A modular lead-oriented synthesis of diverse piperazine, 1,4-diazepane and 1,5-diazocane scaffolds

Article abstract of DOI:10.1039/c3ob42512f

Piperazines are found widely in commercially-available compounds and bioactive molecules (including many drugs). However, in the vast majority of these molecules, the piperazine ring is isolated (i.e. not fused to another ring) and is not substituted on any of its carbon atoms. A modular synthetic approach is described in which combinations of cyclic sulfamidate and hydroxy sulfonamide building blocks may be converted into piperazines and related 1,4-diazepine and 1,5-diazocane scaffolds. By variation of the combinations of building blocks used, it was possible to vary the ring size, substitution and configuration of the resulting heterocyclic scaffolds. The approach was exemplified in the synthesis of a range of heterocyclic scaffolds that, on decoration, would target lead-like chemical space. It was demonstrated that lead-like small molecules based on these scaffolds would likely complement those found in large compound collections. This journal is the Partner Organisations 2014.

Full text of DOI:10.1039/c3ob42512f

Organic &
Biomolecular Chemistry
View Article Online
View Journal | View Issue
PAPER
A modular lead-oriented synthesis of diverse
piperazine, 1,4-diazepane and 1,5-diazocane
scaffolds†‡
Cite this: Org. Biomol. Chem., 2014,
12, 2584
Thomas James,a,b Paul MacLellan,a,b George M. Burslem,a,b Iain Simpson,c
J. Andrew Grant,§c Stuart Warriner,a,b Visuvanathar Sridharana and Adam Nelson*a,b
Piperazines are found widely in commercially-available compounds and bioactive molecules (including
many drugs). However, in the vast majority of these molecules, the piperazine ring is isolated (i.e. not
fused to another ring) and is not substituted on any of its carbon atoms. A modular synthetic approach is
described in which combinations of cyclic sulfamidate and hydroxy sulfonamide building blocks may be
converted into piperazines and related 1,4-diazepine and 1,5-diazocane scaffolds. By variation of the
combinations of building blocks used, it was possible to vary the ring size, substitution and configuration
of the resulting heterocyclic scaffolds. The approach was exemplified in the synthesis of a range of
heterocyclic scaffolds that, on decoration, would target lead-like chemical space. It was demonstrated
that lead-like small molecules based on these scaffolds would likely complement those found in large
compound collections.
Received 16th December 2013,
Accepted 27th February 2014
DOI: 10.1039/c3ob42512f
www.rsc.org/obc
synthetic chemistry literature – do not have lead-like pro-
perties.¶ In addition, chemists have historically explored
Introduction
The development of the field of lead-oriented synthesis, in chemical space rather unsystematically.8 The challenge is thus
which diverse compounds with lead-like molecular properties further exacerbated when diversity considerations are also
are prepared, has recently been framed as a major challenge introduced.
for synthetic chemists.1 The molecular properties of clinical
In this paper, a modular approach to diverse heterocyclic
candidates2–5 – particularly molecular size and lipophilicity5 – scaffolds, including those based on piperazines, 1,4-diaze-
are strongly linked to the probability of successful negotiation panes and 1,5-diazocanes, is described. Piperazines are found
of the development process. Optimisation almost inevitably widely in commercially-available compounds (in around 5.4%
leads to increases in both molecular weight and lipophilicity, of compounds in the ZINC databasek) and bioactive molecules
making it essential to control the properties of lead com- (in around 7.7% of compounds in the ChEMBL databasek).
pounds.6,7 The significant challenges associated with prepar- Indeed, 13 of the 200 best-selling small molecule drugs in
ing diverse lead-like small molecules – i.e. molecules that 2012 contain a piperazine ring (for examples, see Fig. 1).9
would be good starting points for lead optimisation – have However, in the vast majority of these molecules, the piper-
recently been articulated.1
azine ring is isolated (i.e. not fused to another ring) and is not
Sourcing large numbers of lead-like small molecules is a substituted on any of its carbon atoms.10 Significant chemical
major challenge in maintaining large, high quality screening space that is closely related to that known to be biologically-
collections.1 The vast majority of commercially-available relevant therefore remains underexplored.
screening compounds – as well as compounds reported in the
An overview of the proposed modular synthetic approach is
shown in Scheme 1. It was proposed that bi-connective11 build-
ing blocks would be prepared from readily available amino
aSchool of Chemistry, University of Leeds, Leeds, LS2 9JT, UK.
E-mail: a.s.nelson@leeds.ac.uk; Fax: +44 (0) 113 6401; Tel: +44 (0) 113 6502
bAstbury Centre for Structural Molecular Biology, University of Leeds, Leeds, LS2 9JT,
¶In a recent survey, the vast majority of 4.9 M commercially-available com-
pounds (>99%) and compounds reported in the synthetic chemistry literature
(∼98%) failed at least one of the following filters: −1 < c log P < 3; 14 ≤ heavy
atoms ≤ 26; number of aromatic rings, nAr, ≤3; absence of specific chemically-
reactive and redox-active groups (ref. 1).
UK
cAstraZeneca, Mereside, Alderley Park, Macclesfield, Cheshire SK10 4TG, UK
†Dedicated to J. Andrew Grant who initiated this project.
‡Electronic supplementary information (ESI) available. See DOI:
10.1039/c3ob42512f
kDetermined by substructure searches of the ZINC and ChEMBL databases
using Pipeline Pilot version 8.5 (Accelrys).
§Deceased.
2584 | Org. Biomol. Chem., 2014, 12, 2584–2591
This journal is © The Royal Society of Chemistry 2014

