Introduction

Already in the 80s, Harald zur Hausen proposed a role of human papillomaviruses in cancer (zur Hausen, 2009). Since then, five more classes of oncogenic viruses have been identified (Galati et al., 2024a). To date, it is assumed that more than 10% of the worldwide human cancer burden is associated with infectious agents (Galati et al., 2024a), from which about a half is caused by Papillomaviridae (Martel et al., 2017) For this reason, the understanding of viral entry strategies has implications going beyond the classical treatment of acute viral infections.

Human papillomaviruses are small, non-enveloped viruses with a diameter of ̴55 nm. The icosahedral capsid is mainly composed of pentameric L1 capsomers. Together with fewer L2 capsid proteins, capsomers surround a histone core bearing a circular double-stranded DNA (Baker et al., 1991; Ozbun and Campos, 2021). From the more than 200 phylogenetically classified HPV genotypes, the most oncogenic ones are HPV16 and HPV18 (Galati et al., 2024b), which are responsible for about 70-80% of the cervical cancer cases (Christiansen et al., 2015). In addition, they cause other severe cancers such as anogenital, head and neck tumors (Doorbar et al., 2012).

Papillomavirus infection requires a break in the epithelial barrier, through which virions reach mitotically active basal cells of the epithelia (Ozbun and Campos, 2021). Here, virions bind to the linear polysaccharide heparan sulfate (HS) that is an important constituent of the extracellular matrix (ECM) and the plasma membrane. In the latter case, it is attached to membrane proteins, so called heparan sulfate proteoglycans (HSPGs). Positively charged and polar residues of the L1 capsid protein interact with negative charges of HS, resulting in a strong bond (Dasgupta et al., 2011; Giroglou et al., 2001; Joyce et al., 1999; Knappe et al., 2007; Surviladze et al., 2015). While in cell culture virions bind to HS of the cell surface and the ECM, it has been suggested that in vivo they bind predominantly to HS of the extracellular basement membrane (Day and Schelhaas, 2014; Kines et al., 2009; Schiller et al., 2010).

Regardless of where the virions bind to HS, they must detach from the linear polysaccharide before they bind to cell surface receptors and are internalized (Ozbun and Campos, 2021). However, the strong electrostatic bonds do not allow for HS-virion dissociation. Two release mechanisms are discussed, that mutually are not exclusive. In one model, binding to HS structurally results in capsid enlargement and softening (Feng et al., 2024), finally resulting in the exposure and cleavage of capsid proteins (Becker et al., 2018), which is required for further steps in infection. In an alternative model, the direct HS-virion bond remains. However, cleavage of HS/HSPGs by proteinases and heparanases produces fragmented HS that, albeit still bound to the virion surface, no longer immobilizes the virion (Surviladze et al., 2012; Surviladze et al., 2015).

Once the virions are released, they could reach the cell surface simply by passive diffusion. In a co-culture trans-well assay, PsVs from donor cells infect spatially separated receiver cells (Surviladze et al., 2012), demonstrating that free diffusion of virions is sufficient for infection. However, an active transport mechanism has been reported as well. PsVs migrate along actin-rich protrusions from the ECM towards the cell body (Schelhaas et al., 2008; Smith et al., 2008).

While it is generally agreed on that HS is a primary virion attachment site, the molecular identity of the receptor complex on the cell surface is unknown. This complex is most likely a multimeric complex rather than a single molecular component. Possible candidates are proteins crucial for cell entry, as the tetraspanin CD151 (Mikuličić et al., 2019; Scheffer et al., 2013; Spoden et al., 2008), integrin-α6 (Itgα6) (Evander et al., 1997; Yoon et al., 2001), growth factor receptors (Mikuličić et al., 2019; Surviladze et al., 2012), and the annexin A2 heterotetramer (Dziduszko and Ozbun, 2013; Woodham et al., 2012). From these molecules, the tetraspanin CD151 could play a coordinating role, as tetraspanins have been suggested to organize viral entry platforms in several types of viral infections, including infections with coronavirus, cytomegalovirus, hepatitis C virus, human immunodeficiency virus, human papilloma virus, and influenza virus (Bruening et al., 2018; Earnest et al., 2015; Florin and Lang, 2018; Hantak et al., 2019; Hochdorfer et al., 2016; Scheffer et al., 2013; Zona et al., 2013). Tetraspanin entry platforms could form in a slow and stochastic manner, providing an explanation for the asynchronous virion uptake with half-times of above 10 h (Becker et al., 2018). However, it is unclear whether virions associate with CD151 already during virion transfer from the ECM to the cell surface, or perhaps much later, in preparation of the endocytic uptake.

In this study, we explore the actin-dependent transport of HPV-PsV infection by employing the cell permeable mycotoxin and actin polymerization inhibitor cytochalasin D (CytD). We find that CytD causes the trapping of HPV16 pseudovirions (PsVs) in the ECM, where they remain accumulated adjacent to the cell periphery. Upon CytD removal, HS-decorated PsVs move from the ECM to cell surface CD151. The association of PsVs with CD151 persists within the next few hours, whereas the HS coat is stripped off. These findings distinguish a step in the infection cascade with CD151 playing an early role directly after the release of virions from the ECM.

Results

Cytochalasin D arrests the transfer of PsVs from the extracellular matrix to the cell surface

The molecular surface of PsVs is immunologically indistinguishable from HPV virus particles (Doorbar et al., 2012), which makes them a widely used tool for studying host cell entry. In this study, we employ HPV16-pseudovirions with an encapsidated luciferase reporter-plasmid under the control of the HPV16-promotor. Instead of viral-DNA, a luciferase-encoding plasmid enters the nucleus, enabling the analysis of the infection rate via the luciferase activity. Additionally, the plasmids are composed of nucleotides to which fluorophores can be coupled by click-chemistry. This allows for their detection by fluorescence microscopy throughout the entire infection pathway.

