In vivo imaging of lung inflammation with neutrophil-specific 68Ga nano-radiotracer

In vivo detection and quantification of inflammation is a major goal in molecular imaging. Furthermore, cell-specific detection of inflammation would be a tremendous advantage in the characterization of many diseases. Here, we show how this goal can be achieved through the synergistic combination of nanotechnology and nuclear imaging. One of the most remarkable features of this hybrid approach is the possibility to tailor the pharmacokinetics of the nanomaterial-incorporated biomolecule and radionuclide. A good example of this approach is the covalent binding of a large amount of a neutrophil-specific, hydrophobic peptide on the surface of 68Ga core-doped nanoparticles. This new nano-radiotracer has been used for non-invasive in vivo detection of acute inflammation with very high in vivo labelling efficiency, i.e. a large percentage of labelled neutrophils. Furthermore, we demonstrate that the tracer is neutrophil-specific and yields images of neutrophil recruitment of unprecedented quality. Finally, the nano-radiotracer was successfully detected in chronic inflammation in atherosclerosis-prone ApoE−/− mice after several weeks on a high-fat diet.

Non-invasive quantitative detection of lung inflammation is highly desirable for assessing pathogenic processes in the lung. Inflammatory-cell activation is currently assessed by combining anatomical information obtained by computed tomography with molecular and cellular information obtained from invasive lung biopsy, histopathology and bronchoalveolar lavages 1 . This difficult and time-consuming approach explains the numerous attempts to produce a reliable probe for non-invasive in vivo diagnosis 2,3 . Neutrophils are an essential part of the inflammatory cascade, being the first cell type to migrate from the bloodstream to the site of inflammation. One of their most important functions is to eliminate pathogens by engulfing them and liberating enzymes and reactive oxygen species 4,5 . However, neutrophil invasion can cause major tissue damage in acute processes such as lung injury and in chronic diseases such as chronic obstructive pulmonary disease (COPD) and asthma 6 . This multifaceted role underlies interest in designing an effective tool for non-invasive detection, tracking, and quantification of neutrophils. The most promising approach is molecular imaging with a selective tracer, and several probes have been developed to exploit the superb sensitivity of positron emission tomography (PET) and fluorescence tomography or microscopy 1,[7][8][9] .
A particularly interesting compound is the peptide N-cinnamoyl-F-(D)L-F-(D)L-F (cFLFLF), an antagonist of formyl peptide receptor 1 (FPR-1). The very high binding affinity of cFLFLF (K d = 2 nM) is the main reason for its use in neutrophil-specific probes for PET and optical fluorescence 7,8,10 . However, the high hydrophobicity of cFLFLF produces very poor target-to-background ratios 7 . This serious limitation is the reason why to date the amount of radiotracer reaching the inflammation site is below 1% (expressed as the % of injected dose per mass of the organ) 7,10 .
The combined use of nanotechnology and radiochemistry is an attempt to join the best of both approaches: size-dependent properties and unparalleled sensitivity with high in vivo selectivity. This combination has been successfully used for many applications, particularly cancer diagnosis 11,12 . Recently, we showed how iron oxide nanoparticles can be core-doped with 68 Ga for enhanced molecular imaging 13 . Here, we have developed a modified version of these particles, with citrate molecules instead of dextran as the coating. These 68 Ga-based citrate-coated nano-radiotracers ( 68 Ga-NRT) not only provide signal simultaneously in PET and MRI, but, crucially to our approach, also have a very large and highly hydrophilic surface due to the large organic coating per particle 13,14 . We hypothesized that 68 Ga-NRT particles would overcome the solubility problems encountered with cFLFLF, permitting its use for molecular imaging of neutrophil-driven acute inflammation.
