Most of our PET-CT studies to date have used 18F-2-fluoro-2-deoxyglucose (18F-FDG) to image the metabolism of the eukaryotic cells in TB lesions but we are also making attempts to identify the location, abundance and metabolic state of the bacteria in lesions. In an effort to identify small molecules that could be used to specifically label Mtb in vivo, we capitalized on the unusually broad substrate tolerance of the Mtb antigen 85 enzymes, which transfer mycolates onto structurally diverse sugars to form part of Mtbs cell wall. Antigen-85 enzymes are expressed on the exterior of MTbs cell wall and incorporate exogenous trehalose (a nonmammalian disaccharide consisting of a two 1-1 alpha, alpha;-linked glucose monomers) as either the mono- or dimycolate, even tolerating trehalose molecules containing bulky modifications. We have been able to observe 18F-FDG activity in lesions of infected rabbits by PET-CT imaging. A series of different positions and methods for attaching the 18F to the sugar are being explored to see which is most efficiently incorporated. Using trehalose should afford an improvement over the currently used 18F-FDG, as glucose is used by mammalian cells as well as bacterial, causing noise in the scans due to increased metabolism or inflammation in the host. Uptake of trehalose is unique to bacteria;it is not absorbed by mammalian cells, which should limit noise in PET scans. The Davis group has designed three trehalose analogs incorporating fluorine at the 2, epi-4, or 6-position of trehalose for use as PET radiotracers. The 2-fluorotrehalose (2-FDT) synthesis is a biomimetic process, inpired by bacterial synthesis of trehalose from glucose. Chemoenzymatic synthesis of 2-FDT occurs as a one-pot cascade reaction in which hexokinase transfers a phosphate from adenosine triphosphate (ATP) to 18F-FDG (normally glucose in bacterial trehalose synthesis). OtsA then transfers the glucose from the donor UDP-glucose to the acceptor phosphorylated 18F-FDG. Dephosphorylation to give the desired product is effected by OtsB. The entire one-pot process is complete in 45 min. The advantage here is that a relatively technically facile manipulation would convert a commercially available radiotracer to a TB-specific one. We had previously acquired some preliminary PET-CT scan data in rabbits, one healthy, one infected with HN878 Mtb. In the infected animal, four lesions were present. Two were not PET-active, and two were, although all four had similar amounts of colony forming units (CFU). Upon necropsy, the two PET-inactive lesions were extremely rigid and thick-walled, implying that uptake is related more to accessibility (i.e. vasculature) than amount of bacteria present. The t1/2 was approximately 40 min, and the radiotracer was clear from blood at 200 min post-injection, but later animals experienced unexpected toxicity within about 24 hours of the procedure. We were able to trace this toxicity to LPS contamination of the two enzymes OtsA and OtsB. Unfortunately, traditional techniques for LPS removal either before or after radiosynthesis were unsuccessful, likely due to the similar physicochemical properties of LPS, a polysaccharide, and trehalose, a disaccharide. As of last years update, we changed expression systems for OtsA and OtsB from E. coli to a baculovirus/insect cell expression system, which produces LPS-free enzymes in yields of approximately 10 mg/L for each protein. We learned that the C-terminal His-tag on OtsA is an absolute requirement for function and long term stability. N-terminal His-tagged OtsA was non-functional in our chemoenzymatic synthesis, and the absence of a His-tag (the native protein) led to problems replicating the initial expression and purification as well as degradation over time in storage leading to complete loss of function. The baculovirus/insect cell produced OtsA and OtsB gave reaction times comparable (and perhaps slightly shorter) than the E. coli produced enzymes. The current synthesis gives LPS-free radiotracer in approximately 60 min of synthesis time (including operationally simple cartridge-based purification steps) that is reproducibly >98% radiochemically pure, pH 7-8, isotonic, and osmolal. With a reproducible and successful radiosynthesis developed, we analyzed this radiotracer in rabbits. In our two initial trials using New Zealand white rabbits infected with the HN878 strain of Mtb, we were able to observe uptake of the radiotracer in lesions. The t1/2 for this radiotracer (dosed at approx. 2 mCi/kg), was approximately 60 min, and the tracer had cleared by 100 min post-injection. Although we saw uptake of the tracer to Mtb lesions, it appears that the rabbits metabolize as much as 30% of the radiotracer, as determined by analysis of blood and urine samples. Rabbits express significant quantities of trahalase, an enzyme which breaks down trehalose into two molecules of glucose. Trehalose is found in many plant materials typical to a rabbit diet, so rabbits have relatively high levels of trehalase expression, particularly in liver. In the case of our 2-FDT probe, trehalase metabolism generates one equivalent of PET-inactive glucose and one equivalent of 18F-FDG, the radiotracer routinely used by our lab for indirectly imaging disease. Although the lesions did appear to give good signal-to-noise, the possibility of metabolism made it impossible to assess whether this radiotracer was working as planned, because we have already established that 18F-FDG will label lesions, and up to 30% of the radioactivity circulating in plasma consisted of 18F-FDG. We then turned to a non-human primate model of tuberculosis. Our lab has developed a marmoset model of tuberculosis, which is highly similar to human disease in terms of pathology and response to chemotherapy. Primates also express trehalase but at much lower levels than rabbits. We used the 2-FDT in an Mtb-infected marmoset and were gratified to see that the lesions showed good uptake of the probe, and no metabolism was observable, either in blood or in urine samples drawn immediately after the PET-CT scan by both HPLC and autoradiography. The signal-to-noise in the marmoset was approximately 3:1, and even more gratifyingly, there was little uptake in the liver (in contrast to the rabbit) and heart. One of the problems with PET-CT using 18F-FDG has been the issue of hot hearts in both the rabbits and the marmosets due to glucose uptake in these types of tissue. Hot hearts are a problem for analyzing Mtb due to the position of the heart relative to the lung. Conservatively drawn regions of interest in the PET scans frequently do not include all the lung tissue to avoid including signal from heart uptake of 18F-FDG. Additionally, uptake could be seen in muscle of both animals. It appears that the 2-FDT probe is an improvement in both hot hearts and muscles. It appears that the low background of the trehalose probes will be a clear improvement over the currently used 18F-FDG. Finally, in an experiment with a related radiotracer, epi-4-fluoro-4-deoxytrehalose (4-FDT), we have been able to show that the proposed mechanism of labeling appears to be correct. We analyzed rabbit lung tissue and were able to determine that at least some of the radiotracer 4-FDT was converted to the corresponding mono- or dimycolate in vivo. This is strong support that the signal observed is due to bacterial uptake of the trehalose rather than nonspecific concentration of the radiotracer in dense tissue or due to secondary factors such as inflammation.

Project Start
Project End
Budget Start
Budget End
Support Year
Fiscal Year
Total Cost
Indirect Cost
Niaid Extramural Activities
Zip Code
Johnson, Daniel H; Via, Laura E; Kim, Peter et al. (2014) Nuclear imaging: a powerful novel approach for tuberculosis. Nucl Med Biol 41:777-84