Introduction

Head and neck squamous-cell carcinoma (HNSCC) is frequently an aggressive cancer with a high risk of recurrence. HNSCC is the 8th most common cancer in the UK, accounting for 3% of all new cancer cases and 2% of all cancer deaths (2016) (Shaw and Beasley 2016). Currently, surgery, radiotherapy and chemotherapy, in various combinations, are used in the treatment of HNSCC. The final treatment choice is dependent on disease locality, stage at diagnosis, HPV status, potential side effects, patient preference and overall health status (Brizel et al. 1998; Bonner et al. 2006; Vermorken and Specenier 2020; Cooper et al. 2004). For locally advanced HNSCC (LA-HNSCC), concomitant radio-chemotherapy remains the standard-of-care treatment, with HPV-positive tumours displaying greater radiosensitivity compared to HPV-negative tumours. This is reflected in the superior 3year survival data in response to radio-chemotherapy for HPV-negative [HPV ( − )] (57.1%) versus HPV-positive (82.4%) (Göttgens et al. 2018; Mirghani et al. 2015). Despite the fact that concomitant radio-chemotherapy is proven to extend overall survival, yielding improved the quality of life, long-term disease-free survival is rarely achieved due to intrinsic tumour resistance (Perri et al. 2015). This highlights an important area of unmet clinical need to develop innovative strategies for the treatment of HPV( − ) HNSCC.

Radiotherapy is a critical component in cancer management for more than 50% of cancer patients (Joiner and Kogel 2018; Delaney et al. 2005). In head and neck cancer, radiotherapy is widely utilised for treatment of primary disease and as an adjuvant with high local–regional control (Cooper et al. 2004). However, dose-limiting radiotoxicity represents a major challenge that limits the efficacy of the conventional radiotherapy (Bentzen 2006). One approach to overcome these obstacles is radiation dose enhancement, improving dose to the tumour while sparing surrounding normal tissue (Janic et al. 2021). This can be achieved through modulating tumour radiation response using radiosensitisers (Gong et al. 2021). Examples of radiosensitisers include oxygen mimics, cisplatin, hypoxia specific toxins, iodinated DNA targeting agents, inhibitors of DNA damage repair proteins and high atomic number materials such as gold and gadolinium (Gong et al. 2021; Wang et al. 2018; Chen et al. 2020).

Over the past decade, gold nanoparticles (AuNPs) have been widely explored as radiosensitisers due to their excellent physical and chemical properties (Her et al. 2017; Butterworth et al. 2012; Penninckx et al. 2020; Schuemann et al. 2020). AuNPs exploit the X-ray absorption properties of gold, where interactions with radiation destabilise atomic orbital electrons, resulting in a highly localised increase in radiation dose deposition, ultimately augmenting the effects of radiotherapy (Hainfeld et al. 2008). Radiosensitisation by AuNPs remains an active area of research in preclinical settings, with numerous studies demonstrating efficacy both in vitro and in vivo (Hainfeld et al. 2004; Cui et al. 2017; Jeremic et al. 2013). AuNPs are reported to complement the biological and chemical interaction of ionising radiation through the induction of ROS, cell cycle synchronisation, increased DNA damage yields, and via bystander effects (Rosa et al. 2017; Kempson 2021). However, the radiosensitising potential of AuNPs within the hypoxic microenvironment is often compromised, an effect attributed to compromised nanoparticle uptake, a reduction in reactive oxygen species and by the upregulation of various pro-survival hypoxia–inducible genes (Jain et al. 2014). These factors contribute to the development of radiation resistance, leading to tumour relapse and treatment failure. Therefore, a potential strategy to further augment AuNP radiosensitisation is the combined treatment with hypoxia modifying drugs, sensitising this treatment resistant microenvironment.

Hypoxia is a common feature of most solid tumours occurring in approximately 80% of head and neck tumours (Bittner and Grosu 2013). The intrinsic radiation resistance of hypoxic cells represents a significant obstacle to the successful application of radiotherapy (Rockwell et al. 2009). Numerous clinical studies have proven tumour hypoxia as a potent negative prognostic factor in head and neck cancer, with low oxygen levels correlating with reduced treatment efficacy and decreased overall survival (Brizel et al. 1997; Linge et al. 2016; Eschmann et al. 2005). To date, most attempts to ameliorate the negative impact of hypoxia involved either increasing endogenous oxygen levels by perfusion of high concentrations of oxygen, or by targeting specific subpopulations of hypoxic tumour cells using hypoxia activated prodrugs (HAPs) (Curtis et al. 2016). While effective in the experimental setting, these approaches yielded only modest benefit in the clinical context (Spiegelberg et al. 2019). A recent study revealed that a Warburg phenotype in tumour cells, resulted in impaired mitochondrial respiration, enhancing radiosensitivity compared to genetically matched parental cells (Bol et al. 2015). This effect was related to a differential metabolic rate and aberrant oxygen consumption, suggesting that suppression of mitochondrial metabolism could potentially increase susceptibility to radiotherapy (Yasui et al. 2017). Such an approach represents a potential strategy for all hypoxic tumours, not solely restricted to tumours where oxidative phosphorylation activity is upregulated. Furthermore, in silico models also indicate that reduced oxygen consumption holds more potential than elevating blood flow or promoting oxygen delivery, with a 30% decrease in oxygen consumption retaining sufficient molecular oxygen to abolish severe radiobiological hypoxia (Spiegelberg et al. 2019). Conversely, an 11-fold increase in arterial partial pressure of oxygen (pO2) is required to achieve a comparable level of oxygenation (Grimes et al. 2014). Currently, several pharmacological drugs that inhibit cellular oxygen consumption have been investigated for their potential to increase tumour oxygenation and thereby enhance radiosensitivity. Metformin, papaverine, arsenic trioxide and atovaquone are all known to influence tumour cell metabolism by modifying oxygen consumption (Benej et. al 2018; Diepart et al. 2012)

