Facile Synthesis of 7-Alkyl-1,2,3,4-tetrahydro-[1,8]-naphthyridines as Arginine Mimetics Using a Horner–Wadsworth–Emmons Based Approach

Integrin inhibitors based on the tripeptide sequence Arg -Gly-Asp (RGD) are potential therapeutics for the treatment of idiopathic pulmonary fibrosis (IPF). Herein, we describe a n ex peditious three-step synthetic sequence of Horner –Wadsworth– Emmons olefination, diimide reduction and global deprotection to synthesise cores for these compounds in high yields (63 –83% over 3 steps) with no need for chromatography. Key to this transformation is the phosphoramidate protecting group, which is stable to metalation steps. 2

Arginine; Horner–Wadsworth–Emmons; Integrin; Phosphoramidate; Tetrahydronaphthyridine

Tetrahydronaphthyridines are prominent in peptidomimet ic pharmaceuticals as arginine mimetics; they are widely used in Arg-Gly-Asp (RGD) peptide mimetics such as αv integrin inhibitors [1]. Tetrahydronaphthyridines represent less basic but more permeable alternatives to arginine (pKa ≈7 versus 13.8) [1], replicating the side-on salt- bridge binding interaction made between the guanidinium functionality of arginine and an aspartic acid residue in the protein . Consequently, this moiety has been used in various integrin inhibitors (Figure 1) [2-7]. Figure 1. The Arg -Gly-Asp tripeptide sequence and e xamples of tetrahydro-1,8- naphthyridine-containing integrin inhibitors. 3 Current routes to install tetrahydronaphthyridines predominantly revolve around late- stage hydrogenation of fully unsaturated 1,8 -naphthyridine derivatives 3, usually prepared via an acid or base-catalysed Friedlander reaction between 2 - aminonicotinaldehyde 1 and the corresponding ketone 2 (Scheme 1). Both reactions employ harsh conditions with limited functional group tolerance and lack of regiochemical control, which present s considerable purification issues during large - scale synthesis [8-10]. Scheme 1. Commonly used synthetic routes to tetrahydro-1,8-naphthyridine moieties by hydrogenation of saturated naphthyridines 3. Recently, GlaxoSmithKline disclosed a route to such a fluoropyrrolidine 6 using a Wittig reaction between phosphonium salt 4 and aldehyde 5 [2]. Synthesis of phosphonium salt 4 (itself requiring 6 steps [including partial saturation of a 1,8 -naphthyridine moiety)] and the formation of the triphenylphosphine oxide by-product in the Wittig step presented complications on both gram and kilogram scales. Herein, we report a novel synthetic sequence to tetrahydro-1,8-naphthyridines using a Horner -Wadsworth- Emmons reaction using diphosphorylated compound 7, proceeding in high yields and high purities without the need for chromatographic purification (Scheme 2). 4 Scheme 2. Previous synthetic route to fluoropyrrolidine 6 utilising a Wittig reaction and a novel, higher yielding route using a Horner–Wadsworth–Emmons reaction.

