Synthesis of Novel Pyridine‐Carboxylates as Small‐Molecule Inhibitors of Human Aspartate/Asparagine‐β‐Hydroxylase

Abstract The human 2‐oxoglutarate (2OG)‐dependent oxygenase aspartate/asparagine‐β‐hydroxylase (AspH) is a potential medicinal chemistry target for anticancer therapy. AspH is present on the cell surface of invasive cancer cells and accepts epidermal growth factor‐like domain (EGFD) substrates with a noncanonical (i. e., Cys 1–2, 3–4, 5–6) disulfide pattern. We report a concise synthesis of C‐3‐substituted derivatives of pyridine‐2,4‐dicarboxylic acid (2,4‐PDCA) as 2OG competitors for use in SAR studies on AspH inhibition. AspH inhibition was assayed by using a mass spectrometry‐based assay with a stable thioether analogue of a natural EGFD AspH substrate. Certain C‐3‐substituted 2,4‐PDCA derivatives were potent AspH inhibitors, manifesting selectivity over some, but not all, other tested human 2OG oxygenases. The results raise questions about the use of pyridine‐carboxylate‐related 2OG analogues as selective functional probes for specific 2OG oxygenases, and should aid in the development of AspH inhibitors suitable for in vivo use.


Introduction
Following the pioneering identification of the procollagen prolyl-residue hydroxylases (CPHs) as Fe II and 2-oxoglutarate (2OG)-dependent oxygenases, [1] related enzymes, which play important roles in human biology, have emerged; some of these are validated medicinal chemistry targets. [2] Human 2OG oxygenases have roles in lipid metabolism, [3] processing proteins destined for secretion, [1][2] histone/chromatin modifications, [4] DNA/RNA damage repair, [5] and hypoxia sensing. [6] Inhibition of the CPHs was pursued for the treatment of fibrotic diseases, but was suspended due to toxicity issues. [7] The CPH inhibition work pioneered the use of 2OG analogues/ competitors such as pyridine-2,4-dicarboxylic acid (2,4-PDCA or 2,4-lutidinic acid, 1; Figure 1) [8] and N-oxalylglycine (NOG, 2, Figure 1). [9] This approach has been successfully applied to the development of the 2OG-dependent hypoxia-inducible transcription factor-α (HIF-α) prolyl-residue hydroxylases (PHDs) inhibitors, [10] some of which are approved for use in the treatment of anemia in patients suffering from chronic kidney disease (e. g., roxadustat 3, [11] Figure 1). With the notable exception of mildronate (meldonium), [12] an inhibitor of γ-butyrobetaine dioxygenase (BBOX), [13] most 2OG oxygenase inhibitors are active-site Fe II ligands/2OG competitors. [10] Examples include pyridine-carboxylate derivative QC-6352 (4, Figure 1), which inhibits the Jmjc lysine-specific demethylase 4 (KDM4, JMJD2) enzymes and which shows anti-proliferative effects in cancer models. [14] The human 2OG oxygenase aspartate/asparagine-β-hydroxylase (AspH, BAH, HAAH) catalyses the hydroxylation of conserved Asp and Asn residues in epidermal growth factor-like domains (EGFDs), [15] which include the extracellular domains of the notch receptor and its ligands. [16] AspH levels are upregulated in multiple cancers (e. g., hepatocellular carcinoma [17] and pancreatic cancer [18] ). AspH is reported to be translocalised from the endoplasmic reticulum (ER) to the cell surface membrane resulting in enhanced tumour invasiveness and a diminished patient survival rate. [17b,19] AspH is also strongly hypoxically regulated [20] and may have a role in hypoxia sensing. [21] AspH is thus an interesting potential target from a cancer medicinal chemistry perspective. [22] Only a few structure-activity relationship (SAR) studies directed at identifying AspH inhibitors have been reported. Initial studies identified several nonselective small-molecule AspH inhibitors in cell-based experiments [25] and l-ascorbate based AspH inhibitors have been used in cellular and animal experiments. [22a,26] Studies focusing on the development of nitrogen-containing heteroaromatic scaffolds for AspH inhibition have not been reported despite the prominence of these scaffolds in approved active pharmaceutical ingredients (APIs), including anticancer drugs and 2OG oxygenase inhibitors. [27] The discrepancy between the potential of a AspH as a medicinal chemistry target and the limited effort of developing efficient small-molecule AspH inhibitors for clinical use likely relates to the lack of efficient assays for isolated AspH.
Recently, we reported crystal structures of recombinant human AspH using truncated constructs. [23] Together with biochemical studies, these reveal that AspH accepts EGFDs with a noncanonical (Cys 1-2, 3-4, 5-6; Figure 2C) disulfide pattern, rather than the well-characterised canonical (Cys 1-3, 2-4, 5-6; Figure 2B) disulfide pattern. [23] These findings prompted us to develop a semi-automated high-throughput assay using solid phase extraction (SPE) coupled to mass spectrometry (MS) and stable disulfide analogues (e. g., hFX-CP 101-119 ; Figure 2D) to monitor AspH activity. [24] Here, we report SAR studies on nitrogen-containing heteroaromatic small-molecule AspH inhibitors. Novel C-3-substituted derivatives of 2,4-PDCA (1), which is a potent, but non-selective AspH inhibitor competing with 2OG for 2OG binding in the AspH active site ( Figure 2E), [24][25] were synthesised and their inhibitory properties investigated using SPE-MS inhibition assays against AspH and other human 2OG oxygenases. The results reveal the potential for potent and selective AspH inhibition, and raise questions regarding the selectivity of PDCA-type and related 2OG oxygenase inhibitors that have been reported in the literature.
The most pronounced effect on inhibitor potency was observed when the aniline substituent at the C-3 position of 2,4-PDCA was changed to 4-methoxybenzylamine: Compound 18 inhibited AspH efficiently, its IC 50 value was only 20 fold above that of 2,4-PDCA 1 (IC 50~0 .6 μM, Table 1). The significantly higher potency of 18 compared to 2,4-PDCA derivatives 8-17 may relate to the increased rotational freedom enabled by the additional methylene-unit in 18.

