[18F](2S,4R)-4-Fluoroglutamine as a New Positron Emission Tomography Tracer in Myeloma

The high glycolytic activity of multiple myeloma (MM) cells is the rationale for use of Positron Emission Tomography (PET) with 18F-fluorodeoxyglucose ([18F]FDG) to detect both bone marrow (BM) and extramedullary disease. However, new tracers are actively searched because [18F]FDG-PET has some limitations and there is a portion of MM patients who are negative. Glutamine (Gln) addiction has been recently described as a typical metabolic feature of MM cells. Yet, the possible exploitation of Gln as a PET tracer in MM has never been assessed so far and is investigated in this study in preclinical models. Firstly, we have synthesized enantiopure (2S,4R)-4-fluoroglutamine (4-FGln) and validated it as a Gln transport analogue in human MM cell lines, comparing its uptake with that of 3H-labelled Gln. We then radiosynthesized [18F]4-FGln, tested its uptake in two different in vivo murine MM models, and checked the effect of Bortezomib, a proteasome inhibitor currently used in the treatment of MM. Both [18F]4-FGln and [18F]FDG clearly identified the spleen as site of MM cell colonization in C57BL/6 mice, challenged with syngeneic Vk12598 cells and assessed by PET. NOD.SCID mice, subcutaneously injected with human MM JJN3 cells, showed high values of both [18F]4-FGln and [18F]FDG uptake. Bortezomib significantly reduced the uptake of both radiopharmaceuticals in comparison with vehicle at post treatment PET. However, a reduction of glutaminolytic, but not of glycolytic, tumor volume was evident in mice showing the highest response to Bortezomib. Our data indicate that [18F](2S,4R)-4-FGln is a new PET tracer in preclinical MM models, yielding a rationale to design studies in MM patients.


INTRODUCTION
Multiple myeloma (MM) is a hematological disease characterized by the accumulation of malignant plasma cells (PC) into and, more rarely, outside the bone marrow (BM) (1). In the last years, 2-deoxy-2-[ 18 F]fluoro-D-glucose positron emission tomography/computed tomography ([ 18 F]FDG PET/ CT) in MM has attained significant relevance, and it is considered the cornerstone of MM imaging at the initial diagnosis as well as in staging, prognostic evaluation, and monitoring response to therapy (2). Thus, [ 18 F]FDG PET/CT is currently used to assess active bone lesions and extramedullary localizations in MM patients (3). However, [ 18 F]FDG uptake yields both false positive and false negative lesions, and only 60-70% of patients with active MM are positive for [ 18 F]FDG PET (4,5). These data support the need for additional imaging methods to assess skeletal involvement and monitoring the effect of treatment. To this purpose, several other PET tracers, such as choline and methionine, have been proposed (6,7).
Previous studies showed that [ 18 F]4-FGln is taken up by Gln transporters, including ASCT2, in solid tumors (13). Since we have already demonstrated that MM cells have increased expression of several Gln transporters and mainly depend on ASCT2 for Gln transport (11), we hypothesized that [ 18

