Pyrazinamide (PZA) is a first-line drug used for the treatment of Mycobacterium tuberculosis. Despite the important role of PZA in shortening tuberculosis (TB) treatment, the exact mechanism of PZA action in vivo is poorly understood. Although in vitro studies have found PZA only to be active against TB at an acidic pH (e.g. 5.5) (1,2), this observation has not been verified by in vivo studies due to the lack of tools to measure pH of lesions noninvasively. Here chemical exchange saturation transfer (CEST) MRI was implemented to study pulmonary TB lesions to monitor in situ pH in live M. tuberculosis-infected mice, by exploiting the chemical exchange process of amide and amine protons in proteins (3). By assuming that the CEST contrast in a TB lesion is dominated by protein signals and in analogy to data from a protein phantom, we assess the shape of the CEST spectrum in the amide-amine proton shift range in TB lesions to assess lesion pH.

Four-to-six-week-old female C3HeB/FeJ mice were aerosol infected with M. tuberculosis H37Rv as described previously (4). All mice were imaged using in-house designed bio-containment devices to fit the MR coil (5). Five animals were imaged approximately 10 to 15 weeks post infection.

All MRI experiments were performed on a 9.4T Bruker Biospec system (Bruker, Germany). A 40 mm volume resonator was used for transmission and image acquisition. The single slice UTE-CEST scheme repeated a short saturation pulse (30 ms) followed by UTE readout with TE=0.3 ms and TR=40 ms (Fig. 1A). One Gaussian pulse with 0.3 ms pulse width and 15 degree flip angle was used in UTE. Slice thickness=1.5 mm; FOV=3.2 × 3.2 cm2; CEST saturation offsets were -3 kHz to 3 kHz with a step of 200 Hz. Two saturation powers were applied with flip angles of 3.7 µT and 7.4 µT. Post-mortem (immediately after imaging) pH in the TB lesion was also assessed using a micro-pH comb electrode and a benchtop pH meter.

Conventional T2-weighted images showed strong motion artifacts, compared to those acquired with UTE sequences (Fig. 1B&C). CEST spectra of bovine serum albumin (BSA) solutions with different pH values (Fig. 2A) demonstrated that, at acidic pH, the amine group (2 ppm offset) was significantly stronger than the amide peak (3 ppm offset). This amine/amide peak ratio was found to be very sensitive to pH in the acidic pH range. The averaged CEST spectra (n=5) of TB lesions recorded using UTE-CEST with B1 values of 3.7 µT and 7.4 µT are plotted in Figs. 2B & 2C respectively. The TB lesions showed strong saturation effects assumed to be due to high protein concentration, similar to other bacteria (6). Therefore, the shape of the CEST MTRasym signal from the lesion area may be useful to assess pH effects similar to the BSA phantom. Asymmetry analysis of the CEST spectrum both with a 3.7 µT and 7.4 µT B1 field shows negligible effect for offsets above 5ppm, indicating that, different from brain tissue, the semi-solid conventional Magnetization Transfer Contrast (MTC) is highly symmetric in the TB lesion. With higher saturation power (7.4 µT, Fig. 2C), strong peaks were visible in the 0-5 ppm range, tentatively assigned to amine and amide groups. The rNOE contributions mixed in during asymmetry analysis are known not to be as sensitive to pH as amide and amine contrast (see Fig. 2A). Therefore, it can be treated as a constant background in the MTRasym. Assuming dominance of protein signals to the MTRasym spectra (other bacteria show low rNOE)(5), the lack of high intensity signal around 3 ppm indicates that the pH of TB lesion is not in the acidic range (see Figs. 2A,C). A typical MTRasym map of mouse TB lesions and muscle is shown in Fig. 2D. Post-mortem pH of TB lesions measured by microelectrode was 7.40 ± 0.12 (n=5), confirming the imaging data. Assessment of lesion pH is important for activating TB drugs and these preliminary data show that CEST has potential to do this noninvasively in vivo. We have developed a novel UTE-CEST technique that can be used to perform CEST MRI on pulmonary TB lesions noninvasively in mice. The CEST-MRI results suggest that the pH of TB lesions in mice is not acidic.