Heat shock protein 70 (Hsp70) is an important emerging cancer target whose inhibition may affect multiple cancer-associated signaling pathways and, moreover, result in significant cancer cell apoptosis. Despite considerable interest from both academia and pharmaceutical companies in the discovery and development of druglike Hsp70 inhibitors, little success has been reported so far. Here we describe structure-activity relationship studies in the first rationally designed Hsp70 inhibitor class that binds to a novel allosteric pocket located in the N-terminal domain of the protein. These 2,5'-thiodipyrimidine and 5-(phenylthio)pyrimidine acrylamides take advantage of an active cysteine embedded in the allosteric pocket to act as covalent protein modifiers upon binding. The study identifies derivatives 17a and 20a, which selectively bind to Hsp70 in cancer cells. Addition of high nanomolar to low micromolar concentrations of these inhibitors to cancer cells leads to a reduction in the steady-state levels of Hsp70-sheltered oncoproteins, an effect associated with inhibition of cancer cell growth and apoptosis. In summary, the described scaffolds represent a viable starting point for the development of druglike Hsp70 inhibitors as novel anticancer therapeutics.
Heat shock protein 70 (Hsp70) is an important emerging cancer target whose inhibition may affect multiple cancer-associated signaling pathways and, moreover, result in significant cancer cell apoptosis. Despite considerable interest from both academia and pharmaceutical companies in the discovery and development of druglike Hsp70 inhibitors, little success has been reported so far. Here we describe structure-activity relationship studies in the first rationally designed Hsp70 inhibitor class that binds to a novel allosteric pocket located in the N-terminal domain of the protein. These 2,5'-thiodipyrimidine and 5-(phenylthio)pyrimidine acrylamides take advantage of an active cysteine embedded in the allosteric pocket to act as covalent protein modifiers upon binding. The study identifies derivatives 17a and 20a, which selectively bind to Hsp70 in cancer cells. Addition of high nanomolar to low micromolar concentrations of these inhibitors to cancer cells leads to a reduction in the steady-state levels of Hsp70-sheltered oncoproteins, an effect associated with inhibition of cancer cell growth and apoptosis. In summary, the described scaffolds represent a viable starting point for the development of druglike Hsp70 inhibitors as novel anticancer therapeutics.
The
heat shock protein 70 (Hsp70) family members are powerful proteins
with major roles in malignancy, such as inhibition of apoptosis, induction
of resistance to chemotherapy, and regulation of the stability of
oncoproteins.[1−3] Specifically, Hsp70 expression blocks apoptosis at
several levels, and in this respect the chaperone inhibits key effectors
of the apoptotic machinery, and also facilitates proteasome-mediated
degradation of apoptosis-regulatory proteins. The contribution of
Hsp70 isoforms to tumorigenesis is mainly through their role as cochaperones
of heat shock protein 90 (Hsp90), a heat shock protein known to regulate
the transforming activities of several kinases and transcription factors.
In this process, Hsp70 initiates the association of the client protein
with Hsp90 through a bridging protein called HSP-organizing protein
(HOP). These biological functions propose Hsp70 as an important target
whose inhibition or downregulation may result in significant apoptosis
in a wide range of cancer cells and also in inhibition of signaling
pathways involved in tumorigenesis and metastasis. Indeed, simultaneous
silencing of Hsc70 or Hsp70 expression in humancolon cancer cell
lines induced proteasome-dependent degradation of Hsp90 onco-client
proteins, cell-cycle arrest, and tumor-specific apoptosis.[4] Importantly, silencing of Hsp70 isoforms in nontumorigenic
cell lines did not result in comparable growth arrest or induction
of apoptosis, indicating a potential therapeutic window for Hsp70
targeted therapies.The Hsp70’s are a family of highly
homologous proteins composed
of two functional domains: the N-terminal ATPase domain and the C-terminal
client protein-binding domain.[5,6] The unique interplay
between the two domains creates a ligand-activated, bidirectional
molecular switch. For example, ATP binding to the ATPase domain induces
a conformational change that is rapidly propagated to the C-terminal
and that results in accelerated client protein dissociation. Conversely,
client protein binding to the C-terminal domain of ATP-bound Hsp70
induces a conformational change that is propagated to the ATPase domain
and that results in a stimulation of the ATP hydrolysis rate. The
chaperoning activity of Hsp70 is further regulated by cochaperones
(e.g., Hsp40s, BAG, and Hsp110) that catalyze the interconversion
between the ATP- and ADP-bound states and thus regulate chaperone
function. Such structural regulation suggests that Hsp70 may be vulnerable
to most strategies that interfere with its flexibility.Much
effort has recently been dedicated toward the discovery of
Hsp70 inhibitors, and unsurprisingly, molecules from a number of chemical
classes have been reported to interact with Hsp70 through a variety
of modes (Figure 1).[7,8] A
few, such as 15-deoxyspergualin (1) and pifithrin-μ
(2-phenylethynesulfonamide) (2), are believed to target
the C-terminal of Hsp70,[9,10] whereas others, such
as dihydropyrimidines (i.e., 3 (MAL3-101)),[11] are thought to block J-domain-stimulated ATPase
activity of Hsp70. Compounds such as myricetin (4)[12] and 5 (MKT-077)[13] are proposed to interact with a pocket outside the nucleotide-binding
domain, whereas apoptozole (6) may bind to the ATP-binding
pocket of Hsp70.[14]
Figure 1
Chemical structure of
reported potential Hsp70 inhibitors.
