In the activation of HPV-specific human B cells HPV-VLP vaccines mimic membrane-associated antigens

Significance

In an era of predicted emerging pandemics, our ability to produce highly effective vaccines that induce long-lasting protection from infection even following a single dose will be critical. To do so will require a greater understanding of the mechanisms by which highly efficacious vaccines activate human naive B cells. Here, we provide evidence that the current human papillomavirus virus-like particle (HPV-VLP) vaccines that provide long-lasting protective virus-neutralizing antibody responses even after a single dose mimic membrane-presented antigens in their activation of human B cells, requiring the activity of the plasma membrane mechanosensitive cation channel, Piezo1. This finding suggests the possibility that this feature of HPV-VLP vaccines could be exploited to design highly effective human vaccines for pathogen for which we have none.

Abstract

B cell responses to membrane-presented antigens appear to be strongly favored over soluble antigens in vivo suggesting that vaccines that mimic membrane-presented antigens may be highly efficacious. We recently demonstrated that human B cell responses to membrane-associated but not to soluble antigens in vitro depended on the expression and activity of the plasma membrane mechanosensitive ion channel, Piezo1. Here, we provide evidence that the efficacy of the current human papillomavirus virus-like particle (HPV VLP) vaccines may be due in part to their inherent ability to mimic Piezo1-dependent membrane presentation of antigens to B cells. We compared HPV-specific human B cell responses to HPV VLPs versus soluble HPV pentameric capsomeres and showed that although both induced calcium responses, only HPV VLP-induced responses were blocked by Piezo1 inhibitors. The kinetics of internalization of HPV-VLP and capsomeres into HPV-specific B cells were similar and neither required Piezo1 function as shown by small interfering RNA (siRNA)-mediated knockdown of Piezo. However, trafficking of HPV-VLPs into intracellular major histocompatibility complex (MHC) class II, lysosomal associated membrane protein 1 (LAMP1)⁺ antigen-processing compartments was Piezo1-dependent, whereas trafficking of capsomeres was not. In addition, a time course of intracellular trafficking suggested that colocalization of HPV-VLP with MHC classII was more stable over time as compared to capsomeres. Taken together these findings suggest that the ability of HPV-VLP vaccines to mimic Piezo1-dependent membrane antigen presentation may be exploited in the design of highly effective human vaccines.

Human papillomavirus virus-like particle (HPV-VLP) vaccines are remarkably effective in providing long-lasting protection from both HPV infection and HPV-related cancers even following a single vaccine dose (1). Antibodies play a central role in vaccine-induced protection and the efficacy of the HPV-VLP vaccines has been largely attributed to their ability to consistently induce high, stable levels of virus-neutralizing antibodies and to the exceptional susceptibility of HPV to antibody-mediated inhibition of human epithelial target cell infection (2). However, at present, the features of the HPV VLPs that contribute to their effectiveness as vaccines is incompletely understood and filling this gap in our knowledge may benefit the development of highly effective vaccines for pathogens for which we currently have none.

It is well established that B cells respond robustly to multivalent antigens in which epitopes are presented in ordered arrays on particles but only poorly to monomeric antigens in solution (3–6). HPV VLPs are highly multivalent, ~50 nm particles, that form spontaneously from 72 pentamers of the HPV VLP L1 protein (termed capsomeres) (1). An atomic model of HPV VLPs generated from a crystal structure of L1 pentamers, provided evidence that the L1 pentamer formed an ordered display of antigen epitopes at 50 to 100 Å on the VLP surface. The importance of epitope spacing is underscored by the observation of similar epitope spacing on the surfaces of most viruses, including both nonenveloped capsid viruses and enveloped viruses, as well as other microbial structures, including bacterial pili (7). Moreover, viruses that express essential envelope membrane proteins at low, irregular densities, notably HIV, generally fail to induce long-lived protective antibody immunity (8).

In addition to the valency and epitope density of antigens, studies of B cells responding to antigens in the highly specialized microenvironments of secondary lymphoid organs provide evidence that even though B cells can respond to both small soluble antigens and membrane-associated antigens in vitro, the most efficient mode of B cell activation in vitro and the primary mode of B cell activation in vivo appears to be of antigens associated with membranes, in particular on the surfaces of antigen-presenting cells such as follicular dendritic cells (9–11). The preference for activation by antigens associated with membranes indicates that B cells are able to sense differences between soluble and membrane-associated antigens. We recently discovered that human B cell responses to a membrane-associated versus a soluble antigen were distinguished by the requirement for the function of the cell surface mechanosensing ion channel, Piezo1 (12) for full activation. We found that human B cells making contact with a rigid glass surface either by cover glass or glass probe in vitro, independently of B cell receptors (BCRs) engagement of antigen, showed increased plasma membrane tension, and calcium fluxes in Piezo1-dependent manners. Critically, B cell responses to antigens presented on planar lipid bilayers but not to lipid bilayers alone or to soluble antigen resulted in B cell activation blocked by two inhibitors that interfered with Piezo activity by different mechanisms, as measured by accumulation of BCR at the immune synapses, polymerization of actin, and calcium fluxes. In addition to Piezo1 inhibitors, we also provided direct evidence for the role of Piezo1 in B cell responses to membrane-associated antigens showing that a siRNA mediated reduction of 50% in Piezo1 abundance was sufficient to block B cell responses to membrane-associated but not soluble antigen (12). Of note, the siRNA-mediated reduction protocol was the same as that used in the original study that demonstrated that Piezo1 was an essential component of distinct mechanically activated ion channels by showing that a 50% reduction in Piezo1 expression was sufficient to reduce mechanically activated cation channel activity (13). Together, these findings raise the possibility that the efficacy of HPV VLPs may be achieved, at least in part, by mimicking the structural properties of membrane-presented antigens and activating B cells by a Piezo-dependent mechanism.

Antigen-driven B cell responses require not only antigen binding to the BCR that provides signal 1 but in addition a second signal provided by T cell help or innate immune receptor ligands (14). In the absence of a second signal B cells undergo antigen-induced cell death (15). In this two-signal framework, it is possible that the efficacy of HPV-VLP vaccines is due to their ability to provide a second signal for full B cell activation. In fact, it has been suggested that VLP structures evolved to function as danger signals or pathogen associated molecular patterns for B cells (16). Consistent with this possibility, HPV VLPs are unusual in their ability to induce T helper cell–independent antibody responses, including IgG class switching, in mouse models (17). In addition, HPV VLPs have been shown to have the uncommon ability in mice to activate autoreactive B cells that have been put into a state of hyporesponsiveness, termed anergy, following binding soluble, monomeric forms of the self-antigens (18–20).

Here, we describe the response of HPV16-specific human B cells to an HPV-VLP vaccine in vitro and compare these to responses to pentamers of HPV16 termed capsomeres. We provided details of the outcome of human B cell responses to HPV-VLP in terms of providing both signal 1 and signal 2 to HPV16-specific B cells resulting in full activation and internalizing and trafficking HPV VLPs into MHC class II-containing compartments for presentation. We also determined the dependence on Piezo1 of the B cell calcium flux triggered by HPV-VLP vaccine and by soluble HPV pentamers in the initiation of B cell responses.

Results

HPV16 VLPs and HPV16 Capsomeres Compete for Binding to HPV-Specific BCRs.

