Introduction

The development of radiotheranostics, which combine imaging and targeted radionuclide therapy, has experienced significant growth in recent years, driven by the success of products such as Lutathera® and Pluvicto®1,2. Therefore, many radiopharmaceuticals are currently in preclinical stages or clinical trials3. One of the key aspects of radiopharmaceutical development is the choice of radiolabeling strategy. Among the various approaches, chelator-based methods are prevalent in clinical settings. In this method, the selection of a suitable chelator is key for radiopharmaceutical development because it affects both the type of isotopes that can be used and the synthetic design to combine diagnosis and therapy into a single compound. Consequently, the development of adaptable chelators that can incorporate various radiometals, each with different radiochemical properties, is a key objective with longstanding unmet challenges4,5.

Ideally, a chelator for a radiotheranostic pharmaceutical should satisfy the following criteria: i) rapid and quantitative binding to very low concentrations of the radiometal, even in the presence of metal impurities; ii) high stability to retain the radiometal in vivo, preventing unwanted accumulation of the detached radiometal in healthy and non-target tissues; iii) straightforward bioconjugation with vectors to achieve active targeting; and iv) meeting these conditions for a large number of radioisotopes to simultaneously cover imaging and therapeutic responses. This would ensure the application of a unique platform that covers all areas of nuclear medicine regardless of the biomolecules or radioisotopes used. To date, no universal chelating agent has been developed4, with best candidates being able to chelate just a few radiometals6.

When using traditional chelating agents to create radiopharmaceuticals, favorable thermodynamics are crucial for driving the formation of complexes under conditions in which both chelator and metal ion concentrations are extremely low. Polydentate chelators typically form more thermodynamically stable complexes than monodentate analogs. Once circulating in vivo, radiopharmaceuticals are further diluted, and endogenous biomolecules at higher concentrations may compete for binding with the radiometal, whereas endogenous metal ions may compete for binding with chelators. Even the most thermodynamically stable radiometal-chelator complexes can dissociate under these conditions unless the radiolabeled metal complex exhibits high kinetic stability.

With the growing number of radioisotopes suggested for use in imaging and/or therapeutic purposes, the conventional chelator method fails to produce appropriate compounds at a comparable rate7. Although many chelator approaches have been successfully used in vivo, each possesses unique characteristics in terms of the radiolabeling process, in vivo performance, bioconjugation, and a combination of these factors. For example, in the case of radiometals such as gallium-68 (68Ga) and indium-111 (111In), radiochemists have focused on making radiolabeling as simple as possible, emulating the ease of preparing 99mTc complexes but overcoming the tendency of both Ga(OH)3 (above pH 3) and In(OH)3 (above pH 3.5) to precipitate from the solution, which complicates the process8. For copper-64 (64Cu), there is no straightforward answer as to which chelator is the best; although positive results have been shown using sarcophagine9, the development of novel chelation systems for 64Cu with long-term in vivo stability is an active area of research. For instance, traditional radiocopper-labeled complexes of 1,4,7,10-tetraazacyclodecane-1,4,7,10-tetraacetic acid (DOTA) and triethylenetetramine (TETA) have been shown to be relatively unstable in vivo8. Even for clinical radioisotopes such as zirconium-89 (89Zr), the situation can be further improved. The stable complexation of 89Zr is critical because Zr4+ is biologically inert and entrapped within the cell once internalized, complicating patient dosimetry. Current complexation with deferoxamine (DFO) is the gold standard for 89Zr; however, decomplexation remains an important concern because of the moderate thermodynamic stability of the Zr-DFO complex. Furthermore, the most commonly used DFO variant for protein conjugation is a bifunctional variant containing a phenylisothiocyanate group. Initial findings indicate that the thiourea bond formed between isothiocyanate and lysines is vulnerable to ionizing radiation from the radioisotope, causing bond breakage and release of the chelator/radiometal complex10. This, together with a cumbersome radiolabeling protocol, makes it necessary to look for new alternatives. The situation is even more complicated for radiometals such as thallium-201 (201Tl) and radium-223 (223Ra), for which the number of suitable bifunctional chelators is very limited11,12. Similarly, actinium-225 (225Ac) is a radiometal of interest in targeted alpha therapy because of its distinctive cascade of alpha emissions and clinical half-life. However, realizing its full potential requires managing the “recoil effect”, which is a significant hurdle in 225Ac utilization. The recoil effect refers to the possibility that during alpha decay, the energy released by 225Ac can break the direct bond between Ac and its chelator, releasing daughter isotopes into the body. This recoil may lead to the unintended distribution of isotopes, such as francium-221 (221Fr), which can circulate and accumulate in non-target tissues, causing unwanted radiation exposure13.

Nanoparticles are an attractive alternative to traditional chelators. In particular, iron oxide nanomaterials have been proposed in combination with various radioisotopes14. This approach can be divided into two categories: those using a chelator on the surface and those using non-chelator strategies. The nanoparticle-chelator approach is a versatile platform with valuable in vivo results15,16. However, as a platform intended to accommodate a wide range of radioisotopes it may suffer from the same problems as the traditional chelator approach, in addition to some new issues, such as the selectivity of the radiometal for the chelator versus the coating of the nanoparticle, which may lead to non-specific radiolabeling of the nanoparticle. The second alternative, non-chelator labeling, can be performed using several approaches, such as chemical adsorption, radio-mineralization, or the use of hot and cold precursors in nanoparticle synthesis for radiometal core doping17. Overall, non-chelator approaches show several advantages, and among them, the core doping approach stands out, particularly in terms of the stability of labeling17. Numerous examples of nanoparticle–radiometal pairs have been reported using the core-labeling approach, including 89Zr, germanium-69 (69Ge), arsenic-71 (71As), 64Cu, 111In and 68Ga14,18. However, each nanoparticle differs from the others in terms of reaction conditions, physicochemical properties, bioconjugation protocols, and in vivo behavior. In summary, the situation is similar to that for traditional chelators, that is, a range of nanoparticles with similar physicochemical and in vivo properties that cannot be used for several radiometals.

