A Fluorinated Ionic Liquid-Based Activatable ¹⁹F MRI Platform Detects Biological Targets

SUMMARY

¹⁹F magnetic resonance imaging (¹⁹F MRI) is a promising technique for in vivo molecular imaging and clinical diagnosis, benefiting from its negligible background and unlimited tissue penetration depth. However, the development of ¹⁹F probes with good water solubility and versatile functions for bioresponsive and practical applications remains a challenge. Here, we report fluorinated ion liquids (ILs) as a new type of fluorine agents and build a fluorinated ionic liquid-based activatable ¹⁹F MRI platform (FILAMP), which relies on the phase transition of ILs. Upon exposure to environmental stimulation, coating polymer dissolves or degrades to release the fluorinated ILs payload, which rapidly enhances ¹⁹F signal. This “turn-on” response is verified by the successful detection of biological targets (for example, dysregulated pH and MMP overexpression) at the cellular level and in mice, demonstrating the potential of FILAMP as a robust activatable ¹⁹F probe for diagnosis and monitoring of biological and pathological processes.

Graphical Abstract

Fluorinated ion liquids (ILs) act as a new type of fluorine agents to build a fluorinated ionic liquid-based activatable ¹⁹F MRI platform (FILAMP). Upon biological stimulation, the coating polymer dissolves or degrades to release the payload, which rapidly enhances the ¹⁹F signals. This “turn-on” response is verified by successful detection of biological targets in vitro and in vivo. In summary, FILAMP can serve as a type of activatable ¹⁹F probes for diagnosis and monitoring of biological and pathological processes.

INTRODUCTION

Magnetic resonance imaging (MRI) is a noninvasive and tomographic technique for imaging living systems and holds great promise in disease diagnosis.^{1 1}H MRI plays an important role in both clinic and research, because tissues contain 65% to 80% of water, which provides sufficient signals.² However, the high content of water also causes a strong background, which might lead to low contrast and imaging artifacts.^3-7 MRI based on alternative nuclei is a fundamental solution to the background issue. Among these nuclei, ¹⁹F is a promising candidate because of its high magnetic resonance sensitivity (83% relative to ¹H) and 100% natural abundance.⁸ More importantly, endogenous ¹⁹F nuclei mainly exist in solid form within teeth and bones. The immobilized fluorine nuclei have very short spin-spin relaxation time (T₂) and show negligible MR signals.^9-11 Therefore, ¹⁹F MRI by introducing exogenous fluorine nuclei is considered to be a zero-background imaging means with high tissue penetration depth.

Recently, because of the advances of dual ¹H/¹⁹F radiofrequency (RF) coil and ultrafast pulse sequence, the accuracy and sensitivity of ¹⁹F MRI are greatly improved. As for imaging agents, there are two popular kinds of ¹⁹F agents under active development at present: perfluorocarbons (PFCs) and trifluoromethyl compounds (TFMCs). In PFCs, fluorine atoms substitute all the protons in organic compounds, resulting in high ¹⁹F density.^12,13 Kikuchi and co-workers constructed a nanoparticle with a core micelle of liquid perfluoro-15-crown-5-ether (PFCE) and a robust silica shell for in vivo tumor imaging.^14-17 In TFMCs, biomolecules are modified with trifluoromethyl groups or their derivatives, on the premise of no change in functions.¹⁸ TFMCs are more flexible in molecular design for biomedical applications because of diverse modifications. For example, Hamachi and co-workers introduced trifluoromethyl groups onto a self-assembled supramolecule, leading to “quenched” ¹⁹F signals. In the presence of a target protein, the supramolecule disassembled along with the recovery of ¹⁹F signals.¹⁹

However, both PFCs and TFMCs have some limitations that prevent them from broad applications. PFCs, which are hydrophobic and chemically inert, cannot be used as probes in their original form. Encapsulation in emulsion or polymer-based formula are two major strategies to expand their applications.^20-23 For TFMCs, molecular fluorine density is usually insufficient because excessive trifluoromethylation would decrease the hydrophilicity of biomolecules and attenuate their functions.^24,25 Therefore, new ¹⁹F imaging agents are in great demand to circumvent the deficiency of present fluorine sources. Appropriate ¹⁹F agents need to satisfy two criteria: first, the agent should be of high molecular fluorine density at the same chemical shift, thus every fluorine nucleus is detectable and the chemical-shift artifact is avoided; second, the agent should be readily soluble in water for biomedical applications.²⁶ Fluorinated anions, such as tetrafluoroborate $({\text{BF}}_4^{-})$ and trifluoromethanesulfonate (OTf⁻), are potential candidates because of their excellent water solubility and high density of multiple chemically equivalent ¹⁹F nuclei, which would offer strong and sharp peak in nuclear magnetic resonance (NMR) spectra. Ionic liquids (ILs) is one of the most explored compounds that contain these fluorinated anions. The tunable nature facilitates the development of ILs for applications in life sciences and medicine.^27,28 Many ILs with excellent solubility, good biocompatibility, and easy excretion have found applications in biomedical fields.^29,30 Therefore, ILs containing fluorinated anions can act as a new type of probe sources for ¹⁹F MRI when suitable nanocarriers are used to deliver and release these ILs in a controllable fashion.^31,32

