Analysis of Prototype Foamy Virus particle-host cell interaction with autofluorescent retroviral particles

Peptide length and location influence function of tagged PFV Gag

We set out to establish a collection of tagged PFV Gag proteins that retain most of their natural functions essential for FV replication. With these tools we aim to study various steps of the FV replication strategy in host cells by combining different biochemical assays with modern live-cell imaging techniques. Towards this end we generated expression constructs containing different protein tags fused in frame with the PFV Gag ORF (Fig. 1). Recombinant PFV vector particles containing these Gag fusion proteins (Gag-FPs) were produced by transient transfection of 293T cells using a 4-plasmid PFV vector system [29]. Subsequently, cellular protein expression, particle-associated protein composition, and infectivity of recombinant vector particles were examined. Biochemical analysis of cell lysates revealed that all Gag-FPs were expressed and processed at levels slightly lower or similar to untagged PFV Gag (Fig. 2A). Increases in the observed molecular weight of the individual tagged Gag proteins were consistent with the predicted size of the different peptide tags added. For N-terminal tagged Gag proteins, both the p71^Gag and p68^Gag displayed a higher mass in comparison to untagged PFV Gag (Fig. 2A, lane 1-6). In contrast, for the C-terminal tagged Gag proteins, only the p71^Gag precursor protein showed a higher molecular weight because normal C-terminal proteolytic processing led to authentic p68^Gag cleavage products lacking the tag (Fig. 2A, lane 8-13). Initial analysis of particle release, by particle concentration through ultracentrifugation and subsequent Western blot analysis using FV specific antisera, revealed that all of the tagged PFV Gag proteins appeared to support particle egress (Fig. 2B). However, in general, the release of capsid containing N-terminal tagged Gag proteins was significantly decreased in comparison to wild type (Fig. 2B, lane 1-6). Furthermore, in the lysates of the larger N-terminal AFP-tagged Gag protein particle preparations no viral glycoprotein was detectable, evidenced by the lack of PFV Env LP specific signals (Fig. 2C, lane 3-6). In contrast, particle lysates of the smaller N-terminal HA-tagged Gag displayed incorporation of the PFV Env LP subunit (Fig. 2C, lane 2).

To investigate whether detected Gag proteins were particle-associated or extracellular protein aggregates, purified particle samples were digested with the membrane-impermeable protease subtilisin, prior to particle lysis (Fig. 2B; lower panel). By this treatment, all viral protein components not enveloped and protected by a lipid membrane are removed. Indeed, we observed that in all N-terminal Gag-AFP samples the Gag-specific signals detected in duplicates that were mock treated (Fig. 2B, lane 3-6, upper panel) disappeared upon subtilisin digestion (Fig. 2B, lane 3-6; lower panel). All other samples, including N-terminal HA-tagged- and all C-terminal tagged Gag proteins, were unaffected by proteolytic digestion and appear as Gag-specific signals in the Western Blot analysis (Fig. 2B, lanes 1, 2, 7-20; compare upper and lower panel). Remarkably, there was an additional prominent protein band in all C-terminal tagged mCherry-Gag samples, recognized with both Gag-specific and mCherry-specific antibodies (Fig. 2A, B, lane 13, 19; data not shown). This protein most probably is the result of an internal mCherry cleavage, which has been described in the literature, and is thought to be involved in maintaining the functional chromophore of this fluorescent protein [30–32].