To test our hypothesis, we used a well-characterized model of acute inflammation in mice, based on intratracheal administration of lipopolysaccharide (LPS), which produces neutrophil recruitment in the lungs 24 h post-instillation 15 . PET analysis after intravenous injection of the cFLFLF-functionalized 68 Ga-NRT enabled in vivo detection of neutrophils infiltrating the LPS-treated lungs with very high signal to noise ratios. The 68 Ga-NRT signal was absent from the lungs of LPS-treated neutrophil-depleted mice, while it was present in LPS-treated macrophage-depleted mice, demonstrating the specificity of the tracer for neutrophils versus macrophages. Finally, we tested the tracer in ApoE −/− mice. Several reports show that after a few weeks on a high-fat diet (HFD) these mice develop lung inflammation featuring invasion by macrophages and neutrophils. Here, we demonstrate that the 68 Ga-NRT enables detection of this early-stage condition by in vivo PET imaging.

Results
Synthesis and functionalization of 68 Ga nano-radiotracers. We recently reported the use of 68 Ga core-doped nanoparticles as a novel platform for molecular imaging 13 . Here, we have used citrate instead of dextran as the coating agent. Briefly, a mixture of FeCl 3 , sodium citrate, 68 GaCl 3 (eluted from a 68 Ge/ 68 Ga generator), and hydrazine was heated at 100 °C in a synthesis microwave oven (Fig. 1a). After purification by gel chromatography, we obtained hydrophilic and extremely small nano-radiotracers, 2.7 ± 1.0 nm of core size measured by STEM-HAADF (Fig. S1) and 14.5 ± 2.1 nm hydrodynamic size (HD). The yield of the synthesis was 26.0% Fe and a 92% radiolabelling, for a final specific activity of 7.1 GBq/mmol Fe. Ten independent repetitions of the synthesis on different days demonstrated excellent HD reproducibility (Fig. 1b). The particles show a very thick organic coating, 47% weight measured by thermogravimetry (Fig. S2). The small size and very thick organic layer of these NRT favour excellent colloidal stability. This property is essential for our goal to conjugate a highly hydrophobic peptide, which under other conditions would be very difficult to stabilize. Colloidal stability was demonstrated by the maintenance of hydrodynamic size and the absence of NRT aggregation up to 24 hours after dispersion in PBS, saline and mouse serum (Fig. 1c). Testing of the PET and MRI signals in 68 Ga-NRT phantoms demonstrated successful incorporation of the radioisotope, with the signal increasing in proportion with the amount of NRT (Fig. 1d). As we have already demonstrated, with this synthetic protocol and purification, this is an indication of core-doping with the radioisotope 13 . Positive signal in MRI phantom is in agreement with the data measured by relaxometry for r 1 and r 2 , 6.8 mM −1 s-1 and 15.9 mM −1 s −1 respectively.
After characterizing 68 Ga-NRT performance, we covalently attached the neutrophil-specific peptide cFLFLF to citrate carboxyl groups by the classical EDC/Sulfo-NHS reaction, yielding 68 Ga-NRT-cFLFLF. Conjugation of cFLFLF increased NRT hydrodynamic size from 14.5 nm to 83.3 ± 10.5 nm (Fig. 1e). Despite the increased hydrodynamic size, we did not observe aggregation upon storage or in the in vivo experiments; TEM images show a core size of 2.5 ± 0.5 nm, also with no sign of aggregation (Fig. 1f). This increase in hydrodynamic size, but not core size, can therefore be attributed to the large number of highly hydrophobic peptides attached to the NRT surface. Successful peptide attachment was confirmed by the presence of bands on FTIR at 2913 cm −1 (aromatic C-H), 996 cm −1 (olefin) and 770 cm −1 (benzene ring) (Fig. 1g). Further evidence of peptide attachment is provided by the zeta potential, which changed from a mean value of −31.5 mV for 68 Ga-NRT to −14.6 mV for 68 Ga-NRT-cFLFLF, due to the transformation of some of the negatively charged carboxylate groups into amides (Fig. 1h). The thermogravimetric curve shows the typical profile for this type of nanoparticles, with a very small core and a thick organic coating. At 560 °C, a strong exothermic process further indicates the presence of the peptide, accounting the very high figure of 23% of total NRT mass, corresponding to 1 mmol of peptide per 90 mmol Fe.
