Carbocyclic 30-deoxyadenosine-based highly potent bisubstrate-analog inhibitor of basophilic protein kinases
Abstract—Carbocyclic analogs of 30-deoxyadenosine were synthesized as racemates and the resulting stereoisomers were separated by chromatography on a chiral column. The conjugation of obtained compounds with hexa-(D-arginine) via 6-aminohexanoic acid linker led to a highly potent inhibitor of several basophilic protein kinases with some selectivity towards cAMP-dependent protein kinase.
Protein kinases (PKs) play a key role in the regulation of cellular protein functions. More than 400 human dis- eases have been linked to aberrant PK signaling.1 This fact and the recent success with PK inhibitor-based drugs have made PKs the second largest drug target after G protein-coupled receptors.2 Due to apparent ease of development of high-affinity low molecular weight inhibitors targeted to the well-defined hydropho- bic adenine-binding cleft, and despite serious selectivity problems (additionally to 500 coded PKs in human gen- ome 1500 other proteins are able to bind adenine nucle- otides)1 in combination with high concentration of competing ATP in the cellular milieu, most of the devel- oped and studied inhibitors of PKs are competitive with ATP. The high degree of 3D structural conservation of adenine nucleotide binding site of these proteins makes the development of specific inhibitors a challenging task.
The structure of protein substrate-binding domain of ki- nases is more variable and the development of highly selective peptide inhibitors has been described.3 How- ever, longer peptidic structures are needed for achieving nanomolar potency which leads to problems with cellu- lar transport and stability of the compounds.
Keywords: Protein kinases; Bisubstrate inhibitors; Carbocyclic nucleoside; Oligoarginine.
The development of bisubstrate-analog (biligand) inhib- itors that simultaneously associate with both ATP and protein-binding domains of PK could give selective and potent inhibitors of these dual substrate enzymes.4 A linker enabling optimal positioning of the active frag- ments comprising the bisubstrate inhibitor could lead to a conjugate with kinase-binding energy substantially exceeding the sum of binding energies of the fragments.5
Recently6 we described novel bisubstrate-analog-type inhibitors of PK that consisted of adenosine and oligo- (D-arginine) moieties connected via 6-aminohexanoic acid linker (ARCs, adenosine–oligoarginine conjugates). The most active of the compounds, ARC-902 (number 5 in Table 1 of Ref. 6), showed low nanomolar inhibitory potency toward the catalytic subunit (type a) of cAMP- dependent protein kinase (cAPK Ca, also known as PKA Ca), and high proteolytic stability.6 Later we dem- onstrated that such highly potent ARCs really behave as bisubstrate-analog inhibitors and can be displaced from their complex with cAPK Ca by both ATP- and protein substrate-competitive inhibitors.7
The further increase of inhibitory potency of ARC-type inhibitors could be achieved by optimization of struc- tures of the moieties and their spatial positioning. The earlier structure–activity studies with adenosine and adenine nucleotide analogs have shown that contrary to other modifications of the structure that decrease the affinity of adenosine analogs, the removal of the hydroxyl group from the 30-carbon of the ribose moiety of the ribose and the kinase. These facts point to the possibility of the use of chemically and enzymatically more stable9 carbocyclic (cyclopentane) adenosine mim- ics for the design of potent ARC-type inhibitors.
Here we present the synthesis of cyclopentane-based car- bocyclic analogs of 30-deoxyadenosine (30-deoxyariste- romycin),
chromatographic separation of the formed stereoisomers, their conjugation with oligo-(D-argi- nine)-containing peptides, and characterization of the latter compounds as highly potent inhibitors of baso- philic protein kinases.
The shortest synthetic scheme was selected to prepare the derivatives of carbocyclic 30-deoxyadenosine 1a, 1b, 2a, and 2b that are applicable for easy conjugation with peptides. 3-Cyclopentene-1-carboxylic acid (3) was converted to the methyl ester 4 and the obtained es- ter was oxidized to the appropriate epoxide (5a/5b ratio 3:1) with mCPBA (Scheme 1). Cis and trans isomers (5a
and 5b) of the epoxide were separated by column chroently, the low yield of the synthesis resulted from insta- bility of the starting epoxides (5a and 5b) in the reaction medium. The pairs of diastereomers (1a/1b and 2a/2b) were separated by column chromatography on silica gel. The overlapping fractions were purified repeatedly. Structural elucidation of the diastereomers was based on different NOE signals between 10- or 20-protons and 40-proton of the isomers. The diastereomer 1a/1b eluted faster than 2a/2b in case of normal phase chroma- tography, which is consistent with the previous data concerning ribose counterparts of the compounds.14 Enantiomers 1a and 1b were separated by HPLC on a chiral chromatography column (Chiralpak AD-H).