We further observed that the small HA-tag fused to the N-terminus of Gag significantly reduced particle release efficiency in comparison to wild type, which was not observed for the C-terminal HA-tagged Gag-FP (Fig. 2B, lane 1, 2, 8, 9). These effects of HA-tag addition on particle release were in accordance with the calculated relative infectivities depicted in Fig. 2D. Samples of N-terminal HA-tagged particles showed a 10-fold reduction of supernatant-associated infectivity, whereas those of C-terminal HA-Gag-FP particles were almost at wild type levels (Fig. 2D, bar 1, 2, 8, 9). This suggests that the PFV Gag N-terminus is more sensitive to modifications than the C-terminus. Furthermore, addition of different AFPs to the N-terminus of Gag almost completely abolished release of infectious particles (Fig. 2D, bar 3-6). This observation is in line with the inability of these proteins to support release of lipid membrane enveloped Gag protein (Fig. 2B, lane 3-6). In contrast, the range of supernatant infectivity measured for C-terminal Gag-AFPs was between 1 - 8% compared to untagged wild type samples (Fig. 2D, bar 8, 10-13). Since the physical particle release of these samples was almost equal to wild type (Fig. 2B, lane 8-13), this reduction in measurable infectivity indicates that a larger C-terminal fusion tag might interfere with replication steps other than particle release. No major difference in the relative incorporation and processing of Pol was observed in released particles of the individual Gag mutants (data not shown). To examine if untagged wild type PFV Gag protein is able to rescue the particle release and infectivity defects observed for some of the Gag-FP, we cotransfected expression constructs of both type of proteins at various ratios (Fig. 2; and data not shown). In Fig. 2E, the influence of cotransfection of various ratios of wild type Gag with C-terminal tagged Gag-GFP on supernatant infectivity is shown. By increasing the ratio of wild type Gag protein to tagged protein the infectivity could be restored, reaching wild type levels at a 3:1 ratio of wild type to tagged Gag protein and 50% infectivity levels at a 1:1 ratio. For the N-terminal tagged Gag-GFP, cotransfection of wild type Gag was unable to restore supernatant infectivity to wild type levels, even at a 15-fold excess of wild type Gag expression construct (data not shown). This suggests a dominant negative effect of the N-terminal Gag-GFP fusion. Subsequently, physical particle release of all fusion proteins was analyzed at a 1:1 cotransfection ratio and compared to conditions without wild type Gag protein coexpression (Fig. 2A-D; and data not shown). For all C-terminal tagged Gag constructs a similar ratio of tagged and wild type protein was detected in corresponding cell and particle lysates (Fig. 2B, lane 14-20). In contrast, no tagged Gag protein was observed in particle lysates of samples cotransfected with N-terminal AFP-tagged constructs (data not shown). Supernatant infectivities of the C-terminal tagged constructs were restored to 15-100% of wild type levels independent of the specific tag sequence used. The relative differences in infectivities between the various tagged constructs were similar, independent of wild type Gag protein coexpression. Thus, C-terminal, but not N-terminal AFP-tagged PFV Gag proteins, can interact with wild type Gag protein to allow release of mixed particles with greatly improved specific particle infectivity.

C-terminally tagged Gag-AFPs display nearly normal capsid structures and budding characteristics

Due to the apparently decreased infectious titer of several Gag-AFP tagged particles observed, we were interested in taking a closer look at the particle morphology of these fluorescent viruses. Therefore, we used ultrastructural EM (electron microscopy) to analyze 293T cells expressing different GFP-tagged Gag-FPs in the context of the 4-plasmid FV vector system (Fig. 3).

Wild type unmodified Gag proteins were found to assemble into homogenous spherical capsids accumulating intracellularly in large amounts mainly at the MTOC (microtubule organizing center), as previously reported (Fig. 3A, B). Furthermore, particle budding was observed into intracellular vesicles and to a large extent also at the plasma membrane, sometimes associated with capsids aggregating at the plasma membrane (Fig. 3C, D). Similar to wild type PFV Gag, N-terminal tagged Gag-GFP also assembled into capsids with wild type morphology and accumulating mainly at the MTOC (Fig. 3E, F). However, in these samples no budding profiles could be detected (Fig. 3E, F; and data not shown). This is in line with the biochemical analysis (Fig. 2B, C) and indicate that the lack of particle release may be due to a failure of the N-terminal tagged Gag-AFP to successfully interact with PFV Env, an interaction that is essential for capsid-membrane association. In contrast, C-terminal Gag-GFP-FPs were found to bud at the plasma membrane, indicating that a functional Gag-Env interaction occurs and that the GFP tag does not influence late budding events (Fig. 3J, K). In this case, capsid morphology seemed to be slightly more heterogeneous compared to untagged capsids. But capsids were also found to accumulate at the MTOC, and budding structures containing the typical prominent FV Env spike structures at the plasma membrane were observed (Fig. 3G, I, J, K). Remarkably, in some cells in these samples, we detected intracellular accumulation of potentially aberrant capsid structures which might represent sites of protein degradation (Fig. 3H). These curious structures were neither found at the budding site nor in released viruses of C-terminal tagged PFV Gag samples nor in samples of other tagged or wild type Gag constructs. This suggests that C-terminal AFP tags to the PFV Gag protein may result in some minor interference with intracellular capsid assembly, however, all budding and released virions displayed wild type morphology.

