Endogenous and Transfected IFP 35

Final proof for the described unique subcellular localization of IFP 35 was given by transfection studies. HeLa cells were transiently transfected with an expression vector coding for an IFP 35 protein tagged with the hemagglutinin (HA) epitope. As control, cells were transfected with an expression vector coding for HA-tagged B-ATF or with an insertless expression vector. Transfected cells were analyzed by Western blotting, immunofluorescence, and immunoprecipitation studies.

Owing to the insertion of the tag, the HA-tagged IFP 35 showed a slightly higher molecular mass than the endogenous IFP 35, allowing differentiation of endogenous and transfected IFP 35 by SDS-gel electrophoresis and Western blotting using IFP 35-specific antibodies. The HA-tagged IFP 35 partitioned in the same way as the endogenous IFP 35 protein, which is present predominantly in the nuclear fraction of cell extracts (Bange et al. 1994) as demonstrated by Western blot analysis of crude nuclear and cytoplasmic extracts (see Figure 8).

Immunofluorescence studies of transfected cells with an anti-HA MAb showed the same speckled immunofluorescence pattern characteristic for endogenous IFP 35 (compare Figure 9 with Figures 1 and 2). Double immunofluorescence studies using anti-HA MAb and polyclonal anti-IFP 35 antibody suggested co-localization of endogenous and transfected IFP 35 (not shown).

To demonstrate complex formation between endogenous and transfected IFP 35, immunoprecipitation studies were performed. HA-tagged transfected proteins were immunoprecipitated with an anti-HA MAb and immunoprecipitates were subsequently analyzed by Western blotting using IFP 35-specific antibodies (see Figure 10). Co-immunoprecipitation of endogenous IFP 35 was observed in cells transfected with pDCR-IFP 35 but not in cells transfected with pDCR-B-ATF. As a control, aliquots of the cellular extracts used for immunoprecipitation were subjected to SDS-PAGE and submitted to immunoblotting using anti-HA and anti-IFP 35 antibodies. These controls demonstrated that quantitatively similar amounts of endogenous IFP 35 were present in the different cellular extracts investigated and that cells were efficiently transfected with pDCR-IFP 35 and pDCR-B-ATF.

Figure 8

Western blot analysis of cells transfected with HA-tagged IFP 35. HeLa cells were transiently transfected, treated with 1000 U/ml IFN-γ for 24 hr, crude nuclear and cytoplasmic extracts were prepared, and Western blot analysis performed using polyclonal anti-IFP 35. Lane 1, cytoplasmic extract of cells transfected with control vector without insert (pDCR). Lane 2, cytoplasmic extract of cells transfected with pDCR-IFP 35. Lanes 3 and 4, nuclear extracts of cells transfected with pDCR (Lane 3) and with pDCR-IFP 35 (Lane 4).

Discussion

IFP 35 is an IFN-regulated leucine zipper protein of unknown function, which forms homodimers in vitro. Previous studies have shown that IFP 35 was able to pellet in the nuclear fraction of crude cellular extracts (Bange et al. 1994), which also contains membranous and subcellular cytoskeletal fractions. By immunofluorescence, we demonstrate here that IFP 35 localizes to the cytoplasm and is found mainly in conspicuous punctate structures, in addition to more diffuse cytoplasmic staining. This morphological structure was not restricted to a particular cell type but was observed in a variety of cells after IFN treatment.

The speckled pattern found by immunofluorescence microscopy was suggestive of vesicle-like structures. Co-localization experiments using double immunofluorescence failed to demonstrate an association with known organelles. Subcellular fractionation studies using Percoll gradient centrifugation also demonstrated a lack of association of IFP 35 with organelle-containing fractions. IFP 35 was found in the least dense fractions, indicative of a cytoplasmic localization. As a possible clue to the unique subcellular morphology of IFP 35, the native molecular mass of IFP 35 was determined by gel filtration of cytoplasmic extracts. These experiments demonstrated that IFP 35 eluted at a molecular mass of 200-440 kD, suggesting that IFP 35 is part of larger protein complexes. Given that IFP 35 forms homodimers in vitro (Bange et al. 1994), it is possible that these protein complexes contain multimers of IFP 35, a conclusion further supported by electron microscopy and immunoprecipitation studies. An involvement of other proteins in these complexes was suggested by our CHX experiments. These experiments demonstrated a sensitivity of the morphological IFP 35 structures to CHX, i.e., in contrast to the more diffuse cytoplasmic staining, the characteristic speckled pattern observed by immunofluorescence after IFN addition disappeared after a 3-hr treatment with CHX. Western blot analysis demonstrated that the disappearance of the characteristic morphological structures was not due to CHX sensitivity of the IFP 35 protein itself. Apparently an as yet unknown, CHX-sensitive protein distinct from IFP 35 is involved in forming these complexes and in shuttling IFP 35 between a dissociated cytoplasmic pool and an aggregated form.

Figure 9

Immunofluorescence of cells transfected with HA-tagged IFP 35. HeLa cells were transiently transfected and treated with 1000 U IFN-γ/ml for 24 hr. Cells were fixed, permeabilized and stained with monoclonal anti-HA antibody. (A) Cells transfected with control vector pDCR. (B) Cells transfected with pDCR-IFP 35. Cells were examined with a X63 Plan-Neofluor oil immersion objective on a Zeiss Axioscope microscope.

