Engineered αvβ6-targeting peptide‒BSH conjugates for boron neutron capture therapy of pancreatic ductal adenocarcinoma

Abstract

Pancreatic ductal adenocarcinoma (PDAC) is a highly aggressive malignancy characterized by a dense stroma that impedes drug penetration and establishes an immune-“cold” microenvironment. Conventional external beam radiotherapy is often limited by off-target toxicity to surrounding normal tissues due to the complex anatomy of the pancreas. Boron neutron capture therapy (BNCT) is a binary targeted radiotherapy that utilizes a focused thermal neutron beam to trigger the in situ production of high-energy particles, enabling single-cell-scale tumor cell ablation. However, the efficacy of BNCT in PDAC critically depends on the ability of B-10 carriers to effectively penetrate the dense stromal barrier and achieve efficient internalization into tumor cells. Herein, we developed nine αvβ6 integrin-targeting peptide‒BSH conjugates to optimize the BNCT therapeutic index. By optimizing linker structure and the number of BSH moieties, amphiphilic agents with enhanced PDAC cellular uptake were obtained. The lead compound, namely DOTA-2BB, adopts a “trident-shaped” structure that facilitates stromal penetration and exhibits the optimal tumor uptake and retention. In a PDAC mouse model, administration of DOTA-2BB followed by thermal neutron irradiation resulted in suppression of tumor growth with favorable biosafety. These findings establish DOTA-2BB as a promising B-10 carrier for PDAC and provide a strategy for treating stroma-rich tumors.

Graphical abstract

This study reports a trident-shaped peptide–BSH conjugate (DOTA-2BB) that effectively penetrates the dense pancreatic ductal adenocarcinoma (PDAC) stroma to achieve successful boron neutron capture therapy. Created with Biorender.com.

1. Introduction

Pancreatic ductal adenocarcinoma (PDAC) is the most common and aggressive subtype of pancreatic cancer. PDAC ranks among the leading causes of cancer-related deaths in China, with a five-year survival rate of less than 10%¹. Only 15%‒20% of patients are eligible for surgery, and the vast majority, including those with surgical eligibility, receive systemic chemotherapy². However, the dense stromal structure and complex tumor microenvironment (TME) of PDAC severely restrict drug penetration, thereby limiting the efficacy of chemotherapy³. Recently, hydroxychloroquine (HCQ) has been reported to improve intratumoral drug penetration by reducing extracellular matrix components, such as collagen and hyaluronan, through its antifibrotic and autophagy-inhibitory effects⁴. These findings suggest that the rational development of molecules with enhanced tissue-penetrating properties represents a promising strategy for improving drug delivery in desmoplastic tumors. Radiotherapy serves as an adjunct treatment in PDAC but is comparatively less effective than in more radiosensitive cancers such as nasopharyngeal carcinoma or lung carcinoma, where radical radiotherapy is often achievable^5,6. The deep anatomical location of pancreatic tumors and their proximity to radiosensitive organs limit the dose escalation possible with external beam radiotherapy, as doing so may increase the risk of treatment-related toxicity⁷. Moreover, PDAC exhibits only moderate intrinsic radiosensitivity⁸. PDAC is also characterized by early and rapid metastatic progression, often accompanied by nonspecific clinical symptoms—both of which contribute to delayed diagnosis and poor prognosis^9,10. Therefore, there is an urgent clinical need to develop more effective, targeted treatment strategies for PDAC¹¹.

Boron neutron capture therapy (BNCT) is an advanced binary radiotherapy that utilizes nuclear capture reactions, and its clinical efficacy has been demonstrated in various advanced and recurrent cancers^12,13. The therapeutic selectivity of BNCT is attributed to the short path length of the high linear energy transfer (LET) particles emitted following neutron capture by boron-10 (B-10), which is approximately equivalent to the diameter of a single cell (5‒9 μm). Consequently, tumor-targeted BNCT agents enable precise destruction of malignant cells while sparing surrounding healthy tissues. Recently, a variety of tumor-targeted BNCT agents have been developed, and multiple targeting moieties, including small molecules, nanoparticles, peptides and others, have been explored14, 15, 16. For example, Ma et al. ¹⁷ developed a series of erlotinib-based aggregation-induced emission (AIE)-active BNCT agents. Among them, NapE4 achieved a boron concentration of 217.90 ± 5.08 ng/10⁶ cells in A549 tumor-bearing mice at 48 h post-injection (p.i.). Li et al.¹⁸ developed a carboranyl-phosphatidylcholine-based liposome for combined BNCT and chemotherapy, which achieved a boron concentration of 98.9 ppm at 8 h after administration. Notably, peptides are considered particularly promising due to their low molecular weight and favorable pharmacokinetic properties, which enable high boron payload delivery without compromising tumor penetration or target affinity¹⁹. Mishiro et al. ²⁰ demonstrated that cyclic RGD peptides targeting integrins can serve as effective BNCT ligands, achieving over 60% internalization efficiency. Similarly, Nakase et al.²¹ reported an Fc receptor-targeting peptide–dodecaborate conjugate, which is internalized via macropinocytosis, resulting in enhanced boron delivery and improved BNCT efficacy in tumor-bearing mice.

To develop an effective boron carrier to enable BNCT for PDAC, we aim to design novel peptide-based B-10 carriers with enhanced targeting and tissue-penetrating properties. To this end, a hydrophilic peptide targeting integrin αvβ6 was selected as the targeting ligand²², given that integrin αvβ6 is overexpressed in PDAC but minimally expressed in normal pancreatic tissue, making it a highly promising drug target for this disease. Moreover, payloads conjugated to αvβ6-binding peptides can undergo clathrin-mediated endocytosis, thereby promoting continuous internalization and intracellular accumulation²³.

For B-10 delivery, BSH-derived icosahedral carborane clusters were employed as the boron source, as they provide a higher boron payload per molecule than 4-borono-l-phenylalanine (BPA)²⁴. When hydrophobic icosahedral carborane clusters are conjugated to hydrophilic peptides, amphiphilic peptide–BSH conjugates are formed, and this amphiphilic structure is expected to facilitate membrane penetration and tumor tissue infiltration²⁵. This is consistent with the mechanistic evidence that cell-penetrating peptides can directly translocate across pure lipid bilayers via peptide‒lipid interactions in a non-endocytotic, non-pore-forming manner, and retain their membrane-translocating capability even when coupled with large hydrophilic molecular moieties²⁶. Based on this design principle, we systematically optimized the molecular architecture and investigate the effects of both the peptide–BSH linker and the number of BSH moieties on binding affinity and cellular uptake.

In this study, we developed a series of multifunctional B-10 carriers by conjugating αvβ6-targeting peptides, BSH moieties and imaging modules, including 1,4,7,10-tetraazacyclododecane-1,4,7,10-tetraacetic acid (DOTA) and sulfo-cyanine5 (Cy5), which demonstrated strong target affinity and cell-penetrating capacity. For PDAC imaging, Cy5-2BB enabled high-contrast visualization of both subcutaneous and orthotopic tumors at 48 h after intravenous injection (i.v.). More importantly, these molecules overcame the major challenge of clinical boron delivery by significantly prolonging tumor retention time and enhancing boron accumulation, reaching up to 103.3 μg/g in tumors. Consequently, a single administration of DOTA-2BB achieved a superior BNCT therapeutic outcome compared with BSH. Overall, this study provides a new strategy for the development of next-generation BNCT agents with high targeting efficiency and enhanced membrane permeability, underscoring the potential of BNCT for the treatment of PDAC.