EYFP and EGFP are the most convenient tags to analyze PFV capsids by fluorescence microscope techniques

Since the biochemical analysis revealed that all four C-terminal tagged autofluorescent Gag-FPs mediate particle release of infectious virions, we were interested to determine if single fluorescent particles can be imaged by Confocal Laser Scanning Microscopy (CLSM). For this purpose particles purified by ultracentrifugation were spotted onto glass cover slips, fixed and further analyzed by CLSM. The results obtained are summarized in Fig. 4. Whereas EGFP and EYFP tagged PFV particles could be detected very easily, mCherry and mCerulean modified virus particles showed very low signal intensities (Fig. 4A). Although mCerulean and mCherry were incorporated into particles (Fig. 2B, lane 12, 13), they were only detectable by making "blind scans". Subsequent image correction with ImageJ plugins and further modifications of brightness and contrast levels, finally led to the images shown in Fig. 4B. The particle signal intensities calculated from non-modified original scan pictures and the results given as average of the maximum pixel values per particle (n = 30) are shown in Fig. 4A. Furthermore, no GFP signals were detected in mock-purified supernatants of 293T cells, which were cotransfected with pcoPG4 CeGFP in the context of the 4-plasmid vector system lacking an Env expression plasmid (data not shown). Thus PFV Gag-AFP proteins seem to be released in particulate forms in a PFV Env-dependent manner, like the wild type protein.

Gag-GFP labelled PFV particle preparations contain single viruses

We were interested in verifying that autofluorescent PFV particle preparations contain predominantly single virions and not aggregates. For this purpose a comparative ultrastructural analysis on C-terminal Gag-GFP-tagged PFV particle preparations was applied. Labelled virions were harvested by ultracentrifugation and simultaneously fixed in paraformaldehyde. Purified PFV particles were prepared for a combined AFM (atomic force microscopy) and CLSM analyses, performed as described in materials and methods. They were mixed prior to analysis with fluorescent beads (100 nm in diameter) to obtain topographical landmarks useful for alignment of AFM and CLSM scans resulting in three important advantages. First, the same excitation wavelength (488 nm) could be used for Gag-GFP labelled virions and fluorescent beads. Furthermore, CLSM scans nicely show oversaturated beads located next to less intensive GFP-tagged particles, a typical example of which is shown in Fig. 5A. Second, applying distance measurement analysis between beads and particles in the CLSM scan enabled identification of the appropriate GFP-tagged particles in the AFM scan (Fig. 5B). Third, the bead diameter of 100 nm gave us the possibility to compare the size of PFV particles in the AFM scan. In cross section analysis the average height of single PFV particles was calculated as 85 nm (n = 11, standard deviation 13 nm; data not shown). Thus combined AFM- and CLSM analysis confirmed that C-terminal AFP-tagged PFV particle preparation contained predominantly single virions.

PFV particles bind to the host cell surface, but not to extracted host cell lipids

One special feature of the FV life cycle is an extremely broad host range. To date, there are no reports identifying species, tissues or cell types that are not susceptible to FV Env-mediated transmission. This suggests that the FV receptor molecule(s) is evolutionarily well conserved and present on most if not all eukaryotic cell membranes. We were interested in using the functional fluorescently-tagged PFV particles described above as a tool to measure and visualize potential virus-receptor interactions.

