Novel 2,6,9-Trisubstituted Purines as Potent CDK Inhibitors Alleviating Trastuzumab-Resistance of HER2-Positive Breast Cancers

HER2-positive (HER2+) breast cancer is defined by HER2 oncogene amplification on chromosome 17q12 and accounts for 15–20% population of breast-cancer patients. Therapeutic anti-HER2 antibody such as trastuzumab is used as the first-line therapy for HER2-positive breast cancers. However, more than 50% of the patients respond poorly to trastuzumab, illustrating that novel therapy is warranted to overcome the resistance. We previously reported that in the majority of HER2+ breast-cancer patients, CDK12 is co-amplified on 17q12 and involved in developing tumors and trastuzumab resistance, proposing CDK12 as a potential drug target for HER2+ breast cancers. Here, we designed and synthesized novel 2,6,9-trisubstituted purines as potent CDK12 inhibitors showing strong, equipotent antiproliferative activity against trastuzumab-sensitive HER2+ SK-Br3 cells and trastuzumab-resistant HER2+ HCC1954 cells (GI50 values < 50 nM) both of which express a high level of CDK12. Two potent analogue 30d and 30e at 40, 200 nM greatly downregulated the levels of cyclinK and Pol II p-CTD (Ser2), as well as the expression of CDK12 downstream genes (IRS1 and WNT1) in a dose-dependent manner. We also observed structure-property relationship for a subset of potent analogues, and found that 30e is highly stable in liver microsomes with lack of CYP inhibition. In addition, 30d exhibited a synergy with trastuzumab in the both cells, suggesting that our inhibitors could be applied to alleviate trastuzumab-resistance of HER2+ breast cancers and escalate the efficacy of trastuzumab as well. Our study may provide insight into developing a novel therapy for HER2+ breast cancers.


Introduction
Breast cancers are the most prevalent cancers in the world, and categorized into several subtypes based on genetic background and biomarker profiles [1]. So called HER2-positive (HER2+) breast cancers, in which human epidermal growth factor receptor 2 (HER2) oncogene is amplified in cancer genome, account for 15-20% of breast cancers. The HER2 overexpression is associated with the aggressiveness of breast cancers [2,3]. Accordingly, anti-HER2 monoclonal antibody such as trastuzumab is used as a first-line treatment for metastatic HER2+ breast cancers. Both progression-free survival and overall survival The HER2 overexpression is associated with the aggressiveness of breast cancers [2,3]. Accordingly, anti-HER2 monoclonal antibody such as trastuzumab is used as a first-line treatment for metastatic HER2+ breast cancers. Both progression-free survival and overall survival of HER2+ breast-cancer patients were significantly enhanced when trastuzumab was administered in combination with chemotherapy [4,5]. However, more than a half of HER2+ breast-cancer patients responded poorly to trastuzumab [6]. Studies revealed a number of molecular mechanisms associated with the trastuzumab resistance [7,8], including (i) HER2 mutation defective in binding with trastuzumab (i.e., HER2 truncation), (ii) upregulation of HER2 downstream signalings (i.e., PTEN loss), (iii) activation of alternative signaling pathways (i.e., IGF1R stimulation), and iv) failure to trigger antibodymediated anti-cancer immunity (i.e., FcγRIII F158 polymorphism). Thus, the development of a new drug offsetting the resistance mechanisms is warranted.
The HER2 gene is located on chromosome 17q12. It was reported that multiple other genes at 17q12 are also amplified in HER2+ breast cancers, some of which are crucial for growth and survival of breast cancers [9]. Recently, we discovered that cyclin-dependent kinase 12 (CDK12) as a co-amplified gene on 17q12 is involved in tumorigenesis and trastuzumab resistance in HER2+ breast cancers, proposing CDK12 an attractive therapeutic target to escalate the therapeutic activity of trastuzumab and overcome the trastuzumab resistance ( Figure 1) [10]. CDK12 is a transcription-associated CDK-family kinase, and requires binding with cyclinK for activation. CDK12/cyclinK phosphorylates the C-terminal domain (CTD) of RNA PolII at Ser2, which, in turn, promotes transcriptional elongation, plays roles in RNA splicing, and regulates the expression of the genes involved in the DNA damage response and replication [11]. Several CDK12 inhibitors and their anti-cancer activities have been reported ( Figure  2) [12]. Dinaciclib (SCH 727965) was developed as a potent inhibitor of CDK-family kinases [13], and later it was revealed that dinaciclib shows a potent inhibitory activity against CDK12 as well [14]. In the clinical phase II trial for advanced breast cancers, dinaciclib alone showed only a marginal efficacy [15]. However, preclinical studies suggested that dinaciclib might be able to confer a significant efficacy for selected patient cohorts [10]. Additionally, pan-CDK inhibitors such as dinaciclib, selective CDK12 inhibitors (2, 3) were unveiled from a rational designing approach [16,17]. Gray group reported irreversible CDK12 inhibitors such as THZ531 (4) [18], MFH290 (5) [19], and E9 (6) [20], Several CDK12 inhibitors and their anti-cancer activities have been reported ( Figure 2) [12]. Dinaciclib (SCH 727965) was developed as a potent inhibitor of CDK-family kinases [13], and later it was revealed that dinaciclib shows a potent inhibitory activity against CDK12 as well [14]. In the clinical phase II trial for advanced breast cancers, dinaciclib alone showed only a marginal efficacy [15]. However, preclinical studies suggested that dinaciclib might be able to confer a significant efficacy for selected patient cohorts [10]. Additionally, pan-CDK inhibitors such as dinaciclib, selective CDK12 inhibitors (2, 3) were unveiled from a rational designing approach [16,17]. Gray group reported irreversible CDK12 inhibitors such as THZ531 (4) [18], MFH290 (5) [19], and E9 (6) [20], whose electrophilic acrylamide moiety formed a covalent bond with the unique cysteine located in C-terminal region of CDK12/13 (Cys1039 for CDK12, Cys1017 for CDK13). These irreversible inhibitors exhibited prominent anti-cancer phenotypes in cancer cells owing to their highly selective, strong, persistent suppression of CDK12/13, but their in vivo activity could not be evaluated due to their poor stability in vivo. Recently two different CDK12 inhibitor-based PROTACs (PP-C8, BSJ-4-116) were reported, which were derived from 2 and THZ531 (4), respectively, [21,22]. Interestingly, unlike 2 and 4 showing a similar potency against CDK12 and CDK13, the both CDK12 degraders discriminated CDK12 over CDK13 for degradation in cells, providing an approach to achieve CDK12 selectivity. whose electrophilic acrylamide moiety formed a covalent bond with the unique cysteine located in C-terminal region of CDK12/13 (Cys1039 for CDK12, Cys1017 for CDK13). These irreversible inhibitors exhibited prominent anti-cancer phenotypes in cancer cells owing to their highly selective, strong, persistent suppression of CDK12/13, but their in vivo activity could not be evaluated due to their poor stability in vivo. Recently two different CDK12 inhibitor-based PROTACs (PP-C8, BSJ-4-116) were reported, which were derived from 2 and THZ531 (4), respectively, [21,22]. Interestingly, unlike 2 and 4 showing a similar potency against CDK12 and CDK13, the both CDK12 degraders discriminated CDK12 over CDK13 for degradation in cells, providing an approach to achieve CDK12 selectivity. Additionally, a new type of CDK12 inhibitors were reported, including R-CR8 (9) [23], HQ461 (10) [24], and dCeMM2/3/4 (11a/b/c) [25] which act as molecular glues promoting CDK12-DDB1 interaction. The inhibitor-bound CDK12 recruits DDB1, an adaptor protein of Cul4-uibiqutin ligase, through the terminal moiety of the inhibitors (i.e., the pyridyl group in R-CR8, 5-methylthiazol-2-amine group in HQ461). In the drug-induced complex, CDK12 behaves as a substrate binding protein of DDB1-Cul4 E3 ligase, provoking ubiquitination and the subsequent proteasomal degradation of cyclinK. CyclinK degradation significantly enhanced inhibitors' activity to suppress the function of CDK12 in Additionally, a new type of CDK12 inhibitors were reported, including R-CR8 (9) [23], HQ461 (10) [24], and dCeMM2/3/4 (11a/b/c) [25] which act as molecular glues promoting CDK12-DDB1 interaction. The inhibitor-bound CDK12 recruits DDB1, an adaptor protein of Cul4-uibiqutin ligase, through the terminal moiety of the inhibitors (i.e., the pyridyl group in R-CR8, 5-methylthiazol-2-amine group in HQ461). In the drug-induced complex, CDK12 behaves as a substrate binding protein of DDB1-Cul4 E3 ligase, provoking ubiquitination and the subsequent proteasomal degradation of cyclinK. CyclinK degradation significantly enhanced inhibitors' activity to suppress the function of CDK12 in cancer cells. Herein, we wish to report novel potent purine scaffold CDK12 inhibitors acting as a cyclinK degrader that potently suppressed a growth of HER2+ breast cancers. Our inhibitors also deteriorated the growth of traszutumab-resistant HER2+ breast cancers with a similar potency. This study may provide an insight into designing potent CDK12 inhibitors for HER2+ breast cancers.