2. Materials and methods

2.1. Reagents and instruments

The reagents involved in solid-phase peptide synthesis (SPPS) and chelator conjugation were purchased from Sigma‒Aldrich (St. Louis, MO, USA), or Wako Pure Chemical Industries (Osaka, Japan). All reagents for cell culture were purchased from Gibco (Thermo Fisher Scientific, Waltham, MA, USA). Liquid chromatography‒mass spectrometry (LC‒MS) was performed on a Waters SQ Detector 2 mass spectrometer coupled to an acquity ultra-performance liquid chromatography (UPLC) system (Waters, MA, USA). Semi-preparative high-performance liquid chromatography (HPLC) was performed using a Shimadzu LC-16P system (Shimadzu Corporation, Kyoto, Japan), and analytical HPLC was performed using a Shimadzu LC-20AT system (Shimadzu Corporation, Kyoto, Japan). Flow cytometry analysis was performed using an Agilent NovoCyte Penteon Flow Cytometer System (Agilent, Santa Clara, CA, USA). Confocal microscopy was conducted using an BZ-X810 fluorescence microscopy imaging system (Keyence, Osaka, Japan). Gallium-68 was generated by a ⁶⁸Ge/⁶⁸Ga generator (Galliator, New Radiomedicine Technology, Sichuan, China). Radio-HPLC was performed using a Shimadzu LC-2050 system (Shimadzu Corporation, Kyoto, Japan) equipped with a Flow-Count PRO HPLC Remote Detection Unit (Eckert&Ziegler, Berlin, Germany). MicroPET/CT imaging was conducted using IRIS imaging systems (Inviscan SAS, Strasbourg, France). A 2480 Wizard² Automatic Gamma Counter (Revvity, Waltham, MA, USA) was used to measure the radioactivity (counts per minute, CPM) for in vitro and in vivo studies. Fluorescence and luminescence imaging were performed on a multi-modal small animal live imaging system (LASER6000, Guangzhou Guangyi Biotechnology Co., Ltd., Guangzhou, China).

2.2. Cells and animals

The PDAC cell line BxPC-3 (RRID: CVCL_0186) and its luciferase-expressing derivative, BxPC-3 luc (RRID: CVCL_J246), were obtained from the American Type Culture Collection (ATCC). Cells were cultured in Roswell Park Memorial Institute (RPMI) 1640 medium supplemented with 15% fetal bovine serum (FBS, TransSerum FQ Fetal Bovine Serum, China), penicillin (100 U/mL, Gibco), and streptomycin (100 μg/mL, Gibco). Cells were cultured in a humidified environment containing 37 °C and 5% CO₂. To ensure the authenticity and quality of the cell lines, routine monthly mycoplasma testing was performed using a GMyc-PCR Mycoplasma Test Kit (G7312070, Yeasen Biotechnology Shanghai Co., Ltd., China).

BALB/c nude mice (female, 18‒22 g) were provided by the GemPharmatech Co., Ltd. (Beijing, China). The experimental procedures followed the Ethical Norms for Laboratory Animals of the Chinese Academy of Medical Sciences and Peking Union Medical College (No. 00000775 and No. IMM-S-25-0081). Mice were anesthetized with 2% isoflurane. For the subcutaneous tumor model, mice were injected with 2 × 10⁶ BxPC-3 or BxPC-3 luc cells/mouse. In vivo evaluations were performed on mice with tumor volumes of approximately 200 mm³. For the pancreatic orthotopic tumor model, BxPC-3 luc tumor fragments (1‒2 mm³) were surgically implanted into the pancreatic tail of anesthetized BALB/c nude mice via laparotomy. The tumor was secured with absorbable sutures, followed by abdominal closure. Postoperative analgesia was provided for 72 h. Tumor growth was monitored weekly by bioluminescence.

2.3. Synthesis of carborane analogues

Carborane analogues (2C-BSH, 4C-BSH, and 6C-BSH) were synthesized via a two-step protocol²⁷, with 2C-BSH used as an example. Briefly, (11Z)-1,2,3,4,5,6,7,8,9,10-decaboracyclododeca-1,3,5,7,9,11-hexaene (2.0 g, 1.0 equiv.) was treated with n-butyllithium (1.0 equiv., 1.6 mol/L) in anhydrous tetrahydrofuran (THF) at 0 °C for 30 min. Sulfur (1.0 equiv.) was added, after 30 min, the temperature was raised to 25 °C. After another 30 min, the mixture was concentrated in vacuo and extracted with hexane. Without further purification, the crude product was stirred in a solution of hexane/triethylamine (20 mL, 3:1 v/v) for 15 min. The intermediate product was collected by vacuum filtration. In the second step, the intermediate was subsequently reacted with various methyl bromoalkylcarboxylates. The reaction proceeded as follows: methyl 2-bromoacetate (1.0 equiv.) was reacted with the thiolate intermediate in ethanol under an oil-bath reflux for 3 h, followed by acid-mediated hydrolysis (15 mL glacial acetic acid, 15 min; 1:1 H₂SO₄/H₂O, 3 h). The crude product was extracted with hexane, subjected to vacuum distillation, and purified through repeated hexane washes to yield 2C-BSH. Analogues 4C-BSH and 6C-BSH were synthesized similarly by substituting methyl 2-bromoacetate with methyl 4-bromobutyrate or methyl 6-bromohexanoate, respectively. All compounds were characterized by LC‒MS and ¹H-NMR.

2.4. Synthesis of targeting B-10 carriers

The commonly used Fmoc-based SPPS method was applied to synthesize targeting B-10 carriers. The peptide sequence was derived from a known αvβ6-targeting peptide (GNGVPNLRGDLQVLGQRVGRT), which was previously validated for diagnosis and therapy of PDAC. The modified sequence of peptide is KGNGVPNLRGDLQVLGQRVGRT (KKGNGVPNLRGDLQVLGQRVGRT for DOTA/Cy5-2BB)²⁸. For DOTA/Cy5-P4, DOTA/Cy5-PP4, DOTA/Cy5-P12 and DOTA/Cy5-PP12, the building blocks of Fmoc-PEG_4/12-COOH were conjugated to the N-/C-terminus of peptide chain during the process of SPPS. DOTA was conjugated to the N-terminus of the peptide chain. The products were purified by HPLC and conjugated with carborane to yield target molecules. Taking DOTA-2B as an example, a solution of 2C-BSH (2.0 equiv.), N-hydroxysuccinimide (NHS, 3.0 equiv.), and N-(3-dimethylaminopropyl)-N′-ethylcarbodiimide (EDCI, 3.0 equiv.) in anhydrous N,N-dimethylformamide (DMF) was stirred at room temperature for 30 min to activate the carboxyl group. The peptide-DOTA conjugate (1.0 equiv.) and N,N-diisopropylethylamine (DIPEA, 1.5 equiv.) were then added, and the reaction mixture was stirred for an additional 45 min. The crude product was purified by HPLC to yield DOTA-2B. The final product was purified to >95% by HPLC. The Cy5-conjugate was synthesized with a similar route, only replacing the DOTA with Cy5.

2.5. Flow cytometry assay

The binding ability of Cy5-labeled B-10 carriers to BxPC-3 cells was confirmed by flow cytometry. The BxPC-3 cells were treated with 2 μmol/L Cy5-labeled B-10 carriers at 37 °C for 30 min. The cells were subsequently analyzed by flow cytometry. To further obtain the dissociation constant (K_D) value, a range of concentrations (5-fold serial dilution, from 25 μmol/L to 64 pmol/L) of Cy5-labeled B-10 carriers were incubated with BxPC-3 cells for 30 min. The mean fluorescence efficiency (MFE) and mean fluorescence intensity (MFI) were subsequently analyzed by flow cytometry and K_D values were obtained by fitting using GraphPad Prism 10.4.0 based on the results.

2.6. Cell uptake assay

BxPC-3 cells were treated with 2 μmol/L Cy5-labeled B-10 carriers at 37 °C for 30 min. The cells were then rinsed with PBS three times. After fixation with 4% paraformaldehyde for 15 min, DAPI (S9980, Selleck, Houston, TX, USA) was used to stain the nucleus. Images were then captured using a confocal laser-scanning microscope (Keyence, Osaka, Japan). Quantitative analysis of experimental data was performed using ImageJ, and all statistical graphs were generated with GraphPad Prism 10.4.0.

2.7. Tumor penetration assay

A subcutaneous xenograft model of BxPC-3 was established and used for the tumor penetration assay. Briefly, Cy5-2BB molecules were intravenously injected into the mice (15 mg/kg, 100 μL). At 48 h p.i., the mice were euthanized humanely, and the intact tumor tissues were carefully dissected. The harvested tumor tissues were fixed with paraformaldehyde, followed by cryosectioning into 10 μm slices. Subsequently, the tumor sections were imaged using a confocal laser-scanning microscope.

