Synthetic Efforts toward the Synthesis of a Fluorinated Analog of 5-Aminolevulinic Acid: Practical Synthesis of Racemic and Enantiomerically Defined 3-Fluoro-5-aminolevulinic Acid

In 2017, the FDA authorized 5-aminolevulinic acid (5-ALA) for intraoperative optical imaging of suspected high-grade gliomas. This was the first authorized optical imaging agent for brain tumor surgery to enhance the visualization of malignant tissue. Herein we report the synthesis of a racemic and enantiopure fluorinated analog of 5-ALA, i.e., 3-fluoro-5-aminolevulinic acid (3F-5-ALA). We anticipate that these studies will provide the foundation for the future construction of a fluorine-18-labeled 5-ALA PET tracer to be used for functional and metabolic imaging of gliomas.


■ INTRODUCTION
Brain tumors represent a prominent category of severe and lifethreatening malignancies on a global scale.Within this category, gliomas emerge as the most prevalent primary neoplasms within the central nervous system (CNS), constituting approximately 80% of malignant brain tumors. 1 Their molecular and cellular heterogeneity promotes proliferation, invasion, and treatment resistance. 2 Therefore, noninvasive imaging technologies that reveal a tumor's molecular and metabolic information are needed for subtype-specific and individualized treatment for gliomas.Despite the fact that magnetic resonance imaging (MRI) is the clinical gold standard for imaging gliomas due to its superior spatial resolution and soft tissue contrast, its molecular and metabolic imaging applications suffer from low sensitivity. 3In contrast, positron emission tomography (PET) appears promising and offers insight into the molecular biology of tissues across the body. 4However, the currently used fluorodeoxyglucose (FDG) PET has low specificity and is unsuitable for imaging brain tumors due to significant FDG uptake in normal brain tissue.5a,b In contrast, [ 11 C]methionine is valuable for PET imaging in oncology, particularly for gliomas, due to its ability to trace tumor metabolism, but its short half-life poses logistical challenges.5c Currently, there is a significant unmet demand for highly diverse and infiltrating molecular imaging probes for gliomas.
5-Aminolevulinic acid (5-ALA) is the first natural precursor metabolite in the heme biosynthesis pathway; it is typically produced from succinyl-CoA and alanine. 6Protoporphyrin IX (PPIX) is a fluorescent red molecule that represents the final metabolite before heme production. 6During the synthesis of PpIX, 5-ALA is the rate-limiting intermediate step, and oral administration of 5-ALA causes selective accumulation of 5-ALA and PpIX within the pathological tissue, particularly Glioblastomas (GBMs). 7Its preferential accumulation within glioma cells is associated with decreased amounts of the ferrochelatase enzyme and impaired cellular clearance by an ATP-binding cassette transporter (ABCB6). 8It has been demonstrated that 5-ALA-induced PpIX fluorescence is highly selective (98%) for neoplastic tissue in patients with known glioblastoma multiforme. 9Exogenous administration of 5-ALA causes selective accumulation of 5-ALA and PpIX in gliomas, enabling tumor cell-specific intraoperative identification of the tumor margin for tumor resection with optical devices and photodynamic treatment of high grade gliomas (HGGs), 10 both of which have been FDA-approved for the treatment of gliomas.Even though ALA-mediated PDT offers several benefits, it has been utilized mostly for superficial lesions due to poor light penetration within the tissue. 11,12The combination of PET imaging and a radiolabeled form of 5-ALA can overcome the constraints of optical detectability and provide insight into diagnostic and therapeutic techniques (for presurgical planning purposes) with 3D rendering capacity at the molecular level throughout the body.
The possible PET radioisotopes suitable for 5-ALA are 11 C, 13   N, and 15 O.To date, Goodman and co-workers have successfully synthesized [ 13 N] 5-ALA in high radiochemical yield (65%) and validated the rapid tumor-specific uptake of [ 13 N] 5-ALA in rats bearing intracranial 9L glioblastoma. 13owever, the extremely short half-life of the 13 N tracer limits its use for further applications.The longer half-life of 18 F compared to those of 13 N and 11 C (110.8 min for 18 F vs 9.98 min for 13 N and 20.3 min for 11 C) makes it the ideal radionuclide for therapeutic PET imaging; hence, we envisioned producing a fluorine-18 derivative of 5-ALA.In order to develop a stable 18 F PET radioligand of 5-ALA, fluorine-18 can be introduced by substituting hydrogen at the 2-or 3-carbon positions of the ALA core, as illustrated in Figure 1.There have been reports of nonradioactive forms of 2-fluoro-5-ALA (2F-5-ALA) and 2,2′-difluoro-5-ALA. 14 Since 2F-5-ALA is an inhibitor of 5-aminolevulinate dehydratase (K i = 0.5−2 mM), it can only be used to determine the production rate of PpIX and is not suitable for PET imaging.Based on the inhibitory effect of 2F-ALA to aminolevulinate dehydratase, we chose to define and synthesize a nonradioactive version of 3fluoro-5-ALA (3F-5-ALA) and evaluate its optical characteristics to determine if fluorination at 3-position affects glioma cell uptake and metabolism to produce PpIX.In this manuscript, we disclose our synthetic efforts for constructing 3-fluoro-5-ALA and its enantiomers as a prelude to future disclosures of 18F analogs of 5-ALA.