Inhibitor Design
We designed novel 2,6,9-trisubstituted purine scaffold CDK12 inhibitors by hybridizing the reported purine-based CDK12 inhibitor (2) [17] and the cyclinK degrader R-CR8 (9) [23] (Figure 3). The X-ray co-crystal structures displayed two unique hydrogen bond interactions between the imidazole part of 2 and the side chains of Tyr815 and Asp819, which are positioned near the phenyl group of R-CR8. Thus, we replaced the phenyl group of R-CR8 with a pyridyl group, on which a nitrogen atom could be engaged with a hydrogen bond with Tyr815 or Asp819. We also introduced various heteroaryl or aryl moieties to 2 position, to which hydroxyalkyl moieties were attached to emulate the hydroxyalkyl group of CR-8. We expected that any heteroatom at 2 position might be involved in the interaction with nearby residues such as Asp819.
cancer cells. Herein, we wish to report novel potent purine scaffold CDK12 inhibitors acting as a cyclinK degrader that potently suppressed a growth of HER2+ breast cancers. Our inhibitors also deteriorated the growth of traszutumab-resistant HER2+ breast cancers with a similar potency. This study may provide an insight into designing potent CDK12 inhibitors for HER2+ breast cancers.

Inhibitor Design
We designed novel 2,6,9-trisubstituted purine scaffold CDK12 inhibitors by hybridizing the reported purine-based CDK12 inhibitor (2) [17] and the cyclinK degrader R-CR8 (9) [23] (Figure 3). The X-ray co-crystal structures displayed two unique hydrogen bond interactions between the imidazole part of 2 and the side chains of Tyr815 and Asp819, which are positioned near the phenyl group of R-CR8. Thus, we replaced the phenyl group of R-CR8 with a pyridyl group, on which a nitrogen atom could be engaged with a hydrogen bond with Tyr815 or Asp819. We also introduced various heteroaryl or aryl moieties to 2 position, to which hydroxyalkyl moieties were attached to emulate the hydroxyalkyl group of CR-8. We expected that any heteroatom at 2 position might be involved in the interaction with nearby residues such as Asp819.
At first, we investigated the structure-activity relationship (SAR) for 6 position. A series of bipyridyl methaneamines were incorporated to 6 position, while 2 and 9 positions were fixed with 3-pyridyl and isopropyl group, respectively, (Scheme 1). A set of biaryl carbonitriles (12a-g) were prepared using Suzuki coupling reactions between respective pyridyl boronic acids and bromoaryl nitriles. The cyano group was then converted to aminomethylene group (14a-g) using NiCl2/Boc-mediated reduction and the subsequent Boc deprotection [26]. To synthesize the desired products 17a-g, commercially available 2,6dichloropurine was alkylated at 9 position with isopropyl group, and aminated with various bipyridyl methanamines (14a-g) at 6 position. The resulting intermediates (16a-g) were conjugated with 3-pyridyl group at 2 position through Suzuki coupling reactions.
At first, we investigated the structure-activity relationship (SAR) for 6 position. A series of bipyridyl methaneamines were incorporated to 6 position, while 2 and 9 positions were fixed with 3-pyridyl and isopropyl group, respectively, (Scheme 1). A set of biaryl carbonitriles (12a-g) were prepared using Suzuki coupling reactions between respective pyridyl boronic acids and bromoaryl nitriles. The cyano group was then converted to aminomethylene group (14a-g) using NiCl 2 /Boc-mediated reduction and the subsequent Boc deprotection [26]. To synthesize the desired products 17a-g, commercially available 2,6-dichloropurine was alkylated at 9 position with isopropyl group, and aminated with various bipyridyl methanamines (14a-g) at 6 position. The resulting intermediates (16a-g) were conjugated with 3-pyridyl group at 2 position through Suzuki coupling reactions. Compounds 17a-e showed similar IC50 values against CDK12/cyclinK, indicating that the terminal pyridyl group at 6 position is not involved in binding with CDK12 as in the case of CR-8 (Table 1). Interestingly, the N2 on the inner pyridine ring at 6 position increased CDK12 inhibitory activity 2-3 times (17f vs. 17b, 17g vs. 17d), suggesting a potential hydrogen bond between the N2 and a nearby residue within CDK12 as we anticipated. In addition, although a terminal pyridyl group at 6 position made almost no contribution to the enzymatic activity, particular terminal groups present in 17a and 17e substantially augmented cell growth inhibitory activities against the both cells. Compounds 17a-e showed similar IC 50 values against CDK12/cyclinK, indicating that the terminal pyridyl group at 6 position is not involved in binding with CDK12 as in the case of CR-8 (Table 1). Interestingly, the N2 on the inner pyridine ring at 6 position increased CDK12 inhibitory activity 2-3 times (17f vs. 17b, 17g vs. 17d), suggesting a potential hydrogen bond between the N2 and a nearby residue within CDK12 as we anticipated. In addition, although a terminal pyridyl group at 6 position made almost no contribution to the enzymatic activity, particular terminal groups present in 17a and 17e substantially augmented cell growth inhibitory activities against the both cells. Compounds 17a-e showed similar IC50 values against CDK12/cyclinK, indicating that the terminal pyridyl group at 6 position is not involved in binding with CDK12 as in the case of CR-8 (Table 1). Interestingly, the N2 on the inner pyridine ring at 6 position increased CDK12 inhibitory activity 2-3 times (17f vs. 17b, 17g vs. 17d), suggesting a potential hydrogen bond between the N2 and a nearby residue within CDK12 as we anticipated. In addition, although a terminal pyridyl group at 6 position made almost no contribution to the enzymatic activity, particular terminal groups present in 17a and 17e substantially augmented cell growth inhibitory activities against the both cells. Compounds 17a-e showed similar IC50 values against CDK12/cyclinK, indicating that the terminal pyridyl group at 6 position is not involved in binding with CDK12 as in the case of CR-8 (Table 1). Interestingly, the N2 on the inner pyridine ring at 6 position increased CDK12 inhibitory activity 2-3 times (17f vs. 17b, 17g vs. 17d), suggesting a potential hydrogen bond between the N2 and a nearby residue within CDK12 as we anticipated. In addition, although a terminal pyridyl group at 6 position made almost no contribution to the enzymatic activity, particular terminal groups present in 17a and 17e substantially augmented cell growth inhibitory activities against the both cells. Compounds 17a-e showed similar IC50 values against CDK12/cyclinK, indicating that the terminal pyridyl group at 6 position is not involved in binding with CDK12 as in the case of CR-8 (Table 1). Interestingly, the N2 on the inner pyridine ring at 6 position increased CDK12 inhibitory activity 2-3 times (17f vs. 17b, 17g vs. 17d), suggesting a potential hydrogen bond between the N2 and a nearby residue within CDK12 as we anticipated. In addition, although a terminal pyridyl group at 6 position made almost no contribution to the enzymatic activity, particular terminal groups present in 17a and 17e substantially augmented cell growth