A reactivity-based [18F]FDG probe for in vivo formaldehyde imaging using positron emission tomography† †Electronic supplementary information (ESI) available: Synthesis and characterization of probes, animal experiments, and supporting figures. See DOI: 10.1039/c6sc01503d

We present an aza-Cope-based reactivity probe for imaging formaldehyde in vivo using positron emission tomography.


Introduction
Reaction-based chemical probes for selective and non-invasive molecular imaging of biologically important species have attracted signicant attention. By utilizing biocompatible chemical transformations, a variety of small-molecule reagents have been developed to detect a diverse range of analytes in living systems. [1][2][3][4][5] Among the many non-invasive molecular imaging techniques, uorescence is currently the most well studied modality, particularly at the cellular level, owing to its high spatiotemporal resolution, high sensitivity, relative simplicity and the widespread use of confocal and other light microscopy. However, in part because of relatively poor tissue penetration, in vivo imaging with the uorescence modality has had limited clinical translation compared to positron emission tomography (PET), which has been widely applied to oncology, neurology, cardiology and pharmacokinetic studies. 6 As such, new chemical strategies for designing functional PET imaging agents for in vivo use are of interest, and in this context, reaction-based PET probes remain largely underdeveloped compared to radiolabeled ligands for receptors and other biomolecular targets.
One design strategy for bioanalyte sensing using PET relies on caging a clinically utilized PET tracer, as an analogy to reaction-based uorescent probes that uncage useful dyes for light microscopy. In the presence of a specic bioanalyte, the caged species is degraded to the parent tracer, which can subsequently be trapped and accumulated in adjacent cells. We have recently employed this approach with success for PETbased monitoring of hydrogen peroxide 7 and acidic pH. 8 In view of the synthetic ease and wide availability of [ 18 F]uorodeoxyglucose ( 18 F-FDG), the most commonly used PET tracer, we decided to pursue 18 F-FDG as a general platform for developing reaction-based PET probes. In particular, we recognized that 18 F-FDG could be thought of as a latent masked aldehyde and reasoned that the aldehyde group of this open-chain form of 18 F-FDG could be converted to a reactive trigger through suitable chemical modication, which can selectively respond to the bioanalytes of interest and release parent 18 F-FDG. Thus, the engineered 18 F-FDG could be used as a reaction-based PET probe (Scheme 1).
To illustrate this concept with a representative example, we targeted the detection of formaldehyde (FA), a reactive carbonyl species (RCS) involved in a diverse array of processes related to human health and disease. Commonly used as a reagent for tissue preservation owing to its protein cross-linking ability, 9 FA is a known carcinogen 10 and has been associated with neurotoxicity and acute respiratory illness. 11 FA is produced endogenously in the body by demethylation of histones, DNA/RNA, and various metabolites, mediated by enzymes including semicarbazidesensitive amine oxidase (SSAO), 12,13 lysine-specic demethylase 1 (LSD1), 14 and JmjC domain-containing histone demethylases (JHDM). 15,16 More recent studies show that FA is essential for normal brain function, modulating DNA demethylation/methylation events that are critical for memory formation. [17][18][19] FA homeostasis is maintained by the continuous action of FA-metabolizing enzymes, including mitochondrial ALDH2 and cytosolic ADH3, 20,21 resulting in FA concentrations in healthy individuals ranging from 70 mM in blood to 200 mM in brain. 11,19 However, elevation of formaldehyde-generating enzymes has been associated with many types of disease, including Alzheimer's disease, 22,23 multiple sclerosis, 24 heart disease, 25 diabetes 26 and different types of cancer. [27][28][29][30] Indeed, FA levels reaching 700-1000 mM are observed in malignant tissues. 31 These far-ranging roles of FA in healthy and diseased states motivate the development of new technologies for monitoring its spatial and temporal distributions in living systems. However, traditional methods for FA detection require sample processing and/or destruction including colorimetric assays, 32 radiometry, 33 HPLC 34,35 and gas chromatography. [36][37][38] Several uorescent probes based on imine formation have been developed for detecting reactive aldehydes. [39][40][41] As a rst step to tracking FA in living samples, we and others have recently reported FA-responsive uorescent probes based on aza-Cope reactivity that are selective for FA. 42,43 Since then, some other uorescent probes have been developed that can selectively image formaldehyde. [44][45][46] With the goal of creating FA probes with potential for in vivo translation, we turned our attention to PET as a noninvasive nuclear medicine imaging modality. We now report the design, synthesis and application of formaldehyde-caged-[ 18

