l-Amino Acid Transporter-1 and Boronophenylalanine-Based Boron Neutron Capture Therapy of Human Brain Tumors

The system l-amino acid transporter-1 (LAT-1) imports p-boronophenylalanine (BPA) into cells and may play a major role in the effectiveness of BPA-based boron neutron capture therapy. The functional status of LAT-1 and its relationship to cell proliferation were simultaneously examined in the same section of human tumor material using a dual-labeling technique. The uptake of BPA (boron inductively coupled plasma mass spectrometry) was profiled in the presence of agonists and antagonists in fresh tumor explants. The number of LAT-1–expressing cells (mean ± SD) was three times higher than that of proliferating cell nuclear antigen (PCNA)–expressing cells (71.5 ± 17.02% versus 23.8 ± 16.5%; P < 0.0001; n = 38 glioblastoma and metastatic tumors). There was no correlation between PCNA cells and the number of LAT-1/PCNA double-stained cells, and not all PCNA-expressing cells coexpressed LAT-1. Boron uptake reached 30 ± 15 μg/g of wet weight of tissue by 4 hours both in tumor and brain around tumor tissue containing tumor cells compared with time 0 (P < 0.005; n = 4 glioblastoma tumors). This uptake was inhibited by both phenylalanine and 2-aminobicyclo-(2,2,1)-heptane-2-carboxylic acid. These LAT-1 data indicate that BPA-based boron neutron capture therapy might affect up to 70% of tumor cells, representing a three times higher proportion of tumor cells than their cell cycle status might suggest. Cells expressing PCNA, but not LAT-1, will require a different therapeutic strategy. [Cancer Res 2009;69(5):2126–32]

There is increasing interest in the use of boron neutron capture therapy (BNCT) as a tumor-selective treatment for malignant intracerebral tumors where these tumors currently remain incurable despite aggressive treatment with surgery, chemotherapy, and conventional radiotherapy (reviewed in ref. 1). The therapeutic potential of BNCT rests in the selective accumulation of a sufficient amount of ¹⁰B within the cancer cell. Partitioning of ¹⁰B within and outside the cell is also critical because the short spatial range (4–7 μm, roughly one cell diameter) of the α particles and lithium ions, resulting from the capture reaction, limits the ionizing effect to those cells that have taken up the boron (1). The success of BNCT depends on the preferential uptake of boron by the malignant cell. In the case of boronophenylalanine (BPA), a rational carrier for ¹⁰B and used in BNCT trials for the treatment of brain tumors (2–4), as well as currently under further pharmacokinetic investigation for clinical use (ref. 5, and our own clinical study),¹

selective accumulation is thought to be enabled by specific uptake mechanisms involving the system L family of heterodimeric, sodium-independent, amino acid transporters located at the plasma membrane, in particular LAT-1 (ref. 6; for a review of LAT transporters, see refs. 7, 8). LAT-1 has been associated with growth, proliferation, and tissue development and is highly expressed in malignant tumors and tumor cell lines, where it is thought to contribute to growth autonomy by increasing amino acid transport (9, 10). LAT-1 expression has not been studied to any great extent, in the context of BNCT, in human malignant brain tumors where it is likely to play a central role in trafficking BPA selectively to tumor cells. Moreover, expression profiles of LAT-1 in such tumors will provide essential information about the distribution of BPA in the tumor because the effectiveness of BNCT largely depends on the uniformity of distribution of the boron compound within the tumor mass.