Here, we report a fluorinated ionic liquid-based activatable ¹⁹F MRI platform (FILAMP) for stimuli-responsive imaging. We utilized the phase transition of fluorinated ILs with suitable melting point (m.p.) to construct nanoprobes in their molten state by stably confining them in hollow mesoporous silica (HMS) spheres and then sealing the pores with stimuli-responsive copolymers. The entrapment of the ILs in HMS spheres restricts their molecular mobility and accelerates transverse relaxation,^33-35 turning ¹⁹F signals “off” (Figure 1A). Their ¹⁹F signals would be significantly enhanced once the trapped ILs are released after the probes respond to a specific stimulus (e.g., tumor-related factors including low pH and matrix metalloproteinase (MMP) secretion),^36,37 thus turning on the ¹⁹F signal. As a versatile platform, FILAMP is modular with optional ILs and various biological triggers. We further demonstrated that the release kinetics is also influenced by temperature, coating thickness, and IL m.p., which helps in achieving a suitable time window for signal enhancement. These preferred characteristics enable FILAMP to respond to specific stimuli at the cellular level and in vivo and serve as versatile and smart probes for detection and imaging of biological targets by ¹⁹F MRI.

Figure 1. Schematic Illustration of FILAMP for Stimuli-responsive ¹⁹F MRI.

(A) ¹⁹F signal is “On” for free fluorinated ionic liquid (IL). When sealed by silica and polymer coating, ¹⁹F signal turns “Off”. The coating polymer dissolves or degrades by the stimuli, leading to the release of the loaded fluorinated IL from HMS spheres and the ”turn-on” of ¹⁹F signal for both NMR and MRI.

(B) A cartoon illustrating the construction strategy of FILAMP. After dispersed in molten IL, HMS-Acryl nanoparticles can load IL into the cavities. The IL inside the cavities is then trapped by a polymer layer formed in situ. After the mixture is frozen, cold water is added to remove the excess IL and the probes are collected by ultrafiltration. This strategy exploits the phase transition of ILs and can achieve high loading efficiency and stability.

RESULTS AND DISCUSSION

Design and Construction of FILAMP

The design of FILAMP was motivated by three goals: (1) to improve loading efficiency, which is essential for signal intensity, (2) to shorten the T₂ of loaded fluorine nuclei for keeping signals “off” before activation, and (3) to achieve stimuli-responsive ¹⁹F MRI and avoid uncontrolled burst release. To accomplish the above goals, we developed a strategy to load ILs into HMS spheres mainly relying on the phase transition of ILs (Figure 1B). We chose BMMIBF₄ (BMMI, 1-butyl-2,3-dimethyl-imidazolium), EMIOTf (EMI, 1-ethyl-3-methyl-imidazolium), and EPyBF₄ (EPy, 1-ethyl-pyridinium) in this study because of their typical m.p. at 37°C, −9°C, and 53°C, respectively (Figure 2A). HMS spheres as the nanocarrier (Figure S1) was synthesized via a selective etching method.³⁸ After further modification with acryl groups, ILs were loaded into HMS-Acryl followed by in situ surface coating of stimuli-responsive diblock copolymers.

Figure 2. Characterizations of FILAMP.

(A) Chemical structures of three typical fluorinated ILs: BMMIBF₄, EMIOTf, and EPyBF₄.

(B and C) TEM images of (B) HMS spheres and (C) BMMIBF₄-pH(thin).

(D) EDX mapping and EDS of BMMIBF₄-pH(thin) probes.

(E) TEM images of BMMIBF₄-pH(medium), BMMIBF₄-pH(thick), BMMIBF₄-MMP, EMIOTf-pH, and EPyBF₄-pH probes. Scale bars represent 50 nm.