Scheme 1. Synthesis of the carbocyclic nucleoside analogs.
The inhibitory potencies of the synthesized compounds were evaluated against cAPK Ca with the application of fluorometric TLC kinase activity assay,17 and the re- sults are expressed as the IC50 values in Table 1. TAM- RA-kempide (30 lM) and ATP (0.1 or 1.0 mM) were used as substrates. The application of higher concentra- tion of ATP (1.0 mM) in the test system increased the IC50 value and enabled to avoid tight-binding conditions of the assay (IC50 6 Ckinase).18 The racemic nucleoside analogs 1a/1b and 2a/2b showed expectedly weak inhib- itory potencies. The IC50 values for these compounds were comparable to that of adenosine (IC50 = 350 lM). The diastereomer 2a/2b, being structurally less similar to the native adenosine than 1a/1b, showed slightly stron- ger inhibition. The inhibitory potency of the conjugates of these racemates with hexa-(D-arginine) differed by more than 30-fold, and ARC-659-1a/1b (structurally more similar to adenosine) revealed much higher po- tency. This difference in activity may originate from the more effective positioning of the 6-aminohexanoic acid linker and the peptide in complex with the enzyme.
The separated enantiomer 1b was ca. 10-fold more po- tent than 1a. This result indicates the importance of cor- rect positioning of 20-hydroxy group for inhibitor binding to the enzyme. The conjugates with peptides re- vealed similar one-magnitude difference in activity. To the best of our knowledge ARC-659-1b is the most potent bisubstrate-analog inhibitor of protein kinases described so far.
The selectivity of ARC-659-1b was tested (on the com- mercial basis at the Division of Signal Transduction Therapy, University of Dundee) against the panel of protein kinases. Inhibition efficiency was determined at two concentrations of ARC-659-1b (1 and 0.01 lM).The inhibition data for 34 kinases are presented as the percent of residual activity of the kinase in the presence of the inhibitor relative to control incubations where the inhibitor was omitted (Table 2). The analysis of the data reveals that ARC-659-1b most potently inhibits cAPK. The Ki value of 0.5 nM was estimated for this PK (Ta- ble 1; calculated according to the Cheng–Prusoff equa- tion21 with Km of 20 lM for ATP). The compound also showed strong inhibition (ca. 50% at 10 nM concen- tration) of ROCK-II, MSK1, and PRK2. Similar inhibition potency was determined toward Akt/PKB, SGK1, and CAMK1. The moderately inhibited kinases (>90% of inhibition at 1 lM concentration of the inhibitor) were PIM1, S6K1, PAK5, MELK, and AMPK. Some other kinases retained higher activity in the presence of ARC-659-1b and several tested kinases were not inhibited at all. Generally, the inhibition profile of ARC-659-1b was similar to that of the previously re- ported conjugate of adenosine and oligo-(D-arginine), ARC-902.6 However, ARC-659-1b has remarkable cAPK selectivity while its adenosine counterpart ARC- 902 revealed preference for ROCK-II. Several kinases of the CAMK group like AMPK, PIM-2, and CAMK1 were inhibited more strongly by ARC-659-1b, whereas ARC-902 was more potent towards the kinases of the RSK group. PAK5 and MELK kinases (both known as basophilic kinases) were also strongly inhibited by ARC-659-1b.
In conclusion, we have accomplished synthesis of the carbocyclic analog of 30-deoxyadenosine along with sep- aration and characterization of its stereoisomers. The conjugation of these compounds with hexa-(D-arginine) peptide led to potent inhibitors of basophilic PKs. The stereochemistry of the nucleoside part of most active conjugate ARC-659-1b is similar to that of adenosine, but the novel conjugate does not incorporate a native nucleoside with a chemically and enzymatically degrad- able glycosidic bond. The selectivity study revealed that ARC-659-1b strongly inhibited basophilic kinases and was almost inactive toward a number of PKs that do not contain multiple basic amino acids (arginine and/ or lysine) Hexa-D-arginine residues in their consensus phosphorylation sequences.