Abstract
The cone-shaped mature HIV-1 capsid is the main orchestrator of early viral replication. After cytosolic entry, it transports the viral replication complex along microtubules towards the nucleus. Capsid uncoating from the viral genome apparently occurs beyond the nuclear pore. Observation of post-entry events via microscopic detection of HIV-1 capsid protein (CA) is challenging, since epitope shielding limits immunodetection, and the genetic fragility of CA hampers other labeling approaches. Here, we present a minimally invasive strategy based on genetic code expansion and click chemistry that allows for site-directed fluorescent labeling of HIV-1 CA, while retaining virus morphology and infectivity. Thereby, we could directly visualize virions and subviral complexes using advanced microscopy, including nanoscopy and correlative imaging. Quantification of signal intensities of subviral complexes showed that the amount of CA associated with nuclear complexes in HeLa-derived cells and primary T cells is consistent with a complete capsid and revealed that treatment with the small molecule inhibitor PF74 did not result in capsid dissociation from nuclear complexes. Cone-shaped objects detected in the nucleus by electron tomography were clearly identified as capsid-derived structures by correlative microscopy. High-resolution imaging revealed dose-dependent clustering of nuclear capsids, suggesting that incoming particles may follow common entry routes.
Introduction
The cone-shaped capsid that encases the viral RNA genome and replication proteins is a characteristic feature of infectious human immunodeficiency virus type 1 (HIV-1) particles. Data obtained by many research groups over the past decade have revised our understanding of the role of the mature capsid in HIV-1 replication, placing this structure at the center stage of post-entry replication steps (reviewed in e.g., (Aiken and Rousso, 2021; Engelman, 2021; Guedan et al., 2021; James, 2019; Novikova et al., 2019). Upon fusion of the virion envelope with the cell membrane, the capsid, which consists of ∼1,200-1,500 monomers of the capsid protein CA (Briggs et al., 2003), is released into the cytosol. It then usurps host cell factors to traffic towards the nucleus. Reverse transcription of the viral RNA into dsDNA is initiated during passage of the subviral structure through the cytosol. Following import into the nucleus, the viral dsDNA is covalently integrated into the host cell genome by the viral integrase. Prior to integration, the surrounding capsid shell needs to expose the dsDNA in a process termed uncoating. However, the precise mechanisms, location, and timing of capsid uncoating are still under investigation.
Initially, the HIV-1 capsid was presumed to rapidly dissociate upon cell entry, based on little or no CA detected associated with isolated post-entry complexes (reviewed in (Campbell and Hope, 2015)). Rapid or gradual disassembly in the cytosol was also supported by several studies that applied fluorescence imaging to analyze subviral complexes in infected cells (e.g., (Hulme et al., 2011; Mamede et al., 2017; Xu et al., 2013). However, the finding that CA or the capsid lattice, directly interacts with various host factors involved in post-entry replication steps (cytosolic proteins, including proteins involved in microtubular transport, but also nucleoporins and even the nuclear protein CPSF6; reviewed in (Engelman, 2021; Naghavi, 2021; Saito and Yamashita, 2021)) implied involvement of at least a partial lattice structure in later stages of post-entry replication. Furthermore, increasing evidence from imaging-based analyses argued for capsid uncoating at the nuclear pore (Burdick et al., 2017; Francis et al., 2020a; Francis et al., 2016; Francis and Melikyan, 2018), or even indicated passage of (nearly) intact capsids through nuclear pores (Burdick et al., 2020; Li et al., 2021; Muller et al., 2021; Zila et al., 2021). The recent detection of cone-shaped objects in the nuclear pore channel and inside the nucleus by correlative light and electron microscopy (CLEM) (Zila et al., 2021), and intranuclear separation of CA or IN from reverse transcribed dsDNA (Muller et al., 2021) also support the model that the nucleus is the site of HIV-1 uncoating.
One explanation for apparent discrepancies between different studies are the methods that have been used for CA detection in fluorescence microscopy. Since the modification of CA by genetic labeling strategies proved to be challenging, most studies applied immunofluorescence (IF) staining or indirect labeling through a capsid binding protein (e.g. (Burdick et al., 2017; Francis et al., 2016; Hulme et al., 2015; Mamede et al., 2017; Peng et al., 2014; Zila et al., 2019). A limitation of IF is that staining efficiency may vary substantially depending on the antibody and detection conditions used, as well as on differential exposure or shielding of epitopes due to conformational changes or different intracellular environments. We could indeed show previously that immunostaining efficiency of CA in the nucleus of host cells strongly depends on cell type and experimental conditions (Muller et al., 2021). Furthermore, IF is incompatible with live cell analyses. Infectious HIV-1 derivatives carrying fluorescent CA would resolve these limitations and allow the direct observation of entering capsids with quantitative analyses.
Direct genetic labeling of viral capsid proteins is challenging, however. Capsid proteins are generally small proteins that need to assemble into ordered multimeric lattices. The resulting assemblies must be stable during virus formation and transmission to a new target cell, but also ready to disassemble in the newly infected cell, requiring structural flexibility of the protomers. Beyond protein-protein interactions involved in capsid assembly itself, capsid proteins generally undergo crucial interactions with other components of the virion, e.g., the viral genome. Finally, the capsid surface represents an essential contact interface between virus and host cell in the early phase of infection, mediating cell entry in the case of non-enveloped viruses, or interacting with critical host cell dependency or restriction factors in the case of enveloped viruses. Consequently, a large proportion of the surface exposed amino acids of a viral capsid protein is involved in intermolecular contacts that are crucial for virus replication, which renders these proteins highly susceptible to genetic modification. Fusion of a capsid protein to a relatively large genetic label, e.g., green fluorescent protein (GFP) or other fluorescent proteins, is thus generally prone to severely affect virus infectivity.
These considerations also apply to HIV-1 CA. The protein is encoded as a subdomain of the structural polyprotein Gag, from which it is released by the viral protease (PR) concomitant with virus budding to allow for formation of the mature capsid. With a molecular mass of ∼24 kDa, mature CA is of a similar size as GFP. Hexa- and pentamers of CA are the core structural elements of the immature Gag polyprotein shell forming the nascent virus bud in HIV-1 producing cells, as well as of the mature capsid lattice. CA pentamers, immature and mature hexamers employ different protein-protein interfaces; together, these interfaces involve most of the exposed surface of the CA monomer (reviewed in (Mattei et al., 2016)). Accordingly, scanning mutagenesis analyses found HIV-1 CA to be highly genetically fragile (Rihn et al., 2013; von Schwedler et al., 2003), with up to 89% of single amino acid exchanges tested abolishing or severely affecting virus replication (Rihn et al., 2013). It is thus not surprising that the introduction of genetically encoded labels - GFP or even a small peptide tag - at various positions within HIV-1 CA have resulted in loss or severe reduction of infectivity. Complementation with wild-type (wt) virus, from at least equimolar amounts of wt CA to a substantial molar excess, was essential to restore virus infectivity (Burdick et al., 2020; Campbell et al., 2008; Pereira et al., 2011; Zurnic Bonisch et al., 2020). While the use of wt complemented particles can be sufficient for fluorescent labeling, it is unclear whether the modified CA molecules are an integral part of the mature CA lattice; only ∼ 50% of CA molecules present inside the virion are eventually used to form the mature capsid (Briggs et al., 2004; Lanman et al., 2004), and incorporated fusion proteins may be preferentially excluded or less stably integrated into the mature lattice.
