Aza Analogs of the TRPML1 Inhibitor Estradiol Methyl Ether (EDME)

Estradiol methyl ether (EDME) has recently been described by us as a very potent and subtype-specific inhibitor of the lysosomal cation channel TRPML1. Following the principle of bioisosteres, we worked out efficient synthetic approaches to ring-A aza-analogs of EDME, namely a methoxypyridine and a methoxypyrimidine analog. Both target compounds were obtained in good overall yields in six and eight steps starting from 19-nortestosterone via the oxidative cleavage of ring A followed over several intermediates and with the use of well-selected protective groups by re-cyclization to provide the desired hetero-analogs. The methoxypyridine analog largely retained its TRPML1-inhibitory activity, whereas the methoxypyrimidine analog significantly lost activity.


Introduction
TRPML1 is one of three members (TRPML1-3) of the TRPML cation channels group, a subfamily within the transient receptor potential (TRP) superfamily.As a non-selective lysosomal channel permeable to Ca 2+ , Na + , Fe 2+ , Zn 2+ and other cations, it plays an important role in multiple physiological processes but also in several human diseases.A mutation with loss of function of TRPML1 causes Mucolipidosis Type IV, a neurodegenerative lysosomal storage disorder [1].Furthermore, TRPML1 has gained interest as it is associated to be involved in various processes in different cancers, e.g., melanoma [2] and non-small lung cancer [3], and its influence on cardiovascular [4] and neurodegenerative diseases has been discussed [5].
Therefore, obtaining access to inhibitors and activators for this target as pharmacological tools or even as possible future therapeutic options is of great interest.
Phosphatidylinositol 3,5-bisphosphate (PI(3,5)P2), a major constituent of the lysosomal membrane, has been described as an endogenous activator of all TRPML channels, while phosphatidylinositol 4-5-bisphosphate (PI(4,5)P2), which is mainly found in the plasma membrane, has been identified as endogenous inhibitor of TRPML1 and TRPML3 [6].Due to their structural characteristics (polarity), these two molecules are not suitable as pharmacological tools as they cannot permeate cell membranes.
As a consequence, several low-molecular activators and inhibitors of TRPML1 with suitable pharmacokinetic properties have been developed in recent years.While MK6-83 [7], SF-51 and ML-SA1 [8] are examples of unselective TRPML activators, the only known selective activator of TRPML1 is ML1-SA1 [9].Despite the evident need for potent TRPML1 inhibitors, the number of these is still limited.
This inhibitor was identified by random screening of 2,430 compounds on hTRPML1∆NC-YFP, a plasma membrane variant of wild-type TRPML1.Subsequently, we tested other pharmacologically relevant steroidal compounds and found that natural and synthetic steroids lacking aromaticity in ring A (the typical structure motif of estrogens) are virtually inactive (cholesterol, phytosterols, glucocorticoids, mineralocorticoids, antiestrogens, antiandrogens, 5α-reductase inhibitors).In the class of estrogens, the native hormone 17β-estradiol showed significantly reduced inhibition (IC 50 5.3 µM) and only mestranol, a congener of EDME bearing an additional ethinyl group at C-17, showed considerable activity.Inversion of the configuration of the 17α-hydroxy group eliminated inhibitory activity in all cases.Finally, we synthesized ten modified versions of EDME, most of which have a replacement of the methoxy group at C-3 in common with a lipophilic residue.Out of these, the 3-vinylestrane PRU-10 (IC 50 0.41 µM) and the 3-acetyl derivative PRU-12 (IC 50 0.28 µM) (Figure 1) showed stronger TRPML1 inhibition, an improved selectivity profile compared to EDME, and reduced estrogenic activity.
potent (IC50 0.6 μM) and subtype-selective TRPML1 inhibitor in our previous work 1) [13].This inhibitor was identified by random screening of 2,430 compoun hTRPML1ΔNC-YFP, a plasma membrane variant of wild-type TRPML1.Subseq we tested other pharmacologically relevant steroidal compounds and found that and synthetic steroids lacking aromaticity in ring A (the typical structure motif o gens) are virtually inactive (cholesterol, phytosterols, glucocorticoids, mineralocor antiestrogens, antiandrogens, 5α-reductase inhibitors).In the class of estrogens, the hormone 17β-estradiol showed significantly reduced inhibition (IC50 5.3 μM) an mestranol, a congener of EDME bearing an additional ethinyl group at C-17, show siderable activity.Inversion of the configuration of the 17α-hydroxy group elimina hibitory activity in all cases.Finally, we synthesized ten modified versions of EDME of which have a replacement of the methoxy group at C-3 in common with a lip residue.Out of these, the 3-vinylestrane PRU-10 (IC50 0.41 μM) and the 3-acetyl der PRU-12 (IC50 0.28 μM) (Figure 1) showed stronger TRPML1 inhibition, an impro lectivity profile compared to EDME, and reduced estrogenic activity.Based on this preliminary evidence on structure-activity relationships, we fo on additional modifications of ring A of EDME.Due to our own positive experien the synthesis of aza analogs of steroidal lead structures for improving or modulati logical activities [14,15], we decided to investigate a pyridine-type 4-aza analog (1 pyrimidine-type 2,4-diaza analog (2) of EDME.
Following the well-established principle of "bioisosteres" [16], single fun groups in a bioactive molecule can be replaced by other, more or less similar gro order to extend or improve potency, enhance selectivity, alter the physicochemica erties or metabolism, or improve pharmacokinetics or toxicity.The bioisosteric r ment of phenyl rings can be performed in a classical manner with the introduction tral aromatic rings (thiophene, furan) or azaarenes (pyridine, pyrimidine, pyrazin further "nonclassical" biosisosteres (acetylene, bridged aliphatic ring systems) hav developed [18].The azaarene bioisosteres have gained significant interest since th introduce basic properties as well as H-bond acceptor and/or H-bond donor pro and thus improve (or reduce) the interaction with the target protein.Based on this preliminary evidence on structure-activity relationships, we focussed on additional modifications of ring A of EDME.Due to our own positive experience with the synthesis of aza analogs of steroidal lead structures for improving or modulating biological activities [14,15], we decided to investigate a pyridine-type 4-aza analog (1) and a pyrimidine-type 2,4-diaza analog (2) of EDME.
Following the well-established principle of "bioisosteres" [16], single functional groups in a bioactive molecule can be replaced by other, more or less similar groups in order to extend or improve potency, enhance selectivity, alter the physicochemical properties or metabolism, or improve pharmacokinetics or toxicity.The bioisosteric replacement of phenyl rings can be performed in a classical manner with the introduction of neutral aromatic rings (thiophene, furan) or azaarenes (pyridine, pyrimidine, pyrazine) [17], further "nonclassical" biosisosteres (acetylene, bridged aliphatic ring systems) have been developed [18].The azaarene bioisosteres have gained significant interest since they can introduce basic properties as well as H-bond acceptor and/or H-bond donor properties and thus improve (or reduce) the interaction with the target protein.

