C5‐Iminosugar modification of casein kinase 1δ lead 3‐(4‐fluorophenyl)‐5‐isopropyl‐4‐(pyridin‐4‐yl)isoxazole promotes enhanced inhibitor affinity and selectivity

The quest for isoform‐selective and specific ATP‐competitive protein kinase inhibitors is of great interest, as inhibitors with these qualities will come with reduced toxicity and improved efficacy. However, creating such inhibitors is very challenging due to the high molecular similarity of kinases ATP active sites. To achieve selectivity for our casein kinase (CK) 1 inhibitor series, we elected to endow our previous CK1δ‐hit, 3‐(4‐fluorophenyl)‐5‐isopropyl‐4‐(pyridin‐4‐yl)isoxazole (1), with chiral iminosugar scaffolds. These scaffolds were attached to C5 of the isoxazole ring, a position deemed favorable to facilitate binding interactions with the ribose pocket/solvent‐open area of the ATP binding pocket of CK1δ. Here, we describe the synthesis of analogs of 1 ((−)‐/(+)‐34, (−)‐/(+)‐48), which were prepared in 13 steps from enantiomerically pure ethyl (3R,4S)‐ and ethyl (3S,4R)‐1‐benzyl‐4‐[(tert‐butyldimethylsilyl)oxy]‐5‐oxopyrrolidine‐3‐carboxylate ((–)‐11 and (+)‐11), respectively. The synthesis involved the coupling of Weinreb amide‐activated chiral pyrrolidine scaffolds with 4‐ and 2‐fluoro‐4‐picoline and reaction of the resulting 4‐picolyl ketone intermediates ((–)‐/(+)‐40 and (–)‐/(+)‐44) with 4‐fluoro‐N‐hydroxybenzenecarboximidoyl chloride to form the desired isoxazole ring. The activity of the compounds against human CK1δ, ‐ε, and ‐α was assessed in recently optimized in vitro assays. Compound (–)‐34 was the most active compound with IC50 values (CK1δ/ε) of 1/8 µM and displayed enhanced selectivity toward CK1δ.