We therefore established and applied a minimal invasive labeling strategy for HIV-1 CA based on genetic code expansion and click labeling. This method involves the exchange of a selected amino acid residue in the protein of interest with a non-canonical amino acid (ncAA) carrying a highly reactive bio-orthogonal functional group by a process termed amber suppression (Figure 1a); this residue is subsequently covalently coupled to a fluorophore functionalized with a cognate reaction partner (Figure 1a; reviewed in e.g. (Lang and Chin, 2014; Muller et al., 2019; Nikic and Lemke, 2015)). Using this approach, we generated a CA-labeled HIV-1 derivative that largely retained infectivity; in contrast to previous approaches for direct CA labeling, our minimally modified derivative did not require complementation with wt virus. Direct labeling with a bright and photostable chemical dye allows the application of various imaging methods, i.e., live-cell imaging, super-resolution nanoscopy, or CLEM. The virus variant click labeled with a bright and photostable chemical dye thus enabled us to directly assess the amount of CA associated with entering subviral complexes outside and within the nucleus of infected HeLa-derived cells and primary CD4+ T-cells, to visualize CA containing structures in the nucleus by nanoscopy and correlative microscopy and to study the effect of the CA-binding drug PF74 on the nuclear complexes.
Results
Generation of an HIV-1 variant carrying a bio-orthogonal amino acid within CA
To allow for minimally invasive labeling of HIV-1 CA by genetic code expansion (GCE; Figure 1a), we introduced an amber stop codon at a position of interest into the CA coding sequence within the gag open reading frame of the proviral plasmid pNLC4-3 (Bohne and Krausslich, 2004). In order to avoid GCE modification of the viral protein R (Vpr), which is incorporated into the virion in high amounts (Muller et al., 2000), we first exchanged the amber stop codon of vpr to an opal codon (TGA), resulting in plasmid pNLC4-3*. Albeit this mutation did not alter the coding sequence of viral proteins or virion infectivity, the corresponding virus was termed HIV-1* to indicate this modification. Since neither the efficiency of amber suppression in a given sequence context in eukaryotic cells, nor the effect of ncAA incorporation on viral functionality can be predicted with certainty, we tested a panel of 18 amber mutations at sites located towards the outer surface of the capsid lattice for suppression efficiency and virus infectivity (Schifferdecker, Sakin et al., in preparation). Based on a comparison of Gag expression levels and viral infectivity upon ncAA incorporation, we selected a virus variant in which residue alanine 14 in CA was replaced by a non-canonical amino acid (HIV-1*CA14ncAA) for further analyses.
For virus preparation, HEK293T cells were co-transfected with the respective mutant proviral plasmid and pNESPylRS-eRF1dn-tRNA. The latter plasmid encodes for a complete amber suppression system, consisting of modified tRNA, a cognate genetically engineered pyrrolysine aminoacyl-tRNA synthetase (Nikic et al., 2016), and a dominant-negative version of the eukaryotic release factor eRF1 that improves amber suppression efficiency in eukaryotic cells (Schmied et al., 2014). To produce functionalized virus particles, cells were grown in the presence of the small ncAA cyclopropene lysine (CpK). While truncation of Gag at position 14 of CA would prevent virus formation, incorporation of CpK by amber suppression should result in the expression of full-length Gag and thereby promote HIV-1 particle assembly.
Immunoblot analysis of cell lysates indeed demonstrated the presence of full-length Gag polyprotein precursor when HIV-1*CA14TAG expressing cells were grown in the presence of CpK, whereas full-length Gag was not detected when CpK was omitted from the growth medium (Figure S1). Thin-section electron microscopy (EM) revealed late budding sites and immature-as well as mature-like virions at the plasma membrane and in the vicinity of HIV-1*CA14ncAA expressing cells, that were morphologically indistinguishable from typical HIV-1 wild-type (wt) budding sites and virions (Figure 1b). We concluded that Gag expression of HIV-1*CA14TAG is ncAA dependent and the modified CA domain is competent for immature and mature lattice assembly.
Characterization of click labeled HIV-1 virions
We next prepared virus particles from the supernatant of HIV-1*CA14ncAA producing cells and subjected them to click labeling using the membrane-permeable dye silicon rhodamine tetrazine (SiR-Tet; (Lukinavicius et al., 2013)), generating HIV-1*CA14SiR. As a control, HIV-1* wt particles were prepared under amber suppression conditions and stained in parallel. Consistent with the detection of viral assembly sites and particles in electron micrographs (Figure 1b), virus was recovered from the tissue culture supernatant of HIV-1*CA14ncAA expressing cells. Particle yields were somewhat reduced compared to the HIV-1* wt control, in line with the fact that amber suppression is usually incomplete in eukaryotic cells (optimal ncAA incorporation efficiencies in the range of ∼25-50 %; e.g., (Sakin et al., 2017; Schmied et al., 2014). On average, we obtained 5-10-fold lower yields for HIV-1*CA14SiR compared to HIV-1* (Figure 1c, d). Consistent with the observation of morphologically mature particles by EM, click labeled particles displayed regular Gag and GagPol processing products (Figure 1e), with clear bands for mature RT heterodimer (p51, p66) and mature CA (p24). In-gel fluorescence revealed a distinct SiR labeled band corresponding to a mass of approximately 24 kDa for HIV-1*CA14SiR, but not for HIV-1* control particles (Figure 1f). Taken together, these findings indicate specific GCE-dependent labeling of CA via amber suppression at position 14 of HIV-1 CA.
Fluorescence labeling and infectivity of click labeled virions
To test specificity and efficiency of SiR staining, labeled particles adhered to a glass chamber slide were fixed, permeabilized, and immunostained with antiserum raised against HIV-1 CA to validate that detected signals corresponded to virus particles. Confocal micrographs were recorded in the channels corresponding to the CA immunofluorescence (IF) stain (green) and direct CA labeling with SiR (magenta) (Figure 2a). Regions of interest (ROIs) corresponding to the position of virus particles were defined based on CA(IF) signals. Measurement of SiR fluorescence intensities in these ROIs revealed weak background staining in the case of HIV-1* (Figure 2a, left panel). In contrast, distinct SiR signals co-localizing with CA(IF) punctae were detected for HIV-1*CA14SiR (Figure 2a, right panel). Quantitative analyses of images from multiple independent experiments confirmed this visual impression (Figure 2b). Only ∼8.5% of HIV-1* particles were classified as SiR positive, with fluorescence intensities only slightly above the background level (∼1,000 a.u.). In contrast, >95% HIV-1*CA14SiR particles displayed clear SiR staining, with a mean fluorescence intensity of ∼15,000 a.u. Variation in SiR fluorescence intensities between individual particles is expected, since particle size and CA content of HIV-1 virions varies, with ∼1,700-3,100 CA monomers estimated per particle (Carlson et al., 2008)). Beyond that, the range of SiR signal intensities observed also indicates a range of click labeling efficiencies. Despite some variability in the preparation, the vast majority of HIV-1*CA14CpK particles could be efficiently click labeled with SiR, attaining fluorescence intensities suitable for fluorescence microscopy of infected cells.