The issue of cell cycle–specific uptake of BPA is particularly important. Several studies point to a direct relationship between the uptake of BPA and cell replication. In a preclinical study using a mouse melanoma model and comparing boron and [³H]thymidine (TdR; incorporated into cells at the S phase) uptake in double-labeled whole body–sectioned autoradiographs, it was concluded that the l-isomer of BPA accumulated selectively in the actively dividing (TdR-accumulating) cells of the tumor, although at this low resolution it was not possible to discern individual cells (11). Following up on this report, Ono and colleagues (12) also studied the microdistribution of boron compounds in peripheral (thigh) tumors using a squamous cell carcinoma mouse model, with attention particularly given to the quiescent cells, and concluded that the distribution of BPA was heterogeneous and that quiescent cells may not accumulate BPA. Yoshida and colleagues (13), in an in vitro study where G₀-G₁ and G₂-M phase tumor cells were sorted by flow cytometry after exposure to BPA and BSH, found higher levels of BPA in the G₂-M phase, concluding that BPA uptake was dependent on the cell cycle. Langen and colleagues (14), studying the single photon emission computed tomography (SPECT) tracer ¹²³I-α-methyl-l-tyrosine (IMT) in vitro, suggested that the Na⁺-insensitive, L system–mediated uptake of this compound was dependent on the proliferation rate of the human glioma (86HG-39) cells, whereas in a study of 27 glioma patients, Kuwert and colleagues (15) found a positive correlation between the in vivo IMT transport rate, as measured by SPECT, and the histologically determined Ki-67 proliferation index.

If the implication of these reports is correct, that BPA uptake by tumor cells is coupled to their proliferation status, then BPA-based BNCT in malignant tumors will have a very low efficacy because the only cells targetable will be the ones temporally in cycle in the tumor—approximately 10% (16, 17). Therefore, it is important to verify this relationship in the tumor in situ, in the original specimen, a paradigm believed to more closely resemble the real situation. To do this, it is hypothesized that all proliferating cells express functional LAT-1. Therefore, proliferating and LAT-1–expressing cells were simultaneously identified in the same section with a dual-stain immunohistologic technique, and LAT-1 function was probed by measuring the importation of BPA in the presence of LAT-1–specific substrates in biopsied brain tumor tissue slices.

The rabbit polyclonal LAT-1 antibody was provided by Serotec, whereas the biotinylated donkey anti-rabbit immunoglobulin was from Amersham Biosciences UK Limited. The mouse monoclonal proliferating cell nuclear antigen (PCNA; clone PC10), monoclonal and polyclonal glial fibrillary acidic protein (GFAP), goat anti-mouse immunoglobulins, mouse alkaline phosphatase anti–alkaline phosphatase, and StreptABComplex/alkaline phosphatase were supplied by DakoCytomation. The enzyme substrates naphthol AS-MX phosphate, levamisole, Fast Blue BB salt, and aminoethylcarbazole; the solvent, dimethylformamide; and the aqueous mountant, glycerol gelatin, were from Sigma-Aldrich.

Clinical grade BPA was a generous gift from Cancer Research UK and was formulated at 100 mg/mL in 110 mg/mL mannitol. All other chemicals, including 2-aminobicyclo-(2,2,1)-heptane-2-carboxylic acid (BCH), phenylalanine, tyrosine, and the vital stain dyes (ethidium bromide and acridine orange), were from Sigma-Aldrich.

Immunohistochemistry. Routine histologic sections of tumor tissue and nontumorous brain tissue from epileptic patients, as well as freshly excised tumor and brain around tumor tissue, were collected from the operating theatre in accordance with current hospital ethical guidelines. Approval with written consent was obtained from patients undergoing craniotomies. All tumor specimens were nonrecurrent from patients that were unirradiated and chemotherapeutically naive. Sections were stained for LAT-1 and PCNA using a three-layer immunostaining method and antibodies from different species to minimize the possibility of cross-reactivity. Sections of normal term placenta for verification of LAT-1 antibody specificity were obtained from Birmingham Women's Hospital in accordance with the same ethical guidelines. The PCNA immunostaining method (18) was modified for the dual-staining technique to simultaneously visualize LAT-1 and PCNA cells as follows. Briefly, the LAT-1 and PCNA primary antibodies (at 1:30 and 1:80 dilutions, respectively, in 0.3% NSS/T/T) were simultaneously applied without antigen retrieval and incubated in a humidified chamber overnight at 4°C followed by 30-min incubations at room temperature with the secondary [biotinylated antirabbit immunoglobulin (1:100) and goat anti-mouse immunoglobulins (1:50)] and tertiary (StreptABComplex and mouse alkaline phosphatase anti–alkaline phosphatase at 1:50 dilution) layers, with rinsing before application of each layer. Colors were developed separately, first with aminoethylcarbazole and then with Fast Blue BB salt, and sections were mounted in glycerol gelatin without counterstaining. Positive and negative tissue control sections for LAT-1 and PCNA (normal term placenta and nontumorous adult brain) were included in each assay in addition to the controls for each primary antibody (normal nonimmune serum of species donating antibody). Intra-assay controls were included to determine any cross-reactivity between the respective antibody sandwiches.

