Research Article | Open Access
Henryk M. Faas, James L. Krupa, Alexander J. Taylor, Francesco Zamberlan, Christopher J. Philp, Huw E. L. Williams, Simon R. Johnson, Galina E. Pavlovskaya, Neil R. Thomas, Thomas Meersmann, "Accelerated 19F·MRI Detection of Matrix Metalloproteinase-2/-9 through Responsive Deactivation of Paramagnetic Relaxation Enhancement", Contrast Media & Molecular Imaging, vol. 2019, Article ID 4826520, 13 pages, 2019. https://doi.org/10.1155/2019/4826520
Accelerated 19F·MRI Detection of Matrix Metalloproteinase-2/-9 through Responsive Deactivation of Paramagnetic Relaxation Enhancement
Paramagnetic gadolinium ions (GdIII), complexed within DOTA-based chelates, have become useful tools to increase the magnetic resonance imaging (MRI) contrast in tissues of interest. Recently, “on/off” probes serving as 19F·MRI biosensors for target enzymes have emerged that utilize the increase in transverse ( or T2) relaxation times upon cleavage of the paramagnetic GdIII centre. Molecular 19F·MRI has the advantage of high specificity due to the lack of background signal but suffers from low signal intensity that leads to low spatial resolution and long recording times. In this work, an “on/off” probe concept is introduced that utilizes responsive deactivation of paramagnetic relaxation enhancement (PRE) to generate 19F longitudinal (T1) relaxation contrast for accelerated molecular MRI. The probe concept is applied to matrix metalloproteinases (MMPs), a class of enzymes linked with many inflammatory diseases and cancer that modify bioactive extracellular substrates. The presence of these biomarkers in extracellular space makes MMPs an accessible target for responsive PRE deactivation probes. Responsive PRE deactivation in a 19F biosensor probe, selective for MMP-2 and MMP-9, is shown to enable molecular MRI contrast at significantly reduced experimental times compared to previous methods. PRE deactivation was caused by MMP through cleavage of a protease substrate that served as a linker between the fluorine-containing moiety and a paramagnetic GdIII-bound DOTA complex. Ultrashort echo time (UTE) MRI and, alternatively, short echo times in standard gradient echo (GE) MRI were employed to cope with the fast 19F transverse relaxation of the PRE active probe in its “on-state.” Upon responsive PRE deactivation, the 19F·MRI signal from the “off-state” probe diminished, thereby indicating the presence of the target enzyme through the associated negative MRI contrast. Null point 1H·MRI, obtainable within a short time course, was employed to identify false-positive 19F·MRI responses caused by dilution of the contrast agent.
The measurement of enzyme activity in vivo is a major challenge for the development of enzyme-specific chemical probes and will facilitate a deeper understanding of the role of enzymes in biological processes and facilitate drug discovery, synthetic biology, and metabolic engineering research . Matrix metalloproteinases (MMPs), a subclass of proteases, are zinc(II)-dependent enzymes that typically possess two ZnII ions, one for structural purposes and the other critical for catalysis. In humans, there are 24 different genes which code for 23 different MMPs, with the first and last gene coding for the same MMP . Expressed in a latent form prior to activation extracellularly, MMPs are an attractive drug target, due to their selective activation location outside the cell. MMPs remodel the extracellular matrix (ECM) and also play key roles in a range of physiological processes, including wound healing , organogenesis, and modulation of inflammatory processes. MMPs are also involved in pathological processes such as chronic obstructive pulmonary disease (COPD)  and can be used as biomarkers for determining stages of cancer, where the levels of MMPs are correlated to metastatic potential .
Chemical probes to detect MMPs have been developed in the field of fluorescent microscopy. Förster resonance energy transfer- (FRET-) based MMP probes have been available since the early 1990s , but in vivo imaging applications are hampered by the limited tissue penetration of the short wavelength light, inherently used for excitation and emission in fluorescent techniques. As a result, ex vivo studies have been performed, for example, using probes to measure serum MMP levels . An alternative and noninvasive imaging modality is magnetic resonance imaging (MRI). Chemical probes or responsive contrast agents have been previously developed for use in 1H·MRI, providing information on a range of catalytic  and noncatalytic processes such as monitoring pH , redox dysregulation , and levels of metal ions such as zinc , calcium , and copper . However, to date, there have only been a small number of examples of MMP MRI probes . Notable examples include iron oxide nanoparticles which have been used to show changes in T2 relaxation times upon conjugation with MMPs  and contrast agents which use the concept of a solubility switch in which the T1 value is altered due to the precipitation resulting from MMP ligand cleavage .
A key hurdle in the successful application of 1H·MMP probes is the confounding background signal from endogenous water in the body, which makes contrast determination difficult. To overcome this problem, efforts have focussed on 19F, for which there is little detectable 19F background signal in the body (except for bones, teeth and calcified tissue), leading to excellent specificity . This is an important requirement for a molecular probe and a key advantage over 1H·MRI . An exciting approach to harness the effect of paramagnetic GdIII upon 19F relaxation was proposed by Mizukami et al. to study the enzyme activity using a peptide of four amino acid residues that acts as a substrate for caspase-3. A trifluoro aryl ether at the carboxy terminus of the caspase-3 peptide substrate was linked with a GdIII-DOTA chelate at its amine terminus . In the initial uncleaved form, the fluorine signal is weak due to severe line broadening by fast transverse relaxation (i.e., short T2 times). However, in the presence of the specific enzyme, the linker was cleaved and the distance between the GdIII DOTA and fluorine increased, causing a reduction of the paramagnetic influence on fluorine. The decreased transverse relaxation led to an increase of the 19F signal due to line narrowing. The line-narrowing response of the probe molecule—or biosensor molecule—produces detectable MRI signal in regions where the biosensor is co-located with the enzyme target. Using and validating this concept, further 19F biosensor molecules were designed to probe for -galactosidase and -lactamase activities [20–22]. Previous developments also include a dual fluorescence 19F probe  and a dual 1H-19F·MRI probe which in its precleaved form contained a labile carbamate with a trifluoromethyl group in close proximity to a chelated GdIII ion . The 1H-19F probe was then activated by -galactosidase, which cleaved the carbamate resulting in the release of the 19F source, in turn increasing the 19F·MRI signal intensity. For MMP-2 detection, equivalent to the present work, an “off/on” probe with nine equivalent 19F atoms as signal source has been reported by Yue et al., which utilizes this effect of cleavage to alter the 19F transverse relaxation properties [25, 26].
