NefG3C-GFP is incorporated in HIV-1 virions at high levels
The N-terminal palmitoylation significantly increases the localization at the cell membrane rafts of Nef both in its native form [20] and when fused with GFP [22]. Thus, since rafts are the sites from where HIV preferentially buds [15], we expected that the NefG3C-GFP fusion product would be incorporated into virions at levels which would be high enough to result in the HIV-1 particles being highly fluorescent. To validate such a prediction, we compared the levels of incorporation of NefG3C-GFP into virions with those of wt Nef and wt Nef-GFP. In addition, the virion associated amounts of NefG3C-GFP were compared with those of GFP-Vpr, already known to incorporate efficiently in viral particles [9, 10]. We co-transfected 293T cells with the pCMVΔR8.74 HIV-1 packaging construct together with vectors expressing each of the above HIV-1 protein derivatives. Forty-eight hours later, the supernatants were harvested, clarified, and the VLPs purified and analyzed by Western blot for the incorporation of Nef- or GFP-derivatives. We observed that equal amounts of viral particles gave rise to NefG3C-GFP signals much stronger than those produced by both wt Nef and wt Nef-GFP (Figure 1A), and also slightly more intense compared with GFP-Vpr (Figure 1B). This suggests that NefG3C-GFP is incorporated at high levels in HIV-1 viral particles.
Since NefG3C strongly associates with cell membranes [20, 22], it is conceivable that at least part of the NefG3C-GFP we detected was associated with microvesicles and/or exosomes rather than to viral particles. To test this possibility, we transfected 293T cells with the NefG3C-GFP expressing vector alone or together with the pCMVΔR8.74 HIV-1 packaging construct. Forty-eight hours later, the supernatants were harvested, concentrated and purified as above described, and equal volumes of each preparation were analyzed by anti-Nef Western blot. As shown in Figure 1C, no detectable amounts of NefG3C-GFP were found in supernatants from 293T cells transfected in the absence of the HIV-1 packaging construct. This strongly suggests that the most part of NefG3C-GFP molecules we detected in supernatants from the cells co-transfected with the HIV-1 packaging construct were indeed associated with viral particles.
These data strongly support the hypothesis that Nef palmitoylation results in a significant increase in virion incorporation levels which was the basis to attempt the generation of cells stably releasing VLPs with incorporated NefG3C-GFP.
Isolation and induction of the 18-4s cell line stably expressing the NefG3C-GFP fusion product
We have used the 293 Rev-Gag-Pol [23] (henceforth referred to as 293/RGP) as a platform for the generation of stable cells releasing fluorescent HIV-1 based VLPs. 293/RGP cells are packaging cells for HIV-1 derived vectors inducibly expressing and releasing HIV-1 VLPs. In the 293/RGP cells, the HIV-gag-pol gene and the HIV-rev are separately expressed under the control an ecdysone-inducible promoter (for a review, see [24]) so that particle production requires addition of the ecdysone analogue ponasterone A (PonA) to the medium.
293/RGP cells were co-transfected with the vectors expressing the NefG3C-GFP fusion product and the p75 human Nerve Growth factor receptor truncated in its intracytoplasmic domain (ΔNGFr). Indeed, the expression of this marker was useful in the first selection steps, when the cell sorter-based selection led the cells to an unacceptable cell mortality. In this respect, after two anti-NGFr selections followed by two GFP-based cell sorting, we recovered a cell population (18-4s cells) more than 95% positive for the NefG3C-GFP expression (Figure 2A). Of note, at the FACs analysis 18-4s cells actually appeared as two distinct sub-populations in terms of the expression of both NefG3C-GFP and ΔNGFr. 18-4s cell grew with a similar duplication time to that of the parental cells (not shown), but tended to float much more easily. The levels of NefG3C-GFP remained stable during two months period in culture, but progressively declining thereafter (not shown). This appeared to be the consequence of the silencing rather than the loss of the NefG3C-GFP sequences, since treatment with sodium butyrate restored a GFP FACs profile similar to early-passage cells (not shown).