Scheme 1 summarizes the concept of this study. This figure presents a comparison between the traditional labeling approach using radiometals, which uses a wide range of chelating agents for different isotopes, as no single chelator meets all criteria across the periodic table. Each chelator-radiometal pair has its own specific characteristics, protocols, and in vivo performance, and in some cases (such as 223Ra), lacks a gold standard for in vivo use. In contrast, Scheme 1b shows our strategy: incorporating any of the selected radiometals directly into the nanoparticle core in a single, rapid step under uniform conditions, followed by the same purification process, to ensure consistent in vivo stability and biodistribution

Scheme 1 : Radiolabeling approaches.
scheme 1

a Classical labeling approach for radiopharmaceuticals, in which a large number of chelators are proposed. Each chelator shows different labeling and bioconjugation conditions. Despite the wide variety of available chelators, several radiometals still lack an appropriate chelator for effective in vivo applications. b Labeling approach proposed in this study, where one platform with the same physicochemical properties, bioconjugation protocols, and in vivo behavior was used with 10 different radiometals.

Full size image

Based on this, we propose the use of very small iron oxide nanotracers (IONT) as a candidate close to this universal chelator. Based on previous findings from our group, we hypothesized that IONT could incorporate a wide range of radiometals into their core with excellent radiolabeling stability. In this study, we aimed to demonstrate that the core doping versatility, combined with the inherent properties of IONT, makes it a highly adaptable multiplexed probe. To test our hypothesis, we selected 10 radiometals for nuclear imaging and therapy applications: 68Ga, 64Cu, and 89Zr for PET; 99mTc, 201Tl, 111In, and 67Ga for SPECT; and lutetium-177 (177Lu), 223Ra, and 225Ac for therapy. Finally, we conducted five different in vivo proof-of-concept experiments to demonstrate their potential for multimodal imaging, intratumoral therapy, passive and active bioorthogonal targeting, and renal clearance.

Results

Synthesis of xxM-IONT

We employed microwave-assisted (MW) synthesis to obtain small nanoparticles with a core size of ~3 nm, as previously described by our group19,20. This method is particularly appealing in radiochemistry because it ensures safe and contained handling of radioactive samples with short reaction times (10 min) in a highly reproducible manner. After choosing the methodology, the general protocol for the one-step synthesis and radiolabeling of the nanoparticles was as follows: MilliQ water, iron (III) chloride hexahydrate, sodium citrate, and the radiometal of choice were mixed in a high-pressure MW-designed vial containing a magnetic stirrer, and then hydrazine hydrate (55%) was added to the vial. The mixture was heated at 100 °C for 10 min, cooled, and purified using pre-packed columns for size-exclusion chromatography. The general scheme is shown in Fig. 1a. These reaction conditions enable the synthesis of small, uniform nanoparticles with a thick organic layer that ensures colloidal stability under physiological conditions. Using this protocol, we synthesized 10 IONT with the general formula xxM-IONT, where xxM denotes the selected radioisotopes: 68Ga, 64Cu, 89Zr, 99mTc, 201Tl, 111In, 67Ga, 177Lu, 223Ra, or 225Ac (Fig. 1a), and IONT to the nanoparticles formed by a maghemite core coated with citrate molecules.

Fig. 1: Synthesis and characterization of IONT.
figure 1

a General synthetic scheme for the simultaneous synthesis of iron oxide nanoparticles and core incorporation of radiometals coated with a thick organic layer of citrate molecules of approximately 7 nm and a core of approximately 3 nm. In a high-pressure MW-designed vial, milli Q water, iron (III) chloride hexahydrate, sodium citrate, and the radiometal of choice were mixed, and then hydrazine hydrate was added to the vial. The mixture was heated at 100 °C for 10 min, cooled, and purified using pre-packed columns for size-exclusion chromatography. Using this protocol, 10 IONT were synthesized using the selected radioisotopes: 68Ga, 64Cu, 89Zr, 99mTc, 201Tl, 111In, 67Ga, 177Lu, 223Ra, or 225Ac; b Radiolabeling yield for the synthesis of each xxM-IONT, calculated as the percentage of the activity in the purified sample divided by the total activity (sample activity + purification column activity) measured at a given time, each synthesis was repeated at least three times, and the mean and standard deviation are shown; c Average hydrodynamic size for each synthesized xxM-IONT, the mean and standard deviation are shown (N = 3), the results showed the expected size range (average 11.0 ± 0.6 nm) with excellent reproducibility; The radiochemical stability was evaluated at different times, depending on the radiometal half-life (t1/2), by measuring the RCS as the percentage of activity that remained in the IONT after three ultrafiltration steps with PBS 1X. d Radiochemical stability in human serum experiments for radiometals with a short half-life (N = 3). RCS after 1 and 3 h for 68Ga-IONT and after 1, 3, and 24 h for 64Cu-IONT and 99mTc-IONT; e radiochemical stability in human serum experiments for radiometals with a long half-life (N = 3). RCS was measured after 24, 48, 72, and 168 h, except for 201Tl.