To test the effect of introducing a synthetic fluorophore at position 14 on CA functionality, the infectivity of click labeled particles was assessed by titration of labeled particles on TZM-bl cells, followed by immunostaining against the HIV-1 matrix protein (MA) to identify infected cells. As shown in Figures 2c and d, relative infectivity of HIV-1*CA14SiR was only mildly reduced by an average of ∼2-fold compared to HIV-1*, a substantial improvement compared to previous genetically labeled derivatives in the absence of complementation (Burdick et al., 2020; Campbell et al., 2008; Pereira et al., 2011; Zurnic Bonisch et al., 2020). Thus, minimal invasive labeling by GCE allows direct labeling of HIV-1 CA without requiring complementation with wt virus.
Detection of click labeled HIV-1 in infected cells
Having established a suitable labeling strategy, we used labeled particles to infect target cells. Initial experiments were performed in the model cell line HeLa TZM-bl. Cells infected with HIV-1*CA14SiR at an MOI∼0.8 were fixed at 18 h post infection (h p.i.). Immunostaining with antiserum against CA was performed under conditions that allow for immunodetection of cytosolic and nuclear complexes (Muller et al., 2021) to validate that detected SiR signals corresponded to HIV-1 particles. Labeled particles could be visualized by spinning disc confocal microscopy (SDCM) in the cellular environment (Figure S2). Confocal images revealed punctate SiR signals in the cytosol, close to the nuclear envelope and within the nucleus of infected cells. Co-localization with CA(IF) staining confirmed that these signals represented entering viral structures (Figure 3a and Figure S3).
Next, TZM-bl cells infected with HIV-1* or HIV-1*CA14SiR were fixed and analyzed for the presence of click labeled subviral particles inside the nucleus at different time points after infection. Consistent with earlier results (Zurnic 2020, Burdick 2020, Müller 2021), we observed nuclear CA(IF) positive foci in HIV-1* infected cells as early as 6 h post infection (Figure 3b, black), while such signals were absent in noninfected cells (Figure 3b, grey). Importantly, we detected SiR positive complexes in the nucleus of HIV-1*CA14SiR infected cells, with the vast majority also positive for CA(IF) (Figure 3b, magenta). Nuclear entry appeared to be delayed for HIV-1*CA14SiR compared to HIV-1* by approximately 12 h. Nevertheless, comparable numbers of cells with detectable capsid-like objects in the nucleus and the number of objects per cell were reached between 12 and 18 h p.i. (Figure 3b and c). At 12 h p.i., HIV-1*CA14SiR reached the highest number of nuclear particles per cell, with an average of 4.58 ± 4.12, similar to HIV-1* with 5.91 ± 4.11.
Delayed detection of subviral complexes in the nucleus may be due to slower uptake, slower trafficking towards the nuclear envelope, delayed passage through the NPC, or a combination thereof. In order to distinguish between these possibilities, we extended the time-resolved quantification to objects in close vicinity to the nuclear envelope (Figure 3d). This analysis revealed that the HIV-1*CA14SiR derived subviral structures reached the nuclear envelope with similar kinetics to HIV-1* particles (Figure 3d, NE assoc.). A comparable average proportion of CA containing objects was detected at the nuclear envelope in both cases at 6 h, while the numbers of nuclear capsids were lower for HIV-1*CA14SiR at that time (Figure 3d, Nucleus). In contrast, the highest proportion of HIV-1*CA14SiR nuclear objects with 4.20 ± 1.80% was detected at 18 h p.i., while HIV-1* reached similar levels already at 6 h p.i. We conclude that uptake and intracellular trafficking of HIV-1*CA14SiR complexes occurs with similar efficiency as for the wt virus, but transport into the nucleus is slower, offering a possible explanation for the slightly reduced infectivity of HIV-1*CA14SiR virions. This implies that the mechanistic action of the capsid in nuclear import underlies tight margins with respect to its biophysical properties.
Characterization of nuclear CASiR containing complexes
A long-standing question in the field of HIV-1 early replication is the question of when and where capsid uncoating takes place. The possibility to directly detect CA molecules clicked to a synthetic fluorophore enabled us to assess the amounts of CA associated with subviral complexes at different intracellular sites, without the influence of differential epitope accessibility or of a tag domain that potentially confers different properties to a subpopulation of CA molecules. Nevertheless, comparing labeling intensities for nuclear, cytoplasmic, and extracellular particle-associated structures may be confounded in diffraction-limited microscopy by the failure to resolve closely adjacent individual capsids. Clusters of nuclear capsids had indeed been observed by CLEM analyses in our previous study (Muller et al., 2021).
To determine whether nuclear cluster formation occurred under our conditions, we exploited the fact that the chemical dye conjugated to the capsid surface renders the modified virus suitable for super-resolution microscopy. With a lateral resolution of <50 nm, STED nanoscopy allows visual separation of closely adjacent CA objects. TZM-bl cells were infected with HIV-1*CA14SiR at two different MOIs. An MOI of ∼0.8 corresponded to the conditions generally used in our experiments; a 10-fold higher virus dose (MOI ∼8) was applied in a parallel experiment to potentially enhance capsid clustering. At 18 h p.i., cells were fixed, immunostained against CA, and imaged using a STED system in confocal and STED mode (Figure 4). Nuclear CA(IF)/(SiR) double-positive objects were detected under both conditions (Figure 4a, arrowheads). While these objects appeared as individual punctae in diffraction-limited micrographs from the IF and SiR channels at both MOIs (Figure 4b, top and middle row), imaging of the SiR channel in STED mode revealed differences between individual punctae. Some diffraction-limited punctae in the nucleus represented individual capsid-like objects when imaged by STED (Figure 4b, left panel, bottom row). In contrast, other punctae were resolved into small clusters of 2-4 closely apposed CA-containing objects by super-resolution microscopy (Figure 4b, right, bottom panel), consistent with observations made by electron tomography (Muller et al., 2021; Zila et al., 2021). A quantitative analysis of cluster sizes (Figure 4c) revealed that the propensity for capsid clustering in the nucleus correlated with the amount of virus used for infection: at an MOI∼0.8, the vast majority of punctae (∼88%) corresponded to individual capsid-like objects in the nucleus, and clusters of more than two objects were not observed. On the other hand, almost half of the nuclear punctae (∼43%) corresponded to clusters of 2-4 objects when cells were infected with the high MOI∼8. We conclude that nuclear capsid clustering is rarely observed at the MOI of 0.8 used throughout this study. The previously observed capsid clustering in distinct nuclear positions appears to occur preferentially at high MOI.
