Adamantane-Substituted Purines and Their β-Cyclodextrin Complexes: Synthesis and Biological Activity

Cyclin-dependent kinases (CDKs) play an important role in the cell-division cycle. Synthetic inhibitors of CDKs are based on 2,6,9-trisubstituted purines and are developed as potential anticancer drugs; however, they have low solubility in water. In this study, we proved that the pharmaco-chemical properties of purine-based inhibitors can be improved by appropriate substitution with the adamantane moiety. We prepared ten new purine derivatives with adamantane skeletons that were linked at position 6 using phenylene spacers of variable geometry and polarity. We demonstrated that the adamantane skeleton does not compromise the biological activity, and some of the new purines displayed even higher inhibition activity towards CDK2/cyclin E than the parental compounds. These findings were supported by a docking study, which showed an adamantane scaffold inside the binding pocket participating in the complex stabilisation with non-polar interactions. In addition, we demonstrated that β-cyclodextrin (CD) increases the drug’s solubility in water, although this is at the cost of reducing the biochemical and cellular effect. Most likely, the drug concentration, which is necessary for target engagement, was decreased by competitive drug binding within the complex with β-CD.


Introduction
Cyclin-dependent kinases (CDKs) are a family of serine/threonine protein kinases that play a crucial role in controlling the cell division cycle in all eukaryotic cells. The activation of CDKs, which represent the catalytic subunit in the heterodimeric complex, usually requires association with a regulatory cyclin subunit [1]. It was found that the excessive activity of cyclins or loss of expression of endogenous CDK protein inhibitors (Cip/Kip or INK4), which results in the deregulation of CDKs activity, is frequently observed in human malignancies. This initiated extensive search for the low-molecular-weight inhibitors of CDKs [2,3]. Over the past three decades, a number of ATP-competitive CDK inhibitors, differing in their chemical structure or selectivity, have been developed. To date, more than 20 CDK inhibitors have been advanced to some stage of clinical evaluation [4]. In 2015, the low-molecular-weight inhibitor of CDK4 and CDK6 palbociclib was approved by the United States Food and Drug Administration for the treatment of breast cancer in clinical use. This drug was followed by the next two CDK4/6 inhibitors, ribociclib and abemaciclib [5].
Purines, appropriately substituted at C2, C6 and C9 centres, represent a class of compounds that strongly inhibit CDKs or other kinases [6]. Structural modifications of 2 of 20 olomoucine, one of the first selective CDK inhibitors [7], provided a number of 2,6,9trisubstituted purines that are highly active towards CDKs, such as bohemine, roscovitine, purvalanol A and olomoucine II ( Figure 1) [8][9][10]. More recently, a novel series of 2substituted-6-biarylmethylamino-9-cyclopentylpurine derivatives exhibiting strong CDK inhibitory activity and antiproliferative effects was described [11]. The success of roscovitine as an anticancer drug candidate even inspired the synthesis and biological evaluation of several types of bioisosteres [12,13].
CDs are water-soluble, non-toxic chemical substances that are easily obtained by the biotransformation of starch. CDs are used across all industries, but most frequently they are applied in pharmacy, medicine, biotechnology, agriculture, food chemistry, cosmetics and textile chemistry [31]. Natural CDs, generally known as α-, β-and γ-, are cyclic oligosaccharides consisting of six to eight glucose units, which are linked via α-(1,4)-glycosidic bonds. CDs have a truncated cone shape with a hydrophobic cavity interior in which a number of lipophilic guest molecules can be included through non-covalent interactions [32]. The formation of host-guest complexes of natural and/or modified CDs (prepared with the transformation of hydroxyl groups) with potential drug candidates has been extensively studied, with the aim of enhancing the water solubility, biological activity and characteristics of pharmaceutical interests, such as, for example, stability and/or bioavailability [33][34][35][36][37][38][39].
CDs are water-soluble, non-toxic chemical substances that are easily obtained by the biotransformation of starch. CDs are used across all industries, but most frequently they are applied in pharmacy, medicine, biotechnology, agriculture, food chemistry, cosmetics and textile chemistry [31]. Natural CDs, generally known as α-, βand γ-, are cyclic oligosaccharides consisting of six to eight glucose units, which are linked via α-(1,4)glycosidic bonds. CDs have a truncated cone shape with a hydrophobic cavity interior in which a number of lipophilic guest molecules can be included through non-covalent interactions [32]. The formation of host-guest complexes of natural and/or modified CDs (prepared with the transformation of hydroxyl groups) with potential drug candidates has been extensively studied, with the aim of enhancing the water solubility, biological activity and characteristics of pharmaceutical interests, such as, for example, stability and/or bioavailability [33][34][35][36][37][38][39].
Subsequently, the previously prepared adamantylated aromatic amines [41,42], or benzylamine, were introduced to position 6 of the purine 2a. The nucleophilic aromatic substitution of chlorine at C6 led to the formation of compounds 3a-k (Scheme 2). These reactions were carried out in the mixture of dimethyl formamide, the corresponding amine and triethylamine, from 80-100 °C [43]. Purines 3a-k were obtained after purifying the crude material using column chromatography with good purity and yields of approximately 41-87%.
The final substitution of chlorine at C2 of compounds 3a-k with 3-aminopropan-1-ol was relatively difficult. Initially, several reaction conditions, such as the utilisation of 4-(dimethylamino)pyridine [44] and 1-methylpyrrolidin-2-one [45], or the application of a stronger base (Hünig's base) in DMSO, [46] were unsuccessful. Under these conditions, the desired product in the reaction mixture was either in a minor portion or not detected at all. Compounds 4a-k were finally prepared under solvent-free conditions [47] with a significant excess of 3-aminopropan-1-ol at 160 °C in 3-6 h (Scheme 2).