Although significant progress has been made with nonspecific 19F·MRI tracers for cell labelling, including clinical applications [27, 28], the use of specifically targeted 19F·MRI probes for molecular imaging, in particular for in vivo studies, is frustrated by inadequate 19F·MRI signal intensity . The MMP biosensor concept described above utilizes reduced transverse (T2 or ) relaxation; however, the longitudinal relaxation is also reduced, and long T1 times diminish the amount of signal averaging that is feasible within the time span of typical in vivo MRI experiments.
In this work, we have explored a modified MMP 19F biosensor detection protocol where the observation concept is turned around by detecting the intact biosensor instead of the cleaved molecule. The short 19F·T1 relaxation time of the intact biosensor molecule is utilized for paramagnetic relaxation enhancement (PRE) [30–32] to improve the 19F·MRI signal to noise ratio (S/N) through rapid signal averaging. MRI protocols are adjusted to handle the fast transverse relaxation of PRE-activated biosensors. Responsive cleavage of the biosensor by the target protein causes PRE to be “switched off” and the MRI signal to disappear. The resulting negative MRI contrast is demonstrated to serve as an indicator for the presence of MMP biomarker molecules.
2. Materials and Methods
2.1. Synthesis of the Biosensor
To advance the concept of responsive PRE deactivation, two different “on/off” 19F·MRI probes sensitive to MMP cleavage have been synthesized using a short customisable synthetic route. The probes can be split into three segments: (1) the fluorine-containing moiety, which delivers the signal; (2) the paramagnetic moiety for PRE, which consists of a GdIII-bound DOTA-complex; (3) the protease substrate which links together the other two constituents and can be tailored to serve as substrate to specific target for MMPs (see Figure 1 and also Supporting Materials S1 for selected sequences). The main focus of the current work is on the concept of molecular contrast through responsive PRE deactivation using molecular probes that are mechanistically similar to the MMP-2 selective probe previously utilized by Yue et al. for contrast . Some changes in the probe structure and its synthesis have been made compared to . In the previous work, a PEG linker was incorporated for solubility of the fluorine-containing moiety and was attached to the MMP substrate peptide via Michael addition of maleimide to the thiol of the cysteine-containing substrate peptide. In the current work, this type of linker, known to be labile due to thiol exchange in vivo , was avoided, and the fluorine moiety was directly incorporated as amino acid side chain in the peptide sequence; furthermore, GdIII-bound DOTA-complex was linked to the peptide sequence using a stable 1,4-triazole formed using “click” chemistry. The molecule in this work is soluble in 9 : 1 (v/v) water/acetonitrile up to at least 1.2 mM concentration without a PEG linker, and the solvent was acceptable for the proof of concept work and tolerated by the enzyme. Furthermore, Yue et al. used peptide-coupling chemistry to attach an octadentate GdIII-DOTA chelator as the final unit to their peptide. This synthesis required 30 synthetic steps, two of which required purification by high-performance liquid chromatography (HPLC). The probes in the current work required 27 synthetic steps, with purification via HPLC reserved for only the final step adding some efficiency. The synthesis is described in detail in Supplementary Materials S1–S3.
2.2. MMP Preparation
Recombinant human carrier-free MMP-1, MMP-2, MMP-9, and MMP-12 (R&D Systems, Minneapolis, Minnesota) were diluted upon receipt to 75 µg/ml in 50 mM Tris, 10 mM·CaCl2, 150 mM·NaCl, 0.05% (w/v) Brij 35, and pH 7.5. MMPs were activated by incubation with 4-aminophenylmercuric acetate (APMA) at 100 mM in dimethyl sulfoxide (DMSO) (Sigma Aldrich, UK) was added to a final concentration of 1 mM. Activation incubation times at 310 K were 1 hour for MMP1-1 and MMP-2 and 24 hours for MMP-9 and MMP-12. Following incubation, enzymes were aliquoted and frozen at 193 K until required. A 5 μl aliquot was thawed and added to 600 µl of the biosensor solution, immediately prior to the NMR experiments.
2.3. Enzyme Activity Time Course Observed through 19F·NMR Line Shape at 14.1 Tesla
19F·NMR/MRI·MMP probe (0.1 mM in 9 : 1 (v/v) H2O/D2O) was incubated with MMP-9 (5 µL, at 10 µg/mL, final concentration 83 ng/mL) in a total volume 600 µL adjusted to pH 7.5. The time point t = 0 for each time course was right after MMP addition; subsequently, the sample was reintroduced into the NMR magnet and kept at the temperature specified. The biosensor molecule (Figure 1) results to a single peak at –64 ppm (relative to CFCl3 at 0 ppm). The 19F·NMR lineshape measurements (Figure 2) were obtained in a 5 mm sample diameter high-resolution probe head with 19F inner coil for acquisition (1H outer coil for decoupling—not used) tuned to the 19F frequency of 564.6 MHz using a Bruker Avance III 600 MHz (14.1 Tesla) spectrometer. The fluorine linewidth (FWHM) was used as a measure of the transverse relaxation, . Measurements were taken at a temperature of either 298 K or 310 K as indicated in the figure and main text. The time dependence of the signal intensity was analyzed using Lambert functions for a Michaelis-Menten fit to obtain the enzyme efficiency parameter listed in Table 1.
2.4. 19F·NMR T1 Relaxation Measurements at 14.1 Tesla
Longitudinal relaxation rates (T1) for the fluorinated molecular probes were determined in an inversion recovery experiment at 298 K with 14 to 28 inversion recovery times and a repetition time (TR) of up to 8 seconds to insure full magnetization recovery (Table 2). NMR spectrometer and probe head were identical to those described above the linewidth measurements.
H2O as a solvent denotes v/v 9 : 1 H2O/D2O.