Chemical structure of
reported potential Hsp70 inhibitors.The majority of these compounds were discovered in library
screens
that aimed to identify inhibitors of either the ATPase or the folding
capacity of yeast or bacterial Hsp70[2,7,8] or in the case of 6 a cell-based screen
of compounds capable of inducing apoptosis.[15]5 was discovered following optimization efforts[16] that had previously identified such rhodacyanine
dyes as possessing anticancer activity.[17] In the only reported rational design approach to develop Hsp70 inhibitors,
nucleotide mimetics such as the dibenzyl-8-aminoadenosine analogue 7 (VER-155008) were developed to bind into the N-terminal
ATP pocket of Hsp70.[18] While these molecules
are reported to elicit their effects through an Hsp70 mechanism, it
is likely that they also act on multiple other unrelated and as yet
unspecified mechanisms. Furthermore, these molecules have been hindered
by a nontractable structure–activity relationship (SAR), with
subtle changes resulting in drastic changes in activity. While these
molecules have been of some value as tool molecules to offer insight
into the consequences of pharmacological modulation of Hsp70, they
have limited potential to become useful drugs.At this point
in time it is fair to say that Hsp70 has proven to
be a very difficult target to drug. In contrast, Hsp90 has proven
highly amenable with numerous small-molecule ATP-competitive inhibitors
entering into the clinic.[19] In the case
of Hsp90, potent small-molecule inhibitors such as geldanamycin and
radicicol were known even before their precise mode of action was
determined. When X-ray crystal structures showed that they bound to
a clearly specified pocket (i.e., ATP pocket) and behaved as ATP-competitive
inhibitors, structure-based drug design became possible. Unfortunately,
no such molecules that could potentially guide and truly inspire the
development of Hsp70 inhibitors exist for Hsp70.The experiences
of scientists at Vernalis offer some further insight
into the difficulty of targeting Hsp70 and perhaps as to why no natural
product ATP-competitive inhibitor is known.[18,20] In the only rational approach reported to date, they designed a
series of adenosine analogues to act as direct ATP-competitive inhibitors.
However, despite obtaining molecules that bound Hsp70 with high affinity
(Kd = 50 nM), their cellular potency was
disappointing, and a number of reasons were given for this poor correlation.
Hsp70 has a high affinity for ADP (Kd =
0.11–0.5 μM) and coupled with high intracellular ATP
concentrations makes the prospect of competitive inhibition a daunting
task. Furthermore, the binding mode of ATP is such that important
polar contacts are made with its β- and γ-phosphate groups
that lie buried within a polar cavity. Attempts to mimic these interactions
have resulted in highly polar nondruglike molecules such as 7, which despite potent affinity (Kd = 0.3 μM) possess weak cellular activity. Combined, these
factors make the prospect of obtaining compounds with potent in vivo
activity low and make a strong case against the development of reversible
ATP-competitive inhibitors of Hsp70 as a viable therapeutic strategy.[21]We had sought alternative strategies toward
inhibiting Hsp70 which
would have the potential for potent in vivo activity. Specifically,
we had sought and subsequently identified another pocket, outside
of the active site.[22] This allosteric site,
located in the N-terminal domain, was not evident nor entirely predicted
by the available crystal structures of Hsp70 and has been recently
discovered by us through computational analyses.[22] We used the homology model to design ligands that could
bind to this allosteric pocket.[22] Here
we describe structure–activity investigations in this first
reported allosteric pocket inhibitor class. These ligands are 2,5′-thiodipyrimidine
and 5-(phenylthio)pyrimidine scaffold compounds that we show here
and in the accompanying paper[23] to be amenable
to extensive medicinal chemistry and to act on cancer cells through
an Hsp70-mediated mechanism.
Ligand Design and Computational Analysis
of Ligand–Hsp70
Interactions
There are several available crystal structures
of Hsp70, but most
capture a relatively closed nucleotide-binding domain.[24−27] In contrast, recent NMR techniques and molecular dynamics studies
suggest a model in which subdomains of the N-terminus are bridged
and in a close proximity only in the ATP-binding conformation, with
residues of these subdomains rearranging and the cleft opening upon
ADP binding.[28,29] To overcome these limitations
of the currently available crystal structures and to investigate pockets
other than the ATP-binding site, we recently developed a theoretical
model of humanHsp70 (hHsp70).[22] This nucleotide-free
structure of hHsp70 both captured the N-terminal binding cleft in
a conformation that is more open than that captured by X-ray and unveiled
a pocket that contains a potentially reactive cysteine residue (Cys267
in hHsp70) embedded inside the cavity (Figure 2a).[22] The newly identified allosteric
pocket (depicted by the green surface in Figure 2a,b) is located outside the nucleotide-binding domain and is flanked
by subregions IB and IIB of the nucleotide-binding domain (NBD) (Figure 2b).
Figure 2
Rational design of the allosteric pocket ligands. (a,b)
The homology-modeled
hHsp70 structure led to the identification of a novel allosteric pocket
(shown in green) located in the N-terminal domain of human Hsp70.[22] This pocket contains a potentially reactive
cysteine residue (position shown in red lettering in panel a) and
is located outside the ATP-binding site (shown in red in panel b).