An improved synthesis of C-3 aminoalkyl-substituted 2,4-PDCA derivatives
Having identified pyridine 18 as a potent and selective (with respect to KDM4E) AspH inhibitor, we aimed to synthesise other C-3 aminoalkyl-substituted 2,4-PDCA derivatives to generate further SAR data and to investigate the selectivity of the inhibitor series with respect to other human 2OG oxygenases. However, the reported synthesis of C-3-substituted 2,4-PDCA derivatives employs harsh reaction conditions and non-selective reactions as manifested by low overall yields (Scheme 1A). [31] Moreover, the HCl adduct of pyridine 18 slowly degraded over time, even when stored at À 20°C.
An optimised route for synthesis of C-3 aminoalkylsubstituted 2,4-PDCA derivatives was developed which shortens the reported synthesis by one step and overcomes the described shortcomings (Scheme 1B). Thus, commercial 2,3-  [a] Mean average of two independent runs (n = 2; mean � standard deviation, SD). AspH inhibition assays were performed as described in the Experimental Section using 50 nM His 6 -AspH 315-758 and 1.0 μM hFX-CP 101-119 ( Figure 2D) as a substrate.
dichloroisonicotinic acid (23) was converted into its methyl ester, which was then submitted to regioselective Pd-catalysed carbonylation [32] yielding dimethyl ester 24 (90 %, over two steps). 1-and 2-dimensional NMR analysis of the crude reaction product 24 indicated full substrate conversion with exclusive formation of the desired regioisomer in the carbonylation reaction as confirmed by single-crystal X-ray diffraction analysis of purified 24.
Buchwald-Hartwig amination [33] of dimethyl ester 24 with alkylamines required optimisation due to the reduced reactivity of its CÀ Cl bond compared to the CÀ Br bond of dimethyl ester 21 (Scheme 1). Undesired reaction pathways (ester saponification and amide bond formation) occurring instead of Buchwald-Hartwig amination under the reported conditions [31] were circumvented by replacing the inorganic base Cs 2 CO 3 with pyridine. Conversion was further improved by substituting the palladium precatalyst Pd 2 (dba) 3 with Pd(OAc) 2 , retaining Xantphos [34] as ligand. Both linear and α-branched alkylamines were successfully applied in the amination reactions whereas α,α-disubstituted alkylamines did not react, presumably due to steric hindrance. The use of alkylamines containing heteroaromatic moieties (e. g., thiophene, furan) was unsuccessful. Crystallographic analysis confirmed the anticipated regioselectivity of the Buchwald-Hartwig reaction ( Figure S1 and Table S1 in the Supporting Information).
The intermediate dimethyl esters were subjected to lithium hydroxide-mediated saponification to afford the desired C-3 aminoalkyl-substituted 2,4-PDCA derivatives 18 and 25-33 ( Figure 4) in good yields; excess base was removed by acidic ion exchange chromatography to yield the salt-free inhibitors which were stable as solids and in aqueous solution (see the Experimental Section).

