Coupling Adoptive Cell Therapy with Boron Neutron Capture Therapy: Using Functional Tumor-Infiltrating Lymphocytes for Tumor Delivery of Boron Carbide Nanoparticles

Abstract

Nanomedicine offers promising strategies for targeted drug delivery, imaging, and molecular-level therapies. However, the clinical translation of nanomedicine has often been hindered by the complex interactions of nanoparticles (NPs) with biological systems. This study investigates a cell-based delivery platform designed to overcome some of these limitations, using clinical-grade tumor-infiltrating lymphocytes (TILs) as biological carriers of boron carbide (B₄C) NPs in boron neutron capture therapy (BNCT). Biological vectors, such as TILs, could enable selective tumor targeting, leading to highly localized ¹⁰B levels and minimizing off-target accumulation. We evaluated the uptake and retention of composite Fe₂O₃–B₄C NPs (FeBNPs) using both immortalized Jurkat T cells and primary human TILs. Both cell types efficiently internalized FeBNPs without cytotoxic effects, maintained their functionalities, and retained the boron-rich NPs for up to 72 h. Imaging confirmed intracellular localization, and neutron autoradiography demonstrated that TILs accumulated sufficient ¹⁰B for therapeutic efficacy, eliminating the need for isotopically enriched compounds like L-4-boronophenylalanine (BPA) or sodium borocaptate (BSH). Coculture experiments with Jurkat and HeLa cells confirmed that lymphocyte-delivered boron could mediate localized radiation damage via neutron capture. These findings support the concept of TILs as “Trojan Horses” for boron delivery, allowing for overcoming traditional barriers in NP-based therapies and taking advantage of their innate tumor-homing ability. This approach not only enhances BNCT selectivity and efficacy but also supports the integration of nanomedicine with adoptive cell therapy in a combined cancer treatment framework.

Introduction

Boron neutron capture therapy (BNCT) is an innovative form of radiotherapy that enables selective destruction of malignant cells while sparing surrounding healthy tissues. This two-step treatment relies on the delivery of ¹⁰B-enriched compounds to tumor cells, followed by irradiation with low-energy thermal neutrons. The neutron capture in ¹⁰B produces high-linear energy transfer (LET) particles, an α particle, and a lithium ion, which travel only a short distance (4–10 μm), thereby confining the damage to the immediate surroundings of ¹⁰B and ideally targeting only the tumor cells. Successful BNCT requires two critical conditions: (i) a sufficient concentration of ¹⁰B in tumor tissues (generally 20–35 μg ¹⁰B/g of tumor), and (ii) an adequate flux of thermal neutrons (≥109 cm^–2 s^–1). Recent advancements in accelerator-based BNCT have increased the feasibility of clinical applications, enabling the use of hospital-based neutron sources instead of nuclear reactors. , However, the overall effectiveness of BNCT is still limited by the availability of effective ¹⁰B delivery agents that can specifically target tumor cells with minimal systemic toxicity. , Currently, clinical applications rely on small-molecule boron carriers, such as l-4-boronophenylalanine (BPA) and sodium borocaptate (BSH). , While these compounds are widely used in BNCT, their limitations, including poor tumor specificity, short retention times, and off-target toxicity, have hindered their broader clinical application. Notably, the therapeutic efficacy of BNCT depends not only on the total boron concentration but also on its subcellular localization. , Because DNA is the primary target of high-LET particles, boron agents that reach or accumulate near the nucleus are more likely to induce lethal DNA damage. To tackle these challenges, nanotechnology has emerged as a promising tool for developing next-generation ¹⁰B delivery systems. − Nanoparticles (NPs) can be engineered for controlled biodistribution, enhanced cellular uptake, and extended circulation times. Among these, nanostructures based on boron carbide (B₄C) are especially appealing due to their high boron content, chemical stability, and potential to operate without the need for isotopic enrichment. − However, the effective delivery of NPs to tumors remains a challenging task. Systemic delivery has proven problematic because NPs struggle to pass physiological barriers and effectively reach their targets, even when properly functionalized. This is especially important for treating solid tumors, where NP penetration can also be blocked by the complex tumor microenvironment. To address this, we explored a new strategy that uses the tumor-homing abilities of tumor infiltrating lymphocytes (TILs) as “Trojan horses” to deliver NPs, aiming to combine the principles of adoptive cell therapy (ACT) with that of BNCT. ACT, particularly TIL-based therapy, has shown impressive efficacy in several malignancies, including metastatic melanoma. TILs are isolated from the tumor, expanded ex vivo, and then re-infused into the patient to promote anti-tumor activity. Tracking radiolabeled TILs in patients with metastatic melanoma demonstrated that adoptively transferred, ex vivo-expanded TILs can home to and accumulate at tumor sites. In this study, we propose a hybrid proof-of-concept approach in which TILs are employed not only as effectors, but also as carriers for boron-containing NPs. For this application, we specifically developed composite B₄C–Fe₃O₄ NPs, hereafter referred to as FeBNPs, functionalized with polyacrylic acid (PAA) to enhance their aqueous solubility, biocompatibility, and intracellular delivery. Building on our previous work with B₄C nanoparticles in tumor cell lines, we first assessed their compatibility with the Jurkat T cell line as a model for human lymphocytes and subsequently translated those findings to melanoma-derived TILs. We evaluated boron uptake, cell viability, NP retention over time, and the preservation of TIL functional capacity after loading with NPs. In parallel, we determined the minimum incubation time required for efficient loading and used various imaging techniques, including confocal microscopy, scanning electron microscopy (SEM), and transmission electron microscopy (TEM), to assess intracellular uptake and localization of the nanomaterials. Our goal was to assess the in vitro feasibility of using TILs as effective cellular vectors for targeted boron delivery in BNCT.

Results

FeBNPs Synthesis and Characterization

The nanocomposite consists of B₄C NPs with an average size of 50 nm that are clustered with much smaller Fe₃O₄ NPs. B₄C NPs are inherently hydrophobic and therefore must be functionalized to remain stable in aqueous media. This stability is achieved by the combined presence of iron oxide NPs and PAA. As a capping agent, PAA not only provides steric hindrance to prevent NP agglomeration but also imparts a strong negative surface charge (greater than −25 mV), which is essential for maintaining colloidal stability. B₄C NPs were combined with iron oxide NPs for both structural and functional purposes. As mentioned earlier, B₄C NPs are highly hydrophobic, and the surrounding hydrophilic Fe₃O₄ NPs improve the water stability of the composite nanostructures. The ionic nature of iron oxide surfaces also facilitates an effective capping with PAA. Functionally, adding iron oxide allows for MRI imaging of the NP distribution, potentially giving FeBNPs a theranostic trait. High-resolution TEM (HR-TEM) images demonstrated the colocalization of both phases, showing spherical, ultrasmall iron oxide nanoparticles with diameters ranging from 4 to 6 nm near larger B₄C NPs (Figure A). Figure B shows the XRD of the FeBNPs, presenting only the peaks corresponding to the phase of Fe₃O₄ with a spinel structure, presenting a lattice constant, determined by Rietveld analysis, of 0.837 nm. This lattice parameter falls between those of maghemite (0.83515 nm) and magnetite (0.8396 nm), probably because of an incomplete oxidation of Fe²⁺ to Fe³⁺ during the synthesis. , The B₄C phase is not visible in this pattern because its scattering power is very low compared to the iron oxide, which covers its surface.

1.

Characterization of FeBNPs. (A) HR-TEM pictures of FeBNPs showing the colocalization of the two phases (B₄C and Fe₃O₄ NPs). (B) XRD pattern, peaks corresponding to the Fd-3m space group attributable to Fe₃O₄ NPs. (C) DLS analysis of FeBNPs, average hydrodynamic diameter 55(±13) nm, shown with DLS measurements conducted following FeBNPs incubation in water and in cell medium complete with 10% FBS for 30 min and 2 h. (D) Table showing FeBNPs hydrodynamic diameter measurements conducted in different media.

The peak broadening of the pattern in Figure B allows us to estimate a particle size for Fe₃O₄ of about 3.7 nm (±0.4 nm), as determined using the Debye–Scherrer equation on (220), (311), and (400) peaks. This value aligns with the HR-TEM characterization. The presence of the PAA capping is confirmed by zeta potential measurements, which indicate a negatively charged surface (−36 ± 0.9 mV). Hydrodynamic diameter for the FeBNP was evaluated by DLS (Figure C), conducted in both water and following a 30 min and 2 h incubation in complete cell culture medium at 37 °C, 5% CO₂. Hydrodynamic size increases upon incubation in the complete medium, including FBS (Figure D). All FeBNP suspensions remained stable for several weeks. The evaluation of boron content conducted through inductively coupled plasma-optical emission spectroscopy (ICP-OES) revealed a final ¹⁰B concentration of 45 μg ¹⁰B/mL.

