Verification of radiodynamic therapy by medical linear accelerator using a mouse melanoma tumor model

Combined treatment with 5-aminolevulinic acid (5-ALA) and X-rays improves tumor suppression in vivo. This is because the accumulated protoporphyrin IX from 5-ALA enhances the generation of ROS by the X-ray irradiation. In the present study, a high-energy medical linear accelerator was used instead of a non-medical low energy X-ray irradiator, which had been previously used. Tumor-bearing mice implanted with B16-BL6 melanoma cells were treated with fractionated doses of irradiation (in total, 20 or 30 Gy), using two types of X-ray irradiator after 5-ALA administration. Suppression of tumor growth was enhanced with X-ray irradiation in combination with 5-ALA treatment compared with X-ray treatment alone, using both medical and non-medical X-ray irradiators. 5-ALA has been used clinically for photodynamic therapy. Thus, “radiodynamic therapy”, using radiation from medical linacs as a physical driving force, rather than the light used in photodynamic therapy, may have potential clinical applications.

For cancer treatment, radiotherapy is preferred to surgical resection because it is non-invasive, and preserves organ structure and function. However, acute toxicity and potential long-term adverse effects often limit the dose of radiation to levels that are insufficient for controlling tumors, particularly when a tumor resides in close proximity to a radiosensitive organ. Use of tumor-selective agents that enhance radiation effects in tumors, but spare normal tissue, can improve therapeutic efficacy, by ameliorating local control of the tumor without increasing the radiation dose. Some possible radiosensitizers induce apoptosis by inhibiting nucleic-acid synthesis, angiogenesis, DNA repair, and cell signaling 1,2 . Others affect cellular redox stress 3,4 . To date, no radiosensitizers with clinical benefits have been discovered.
We reasoned that the effects of radiation might be enhanced by amplifying the radiation dose by physico-chemical reaction, because it is easy to control the radiation dose. Organic scintillators are materials that exhibit scintillation, the property of luminescence, when excited by ionizing radiation during physicochemical reactions. The most common scintillators are anthracene, stilbene, and naphthalene 5 ; however, these can be toxic to cells. Much less work has been done on the use of the physicochemical reactions of organic materials under ionizing radiation for radiosensitization.
Previously, we investigated protoporphyrin IX (PpIX) as a candidate oncotropic radiosensitizer. Specifically, we measured the types and amount of reactive oxygen spices (ROS) generated by X-ray irradiation using ROS detection reagents, with ethanol as a quencher of ROS in solutions containing different concentrations of PpIX. We estimated the amount of the major ROS-the hydroxyl radical (•OH), superoxide anion (O 2 •− ), and singlet oxygen ( 1 O 2 )-resulting from PpIX treatment 6 . Systemic application of 5-aminolevulinic acid (5-ALA) leads to selective accumulation of PpIX in tumor cells. Subsequent excitation with light forces PpIX into its singlet state, which emits fluorescence upon returning to the ground state, and producing ROS 7,8 . 5-ALA has been used clinically for fluorescence-guided surgery and photodynamic therapy (PDT) 9,10 . Our findings suggest that it is possible to use X-rays as an energy source instead of light, as used in PDT.
5-ALA pre-treatment increases the efficacy of radiotherapy with multi-dose ionizing radiation with 100-160 kV X-rays in animal cancer models 11,12 . The most commonly used radiotherapy device in practice today is Microarray analysis. Microarray analysis of tumor samples irradiated with the linear accelerator was performed as previously described 14 . Data were normalized by quantile normalization, and were analyzed using GeneSpring GX software version 10.0.1 (Agilent Technologies, CA, USA). The gene ontology (GO) database (http://www.geneontology.org/) was used to functionally categorize the gene expression profiles. GO terms were obtained from Agilent Technologies eArray (https://earray.chem.agilent.com/earray/).
Quantitative RT-PCR (qRT-PCR). Total RNA was subjected to reverse transcription using iScript ™ reverse transcription super-mix (Bio-Rad Laboratories, CA, USA), and used for RT-PCR in the 2-step qRT-PCR assays. Gene-specific primers were used to amplify target transcripts using an MJ mini personal thermal cycler (Bio-Rad Laboratories, CA, USA), using Sso Fast Eva Green Supermix (Bio-Rad Laboratories, CA, USA), according to the manufacturer's instructions. Primer sequences for amplification of mouse Actb (ID 6671509a1), mouse p21 (ID 162287332c1), mouse Gadd45a (ID 6681149a1), mouse Sod2 (ID 31980762a1), and mouse Gpx7 (Primer Bank ID 13195626a1) were obtained from Primer Bank, a public resource for PCR primers 15 . Relative transcript quantities were calculated using the ΔΔCt method with Actb as the reference gene.

