Imaging of Pseudomonas aeruginosa infection with Ga-68 labelled pyoverdine for positron emission tomography

Pseudomonas aeruginosa is an increasingly prevalent opportunistic pathogen that causes a variety of life-threatening nosocomial infections. Novel strategies for the development of new antibacterial treatments as well as diagnostic tools are needed. One of the novel diagnostic strategies for the detection of infection could be the utilization of siderophores. Siderophores are low-molecular-weight chelators produced by microbes to scavenge essential iron. Replacing iron in siderophores by suitable radiometals, such as Ga-68 for positron emission tomography (PET) imaging, opens approaches for targeted imaging of infection. Here we report on pyoverdine PAO1 (PVD-PAO1), a siderophore produced by P. aeruginosa, labelled with Ga-68 for specific imaging of Pseudomonas infections. PVD-PAO1 was labelled with Ga-68 with high radiochemical purity. The resulting complex showed hydrophilic properties, low protein binding and high stability in human serum. In vitro uptake of 68Ga-PVD-PAO1 was highly dependent on the type of microbial culture. In normal mice 68Ga-PVD-PAO1 showed rapid pharmacokinetics with urinary excretion. PET imaging in infected animals displayed specific accumulation of 68Ga-PVD-PAO1 in infected tissues and better distribution than clinically used 18F-fluorodeoxyglucose (18F-FDG) and 68Ga-citrate. Ga-68 labelled pyoverdine PAO1 seems to be a promising agent for imaging of P. aeruginosa infections by means of PET.


Results
Radiolabelling and in vitro characterization. Pyoverdine PAO1 (PVD-PAO1) ( Fig. 1) was labelled with Ga-68 with high radiochemical purity (>95%). The resulting complex showed hydrophilic properties (log P = −3.07 ± 0.08) with low values of protein binding even 120 min after incubation in human serum (<3%). 68 Ga-PVD-PAO1 displayed excellent stability in human serum (~95%) as well as in DTPA solution (~90%) and rapid in vitro instability in highly concentrated FeCl 3 solution. A summary of the analytical data and in vitro characteristics of 68 Ga-PVD-PAO1 is given in Supplementary Fig. S1 and Supplementary Table S1.
In vitro uptake of 68 Ga-PVD-PAO1 by microbial cultures. Uptake of 68 Ga-PVD-PAO1 by P. aeruginosa was highly dependent on the bacterial culture conditions. Iron-deficient but not iron-sufficient cultures displayed high uptake that could be blocked with an excess (10 µM) of cold iron PVD-PAO1 and partly with NaN 3 , indicating a specific and energy-dependent uptake mechanism ( Fig. 2A). Uptake was observed up to 90 min after incubation without saturation, again sensitive to blocking with Fe-PVD-PAO1 (Fig. 2B). Uptake study of different siderophores not produced by P. aeruginosa radiolabelled with Ga-68 in iron-deficient P. aeruginosa cultures showed negligible uptake of these compounds compared to 68 Ga-PVD-PAO1 (Fig. 3A). Figure 3B summarizes the uptake of 68 Ga-PVD-PAO1 in different microorganisms. Under iron-deficient conditions 68 Ga-PVD-PAO1 showed much higher uptake by P. aeruginosa compared to other tested microbial cultures in which minor or no uptake was observed, indicating high specificity of 68 Ga-PVD-PAO1 for P. aeruginosa.

Ex vivo biodistribution in mice and rats.
In non-infected Balb/c mice ( Supplementary Fig. S2A), 68 Ga-PVD-PAO1 displayed rapid excretion via the renal system and showed minimal retention in blood and other organs, even at short times (30 and 90 min) after injection. Biodistribution of 68 Ga-PVD-PAO1 in intratracheally infected and non-infected Lewis rats 45 min p.i. (Fig. 4) did not show any significant difference in organ uptake, except for lung uptake (1.35 ± 0.23%ID/g in infected rats; 0.23 ± 0.03%ID/g in non-infected rats; P < 0.005).
