Introduction

Medical imaging in early detection has become crucial in the fight against cancer. Among various molecular imaging techniques, magnetic resonance imaging (MRI) and fluorescence optical imaging (FOI) are the most commonly used imaging approaches1,2,3,4, which can distinguish fine variations in tissues within the body and provide detailed information on the dynamics of cellular interactions5,6,7. However, sometimes, the difference between normal and abnormal tissue is too small to be distinguished, which raises the demand of contrast agents (CAs) to differentiate between target and background for in-vivo bio-imaging8. CAs can depict anatomic details conserving high spatial resolution without the loss of signal and finally maximizing the ability of MRI due to the signal-enhancing positive contrast ability9. Two types of MRI contrast agents, paramagnetic coordination complexes and superparamagnetic nanoparticles, are currently used in practice10. For example, paramagnetic gadolinium (Gd3+) complexes (e.g. gadolinium diethylene-triaminepentaacetate (Gd-DTPA) (r1 = 4 mM−1 sec−1) are most widely used as T1 MRI CAs in clinical practice, which can provide outstanding positive MR images with high resolution11,12,13. Unfortunately, some discoveries have shown Gd3+ could be involved in nephrogenic systemic fibrosis (NSF), which would limit their megadose clinical applications14. Other biocompatible materials such as manganese oxide nanoparticles (e.g. water-dispersible MnO with r1 = 0.21 mM−1 sec−1 and MnO hollow particles with r1 = 0.353 mM−1 sec−1) could also be used as T1 contrast agents. However, extremely low longitudinal relaxivity (r1) of manganese oxide nanoparticles has severely hindered their applications as CAs for the cytotoxicity associated with a high-dose nanoparticles required15,16. Typical T2 CAs are superparamagnetic iron oxide nanoparticles (SPIO), which can provide very sensitive T2-weighted images to detect lesions from normal tissues17,18. However, T2 agents provide dark negative signal intensity in images, which is often confused with the signals from bleeding and calcification19. To overcome these disadvantages, major efforts have been made in recent years to integrate T1 and T2 CAs together to form dual modal contrast agents (DMCAs), T1-weighted imaging is very sensitive at finding a diseased tissue, while T2-weighted imaging can define tumors in clinic applications20. Currently, DMCAs are prepared through the integration of iron oxide nanoparticles and Gd species in a “core-shell” format21. Recently, Fe3O4/MnO hybrid nanocrystals have also been prepared as DMCAs22. Unfortunately, it was reported that the “core-shell” structure is apt to cause undesirable interactions between the superparamagnetic and paramagnetic nanoparticles, which can decrease the relaxivity of T1 signal23. Although it has been reported that monodisperse Gd2O3-embedded iron oxide (GdIO) nanoparticles as DMCAs with high r1 values is a solution to this problem, the mixture of precursors cannot ensure Gd2O3 uniformly distributed in iron oxide nanoparticles. Furthermore, the synthetic process of core-shell DMCAs is complicated and the surfactants used to assist synthesis is difficult to be removed, which may increase potential hazards to human health.

Two-photon laser scanning microscopy (TPLSM) has also become an indispensable tool for direct observation of cellular structure and biological process with the advantages of larger penetration depth in biological tissues, more reduction of photobleaching and weak autofluorescence24. The cellular imaging application of TPLSM has been investigated using inorganic quantum dots (QDs), metallic nanoparticles and organic dyes or pigments as probes25,26,27. Even though various targeting ligands have been labeled with these probes to delineate tumor margins to improve surgical outcome, preoperative diagnostic images cannot be correlated with intraoperative pathology28.