Full size image

The experimental details for each radiometal are shown in the Supporting Information, explaining the differences in each radioisotope, such as generator/cyclotron production, conversion of the precursor to a more reactive chemical form, or quenching of the salts present in the precursor compound. The characterization of the different xxM-IONTs started with the assessment of the radiolabeling yield (RLY), calculated as the percentage of activity in the purified sample divided by the total activity (sample activity + purification column activity) measured at a given time. The results obtained for the different radiometals are shown in Fig. 1b. The RLY values were higher than 80% for most radiometals, with only 64Cu and 99mTc showing values below 50%. Moreover, the RLYs were consistent for all radiometals, with a standard deviation of less than 5%, confirming the reproducibility of the methodology (Table S1). The anticipated lower RLY for 64Cu can be attributed to the complex reactions that occur when Cu, hydrazine hydrate, and citric acid are combined and heated. This mixture can yield various complexes between hydrazine and copper (i.e., hydrazine acts as both a reductant and complexing agent, forming stable complexes), including copper(I) complex (N2H4)CuCl, mixed-valence copper complex (N2H5)2Cu2Cl6, and copper(II) complexes (N2H5)2CuCl4 2H2O and (N2H5)CuCl321. Moreover, similar conditions have been reported for the synthesis of Cu(0) nanoparticles22. This chemistry explains that the availability of free 64Cu2+ is reduced in comparison to the other radiometals demanding natural “cold” Cu to increase the RLY (see supp. info, 64Cu-IONT). Similarly, 99mTc complexes have been reported with oxidation states ranging from -1 to +723. The diversity of oxidation states is affected by several factors, such as the pH, type and strength of the reducing agent, and nature of the coordinating ligand. For example, some of these oxidation states (e.g., Tc(IV) or Tc(V)) are unstable or reactive for integration into iron oxide. Additionally, these Tc ions form stable citric acid complexes, which can clearly affect the final RLY, with citric acid being the main component of the IONT24. All of these factors make a complicated mixture, limiting the availability of 99mTc for incorporation in the core of the IONT.

The hydrodynamic diameters of all xxM-IONTs were measured in water after decay. The results showed the expected size range (average 11.0 ± 0.6 nm) with excellent reproducibility (Fig. 1c and Table S1), consistent with reported studies19,25,26. The radiochemical stabilities (RCS) of the radiolabeled IONT were tested over time using human serum samples. 150 µL of radiolabeled IONT were incubated with 150 µL of human serum at 37 °C. The stability was evaluated at different times, depending on the radiometal half-life (t1/2), by measuring the RCS as the percentage of activity that remained in the IONT after three ultrafiltration steps with PBS 1X. Due to the differences in the time points analyzed, the results are shown separately depending on their t1/2 in short half-life (<24 h, Fig. 1d) and medium/long half-life (>24 h, Fig. 1e). The RCS was ≥ 80% for the 10 xxM-IONT. This is particularly relevant for samples radiolabeled with 223Ra and 225Ac, because their application in targeted radiotherapy is currently hindered by the lack of efficient chelators that can provide sufficient stability. Using this methodology, both 223Ra-IONTs and 225Ac-IONTs showed high RLYs ( ≥ 85%) and an unprecedented RCS ( ≥ 97%) after 7 days in serum, providing a straightforward and reproducible solution to overcome the longstanding stability concerns of these therapeutic radiometals. Interestingly, even though the RLY values for 64Cu and 99mTc were lower, their hydrodynamic sizes and RCS values matched those of other radiometals. This highlights a crucial aspect of this study: the chemical platform remains consistent regardless of the radiometal used.

Following the key characteristics of the nano-radiolabeling approach, we selected five radiometals for additional evaluation. Our selection comprised two PET radiometals (68Ga and 89Zr), two SPECT radiometals (67Ga and 99mTc), and one therapeutic radiometal (177Lu). The inclusion of 99mTc has an additional reason: it allows the testing of these features for radiometals with a lower RLY. First, we analyzed the samples using high-angle annular dark-field scanning transmission electron microscopy (HAADF-STEM). (Fig. 2a–e and Figs. S1–S5). The nanoparticle size and shape remained consistent across all radiometals (Fig. 2f), mirroring the consistency observed for the other parameters examined. The images revealed very small nanoparticles with a core size of 3.7 ± 0.3 nm for 68Ga-IONT; 3.8 ± 0.4 nm for 67Ga-IONT; 3.8 ± 0.5 nm for 89Zr-IONT; 4.1 ± 0.4 nm for 99mTc-IONT and 3.9 ± 0.5 nm for 177Lu-IONT and a quasi-spherical shape, independent of the radiometal used.

Fig. 2: Characterization and radiolabeling in challenging conditions.
figure 2

a–e Representative HAADF-STEM images of 68Ga-IONT, 89Zr-IONT, 67Ga-INOT, 99mTc-IONT, and 177Lu-IONT after radioactivity decay (scale bar is 20 nm); f Average core size values are 3.7 ± 0.3 nm for 68Ga-IONT; 3.8 ± 0.4 nm for 67Ga-IONT; 3.8 ± 0.5 nm for 89Zr-IONT; 4.1 ± 0.4 nm for 99mTc-IONT and 3.9 ± 0.5 nm for 177Lu-IONT (N = 10, One-way ANOVA) data are shown as mean plus standard deviation; g RLY for 68Ga-IONT, 89Zr-IONT, 67Ga-INOT, 99mTc-IONT at different activity ranges: 100 – 400 μCi for 68Ga, 15 – 65 μCi for 89Zr, 40 – 210 μCi for 99mTc, 30–140 μCi for 67Ga and 25 – 220 μCi for 177Lu (N = 3, One-way ANOVA with Geisser-Greenhouse correction); and h RLY for 68Ga-IONT, 89Zr-IONT, 67Ga-INOT, 99mTc-IONT in the presence of 500 μM concentration of “impurities” of Zn in each synthesis (N = 3, Welch’s t test).

Full size image

A crucial factor in developing effective chelators for radiopharmaceuticals is their radiolabeling performance under challenging conditions, such as low concentrations of radiometals or the presence of metal impurities. We aimed to assess the performance of our approach under these demanding conditions. To evaluate this, we initially tested our synthetic protocol using the very low activities of the five selected radiometals. As illustrated in Fig. 2g, the results demonstrated that the quantity of radiometal used did not affect the RLY of our protocol, even in the case of 99mTc. To evaluate our approach in the presence of metallic impurities, we challenged our methodology by adding a substantial quantity (500 μM) of Zn2+, a common metallic impurity for many radioisotopes, and investigated its effect on RLY. Under these conditions, we did not observe any significant effect on RLY when challenged with Zn impurities (Fig. 2h).

