¹⁹F Magnetic Resonance Imaging Signals from Peptide Amphiphile Nanostructures Are Strongly Affected by Their Shape

Abstract

Magnetic resonance imaging (MRI) is a noninvasive imaging modality that provides excellent spatial and temporal resolution. The most commonly used MR probes face significant challenges originating from the endogenous ¹H background signal of water. In contrast, fluorine MRI (¹⁹F MRI) allows quantitative probe imaging with zero background signal. Probes with high fluorine content are required for high sensitivity, suggesting nanoscale supramolecular assemblies containing ¹⁹F probes offer a potentially useful strategy for optimum imaging as a result of improved payload. We report here on supramolecular nanostructures formed by fluorinated peptide amphiphiles containing either glutamic acid or lysine residues in their sequence. We identified molecules that form aggregates in water which transition from cylindrical to ribbon-like shape as pH increased from 4.5 to 8.0. Interestingly, we found that ribbon-like nanostructures had reduced magnetic resonance signal, whereas their cylindrical counterparts exhibited strong signals. We attribute this drastic difference to the greater mobility of fluorinated tails in the hydrophobic compartment of cylindrical nanostructures compared to lower mobility in ribbon-like assemblies. This discovery identifies a strategy to design supramolecular, self-assembling contrast agents for ¹⁹F MRI that can spatially map physiologically relevant changes in pH using changes in morphology.

Graphical abstract

Magnetic resonance imaging (MRI) is an invaluable clinical and research tool used to noninvasively produce three-dimensional images of living tissues with excellent spatiotemporal resolution.¹ MR probes capable of detecting specific biological events, including changes in pH,^2,3 ion concentration,^4–6 and the activity of enzymes⁷ have been developed to expand MRI’s capabilities. Responsive MR agents are often based on modulating magnetic effects on the ¹H nuclei of water.^8,9 The challenge of detecting MR contrast agents that modulate the ¹H signal in biological systems where the background water signal is large¹⁰ has led to the development of hyperpolarization techniques^11–16 and improvement in contrast agent design.^17,18 One of the strategies to improve MRI has been to develop probes based on nuclei other than ¹H. ¹⁹F is particularly attractive nucleus because of its large gyromagnetic ratio and natural abundance, resulting in an innate MR signal comparable to ¹H that is not obscured by endogenous proton signal of biological systems.^19–23

Despite improvements in MR sensitivity, ¹⁹F probes must present in millimolar quantities of fluorine atoms to be detectable.²⁴ Recently reported ¹⁹F probes achieve these concentrations by employing nanoscale chemical strategies such as design of emulsions,^20,25 dendrimers,²⁶ and micellar structures.^19,27,28 In addition to higher fluorine payloads, micellar constructs can be designed to respond to chemical stimuli, switching between MR signal “on” and “off” states.²⁹ One specific mechanism is silencing the ¹⁹F MR signal by shortening the T₂ (spin–spin) relaxation time with aggregate formation or interaction with a paramagnetic species.^30–32 Aggregation-based T₂ quenching works by increasing local molecular correlation times, resulting in rapid spin–spin relaxation and signal reduction. We hypothesized that ¹⁹F-containing nanoscale supramolecular assemblies that are more highly structured than spherical micelles could have MR signal response to stimuli based on differences in their internal dynamics.

Peptide amphiphiles (PAs) contain a short peptide sequence covalently grafted to a hydrophobic tail and can be programmed to generate one-dimensional assemblies such as cylinders and ribbons.^33–35 PAs have a modular design that has been functionalized with an array of biomolecular epitopes and bioactive substituents for any number of applications in regenerative medicine,^36–38 drug delivery,^39–42 and as Gd(III)-based MRI contrast agents.^43–45 These nanostructures exhibit structural responses to thermal annealing,⁴⁶ light,⁴⁷ pH,^35,48,49 and enzymatic activity.³⁹ These structural responses to external stimuli introduce functions in PA biomaterials.

An important biomedical target is the development of systems that sense pH as an indicator of hypoxia in ischemic or cancerous tissue.^50,51 In this context, it is possible to vary the number of charged amino acids in peptide amphiphiles in order to gain structural sensitivity to changes in pH.^35,49,52 It follows that differences in local dynamics between different nanostructure geometries could result in a measurable ¹⁹F MRI signal response, but this is not characterized in the literature to our knowledge. The present work aims to explore hydrophobicity and charge to examine ¹⁹F MRI response to pH in peptide amphiphile aggregates. Using this design strategy, we have developed a series of peptide amphiphiles as responsive MR contrast agents. The series varies the hydrophobicity of the perfluoroalkyl tail by amidating either perfluorooctanoic acid or perfluoroheptanoic acid to the amino terminus of the peptide sequence and varies charge in the peptide sequence using either E₂, K₂, or K₃ sequences.

