Normal human dermal-derived fibroblast (NHDFs) and red fluorescent protein (RFP)-labeled human pancreatic cancer cell line BxPC3 used in the experiment were purchased from LONZA (Walkersville, MD) and Anti-Cancer Japan (Ibaraki, Japan), respectively. Dulbecco’s modified Eagle’s medium (DMEM) (Wako, Osaka, Japan) containing 10% fetal bovine serum (FBS) (Nichirei, Tokyo, Japan) was used to proliferate cells prior to construction of the tumor tissue model. The cells were cultivated at 37˚C, 5% carbon dioxide. Bovine plasma-derived fibronectin (FN) and porcine skin gelatin (G) were purchased from Sigma-Aldrich (St. Louis, MO) and Wako Pure Chemical Industries, Ltd. (Osaka, Japan), respectively. Transwell inserts with porous polyester bottom (pore size: 0.4 µm) for 12-well culture plate (12 mm diameter, 112 mm² area, cat. No. 3401) were purchased from CORNING Inc. (New York, NY). The boron drug ¹⁰BPA (boronophenylalanine; C₉H₁₂BNO₄, molecular weight 209.01, fructose complex) was kindly gifted from Interpharma Praha, a.s. (Komorany, Czechia). A solid state nuclear track detector CR-39, an optical plastic material with composition C₁₂H₁₈O₇ [6] [7], was purchased from Cokin (Tokyo, Japan).

Human three-dimensional tumor tissue was prepared by cell accumulation method [10] [11] [12] [13]. First, connective tissue-like structures were fabricated by three-dimensional lamination of NHDFs. As shown in Figure 1(a), NHDFs are cultured under the conditions mentioned above, then ECM-nano film (about 10 nm thick) was formed on each cell surface by coating cells with fibronectin and gelatin dissolved in Tris-HCl buffer according to previously published methods [10] [11] [12] [13]. The cells are seeded on the transwell inserts at a density of 27.2 × 10⁵ cells/insert (8 layers) and cultured under the conditions of 5% carbon dioxide at 37˚C for 12 to 24 hours. We regarded this connective tissue-like structure as an artificial human normal tissue model, termed as NHDF3D.

Next, RFP-labeled BxPC3 cells which have been cultured and proliferated were collected by trypsin treatment, washed, and uniformly seeded on the upper surface of NHDF3D at a density of 300 cells/mm², then further cultured for 24 hours under the above culture conditions. As shown in Figure 2(e) and Figure 2(f), BxPC3 cells proliferated on the surface of NHDF3D forming the groups with flat shape. We regarded this cancer-loaded NHDF3D as an artificial human tumor tissue model, termed as NHDF3D/BxPC3. In the present study, one case of NHDF3D and two cases of NHDF3D/BxPC3 were prepared by the methods mentioned above and were used for the experiments.

The boronophenylalanine (BPA) solution (3% w/v) was diluted to a concentration of 40 ppm with DMEM containing 10% FBS. This is referred to as a BPA treatment solution. After removing the culture solution of NHDF3D or NHDF3D/BxPC3, 750 μL of BPA treatment solution was added, and incubated for 2 hours (BPA exposure) under conditions of 5% carbon dioxide at 37˚C (Figure 1(b)). After that, the BPA treatment solution was removed and the tissues were washed 3 times with 0.01 M phosphate-buffered saline (PBS, pH 7.3). Subsequently, the tissues were fixed by 4% paraformaldehyde/0.1M phosphate buffer (pH 7.3) for 30 minutes at room temperature shading the light. After the fixation, the cellular nucleus was stained by 4’,6-diamidino-2-phenylindole (DAPI).

After the fixation, three holes were provided on the tissue using an 18 G injection needle in order to provide alignment marks. Then the top surface of NHDF3D/BxPC3 was observed by fluorescence microscope BZ-X700 (Keyence, Osaka, Japan). The distribution of BxPC3 cells was visualized as a fluorescence image by RFP excitation (Figure 1(c)). A low magnification image including the entire tissue and high magnification images of various parts of each tissue were respectively obtained. A fluorescence image of NHDF3D was also obtained as a control.