Many PsVs in the ECM are too far away from the cell surface to allow for direct physical contact between PsVs and entry receptors. In order to reach the cell surface, virions migrate by an actin-dependent transport along cell protrusions towards the cell body (Schelhaas et al., 2008; Smith et al., 2008). We aimed for blocking this transport in HaCaT cells, a human keratinocyte cell line that is widely used as a cell culture model for HPV infection. Cells are incubated for 5 h with PsVs in the absence or presence of 10 µg/ml CytD. Previously, it has been shown that this CytD concentration stops cell migration only after a few minutes (Peng et al., 2011). Therefore, we assume that CytD does not immediately stop PsV translocation. Figure 1A shows cells that, after the 5 h incubation, are fixed and stained by a membrane marker (the dye TMA-DPH) to mark the main cell body (gray), for PsVs (by click-chemistry; magenta), HS (antibody staining; cyan), and the membrane protein Itgα6 (antibody staining; green). In the control, PsVs frequently locate within the area of the cell body (Figure 1A, upper row; see outline based on TMA-DPH staining). In contrast, when CytD was present, the cell body is largely devoid of PsVs (Figure 1A, lower row; please note that the brightness in the center is mainly caused by autofluorescence). Instead of being transported to the cell body as in the control, PsVs accumulate adjacent to the cell body. This region is likely to be the ECM as it is up to several µm wide and rich in HS. We conclude that, upon inhibition of actin-dynamics by CytD, PsVs remain accumulated in the ECM, that in this assay is clearly distinguishable from the cell body.

Figure 1.

Comparing the total PsV intensities in the upper (control) and lower (CytD) magenta panels in Figure 1A suggests that the amount of accumulated PsVs at the periphery does not differ greatly from the amount of PsVs in the control. This is surprising, as CytD should not inhibit enzymes involved in capsid- or HS/HSPGs-processing, which is supposedly sufficient for the release of PsVs. Therefore, we conclude that potentially releasable PsVs remain in association with the ECM or the cell body. However, a quantitative comparison is not possible in this experiment because of the central autofluorescence (Figure 1A, large magenta area in the center) overlapping with some of the PsVs. For that reason, we employed unroofed cells (Heuser, 2000), also called membrane sheets. After treatment of cells with PsVs and CytD as above, membrane sheets are generated by brief ultrasound pulses that remove the upper cellular parts (and thus any intracellular autofluorescence), leaving behind the extracellular matrix and the basal membrane along with the bound PsVs. During imaging, membrane sheets are identified via staining of F-actin with fluorescent labelled phalloidin (Figure 1B, green). As in the experiment above (Figure 1A), PsVs remain accumulated after CytD treatment, adjacent to the now isolated basal membrane (Figure 1B, magenta).

The PsV maxima density (Figure 1B, magenta, detected as local maxima of antibody stained L1) and the PsV intensity are quantified in an area covering the cell body and the periphery. Compared to the control, CytD reduces the PsV maxima density by ̴ 30% (the PsV maxima density in the control and CytD treated cells is 0.7 and 0.5 PsVs/µm², respectively), whereas the maxima intensity increases by ̴ 40% (Figure 1C). A histogram of the PsV maxima intensities illustrates that CytD broadens the intensity distribution towards many PsVs that are several-fold brighter than the most frequent maxima intensity of the control (Figure 1D). Brighter PsVs resulting from CytD treatment can also be seen in the L1 images of Figure 1B (shown at the same scaling). We assume that this observation is due to the limited microscopic resolution, as CytD is unlikely to act directly on PsVs (e.g. by fusing them to larger and brighter particles). The accumulation of PsVs results in overlapping and thus poorly resolved maxima which are ̴ 40% brighter but of which there are ̴ 30% fewer. The decrease in maxima density is in the range of the increase in intensity, which yields a similar total signal in both conditions.

We addressed whether PsVs would still be accumulated when adding CytD 1 h after adding the PsVs rather than simultaneously (Figure 1B, lower panel, CytD after 1 h). As in the CytD condition, we see accumulated PsVs adjacent to the isolated basal membrane (Figure 1B, arrows point towards PsV accumulations). Furthermore, we find a broader PsV maxima intensity distribution compared to the control but a narrower intensity distribution compared to the sample where CytD is present during the entire 5 h incubation (Figure 1D). We conclude that without CytD it takes longer than 1 h for the PsVs to be translocated to the cell body. However, the preparation of PsVs for the translocation step likely is completed after 5 h.

Hence, it appears that CytD only arrests PsVs on their journey towards the cell surface. However, our assay is not very precise, and we do not know how many PsVs in the control were already endocytosed, which would reduce the reference value. Therefore, we can only say that within the 5 h incubation period we do not lose a major part of accumulated PsVs after processing in the ECM.

The blocking of PsV endocytosis by cytochalasin D is reversible

CytD not only arrests the transport of PsVs to the cell but is known to block other actin-dependent processes, which strongly affects the physiology of the cell. Therefore, we investigated whether PsVs would proceed normally on their infection pathway, after removal of CytD. In order to allow for recovery from CytD, cells were washed after the 5 h PsV/CytD treatment, then incubated for another 19 h and the infection rate was subsequently determined by measuring the luciferase activity in the cell lysate. Compared to the control, CytD reduced luciferase activity by 30% (Figure 1E; left). A certain degree of reduced infectivity is to be expected, as PsVs have ~20% less time to complete infection as compared to PsVs infecting cells that were not treated with CytD. When PsVs were not washed off and incubated for 24 h (5 h + 19 h), the infection rate increased by 50 %, whereas continuous treatment with CytD results in no infection (Figure 1E; right). The latter is in line with previous studies (Schelhaas et al., 2008; Selinka et al., 2002; Selinka et al., 2007; Spoden et al., 2013) showing that CytD is a strong inhibitor of HPV infection. In any case, PsVs proceed on the infection pathway when CytD is removed. The reduced infection rate of 30% can be explained, at least in part, by the 5 h delay by which PsVs reach the cell surface. Therefore, we propose that CytD is suitable for transiently arresting PsVs in a state between primary attachment to HS and cell surface binding.

The transfer of PsVs from the ECM to the cell body is fast and efficient

Next, we studied the time course of PsV translocation. Cells were treated as in Figure 1A (in Figure 2 the 0 min time point is identical), followed by washing off of PsVs/CytD and fixation after 0, 15, 30 and 60 min. The integrated density of PsVs locating at the cell periphery (Figure 2A, the cell periphery is defined by the white lines) is quantified and plotted against time. Compared to the control, CytD treatment results in a 6-fold increase of PsVs in the periphery when cells are immediately fixed after washing off of CytD (0 min time point). This increase is more than halved when cells were incubated for another 15 min after washing off of CytD and after 30 min, the level of the control is reached (Figure 2B). Hence, the half-time of PsV translocation from the periphery to the cell body is about 15 min, demonstrating that translocation is fast and efficient. Considering the fact that prior to endocytosis virions remain several hours on the cell surface, the duration of translocation is short, making it unlikely that it contributes to the asynchronous virion uptake that is observed for HPV-PsVs.