2.5. 19F·NMR T1 and T2 Relaxation Measurements at 9.4 Tesla
All bulk relaxation measurements (i.e., without spatial resolution) at 9.4 Tesla were obtained for samples contained within a single 5 mm NMR sample tube located in the centre (i.e., axial location) of a 30 mm (sample diameter) 19F microimaging probe head (Bruker) tuned to 376.5 MHz using a Bruker Avance III 400 MHz (9.4 Tesla) spectrometer and microimaging system. The temperature for all 9.4 Tesla measurements was 293 K. The T1 and T2 relaxation times of intact (PRE activated) 19F biosensor dissolved at concentrations ranging from 0.3–1.2 mM in 9 : 1 (v/v) water/acetonitrile (ACN) were obtained using standard inversion recovery and spin echo experiments, respectively. A 82 rectangular 90° pulse was used for excitation, and 164 rectangular 180° pulses was used for inversion and refocusing. The integrated signal intensity and signal heights were obtained from 12–14 inversion recovery and spin echo measurements with a signal averaging of 1024 scans. Likewise, inversion recovery and spin echo measurements were also performed with the cleaved biosensor after completed reaction with the MMP. Recovery times of at least 5 T1 times were used for all measurements (up to 2.5 s) at the range of recovery and spin echo times that allowed for at least 90 % recovery or 90% decay, respectively. The resulting curves were analysed using Igor Pro 7, and the average from signal intensity curves and integrated intensities are reported in Figure 3.
2.6. 19F·MRI Protocols at 9.4 Tesla to Monitor MMP Activity
All MRI were recorded using a Bruker Avance III 400 MHz (9.4 Tesla) microimaging system with a 19F frequency of 376.4 MHz at 293 K. The sample for MRI were contained within three or five (see Figures 4 and 5, respectively) 5mm NMR sample tubes located in a custom-made 30 mm sample holder placed into a 30 mm 19F Bruker microimaging probe head tuned to 376.5 MHz. The samples in each tube are described in the figure captions.
2D transverse ultrashort echo time (UTE) 19F·MR images (30 mm slice thickness) of a sample containing three 5 mm NMR tubes (Figure 4) were obtained from 202 projections and used a polar undersampling factor of 1. Images were reconstructed to a 64 × 64 data point resolution. The field of view was FOV = 25 mm × 25 mm. A 300 s Gaussian 90° excitation pulse was used, TE was 191 s, repetition time was 5 ms, and 360 averages were taken that lead to a total scanning time of 6 min.
2D transverse spin echo 19F·MRI (30 mm slice thickness) of sample containing three 5 mm NMR tubes (Figure 4) used a 1 ms Gaussian 90° excitation pulse, 10.29 ms echo time, 200 ms repetition time, and 360 averages. FOV = 30 mm × 30 mm at 32 × 32 point resolution (raw data) recorded at a total acquisition time of 77 min. For processing, a sinebell apodization and zero filling to 64 × 64 data points were applied.
2D transverse gradient echo 19F·MRI (50 mm slice thickness) of sample containing five 5 mm NMR tubes (Figure 5) used 873 s echo time, 20 ms repetition time, and 1536 averages. FOV = 30 mm × 30 mm at 32 × 32 point resolution (raw data) recorded at a total acquisition time of 16 min. For processing, a sinebell apodization and zero filling to 64 × 64 data points were applied.
2.7. 1H·MRI Protocols at 9.4 Tesla
The 2D gradient echo (GE) 1H·MRI (2 mm slice thickness) of sample containing five 5 mm NMR tubes (Figure 5) used a standard GE protocol with 1.4 ms rectangular 90° excitation pulse, 2.5 ms echo time, 2000 ms repetition time, and no signal averaging. FOV = 30 mm × 30 mm at 64 × 64 data point (raw) resolution recorded at a total acquisition time of 2½ min at 293 K. For processing, a sinebell apodization and zero filling to 128 × 128 points were applied.
The 2D null point 1H·MRI (2 mm slice thickness) of the sample in Figure 5, containing five 5 mm NMR tubes, used an inversion recovery protocol with 4.84 ms rectangular 180° inversion pulse, 220 ms inversion recovery time, 2.5 ms echo time, 2000 ms repetition time, and no signal averaging. FOV = 30 mm × 30 mm at 64 × 64 data point resolution recorded at a total acquisition time of 2½ min at 293 K. For processing, a sinebell apodization was applied with zero filling to 128 × 128 points.
3. Results and Discussion
3.1. Following Enzymatic Activity through NMR Linewidth
Intramolecular paramagnetic relaxation, the dominating relaxation mechanism for the 19F nuclear spins in the intact biosensor molecule, will be replaced by intermolecular paramagnetic relaxation after biosensor cleavage through MMP (Figure 1). The spatial separation r between the 19F-containing unit and the paramagnetic Gd DOTA complex will substantially increase by the cleavage, causing a dramatic reduction in the 1/r6-dependent paramagnetic relaxation for 19F. This means that 19F·PRE is effectively “turned off” by the biosensor response to MMP.
The 19F·NMR linewidth narrowing, associated with cleavage, as previously utilized by Yue et al. [25, 26], was used in this work for the initial testing of the biosensor and to explore both temperature and also isotope effects on the sensor kinetics. The isotope effect of deuterium on the cleavage reaction is of interest as D2O is typically used as a lock solvent in NMR spectroscopy. The MMP-2/-9 selective probe was incubated with APMA- (4-aminophenylmercuric acetate-) activated MMP-9 under three different sets of conditions (i.e., D2O at 298 K, H2O at 298 K, and H2O at 310 K). Before the addition of MMP-9 the 19F·NMR peaks observed were broad, with a full width at half maximum (FWHM) of ∼60 Hz (310 K). After addition of MMP-9, a time series of 19F·NMR experiments was performed to monitor the progress of the enzymatic reaction. The observed line narrowing with reaction time in Figure 2(a) indicates the progressive cleavage of the biosensor with associated reduction in the transverse paramagnetic relaxation.
