As a molecular aggregate with precise structural information and unique electronic characteristics, lanthanide nanoclusters have become a bridge between microscopic molecules and macroscopic nanoparticles, and occupy an irreplaceable position in the fields of synthetic chemistry and material chemistry [1–5]. Compared with other nanomaterials, lanthanide nanoclusters have a strictly defined composition and geometric configuration, allowing their photophysical properties to establish a clear correlation with the structure, thus providing a solid foundation for the study of imaging mechanisms. Through ideas such as ligand design, core number regulation and assembly strategy optimization, a series of lanthanide nanoclusters with high symmetry, multi-core structure and precise coordination environment have been successfully constructed [6–11]. Thanks to the special configuration of the 4f electron layer, the lanthanide nanoclusters have shown excellent photoelectromagnetic properties and have made significant progress in the fields of ultra-low temperature magnetic refrigeration and solid-state luminescence [12–15]. However, compared with the widespread application of the solid-state field, the application expansion of lanthanide nanoclusters in solutions is relatively lagging, mainly because they face a series of problems such as poor stability, poor dispersion and luminescence quenching in solutions [16–18].

In recent years, our research team has proposed a multidentate chelate coordination (MCC) for directed construction of stable lanthanide nanoclusters in solution [19,20]. Based on the MCC, a series of stable lanthanide nanoclusters have been obtained and their optical imaging and sensing properties in aqueous solutions have been expanded [21–23]. In 2024, our team used the MCC to synthesize a case of Eu₉Zn₈ nanocluster with dual antenna effects and achieved high-resolution cell imaging effects [22]. In 2025, we used polyoxygenate Mo₈ as an antenna to synthesize luminescent EuMo₁₆, achieving specific antibiotic-enhanced antenna effect behavior [23]. Contrast agents (CAs) are substances that can significantly enhance image contrast by altering the physical or chemical properties of tissues or organs. Clinically, Gd(Ⅲ) chelates are the most common CAs for nuclear magnetic resonance imaging (MRI) [24,25]. So far, the relaxation rate of these Gd(Ⅲ) chelates is low, and the need for high doses has caused a series of problems such as nephrotoxicity [26]. According to the Solomon-Bloembergen-Morgan (SBM) relaxation theory, multi-core Gd(Ⅲ) clusters constructed using multi-component integration strategy can significantly improve relaxation efficiency, thereby effectively reducing the risk of CAs in clinical applications [27]. At present, the construction of lanthanide nanoclusters with both MRI and optical imaging functions is still in its infancy [28,29]. Obviously, the expansion of lanthanide nanoclusters with both MRI and optical imaging functions provides new imaging methods for early diagnosis of diseases.

Herein, three yellow block crystals were obtained by reacting 2-aminobenzyl alcohol, o-vanillin and Ln(NO₃)₃·6H₂O (Ln = Eu, Dy, and Gd) under solvent thermal conditions, and their molecular formulas were [Ln₉(L¹)₄(L²)(NO₃)₁₀(CH₃O)(μ₂—OH)₂(μ₃—OH)₃(CH₃OH)₅](NO₃)₂·xCH₃OH·yCH₃CN (Ln = Eu, x = 7, y = 9; Dy, x = 7, y = 4; Gd, x = 7, y = 9; H₂L¹ = 2–methoxy-6-[(E)-2′-hydroxymethyl-phenyliminomethyl]phenol; HL² = o-vanillin). High resolution electrospray mass spectrometry (HRESI-MS) results prove that nanocluster Ln₉ has high stability in solution. In addition, time-dependent HRESI-MS tracks the formation process of Dy₉ clusters, identifying changes in multiple intermediates, and revealing that it may be formed through a multi-template assembly mechanism. It is worth noting that the lanthanide nanocluster Eu₉, which is low in biotoxicity, uniform in size and positive charge, can still achieve excellent imaging effects under low concentration incubation, and exhibit excellent mitochondrial targeting capabilities in both MDA-MB-231 and MCF-7 cells (Scheme 1). To the best of our knowledge, this is the first application of lanthanide nanoclusters for mitochondria-specific fluorescence imaging, and this finding provides an important reference for the development of novel lanthanide nanoclusters for subcellular organelle localization. In addition, in zebrafish and its embryo models, Eu₉ also showed selective marking ability to the egg envelope, yolk sac and digestive system. The low concentration of Gd₉ has a high relaxation rate (r₁ = 39.42 mmol L⁻¹ s⁻¹, and r₂ = 50.58 mmol L⁻¹ s⁻¹) at a magnetic field intensity of 0.5 T, indicating its excellent MRI application potential. Ln₉ offers new insights into developing lanthanide nanoclusters with potential for high-resolution imaging applications.