Figure 2.

Translocation of PsVs to CD151

As shown above, after CytD removal, the accumulated PsVs approach the cell surface. During this process, they may be coated with ECM cleavage products, and at some point, bind to cell surface HPV entry factors, like the tetraspanin CD151. For studying these events in detail, we employ superresolution STED microscopy, analyzing the association of PsVs with HS and CD151 over time. Cells are treated and monitored as in Figure 2, but with an extended time window of up to 180 min, as cell surface processes are expected to take longer than an hour.

After fixation, PsVs and CD151 were double-stained with antibodies against L1 and CD151, without prior cell permeabilization (cell surface staining, albeit fixation perforates the cell membrane to some extent). In addition, we stained F-actin with fluorescent labelled phalloidin. We simultaneously image L1 and F-actin in the confocal and CD151 in the STED channel.

As shown in Figure 3A, CD151 concentrates in spots scattered across the cell surface. It is also present at cell protrusions that are rich in F-actin and vary strongly in number and shape (Figure 4A shows anecdotal images of these filopodia). The antibody staining against PsVs, like the PsV click-chemistry staining in Figure 1A and 2A, is highly variable. CytD conserves the accumulated PsVs at the cell periphery and at early time points PsVs appear brighter (Figure 3A). Similar to Figure 2, we analyzed the time course of the diminishment of accumulated PsVs but focused on PsVs that are very close to or have already approached the cell surface (Figure 4B and C, for details see Figure legend of 4B). Moreover, we counted the number of PsVs instead of measuring the integrated density. As shown in Figure 4C, the PsVs accumulated in the cell border region diminish with about the same rate as PsVs in the periphery (Figure 2B).

Figure 3.

Figure 4.

Just like the L1 staining of PsVs, the intensity of the CD151 staining between cells is highly variable (Figure 3A and B). In the control, the mean intensity does not change over the 180 min time course, whereas a trend towards a lower mean intensity is observed in the cells treated with CytD, with the last time point being significantly lower than the control (Figure 3B). The decrease of the CD151 staining intensity points towards the possibility that after CytD wash off some CD151 is internalized, presumably due to co-internalization with endocytosed PsVs. This idea is supported by the observation that in particular at 180 min/CytD we observe occasionally CD151/PsV agglomerations (Figure 3A, see structures marked by arrows at 180 min/CytD, for more examples see Supplementary Figure 1A). We did not study this issue systematically, but some of these structures have clear three-dimensional extension (see Supplementary Figure 1B for axial scans) and therefore likely are tubular structures filled with several PsVs, as previously described by electron microscopy (Schelhaas et al., 2012). We found fewer of such structures at 180 min/control, as cells might have been actively interacting with PsVs for altogether 8 h, opposed to 3 h in the CytD treated cells. Hence, in the control, PsV endocytosis is less synchronized compared to CytD and therefore we do not find a notable CD151 diminishment.

For studying the association between PsVs (L1) and CD151, the Pearson correlation coefficient (PCC) between the channels is calculated. In theory, the PCC would equal 1 in the case of a perfect overlap and −1 for an image and its negative. We should not obtain large PCC values, as in most images PsVs are less abundant than CD151. As expected, the PCCs are around 0.1, with the exception of the 0 min/CytD value that is significantly lower and even slightly negative (Figure 3C, for control PCC values of flipped images see Supplementary Figure 2). This reflects the partially mutual exclusion of the two stainings; PsV accumulations are at the cell periphery and the CD151 staining is mainly at the cell body. At 30 min/CytD, the PCC has reached the level of 0.1 as in the other conditions, in line with Figure 2B showing that the accumulated PsVs diminish within 15 min.

PsV translocation to the cell surface should increase the number of PsVs that are tightly associated with CD151. As criteria for tight association, we define a distance of 80 nm between PsV and CD151 maxima, a value which is close to the resolution limit of the used microscope (Finke et al., 2020). In the control, the fraction of PsVs tightly associated with CD151, after correction for random background association (see Supplementary Figure 3) is about 10% (Figure 3D, control). At 0 min/CytD, we start with a fraction of 5.5%, a value that more than doubles in the next 60 min, although these changes are not significant (Figure 3D, CytD).

In summary, we conclude that within 180 min after a 5 h pre-incubation with PsVs (Figure 3D, control) around 10% of the PsVs associate tightly with CD151, with no trend towards more/less association over time. With the exception of the lower 0 min/CytD value of 5.5%, CytD has no effect on the association (Figure 3D, CytD). The 0 min/CytD value could be an underestimation, because (i) PsVs are optically less well-resolved (which increases the distance of a PsV to CD151) and (ii) this time point in particular is likely to be overcorrected for background association (see legend of Supplementary Figure 3). Despite of these uncertainties, the data shows that PsVs establish contact to CD151 assemblies either while still in the ECM, or in the moment of leaving the ECM.

PsV association with HS

Next, we studied the association between PsVs and HS, including a reference staining for the cell body as in Figure 1. However, we cannot image TMA-DPH with our STED microscope. Instead of TMA-DPH, we used Itgα6 as a reference marker for the cell body as it is not detected at cell protrusions (compare TMA-DPH and Itgα6 in Figure 1A). PsVs are visualized by click-chemistry and imaged in the confocal channel. HS and Itgα6 were stained by antibodies without prior permeabilization, and imaged at STED microscopic resolution. The three stainings were simultaneously recorded.

As shown in Figure 5A (green), the Itgα6 staining results in a densely spotted pattern. The Itgα6 intensity does not change over time (Supplementary Figure 4E). The pattern of the HS staining is variable regarding the overlap of HS with both, PsVs and Itgα6 (Figure 5A). Moreover, CytD treatment increases the HS intensity (Figure 5B, also visual when comparing the brightness of HS in Figure 5A upper and lower images). This increase of intensity is particularly notable at the 0 min time point, where the samples treated with CytD have a more than two-fold higher intensity and differ significantly from the control. This suggests that CytD prevents loss of HS occurring during or after PsV translocation.

Figure 5.