Interaction of FeBNPs with Jurkat Cells

Preliminary tests on the compatibility of FeBNPs with lymphocytes were conducted using the Jurkat cell line. The effect of exposing Jurkat cells to different concentrations of FeBNPs was initially assessed by considering the resulting cytotoxic effects. Cytotoxicity was assessed using the 3-(4,5-dimethylthiazolyl-2)-2,5-diphenyltetrazolium bromide (MTT) assay for cells incubated with FeBNPs for 30 min (TR₃₀) and 2 h (TR₁₂₀). The incubation conditions (FeBNP concentration and time) were chosen based on a previous parametric study. , Additionally, long-term cell survival was also assessed through pulse-chase experiments. In this case, Jurkat cells viability was evaluated after a chase of 24 h in fresh media without FeBNPs (TR_30+24h and TR_120+24h). In all cases, no significant cytotoxicity was observed for any FeBNP concentration (Figure A,B). The interaction of FeBNPs with Jurkat cells was also examined using flow cytometry, SEM, confocal microscopy, and TEM. Quantitative information about FeBNP uptake, whether internalized or on the cell surface, was obtained through flow cytometry (Figure C). In this case, Jurkat cells were incubated with a fluorescent variant of FeBNPs (TR₃₀ and TR₁₂₀), which included the fluorophore DiI obtained using a procedure described previously. The figure indicates that the signal associated with the presence of FeBNPs in Jurkat cells increases with both incubation time and FeBNP concentration. This result was corroborated by SEM (Figure D) and confocal microscopy (Figure E) observations. SEM analysis demonstrated a strong adhesion of FeBNP to the Jurkat cell membrane. Images in Figure D were acquired with a backscatter electron detector (BS), which enhanced the contrast between the Fe₃O₄ present in the FeBNPs and the biological material, which consists only of light atoms. In these images, the FeBNPs appear as light agglomerates attached to the external membrane of the cells. Their position is indicated by the red arrows. From the SEM backscattered (SEM-BS) images, it can be appreciated the increase in FeBNP adhesion onto Jurkat cells membrane when the incubation time is increased from 30 min (TR₃₀) to 2 h (TR₁₂₀). Confocal microscopy confirmed these observations (Figure E), while also providing evidence that the FeBNPs can persist inside the cells for at least 24 h, as shown by the images from the pulse-chase experiments (TR_30+24h and TR_120+24h). This information is critical, as the ability of Jurkat cells to uptake FeBNPs and retain them for at least 24 h is essential for future in vivo applications, where cells loaded with NPs must circulate in the organism for several hours before reaching malignant cells. TEM analysis was conducted to reveal the details of the interaction between the FeBNPs and the cells. The results are shown in Figure F. Compared to control cells (CTRL), after 30 min of incubation (TR₃₀), FeBNPs adhere to the cell membrane and microvilli, while clathrin-coated pits allow the internalization of the NPs. After 2 h of incubation (TR₁₂₀), the concentration of NPs attached to the cell membrane and microvilli increases, and NPs can also be found in both the cell endosomes and autophagosomes, confirming the successful incorporation of the NPs. To quantify the amount of ¹⁰B taken up by Jurkat cells due to their interaction with the FeBNPs, the total boron content of the cells was determined by ICP-OES.

2.

Evaluation of FeBNPs interaction with Jurkat cells. (A) Cell viability of Jurkat cells incubated with different concentrations of FeBNPs for 30 min (TR₃₀), and 2 h (TR₁₂₀). (B) Cell viability of Jurkat cells in pulse-chase conditions. Cells were incubated with different concentrations of FeBNPs for 30 min and 2 h (pulse) plus 24 h chase (TR_30+24h and TR_120+24h). (C) Cytofluorimetric analysis results showing fluorescence of adhered and/or internalized DiI-functionalized FeBNPs within Jurkat cells. Fluorescence intensity increases with both increasing FeBNPs concentration (1.5, 2.25, and 4.5 μg ¹⁰B/mL) and incubation time (30 min and 2 h). (D) SEM-BS images of control cells (CTRL) and cells incubated with FeBNPs (4.5 μg ¹⁰B/mL) for different incubation times (TR₃₀ and TR₁₂₀). The presence of adhering FeBNPs is indicated by the red arrows. (E) Confocal microscopy showing FeBNP interaction and persistence with Jurkat cells. Upper row, cells not incubated with FeBNPs (CTRL) and cells incubated with FeBNPs for 30 min (TR₃₀) and 2 h (TR₁₂₀). Lower row, persistence of FeBNP after chase: cells not incubated with FeBNPs (CTRL₊₂₄) and cells incubated with FeBNPs for 30 min and 2 h (pulse) following by 24 h chase (TR_30+24h and TR_120+24h). Each picture is a merge of 25 confocal planes. (F) TEM images of Jurkat cell not incubated with FeBNPs (CTRL), with magnification of the boxed area; cells incubated for 30 min with 4.5 μg ¹⁰B/mL (TR₃₀) in cell culture media, with magnification of the boxed area showing FeBNPs being internalized through clathrin-coated pits; cells incubated for 2 h with FeBNPs (TR₁₂₀), with magnification of boxed areas showing FeBNPs adhering to the cells microvilli, membrane, and present within pits, autophagosomes and endosomes. N: nuclei, M: mitochondria, rER: rough endoplasmic reticulum, G, Golgi, A: autophagosomes, E: endosomes.

The amount of ¹⁰B was evaluated, considering that it represents 19.9% (¹⁰B isotopic abundance) of the total boron amount determined by the ICP-OES analysis. The results showed that Jurkat cells incorporated up to 80 μg ¹⁰B/g (ppm), which is significantly higher than the typical boron concentration achieved for treatment, thus representing promising vectors for delivering a useful boron concentration in the tumor (Table S1). , It is important to note that this result was obtained using a boron source that was not isotopically enriched in ¹⁰B.

The intracellular localization and the distribution of ¹⁰B across the cell population were also investigated using intracellular neutron autoradiography. Using the technique reported by Portu et al., it is possible to reveal not only the tracks generated by the secondary particles (⁴He, ⁷Li) produced by neutron capture reactions, but also the original disposition of the cells and their internal structure, allowing to obtain the distribution of ¹⁰B within the cells. The particle-produced tracks appear as pits on the solid-state nuclear track detectors (SSNTD) generated by the high LET secondary particles (latent tracks) that are made visible by a subsequent chemical attack. The cellular imprints, on the other hand, are generated by exposing the SSNTD to a UV-C source in the presence of the cells stained with hematoxylin (Figure S1). Figure A shows optical microscopy images from intracellular neutron autoradiography conducted on Jurkat cells. In these images, the particle tracks (red arrows) appear as darker, round pits of approximately 1 μm in diameter, situated within or near the original cell locations (green arrows). Comparing the image of control cells (CTRL) with that of cells loaded with 4.5 μg ¹⁰B/mL, it is evident that increasing the incubation time from 30 min (TR₃₀) to 2 h (TR₁₂₀) results in a rise in ¹⁰B uptake, as indicated by the increase in track density.

3.

Intracellular neutron autoradiography of Jurkat cells. (A) Optical and (B) SEM microscopy images of Jurkat cells imprints and tracks on the SSNTD. Upper row: cells control sample (CTRLabsence of FeBNPs) and cells incubated with 4.5 μg ¹⁰B/mL for 30 min (TR₃₀), and 2 h (TR₁₂₀). Lower row: cells control sample (CTRLabsence of FeBNPs) and cells incubated with FeBNPs for 30 min and 2 h followed by 24 h chase (TR_30+24h and TR_120+24h). For track acquisition, images of optical microscopy were taken when tracks appeared as dark round pits of about 1 μm in size, but because cell imprints and tracks can only be focused separately, this might result in blurring of cell imprints. Samples were irradiated with a 10¹³ n/cm² fluence.

These results agree with ICP measurements, which indicated in all cases an average content higher than 30 μg ¹⁰B/g of cells (Table S1). The ¹⁰B cellular uptake could not be determined based on the number of tracks observed. This procedure, commonly used in the case of borate molecules, cannot be applied here, as the high track density, which leads to the formation of clusters, prevents tracks from being individually distinguished. SEM analysis confirmed these observations. Figure B shows the SEM images acquired for the same incubation conditions of Jurkat cells reported in Figure A. The track overlaps are very evident. Nevertheless, both techniques confirmed the significant ¹⁰B uptake within the Jurkat cells, as well as the FeBNPs persistence following the chase phase (TR_30+24h and TR_120+24h).