The amplitude of the single 19F peak at −64 ppm (relative to CFCl3 at 0 ppm) was measured and used to plot peak amplitude versus time. Postcleavage, the FWHM was reduced to 19 Hz (310 K), caused by the increase in distance between the GdIII ion and the 19F signal source. These experiments were repeated with the broad-range probe under identical conditions, and a FWHM of ∼53 Hz precleavage, and 13 Hz postcleavage (310 K) was observed (not shown).
The time course of the NMR signal under the three different sets of conditions (varying solvent and temperature) is shown for the MMP-2/-9 probe in Figure 2(b) and for the broad-range probe in Figure 2(c). In order to compare the relative conditions and probes, we calculated the specificity constant (enzyme efficiency), kcat/KM, for both probes under each set of condition (Table 1). Here, kcat is the turnover number (per second) and KM the apparent (Michaelis) binding constant. The enzyme efficiency was calculated by mapping the time dependence of amplitude of the signal peak and then applying a Michaelis-Menten based fit using Lambert functions and least squares fitting .
aThe value was calculated using the relation, , where is the linewidth of the fluorine peak at full width half maximum. bThe inability to properly identify the exact linewidth of the precleaved state due to the broad linewidth; hence, the maximum linewidth is used, yielding a minimum value.
At physiological temperatures, the MMP-2/-9 probe was almost completely cleaved within an hour. The enzyme was fivefold more efficient (seen in the change in kcat/KM) if the solvent contained 10% v/v D2O rather than 100% D2O (at 298 K). A similar trend was observed with the broad-range probe, where the enzyme was twice as efficient in 10% v/v D2O rather than pure D2O. Intriguingly, we observed that, in contrast to the specific MMP-2/-9 probe, the turnover of the broad-range probe was slower at physiological conditions than at lower temperatures, with kcat/KM at 310 K for the broad-range probe reaching only a third its value at 298 K. One explanation may be that at higher temperature, the broad-range probe may occupy a different conformation, making it less readily accepted by MMP-9 and thereby increasing KM and/or decreasing kcat for the generic probe.
When compared to the FRET substrate described by Knight et al. , the NMR probe is substantially more slowly turned over, an observation which can potentially be attributed to the relatively bulky GdIII-DOTA chelate reducing the MMP-9 binding affinity for the NMR probe.
3.2. Selectivity of the Probe
To test the specificity of the probe, the MMP-2/-9 probe was incubated with MMP-1 and MMP-12, leading to no observable cleavage, as confirmed by mass spectrometry to detect the cleaved products (data not shown). The broad-range probe however responded as expected to all MMPs tested (MMP-1, MMP-2, MMP-9, and MMP-12).
3.3. Transverse and Longitudinal 19F Relaxation of the 0.1 mM Biosensor in Aqueous Solution
Longitudinal 19F relaxation times T1 for the fluorinated probes and transverse relaxation times , determined by inversion recovery experiments and from the fluorine linewidth (FWHM) , respectively, are listed in Table 2. Upon cleavage, the two different MMP probes display drastic changes in their T1 and times. The MMP-2/-9 probe shows an increase in T1 by a factor of over 60, whereas the generic probe increases the T1 time by a factor of 75. Cleavage also leads to an increase in by a factor of about 7 and 10 for the MMP-2/-9 probe and the generic probe, respectively, and by a factor of 20 in the previous work at 7 Tesla [25, 26]. The linewidth effect, utilized in previous literature, may therefore lead to a less pronounced change in the MRI contrast than that caused by the reduction of the longitudinal relaxation. Note that however the relaxation differences between the intact and the cleaved sensor are strongly dependent on the sensor concentration as explored in detail further below.
3.4. 19F Relaxation Behaviour of the Biosensor at 9.4 Tesla
The data in Figure 2 and Table 2 were obtained with a 5 mm high-resolution probe at 14.1 Tesla (564.6 MHz 19F resonance frequency) and a biosensor concentration of 0.1 mM. The biosensor concentration was increased to ≥0.4 mM to provide sufficient signal intensity for molecular 19F·MRI at 9.4 Tesla (376.4 MHz 19F resonance frequency) using a 30 mm 19F microimaging coil for excitation and detection. To dissolve the biosensor up to a 1.2 mM concentration, a 9 : 1 (v/v) H2O/acetonitrile mixture was used as a solvent at 293 K. The effect of acetonitrile on the kinetics of biosensor cleavage by the enzyme MMP-2 was studied in assays (not shown here). Although the presence of 10% ACN reduced the reaction kinetics, it still allowed for cleavage of the sensor with MMP-2. All relaxation measurements at 9.4 T were performed at a temperature of 293 K.
The biosensor concentration [Rx] dependence of the 19F relaxation rate, i.e., the biosensor 19F relaxivity, is crucial for the concept development of molecular MRI contrast explored here. As shown in Figure 3 and listed in Table 3, the intact MMP-2/-9 biosensor does not exhibit a concentration dependence for the longitudinal relaxation with T1 = 10.7 ± 1.0 ms. The observed value is about 30% shorter than the value listed in Table 2 with T1 = 15.1 ms for the intact biosensor at 0.1. mM concentration obtained in aqueous solution (containing 10% D2O) at 14.1 Tesla magnetic field strength and 298 K. Similarly, the transverse relaxation of the intact biosensor in Figure 3 exhibited little concentration dependence, and the average value was found to be T2 = 3.46 ± 0.42 ms (note that T2 relaxation times and rates, as determined by spin echo measurements, are reported here as they provide more precise data than measurements that are affected by sample susceptibility). Upon catalytic cleavage with MMP, the relaxation rates are dramatically reduced. At the same time, the relaxation rates start to exhibit a strong dependence upon (cleaved) biosensor concentration [Rx]. However, even at a relatively high concentration of [Rx] = 1.0 mM, the 19F longitudinal relaxation rate is slowed down by a factor of 25 upon cleavage, leading to a longitudinal relaxation time of T1 = 267 ± 37 ms. Similarly, transverse relaxation is slowed 20-fold upon cleavage, resulting to a transverse relaxation time of T2 = 72.7 ± 2.5 ms.
Data are for biosensor with MMP-2/-9-specific sequence using MMP-2 as the cleavage enzyme.