In the control, the PCC between PsVs and HS is close to zero (Figure 5C, for control PCC values of flipped images see Supplementary Figure 4). The largest PCC is found at 0 min/CytD, which reflects the finding that at this time point both PsVs and HS preferentially locate at the cell periphery. However, over time the accumulated PsVs diminish by translocating to the cell surface, which is accompanied by the PCC decreasing to zero, the same value as in the control (Figure 5C). Additionally, we analyze the PCC between PsVs and HS specifically in the cell body region, excluding the cell periphery (Figure 5D). The PCC in the cell body region is close to zero for the control. After CytD, we observe a trend towards a higher PCC between 0 min and 30 min. The fraction of PsVs tightly associating with HS at 0 min is increased by more than 3-fold by the CytD treatment. Over the next 180 min, the fraction decreases until it reaches the control value (Figure 5E).

Altogether, the analysis of the PCC between HS and PsVs (Figure 5C) and the fraction of tightly HS associated PsVs (Figure 5E) shows that CytD increases at 0 min the HS/PsV association, because PsVs remain accumulated in the ECM. After CytD removal, this association is lost, only in the region of the cell body is a trend towards a higher PCC at 30 min (Figure 5D) that could reflect HS-coated virions translocating to the cell surface.

Next, we plot for each PsV their distance to the next nearest Itgα6 maximum against the distance to their next nearest HS maximum. In the control, the distances remain unchanged over the entire observation time; 59% - 65% of the PsVs have a short distance (< 250 nm) to both Itgα6 and HS (Figure 6B, black; see also Supplementary Table 1), and another 12 - 15% a short distance (< 250 nm) to HS but not to Itgα6 (> 250 nm) (Figure 6B, green). The distance plots shown in Figure 6C are in line with the observation that the distance patterns of PsVs in untreated cells do not change over the 180 min observation time (Figure 6C, please compare the upper distance plots from left to right). In cells treated with CytD, a larger fraction of PsVs is accumulated at the periphery at the 0 min time point, as marked by a larger fraction (46% as opposed to 12% in the control; Figure 6B, green) of PsVs with a short distance (< 250 nm) to HS and a large distance (> 250 nm) to Itgα6. Over time, the PsVs in the CytD treated cells (Figure 6C, lower row) acquire a shorter distance to Itgα6 and a larger distance to HS. After 180 min, the distances are similar to the untreated control.

Figure 6.

If HS would bind irreversibly to PsVs, the population with short HS-PsV distances is expected to be constant. However, we observe that this population diminishes. This can only be explained by HS uncoating. In fact, between 60 min/CytD and 180 min/CytD, the fraction of PsVs with a large distance (> 250 nm) to HS and a short distance (< 250 nm) to Itgα6, representing PsVs at the cell body without HS, increases from 9.6% to 25.2% (Figure 6B, pink). The observation that PsVs lose their HS coat after translocating to the cell surface is also supported by the lower HS intensity in the control (Figure 5B) and the transient increase in the PCC between HS and PsVs in the cell body at 30 min (Figure 5D).

The movement of HS towards the cell body is demonstrated by a shortening of the HS-Itgα6 distance over time (Supplementary Figure 4D). Moreover, the PCC between PsVs and HS in the cell body area increases (Figure 5D). Altogether, the data suggest that HS-coated PsVs migrate towards the cell body. A majority of translocating PsVs shed HS within one hour after washing off of CytD (Figure 6C, lower panels, showing increasing PsV-HS distances). This is the time window where we notice the occurrence of endocytic PsV/CD151 structures (see above and Supplementary Figure 1).

Discussion

In this study, we investigate the early stages of HPV16 infection that occur prior to virus endocytosis. Our findings reveal that the dynamics of actin in filopodia, rather than the diffusion of released particles, are crucial for the release of viruses from the adhesive ECM and their subsequent transfer to the cell body. When actin-dependent processes are inhibited, virus particles remain trapped in the ECM, allowing for a more synchronized observation of the infection process. Upon removing the transfer blockade, we observe a rapid translocation of heparan sulfate (HS)-coated virus capsids from the ECM to CD151 assemblies and the cell body. Subsequently, the capsid sheds its HS coat and is endocytosed in association with CD151.

A reversible block of PsV translocation from the ECM to the cell body by CytD

Many viruses use cell surface HS for primary attachment, including herpes simplex virus type 1, human cytomegalovirus, human immunodeficiency virus type 1, adenovirus type 2, dengue virus, hepatitis B virus, and vaccinia virus (Cagno et al., 2019). HPVs could be different as they bind as well to HS of the extracellular basement membrane, at least in vivo during the wounding and healing processes required for infection (Day and Schelhaas, 2014; Kines et al., 2009; Schiller et al., 2010).

For studying the translocation of virions from the ECM to the cell surface, we arrest the translocation with the actin-transport inhibitor CytD. After 5 h, we find a 6-fold increase of PsVs in the cell periphery compared to the control (Figure 2B). After washing off of CytD, the accumulated PsVs diminish with a half-time of about 15 min. This demonstrates fast translocation of PsVs to the cell surface, further supported by the increase in the PCC between PsVs and CD151 from 0 min to 30 min (Figure 3C).

Infectivity is diminished by 30% (Figure 1E) when cells are allowed to recover for 19 h after washing off of CytD and PsVs. In comparison to the control, this effect appears moderate, considering that the 5 h incubation time without CytD allows for continuous infection by the PsVs. In addition, slightly harmful effects of the CytD treatment cannot be excluded. The rapid onset of PsV translocation and the moderate effect on infection indicate that CytD only transiently prevents PsV uptake.

It is possible that accumulated PsVs are merely diminished by the wash off, however, the effect of this could not have been very strong, considering the 6-fold accumulation of PsVs after the 5 h incubation with CytD. Moreover, in Figure 4C, we measure the diminishment of accumulated PsVs in the cell border region, in this case as percentage of all PsVs. The diminishment in the cell border does not reciprocally reflect an increase of PsVs at the cell body. However, assuming that those PsVs not in the cell border region nor within the boundaries of the cell are not increasing, we conclude that over time the number of PsVs at the cell body increases. Hence, the PsV diminishment at the cell border region is accompanied by an increase of cell body PsVs (provided background PsVs are constant), documenting that PsVs indeed translocate.

