Are We There Yet? Applying Thermodynamic and Kinetic Profiling on Embryonic Ectoderm Development (EED) Hit-to-Lead Program

Article abstract of DOI:10.1021/acs.jmedchem.7b00576

It is advocated that kinetic and thermodynamic profiling of bioactive compounds should be incorporated and utilized as complementary tools for hit and lead optimizations in drug discovery. To assess their applications in the EED hit-to-lead optimization p

Full text of DOI:10.1021/acs.jmedchem.7b00576

Subscriber access provided by University of Sussex Library
Article
Are We There Yet? – Applying Thermodynamic and Kinetic Profiling
on Embryonic Ectoderm Development (EED) Hit-to-Lead Program
Ying Wang, Rohinton P. Edalji, Sanjay C. Panchal, Chaohong C. Sun, Stevan W Djuric, and Anil Vasudevan
J. Med. Chem., Just Accepted Manuscript • DOI: 10.1021/acs.jmedchem.7b00576 • Publication Date (Web): 19 Sep 2017
Downloaded from http://pubs.acs.org on September 19, 2017
Just Accepted
“Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted
online prior to technical editing, formatting for publication and author proofing. The American Chemical
Society provides “Just Accepted” as a free service to the research community to expedite the
dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts
appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been
fully peer reviewed, but should not be considered the official version of record. They are accessible to all
readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered
to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published
in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just
Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor
changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers
and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors
or consequences arising from the use of information contained in these “Just Accepted” manuscripts.
Journal of Medicinal Chemistry is published by the American Chemical Society. 1155
Sixteenth Street N.W., Washington, DC 20036
Published by American Chemical Society. Copyright © American Chemical Society.
However, no copyright claim is made to original U.S. Government works, or works
produced by employees of any Commonwealth realm Crown government in the course
of their duties.

Page 1 of 56
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Are We There Yet? – Applying Thermodynamic
and Kinetic Profiling on Embryonic Ectoderm
Development (EED) HitꢀtoꢀLead Program
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
,
‡
‡
‡
Ying Wang* , Rohinton P. Edalji , Sanjay C. Panchal , Chaohong Sun, Stevan W. Djuric and
Anil Vasudevan
AbbVie Inc., 1 North Waukegan Road, North Chicago, Illinois 60064, United States
‡
These authors contributed equally
KEYWORDS. StructureꢀKinetics Relationship, StructureꢀThermodynamics Relationship, EED,
ITC, k , k , enthalpy, entropy, SPR
off
on
ABSTRACT. It is advocated that kinetic and thermodynamic profiling of bioactive compounds
should be incorporated and utilized as complementary tools for hit and lead optimizations in drug
discovery. To assess their applications in the EED hitꢀtoꢀlead optimization process, large amount
of thermodynamic and kinetic data were collected and analyzed via isothermal titration
calorimetry (ITC) and surface plasmon resonance (SPR), respectively. Slower dissociation rates
(k ) of the lead compounds were observed as the program progressed. Analysis of the kinetic
off
data indicated that compound cellular activity correlated with both K and k . Our analysis
i
off
ACS Paragon Plus Environment
1

Journal of Medicinal Chemistry
Page 2 of 56
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
revealed that ITC data should be interpreted in the context of chiral purity of the compounds.
The thermodynamic signatures of the EED amino pyrrolidine compounds were found to be
mainly enthalpy driven with improved enthalpic contributions as the program progressed. Our
study also demonstrated that significant challenges still exist in utilizing kinetic and
thermodynamic parameters for hit selection.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
INTRODUCTION
Drug discovery is a complex process in which multiple factors of bioactive compounds need to
be optimized in parallel to lead to a clinical candidate with balanced efficacy and selectivity. For
nearly a century, the binding affinity of a compound for the target of interest has been one of the
major guiding principles in drug discovery, coined by Paul Ehrlich in 1913, “Substances do not
act unless bound”. However, it has also been long recognized that binding affinity alone is not
sufficient to predict a successful clinical outcome. As such, a number of metrics focusing on the
a compound’s characteristics, such as Lipinski rule of five and its variants, have been correlated
to clinical observations, and subsequently incorporated routinely in the hit and lead optimization
1
processes. As we continued to look for better or complementary principles to guide drug
discovery efforts, we became aware of studies that discuss the importance of incorporating
binding kinetics and thermodynamics into drug discovery compound triage over the last decade.
It has been proposed that efficacy and selectivity may also be driven by kinetics and
2
,3,4,5
thermodynamics, often overlooked dimensions in conventional SAR profiling.
Association and dissociation rate constants (k and k ) of a compound binding to a biological
on
off
target are not intrinsically related to one another, although they are connected by dissociation
ACS Paragon Plus Environment
2

Page 3 of 56
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
equilibrium constant K (K = k /k ). In theory, they are correlated with the structural
D
D
off on
6
,7
elements of the compoundꢀprotein complex during the association and dissociation processes.
The kinetic profile of structurally different compounds could reveal features and aspects of
compoundꢀprotein interactions and thus aid optimization. This is defined as structureꢀkinetics
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
relationships (SKR). It is postulated that in many cases, k is a better predictor for in vivo
off
efficacy and selectivity, and has the potential to differentiate otherwise indistinguishable
8
,9
compounds with similar affinities.
Similarly, the relative contribution of the binding enthalpy ꢀH and entropic term ꢀTꢀS of a
10,11
compound is claimed to be of importance to predict its selectivity in clinical studies.
Structureꢀthermodynamics relationships (STR) could be defined in theory as well. A large
negative value of ꢀH is believed to be desirable as it reflects the strength of specific compoundꢀ
ligand interactions such as hydrogen bonds, while large negative entropic term ꢀTꢀS mainly
reflects nonspecific lipophilic interactions.
Recommendations were thus made that a more holistic approach including binding kinetic and
thermodynamic data should be incorporated as early as possible in the drug discovery process. It
has been claimed that this may enable medicinal chemists to identify and prioritize hit and lead
2
,3,4,5
compounds with bestꢀinꢀclass potentials.
However, many of these claims were rationalized
from retrospective analyses of drugs in the clinic, giving little guidance on their utilization in the
hit and lead optimization processes. In addition, there exists a lack of thermodynamic and kinetic
data in the drug discovery phase to sustain many of the underlying assumptions. General
validation of these claims remains to be established. In fact, some of the claims are not without
1
2, 13
controversies and have been challenged by several recent articles.
The proposed impact of
ACS Paragon Plus Environment
3

Journal of Medicinal Chemistry
Page 4 of 56
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
binding kinetics and thermodynamics that will ultimately lead to clinical candidates, is yet to be
1
4
demonstrated. To gauge the applications of thermodynamic and kinetic profiling in drug
discovery, we decided to investigate these parameters in a hitꢀtoꢀlead medicinal chemistry
program.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Polycomb repressive complex 2 (PRC2) catalyzes trimethylation of histone H3 at lysine 27
1
5
(
H3K27me3). It plays an essential role in gene regulation and is a known epigenetic drug target
for cancer therapy. EZH2 (enhancer of zeste homolog 2) is the catalytic subunit of PRC2 that is
responsible for histone methylation. As such, EZH2 has received a lot of attention within the
pharmaceutical industry as a potential drug target. To date, a number of small molecule EZH2
1
6
inhibitors have been developed and advanced into cancer clinical trials. EED (embryonic
ectoderm development), another core component of PRC2, is at least partially responsible for the
recognition of H3K27me3. The direct association of the PRC2 complex to histone H3K27me3
through EED stimulates the successive methyltransferase activity of PRC2, which results in the
propagation of the H3K27me3 histone mark and silencing of target genes through generation of
repressed chromatin. Therefore, blocking this process allosterically via inhibiting the H3K27me3
binding function of EED seems to be an attractive approach for cancer therapy. Indeed, we and
Novartis have very recently disclosed novel EED inhibitors that demonstrated robust efficacy
1
7,18
against human tumor xenograft models.
Our medicinal chemistry efforts started with the identification of amino pyrrolidine compound 1
19
from a high throughput thermal shift assay (TSA) based screen (Figure 1). Compound 1 was
confirmed by EED TRꢀFRET assay with a K of 600 nM. The Xꢀray crystal structure of EED
i
WD40 domain with compound 1 at 1.28 Å suggested that compound 1 bound in the histone
peptide binding site of EED. Interestingly, when compared to the previously published structure
ACS Paragon Plus Environment
4