To compare the ability of HPV16 VLPs versus HPV16 capsomeres to activate HPV16-specific human B cells we generated a stable Ramos B cell line that expressed human HPV16-specific IgM⁺ κ⁺ BCRs (termed A24M04) (SI Appendix, Fig. S1 A–G) (21). Several V_H and V_K gene pairs including A24M04 were cloned from IgG⁺ HPV16-specific memory B cells obtained from peripheral blood of individuals vaccinated with Gardasil-4 containing HPV16 VLPs as described (SI Appendix, Fig. S2) (22). Among the cloned HPV16-specific monoclonal antibodies (mAbs), several were expressed as soluble antibodies and these showed similar binding to HPV16 VLPs and to HPV16 capsomeres by ELISA (SI Appendix, Fig. S2). A comparison of the cell surface marker phenotype of wild type (WT) Ramos and A24M04 Ramos showed these to be similar (SI Appendix, Fig. S1H).

To compare the activation of HPV16-specific B cells to HPV16 VLPs and HPV16 capsomeres it was critical to demonstrate that HPV16 VLPs and HPV16 capsomeres bound exclusively to B cells expressing HPV16 specific BCRs and not to nonspecific BCRs or other structures on the surfaces of B cells that could potentially transduce activating signals. To do so we determined the specificity of binding of fluorescently labeled HPV16 VLPs and HPV16 capsomeres to three different human B cell populations including B cells in human peripheral blood mononuclear cells (PBMCs) obtained from HPV16-seronegative individuals in which we determined the frequency of HPV-specific B cells to be ~1 in 2 × 10³ B cells; WT Ramos expressing only an endogenous mIgM⁺λ⁺ nonspecific BCR and A24M04 Ramos (referred to as A24 Ramos) expressing a mIgM⁺κ⁺ HPV16-specific BCR. In each case, cells were incubated with HPV16 VLPs labeled with two different fluorophores, HPV16 VLP-Alexa Fluor 647 (HPV16 VLP-AF647) and HPV16 VLP-Alexa Fluor 488 (HPV16 VLP-AF488) and analyzed by flow cytometry (Fig. 1A). Neither of the fluorescently labeled HPV16 VLPs showed detectable binding to B cells in PBMC samples or to WT Ramos. However, both HPV16 VLP-AF647 and HPV16 VLP-AF488 bound similarly to over 95% of A24 Ramos resulting in double positive cells in flow cytometry. Double positive cells were also observed for A24 Ramos binding to combinations of HPV16 VLP-AF647 and HPV16 capsomere-Alexa Fluor 488 (Capsomere-AF488), including HPV16 VLP-AF647 versus Capsomere-AF488 and HPV16 capsomere-Alexa Fluor 647 (Capsomere-AF647) versus Capsomere-AF488 demonstrating equivalent binding of HPV16 VLPs and HPV16 capsomeres to A24 Ramos BCRs (Fig. 1B).

Fig. 1.

HPV16 VLPs and HPV16 capsomeres bind exclusively to HVP16-specific B cells. (A) Flow cytometry plots show the binding of HPV16 VLP-AF647 and HPV16 VLP-AF488 to PBMCs alone (Left), WT Ramos (Middle), or A24 Ramos (Right). (B) Flow cytometry plots showing binding of either HPV16 VLP-AF488 and HPV16 VLP-AF647 (Left), HPV16 VLP-AF647 and capsomere-AF488 (Middle), or capsomere-AF647 and capsomere-AF488 (Right) to A24 Ramos diluted into PBMCs at a ratio of 1 A24 Ramos:100 PBMCs. (C) Representative flow cytometry plots of A24 Ramos diluted into PBMCs at a ratio of 1 A24 Ramos:100 PBMCs are gated on A24 Ramos labeled with fluorescent Abs specific for CD19, CD27, and CD10 (CD19+CD27^brightCD10^bright). A24 Ramos were preincubated without competitor or with the indicated unlabeled competitors HPV16 VLPs, BPV, or HPV58 VLPs, at concentrations of 4 or 20 µg/mL followed by incubation with HPV16 VLP-AF647 and HPV16 VLP-AF488. (D) A summary of the data as in C is given, representing replicates in three independent experiments. The data were analyzed by a Kruskal–Wallis test with a Dunn’s correction and P-values > 0.05 were considered not significant (ns). **P < 0.01. (E) Representative flow cytometry plots of A24 Ramos diluted into PBMCs at a ratio of 1 A24 Ramos:100 PBMCs and gated on A24 Ramos as in C. Cells were preincubated without competitor or with the indicated concentrations of unlabeled HPV16 capsomeres, followed by incubation with HPV16 VLP-AF647 and HPV16 VLP-488. (F) A summary of the data in E is shown, representing replicates in three independent experiments. The data were analyzed by a Kruskal–Wallis test with a Dunn’s correction. ns, P > 0.05, ***P < 0.001.

We next determined whether A24 Ramos discriminated VLPs composed of L1 proteins of the HPV16 serotype versus VLPs from related but non-cross-reacting HPV58 or to the unrelated bovine papillomavirus (BPV). To do so, before the addition of HPV16 VLP-AF647 and HPV16 VLP-AF488 to A24 Ramos, the cells were briefly preincubated with competitors either unlabeled HPV16 VLP, BPV VLP, or HPV58 VLP at equivalent concentrations to the HPV16 VLP-AF488 and HPV16 VLP-AF647 (4 µg/mL) or at a fivefold higher concentration (20 µg/mL) (Fig. 1C). Of note, concentrations of HPV VLPs and HPV capsomeres are given as µg/mL L1 protein. Shown are representative flow cytometry plots of the competition for binding of the labeled HPV16 VLPs to A24 Ramos (Fig. 1C). Competition was only observed for the unlabeled HPV16 VLPs at a fivefold higher concentration (20 μg/mL) and not for BPV VLPs or HPV58 VLPs at either concentration tested (Fig. 1C). Quantification of results from several experiments showed that unlabeled HPV16 VLPs reduced the binding of the AF647- and AF488-labeled HPV16 VLPs to A24 Ramos cells in a dose-dependent fashion to less than 30% of A24 Ramos whereas there was no significant effect of BPV VLPs or HPV58 VLPs on AF647- and AF488-HPV16 VLPs binding at either concentration tested (Fig. 1D). Similar results were obtained for the inhibition of HPV16 VLP-AF647 and HPV16 VLP-AF488 by unlabeled capsomeres (Fig. 1 E and F). Taken together these data provide strong evidence that the binding of HPV16 VLPs and HPV16 capsomeres was restricted to HPV16-specific B cells expressing HPV16-specific BCRs and moreover, binding to the VLP versus capsomere forms of HPV16 was indistinguishable.

Ramos A24 Calcium Responses to HPV16 VLPs Are More Robust than to HPV16 Capsomeres and Require Piezo1.