View Article Online
Organic & Biomolecular Chemistry
Paper
Fig. 1 Examples of best-selling drugs that contain a piperazine ring
system.
alcohols (panel A). Thus, ring-opening of a cyclic sulfamidate12
(e.g. 2) with a hydroxy sulfonamide (e.g. 1) would yield an inter-
mediate (e.g. 3) in which the functionality needed for cyclisa-
tion had been revealed. Finally, cyclisation would yield a
specific heterocyclic scaffold (e.g. 4). It was envisaged that, by
varying the building blocks used, it would be possible to vary
both the size of the new heterocyclic ring, and the substituents
on specific carbon atoms. The overall synthesis – from the bi-
connective building blocks (e.g. 1 and 2) to the product
scaffold (e.g. 4) – may be classified as a bi/bi process11 or as an
ambiphile pairing13 process.
Scheme 1 Overview of the proposed modular approach to diverse
lead-like heterocycles. The provenance of atoms from building blocks is
indicated using colour, and new bonds are shown in black. Panel A:
Combination of two bi-connective building blocks – for example the
hydroxy sulfonamide 1 and the cyclic sulfamidate 2 – would yield the
piperazine 4. Panel B: Examples of additional scaffolds that might also
be prepared using the approach.
Results and discussion
Development of a modular synthesis of diverse heterocyclic
scaffolds
Initially, a modular synthesis of diverse heterocyclic scaffolds
was developed. Specifically, it was decided to investigate the
effect of both the ring size and substitution of the product
scaffold. A range of cyclic sulfamidates was prepared by reduc-
tive amination of the corresponding amino alcohol (e.g. 5→6),
followed by cyclic sulfamidite formation and ruthenium-cata-
lysed oxidation (e.g. →7a) (Scheme 2). In addition, a range of
hydroxy sulfonamides was prepared by sulfonylation of the
corresponding hydroxy alcohols (Scheme 3).
Initial work focused on the combination of the hydroxy sul-
fonamide 8a and the cyclic sulfamidate 7a (Scheme 4). It was
decided that the ideal method would minimise the need for
purification of the intermediate 9 and the product 10a using
silica gel column chromatography. The hydroxy sulfonamide
Scheme 2 Synthesis of cyclic sulfamidate building blocks 7. PMB:
p-methoxybenzyl. Panel A: Synthesis of the building block 7a. Panel B:
Additional building blocks prepared using the same approach. aYields for
8a was treated with sodium hydride in DMF, reacted with the reductive amination and cyclic sulfamidate formation respectively.
This journal is © The Royal Society of Chemistry 2014
Org. Biomol. Chem., 2014, 12, 2584–2591 | 2585