Page 5 of 56
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
of EED binding the H3K27mes peptide, significant remodeling of the peptide binding pocket to
accommodate compound 1 was observed, despite similar global binding of peptide and
1
7
compound 1 to EED. Both Trp364 and Tyr365 flipped up to expose a larger and more
druggable binding pocket that allowed the 2ꢀmethoxybenzyl group to slide deep into the
hydrophobic pocket (Figure 2a). Additional key interactions included a hydrogen bond from the
dimethylamino group to highly coordinated water. This dimethylamino moiety is, thus, largely
nonꢀreplaceable (Figure 2b). Other interactions include the hydrophobic and van der Waals
interactions between the rings of the indole and protein. The initial complex structure suggested
that the both the indole (R ) and Nꢀbenzyl moieties (R ) in compound 1 would be amenable for
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
1
2
1
9
SAR studies (Figure 3).
The study of structureꢀkinetics and structureꢀthermodynamics relationships (SKR and STR),
needless to say, will be most fruitful when considering analogs within a specific chemical series
for a given target. The availability of the high resolution xꢀray structures and biophysical
confirmation methods, coupled with synthetic tractability of the amino pyrrolidine series
prompted us to study the SKR and STR at the onset of EED hitꢀtoꢀlead program. This case study
was aimed at gauging the feasibility of applying kinetic and thermodynamic information for the
prospective design in the hitꢀtoꢀlead optimization stage.
RESULTS AND DISCUSSION
DATASET. About one hundred amino pyrrolidine EED inhibitors were profiled in ITC and SPR
assays in addition to EED TRꢀFRET assay. These analogs were carefully selected to cover a
broad range of biochemical affinities (K = 0.0005 µM to 50 µM). Analogs with chemical
i
ACS Paragon Plus Environment
5

Journal of Medicinal Chemistry
Page 6 of 56
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
structural variations were systematically selected and included in the dataset (Figure 3). Data
from cellular assays that measured selective inhibition of H3K27 trimethylation and proliferation
in tumor cell lines were collected on half of the compounds. Xꢀray crystal structures were
obtained on twenty percent of the compounds. This represents one of the most comprehensive
thermodynamic and kinetic profiling datasets in the literature that focuses on one chemical series
for one target.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
STRUCTUREꢀKINETICS RELATIONSHIP. A Biacore T200 was used to collect the
compound binding data. EED (h) (78ꢀ441) was minimally biotinylated and captured on a
neutravidin surface for all SPR binding studies. Compounds were then assayed in the SPR runs.
3
7
As per manufacturer guidance, the detection ranges of the SPR used in our study are k 10 ꢀ10
on
ꢀ1 ꢀ1
ꢀ5
ꢀ1
M s and k 10 ꢀ 0.5 s , respectively. Roughly 25% of the EED compounds studied had both
off
7
ꢀ1 ꢀ1
ꢀ1
fast k ( >10 M s ) and k ( > 0.5 s ) rates that were beyond our biacore detection limit. In
on
off
these cases, steadyꢀstate K values were obtained instead and ranged from 8 nM to 50 µM.
D
3
7
ꢀ1 ꢀ1
The k values spanned more than three log unit, from 4.2×10 to > 1×10 M s . Similarly, the
on
ꢀ
4
ꢀ1
k spanned more than two log units, from 7×10 to > 0.5 s . Our data clearly showed that k
off
on
could vary dramatically even among close analogs. Recent literature examples suggested in
20,21
certain cases k may also play an important role in downstream processes.
on
The binding affinities determined from SPR studies, K , correlated well with the binding
D
affinities, K , obtained from our TRꢀFRET biochemical assay (Figure 4). The SPR data, thus,
i
directly validated the binding affinity K readout from our EED TRꢀFRET assay.
i
From the SAR studies, the R group, where the original 2ꢀmethoxybenzyl group in compound 1
2
was at, had a strong preference for 2,6ꢀdisubstitution (Figure 3). This led to the discovery of 2ꢀ
ACS Paragon Plus Environment
6

Page 7 of 56
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
bromoꢀ6ꢀfluorobenzyl (A) as the initial preferred R group. More than half of the compounds
2
profiled in SPR were analogs based on R as A while variation in R groups helped map out the
2
1
initial SAR. From the Xꢀray structures, the R group projected out towards waterꢀexposed EED
1
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
protein and was considered to potentially accommodate a variety of substituents. In practice,
many aryl and heteroaryl groups were tolerated at the R position and afforded potent EED
1
amino pyrrolidine analogs.
Extensive analoging via parallel library synthesis in efforts to find other suitable R groups led to
2
the discovery of 2ꢀfluoroꢀ6ꢀmethylbenzyl (B) analogs. These analogs generally had equal or
improved potency to the 2ꢀbromoꢀ6ꢀfluorobenzyl analogs (A), but the 2ꢀfluoroꢀ6ꢀmethylbenzyl
(B) also had much better overall physicochemical properties (Figure 3). A handful of these
compounds were profiled in SPR binding measurements with optimal R groups identified from
1
earlier SAR studies.
As the project progressed, based on Xꢀray crystal structures, we reasoned that restricting the
conformational freedom of the benzyl moiety at the R position may enforce the favorable
2
1
9
binding conformation observed in the Xꢀray structures. This led to the identification of the 7ꢀ
fluoroꢀ2ꢀindanyl group at the R position (C, Figure 3). Fine tuning of the R group eventually
2
1
led to the discovery of compound 2 as the lead compound, which provided onꢀtarget tumor
1
9
growth inhibition in a mouse xenograft model (Figure 1).
The kinetic plot of the compounds with these three different R groups is shown in Figure 5.
2
Compounds with fast k or k were not depicted in Figure 5. Slower k values were observed
on
off
off
generally when R progressed from A to B then to C. However, no such trend was observed for
2
k for this R progression. Instead, a similar range of k values were observed going from B to
on
2
on
ACS Paragon Plus Environment
7

Journal of Medicinal Chemistry
Page 8 of 56
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
C. The broader coverage of k and koff values of analogs with R = A was as expected as we
on
2
used R = A to map out the SAR at R at the beginning of the hit to lead program. This broader
2
1
coverage was also reflected in the wider range of the K values with R = A (Figure 5). As the
D
2
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
program progressed, K and k were simultaneously improved for the EED amino pyrrolidine
D
off
series. This trend was also observed when comparing matched pairs, such as compound 3 vs 4
and compound 5, 6 vs 7 (Table 1), where the trend of k values tracked the binding affinity KD
off
trend, although not linearly.
Without sufficient detailed knowledge to unravel all of the binding events that happened during
the formation of the proteinꢀcompound complex, especially information on the transition states
in the complex formation process, we did not attempt to quantify the observed phenomenon. Yet,
we hypothesized that the optimal interactions between the rigid indanyl group and the
hydrophobic protein pocket as evidenced from the static Xꢀray structures were likely responsible
for the observed slower k in the 7ꢀfluoroindanyl analogs.
off
Enantiomers with the (3R, 4S) configuration were preferred for the R groups studied, generally
2
several fold more potent than the corresponding enantiomers with the (3S, 4R) configurations.
Compound 8, the less potent enantiomer, had fast k and k rates. The more potent enantiomer
on
off
9
1
though, had slower measurable k and k . A similar trend was observed for compound 10 and
on off
1. This indicated that enantiomers can have very different kinetic signatures despite their
structural similarities.
In our case, the effect of the polar group on kinetic parameters depended on its position and
interaction within the Xꢀray structure. Compound 12, lacked a hydroxyl group at the R position
1
compared to compound 13, but showed similar binding affinity. Both compounds had fast kon
ACS Paragon Plus Environment
8