The above-mentioned NHDF3D or NHDF3D/BxPC3 in the transwell inserts were cut out with a knife together with the polyester base, mounted on the solid track detector CR-39 [6] [7] with close contact, and used as a sample for track image acquisition (Figure 1(d)). Irradiation experiments using these samples were conducted at the Heavy Water Neutron Irradiation Facility of Kyoto University Reactor (KUR), and irradiation was performed for 30 minutes under an irradiation flux of 1.4 × 10⁹ n/cm²/s (total flux = 2.25 × 10¹² n/cm²) (Figure 1(d)). After the neutron irradiation, the above sample was etched (6 N NaOH, 70˚C × 2 hours) to visualize the α-ray/recoiled Li particle tracks generated on the CR-39 surface.

The α-ray/recoiled Li particle track image of etched CR-39 was taken using the bright field function of the fluorescence microscope BZ-X700 (Figure 1(e)). A low magnification image including the entire tissue mount and 10 random high magnification images were obtained respectively, and the following analysis was performed.

1) The whole tissue images of fluorescent BxPC3 cell distribution and the α-ray/recoiled Li particle track distribution were compared referring to the position of three-hole markers, and the relationship of these distributions was observed.

2) Alpha-ray/recoiled Li particle tracks in the high magnification images were regarded as the particles, and quantitatively analyzed by using software FIJI (https://fiji.sc). Briefly, after binarizing the image, the tracks with more than 4 μm diameter were detected, and their number per unit area (0.01 mm²) and the track size were measured. Significant difference between the data from the artificial normal tissue model (NHDF3D) and artificial tumor tissue model (NHDF3D/BxCP3) was confirmed using student t-test.

For measurement of nuclear size, BxPC3 cells cultivated on the cell culture dish were fixed by 4% paraformaldehyde/0.1 M phosphate buffer (pH 7.3) for 30 minutes at room temperature shading the light. After the fixation, the cellular nuclei were stained by 4’,6-diamidino-2-phenylindole (DAPI). The images of the cells were obtained by fluorescence microscope as shown in Figures 2(a)-(d). The photo images of nucleus with DAPI fluorescence were used for analysis of their area size using software FIJI (https://fiji.sc), then their diameter was estimated regarding the shape like circle.

For analysis of BxPC3 cell size, the high magnification fluorescent images of NHDF3D/BxPC3, as shown in Figure 2(e) and Figure 2(f), were used. Briefly, the shapes of RFP-fluorescent BxPC3 cells were extracted from the images, then the area size and putative square size were analyzed as the methods mentioned above.

Low-magnification images of NHDF3D and NHDF3D/BxPC3 (two cases) detecting RFP fluorescence were shown in Figures 3(a)-(c). Although the BxPC3 cells were seeded over the artificial tissue in regular cell concentration, slight heterogeneity of cell distribution was observed after the cultivation for 24 hours. The parts with intense fluorescence denoting high density of BxPC3 cells were indicated by yellow circles in the figures. The α-ray/recoiled Li particle tracks in corresponded parts of Figures 3(a)-(c) were shown in Figures 3(a’)-(c’). Asterisks indicate the marking holes to match the position of the tissue images and the track images. The concentration of α-ray/recoiled Li particle tracks in NHDF3D/BxCP3 is higher than those of NHDF3D. Moreover, relatively high concentration of the α-ray/recoiled Li particle tracks was observed at the yellow circles in Figure 3(b) and Figure 3(c). These results suggested that the preferential incorporation of ¹⁰B into BxPC3 cells and their distribution on the tumor tissue model were reflected by the α-ray/recoiled Li particle track images.

Figure 4(b), Figure 4(b’), Figure 4(c), and Figure 4(c’) showed the higher magnification images of the white boxes on NHDF3D/BxPC3 in Figure 3. Figure 4(a) and Figure 4(a’) were the images of NHDF3D as control. In Figure 4(b)

and Figure 4(c) with RFP fluorescence, BxPC3 cells were observed as small fluorescent clusters. In the control image Figure 4(a), the fluorescent clusters were not observed. In Figures 4(a’)-(c’), the α-ray/recoiled Li particle tracks were observed as small dots, reflecting the intensity of the emission. Since the contrast between NHDF3D (a’) and NHDF3D/BxPC3 (b’ or c’) was obvious, the BxPC3 cells-dependent incorporation of ¹⁰B resulting the α-ray/recoiled Li particle emission was clearly detected in this system. Furthermore, as shown by the area surrounded with yellow and red hatched lines, the high density of BxPC3 cells on the tissue in Figure 4(b) and Figure 4(c) was comparative with the high intensity of the α-ray/recoiled Li particle emission in (b’) and (c’). From these results, it was suggested that ¹⁰BPA was abundantly taken in proportion to the tumor concentration locally at a high tumor concentration site.