RESULTS AND DISCUSSION

The fluorinated peptide amphiphiles (F-PAs) used in this study were designed to probe the effect of varying fluorophilic interactions and the repulsive effects of charge on NMR spectra and consequently on MRI signals. We expected that acidic and basic headgroups would exhibit NMR signals that varied as a function of pH. We designed a series of F-PAs with either glutamic acid or lysine headgroups to explore the responsiveness of the acidic and basic side chains to pH (Figure 1). β-sheet-promoting sequences like V₂A₂ are known to direct assembly into one-dimensional nanostructures.^53,54 Short, C₇ and C₈ perfluoro chains were selected to balance hydrophobic collapse with solubility in organic solvents needed for synthesis. All of the peptides were synthesized using conventional solid-phase peptide synthesis techniques (see Supporting Information). Each perfluoroalkyl tail was added via the Schotten–Baumann reaction as the acyl chloride dissolved in chloroform with excess N,N-diisopropylethylamine. The final peptide amphiphiles were obtained in approximately 50% overall yield after HPLC purification.

Figure 1.

Chemical structures of peptide amphiphiles used in this report. The amino acid hydrogen bonding sequence V₂A₂ is kept constant while sequence charge (blue) and tail length (red) are varied. Throughout this report, compounds are referenced by their tail length (C₇ or C₈) and their charged residues (K₂, K₃, or E₂).

In order to optimize amphiphile concentration for NMR detection, we obtained the ¹⁹F NMR spectra of each F-PA at concentrations of 1, 1.5, 2, and 3 mM and the signal-to-noise ratio (SNR) was assessed. MRI is innately less sensitive than NMR spectroscopy, because MRI is usually performed at lower field strengths and with much greater magnet bore size. Therefore, we targeted SNR values greater than 100 when performing NMR to ensure MRI detection of a given sample.

For ¹⁹F MR contrast agents to provide quantitative imaging information, peak intensity must increase predictably with concentration from a few hundred micromolar to several millimolar.^24,32 The NMR spectra (Figure 2) of the agents in Figure 1 indicate clearly that the SNR is sensitive to the length of the F-PA tail segment. For example, C₇E₂ produced sharp NMR peaks at each concentration tested, indicative of high tail mobility. The terminal CF₃ group peak at −80.8 ppm maintained a peak width of 9–10 Hz at over 100 SNR at each concentration tested. For C₈E₂, the CF₃ signal at −80.8 ppm and 103 SNR broadened and shifted to −83.0 ppm and 11 SNR above 1.5 mM (Figure 2), indicating a transition from high to low mobility. This trend of high SNR for C₇ and low SNR for C₈ peptides held for all sequences studied. Comparing positively charged headgroups, C₈K₂ produced high (>100) SNR only at 1 mM, whereas C₇K₂ produced high SNR spectra up to 1.5 mM concentration. C₇K₃ exhibited sufficient signal for imaging (greater than 100 SNR) over the range of concentrations tested, whereas C₈K₃ produced low SNR peaks (SNR < 10) through the concentrations tested. Based on these observations, the C₇ F-PAs are superior ¹⁹F MRI candidates compared to the C₈ F-PAs because the C₈ compounds have low SNR at the range of concentrations tested, while C₇ compounds had high SNR. Moving forward, we considered only C₇ compounds.

Figure 2.

Comparison of NMR spectra for each F-PA as a function of peptide concentration: 1.0 mM (blue), 1.5 mM (green), 2.0 mM (orange) and 3.0 mM (red). C₇E₂ and C₇K₃ maintain sharp peaks at the higher concentrations needed to obtain ¹⁹F MR images. The other conjugates exhibited sharp spectral features only at concentrations lower than 2 mM or always exhibited broad features. Measurements conducted at pH 6.5.

Negatively charged glutamic acid residues were compared with positively charged lysine residues to determine if charge had an influence on the effect of concentration on signal intensity. The C₇E₂ spectral signal was unaffected by increasing peptide concentration through the range tested. C₇K₂ experienced significant peak broadening at concentrations above 1.5 mM, with SNR in the NMR spectra remaining less than 15 at 2 mM. The increased intramolecular repulsion of adding an additional lysine residue in the C₇K₃ sequence led to a nanostructure with an NMR signal that is linear with respect to concentration. On the basis of these concentration studies, C₇E₂ and C₇K₃ showed the most promise as contrast agents, and a concentration of 2 mM was chosen for all future experiments.

