YC-1 sensitizes the antitumor effects of boron neutron capture therapy in hypoxic tumor cells

Abstract The uptake of boron into tumor cells is a key factor in the biological effects of boron neutron capture therapy (BNCT). The uptake of boron agents is suppressed in hypoxic conditions, but the mechanism of hypoxia-induced modulation of suppression of boron uptake is not clear. Therefore, we evaluated whether hypoxia-inducible factor 1α (HIF-1α) contributes to attenuation of the antitumor effects of BNCT in hypoxic tumor cells. We also tested whether YC-1, a HIF-1α-targeting inhibitor, has therapeutic potential with BNCT. To elucidate the mechanism of attenuation of the effects of BNCT caused by hypoxia, deferoxamine (DFO) was used in experiments. Cells were incubated in normal oxygen, hypoxic conditions (1% O2) or 5 μM DFO for 24 h. Then, cells were treated with 10B-boronophenylalanine (BPA) for 2 h and boron accumulation in cells was evaluated. To clarify the relationship between HIF-1α and L-type amino acid transporter 1 (LAT1), gene expression was evaluated by a using HIF-1α gene knockdown technique. Finally, to improve attenuation of the effects of BNCT in hypoxic cells, BNCT was combined with YC-1. Boron uptake was continuously suppressed up to 2 h after administration of BPA by 5 μM DFO treatment. In cells treated with 5 μM DFO, LAT1 expression was restored in HIF-1α-knocked down samples in all cell lines, revealing that HIF-1α suppresses LAT1 expression in hypoxic cells. From the results of the surviving fraction after BNCT combined with YC-1, treatment with YC-1 sensitized the antitumor effects of BNCT in cells cultured in hypoxia.


INTRODUCTION
Boron neutron capture therapy (BNCT) is a unique treatment technique in which cancer cells are damaged by alpha particles ( 4 He) and 7 Li nuclei, which are both high linear energy transfer particles derived from the nuclear transmutation reaction of 10 B and thermal neutrons. The path lengths of these high linear energy transfer particles are about 9 and 5 μm, respectively, which are almost the same as the tumor cell diameter. Therefore, if 10 B accumulates selectively in cancer cells, theoretically, only cancer cells could be destroyed without damaging surrounding normal cells. p-Boronophenylalanine (BPA) is generally used as a boron agent for BNCT. For effective antitumor activity, 20 ppm or more 10 B must be taken up into tumor cells [1]. Thus, a large amount of boron drug must be administered to the patient. This can be achieved because BPA has almost no pharmacological effect or toxicity.
Damage to normal tissue adjacent to the tumor in the irradiation field is relatively low compared to photon therapy and particle beam therapy, depending on tumor cell selectivity for BPA [2][3][4]. Therefore, BNCT is considered less burdensome on heavily pretreated cancer patients. Especially for recurrent tumors following radiotherapy, re-irradiation with conventional radiotherapy brings a significant risk, as the cumulative dose of surrounding normal tissue has already reached near the tolerable dose by first radiotherapy. However, BNCT is considered to be relatively safe for re-irradiation because it can minimize the dose to normal cells adjacent to tumors, depending on low boron uptake [5]. Therefore, recurrent glioblastoma and recurrent head and neck cancer after standard treatment are considered good indications for BNCT. 10 B-BPA, which is used most frequently in BNCT, is a chemically modified phenylalanine with a boronic acid residue that is taken up • 524 YC-1 sensitizes the effects of BNCT in hypoxia • 525 by cancer cells through the L-type amino acid transporter 1 (LAT1), which mediates retrograde transport of other amino acids [6]. When treating patients, administration is started 2 h before neutron irradiation, and BPA selectively accumulates in tumor cells via LAT1. The boron concentration in blood is maintained at 20-40 ppm and irradiation is then performed. 18 F-BPA, in which 18 F is added to BPA, accumulates in tumors identically to BPA. Recently, a 4-borono-2-18 Ffluoro-phenylalanine positron emission tomography (PET) scan was performed to evaluate approximately whether 10 B-BPA accumulates in the tumor before BNCT [7,8].
Tumors have hypoxic regions with poor oxygen supply due to abnormal angiogenesis and rapid growth leading to insufficient blood flow [9]. In hypoxic tumor cells, hypoxia-inducible factor 1α (HIF-1α) accumulates as an adaptive response to hypoxia. As a result of gene expression that is regulated by HIF-1α, changes that promote tumor growth, such as adaptation of nutrient metabolism, hyperangiogenesis and enhancement of infiltration ability, occur [10,11], and resistance to therapeutics such as chemotherapeutic agents and molecular targeted agents is acquired.
A few papers have reported the effects of hypoxia on BNCT treatment. BPA uptake into tumor cells incubated in chronic hypoxic conditions is suppressed significantly depending on the oxygen concentration [12,13]. Thus, hypoxia may suppress LAT1 expression, resulting in suppression of BPA uptake into tumor cells and attenuation of the antitumor effects of BNCT. If the molecular mechanism of hypoxia-induced modulation of LAT1 expression is clarified, new therapeutic targets may be developed that enhance the treatment effect of BNCT. Therefore, in this study, we investigated whether HIF-1α contributes to attenuation of the antitumor activity of BNCT and evaluated whether inhibitors targeting HIF-1 have therapeutic potential.