Atovaquone, an FDA approved drug, is a complex III inhibitor, originally used for the prevention and treatment of pneumocystis pneumonia and toxoplasmosis in HIV patients (Fry and Pudney 1992; Araujo et al. 1991). Atovaquone has recently been repurposed as hypoxia modifier, where atovaquone treatment induced an 80% reduction in tumour hypoxia in a range of tumour cell models through targeted inhibition of mitochondrial complex III activity. Importantly, this effect contributed to a significant tumour growth delay when combined with radiotherapy (Ashton et al. 2016). In addition, atovaquone also showed efficacy in the treatment of ovarian and breast cancer stem cell populations, a rare sub-population of cells linked with development of radioresistance and tumour relapse (Fiorillo et al. 2016; Guo et al. 2021). While current work with atovaquone as hypoxia modifier is still at early stage, the ability of atovaquone to alleviate tumour hypoxia appears promising. A recent clinical trial (NCT02628080) demonstrated that atovaquone reduced tumour hypoxia by up to 55% in non-small cell lung cancer patients (Skwarski et al. 2021). Encouragingly, the authors also reported that the most intensely hypoxic tumour voxels, i.e. the most radiologically resistant regions, were subjected to the most pronounced reduction in hypoxia, providing important insights into the action of atovaquone as a novel radiosensitiser (Bourigault et al. 2021).

In this study, we investigate the effect of targeting multiple pathways of radiosensitisation as an approach to enhance the radiotherapeutic potential in HPV( − ) HNSCC. We hypothesise that the combination of AuNPs, which endow physical dose enhancement, with atovaquone as metabolic modulator should result in a greater degree of radiosensitisation compared to either agent alone. To test our hypothesis, we evaluated the radiosensitising effects of AuNPs in combination with atovaquone. The effect on nanoparticle uptake, apoptosis, and radiation-induced DNA damage will be evaluated to elucidate the underlying mechanisms of radiosensitisation conferred by the combined treatment.

Materials and methods

Chemicals

Atovaquone was purchased from Sigma-Aldrich [UK) and was dissolved in 100% DMSO to make a 30 µM stock solution. All drug stock solutions were stored at − 20 °C for future experimental use.

Cell lines and reagents

CAL27 cells were obtained from the American Type Culture Collection [ATCC), and FaDu and CAL33 cells obtained from DSMZ [Brauschweig, Germany). CAL27 cells were grown in Dulbecco’s modified Eagle’s medium [DMEM, Sigma-Aldrich) supplemented with 2% HEPES (1 M, PAA laboratories), 1% sodium pyruvate (100 mM, Sigma), 1% non-essential amino acids (100x, Sigma), 10% foetal bovine serum (FBS, Gibco) and 1% L-glutamine (200 mM, Sigma-Aldrich). FaDu and CAL33 cells were maintained in Eagle’s minimum essential medium (EMEM, ATCC) with 10% FBS. All routine culturing was carried out at 37 °C and 5% CO2/95% air. Short tandem repeat profiling and mycoplasma testing (Lonza) were conducted routinely for these cell lines.

Synthesis of AuNPs

AuNPs were prepared according to the classical Turkevich method (Turkevich et al. 1951). Briefly, 400 mL of 0.01% chloroauric acid (HAuCl4.4H2O) solution was heated and refluxed in a round-bottom flask. Next, 9.6 mL of 1% sodium citrate solution was added to the boiling solution to obtain the AuNPs. The reduction of gold chloride by sodium citrate was completed after 5 min. The solution was further boiled for 30 min, then left stirring overnight at room temperature. Finally, the resultant colloidal solution was centrifuged at 12,000 rcf for 90 min and washed with distilled water.

Quantitative evaluation of AuNP uptake in cell monolayers

Cells were seeded in 6-well plates at a density of 4 × 105 cells/well and incubated overnight. Growth media was then replaced with fresh media containing different concentrations of AuNPs (0, 5, 10, 50, 100 µg/mL of gold). Following a 24 h treatment period, media containing excess extracellular AuNPs was removed, cells were then washed three times with PBS and harvested using 1×trypsin. Cells were counted and centrifuged at 500 g for 5 min, supernatant removed, and the cell pellet digested using 1 mL aqua regia. Samples were then made to a total volume of 5 ml using Milli-Q water. For hypoxic uptake experiments, cells were seeded into 6-well plates and allowed to adhere under atmospheric oxygen conditions overnight. The following day cells were transferred to a hypoxic workstation station (In Vivo2 400, Baker Ruskinn, UK) for 4 h, then treated with AuNPs at differing concentrations for a further 24 h. The concentration of internalised gold was measured by inductively coupled plasma atomic emission spectroscopy (ICP-AES) and reported as concentration of gold (pg) per cell.

Qualitative evaluation of AuNPs internalisation using hyperspectral microscopy

Enhanced darkfield/hyperspectral Cytoviva microscopy was employed for the optical observation and spectral confirmation of nanoparticle internalisation. Within these optical images AuNPs appear bright due to their high scattering cross section. To confirm the presence of AuNPs, the plasmonic resonance properties of gold was detected by spectra angle mapping (SAM) using the Cytoviva hyperspectral imaging system. This system confirms internalisation and intracellular location by comparing the unknown spectra in the acquired hyperspectral image against a user defined spectral library, i.e. that of AuNPs. Briefly, 4 × 104 cells were seeded into a 4-well glass chamber slide (Corning, USA). After 48 h, cells were treated with media containing AuNPs for 24 h, washed three times with PBS and fixed using 10% formalin. Under hypoxic conditions cells were transferred to a hypoxia chamber (0.5% O2) and incubated for 4 h before a 24 h AuNPs treatment, then washed, fixed and imaged.

