Fluorine-18 labeled maltohexaose images bacterial infections by PET

Abstract

A new positron emission tomography (PET) tracer, composed of ¹⁸F labeled maltohexaose (MH¹⁸F), can image bacteria in vivo with a sensitivity and specificity that is orders of magnitude better than fluorodeoxyglucose (¹⁸FDG). MH¹⁸F can detect early stage infections composed of as few as 10⁵ E.coli colony forming units (CFUs), and can identify drug resistance in bacteria in vivo. MH¹⁸F has the potential to improve the diagnosis of bacterial infections given its unique combination of high specificity and sensitivity for bacteria.

The diagnosis of bacterial infections remains a central challenge in medicine. Infections are currently diagnosed via culturing of tissue biopsies or blood samples.^[1] However these methods can only detect late stage infections that are challenging to treat. Bacterial infections therefore cause an enormous medical burden, for example, the mortality caused by bacterial infections was greater than the mortality caused from AIDs, breast cancer and prostate cancer combined.^[2] Bacterial infections can be treated effectively, if diagnosed and treated at an early stage, and if the presence of drug resistance is also identified. However, this is challenging at present because the symptoms of infections look identical to a variety of other illnesses, such as cancer and inflammation.^[3] Thus in this clinical environment, an imaging technology that can identify and localize bacterial infections with high sensitivity and specificity has the potential to have a significant impact on medicine.

Positron emission tomography (PET) imaging has the potential to significantly improve the diagnosis of bacterial infections due to its unparalleled sensitivity.^[4] However, ¹⁸FDG is currently the only PET contrast agent available for clinical imaging of infections, and is problematic because it lacks specificity for bacteria and has high uptake in mammalian cells.^{[5] 18}FDG therefore cannot distinguish bacterial infections from other pathologies such as cancer and inflammation, and cannot diagnose bacterial infections at an early stage.^{[5a, 6]} Although numerous experimental PET contrast agents have been developed for imaging bacterial infections, such as radiolabeled antibiotics,^[7] antimicrobial peptides,^[1a] antibodies^[8] or white blood cells,^[9] these agents have had minimal clinical impact. Several factors have contributed to the lack of success of bacterial imaging agents, such as poor clearance due to non-specific adsorption, low target receptor expression on bacteria, or complicated radiochemical synthesis, which are challenging to perform in clinical radiochemistry labs.^[10] Therefore, there is a great need for the development of new PET contrast agents that can image small numbers of bacteria with high specificity in vivo.^[11]

Herein, we present a new PET tracer, composed of ¹⁸F labeled maltohexaose (MH¹⁸F) that can image bacterial infections in vivo with unprecedented sensitivity and specificity (see Scheme 1). MH¹⁸F targets the bacteria-specific maltodextrin transporter, which internalizes alpha 1,4 linked glucose oligomers (maltodextrins) as a source of glucose.^[12] The maltodextrin transport system is an ideal target for imaging bacteria because of its high uptake of maltodextrins (K_m of 130 μM),^[13] great specificity for bacteria, and the rapid clearance of maltodextrins from un-infected tissues.^[14] In addition, the maltodextrin transporter is only functional in metabolically active bacteria and MH¹⁸F uptake is therefore an indicator of bacterial viability,^{[14b, 15]} and potentially antibiotic efficacy. Finally, MH¹⁸F should have minimal toxicity in humans because maltodextrins are a commonly used food additive.^[16]

Scheme 1.

Synthesis of MH¹⁸F. MH¹⁸F is composed of ¹⁸F-fluoride conjugated to maltohexaose and was synthesized by one-step nucleophilic ¹⁸F-fluorination of brosylate-maltohexaose 3.

A synthetic strategy was devised to synthesize MH¹⁸F via nucleophilic 18-Fluorination of the maltohexaose-brosylate precursor (3) with K¹⁸F in the presence of kryptofix k222 (see Scheme 1). The reducing end of maltohexaose was selected for fluorination because the maltodextrin transporter recognizes the non-reducing end of maltodextrins and should therefore tolerate substitutions at the reducing end.^[17] Azide functionalized maltohexaose 1 was synthesized from maltohexaose in 4 steps following established methods,^[14b] and was conjugated with pent-4-yn-1-yl 4-bromobenzenesulfonate 2 using the Cu(I) catalyzed Huisgen cycloaddition, to afford the brosylate-maltohexaose precursor 3.^[18] Radiochemical synthesis of MH¹⁸F was carried with cryptate-mediated nucleophilic substitution of the brosylate precursor 3 with potassium ¹⁸F-fluoride (K¹⁸F), followed by basic hydrolysis with NaOH and acid neutralization. A decay corrected yield of 4.2% was obtained for this synthetic procedure, starting from ¹⁸F-fluoride, with an 87 % radiochemical purity, based on radiometric HPLC (see Supplementary Figure S5).^[19] The protocol for the synthesis of MH¹⁸F had a synthesis time of 100 minutes, and follows the same procedures used to make ¹⁸FDG,^[20] and should therefore be achievable in clinical radiochemistry laboratories. In addition, we anticipate that the radiochemical yield of MH¹⁸F can be increased using new F-18 fluorination methodologies.^[19]