PET/CT imaging in mice and rats. MicroPET/CT imaging of Balb/c mice injected with 68 Ga-PVD-PAO1 confirmed the data from ex vivo biodistribution studies. 68 Ga-PVD-PAO1 was rapidly cleared from the bloodstream with the major excretion route via kidneys ( Supplementary Fig. S2B). PET/CT imaging in the rat respiratory infection model showed focal accumulation of 68 Ga-PVD-PAO1 in the lung (Fig. 5B). No uptake in the lung region was detected in non-infected rats in which the only visible organs were the kidneys and bladder (Fig. 5A). In vivo specificity of 68 Ga-PVD-PAO1 for Pseudomonas infection was studied in the mouse muscle infection model. PET/CT images summarized on Fig. 6 display specific accumulation of 68 Ga-PVD-PAO1 in P. aeruginosa infection and much better biodistribution compared to radiopharmaceuticals clinically used for infection imaging. 68 Ga-PVD-PAO1 showed not only great in vivo specificity but also high sensitivity enabling to detect Pseudomonas infection in the dose of 5 CFU (Fig. 7).

Discussion
Novel reliable approaches enabling specific, sensitive and early diagnosis as well as rapid monitoring of infectious diseases are constantly searched for also in the field of nuclear medicine imaging 18 . Currently the diagnosis of infection with radiopharmaceuticals is mainly performed with 99m Tc-or 111 In-labelled leukocytes, 67 Ga/ 68 Ga-citrate, 99m Tc-diphosphonates and 18 F-FDG 19 . However, all of these compounds lack the specificity to discriminate among different infectious pathogens and even between infectious and non-infectious inflammation. The limitations of 18 F-FDG and 68 Ga-citrate in terms of pharmacokinetics and specificity were demonstrated in Fig. 6. Recently, various attempts were made to develop specific nuclear imaging agent for infection detection including radiolabelled antimicrobial agents (e.g. peptides, antibiotics), monoclonal antibodies, cytokines, nucleoside analogues (e.g. fialuridine), saccharides (e.g. fluorodeoxysorbitol, maltohexaose, maltose, trehalose), oligomers, polyethyleneglycol liposomes, fluorophores and siderophores 18,[20][21][22] . However, none of these probes was translated to routine clinical practice so far. Imaging agent with clinical translation potential is expected to have the following unique characteristics: (i) high binding affinity to target, (ii) high specificity to target, (iii) high  sensitivity, (iv) high contrast, (v) high in vivo stability, (vi) low immunogenicity and toxicity and (vii) economical and production feasibility 23 . We believe that from the agents currently under investigation, especially siderophores hold the promise to fulfil these criteria.
Siderophores have high affinity to siderophore transporters that are highly upregulated during infection and are not present in human cells 13,24 . The energy-dependent active uptake of siderophores leads to their accumulation in the target tissue 16 . The low molecular mass of siderophores and their high hydrophilicity ensure rapid diffusion from the circulation into infected tissues, but also rapid clearance from non-target tissue and elimination via renal excretion 16,17 . Radiolabelling of siderophores can be achieved easily by replacing Fe 3+ in the siderophore by a suitable radionuclide 15,25 . There is no isotope of iron with suitable properties for imaging in terms of half-life and photon emission, however, Ga 3+ is an isosteric diamagnetic substitute for Fe 3+ and has been used extensively to characterize siderophore complexes 15,22 . Recently, interest in the isotope Ga-68, a positron emitter, has increased tremendously with the establishment of PET as a clinical imaging modality. Ga-68 can be obtained from a 68 Ge/ 68 Ga generator without the requirement of cyclotron installations and with a half-life of 68 minutes exhibits a very low radiation burden to the patient, typically about half of that of F-18, the most widely used isotope for PET. Moreover, some siderophores can also be labelled with another cyclotron produced positron emitter Zr-89 (t 1/2 = 78.4 h), which could allow long-term follow-up of infectious process 25 . Typical example of siderophore that can bind Zr-89 is desferrioxamine (DFO). DFO was approved for medical use in the US in 1968 and was clinically used to counteract iron and aluminium overload under the brand name Desferal 26 . Currently, DFO is the only bifunctional chelator used in the clinic for the Zr-89 labelling of monoclonal antibodies 27 .  Radiolabelled siderophores have been investigated in radiopharmaceutical research already in early 1980s 28,29 and recently were successfully used for the detection of Staphylococcus aureus infection 22 and for preclinical imaging of Invasive Aspergillosis (IA) caused by the fungal pathogen Aspergillus fumigatus 16,17 . Triacetylfusarinine  infected muscles were calculated by PMOD software (PMOD Technologies Ltd., Zurich, Switzerland) and were 6.01 ± 0.70%ID/g for 5 × 10 7 CFU/dose; 4.60 ± 0.58%ID/g for 5 × 10 3 CFU/dose and 0.82 ± 0.22%ID/g for 5 CFU/dose. SCIeNTIFIC REPORtS | (2018) 8:15698 | DOI:10.1038/s41598-018-33895-w C (TAFC), siderophore produced by A. fumigatus, was labelled with Ga-68 with high affinity and stability. 68 Ga-TAFC showed high metabolic stability, favourable pharmacokinetics with rapid renal excretion and high specific uptake in A. fumigatus cultures 16,17,30 . Promising properties of 68 Ga-TAFC were confirmed in imaging studies in a rat IA model that showed high focal uptake in infected lung tissue corresponding to pathological findings seen on CT. These encouraging results have led us to undergo the concept of radiolabelled siderophores for specific infection imaging to test with other pathogens.
