Triazoles as T₂-Exchange Magnetic Resonance Imaging Contrast Agents for the Detection of Nitrilase Activity

Abstract

We characterized the T₂-exchange (T_2ex) magnetic resonance imaging (MRI) contrast of azole protons that have large chemical shifts from the water proton resonance as a function of pH, temperature, and chemical modification. Our results showed that 1,2,4-triazoles could be tuned into excellent diamagnetic T_2ex contrast agents, with an optimal exchange-based relaxivity r_2ex of 0.10 s⁻¹ mm⁻¹ at physiological pH and B₀ = 9.4 T. A fit of r_2ex data to the Swift–Connick equation indicated that imino proton exchange of triazoles is dominated by a base-catalyzed process at higher pH values and an acid-catalyzed process at lower pH. The magnitude of r_2ex was also found to be heavily dependent on chemical modifications, that is, enhanced by electron-donating groups, such as amines and methyls, or by intramolecular hydrogen bonding between the imino proton and the carboxyl, and weakened by electron-withdrawing groups like bromo, cyano, and nitro. In light of these findings, we applied T_2ex MRI to assess the activity of nitrilase, an enzyme catalyzing the hydrolysis of 1,2,4-triazole-3-carbonitrile to 1,2,4-triazole-3-carboxylic acid, resulting in the enhancement of R_2ex. Our findings suggest that 1,2,4-triazoles have potential to provide sensitive and tunable diagnostic probes for MRI.

Introduction

¹H magnetic resonance imaging (MRI) is one of the most versatile medical imaging modalities. The sensitivity and specificity of MRI can be boosted dramatically by the use of contrast agents. Currently, T₁ and T₂/T₂* contrast agents are most widely used, for example, gadolinium chelates and iron oxide nanoparticles, respectively, modulating the water proton longitudinal relaxation times and transverse relaxation times to generate detectable contrast.^[1] Targeted molecular imaging can be achieved by combining these agents with specific moieties or functional groups.^[2] Recently, chemical exchange saturation transfer (CEST)^[3] and chemical exchange-sensitive spin-lock (CESL)^[4] have been developed as new MRI approaches to visualize low-concentration diamagnetic molecules containing exchangeable protons. Numerous diamagnetic agents have been reported for CEST and CESL MRI detection in a broad spectrum of biomedical applications.^[5] An invaluable merit of CEST/CESL is the possibility to noninvasively visualize bio-relevant compounds (diamagnetic agents), including clinically accessible ones, thus allowing their use as MRI contrast agents. This feature has the potential to greatly simplify the process for development of new imaging agents and their clinical translation. It is noted that CEST/CESL MRI requires the exchange rate (k_ex) of exchangeable protons to fall in the slow-to-intermediate range, that is, k_ex ≤ Δω where Δω is the chemical shift difference (in radians s⁻¹) between the resonance frequencies of labile protons and water protons.^[6] This condition limits the applications of CEST MRI in a broad spectrum of biological molecules whose exchangeable protons fall in the fast regime at physiological pH.

To improve the detectability of exchangeable protons whose k_ex is in intermediate-to-fast range, T₂-exchange (T_2ex) MRI has been developed. It has been known for a long time that the proton exchange of agents with bulk water can reduce water transverse (T₂) relaxation times without affecting the T₁.^[7] The fast exchanging protons hence can be utilized to modulate water T₂ relaxation times to generate T₂ contrast,^[8] extending MRI detection of labile protons from the slow-to-intermediate range (CEST/CESL) to the intermediate-to-fast range. At a given B₀, the T_2ex effect (quantified by the transverse relaxivity, r_2ex) of 1 mm of exchangeable protons can be described by the Swift–Connick equation^[9] [Eq. (1) and Figure S1 in the Supporting Information].

$$r_{2\text{ex}}=9.08\times {10}^{-6}\frac{k_{\text{ex}}\mathrm{\Delta }{\omega }^2}{{k_{\text{ex}}}^2+\mathrm{\Delta }{\omega }^2}$$ (1)

