Imaging of pH in vivo using hyperpolarized 13C-labelled zymonic acid

Natural pH regulatory mechanisms can be overruled during several pathologies such as cancer, inflammation and ischaemia, leading to local pH changes in the human body. Here we demonstrate that 13C-labelled zymonic acid (ZA) can be used as hyperpolarized magnetic resonance pH imaging sensor. ZA is synthesized from [1-13C]pyruvic acid and its 13C resonance frequencies shift up to 3.0 p.p.m. per pH unit in the physiological pH range. The long lifetime of the hyperpolarized signal enhancement enables monitoring of pH, independent of concentration, temperature, ionic strength and protein concentration. We show in vivo pH maps within rat kidneys and subcutaneously inoculated tumours derived from a mammary adenocarcinoma cell line and characterize ZA as non-toxic compound predominantly present in the extracellular space. We suggest that ZA represents a reliable and non-invasive extracellular imaging sensor to localize and quantify pH, with the potential to improve understanding, diagnosis and therapy of diseases characterized by aberrant acid-base balance.


Supplementary Figure 4 | Stability of ZA in D 2 O as a function of time.
Slow chemical decay of ZA into parapyruvic acid (PP) in D 2 O at pH = 7.54 ± 0.01. 1 H spectra were acquired at 1 T and 27 °C over 20 h, each spectrum averaged over 60 scans within 10 min. (a) The single proton H(ZA) attached to carbon ZA 3 can be seen at ~5.9 ppm, HDO at 4.7 ppm, DMSO at 2.7 ppm, the methyl group H 3 (ZA) attached to carbon ZA 5 at 1.55 ppm and the methyl group H 3 (PP) of parapyruvate hydrate at 1.35 ppm. The single proton attached to carbon ZA 3 is quickly exchanged for a deuteron by keto-enol-tautomerism and can thus only be observed in the first few spectra. (b) Peak amplitudes and exponentially fitted curves of the methyl groups of ZA and parapyruvic acid (PP) from a showing that ZA decays into PP with a half-life of t 1/2 = 2.27 ± 0.04 h.
Supplementary Figure 5 | 13 C biosensor pH of the same buffer phantom measurement evaluated from ZA with and without considering the additional urea peak used as chemical shift reference at 7 T. The 13 C biosensor pH was backcalculated based on the chemical shift difference of both 13 C-labeled ZA positions (a) with respect to the pH insensitive 13 C urea and (b) based on the chemical shift difference between the two 13 C-labeled ZA positions only. (c) The pH values extracted from the two 13 C pH maps correlate well with the electrode pH (in white in a and b). At the limit of its sensitivity (at pH ≈ 5), the back-calculation of the 13 C biosensor pH is improved by taking the urea peak into account as pH insensitive chemical shift reference. Scale bars, 1 cm.

Supplementary Figure 6 | Longitudinal relaxation time T 1 of hyperpolarized natural abundance ZA in vitro at 3 T.
A three-parameter monoexponential curve was fitted to each dataset and the mean and standard deviation was calculated from the resulting decay constants of 50 mM ZA in 80 mM Tris buffer in H 2 O adjusted with 1M NaOH to an average pH of 6.53 ± 0.03 at 27 °C. The close proximity of the frequently and fast exchanging proton of the hydroxy group attached to carbon number two of ZA most likely causes the shorter T 1 of carbon number one ( 13 ZA 1, a) compared to carbon number five ( 13 ZA 5 , b) of ZA in vitro.

Supplementary Figure 7 | Cytotoxicity tests show that ZA is non-toxic within experimentally relevant concentration ranges.
Typical concentrations of hyperpolarized substances injected into animals are on the order of 60-100 mM at a dose of approximately 5 mL kg -1 , resulting in an end concentration of the substance in the blood of 6-10 mM assuming a ratio of injected volume to blood volume in the order of 1:10. Therefore, 5000 HeLa cells each in 100 µL cell culture medium were incubated with ZA for 24 h at concentrations 0.4-12.5 mM (a) without and (b) with a Zn catalyst being used in the synthesis of ZA before being purified using reversed phase HPLC. No marked reduction in cell viability was observed (n = 3, mean ± s.d.).