The transverse and longitudinal relaxivity (i.e., [Rx] dependence of 19F relaxation) obtained from data in Figure 3 is listed in Table 3. The T1 relaxivity for the cleaved biosensor interpolated for 9 : 1 (v/v) H2O/ACN solutions to a concentration of [Rx] = 0.1 mM (Figure 3 and Table 3) is T1 = 796 ms, a value that falls about 17% short of the value T1 = 960 ms listed in Table 2 for the aqueous solution (10% D2O), indicating a qualitatively very similar relaxation behaviour of the cleaved biosensor within the two solvents.
3.5. Molecular Imaging Using Negative T1 Weighted MRI Contrast through PRE Deactivation
The results of Figure 3 show that, upon cleavage, both transverse and longitudinal relaxations slow down by more than an order of magnitude, even if very high concentrations of up to 1.2 mM biosensor are being used. The effect on transverse relaxation has been exploited in the past to generate positive MRI contrast through increasing 19F biosensor signal intensity that indicates the presence of MMP. Figure 4 shows a repeat of this concept with the MMP-2/-9-specific biosensor (Figure 4(c)) but also demonstrates (Figure 4(b)) that T1 contrast can be exploited through ultrashort imaging time (UTE) 19F·MRI [28, 36, 37]. The T1 weighted MRI produces a negative contrast that is caused by the deactivation of the paramagnetic relaxation effect (PRE) on the 19F signal upon cleavage through MMP. An important advantage of the T1 weighted contrast, as for PRE in general, is the fast data acquisition that significantly reduces the required experimental time.
The UTE images in Figure 4 were employed to allow for MRI of the intact biosensor despite the short T2 times and fast repetition times (5 ms recycle delay) enabled recording within 6 min each, whereas the spin echo MR images that used T2 contrast took 75 min each due to the long recycle delay of 200 ms to allow for sufficient, but still incomplete, T1 relaxation of the cleaved sensor. T1 weighted 19F·MRI contrast of the intact biosensor with active PRE benefits from the UTE methodology that enables the recording of signals with very short T2 times. However, as shown in Table 3, the intact biosensor has a fairly large transverse to longitudinal relaxation ratio of T2/T1 0.3 with T2 = 3.46 ms and UTE MRI, although helpful, is not absolutely required. UTE MRI is an excellent methodology for systems with very short transverse relaxation times, but it is experimentally demanding (i.e., requires trajectory calibration), in particular if used for systems with inherently low signal intensity. Figure 5 demonstrates that, for the particular molecule used in this study, 19F biosensor imaging is also feasible through a simple gradient echo (GE) sequence. The 19F·GE·MR images shown in Figures 5(b) and 5(d) were each recorded within 16 min using an echo time of TE = 0.873 μs (the shortest possible with the hardware used), signal averages NS = 1536, and a repetition time of TR = 20 ms. Figure 5(b) shows three sample tubes with 0.6 mM biosensor solution. The biosensor in all three sample tubes was 0.6 mM (i.e., half the concentration of the samples used for UTE MRI in Figure 4).
The appearance of negative 19F contrast of the biosensor obtained through the T1 weighted protocols described in this work indicates the presence of MMP through responsive PRE deactivation. However, transport phenomena within in vivo organisms may cause negative contrast through biosensor dilution and therefore would produce a false-positive response. To distinguish the MMP catalysed biosensor reaction from the false positive due to biosensor dilution, it is instructive to utilize additional T1 sensitive 1H·MRI measurements. Figure 5(b) shows the 19F·MRI of three sample tubes, (i), (ii), and (iii) that contain the intact biosensor at 0.6 mM concentration. In Figure 5(d), only sample tube (i) contains the intact biosensor at this concentration while MMP-2 was administered to sample (ii), and sample tube (iii) was diluted to 0.4 mM concentration of the intact biosensor. The 19F·MR image in Figure 5(d) shows very similar reduction in intensity for the two samples (ii) and (iii). After >5 h incubation with MMP, the 19F longitudinal relaxation time of sample (ii) displayed an increase from T1 = 10.7 ms for the intact sensor to approximately T1 = 275 ± 15 ms (i.e. using single exponential fit), a value that remained stable throughout the further experiments. The observed 19F relaxation time value falls a little short of T1 = 380 ms, the expected value from Table 3 for the completely reacted sensor at 0.6 mM concentration. The cause of this deviation was not further investigated as it is of little consequence for the proof of concept demonstrated here. For better visualization, Figure 5(f) shows the result from the subtraction of the MR image of Figure 5(d) from Figure 5(b).
The T1 relaxation in 1H·MRI is strongly affected by the paramagnetic Gd(III) concentration, but the relaxation is always intermolecular in nature. Therefore, unlike 19F·T1 relaxation, the 1H·T1 relaxation is not affected by the cleavage of the gadolinium group from the 19F containing moiety. The 1H longitudinal relaxation of the 0.6 mM intact sensor in (ii) was found to be T1 = 320 ms both before and after incubation with MMP. Figure 5(a) shows the 1H gradient echo MRI of three tubes with 0.6 mM biosensor before MMP and dilution; tube (vi) contains 0.4 mM of a completely cleaved biosensor (recycled from a previous experiment) that causes a weak signal in Figures 5(b) and 5(d); and sample tube (v) contains only PBS solution without 19F or gadolinium. Figure 5(c) depicts 1H null point MRI using inversion recovery protocol with an inversion recovery time TI = 220 ms that causes all signals from tubes containing 0.6 mM biosensor solution to vanish due to zero crossing.
Figure 5(e) shows the result from an identical 1H null point MRI protocol but with the samples as in Figure 5(d) (i.e., (ii) partially reacted through MMP and (iii) diluted to 0.4 mM intact biosensor). As expected, there is no observable change in 1H·T1 relaxation of the reacted biosensor in (ii), but the diluted sample in (iii) leads to a strong inversion recovery signal increase due to reduced 1H·T1 relaxation that produces a clearly visible 1H·MRI response. The 1H·MRI response, shown in Figure 5(e), identifies the 19F·MRI response for sample (iii) as false-positive. For better visualization, Figure 5(g) depicts the 1H null point MRI difference between Figures 5(e) and 5(c).