Interaction of FeBNP-Loaded Jurkat Cells with HeLa Cells

Preliminary indications on the ability of FeBNP-loaded lymphocytes to deliver a dose of radiation via their ¹⁰B load have been obtained using cocultures of FeBNP-loaded Jurkat cells (carrier cells) with HeLa cells (target cells).

The SEM images in Figure A show that after 45 min of coculture, Jurkat cells can adhere to HeLa cells, both in the absence of FeBNPs (control sample) and in their presence. Notably, FeBNP-loaded Jurkat cells maintained their ability to interact with HeLa cells even when their membrane was heavily coated with FeBNPs (purple arrows). FeBNP persistence was also verified following the chase phase (TR_120+24h). In some instances, FeBNPs were observed on the surface of the HeLa cells (green squares), possibly due to the transfer of FeBNPs from the Jurkat cell surface. Intracellular neutron autoradiography results from the 2 h incubation with FeBNPs (TR₁₂₀) are presented in Figure B. The tracks resulting from the distribution of FeBNPs within the two cell types significantly overlap, indicating that the secondary particles produced by the neutron capture reaction in FeBNPs-loaded Jurkat cells can reach HeLa cells (being the combined range of charged secondary products of neutron capture equal to 14 μm). FeBNPs can also eventually be transferred from Jurkat cells to tumor cells. Overall, this experiment, which mimics in vivo lymphocyte-tumor cell interactions, confirms the feasibility of using FeBNPs delivered through lymphocytes to target specific cells. Specifically, it confirmed that radiation dose can be delivered either by the proximity of the lymphocyte to the tissues or by the transfer of nanoparticles.

4.

Characterization of Jurkat-HeLa cocultures. (A) SEM-BS images of Jurkat cells adhering to HeLa cells in the absence (CTRL) of FeBNPs. In the presence of FeBNPs (TR₁₂₀ and TR_120+2h) Jurkat cells appear covered in FeBNP. In the case of TR_120+24h, the presence of free FeBNPs adhering to HeLa cells can be appreciated (green squares). (B) Intracellular neutron autoradiography. Optical microscopy pictures of intracellular neutron autoradiography results of HeLa-Jurkat cells cocultures, following the 2 h loading of Jurkat cells with 4.5 μg ¹⁰B/mL in cell culture media. ROIs are first identified, and pictures were acquired at 40X (left). Jurkat cells appear smaller (about 10–15 μm of diameter) and round, strongly stained by hematoxylin given the large size of their nuclei when compared to the overall volume of the cell; HeLa cells are bigger, adherent, and present a lighter staining when compared to the Jurkat cell line, given the high volume of cytoplasm compared to the size of their nuclear region. Jurkat cells are highlighted by blue contours, HeLa cells by green contours. In the middle, an optical microscopy picture of cell imprints corresponding to the previously identified ROI is shown. On the right, the same area is shown in a SEM image. Tracks localized within Jurkat cell imprints are indicated by blue arrows, while tracks localized within the HeLa cell imprints are highlighted by green arrows. Enlargements in the top right corner are reported in the top right corner. The presence of tracks can be observed in both cell line imprints, indicating that secondary particles reach HeLa cells when FeBNPs are delivered through lymphocytes.

Interaction of FeBNP with TILs

Building on preliminary findings obtained with Jurkat cells, we applied the same NP loading strategy to REP-expanded TILs derived from patients with metastatic melanoma. To this end, TILs were incubated with FeBNPs (1.5 μg ¹⁰B/mL) for 30 min (TR₃₀) or 2 h (TR₁₂₀). To assess the persistence of NP associated with cells over time, we performed pulse-chase experiments like those previously conducted in Jurkat cells. Specifically, TILs were incubated with FeBNPs (pulse, TR₃₀ and TR₁₂₀), then washed and cultured in fresh NP-free medium for up to 72 h (chase, TR_{30+24h/48h/72h} and TR_{120+24h/48h/72h}). Flow cytometry analysis confirmed NP loading and viability. Viability was assessed using LIVE/DEAD staining (Figure A,B), and NP retention was evaluated using DiI-labeled FeBNPs (Figure C,D). TILs not incubated with FeBNPs (CTRL) were included and maintained under identical culture conditions. As shown in Figure B, normalized viability was calculated based on the percentage of LIVE/DEAD-negative cells (doublet-excluded), with the viability of CTRL at the end of pulse set to 100% (see Section Materials and Methods). No significant viability changes, compared to CTRL, were detected up to 72 h for both TR₃₀ and TR₁₂₀ across three independent patient-derived TIL products (#TIL1, #TIL2, #TIL3). FeBNP retention was monitored by measuring the FeBNP fluorescence intensity in live-gated cells (Figure C–G). The FeBNP signal increased proportionally with incubation time, being higher in TR₁₂₀ than TR₃₀, and absent in CTRL (Figure D). Fluorescence intensity remained stable over time for both TR₃₀ and TR₁₂₀ (Figure E). These trends were consistent across the three independent TIL samples under 72 h chase conditions (Figure F). The GeoMFI ratio summarized in Figure G further confirmed comparable FeBNP retention across conditions and time points, without significant differences across time points within the TR₃₀ and TR₁₂₀ conditions. To further dissect the distribution of NP uptake, dot plot analyses allowed us to distinguish negative (−), positive (+), and highly positive (++) TIL populations, confirming that more than 70% of cells consistently displayed positive-to-high FeBNP signals across all three TIL products (Figure S2A). Additionally, the reproducibility of FeBNPs uptake was confirmed using three independent batches of FeBNPs, each tested at both dilutions on the same TIL product (#TIL3), across three separate experiments. Fluorescence intensity was measured at the end of the pulse (TR₃₀ and TR₁₂₀) and after a 24 h chase (TR_30+24h and TR_120+24h), consistently confirming the retention of the NPs (Figure H). This batch-to-batch comparison further demonstrated that the frequency of positive TILs (+ and ++) remained above 70% regardless of the FeBNP batches used, supporting the robustness of the loading protocol (Figure S2B).

5.

NP-loaded TILs remain viable and retain fluorescent FeBNPs over time. (A) Gating strategy used to identify live TILs (LIVE/DEAD-negative). (B) Line graph showing viability of each TIL product (#TIL1, #TIL2, #TIL3), expressed as normalized viability, with CTRL TILs at time 0 set to 100%. (C) Gating strategy for assessing live TILs retaining DiI-labeled FeBNPs. (D) Representative histogram plots of FeBNP fluorescence intensity (PE channel) comparing TR₃₀ (30 min FeBNP loading; light red), TR₁₂₀ (2 h FeBNP loading; light blue), and CTRL (gray). (E) Fluorescence intensity histograms showing FeBNP PE signal retention in TR₃₀ (left) and TR₁₂₀ (right) across all time points. (F) Histogram plots showing DiI-FeBNP retention at 72 h postloading in TR₃₀ (TR_30+72h, left) and TR₁₂₀ (TR_120+72h, right) across the three TIL products. (G) Graph showing ΔGeoMFI (PE) values at each time point for treated versus CTRL TILs. Different shapes represent individual TIL products: #TIL1 (square), #TIL2 (triangle), and #TIL3 (circle). (H) Histogram plots of DiI-FeBNP (PE) signal for CTRL, TR₃₀, TR₁₂₀, and after 24 h chase (CTRL_24h, TR_30+24h, and TR_120+24h), in #TIL3, following loading with three independent FeBNP batches. Statistical analysis was performed using two-way ANOVA with Sidak’s correction for multiple comparisons; only statistically significant differences are reported.