To determine pH response profiles of the F-PAs, pH titration ¹⁹F NMR experiments were conducted. Solutions were allowed to equilibrate overnight at room temperature before measurement, and were pH-adjusted between measurements by addition of dilute NaOH and HCl solutions (allowed to equilibrate at least 15 min between measurements; for each pH value a 1D ¹⁹F NMR spectra was acquired). For both C₇E₂ and C₇K₃, peak intensity could be modulated reversibly by pH as measured against the trifluoroethanol (TFE) internal standard (Figure 3A and B). Throughout these titrations, pH values corresponding to higher peptide charged states were expected to cause greater intramolecular repulsion, which in turn would increase intramolecular distance, leading to longer T₂ relaxation times and greater integrated signal.

Figure 3.

¹⁹F NMR pH response of (A) C₇E₂ and (B) C₇K₃. Both C₇E₂ and C₇K₃ line shapes were found to be robust to pH changes during titration. (C) C₇K₃ and C₇E₂ both showed an increase in NMR signal in response to pH starting around 5.5. C₇E₂ signal reached maximum at pH 6.5, whereas C₇K₃ signal increased slowly through the pH values tested. (D) Nile red fluorescence assay of C₇E₂ and C₇K₃. Concentration of F-PA is plotted versus blue shift of the Nile red florescence peak (measured at 658 nm). Dotted line is fluorescence maximum of Nile red in water. Both compounds exhibit detectable blue shifts (and therefore critical aggregation concentrations) at or below 0.1 mM. This implies that, at the 2 mM concentration and pH 7.5, less than 5% of compound is in the monomeric state.

Because both C₇K₃ and C₇E₂ exhibited MR signal that increased as a function of pH, they represent the best candidate compounds as pH-responsive ¹⁹F MRI agents. The transition from NMR-inactive to NMR-active occurred across a narrower pH range for C₇E₂, rising from baseline to 100% entirely in the pH window 5.5–7.0 (Figure 3C). The same transition was more gradual and much smaller for C₇K₃, approximating a linear relationship between pH 5.5 and pH 8.5 and rising from 80% activation to 100% over that range. pH response reinforces C₇E₂ as the best candidate F-PA contrast agents since this compound exhibits signal enhancement in the biologically relevant pH window. Interestingly, both C₇E₂ and C₇K₃ exhibit an increase in signal with pH despite of their opposite charge. This unexpected correlation between pH and ¹⁹F signal led us to hypothesize that local tail environment, perhaps as a function of nanostructure morphology, was contributing to ¹⁹F NMR signal intensity in unexpected ways. pH titration analysis by ¹⁹F NMR for the other investigated compounds at 2.0 mM are shown in Figures S1–S4. Interestingly, C₇K₂, C₈E₂, and C₈K₂ were found to exhibit sharp, low intensity peaks at intermediate pH values (Figure S1, S3, and S4). Although this is not ideal as ¹⁹F NMR agents, because signal intensity is too low to be detected in most imaging experiments, it is interesting that intermediate charge states might produce high tail mobility nanostructures in solution. C₇E₂ was shown to exhibit similar signal variation when dissolved in Tris-buffered saline rather than saline (100 mM NaCl and 30 mM Tris) and pH adjusted (Figure S5A–B). The same phenomenon was not observed when that buffered solution was diluted 1:1 in fetal bovine serum (FBS). We believe that this loss of function is likely due to protein-assembly interactions that disrupt the nanostructures through nonspecific interactions, because the resulting NMR signal was broadened (Figure S5C). Future applications of these compounds in vivo will require optimization of application and processing techniques as well as assessment of the agent in solid tissue, which are beyond the scope of the current work.

In order to prove that the observed signal increase with pH for C₇K₃ and C₇E₂ did not arise from unassembled (i.e., monomeric) F-PA, a Nile red assay was carried out to obtain the critical micelle concentration (CMC) for C₇E₂ and C₇K₃ (Figure 3D). We chose pH 7.5 for the assay because it is relevant to biological media and both compounds show high signal intensity at this value. Both C₇E₂ and C₇K₃ were found to induce a blue shift in the Nile red florescence at a concentration below 0.1 mM, corresponding to a CMC less than or equal to 0.1 mM for both compounds. These data support an NMR signal contribution of less than 5% for monomers in solution, indicating the NMR signal increase observed as a result of greater pH is largely the result of the nanostructures formed by both compounds.