Consequently, an optimal r_2ex is obtained when k_ex ≈ Δω. To date, the T_2ex principle has been successfully extended to a few diamagnetic compounds, including glycogen,^[10] iopamidol,^[11] glucose,^[12] and phenols.^[13] The feasibility of the in vivo applications of T_2ex agents has been demonstrated by a recently published T₂-weighted dynamic contrast-enhanced MRI (DCE-MRI) study in which d-maltose was used as the contrast agent to characterize the permeability of tumors.^[14] However, the limited Δω (typically < 6 ppm) creates a sensitivity barrier for most diamagnetic agents (r_2ex < 0.1 s⁻¹ mm⁻¹). According to Equation (1), the magnitude of the maximum r_2ex increases linearly with Δω, indicating one way to enhance the T_2ex effect is to use agents whose exchangeable protons have chemical shifts far from water.

Azoles are nitrogen containing heterocyclic compounds, widely used in the development of medicinal drugs such as antifungal, anticancer, antiviral agents, and so on.^[15] Their N–H protons can exhibit a large chemical shift (up to 15.3 ppm, or about 10.6 ppm from water). Additionally, azoles are often incorporated into polymers towards the fabrication of proton-exchange membranes because of their fast exchange property.^[16] We thereby hypothesized that azoles could be tuned as a new class of diamagnetic T_2ex agents.

Results and Discussion

Selection of 1,2,4-triazole over other azoles for the T_2ex characterization

We first chose 1,2,4-triazole (1), 1,2,3-triazole (2), and imidazole (3) as model compounds to study T_2ex contrast. As evident from the NMR spectra (Figure S2 in the Supporting Information), their exchangeable imino proton, respectively resonates at 9.2, 10.3, and 7.3 ppm from the water proton frequency (at 4.7 ppm). It turned out that 1 is the best azole scaffold, showing the largest T_2ex effect (Figure S3), and thus we took it for more detailed characterization. As shown in Figures 1 A–B, 1 exhibited a concentration-dependent T₂ effect, with a r_2ex estimated to be 0.045 s⁻¹ mm⁻¹ in PBS at pH 7.4 and at 37 °C. The r_2ex is strongly dependent on pH above pH 7.0 (Figure 1 C). Interestingly, the pH-dependence exhibited two regional maxima, which we speculated to be due to the involvement of two catalytic mechanisms, that is, base-catalysis when pH > 6.5 with the regional maximum r_2ex occurring at pH 7.0, and acid-catalysis when pH< 6.5 with the regional maximum r_2ex occurring at pH 6.3. In order to determine the k_ex values, we measured r_2ex values at three temperatures (i.e., 20, 30, and 37 °C) and fitted them to Equation (1) at each pH (Figures 1D and E), under the assumption that k_ex increases with temperature (Figure S4).^[12] The two-phase pH dependence of k_ex of 1 is markedly different from other reported protons such as phenol.^[13] At pH 7.0, the k_ex was estimated to be 27.3 ± 4.0 kHz, which is close to the Δω of 1 [i.e., 2π × 9.2 (ppm) × 400 (Hz ppm⁻¹) = 23.1 × 10³ rad s⁻¹], enabling the highest enhancement (r_2ex = 0.10 s⁻¹ mm⁻¹). It should be noted that our assumption that the temperature dependence of the exchangeable protons follows the Arrhenius equation is valid only for one-step exchange model and caution should be taken when the exchange model involves two steps such as those of salicylic acid.^[17]

Figure 1.