Single crystal X-ray diffraction (SCXRD) indicates that nonanuclear Ln₉ crystallizes in the P-1 space group of the triclinic crystal system (Tables S1 and S2 in Supporting information). Ln₉ consists of 9 Ln(Ⅲ) ions, 4 ligands (L¹)²⁻ that remove two protons, 1 ligand (L²)⁻ that removes one proton, 10 NO₃⁻, 1 CH₃O⁻, 5 OH⁻, 5 CH₃OH, and peripheral guest molecules. Their molecular formulas are [Ln₉(L¹)₄(L²)(NO₃)₁₀(CH₃O)(μ₂—OH)₂(μ₃—OH)₃(CH₃OH)₅](NO₃)₂·xCH₃OH·yCH₃CN (Ln = Eu, x = 7, y = 9; Dy, x = 7, y = 4; Gd, x = 7, y = 9; H₂L¹ = 2–methoxy-6-[(E)-2′–hydroxy methyl-phenyliminomethyl]-phenol; HL² = o-vanillin) (Figs. 1a and b). The nonanuclear Ln₉ cluster core can be viewed as a triangular framework that extends in three different directions, generating a binuclear structure. Generally, due to the unique spatial extension mode of electrons and the high coordination number of Ln³⁺ ions, the nonanuclear lanthanide clusters tend to form highly symmetric "hourglass-type" cluster nuclei [30–32]. The lowest energy effect brought by high symmetry gives "hourglass" clusters an absolute advantage in solution self-assembly and crystal stacking. However, the structural characteristics of nonanuclear Ln₉ clearly deviate from this common pattern: the cluster core of Ln₉ can be abstracted into a triangular frame, in which three vertices are occupied by an 8-coordinated and two 9-coordinated Ln³⁺, and the three sides extend outward from the dual-core unit, forming three radial arms. This structural feature forms a sharp contrast with the "hourglass" cluster core. The difference between nonanuclear lanthanide clusters Ln₉ and the highly symmetric "hourglass" cluster provides new topological sites for solvent molecules, counter ions and guest molecules, enriching the structural database of lanthanide metal complexes. The {Dy/O} cluster nucleus of nonanuclear Ln₉ is formed by the oxygen atoms on the deprotonated ligand, OH⁻ and NO₃⁻ bridging the nine Ln(Ⅲ) ions through the co-formation (Figs. 1c and d). The (L¹)²⁻ ligand within the nonanuclear Ln₉ structure chelates three Ln(Ⅲ) ions, with a coordination mode of μ₃-η²: η¹: η²: η¹ (Fig. 1e). In addition, the (L²)⁻ ligand chelates two Ln (Ⅲ) ions separately, with a coordination mode of μ₂-η¹: η²: η¹ (Fig. 1f). The nine Ln(Ⅲ) ions within the nonanuclear Ln₉ structure possess five different coordination environments, indicating its relatively low symmetry (Tables S3-S7 in Supporting information) [33]. The thermogravimetric analysis test proved that the number of free guest molecules outside the nonanuclear Ln₉ structure was consistent with the molecular formula (Fig. S2 in Supporting information). The absorption peaks of the Fourier transform infrared (FT-IR) absorption spectrum of nonanuclear Ln₉ were located at 1460, 1384, 1295, 1231, and 738 cm⁻¹, respectively, which can be attributed to the bending vibration of the C—H bond, symmetric deformation vibration of -CH₃, out-of-plane bending vibration of the C—H bond of the benzene ring, the stretching vibration of the C—O bond, and the out-of-plane bending vibration (Fig. S3a in Supporting information). The observed and simulated values of powder X-ray diffraction (PXRD) of nonanuclear Ln₉ at ambient temperature and pressure are in high agreement, proving that they are all pure phases (Fig. S3b in Supporting information). Scanning electron microscopy (SEM) also verified the cleanliness of the nonanuclear Dy₉ surface (Fig. S4 in Supporting information).