■ RESULTS AND DISCUSSION
Our first approach to 3-fluorinated 5-ALA began with the substitution of the bromine atom of the requisite functionalized levulinate 1 (prepared by a known protocol 15 ) with nucleophilic fluoride (Scheme 1).However, many attempts to fluorinate bromide 1 under the required basic conditions, including (CsF/tBuOH, KF/18-crown-6/ACN, TBAF/THF, etc.) were thwarted by the formation of the olefinic byproduct via elimination of HBr, undoubtedly facilitated by the presence of the ester group.
Based on these initial observations, we concluded that the oxo and carboxylic acid functionalities of 3-F-5-ALA should be revealed later in the synthetic scheme to favor S N 2 formation of the desired fluorinated product over the competitive βelimination product since these activating groups enhance the formation of a double bond.We thus sought out appropriate protecting groups for the oxo and carboxylic acid groups that could be exposed post fluorination.After an extensive investigation, we were led to protectively embed both the amino and oxo functionalities of 3F-5-ALA within an oxazole precursor 16 as depicted at the top of Scheme 2 (I−II−II).We anticipated a gentle fluorodecarboxylation 17 to introduce the fluorine atom (bottom of Scheme 2) followed by a hydrolytic reveal of the 5-ALA skeleton.
Thus, we began our plans with the construction of a functionalized derivative of oxazole that would be amenable to decarboxylative fluorination (7 in Scheme 2).In accordance Scheme 3. Synthesis of rac-3F-5-ALA Scheme 4. Synthesis of 3F-5-ALA Enantiomers from Fluorosuccinic Anhydride 15 The Journal of Organic Chemistry with the existing literature, commercially available ethyl oxazole-4-carboxylate 3 was selectively brominated to give ethyl 5-bromooxazole-4-carboxylate 4. 18 This underwent S N Ar displacement with the sodium enolate of malonate 5, yielding the triester intermediate 6 in 76% yield.One-pot hydrolysis of the tertiary butyl esters of 6 followed by decarboxylation to the monoacid 7 proved challenging due to overdecarboxylation resulting in the formation of 9. Fortunately, when compound 6 was treated with a 1:1 TFA/CH 2 Cl 2 mixture for 30 min at room temp, monoacid 7 was produced in 58% yield (Scheme 2).
With the requisite monoacid 7 in hand, we focused on the crucial decarboxylative fluorination reaction.Unfortunately, numerous attempts to fluorinate 7 with electrophilic 17a,b and nucleophilic fluorinating 17c agents were futile, all resulting in the formation of the decarboxylative product 9.We postulated that the electron-deficient oxazole ring might activate the benzylic carboxylic acid 7 toward decarboxylation leading to 9.
As an alternative, a straightforward approach for synthesizing rac-3F-5-ALA using electrophilic fluorination of intermediate 11 with Selectfluor followed by decarboxylation under acidic conditions is depicted in Scheme 3. N-Cbz γ-amino-β-keto ester 10 (Scheme 3) is readily accessible in two steps from commercially available Cbz-glycine. 19Treatment of N-Cbz γamino-β-keto ester 10 with NaH and benzyl bromoacetate in tetrahydrofuran led to the formation of keto diester 11 in 62% yield along with the recovery of starting material.Keto diester 11 was fluorinated using Selectfluor in the presence of sodium hydride (76% yield).The resulting product was then refluxed with trifluoroacetic acid in dichloromethane to afford monofluoroketoester 13 in 85% yield.Finally, hydrogenation of the monofluoroketoester 13 in a mixture of THF/aq. 2 M HCl (4:1) solvent system furnished the desired rac-3-fluoro-5-Scheme 5.A New Approach to the Synthesis of 5-ALA Scheme 6. Proposed Scheme for the Synthesis of 14a from 2-Deoxy-D-ribose The Journal of Organic Chemistry aminolevulinic acid hydrochloride (3-F-5-ALA•HCl) 14 in 75% yield.