Chemistry
As we intended to synthesize both target compounds in an enantiomerically pure form, we selected a "chiral pool" approach [19] for our syntheses.Estradiol or EDME were not suitable starting materials for this approach due to the lack of feasible synthetic methods for the conversion of phenols/phenol ethers into pyridines and pyrimidines.The same holds for non-estrogenic sterols bearing a methyl group (C-19) at C-10 since this residue would prevent aromatization of ring A. For our purposes, 19-nortestosterone (nandrolone; 3) was identified as the best precursor for a couple of reasons: this (commercially available and affordable) homochiral compound already has the required configurations at the stereocenters in rings C and D, its ring A is a cyclohexenone that can be cleaved by oxidation, and, last but not least, there is no methyl group at C-10.The published oxidative degradation of 19-nortestosterone (3) under the cleavage of the C-4,C-5 bond and decarboxylation yields a ketocarboxylic acid of type A [20].The formal integration of ammonia and oxidative aromatization should provide a ring A pyridone, which was then to be O-methylated to provide the desired 4-aza analog 1 of EDME.
Oxidative degradation of the propionate side chain in A [20] would provide a ketoaldehyde of type B, which, upon treatment with O-methylisourea, should provide the methoxypyrimidine target compound 2. In both series, temporary protection of the 17-OH group had to be considered (Figure 2).

