BC cell lines, MDA-MB-231 and MCF7 were sourced from ATCC (Manassas, VA, USA). MDA-MB-231 cells are characterized by their triple-negative status (ER-/PR-/HER2-) and TP53 deficiency, while MCF7 cells are estrogen receptor-positive (ER+) and progesterone receptor-positive (PR+), with intact TP53 function. Both cell lines were cultured in DMEM medium (PanEko, Moscow, Russia) containing 4.5 g/L glucose and 1% L-glutamine, supplemented with 10% fetal bovine serum (FBS), 100 U/mL penicillin, 100 U/mL streptomycin, and sodium pyruvate (PanEko, Moscow, Russia). The cells were maintained at 37 °C in a humidified atmosphere with 5% CO₂. Cell line authenticity was verified through STR profiling, and they tested negative for mycoplasma contamination using the MycoReport Mycoplasma Detection Kit (Cat. # MR001, Evrogene, Russia).

Cells were irradiated at the Russian Proton Accelerator “Prometheus” of A. Tsyb Medical Radiological Research Centre (MRRC, Obninsk, Russia). Irradiation was carried out at the middle spread-out Bragg peak (SOBP, 15 mm width), with doses of 2, 4, and 6 Gy, with 70 $\times$ 83 mm² field size (energy range 94.40–105.30 MeV, LET 5.2 kEv/µm). Cell culture flasks were positioned upright and irradiated by directing the collimated beam at the flask surface. To ensure uniform exposure, the flasks were covered by delivering multiple shots that collectively encompassed the entire surface area. Flasks were placed on a piece of special plastic equipment at a reference depth in the center of the field with the help of laser centralizers. To ensure accurate radiation delivery, depth dose profiles were determined, and dosimetric calibrations were conducted using a water phantom PTW MP3-P Water tank using a plane-parallel ionization chamber Advanced Markus Chamber Type 34045 (SE 002617) (PTW, Freiburg im Breisgau, Germany). The dosimetric system was calibrated under reference conditions [35]. For $\gamma$-irradiation, the Rokus-AM facility (Rawenstvo, Russia) was used (Co⁶⁰ source, dose-rate 0.613 Gy/min^–⁢1).

Cells were seeded in T25 tissue culture flasks 24 h prior to irradiation at 3–5 $\times$ 10⁵ cells per flask. After irradiation, cells were seeded on 6-well plates at opportune densities (100, 500, 1000, and 2000 cells/well for doses of 0, 2, 4, and 6 Gy, respectively) according to the dose delivered to assay. Untreated cells served as a control to assess the plating efficiency (PE). Cells were cultured under standard conditions for 14 days, forming colonies. They were then fixed with methanol and stained with Giemsa. Colonies containing more than 50 cells were manually counted using a Zeiss Axiovert microscope with phase-contrast capabilities (Carl Zeiss, Göttingen, Germany). Dose–response data were generated from the mean of three independent experiments. Each biological repeat contained at least three replicates.

PE and survival fraction (SF) were subsequently calculated using the formula:

(1)$$PE=\frac{\text{number of colonies}}{\text{number of cells seeded in a well}}\times 100\%$$

(2)$$SF=\frac{\text{number of colonies formed after irradiation}}{(\text{number of cells seeded}\times PE)}$$

RBE is defined as the ratio of the absorbed dose of the reference radiation (gamma-irradiation, Со⁶⁰, Dref) to the absorbed dose of the test radiation (proton Beam, D) required to achieve the same biological effect (Eqn. 3).

(3)$$RBE=\frac{\text{Dref}}{D}$$

The biological endpoint is assessed through in vitro clonogenic survival, which is typically modeled using the linear quadratic model (LQM).

(4)$$S=\exp \left(-\alpha D-\beta D^2\right)$$

where the surviving fraction S as a function of the absorbed dose D, $\alpha$ and $\beta$ are exposure- and cell-specific fitting parameters.

Thus, the RBE was evaluated as a function of the surviving fraction S (RBES) using Eqn. 5:

(5)$$RBES=\frac{\alpha +\sqrt{{\alpha }^2-4\beta ln(S)}}{\alpha ref+\sqrt{{\alpha }^{2}ref-4\beta refln(S)}}$$

where $\alpha$, $\beta$, $\alpha$ref, and $\beta$ref are the linear and quadratic terms of the LQM for the radiation under investigation and the reference photon exposure, respectively.

Prior to irradiation, cells were seeded at a density of 0.1 $\times$ 10⁵ cells/0.32 cm² in 96-well plates. Cells were exposed to 2 Gy PBI (10 mm modified Bragg peak, 90 $\times$ 130 mm² field size, energy range 96.00–103.15 MeV).