2.8. ⁶⁸Ga labeling

The ⁶⁸Ga was produced using a ⁶⁸Ge/⁶⁸Ga generator. The generator was eluted with 5 mL 0.1 mol/L HCl to collect ⁶⁸GaCl₃. The eluate was then mixed with 475 μL of 1.25 mol/L NaOAc solution containing 20 μg of precursors, and the pH was adjusted to 4.0. The mixture was incubated at 95 °C for 15 min. After incubation, the products were purified by a Sep-Pak C18 cartridge to collect the radiolabeled peptides. The crude products were filtered through a 0.22 μm polyvinylidene fluoride (PVDF) membrane filter. The radiochemical purity and radiolabeling stability of ⁶⁸Ga-labeled B-10 carriers were analyzed using radio-HPLC.

2.9. Lipid‒water partition coefficient

The lipid‒water partition coefficient (LogP) of each ⁶⁸Ga-labeled B-10 carrier was determined using the shake-flask method. Briefly, 1 mL of water and 1 mL of n-octanol were mutually pre-saturated by vortexing for 30 min and allowed to separate overnight. Approximately 100 μL (∼1.0 MBq) of the ⁶⁸Ga-labeled compound was added to the mixture in a 2 mL microcentrifuge tube. The mixture was vigorously vortexed for 3 min at room temperature to facilitate partitioning and then centrifuged at 5000×g for 10 min to achieve complete phase separation. Aliquots of 100 μL were carefully taken from both the water and organic phases and transferred to separate tubes. The radioactivity in each sample was measured using a 2480 Wizard² Automatic Gamma Counter. The LogP value was calculated as the logarithm (base 10) of the ratio of CPM in the organic phase to that in the water phase. Measurements were performed in triplicate (n = 3) for each compound.

2.10. PET/CT imaging assay

PET/CT imaging was performed when the BxPC-3 tumors reached 200 mm³. ⁶⁸Ga-labeled B-10 carriers (100 μCi in 100 μL saline) were injected into each mouse i.v. For the blocking assay, each mouse was pre-injected with 100 μg of 5G peptide 30 min prior to the administration of ⁶⁸Ga-labeled B-10 carriers (100 μCi in 100 μL normal saline). During the scan, all mice were anesthetized with 2% (v/v) isoflurane and kept in a prone position. PET/CT scans were performed at 1, 2 and 4 h p.i. Imaging analysis was performed at the IMALYTICS Preclinical 3.1. The uptake levels of radiotracers in tumors and organs of interest are presented as percentage of the injected dose per milliliter (%ID/cc).

2.11. In vivo small animal fluorescence and bioluminescence imaging

Fluorescence and luminescence images of whole bodies of mice and isolated organs were obtained using the multi-modal small animal live imaging system. For fluorescence imaging, after being injected with Cy5-labeled B-10 carriers (15 mg/kg, i.v.), mice underwent the whole-body fluorescence imaging and then were sacrificed to isolate organs for the ex vivo fluorescence imaging (n = 3). Excitation and emission wavelengths were set at 630–660 nm and 680–710 nm, respectively. Besides, luciferase-expressing tumor-bearing mice were injected with 100 μL of d-luciferin (i.v.) and immediately anesthetized for imaging. Luminescence and fluorescence imaging were performed at the multi-modal small animal live imaging system (LASER6000). The analysis of images was performed using OI-IMAGE software.

2.12. Biodistribution and quantification of boron

Boron concentrations of the organs were quantified through inductively coupled plasma-mass spectrometry (ICP-MS, Agilent 7800, USA). 4, 12, 24, and 48 h after injection of DOTA-2BB (50 mg/kg), the mice were sacrificed to isolate organs. Organ samples were digested in 12 mol/L HNO₃ at 95 °C for 2 h. Subsequently, digestion was conducted by MARS 6 Microwave Digestion System (CEM, NC, USA) at 180 °C for 15 min under 600 psi pressure, followed by dilution to 10 mL with ultrapure water for ICP-MS analysis.

2.13. Neutron irradiation on tumor cells and tumor-bearing mice

All cell and animal irradiation experiments were performed at BNCT Center, Dongguan People's Hospital using the accelerator-based CASBNCT-50kW-01 BNCT system, produced by Guoke Zhongzi Medical Technology Co., Ltd. (CASBNCT). For in vitro studies, the BxPC-3 cells were seeded in 6-well plates at 5 × 10⁵ cells/well, incubated with DOTA-2BB or BSH at 250 μg/mL for 24 h, washed with PBS after drug removal, detached via trypsinization, and irradiated at 7 × 10⁸ n/(cm²·s) for 30 min. For in vivo therapeutic assessment, the BxPC-3 tumor-bearing mice received either: (1) intratumoral (i.t.) administration of DOTA-2BB (50 mg/kg, 100 μL) 1 h pre-irradiation, (2) i.v. administration of DOTA-2BB (50 mg/kg, 100 μL) 24 h pre-irradiation, (3) i.t. administration of natural boron BSH (5 mg/kg, 100 μL) 1 h pre-irradiation, or (4) i.t. administration of saline (100 μL) 1 h pre-irradiation (n = 5), followed by uniform neutron exposure [7 × 10⁸ n/(cm²·s), 30 min of irradiation time]. Tumor volumes were calculated daily for 14 days using the formula (length × width²)/2, with body weight monitored gravimetrically; humane endpoints were strictly enforced at tumor volume ≥2000 mm³ or >20% body weight loss.

2.14. Blood routine and biochemical test

Thermal neutron-irradiated mice were monitored until the therapeutic endpoint. Blood routine tests and serum biochemical analyses were subsequently performed at designated time points using standard clinical laboratory procedures.

2.15. Transcriptomic analysis

At the BNCT endpoint (day 14), subcutaneous BxPC-3 tumors were harvested from the following groups: DOTA-2BB-i.t. + N, BSH-i.t. + N, and neutron (n = 3). Tumor tissues were snap-frozen in liquid nitrogen and stored at −80 °C until RNA extraction. RNA sequencing was performed by Shanghai Majorbio Bio-pharm Technology Co., Ltd. Clean reads were aligned to the human genome using HISAT2 v2.2.1 and gene expression levels were quantified using featureCounts v2.0.1. Genes with low expression (mean count <10) were excluded. Differential expression analysis was performed using DESeq2 v1.30.1 in R v4.1.0, with significance thresholds set at adjusted P < 0.05 and |log₂ fold change| > 1. These genes were subsequently analyzed for Gene Ontology (GO) enrichment using Metascape v3.5, applying thresholds of P < 0.01, minimum overlap of 3 genes, and an enrichment factor >1.5. Redundant GO terms were consolidated through semantic similarity clustering with a cutoff of 0.7. Measurements were performed in triplicate (n = 3) for each group.

2.16. Statistical analysis

Data were presented as mean ± standard error of the mean (SEM), with sample size (n ≥ 3) specified in the figure legends. Statistical analyses were performed using GraphPad Prism 10.4.0, applying two-tailed unpaired Student's t-test for two-group comparisons. The specific statistical tests and corresponding P values were provided in the figure legends. Statistical significance was indicated in the figures as ∗P < 0.05, ∗∗P < 0.01, ∗∗∗P < 0.001, and ∗∗∗∗P < 0.0001; ^#P < 0.05, ^##P < 0.01, ^###P < 0.001, and ^####P < 0.0001.

3. Results and discussion

3.1. Design and synthesis of targeting B-10 carriers

To overcome the challenge of achieving efficient and specific delivery of B-10 carriers in PDAC, we designed and synthesized a series of peptide-guided theranostic boron agents. All molecules comprise a tumor-targeting peptide, a DOTA chelator (or Cy5) and an/two icosahedral carborane cluster(s). Among them, the αvβ6-targeting peptide warhead enabled precise tumor cell targeting. The targeting peptide 5G was developed by Sutcliffe et al. based on A20FMDV2, a peptide derived from foot-and-mouth disease virus that was identified as a potent and highly selective inhibitor of integrin αvβ6²⁹. To improve hydrophilicity and pharmacokinetic properties while preserving binding affinity, non-critical amino acids in A20FMDV2 were systematically substituted with glycine, yielding the optimized peptide 5G. Peptide 5G has shown promising performance in PDAC. In preclinical studies, [⁶⁸Ga]Ga-DOTA-ABM-5G exhibited tumor uptake of 10.4 ± 2.6%ID/g at 4 h p.i., while [¹⁷⁷Lu]Lu-DOTA-ABM-5G significantly extended median survival by 57 days²⁸. A Phase I clinical trial further demonstrated the safety and theranostic potential of [⁶⁸Ga]Ga-DOTA-5G and [¹⁷⁷Lu]Lu-DOTA-ABM-5G³⁰. The incorporation of a DOTA chelator into the 5G scaffold enabled radiolabeling and imaging functionality, a feature that was retained in the design of the B-10 carriers. The icosahedral carborane cluster, a structural analogue of BSH, is characterized by a high boron density (10 boron atoms/molecule) and enhanced functionalization feasibility, which has been widely used in BNCT studies³¹.