A small amount of Dy₉ crystals was dissolved in chromatographically pure acetonitrile and subsequently diluted extensively with chromatographically pure anhydrous methanol for high-resolution electrospray mass spectrometry (HRESI-MS) testing [16,19]. HRESI-MS analysis showed that molecular ion peaks highly consistent with the nonanuclear Dy₉ body frame were detected, indicating that Dy₉ exhibited high stability in the above solutions. As shown in Fig. 2, the molecular ion peaks of HRESI-MS of Dy₉ cluster were attributed and fitted as follows: [Dy₉(L¹)₄(L²)(CH₃O)₂(OH)₁₆]²⁺ (Cal. 1485.32, Exp. 1485.40), [Dy₉(L¹)₄(L²)(CH₃O)₅(NO₃)₃(OH)₁₀]²⁺ (Cal. 1573.85, Exp. 1573.92), [Dy₉(L¹)₄(L²)(NO₃)₂(OH)₁₆(H₂O)(CH₃OH)(CH₃CN)₂]²⁺ (Cal. 1582.34, Exp. 1582.44), [Dy₉(L¹)₄(L²)(NO₃)₃(OH)₁₅(H₂O)₄(CH₃OH)]²⁺ (Cal. 1590.85, Exp. 1590.92), [Dy₉(L¹)₃(L²)(NO₃)₇(OH)₁₃(CH₃OH)₃]⁺ (Cal. 3131.67, Exp. 3131.70), [Dy₉(L¹)₄(L²)(NO₃)₄(OH)₁₂(CH₃O)₂]⁺ (Cal. 3149.68, Exp. 3149.78), [Dy₉(L¹)₄(NO₃)₇(OH)₁₂(CH₃CN)(CH₃OH)₂]⁺ (Cal. 3227.69, Exp. 3227.75). Combined with structural analysis, the peripheral ligands (L¹)²⁻, (L²)⁻ and NO₃⁻ all form a tight protective shell for the cluster nuclei of nonanuclear Ln₉ clusters, enhancing the ability of the clusters to withstand attack by external guest molecules (Fig. S5 in Supporting information). Clearly, this structural feature provides a foundation for exploring the assembly mechanisms and potential applications of such lanthanide clusters in solution.