MH¹⁸F is designed to selectively target bacteria due to the presence of maltodextrin transporters in bacteria, and their absence in mammalian cells. We therefore investigated if MH¹⁹F has specificity for bacteria over mammalian cells, and if it is internalized via the maltodextrin transporter LamB, using F19-NMR. Bacteria (E.coli) and mammalian cells (hepatocytes) were incubated with a 500 μM concentration of MH¹⁹F for one hour, washed with PBS, lysed, and the cellular supernatant was analyzed using F19-NMR. Figures 1a and 1b demonstrate that MH¹⁹F has high specificity for bacteria over mammalian cells and is robustly internalized. For example, under these conditions, E.coli had accumulated 2 orders of magnitude more MH¹⁹F than hepatocytes, and reached millimolar intracellular concentrations. In addition we performed maltohexaose competition experiments and experiments with LamB mutant E.coli to determine if MH¹⁹F was being internalized via the maltodextrin transport pathway. Figure 1a demonstrates that the uptake of MH¹⁹F in E.coli could be inhibited by an excess of maltohexaose, and that there is minimal uptake of MH¹⁹F in LamB mutants, demonstrating that MH¹⁹F enters E.coli via the maltodextrin transport pathway.

Figure 1.

MH¹⁹F has high specificity for bacteria and is robustly internalized by bacteria. a, MH¹⁹F has high specificity for bacteria over hepatocytes. E. coli (EC), EC with LamB mutation (LamB) and mammalian cells were incubated with 500 μM MH¹⁹F for 1 hour in the presence or absence of 50 mM maltohexaose (MH). The intracellular MH¹⁹F concentration was determined and normalized to protein content. Bacteria robustly accumulate MH¹⁹F whereas hepatocytes have negligible uptake. The uptake of MH¹⁹F in EC is inhibited by a large excess of maltohexaose, and the uptake of MH¹⁹F in LamB mutants is significantly reduced. The results are expressed as mean micromoles per gram of protein ± s.e.m. for n = 3 per group. b, The accumulation of MH¹⁹F in EC reaches millimolar concentrations. EC were incubated with 500 μM MH¹⁹F, and the intracellular concentration of MH¹⁹F was determined at different time points, n = 3 per group.

We investigated the ability of MH¹⁸F to image bacterial infections in rats. E.coli (10⁷ CFUs) were injected into the left triceps muscle of rats, and the right triceps muscle was injected with PBS as a control. Two hours later, the rats were injected with 250 μCi of MH¹⁸F via the tail vein, and dynamic PET scans were performed using an Inveon micro PET/CT Preclinical Scanner (Siemens). Figures 2a and b demonstrates that MH¹⁸F clears well from healthy tissue but is retained in infected muscle. For example, bacterial infections were clearly visible as early as 10 min after MH¹⁸F injection and after seventy minutes had a high target-to-control contrast of 8.5, allowing bacterial infections to be easily visualized in vivo.

Figure 2.

In vivo PET imaging of rats infected with E.coli (10⁷CFUs). a. Rats were infected in the left triceps muscle with 10⁷ E.coli, injected with MH¹⁸F, and dynamic PET scans were performed for 90 minutes using a microPET/CT. Infected muscles can be easily visualized after 90 min. b. Time activity curves of decay-corrected MH¹⁸F activity in the infected rat, generated from Figure 2a. Infected muscle has an 8.5 fold increase in radioactivity over PBS injected muscle. Arrows indicate the location of infected muscle (EC), PBS injected muscle (PBS) and healthy tissue (HT).

A key challenge in imaging bacteria is developing probes that have high sensitivity for bacteria.^[21] Current PET tracers for imaging bacteria, such as ¹⁸FDG and radiolabeled antibiotics and antibodies, can only image 10⁷-10⁹ bacterial CFUs in vivo, and cannot detect infections at an early stage.^[22] MH¹⁸F has the potential to detect small numbers of bacteria because of its fast transport into bacteria and its rapid clearance from un-infected tissues. We investigated the ability of MH¹⁸F to image early stage bacterial infections. E.coli (10⁵ CFUs) were injected into the left triceps muscle of rats and imaged with MH¹⁸F as described above. Figure 3b demonstrates that MH¹⁸F is capable of detecting as few as 10⁵ bacterial CFUs in vivo, for example, rat triceps muscles infected with 10⁵ bacterial CFUs had a 2.7 fold increase in radioactivity over un-infected controls. Thus, MH¹⁸F's unique combination of robust transport into bacteria and clearance from healthy tissues, allows it to image bacteria with high sensitivity.