Chemistry
As we intended to synthesize both target compounds in an enantiomerically pure form, we selected a "chiral pool" approach [19] for our syntheses.Estradiol or EDME were not suitable starting materials for this approach due to the lack of feasible synthetic methods for the conversion of phenols/phenol ethers into pyridines and pyrimidines.The same holds for non-estrogenic sterols bearing a methyl group (C-19) at C-10 since this residue would prevent aromatization of ring A. For our purposes, 19-nortestosterone (nandrolone; 3) was identified as the best precursor for a couple of reasons: this (commercially available and affordable) homochiral compound already has the required configurations at the stereocenters in rings C and D, its ring A is a cyclohexenone that can be cleaved by oxidation, and, last but not least, there is no methyl group at C-10.The published oxidative degradation of 19-nortestosterone (3) under the cleavage of the C-4,C-5 bond and decarboxylation yields a ketocarboxylic acid of type A [20].The formal integration of ammonia and oxidative aromatization should provide a ring A pyridone, which was then to be Omethylated to provide the desired 4-aza analog 1 of EDME.
Oxidative degradation of the propionate side chain in A [20] would provide a ketoaldehyde of type B, which, upon treatment with O-methylisourea, should provide the methoxypyrimidine target compound 2. In both series, temporary protection of the 17-OH group had to be considered (Figure 2).

4-Aza Analog of EDME
Our chiral pool approach started with the oxidative cleavage of ring A of 19nortestosterone (3) to provide ketocarboxylic acid 4. While Holt et al. [20] described an ozonolysis protocol with a yield of 50%, we obtained 4 in a yield of 94.5% by using NaIO4/KMnO4 as the oxidant, a method established for a related degradation of a 19-methyl steroid in the course of the synthesis of the drug finasteride [21].Subsequent treatment with ammonium acetate in acetic acid under reflux [22] resulted in ring closure to two poorly separable unsaturated lactams, 5a with a Δ5,10-and 5b with a Δ5,6 double bond (yield: 87.4%, ratio 5a:5b: 15:85).Unfortunately, we could not find a suitable oxidant for direct dehydrogenation of these lactams to the ring A pyridone 6 (Scheme 1).Most likely, the unprotected 17-OH group interfered with the examined oxidants (DDQ, iodinebased reagents, and others).As a consequence, we examined protective groups for 17-OH.