To investigate detailed α-ray/recoiled Li particle track distribution and number of tracks in NHDF3D and NHDF3D/BxCP3 samples, high magnification images of α-ray/recoiled Li particle tracks on CR-39 (Figures 5(a)-(c), photo) were quantitatively analyzed. After the process of binarization with the track detection as the particles (Figures 5(a)-(c), binary), those with more than 4 µm diameter were selected, and their total number and sizes were measured (Figures 5(a)-(c), particle analysis). The results are shown in Figure 4(d) and Figure 4(e). The number of α-ray/recoiled Li particle tracks of NHDF3D/BxCP3 was about 1.6 times higher than that of NHDF3D (Figure 5(d)).

Next, we evaluated the hit number of α-ray/recoiled Li particle tracks per single cell on the top layer of NHDF3D and NHDF3D/BxCP3 as shown in Table 1. In the control tissue NHDF3D, the number of α-ray and recoiled Li particles tracks (H_B) was 51.52/0.01 mm² as the value of Figure 5(d). Whereas, H_B in NHDF3D/BxCP3 (1) and NHDF3D/BxCP3 (2) was 81.10 and 87.68/0.01 mm², respectively. The increased α-ray and recoiled Li particles tracks by seeded and grown BxPC3 cells (ΔH_B) were calculated by [H_B (NHDF3D) - H_B (NHDF3D/BxPC3)]. The cell number (H_cell) of NHDF in the top layer of NHDF3D was estimated as 30.09 cells/0.01 mm² according to the seeded cell number at the construction of tissue model. The H_cell of BxPC3 was estimated as 6.02 cells/0.01mm² in NHDF3D/BxCP3 (1) and (2) according to the seeded cell number and duplication at 24 hours after seeding that was confirmed in previous study (data not shown).

By using these data, we estimated tracks per single NHDF [(H_B/H_cell)_NHDF] in top layer of NHDF3D as 1.71, and that tracks par single BxPC3 cell [(H_B/ΔH_cell)_BxPC3] in NHDF3D/BxCP3 (1) and (2) as 4.91 and 6.01 (average: 5.46). Therefore the ratio was estimated as 2.87 [(H_B/ΔH_cell)_BxPC3(1)/(H_B/H_cell)_NHDF] and 3.51 [(H_B/ΔH_cell)_BxPC3(2)/(H_B/H_cell)_NHDF] respectively, and the averaged value was 3.19. These values corresponded with the tumor and normal tissue ratio (T/N ratio) of BPA as 2 - 4 that reported in the previous studies including animal

H_B: the number of α-ray and recoiled Li particle tracks per 0.01 mm²; H_cell: cell number per 0.01 mm².

experiments [16].

From these results, it was suggested that our in vitro method can reproduce the in vivo T/N ratio of BPA and available for evaluation system of pharmacokinetics.

In conclusion, our in vitro model of tumor tissue for BNCT demonstrated the pharmacokinetics of BPA and the efficacy of neutron irradiation by direct observation of α-ray/recoiled Li particle tracks that are corresponding to the distribution of BxPC3 cells. Moreover, the evaluated number of α-ray/recoiled Li particle tracks per single BxPC3 cell or NHDF provided the comparable value with T/N ratio of BPA in the previous studies. These results suggested that our in vitro model can be applied for a reproducible and high throughput method which enables to develop new boron-containing reagents from many drug candidates for BNCT without any animal experiments. Our model also can be used to evaluate the optimum conditions of BNCT such as dose of boron-containing drugs and the irradiated intensity (power and time) of neutron to the patients. Particularly, the examination using cancer cells obtained from each individual patient may enable to provide the safety and effectiveness for personalized BNCT.

As a limitation of this system, the detectable α-ray/recoiled Li particle tracks by CR-39 is restricted only those from the top layer of the artificial tissue, so that the other cancer cells such that rapidly infiltrate into the tissue cannot be applied. Three-dimensional detection of boron-containing drugs or irradiation damages distribution may extremely increase the value of this evaluation system as a tool for the development of BNCT and related medical studies.