Rapid access to compound libraries through flow technology: Fully automated synthesis of a 3-aminoindolizine library via orthogonal diversification

Article abstract of DOI:10.1021/co300094n

A novel methodology for the synthesis of druglike heterocycle libraries has been developed through the use of flow reactor technology. The strategy employs orthogonal modification of a heterocyclic core, which is generated in situ, and was used to construct both a 25-membered library of druglike 3-aminoindolizines, and selected examples of a 100-member virtual library. This general protocol allows a broad range of acylation, alkylation and sulfonamidation reactions to be performed in conjunction with a tandem Sonogashira coupling/cycloisomerization sequence. All three synthetic steps were conducted under full automation in the flow reactor, with no handling or isolation of intermediates, to afford the desired products in good yields. This fully automated, multistep flow approach opens the way to highly efficient generation of druglike heterocyclic systems as part of a lead discovery strategy or within a lead optimization program.

Full text of DOI:10.1021/co300094n

Research Article
pubs.acs.org/acscombsci
Rapid Access to Compound Libraries through Flow Technology: Fully
Automated Synthesis of a 3‑Aminoindolizine Library via Orthogonal
Diversification
Paul P. Lange and Keith James*
Department of Chemistry, The Scripps Research Institute, 10550 North Torrey Pines Road, La Jolla, California 92037, United States
S
* Supporting Information
ABSTRACT: A novel methodology for the synthesis of
druglike heterocycle libraries has been developed through the
use of flow reactor technology. The strategy employs
orthogonal modification of a heterocyclic core, which is
generated in situ, and was used to construct both a 25-
membered library of druglike 3-aminoindolizines, and selected
examples of a 100-member virtual library. This general
protocol allows a broad range of acylation, alkylation and
sulfonamidation reactions to be performed in conjunction with a tandem Sonogashira coupling/cycloisomerization sequence. All
three synthetic steps were conducted under full automation in the flow reactor, with no handling or isolation of intermediates, to
afford the desired products in good yields. This fully automated, multistep flow approach opens the way to highly efficient
generation of druglike heterocyclic systems as part of a lead discovery strategy or within a lead optimization program.
KEYWORDS: flow chemistry, library synthesis, indolizine, conjure, cross-coupling, enabling technology
capable of performing multiple modifications on one molecule
would be a significant advance for the field of combinatorial
chemistry, since libraries with increased size and diversity could
be generated efficiently, while benefitting from the advantages
of flow technology.8 To date, only two reports have appeared in
the recent literature that describe a two-step modification to
create library compounds in flow. A “catch and release”
protocol was developed by researchers at GSK for the
generation of a 48-membered library of secondary sulfonamides
via deprotection and subsequent alkylation of the primary
sulfonamide.9 However, the protocol only allows two
modifications of the substrate at one reactive site. Ulven et al.
have shown that a series of reactors and scavenging cartridges
can be used to create a small library of receptor ligands via
subsequent modification of mono-Cbz-protected piperazine.10
However, the methodology is limited to the formation of ureas
generated in the first step, followed by Cbz-deprotection and
N-alkylation in the second step. Yields for the final products are
generally low. Thus, a true protocol for the orthogonal
diversification of substrates to afford compound libraries with
substantial size and diversity has yet to be developed.
INTRODUCTION
■
The continued growth of unexplored molecular targets,
coupled with widely available high-throughput screening
(HTS) technology, provides continued impetus for the efficient
generation of novel, druglike molecular structures that could
potentially serve as starting points for drug-discovery
programs.1 To meet this demand for access to collections of
novel molecules that already possess druglike structures and
properties, new and efficient strategies for generating
compound libraries, ideally in an automated fashion, are
needed.
In this regard, flow chemistry in particular has received
increased attention because of its intrinsic benefits over
traditional batch techniques; these include excellent heat and
mass transfer for better reaction control, safe operation at high
temperatures/pressures, and when employing toxic or noxious
reaction components, as well as automated reaction execution,
purification and product isolation. Thus, increased efforts have
been undertaken to incorporate flow technology into
combinatorial chemistry and library synthesis.2 Recent seminal
reports on flow-based library synthesis include the combination
of batch and flow techniques,3 fully automated flow reactor
systems,4 stepwise synthesis of compounds under continuous5
and segmented flow conditions,6 and parallel synthesis of
different compounds in multiple flow channels.7 Although
substantial improvement has been achieved through the
incorporation of flow technology into library synthesis, current
developments have primarily focused on one-step reactions in
flow to afford the desired library members, thus limiting the
potential size and diversity of the library. A flow methodology
In the course of our program to develop novel synthetic
strategies for the construction of druglike molecules in batch11
and flow,12 we have recently reported a flow-based protocol for
the expedient synthesis of 3-aminoindolizines 3 via a tandem
Received: August 14, 2012
Revised: September 3, 2012
Published: September 6, 2012
© 2012 American Chemical Society
570
dx.doi.org/10.1021/co300094n | ACS Comb. Sci. 2012, 14, 570−578

ACS Combinatorial Science
Research Article
Sonogashira coupling/cycloisomerization reaction (Scheme 1,
In that report, we described a flow-based protocol that offers
distinct advantages in comparison to analogous batch
procedures,14 and highlights a substantially increased substrate
scope, increased speed of reaction and relative ease and safety
of scale-up. Moreover, we alluded to the potential of this new
methodology, in conjunction with the flow reactor system
employed, to access libraries of druglike molecules via
orthogonal diversification of a suitable heterocyclic core
(Scheme 1, (II)). On the basis of these initial findings, we
herein present the development of the first fully automated,
flow-based synthesis of a diverse 25-membered library of
druglike 3-aminoindolizines. By employing the Conjure flow
reactor system,15 a protocol has been developed that allows the
in situ construction and orthogonal diversification of an
aminoindolizine in a multistep sequence to afford the desired
library compounds in good yields. All modifications are
performed successively within the flow reactor, thereby
avoiding the need to handle or isolate of any intermediates.