In this study, we integrated three functional components into a single molecular entity. The peptide was synthesized using the Fmoc-SPPS strategy and subsequently conjugated with BSH analogues after cleavage (Fig. 1A). To address the long-standing challenge of poor drug penetration in stroma-rich PDAC, the αvβ6-targeting peptide was deliberately selected not only for its high tumor specificity but also for its contribution to tissue permeability. First, conjugation of hydrophobic BSH analogues to the intrinsically hydrophilic peptide backbone generated an amphiphilic molecular architecture, which has been shown to facilitate penetration across biological barriers and improve intratumoral distribution in desmoplastic tumors²³. In addition, αvβ6 is highly overexpressed in PDAC while being minimally expressed in normal pancreatic tissue, and undergoes clathrin-mediated endocytosis, enabling continuous internalization and intracellular accumulation of bound ligands²⁵. Targeting such an internalizing receptor therefore provides a promising strategy to enhance boron delivery efficiency and BNCT efficacy in PDAC.

Figure 1.

Synthesis, optimization, and chemical structures of αvβ6-targeting B-10 carriers. (A) General synthetic route of αvβ6-targeting B-10 carriers. (B) Schematic diagram of the “three-round” optimization based on BNCT-TI and the chemical structures of αvβ6-targeting B-10 carriers.

To facilitate rational screening and optimization of B-10 carriers, we introduced the BNCT therapeutic index (BNCT-TI), defined as the product of boron mass ratio and tumor uptake (BNCT-TI = boron mass ratio% × tumor uptake %ID/cc), as a heuristic metric for early-stage candidate evaluation. A three-round optimization strategy was implemented to maximize BNCT-TI while preserving favorable pharmacokinetic and penetration properties (Fig. 1B). In the first round, to increase the boron mass ratio, we shortened the PEG chains or retained only N-terminal mono-PEGylation based on the peptide 5G scaffold. Although prior study indicated that bi-PEGylation of the αvβ6-targeting peptide enhanced pharmacokinetic properties and tumor uptake^32,33, our subsequent biological evaluations confirmed that reduced PEGylation enhanced tumor retention of the B-10 carriers, and indeed increased the BNCT-TI. In the second round, the length of fatty acid linker was explored. The icosahedral carborane clusters were functionalized with short-to medium-length fatty acid chains (C2, C4, and C6) to form BSH analogues 2C-BSH, 4C-BSH, and 6C-BSH (Supporting Information Fig. S1). The structures were confirmed by ¹H NMR and LC‒MS, and their purity was verified by HPLC (>95%) (Supporting Information Figs. S2–S4).

Subsequently, BSH analogues were conjugated to the peptide-DOTA scaffold, generating DOTA-2B, DOTA-4B, and DOTA-6B. Although elongation of the fatty acid chain moderately increased cellular uptake in subsequent biological evaluations, it also led to a slight reduction in boron content. Consequently, this modification provided negligible improvement in the overall BNCT-TI. The BNCT-TI values were 26.82 for DOTA-2B, 35.81 for DOTA-4B, and 29.58 for DOTA-6B, respectively (Supporting Information Table S2). Notably, variation in linker length did not significantly enhance the BNCT-TI.

In the third round, to further increase boron payload without compromising permeability, two 2C-BSH moieties were conjugated within a single molecular framework, generating the lead candidate 2BB. This molecule adopted a distinctive “trident-shaped” architecture, which combines high boron mass ratio, amphiphilic balance, and αvβ6-mediated internalization, ultimately yielding the highest BNCT-TI among all candidates and demonstrating strong potential as an effective B-10 carrier. As a control group, DOTA-BSH was synthesized, and its structure is shown in Supporting Information Fig. S5. The synthetic route of these molecules is shown in Supporting Information Fig. S6. Following HPLC purification of the crude products, the target compounds were obtained with >95% purity and their structures were verified by LC‒MS analysis (Supporting Information Figs. S7–S24). Notably, BNCT-TI was intended as a simplified in vitro screening parameter rather than a comprehensive predictor of in vivo or clinical BNCT efficacy. This index does not incorporate critical pharmacokinetic factors such as tumor-to-normal tissue (T/N) boron ratios, boron retention time, or biodistribution, which are essential determinants of therapeutic outcome. These parameters were therefore systematically evaluated in subsequent in vivo imaging, biodistribution, and BNCT efficacy studies described below.

3.2. In vitro cellular uptake and membrane penetration studies

The binding affinity of DOTA/Cy5-labeled targeting B-10 carriers was evaluated by molecular docking and cellular uptake assays in αvβ6-overexpressing BxPC-3 cells. Molecular docking analysis revealed that DOTA-2BB exhibited the most favorable binding free energy (−5.41 kcal/mol), followed by DOTA-2B (−4.62 kcal/mol), indicating a pronounced enhancement in predicted binding affinity compared with the parent compound DOTA-O (−2.19 kcal/mol). In contrast, DOTA-P4 showed a binding free energy of −2.86 kcal/mol, comparable to that of DOTA-O, suggesting only marginal improvement in receptor interaction (Supporting Information Fig. S25).

Flow cytometry results revealed that all modified analogues exhibited measurable cellular binding affinity (Fig. 2A), with a time-dependent increase in cellular internalization (Supporting Information Fig. S26A). A gradual decrease in PEG chain length correlated with a progressive increase in cellular affinity for the Cy5-labeled targeting B-10 carriers, demonstrating the efficacy of this modification strategy. Among fatty acid-modified derivatives, Cy5-6B demonstrated 2.3-fold and 1.7-fold higher MFI than Cy5-2B and Cy5-4B, respectively, suggesting a chain-length-dependent enhancement in cellular affinity. Conversely, PEGylated molecules showed no significant improvement in αvβ6 binding affinity compared with other modification strategies, and exhibited lower MFI values than the fatty acid-modified molecules.

Figure 2.

Cellular uptake and membrane penetration of Cy5-labeled B-10 carriers in αvβ6-positive BxPC-3 cells. (A) Flow cytometry histograms showing cellular uptake after incubation with 2 μmol/L Cy5-labeled B-10 carriers for 30 min. (B) Quantitative analysis of MFI of flow cytometry assay of main Cy5-labeled B-10 carriers. (C) Co-localization analysis based on confocal microscopy images of Cy5-O, -P4, -2B and -2BB. Fluorescence intensity profiles were quantified using line-scan analysis across cells, as shown in panel (D). (D) Confocal microscopy images of BxPC-3 cells incubated with 2 μmol/L Cy5-labeled B-10 carriers for 30 min. Red and blue indicate Cy5 signal and nuclei, respectively. Scale bar: 10 μm. (E) Flow cytometry histograms of Cy5-2BB. Cells were incubated with decreasing ligand concentrations (5-fold serial dilution, from 25 μmol/L to 64 pmol/L) for 30 min at 37 °C. (F) Equilibrium binding curves of Cy5-O, Cy5-2B, Cy5-P4, and Cy5-2BB to BxPC-3 cells, as measured by flow cytometry saturation assays. Data were fitted to a one-site specific binding model (R² > 0.94). (G) K_D values of Cy5-O, -2B, -P4 and -2BB. (H) Confocal fluorescence images of BxPC-3 tumor sections collected after i.v. administration of Cy5-2BB and stained with DAPI. Scale bar: 120 μm in the full images, 50 μm in the magnified insets. “C-” is the abbreviation of “Cy5-”, stands for the Cy5 conjugated targeting B-10 carriers. All results are expressed as means ± SEM, as indicated in at least three independent experiments (n = 3). A two-tailed unpaired Student's t-test was used when two groups were compared. The symbol “∗” represents differences compared with O; ∗∗∗∗P < 0.0001. ns indicates non-significant.