ESI-MS Analyses
The ESI-MS was applied, not only for the study of host-guest complexes, but also to confirm the structure of the prepared compounds.
In the positive-ion first-order mass spectra of single purines, the dominant signal at m/z corresponding to the protonated molecule [M + H] + was accompanied by the [M + Na] + ion. It should be noted that the ion corresponding to the formation of dimers was not observed for any prepared purine in the gas phase. On the other hand, the existence of dimers, linked via H-bonds, was confirmed to be in a solid state for purines 3f, 3j and 3k using X-ray diffraction analyses [49,50].
The equimolar mixtures of individual purines with β-CD were prepared immediately before each experiment. In addition to the protonated molecule of the guest [M + H] + , its sodium adduct [M + Na] + and the sodium adduct of β-CD [β-CD + Na] + , and two types of purine·β-CD aggregates (namely [M·β-CD + H] + and [M·β-CD + Na] + ) were detected for all purines bearing the adamantane moiety ( Figure 2a). In the first-order mass spectra of purine 4k (bohemine), only a signal of low intensity, which can be related to the protonated purine·β-CD aggregate, was detected. The existence of host-guest aggregates in the gas phase was confirmed using MS/MS experiments of [M·β-CD + H] + and [M·β-CD + Na] + ions under collision-induced dissociation (CID) conditions. The loss of neutral β-CD from the complex, and the formation of a singly charged product ion at m/z corresponding to Scheme 2. Synthesis of 2,6,9-trisubstituted purines bearing the adamantane moiety.
The alkylation of the commercially available 2,6-dichloro-9H-purine (1) was carried out with 2-iodopropane and potassium carbonate in dimethyl sulfoxide (DMSO) between 15 and 18 • C (Scheme 1), according to the previously published procedure [40]. Although the authors of the original paper claimed that the precise control of the reaction temperature led to the regioselective N9 alkylation, we detected both N9 and N7 isomers in the reaction mixture. The gas chromatography analyses showed a relative ratio for N9 and N7 isomers from 89% to 11%. The required 2,6-dichloro-9-isopropyl-9H-purine was separated from the mixture of both isomers using crystallisation from ethyl acetate/hexane with a 60% yield.
Subsequently, the previously prepared adamantylated aromatic amines [41,42], or benzylamine, were introduced to position 6 of the purine 2a. The nucleophilic aromatic substitution of chlorine at C6 led to the formation of compounds 3a-k (Scheme 2). These reactions were carried out in the mixture of dimethyl formamide, the corresponding amine and triethylamine, from 80-100 • C [43]. Purines 3a-k were obtained after purifying the crude material using column chromatography with good purity and yields of approximately 41-87%.
The final substitution of chlorine at C2 of compounds 3a-k with 3-aminopropan-1-ol was relatively difficult. Initially, several reaction conditions, such as the utilisation of 4-(dimethylamino)pyridine [44] and 1-methylpyrrolidin-2-one [45], or the application of a stronger base (Hünig's base) in DMSO, [46] were unsuccessful. Under these conditions, the desired product in the reaction mixture was either in a minor portion or not detected at all. Compounds 4a-k were finally prepared under solvent-free conditions [47] with a significant excess of 3-aminopropan-1-ol at 160 • C in 3-6 h (Scheme 2).