Compared to Figure 2B, in Figure 4C the half-time of diminishment appears longer. We attribute this to an underestimation of the number of PsVs at 0 min/CytD. If the resolution of the accumulated PsVs in Figure 4C had been higher, their fraction might have been larger than 36%. Consequently, the half-time of PsV diminishment would be shorter yet. In contrast, the microscope resolution in Figure 2 is not a limiting factor for the time course measurements, as instead of counting PsVs (Figure 4C) we quantify their integrated fluorescence intensity (only possible at the periphery, as autofluorescence precludes quantification of PsVs in the cell body region). Despite of these minor uncertainties, we conclude that the translocation step is fast and efficient.

Some trends in the data suggest that the CytD treatment synchronizes HPV uptake. On the one hand, we see endocytic structures after CytD treatment more frequently (for examples see Supplementary Figure 1), in particular between 60 min and 180 min. On the other hand, the CD151 intensity diminishes over time only after the CytD treatment (Figure 3B). We speculate that the processing of the PsVs in the ECM promotes desynchronization and is largely completed after the 5 h pre-incubation with CytD. Hence, after CytD removal most PsVs are ready at the same time for cell surface binding. Compared to the control, 6-fold more PsVs (Figure 2B) synchronously translocate to the cell surface. Thus, in the following 3 h more endocytic events occur than in the control, which is noticeable in a diminishment of CD151 being co-internalized with the PsVs.

The role of actin-driven virion transport

As outlined above, the electrostatic HPV-HS bond is strong, and thus requires processing of the capsid surface/HS in order to proceed on the infection pathway. This hypothesis is supported by several studies showing that e.g., the activity of HS cleaving enzymes facilitates the virus release from the ECM and infection (Surviladze et al., 2015). CytD supposedly does not inhibit the activity of these enzymes that may be present in the ECM (Jamieson et al., 2021). It is possible that not at every ECM location all enzymes are available, therefore PsVs must be pulled through the ECM to the required enzymes. If that is the case, CytD is indeed not likely to inhibit the processing enzymes, but rather inhibits the transport of PsVs and thus prevents the contact to the necessary enzymes. Alternatively, processing might be completed but virions remain associated with the ECM by electrostatic/van-der-

Waals forces. In this case, for ECM release actin-driven transport would be required to transport the virion out of the sticky matrix to the cell surface. Here, detachment from the ECM may be facilitated by new bonds established between the PsVs and cell surface receptors. Now, the PsVs and cell surface receptors move towards the central cell body by intracellular dynamics and we ultimately observe endocytosis (see Figure 3A).

In our experiments, many PsVs reach the cell body after crossing a distance of several micrometers (see the analyzed cell periphery area in Figure 2A). It has been previously described that HPV-PsVs are taken up from the ECM via transport along actin-rich protrusions (Schelhaas et al., 2008; Smith et al., 2008). In HeLa cells, the speed of the fastest, straight migrating PsVs is in the range of several micrometer per minute (Schelhaas et al., 2008). Considering that only the fastest PsVs move at that speed, we propose our observed time course is in accordance with the previously reported migration of PsVs along actin-rich protrusions.

If cells require some time to recover from CytD before PsV migration restarts, the translocation half-time of 15 min is likely to be overestimated. In any case, the translocation is fast compared to the entire infection process and therefore cannot largely contribute to the asynchronous HPV uptake.

Inhibition of transport diminishes infection only of subconfluent but not of confluent HaCaT cells by about 50% (Schelhaas et al., 2008). Therefore, actin-dependent translocation along protrusions may be dispensable for infection at a high cell density (Schelhaas et al., 2008) and simply increases the exposure of cells to virions (Smith et al., 2008). However, this increased exposure could be relevant in vivo, as wounding of the epidermis results in upregulation of filopodia formation (Vasioukhin et al., 2000). Filipodia usage not only facilitates infection but also increase the likelihood of virions to reach their target cells during wound healing, the filopodia-rich basal dividing cells. In fact, several viruses exploit filopodia during virus entry (Chang et al., 2016), wherefore actin-driven virion transport could have a more important role in HPV infection than currently assumed.

Translocating PsVs are decorated with HS which they lose over time

At 0 min/CytD, 15% of the PsVs are tightly associated with HS (Figure 5E), i.e. the distance between PsVs and HS is ≤ 80 nm. 80 nm is above the theoretical distance between an HS fragment adhering to the surface of a PsV and the PsV position. Considering the important role of HS as a primary attachment site, the fraction of 15% appears unexpectedly small. One factor contributing to an underestimation is that accumulated PsVs are not well-resolved (see above). Therefore, some PsV maxima positions actually represent the positions of several merged PsV maxima. Instead of several correct positions that may be within a distance of 80 nm to HS, we obtain a displaced position further away from HS and then these PsVs would not only not contribute to the fraction of tightly associated PsVs, but increase the fraction of non-tightly associated PsVs. This error decreases over time, as the accumulated PsVs diminish and individual PsVs become better resolved. Another reason for underestimation could be that the HS antibody does not reach all HS epitopes in the dense matrix, or that PsV binding and antibody binding to HS is to some extent mutually exclusive. Finally, not all PsVs bound to the ECM are expected to bind to HS but could also bind to laminin 332 (Culp et al., 2006).

The shortening of the HS-Itgα6 distance (Supplementary Figure 4D) suggests movement of HS towards the cell body. Moreover, in CytD treated cells, after 30 min, the increase in the PCC between PsVs and HS specifically measured in the cell body region (Figure 5D) implies co-migration of PsVs and HS. PsV-HS association decreases in 180 min from 15% to 5% (Figure 5E), with 5% being equal to the association in the control (Figure 5E). Hence, over time PsVs lose their HS coat. Loss of HS is also visible in Figure 6C, showing that over time more PsVs have very long distances to HS. Additionally, the lower HS intensity level in the control (Figure 5B) suggests that those PsVs that are actively taken up lose their HS coat. These observations are in line with a previous study showing that the colocalization of HS with cell surface and endocytosed virions decreases over time unless virions are directed into a non-infectious entry pathway (Selinka et al., 2007).

In conclusion, as previously postulated by others (Ozbun and Campos, 2021), our data confirms that PsVs released from the ECM are decorated with HS. However, they lose HS within a few hours after translocation, perhaps due to further structural changes of the capsids or when they bind to receptors for endocytosis.