Cryogenic electron microscopy (Cryo-TEM) experiments were carried out to investigate the effect of the amphiphile structure on nanoscale aggregates. Sample preparation for Cryo-TEMs used identical buffer conditions described for NMR measurements but omitted the use of deuterated solvent and internal standard. Results from these experiments are shown in Figures 4 and S5. C₇E₂ formed nanoribbon structures at pH 4 and 7 with widths of 25 ± 6 nm (13 ribbons measured) and 19 ± 5 nm (20 ribbons measured), respectively, and sparse cylindrical nanofibers at pH 9 with diameters of 11 ± 2 nm (16 fibers measured). C₇K₃ formed only very short, sparse ribbons at both pH 4 and 7, and long cylindrical fibers and some ribbons at pH 9 with a diameter of 14 ± 3 nm (16 ribbons measured). These observations across the C₇ series were consistent with relatively weak hydrophobic interactions, resulting in structures that respond strongly to the effect of charge at the peptide terminus and, therefore, pH (Figure 4). Interestingly, the strongest signals at various values of pH for C₇K₃ and C₇E₂ were observed for cylindrical nanofibers (a larger view of C7K3 nanostructures can be found in Figure S7). This suggests a relationship between nanostructure morphology and ¹⁹F MRI signal, with cylindrical nanostructures producing higher integration NMR signal peaks and therefore higher MRI signals.

Figure 4.

Representative Cryo-TEM images of perfluorinated peptide amphiphiles at pH 4, 7, and 9. Scale bar is 100 nm. C₇K₂ exhibits sparse, disordered structures at pH 4 (panel A), and ribbons at pH 7 (B) and 9 (C). C₇E₂ produces ribbon nanostructures at pH 4 (D) and 7 (E), transitioning to cylindrical fibers at pH 9 (F). Finally, C₇K₃ forms sparse ribbon structures at 4 (G) and 7 (H), transitioning to cylindrical fibers at pH 9 (I).

In contrast to the C₇ series, the C₈ series did not exhibit pH sensitivity at the concentration ranges tested, and signal-to-noise for the NMR signal of these compounds proved prohibitively low for MR imaging. All C₈ compounds, differing by only one CF₂ unit from the C₇ series, exhibited flat nanostructures that did not change with pH and were more likely to show nanostructure aggregation (Figure S6). C₈E₂ produced aggregated ribbons at pH 4, consistent with the viscous and turbid solutions. At pH 7, ribbon nanostructures were observed, and sparse nanoribbons at pH 9. For C₈K₂ at pH 4, a mixture of ribbons and fibers was observed and at pH 7 and 9 these transitioned to only ribbon nanostructures. Consistent with results obtained with the C₇ series, C₈K₂ showed broad NMR signals at low pH which lost intensity at high pH. To summarize, Cryo-TEM images indicate that the C₇ series is significantly more sensitive to pH and also that cylindrical nanostructures correlate with enhanced NMR signals. This result indicates that a ribbon to fiber transition in these supramolecular aggregates may be responsible for the pH response observed.

Peptide secondary structure may play a role in restricting tail mobility and reducing the signal observed. To assess this possibility, a series of circular dichroism (CD) experiments were conducted to determine if a correlation exists between NMR signals and peptide secondary structure (Figures 5 and S8–S11). Each of the spectra shows the decay of a random coil signal appearing at 200 nm and the growth of a β-sheet signal as a negative peak at 220 nm as the nanostructure becomes less charged as a result of pH. For C₇E₂, β-sheet character increases with decreasing pH. For C₇K₃, β-sheet character increases with increasing pH. Despite this inversion, pH increase exhibits a positive correlation with ¹⁹F NMR signal in both instances (Figure 4). The pH trends in peptide secondary structure for these samples therefore imply that peptide secondary structure does not directly result in higher or lower ¹⁹F NMR or ¹⁹F MRI signal.

Figure 5.

Circular dichroism spectra of (A) C₇E₂ and (B) C₇K₃ at different values of pH. ranges. β-sheet component increase for C₇K₃ coincides with NMR signal increase, whereas the opposite trend is observed for C₇E₂. This implies that β-sheet signal is not predictive of NMR signal enhancement. Spectra for other compounds investigated can be found in Figures S7–S10.