Characterization of the T_2ex contrast of 1,2,4-triazole (1) in PBS solution. (A) T₂-weighted MR images at echo times (TE) of 20 and 2560 ms and the corresponding R₂ map at different concentrations at pH 7.4. (B) The dependence of transverse relaxation rates (R₂) on concentration at pH 7.4. (C) The exchange relaxivity (r_2ex) as a function of pH. (D, E) The estimated k_ex at different pH values and their positions on the theoretical r_2ex–k_ex curve calculated using the Swift–Connick equation, assuming that k_ex increases with temperature. Conditions: 37 °C, B₀ = 9.4 T, and data points are based on three measurements (experimental details in Supporting Information).

T_2ex effects of 1,2,4-triazole derivatives

A library of 1,2,4-triazole analogues was then tested and their measured r_2ex, Δω, and k_ex values are listed in Scheme 1. When electron-donating groups such as amine- (4) and methyl- (5 and 6) were introduced, the proton exchange slowed down towards the optimum (4 and 5), or behind the optimum (6). However, the substitution of electron-donating groups also reduced the Δω (7.2 ppm for 4, 8.8 ppm for 5, and 8.1 ppm for 6). The overall effect reached maximum for 6 and the r_2ex was doubled at pH 7.4 (0.088 vs. 0.045 s⁻¹ mm⁻¹) with a k_ex of 15 kHz. As expected, the introduction of electron-withdrawing groups, such as bromo (7) and cyano (8) (chloro- and nitro-substituted compounds were also tested, Figure S6) dramatically increased proton exchange rates, leading to strongly weakened T_2ex contrast. Interestingly, our results also showed that the introduction of carboxyl (9) was favorable and a r_2ex of 0.059 s⁻¹ mm⁻¹ was observed. We speculated the considerably decreased k_ex (77 kHz) was a result of the formation of an intramolecular hydrogen bonding (or possibly intermolecular bonding when concentration is high) between the carboxyl and the exchangeable imino proton.^[18]

Scheme 1.

Relaxivities (r_2ex) of azoles and their derivatives. Experimental conditions: pH 7.4, 37 °C and B₀ = 9.4 T. Δω data were obtained from NMR spectra of each compound dissolving in [D₆]DMSO (Figure S5)*, r_2ex values were obtained from the concentration-dependent R₂ fitting line (Figure S6), and k_ex data were estimated by fitting r_2ex values using the Swift–Connick equation, assuming that k_ex increases with temperature. *We also tested the NMR spectra of each compound in H₂O (containing 10 % D₂O by volume). The results did not show the chemical shifts of the imino protons due to the fast exchange with water.

We next characterized the T_2ex effects of 9 at different pH (5.3 to 7.4) and temperatures (20, 30, and 37 °C), and studied the relationship among r_2ex, pH, and k_ex, as shown in Figure 2. At 37 °C (Figure 2 A), the maximum r_2ex (0.092 s⁻¹ mm⁻¹) was observed at pH 6.0, which is slightly smaller than the theoretical maximum (0.106 s⁻¹ mm⁻¹) calculated using the Swift–Connick equation, indicating that the proton exchange is still a bit fast compared to the shift difference with water. The theoretically maximal r_2ex could be attained at pH 6.2 at 20 °C (Figure 2 B). Based on the measurements at different temperatures, we estimated the k_ex values at different pH by fitting the r_2ex data at 20, 30, and 37 °C to the Equation (1) (Figure S7). Figures 2 C, D show the correlation between r_2ex and k_ex. At pH > 6.0, the exchange rate increases with pH, implying the proton exchange is base-catalyzed, and at pH < 6.0, the exchange rate decreases with decreasing pH, suggesting that the proton exchange is dominated by an acid-catalyzed process. This two-phase pH dependence of k_ex is consistent with what we have observed for 1.

Figure 2.

T_2ex contrast of 1,2,4-triazole-3-carboxylic acid (9) in PBS. (A, B) The exchange relaxivity (r_2ex) determined as a function of pH at 37 °C and 20 °C. (C, D) The estimated k_ex values as a function of pH at 37 °C and their positions on the theoretical r_2ex-k_ex curve calculated using the Swift–Connick equation. B₀ = 9.4 T.