The construction process was straightforward (Figure 1B). The two steps, loading and sealing, were carried out in one pot. For example, the preparation of BMMIBF₄-based probes started with the dispersion of as-prepared HMS-Acryl in molten BMMIBF₄. Repeated freeze-thaw cycles were conducted in vacuo to ascertain the cavities were fully filled with ILs. Subsequently, monomers and azodiisobutyronitrile (AIBN) as an initiator were introduced and mixed. Polymerization occurred on the HMS surface for coating when the mixture was heated to 50°C–60°C. When the reaction was completed, the mixture was cooled with ice until it froze. Cold water was then added to dissolve the solid mixture and redisperse the probes. Finally, the probes were separated and purified by ultrafiltration to remove the excess ILs. This construction strategy smartly exploited the phase-transition of ILs, including molten ILs as the solvent for polymerization and solidified ILs as the cargo. To investigate the functions and stability of these nanoprobes, we prepared several probes with different polymer thicknesses by adjusting the amount of pH-responsive monomer 1,3-di-4-piperidylpropane-di(ethylene glycol) diacrylate (DPP-DEGDA, Figure S2). DPP is a pH-responsive component that could be protonated on its tertiary amines and become soluble at acidic pH.³⁹ Three probes with different polymer thicknesses were obtained, termed as BMMIBF₄-pH(thin), BMMIBF₄-pH(medium) and BMMIBF₄-pH(thick). This strategy was applied to two other ILs to construct two pH-responsive probes, EMIOTf-pH and EPyBF₄-pH. We also constructed a MMP-responsive ¹⁹F probe BMMIBF₄-MMP by using Peptide-PEGDA monomers (Figure S3). The peptide contains a PLGLAG fragment as an MMP (e.g., MMP-2) substrate, and two cysteine units at each end for conjugation with poly (ethylene glycol) diacrylate (PEGDA) via Michael addition. A control sample BMMIBF₄-MMP(d) was synthesized similarly except that an uncleavable plglag (lowercase letters for d-amino acids) fragment was used. Detailed protocols are included in the Supplemental Information.

Characterization of FILAMP

BMMIBF₄ based probes were characterized by transmission electronic microscopy (TEM). The as-prepared BMMIBF₄-pH(thin) showed an encapsulated silica shell by a rough polymer layer, which was in sharp contrast to uncoated HMS spheres (Figures 2B and 2C). Both TEM-associated energy dispersive X-ray element mapping (EDX mapping) and X-ray spectroscopy (EDS) revealed distinct fluorine signals (Figure 2D), suggesting the successful loading of the ILs. In EDX mapping, fluorine signals preferred to accumulate in the silica shell rather than in the core of HMS spheres, which is ascribed to the melting and sputtering of BMMIBF₄ upon focused X-ray irradiation. The synthesis process was characterized by infrared spectroscopy (IR) and X-ray photoelectron spectroscopy (XPS) (Figure S4). With continued addition of DPP-DEGDA monomers, BMMIBF₄-pH(medium), and BMMIBF₄-pH(thick) were obtained (Figure 2E) with their morphologies much different from BMMIBF₄-pH(thin). Interestingly, the polymer layer seemed to cover the shell surface first, then polymerization continued inward and formed a vacuole (BMMIBF₄-pH(medium)). With increasing monomers, a crisscrossing network structure embracing the vacuole was formed (BMMIBF₄-pH(thick)). The MMP-sensitive probe and its negative control (i.e., BMMIBF₄-MMP and BMMIBF₄-MMP(d)) exhibited only a loose polymer layer outside the silica shell (Figure 2E). In addition, we could not obtain a medium or thick layer even though more monomers were added. This was probably because the longer chain of Peptide-PEGDA (average molecular mass at 2,672.4) prevented it from entering the cavity through the mesopores. EMIOTf- and EPyBF₄-based probes were constructed in a manner similar to BMMIBF₄-pH(thin) and exhibited comparable morphologies with thin polymer layers (Figure 2E). EDX mapping also validated the successful loading of ILs in these probes (Figure S5).

Relaxation Times and Loading Efficiencies

We measured the ¹⁹F relaxation times of the probes and the corresponding ILs on an NMR spectrometer (564 MHz for ¹⁹F) with inversion recovery (IR) and Carr-Purcell-Meiboom-Gill (CPMG) sequences at 20°C (Figure S6; Table 1). Free fluorinated ILs had extremely long T₁ and T₂ relaxation times within a range of 3–5 s. However, BMMIBF₄-pH(thin) probes showed a similar T₁ but a much shorter T₂ (0.09 s) with T₂/T₁ at 0.020. This shortened T₂ could be attributed to the spin-spin relaxation enhancement, as the loaded IL was confined in solid or glass state in HMS spheres with restricted molecular motion. Moreover, BMMIBF₄-pH(medium) and BMMIBF₄-pH(thick) exhibited even shorter T₂ values than BMMIBF₄-pH(thin), which were 0.084 and 0.069 s, respectively. This was due to the further restricted motion by a thicker polymer coating. Similar to pH-sensitive probes, BMMIBF₄-MMP also showed a short T₂ of 0.119 s with T₂/T₁ at 0.025. Interestingly, although EMIOTf-pH and EPyBF₄-pH were fabricated similarly to BMMIBF₄-pH(thin) and also had a thin coating, they exhibited distinct T₂ values of 0.217 and 0.055 s, respectively. This result could be attributed to their different physical states after loading as well as the intrinsic relaxation property of each IL. The loaded EPyBF₄ was more likely to solidify with high viscosity because of its higher m.p. at 53°C, whereas EMIOTf was probably still liquid. We also compared the ¹⁹F relaxation times at different magnetic fields (376 and 564 MHz) and temperatures (20°C and 37°C) and found slight changes in ¹⁹F relaxation times (Table S1).

