Electrostatic Surface Properties of Blood and Semen Extracellular Vesicles: Implications of Sialylation and HIV-Induced Changes on EV Internalization

2.1. Ethical Approvals

The Institutional Review Board (IRB) of the University of Iowa and Stony Brook University approved the use of human specimens (IRB numbers 201608703 and 1106876, respectively). HIV- and HIV+ subjects consented to participate in this study via written informed consent. All samples were received unlinked to any identifiers. All experiments were performed in accordance with the approved university guidelines and regulations.

2.2. Participants

The clinical and demographic characteristics of HIV+ (n = 13) and HIV- (n = 3) donors of matched blood and semen used in this study were previously published [38]. HIV- donors had no history of HIV, hepatitis B virus (HBV), and hepatitis C virus (HCV) infection at time of participation. HIV+ donors were ART-suppressed with viral load <50 copies/mL at the time of specimen donation [38].

2.3. Cells

TZM-bl cells were obtained through the NIH AIDS Reagent Program and maintained in Dulbecco’s modified Eagle’s medium (DMEM) (Gibco-BRL/Life Technologies) supplemented with 10% fetal bovine serum (FBS) (Atlanta Biologicals, Flowery Branch, GA, USA) that was EV-depleted by 16 h ultracentrifugation at 100,000× g, 1% Penicillin-streptomycin (Thermofisher, Grand Island, NY, USA), 1 µg/mL Amphotericin B (Thermofisher), 2 mM sodium pyruvate (Corning, Corning, NY, USA), 1% of glutamate (Thermofisher), and 10 mM HEPES buffer (Fisher Biotech, Fair Lawn, NJ, USA) at pH 8. Isolation of primary cells was performed as previously described [37,39,40]. Frozen primary peripheral blood mononuclear cells (PBMCs), isolated from blood collected from 3 healthy donors, were thawed at 37 °C and individually washed with complete Roswell Park Memorial Institute (RPMI) media (Corning). Cell viability was determined by the trypan blue method to be >90%. Cells were plated in 6-well plate (Cyto-One, USA Scientific) at 3 × 10⁶/mL, and allowed to adhere in a 5% CO₂ incubator at 37 °C for 2 h. Non-adherent cells were collected as Peripheral blood lymphocytes (PBLs) and cultured in a separate plate. Adherent cells were carefully washed with RPMI and cultured as primary monocytes.

2.4. EV Purification from Human Semen and Human Blood

The isolation and characterization of the EVs used in this study was recently described [38]. Briefly, blood or seminal plasma were centrifuged at 2000× g for 10 min and 10,000× g for 30 min, to pellet cellular debris. ExoQuick (EQ, System Biosciences, Palo Alto, CA, USA) was added at a ratio of 4:1 (plasma: EQ) and incubated at 4 °C overnight. The mixture was then centrifuged at 1500× g for 30 min at 4 °C and the EQ/EV-free seminal plasma was removed. Residual EQ was removed by centrifuging the EV pellet at 1500× g for 5 min and discarding the supernatant. The EV pellet was re-suspended in PBS in 1/10 of the original volume. Protein quantification was determined by NanoDrop absorbance at 280 nm, and samples were frozen at −80 °C until use.

2.5. ζ-Potential Measurements

ζ-potential measurements were performed using ZetaView PMX 110 (Particle Metrix, Mebane, NC) and corresponding software (version 8.04.02). EVs were diluted in ultrapure water (1:20,000–1:320,000) to reach to a particle number per frame of 50 to 300, ideal for Nano Tracking Analysis (NTA). Temperature was set at 25 °C and 5 cycles at high sensitivity setting were performed per measurement for a total of three to ten measurements per sample. As a control, 10 measurements were collected for representative samples of each group to test sample stability during measurements. Camera settings for all measurements were fixed at a sensitivity and a shutter of 91 and 70, respectively. The high sensitivity settings were recommended by the manufacturer to capture the most EVs.

2.6. pH Dependence

In total, 2 µL of pooled (n = 10) HIV- or HIV+, BEV or SEV, diluted 1:20,000 in 40 mL H₂O, were titrated with 10 µL NaOH (10 mM) at a time. The pH was recorded, and the ζ-potential was measured in quintuplets.