SAR studies on the inhibition of AspH by C-3 aminoalkyl-substituted 2,4-PDCA derivatives
The SAR studies reveal that increasing the length of the C-3 aminoalkyl chain from methylene to butylene has a detrimental effect on potency (25-28, Figure 5 and Table 2). Introducing a substituent α to the amino group of the alkyl chain increases potency; the effect appears to be more pronounced for shorter alkyl groups (methyl, 29, IC 50~0 .3 μM versus ethyl, 30, IC 501 .2 μM; Table 2). Note that for pyridines 29 and 30 racemic mixtures were tested. Potency was reduced with a saturated C-3 aminoalkyl chain (33) compared to a C-3 aminoalkyl side chain bearing an aromatic residue (25). This might relate to increased steric bulk or lack of π-π stacking effects. Overall para-substituted aromatic moieties in the C-3 aminoalkyl side chain (18,31,32) increase potency compared to nonsubstituted aromatics (25, 26). This effect is more pronounced Scheme 1. A) The reported synthesis of C-3 aminoalkyl-substituted 2,4-PDCA derivative 18 [31]

Selectivity studies
To investigate the selectivity of the C-3 functionalised 2,4-PDCA derivatives for AspH, their inhibitory activities against the isolated human 2OG oxygenases PHD2, which is a validated medicinal chemistry target, [35] factor inhibiting HIF-α (FIH), and the Jmjc histone demethylase KDM4E were determined. The inhibition assays all employed SPE-MS to help ensure comparability with the AspH results and minimise possible discrepancies due to different assay methods ( Table 3).

Conclusions
The results described herein demonstrate that it is possible to efficiently modulate the potent AspH inhibitor 2,4-PDCA (1, Figure 1) using a novel four-step synthetic sequence (Scheme 1), which includes two palladium-catalysed reactions, to identify compounds of approximately equal potency. Out of the eight most potent identified AspH inhibitors, 29 is the most efficient AspH inhibitor (IC 50~0 .3 μM); C-3 aminoalkyl-substituted 2,4-PDCA derivative 26 (IC 50~7 .7 μM) is the least efficient (Table 3). Although further work is required, the results reveal that it should be possible to develop small-molecule AspH inhibitors of suitable potency and selectivity for in vivo studies aimed at validating AspH as a medicinal chemistry target to develop novel cancer therapeutics and at exploring the function of EGFD hydroxylation in greater detail. Considering that AspH is translocated to the cell membrane of invasive cancer cells, [19] the dicarboxylic acid motif of the 2,4-PDCA derivatives synthesised might be beneficial to minimise the cell-wall permeability of the inhibitors and thus reducing the possibility of undesired off-target effects through inhibiting other 2OG oxygenases, such as the Jmjc KDMs, in cells.
In terms of the selectivity of the C-3-substituted 2,4-PDCA derivatives, the results presented here demonstrate a substantial overlap between AspH inhibition with that of KDM4E (Table 3). By implication, this will likely extend to at least the other Jmjc KDM4 enzymes (i. e., human KDM4A-D). Crystallographic analysis of the active site structures of the two types of 2OG oxygenases suggest that it should be possible to develop selective inhibitors for AspH or the KDM4 class of 2OG oxygenases. [23,43] Now that a reliable assay for isolated AspH has been established [24] and Jmjc KDMs including KDM4E are actively being pursued as medicinal chemistry targets with several pyridine-based and related small-molecule inhibitors for cancer treatment, [44] AspH should be included in Jmjc KDM selectivity counter-screens in order to develop improved inhibitors and safe medicines.
It should also be noted that 2,4-PDCA (1) [24] and 2,3-PDCA (quinolinic acid, 5) are potent AspH inhibitors. The latter observation raises the possibility of natural inhibition of AspH by small-molecules such as quinolinic acid. A prodrug form of 2,4-PDCA might be useful in probing AspH function in vivo bearing in mind possible "off-target" effects through the inhibition of other 2OG oxygenases including the Jmjc KDMs (Table 3). In this regard, it is notable that dimethyl Noxalylglycine (DMOG), a prodrug form of NOG (2, Figure 1), has been used to mimic the cellular hypoxic response by inhibiting the PHDs, though it inhibits other 2OG oxygenases and other enzymes.
[20] Like 2,3-PDCA, NOG is a natural product, at least in plants. [45] NOG is a less potent inhibitor of AspH and the Jmjc KDMs than 2,4-PDCA; [24] conversely, 2,4-PDCA inhibits the PHDs significantly less efficient than AspH (Table 3). Thus, providing care is taken to consider "off-target" effects, incompletely selective small-molecule inhibitors can be of use in biological functional assignment work.