Table 1.

The ¹⁹F Relaxation Times and Loading Efficiencies of FILAMP

ILs or Probes	Longitudinal Relaxation Time (T1) [s]	Transverse Relaxation Time (T2) [s]	Loading Efficiency (wt %)	Fluorine Content (wt %)
Free BMMIBF4	4.908	4.446	n/a	31.7
BMMIBF4-pH(thin)	4.467	0.090	57.4 ± 5.2	18.2 ± 1.6
BMMIBF4-pH(medium)	4.194	0.084	36.2 ± 6.8	11.5 ± 2.2
BMMIBF4-pH(thick)	3.916	0.069	32.6 ± 5.5	10.3 ± 1.7
BMMIBF4-MMP	4.719	0.119	48.5 ± 4.4	15.4 ± 1.4
Free EMIOTf	3.665	3.650	n/a	21.9
EMIOTf-pH	3.144	0.217	53.1 ± 4.7	11.6 ± 1.0
Free EPyBF4	4.364	3.817	n/a	39.0
EPyBF4-pH	3.277	0.055	46.6 ± 3.3	18.2 ± 1.3

Refer to Figure S6 for relaxation determination and curve fitting on an NMR spectrometer (564 MHz for ¹⁹F). Compared with free ILs, the ¹⁹F probes show similar T₁ but much shorter T₂, which is ascribed to the spin-spin relaxation of loaded ILs. With the construction strategy that exploits molten ILs, these probes feature high loading efficiencies and fluorine contents. Data are shown as mean ± SD (n = 3).

To quantify the loading efficiency, we measured the amount of the loaded fluorine nuclei by quantitative NMR (qNMR) with CF₃COONa as an internal reference (AQARI method). Briefly, standard curves of free ¹⁹F agents (BMMIBF₄ and EMIOTf) were established by the integrations relative to the reference (Figure S7). The loaded IL amount was measured after the silica shell was destroyed by HF etching, and the IL was completely released. The loading efficiency was determined to be as high as 57.4% (by IL mass, 18.2% by fluorine mass) for BMMIBF₄-pH(thin) (Table 1), and decreased with the increasing coating thickness, 36.2% for BMMIBF₄-pH(medium) and 32.7% for BMMIBF₄-pH(thick), but remained at relatively high levels. We also obtained loading efficiencies of 48.5% for BMMIBF₄-MMP, 53.1% for EMIOTf-pH, and 46.6% for EPyBF₄-pH. No appreciable release was found in these probes at 4°C over 6 months. These results illustrated that the loading strategy, in which the ILs acted as both cargo and solvent at the same time, conferred high loading efficiency and excellent stability for these probes.

Stimuli-Activated IL Release

At a given pulse sequence and operating parameters, the characteristics of ¹⁹F NMR peaks are highly dependent on relaxation times. For our probes, the significant T₂ shortening with low T₂/T₁ ratio would lead to a dramatically accelerated signal decay and severe line broadening, as demonstrated by several previously reported probes exploiting the same mechanism (enhanced spin-spin relaxation)^19,25,40 or other effects (such as paramagnetic relaxation enhancement, PRE).^11,15,41,42 We acquired the NMR spectra of BMMIBF₄-pH(thin) and observed that the ¹⁹F NMR peak was short with severe line broadening and became sharper and taller overtime during incubation at pH 5.0, and Δν₀ (natural Lorentzian line-width at half-height) decreased from 63.4 to 5.0 Hz after 6 h incubation (Figure S8). To quantitatively evaluate the IL release kinetics, we used qNMR to record the intensity of discharged ¹⁹F agents (CF₃COONa as an internal reference) with 200 ms dead time to filter off the broad signal of the unreleased ¹⁹F agents. The as-synthesized probes with fluorine concentration ([F]) at 200 mM were incubated (37°C) under various physiologically relevant pH conditions, including tumor microenvironment (pH 6.4~6.8), lysosomes (pH 5~6), and blood (pH 7.4). Using the same operating parameters, the spectra were acquired during the incubation and the ¹⁹F signals were evaluated by relative integrations for release kinetics analysis. As shown in Figure 3A, the ¹⁹F peaks remained short at pH 7.4, and the unchanged integration suggested that less than 10% of IL was released after 6 h. However, more IL was rapidly released from nanoparticles in an acidic environment, discharging 35.1% at pH 6.8, 54.0% at pH 6.4, 69.9% at pH 5.8, and 81.4% at pH 5.0 after 6 h (Figure 3B). Besides, the release kinetics was temperature-dependent (Figures 3C and 3D), probably due to the enhanced dissolution at higher temperatures. We also studied the release kinetics of BMMIBF₄-pH(medium) and BMMIBF₄-pH(thick) (Figures S9 and S10) and found that the release was negatively correlated with the polymer thickness, consistent with the assumption that thicker coating takes longer to rupture.

