Mapping Aldehyde Dehydrogenase 1A1 Activity using an [18F]Substrate‐Based Approach

Abstract Aldehyde dehydrogenases (ALDHs) catalyze the oxidation of aldehydes to carboxylic acids. Elevated ALDH expression in human cancers is linked to metastases and poor overall survival. Despite ALDH being a poor prognostic factor, the non‐invasive assessment of ALDH activity in vivo has not been possible due to a lack of sensitive and translational imaging agents. Presented in this report are the synthesis and biological evaluation of ALDH1A1‐selective chemical probes composed of an aromatic aldehyde derived from N,N‐diethylamino benzaldehyde (DEAB) linked to a fluorinated pyridine ring either via an amide or amine linkage. Of the focused library of compounds evaluated, N‐ethyl‐6‐(fluoro)‐N‐(4‐formylbenzyl)nicotinamide 4 b was found to have excellent affinity and isozyme selectivity for ALDH1A1 in vitro. Following 18F‐fluorination, [18F]4 b was taken up by colorectal tumor cells and trapped through the conversion to its 18F‐labeled carboxylate product under the action of ALDH. In vivo positron emission tomography revealed high uptake of [18F]4 b in the lungs and liver, with radioactivity cleared through the urinary tract. Oxidation of [18F]4 b, however, was observed in vivo, which may limit the tissue penetration of this first‐in‐class radiotracer.


N-(4-(1,3-dioxan-2-yl)benzyl)-N-ethyl-5-fluoronicotinamide 16
A dry 100 mL round-bottom flask was charged with 5-fluoronicotinic acid (0.423 g, 3.0 mmol), evacuated and refilled with argon (3×). Dry toluene (10 mL) followed by oxalyl chloride(0.380 mg, 0.26 mL, 1.0 equiv), and DMF (2 drops) were injected. The reaction was stirred at room temperature for 5 minutes and at 40 °C for a further hour. Dry methylene chloride (10 mL) was injected, and the reaction cooled to -78 °C. triethyl amine (1.7 mL, 12.0 mmol, 4.0 equiv) was injected followed by a solution of N-(4-(1,3-dioxan-2-yl)benzyl)ethanamine, B, (0.664 g, 3.0 mmol, 1.0 equiv) in dry methylene chloride (2 mL). The reaction was allowed to reach room temperature overnight, and then quenched with water (20 mL) and K 2 CO 3 (10 mL, sat. aq), and then extracted with methylene chloride (50 mL × 3). The combined methylene chloride phases were rinsed with water (30 mL), brine (30 mL) and then dried over MgSO 4 . The organic solvents were removed under reduced pressure to yield a yellow oil.       13   °C. Lithium aluminium hydride (2.0 mL, 1M in THF 2.0 mmol, 2.0 equiv) was injected over 5 minutes and then stirred for one hour more at 0 °C. The reaction was carefully quenched at 0 °C with water (0.1 mL) followed by NaOH (0.1 mL, 15% aq.) and water (0.3 mL). The reaction was stirred for a further 30 minutes at room temperature, diluted with THF (30 mL) and then dried with MgSO 4 . This was filtered over a plug of celite (3 cm × 3 cm) to remove the aluminum salts. The organic solvents were removed under pressure to yield a yellow oil that was dissolved in wet THF (10 mL) and added to a 50 mL round-bottom flask. Conc. HCl was added until acidic (pH~1), and the resulting solution was stirred overnight at room temperature. The reaction was quenched with aqueous K 2 CO 3 (sat.) until alkaline (pH~11), and then extracted with methylene chloride (50 mL × 3). The combined organic phases

