In vitro imaging of bacteria using 18F-fluorodeoxyglucose micro positron emission tomography

Positron emission tomography (PET) with fluorine-18-fluorodeoxyglucose (18F-FDG) can be applied to detect infection and inflammation. However, it was so far not known to what extent bacterial pathogens may contribute to the PET signal. Therefore, we investigated whether clinical isolates of frequently encountered bacterial pathogens take up 18F-FDG in vitro, and whether FDG inhibits bacterial growth as previously shown for 2-deoxy-glucose. 22 isolates of Gram-positive and Gram-negative bacterial pathogens implicated in fever and inflammation were incubated with 18F-FDG and uptake of 18F-FDG was assessed by gamma-counting and µPET imaging. Possible growth inhibition by FDG was assayed with Staphylococcus aureus and the Gram-positive model bacterium Bacillus subtilis. The results show that all tested isolates accumulated 18F-FDG actively. Further, 18F-FDG uptake was hampered in B. subtilis pts mutants impaired in glucose uptake. FDG inhibited growth of S. aureus and B. subtilis only to minor extents, and this effect was abrogated by pts mutations in B. subtilis. These observations imply that bacteria may contribute to the signals observed in FDG-PET infection imaging in vivo. Active bacterial FDG uptake is corroborated by the fact that the B. subtilis phosphotransferase system is needed for 18F-FDG uptake, while pts mutations protect against growth inhibition by FDG.

Enterococcus faecium were cultured in brain-heart infusion (BHI), while all other bacteria were cultured in tryptic soy broth (TSB). All bacteria were grown overnight at 37 °C under constant agitation (250 rpm), with the exception of S. pneumoniae which was grown in standing culture at 5% CO 2 . The next day, fresh cultures were prepared by dilution of overnight cultures in 9 mL TSB or BHI, and growth was continued up to an optical density (OD) of 1-5 McFarland units. Heat-killed Staphylococcus aureus and Escherichia coli were obtained by incubating the respective cultures for 30 min at 99 °C. To assess 18 F-FDG uptake, the bacterial cultures were incubated with 5-10  18 F-FDG dissolved in 0.5 mL NaCl (0.9%) at 37 °C for 5 min. Subsequently, 2 mL samples were 'washed' twice by centrifugation and resuspension of the bacterial pellets in phosphate buffered saline (PBS).
To image 18 F-FDG uptake, the bacteria were pelleted by centrifugation, and four Eppendorf tubes containing bacterial pellets were positioned at the center of the ring system of a micro-PET (µPET) Focus 220 scanner (Siemens Medical Solutions, TN, US) with the bacterial pellets in the field of view. Images were recorded in 30 min and analyzed with the AMIDE software package (Amide's a Medical Imaging Data Examiner). To quantify 18 F-FDG uptake, the bacterial pellets were placed in a calibrated gamma-counter (CompuGamma CS1282, LKB-Wallac, Turku, Finland). The radioactivity in each sample was measured, corrected for background and converted into kilobecquerel (kBq

Growth of S. aureus and B. subtilis in the presence of non-labeled FDG.
To assess the toxicity of FDG, the laboratory strain S. aureus HG001 and different B. subtilis strains ( Table 2) were grown overnight in TSB or LB, respectively. Where appropriate the media were supplemented with chloramphenicol or erythromycin at a concentration of 5 µg/mL. The overnight cultures were, in triplicate, diluted 1:50 in 100 µL fresh TSB or LB in 96-well microtiter plates, and the plates were incubated for 3 h at 37 °C with vigorous agitation (800 rpm). Lastly, each culture was again diluted 1:50 in 100 µL of the respective growth medium supplemented with 2 µg/mL, 20 µg/ mL, 200 µg/mL or 2 mg/mL non-radioactive FDG, or without FDG. Cultures were incubated at 37 °C in a Biotek synergy 2 plate reader with shaking and the OD at 600 nm (OD 600 ) was recorded at 10 min intervals. Each growth experiment was performed at least twice.

Statistical analyses.
Kruskal Wallis and Mann-Whitney U tests were performed using IBM SPSS Statistics 23. P-values of 0.05 or less were considered significant.