Inclusion Complexes with β-CD 2.2.1. ESI-MS Analyses
The ESI-MS was applied, not only for the study of host-guest complexes, but also to confirm the structure of the prepared compounds.
In the positive-ion first-order mass spectra of single purines, the dominant signal at m/z corresponding to the protonated molecule [M + H] + was accompanied by the [M + Na] + ion. It should be noted that the ion corresponding to the formation of dimers was not observed for any prepared purine in the gas phase. On the other hand, the existence of dimers, linked via H-bonds, was confirmed to be in a solid state for purines 3f, 3j and 3k using X-ray diffraction analyses [49,50].
The equimolar mixtures of individual purines with β-CD were prepared immediately before each experiment. In addition to the protonated molecule of the guest [M + H] + , its sodium adduct [M + Na] + and the sodium adduct of β-CD [β-CD + Na] + , and two types of purine·β-CD aggregates (namely [M·β-CD + H] + and [M·β-CD + Na] + ) were detected for all purines bearing the adamantane moiety ( Figure 2a). In the first-order mass spectra of purine 4k (bohemine), only a signal of low intensity, which can be related to the protonated purine·β-CD aggregate, was detected. The existence of host-guest aggregates in the gas phase was confirmed using MS/MS experiments of [M·β-CD + H] + and [M·β-CD + Na] + ions under collision-induced dissociation (CID) conditions. The loss of neutral β-CD from the complex, and the formation of a singly charged product ion at m/z corresponding to the protonated guest molecule, represent a typical fragmentation pathway of protonated purine·β-CD aggregates ( Figure 2b). In contrast, the fragmentation of the [M·β-CD + Na] + ion led exclusively to the sodium adduct of β-CD (m/z 1157) in all cases ( Figure 2c). This means that the neutral molecule of the guest was lost by this type of complex. The ESI mass spectra of host-guest complexes for all prepared purines 4a-k are given in the Supplementary Materials ( Figures S31-S40).

1 H-1 H Rotating Frame Nuclear Overhauser Effect Spectroscopy (ROESY) NMR Experiments
Our attempts to demonstrate the supramolecular properties of our purines were compromised by the low solubility of purines in water. Thus, isothermal calorimetric titrations and NMR titration experiments were disabled. However, we succeeded in 2D ROESY experiments to show, unambiguously, the supramolecular nature of purine complexes with β-CD. Figure 3 clearly shows cross-peaks, indicating that hydrogen atoms from the adamantane cage, that is, H(a), H(b) and H(c), interact with H5 and H3 atoms, which are located on the inside of the CD cylinder. This data concerts with the typical arrangement of inclusion complexes of adamantane derivatives and β-CD, where the adamantane cage is positioned inside the cavity of the CD. Combining the NMR and MS data, we can infer that our purines are able to form relatively stable inclusion complexes in water solutions.

1 H-1 H Rotating Frame Nuclear Overhauser Effect Spectroscopy (ROESY) NMR Experiments
Our attempts to demonstrate the supramolecular properties of our purines were compromised by the low solubility of purines in water. Thus, isothermal calorimetric titrations and NMR titration experiments were disabled. However, we succeeded in 2D ROESY experiments to show, unambiguously, the supramolecular nature of purine complexes with β-CD. Figure 3 clearly shows cross-peaks, indicating that hydrogen atoms from the adamantane cage, that is, H(a), H(b) and H(c), interact with H5 and H3 atoms, which are located on the inside of the CD cylinder. This data concerts with the typical arrangement of inclusion complexes of adamantane derivatives and β-CD, where the adamantane cage is positioned inside the cavity of the CD. Combining the NMR and MS data, we can infer that our purines are able to form relatively stable inclusion complexes in water solutions.