Time-dependent HRESI-MS was utilized to monitor changes in species at different time points during Dy₉ formation. The resolution and intensity changes of the molecular ion peaks were utilized to speculate on the possible assembly mechanism of Dy₉. As shown in Fig. 3, molecular ion peaks were collected at multiple time points (i.e., 0, 30 min and 1, 2, 3, 6, 12, 24, 48 h) in the m/z = 200~4000 range, and 13 relevant molecular ion peaks were monitored. At the start of the reaction (at 0 min), the molecular ion peak of the organic ligand HL² was monitored (m/z = 266.25). As the reaction progresses, the signal intensity associated with the HL² ligand peaks and then decreases in intensity. This phenomenon suggests that the HL² ligand is involved in coordination in the reaction system and subsequently in the assembly process. The above debris can be attributed: [(C₈H₈O₃)(CH₃CN)(H₂O)₄]⁺ (Cal. 266.25, Exp. 266.12) (Fig. S6a in Supporting information). As the reaction progressed to 2 h, molecular ion peaks associated with the organic ligand H₂L¹ began to be detected, with fragments attributed to the (C₁₅H₁₅NO₃)⁺ (Cal. 258.12, Exp. 258.11) (Fig. S6b in Supporting information). However, at the beginning of the reaction initiation (0 min), a fragmentation peak of DyL¹ was already detected, indicating that the organic ligand H₂L¹ rapidly coordinated with the Dy(Ⅲ) ion to form a fragment of DyL¹. Therefore, the molecular ion peak associated with H₂L¹ was not observed at the onset of the reaction. The intensity of the DyL¹ fragment peaked at 3 h and diminished with reaction time. These molecular ion peaks are attributed to [Dy(L¹)(CH₃O)(CH₃CN)₂]⁺ (Cal. 532.99, Exp. 533.10), [Dy(L¹)(NO₃)(CH₃CN)₃(CH₃OH)(H₂O)₈]⁺ (Cal. 780.26, Exp. 780.20) (Fig. S7 in Supporting information). In the range of m/z = 426.95~521.95, three molecular ion peaks associated with DyL² fragments were detected, which were formed by chelating Dy(Ⅲ) ions with the organic ligand HL². The molecular ion peak of DyL² can correspond to the following species: [Dy(L²)(NO₃)₂(CH₃CN)₂]⁺ (Cal. 521.95, Exp. 522.00), [Dy(L²)(NO₃)(OH)(CH₃OH)(H₂O)]⁺ (Cal. 443.94, Exp. 444.00), [Dy(L²)(NO₃)(OH)(CH₃OH)]⁺ (Cal. 426.95, Exp. 426.99) (Fig. S8 in Supporting information). Finally, six major backbone peaks of [Dy₉(L¹)₄(L²)] were detected in the interval of m/z = 1485.32~3227.69. These characteristic fragments were first observed at 1 h and as the reaction progressed, their number intensity gradually increased and reached a peak at 48 h. Attribution of these molecular ion peaks: [Dy₉(L¹)₄(L²)(CH₃O)₂(OH)₁₆]²⁺ (Cal. 1485.32, Exp. 1485.40), [Dy₉(L¹)₄(L²)(CH₃O)₅(NO₃)₃(OH)₁₀]²⁺ (Cal. 1573.85, Exp. 1573.92), [Dy₉(L¹)₄(L²)(NO₃)₂(OH)₁₆(H₂O)(CH₃OH)(CH₃CN)₂]²⁺ (Cal. 1582.34, Exp. 1582.44), [Dy₉(L¹)₄(L²)(NO₃)₃(OH)₁₅(H₂O)₄(CH₃OH)]²⁺ (Cal. 1590.85, Exp. 1590.92), [Dy₉(L¹)₃(L²)(NO₃)₇(OH)₁₃(CH₃OH)₃]⁺ (Cal. 3131.67, Exp. 3131.70), [Dy₉(L¹)₄(L²)(NO₃)₄(OH)₁₂(CH₃O)₂]⁺ (Cal. 3149.68, Exp. 3149.78), [Dy₉(L¹)₄(NO₃)₇(OH)₁₂(CH₃CN)(CH₃OH)₂]⁺ (Cal. 3227.69, Exp. 3227.75). HRESI-MS results show that the [DyL¹] template primitive and the [DyL²] template primitive may be assembled through a multi-template assembly mechanism to form the final cluster [Dy₉(L¹)₄(L²)].

Exploration of the optical properties of metal complexes is crucial for their application in high-resolution imaging [34]. Under irradiation of 365 nm ultraviolet lamps, the bright red luminescence of Eu₉ solution was observed, proving its potential application value in the field of optical. The UV–visible (UV–vis) absorption spectrum of Eu₉ showed a wide absorption peak at 280-430 nm (Fig. 4a). In addition, under 365 nm excitation, Eu₉ showed obvious characteristic emission peaks of Eu(Ⅲ) ions at 592, 615 and 698 nm, respectively, which can be attributed to the characteristic energy level transitions of ⁵D₀→⁷F₁, ⁵D₀→⁷F₂ and ⁵D₀→⁷F₄ of Eu(Ⅲ) ions, respectively (Fig. 4b). As shown in Fig. 4c, within the range of the excitation wavelength gradually increased from 370 nm to 410 nm, Eu₉ clearly showed the characteristic emission of Eu(Ⅲ) ions. The above results show that ligands around the Eu₉ cluster can act as efficient antennas and sensitize the emission of Eu(Ⅲ) ions through antenna effects.