(c) The geometry of the N-terminal allosteric pocket of hHsp70 as
predicted by SiteMap (shown in yellow) and residues lining the pocket
and their relative location (depicted as colored circles) are presented.
Ligands based on the 2,5′-thiopyrimidine and 5-(phenylthio)pyrimidine
chemical scaffolds were designed to fit into the allosteric pocket.
They adopt the conformation required for a proper fit and position
functionalities toward the shown pocket residues. An acrylamide positioned
toward Cys267 was also incorporated into these scaffolds.
Rational design of the allosteric pocket ligands. (a,b)
The homology-modeled
hHsp70 structure led to the identification of a novel allosteric pocket
(shown in green) located in the N-terminal domain of humanHsp70.[22] This pocket contains a potentially reactive
cysteine residue (position shown in red lettering in panel a) and
is located outside the ATP-binding site (shown in red in panel b).
(c) The geometry of the N-terminal allosteric pocket of hHsp70 as
predicted by SiteMap (shown in yellow) and residues lining the pocket
and their relative location (depicted as colored circles) are presented.
Ligands based on the 2,5′-thiopyrimidine and 5-(phenylthio)pyrimidine
chemical scaffolds were designed to fit into the allosteric pocket.
They adopt the conformation required for a proper fit and position
functionalities toward the shown pocket residues. An acrylamide positioned
toward Cys267 was also incorporated into these scaffolds.This cavity is larger and potentially more druglike
than the nucleotide-binding
pocket (depicted by the red surface in Figure 2b), because it contains a balanced number of hydrophilic and hydrophobic
residues (Figure 2c). At the lower part of
the allosteric site lies mainly a hydrophilic cavity containing Lys271
and Glu268 and further apart lined by Thr13, Thr14, Tyr15, Lys56,
Asp234, Asn235, Arg264, Thr265, and Arg269 (numbering as in hHsp70)
(site A, Figure 2c). Hydrophobic amino acids
such as Leu237 and Val238 also line site A. This subpocket also contains
the potentially reactive cysteine residue, Cys267, that could covalently
link to a ligand containing the appropriate Cys-reactive functionality,
such as an acrylamide.[30] Adjacent to site
A, and placed in the middle, is a larger cavity comprised of both
nonpolar and polar amino acid residues, such as Tyr41, Val59, Pro63,
Thr66, Asp69, Phe68, Arg72, Glu231, Val260, and Arg261 (site B, Figure 2c). Placed at the upper part of the binding site,
and also potentially providing interactions to a small-molecule ligand,
are Lys88, His89, Trp90, Pro91, and Phe92 (site C, Figure 2c).We analyzed the geometry and the environment
of the computationally
identified allosteric pocket to provide a starting point for the rational
design of ligands (Figure 2c). We concluded
that ligands built around the 2,5′-thiodipyrimidine and 5-(phenylthio)pyrimidine
scaffolds, both little explored chemical spaces, would adapt the necessary
conformation and provide the required attachment sites for ligand
functionalization (Figure 2c). Each designed
ligand used this pharmacophore template to attach X3–7 functionalities pointed toward several amino acids lining the Hsp70
pocket (Figure 2c).Because our homology
model may not entirely mimic the native protein
structure, we included an additional binding hook, i.e., an acrylamide
functionality, to probe possible covalent bond formation between the
inhibitor and Cys267 upon protein binding (Figure 2c). Due to the location of Cys267 deep inside the cavity,
such bond formation would be possible only after the ligand was inserted
into the pocket and has achieved a proper fit. By gaining affinity
through a covalent linkage in addition to enthalpy, we hoped to increase
the ligand’s apparent affinity for the protein in the event
the fit would be less than optimal. We reasoned that if the initial
ligand would be of sufficiently good fit, additional enthalpy could
be gained by properly modifying the appended functionalities, such
as we proceed to do here and in the adjoining paper.[23]
Chemistry
The synthesis of all designed compounds evaluated
in this study
is shown in Schemes 1–3 and described below. Reaction of 2-amino-4,6-dimethoxypyrimidine
(8a) with p-methoxybenzyl chloride ((PMB)Cl)
resulted in PMB-protected pyrimidine 9a in 95% yield
(Scheme 1). This
was iodinated with N-iodosuccinimide (NIS) to give 10a in 98%, which was further coupled to 4,6-diamino-2-mercaptopyrimidine
using CuI/neocuproine to give 2,5′-thiodipyrimidine 11a in 65% yield. Acetylation of the amino groups was accomplished with
Ac2O/DMAP at 110 °C for 2 h to give 12a in 91% yield. Selective removal of the PMB groups occurred by heating 12a in a 1:1 mixture of TFA/CHCl3 at 62 °C
for 24 h to give 13a in 95%. Fluorodediazoniation was
accomplished with NaNO2/HF/pyridine to give 14a in 54% yield, which was then reacted with a variety of amines to
yield 15a–c. Deacetylation of these
intermediates followed by reaction with acryloyl chloride resulted
in target compounds 17a–c. Derivatives 20a–c were prepared following a similar
route from 2-amino-4,6-diethoxypyrimidine (8b), which
was prepared by refluxing a mixture of 2-amino-4,6-dichloropyrimidine
and NaH in ethanol in 89% yield (Scheme 1).