Seahorse X-Fe 96 metabolic flux analysis

The bioenergetic function of HPV( − ) HNSCC cells treated with atovaquone was determined using Seahorse Extracellular Flux analyser (Seahorse Bioscience, MA, USA). In brief, 2 × 104 cells per well were seeded into seahorse X-Fe 96-well culture plates and incubated overnight to allow attachment. Cells were then washed with pre-warmed XF assay DMEM medium for OCR (oxygen consumption rate) measurement. XF assay media was supplemented with 10 mM glucose, 1 mM sodium pyruvate, 2 mM L-glutamine and adjusted to pH = 7.4. Cells were maintained in 180 µL/well of XF assay media, in a non-CO2 incubator for 1 h. During the incubation period, 20 µL of (150, 300 µM) diluted in seahorse XF medium was loaded into the injection ports in the XFe-96 senor cartridge. After establishing a baseline OCR and extracellular acidification rate (EACR), atovaquone was administered thorough an automated pneumatic injection port. Subsequent changes in OCR and ECAR were monitored over a 3 h period. The resulting effects on OCR and ECAR are shown as percentages of the baseline for respective treatments. For hypoxia OCR and ECAR measurements, all manipulations including seahorse analyses, were carried out within a hypoxic chamber at 1% O2.

Clonogenic assays

Clonogenic assays were performed according to our previously published reports [50]. In brief, 4 × 105 cells were plated in a 35 mm2 dish and allowed to adhere for 24 h. Cells were then treated with 100 µg AuNPs and/or atovaquone for 24 h, then irradiated with a single dose of 2, 4 or 6 Gy. For hypoxic studies, cells were seeded and allowed to attach at 37 °C, 5% CO2/95% air overnight before being transferred to hypoxic station with 0.5% O2, 5% CO2, 95% nitrogen for 4 h allowing equilibrium. Cells were then treated with either AuNPs (100 µg), atovaquone or combination of AuNPs and atovaquone for 24 h, prior to radiation treatment at 0.5% O2, 5% CO2. Following radiation treatments, cells were washed twice in PBS, trypsinised, counted, and reseeded in 6-well plates before incubating for a further 9–14 days under atmospheric oxygen conditions, allowing colony formation. Colonies were then fixed and stained using 70% methanol and 0.4% crystal violet. Colonies with more than 50 cells were counted to determine plating efficiency (PE), where PE is defined as the number of colonies formed divided by the number of cells seeded. Surviving fractions (SF) were then calculated relative to non-irradiated controls and fitted to the linear quadratic (LQ) equation (Eq. 1).

$$\bf {\text{Linear quadratic:}}\,\,{\text{S = exp}}(-\alpha {\text{D}}-\beta {\text{D}}^{2} )$$
(1)

LQ fits were calculated using least-square regression in prism 9.0 (GraphPad Software, CA, USA). The area under the curve (AUC), which represents the mean inactivation dose (MID), was obtained and the sensitiser enhancement ratio (SER) calculated by dividing the MID of non-exposed cells with AuNPs or atovaquone treated cells. α and β components have been derived from the linear quadratic fit, representing the ratio of either direct (α) and indirect (β) cell damage.

Cell cycle analysis

Cells were seeded in 6-well plates at a density of 8 × 105 cells/well, then incubated overnight. The next day growth medium was replaced with fresh medium containing atovaquone, AuNPs (100 µg/mL) or the combination of AuNPs and atovaquone for a further 24 h. Cells were then irradiated using a single 4 Gy dose and returned to the incubator for a further 24 h before fixing. Monodisperse cells were washed in PBS and resuspended in 0.5 mL of propidium iodide (containing 50 µg/mL PI and 10 µg/mL RNase) for 30 min at 4 °C. Analysis was performed using a BD FACSCalibur flow cytometer. Cell cycle analysis was performed in triplicate with data from a minimum of 10,000 cells per sample collected. Cell cycle data analysis was performed using FlowJO software.

Cell apoptosis assay

Quantitative analysis of apoptosis was performed by flow cytometry using an Annexin-FITC-propidium iodide (PI) apoptosis detection kit (BD Biosciences, USA). In short, cells were incubated at a density of 2 × 105 cells/well in 6-well plates, and allowed to attach for 24 h. Cells were then exposed to either hypoxic (0.5% O2) or atmospheric oxygen (21% O2) conditions and treated with medium containing atovaquone, AuNPs (100 µg/mL) or the combination of AuNPs and atovaquone for a further 24 h prior to radiation treatment. As with cell cycle analysis, apoptosis assays were performed 24 h post-radiation treatment. The subsequent procedure was performed in accordance with the manufacturer protocol. Briefly, cells were washed twice with cold PBS, then suspended in 195 μL binding buffer at a cell density of 5 × 105 cells/mL. After a 10-min incubation with 5 μL of Annexin V-FITC, cells were resuspended in 190 μL binding buffer including 10 μL of PI. After incubation for 15 min at room temperature in the dark, samples were analysed by flow cytometry (BD FACSCalibur). Data analysis was performed using BD FACSuiteTM software.

DNA damage analysis by immunofluorescence assay

Immunofluorescence assays were performed as described previously [50]. Cells were plated in 4-well chamber slides at a density of 4 × 104 cells per well and left to adhere for 48 h. Cells were then treated with atovaquone (30 μM), and/or AuNPs (100 µg/mL) for a further 24 h before IR treatment using a single fraction dose of 2 Gy (Faxtrion CP-160 Arizona USA). After IR treatment cells were returned to the incubator for either 2 h or 24 h, allowing partial DNA damage repair. Cells were then washed with PBS and fixed with 4% formaldehyde, permeabilised using 0.1% PBS-Tween 20, and blocked with 1% BSA for 1 h. Cells were then incubated with primary 53BP1 antibody (1:1000 dilution in 1% BSA). As indirect immunofluorescence was used, cells were then washed three times with PBS for 5 min and incubated with Alexa flour 488 secondary antibody (1:1000 dilution in 1% BSA) at room temperature in the dark. Finally, cells were washed three times with PBS before drying and mounting with DAPI (Abcam, UK). With respect to quantifying residual DNA damage, for each replicate 53BP1 foci from 50 cells per treatment were scored.

Statistical analysis

All statistical analysis were performed using GraphPad Prism 9 software. Scientific data were expressed as mean ± SEM. For clonogenic assays, two-way analysis of variance (ANOVA) was performed with Turkey multiple tests to determine statistical differences between treatment groups. For all other experiments, comparisons between multiple groups were performed using one-way analysis of variance with Turkey multiple tests. Data are shown as statistically significant when p-value of equal to or less than 0.05 were recorded.

