FMI of the ATP-responsive property of SPIOs@A-T NPs
A Cy3-labeled double-stranded DNA (dsDNA-Cy3) probe was first designed and incubated with a series concentration of ATP (0, 0.5, 1, 2, 3, 4, 6, 8 and 10 mM) to verify the ATP-responsive ability. Fluorescence intensity was nearly undetectable in the absence of ATP, but fluorescent signals gradually increased with increasing ATP concentrations (Fig. S1a, S1b). Control groups with CTP, GTP, or UTP exhibited weak responses to the DNA probe even with increasing ATP concentrations. The results suggest that the dsDNA-Cy3 probe can specifically respond to ATP. The fluorescence signal changes showed a linear correlation with the concentration of dsDNA-Cy3 probe (from 0.1 to 2.0 μM) in response to 5 mM ATP (R2=0.997, Fig. S1c).
Thereafter, the ATP-responsive aptamer (dsDNA-Cy5.5) and targeting peptide were conjugated onto the SPIOs (SPIOs@A-T NPs). The transmission electron microscopy (TEM) image showed that SPIOs@A-T NPs were mono-dispersed and homogeneous (Fig. 1a), without any obvious change in the shape compared to unmodified SPIOs (Fig. S2a). The ultraviolet-visible (UV-vis) absorption spectra showed that SPIOs@A-T had an absorption peak at 260 nm (Fig. 1b) and a positive potential of 4.24 ± 0.83 mV. After modification with CREKA and dsDNA, the zeta potential changed to −9.36 ± 2.04 mV (Fig. 1c). The hydrodynamic sizes of SPIOs and SPIOs@A-T were 51.63 ± 16.30 and 58.58 ± 19.34 nm, respectively (Fig. S2b). The abovementioned results verified the successful synthesis of SPIOs@A-T NPs. Both targeted, ATP-nonresponsive nanoparticles (SPIOs@nA-T NPs) and non-targeted, ATP-responsive nanoparticles (SPIOs@A-nT NPs) were synthesised as control groups.
To verify the ATP-responsive fluorescence imaging property of SPIOs@A-T NPs, the same experiments were performed to test whether SPIOs@A-T can respond to ATP, with SPIOs@nA-T in the control group. As shown in Fig. 1d and 1e, SPIOs@A-T manifested a significant increase in fluorescence signal intensity with increasing ATP concentrations, whereas they showed no change in the response with the addition of CTP, GTP or UTP. In contrast, SPIOs@nA-T showed a weak fluorescence response to ATP as well as to CTP, GTP, and UTP under the same experimental condition (Fig. S3a, S3b). SPIOs@A-T showed a significantly higher fluorescence intensity than SPIOs@nA-T at the same ATP concentration (****p< 0.0001 in Fig. 1f). Furthermore, fluorescence intensity maintained a good linear relationship with the SPIOs@A-T concentration at 5 mM ATP (R2=0.990, Fig. 1g). Thus, our results revealed that SPIOs@A-T possesses excellent ATP-responsive ability.
MPI imaging property of SPIOs@A-T NPs
SPIOs@A-nT and Vivotrax (Magnetic Insight Inc., USA) were used as two controls to examine the MPI-based imaging property of SPIOs@A-T. We tested MPI signals of SPIOs@A-T, SPIOs@A-nT, and Vivotrax within an Fe concentration range of 0.0625–1.0000 mM. In Fig. 1h, there is no obvious difference of MPI signal between SPIOs@A-T and SPIOs@A-nT, and the quantitative value of MPI signals were both linearly related to the Fe concentration (Fig. 1i). Moreover, the MPI signal of SPIOs@A-T was relatively higher than that of commercial Vivotrax at the same Fe concentration (Fig. 1h, 1i), suggesting that SPIOs@A-T possess good MPI performance. We examined the MRI imaging property of SPIOs@A-T, which may provide in vivo anatomical information about lymph nodes. With increasing Fe concentrations, the T2-weighted MRI images gradually darkened (Fig. S4a). Furthermore, the r2 relaxation time of SPIOs@A-T was higher than that of Vivotrax in the same Fe concentration (Fig. S4b). Thus, we successfully developed SPIOs@A-T and proved that SPIOs@A-T have good ATP-responsive fluorescence and MPI/MRI imaging characteristics.
Characterization of cellular targeting and cytotoxicity of SPIOs@A-T NPs
The specific expression of fibronectin in malignant lymph node (MLN) tissues was examined using Western blotting analysis of 4T1 breast tumour-bearing mice. Normal lymph node (NLN) from healthy mice were used as the control. MLN showed a 1.8-fold higher expression of fibronectin compared to NLN (Fig. 2a, 2b, *p < 0.05), suggesting that fibronectin is highly expressed in metastatic tumour cells within LN.