Figure 3. ¹⁹F NMR Signal Intensity Analysis and Release Kinetics of FILAMP.

(A and B) ¹⁹F qNMR spectra (A) and ¹⁹F agent release kinetics (B) of BMMIBF₄-pH(thin) at pHs 7.4, 6.8, 6.4, 5.8, and 5.0 (37°C). Signal enhancement at −150.0 ppm associates with BMMIBF₄ release because of the dissolution of the coating layer under acidic conditions.

(C and D) ¹⁹F qNMR spectra (C) and release kinetics analysis (D) of BMMIBF₄-pH(thin) at different temperatures (pH 6.4).

(E and F) ¹⁹F qNMR spectra (E) and ¹⁹F agent release kinetics (F) of BMMIBF₄-MMP and BMMIBF₄-MMP(d) when incubated with 50 nM MMP-2.

(G and H) ¹⁹F qNMR spectra (G) and ¹⁹F agent (−78.7 ppm) release kinetics (H) of EMIOTf-pH at different pHs.

(I and J) ¹⁹F qNMR spectra (I) and release kinetics analysis (J) of EMIOTf-pH at different temperatures (pH 6.4).

(K and L) ¹⁹F qNMR spectra (K) and ¹⁹F agent (−150.0 ppm) release kinetics (L) of EPyBF₄-pH at different pHs.

CF₃COONa at −75.4 ppm was used as an internal reference. Data are shown as mean ± SD (n = 3).

Apart from pH-activation, FILAMP was also designed to serve as an MMP-sensitive probe. BMMIBF₄-MMP probes were constructed with the as-synthesized Peptide-PEGDA monomers, which could serve as a substrate of MMP.^43,44 We investigated the MMP-triggered release of BMMIBF₄-MMP probes ([F] = 200 mM) after incubation with 50 nM active MMP-2, a main member of MMP family (Figures 3E and 3F). The loaded IL quickly released into the solution within 2 h, and then reached a plateau, indicating a burst release by enzymatic hydrolysis at the early stage. We also conducted a control experiment by using BMMIBF₄-MMP(d) probes, which was composed of the cleavage-resistant coating polymer. As expected, BMMIBF₄-MMP(d) did not show apparent release during incubation. These data indicated the high sensitivity and specificity of FILAMP for the detection of MMP, which could be easily extended to the detection of various biological targets.

To further validate the versatility of FILAMP with different ILs, we conducted the pH-response experiment on EMIOTf-pH and EPyBF₄-pH (Figures 3G-3L and S11). Interestingly, although the probes were all coated with thin polymer layers, EMIOTf-pH showed a faster release than BMMIBF₄-pH(thin), whereas EPyBF₄-pH was the slowest. This phenomenon could be ascribed to the states of ILs, i.e., the melt EMIOTf with low viscosity facilitated fast release, whereas the high m.p. prevented the discharge of EPyBF₄.

Stimuli-Activated ¹⁹F MRI

We next examined ¹⁹F MRI of FILAMP by using a fast low angle shot (FLASH) sequence. In ¹⁹F MRI with an FLASH sequence, ¹⁹F signal intensity depends on T₂* according to

$$S_{\text{avg}}\propto e^{\frac{-T_{\mathrm{E}}}{T_2^{\ast }}}$$

where S_avg represents the average signal intensity, and T₂* stands for the time constant for the relaxation caused by transverse relaxation and external magnetic field inhomogeneity.^45,46

If T_E is longer than T₂* of fluorine nuclei, the ¹⁹F MRI signal becomes weak because of the exponential decay relationship between S_avg and T_E. We evaluated the impact of T_E on the signal intensity of the probes and found that the T₂* of loaded BMMIBF₄ (intact BMMIBF₄-pH(thin) probes) was extremely short and undetectable on our 9.4 T MRI (Bruker), whereas the T₂* of released BMMIBF₄ (BMMIBF₄-pH(thin) probes incubated at pH 5.0 for 6 h) was measured to be 9.90 ms. We then acquired the corresponding ¹⁹F MRI phantoms of BMMIBF₄-pH(thin) after 37°C incubation at pH 7.4, 6.4, and 5.0 with an FLASH sequence (T_R/T_E = 1,600/1.3 ms, flip angle (FA) = 30°, average = 16, acquisition time =13 min). As shown in Figure 4A, the ¹⁹F “hot spot” signal was almost invisible at pH 7.4 because of the extremely short T₂* of loaded BMMIBF₄ (T₂* < T_E) that led to the very weak ¹⁹F signal in MRI. However, the samples at pH 6.4 and 5.0 gradually brightened over time, suggesting that the pH-triggered release of fluorine agent resulted in T₂* recovery and “turn-on” ¹⁹F “hot spot” signals in MRI. These results validated our original design of pH-responsive ¹⁹F MRI in an acidic environment. We traced the morphology change during the incubation at pH 6.4 (Figure 4B). With OsO₄ staining, the TEM images clearly showed the entire activation process that the pH-sensitive layer peeled off, and finally, the smooth silica shell was exposed. As shown in Figure 4C, the sample at pH 5.0 showed higher signal-to-noise ratio (SNR) (up to 188) than that at pH 6.4 (up to 121).

