Single-particle analysis of small extracellular vesicles from human follicular fluid unveils immunomodulatory PD-L1⁺ subpopulations and potentially fertility biomarkers

Patient recruitment

Ten patients who underwent assisted reproductive technology (ART) treatment were enrolled in this study (Table 1). Following approval by the Institutional Review Board of the Institute for Maternal and Child Health IRCCS Burlo Garofolo, Trieste, Italy (IRB-BURLO 07/2023 dd. 31.08.2023), patients were asked to sign an informed consent form. The FF from the first and largest punctured follicle from the bilateral ovaries was collected during the oocyte retrieval procedure (transvaginal follicular aspiration). Each ovarian follicle was aspirated independently.

Inclusion criteria: Follicular fluid samples were collected from infertile patients undergoing in vitro fertilization. Exclusion criteria: To ensure sample purity, the presence of blood contamination was assessed by visual inspection. Samples that appeared cloudy or blood-stained were excluded from the study. Only uncontaminated samples were processed for analysis.

Single-particle interferometric reflectance imaging sensing: ExoView R100 analysis

The investigation was performed via single-particle interferometric reflectance imaging sensing (SP-IRIS) via the ExoView R100 system, which enables the detection of particles as small as 50 nm through optical interferometry on Si/SiO₂ substrates integrated into the microarray chip (Mizenko et al., 2021). Initially, 35 µL of each sample, diluted 1:10, were applied to the microarray chip functionalized with capture antibodies specific for CD9, CD81, and CD63 (ExoView™ Tetraspanin Kit, CD9, CD63, CD81 + IgG Negative Control (mIgG); Nano View Bioscience, Boston, MA, USA) and incubated for 16 h at room temperature. The chips were washed three times with solution A from the ExoView Human Tetraspanin Kit. Each binding event is imaged using visible light interference to determine particle size and concentration. Following EV capture, fluorescently labeled antibodies, anti-CD9 CF488, anti-CD81 CF555 (from the ExoView™ Tetraspanin Kit), and anti-PD-L1 [28-8] Alexa Fluor^® 647 (ab209960, Abcam, Cambridge, UK), were applied to detect surface markers. The antibodies were first diluted 1:500 in Solution A (Nano View Bioscience), and then mixed 1:1 with fresh Solution A, achieving a final dilution of 1:1000. Chips were incubated with this antibody mix for 60 min at ambient temperature, then washed and dehydrated. Image acquisition was performed using the ExoView R100 reader and ExoView Scanner 3.0 software. Data were subsequently analyzed using ExoView Analyzer 3.0.

To ensure the reliability and specificity of the immunofluorescence labelling results, the ExoView R100 system integrates crucial internal controls. Fluorescence specificity and intensity cut-offs were established using isotype control mouse IgG (mIgG) in accordance with the manufacturer’s guidelines. These isotype-matched negative controls assess non-specific antibody binding, ensuring that detected signals specifically originate from tetraspanins and PD-L1 expressed on sEVs. Furthermore, the system’s software performs real-time quality checks during data acquisition, automatically flagging samples with elevated background levels. This automated detection is vital for identifying and mitigating technical artifacts such as inadequate washing, particle aggregation, or sample/reagent impurities, thereby enabling researchers to exclude unreliable data or optimize experimental conditions. Anomalous spots were excluded to reduce analytical variability. Co-localization analysis was also performed to determine marker overlap across individual particles (Breitwieser et al., 2022).

Size exclusion chromatography isolation on qEVoriginal 35 nm columns, IZON

sEVs were isolated via size exclusion chromatography (SEC) following the manufacturer’s protocol. Briefly, single qEV original 35 nm columns (Izon Science; distributed by Shaefer-Tec, Italy) were brought to room temperature and equilibrated with 17 mL of freshly filtered (0.22 µm) phosphate-buffered saline (PBS 1X, Corning, Glendale, AZ, USA). FF samples were initially centrifuged at 2,000 × g for 30 min. Subsequently, 500 µL of the FF supernatant was carefully loaded onto the top of the column. Once the sample had fully entered the resin bed, 0.22-µm-filtered PBS 1X was added to the column to initiate elution of EVs. Fractions were manually collected as follows: the void volume (F0) was 1.6 mL, followed by six subsequent fractions (F1 to F6), each consisting of 400 µL. Fractions F1–F6, which are expected to contain sEVs, were pooled and concentrated via ultracentrifugation at 100,000 × g for 70 min. Following SEC, samples were frozen at −80 °C immediately. The fractions were analyzed immediately after thawing, ensuring minimal degradation and aggregation prior to the AFM analysis.