In this study we evaluated the potential of radiolabelled Pseudomonas produced siderophore, pyoverdine PAO1, as specific agent for bacterial pathogen P. aeruginosa infection imaging. The basic mechanism of pyoverdine uptake by P. aeruginosa was described in detail in various reviews [31][32][33][34] . The ferric-pyoverdine is specifically recognized and subsequently transported by FpVa, specific outer membrane Siderophore Transporters (SITs, also called siderophore receptors) for iron utilization. The energy for this process is provided by the TonB protein, which can span the entire periplasm and functions in coordination with inner membrane proteins ExbB and ExbD during energy transduction. After transport into the periplasm the ferric-pyoverdine is bound by a binding protein that delivers its cargo via the cognate ATP-binding cassette (ABC) transporter into the cytoplasm, where the iron is removed from the complex. Here we attempted to investigate if the Ga-68 labelled pyoverdine can behave similarly to ferric-pyoverdine complex in P. aeruginosa. 68 Ga-PVD-PAO1 showed hydrophilic properties, low protein binding and high stability in human serum. In vitro assays displayed rapid and high uptake increasing over time (up to 90 min) by P. aeruginosa under iron-deficient conditions, which could be blocked with excess of Fe-PVD-PAO1 or sodium azide. The lower extent of uptake reduction under the addition of sodium azide confirmed the phenomenon that especially in iron-deficient conditions not only transporters are up-regulated, but also siderophore binding proteins, leading to increased cell surface binding of Fe/ 68 Ga-siderophores 15 .
It is known that numerous microorganisms possess specific uptake systems not only for native siderophores, but also for siderophores synthesized exclusively by other pathogens, so called "xenosiderophores" 35 . Therefore, in vitro uptake studies of selected 68 Ga-siderophores in P. aeruginosa and 68 Ga-PVD-PAO1 in different microorganisms were performed. None of tested 68 Ga-siderophores was significantly taken up by P. aeruginosa and the uptake of 68 Ga-PVD-PAO1 was highly specific for P. aeruginosa cultures. In animals, 68 Ga-PVD-PAO1 showed excellent pharmacokinetic properties with rapid renal elimination from non target tissue, selective accumulation in infected tissues and great sensitivity enabling the detection of only five viable cells of P. aeruginosa. Moreover, 68 Ga-PVD-PAO1 displayed better specificity and pharmacokinetic properties than other, clinically used radiopharmaceuticals.

Materials and Methods
Chemicals. All reagents were purchased from commercial sources as analytical grade and used without further purification. Pyoverdine PAO1 (PVD-PAO1) isolated from P. aeruginosa ATCC 15692 and other siderophores in the study were purchased from EMC Microcollections GmbH (Tuebingen, Germany). 68 GaCl 3 was eluted from a 68 Ge/ 68 Ga generator (Eckert & Ziegler Eurotope GmbH, Berlin, Germany) with 0.1 N HCl using the fractionated elution approach. 18  Radiolabelling. Radiolabelling of PVD-PAO1 with Ga-68 was performed as follows: 20 μg of PVD-PAO1 dissolved in water (1 μg/μl) were mixed with 30 μl of sodium acetate (155 mg/ml in water) and 300 μl of generator eluate (10-100 MBq of 68 GaCl 3 ). The reaction mixture (pH 3-4) was incubated at 80 °C for l5 min. After the reaction, 100 μl of sodium acetate was added to increase the pH to 5-6. Radiochemical purity (RCP) of 68 Ga-PVD-PAO1 was analyzed by reversed-phase high-performance liquid chromatography (RP-HPLC) or using instant thin-layer chromatography on silica gel impregnated glass fibres (ITLC-SG).