The detection of nitrilase activity by T_2ex MRI

T_2ex MRI contrast can be utilized to develop responsive agents for specific detection of biomarkers and enzymes, for example, nitric oxide^[19] and tyrosinase.^[13] Such an MRI detection can be achieved when a biomarker or enzyme acts on the exchangeable protons, either by enhancing (by the so-called “turn-on” mechanism) or attenuating (by the “turn-off” mechanism) the T2ex contrast. In the present study, we designed a “turn-on” mechanism for detecting nitrilase using the low r_2ex compound 1,2,4-triazole-3-carbonitrile (8) as the probe and also the substrate of nitrilase, which is converted to high r_2ex compound 9 by the enzyme (Figure 3 A). Nitrilases (EC 3.5.5.1) are a family of enzymes^[20] that are abundant in microorganisms, especially in bacteria, filamentous fungi, yeasts, and plants, catalyzing the hydrolysis of organonitriles to related carboxylic acids under mild conditions.^[21] Nitrilase-mediated biocatalysis is widely used in industry as a “green method” for the production of a number of commercial compounds, which underlines the critical importance of developing strategies to assess the activity of nitrilase for screening nitrilase-expressing organisms. Conventional assays rely on the determination of the amounts of NH₃ produced equimolarly with acids using solution-based and instrument-based methods.^[20] High-throughput screening strategies have been developed either by the formation of chromophores or fluorophores or by pH indicators.^[22] On the basis of the large difference in T_2ex between the substrate (8) and the product (9), we hypothesized that the activity of nitrilase can be monitored with MRI, and since the R₂ relaxation rate is increased with the conversion, the enzymatic reaction can be temporally imaged via from hyperintense T₂ states to hypointense T₂ states.

Figure 3.

MRI detection of nitrilase activity by using T_2ex contrast difference of triazoles. (A) Schematic illustration of the enzymatic conversion of 8 (low T_2ex contrast or baseline MRI signal on the T_2w image) to 9 (high T_2ex contrast or hypointense MRI signal on the T_2w image) by the catalysis of nitrilase. (B) T_2w image and pseudo-colored R₂ map of 10 mm of 8 (in PBS, 10 mm, pH 7.2), containing nitrilase after reaction for 4h at 37 °C, and being measured at 20 °C. The concentrations of nitrilase, from left to right, are 0, 0.24, 0.48, 0.72, 0.96, 1.2, 1.8, and 2.4 U, respectively. (C) Transverse relaxation rate and conversion ratio as a function of nitrilase concentration. Note: U is the unit of nitrilase concentration per mL. According to the producer, 1 unit corresponds to the amount of nitrilase enzyme that liberates 1 mmol ammonia per minute at pH 7.2 and 25 °C with the conversion of acrylonitrile to acrylic acid. (D) The dependence of reaction velocity on substrate concentrations and the curve fitting using steady state kinetic analysis using the Michaelis–Menten model. Noted that since the enzymatic reaction did not follow a linear change throughout the time (Figure S8B), we herein chose 10 min as the reaction time to obtain the initial reaction velocity by supposing that the conversion is linearly correlated with time.

