Synthesis and characterization of ES-MIONs
The ES-MIONs were synthesized by a method of co-precipitation, and reaction conditions were optimized to obtain high quality ES-MIONs with high r1 and r2/r1 (Additional file 1: Table S1). PASP was used as a stabilizer for the ES-MIONs preparation, which gives the obtained ES-MIONs excellent water dispersibility. Four concentrations of PASP solutions were used for synthesis of ES-MION1-4. The Fe concentration of ES-MIONs was determined by inductively coupled plasma-optical emission spectrometry (ICP-OES), and the ES-MION2 has the largest Fe recovery of 96.6% (Additional file 1: Table S1). T1 and T2 relaxation rates (3.0 T) versus Fe concentration of ES-MION1-4 are shown in Fig. 1A, B. The r1 and r2 values are obtained from the linear line slopes, which are summarized in Fig. 1E and Additional file 1: Table S1. The ES-MION2 has a r1 value of 1.6 mM−1 s−1 and r2/r1 ratio of 8.8. Though the r1 of ES-MION3, 4 is larger than ES-MION2, the r2/r1 values of ES-MION3, 4 are also much higher than that of ES-MION2, which are not good for T1 imaging. The r2/r1 value of ES-MION1 is lower than that of ES-MION2, but the r1 value is also lower than that of ES-MION2. Therefore, 2.0 mg/mL of PASP solution was considered as the optimal concentration for the synthesis of ES-MIONs.
Furthermore, 0.5–8.0% of ammonia solutions were used to synthesize ES-MION5-8, whose T1 and T2 relaxation rates (3.0 T) as a function of Fe concentration are shown in Fig. 1C, D. As shown in Fig. 1F and Additional file 1: Table S1, the r1 and r2/r1 of ES-MION6 are comparable to those of ES-MION5, but much better than ES-MION 7, 8. Therefore, 4.0% of ammonia solution was chosen as the optimal condition.
In addition, based on the optimized conditions for ES-MION6 synthesis, the concentration of PASP and iron precursors (FeCl3 plus FeSO4) were all decreased to synthesize ES-MION9-11. From Fig. 1C, D, F and Additional file 1: Table S1, it can be found that ES-MION9 has a highest r1 value of 7.0 ± 0.4 mM−1 s−1 (3.0 T) and a lowest r2/r1 value of 4.9 ± 0.6 (3.0 T) compared with ES-MION6, 10, 11. According to Eq. (1) [28], the signal intensity of MRI is depended on gradient intensity (M0), echo time (TE), repetition time (TR), flip Angle (α), R2* and R1. The factors of M0, TE, TR, and α could be regulated by MRI scanners, while R2* and R1 depend on contrast agents. The R2* can be considered a valid R2 and is always greater than or equal to R2. It can be concluded that the T1 MRI signal intensity is proportional to r1 value, but inversely proportional to r2/r1 ratio. Thus, the synthesis conditions of ES-MION9 should be optimal to obtain a high T1 MRI capability with a high r1 and low r2/r1.
$${\text{Signal intensity = M}}_{{0}} {\text{sin(}}\alpha {)}\frac{{1 - {\text{e}}^{{ - {\text{R}}_{{1}} \cdot {\text{TR}}}} }}{{1 - {\text{cos(}}\alpha {)} \cdot {\text{e}}^{{ - {\text{R}}_{{1}} \cdot {\text{TR}}}} }}{\text{e}}^{{ - {\text{R}}_{{2}}^{*} \cdot {\text{TE}}}}$$
(1)
Besides, Fe recoveries of ES-MION1-11 tested by ICP-OES are all above 85%, indicating high utilization rates of raw materials and low cost for ES-MIONs synthesis, which are beneficial for clinical transformation.