Host cell lipids, in addition to proteins and carbohydrates, are the major constituent of cellular membranes and are also implicated in uptake mediated by VSV-G, a viral glycoprotein displaying a broad host range similar to the FV Env protein [33, 34]. The potential involvement of host cell lipids for FV Env mediated entry was tested using two approaches. First, synthetic lipids or a lipid mixture extracted from the FV susceptible human cell line HeLa were spotted onto a glass slide. Subsequently, several differently tagged viral particle preparations, normalized for physical particle concentration, which was determined by FCS, were incubated with the spotted lipids. After extensive washing, particle binding was examined by CLSM (Fig. 6A). GFP-tagged HIV-VSV-G pseudoparticle binding was detectable for HeLa lipids containing phosphatidylserine (PS) and to a slightly lower extent for a mixture containing 30% synthetic PS (DOPS, dioleoyl phosphatidylserine) and 70% DOPC (dioleoyl phosphatidylcholine), but not for DOPC alone (Fig. 6A+B, left column). In contrast, both GFP-tagged HIV virions lacking a viral glycoprotein and GFP-tagged PFV virions displayed minimal or no binding capacity to any of the lipids examined (Fig. 6A+B, center and right column). In a second approach, HeLa cell lipid extracts were used to generate giant unilamellar vesicles (GUV). Control experiments showed that these lipid extracts contained both charged lipids as PS and glycosylated lipids as GM1 (data not shown). But incubation of these GUVs with purified EGFP-tagged PFV virions for up to 30 minutes followed by CLSM analysis of the samples resulted in no indication of FV particle attachment to the GUV surface (Fig. 6C), whereas HIV-VSV-G pseudotype particle binding was clearly detectable (data not shown). Labelled PFV virion signals were only detectable in the liquid surrounding the GUVs (Fig. 6C). Thus, neither lipids extracted from susceptible cells by the method employed nor selected synthetic lipids seem to contribute to PFV particle attachment.

Second, we examined the capacity of fluorescent PFV particles to bind to target cells. For this purpose HeLa cells were incubated with concentrated GFP-tagged PFV virions, followed by extensive washing and subsequent investigation by CLSM analysis. Binding of Gag-GFP-labelled particles to the surface of HeLa cells was readily detectable (Fig. 6D, PGwt +Env). Since particle release of FVs is strictly glycoprotein-dependent, we were unable to assess the binding capacity of FV VLP lacking FV Env. Therefore we made use of a PFV Gag mutant (PGM3) that contains a heterologous N-terminal membrane-targeting signal to examine the FV Env-independent binding capacity of FV virions. Similar PFV Gag proteins were reported previously to enable Env-independent PFV particle release [8, 9]. As illustrated in Fig. 6D GFP tagged PGM3 virions harboring PFV Env (PGM3 +Env) were capable of attaching to the HeLa cell surface whereas GFP tagged PGM3 virions generated in the absence of PFV Env coexpression (PGM3 ΔEnv) had a strongly reduced binding capacity. Thus, specific binding of GFP-tagged virus to target cells was observed.

Subsequently, a more quantitative and sensitive flow cytometric assay to assess target cell binding of GFP-tagged PFV virions was established. A clear shift in the mean fluorescence intensities was observed upon incubation of HeLa cells with wild type Gag-GFP-labelled particles (PG-GFP) in comparison to mock treated cells (mock) (Fig. 7A). Further this shift was also obtained for PGM3-GFP labelled particles harboring PFV Env (PGM3-GFP +Env) in comparison to those lacking PFV Env (PGM3-GFP ΔEnv) or mock-treated cells (mock) (Fig. 7B). However, a significant binding activity of Env-deficient PGM3-GFP particles (PGM3-GFP ΔEnv) was detected on HeLa cells in comparison to mock-incubated cells (mock), implying an Env-independent component of FV particle attachment to target cells similar to previous reports for other retroviruses [35]. Target cell attachment of Gag-GFP labelled PFV virions was dose-dependent (Fig. 7C) and could be competed for by untagged PFV particles (Fig. 7D).