4-Aza Analog of EDME
Our chiral pool approach started with the oxidative cleavage of ring A of 19-nortestosterone (3) to provide ketocarboxylic acid 4. While Holt et al. [20] described an ozonolysis protocol with a yield of 50%, we obtained 4 in a yield of 94.5% by using NaIO 4 /KMnO 4 as the oxidant, a method established for a related degradation of a 19-methyl steroid in the course of the synthesis of the drug finasteride [21].Subsequent treatment with ammonium acetate in acetic acid under reflux [22] resulted in ring closure to two poorly separable unsaturated lactams, 5a with a ∆5,10and 5b with a ∆5,6 double bond (yield: 87.4%, ratio 5a:5b: 15:85).Unfortunately, we could not find a suitable oxidant for direct dehydrogenation of these lactams to the ring A pyridone 6 (Scheme 1).Most likely, the unprotected 17-OH group interfered with the examined oxidants (DDQ, iodine-based reagents, and others).As a consequence, we examined protective groups for 17-OH.Scheme 1.First attempt for the synthesis of the 4-aza analog 1 of EDME.
Our first attempts to utilize MOM protection of the starting material 19-nortestosterone (3) failed early due to problems with introducing this protective group.The following experiments using TBDMS protection gave promising results in the early steps (for details, see Supporting Information) but failed due to the instability of the TBDMS ether as soon as experiments were performed under acidic conditions (no details shown).
O-Methylation of 10 to provide the methoxypyridine derivative 11 was achieved in 47.2% yield using iodomethane/Ag2CO3 [32].As the final step, the benzyl protective group at 17-OH had to be removed without affecting the methoxypyridine unit.This step turned out to be more difficult than expected.Under standard O-debenzylation conditions (hydrogenolysis under Pd catalysis), no conversion was achieved; an alternative Pd-catalyzed method using Et3SiH as the reductant [33] failed as well.A published method for the selective cleavage of benzyl ethers utilizing CrCl2/LiI [34] surprisingly led to the cleavage of the methyl ether at the pyridine ring and left the benzyl ether untouched.Pyridone 10 (the precursor of 11) was obtained in a 90% yield.Finally, the desired O-debenzylation was Scheme 1.First attempt for the synthesis of the 4-aza analog 1 of EDME.
Our first attempts to utilize MOM protection of the starting material 19-nortestosterone (3) failed early due to problems with introducing this protective group.The following experiments using TBDMS protection gave promising results in the early steps (for details, see Supporting Information) but failed due to the instability of the TBDMS ether as soon as experiments were performed under acidic conditions (no details shown).
O-Methylation of 10 to provide the methoxypyridine derivative 11 was achieved in 47.2% yield using iodomethane/Ag 2 CO 3 [32].As the final step, the benzyl protective group at 17-OH had to be removed without affecting the methoxypyridine unit.This step turned out to be more difficult than expected.Under standard O-debenzylation conditions (hydrogenolysis under Pd catalysis), no conversion was achieved; an alternative Pd-catalyzed method using Et 3 SiH as the reductant [33] failed as well.A published method for the selective cleavage of benzyl ethers utilizing CrCl 2 /LiI [34] surprisingly led to the cleavage of the methyl ether at the pyridine ring and left the benzyl ether untouched.Pyridone 10 (the precursor of 11) was obtained in a 90% yield.Finally, the desired O-debenzylation was achieved by means of BCl 3 [35].The carbinol 1 was obtained in a 90.5% yield, and the methoxy group at C-3 was not affected (Scheme 2).
Molecules 2023, 28, x FOR PEER REVIEW 5 of 17 achieved by means of BCl3 [35].The carbinol 1 was obtained in a 90.5% yield, and the methoxy group at C-3 was not affected (Scheme 2).
Scheme 2. Successful approach to the 4-aza analog 1 utilizing benzyl protection at 17-OH.

2,4-Diaza (pyrimidine) Analog of EDME
As mentioned above (Figure 2), the methoxypyrimidine motif of the target compound 2 should be built up by cyclocondensation of a ketoaldehyde of type B with Omethylisourea.This approach has, in principle, been published before in a French patent claimed by Roussel Uclaf in 1967 [36]; however, this route started with a fully synthetic precursor [37] of undefined stereochemistry (most likely racemic), and neither full experimental details nor acceptable spectroscopic data on the characterization of intermediates and the final product were presented.
Our chiral pool approach started once again with 19-nortestosterone (3).For this new purpose, the ketocarboxylic acid 4 obtained by oxidative cleavage of ring A (Scheme 1) needed to be degraded further in order to convert the propionate side-chain into a formyl group (see Figure 2) following, in general, a poorly detailed protocol published by Holt et al. [20].For this purpose, ketocarboxylic acid 4 was first converted into its methyl ester 12 via a higher-yielding protocol utilizing iodomethane/Cs2CO3 (96.9% yield), followed by conversion of the keto group into the dioxolane 13 (67.5% yield).Next, and distinct from the Holt protocol, the 17-OH group was protected by conversion into the TBDS ether 14 (74.4% yield), in order to circumvent interference of the acidic 17-OH group with the strong base LDA required for the following step.Then, the methyl propionate side chain was converted into the α,β-unsaturated ester 15 in 72.2% yield by a selenation-selenoxide elimination protocol including treatment with LDA/diphenyldiselenide and oxidation Scheme 2. Successful approach to the 4-aza analog 1 utilizing benzyl protection at 17-OH.