According to previous reports, Fe3O4 nanoparticles with size below 5.0 nm can be used as T1 CAs [24]. Furthermore, Fe3O4 nanoparticles with large particle size are easily taken up by the spleen and liver, which seriously affects tumor images. The images of transmission electron microscopy (TEM, Fig. 2A–K) indicate our ES-MION1-11 have excellent water dispersibility. It is found from the TEM images (Fig. 2A–D) and size distributions (Additional file 1: Fig. S1A–D) measured from TEM images that the concentration of PASP has a large influence on the sizes of ES-MIONs. The sizes of ES-MION1-4 are respectively 2.7, 2.5, 6.0 and 8.0 nm, whose r1 is 1.0, 2.0, 4.7, and 5.4 mM−1 s−1, and the r2/r1 is 1.9, 7.0, 19.0, and 28.3. These results demonstrate that Fe3O4 nanoparticles with size below 5.0 nm have potential as T1 CAs, while those with size larger than 5.0 nm can be only utilized as T2 CAs due to the high r2/r1 ratios. Figure 2E–K and Additional file 1: Fig. S1E–K show that both the concentration of ammonia solution and the whole concentrations of feeding materials have a slight influence on the size of ES-MIONs. The relationships between the particle size and r1 value (or r2/r1 ratio) (Fig. 2L) show that the best particle size is 3.7 nm (ES-MION9).
Three batches of ES-MION9 were synthesized and the T1/T2 relaxation rates were determined by a 3.0 T (Additional file 1: Fig. S2) and 7.0 T MRI scanner (Additional file 1: Fig. S3), whose similar r1 and r2 data for different batches demonstrate the good repeatability for ES-MION9 synthesis. At 3.0 T, the ES-MION9 has a larger r1 (7.0 ± 0.4 mM−1 s−1) than Gadavist (4.9 ± 0.1 mM−1 s−1), indicating a stronger T1 MRI capability of our ES-MION9.
The related T1-weighted MR images (3.0 T) of ES-MION1-11 are shown in Additional file 1: Figs. S4A, S5A, and S6A. The corresponding SNR and ΔSNR values were calculated according to Eqs. (2) and (3) [29, 30], and shown in Additional file 1: Figs. S4B, S5B, and S6B, which reinforce that the signal intensities of MR images increase with the increase of Fe concentration with a strong concentration gradient dependence, showing good T1-weighted MR capabilities of ES-MION1-11.
$${\text{SNR}} = \frac{{{\text{SI}}_{{{\text{mean}}}} }}{{{\text{SD}}_{{{\text{noise}}}} }}$$
(2)
$$\Delta {\text{SNR}} = \frac{{({\text{SNR}}_{{{\text{sample}}}} - {\text{SNR}}_{{{\text{control}}}} )}}{{{\text{SNR}}_{{{\text{control}}}} }} \times 100\%$$
(3)
It is obvious that the ΔSNR value of ES-MION9 is the maximum up to 5500% when the Fe concentration of is 1.0 mM (Additional file 1: Fig. S6B), which further demonstrate 3.7 nm is the best diameter of ES-MIONs for T1 MRI.
The T1 images (3.0 T) of ES-MION9 solution at 1.0 mM were further compared with the commercial Gadavist at 1.0 mM of Gd concentration (Fig. 3A). It can be seen from Fig. 3B that the ΔSNR (5400%) of ES-MION9 is higher than that (4600%) of Gadavist (***P < 0.001), which demonstrates the better MR imaging capability of our ES-MION9 (r1 is 7.0 mM−1 s−1, r2/r1 is 4.9, 3.0 T) compared with the Gadavist.
A 7.0 T of MRI scanner was also used to double confirm the T1-weighted MRI contrast of ES-MION9 solutions at various concentrations compared with pure water (Additional file 1: Fig. S7A). The corresponding ΔSNR values (Additional file 1: Fig. S7B) also show a strong concentration gradient dependence, indicating a strong MRI capability at 7.0 T.