Page 9 of 56
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
and koff rates. Compound 14, with a polar substituent at R , a pyridyl group, also had fast kon
1
and koff rates. On the contrary, comparing compound 15 to compound 16, which differs by a
hydroxyl group at R position, the k was largely unchanged while the k differed by several
2
on
off
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
fold. From Xꢀray structures of similar analogs, we knew that the hydroxyl group at the para
position on the benzyl substituent at the R position, such as in compound 16, forms an Hꢀbond
2
to the carbonyl of Asp310 in the protein complex. This could at least partially explain the slower
koff observed. In contrast, the hydroxyl group of compound 13 is projected out towards the waterꢀ
exposed EED protein without forming a hydrogen bond with the EED protein, and one would
expect little influence on the kinetic parameters.
A proliferation assay featuring the Pfeiffer (human lymphoblast) tumor cell line was used to
evaluate the inhibition of EED by the amino pyrrolidine compounds. The IC of the amino
50
pyrrolidine EED compounds from this cellular assay were binned into three categories: IC less
5
0
than 0.1µM, more than 1µM and IC in between. From the box plots in Figure 6, the cellular
50
activity improved with improved median value of K from the TRꢀFRET assay (Figure 6a). We
i
also noticed a correlation between slower k and improved cellular activity (Figure 6b). On the
off
contrary, the trend was much less obvious with k (Figure 6c). The median value for k only
on
on
increased slightly with improved cellular activity. It should be pointed out that although the koff
values could potentially be used to predict cellular activity, the same conclusion could also be
reached by using the binding affinity K from the TRꢀFRET assay in our case. The lead
i
compound identified (compound 2), had the best cellular activity, and the best K and the slowest
i
k value of all the compounds tested, but not the fastest k .
off
on
We also investigated if there was any correlation between kinetic parameters (k and k ) and
on
off
physicochemical parameters. The wide physicochemical parameter distribution of amino
ACS Paragon Plus Environment
9

Journal of Medicinal Chemistry
Page 10 of 56
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
pyrrolidine series in the hitꢀtoꢀlead stage (e.g. MW 333ꢀ 533, ClogP 1.8 ꢀ 6.4 and tPSA 5ꢀ67)
made such an analysis feasible. No correlation could be found in our case. However, it has been
previously reported that a correlation exists between the k and the molecular weight of
off
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
22
compounds. The correlation though was derived from a study of compounds for more than 40
targets. It is plausible, in our case, that local effects of a specific structural moiety in the context
of the protein target of interest could be dominant over the physicochemical descriptors in
influencing kinetic signatures. The lack of correlation with physicochemical parameters in our
case also indicated that kinetic signatures could be considered as independent parameters in the
compound optimization process.
It was proposed that kinetic profiling could be used to distinguish and prioritize chemical series
in the early stage of the medicinal chemistry programs. The idea was put forward that hits with
slower k values should be prioritized for further optimization. To test this claim, hits identified
off
from our EED HTS campaign were also profiled in the SPR binding. Selected examples were
given in Table 1 (entry 1, 17 and 18). The majority of the initial EED HTS hits identified in our
case possessed fast k and k rates. This is not surprising as the initial hits identified from HTS
on
off
in general were not designed specifically for the target of interest, thus the compounds were not
optimized to display optimal kinetic profiles for the EED target. Based on our experience, it is
very common for the initial HTS hits of a given target to have fast k and k values.
on
off
Furthermore, compound 1 was the original pyrrolidine EED hit with fast k and k rates. Many
on
off
analogs designed based on compound 1 continued to possess fast k and k rates. Thus we were
on
off
not able to use the kinetic parameters for initial compound selection. In order to fully explore the
potential utility of kinetic parameters in hitꢀtoꢀlead stage, further advances in instrumentation
technology for kinetic measurements are warranted.
ACS Paragon Plus Environment
10

Page 11 of 56
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
STRUCTUREꢀTHERMODYNAMICS RELATIONSHIP. It has been proposed that enthalpy
optimization maximizes the influence of forces other than hydrophobicity and compounds with
more favorable enthalpy contribution will likely exhibit better efficacy and selectivity in clinical
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
4,5
trials from retrospective analysis. It has also been shown that in general, medicinal chemistry
programs optimize binding affinities via entropy optimization by increasing the hydrophobicity
of the compounds through the hit and lead optimization processes. As such, it is advocated that
hits with more enthalpic contribution to affinity should be prioritized at the early stage of the
medicinal chemistry campaign, as entropic optimization will likely dominate during the hit and
lead optimization process.
ITC analysis is the primary technology used for directly measuring the enthalpic contribution
ꢀ
H. The binding affinity K is obtained from an ITC experiment via curve fitting of the binding
D
isotherm, and consequently the entropic component (ꢀTꢀS) can be derived.
Both the enthalpic and entropic contributions of the EED compounds tested spanned more than
ten log units, while the corresponding K from ITC experiments and K from TRꢀFRET spanned
D
i
less than five log units. Globally, all the EED inhibitor tested were found to have favorable
enthalpic contributions (ꢀH) and unfavorable entropic contributions (ꢀTꢀS). To our surprise,
there’s no correlation between the K determined from ITC experiments and the K determined
D
i
from the TRꢀFRET assay. The ratios of the K values from ITC experiments versus the K values
D
i
from the TRꢀFRET assay scattered from less than one fold to greater than 3000 fold. The
difference was especially drastic with the 7ꢀfluoroꢀindanyl analogs (Figure 7). Whilst most of the
analogs with the 7ꢀfluoroꢀindanyl group at R position were below 10 nM in the EED TRꢀFRET
2
assays, these compounds had apparent K values in the micromolar ranges in ITC experiments,
D
ACS Paragon Plus Environment
11

Journal of Medicinal Chemistry
Page 12 of 56
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
including lead compound 2 (Figure 7). The stark contrast of binding affinity values from two
measurements prompted us to take a closer examination of the ITC experiments.
We set out to examine the potential influence of various ITC experimental parameters on the
data readout. It is recognized that these parameters may have a profound influence on the data
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
2
3,24,25
obtained from ITC experiments.
The majority of ITC experiments were done in the reverse
mode. However, a few titrations in the forward mode were also carried out and confirmed the
thermodynamic data obtained with reverse mode. Our ITC experiments on EED program had
indicated a minimal effect of pH on the K values within the physiologic pH range. For this
D
dataset all experiments were performed using the same lot of EED protein. Concentrations of the
compounds tested were reconfirmed. None of these studies could explain the huge discrepancies
in the binding affinities observed. Instead, these studies further confirmed and validated the
original ITC experimental data we obtained. Perplexed by these observations, we took a closer
look at Figure 7, interestingly, although globally there’s no correlation, the K from ITC
D
experiments for compound 19 and 22 matched well with the K from TRꢀFRET (Table 2).
i
Contrary to our lead compound 2, which is a mixture of two diastereomers with epimeric center
1
9
at the 7ꢀfluoroindanyl group (Figure 1), compounds 19 and 22 are enantiomers with high ee
values (see experimental section).
Separation of the achiral 7ꢀfluoroindane analogs to the individual four enantiomerically pure
stereoisomers turned out to be very challenging in most cases, especially on preparative scales.
From a few successful chiral separations and the corresponding Xꢀray crystal structures, we
1
9
knew the preferred stereochemistry at the three stereogenic centers, as depicted in compound 19
Table 2). We were able to get chiral compounds 19 and 22 in high ee values while compound 20
(
and 21 were contaminated with 5% of other stereoisomers. The binding affinities for Compound
ACS Paragon Plus Environment
12

Page 13 of 56
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
1
9 matched well in TRꢀFRET, SPR and ITC experiments, as did compound 22. In stark contrast,
compounds 20 and 21, gave measured K values in the micromolar range in ITC experiments
D
even though they had measured K values of 3 and 42 nM respectively, in the TRꢀFRET assay.
i
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
The achiral compound 4, also had a K of 2.1 µM in ITC experiments while the measured
D
affinities were found to be 7 nM in TRꢀFRET and 4 nM in SPR experiments. These results,
therefore, implied that chiral purity of a compound could potentially have huge influence on the
apparent K obtained from ITC experiments, which is not the case for TRꢀFRET and SPR
D
experiments. Compound 4, a mixture of four diastereomers, its binding affinity reflected the
most potent single diastereomer, compound 19, in SPR and TRꢀFRET assays, but not in ITC
assay (Table 2).
To test our hypothesis, we selected all the chiral compounds with high ee values (>95%) and
looked at the K values of these compounds from the ITC experiments. A good correlation
D
between the ITC K and TRꢀFRET K values was obtained with this subset of compounds
D
i
(Figure 8). Most of the literature reports on thermodynamic studies using ITC were studies using
late stage compounds with high chiral purity or drugs on the market, as well as compounds
devoid of chiral centers. The lack of a large ITC dataset on close related analogs also would
make such discovery unlikely. It has also been reported recently that compound impurities could
2
3
lead to pronounced changes in the measured thermodynamic parameters. Here we showed that
the thermodynamic readout from ITC can be further complicated by the chiral purity of the
compounds of interest. This complication indicated that it would be more likely to get false
negative results for binding affinities in ITC experiments compared to TRꢀFRET assay and SPR
binding measurements, where the potency readout will reflect the most potent diastereomer
within the compound tested It seems that when a compound of interest is composed of more
ACS Paragon Plus Environment
13