To support these findings with morphological evidence, SEM was performed on TR₃₀ and TR₁₂₀ TILs, revealing successful NP association (Figure A). The SEM analysis indicated that TILs, for both incubation times, showed FeBNPs adhering to their cell membranes. SEM allows for the assessment of a certain degree of variability in the extent of FeBNPs associated with the cells, with some TILs being heavily coated by the NPs, while others exhibit only a minimal amount. By increasing the incubation time to 2 h, it can be observed that most TILs now exhibit FeBNPs adhering to their surface (Figure A, green arrows), while others show a reduction of FeBNPs adhesion on their membrane, but with more NPs internalized (Figure A, red arrows). Figure also shows SEM-BS images of TILs that were incubated with FeBNPs for 30 min and 2 h followed by 24 h chase in fresh media without nanoparticles (TR_30+24h and TR_120+24h). The images indicate that cells can retain nanoparticles, either adhering to the external surface or internalized. Many cells still show nanoparticles attached to their surface, indicating a strong interaction. The interaction between FeBNPs and TIL was also examined using confocal microscopy with fluorescent nanoparticles. Figure B presents confocal microscopy images of FeBNPs-loaded TILs, confirming successful interaction, as indicated by the red fluorescence signal corresponding to the DiI-functionalized FeBNPs. Pulse-chase experiments confirm the persistence of the FeBNPs in TILs over a 24 h chase period. Ultimately, intracellular neutron autoradiography was performed to evaluate the microdistribution of ¹⁰B within the TIL population and the retention of nanoparticles after the pulse-chase experiment. Under all incubation conditions of 30 min and 2 h, both optical microscopy and SEM-BS images of TILs cell imprints (green contours) on the SSNTD detector revealed that significant amounts of ¹⁰B were effectively delivered (Figure C,D), as indicated by large track clusters (red arrows) localized within the imprints.

6.

Characterization of efficient TIL loading with FeBNPs. SEM, confocal, and intracellular neutron autoradiography analyses were conducted on TILs not incubated with FeBNP (CTRL), incubated with FeBNPs for 30 min (TR₃₀) or 2 h (TR₁₂₀), and incubated for 30 min and 2 h (pulse) followed by a 24 h chase (TR_30+24h and TR_120+24h). (A) SEM-BS. TILs with adhering FeBNPs are visible in both TR₃₀ and TR₁₂₀; the longer incubation time shows an increase in the number of cells with FeBNPs compared to those without NPs (appearing light gray). After a 2 h incubation, there is a significant rise in the number of TILs interacting with FeBNPs. While some TILs show no NPs, most are positive for the presence of nanoparticles. When testing for FeBNP persistence, both TR_30+24h and TR_120+24h show cells with NPs on their surface or internalized. Magnifications in the second column show details and characteristics of nanoparticle-loaded TILs. Some nanoparticles can be seen adhering to the cell membrane, while others appear to be located beneath the membrane, indicating internalization. Note that the difference in brightness correlates with NPs’ localization: high intensity indicates NPs on the surface, while low intensity shows NPs beneath the surface. (B) Confocal microscopy pictures of TILs loaded with fluorescent FeBNPs. Nuclei are stained with Hoescht, and membranes are stained with MemGlow. Bar, 3 μm. (C) Optical and (D) SEM microscopy images of intracellular neutron autoradiography. Cell imprints are contoured in green. Tracks left on the SSNTD appear as round, black pits (1 μm). Presence of tracks and track clusters (red arrows) is shown for FeBNPs-loaded TILs. Both TR_30+24h and TR_120+24h result in FeBNPs retention within the TILs. Bars in SEM images insets, 10 μm.

The retention of FeBNPs following the pulse-chase experiment was confirmed, corroborating the results of confocal microscopy. The significant uptake of ¹⁰B observed in TILs, as evidenced by intracellular neutron autoradiography, was confirmed by ICP-OES analysis (Table S1). The fact that neutron irradiation products can also reach neighboring cells, and the evidence that FeBNPs are transferred to surrounding tumor cells, both support the concept of BNCT mediated by immune cells.

Evaluation of Functional Competence in FeBNP-Loaded TILs

After confirming effective FeBNP loading in melanoma-derived REP-TILs, we next evaluated whether NP-functionalized TILs maintained their functional capacity, particularly in response to mitogenic stimulation. To this end, both FeBNP-treated TILs and CTRL were stimulated with either anti-CD3/CD28 beads or allogeneic dendritic cells (alloDCs) in combination with OKT3. After 24 h, TIL activation was assessed by multiparametric flow cytometry.

Figure A outlines the gating strategy: debris, doublets, and dead cells were excluded, and the analysis focused on CD3⁺ T cells. Considering that TIL products can exhibit variable proportions of CD4⁺ and CD8⁺ T cells, both NP-functionalized and control TILs maintained comparable CD4⁺ and CD8⁺ subset distributions after stimulation (Figure B). Evaluation of early activation markers CD25 and CD69 (Figure C,D) showed no significant differences in the percentages of CD25⁺ and CD69⁺ cells within the CD3⁺ gate among TR₃₀, TR₁₂₀, and CTRL conditions, within each specific stimulation condition. Importantly, the preservation of TIL functionality was confirmed using three independent FeBNP batches tested on the same REP-TIL product (Figure E). To further investigate the impact of FeBNPs on effector function, cytokine production was assessed using a secretome profiling assay after 24 h of stimulation. Levels of four cytotoxic T cell-associated analytes were measured and reported as fold change relative to the pg/mL values of CTRL. Except for TNF-α in #TIL2 (which remained undetectable across CTRL, TR₃₀, and TR₁₂₀ before stimulation), fold changes for IFN-γ, TNF-α, and IL-2 were close to or exceeded 1 across all stimulated conditions in both TR₃₀ and TR₁₂₀ samples. In contrast, GZMB levels were consistently reduced in both unstimulated and stimulated conditions following FeBNP treatment (Figure F).

7.

NP-loaded TILs retain functional responsiveness upon stimulation. (A) Gating strategy applied for TIL flow cytometry analysis. (B) Proportions of CD4⁺ and CD8⁺ cells within live, CD3⁺ gated populations. (C) Histogram plots showing CD25 (right) and CD69 (left) expression in live CD3⁺ TILs (#TIL2) after 24 h stimulation with anti-CD3/CD28 beads (bead:T ratio 1:25; filled histograms) or with allogeneic dendritic cells plus OKT3 (alloDC +OKT3) (alloDC:T ratio 1:3; open histograms), compared to unstimulated controls (dotted-line histograms). (D) Violin plots showing the percentage of CD25⁺ and CD69⁺ T cells across three TIL products: #TIL1 (square), #TIL2 (triangle), and #TIL3 (circle). (E) Histogram plots showing CD25 and CD69 expression in #TIL3 following stimulation with anti-CD3/CD28 beads, upon upload with three different FeBNP batches. (F) Graphs showing fold change in TIL-derived soluble factors (TNF-α, IFN-γ, IL-2, GZMB) calculated by dividing the concentration (pg/mL) in TR₃₀ or TR₁₂₀ samples by that in CTRL samples under each stimulation condition. #TIL1 (square), #TIL2 (triangle), and #TIL3 (circle). Statistical comparisons were performed using two-way ANOVA with Sidak’s correction for multiple comparisons; only statistically significant differences are indicated.

Evaluation of CXCR3 Expression and Chemotactic Function in FeBNP-Loaded TILs

TILs are known for their intrinsic tumor tropism, a key feature supporting their potential use as boron carriers for targeted BNCT. This chemotactic ability is largely mediated by the expression of chemokine receptors, particularly CXCR3, the receptor for CXCL10 and CXCL11, which has been shown to play a fundamental role in tumor homing, including melanoma. − As expected, CXCR3 expression on TILs may vary due to differences in their functional orientation and the influence of the applied in vitro expansion protocols. CXCR3 is typically associated with CD4⁺ and CD8⁺ effector T cells exhibiting a Th1 profile and, consequently, potential antitumor activity. We observed CXCR3 expression in two out of three TIL products, #TIL2 and #TIL3, with #TIL2 showing the highest levels (Figure A). As previously reported, CXCR3 expression declined upon activation and was more prominent in resting TILs (data not shown). Further analysis revealed that CXCR3 was predominantly expressed on CD3⁺ CD4^– T cells in #TIL2, while in #TIL3 it was mainly restricted to the CD3⁺ CD4⁺ subset (Figure B). To assess whether FeBNP upload affected chemokine receptor expression, we evaluated CXCR3 levels under TR₃₀ and TR₁₂₀ conditions using #TIL2 as a representative TIL product. No differences were observed in the percentage of CXCR3⁺ cells between the treated and untreated conditions (Figure B). To further assess the preservation of functionality, we evaluated the chemotactic behavior of NP-functionalized TILs. Using #TIL2 as a model system, we tested their ability to respond to a CXCL10 chemotactic gradient within the MIVO millifluidic organ-on-chip platform (Figure C). Semiquantitative analysis demonstrated that FeBNP-loaded TILs retained a robust migratory capacity, moving upward (against gravity) through an “extravasation-like” process in response to the CXCL10 gradient (Figure D). These findings confirm that TILs retain their functional properties and homing potential after FeBNP loading, further supporting their suitability as cellular carriers for targeted BNCT.