Identification of cell lines resistant to PFV-Env mediated vector transduction

Previous attempts to identify cell lines non-permissive for FV infection proved to be unsuccessful [24, 25]. We extended the analysis of FV-Env mediated host range further by challenging target cells of various origins with high-titer supernatants of PFV vectors and HIV-1 VSV-G or PFV Env pseudotypes (Fig. 8A, B). First, we examined whether proteoglycans are essential for PFV transduction by comparing the transduction efficiency of mouse L-cell and a proteoglycan synthesis-deficient subclone thereof called Sog9 [36]. As shown in Fig. 8A, Sog9 cells were 2-3 fold better transduced by HIV-1 VSV-G pseudotypes than parental mouse L cells. In contrast, PFV Env-mediated transduction of PFV or HIV-1 PFV Env pseudotypes was diminished about 10-fold on the Sog9 cell line in comparison to parental mouse L cells; nevertheless Sog9 cells were still clearly susceptible to PFV Env-mediated entry. This indicates that proteoglycans are not absolutely essential for FV susceptibility, although they seem to contribute to significant extent to PFV Env-mediated infection efficiency. Second, we examined transduction efficiencies of various other target cells including cells of the human hematopoietic lineage and other species (Fig. 8A, B; and data not shown). All target cell types examined were clearly susceptible to VSV-G mediated marker gene transfer (Fig. 8A, B; and data not shown). However, the extensive analysis led to the identification of two cell lines apparently resistant to PFV Env-mediated vector transduction (Fig. 8A, B). No infectivity of PFV Env containing vector supernatants was detectable on the zebrafish cell line Pac2 (Fig. 8A) and the human erythroid precursor cell line G1E-ER4 (Fig. 8B) even after transduction by spinoculation. In contrast, VSV-G pseudotype titers were 500-fold above the detection limit.

Finally, we examined these two cell lines, along with susceptible adherent and suspension cell lines as controls, for their retroviral particle binding capacity. Therefore the flow cytometric assay using various GFP-tagged retroviral virions as described above was used to get a first idea at which step of the entry process the block might occur (Fig. 8C, D). Most adherent and suspension cell lines examined clearly bound VSV-G containing HIV particles, in line with the transduction data (Fig. 8C, D), although significant cell-type specific differences in relative binding capacities of corresponding Env-less or Env-containing particles were observable. Binding of FV Env containing particles but not of VSV-G pseudotypes was diminished on the proteoglycan synthesis deficient Sog9 cells in comparison to the parental mouse L-cells (Fig. 8C). This is in line with the transduction susceptibility data of these cell lines. Interestingly, Pac2 cells still specifically bound different FV Env containing PFV particles (Fig. 8C). The same was observed for G1E-ER4 cells; however, here clearly specific binding of FV Env-containing virus was only observable by using spinoculation (spin) during particle incubation with the target cells (Fig. 8D). Taken together, these data suggest that both cell lines still demonstrate a FV Env specific attachment of retroviral particles, and a currently unknown post-attachment block might be responsible for the failure of FV Env-mediated marker gene transfer and expression by different retroviral vectors in these target cells.

Cells

The human kidney cell line 293T [42], the human fibrosarcoma cell line HT1080 [43], the mouse L, the sog9 [36] and the HeLa cell line [44] were cultivated in Dulbecco's modified Eagle's medium supplemented with 10% heat-inactivated fetal calf serum and antibiotics. HeLa cells were cultivated in phenol red free media. The zebrafish embryonic fibroblast cell line Pac2 was cultivated in Leibovitz media L15 supplemented with 20% heat-inactivated fetal bovine serum and antibiotics at 28°C [45, 46]. The suspension T-cell line Jurkat [47] was cultivated in RPMI-1640 media supplemented with 10% heat-inactivated fetal calf serum and antibiotics. The immortalized, erythroid suspension cell line G1E-ER4 was cultivated in Iscove's modified Dulbecco's medium, supplemented with 15% heat-inactivated fetal calf serum, recombinant human erythropoietin (2 U/ml) and recombinant rat SCF (50 ng/ml) [48].