Journal of Medicinal Chemistry
Page 14 of 56
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
than one stereoisomer, albeit in small amount, simultaneous binding of the each stereoisomer to
the protein in ITC experiments influence the way the instrument models the binding and handles
the data analysis, which could lead to significant changes in the data readout. Advanced software
for ITC data handling and more inꢀdepth studies are needed to decipher the mechanisms behind
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
2
6,27
this observed phenomenon ( See Supporting Information).
Even with this subset of 23 chirally pure compounds, huge ranges of enthalpy and entropy
contributions were observed. The enthalpy component ꢀH ranged from ꢀ20,100 to ꢀ11,640
2
8
cal/mol, corresponding to more than a six log unit change. Similarly, the entropy component ꢀ
TꢀS ranged from 11,026 to 3,487 cal/mol, roughly a six log unit difference. In sharp contrast,
K values ranged from 0.005 µM to 1.1 µM, only three log units. All amino pyrrolidine EED
D
compounds studied had favorable enthalpic contribution and unfavorable entropic contribution.
2
5
Enthalpyꢀentropy compensation can be clearly seen from the thermodynamic plot (Figure 9).
Notably most of these analogs were observed to be in the upper left and lower right quadrants
along the diagonal line in this figure. As compound optimization progressed, we observed
enhanced enthalpic contributions and reduced entropic contributions to the overall binding
2
9
affinity, as shown in Figure 9 from A to B to C. This finding was also illustrated in the direct
comparison of compounds 9, 23 and 24 (Table 3). The more than one log unit change in enthalpy
from compound 9 to 23 clearly showed the potency boost came mainly from the enthalpic
contribution. It is also interesting to note that although compounds 23 and 24 had similar K as
D
compound progressed, the thermodynamic signatures were very different. Compound 24 had
more favorable enthalpic contribution while compound 23 had much more favorable entropy
contribution which offset the enthalpy difference. Although no definite answers could be
ACS Paragon Plus Environment
14

Page 15 of 56
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
provided to explain the differences from Xꢀray crystal structures of these compounds alone, as
the hit to lead EED project progressed, we saw dramatically improved enthalpic contribution.
When R is A or B (Figure 2), the desired enantiomers with the (3R, 4S) configurations had both
2
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
more favorable enthalpic and entropic contributions, compared to the less potent enantiomers
with the (3S, 4R) configurations, where enthalpy played a major role in the observed potent
differences (compound 8ꢀ11 in Table 3). No attempts were made to explain the observed
thermodynamic signature differences of the enantiomers tested based on the xꢀray crystal
structures, although in theory it’s plausible, for example, as reported in the case of enantiomers
30
binding to acetylcholinesterase. When R is 7ꢀfluoroindanyl C, we also observed favorable
2
enthalpy change from compound 22 to 19 in Table 2. Enthalpy again largely contributed to the
observed potency difference, despite the unfavorable entropic change in this case. It seemed that
small chemistry structure and stereochemistry changes could alter the observed thermodynamic
signature significantly in these cases.
We discussed previously that the fast k and k rates of the EED HTS hits makes the hit
on
off
selection based on kinetic signatures at the hit to lead stage less attractive. In the same manner,
we also examined the thermodynamic signature of several of the EED HTS hits. It has been
recommended that hits with more enthalpic contributions should be prioritized for further
development. However, the challenges of doing this would be severalꢀfold based on our studies.
First, the potential requirement for analogs that are enantiomerically pure and of high purifies in
order to get reliable K values from ITC experiments are much less feasible in the hit to lead
D
stage where impurities may commonly present in the HTS hits and chirally pure compounds with
high ee values may not be readily available for screening. This means that false negative results
for hit confirmation will likely be common from ITC experiments. Secondly, slight
ACS Paragon Plus Environment
15

Journal of Medicinal Chemistry
Page 16 of 56
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
modifications to the chemical structure of the hits may cause significant change in the
thermodynamic signature, thus making selection of a hit as a chemical series to work on less
feasible. For example, by simply adding a dimethylamino group onto one of the HTS hits
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
(compound 18 to 25), we saw more than a four log unit change in both enthalpic and entropic
contributions, with similar binding affinities however. The outcome of the prioritization of hits
based on the thermodynamic signature would have been very different whether compound 18 or
25 was considered. Lastly, the lower throughput of ITC compared to other screening techniques
such as TRꢀFRET assay and SPR binding measurement would make this technology less
applicable for massive data collection in the hitꢀtoꢀlead stage.
CONCLUSION. Undeniably, binding kinetics and thermodynamics are intrinsically connected
with interactions between compounds and their protein targets. These interactions are essential
for the function of all biomolecules, presumably also eventually the development of effective
drugs with high efficacy and selectivity. The studies of the kinetic and thermodynamic signatures
of compounds interacting with their protein targets in the context of the chemical structures have
the potential to identify the structural features of the both the compound and the target protein
that influence the formation and dissociation of the compoundꢀprotein complexes, which in turn
aids the design of the compounds to maximize the compoundꢀprotein interaction.
In our study using the EED aminopyrrolidine series, the values of kinetic and thermodynamic
parameters covered a broad range, making SKR and STR studies possible. As the program
progressed, slower k rates were observed along with improved binding affinities. Both slower
off
koff rates and better binding affinity correlated with better cellular activities.
ACS Paragon Plus Environment
16

Page 17 of 56
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Our thermodynamic study of the EED chirally pure compounds showed that this series of
compounds all exhibited favorable enthalpic contributions and unfavorable entropic
contributions to binding affinity. The EED amino pyrrolidine series appeared to be largely
enthalpy driven, which accounted for 60ꢀ80% of the observed binding affinity. As the program
progressed, in general, we saw enthalpyꢀ entropy compensation phenomenon. Moreover, in stark
contrast to the view that ligands that have originated from medicinal chemistry optimization tend
to display increased affinities due to enhancement in entropic contributions, we observed
improved enthalpic contribution on our hitꢀtoꢀlead EED program.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
From our studies to incorporate thermodynamic and kinetic profiling on the EED hitꢀtoꢀlead
program, it became obvious that in order to evaluate the role of thermodynamic and kinetic data
in the hit and lead optimization process, the first step would be to increase the throughput of the
data generation. Biophysical technologies are now being increasingly applied to drug discovery
processes. They are incorporated as part of an integral process in the HTS hit characterization
3
1
and validation phase at many pharmaceutical companies. For example, SPR is one of the most
widely used biophysical techniques for obtaining binding affinities orthogonally, to confirm that
compound of interest binds directly to the protein target of interest. However, significant
challenges still exist however in getting SPR and ITC data routinely with acceptable cycle times
comparable to the traditional biochemical assays, especially in the case of ITC. More than 800
EED amino pyrrolidine compounds were synthesized and profiled in the TRꢀFRET assay with
fast turnaround times enabling rapid SAR iterations. However, less than 15% of these analogs
were profiled in SPR and ITC, due to the longer testing cycles needed for data collection and
processing, making the timely incorporation of this data in compound selection consideration
much less feasible. This in turn, limited the practical use of kinetic and thermodynamic
ACS Paragon Plus Environment
17

