Phenols as Diamagnetic T₂-Exchange Magnetic Resonance Imaging Contrast Agents

Abstract

While T₂-exchange (T_2ex) NMR phenomena have been known for decades, only recently there has been a resurgence of interest to develop T_2ex MRI contrast agents. One indispensable advantage of T_2ex MR agents is the possibility of using non-toxic and/or bio-important diamagnetic compounds with intermediate exchangeable protons. In this study, we screened a library of phenol-based compounds and determined their T_2ex contrast (exchange relaxivity, r_2ex) at 9.4 T. Our results showed that the T_2ex contrast of phenol protons allows them to be detected directly by MRI at a mM concentration level. We also studied the effect of chemical modification of the phenol on the T_2ex MRI contrast through modulation of exchange rate and chemical shift. This study provides a guideline for using endogenous and exogenous phenols for T_2ex MRI contrast. As a proof-of-principle application, we demonstrated phenol T_2ex contrast can be used to detect enzyme activity in a tyrosinase-catalyzed catechol oxidation reaction.

Graphical Abstract

Phenol-based T₂-exchange magnetic resonance imaging contrast agents are screened. With the inherent T₂-exchange effect, catechol is utilized as the substrate for the detection of tyrosinase activity in a label-free manner.

The utility of magnetic resonance imaging (MRI) contrast agents is indispensable for both clinical diagnoses and research. Among the current available agents that have been developed in the last four decades, relaxivity agents, i.e., T₁ and T₂/T₂* MRI contrast agents, are the two most widely used types. Most of these agents are metal-based, either lanthanide or transition metal, which raises more concerns recently on their potential toxicity or disturbance of these metal ions and imposes the extra difficulty for the clinical translation of new agents. Recently, a renewed interest has been drawn to a category of compounds that possess labile protons that can efficiently exchange with surrounding water molecules and result in an observable change in water T₂ relaxation time.^[1] This type of MRI contrast agents, namely T₂-exchange (T_2ex) agents, includes both lanthanide-based agents^[2] and non-metallic diamagnetic compounds such as iopamidol^[3], glycogen^[4], and glucose^[5] to generate MRI contrast, which paves a new avenue for directly using readily translatable diamagnetic agents for biomedical imaging. This can be considered as a natural extension of diamagnetic chemical exchange saturation transfer (CEST) contrast.^[6] The relationship between chemical properties (chemical shift difference from water Δω and proton exchange rate k_ex) and T_2ex contrast has been described using the Swift-Connick equation (Supporting information, Eq. S4–S5).^[7] We simulated the equation in Figures 1A and S1, which describes the dependence of the relaxivity r_2ex of an agent on its offset frequency from the exchanging partner (water protons here) Δω and the exchange rate k_ex. As shown, at a given field strength, both Δω and k_ex can strongly affect relaxivity and thus the T_2ex based contrast. A high Δω is always favorable for generating strong T_2ex contrast, while there is an optimal k_ex at each Δω (Figure 1B). According to Figure 1C, for Δω values ranging from 1 ppm to 12 ppm, the optimal k_ex ranges from 2.5 kHz (for 1 ppm) to 30.2 kHz (for 12 ppm) at 9.4 T, which is faster than the favorable exchange rate to generate CEST contrast. Compared with the much larger shifts in paramagnetic shift agents,^[2c] the limited Δω creates a sensitivity barrier for most diamagnetic agents (mostly <6 ppm). For example, as shown in Figure 1B, the theoretical maximum r_2ex values at 1.5, 4.6, 9.3 and 12 ppm are 0.017 s⁻¹ mM⁻¹ (k_ex = 3.8 kHz), 0.053 s⁻¹ mM⁻¹ (k_ex = 11.6 kHz), 0.106 s⁻¹ mM⁻¹ (k_ex = 23.4 kHz), and 0.137 s⁻¹ mM⁻¹ (k_ex = 30.2 kHz), respectively.

Figure 1.

Simulations with the Swift-Connick equation showing the dependence of T_2ex contrast on exchange rates and chemical shifts. A) Simulated r_2ex as a function of labile proton Δω and k_ex at B₀ = 9.4 T; B) Simulated r_2ex dependence on k_ex for Δω values of 1.5, 4.6, 9.3 and 12 ppm respectively; C) Optimal k_ex for the Δω range from 1 ppm to 12 ppm.