View Article Online
Paper
Organic & Biomolecular Chemistry
Table 1 Investigation into the effect of product ring size and substi-
tution on the efficiency of heterocycle synthesis
Entry
1
Building blocks
Producta
Yieldb/%
69
7a, 8ac
2
3
7b, 8a
7b, 8b
40d
60d
Scheme 3 Synthesis of hydroxy sulfonamide building blocks 8. Ns:
o-nitrophenylsulfonyl. Panel A: Synthesis of the building block 8a. Panel
B: Additional building blocks prepared using the same approach.
4
7a, 8e
68
5
6
7
7e, 8a
7e, 8b
7c, 8e
72
44
0e
Scheme 4 Modular synthesis of the piperazine 10a.
cyclic sulfamidate 7a at room temperature and finally treated
with aqueous acid; after work-up, and strong ion exchange
chromatography (SCX), the ring-opened intermediate 9 was
obtained in 74% yield. The successful purification of 9 using
SCX stems from the basicity of 9 (but not 7a or 8a). Treatment of
9 with triphenylphosphine (1.4 eq.) and DEAD (1.4 eq.) triggered
cyclisation14 to give the piperazine 10a; here, SCX was used suc-
cessfully to facilitate the purification of the (basic) product 10a
from the by-products of the reaction. The use of hydrogen-
borrowing chemistry to promote the cyclisation of intermediates
such as 9 was also investigated, but without success.
8
7d, 8c
41 (33f)
The scope and limitations of the modular synthetic
approach were investigated (Table 1). Initially, the effect of the
ring-size of the product scaffold was investigated (compare
entries 1–3). Remarkably, the yields of the 1,4-diazepane 10b
(entry 2) and the 1,5-diazocane 10c (entry 3) were only margin-
ally lower than that of the piperazine 10a (entry 1). It should
be noted, however, that ring opening of the six-membered
cyclic sulfamidate 7b required heating (70 °C).15
The effect of substitution on the yield of the product
scaffold was investigated using the valinol-derived building
blocks 7e and 8e (Table 1; entries 4–6). The effect of a single
iso-propyl substituent on the yield of the product scaffold was
small, both in the synthesis of piperazines (compare entries 4
and 5 with entry 1) and 1,4-diazepanes 10f (compare entry 6
with entry 4). However, with the hindered hydroxy sulfonamide
8e, no product was obtained with either 4- or 5-substituted
cyclic sulfamidates (7c, 7d or 7e) (entry 7). Here, the specific
combination of a hindered electrophile and a hindered nucleo-
phile prevented efficient ring-opening of the cyclic sulfamidate.
Finally, the possibility of matched-mismatched16 effects
was investigated in the synthesis of diastereomeric piperazines
9
7d, 8d
7e, 8d
36 (29f)
10
40
2586 | Org. Biomol. Chem., 2014, 12, 2584–2591
This journal is © The Royal Society of Chemistry 2014