2.7. EV Labelling

In total, 300 µg pooled (n = 10) HIV- or HIV+, BEV or SEV were diluted with 1× PBS to 499 µL, to which 1 µL ExoGlow protein labelling dye (System Biosciences) was added, and the mixture was incubated at 37 °C for 20 min with constant shaking (1000 rpm). A 1x PBS solution was used as a negative control. Labelled EVs were precipitated by 167 µL EQ TC (System Biosciences) for 2 h at 4 °C and re-concentrated by centrifugation at 10,000 g for 10 min. Labelled EV pellets were dissolved in 1x PBS and their particle concentration was measured by NTA.

2.8. EV Internalization

TZM-bl cells were seeded overnight in a 96-well plate (Cyto-One, USA Scientific) at 10,000 cells per well. ExoGlow labelled EVs (100 µg/mL, corresponding to ~2–3 × 10¹⁰ particles/mL) were diluted in complete DMEM to which 20 µL/mL NucBlue Live ReadyProbes (EasyProbes, Thermofisher) was added. Kinetic imaging of live cells was performed using LionHeart (BioTek, Winooski, VT, USA) with a 10× objective and images of 4 adjacent fields of view were taken. Images were stitched and fluorescence in the Green-fluorescence protein (GFP) channel was measured within a secondary mask that was set at 6 µm expanding from a primary mask, which was set in the DAPI channel and reduced by 4 µm. This quantification strategy reduced the background signal from non-internalized EVs and accounted for any potential differential cell growth. Subsequently, cells’ background was subtracted, and fluorescence values were normalized relative to labelled PBS control. The luciferase assay post-internalization was performed according to a published procedure [41].

PBLs and monocytes were activated overnight by plate-bound anti-CD3 (10 µg/mL) and soluble anti-CD28.8 (2 µg/mL) (BioLegend, San Diego, CA, USA). To activated PBLs and monocytes, ExoGlow labelled HIV- BEV or SEV, or labelled vehicle control, were added at 100 µg/mL (~2–3 × 10¹⁰ particles/mL) and kinetic imaging of live cells were recorded every three hours for 24 h. Total green fluorescence was measured, cells background was subtracted, and values were normalized to labelled PBS control. The MTT viability assay post-internalization was performed according to published procedure [41].

2.9. α-Neuraminidase Treatment

In total, 100 µg of pooled (n = 10) HIV- or HIV+, BEV or SEV was mixed with 250 units of α2-3,6,8 Neuraminidase (New England BioLabs, Ipswich, MA), or equivalent volume of PBS control, in 100 µL GlycoBuffer1 (5 mM CaCl₂, 50 mM sodium acetate, pH 5.5 at 25 °C) and the reaction was incubated at 37 °C for 1 h. Samples were then inactivated at 65 °C for 10 min, according to the manufacturer’s instructions, before measurement of ζ-potential, EV labelling, and internalization assessment.

2.10. Statistics

Graphpad Prism (v 8.4.2) was used to plot all the graphs and to determine the statistical significance. For a two-group comparison, an unpaired t-test with Welch’s correction was used to determine the differences between the groups (Figures 2c,e,f, 5b,d and S1a,c,d), unless all samples were matched in which case a paired and parametric t-test was employed (Figures 2b and S1b). For multiple comparisons, ordinary one-way ANOVA test with Dunnett’s correction was used to determine the differences between the groups as compared to the control (Figures 3d, 4c,d and 5d). For 2-variable multiple comparisons, two-way ANOVA with Sidak’s correction was used and the predicted mean difference with adjusted p-value reported (Figure 4e). Error bars represent either standard deviation (S.D.) across multiple measurements, or standard error of the mean (S.E.M.) across independent experiments.

3.1. ζ-Potential of BEV and SEV Is Stable and Time- and Concentration- Independent