Response and selectivity
With these probes in hand, we then evaluated the reactivity of [ 18 F]FAC-FDG-1 with FA and a variety of reactive carbonyl species (RCS) by monitoring its conversion to [ 18 F]FDG using radio-HPLC (Fig. 2) (Fig. 2). [ 18 F]FAC-FDG-1 shows a small response to superphysiological level (1000 mM) of methylglyoxal, but is not responsive to 10 mM of this RCS, which is above its single-digit micromolar physiological range. 56

Cellular FA detection with [ 18 F]FAC-FDG-1
We next tested whether [ 18 F]FAC-FDG-1 could respond to changes in FA levels using PC3 prostate cancer and U87-MG glioblastoma cells, as these cell lines exhibit high FDG avidity. [ 18 F]FAC-FDG-1 responses to added FA concentrations ranging from 0-1000 mM showed a FA dose-dependent (Fig. 3a) and time-dependent accumulation in cells (Fig. 3b), with a 4.4 fold increase in signal from 1.3 AE 0.2% cell associated activity at 0 mM FA to 5.7 AE 0.4% cell associated activity at 1000 mM FA at 1 h. Similarly, in U87-MG cancer cells, a 5.5-fold increase in signal was observed ( Fig. S2 and S3 †). The ctrl experiments showed that uptake of FDG in the same cell lines was not affected by varying FA concentrations (Fig. S5 †). Moreover, [ 18 F] Ctrl-FAC-FDG-1 did not exhibit a signicant change in accumulation at 1 h AE 1 1000 mM FA (Fig. S4 †). Also, at 1 h with 1000 mM FA, cell uptake of [ 18 F]FAC-FDG-1 is effectively blocked by the addition of cytochalasin B, 57 a known GLUT inhibitor, suggesting that [ 18 F] accumulation occurs by GLUT-dependent transport. These data suggest that [ 18 F]FAC-FDG-1 reacts with FA mainly via an extracellular process and the resulting [ 18 F] FDG is transported intracellularly by GLUT and is then trapped by hexokinase.
In vivo imaging of FA Finally, we evaluated the ability of [ 18 F]FAC-FDG-1 to image changes in FA levels in vivo using a murine cancer model. Specically, [ 18 F] PET imaging was performed 7-8 weeks following implantation of PC3-derived xenogra tumors on the anks of nu/nu mice. [ 18 F]FAC-FDG-1 shows detectable uptake within the PC3-derived tumor as revealed by [ 18 F] imaging in living mice (Fig. 4a), and the signal increases upon intratumoral injection of FA (Fig. 4b). As anticipated, the control probe [ 18 F] Ctrl-FAC-FDG-1 does not exhibit signicant uptake within tumor, with only hepatobiliary and renal clearance observed (Fig. 4c)

Conclusions
To close, we have presented the design, synthesis, and cellular and in vivo properties of [ 18 F]FAC-FDG-1, a unique reactivitybased PET probe for selective imaging of FA in living animals. [ 18 F]FAC-FDG-1 reacts with FA via a 2-aza-Cope rearrangement to uncage the clinically-used PET tracer [ 18 F]FDG in a FAdependent manner, allowing for detection of changes of FA in living cells and animals most likely via an extracellular pathway. While we are encouraged by these proof-of-principle results, potential limitations may include the short 18 F lifetime vs. FA uncaging as well as the short circulation time of the probe, and, therefore, future improvements will seek to improve probe kinetics by tuning the reactive trigger and optimize the pharmacokinetic properties of the probe. Current efforts are underway to apply [ 18 F]FAC-FDG-1 and related reactivity-based imaging probes to various preclinical models, with particular interest in the epigenetic modications seen in cancer and neurodegeneration. 55,58,59 The use of aldehyde-caged [ 18 F]FDG tracers provides a general synthetic platform for the potential design of a wide variety of responsive molecular imaging probes.