Characterisation of extracellular vesicles. In the present study, MCF10A cells were exposed to either MDA-TEVs or EVs derived from non-tumuorigenic human embryonic kidney (HEK)293 cells (HEK-EVs), to demonstrate that the associated EMT-inducing effect was caused by TEVs specifically, rather than EVs in general – a so-called specificity check. EVs were isolated by ultracentrifugation, but without further purification to avoid compromising particle yield and may therefore be described as crude EVs. EVs were characterised following the guidelines of The International Society for Extracellular Vesicle (ISEV) for EV research57, which include EV quantification by particle number and total protein content, protein markers expected and not expected to be present in enriched EV samples, and single vesicle analysis. Samples were enriched in the classic EV protein biomarkers CD9 and CD63. Samples did not contain calreticulin; a protein located in the endoplasmic reticulum, nor annexin A1; a microvesicle (150-1,000 nm sized EVs) biomarker58 (Fig. 2a), as expected. Nanoparticle tracking analysis (NTA) revealed heterogenous EV populations, with peak particle sizes of 153 ± 5.1 nm and 112 ± 12.7 nm for MDA-TEVs and HEK-EVs, respectively (Fig. 2b). Although such EVs may qualify as exosomes, without confirmation of their endosomal origin, they should be characterised as CD63-positive EVs57. Interestingly, EV particle concentration varied by a factor of >6 between the cell types, with 145 x 109 ± 2.3 x 109 MDA-TEVs and 22.7 x 109± 0.8 x 109 HEK-EVs derived from the same number of cells. This is consistent with reports that cancer cells produce an overabundance of EVs compared to healthy cells59,60, with their biogenesis possibly enhanced by hypoxic conditions prevalent in the tumour microenvironment (TME)61. This difference was equally reflected in the total protein amount, (Table 1). EVs had particle zeta potentials of -12.6 ± 0.62 mV for MDA-TEVs and -6.2 ± 0.25 mV for HEK-EVs, as expected62. EVs displayed characteristic spherical and cup-shaped morphologies on TEM micrographs, and size heterogeneity is apparent with vesicles ranging from 50-200 nm in diameter (Fig. 2c). Immunogold labelling of MDA-TEVs was performed to confirm the presence of CD63 on the EV’s surface (Fig. 2cii).
EV
|
Size (nm)
|
ζ-potential s(mV)
|
Concentration (billion particles/ml)
|
Protein content (μg/ml)
|
Ratio of
μg protein to
billion particles
|
MDA-TEV
|
153 ± 5.1
|
-12.6 ± 0.62
|
145 ± 2.3
|
2411 ± 23.2
|
16.6
|
HEK-EV
|
112 ± 12.7
|
-6.2 ± 0.25
|
22.7 ± 0.8
|
350 ± 3.5
|
15.4
|
Table 1: EV characterisation data. Mean ± s.e.m.; values calculated from measurements of EVs collected from n=2 and 3 different cell passages for each cell line for NTA/zeta potential and protein concentration, measured using a Qubit 4 fluorometer, respectively.
Continuous OECT-based monitoring of TEV-induced EMT. Current cell-based approaches and accompanying sensing technologies suffer from limitations that result in the loss of useful information and hamper the development of therapeutic interventions against EMT. End-point assays that yield discrete data require multiple time-points to be incorporated into the experimental design to track the dynamic changes that occur during EMT. While functional assays, such as migration or invasion assays, allow functional changes to be assessed, these are often limited to optical techniques, which are inherently low-throughout and semi-quantitative at best. To address this, we developed a novel functional readout of EMT using dynamic impedance-based monitoring on OECTs to probe changes in epithelial cell barrier integrity resulting from malignant transformation. The electrical “tightness” of a cellular monolayer is reflects the ionic conductance of the paracellular and transcellular pathways in an epithelial monolayer64. Although strictly speaking this should be a measured on a Transwell format, the resistance of cells adhered on substrates has previously been measured and correlates with barrier cell differentiation to epithelial phenotypes48. Resistance of the cell layer is a quantitative measure of barrier integrity and permeability, and, as such, can be correlated to the EMT status of cells (Fig. 1). We postulated that the ability to continuously and non-invasively monitor cell phenotype during EMT could help shed light on the incidence and implication of EMT hybrid states in metastasis and elucidate the timescales of TEV function; a crucial insight for developing potent therapeutic interventions (see Supplementary Discussion 2).
