18F-meta-fluorobenzylguanidine (18F-mFBG) to monitor changes in norepinephrine transporter expression in response to therapeutic intervention in neuroblastoma models

Targeted radiotherapy with 131I-mIBG, a substrate of the human norepinephrine transporter (NET-1), shows promising responses in heavily pre-treated neuroblastoma (NB) patients. Combinatorial approaches that enhance 131I-mIBG tumour uptake are of substantial clinical interest but biomarkers of response are needed. Here, we investigate the potential of 18F-mFBG, a positron emission tomography (PET) analogue of the 123I-mIBG radiotracer, to quantify NET-1 expression levels in mouse models of NB following treatment with AZD2014, a dual mTOR inhibitor. The response to AZD2014 treatment was evaluated in MYCN amplified NB cell lines (Kelly and SK-N-BE(2)C) by Western blot (WB) and immunohistochemistry. PET quantification of 18F-mFBG uptake post-treatment in vivo was performed, and data correlated with NET-1 protein levels measured ex vivo. Following 72 h AZD2014 treatment, in vitro WB analysis indicated decreased mTOR signalling and enhanced NET-1 expression in both cell lines, and 18F-mFBG revealed a concentration-dependent increase in NET-1 function. AZD2014 treatment failed however to inhibit mTOR signalling in vivo and did not significantly modulate intratumoural NET-1 activity. Image analysis of 18F-mFBG PET data showed correlation to tumour NET-1 protein expression, while further studies are needed to elucidate whether NET-1 upregulation induced by blocking mTOR might be a useful adjunct to 131I-mIBG therapy.


Results
Internalisation of 18 F-mFBG and 123 I-mIBG. To assess the NET-1-mediated uptake of 18  When incubated with either 123 I-mIBG or 18 F-mFBG, the cell-associated radioactivity reflected NET-1 protein levels with SK-N-BE(2)C cells showing the greatest uptake and Kelly showing the least uptake (Fig. 1c). In contrast, uptake in cells with intermediate NET-1 expression did not reflect the total protein level measured by WB, perhaps due to the lower number of transporters available on the membrane for radiotracer binding. Importantly, the uptake of 18 F-mFBG paralleled that of 123 I-mIBG and the uptake of each product was inhibited by desipramine (DMI), a NET-1 inhibitor, confirming the requirement for NET-1 activity in cellular tracer uptake (Fig. 1c).
In vitro mTOR inhibition and modulation of radiotracer uptake. To assess the cellular response to mTORC1/2 inhibition, Kelly and SK-N-BE(2)C cell lines, both of which over-express MYCN protein (Fig. 1b), were incubated with AZD2014 (0 to 1 µM) for 72 h. The GI 50 for AZD2014 (concentration that reduced cell viability by 50%) was 476 nM in Kelly and 307 nM in SK-N-BE(2)C cells ( Supplementary Fig. 5). Key substrates of AZD2014-induced mTORC1/2 inhibition were assessed by WB. The immunoblots showed a concentration dependent decrease in phosphorylation of both ribosomal S6 protein (S6) and eukaryotic translation initiation factor 4E-binding protein 1 (4EBP1) in Kelly and SK-N-BE(2)C cells (Fig. 4a, Supplementary Fig. 6 and 7), key indicators of mTORC1 inhibition. MYCN protein expression was not robustly reduced ( Fig. 4a, Supplementary  Fig. 6 and 7). In both cell lines, a concentration-dependent downregulation of p-Akt S473 was evident at early time points (3-12 h) (Supplementary Fig. 6 and 7), but inhibition was not sustained after 72 h incubation with the compound (Fig. 4a). Given the potential role of Akt in NET-1 protein synthesis and surface expression 43 , we further looked at NET-1 protein levels following AZD2014 treatment. In SK-N-BE(2)C cells, a prominent and persistent increase in total NET-1 protein expression was seen at 24 h post-incubation ( Supplementary Fig. 6) and was maintained at 72 h (Fig. 4a). In Kelly cells, the WB analysis was not sensitive enough to detect and accurately quantify the inherently low expression of the target protein.
