Abstract

Many bacteria, but not mammalian cells, can directly import and metabolize extracellular gluconate via gluconate permease (GntP) and kinase (GntK). We hypothesized that fluorine-18 labeled gluconic acid ([¹⁸F]FGA) could leverage unique bacterial gluconate metabolism for pathogen-specific PET imaging. Here, we report [¹⁸F]FGA as a new PET tracer targeting bacterial gluconate metabolism. The chemoenzymatic radiosynthesis of [¹⁸F]FGA was simple, only requiring readily available [¹⁸F]FDG and glucose oxidase. [¹⁸F]FGA showed broad detection sensitivity across multiple bacterial species, including multidrug-resistant clinical isolates. Staphylococcus aureus transposon mutant studies showed substantial loss of [¹⁸F]FGA uptake in GntP and GntK mutants, supporting high specificity for bacterial gluconate metabolism. [¹⁸F]FGA PET selectively highlighted live bacterial infection, with high target-to-nontarget ratios in S. aureus (36-fold) and Escherichia coli (30-fold) in a murine model of myositis. Unlike [¹⁸F]FDG, no appreciable uptake of [¹⁸F]FGA was detected at sites of sterile inflammation. In a Klebsiella pneumoniae pneumonia model, [¹⁸F]FGA showed significantly higher accumulation in the infected lung (5.2 ± 0.8%IA/cc) than in the uninfected lung (0.1 ± 0.0%IA/cc), further supporting its potential for imaging challenging clinical infections. Taken together, [¹⁸F]FGA PET may be a useful tool for pathogen-targeted imaging in clinical practice.

Results from nanoScan® PET/CT

To assess in vivo distribution and elimination profile of [¹⁸F]FGA, PET data were acquired on a 90 min list mode immediately after tracer administration in uninfected mice (n = 4, unconscious). Representative time-course PET images indicate rapid distribution and clearance of [¹⁸F]FGA at early time after tracer injection, followed by rapid urinary excretion (Figure 1A). In a different mice cohort, we performed a 5 min static scan of [¹⁸F]FGA at 85 min after tracer injection to assess if physiological states affect in vivo profile of tracer (Figure 1A). Unlike [¹⁸F]FDG PET, which is highly affected by the imaging conditions, [¹⁸F]FGA PET showed comparable biodistribution under different physiological conditions (conscious vs unconscious) in both in vivo and ex vivo analysis, with only minimal differences likely attributable to slower blood circulation under the unconscious state (Figure 1B,C).

Figure 1.

To investigate the uptake pathway, we compared in vitro uptake of [¹⁸F]FGA in a S. aureus JE2 strain and two isogenic JE2 transposon mutants (ΔgntP and ΔgntK). Fluorine-18 labeled maltose ([¹⁸F]FDM) was selected as a control tracer, as our previous transposon mutant study demonstrated that its uptake is mediated by the maltose/maltodextrin transport system (malE, malF, malG, and malK) (Figure 2A).
The authors observed no appreciable tracer uptake in the ΔgntP mutant, suggesting that [¹⁸F]FGA uptake in S. aureus requires GntP (Figure 2B). [¹⁸F]FGA uptake was significantly reduced in the ΔgntK mutant, suggesting that intracellular phosphorylation via GntK may contribute to tracer retention. The uptake of [¹⁸F]FDM was not affected by disruption of gntP or gntK, suggesting that the gluconate metabolic pathway is specific for [¹⁸F]FGA in S. aureus. We further tested the use of [¹⁸F]FGA PET in a murine model of myositis (Figure 2C). Mice were infected with S. aureus JE2 strain in the right triceps muscle and ΔgntP mutant in the left triceps muscle. [¹⁸F]FGA PET was conducted without any metabolic inhibitor, whereas [¹⁸F]FDM PET was conducted by co-injection with voglibose (α-glucosidase inhibitor, 1 mg per head) to prevent in vivo degradation. [¹⁸F]FGA showed 15.2 ± 4.3 fold higher uptake at the JE2 infected muscle (0.90 ± 0.17%IA/cc) than ΔgntP mutant infected muscle (0.06 ± 0.01%IA/cc, p = 0.0026). The radioactivity of [¹⁸F]FGA PET at the site of ΔgntP mutant infection was not significantly different with that in uninfected mice (p = 0.4250). In contrast, uptake of [¹⁸F]FDM was not significantly different between JE2 and ΔgntP mutant infected muscles (1.51 ± 0.17 vs 1.37 ± 0.27%IA/cc, p = 0.2996) (Figure 2D). Ex vivo CFU quantification data confirmed comparable bacterial burdens between JE2 and ΔgntP mutant, further supporting that the observed uptake of [¹⁸F]FGA in S. aureus JE2 strain were not driven by difference in bacterial burden or host inflammation (Figure 2E). Taken as a whole, these data support both specificity and selectivity of [¹⁸F]FGA for gluconate-dependent metabolism in active S. aureus.