Evaluation of ROS generation in vivo.
To evaluate ROS formation, we measured luminescent signals produced by coelenterazine, which is used as a probe for O 2 •− in vivo 16,17 . Native or unmodified coelenterazine was purchased from Cyman Chemical (MN, USA). Coleneterazine was dissolved in ethanol and diluted using PBS (final ethanol concentration, <2%). To evaluate the luminescent signals of coelenterazine in vitro, PpIX concentrations ranging from 0 to 1 µg/mL were exposed to 1 µM colenterazine, and irradiated (0, 3, 10, or 30 Gy X-ray) in a microplate containing only reagents and no cells. To detect ROS generation in vivo, X-ray irradiation was performed after the tumor volume reached approximately 250 mm 3 . The 5-ALA treatment groups were locally administrated 5-ALA at a concentration of 50 mg/kg, about 4 to 5 hours before X-ray irradiation. Mice were anesthetized using isoflurane and administered 0.1 mg/kg native coelenterazine locally, 10 min before subjecting them to 3 Gy irradiation using the Faxitron CP-160 irradiator. Luminescent signals were measured before and after X-ray irradiation using IVIS 100 (Xenogen, CA, USA), which consists of a light-tight chamber equipped with a cooled CCD camera. Data are represented as the total flux (p/s). Statistical analysis. Tumor volume, tumor weight, body weight, gene expression based on qRT-PCR or microarray, cell cycle distribution, and chemiluminescent signal produced by coelenterazine were analyzed by SCIeNTIFIC RepoRts | (2018) 8:2728 | DOI:10.1038/s41598-018-21152-z one-way factorial ANOVA followed by Tukey-Kramer multiple comparisons test. In cases where the variance was not homogenous, the Games-Howell post hoc test was used. Differences were considered statistically significant at p < 0.05. Pearson's correlation coefficient was calculated to identify correlations in gene expression among individual microarray data sets. Before analysis by these multiple comparisons tests, Fisher's Z-transform was used to fit the data to a normal distribution. Hierarchical clustering using the Euclidean distance method and Ward's linkage was also performed on the 22 arrays.

Results
Combined treatment with 5-ALA and X-ray irradiation. A C57BL/6J melanoma tumor model was used to evaluate the effect of irradiation dose in combined treatment with 5-ALA on tumor suppression in vivo. In the present study, we used two types of X-ray irradiation devices: a linear accelerator with 4 MeV beam energy, and a cabinet type X-ray irradiator with 160 kV nominal X-ray tube voltage. The tumor size decreased with increasing total irradiation dose, and was notably reduced in mice receiving 5-ALA treatment plus a total dose of 20 Gy of irradiation, regardless of X-ray device used (Fig. 1A,B). Tumor weight decreased in a similarly , and a CP-160 Cabinet X-Radiator ™ System (nominal X-ray tube voltage: 160 kV) (B,D,F). C57BL/6J mice with B16-BL6 cells were divided into the following 6 groups: NT, no treatment; ALAT, 5-ALA treatment; 20XT, 2 Gy/day for 10 days; ALA-20XT, 5-ALA treatment followed by 2 Gy/day for 10 days; 30XT, 3 Gy/day for 10 days; ALA-30XT, 5-ALA treatment followed 3 Gy/day for 10 days. Data are shown as means ± SD (n = 4 or 5, *p < 0.05, **p < 0.01). Tumor size was measured after inoculation with B16-BL6 at day 0 (A,B). Tumors were excised from animals and weighed after 10 sessions of fractionated irradiation (C,D). Body weights were measured 3 times a week (E,F).
SCIeNTIFIC RepoRts | (2018) 8:2728 | DOI:10.1038/s41598-018-21152-z dose-dependent fashion, and was notably reduced in mice receiving 5-ALA plus a total dose of 20 Gy and 30 Gy of irradiation (Fig. 1C,D). The measurement of carefully cleaned tumor weight provided a more precise measurement of tumor growth suppression, as some skin was included in tumor size measurements using calipers. 5-ALA plus irradiation synergistically suppressed B16-BL6 tumor growth at the integral doses of 20 Gy and 30 Gy using both X-ray irradiation devices, without decreasing body weight (Fig. 1E,F).