(I)).13
Scheme 1. Indolizine and Aza-indolizine Synthesis in Flow
Scheme 2. Schematic (a) and Actual (b) Set-up of the Conjure Flow System
571
dx.doi.org/10.1021/co300094n | ACS Comb. Sci. 2012, 14, 570−578

ACS Combinatorial Science
Research Article
Scheme 3. Library Design Strategy
the second reaction step. The reaction segment is detected by
UV upon exiting the reactor and subsequently collected.
With these unique features and design, the Conjure flow
synthesis system is ideally suited for our development of a
methodology for library synthesis of druglike heterocycles. In
particular, we chose to exploit the fact that two-step sequences
can be conducted automatically and that our previously
described methodology could be incorporated as the second
step in this approach.
RESULTS AND DISCUSSION
■
We began our investigations employing the commercially
available Accendo Conjure flow reactor system, in light of its
ability both to undertake rapid reaction parameter screening
and the automatic execution of two-step reaction sequences.
The Flow Reactor. The Conjure flow system consists of
three units, the pump module, reactor module and fluid prep
module, which will all be explained in further detail, to
emphasize our choice to employ this particular flow device. All
three modules can be individually controlled and settings
adjusted by the user with the Conjure Software. The whole
system is connected to a back pressure regulator (bpr), an
Agilent UV-detector and a Gilson fraction collector (Scheme 2,
UV-detector and fraction collector not shown in 2b).
Reactions are conducted in individual segments, which allow
highly time-efficient screening of reaction parameters with
minimal quantities of reagents, excellent reproducibility for
scale-up and rapid generation of our 25-membered library.
The Pump Module. The pump module is directly
connected to both the reactor and fluid prep module and
allows the selective aspiration of three different system solvents
(pump speeds 0.1−3.5 mL·min−1).
The Reactor Module. The reactor coil used in the Conjure
flow reactor is constructed from Hastelloy tubing (0.75 mm i.
d., 4.5 m length, 2.0 mL i. v.) and enclosed in a reactor diskette
(see ESI for details). Within the heating unit of the reactor
module, the diskette can be heated to a maximum temperature
of 300 °C. The inlet of the reactor is connected to a valve
directing flow of the system solvent either directly from the
solvent reservoir or via the fluid prep module. The outlet of the
reactor is connected to a back pressure regulator immediately
followed by the UV-detector and fraction collector. As soon as
a reaction segment is detected by UV, it is automatically
collected in an individual vial.
The Fluid Prep Module. The most intricate and unique
part of the Conjure reactor system is the fluid prep module,
which contains a rotating 40-vial carousel, an incubation
chamber and four syringe pumps to aspirate up to four different
reagent stock solutions (25−300 μL each, see ESI for details).
Segments are prepared in a user-predetermined order and
composition from the respective source vials containing the
reagent stock solutions. Each segment is preceded and followed
by an immiscible liquid spacer that prevents segment diffusion
and reaction trailing.16 The segment preparation is fully
automated once the stock solutions are in place and the
carousel is loaded. Segments can be prepared from up to four
different reagent solutions simultaneously, drawing from a
contingent of 36 vials held on the carousel.17 Alternatively, as in
this project and exemplified in our initial protocol,11 a first step
reaction segment can be prepared and positioned in one of the
sample loops housed in the incubation chamber.18 The mixture
can be heated up to (incubated at) 100 °C for the desired
reaction time and is then reaspirated, mixed with up to three
different reagents19 and subjected to the flow stream to process
Library Design. Our strategy for constructing the library
involved introduction of three elements of diversity, as
illustrated in Scheme 3. The first point of diversification is in
the selection of a heteroaryl bromide core I, which is capable of
both undergoing a derivatization reaction via a free amino
substituent, and a Sonogashira reaction (followed by a
cycliosomerization to generate a 3-aminoindolizine) via the
bromo substituent. The second point of diversification is
derivatization of the pendant amino substituent via, for
example, an acylation reaction, to yield intermediate II. The
final point of diversification is through the incorporation of a
range of substituted propargyl amines in the Sonogashira
coupling step to yield the 3-aminoindolizine product III.
We chose to exemplify the approach by initially employing
one heteroaryl bromide core and a 5 × 5 matrix of amine
substituents and propargyl amines, as the basis of a 25-
membered library. We also went on to illustrate the
applicability of the approach to three other heteroaryl core
fragments, as well as a further variant of the protocol based
upon derivatization of the propargylamine component. Thus,
we have demonstrated that a corresponding virtual library of
100 molecules (4 × 5 × 5) could potentially be accessed using
our approach.
Library Protocol Optimization. We chose heterocyclic
building block 10, which is easily synthesized in gram-quantities
from commercially available starting materials (Scheme 4), as
our initial core fragment to optimize the multistep flow
protocol.
Scheme 4. Straightforward Synthesis of the Common
Building Block 10
We began our investigations with the optimization of the
reaction conditions for the first step modification on the
piperazine substituent. We reasoned that more thoroughly
optimized conditions would allow a multitude of viable
transformations under flow conditions and lead to much better
yields for the final library compounds. We focused on two
major issues regarding the first step: (1) The conditions
required to effect a rapid and versatile N-modification must not
inhibit the follow-up cross-coupling/cycloisomerization se-
572
dx.doi.org/10.1021/co300094n | ACS Comb. Sci. 2012, 14, 570−578

ACS Combinatorial Science
Research Article
quence and (2) all desired N-modifications must be amenable
to flow chemistry. At first, we analyzed whether any by-
products inevitably formed during a first step N-acylation
interfered with the follow-up indolizine synthesis. The coupling
of acylated heteroaryl bromide 11{1} and propargyl amine
2{1} served as our model reaction for this analysis (Table 1).
salt 10, along with the required amounts of base and acetic
anhydride. The prepared mixture was positioned in the
incubation chamber at 25 °C for 10 min, then quenched with
MeOH (200 μL) upon reaspiration, and passed through the
reactor coil at room temperature. LC-MS analysis of the
collected segment showed full conversion of the amine to the
amide without any notable side reactions in either DMF or
MeCN solvent. For our further investigations, we chose to
conduct all further N-modifications in MeCN because of its
more environmentally benign properties, thermal stability in
comparison to DMF, and higher volatility, thus simplifying
purification and isolation of the final product.