View Article Online
Organic & Biomolecular Chemistry
Paper
Table 1 (Contd.)
Entry
11
Building blocks
Producta
Yieldb/%
40
7e, 8c
a See Scheme 4 for the method used. b Yield of product over 2 steps.
c See Scheme 4 for this specific combination of building blocks.
d Opening of cyclic sulfamidate 7b were performed at 70 °C. e The
required product was also not obtained when the building block 8e
was combined with either 7d or 7e. f Yield of the enantiomeric product
prepared from the enantiomeric building blocks.
(Table 1; entries 8–11). Remarkably, the yields of diastereo-
meric pairs of piperazines (10h/10i and 10j/10k; compare
entries 8–9 and entries 10–11) were broadly similar. In each of
these cases, a single diastereomeric product was obtained,
demonstrating that both the ring-opening and cyclisation
steps were stereospecific.
Exemplification in the synthesis of lead-like scaffolds
The modular synthetic approach was exemplified through the
synthesis of a range of molecular scaffolds. The required build-
ing blocks were selected carefully such that diverse scaffolds
might be prepared which, on derivatisation, would yield mole-
cules with lead-like molecular properties. Crucially, different
combinations of the building blocks would yield scaffolds (a)
in which the size of the new heterocycle was varied; (b) with
contrasting carbon-based substituents; and, (c) in which, in
some cases, the new heterocycle was fused to another ring.
The required building blocks were prepared from readily-
available starting materials (Schemes 5 and 6). The amino
alcohols 12, 15 and 17 were prepared respectively by reduction
of the amino acid 11 (→12), by substitution of the bromo
alcohol 13 (→15), and by ring-opening17 of the epoxide 16
(→17) (Scheme 5). In each case, the amino alcohols were con-
verted into the corresponding cyclic sulfamidites which were
then oxidised to give the required cyclic sulfamidates 7. The
hydroxy sulfonamide building blocks 8f–8i were prepared by
sulfonylation of the corresponding amino alcohols (Scheme 6).
Pairs of building blocks were combined to yield a range of
lead-like molecular scaffolds (see Scheme 7 and Table 2). In
each case, a hydroxy sulfonamide building block 8 was treated
with sodium hydride in DMF, reacted with a cyclic sulfamidate
7 and, finally, treated with aqueous acid; after work-up, SCX
yielded the crude ring-opened intermediate. Subsequently,
the ring-opened intermediates were treated with triphenyl-
phosphine (1.2 eq.) and DEAD (1.1 eq.). A wide range of mole-
cular scaffolds was obtained that display significant structural
diversity. The new heterocyclic rings were six-, seven- or eight-
Scheme 5 Synthesis of cyclic sulfamidate building blocks.
The efficiency of the modular approach depended on the
specific building blocks used (Table 2). Diverse molecular
scaffolds were prepared efficiently from the 5-substituted cyclic
sulfamidate 7f: 2-substituted and 2,7-disubstituted 1,4-di-
azepanes (10l and 10o respectively; entries 1 and 4), and 2,5-
and 2,6-disubstituted piperazines (10m and 10n respectively;
entries 2–3). However, the benzo-fused 1,4-diazepane 10p was
obtained in low yield (entry 5). With the 6-membered cyclic
sulfamidate building block 7g, low yields of the 1,5-diazocanes
10q and 10r were obtained (entries 6–7). With the 4-substi-
tuted cyclic sulfamidate 7h, the success of the approach
depended on the specific hydroxy sulfonamide building block
used; as had previously been observed with a 4-substituted
cyclic sulfamidate (see entry 7, Table 1), the approach was less
successful when more hindered hydroxy sulfonamides were
used (compare entries 8–10, Table 2).
Assessment of the value of the molecular scaffolds
membered; and substituents were introduced on contrasting A virtual library of compounds was enumerated from combi-
carbons on the new ring. In addition, in some cases, the new nations of the deprotected scaffolds (derived from 10l–u) and a
heterocyclic ring was benzo-fused.
range of exemplar medicinal chemistry reagents for amine
This journal is © The Royal Society of Chemistry 2014
Org. Biomol. Chem., 2014, 12, 2584–2591 | 2587

View Article Online
Paper
Organic & Biomolecular Chemistry
Table 2 Synthesis of diverse lead-like molecular scaffolds
Entry
1
Building blocks
Producta
Yieldb/%
87
7f, 8b′c
2
3
4
7f, 8f
7f, 8g
7f, 8h
44
55
80
5
6
7f, 8i
11
13
Scheme 6 Synthesis of hydroxy sulfonamide building blocks. Ns’:
p-nitrophenylsulfonyl.
7g, 8b′
7
8
7g, 8i
14
38
7h, 8h
Scheme 7 Modular synthesis of the lead-like scaffold 10l. Ns: o-nitro-
phenylsulfonyl. Ns’: p-nitrophenylsulfonyl.
decoration (see Scheme 8 and ESI†). The resulting library com-
prised 977 likely synthetically-accessible small molecules. The
value of the scaffolds was assessed by analysis of this virtual
library.
2588 | Org. Biomol. Chem., 2014, 12, 2584–2591
This journal is © The Royal Society of Chemistry 2014

