Subcellular elements responsive to the biomechanical activity of triple-negative breast cancer-derived small extracellular vesicles

Cell cultures

Both MDA-MB-231 and MCF7 breast cancer cell lines were cultivated in DMEM (Dulbecco’s Modified Eagle’s Medium High Glucose with Sodium Pyruvate with L-Glutamine, EuroClone, ECM0728L) supplemented with 10% FBS (Fetal Bovine Serum South America origin EU, EuroClone, ECS0180L) and 1% penicillin/streptomycin (100X, EuroClone, ECB3001D). Cells were grown at 37°C in a humidified 5% CO₂ incubator and split every 2–3 days according to their confluence. Culture and harvesting conditions (such as passage number and seeding confluence) were maintained the same, and regular checks for Mycoplasma contamination were performed.

sEV isolation

For sEV isolation, MDA-MB-231 cells (2 × 10⁶) were grown in 175 cm² flask in DMEM with 10% FBS for 3 days, and then, the cells were washed twice with PBS and then three times with DMEM without serum. sEVs released by the cells in this medium (DMEM without serum) in the last 24 h were collected. Cell and cell debris present in the medium were centrifuged at 300g for 10 minutes at 4°C. The resulting supernatant was filtered using a 0.2 µm filter and transferred into Amicon Ultra-15 centrifugal filters (Ultracel-PL PLHK, 100 kDa cutoff, Merck Millipore, UFC9100) and centrifuged at 4,000g for 40 min at 4°C. The ultracentrifuge tubes (Beckman Coulter, 361623) were filled with the concentrated samples and PBS to reach the final volume and were ultracentrifuged at 120,000g for 120 min at 4°C (70.1 Ti rotor, k-factor 36, Beckman Coulter, Brea, CA, USA). Finally, the supernatant was removed, the pellet was resuspended in PBS, and sEVs were stored at +4°C or −80°C for short-term periods.

Functional experiment: small EV treatment of target cell

MCF7 cells (2.5 × 10⁴) were seeded in a 24-well plate on 13 mm Ø glass slices or CaF₂ optical windows (only for Fourier transformed infrared spectroscopy (FTIR) analyses) and were allowed to grow for 24 h. Afterward, cells were washed with PBS, and fresh culture medium containing 0.2 μg/μL of 231_sEVs (quantified through Bradford assay) was added (231_sEV-treated MCF7). PBS was used as a negative control for both MDA-MB-231 and MCF7 cells. Cells were allowed to incubate with vesicles or PBS for 48 h. Then, cells were fixed in 4% paraformaldehyde (PFA) for 20 minutes and washed twice in PBS. Samples were stored in PBS buffer with 1% penicillin/streptomycin at +4°C for short-term periods.

Immunofluorescence

Immunofluorescence images were taken using a microscope (Inverted Research Microscope Eclipse Ti2, Nikon) equipped with an epifluorescence illuminator and a highly sensitive sCMOS camera (Prime BSI, Teledyne Photometrics). For sample preparation, fixed cells were permeabilized with 0.5% PBS-TWEEN for 10 minutes and 0.1% PBS-TWEEN for 5 minutes (three times). Subsequently, cells were blocked in 1% bovine serum albumin in 0.1% PBS-TWEEN for 60 minutes. Antigen recognition was performed by incubating phalloidin in a humidified chamber or in agitation at room temperature (RT) for 45 minutes (Phalloidin, Invitrogen, A12381). Nuclei were stained with DAPI (Sigma Aldrich). Images were analyzed by using ImageJ^®.

AFM

AFM images were acquired by using a commercially available microscope (MFP-3D Stand Alone AFM from Asylum Research, Santa Barbara, CA) and a NSC36 tip (cantilever A with a typical resonance frequency of 90 kHz and a spring constant of 1 N/m, Mikromasch, radius of curvature < 10 nm) tip. Measurements were carried out in air at RT in a dynamic AC mode. For AFM imaging, fixed cells were washed twice in H₂O Milli-Q and decreasing percentages of absolute ethanol (40-60-80-98-100% each for 5 minutes) were used to dehydrate samples. Images with 55 µm × 55 µm of scan size and with a resolution of 512 × 512 pixels were acquired. The AFM images were analyzed with the ImageJ^® software.