In order to examine MRI contrast agent properties in these compounds, C₇E₂ and C₇K₃ were imaged at pH 4, 6, 7, 8, and 9 using the same solution conditions as the NMR studies (Figure 6). The NMR signal peak corresponding to the terminal CF₃ group of the perfluorocarbon tail was selected for imaging, due to its relatively high intensity. MRI signal for C₇E₂ was shown to increase until pH 7, as predicted by data shown in Figure 3. C₇K₃ was imaged, but imaging artifacts from the TFA counterions required dialysis prior to imaging. In these samples, measurable signals that showed pH response were observed, but the data did not fit the trend described in Figure 3 (Figure S12). One possible explanation is that the dialysis process changes the nature of the assembly and therefore the pH response profile.

Figure 6.

¹H and ¹⁹F MRI solution images of C₇E₂ in 150 mM NaCl acquired at 9.4T and ambient temperature. Each image is 25 mm on edge. Sample preparation was identical to NMR experiments. Phantoms were imaged simultaneously with a trifluoroethanol standard to allow intensity comparison. Fluorine images increase in intensity as pH increases, reaching a maximum after pH 7. This is analogous with behavior observed in Figure 3.

The morphology of nanostructures formed by C₇E₂ and C₇K₃ molecules is controlled by the balance of repulsive and attractive intermolecular forces sensitive to the acid–base equilibrium of charged amino acids in each peptide sequence. Altering the peptide sequence by varying the nature of charged residues or hydrophobic tail structure results in loss of pH sensitivity. Morphologically, both C₇E₂ and C₇K₃ formed cylindrical nanofiber structures when producing NMR signal while molecules of the other compounds tested aggregated into ribbon morphologies. The implication is that the local mobility of perfluorinated tails is greater in cylindrical nanostructures than in nanoribbons. One hypothesis for this difference in dynamics between the two different types of supramolecular aggregates would be a denser packing of tail segments in the case of ribbon-like nanostructures, which possess less local curvature, and less dense packing in the hydrophobic domains of nanofibers, which possess more local curvature.

CONCLUSIONS

A series of self-assembling fluorinated peptide amphiphiles were designed to optimize ¹⁹F MRI signals in response to pH changes by varying their content of charged amino acids and number of fluorine atoms in hydrophobic segments. We found that the supramolecular aggregates formed by some of the amphiphiles undergo a morphological transition from low-curvature nanoscale ribbons to cylindrical nanofibers as pH increases that results in enhanced MR signals. This work suggests that morphological transitions in aggregates that affect the mobility of fluorinated segments represent a good strategy to design stimulus-responsive supramolecular ¹⁹F MRI contrast agents.

METHODS

Synthesis

Peptides were synthesized using standard solid-phase peptide synthesis techniques on an automated CSBio Synthesizer, model number CS136. Amino acids were obtained from Anaspec (Fremont, CA). DIC and Oxyma pure were obtained from ThermoFischer Sciences (Waltham, MA) and BAChem (Bubendorf, Switzerland), respectively. Dimethylformamide, dichloromethane, diisopropylethylamine, and 4-methylpiperidine were obtained and used without further purification from VWR. Synthesis was conducted on a rink amide resin on MBHA support (EMD Millipore, Billerica Massachusetts) at 0.5 mM scale. The perfluoroalkane tails were added on-resin via the Schotten-Baumann reaction. Acyl chlorides were obtained from Sigma-Aldrich. Acyl chloride (300 μL) was dissolved in 10 mL of solvent system-dried chloroform along with 600 μL of diisopropyl ethyl amine. The solution was directly added to a fritted shaker vessel along with the peptide on-resin and agitated for 30–45 min. The addition was repeated until ninhydrin test was negative. The peptide was cleaved from resin support in a solution of 95% trifluoroacetic acid, 2.5% water, and 2.5% triisopropyl silane over 3 h and precipitated in cold ether stored at −20 °C. The precipitate was purified using reverse-phase HPLC. Overall yield was typically 40%– 50%.

Purification

All peptides were purified using reverse-phase high pressure liquid chromatography on a Phenomenex Gemini C18 column (5 μm particle size) using a 2%–100% acetonitrile gradient program. To each solution was added 0.1% saturated NH₄OH solution in water (for glutamic acid-containing sequences) or 0.1% trifluoroacetic acid (for lysine-containing sequences). System eluent was analyzed with an Agilent 6510 Q-TOF MS to identify product peaks. Acetonitrile was removed with rotary evaporation and water was removed via lyophilization. The resultant white, fluffy powder was stored at −20 °C.