inhibitory activities against the both cells. Compounds 17a-e showed similar IC50 values against CDK12/cyclinK, indicating that the terminal pyridyl group at 6 position is not involved in binding with CDK12 as in the case of CR-8 (Table 1). Interestingly, the N2 on the inner pyridine ring at 6 position increased CDK12 inhibitory activity 2-3 times (17f vs. 17b, 17g vs. 17d), suggesting a potential hydrogen bond between the N2 and a nearby residue within CDK12 as we anticipated. In addition, although a terminal pyridyl group at 6 position made almost no contribution to the enzymatic activity, particular terminal groups present in 17a and 17e substantially augmented cell growth inhibitory activities against the both cells. Compounds 17a-e showed similar IC50 values against CDK12/cyclinK, indicating that the terminal pyridyl group at 6 position is not involved in binding with CDK12 as in the case of CR-8 (Table 1). Interestingly, the N2 on the inner pyridine ring at 6 position increased CDK12 inhibitory activity 2-3 times (17f vs. 17b, 17g vs. 17d), suggesting a potential hydrogen bond between the N2 and a nearby residue within CDK12 as we anticipated. In addition, although a terminal pyridyl group at 6 position made almost no contribution to the enzymatic activity, particular terminal groups present in 17a and 17e substantially augmented cell growth inhibitory activities against the both cells. Compounds 17a-e showed similar IC50 values against CDK12/cyclinK, indicating that the terminal pyridyl group at 6 position is not involved in binding with CDK12 as in the case of CR-8 (Table 1). Interestingly, the N2 on the inner pyridine ring at 6 position increased CDK12 inhibitory activity 2-3 times (17f vs. 17b, 17g vs. 17d), suggesting a potential hydrogen bond between the N2 and a nearby residue within CDK12 as we anticipated. In addition, although a terminal pyridyl group at 6 position made almost no contribution to the enzymatic activity, particular terminal groups present in 17a and 17e substantially augmented cell growth inhibitory activities against the both cells.  To test whether the significant improvement of cellular activity from 17a and 17e originated from cyclinK degradation, we selected four analogues (17a/b/d/e) showing similar in vitro CDK12 potency, and compared their ability in downregulating the levels of cyclinK and PolII p-CTD (Ser2) in SK-Br3 cells ( Figure 4). All of the four compounds at 0.2, 1 μM downregulated cyclinK in a dose-dependent manner, and among them, 17a induced the most prominent cyclinK degradation. However, all four compounds exhibited a similar level of PolII p-CTD (Ser2) suppression, suggesting that the higher growth inhibitory activity of 17a and 17e might originate from off-target effects irrespective of CDK12 or cyclinK. Conversely, dinaciclib at 0.2, 1 μM strongly suppressed PolII p-CTD, but showed a marginal activity in downrgulating cyclinK in cells. To test whether the significant improvement of cellular activity from 17a and 17e originated from cyclinK degradation, we selected four analogues (17a/b/d/e) showing similar in vitro CDK12 potency, and compared their ability in downregulating the levels of cyclinK and PolII p-CTD (Ser2) in SK-Br3 cells ( Figure 4). All of the four compounds at 0.2, 1 µM downregulated cyclinK in a dose-dependent manner, and among them, 17a induced the most prominent cyclinK degradation. However, all four compounds exhibited a similar level of PolII p-CTD (Ser2) suppression, suggesting that the higher growth inhibitory activity of 17a and 17e might originate from off-target effects irrespective of CDK12 or cyclinK. Conversely, dinaciclib at 0.2, 1 µM strongly suppressed PolII p-CTD, but showed a marginal activity in downrgulating cyclinK in cells. To test whether the significant improvement of cellular activity from 17a and 17e originated from cyclinK degradation, we selected four analogues (17a/b/d/e) showing similar in vitro CDK12 potency, and compared their ability in downregulating the levels of cyclinK and PolII p-CTD (Ser2) in SK-Br3 cells ( Figure 4). All of the four compounds at 0.2, 1 μM downregulated cyclinK in a dose-dependent manner, and among them, 17a induced the most prominent cyclinK degradation. However, all four compounds exhibited a similar level of PolII p-CTD (Ser2) suppression, suggesting that the higher growth inhibitory activity of 17a and 17e might originate from off-target effects irrespective of CDK12 or cyclinK. Conversely, dinaciclib at 0.2, 1 μM strongly suppressed PolII p-CTD, but showed a marginal activity in downrgulating cyclinK in cells.  Next, we investigated SAR for 2 position. A variety of aryl or heteroaryl groups were attached to 2 position of the intermediate 16f to afford 18a-n, 19a-c, and 21a-b (Scheme 2). The aniline NH 2 group of 19a-c was subjected to alkylation to produce N-hydroxyalkyl analogues 20a-d. Additionally, α-fluoro pyridyl compounds 21a-b reacted with hydroxyalkyl amines to generate 22a-d. The chloro group of 16f was also replaced with hydrazine group (23), which then reacted with 3-oxobutanenitrile to form an aminopyrazol-containing analogue 24. Additionally, 23 was transformed to an azido intermediate (25) through sequential reactions using NaNO 2 /HCl and NaN 3 , which then underwent click chemistry to generate 26a-b containing hydroxyalkyl triazoles at 2 position. analogues 20a-d. Additionally, α-fluoro pyridyl compounds 21a-b reacted with hydroxyalkyl amines to generate 22a-d. The chloro group of 16f was also replaced with hydrazine group (23), which then reacted with 3-oxobutanenitrile to form an aminopyrazolcontaining analogue 24. Additionally, 23 was transformed to an azido intermediate (25) through sequential reactions using NaNO2/HCl and NaN3, which then underwent click chemistry to generate 26a-b containing hydroxyalkyl triazoles at 2 position. The SAR result in Table 2 showed that m-amino and p-amino group of the phenyl at 2 position augmented CDK12 inhibitory activity 3-4 times (19b vs. 18a, 19c vs. 18a). A similar level of improvement was observed from their isosteres such as indole (18d) and indazole (18e, 18f). N-hydroxypropyl modification of the aniline groups (20b, 20c, 20d) diminished the inhibitory activity, indicating that this variation might be sterically unfavorable for binding with CDK12. The SAR result in Table 2 showed that m-amino and p-amino group of the phenyl at 2 position augmented CDK12 inhibitory activity 3-4 times (19b vs. 18a, 19c vs. 18a). A similar level of improvement was observed from their isosteres such as indole (18d) and indazole (18e, 18f). N-hydroxypropyl modification of the aniline groups (20b, 20c, 20d) diminished the inhibitory activity, indicating that this variation might be sterically unfavorable for binding with CDK12. analogues 20a-d. Additionally, α-fluoro pyridyl compounds 21a-b reacted with hydroxyalkyl amines to generate 22a-d. The chloro group of 16f was also replaced with hydrazine group (23), which then reacted with 3-oxobutanenitrile to form an aminopyrazolcontaining analogue 24. Additionally, 23 was transformed to an azido intermediate (25) through sequential reactions using NaNO2/HCl and