Following incubation for 1, 12, 24, and 48 hours, cells were fixed in 4% paraformaldehyde in PBS (pH 7.4) for 20 minutes at room temperature. After two PBS rinses, cells were permeabilized with 0.3% Triton-X100 (PanEko, Moscow, Russia) in PBS supplemented with 2% BSA to prevent nonspecific antibody binding. Cells were then incubated with primary antibodies against $\gamma$H2AX (dilution 1:200, clone JBW301, MerckMillipore, Burlington, VT, USA), Rad51 (1:200 dilution, Cat. # ABE257, Merck Millipore, Burlington, VT, USA), and 53BP1 (1:200 dilution, Cat. # MAB3802, Merck-Millipore, Burlington, VT, USA) for 1 hour at room temperature. The primary antibodies were diluted in PBS with 1% BSA. Following three PBS washes, cells were incubated with secondary antibodies conjugated to Alexa Fluor 488 (1:800 dilution; Cat. #A-21424, Merck-Millipore, Burlington, VT, USA) and Cy3 (1:500 dilution; Cat. #AP124C, Merck Millipore, Burlington, VT, USA) for 1 hour. Nuclei were stained with Hoechst 33342 (Thermo Fisher Scientific, Waltham, MA, USA). Images were captured using a JuLI Stage automatic cell analyzer (NanoEntek, Seoul, Korea) with a 40$\times$ objective. Foci were manually counted, and data were derived from the mean of three independent experiments, each containing two replicates, with at least 100 cells analyzed per sample.

The proportion of cells expressing senescence-associated $\beta$-galactosidase (SA-$\beta$gal) was determined using a commercial Cellular Senescence Assay kit (EMD Millipore, Burlington, VT , USA, Cat. # KAA002). Cells were stained according to the manufacturer’s instructions, and nuclei were counterstained with Hoechst 33342. Visualization was performed using the EVOS FL Auto Imaging System (Fisher Scientific, Pittsburgh, PA, USA) with a 20$\times$ objective lens. The percentage of SA-$\beta$gal-positive cells, along with entotic cells, was manually calculated from three independent experiments, each containing three replicates, with at least 200 cells analyzed per sample. Statistical analysis was conducted using the Kruskal–Wallis test via GraphPad Prism software 9.0.2.161 (GraphPad Software, San Diego, CA, USA).

Cells were seeded in 3D fibrin gel using the protocol described previously [36]. Fibrin gels were fabricated using human fibrinogen and thrombin (NPO RENAM, Moscow, Russia). The fibrinogen powder was dissolved in sterile 0.9% saline solution at 37 °C to get 20 mg/mL concentration, aliquoted and stored at –80 °C. Thrombin stock (0.1 U/µL) was prepared by dissolving 1000 U thrombin powder in 10 mL T7 buffer (50 mM Tris, 150 mM NaCl, pH 7.4) and stored at –80 °C in 150 µL aliquots. Gel of 90 and 1050 stiffness was prepared according to Table 1.

For live 3D colony growth monitoring, 5000 MCF7 cells and 8000 MDA-MB-231 cells were seeded into 3D fibrin gels per well in a 24-well plate. Two wells were made for each condition and at least 20 images of colonies were captured in each well every two days on an EVOS M5000 fluorescent microscope (Thermo Fisher Scientific, Waltham, MA, USA) in a transmitted light channel with a 20$\times$ objective. The colony sizes were quantitatively estimated from digital images using the ImageJ software IJ 1.46r (Laboratory for Optical and Computational Instrumentation, University of Wisconsin, Madison, WI, USA).

Cells were seeded in 384-well plates at a concentration of 2.5 $\times$ 10³ cells/well. On the next day, cells were fixed with 4% paraformaldehyde for 15 min and rinsed three times with 1$\times$ PBS (pH 7.4). After blocking and permeabilization with 2% BSA and 0.3% Triton in 1$\times$ PBS (pH 7.4) for 45 min at room temperature, cells were incubated with primary antibodies against E-Cadherin (1:400 dilution, Cat. # ab40772, Abcam, Cambridge, UK), N-Cadherin (1:400 dilution, Cat. # ab76011, Abcam, Cambridge, UK), Vimentin (1:500 dilution, Cat. # ab92547, Abcam, Cambridge, UK), and SNAIL (1:500 dilution, Cat. # ab85936, Abcam, Cambridge, UK) diluted in 1$\times$ PBS with 2% BSA and 0.3% Triton X100, overnight at +4 °C. Goat anti-rabbit IgG Alexa Fluor 488 (1:1000 dilution, Cat. # ab150077, Abcam, Cambridge, UK) was used as a secondary antibody (2 µg/mL) for 3 hours at room temperature. Nuclei were counterstained with Hoechst 33342 (6 µg/mL, Thermo Scientific, Rockford, IL, USA).

To prepare the cells for analysis of filamentous actin (F-actin), they were fixed using a solution containing 4% paraformaldehyde diluted in 1$\times$ PBS for a period of 15 minutes. Following this fixation step, the cells were permeabilized with 0.2% Triton X-100, allowing for a consistent permeabilization process that lasted for 20 minutes. The cells were subsequently blocked using a 2% solution of BSA in 1$\times$ PBS (PBS-2%BSA) for one hour at room temperature. Following this, F-actin staining was performed using the Phalloidin-iFluor 488 Reagent (Abcam, Cat. # ab176753, Cambridge, UK) in PBS, adhering to the manufacturer’s protocol. Cells were imaged using the EVOS M5000 fluorescent microscope with a 20$\times$ objective. The results were obtained from three independent experiments with three replicates.