Atomic force microscopy morphometry

Samples obtained via SEC isolation were analyzed via atomic force microscopy (AFM) as described elsewhere (Ridolfi et al., 2020; Ridolfi et al., 2023; Bortot et al., 2022). Briefly, images were taken in PeakForce mode on a Bruker Multimode8 equipped with a Nanoscope V controller, a sealed fluid cell and a type JV piezoelectric scanner using Bruker ScanAsystFluid+ probes (triangular cantilever, nominal tip curvature radius 2–12 nm, nominal elastic constant 0.7 N/m) calibrated with the thermal noise method. Image analysis was performed with a combination of Gwyddion 2.58 (Nečas & Klapetek, 2012) and custom Python scripts to recover the equivalent spherical diameter of individual particles. All AFM images were acquired at combinations of scan size and pixel size resulting in a fixed resolution of 9.77 nm/pixel. Scanning speed was kept below five µm/s. Particles were selected as described in (Ridolfi et al., 2020; Ridolfi et al., 2023); briefly, a height threshold of eight nm was applied to single out all particles from the background, then putative vesicles were selected for subsequent morphometrical analysis by removing particles that (i) touched any edge of the image, (ii) had a maximum inscribed disc radius below eight nm, (iii) had a maximum value below 10 nm. At least 100 individual particles were measured for each sample.

Statistical analysis

All statistical analyses were performed using Python, leveraging the scipy.stats and statsmodels libraries. Graphs were performed using GraphPad Prism 10.5.0. Data are presented as mean ± standard deviation (SD) or as percentages, as indicated in the figure legends and text.

For comparisons of particle counts between total particles and those detected via imaging mode (as illustrated in Figs. 1 and 2), paired t-tests were applied to triplicate analyses for individual samples and across the entire cohort, respectively. To assess differences in total particle numbers captured by different tetraspanin antibodies (CD63, CD81, CD9), as shown in Fig. 3, paired t-tests were utilized for direct comparisons between specific capture antibody groups. For the analysis of PD-L1 co-expression ratios across CD63, CD81, and CD9 (as presented in the relevant results section), a repeated measures Analysis of Variance (ANOVA) was conducted to determine if there was an overall statistically significant difference among the mean values of these ratios. Following a significant ANOVA result, paired-samples t-tests were performed as post-hoc analyses to investigate specific pairwise differences between the ratios. Coefficients of variation (CV) were calculated for PD-L1 co-expression ratios across the 10-patient cohort to assess inter-individual variability and consistency of sEV subpopulations. The CV is a standardized measure of relative variability, calculated as the ratio of the standard deviation to the mean, expressed as a percentage. A lower CV indicates greater consistency within the data set. A significance level of α = 0.05 was set for all statistical tests. Specific p-values are reported in the Results section.

ExoView R100 analysis reveals the presence of sEVs with a distinct pattern of PD-L1 and tetraspanin profiles in FF

A comprehensive characterization of the tetraspanin expression profile in FF sEVs has not been accomplished; hence, we performed this investigation using a platform that minimized biases derived from the sEV isolation step and was capable of characterizing vesicles smaller than 50 nm (Mizenko et al., 2021). We have associated the analysis of tetraspanins with that of our target marker, PD-L1. The ExoView R100 platform was used to quantify distinct EV populations by combining two complementary systems: antigenic capture via interferometric imaging and fluorescent antigenic detection of the captured EV. Single-particle interferometric reflectance imaging (IM) enables the quantification of particles bound to the capture spot between 50 and 200 nm (Avci et al., 2015), whereas fluorescence mode was implemented to quantify the particles as small as 35 nm. The concurrent use of three distinct fluorescence-labelled antibodies enabled the assessment of antigen distribution on a single sEV (Fig. 1).

As shown in Fig. 1, a representative example of the EXOVIEW 100 output from one patient sample, the number of particles detectable in imaging mode was significantly lower than the total number of particles detected in fluorescence mode, consistent with the fact that particles larger than 50 nm accounted for less than 10% of the total population. A paired t-test was applied to triplicate analysis comparing total particles and those detected via imaging mode, revealing statistically significant differences. Specifically, in this representative example, CD63 yielded p = 1.40  × 10⁻⁶ (imaging mode: 9%), CD81 p = 9.92  × 10⁻⁷ (imaging mode: 5%), and CD9 p = 1.81  × 10⁻⁵ (imaging mode: 7%). Furthermore, analysis of the entire cohort (Fig. 2) consistently demonstrated that the majority of sEVs were smaller than 50 nm. Specifically, particles under 50 nm constituted 10.38% of the total for CD63-captured EVs (p = 5.60  × 10⁻⁷), 5.15% for CD81-captured EVs (p = 3.02  × 10⁻⁹), and 8.86% for CD9-captured EVs (p = 5,3  × 10⁻⁸).

