Synthesis-free PET imaging of brown adipose tissue and TSPO via combination of disulfiram and 64CuCl2

PET imaging is a widely applicable but a very expensive technology. On-site synthesis is one important contributor to the high cost. In this report, we demonstrated the feasibility of a synthesis-free method for PET imaging of brown adipose tissue (BAT) and translocator protein 18 kDa (TSPO) via a combination of disulfiram, an FDA approved drug for alcoholism, and 64CuCl2 (termed 64Cu-Dis). In this method, a step-wise injection protocol of 64CuCl2 and disulfiram was used to accomplish the purpose of synthesis-free. Specifically, disulfiram, an inactive 64Cu ligand, was first injected to allow it to metabolize into diethyldithiocarbamate (DDC), a strong 64Cu ligand, which can chelate 64CuCl2 from the following injection to form the actual PET tracer in situ. Our blocking studies, western blot, and tissue histological imaging suggested that the observed BAT contrast was due to 64Cu-Dis binding to TSPO, which was further confirmed as a specific biomarker for BAT imaging using [18F]-F-DPA, a TSPO-specific PET tracer. Our studies, for the first time, demonstrated that TSPO could serve as a potential imaging biomarker for BAT. We believe that our strategy could be extended to other targets while significantly reducing the cost of PET imaging.


Introduction
Positron emission tomography (PET) has been widely used for clinical and preclinical studies, including disease diagnosis, treatment monitoring, and drug development.
Compared to other imaging modalities, such as MRI, ultrasound, and optical imaging, PET is highly sensitive and quantitative (1). However, the high cost of PET imaging has been the key roadblock for its widespread routine use. The high cost originates from expensive radionuclide production and tracer preparation, which must be conducted onsite (1). Here, we demonstrated the feasibility of a synthesis-free PET imaging method for brown adipose tissue (BAT). With this method, the cost of PET imaging could be dramatically reduced, thus allowing for a widespread application of this technology.
For the purpose of synthesis-free PET imaging, we considered the following criteria: 1) the radionuclide should have a suitable lifetime for delivery and transportation; 2) synthetic components (excluding the radionuclide) should be available in a convenient kit; and 3) the generation of the actual tracers should be very fast or the actual tracers should be formed in vivo. To meet the above requirements, we selected 64 Cu as the radionuclide due to its reasonable decay lifetime (12.7 hours) and widespread availability. For our proof-of-principle studies we selected BAT as the biological target, because BAT has the following unique features making it a suitable imaging model (2, 3): 1) BAT in mice is situated away from large organs such as liver, heart, and stomach, and thus signal interference from these large organs is minimal; 2) BAT is a whole mass organ; 3) BAT has a unique triangular physical shape which is easy to distinguish from other tissues.
BAT is a specialized tissue for thermogenesis in mammals, whose function is to dissipate large amounts of chemical/food energy as heat, thus maintaining the energy balance of the whole body (9)(10)(11). In spite of the fact that investigations of BAT have been ongoing for 70 years, it had been assumed that BAT disappears from the body of adults and has no significant physiological relevance in adult humans (9,(12)(13)(14). This "non-existence" assumption is partially due to the lack of proper imaging methods to "see" the small BAT depots in vivo, as only 3%-8% of BAT depots in adults could be clearly visualized with [18F]-FDG (the most used imaging method) if no cold or drug stimulation is applied (6,(15)(16)(17). However, under stimulated conditions, [18F]-FDG PET imaging has shown that BAT is still present in 95% of healthy adults in the upper chest, neck, and other locations (4)(5)(6). This remarkably large difference between unstimulated and stimulated conditions strongly indicates that [18F]-FDG PET imaging primarily reflects the activation of BAT, but not BAT mass. Various other imaging methods for BAT are available for preclinical and clinical studies, however most of them are dependent on BAT activation (18)(19)(20)(21)(22)(23)(24)(25)(26)(27).
Therefore, an imaging probe that can consistently report on BAT mass is highly desirable.
In this report, we first conducted a top-down screening, and found that the combination of 64 CuCl 2 and Disulfiram (termed 64 Cu-Dis) provided high contrast for BAT, which was not affected by BAT activation. We also found that TSPO, a transport protein located on the outer mitochondrial membrane (28), was the binding target of 64 Cu-Dis. We further validated that TSPO is an excellent but unexpected imaging biomarker for BAT using Western blot, histology, and PET imaging with TSPO-specific ligand [18F]-F-DPA.