Experimental Section
General information: All reagents were from commercial sources (Sigma-Aldrich, Inc.; Fluorochem Ltd; Tokyo Chemical Industries; Alfa Aesar) and used as received unless otherwise stated. 2,4-PDCA (1) and its regioisomers (5-7) were from Sigma-Aldrich. The synthesis and characterisation of 2,4-PDCA derivatives 8-17 has been described elsewhere. [31] Anhydrous solvents were from Sigma-Aldrich, Inc. and kept under an atmosphere of nitrogen. Solvents, liquids, and solutions were transferred using nitrogen-flushed stainless steel needles and syringes. All reactions were carried out under an atmosphere of nitrogen unless stated otherwise. Milli-Q ultrapure (MQ-grade) water was used for buffers; LCMS grade solvents (Merck) were used for solid phase extraction (SPE)-MS.
Purifications were performed using an automated Biotage Isolera One purification machine (wavelength monitored: 254 and 280 nm) equipped with pre-packed Biotage® SNAP KP-Sil or Biotage® SNAP Ultra flash chromatography cartridges. The cartridge size and solvent gradients (in column volumes, CV) used, are specified in the individual experimental procedures. HPLC grade solvents (ethyl acetate and cyclohexane; Sigma-Aldrich Inc.) were used for purifications, reaction work-ups, and extractions.
Thin-layer chromatography (TLC) was carried out using Merck silica gel 60 F 254 TLC plates and visualised under UV light. Melting points (MP) were determined using a Stuart SMP-40 automated melting point apparatus. Infrared (IR) spectroscopy was performed using a Bruker Tensor-27 Fourier transform infrared (FTIR) spectrometer. High-resolution mass spectrometry (HRMS) was performed using electro-spray ionisation (ESI) mass spectrometry (MS) in the positive or negative ionisation modes employing a Thermo Scientific Exactive mass spectrometer (ThermoFisher Scientific); data are presented as a mass-to-charge ratio (m/z) in Daltons.
Single crystal X-ray diffraction data were collected using an Oxford Diffraction SuperNova diffractometer (Rigaku). Structures were solved using SUPERFLIP software [46] before refinement with the CRYSTALS software suite. [47] Crystallographic data can be obtained from the Cambridge Crystallographic Data Centre (CCDC 1988141-42).
Nuclear magnetic resonance (NMR) spectroscopy was performed using a Bruker AVANCE AVIIIHD 600 machine equipped with a 5 mm BB-F/1H Prodigy N2 cryoprobe. Chemical shifts for protons are reported in parts per million (ppm) downfield from tetramethylsilane and are referenced to residual protium in the NMR solvent (CDCl 3 : δ = 7.28 ppm; D 2 O: δ = 4.79 ppm). For 13 C NMR, chemical shifts are reported in the scale relative to the NMR solvent (i. e., CDCl 3 : δ = 77.00 ppm). For 19 F NMR, chemical shifts are reported in ChemMedChem Full Papers doi.org/10.1002/cmdc.202000147 the scale relative to CFCl 3 ; coupling constants are accurate to 0.1 Hz. The number of C atoms in brackets indicates overlapping signals in 13 C NMR; chemical shift numbers in brackets indicate close signals that can be differentiated taking into account second respectively third decimal numbers.