Stained cells were quantified in high-power fields (40× objective) using an eye-piece reticule. Counting was confined to informative areas where the staining was unambiguous. Usually several informative areas were scanned, and at least 1,000 stained cells counted for each specimen; unstained cells could not be counted because the sections were not counterstained. The ratios of the stained cell populations were calculated from the enumerated cells. A total of 44 specimens were analyzed, including 29 glioblastomas, 9 metastases, 4 meningiomas, and 1 each of nontumoral agliotic cortex and placenta.

In vitro BPA uptake experiments. Brain tumor and brain around tumor tissue were collected immediately after excision from four patients with glioblastoma undergoing craniotomies, with the diagnosis subsequently confirmed histologically, in ice-cold Ringer solution (125 mmol/L NaCl, 4.8 mmol/L KCl, 1.3 mmol/L CaCl₂, 1.2 mmol/L MgSO₄, 25 mmol/L HEPES-Tris, 1.2 mmol/L KH₂PO₄, and 5.6 mmol/L glucose). These were washed free of coagulated blood and cut into approximately 5 × 5 × 1 mm thick slices in the same buffer, washed thrice, and divided into groups for the experiment, the basic design of which was to monitor the temporal uptake of BPA in the presence and absence of system L competitors, inhibitors, and stimulators. Accordingly, four groups were set up: (a) BPA, (b) BPA plus BCH, (c) BPA plus phenylalanine, and (d) BPA plus tyrosine, with at least two slices per time point depending on tissue availability. Tissue slices were incubated in Ringer's solution containing equimolar amounts of amino acids with respect to the BPA, which was added at 10 mmol/L final concentration (110 μg/mL boron equivalent; ref. 19) at 37°C. All amino acids were prepared in the same format as BPA (i.e., in mannitol). Control slices received an equivalent amount of mannitol. At various time points (0, 1, 2, 4, and 8 h), uptake was terminated by removing the tissue slices from the incubation buffer and washing thrice in Na⁺-free Ringer's solution (sodium chloride was replaced by choline chloride) because other analyses were to be done on the material that required sodium-free conditions. The slices were then blotted dry, weighed, and processed for analysis in an inductively coupled plasma mass spectrometer (ICP-MS) to measure the ¹⁰B content. Briefly, analysis was done using an Agilent 7500c ICP-MS in normal (noncollision) mode. The instrument was equipped with a Babington nebulizer, quartz spray chamber, and nickel cones. Calibration used matrix-matched aqueous standards prepared from boron-10–enriched boric acid. Standards and specimens to be assayed were diluted with an aqueous solution containing 2% butanol, 0.05% EDTA, 1% NH₄OH, and 0.05% Triton X-100. Beryllium was used as the internal standard (20). Perished tissue (0% viability; see below) served as control for nonspecific uptake/entrapment of BPA. Boron values were corrected for nonspecific uptake and are expressed as micrograms of boron per gram of wet weight of tissue.