Table 1. Analysis of Potential Factors Inhibiting the
a
Indolizine Synthesis
Having found conditions for the first reaction step that met
all the aforementioned criteria, we combined this first step with
the previously developed indolizine flow protocol and
examined a range of base, solvent and reaction time variants
in order to determine the optimal conditions for the multistep
protocol (Table 2).
b
c
entry
additive
12{1,1} (%)
1
2
3
4
5
6
7
8
87
85
84
0
DMF
MeCN
CHCl3
Table 2. Optimization of the Two-Step Protocol for Library
a
Synthesis
NEt3·HOAc/DMF (1.0)
NEt3·HCl/MeCN (1.0)
NEt3·TFA/DMF (1.0)
Ac2O/DMF (0.2)
83
76
84
87
a
Reaction conditions: Accendo Conjure Flow Reactor, Hastelloy
tubing (0.75 mm i. d., 2.0 mL internal vol.), 400 μL segment: 0.05
mmol heteroaryl bromide 11{1}, 0.10 mmol alkyne 2{1}, 3 mol %
PdCl2(PPh3)2, 10 mol % CuI. 200 μL of solution added to the
b
c
b
entry
base step 2
solvent step 2 time min 12{1,1} (%)
segment. Number in parentheses corresponds to equivalents of
additive. Percent UV from LC-MS analysis.
1
2
3
4
5
6
7
NEt3 (1.5), DBU (3.0)
DMSO
5
7
52
55
62
75
92
c
10
5
NEt3 (1.5), DBU (5.0)
DBU (5.0)
Indolizine 12{1,1} was formed in good yield when using the
developed protocol in the absence of any additives (entry 1).
The addition of DMF or MeCN solvent had no negative effect
on the outcome of the reaction (entries 2 and 3). However,
chloroform, a standard solvent for acylation reactions,
completely inhibited the catalyst system and only heteroaryl
bromide 11{1} along with a complex reaction mixture was
observed (entry 4). As a result, we ruled out conducting the
first step modification in this solvent. We also investigated the
effect of stoichiometric amounts of ammonium salts that would
be formed in the first step when using a typical base, such as
triethylamine. Only triethylammonium chloride, which was
added in MeCN solution due to its insolubility in DMF, had a
slightly detrimental influence on the yield (entries 5−7).
Moreover, the chloride salt is a highly crystalline solid that was
found to easily precipitate from concentrated reaction solutions
and subsequently block the fluidics of the reactor system. Thus,
we investigated the effect of substoichiometric amounts of
acetic anhydride, instead of acetyl chloride, as the potential
acylating agent, and were delighted to find no diminution in
yield (entry 8).
With these encouraging results in hand, we proceeded with
the optimization of the amide formation. In preliminary batch
experiments, we found that in the presence of a slight excess of
acetic anhydride (1.2 equiv.) and triethylamine (3.0 equiv.)
quantitative acylation of TFA salt 10 was observed at room
temperature within only 10 min. DMF and MeCN were equally
suitable solvents for this reaction, and no precipitation was
observed in either case (see ESI for details). We then
transferred this approach into the flow reactor by preparing a
segment (200 μL, 0.25 M, MeCN or DMF) containing TFA
7
93
d
MeCN
92 (59)
a
Reaction conditions: Accendo Conjure Flow Reactor; step 1 200 μL
segment: 0.05 mmol 10, 0.06 mmol Ac2O 5{1}, 0.15 mmol NEt3, PFA
tubing (0.75 mm i. d., 0.7 mL internal vol.); step 2 400 μL segment:
0.05 mmol 11{1}, 0.10 mmol alkyne 2{1}, 3 mol % PdCl2(PPh3)2, 10
mol % CuI, Hastelloy tubing (0.75 mm i. d., 2.0 mL internal vol.).
Number in parentheses corresponds to equivalents of base. Percent
UV from LC-MS analysis. Isolated yield.
b
c
d
We were delighted to find that our first attempt to combine
both steps did indeed lead to the formation of indolizine
12{1,1} in 52% yield (entry 1). In this case, we observed full
conversion of the aryl bromide intermediate 11{1} to a 1:1
mixture of the desired indolizine and Sonogashira coupling
product. Further optimization showed that an increased
residence time did not affect this ratio (entries 2 and 3).