N-ethyl-5-fluoro-N-(4-formylbenzyl)pyridine-3-sulfonamide 6
A dry 50 mL round-bottom flask was charged with 5-fluoropyridine-3-sulfonyl chloride, 19, (0.25 g, 1.27 mmol), evacuated and refilled with argon (3×). Dry methylene chloride (10 mL) was injected and the resulting solution was cooled to 0°C. Triethylamine (0.53 mL, 0.386 g 3.8 mmol, 3.0 equiv) was injected followed by a solution of N- in dry methylene chloride (2 mL). The reaction was allowed to reach room temperature overnight, and then quenched with water (15 mL) and K 2 CO 3 (15 mL, sat. aq), and then extracted with methylene chloride (50 mL × 3). The combined organic phases were rinsed with water (30 mL), brine (30 mL) and then dried over MgSO 4 .The organic solvents were removed under reduced pressure to yield a yellow oil that was dissolved in wet THF (10 mL) and added to a 50 mL round-bottom flask. Conc. HCl was added until acidic (pH~1), and the resulting solution was stirred overnight at room temperature. The reaction was quenched with aqueous K 2 CO 3 (sat.) until alkaline (pH~11), and then extracted with methylene chloride (50 mL × 3). The combined organic phases were dried over MgSO 4
Potassium permanganate (0.056 g, 0.357 mmol, 1.25 equiv) in water (1.5 mL) followed by NaOH (1.5 mL, 10%, aq) were injected. The cooling bath was removed and the reaction was stirred at room temperature overnight. The reaction was quenched with sodium thiosulfate (20% aq) until the purple colour disappeared, and the pH was adjusted to 2-3 by the addition of HCl (appx. 2 mL, 2N, aq), and then extracted with ethyl acetate (25 mL × 3). The combined organic phases were dried  13   Ar-H). 13 (Table S1). After cooling, the reaction was quenched with water (1.0 mL) and the mixture was analysed by radio-HPLC on an Eclipse® C18 4.6 × 100 mm, 5 µm HPLC column using water and methanol (gradient elution with a flow rate of 1.8 mL/min starting with 25% methanol content, then increased to 50% in 7 min; Rt ≈ 6.2 min). After removing the solvent by heating at 90 °C under a stream of nitrogen, dry acetonitrile (0.5 mL) was added, and the distillation was continued at 90 °C. This procedure was repeated once and the reaction vial was subsequently capped. The chloro-precursor 7 (0.5-5 mg) dissolved in anhydrous DMSO (0.6 ± 0.05 mL) was added and the mixture was stirred for 25 min at 150 °C. After cooling the reaction to 110 °C, aqueous HCl (0.5 mL, 1M) was injected and the reaction stirred for 5 minutes. The reaction was cooled to room temperature following which it was quenched by the addition of aqueous NH 4 OAc (1.0 mL, 0.02M) and purified via semi preparative HPLC using a ZORBAX ® StableBond 300 C18, 9.4 x 250 mm, 5 µm HPLC column at room temperature and with a flow rate of 3.5 mL/min. The mobile phase consisted of aqueous NH 4 OAc (0.02M) and methanol. Isocratic elution with 36.5% methanol content allowed for isolation of the radioactive product (Rt ≈ 18 min). The isolated fraction was diluted with water (20 mL) containing Na-ascorbate (5 mg and methanol. Isocratic elution with 36.5% methanol content allowed for isolation of the radioactive product (Chromatogram S1, Rt ≈ 18 min).
The decay corrected isolated RCY was 43 ± 1% (n = 3) at the end of HPLC purification and cartridge formulation. The identity of the radiochemical product was confirmed by co-elution with the non-radioactive analogue (Chromatogram S1 and S2). The radiochemical purity of [ 18 F]4b was > 99% and the molar activity of the tracer was 0.6-7.2 GBq/μmol.

DOCKING STUDIES
Sequences for all 19 isozymes of ALDH1A1 were collected form Uniprot [6] and aligned using MUSCLE. [7] Protein structures were compared in UCSF Chimera. [8] Fpocket2.0 [9] was utilized to analyse features of the pocket environment. The diameter of the substrate pocket entrance tunnel was investigated using HOLE2. [10] The substrates of interest (compounds 2, 3a, 3b, 4a, 4b, 5, 6, 9-cis-retinal, 13-cis-retinal and all-trans-retinal) were docked into the receptor structure PDBID: 4WB9 corresponding to ALDH1A1 in the cofactor bound state. PDBID: 4L2O corresponding to inhibitor bound ALDH3A1, and PDBID: 1O01 corresponding to crotonaldehyde and cofactor bound ALDH1A1 were superposed on 4WB9 using the chimera matchmaker command (6). for further analysis were 4WB9, 1O01 and 4L2O for ALDH1A1, ALDH2 and ALDH3A1 respectively. This was because each of these structures contained the cofactor and were thought to be closest to the state that the enzyme would be in when a substrate was bound.
ALDH1A1 was found to share 68% sequence identity with ALDH2 and 27% sequence identity with ALDH3A1 while ALDH3A1 and ALDH2 shared 26% sequence identity, suggesting that ALDH1A1 and ALDH2 are much more similar to each other than they are to ALDH3A1. Analysis of the active site features of the chosen structures with Fpocket2.0 (4) revealed that the ALDH1A1 substrate pocket is slightly higher in hydrophobic density than that of ALDH2 or ALDH3A1 and that the ALDH3A1 substrate pocket has a reversal in polarity compared with ALDH1A1 and ALDH2, from overall positively charged pocket to a negatively charged pocket.
ALDH1A1 was also found to have the widest diameter entrance tunnel for the substrate binding pocket, similarly the tunnel remained wider than that of ALDH2 or ALDH3A1 until very close to the active cysteine residue (Fig. S6). The observation that ALDH1A1 has the largest access tunnel to the active site residue (Fig. S6) might suggest that bulkier and more rigid substrates would be preferentially turned over by ALDH1A1 vs. ALDH2 or ALDH3A1. This may also help to explain why reducing the amide in compound 4a or 4b to a secondary amine in compound 5, maintains enzymatic efficiency at ALDH1A1 to a large extent but significantly increases efficiency at ALDH2 and ALDH3A1. Given the higher degree of flexibility for compound 5, it may be more able to squeeze down a narrow entrance tunnel to the substrate binding site, while the more rigid and bent conformation of compounds 4a and 4b, preclude their access. The wider access tunnel to ALDH1A1 allows it to accommodate bulkier and less flexible ligands. The addition of the pyridyl ring allows the formation of a π-stacking interaction with Y296 in ALDH1A1 leading to higher binding affinities for substrates containing the pyridyl substituent. These