Synthesis of two ‘heteroaromatic rings of the future’ for applications in medicinal chemistry

Abstract

In a computational study, the 1H-pyrazolo[3,4-c]pyridin-5-ol and 2,6-naphthyridin-3-ol heterocycles were identified as unknown heteroaromatic ring systems of potential value for medicinal chemistry. Here we report robust and concise synthetic protocols that provide access to these two scaffolds on a multigram scale.

---

Heterocycles are an indispensable component of life occurring in a vast array of natural small molecules as well as biopolymers, such as proteins and nucleic acids. Consequently, it is not surprising that medicinal chemists are heavily dependent on heterocycles for drug discovery. In addition to mimicking natural motifs, heterocycles serve important functions as bioisosteres and their inclusion can profoundly influence a drug's physicochemical properties. Their prevalence can be appreciated by examining the top 200 pharmaceutical products by US retail sales in 2012.¹ Among the 147 small molecule entries, 97 (66%) incorporate at least one heterocycle.

A recent publication by Taylor et al. has identified the heterocycles that are most frequently present in drug molecules.² The adoption of a given heterocycle by the medicinal chemistry community has largely relied on synthetic tractability and historical precedents with known scaffolds. While this conservative strategy may be a valid choice given the high rate of attrition in drug discovery, it is not clear if popular heterocycles are inherently superior or if less explored alternatives could be equally profitable for the discovery of drug-like compounds. In other words, in addition to ‘privileged scaffolds’, are there ‘underprivileged scaffolds’ that could enable access to unexploited areas of chemical space and have the added advantage of avoiding overlap with the patent coverage of competitors? In order to access such rare heterocycles, it is first necessary to design reliable synthetic routes for their preparation. For example, we have recently reported the solution- and solid-phase synthesis of 2,4,5-trisubstituted-1,2,4-triazine-3,6-diones.³ This heterocyclic ring system can be considered an aza-analogue and bioisostere of diketopiperazines, but has far fewer reports than the latter in the literature.

Here we describe our efforts in the synthesis of two scaffolds that belong to the underprivileged category. Our starting point was the computational study conducted by UCB Celltech, which sought to identify ‘Heteroaromatic rings of the future’.⁴ A virtual library of all the 24847 possible small heteroaromatic ring systems was generated. Of these heterocycles, only 1701 were described in the literature and over 3000 synthetically tractable unpublished heteroaromatic ring systems were identified using a machine learning approach. These were further filtered down to 22 heterocycles containing four or fewer heteroatoms that were described as a ‘challenge to creative organic chemists to either make or explain why they cannot be made!’ We have risen to the challenge and targeted two of these heterocycles, 1,6-dihydro-5H-pyrazolo[3,4-c]pyridin-5-one 1a and 2,6-naphthyridine-3(2H)-one 2a, as depicted in the UCB Celltech publication and shown here with their alternative tautomeric forms 1 and 2 (Fig. 1).

Fig. 1 Structures of two scaffolds 1 and 2 identified as ‘Heteroaromatic rings of the future’ and of bioisosteric scaffolds 3–6.

These heterocycles represent a complementary pair containing a pyridine-2-ol unit fused to either a π-excessive or a π-deficient heterocycle. While neither of the parent heterocycles, 1 and 2, have previously appeared in the medicinal chemistry literature, they are isosteres of more familiar fused-ring heteroaromatic systems such as 3–6.

1H-Pyrazolo[3,4-c]pyridin-5-ol (1)

Chapman and Hurst's investigations on pyrazolopyridines inspired our synthesis of scaffold 1. As part of this work, a 1980 publication reported the acetylation of pyridine 7 to 8 and its conversion to 12. It is proposed that this conversion proceeds via the fragmentation of intermediate N-nitroso pyridine 9 to diazonium salt 10, followed by cyclization and acetylation to 12.⁵ Hydrolysis then afforded 11, which is the methoxy derivative of scaffold 1 (Scheme 1). We repeated this procedure by replacing nitrosyl chloride with the safer and more easily handled sodium nitrite for nitrosation, as in Zhu's analogous cyclization.⁶ Although Chapman and Hurst do not report the scale of the final reactions, we were able to prepare 11 in multigram quantities using this method. In a similar fashion, we synthesized the chloride derivative 15. In this case, the acetylation, nitrosation, and cyclization of 13 to 14 and acetate hydrolysis to 15 were accomplished by one-pot simplification of the original procedure. However, we were unable to reach the parent heterocycle 1, either by demethylation of 11 or by displacement of the chloride in 15, under a variety of reaction conditions. Although the demethylation of 11 with TMSCl was recently reported as part of an NMR study by Tsikouris et al.,⁷ we were unsuccessful in similar attempts. Additionally, demethylation methods with boron tribromide,⁸ CH₃SNa,⁹ and AlCl₃ ¹⁰ were also fruitless. For chloride replacement in 14 we used several aromatic substitutions described for 2-chloropyridines using both acidic^11–13 and basic catalysis;¹⁴ however, none of these methods were successful.