3.6. Towards In Vivo MRI: Tasks for MMP Biosensor Development
The results presented in Figures 4 and 5 demonstrate the effectiveness of responsive PRE contrast for molecular imaging but also indicate where future development is required. False-positive signals due to biosensor dilution can be eliminated by fast null point 1H·MRI measurements of H2O that was completed within a total recording time of 2½ min in Figure 5. Within an organism, the biosensor localization may not be known quantitatively and, additionally, dilution due to transport mechanisms will occur over time. Therefore, 1H·MRI·T1 maps may provide detailed insights into biosensor concentration throughout the organism. To generate “snapshots” of the concentration distribution, the T1 maps need to be recorded quickly which should not be a problem as abundant water molecules are being detected and improved 1H·MRI protocols and/or reduced resolution may accelerate the acquisition further. Detailed GdIII concentration maps will then enable 19F·PRE·MRI data analysis to extract quantitatively the extent of the reaction. It is important to note that the previously explored transverse () relaxation-based biosensor detection concept can also benefit from 1H·MRI·T1 maps. In analogy to a false-positive response with a PRE-based biosensor, a false-negative response may occur for biosensors because dilution will diminish the 19F·MRI signal, and this can mask the presence of the target enzyme. As a word of caution for NMR spectroscopic studies, the rate of enzymatic hydrolysis of the probes was found to be significantly affected by the percentage of D2O present in the solution.
To advance the responsive PRE deactivation concept towards in vivo applications, future biosensor synthesis will need to address three key issues: (1) The biosensors will need to be fully water soluble to enable the concentration required for MRI. At the same time, the water soluble biosensor needs to be stable within the in vivo environment. (2) A relatively high concentration of the probe molecule was required in order to perform the MR imaging experiment in the current and previous studies. The 19F signal intensity  requires further improvement to be viable for in vivo applications. Generally, various successful efforts have been made to improve the signal intensity by increasing the 19F spins per molecule unit [38, 39], and these approaches need to be explored to make responsive PRE deactivation more biocompatible by reducing the required concentration of biosensor molecules. Furthermore, reducing the required biosensor concentration will also enhance the molecular MRI contrast due to an increased T1 relaxivity gap between the PRE-activated and the PRE-deactivated probe molecule. As shown in Figure 3, the relaxivity of the “off-state” exhibits a strong concentration dependence while the “on-state” relaxivity is largely unaffected by concentration changes. (3) A paramagnetic group producing fast T1 but relatively slow T2 relaxation (i.e., a high ratio close to the limit T2/T1 = 1) will further improve the presented concept. The paramagnetic system used in this work, GdIII, generally produces very unfavourable T2/T1 ratios that are also dependent on the separation between the paramagnetic centre and the fluorine spins . Distance cannot always be freely selected in functionalized sensors but a variety of better options than GdIII for the paramagnetic centre, such as FeII, TmIII, and HoIII, have been explored to improve 19F·MR signal intensities through increased the T2/T1 ratios [34, 39, 40]. Combining high T2/T1 ratios with an increased number of 19F spins, Kislukhin and coworkers have reported promising results with paramagnetic fluorinated nanoemulsions .
Lastly, the presented concept may also work for molecular imaging with hyperpolarized (hp) 129Xe, a contrast agent that is becoming more readily available [41, 42]. A recent study demonstrated a conceptual “on/off” probe with a cryptophane cage that temporarily binds xenon atoms and brings them into close proximity to the paramagnetic GdIII centre. Deactivation of the paramagnetic relaxation of the probe molecule caused an eight-fold reduction in the T1 relaxation of 129Xe in the solvent . Note that hp 129Xe can in principle be added long after the probe molecule has been administered to an organism and after any deactivating reaction has occurred in regions with up-regulated biomarker targets. Replacement of 19F with exogenous hp 129Xe for MMP biosensor probe molecules may therefore be an option to significantly increase the signal from such enzymatic probes.
The usage of responsive PRE deactivation for T1 weighted 19F·MRI contrast, presented in this work, enables accelerated observation of MMP enzyme activity by taking advantage of the significantly reduced imaging time compared to the previously utilized T2 or weighted MRI. UTE 19F·MRI can be employed if problems associated with short 19F·T2 relaxation need to be overcome. In this work, UTE·19F·MRI allowed for a more than 12-fold reduction in the total acquisition time compared to a T2 weighted spin echo MRI protocol. Alternatively, even a simple T1 weighted gradient echo sequence enabled molecular MRI of MMP-2 within a short acquisition time and with a reasonable signal to noise ratio despite the unfavourable T2/T1 ratio of the GdIII system used in this proof of concept study. Responsive PRE deactivation for T1 weighted MRI leads to a negative contrast where the intact biosensor is observed and the disappearance of the MRI signal indicates cleavage due to enzyme activity. False-positive identification, where vanishing 19F·MRI signals are caused by biosensor dilution, can be identified through additional T1 weighted 1H·MRI that is not affected by the biosensor cleavage. More progress is needed to advance targeted biosensors with responsive PRE deactivation towards in vivo and clinical applications. This includes improved molecular design of 19F biosensors but also the advancement of hyperpolarized 129Xe biosensors with responsive deactivation of paramagnetic relaxation . If successful, this concept offer significant prospects for monitoring disease progression and treatment impact with much improved precision and therefore could play an important role in personalized medicine.
The chemical synthesis data used to support the findings of this study are included in the supplementary information. The NMR spectroscopic and MRI data used to support the findings of this study are available from the corresponding author upon request.
Alexander J. Taylor is currently at Medical Physics, Nottingham University Hospitals NHS Trust, Nottingham NG7 2UH, UK, and Francesco Zamberlan is currently at School of Chemistry, University of Lincoln, Brayford Pool, Lincoln LN6 7TS, UK.
Conflicts of Interest
The authors declare that there are no conflicts of interest regarding the publication of this paper.
We would like to thank Dr. Weng Chan for access to a solid phase peptide synthesizer and Dr. Joshua Bray for technical MRI support. This work was supported by the Biotechnology and Biological Sciences Research Council (grant nos. BB/N021460/1 and BB/F017154/1), the BBSRC Tools and Resources Development Fund, and a BBSRC DTG studentship (J. L. Krupa), respectively; the Engineering and Physical Sciences Research Council via the EPSRC Bridging the Gaps award and the University of Nottingham (J. L. Krupa and A. J. Taylor); and the Medical Research Council by a MRC PhD studentship (A. J. Taylor).