Cell culture and hypoxic conditions
The human glioblastoma cell line T98G, human oral squamous cell carcinoma cell line HSC-3 and human breast adenocarcinoma cell line MCF-7 were provided by the Cell Resource Center for Biomedical Research, Institute of Development, Aging and Cancer, Tohoku University (Sendai, Japan). T98G cells were cultured in serum-free Dulbecco's modified Eagle's medium/Ham's nutrient mixture F-12 (DMEM/F12 1:1; FUJIFILM Wako Pure Chemical, Osaka, Japan), and HSC-3 and MCF-7 cells were cultured in serum-free DMEM (FUJIFILM Wako Pure Chemical). All media were supplemented with 10% fetal bovine serum (Sigma-Aldrich) and 1% penicillin/streptomycin (Gibco Life Technologies, Waltham, MA, USA), and cells were maintained at 37 • C in a 5% CO 2 atmosphere. Hypoxic conditions were achieved by culturing cells in modular incubator chambers (Billups-Rothenberg Inc., Del Mar, CA, USA). The chambers were flushed with mixed gas (95% N 2 and 5% CO 2 ) to achieve hypoxic conditions (1% O 2 ). After flushing, the oxygen concentration was confirmed using a JKO-02 Ver. III oxygen monitor ( JIKCO, Tokyo, Japan), and the chambers were sealed and incubated at 37 • C. Cells were seeded at a density of 3.0-5.0 × 10 5 cells depending on the growth rate, and after 12 h, cells were incubated in normal oxygen (21% O 2 ) or hypoxic conditions. The study protocol is shown in Fig. 1.

Trypan blue dye exclusion test
Cells were plated into 24-well plates at 1.0 × 10 5 cells/well and cultured with each DFO and YC-1 concentration. After 24 and 48 h of incubation, cells were stained using 0.5% Trypan Blue Stain Solution (Nacalai Tesque, Inc., Kyoto, Japan) and cell viability was assessed by counting the number of unstained cells with a hemocytometer.

Fluorescence imaging
To evaluate the intracellular oxygen state in a viable cell, fluorescence imaging was performed using a hypoxia live imaging agent, mono azo rhodamine (MAR; GORYO Chemical, Inc., Hokkaido, Japan). Cells were plated onto 8-well chamber slide at 1.0 × 10 5 cells/well and incubated for 24 h, then 1.0 μM MAR was added into the dishes and cells were incubated in normal oxygen conditions, hypoxic conditions or treated with 5 μM DFO in normal oxygen conditions for 6 h. Total RNA extraction and quantitative real-time polymerase chain reaction Total RNA was extracted using NucleoSpin ® RNA (Takara Bio Inc., Otsu, Shiga, Japan) according to the manufacturer's protocol. Firststrand cDNA was synthesized with an iScript RT Supermix for RT-qPCR ® (Bio-Rad, Hercules, CA) from extracted total RNA according to the manufacturer's protocol. Gene expression was assessed using quantitative real-time polymerase chain reaction (qRT-PCR) with SsoAdvanced Universal SYBR Green Supermix ® (Bio-Rad), with typical amplification parameters (95 • C for 30 s followed by 60 cycles at In both time courses, after incubation for 24 h, total RNA was extracted and first-strand cDNA was synthesized. LAT1 expression was assessed by qRT-PCR. In the time course in (A), BPA was added to the culture medium with short-term interruption of the hypoxic condition, and then returned to the previous condition. After treatment with BPA for 2 h, boron accumulation was measured by inductively coupled plasma atomic emission spectroscopy. To evaluate the surviving fraction, after treatment with BPA for 2 h, cells were exposed to neutron beams. After neutron irradiation, cells were resuspended by trypsinization, plated onto dishes, cultured for 10-13 days and then used for the clonogenic assay.  hypoxia, the cDNA concentration of each sample after the reverse transcription reaction of extracted total RNA was normalized using Qubit 4 Fluorometer (Invitrogen; Thermo Fisher Scientific, Inc.) and then RT-PCR was performed. The oligonucleotide primer sets used for qRT-PCR purchased from Takara Bio Inc. are shown in Table 1.