However, increasing the amount of DBU base from three to
five equivalents showed an improved yield of 75% for the
desired indolizine (entry 4). Prolonging the residence time
under these conditions had a profound effect on the
intramolecular cycloisomerization and a 92% yield was obtained
(entry 5). In the presence of a larger excess of DBU base we
found that triethylamine, previously added to solubilize the
catalyst precursors, was no longer required, and identical results
were obtained in its absence (entry 6). Furthermore, due to the
nondetrimental presence of MeCN in the reaction segment
from the incubation step, we explored conducting the entire
reaction sequence in this preferred solvent and were pleased to
find that similar results were obtained when no DMSO was
573
dx.doi.org/10.1021/co300094n | ACS Comb. Sci. 2012, 14, 570−578

ACS Combinatorial Science
Research Article
a
Table 3. Synthesized 25-Membered 3-Aminiondolizine Library
a
Reaction conditions: Accendo Conjure Flow Reactor; step 1 200 μL segment: 0.05 mmol 10, 0.06 mmol 2, 0.15 mmol NEt3, PFA tubing (0.75 mm
i. d., 0.7 mL internal vol.); step 2 400 μL segment: 0.05 mmol 11, 0.10 mmol alkyne 5, 3 mol % PdCl2(PPh3)2, 10 mol % CuI, Hastelloy tubing (0.75
b
c
mm i. d., 2.0 mL internal vol.). Isolated yields. Step 1 conducted in MeCN/H2O (9:1). 7 equiv of DBU.
used for the cross-coupling step (entry 7), leading to a much
more environmentally benign process. Moreover, the absence
of DMSO facilitated a much faster and more reliable isolation
of the final product. The volatile components, mainly MeCN
solvent and polyfluorinated spacer, were removed under
reduced pressure and the residue was purified by preparative
TLC to afford the desired indolizine 12{1,1} in 59% isolated
yield.
Library Construction. Using these optimized conditions,
we undertook the construction of a library of indolizine
derivatives based upon the core fragment 10 (Table 3). We
chose to employ a series of propargyl amines (reagent chemset
2) for the generation of our library, including aromatic (2{1−
2}) and aliphatic (2{3−4}) substituents of different polarity, as
well as the N-Boc-protected propargylamine 2{5}.
Diversity Through Amine Acylation. Building upon our
pilot studies using an acetylation reaction as the amine
diversification step, we generated the first wave of the library
using a series of acylation reactions. Thus, starting from TFA
salt 10 the desired compounds were formed in good isolated
yields (12{1,1−1,5}, 59−78%, 65% average yield) after
undergoing three sequential modifications (amidation, Sonoga-
shira coupling, cycloisomerization). We were able to quickly
expand our library with two other commercially available
574
dx.doi.org/10.1021/co300094n | ACS Comb. Sci. 2012, 14, 570−578

ACS Combinatorial Science
Research Article
Figure 1. Library members featuring alternative heterocyclic cores.
anhydrides, benzoic anhydride 5{2} and butyric anhydride
5{3} using this general protocol. In these cases, all desired
library compounds were obtained in good yields and excellent
purity after preparative TLC purification (52−76%, 59%
average yield for chemset 12{2,1−2,5}; 51−65%, 59% average
yield for chemset 12{3,1−3,5}). Thus, we were able to rapidly
generate 15 members of the library, covering a range of physical
properties, illustrating how the properties of the library can
readily be tuned through the choice of substituents at the two
points of diversity on the core heteroaryl bromide fragment.
Notably, N-Boc-protected amines are very well tolerated under
the reaction conditions (12{1,5−5,5}), allowing the future
possibility of subsequent deprotection and further derivatiza-
tion of the resultant amine.
sulfonamidation with tosyl chloride 5{5} without precipitation
of the corresponding ammonium salts.24 Moreover, we found
that these conditions were compatible with the overall
indolizine formation protocol, thereby opening the way to yet
further diversification of our library. The presence of water (20
μL of 400 μL for the cross-coupling step) had a slightly
detrimental effect on the cycloisomerization to the indolizine,
and thus yields were generally lower. However, in case of very
low yields, a greater excess of DBU base (7 equiv) may be
employed to increase the isolated yields (12{4,1} and
12{5,1}).
The library members bearing a tertiary amine (resulting from
an alkylation step) were mostly isolated in lower yields
(chemset 12{4,1−4,5}, 24−47%, 35% average yield). The
reason for this is unclear, but we presume that the tertiary
amine moiety might undergo decomposition under the basic
aqueous reaction conditions, as LC-MS analysis usually showed
a much more complex reaction mixture for these trans-
formations. However, because of the high reaction concen-
trations of the developed protocol, 10−20 mg of pure material
could be isolated from only 2 segments for each compound. In
contrast, the sulfonamides were usually formed in good yields
(chemset 12{5,1−5,5}, 42−54%, 44% average yield).
Diversity Through Heteroaryl Bromide Core. To
further exemplify the applicability of this protocol and the
potential size the corresponding virtual library, we carried out
additional experiments using several other amine substituted
heterocyclic bromide cores (Figure 1). Compound 7{1,1}
represents our prototypical library member,11 resulting from
acetylation of core fragment 4 and coupling with propargyl-
amine 2{1}. This agent was resynthesized using our optimized
multistep flow protocol, resulting in an improved yield. This
indicates that an analogous 25-member library, employing the
variations illustrated in Table 3, would be accessible from this
core fragment. Similarly, 13{1,1} could be prepared using the
same acetylation/coupling protocol, again indicating access to
yet another 25-member library featuring the substitution
patterns illustrated in Table 3. Finally, 14{1,1}, a regioisomer
of 12{1,1}, could be prepared, indicating its compatibility with
generation of a matrix of analogs along the lines illustrated in
Table 3. Thus, based on the already generated 25-membered
library in Table 3, a virtual library of at least 100 compounds is
now accessible in less than three days using our protocol.
Diversity Through Further Propargylamine Variants.
Importantly, our optimized method is not limited to orthogonal
modification of the heteroaryl bromide core. Thus, the readily
synthesized TFA salt of N-propargylpiperazine can also serve as
a building block for a further variant of the library. Again, N-
modification is carried out according to the described
procedure in the incubation chamber, but this time a secondary
amine within the alkyne coupling partner was acylated and then
coupled with methyl 6-bromonicotinate in the Sonogashira/
cycloisomerization sequence to afford indolizine 15{1,1} in
good yield (Figure 2).