Having established a successful route for the synthesis of racemic 3F-5-ALA, we then turned our attention to the construction of enantiopure 3F-5-ALA.A literature search revealed four examples describing the synthesis of 5-ALA by ring-opening of succinic anhydride with different nucleophiles (N,N-diphenylmethylidene glycine ethyl ester, KCN, ethyl hippurate, and methyl nitro acetate). 20Based on this precedent, it was anticipated that both enantiomers of 3F-5-ALA would be accessible via the regioselective ring-opening of the appropriate enantiomeric 2-fluorosuccinic anhydride 15 with N,N-diphenylmethylidene glycine ethyl ester in the presence of a base, followed by acidic hydrolysis to give 3F-5-ALA•HCl (14a or 14b), as depicted in Scheme 4.
The synthesis started with the preparation of S-fluorosuccinic anhydride 15 (Scheme 4), which is readily accessible from L-aspartic acid, in two steps. 21Unfortunately, all attempts reacting 15 with any of the nucleophiles 20 mentioned earlier, including the use of different solvent systems and moderating additives (MgBr 2 or ZnCl 2 ), did not provide any discrete ringopening products.
We next examined a model reaction adding ethyl isocyanate to succinic anhydride following Wilfred's strategy, 22 as depicted in Scheme 5. Interestingly, an oxazole intermediate was formed, which after hydrolysis with 6 N HCl, afforded 5-ALA in 64% yield for two steps (Scheme 5a).Thus, this set of reactions provides a new, high-yield synthesis of 5-ALA•HCl 21 from readily available, inexpensive starting materials.Unfortunately, fluorosuccinic anhydride 15 did not survive under the reaction conditions (Scheme 5b).
Next, a different strategy for the synthesis of R-and S-3F-5-ALA was envisaged from 2-deoxyribose, as shown in Scheme 6.In this approach (the Scheme shows only one of the enantiomers), the enantiomers of 3F-5-ALA would be obtained by overoxidation of 27 using Jones' reagent followed by reduction of the azide to the amine in an acidic medium (Scheme 6).The precursor 27 would be derived from the nucleophilic fluorination of 26, the latter produced from 2deoxyribose in four steps. 23As in our previous strategy, the addition of various enolates to fluorosuccinic anhydride proved difficult (Scheme 4).However, contrary to the previous approach, protecting the 2-deoxyribose as cyclic acetal did prevent the elimination and provided the necessary framework.
As outlined in Scheme 6, the new synthesis commenced with the preparation of the known intermediate 25 in three steps.Thus, following the protocol detailed by Gottschaldt, 23 commercially available 2-deoxy-D-ribose was treated with catalytic amounts of sulfuric acid in dry methanol, and furnished the cyclic acetal 23 (74%) as a mixture of α+β anomers.In practice, both anomers were utilized since the anomeric center is oxidized later in the synthesis.Conversion of compound 23 into the corresponding 5-O-tosyl-intermediate 24 (78%) was readily achieved under standard conditions, and this was then treated with lithium azide in DMF to afford the 5-azido-derivative 25 in 80% yield. 23ubsequent treatment of 25 with triflic anhydride in CH 2 Cl 2 in the presence of pyridine at 0 °C afforded the highly unstable triflate 26, which was used without purification in the next reaction.Attempts to fluorinate triflate 26 using CsF/tBuOH 24 or KF/18-crown-6 25 were unsuccessful, instead leading to the formation of unidentified products.However, reaction of the triflate 25 with TBAF/THF did produce the required product 27 but never in yields better than 10% (Scheme 6).Although we were able to access the fluorinated cyclic acetal 27 using this route, the poor reaction yield precluded its viability.We therefore investigated amino ribose derivatives 30a toward nucleophilic fluorination followed by oxidation, as detailed in Scheme 7.
Thus, tosylate 24 was treated with potassium phthalimide to produce the protected amino ribose derivative 29 (Scheme 7), which was then treated with triflic anhydride to produce the unstable triflate 30a, which could not be purified.Efforts to replace the triflate moiety of 30a with fluoride led to the formation of olefinic byproducts.We hypothesized that a more stable leaving group might survive under the reaction Scheme 7. Attempts to Fluorinate Amino Ribose Derivatives 30a and 30b The Journal of Organic Chemistry conditions and produce the desired product.Consequently, compound 29 was treated with TsCl in the presence of DMAP to produce a stable tosylate 30b in 60% yield (Scheme 7).Then, to obtain the desired fluorinated product 31, we explored the conversion of 30b under various nucleophilic fluorination conditions (CsF/tBuOH, TBAF/ACN at 80 °C and CsF/DMSO and KF/18-crown-6 in ACN 100 °C), but the reaction resulted in either recovery of 30b or formation of elimination byproducts.