ζ-potential corresponds to the electric potential near the surface of a charged particle [42] (Figure 1A), and represents the charge of the particle in relation to its surrounding milieu. ζ-potential is measured in ranges where values between −10 to +10 mV are considered neutral while values > +30 mV or < −30 mV are considered strongly cationic or strongly anionic, respectively [21] (Figure 1B). ζ-potential of EVs can be measured using Nanoparticle Tracking Analysis (NTA) [43]. NTA, in general, requires diluted samples, in the range of 10⁷ to 10⁹ particles/mL [44], which corresponds to ~0.1 or 10 µg protein/mL in the case of BEVs and SEVs, respectively. Therefore, purified and pooled BEVs and SEVs from HIV-infected (HIV+, hereafter designated as HIV+ BEV, HIV+ SEV) and uninfected (HIV-, hereafter designated as HIV- BEV, HIV- SEV) donors were diluted 1:20,000 to 1:160,000 in filtered ultrapure water until the concentration was within the range for suitable measurements. This range corresponds to 50–300 average particles per frame when using ZetaView (Table S1). First, to ensure that the BEV and SEV preparations were suitable for ζ-potential measurements, we acquired 10 measurements of 5 cycles each per sample (Figure 1C). Numerous acquisitions are typical for testing stability of samples during measurement [45]; if ζ-potential values fluctuate during measurement, it is an indication that the sample is unstable i.e., dissociating, aggregating, sedimenting, or solubilizing [46]. We found that the variation in ζ-potential for all four samples (pool of n = 10, HIV−BEV, HIV−SEV, and HIV + BEV, HIV + SEV) was <5 mV (Figure 1C), indicating that the EVs (BEVs and SEVs) were stable during the measurement period. Second, we confirmed that BEV and SEV ζ-potential measurements are within the suitable concentration range by performing two measurements for each sample at the high- and low- ends of the concentration range recommended by the manufacturer. Although Debye–Hückel–Henry’s equation for ζ-potential determination does not have concentration as a variable [47], it is only applicable in a narrow range of concentration. Thus, drastic dilutions, which are required for NTA, may largely affect the quality of the measurements leading to potential erroneous interpretations [46,48]. Nonetheless, we observed no significant differences between the two dilutions (Figure 1D), indicating that BEV and SEV ζ-potential measurements were indeed concentration-independent. Third, ζ-potential depends on the protonation state at the surface of the measured particle and is thus largely affected by the pH of the solution. Therefore, we constantly monitored the pH of all measured samples which were diluted with ultrapure water. Buffered solution such as PBS was not used because the salts in PBS would build up through the tubing and connections of the NTA machine to skew measurements. The pH during measurements ranged between 5.6 and 5.9, similar to the measured pH of ultrapure water (pH = 5.8), which is slightly acidic because of the presence of carbonic acid that results from the dissolution of atmospheric carbon dioxide [49]. In summary, ζ-potential measurement of BEV and SEV is stable and the observed ζ-potential was not influenced by unspecific interactions. Finally, it is known that the size of vesicles is an important parameter that dictates the amount of lipids, surface glycan, and glycoprotein content of a vesicle, and potentially the vesicle surface charge. Consequently, we measured the particle sizes of BEVs and SEVs for all samples individually. The result shows no significant differences between BEVs and SEVs size irrespective of donor HIV status (Figure S1).

3.2. BEVs Bear Neutral Surface Compared to the More Anionic SEVs

ζ-potential of purified HIV- BEV and HIV- SEV from 10 unmatched HIV- donors was measured using ZetaView and quintuplet measurements for each sample were recorded. HIV- BEVs were neutral (−5–+4.8 mV), whereas HIV- SEV were negatively charged (−33.5–−8.8 mV) Figure 2a and Figure S2). To determine whether the difference in ζ-potential between BEV and SEV is universal or donor-dependent, we analyzed 3 additional HIV- BEV and HIV- SEV from matched donors (Figure 2b and Figure S2). Statistical analysis confirmed that HIV- BEV are generally neutral (mean ± SD = 0.67 ± 3.7) while HIV- SEV are negatively charged (mean ± SD = −17.09 ± 8.7), with a p-value of the difference <0.0001 (Figure 2c). This result suggests that the difference in ζ-potential between HIV- BEVs and HIV- SEVs may be due to membrane compositional differences between HIV- BEVs and HIV- SEVs.