Tissue slice viability was monitored over the course of the experiment using a vital stain method described elsewhere (21). Briefly, imprints of the slices were exposed to the vital stain (containing 3.4 mg/mL each of ethidium bromide and acridine orange dyes), and the live (green) and dead (orange/red) cells counted immediately under a fluorescence microscope.

Statistical analysis. Relative differences in the stained cell populations were analyzed by Student-Newman-Keuls test using the GraphPad Instat program, whereas differences in BPA uptake were analyzed by unpaired t test. Significance was set at P < 0.05.

LAT-1 expression and profiles. Immunohistochemical LAT-1 and PCNA profiles of human brain tumors in situ were examined to verify the relationship between LAT-1 expression and cell proliferation. For this, archival brain tumor tissue was screened using immunohistology with a LAT-1 antibody. The specificity of the LAT-1 antibody was confirmed using human placental and cerebral cortical tissues. In the placenta, staining with this antibody was confined mostly to the apical surface of the syncytiotrophoblasts, consistent with previous reporting (22), whereas the cerebral cortical specimen showed no staining. Histologic sections were then subjected to the LAT-1/PCNA dual-staining procedure described above to concurrently visualize individual and dual-expressing LAT-1 and PCNA cells. The PCNA protein, which is present in the nucleoplasm of continually cycling cells throughout the cell cycle (23), was used as a marker of proliferating cells. Because the normal cortex specimen showed no staining for PCNA, it was screened for LAT-1 and GFAP immunoreactivity instead to verify the double-labeling technique.

The LAT-1/PCNA staining profiles of placenta and cortex are shown in Fig. 1A and B. In the placenta, the apical surface syncytiotrophoblasts were highly positive for LAT-1 expression (red stain) but were negative for PCNA immunoreactivity (blue stain), which was confined to the luminal trophoblasts; there was no staining of fibrous areas (Fig. 1A,, bottom right). In contrast, the specimen of normal cortex was negative for LAT-1 immunoreactivity, except for the occasional lymphocyte (Fig. 1B , middle left), but positive for GFAP (blue cells). These observations of LAT-1 immunoreactivity in placental syncytiotrophoblasts and of the inability to detect PCNA in the cortex were taken as confirmation of the specificity of the antibodies and also of the validity of the dual-staining technique.

In the tumors, at low magnification, LAT-1 and PCNA staining was seen to follow the distribution of viable areas of the specimen (Fig. 2A and B), analogous to the pattern of PCNA immunoreactivity described previously (18). Fibrous, necrotic, and hematogenous areas were negative for both LAT-1 and PCNA immunostaining except for lymphocytes, which were highly reactive for LAT-1, particularly in the metastatic tumors (Fig. 2B). The glioblastomas and the metastases were highly positive for both LAT-1 and PCNA expression, whereas the meningiomas were mostly positive for LAT-1 only, but all tumors showed LAT-1 expression.

At a higher magnification, three distinct types of cells could be distinguished: red cells (LAT-1⁺), blue cells (PCNA⁺), and purple/black or red and blue (LAT-1⁺PCNA⁺ dual-stained) cells, the latter showing clearly a blue nucleus surrounded by a red cytoplasm (Fig. 2C and D). All three types of cells were found interspersed and were readily distinguishable. The numbers of these stained cells were recorded in each specimen and their relative proportions are shown in Fig. 3A, to C and summarized in Table 1. Lymphocytes, which stained positive for LAT-1 (Fig. 2B), were excluded from analysis. No double-stained cells were found in the placenta or cortex. There was no significant difference in the proportion of LAT-1 cells between glioblastoma and the metastatic tumors (72.6% versus 71.2%), but there was a relatively higher proportion of LAT-1⁺ cells in the meningiomas (P < 0.01; Table 1). There was no difference between the proportion of PCNA cells in glioblastoma and metastatic tumors (22.8% versus 23.6%), but in contrast to LAT-1, the proportion of PCNA⁺ cells in meningiomas was significantly lower (23.8 ± 16.5% versus 2.8 ± 1.6%; P < 0.02). Of the three types of cells identified with the double-labeling procedure, the LAT-1⁺PCNA⁺ dual-stained cells were the least abundant, numbering only 1% of detected cells in the meningiomas and 4.8% and 5.1% in the glioblastoma and metastases, respectively (Table 1). Overall, there were three times as many LAT-1⁺ cells as PCNA⁺ cells in the malignant tumor groups (72.6% versus 22.8% and 71.2% versus 23.6% in the glioblastoma and metastases, respectively; P < 0.001).