View Article Online
Organic & Biomolecular Chemistry
Paper
Table 2 (Contd.)
Entry
9
Building blocks
Producta
Yieldb/%
8
7h, 8f
10
7h, 8i
28
a See Scheme 7 for the method used. b Yield of product over 2 steps.
c See Scheme 7 for this specific combination of building blocks. Ns′:
p-nitrophenylsulfonyl.
Fig. 2 Lead-likeness of a virtual library of compounds derived from the
scaffolds 10l–u (see Scheme 8). Data for compounds with lead-like pro-
perties is highlighted in green. Panel A: Distribution of the number of
heavy atoms (nHA) and the lipophilicity (A log P). Panel B: Distribution of
the number of aromatic rings (nAr).
(nAr) were determined (Fig. 2). With these properties deter-
mined, the lead-likeness of the compounds was assessed.
Remarkably, this analysis showed that 338 (∼34%) of the com-
pounds in the virtual library had lead-like molecular properties
(14 ≤ nHA ≤ 26; −1 ≤ A log P ≤ 3; nAr ≤ 3).
Second, the shape diversity of the 338 lead-like compounds
from the virtual library was assessed using two alternative
methods. In order to compare these compounds with com-
pounds of
a similar size, ∼100 000 compounds with
14–26 heavy atoms were randomly selected from the ZINC18
database. To allow shape analysis, low-energy conformations
were determined for all compounds using CORINA.
Initially, the three principal moments of inertia (PMI) were
determined each compound, which were then used to calcu-
late two normalised PMI values.19 These data are presented in
the form of a triangular plot (Fig. 3) in which the vertices are
defined by rod, disk and spherical shapes. The analysis
showed that the virtual library derived from the scaffolds 10l–u
was significantly more three-dimensional than most of the
Scheme 8 Enumeration of a virtual library of likely synthetically-acces-
sible small molecules based on the molecular scaffolds 10l–u.
First, the lead-likeness of the members of the virtual library
was assessed. For each compound, the number of heavy atoms
(nHA), lipophilicity (A log P) and number of aromatic rings
This journal is © The Royal Society of Chemistry 2014
Org. Biomol. Chem., 2014, 12, 2584–2591 | 2589

View Article Online
Paper
Organic & Biomolecular Chemistry
Fig. 3 Principal moment of inertia analysis of small molecules. The two
normalised PMI values are shown, both for the 338 lead-like compounds
derived from the scaffolds 10l–u (see Scheme 8) (blue, enlarged for
clarity) and ∼100 000 compounds derived from the ZINC database
(green).
commercially-available compounds selected from the ZINC
database.
Secondly, a more sophisticated shape analysis was under-
taken using the ROCS (Rapid Overlay of Chemical Shapes)
tool. Here, each compound was compared with a set of refer-
ence shapes. The set of reference shapes, generated using an
established procedure,20 describes the shape diversity of both
the 338 lead-like compounds within the virtual library and the Fig. 4 Shape-diversity of lead-like small molecules. The proportion of
compounds that have similar shape to each reference shape is shown.
The reference shapes are ordered by the proportion of the ∼100 000
compounds from the ZINC database that map onto each shape. Panel A:
Mapping of ∼100 000 compounds from the ZINC database. Panel B:
∼100 000 compounds within the ZINC database.** The shape
of each compound was then compared to that of each refer-
ence shape and, using a shape Tanimoto cut-off of 0.7, the
number of compounds that match each reference shape
recorded (Fig. 4). Crucially, the 338 lead-like compounds
based on the scaffolds 10l–u target a large number of the refer-
ence shapes, including many that are targeted poorly by the
∼100 000 compounds from the ZINC database (compare
panels A and B). The shapes of compounds based on the
scaffolds 10l–u are therefore diverse, and complement those of
commercially-available compounds of similar size.
Mapping of 338 lead-like compounds derived from the scaffolds 10l–u.
carefully designed such that, after decoration, many small
molecules with lead-like molecular properties could be
obtained. A virtual library of small molecules was enumerated
from combinations of the deprotected scaffolds and a range of
exemplar medicinal chemistry reagents for amine decoration.
Remarkably, ∼34% of the resulting virtual compounds had
lead-like molecular properties. These lead-like compounds
were shown to complement the three-dimensional shape of
commercially-available compounds of similar size. Thus, the
modular synthetic approach was exemplified in the synthesis
of lead-like molecular scaffolds. Lead-like small molecules
based on these scaffolds would likely complement those found
in existing small molecule screening collections.
Conclusions
In summary, an approach to the synthesis of diverse hetero-
cyclic scaffolds has been developed and exemplified. The
modular approach involved ring-opening of cyclic sulfamidates
with hydroxy sulfonamides, followed by cyclisation. Crucially,
by variation of the combinations of building blocks used, it
was possible to vary the ring size, substitution and configur-
ation of the resulting heterocyclic scaffold.
The synthetic approach was exemplified in the synthesis of
a range of heterocyclic scaffolds (10l–u). The scaffolds were
Acknowledgements
We thank AstraZeneca and EPSRC for funding, and OpenEye,
Accelrys, Molecular Networks and Dotmatics for academic soft-
ware licenses.
**Each pair of reference shapes has a distinct shape with shape Tanimoto simi-
larity score <0.8 (calculated using ROCS).
2590 | Org. Biomol. Chem., 2014, 12, 2584–2591
This journal is © The Royal Society of Chemistry 2014