Morphological observation of tumor tissues.
The morphological characteristics of the tumors were obtained by HE and PI staining. The control NT and ALAT tumor tissue cells were fairly uniform ( Fig. 2A,B), whereas the X-ray treated tumor cells were not (Fig. 2C-F). X-ray treatment resulted in the formation of giant cells with aberrant nuclear morphologies due to mitotic catastrophe. The NT and ALAT cells had uniform nuclei (Fig. 2G,H), while cells with enlarged or shrunken nuclei were observed in the X-ray treated tumor tissue. Cells with fragmented nuclei, or cells without stained nuclei, were typically observed in the ALA-20XT, 30XT, and ALA-30XT tissue (Fig. 2J-L).

Analysis of gene expression profiles by microarray.
We further characterized the effects of 5-ALA and linac X-ray irradiation on gene expression by microarray. Each sample (NT, ALAT; n = 3 20XT, ALA-20XT, 30XT, ALA-30XT; n = 4) was analyzed using a mouse gene expression microarray consisting of 43,379 oligonucleotide probes.
We evaluated the variation in the correlation coefficients of individual samples both within and between treatment groups. Pearson's correlation coefficient was used for the correlation analysis. The mean correlation coefficients for gene expression profiles among individuals within a treatment group are shown in Fig. 3A, and reflect the individual differences in gene expression. The correlation coefficients of the gene expression in tumors in the 30XT group were higher than in other treatment groups, which is consistent with reduction in tumor weight variation observed in this group compared to that of the other groups. We also analyzed the expression profile correlation between the different treatment groups vs. that of ALA-20XT (Fig. 3B). The differences in correlation coefficient between the groups without X-ray treatment and the groups with X-ray treatment were statistically significant. The differences in correlation coefficient between 30XT vs. ALA-20XT, or ALA-30XT vs. ALA-20XT, were higher than between 20XT vs. ALA-20XT. This suggests that the gene expression in the ALA-20XT samples was more similar to that in the 30XT or ALA-30XT samples than in the 20XT samples.
Hierarchical clustering was performed on the 22 arrays using the Euclidean distance method and Ward's linkage. The samples clustered into three clusters (Fig. 3C): one containing the non-irradiated groups (NT and ALAT), a second containing 20XT and half of 2ALA-20XT, and a third containing 30XT, ALA-30XT and half of ALA-20XT. We therefore surmised that the gene expression profile of ALA-20XT was located in the middle of the gene expression profile of 20XT and 30XT/ALA-30XT.
Functional analysis and functional validation using marker genes. We selected genes with altered expression in each treatment group compared to the non-irradiated groups (NT and ALAT) based on their p-values (p < 0.01). Based on treatment, these genes were separated into up-and down-regulated groups (Supplementary Table 1). The selected genes were functionally annotated in DAVID. We then extracted GO terms based on the number of genes in each GO category. Supplementary Table 2 shows the differentially represented (B) Correlation coefficients between different treatment groups. The mean correlation coefficient between individuals in the NT and ALA-20XT groups is represented as "NT vs. ALA-20XT". The lower and upper limits of the boxes represent the 25th and 75th percentiles, respectively. The lower and upper whiskers denote the minimum and maximum values, respectively. Fisher's Z-transformation was used to normalize the correlation distribution, and continuous variables were analyzed using one-way factorial ANOVA followed by the Tukey-Kramer multiple comparisons test. In cases where the variances were not homogenous, a Games-Howell test for multiple groups was performed (*p < 0.01). (C) Hierarchical clustering of 22 arrays from 6 groups, with Euclidean similarity measure and Ward's linkage to visualize the expression profiles of genes between groups. The heat map shows the gene expression for arrays in rows, and the dendrogram representing their similarity. Clustering was performed using normalized signal intensity values for all 22 arrays. GO terms in the treated groups vs. the non-irradiated groups (FDR < 0.01). The GO terms associated with the downregulated genes in the treated groups were mostly related to cell cycle, including DNA metabolic processes and RNA metabolic processes. The GO terms associated with the upregulated genes included oxidation-reduction processes. We investigated whether the cell cycle was disrupted by 5-ALA treatment prior to X-ray irradiation in vitro. The population in G2/M phase increased with increasing X-ray irradiation doses, and this phenomenon was further enhanced by 5-ALA treatment (Supplementary Figure 1).