Cytotoxicity analysis of 68 Ga-NRT-cFLFLF in primary hematopoietic progenitor cells and mature neutrophils. The potential cytotoxic effects of 68 Ga-NRT-cFLFLF were studied by in vitro colony forming cell (CFC) assay in primary hematopoietic progenitor cells, as highly-proliferative primary cells are particularly sensitive to toxic compounds. To induce the formation of myeloid and erythroid colonies, we cultured freshly isolated cord blood CD34 + cells from healthy donors in methylcellulose medium in the presence of cytokines. These assays were conducted in the absence or presence of different concentrations of 68 Ga-NRT-cFLFLF. The NRT concentrations used in the CFC assays were up to 50 times the expected in vivo-infused NRT dose; 3 μg of Fe, 15 μg of Fe and 150 μg of Fe.
The total number of colonies generated at any 68 Ga-NRT-cFLFLF dose was similar to the number generated in unexposed cultures (Fig. 2a), indicating that 68 Ga-NRT-cFLFLF is not toxic to human hematopoietic progenitor cells even at the highest concentration used. Data show the different values per cord. Values show a large standard deviation, which is expected in cord blood samples 16 . However, it is clear the lack of effect due to the NRT in each cord. Statistical analysis by ANOVA rendered no significant differences between control and experimental groups (P > 0.05, N = 3). In the CFC assays, we distinguished three types of colonies: erythroid burst-forming units (BFU-E), granulocyte-monocyte colony-forming units (CFU-GM), and granulocyte, erythrocyte, monocyte, megakaryocyte colony-forming units (CFU-GEMM). The colony-type profile was not affected by NRT exposure, additionally indicating that lineage commitment and differentiation were unaffected by 68 Ga-NRT-cFLFLF. However, we have to take into account that at the highest concentration of 68 Ga-NRT-cFLFLF, due to the presence of high concentrations of iron, it is difficult to distinguish red coloured erythroid colonies.
Since the cFLFLF peptide targets the formyl peptide receptor-1 (FPR-1) on neutrophils, it was important to determine whether binding of 68 Ga-NRT-cFLFLF altered neutrophil function, in particular the capacity for the oxidative burst in response to pathogens. We therefore differentiated freshly isolated CD34+ cells into neutrophils in vitro. These neutrophils were then subjected to a chemiluminescence oxidative burst assay (Fig. 2b). Also in this case, statistical analysis by ANOVA rendered no significant differences between control and experimental groups (P > 0.05, N = 3). Treatment with 68 Ga-NRT-cFLFLF in the absence of zymosan stimulation induced no response in in vitro differentiated neutrophils, indicating that the tracer by itself is unable to stimulate the oxidative burst. In response to zymosan, the oxidative burst was similar in the absence and presence of 68 Ga-NRT-cFLFLF (P > 0.05, Mann-Whitney non parametric test), indicating that the presence of the tracer does not alter cell function (Fig. 2c).
Inflammation model in C57BL/6 mice. The endotoxin lipopolysaccharide (LPS) is commonly used to experimentally induce acute lung inflammation. In this model, an acute transient inflammation elicits maximum neutrophil recruitment 24 hours after intratracheal LPS administration 17,18 . This approach has been used to model conditions such as chronic obstructive pulmonary disease (COPD) and asthma 19 . LPS (50 µg) was administered by intratracheal instillation to 12-week-old C57BL/6 mice, followed by perfusion and excision of the lungs at 24 h post treatment. Neutrophil accumulation was assessed by flow cytometry of the perfusate. The lungs of LPS-instilled mice had a markedly higher neutrophil content than control mice instilled with PBS, demonstrating LPS-induced neutrophil recruitment (Fig. S3). Histological analysis of lung sections confirmed hyaline membrane formation and neutrophil infiltration, a typical feature of LPS-induced lung injury (Fig. S3).