Synthesis and multimodal application of xxM/xxN-IONT

Building on the positive results shown in the previous section, we further aimed to test the use of IONT as a general platform for simultaneous dual labeling in a single step. We hypothesized that it would be possible to perform double-core doping of IONT, synthesizing a nano-radiomaterial with two different radioisotopes simultaneously incorporated, which could, for example, provide a signal for two imaging techniques or combine imaging and therapeutic features. To achieve this goal, we synthesized the following double-doped nanotracers: 68Ga/177Lu-IONT, 68Ga/67Ga-IONT, and 67Ga/177Lu-IONT. The same synthetic protocol was followed, but with simultaneous addition of the two radiometals to the starting materials. The characterization of the labeling efficiency was performed in a similar way as for single doping, and the results are shown in Fig. 3 and Table S2. Due to the use of two radiometals, there is a major difference in assessing the labeling efficiency of these IONT, that is, the difficulty in simultaneously measuring two radiometals with distinct energy profiles in a dose calibrator. Although it is operationally difficult to quickly and reliably determine the total activity of each isotope at the time of measurement, it is possible to evaluate the percentage of total activity that is incorporated (RLY) and remains during serum stability studies. To evaluate the percentage of activity corresponding to each radioisotope, it was necessary to wait for the activity to decay and calculate it according to the t1/2 of each radioisotope.

Fig. 3: Multimodal imaging using a single IONT.
figure 3

a Radiolabeling yield in the synthesis of 68Ga/177Lu-IONT; 68Ga/67Ga-IONT and 67Ga/177Lu-IONT (N = 3 for 68Ga/177Lu-IONT and N = 2 for 68Ga/67Ga-IONT and 67Ga/177Lu-IONT); b) Serum stability for 68Ga/177Lu-IONT (1 h to 168 h); 68Ga/67Ga-IONT (1 h to 168 h) and 67Ga/177Lu-IONT (24 h to 672 h), (N = 3, One-way ANOVA with Geisser-Greenhouse correction)and c Representative CT/PET/SPECT and triple fusion images of 68Ga177Lu-IONT after i.v. injection in healthy mice. PET images were acquired for 30 min immediately after injection using a small-animal PET/CT system, followed by SPECT studies for 15 min. Finally, computed tomography (CT) was performed. Typical distribution of nanoparticles of this hydrodynamic size, with most of the activity ending in the liver and a minor portion in the bladder due to the size of the particles. In the case of PET images, it is possible to observe the circulation of the nanoparticles due to their biodistribution time.

Full size image

The RLY values for the double-doped IONT were larger than 85%, in agreement with the results of the single-doped experiments (Fig. 3a), and the serum stability study showed values larger than 92% for the three dual-labeled IONT (Fig. 3b and Table S2).

In vivo application of xxM/xxN-IONT

We selected 68Ga/177Lu-IONT to demonstrate its in vivo use for multiplexed PET/SPECT imaging, leveraging the presence of 68Ga and 177Lu for PET and SPECT imaging, respectively. 68Ga177Lu-IONT was synthesized and intravenously injected (45-80 µL; 2.2 mg/mL of Fe concentration, see Supporting Information for injected doses) into healthy 6-week-old male mice (C57BL/6 J). In vivo PET images were acquired for 30 min immediately after injection using a small-animal PET/CT system, followed by SPECT studies for 15 min. Finally, computed tomography (CT) was performed. Representative images are shown in Fig. 3c. The biodistribution of IONT is typical of these nanomaterials when not functionalized with a targeting vector, with most of the activity ending in the liver and a minor portion in the bladder due to the size of the particles (vide infra). In the case of PET, the circulation of the IONT can be observed because the images were acquired 30 min post-injection, and the circulation time of these IONT is typically around 90 min20. SPECT images similarly show the uptake of IONT in the liver due to gamma rays from 177Lu.

In vivo targeting with xxM-IONT

Passive targeting in glioblastoma model

In previous studies, we demonstrated that these iron oxide nanoparticles provide a strong signal in T1-MRI due to their small size and paramagnetic properties25. Here, we aimed to extend this use to combined PET and T1-MRI and evaluate whether our optimized nano-radioplatform can be used to diagnose a condition as complex as glioblastoma. Therefore, we performed dual PET/MR imaging using 68Ga-IONT in an orthotopic glioma model generated in C57BL6/J mice aged 8–10 weeks, using GL261 cells (see Supporting Information). 68Ga-IONT were i.v. injected through a previously inserted cannula in the tail, and PET and MRI acquisitions were performed simultaneously from the time of injection to 125 min on a 3.0 T system equipped with a PET insert.

The uptake of 68Ga-IONT in glioblastoma is shown in the PET image and, using positive contrast, in the MR image with excellent co-localization when fused (Fig. 4a). Further 3D reconstruction of dynamic PET images over 90 min post-injection showed circulation and different uptake of IONT in the liver and glioblastoma. For all injected animals (N = 11), the regions of interest (ROIs) of the brain and tumor were analyzed to determine the actual injected dose that eventually reached the tumor region and its variation over time. To analyze uptake in the whole organ under study (brain and heart to account for circulating activity), the ROI was expanded three-dimensionally to study the volume of interest (VOI). Figure 4c, d display the uptake over time for all the studied animals (see also Table S3). Owing to the significant influence of tumor size, the graphs were categorized into four sections based on tumor size. We observed that the larger the tumor, the higher the uptake (Fig. 4c), whereas no correlation was evident in the analyzed heart area (Fig. 4d). To extract more information from the obtained data, the %ID in the tumor (Fig. 4e) and heart (Fig. 4f) at 90 min (the standard circulation time for these IONT) were plotted against the size of each tumor. It can be seen that at very small tumor sizes, the uptake is practically non-existent and circulating activity is high, while from a certain value, estimated at 23 mm3 based on the linear adjustment, the passive uptake of the IONT begins to be significant. This value may indicate the minimum amount of tumor tissue that can break the BBB for IONT to accumulate passively under these experimental conditions.