To further evaluate the cellular uptake of targeted SPIOs@A-T, 4T1 cells were incubated with SPIOs@A-T for confocal microscopic imaging. The activation of Cy5.5 red fluorescence represents an ATP-responsive reaction of SPIOs@A-T in the presence of 4T1 tumour cells, whereas a blue colour represents the cellular nucleus (DAPI staining). An obvious red fluorescence signal was observed in the SPIOs@A-T group, and the merged confocal images indicated good SPIOs@A-T uptake by 4T1 cells (Fig. 2c). Non-targeted SPIOs@A-nT and ATP-nonresponsive nanoparticles (SPIOs@nA-T) were used as controls. The red fluorescence signals were weak for both SPIOs@nA-T and SPIOs@A-nT due to their non-ATP-responsive and non-targeting properties, respectively (Fig. 2c).
The cytotoxicity of the nanoparticles on 4T1 cells was examined by co-culturing the cells with serial concentrations of SPIOs@A-T, SPIOs@nA-T, and SPIOs@A-nT for 24 h. The data demonstrated that cell viability was 90–100% for these three groups, indicating that nanoparticles have good biosafety (Fig. S5).
In vivo FMI/MPI dual-modality imaging of orthotopic breast tumour
First, to evaluate the specific and targeted imaging of SPIOs@A-T in vivo, an orthotopic 4T1 breast cancer murine model was created and imaged after intratumoural injection of SPIO@A-T, SPIOs@A-nT, and SPIOs@nA-T (n = 3 mice in each group), respectively. FMI image acquisition was carried out sequentially at time points both before and after the injection of nanoparticles. The results showed that the SPIOs@A-T group showed a higher fluorescence intensity than the non-targeted SPIOs@A-nT for a 24-h post-injection observation period, indicating that, in addition to the EPR effect in vivo, the targeting CREKA peptide could effectively facilitate the targeted binding of nanoparticles to the tumour site (Fig. 3a, 3b). The fluorescence intensity of the SPIOs@A-T group at tumour sites was significantly stronger than that of the SPIOs@nA-T group at all time points, which suggested that these novel nanoparticles could activate fluorescence signalling in the presence of ATP in breast tumours in vivo. Moreover, after the in vivo observation, the tumours were dissected for ex vivo FMI observation (Fig. 3a), and the results suggested that mice injected with SPIOs@A-T exhibited the highest fluorescence signal intensity among the three groups (Fig. 3c).
Furthermore, the MPI was further undertaken on small breast tumours in order to test the feasibility of sensitive imaging of small breast tumours (diameter < 4 mm) (37). As shown in Fig. 3d and 3e, the MPI signal was detected at the tumour site 4 h after the injection of SPIOs@A-T, SPIOs@A-nT, and Vivotrax, and the signal intensity peaked at 8 h after the injection. The MPI signal of SPIOs@A-T was notably higher than those of SPIOs@A-nT and Vivotrax at all time points (n = 3 mice). The ex vivo tumour images from MPI revealed that mice treated with SPIOs@A-T showed significantly higher signal intensities than mice in the other groups (Fig. 3d, 3f).
In vivo FMI/MPI dual-modality imaging of MLNs of breast cancer
Based on the results of the above-described experiment, we conducted FMI of MLNs in the breast tumour model in four groups that received an injection of SPIOs@A-T, SPIOs@nA-T, and SPIOs@A-nT into the left hind paws of breast-tumour-bearing mice as well as the NLN mice which were injected with SPIOs@A-T. After the intradermal injection of SPIOs@A-T in MLN mice, the fluorescence intensity augmented notably and peaked at 12 h, and the signal was detectable for 24 h (Fig. 4a). In contrast, only a very weak fluorescence signal was detected in the remaining three groups, suggesting that SPIOs@A-T augment the targeted and specific metastatic breast tumour cell detection capability. Due to the lack of ATP-responsive fluorescence in the TME, mice injected with SPIOs@nA-T presented almost no fluorescence signal at every time point. Similarly, due to the lack of the CREKA targeting peptide, mice injected with SPIOs@A-nT showed a relatively lower signal intensity than those treated with targeted SPIOs@A-T. Quantitative analysis showed that the average SBR of the SPIOs@A-T group was 222.02 ± 9.33, which was 14.3- and 1.96-fold higher than those of the SPIOs@nA-T (15.51 ± 1.22) and SPIOs@A-nT (113.01 ± 13.84) groups, respectively, at 12 h post-injection. To verify the specific detection of SPIOs@A-T for MLN, the mice with NLN were utilized as controls. The NLN group did not manifest an observable fluorescence signal, further demonstrating that SPIOs@A-T could specifically detect malignant LN (Fig. 4a). The average SBR of SPIOs@A-T (222.02 ± 9.33) was 4.99-fold higher than that of the group with NLN mice (44.47 ± 3.53) 12 h after FMI in vivo (Fig. 4b). Moreover, 24 h after injection of different NPs, the LN were dissected out for FMI ex vivo. As expected, MLN in the SPIOs@A-T group demonstrated the most obvious fluorescence signal compared to the other two groups. In the SPIOs@A-T group, the corresponding quantitative fluorescence intensity of LNs ex vivo was 4.5 ± 0.2 ×108, which was 16.7- and 1.93-fold higher than that of the SPIOs@nA-T (0.27 ± 0.06 ×108) and SPIOs@A-nT (2.33 ± 1.30 ×108) groups (Fig. 4c). The sizes of MLN and NLN are shown in Fig. S6a. Histological examination of LN further revealed findings that were consistent with in vivo and ex vivo FMI observation (Fig. S6a; tumour cells in LN are indicated by black arrows).