To investigate potential changes in NET-1 function following an AZD2014-induced increase in protein expression, 18 F-mFBG and 123 I-mIBG cell uptake assays were performed in Kelly and SK-N-BE(2)C cells incubated with the drug (0-500 nM) for 24 and 72 h. A concentration-dependent increase of 123 I-mIBG and 18 F-mFBG uptake was observed in both cell lines in the 24 h treatment-time point ( Supplementary Fig. 8b), but this was www.nature.com/scientificreports/ more evident at 72 h as compared to untreated control cells (Fig. 4b,c). Following pre-treatment with 200 nM and 500 nM AZD2014, uptake of 18 F-mFBG in Kelly cells significantly increased by 128 ± 23% and 292 ± 27%, respectively (p < 0.0001, n = 2 performed in triplicate). Perhaps due to the already high expression of NET-1, increase in radiotracer uptake was less evident in SK-N-BE(2)C cells (39 ± 9% and 62 ± 5%, respectively). A similar trend was seen in 123 I-mIBG incubated cells pre-treated with AZD2014 ( Fig. 4c). Desipramine (500 nM) reduced the radioactivity signal in AZD2014 pre-treated cells confirming that uptake of both radiotracers was NET-1 transporter-specific (Fig. 4b,c Supplementary Fig. 8a).

F-mFBG to monitor changes in NET-1 expression in vivo.
In light of the effects of AZD2014 treatment on NET-1 expression observed in vitro, we investigated whether 18 F-mFBG could capture changes in tumour NET-1 expression following AZD2014 (25 mg/kg/day) treatment in Kelly xenografts (low NET-1). After 3 days post-treatment initiation mild toxicities were observed (i.e. mice treated with the compound lost ~ 10% of body weight compared to the vehicle control) ( Supplementary Fig. 9a). Moreover, tumour growth was impeded in AZD2014-treated xenografts (Supplementary Fig. 9c) (n = 3, p < 0.0001). Biodistribution studies performed 4 h post-radiotracer injection demonstrated no change in radiotracer uptake in organs such as the heart or small intestine, which are known to be innervated by NET-1-positive sympathetic ganglia 44 . Further, we observed an increase in 18 F-mFBG uptake in drug treated tumours (0.74 ± 0.25%ID/g) as compared to controls (0.39 ± 0.13), but this difference was not significant (p = 0.85, n = 3). The tumour-to-blood and tumour-to-muscle ratios were 1.75 ± 0.29 and 0.53 ± 0.01 in the vehicle group, as compared with 4.16 ± 1.89 and 1.23 ± 0.61 in the AZD2104 treated mice, respectively (Table 1). Following these observations, we evaluated a lower dose (20 mg/kg/day) of AZD2014 (or vehicle control) for 1, 3 or 7 days in SK-N-BE(2)C tumour bearing mice to ascertain longitudinal tumour NET-1 status. No toxicity was observed during the treatment period. No change in non-target organ distribution was observed at any time-point (Table 2). A prominent difference in tumour radiotracer accumulation was observed on day 3 in the treatment group (3.95 ± 1.53%ID/g) as compared to the vehicle controls (2.39 ± 0.01%ID/g) resulting in a greater tumour-to-blood and tumour-to-muscle ratio (Table 2), however this was statistically insignificant (p = 0.81 and p > 0.99, respectively, n = 3). These results were concordant with image-derived tumour radiotracer uptake studies (Fig. 5a, Supplementary Fig. 4b). We observed a decrease in radioactivity in treated tumours on day 7 post-treatment (2.61 ± 1.19%ID/g) that was statistically indistinguishable from controls (2.17 ± 1.01%ID/g) (p > 0.99, n = 3). These statistically insignificant changes in radiotracer uptake between vehicle and AZD2014treated xenografts were corroborated by ex vivo analysis of tumour samples. Western blot data showed an increase in NET-1 expression in SK-N-BE(2)C tumour lysates (Fig. 5b), but this difference was not significant (p = 0.07,  18 F-mFBG uptake in Kelly and SK-N-BE(2)C cells pre-treated with AZD2014 for 72 h 0-500 nM, and the same conditions for (c) 123 I-mIBG uptake. Data presented as mean ± SEM, n ≥ 2, performed in triplicate. Inhibition of radiotracer uptake n = 1, performed in triplicate. *p < 0.05, ***p < 0.001, ****p < 0.0001; 2-way ANOVA with Tukey post-hoc test. DMI = desipramine. Graphs are generated using GraphPad Prism (v 8.4.1), https ://www.graph pad.com.  