Results

Cell uptake and location of AuNPs

Intracellular concentration and localisation of AuNPs are two important factors that markedly influence tumour cell sensitivity to radiation treatment. AuNP internalisation was assessed both quantitatively and qualitatively using ICP-AES and hyperspectral imaging. Quantitative assessment of AuNPs uptake demonstrated a clear positive correlation with treatment concentration under both hypoxic and normoxic conditions, across all three cell lines tested (Fig. 1). More specifically, AuNPs internalisation proved to be significantly (p < 0.01) increased at 100 µg/ml AuNP treatment over 10 µg/ml or below, an effect consistent to all cell models. Intracellular Au concentrations of 64 ± 24 pg/cell, 68 ± 17 pg/cell and 75 ± 23 pg/cell, were achieved in FaDu, CAL27 and CAL33 cells (Fig. 1a–c), respectively. Hypoxic stress had no obvious impact on AuNP uptake, with total intracellular Au levels slightly increased in all three cell lines over cells maintained under environmental oxygen conditions (FaDu—86 ± 19 pg/cell, CAL27—80 ± 2 pg/cell, and CAL33—97 ± 19 pg/cell, respectively).

Fig. 1
figure 1

Cell uptake of AuNPs in FaDu (a), CAL27 (b) and CAL33 (c) cells under atmospheric (21% O2) or hypoxic (0.5% O2) conditions. One-way ANOVA followed by Turkey’s multiple tests was used to compare within different concentrations, * p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. Data were obtained from three independent experiments performed in triplicate ± SEM

Full size image

Intracellular AuNP uptake and localisation was also visualised using hyperspectral imaging. Darkfield imaging indicates significant nanoparticle internalisation by the presence of multiple bright focal regions following 24 h nanoparticle treatment. Using integrated pixel-by-pixel hyperspectral filtering against the untreated control, false positive signals are eliminated, producing a spectral angle map (SAM) confirming AuNP intracellular locality. At this resolution, predominant AuNP intracellular accumulation appears to co-localise within a perinuclear region (Fig. 2), mostly likely within endoplasmic reticulum due to endosomal vesicle trafficking. As with quantitative uptake studies, similar observations were also observed under hypoxic conditions (Additional file 1: Figure S1).

Fig. 2
figure 2

Intracellular uptake and localisation of AuNPs within FaDu, CAL27 and CAL33 cells under atmospheric oxygen conditions using Cytoviva darkfield/hyperspectral imaging. Cells were treated with AuNPs for 24 h, fixed and mounted with DAPI before imaging. Representative images include enhanced dark field and the corresponding hyperspectral images, overlaid with a spectral angle map (SAM) representing AuNPs in red

Full size image

Effect of atovaquone on bioenergetic function in HPV( − ) HNSCC cells

Atovaquone mediated inhibition of respiratory capacity would be expected to significantly alter the tumour cell metabolic profile, enabling adaption to treatment related stress. To test this, we measured the bioenergetic function of FaDu, CAL27 and CAL33 cells in response to atovaquone using the Seahorse Bioscience XF96 analyser, with experiments conducted under 21% O2 and 1% O2. Cells were analysed for both oxygen consumption rate (OCR) and extracellular acidification (ECAR—a surrogate readout of lactic acid efflux). Atovaquone greatly reduced OCR in all cells under normoxic (Fig. 3) and hypoxic (Additional file 1: Figure S2) conditions, with hypoxia notably decreasing OCR by roughly 50% in the control group. More specifically, a marked OCR decrease was observed in FaDu cells using atovaquone at both 15 and 30 µM under normoxia (Fig. 3a), reducing OCR by 76% over untreated controls. A similar degree of OCR suppression was also observed in CAL27 and CAL33 cells (Fig. 3b–c), reducing OCR by approximately 80%. Encouragingly, if adopting this strategy to help alleviate hypoxia induced radioresistance, a similar pattern was also observed under hypoxic conditions following atovaquone treatment (Additional file 1: Figure S2a–c), maximally suppressing OCR by 87.6%. This meant that atovaquone treatment effectively doubled the reduction in OCR relative to hypoxia alone, proving effective even under hypoxic conditions.

Fig. 3
figure 3

Bioenergetic profile of HNSCC cells treated with atovaquone (AQ) under 21% O2. Changes in oxygen consumption rate (OCR) and extracellular acidification rate (ECAR) were monitored at 37 °C for 3 h. The resulting effect of atovaquone on OCR and ECAR is shown as a percentage of the baseline measurement after each treatment. Data shown are the means ± SEM (n = 5 per treatment group)

Full size image

Parallel to the changes in the mitochondrial respiration, we also observed that atovaquone stimulated a concomitant increase in ECAR, indicating a switch to glycolytic metabolism, likely compensating for ATP depletion due to respiratory stress. Atovaquone elevated the extracellular acidification by 2.3-fold in FaDu cells over untreated controls (Fig. 3d). This effect was even more pronounced in CAL27 and CAL33 cells, resulting in an approximate fourfold increase in ECAR, irrespective of drug concentration (Fig. 3e–f). Under hypoxic conditions, similar effects were also observed in all three cell lines tested (Additional file 1: Figure S2), with the greatest effect seen in CAL33 cells, yielding a 2.5-fold ECAR increase (Additional file 1: Figure S2f).