AFM nanoindentation

AFM nanoindentation measurements were performed using a NanoWizard III AFM (JPK Instruments AG, Germany) mounted on top of an IX-81 inverted microscope (Olympus Corp., Japan). Fixed cells were measured in PBS buffer with 1% penicillin/streptomycin at RT. Despite PFA, fixation can induce alteration in the cell stiffness [26], and it is known that the relative variations in stiffness after treatments remain statistically significant even after fixation [27]. Standard pyramidal tipped cantilevers (MLCT-E, Bruker Corp.) with a theoretical spring constant of 0.2 N/m were calibrated by the thermal noise method before each experiment. Force–distance curves were collected with a force load of 0.4 nN at a rate of 2.5 μm/s. Sixteen curves were acquired for every cell in a spiral pattern over the nuclear region, with a boundary of 2 μm. Force–distance curves were analyzed using the JPK data processing software to level and offset the baseline, find the contact point, and subtract the lever bending, thus obtaining a pre-processed force-indentation (F–I) curve. Finally, each F–I curve was fitted to the modified Hertz-Sneddon model using a custom procedure on IGOR PRO 9.0 (Wavemetrics Inc.), which led to Y1 and Y2 values expressed as an average for each cell.

Fourier transformed infrared spectroscopy

For the purpose of these measurements, fixed cells were analyzed. Fixed cells were measured considering that the PFA fixation does not introduce significant spectral distortions and is consistent with immunofluorescence and AFM analyses [28]. The analysis was performed in air after some washes in H₂O Milli-Q to remove the PBS contribution. At least 500 single cells were acquired in two independent sessions, and in each session for each sample, it was acquired in triplicate. FTIR spectra of single cells were acquired at the infrared beamline SISSI at Elettra Sincrotrone Trieste, on a Vis-IR microscope Hyperion 3000 coupled to an interferometer Vertex 70 v (Bruker Optics GmbH, Ettlingen, Germany). Microscope knife-edge apertures were set to 30 µm × 30 µm to fit a single cell. An MCT detector with a 100 µm sensitive element was used. Two hundred fifty-six scans were averaged for each measurement in the 4,000–850 cm⁻¹ spectral region in a transmission mode using a 15× condenser/objective with a 4 cm⁻¹ spectral resolution. Quasar [29,30] and Origin (Version 2021b, OriginLab Corporation, Northampton, MA, USA) were used for the data plotting and analysis. Integrals were calculated by considering the area from the baseline in the following spectral ranges: lipids C-H stretching areas, 3,000–2,800 cm⁻¹; the protein region – Amide I and Amide II: 1,718–1,588 and 1,588–1,482 cm⁻¹, respectively; the nucleic acids PO₂ ⁻ stretching region, 1,285–1,188 and 1,135–998 cm⁻¹. To follow the method of analysis of Morrish et al., on estimating the DNA/protein ratio, half of the symmetric PO₂ ⁻ stretching region (1,188–1,070 cm⁻¹) and only Amide II band was considered.

Data processing and statistics

The significance of data differences was established via the analysis of variance (ANOVA) Kruskal–Wallis test considering the nonnormally distributed values (*p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001, respectively). For spectral data, the significance of band ratio differences was established via one-way ANOVA with the Tukey test (*p < 0.05; **p < 0.01; ***p < 0.001).

Small-EV isolation and characterization

Experimental conditions and procedures for the cell growth, isolation, and characterization were fine-tuned, as suggested by MISEV2018 guidelines [3], and reported in detail in our previous publication [24]. sEVs were isolated from MDA-MB-231 cells through ultracentrifuge and characterized from a morphological, dimensional, and biomolecular point of view, in the same way as shown in ref. [24]. In Figure S1, we just report a representative AFM image of one typical vesicle with a round shape and a diameter of 60 nm, perfectly in line with the expected sEVs dimensional range (40–200 nm).