Results
In vitro uptake of 18 F-FDG. To evaluate whether bacterial pathogens are capable of active uptake of 18 F-FDG, a range of different Gram-positive and Gram-negative bacteria including mostly clinical isolates were incubated with 18 F-FDG, and uptake was determined with a calibrated gamma-counter. The 18 F-FDG PET signals were quantified and the absorbed activity per CFU was determined as shown in Fig. 1(A and B). All bacterial isolates showed significant 18 F-FDG uptake. No major differences were observed for most of the investigated Gram-positive and Gram-negative bacteria (p = 0.60; not shown). Nonetheless, statistically significant differences in 18 F-FDG uptake were observed for different bacterial species (p < 0.01) as illustrated in Fig. 1. The Grampositive bacterium Streptococcus pyogenes displayed overall the highest uptake activity per 10 7 CFUs ( Fig. 1A and B). E. coli and Citrobacter freundii showed the highest 18 F-FDG uptake amongst the tested Gram-negative species (Fig. 1B). In contrast, S. pneumoniae, E. faecium and P. mirabilis showed relatively low 18 F-FDG uptake, but as shown for S. pneumoniae the 18 F-FDG uptake was still significantly higher than the negative controls (p = 0.01; Fig. 1C). Importantly, other living bacteria, such as S. aureus and E. coli accumulated significantly higher 18 F-FDG levels than the respective heat-killed bacteria, bacteria incubated in the absence of 18 F-FDG, or medium controls (p = 0.004).  To evaluate whether bacterial 18 F-FDG uptake can also be detected by µPET imaging, the 18 F-FDG uptake by several Gram-negative and Gram-positive clinical isolates was qualitatively assessed with a µPET Focus 220 scanner. Indeed, the resulting µPET images clearly show 18 F-FDG uptake by the tested bacteria (Fig. 2). S. pneumoniae showed the least uptake activity, which might relate to the high autolytic activity of this bacterium 21 . As expected, little if any 18 F-FDG uptake was detectable for heat-killed S. aureus and E. coli, as was the case for non-incubated bacteria and medium controls. Notably, there were some apparent differences in the relative signals observed in Figs 1 and 2 for, respectively, the E. faecium and E. faecalis samples, and for the S. aureus and S. epidermidis samples. This may relate to the fact that counting with the calibrated gamma-counter as in Fig. 1 provides quantitative data on 18 F-FDG uptake, whereas the µPET scans as in Fig. 2 give semi-quantitative information on 18 F-FDG uptake. For example, there may have been some variation in the distribution of the pelleted bacteria in the samples that were imaged by µPET scans. Furthermore, the observed variations may have been due, to some extent, to differing numbers of metabolically active bacteria in the samples respectively used for Fig. 1 and 2.
To approximate the time needed for uptake of 18 F-FDG, time course experiments were performed with S. aureus and E. coli, two bacterial species showing significant differences in 18 F-FDG uptake (Fig. 1). The results presented in Fig. 3 show indeed significant differences in the rates of uptake by both species, E. coli reaching saturation almost instantaneously while 18 F-FDG uptake by S. aureus was relatively slow. Within 60 min of incubation, E. coli absorbed significantly more 18 F-FDG than S. aureus (p < 0.001), while E. coli and S. aureus both showed higher signals than the respective heat-killed controls (p < 0.01).
Potential mechanism of 18 F-FDG uptake. In order to assess the possible mechanism of 18 F-FDG uptake, we applied the Gram-positive model bacterium B. subtilis for which mutations in the phosphotransferase system (PTS) that facilitates glucose uptake are available 22 . Indeed, B. subtilis ptsI and ptsG deletion strains displayed a strongly impaired uptake of 18 F-FDG compared to the parental wild-type strain B. subtilis 168 (p < 0.001; Fig. 3).  18 F-FDG, and medium controls. The indicated standard deviations include the variations introduced by the use of at least two different isolates per tested species. Note that the detection limit for a bacterial species will lie between the nonspecifically absorbed activity measured for heat-killed bacteria, such as E. coli and S. aureus, and the lowest level of 18 F-FDG uptake by the living bacteria, such as S. pneumoniae. This implies that under the tested conditions, the detection limit is ~2.4 kBq. *p ≤ 0.01. Values represent mean ± SEM (n = 8).
Of note, the ptsI deletion mutant showed a somewhat higher uptake than the ptsG deletion mutant (p = < 0.001), which is in line with the fact that PtsG is the major glucose transporter 23 , while PtsI is involved in the phosphorylation of internalized glucose making its uptake by PtsG irreversible 24 . These findings show that, at least in B. subtilis, 18 F-FDG is taken up via the PTS system.

