Esterase-catalyzed production of hyperpolarized ¹³C-enriched carbon dioxide in tissues for measuring pH

Abstract

¹³C Magnetic resonance imaging of hyperpolarized (HP) ¹³C-enriched bicarbonate (H¹³CO₃⁻) and carbon dioxide (¹³CO₂) is a novel and sensitive technique for tissue pH mapping in vivo. Administration of the HP physiological buffer pair is attractive, but poor polarization and the short T₁ of ¹³C-enriched inorganic bicarbonate salts are major drawbacks for this approach. Here, we report a new class of mixed anhydrides for esterase-catalyzed production of highly polarized ¹³CO₂ and H¹³CO₃⁻ in tissue. A series of precursors with different alkoxy and acyl group were synthesized and tested for chemical stability and T₁. ¹³C-enriched ethyl acetyl carbonate (¹³C-EAC) was found to be the most suitable candidate due to the relatively long T₁ and good chemical stability. Our results showed that ¹³C-EAC can be efficiently and rapidly polarized using BDPA. HP ¹³C-EAC was rapidly hydrolyzed by esterase to ¹³C-enriched monoacetyl carbonate (¹³C-MAC), which then decomposed to HP ¹³CO₂. Equilibrium between the newly produced ¹³CO₂ and H¹³CO₃⁻ was quickly established by carbonic anhydrase, producing a physiological buffer pair with ¹³C NMR signals that can be quantified for pH measurements. Finally, in vivo tissue pH measurements using HP ¹³C-EAC was successfully demonstrated in the liver of healthy rats. These results suggest that HP ¹³C-EAC is a novel imaging probe for in vivo pH measurements.

Graphical Abstract

Carbon dioxide (CO₂), the metabolic product of catabolism in all tissues, is rapidly converted to carbonic acid (H₂CO₃) by the ubiquitous enzyme, carbonic anhydrase.^1,2 Even though carbonic acid is considered a weak acid (pK_a = 6.35), the amount of bicarbonate (HCO₃⁻) present in blood near physiological pH values (~25 mM) contributes significantly to the total buffering capacity of blood. The equilibrium between CO₂ and HCO₃⁻ is altered most dramatically in tissues that produce excess acid. Consequently, the ratio of CO₂ to HCO₃⁻ has been used as an index of tissue pH using hyperpolarized (HP) ¹³C MRI.³ HP ¹³C MRI is a novel imaging method that allows for insensitive nuclei such as ¹³C to be imaged, owing to the significantly improved NMR sensitivity afforded by dynamic nuclear polarization (DNP).⁴ The hyperpolarized spin state is non-equilibrium and returns to thermal equilibrium as a function of spin-lattice relaxation time, T₁. Using DNP, various salts of ¹³C-enriched bicarbonate (H¹³CO₃⁻) have been hyperpolarized and upon dissolution, the resulting HP H¹³CO₃⁻ rapidly equilibrates to form HP ¹³CO₂. This approach is attractive because it involves administration of physiological base bicarbonate. Another major advantage of this approach is the excellent MR imaging sensitivity of HP ¹³C MRI afforded by dynamic nuclear polarization (DNP). This approach is attractive because it involves administration of a physiological base, bicarbonate, with excellent MR imaging sensitivity afforded by DNP.