To demonstrate this, we incubated 10 mm of 8 with nitrilase at a range of concentrations (0, 0.24, 0.48, 0.72, 0.96, 1.2, 1.8, and 2.4 U, U = units mL⁻¹) for 4h at 37 °C in PBS (pH 7.2, 10 mm), followed by T₂ MRI measurement (20 °C, pH 7.2). The reaction time and temperature were chosen to obtain the optimal enzymatic conversion (Figure S8). The T₂ test was performed at 20 °C due to a more appreciable T_2ex effect of 9 (Figure 2 B). At 20 °C and pH 7.2, the r_2ex of 9 was determined to be 0.086 s⁻¹ mm⁻¹ vs. 0.063 s⁻¹ mm⁻¹ at 37 °C (Figure S9). The R₂ map of Figure 3B clearly shows that the presence of nitrilase increases the R₂ relaxation rate, which is proportional to the enzyme concentration (Figure 3 C, round mark), thereby allowing the direct quantification of enzyme activity and simplifying the detection. Because neither the substrate nor nitrilase produced noticeable T_2ex effects (Figure S10), the net differences in R₂ between the samples before and after reaction [ΔR₂ = R₂ (after reaction)–R₂ (before reaction)] can be used to quantify 9, that is, C(9) = ΔR₂/r_2ex(9), in which C(9) is the final concentration of 9 (See details in Supporting Information). As a consequence, the conversion ratio can be calculated for each enzyme concentration (Figure 3 C, square mark), showing that a maximum conversion ratio of 65 % was obtained under the present circumstances. It also allowed us to estimate the kinetic parameters, K_M (7.5 mm), V_max (0.19 mm min⁻¹), k_cat (33 min⁻¹), and k_cat/K_M (4.4 mm⁻¹ min⁻¹), using the Michaelis–Menten kinetics model, as depicted in Figure 3 D. Given that compound 8 has not been used as substrate for nitrilase before, we could not find other kinetics data for comparison. Overall, these data demonstrate that the activity of nitrilase can be assessed quantitatively by the inherent T_2ex MRI contrast of triazoles.

It is worth noting that the T_2ex MRI is similar to CEST in that it exploits the chemical exchange of protons of agents with surrounding water protons. CEST is more suitable for detecting agents whose proton exchange rates fall in the slow to intermediate regime, whereas T_2ex is useful for detecting those with intermediate and fast exchange rates. For diamagnetic agents, the sensitivity of T_2ex agents is on the same order (≈ mm) as CEST agents. In addition, since the R₁ relaxation rate is characteristically not affected by exchange for diamagnetic T_2ex agents, the use of a R₂/R₁ ratio can provide an alternative way of quantification.^[19] As CEST MRI is being demonstrated in more and more biomedical applications, we expect T_2ex MRI to be developed similarly, but as a unique tool complementary to CEST. For example, the T_2ex strategy might be used for in vivo detection of enzymatic activity using non-toxic substrates (and non-toxic products) that can be administered at a relatively high dose to generate high local concentrations in the targeted tissues such as tumors^[5k,23] (typically with long T₂ times) or kidney (high local concentrations can be reached^[12]). It should be noted that, as shown in Figure 2, pH has a strong effect on T2ex contrast, indicating the precise determination of the concentration of T_2ex agents requires a priori knowledge of local pH. On the other hand, similar to CEST agents, the pH-dependence of T_2ex agents may be utilized to measure local pH.

Conclusion

We characterized the T_2ex effects of a series of triazoles by modulating proton exchange properties with respect to pH, temperature, and chemical modification. Exchange rates were studied using the Swift–Connick equation to optimize the T_2ex effect. In addition, as a proof-of-concept study, we applied the T2ex differences between compounds to detect the enzymatic activity of nitrilase. This type of detection for enzyme activities by exploiting the signal difference between the substrate and the product is a common advantage characteristic of CEST MRI and T_2ex MRI. Given its simplicity and compatibility with multiple analysis,^[24] this triazoles-based T_2ex strategy can for instance be used as a high-throughput screening method for discovering nitrilase-expressing organisms. Taken together, the T_2ex MRI is a useful extension of CEST MRI for the detection of diamagnetic compounds with fast exchangeable protons, potentially enabling the label-free detection of many drugs and enzymes.

Materials and Methods

Chemicals

Imidazole, 1,2,3-triazole, 1,2,4-triazole, 3-amine-1,2,4-triazole, 3-methyl-1,2,4-triazole, 3,5-dimethyl-1,2,4-triazole, 1,2,4-triazole-3-carboxylic acid, 3-bromo-1,2,4-triazole, 3-chloro-1,2,4-triazole, 1,2,4-triazole-3-carbonitrile, and 3-nitro-1,2,4-triazole were purchased from Combi-Blocks (San Diego, CA). Nitrilase (recombinant, expressed in E. coli, 4.8 unit mg⁻¹) was purchased from Sigma–Aldrich (St. Louis, MO).