The ES-MION9 HR-TEM image is presented in Additional file 1: Fig. S8A. The lattice planes of 311 and 220 can be confirmed by the 0.51 and 0.301 nm of interplanar distances [31], indicating a crystalline structure of ES-MION9. The characteristic peaks of O and Fe can be found in the EDS (Additional file 1: Fig. S8B), demonstrating the component of iron oxide for ES-MION9 [32]. To further demonstrate the successful synthesis of Fe3O4 nanoparticles, the X-ray photoelectron spectroscopy (XPS) of ES-MION9 is performed in Additional file 1: Fig. S8C. The primary peaks at 723.8 and 710.3 eV correspond to the energy of Fe 2p3/2 and Fe 2p1/2 [33, 34], indicating the Fe3O4 component of our ES-MION9 [23]. Additional file 1: Fig. S8D shows the XRD of ES-MION9. Four characteristic peaks (2θ ≈ 30.0°, 35.2°, 42.8°, and 53.0°) match with the indices [(220), (311), (400), and (511)]. The crystal structure of ES-MION9 matches the pristine of Fe3O4, demonstrating the high crystalline purity of our ES-MION9. The field dependent magnetization curve (Additional file 1: Fig. S8E) indicates the ES-MION9 is superparamagnetic with 16.0 emu/g of saturation magnetization (Ms). All these results indicate that the ES-MION9 we synthesized is superparamagnetic Fe3O4 nanocrystals.
Because the Ms values of ES-MIONs increase with the increasing particle sizes [28], the small Ms value of ES-MION9 indicates its small particle size. In Eq. (4), the r is the magnetic core radius and Ms is the saturation magnetization. According to Eq. (4), both the extremely small particle size (3.7 nm) and small Ms (16.0 emu/g) lead to a very low r2, which results in a very low r2/r1. Therefore, our exceedingly small ES-MION9 can be used as T1 CA.
$$\frac{1}{{{\text{T}}_{2} }} = \frac{{(256\uppi ^{2}\upgamma ^{2} /405){\text{V}}^{*} {\text{M}}_{{\text{S}}}^{2} {\text{r}}^{2} }}{{{\text{D}}(1 + {\text{L}}/{\text{r}})}}$$
(4)
The high r1 value of ES-MION9 is mainly due to the following two reasons: (1) ES-MION9 has a small particle size (3.7 nm), which gives ES-MION9 a larger specific surface area. In accordance with the mechanism of inner-sphere, larger specific surface area means there are more naked iron on ES-MION9 surfaces, which can fully interacts with hydrogen protons in H2O molecules, resulting in a high r1 value. (2) There are excessive carboxyl groups on ES-MION9 surfaces, and these carboxyl groups are derived from PASP, which greatly improves the water dispersion of ES-MION9. This leads to more H2O in the inner sphere that can interact with the naked iron on the ES-MION9 surface, which causes a large number of bound H2O (q) and mole fraction of H2O coordinated to Fe (Pm) in Eq. (5) [16]. The large q and Pm result in a large r1 value for ES-MION9.
$$\frac{1}{{{\text{T}}_{1} }} = \frac{{{\text{q P}}_{{\text{m}}} }}{{{\text{T}}_{{1{\text{m}}}} + \tau_{{\text{M}}} }}$$
(5)
The T1/T2 relaxation rate (1/T1 or 1/T2) is plotted versus concentration for contrast agents, and the r1 and r2 values are calculated from the slopes of the corresponding fitting lines. T1 CAs increase signal intensity of T1 images by shortening the longitudinal relaxation time (T1) of protons, which leads to high r1 values. The Fe3O4 nanoparticles with size below 5.0 nm have low Ms values causing low r2 values according to Eq. (4). Both high r1 and low r2 result in low r2/r1. Therefore, the 3.7 nm of ES-MION9 (< 5.0 nm) could be utilized for T1 MRI [35, 36].