Siderophores for in vitro uptake study in P. aeruginosa cultures were radiolabelled with Ga-68 and analyzed as described previously 15,25 , using similar conditions as for PVD-PAO1 labelling. The preparation of 68 Ga-citrate for in vivo specificity challenge was reported elsewhere 36 . Briefly, 300 µl of 68 Ge/ 68 Ga generator eluate were mixed with 80 µl of 0.5 M sodium citrate (pH ∼ 5). The reaction mixture was incubated for 15 min at room temperature. was determined by measuring the activity distributed between the column (non-protein-bound) and the eluate (protein-bound) using a γ-counter (2480 Wizard 2 automatic gamma counter; PerkinElmer, Waltham, USA).
Partition coefficient (log P) of 68 Ga-PVD-PAO1 was determined as follows. Radiolabelled pyoverdine A in 0.5 ml phosphate-buffered saline (PBS) pH = 7.4 was added to 0.5 ml octanol and the mixture was vigorously vortexed for 15 min. The aqueous and organic solvents were separated by centrifugation and 50 μl aliquots of both layers were collected and measured in a γ-counter. Log P values were calculated from obtained data (mean of n = 6).
All human samples were derived and processed under general ethical criteria accepted at the University Hospital in Olomouc. The informed consent of human participants was obtained in written for usage of biological sample for research purpose in the future. The experiments with human samples were conducted with the approval of Ethics Committee of the University Hospital and the Faculty of Medicine and Dentistry of Palacky University and in Olomouc, Czech Republic.
Microbial strains and growth conditions. Microbial strains used in the study are listed in Supplementary   Table S2. For in vitro uptake experiments, strains were grown in 10 ml of Yeast Extract-Peptone-Glucose (YPG) medium (1% of each peptone and yeast extract and 2% of glucose) in 25 ml Erlenmeyer flasks closed by cotton stoppers. Flasks were shaken at 90 rpm at 30 °C for 16-24 hours to prepare an iron-limited culture. A control iron-replete culture was prepared at the same conditions except for the YPG medium being supplemented with 3 µM FeSO 4 . Pseudomonas aeruginosa ATCC 15692 and Escherichia coli ATCC 10536 for in vivo experiments were cultivated in Petri dishes containing the solid medium Luria-Bertani (LB; 1% tryptone, 0.5% yeast extract, 1% NaCl, 2% agar) for 24 h at 30 °C. Then, the single colony from the Petri dish was transferred to a 500 ml Erlenmeyer flask containing 100 ml of LB broth and incubated on a shaker (200 rpm) at 30 °C. After 24 h, the cell suspension was centrifuged (8000 g, 10 min, 10 °C) and the pellet was diluted with PBS to reach the final number of viability cells of 10 9 CFU/ml.
In vitro uptake assays in P. aeruginosa cultures. For the monitoring of uptake over time, 68 Ga-PVD-PAO1 was incubated in triplicates with iron-deficient P. aeruginosa culture for 10, 20, 30, 45, 60, and 90 min at 37 °C with or without blocking solution (Fe-PVD-PAO1) in Eppendorf tubes. Incubation was interrupted by 5-min centrifugation at 21000 g, supernatant removal and rapid rinsing with ice-cold Tris buffer. The same procedure was repeated twice. After the last centrifugation and supernatant removal Eppendorf tubes containing the microbial sediment were weighed and counted in a γ-counter. Results were expressed as percentage of applied dose per gram of microbial culture (%AD/g).
For the uptake assays, 68 Ga-PVD-PAO1 was incubated in triplicates with iron-sufficient or iron-deficient P. aeruginosa for 45 min at 37 °C with and without sodium azide or excess of Fe-PVD-PAO1. After the incubation samples were treated as described above. In vitro uptake of different Ga-68 labelled siderophores (triacetylfusarinine C (TAFC), ornibactin (ORNB), ferrioxamine E (FOXE), ferrichrome A (FCHA) and aerobactin (AERO)) was tested in P. aeruginosa under iron-deficient conditions. 68 Ga-siderophores were incubated in P. aeruginosa in triplicates for 45 min at 37 °C. After the incubation samples were treated as described above.