We next proceeded to SiR fluorescence intensity measurements, comparing the signal intensity of extranuclear HIV-1 particles to that of subviral structures in the nucleus. Staining of the plasma membrane with mCling before infection revealed that under our conditions most cell-associated particles in the cytosolic region represented virions present in endosomes, corresponding to a pre-fusion state of the virus (Figure S4). To ensure that these extranuclear punctae represented single objects, cytoplasmic foci were analyzed in STED mode. We found that ∼95% (n=79) of analyzed objects corresponded to an individual object, while only ∼5% (n=4) of these foci were resolved into two objects by nanoscopy (Figure S5). As illustrated by the cartoon in Figure 5a, complete virions comprise on average ∼2,400 CA molecules, while only ∼1,200-1,500 of these are part of the mature fullerene capsid (Briggs 2003, Carlson 2008, Lanman 2004) that represents a post-fusion state. Assuming equal click labeling efficiency of CA14ncAA for molecules that are part of the mature lattice and those that remain free in the viral volume, the average SiR intensity of complete capsids would be expected to correspond to ∼60% of the average intensity of complete virions from the same preparation. We infected TZM-bl cells at an MOI of 0.8 and quantified the SiR intensity of >6,000 virions attached to the cell or in the cytosolic region and of >100 nuclear punctae. The average SiR intensity of cell-attached and (mostly) endosomal particles in the cytosolic region exhibited an average of 17,649 a.u.. In contrast, the SiR intensity of nuclear subviral structures averaged 9,835 a.u. (Figure 5b), i.e., ∼56% of the average intensity of complete virions, in line with the predicted relative CA content of the mature capsid. Based on these findings, we conclude that the CA(SiR) containing objects in the nuclei of these cells correspond approximately to a full complement of the mature capsid.
The small molecule inhibitor PF74 (Blair et al. 2010) binds to the HIV-1 capsid in a pocket overlapping the binding sites for the FG motifs of various nucleoporins and for the nuclear host protein CPSF6; the compound inhibits HIV-1 replication by multiple mechanisms (reviewed in (Thenin-Houssier and Valente, 2016)). Treatment with high concentrations of PF74 has been reported to destabilize the capsid ((Blair et al., 2010; Price et al., 2014; Shi et al., 2011), but data obtained using CA(IF) detection argued against a PF74 induced loss of CA from nuclear complexes (Müller 2021). Since we cannot exclude that results obtained by immunodetection are influenced by differential CA epitope exposure, we re-addressed this issue employing direct CA labeling. TZM-bl cells were infected with HIV-1*CA14SiR particles for 17 h and treated with 15 µM PF74 or DMSO for 1 h, followed by fixation, permeabilization, methanol extraction, and SDCM imaging. As shown in Figure 5c, CPSF6 was removed from the subviral complexes, in accordance with earlier results (Muller et al., 2021). In contrast, mean CA(SiR) intensity remained unaltered, indicating that the capsid remains largely stable under these conditions.
Detection of directly labeled HIV-1 capsids in primary cells
To validate our results in a physiologically relevant cell type, primary human CD4+ T cells from healthy blood donors were infected, subjected to IF staining against CA, and imaged by SDCM at 24 h p.i. (Figure 6a and Figure S6). We readily detected nuclear subviral SiR positive structures in HIV-1*CA14SiR infected cells, indicating that nuclear replication complexes retained CA also in these primary cells (Figure 6a). Consistent with prior observations made in T cells (Zila et al., 2019) the majority of SiR-positive objects were not associated with CA(IF) signals (9/11 particles; Figure 6a, left) when fixation and immunostaining were performed under standard conditions. As outlined above, treatment with 15 µM PF74 for 1 h dissociates the large clusters of CPSF6 from nuclear subviral complexes. We observed that this in turn renders nuclear CA accessible for IF detection in T cells, presumably by exposure of CA epitopes upon CPSF6 displacement (Muller et al., 2021). Accordingly, brief PF74 treatment allowed for detection of CA(IF) signals co-localizing with nuclear CA(SiR) punctae (13/16; Figure 6a, right). We conclude that the direct CA labeling strategy presented here overcomes technical artifacts that hamper IF analyses.
Further quantitative analyses using primary CD4+ T cells prepared from six blood donors revealed similar numbers of nuclear capsid structures in cells infected with HIV-1*CA14SiR than in cells infected with HIV-1* at 24 h p.i. (Figure 6b). SiR intensity measurements were only performed for intranuclear objects in this case since high background due to SiR accumulation in the narrow cytoplasm of T cells precluded reliable analysis of individual particles in the extranuclear region (see Figure 6a and Figure S6). Quantitation of SiR intensities of nuclear punctae in cells infected with an MOI∼0.8 yielded similar average intensities as measured in TZM-bl cells (mean=12,485 a.u.), indicating the presence of a complete or nearly complete mature capsid in the nuclear complexes in primary T cells (Figure 6c). Cells infected with an MOI of ∼8 displayed higher CA(SiR) intensities of diffraction-limited nuclear objects (mean=39,502 a.u.), suggesting intranuclear clustering of capsids, as observed in TZM-bl cells (Figure 4).
Our findings from CA(SiR) intensity measurements argue for the presence of a full capsid complement at subviral structures in the nucleus. These data strengthen conclusions from several recent studies suggesting that the mature capsid lattice may be completely or largely intact on nuclear subviral objects (Burdick et al., 2020; Muller et al., 2021; Zila et al., 2021). However, fluorescence signals do not yield information on the architecture of nuclear CA14SiR containing objects. Therefore, we complemented our analyses by performing CLEM of infected SupT1 T cells. In order to maximize the number of nuclear objects, infection was synchronized by the attachment of particles to the cells for 3 h at a low temperature (16°C) to prevent particle uptake by membrane fusion or endocytosis ((Melikyan et al., 2000; Weigel and Oka, 1981). Virus entry was initiated by temperature shift to 37°C. At 24 h post temperature shift, specimens were prepared by high-pressure freezing (HPF) and freeze substitution, and 250 nm thick resin sections were subjected to SDCM in order to localize CA(SiR) containing structures, followed by correlative electron tomography (CLEM-ET) analysis. CA(SiR) positive objects could be identified by SDCM in the sections (Figure 6d), demonstrating that the brightness of signals derived from direct CA(SiR) labeling is sufficient for CLEM detection of cytosolic and nuclear (sub)viral structures. ROIs were defined based on the SiR signals and subjected to correlative ET analysis. Figure 6e shows an exemplary tomogram obtained from a ROI located within the nucleus. It reveals several closely attached electron-dense structures at the position of the SiR label, whose shape and dimension match those of intact or largely intact mature HIV-1 capsids (Figure 6f and supplementary movie 1). Such structures were recently identified in nuclei of infected cells by CLEM using fluorescently labeled HIV-1 IN as an indirect marker for subviral structures (Muller et al., 2021; Zila et al., 2021) and were interpreted as capsid shells based on their morphology. Here we demonstrate that such structures co-localize with nuclear foci comprising a high number of click labeled CA molecules, thereby providing direct evidence that the cone-shaped objects are HIV-1 capsids that have entered the nucleus of infected cells.
Discussion
Here we present a direct labeling approach for the HIV-1 CA protein that yields infectious and morphologically mature viral particles. The minimally invasive GCE/click labeling approach used here represents an ideal strategy for the versatile labeling of genetically fragile viral capsid proteins in principle, but its potential for virus imaging has not been fully exploited. GCE has previously been explored to generate conditionally replication-competent live attenuated viruses as vaccine candidates (Si et al., 2016; Yuan et al., 2017). The combination of GCE and subsequent functionalization of a viral capsid protein by click chemistry has so far only been applied to the non-enveloped adeno-associated virus (AAV) (e.g., (Kelemen et al., 2018; Zhang et al., 2018). However, the capsid of AAV, unlike HIV-1 CA, can tolerate peptide insertions and larger modifications (e.g.,(Borner et al., 2020; Chandran et al., 2017; Feiner et al., 2019; Varadi et al., 2012). Here, we demonstrate that GCE in conjunction with click labeling can also be applied to an enveloped virus with a highly multifunctional and extremely genetically fragile capsid protein. All previously described genetic tagging strategies for HIV-1 CA (Burdick et al., 2020; Campbell et al., 2008; Pereira et al., 2011; Zurnic Bonisch et al., 2020) require complementation with a molar excess of wt protein or virus. Since the mature HIV-1 capsid is assembled from less than half of the ∼2,500 CA molecules packaged in the virion (Briggs et al., 2004; Lanman et al., 2004), it cannot be ascertained in this case whether the subset of genetically tagged CA molecules is an integral part of the mature capsid lattice. In contrast, we found that the strategy described here allowed genetic labeling of HIV-1 CA in the proviral context while retaining infectivity in the absence of complementation.