TNBC-sEVs promote actin rearrangements and protrusions in target cells

The biomechanical behavior of cells is largely determined by the cytoskeleton network. Several studies have shown that elastic properties of cells are mainly influenced by actin structures [31,32,33]. Indeed, in our previous study, we demonstrated that filamentous actin formation in MCF7 target cells was promoted by the addition of 231_sEVs [24], as clearly visible from the epifluorescence images shown in Figure 1, Row a. On the basis of this result, we here further investigated actin rearrangements in MCF7 cells upon 231_sEV treatment. Cell growth is different for MDA-MB-231, MCF7, and 231_sEV-treated MCF7 cells, as observed from epifluorescence images: MDA-MB-231 and 231_sEV-treated MCF7 cells grow one on top of the other, while the untreated MCF7 cells are well separated from each other (Figure 1, row a). In the literature, the uncontrolled way of growth typical of MDA-MB-231 cells is associated with a high expression of Yap, low levels of E-cadherin, and the consequent lack of contact inhibition [34]. The acquisition of the same growth properties by 231_sEV-treated MCF7 cells represents a direct demonstration of the 231_sEV activity in target cells.

Figure 1

Cellular protrusions of MDA-MB-231, MCF7, and 231_sEV-treated MCF7 cells. Representative epifluorescence images of cells stained with phalloidin (red) and DAPI (blue) and the corresponding boxplots on bottom (Row a–b–c–g); scale bar indicates 50 µm; representative heights (Row d) and amplitude (Row e–f) images obtained from AFM imaging of cells and the corresponding boxplots in the bottom (Row g); dotted images and white arrows show in detail cellular protrusions. The lower and the upper boundaries of the box represent Q1 (25 percentile) and Q3 (75 percentile) of the data, respectively; the horizontal bar inside the box represents the median of the data. The significance of data differences was established via ANOVA Kruskal-Wallis test (*p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001, respectively).

Going further into details, epifluorescence analyses pointed out clear differences in cell morphology at the extremities of the analyzed cells. A significantly higher number and length of cellular protrusions, associable with actin-rich structures beneath the plasma membrane, aimed at exploring and moving into the surrounding environment [35], were found in 231_sEV-treated MCF7 cells compared to the MCF7 cells (indicated by white arrows in Figure 1, row a, highlighted in the magnified dotted images in Figure 1, Row c and quantified in the boxplot in Figure 1, Row g; n = 29 MDA-MB-231 cells , n = 25 MCF7 cells, and n = 23 231_sEV-treated MCF7 cells). Surprisingly, no differences in the number and extension of cellular protrusions between MDA-MB-231 and MCF7 cells emerged (Figure 1, Row b–c–g). The presence, numbers, and length of cellular protrusions were also confirmed by the AFM topographic imaging of cells (Figure 1, Row d–g; n = 9 MDA-MB-231 cells, n = 9 MCF7 cells, and n = 9 231_sEV-treated MCF7 cells).