To achieve this, we had to adapt our OECT platform to ensure that it was compatible with long-term monitoring studies and could be integrated seamlessly with the EMT-model. Previous studies had functionalised the OECT surface with collagen type I to promote cell adhesion32,52. However, collagen type I promotes EMT and tumour invasion in cancer cells65,66. Therefore, to avoid any confounding variables, we forewent any surface functionalisation and instead incubated the OECTs with normal MCF10A culture media overnight prior to cell seeding to promote adherence of cells. Moreover, the commonly used external Ag/AgCl electrodes52,53 are cytotoxic67–69 and thus inappropriate for long-term studies. Instead, we opted to use a planar Au gate electrode to ensure that the devices produced stable outputs over extended periods of time. The absence of any external electrode also reduced the risk of contamination considerably and improved data acquisition times.
The addition of an insulating cell layer between the channel (Fig. 3a) and the gate affects the OECT gating efficiency, rendering the device slower to respond to a given applied gate bias35. The modulation of the device response time serves as a measure of the epithelial barrier integrity and is defined as the cut-off frequency, which corresponds to the frequency at 70.7% of the maximum transconductance48. The cut-off frequency is a figure of merit that define the regimes of high and low ionic transport43. Another important readout relevant to cells undergoing EMT can be derived from impedance measurements of the OECT channel as a function of frequency. As shown in the inset of Figure 3b, the impedance of the channels increases for frequencies over ~2KHz. At this frequency range, the impedance of our system is dominated by the resistance of the cell layer with the effect of capacitance being negligible70. Therefore, we can directly correlate this increase of impedance with the cell layer resistance.
To assess the ability of MDA-TEVs to promote EMT in MCF10A cells, four experimental conditions were tested: (1) non-treated cells cultured with normal medium; (2) cells treated with MDA-TEVs representing the treatment condition; (3) cells treated with transforming growth factor beta 1 (TGF-β1) representing a positive control, as TGF-β1 is commonly used to induce EMT in epithelial cells71–73; and (4) cells treated with HEK-EVs representing a negative control and a specificity check. The culture medium (normal or supplemented with MDA-TEVs, HEK-EVs, or TGF-β1) was refreshed every 48 hours, thereby exposing the cells to fresh EVs or TGF-β1 every two days (see Schematic S1). Throughout the experimental period, the activity of cells growing on the devices was monitored via daily OECT measurements. Figure 3a (see also Fig. S2) illustrates highly viable cells on the devices at the end of the experimental period, assessed by cytotoxicity/viability assays (LIVE/DEAD), assay) revealing an exceedingly high ratio of live-to-dead cells. This also highlights the added benefit of using the OECT which has a transparent channel that can be used to image the cells.
Cells present on the transistor channel and the gate electrode induce a shift in the cut-off frequency (Fig. 3c), which decreases as the cells continue to grow and differentiate, creating a barrier to the ion flux and slowing the response of the device. This is equally reflected in the extracted cell layer resistance () data, which increases over time (Fig. 3c, Fig. S3). The cut-off frequency was normalised to treatment day 0 for each condition (Fig. 3d), as the treatments were started at this time and to account for device-to-device performance variation.