In a word, each imaging modality has advantages and limitations regarding sensitivity, resolution and tissue penetration of signal. Dual modal imaging combining MRI/fluorescent probes can overcome the limitations of each modality29. Researchers usually prefer to conjunct fluorescent organic dyes with already prepared MRI CAs to develop such dual probes. For example, such probes have been synthesized by combining mesoporous silica-coated hollow manganese oxide nanoparticles with rhodamine B isothiocyanate (RITC) or Ru(bpy)32+ (fluorescent probes)30,31, doping SPIO nanocrystals with fluorescein isothiocyanate (FITC)32 and coating Fe3O4 cores or MnO nanoflowers with Au shells33,34. It is reported that the combination of SPIO nanoparticles with QDs or noble metal nanoparticles (e.g. Ag) can also produce dual functional CAs35,36. However, there are several drawbacks of such composite nanoparticles, which include: 1) The combining process is complicated; 2) The surfactants used during the preparation, e.g. CTAB and EDAC, are known to be cytotoxic37,38; 3) The attached fluorescent probes can drop off, with the potential of causing damage to human health; 4) The organic dye is not stable enough, for example, FITC is easily quenched in an acidic buffer; 5) Even though various post-synthesis modifications have been done, the cytotoxicity of some composite nanoparticles is still high.

The best way to overcome the problems highlighted above is to combine all those MRI and FOI properties in single-material nanoparticles instead of composite nanoparticles. The overall goal of this study is to develop a single-material nanoparticles-based platform for multimodal imaging. It has been reported that Prussian blue nanoparticles exhibit a T1 contrast effect, the unique structure of metal organic frameworks (MOFs) may bring strong interactions between ion centers and water protons, inducing MRI contrast enhancement39. Taking into consideration the MRI properties of hollow MnO nanoparticles40, it is reasonable to suggest that MOF compounds containing Mn2+ ions have the potential to be used as DMCAs. Moreover, Mn2+ doped QDs represent a class of phosphors that have already been utilized for many applications41,42. Thus, nanocubes of Mn3[Co(CN)6] were prepared and their applications as multimodal imaging contrast agents were explored. To further enhance their biocompatibility, silica coating process has been applied to form core-shell nanocubes, Mn3[Co(CN)6]2@SiO2.

Results

Mn3[Co(CN)6]2 nanocubes were synthesized at room temperature43 and coated by treating the nanocubes with TEOS under sol-gel conditions (Supporting Information, Scheme 1). Field-effect scanning electron microscope (FE-SEM) images showed that uniform Mn3[Co(CN)6]2 and Mn3[Co(CN)6]2@SiO2 nanocubes with the size around 150–180 nm and 160–200 nm were obtained, respectively ( Figure 1 ). Some cracks (shown by the red arrow in Figure 1b ) on the surface appeared along with the increase of pore size after silica coating to form Mn3[Co(CN)6]2@SiO2 nanocubes ( Supplementary Table 1 ). The molar ratio of Mn to Co was measured to be approximately 3:2 ( Supplementary Table 2 ) by inductively coupled plasma-atomic emission spectrometry (ICP-AES) and was consistent with the X-ray powder diffraction (XRD) phase analysis ( Supplementary Fig. 1 ). The silica coating cannot be detected by XRD due to its amorphous nature with a very weak and broad diffraction band at 20°, however, the peak at 102.90 eV in X-ray photoelectron spectroscopy (XPS) spectra corresponds to chemical bonding of Si-O, which confirmed the existence of silica ( Supplementary Fig. 2 ). Transmission electron microscopy (TEM) images were applied to substantiate the core-shell format of Mn3[Co(CN)6]2@SiO2 nanocube ( Supplementary Fig. 3a, b and inset). Energy Dispersive Spectrometer (EDS) pattern comfirmed the surface element distribution of manganese (Mn), cobalt (Co), silicium (Si) and oxygen (O) in Mn3[Co(CN)6]2@SiO2 nanocube, showing the existence of the silica layer ( Supplementary Fig. 3c ). Both uncoated and silica-coated Mn3[Co(CN)6]2 nanocubes displayed relatively high surface area and large pore size, measured by nitrogen absorption–desorption isotherms at 77 K ( Supplementary Fig. 4 ). The nanocubes can form stable colloidal suspensions in aqueous media, which can be used days later ( Supplementary Fig. 5 ).