3.3. HIV Infection Increases the Net Negative Charge on the Surface of BEVs and SEVs

To identify the effect of HIV infection on the electrostatic surface properties of BEV and SEV, purified HIV+ BEV and HIV+ SEV from 13 matched HIV infected ART-suppressed donors were analyzed for ζ-potential. HIV+ SEV were more negatively charged (−25.83 ± 4.4 mV) compared to HIV + BEV (−2.06 ± 3.2 mV) as expected (Figure 2d,e, Figure S2). Although BEVs remained generally neutral and SEVs negatively charged, a comparison of the ζ-potential values of HIV + EV with those from HIV- EVs indicated that HIV+ EVs tend to be slightly more negatively charged compared to HIV- EVs (Figure 2f). A two-tailed unpaired t-test between the two groups indicated that the observed difference was significant (p = 0.0462 and p = 0.0062, for BEVs and SEVs, respectively). To confirm this result, a pH dependence experiment was conducted in which a pool of EVs from each experimental group (n = 10) was pH-titrated and ζ-potential measured (Figure 2g). To assign numerical values to the surface charge of EVs, the isoelectric point (IEP) which corresponds to the pH at which the surface is neutral was determined graphically and is summarized in Table 1. The result confirmed the differences between HIV- and HIV+ groups for both BEVs and SEVs and indicated that HIV infection may indeed alter the EV membrane surface charge by increasing its net negative charge. The reasons for the differences between HIV- and HIV+ are unknown but may include HIV-induced (i) differences in EV composition or (ii) differences in the ratios of EV subtypes release. It is also possible that these differences may be related to the use of ART, since all HIV+ donors are virally suppressed by ART.

3.4. SEVs Internalization by Epithelial Cells is more Efficient Compared to BEVs and EVs Internalization Correlates with Their IEPs

To determine if HIV-induced surface charge modification of EVs impact their downstream function, we investigated the internalization of HIV- and HIV+ BEVs and SEVs. To this end, EVs from HIV+ and HIV- groups were separately pooled and labelled using an EV protein labelling dye. To confirm that labelling does not affect the EVs surface charge, ζ-potential and pH were measured before and after EV labeling (Table 2). Although the numbers varied, the labeling process produced no significant effect on ζ-potential; i.e., HIV- and HIV+ BEV were consistently in the slightly negative to neutral range (−10 to 0 mV), whereas HIV- and HIV+ SEV remained significantly more negative (−20 to −30 mV).

Labeled HIV- and HIV+ BEVs and SEVs or labeled PBS control were mixed with DMEM media containing NucBlue stain and the mixture was added to TZM-bl cells. Kinetics of EV internalization by cells were monitored in Green-fluorescence protein (GFP), bright field, and DAPI channels every hour for 24 h using an automated scope. Representative images of the cellular internalization are shown in Figure 3a. During measurements, the parameters of the GFP acquisition channel were kept constant (LED = 5, acquisition time = 100 ms, and gain = 1), the GFP intensity per cell was determined using a secondary mask strategy set in the GFP channel whereas a primary mask was set in the DAPI channel (Figure 3b). The average intensity per cell in each treatment was normalized to time zero to correct for background and then normalized to that of labelled PBS control, which was set to 1, and the relative GFP intensity was plotted as a function of time (Figure 3c). Internalization efficiency was higher for SEV compared to BEV (Figure 3c). Rate of internalization plateaued earlier with BEVs (at hour 12) compared to SEVs that continued the upward trend up to end of the experiment at hour 24. To understand whether the difference in EV internalization is affected by the presence of infectious HIV particles in the EV preparation, we assessed HIV infectivity by measuring the relative luciferase intensity at the end of the internalization experiment, since TZM-bl cells contain LTR-luciferase reporter used for assessing HIV infectivity. No signal beyond basal luciferase levels was detected in the HIV- or the HIV+ EV treated cells (Figure 3d), indicating that the observed differences in EV internalization efficiency between the two groups do not result from the presence of free infectious HIV. This observation is not surprising because all HIV+ donors were ART-adherent and virally suppressed for 5+ years. Interestingly, we noticed that the relative internalization of BEVs and SEVs significantly correlate with the ζ-potential of the labeled EVs (Pearson correlation: R squared = 0.9264, two-tailed p-value = 0.0375) (Figure 3e). This correlation suggests that EV internalization may be influenced by the EV surface charge with more negatively charged vesicles being more efficiently internalized by cells.