Non-irradiated and irradiation-survived MCF7 and MDA-MB-231 cells were seeded at a density of 5 $\times$ 10⁵ cells per well in a 6-well plate. The cells were cultured in a humidified atmosphere with 5% CO₂ at 37 °C to form a confluent monolayer. After 24 hours in depleted medium (1% FBS), the cell monolayer was wounded using a sterile micropipette tip, creating a straight line scratch. The monolayer was then washed with PBS to remove cellular debris. The initial and final wound areas were imaged at 0 and 24 hours post-wounding using an EVOS M5000 microscope with a 10$\times$ objective. Wound healing was quantified as a percentage using the Eqn. 6:

(6)$$\text{Wound healing, }\%=((\mathrm{WA0h})-(\mathrm{WA}\mathrm{\Delta }\mathrm{h}))$$ $$/(\mathrm{WA0h})\times 100\%$$

where WA0 h is the initial wound area, measured immediately after creating the wound, and WA$\mathrm{\Delta }$ h is the area after 24 hours. The percentage of wound healing was calculated as the mean with standard error of the mean (SEM) from two independent experiments, each with three replicates. Statistical analysis was performed using the Kruskal–Wallis test via GraphPad Prism software to compare the healing rates of parental and irradiated cells.

The test was performed according to our previously published protocol [37]. Briefly, carboxylate-modified fluorescent 200 nm in diameter particles (excitation/emission 505/515 nm) (Invitrogen life technologies, Carlsbad, CA, USA) were used to evaluate the adhesion and encapsulation efficiency of the cells. Cells were seeded in 96-well plates (Corning Incorporated, Corning, NY, USA) at a concentration of 5000 cells per well. Nanoparticles (NPs) were added to the cells to achieve the final concentration of approximately 2000 particles per cell. Nuclei were stained with Hoechst 33342. Cells were incubated for 1 hour at 37 °C, then washed with PBS and treated with trypsin to remove non-internalized particles. Cell viability was confirmed using a live/dead nuclei fluorescent stain. Negative controls, consisting of cells without nanoparticles, were also evaluated under the same conditions. Data were obtained from 3 independent experiments, each sample contained at least 3 replicates.

For the analysis of cell migration through physically confined environments cells were serum-starved overnight and then seeded in the upper compartment of Transwell inserts (Cat.# 3428 , EDM Milli-pore, Billerica, MA, USA) containing 0.1 mL of serum-free medium specific to the cell type. The inserts had 8 µm pores and were placed in wells with serum-supplemented growth medium in the lower chamber, which served as a chemoattractant. Cells were incubated for 72 hours at 37 °C and 5% CO₂. Following incubation, cells from both compartments were collected by trypsinization and counted using a hemocytometer after neutralization with a serum-containing medium. For each cell type, the experiment was repeated 6 to 9 times, with three replicates per experiment, to determine the percentage of cancer cells that migrated through the membrane.

Scanning ion-conductance microscopy (SICM) was employed to evaluate the mechanical stiffness of cell membranes. This advanced technique utilizes a nanoscale glass capillary filled with an electrolyte solution, which performs hopping movements to effectively map the topographical and mechanical properties of biological samples adhered to a substrate also immersed in the same solution.

The measurement of ion current occurs through the capillary tip, which is positioned between two AgCl electrodes that are submerged in the electrolyte solution, both inside and outside the capillary probe. This configuration allows for precise control over the interaction between the probe and the sample surface. For topographical mapping, a setpoint for an ion current decrease of 0.4% was utilized, while 1% and 2% were applied to assess stiffness. The Young’s modulus was subsequently calculated using the methodology established by Rheinlaender and Schäffer [38].

The SICM measurements were conducted using an instrument from ICAPPIC Ltd. (Moscow, Russia), which was equipped with a 1.2 mm borosilicate glass capillary that had been shaped using a P-2000 laser puller (Sutter Instrument, Novato, CA, USA). The inner diameter of the capillary tip was approximately 50 nm, as determined by measuring the ion current in a buffer solution.

For these SICM assessments, cells were seeded at a density of 100,000 cells per 35 mm Petri dish and incubated overnight at 37 °C in a CO₂ incubator. Prior to conducting measurements, the culture medium was replaced with an HBSS buffer solution that served as the electrolyte.

Fertilized specific pathogen-free (SPF) eggs were purchased from a local SPF hatchery. The surface of the eggs was cleaned with paper towels and sterile water, and then placed horizontally in an incubator set to 37 °C and 70% relative humidity, with the automatic rotation function activated. This day was designated as embryonic day 0 (EMD 0). On embryonic day 3 (EMD 3), the eggs were removed, and the air sacs and major blood vessels were identified and marked using an egg candler. Approximately 3 mL of egg white was then withdrawn, and upon observing the chorioallantoic membrane (CAM) drop, a 2 cm diameter circle was drawn at the drop location. The eggshell was cut along the marked edges using a mini chainsaw, and the window was sealed with 3M semipermeable film before the eggs were returned to the incubator, with the automatic rotation function turned off. On EMD 9, tumor cells were seeded. The eggs were removed, the semipermeable membrane was peeled off, and an O-ring (polytetrafluoroethylene (PTFE) O-ring, with an inner diameter of 6 mm and an outer diameter of 9 mm) was placed over the window near the densely vascularized area of the CAM. Twenty-five µL of a serum-free cell suspension containing 5 $\times$ 10⁶ cells was injected into the ring, after which the window was resealed with the semipermeable membrane, and the eggs were returned to the incubator. Eggs were collected on EMD 16, and tumors were imaged using a Leica M60 microscope (Leica Microsystems, Singapore, Republic of Singapore). To evaluate tumor growth characteristics, image analysis was performed using the ImageJ software. To account for potential variations in camera height during imaging, which could introduce discrepancies in absolute area measurements, the inner ring region with known size was used as a reference. This approach minimizes size deviations, enhancing the accuracy and comparability of the data. Additionally, the mean grayscale value of the target region was measured, where grayscale intensity in digital images, which reflects the optical density or thickness, serves as an approximate indicator of tumor tissue density or thickness in this study. The tumor size was estimated by calculating the ratio of the tumor area to the outer ring area, which was then multiplied by the normalized grayscale value, represented on the graph in arbitrary units (a.u.). Data were obtained from 3 individual tumors per sample.