NaN3, which then underwent click chemistry to generate 26a-b containing hydroxyalkyl triazoles at 2 position. The SAR result in Table 2 showed that m-amino and p-amino group of the phenyl at 2 position augmented CDK12 inhibitory activity 3-4 times (19b vs. 18a, 19c vs. 18a). A similar level of improvement was observed from their isosteres such as indole (18d) and indazole (18e, 18f). N-hydroxypropyl modification of the aniline groups (20b, 20c, 20d) diminished the inhibitory activity, indicating that this variation might be sterically unfavorable for binding with CDK12. analogues 20a-d. Additionally, α-fluoro pyridyl compounds 21a-b reacted with hydroxyalkyl amines to generate 22a-d. The chloro group of 16f was also replaced with hydrazine group (23), which then reacted with 3-oxobutanenitrile to form an aminopyrazolcontaining analogue 24. Additionally, 23 was transformed to an azido intermediate (25) through sequential reactions using NaNO2/HCl and NaN3, which then underwent click chemistry to generate 26a-b containing hydroxyalkyl triazoles at 2 position. The SAR result in Table 2 showed that m-amino and p-amino group of the phenyl at 2 position augmented CDK12 inhibitory activity 3-4 times (19b vs. 18a, 19c vs. 18a). A similar level of improvement was observed from their isosteres such as indole (18d) and indazole (18e, 18f). N-hydroxypropyl modification of the aniline groups (20b, 20c, 20d) diminished the inhibitory activity, indicating that this variation might be sterically unfavorable for binding with CDK12. analogues 20a-d. Additionally, α-fluoro pyridyl compounds 21a-b reacted with hydroxyalkyl amines to generate 22a-d. The chloro group of 16f was also replaced with hydrazine group (23), which then reacted with 3-oxobutanenitrile to form an aminopyrazolcontaining analogue 24. Additionally, 23 was transformed to an azido intermediate (25) through sequential reactions using NaNO2/HCl and NaN3, which then underwent click chemistry to generate 26a-b containing hydroxyalkyl triazoles at 2 position. The SAR result in Table 2 showed that m-amino and p-amino group of the phenyl at 2 position augmented CDK12 inhibitory activity 3-4 times (19b vs. 18a, 19c vs. 18a). A similar level of improvement was observed from their isosteres such as indole (18d) and indazole (18e, 18f). N-hydroxypropyl modification of the aniline groups (20b, 20c, 20d) diminished the inhibitory activity, indicating that this variation might be sterically unfavorable for binding with CDK12.  Noticeably, the SAR result in Table 3 indicated that 6-membered heteroaryl moieties at 2 position significantly improved inhibitory activity. Compounds containing aminopyridyl (18i, 18j), pyridyl (18k), or aminopyrimidyl (18l) group at 2 position showed a profound inhibition of CDK12/cyclinK with IC50 values < 100 nM. Their inhibitory activity was tolerated by N-hydroxyethyl group (22a, 22c), but was compromised by N-hydroxypropyl modification (22b, 22d). Unlike the 6-membered heteroaryl groups, 5-membered heteroaryl groups based on pyrazole (18m, 18n, 24) or triazole (26a, 26b) did not improve CDK12 inhibition. Noticeably, the SAR result in Table 3 indicated that 6-membered heteroaryl moieties at 2 position significantly improved inhibitory activity. Compounds containing aminopyridyl (18i, 18j), pyridyl (18k), or aminopyrimidyl (18l) group at 2 position showed a profound inhibition of CDK12/cyclinK with IC50 values < 100 nM. Their inhibitory activity was tolerated by N-hydroxyethyl group (22a, 22c), but was compromised by N-hydroxypropyl modification (22b, 22d). Unlike the 6-membered heteroaryl groups, 5-membered heteroaryl groups based on pyrazole (18m, 18n, 24) or triazole (26a, 26b) did not improve CDK12 inhibition.  Noticeably, the SAR result in Table 3 indicated that 6-membered heteroaryl moieties at 2 position significantly improved inhibitory activity. Compounds containing aminopyridyl (18i, 18j), pyridyl (18k), or aminopyrimidyl (18l) group at 2 position showed a profound inhibition of CDK12/cyclinK with IC50 values < 100 nM. Their inhibitory activity was tolerated by N-hydroxyethyl group (22a, 22c), but was compromised by N-hydroxypropyl modification (22b, 22d). Unlike the 6-membered heteroaryl groups, 5-membered heteroaryl groups based on pyrazole (18m, 18n, 24) or triazole (26a, 26b) did not improve CDK12 inhibition. Noticeably, the SAR result in Table 3 indicated that 6-membered heteroaryl moieties at 2 position significantly improved inhibitory activity. Compounds containing aminopyridyl (18i, 18j), pyridyl (18k), or aminopyrimidyl (18l) group at 2 position showed a profound inhibition of CDK12/cyclinK with IC50 values < 100 nM. Their inhibitory activity was tolerated by N-hydroxyethyl group (22a, 22c), but was compromised by N-hydroxypropyl modification (22b, 22d). Unlike the 6-membered heteroaryl groups, 5-membered heteroaryl groups based on pyrazole (18m, 18n, 24 Noticeably, the SAR result in Table 3 indicated that 6-membered heteroaryl moieties at 2 position significantly improved inhibitory activity. Compounds containing aminopyridyl (18i, 18j), pyridyl (18k), or aminopyrimidyl (18l) group at 2 position showed a profound inhibition of CDK12/cyclinK with IC50 values < 100 nM. Their inhibitory activity was tolerated by N-hydroxyethyl group (22a, 22c), but was compromised by N-hydroxypropyl modification (22b, 22d). Unlike the 6-membered heteroaryl groups, 5-membered heteroaryl groups based on pyrazole (18m, 18n, 24 Noticeably, the SAR result in Table 3 indicated that 6-membered heteroaryl moieties at 2 position significantly improved inhibitory activity. Compounds containing aminopyridyl (18i, 18j), pyridyl (18k), or aminopyrimidyl (18l) group at 2 position showed a profound inhibition of CDK12/cyclinK with IC50 values < 100 nM. Their inhibitory activity was tolerated by N-hydroxyethyl group (22a, 22c), but was compromised by N-hydroxypropyl modification (22b, 22d). Unlike the 6-membered heteroaryl groups, 5-membered Noticeably, the SAR result in Table 3 indicated that 6-membered heteroaryl moieties at 2 position significantly improved inhibitory activity. Compounds containing aminopyridyl (18i, 18j), pyridyl (18k), or aminopyrimidyl (18l) group at 2 position showed a profound inhibition of CDK12/cyclinK with IC50 values < 100 nM. Their inhibitory activity was tolerated by N-hydroxyethyl group (22a, 22c), but was compromised by N-hydroxypropyl modification (22b, 22d). Unlike the 6-membered heteroaryl groups, 5-membered Noticeably, the SAR result in Table 3 indicated that 6-membered heteroaryl moieties at 2 position significantly improved inhibitory activity. Compounds containing aminopyridyl (18i, 18j), pyridyl (18k), or aminopyrimidyl (18l) group at 2 position showed a profound inhibition of CDK12/cyclinK with IC50 values < 100 nM. Their inhibitory activity was tolerated by N-hydroxyethyl group (22a, 22c) Noticeably, the SAR result in Table 3 indicated that 6-membered heteroaryl moieties at 2 position significantly improved inhibitory activity. Compounds containing aminopyridyl (18i, 18j), pyridyl (18k), or aminopyrimidyl (18l) group at 2 position showed a profound inhibition of CDK12/cyclinK