Figure 1
figure 1

(a) SEM image of Mn3[Co(CN)6]2 nanocubes; (b) SEM image of Mn3[Co(CN)6]2@SiO2 nanocubes.

Full size image

Biocompatibility without cytotoxicity is an important property of materials for biomedical applications. To evaluate the cytotoxicity of the as-prepared nanocubes, 3-(4,5-Dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) assays were carried on HeLa, HepG2, A549 and glioma C6 cell lines. As shown in Figure 2 , these cell lines remain highly viable after being incubated for 24 h, indicating that for the tested concentrations there is no cytotoxicity. Although Mn, Co and CN− may be toxic, their toxicity can be significantly reduced when forming coordination compounds such as Mn3[Co (CN)6]2 nanocubes that are composed by stable chemical bonds. What's more, the biotoxicity of silica-coated Mn3[Co(CN)6]2 nanocubes can be further reduced, especially at higher concentrations.

Figure 2
figure 2

Cytotoxicity test of as-prepared Mn3[Co(CN)6]2 and Mn3[Co(CN)6]2@SiO2 nanocubes on the viability of HeLa, HepG2, A549 and glioma C6 cells, as measured by MTT assay.

Full size image

Uncoated and silica-coated Mn3[Co(CN)6]2 nanocubes as DMCAs for MRI have been investigated both in-vitro and in-vivo. The MRI signal intensity of samples was measured by a 3T MRI scanner. Uncoated and silica-coated Mn3[Co(CN)6]2 nanocubes were respectively dispersed in pH 7.4 phosphate buffered solutions (PBS) with different concentration of Mn2+ to determine the longitudinal (r1) and transverse (r2) relaxivities. To quantitatively characterize their magnetic resonance properties, the concentration of Mn2+ was measured by ICP-AES. The in-vitro MR images ( Figure 3a, b, c, d ) demonstrated the relationship between the concentration of Mn2+ ions and MR imaging performance. On a per mM Mn2+ basis, uncoated and silica-coated Mn3[Co(CN)6]2 nanocubes exhibited r1 of 4.798 mM−1s−1 and 10.52 mM−1s−1 and r2* of 89.5572 mM−1s−1 and 173.09 mM−1s−1, respectively. It is particularly notable that the minimum dosage that can manifest the contrast over water proton is only 2.5 μg/mL for both types of nanocubes, which shows negligible toxicity at this concentration in cytotoxicity test. HeLa cells were also incubated with various concentrations of Mn3[Co(CN)6]2@SiO2 nanocubes for 24 hours ( Supplementary Fig. 6 ) for further MRI tests. The results obtained are in accordance with results from the in vitro MRI tests. It is demonstrated for the first time that Mn3[Co(CN)6]2 nanocubes have the potential to be used as effective T1 & T2 MRI contrast agents.

Figure 3
figure 3

In vitro MRI test.

(a), (b) T1 and T2* weighted MR images of Mn3[Co(CN)6]2 and Mn3[Co(CN)6]2@SiO2 nanocubes with different concentrations of Mn, respectively; c), d) T1 and T2* relaxation rates as a function of Mn concentration of Mn3[Co(CN)6]2 and Mn3[Co(CN)6]2@SiO2 nanocubes, respectively.