Western blot analysis
Cells were incubated in 35-mm dishes at a density of 5.0 × 10 5 cells.   Boron accumulation study 10 B-Enriched L-BPA solution and yttrium inductively coupled plasma standard solution were kindly supplied by Stella Pharma Corporation (Osaka, Japan). Cells were plated into 6-well plates at 3.0 × 10 5 cells/well and cultured with or without 5 μM DFO for 24 h in normal oxygen conditions at 37 • C. After 24 h of incubation, each sample was exposed to BPA at 30 μg 10 B/mL in the medium for up to 2 h. After BPA treatment, the culture medium containing BPA was removed and each sample was digested with perchloric acid (HClO 4 ) and hydrogen peroxide (H 2 O 2 ) for 2 h at 50 • C. To 500 μL of digested cell solution, 500 μL of standard yttrium solution was added and diluted with 4 mL of distilled water. 10

Neutron irradiation and clonogenic assay
Cells were cultured in normal oxygen or hypoxic conditions for 24 h and then treated with BPA for 2 h. Then, they were placed on an acrylic phantom, which was made of 2-cm-thick acrylic plates for cell irradiation and set as close as possible on the irradiation port, and exposed to neutron beams with 0.6 C as the proton charge using the acceleratorbased BNCT system at Southern Tohoku BNCT Research Center. The acrylic phantom is shown in Fig. 2. The dose components of the physical dose of neutron irradiation for BNCT are shown in Table 2. After neutron irradiation, cells were resuspended with trypsinization and plated onto 60-mm dishes at 100-5000 cells/dish depending on the results of preliminary experiments. After culturing for ∼10 days, the cells were fixed in 100% methanol and stained with Giemsa stain solution (Nacalai Tesque). Colonies composed of >50 cells were counted. The surviving fraction was determined by dividing the number of colonies from the irradiated culture with that of the non-irradiated control culture treated with BPA.

Statistical analysis
All experiments except the cytotoxicity of DFO and YC-1 were performed at least three times and results are expressed as means ± standard error. Experiments for the cytotoxicity of DFO and YC-1 were performed twice and results are expressed as means ± standard deviation. Statistical significance was estimated using the Mann-Whitney U-test, Student's t-test or Welch's t-test. The Statcel 3 addin (OMS Publishing, Saitama, Japan) for Microsoft Excel was used for the statistical analysis. Probability values of P < 0.05 were considered statistically significant.

Attenuation of the antitumor effects of BNCT under hypoxic conditions
In our previous study, hypoxic conditions with 1% O 2 have been reported to induce a decrease of LAT1 mRNA expression in glioblastoma cell lines [12]. Therefore, at first, we evaluated the impact of hypoxia with 1% O 2 on the toxicity of BNCT with 30 ppm 10 B-BPA. For cells cultured in normal and hypoxic conditions and treated with 30 ppm 10 B-BPA, the surviving fractions after neutron irradiation were 0.414 ± 0.049 and 0.600 ± 0.048 (P = 0.027) for T98G cells, 0.062 ± 0.011 and 0.107 ± 0.012 (P = 0.024) for HSC-3 cells and 0.036 ± 0.016 and 0.045 ± 0.025 (P = 0.396) for MCF-7 cells, respectively (Fig. 3). In agreement with previous reports, we confirmed that the antitumor effects of BNCT were attenuated in hypoxic glioblastoma and head and neck cancer cells with a potential indication for BNCT.