Screening for BAT-binding ligands
In our previous report (3), we have demonstrated that a top-down screening approach could be used for seeking near infrared fluorescence (NIRF) imaging probes for BAT. In that study, we screened 38 NIRF dyes resulting in two hits that we further optimized for high BAT selectivity (3). In the present report, we used a similar top-down strategy for fast screening of a library of copper ligands, which could be used for fast coordination chemistry with no need for purification. Among the 16 screened ligands, four compounds were considered as positive hits (SI Fig.1), including Disulfiram, diethyldithiocarbamate (DDC, a metabolic product of Disulfiram in vivo (29)), cysteamine, and salicylaldoxime.

PET imaging
One of the lead ligands, Disulfiram, caught our attention, because it is an FDA approved drug for alcoholism (30). Disulfiram, a disulfide compound, does not have a strong affinity to copper. However, when injected in vivo, it is reduced to the monomer and releases the thiol group producing diethyldithiocarbamate (DDC), which has a high affinity for copper (II) (31). Considering that DDC, a metabolic product of Disulfiram, is an active copper chelator, we proposed a step-wise injection strategy to realize synthesisfree PET imaging (Fig.1a). To this end, we injected Disulfiram intraperitoneally (40 mg/kg) in mice and allowed 60 minutes for it to metabolize into DDC, the active copper binding form. Next, we intravenously injected 64 CuCl 2 , which can quickly chelate with DDC in vivo. Strikingly, we found that BAT could be easily identified by PET imaging as early as 10 minutes (Fig.1b), and as late as 48 hours after injections. The uptake of the tracer at 30 minutes after 64 CuCl 2 injection was 10.6 %ID/g (Fig.1c). More importantly, no apparent signal could be observed from WAT (white adipose tissue), including inguinal and epidermal areas ( Fig.1c and SI Fig.2), indicating that this method has high BAT selectivity over WAT.
Attaining high BAT/WAT selectivity has been one of the challenges for developing imaging probes for BAT. To confirm the data on BAT/WAT selectivity obtained from in vivo PET imaging, we conducted ex vivo bio-distribution of BAT and WAT tissues 6 hours after 64 CuCl 2 injection. We found that interscapular BAT uptake was 13.1 %ID/g, which was 10-fold higher than that of interscapular WAT, inguinal WAT, or gonadal WAT. (Fig.1d). Taken together, our in vivo imaging data were consistent with the ex vivo data, suggesting that 64 Cu-Dis was highly selective for BAT over WAT.
We also conducted control experiment with 64 CuCl 2 only, and found that BAT uptake was about 1.0% ID/g (SI Fig.3), which was much lower than that with 64 Cu-Dis, suggesting that Disulfiram is necessary for high BAT uptake.

In vivo BAT imaging with 64 Cu-Dis
To investigate whether the "synthesis-free" method can be used to consistently image BAT mass under different conditions, we imaged mice under a normal condition and under cold exposure, which is a standard protocol for BAT activation (2,32). For cold exposure, the mice were placed in a cold room at 4°C for 2 hours before and 1 hour after Disulfiram administration. Next, the mice were injected with 64 CuCl 2 and kept at 4°C for 30 minutes followed by imaging for 30 minutes (Fig.2a). BAT uptake was about 10.0 % ID/g for normal conditions at 1 hour after 64 CuCl 2 injection with no significant decrease Cold Normal c at 6 hours (SI Fig.4). Importantly, we found that there was no significant difference in uptake between normal and cold treated groups (Fig.2b,c). This is contrary to [18F]-FDG PET and other imaging methods, in which BAT contrast is highly dependent on its activation status. Our data suggested that this synthesis-free method could be used for reporting BAT mass regardless of BAT activation status.
It has been reported that different anesthesia regimens have significant suppression impact on BAT activity (7,33,34). To investigate this, we kept mice under light anesthesia for 20 minutes before and 40 minutes after Disulfiram injection, followed by 64 CuCl 2 injection. We found that there was no significant difference in BAT uptake in animals under 20-minute anesthesia compared to animals under normal conditions (SI