PsVs-CD151 association is an early contact to the cell surface

CD151 is crucial for cell entry (Scheffer et al., 2013; Spoden et al., 2008). Hence, PsVs must establish contact to CD151 assemblies sometime between the first cell surface contact and endocytosis. As shown in Figure 3D, about 10% of the PsVs are tightly associated with CD151 over the entire incubation period if cells are not treated with CytD. We find a similar effect in samples treated with CytD, with the exception of the 0 min/CytD time point, where the value of associated PsVs is almost halved. The reduction could be in some part due to background overcorrection (see legend of Supplementary Figure 3), but also mainly because not all PsVs have translocated via actin-driven transport to CD151 receptor assemblies.

In any case, the data suggest that PsVs establish contact to CD151 assemblies at the moment of translocation to the cell surface, and from then on, the fraction of tightly associated PsVs remains on a 10% level. This implies that, after the initial formation of the PsV-CD151 assemblies, they do not change in nature. Yet, we observe agglomerated CD151 maxima with PsVs at later time points that likely are endocytic structures (see arrow in Figure 3A, Supplementary Figure 1). Therefore, some further CD151 reorganization into larger entry platforms must occur (Florin and Lang, 2018). This is in accordance with the observation that virions enter the cell in crowds with many of them occupying one endocytic organelle (Schelhaas et al., 2012). However, we do not resolve these events in our assay, either because they are too rare or they are not associated with a change in the PsV-CD151 distance.

For the PsVs tightly associated with CD151, one may have expected a larger fraction than 10%. Aside from the issues regarding the PsV-HS association, for the PsV-CD151 association our 80 nm distance criteria may be too strict, because PsVs are not expected to bind directly to CD151 but to larger plasmalemmal assemblies containing, among other proteins, the yet unknown PsV cell surface receptor.

Altogether, we propose that the initial contact between PsVs and the cell surface involves an assembly defined by CD151 (Figure 7). Moreover, our data show that after cell surface contact the virions are decorated with HS fragments.

Figure 7.

Integration of our data into the cascade of steps underlying HPV infection

HPV infection is the result of several steps, starting with the initial binding of virions via electrostatic and polar interactions (Dasgupta et al., 2011) to the primary attachment site HS, which induces capsid modification (Feng et al., 2024) and HS cleavage (Surviladze et al., 2015), enabling the virion to be released from the ECM or the glycocalyx. Next, virions bind to the cell surface (to a secondary receptor or an early version of it), and become internalized via endocytosis, before they are trafficked to the nucleus (Ozbun and Campos, 2021).

We propose that after 5 h of CytD treatment, glycan-induced structural activation of the capsid and HS cleavage essentially are completed ((i) in Figure 7). Subsequently, (ii) the HS-decorated virion is transported from the ECM to the cell body on filopodia, which takes about 15 min. During this process, PsV-CD151 association occurs early. (iii) In the samples recorded 30 – 180 min after CytD wash off, we observe that PsVs lose their HS coat, and individual PsV-CD151 assemblies seem to merge into larger structures that are subsequently endocytosed.

Materials and Methods

Antibodies and PsVs

We used the following primary antibodies in immunostainings. For the capsid protein L1, a rabbit polyclonal antibody (pAb) K75 (1:1000) as described previously (Rommel et al., 2005), for CD151 a mouse monoclonal antibody (mAb) (1:200) (Biorad, cat# MCA1856GA), for Heparan sulfate (HS) a mouse IgM monoclonal antibody (1:200) (amsbio, cat# 370255-S) and for Itgα6 a rabbit polyclonal antibody (1:200) (Invitrogen™, cat# PA5-12334). As secondary antibodies we used a STAR GREEN-coupled goat-anti-rabbit (1:200) (Abberior, cat# STGREEN-1002-500UG), an AlexaFluor™ 594- coupled donkey-anti-rabbit (1:200) (Invitrogen™, cat# A21207), an AlexaFluor™ 594-coupled donkey- anti-mouse (1:200) (Invitrogen™, cat# A21203), an AlexaFluor™ 594-coupled donkey-anti-mouse IgM (1:200) (Invitrogen™, cat# A21044), a STAR RED-coupled goat-anti-mouse (1:200) (Abberior, cat# STRED-1001-500UG) and a STAR RED-coupled goat-anti-rabbit (1:200) (Abberior, cat# STRED-1002- 500UG). Additionally, phalloidin iFluor488 (1:1000 of ready to use solution) (abcam, cat# ab176753) or phalloidin iFluor647 (1:1000) (abcam, cat# ab176759) was used to stain F-actin. Staining of the PsV plasmid DNA was performed by click-labeling according to the manufacturer’s instructions with the dye 6-FAM Azide (Baseclick EdU 488 kit, Carl Roth, cat# 1Y67.1).

HPV16 PsVs were produced following established procedures (Buck et al., 2004; Spoden et al., 2012). HEK293TT cells were cultured in 175 cm² flasks and transfected with polyethylenimine (PEI), using equimolar amounts of codon-optimized HPV16 L1/L2 (pShell 16L1/L2wt) and a reporter plasmid (pGL4.20-puro-HPV16 LCR). For the production of 5-ethynyl-2’-deoxyuridine (EdU)-labeled PsVs, the culture medium was replaced 5 hours post-transfection with fresh medium containing 30 μM EdU. After 48 hours, cells were harvested, centrifuged, and washed twice with phosphate-buffered saline (PBS) supplemented with 9.5 mM MgCl₂ (1x PBS/ MgCl₂). Following a final centrifugation step, the pellet was resuspended in 1x PBS/ MgCl₂ containing 0.5% Brij58 (Sigma-Aldrich), and 0.2% Benzonase (Merck Millipore), and incubated for 24 hours at 37°C on a rotating platform. Subsequently, lysates were chilled on ice and the NaCl concentration was adjusted to 0.85 M. After clarification by centrifugation, the supernatant was loaded onto an iodixanol (Optiprep) gradient consisting of 39%, 33%, and 27% layers (bottom to top). Gradients were equilibrated for 90 minutes at room temperature before ultracentrifugation at 55,000 rpm for 3.5 hours at 16°C. Fifteen fractions of 300 µl each were collected from the top and analyzed via luciferase reporter assay to identify peak fractions. PsV titers were determined based on packaged genomes (viral genome equivalents, vge) as previously described (Spoden et al., 2012). The concentrations of stock solutions are 7,7 x 10⁶ vge/µl (for microscopy experiments) and 14,1 x 10⁶ vge/µl (for the luciferase assay).