Journal of Medicinal Chemistry
Page 18 of 56
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
information for prospective design, compared to other parameters that are easily obtainable.
Significant instrumentation improvements have been seen in recent years. For example, biacore
8K can potentially characterize 64 interactions in 5 hours.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Our study also aimed to raise the awareness of the complexity of the ITC data interpretation.
When compounds tested contain several stereoisomers, which have different binding affinities to
the protein target tested, the apparent K determined from ITC experiments could be much
D
higher than the K for the individual stereoisomer. In some cases, this difference could appear
D
more than several thousand fold less potent. Most strikingly, we observed the presence of a small
percentage of another stereoisomer could alter the apparent K dramatically and mask the true
D
binding affinity of the compound of interest, which is not the case for TRꢀFRET and SPR
measurements. Although inꢀdepth study and more data are needed to elucidate the striking
differences observed, we suspect the prevalence of such phenomenon, especially when using ITC
for binding affinity determination in the early stage of the compound optimization process. Thus,
ITC experiments must be interpreted in the context of chiral purities of the compounds.
Our study also demonstrated the difficulty in utilizing kinetic and thermodynamic information in
hit selection and optimization. The prevalence of fast kinetics amongst the HTS hits made the
compounds of interest kinetically indistinguishable. We have also suggested that the
recommendation that a hit with more enthalpic binding affinity should be prioritized for hit
optimization should be revisited. Clearly, small changes in structures can have profound changes
in the enthalpic contribution.
We had attempted to interpret the observed SKR and STR based on high resolution Xꢀray crystal
structures. The effects of polar groups and hydrogen bond acceptors/donors as mentioned
ACS Paragon Plus Environment
18

Page 19 of 56
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
previsouly, as well as the positions of the various substituents and chirality in the context of the
Xꢀray crystal structures were investigated. However, only limited speculations can be made due
to the lack of the accurate information on compoundꢀprotein complex formation as well as
available computational tools. For example, it should be noted that the observed ꢀH and
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
calculated ꢀTꢀS are global parameters, containing a mix of many contributions. It is widely
acknowledged that the observed thermodynamic values reflect the sum of all coupled processes
accompanying the binding process including such as solvation and desolvation stages of the
3
2,33
compound, protein, and the complex formed.
These factors are not revealed by the static
high resolution xꢀray structures and may not be easily obtainable or identifiable. Our
understanding of the multiple factors that influence compoundꢀprotein interactions is still rather
rudimentary despite the enormous research efforts that went into the characterization of these
3
4, 35
factors.
Similarly, a major challenge with optimization of the kinetic signature of a
compound is the fundamental difficulty in characterizing transition states during the compound
protein complex formation. Overall, the complexity of the compoundꢀprotein interactions and
our limited knowledge in extricating the specific factors influencing kinetic and thermodynamic
readouts at the macroscopic level means more research is warranted in these fields for their
potential application in prospective design.
The dearth of the existing kinetic and especially thermodynamic data made the understanding of
factors that govern the data readout particularly challenging. The limited datasets reported in the
literature is evidently nowhere near the plethora of the available binding affinity data in
2
2,34
literature.
ACS Paragon Plus Environment
19

Journal of Medicinal Chemistry
Page 20 of 56
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
As such, it is our view that significant efforts are still needed to collect, analyze and understand
3
6
kinetic and thermodynamic data. Although both fields are still in their infancies with very few
successful examples showing the essential role of thermodynamic and kinetic profiling in
prospective design, the ultimate hope is that a multifaceted strategy will be developed to achieve
a complete understanding of the interplay between all of these parameters to facilitate rational
prospective molecule design in drug discovery.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
EXPERIMENTAL SECTION
Descriptions of EED TRꢀFRET assay, cellular assay and Xꢀray crystallography have been
1
7,19
reported previously.
ITC experiment. ITC experiments were carried out on the Microcal AutoꢀITC 200 system. The
calorimeter has a sample cell volume of 200 ul and a syringe volume of 40 ul. However the
fluidics system installed for automated sample handling requires a minimum of 370 ul for the
cell and 120 ul for the syringe. A temperature controlled tray holding 96 well plates is integral to
the instrument for storing samples; in this case at 10˚ C. Experiments were designed to enable
the robotic arm to transfer both ligand and protein from the tray which was kept in the chamber
at 10˚C. ITC runs were done at 25˚C using twenty 2 µl injections, each over 4 seconds with a
delay of 3 minutes between injections. Data analysis was performed using the accompanying
Origin software.
Protein preparation for the ITC experiments: Residues (77–441) of human EED were cloned into
pET28b vector and expressed in Escherichia coli. BL21(DE3)ꢀT1R strain. Hisꢀtagged protein
was purified by Ni–Sepharoseꢀ6 affinity chromatography and finally on Superdex 200 column
ACS Paragon Plus Environment
20

Page 21 of 56
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
(
GE Healthcare) equilibrated with 20 mM Tris, 150 mM NaCl, 1 mM TCEP, pH 8.0. This buffer
had been found optimal for EED long term stability from other studies and was also adopted for
ITC, in spite of its large value for heat of ionization. Protein was defrosted from ꢀ80˚C and
dialyzed against fresh column buffer overnight and concentrated next day on a 10,000 MWCO
membrane centrifugal concentrator. Protein concentration was determined by absorbance at 280
nm. The majority of titrations were done in the reverse mode with 200 µM EED in the syringe,
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
2
0 µM ligand in the cell. Ligands were soluble at this concentration. To confirm, some titrations
were also done with 20 µM protein in the cell and 200 µM ligand in the syringe, the forward
mode. No attempt was made to artificially set the stoichiometry to 1.0 during curve fitting, and it
was allowed to float during curve fitting with Origin.
SPR Binding Experiment. The binding kinetics of EED compounds were assayed via Surface
Plasmon Resonance (SPR) using a Biacore T200 instrument (GE Healthcare and the software
provided). Biotinylated EED (h)(78ꢀ441) was captured on neutravidin in flowcell 2. A blank
surface had neutravidin saturated with biotin in flowcell 1 and used as a reference surface during
binding assays. Briefly, the running buffer during neutravidin immobilization was HSBꢀP+
Buffer (10 mM Hepes, 150 mM NaCl, 0.05% (vol/vol) surfactant P20, pH 7.4) and the coupling
procedure was run at a flow rate of 5uL/min. A neutravidin coupling solution was prepared by
diluting neutavidin(Prod#31000 from Thermo Scientific) to 25 ug/mL in 10 mM acetate solution
(pH 5.5). Carboxyl groups on the dextran layer of the chip were activated by injecting a 1:1
mixture of 0.4 M NꢀethylꢀNꢀ(3dimethylaminopropyl)carbodiimide and 0.1 M Nꢀ
hydroxysuccinimide for 7 min. The neutravidin coupling solution was injected over the
activated chip surface for 300 sec at 10 ul/min to achieve a immobilization level of 14000
resonance units (RU). Remaining free activated carboxyl groups were blocked by injecting a
ACS Paragon Plus Environment
21

Journal of Medicinal Chemistry
Page 22 of 56
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
solution 1 M ethanolamine for 7 min. Binding affinities measurements were performed at flow
rate of 100 uL/min using 20 mM Phos, pH 6.5, 150 mM NaCl, 0.05% Tweenꢀ20, containing 1
mM DTT and 3 % DMSO. Compounds were assayed using multiꢀcycle kinetics mode as
provided by the Biacore T200 control software and were recorded at a frequency of 10 Hz. The
compounds were diluted in the running buffer and injected in a series of increasing
concentrations for contact time of 210 sec each and dissociation was monitored for up to 500 sec.
Sensograms were processed and analyzed using Biacore T200 evaluation software, the binding
curves were fit to determine the equilibrium dissociation constant (KD).
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Chemistry. Unless otherwise noted, all materials were obtained from commercial suppliers
and used without further purification. Flash chromatography was performed using a
Combiflash® Rf automated purification system. Preparative HPLC was performed on either an
automated preparativeꢀscale purification system equipped with a Waters Sunfire C8 5m column
(150 x 30 mm) or on a Phenomenex Luna C8 5m 100Å AXIA column (50mm × 21.2mm). A
gradient of acetonitrile (A) and 0.1% trifluoroacetic acid in water (B) was used, at a flow rate of
3
0 mL/min (0ꢀ0.5 min 5% A, 0.5ꢀ6.5 min linear gradient 5ꢀ100% A, 6.5ꢀ8.5 min 100% A, 8.5ꢀ
.0 min linear gradient 100ꢀ5% A, 9.0ꢀ10 min 5% A). Proton nuclear magnetic resonance spectra
9
1
(
H NMR, 500 or 400 MHz) were obtained in deuterodimethylsulfoxide (DMSOꢀd ) with
6
residual solvent as the internal standard unless otherwise noted. Mass spectra (MS) were
obtained by ionizing samples via positive electron spray ionization (ESI) or desorption chemical
ionization (DCI) with TOF as the mass analyzer. Preparative chiral SFC was performed on a
THAR/Waters SFC 80 system running under SuperChrom software control. The sample was
dissolved in a 8:1:1 mixture of methanol/dichloromethane/dimethylsulfoxide at a concentration
of 15 mg/mL. The sample was loaded into the modifier stream in 1 mL (15 mg) injections. The
ACS Paragon Plus Environment
22