with IC50 values < 100 nM. Their inhibitory activity was tolerated by N-hydroxyethyl group (22a, 22c), but was compromised by N-hydroxy- Noticeably, the SAR result in Table 3 indicated that 6-membered heteroaryl moieties at 2 position significantly improved inhibitory activity. Compounds containing aminopyridyl (18i, 18j), pyridyl (18k), or aminopyrimidyl (18l) group at 2 position showed a profound inhibition of CDK12/cyclinK with IC50 values < 100 nM. Their inhibitory activity  Noticeably, the SAR result in Table 3 indicated that 6-membered heteroaryl moieties at 2 position significantly improved inhibitory activity. Compounds containing aminopyridyl (18i, 18j), pyridyl (18k), or aminopyrimidyl (18l) group at 2 position showed a profound inhibition of CDK12/cyclinK with IC50 values < 100 nM. Their inhibitory activity Noticeably, the SAR result in Table 3 indicated that 6-membered heteroaryl moieties at 2 position significantly improved inhibitory activity. Compounds containing aminopyridyl (18i, 18j), pyridyl (18k), or aminopyrimidyl (18l) group at 2 position showed a profound inhibition of CDK12/cyclinK with IC 50 values < 100 nM. Their inhibitory activity was tolerated by Nhydroxyethyl group (22a, 22c), but was compromised by N-hydroxypropyl modification (22b, 22d). Unlike the 6-membered heteroaryl groups, 5-membered heteroaryl groups based on pyrazole (18m, 18n, 24) or triazole (26a, 26b) did not improve CDK12 inhibition.            In order to study SAR for 9 position, we prepared the analogues, which have 3pyridyl at 2 position and 14g at 6 position in common, but different alkyl groups at 9 position (Scheme 3). The SAR result in Table 4 showed that the ethyl group (28a) conferred a significantly better CDK12 inhibitory activity (CDK12/cyclinK IC50 = 16 nM) and cell growth inhibitory activity compared with isopropyl group (17f) (CDK12/cyclinK IC50 = 221 nM). However, replacement with a cyclopentyl group (28b) slightly decreased the activity and that with mopholinyl group (28c) eliminated inhibitory activity, suggesting that the pocket around the 9 position prefers a small-size moiety such as an ethyl group and does not accommodate the bulky groups.  In order to study SAR for 9 position, we prepared the analogues, which have 3pyridyl at 2 position and 14g at 6 position in common, but different alkyl groups at 9 position (Scheme 3). The SAR result in Table 4 showed that the ethyl group (28a) conferred a significantly better CDK12 inhibitory activity (CDK12/cyclinK IC50 = 16 nM) and cell growth inhibitory activity compared with isopropyl group (17f) (CDK12/cyclinK IC50 = 221 nM). However, replacement with a cyclopentyl group (28b) slightly decreased the activity and that with mopholinyl group (28c) eliminated inhibitory activity, suggesting that the pocket around the 9 position prefers a small-size moiety such as an ethyl group and does not accommodate the bulky groups.  In order to study SAR for 9 position, we prepared the analogues, which have 3pyridyl at 2 position and 14g at 6 position in common, but different alkyl groups at 9 position (Scheme 3). The SAR result in Table 4 showed that the ethyl group (28a) conferred a significantly better CDK12 inhibitory activity (CDK12/cyclinK IC50 = 16 nM) and cell growth inhibitory activity compared with isopropyl group (17f) (CDK12/cyclinK IC50 = 221 nM). However, replacement with a cyclopentyl group (28b) slightly decreased the activity and that with mopholinyl group (28c) eliminated inhibitory activity, suggesting that the pocket around the 9 position prefers a small-size moiety such as an ethyl group and does not accommodate the bulky groups.  In order to study SAR for 9 position, we prepared the analogues, which have 3pyridyl at 2 position and 14g at 6 position in common, but different alkyl groups at 9 position (Scheme 3). The SAR result in Table 4 showed that the ethyl group (28a) conferred a significantly better CDK12 inhibitory activity (CDK12/cyclinK IC50 = 16 nM) and cell growth inhibitory activity compared with isopropyl group (17f) (CDK12/cyclinK IC50 = 221 nM). However, replacement with a cyclopentyl group (28b) slightly decreased the activity and that with mopholinyl group (28c) eliminated inhibitory activity, suggesting that the pocket around the 9 position prefers a small-size moiety such as an ethyl group and does not accommodate the bulky groups.  In order to study SAR for 9 position, we prepared the analogues, which have 3pyridyl at 2 position and 14g at 6 position in common, but different alkyl groups at 9 position (Scheme 3). The SAR result in Table 4 showed that the ethyl group (28a) conferred a significantly better CDK12 inhibitory activity (CDK12/cyclinK IC50 = 16 nM) and cell growth inhibitory activity compared with isopropyl group (17f) (CDK12/cyclinK IC50 = 221 nM). However, replacement with a cyclopentyl group (28b) slightly decreased the activity and that with mopholinyl group (28c) eliminated inhibitory activity, suggesting that the pocket around the 9 position prefers a small-size moiety such as an ethyl group and does not accommodate the bulky groups. In order to study SAR for 9 position, we prepared the analogues, which have 3pyridyl at 2 position and 14g at 6 position in common, but different alkyl groups at 9 position (Scheme 3). The SAR result in Table 4 showed that the ethyl group (28a) conferred a significantly better CDK12 inhibitory activity (CDK12/cyclinK IC50 = 16 nM) and cell growth inhibitory activity compared with isopropyl group (17f) (CDK12/cyclinK IC50 = 221 nM). However, replacement with a cyclopentyl group (28b) slightly decreased the activity and that with mopholinyl group (28c) eliminated inhibitory activity, suggesting that the pocket around the 9 position prefers a small-size moiety such as an ethyl group and does not accommodate the bulky groups. In order to study SAR for 9 position, we prepared the analogues, which have 3-pyridyl at 2 position and 14g at 6 position in common, but different alkyl groups at 9 position (Scheme 3). The SAR result in Table 4 showed that the ethyl group (28a) conferred a significantly better CDK12 inhibitory activity (CDK12/cyclinK IC 50 = 16 nM) and cell growth inhibitory activity compared with isopropyl group (17f) (CDK12/cyclinK IC 50 = 221 nM). However, replacement with a cyclopentyl group (28b) slightly decreased the activity and that with mopholinyl group (28c) eliminated inhibitory activity, suggesting that the pocket around the 9 position prefers a small-size moiety such as an ethyl group and does not accommodate the bulky groups. Table 4 showed that the ethyl group (28a) conferred a significantly better CDK12 inhibitory activity (CDK12/cyclinK IC50 = 16 nM) and cell growth inhibitory activity compared with isopropyl group (17f) (CDK12/cyclinK IC50 = 221 nM). However, replacement with a cyclopentyl group (28b) slightly decreased the activity and that with mopholinyl group (28c) eliminated inhibitory activity, suggesting that the pocket around the 9 position prefers a small-size moiety such as an ethyl group and does not accommodate the bulky groups.    * These values are means and standard deviations from three independent assays.