The CAM xenograft experiments in this study were conducted in strict compliance with the guidelines and regulations of the Institutional Animal Care and Use Committee (IACUC). Chicken embryos are not classified as living animals until 17 days of incubation [39, 40], and our experiments with CAM were completed by embryonic day 16 (E16), which were exempt from IACUC regulation. Additionally, in adherence to the Russian Animal Experiment and Welfare Guidelines, ethical approval was not required. Daily assessments were carried out to ensure embryonic viability.

Statistical analysis was performed using Student’s t-test, the Mann–Whitney U-test, and two-way ANOVA with Tukey’s post-hoc test for multiple comparisons, all facilitated by GraphPad Prism 8 software. Results are presented as means $\pm$ standard deviation (SD) from three independent experiments unless otherwise specified. Significance levels were indicated by asterisks: * for p 0.05, ** for p $\le$ 0.01, *** for p $\le$ 0.001, and **** for p $\le$ 0.0001.

In this study, we first investigated the RBE of PBI for two BC cell lines, MCF7 and MDA-MB-231 (Fig. 1a), using clonogenic assay. Our goal was to examine the unique interplay between biological factors influencing physical parameters controlling the LET effect and innate radiosensitivity in determining a cell’s reaction to protons.

Fig. 1.

Cell survival clonogenic assay was performed to evaluate the RBE of proton beam irradiation. Representative pictures of colonies grown after each irradiation dose of gamma and proton beam irradiation of MCF7 and MDA-MB-231 (a) cells. Survival curves of MCF7 and MDA-MB-231 (b) cell lines after proton and gamma (Co⁶⁰ source) irradiation at doses of 2, 4, and 6 Gy were calculated. Data are means $\pm$ SD of three independent experiments. Statistical significance between groups indicated on the graphs was calculated with Student’s t-test. Scale bar 5000 µm. RBE, relative biological effectiveness.

RBE is a fundamental concept in radiation biology that quantifies the effectiveness of different types of ionizing radiation in inducing biological effects, typically in comparison to standard reference radiation, such as X-rays or gamma rays [41]. Using gamma irradiation from the Co⁶⁰ source as a reference, RBE was calculated based on the linear quadratic model (LQM), the most widely used model for predicting cell survival in radiation therapy. Following PBI, the radiosensitivity of MCF7 cells, as indicated by survival fraction, significantly decreased compared to gamma irradiation (p 0.05 for 2, 4, and 6 Gy) (Fig. 1b). In MDA-MB-231 cells, proton irradiation resulted in an even more pronounced reduction in radiosensitivity. However, statistical significance was only reached at 6 Gy compared to gamma irradiation (p 0.05) (Fig. 1b). The RBE of PBI at 10% cell survival fraction (i.e., SF = 0.1) was found to be 1.7 for MCF7 cells and 1.75 for MDA-MB-231 cells, confirming the superior efficacy of PBI in eliminating cancer cells compared to gamma irradiation. Notably, MDA-MB-231 cells exhibited the lowest clonogenic survival rates following high-LET irradiation, indicating a heightened sensitivity to protons compared to MCF7 cells.

After a continuous cultivation period of 3 to 4 weeks in T-25 tissue culture flasks, we observed that a distinct subset of cells that survived high-LET radiation from each parental cell line successfully generated viable progeny. These progeny were designated as the MDA-MB-231RP and MCF7RP sublines for further investigation (see follow-up Sections 3.4–3.8). All subsequent experiments involving irradiation-survived progeny were conducted within 10 passages.

After radiation exposure, DNA damage response (DDR) proteins are rapidly recruited and/or modified at DSB sites leading to their accumulation at the damage site. This results in the formation of nuclear foci, known as radiation-induced foci (RIF), which are detectable under fluorescent microscopy. RIF quantification serves as an indirect method for assessing DNA damage and repair by monitoring the molecular response to DNA alterations, ultimately reflecting the resolution of RIF. Typically, RIF measurements are performed on fixed samples at defined time points following ionizing radiation (IR) exposure [42].