Phenols represent a class of diamagnetic agents potentially well suited for T_2ex contrast. The aromatic ring and lower pK_a for the phenolic proton, compared with alcohols, enable a higher Δω and k_ex.^[8] With judicious modification of chemical structures, Δω values of up to 12 ppm have been achieved.^[9] In addition, phenol protons are present in a variety of biologically important compounds, such as amino acids, neurotransmitters (e.g. dopamine and serotonin), and commonly used drugs (e.g. acetaminophen). The establishment of controllable exchange for the optimal T_2ex effect could potentially help to balance the sensitivity and specificity, promoting the development of clinically useful diamagnetic phenols as T_2ex contrast agents.

With phenol (1) as the model compound, we first characterized its T_2ex contrast ability at 9.4 T, a field strength commonly used for pre-clinical studies. Phenol has an exchangeable hydroxyl proton 4.6 ppm shifted from water (or 9.3 ppm in the NMR spectrum versus TMS, Figure S2), with a pK_a of 10. As shown in Figures 2A–2B, phenol clearly exhibits concentration-dependent and pH-dependent T_2ex effects, with an observed transverse exchange relativity (r_2ex) of, for example, 0.023 s⁻¹ mM⁻¹ at pH 7.4 and 37 °C. It should be noted that phosphate can catalyze proton exchanges,^[10] which is negligible when exchange is fast (i.e., pH> 7.0) but become significant at lower pHs where exchange is slow (Figure S3). Therefore, we performed the pH study using water but not PBS solutions. Figure 2C shows that the maximum contrast was obtained at pH 6.5. The k_ex values of phenol were then estimated by fitting the measured r_2ex values at three temperatures (20, 30, and 37 °C) to the Swift-Connick equation (Figure S4, Eqs. S4&S5), under the assumption that k_ex increases at higher temperature.^[7] As shown in Figure 2D, the k_ex of phenol was estimated to be 23.4 and 44.7 kHz for pH 7.0 and pH 7.4 respectively, which falls in the fast regime of r_2ex-k_ex curve, and 1.6 and 8.0 kHz for pH 6.0 and pH 6.5 respectively, which falls in the slow regime of r_2ex-k_ex plot. This phenomenon can be explained by base-catalyzed proton exchange characteristic of exchangeable protons.^[10] At pH 6.5, the k_ex is most close to the Δω difference (i.e., 2π ×4.6 (ppm)× 400 (Hz/ppm)=11.6 × 10³ rad s⁻¹) to generate the maximal T_2ex effect. Both fast and slow rates of exchange will reduce the magnitude of T_2ex effect. Interestingly, when k_ex decreased with dropping pH, CEST signals from the phenolic proton appeared (Figure 2E). This process is in line with the condition of k_ex ≪ Δω, which is a prerequisite for observing CEST effect. At pH 6.0, there was a strong CEST signal at 4.6 ppm (Figure 2E), consistent with the chemical shift value (i.e., 9.3 ppm, Supporting information S2) measured using NMR spectroscopy (compared to the water signal at 4.7 ppm).

Figure 2.

T_2ex contrast of phenol (1). A) T₂-weighted MR images at echo times (TE) of 20 and 2560 ms as well as the corresponding T₂ maps as a function of concentration; B) The concentration dependence of transverse relaxation rates (R₂) of phenol in water at the concentration ranging from 2 to 20 mM at 37 °C at four different pHs; C) The measured r_2ex of phenol at different pHs; D) The estimated k_ex of phenols at different pHs and their positions on the r_2ex-k_ex theoretical curve obtained by using the Swift-Connick equation. All data points are shown as the average of three samples (see Supporting Information for experimental details); E) CEST MRI signals of phenol at different pHs as quantified by MTR_asym plots from 0 to 6 ppm.

The relaxivity results of studying a series of phenol analogues in PBS at pH 7.4 and 37 °C are listed in Scheme 1. Electron donating substitution (NH₂, OMe, Me) para- to the phenol OH (2–4) slightly reduced Δω and more significantly reduced the k_ex towards the optimum, resulting in an overall increase of r_2ex. In sharp contrast, electron withdrawing substitution (Cl, COOH) para- to the phenol OH (6 and 7) dramatically increased the k_ex against the optimum value and making these agents unsuitable for generating T_2ex contrast. Both electron donating and withdrawing modification (OH and COOH) at meta-position were tolerated with only slightly changes in overall r_2ex (9 and 10).