The required laboratory work for the synthesis of these first
15 library members involved preparation of the corresponding
ten stock solutions (TFA salt 10 and NEt3, catalyst, reagent
chemsets 2{1−3} and 5{1−5}) and programming of the
Conjure software to perform the correct loading of source vials
and aspiration of stock solutions. Each of the library
compounds in Table 3 was isolated from two identical
segments, allowing straightforward purification and character-
ization of each compound. For the entire sequence involved in
executing these two experiments/segments, the reactor requires
approximately 30 min from loading the first vial to completion
of collection.20 As a result, the first 15 library compounds
12{1,1−3,5} were prepared and separately collected within 7.5
h in fully automated fashion. Final isolation and purification was
achieved by preparative TLC. For lead discovery purposes
however, single segment runs would be sufficient to generate
the necessary material due to the high reaction concentrations
(0.125 M for step 2), thus decreasing the overall reactor time to
approximately 5 h.21 Furthermore, in principle, the reactor
could be combined with a preparative HPLC-MS device for
automatic purification of the desired compounds, as was
recently disclosed by researchers at Abbott Laboratories.22
Diversity Through Other Amine Derivatizations. At
this point we investigated whether our optimized protocol was
suitable for N-modifications other than amide formation. A
variety of electrophilic reagent chemsets were therefore
screened for their suitability. Electrophiles such as alkyl halides
and sulfonyl chlorides readily afforded the desired tertiary
amines and sulfonamides respectively, within 10 min in MeCN.
However, both reactions were routinely accompanied by rapid
formation of voluminous precipitates (the corresponding
ammonium halide salts), thus making them unsuitable for a
flow chemistry sequence. Since the reactivity was acceptable in
each case, we re-evaluated the solvent system in the hope of
avoiding the precipitation issue. We reasoned that the addition
of minimal amounts of water might suffice to efficiently
solubilize the ammonium salts while having no adverse effect
on the first reaction and, more importantly, on the subsequent
Sonogashira coupling.23 We were delighted to find that a
mixture of MeCN and water (9:1) resulted in full conversion
for both the alkylation using benzyl bromide 5{4} and the
575
dx.doi.org/10.1021/co300094n | ACS Comb. Sci. 2012, 14, 570−578

ACS Combinatorial Science
Research Article
0.05 mmol for the TFA salt, 0.15 mmol for NEt3 or 0.25 mmol
for pyridine) and electrophilic reagent (1.20 M in MeCN, 50
μL, 0.06 mmol) were aspirated from their respective source
vials located in position C (TFA salt/base) and position D
(electrophilic reagent), mixed through a PFA mixing tube (0.2
mm i. d.), and loaded into a sample loop in the incubation
chamber to remain for 10 min at room temperature. The alkyne
(2.00 M in MeCN, 50.0 μL, 0.10 mmol) and catalyst stock
solution (1.5 μmol PdCl2(PPh3)2, 5 μmol CuI, 0.25 mmol
DBU in MeCN, 150 μL) were aspirated from their respective
source vials located in position A (alkyne) and position B
(catalyst), mixed with the segment containing the N-modified
material in MeCN through a PFA mixing tube (0.2 mm inner
diameter), and loaded into an injection loop. The reaction
segment (400 μL) was injected into the flow reactor (Hastelloy
coil, 0.75 mm i. d., 2.0 mL internal vol.) set at 180 °C and
passed through the reactor at a flow-rate of 286 μL·min−1 (7
min residence time). A total of 2 reaction segments prepared in
this manner were collected.
Figure 2. Prototypical library member featuring diversification via the
propargyl amine fragment.
As a result, the size and diversity of the alternative library
generated using this approach would be determined by
acylation, alkylation and sulfonamidation reactions of the
propargylamine, and availablility of heteroaryl bromides for
the second step cross-coupling. These can include a variety of
2-bromopyridines, -pyrimidines, -pyrazines, and -quinolines, as
described in our earlier report.11
CONCLUSION
■
In summary, we have developed a novel flow methodology for
the orthogonal modification of heterocyclic building blocks and
its subsequent application for automated library synthesis of
druglike molecules. All reaction steps are conducted within the
flow reactor and products are isolated in good yields following
three sequential transformations. We have exemplified this
methodology through generation of a diverse 25-membered
library of druglike 3-aminoindolizines. We were able to show
that the optimized conditions allow a variety of quantitative
transformations to take place on the secondary amine within 10
min, including acylations, alkylations and sulfonamidations. The
latter two modifications require the addition of water (10%) to
solubilize ammonium halide salts. Under optimized flow
conditions the subsequent tandem Sonogashira/cycloisomeri-
zation reaction can be carried out with a variety of propargyl
amines within 7 min at 180 °C in MeCN to afford the desired
compounds in good yields. All products were isolated by
preparative TLC and fully analyzed and characterized.
Upon complete collection, the volatile reaction components
were removed from the reaction solution in a stream of
nitrogen. The crude oily residue was purified by preparative
TLC (silica gel, hexanes/ethyl acetate or dichloromethane/
methanol) to afford the desired library compounds.
Compound 12{1,1}. Compound 12{1,1} was synthesized
according to the general flow procedure using 2 segments (400
μL each) containing TFA salt 10 and NEt3 (0.33 M in MeCN,
150 μL, 0.05 mmol for the TFA salt, 0.15 mmol for NEt3),
acetic anhydride 5{1} (1.20 M in MeCN, 50 μL, 0.06 mmol),
N-methyl-N-propargylbenzylamine 2{1} (2.00 M in MeCN,
50.0 μL, 0.10 mmol) and catalyst stock solution (in MeCN, 150
μL) and a flow-rate of 286 μL min−1 (7 min residence time).