Cell culture

For microscopy, human immortalized keratinocytes (HaCaTs) (Cell Lines Services, Eppelheim, Germany) are cultured in high glucose DMEM (Gibco®, cat# 61965-026) medium supplemented with 10% FBS (PAN Biotech, cat# P30-3031) and 1% Penicillin/Streptomycin (10,000 U/ml Penicillin, 10 mg/ml Streptomycin; PAN Biotech, cat# P06-07100) at 37 °C with 5 % CO₂. For the luciferase infection assay, cells were grown at 37°C in Dulbecco’s modified Eagle’s medium (DMEM + GlutaMAX (Thermo Fisher Scientific), supplemented with 10% FBS (Sigma-Aldrich), 1% minimum essential medium non-essential amino acids (MEM non-essential amino acids (Thermo Fisher Scientific)), and 5 µg/ml ciprofloxacin (Fresenius Kabi). For experiments, the antibiotics in the medium are omitted.

Sample preparation and immunostaining

About 150,000 HaCaT cells were plated onto 25 mm diameter poly-L-lysine (PLL) coated glass coverslips in 6-well plates and incubated for 24 h at 37°C and 5% CO₂. The next day, cells were incubated at 37°C and 5% CO₂ for 5 h with PsVs (50 vge/plated cell) and 10 µg/ml cytochalasin D (CytD; stock solution 10 mg/ml in dimethyl sulfoxide (DMSO)) (Life Technologies, cat# PHZ1063) in DMEM supplemented with 10% FBS. For controls, we added the same amount of DMSO without CytD. Cells were washed with PBS (137 mM NaCl, 2.7 mM KCl, 1.76 mM KH₂PO₄, 10 mM Na₂HPO₄, pH7.4) and fresh DMEM supplemented with 10% FBS was added. Cells were incubated further at 37°C and 5% CO₂ for 0 min (here the medium was added followed by immediate removal), 15 min, 30 min, 60 min and 180 min. Cells were washed twice with PBS and fixed at room temperature (RT) with 4% PFA in PBS for 30 min, unless membrane sheets are generated. In this case, the coverslips were placed in ice cold sonication buffer (120 mM KGlu, 20 mM KAc, 10 mM EGTA, 20 mM HEPES, pH 7.2) and a 100 ms ultrasound pulse at 100% power was applied. This was repeated until in total about 10 pulses are applied at different locations of the coverslip. Then, membrane sheets were fixed like cells at RT with 4% PFA in PBS for 30 min. PFA is removed, and residual PFA is quenched by 50 mM NH₄Cl in PBS for 30 min. Afterwards, samples are blocked with 3% BSA in PBS for 30 min. Staining of PsVs was performed by click-labeling with the dye 6-FAM for 30 min at RT according to the manufacturer’s instructions. Then, samples were washed three times with PBS. In case of no PsV labeling by click-chemistry, directly after blocking, the respective primary antibodies are added: mouse IgM mAb against HS (1:200), rabbit pAb against Itgα6 (1:200), mouse mAb against CD151 (1:200) and rabbit pAb K75 against L1 (1:1000) in 3% BSA in PBS for 2 h. Samples were washed three times with PBS before adding the respective secondary antibodies and fluorescent labelled phalloidins in 3% BSA in PBS for 1 h: for HS we add AlexaFluor™ 594 coupled to donkey-anti-mouse IgM (1:200), for Itgα6 STAR RED coupled to goat-anti-rabbit (1:200), for CD151 AlexaFluor™ 594 coupled to donkey-anti-mouse (1:200) or STAR RED coupled to goat-anti-mouse (1:200), for L1 AlexaFluor™ 594 coupled to donkey-anti-rabbit (1:200) or STAR GREEN coupled to goat-anti-rabbit (1:200) and for F-actin phalloidin iFluor488 (1:1000) or phalloidin iFluor647 (1:1000). Afterwards, samples are washed three times with PBS, followed by mounting of the coverslips onto microscopy slides using ProLong^® Gold antifade mounting medium (Invitrogen, cat# P36930).

Confocal and STED microscopy

For confocal and STED microscopy, samples were imaged employing a 4-channel STED microscope from Abberior Instruments (available at the LIMES institute imaging facility, Bonn, Germany). The microscope is based on an Olympus IX83 confocal microscope equipped with an UPlanSApo 100x (1.4 NA) objective (Olympus, Tokyo, Japan). Confocal and STED micrographs are recorded simultaneously. For confocal imaging, a 485 nm laser is used for the excitation of 6-FAM, STAR GREEN and phalloidin iFluor488, and emission is detected at 500–550 nm. For STED imaging, a 561 nm laser (at 45%) is used for the excitation of AlexaFluor™ 594 (detection at 580–630 nm) and a 640 nm laser (at 45%) for the excitation of STAR RED and phalloidin iFluor647 (detection at 650- 720 nm), in combination with a 775 nm laser for depletion (at 45%). The pixel size is set to 25 nm.

Analysis of confocal and STED micrographs

Images are analyzed using Fiji ImageJ in combination with a custom written macro, essentially as described previously (Schmidt et al., 2024). When analyzing double stainings of Alexa594 (red channel) and STAR RED (long red channel), we corrected for crosstalk from the red into the long red channel by subtracting 40% of the intensity of the red channel image from the long-red channel image. First, images are smoothed with a Gaussian blur to improve maxima detection. For confocal PsV-DNA micrographs, we employed a Gaussian blur of σ = 1, and for confocal PsV-L1 and STED micrographs a Gaussian blur of σ = 0.5. Local maxima were detected using the ‘Find Maxima’ function (noise tolerance 60 for L1 if stained together with phalloidin iFluor488, and 80 if stained together with CD151, 3 for click-labeled PsVs, 8 for CD151, 6 for HS and 15 for Itgα6), yielding maxima positions in pixel positions. For analysis, rectangular regions of interest (ROIs) are defined. To measure the L1 maxima intensity (Figure 1C and D) the ROIs covered the whole image. For CytD wash off experiments the rectangular ROIs covered mainly the cell body (75%) but included the cell periphery as well (25%).