Blocking studies with TSPO-specific ligand F-DPA
Our in vivo and ex vivo results revealed that the 64 Cu-Dis combination was indeed a synthesis-free method for PET imaging of BAT, however, it was not clear what the molecular binding target was. By surveying the publications reporting on the Disulfiram mechanism of action, and found that TSPO could be a target candidate (35,36). To investigate whether the uptake of Disulfiram or DDC is related to TSPO, we conducted blocking studies using the TSPO specific ligand F-DPA (Ki = 9nM) (Compound 3K in reference 37, and structure in SI Fig.6c) (37). F-DPA is an analogue of [F18]-DPA-714 (38), and is a highly selective ligand for TSPO (37,39). In these experiments, we intravenously injected 3.0 mg/kg of F-DPA after ip injection of Disulfiram, and imaged the mice using the same protocol as above. Images were captured at 1 hour and 6 hours after 64 CuCl 2 injection. Indeed, we found that F-DPA could effectively block BAT signal at both time points (Fig.3a,b), with a 35.0% and 60.0% decrease respectively (Fig.3c,d).
Interestingly, we also observed a significant decrease in PET signal from other organs such as heart and kidney (Fig.3e,f and SI Fig.6a,b), in which TSPO has reportedly high expression (40). Therefore, our data strongly suggested that high BAT contrast was associated with TSPO.

Biological validation of TSPO as a specific biomarker for BAT imaging
To further confirm that TSPO can be used as a specific imaging biomarker for BAT, we performed biological analyses. TSPO, also called translocator protein or peripheral benzodiazepine receptor (PBR) (28,40), is located on the outer mitochondrial membrane, and one of its characteristic features of BAT is its high abundance of mitochondria compared to WAT (9)(10)(11). It is likely that the abundance of TSPO in BAT is much higher than in WAT. Nonetheless, no experimental data are available indicating whether TSPO expression in BAT and WAT is significantly different, which is crucial for determining whether TSPO can be used as a specific imaging biomarker for BAT. To answer this question, we compared the qPCR data of TSPO in BAT and WAT, and found that the mRNA level in BAT was about 1.5-fold higher than in WAT (SI Fig.7). To further confirm different levels of TSPO protein expression in BAT and WAT, we performed  Fig. 4a, protein level in BAT was about 20-fold higher than in WAT (Fig.4a,b). We next investigated the abundance of mitochondria in BAT and WAT tissues using a mitochondria specific dye (MitoTrack deep red FM). As expected, BAT showed much higher fluorescence intensity than WAT (Fig.4c). In addition, we performed immunological staining of BAT and WAT tissues with anti-TSPO antibody. Fig. 4d demonstrated that the abundance of TSPO in BAT was significantly higher than that in WAT. In conclusion, our data for the first time suggest that TSPO could be used as a specific imaging biomarker for BAT.