Full size image

Effects of Mn3[Co(CN)6]2@SiO2 nanocubes on in-vivo MR imaging were studied in mice with C6 glioma bearing brain tumor with a suspension of 0.5 mL Mn3[Co(CN)6]2@SiO2 nanocubes injected (1.25 mg nanocubes per kilogram of mouse body weight). As the blood brain barrier is destroyed as a result of the tumor formation in this animal model, Mn3[Co(CN)6]2@SiO2 nanocubes enter the tumor readily and are retained there for long periods. Thus, the brain cancer cells were selectively enhanced in T1 MRI because Mn3[Co(CN)6]2@SiO2 nanocubes were injected through tail vein and accumulated at the tumor by strong cytophagy effect of the cancer cells. The localization of nanoparticles in brain tumor tissue was confirmed by ex-vivo ICP measurements, with the concentrations of Mn and Co at approximately 0.19 μg/ml and 0.13 μg/ml, respectively. Based on the images taken from coronal plane ( Figure 4 ), within 30 minutes after injection, obvious T1 contrast effects can be observed at the tumor and remain for 24 hours as the nanocubes accumulating. However, influenced by the bleeding and edema in brain tissue, T2 effect was not clearly detected. Therefore, in order to evaluate T2 contrast in tumor imaging of the nanocubes, a perfusion weighted imaging (PWI) was used ( Supplementary Fig. 7 ). The T2 signal was measured every second after the injection of Mn3[Co (CN)6]2@SiO2 nanocubes through tail vein. The signal decreased sharply 11 s post injection, reached the lowest value at 16 s, then returned back immediately, confirming the T2 relaxation effects of as-prepared nanocubes. When we compared the signal intensity of brain tumors with normal brain tissue, it was observed that the signal amplitude of T2 decreased over time because of the abundant vessel and large blood volume at the tumor. It needs to be emphasized that the nanoparticles appeared unlikely to cause acute or severe toxic effects to cause sudden death in mice during in-vivo experiments. No obvious dose-dependent body weight loss in mice was observed. Mn3[Co(CN)6]2@SiO2 nanocubes demonstrated low acute toxicities, thus, can be used as contrast agents.

Figure 4
figure 4

MRI signal time courses of T1 (upper) and T2 (lower) of a mouse brain bearing the glioma C6 brain tumor that was intravenously injected a suspension of 0.5 mL Mn3[Co(CN)6]2@SiO2 nanocubes in preinjection, 1 h and 24 h after injection.

Full size image

In order to further demonstrate Mn3[Co(CN)6]2@SiO2 nanocubes with a high r1 relaxivity, we also investigated in-vivo vessel wall imaging of a mouse. Mn3[Co(CN)6]2@SiO2 nanocubes (1.25 mg nanocubes per kilogram of mouse body weight) were injected into a mouse via the tail vein and measured by MR angiography immediately. Blood vessels were clearly seen on T1-weighted images, displaying outstanding T1 relaxation properties of Mn3[Co(CN)6]2@SiO2 nanocubes as intravascular MRI CAs ( Figure 5b ). Dynamic changes in signal intensity in the kidney were also clearly observed on T1-weighted imaging after the injection ( Figure 5a ). By comparing each pair of MRI images before and after Mn3[Co(CN)6]2@SiO2 nanocubes injection, a signal increase is observed on T1-weighted MR images of a mouse kidney 5 minutes post injection, then the image signals remain stable for 30 minutes. These results demonstrate for the first time that Mn3[Co(CN)6]2@SiO2 nanocubes have potential for use as T1 & T2 MRI contrast agents.

Figure 5
figure 5

In vivo MRI images of a mouse injected with Mn3[Co(CN)6]2@SiO2 nanocubes.

(a) The time course of the signal enhancement in T1 MRI of a mouse kidney after intravascular injection of Mn3[Co(CN)6]2@SiO2 nanocubes; (b) Intravascular MR angiography images in postcava of the mouse after injecting Mn3[Co(CN)6]2@SiO2 nanocubes through tail vein.

Full size image

The nanocubes were incubated with HeLa cell lines for 24 h without any further dyeing for TPLSM test by confocal laser scanning microscopy (CLSM). It was observed that uncoated and silica-coated Mn3[Co(CN)6]2 nanocubes were localized in cytoplasm and the fluorescence from unstained cells excited at various wavelengths were caused by the nanocubes themselves ( Figure 6 ). The emission spectrum in the wavelength range 400 to 720 nm (λex = 720 nm, two photons excitation) was measured by two-photon excited fluorescence microscopy (TEFM) ( Supplementary Fig. 8 ). Fluorescence spectra of uncoated and silica-coated Mn3[Co(CN)6]2 nanocubes in the concentration of 250 μg/mL also confirmed that the fluorescence was caused by Mn2+ ion, whose emission peaks were around 550–600 nm. The excitation spectra were determined by varying the excitation wavelength to find the maximum emission intensity at 571 nm ( Supplementary Fig. 9a,b ). Two peaks at 355 nm and 475 nm detected in the excitation spectrum were assigned with 6A1(6S) → 4E(4D) (350 nm, 3.49 eV) and 6A1(6S) → 4E, 4A1(4G) (475 nm, 2.61 eV) transition of Mn2+, respectively ( Supplementary Fig. 9c,d ). Thus, the luminecence mechanism of as-prepared CAs can be well explained.