Fig. 10). Moreover, there were no apparent differences in phosphorylated Akt between control and treated tumours of both tumour types. Furthermore, staining of phosphorylated S6 or 4EBP1 was identical in SK-N-BE(2)C tumours treated with AZD2014 and vehicle (Fig. 5c). In Kelly xenografts, there was no difference between NET-1 expression levels in Kelly tumour lysates between control and treated animals ( Fig. 5d) (p = 0.84, n = 2). Further, IHC showed no changes in p-S6 S240/244 or p-4EBP1 T37/46 in Kelly xenografts (Fig. 5e) indicating that the dose of AZD2014 in this regimen failed to inhibit mTORC1/2. Quantitative PET analysis did demonstrate good correlation with tumour NET-1 status in the vehicle treated tumours (r 2 = 0.98), but in the AZD2014 treated animals, the slope of the line of best fit was perturbed (r 2 = 0.53) (Fig. 5f).  51 . Therefore, a combined strategy by which MYCN is inhibited and NET-1 levels could be primed may be of particular benefit. The radioconjugates 123 I-and 131 I-mIBG are widely used for the imaging and therapy of NB patients owing to the selective presence of NET-1, and a 123 I-mIBG baseline scan is a necessary prerequisite of 131 I-mIBG therapy. 123 I-mIBG is also used to monitor disease progression, and increased accumulation of the radiotracer is associated with an unfavourable outcome 52 . Furthermore, an early response to chemotherapy captured on mIBG imaging correlates with good prognosis, improved EFS and OS in advanced and HR-NB 53 . However, 123 I-mIBG imaging is a descriptive technique, making clinical interpretation and accurate quantitation of a the dose:response relationship challenging. Semi-quantitative scoring systems have been developed (e.g. CURIE or SIOPEN visual scoring system) to provide an objective and uniform way for evaluation of disease burden and efficacy of therapies [54][55][56] . The availability of quantitative PET imaging agents to measure NET-1 activity would improve detection and quantitation of the disease.

Scientific Reports
Promising data in 5 NB patients indicates that a fluorinated guanidine analogue, 18 F-mFBG, has similar biodistribution to that of 123 I-mIBG, faster clearance, higher imaging resolution, and improved assessment of lesion radiotracer uptake 31 . Of note, a new European clinical trial is presently investigating 18 F-mFBG imaging in NET-1 expressing tumours [NCT02348749]. However, these trials have yet to introduce quantifiable measures of 18 F-mFBG uptake in specific tumour lesions. The studies presented herein strengthen the case that 18 F-mFBG imaging could be a useful modality for non-invasive assessment of NET-1 status in NB during therapeutic intervention. We have implemented simplified radio-synthetic approaches that have been developed to produce 18 F-mFBG via fluorination of electron-rich aromatics using copper-mediated fluorodeboronation 42 .
Traditional radiochemical strategies have focused on electron-poor aromatics to carry out nucleophilic aromatic substitution reactions (e.g. Balz-Schieman reaction). However, new radiosynthetic strategies have become recently available. For example, Rotstein et al.reported the synthesis of 18 F-mFBG via a spirocyclic iodonium(III) ylide (~ 14% RCY) 39 . The groups of Sanford and Gouverneur published approaches to the fluorination of electronrich aromatics via copper-mediated destannylation or deborylation 42,57 . This has proved a versatile method to www.nature.com/scientificreports/ access highly functionalised electron-rich PET radiotracers including 18 F-mFBG 58 . These highly promising results prompted us to explore the fully automated cassette-based radiosynthesis of 18 F-mFBG to support an intensive programme of research at our institution. As expected, and in line with previously reported studies 38 , uptake of both 18 F-mFBG and 123 I-mIBG correlated with the NET-1 expression levels in vitro and allowed for clear delineation of NET-1 expressing SK-N-BE(2)C xenografts after 4 h p.i..