We next determined whether A24 Ramos cells responded similarly to HPV16 VLPs versus HPV16 capsomeres by fluxing calcium. We compared the calcium responses of HPV16-specific A24 Ramos to HPV16 VLPs versus HPV16 capsomeres over a concentration range that induced maximal calcium responses. Ramos cells were incubated with increasing concentrations of HPV16-VLP or HPV16 capsomeres (0.1 µg/mL to 5 µg/mL) or with 20 μg/mL anti-κ. Shown are the data from a representative experiment (Fig. 2A) and the quantification of results from two to four independent experiments (Fig. 2B). The calcium response of A24 Ramos cells to HPV16 VLPs at each concentration tested was greater than the responses to HPV16 capsomeres and persisted for longer. The response to HPV16 capsomeres at 5 µg/mL was similar in magnitude and duration to the control response to anti-κ at 20 µg/mL.

Fig. 2.

HPV16 VLPs induce more robust and prolonged calcium fluxes as compared to HPV16 capsomeres in a Piezo1-dependent manner. A24 Ramos cells were loaded with a calcium indicator dye, CAL520, and baseline calcium levels were recorded by flow cytometry. Upon addition of 0.1, 1, and 5 µg/mL of HPV16 VLPs, HPV16 capsomeres or 20 µg/mL goat F(ab′)2 anti-Igκ Abs (anti-κ) calcium fluxes were recorded with time. (A) Shown is a representative plot of the median CAL520 fluorescence intensities (MFI) over time at the indicated concentrations of HPV16 VLP or HPV16 capsomere or anti-κ. (B) Quantifications of the fold change in CAL520 MFI during BCR stimulation relative to the baseline at each indicated concentration of HPV16 VLPs or HPV16 capsomere are given as the mean ± SEM for 2-4 independent experiments. (C–H) A24 Ramos B cells were either untreated or pretreated with GsMTx4 (5 µM) for 30 min or OB-1 (20 µM) for 1 h at 37 °C and calcium fluxes were recorded upon addition of HPV16 VLP (C and D), HPV16 capsomere (E and F), or anti-κ (G and H). Shown for each antigen/inhibitor combination are the representative kinetics plots of the calcium fluxes (Left) and the fold change (Right) as the mean ± SEM from 2 to 3 independent experiments. Statistical significance was determined by the two-tailed paired t-test. ns, P > 0.05, *P ≤ 0.05, **P ≤ 0.01, and ***P ≤ 0.001.

We next determined whether the observed calcium responses required the activity of Piezo1 as we previously demonstrated was necessary for the response of human B cells to membrane-associated antigens but not to soluble antigens (12). We previously demonstrated that Ramos cells expressed Piezo1 independently of BCR expression by real time RT-PCR and by flow cytometry using Abs specific for Piezo1 (12). A24 Ramos were incubated with Piezo1 inhibitors that had different mechanisms of action, either GsMTx4 (23) (Fig. 2 C, E, and G) or OB-1 (24) (Fig. 2 D, F, and H), prior to the addition of HPV16 VLPs or HPV16 capsomeres, or anti-κ. Both inhibitors reduced the magnitude of the calcium responses induced by HPV16 VLPs (Fig. 2 C and D) to levels similar to those induced by capsomeres. In contrast, neither GsMTx4 nor OB-1 had a detectable effect on responses to HPV16 capsomeres (Fig. 2 E and F) or to anti-κ (Fig. 2 G and H). Thus, activation of Ramos A24 to HPV VLPs but not to capsomeres or to anti-κ crosslinking appeared to share the functional phenotype of B cell responses to membrane-associated antigen in requiring the activity of Piezo1.

To conclude that Ramos cells responded differently to the HPV16 VLP versus HPV16 capsomere, it was important to rule out the possibility that A24 Ramos cells were responding to HPV16 VLP presented by by-stander A24 Ramos cells in cultures. To do so we determined whether addition of nonspecific WT Ramos cells (at ratios of either 1 A24 Ramos:5 WT Ramos or 1 A24 Ramos:20 WT Ramos) had an impact on the response of A24 Ramos cells to HPV16 VLP, to HPV16 capsomere, or to anti-κ (SI Appendix, Fig. S3). The results showed no detectable differences between the undiluted versus diluted A24 Ramos samples responding to either HPV16 VLP, HPV16 capsomeres or anti-κ providing evidence that A24 Ramos calcium responses did not depend on antigen presentation.

HPV16 VLPs and HPV16 Capsomeres Fail to Provide Second Signals to B Cells.

We showed recently that activation of B cells through the BCR alone providing signal 1 in the absence of signal 2, resulted in activation-induced mitochondrial dysfunction characterized by gradually increasing intracellular calcium levels and excessive production of mitochondrial reactive oxygen species (ROS), resulting in activation-induced cell death (AICD) (15, 25). In contrast, when a signal 2 was provided by T cell help or a TLR9 agonist within 12 h of BCR stimulation, calcium and ROS levels were kept under control and the B cells remained viable. We concluded that the binding of antigen to the BCR activated a “metabolic clock” that limited the time during which B cells must receive a second signal to survive. Using this model, we asked whether HPV16 VLPs and/or HPV16 capsomeres have the inherent ability to provide signal 2 to prevent HPV16-induced mitochondrial dysfunction in A24 Ramos cells. We first determined that the A24 cells were susceptible to signal 1-induced AICD by stimulation with anti-human IgM antibodies and that they could be partially rescued from AICD by the addition of a second signal the TLR9 agonist, cytosine guanine dinucleotide (CpG), as previously described (15). Stimulation with anti-human IgM alone induced calcium flux and increased mitochondrial damage by increasing ROS production, indicated by increased MitoSox staining and increased mitochondrial swelling indicated by increased MitoTracker Green staining (SI Appendix, Fig. S4 E–G) that were partially restored by the addition of CpG (SI Appendix, Fig. S4). Thus, A24 Ramos appeared to be susceptible to signal 1–induced AICD that was rescued by CpG provided signal 2.

We next confirmed that both HPV16 VLPs and HPV16 capsomeres activated A24 Ramos cells as evidenced by increased expression of the activation marker CD69 (SI Appendix, Fig. S4A). Both HPV16 VLPs and HPV16 capsomeres induced expression of CD69 that was significantly greater than the expression of CD69 by untreated A24 Ramos cells, although the CD69 levels in HPV16 VLP-stimulated A24 Ramos cells were less as compared to HPV16 capsomere-stimulated A24 Ramos cells. We then showed that both HPV16 VLPs and HPV16 capsomeres stimulated A24 Ramos cells similarly to increase the production of mitochondrial superoxide as quantified by MitoSox staining (SI Appendix, Fig. S4B) and the levels of intracellular calcium (SI Appendix, Fig. S4C) resulting in decreased viability (SI Appendix, Fig. S4D). Consistent with the lower CD69 expression in B cells activated by HPV16 VLP versus capsomeres, the HPV16 VLP-treated B cells were also less viable than capsomere-stimulated B cells which may be attributed to stronger BCR crosslinking and induced AICD in the absence of a second signal. Indeed, we showed earlier that BCR-antigen-induced mitochondrial dysfunction of B cells correlated with the strength of the BCR stimulation (15). There may also be differences in the kinetics of the responses to HPV-VLP versus capsomere not reflected in the single 48 h time frame.