Figure 2.

The authors further evaluated [¹⁸F]FGA PET in murine myositis models (n = 4 each) to assess its ability to distinguish bacterial infection from sterile inflammation. Infection was induced by intramuscular inoculation of live S. aureus or E. coli (10⁶ cells) into the left triceps muscle, with a 10-fold higher concentration of heat-killed bacteria injected into the contralateral muscle, 6 h before imaging study. For sterile inflammation, LPS (5 mg/kg) was injected into the left triceps muscle 96 h prior to imaging study. At 80 min post injection of [¹⁸F]FGA, PET data were acquired using a 10 min static scan. In the LPS-induced myositis model, no significant [¹⁸F]FGA uptake was observed at the site of LPS injection (0.06 ± 0.02%IA/cc), similar to that observed in the uninfected mice (0.06 ± 0.01%IA/cc) (Figure 3A). Quantitative analyses of the left triceps muscle showed no significant difference between uninfected mice and LPS-treated mice both in vivo (p = 0.9470) and ex vivo (p = 0.7777) (Figure 3B). In contrast, [¹⁸F]FDG PET showed 4.1 ± 0.7 fold higher activity at the site of LPS injected muscle (9.5 ± 2.0%IA/cc; p = 0.004) than in the contralateral triceps muscle (2.3 ± 0.4%IA/cc) (Figure 3A). In bacterial myositis models, [¹⁸F]FGA showed focal uptake in muscles infected with live S. aureus (2.5 ± 0.4%IA/cc) and E. coli (1.9 ± 0.3%IA/cc), whereas negligible uptake was observed in muscles injected with heat-killed bacteria (heat-killed S. aureus: 0.07 ± 0.01%IA/cc, heat-killed E. coli: 0.07 ± 0.01%IA/cc). The ROI-derived target-to-nontarget ratios (Live vs HK) were markedly higher for S. aureus (36.4 ± 4.6) and E. coli (29.6 ± 7.3) than for LPS-treated mice (1.1 ± 0.2) (Figure 3C,D).

Figure 3.

Next, given the short plasma elimination half-life and low nonspecific uptake of [¹⁸F]FGA in the lung (Table 1), the authors anticipated that [¹⁸F]FGA PET would be suitable for imaging pulmonary infections, in which interpretation could be confounded by blood pool background associated with pulmonary perfusion. Pulmonary infection was induced by intratracheal inoculation of K. pneumoniae (10⁶ cells). A 10 min [¹⁸F]FGA static scan was performed 80 min after tracer administration. In a K. pneumoniae pulmonary infection model, [¹⁸F]FGA showed significantly higher tracer accumulation in the infected lung (5.16 ± 0.80%IA/cc) than in the uninfected lung (0.09 ± 0.02%IA/cc). Representative whole-body MIP image showed a focal [¹⁸F]FGA signal in the lung (Red arrows). Co-registered µPET/CT images indicated that the focal [¹⁸F]FGA uptake localized to the region of lung infiltration identified on CT (Yellow arrows) (Figure 4A). Quantitative analyses indicated significantly higher [¹⁸F]FGA uptake in the infected lung than in the uninfected lung both in vivo (p = 0.001) and ex vivo (p = 0.0035) (Figure 4B).

Figure 4.

Full article on onlinelibrary.wiley.com