As a further proof to validate our observations on the subcellular localization of IFP 35, transfection experiments were carried out. In these experiments, an HA-tagged expression construct of IFP 35 was used. Staining of transfected cells with an HA-specific MAb showed the same speckled cytoplasmic expression pattern as observed with endogenous IFP 35 (compare Figure 9B with Figures 1 and 2). As expected from the immunofluorescence results, separation of transfected cells into crude cytoplasmic and nuclear extracts demonstrated an identical partitioning for endogenous and transfected IFP 35 (see Figure 8). Immunoprecipitation of transfected cells using an HA-specific MAb and subsequent Western blot analysis of the immunoprecipitates demonstrated complex formation between endogenous and transfected IFP 35. This finding supports our previous observation that IFP 35 forms homodimers in vitro. Presumably, this interaction is at least in part mediated by the leucine zipper of IFP 35. The results from the immunoprecipitation experiments support the conclusion that the characteristic morphological structures of IFP 35, as observed by immunofluorescence and electron microscopy, consist of multimers of IFP 35 with or without other proteins. Our immunocytochemical and biochemical characterization strongly suggests that these structures represent large cytosolic protein aggregates. These structures are not associated with membrane-bound vesicles, as evident from subcellular fractionation and electron microscopic studies.

More recently, it has been suggested that IFP 35 forms complexes with B-ATF, a member of the AP1 family of nuclear, basic leucine zipper proteins (Dorsey et al. 1995; Wang et al. 1996). In these experiments, it was demonstrated that IFP 35 interacts with B-ATF in the yeast two-hybrid system as well as in vitro using pull-down assays with bacterially expressed proteins and that, on preparation of crude cytoplasmic and nuclear extracts from transfected cells, both IFP 35 and B-ATF partitioned predominantly in the nuclear fraction. Although we continue to observe in vitro interaction between IFP 35 and B-ATF, a more direct analysis of the potential of these two proteins to interact in vivo by double immunofluorescence studies failed to demonstrate co-localization of IFP 35 and B-ATF in HeLa cells (not shown). Transfected HA-tagged B-ATF was exclusively a nuclear protein which is in stark contrast to the cytoplasmic expression of IFP 35. To further support these observations, immunoprecipitation experiments from transfected HeLa cells failed to detect complex formation between B-ATF and IFP 35 in vivo (see Figure 10).

An important question concerns the biological implications of our findings. Although no firm conclusions are yet possible in this respect, preliminary immunofluorescence experiments suggest a partial co-localization of IFP 35 with cytokeratins (M.B. Omary, personal communication; and our own unpublished results). The association of cytokeratins with proteins involved in signal transduction, such as the 14-3-3 protein (Liao and Omary 1996) and PKC_∊-related kinases (Omary et al. 1992), and ligand-receptor binding-induced phosphorylation of cytokeratins (Baribault et al. 1989; Basso et al. 1994) have evoked the view that keratin intermediate filaments may play a role in signal transduction. It is tempting to speculate that the association of IFP 35 with cytokeratins might be involved in some early or late signaling events of IFNs (for review see Schindler and Darnell 1995; Pellegrini and Dusanter-Fourt 1997).

Figure 10

Immunoprecipitation analysis of cells transfected with HA-tagged IFP 35. HeLa cells were transiently transfected, treated with 1000 U IFN-γ/ml, extracts were prepared, and immunoprecipitation was performed using monoclonal anti-HA antibody. Cellular extracts before immunoprecipitation were separated by SDS-PAGE, blotted, and stained with monoclonal anti-IFP 35 (A) or anti-HA (B) antibody to control for the presence of endogenous IFP 35 and transfected IFP 35 and B-ATF, respectively. Cellular extracts were immunoprecipitated using polyclonal anti-IFP 35 (C) or anti-HA (D) antibody. Immunoprecipitates were separated on SDS-PAGE, blotted, and stained with monoclonal anti-IFP 35 antibody. Lane 1, cells transfected with the pDCR plasmid. Lane 2, cells transfected with pDCR-IFP 35. Lane 3, cells transfected with pDCR-B-ATF. The arrow marks the transfected IFP 35 tagged with the hemagglutinin epitope; an open arrow marks the HA-tagged B-ATF. The endogenous IFP 35 is indicated by an asterisk.

There is precedence for binding of IFN-induced proteins to the cytoskeleton. The IFN-induced ubiquitin-crossreactive protein, a protein of 15 kD that has been suggested to target a set of proteins for degradation via the ubiquitin-dependent degradative pathway, has been shown to distribute in a cytoskeletal pattern (Loeb and Haas 1994). In addition, IFNs directly affect cellular compartmentalization and cytoskeletal structure, e.g., by promoting phagolysosome fusion (Sibley et al. 1985, 1987; Kagaya et al. 1989) and by de novo expression of cytokeratin K17 (Flohr et al. 1992; Jiang et al. 1994; Carroll et al. 1997). It remains to be determined whether IFP 35-related cellular reorganization plays any role in signal transduction of interferons, in targeted protein degradation/sequestration, or in interferon-mediated changes of the cytoskeleton.