2,4-Diaza (pyrimidine) Analog of EDME
As mentioned above (Figure 2), the methoxypyrimidine motif of the target compound 2 should be built up by cyclocondensation of a ketoaldehyde of type B with O-methylisourea.This approach has, in principle, been published before in a French patent claimed by Roussel Uclaf in 1967 [36]; however, this route started with a fully synthetic precursor [37] of undefined stereochemistry (most likely racemic), and neither full experimental details nor acceptable spectroscopic data on the characterization of intermediates and the final product were presented.
Our chiral pool approach started once again with 19-nortestosterone (3).For this new purpose, the ketocarboxylic acid 4 obtained by oxidative cleavage of ring A (Scheme 1) needed to be degraded further in order to convert the propionate side-chain into a formyl group (see Figure 2) following, in general, a poorly detailed protocol published by Holt et al. [20].For this purpose, ketocarboxylic acid 4 was first converted into its methyl ester 12 via a higher-yielding protocol utilizing iodomethane/Cs 2 CO 3 (96.9%yield), followed by conversion of the keto group into the dioxolane 13 (67.5% yield).Next, and distinct from the Holt protocol, the 17-OH group was protected by conversion into the TBDS ether 14 (74.4% yield), in order to circumvent interference of the acidic 17-OH group with the strong base LDA required for the following step.Then, the methyl propionate side chain was converted into the α,β-unsaturated ester 15 in 72.2% yield by a selenation-selenoxide elimination protocol including treatment with LDA/diphenyldiselenide and oxidation with H 2 O 2 [38], followed by spontaneous elimination.Two-carbon degradation was then per-formed by ozonolysis followed by work-up with dimethyl sulfide to provide the aldehyde 16 in 75.8% yield.The treatment of 16 with acetic acid in THF-water resulted in simultaneous deprotection of the dioxolane and the TBDS ether to provide the ketoaldehyde 17 in 79.2% yield.Finally, treatment with O-methylisourea gave the target methoxypyrimidine 2 in 37.7% yield (Scheme 3).
Molecules 2023, 28, x FOR PEER REVIEW 6 of 17 with H2O2 [38], followed by spontaneous elimination.Two-carbon degradation was then performed by ozonolysis followed by work-up with dimethyl sulfide to provide the aldehyde 16 in 75.8% yield.The treatment of 16 with acetic acid in THF-water resulted in simultaneous deprotection of the dioxolane and the TBDS ether to provide the ketoaldehyde 17 in 79.2% yield.Finally, treatment with O-methylisourea gave the target methoxypyrimidine 2 in 37.7% yield (Scheme 3).

Biological Testing
The two target compounds, pyridine analog 1 and pyrimidine analog 2, as well as the inadvertently obtained pyridone analog 6 were submitted to our previously described [13] test for inhibition of TRPML1 on hTRPML1ΔNC-YFP, a plasma membrane variant of wild-type TRPML1 by means of a fluorimetric Ca 2+ influx assay.The results are shown in Table 1.

Biological Testing
The two target compounds, pyridine analog 1 and pyrimidine analog 2, as well as the inadvertently obtained pyridone analog 6 were submitted to our previously described [13] test for inhibition of TRPML1 on hTRPML1∆NC-YFP, a plasma membrane variant of wild-type TRPML1 by means of a fluorimetric Ca 2+ influx assay.The results are shown in Table 1.Compared to EDME, the 4-aza analog 1 showed slightly reduced TRPML1-inhibitory activity (factor <2 less potent), the 2,4-diaza analog 2; however, it is only a very weak inhibitor, and the pyridone analog 6 is virtually inactive.