Using two MYCN amplified cell lines (Kelly and SK-N-BE(2)C), we initially focused on assessing the ability of the mTORC1/2 inhibitor AZD2014 to target the mTOR/Akt axis in vitro and subsequently investigated whether mTORC1/2 inhibition will modulate NET-1 expression. In vitro, we observed the concentration-dependent inhibition of mTORC1 substrates p-4EBP1 T37/46 and p-S6 S240/244 . In addition, consistent with previously described results, p-Akt S473 was also depressed 59,60 , however this was only a transient effect in our experiments. Dual inhibition of both mTOR complexes is associated with MYCN protein degradation 9 , which was mild in Kelly and SK-N-BE(2)C cells treated with 500 nM AZD2014. Although higher drug concentrations may result in further MYCN depression, this would result in greater cell death and impede further analysis.
Importantly, WB analysis revealed a concomitant increase in NET-1 expression in SK-N-BE(2)C cells after AZD2014 treatment; an effect attributable to the drug inhibition of p-Akt S47343 . Of note, an enhanced NET-1 function was also highlighted in both Kelly and SK-N-BE(2)C by the increased cell uptake of 123 I-mIBG and 18 F-mFBG, in a concentration-dependent manner. The maximum change in NET-1 activity was observed in Kelly cells and was not clearly evident in SK-N-BE(2)C most likely due to high levels of inherent NET-1 expression.
Changes in NET-1 expression level post-mTOR inhibition have not been investigated yet by molecular imaging. Recently, T1-weighted MRI has been used to detect apoptotic responses 24 h following AZD2014 treatment in NB tumours 61 . Therefore, we assessed whether PET imaging could be used to monitor AZD2014-induced alterations in 18 F-mFBG tumour uptake by upregulating NB NET-1 expression levels in vivo. Following our in vitro findings, Kelly xenografts were used in vivo as a model system. Inhibitor dose regimens were selected based on previous reports 61,62 . After 3 days treatment, tumour growth plateaued and 18 F-mFBG biodistribution studies showed increased tumour-to-blood and tumour-to-muscle ratios in AZD2014-treated animals compared to the vehicle-treated controls, however PET signals were still too low for robust quantification and mild toxicities were noted. Therefore, a lower AZD2014 dose was investigated in the SK-N-BE(2)C model. There was a large variation in the treated SK-N-BE(2)C response according to PET and these results were corroborated by ex vivo analysis of tumour tissue. In both Kelly and SK-N-BE(2)C tumour lysates, WB results demonstrated only small differences in NET-1 protein levels and there was no robust inhibition of p-Akt S473 in treated tumours. Similarly, IHC staining did not detect any reduction of p-S6 S240/244 and p-4EBP1 T37/46 , which confirmed that the selected dose regimens did not fully inhibit the mTOR signalling.
Interestingly, a clear relationship between 18 F-mFBG uptake and NET-1 expression was observed in the vehicle treated tumours. However, in the AZD2014 treated group, accumulation of the radiotracer within the tumour was inhibited. This could suggest that the mTORC1/2 inhibitor was affecting the delivery of the radiotracer, which could be due to anti-angiogenic effects of mTOR inhibition 63 . Further optimisation of this approach, including B.I.D. dosing of AZD2014 or using alternative mTORC1/2 inhibitors, is needed to address the potential benefits of mTORC1/2 inhibition and 131 I-mIBG combinations.

Conclusion
In this work, we have shown that dual mTORC1/2 inhibitor (AZD2014) potentiates NET-1 expression in NB cells in vitro. Although the AZD2014 exhibited suboptimal activity in vivo and more studies are required to validate a more effective primary therapeutic strategy to inhibit tumour progression whilst sensitising cells to 131 I-mIBG therapy, the ability of 18 F-mFBG to quantify the NET-1 expression was highlighted. This work supports the potential of 18 F-mFBG to use this tracer in future studies for image-guided therapeutic strategies leading to more robust and durable responses to 131 I-mIBG radiotherapy. Western blot. Western blotting (WB) was performed as previously described 64 . Antibodies were obtained from Cell Signalling Technologies (London, UK) unless stated otherwise: GAPDH, beta-actin, p-Akt S473 , Akt, p-4EBP1 T37/46 , p-S6 S240/244 , S6, N-MYC (Merck Millipore, Watford, UK), NET-1 (mAbTechnologies, Stone Mountain, Georgia), anti-rabbit HRP, anti-mouse HRP. Blots were visualised with SuperSignal West Pico and Femto Chemiluminescent Substrate (Thermo Fisher Scientific, Loughborough, UK) and ChemiDoc XRS + System (BioRad, Watford, UK). Data were processed and band density analysed with ImageJ and Image Lab 6.0 (BioRad, Watford, UK).