Modified radiation sensitivity using AuNPs and atovaquone

AuNPs are proven radiosensitisers, nevertheless the magnitude of effect is often compromised under hypoxic conditions (Jain et al. 2014). Herein we investigated whether AuNPs, which act as both physical and chemical radiation dose modifiers, combined with atovaquone, increase the radiosensitivity of HNSCC cells compared to either agent alone. In the absence of radiation, AuNP, atovaquone or the combined treatment showed no significant toxicity as evidenced by the minimal changes in SF0 values under both normoxia and hypoxia (Additional file 1: Figure S3). Radiation survival curves for all treatment groups along with dose enhancement factors (DEF) at 2 Gy and 4 Gy are provided in Fig. 4 and in the supporting information (Additional file 1: Table S1). AuNPs alone significantly sensitised all three cell lines to radiation treatment under atmospheric oxygen conditions, producing mean sensitiser enhancement ratios (SER) of 1.37, 1.33 and 1.37 for FaDu (p = 0.0122), CAL27 (p < 0.0001) and CAL33 cells (p = 0.0002), respectively. Atovaquone alone significantly increased radiation sensitivity in FaDu cells (p = 0.0463), generating a SER of 1.43 (Fig. 4a). However, drug alone failed to modulate radiation response in CAL27 and CAL33 cells (Fig. 4b–c). Using an equivalent concentration of atovaquone combined with AuNPs markedly augmented radiation sensitivity (p = 0.0002), yielding an SER of 1.72 in FaDu, equating to an additional 35% increase in radiation sensitivity over single agent treatment with AuNPs alone. Conversely, the combined treatment of AuNP and atovaquone failed to significantly enhance radiation sensitivity over AuNPs alone in CAL27 and CAL33 cells (Fig. 4b–c), implying that in these cells radiation dose enhancement is predominately mediated by AuNPs. Interestingly, in an era where hypofractionation and stereotactic radiotherapy are increasingly pushing the clinical limits of dose fractionation, the radiosensitising potential of combining these two agents appear to increase with dose, evidenced by an increased DEF of 2.64 at 4 Gy in FaDu cells (Additional file 1: Table S1).

Fig. 4
figure 4

Clonogenic radiation survival curves for FaDu (a and d), CAL27 (b and e) and CAL33 (c and f) cells treated with either AuNPs (100 μg), atovaquone (AQ, 30 μM) or the combination of both under normoxic (a, b and c) and hypoxic (d, e and f) conditions. Two-way ANOVA with Turkey multiple comparison test was used to compare within different treatments. Significant differences are represented by *p < 0.05, **p < 0.01, ***p < 0.001. Data were obtained from three independent experiments performed in triplicate ± SEM

Full size image

The radiosensitising effect for the combined use of AuNPs and atovaquone was also investigated under hypoxic conditions. Unsurprisingly, hypoxia severely impaired radiation sensitivity, with an approximate oxygen enhancement ratio (OER) of 2, indicating that hypoxia was maintained during radiation treatment, and that hypoxia induced radioresistance was observed. Encouragingly, under hypoxic conditions, the combination of AuNP and atovaquone again resulted in a significant (p = 0.049) increase in FaDu radiosensitivity (SER 2.1) over treatment with AuNP alone (Fig. 4f). As observed under normoxia, atovaquone alone failed to significantly modulate the response to radiotherapy in CAL27 and CAL33 cells, an effect recapitulated in the combined treatment group.

Effect of AuNP and atovaquone treatment on cell cycle distribution

To explore further the potential mechanism underpinning AuNP/atovaquone radiosensitisation, the effect of treatment on cell cycle distribution was evaluated. Combinations of AuNP and atovaquone induced a significant (p < 0.05) increase in the ratio of S-phase FaDu cells under 21% O2 (Fig. 5a, Additional file 1: Fig S4a). Furthermore, combined with a 4 Gy radiation dose, atovaquone significantly (p = 0.014) decreased the G0/G1 population compared to AuNPs + 4 Gy treated cells, from 55 to 18% (p = 0.0143). This response drove a concomitant increase in the proportion of G2/M cells, doubling from 20.7% to 40.6% (p = 0.0141), Despite that no significant difference in clonogenic cell survival was observed in CAL33 cells, similar significant (p = 0.0216) drug/nanoparticle induced alterations in cell cycle distribution were observed, with a > 10% increase in G2/M accumulation following atovaquone and radiation (Fig. 5c and Additional file 1: Fig S4a), an effect that was surprisingly impaired (42.9% to 15.9% (p < 0.001)) when AuNPs were added to the treatment combination. No statistically significant cell cycle alterations were observed in CAL27 cells irrespective of treatment (Fig. 5b and Additional file 1: Fig S4a). In a similar manner, under hypoxia, the various combinations of atovaquone/AuNP treatment had minimal impact on cell cycle distribution (Fig. 5d–e, and Additional file 1: Fig S4b), except that we again observed a significant decrease in the G2/M population of CAL33 cells following the combination treatment (Fig. 5f) (p = 0.0126).

Fig. 5
figure 5

Cell cycle distribution of HNSCC cells 24 h after treatment with AuNPs (100 μg) and/or atovaquone (AQ 30 μM), alone or in combination with radiation (4 Gy) under both 21% O2 and hypoxic conditions (0.5% O2). One-way ANOVA followed by Turkey’s multiple tests was used to compare within different treatments, significant differences are represented by *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. Data were obtained from three independent experiments performed in triplicate ± SEM

Full size image

Effects of AuNPs and atovaquone inhibitor treatment in apoptosis induction

Apoptosis significantly contributes to radiation-induced cell death (Rupnow and Knox 1999; Dewey et al. 1995; Verheij and Bartelink 2000). In the current study, the apoptotic response of HNSCC cells in response to atovaquone, AuNPs or the combination with and without radiation was evaluated using Annexin-FITC/PI staining (Fig. 6 and Additional file 1: Fig S5). Unsurprisingly, the combined treatment with radiation induced the highest proportion of apoptotic cells in all cells tested, proving statistically significant in FaDu (27%, p < 0.01—Fig. 6a) and CAL27 cells (25%, p = 0.012—Fig. 6b) compared to either AuNP or atovaquone as a monotherapy combined with radiation. In CAL33 cells the magnitude of response was similar, although not significant, with the observed increase in apoptosis predominantly driven by atovaquone (Fig. 6c). Interestingly, under hypoxia there was a trend towards increased basal apoptosis, likely triggered by hypoxia induced stress, with the overall combination effect of atovaquone and AuNP strongly attenuated (Fig. 6d–f) across all three cell lines.