We then sought to establish the best experimental conditions for the induction of the expression of HIV-1 products in 18-4s cells. Following the methods previously described for the 293/RGP cells [23], we treated 18-4s cells with 5 mM sodium butyrate and/or PonA 2 μM, and monitored the expression of HIV-1 CAp24 and NefG3C-GFP by FACs two and three days later. Specifically, 293/RGP and 18-4s cells were treated with: i) PonA alone; ii) both PonA and sodium butyrate for one day and, thereafter, with PonA alone, or iii) with PonA and sodium butyrate. As shown in Figure 2B, the treatment with PonA alone resulted in inefficient expression of Gag products in 18-4s cells, while more than 50% of the parental 293/RGP cells was induced to express Gag products. The addition of sodium butyrate was mandatory for significant Gag induction in 18-4s cells. In fact, a pulse of sodium butyrate for 24 h resulted in 60% of the cells expressing Gag whereas its continuous presence for the 48 h induction period further increased the percentage of positive cells to 77%. Additionally, the expression of NefG3C-GFP was increased 3–4 fold. Sodium butyrate treatment increased also the percentage of Gag positive 293/RGP cells from 57% to more than 80%, and improved the relative mean fluorescence intensity by 2–2.5 fold. Of note, in 18-4s cells the treatment with sodium butyrate alone led to an increase of the NefG3C-GFP expression similar to that induced also in the presence of PonA (not shown). No remarkable differences in terms of expression of both CAp24 and NefG3C-GFP were observed between the cells harvested at the day 2 and at the day 3 (not shown). At this time, however, the cell cultures suffered of some cell mortality (not shown).
Finally, the intracellular localization of NefG3C-GFP we analyzed by fluorescence microscope seemed qualitatively similar to that already described for wt Nef [25] in the presence, however, of a stronger localization at the cell membrane (not shown).
Dose-response of the PonA induced production of VLPs from 18-4s cells
Clearly, the most relevant feature of 18-4s cells was expected to be the ability to produce fluorescent VLPs. In this regard, the VLP production efficiency was assayed by inducing 18-4s cells with increasing concentrations of PonA in the presence of the optimal doses of sodium butyrate (i.e., 5 mM) that was mandatory for the expression of HIV-1 CAp24 and related products. As for 293/RGP, 18-4s cells produced VLPs at concentrations that increased with the doses of PonA up to 2 μM (Figure 3). Higher doses of PonA did not further increase the amounts of produced VLPs (not shown). At the highest concentrations of PonA, 18-4s cells produced amounts of VLPs equivalent to up to 4 μg/ml. However, these levels were still 2–3 fold lower compared with those from the parental cells. This could be the consequence of the incorporation of the high number of NefG3C-GFP molecules that may in some way decrease the efficiency of the HIV assembling/budding process.
Characterization of NefG3C-GFP VLPs
Next, we molecularly characterized the VLPs released by 18-4s cells. The Western blot analysis for the cell associated HIV-1 Gag-related products did not show significant differences compared with the parental cells, while the anti-Nef Western blot analysis, consistently with the here above reported results, revealed an induction-dependent apparent increase of the NefG3C-GFP expression (Figure 4A). Concerning the VLP molecular composition, no significant differences were detected in the contents of Gag-related products between HIV-1 and VLPs from 293/RGP or 18-4s cells. On the other hand, and of a major relevance, remarkable amounts of NefG3C-GFP molecules appeared to be associated with the VLPs from 18-4s cells, as also confirmed by the anti-GFP Western blot analysis (not shown). However, in order to ensure that the NefG3C-GFP molecules we detected were indeed incorporated and not just associated with their surface, we additionally performed Western blot analysis after a proteolytic treatment of the VLPs. Thus, VLPs from 18-4s which had been pseudotyped with the glycoprotein receptor of the Vesicular Stomatitis virus (VSV-G) were treated with the subtilisin A protease and analyzed for the presence of both NefG3C-GFP and VSV-G. Such a treatment was expected to degrade the viral surface proteins as well as any products non specifically associated with the viral envelope, but should not affect genuinely incorporated proteins. As shown in Figure 4B, while the NefG3C-GFP related signal appeared substantially unaffected by the subtilisin A treatment, the VSV-G receptor resulted efficiently degraded. To further support the idea that the Nef-related signals we detected by Western blot analysis indeed originated from virion incorporated NefG3C-GFP molecules, VLPs were purified through a iodixanol gradient, and the recovered fractions analyzed by Western blot (Figure 4C). The co-sedimentation of NefG3C-GFP and CAp24 strengthens the conclusion that NefG3C-GFP molecules are indeed incorporated in VLPs.