TNBC-sEVs modulate the elastic properties of cell body in target cells

Since we observed an increment of cellular protrusions in MCF7 cells upon 231_sEV addition, we wondered if the biomechanical changes observed at the whole cell level were mostly ascribable to the contribution of the external cellular layer, i.e., plasma membrane rigidity, composition of the cellular brush, and the associated actin cortex. Toward this goal, we employed AFM force spectroscopy, a method widely used in the field of cell mechanobiology. Many reports in the literature, including our previous contribution [24,36] made use of micron-sized silica spheres to investigate the biomechanics of the whole cells [36,37,38]. By indenting the cells up to the 10% of the whole thickness, AFM force–distance curves are generally fitted by using the Hertz model to obtain a Young’s modulus, which reflects the cellular elastic properties. However, the coexistence of different structural layers within the cell makes it more elaborate than a simple, homogeneous mass. Vahabikashi et al. demonstrated that the large spherical probes are poorly sensitive to the cortical actin elasticity, being the interaction dominated by the cell body [39]. Instead, when indenting with a nanometer-sized pyramidal probe, the tip is more sensitive to the cell cortical stiffness [39], allowing to distinguish the contributions of the external layers and of intracellular components, respectively. Therefore, in the present work, we have performed AFM nanoindentations by pressing a pyramidal tip on the nuclear region. We observed the coexistence of two indentation slopes, the first one being dominated by the cell boundary and a second steeper one accountable for the intracellular organization, which adds up to the first. Accordingly, we have fitted such a trend to a variation of the Sneddon–Hertz model that considers a purely elastic deformation of both layers separately [40]. Finally, we have computed two Young’s moduli: Y1 represents the first ∼200 nm of deformation, while Y2 is relative to the segment ∼200–500 nm. Considering the cell architecture, we have approximated Y1 to the stiffness of the membrane/actin cortex complex and Y2 to the cellular body/nucleus. For each cell, 16 points above the nucleus were chosen in a spiral pattern to acquire force curves. For comparison, the Young’s modulus derived from the measurements with the spherical microprobe performed on MCF7, MDA-MB-231, and 231_sEV-treated MCF7 cells reported in our previous work is shown in Figure S2. In Figure 2a, we report representative force–indentation curves measured with the pyramidal nanoindenter in correspondence to the nucleus of each type of cells, highlighting the different elasticity. Two slopes are clearly identified for each representative curve. By fitting separately each part of the curve, the relative Young’s moduli Y1 and Y2 can be extracted (Figure 2b). As expected, both the external layers and the nuclear body of breast cancer cells are different from metastatic and nonmetastatic ones. In particular, both Y1 and Y2 of MDA-MB-231 cells resulted significantly lower compared to MCF7 cells (Figure 2b; n = 36 MDA-MB-231 cells, n = 39 MCF7 cells, and n = 34 231_sEV-treated MCF7 cells). In the case of 231_sEV-treated MCF7 cells, the elastic modulus Y2, corresponding to the nuclear body is significantly lower than that of the MCF7 cells, while the respective Y1 values are comparable.

Figure 2

Young’s moduli obtained through AFM force spectroscopy of MDA-MB-231, MCF7, and 231_sEV-treated MCF7 cells. Representative AFM force-indentation curve of cells by using a pyramidal tip (positioned on the nucleus) and by using the two-slope modified Hertz-Sneddon model to fit the force curves (a); representative AFM force-indentation curve and boxplots showing the elastic moduli Y1 (external layer) and Y2 (cell body) of the cells, obtained as described in A (b). The lower and the upper boundaries of the box represent Q1 (25 percentile) and Q3 (75 percentile) of the data, respectively; the horizontal bar inside the box represent the median of the data. Significance of data differences was established via ANOVA Kruskal–Wallis test (*p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001, respectively).

TNBC-sEVs mainly affect protein content and promotes chromatin decondensation in target cells