Figure 3.

MH¹⁸F can detect as few as 10⁵ CFUs of E.coli (EC) in muscle infections. a1, MH¹⁸F can detect 10⁷ E.coli CFUs in rats. Rats were infected with 10⁷ E.coli and imaged with MH¹⁸F using a microPET/CT. The rat image is a representative result of four experiments, and identifies the infection site. a2, MH¹⁸F generates a 6 fold increase in radioactivity in infected muscles. b1, MH¹⁸F can detect as few as 10⁵ E.coli in rats. Rats were infected with 10⁵ E.coli CFUs and imaged with MH¹⁸F using a microPET/CT. The rat image is a representative result of four experiments, and identifies the infection site. b2, MH¹⁸F generates a 2.7 fold increase in radioactivity in infected muscles. Regions of interest (ROIs) including the infected muscles (target) or PBS injection areas (control) and healthy tissues (background) were identified and integrated using ASI Pro VM™ micro PET analysis software. The results in a2 and b2 are expressed as the target or control to background ratio (ROI ratio) ± s.e.m. for n = 4 per group. The ROI ratio is defined as the mean radioactivity in the target/the mean radioactivity in the background. The statistical significances in a2 and b2 were determined using a two-sample Student's t-test (*p ≤ 0.05 and ***p ≤ 0.001).

¹⁸FDG is currently the only PET radiopharmaceutical available for imaging bacterial infections; however, ¹⁸FDG has significant limitations due to its high uptake in mammalian cells.^[23] To determine the translational potential of MH¹⁸F, a biodistribution study was performed with MH¹⁸F and ¹⁸FDG to compare their specificity for bacteria and non-specific adsorption in healthy tissues. Rats were infected with 10⁹ colony forming units (CFUs) of E.coli and intravenously injected with either MH¹⁸F or ¹⁸FDG. After one hour post administration, the various organs were harvested and their radioactivty was measured. Figure 4 demonstrates that MH¹⁸F is specific for bacteria and has excellent clearance from healthy tissues. For example, MH¹⁸F generated a 30 fold difference in accumulation between infected muscles versus healthy muscles, and in contrast, ¹⁸FDG generated only a 1.5 fold difference. The improved biodistribution pattern of MH¹⁸F over ¹⁸FDG is due to the exclusive expression of maltohexaose transporters in bacteria in contrast to the high expression of glucose transporters in mammalian cells.^{[12c, 24]} This allowed MH¹⁸F to clear from all of the major organs including heart, lung, brain, liver, bone and muscle, whereas ¹⁸FDG had significant accumulation within these tissues. For example, in infected rats, the ratio of accumulation of MH¹⁸F in infected muscle versus liver was 5, whereas for ¹⁸FDG, this ratio was only 0.3, and for other reported PET contrast agents the infected muscle to liver ratio is also generally less than 1.^{[1a, 7-8]} The excellent clearance of MH¹⁸F allowed it to target bacteria much better than ¹⁸FDG, and MH¹⁸F therefore has the potential to image bacterial infections in a variety of anatomical areas.

Figure 4.

MH¹⁸F is more effective than ¹⁸FDG at imaging bacterial infections. A biodistribution study was performed with either MH¹⁸F or ¹⁸FDG in rats infected with 10⁹ E.coli CFUs. MH¹⁸F is efficiently cleared from un-infected tissues, whereas ¹⁸FDG has significant accumulation within the major organs. The results are expressed as % injected dose/gram tissue ± s.e.m. for n = 4 per group. Statistical significance was determined using a two-sample Student's t-test (*p ≤ 0.05 and ***p ≤ 0.001).

At present, there is no direct method available to monitor the efficacy of antibiotic treatment, and doctors therefore have to rely on non-specific and imprecise clinical indicators to guide antibiotic therapy.^[25] MH¹⁸F has the potential to image bacterial drug resistance because it targets ATP binding cassette (ABC) transporters,^[26] which require ATP for internalizing their substrates, connecting MH¹⁸F's uptake with cellular metabolism and bacterial viability. We therefore investigated if MH¹⁸F could distinguish between live versus dead bacteria and identify resistance to therapy. We first performed PET imaging with MH¹⁸F and ¹⁸FDG, and compared their ability to monitor bacterial metabolic activity in vivo. Rats were injected with 10⁹ CFUs of live E.coli in their left triceps and 10⁹ CFUs of metabolically inactive E.coli (sodium azide treated) in their right triceps. Two hours later, the rats were injected with 250 μCi of either MH¹⁸F or ¹⁸FDG via the tail vein, and imaged using an Inveon micro PET/CT Scanner. Figure 5a shows that MH¹⁸F can distinguish between live versus metabolically inactive bacteria, for example metabolically active E.coli had a 7 fold increase in relative radioactivity over sodium azide treated metabolically inactive bacteria, demonstrating that MH¹⁸F is being actively transported by bacteria in vivo. In contrast, Figure 5b shows that ¹⁸FDG could not distinguish between live versus dead bacteria, due to its high uptake by inflammatory cells.