One drawback of the dual-staining method is that because sections could not be counterstained without affecting the specific staining, then all residual cells could not be enumerated. However, this does not seem to be a major problem because compared with the PCNA indices reported by Detta and colleagues (18), the relative proportions of PCNA cells found in the present study are similar, suggesting that the proportion of unstained and uncountable cells was relatively small.

Association of LAT-1 expression with proliferation. If there was a direct relationship between cell proliferation and LAT-1 expression, a higher proportion of LAT-1⁺PCNA⁺ (proliferating LAT-1, double-labeled cells) would be expected with a correspondingly lower proportion of PCNA⁺ (proliferating-only, single-labeled) cells. To verify such a relationship, data related to the malignant tumor categories (glioblastoma and metastases) were pooled and PCNA⁺ values were plotted as a function of those with dual-staining LAT-1⁺ and PCNA⁺ cells falling above the lower 95% confidence interval (n = 30). The graph showed no significant correlation between the two (i.e., LAT-1 expression was not related to proliferation; Fig. 4). There was also no relationship between patient's age and LAT-1 or PCNA expression (data not shown).

BPA uptake (LAT-1 function). The functional integrity of the LAT-1 transporter was examined in malignant brain tumor specimens by measuring the importation of BPA in the presence of LAT-1–specific substrates in biopsied brain tumor tissue slices to preserve and maintain the original tissue integrity and viability. Uptake of BPA was measured in the presence and absence of the system L–specific agonist, phenylalanine; the antagonist, BCH (24); and the augmentor, tyrosine (25). Tissue slices were incubated with BPA 1 hour after exposure and removal of tyrosine (25), 10 minutes after BCH (without removal; refs. 24, 26), and simultaneously with phenylalanine for the time periods shown in Fig. 5, after which they were washed and analyzed for boron content. Tissue viability was monitored at the start and end of the experiment. The boron data did not require correction for any differences in viability because such changes were not significant (Fig. 5A  and B, topmost line). Viability scores [expressed as the ratio of live (green) to dead (red) cells ×100] ranged from 90.7 ± 4.2% to 87.3 ± 4.0% and from 89.3 ± 5.6% to 85.5 ± 7.5% for the tumor and brain around tumor groups at the start and end of the experiment, respectively.

The uptake profiles of boron (BPA) in tumor and brain around tumor tissue slices are shown in Fig. 5A and B. The accumulation of boron was curvilinear over the time course, reaching significantly elevated levels by 2 hours in both tumor and brain around tumor tissue (P = 0.0191 and P = 0.008, respectively, compared with time 0) and approximating a plateau level of 25 to 30 μg/g tissue by 4 hours in both tissue groups (P = 0.0015 and P = 0.005 in tumor and brain around tumor, respectively), after which levels tended to decrease. Values were corrected for nonspecific uptake/entrapment. The profile of BPA accumulation and inhibition was similar in both tissue groups and there was no significant difference between the peak accumulated BPA levels (Fig. 5A and B). Substrate selectivity of BPA uptake was assessed by measuring its accumulation in the presence of 10 mmol/L phenylalanine. The uptake was markedly and effectively inhibited by both phenylalanine and the antagonist, BCH, over the initial 2 hours of the time course in both tumor and brain around tumor tissue (P = 0.013, 4 hours BPA versus 2 hours phenylalanine or BCH) and tended to increase thereafter, whereas it was not affected by pretreatment with tyrosine.