Figure 6
figure 6

CLSM images of Mn3[Co(CN)6]2 (a) and Mn3[Co(CN)6]2@SiO2 (b) nanocubes in bright-field, two-photon fluorescence excited by 720 nm femtosecond laser pulses and CLSM excited with 488 nm and 403 nm laser beams, respectively.

Full size image

Discussion

MRI is currently one of the most powerful imaging technologies both in biological investigations and medical diagnosis. As a new MRI CA, the r1 value of Mn3[Co(CN)6]2@SiO2 nanocubes is much higher than Mn-based CAs. Even though the relaxivity is not as high as that of Gd3+ with various ligands (higher than 10 mM−1s−1)44, it is higher than the average value of MnO (lower than 1 mM−1s−1)45,46. It is interesting to understand why the high r1 value and r2/r1 relaxivity ratio can be generated in the nanocubes. As we know, the interaction dynamics between water and Mn2+ complexes highly affects the relaxivity, while the mechanism remains unclear. Based on Solomon-Bloembergen-Morgan (SBM) theory47,48, direct water coordination to the paramagnetic metal centers is the major contributor to T1 inner-sphere relaxivity49. Mn3[Co(CN)6]2 nanocubes have a face-centered cubic structure in which two different ion centers Mn2+ and Co3+ are bridged by CN− groups to form an extended 3D covalent coordination lattice structure39. As a result, there are a large amount of original pores and voids inside the structure confirmed by pore size distribution measurements. The vacant sites in the lattice are usually occupied by water molecules, which are accessible for water motion, therefore, a well-working inner sphere T1 relaxation of the nanocubes and water molecules can be generated. What's more, the Mn2+ ions are positioned in the body centers and edge centers in the face-centered cubic structure, which is free to adjunct with water molecules to form a coordination sphere with each Mn2+ ion surrounded with water molecules, thus, the outer sphere T1 relaxation has also been generated. With the simultaneous action of inner and outer sphere relaxation, Mn3 [Co(CN)6]2 nanocubes manifest an enhanced relaxivity. Moreover, it is suggested that the higher relaxivity of Mn3[Co(CN)6]2 @SiO2 nanocubes is benefited from the silica coating, which allows better water molecules exchange with the metal ions. It has been reported that water molecules diffusing anisotropically inside the silica, with the fastest diffusion component occurring along the channels in the silica, is more rapid than that of the isotropic diffusion in the surrounding environment30. Nanopores in Mn3[Co(CN)6]2@SiO2 nanocubes reveraled by nitrogen absorption isotherms measurement allow water molecules diffusion to the magnetic cores, leading to efficient relaxation of water molecules in the vicinity of the nanoparticles.

Upon biphotonic excitation at 720 nm, both Mn3[Co(CN)6]2 and Mn3[Co(CN)6]2@SiO2 nanocubes displayed outstanding fluorescence in the visible spectrum. This TPF imaging technique provides the possibility of long-term imgaing of cellular process with reduced photodamage compared to UV-exicited imaging48. Recently, it is reported that the photoluminescence of 4T1 → 6A1 transition of Mn2+ can provide efficient two-photon fluorescent in doped QDs, such as Mn-doped ZnS QDs42, which demonstrates the possibility of Mn2+ ions used in TPF imaging. Compared with Au nanodots with the same fluorescent intensity50, the concentrations of uncoated and silica-coated Mn3[Co(CN)6]2 nanocubes are much lower than that of the reported human insulin-Au nanodots for cellular imaging. Thus, all the results above demonstrate that both uncoated and silica-coated Mn3[Co(CN)6]2 nanocubes can be efficiently used as TFP imaging probes. The use of near-infrared-emitting probes could enhance both the tissue depth penetration and imaging sensitivity.