Scheme 1.

r_2ex of phenols tested. Experimental conditions: pH = 7.4, 37 °C and B₀ = 9.4 T. r_2ex value were derived from the concentration-dependent R₂ fitting line, and Δω was obtained from either NMR spectra or CEST spectra. See Figure S5–S7 (Supporting Information) for details. ^a) r_2ex data for compounds with multiple OHs were normalized for the contribution of single OH. ^b) Due to interaction between the two ortho- OHs, k_ex calculated via treating them independently may have significant error. ^c) k_ex were not estimated due to the potential minor contribution from other exchangeable protons.

Intramolecular hydrogen bonded phenols have been reported to achieve higher Δω (8.6–12.0 ppm) and applied for CEST MR imaging.^[11] For example, salicylic acid (11) can achieve a Δω of 9.3 ppm. However, its exchange rate is too slow (k_ex ~ 0.4 kHz^[11]) to generate T_2ex contrast (r_2ex = 0.003 s⁻¹ mM⁻¹), despite its favorable chemical shift. Attempts to increase the k_ex by the introduction of ortho-/para- electron withdrawn groups proved to be successful in 12–15. Higher T_2ex contrast was found for these, for example a r_2ex of 0.059 s⁻¹ mM⁻¹ and k_ex of 7.1 kHz for 12 at Δω of 9.0 ppm, and a r_2ex of 0.046 s⁻¹ mM⁻¹ and k_ex of 5.2 kHz for 15 at Δω of 12.0 ppm. In 10 mM PBS at pH 4.5, the r_2ex for 15 was 0.11 s⁻¹ mM⁻¹, close to its theoretical maximum (0.137 s⁻¹ mM⁻¹). This experimental largest r_2ex is comparable in magnitude to that of glucose at 9.4 T,^[5] suggesting the potential opportunity to be used for relevant biomedical studies.

To further demonstrate potential applications, we applied T_2ex MRI to monitor the bioactivity of tyrosinase (TYR) in a label-free manner. Tyrosinase is an enzyme catalyzing the hydroxylation of phenolic substrates to catechol derivatives, further oxidized to ortho-quinone products (Figure 3A).^[12] These reactions have been recognized as key processes in the biosynthetic pathway of some natural pigments, making TYR the targeting enzyme for treating hypopigmentation-related problems.^[13] Recent studies also showed that the accumulated TYR can be considered as an important biomarker of melanoma cancer,^[14] and TYR imbalance is related to Parkinson’s disease due to its effect on dopamine neurotoxicity.^[15] Therefore, the non-invasive detection of TYR activity is pivotal for diagnosis and treatment monitoring. Our results show that the natural substrate, catechol, has a r_2ex of 0.077 s⁻¹ mM⁻¹ at pH 6.5 and 25 °C (Figure S8A), while the product, ortho-quinone, has no T_2ex contrast-enhancing ability due to the lack of exchangeable hydroxyl protons, which allows the direct detection of TYR activity by observing the changes in the T₂ relaxation time of the system. To demonstrate that, we incubated 10 mM catechol (pH 6.5, 25 °C) with TYR at different concentrations (U, units/mL). As shown in Figure 3B, the presence of TYR significantly decreased the R₂ rates of the samples. Because TYR itself did not produce noticeable T_2ex effects (Figure S8B), the differences in T_2ex between the samples incubated with different concentrations of TYR were attributed the conversion of catechol to ortho-quinone. The change in T_2ex thus allows the quantitative detection of enzyme activity without the need for additional agents. For example, as shown in Figure 3C, we calculated the conversion ratio based on the T_2ex of the samples. This is just a first example of using the T_2ex contrast not only to detect the presence of phenol derivatives but also to monitor their changes and reactions.

Figure 3.

The Detection of tyrosinase enzyme activity using T_2ex MRI. A) Schematic illustration of the conversion of catechol (high T_2ex contrast or hypointense MRI signal on the T2w image shown below) to benzoquinone (no T_2ex contrast or hyperintense MRI signal on the T2w image shown below) by the catalysis of tyrosinase. B) Pseudo-colored R₂ maps of 10 mM catechol solutions (in PBS, 10 mM, pH 6.5), containing different concentration of tyrosinase after reaction for 1.5 h at 25 °C. C) Correlation of transverse relaxation rate and conversion ratio with concentration of tyrosinase. Note: U is the unit of tyrosinase concentration and stands for unit/mL.