3.5. SEVs Internalization Efficiency by Primary Immune Cells is Higher Compared to BEVs

TZM-bl cells studied in the previous section are epithelial cells from a cervical carcinoma cell line. They may thus be naturally more receptive to SEVs than BEVs. Hence, we sought to assess internalization efficiency of BEVs and SEVs in different cell lines, namely primary immune blood cells, PBLs, and monocytes. Labelled PBS control or labelled HIV- BEV or SEV were mixed with RPMI media at a concentration of 100 µg/mL and the mixture was added to primary PBLs or monocytes isolated from three healthy blood donors. The fluorescence intensity was recorded every 3 h for 24 h using an automated microscope. Total GFP intensity in each treatment was normalized to that of labeled PBS which was set to 1, and the relative GFP intensity was plotted as a function of time (Figure 4a,b). At the end of the internalization experiment, cell viability was measured with the MTT assay (Figure 4c,d). The results show that EVs entered the cells in the first three hours, with SEV being more efficiently internalized compared to BEVs both in PBLs (Figure 4a) and monocytes (Figure 4b), suggesting cell-type independency in EV internalization rate and efficiency. The MTT viability test showed no significant difference between the EV and PBS treated PBLs (Figure 4c) and monocytes (Figure 4d), suggesting that the observed differential internalization efficiency between SEVs and BEVs is independent of cell death. Comparison of internalization efficiency at 24 h amongst the three different cells types shows higher internalization of HIV- BEVs by immune cells as compared to TZM-bl (PBLs mean diff. = 0.278, Adj. p-value = 0.0270; monocytes mean diff 0.2591, Adj. p-value = 0.0449) (Figure 4e). Interestingly, SEV internalization at 24 h was also higher for PBLs and monocytes compared to TZM-bl (PBLs mean diff. = 1.016, Adj. p-value <0.0001; monocytes mean diff. = 1.177, Adj. p-value <0.0001). Taken together, these results suggest that immune cells internalize EVs to a greater degree than epithelial cells, irrespective of the EV type or source. This is not surprising, since immune cells are highly phagocytic. Additionally, EV internalization by target cells appears to be dictated by the charge on the surface of EVs.

3.6. Neuraminidase Treatment Decreases SEV Absolute ζ-Potential and Internalization Efficiency

Considering our findings that EV surface charge correlates with cellular internalization efficiency with more negatively charged EVs being more efficiently internalized, we hypothesized that SEVs, but not BEVs, are enriched with sialic acid residues conferring to SEVs their net negative charge. Indeed, EVs were found to bind with high affinity sialic-acid-binding lectins [31] and sialic acids have a low pK_a of 3.76 in aqueous solution [50]. EV pools from HIV- and HIV+ BEVs and SEVs were subjected to α2-3,6,8 neuraminidase treatment to hydrolyze α2-3, α2-6, and α2-8 linked sialic acid residues from surface glycoproteins. Three types of membrane molecules contain sialic acids: O-glycan, N-glycan and glycolipids such as sphingolipids (Figure 5a). It is important to note that neuraminidase treatment is minimally invasive and does not alter the glycocalyx coat nor the membrane proteins, hence preserving the EV membrane integrity [32]. As expected, sialic acid digestion decreased the net negative charge on HIV- and HIV+ SEVs but did not significantly change the ζ-potential of HIV- or HIV+ BEVs (Figure 5b). Intact and digested EVs were labeled with EV protein labelling dye and labelled EVs were used for internalization experiment in TZM-bl cells. Then, 24 h later, cells were imaged using an automated microscope at a fixed GFP intensity. Primary and secondary masks were set in the DAPI and GFP channels, respectively, allowing GFP intensity tracing per cell (Figure 5c). Analysis of the GFP data shows that SEVs were more significantly internalized as compared to BEVs (Figure 5d, filled bars). Upon neuraminidase treatment, internalization efficiency of both HIV- and HIV+ SEVs significantly decreased, whereas BEVs internalization was not affected (Figure 5d, two-group comparisons).

To gain insight into the differential response of BEVs and SEVs to neuraminidase, we performed a secondary analysis of our previously published Kaddour et al., proteomics dataset of BEVs and SEVs isolated from uninfected or HIV-infected ART-suppressed individuals [51]. The Kaddour et al., study identified 218 and 246 differentially present proteins (DPPs) in SEV as compared to BEVs in HIV- and HIV+ groups, respectively. A 2-way Venn analysis of the Kaddour dataset identified a total of 310 unique SEV-enriched DDPs, of which 154 proteins were common to HIV- and HIV+ SEVs, 64 and 92 proteins were enriched in HIV- and HIV+ SEVs, respectively (Figure 5e). To identify molecular pathways potentially perturbed by the DPPs, the 310 modulated proteins were mapped to KEGG database. Notably, the top two pathways with the highest enrichment ratio (Table 3) include the following terms: other glycan degradation and glycolysis/glycogenesis. This analysis revealed 21 glycocalyx-related SEV-enriched DPPs, including sialidase-1 (NEU1) known to be shed from sperm during capacitation [52] (Figure 5f).