Scheme S1: synthesis of fluorinated amino acid (4). Scheme S2: synthesis of “clickable” GdIII chelate. Scheme S3: SPPS of the broad-range probe and attachment of the GdIII-DOTA chelator (1). (Supplementary Materials)
- D. V. Hingorani, B. Yoo, A. S. Bernstein, and M. D. Pagel, “Detecting enzyme activities with exogenous MRI contrast agents,” Chemistry–A European Journal, vol. 20, no. 32, pp. 9840–9850, 2014.
- J. O. McIntyre and L. M. Matrisian, “Molecular imaging of proteolytic activity in cancer,” Journal of Cellular Biochemistry, vol. 90, no. 6, pp. 1087–1097, 2003.
- L. M. Coussens, B. Fingleton, and L. M. Matrisian, “Matrix metalloproteinase inhibitors and cancer: trials and tribulations,” Science, vol. 295, no. 5564, pp. 2387–2392, 2002.
- M. G. Belvisi and K. M. Bottomley, “The role of matrix metalloproteinases (MMPs) in the pathophysiology of chronic obstructive pulmonary disease (COPD): a therapeutic role for inhibitors of MMPs?” Inflammation Research, vol. 52, no. 3, pp. 95–100, 2003.
- L. A. Liotta, K. Tryggvason, S. Garbisa, I. Hart, C. M. Foltz, and S. Shafie, “Metastatic potential correlates with enzymatic degradation of basement membrane collagen,” Nature, vol. 284, no. 5751, pp. 67-68, 1980.
- C. G. Knight, F. Willenbrock, and G. Murphy, “A novel coumarin-labelled peptide for sensitive continuous assays of the matrix metalloproteinases,” FEBS Letters, vol. 296, no. 3, pp. 263–266, 2002.
- E. Song, D. Cheng, Y. Song, M. Jiang, J. Yu, and Y. Wang, “A graphene oxide-based FRET sensor for rapid and sensitive detection of matrix metalloproteinase 2 in human serum sample,” Biosensors and Bioelectronics, vol. 47, pp. 445–450, 2013.
- A. Y. Louie, M. M. Hüber, E. T. Ahrens et al., “In vivo visualization of gene expression using magnetic resonance imaging,” Nature Biotechnology, vol. 18, no. 3, pp. 321–325, 2000.
- D. L. Longo, P. Z. Sun, L. Consolino, F. C. Michelotti, F. Uggeri, and S. Aime, “A general MRI-CEST ratiometric approach for pH imaging: demonstration of in vivo pH mapping with iobitridol,” Journal of the American Chemical Society, vol. 136, no. 41, pp. 14333–14336, 2014.
- E. M. Gale, S. Mukherjee, C. Liu, G. S. Loving, and P. Caravan, “Structure-redox-relaxivity relationships for redox responsive manganese-based magnetic resonance imaging probes,” Inorganic Chemistry, vol. 53, no. 19, pp. 10748–10761, 2014.
- Y.-Q. Xu, J. Luo, and Z.-N. Chen, “Zn2+-responsive bimodal magnetic resonance imaging and fluorescence imaging agents and their interaction with human serum albumin,” European Journal of Inorganic Chemistry, vol. 2014, no. 20, pp. 3208–3215, 2014.
- K. D. Verma, A. Forgács, H. Uh et al., “New calcium-selective smart contrast agents for magnetic resonance imaging,” Chemistry–A European Journal, vol. 19, no. 52, pp. 18011–18026, 2013.
- E. L. Que, E. Gianolio, S. L. Baker, A. P. Wong, S. Aime, and C. J. Chang, “Copper-responsive magnetic resonance imaging contrast agents,” Journal of the American Chemical Society, vol. 131, no. 24, pp. 8527–8536, 2009.
- C. V. Gringeri, V. Menchise, S. Rizzitelli et al., “Novel Gd(III)-based probes for MR molecular imaging of matrix metalloproteinases,” Contrast Media & Molecular Imaging, vol. 7, no. 2, pp. 175–184, 2012.
- J. Gallo, N. Kamaly, I. Lavdas et al., “CXCR4-Targeted and MMP-responsive iron oxide nanoparticles for enhanced magnetic resonance imaging,” Angewandte Chemie International Edition, vol. 53, no. 36, pp. 9550–9554, 2014.
- M. Lepage, W. C. Dow, M. Melchior et al., “Noninvasive detection of matrix metalloproteinase activity in vivo using a novel magnetic resonance imaging contrast agent with a solubility switch,” Molecular Imaging, vol. 6, no. 6, pp. 393–403, 2007.
- K. Matsuo, R. Kamada, K. Mizusawa et al., “Specific detection and imaging of enzyme activity by signal-amplifiable self-assembling 19F MRI probes,” Chemistry–A European Journal, vol. 19, no. 38, pp. 12875–12883, 2013.
- E. McFarland, J. A. Koutcher, B. R. Rosen, B. Teicher, and T. J. Brady, “In vivo 19F NMR imaging,” Journal of Computer Assisted Tomography, vol. 9, no. 1, pp. 8–15, 1985.
- S. Mizukami, R. Takikawa, F. Sugihara et al., “Paramagnetic relaxation-based 19F MRI probe to detect protease activity,” Journal of the American Chemical Society, vol. 130, no. 3, pp. 794-795, 2008.
- A. Keliris, T. Ziegler, R. Mishra et al., “Synthesis and characterization of a cell-permeable bimodal contrast agent targeting β-galactosidase,” Bioorganic & Medicinal Chemistry, vol. 19, no. 8, pp. 2529–2540, 2011.
- S. Mizukami, H. Matsushita, R. Takikawa, F. Sugihara, M. Shirakawa, and K. Kikuchi, “19F MRI detection of β-galactosidase activity for imaging of gene expression,” Chemical Science, vol. 2, no. 6, pp. 1151–1155, 2011.