A two-way ANOVA was performed to analyse the effect of treatment condition and treatment time on cut-off frequency. Within conditions, a two-way ANOVA showed that there was not a statistically significant difference in mean normalised cut-off frequency between day 0 and day 9 for non-treated cells (F(1,45)=54.88, p=.92) with Mdiff=0.17 (-17%) and cells treated with HEK-EVs (F(1,45)=54.88, p=.86) with Mdiff=0.20 (-20%). A small decrease in normalised cut-off frequency is observed in both conditions, which may be caused by the tightening of the lateral cell-cell junctions, as cells continue to differentiate and form an insulating barrier on the device. An asymptotic regression analysis revealed that the normalised cut-off frequency of non-treated MCF10A cells tends towards 0.88 0.034 (Fig. S4a). There was, however, a statistically significant difference in mean normalised cut-off frequency between day 0 and day 9 for cells treated with MDA-TEVs (F(1,45)=54.88, p<.000) with Mdiff=1.24 (+124%) and cells treated with TGF-β1 (F(1,45)=54.88, p<,000) with Mdiff=1.09 (+109%). This increase in cut-off frequency indicates an increase in the “leakiness” of the cell monolayer. This is in line with the degradation of cell-cell junctions74, characteristic of cells undergoing EMT, leading to an increase in paracellular ion flux into the PEDOT:PSS channel75. Optical imaging illustrates that cells are present and alive on the channel (Fig. 3a, 4b), , although the changes in paracellular ion flux are not visible, in contrast to the highly sensitive OECT-based measurements of changes in paracellular flux caused by EMT, enabling real-time monitoring of cell phenotype transition. Interestingly, it appears that MDA-TEVs initially promote degradation of epithelial barrier integrity faster than TGF-β1, during treatment day 0 to 4, after which point the effect of both treatments are comparable. This is in line with previously reported timescales of TEV and TGF-β1-indued EMT72,76,77.
Additionally, mean normalised cut-off frequency was compared between treatment conditions on day 9 (Fig. 3e). A two-way ANOVA showed that there was a statistically significant difference in mean normalised cut-off frequency on day 9 between non-treated cells and cells treated with either MDA-TEVs (F(3,45)=35.20, p<.000) with Mdiff=1.41 (171%) or TGF-β1 (F(3,45)=35.20, p<.000) with Mdiff=1.27 (153%). Whereas there was not a statistically significant difference in mean normalised cut-off frequency on day 9 between non-treated cells and cells treated with HEK-EVs (F(3,45)=35.20, p=1.) with Mdiff=0.021 (-3%). This demonstrates our ability to compare different treatment conditions conducted on different OECT devices to one another, which holds promise for scaled-up operation and confirms that the observed effect is caused by the intervention (MDA-TEV or TGF-β1). Lastly, a simple linear regression was used to test if MDA-TEV treatment time significantly predicted the mean normalised cut-off frequency (Fig. S5). The fitted regression model was:
Normalised cut-off frequency = 1 + 0.134 (±0.008) x MDA-TEV treatment time
|
(1)
|
The overall regression was statistically significant (R2=.978, F(1,69)=1527.60, p< .000) and it was found that MDA-TEV treatment time significantly predicted mean normalised cut-off frequency (β=0.134, p<.000). This indicates that the kinetics of the degradation of epithelial barrier integrity in this regime (0-9 days of MDA-TEV treatment) can be represented by a linear approximation. We would assume that the linear regime only represents a slice of the epithelial barrier degradation process, as the cells progress through EMT. The overall kinetics could potentially be represented by a logistic regression model (with its characteristic S-shape) with a maximum corresponding to the cut-off frequency when no cells are present in the channel. As cell barrier integrity invariably is associated with cell phenotype48 and by extension EMT status12,74, our platform is thus able to provide insight into the kinetics governing TEV-mediated EMT. This is highly applicable to predicting therapeutic response and modelling the relationship between TEV/drug treatment dose/time and epithelial barrier integrity.