Notably, the dual-carborane cluster Cy5-2BB achieved 9.9-fold greater MFI than Cy5-O, indicating its potential as a highly efficient B-10 carrier (Fig. 2B). Confocal microscopy further confirmed these findings, demonstrating specific cellular uptake of all Cy5-labeled B-10 carriers in BxPC-3 cells compared to the free-Cy5 control. Among them, Cy5-2B and Cy5-2BB, which were conjugated with one or two 2C-BSH moieties respectively, exhibited increased fluorescence intensity in the cytoplasm relative to the unmodified Cy5-O (Fig. 2C). This enhancement was attenuated in the PEGylated derivative Cy5-P4. Consistently, all molecules were predominantly localized in the cytoplasm (Fig. 2D and Fig. S26C).

These results indicate that the conjugation of BSH analogues enhances the cellular affinity of B-10 carriers, and this improvement correlates with the molecular hydrophobicity. Specifically, increased fatty acid chain length, shorter PEG chains, and a greater number of BSH modules respectively contributed to increased molecular hydrophobicity³⁴. Therefore, we propose that, in addition to receptor-mediated endocytosis, the incorporation of BSH-derived hydrophobic groups facilitates the uptake of Cy5-labeled B-10 carriers into αvβ6 high-expressing cells. The amphiphilic nature conferred by the hydrophobic BSH moieties enables the peptide-based agents to adopt a structure with distinct hydrophilic and hydrophobic regions, which may promote efficient translocation across the lipid bilayer of the cell membrane^18,26. The efficacy of BNCT relies on the alpha particles (⁴He²⁺) and lithium nucleus (⁷Li³⁺) emitted from B-10 atoms, which are capable of inducing DNA double-strand breaks within the cell nucleus³⁵. Since these particles have a path length limited to approximately one cell diameter (5‒9 μm), B-10 carriers with enhanced membrane permeability may enable more efficient nuclear DNA damage and improved cytotoxic efficacy^35,36.

We further selected representative molecules from each modification strategy for quantitative binding assessment (Fig. 2E and Fig. S26B). Saturation flow cytometry analysis revealed that Cy5-2BB possessed superior affinity (K_D = 1.23 μmol/L) compared to Cy5-O (48.91 μmol/L), Cy5-P4 (93.29 μmol/L), and Cy5-2B (5.51 μmol/L) (Fig. 2F‒G). Consistent with these findings, confocal imaging of subcutaneous tumor slides further demonstrated that Cy5-2BB was able to penetrate deeply into tumor tissue (Fig. 2H). This behavior is thought to arise from its amphiphilic “trident-shaped structure”, where spatially segregated hydrophilic and hydrophobic domains may facilitate efficient diffusion.

3.3. In vivo PET/CT imaging of ⁶⁸Ga-labeled B-10 carriers

To evaluate the in vivo targeting efficacy and therapeutic potential of the molecules, we conducted PET/CT imaging with ⁶⁸Ga-labeled B-10 carriers using a BxPC-3 subcutaneous tumor-bearing mouse model³⁷. All ⁶⁸Ga-labeled B-10 carriers exhibited radiochemical purity over 95% measured by radio-HPLC, and their in vitro and in vivo stability were evaluated (Supporting Information Figs. S27‒S29). Various modification strategies influenced hydrophobicity differentially. Given the established correlation between lipophilicity and membrane permeability, we evaluated the partition coefficients (LogP) of ⁶⁸Ga-labeled B-10 carriers to characterize their hydrophobicity³⁸. The results indicated that PEGylated analogues showed no substantial change in lipophilicity compared with [⁶⁸Ga]Ga-DOTA-O. In contrast, [⁶⁸Ga]Ga-DOTA-4B and [⁶⁸Ga]Ga-DOTA-6B exhibited significantly higher lipophilicity than [⁶⁸Ga]Ga-DOTA-O. Notably, [⁶⁸Ga]Ga-DOTA-2BB was demonstrated with the highest hydrophobicity among all the compounds tested (Supporting Information Table S1).

Various modification strategies significantly altered the in vivo metabolic profiles of ⁶⁸Ga-labeled B-10 carriers (Fig. 3A). PEGylated ⁶⁸Ga-labeled B-10 carriers exhibited reduced non-specific accumulation, particularly in liver ([⁶⁸Ga]Ga-DOTA-PP12: 5.01 ± 0.40 %ID/cc at 1 h p.i.). Moreover, agents with longer PEG chains showed accelerated clearance and rapid urinary excretion, which was attributed to significantly enhanced kidney uptake ([⁶⁸Ga]Ga-DOTA-PP12: 61.20 ± 2.36 %ID/cc at 1 h p.i., Fig. 3A and B and Supporting Information Fig. S31). According to previous research by Julie et al.³², poly-PEG on N and C terminals of αvβ6-targeting peptide A20FMDV2, which is the precursor to peptide 5G, can enhance the pharmacokinetic properties of the ligand.

Figure 3.

In vivo PET/CT imaging of ⁶⁸Ga-labeled B-10 carriers in BxPC-3 subcutaneous tumor-bearing mice. The images only show the lower half of the mice, and the upper half of the mice are separately displayed in Fig. 4 (A) Representative PET/CT images of the liver and kidneys at 1, 2, and 4 h p.i. (B, C) Quantitative analysis of kidney (B) and heart (C) uptake (%ID/cc) at 4 h p.i. “D-” is the abbreviation of “DOTA-”, stands for the DOTA conjugated targeting B-10 carriers. All results are expressed as means ± SEM, as indicated in at least three independent experiments (n = 3). A two-tailed unpaired Student's t-test was used when two groups were compared. The symbol “∗” represents differences compared with O; ∗∗∗P < 0.001. ∗∗∗∗P < 0.0001. ns indicates non-significant.

Fatty acid-modified ⁶⁸Ga-labeled B-10 carriers exhibited chain length-dependent hepatobiliary accumulation. Hepatic uptake at 1 h p.i. increased progressively from 8.08 ± 0.25 %ID/cc ([⁶⁸Ga]Ga-DOTA-2B) to 13.02 ± 1.02 %ID/cc ([⁶⁸Ga]Ga-DOTA-4B) and 20.33 ± 0.61 %ID/cc ([⁶⁸Ga]Ga-DOTA-6B). This trend may be attributed to the progressive increase in lipophilicity, which potentially promoted hepatic uptake. The dual-carborane construct [⁶⁸Ga]Ga-DOTA-2BB demonstrated prolonged systemic retention with multi-organ deposition. It should be noted that BNCT differs from internal radionuclide therapy. In BNCT, the treatment outcome is determined by the absolute tumor B-10 concentration and the tumor-to-normal (T/N) ratio at the time of neutron irradiation. In our study, PET-derived SUV and the corresponding T/N ratios primarily reflect the biodistribution and pharmacokinetics of the whole molecules rather than that of B-10, and therefore do not directly translate into BNCT therapeutic efficacy. This is because each molecule has a different B-10 mass ratio. Consequently, the BNCT-TI is an integrated metric that takes into account the PET SUV value, the SUV-derived T/N ratio, and the B-10 mass ratio.

Subsequently, the cellular uptake capacity of representative ⁶⁸Ga-labeled B-10 carriers was evaluated (Supporting Information Fig. S30). The results demonstrated that [⁶⁸Ga]Ga-DOTA-2BB exhibited a markedly enhanced binding affinity compared to [⁶⁸Ga]Ga-DOTA-2B, with EC₅₀ (half maximal effective concentration) values of 5.46 and 25.49 nmol/L, respectively. Consistent trends were observed in the competitive inhibition assay, where [⁶⁸Ga]Ga-DOTA-2BB showed a lower IC₅₀ (half maximal inhibitory concentration) value (8.90 nmol/L) than [⁶⁸Ga]Ga-DOTA-2B (40.49 nmol/L). For PET/CT imaging, all radioligands demonstrated measurable tumor targeting specificity in αvβ6 overexpressing BxPC-3 xenografts compared to the control [⁶⁸Ga]Ga-DOTA-BSH (Fig. 4A). Among these candidates, DOTA-2BB showed the highest tumor uptake at 4 h p.i. To further confirm the αvβ6-specific binding capacity, DOTA-2BB was chosen for a blocking study using an excess of the 5G peptide. As a result, tumor uptake of [⁶⁸Ga]Ga-DOTA-2BB at 4 h p.i. markedly decreased from 10.98 to 0.26 %ID/cc, demonstrating the high αvβ6 specificity of DOTA-2BB (Fig. 4B). PEGylated B-10 carriers were excluded from further optimization due to their consistently lower general uptake compared to other modified variants. Fatty acid-modified derivatives exhibited chain length-dependent tumor interactions. Tumor uptake values at 1 h p.i. were as follows: [⁶⁸Ga]Ga-DOTA-2B, 9.51 ± 0.71 %ID/cc; [⁶⁸Ga]Ga-DOTA-4B, 9.83 ± 0.03 %ID/cc; and [⁶⁸Ga]Ga-DOTA-6B, 13.14 ± 0.85 %ID/cc (Supporting Information Fig. S31). Although [⁶⁸Ga]Ga-DOTA-6B exhibited higher initial tumor uptake at 1 h (13.14 ± 0.85 %ID/cc), its accumulation decreased significantly by 4 h (8.19 ± 0.24 %ID/cc). In contrast, [⁶⁸Ga]Ga-DOTA-2BB showed a continuous increase in tumor over time, rising from 8.47 ± 0.78 %ID/cc at 1 h to 10.98 ± 0.17 %ID/cc at 4 h, demonstrating superior retention compared to DOTA-6B (Fig. 4C and Fig. S31).