To explore these possibilities, we carried out a time course of unstimulated B cells and cells treated with HPV-VLPs or capsomeres alone (signal 1) or T cell help alone (signal 2) or with both HPV-VLPs or capsomeres with T cell help (signal 1 plus signal 2) for 6 h, 12 h, or 24 h measuring either CD69 expression or viability (Fig. 3). The results showed that at each time point activation of CD69 expression was increased over that of unstimulated cells by capsomeres and HPV VLPs alone and that addition of T cell help resulted in higher CD69 expression for HPV-VLPs as compared to capsomeres, consistent with a stronger activating effect of HPV-VLP. However, HPV-VLP stimulation alone resulted in decreased viability as compared to capsomere stimulation that was partially rescued by T cell help at 24 h (Fig. 3B).

Fig. 3.

HPV16 VLP-induced B cell activation results in loss of viability that can be partially rescued by T cell help. A24 Ramos cells were cultured in the absence or presence of HPV16 capsomeres (10 µg/mL) or HPV16 VLPs (10 µg/mL) in the absence or presence of T cell help (Th) media for the indicated time. Shown are the graphs with mean ± SD from flow cytometry for: (A) the percentage of surface CD69-expressing cells; (B) the percent of viable cells (Viability) measured using a Live/Dead fixable viability dye. Each symbol is an individual replicate. Data are representative of two independent experiments, and statistical significance was determined by one-way ANOVA with Tukey’s multiple comparison test. ****P < 0.0001, ***P < 0.001, **P < 0.01, ns, not significant. ####P < 0.0001 between unstimulated and either HPV16 capsomeres or HPV16 VLPs-stimulated cells in A.

Taken together these findings provide evidence that although the antigen components of HPV16 VLPs and HPV16 capsomeres are able to trigger A24 Ramos activation (signal 1), neither HPV16 VLPs nor HPV16 capsomeres were inherently able to provide a signal 2 to rescue A24 Ramos cells from BCR-induced mitochondrial dysfunction, AICD, and reduced viability. In light of these findings, we conclude that the efficacy of HPV16 VLP vaccines is unlikely due to their ability to provide signal 2 to rescue B cells from AICD.

Both HPV-VLPs and Capsomeres Are Readily Internalized from the Cell Surface by HPV-Specific B Cells.

The binding of antigen to B cell surface expressed BCRs has the potential to induce both the initiation of BCR signaling cascades and the subsequent signaling-dependent internalization of BCR-bound antigen into specialized intracellular compartments for processing and presentation to antigen-specific CD4⁺ T_H cells. We first determined whether the internalization of HPV16 VLPs and HPV16 capsomeres by HPV16-specific B cells were similar and if internalization was dependent on Piezo-1 function. To do so we transfected A24 Ramos with either PIEZO1-specific siRNA or control siRNA after optimizing with siGlo-Red, an indicator of transfection efficiency (Fig. 4A). The transfection with PIEZO1-specific siRNA resulted in a decrease in Piezo1 expression in A24 Ramos cells by approximately 57%, as compared to control siRNA, determined by staining with rabbit anti-Piezo1-fluorescently labeled polyclonal antibodies (Fig. 4 B and C). Of note, we previously demonstrated a 50% decrease in Piezo1 expression was sufficient to block human B cells calcium flux in response to the Piezo1 agonist YODA1 and to membrane-associated anti-Ig but not to soluble anti-Ig (12).

Fig. 4.

Piezo1 down-expression has no influence on both HPV16 VLPs and capsomeres-driven BCR down-modulation. A24 Ramos cells were transfected with 1 µM of either control or PIEZO1 siRNA using Amaxa nucleofection system as described in Method and cultured for 3 d at 37 °C in the 5% CO₂ incubator. (A) siGlo-Red transfection indicator, a red fluorescent oligonucleotide duplex, was used in three different concentrations (0.5, 1.0, and 1.5 µM) for the optimization of siRNA transfection efficiency in A24 Ramos cells. (B and C) Representative flow cytometry plot for total Piezo1 expression stained by rabbit anti-Piezo1-Janelia Fluor 549 polyclonal antibodies (Piezo1-JF549) for either Control (Control)–or PIEZO1 siRNA (PIEZO1)-transfected cells (B). The level of Piezo1 expression in PIEZO1 siRNA transfected cells is given as a percent of Piezo1 expressed in Control cells (C). Shown is a bar graph with mean ± SD with dots, indicating data from three independent experiments. One-sample t test was performed for the statistical analysis. **P < 0.01. (D) Surface BCR downregulation was analyzed for HPV16 VLPs and capsomeres in Control and Piezo1 knock-down (KD) cells (Piezo1 KD-VLP and Piezo1 KD-Capsomere) by Control and PIEZO1 siRNA, respectively. HPV16 VLPs or capsomeres (5 µg/mL) were added to Control or Piezo1 KD cells, allowed to bind to the A24 BCR on ice for 30 min and then chased for the indicated time at 37 °C and fixed with 4% paraformaldehyde (PFA). Surface BCR was labeled with either goat IgG anti-Igµ heavy chain (IgM HC) or anti-Igκ light chain (Igκ LC) and analyzed for the percent of Piezo1 expression in control cells in surface level of BCR by flow cytometry. Ratios of surface IgM HC or Igκ LC gMFIs at each time point (T = t) relative to those on ice (T = 0) was calculated (T = t/T =0). Statistical significance (95% CI) was assessed with the “compareGrowthCurves” function from the statmod statistical modeling package for three independent experiments (26).

We next determined the kinetics of BCR internalization by HPV16 VLP versus capsomeres. To do so, HPV16 VLPs or capsomeres were added to Control or Piezo1 knock-down (KD) cells on ice for 30 min and then chased for up to 90 min at 37 °C. Cells were fixed and stained with Alexa Fluor 488-conjugated Fab of Igµ heavy chain-specific antibodies or Igκ light chain-specific antibodies to quantify BCR internalization. As shown the kinetics of internalization of the BCR following HPV16-VLP or capsomere binding in control cells and in Piezo1 KD cells were equivalent (Fig. 4D). We observed differences in the detection of BCRs by anti-IgM versus anti-κ antibodies which we attribute to different accessibilities of the two antibodies to the antigen bound BCRs.

We confirmed BCR internalization from the B cell surface with HPV16 VLP by three-dimensional (3D) imaging. To do so, A24 Ramos cells were stained with Alexa Fluor 488-conjugated Fab of Igμ heavy chain-specific antibodies to visualize BCR and with BV421-conjugated mouse IgG specific for CD45 to visualize the cell surface. Cells were placed on a chambered cover glass and allowed time to settle on the chamber surface and Alexa Fluor 647-conjugated HPV16 VLPs were added to the chambers and the cells incubated at 37 °C for 5, 30, or 120 min. The cells were fixed, and images were obtained by high-resolution Nikon AXR confocal microscope equipped with a Nikon Spatial Array Confocal (NSPARC) detector. Shown are orthogonal views by XY and XZ planes of three-color merged reconstructed 3D images of representative A24 Ramos cells stimulated with HPV16 VLPs (SI Appendix, Fig. S5A). The BCRs and HPV16 VLPs associated and formed clusters over the cell surface by 5 min and remained associated for the 120 min of imaging during BCR-HPV16-VLP internalization (SI Appendix, Fig. S5A and Movie S1). The amount of internalized antigen increased with time (SI Appendix, Fig. S5B) consistent with the finding in the images provided (SI Appendix, Fig. S5A).