General Conditions

Unless otherwise specified, all F–PA samples were analyzed at 2 mM concentration in 150 mM NaCl in Millipure water. All solutions were allowed to stand at room temperature for a period of not less than 16 h and not more than 24 h prior to NMR, CD, Cryo-EM, MRI, Nile red assay, and analytical HPLC.

Analytical HPLC

All lysine-based sequences were analyzed in acidic mobile phase and all glutamic acid–based sequences were analyzed in basic mobile phase. A Phenomenex Jupiter C18 and Phemomenex Gemini C18 column were used for acidic and basic condition separations, respectively, on an Agilent 6520 LC/MS system. Each column measured 1.0 × 150 mm, and contained a 5 μm particle size. A gradient of 5–95% acetonitrile over 30 min and 1 mL/min flow rate was employed. Mobile phases were doped with either 0.1% NH₄OH solution (if basic) or 0.1% formic acid (if acidic).

Nuclear Magnetic Resonance Spectroscopy

All NMR spectra were acquired using a Varian 500 MHz NMR spectrometer at 25 °C. Spectra represent the average of 128 scans with a 30° flip angle and 1 s delay time. Precision, thin-walled Shigemi Tubes were used for NMR measurements. PA solutions were dissolved in 150 mM NaCl in Millipure water with 1.8 mM trifluoroethanol and 10% D₂O as the lock solvent and aged overnight prior to measurement to promote structure stability. pH adjustment was conducted using a VWR SB21 pH meter between each measurement, using concentrated (0.1 or 0.01 M) solutions of NaOH and HCl in order to minimize concentration change through the pH adjustment process. Each sample at each pH point was equilibrated for 15 min in the scanner at 25 °C prior to spectral acquisition.

Critical Micelle Concentration Determination

Solutions of C₇E₂ and C₇K₃ were prepared as specified in the General Conditions section and adjusted to pH 7.5 using 0.1 M NaOH and HCl solutions. On the day of measurement, these solutions were serial diluted to triplicate sets of 1 mM, 0.5 mM, 0.25 mM, 0.1 mM, 0.04 mM, and 0.008 mM concentrations using a preadjusted 150 mM NaCl solution with a total volume of 1 mL. A solution of Nile red was prepared at 150 μM in ethanol. A micropipette calibrated to the density of ethanol was used to dispense 1 μL of Nile red solution in each sample. These samples were then agitated briefly and allowed to incubate at room temperature for 2 h. The solutions were then analyzed using a Hitachi F-4500 fluorimeter operating at a 550 nm excitation wavelength. Each acquisition was the average of three scans in a 5 × 5 mm quartz cuvette.

Circular Dichroism UV Spectroscopy

General conditions were used as described above. Measurements were performed with a AJ-715 Jasco circular dichroism spectrometer at 25 °C. Slit width was 2 nm and the scan speed was 100 nm/min. Samples were loaded into 0.1 mm path length plates. Each spectra is the average of three scans.

Cryo-TEM

Cryo-TEM samples were prepared according to General Conditions described above. Images were acquired using a JEOL 1230 microscope operating at 100 kV. Specimens were cast on plasma-cleaned 300 mesh copper grids with a lacey carbon support obtained from Electron Microscopy Sciences and frozen using a Vitrobot Mark IV plunging into liquid ethane. Images were acquired using a Gatan 831 CCD camera.

Magnetic Resonance Imaging (MRI)

MR images were acquired on a 9.4T Agilent (Varian) horizontal bore MRI scanner. For ¹H images, a quad-volume coil of 63 mm diameter was used as both excitation and receiving RF device. The ¹H channel was used to shim and to acquire the ¹H images. After this, it was switched to a surface coil of 20 mm diameter for ¹⁹F images. For ¹⁹F MRI, a steady-state free procession (SSFP) sequence was used to selectively excite the desired ¹⁹F signal (ca. −85 ppm).^{29 19}F MRI parameters: the repetition time (TR = 30 ms), the echo time (TE = 3.01 ms), the small flip angle (FA = 50), the number of transient (NT = 4096), the SINC excitation pulse length (P_sinc = 1.5 ms), the matrix size = 32 × 32, which was zero-filling to 64 × 64 before Fourier transform, the field of view (FOV = 30 × 30 mm), and the slice thickness (THK = 11–15 mm), respectively.