However, multiple challenges remain before this method can be applied for routine in vivo pH measurements. First, the HP ¹³C MR signal of H¹³CO₃⁻ is relatively short-lived because of fast T₁ relaxation (T₁ = ~10 s). DNP of inorganic bicarbonate salts is also challenging, generally resulting in low polarization levels (~16%).⁵ Gallagher et al achieved an improved ¹³C polarization level using cesium [¹³C]bicarbonate and demonstrated pH imaging of subcutaneous EL4 tumors in mice.⁶ The alkaline metal ions were removed by ion exchange prior to administration due to potential Cs toxicity.⁷ Several studies have evaluated the use of molecular precursors to produce HP ¹³C imaging probes in situ either by chemical or enzymatic decompositions of the precursors. These approaches involve the polarization of chemically stable ¹³C-enriched compounds with a long T₁ and the desired molecular sensing probes are produced after the dissolution of the HP agents.^8–13 So far, two molecular precursors for HP ¹³CO₂/H¹³CO₃⁻ have been evaluated for pH imaging applications. Ghosh et al produced HP ¹³CO₂ by oxidation of HP ¹³C α-keto acids with H₂O₂.⁹ In another study by Korenchan et al., HP H¹³CO₃⁻ and ¹³CO₂ were produced by chemical hydrolyses of HP ¹³C organic carbonate compounds.¹² Although the longer T₁ and higher polarization levels of the precursor molecules may be beneficial, the required decarboxylation delayed the probe injection, again resulting in significant HP ¹³C signal losses. More importantly, these approaches still involve administration of HP H ¹³CO₃⁻ and ¹³CO₂, probes with very short T₁ values, offsetting any advantages the proposed HP ¹³CO₂-precursor molecules offer. HP ¹³CO₂ can also be produced in tissues via the decarboxylation of HP ¹³C-pyruvate by pyruvate dehydrogenase (PDH).^14,15 However, this method requires high flux through PDH which may not occur in vivo even in highly oxidative tissues because of the presence of competing substrates such as fatty acids.¹⁶ Furthermore PDH activity is inherently low in some tissues such as cancers.^17,18 Consequently, administration of HP [1-¹³C]pyruvate is not a reliable method for generating HP ¹³CO₂.

Here, we report the development of mixed anhydride compounds as precursors of HP ¹³CO₂ and H¹³CO₃⁻ for tissue pH measurement. Typical mixed anhydrides consist of an alkoxy (R′O−) group and an acyl (R″CO−) moiety (Scheme 1). We hypothesized that the ester functional group of a mixed anhydride would be readily hydrolyzed by carboxyl esterase (Step 1). The hydrolyzed product monoacyl carbonate would then undergo a unimolecular decomposition to produce CO₂ (Step 2). Since these processes take place in vivo, the newly produced CO₂ will be rapidly equilibrated to HCO₃⁻ by carbonic anhydrase (Step 3) and the resulting HCO₃⁻/CO₂ ratio will reflect the local tissue pH. Based on the design of these imaging probes, ¹³C-enrichment of only the carbonate carbon is required for in situ HP ¹³CO₂ production. Given the high esterase activity in plasma,^19,20 we expect that the mixed anhydride compounds would be readily hydrolyzed in the circulation soon after being administered. Mixed anhydride molecules that survive the hydrolysis by circulating esterase can also be hydrolyzed in esterase-rich organs such as the liver and kidneys.²¹ We also expect that the sequence of chemical reactions following the hydrolysis would be relatively fast and produce enough H ¹³CO₃⁻ and ¹³CO₂ for pH measurements. This approach also allows for administration of relatively stable HP ¹³C compounds with high polarization levels. The production of HP ¹³CO₂ and H¹³CO₃⁻ is also expected to primarily take place only after the injection of the HP ¹³C imaging agent.

Scheme 1.

Proposed mechanism of esterase-catalyzed production of HP ¹³CO₂ and H¹³CO₃⁻ from ¹³C-enriched mixed anhydrides. CA = carbonic anhydrase.

To test our hypotheses, we first synthesized ¹³C-enriched ethyl acetyl carbonate (¹³C-EAC) from ¹³C-enriched ethyl chloroformate and acetic acid in diethyl ether using N-methylmorpholine as a base (Supplementary Information). We found that ¹³C-EAC was relatively stable when stored in an aprotic solvent at low temperature but gradually hydrolyzed in water under ambient conditions. During initial trials to polarize ¹³C-EAC using the trityl radical OX063, it was found that the radical itself quickly hydrolyzed the compound. However, the hydrophobic and relatively inert radical BDPA^22,23 was found to be fully soluble in ¹³C-EAC without causing any decomposition. Moreover, the resulting neat-EAC-BDPA (40 mM) formed a glassy solid when frozen and nicely polarized without a glassing matrix. From a polarization buildup of ¹³CEAC/BDPA (Fig. 1a), it is clear that ¹³C-EAC polarization reached the maximum level in a very short time (~10 min) confirming that BDPA is an excellent radical for ¹³C-EAC polarization. Dissolution of HP ¹³C-EAC was first done with ethanol to avoid potential hydrolysis. Arrayed ¹³C NMR spectra acquired every 5 s of 2 mM ¹³C-EAC in ethanol are shown in Fig. 1b. The T₁ of the ¹³C-enriched carbonate at 9.4 T was 30 s. The liquid state signal enhancement calculated from signal intensities of the first HP spectrum and thermally polarized signal (Fig. S3) was ~30,000-fold, corresponding to ~25% polarization. In addition to EAC, other mixed anhydride compounds with different alkoxy and acyl groups were also synthesized to evaluate the chemical stability and T₁’s of this class of compounds (Table S1, ESI). We found that ¹³C-EAC possesses the most favorable properties among those molecules for in situ production of HP ¹³CO₂. Larger mixed anhydrides are more stable, but the shorter T₁ and lower water solubility are major drawbacks. Despite the longer T₁, methyl acetyl carbonate is less stable and more importantly produces toxic methanol as a byproduct after the esterase hydrolysis. Biocompatible ethanol and acetate are produced as byproducts following the esterase hydrolysis of HP ¹³CEAC, thereby minimizing toxicity concerns.