CDK Inhibition Activity
The inhibitory potency of purines 4a-k was tested with the recombinant heterodimeric complex CDK2/cyclin E. As can be seen in Table 1, the CDK inhibition activity was studied with both the single purines and their mixtures with β-CD (purine·β-CD, 1:1 and 1:10, molar ratio). Purine 4k (bohemine) was used as a reference compound. Moreover, the activity of single compounds 4a-k was compared with the previously published IC50

CDK Inhibition Activity
The inhibitory potency of purines 4a-k was tested with the recombinant heterodimeric complex CDK2/cyclin E. As can be seen in Table 1, the CDK inhibition activity was studied with both the single purines and their mixtures with β-CD (purine·β-CD, 1:1 and 1:10, molar ratio). Purine 4k (bohemine) was used as a reference compound. Moreover, the activity of single compounds 4a-k was compared with the previously published IC 50 value of roscovitine [19]. The strongest activity has been observed for purines 4b (IC 50 = 0.21 βM) and 4d (IC 50 = 0.58 βM). The activity of purine 4b on CDK2/cyclin E is 3.5-fold higher than bohemine and is essentially the same as that of roscovitine (Table 1). Compounds 4a, 4e, 4g and 4j inhibit CDK2/cyclin E in the micromolar range (IC 50 = 0.92-3.26 βM). On the other hand, purines 4c and 4i displayed relatively weak inhibitory activity against CDK2/cyclin E in concentrations higher than 10 µM. The results mentioned above show that the presence of polar oxo, or the hydroxy function group, in the substituent at C6 of the purine ring positively influences the inhibitory activity of tested compounds. This point is further supported by the results of the inhibitory activity obtained for purines 4c, 4f and 4h with a non-polar linker between the adamantane scaffold and the benzene ring. These compounds displayed weak or no inhibition activity towards CDK2/cyclin E. In the case of compounds with adamantylated benzylamine in position 6 (4i and 4j), the para substituted derivative, 4j, displayed 4.3-fold higher activity than its meta-analogue, 4i. Of note, two purines bearing the adamantane moiety, linked directly by the N-atom at position 6 (4m and 4n in Figure 1), were previously tested for their CDK inhibition activity with no positive results [18,20]. The presence of a bulky adamantane cage close to the purine ring was assumed to be the reason for the low activity. The CDK inhibition activity was also studied for β-CD·purines 4a-k complexes (1:1 and 10:1, molar ratio). The activity of the equimolar mixtures of β-CD with purines 4b, 4e, 4g and 4k was essentially consistent with their non-complexed forms. In the case of compounds 4a and 4d, the complexation with β-CD (1:1) resulted in an approximately 1.5-2.5 times decrease of their inhibitory activity. Purines 4f and 4h had no activity on CDK2/E, neither as single compounds nor as complexes with β-CD (both 1:1 and 1:10). The preparation of complexes purine·β-CD (1:10) led, with the exception of compounds 4b and 4k, to the dramatic decrease or total loss of inhibition activity. These results can be explained by the competition of β-CD and CDK for the purine ligand.

In Vitro Cytotoxicity
The antiproliferative activity of purines 4a-k was assayed on two types of human tumour cell lines (K-562, chronic myelogenous leukaemia and MCF-7, breast adenocarcinoma). Analogous to that of CDK inhibition activity, the cytotoxicity was tested on single purines 4a-k, as well as their mixtures with β-CD (Table 2). Purine 4d was 17.8and 9.2-fold more potent at inhibiting cell proliferation on K-562 than bohemine and roscovitine, respectively. In the case of MCF-7, the efficacy of 4d was 4.7-and 2.2-fold higher to that of bohemine and roscovitine, respectively. Additionally, compounds 4a, 4e and 4f showed lower growth inhibitory values (GI 50 ) towards leukaemia and/or breast cancer cells in comparison with bohemine and roscovitine. Unfortunately, the study of the antiproliferative activity of single purines 4a-j was compromised by the low solubility of the prepared purines. However, the solubility of compounds 4a-j rapidly increased when they were mixed with β-CD. The cytotoxic effects of the equimolar mixtures of β-CD with adamantane-bearing purines, 4a-j, on leukaemia and breast cancer cells were, in most cases, significantly higher to that of the β-CD·bohemine (4k) complex. Antiproliferative activity of 1:1 complexes of β-CD with purines 4a, 4d, 4e and 4f was 2-3 times lower than their non-complexed form. In the case of purine·β-CD (1:10) complexes, no cytotoxic effects were observed for all of the prepared compounds (except to purine 4k towards MCF-7).