Scheme 1
Scheme 3
The chemistry
used to prepare 27a–d is similar
and shown in Scheme 1. Derivative 10a was coupled with 4-amino-2-mercaptopyrimidine to give 21 in 92% yield. Acetylation followed by PMB removal resulted
in 23. Reaction with NaNO2/HF/pyridine gave
fluoro derivative 24, which was reacted with a variety
of amines to give 25a–d. Deacetylation
followed by reaction with acryloyl chloride resulted in target compounds 27a–d. Reaction of 27a with m-CPBA at −78 °C resulted in N-oxide 28. Reaction of 24 with NaOMe/MeOH
followed by deacetylation resulted in 29, which was reacted
with acryloyl chloride to give 30 (Scheme 2). Reaction of 11a with acryloyl chloride followed
by treatment with TFA resulted in 31 (Scheme 2).
Scheme 2
Thioethers shown in Scheme 3 were prepared from chloropyrimidine 32 or 38. Reaction of 32 with N-methylpiperazine or morpholine resulted in 33a and 33b, respectively. Iodination of these derivatives,
followed
by coupling with 3-aminothiophenol resulted in 35a and 35b, respectively. These compounds were then reacted with
a variety of unsaturated acid chlorides to give target compounds 36a,b and 37. Target compounds 42a–c and 43 were prepared
similarly starting from chloropyrimidine 38 (Scheme 3).
Biological Evaluation in
the Hsp70 Inhibitor Series
Recent studies suggest that, unlike
normal cells, cancer cells
express modified heat shock protein species characterized by both
enhanced cochaperone recruitment and distinct post-translational modifications.[31−33] Evidence also indicates that certain small molecules show preferential
binding to these tumor-modified heat shock proteins.[32] Our goal being to develop Hsp70 inhibitors for cancer treatment,
we accordingly employed here a testing strategy that used phenotypic
assays to read fingerprints of Hsp70 inhibition in tumor cells rather
than a classical recombinant protein-based approach (Figure 3).
Figure 3
Known Hsp70 functions used to design the biological testing
of
the Hsp70 inhibitors in cancer cells. Hsp90 in concert with Hsp70
maintains the transforming capacity of several oncoproteins whose
aberrant activity leads to increased cell proliferation and survival.
The function of the Hsp90 protein complex requires the HSP-organizing
protein (HOP), involved in the formation of the active chaperone complex.
Altering the formation of the Hsp90–HOP–Hsp70 complex
leads to degradation of the oncoprotein via a proteasome-mediated
pathway and is associated with cell growth inhibition and/or cell
death. To confirm an Hsp70-mediated mechanism, it is expected that
compounds will be active at similar concentrations in this battery
of assays.
Known Hsp70 functions used to design the biological testing
of
the Hsp70 inhibitors in cancer cells. Hsp90 in concert with Hsp70
maintains the transforming capacity of several oncoproteins whose
aberrant activity leads to increased cell proliferation and survival.
The function of the Hsp90 protein complex requires the HSP-organizing
protein (HOP), involved in the formation of the active chaperone complex.
Altering the formation of the Hsp90–HOP–Hsp70 complex
leads to degradation of the oncoprotein via a proteasome-mediated
pathway and is associated with cell growth inhibition and/or cell
death. To confirm an Hsp70-mediated mechanism, it is expected that
compounds will be active at similar concentrations in this battery
of assays.To design a tumor–Hsp70
tailored testing module, we took
advantage of known biological activities mediated by Hsp70 in cancer.
As mentioned, Hsp70 is both an antiapoptotic molecule[4] and a transforming protein that acts together with Hsp90
to regulate the altered activity of several cancer kinases.[1,2,34] Thus, our testing strategy incorporated
assays that measured induction of apoptosis (i.e., the ability to
activate caspase-3,7, Tables 1 and 2 and induce PARP cleavage, Table 3) and those that read a phenotypic outcome similar to that
of Hsp90 inhibitors (i.e., degradation of Hsp90/Hsp70-sheltered oncoproteins
such as HER2 and Raf-1 in the proper genetic background, Table 3) (see tests 1 and 2 in Figure 3). Because a compound that interferes with nucleotide binding
to the ATP-binding pocket of Hsp90, either competitively or allosterically,
could result in a phenotype closely resembling Hsp70 inhibition, we
also included an assay to test for possible direct binding to Hsp90
(Table 3).
Inhibition of growth
measured in
Kasumi-1 acute myeloid leukemia cells. Values are the mean ±
SEM.
Caspase-3,7 activation
measured
in MOLM13 acute myeloid leukemia cells. Values are the mean ±
SEM.
Table 2
Inhibition of growth measured in
Kasumi-1 acute myeloid leukemia cells. Values are the mean ±
SEM.
Caspase-3,7 activation
measured
in MOLM13 acute myeloid leukemia cells. Values are the mean ±
SEM.
Table 3
a
compd
HER2b
Raf-1b
cPARPb
growth inhibitionb
Hsp90 bindingc
4
>100
>100
>100
NA
13.5
PU24FCld
2
2.5
NA
2.9
0.66
17a
0.7
1.7
2.0
0.8
>500
17c
2.5
7.5
5
9
>250
20a
1
2.5
1.5
1.1
>500
20c
2
3.1
10
7
>500
27a
7.5
5
7.5
1.2
>500
27d
20
17
25
5
>500
31
5
3.5
7.5
3.2
>500
37
15
20
10
16.5
>500
42a
4
5
5
1.8
>500
42c
30
50
40
16.7
>500
44a
2.5
7.5
10
7.5
>500
45a
75
60
100
50
>500
All values are in micromolar units.