Based on the functional analysis results, we examined the expression level of genes that are representative of cell-cycle regulation and oxidation-reduction. We selected p21 and Gadd45a as marker genes representative of cell-cycle regulation, and Sod2 and Gpx7 as marker genes representative of oxidation-reduction (Fig. 4A-D). As a whole, the expression changes identified by the microarray analysis were confirmed by the qRT-PCR analysis. The expression of p21 increased following X-ray irradiation, regardless of irradiation dose or 5-ALA treatment. Gadd45a and Sod2 expression increased with increasing irradiation dose, while Sod2 expression was further enhanced by 5-ALA treatment. Gpx7 was not expressed in cells exposed to 20XT treatment, but was expressed in cells exposed to ALA-20XT, 30XT, and ALA-30XT treatment. Luminescent signals produced by coelenterazine detected in response to X-ray irradiation. We tried to evaluatedmeasure ROS production in tumor caused by X-ray irradiation. Chemiluminescence could be detected, and was observed to be dependent on the PpIX concentration and X-ray irradiation dose in a microplate containing only reagents and no cells (Fig. 5A). For evaluating ROS generation in vivo, we measured the luminescent signals produced by coelenterazine after 3 Gy X-ray irradiation. Immediately after X-ray irradiation, the intensity of luminescence increased and subsequently decreased with time, with or without 5-ALA treatment. When irradiation was subsequently repeated, the signal again increased and decayed over time (Fig. 5B). It was observed to be more conspicuous in tumor treated with 5-ALA (Fig. 5C,D). Since coelenterazine detects superoxide 15,16 , 5-ALA treatment potentially increased superoxide production.

Discussion
The most common form of radiation used in practice today is the high-energy photon, which is created electronically by devices such as medical linear accelerators. The interaction between photon and tissue differs with photon energy. The Compton effect is the most common clinically occurring interaction, as most radiation treatments are performed at energy levels of about 6-20 MeV 18 . We previously studied the combined effect of 5-ALA and X-ray irradiation using a tube voltage of 100-160 kV, in which the photoelectric effect predominates in tissue 11,14 . In the present study, we evaluated the tumor suppression effects of two kinds of devices that produce different energy: a linear accelerator with 4 MeV beam energy, and a cabinet type X-ray irradiator with 160 kV nominal X-ray tube voltage. The effective energy of the linear accelerator was estimated to be about 2 MeV 19 . The operating dose rate was 2.5 Gy/min. A 0.5 mm thick aluminum plate was used for CP-160, and the dose rate was 1 Gy/min. The X-ray spectrum of this source is broad, with a mean X-ray energy of 75 keV 20 . Although two devices had large differences in their energy outputs, tumor suppression improved similarly in animals treated by both (Fig. 1). A combined treatment with 5-ALA seems to improve the effect, for the dose rates used in this study.
The mechanism of tumor suppression by combined treatment with X-ray and 5-ALA is not fully understood. Excess exogenous 5-ALA leads to a build-up of PpIX, which accumulates selectively in epithelial tissues and tumors 9 . PpIX not only contributes to enhanced generation of •OH in the presence of X-ray irradiation but also •− and 1 O 2 by physicochemical reactions 6 . ROS generated by PpIX accumulating in tumor cells caused by X-ray irradiation caused cell cycle arrest 14 . 5-ALA is thus thought to improve the efficacy of cancer radiotherapy by acting as a radiomediator 11 . Although the reactive oxygen species are different, this method, named "radiodynamic therapy", used radiation as a physical driving force instead of the laser light used in photodynamic therapy (Fig. 6).
Kennedy reported topical application of 5-ALA for sensitization of cutaneous basal cell carcinomas for PDT 21 . Grant showed that oral squamous cell carcinoma can synthesize and accumulate photosensitizing levels of PpIX 22 . For other cancers, such as brain tumors, bladder cancer, and prostate cancer, the accumulation of PpIX following systemic administration of 5-ALA is considered to be a common characteristic of cancer 23 . Because radiation can treat cancer at a greater depth than laser light, radiodynamic therapy could more useful for many additional types of cancer than photodynamic therapy.
Microarrays allow for the detection of genome-wide perturbations during treatment, and the measurement of responses by gene probes. Previously, we examined the effect of combined treatment with X-ray irradiation and 5-ALA using CP-160, which induced changes in the expression of genes related to cell-cycle arrest 14 . In the present study, we achieved similar results using linac (Supplementary Table 2). We also investigated the effect of radiation dose, and found that the gene expression profile of ALA-20XT is located between that of 20XT and 30XT/ALA-30XT according to clustering and correlation analysis (Fig. 3B,C).