Fig. 6
figure 6

Annexin V/PI staining for apoptosis in FaDu (a, d), CAL27 (b, c) and CAL33 (e, f) cells treated with AuNPs (100 μg), atovaquone (AQ 30 μM) or radiation (4 Gy) or the combination. Cells were maintained under atmospheric oxygen conditions (21% O2—panels a, b and c) or hypoxia (0.5% O2—panels d, e and f). Data presented are from at least three independent replicates performed in triplicate ± SEM. One-way ANOVA followed by Tukey’s multiple comparisons test was used to compare within different treatments, significant differences are represented by *p < 0.05, **p < 0.01, respectively

Full size image

Effects of AuNPs and atovaquone treatment on the induction of DNA DSBs

As DNA damage is the primary mechanism by which radiation triggers cell death, we next aimed to investigate if the combined effect of atovaquone and AuNP with radiation treatment increased DNA damage yields or interfered with DNA damage repair. 53BP1 foci were scored and used as a surrogate marker of DNA double strand breaks 2 h and 24 h post-radiation treatment. Figure 7a is representative images of FaDu 53BP1 foci. Radiation combined with either AuNP or atovaquone appeared to increase residual 53BP1 foci compared to 2 Gy alone, an effect that was augmented using combined AuNP/atovaquone, significantly (p = 0.0124) increasing residual DNA damage yields by 31% over radiation alone. Similar effects were also seen in CAL33 cells (Additional file 1: Figure S6), where the combination significantly (p = 0.0338) attenuated DNA damage over AuNPs alone, increasing unresolved DNA damage yields by 60%. This response was not observed in CAL27 cells (Additional file 1: Figure S7).

Fig. 7
figure 7

(a) Representative immunofluorescence images of 53BP1 foci in FaDu cells 24 h post-radiation (2 Gy) under 21% O2 and 0.5% O2. Cells were treated with either AuNPs (100 μg), atovaquone (AQ 30 μM) or the combination. Panels b and c represent the mean 53BP1 foci 24 h post-radiation (0 or 2 Gy) in FaDu cells after treatment with AuNPs, atovaquone or the combination under atmospheric oxygen levels (b) and hypoxia (c). Data presented are from at least three independent replicates performed in triplicate ± SEM. One-way ANOVA followed by Turkey’s multiple comparison test was used to compared within different treatment. * represents significant difference at p < 0.05, ** at p < 0.01, *** at p < 0.001

Full size image

Hypothesising that reduced oxygen consumption would increase intracellular oxygen levels, therefore leading to further DNA damage, we next assessed the impact of the combined treatment on radiation-induced DNA damage under hypoxia at both 2 h and 24 h post-radiation. Additional file 1: Figure S8 shows representative and quantified data of 53BP1 foci in FaDu, CAL27 and CAL33 cells. The radiation treatment induced a significant increase in DNA damage, producing 15.6 ± 1.7, 12.0 ± 2.8 and 13.6 ± 3.9 for FaDu, CAL27 and CAL33 cells, respectively. Interestingly, the combined treatment with radiation and atovaquone further enhanced DNA damage yields over radiation alone, increasing double strand break formation by 47%. Figure 7 a&c are representative images and quantified data of 53BP1 foci in FaDu cells under 0.5% O2, only this time 24 h post-irradiation, allowing time for DNA damage repair. Surprisingly, the mean residual DNA damage yields were elevated in the combination treatment, again by exactly 47%, indicating that the combination treatment increases DNA damage rather than interfering with DNA damage repair. In CAL27 and CAL33 cells, the combination of atovaquone and AuNPs plus radiation only slightly increased DNA DSB yields (Additional file 1: Figures S6, S7), failing to prove statistically significant.

Discussion

In the current study, AuNPs have been explored as an experimental radiosensitiser to enhance the biological effectiveness of conventional photon therapy due to their desirable physical and chemical properties (Cui et al. 2017; Choi et al. 2020) Various existing in vitro and in vivo studies have demonstrated the potential of AuNPs as effective radiosensitisers, capable of achieving dose enhancements comparable to those of approved radiosensitisers in routine clinical practice (López-Valverde et al. 2022; Khoo et al. 2017) However, clinical translation of AuNPs has been slow, attributed to several reasons including less well understood PK/PD properties over conventional small molecules, poor target cell internalisation and inconsistent impacts on redox activity, rendering these experimental agents less effective under hypoxic conditions(Jain et al. 2014; Cui et al. 2014). To overcome this challenge, we present a relatively unexplored strategy for increasing the radiosensitisation of AuNPs through the combined use of AuNPs with atovaquone, a metabolic modulator which could lead to increased tumour radiosensitivity through improved oxygenation. This approach could constitute a novel strategy in which tumour radiation sensitivity is improved secondary to metabolic remodelling rather than by altering intrinsic radiosensitivity.

AuNPs internalisation was concentration-dependent, displaying a clear positive correlation with treatment concentration under both normoxic and hypoxic conditions (Fig. 1). Of note, AuNP uptake under hypoxic conditions was at least comparable if not greater than that observed under aerobic conditions, indicating that hypoxic stress does not present a barrier to active AuNP internalisation. Indeed, emerging evidence suggests that prolonged hypoxic exposure increases the retention of AuNPs within cells due to a reduction in intracellular vesicle trafficking and exocytosis (Neshatian et al. 2015). Differences in the NPs uptake are largely determined by specific cellular processes including endocytosis, exocytosis and autophagy (Chithrani et al. 2006; Gao et al. 2005; Chithrani and Chan 2007). When nanoparticles enter cells by receptor-mediated endocytosis, normally they remain constrained within endosomes prior to lysosomal fusion, degradation, and exocytosis. Prolonged hypoxic exposure results in a reduction in the supply of nutrients and energy, suppressing energy dependent processes such as autophagy and exocytosis (Shorer et al. 2005; Ma et al. 2011). Our results are consistent with findings which report that prolonged hypoxic exposure results in higher AuNP uptake in MCF-7 and HeLa cells, compared to equivalent normoxic conditions.(Neshatian et al. 2015) Another plausible explanation is due to an increase in AuNP agglomeration under hypoxic conditions. Hypoxia, and in particular anaerobic metabolism is recognised as a common feature of most solid tumours, increasing extracellular lactic acid levels. This metabolic effect elevates the stromal pH of the tumour microenvironment, exposing AuNPs to environmental conditions that favour agglomeration.(Neshatian et al. 2015; Sonveaux et al. 2008).