8.

Assessment of CXCR3^high TIL chemotaxis in the MIVO system. (A) Overlay histograms of CXCR3 expression in live CD3⁺ cells across three independent TIL products (left), and dot plots of CXCR3 versus CD4 expression in the CXCR3⁺ subsets of #TIL2 and #TIL3 (right). (B) CXCR3 expression in CTRL, TR₃₀-, and TR₁₂₀-treated #TIL2. (C) Schematic representation of the chemotaxis assay in the MIVO organ-on-chip platform. A 3D alginate scaffold was functionalized with rhCXCL10 and maintained in the MIVO device, while CFSE-labeled CXCR3^high TILs circulated under physiological flow for 24 h. (D) Quantification of migrated CXCR3^high TILs and NP-loaded (TR₁₂₀) CXCR3^high counterparts (NP_CXCR3^high TILs) toward the functionalized hydrogel chamber. Migration was assessed by counting cells in at least five different regions of interest (ROIs), and results are expressed as the mean number of migrated TILs per ROI. Dots indicate individual experimental replicates. Statistical comparisons were performed using one-way ANOVA with Dunnett’s correction for multiple comparisons.

Discussion

This study demonstrates that FeBNPs can be efficiently loaded into both immortalized Jurkat cells and patient-derived TILs at therapeutically relevant concentrations of ¹⁰B. Thus, the potential of this method for NP delivery in general and boron delivery in BNCT specifically is confirmed. We previously reported that FeBNPs can be taken up to substantial ¹⁰B concentrations in the HeLa cell line. This inorganic boron-containing nanomaterial based on B₄C presents the advantage of delivering high amounts of boron locally without depending on expensive and hard-to-obtain isotopically enriched compounds. Additionally, they are chemically inert and can stay in cells longer compared to borated molecules like BPA, which are metabolized quickly. We have demonstrated that FeBNPs can be retained within cells for up to 72 h, indicating that these B₄C-based NPs possess favorable intracellular stability. The ability of FeBNPs to be easily internalized by TILs opens the possibility of using these biological carriers to deliver boron-rich NPs and other types of NPs to tumor tissues. Although cellular endocytosis is very effective at facilitating NPs uptake, we observed that it also causes some inhomogeneity in the uptake of NPs within the cell population. To enhance homogeneity, alternative NP-loading methods, such as mechanoporation and electroporation, could be explored in the future. These methods have proven effective in boosting intracellular NPs uptake by creating temporary pores in the membrane through alterations in permeability, achieved with short electrical pulses or mechanical forces, like membrane deformation or microfluidic constriction. ,

Coculture experiments using FeBNP-loaded Jurkat cells along with tumor HeLa cells provided a proof of concept that the T cells can serve as delivery vehicles for localized NPs treatment, allowing to bring the boron-loaded NPs in proximity to the target cells. Unexpectedly, neutron autoradiography also evidenced the presence of tracks generated by neutron capture within HeLa cells following neutron irradiation, suggesting that the transfer of NPs from the carrier to the target cells can also happen. These FeBNPs observed on HeLa cells could result from the transfer of material between cargo and target cells through temporary direct contact. Although this passive process seems most likely, other more active mechanisms, like trogocytosis, could also be considered. Vesicle-mediated transfer appears unlikely, although it might depend on the type and duration of contacts. However, regardless of the mechanism responsible for the transfer of FeBNPs on the target cells, which might be more prominent in in vitro conditions, we believe that the effectiveness of our approach mainly depends on the proximity of cargo to target cells, facilitated by cell type recognition, and is less reliant on FeBNPs transfer to the target cells, especially in in vivo conditions.

To assess clinical relevance, we shifted from Jurkat cells to GMP-grade, ex vivo expanded melanoma TILs. Also in this case, SEM analysis confirmed the successful adhesion of FeBNPs to the TIL membrane, and confocal microscopy validated their internalization. Importantly, FeBNP loading did not affect TIL viability, CD4/CD8 composition, or the activation potential of TILs. Upon mitogenic stimulation, treated TILs retained the ability to upregulate key activation markers (CD25, CD69), indicating that TCR and costimulatory pathways remain functional. Previous studies have explored boron delivery using phagocytic immune cells such as macrophages or dendritic cells; however, these cell types lack inherent tumor specificity unless genetically modified (e.g., via CARs). TILs, on the other hand, naturally target tumor tissue, providing a high tumor-to-normal cells ratio, which is a crucial feature for BNCT success. Tumor infiltration by TILs is often mediated by chemokine receptors such as CXCR3. We demonstrate that CXCR3 expression was not affected by NPs loading, and this was corroborated by chemotaxis experiments in the MIVO system, where NP-loaded CXCR3^high TILs retained their ability to migrate toward CXCL10-functionalized scaffolds. Moreover, the superparamagnetic iron oxide NPs present in FeBNPs could offer a secondary advantage: enabling noninvasive MRI tracking of boron-loaded cell distribution. Interestingly, although FeBNP-loaded TILs preserved viability and cytokine production (IFN-γ, TNF-α, IL-2), a significant reduction in GZMB secretion was observed. This suggests that while proximal TCR signaling and general effector function remain active, the production of cytotoxic granules may be selectively impaired. Since internalization occurred via clathrin-mediated endocytosis, it is plausible that NP uptake, depending on size and chemical properties, could saturate or interfere with trafficking pathways required for cytotoxic granule biogenesis or secretion. This functional alteration, although relevant in the context of conventional ACT, may be less important for BNCT, where TILs act primarily as “Trojan horses” to deliver boron rather than exerting direct cytotoxicity. Importantly, the potential synergy between ACT and BNCT may justify repeated TIL infusions. Indeed, serial administrations of CD8⁺ T cells have been shown to improve tumor infiltration, suggesting a strategy in which standard and NP-loaded TILs are combined to enhance both immunologic and radiologic tumor targeting. Despite the promise of ACT, its efficacy in solid tumors is often limited by immune escape mechanisms. Indeed, while still present and reactive within the tumor tissue, the therapeutic index of TILs could be weakened by numerous resistance mechanisms, such as the alterations of the major histocompatibility complex (MHC) class I antigen-processing and presentation machinery of tumor cells and T cell exhaustion. , Incorporating BNCT offers a complementary, immune-independent cytotoxic mechanism that can overcome some of these limitations. We therefore can envisage a dual-modality approach that couples antigen-specific immune recognition with localized radiation-induced tumor damage. This strategy may enhance therapeutic outcomes by addressing both immune-sensitive and immune-resistant tumor subclones.

Moreover, it is well established that TIL antitumor reactivity varies across tumor histologies, with the most robust clinical benefits to date observed in melanoma, while efficacy in other solid tumors remains limited. This integrated approach with BNCT could help broaden the therapeutic applicability of TIL-based treatments beyond melanoma by adding a second, nonimmunologic cytotoxic mechanism. Importantly, this study enabled the optimization of key technical parameters, such as FeBNP concentration, incubation time, and cellular uptake conditions, establishing a reproducible protocol for loading human T cells with ¹⁰B while preserving their functionality. Additional work will be necessary to evaluate the therapeutic effectiveness of this approach. However, to obtain a preliminary indication that BNCT mediated by boron-enriched immune cells, acting as a biological carrier, is effective, we irradiated a coculture composed of a tumor cell line and a 1:1 mixture of Jurkat cells that had been previously administered FeBNP, as described above. Cells were irradiated in a well-characterized thermal neutron field with a typical neutron fluence employed in BNCT radiobiological studies. MTT tests were performed after 96 h, showing that irradiation with neutrons without FeBPN in Jurkat cells resulted in cell vitality comparable to that of the nonirradiated control (as expected, given the low radiation dose due to the delivered fluence). Cells irradiated in the presence of FeBNP-loaded Jurkat cells displayed a vitality reduced to 50% (Table S2). Although still preliminary, these results indicate that boron delivered by Jurkat cells effectively induces measurable and significant cell death in the target cells.

Future studies will focus on coculture models involving boron-loaded TILs and tumor targets exposed to neutron irradiation, to quantify the resulting cytotoxic effects by measuring clonogenicity as a function of the absorbed dose. This will allow comparisons with standard BNCT radiobiological results, obtained by directly administering boron to the tumor cell cultures. In vivo models will also be crucial to assess tumor homing, boron delivery efficiency, and therapeutic outcomes within the context of intact immune and vascular systems. Overall, our findings provide strong evidence of concept for using FeBNP-functionalized TILs as active delivery vectors of ¹⁰B in BNCT, supporting further development of this platform to enhance the effectiveness of adoptive cell therapy in solid tumors.