The hydrodynamic size (dh) of ES-MION9 is 13.7 nm (Additional file 1: Fig. S9A), which is larger than renal filtration threshold (~ 8 nm). The slightly larger hydrodynamic diameter prolongs blood circulation time overcoming the limited MRI time window problem of commercial Gd chelates. The zeta potential of ES-MION9 was measured to be − 55.0 mV (Additional file 1: Fig. S9B), which is due to the presence of excessive carboxyl groups on the surface. Charge plays a key role in the behavior of intravenously injected nanoparticles and pharmacokinetics. For example, nanoparticles agglomerate under charge-mediated nonspecific binding to serum proteins. Sufficient negative charges can avoid the agglomeration of ES-MION9 while avoiding uptake of the nanoparticles by normal cells during blood circulation, resulting in more accumulated ES-MION9 in tumors. Additional file 1: Fig. S9C shows that the hydrodynamic diameter of ES-MION9 do not change significantly during storage in water, 10.0% FBS and 0.9% NaCl solution for 1 week, demonstrating the great stability of ES-MION9.
Additional file 1: Fig. S10 shows UV–vis absorption spectra for ES-MION1-11, which are similar with that of reported ES-MIONs stabilized with other polymers [23]. Additional file 1: Fig. S11 shows the FT-IR of PASP and ES-MIONN9. The stretching vibration peak of –CH2– at 1400.6 cm−1 can be seen from the FT-IR of PASP and ES-MION9, indicating the existence of PASP on the surface of ES-MION9 [37]. In addition, the stretching vibration peak of Fe–O at 604.5 cm−1 can be seen from the FT-IR of ES-MION9, but not in the FT-IR of PASP, indicating the existence of iron in ES-MION9. These results prove the successful synthesis of Fe3O4 [38]. Additional file 1: Fig. S12 presents the curves of thermogravimetric analysis (TGA) and differential thermogravimetry (DTG) for ES-MION9. As the temperature increases, the mass of ES-MION9 continues to decrease, and becomes stable at 37.8% of remaining mass. This is similar to 40.1% of Fe3O4 loading content for ES-MION9 measured by ICP. This result further demonstrates the existence of PASP on the ES-MION9 surface.
Cellular uptake, cytotoxicity assay and T1-weighted imaging of cells
To evaluate the biosafety of ES-MION9, its cytotoxicity was examined by thiazolyl blue tetrazolium bromide (MTT) assay on MCF-7 cells (Human breast cancer cells) and 4T1 cells (Mouse breast cancer cells). Figure 4A, B shows that when the Fe concentration of ES-MION9 reaches 0.8 mM, the cell viability of MCF-7 cells and 4T1 cells was higher than 95.0%. This result indicates that ES-MION9 is almost not cytotoxic due to its biocompatible components (i.e., Fe3O4 and PASP). Although Gd3+ can cause nephrogenic systemic fibrosis and can be deposited in the human brain and body [39], Fig. 4A, B shows that the Gadavist is also non-toxic at the Gd concentration of 0.8 mM. That’s because Gd3+ leads to long-term toxicity, which cannot be revealed in the short-term MTT assay.
To further demonstrate the non-cytotoxicity of ES-MION9, live/dead cytotoxicity analysis was used to evaluate the toxicity of ES-MION9 to 4T1 cells and MCF-7 cells (Additional file 1: Figs. S13, S14). The PBS treated cells were used as a control. Green dots represent live cells and red dots represent dead cells. Obviously, almost no dead cells are found for ES-MION9-treated 4T1 cells and MCF-7 cells, showing good biosafety of ES-MION9. That’s because the main component aspartic acid (ASP) is one of the 20 essential amino acids and iron is one of the essential elements in the human body.