The integration of conventional MRI and TPF probes into a single material such as MOF structured compounds provide a new way to design and synthesis CAs. The advantages of the as-prepared nanoparticles is that they have low toxicity and long circulation time, suggesting their suitability for clinical applications. Most fluorescent probes might be harmful. For instance, Cd-based QDs are used as excellent fluorescent probes, however these nanocrystals have two intrinsic limitations: photooxidation of QDs can cause the release of cadmium and photoexcited QDs can produce reactive oxygen species (ROS). As a result, a doubt has been shed on the future applicability of Cd-containing QDs, particularly in view of environment regulations51. While, in Mn3[Co(CN)6]2 nanocubes, Mn, Co and CN− were composed by strong chemical bonds, which can effectively get rid of the toxicity, resulting in better biocompatibility than some reported CAs52. Furthermore, after coated with SiO2, the toxicity decreases remarkably. Since all the CAs investigated nowadays might aggregate in body to cause some long term toxicity, a single material with low cytotoxicity would be more friendly in bio-applications.

In conclusion, we have prepared efficient CAs, Mn3[Co(CN)6]2 nanocubes, which show dual contrast enhancement of both T1- and T2-weighted MRI and TPLSM fluorescence imaging. This is the first report that an individual material can act as multimodal contrast agents, overcoming major limitations of current multimodal imaging probes prepared by incorporating QDs and magnetic nanoparticles or paramagnetic Gd ions. Surface modification with silica to form Mn3[Co(CN)6]2@SiO2 core-shell nanocubes was achieved, much better biocompatibility was observed in the core-shell nanocubes by cellular cytotoxicity test, making them suitable for biomedical applications.

Methods

In vitro MR imaging

Magnetic Resonance Images (MRI) were acquired by magnetic resonance scanner (GE Signa HDxt 3.0 Tesla MRI system). T1-weighted MR images were taken by using a saturation recovery using spin-echo sequence (TE = 10 ms, TR = 4000, 2000, 1000, 500, 200, 100 ms,respectively). T2*-weighted images were also obtained by Carr-Purcell-Meiboom-Gill method with the RARE sequence using the parameter of TR = 120, TE = 2.328, 6.112, 9.896, 13.68, 17.46, 21.24 ms, the filp angle = 30°, bandwidth = 31.25 Hz, FOV 180 × 180 mm2, slice thickness = 3 mm without gap.

In vivo MR imaging

Sprague-Dawley (SD) mice weighting 250–280 g were imaged on a 3 T MRI system (GE Signa HDX 3.0 T). High-resolution Mn3[Co(CN)6]2@SiO2 nanocube contrast enhanced multi-slice MR images were obtained from each mouse brain using a fast spin-echo T1-weighted MRI sequence (repetition time (TR)/echo time (TE) = 780/19.6 ms, number of exitations (NEX) = 2, echo train length = 2, 0.188 × 0.188 mm2 in plane resolution with a slice thickness of 2 mm and 10 slices) and a fast spin-echo T2-weighted MRI sequence (repetition time (TR)/echo time (TE) = 3000/110 ms, number of exitations (NEX) = 2, echo train length = 2, 0.188 × 0.188 mm2 in plane resolution with a slice thickness of 2 mm and 10 slices).

All the animal experiment were permitted by the Ethical Committee of the experimental animal center of Medical University of Anhui, China. All the animal experiments and feeding were carried out in accordance with the guidelines of the Ethical Committee of the experimental animal center of Medical University of Anhui, China.

CLSM and TPF measurement

HeLa cells were seeded onto sterile, acid-treated 12-mm coverslips in 24-well plates (Corning Glass Works, Corning, NY, USA). Images were taken with a laser scanning microscope (Zeiss L SM 710) using a 63_1.3 numerical aperture PlanApo objective.