and Discussion Investigation initially began using commercially available N-Boc protected tetrahydro- 1,8-naphthyridine 8; however, upon deprotonation and quenching with diethyl chlorophosphate, migration of the Boc group from the nitrogen atom to the exocyclic methyl group was observed, affording phosphoramidate 9 in low yield with no formation of phosphonate 10 seen (Scheme 3). Scheme 3. Synthesis of phosphoramidate 9 from tetrahydro -1,8-naphthyridine 8. Conditions: sec-BuLi (3 eq.), diethyl chlorophosphate (1.1 eq.), THF, −42 °C, 14% yield. 5 It was proposed that deprotonation of 7 -methyl-1,2,3,4-tetrahydro-1,8-naphthyridine 11 with two equivalents of sec-BuLi would afford phosphonate 12 upon quenching with diethyl chlorophosphate via formation of the dianion. This could then be used in a subsequent Horner–Wadsworth–Emmons reaction to construct the carbon skeleton of amine 6. Up on addition of a single equivalent of diethyl chlorophosphate, phosphoramidate 13 was obtained exclusively at both −42 and −78 °C. Addition of two equivalents of the chlorophosphate yielded diphosphorylated compound 7, albeit in poor yield (Scheme 4). Scheme 4. Mono- and di -phosphorylation of tetrahydro -1,8-naphthyridine 11. Conditions: (i) sec-BuLi (2 eq.), diethyl chlorophosphate (1 eq.), THF, −78 °C, 44% yield; (ii) sec-BuLi (2 eq.), diethyl chlorophosphate (2 eq.), THF, −78 °C, 27% yield. Deprotonation of phosphonate 7 and reaction with aldehyde 5 (formed in situ by oxidation of alcohol 14 using T3P®) [11] yielded olefin 15 in 93% yield as a 94:5 mix of stereoisomers (presumably E/Z, although this is not conclusive from the 1H NMR spectrum) Reduction to compound 16 using diimide, generated in situ, proceeded in 80% yield and was followed by single-pot carbamate and phosphonate deprotection to afford arginine mimetic 6 in 86% yield. This represents a 64% overall yield which was increased to 68% when no column chromatography was undertaken between transformations, with no loss of purity (Scheme 5). 6 Scheme 5. Synthesis of amine 6 from phosphonate 7 and aldehyde 5. Conditions: (i) T3P® (50% w/w in DCM) (3 eq.), DMSO (3 eq.), DIPEA (2.5 eq.), DCM, 0 °C; (ii) KOtBu (6 eq.), THF, 0 °C, 93% yield (relative to phosphonate 7); (iii) PhSO 2NHNH2 (3 eq.), K2CO3 (4 eq.), DMF, 100 °C, 80% yield; (iv) 7.4 M HCl, 100 °C, 86% yield. Olefin reduction and Cbz deprotection could not be performed simultaneously by palladium-catalysed hydrogenation as this results in defluorination , presumably via a Tsuji–Trost-like elimination of the allylic fluoride [12-13]. This sequence represents a marked improvement from the Wittig-including route, lowering the number of synthetic steps and increasing overall yield [2]. Furthermore, no problematic by -products are formed, and good purity is obtained without the use of any chromatography, which is ideal for large-scale processes. Optimisation of the Synthesis of Phosphoramidate 13 and Phosphonate 7 Having been shown to be a feasible intermediate, attention turned to improving the synthesis of bis-phosphonate 7 via a two-step process, exploiting the base stability of the phosphoramidate protecting group. A variety of bases were trialed at 0 °C for the initial N-phosphorylation, with 10 minutes allowed for complete deprotonation to occur (Table 1). 7 Table 1. Bases surveyed for the formation of phosphoramidate 13a Entry Base Amount of phosphoramidate 13 / LCMS a/a% 1 KOtBu (1 M in THF) 4 2 LiHMDS (1 M in THF) 58 3 LDA (2 M in hexanes/benzene) 58 4 sBuLi (1.4 M in cyclohexane) 65 5 iPrMgCl (2 M in THF) 88 aReactions performed on 0.7 mmol scale of compound 11 Minimal phosphorylation was observed when using potassium tert-butoxide (Entry 1); this may be due the disparity in p Ka between the base and the tetrahydro-1,8- naphthyridine (based on 2 -aminopyridine, th e p Ka of the saturated ring nitrogen is expected to be ≈28) [14]. Similarly, nitrogen -centred bases (Entries 2,3) gave moderate conversions to phosphoramidate 13 due to a close match of the p Ka of tetrahydronaphthyridine 11 and the pKaH of the base; sec-BuLi also gave reasonable conversion to (Entry 4) albeit with some