position (Scheme 3). The SAR result in
Having the comprehensive SAR results in hands, we finally synthesized several analogues that were anticipated to have potent activities against both CDK12/cyclinK enzyme and cell growth inhibition. We fixed 9 position with the ethyl group and introduced the selected heteroaryl groups at 2 position, conferring a potent CDK12/cyclinK inhibition (Tables 3 and 4). Additionally, two different bipyridyl methaneamines (14h, 14i) were incorporated at 6 position that were expected to confer relatively more effective in CDK12 inhibition and cell-growth inhibition based on the results in Table 1. Synthesis was carried out using the same synthetic routes as before, but for 2,2′-bipyridyl carbonitrile (12i) that was prepared using a modified Negishi coupling reaction [27,28] (Scheme 4). Compounds 30d (CDK12/cyclinK IC50 = 21 nM) and 30e (CDK12/cyclinK IC50 = 85 nM) containing the 2′-pyridyl terminal moiety at 6 position showed less potent enzymatic activity compared with 28a, 30a, 30b (IC50 values < 30 nM), but showed 2-3 fold higher activity in cancer cellgrowth inhibition (GI50 values = 34-52 nM) (Table 5), which is consistent with the SAR trend shown in Table 1.  Having the comprehensive SAR results in hands, we finally synthesized several analogues that were anticipated to have potent activities against both CDK12/cyclinK enzyme and cell growth inhibition. We fixed 9 position with the ethyl group and introduced the selected heteroaryl groups at 2 position, conferring a potent CDK12/cyclinK inhibition (Tables 3 and 4). Additionally, two different bipyridyl methaneamines (14h, 14i) were incorporated at 6 position that were expected to confer relatively more effective in CDK12 inhibition and cell-growth inhibition based on the results in Table 1. Synthesis was carried out using the same synthetic routes as before, but for 2,2′-bipyridyl carbonitrile (12i) that was prepared using a modified Negishi coupling reaction [27,28] (Scheme 4). Compounds 30d (CDK12/cyclinK IC50 = 21 nM) and 30e (CDK12/cyclinK IC50 = 85 nM) containing the 2′-pyridyl terminal moiety at 6 position showed less potent enzymatic activity compared with 28a, 30a, 30b (IC50 values < 30 nM), but showed 2-3 fold higher activity in cancer cellgrowth inhibition (GI50 values = 34-52 nM) (  * These values are means and standard deviations from three independent assays. Having the comprehensive SAR results in hands, we finally synthesized several analogues that were anticipated to have potent activities against both CDK12/cyclinK enzyme and cell growth inhibition. We fixed 9 position with the ethyl group and introduced the selected heteroaryl groups at 2 position, conferring a potent CDK12/cyclinK inhibition (Tables 3 and 4). Additionally, two different bipyridyl methaneamines (14h, 14i) were incorporated at 6 position that were expected to confer relatively more effective in CDK12 inhibition and cell-growth inhibition based on the results in Table 1. Synthesis was carried out using the same synthetic routes as before, but for 2,2′-bipyridyl carbonitrile (12i) that was prepared using a modified Negishi coupling reaction [27,28] (Scheme 4). Compounds 30d (CDK12/cyclinK IC50 = 21 nM) and 30e (CDK12/cyclinK IC50 = 85 nM) containing the 2′-pyridyl terminal moiety at 6 position showed less potent enzymatic activity compared with 28a, 30a, 30b (IC50 values < 30 nM), but showed 2-3 fold higher activity in cancer cellgrowth inhibition (GI50 values = 34-52 nM) (Table 5), which is consistent with the SAR trend shown in Table 1.  * These values are means and standard deviations from three independent assays.
Having the comprehensive SAR results in hands, we finally synthesized several analogues that were anticipated to have potent activities against both CDK12/cyclinK enzyme and cell growth inhibition. We fixed 9 position with the ethyl group and introduced the selected heteroaryl groups at 2 position, conferring a potent CDK12/cyclinK inhibition (Tables 3 and 4). Additionally, two different bipyridyl methaneamines (14h, 14i) were incorporated at 6 position that were expected to confer relatively more effective in CDK12 inhibition and cell-growth inhibition based on the results in Table 1. Synthesis was carried out using the same synthetic routes as before, but for 2,2′-bipyridyl carbonitrile (12i) that was prepared using a modified Negishi coupling reaction [27,28] (Scheme 4). Compounds 30d (CDK12/cyclinK IC50 = 21 nM) and 30e (CDK12/cyclinK IC50 = 85 nM) containing the 2′-pyridyl terminal moiety at 6 position showed less potent enzymatic activity compared with 28a, 30a, 30b (IC50 values < 30 nM), but showed 2-3 fold higher activity in cancer cellgrowth inhibition (GI50 values = 34-52 nM) (  * These values are means and standard deviations from three independent assays. Having the comprehensive SAR results in hands, we finally synthesized several analogues that were anticipated to have potent activities against both CDK12/cyclinK enzyme and cell growth inhibition. We fixed 9 position with the ethyl group and introduced the selected heteroaryl groups at 2 position, conferring a potent CDK12/cyclinK inhibition (Tables 3 and 4). Additionally, two different bipyridyl methaneamines (14h, 14i) were incorporated at 6 position that were expected to confer relatively more effective in CDK12 inhibition and cell-growth inhibition based on the results in Table 1. Synthesis was carried out using the same synthetic routes as before, but for 2,2′-bipyridyl carbonitrile (12i) that was prepared using a modified Negishi coupling reaction [27,28] (Scheme 4). Compounds 30d (CDK12/cyclinK IC50 = 21 nM) and 30e (CDK12/cyclinK IC50 = 85 nM) containing the 2′-pyridyl terminal moiety at 6 position showed less potent enzymatic activity compared with 28a, 30a, 30b (IC50 values < 30 nM), but showed 2-3 fold higher activity in cancer cellgrowth inhibition (GI50 values = 34-52 nM) (Table 5), which is consistent with the SAR trend shown in Table 1. Having the comprehensive SAR results in hands, we finally synthesized several analogues that were anticipated to have potent activities against both CDK12/cyclinK enzyme and cell growth inhibition. We fixed 9 position with the ethyl group and introduced the selected heteroaryl groups at 2 position, conferring a potent CDK12/cyclinK inhibition (Tables 3 and 4). Additionally, two different bipyridyl methaneamines (14h, 14i) were incorporated at 6 position that were expected to confer relatively more effective in CDK12 inhibition and cell-growth inhibition based on the results in Table 1. Synthesis was carried out using the same synthetic routes as before, but for 2,2 -bipyridyl carbonitrile (12i) that was prepared using a modified Negishi coupling reaction [27,28] (Scheme 4). Compounds 30d (CDK12/cyclinK IC 50 = 21 nM) and 30e (CDK12/cyclinK IC 50 = 85 nM) containing the 2 -pyridyl terminal moiety at 6 position showed less potent enzymatic activity compared with 28a, 30a, 30b (IC 50 values < 30 nM), but showed 2-3 fold higher activity in cancer cell-growth inhibition (GI 50 values = 34-52 nM) (Table 5), which is consistent with the SAR trend shown in Table 1.