Our observed differences in the sensitivity of TNBC (MDA-MB-231) and ER-positive (MCF7) cell lines to PT (Fig. 1) raised important questions regarding the DDR protein signatures associated with cell sensitivity to high- versus low-LET radiation. We further aimed to explore the molecular mechanisms underlying the heightened biological impact of protons, as well as potential proton-specific resistance mechanisms. Using quantitative immunofluorescence analysis, we calculated the kinetics of RIF formation for key DDR proteins, $\gamma$H2AX (Fig. 2a), 53BP1 (Fig. 2c), and Rad51 (Fig. 2e) in BC cell lines following 2 Gy PBI. At 1 hour post-irradiation, the number of $\gamma$H2AX RIF in MCF7 and MDA-MB-231 cells was nearly identical (Fig. 2b). However, extending the incubation period to 12 hours revealed significant differences between the two cell lines. MDA-MB-231 cells exhibited approximately twice as many $\gamma$H2AX RIF per nucleus compared to MCF7 cells (p 0.05). Furthermore, in MCF7 cells, $\gamma$H2AX RIF numbers did not return to baseline even after 48 hours of incubation. Notably, non-irradiated MDA-MB-231 cells displayed a higher intrinsic level of spontaneous $\gamma$H2AX foci at both 1 hour and 12 hours post-treatment compared to MCF7 cells (Fig. 2b).

Fig. 2.

Immunofluorescence images and quantitative immunofluorescence data illustrating the kinetics of DNA DSBs repair proteins. Quantitative immunofluorescence analysis was performed to calculate the kinetics of RIF numbers of key DDR proteins, namely $\gamma$H2AX (a), 53BP1 (c), and Rad51 (e). Kinetics of changes in the number of $\gamma$H2AX (b), 53BP1 (d), and Rad51 (f) foci in the cell’s nucleus after 2 Gy of proton beam irradiation. Statistical significance calculated by Student’s t-test. * p 0.05, ** p $\le$ 0.01, *** p $\le$ 0.001, **** p $\le$ 0.0001. Data are means $\pm$ SEM of three independent experiments. Scale bar 10 µm. RIF, radiation-induced foci; DDR, DNA damage response; $\gamma$H2AX, phosphorylated histone H2AX; 53BP1, p53-binding protein 1; SEM, standard error of the mean.

MCF7 cells exhibited a significantly higher number of 53BP1 RIF at the initial time point (1 hour) post-irradiation compared to MDA-MB-231 cells (p 0.05) (Fig. 2d). In MCF7 cells, this key marker of the NHEJ pathway for DNA DSB repair was resolved within approximately 12 hours, returning to baseline levels. In contrast, MDA-MB-231 cells displayed a different repair kinetics pattern. At 12 hours post-irradiation, the number of 53BP1 RIF had decreased to only 50% of the levels observed at 1 hour. Strikingly, by 48 hours post-irradiation, the number of 53BP1 RIF in irradiated MDA-MB-231 cells increased rather than resolving. Additionally, in non-irradiated MDA-MB-231 cells, 53BP1 foci remained consistently high throughout the 1–12 hour period (Fig. 2d).

In irradiated MCF7 cells, the initial peak of Rad51 RIF, a key marker of the homologous recombination (HR) pathway, reached approximately 5 foci per nucleus (Fig. 2e) and gradually declined to baseline levels by 24 hours (Fig. 2f). Increasing the incubation time to 48 hours resulted in more Rad51 RIF in both irradiated cell lines. In irradiated MDA-MB-231 cells, the number of Rad51 RIF remained above control levels at all examined time points (Fig. 2f). Notably, at 48 hours post-irradiation, Rad51 RIF levels were significantly higher compared to control levels (p 0.01).

The emergence of IR-induced cell senescence, a form of stress-induced premature senescence (SIPS) [43], represents a potential method to assist cancer cells in overcoming RT, and it may exacerbate the biological behavior of tumor cells after IR treatment [44]. Therefore, we sought to determine whether SIPS also contributes to the differences in radiosensitivity and DDR network functionality induced by high-LET radiation between MCF7 and MDA-MB-231 cells. Specifically, we aimed to investigate whether SIPS is associated with the inaccurate repair of DNA damage following high-LET irradiation.

To address this, we analyzed the proportion of senescent cells on days 1, 3, and 7 post-irradiation using analysis of the senescence-associated (SA) $\beta$-galactosidase (SA-$\beta$-gal). $\beta$-galactosidase is a lysosomal hydrolase, which accumulates in the cytoplasm of senescent cells giving green-blue staining at pH = 6 during analysis [45]. In non-irradiated cells, the fraction of $\beta$-galactosidase positive (SA-$\beta$-gal+) cells on the first day of analysis was 10% and 30% in MCF7 and MDA-MB-231 (Fig. 3a) cells, respectively, while not exceeding 2% during the subsequent 7 days of incubation.

Fig. 3.

Quantitative analysis of senescence-associated $\beta$-galactosidase (SA-$\beta$-gal) staining. The proportion of the senescent cells in MCF7 and MDA-MB-231 (a) populations after proton beam irradiation. Representative image of entotic cells in MCF7 irradiated by 6 Gy of proton beam (indicated by black arrows) (b). Cytoplasm of $\beta$-galactosidase positive cells stained in green. Nuclei (blue) stained by Hoechst 33342. Changes in the proportion of entotic cells in proton irradiation-survived (RP) MCF7 population by day 1, 3, and 7 (c). Statistical significance calculated by Student’s t-test. * p 0.05, ** p $\le$ 0.01, *** p $\le$ 0.001, ns, not significant. Data are means $\pm$ SEM from three independent experiments. Scale bar 25 µm.