Figure 4. Phantom Imaging of FILAMP by ¹H MRI and ¹⁹F “Hot Spot” MRI.

(A) ¹⁹F MRI of BMMIBF₄-pH(thin) after incubation at pHs 7.4, 6.4, and 5.0 for 0, 0.5, 1, 2, 3, 4, and 6 h.

(B) Morphology changes of BMMIBF₄-pH(thin) at pH 6.4 by TEM after incubation for 0, 1, 3, and 6 h with OsO₄ staining. Scale bar represents 100 nm.

(C) SNR analysis of ¹⁹F phantom imaging of BMMIBF₄-pH(thin) after incubation at pHs 7.4, 6.4, and 5.0. Data are shown as mean ± SD (n = 3).

(D and E) ¹⁹F phantom imaging (D) and SNR analysis (E) of BMMIBF₄-MMP and BMMIBF₄-MMP(d) after incubation with 50 nM MMP-2 at 37°C. Data are shown as mean ± S.D. (n = 3).

MMP-activated ¹⁹F MRI was carried out with BMMIBF₄-MMP and BMMIBF₄-MMP(d) (Figures 4D and 4E). Upon incubation with MMP-2, ¹⁹F phantom imaging indicated that the BMMIBF₄-MMP signal was gradually turned on with the highest SNR of 152 at 3 h. However, no visible ¹⁹F “hot spot” signal could be observed with BMMIBF₄-MMP(d) because of the presence of hydrolysis-resistant d-peptide in the coating polymers.

Cellular and In Vivo ¹⁹F MRI Study

The success of in vitro activatable ¹⁹F MRI prompted us to explore the application of FILAMP at the cellular level and in vivo. Prior to ¹⁹F MRI experiments, we evaluated the biocompatibility and toxicity of FILAMP and BMMIBF₄ (Figures S12-S14; Table S2). Cytotoxicity of BMMIBF₄-pH(thin) and BMMIBF₄-MMP probes was examined on HT-1080 cells by lactate dehydrogenase (LDH) and 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) assays. BMMIBF₄-pH(thin) and BMMIBF₄-MMP probes showed negligible cytotoxicity in both LDH and MTT assays (Figure S12), manifesting the good biocompatibility of the silica shell and PEG coating. Tissue toxicity was assessed by the histological TdT-mediated dUTP nick-end labeling (TUNEL) assay, showing no apoptosis caused by the probes in tumors after intratumoral injection of the probes (Figure S13). To further assess systemic biocompatibility and toxicity in animals, we conducted hematoxylin and eosin (H&E) staining and biochemistry index analysis after intravenous injection of the probes to mice with S180 tumors (Figure S14). All major organs maintained their typical structures and did not display any appreciable microscopic lesions. The major biochemistry indices, including albumin (ALB), alanine aminotransferase (ALT), aspartate transaminase (AST), creatinine (CRE), total protein (TP), cholesterol (CHOL) and urea (UREA), were maintained at levels comparable to those of the control. These results suggested that the probes had minimal side effects and good biocompatibility.

We chose HT-1080 cells for MMP-response experiments because of their high MMP expression level.^47,48 The MMP concentration in the media after 4 h culture was determined to be 76.8 nM by gelatin zymograms (Figure S15). Although cell uptake of the probes was rather low (Figure S16), the extracellular MMP expression was high enough to cleave the peptide substrates on the probes. BMMIBF₄-MMP, BMMIBF₄-MMP(d), and BMMIBF₄-MMP + Batimastat were individually added to the media ([F] = 200 mM). Batimastat is a potent, broad-spectrum inhibitor of MMP.⁴⁹ After incubation for 2 h, the media were collected and subjected to ¹⁹F MR imaging (T_R/T_E = 1,600/1.3 ms, FA = 30°, average = 20, acquisition time = 17 min). Apparent recovery of ¹⁹F “hot spot” signal was observed for the BMMIBF₄-MMP group with the SNR of 106 (Figure 5), whereas the groups of BMMIBF₄-MMP + Batimastat and BMMIBF₄-MMP(d) exhibited weak signal and almost no signal with their SNRs at 29.7 and 5.5, respectively. Besides cellular experiments, BMMIBF₄-MMP and BMMIBF₄-MMP(d) were also intratumorally injected in mice bearing HT-1080 xenograft tumors at a dose of 50 μL ([F] = 200 mM). MR images were acquired at 2 h post-injection. Distinct ¹⁹F MRI signals for mice injected with BMMIBF₄-MMP (SNR = 93.7) suggested the successful activation by MMP (Figure 5). The overexpressed MMP in the tumor microenvironment hydrolyzed the coating polymers and released the ILs. On the contrary, signal enhancement was not obvious for mice injected with BMMIBF₄-MMP(d) (SNR = 4.3). These results demonstrated the success of specific MMP-activated ¹⁹F MRI with the negligible background by using BMMIBF₄-MMP probes both at the cellular level and in vivo.