MRI

Aqueous solutions of each chemical were prepared in PBS (1x), with pH adjusted by using 1 m NaOH or HCl solutions. Sample phantoms were prepared by placing the solutions into 1 mm-diameter glass capillaries, and assembled in a customized holder for MR imaging on a Bruker 9.4 T vertical scanner with a 20-mm birdcage transmit/receive coil. We acquired T₂ relaxation times using a Carr-Purcell-Meiboom-Gill (CPMG) method. Briefly, a T₂ preparation module was added in front of a fast spin-echo imaging readout, i.e., Rapid Acquisition with Relaxation Enhancement (RARE) pulse sequence. The T₂ preparation period consisted of a series of single spin-echo elements with t_CPMG = 10 ms together forming a CPMG pulse train. We used loop numbers ranging from 2 to 512, making echo times from 20 ms to 5.12 sec. The imaging parameters were: TR/TE = 25 s/4.3 ms, RARE factor = 16, a 64 × 64 acquisition matrix with a spatial resolution of 0.25 × 0.25 mm², and a slice thickness of 2 mm. The acquisition time for each T₂-weighted image was 1 min 40 s. To obtain the exchange-based relaxivity, r_2ex, of each compound, the R₂ = 1/T₂ water proton relaxation rates of the compound solutions were measured at different concentrations, i.e., 1.25, 2.5, 5 and 10 mm, and fitted to Eq. (2).

$$R_2=R_2^0+r_{2\text{ex}}\times \left[C\right]$$ (2)

Where R₂⁰ is the inherent relaxation rate of the water protons, R₂ the relaxation rate of the solutions, and [C] the concentration of the agent.

NMR

¹H-NMR spectra of the compounds dissolved in [D₆]DMSO were collected on a Bruker Avance III 500 MHz NMR spectrometer equipped with an autosampler. The chemical shifts are reported as δ values (ppm) relative to TMS.

Theory

The relationship between Δω, k_ex, and T_2ex relaxation times can be described by the Swift–Connick equation (Eq. 3),^[9a,b] in which R_2B is the transverse relaxation rate of the exchangeable solute proton and P_B is the mole fraction of exchangeable protons. The Δω in the unit of rad s⁻¹ is the shift difference between the water frequency and the chemical shift of the exchangeable proton, and k_ex is the rate of exchange.

$$R_{2\text{ex}}=k_{2\text{ex}}\times P_{\text{B}}\frac{R_{2\text{B}}{}^2+R_{2\text{B}}{k}_{\text{ex}}+\mathrm{\Delta }{\omega }^2}{{\left(R_{2\text{B}}+k_{\text{ex}}\right)}^2+\mathrm{\Delta }{\omega }^2}$$ (3)

In the case of $\mathrm{\Delta }\text{ω}\gg R_{2\text{B}}{k}_{\text{ex}}\gg R_{2\text{B}}$ and 1 mm of labile proton, this can be simplified to Equation (1).

Analysis of enzyme kinetics

At 37 °C in PBS (pH 7.2, 10 mm), 1.2 U of nitrilase was incubated with various concentrations of 8 ranging from 2 to 40 mm for 10 min. The reaction was halted by refrigerating the samples, after which solution pH was adjusted to 7.2 and T₂ MRI was measured at 20 °C. We assume that the conversion of substrate is linearly correlated with time over this short time period. The correlation of reaction velocity with the concentration of substrate can be fitted by the Michaelis–Menten model, based on which the Michaelis constant K_M and the maximum rate V_max are obtained. The k_cat was calculated from the V_max/[enzyme], where [enzyme] was measured to be 5.8 µm.