We next determined whether early events in the BCR-induced signaling initiated by antigen binding that accompany internalization, including phosphorylation of the BCR Igα chain, Syk, and The extracellular signal-regulated kinase, were similar in response to HPV16 VLPs and to HPV16 capsomeres and if signaling was affected by the Piezo1 inhibitor GsMTx4 (SI Appendix, Fig. S6). The signaling events induced by HPV16 VLPs and HPV16 capsomeres were similar and neither appeared to be sensitive to the Piezo 1 inhibitor GsMTx4. Taken together these results provide strong evidence that both HPV-VLPs and capsomeres induce early signaling events and BCR internalization from the B cell surface that are independent of the function of Piezo1.

The Intracellular Trafficking of HPV16 VLPs But Not Capsomeres into Intracellular LAMP-1⁺ Human Leucocyte Antigen-DR⁺ Antigen Processing Compartments Is Piezo1-Dependent.

We used 3D high-resolution confocal microscopy to visualize the intracellular trafficking of HPV16 VLPs versus capsomeres in Control versus Piezo1-KD A24 Ramos cells to determine the dependence of trafficking on Piezo1 function. Cells were placed in chambers and incubated with either Alexa Fluor 488-HPV16 VLPs or capsomeres for 120 min at 37 °C, fixed and permeabilized with Triton-X100, and labeled with Alexa Fluor 549-IgG anti-LAMP1 and Alexa Fluor 647-IgG anti-human leucocyte antigen (HLA)-DR. 3D images were obtained by Z-stack confocal microscopy followed by reconstruction of the images. Shown are representative 3D surface views of HPV VLPs and capsomeres (Antigen), LAMP1, HLA-DR, and merged images of the three for Control and Piezo1-KD Ramos A24 cells (Fig. 5A). To visualize more clearly the colocalization of the three markers within the representative cells shown in Fig. 5A, we also provided orthogonal views by XY, XZ, YZ planes of three-color merged 3D images of the same cells (Fig. 5B). Colocalization of antigens, LAMP1, and HLA-DR were quantified by Manders overlap coefficient. Shown are colocalization of antigens per cell with LAMP1 (Fig. 5C); or with HLA-DR (Fig. 5D) and colocalization of LAMP1 and HLA-DR (Fig. 5E). The 3D images (Fig. 5 A and B) and their quantification (Fig. 5 C–E) together provide strong evidence that HPV-VLPs required Piezo1 activity to colocalize with LAMP1 (Fig. 5C) and with HLA-DR (Fig. 5D) whereas colocalization of capsomeres with LAMP1 and HLA-DR were similar in control and Piezo1-KD cells. Of note, LAMP1 and HLA-DR colocalize similarly in Control versus PIEZO1 KO cells (Fig. 5E) and form highly integrated structures indicating that these structures themselves are not dependent on Piezo1. Rather, Piezo1 function appears to be required to facilitate the colocalization of HPV16 VLPs, not capsomeres, with the intracellular vesicles containing both LAMP1 and HLA-DR (Fig. 5F).

Fig. 5.

Piezo1 down-expression influences on the trafficking of HPV16 VLPs into the LAMP1⁺HLA-DR⁺ organelles. Control or Piezo1-KD A24 Ramos cells were allowed to settle on the chambered cover glass for 20 min at 37 °C and incubated with 5 µg/mL of Alexa Fluor 488-conjugated HPV16 VLPs (VLP) or HPV16 capsomeres (Capsomere) for 120 min at 37 °C in the CO₂ incubator. The cells were fixed with 4% PFA, permeabilized with 0.1% Triton-X100, and labeled with mouse mAbs specific for LAMP1 and HLA-DR conjugated with Alexa Fluor 549 and Alexa Fluor 647, respectively. 3D surface and sectional views of single cell images were obtained by Z-stack confocal imaging using a ZEISS 880 confocal laser scanning microscope followed by 3D reconstruction using Imaris and Huygens software as described in Materials and Methods. (A and B) Shown are 3D surface views of HPV16 VLPs or capsomeres (Antigen, green), LAMP1 (red), or HLA-DR (Blue) or three-color merged images of the representative cells (A). Also shown are orthogonal views by XY, XZ, and YZ planes of three-color merged 3D images of the same cells shown in panel A to provide a clear image of the colocalization of the three markers in intracellular vesicles (B). Yellow arrowheads indicate the vesicles containing all three markers: antigen, HLA-DR, and LAMP1. Fluorescence intensity is not comparable between different cells. (Scale bar, 2 µm.) (C–E) The ratio of antigens colocalized with LAMP1 or HLA-DR, or that of LAMP1 colocalized with HLA-DR was quantified among different experimental conditions by Manders overlap coefficient per cell between antigens and LAMP1 (C), between antigens and HLA-DR (D) or between LAMP1 and HLA-DR (E) using Huygens software. Over ninety cells per condition combined from two independent experiments were analyzed. Graphs are shown as box plots with median and Tukey whiskers. ****P < 0.0001, **P < 0.01, ns, not significant; Statistical significance was performed by the Kruskal–Wallis test in all four groups followed by Dunn’s multiple comparison tests between Control and Piezo1 KD cells treated with either HPV16 VLPs or HPV16 capsomeres, or between HPV16 VLPs and HPV16 capsomeres within Control or Piezo1 KD cells. (F) HPV16 VLPs or capsomeres colocalized with LAMP1 and HLA-DR double positive (LAMP1⁺HLA-DR⁺) organelles were quantified and compared between Control and Piezo1 KD cells as the percentage of HPV16 VLPs or capsomeres colocalized with LAMP1⁺HLA-DR⁺ vesicles by Manders overlap coefficient. Data are shown as box plots with median and Tukey whiskers for over 40 cells per condition from two independent experiments. Statistical significance was compared between VLP- and Capsomere-treated cells in Control siRNA-transfected cells, and between Control and Piezo1 KD cells in either VLP or Capsomere-treated cells by the two-tailed Mann–Whitney test *P < 0.05, ns, P > 0.05.

We imaged two additional markers for early endosomes (EEA1 and TfR) in A24 Ramos cells to better characterize the early internalization of HPV-VLP and capsomeres into B cells. To do so cells were activated with either HPV-VLP or capsomeres for 30 or 60 min at 37°, fixed and labeled with anti-CD45 to stain the cell surface, and then permeabilized and labeled with EEA1- and TfR-specific antibodies and imaged by 3D confocal microscopy. Shown are representative three-color images of CD45, Antigen, and EEA1 (SI Appendix, Fig. S7A) and quantification of overlap of antigen and EEA1 and TfR overlap with EEA1 for HPV VLP and capsomere-treated cells (SI Appendix, Fig. S7B). The results showed that the associations of HPV VLP and capsomeres with EEA1 were similar at 30- and 60-min following activation. Moreover, the association of TfR and EEA1 was highly similar in B cells stimulated with HPV VLP versus capsomere (SI Appendix, Fig. S7 A and B).

We also carried out a time course (0 to 120 min) of cells activated by HPV VLP or capsomeres and stained with HLA-DR- and HLA-DM-specific antibodies. Shown are three color images of CD45, Antigen and colocalization of Antigen and HLA-DR or HLA-DM for representative images of cells activated by HPV VLP or capsomeres (SI Appendix, Fig. S8A). The time course showed that the association of HPV VLP with HLA-DR was more stable over the 30 to 60 min time course as compared to the association of capsomeres with HLA-DR that dropped significantly at 60 min and then reassociated at 90 min. In controls the association of HPV VLP and Capsomeres with HLA-DM were similar.