- H. Matsushita, S. Mizukami, Y. Mori et al., “19F MRI monitoring of gene expression in living cells through cell-surface β-lactamase activity,” ChemBioChem, vol. 13, no. 11, pp. 1579–1583, 2012.
- S. Mizukami, R. Takikawa, F. Sugihara, M. Shirakawa, and K. Kikuchi, “Dual-function probe to detect protease activity for fluorescence measurement and 19F MRI,” Angewandte Chemie International Edition, vol. 48, no. 20, pp. 3641–3643, 2009.
- A. Keliris, I. Mamedov, G. E. Hagberg, N. K. Logothetis, K. Scheffler, and J. Engelmann, “A smart 19F and 1H MRI probe with self-immolative linker as a versatile tool for detection of enzymes,” Contrast Media & Molecular Imaging, vol. 7, no. 5, pp. 478–483, 2012.
- X. Yue, Z. Wang, L. Zhu et al., “Novel 19F activatable probe for the detection of matrix metalloprotease-2 activity by MRI/MRS,” Molecular Pharmaceutics, vol. 11, no. 11, pp. 4208–4217, 2014.
- X. Yue, Z. Wang, L. Zhu et al., “Correction to “Novel 19F activatable probe for the detection of matrix metalloprotease-2 activity by MRI/MRS”,” Molecular Pharmaceutics, vol. 14, no. 4, pp. 1317-1318, 2017.
- E. T. Ahrens, B. M. Helfer, C. F. O'Hanlon, and C. Schirda, “Clinical cell therapy imaging using a perfluorocarbon tracer and fluorine-19 MRI,” Magnetic Resonance in Medicine, vol. 72, no. 6, pp. 1696–1701, 2014.
- A. A. Kislukhin, H. Xu, S. R. Adams, K. H. Narsinh, R. Y. Tsien, and E. T. Ahrens, “Paramagnetic fluorinated nanoemulsions for sensitive cellular fluorine-19 magnetic resonance imaging,” Nature Materials, vol. 15, no. 6, pp. 662–668, 2016.
- A. J. Taylor, J. Granwehr, C. Lesbats et al., “Probe-specific procedure to estimate sensitivity and detection limits for 19F magnetic resonance imaging,” PLos One, vol. 11, no. 10, Article ID e0163704, 2016.
- K. H. Chalmers, E. DeâLuca, N. H. Hogg et al., “Design principles and theory of paramagnetic fluorine-labelled lanthanide complexes as probes for 19F magnetic resonance: a proof-of-concept study,” Chemistry–A European Journal, vol. 16, no. 1, pp. 134–148, 2010.
- K. H. Chalmers, A. M. Kenwright, D. Parker, and A. M. Blamire, “19F-lanthanide complexes with increased sensitivity for 19F-MRI: optimization of the MR acquisition,” Magnetic Resonance in Medicine, vol. 66, no. 4, pp. 931–936, 2011.
- F. Schmid, C. Höltke, D. Parker, and C. Faber, “Boosting 19F MRI-SNR efficient detection of paramagnetic contrast agents using ultrafast sequences,” Magnetic Resonance in Medicine, vol. 69, no. 4, pp. 1056–1062, 2012.
- S. D. Fontaine, R. Reid, L. Robinson, G. W. Ashley, and D. V. Santi, “Long-term stabilization of maleimide-thiol conjugates,” Bioconjugate Chemistry, vol. 26, no. 1, pp. 145–152, 2014.
- K. H. Chalmers, M. Botta, and D. Parker, “Strategies to enhance signal intensity with paramagnetic fluorine-labelled lanthanide complexes as probes for 19F magnetic resonance,” Dalton Transactions, vol. 40, no. 4, pp. 904–913, 2011.
- C. T. Goudar, S. K. Harris, M. J. McInerney, and J. M. Suflita, “Progress curve analysis for enzyme and microbial kinetic reactions using explicit solutions based on the Lambert W function,” Journal of Microbiological Methods, vol. 59, no. 3, pp. 317–326, 2004.
- M. D. Robson, P. D. Gatehouse, M. Bydder, and G. M. Bydder, “Magnetic resonance: an introduction to ultrashort TE (UTE) imaging,” Journal of Computer Assisted Tomography, vol. 27, no. 6, pp. 825–846, 2003.
- D. J. Tyler, M. D. Robson, R. M. Henkelman, I. R. Young, and G. M. Bydder, “Magnetic resonance imaging with ultrashort TE (UTE) PULSE sequences: technical considerations,” Journal of Magnetic Resonance Imaging, vol. 25, no. 2, pp. 279–289, 2007.
- W. Yu, Y. Yang, S. Bo et al., “Design and synthesis of fluorinated dendrimers for sensitive 19F MRI,” The Journal of Organic Chemistry, vol. 80, no. 9, pp. 4443–4449, 2015.
- K. L. Peterson, K. Srivastava, and V. C. Pierre, “Fluorinated paramagnetic complexes: sensitive and responsive probes for magnetic resonance spectroscopy and imaging,” Frontiers in Chemistry, vol. 6, 2018.
- P. Harvey, I. Kuprov, and D. Parker, “Lanthanide complexes as paramagnetic probes for 19F magnetic resonance,” European Journal of Inorganic Chemistry, vol. 2012, no. 12, pp. 2015–2022, 2011.
- D. M. L. Lilburn, G. E. Pavlovskaya, and T. Meersmann, “Perspectives of hyperpolarized noble gas MRI beyond 3He,” Journal of Magnetic Resonance, vol. 229, pp. 173–186, 2013.
- N. J. Rogers, F. Hill-Casey, K. F. Stupic et al., “Molecular hydrogen and catalytic combustion in the production of hyperpolarized 83Kr and 129Xe MRI contrast agents,” Proceedings of the National Academy of Sciences, vol. 113, no. 12, pp. 3164–3168, 2016.
- F. Zamberlan, C. Lesbats, N. J. Rogers et al., “Molecular sensing with hyperpolarized 129Xe using switchable chemical exchange relaxation transfer,” ChemPhysChem, vol. 16, pp. 2294–2298, 2015.
Copyright © 2019 Henryk M. Faas et al. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.