Chemical Synthesis and Characterization
The synthesis of the four stereoisomers of 4-FGln and their 18 Flabeled counterparts, together with data on the uptake of these compounds in 9L and SF188-Bcl-xL tumor cells (Gln-addicted glioblastoma-derived tumor cells), has been previously reported (14). In that work, the radiolabeled (2S,4R)-configured 4fluoroglutamine [ 18 F]4-FGln ( 18 F-1), a fluorinated analogue of natural L-glutamine, displayed the best uptake properties in tumor cells as compared to the other stereoisomers.
Based on these precedents, we carried out the stereospecific synthesis of 4-FGln (1), to be used both for in vitro biological assays and as a reference compound in radio-HPLC analyses.
The synthesis procedure moved upon the previously reported footsteps (14), apart from some optimization variants. Thus, as shown in Supplemental Figure 1, starting from commercially available 2S-configured protected homoserine 2, the synthesis path proceeded uneventfully, providing the tosyl product 10 in 29% overall yield. This advanced intermediate was ready for either the subsequent radiofluorination step to [ 18 F]4-FGln, or its transformation to "cold" target 1.
Following the reported procedure for fluorination step of 10 to 11, we were able to isolate but low amounts of the desired product (32% yield instead of the reported 77%). Thus, we proceeded to slightly modify the procedure by adopting the following conditions: (i) TASF (5 equiv) and Et 3 N·(HF) 3 (3 equiv) till reaching pH 5; (ii) solution concentration was a critical parameter and 0.1 M was judged optimal; and (iii) dry solvents were necessary to avoid C2 epimerization. Under these conditions, compound 11 was isolated in a very good 86% yield with complete stereochemical integrity.
Finally, global removal of the acid-sensitive protecting groups within 11 was carried out by employing trifluoroacetic acid and dimethyl sulfide, to provide crude fluoroglutamine 1. The purification procedure of the target compound 1 turned out to be more troublesome than expected. According to the reported procedure, a first column purification step using Dowex 50WX8-200 (H + form) resin followed by recrystallization from EtOH/ H 2 O should have provided the final product in a good yield. In our hands, this method did not furnish any precipitate; after resin column, the eluted fraction was purified by reverse phase HPLC (using H 2 O/TFA 0.1% and acetonitrile as eluent mixture), giving an unknown fluorinated product whose mass spectrum (ESI + ) coincided with that of the target, but the characterization data ( 1 H/ 13 C/ 19 F NMR spectra, HPLC retention time), though similar, did not perfectly matched those reported for the target. Further attempts of purifications of the crude using the same acidic resin and avoiding the HPLC analysis gave, again, unsuccessful results. The 19 F-NMR analysis performed directly on the fraction eluted from the Dowex, without concentrating it, revealed the presence of several byproducts. Therefore, the purification procedure was modified, and it was found that evaporation of the reaction mixture to eliminate TFA and direct purification via reverse-phase HPLC eluting with H 2 O (+0.1% formic acid) and acetonitrile lastly provided the desired fluoroglutamine 1 (41% yield), which was perfectly consistent with the reported data (optical activity, 1 H/ 13 C/ 19 F NMR spectra, HPLC retention time, Figures S2 and S3).
It was concluded that the coexistence of many reactive functional groups (amide, carboxylic acid, amine, fluorine) within the small molecule, together with the presence of two stereocenters, renders the purification of this molecule highly challenging, and particular caution to both basic and acidic conditions has to be paid. We may hypothesize that the use of the acidic Dowex resin in the last stage and/or the 5% aqueous ammonia used to elute the product could be responsible for the undesired formation of less polar, cyclic imide product 12 (Supplemental Figures 1,[4][5], possibly deriving from 1 via intramolecular dehydration closure. The data collected for the unknown compound are compatible with the structure of putative imide 12, while its mass spectrum, coinciding with that of target 1, could be generated by reopening of imide 12 under the mass source conditions. In conclusion, 4-FGln 1 was obtained in 10% overall yield over six steps from homoserine 2. (2S,4S)-tert-Butyl 2-(tert-Butoxycarbonylamino)-4-Hydroxy-5-Oxo-5-(2,4,6-Trimethoxybenzylamino) Pentanoate (9) The title compound was prepared from compound 8, thiourea and sodium bicarbonate, according to a reported procedure (1). The crude reaction mixture containing a 1:1 mixture of two diastereoisomeric alcohols 9 and 9' was purified by flash chromatography on silica gel using CH 2 Cl 2 /MeOH as eluent (gradient from 99:1 to 98.5:1.5). Pure product 9 was isolated in a 44% yield as a white solid. (2S,4R)-tert-Butyl 2-(Tert-Butoxycarbonylamino)-4-Fluoro-5-Oxo-5-(2,4,6-Trimethoxybenzylamino) Pentanoate (11) To a stirred solution of tris(dimethylamino)sulfonium difluorotrimethylsilicate (TASF, 160 mg, 0.582 mmol) in dry DCM/THF (0.7:0.7 ml), under nitrogen atmosphere, Et 3 N·(HF) 3 (6 ml) was added. After that, tosylate 10 (76 mg, 0.116 mmol) in dry DCM/THF (0.7:0.7 ml) was added to the TASF solution. The reaction mixture was heated to 50°C and kept under inert atmosphere. Some of the solvent was removed by a nitrogen flux in order to have a final volume of 1 ml (final concentration 0.1 M). After 16 h, the oil bath was removed and EtOAc (8 ml) was added. The organic phase was extracted with NaHCO 3 solution (0.5 M) (1×), water (1×), and brine (1× (2S,4R)-4-Fluoroglutamine (1) Protected 4-fluoroglutamine 11 (0.104 g, 0.208 mmol) was put in a two-neck flask, kept under N 2 atmosphere and cooled at 0°C with an ice bath. Dimethylsulfide was added (53 ml, 0.726 mmol), and immediately after TFA was added dropwise (5 ml, 3.53 mmol). The reaction was then removed by the ice bath and kept under stirring for 2.5 h at room temperature. TFA was evaporated, and the crude was washed with DCM (2 × 1 ml) and Et 2 O (2 × 1 ml) removing the solvent with a Pasteur pipette.
For the discrimination of the transporters involved in 4-FGln uptake, cells were incubated for 1 min at pH 7.4 in EBSS in the presence of [ 3 H]-Gln (10 mCi/ml, 0.1 mM) or 4-FGln (0.1 mM), and the influx was measured in the absence or in the presence of a-methylaminoisobutyric acid (MeAIB, 5 mM), L-threonine (Thr, 5 mM), or 2-aminobicyclo[2.2.1]heptane-2-carboxylic acid (BCH, 5 mM), as specific or preferential competitive inhibitors of SNAT1/2, ASCT2, or LAT1 transporters, respectively (11). At the end of the uptake, cells were rapidly washed with ice-cold urea (300 mM), and the intracellular amino acid content was extracted with 100 ml of cold absolute ethanol. For cells incubated with radiolabeled Gln, extracts were mixed with 200 ml of scintillation fluid and counted as described above, while the intracellular content of 4-FGln was quantified by high performance liquid chromatography coupled to tandem mass spectrometry (HPLC-MS/MS, see below). For both Gln and 4-FGln, influx data are expressed as pmol/mg prot/min.