“Metabolic radiosensitisation” is conceptually based on altering the metabolic demand for oxygen consumption primarily by downregulating mitochondrial oxidative metabolism (Ashton et al. 2018). Increasing evidence suggests that decreasing oxygen consumption contributes to a reduction in hypoxia induced radioresistance (Ashton et al. 2016; Mey et al. 2018); Atovaquone targets and inhibits the function of mitochondrial complex III, as such we selected this anti-malarial drug to assess its ability to modulate radiation sensitivity. Our results demonstrate that atovaquone efficiently and rapidly (within two hours) suppresses OCR across all three cell lines, corroborating observations from existing reports in alternative tumour models.(Kapur et al. 2022; Ashton et al. 2016) We also observed that atovaquone concomitantly triggered an increase in ECAR (Fig. 3d–f), indicating increased glycolytic function in response atovaquone mediated suppression of ATP production via oxidative phosphorylation. Such metabolic flexibility enables activation of various signalling pathways, compensating for loss of critical functions such as electron transport chain function, ultimately enabling survival under nutrient/oxygen deprivation. Additionally, the metabolic shift from oxidative metabolism to anaerobic glycolysis is recognised as contributing to undesirable effects with respect to the response to the radiotherapy, through elevated DNA damage repair and cancer stem cell enrichment (Bhatt et al. 2015; Li et al. 2016). Given the recognised role which glycolysis confers in mediating therapeutic resistance, glycolysis inhibitors could be used in combination with atovaquone to maximise the sensitising effect of these compounds in future studies.

AuNPs yielded significant positive radiation dose modifying effects in all three tumour cell models tested. Hypoxia is recognised as the main driver for the development of radioresistance via a series of mechanism, including impaired oxygen fixation at the time of radiation delivery, changes in cell cycle regulation, and/or the induction of anti-apoptotic genes (Boulefour et al. 2021). However, in our study, we found that hypoxic (0.5% O2) cells were still sensitised to radiation by AuNPs. AuNP uptake is recognised as a key factor influencing radiosensitising potential. Zhang et al. (2008) highlighted the role of surface functional groups on prostate cancer radiosensitivity comparing the uptake and radiosensitivity of thiol-glucose-capped AuNPs (Glu-AuNPs) and the neutral AuNPs (TGS-AuNPs), demonstrating that Glu-AuNPs resulted in a threefold increase in nanoparticle internalisation over TGS-AuNPs (Zhang et al. 2008). In keeping with the assertion that more intracellular gold results in greater radiosensitisation, Glu-AuNPs were more than twofold more effective when combined with radiotherapy compared with TGS-AuNPs. Our qualitative study using hyperspectral microscopy demonstrated that prolonged hypoxic exposure does not impair the nanoparticle uptake, nor differentially influence intracellular localisation (Fig. 1 and Additional file 1: Fig S1). Taken together, our quantitative and qualitative study of AuNPs uptake provide a solid rationale for AuNPs mediated radiosensitisation under hypoxic conditions.

The colony forming assay was used to establish both the direct impact of each individual component treatment on viability both ± radiation. Importantly, in the absence of radiation, neither AuNP treatment, atovaquone or combined AuNP/atovaquone treatment had any significant impact on colony formation, indicating that subsequent radiosensitisation effects are indeed a synergistic interaction as opposed to cumulative toxicity resulting from individual components (Additional file 1: Figure S3). Interestingly, treatment with atovaquone alone resulted in radiosensitisation in FaDu cells, leading to a significant increase in radiosensitivity (Fig. 4a), with greater effects observed when combined with AuNPs. More encouragingly, at the clinically relevant dose of 2 Gy, the combination of AuNPs and atovaquone significantly increased cell killing compared to radiation alone, producing a dose enhancement factor (DEF) of 1.5 over radiation alone (Additional file 1: Table S1). Similarly, the combination treatment also yielded the greatest effect under hypoxia (Additional file 1: Table S2), generating a mean DEF2Gy of 1.25. In a clinical setting, dose enhancement on this scale could enable dose de-escalation helping to spare surrounding normal tissue from the adverse off-target effect of radiotherapy, or conversely if retaining an equivalent treatment plan markedly reduce the risk of treatment failure.

Next, we investigated the effect of combined AuNP and atovaquone treatment on cell cycle distribution, apoptosis and radiation-induced DNA damage repair. As outlined above, AuNP radiosensitisation is predominantly attributed to physical interactions between photons and high-Z material.(Carter et al. 2007; Ghita et al. 2017) However, growing evidence suggests that AuNPs may also contribute to biological responses that arise following radiation interaction.(Rosa et al. 2017) At present, four potential mechanisms contributing to the biological influence on AuNP radiosensitisation have been proposed including: cell cycle synchronisation, increased production of ROS and oxidative stress, DNA damage induction and an enhanced radiation bystander effect.(Rosa et al. 2017); Roa et al. (2009) demonstrated AuNP mediated cell cycle acceleration through G0/G1 with an accumulation of cells in G2/M via the activation of CDK kinase (Roa et al. 2009). The consequences of CDK activation were shown to deliver striking radiosensitisation, suggesting that cell cycle synchronisation played an important role in mediating such effect. Similarly, Geng et al. (2011) found that thiol-glucose coated AuNPs promoted G2/M accumulation, contributing to an enhanced SKOV-3 cell sensitivity against MV X-ray radiation.(Geng et al. 2011) The current study demonstrated that combined AuNPs and atovaquone treatment induced normoxic S-phase accumulation (Fig. 5a). The sensitivity and subsequent biological impact of radiation is highly cell cycle dependent, with G2/M recognised as the most sensitive phase to radiation (Pawlik and Keyomarsi 2004). Herein, the increase of S-phase cells appears to positively correlate with the radiosensitising effect, however, a mechanism for S-phase arrest remains unknown. Preliminary work suggests that functionality of the master tumour suppressor gene TP53 is believed to play a key role in mediating S-phase specific cell death (Ostruszka and Shewach 2000). All three cell lines used in the current study are TP53 mutant, therefore, these cells do not possess a competent G1-S cell cycle checkpoint. As such, cell cycle progression is not stalled, limiting the opportunity for an increase in p21 to resolve or repair the damaged DNA. The ability to progress through cell cycle after radiation-induced damage is thought to synergistically enhance cell death, presenting a biological mechanism of radiosensitisation.(Bélanger et al. 2014) As S-phase cells are considered more radioresistant than G1-S or G2-M cells, it can therefore be concluded that the radiosensitising effects caused by the combination of AuNPs and atovaquone are unlikely to be directly linked to the redistribution of cells to a more radiosensitive cell cycle phase (Pawlik and Keyomarsi 2004). Despite this, increased S-phase radiosensitivity is not unique to the current study. Kuno and Shinomiya (2000) demonstrated enhanced S-phase tumour killing activity following the treatment with the hypoxic radiosensitiser (PR-000350) (Kuno and Shinomiya 2000). Additionally, S-phase specific apoptosis is directly associated with a defect in nucleotide excision repair (NER), an important DNA damage repair process. Bélanger et al. (2014) reported that defective NER during S-phase increases apoptosis and reduces clonogenic survival in human melanoma cell lines following UV radiation treatment, an effect is mediated via the diminished activation of ataxia telangiectasia and Rad 3-related (ATR) kinase (Bélanger et al. 2014).