Figure 5. Cellular and In Vivo ¹⁹F MRI with FILAMP.

(A) MMP-responsive MRI of HT-1080 cells and mice with xenograft tumors. ¹H and ¹⁹F phantom imaging of BMMIBF₄-MMP, BMMIBF₄-MMP(d), and BMMIBF₄-MMP + Batimastat after incubation with the culture media of HT-1080 cells. ¹H MRI and ¹⁹F MRI of mice with HT-1080 xenograft tumors after injection of BMMIBF₄-MMP probes or BMMIBF₄-MMP(d). The pH-responsive ¹⁹F MRI was performed on mice with S180 xenograft tumors after injection of BMMIBF₄-pH(thin).

(B) Comparison of SNRs in regions of interest when performed on tumor cells and mice with xenograft tumors after different treatments. Data are shown as mean ± SD (n = 3/group), *p < 0.05, **p < 0.01, ***p < 0.005, paired Student’s t test.

To elucidate the versatile feature of our FILAMP for in vivo imaging, we further evaluated the pH-activated ¹⁹F MRI with BMMIBF₄-pH(thin) probes in mice bearing subcutaneous S180 tumors. The mildly acidic (pH 6.4~6.8) extracellular microenvironment of tumor tissues could also serve as a biomarker for solid tumors. The mice were intratumorally injected with 50 μL of the probes ([F] = 200 mM) and ¹H/¹⁹F MRI was acquired at 3 h post-injection. Similar to MMP-response experiments, clear ¹⁹F “hot spot” signals with an SNR of 80.7 were observed in the tumor region (Figure 5). However, the ¹⁹F signal started to fall at 6 h with an SNR of 57.3 (Figure S17). This could be probably ascribed to the diffusion of the released ILs in tissue fluid because of the excellent solubility of the ILs. These results indicated the successful pH-triggered release of the loaded ILs in the acidic tumor microenvironment that enabled pH-activated ¹⁹F MRI.

FILAMP has been demonstrated as an effective platform for stimuli-triggered ¹⁹F MRI, both in vitro and in vivo. However, before the successful biomedical applications of FILAMP, there are lots of challenges, including rapid clearance, unpredictable biodistribution, and restricted delivery efficiency to targets, which are inevitable for most nanoparticle-based probes. Fortunately, FILAMP is operative with modular design integrating optional ILs and various stimuli-sensitive polymer coatings, which might provide potential solutions to the existing challenges. Incorporated strategies can also be optimized, including size engineering and surface modification for improved circulation time in the blood pool, and/or conjugation of targeting molecules for selective delivery and significant reduction of nonspecific uptake. Given that FILAMP is still at its infant stage, further improvements are necessary for future applications.

In summary, we developed a novel FILAMP technique, which, to the best of our knowledge, is the first report on using fluorinated ILs as new fluorine sources to develop activatable probes for ¹⁹F MRI. Based on the phase-transition, several ILs, such as BMMIBF₄, EMIOTf, and EPyBF₄, could be used to construct the probes with high loading efficiency and remarkable stability. The high water solubility and multiple chemical equivalent ¹⁹F nuclei of ILs warrant sufficient signal intensity for ¹⁹F NMR and MRI. More importantly, our platform could achieve responsive ¹⁹F signal activation triggered by various stimuli that are pathological hallmarks of tumors, such as dysregulated pH and abnormal MMP overexpression, and have tunable response behaviors via altering the coating thickness of the probes. These probes are applicable for in vivo sensing of the low pH and overexpressed MMP in tumors, and offer a sensitive means for cancer diagnosis and metastasis monitoring. Compared with conventional ¹H MRI, ¹⁹F MRI using FILAMP shows a major advantage of no background for in vivo imaging. Furthermore, fluorinated ILs with various chemical shifts hold great potential as a series of ¹⁹F agents, which would be suitable for many applications in biomedical imaging, such as multicolor imaging and cell tracking. We believe that FILAMP, with its modular design that incorporates optional ILs and biological triggers, would serve as a powerful and versatile tool for non-invasive in vivo imaging of biological and pathological indicators.