The biological effect of radiotherapy is caused by damage from either direct or indirect ionization of the atoms that make up the DNA chain. Indirect ionization occurs because of the ionization of water, forming free radicals, notably •OH. PpIX, which accumulates in cell following the administration of 5-ALA, causes the production of 1 O 2 or O 2 •− during X-ray irradiation. GADD45 is a growth arrest and DNA-damage-inducible protein, while P21 is a cyclin dependent kinase inhibitor that is induced following DNA damage. Both p21(Cdkn1a) and Gadd45a are typically induced by X-ray irradiation. These genes have the largest and most persistent responses to X-ray exposure, and have previously been used for radiation dosimetry by expression analysis in blood from 20 normal healthy human donors 24 . In the present study, the mRNA levels of p21 and Gadd45a increased following X-ray irradiation, regardless of dose or 5-ALA treatment (Fig. 4A,B). The expression of antioxidant enzymes Sod2 and Gpx7 increased with increased dose, and with 5-ALA treatment (Fig. 4C,D).
In the in vitro cell cycle assay, we observed that 2-Gy X-ray irradiation increased cell cycle arrest at the G2/M phase. This phenomenon increased slightly with 3-Gy irradiation, while the cell cycle arrest increased in both  Figure 1). Since the in vitro irradiation dose is equal to one day's dose in vivo, the effect of irradiation in vivo should be 10 times that of what is seen in vitro. Thus, 5-ALA treatment seems to enhance the suppression of tumor growth with X-ray treatment regardless of irradiation dose by enhancing the cell cycle arrest.
The lifetime of ROS is short. Also, it is difficult to operate measuring devices simultaneously with X-ray irradiation. For this reason, detection of ROS generated due to X-rays is extremely difficult. In the present study, the luminescent signals produced by coelenterazine after 3 Gy X-ray irradiation increased and subsequently decreased with time, and became more conspicuous in tumor with 5-ALA treatment (Fig. 5C,D). If the signal is due to ROS, O 2 •− should be measured 16,17 . Meanwhile, we observed PpIX contributes towards the enhanced generation of •OH, O 2 •− , and 1 O 2 in the presence of X-ray irradiation 6 . Further studies are required for understanding whether ROS are directly generated by X-rays or generated secondarily, or what biological effect the different kinds of ROS have as a result of 5-ALA treatment.
Melanomas are known to be resistant to radiation. This could be explained by the activation of certain oncogenes, which decreases the sensitivity of cells towards ionizing radiation 25 . Modern techniques, such as stereotactic radiosurgery and stereotactic radiotherapy, are useful for treating high-risk melanomas to prevent locoregional recurrence; however, these techniques have failed to improve the overall survival rate. A number of case reports on the use of radiotherapy combined with targeted therapies, such as the use of BRAF inhibitor or lpilimumab, were published 26,27 . They can increase the radiosensitivity of melanomas cells and normal tissues that can lead to an increased risk of toxicity. On the other hand, 5-ALA, which shows high selectivity for cancer cells with very low toxicity, has been approved for use in high-grade glioma surgery as a precursor of the natural photosensitizer PpIX in PDD and/or PDT in Europe, Japan, and USA 28 . Further in vivo studies on all types of melanomas and clinical trials are needed to investigate the usefulness of combined treatment of melanoma with radiotherapy and 5-ALA.
In summary, we have demonstrated that 5-ALA works as a radiosensitizer with X-ray irradiation, using both a high-energy medical linac accelerator as well as a low energy source. 5-ALA treatment improved tumor suppression when combined with X-ray irradiation in the ALA-20XT group regardless of X-ray apparatus. The gene expression profiles of the ALA-20XT group resemble those of 30XT and ALA-30XT along with shrinking of tumor volume. In terms of molecular response, X-ray irradiation seems to cause cell-cycle arrest, and 5-ALA treatment enhanced the tumor suppression at both irradiation doses. 5-ALA has been used clinically for fluorescence-guided surgery and photodynamic therapy. Thus, radiodynamic therapy, which uses radiation as the physical driving force instead of the laser light used in photodynamic therapy, might practical clinical applications. Further studies must first establish practical procedures to optimize the use of radiodynamic therapy with combined 5-ALA treatment. , and 1 O 2 in the presence of X-ray irradiation. 1 O 2 is thought to be a major ROS produced by photodynamic therapy.