| INTRODUCTION
Since the approval of the first protein kinase inhibitor, imatinib, for clinical use in 2001, the search for new kinase inhibitors has become an area of intense research activity. [1,2] Kinases are physiologically important proteins that are essential for the transmission of cellular signals, for example, from the cell surface to the nucleus. As such, kinases are at the heart of many signaling pathways that regulate a plethora of cellular processes. [3] Among the 555 known kinases, casein kinase 1 δ (CK1δ) is a member of the casein kinase 1 (CK1) family, which consists of seven highly conserved and widely expressed serine/threonine kinase isoforms (α, β, γ1, γ2, γ3, δ, and ε). [4,5] In concert with its family members CK1δ is involved in the regulation of cell cycle progression and apoptosis induction, mitosis, DNA damage response, and circadian rhythm as well as several developmental pathways, such as Hippo and Wnt/β-catenin-mediated signaling. Aberrant CK1δ activity, for example, due to genetic and epigenetic alteration or excessive activation, has been demonstrated in tumorigenesis and tumor progression, as well as in various neurodegenerative diseases, including Alzheimer's and Parkinson's disease, familial advanced sleep phase syndrome, amyotrophic lateral sclerosis, and frontotemporal lobar degeneration. [5][6][7][8][9][10] Therefore, CK1δ is pursued as an attractive therapeutic target for the treatment of various types of cancer and the aforementioned neurovegetative diseases.
However, an isoform-selective ATP-competitive CK1 inhibitor has yet to be approved for clinical use. [11] Our approach, to build inhibitors with greater selectivity toward CK1δ, is to utilize scaffolds with instilled chirality (i.e., iminosugars), to confer binding selectivity toward the highly homologous kinase domains.
Moreover, a recently solved ligand-protein complex of a structurally similar imidazole-based inhibitor (Figure 2a; pdb code 5MQV) revealed the presence of a water network bridging binding interactions between the ligand and the ribose pocket. [13] The presence of such a water network at the solvent open area of the ATP binding pocket is not uncommon and has also been observed by others (e.g., pdb code 3UYT, not shown). [14] In an attempt to harness these water-mediated interactions for the development of an isoform-selective CK1δ inhibitor, we subsequently explored the possibility of extending the central pharmacophore toward interaction with the solvent-oriented ribose pocket by modifying compound 2 with iminosugars. [15] Thus, we designed compounds 3 and 4 in which the double bond of the inherently unstable acrylamide side chain of 2 was replaced with a pyrrole, whereby the pyrrole also served as an anchor point for the iminoribitol moiety ( Figure 1). As a result, compounds 3 and 4 emerged as nanomolar dual inhibitors of CK1δ/ε. Although we were able to show that the iminosugar motif was important for activity, application of the enantiomeric pure iminoribitols imparted only a negligible impact on affinity and isoform selectivity. Compounds 5 F I G U R E 1 The vicinal pyridin-4-yl/4-fluorophenyl isoxazole motif as core structure toward the development of optimized ATP-competitive CK1 inhibitors by means of attaching chiral scaffolds acting as specificity-mediating moieties F I G U R E 2 X-ray-defined binding poses of hit compound (a) (pdb 5MQV) showing a water network toward the ribose pocket, and compound 5 (b) (pdb 6F1W) as ligands in the ATP binding pocket of CK1δ. (c) Docking pose of the designed compound (-)-34 in the ATP binding pocket based on CK1δ pdb structure 5MQV. The chiral hydroxy-pyrrolidine scaffold attached to the central isoxazole pharmacophore is involved in H-bonds within the ATP binding pocket, addressing Asp91 toward the ribose pocket, and Ile15 at the ceiling of CK1δ's active site von DRATHEN ET AL. | 3 of 22 and 6, which were obtained from 3 and its enantiomer 4 by a Pictet-Spengler cyclization, were found to be more potent and more isoform-selective (five-to an eight-fold preference for CK1δ over CK1ε) than their noncyclized counterparts. However, X-ray crystallographic analysis of the respective ligand-protein complexes revealed that the chiral regions of the molecules (5/6) were positioned too far from the ATP binding site to undergo effective binding interactions ( Figure 2b). These results prompted us to pursue relocating the iminosugar unit from the side chain (position A) into position B (C5) of the isoxazole ring ( Figure 3). Thus, the polar iminosugar moiety was deemed to be well-positioned for interaction with the solvent-open area of the ATP binding pocket. This notion was supported by an initial in-silico study that showed compound (-)-34 was able to undergo hydrogen bonding with Asp91 and Ile15 (Figure 2c).
Similarly, docking of its enantiomer (+)-34 predicted a reversed bonding pattern, whereby the hydroxy group of the iminosugar underwent hydrogen bonding with Asp132 and the pyrrolidine nitrogen with Ile15 ( Figure S1).
Motivated by the in-silico results, we pursued the synthesis of these chiral compounds, (-)-and (+)-34, as well as experimental evaluation of the predicted ligand affinity. Accordingly, in this study, we report on our efforts to synthesize analogs of 1, in which the 5-iso-propyl substituent is replaced with an iminosugar (7) (Figure 3).
Our synthetic strategy relies on establishing the isoxazole ring at the final stages of the synthesis, by reaction of the 4-picolyl ketone intermediate 9 with 4-fluoro-N-hydroxybenzenecarboximidoyl chloride (8), using our previously described reaction conditions. [16] To obtain the required ketone intermediate 9, we initially explored the addition of 4-picoline to the corresponding acetaldehyde-equipped iminosugar 10a followed by subsequent oxidation of the resulting alcohol to ketone 9. [17] However, due to a lack of success in executing this envisaged oxidation, we instead employed the respective Weinreb amide (10b) to enable the formation of 9. [18] Aldehyde and Weinreb amides 10a-b were prepared from the enantiomeric ethyl esters (−)-11 and (+)-11 using a sequence of standard transformations.
Finally, all iminosugar-modified analogs of the lead compound 1 were tested for their activity against human CK1δ, -ε, and -α in established in vitro assays. [19] 2 | RESULTS AND DISCUSSION
Separation of the two enantiomers was achieved by selective, enzymatic ester hydrolysis with Novozyme 435. The resultant carboxylic acid (+)-15 was subsequently re-esterified under standard reaction conditions. Both ethyl esters ((-)-11/(+)-11) were then silyl protected and the lactams reduced with borane dimethyl sulfide in tetrahydrofuran at 70°C. [21,22] The reduction, however, gave rise to a mixture of diastereomeric amine-borane adducts that could be isolated as stable compounds. [21,22] Although methanol has been reported as a suitable means to remove the excess borane, we found that decomposition of the adduct worked equally well when a solution of the crude reduction product in N,N-dimethylformamide was F I G U R E 3 Conceptualization and synthetic strategy heated in the presence of water. [23] The resulting alcohols ((-)-18 /(+)-18) were then activated as their corresponding mesylates ((-)-20/(+)-20) and displacement with cyanide produced the respective nitriles (-)-21 and (+)-21 in good yields. [24] Treatment of the nitriles with diisobutylaluminum hydride (DIBAL) furnished the desired aldehydes (-)-22 and (+)-22 and subsequent reaction of the aldehyde (-)-22 with the anion of 4-picoline resulted in the formation of a 1.2:1 diastereomeric mixture of the alcohols 23a and 23b. [17,25] Both alcohols could be separated chromatographically and hence were individually characterized, but the exact configuration of the newly formed alcohol function in each of the diastereomers remained elu-  [26][27][28] And exchanging the N-protecting group did not facilitate the formation of any oxidized product either.

| Biology
To examine the inhibitory properties of compounds 34, 39, and 48 toward the casein kinase 1 family members CK1δ, ε, and α, we employed suitable in vitro kinase assay conditions that have recently been optimized for the evaluation of CK1. [19] The assay enables the differentiation between kinase autophosphorylation and substrate phosphorylation, thus limiting the accumulation of false-positive responses. In essence, the assay relies on the transfer of the radiolabeled phosphate group from [γ-32 P]-ATP to a substrate protein and careful measurement of each kinases' initial velocity region as well as defining an ATP concentration equal to the corresponding K m (ATP) . Accordingly, the enzymatic reaction was performed by taking these latter parameters into account, thereby achieving a high level of comparability.
Among all compounds tested in this study, we identified com- To a stirred solution of ethyl ester (-)-11 (19.6 g, 74.4
The residue was re-dissolved in a mixture of N,N-dimethylformamide (280 ml) and water (7 ml), and the solution was heated at 100°C for 1 h.