HER2 and Raf-1 steady-state levels,
PARP cleavage and inhibition of growth measured in SKBr3 breast cancer
cells.
Binding to SKBr3
cell extracts.
Structure
and activity reported
in ref (38).
Inhibition of growth
measured in
Kasumi-1 acute myeloid leukemia cells. Values are the mean ±
SEM.Caspase-3,7 activation
measured
in MOLM13 acute myeloid leukemia cells. Values are the mean ±
SEM.Inhibition of growth measured in
Kasumi-1 acute myeloid leukemia cells. Values are the mean ±
SEM.Caspase-3,7 activation
measured
in MOLM13 acute myeloid leukemia cells. Values are the mean ±
SEM.aAll values are in micromolar units.HER2 and Raf-1 steady-state levels,
PARP cleavage and inhibition of growth measured in SKBr3breast cancer
cells.Binding to SKBr3
cell extracts.Structure
and activity reported
in ref (38).HER2 is a key transforming protein
in the HER2-overexpressing breast
cancer cells SKBr3, and its degradation or inhibition is sufficient
to inhibit the growth of these cells. Thus, in addition to testing
the effect of these derivatives on HER2 expression, we also measured
their activity on cell growth (Table 3).[35,36] In addition, the compounds were further tested for their ability
to inhibit the growth of cancer cells of distinct genetic origin,
including the P-gp/MDR-1-expressing Kasumi-1 acute myeloid leukemia
cells (Table 1) (see test 3 in Figure 3).For select derivatives we probed their
ability to bind Hsp70 from
tumor extracts (Figure 4a) and from live cancer
cells (Figure 4b,c) and investigated their
ability to alter the Hsp70–HOP complex in cancer cells (Figure 4d; see test 4 in Figure 3).[5,6,32]
Figure 4
Designed ligands interact
specifically with Hsp70 in cancer cells.
(a) Streptavidin beads were incubated with the indicated concentrations
of 44b, 45b, and d-biotin, the
unbound agent was washed off, and the resulting beads carrying 44b, 45b, or d-biotin were probed with
SKBr3 cell extracts (500 μg). Hsp70 isolated on the beads was
probed by Western blot (WB). A representative blot is presented (top).
Blots were quantified by densitometry and values, in relative luminescence
units, graphed against the concentration of added biotinylated Hsp70
inhibitor (bottom). Results from three independent experiments were
graphed to determine the relative binding affinity (Kd) of 44b and 45b using equations
as implemented in the GraphPad Prism software. Key: points, mean;
bars, SD. (b) Cells were treated with the indicated concentrations
of 44b for 6 h prior to lysing and precipitation of protein
complexes on streptavidin beads. Beads were washed with high-salt
(1 M NaCl) buffer before elution of proteins on a denaturing gel and
silver staining. BB70 is an antibody specific for Hsp70. This antibody
also recognizes Grp78 and Grp75, the endoplasmic reticulum and the
mitochondrial Hsp70 paralogues, respectively. (c) Protein complexes
were isolated as indicated in (b) in cells pretreated with 17a. CP = chemical precipitation. (d) Analysis of the Hsp70–HOP
complex. SKBr3 cells were treated for 24 h with vehicle or indicated
concentrations of 20a. Upon cell lysing, Hsp70 complexes
were isolated with an anti-Hsp70 antibody (IP BB70) and analyzed by
WB. Specificity of binding was tested with a control IgG. Gels were
quantified by densitometry, values normalized to the control (vehicle
only treated cells), and data graphed against the concentration of 20a. Error bars represent the SEM (n = 2).
Designed ligands interact
specifically with Hsp70 in cancer cells.
(a) Streptavidin beads were incubated with the indicated concentrations
of 44b, 45b, and d-biotin, the
unbound agent was washed off, and the resulting beads carrying 44b, 45b, or d-biotin were probed with
SKBr3 cell extracts (500 μg). Hsp70 isolated on the beads was
probed by Western blot (WB). A representative blot is presented (top).
Blots were quantified by densitometry and values, in relative luminescence
units, graphed against the concentration of added biotinylated Hsp70
inhibitor (bottom). Results from three independent experiments were
graphed to determine the relative binding affinity (Kd) of 44b and 45b using equations
as implemented in the GraphPad Prism software. Key: points, mean;
bars, SD. (b) Cells were treated with the indicated concentrations
of 44b for 6 h prior to lysing and precipitation of protein
complexes on streptavidin beads. Beads were washed with high-salt
(1 M NaCl) buffer before elution of proteins on a denaturing gel and
silver staining. BB70 is an antibody specific for Hsp70. This antibody
also recognizes Grp78 and Grp75, the endoplasmic reticulum and the
mitochondrial Hsp70 paralogues, respectively. (c) Protein complexes
were isolated as indicated in (b) in cells pretreated with 17a. CP = chemical precipitation. (d) Analysis of the Hsp70–HOP
complex. SKBr3 cells were treated for 24 h with vehicle or indicated
concentrations of 20a. Upon cell lysing, Hsp70 complexes
were isolated with an anti-Hsp70 antibody (IP BB70) and analyzed by
WB. Specificity of binding was tested with a control IgG. Gels were
quantified by densitometry, values normalized to the control (vehicle
only treated cells), and data graphed against the concentration of 20a. Error bars represent the SEM (n = 2).For a specific Hsp70-mediated
biological effect it is expected
that compounds should act with a similar potency in the above-described
battery of assays (for example, the IC50 measured in the
client degradation assay should be near the IC50 measured
in Hsp70–HOP complex alteration and so on).[2,35,36] Inclusion in the testing module of three
different Hsp70-addicted cell lines, including one P-gp/MDR-1 expressing,
increased our likelihood to eliminate early on poorly permeable compounds
and those marred by potential for P-gp/MDR-1-induced resistance.