impurities. This can likely be attributed to the temperature instability of the base and/or lithiated tetrahydronaphthyridine in THF at this temperature. Use of iPrMgCl, a strong but room temperature-stable base, gave the greatest conversion to phosphoramidate 13. Further investigation into the use of iPrMgCl found that quantitative conversion was achieved at ambient temperature with a metalation time of <1 minute. Pleasingly, premixing of tetrahydronaphthyridine 11 8 with diethyl chlorophosphate, followed by drop -wise addition of iPrMgCl gave clean and total conversion to phosphoramidate 13 as seen by LC-MS. When performed on a multi -gram scale, a 94% yield of compound 13 was obtained. The re action also proceeded well (91% isolated yield) in 2-MeTHF, which offers a preferred alternative if performed on larger scale due to better partitioning with wa ter, stability and sustainability of production [15]. Of the bases trialled, only sec-BuLi was efficient in promoting C-phosphorylation. Optimisation of the use of this base was then investigated further (Table 2). Table 2. Optimisation of formation of phosphonate 7 using sec-BuLi as a basea Entry T / °C Eq. sec-BuLi Lithiation time / min Amount of phosphonate 7 / LCMS a/a% 1 −78 1.5 90 53 2 −42 2.0 20 62 3 −42 2.5 20 81 4 −42 3.0 20 86 5 −42 3.5 20 67 6 −42 4.0 20 48 7 −42 3.0 (+ 3 eq. TMEDA) 20 59 8 0 2.0 10 0 aReactions performed on 4.6 mmol scale (Entry 1) , 0.5 mmol scale (Entries 2 –6), 0.3 mmol (Entry 7) or 0.4 mmol (Entry 8). 9 Optimal deprotonation and phosphorylation was found to occur when an excess (3 eq.) of base was used at −42 °C (Entry 4), with a lithiation time of 20 min. Addition of TMEDA was detrimental to this conversion with a large proportion of starting material 13 remaining (Entry 7). No product was formed at higher temperatures (Entry 8), likely due to instability of the C-lithiated species at elevated temperatures as degradation was observed. It is believed that the excess base loading is required to account for deprotonation of the more acidic phosphonate product 7 versus the starting material 13 and potential lithium sequestration by chelation between an oxygen atom of the phosphonate and the nitrogen atom of the unsaturated ring. Monitoring of deprotonation followed by quenching with CD 3OD by 1H and 13C NMR spectroscopy indicated mono -labelled 17 as the major product, demonstrating that lithiation only occurs at a single position of compound 13 (Scheme 6). Scheme 6. Monodeuteration of 13 as observed by 1H and 13C NMR. Conditions: sec- BuLi (3 eq.), THF, −42 °C then CD3OD (14 eq.). When performed on a multi-gram scale, phosphonate 7 was synthesised in 68% yield after purification by chromatography on silica. This, combined with the formation of phosphoramidate 13 in 94% yield, represents a marked improvement to the initial simultaneous diphosphorylation (Scheme 7). When performed sequentially in a single- pot, dip hosphorylated compound 7 was not observed, with phosphoramidate 13 accounting for the majority of product formed [16]. 10 Scheme 7. Sequential diphosphorylation of tetrahydronaphthyridine 11. Conditions: (i) iPrMgCl (1.5 eq.), THF, then diethyl chlorophosphate (1.2 eq.), 94% yield; (ii) sec-BuLi (3 eq.), THF, −42 °C then (EtO)2P(O)Cl (1.1 eq.), 68% yield Reaction Scope The sequence of olefination, reduction and deprotection was tested on othe r arginine mimetics of varying amine structure, constituting potential Arg-Gly components of Arg- Gly-Asp inhibitors (Table 3). Where the aldehyde was not commercially available, N- Boc-protected alcohols were oxidised using IBX in refluxing ethyl acetate an d used crude. All amines were formed in high NMR and LC -MS purity without the need for purification by column chromatography. Table 3. Cores synthesised by the sequence of olefination, reduction and deprotection and the corresponding starting alcohols and aldehydes.a Entry Alcohol Aldehyde Product Yield 1 82%1 11 2 83%1 3 - 79%2 4 - 63%3 *Conditions: 1: IBX, EtOAc, reflux then KO tBu, THF, 0 °C then K 2CO3, PhSO 2NHNH2, DMF, 100 °C then 7.4 M HCl, 100 °C; 2: KOtBu, THF, 0 °C then K2CO3, PhSO2NHNH2, DMF, 