The detection of a label covalently attached to CA is independent of cellular context, sample treatment, or exposure of CA epitopes. Thereby, the method overcomes limitations of IF detection that had previously resulted in different conclusions regarding the presence of CA on subviral complexes. The use of synthetic dye molecules also renders the labeling strategy compatible with a wide range of fluorescence imaging approaches, including live-cell microscopy, correlative imaging and super-resolution fluorescence microscopy techniques (Wang et al., 2019).
Our approach allowed for direct, quantitative analysis of containing objects and CA amounts associated with viral complexes in microscopic images of infected cells. While time-lapse experiments showed some delay in nuclear import kinetics for labeled capsid-like objects, the infectivity of highly labeled preparations was reduced by only twofold, and the number of nuclear objects reached was similar to that detected in cells infected with wt virus. Thus, site-specific introduction of a synthetic fluorophore can be compatible with capsid functionality in HIV-1 post-entry processes. CA amounts approximately corresponding to a full complement of a mature capsid were found to be associated with subviral complexes in nuclei of a HeLa-derived cell line and primary human CD4+ T cells, also upon inhibition of cell division by aphidicolin treatment. By applying correlative imaging, we provide direct evidence that nuclear capsid-shaped objects, as recently detected by correlative ET before and after separation of the viral genome from the bulk of viral proteins (Muller et al., 2021; Zila et al., 2021) indeed represent HIV-1 capsids or capsid-like remnants. Taken together, these results argue against (partial) capsid uncoating prior to entering the nucleoplasm, as had been concluded earlier based on low or lacking CA IF signals associated with nuclear subviral complexes in certain cell types (e.g. (Burdick et al., 2017; Hulme et al., 2015; Peng et al., 2014; Zila et al., 2019), or based on the loss of the fluorescently labeled capsid binding protein CypA at the nuclear envelope ((Francis et al., 2016); Francis 2018; Francis and Melikyan 2020). The apparent discrepancy between these previous IF results and data from direct CA quantification may be explained by differential accessibility of capsid epitopes under different IF conditions. The indirect label CypA, on the other hand, might be displaced from capsids at the nuclear pore, possibly by competition between fluorescent CypA and the outer NPC protein Nup358, which also carries a binding site for the CypA binding loop of CA (Schaller et al., 2011). Our data suggest nuclear capsid uncoating in a model cell line, as well as in primary T cells, in agreement with recent findings from us and others, which indicated that the nuclear pore channel is wider than assumed earlier, allowing HIV-1 capsids to pass the intact NPC (Zila et al., 2021), and that HIV-1 uncoating occurs after nuclear import (Burdick et al., 2020; Dharan et al., 2020; Li et al., 2021; Selyutina et al., 2020), apparently by separation of the viral genome from a broken capsid remnant (Muller et al., 2021).
Small clusters of CA positive objects were detected by STED nanoscopy in nuclei of TZM-bl cells, consistent with the reported detection of nuclear clusters containing multiple HIV-1 replication complexes (Francis et al., 2020b), multiple viral genomes (Rensen et al., 2021), or even several intact or partly intact capsid-like structures (Muller et al., 2021) in various cell types. Our analyses revealed that the observed clustering is dependent on the amount of virus used for infection. Most nuclear signals represented single capsids at a lower MOI, whereas frequent clustering was observed at high MOI. This observation suggests that capsids enter the nucleus individually, but traffic via a limited number of routes and accumulate at defined sites of uncoating. This raises the question whether HIV-1 capsids use a ‘specialized’ subset of nuclear pores for nuclear entry; the answer would not only be relevant in the context of HIV-1 replication, but also with respect to an understanding of the nuclear import process. Intracellular Nup levels and presumably NPC composition have been reported to influence HIV-1 replication (Kane et al., 2018), but compositional and structural variability of NPCs between different cell types, or within an individual cell, is incompletely understood (reviewed in (Knockenhauer and Schwartz, 2016). The route, mechanism and functional consequences of intranuclear trafficking of HIV-1 complexes also warrant further analysis. Growing evidence from recent studies suggests that incoming viral replication complexes accumulate at nuclear speckles in a CA and CPSF6-dependent manner, and that reverse transcription may only be completed near the site of integration (Burdick et al., 2020; Francis et al., 2020a; Rensen et al., 2021; Selyutina et al., 2020). Combining the direct CA labeling described here with the recently developed fluorescence detection of the reverse transcribed genome (Blanco-Rodriguez et al., 2020; Muller et al., 2021) will provide us with the possibility to study the uncoating process in more detail using a combination of confocal imaging, nanoscopy, and correlative imaging.
The direct labeling approach also allowed us to investigate the effect of the capsid inhibitor PF74 (Blair et al., 2010), whose detailed mode of action is still under investigation, on nuclear capsids. We found that displacement of CPSF6 from nuclear subviral structures was not accompanied by a loss of CA signal. This finding disagrees with the recently reported rapid CA dissociation from nuclear complexes upon PF74 addition, that was based on imaging of HIV-1 particles containing eGFP-CA complemented by a molar excess of wt CA (Burdick et al., 2020). The apparent discrepancy may suggest that the subset of eGFP-tagged CA molecules is not an integral part of the mature capsid lattice, resulting in premature loss of the labeled molecules. Our findings are in line with the observation that PF74 treatment does not lead to a loss of CA IF signal on nuclear complexes but rather enhances immunostaining efficiency ((Francis et al., 2020b; Muller et al., 2021)) and with in vitro findings that indicate breakage of lattice integrity by PF74, but stabilization of the remaining lattice (Marquez et al., 2019; Rankovic et al., 2018).
In conclusion, direct click labeling of HIV-1 CA is a versatile approach that substantially expands the possibilities to study the early events in HIV-1 replication with high temporal and/or spatial resolution using advanced fluorescence microscopy methods. Application for the HIV-1 CA provided direct proof that the capsid stays largely intact upon passage of the subviral complex into the nucleus and directly identified nuclear capsid-like structures that morphologically resembled the virion capsid by CLEM-ET. The fact that the combination of GCE and click chemistry could successfully be applied to a notoriously genetically fragile capsid protein of an enveloped virus opens the perspective that this strategy may also advance and expand fluorescence labeling of a broad range of other viruses.
Materials and Methods
Plasmids
Plasmids were cloned using standard molecular biology techniques and verified by commercial Sanger sequencing (Eurofins Genomics). PCR was performed using Q5 High-Fidelity DNA Polymerase (New England Biolabs) or Phusion DNA Polymerase (New England Biolabs) according to the manufacturer’s instructions using primers purchased from Eurofins Genomics. Plasmid amplification was carried out in E. coli Stbl2 (Thermo Fisher Scientific) cells.