Compared to previous studies, which assessed the effects of MDA-TEVs up to 48 hours post treatment76,77, we demonstrate continuous monitoring of MDA-TEV-induced EMT over a 9 day period and elucidate/establish a linear relationship between MDA-TEV treatment time and epithelial barrier integrity/EMT status. The ability to continuously monitor a transient process in a dynamic system is ideal for assessing drug response over time. To validate the electrical readouts, biomolecular assays were carried out to assess the change in cell phenotype and determine EMT.
Biological validation of OECT measurements by immunofluorescence and immunoblots. In our study, to validate the electrical readouts and establish cut-off frequency and impedance as a credible figure of merit for EMT, we determined cell phenotype and defined EMT status by analysis and quantification of gene expression of classic EMT markers using immunoblots and ELISA. Additionally, we performed IF confocal imaging to assess protein expression and location profiles and measure cell height and morphology. IF confocal imaging was performed both on samples fixed 48 hours after treatment, as previous studies report TEV-induced changes within this timeframe76,77, and on samples fixed on OECTs treatment day 9. An N-cadherin based ELISA was performed on treatment day 7 and whole-cell lysates were collected on treatment day 9 for immunoblot analysis.
After 48 hours of exposure to MDA-TEVs or TGF-β1, vimentin expression markedly increased in the treated cells and remodelling of F-actin is apparent with the appearance of actin stress fibres inside the cells, as opposed to the cortical thin actin bundles in the non-treated cells (Fig. 4a, see supplementary discussion 3). After 9 days of treatment, F-actin stress fibres are still present (Fig. 4b). Additional confocal IF images of MCF10A cells stained for E-cadherin,N-cadherin, and nuclei are available in Fig. S6. Cell density, as measured by the number of nuclei per area, decreased by 23% and 44% with MDA-TEV and TGF-β1 treatment, respectively (Fig. 4c). This is caused by individual cells occupying a larger area, which indicates a flattening of the cells, consistent with the transformation from the apical-basal polarity of epithelial cells to the back-front polarity of migratory mesenchymal cells74. The accompanying functional changes to the cells was captured by the OECT measurements, as cut-off frequency increased in the same 48-hour period (Fig. 4d), indicating that changes to lateral cell-cell adhesion have already begun and is readily detectable by the OECT. This holds promise for the utilisation of the OECT platform to provide better temporal resolution of TEV-induced EMT.
To further substantiate these findings, confocal microscopy was employed to evaluate the height of cells hosted on the OECT channel, as changes in cell height can be used to imply phenotypic switching during EMT79. Z-stacked confocal images of cells in the OECT channel, labelled for F-actin and nuclei, were obtained on day 9 of the experimental period (Fig. 4b, Fig. S7, Supplementary Videos 1-3). Cell height varied between non-treated and treated cells, mirroring the trend of cell layer resistance derived by OECT monitoring (Fig. 4e), and clearly demonstrating the unique ability of our platform to directly correlate a biological event with a corresponding electrical readout, both occurring in the channel. To assess the degradation of cell-cell junctions and define EMT, the samples were probed for several hallmark EMT proteins. Treated cells had increased expression of vimentin, N-cadherin, and fibronectin, (mesenchymal markers) and decreased expression of E-cadherin (epithelial marker), while F-actin expression remain constant (Fig. 4f-h). This evidences the occurrence of the cadherin switch (from E- to N-cadherin) and corroborates vimentin upregulation as seen on confocal images (Fig. 4a). ELISA-based N-cadherin protein expression revealed that N-cadherin on treatment day 7 in the MDA-TEV- and TGF-β1-treated cells was 19- and 15-fold higher, respectively, compared to the non-treated cells (Fig. 4f). Cells treated with HEK-EVs did not exhibit differential expression of these markers compared to non-treated cells (Fig. 4g-h), indicating that it is specifically MDA-TEVs that mediate EMT.