Fig. 4: Glioblastoma detection.
figure 4

a Representative PET/MRI images 90 min after the i.v. injection of 68Ga-IONT in an orthotopic glioblastoma tumor model for three different mice showing uptake in the tumor by passive accumulation; b 3D PET reconstruction over 90 min; injected dose uptake with time in c tumor and d heart with time and for different tumor sizes, starting at 5 min due to PET acquisition time, injected dose, and volume are indicated in Table S3. Blue indicates a tumor size larger than 200 mm3, garnet indicates a size between 100 mm3 and 200 mm3, green indicates tumors between 50 mm3 and 100 mm3, and red/orange indicates a tumor size smaller than 50 mm3. e % of the injected dose accumulated in the tumor versus the tumor size. f % of injected dose circulating in the heart versus tumor size. Both e and f show a reference value of 23 mm3, from which the tumor was detected.

Full size image

Active targeting thrombosis by biorthogonal chemistry

The previous section demonstrated how xxM-IONT can achieve passive targeting under suitable conditions. We also sought to explore their potential for active targeting. In this context, we employed a pretargeting strategy based on bioorthogonal chemistry—specifically tetrazine ligation—which offers several advantages over conventional active targeting approaches. Although pretargeting differs from classical ligand–receptor binding, we consider it as an active targeting method because it relies on specific, engineered interactions rather than passive distribution. As an animal model, we selected a carotid crush injury model in mice, aiming to detect thrombus formation by targeting the responding platelets, a phenomenon that is difficult to detect using conventional targeting strategies with short half-life radioisotopes. For this, the antibody (anti-platelet, antiCD41) was modified by the covalent addition of trans-cyclooctene and the 68Ga-IONT surface was modified by covalently attaching a tetrazine group, as previously described19.

Figure 5a shows that the hydrodynamic size profile of the IONT did not change after covalent coupling with tetrazine. Similarly, Fig. 5b shows the reproducibility of the functionalization and how there is a small, though significant difference, between the two groups of IONT (with and without tetrazine) with a small increase (from an average of 14.5 nm to 15.6 nm) due to the incorporation of tetrazine molecules on the surface. Figure 5c shows the change in the zeta potential value for four repetitions of each type of IONT and how the value was significantly reduced owing to the incorporation of the tetrazine molecule and the loss of carboxylate groups. Figure 5d summarizes the results of in vivo experiments. Carotid crush injury was induced in healthy C57BL6 mice, followed by intravenous injection through the tail vein of the modified anti-CD41-TCO. After 24 h, to allow for the antibody biodistribution, 68Ga-IONT-TZ was synthesized and intravenously injected through the tail vein, and PET/CT images were acquired 120 min later. Control mice underwent the same protocol but were injected with saline instead of anti-CD41-TCO. Figure 5e shows the representative results of thrombus formation detection in the carotid crush injury mouse model. The PET/CT images of the control did not show any specific accumulation in the carotid area (Fig. 5e, top row), whereas the signal in the carotid area was clearly visible in the thrombotic lesion because the particles successfully underwent a biorthogonal reaction (Fig. 5e, bottom row). These results confirm the suitability of our platform for detecting complex lesions using pretargeting approaches.

Fig. 5: Pretargeted imaging.
figure 5

a Hydrodynamic size curves (volume %) for 68Ga-IONT before and after covalent coupling of the tetrazine moiety; b Z-average size values for six different syntheses of 68Ga-IONT and 68Ga-IONT-TZ analyzed by two-tailed unpaired t-test (N = 6); c Zeta potential values for different samples of 68Ga-IONT and 68Ga-IONT-TZ analyzed by two-tailed unpaired t-test (N = 4); d Experimental setup in the pretargeted imaging experiment. Carotid crush injury was induced in healthy C57BL6 mice, followed by intravenous injection through the tail vein of the modified anti-CD41-TCO. After 24 h, 68Ga-IONT-TZ was intravenously injected through the tail vein, and PET/CT images were acquired 120 min later; and e CT, PET, and PET/CT images of the mouse model injected with saline solution (bottom row), without any specific accumulation in the carotid area, or with TCO-modified anti-CD41 antibody (top row), with a clearly visible signal in the thrombotic lesion because the particles successfully underwent the biorthogonal reaction. In both conditions, the animals were injected with 68Ga-IONT-TZ.

Full size image

Glioblastoma treatment with 177Lu-IONT

Once the capacity of the IONT for passive and active targeting was proven, we prepared 177Lu-IONT to evaluate their therapeutic potential in a glioblastoma model. The animals were divided into two groups: control (Group C) and intratumorally treated (Group IT). The four animals from GROUP IT that were sacrificed at t = 0 to confirm that the 177Lu-IONT effectively reached the tumor were also used as a reference for the tumor size at t = 0. Thus, at point 1, there were six animals, of which three were injected and three were not (Table S4). The biodistribution of GROUP IT showed that most of the nanoparticles remained in the tumor for 14 days after intratumoral injection (Fig. 6a).

Fig. 6: Glioblastoma Therapy.
figure 6

a Gamma counter biodistribution of 177Lu-IONT for intratumoral injection (GROUP IT) represented as the percentage of injected dose per gram of tissue; b Tumor mass at t = 0 and t = 14 d for each control and experimental animal after treatment with 177Lu-IONT; injected doses are indicated in Table S4; c Statistical analysis of the data collected for tumor weight in control and experimental groups 14 days after the intratumoral injection (mean ± standard deviation; statistical analysis by ANOVA test, N = 3–6; ****P 0.0001).