Furthermore, we were interested in estimating the number of molecules incorporated in the VLPs released from induced 18-4s cells. To this end, we compared the intensity of the signals from decreasing amounts of recombinant (r) Nef with those from decreasing amounts of VLPs whose contents in CAp24 were preventively and rigorously established by quantitative ELISA (Figure 4D). On the basis of the densitometric analysis carried out on the Western blot films (not shown), we estimated that 20 ng of 18-4s VLPs contain approximately 1.25 ng of NefG3C-GFP molecules. Considering the purity of rNef preparation (i.e., 95%), the differences in the molecular weight between CAp24 and NefG3C-GFP, and that it was reported that HIV-1 particles incorporate approximately 5,000 molecules of CAp24 per virion [26], we calculated the presence of about 150 NefG3C-GFP molecules per VLP. Such an estimation cannot take account of possible differences in the antibody recognition between NefG3C-GFP and wt rNef. These, however, are not expected to be of a major relevance since the anti-Nef polyclonal Abs preparation we employed was recovered using a full-length wt rNef derivative as immunogen.
Finally, we analyzed the morphology of the VLPs released by induced 18-4s cells by electron microscope as compared with VLPs from parental 293/RGP cells. Clearly, 18-4s cells released VLPs with a morphology very similar to that of the VLPs from parental cells (Figure 5A–B), thus suggesting that the NefG3C-GFP incorporation did not grossly influence the VLP morphology significantly. Both VLP populations appeared composed of immature and mature viral particles, the latter bearing the typical electron dense conical cone structures. Of note, in both cell cultures, the presence of immature VLPs raised significantly with the cell passages (not shown)
In conclusion, our data indicate that, upon induction, 18-4s cells produce HIV-1 VLPs whose morphology was not influenced by the high levels of incorporated NefG3C-GFP molecules.
Cytofluorimetric detection of fluorescent NefG3C-GFP VLPs
Next, we developed a rapid and sensitive assay for the detection of the VLP-associated fluorescence based on the ability of aldehyde latex beads to specifically bind lipid enveloped micro- and nanoparticles. Similar tests were previously applied to the analysis of exosomes [27], and here we originally provide evidence that such a method can be fruitfully exploited for the analysis of VLPs. In detail, we incubated clarified supernatants from induced 18-4s cells with the aldehyde latex beads at r.t. that, 1–2 hours later, were extensively washed and analyzed by FACs. The representative results reported in Figure 6A demonstrate that this assay can detect as few as 25 ng of 18-4s VLPs measured as CAp24 contents. The possible presence of free NefG3C-GFP molecules did not interfere with the FACs measurements since no fluorescence increase was found associated to the beads incubated with the supernatants from induced 18-4s cells previously cleared from VLPs by ultracentrifugation as compared with the control supernatants (Figure 6B). In addition, the actual binding of VLPs to the beads was confirmed by the specific labeling of either virion-associated CAp24 or Nef with monoclonal Abs we observed only upon a permeabilization step (Figure 6C).
We exploited such VLP analytical method to compare the fluorescence associated with NefG3C-GFP VLPs with that of VLPs incorporating GFP-Vpr. To this aim, 293T cells were co-transfected with a HIV-1 packaging vector together with the vector expressing the respective GFP-related product, and the released VLPs purified on a 20% sucrose cushion. Then, equivalent amounts of VLPs, as measured by a quantitative anti-CAp24 ELISA, were incubated with aldehyde latex beads, and the fluorescence evaluated by FACs. As shown in Figure 7A, beads binding NefG3C-GFP VLPs showed more than 3-fold increased fluorescence as compared with the GFP-Vpr VLPs. This was not the result of different transfection efficiencies in the respective producer cells, where the GFP-related products appeared equally expressed (not shown), but is consistent with the VLP Western blot analysis depicted in Figure 1B.