FDG-impaired bacterial growth. To test whether FDG could be toxic for bacteria growth experiments
were performed with S. aureus HG001 and B. subtilis 168, where the growth media were supplemented with increasing concentrations of FDG up to 2 mg/mL. As shown in Fig. 4A, S. aureus HG001 reached somewhat lowered OD 600 values in the stationary phase when the medium contained 200 µg/mL FDG or more. Similarly, B. subtilis reached a lowered optical density in the stationary phase when cultured in the presence of 2 mg/mL FDG (Fig. 4B). This indicates that FDG is mildly toxic for bacteria, such as S. aureus and B. subtilis, at high concentrations. To verify this, the B. subtilis ptsI and ptsG deletion mutants were also cultured in the presence of FDG. The results show that both mutations significantly diminished the growth impairment caused by 2 mg/mL FDG ( Fig. 4C and D). Together, these findings show that active FDG uptake may be toxic for bacteria, but only at very high concentrations that will not be reached during 18 F-FDG-PET imaging in vivo.

Discussion
The current study shows for the first time that clinical isolates of many major Gram-positive and Gram-negative bacterial pathogens can actively take up 18 F-FDG. This is consistent with the previous finding of Weinstein et al. 25 , who reported that type strains of E. coli, Klebsiella pneumoniae and S. aureus can take up this compound. The combined findings imply that bacterial pathogens may contribute to the signal observed in 18 F-FDG PET imaging   of bacterial infections. Furthermore, the observed 18 F-FDG uptake was, at least in B. subtilis, shown to be facilitated by the PTS system for glucose uptake. Lastly, our present observations show that FDG may be toxic for bacteria but only at very high doses that exceed the physiological levels during 18 F-FDG PET imaging about 100-to 1000-fold.
Notably, of all presently tested bacterial species, S. pyogenes displayed the highest uptake of 18 F-FDG. This bacterium is notorious for causing necrotizing fasciitis, a rapidly progressive infection of deep layers of the skin and subcutaneous tissues, which requires immediate antibiotic therapy and aggressive surgery 26 . The high uptake levels of 18 F-FDG suggest that S. pyogenes may have been the metabolically most active bacterium under the tested conditions. All other bacteria showed lower 18 F-FDG uptake in vitro. In view of the fulminant pathology of invasive S. pyogenes infections, it is conceivable that this bacterium displays also a high metabolic activity in vivo.
In line with the finding that 18 F-FDG is actively taken up by the investigated bacteria, the absence of the glucose transporter PtsG abrogated FDG uptake by B. subtilis. The absence of PtsI, also known as Enzyme I of the PTS system 22 , had a strong but less drastic negative effect on 18 F-FDG uptake suggesting that its contribution to FDG uptake can to some extent be bypassed. Further, our present results show that FDG is toxic for S. aureus and B. subtilis, but only at high concentrations that are not reached during 18 F-FDG PET imaging in vivo. Yet, the observed toxicity of FDG is reminiscent of the toxic effects of 2-DG. In the latter case, the PTS system recognizes 2-DG and phosphorylates it into 2-DG-6-phosphate, which is subsequently dephosphorylated by hexose-6-phosphatase. This cycle of 2-DG phosphorylation and de-phosphorylation depletes the bacterial cell of phosphoenolpyruvate and eventually adenosine triphosphate (ATP) 27 . Thus, 2-DG has the ability to de-energize bacteria. This may also be the case for FDG, which would explain the observed growth inhibition at high FDG concentrations. Of note, the growth of B. subtilis ptsG or ptsI deletion mutants was much less affected by FDG than growth of the parental strain, which is consistent with the reduced uptake of 18 F-FDG.