EXPERIMENTAL PROCEDURES

Construction of ¹⁹F Probes

BMMIBF₄-pH(thin) was constructed as follows: HMS-Acryl was dispersed in 50 mL acetonitrile, and the suspension was added to 100 mL acetonitrile solution containing 30 g BMMIBF₄. Subsequently, 80 mg DPP-DEGDA and 3 mg AIBN were dissolved in 10 mL DCM solution containing 5 g BMMIBF₄. Acetonitrile and DCM were removed from these two solutions via concentration in vacuo and repeated freeze-thaw cycles between −20°C and 40°C. The IL solution containing HMS-Acryl was gently stirred at 50°C in N₂, and the IL solution containing DPP-DEGDA and AIBN was added slowly over 5 h. The reaction was maintained at this temperature for another 5 h before 5 mg poly (ethylene glycol) methyl ether acrylate (mPEG-acryl, M_n = 1,000) was added. The mixture was allowed to react for another 4 h before it was frozen at 0°C. Subsequently, an appropriate amount of water at 0°C was added. BMMIBF₄-pH(thin) probes were separated from the IL solution by repeated ultrafiltration and finally stored in 1× PBS. Other probes were prepared with a protocol similar to the preparation of BMMIBF₄-pH(thin), please refer to Supplemental Information for details.

Relaxation Time and Quantitative NMR (qNMR)

Relaxation measurements were performed on NMR spectrometers (564 and 376 MHz for ¹⁹F) with 10% D₂O for shimming. The longitudinal relaxation times (T₁) were measured using an inversion recovery (IR) sequence. The transverse relaxation times (T₂) were measured using a Carr-Purcell-Meiboom-Gill (CPMG) sequence. Other parameters consisted of repetition time = 30 s, temperature = 20 or 37°C, sample rotation = 15 Hz, scan sweep = 60 ppm, NS (number of scans) = 1, and receive gain = 10.

T₁ was calculated by curve fitting with the following equation:

$$\mathrm{M}(\tau )=M_0\left(1-\text{exp}\left(-\frac{\tau }{T_1}\right)\right)$$

T₂ was calculated by curve fitting with the following equation:

$$\mathrm{M}(\tau )=M_0\text{exp}\left(-\frac{\tau }{T_2}\right)$$

qNMR was carried out using an AQARI method (accurate quantitative NMR with internal reference substance) a 564 MHz NMR Spectrometer with 10% D₂O for shimming and sodium trifluoroacetate ([F] = 50 mM) as internal standard.⁵⁰ Samples were incubated under certain conditions (pH, temperature and MMP-2) for different durations. qNMR spectra were acquired with the following parameters: 90° pulse = 9.4 μs, repetition time ≥ 7×T₁, flip angle = 90°, dead time = 200 ms, sample rotation = OFF, NS = 8, and receive gain = 56.

In vivo ¹⁹F MRI Studies

All animal procedures were conducted in accordance with the National Institute of Health Guidelines for the care and use of laboratory animals and were approved by the Institutional Animal Care and Use Committee of Xiamen University. BALB/c nude mice (male, 6 weeks old) were obtained from the Laboratory Animal Center of Xiamen University. S180 or HT-1080 cells (5 × 10⁶ in 100 μL PBS) were injected subcutaneously into the right rear flank areas of the mice. The mice were used when the tumor grew to ~5 mm in diameter. The mice were then intratumorally injected with 50 μL probes ([F] = 200 mM, BMMIBF₄-pH(thin) into S180 tumors, BMMIBF₄-MMP and BMMIBF₄-MMP(d) into HT-1080 tumors). No imaging guidance was needed. ¹H/¹⁹F MR images of the same slices were acquired before and 2 h after injection on a 9.4 T MRI scanner (BioSpec 94/20 USR, Bruker) equipped with commercially available ¹H/¹⁹F MRI coils. ¹H MRI was acquired using a RARE sequence: T_R/T_E = 2,500/33 ms, 256 × 256 matrices, slices = 9, thickness = 1 mm, average = 1, FOV = 40 × 40 mm². ¹⁹F MRI was acquired using an FLASH sequence: T_R/T_E = 1,600/1.3 ms, FA = 30°, 32 × 32 matrices, thickness = 10 mm, average = 20, FOV = 40 × 40 mm², acquisition time = 17 min.

HIGHLIGHTS.

Fluorinated ion liquids act as a new type of fluorine agents for ¹⁹F MRI

The ¹⁹F probes are constructed by a modified ion liquids phase-transition method

Activatable ¹⁹F probes are able to detect and image biological targets

The Bigger Picture.

Clinical ¹H MRI often suffers from low contrast and imaging artifacts because of intrinsic ¹H signals from endogenous water molecules. ¹⁹F MRI is emerging as a promising complement to conventional ¹H MRI because ¹⁹F MRI by introducing exogenous fluorine nuclei is considered to be a zero-background imaging means. Here, we report a fluorinated ionic liquid-based activatable ¹⁹F MRI platform for development of smart ¹⁹F probes. Several types of fluorinated ion liquids in molten state are loaded in the carriers and sealed with various stimuli-responsive diblock copolymers, leading to an “off” ¹⁹F signal. Upon activated by stimuli, a “turn-on” ¹⁹F signal lights up the biological targets in the living animals without background. This work significantly contributes to the development of stimuli-responsive ¹⁹F probes, which is highly important in ¹⁹F MRI and disease diagnosis.