Our biological validation data strongly indicate that MDA-TEV treatment promotes EMT in MCF10A cells. Additionally, we demonstrate the multiplexing capability of the OECT platform by correlating the height of cells in the transistor channel to the cell layer resistance, Rcell, measured in the same channel. This ability to directly correlate an electrical signal with a biological readout is highly beneficial for furthering our understanding of EMT.
MDA-TEVs do not influence TWIST1 and TFPI2 DNA methylation levels. Next, we sought to gain mechanistic insight into MDA-TEV-induced EMT via epigenetics analysis. Epigenetic remodelling is prevalent during breast cancer metastasis80 and DNA methylation of tumour and metastasis suppressor genes is a hallmark of circulating tumour cells81. Gene promoters, especially key tumour suppressor genes, are unmethylated in normal tissues and highly methylated in cancer tissues82. In this way, gene promoter hypermethylation is a pathway for repression of gene transcription (transcriptional silencing), and conversely, gene promotor hypomethylation (loss of DNA methylation) may promote gene expression83. Oncogene BCR-ABL1-positive EVs released from leukaemia cells have been demonstrated to increase global DNA methylation levels in recipient cells84. With this in mind, we sought to determine whether MDA-TEVs influence DNA methylation by analysing the methylation status of the tumour suppressor gene tissue factor pathway inhibitor 2 (TFPI2) and the pro-metastatic transcription factor TWIST1. TWIST1 represses E-cadherin and promotes EMT76 and is negatively associated with TFPI2 in breast cancer patients, as TFPI2 suppresses breast cancer progression through inhibiting TWIST-integrin α5 pathway85. TFPI2 is downregulated in breast cancer cells lines compared to MCF10A cells and methylation in the TFPI2 promotor has been found in highly invasive breast cancer cells86. TWIST1 transcripts were reportedly upregulated after MDA-TEV treatment76,77 (Fig. 5a) and we observed a down-regulation of E-cadherin between treatment and non-treatment conditions (Fig. 4g-h), indicating E-cadherin repression. However, MDA-TEVs do not appear to carry TWIST1 (Fig. 5b)
A MethyLight-based approach was used to determine the methylation of these markers87. This was based on previous work published by Chettouh et al (2018), which looked at markers of Barrett’s oesophagus for methylation88. We found that neither TFPI2 nor TWIST1 was significantly differentially methylated between non-treated and MDA-TEV-treated cells (Fig. 5c). Surprisingly, TFPI2 displayed higher levels of methylation compared to TWIST1 across both conditions. This raises the question of whether TFPI2 (and TWIST1) methylation status in MCF10A cells could potentially contribute to the cell line’s intrinsic phenotypic plasticity for mesenchymal transition.
Interesting, immunoblot analysis revealed that TWIST1 protein was differentially expressed between the two conditions (Fig. 5a, d), which could suggest that TWIST1 protein expression is being repressed post-transcriptionally in non-treated MCF10A cells, e.g. by CPEB1/2 and miR-58089. miR-580 acts as a negative regulator of TWIST1 expression in MCF10A cells and is downregulated in MCF10A cells which have undergone EMT89. As cells treated with MDA-TEVs exhibit relatively higher levels of TWIST1, TWIST1 protein expression could be upregulated post-transcriptionally by other regulatory mechanisms, e.g. MDA-TEV-delivered miR-580-siliencing circular RNAs90, or even be induced by upregulation of transcriptional factors, e.g. HMGA291. Further immunoblot analysis revealed that TWIST1 was not part of MDA-TEV cargo, although highly expressed in MDA-MB-231 cells (Fig. 5b). These results collectively indicate that MDA-TEVs are dysregulating endogenous TWIST1-repression in MCF10A cells via mechanisms other than epigenetic remodelling or direct protein transfer.