Full size image

Subsequently, the efficacy of the treatment was analyzed. The tumors were weighed immediately after the animals were sacrificed at 7 and 14 days, and the evolution of tumor weight was compared. As shown in Fig. 6b (for individual tumors) and 6c (for average weight), untreated tumors increased in weight by more than 500%, whereas tumors intratumorally treated with 177Lu-IONT did not increase in weight. Figure S6 shows images of the tumors for untreated and treated mice at 7 and 14 days of treatment, depicting clearly observable differences in tumor size.

Renal clearance of xxM-IONT

Although our results confirmed the potential of xxM-IONT as an efficient radio-theranostic agent for solid tumors following intravenous and intratumoral administration, these therapeutic applications require multiple injections for a complete treatment regime, raising concerns regarding the potential excessive accumulation of IONT in the liver and spleen. Although this is not a problem for imaging applications, it poses a twofold risk for therapy: first, the possibility of an excessive radiation dose to healthy elimination organs, and second, the risk of substantial iron accumulation in these tissues, which might present a significant hazard. We addressed these concerns by leveraging the tunability of nanomaterials to decrease their size, enabling renal excretion, and minimizing both biodistribution time and IONT accumulation in elimination organs.

To achieve this, we modified the reaction conditions, with a particular focus on the reaction temperature and time, to reduce the total size of the nanoparticles (core plus coating) as much as possible while maintaining the integrity of the particles and radiolabeling. A library of 11 novel IONT was initially evaluated in the absence of radioactivity to assess their hydrodynamic size and zeta potential. Comprehensive characterization of these properties and their stability in water at 25 °C and 37 °C was performed (Fig. S8). IONT with the smallest hydrodynamic size (≤5.5 nm) were selected for the next stage. Based on this, 8 of the 11 IONT progressed to the next phase, which involved their synthesis incorporating 68Ga in their core. The objective of this study was to develop IONT that exhibit renal clearance and favorable properties in terms of colloidal and radiolabeling stability. To achieve this, specific criteria were established for each 68Ga-IONT: a hydrodynamic size ≤5 nm, a negative zeta potential ≤ -25 mV, a radiolabeling yield greater than 25%, and radiolabeling stability higher than 90%. Table 1 summarizes the results of the physicochemical and radiochemical characterizations of the 8 68Ga-IONT.

Table 1 Summary of physicochemical properties of 68Ga labeled IONT for renal clearance
Full size table

For a better comparison, we used the initial 68Ga-IONT, which showed common elimination in the liver and spleen, as a reference. The most significant changes were observed in hydrodynamic size and RLY. Six samples had sizes below the 5 nm threshold (Fig. 7a). In terms of Z-potential, all samples showed similar results, ensuring good colloidal stability (Fig. 7b). As expected, the RLY values were smaller than those of the standard 68Ga-IONT, as the temperature and reaction time play crucial roles in the nucleation of the particles and, hence, in the incorporation of the radiometal. We set a minimum RLY of 25% to obtain a sufficient signal for imaging and/or therapeutic experiments, considering the initial activities used in this work. Nevertheless, further experiments should focus on improving this value to achieve a higher specific activity. Finally, the RCS was excellent for all samples, confirming the stable incorporation of the isotope into the nanoparticles, even with IONT that were significantly smaller (Fig. 7d). Among the developed particles, the sample synthesized at 65 °C for 30 min (65_30) was selected as the best sample for in vivo testing. This sample showed a low hydrodynamic size, good Z potential value, decent RLY, and most importantly, the highest RCS of all the samples. Considering the characterization values, other candidates, such as 80_10 and 120_1, may be worth exploring in future applications.

Fig. 7: Characterization of the eight IONT samples candidates for renal clearance.
figure 7

For simplicity, 68Ga-IONT samples are indicated referring to the temperature and time for the synthesis (T_t) a Hydrodynamic size (nm) measured by DLS (volume %), the dashed line indicates the 5 nm threshold for the eight IONT, six samples have sizes below the 5 nm threshold; b Z potential (mV) measured by DLS, dashed line indicates the -25 mV minimum threshold for the eight IONT, all the samples show values below the threshold; c RLY of the eight 68Ga-IONT synthesized, dashed line indicates the 25% minimum threshold, five samples show values larger than the threshold; d Radiolabeling stability in serum of the eight samples, dashed line indicates the 90% stability threshold, all the samples show RCS values similar to or larger than the threshold; e HAADF-STEM images for selected 68Ga-IONT_65_30 (scale bars are 20 and 10 nm); f HAADF-STEM images for 68Ga-IONT (scale bars are 20 and 10 nm); g Selected PET MIP images of two different mice after the i.v. injection of 68Ga-IONT_65_30; h DLS curves (volume %) for 68Ga-IONT_65_30 (blue) and 68Ga-IONT (green) and i Selected in vivo PET MIP images of two different mice after the i.v. injection of 68Ga-IONT.

Full size image

68GaIONT_65_30 was characterized by HADDF-STEM to compare its morphology with that of the standard 68GaIONT (Fig. 7e, f). We observed a similar quasi-spherical shape with an average core size of 1.7 ± 0.3 nm for the 65_30 sample and 3.2 ± 0.4 (N = 100) for 68GaIONT. These two samples were then used in in vivo PET to study their biodistribution and excretion routes. Different distributions of the two types of nanotracers were clearly observed. Sample 65_30 was rapidly eliminated through the kidneys, with the activity concentrated in the bladder 5 min post-injection, while the traditional 68Ga-IONT showed the main accumulation in the liver and spleen (Fig. 7g). As expected, the different excretion pathways correlated with the differences in the hydrodynamic sizes of the two nanomaterials, as shown by the DLS curves (Fig. 7g).