Fig. 1.

a) A DNP polarization buildup of neat ¹³C-EAC polarized with BDPA (40 mM) in a HyperSense polarizer; b) Arrayed ¹³C NMR spectra of HP ¹³C-EaC (2 mM in ethanol) following ~1 h of polarization acquired every 5 s with a 10-deg flip angle at 400 MHz.

We next investigated the applicability of HP ¹³C-EAC as a precursor for in situ production of HP ¹³CO/H¹³CO₃⁻. The main goals were to (1) evaluate the chemical stability of HP ¹³C-EAC in a physiological buffer solution; (2) assess whether HP ¹³C-EAC is sensitive to esterase activity; and (3) evaluate whether the esterase-hydrolyzed product decomposes within the HP ¹³C signal timeframe to produce HP ¹³CO₂. First, we collected an array of ¹³C NMR spectra of HP ¹³C-EAC dissolved in PBS buffer with each spectrum acquired every 5 s (pH = 7.4, Fig. S4). A summed ¹³C NMR spectrum over the course of the array (lower panel of Fig. 2a) shows some degree of HP ¹³C-EAC hydrolysis in PBS. The hydro-lyzed product, ¹³C-enriched monoacetyl carbonate (¹³C-MAC), appeared at 160.0 ppm. ¹³CO₂ and H¹³CO₃⁻ resonances are also visible at 125.0 ppm and 160.6 ppm, respectively, confirming ¹³CO₂ production from ¹³C-MAC decomposition. The H ¹³CO₃⁻ peak is small because carbonic anhydrase was not present. We then evaluated the esterase-catalyzed hydrolysis of HP ¹³C-EAC in the presence of isolated porcine liver esterase and in rat plasma and liver homogenate solutions, both of which contain carboxyl esterase.^20,24 Arrayed ¹³C NMR spectra of HP ¹³C-EAC in esterase solution, plasma, and liver homogenate (Fig. S4), show the evolution of ¹³C-MAC, ¹³CO₂, and H ¹³CO₃⁻ peaks soon after mixing. Summed ¹³C spectra of HP ¹³C-EAC in these samples are shown in Fig. 2a. The spectra confirm higher production of these hydrolyzed ¹³C-species in the samples that contain carboxyl esterase, as expected. The presence of enzyme esterase resulted in rapid hydrolysis of HP ¹³C-EAC to HP ¹³C-MAC. Plots of signal intensities of ¹³C-MAC, ¹³CO₂, and H ¹³CO₃⁻ normalized to the total HP ¹³C signal over the course of the NMR acquisition are shown in Fig. 2b. The signal-intensity plots confirm that HP ¹³C-MAC was produced instantly in esterase, plasma, and liver homogenate samples following the mixing.

Fig. 2.

a) Representative ¹³C NMR spectra of HP ¹³C-EAC (2 mM) in PBS (pH = 7.4), isolated esterase solution, rat plasma, and rat liver homogenate. Each spectrum is a sum of all spectra acquired over the course of HP ¹³C-signal lifetime (~3 min). A small peak at 156.3 ppm is the non-¹³C-enriched acetyl carbon. b) HP ¹³C NMR signal-time curves of ¹³C-MAC, ¹³CO₂, H¹³CO₃⁻, and ¹³C-EAC (divided by 5) normalized to the total HP ¹³C signal from the same samples.