In vitro uptake in various microorganisms. In vitro uptake was studied in microbial iron-deficient cultures listed in Supplementary Table S2. Microbial cultures were incubated in Eppendorf tubes in triplicates with 68 Ga-PVD-PAO1 at 37 °C for 45 min. The uptake was interrupted by 5-min centrifugation at 21000 g. The supernatant was removed and the sediment was disturbed by 1 ml of ice-cold Tris buffer and subsequent whirling. The same procedure was repeated twice. After the last centrifugation and supernatant removal Eppendorf tubes containing the microbial sediment were weighed and counted in a γ-counter. Results were expressed as percentage of applied dose per gram of microbial culture (%AD/g). Ex vivo biodistribution in mice and rats. Non-infected mice as well as P. aeruginosa infected and non-infected rats were retro-orbitally (r.o.) injected with 68 Ga-PVD-PAO1 (1-2 MBq and 0.5-1 µg of pyoverdine PAO1 per mouse or rat). Mice were sacrificed by cervical dislocation 30 and 90 min after injection, while rats were killed by exsanguination 45 min post-injection (p.i.). Organs and tissues of interest (blood, spleen, pancreas, stomach, intestines, kidneys, liver, heart, lung, muscle and femur) were removed and weighed. The amount of radioactivity in the samples was measured in a γ-counter. Results were expressed as percentage of injected dose per gram of organ (%ID/g).

Animal infection models.
In vivo uptake of radiolabelled tracers was studied in acute respiratory 37  by intramuscular (i.m.) injection of P. aeruginosa ATCC 15692 (5 to 5 × 10 7 CFU/dose in 50 µl) in mice 5 hours before PET/CT imaging. For in vivo specificity challenge the infection was induced in the left hind muscle with P. aeruginosa (5 × 10 7 CFU/dose in 50 µl 5 h prior imaging) and in the right hind muscle with E. coli ATCC 10536 (5 × 10 7 CFU/dose in 50 µl 5 h prior imaging) or with turpentine oil (50 µl 24 h prior imaging) in mice.
Animal imaging studies. MicroPET and CT images were acquired with an Albira PET/SPECT/CT small animal imaging system (Bruker Biospin Corporation, Woodbridge, CT, USA). Mice and rats were r.o. injected with radiolabelled tracer in a dose of 5-10 MBq corresponding to 1-2 μg of pyoverdine PAO1 per animal. Anaesthetized animals were placed in a prone position in the Albira system before the start of imaging. Static PET/CT images were acquired over 40 min starting 30 and 90 min after injection for normal biodistribution studies and 45 min post-injection for infection models imaging. A 10-min PET scan (axial FOV 148 mm) was performed, followed by a triple CT scan (axial FOV 65 mm, 45 kVp, 400 μA, at 400 projections). Scans were reconstructed with the Albira software (Bruker Biospin Corporation, Woodbridge, CT, USA) using the maximum likelihood expectation maximization (MLEM) and filtered backprojection (FBP) algorithms. After reconstruction, acquired data was viewed and analyzed with PMOD software (PMOD Technologies Ltd., Zurich, Switzerland). 3D volume rendered images were obtained using VolView software (Kitware, Clifton Park, NY, USA).
Statistical analysis. Student's t-test (level of significance, P < 0.005) was used to determine the significance of differences in the ex vivo biodistribution data of infected and non-infected rats. Analysis was performed using Microsoft Office Excel 2007.

Conclusion
We have shown that 68 Ga-PVD-PAO1 can be used for the detection of P. aeruginosa infection with high specificity and sensitivity. The high and specific uptake of 68 Ga-PVD-PAO1 by P. aeruginosa was confirmed both in vitro and in vivo, proving the potential of pyoverdine PAO1 for specific imaging of Pseudomonas infections. Biodistribution studies in non-infected animals showed excellent in vivo behaviour of 68 Ga-PVD-PAO1 with rapid renal excretion and no accumulation in any organ. Animal infection models displayed focal uptake of 68 Ga-PVD-PAO1 with high sensitivity in infected tissue and much better pharmacokinetics than clinically used radiopharmaceuticals. Thus, 68 Ga-PVD-PAO1 seems to be a promising new PET agent for specific P. aeruginosa imaging. Moreover, we have proved that the concept of radiolabelled siderophores for specific infection imaging is transferable to different pathogens.

Data Availability
All data generated or analyzed in this study are included in this published article and its supplementary information files.