As S-phase specific death is linked to the activation of apoptosis, we also investigated the combined treatment effect of AuNPs and atovaquone on apoptotic fraction (Ray et al. 2007). Surprisingly, our work indicated that at the doses used atovaquone, AuNP or radiation alone did not significantly increase the apoptotic fraction, despite observing increased clonogenic radiosensitisation. This could likely be explained by delayed radiation cell death due to mitotic catastrophe. Firat et al. (2011) reported similar findings in a stem-like glioma cell (SLGCs) model that failed to trigger apoptosis four days after exposure to a single 10 Gy dose.(Firat et al. 2011) However, four days post-radiation treatment 50% of the SLGCs tumours treated with 5 Gy or more observed delayed cell death. This effect was predominantly associated with cells lacking wild type TP53, attributing cell death to mitotic catastrophe. This finding highlights the important role of p53 in regulating the response to genotoxic damage. Interestingly, the combined effect of atovaquone, AuNP and radiation did significantly trigger the apoptosis in FaDu and CAL27 cells, within a comparatively short experimental timeframe. Gao et al. (2018) demonstrated that atovaquone significantly induced apoptosis in a concentration-dependent manner, inhibiting both cell proliferation and angiogenesis in vivo (Gao et al. 2018). Furthermore, Fu et al. (2020) reported that atovaquone triggers apoptosis in EpCAM+CD44+ tumour initiating HCT116 cells under hypoxia (Fu et al. 2020). Here we observed that AuNPs treatment alone is not enough to elicit strong apoptotic response, however, the apoptotic fraction was increased following the combination treatment (Fig. 6a). This implies that stress caused by the inhibition of mitochondrial respiration by atovaquone may prime cells toward radiation-induced apoptosis.

Impaired DNA damage repair is another fundamental process that may contribute to the observed effects. Our data indicate that combined AuNPs and atovaquone results in more DNA damage at 2 and 24 h post treatment compared to cells treated with AuNP alone (Fig. 7 and Additional file 1: Fig S8). This effect is consistent with a previous report that 50 nm citrate capped AuNPs induce significant radiosensitisation in HeLa cells (Chithrani et al. 2010). To date, there are no direct studies evaluating DNA damage levels after atovaquone and radiation treatment. However, some reports suggests that atovaquone exhibits antitumour effects by increasing DNA damage. Ashton et al. (2016) reported that atovaquone increased radiosensitivity by inhibiting pyrimidine synthesis, thus negatively impacting both DNA damage repair and replication. Additionally, Gao et al. (2018) showed that atovaquone inhibits hepatoma cell proliferation by inducing double stranded DNA breaks, which contribute to the sustained ataxia -telangiectasia mutated (ATM) activation, and downstream molecules including cell cycle checkpoint kinase-2 (CHK2) and H2AX (Gao et al. 2018). In our work, we consistently observed a significant increase in the induction of lethal DNA double strand breaks (DSBs) after treatment with the combination of AuNP and atovaquone at both 2 h and 24 h (Fig. 7 and Additional file 1: Fig S8), an effect consistent with our clonogenic survival data. Interestingly, we observed that combination of atovaquone and AuNPs induce 47% increase in the yield of DSB induction for both initial (2 h) and residual damage (24 h) under hypoxia, indicating that the enhanced radiosensitivity from the combination of AuNPs and atovaquone may be due to the induction of greater DNA damage rather than negatively impacting the DNA damage repair process.

Conclusions

Overall, this work demonstrates the potential of combining AuNPs and atovaquone as a novel approach to enhance the radiotherapeutic effect in HPV( − ) HNSCC. Treatment with atovaquone greatly reduced oxygen consumption in HNSCC cells, inducing a shift to a more glycolytic type, providing the basis for metabolic radiosensitisation. This approach appears to complement physical radiosensitisation conferred by AuNPs, inducing S-phase accumulation, contributing to greater DNA damage yields. Importantly, the combination results in superior radiotherapeutic effects compared to either agent alone. While this work provides a strong rationale for the combined use of physical and metabolic radiosensitisers, clinical translation will require careful consideration of critical parameters such as route of administration to avoid enhanced radiosensitivity of surrounding normal tissue. Given that external beam radiotherapy is delivered in a highly conformal manner, tumour specific radiosensitisation is by default achieved through accurate radiotherapy delivery. Furthermore, by adopting direct intra-tumoural injection as the means of delivery, the risk of off-target nanoparticle radiosensitisation can be further mitigated against. Indeed, this is the approach utilised by successful radiosensitising clinical studies using hafnium oxide nanoparticles, and in our opinion the option most likely to ultimately deliver clinical benefit from similar nanotechnologies.