Full Paper
Figure 4. Structures of the building blocks used.
approach enabled us to secure a shape-diverse fragment set
that met our three objectives.
cyclic sulfamidate 6d, was poorly diastereoselective, presuma-
bly stemming from epimerisation by enolisation: the amino al-
cohol 33 was nonetheless obtained in 25% yield. However, the
intramolecular Mitsunobu reaction of 33 was extremely poor,
and the piperazine 34 was obtained in just 8% yield. Related
syntheses of similarly substituted piperazines have also been
disappointing.[12] This lead-oriented approach was not, there-
fore, ultimately exploited in fragment syntheses. Finally, Ir-cata-
lysed amination[10] was also not ultimately exploited. The re-
quired Ir-catalysed substitutions of allylic carbonates with
amine building blocks were either low yielding or were poorly
regioselective. For example, with azetidine as nucleophile, the
unwanted terminal regioisomer was obtained predominantly.
In these cases, to ensure that a shape-diverse fragment set
was assembled, the targeted fragments were revised (see
above).
The required scaffolds were prepared from pairs of building
blocks (Figure 4 and Scheme 1) in two stages: the pairs of
building blocks were initially combined to yield a cyclisation
precursor which was then cyclised (often via an intermediate)
to yield a scaffold.
Five scaffolds were prepared by cyclisation of a-allyl a-
amino ester precursors (Scheme 1).[11] The ureas 9, 11 and 13
were prepared by reaction between an a-allyl a-amino ester 1
and an isocyanate 2; base-mediated cyclisation then gave the
hydantoins 10, 12 and 14. Alternatively, acylation of the a-allyl
a-amino ester 1d gave the amide 15 which was Boc-depro-
tected and cyclised to give the diketopiperazine 16. Finally, in-
tramolecular Heck reaction of the o-bromobenzyl-substituted
a-amino ester 17, prepared by reductive amination, gave the
benzo-fused azepane 18.
Ten further scaffolds were prepared by using Ugi reactions
(Scheme 1).[13] The required cyclisation precursors (19, 22, 25
and 28) were generally prepared by ring opening of a cyclic
sulfamidate 6 with a propargylic sulfonamide 5. The outcome
of the reaction of these cyclisation precursors with 5 mol%
Au(PPh3)Cl and 5 mol% AgSbF6 in dioxane at 1008C depended
on substitution.[13a] Thus, with the terminal alkynes, 19, 22 and
25, the Boc-protected amine cyclised onto promixal alkyne
carbon to yield the corresponding tetrahydropyrazines 20, 23
and 26. In contrast, the 1,2-disubstituted alkyne 28 underwent
hydration at the remote alkyne carbon to yield the open-chain
b-keto amine derivative 29. The related b-keto amine deriva-
tive 31 was also prepared by conjugate addition of the amine
8 to the enone 7.[13b] After Boc protection, the tetrahydropyra-
zines 20, 23 and 26 and the b-keto amine derivatives 29 and
31 were all competent substrates in Ugi reactions: subsequent
reaction with an isocyanide in ethanol to yielded either substi-
tuted 2-piperazinyl carboxamides (21a–c, 24a–c and 27) or 2-
(1,4-diazepanyl) carboxamides (30a,b or 32). Notably, with the
chiral tetrahydropyrazines 23 and 26, the corresponding Ugi
reaction products were obtained, like related examples,[13b]
with high diastereoselectivity.
The synthesis of the 20 fragments that contributed to the
final shape-diverse fragment set is summarised in Table 1. In
most cases, it was initially necessary to unmask functionality in
the starting scaffold, for example by deprotection or by oxida-
tive alkene cleavage. Four different reactions were exploited in
decoration steps: reductive amination, sulfonylation, acylation,
and urea formation. The synthesis of 13 of the fragments was
influenced by the propensity of trifluoroacetylated Ugi prod-
ucts to undergo based-catalysed hydantoin formation:[13b] to
minimise this undesired reaction pathway, trifluoroacetylated
a-amino esters were deprotected by treatment with sodium
borohydride in methanol.
The shapes and molecular properties of the compounds in
the final fragment set are presented in Figure 3C,D. The frag-
ment set is significantly more three-dimensional than typical
commercial fragment sets,[5a] with only two of the eighty frag-
ments lying close to the rod–disk axis of the PMI plot
(panel C). Fragment sets with distinctive three dimensionality
have previously also already been designed through judicious
selection of commercially available compounds.[5]
The shape diversity of our fragment set is significantly en-
riched by the 20 synthesised fragments which are more three-
dimensional, and are thus complementary to, the 60 pur-
chased fragments (Figure 3C). We note that, whilst the 20 syn-
thesised and 60 purchased fragments have a similar mean
number of rings (m=2.4 and 2.2 respectively), the synthesised
fragments generally have fewer aromatic rings (m=0.6 com-
The synthesis of the piperazine 34 (Scheme 1) emphasised
some of the shortcomings of a lead-oriented synthesis of di-
verse saturated nitrogen heterocycles.[12] First, treatment of the
sulfonamide 5c with sodium hydride, and reaction with the
&
&
Chem. Eur. J. 2019, 25, 1 – 10
4
ꢀ 2019 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ÝÝ These are not the final page numbers!