Figure 4.

In vivo PET/CT imaging of ⁶⁸Ga-labeled B-10 carriers in BxPC-3 subcutaneous tumor-bearing mice. Only the upper half of mice were displayed (referred to Fig. 3A). (A) Representative PET/CT images of the tumor region at 1, 2, and 4 h p.i. (B) Representative PET/CT images of the blocking study of [⁶⁸Ga]Ga-DOTA-2BB at 1, 2, and 4 h p.i. (C) Quantitative analysis of tumor uptake (%ID/cc) at 4 h p.i. of ⁶⁸Ga-labeled B-10 carriers. (D) Biodistribution analysis of [⁶⁸Ga]Ga-DOTA-2BB in BxPC-3 tumor-bearing mice. “D-” is the abbreviation of “DOTA-”, stands for the DOTA conjugated targeting B-10 carriers. All results are expressed as means ± SEM, as indicated in at least three independent experiments. A two-tailed unpaired Student's t-test was used when two groups were compared. The symbol “∗” represents differences compared with O; ∗∗∗P < 0.001. ∗∗∗∗P < 0.0001. ns indicates non-significant.

However, notable off-target accumulation was observed in [⁶⁸Ga]Ga-DOTA-2BB injected mice. Although [⁶⁸Ga]Ga-DOTA-2BB showed high cardiac region uptake (exceeding 20.0 %ID/cc within 4 h p.i., Fig. 4A, B and Fig. S31), ex vivo biodistribution revealed low actual heart tissue accumulation, indicating that radioactivity was primarily confined to the blood pool and thus was likely associated with minimal cardiotoxicity (Fig. 4D and Supporting Information Fig. S32)³⁹. This was consistent with the known albumin-binding tendency of BSH analogues^40,41. In our previous study using ⁶⁴Cu-labeled DOTA-2BB, strong plasma albumin binding was experimentally confirmed, resulting in markedly prolonged blood circulation and exceptionally high tumor AUC (410.2 %ID·cc⁻¹·h)⁴². Therefore, the elevated heart-region %ID/cc observed for [⁶⁸Ga]Ga-DOTA-2BB primarily reflects albumin-mediated blood retention. This prolonged circulation enables sustained tumor exposure and continuous tumor uptake over time.

To this end, based on the calculated BNCT-TI and T/B ratio for each agent at 4 h p.i (Supporting Information Table S2), two candidates, [⁶⁸Ga]Ga-DOTA-2BB, exhibiting the highest BNCT-TI, and [⁶⁸Ga]Ga-DOTA-2B, showing a BNCT-TI above 20 and the highest T/B ratio, were selected for further biological evaluation.

3.4. In vivo fluorescence imaging of Cy5-labeled B-10 carriers

Because of the short half-life of ⁶⁸Ga, which precludes longitudinal tracking of pharmacokinetics beyond 4 h, we employed fluorescence imaging with Cy5-labeled B-10 carriers to evaluate long-term biodistribution profiles. Tumor-bearing mice received intravenous injections of representative Cy5-labeled B-10 carriers, enabling semi-quantitative assessment of tumor targeting efficiency and organ-specific accumulation⁴³.

In vivo fluorescence imaging revealed significantly enhanced tumor-targeting ability for Cy5-2B and Cy5-2BB compared with the unmodified peptide (Fig. 5A). Moreover, pharmacokinetic variations were observed: while the original peptide was rapidly cleared by kidney, with near-complete elimination within 24 h, Cy5-2B and Cy5-2BB exhibited predominantly hepatic elimination. Notably, Cy5-2BB demonstrated enhanced tumor targeting efficiency and significantly prolonged retention, peaking as late as 48 h p.i., with 2.42 ± 0.28 × 10⁷ MFE, representing a 44% increase over Cy5-2B's signal (MFE = 9.34 ± 0.05 × 10⁶) (Fig. 5A and B). Crucially, non-target organ deposition was minimized at this therapeutic window around 48 h, yielding optimal tumor-to-background contrast. This biodistribution profile established Cy5-2BB as a promising delayed-phase imaging agent, with ex vivo organ fluorescence confirming spatial consistency with in vivo observations (Fig. 5C‒E and Supporting Information Fig. S33).

Figure 5.

In vivo fluorescence imaging of BxPC-3 luc tumor-bearing mice. (A) Representative fluorescence images of subcutaneous BxPC-3 luc tumor-bearing mice after injection of 15 mg/kg Cy5-2BB, Cy5-2B, and Cy5-O (i.v.) at various time points. (B) Quantitative analysis of MFE in subcutaneous tumors after 15 mg/kg Cy5-2BB, Cy5-2B, and Cy5-O i.v. administration. (C) Ex vivo fluorescence imaging of subcutaneous tumors and major organs from the treated mice at 48 h p.i. (D) Time-dependent MFI curves in kidneys after injection of Cy5-labeled B-10 carriers. (E) Time-dependent MFI curves in liver after injection of Cy5-labeled B-10 carriers. (F) Co-localization of luciferase bioluminescence and Cy5 fluorescence signals in an orthotopic BxPC-3 luc mouse model at 48 h p.i. (G) Ex vivo fluorescence and bioluminescence imaging of major organs, including the tumor-bearing pancreas from the orthotopic model, harvested at 48 h p.i. (H) Quantitative analysis of orthotopic PDAC tumor uptake (MFI) at 48 h p.i. “H” stands for heart, “Li” stands for liver, “Sp” stands for spleen, “Lu” stands for lung, “Ki” stands for kidney, “Sto” stands for stomach, “I” stands for intestine, “Pan” stands for pancreas, “Br” stands for brain, “Tu” stands for tumor. All results are expressed as means ± SEM, as indicated in at least three independent experiments (n = 3). A two-tailed unpaired Student's t-test was used when two groups were compared. The symbol “∗” represents differences compared with O; ∗∗∗∗P < 0.0001.

Building on promising subcutaneous tumor results, we evaluated the Cy5-labeled B-10 carriers’ targeting efficacy in an orthotopic pancreatic tumor model using dual-modality fluorescence-bioluminescence imaging. Cy5-2B and Cy5-2BB exhibited precise co-localization with luciferase-expressing tumors (Fig. 5F). Ex vivo analysis at 48 h p.i. confirmed selective accumulation in malignant pancreatic tissues versus minimal uptake in adjacent normal pancreas (Fig. 5G). Quantitative results from the orthotopic pancreatic cancer model revealed that Cy5-2BB exhibited the strongest fluorescence signal in the tumor region (MFI 9.56 × 10⁶ ± 0.28 × 10⁶), followed by Cy5-2B (MFI = 6.30 ± 0.07 × 10⁶), both significantly higher than that of Cy5-O (MFI = 2.81 ± 0.00 × 10⁶) (Fig. 5H).

The enhanced and BSH moiety-dependent tumor accumulation observed with Cy5-2B and Cy5-2BB in orthotopic pancreatic tumors further supported the critical role of lipophilicity in promoting membrane permeability and stromal diffusion. PDAC is renowned for its highly dense and desmoplastic extracellular matrix, which profoundly restricts drug delivery and diffusion, contributing to poor treatment outcomes^44,45. The stepwise increase in fluorescence from Cy5-2B to Cy5-2BB suggested that BSH-mediated lipophilicity not only enhanced cellular uptake but also facilitated penetration through this mechanically restrictive TME. Consistent with these findings, in vivo fluorescence imaging confirmed the superior performance of Cy5-2BB over Cy5-2B, leading to its selection as the lead B-10 carrier for subsequent boron delivery studies using DOTA-2BB. These findings validated a rational chemical strategy, modulating lipophilicity via BSH conjugation to overcome stromal barriers and improve boron delivery in pancreatic cancer, supporting further development of stroma-compatible B-10 carriers.