Docking Study
The docking study involved two CDK2 complexes (whose structures are available in the Protein Data Bank), the CDK2 (2A4L) and CDK2/cyclin E complex (5L2W), and ten purine derivatives, 4a-j. For chiral compounds 4b, 4e and 4g, both enantiomers were docked separately. Initially, we validated the reliability of the docking protocol by redocking the optimised dinaciclib structure in the CDK2/cyclin E active site. The superimposed experimental [52] and docked complex is shown in Figure S41 to demonstrate the essentially equal positioning of the two structures inside the binding site.
Considering the lowest energy geometries for each protein-ligand pair, three distinct orientations of the ligand inside the ATP binding pocket were detected in the case of the CDK2/cyclin E complex (5L2W). The most frequent typical orientation (i.e., the adamantane cage deep inside the binding pocket) can be seen in Figure 4b for the ligand 4a. In this orientation, the adamantane cage contributes to complex stabilisation through non-polar interactions with Val 19, Phe 81 and Leu 135. The N9-isopropyl group is located near the portal of the binding site and displays non-polar interactions with Phe 83. The purine core is linked with a number of H-bond interactions to residues Ile 11, Leu 84, His 85, Asp 87 and Lys 90. In the case of three compounds, a flipped orientation of the ligand molecule (i.e., with the adamantane cage near the portal of the binding pocket) displayed minimum energy (for an example, see 4f@CDK2/E in Figure S42). This orientation is stabilised through non-polar interactions with Val 65, Phe 81, Phe 83, Gln 132 and Leu 135, and H-bonds with Gln 132, Leu 84 and Asp 87.
tially equal positioning of the two structures inside the binding site.
Considering the lowest energy geometries for each protein-ligand pair, three distinct orientations of the ligand inside the ATP binding pocket were detected in the case of the CDK2/cyclin E complex (5L2W). The most frequent typical orientation (i.e., the adamantane cage deep inside the binding pocket) can be seen in Figure 4b for the ligand 4a. In this orientation, the adamantane cage contributes to complex stabilisation through nonpolar interactions with Val 19, Phe 81 and Leu 135. The N9-isopropyl group is located near the portal of the binding site and displays non-polar interactions with Phe 83. The purine core is linked with a number of H-bond interactions to residues Ile 11, Leu 84, His 85, Asp 87 and Lys 90. In the case of three compounds, a flipped orientation of the ligand molecule (i.e., with the adamantane cage near the portal of the binding pocket) displayed minimum energy (for an example, see 4f@CDK2/E in Figure S42). This orientation is stabilised through non-polar interactions with Val 65, Phe 81, Phe 83, Gln 132 and Leu 135, and Hbonds with Gln 132, Leu 84 and Asp 87.  In the binding pocket of CDK2 (2A4L), the most frequent arrangement of the ligand is stabilised via non-polar interactions between the adamantane cage and Val 18, Lys 33, Val 64, Phe 80, Leu 134 and Ala 144 (for the orientation of 4d, see Figure 4a). In addition, N9isopropyl has non-polar interactions with Val 164 and Glu 162, and the aminopurine ring interacts through the H-bonds with Gly 13 and Gln 131. Conformation of the C2-sidearm was stabilised via the H-bond between the OH group and Asp 86. The other orientations with minimum energy are shown in Figure S43 for 4a and Figure S44 for 4j.
A complete list of the intermolecular contacts recognised by PLIP software, and energies of the most stable arrangements of the ligand with CDK2 and CDK2/cyclin E (according to PyRx), are given in Tables S1 and S2, respectively.
Our docking study implicates three interesting conclusions. First, the binding pocket of the CDK2 is large enough to accommodate not only the bulky adamantane moiety but all of the molecules of our ligands. Therefore, the appropriate modification of the purine-based ligands (i.e., with a sufficiently long spacer at position C6) with an adamantane scaffold, which is motivated by the improvement of the pharmacologic and/or pharmacokinetic properties, does not disable the ligand binding to the active site. Second, in both examined protein structures, there are two distinct domains in which the adamantane cage can be accommodated. In the case of CDK2, the ligand orientation depends on the length of the spacer between the adamantane cage and purine ring. Ligands with longer spacers (4g-4i) prefer a positioning of the adamantane cage in the Val 164 domain, whereas ligands with shorter spacers (4a-4f) prefer an orientation with the adamantane cage in the Phe 80 domain. In the case of CDK2/cyclin E, the positioning of the adamantane cage in the Phe 81 domain predominates (9:4), with no clear correlation to the ligand geometry. Finally, we docked the previously published 6-[(1-adamantyl)amino]-2-[(3-hydroxypropyl)amino]-9-(propan-2-yl)purine (4m) [18] into the binding site of CDK2. This compound, 4m, displayed no inhibition activity (IC 50 > 0.1 mmol dm −3 ) towards CDK2 and CDK1. We compared the most stable arrangement of the inactive 4m with those of our ligands, 4a-4f, to clarify the role of the adamantane moiety. The superposed image is shown in Figure S45. As can be seen, the positioning of the compound 4m inside the CDK2 binding site is driven by interactions of the adamantane moiety in the Phe 80 domain. Since the adamantane cage occupies essentially the same position in the case of guests 4a-4f and 4m, the purine ring of compound 4m does not interact efficiently via H-bonds with residues within the pocket. This leads to significantly lower binding strength, as demonstrated in Table S1.