To understand the energy transfer pathway of Eu₉, a Jablonsky energy level diagram was drawn (Fig. 4d). Following Reinhoudt's empirical guidelines, the inter-system crossing (ISC) process will be efficient when the energy level difference between the lowest singlet excited state (S₁) and the lowest excited triplet state (T₁) of the ligand reaches or exceeds 5000 cm^-1 [35–37]. Based on the solid-state UV–vis absorption spectroscopy data of H₂L¹, the energy level of S₁ is 30,211 cm⁻¹ (Fig. S9 in Supporting information). We obtained the solid phosphorescence spectrum of Gd₉ at 77 K and performed Gaussian curve fitting to derive the T₁ energy value for H₂L¹, which was 21,097 cm⁻¹ (Fig. S10 in Supporting information). Therefore, the energy difference (ΔE₁) between S₁ and T₁ of H₂L¹ is 9114 cm⁻¹. Obviously, the above ΔE is significantly greater than 5000 cm⁻¹, confirming that there is an efficient ISC process between the S₁ and T₁ energy levels of H₂L¹. According to Latva's rule of thumb, the optimal energy level difference (ΔE) between the ⁵D₀ energy level of Eu(Ⅲ) ions and T₁ capable of efficient energy transfer processes is 2000~5000 cm⁻¹ [38,39]. The energy level of the ⁵D₀ energy level of Eu(Ⅲ) ion is 17,500 cm⁻¹. Obviously, the ΔE between the T₁ energy level of H₂L¹ and the ⁵D₀ energy level of Eu(Ⅲ) ion is 3597 cm⁻¹, proving that H₂L¹ can emit light as a characteristic of efficient antenna-sensitized Eu(Ⅲ) ion.

Compared with single-metal or low-core lanthanide complexes, high-nuclear lanthanide clusters not only have the quantum size effect brought by nanosize, but also highly concentrated metal centers within the clusters can often promote the development of new functions. Dynamic light scattering (DLS) results indicate that Eu₉, Dy₉, and Gd₉ clusters form uniformly dispersed nanoparticles in aqueous solution, with hydrated diameters of 118.2, 88.89, and 111.37 nm, respectively (Fig. 5a and Fig. S11 in Supporting information). The stability of Eu₉ and Gd₉ in physiological media was confirmed by dispersing them in PBS, complete RPMI-1640, and Dulbecco's Modified Eagle Medium. Over five days, their UV–vis absorbance at 406 nm exhibited negligible variation, exhibiting excellent stability under biologically relevant conditions (Fig. S12 in Supporting information). Furthermore, as shown in Fig. 5a, the zeta potential of the Eu₉ cluster in the aqueous solution was 22 mV, confirming that its nanoparticles have high stability in the aqueous solution. The potential toxicity of concentration-dependent Eu₉ solutions (0, 6.25, 12.5, 25, 50, 100 μg/mL) to different cells was evaluated using 3-(4,5-dimethylthiazole-2-yl)-2,5-diphenyltetraazole bromine salt (MTT). As shown in Fig. 5c and Fig. S13 (Supporting information), after incubating HeLa, MDA-MB-231, HepG2, 4T1, SK-OV-3, MCF-7 and WI-38 cells with concentration-dependent Eu₉ solutions for 16 h, the cell survival rate exceeded 75%, demonstrating that Eu₉ has low cytotoxicity. The performance of Eu₉ in the optical imaging field of HeLa, MDA-MB-231, HepG2, 4T1, SK-OV-3, MCF-7 and WI-38 cells was further explored. Cells incubated in Eu₉/DMEM solution at a concentration of 25 μg/mL for 16 h were analyzed by in vitro cell imaging using two-photon confocal laser scanning microscopy (CLSM). The results show that the Eu₉ cluster can be effectively uptaken by a variety of different cells (Figs. 5d-f, Figs. S14 and S15 in Supporting information). Quantitative analysis of single-cell fluorescence intensity showed that the fluorescence intensity of Eu₉ showed a decreasing trend in cells from HeLa to WI-38, and the specific fluorescence intensity was ranked as HeLa > MDA-MB-231 > MCF-7 > SKOV3 > 4T1 > HepG2 > WI-38. Obviously, different cells have differentiated uptake capabilities for Eu₉ nanoclusters, and the Eu₉ nanoclusters have high-resolution optical imaging effects on multiple cells, indicating the application potential of lanthanide nanoclusters in early cancer diagnosis.