The relationship between LAT-1 and PCNA expression and the functionality of the transporter, crucial to our understanding of BPA-based BNCT, was explored in human malignant brain tumors in situ. Our results indicate that the situation in vivo is complicated by the considerable heterogeneity of human brain tumors with microscopic areas of viable and nonviable cells (18), making direct comparisons with clonal animal tumor models difficult. Contiguous positive LAT-1 and PCNA staining was naturally restricted to viable tumor areas. In this study focusing on viable tumor areas, there was no correlation between the cellular expression of LAT-1 and PCNA. Additionally, examination of the significance of LAT-1 in primary explants of these tumors, which represent a closer model of the in vivo handling of BPA, showed that BPA uptake was specifically antagonized by pretreatment with phenylalanine or the specific inhibitor of LAT-1, BCH. Furthermore, the suppression of overall uptake of BPA by BCH indicates the apparently critical role of LAT-1 in determining the uptake of BPA. The system L transporter (LAT-1) mediates the transport of l-amino acids into cells and likely plays a major role in the uptake of BPA in BPA-based BNCT for malignant brain tumors.

LAT-1 expression. LAT-1 expression has been detected in several tissue types including the blood-brain barrier and glial cells (10, 27). The expression of LAT-1 has been shown to be markedly elevated in various tumor cells (10) and notably in glioma cells (28, 29). Up-regulation of LAT-1 thus seems to be a common feature of tumor cells and of interest in targeting treatments dependent on LAT-1 function, such as transported cytotoxic drugs (e.g., melphalan; ref. 26) or enhancement of BNCT by increased uptake of boron using BPA. Clinical evidence of the role of LAT-1 has been implied by Langen and colleagues and Kuwert and colleagues in studies (14, 15) that showed increased tracer (IMT) concentration in progressively more active tumors, suggesting a link between tracer uptake (l-amino acid handling) and proliferative activity and, at the same time, calling into question the use of the LAT-1/BPA system for BNCT treatment of malignant tumors where the proliferation rate in vivo has been shown to be low (16, 17). Whereas a relationship between L amino acid uptake and proliferation might be expected, the apparent correlation of IMT uptake with proliferative indices is explained by the occurrence of contiguous LAT-1 and PCNA-stained cells found in this study. Indeed, other lines of evidence suggest that the linkage between proliferation and LAT-1 activity is less clear. For example, De Wolde and colleagues (30), in an attempt to correlate the protein synthesis rate of human brain tumors with histologic parameters of proliferation (Ki-67 and AGNOR counts), found no correlation between l-tyrosine positron emission tomography and Ki-67 in human brain tumors. Similarly, Sasajima and colleagues (31), in exploring the relationship between proliferative activity, amino acid transport, and glucose metabolism in glioma cell lines, found an inverse relationship between the influx of a nonmetabolizable system L–specific amino acid (aminocylopentane carboxylic acid; also known as cycloleucine) and proliferation index and TdR accumulation, and a recent in vitro study on the localization of ¹⁰B (BPA) in metaphase and interphase T98G human glioblastoma cells using secondary ion mass spectrometry (SIMS) reported reduced levels of ¹⁰B in the cytoplasm of mitotic cells compared with interphase cells (32). Our data also seem to support a lack of an association between amino acid uptake and proliferative activity. More importantly, the new finding of more than 70% of viable tumor cells expressing LAT-1 regardless of PCNA expression (cell cycle status) may offer considerable benefit to BPA uptake in these cells, whether they are localized within the tumor mass or dispersed in the brain around tumor. Accordingly, functional imaging biomarkers such as IMT or [methyl-¹¹C]methionine (amino acids) may be superior in delineating the radiation therapy target volume than proliferation-based markers, such as bromodeoxyuridine, or nonspecific metabolic markers such as [¹⁸F]fluorodeoxyglucose. It is also important to note that a number of PCNA-positive (proliferating) tumor cells (up to 25%) do not express detectable LAT-1 and, hence, are unlikely to accumulate BPA and will require a different therapeutic approach than BNCT with this agent.