In vivo detection of neutrophils by PET in LPS-treated model. The efficiency of 68 Ga-NRT-cFLFLF as a neutrophil-specific radiotracer was assessed in vivo PET/CT imaging experiments (Fig. 3). We first conducted two control analyses. In one, 68 Ga-NRT-cFLFLF particles were administered intravenously to healthy C57BL/6 mice, and images were acquired one hour post-injection. No 68 Ga signal was observed in the lungs (Fig. 3a), whereas strong signals appeared in the liver and spleen. This biodistribution is expected for nanoparticles with this size and coating, and demonstrates that the particles do not accumulate passively in the target organ (lung) due to an effect of hydrodynamic size or through non-specific cellular recognition of the peptide. In the second control, three LPS-treated mice received systemic injections of 68 Ga-NRT with no peptide attached (Fig. S4). This NRT control allowed us to demonstrate that the particles do not passively accumulate in the lungs of LPS-treated mice. As we expected, the precursor nanoparticles were removed from the circulation through the normal clearance organs, mainly liver and bladder, due to their small hydrodynamic size, at the limit between renal filtration and hepatic elimination. With these controls completed, we proceeded to the study of 68 Ga-NRT-cFLFLF in mice (N = 5) with LPS-induced lung inflammation. The results show a clear uptake of 68 Ga-NRT-cFLFLF in the lungs together with accumulation in the major reticuloendothelial system organs (Fig. 3b).
The biodistribution data is quite revealing, in agreement with imaging results, showing a very large accumulation in the lungs when 68 Ga-NRT-cFLFLF is used in the pulmonary inflammation model, with a percentage injected dose per gram (%ID/g) of almost 30% compared to 3% in the controls (Fig. 3d). Furthermore, in a different experiment with the same experimental protocol, neutrophils in the lungs were isolated 1 hour post i.v. injection of dye-labeled 68 Ga-NRT-cFLFLF and their labelling calculated mounting for and outstanding 15% of in vivo labelled neutrophils, compared to less than 3% in control experiment (Fig. 3e).
The results up to this point seemed to confirm that our NRT was able to detect the inflammatory process after LPS treatment and that, in this detection, the in vivo labelling of neutrophils was a major contributor. However, to demonstrate selectivity towards neutrophils two further experiments were designed. It is well known the avidity of macrophages for nanosized particles and the important role and accumulation of these cells in inflammatory processes and imaging 20 . Thus, it could be argued that the increased signal intensity in the lungs could be due to pronounced presence of macrophages capturing the NRT after the adsorption of opsonins in systemic circulation, and as result they could not show specificity towards neutrophils. Supporting this contention, some studies have also shown some affinity towards macrophages by the cFLFLF peptide 21,22 . To check this option, first we performed an additional experiment in which LPS-instilled mice were further treated with Ly6G antibody for neutrophils depletion 23 . In this case, without the transient presence of neutrophils, the specific uptake of the neutrophilic targeted nanoprobe in the lungs should be much lower or, ideally, disappear. The complete disappearance of signal uptake in the lungs after neutrophilic depletion in acute inflammatory animals (Fig. 3c) demonstrated the specificity of nanotracers comprising a formyl peptide receptor binding moiety toward this type of leukocytes. Moreover, we found that, after neutrophil depletion, the NRT circulate for longer times in the bloodstream with a clear signal in the heart not displayed previously (Fig. 3c, axial view).
After imaging we measured radiotracer accumulation in isolated organs was measured ex vivo with a gamma counter (Fig. 3f). In agreement with the imaging results, radiotracer accumulation in the lungs revealed a very high accumulation of 68 Ga-NRT-cFLFLF in inflamed lungs, with a 27.6%ID/g in the lungs of LPS-instilled mice compared with 5%ID/g in neutrophil-depleted mice and 3%ID/g in mice injected with non-peptide-conjugated 68 Ga-NRT (Fig. 3f). Consistent with the in vivo and gamma counter results, Prussian blue staining on histological sections after decay of the radioactive signal revealed iron-containing 68 Ga-NRT-cFLFLF particles (blue dots) in the lungs of mice with LPS-induced lung inflammation (Figs 3g and S5) but not those of similarly treated mice depleted of neutrophils (Figs 3h and S5) or of mice without induced neutrophil recruitment (Fig. S5). After this, we performed a second experiment in which LPS-instilled mice were previously treated with clodronate, for macrophage depletion 23 . This treatment transiently depletes macrophages but, importantly, neutrophil recruitment to the site of infection is strongly reduced [24][25][26] . In this model and with a NRT specific for neutrophils, we predicted 68 Ga-NRT-cFLFLF accumulation in the lungs after LPS treatment, but with a reduced uptake compared with macrophage-retaining mice, since the amount of neutrophils reaching the area of inflammation should be lower. The results confirmed our prediction; images (Fig. 4a) and quantification (Fig. 4b) of the uptake in the lungs of macrophages-depleted mice after LPS show robust uptake compared to control mice, but much lower than mice with intact macrophages, in agreement with our hypothesis 26 .