Structure–Activity
Relationship in the Hsp70 Inhibitor
Series
In addition to ligand design, we used the homology
model to investigate
structure–activity relationships in the designed Hsp70 inhibitor
series (Figure 5). To study potential ligand–protein
interactions, we docked each derivative into the allosteric site.
This strategy, combined with the phenotypic, target-derived biological
investigation (Figure 3), as described above,
was used to understand differences in tumorHsp70 inhibitory activity
among the designed derivatives.
Figure 5
Binding interactions of hHsp70 with derivative 17a, as predicted by Glide (Schrodinger LLC, New York). Hydrogen
bonds
are shown as dotted red lines, and the interaction distance is shown
by dotted purple lines.
Binding interactions of hHsp70 with derivative 17a, as predicted by Glide (Schrodinger LLC, New York). Hydrogen
bonds
are shown as dotted red lines, and the interaction distance is shown
by dotted purple lines.
Site C
Attached to ring A is substituent X7 pointing
toward the exit of the binding site (Figure 2c). At this position we could accommodate a variety of substituents
such as piperazine, morpholine, piperidine, pyrrolidine, methoxy,
and amino (Table 1). Methylpiperazine, however,
was preferred over all others (17a versus 17b, 17c, and 31; 20a versus 20b and 20c; 27a versus 27b, 27c, 27d, and 30; 36a versus 37). At physiological pH, methylpiperazine
is likely to be protonated, and this H is predicted by docking to
form a hydrogen bond interaction with the backbone carbonyl of His89
(Figure 5). Indeed, an approximately 10-fold
drop in activity was observed when the N-Me of methylpiperazine
was substituted with O (as in morpholine) or C (as in piperidine).
Additionally, when this amine was oxidized to the corresponding N-oxide, as in derivative 28, activity was
almost abolished, potentially due to the unfavorable interaction of
the N-oxide oxygen with the backbone carbonyl of
His89. The pyrrolidine-containing derivative 27d had
a poor aqueous solubility, which likely accounted for the erratic
cellular activity we observed for this compound.
Site B
Substituents X5 and X6 of ring A are placed
in the middle cavity of the binding site comprised
of both nonpolar and polar amino acid residues, as described above
(Figure 2c). Substituents at these positions,
such as methoxy, ethoxy, and methyl, were well tolerated (compare 17a (methoxy), 20a (ethoxy), and 42a (methyl)). Modeling predicts for such substituents a potential for
hydrophobic (alkyl portion of these substituents with Val59) and electrostatic
(oxygen of methoxy and ethoxy with Arg261) interactions (Figure 5), and all derivatives containing these substituents
had favorable biological activity. Modeling also indicated the potential
for steric clash for X5 and X6 substituents
of larger size, such as in 45a, and indeed, we observed
a substantial loss of activity for this derivative (Table 1).
Site A
At the bottom of the allosteric
pocket lies
Cys267, which is predicted to form a covalent bond with the acrylamide
positioned at X4 on ring B (Figure 5). The contribution of the acrylamide moiety toward covalent binding
with the protein can be seen by comparing compounds in which it is
attached to either a pyrimidine or phenyl ring. Since the two nitrogen
atoms of the pyrimidine ring render the ring more electron deficient,
acrylamide functions attached to it would be expected to be more active
Michael acceptors than in the corresponding phenyl compounds. Indeed,
a trend for increased activity was seen for the pyrimidine-based compounds
when compared to the similar phenyl derivatives (Table 1, 27a and 27b versus 36a and 37, respectively).In addition to covalent
Cys modification, the acrylamide substituent at position X4 is poised to form several interactions with the adjacent protein
residues (Figure 5). Specifically, the alkene
group is positioned to establish hydrophobic interactions with Leu237
and Val238. The carbonyl group is poised to form electrostatic interactions
with the guanidine group of Arg264, ε-NH2 of Lys271,
and backbone −NH of Glu268 and could potentially also form
hydrogen bonds with the surrounding residues.To further understand
the contribution of the acrylamide to binding,
we prepared several methyl-substituted acrylamide derivatives (Table 2). Each of these derivatives was found to be less
active than the unsubstituted acrylamide derivatives (36a versus 36b; 42a versus 42b and 42c). This decrease in reactivity can be explained
by both steric and electronic effects. The methyl group could act
to sterically hinder nucleophilic attack by the reactive cysteine
residue (Cys267). Furthermore, the methyl group is a mild electron-donating
group and would decrease the electrophilicity of the Michael acceptor.Ring B occupies the lower part of the binding site which also contains
Arg264. Thus, the aromatic rings that were designed to compose ring
B, such as phenyl and pyrimidine, are predicted to be stabilized by
cation−π interactions with the guanidine group of this
arginine residue (Figure 5). Ring B also orients
substituents toward several important pocket residues. Specifically,
at the lower part of the pocket are Asp234 and Glu268, which have
the potential to interact with substituents on ring B, such as those
located at position X3 (Figure 5). Accordingly, derivatives for which X3 was NH2 (i.e., 17a) were more potent than similar derivatives
having a hydrogen at this position (i.e., 27a) potentially
because NH2 could form hydrogen bonds with the carboxylate
groups of Asp234 and Glu268.