3.5. Targeted boron delivery for BNCT in PDAC

According to the PET/CT and fluorescence imaging results, DOTA-2BB was selected as the optimal candidate with satisfactory BNCT-TI and biosafety for further therapeutic study. We first assessed the tumor-killing efficacy of DOTA-2BB at the cellular level using a clonogenic survival assay. Cells were seeded in culture dishes, irradiated with thermal neutron beams, and allowed to form colonies over 14 days before being stained with crystal violet (Fig. 6A). The stained area was quantified to determine cell survival fractions and viability. DOTA-2BB combined with neutron irradiation reduced colony formation by 17.8-fold, significantly outperforming BSH (5.9-fold reduction). Furthermore, the higher colony formation area in the non-irradiated DOTA-2BB group compared with the BSH group further indicates a superior safety profile of DOTA-2BB (Fig. 6B and C).

Figure 6.

Preclinical assessment of BNCT efficacy and boron biodistribution. (A) Schematic illustration of the clone formation assay. Created with Biorender.com. (B) Representative images of clonogenic assay results. Purple staining indicates viable cell colonies; unstained areas represent inhibition of cell growth. (C) Quantitative analysis of clonogenic survival after BNCT treatment. (D) Boron concentration profiles in tumors and blood of BxPC-3 subcutaneous tumor-bearing mice within 48 h after intravenous injection of DOTA-2BB. (E) Biodistribution of boron in tumors and major organs at 24 h p.i. All results are expressed as means ± SEM, as indicated in at least three independent experiments (n = 3). Statistical significance was determined by two-tailed unpaired Student's t-test (∗∗∗∗P < 0.0001, compared with control; ^####P < 0.0001, compared with each other).

Given its potential as a B-10 carrier, we further evaluated the biodistribution of boron delivered by DOTA-2BB using ICP-MS. Tumor boron levels were low within the first 12 h after injection, increased sharply to 103.3 μg/g at 24 h, and then slightly decreased by 48 h while still remaining above the minimum level required for effective BNCT (Fig. 6D). Concurrently, boron concentrations in blood remained consistently low, yielding a T/B ratio of 40.6 at 24 h (Fig. 6E). These results demonstrated that DOTA-2BB achieved significantly higher tumor boron accumulation and improved T/N ratios compared with ¹⁰B-BPA (Supporting Information Fig. S34A). It should be noted that this boron biodistribution profile differs from the fluorescence imaging results obtained with Cy5-2BB, as the introduction of the Cy5 fluorophore is known to alter in vivo metabolic behavior⁴⁶. Given that DOTA-2BB was used in the subsequent BNCT therapeutic studies, boron distribution derived from ICP-MS was considered more relevant for evaluating BNCT efficacy.

The T/N ratio is a critical parameter for evaluating B-10 carriers, as it directly determines the therapeutic window and radiation damage to healthy tissues⁴⁷. Our boron biodistribution results demonstrate that DOTA-2BB achieves an exceptionally high T/B ratio of 40.6 at 24 h p.i. This was attributed to its prolonged systemic retention combined with continuous tumor-selective uptake. While boron content in blood and normal organs decreased significantly over time due to metabolic clearance, tumor tissues exhibited sustained accumulation, peaking at 24 h (103.3 μg/g). Such a high and delayed T/N ratio is advantageous for practical BNCT applications, as it allows sufficient time for selective boron buildup within tumors and clearance from radiosensitive organs before neutron irradiation⁴⁸.

3.6. In vivo evaluation of targeting BNCT utilizing DOTA-2BB in a PDAC mouse model

Consequently, we established a subcutaneous BxPC-3 tumor-bearing mice model to evaluate the therapeutic efficacy of DOTA-2BB. During irradiation, anesthetized mice were arranged in a circular pattern with tumors targeted toward the neutron beam center, while abdominal organs were shielded from radiation (Fig. 7A). Groups received i.t. administration of DOTA-2BB (50 mg/kg), BSH (5 mg/kg), or saline (100 μL) 1 h before irradiation, and another group received i.v. administration of DOTA-2BB (50 mg/kg) 24 h before. All groups were irradiated with a thermal neutron beam [7 × 10⁸ n/(cm²·s), 30 min of irradiation time] (Fig. 7B). Non-irradiated control groups showed over 6-fold tumor growth by Day 12, after which the mice were euthanized. In contrast, DOTA-2BB combined with neutron irradiation significantly suppressed tumor progression, with intratumoral and intravenous groups exhibiting only 2.1- and 2.2-fold increases in tumor volume, respectively. However, no statistically significant difference in tumor volume reduction was observed between the DOTA-2BB (both i.t. and i.v.) and BSH treatment groups (Fig. 7D‒E and Fig. S34B and S34C). These results indicated that DOTA-2BB potently enhanced the therapeutic efficacy of thermal neutron irradiation and achieved antitumor performance superior to that of BSH. Additionally, all animals in the treatment groups survived until the end of the 14-day study, except for those that reached predefined ethical tumor volume limits. Body weights remained within normal fluctuations throughout the experiment, suggesting minimal systemic toxicity (Fig. 7C).

Figure 7.

In vivo BNCT for DOTA-2BB. (A) Photograph of mice receiving a localized thermal neutron beam irradiation (n = 5). The blue dashed box indicates the irradiation field. (B) Schematic diagram of the BNCT treatment and dosing schedule. Created with Biorender.com. (C) Body weight change (BWC) curves of mice following treatment (n = 5). (D) Relative tumor volume (RTV) growth curves over 14 days post-irradiation (n = 5). The dashed line represents the mean tumor growth trajectory. (E) Summary of tumor growth curves across all treatment groups during the 14-day observation period. (F) Representative histology and immunofluorescence staining of tumor tissues at the experimental endpoint, including hematoxylin-eosin (HE), Ki67 (proliferation), γ-H2AX (DNA double-strand breaks), and αvβ6 (target expression). Scale bars in these groups are 75, 50, 10, and 50 μm, respectively. (G) Blood biochemistry results of groups exposed to thermal neutrons at 14 days post-irradiation, including AST (aspartate aminotransferase), ALT (alanine aminotransferase), UREA (urea), and CREA (creatinine). (H) Blood routine test results of groups exposed to thermal neutrons at 14 days post-irradiation, including WBC (white blood cell), RBC (red blood cell), HGB (hemoglobin), HCT (hematocrit), MCV (mean corpuscular volume), MCH (mean corpuscular hemoglobin), MCHC (mean corpuscular hemoglobin concentration), PLT (platelet). “D-” is the abbreviation of “DOTA-”, stands for the DOTA conjugated targeting B-10 carriers. All results are expressed as means ± SEM, as indicated in at least five independent experiments (n = 5). A two-tailed unpaired Student's t-test was used when two groups were compared. “ns” represents no significant difference. The symbol “∗” represents differences compared with the neutron group. ∗P < 0.05. ∗∗P < 0.01.

We further performed HE, Ki67, γ-H2AX, and αvβ6 immunofluorescence staining on tumor tissues. HE staining revealed pronounced tissue damage in the thermal neutron irradiated groups, with the most severe injury observed in the i.t. administration of DOTA-2BB group, followed by the i.v. administration of DOTA-2BB group, and then the i.t. administration of BSH group; no significant histopathological changes were detected in neutron-only irradiated group or non-irradiated controls. Ki67 and γ-H2AX staining showed similar trends, indicating that the i.t. administration of DOTA-2BB group showed the most substantial reduction in tumor proliferation and the highest level of DNA double-strand breaks, followed by the intravenous DOTA-2BB and i.t. administration of BSH groups. Furthermore, αvβ6 expression decreased most markedly in both i.t. and i.v. administration of DOTA-2BB-treated tumors, while no obvious change was observed in the other groups, further confirming the targeting specificity of DOTA-2BB (Fig. 7F and Fig. S34).