PBI caused a significant increase in the proportion of SA-$\beta$-gal+ cells by ~55% and 50% in MCF7 and MDA-MB-231 cells, respectively (Fig. 3a). On day 3 after irradiation, the proportion of the SA-$\beta$-gal+ MCF7 cells was 27%, which further increased to 55% on day 7 (Fig. 3a). In contrary, only 5% of MDA-MB-231 cells were SA-$\beta$-gal+ on day 3, increasing to 20% (Fig. 3a) on day 7 after irradiation.

Entosis has recently been proposed as another potential survival mechanism in cancer cells exposed to radiation-induced stress [46]. BC has been found to have a high level of entosis competence, with entotic figures being observed in both primary tumors and metastatic lesions [47, 48, 49]. Despite this, little is currently known about the extent and effectiveness of the entosis process in human BC cell lines following proton irradiation.

Non-irradiated MCF7 cells exhibited a very low fraction ($\le$0.5%) of entotic cells when cultured under adherent and serum-rich conditions. As expected, entosis was exclusively observed in proton-irradiated MCF7 cells (Fig. 3b), but was absent in MDA-MB-231 cells under the same conditions. Interestingly, the majority of entotic MCF7 cells were positive for $\beta$-galactosidase, indicating an association with the SIPS phenotype (Fig. 3b, indicated by arrows). Although the proportion of entotic cells remained low under adherent and serum-enriched conditions, it significantly increased over time following proton irradiation (Fig. 3c).

Thus, the lower levels of the SA-$\beta$-gal senescence marker (Fig. 3a), the absence of entosis, and the less efficient DNA DSB repair mechanisms likely contribute to the heightened sensitivity of MDA-MB-231 cells to high-LET irradiation compared to MCF7 cells.

The remarkable biophysical properties of metastatic migrating cells, such as their exceptional motility and deformability, enable them to navigate through physical confinements created by neighboring cells or extracellular matrix (ECM) [29, 50]. The “scratch” test, also known as the wound healing assay, is a widely employed in vitro technique serving as a valuable tool allowing for the real-time visualization and quantification of 2D cell migration inside a monolayer cultured in serum-depleted medium (1% FBS) for 24 hours post-wound formation.

Non-irradiated parental MDA-MB-231 cells exhibited nearly double the migration within a monolayer as compared to non-irradiated MCF7 cells (Fig. 4a,b). Both MDA-MB-231RP and MCF7RP cell sublines demonstrated enhanced wound healing abilities compared to their non-irradiated parental cell lines (Fig. 4b).

Fig. 4.

Collective (wound healing assay) and confined (through 8 µm pores of Boyden chamber membranes) migration of parental (P) and proton irradiation-survived (RP) MCF7 and MDA-MB-231 cell lines. Representative images at 24 hours after wound formation (a). Scale bar 750 µm. Bar-graphs of changes in wound healing percentage (b). Representative pictures of Giemsa-stained cells on the outer surface of the Boyden chamber membranes at 48 hours of confined migration (c). Scale bar 50 µm. Bar graphs of the fraction of cells migrated into the lower Boyden chamber (d). Data are means of three replicates $\pm$ SD. Scale bar 50 µm. Statistical significance calculated by Student’s t-test. * p 0.05; *** p $\le$ 0.001; **** p $\le$ 0.0001. SD, standard deviation.

Using the classical transwell or Boyden Chamber assay, we evaluated a fraction of cancer cells that were confined in a physically restricted microenvironment. These cells were allowed to migrate through a membrane with 8 µm pores, which are significantly smaller than their size (19–21 µm), while following a gradient of serum concentration formed between the upper and lower chambers (Fig. 4c). In concordance with the 2D monolayer motility data, the non-irradiated MCF7 cells possess a significantly lower fraction of confined-migrating cells compared to MDA-MB-231 cells (Fig. 4d). In contrast to the 2D monolayer motility data, the MDA-MB-231RP subline, which survived high-LET exposure, exhibited a most prominent (14-fold) decrease in the number of confined-migrating cells compared to the MCF7RP (10-fold) subline (Fig. 4d).

EMT is a crucial biological process involved in various physiological and pathological conditions including cancer metastasis and invasion. EMT is characterized by the transition of epithelial cells into a mesenchymal phenotype, leading to changes in cellular morphology, gene expression, and functional properties. During classical EMT, epithelial markers such as E-cadherin are downregulated, while mesenchymal markers, including N-cadherin, Vimentin, and SNAIL are upregulated [51].

Unexpectedly, high-LET exposure led to an increase in E-cadherin expression in MCF7RP cells, while no significant effect was observed in MDA-MB-231RP cells compared to their respective non-irradiated parental controls (Fig. 5a, Supplementary Fig. 1a). Additionally, we observed a significant decrease in N-cadherin expression in MCF7RP cells (p 0.05), whereas its expression was significantly increased in MDA-MB-231RP cells compared to parental cells (Fig. 5b, Supplementary Fig. 1b). Vimentin expression was undetectable in both parental and MCF7RP cells, while its high expression remained unchanged between parental and MDA-MB-231RP cells (Fig. 5c, Supplementary Fig. 1c). SNAIL, a key transcriptional repressor of E-cadherin expression [52], showed an unexpected increase in expression in both MCF7RP and MDA-MB-231RP cells, although the change did not reach statistical significance (Fig. 5d, Supplementary Fig. 1d).

Fig. 5.