Page 23 of 56
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
mobile phase was held isocratically at 30% methanol:CO . The instrument was fitted with a
2
Chiralpak IA column with dimensions 21 mm i.d. x 250 mm length with 5 µm particles. Chiral
purities were determined by chiral SFC columns. All compounds described and characterized
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
1
herein are of ≥95% purity as assessed by H NMR and LCꢀMS.
1
7,19
Detailed synthesis and characterization data of compounds 1ꢀ2 and 23ꢀ24 has been reported.
Other compounds 3ꢀ16 and 19ꢀ22 follow the similar synthetic routes as described for compound
2
.
1ꢀ(4ꢀ(4ꢀ(4ꢀ(dimethylamino)ꢀ1ꢀ(2ꢀfluoroꢀ6ꢀmethylbenzyl)pyrrolidinꢀ3ꢀyl)phenyl)piperazinꢀ
1
ꢀyl)ethanone (3). Step 1. A mixture of 1ꢀbenzylꢀ4ꢀ(4ꢀbromophenyl)pyrrolidinꢀ3ꢀamine (55 g,
1
66 mmol)17 and diꢀtertꢀbutyl dicarbonate (73 g, 332 mmol) were stirred in a round bottom flask
in 500 ml ethanol at room temperature overnight. The reaction mixture was concentrated and
purified by flash chromatography to give 35g tertꢀbutyl (1ꢀbenzylꢀ4ꢀ(4ꢀbromophenyl)pyrrolidinꢀ
3ꢀyl)carbamate with 50% isolated yield.
Step 2. To a solution of tertꢀbutyl (1ꢀbenzylꢀ4ꢀ(4ꢀbromophenyl)pyrrolidinꢀ3ꢀyl)carbamate
from step 1 (5 g, 11.6 mmol) in dichloroethane 60 ml was added triethylamine (7.0 g, 69.5
mmol) followed by 1ꢀchloroethyl carbonochloridate (9.9 g, 69.5 mmol). The mixture was stirred
at 80 °C overnight and extracted by dichloromethane followed by saturated NH Cl solution. The
4
combined organic phases were dried over Na SO and concentrated to give the crude material as
2
4
brown oil. This crude material was refluxed in 60 ml MeOH for 2 hr, concentrated and
partitioned between dichloromethane and pH = 1 HCl aqueous solution. After separation, the
aqueous phase was basified to pH = 9 with sodium bicarbonate and extracted with
ACS Paragon Plus Environment
23

Journal of Medicinal Chemistry
Page 24 of 56
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
dichloromethane to give tertꢀbutyl (4ꢀ(4ꢀbromophenyl)pyrrolidinꢀ3ꢀyl)carbamate(1.32 g, 3.9
mmol) with 33% yield after purification.
Step 3. To tertꢀbutyl (4ꢀ(4ꢀbromophenyl)pyrrolidinꢀ3ꢀyl)carbamate (800 mg, 2.3 mmol) from
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
step 2 in pH = 4 buffer solution in MeOH, sodium cyanoborohydride (221 mg, 3.52 mmol) and
2ꢀfluoroꢀ6ꢀmethylbenzaldehyde (356 mg, 2.58 mmol) were added and stirred at room
temperature for 2 hr. The reaction mixture was partitioned between water and dichloromethane.
The organic layer was concentrated and purified via flash chromatography to give tertꢀbutyl 4ꢀ
(4ꢀbromophenyl)ꢀ1ꢀ(2ꢀfluoroꢀ6ꢀmethylbenzyl)pyrrolidinꢀ3ꢀyl)carbamate as a white solid (870
mg, 1.88 mmol) with 80% isolated yield.
Step
4.
To
tertꢀbutyl
4ꢀ(4ꢀbromophenyl)ꢀ1ꢀ(2ꢀfluoroꢀ6ꢀmethylbenzyl)pyrrolidinꢀ3ꢀ
yl)carbamate obtained from step 3 (870 mg, 1.88 mmol) was added 4 ml dioxane and 3 ml 4N
HCl in dioxane. The mixture was stirred at room temperature for 2 hr and then concentrated
down to afford 4ꢀ(4ꢀbromophenyl)ꢀ1ꢀ(2ꢀfluoroꢀ6ꢀmethylbenzyl)pyrrolidinꢀ3ꢀamine as HCl salt
with quantitative yield. To this material, 8 ml pH=4 buffer in MeOH, formaldehyde (0.7 ml, 9.3
mmol) and sodium cyanoborohydride (176 mg, 2.8 mmol) were then added. The reaction
mixture was stirred at room temperature for 3 hr then partitioned between dichloromethane and
saturated sodium bicarbonate solution. The organic layer was separated, dried and concentrated
down to give 4ꢀ(4ꢀbromophenyl)ꢀ1ꢀ(2ꢀfluoroꢀ6ꢀmethylbenzyl)ꢀN,Nꢀdimethylpyrrolidinꢀ3ꢀamine,
which was used in the next step without further purification.
Step 5. A microwave vial was charged with 4ꢀ(4ꢀbromophenyl)ꢀ1ꢀ(2ꢀfluoroꢀ6ꢀmethylbenzyl)ꢀ
N,Nꢀdimethylpyrrolidinꢀ3ꢀamine (60 mg, 0.15 mmol), Pd (dba) (14 mg, 0.015mmol), 2ꢀ
2
3
(dicyclohexylphosphine)ꢀ2',4',6'ꢀtriꢀisopropylbiphenyl (14.6 mg, 0.031 mmol) and sodium tertꢀ
ACS Paragon Plus Environment
24

Page 25 of 56
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
butoxide (22 mg, 0.23 mmol) under nitrogen in 0.6 ml dioxane. This mixture was heated at 110
°C for 0.5 hr and then diluted with ethyl acetate followed by filtration through a pad of celite.
The filtrate was concentrated and dissolved in 1.5 ml DMSO/MeOH (1:1). Preparative HPLC
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
1
purification afforded compound 3 as TFA salts with 78% isolated yield. H NMR (500 MHz,
DMSOꢀd6) δ 7.30 (m, 2H), 7.12 – 7.00 (m, 3H), 6.95 (d, J = 8.8 Hz, 2H), 3.74ꢀ4.0 (m, 4H), 3.57
(
m, 6H), 3.11 (m, 5H), 2.71 (s, 6H), 2.70 (m, 1H), 2.43 (s, 3H), 2.04 (s, 3H). MS (ESI) m/z =
439 (M+H)+.
1
ꢀ(4ꢀ(4ꢀ((ꢀ(dimethylamino)ꢀ1ꢀ(7ꢀfluoroꢀ2,3ꢀdihydroꢀ1Hꢀindenꢀ1ꢀyl)pyrrolidinꢀ3ꢀ
yl)phenyl)piperazinꢀ1ꢀyl)ethanone (4). Compound 4 was synthesized using the similar
synthetic route as for compound 3, substituting 7ꢀfluoroꢀ2,3ꢀdihydroꢀ1Hꢀindenꢀ1ꢀone for 2ꢀ
fluoroꢀ6ꢀmethylbenzaldehyde. The compound was an approx. 67:33 mixture of epimers as
1
1
assessed by H NMR: H NMR (400 MHz, DMSOꢀd6) δ 7.31 –7.21 (m, 1H), 7.16 (dd, J = 8.8,
2.3 Hz, 2H), 7.09 (d, J = 7.4 Hz, 1H), 6.95 (td, J = 8.8, 5.2 Hz, 1H), 6.86 (dd, J= 8.7, 3.3 Hz,
2
H), 4.18 (dd, J = 7.0, 1.7 Hz, 0.33H), 4.19 (dd, J = 6.9, 1.7 Hz, 0.67H), 3.61ꢀ3.50 (m, 4H), 3.15
–
2.92 (m, 6H), 2.89ꢀ2.73 (m, 4H), 2.65ꢀ2.52 (m, 1H), 2.48 – 2.35 (m, 1H), 2.34 – 2.11 (m, 1H),
+
2.14 – 1.93 (m, 7H), 1.88 (s, 3H). MS (ESI) m/z 451 (M+H) .
1
ꢀ(2ꢀbromoꢀ6ꢀfluorobenzyl)ꢀ4ꢀ(3,4ꢀdimethoxyphenyl)ꢀN,Nꢀdimethylpyrrolidinꢀ3ꢀamine
5). Step 1. To a solution of (E)ꢀ1,2ꢀdimethoxyꢀ4ꢀ(2ꢀnitrovinyl)benzene (40 g, 191 mmol) in
dichloromethane 500 ml was added Nꢀbenzylꢀ1ꢀmethoxyꢀNꢀ
(trimethylsilyl)methyl)methanamine (54.5 g, 229 mmol). The resulting solution was cooled to ꢀ
(
(
o
1
0 C using saltꢀice bath and then TFA (1.473 mL, 19.12 mmol) was added dropwise (Caution:
o
severely exothermic). After addition, the reaction was stirred at ꢀ10 C for 1 hour, and then
allowed to warm to room temperature and stirred at this temperature for 16 hr. The contents were
ACS Paragon Plus Environment
25