LAT-1 and BPA uptake. Based on BPA import and efflux measurements in the presence of system L– and system A–specific substrates in the rat 9L gliosarcoma in vitro model, Wittig and colleagues (6) concluded that BPA was transported by the L system, a mechanism that is likely to play a major role in BPA-based BNCT. The capacity of our explant system to take up BPA has not been evaluated in this study, but the peak levels of boron achieved, 30 μg/g, are in the range estimated to enable the therapeutic advantage of BNCT (33, 34) despite the heterogeneity of the explants used. This still clinically relevant level of boron is not unexpectedly lower than that achieved in the more homogeneous rat 9L gliosarcoma in vitro model (6). Interestingly, in the clinical samples used here, the similarity in BPA uptake profiles of tumor and brain around tumor may be a reflection of higher LAT-1 expression by the infiltrating neoplastic cells in the brain around tumor (35), although the number of neoplastic cells in this region, by definition, will be lower than their core mass counterparts. To confirm this requires quantitative assessment of LAT cell numbers and their corresponding expression levels correlated with BPA uptake, which is beyond the scope of this study. Importantly, the uptake profiles of tumor and brain around tumor indicate the presence of LAT-1 tumor cells dispersed in brain around tumor tissue and that they are accessible to available BPA through LAT-1 activity.

Implications for BNCT. This study does not consider the issue of bioavailability of BPA for access to LAT-1 for its uptake into tumor cells. The uptake profiles were, however, measured in the presence of excess BPA and showed a delayed time course to peak ∼4 hours, which may be relevant to in vivo studies of BPA uptake and sampling. It remains to be seen whether prolonged infusions, elevated BPA concentrations, or blood-brain barrier manipulation will enhance bioavailability in vivo to the extent of increasing tumor boron concentrations through LAT-1 transporter activity. These data suggest that a much larger proportion of the tumor, regardless of cell cycle status, will be available for BNCT, with BPA uptake being mediated by LAT-1, and that consideration of the expression of LAT-1 and its distribution may be needed to interpret the variability in clinical response to BNCT using BPA. Furthermore, our study has identified a proportion of tumor cells that are cycle dependent but will escape the neutron reaction by failing to accumulate available BPA due to the absence of the transporter LAT-1. These cells will require a different treatment strategy, such as cycle-sensitive conventional radiation or chemotherapy.

In summary, LAT-1 seems to be more widely expressed in malignant brain tumors than expected and is unrelated to the lower number of PCNA-expressing cells. BPA-based BNCT might be expected to affect an average 70% of detected tumor cells, which is three times the proportion suggested by their cell cycle status. In addition, not all PCNA-expressing cells coexpressed LAT-1. The overall uptake of BPA could be specifically inhibited by LAT 1-dependent substrates and known inhibitors, indicating the essential role of this transporter for BPA uptake in these tumors. Our data also underline the issues of heterogeneity in these tumors and the value of LAT-1 determination in planning BPA-based BNCT studies.

No potential conflicts of interest were disclosed.

Grant support: Department's Brain Tumour Trust Fund.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

We thank the Division of Pathology for provision of histologic sections of cerebral tissue and definitive diagnoses, the Birmingham Maternity Hospital for placental tissue sections and the Department of Medicine for laboratory facilities; M. Khan of the Department of Immunology for assistance with the dual-staining technique, T. Sheehan of the Regional Laboratory of Toxicology for the boron measurements, Cancer Research UK for the provision of BPA, and Z. Ghani for help with the graphs; and Professor J. Hopewell of Oxford University for critically reading the manuscript.