The decomposition of HP ¹³C-MAC to HP ¹³CO₂ in both biological samples was quite fast as confirmed by the HP ¹³CO₂ signal-time curves. Interestingly, the HP ¹³CO₂ curve of the esterase sample has a distinctively different kinetics compared to those of the two biological samples. In plasma and liver homogenate, the signals reached an apex at ~20 s before decaying due to T₁ relaxation. In the esterase solution, the signal intensity of HP ¹³CO₂ increased much more slowly and peaked at ~50 s. Given that decomposition of ¹³C-MAC is not enzymatically catalyzed, one would expect similar kinetics for the appearance of HP ¹³CO₂ in all three samples. However, the apparent decomposition rates of HP ¹³C-MAC follow the decreasing trend as liver homogenate > plasma > esterase. Much faster decomposition of HP ¹³C-MAC was observed in the liver homogenate sample as evident from the sharp decay of the HP ¹³C-MAC signal curve. These results suggest that the presence of proteins and divalent cations present in the two biological samples may facilitate the hydrolysis of ¹³C-MAC, but the exact mechanism is yet to be delineated. Nevertheless, the rapid decomposition of HP ¹³C-MAC is highly beneficial for the purpose of producing HP ¹³CO₂ for imaging tissue pH. The HP H ¹³CO₃⁻ peaks in Fig 2a and signal-time curves in Fig. 2b demonstrate that much of the newly generated HP ¹³CO₂ in the liver homogenate sample was rapidly converted to HP H¹³CO₃⁻ by carbonic anhydrase. While carbonic anhydrase is not expected in the plasma sample,^25,26 a small amount of the enzyme may be present due to lysed red blood cells during the plasma preparation. Consequently, the H ¹³CO₃⁻ signal in the plasma was much lower. The isolated esterase solution did not contain carbonic anhydrase enzyme. Therefore H ¹³CO₃⁻ was produced solely by chemical hydration of ¹³CO₂ in this sample. Despite the presence of carbonic anhydrase in both plasma and liver homogenate samples, the pH values estimated from H ¹³CO₃⁻¹³CO₂ ratios were much lower than the values measured by a glass electrode of the same solutions. This is likely due to the high concentrations of newly produced ¹³CO₂ in these static systems. The ¹³CO2/H¹³CO₃⁻ ratio was therefore not in an equilibrium within the HP ¹³C signal timeframe.⁵ Our results demonstrate that HP ¹³C-EAC is quite stable with slight hydrolysis in buffer media. The presence of esterase enzyme in plasma and liver homogenate significantly accelerated HP ¹³C-EAC hydrolysis, producing HP ¹³CO₂ that can be readily converted to H¹³CO₃⁻ by carbonic anhydrase for tissue pH measurements.

¹³C chemical shift imaging (CSI) of a phantom containing solutions of HP ¹³C-EAC in PBS and liver homogenate are shown in Fig. 3. In this phantom, two small vials inside a 20-mL beaker filled with water were added a solution of HP ¹³C-EAC in PBS or HP ¹³C-EAC in liver homogenate ([¹³C-EAC] = 15 mM). Fig. 3a–c show HP ¹³C signal intensity maps of ¹³CO₂, H¹³CO₃⁻, and ¹³CMAC acquired ~15 s after mixing overlaid on a ¹H reference image. Higher intensity of ¹³CO₂ is shown in the left vial containing the liver homogenate solution than the right vial of HP ¹³C-EAC in PBS. This confirms the rapid hydrolysis of HP ¹³C-EAC in the presence of esterase enzyme coupled with faster ¹³C-MAC decomposition in the liver homogenate sample, as previously discussed. Intensity maps of ¹³C-MAC and H¹³CO₃⁻ are not distinguishably different in these two vials. However, the ¹³C-EAC maps in Fig. 3d show a noticeably lower signal in the liver homogenate vial due to fast ¹³C-EAC decomposition by esterase hydrolysis. The results show that the production of HP ¹³CO₂ and H¹³CO₃⁻ from esterase-catalyzed hydrolysis of HP ¹³C-EAC can be imaged by chemical shift imaging. The H ¹³CO₃⁻ and ¹³C-MAC peaks can be resolved by peak-fitting both peaks during post-image processing and signal intensity maps of these two hydrolyzed products can be generated. It must be stated that this relatively small chemical shift separation (Δδ = 0.6 ppm) could become a challenge in the in vivo settings because of field inhomogeneities. Nonetheless, this concern could be alleviated with increasing interest in high-field MRI where larger chemical shift separation can be achieved.