Docking Analysis
We performed a docking analysis for 30d using the reported co-crystal structure of R-CR8-bound CDK12-DDB1 complex [23] (Figure 5). As shown in R-CR8, hydrogen bonds were expected between a pair of N7 and NH on the purine ring and the hinge Met816 backbone. The Ethyl group at 9 position sits in the small hydrophobic pocket created by three hydrophobic side chains of Val787, Phe813, and Leu866. The aminopyridine group at 2 position forms a hydrogen bond with the carbonyl backbone of Glu735, as well as hydrophobic interactions with Ile733 and Val741. The inner pyridine at 6 position is predicted to form a hydrogen bond with the side chain of Tyr815 and a hydrophobic contact with Ile733. In addition, the terminal pyridine at 6 position interacts with DDB1 through a hydrogen bond with Asn907 and hydrophobic interactions with the hydrophobic side chains of Ile909 and Arg928.

Docking Analysis
We performed a docking analysis for 30d using the reported co-crystal structure of R-CR8-bound CDK12-DDB1 complex [23] (Figure 5). As shown in R-CR8, hydrogen bonds were expected between a pair of N7 and NH on the purine ring and the hinge Met816 backbone. The Ethyl group at 9 position sits in the small hydrophobic pocket created by three hydrophobic side chains of Val787, Phe813, and Leu866. The aminopyridine group at 2 position forms a hydrogen bond with the carbonyl backbone of Glu735, as well as hydrophobic interactions with Ile733 and Val741. The inner pyridine at 6 position is predicted to form a hydrogen bond with the side chain of Tyr815 and a hydrophobic contact with Ile733. In addition, the terminal pyridine at 6 position interacts with DDB1 through a hydrogen bond with Asn907 and hydrophobic interactions with the hydrophobic side chains of Ile909 and Arg928. Figure 5. Docking result of 30d in CDK12-DDB1 complex. Docking was performed based on the Xray co-crystal structure of R-CR8 (pdb id: 6td3), and visualized using Pymol2.5 software (Schrödinger, New York, NY, USA). CDK12, DDB1, and 30d are colored in cyan, green, and purple, respectively. The labeled residues are predicted to interact with 30d. Predicted hydrogen bonds are highlighted in yellow dotted lines.