Expression constructs

The original 4-plasmid PFV vector system consisting of the PFV Gag expression vector pcziPG4, the PFV Pol expression vector pcziPol, the PFV Env expression construct pczHFVenvEM002, and the enhanced green fluorescent protein (EGFP)-expressing PFV transfer vector pMD9, has been described previously [29]. In this study an expression-optimized PFV Gag construct pcoPG4 (PG) was used instead of the original pcziGag4 construct that contains the wild type PFV Gag ORF. Expression-optimization and gene synthesis was done by Geneart, Regensburg, Germany. Furthermore, a variant transfer vector puc2MD9 was used, containing a pUC19 backbone with a SV40 ori instead of the pcDNA3.1 zeo backbone of the original pMD9 vector. For some experiments the PFV transfer vector pMD11, encoding lacZ as reporter gene, was used [29]. A schematic outline of the PFV Gag constructs used in this study is shown in Fig. 1. Expression vectors for Gag fusion proteins were cloned by fusing the tag sequence (HA, EGFP, EYFP, mCerulean or mCherry), together with a flexible glycine-serine (G/S) linker in between, either N- (e.g. pcoPG4 NeGFP) or C-terminal (e.g. pcoPG4 CeGFP) to the PFV Gag ORF in pcoPG4. Further modification of pcoPG4 CeGFP by addition of an N-terminal tag comprising the c-src membrane-targeting signal (aa 1-10), a HA tag, a flexible G/S linker and an additional N-terminal Gag p68/p3 proteolytic cleavage site to the full-length PFV Gag ORF resulted in the generation of the PGM3 mutant, allowing PFV Env-independent membrane-targeting (Lindemann, unpublished data). All Gag fusion protein (FP) expression constructs were generated using standard PCR cloning techniques and mutagenesis primers and were verified by sequencing analysis. Details are available upon request. In some transduction experiments a replication-deficient lentiviral vector system was used. HIV-1 pseudotyped viruses were generated by cotransfection of the constitutively EGFP expressing lentiviral transfer vector p6NST90 (Lindemann, unpublished) and the packaging plasmids pCD/NL-BH and pczVSV-G encoding for HIV-1 Gag/Pol and the VSV-G protein (Vesicular Stomatitis Virus glycoprotein G), respectively [49, 50].

Generation of viral supernatants and analysis of transduction efficiency

FV supernatants containing recombinant viral particles were generated essentially as described previously [51, 52]. Briefly, FV supernatants were produced by cotransfection of 293T cells with transfer vector (puc2MD9 or pMD11), Env- (pczHFVenvEM002), Pol - (pcziPol), and Gag packaging plasmid (pcoPG4 or PG mutants thereof as indicated) at a ratio of 4:4:4:1 using polyethyleneimine (PEI) or Polyfect transfection reagents. At 24 h posttransfection, sodium butyrate (final concentration, 10 mM) was added to the growth medium for 8 h. Subsequently, the medium was replaced, and viral supernatants were harvested an additional 16 h later. Lentiviral supernatants were generated by cotransfection of transfer vector (p6NST90), Gag/Pol packaging plasmid (pCD/NL-BH), and an Env packaging plasmid (pczVSV-G or pczPE01) at a ratio of 1:1:1 and harvested as described above.

Transductions of recombinant EGFP expressing PFV vector particles containing various PFV Gag proteins were performed by infection of 2 × 10⁴ HT1080 cells, plated 24 h in advance in 12-well plates. For the infection 1 ml of the viral supernatant or dilutions thereof were incubated with the target cells for 4 to 6 h. The percentage of EGFP-positive cells was determined by fluorescence-activated cell sorter (FACS) analysis 72 h after infection. All transduction experiments were performed three times, and in each independent experiment the values obtained with the wild-type construct pcoPG4 were arbitrarily set to 100%. Analysis of tissue tropism of different PFV vector particles or HIV-1 vector pseudotypes was performed on various target cell lines. Adherent target cells (HT1080, HeLa, mouseL, Sog9, Pac2), which were plated one day in advance at a density of 2 × 10⁴ cells in 12-well plates, were infected with one ml of viral cell culture supernatant or dilutions thereof. Target cells growing in suspension (Jurkat, G1E-ER4) were infected by resuspending 1 × 10⁵ target cells in one ml of viral cell culture supernatant or dilutions thereof. Afterwards they were transferred into a 6-well plate and either incubated at 37°C, 5% CO₂ in a humidified incubator or centrifuged for 1 h at 2000 rpm (30°C) before the incubation step. Sixteen hours later the viral cell culture supernatant was replaced for both types of target cells by fresh media and the transduction efficiency was determined by flow cytometry 72 - 96 h after infection as described above.

Biochemical analysis of PFV particles

Biochemical analysis of purified PFV particles was essentially performed as described previously [15, 53]. Briefly, the cell-free viral supernatant, generated by transient transfection of 293T cells as described above, was harvested by sterile filtration (pore size, 0.45 μm) and centrifuged at 4°C and 25,000 rpm for 3 h in an SW40 or SW28 rotor through a 20% sucrose cushion. The supernatant was discarded, and the viral pellet was gently resuspended in 50 μl phosphate-buffered saline (PBS). An equal volume of 2× sodium dodecyl sulfate (SDS) protein sample buffer (2×PPPC) was added to the samples, which were separated by SDS-polyacrylamide gel electrophoresis (PAGE) and analyzed by Western Blotting as described below. In some experiments viral particles were resuspended in a larger volume (134 μl PBS). Subsequently, 67 μl of each sample were proteolytically digested with subtilisin (0.5 mg/ml), the other part was incubated with PBS instead, for 2 h at 37°C. Digestion was terminated by addition of 2 μl phenylmethylsulfonyl fluoride (20 mg/ml) and 6×PPPC.