Evaluation of Selectivity
and Target-Mediated Cellular activity
From these studies,
derivatives 17a and 20a emerged as most
active (Tables 1 and 3). Next, we attached 17a to biotin
with two aims in mind, first, to demonstrate effective and selective
isolation of Hsp70 from cancer cells by 17a and, second,
to further validate that the acrylamide was not the sole contributor
to ligand–protein binding. Specifically, we have created two
derivatives of 17a by attaching to it linkers that allow
for biotin modification of 17a (Figure 4a). In derivative 44a, the linker was attached
from the piperazine ring and thus predicted to point outside the allosteric
pocket. Indeed, 44a retained biological activity similar
to that of 17a (Tables 1 and 3). In derivative 45a, the linker was
attached on ring A through substituent X5, thus at a position
that would create steric clashes between the substituent and Hsp70
(Figure 4a). Indeed, 45a was of
much reduced activity when compared to 17a (Tables 1 and 3). We next attached 44a and 45a to biotin and then immobilized, dose-dependently,
the resulting 44b and 45b biotin-labeled
ligands onto streptavidin-containing beads. We then used these beads
to isolate Hsp70 from cancer cell extracts.As expected from
a mode of binding in which enthalpy substantially
contributes to binding, efficient isolation of Hsp70 from a cancer
cell extract was observed for the beads containing derivative 44b but not 45b (Figure 4a). If binding were solely driven by covalent modification
through the acrylamide functionality, both 44b and 45b would interact equally well with Hsp70 and, thus, isolate
similar amounts of Hsp70 from the cell extract. In conclusion, a proper
fit into the protein pocket in addition to covalent modification was
important for the ability of these ligands to bind to and inhibit
Hsp70.To demonstrate the selectivity of interaction, we incubated
cancer
cells with 44b and then proceeded to isolate the captured
proteins on streptavidin beads. Upon washing the affinity-purified
complexes and running them on a denaturing gel, the selectivity of 44b toward Hsp70 was demonstrated by the presence of a single
band upon silver staining (Figure 4b). This
band as demonstrated by its size of 70 kDa, its ability to run at
a position identical to that of the protein isolated by an anti-Hsp70
antibody (Figure 4b, BB70), and its recognition
by an Hsp70-specific antibody (Figure 4a) and
mass spectrum analysis[22] corresponds to
Hsp70. Furthermore, preincubation of cells with soluble 17a prior to affinity purification led to a dose-dependent reduction
of immobilized Hsp70, demonstrating the specificity of the interaction
(Figure 4c). Altogether, these findings confirm
that the designed ligands act in cells by specifically binding to
Hsp70.When cancer cells were incubated with select derivatives
(i.e., 20a), we observed a dose-dependent alteration
in the formation
of Hsp70–HOP complexes in cancer cells (Figure 4d). As mentioned above, HOP is a cochaperone that bridges
Hsp70 and Hsp90 to form a megachaperone complex (see test 4 in Figure 3). This chaperone machinery regulates the stability
of several onco-client proteins, such as HER2 and Raf-1, whose degradation,
as caused by chaperone inhibition, was tested here (Table 3). Of importance, alteration in the Hsp70–HOP
complex formation occurred at the same low concentrations (IC50 = 1.73 μM, Figure 4d) where
we also observed degradation of HER2 and Raf-1 by this inhibitor (1
and 2.5 μM, respectively, Table 3), further
supporting the Hsp70-mediated mechanism of action of these ligands.We noted no effect on Hsp90 for the ligands reported here at concentrations
as high as 500 μM (Table 3). Specifically,
unlike the direct Hsp90 inhibitor PU24FCl,[38] a compound with cellular activity comparable to that of the most
active Hsp70 inhibitors described here, such as 17a and 20a (Table 3, HER2, Raf-1), the Hsp70
inhibitors failed to compete with a fluorescently labeled geldanamycin
derivative, GM-Cy3B[39] for Hsp90 binding
(Table 3). Geldanamycin is an Hsp90 inhibitor
that binds to the Hsp90 regulatory pocket[37] located in the N-terminal domain of Hsp90. The fluorescence polarization
based Hsp90 assay that incorporates GM-Cy3B is designed in such a
way that it probes not only for direct binders to the ATP pocket but
also for those compounds that may allosterically interfere with the
conformation of Hsp90 to block compound access to the ATP pocket.We also tested myricetin (4; Figure 1) in these assays (Table 3). Using
NMR techniques,[12] this compound was proposed
to interact with the bacterial Hsp70 homologue, DnaK, at a site potentially
close to that occupied by our designed Hsp70 inhibitors. 4 was also recently reported to inhibit the DnaK–DnaJ (bacterial
Hsp70–Hsp40) complex formation (IC50 = 14.5 μM).[12] We could however not measure Hsp70-mediated
biological effects for 4 in the SKBr3cancer cells at
concentrations as high as 100 μM, possibly due to the poor stability
of 4 in the cells. Indeed, LC/MS–MS analyses of
cellular extracts demonstrated rapid myricetin degradation with the
agent virtually undetectable after 10 min of incubation (not shown).