Currently, there is ample evidence that 18 F-FDG PET is a very useful diagnostic tool for many infectious indications, including fever of unknown origin, endocarditis, spondylodiscitis, and vascular graft infections 8,11,12 . This approach can help in defining the location and extent of infection, thereby providing guidance for biopsy and for therapy follow-up. In general, the detection limit in the imaging of infections by PET will depend on the sensitivity and resolution of the PET camera, on the volume of the bacterial lesion, on the surrounding tissue, and on the target-to-normal tissue (T/N) ratios of radioactivity at sites of infection. Our present study shows that 18 F-FDG uptake by infecting bacteria potentially contributes to the overall signal detected by PET imaging of bacterial infections. However, it remains difficult to approximate the contribution of the bacterial 18 F-FDG uptake to the total 18 F-FDG uptake signal measured during in vivo PET imaging of an infection, because the bacterial contribution to the total signal will vary depending on the bacterial species that causes the infection, on the metabolic state of the infecting bacteria, and on the total number of infecting bacteria. Moreover, the inflammatory status of the infected tissue will impact on the total signal, especially because white blood cell migration may vary depending on the extent of inflammation, and also because inflammation may lead to increased blood flow and vascular permeability. Furthermore, the total signal is likely to depend on the particular body site that is infected. Altogether, this means that, in practice, the relative bacterial contribution to the total 18 F-FDG signal will be different for individual patients and for different types of infections. For example, chronic infections can be visualized with 18 F-FDG, but the signal contributed by the bacteria themselves may be lower than in acute infections if their metabolic activity is lower, which would result in lower uptake of 18 F-FDG. In such cases, an option to improve the bacterial signal could be to start imaging at later time points after tracer injection. Clearly, the conditions in vivo are likely to be different compared to the conditions applied in our in vitro study. It is therefore presently not possible to make reliable calculations of the detection limit for infections in vivo.
Lastly, with 18 F-FDG PET it is not possible to differentiate between sterile inflammation and infection. Also, 18 F-FDG PET will neither tell us whether bacteria are involved in an infection, nor which bacteria are involved. Specific bacteria-targeted PET tracers are therefore desirable. For example, these could include labeled antibiotics like 18 F-or 99m Tc-ciprofloxacin (Infecton ® ) 6, 28-31 . A very promising alternative for 18 F-FDG may be 2-[18 F]-fluorodeoxysorbitol ( 18 F-FDS), since sorbitol is a sugar alcohol that is mainly metabolized by the Gram-negative Enterobacteriaceae 5,25 . Another attractive approach appears to be the detection of bacteria with 6-[ 18 F]-fluoromaltose ( 18 F-FM) 32 , because maltose and maltodextrins are metabolizable by all classes of bacteria, in contrast to mammalian cells. Thus, 18 F-FM may allow the distinction between bacterial infection and other causes of inflammation. On this basis it is well conceivable that the combined use of 18 F-FDS and 18 F-FM will facilitate the detection of a wide spectrum of bacteria causing infections. With the aid of both substrates, it should not only be possible to specifically image infecting bacteria over mammalian cells, but also to discriminate infecting Gram-positive and Gram-negative bacteria. The latter would then guide effective antimicrobial therapy. Yet, despite these evident advantages, no clinical studies on infection imaging with 18 F-FDS and 18 F-FM have thus far been reported.

Conclusion
To date, 18 F-FDG is one of the few tracers that may be clinically applied in infection imaging. We now show that this tracer can be taken up by a wide range of commonly encountered bacterial pathogens. We therefore hypothesize that bacterial pathogens may also take up 18 F-FDG in vivo, thereby contributing to the FDG-PET signal observed in infection imaging. Future experiments are needed to determine the proportion of 18 F-FDG that is taken up by infecting bacteria as compared to the 18 F-FDG taken up by inflammatory cells. This will unveil the significance of our present findings in the clinical context.