Antisera, Western blot expression analysis

Western blot expression analysis of cell- and particle-associated viral proteins was performed essentially as described previously [53]. Polyclonal antisera used were specific for PFV Gag [54] or the LP of PFV Env, aa 1 to 86 [53]. The chemiluminescence signal was digitally recorded using a LAS-3000 imager.

FACS based analysis of viral particle target cell binding

A flow cytometric assay was applied to determine and compare the capability of individual particle preparations to specifically bind to selected target cells. GFP-tagged PFV PG or PFV PGM3 derived particles of 10 ml viral supernatant, generated by transient transfection of 293T cells as described above, were harvested, concentrated by ultracentrifugation and resuspended in 100 μl PBS (phosphate-buffered saline) containing 10% inactivated fetal calf serum. HIV VSV-G pseudoparticles were generated as described previously [40]. Adherent target cells were detached with PBS-EDTA. Subsequently, 5 × 10⁴ cells were incubated with virus or dilutions thereof in a total volume of 100 μl. After 30 min incubation on ice, cells were washed two times with FACS buffer (phosphate buffered saline, 1% inactivated fetal calf serum). For the competition experiment, untagged PFV particles or dilutions thereof were incubated 30 min on ice, washed two times with FACS buffer and subsequently incubated another 30 min with GFP-tagged PFV particles. After a fixation step with 6% formaldehyde, the cell pellets were resuspended in 150 μl FACS buffer, stored on ice, and analyzed by flow cytometry using a FACS Calibur (Becton Dickinson). The mean fluorescence of 8,000 - 10,000 events per sample was subsequently determined using the Cell Quest software package (Becton Dickinson).

Lipid binding assays

Lipid binding assays were performed with purified PFV particles or HIV pseudoparticles, generated as described above. Synthetic lipids were obtained from Avanti Polar Lipids, Alabaster, USA. Lipid extracts were prepared from non-transfected HeLaP4 cells according to Dreyfus et al. [55]. This is a modified method according to Folch-Pi et al. with optimizations for recovering highly polar lipids like Gangliosides [56]. Briefly, cells were grown to near confluency, washed with PBS and 150 mM NaCl and were harvested by gentle scraping. 5 × 10⁷ cells were suspended in a total volume of 450 μl of 150 mM NaCl. All following solvent extraction steps were carried out in a glass vial. 5 ml of chloroform/methanol (1:1, v/v) was added. To facilitate extraction of lipids, the suspension was sonicated for 30 min in a bath sonicator interrupted by 2 × 5 min of vortexing. Insoluble material was pelleted by centrifugation at 3200 × g for 10 min and subsequently subjected to 2 further extraction steps. Three ml of chloroform/methanol (1:1) or chloroform/methanol/H₂O (48:35:10, v/v), respectively, were added and extraction and centrifugation were performed as before. All 3 supernatants were collected, dried under nitrogen and redissolved in 1.5 ml chloroform/methanol/H₂O (40:20:3, v/v).

For the lipid spot binding assay small spots of lipid solutions of about 25 mg/ml in organic solvent were dried on cover slips. Remaining traces of solvent were removed by applying vacuum for at least 1 h. The spots were covered by PBS, which contained fluorescent viral particles. The concentration of viral particles had been measured by fluorescence correlation spectroscopy (FCS) before and adjusted accordingly. Confocal laser scanning images of the surface of the lipid spots were obtained on a Zeiss LSM 510 and evaluated by ImageJ and custom made Perl scripts. In short, particles were detected and counted in 3 subsequent images after smoothing and intensity thresholding. Diffusing particles that appear blurred due to slow scanning were discarded based on their low circularity (sqrt(area)/perimeter < 0.21). For a quality control of the lipid extract the binding of cholera toxin subunit B and Annexin V was observed in PBS or PBS containing 2 mM CaCl₂, respectively. The toxin and Annexin V both bound to the spotted lipid extract but not to synthetic DOPC and only Annexin V bound to DOPS. This shows that both charged lipids as phosphatidyl serine (detected by Annexin V) and highly polar glycolipids like GM1 (detected by cholera toxin) were extracted.