DFO as well as hypoxia attenuates the antitumor effects of BNCT
To elucidate the mechanism of attenuation of the effects of BNCT caused by hypoxia, DFO, which is a hypoxic-mimetic agent, was used in experiments. For cytotoxicity of DFO for 24 h, the viabilities of T98G and MCF-7 cells were maintained in up to 5 μM DFO ( Fig. 4A and C). In HSC-3 cells, the viability decreased according to the density of DFO (Fig. 4B). However, under treatment with 5 μM DFO, a significant increase in dead cells in HSC-3 cell lines was not confirmed for up to 48 h of incubation, and the absolute number of cells increased over time. Thus, it was considered that not much cell death, but cell growth suppression, was caused as the concentration of DFO increased. The simulated hypoxic environment induced by DFO was confirmed by enhanced expression of HIF-1α. In cells treated with 5 μM DFO, the expression of HIF-1α was enhanced compared with cells in the normal oxygen conditions (Fig. 4D). In the fluorescence imaging of hypoxic conditions in a viable cell using MAR, hypoxia was observed within a couple of hours in samples under hypoxia of 1% O 2 and treatment with 5 μM DFO. The fluorescence intensity of DFO-treated samples was higher than that of samples under normal oxygen and lower than that of hypoxia (Fig. 4E). Thus, we confirmed that 5 μM DFO induces a pseudo-hypoxic environment. For gene expression of LAT1, relative LAT1 mRNA expression under hypoxia was 0.637 ± 0.082 (P = 0.008) for T98G cells, 0.687 ± 0.061 (P = 0.012) for HSC-3 cells and 0.519 ± 0.083 (P = 0.004) for MCF-7 cells, respectively (Fig. 5A-C). It is widely known that GAPDH is activated under hypoxic conditions in some cell lines. To evaluate GAPDH expression in hypoxic conditions, the cDNA concentration of each sample after the reverse transcription reaction of extracted total RNA was normalized, and then RT-PCR was performed. GAPDH expression made little difference in normal oxygen and hypoxia (data not shown). Therefore, according to the previous report and our study, using GAPDH as a house-keeping gene under hypoxic conditions did not seem to affect the present results. DFO-treated cells showed reduced gene expression of LAT1 [0.748 ± 0.149 (P = 0.180) for T98G cells, 0.360 ± 0.014 (P = 0.0003) for HSC-3 cells and 0.551 ± 0.034 (P = 0.008) for MCF-7 cells, respectively, normalized to DFO non-treated cells (Fig. 5D-F)]. Boron uptake was continuously suppressed up to 2 h after BPA addition by treatment with 5 μM DFO (Fig. 5G). The amount of intracellular boron 2 h after administration of BPA was 0.799 ± 0.033 normalized to the value in DFO non-treated cells (P = 0.003). In addition, boron uptake was also suppressed in HSC-3 and MCF-7 cells treated with 5 μM DFO [0.903 ± 0.043 (P = 0.025) for HSC-3 and 0.940 ± 0.037 (P = 0.066) for MCF-7, respectively] (Fig. 5H).

Influence of HIF-1α accumulation on expression of LAT1
HIF-1α, which accumulates intracellularly in hypoxic conditions, is a key factor in the cellular hypoxic response. To clarify the relationship between HIF-1α and LAT1, HIF-1α was knocked down with siRNA. HIF-1α was successfully reduced in all cell lines to levels that were 20-40% of that in the siRNA control conditions (Fig. 6A-C). In cells treated with 5 μM DFO, LAT1 expression was restored in all cell lines treated with siRNA (P = 0.102 for T98G cells, P = 0.004 for HSC-3 cells, P = 0.030 for MCF-7 cells) (Fig. 6D-F), revealing that HIF-1α suppresses LAT1 expression in hypoxic tumor cells. LAT1 expression in T98G cells transfected with HIF-1α siRNA was not so improved compared with the samples treated with control siRNA. It was suggested that the pseudo-hypoxic effect of DFO in T98G cells may have been small and accumulation of HIF-1α in cells may have been lower than in HSC-3 cells. These results suggest that HIF-1α may be a therapeutic target for enhancing the antitumor effects of BNCT in tumors with hypoxic fractions.
The HIF inhibitor YC-1 sensitizes the antitumor effects of BNCT in hypoxic tumor cells YC-1 at 0.5 μM or lower showed no toxicity ( Fig. 7A and B). In T98G cells treated with 0.5 μM YC-1, HIF-1α protein expression after 24 h incubation under hypoxia decreased compared with YC-1 non-treated cells. In HSC-3 cells, HIF-1α protein expression of 0.5 μM YC-1treated samples slightly decreased (Fig. 7C). The surviving fraction after neutron irradiation in cells treated with BPA in the presence or absence of YC-1 is shown in Fig. 7D and E. Treatment with YC-1