Thus, taken together these findings provide evidence that even though the triggering of early signaling events and initial internalization of HPV-VLPs and capsomeres are similar there are clear differences in intracellular trafficking. Only HPV VLP trafficking to processing compartments required the function of Piezo1 and the association of HPV-VLP with HLA-DR appeared more stable over time as compared to that of capsomeres. Such differences in trafficking may be important in having the potential to impact antigen presentation of HPV VLP versus capsomeres.

Discussion

An important clue as to how HPV VLP vaccines might be so effective in inducing human B cells’ protective immunity came from a relatively recent understanding that even though B cells can be activated by soluble antigens in vitro, B cells activation in vivo is predominantly by antigens presented on the surfaces of antigen-presenting cells (9). Moreover, the mechanisms underlying B cell responses to antigens on membranes were demonstrated to be distinct from B cell responses to antigens in solution (27). For example, we recently showed that the requirement for the activity of Piezo1 distinguished human B cell responses to membrane-associated antigens versus soluble antigens, providing a unique functional phenotype for human B cell responses to membrane-associated antigens. Here, we compared the response of HPV16-specific Ramos cells to HPV16 VLPs versus HPV16 capsomeres and provided evidence that these responses could be distinguished based on the requirement for the activity of Piezo1. Two different Piezo1 inhibitors with different mechanisms of action reduced the B cell calcium flux in response to HPV16 VLP but not in response to HPV16 capsomeres or anti-κ leading us to speculate that the HPV VLPs may mimic the features of membrane-associated antigens and result in the opening of Piezo1. These findings may define a quantifiable functional characteristic of VLPs making it possible to predict the efficacy of other VLP constructs by their ability to activate Piezo1 expressed by human B cells. We also provided evidence that even though HPV16 VLPs provided signal 1 they were unable to rescue B cells from signal 1-induced AICD by providing signal 2. Presumably B cells responding to HPV16 VLP vaccines acquire an activating second signal in vivo by other means. We also determined that even though both HPV-VLPs and HPV capsomeres stimulated early BCR signaling and were internalized from the B cell surface similarly, independently of Piezo-1 function, HPV-VLPs, but not capsomeres, required Piezo-1 for intracellular trafficking to LAMP1⁺ HLA-DR⁺ antigen processing compartments. In addition, a time course of intracellular trafficking measuring HPV VLP versus capsomere association with HLA-DR provided evidence that the association of HPV VLP with HLA-DR was more stable as compared to the association of capsomeres and HLA-DR. Taken together these results provide evidence that the intracellular trafficking of HPV VLP and capsomeres are different, raising the possibility that such differences may impact antigen presentation to T helper cells. The dependence of trafficking to LAMP1⁺HLA-DR⁺ processing compartments on the function of Piezo1 may also suggest potential targets for enhancing vaccine effectiveness.

At present, there are a variety of VLPs that have in common the display of antigenic epitopes in dense arrays and have shown varying degrees of efficacy as human vaccines. It would seem likely that the mechanisms that underlie their efficacy may differ from particle to particle and different particle types may result in qualitatively different B cell responses. For example, Brooks et al. (28) recently described viral size liposomes termed synthetic virus-like structures (SVLS) engineered to display hen egg lysozyme (HEL), a model antigen in mice. These HEL-SVLS induced potent all-or-none B cell responses, that were particularly long-lived and were dependent on the density of the displayed antigen but were affinity independent. The HEL-SVLS did not require the B cells’ expression of CD19 as has been described for the response of B cells to membrane-associated antigens (29). Of interest, these HEL-SVLS propagated signals downstream of the BCR by evading LYN-dependent inhibitory pathways. Similarly, we showed that the ability of membrane-associated antigens to activate human atypical memory B cells, that were refractory to activation by soluble antigens was through a mechanism that excluded the B cell inhibitory FcγRIIB from the immune synapse (30).

In summary, in an era of predicted recurring pandemics our ability to produce highly effective, pathogen-specific vaccines that provide long-lived protection in nonimmune populations will be critical. Meeting the challenge of translating our knowledge of the cellular and molecular mechanisms underlying antigen-driven B cell activation to vaccine development may be key to our success.

Materials and Methods

Cell Lines and Cell Culture.

The human Burkitt’s lymphoma cell line Ramos.2G6.4C10 (ATTC, CRL-1923, CVCL_1646) and its derivatives were cultured in the Roswell Park Memorial Institute (RPMI) complete media containing 10% Fetal Bovine Serum (FBS) (HyClone Cytiva), RPMI 1640 supplemented with 2 mM L-glutamine, 50 µM 2-mercaptoethanol, 100 U/mL penicillin-streptomycin, 10 mM 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid, 1 mM sodium pyruvate, and 1× MEM Non-Essential Amino Acids (Gibco). Lenti-X 293T cells (Fisher Scientific) were maintained in 10% FBS (Tet system approved, Takara Bio)-contained Dulbecco’s Modified Eagle Medium supplemented with L-glutamine, penicillin-streptomycin, sodium pyruvate, and MEM Non-Essential Amino Acid.

Antigens, Antibodies, and Other Reagents.

VLP 16L1, BPVL1, and VLP 58L1 were generated as previously described (31). Briefly, 293TT cells were transfected with the plasmid expressing 16L1 (p16L1h), BPVL1 (pCAL) see https://ccrod.cancer.gov/confluence/display/LCOTF/Support or 58L1 (p58L1h) (32). Cells were grown for 48 h prior to harvest and lysis. VLP lysates were incubated overnight at 37 °C in Dulbecco’s phosphate-buffered saline (DPBS) containing 9.5 mM MgCl₂, 0.5% Triton-X-100, 25 mM ammonium sulfate, 0.1% Benzonase (Sigma), and 0.1% Plasmid-safe (Epicentre). Lysates were salt extracted, clarified, and purified on OptiPrep (Sigma) density gradients. Fractions were collected from the gradients and analyzed by loading on a polyacrylamide gel for L1 protein detection and quantitation. A 16L1 capsomere expression plasmid was constructed by insertion of synthesized 16L1 DNA containing the ∆N10∆414-431 mutation (33) and engineered restriction sites (3’ Xba1 and 5’ HindIII) for cloning into the p16L1h plasmid. Generation of capsomeres was performed as described for VLPs apart from the density gradient purification. For capsomere purification, a 50% sucrose stock in DPBS/0.8 M NaCl was prepared and diluted in DPBS/0.8 M NaCl to 5%, 10%, 15%, 20%, and 25%. 0.8 mL steps of the 5 to 25% were layered into thin wall polyallomer 5 mL tubes (Beckman 326819). Salt extracted, clarified lysates were layered on top of the gradient and the tube filled with DPBS/0.8 M NaCl. Gradients were centrifuged at 28K rpm for 22 h at 16 °C (slow acceleration and deceleration) using a SW55TI rotor in an Optima L-90K Beckman ultracentrifuge. Fractions were collected from the gradient and analyzed by electron microscopy and loading on a polyacrylamide gel for L1 protein detection and quantitation. Alexa Fluor 488 (A20000) and Alexa Fluor 647 (A20006) (ThermoFisher) were used to label VLPs and capsomeres as described here: https://ccrod.cancer.gov/confluence/display/LCOTF/DyeConjugation. The dye-coupled VLPs and capsomeres were purified postlabeling on Optiprep or sucrose gradients, respectively. Fractions were collected from the gradient and analyzed by electron microscopy and loading on a polyacrylamide gel for L1 protein detection and quantitation.