General Procedure B:
To a solution of the dimethyl 3-aminoalkylpyridine-2,4-dicarboxylate (1.0 equiv) in methanol (0.2 M, HPLC grade) was added an aqueous solution of lithium hydroxide (0.4 M, 2.8 equiv) under an ambient atmosphere at 0°C. The reaction mixture was allowed to slowly warm to ambient temperature overnight (14-18 h); the methanol was then removed under reduced pressure. The solution was extracted three times with CH 2 Cl 2 (the organic extracts were discarded); the aqueous phase was acidified (pH~7.7) by adding Dowex® 50XW8 (H + -form, mesh 200-400), filtered, and lyophilised to afford the solid C-3-aminoalkyl-substituted pyridine-2,4-dicarboxylic acid. The crude product was sufficiently purified as judged by 1 H and 13 C NMR and used without further purification in the biological assays.

Methyl 2,3-dichloroisonicotinate (23):
To a solution of 2,3dichloroisonicotinic acid (12.3 g, 63.8 mmol, 1.0 equiv) in anhydrous methanol (125 mL) was added dropwise thionyl chloride (7.0 mL, 95.6 mmol, 1.5 equiv), at ambient temperature under a nitrogen atmosphere. The reaction mixture was stirred under reflux for 2 h, then cooled to ambient temperature and concentrated. The residue was dissolved in ethyl acetate, washed twice with saturated aqueous NaHCO 3 solution, then once with brine; The organic solution was dried over anhydrous Na 2 SO 4 , filtered, evaporated, and purified by column chromatography (100 g KP-Sil; 80 mL/min; 100 % cyclohexane (2 column volumes, CV) followed by a linear gradient (7 CV): 0!40 % ethyl acetate in cyclohexane) to afford 11.6 g (89 %) of purified methyl ester 23. White solid, m.  ures. The Schlenk tube was then sealed under CO-pressure (~1.5-2.0 atm) and placed in a sand bath, which was heated under stirring behind a safety shield at 100°C for 12 h. The reaction mixture was cooled to ambient temperature, then concentrated and purified by column chromatography (50 g KP-Sil; 60 mL/min; 100 % cyclohexane (3 CV) followed by a linear gradient (10 CV): 0!30 % ethyl acetate in cyclohexane) to afford 2.16 g (94 %) the pure dimethyl ester 24. Single-crystals suitable for X-ray diffraction analysis were obtained from a concentrated solution of analytically pure 24 in cyclohexane/CH 2 Cl 2 by slow solvent evaporation at ambient temperature and atmosphere. CCDC 1988141 contains the complete supplementary crystallographic data file; selected crystallographic data are shown in Table S1. White solid, m.p.: 60-62°C; 1

rac-Dimethyl
Solutions of the 2,4-PDCA derivatives (100 % DMSO) were dry dispensed across 384-well polypropylene assay plates (Greiner) in a 3-fold and 11-point dilution series (100 μM top concentration) using an ECHO 550 acoustic dispenser (Labcyte). DMSO and 2,4-PDCA (1) were used as negative and positive controls. The DMSO concentration was kept constant at 0.5 % (v/v) throughout all experiments (using the DMSO backfill option of the acoustic dispenser). Each reaction was performed in technical duplicates in adjacent wells of the assay plates; additionally, assays were performed in two independent duplicates on different days using different DMSO inhibitor solutions.
The Enzyme Mixture (25 μL/well), containing 0.1 μM His 6 -AspH 315-758 in 50 mM HEPES buffer (pH 7.5), was dispensed across the inhibitorcontaining 384-well assay plates with a multidrop dispenser (ThermoFischer Scientific) at 20°C under ambient atmosphere. The plates were subsequently centrifuged (1000 rpm using a Heraeus Megafuge 16 centrifuge equipped with a M-20 rotor, 15 s) and incubated for 15 min. The Substrate Mixture (25 μL/well), containing 2.0 μM hFX-CP 101-119 , 200 μM LAA, 6.0 μM 2OG, and 4.0 μM FAS in 50 mM HEPES buffer (pH 7.5), was added using the multidrop dispenser. Note: The multidrop dispenser ensured proper mixing of the enzyme and the substrate mixtures which was essential for assay reproducibility. The plates were centrifuged (1000 rpm using a Heraeus Megafuge 16 centrifuge equipped with a M-20 rotor, 15 s) and after incubating for 7 min, the enzyme reaction was stopped by addition of 10 % (v/v) aqueous formic acid (5 μL/well). The plates were centrifuged (1000 rpm using a Heraeus Megafuge 16 centrifuge equipped with a M-20 rotor, 60 s) and analysed by MS.