100 °C; 7.4 M HCl, 100 °C; 3: NaH, THF, 0 °C; then K2CO3, PhSO2NHNH2, DMF, 100 °C then 7.4 M HCl, 100 °C Pleasingly, the sequence proceeded in high yields (63–83%) for all substrates with no chromatography required. Despite relatively high yield s, significant racemisation was seen in the synthesis of piperidine 20 and pyrrolidine cores 23. Pyrrolidine 23 was obtained with an e.e. of only 24%, representing a serious loss of enantiopurity. It is believed that racemisation occurs during the olefination step, caused by base mediated keto-enol tautomerisation. As such, this sequence, as currently performed, is suitable for substrates lacking an acidic α-proton (fluoropyrrolidine core 6) or achiral aldehydes (azetidine cores 25 and 27). Furthermore, key to success of the Horner–Wadsworth–Emmons olefination is premixing of the aldehyde and phosphonate 7 prior to addition of KOtBu. Upon deprotonation, phosphonate 7 (in the absence of aldehyde) undergoes dimerisation to olefin 28. While the exact mechanism is not known, it is likely to involve a reacti ve carbene intermediate formed by α -elimination of the phosphonate as described in previous reports (Scheme 8) [17]. 12 Scheme 8. Possible mechanism for the formation of dimer 28 via a reactive carbene intermediate. In order to circumvent racemisation of aldehyde 22 during the Horner–Wadsworth– Emmons olefination, alkylation of phosphoramidate 13 was explored using commercially available iodide 29. Formation of compound 30 proceeded in 21% yield, with alcohol 31 and dimer 32 also formed in 20% and 5% yield respectively ( Scheme 9). Indeed, when iodide 29 was replaced with bromide 33 and tosylate 34 no formation of compound 30 was observed, with alcohol 31 and dimer 32 accounting for the major products. Acidic deprotection of phosphoramidate 30 afforded amine (R)-23 in 92% yield in 99% e.e., offering an alternative route to the Horner –Wadsworth–Emmons- based approach. Scheme 9. Alkylation of phosphoramidate 13 by iodide 29 to afford compound 30 and by-products alcohol 31 and dimer 32. Use of bromide 33 or tosylate 34 afforded only compounds 31 and 32. Conditions: (i) sec-BuLi (1.3 eq.), iodide 29, THF, −78 °C; (ii) 7.4 M HCl, 100 °C, 92% yield. 13 The mechanism of the formation of alcohol 31 and dimer 32 was not fully explored; however, when iodide 29 was replaced by a superior oxidant in 1,2-dibromoethane, formation of dimer 32 increased (17% isolated yield). This supports previously reported proposals that oxidative coupling of the anion can take place, involving a radical pathway [18-20].

In conclusion, a novel method for the assembly of 7-alkyl-1,2,3,4-tetrahydro-[1,8]- naphthyridine-based arginine mimetics has been developed. Synthesis of phosphonate 7 has been optimised, with a sequential diphosphorylation process using commercially available starting materials affording the desired comp ound in 64% overall yield. A Horner –Wadsworth–Emmons/reduction/deprotection procedure has been used to synthesise amines in good yield requiring no chromatography. This methodology utilised the underused base -stable phosphoramidate protecting group , which was superior to the more commonly applied Boc protecting group which was unstable to the lithiation. This synthetic route replaces traditional Wittig and tandem alkylation/reduction methodologies , which suffer from complications arising from troublesome by-products and reaction selectivity; the new procedure proceeds in a higher yield than previously obtained, providing benefits in large-scale manufacture of integrin inhibitors and other arginine peptidomimetics. Supporting Information Supporting Information File 1 14 Detailed experimental procedures, and product characterisation data, along with 1H and 13C NMR spectra.

Thanks to Sean M. Lynn for assistance with NMR spectroscopy analysis and the GSK UK Discovery Analytical team for acquiring HRMS data Funding Financial support for this work was provided by GSK via the GSK/University of Strathclyde Centre for Doctoral Training in Synthetic and Medicinal Chemistry. We thank EPSRC for further funding via Prosperity Partnership EP/S035990/1.