Conclusions

This work explored a new concept of BNCT, proposing the use of a biological vector for composite boron-containing NPs. Our goal was to enhance the selectivity and eventually the efficacy of BNCT by using tumor-primed lymphocytes delivering a suitable boron concentration into the malignancy. This method aims for virtually zero boron concentration in surrounding healthy tissues. This is a paradigm shift that, once tested in vivo, could enable significantly more effective neutron irradiation. Currently, in clinical BNCT, the radiation dose is prescribed to the most radiosensitive healthy tissue within the irradiation field. By achieving a negligible boron concentration in the surrounding tissues of tumors, a higher neutron fluency can be provided, thereby ensuring a better tumor control probability. In summary, this study provides the first compelling preclinical evidence supporting the use of TILs as intelligent delivery vectors for boron-based cancer radiotherapy. We demonstrate that boron carbide-based NPs can be effectively loaded into both immortalized and patient-derived lymphocytes, achieving stable intracellular ¹⁰B levels without impairing T cell viability and migration. We also included a preliminary proof that boron contained in the vector cells is responsible for a sensitive decrease in tumor cell viability. By combining the tumor specificity of TILs with the high cytotoxic potential of BNCT, this approach offers a promising strategy to enhance the efficacy of adoptive T-cell therapies in solid tumors, thus overcoming some of the key barriers of current immunotherapies. This dual-action therapeutic platform holds potential to reshape the landscape of targeted cancer treatment. In more general terms, this study confirms the suitability of T cells as a platform for delivering nanoparticles to specific targets, thereby avoiding physiological barriers.

Materials and Methods

FeBNP Synthesis and Characterization

Composite nanoparticles were synthesized starting from commercial B₄C nanopowders (Iolitec Nanomaterials, Heilbronn, Germany). The synthesis of composite nanoparticles followed a procedure reported previously. For high resolution (HR)-TEM analysis, the NPs were dispersed in ethanol and then deposited by drop casting onto an ultrathin carbon film. The HR-TEM imaging was performed by a FEI Tecnai F20 microscope, operating with an acceleration voltage of 200 kV, and equipped with a Schottky electron source, an S-TWIN objective lens, and a Gatan Orius CCD camera.

XRD was carried out using a Bruker D2 diffractometer (Bruker Corp., Billerica, MA, USA) equipped with a copper X-ray source. We used a θ–θ′ configuration, with a step size of 0.03° in 2θ and an acquisition time of 2 s for each step. The hydrodynamic diameter of the NPs was determined by DLS using a Nano ZS90 DLS analyzer (Malvern Instruments, Malvern, UK). The suspension was diluted in water to obtain a concentration of about 1 mg/mL. Three measurements were acquired, each including 11 runs.

Commercial Cell Lines and Media

HeLa cells, derived from cervical cancer, were grown in DMEM high-glucose medium, added with 10% fetal bovine serum and 2 mM glutamine (all from Euroclone, Italy). Jurkat cells, human T lymphocyte cell line, were grown in Roswell Park Memorial Institute (RPMI) medium, with 10% FBS, 2 mM glutamine, and streptomycin. All cell lines were maintained at 37 °C in a 5% CO₂ humidified atmosphere.

FeBNP Cell Loading

For FeBNPs cell loading, Jurkat or TIL cells were incubated at a density of 3 × 10⁵ cells/mL with FeBNPs in complete cell culture medium at different concentrations (4.5, 2.25, and 1.5 μg ¹⁰B/mL) and incubation times (30 min, TR₃₀; 2 h, TR₁₂₀) specified on a case-by-case basis. After incubation, the cells were pelleted, washed twice with PBS to remove excess nanoparticles, and resuspended for subsequent analyses. Pulse-chase experiments were performed following TR₃₀ and TR₁₂₀ for an additional 24 h without FeBNPs (TR_30+24h and TR_120+24h). In the case of TIL, the chase period was also extended to 48 and 72 h (see below).

HeLa–Jurkat Coculture

HeLa cells were seeded at a density of 75,000 cells/mL. Meanwhile, Jurkat cells were incubated with FeBNPs. At the end of the incubation, the cells were washed, resuspended in PBS, and then seeded with HeLa cells. After 45 min in PBS, the cells were fixed for further analysis.

Cytotoxicity Assay

Cells exposed to FeBNPs (concentrations and incubation times as specified in Section Results) were washed and incubated with MTT (3-(4,5-dimethylthiazolyl-2)-2,5-diphenyltetrazolium bromide) at a concentration of 0.5 mg/mL in complete medium for 90 min. After several washings, DMSO was added to facilitate the dissolution of the formazan crystal. Absorbance at 570 nm was measured using the plate reader (CLARIOstar Plus, BMG Labtech, Ortenberg, Germany).

Flow Cytometry

For Jurkat cells, a BD FACSLyric flow cytometer was used. At least 30,000 events were acquired for each sample. Cells were identified based on forward scatter (FSC) and side scatter (SSC) parameters. The emission from the FeBNPs was stimulated with the blue laser, and fluorescence was detected using the band-pass filter (590/40 nm). Data analysis was performed using FlowJo software v10.10 (BD Biosciences).

Scanning Electron Microscopy

At the end of the treatments, Jurkat cells and TILs were pelleted and resuspended in PBS to achieve a density of 10⁶ cells/mL, then seeded in poly lysinated 35 mm Petri dishes to promote adhesion. A similar procedure was applied for the Jurkat–HeLa coculture, with HeLa cells already plated in the Petri dish. After 10 min at room temperature, the PBS was discarded, and the cells were fixed with 4% glutaraldehyde for 2 h at room temperature. The samples were left to dry, then attached to the stub using conductive graphite tape and coated with graphite via high vacuum evaporation. Analyses were performed with a TESCAN Mira 3 XMY microscope (TESCAN ORSAY HOLDING s.a, Czech Republic), using a backscattered detector (SEM-BS) to visualize NPs.

Confocal Microscopy

Jurkat cells and TILs were centrifuged at the end of FeBNP incubation, resuspended in PBS, and seeded on 18 × 18 mm coverslip glass. After 45 min, the cells were fixed with 4% formaldehyde in PBS for 15 min at room temperature, and the nuclei were stained with Hoechst 333342. A Leica TCS SP8 confocal microscope, equipped with PL APO 40x/1.25 NA or 63x/1.40 NA objectives, was used.

Transmission Electron Microscopy

FeBNPs-treated cells were fixed in 2% glutaraldehyde in cacodylate buffer 0.1 M, pH 7.3, for 20 min at room temperature and then centrifuged for 1 min at 300g. The cell pellets were further fixed for 24 h at 4 °C and postfixed for 1 h in a 2% osmium tetroxide water solution. After several washings in bidistilled water, the pellets were placed in a 1% uranyl acetate solution and left for 1 h at room temperature. Samples were dehydrated using an increasing ethanol series (70%, 80%, 90%, 100%, and acetone) for 10 min each and then embedded in Epon resin. After curing at 60 °C for 48 h thin sections were obtained by cutting the embedded samples using an UltraCut E Ultramicrotome (Reichert) and placed on 300-mesh copper grids. These grids were then examined using a Talos L120C G2 transmission electron microscope (Thermo Fisher Scientific Inc. Waltham, MA, USA) working at an acceleration voltage of 120 kV, equipped with a lanthanum hexaboride thermionic source and a bottom-mount Ceta Thermofisher 4kx4k CMOS camera.

Boron Uptake Determination

The quantification of boron uptake from Jurkat cells and TILs was performed using ICP-OES analysis. Jurkat cells were grown in T10 flasks and diluted to obtain a concentration of 6 × 10⁵ cells/mL. Jurkat cells were incubated with 1.5 or 4.5 μg ¹⁰B/mL added directly to the medium. Following incubation, the suspension was centrifuged at 180g for 5 min, and the supernatant was discarded. After washing, the cells were resuspended in 1.5 mL of PBS and then digested in 6 mL of a 1:1 = HNO₃/H₂SO₄ mixture in 50 mL Teflon vessels. Digestions were performed using the Mars Microwave at 220 °C for 15 min at a pressure of 80 bar. After digestion, 790 mg of mannitol was added to each sample to prevent the loss of volatile boron compounds. The samples were diluted, and the method of standard additions was performed.