Discussion

The development of radiopharmaceuticals is an active research field owing to the promising features of such compounds in nuclear medicine and theranostics. One of the bottlenecks in such development is the lack of suitable chelators that can be safely used in vivo, especially because of the growing number of radiometals of interest. In this study, we demonstrate how the use of nanomaterials can solve many of these problems.

This study demonstrates the potential of iron oxide nanotracers (IONT) as a versatile platform for radiolabeling a wide range of radiometals used in nuclear medicine. Ten different nano-radiotracers were synthesized using a one-step protocol, which showed excellent RLY and RCS values for all tested radiometals. Different proof-of-concept experiments have demonstrated the versatility and tunability of this nanoplatform, addressing several challenges in radiopharmaceutical development, from passive to active targeting, imaging to therapy, and easily tunable biodistribution. Overall, the IONT platform shows great promise as a robust and adaptable tool for the development of next-generation radiopharmaceuticals for various diagnostic and therapeutic applications in nuclear medicine. Further studies are warranted to optimize specific formulations and evaluate their clinical translation.

Methods

Chemistry

Synthesis of xxM-IONT and RLY evaluation

Iron (III) chloride hexahydrate (1 eq), sodium citrate tribasic dihydrate (1 eq), and radiometal water solution were added to a high-pressure MW-designed vial using a magnetic stirrer. Hydrazine hydrate 50–60% (10% vol) was then added to the vial. The vial was closed with a septum cap and stirred under MW irradiation at 100 °C for 10 min. Subsequently, the reaction was cooled to 50 °C using pressurized air, and the mixture was purified by size-exclusion chromatography.

Radiolabeling yield and stability evaluation

The radiolabeling yield (RLY) was measured as the percentage of purified sample activity divided by the total activity (sample activity + purification column activity). The in vitro stability of the xxM-IONT was tested in human serum. For this purpose, 150–250 µL of radiolabeled nanoparticles was incubated with 150 µL of human serum (Sigma Aldrich) at 37 °C. The stability was evaluated at 1 h and 3 h by measuring the serum stability as the percentage of activity that remained in the filter after three steps of ultrafiltration with AMICON 30 kDa with PBS 1X as the solvent for re-suspending the sample remaining between filtrations.

Labeling competition vs Zn

The synthesis was performed under the same conditions as for the other xxM-IONT, but with the addition of ZnCl2 (0.05 mM).

Renal clearance optimization study

The synthesis was performed in a silicone bath that was preheated to the target temperature using a high-pressure microwave vial closed with a cap and silicon septum. Iron (III) chloride hexahydrate (1 eq), sodium citrate tribasic dihydrate (1 eq), and 68GaCl3 were dissolved in water (Milli-Q grade, 90% vol) in a 10 ml reaction vial. The solution was placed inside a silicone oil bath, and after reaching the reaction temperature, hydrazine monohydrate 50–60% was quickly added (10% vol) through the septum. The reactions were stirred for different times depending on the synthesis and were rapidly stopped by placing the mixture in an ice–water bath. Purification was performed by size-exclusion chromatography using pre-packed Sephadex G-25 M (PD10) columns.

The stability of IONPs in water over time was evaluated by studying the variations in two key structural parameters: hydrodynamic size and Z potential. A Zetasizer NanoZS90 (Malvern Panalytical, United Kingdom) was used for this purpose.

To carry out the most complete evaluation possible, the study was carried out at two different temperatures, 25 °C (rt) and 37 °C (physiological conditions, in a dry bath, ThermoMixer®) over a period of 7 days (analyzing the samples at 0, 1, 3, and 7 days). At each time point, each sample was analyzed in triplicate. For the two candidates selected as optimal for use in renal elimination, the same stability study was performed in PBS1x.

In vivo PET/SPECT/CT

PET and micro-CT imaging acquisitions were sequentially performed using the β (PET), γ (SPECT) and X-cube (CT) microsystems of Molecubes® (MOLECUBES NV), respectively. 4 healthy mice were i.v. injected through the tail vein with 68Ga/177Lu-IONT (5-6 MBq 68Ga, 48-52 MBq 177Lu) under isoflurane anesthesia (5% induction, 3% maintenance in medicinal gas). PET images were acquired for 30 min immediately after injection using a small-animal PET/CT system, followed by SPECT studies for 15 min. Finally, computed tomography (CT) was performed. For each animal, PET-SPECT-CT images were co-registered and analyzed using PMOD image analysis software (PMOD Technologies Ltd., Zürich, Switzerland). The volumes of interest (VOIs) were manually outlined on the organs of interest in the CT images. VOIs were transferred to PET images, and activity values normalized to the initial amount of radioactivity were obtained as the mean standard uptake values (SUVmean).

In vivo PET/MRI

PET imaging post-processing was performed using PMOD software (PMOD Technologies GmbH, Version 4.302). The volumes of interest (VOIs) were established using MRI images and transferred to the reconstruction of the 5 min window dynamic PET images. The percentage of injected dose per gram in the brain and heart was measured.

PET-MRI studies were conducted using a horizontal system equipped with a PET insert for simultaneous acquisition (MRS*DRYMAG 3T-7T variable field, MR Solutions Ltd., Guildford, UK) and a 38 mm-diameter birdcage coil for MRI studies. Images were acquired using Preclinical Scan software.

Animals were anesthetized as described in the tumor generation protocol, and anesthesia was maintained using a nose mask throughout the imaging experiment. Body temperature was maintained at 37 °C using an air-heated bed, and the respiratory rate was monitored during the MRI study.