General Data
All starting compounds, reagents and solvents were purchased from commercial sources in analytical quality and were used without further purification. Adamantylated aromatic amines were prepared following previously published procedures [41,42,53]. Melting points were measured on a Kofler block and were not corrected. Only peaks of the relative abundance exceeding 5% are listed. The electrospray mass spectra were recorded on an Esquire LC ion-trap mass spectrometer (Bruker Daltonics, Bremen, Germany) equipped with an ESI source. Sample solutions (8.8 µM in methanol/water, 1/1, v/v) were infused into the ion source using a syringe pump at a constant flow rate of 3 µL min −1 . The other instrumental conditions were as follows: a capillary voltage of −3.5 kV, a drying gas temperature of 250 • C, a drying gas flow of 5 dm 3 min −1 and a nebuliser pressure of 96.53 kPa. Nitrogen was used as both nebulising gas and drying gas. The nozzle-skimmer potential and octopole potential were modified and optimised before each experiment. Tandem mass spectra were carried out using CID with He as the collision gas after isolation of the required ions.

Molecular Docking Study
Molecules of the purines were built up in Avogadro v1.0.3 and optimised using a MMFF94s force field [54]. Previously reported crystal structures of CDK2 (2A4L) [55] and CDK2/cyclin E (5L2W) [52] were used as target proteins. Water molecules and ligands from the active site were removed in Avogadro. Molecular docking was performed using AutoDock Vina [56] implemented in PyRx v0.8 software [57]. The intermolecular contacts between the ligand and protein were analysed using an online tool, PLIP v2.2.0 [58].

General Procedure for Preparation of 6-"Amino"-2-chloro-9-isopropyl-9H-purines (3a-k)
Compounds 3a-k were prepared following a slightly modified literature procedure [43]. Purine 2a (100 mg, 0.43 mmol) was dissolved in DMF (2 cm 3 ), and then corresponding amine (1.05 equivalent) and triethylamine (1.1 equivalent for a free base and 2 equivalents for hydrochloride) were added. The reaction mixture was stirred at 90 • C till the TLC indicated the consumption of all starting material (2-53 h). Subsequently, the mixture was diluted with water (a formation of soft precipitation was observed) and extracted with ether (4 × 10 cm 3 ). Combined organic layers were washed twice with brine and dried over sodium sulphate. The solvent was removed by the evaporation in vacuo. Purification of the crude product using column chromatography (silicagel, petroleum ether/ethyl acetate, 1/1, v/v) resulted in the desired product.

Conclusions
Our initial motivation was to investigate if the adamantane scaffold (Ad) can be used to improve purine-based CDK inhibitors. We assumed there would be improved water solubility and related bioavailability of Ad-modified drugs via complexation with suitable macrocyclic carriers. We prepared a series of ten new 2,6,9-trisubstituted purines bearing isopropyl, 3-hydroxypropylamino and various adamantylated anilines and benzylamines in positions 9, 2 and 6, respectively. Although previous studies demonstrated that substituting purine at position 6 with adamantylamine would switch off the inhibition activity towards CDK1 and CDK2, we reached IC 50 (CDK2/E) = 0.21 ± 0.02 µM for adamatylated purine 4b, which is even lower than that of its parent, bohemine 4k. In concert with the supramolecular nature of the interaction between adamantylated purines and CD carries, we showed that IC 50 (CDK2/E) and GI 50 (K-562, MCF-7) increased with a portion of β-CD in tested mixtures. On the other hand, the application of CD carriers allowed us to determine the GI 50 values, even in the cases of sparingly soluble purines. In addition, we performed a docking study to clarify that the inactivity of previously published adamantylated purines is not related to the presence of the bulky adamantane moiety, which was believed to prevent the entrance of the inhibitor to the active site. We demonstrated that entire molecules of purine derivatives with the adamantane cage, linked via appropriately long spacers, can be buried inside the binding pocket of CDK2 to form tight complexes. We believe that our results will encourage scientists to use the adamantane scaffold for improving drug properties.