While the T₂ contrast-enhancement ability of diamagnetic agents is weaker than those of paramagnetic metallic agents, this new approach has several important advantages for in vivo applications. Firstly, similar to CEST MRI, T_2ex contrast can be used to detect enzyme activity^[16] in a label-free way. Evidenced by the TYR study, the changes in many physiologically or pathologically important molecules (in response to treatments) can be observed directly without disturbing the system. Indeed, we showed that two naturally abundant compounds (tyrosine 16, r_2ex = 0.10 s⁻¹mM⁻¹ and serotonin 18, r_2ex = 0.08 s⁻¹mM⁻¹) can be used as T_2ex agents. The concentrations generating 1% MRI contrast, which is on the same order of magnitude as functional MRI signal changes, are 2 and 2.4 mM, respectively, in short T₂ tissues (e.g. muscle, T₂~50 ms^[17]), or 1 and 1.2 mM respectively, in long T₂ tissues (e.g. grey matter, T₂~100 ms^[17]), implying the potential for in vivo applications. It should be noted that the detection limit in a MRI study is defined by the contrast-to-noise ratio (CNR) and a threshold of CNR of 2√2 is typically used to achieve a 95% probability that the contrast before and after the injection of an agent is different^[18]. As CNR is defined by ΔS% times signal-to-noise ratio (SNR), the minimally required SNR for reliably detecting 1% signal change is 282, suggesting a high SNR (i.e., ~300) should be used to reliably detect a 1% signal change. On the other hand, previous studies showed that the local concentration of exogenously injected agents can reach >4 mM in the tumor^[19] and >5 mM in the blood^[20] in animal studies, and > 69 mM in the kidney in human studies^[21], indicative of the possibility of achieving sufficient local concentrations for generating T_2ex contrast. Finally, while pH may be a confounding factor for quantifying the concentration in the targeted tissue, its effect on quantification can be minimized by using the dynamic contrast enhanced imaging scheme, an approach widely used in CEST MRI studies for quantitatively measuring tissue uptake of pH sensitive CEST agents.^[22] Nevertheless, this approach is particularly useful for monitoring the ex vivo drugs or agents, for example, in tumors (typically have longer T₂ values), without any additional imaging labeling, in a theranostic manner.^{[22a, 23]}

In summary, we screened a library of phenol-based compounds and determined their T_2ex effects. The T_2ex contrast was delicately modulated by means of chemical modification, which affected exchange rate and chemical shift. Current results show that the T_2ex effect of phenols can be optimized for MR imaging directly at a mM concentration level. Understanding the T_2ex mechanism of phenolic protons shall contribute to the interpretation of endogenous T₂ signals and the development of new exogenous T_2ex agents. As the first proof-of-concept, we applied this contrast mechanism to the detection of enzymatic activity of tyrosinase by directly utilizing the changes in T_2ex relaxation times when the natural substrate is converted to the product. These phenols are also expected to affect the parameter T_1ρ, the longitudinal relaxation time in the rotating frame, which is also exchange dependent.^[24]

Experimental Section

Aqueous solutions of each chemical were freshly prepared prior to each MRI measurements either in water or phosphate buffered saline (PBS). When not otherwise noted, the sample phantoms were warmed to 37 °C. All MRI measurements were performed using a Bruker 9.4 T vertical MR scanner with a 20-mm birdcage transmit/receive coil. T₂ relaxation times were acquired using a Carr-Purcell-Meiboom-Gill (CPMG) method as previously described^[5]. Briefly, a T₂ preparation module, with 16 echo times ranging from 20 ms to 10.24 sec, was added in the front of a Rapid Acquisition with Relaxation Enhancement (RARE) pulse sequence. The imaging parameters were: TR/TE = 25000/4.3 ms, RARE factor = 16, a 64x64 acquisition matrix with a spatial resolution of c.a. 250x250 μm², and slice thickness of 1 mm. The acquisition time for each T₂-weighted image was 1 min 40 s.