HIV-1 plasmids were based on the proviral plasmid pNLC4-3 (Bohne and Krausslich, 2004) that expresses the authentic genomic RNA from HIV-1NL4-3 (Adachi et al., 1986) under the control of the cytomegalovirus promoter. To avoid unwanted ncAA incorporation into the virion component Vpr, the amber stop codon of the vpr ORF of pNLC4-3 was mutated into an opal stop codon (TGA) via site-directed mutagenesis. See primer list for sequences of primers used. PCR1 (primers VprTGA a and VprTGA b) and PCR2 (primers VprTGA c and VprTGA d) were performed in parallel to generate two overlapping single stranded PCR products. Using a combination of both products of these reactions as new templates, PCR3 with primers VprTGA a and VprTGA d resulted in PCR fragments comprising the respective mutation. These fragments were subcloned into pNLC4-3 using unique PflMI/NheI restriction sites, resulting in pNLC4-3* (HIV-1*).
To allow for site-specific GCE the codon for amino acid A14 of CA was mutated into TAG via overlap PCR. PCR1 (primers CA14BssHII fwd 1, CA14TAG rev 1) and PCR2 (primers CA14TAG fwd 2, CA14ApaI rev 2) were performed in parallel to generate two overlapping single-stranded PCR products. PCR3 with primers CA14BssHII fwd 1 and CA14ApaI rev 2 result in the PCR fragment comprising the mutation, which was subcloned into pNLC4-3* using unique BssHII/ApaI restriction sites, resulting in pNLC4-3*CA14TAG (HIV-1*CA14TAG). Plasmid pNESPylRS-eRF1dn-tRNA (Schifferdecker, Sakin et al., in preparation) is based on pEA168 ((Cohen and Arbely, 2016), kindly provided by Eyal Arbely, Ben-Gurion University of the Negev, Israel), a eukaryotic vector that comprises expression cassettes for two proteins and four tRNA molecules. The coding sequence for a modified pyrrolysine tRNA synthetase was PCR amplified from plasmid tRNAPyl/NESPylRSAF (Nikic et al., 2016) and cloned into a CMV promoter driven cassette in pEA168 using HindIII/XbaI restriction sites, resulting in plasmid pEA168-CMV-aaRS-4xU6tRNA. A PCR fragment encoding a dominant version of the eukaryotic release factor 1 (eRF1(E55D)) amplified from plasmid peRF1-E55D (Schmied et al., 2014) was subsequently inserted into an expression cassette driven by the EF1 promotor into pEA168-CMV-aaRS-4xU6tRNA using KpnI/MluI restriction sites, yielding pNESPylRS-eRF1dn-tRNA.
Cell culture
HEK293T (Pear et al., 1993) and HeLa TZM-bl indicator cells (Wei et al., 2002) were maintained in Dulbecco’s Modified Eagle’s medium (Thermo Fisher Scientific) supplemented with 100 U/ml penicillin, 100 μg/ml streptomycin (PAN Biotech, GER) and 10% fetal calf serum (FCS, Sigma Aldrich, USA). Both cell lines were regularly monitored for mycoplasma contamination using the MycoAlert mycoplasma detection kit (Lonza Rockland, USA). Primary CD4+ T cells were cultured in RPMI 1640 containing L-glutamine supplemented with 100 U/ml penicillin, 100 μg/ml streptomycin (PAN Biotech), 10% heat-inactivated FCS, and 5% human AB serum (Sigma Aldrich).
Isolation of primary cells
Primary human CD4+ T cells were isolated from buffy coats obtained from healthy and anonymous blood donors at the Heidelberg University Hospital Blood Bank following the regulations of the local ethics committee. CD4+ T cells were isolated using EasySep™ Direct Human T Cell Isolation Kit (Stemcell technologies, GER) according to the manufacturer’s instructions and activated by incubation in the presence of 100 U/ml IL-2 (Sigma Aldrich) and T Cell TransAct™ human (Miltenyi Biotec, GER) for 72 h.
Virus particle production
HEK293T cells were seeded in T175 tissue culture flasks the day before (∼15 Mio. cells) and transfected using calcium phosphate precipitation according to standard procedures (∼80 % confluency). Cells were co-transfected with 50 µg / flask total DNA of pNLC4-3* (HIV-1*) or pNLC4-3*CA14TAG (HIV-1*CA14TAG) and plasmid pNESPylRS-eRF1dn-tRNA in a molar ratio of 2.22:1. At 6 h p.t., medium was removed, and fresh complete DMEM containing a final concentration of 500 µM CpK (SiChem; stock solution of 100 mM was pre-diluted 1:4 in 1M HEPES shortly before use), and 100 µM ascorbic acid (Sigma Aldrich; stock solution 10 mM) was added. At 48 h p.t. the tissue culture supernatant was harvested and filtered through 0.45 µm nitrocellulose filters. For labeling the CA protein, 250 nM Tetrazine-SiR (Spirochrome; stock solution 1 mM) was added to the filtered supernatant, and samples were incubated at 37°C for 30 min. Particles were then concentrated by ultracentrifugation through a 20% (w/v) sucrose cushion at 28,000 rpm using a Beckman TLA-100 fixed angle-rotor (Beckman Coulter, GER) for 90 min at 4°C. Pellets were gently resuspended in phosphate-buffered saline (PBS) containing 10% FCS and 10 mM HEPES (pH 7.5) and stored in 5 µl aliquots at −80°C.
Immunoblotting and In-gel fluorescence
Virus samples were mixed 1:10 with SDS sample buffer (150 mM Tris HCl, pH 6.8, 6% (w/v) SDS, 30% Glycerin, 0.06% bromophenol blue, 20% β -Mercaptoethanol) and boiled at 95°C for 15 min. 10 µl HIV-1* and 40 µl HIV-1*CASiR lysates were subjected to SDS-PAGE (15 %; acrylamide:bisacrylamide 200:1). Cell lysates were generated from transfected HEK293T cells. At 40 h p.t. cells were washed with PBS, trypsinized and resuspended in PBS. 1 ml of cell suspension was mixed with 300 µl SDS sample buffer and boiled at 95°C for 15 min. 10 µl cell lysate was subjected to SDS-PAGE. Proteins were transferred to a nitrocellulose membrane (Millipore) by semi-dry blotting for 1 h at 0.8 mA/cm2. Viral antigens were stained with the indicated antisera in PBS/0.5% bovine serum albumin (BSA) (sheep αCA, polyclonal 1:5 (in-house); rabbit αMA, polyclonal 1:1,000 (in-house); rabbit αRT, polyclonal, 1:1,000 (in-house), mouse αlaminA/C, monoclonal antibody 1:100 (Cat# sc-7292, Santa Cruz Biotechnology), mouse αlaminB1, monoclonal 1:100 (Cat# sc-365962, Santa Cruz Biotechnology)) followed by staining with corresponding secondary antibodies IRDye™ in PBS/0.5% BSA (anti-sheep 680CW (1;10,000); Rockland, or anti-rabbit 800CW (1:10,000); Li-COR Biosciences). Detection was performed using a Li-COR Odyssey CLx infrared scanner (Li-COR Biosciences) according to manufacturer’s instructions. CA quantification was performed with ImageStudio LITE software (Li-COR Biosciences) via intensity measurements of CA bands and a serial dilution of recombinant purified CA standard (2.5 ng/µl; in-house) on the same membrane. For in-gel fluorescence, the acrylamide gels were directly scanned using a Li-COR Odyssey CLx infrared scanner (Li-COR Biosciences) set at an emission wavelength of 700 nm.