Intracellular Neutron Autoradiography

To perform intracellular neutron autoradiography and conduct ¹⁰B microdistribution studies, cells were grown over SSNTD (Lexan polymer), which were then subjected to neutron irradiation, as described by Portu et al. This technique enables the simultaneous observation of cellular distribution, as shown by imprints formed on the SSNTD by UV exposure, and ¹⁰B distribution, revealed by tracks generated by secondary particles (⁴He, ⁷Li) produced by neutron capture upon irradiation of the sample. The tracks appear as pits in the polymer resulting from the damage caused by the energy deposition of high-LET secondary particles, which disrupt the polymeric chains. These nanometric latent tracks can be visualized using optical microscopy after a treatment with a strong alkaline solution. The procedure for the intracellular neutron autoradiography can be divided into four steps: cell attachment and fixation, irradiation, cell imprint formation, and latent track development. After treatments with NPs, cell pellets were resuspended in PBS and seeded onto poly­(lysine)-coated SSNTD. After 1 h, the PBS was removed, and the cells were fixed with 4% glutaraldehyde for 15 min at room temperature. The fixed cells were washed three times with absolute ethanol, dried, and stored at room temperature. Samples were irradiated at the RA3 reactor in Ezeiza (Argentina), exposed to two different neutron fluences of 10¹² cm^–2 and 10¹³ cm^–2, or at the Triga Mark II reactor of LENA (Laboratory of Applied Nuclear Energy) in Pavia (Italy) with a fluence of 10¹³ cm^–2. To develop cellular imprints, cells were first stained with hematoxylin for 15 min. This is a nuclear stain that absorbs UV light and protects SSNTD from photodegradation. Once the samples were stained and dried, they were exposed to UV-C radiation with an average wavelength of 254 nm for 5 min using a 15 W, TUV G15T8 lamp (Philips, Holland). The irradiance (5.7 mW/cm²) was measured using a radiometer. (International Light Technologies ILT77). This process created cellular imprints on SSNTD, deriving from Fries photodegradation and photooxidation of the SSNTD polymer in areas not protected by the stained cells. After UV-C sensitization, the organic tissues were removed by trypsinization, and the SSNTD underwent chemical etching with PEW solution (consisting of 30 g of KOH, 80 g ethyl alcohol, and 90 g distilled water) for 4 min at 70 °C. This treatment enlarged the latent tracks, making them visible by optical microscopy. Photodegradation and photooxidation processes during UV-C sensitization can lead to track fading. Saint Martin et al. estimated that track fading accounts for 27%, indicating that the number of tracks observed represents 73% of the actual number of tracks had the SSNTD not been exposed to UV-C radiation. Images were acquired using an optical microscope (Olympus BX51), coupled to a CCD camera (Olympus DP70).

Melanoma Patient-Derived TIL Expansion

The protocols allowing TIL isolation and expansion from tumor specimens were approved by the institutional Medical Ethical Review Board (Protocol IRSTB099, cod. L3P1919, Approval date from the Romagna Ethics Committee (CEROM) 13/03/2019 and Protocol IRSTB137, cod. L3P2767, Approval date from the Romagna Ethics Committee (CEROM) 14/03/2024) and the study was conducted in accordance with the ethical standards laid down in the 1964 Declaration of Helsinki and later versions. Written informed consent was obtained from all patients. Fresh tumor specimens were obtained from patients with metastatic melanoma undergoing standard-of-care surgical procedures. Tumors were dissected from adjacent tissues and cut into multiple fragments (1–3 mm³ each) using a scalpel. TIL isolation and expansion were performed according to the original protocol by Dudley et al. and subsequent adaptations. − In the initial phase, tumor fragments or digested tumor tissue were cultured in 10M-GREX vessels (5 to 8 fragments per vessel) for 3 weeks in serum-free TexMACS medium (Miltenyi Biotec) supplemented with recombinant human interleukin-2 (rhIL-2, 6000 IU/mL; Proleukin). This supported the outgrowth of young TILs (Y-TILs), which were then pooled, counted, and cryopreserved as needed. Subsequently, a 14 day rapid expansion protocol (REP) was carried out in GREX100 M vessels using either freshly generated or cryopreserved Y-TILs (5 × 10⁶ cells per vessel). Cells were activated with functional grade anti-CD3 antibody (OKT3, 30 ng/mL; Miltenyi Biotec) and cocultured with allogeneic irradiated (40 Gy) peripheral blood mononuclear cells (PBMCs) at a 1:100 TIL/PBMC ratio in serum-free TexMACS medium supplemented with rhIL-2 (3000 IU/mL). REP-TILs were harvested between day 14 and 21, counted, and cryopreserved in aliquots for downstream applications. All procedures were approved by the Romagna Ethics Committee (CEROM), and written informed consent was obtained from all patients before any procedure, in accordance with the Declaration of Helsinki. Table S3 illustrates the clinical and experimental characteristics of the three-melanoma patient-derived TIL products used in this study.

Preparation of NP-Loaded TILs

Cryopreserved REP-TILs were thawed in Iscove’s modified Dulbecco’s medium (IMDM) supplemented with 10% fetal bovine serum (FBS) and DNase I (Sigma-Aldrich). Viable cells were counted using Trypan Blue exclusion. For NP loading, TILs were incubated with FeBNPs to obtain ¹⁰B concentration in cell culture media equivalent to 1.5 μg ¹⁰B/mL for either 30 min (TR₃₀) or 2 h (TR₁₂₀). Specifically, 6 × 10⁶ cells were seeded in T75 flasks in 10 mL of IMDM + 10% FBS, and incubated at 37 °C, without agitation. After incubation, the cells were centrifuged at 1300 rpm for 10 min at room temperature (RT), and the supernatant was discarded. The pellets were then resuspended in PBS. Following a second centrifugation under the same conditions, cell pellets were resuspended in TexMACS medium (Miltenyi Biotec). Cell concentrations were reassessed, and NP-loaded TILs were used in downstream in vitro assays. Untreated control TILs (CTRL), resuspended at the same concentration but not incubated with NPs, were processed in parallel.

Chase Experiment with Fluorescent NP-Loaded TILs

To monitor NP retention and T cell viability over time, a fluorescent variant of FeBNP that incorporates the DiI fluorophore was used. After incubation, TR₃₀ and TR₁₂₀ REP-TILs were washed twice in PBS, resuspended in TexMACS medium supplemented with rhIL-2 (50 IU/mL), and plated at 1 × 10⁶ cells per well in 1 mL in 48-well plates. Flow cytometry was performed at 24, 48, and 72 h postloading. At each time point, cells were harvested and stained with the LIVE/DEAD Fixable Yellow Dead Cell Stain Kit (Thermo Fisher Scientific, Carlsbad, CA, USA), following the manufacturer’s protocol. Fluorescence signals corresponding to DiI (PE channel, excited by the yellow laser YL1) and the viability stain (excited by the violet laser VL3) were acquired using an Attune NxT Flow Cytometer (Thermo Fisher Scientific), equipped with 405 nm (violet), 488 nm (blue), 561 nm (yellow), and 637 nm (red) lasers. Data were analyzed using FlowJo software v10.10 to quantify both the percentage of NP-positive cells and viable (LIVE/DEAD-negative) cells over time, comparing TR₃₀ and TR₁₂₀ conditions. Viability over time was calculated with the following formula, setting 100 the viability of CTRL cells at Time 0 (which coincides with the end of pulse for TR₃₀ e TR₁₂₀)­

$$\mathrm{n}\mathrm{o}\mathrm{r}\mathrm{m}\mathrm{a}\mathrm{l}\mathrm{i}\mathrm{z}\mathrm{e}\mathrm{d}\mathrm{v}\mathrm{i}\mathrm{a}\mathrm{b}\mathrm{i}\mathrm{l}\mathrm{i}\mathrm{t}\mathrm{y}(\%)=\left(\frac{{\mathrm{v}\mathrm{i}\mathrm{a}\mathrm{b}\mathrm{i}\mathrm{l}\mathrm{i}\mathrm{t}\mathrm{y}}_{\mathrm{s}\mathrm{a}\mathrm{m}\mathrm{p}\mathrm{l}\mathrm{e},t}}{\mathrm{v}\mathrm{i}\mathrm{a}\mathrm{b}\mathrm{i}\mathrm{l}\mathrm{i}\mathrm{t}{\mathrm{y}}_{\mathrm{C}\mathrm{T}\mathrm{R}\mathrm{L},t=0}}\right)\times 100$$

FeBNP loading was measured as geometric mean fluorescence intensity (GeoMFI) on the PE channel and is reported as Delta GeoMFI, calculated using the following formula

$$\mathrm{\Delta }\mathrm{G}\mathrm{e}\mathrm{o}\mathrm{M}\mathrm{F}\mathrm{I}=\frac{\mathrm{G}\mathrm{e}\mathrm{o}\mathrm{M}\mathrm{F}{\mathrm{I}}_{\mathrm{T}{\mathrm{R}}_{30}\mathrm{o}\mathrm{r}\mathrm{T}{\mathrm{R}}_{120}}}{\mathrm{G}\mathrm{e}\mathrm{o}\mathrm{M}\mathrm{F}{\mathrm{I}}_{\mathrm{C}\mathrm{T}\mathrm{R}\mathrm{L}}}$$

GeoMFI values were obtained at each time point to assess the relative uptake of NPs in treated versus untreated TILs.