PET imaging of BAT with TSPO-specific tracer [18F]-F-DPA
To further validate whether TSPO is a specific imaging biomarker for BAT, we investigated whether existing TSPO-specific PET tracers could provide high contrast for BAT over WAT. To this end, we used 18F-labeled TSPO-specific ligand [18F]-F-DPA (37)(38)(39). We first performed PET imaging in Balb/c mice using a 20-minute static scan.
As seen in Fig. 5a-c, there was a readily identifiable contrast from BAT with a 16% ID/g uptake 30 minutes after the injection. Moreover, we also compared images at locations with high WAT abundance and found that inguinal and gonadal WAT had minimal uptake (SI Fig.8).
To further validate the uptake of [18F]-F-DPA in BAT, we conducted a bio-distribution study in Balb/c mice at 45 minutes after the probe injection. BAT, inguinal WAT and gonadal WAT, brain, liver, and heart were harvested (SI Fig.9). The bio-distribution data confirmed a higher uptake of the tracer in BAT compared to other organs, suggesting that

18F-DPA uptake is not influenced by BAT activation
To investigate whether [18F]-F-DPA can be a reliable PET tracer for BAT mass, we conducted PET imaging under BAT activation with cold exposure. Mice were exposed to 4°C for 2 hours before imaging. Our imaging data showed that there were no significant differences between cold treatment and normal condition, indicating that [18F]-F-DPA could serve as a reliable tracer for BAT mass independent of its activation (Fig.5d).
Thompson et al have recently showed that TPSO expression could not be altered by acute cold exposure, which is consistent with the data obtained in our experiment (41). This also suggests that TSPO is a reliable biomarker for BAT mass.

Discussions and conclusions
PET imaging is immensely useful for both clinical and preclinical applications, however its high cost prevents it from being widely applicable. Requirements for an on-site cyclotron and on-site synthesis are among the major contributors of the high cost. In this report, we demonstrated the feasibility of a synthesis-free PET imaging. Although we demonstrated the application of this method for a specific case that involved imaging of BAT mass, we believe that "pseudo-synthesis-free" and "synthesis-free" methods could be used for other applications as well. "Pseudo-synthesis-free" methods can be realized with fast and strong chelation reaction of ligands with suitable radionuclides. "Synthesisfree" could be feasible if some compounds, which have strong coordination with the radionuclides, could pre-target receptors/enzymes, or if peptides/proteins that have strong chelation with radionuclides could be engineered into specific targets.
In the present study, Disulfiram was used as a precursor of a copper ligand, and the actual copper delivery drug to the brain in Menkes disease, which is characterized by dysfunction in copper transporting (42). Brain images from our study (Fig.1) showed a significant amount of copper accumulated in the brain, which is consistent with the reported results (42 Our studies indicate that the BAT contrast is due to the binding of 64 Cu-Dis to TSPO, which is consistent with previous reports (35,36). Katz (36). In spite of the fact that TSPO has been an imaging target for a long time, particularly for brain imaging, very few reports have emerged utilizing it for peripheral target imaging, and biological evaluation of the high TSPO expression level in BAT is rare (41). In the present study, for the first time, we validated TSPO as a biomarker for imaging BAT.
"Browning", a process of turning WAT into BAT (43)(44)(45) represents an exciting approach for converting "bad fat" to "good fat" (BAT and beige fat). Since the "browning" process could result in more mitochondria, and our method is capable of detecting the abundance of TSPO in mitochondria, it is reasonable to speculate that our methods have the potential to monitor the browning process, in which the abundance of TSPO increases with the increase of BAT-like cells.
Although TSPO was discovered nearly 40 years ago, its functions in obesity and adipocytes are not well explored. Our studies discovered and confirmed high TSPO expression in BAT, which opened a new avenue for basic BAT research aimed at investigation of biological functions of TSPO. In addition, since TSPO is tightly associated with inflammation (28), it is conceivable that 64 Cu-Dis is also suitable for imaging inflammation in different diseases.
In summary, we demonstrated that a synthesis-free method for PET imaging was feasible, and the combination of 64 CuCl 2 and Disulfiram could be used for BAT imaging. We also validated, for the first time, TSPO as an imaging biomarker for BAT. We believe that our method can be widely applied for TSPO and BAT imaging, and that the synthesis-free strategy could significantly reduce the cost of PET imaging.

Materials and Methods
All of the chemicals were purchased from commercial vendors and used without further  Imaging analysis was conducted with Amide.

Supplemental Figures
SI Fig.1 Top-down screening of Copper (II) ligands for BAT imaging.