Mitochondria play a central role in cellular metabolism, serving not only as key sites for energy conversion but also participating in the regulation of apoptosis, signal transduction, metabolic homeostasis, and immune defense [40,41]. These functions make mitochondria key to studying cellular function and disease mechanisms. To explore the precise positioning of Eu₉ cluster for specific organelles in cells, commercial mitochondrial green fluorescent probes (Mito-Tracker Green) and lysosomal green fluorescent probes (Lyso Tracker Green) were used for colocalization experiments with Eu₉ cluster (Fig. 6, Figs. S16 and S17 in Supporting information). The experimental results show that the Eu₉ cluster and Mito-Tracker Green show bright red and green fluorescence in the red and green channels, respectively, and the fluorescence signals of the two show extremely high overlap. The correlation of fluorescence intensity of these two channels was quantified by colocalization coefficient, which was as high as 0.96 in MCF-7 cells (Fig. 6c) and 0.93 in MDA-MB-231 cells (Fig. 6d), indicating that there is a strong correlation between the Eu₉ cluster and Mito-Tracker Green (Figs. S17 and S18 in Supporting information). At the same time, we also studied the colocalization imaging experiment of Eu₉ in murine-derived cell 4T1. As shown in Fig. S18, the co-localization coefficient of Eu₉ clusters and Mito-Tracker Green in 4T1 cells was 0.87. In addition, comparative experiments were performed on HeLa cells using Eu₉ cluster and Lyso Tracker Green. Experimental results show that there is only a small overlap area between the red fluorescence generated by the Eu₉ cluster and the green fluorescence generated by Lyso Tracker Green. The calculated colocation coefficient is only 0.4 (Fig. S17). The above experimental results show that the Eu₉ cluster has excellent mitochondrial localization ability and can specifically label mitochondria in cells. In general, mitochondria-specific fluorescent labeled dyes are designed based on the positive electrical properties of cationized molecules such as pyridine derivatives and quaternary ammonium salts. As a new cationic luminescent material—cationic cluster, cluster Eu₉ can be attracted by the mitochondrial membrane potential due to its positive charge characteristics and then gather in the mitochondria. To our knowledge, this is the first time that specific fluorescent labeling application of lanthanide cluster mitochondrial organelles has been realized. This discovery not only enriches the scope of mitochondria-specific fluorescent probes, but also opens a door for the application of lanthanide nanoclusters in suborganelle imaging.

In the field of biomedical science, choosing the right biological model is crucial for biological imaging, disease mechanism exploration, drug response research, and therapeutic method development. Zebrafish and their embryos have significant advantages in biomedical research, which are mainly reflected in the following aspects: highly similar biological characteristics to humans, transparent embryos for easy observation, suitable for high-throughput experiments, and low research costs [42,43]. These advantages make zebrafish an ideal model for studying human disease and drug response. Here, we exposed fertilized zebrafish embryos to 25 μg/mL of Eu₉ and incubated for 1, 2, 3, 6 and 12 h, respectively, to evaluate the imaging ability and distribution characteristics of Eu₉ in zebrafish embryos. As shown in Fig. 7a, after fertilized zebrafish embryos, after 3 h incubation with Eu₉, a slight red fluorescence accumulation occurred in the egg envelope and yolk sac areas, indicating that Eu₉ has the ability to enter the embryo through the chorionic. As the incubation time is extended, the fluorescence intensity gradually increases and the fluorescence distribution range is also expanded accordingly, which indicates that the increase of Eu₉ entering the embryo (Fig. 7a and Fig. S19 in Supporting information). The significant difference in fluorescence intensity of the egg envelope, yolk sac and intraembryonic region revealed that Eu₉ has selective affinity for different tissues inside the embryo.

Immediately afterwards, newly hatched zebrafish were exposed to 25 μg/mL of Eu₉ and incubated for 1, 2, 3, 6 and 12 h, respectively, to evaluate the imaging ability and distribution characteristics of Eu₉ in zebrafish [44]. Fig. 7b shows that newly hatched zebrafish exhibited a slight red fluorescence aggregation after 2 h of incubation with Eu₉, mainly concentrated in the digestive system and yolk sac region of the zebrafish. This phenomenon shows that Eu₉ can enter zebrafish through swallowing and skin absorption mechanisms. At the same time, similar to the zebrafish embryo imaging phenomenon, zebrafish also showed that with the increase in incubation time, the fluorescence intensity in vivo gradually increased, and the fluorescence distribution range expanded accordingly (Fig. 7b). When the incubation time reached 12 h, the zebrafish head showed a faint red fluorescence signal. This may be caused by the fact that Eu₉ is transported to the head area through blood circulation and other channels after being absorbed by the digestive system. The above experimental results show that Eu₉ can efficiently enter zebrafish and its embryos and show optical imaging effects specific to specific tissues. At present, there are still few cases of applying lanthanide nanoclusters to the field of zebrafish imaging. This work not only develops excellent optical imaging probes based on lanthanide nanoclusters, but also opens new horizons for the biomedical application of lanthanide nanoclusters.