Figure 5.

MH¹⁸F can distinguish between live versus dead bacteria and can discriminate infections from inflammation. a1, MH¹⁸F can distinguish between live versus dead bacteria in vivo. Rats were infected with 10⁹ live and dead E.coli and imaged with MH¹⁸F using a microPET/CT. The rat image is a representative result of four experiments, and demonstrates that MH¹⁸F does not accumulate in dead bacteria. a2, E.coli infected tissues had a 7 fold increase in radioactivity over muscles treated with dead bacteria. b1, ¹⁸FDG cannot distinguish between live and dead E.coli infected tissues. Rats were infected with 10⁹ live and dead E.coli CFUs and imaged with ¹⁸FDG using a microPET/CT. The image is a representative result of four experiments, and demonstrates that ¹⁸FDG cannot discriminate live bacteria from dead bacteria. b2, ¹⁸FDG accumulates in both live and dead bacteria infected tissues. ROIs including the infected muscles (target) and healthy tissues (background) from a1 and b1were identified and integrated using ASI Pro VM™ micro PET analysis software. The results in a2 and b2 are expressed as ROI ratio ± s.e.m. for n = 4 per group. The ROI ratio is defined as the mean radioactivity in the target/the mean radioactivity in the background. The statistical significance in a2 was determined using a two-sample Student's t-test (***p ≤ 0.001).

Based on these results we investigated if MH¹⁸F could identify bacterial drug resistance in vivo and measure antibiotic efficacy. Rats were infected with ampicillin-resistant E.coli (10⁹ CFUs) and wild-type E.coli (10⁹ CFUs), treated with ampicillin and imaged with MH¹⁸F. Figure 6a demonstrates that MH¹⁸F can measure the efficacy of antibiotics in vivo and rapidly identify drug resistance. For example, ampicillin-resistant E.coli generated an 8.2 fold increase in PET signal intensity over susceptible E.coli, due to their increased survival under antibiotic treatment. In addition, we investigated if MH¹⁸F could monitor the treatment of ampicillin resistant bacteria with ciprofloxacin, and be used as a real time methodology to assess antibiotic efficacy. Figure 6b demonstrates that in rats treated with ciprofloxacin, both ampicillin-resistant E.coli (10⁹ CFUs) and wild-type E.coli (10⁹ CFUs) infected tissues have very low accumulation of MH¹⁸F, indicating that MH¹⁸F can quantify the effects of antibiotics, and can be used to guide the selection of antibiotics.

Figure 6.

MH¹⁸F can measure drug resistance and monitor the therapeutic effect of antibiotics in vivo. a1, MH¹⁸F can identify drug resistance in bacteria in vivo. Rats were infected with 10⁹ CFUs of ampicillin-resistant E.coli (DR EC) and wild-type E.coli (EC), treated with ampicillin and imaged with MH¹⁸F using a microPET/CT. The rat image is a representative result of four experiments, and demonstrates that MH¹⁸F only accumulates in DR EC infected muscles. a2, DR EC generated a 10 fold increase in radioactivity over EC. b1, MH¹⁸F can monitor the therapeutic effect of antibiotics. Rats were infected with DR EC and EC, treated with ciprofloxacin and imaged with MH¹⁸F using a microPET/CT. The rat image is a representative result of four experiments, and demonstrates that both DR EC and EC infected muscles have weak accumulation of MH¹⁸F. a2, Both infected tissues have weak radioactivity. ROIs including the infected muscles (target) and healthy tissues (background) from a1 and b1 were identified and integrated using ASI Pro VM™ micro PET analysis software. The results in a2 and b2 are expressed as ROI ratio ± s.e.m. for n = 4 per group. The ROI ratio is defined as the mean radioactivity in the target/the mean radioactivity in the background. The statistical significance in a2 was determined using a two-sample Student's t-test (***p ≤ 0.001).

In conclusion, in this report we present a bacterial targeted PET tracer, termed MH¹⁸F, which can image bacteria in vivo with sensitivity and specificity that is orders of magnitude better than previously reported PET tracers. MH¹⁸F can also identify drug resistance and can therefore potentially assist physicians in prescribing antibiotics. Finally, MH¹⁸F can be synthesized in one radiochemical step from clinically available K¹⁸F, and therefore has the potential to rapidly enter into clinical trials.