Radiosynthesis of [ 18 F]4-FGln
Radiosynthesis of [ 18 F]4-FGln was obtained as described in literature (15), adapting the reported procedure to the in-house modified GE TRACERlab FX Module.   (16). This index was defined as Tumor Volume Response (TVR) and calculated as the percentage change in median tumor volume measured by caliper at the end of treatment over the median tumor volume before treatment. Treatment response was defined as Partial Response (PR) (TVR > −30%); Stable Disease (SD) (TVR < −30% and < +20%), and Progressive Disease (PD), (TVR > +20%). In this study, PR and SD mice were grouped as responders.

PET Acquisition
PET studies were performed on a YAP-(S)-PET II (ISE S.r.l., Pisa, Italy) (17 (18,19). Mice were positioned prone on a "handmade" mold on the PET bed to maintain the position with the tumor centered in field of view. For [ 18 F]4-FGln pilot study, mice (n = 3) underwent a 90-min dynamic acquisition (frames: 4 of 2.5 min, 4 of 5 min, 6 of 10 min each) to determine the optimal imaging time for uptake measures. In the subsequent studies, dynamic PET acquisitions started at 15 min lasting 30 min (six scans of 5 min each). Mice were positioned with the tumor centered in field of view (20) and anesthetized (isoflurane 2%) during acquisition.

Images Quantification
All images were calibrated with a dedicated phantom, corrected for the radionuclide half-life decay, and then quantified with PMOD 3.2 (Zurich, Switzerland). To set acquisition time and verify reversibility of [ 18 F]4-FGln (21), we applied the simplified Logan analyses using muscle as input function (21,22) to three JJN3 mice. Regions of interest (ROIs) were drawn on tumor and thorax muscle and data expressed as %ID/g to obtain timeactivity curves (TACs) of [ 18

Murine Serum Protein Electrophoresis
Blood was collected in Eppendorf by retro-orbital sampling. Semiautomated electrophoresis was performed on the Hydrasys instrument (Sebia, Lissex, France). According to the manufacturer's instructions, 10 ml of undiluted serum were manually applied to the Hydragel agarose gels (Sebia). The subsequent steps, electrophoresis (pH 9.2, 20 W constant current at 20°C), drying, amidoblack staining, de-staining, and final drying, were carried out automatically. The use of Hydrasys densitometer and Phoresis software (Sebia) for scanning resulting profiles provided accurate relative concentrations (percentage) of individual protein zones. M-spike levels were calculated as total gamma globulins/albumin ratio (G/A).

Statistics
In vivo data are presented as means ± standard deviation (SD). All statistical analyses were performed using Prism 8 (GraphPad Software, Inc., USA). Statistics are detailed in figure legends. Differences were considered significant when p < 0.05.

Chemical and Radiochemical Syntheses
Based on previously reported procedures (14), we carried out the stereospecific synthesis of 4-FGln ( Supplementary Figures 1-5) to be used both for in vitro biological assays and as a reference compound in radio-HPLC analyses. This section as well as radiolabeling have been extensively described in Methods.

Characterization of 4-FGln Uptake
Gln and 4-FGln were rapidly accumulated in either RPMI8226 or JJN3 HMCLs, reaching a maximum at 5 min of incubation ( Figure 1A). 4-FGln accumulation remained fairly stable up to 60 min, while showed a partial decrease at 120 min ( Figure 1B). To verify if 4-FGln enters MM cells through the same transporters used by Gln, radiolabeled Gln uptake was performed in the presence of increasing doses of either 4-FGln or Gln in RPMI8226. As shown in Figure 1C Figures 1D-E). Moreover, all the inhibitors tested were able to hinder both 4-FGln and Gln uptake, with Thr showing the highest and MeAIB the lowest inhibitory activity (Figures 1D-E). BCH inhibited more 4-FGln than Gln uptake, suggesting that 4-FGln exploits sodiumindependent system L transporters more than the natural amino acid.