Magnetic resonance imaging (MRI) plays an important role in clinical medical diagnosis as a mature medical imaging technology. However, because of the problem of insufficient contrast in some cases, it is often necessary to enhance the imaging effect by using CAs. Therefore, it is necessary to design and develop CAs with high resolution and clear imaging effects [45–47]. T₁-weighted MRI CAs, which are now widely used as clinical diagnostics, are gadolinium-based chelates, and their superior imaging is dependent on higher doses. To explore the feasibility of Gd₉ as a T₁-weighted MRI CA, we tested the relaxation time with the Gd(Ⅲ) concentration changes at 0.5 T magnetic field strength and obtained the longitudinal and transverse relaxation rates (r₁ and r₂, respectively) (Fig. 8). The results show that the values of r₁ and r₂ values of Gd₉ are 39.42 and 50.58 mmol L⁻¹ s⁻¹, respectively, at 0.5 T magnetic field. Gd₉ with a highly aggregated Gd(Ⅲ) exhibited higher relaxation rates compared to the conventional Gd chelates (Fig. 8a). r₂/r₁ = 1.28 (r₂/r₁ < 2) indicated that Gd₉ is a potential T₁-weighted MRI CA candidate. With the gradual increase of the Gd(Ⅲ) concentration of Gd₉ in the aqueous solution, the T₁-weighted image gradually became brighter under 0.5 T magnetic field (Fig. 8b). In contrast, the change in T₂-weighted images was not significant, indicating that Gd₉ exhibits good T₁ imaging in solution. The T₁ mapping images displayed by different Gd₉ concentrations at 0.5 T magnetic field also depicted its great potential as an MRI CA in biomedical diagnosis. Furthermore, we used the MTT method to evaluate the potential toxicity of concentration-dependent Gd₉ solutions (0, 6.25, 12.5, 25, 50, 100 μg/mL) to HeLa and WI-38, respectively. As shown in Fig. S20 (Supporting information), after co-incubation of HeLa and WI-38 cells with concentration-dependent Gd₉ solution for 16 h, the cell survival rate exceeded 75%, demonstrating that Gd₉ has low cytotoxicity (Fig S20 in Supporting information). At the same time, Gd₉ (100 μg/mL, 50 μL/healthy mice) was injected into healthy mice through the tail vein for 48 h, followed by routine blood tests and pathological analysis of H&E staining of major organs. The results of in vivo safety evaluation showed that Gd₉ showed excellent biocompatibility (Figs. S21 and S22). Obviously, high-nuclear gadolinium clusters that highly cluster Gd(Ⅲ) at the molecular level have more attractive prospects in MRI CAs. Therefore, it is crucial to construct high-nuclear gadolinium clusters that are stable and suitable in aqueous solutions for application as novel CAs for MRI.

In summary, the chelating ligand protection strategy was used to synthesize nonanuclear lanthanide nanoclusters Ln₉ that are stable in aqueous solution under solvothermal "one-pot" conditions. The nanosized nonanuclear Eu₉ still shows bright characteristic red emission of Eu(Ⅲ) ions in aqueous solution and can show excellent fluorescence imaging performance for many different types of cells. Notably, the cationic Eu₉ nanocluster can specifically target and label mitochondria in cells. To our knowledge, this is the first time that the cationic properties of lanthanide nanoclusters have been used to achieve specific fluorescence labeling of mitochondria. In addition, nonanuclear Eu₉ can also be taken up by zebrafish eggs and larvae, showing high-resolution in vivo imaging. The longitudinal and transverse relaxation rates r₁ and r₂ of the nonanuclear nanocluster Gd₉ with highly aggregated Gd(Ⅲ) ions are 39.42 and 50.58 mmol L⁻¹ s⁻¹, respectively, showing great potential as an excellent T₁-weighted MRI CA. This work not only opens a door for the application of lanthanide nanocluster emitters in the field of organelle imaging, but also has the potential to develop multimodal imaging applications based on lanthanide nanocluster emitters.