Fig. 3.

CSI of a phantom containing HP ¹³C-EAC (15 mM) in PBS (right tube) and liver homogenate (left tube), a-d) Signal intensity images of HP ¹³CO₂, ¹³C-MAC, H¹³CO₃⁻, and ¹³C-EAC, A total HP ¹³C intensity map and a ¹H reference image are shown in e) and f), respectively. g-h) ¹³C spectra of the region-of-interest (ROI) depicted in e). The left and right intensity histograms are for images a-c and images d-e, respectively. CSI parameters: matrix = 8×8, FOV = 45×45 mm2, slice thickness = 10 mm, flip angle = 10 deg, number of points = 512; spectral width = 5 kHz.

Finally, HP ¹³C-EAC was also presented to live animals. In this experiment, rats were administered with a solution of HP ¹³C-EAC in PBS (3 mL, ~80 mM) via a tail vein catheter after a quick removal of BDPA by filtration. Sequential ¹³C NMR spectra collected form the liver region only (n= 3) were acquired every 3 s using 20-deg pulses. A HP ¹³C spectrum (sum of 7 spectra) over 21-s period after the HP ¹³C-EAC injection is shown in Fig. 4a, showing all expected HP ¹³C species. Our results showed that HP ¹³C-MAC, ¹³CO₂, and H¹³CO₃⁻ peaks appeared soon after the injection of HP ¹³C-EAC (Fig. 4b). The signal of the parent compound HP ¹³C-EAC was very small in every spectrum compared to its downstream metabolites. This indicates that the hydrolysis of ¹³C-EAC by esterase is very rapid in vivo. Although the HP ¹³C-MAC and HP H¹³CO₃⁻ peaks in this in vivo result are somewhat overlapped due to greater field inhomogeneity, the two resonances are resolved and the peak intensities of the two HP compounds can be quantified by peak fittings using Gaussian and Lorentzian functions (Fig. S6). The intensities of the H¹³CO₃⁻ and ¹³CO₂ signals are plotted as a function of time in Fig. 4b. A ratio between HP H ¹³CO₃⁻ and ¹³CO₂ signal intensities was calculated over 21 s (7 time points) while HP ¹³CO₂ signal intensity can be reliably measured. A bar graph showing an average value of these ratios are shown in Fig. 4c. From these values, the average HP H¹³CO₃⁻/¹³CO₂ signal intensity ratio was 11.9:1 and the average pH value was 7.24 ± 0.08 (Fig. 4d). This value agrees very well with the pH of rat livers reported previously.²⁷ It is important to note that in all three rats injected with HP ¹³C-EAC, there were no visible signs of acute toxicity following the injection as observed from the normal heart and respiratory rates. All rats also recovered normally after the experiments, suggesting that ¹³C-EAC is biocompatible.

Fig. 4. In vivo pH measurements of a rat liver.

a) A ¹³C NMR spectrum of rat liver following HP ¹³C-EAC (80 mM in PBS) administration. The spectrum is a sum of 7 spectra acquired every 3 s over 21 s with 10-deg pulses at 4.7T; b) signal intensities of HP H¹³CO₃⁻ and ¹³CO₂ from the rat liver; c) average HP H ¹³CO₃⁻/¹³CO₂ signal ratio calculated over 21 s; d) average liver pH value estimated from the HP H¹³CO₃⁻/¹³CO₂ ratio; and e) a ¹H reference image showing the liver for HP ¹³C MRS.

In conclusion, we have demonstrated that tissue pH can be measured in vivo using HP ¹³C-EAC. Taking advantage of the abundant esterase and carbonic anhydrase activities in mammalian tissues, HP ¹³CO₂ and H¹³CO₃⁻ can be rapidly produced in situ and tissue pH can be calculated from HP ¹³C signal ratios of the physiological buffer pair. The T₁, reasonable chemical stability, and high polarization level of HP ¹³C-EAC are also advantageous for HP ¹³C imaging applications, potentially allowing for pre-injection quality control procedures such as radical removal without a significant signal loss. The results suggest that HP ¹³C-EAC is an attractive pH imaging agent for in vivo assessment of abnormal tissue pH associated with many diseases.