Upon completion of the reaction, the volatile reaction
components (mainly MeCN, NEt3, and fluorous spacer) were
removed in a stream of nitrogen. The oily residue was purified
by preparative TLC (silica gel, dichloromethane/methanol
19:1) to afford 23.1 mg (59%) of a yellow solid. 1H NMR (500
MHz, CHLOROFORM-d) δ ppm 2.14 (s, 3 H) 2.69 (s, 3 H)
3.51 (br. s., 2 H) 3.63 (d, J = 8.44 Hz, 4 H) 3.69 (br. s., 2 H)
4.08 (s, 2 H) 6.42 (d, J = 4.04 Hz, 1 H) 6.52 (d, J = 4.03 Hz, 1
H) 6.58 (d, J = 9.17 Hz, 1 H) 7.24 - 7.28 (m, 1 H) 7.31 (d, J =
4.40 Hz, 5 H) 8.21 (s, 1 H); 13C NMR (126 MHz, chloroform-
d) δ ppm 21.3, 41.4, 41.5, 46.1, 60.6, 98.4, 104.8, 114.0, 117.4,
118.9, 122.3, 127.3, 128.3, 128.5, 131.9, 135.8, 137.7, 169.1,
169.4; HRMS (ESI-TOF): C23H26N4O2 [M + H]+ calculated
391.2128, found 391.2138.
The size and diversity of the virtual library accessible via this
protocol is substantial, since it allows a variety of N-
modifications including acylations, alkylations and sulfonami-
dations to be carried out in short time. Moreover, the second
step coupling can be conducted within minutes employing a
broad range of propargyl amines and amides that bear aromatic
and aliphatic groups of varying polarities. We have also
demonstrated the ability to introduce further diversity through
variation of the amino-substituted heteroaryl bromide core
fragment, where the three further variants examined indicate
that a total of 100 3-aminoindolizines could be prepared within
several days.
All compounds in Table 3 and Figures 1 and 2 were prepared
accordingly. For full experimental procedures and analytical
data, see the ESI.
Building upon the robust first step reaction conditions,
expansion of the second step protocol to include other cross-
coupling reactions appears viable, thus potentially expanding
the accessible chemical space still further with one general
protocol. The option of combining the flow reactor employed
in this program with an automated purification system has the
potential to achieve a highly efficient production of large
libraries of druglike heterocycles. This will be the subject of
future investigations.
ASSOCIATED CONTENT
■
S
* Supporting Information
Conjure flow reactor information, additional screening experi-
ments, and analytical data for all compounds is available online.
This material is available free of charge via the Internet at
http://pubs.acs.org.
AUTHOR INFORMATION
EXPERIMENTAL SECTION
■
■
Corresponding Author
*Phone: (1) 858 784 2507. Fax: (1) 858 784 7550. E-mail:
kjames@scripps.edu.
General Procedure for the Synthesis of Library
Compounds Using a Conjure Flow Reactor. The
corresponding TFA salt and base (0.33 M in MeCN, 150 μL,
576
dx.doi.org/10.1021/co300094n | ACS Comb. Sci. 2012, 14, 570−578

ACS Combinatorial Science
Research Article
flow. Synthesis of an unsymmetrically substituted 3,5-diamino-
benzonitrile library. Mol. Diversity 2011, 15, 631−638.
Author Contributions
The manuscript was written through contributions of all
authors. All authors have given approval to the final version of
the manuscript.
(6) (a) Bogdan, A. R.; Sach, N. W. The use of copper flow reactor
technology for the continuous synthesis of 1,4-disubstituted 1,2,3-
triazoles. Adv. Synth. Catal. 2009, 351, 849−854. (b) Thompson, C.
M.; Poole, J. L.; Cross, J. L.; Akritopoulou-Zanze, I.; Djuric, S. W.
Small molecule library synthesis using segmented flow. Molecules 2011,
16, 9161−9177. (c) Tu, N. P.; Hochlowski, J. E.; Djuric, S. W.
Ultrasound-assisted click chemistry in continuous flow. Mol. Diversity
2012, 16, 53−58.
Notes
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
■
The authors would like to thank Pfizer Worldwide R&D for
funding and generous donation of chemicals.
(7) Wiles, C.; Watts, P. Parallel synthesis in an EOF-based micro
reactor. Chem. Commun. 2007, 4928−4930.
(8) For a review on multi-step organic synthesis of single molecules
in flow, see: Webb, D; Jamison, T. F. Continuous flow multi-step
organic synthesis. Chem. Sci. 2010, 1, 675−680.
REFERENCES
■
(1) (a) Kennedy, J. P.; Williams, L.; Bridges, T. M.; Daniels, R. N.;
Weaver, D.; Lindsley, C. W. Application of combinatorial chemistry
science on modern drug discovery. J. Comb. Chem. 2008, 10, 345−354.
(b) Verdonk, M. L.; Rees, D. C. Group efficiency: A guideline for hits-
to leads chemistry. ChemMedChem 2008, 3, 1179−1180. (c) Baxendale,
I. R.; Hayward, J. J.; Ley, S. V.; Tranmer, G. K. Pharmaceutical strategy
and innovation: An academics perspective. ChemMedChem 2007, 2,
768−788. (d) Dar, Y. L. High-throughput experimentation: A
powerful enabling technology for the chemicals and materials industry.
Macromol. Rapid Commun. 2004, 25, 34−47. (e) Lombardino, J. G.;
Lowe, J. A., III. The role of the medicinal chemist in drug discovery
Then and now. Nat. Rev. Drug Discovery 2004, 3, 853−862.
(9) Griffiths-Jones, C. M.; Hopkin, M. D.; Jonsson, D.; Ley, S. V.;
̈
Tapolczay, D. J.; Vickerstaffe, E.; Ladlow, M. Fully automated flow-
through synthesis of secondary sulfonamides in a binary reactor
system. J. Comb. Chem. 2007, 9, 422−430.