Figure 3 illustrates the overall abundance and distribution of sEVs captured on different tetraspanin antibodies from the full patient cohort. The bar chart, titled “Total particle number,” displays the average particle counts on the CD63, CD81, and CD9 antibody capture spots. Each bar represents the mean particle count, with individual data points (dots) showing the counts from each patient in the cohort, and error bars indicating the standard deviation. Statistical analysis reveals a significant difference in the total number of captured particles: specifically, the number of particles captured by anti-CD9 antibodies is significantly lower than those captured by anti-CD81 antibodies (p = 0.0059) and anti-CD63 antibodies (p = 0.0047).

Figure 4 illustrates the multi-marker co-expression profiles of sEVs within the follicular fluid, derived from the full patient cohort. It provides a detailed breakdown of the percentage of sEVs exhibiting various combinations of tetraspanins (CD63, CD81, CD9) and PD-L1, relative to their respective capture antibodies. A small but detectable proportion of sEVs exhibited PD-L1 expression on the CD9, CD63, and CD81 antibody capture spots. This highlights the presence of PD-L1-bearing sEVs within the follicular microenvironment, a finding of particular interest given their potential immunomodulatory roles. While PD-L1 expression is observed, its overall fraction remains lower compared to the robust tetraspanin co-localization. The detailed breakdown across various capture and detection combinations provides nuanced insights into the complexity of the sEV proteome in follicular fluid.

Characterization of PD-L1 expression profile of sEVs in FF

The analysis of the PD-L1 expression profile revealed a preference for colocalization with CD9⁺ sEVs (Fig. 5). In terms of the three capture spots, CD63, CD81, and CD9, the percentages of EVs expressing PD-L1 were 2,43%, 2,62% , and 4,77% respectively (Fig. 5). A repeated measures analysis of variance (ANOVA) was conducted to assess if there was an overall statistically significant difference among the mean values of these three PD-L1 ratios. The ANOVA revealed a statistically significant overall difference among the markers (F(2,18) = 5.71, p = 0.0120). To further investigate these differences, paired-samples t-tests were subsequently performed as post-hoc analyses. The findings indicate that while the PD-L1/CD63 and PD-L1/CD81 ratios do not significantly differ from each other (t(9)= −0.51, p = 1.00), both are significantly different from the PD-L1/CD9 ratio. Specifically, a statistically significant difference was observed for PD-L1/CD63 vs. PD-L1/CD9 (t(9)= −4.99, p = 0.0024). Similarly, a statistically significant difference was also found for PD-L1/CD81 vs. PD-L1/CD9 (t(9)= −4.45, p = 0.0045) (Fig. 6). These results confirm a preferential association of PD-L1 with CD9⁺ sEVs.

We observed that PD-L1 mostly colocalized with a single tetraspanin, with percentages of 1.88%, 1.95% and 3.98%, respectively, while the percentage of EVs exhibiting PD-L1 in combination with multiple tetraspanins was less than 1% (Fig. 6). Coefficients of variation (CV) across the 10-patient cohort were 0.64 for PD-L1/CD63, 0.62 for PD-L1/CD81, and 0.18 for PD-L1/CD9, indicating that PD-L1/CD9 co-expression was the most consistent among patients. This lower CV suggests reduced inter-individual variability and a more stable association between PD-L1 and CD9⁺ sEVs, potentially reflecting a conserved biological subpopulation. In contrast, the higher CVs for PD-L1/CD63 and PD-L1/CD81 point to greater heterogeneity, which may be influenced by patient-specific factors or follicular microenvironmental conditions. These findings suggest that PD-L1⁺ sEVs are predominantly associated with individual tetraspanin subpopulations, particularly CD9⁺ vesicles, and that multi-marker PD-L1⁺ sEVs are rare and potentially more heterogeneous.

It has recently been demonstrated that EVs exhibit a distinct nanomechanical fingerprint that can be identified using force spectroscopy. This fingerprint can be used to distinguish between various EV populations, making atomic force microscopy (AFM)-based nanomechanics a useful tool for evaluating EV identity, purity, and function (Bortot et al., 2021; Ridolfi et al., 2023). Single-particle AFM morphometry performed on SEC-enriched sEV samples revealed that over 50% of the particles had diameters smaller than 50 nm, while no significant amounts of vesicles with diameters above 100 nm were observed (Fig. 7).