Analysis of EMT marker expression in parental (P) and proton irradiation-survived (RP) MCF7 and MDA-MB-231 cells. The expression changes of E-cadherin (a), N-cadherin (b), Vimentin (c), and SNAIL (d) were analyzed using high-content immunofluorescent imaging and analysis. Statistical significance calculated by Student’s t-test. * p 0.05. Data are means $\pm$ SD from three independent experiments.

Soft 3D fibrin gels can select highly tumorigenic cells and support their growth without differentiation. Only soft tumor cells, not stiff ones, can form spheroid colonies [53]. In this study, untreated parental MCF7 and MDA-MB-231 cells, along with their isogenic proton irradiation-survived descendants were seeded in 3D fibrin gels with stiffness levels of 90 Pa and 1050 Pa. Previous research has demonstrated that stiffness of 90 Pa promotes cancer stemness [54], whereas 1050 Pa approximates the stiffness of healthy breast tissue [55]. Bright-field digital images of growing spheroids were captured under a microscope on days 2, 4, and 6 post-seeding (Fig. 6a). The area of each spheroid colony was subsequently measured using ImageJ software.

Fig. 6.

Analysis of tumor spheroid growth in a 3D fibrin matrix. Images (a) and quantification of the area (b) of multicellular tumor spheroids grown within 90 and 1050 Pa 3D fibrin gels during the culture course of parental (P) and proton irradiation-survived (RP) MCF7 and MDA-MB-231 cells on days 2, 4, and 6. Statistical significance calculated by two-way ANOVA with Tukey correction for multiple comparisons. ** p $\le$ 0.01, *** p $\le$ 0.001, **** p $\le$ 0.0001. Data are means $\pm$ SD from three independent experiments. Scale bar 20 µm.

The mean area of compact MCF7 spheroids (of both irradiation-survived and parental cells) increased with time becoming significantly (p 0.0001) higher in low (90 Pa) compared to higher (1050 Pa) stiffness gel by day 6 after seeding (Fig. 6b). When cultured on a fibrin gel with a stiffness of 90 Pa, MCF7RP cells exhibited a remarkable (p 0.01) reduction in spheroid size when compared to non-irradiated parental cells. This observation suggests a reduced soft cell population that survives after proton exposure. In contrast, it was not evident in the gel with a stiffness of 1050 Pa, where no difference in colony area was observed at any time after seeding (Fig. 6b).

Meanwhile, MDA-MB-231 cells, both non-irradiated parental and irradiation-survived, formed only small diffuse spheroid colonies that remained almost unchanged in size over time, regardless of fibrin gel stiffness (Fig. 6b). By day 4 post-seeding, these cells had even degraded the fibrin gel and were predominantly found adherent to the bottom of the 6-well plate. Since cancer cells are known to secrete matrix metalloproteinases (MMPs) to degrade the basement membrane and facilitate invasion into the stromal matrix, we hypothesized that MDA-MB-231 cells mediate protease-driven degradation of the fibrin gel. However, the addition of the MMP inhibitor Marimastat on the first day of culturing had little to no effect on fibrin gel degradation in both parental and irradiation-survived MDA-MB-231 cells (data not shown).

After observing the differing abilities of proton irradiation-survived MCF7RP and MDA-MB-231RP cells in 2D migration in a monolayer and confined migration (Fig. 4), we aimed to further evaluate their metastatic potential in comparison to their parental counterparts. One emerging method for assessing metastatic potential is the NP test, which examines cellular nanoparticles (NP) uptake. This approach explores the relationship between cell migration and cancer cell endocytosis, emphasizing the role of the actin cytoskeleton [37].

To investigate the well-documented influence of ECM on cytoskeletal changes, cells were either pre-cultivated on rigid plastic surfaces ($>$10⁷ kPa) or allowed to grow in a pliable 3D fibrin gel (90 Pa). Following this, the amount of cell-associated NPs was evaluated in two-dimensional (2D) settings, where cells pre-cultivated on plastic surfaces were adhered to a rigid plastic surface and cells from 3D fibrin gel were seeded onto a soft fibrin gel layer (90 Pa). The amount of single cell-associated NPs was quantified subsequently using high-content imaging and analysis techniques, as outlined in our prior research [56]. The cells without NPs were used as the negative controls (see Fig. 7a).

Fig. 7.

Analysis of nanoparticles association and F-actin expression by parental (P) and proton irradiation-survived (RP) MCF7, and MDA-MB-231 cells seeded on 2D stiff plastic or a soft fibrin (90 Pa) layer following their pre-cultivation on either 2D stiff (plastic) or 3D soft fibrin gel, respectively. Representative pictures of cellular nanoparticles (NP) association test and F-actin phalloidin staining (a). The quantitative comparison of cellular NP uptake efficiency was calculated based on quantitative co-localization (Pearson’s and overlap coefficients, scatter slope) of fluorescent nanoparticles, and integrated fluorescence intensities (RFUs) of F-actin filaments (b). Scale bar 75 µm. Topography (left) and stiffness (right) mapping images obtained using scanning ion-conductance microscopy (SICM) on breast cancer cell lines (c). Scale bar 10 µm. Average relative Young’s modulus calculated as a weighted mean for each cell and then averaged across all measured cells (d). Statistical significance calculated by Student’s t-test. * p 0.05, ** p $\le$ 0.01, *** p $\le$ 0.001, **** p $\le$ 0.0001, ns, not significant. Data are means of three independent experiments $\pm$ SD. F-actin, filamentous actin.