Tumor development was monitored using T2W MRI twice a week. Axial T2W images were obtained using a Fast Spin Echo sequence, with 14 slices of 1 mm slice thickness, a matrix size of 256 × 256, FOV = 30 × 30 mm2, corresponding to an in-plane resolution of 117 × 117 µm2, TR = 2500 ms, TE effective = 45 ms, RARE factor = 8, number of averages = 6, and total ACQ time = 8 min.

Three weeks after tumor generation, the mice were subjected to imaging experiments. The animals were positioned in the scanner with a tail vein catheter. A series of 12 consecutive T1W images in the axial orientation, using the same geometry as the T2W images, were acquired using a Spin Echo sequence. The parameters were TR = 360 ms, TE effective = 11 ms, 6 averages, and an acquisition time of 9 min and 13 s. The first T1W image was acquired before nanotracer injection (baseline), with subsequent images acquired after nanotracer injection simultaneously with PET acquisition for 120 min. Animals were sacrificed in a CO2 chamber.

GBM model

Authenticated glioma GL261 cells were cultured in RPMI supplemented with 10% fetal bovine serum and antibiotics (1% penicillin/streptomycin). The cells were maintained in an incubator at 37 °C and 5% CO2 until they reached confluence, detached, and counted to generate the glioma model.

An orthotopic glioma model was generated in C57BL6/J mice aged 8–10 weeks, as previously described25. Briefly, the mice were placed in a stereotaxic device, where anesthesia was maintained using a nose mask (1–1.5% isoflurane/O2), and the eyes were covered with Vaseline to prevent drying. A midline incision on the skull was made using a scalpel, and a burr hole was created 0.23 mm right of the bregma using a 25 G needle. Then, 105 GL261 cells in 4 μL of RPMI with 30% Matrigel were injected to a depth of 0.33 mm into the right caudate nucleus using a Hamilton syringe. After 5 min, the syringe was carefully removed, the hole was sealed with bone wax, and the scalp was sutured. The animals were injected subcutaneously with buprenorphine (0.1 mg/kg) for analgesia 30 min before the procedure and for the following 2 days. Animals were sacrificed in a CO2 chamber.

177Lu-IONT in vivo therapy experiment

Authenticated glioma U251 cells were resuspended in DMEM (Dulbecco´s Modified Eagle Medium 1x) supplemented with GlutaMAXTM-I + 4.5 g/L D-Glucose + Pyruvate + 5% Bovine Fetal Serum + antibiotics 1% penicillin/streptomycin. The cells were maintained in an incubator at 37 °C and 5% CO2 until they reached confluence, detached, and counted to generate the glioma model.

Inoculation of female immunodeficient Athymic Nude 7–8-week-old mice (Charles Rivers) with 177Lu-IONT was performed with an insulin syringe in the right lower flank (1 x 106 cells), and the tumors were left to grow for 19 days before starting the study.

Carotid crush injury model

The day before PET acquisition, wild-type mice with a hybrid C3H/He-C57BL/6 genetic background were operated on to surgically induce a carotid crush injury, working as positive controls of the probes. The mice were approximately 70 weeks old and 25–33 g of weight. The procedures were conducted at the animal facility of the Centro Nacional de Investigaciones Cardiovasculares (CNIC, Madrid - Spain) according to the European Communities Council Directive 2010/63/EU, following the ARRIVE guidelines, and approved by the Ethics Committee in Animal Experimentation and corresponding authority (PROEX 147.7/20).

Mice were anesthetized with sevoflurane (5% induction, 3% maintenance in medicinal gas) and placed on a heating pad (37 °C) to avoid hypothermia. The surgical area was depilated and prepared using antisepsis (chlorhexidine 4% topical solution) before making an incision to expose the left common carotid artery. The vessel was isolated by resecting the surrounding tissues, and the blood flow was occluded with a surgical mosquito clamp to promote coagulation mechanisms resulting from vascular injury. After 15 min, the blood flow was restored by releasing the clamp closure, the incision was sutured, and topical antisepsis in the wound and postoperative IP analgesia (buprenorphine, 0.1 mg/kg) were administered. Finally, immediately after the completion of surgery, animals were intravenously injected through the tail vein with AntiCD41-TCO (70 mg/kg, positive controls – C + /C + ) or saline serum (negative controls – C + /C-). Animals were sacrificed in a CO2 chamber.

Pretargeting experiment

Immediately after finishing the surgeries, animals were intravenously injected through the tail vein with the antibody anti-CD41-TCO (70 mg/kg) under isoflurane anesthesia (5% induction, 3% maintenance in medicinal gas). The next day, animals were i.v. injected with 68Ga-IONP-TZ (~10 MBq/mice) through the tail vein and, after 2 h of uptake, PET images were acquired.

PET/CT studies were performed in the facilities of the Distributed Biomedical Imaging Network (ReDIB, Spain) at the Spanish Center for Cardiovascular Research (CNIC) using a dedicated nanoScan® PET/CT scanner (Mediso, Hungary) for small animals. Temperature and respiration (through a respiratory pad) were registered during the entire imaging session. The whole body of each mouse was scanned in each imaging protocol, including a 10 min CT image followed by a 30 min PET acquisition. PET images were reconstructed with a 3D-OSEM algorithm (4 iterations, 6 subsets, isotropic voxel size of 0.4 mm per axis), and CT studies using an isotropic voxel size of 0.14 mm per axis. Corrections for radioactive decay and dead time were applied.

Animals

All experiments were performed in mice kept in specific pathogen-free facilities at Centro Nacional de Investigaciones Cardiovasculares (CNIC), Instituto de Investigaciones Biomédicas Sols-Morreale (IIBM), and CIEMAT, under a 12 h light/12 h dark schedule (lights on at 7am, o at 7 pm), with water and chow available ad libitum. All experimental procedures were approved by the Animal Care and Ethics Committee of the CNIC, IIBM, CIEMAT, and Madrid regional authorities.