Infectivity assay
Virus amounts were quantified via SYBR Green based Product Enhanced Reverse Transcription assay (SG-PERT; (Pizzato et al., 2009)). To determine the effect of incorporating CpK and Tet-SiR labeling on virus infectivity, HIV-1* and HIV-1*CA14SiR viral particles (normalized by RT activity) were titrated on TZM-bl cells seeded in 15-well ibidi µ-Slide angiogenesis dishes. At 6 h p.i. 50 µM T-20 (Enfuvirtide; Roche, GER; stock solution 20 mM) was added to prevent second-round infection. Infection rates were scored at 48 h p.i.. For this, cells were fixed in 4% paraformaldehyde (PFA; Electron Microscopy Sciences, USA; stock solution 16%) for 15 min, followed by 20 min incubation in PBS/0.5% (v/v) Triton X-100 at room temperature. Immunostaining was performed using an in-house polyclonal rabbit antiserum raised against recombinant HIV-1 MA (1:1000) in PBS/0.5% BSA) 1 h at room temperature. Secondary antibody Alexa Fluor 488 donkey anti-rabbit (1:1,000; Thermo Fisher Scientific) in PBS/0.5% BSA was added for 45 min at room temperature. Samples were imaged by SDCM. The mean intensity of the 488 channel (MA(IF)) was quantified in from the non-infected samples imaged in parallel and subtracted as background in each image. The proportion of IF-positive cells was counted in 12 randomly selected fields of view using Fiji (Schindelin et al., 2012). To determine the infectivity of virus particle preparations, the number of infected cells per well was calculated by multiplying the percentage of infected cells detected with the number of cells per well (double of seeded cell number the day before). Division by the volume of virus suspension used for infection yielded the number of infectious units (IU) / ml.
Fixation and immunofluorescence staining of infected cells
3.33 × 103 TZM-bl cells were seeded into 15-well µ-Slides angiogenesis dishes (ibidi, GER; cat. 81507) the day before infection. Infection at 37°C was performed with an MOI ∼0.8 for 6,9,12 or 18 h. Subsequently, cells were incubated for 1 h with 15 µM PF74 (Sigma Aldrich; stock solution 10 mM in DMSO) in DMEM to allow for efficient detection of nuclear CA by IF (Muller et al., 2021). Samples were washed with PBS, fixed in 4% PFA for 15 min and permeabilized with PBS/0.5% (v/v) Triton-X100 for 20 min, and washed again with PBS. Cells were extracted using ice-cold 100% methanol for 10 min. Afterward, samples were blocked with PBS/2.5% BSA for 15 min, followed by incubation with primary antibodies in PBS/0.5% BSA for 1 h at room temperature. After washing three times with PBS, secondary antibodies diluted in PBS/0.5% BSA were added for 45 min at room temperature. Samples were washed and stored in PBS at 4°C. For infection of primary CD4+ T cells, 20,000 cells were infected with HIV-1* or HIV-1*CA14SiR in a 96-well v-bottom microplate (Greiner Bio-one, cat. #650161) in a volume of 40 µl RPMI and transferred at 22 h p.i. onto a PEI-coated 15-well µ-Slide angiogenesis dishes (ibidi). Cells were allowed to adhere for 1 h at 37°C, and PF74 diluted in fresh growth medium was added to a final concentration of 15 µM. Extraction, fixation, and immunostaining were performed after 1 h at 37°C as described above. For the detection of endosome-associated particles, 2 µM mCLING ATTO488 (Synaptic Systems; stock 50 µM) was added to TZM-bl cells seeded in 15-well µ-Slides Angiogenesis and incubated at 16°C for 30 min. Subsequently, the fluorescent probe was removed, HIV-1*CA14SiR particles were added in fresh growth medium, and cells were incubated for an additional 3 h at 37°C (MOI∼0.8). Cells were fixed for 90 min at room temperature in 4% PFA and 0.2% glutaraldehyde to ensure retention of mCLING at cellular membranes. Nuclei were stained with 5 µg/ml Hoechst (Merck) in PBS for 30 min.
Cell viability assay
To test the effect of mCLING ATTO488 (Synaptic Systems) staining on cell viability, TZM-bl cells were seeded into a 96-well plate (9×103 cells/well; flat bottom Greiner Bio-one) the day before and incubated in medium supplemented with the indicated concentration of mCLING ATTO488 for 30 min at 16°C. After staining, cells were trypsinized, stained with Trypan blue using standard procedures and analyzed with a TC20™ Automated Cell Counter (BioRad).
Labelling efficiency of immobilized particles
15-well µ-Slide angiogenesis dishes (ibidi) were coated with 30µl/well polyethyleneimine (PEI; 1mg/ml) for 30 min at room temperature and washed with PBS. Pre-labeled HIV-1* and HIV-1*CA14SiR particles were incubated in PBS on PEI-coated microscopy slides for 1 h at 37°C. Subsequently, samples were washed with PBS, fixed in 4% PFA for 15 min and permeabilized with PBS/0.05% (v/v) Triton X-100 for 20 min at room temperature. Immobilized particles were blocked with PBS/2.5% BSA for 15 min and polyclonal rabbit antiserum raised against recombinant HIV-1 CA protein (in-house) was added (1:1000 in PBS/0.5% BSA for 1 h at room temperature). After washing three times with PBS, secondary antibody Alexa Fluor 488 donkey anti-rabbit (Thermo Fisher Scientific) 1:1000 in PBS/0.5% BSA was added for 45 min at room temperature. Samples were washed and stored in PBS at 4°C.
Confocal microscopy (SDCM)
Multichannel z-series with a z-spacing of 200 nm, spanning the whole cell volume (3D), were acquired using a PerkinElmer Ultra VIEW VoX 3D spinning disk confocal microscope (SDCM; Perkin Elmer). A 60x oil immersion objective (numeric aperture [NA] 1.49; Perkin Elmer) was used for imaging of TZM-bl cells or 100x oil immersion objective ([NA] 1.49; Perkin Elmer) for primary CD4+ T cells and immobilized particles. Images were recorded in the 405-, 488-, 561-, and 640 nm channels.
STED microscopy
STED nanoscopy was performed using a λ = 775 nm STED system (Abberior Instruments GmbH) equipped with a 100x oil immersion objective (NA 1.4; Olympus UPlanSApo). STED images were acquired using the 640 nm excitation laser lines while the 488 and 590 laser line was acquired in confocal mode only. Nominal STED laser power was set to 20% of the maximal power (1250 mW) with pixel dwell time of 10 µs and 15 nm pixel size. STED images were deconvolved using the software Imspector (Abberior Instruments GmbH) and Huygens Professional Deconvolution (Scientific Volume Imaging).