Materials and methods
Preparation of radiotracers. 123 I-mIBG was purchased from GE Healthcare (AdreView Amersham, UK). 18 F-mFBG was prepared using a Trasis All in One (AiO) synthesiser (Trasis SA, Liege, Belgium) housed in a In vitro uptake of 123 I-mIBG and 18 F-mFBG. To evaluate the radioactive agent uptake specificity, approximately 3.0 × 10 5 adherent cells were incubated with either 18 F-mFBG (150 kBq) or 123 I-mIBG (5 kBq) for 1 h at 37 °C, in the presence or absence of the NET-1 specific inhibitor, desipramine (DMI, 50 µM, Sigma Aldrich, Gillingham, UK). Subsequently, the cells were washed with PBS, trypsinised and the radioactivity was measured using a γ-counter (2480 WIZARD 2 , PerkinElmer, Beaconsfield, UK). For each cell line the cell-associated radioactivity was normalised to the number of cells and then each group was presented as a percentage of the signal acquired for SK-N-BE(2)C cells (mean of n = 3 independent experiments performed in triplicate ± SEM).
To test the response of cell uptake using the mTOR1/2 inhibitor AZD2014 (0-500 nM) (vistusertib, Stratech Scientific, Ely, UK), a stock solution of the drug (10 mM) in dimethyl sulfoxide (DMSO, Sigma Aldrich, Gillingham, UK) was initially prepared and then diluted in culture medium to a final DMSO concentration of < 0.1%. Radiotracer uptake was assessed in cells 24 and 72 h after incubation with AZD2014. The uptake was normalised to cell viability as performed using the CellTiter-Glo assay (Promega, Southampton, UK). The data are expressed as the average of n = 3 independent experiments (performed in triplicate) ± SEM. For the treatment studies, AZD2014 stock solution in DMSO was diluted in PEG400 to a final concentration of 2 mg/mL (10% DMSO maximum). When tumours reached ~ 100 mm 3 (calliper measurements using formula: Volume = (Width(2) × Length)/2), mice were treated daily, for a total of either 1, 3 or 7 days, by oral gavage with AZD2014 (20 or 25 mg/kg/day) or vehicle control. Mice were monitored daily for body weight or other adverse effects.
Biodistribution and imaging. When the tumours reached approximately 100-200 mm 3 , mice were anaesthetised with an isoflurane/O 2 mixture (1.5-2.0% v/v) and intravenously injected with either 18 F-mFBG (~ 7.5 MBq) or 123 I-mIBG (~ 20 MBq). Mice were then imaged using an Albira PET/SPECT/CT scanner (Bruker, Coventry, UK). Whole body PET static images were acquired after 1 and 4 h post-injection (p.i.) for 10 min with an energy window of 358 to 664 keV followed by CT acquisition. Acquisition, reconstruction and image analysis were performed as described previously 66 . Following the final scan, the mice were sacrificed and the major organs were excised, weighed and their associated radioactivity was measured using a γ-counter. Biodistribution and image quantification were expressed as a percentage of injected dose per gram of tissue (%ID/g), whereby 1 cm 3 on PET acquisitions were assumed to equal 1 g (n ≥ 3 mice ± SD).
Autoradiography. Dissected tumours were set in Tissue-Tek optimal-cutting-temperature compound (Sakura, Torrance, California, USA) and snap-frozen in liquid nitrogen. The tumours were then sectioned to a thickness of 10 µm using a cytomicrotome (Thermo Fisher Scientific, Loughborough, UK) and mounted on slides, that were exposed to X-ray film for 24 h and scanned using a Typhoon FLA 7000 phosphorimager (GE Healthcare Life Sciences, Amersham, UK).