To obtain more information on the biochemical changes that occurred in MCF7 cells after the addition of 231-sEVs, we employed FTIR. It is well known in fact that infrared spectroscopy is useful for estimating protein-to-lipid ratio and to distinguish different subpopulations of extracellular vesicles [41,42,43,44,45]. Moreover, studies have demonstrated that FTIR spectroscopy is able to provide important clues regarding molecular differences between healthy and cancer cells, in terms of the distribution and structure of lipids, proteins, and nucleic acids [46,47]. Kumar et al. performed FTIR analyses on normal fibroblasts upon the addition of medium derived from three different breast cancer cell lines. They observed significant changes in chemical composition in cancer-stimulated fibroblasts compared to the control, in particular, changes in the regions of lipids (2,950 cm⁻¹) and in the area assigned to phosphate vibrations (nucleotides) (1,230 cm⁻¹) [48]. Among the various factors/molecules released from different breast cancer cell lines, extracellular vesicles in particular might be the major cause of the observed target cell changes. To our knowledge, infrared (IR) spectromicroscopy analysis on single cells upon the addition of extracellular vesicles represents an unexplored field. Therefore, we performed FTIR measurements on MDA-MB-231 cells and 231_sEV-treated/untreated MCF7 cells to assess the main molecular changes related to the sEVs activity, associated with the morphological and biomechanical modifications evidenced via epifluorescence and AFM measurements. Comparison of the mean absorbance spectra apparently revealed no particular differences among the different cells, as shown in Figure 3a (n = 516 MDA-MB-231 cells, n = 625 MCF7 cells, and n = 683 231_sEV-treated MCF7 cells). However, a more detailed analysis revealed some significant variations. The main spectral regions considered for these analyses were the lipids C-H stretching areas (3,000–2,800 cm⁻¹), the protein region (Amide I and Amide II: 1,718–1,588 and 1,588–1,482 cm⁻¹), and the nucleic acids PO₂ ⁻ symmetric stretching region (1,135–998 cm⁻¹). These regions were integrated, and relative ratios were calculated (results are shown in Figure 3, Row b). First, our results showed a significant increase in the proteins/lipids ratio for 231_sEV-treated MCF7 cells compared to both MCF7 and MDA-MB-231 cells. No significant differences were found instead for the sDNA/lipids ratio between 231_sEV-treated and untreated MCF7 cells. Instead, the metastatic MDA-MB-231 cells showed lower sDNA/lipid ratio compared to both treated/untreated MCF7 cells, which can be a hint of the enhanced lipid synthesis, widely proved as a marker of malignancy [49,50] and therefore in agreement with these results. In the previous sections, it was shown that the sEV treatment caused variations in cellular stiffness. The nucleus is the biggest and stiffest component of the cell [51], and chromatin condensation has been extensively identified as a modulator of biomechanics [52]. Specifically, chromatin decondensation and nuclear irregularity have been associated with high tumoral invasiveness of cancer cells [36,51]. In this direction, Morrish and coworkers demonstrated that the evaluation of DNA-to-protein peak ratio derived from an FTIR analysis can represent a viable signature of chromatin condensation at the single-cell level [53]. In their work, DNA decondensation was induced using different approaches either chemical or biological in CH12F3 cell line and in B primary cells from mice. They demonstrated that chromatin decondensation is not associated with changes of the overall DNA content, but to a decrease in DNA-to-protein peak ratio [53]. We evaluated the DNA-to-proteins ratio from the integrated intensity of the complete symmetric PO2⁻ stretching band and the Amide I and II peaks (and shown in Figure 3, Row B) and compared with the results obtained by the application of the same protocol suggested by Morrish et al. [53] (involving half of the PO2⁻ stretching band and Amide II only), shown in Figure S3. In both cases, our results showed a significant decrease in DNA-to-protein peak ratio in MDA-MB-231 and in 231_sEV-treated MCF7 cells with respect to the untreated MCF7 cells in line with the findings of Morrish et al. [53]. Therefore, we can hypothesize that 231_sEVs induce a chromatin decondensation in MCF7 cells.

Figure 3

FTIR spectroscopy of MDA-MB-231, MCF7, and 231_sEV-treated MCF7 cells. FTIR spectra of the average absorbance used for the calculation of different peak ratios (a). Boxplots of the ratios obtained from the respective integrated area peaks (Row b). The DNA-to-protein peak ratio is associated with the chromatin state at the single-cell level and a cartoon representing the compact or decondensed chromatin state (c). The lower and the upper boundaries of the box represent the SD of the data; the ▫ symbol and the horizontal bar inside the box represent the mean and median, respectively. The significance of data differences was established via ANOVA Tukey test (*p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001, respectively).