1. Hatley, R., J. D.; Macdonald, S. J. F.; Slack, R. J.; Le, J.; Ludbrook, S. B.; Lukey, P. T.; Angew. Chem. Int. Ed., 2018, 57, 3298–3321. 2. Anderson, N. A.; Campbell -Crawford, M. H. J.; Hancock, A. P.; Pritchard, J. M.; Redmond, J. M.; Patent WO2016046226 (31 Mar 2016). 3. Procopiou, P. A.; Barrett, T. N.; Copley, R. C. B.; Tame, C. J.; Tetrahedron Asymmetry, 2017, 28, 1384–1393. 4. Procopiou, P. A.; Anderson, N. A.; Barrett, J.; Barrett, T. N.; Crawford, M. H. J.; Fallon, B. J.; Hancock, A. P.; Le, J.; Lemma, S.; Marshall R. P.; Morrell, J.; Pritchard, J. M.; Rowedder, J. E.; Saklatvala, P.; Slack, R. J.; Sollis, S. L.; Suckling, C. J.; Thorp, L. R.; Vitulli, G.; Macdonald, S. J. F.; J. Med. Chem., 2018, 61, 8417–8443. 15 5. Hutchinson, J. H.; Halczenko, W.; Brashear, K. M.; Breslin, M. J.; Coleman, P. J.; Duong, L. T.; Fernandez-Metzler, C.; Gentile, M. A.; Fisher, J. E.; Hartman, G. D.; Huff, J. R.; Kimmel, D. B.; Leu, C. -T.; Meissner, R. S.; Merkle, K.; Nagy, R.; Pennypacker, B.; Perkins, J. J.; Prueksaritanont, T.; Rodan, G. A.; Varga, S. L.; Wesolowski, G. A.; Zartman, A. E.; Rodan, S. B.; Duggan, M. E.; J. Med. Chem., 2003, 46, 4790–4798. 6. Cox, J. M.; Ma, L.; Zhou X.; Haimbach, R. E.; Coleman, P. J.; Zhou; H.; Kelley, D. E.; Stoch S. A.; Duong, L. T.; Hoek M.; Patent WO2016154369 (29 Sep 2016). 7. Devasthale, P.; Moore, F.; Zhao, G.; Pieniazek, S. N.; Selvakumar, K.; Dhanusu, S.; Panda, M.; Marcin, L. R.; Patent WO2018089355 (17 May 2016). 8. Yasuda, N.; Hsiao, Y.; Jensen, M. S.; Rivera, N. R.; Yang, C.; Wells, K. M.; Yau, J.; Palucki, M.; Tan, L.; Dormer, P. G.; Volante, R. P.; Hughes, D. L.; Reider, P. J.; J. Org. Chem., 2004, 69, 1959–1966. 9. Anderson, N. A.; Campbell, I. B.; Fallon, B. J.; Lynn, S. M.; Macdonald, S. J. F.; Pritchard, J. M.; Procopiou, P. A.; Sollis, S. L.; Thorp, L. R.; Org. Biomol. Chem., 2016, 14, 5992–6009. 10. Dormer, P. G.; Eng, K. K.; Farr, R. N.; Humphrey, G. R.; McWilliams, J. C.; Reider, P. J.; Sager, J. W.; Volante, R. P.; J. Org. Chem., 2003, 68, 467–477. 11. Raghavendra, G. M.; Ramesha, A. B.; Revanna, C. N.; Nandeesh, K. N.; Mantelingu, K.; Rangappa, K. S.; Tetrahedron Lett., 2011, 52, 5571–5574. 12. Hudlicky, M.; J. Fluor. Chem., 1989, 44, 345–359. 13. Hintermann, Läng, F.; Maire, P.; Togni, A.; Eur. J. Inorg. Chem., 2006, 1397–1412. 14. Bordwell, F. G.; Algrim, D. J.; J. Am. Chem. Soc., 1988, 110, 2964–2968. 15. Aycock, D. F.; Org. Process Res. Dev., 2007, 11, 156–159. 16. LC-MS shows peaks consistent with phosphoramidate 13 (m/z = 285 [M+H]+) and a triphosphorylated species (m/z = 557 [M+H]+). 16 17. Subramanyam, V.; Silver, E. H.; Soloway, A. H.; J. Org. Chem., 1976, 41, 1272– 1273. 18. Lehn, J.-M.; Ziessel, R.; Helv. Chim. Acta, 1988, 71, 1511–1516. 19. Zorin, A. V.; Zaynashev, A. T.; Chanysheva, A. R.; Zorin, V. V.; Russ. J. Gen. Chem., 2015, 85, 1382–1385. 20. Barham, J. P.; Coulthard, G.; Kane, R. G.; Delgado, N.; John, M. P.; Murphy, J. A. Angew. Chem. Int. Ed., 2016, 55, 4492-4496.