PET detection of neutrophils during chronic inflammation in vivo.
After demonstrating the ability of 68 Ga-NRT-cFLFLF to non-invasively detect neutrophils during acute inflammation, we wanted to test its performance in a mouse model of mild chronic inflammation. Previous studies have shown neutrophilia and monocytosis in ApoE −/− mice fed a high-fat diet (HFD) 27,28 . Moreover, this obesogenic diet induces lung remodelling in ApoE −/− mice, featuring recruitment of monocytes and a small number of neutrophils to the lungs 29 . We carried out imaging experiments in aged ApoE −/− mice (40 weeks old) fed a standard chow diet and in young ApoE −/− mice (16 weeks old) fed the HFD and injected with 68 Ga-NRT-cFLFLF. It has been shown that the older mice would have higher levels of neutrophils than C57BL/6 mice but without any marked alteration in the lungs, whereas the HFD-fed mice would only show modest recruitment of neutrophils to the lungs. Accordingly, the 40-week-old ApoE −/− mice showed a similar biodistribution to controls, with no significant signal in the lungs (Fig. 5a,b). In contrast, the 16-week-old HFD-fed mice showed a strong lung signal (Fig. 5c,d). Ex vivo quantification matched the imaging results, showing a higher overall neutrophil content in ApoE −/− mice than in controls and a clear increase in the lungs after feeding a high-fat diet.

Discussion
Inflammation is a central feature of many clinical conditions and can be broadly classified as acute or chronic. Acute inflammation typically occurs over a timescale of minutes to hours, whereas chronic inflammation can develop over several days to months. The predominant cell type in acute inflammation is the neutrophil, whereas chronic inflammation features a greater variety of cell types. The key role played by inflammation in many diseases places a premium on the development of methods for its in vivo diagnosis, especially methods with specificity towards a particular receptor up-regulated by inflammatory stimuli or for the accumulation of different cell types in the area under study 30 . The interest in neutrophil identification has attracted attention to cFLFLF for the development of molecular imaging probes. This formyl peptide receptor-1 antagonist shows selectivity towards neutrophils with very high binding affinity (K d = 2 nM). However, its use has been limited due to its hydrophobic nature, which has produced poor signals and very low in vivo labelling. We reasoned that this problem could be overcome through the incorporation of cFLFLF into highly hydrophilic nanoplatforms and their detection by nuclear imaging. We used a new generation of nano-radiomaterials in which the radioisotope is incorporated in the nanoparticle core by very fast temperature ramping in a synthesis microwave. This resulted in very high peptide incorporation in the NRT (1 mmol per 90 mmol Fe by TG analysis and FTIR spectroscopy) and a reduction of the zeta potential from −31.5 mV to −14.6 mV. Even with the mild increase hydrodynamic size due to the incorporated peptide, these physiochemical features ensure NRT stability and its successful use in vivo. This is due not only to the large amount of peptide incorporated per nanoparticle, but also and more importantly, to the good value obtained for the specific activity (700 GBq/mmol peptide). The utility of this approach is demonstrated in a model of acute inflammation in the lung, in which the NRT particles selectively accumulate in the lungs of animals treated with LPS, producing very clear images and labelling about 15% of neutrophils in vivo, the highest reported value for this type of approach. This high labelling permits the unambiguous non-invasive identification of acute inflammation in the lungs. Furthermore, the 68 Ga-NRT-cFLFLF tracer shows high selectivity towards neutrophils. Indeed, cFLFLF is highly neutrophil selective; however, some reports also show interaction with macrophages. To check the situation in our model, we depleted LPS-administered mice of neutrophils. Interestingly, the neutrophil depletion method requires the presence of macrophages. Imaging confirmed the loss of the LPS-dependent 68 Ga-NRT-cFLFLF signal in the lungs of neutrophil-depleted mice. Furthermore, in a macrophage-depleted model the signal in the lungs was still statistically significant compared to control mice and images showed clear accumulation, this is particularly remarkable since neutrophil recruitment is hampered in this model.