Scheme 1 Synthetic strategies to develop 1.

Faced with the above-mentioned difficulties, we turned to an alternative starting material lacking the methoxy substitution present in 7. RANEY® nickel reduction of nitropyridine 16 (Scheme 1), which is coincidentally less expensive than 7, afforded pyridone 17, which was subjected to one-pot acetylation and nitrosation. This step yielded a complex mixture of products, from which the acetylated pyrazolopyridine 18 (32%) was isolated, alongside traces of 1 and monoacylated derivatives.

This three-step sequence from an inexpensive starting material can also be conveniently carried out using the one-pot methodology. This one-pot procedure proved to be more effective, yielding 82% of 1 as its hydrochloride salt without requiring any chromatographic purification of the intermediates, even on scales larger than 1 g. In solution, IR and NMR spectroscopy reveal that 1 is present in the hydroxy pyridine tautomer with a characteristic O–H stretch at 3154 cm⁻¹.

2,6-Naphthyridin-3-ol (2)

Our retrosynthetic approach to the unknown heterocycle 2 was based on oxidation of dihydro derivative 19. We assumed that 19 would be formed via spontaneous intramolecular lactamization upon reduction of nitrile 20, a recently described intermediate in a Novartis patent.¹⁵ In practice, the ethyl ester 21 (Scheme 2) of homonicotinic acid was converted to N-oxide 22 and alkylated to give 23 in 87% overall yield. As reported by Novartis, nucleophilic addition of cyanide and concomitant rearomatization¹⁶ provided the desired 4-cyano derivative 20 of the homonicotinate. Catalytic hydrogenation of the nitrile proceeded as envisioned, with the intermediate amine undergoing cyclization to lactam 19.

Scheme 2 Synthesis of a ‘Heteroaromatic ring of the future’ 2 from 21.

The dehydrogenation of 19 to heteroaromatic ring system 2 proved to be challenging. A number of investigated methods, such as Pd/C,¹⁷ DDQ,¹⁸ and MnO₂,¹⁹ did not lead to the desired product. Success was finally achieved with Hayashi's method using activated charcoal in an oxygen-rich atmosphere.²⁰ Overall, this five step route is amenable to the preparation of 2 in good yield, and this experimental procedure is carried out on multigram scale. As was the case for 1, the spectroscopic evidence indicated that the hydroxy tautomer 2 was present in solution with an O–H stretching band at 3261 cm⁻¹ in the IR spectrum and a broad singlet for the OH proton at δ 11.25 in the ¹H NMR spectrum in DMSO-d₆.

Physicochemical properties

In addition to serving as a template for further functionalization, heterocycles 1 and 2 should be of interest for researchers involved in fragment-based drug discovery. The physicochemical properties of both heterocycles fit with the rule of three²¹ proposed for fragments (Table 1), and the values are more compliant than those for the isosteric heterocycles indazole (3), 3-hydroxyisoquinoline (5), and isoquinoline (6). This comparison is particularly evident with respect to lipophilicity and ionization at physiological pH (7.4).

Table 1 Predicted physicochemical properties for the synthesized compounds

Property	1	2	3	5	6
a Properties calculated with VORTEX® (DOTMATICS, UK): MW (molecular weight), HBA (hydrogen bond acceptor), HBD (hydrogen bond donor), TPSA (topological polar surface area), HAC (heavy atoms count) and alog P (calculated logarithm of partition-coefficient), and alog S (logarithm of solubility-coefficient).b Properties calculated with Marvin Sketch 15.7.20.0 (CHEMAXON Ltd): log P (calculated lipophilicity), log D 5.5–7.4 (logarithm of distribution-coefficient).
MW^a	135	146	118	145	129
HBA^a	4	3	2	2	1
HBD^a	2	1	1	1	0
TPSA^a	61.8	46	28.7	33.1	12.9
HAC^a	10	11	9	11	10
alog P^a	0.05	0.85	1.61	2.5	2.14
log P^b	0.57	0.84	1.47	2.15	1.74
log D5.5^b	0.57	0.84	1.47	2.15	1.57
log D7.4^b	0.57	0.84	1.30	2.15	1.73
alog S^a	−1.3	−1.6	−0.75	−1.79	−1.66

---

We employed a spectrophotometric procedure to detect the protonation/deprotonation processes of these heterocycles. Based on the spectral changes observed in aqueous solutions at different pH, we could detect two pK_a values for 1, at 4.5 and 11.5. For heterocycle 2, we observed one pK_a above 12.

For drug discovery applications, the aqueous solubility of the heterocycles is of paramount importance. The experimentally determined aqueous solubilities of the heterocycles 1 and 2 were 301 g L⁻¹ and 2.1 g L⁻¹, respectively, which are higher than those of the common isosteres (ESI Table 1†).