View Article Online
Organic & Biomolecular Chemistry
Paper
E. M. Carreira, Angew. Chem., Int. Ed., 2013, 52,
11908.
Notes and references
11 P. MacLellan and A. Nelson, Chem. Commun., 2013, 2383.
12 J. F. Bower, J. Rujirawanich and T. Gallagher, Org. Biomol.
Chem., 2010, 8, 1505.
13 A. Rolfe, T. B. Samarakoon and P. R. Hanson, Org. Lett.,
2010, 12, 1216.
14 For examples of cyclisations under Mitsunobu-conditions
see: (a) N. Saha, T. Biswass and S. K. Chattopadhyay, Org.
Lett., 2011, 13, 5128; (b) M. Chaumontet, R. Piccardi and
O. Baudoin, Angew. Chem., Int. Ed., 2009, 48, 179.
15 C. G. Espino, P. M. When, J. Chow and J. Du Bois, J. Am.
Chem. Soc., 2001, 123, 6935.
16 (a) S. Masamune, W. Choy, J. S. Petersen and L. R. Sita,
Angew. Chem., Int. Ed., 1985, 24, 1; (b) C. H. Heathcock and
C. T. J. White, J. Am. Chem. Soc., 1979, 101, 7076.
17 M. Breuning, M. Steiner, C. Mehler, A. Paasche and
D. Hein, J. Org. Chem., 2009, 74, 1407.
1 A. Nadin, C. Hattotuwagama and I. Churcher, Angew.
Chem., Int. Ed., 2012, 51, 1114.
2 M. C. Wenlock, R. P. Austin, P. Barton, A. M. Davis and
P. D. Leeson, J. Med. Chem., 2003, 46, 1250.
3 P. D. Leeson and B. Springthorpe, Nat. Rev. Drug Discovery,
2007, 6, 881.
4 T. J. Ritchie and S. J. F. Macdonald, Drug Discovery Today,
2009, 14, 1011.
5 F. Lovering, J. Bikker and C. Humblet, J. Med. Chem., 2009,
52, 6752.
6 T. I. Oprea, A. M. Davis, S. J. Teague and P. D. Leeson,
J. Chem. Inf. Comput. Sci., 2001, 41, 1308.
7 G. M. Keserü and G. M. Makara, Nat. Rev. Drug Discovery,
2009, 8, 203.
8 A. H. Lipkus, Q. Yuan, K. A. Lucas, S. A. Funk, W. F. Bartelt,
R. J. Schenck and A. J. Trippe, J. Org. Chem., 2008, 73, 4443.
9 The structures of the top 200 selling drugs can be down-
loaded from: http://cbc.arizona.edu/njardarson/group;
N. A. McGrath, M. Brichacek and J. T. Njardarson, J. Chem.
Educ., 2010, 87, 1348.
18 J. J. Irwin, T. Sterling, M. M. Mysinger, E. S. Bolstad and
R. G. Coleman, J. Chem. Inf. Model., 2012, 52, 1757.
19 W. H. Sauer and M. K. Schwarz, J. Chem. Inf. Comput. Sci.,
2003, 43, 987.
20 J. A. Haigh, B. T. Pickup, J. A. Grant and A. Nicholls,
J. Chem. Inf. Model., 2005, 45, 673.
10 For a recent approach to piperazines with substituted
carbon atoms, see: S. A. Ruider, S. Müller and
This journal is © The Royal Society of Chemistry 2014
Org. Biomol. Chem., 2014, 12, 2584–2591 | 2591