Supplementary Figure 8 | Dose escalation study testing for in vivo toxicity of ZA in three rats. ZA was dissolved in 80 mM
Tris buffer solution, neutralized to normal blood pH ≈ 7.4 using NaOH, sterile filtered and injected into the tail vain at t = 0 min at a final injected concentration of 40 mM and 80 mM, at a dose of 5 mL kg -1 and a rate of 0.17 mL s -1 . For all three rats (Lewis, male, Charles River, average weight 319 ± 1 g), (a) heart rate, (b) breathing rate and (c) blood oxygenation were monitored for five minutes before and after injection. No abnormalities with respect to the injection of ZA were detected.  (d) thrombocytes. All levels are close to the reference levels (erythrocytes: 5.5-9.3 × 10 3 L -1 , hemoglobin: 106-156 g L -1 , leucocytes: 3.3-8.7 × 10 9 L -1 , thrombocytes: 500-1300 × 10 9 L -1 ) provided by the supplier (Charles River) and no significant difference between ZA and NaCl injected animals can be detected. 14 animals (7 female / 7 male) were used, 10 of them received a tail vein injection of 5 mL kg -1 with a concentration of 250 mM ZA (5 times the dosage used for the imaging experiments), 4 animals served as controls with a tail vein injection of saline (0.09 % w/v of NaCl) (see also Methods).  , c) Proton anatomical images of the three axial slices through intestines, kidney and liver where pH and T 1 measurements were performed within two animals. (b, d) Time resolved slice selective spectra of hyperpolarized ZA and 13 C urea report consistent pH values within the physiologically expected pH range after tail vain injection. (e) Monoexponential decay curves fitted to the peak maxima of ZA 1 , ZA 5 and urea result in slightly longer apparent T 1 times for ZA compared to urea and no significant differences in apparent T 1 between the three slices. For each in vivo experiment, spectra from 64 time steps were recorded using a flip angle of 10 ° and a TR of 3 s. The pH values were calculated from the difference of the chemical shifts of ZA 1 , ZA 5 and urea. The 13 C pH values are given as mean ± standard deviation of the n spectra that could be used for the back-calculation of the pH due to sufficient SNR. Scale bars, 1 cm.

Supplementary Figure 15 | Representative fits and fitting residuals for multiple tissue compartments in the kidneys and in the tumor.
(a-c) In the kidneys, increasing the number of fitted zymonic acid peak pairs from one (a, R 2 = 0.77) to two (b, R 2 = 0.92) to three (c, R 2 = 0.95) results in a reduction of the fitting residuals (red line) and an improved coefficient of determination R 2 . (d-e) Analogously, increasing the number of fitted zymonic acid peaks pairs from one (e, R 2 = 0.83) to two (f, R 2 = 0.88) results in a reduction of the fitting residuals (red line) and an improved coefficient of determination R 2 in the tumor. Urea (0 ppm) and parapyruvate hydrate (15.7 ppm) are fitted in all spectra.
Supplementary Figure 16 | Distribution of 12 C ZA in rat kidney and extravasation into kidney tissue was confirmed by matrix-assisted laser desorption/ionization mass spectrometry imaging (MALDI-MSI). (a) H&E stains of analyzed tissue sections. (b) Distribution of 12 C ZA (m/z 157.0142) in kidney section 30 seconds after injection of 5 mL kg -1 250 mM 12 C-ZA. The control rat was administered with isotonic saline. A positive control, 1 mM 12 C-ZA droplet on slide, and a non-tissue measurement region acting as a background control were included in the measurement. All samples were coated with 9aminoacridine and analyzed in negative mode on a FT-ICR MS. Data was acquired at a spatial resolution of 150 µm and normalized by root mean square. Dashed lines mark the measurement regions. (c) Higher magnification image showing arterial and venous blood vessels (black and white arrows) and high abundant signals of ZA in the medulla of the kidney. (d) Mean 12 C ZA concentrations in renal cortex and medulla. MALDI-MSI represents the distribution of ZA within the kidney fixed 2-3 minutes after injection whereas hyperpolarized MRI shows the distribution of ZA within the kidney 10 s after injection. Whereas in the hyperpolarized MR image, shortly after injection, the cortex exhibits the largest contribution to the overall signal, in MALDI-MSI, much longer after injection, more ZA is already involved in the renal filtering process and thus the area containing the medulla and calyx show the largest signal contribution. Scale bars, 2 mm. Fig. 5 bearing a Mat B III tumor (arrow). (a-p) Proton images with a field of view of 6 cm were acquired every 1 mm using a fast spin echo sequence (see Methods). (g-k) The five proton images contained within the 5 mm thick hyperpolarized 13 C image are marked with a blue box. Image (i) represents the central tumor slice and coincides with the center of the 5 mm thick hyperpolarized 13 C image. Scale bars, 1 cm.