In order to evaluate the sensitivity for the detection of MLN, we performed an MPI with SPIOs@A-T, and used the non-targeted SPIOs@A-nT and Vivotrax as controls. As shown in Fig. 4d, at 4, 12, and 24 h after the injection, the MPI signal of SPIOs@A-T was localised to the popliteal LN, gradually peaked and remained stable, respectively. Quantitative assessment at 12 h of the non-targeted SPIOs@A-nT and commercial Vivotrax showed a value of 577.68 ± 19.99 for SPIOs@A-T, which was 1.52- and 1.61-fold stronger than those of SPIOs@A-nT (366.15 ± 9.25) and Vivotrax (359.13 ± 14.03), respectively (Fig. 4e). Moreover, the NLN group was used as the control group. MPI signal values of the SPIOs@A-T group was 1.87-fold higher at 12 h than that of the NLN group due to the specific targeting effect of CREKA (Fig. 4e). After 24 h in vivo observation, the lymph nodes were dissected out for further ex vivo MPI. The results were consistent with that of the in vivo observation (Fig. 4d, 4f). Moreover, T2-weighted MRI provided anatomical structural information of LN, which exhibited a darkening signal at the left popliteal LN site 2 h post-injection with SPIOs@A-T, and the location was nearly consistent with the FMI and MPI signal areas (Fig. S7).
To verify the in vivo observation, Prussian blue staining and histology were performed for LN to test the biodistribution of SPIOs@A-T NPs and showed more Fe-positive staining in the MLN of the SPIOs@A-T NPs group compared to the other groups (Fig. S6b), which was consistent with the findings of the in vivo observation.
Interestingly, MPI is well known for the superior performance with high sensitivity and the absence of an imaging-depth limitation and, therefore, MPI can be utilised for quantitative analysis. To validate this aspect in our study model, we performed the FMI and MPI on the same mice and the dual modality imaging scan on the same MLN both before and after skin incisions. Mice were injected with SPIOs@A-T NPs and scanned at 12 h post-injection. Compared to before exposure, the fluorescence signal was significantly higher after the skin exposure, although there was no remarkable difference for MPI signal intensities between the before and after skin incision evaluations (Fig. 5). This suggests that MPI is a promising method for the detection of metastatic breast cancer cells in LN or even in deep metastases to other organs.
To validate the specific and targeted binding of SPIOs@A-T to the metastatic breast cancer cells in LN, the MLN tissues were cryosectioned and fluorescence-stained for fibronectin. The confocal microscopy images showed that SPIOs@A-T displayed the strongest fluorescence intensity, compared to SPIOs@nA-T and SPIOs@A-nT. Moreover, SPIOs@A-T specifically bound to metastatic cancer cells, showing merged images (denoted by white arrows). In general, we concluded that SPIOs@A-T NPs can specifically discern and bind to the metastases of breast cancer in LN (Fig. S8).
In vivo biosafety assessment of SPIOs@A-T NPs
In order to estimate the biosafety of NPs in vivo, healthy mice were divided into four groups treated with injection of SPIOs@A-T, SPIOs@nA-T, SPIOs@A-nT NPs, and saline, respectively. There were no observable pathological changes for major organs, including the heart, liver, spleen, lungs, and kidneys, in all groups (Fig. S9). In addition, the serum biochemical indicators, including HDL-C as heart function biomarkers, BUN and CREA as kidney function biomarkers, and ALT and AST as liver function biomarkers were measured (Fig. S10). The data indicated no significant differences between nanoparticle-treated groups and the normal saline group. Thus, SPIOs@A-T, SPIOs@nA-T, and SPIOs@A-nT NPs showed good in vivo biocompatibility and biosafety.