Liver Microsomal Stability and CYP Inhibition
We then examined the five potent analogues in Table 4 for in vitro metabolic stability in liver microsomes from three different species (human, dog, mouse) and for inhibitory activity against the 5 representative cytochrome P450 enzymes (CYPs) ( Table 6). We found that for liver microsomal stability, the 2 -pyridyl group is more suitable than α-methyl-4pyridyl group as the terminal aromatic group at 6 position (30d vs. 30a, 30e vs. 30b), and the aminopyrimidyl group is superior to the aminopyridyl group as a substituent at 2 position (30b vs. 30a, 30e vs. 30d). The five analogues showed a similar, desirable CYP inhibition profile except for CYP3A4, suggesting that they could be readily used in combination with other agents. Among them, the analogues containing aminopyrimidine group at 2 position (30b, 30e) exhibited only a slight inhibition against all 5 CYPs, implying that the aminopyrimidine is the most suitable substituent at 2 position to avoid the inhibition of CYPs. Together, 30e was the best analogue in terms of in vitro liver metabolic stability and CYPs' activity conservation.

Suppression of Cyclink and PolII CTD Phosphorylation
We examined the intracellular target inhibition in SK-Br3 and HCC1954 cells after treatment of potent analogues 30d and 30e at 40 and 200 nM for 2 h ( Figure 7A). In both cells, cyclinK was greatly downregulated in a dose-dependent manner, suggesting that both compounds act as a potent cyclinK degrader. Additionally, 30d and 30e showed a strong, dose-dependent suppression of the levels of PolII p-CTD (Ser2). Compared with 30d and 30e, dinaciclib at 40 nM showed only a marginal downregulation of cyclinK, but exhibited a more potent inhibition of PolII CTD phosphorylation in both cells. We also examined the expression levels of CDK12 downstream genes (IRS1 and WNT1) [10] following treatment of the same doses of compounds for 24 h ( Figure 7B). Both 30d and 30e showed a strong suppression of IRS1 and WNT1, but 30e was more prominent downregulation as comparable to dinaciclib.

Synergism between 30d and Trastuzumab
Additionally, we examined a combination effect in both SK-Br3 and HCC1954 cells. We treated multiple doses of trastuzumab for 72 h in the absence or presence of 30d at a single dose (40 nM) around its GI50 value ( Figure 8). We observed a slight gain in the inhibitory activity of trastuzumab when it was co-treated with 30d in both cell lines, demonstrating a modest level of synergism between 30d and trastuzumab in growth inhibition of HER2+ breast cancer cells, regardless of their trastuzumab sensitivity.

Synergism between 30d and Trastuzumab
Additionally, we examined a combination effect in both SK-Br3 and HCC1954 cells. We treated multiple doses of trastuzumab for 72 h in the absence or presence of 30d at a single dose (40 nM) around its GI 50 value ( Figure 8). We observed a slight gain in the inhibitory activity of trastuzumab when it was co-treated with 30d in both cell lines, demonstrating a modest level of synergism between 30d and trastuzumab in growth inhibition of HER2+ breast cancer cells, regardless of their trastuzumab sensitivity.

Synergism between 30d and Trastuzumab
Additionally, we examined a combination effect in both SK-Br3 and HCC1954 cells. We treated multiple doses of trastuzumab for 72 h in the absence or presence of 30d at a single dose (40 nM) around its GI50 value ( Figure 8). We observed a slight gain in the inhibitory activity of trastuzumab when it was co-treated with 30d in both cell lines, demonstrating a modest level of synergism between 30d and trastuzumab in growth inhibition of HER2+ breast cancer cells, regardless of their trastuzumab sensitivity.

Chemistry
All reagents and solvents purchased from commercial sources were used without further purification. Nuclear magnetic resonance (NMR) solvents were purchased from Cambridge Isotope Laboratories Inc. All reactions were monitored under UV light (254 nm) using a thin layer chromatography on pre-coated silica gel glass plates from Merck. Flash column chromatography was performed using silica gel (Kieselgel 60 Art. 9385, 230-400 mesh) from Merck. 1 H and 13 C NMR spectra were recorded on Bruker 400 MHz FT-NMR. Chemical shifts are reported in parts per million (ppm, δ) using peaks from an NMR solvent (CDCl 3 , CD 3 OD, or DMSO-d 6 ) as a reference. Coupling constants (J) are reported in Hertz (Hz), and the multiplicities of peaks are abbreviated as s: singlet, br: broad singlet, d: doublet, t: triplet, q: quartet, dd: doublet of doublet, dt: doublet of triplet, and m: multiplet. High-resolution mass-spectral results were obtained using Orbitrap Eclipse™ Tribrid™ Mass Spectrometer (ThermoFisher Scientific, Waltham, MA, USA).

General Procedure for Synthesis of 12a-h
To a stirred solution of corresponding aryl bromide (1 eq.), pyridyl boronic acid/ester (1.2 eq.), and Pd(PPh 3 ) 4 (0.05 eq.) in 1,4-dioxane (45 mL) under N 2 atmosphere, was added 5 mL of a 2 M aq. K 2 CO 3 solution. The vigorously stirred mixture was heated to 100 • C for 12 h. After cooling, the mixture was combined with EtOAc, and washed with water and brine. The organic layer was dried over MgSO 4 , filtered, and concentrated under a reduced pressure. The residue was purified by column chromatography (n-hexane:EtOAc = 9:1) to afford bipyridine carbonitriles as white solids. 12i was obtained following the literature procedure [28].

General Procedure for Synthesis of 14a-i
Intermediate 12a-i (ca. 20 g) was added to methanol (120 mL) and cooled to 0 • C. Di-tert-butyl dicarbonate (2 eq.) was added and the suspension was stirred for 15 m. Then, NiCl 2 ·6H 2 O (0.3 eq.) was added and stirred for 5 m. Next, NaBH 4 (3.5 eq.) was added in portion-wise for 30 m. After the addition was completed (ca. 30 m), the ice bath was removed and the mixture was stirred with warming to rt overnight. After the reaction was completed, diethylenetriamine (1 eq.) was added to the stirring mixture. After 15 min, methanol was evaporated and we added 100 mL of aq. NaHCO 3 . After the extraction, using EtOAc (3 × 80 mL), the organic layer was dried over MgSO 4 , evaporated under a reduced pressure, and subjected to flash chromatography (2% methanol in dichloromethane) to afford a Boc-protected intermediate (13a-i). Then, the intermediate was dissolved in 50 mL of dichloromethane and cooled to 4 • C, to which we slowly added 10 mL of 4N HCl in 1,4-dioxane and stirred for 1 h at rt. The eluted solid in dichloromethane was filtered and dried to afford a light brown salt as a pure product (14a-i).

General Procedure for Synthesis of 15a-c
Both 2,6-dichloro-9H-purine (10.0 g, 53.2 mmol) and K 2 CO 3 (21.9 g, 159 mmol) were dissolved in 70 mL anhydrous DMSO. Alkylbromide (133 mmol) was added dropwise to the reaction mixture at rt, and stirred overnight. Upon completion of the reaction, the reaction mixture was poured into ice water and extracted with EtOAc, and dried over MgSO 4 . The concentrated mixture was subjected to a column chromatography using nhexane/EtOAc (3:1) as eluent. Pure products were obtained with yield 40-70% based on alkyl groups.

General Procedure for Synthesis of 20a-d
The anilino compound 19a-c (70 mg, 1.0 eq.) and trimethylamine (2.0 eq.) were added to a solution of respective bromoalkylalcohol (1.5 eq) in n-butanol (1.0 mL) at rt. The reaction mixture was heated with stirring at 110 • C for 12 h. After the reaction was completed, the mixture was cooled to rt and the solvent was evaporated. The residue was diluted with water and extracted with EtOAc (3 × 50 mL). The combined organic extracts were dried over MgSO 4 , filtered, and concentrated under a reduced pressure. The desired product was obtained by a column chromatography using 5% methanol in dichloromethane as eluent.