Journal of Medicinal Chemistry
Page 26 of 56
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
then poured into saturated sodium bicarbonate solution. The organic layer was separated,
washed with water and brine respectively, and then concentrated under reduced pressure. The
residue was purified by flash column chromatography to afford 1ꢀbenzylꢀ3ꢀ(3,4ꢀ
dimethoxyphenyl)ꢀ4ꢀnitropyrrolidine (35 g, 102 mmol )as a yellow oil with 54% yield.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Step 2. To a solution of 1ꢀbenzylꢀ3ꢀ(3,4ꢀdimethoxyphenyl)ꢀ4ꢀnitropyrrolidine (60 g, 175
mmol) in EtOH 2000 ml was added Raney Ni 20g. The mixture was stirred under H and 15 psi
2
at 25 °C for 2 hr. The mixture was filtered and concentrated to give 1ꢀbenzylꢀ4ꢀ(3,4ꢀ
dimethoxyphenyl)pyrrolidinꢀ3ꢀamine as a yellow oil. The crude material was dissolved in 500 ml
EtOH. To this, Boc O (53.5 mL, 230 mmol) was added and stirred at 25 °C for 12 hr, The
2
mixture was concentrated down to give tertꢀbutyl (1ꢀbenzylꢀ4ꢀ(3,4ꢀdimethoxyphenyl)pyrrolidinꢀ
3ꢀyl)carbamate (20 g, 48.5 mmol) as a white solid with a 25% yield.
Step 3. a suspension of tertꢀbutyl (ꢀ1ꢀbenzylꢀ4ꢀ(3,4ꢀdimethoxyphenyl)pyrrolidinꢀ3ꢀ
yl)carbamate (20 g, 48.5 mmol) and Pd/C (5 g) in MeOH 1000 ml was stirred at 25 °C for 24 hr
under Hydrogen (15 psi). After filtration and solvent removal, The crude product was
recrystallized to give tertꢀbutyl (4ꢀ(3,4ꢀdimethoxyphenyl)pyrrolidinꢀ3ꢀyl)carbamate (13.5 g, 41.9
mmol) as white solid with 86% yield.
Step 4. To a solution of tertꢀbutyl (4ꢀ(3,4ꢀdimethoxyphenyl)pyrrolidinꢀ3ꢀyl)carbamate (35 mg,
0
.11mmol) in 1 mL pH=4 buffer solution in MeOH, 2ꢀbromoꢀ6ꢀfluorobenzaldehyde (33 mg,
.16 mmol) and sodium cyanoborohydride (6.4 mg, 0.16 mmol) were added. The mixture was
0
stirred at room temperature for 4 hr. The reaction mixture is partitioned between
dichloromethane and saturated sodium bicarbonate solution. The organic layer was separated,
dried and concentrated down. To the crude material thus obtained, 1 ml dioxane and 0.5 ml 4 N
ACS Paragon Plus Environment
26

Page 27 of 56
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
HCl in dioxane were added. After 2 hr at room temperature, the mixture was dried down and 1
ml pH=4 buffer in MeOH, formaldehyde (0.04 ml, 0.55 mmol) and sodium cyanoborohydride
(6.4 mg, 0.16 mmol) were added. The reaction mixture was stirred at room temperature for 3 hr,
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
then partitioned between dichloromethane and saturated sodium bicarbonate solution. The
organic layer was separated, dried and concentrated down. The crude material was purified by
reverse phase HPLC to give 11 mg of 1ꢀ(2ꢀbromoꢀ6ꢀfluorobenzyl)ꢀ4ꢀ(3,4ꢀdimethoxyphenyl)ꢀ
1
N,Nꢀdimethylpyrrolidinꢀ3ꢀamine as a TFA salt with 11% overall yield. H NMR (500 MHz,
Pyridineꢀd ) δ 7.48 – 7.42 (m, 1H), 7.19 – 7.06 (m, 3H), 7.01 (dd, J = 8.2, 2.1 Hz, 1H), 6.92 (d, J
5
=
8.3 Hz, 1H), 3.91 (q, J = 6.1 Hz, 1H), 3.81 – 3.66 (m, 9H), 3.28 – 3.18 (m, 3H), 2.78 (s, 6H),
+
2
.70 (dd, J = 9.1, 5.7 Hz, 1H). MS (ESI) m/z 439 (M+H) .
4ꢀ(3,4ꢀdimethoxyphenyl)ꢀ1ꢀ(2ꢀfluoroꢀ6ꢀmethylbenzyl)ꢀN,Nꢀdimethylpyrrolidinꢀ3ꢀamine
(6). Compound 6 was synthesized using the similar synthetic route as for compound 5,
1
substituting 2ꢀfluoroꢀ6ꢀmethylbenzaldehyde for 2ꢀfluoroꢀ6ꢀbromobenzaldehyde. H NMR (500
MHz, Pyridineꢀd ) δ 7.21 – 7.15 (m, 1H), 7.13 (d, J = 2.1 Hz, 1H), 7.04 – 6.96 (m, 3H), 6.92 (d,
5
J = 8.3 Hz, 1H), 3.96 (dt, J = 7.6, 5.1 Hz, 1H), 3.77 (m, 7H), 3.68 – 3.60 (m, 2H), 3.29 – 3.12
+
(
m, 3H), 2.79 (s, 6H), 2.62 (dd, J = 9.1, 6.2 Hz, 1H), 2.41 (s, 3H). MS (ESI) m/z 373 (M+H) .
4ꢀ(3,4ꢀdimethoxyphenyl)ꢀ1ꢀ(7ꢀfluoroꢀ2,3ꢀdihydroꢀ1Hꢀindenꢀ1ꢀyl)ꢀN,Nꢀdimethylpyrrolidinꢀ
3ꢀamine (7). Compound 7 was synthesized using the similar synthetic route as for compound 5,
substituting 7ꢀfluoroꢀ2,3ꢀdihydroꢀ1Hꢀindenꢀ1ꢀone for 2ꢀfluoroꢀ6ꢀbromobenzaldehyde. The
1
1
compound was an approx. 66:34 mixture of epimers as assessed by H NMR: H NMR (400
MHz, DMSOꢀd ) δ 7.26 (tdd, J = 7.6, 5.2, 2.1 Hz, 1H), 7.10 (d, J = 7.3 Hz, 1H), 6.96 (dd, J =
6
12.4, 5.5 Hz, 2H), 6.80 (d, J = 3.1 Hz, 2H), 4.25 (m, 1H), 3.73 – 3.67 (m, 6H), 3.10 – 2.95 (m,
ACS Paragon Plus Environment
27