Unlabeled goat Ig F(ab′)2 anti-human Kappa (Southern Biotech) was used as a surrogate antigen. Antibodies used for flow cytometry are as follows: goat Fab anti-human Igµ heavy chain-Alexa Fluor 647 or–Alexa Fluor 488 (Jackson ImmunoResearch), anti-human Igκ light chain-PE (BD Biosciences), and κ + λ-Alexa Fluor 488 by unlabeled goat Ig F(ab′)2 anti-human Igκ and λ (Southern Biotech) 1:1-mixed and then conjugated with Alexa Fluor 488 dye using Lightening-Link Rapid antibody labeling kit (Novusbio) for surface BCR labeling, CD19-BV605, CD27-PE/Cy7, CD10-BV711, CD27-BV785 (clone O323), IgM-BV421 (clone MHM-88), CD95-PEDazzle™594 (clone DX2), CD38-BV711 (clone HIT-2), CD69-PE-Dazzle 594 (Clone FN50) (BioLegend), CD19-AF700 (clone SJ25-C1, Southern Biotechnology), CD11c-BUV737 (clone HL3) and IgD-BV480 (clone IA6-2), IgG-BUV395 (clone G18-145) from BD Biosciences and Brilliant Stain buffer (BD Horizon) for A24 Ramos cell characterization, and Janelia Fluor 549-conjugated rabbit IgG anti-Piezo1 (Novusbio, NBP1-78446JF549) for the quantification of total Piezo1 protein expression in Piezo1-KD cells. For the isolation of A24M04 BCR⁺ Ramos B cells using magnetic bead, anti-human Igµ heavy chain-PE, and Igκ light chain-PE (BD Biosciences) are used. Reagents used for flow cytometry are a calcium indicator dye, CAL520, sodium salt (AAT Bioquest), Brilliant Stain buffer (BD Horizon), and LIVE/DEAD^TM Fixable Near-IR (780) or violet dye (ThermoFisher) for dead cell exclusion. Piezo1 inhibitors, GsMTx4 and OB-1, were obtained from Tocris Bioscience. For the confocal imaging, Alexa Fluor 488-conjugated HPV16 VLPs and HPV16 capsomeres, mouse mAbs anti-LAMP1-Alexa Fluor 549 (Santa Cruz, Clone H4A3) and anti-HLA-DR-Alexa Fluor 647 (BioLegend, Clone LN3) were used. For Piezo1 KD, ON-TARGETplus Human PIEZO1 (9780) siRNA-SMARTpool (Dharmacon, L-020870-03-0005) was used. As a control siRNA, ON-TARGETplus Non-targeting Pool (Dharmacon, D-001810-10-15) was used. Culture media containing purified mouse mAb anti-human CD40 (Biolegend, CL 5C3), IL-4 (25 ng/mL, R&D), and IL-21 (100 ng/mL, R&D) was used as T helper (Th) media as described before (34) except for the modification of 3 µg/mL anti-human CD40 mAb.

Generation of A24M04-Expressing Ramos B Cell Line.

Generation of A24M04-expressing Ramos cells was described in detail in SI Appendix, Materials and Methods. Briefly, BCR-negative Ramos cells were isolated from Ramos.2G6.4C10 B cells (WT Ramos) as described (12, 21). The lentiviral transduction was performed for the generation of stable Ramos cell line expressing HPV16-specific A24M04 mIgM heavy chain and κ light chain (A24 Ramos) using the Lenti-X 293 T cell line with Lenti-X packaging system (Takara Bio). The cells expressing A24M04 IgM BCR were isolated using magnetic beads, first, by enrichment with anti-human Igµ heavy chain-PE (BD Biosciences) followed by the EasySep Human PE immunomagnetic positive selection kit (STEMCELL technologies 17664), and then by anti-human Igκ light chain-PE followed by the EasySep Human PE immunomagnetic positive selection kit according to the manufacturer’s instructions.

Piezo1 KD by PIEZO1-Specific siRNAs.

For transient Piezo1 KD in A24 Ramos cells, Amaxa 4D nucleofector with solution SG was used with Piezo1-specific siRNAs according to the protocol. Cells were prepared for the nucleofection by replacing with fresh culture media at the density of 0.5 × 10⁶ /mL a day before the nucleofection. For the optimization of siRNA concentration in A24 Ramos cells, first, siGlo-Red transfection indicator, which is a fluorescent oligonucleotide duplex that localizes to the nucleus (Horizon), was used at three different concentrations (0.5, 1.0, and 1.5 µM) and 1 µM of either control or Piezo1 siRNAs was used for the Piezo1 KD. Briefly, cells were centrifuged with 100× g for 10 min and completely removed the solution and resuspended with 100 µL supplementary reagent-mixed SG solution and 2 µL of control or Piezo1 siRNA, transferred into a cuvette, and transfected using program CA-137 in 4D nucleofector X unit. Cells were cultured for 3 d. Piezo1 down-expression was confirmed by flow cytometry using Piezo1-specific antibodies in the cells. These cells were used for the BCR down-modulation and confocal imaging experiments.

Competitive Antigen-Binding Assays.

Cryopreserved PBMCs were thawed, washed in fluorescence-activated cell sorting (FACS) buffer (2.5% FBS-PBS) at 4 °C, filtered, and counted. A24 Ramos cells were diluted into the PBMCs at a ratio of 1:100 and then washed and the mixed cells were resuspended in 50 µL of FACS buffer and were incubated with unlabeled HPV16 VLPs, BPV VLPs, or HPV58 VLPs (4 µg/mL or 20 µg/mL) for 10 min on ice in the dark. Cells were washed with 2 mL FACS buffer, then incubated with 0.1 µg of each labeled VLP-AF647 or VLP-AF488 for 45 min on ice in the dark. After cell surface markers staining, data were acquired with the BD FACSymphony cytometer and analyzed with the BD FACSDiva software. 50,000 CD19⁺ live single B cells were acquired for each sample. HPV-specific Ramos B cells were defined as CD10^hi, CD27^hi in contrast to PBMC-derived B cells, which expressed lower levels of CD10 and CD27. For unlabeled-capsomere inhibition of labeled-HPV16 VLPs binding assay, the process above was followed except that cells were resuspended in 50 µL of FACS buffer and were incubated with either 2 µg/mL or 4 µg/mL of unlabeled capsomere for 45 min on ice in the dark. Cells were washed with 2 mL FACS buffer, then cells were incubated with 0.1 µg of each labeled HPV16 VLPs (VLP-AF647 or VLP-A488) for 45 min on ice in the dark. Cell surface markers-labeled- and cell viability indicator dye-stained cells were analyzed with the BD FACSymphony cytometer as described above.

Calcium Flux Assay by Flow Cytometry.