Molecular orbital calculations (Table 2) indicate that both heterocycles have similar electronic distributions and energies for their HOMO and LUMO. The density maps suggest that both 1 and 2 can react with electrophiles and nucleophiles at carbon sites as well as with nitrogen and oxygen moieties; this feature permits further functionalization of the heterocycle towards elaborate fragments.

Table 2 Calculated highest occupied molecular orbitals (HOMOs) and lowest unoccupied molecular orbitals (LUMOs) for heterocycles 1 and 2 with maps indicating the electron density of the LUMO (color range: 0 = red to 0.026 = blue) using Spartan'10 software (B3LYP/6-31g*)

	HOMO	HOMO map	LUMO	LUMO map
1
2

---

Conclusions

In summary, we report a short and inexpensive route to heterocycle 1 and to the previously unknown heterocycle 2, both of which were considered synthetic challenges and were identified as promising scaffolds for medicinal chemistry. These heterocycles have high aqueous solubility and are not significantly ionized at physiological pH. Therefore, we recommend that medicinal chemists consider them for diverse applications and carry out further study. Currently, we are exploring the incorporation of the parent heterocycles into drug-like molecules and the results will be reported in due course.

Acknowledgements

The authors are grateful to the funding agencies São Paulo Research Foundation (FAPESP – grants #2011/23342-9; #2012/20990-2 and #2013/26485-0), CAPES (grant SVR 119/2012), and CNPq for grants and scholarships.

Notes and references

1. N. A. McGrath, M. Brichachek and J. T. Njardarson, J. Chem. Educ., 2010, 87, 1348.
2. R. D. Taylor, M. McCoss and A. D. G. Lawson, J. Med. Chem., 2014, 57, 5845.
3. D. Bonnet, J. F. Margathe, S. Radford, E. Pflimlin, S. Riché, P. Doman, M. Hibert and A. Ganesan, ACS Comb. Sci., 2012, 14, 323.
4. W. R. Pitt, D. M. Parry, B. G. Perry and C. R. Groom, J. Med. Chem., 2009, 52, 2952.
5. D. Chapman and J. Hurst, J. Chem. Soc., Perkin Trans. 1, 1980, 2398.
6. G. D. Zhu, J. Gong, V. B. Gandhi, K. Woods, Y. Luo, X. Liu, R. Guan, V. Klinghofer, E. F. Johnson, V. S. Stoll, M. Mamo, Q. Li, S. H. Rosenberg and V. L. Giranda, Bioorg. Med. Chem., 2007, 15, 2441.
7. O. Tsikouris, T. Bartl, J. Tousek, N. Lougiakis, T. Tite, P. Marakos, N. Pouli, E. Mikros and M. Radek, Magn. Reson. Chem., 2008, 46, 643.
8. J. F. W. McOmie, M. L. Watts and D. E. West, Tetrahedron, 1968, 24, 2289.
9. G. I. Feutrill and R. N. Mirrington, Tetrahedron Lett., 1970, 11, 1327.
10. C. Hartmann and L. Gatterman, Chem. Ber., 1892, 25, 3531.
11. Y. Hirata, S. Ito, H. Kawamoto, T. Kimura, Y. Nagatomi, H. Ohta, A. Sato, A. Satoh and G. Suzuki, Diaryl-substituted five-membered heterocycle derivative, World patent 85214, 2005.
12. S. D. Taylor, F. Mirzaei, A. Sharifi and S. L. Bearne, J. Org. Chem., 2006, 71, 9420.
13. E. Fukuda, Y. Takahashi, N. Hirasawa, O. Sugimoto and K. Tanji, Heterocycles, 2009, 77, 1163.
14. G. Lavecchia, S. Berteina-Raboin and G. Guillaumet, Tetrahedron Lett., 2004, 45, 6633.
15. M. G. Bock, C. Gaul, V. R. Gummadi, H. Moebitz and S. Sengupta, 17 alpha-Hydroxylase/C17,20-lyase inhibitors. World patent 35078, 2012.
16. W. E. Feely and E. M. Beavers, J. Am. Chem. Soc., 1959, 81, 4004.
17. E. C. Horning, J. Org. Chem., 1945, 10, 263.
18. E. A. Braude, A. G. Brook and R. P. Linstead, J. Chem. Soc., 1954, 3569–3574.
19. A. B. A. Jansen, J. M. Johnson and J. R. Surtees, J. Chem. Soc., 1964, 5573.
20. T. Tanaka, K. Okunaga and M. Hayashi, Tetrahedron Lett., 2010, 51, 4633.
21. M. Congreve, R. Carr, C. Murray and H. Jhoti, Drug Discovery Today, 2003, 8, 876.

---

Footnote

† Electronic supplementary information (ESI) available: Synthetic procedures; solubility, pK_a, and computational data. See DOI: 10.1039/c6ra01099g

---