́
(10) Petersen, T. P.; Ritzen, A.; Ulven, T. A multistep continuous-
flow system for rapid on-demand synthesis of receptor ligands. Org.
Lett. 2009, 11, 5134−5137.
(11) (a) Chouhan, G.; James, K. CuAAC macrocyclization: high
intramolecular selectivity through the use of copper−tris(triazole)
ligand complexes. Org. Lett. 2011, 13, 2754−2757. (b) Meyer, F.-M.;
Collins, J. C.; Borin, B.; Bradow, J.; Liras, S.; Limberakis, C.;
Mathiowetz, A. M.; Philippe, L.; Price, D.; Song, K.; James, K. Biaryl-
bridged macrocyclic peptides: conformational constraint via carbo-
genic fusion of natural amino acid side chains. J. Org. Chem. 2012, 77,
3099−3114.
(2) For reviews, see: (a) Watts, P. The application of microreactors
in combinatorial chemistry. QSAR Comb. Sci. 2005, 24, 701−711.
(b) Watts, P.; Haswell, S. J. The application of micro reactors for
organic synthesis. Chem. Soc. Rev. 2005, 34, 235−246. (c) Jones, R. V.;
̈
Csajagi, C.; Szekelyhidi, Z.; Kovacs, I.; Borcsek, Be.; Urge, L. Flow
(12) (a) Bogdan, A. R.; James, K. Efficient access to new chemical
space through flowConstruction of druglike macrocycles through
copper-surface-catalyzed azide−alkyne cycloaddition reactions.
Chem.Eur. J. 2010, 16, 14506−14512. (b) Bogdan, A. R.; James,
K. Synthesis of 5-iodo-1,2,3-triazole-containing macrocycles using
copper flow reactor technology. Org. Lett. 2011, 13, 4060−4063.
(c) Bogdan, A. R.; Jerome, S. V.; Houk, K. N.; James, K. Strained
cyclophane macrocycles: Impact of progressive ring size reduction on
synthesis and structure. J. Am. Chem. Soc. 2012, 134, 2127−2138.
(13) Lange, P. P.; Bogdan, A. R.; James, K. A new flow methodology
for the expedient synthesis of druglike 3-aminoindolizines. Adv. Synth.
Catal. 2012, DOI: 10.1002/adsc.201200316.
(14) (a) Kel’in, A. V.; Sromek, A. W.; Gevorgyan, V. A novel Cu-
assisted cycloisomerization of alkynyl imines: Efficient synthesis of
pyrroles and pyrrole-containing heterocycles. J. Am. Chem. Soc. 2001,
123, 2074−2075. (b) Schwier, T.; Sromek, A. W.; Yap, D. M. L.;
Chernyak, D.; Gevorgyan, V. Mechanistically diverse copper-, silver-,
and gold-catalyzed acyloxy and phosphatyloxy migrations: Efficient
synthesis of heterocycles via cascade migration/cycloisomerization
approach. J. Am. Chem. Soc. 2007, 129, 9868−9878. (c) Liu, Y.; Song,
Z.; Yan, B. General and direct synthesis of 3-aminoindolizines and
their analogues via Pd/Cu-catalyzed sequential cross-coupling/cyclo-
isomerization reactions. Org. Lett. 2007, 9, 409−412. (d) Chernyak,
D.; Gadamsetty, S. B.; Gevorgyan, V. Low temperature organocopper-
mediated two-component cross coupling/cycloisomerization approach
toward N-fused heterocycles. Org. Lett. 2008, 10, 2307−2310.
(15) www.accendocorporation.com
(16) Every segment that is prepared in the fluid prep module of the
Conjure reactor is preceded and followed by an immiscible liquid
spacer (perfluoro(methyldecalin), 50 μL) and is then introduced into
the flow stream and passed through the reactor as an isolated reaction.
(17) The 40-vial carousel is designed to hold 36 reagent solutions
and 4 rinse vials.
(18) The incubation chamber contains five individual sample loops
allowing the preparation of five different first step segments.
(19) The number of additional reagents is limited to three for the
second step, as the fourth syringe is used to reaspirate the first step
reaction segment.
reactors for drug discoveryFlow for reaction optimization, library
synthesis, and scale up. Chim. Oggi. 2008, 26, 10−14.
(3) (a) Baumann, M.; Baxendale, I. R.; Kuratli, C.; Ley, S. V.; Martin,
R. E.; Schneider, J. Synthesis of a druglike focused library of
trisubstituted pyrrolidines using integrated flow chemistry and batch
methods. ACS Comb. Sci. 2011, 13, 405−413. (b) Spencer, J.; Patel,
H.; Callear, S. K.; Coles, S. J.; Deadman, J. J. Synthesis and solid state
study of pyridine- and pyrimidine-based fragment libraries. Tet. Lett.
2011, 52, 5905−5909.
(4) (a) Baumann, M.; Baxendale, I. R.; Ley, S. V.; Smith, C. D.;
Tranmer, G. K. Fully automated continuous flow synthesis of 4,5-
disubstituted oxazoles. Org. Lett. 2006, 8, 5231−5234. (b) Jones, R.;
̈
́ ́ ́
Godorhazy, L.; Szalay, D.; Gerencser, J.; Dorman, G.; Urge, L.; Darvas,
̈
̈
F. A novel method for high-throughput reduction of compounds
through automated sequential injection into a continuous-flow
microfluidic reactor. QSAR Comb. Sci. 2005, 24, 722−727.
(c) Knudsen, K. R.; Holden, J.; Ley, S. V.; Ladlow, M. Optimisation
of conditions for O-benzyl and N-benzyloxycarbonyl protecting group
removal using an automated flow hydrogenator. Adv. Synth. Catal.