Calcium flux assay using CAL520 calcium dye in B cells was described previously (12, 35). Briefly, CAL520 at 2 µM final concentration and Live/Dead Fixable violet viability dye-labeled A24 Ramos cells were resuspended at 5 × 10⁶/mL density in Medium 199 (ThermoFisher Scientific) and recorded for the baseline fluorescence for ~30 to 60 s followed by data acquisition after adding the specified antigens up to ~600 s by flow cytometry. Calcium fluxes were calculated as a fold change of the CAL520 MFI per unit time of area under the curve after antigen stimulation (MFI_antigen) relative to the CAL520 MFI per unit time for the period of the baseline (MFI_baseline), from the kinetic calcium graph using FlowJo™ 10 software (FlowJo, LLC). For the measurement of calcium flux in the presence of Piezo1 inhibitors, cells were pretreated with the Piezo1 inhibitors OB-1 (20 µM) for 1 h, or GsMTX4 (5 µM) for 30 min at 37 °C. Bar graphs for the fold changes were constructed using Prism (GraphPad).

Testing Cell Activation and Death by Flow Cytometry.

To test HPV16 VLPs or capsomeres-induced B cell activation and death, and the cell rescue from AICD by Th media, 2 × 10⁵ cells per well of A24 Ramos cells were placed in 96-well plate and incubated with RPMI culture media only, Th media only, and 10 µg/mL HPV16 VLPs or HPV16 capsomeres in the absence or presence of Th media for 6, 12, 24 h. Then, cells were placed on ice, washed once with ice-cold 1× PBS and stained with mouse mAb anti-CD69-PE-Dazzle and LIVE/DEAD™ Fixable Near-IR for 30 min on ice. B cell activation was analyzed by the percentage of CD69-expressing cells gated against unstimulated cells within singlet, live cell population and the live cells by the percentage of LIVE/DEAD™ Fixable Near-IR dye-excluded cells within total singlet population excluding debris using flow cytometry.

BCR Down-Modulation Assay by Flow Cytometry.

Control- or Piezo1-KD A24 Ramos cells were placed in wells of 96-well plates with 2 × 10⁵ cells per well in Medium199 and incubated with 5 µg/mL of either HPV16 VLPs or capsomeres on ice for 30 min. Then, warm Medium 199 was added and cells were incubated for either 5, 30, 60, or 90 min, followed by fixation with 4% PFA. After washing with 1× PBS, surface BCRs were labeled with 10 µg/mL of goat Fab anti-Igµ-Alexa Fluor 488 (Jackson ImmunoResearch) and goat Fab′₂ anti-Igκ-Alexa Fluor 647 (Southern Biotech) for the detection of IgM HC and Igκ LC specific for A24 BCR on cells. Surface Igµ and Igκ gMFI was analyzed by flow cytometry and the ratio of the surface IgM HC and Igκ LC at each time point (T = t) relative to those on ice (T = 0) was calculated (T = t/T =0).

Confocal Microscopy and Image Analysis.

For confocal microscopy to investigate antigen trafficking into the LAMP1 and/or HLA-DR positive organelles in control or Piezo1-KD A24 Ramos cells, the cells resuspended in serum-free culture media were allowed to attach on poly-L-Lysine-coated 18-well-chambered cover glass (ibidi) by placing in 37 °C CO₂ incubator for 20 min, and incubated with 5 µg/mL of Alexa Fluor 488-conjugated HPV16 VLPs or capsomeres for 120 min in Med199. The cells were fixed with 4% PFA for 20 min at RT, permeabilized with 0.1 % Triton-100 for 10 min at RT and then incubated with 10% normal mouse serum (Southern Biotech) for 10 min at RT for blocking nonspecific binding followed by the incubation with Alexa Fluor 549-conjugated mouse mAb anti-LAMP1 (1:50 dilution) and Alexa Fluor 647-conjugated mouse mAb anti-HLA-DR (1:10 dilution) at 4 °C overnight. Confocal imaging was performed on the ZEISS 880, AxioObserver confocal laser scanning microscope (LSM). Three color 3D images were collected by the excitation of Alexa Fluor 488, Alexa Fluor 549, and Alexa Fluor 647 fluorophores with 488, 561, and 633 nm laser lines, and the emission with 500 to 532 nm, 568 to 630 nm, and 660 to 758 nm, respectively at 2,048 × 2,048 pixels scanned with a pixel size of 43.9 nm and 43.9 nm for X and Y direction and 130 nm z sections covering whole cells, using a 1.4 NA plan apochromat 63× oil objective. Each pixel was scanned four times, and the average value per pixel was recorded. Images were captured with Zen black software. Huygens professional software (version 24.04, Scientific Volume Imaging) was used for the deconvolution of the z stack images obtained from ZEISS 880 confocal LSM according to the deconvolution wizard program with default setting except for setting the maximum iteration number to meet the criterion as 40. When automatic signal-to-noise ratio (SNR) was not good quality, the SNR value was manually adjusted using mean fluorescence intensities of signal-positive pixels and background area. Deconvolved images were cropped for single cells and Manders overlap coefficient per cell was analyzed using colocalization analysis program with optimized background search option in Huygens software. The optimized background search option was used to set the background threshold by the algorithm that finds the point where the Pearson coefficient of the area below the background setting is zero to avoid bias influencing the quantification of ratio of antigen with LAMP1 or HLA-DR and the ratio of LAMP1 colocalized with HLA-DR. For the acquired images, we used Costes’ significance test in coloc2 of Fiji software (Fiji contributors) for quality check of overlap by chance.

Imaris software (Andor Technology Ltd.) was used for the reconstruction of 3D surface views of HPV16 VLPs or HPV16 capsomeres, LAMP1, and HLA-DR, and XY, XZ, and YZ sectional views for the representative single cell. For the quantification of the colocalized antigen with LAMP1 and HLA-DR double positive organelles, first LAMP1 and HLA-DR colocalization analysis was performed to obtain the colocalized voxels between LAMP1 and HLA-DR. Then, Manders overlap coefficient was calculated using the antigen images and the colocalized voxel images between LAMP1 and HLA-DR according to the colocalization program in Imaris with background values taken from mean values on the area without cells.

Statistical Analysis.

Details of the statistical methods that were used for each experiment were indicated in the relevant figure legend. For data expressed as fold, values were transformed to the log scale and statistical analysis performed and P < 0.05 was considered statistically significant. The Prism software (GraphPad) was used to calculate all statistical analyses. For BCR down-modulation assay, statistical significance (95% CI) was assessed with the compareGrowthCurves function from the statmod statistical modeling package (26).

Supplementary Material

Appendix 01 (PDF)

Dataset S01 (XLSX)

Movie S1.

Time-lapse confocal imaging shows stable association and clustering between the BCR and HPV16 VLPs on A24 Ramos B cell. BCR- and CD45-labeled A24 Ramos B cells were placed in 18-well chambered cover glass and live cell confocal imaging was performed using AXR NSPARC confocal microscope. Time-lapse images were obtained with 30 s interval for 30 min at 37 ° C, beginning with 2 min after adding 5 μg/mL concentration of Alexa Fluor 647-conjugated HPV16 VLPs. Shown is a 2-color time-lapse video for BCR (green) and HPV16 VLPs (red) made at 10 fps for representative cell of two experiments.

Data, Materials, and Software Availability

All study data are included in the article and/or supporting information.