Tumor and Muscle Time Activity Curves of [ 18 F]4-FGln
Firstly, we characterized [ 18 F]4-FGln kinetics in vivo using the NOD.SCID JJN3 xenograft model ( Figure 2). Figure 2A shows JJN3 tumor and muscle TACs. In tumor, [ 18 F]4-FGln uptake peaked at 25 min slowly declining thereafter ( Figure 2B). The Logan plot ( Figure 2C) confirmed linearity starting at 15 min from injection, consistent with largely reversible tracer exchange (21 Upon injection into C57BL/6 mice, Vk12598 cells colonize the BM, without lytic lesions, and the spleen, generating an aggressive MM (24). The expression of the glutamine transporter ASCT2 was checked in the BM (femur) of Vk12598 MM bearing mice. As shown in Figure 3A, the expression of the transporter increased along with MM progression, monitored with M-spike in blood samples collected at weeks 3, 4, and 5 after cell injection ( Figure 3B) Figures 4A, B). In any case, [ 18 F]FDG displayed a higher uptake than [ 18 F]4-FGln. Similar results were obtained considering SUVmax ( Figure 4C).
To verify if metabolic modifications were related with tumor response, mice in the Bortezomib-treated group were classified as responders and non-responders based upon the adapted RECIST score ( Figure 6A). Three Bortezomib-treated mice were classified as responders (#7G, #10G, #5G). These mice displayed a reduction of [ 18 F]4-FGln T/M ratio, [ 18 F]4-FGln-related tumor volume, and TLGln ( Figure 6B). On the contrary, independently from the response, all mice displayed increased [ 18 F]FDG parameters ( Figure 6C) Figures 7A, B). This strategy was applied to analyze in vivo, in the same animals, the modification in the volume of distribution of the two radiotracers and the presence of regional  Figure 7D). The tracer overlap increased from 32 to 43% after 6 days in control and from 26 to 45% in non-responder mice. On the contrary, the area of tracer overlap decreased at about 9% in responder mice. After 6 days, the area of exclusive  (8,9,21,(30)(31)(32), supporting its possible use as imaging biomarker. Moreover, [ 18 F]4-FGln could be used to study the metabolic profile and the Gln addiction of MM cells in order to design a metabolic-based therapeutic approach.
Recently, we have demonstrated that MM cells are strictly Gln addicted and lack of a sizable expression of the enzyme Glutamine Synthetase (11). Therefore, MM cells only rely on extracellular Gln. Consequently, MM cells are endowed with fast Gln uptake due to high expression of at least three different types of Gln transporters and, in particular, of the sodium-dependent carrier ASCT2, which is overexpressed in several types of Gln-dependent cancers (33). Moreover, we have also recently shown that the massive Gln uptake by MM cells decreases Gln concentration in the tumor microenvironment, contributing to the bone remodeling process induced by MM cells (34). Overall, this evidence provides a rational basis to exploit Gln metabolism in the design of a PET tracer for MM diagnosis and patient stratification.
Here  (36). In our study, with the signal acquired between 15 and 45 min, bone signal was not evident. To better understand this issue, we analyzed femur uptake in the syngeneic model. We performed a quantification of uptake in femur of the syngeneic mice where bone signal could be increased also by the presence of myeloma. We did not observe any significant difference between the two radiotracers when myeloma was not present (healthy condition). However, at the 4th week we observed a significant increase of [ 18  represents a major factor in Bortezomib efficacy. This confirms the metabolic heterogeneity of MM and suggests that modification in glutaminolysis represents a major event in Bortezomib efficacy. For the reasons above, our results suggest that [ 18 F]4-FGln may be a promising radiopharmaceutical for PET molecular imaging of the outcome of metabolically based target therapy acting on glutaminolysis.
Metabolic changes parallel MM progression (41). For instance, glutamine-dependent anaplerosis of the TCA cycle increases from MGUS to myeloma (12). Moreover, a sizable degree of metabolic heterogeneity may be present in the same MM (42), suggesting that subpopulations of MM cells may respond in a different way to therapeutic treatments. For these reasons, the combined use of distinct metabolically related probes, such as [ 18 F]4-FGln and [ 18 F]FDG, may yield clinically important information.
In conclusion, our data indicate that [ 18 F]4-FGln may be a new tracer to detect MM cells in preclinical in vivo models. [ 18 F] 4-FGln might help to explore the potential use of PET to better define the metabolic phenotype of the tumor and the modifications induced by therapy, particularly as a potential marker of treatment response to proteasome inhibitors. Moreover the in vivo study of the metabolic profile of myeloma cells by [ 18 F]4-FGln could be useful to design future metabolic-based therapeutic approach and for the clinical management of MM patients.