2007, 349, 535−538. (d) Clapham, B.; Wilson, N. S.; Michmerhuizen,
M. J.; Blanchard, D. P.; Dingle, D. M.; Nemcek, T. A.; Pan, J. Y.; Sauer,
D. R. Construction and validation of an automated flow hydrogenation
instrument for application in high-throughput organic chemistry. J.
Comb. Chem. 2008, 10, 88−93. (e) Garcia-Egido, E.; Spikmans, V.;
Wong, S. Y. F.; Warrington, B. H. Synthesis and analysis of
combinatorial libraries performed in an automated micro reactor
system. Lab Chip 2003, 3, 73−76.
(5) (a) Schwalbe, T.; Kadzimirsz, D.; Jas, G. Synthesis of a library of
ciprofloxacin analogues by means of sequential organic synthesis in
microreactors. QSAR Comb. Sci. 2005, 24, 758−768. (b) Csajgi, C.;
̈
Borcsek, B.; Niesz, K.; Kovcs, I.; Szkelyhidi, Z.; Bajk, Z.; Urge, L.;
Darvas, F. High-efficiency aminocarbonylation by introducing CO to a
pressurized continuous flow reactor. Org. Lett. 2008, 10, 1589−1592.
(c) Zhang, Y; Jamison, T. F.; Patel, S.; Mainolfi, N. Continuous flow
coupling and decarboxylation reactions promoted by copper tubing.
Org. Lett. 2011, 13, 280−283. (d) Lengyel, L.; Gyol
́
lai, V.; Nagy, T.;
̈
Dorman, G.; Terleczky, P.; Hada, V.; Nograd
́
́
́
́
i, K.; Sebok
́
, F.; Urge, L.;
Darvas, F. Stepwise aromatic nucleophilic substitution in continuous
577
dx.doi.org/10.1021/co300094n | ACS Comb. Sci. 2012, 14, 570−578

ACS Combinatorial Science
Research Article
(20) Once the first segment is subjected to the flow stream to process
the second reaction step, the fluid prep module immediately begins the
required operations to prepare the first step of the second segment.
(21) The overall reaction time is 17 min (10 min for the first step and
7 min residence time in the flow reactor coil). An additional 3 min is
required for operations of the fluid prep module including loading of
vials and aspirating and mixing of stock solutions.
(22) (a) Hochlowski, J. E.; Searle, P. A.; Tu, N. P.; Pan, J. Y.;
Spanton, S. G.; Djuric, S. W. An Integrated Synthesis−Purification
System to Accelerate the Generation of Compounds in Pharmaceutical
Discovery. J. Flow Chem. 2011, 2, 56−61. (b) For an early approach to
automated library synthesis under continuous flow conditions on a
microchip device, see: Wong-Hawkes, S. Y. F.; Chapela, M. J. V.;
Montembault, M. Leveraging the advantages offered by microfluidics
to enhance the drug discovery process. QSAR Comb. Sci. 2005, 24,
712−721.
(23) For recent reports on Sonogashira couplings in water, see:
(a) Raju, S.; Kumara, P. R.; Mukkantic, K.; Annamalai, P.; Pal, M.
Facile synthesis of substituted thiophenes via Pd/C-mediated
Sonogashira coupling in water. Bioorg. Med. Chem. Lett. 2006, 16,
6185−6189. (b) Shi, S.; Zhang, Y. Palladium-catalyzed copper-free
Sonogashira coupling reaction in water and acetone. Synlett 2007,
1843−1850. (c) Bakherad, M.; Keivanloo, A.; Bahramian, B.; Hashemi,
M. Copper-free Sonogashira coupling reactions catalyzed by a water-
soluble Pd−salen complex under aerobic conditions. Tetrahedron Lett.
2009, 50, 1557−1559. (d) Kamali, T. A.; Bakherad, M.;
Nasrollahzadeh, M.; Farhangi, S.; Habibi, D. Synthesis of 6-substituted
imidazo[2,1-b]thiazoles via Pd/Cu-mediated Sonogashira coupling in
water. Tetrahedron Lett. 2009, 50, 5459−5462. (e) Kamali, T. A.;
Habibi, D.; Nasrollahzadeh, M. Synthesis of 6-substituted imidazo[2,1-
b][1,3]thiazoles and 2-substituted imidazo[2,1-b][1,3]benzothiazoles
via Pd/Cu-mediated Sonogashira coupling. Synlett 2009, 2601−2604.
(f) Suzuka, T.; Okada, Y.; Ooshiro, K.; Uozumi, Y. Copper-free
Sonogashira coupling in water with an amphiphilic resin-supported
palladium complex. Tetrahedron 2010, 66, 1064−1069. (g) Teratani,
T.; Ohtaka, A.; Kawashima, T.; Shimomura, O.; Nomura, R. Copper-
free Sonogashira coupling in water with linear polystyrene-stabilized
PdO nanoparticles. Synlett 2010, 2271−2274. (h) Bakherad, M.;
Keivanloo, A.; Kalantar, Z.; Jajarmi, S. Pd/C-catalyzed heterocycliza-
tion during copper-free Sonogashira coupling: synthesis of 2-
benzylimidazo[1,2-a]pyrimidines in water. Tetrahedron Lett. 2011,
52, 228−230. (i) Liu, N.; Liu, C.; Xu, Q.; Jin, Z. Thermoregulated
copper-free Sonogashira coupling in water. Eur. J. Org. Chem. 2011,
4422−4428.
(24) One of the reviewers of this manuscript pointed out the superior
solubility of DIPEA·HCl in MeCN in comparison to NEt3·HCl. In
cases when the addition of water is not viable, DIPEA can be
employed instead of NEt3 for the first step N-modification.
578
dx.doi.org/10.1021/co300094n | ACS Comb. Sci. 2012, 14, 570−578