In order to define the cell border region (Figure 4B), the ImageJ ‘Make Binary’ function is used on CD151 STED micrographs for generating a binary mask. When possible, the ‘Wand’ tool is used to further outline the area covered by the cell, otherwise the same was done manually using the original micrograph as a reference. From the outline in the binary mask, a ROI is created, that is then filled in white with the ‘Flood Fill’ Tool. The ROI is shrunk by 20 pixels, and cleared inside, what leaves a 20 pixels broad, closed ribbon marking (i) the intracellular side of the cell and (ii) arbitrary edges produced in the process above. These edges are removed by manual adjustments with the ‘Pencil’ tool and the clear function. Afterwards, the remaining border region defines the intracellular side of the cell border. From this border region a ROI is created by applying the ‘Wand’ tool, that is symmetrically broadened to 40 pixels. From the 40 pixels, 30 pixels cover the intracellular and 10 pixels the extracellular side. Using the ImageJ ‘Find Maxima’ function, PsV-L1 maxima are detected (at a noise tolerance of 80) both in the cell border region ROI and in a ROI covering the entire micrograph. From these values, the percentage of PsVs within the cell border region is calculated.

For measuring e.g. the CD151 intensity over time, we used ROIs as illustrated in Supplementary Figure 2A, and measured in the ROI the mean gray value. For background correction, we subtracted the mean gray value measured in a ROI next to the cell.

For measuring maxima intensity, a ROI with a diameter of 125 nm (5 pixels) was placed onto the determined maxima positions (see above). Using these ROIs, the average mean gray value of each maximum is measured. The average background mean gray value is measured in a ROI placed next to the cell and subtracted from the average mean gray value of the maxima.

To measure the shortest distance, e.g. between PsV and CD151 maxima, as a quality control for the PsV maxima (CD151 maxima do not undergo the quality control step) we placed onto each maximum position a horizontal and a vertical linescan (31 pixels long × 3 pixels width). Afterwards, we fitted a Gaussian distribution to the intensity distribution of each maximum. Only if at least one of the fits exhibits a fit quality of R² > 0.8 and if the Gaussian maximum locates central to the intensity distribution the maximum is included in the further analysis. Moreover, using a 125 nm ROI at the PsV and CD151 maxima positions, the center of mass of fluorescence is determined, yielding the maxima positions in sub-pixels. Based on these positions we further calculated the shortest distance of a PsV maximum to the nearest CD151 maximum.

Using the shortest distances, we either calculated the fraction of tightly associated PsVs over time (the fraction of PsVs with a distance ≤ 80 nm to e.g. CD151), or the average shortest distance over time. The fraction of tightly associated PsVs is further corrected for random background association. Random background association increases with the maxima density. The relationship between background association and maxima density is expressed in a linear calibration line. For the calibration line, we use the same images and determine the fraction of PsVs with a distance to e.g. CD151 measured on horizontally and vertically flipped (randomized) ROI-defined images (see also Supplementary Figure 2A). The fraction is plotted against the maxima density. Fitting of a linear regression line yields the mathematical relationship between the fraction of tightly associated PsVs and maxima density, from which we obtain the value to be subtracted for background correction.

The Pearson correlation coefficient (PCC) was calculated with a custom written ImageJ macro in ROIs as illustrated in e.g. Supplementary Figure 2A. As a control we performed the analysis on horizontally and vertically flipped ROI-defined images as well (see e.g. Supplementary Figure 2A). The cell body PCC between PsVs and HS is measured in smaller ROIs covering exclusively the cell body region (this is confirmed with the Itgα6 image as reference).

Per condition and biological replicate, 14 - 15 images are analyzed. For images from one set of experiments, in the figure the same channels are shown at the same scaling and same lookup-tables.

Epi-fluorescence microscopy and image analysis

For epifluorescence microscopy, we used microscopic equipment and settings as previously described (Homsi et al., 2014) except of the illumination system, which is replaced by a Lumencor Light engine® Spectra 90-10034 (Lumencor, USA). In brief, PFA-fixed cells are imaged in PBS containing TMA-DPH [1-(4-tri-methyl-ammonium-phenyl)-6-phenyl-1,3,5-hexatriene-p-toluenesulfonate (T204; Thermofisher)] for visualizing the membranes in the blue channel. PsVs, HS and Itgα6 were imaged in the green, red and far-red channels, respectively.

For analysis, rectangular ROIs of ≈ 45 x 10³ pixel are placed onto the cell body such that one half of the ROI covers roughly the ECM area and the other one the cell body. The cell periphery is defined using the TMA-DPH image from which a binary mask is created. With reference to the binary mask, a band of 30 pixels width is created beginning at the cell body and reaching out towards the cell periphery. Using this ROI, in the PsV-DNA image the size of the ROI and the mean grey values are measured from which the integrated intensities are calculated after background correction.

Luciferase infection assay

HaCaT cells are grown in 24-well plates and allowed to adhere overnight. The following day, cells are treated with HPV16 PsVs at a concentration of 100 vge per cell, in the presence or absence of 10 µg/ml cytochalasin D (CytD) in DMSO or an equivalent amount of DMSO (control). In one condition, PsVs/CytD are applied for 5 hours, after which the medium is replaced with fresh medium lacking the compounds, and incubation is continued for additional 19 hours (in total 24 h). In another condition, cells are exposed to PsVs/CytD continuously for the full 24 h period. Then, cells are washed with PBS and lysed using 1 x Cell Culture Lysis Reagent (Promega, cat# E153A). Following high-speed centrifugation, luciferase activity in the cleared lysates is quantified using an LB 942 Tristar 3 luminometer (Berthold Technologies). Cytotoxic effects, accompanied by a loss of membrane integrity (indicated by released lactate dehydrogenase (LDH)), are determined by measuring the LDH levels in the cell lysates. LDH activity is assessed using the CytoTox-ONE™ Homogeneous Membrane Integrity Assay (Promega, cat# G7891). LDH fluorescence is quantified according to the manufacturer’s instructions using the LB 942 Tristar 3 luminometer.

Statistics

Data sets were based on three biological replicates, including 14 - 15 images per replicate. For infection assay, data sets were based on three biological replicates, with three technical replicates from each biological replicate. Data was tested for significance with a two-tailed, unpaired student’s t test with significance * = p < 0.05, ** = p < 0.01 and *** = p < 0.001.

Supplementary information

Supplementary Figure 1.

Supplementary Figure 2.

Supplementary Figure 3.

Supplementary Figure 4.

Supplementary Figure 5.

Supplementary Table 1.

Acknowledgements

LF and TL were funded by the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation), Projektnummer 322863883, FL696/3-2 (LF); LA 1272/8-2 (TL).