Journal of Medicinal Chemistry
Page 28 of 56
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
2
H), 2.95 – 2.74 (m, 4H), 2.69 – 2.51 (m, 1H), 2.50 – 2.39 (m, 1H), 2.32 – 2.14 (m, 1H), 2.14 –
+
1.93 (m, 7H). MS (ESI) m/z 385 (M+H) .
(3S,4R)ꢀ1ꢀ(2ꢀbromoꢀ6ꢀfluorobenzyl)ꢀN,Nꢀdimethylꢀ4ꢀ(1ꢀmethylꢀ1Hꢀindolꢀ3ꢀyl)pyrrolidinꢀ
ꢀamine (8). The racemic compound 1ꢀ(2ꢀbromoꢀ6ꢀfluorobenzyl)ꢀN,Nꢀdimethylꢀ4ꢀ(1ꢀmethylꢀ
Hꢀindolꢀ3ꢀyl)pyrrolidinꢀ3ꢀamine (compound 15) was synthesized using the similar synthetic
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
3
1
route as for compound 5, substituting (E)ꢀ3ꢀ(2ꢀnitrovinyl)ꢀ1Hꢀindole for (E)ꢀ1,2ꢀdimethoxyꢀ4ꢀ(2ꢀ
nitrovinyl)benzene. The racemic material (594 mg of compound 15) was chirally separated on a
Chiralpak column (retention time = 6.7 min) to give 215 mg (3S,4R)ꢀ1ꢀ(2ꢀbromoꢀ6ꢀ
1
fluorobenzyl)ꢀN,Nꢀdimethylꢀ4ꢀ(1ꢀmethylꢀ1Hꢀindolꢀ3ꢀyl)pyrrolidinꢀ3ꢀamine (ee = 100%). H
NMR (400 MHz, Chloroformꢀd) δ 7.91 (dt, J = 8.0, 1.0 Hz, 1H), 7.41 (dt, J = 8.0, 1.1 Hz, 1H),
7.32 – 7.20 (m, 2H), 7.16 – 7.07 (m, 2H), 7.02 (ddd, J = 9.5, 8.3, 1.2 Hz, 1H), 6.97 (s, 1H), 3.88
(d, J = 2.7 Hz, 2H), 3.74 (s, 3H), 3.48 (dt, J = 8.4, 5.0 Hz, 1H), 3.19 – 3.02 (m, 3H), 2.86 – 2.67
+
(m, 2H), 2.24 (s, 6H). MS (ESI) m/z 432 (M+H) .
(3R,4S)ꢀ1ꢀ(2ꢀbromoꢀ6ꢀfluorobenzyl)ꢀN,Nꢀdimethylꢀ4ꢀ(1ꢀmethylꢀ1Hꢀindolꢀ3ꢀyl)pyrrolidinꢀ
3
ꢀamine (9). The racemic compound 1ꢀ(2ꢀbromoꢀ6ꢀfluorobenzyl)ꢀN,Nꢀdimethylꢀ4ꢀ(1ꢀmethylꢀ
1Hꢀindolꢀ3ꢀyl)pyrrolidinꢀ3ꢀamine (compound 15) was synthesized using the similar synthetic
route as for compound 5, substituting (E)ꢀ3ꢀ(2ꢀnitrovinyl)ꢀ1Hꢀindole for (E)ꢀ1,2ꢀdimethoxyꢀ4ꢀ(2ꢀ
nitrovinyl)benzene. The racemic material (594 mg) was chirally separated on a Chiralpak
column (retention time = 7.5 min) to give 226 mg (3R,4S)ꢀ1ꢀ(2ꢀbromoꢀ6ꢀfluorobenzyl)ꢀN,Nꢀ
1
dimethylꢀ4ꢀ(1ꢀmethylꢀ1Hꢀindolꢀ3ꢀyl)pyrrolidinꢀ3ꢀamine (ee = 99%). H NMR (400 MHz,
Chloroformꢀd) δ 7.91 (dt, J = 8.0, 1.0 Hz, 1H), 7.41 (dt, J = 8.0, 1.1 Hz, 1H), 7.32 – 7.20 (m,
2H), 7.16 – 7.07 (m, 2H), 7.02 (ddd, J = 9.5, 8.3, 1.2 Hz, 1H), 6.97 (s, 1H), 3.88 (d, J = 2.7 Hz,
ACS Paragon Plus Environment
28

Page 29 of 56
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
2
H), 3.74 (s, 3H), 3.48 (dt, J = 8.4, 5.0 Hz, 1H), 3.19 – 3.02 (m, 3H), 2.86 – 2.67 (m, 2H), 2.24
+
(s, 6H). MS (ESI) m/z 432 (M+H) .
(3S,4R)ꢀ1ꢀ(2ꢀfluoroꢀ6ꢀmethylbenzyl)ꢀN,Nꢀdimethylꢀ4ꢀ(4ꢀ
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
(
methylsulfonyl)phenyl)pyrrolidinꢀ3ꢀamine (10). The racemic compound 1ꢀ(2ꢀfluoroꢀ6ꢀ
methylbenzyl)ꢀN,Nꢀdimethylꢀ4ꢀ(4ꢀ(methylsulfonyl)phenyl)pyrrolidinꢀ3ꢀamine was synthesized
using the similar synthetic route as for compound 5, substituting (E)ꢀ1ꢀ(methylsulfonyl)ꢀ4ꢀ(2ꢀ
nitrovinyl)benzene
for
(E)ꢀ1,2ꢀdimethoxyꢀ4ꢀ(2ꢀnitrovinyl)benzene
and
2ꢀfluoroꢀ6ꢀ
methylbenzaldehyde for 2ꢀbromoꢀ6ꢀfluorobenzaldehyde. The racemic material (70 mg) was
chirally separated on a RegisPack 250mm column (retention time = 2.9 min) to give 20 mg
(3S,4R)ꢀ1ꢀ(2ꢀfluoroꢀ6ꢀmethylbenzyl)ꢀN,Nꢀdimethylꢀ4ꢀ(4ꢀ(methylsulfonyl)phenyl)pyrrolidinꢀ3ꢀ
1
amine (ee = 100%). H NMR (400 MHz, Methanolꢀd ) δ 7.99 – 7.85 (m, 2H), 7.74 – 7.60 (m,
4
2
H), 7.21 (td, J = 7.9, 5.8 Hz, 1H), 7.04 (d, J = 7.6 Hz, 1H), 6.92 (t, J = 9.1 Hz, 1H), 4.14 – 3.96
(m, 1H), 3.71 (q, J = 7.4 Hz, 1H), 3.20 (m, 5H), 3.11 (s, 3H), 2.82 (s, 6H), 2.61 (t, J = 8.5 Hz,
+
1H), 2.46 (s, 3H). MS (ESI) m/z 392 (M+H) .
(3R,4S)ꢀ1ꢀ(2ꢀfluoroꢀ6ꢀmethylbenzyl)ꢀN,Nꢀdimethylꢀ4ꢀ(4ꢀ
(
methylsulfonyl)phenyl)pyrrolidinꢀ3ꢀamine (11). The racemic compound 1ꢀ(2ꢀfluoroꢀ6ꢀ
methylbenzyl)ꢀN,Nꢀdimethylꢀ4ꢀ(4ꢀ(methylsulfonyl)phenyl)pyrrolidinꢀ3ꢀamine was synthesized
using the similar synthetic route as for compound 5, substituting (E)ꢀ1ꢀ(methylsulfonyl)ꢀ4ꢀ(2ꢀ
nitrovinyl)benzene
for
(E)ꢀ1,2ꢀdimethoxyꢀ4ꢀ(2ꢀnitrovinyl)benzene
and
2ꢀfluoroꢀ6ꢀ
methylbenzaldehyde for 2ꢀbromoꢀ6ꢀfluorobenzaldehyde. The racemic material (70 mg) was
chirally separated on a RegisPack 250mm column (retention time = 4.6 min) to give 25 mg
(3R,4S)ꢀ1ꢀ(2ꢀfluoroꢀ6ꢀmethylbenzyl)ꢀN,Nꢀdimethylꢀ4ꢀ(4ꢀ(methylsulfonyl)phenyl)pyrrolidinꢀ3ꢀ
ACS Paragon Plus Environment
29