Regardless of microenvironment stiffness during pre-cultivation and NP association, the highly metastatic parental MDA-MB-231 cells exhibited significantly higher NP association efficiency compared to non-metastatic parental MCF7 cells (Fig. 7b). In contrast, the efficiency of NP association in PBI-survived MCF7RP and MDA-MB-231RP cells displayed a clear dependence on microenvironment stiffness. MDA-MB-231RP cells exhibited a marked reduction in NPs association compared to their parental counterparts (p 0.01) when both exposure and NP-cell interactions occurred in a soft microenvironment (Fig. 7b, lower panel). However, in stiff conditions, no significant changes in NP association were observed (Fig. 7b, upper panel). Conversely, MCF7RP cells demonstrated a striking increase in NP uptake compared to their parental cells (p 0.001) under soft microenvironment conditions (Fig. 7b, lower panel), whereas NP association remained unchanged in stiff environments (Fig. 7b, upper panel). These findings indicate that high-LET irradiation alters the mechanosensation of PBI-survived MCF7RP and MDA-MB-231RP cells, modifying their ability to perceive the microenvironment stiffness. This, in turn, results in substantial changes to their mechano-biological properties, ultimately influencing their metastatic potential.

Next, we investigated the hypothesis that the microenvironment stiffness experienced by BC cell lines during culture influences their own cytoskeletal changes. Given that filamentous actin (F-actin) plays a crucial role in determining cell stiffness, one of the important biophysical features of malignant cells [56], we undertook a comprehensive analysis utilizing phalloidin staining of F-actin.

Both parental (P) MDA-MB-231 and proton irradiation-survived (RP) MCF7 cells cultured on a stiff plastic (Fig. 7a,b upper panels) exhibited significantly greater levels of cellular F-actin expression than those grown on a soft fibrin layer (Fig. 7a,b lower panels). High-LET irradiation-survived MCF7RP cells demonstrated a notable increase in softness compared to their parental cells, as evidenced by a significant reduction in F-actin expression, but only within a softer microenvironment (Fig. 7b). Conversely, MDA-MB-231RP cells exhibited greater stiffness in comparison to their parental counterparts, although the difference was not statistically significant when maintained in the same microenvironment (Fig. 7b, lower panels).

SICM was used to confirm the mechanical properties of the cell membrane by obtaining both topographical and stiffness mapping images of parental and PBI-survived cell sublines (Fig. 7c). We did not observe significant morphological changes following PBI exposure. The characteristic Young’s modulus revealed that parental MCF7 is significantly stiffer than MDA-MB-231 cells, which corroborates previous studies [57, 58]. Furthermore, both the PBI-survived cell sublines MCF7RP and MDA-MB-231RP exhibited a statistically significant decrease in stiffness following irradiation (Fig. 7d). This finding aligns with our data on NP uptake and F-actin expression in cells cultured within a stiff microenvironment (Fig. 7b, upper panel), suggesting a correlation between reduced stiffness, F-actin expression and increased 2D collective migration of PBI-survived cells (Fig. 4b).

The chick embryo CAM model presents a compelling in vivo alternative for tumor engraftment. Its vascularized membrane fosters robust tumor growth while the chick embryo’s immune system remains underdeveloped until embryonic day 16 (E16), making it an ideal environment for such research [59, 60, 61]. In this in vivo model, we sought to investigate the comparative tumorigenic potential of parental (P) MDA-MB-231 and MCF7 cells alongside their proton irradiation-survived counterparts (RPs).

By E16, the parental (P) MDA-MB-231 cells formed a substantially larger tumor mass organization characterized by a well-organized tumor–stroma that exhibited a white-opaque appearance and marked revascularization through CAM vessels (see Fig. 8, lower panel, marked by the black arrow). The implanted parental MCF7 cells exhibited a well-organized structure within small clusters, particularly near the CAM vessels. There was little indication of tumor mass organization, accompanied by surrounding mesenchymal and inflammatory reactions (refer to Fig. 8, upper panel).

Fig. 8.

The chick embryo chorioallantoic membrane (CAM) model was employed to assess the tumorigenic potential of BC cells. Representative images depict tumors that developed by embryonic day 16 in an in vivo CAM tumorigenicity assay, comparing parental (P) MDA-MB-231 and MCF7 cells with their proton irradiation-survived counterparts (RPs). Black arrows indicate tumor formation sites. Scale bar 2000 µm. The right panel illustrates the comparative tumor sizes between the P and RP groups. Statistical significance calculated by the Mann–Witney U-test. Values represent data from individual tumors with the mean $\pm$ SEM (n = 3 per sample).

Fig. 8 clearly illustrates that the proton irradiation-survived (RP) variant of MDA-MB-231RP cells displayed only minimal tumor growth (indicated by black arrow), highlighting a ~6-fold decrease in their tumorigenic potential in vivo compared to parental MDA-MB-231 cells. On the other hand, MCF7RP cells exhibited a more pronounced (~3-fold higher size) organization of tumor mass compared to their parental cell line. These data suggest that high-LET proton irradiation substantially impacts the tumorigenic ability of these cells when compared to their syngeneic parental counterparts.