TIL Functional Assay

To evaluate the responsiveness of FeBNP-functionalized TILs to mitogenic stimulation, cells were cultured in the presence of either CD3/CD28 Dynabeads (Thermo Fisher Scientific) or allogeneic dendritic cells (alloDCs) plus soluble anti-CD3 antibody (OKT3, 1 μg/mL). For CD3/CD28 stimulation, TILs (2.5 × 10⁵ cells/well in 200 μL) were incubated with Dynabeads at a final bead-to-T cell ratio of 1:25. For the allogeneic DC condition, TILs were cocultured at a 3:1 TIL/DC ratio in the presence of soluble OKT3 (10 ng/mL final concentration). In both conditions, functionalized (TR₃₀ and TR₁₂₀) and CTRL TILs were tested in parallel in TexMACS medium without serum and IL-2 supplementation. After 24 h, activation markers and cytokine production were assessed to evaluate the functional competence of TILs following NP loading.

TIL Phenotypic Analysis

Flow cytometry was carried out using an Attune NxT Flow Cytometer (Thermo Fisher Scientific). A minimum of 30,000 events per sample was recorded. Lymphocytes were identified based on forward scatter (FSC) and side scatter (SSC) parameters, both acquired on a linear scale. Data analysis was performed using FlowJo software v10.10 (BD Biosciences). The antibody panel included CD3-VioBlue (clone REA613, Cat. No. 130-144-519), CD4-VioBright FITC (clone REA623, Cat. No. 130-113-229), CD8-APC-Vio700 (clone REA734, Cat. No. 130-110-819), CD25-PE-Vio615 (clone REA945, Cat. No. 130-115-631), CD69-APC (clone REA864, Cat. No. 130-112-614), all purchased from Miltenyi Biotec. CXCR3-PE (clone CEW33D, Cat. No. 12-1839-41) was purchased from Invitrogen (Thermo Fisher Scientific). Dead cells were identified using the LIVE/DEAD Fixable Yellow Dead Cell Stain Kit (Thermo Fisher Scientific) and excluded from all analyses.

Cytokine Analysis by Ella Automated Immunoassay System

Cell culture supernatants were collected 24 h after stimulation to quantify Interferon-γ (IFN-γ), Interleukin-2 (IL-2), Granzyme B (GZMB), and Tumor Necrosis Factor-α (TNF-α). Levels of these analytes were measured using the Ella Automated Immunoassay System (Bio-Techne, Minneapolis, MN, USA), with a 1:10 sample dilution performed according to the manufacturer’s instructions. Results were expressed as fold change relative to the CTRL condition and calculated using the following formula

$$\mathrm{f}\mathrm{o}\mathrm{l}\mathrm{d}\mathrm{c}\mathrm{h}\mathrm{a}\mathrm{n}\mathrm{g}\mathrm{e}=\frac{{\mathrm{c}\mathrm{y}\mathrm{t}\mathrm{o}\mathrm{k}\mathrm{i}\mathrm{n}\mathrm{e}}_{\mathrm{T}{\mathrm{R}}_{30}\mathrm{o}\mathrm{r}\mathrm{T}{\mathrm{R}}_{120}}}{\mathrm{c}\mathrm{y}\mathrm{t}\mathrm{o}\mathrm{k}\mathrm{i}\mathrm{n}{\mathrm{e}}_{\mathrm{C}\mathrm{T}\mathrm{R}\mathrm{L}}}$$

MIVO Organ-on-Chip-Based TIL Migration Assay

The chemotactic activity of CXCR3^high TILs (#TIL2) toward recombinant human (rh) CXCL10 was evaluated using the MIVO millifluidic organ-on-chip platform (React4life S.p.A., Genoa, Italy), designed to reproduce the complexity of a dynamic 3D microenvironment. To prepare CXCL10-containing hydrogels, rhCXCL10 was mixed with a 0.5% (w/v) alginate solution to reach a final concentration of 2.5 μg/mL. Aliquots of 100 μL (corresponding to 250 ng of CXCL10) were dispensed into 24-well inserts with 8 μm pores and cross-linked in 0.5 M CaCl₂. After gelation, hydrogels were washed with sterile physiological solution and transferred into MIVO chambers for dynamic coculture. CXCR3^high TILs were counted, stained with the CFSE Cell Proliferation Kit (ThermoFisher Scientific), resuspended in TexMACS medium, and loaded into the circulation of the MIVO device (2 × 10⁶ cells/MIVO). The platform was connected to a pumping system delivering a physiological flow rate (0.3 mL/min), allowing continuous recirculation of immune cells and mimicking tumor-like circulatory dynamics. After 24 h of culture within the MIVO system, the ability of CXCR3^high TILs to migrate upward (against gravity) through an “extravasation-like” process (Figure C), following the CXCL10 chemotactic gradient, was assessed by fluorescence microscopy. Semiquantitative analysis was performed by counting migrated cells in ≥5 different ROIs (Regions of Interest), and results were expressed as the mean number of migrated TILs per ROI. As a negative control, CXCR3^high TILs were cultured in MIVO chambers with alginate hydrogels lacking rhCXCL10. To investigate the impact of NP conjugation on migration, a second set of experiments was conducted by perfusing either CXCR3^high TILs or CXCR3^high NP-loaded TILs (TR₁₂₀) into the MIVO system in the presence of rhCXCL10-containing hydrogels. In all experiments, CXCR3^high TILs cultured in MIVO with alginate hydrogels lacking rhCXCL10 served as a negative control.

Statistical Analysis

Statistical significance between groups was assessed by two-way ANOVA with Sidak’s multiple comparisons test, or by one-way ANOVA with Dunnett’s correction for multiple comparisons. A p-value < 0.05 was considered statistically significant. All analyses were performed using GraphPad Prism v10.4.2 (GraphPad Software, Inc.). Statistical tests applied are provided in the figure legends.

All the data supporting the findings of this study are available within the article and its Supporting Information files and from the corresponding author upon request. The source data underlying Figures D–F, A–C, B,G, D,F, D and S2 and Tables S1 and S2 are provided as a Source data file. Graphical Abstract was created using BioRender (https://biorender.com/).

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acsnano.5c12640.

• Neutron irradiation of tumour cellslymphocyte co-cultures and workflow description of intracellular neutron autoradiography (Figures S1). Analysis of percentage of NP-uptake by TILs (Figure S2).Tables S1–S3: Table of ¹⁰B quantification in Jurkat cells and TILs; Table of cell viability following neutron irradiation; Table of melanoma patient-derived TIL clinical data (PDF)

• Source data (XLSX)

⊥⊥.

M.T. and U.A.-T. are cosenior authors. M.P.D., P.S., G.P., S.B., M.T., and U.A.-T. contributed to the conceptualization of the study. M.P.D., P.S., N.R., A.P., M.G., I.P., A.C., A.F., M.T., and U.A.-T. were involved in conducting experiments/performing acquisition. M.P.D., P.S., and M.T. drafted the original manuscript, while S.B., U.A.-T., and G.P. contributed to its review and editing. M.E.F.P and S.S. conducted TIL migration experiments. F.T. was responsible for clinical sample collection. Supervision was provided by M.T. and U.A.-T. Funding acquisition and provision of resources were supported by S.B., M.T., and U.A.-T. Visualization, formal analysis, and data curation were performed by M.P.D. and M.T. All authors read and approved the final manuscript.

This work was partially funded by the National Plan for NRRP Complementary Investments (PNC, established with the decree-law 6 May 2021, n. 59, converted by law n. 101 of 2021) in the for the funding of research initiatives for technologies and innovative trajectories in the health and care sectors (Directorial Decree n. 931 of 06-06-2022)project n. PNC0000003AdvaNced Technologies for Human-centrEd Medicine (project acronym: ANTHEM), through its cascade call GIOCONDA. This work reflects only the authors’ views and opinions, neither the Ministry for University and Research nor the European Commission can be considered responsible for them.

The authors declare no competing financial interest.