Electron microscopy
HEK293T cells (4×105) were seeded in a glass coverslip-bottom petri dish (MatTek, MA, USA), cultured for 16 h at 37°C and then co-transfected with pNLC4-3*CA14TAG and pNESPylRS-eRF1dn-tRNA by using calcium phosphate precipitation. At 6 h p.t., medium was removed and fresh complete DMEM containing a final concentration of 500 µM CpK (SiChem; stock solution 100 mM was pre-diluted 1:4 in 1M HEPES shortly before use), and 100 µM ascorbic acid (Sigma Aldrich; stock solution 10 mM) was added. At 44 h p.t., cells were fixed with pre-warmed 2% formaldehyde + 2.5% glutaraldehyde in 0.1 M cacodylate buffer (pH 7.4) for 1.5 h at room temperature, then washed in 0.1 M cacodylate buffer and post-fixed with 2% osmium tetroxide (Electron Microscopy Sciences) for a 1 h on ice. Cells were subsequently dehydrated through an increasing cold ethanol series (30, 50, 70, 80, 90, and 100%; on ice) and two anhydrous acetone series (at room temperature). The coverslip with cells was then removed from the dish, and cells were flat embedded in Epon resin. 70-nm thin sections were cut with an ultramicrotome (Leica EM UC6), collected on formvar-coated 100-mesh copper EM grids (Electron Microscopy Sciences) and stained with a 3% uranyl acetate in 70% MetOH (10 min), and lead citrate (7 min). Cells sections were observed with a JEOL JEM-1400 electron microscope operating at 80 kV (Jeol Ltd., JPN), equipped with a bottom-mounted 4K by 4K pixel digital camera (TemCam F416; TVIPS GmbH, GER).
CLEM and electron tomography
SupT1 cells were distributed in a 96-well plate (2×105 cells/well; U-bottom; Greiner Bio-one, 650180) and pre-incubated for 16 h with 1 µm aphidicolin (APC; Merck). Cells were pelleted (200 x g, 3 min) and resuspended in complete RPMI medium containing HIV-1*CA14SIR particles (MOI∼0.4). Cells were incubated with viral particles for 120 min at 16°C to adsorb the virus and synchronize virus entry. Samples were then processed for CLEM and ET as described previously (Zila et al., 2021). In brief, cells were transferred to glass-bottomed ‘microwell’ of a MatTek dish (MatTek, USA) containing carbon-coated and retronectin-coated sapphire discs (Engineering Office M. Wohlwend, SUI). Samples were high pressure frozen, and sapphire discs were then transferred from liquid nitrogen to the freeze-substitution (FS) medium (0.1% uranyl acetate, 2.3% methanol and 2% H2O in acetone) tempered at −90°C. Samples were FS-processed and embedded in Lowicryl HM20 resin (Polysciences, USA) according to a modified protocol of Kukulski et al. (Kukulski et al., 2011). For CLEM-ET, thick resin sections (250 nm) were cut and placed on a slot (1 × 2 mm) EM copper grids covered with a formvar film (Electron Microscopy Sciences, FF2010-Cu). Grids were decorated with fiducial marker and stained with Hoechst to visualize nuclear regions.
Light microscopy Z stacks of sections were acquired by PerkinElmer UltraVIEW VoX 3D Spinning-disc Confocal Microscope (Perkin Elmer) using a 100 × oil immersion objective (NA 1.49; Nikon), with a z-spacing of 200 nm and excitation with the 405-, 488-, 561- and 633-nm laser line. Acquired z stacks were visually examined using Fiji software (Schindelin et al., 2012) and intracellular CA(SiR) positive signals were identified. EM grids were decorated with 15 nm protein-A gold particles for tomogram alignment and stained with uranyl acetate and lead citrate. Grids were loaded to a Tecnai TF20 (FEI) electron microscope (operated at 200 kV) equipped with a field emission gun and a 4K by 4K pixel Eagle CCD camera (FEI). Positions of CA(SiR) signals were pre-correlated with imported SDCM images in SerialEM as described previously (Schorb et al., 2017). Single-axis electron tomograms were carried out. Tomographic tilt ranges were typically from –60° to 60° with an angular increment of 1°. The pixel size was 1.13 nm. Alignments and 3D reconstructions of tomograms were done with IMOD software (Kremer et al., 1996). Post-correlation was performed using eC-CLEM plugin in Icy software (de Chaumont et al., 2012).
Image analysis
Microscopy images were screened and filtered in Fiji/ImageJ (Schindelin et al., 2012) with a mean filter and background subtraction. Infected cells were quantified in Fiji via segmentation and counting of nuclei and the cell counter to manually quantify the number of positive cells. To determine labeling efficiency of click labeled particles, CA(SiR) intensities of detected immobilized particles based on CA(IF) were quantified using the spot detector of the software Icy (de Chaumont et al., 2012). Five ROIs without particles were measured, and mean intensity in the SiR channel was subtracted as background. The threshold was set to t = 1,000 a.u.. CA(IF) detected spots with intensities above the threshold were classified as CA(SiR) positive.
To analyze particle distribution and intensity measurements throughout the entire volume of cells, z-image series were reconstructed in 3D space using Imaris 9.2 software (Bitplane AG). Individual HIV-1 CA(IF) objects were automatically detected using the spot detector Imaris module, which created for each fluorescent signal a 3D ellipsoid object with 300 nm estimated diameter in x-y dimensions and 600 nm in z. The local background of each individual spot was subtracted automatically. Subsequently, the mean signal intensity in the CA(SiR) channel was quantitated within all objects. The threshold for SiR intensity was set to t = 7,000 a.u. and adjusted manually for each image by visual inspection. Spots detected in SiR-clusters were excluded. Nuclear objects were manually identified based on the laminA/C staining. NE-associated objects were classified based on laminA/C intensities. Every image was manually inspected and a threshold for NE-associated objects was set in the range of 6,300-9,100 a.u.. All other particles were classified as PM/cytoplasm (= in the cytoplasm/at plasma membrane).
To identify post-fusion cores by mCLING ATTO488 staining, HIV-1 CA(SiR) positive objects were automatically detected and the mCLING ATTO488 mean signal intensity co-localizing with each object was quantitated. The threshold was set to t = 5,900 a.u. based on the lowest mCLING intensity detected in a T-20 control sample. Particles associated with mCLING intensity above background were classified as endosome associated. Fiji standard ‘greyscale’ lookup table (LUT) was used to visualize single channel images and ‘Fire’ for single channel STED images.
Data visualization and statistical analysis
Statistical significance was assessed using Prism v9.1.0 (GraphPad Software Inc, USA). A two-tailed non-paired Mann-Whitney test (α = 0.05) was used to assess the statistical significance of non-parametric data. Data were plotted using Prism v9.1.0 or the Python statistical data visualization package seaborn v.0.10.0 (Waskom 2020) Graphs show mean/median with error bars as defined in the figure legends.
Competing financial interests
The authors declare no competing financial interests.
Acknowledgements
We acknowledge microscopy support from the Infectious Diseases Imaging Platform (IDIP) at the Center for Integrative Infectious Disease Research, Heidelberg. We thank Eyal Arbely (The National Institute for Biotechnology in the Negev, Beer-Sheva, Israel) for providing pEA168. pNESPylRS-eRF1dn-tRNA was kindly provided by Anna-Lena Schäfer (University Hospital Heidelberg, Heidelberg, Germany).
This work was funded by the Deutsche Forschungsgemeinschaft (DFG; German Research Foundation), Projektnummer 240245660, SFB 1129 project 6 (B.M.) and project 5 (H.G.K.).