DISCUSSION
To elucidate the mechanism of attenuation of the antitumor effects of BNCT in hypoxic conditions, we created a simulated hypoxic environment using DFO. DFO, which is clinically used as an iron chelator for treatment of patients with iron overload disease, inhibits the activity of prolyl hydroxylase, suppresses the transcriptional activity of HIF-1α and stabilizes HIF-1α [14]. The advantage of using DFO is that a hypoxic state can be created easily by simply adding DFO to the cell culture. In addition, construction of a pseudo-hypoxic condition using DFO in vitro has already been performed in many previous studies, and therefore, treating cultured cells with DFO for analysis of hypoxia in this study is appropriate [15,16]. On the other hand, the disadvantage of DFO is that the intracellular oxygen state induced by DFO is not known. Furthermore, the chelating effect of DFO and the hypoxia load in cultured cells may produce different effects on organelles. However, evaluation of the HIF-1α protein expression level showed a similarity between pseudo-hypoxic conditions induced by DFO and hypoxic conditions induced by reduced oxygen (Fig. 4D). In addition, from the fluorescence imaging of hypoxic conditions using MAR, it was found that we could evaluate visually the intracellular oxygen state induced by DFO (Fig. 4E). Furthermore, regarding the gene expression of LAT1, which is involved in BPA uptake, a decrease in LAT1 expression was confirmed following DFO administration compared to normal oxygen conditions (Fig. 5D-F). Therefore, administration of DFO appears to create hypoxia-like conditions. To clarify the relationship between HIF-1α accumulation in hypoxic cells and LAT1 expression, we evaluated the mRNA expression of HIF-1α and LAT1 after treatment with HIF-1α siRNA. In the pseudo-hypoxic condition using DFO, the gene expression of LAT1 increased in cells transfected with HIF-1α siRNA compared with the control (Fig. 6D-F). Therefore, the LAT1 expression level may recover by inhibiting HIF-1α expression. Our study showed for the first time that LAT1 expression is controlled by HIF-1α, the key factor in the cellular hypoxic response. Restoration of LAT1 expression in hypoxic cells may lead to increased boron uptake in cells and decreased cell survival after BNCT, resulting in improvement in therapeutic outcomes following BNCT. Introduction of siRNA is involved in the toxicity and the metabolism of the cell can thereby decrease, and it is suggested that BPA uptake may have been masked in both sicontrol-and siHIF-induced samples. Therefore, it was difficult to show the changes in boron concentration in HIF-1α-depleted cells.
Finally, we evaluated the possibility of sensitization of cells to the therapeutic effects of BNCT by using a HIF inhibitor in hypoxic conditions. It was confirmed that the gene expression of LAT1 recovered under HIF-1α knockdown conditions in all cells that we evaluated. However, in the results of the surviving fraction after neutron irradiation for hypoxic cells treated with BPA, a meaningful difference was not recognized between normal oxygen conditions and hypoxia in MCF-7 cells (Fig. 3). In this study, all cell lines were irradiated under the same neutron beam conditions. Therefore, it was suggested that the sensitivity of MCF-7 cells to BNCT may have been higher than that of the other cell lines depending on cell-specific relative biological effectiveness or BPA uptake. This result might have revealed that the impact of hypoxia on BPA uptake depends on the original sensitivity to BNCT. YC-1 inhibits platelet aggregation and is used pharmacologically [17,18]. The details of the mechanism of YC-1 are not clear but YC-1 suppresses the activity of HIF-1α in cancer cells [19], and in this study, the decrease of HIF1-α protein in YC-1-treated cells was confirmed (Fig. 7C). In addition, YC-1 has a radiosensitization effect on tumors with hypoxic fractions when used in combination with radiation therapy [20,21]. To evaluate the hypoxic area in head and neck cancer, tumor hypoxia imaging with 18 F-fluoromisonidazole ( 18 F-FMISO) PET is useful [22][23][24][25]. Positive findings on 18 F-FMISO PET also correlate with poor treatment outcomes following conventional radiotherapy and chemotherapy [26,27]. Therefore, to improve the therapeutic outcomes of these tumors with hypoxic fractions, enhancing the radiosensitivity and drug sensitivity of the hypoxic region of these tumors is necessary. In this study, addition of YC-1 to glioblastoma and oral squamous cell carcinoma cell lines enhanced the cell killing effect of BNCT in hypoxic cells (Fig. 7D and E). This indicates that administration of YC-1 may enhance the therapeutic response to BNCT in patients with cancer with hypoxic fractions that have responded poorly to radiotherapy and chemotherapy.
In this study, we showed that hypoxia suppressed LAT1 expression through accumulation of HIF-1α. Concomitant use of HIF inhibitors may improve treatment resistance to BNCT for cancers containing hypoxic regions.