Giant unilamellar vesicles (GUVs) were produced by electroformation [57]. In short, lipids were dried on opposed indium tin oxide (ITO) coated coverslips and the resulting chamber was filled with 465 mM sucrose/2 mM EDTA. An alternating voltage of 2 V/10 Hz was applied for 2 h in order for the GUVs to form. These vesicles were sedimented in double the volume of 1.5 × PBS. Virus particles were added before observation of binding in confocal microscopy.

Combined Atomic Force Microscopy (AFM) and Confocal Fluorescence Microscopy

PFV particles were harvested as described above and additionally fixed by addition of PFA (paraformaldehyde) at a final concentration of 2% to the cell culture supernatant and to the 20% sucrose cushion prior to ultracentrifugation.

AFM and fluorescence imaging were performed at room temperature using the same experimental apparatus. It consisted of a NanoWizard AFM (JPK Instruments, Berlin, Germany) mounted on a Laser scanning microscope (LSM) 510 Meta (Zeiss, Jena, Germany). For AFM imaging, uncoated silicon cantilevers (MikroMasch, Spain) with typical spring constant of 0.3 N/m (manufacturer specified) were used in intermittent contact mode. The cantilever oscillation was tuned to a frequency of ~5 kHz, with a maximum amplitude set to 0.1- 0.15 V (5-7.5 nm). The scan rate was set to 0.7- 1 Hz. The height, error, and phase-shift signals were collected simultaneously in both trace and retrace directions. Images were line-fitted as required. Isolated scan lines were occasionally removed. For confocal fluorescence microscopy, the excitation light of an Argon laser at 488 nm was reflected by a dichroic mirror (HFT 490) and focused onto the sample by a Zeiss C-Apochromat 40×, NA = 1.2 UV-VIS-IR water immersion objective. Fluorescence signal was then recollected by the same objective and, after passing through a 530/30 bandpass filter, measured by a photomultiplier (PMT). The confocal geometry was ensured by a 70 nm pinhole in front of the PMT.

Precise spatial alignment of fluorescence imaging and AFM was achieved using 0.1 μm size fluorescent carboxyl-modified polystyrene beads (F-8888, Molecular Probes, Eugene, OR) as "topographical landmarks". Viral particles and fluorescent beads were mixed in phosphate buffer (1 mM KCl, 0.5 mM KH₂PO₄, 2.7 mM Na₂HPO₄, 50 mM NaCl, pH 7.2) and deposited on a thin fresh cleaved mica sheet previously treated for 10 minutes with a 0.1 mg/ml poly-D-lysine (P7280, Sigma) solution. After ca. 20 minutes, the sample was rinsed with the same phosphate buffer and then ready for imaging.

CLSM analysis of PFV binding to host cells

Host cells were seeded at a density of 1.5 × 10⁴ cells/well into 8 well chamber slides. After 24 h cells were cooled down and incubated on ice with fluorescent PFV particle preparations for 30 min. Subsequently cells were washed with cold PBS and either fixed with 3% PFA or incubated an additional 30 min at 37°C before fixation. After fixation the cell nuclei were stained with DAPI for 5 min. Finally the cells were covered with 50% glycerol (in water). Confocal laser scanning images were obtained on a Zeiss LSM 510 and evaluated by ImageJ. The excitation light of an Argon laser at 488 nm or a Diode laser at 405 nm was reflected by a dichroic mirror (HFT 405/488/561) and focused onto the sample by a Zeiss Apochromat 63×, NA = 1.4 oil immersion objective. Fluorescence signal was then recollected by the same objective, splitted by a dichroic mirror (NFT 490) and after passing through either a 520/30 or 420/60 bandpass filter, measured by a photomultiplier (PMT). The confocal geometry was ensured by a 1 Airy Unit pinhole in front of the PMT.

Electron microscopy analysis

At 48 h post transfection, the 293T cells were harvested and processed for electron microscopy analysis as described previously [58].