The utility of 68 Ga-NRT-cFLFLF for the detection of chronic inflammation was evaluated in ApoE −/− mice, which show a mild recruitment of neutrophils that is exacerbated in the lungs upon feeding a high-fat diet for several weeks. Even in the setting of this less pronounced lung remodelling and neutrophil recruitment, the 68 Ga-NRT-cFLFLF radiotracer revealed clear differences between aged and young HFD-fed ApoE −/− mice and between aged ApoE −/− mice fed the HFD and those fed a normal diet. The findings presented here demonstrate how nanotechnology and nuclear imaging can be combined to overcome the limitations of traditional approaches, yielding a new tool for the non-invasive detection of inflammation with in vivo selectivity towards neutrophils. These results pave the way to the non-invasive identification of neutrophils in a number of highly relevant inflammatory disorders.
Synthesis of 68 Ga-NRT. FeCl 3 × 6 H 2 O (75 mg, 0.28 mmol), sodium citrate hydrate (80 mg, 0.27 mmol) and 1280 MBq of 68 GaCl 3 in HCl (0.05 M, 4 mL) were dissolved in water (5 mL) in a microwave-adapted flask, followed by addition of 1 mL hydrazine hydrate. The solution was ramped to 100 °C over 54 s and held at this temperature for 10 minutes (240 W) in a Monowave 300 microwave reactor equipped with an internal temperature probe and an external IR probe (Anton Paar, GmbH73760, Ostfildern-Scharnhausen, Germany). The reaction mixture was then cooled to 60 °C and the 68 Ga-NRT producte was purified by passing the mixture through a PD-10 column to eliminate excess small reagents, including all unincorporated radiotracer. This purification process provided 9 mL of 68 Ga-NRT with a total activity of 781 MBq (measured 40 minutes after starting the reaction), a radiolabelling yield of 92%. In vivo PET/CT imaging in mice was performed with a nanoPET/CT small-animal imaging system (Mediso Medical Imaging Systems, Budapest, Hungary). List-mode PET data acquisition commenced 1 hour after injection of a bolus of 10 MBq of 68 Ga-NRT-cFLFLF through the tail vein and continued for 30 minutes. At the end of PET, microCT was performed for attenuation correction and anatomic reference. The dynamic PET images in a 105 × 105 matrix (frame rates: 3 × 10 min, 1 × 30 min, 1 × 60 min) were reconstructed using a Tera-Tomo 3D iterative algorithm. Images were acquired and reconstructed with proprietary Nucline software (Mediso, Budapest, Hungary). Images were analyzed using Osirix software (Pixmeo, Switzerland).

Ex vivo biodistribution. Biodistribution was studied with a Wizard 1470 gammacounter (Perkin Elmer).
Animals were sacrificed in a CO 2 chamber, after which blood was extracted and the animals perfused with 8 mL PBS. Organs were extracted and counted in the gammacounter for 1 min each. Readings were decay corrected presented as the percentage injected dose per gram (%ID/g).
Purification of cord blood CD34 + cells. Cord Blood samples from healthy donors were obtained from the Madrid Community Transfusion Centre. Mononuclear cells were purified by density gradient centrifugation in Ficoll-Paque PLUS medium (GE Healthcare, Fairfield, USA). CD34 + cells were selected using the CD34 MicroBead Kit. Magnetic-labelled cells were positively selected first with an LS column in a QuadroMACS ™ separator and then with an MS column in an OctoMACS ™ separator (all from MACS, Miltenyi Biotec, Bergisch Gladbach, Germany). FACS analysis routinely revealed a CD34 + purity of 80-95%.