Chemistry
All NMR spectra ( 1 H, 13 C, DEPT, H-H-COSY, HSQC, HMBC) were recorded at 23 • C on an Avance III 400 MHz Bruker BioSpin or Avance III 500 MHz Bruker BioSpin instrument (Bruker, Billerica, MA, USA) unless otherwise specified.Chemical shifts δ are stated in parts per million (ppm) and are calibrated using residual protic solvents as an internal reference for proton (CD 2 Cl 2 : δ = 5.32 ppm, MeOD δ = 3.31 ppm, DMSO: δ = 2.50 ppm) and for carbon the central carbon resonance of the solvent (CD 2 Cl 2 : δ = 53.84ppm, MeOD δ = 49.00 ppm, DMSO: δ = 39.52 ppm).Multiplicity is defined as s-singlet, d-doublet, t-triplet, q-quartet, and m-multiplet.NMR spectra were analyzed with the NMR software MestReNova, version 12.0.1-20560(Mestrelab Research S.L., Santiago de Compostela, Spain).Numbering of the carbon atoms in seco-steroids: For the sake of comparability, we kept using the numbering the single carbon atoms had in the intact steroids, since, in the following, the seco-steroidal intermediates were cyclized to the azasteroids later.High-resolution mass spectra were performed by the LMU Mass Spectrometry Service applying a Thermo Finnigan MAT 95 (Thermo Fisher Scientific, Waltham, MA, USA) or Joel MStation Sektorfeld instrument (Peabody, MA, USA) at a core temperature of 250 • C and 70 eV for EI or a Thermo Finnigan LTQ FT Ultra Fourier Transform Ion Cyclotron Resonance device (Thermo Fisher Scientific, Waltham, MA, USA) at 250 • C for ESI.IR spectra were recorded on a Perkin Elmer FT-IR Paragon 1000 instrument (Perkin Elmer, Hong Kong, China) as neat materials.The absorption bands were reported in wave number (cm −1 ) with ATR PRO450-S.Melting points were determined by the open tube capillary method on a Büchi melting point B-540 apparatus and are uncorrected.The HPLC purities were determined using an HP Agilent 1100 HPLC (Agilent, Santa Clara, CA, USA) with a diode array detector at 210 nm and an Agilent Poroshell column (120 EC-C18; 3.0 × 100 mm; 2.7 micron) with acetonitrile/water as eluent.Values for specific rotation (α) were measured at 23 • C at a wavelength of λ = 589 nm (Na-D-line) using a Perkin Elmer 241 Polarimeter instrument (Perkin Elmer, Hong Kong, China).All samples were dissolved in dichloromethane (layer thickness l = 10 cm, concentration c = 0.1 mg/100 mL).All chemicals used were of analytical grade.Isohexane, ethyl acetate and methylene chloride were purified by distillation.All reactions were monitored by thin-layer chromatography (TLC) using pre-coated plastic sheets, POLYGRAM ® SIL G/UV254 from Macherey-Nagel (Düren, Germany).Flash column chromatography was performed on Merck silica gel Si 60 (0.015-0.040 mm).Ozonolysis was performed on an Ozonova Type OG700-10WC (Jeske Ozontechnik, Ruderserg, Germany).

Biological Testing
hTRPML1∆NC-YFP, a plasma membrane variant of wild-type TRPML1, was obtained from HEK293 cells stably expressing plasma membrane-targeted TRPML1 by trypsination and after resuspending in HEPES buffered solution.IC 50 values for TRPML1 inhibition were determined on a fluorescence imaging plate reader built into a robotic liquid handling station (Freedom Evo 150, Tecan, Mannedorf, Switzerland) using the calcium dye Fluo-4/AM (Invitrogen, Thermo Fisher Scientific, Waltham, MA, USA) according to the test procedure described in our previous work [13] in the section Compound screening and generation of concentration-response curves.

Table 1 .
TRPML1-inhibitory activities of lead structure EDME and the three synthesized aza analogs.

Table 1 .
TRPML1-inhibitory activities of lead structure EDME and the three synthesized aza analogs.
acid (60 mL) was stirred and heated at reflux for 4 h.After cooling, it was concentrated under reduced pressure and the remaining residue was poured into water.The precipitate was filtered, washed with water (20 mL) and dissolved in dichloromethane (40 mL).The resulting solution was washed with NaOH (1M, 3 × 20 mL), water (20 mL) and brine (20 mL), filtered over a hydrophobic filter and concentrated in vacuo.