Next, to assess whether the NefG3C-GFP system could be applied also to other retroviruses, we evaluated the levels of incorporation of NefG3C-GFP molecules in Moloney Leukemia Virus (MLV) particles, that were shown to efficiently incorporate HIV-1 Nef [28]. To this aim, we compared the fluorescence associated to VLPs produced in the presence of wt Nef-GFP with that of NefG3C-GFP. 293T cells were co-transfected with the vector expressing the respective GFP-related product together with a Gag-Pol MLV packaging construct, and equivalent amounts of purified VLPs, as measured by the Mn++ dependent reverse transcriptase assay, were incubated with aldehyde latex beads. The FACs analysis (Figure 7B) clearly showed low levels of fluorescence associated with wt Nef-GFP MLV VLPs, while strong GFP-related signals arose from the beads binding the NefG3C-GFP MLV VLPs. This result represents the proof-of-principle that the NefG3C-GFP approach could be of utility also for non HIV-1 retroviral systems.
Pseudotyped 18-4s VLPs efficiently deliver NefG3C-GFP molecules in target cells
Finally, we were interested in establishing whether and how efficiently 18-4s VLPs can fluorescently mark the target cells upon viral entry. To this aim, 18-4s VLPs pseudotyped with either the VSV-G (in its wt form or with a mutant defective for the fusion activity) [29] or the HIV-1 X4 Env receptors were produced by transfecting 18-4s cells with the appropriate expression vectors and, 6 to 8 hours later, treating the cells with PonA and sodium butyrate. Of note, the liposome-mediated transfection worked efficiently in 18-4s as well as in 293/RGP parental cells, ranging the expression of transfected receptors from 60 to 95%, as measured by FACs (not shown). Thus, 18-4s cells remained a good recipient for the expression of ectopic sequences. VLP preparations were preventively characterized in terms of their contents of both NefG3C-GFP molecules and the respective receptors (Figure 8A). Next, we assessed that 250 ng/105 cells of VLPs pseudotyped with the wt but not with the fusion-mutant VSV-G rendered the majority of challenged cells fluorescent at the FACs analysis as early as 2 hours after the VLP treatment. (Figure 8B). In addition, the pre-treatment of target cells with either bafilomycin A1 or chloroquine, i.e. two drugs inhibiting the VSV-G mediated endocytic fusion by raising the endosomal/lysosomal pH, led to a clear reduction of the fluorescence of the challenged cells (Figure 8B). We interpret the residual cell-associated fluorescence as the product of the accumulation of fluorescent VLPs in not functional endosomes/lysosomes rather than of a suboptimal cell response to the inhibitors. In fact, similar outcomes were obtained also by raising the drug concentrations up to 5-fold (not shown).
Conversely, and as expected, the viral entry driven by HIV-1 Env appeared significantly less efficient, considering that 1 μg/105 cells of HIV-1 Env pseudotyped VLPs rendered fluorescent about 40% of target cells that also internalized lower amounts of NefG3C-GFP, as indicated by the lower mean fluorescence intensity (Figure 8C). The idea that the fluorescence detected in target cells indeed relied on authentic viral fusion events was strengthen by the inhibition we observed upon the pre-incubation of HIV-1 Env VLPs with either the 17b anti-Env gp 120 (Figure 8C), or the 2F5 anti-Env gp41 neutralizing mAbs (not shown).
The analysis at the fluorescence microscope of the target cells (Figure 9A) showed a strong fluorescence associated with (VSV-G) NefG3C-GFP VLP challenged cells, while the fluorescence levels appeared significantly fainter in the cells challenged with the X4 Env pseudotyped VLPs. Finally, the confocal microscope analysis formally confirmed that the fluorescence associated with the target cells originated from an authentic cell internalization of fluorescent VLPs pseudotyped with VSV-G (Figure 9B), or with HIV-1 Env receptors (not shown). Besides the above described challenges carried out with VLPs lacking functional receptors, such results were validated also by the lack of fluorescence we observed in cells challenged with either VSV-G or X4 Env pseudotyped VLPs and incubated at 4°C (not shown). Finally, the results obtained in CEMss cells were also reproduced in U937 cells, primary CD4+ lymphocytes, or human primary monocyte-derived macrophages (not shown).
Taken together, our data support the idea that the 18-4s cells could be a useful reagent for the study of early events correlating with the viral entry mediated also by heterologous virus receptors.