Heparin treatment blocks TEV-induced EMT. Given that EMT determines the most lethal feature of cancer, metastasis, it represents an attractive target in oncology. However, direct targeting of EMT effector molecules is, in most cases, pharmacologically challenging92. Previous anti-EMT strategies have targeted signalling pathways, molecular drivers, or the mesenchymal cells themselves93, and several clinical trials targeting EMT are ongoing94. Recent efforts using nitrofen and its analogues to interfere with the process of EMT have proven efficacious in blocking TNBC metastasis and invasiveness both in vivo and in vitro95. Such reports underscore the importance of identifying pharmacological targets against EMT to prevent metastasis. Moreover, as EMT increases drug resistance96, targeting EMT could also attenuate cancer cell stemness and increase the effectiveness of more classic chemotherapeutic97. There is therefore an imminent need to elucidate the role of TEVs in EMT to identify new anti-cancer strategies that target TEVs.
We selected heparin as a possible TEV-targeting, anti-EMT treatment and leveraged our platform to screen this candidate drug. Heparin is a glycosaminoglycan commonly used as an anticoagulant drug. However, it is reported to have anti-cancer effects in humans98–101 and has been shown to decrease metastasis in animal models102. The mechanism for heparin’s anti-metastatic effect has been proposed to be a blocking function in tumour cell/platelet interactions103,104. In addition to this mechanism, heparin’s anti-cancer role may involve regulating TEV uptake into recipient cells105. EVs depend on cell-surface heparan sulfate proteoglycans (HSPGs) for their internalisation and functional activity and heparin is a competitive inhibitor of cell surface receptors dependent on HSPG coreceptors7,105,106. Heparin treatment of cells and/or EVs is known to interfere with TEV binding to the cell surface7,105,106 (Fig. 6a), and heparin has been reported to partially block EV-transfer of EGFRvIII mRNA into recipient cells105 and significantly reduce EV-mediated stimulation of cancer cell migration and invasion4,7. This function makes HSPGs a potential target for inhibition of TEV-mediated tumour development and proliferation. To test the efficacy of heparin in preventing MDA-TEV-induced EMT, a two-factor factorial experiment was designed, where cells were incubated with or without MDA-TEVs in the presence or absence of heparin (Fig. 6b).
Heparin treatment of cells effectively ablated the EMT-promoting effect of MDA-TEVs and heparin itself does not appear to adversely affect the barrier forming properties of the MCF10A cells (Fig. 6c, Fig. S10), the expression of several mesenchymal and epithelial markers (Fig. 6d-e), nor the viability of cells (Fig. S2). We hypothesize that if heparin treatment blocks MDA-TEV-induced EMT, then it may have an impact on cancer metastasis and could therefore be a putative starting point in developing an effective new drug for treating metastatic TNBC. An asymptotic regression was used to test if heparin treatment time significantly predicted the normalised cut-off frequency (Fig. 6f). The fitted regression model was:
|
Normalised cut-off frequency = ![](data:image/png;base64,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)
|
(2)
|
The overall regression was statistically significant (R2=.872, F(2,62)=210.63, p< .000) and it was found that heparin treatment time significantly predicted normalised cut-off frequency (α=0.92, β=-3.52, γ=0.16, p=0.0013), which compares well with the same model applied to the non-treated condition (Fig. S5). This demonstrates our ability to readily model the transient drug response and compare it to a control condition. Furthermore, it has been reported that persistent heparin treatment is necessary to abrogate the malignant effects of TEVs and inhibit TEV-induced tumour progression107. To test the transient effect of heparin treatment, cells were concurrently treated with heparin and MDA-TEVs for 3 days, after which the cells were washed in PBS and heparin treatment was ceased while MDA-TEV treatment continued (Fig. 6g.). Our results confirm the transient effect of heparin treatment and demonstrate that persistent treatment of cells with heparin concurrently with exposure to MDA-TEVs is necessary to prevent MDA-TEV-induced EMT. This also demonstrates the ability of the OECT to continuously monitor drug response and investigate the temporal relationship between measured pharmacodynamic response and pharmacokinetics non-invasively and in a facile manner.