Treatment safety was evaluated through HE staining, hematological, and biochemical analyses. HE staining of major organs revealed no significant evidence of organ damage (Supporting Information Figs. S35 and S36). Furthermore, comprehensive hematological and serum biochemical analysis showed no significant abnormalities in key parameters across all groups, indicating that the combined regimen of B-10 carriers administration and neutron irradiation was well tolerated without significant systemic toxicity (Fig. 7G‒H and Supporting Information Fig. S37).

The comparable therapeutic efficacy observed between the intravenous and intratumoral administration routes of DOTA-2BB may be attributed to two key factors. First, DOTA-2BB exhibits prolonged retention in the systemic circulation, which enables continuous and efficient tumor uptake over an extended period. This high bioavailability following intravenous injection allowed sufficient boron accumulation in tumor tissues by the time of neutron irradiation⁴⁹. Second, i.t. administration often suffers from uneven distribution of the agent, particularly in poorly perfused or necrotic regions of the tumor. Given that the radiation range of boron neutron capture reaction is limited to a single cell diameter, heterogeneous distribution significantly compromises treatment efficacy, as some areas receive inadequate radiation dose. Conversely, the intravenous approach achieves more thorough and homogeneous tumor coverage.

It is important to note that in this study, DOTA-2BB was synthesized using natural boron (∼20% B-10) rather than B-10-enriched boron (>96% B-10). The resulting lower B-10 density likely limited the therapeutic outcome, as the neutron capture cross section is directly proportional to the concentration of B-10 atoms. This limitation suggests that further improvement in therapeutic efficacy is possible. Numerous chemical and biological methods have been developed to incorporate B-10-enriched boron carriers into targeted molecules^50,51. Applying these strategies to DOTA-2BB, for instance, using B-10-enriched decaborane or carborane precursors in synthesis, could substantially increase the B-10 payload per molecule, thereby enhancing neutron capture efficiency and maximizing cytotoxic effects without altering the targeting capability⁵².

3.7. Transcriptomic profiling of PDAC tumor tissues following BNCT treatment

Transcriptomic analysis revealed significant alterations in gene expression profiles across different treatment groups, demonstrating the therapeutic effects of neutron irradiation in combination with boron agents. Compared to the only neutron-irradiated group, both DOTA-2BB + N and BSH + N treatment groups exhibited widespread differential gene expression (Fig. 8A and Supporting Information Fig. S38A and S38B). Notably, the heatmap of representative differentially expression genes associated not only with tumor cell destruction but also with regulation of the TME, suggesting that DOTA-2BB + N may inhibit tumor growth by remodeling the TME of PDAC (Fig. 8A)⁵³.

Figure 8.

Transcriptomic profiling of PDAC tumor tissues following BNCT treatment. (A) Representative differentially expressed genes between D-2BB-i.t. + N, BSH + N, and neutron group. (B) Overview of differentially expressed genes between D-2BB-i.t. + N and BSH + N groups. (C) Bubble plot of representative pathways enriched in D-2BB-i.t. + N group versus BSH + N group. (D) Representative differentially expressed genes between D-2BB-i.t. + N and BSH + N. (E) Immunofluorescence images of MMP2 and ACTA2 protein expression in tumors from the D-2BB-i.t. + N and BSH + N groups. (F) Quantitative mean fluorescence intensity analysis of immunofluorescence images in Fig. 8E. “D-” is the abbreviation of “DOTA-”, stands for the DOTA conjugated targeting B-10 carriers. Scale bars in Fig. 8E stand for 50 μm. All results are acquired at least three independent experiments (n = 3). A two-tailed unpaired Student's t-test was used when two groups were compared.

Despite the similar therapeutic efficacy of DOTA-2BB + N and BSH + N treatment, further comparative transcriptomic profiling between the DOTA-2BB + N and BSH + N treatment groups identified 453 differentially expressed genes (Fig. 8B), indicating difference in the mechanisms of cell killing effects. Enrichment analysis highlighted pronounced changes in biological processes related to DNA damage response, immune activation, and TME remodeling (Fig. S38C and S38D). GO enrichment analysis presented as bubble plots demonstrated that irradiation after DOTA-2BB administration exerted antitumor effects through both direct cytotoxicity and modulation of the TME (Fig. 8C).

Firstly, genes involved in cell cycle regulation and angiogenesis were significantly downregulated in the DOTA-2BB + N group (Fig. 8D). For instance, SPDYA (a regulator of cell cycle progression)⁵⁴ and TIE1 (an endothelial receptor involved in angiogenesis)⁵⁵ were markedly suppressed, consistent with extensive tumor cell damage induced by high LET particles during neutron irradiation⁵⁶. Additionally, DOTA-2BB treatment was associated with remodeling of the stromal tumor microenvironment. Fibrosis-related genes, including ACTA2 (a marker of activated cancer-associated fibroblasts driving desmoplasia)⁵⁷, MMP2 (an extracellular matrix-degrading enzyme promoting tumor invasion)⁵⁸, and TAGLN (a cytoskeleton-associated protein linked to metastasis and immunosuppression)⁵⁹, were significantly downregulated, indicating reduced stromal density. This change may alleviate the physical barrier of the PDAC stroma and facilitate drug penetration (Fig. 8D).

To validate the transcriptomic findings at the protein level, immunofluorescence staining was performed to assess the expression of MMP2 and ACTA2. MMP2 and ACTA2 expression levels were markedly decreased in the D-2BB-i.t. + N group compared with the BSH + N group (Fig. 8E and F). Taken together, DOTA-2BB not only exhibited enhanced tissue penetration capabilities but also remodeled the TME post-treatment, potentially promoting deeper drug delivery and improved therapeutic response⁶⁰. This dual mechanism underscores its significant therapeutic potential for cancers with dense extracellular matrix barriers, such as PDAC⁶¹.

4. Conclusions

This study reports the development and comprehensive evaluation of a series of αvβ6-targeted amphiphilic boron delivery agents engineered for BNCT in PDAC. By incorporating an αvβ6-specific peptide warhead, DOTA-2BB achieved enhanced cellular internalization, prolonged tumor retention, and significantly improved boron accumulation in tumors. The agent exhibits high targeting affinity for αvβ6-positive BxPC-3 cells, ensuring effective and specific receptor engagement. In vivo studies confirmed exceptional tumor targeting and retention properties of DOTA-2BB. Fluorescence imaging further validated prolonged tumor accumulation with superior contrast in an orthotopic PDAC model. Most notably, intravenous administration resulted in a tumor boron concentration of 103.3 μg/g at 24 h p.i. and a T/B ratio of 40.6, exceeding the therapeutic threshold for effective BNCT. These pharmacokinetic properties enabled effective boron delivery through systemic administration, eliminating the need for intratumoral injections. In addition to efficient boron delivery and BNCT efficacy, transcriptomic analysis revealed that DOTA-2BB treatment not only induced tumor cell death but also remodeled the tumor microenvironment, including stromal suppression and immune-related pathway activation. This suggests that the therapeutic effects of BNCT using DOTA-2BB may involve both direct cytotoxicity and modulation of the tumor microenvironment.

This study presents a multifunctional boron delivery strategy that integrates tumor targeting, penetration, and high boron payload, resulting in efficient boron accumulation and effective BNCT therapy following intravenous administration. Collectively, these findings demonstrate that DOTA-2BB is a promising candidate for clinical translation in BNCT for pancreatic cancer.

Author contributions

Linger Li, Xin Gao, and Siqi Zhang were responsible for the main experimental design, performance of experiments, and manuscript writing. Xingkai Wang and Xueyao Chen synthesized the BSH molecules. Yuan Miao performed the synthesis and purification of BSH molecules, prepared the therapeutic 2BB molecules, established mouse tumor models, and organized data from confocal microscopy, HPLC, and MS. Tongtong Zhang conducted the omics analysis. Mengjie Yang analyzed HPLC data, performed slice IF staining, and organized data from HPLC, MS, PET/CT, fluorescence, and slices. Can Liu assisted in performing the PET/CT and biodistribution experiments. Tianjiao Liang and Tiancheng Yi carried out the BNCT irradiation. Shuo Jiang co-performed the biodistribution experiments and assisted with labeling. Hongyi Huang performed part of the PET/CT imaging. Jianfei Tong, Rui Wang, and Kuan Hu conceived and designed the study, and revised the manuscript. Kuan Hu supervised the project as the principal investigator.

Conflicts of interest

The authors declare no conflicts of interest.

Appendix A. Supporting information

The following is the Supporting Information to this article.