Once again, we needed to turn to an alternative synthetic strategy.At this point, it was clear that elimination to olefinic products under the basic nucleophilic fluorination conditions was the most significant obstacle to achieving our synthetic goal.Consequently, we focused on developing an efficient alternative strategy for synthesizing enantiopure 3F-5-ALA derivatives by maintaining strictly neutral or slightly acidic reaction conditions.We thus turned our attention to construction of the ketonic moiety of 5-ALA using the pH-neutral reaction conditions of the Liebeskind−Srogl crosscoupling reaction as shown in Scheme 8. 26,27 To explore this potential, we conducted reactions using fluoro thioester 35 with alpha aminostannane 37 (Scheme 8b).The required ketonic precursor 36 was prepared from the known alcohol 33, which can be obtained from commercially available benzyl L-(or D-) aspartate according to a literature procedure 28 (Scheme 8a).
Fluorination of compound 33 in dichloromethane at −20 °C with DAST resulted in the formation of required product 34 in 48% yield along with the dehydrated olefinic product.However, switching the solvent from dichloromethane to chloroform and maintaining the reaction temperature at 0 °C throughout resulted in a significant improvement in the yield to 69%.Thus, compound 34 was synthesized in 3 steps with one chromatographic purification in 63% overall yield.Substrate 34 was subsequently converted into thioester 35 in a two-step sequence involving hydrolysis of the methyl ester Scheme 8. Successful Synthesis of R-and S-3F-5-ALA derivatives (14a and 14b) The Journal of Organic Chemistry and coupling the resulting acid group with 4-methylthiophenol using EDC•HCl (Scheme 8a).While we attained thioester in 39% overall yield over 5 steps, protection and deprotection of the acid in Scheme 8a were considered inefficient and unnecessary.
Therefore, we slightly modified the scheme by directly coupling carboxylic acid 32 and 4-methylthiophenol using DCC to furnish the thioester 36 in 75% yield (Scheme 8b).Treatment of alcohol 36 with DAST led to the formation of the fluoro thioester 35 (73%) with inversion of configuration (confirmed by X-ray crystallography as shown below).Benzyl tri-n-butylstannylmethylcarbamate 37 was synthesized from trin-butylstannylmethyl iodide in three steps following the literature precedent. 29Then, a pH-neutral Liebeskind-Srogl cross-coupling using 2 equiv of [copper(I) diphenylphosphinate (CuDPP, 2 equiv) 27 in THF at 50 °C provided the desired ketonic coupling product 13a in 91% yield.Compound 13a was then subjected to deprotection hydrogenolysis in an acidic medium to furnish the desired R-3-F-5-ALA 14a in 75% yield.Separately, S-3F-5-ALA 14b was synthesized from benzyl D-aspartate using the same synthetic Scheme 8.The stereochemistry of the R-and S-3-F-5-ALA isomers was assigned by X-ray crystallography, as shown in Figure 2.

■ CONCLUSION
This study reveals novel and scalable synthetic methods for producing racemic and enantiopure 3-fluoro-5-aminolevulinic acid hydrochloride (3F-5-ALA).The synthesis of racemic 3F-5-ALA•HCl (14) involves a six-step process utilizing electrophilic fluorination, resulting in a satisfactory overall yield of 16% from N-benzyloxycarbonylglycine (Z-Gly-OH).Additionally, both enantiomers of 3-F-ALA•HCl (14a and 14b) were successfully constructed in a five-step procedure starting from benzyl L and D-aspartate, respectively, using a Liebeskind− Srogl cross-coupling as the crucial step.Notably, this enantioselective synthesis provided an overall yield of 34%.Moreover, we demonstrate a scalable approach for synthesizing 5-ALA from widely accessible, inexpensive starting materials in 2 steps in 64% yield.The biological evaluation of the 3-F-5-ALA derivatives will be pursued separately and published elsewhere.
■ EXPERIMENTAL SECTION General Information.Unless otherwise noted, all reagents were used as received from commercial suppliers.All reactions were performed under a nitrogen atmosphere and in flame-dried or ovendried glassware with magnetic stirring.All solvents were dried before use, following the standard procedures.Reactions were monitored using thin-layer chromatography (SiO 2 ).TLC plates were visualized with UV light (254 nm), iodine treatment, or ninhydrin stain.Column chromatography was carried out using silica gel (60−120 mesh and 100−200 mesh) packed in glass columns.NMR spectra were recorded at 300, 400, 500 MHz ( 1 H) and at 75, 100, 125 MHz ( 13 C), respectively.Chemical shifts (δ) are reported in ppm, using the residual solvent peak in CDCl 3 ( 1 H, δ = 7.26; 13 C, δ = 77.16ppm) as internal standard, and coupling constants (J) are given in Hz.HRMS were recorded using ESI-TOF techniques.