Figure 4C shows the LSCM images of 4T1 cells treated with ES-MION9@R6G. The red signal represents R6G@ES-MION9. After 2 h of co-incubation with 4T1 cells, lots of ES-MION9 nanoparticles were found inside the cells (Fig. 4C). The uptake of ES-MION9 by 4T1 cells was further investigated by flow cytometry. After 2 h of co-incubation with 4T1 cells, the fluorescence intensity (Additional file 1: Fig. S15A, B) of R6G-labeled ES-MION9 was almost two orders of magnitude higher than that of the control group with a statistical P value smaller than 0.001, indicating that ES-MION9 is easily taken up by 4T1 cells. The results of flow cytometry are consistent with the LSCM results. In addition, the T1-weighted MR images (7.0 T) (Additional file 1: Fig. S16) show that ES-MION9-treated tumor cells have much stronger MRI signals compared to the control groups, and the MR signal also increases with the increase of incubation time from 1.0 to 2.0 h. These results demonstrate the excellent MR imaging capability of our ES-MION9 at the cellular level.
In vivo MR imaging
MRI can be used for soft tissue imaging, especially for tumor diagnosis. MR contrast agents can improve the signal-to-noise ratio and sensitivity of MRI. We tested the imaging ability of ES-MION9 in 4T1 tumor-bearing mice. 4T1 cells were seeded subcutaneously into BALB/c mice to build 4T1 tumor models. The commercial Gadavist and our ES-MION9 were i.v. injected into the 4T1 tumor-bearing mice for MR imaging (Fig. 5A, B). It can be seen from the MR images that after the administration of Gadavist or ES-MION9, the tumor becomes brighter than that of control (pre-injection), and reaches the brightest at 30 min or 3.0 h post-injection, respectively. MR images of different slices were obtained at each time point, and the brightest one of different slices at each time point was selected to characterize the MR imaging capabilities. Because the contrast difference between tumor and normal tissue is usually hard to be identified by the naked eyes, the signal changes in tumors at various time points after the administration of contrast agents are quantified using ΔSNR as shown in Fig. 5C, D, which is calculated according to the Eq. (6):
$$\Delta {\text{SNR}} = \frac{{({\text{SNR}}_{{{\text{post}}}} - {\text{SNR}}_{{{\text{pre}}}} )}}{{{\text{SNR}}_{{{\text{pre}}}} }} \times 100\%$$
(6)
The ΔSNR value is up to 93.4% at 3.0 h after administration of ES-MION9 (Fig. 5D), which is significantly larger than that of the tumor at 30 min post-injection of Gadavist (57.2%, Fig. 5C). The above results demonstrate that our ES-MION9 can be utilized as a stronger MRI CAs compared with the clinically used Gd chelates.
Pharmacokinetics, biodistribution and biosafety evaluation in vivo
To verify that our ES-MION9 is more biocompatible and safer than Gadavist, the pharmacokinetics, biodistribution and biosafety were evaluated in vivo. Figure 5E shows that the blood half-life of ES-MION9 is about 2.3 h due to the small nanoparticle size (3.7 nm). The best time window for MRI in clinic is close to the half-life (10–15 min) of commercial Gd chelates, which is a little bit tight for MRI after administration of the Gd chelates [40]. The slightly longer half-life of our ES-MION9 overcomes the limited MRI time window problem of commercial Gd chelates.
To evaluate the biodistribution of ES-MION9 in vivo, the Fe contents in the heart, liver, spleen, lung, kidney and tumor of mice were measured at 0 h pre-injection and 12.0 h post-injection of ES-MION9, and the differences are shown in Fig. 5F. It is found the ES-MION9 accumulation inside tumors is very high compared with other normal tissues because of the enhanced permeability and retention (EPR) effect, which is the key reason for the highly enhanced MRI signal of tumors after ES-MION9 injection.
Additional file 1: Fig. S17 shows the representative optical microscopic pictures of the H&E-stained main organs from the normal mice without tumors (control), or that with i.v. injection of ES-MION9 (CFe = 5.0 mg/kg). Compared with controls, ES-MION9-treated mice showed no obvious pathological abnormalities in major organs (heart, liver, spleen, lung, and kidney), indicating that our ES-MION9 does not lead to systemic toxicity.