Interestingly, when tested for its potential to bind Hsp90, we found 4 to alter geldanamycin’s binding to Hsp90 (Table 3). Polyphenols such as 4 are recognized
for their high propensity to bind nonspecifically to proteins. As
such, caution must be taken when interpreting biological and biochemical
results derived through the use of such compounds.Next, we
tested select lead compounds in vitro for potential to
interfere with other proteins. At the physiologically relevant concentration
of 10 μM, derivative 17a was inert when tested
against the scanMAX 402 kinase panel (Figure 6). This panel contains a set of kinases covering the AGC, CAMK, CMGC,
CK1, STE, TK, TKL, lipid, and atypical kinase families, plus important
mutant forms. Developed by Ambit Biosciences, it employs proprietary
active-site-dependent competition binding assays to determine how
compounds bind to kinases. It is based on a competition binding assay
that quantitatively measures the ability of a compound to compete
with an immobilized, active-site-directed ligand and can be used in
detection of multiple inhibitor types (e.g., types I and II and non-ATP-competitive).[40,41]
Figure 6
Derivative 17a (at 10 μM) was tested in the
scanMAX screen (Ambit) against 402 kinases. The TREEspot interaction map for 17a is presented. Only c-Met (red
dot on the kinase tree) appears as a potential low-affinity kinase
hit of 17a. KINOMEscan’s selectivity
score (S) is a quantitative measure of compound selectivity.
It is calculated by dividing the number of kinases that bind to the
compound by the total number of distinct kinases tested, excluding
mutant variants. S(35) = (number of nonmutant kinases
with %Ctrl < 35)/(number of nonmutant kinases tested).
Derivative 17a (at 10 μM) was tested in the
scanMAX screen (Ambit) against 402 kinases. The TREEspot interaction map for 17a is presented. Only c-Met (red
dot on the kinase tree) appears as a potential low-affinity kinase
hit of 17a. KINOMEscan’s selectivity
score (S) is a quantitative measure of compound selectivity.
It is calculated by dividing the number of kinases that bind to the
compound by the total number of distinct kinases tested, excluding
mutant variants. S(35) = (number of nonmutant kinases
with %Ctrl < 35)/(number of nonmutant kinases tested).
Conclusions
In summary, we describe
here SAR studies in the first rationally
designed scaffolds with binding potential to a novel allosteric pocket
in humanHsp70. This site is located in the N-terminal domain of the
chaperone and was recently identified by our homology modeling studies.[22] It is not revealed by reported crystal structures
of Hsp70, as these capture the protein in a closed conformation in
which the reactive Cys267 is unavailable for ligand binding and the
pocket is narrow.[24−27] The designed ligands are of a yet unexplored chemical space based
on the 2,5′-thiodipyrimidine and 5-(phenylthio)pyrimidine scaffolds,
and thus, we are also first to report here a chemical strategy that
allows for the assembly of such compounds.The weight of the
evidence in this paper and in our previous report[22] indicates that these compounds are acting on
Hsp70 through binding to the allosteric pocket. Only if reasonably
correct could the homology model enable the rational design of a ligand
that, when incubated with the thousands of proteins expressed in a
cancer cell, substantially affinity purifies one, Hsp70. Seconded
by a biological investigation of these ligands based on specific modulation
of the target, Hsp70, in the context of a cancer cell, our data clearly
define the mode of action of these ligands, in the cancer cell, through
inhibition of Hsp70. From these studies, we identify derivatives 17a and 20a as being the most active of the series.
Addition of high nanomolar to low micromolar concentrations of these
agents to several cancer cells led to a reduction in the steady-state
levels of Hsp70–HOP–Hsp90 complex chaperoned oncoproteins,
an effect associated with inhibition of cell growth and apoptosis.Because Hsp70 contains an active cysteine embedded in the allosteric
pocket, the derivatives presented herein contain an acrylamide functionality
that may create a covalent bond with such a residue upon Hsp70 protein
binding. We have used several techniques to demonstrate the potential
formation of a covalent bond between the inhibitor and the protein.[22] First, when the biotinylated analogue of 17a, 44b, was incubated with cancer cells, we
noted a time-dependent progressive increase in the amount of immobilized
Hsp70. This profile is indicative of time-dependent covalent modification
of the protein and is specific for compounds such as 17a, where irreversible binding plays a role. Second, mass spectrum
analysis of the trypsin digest of the 44b–Hsp70
complex identified a major m/z peak
at 1867.915.[22] This corresponds to 44b bound to LRTAC267ERAK and confirms that the
site of interaction of 17a with Hsp70 is within site
1, where Cys267 is located and, furthermore, that 17a forms an irreversible bond with the protein upon binding. Third,
when we probed binding of 44b to both the WT Hsp70 protein
and the C267S mutant, we noted that high-salt, high-detergent conditions
eluted preferentially the Hsp70-C267S mutant protein over the WT form
from preformed Hsp70–44b complexes.There
is precedent for the use of an acrylamide “warhead”
in the development of several irreversible kinase inhibitors currently
in clinical trials for cancer.[42−46] There is, of course, a concern that such an entity could indiscriminately
react with non-target-related proteins, resulting in unwanted biological
effects. However, in spite of the presence of an acrylamide, derivatives
described here are not excessively reactive and, in addition, have
an appropriate fit in the active site of the target, anticipating
a favorable enthalpic effect. It is also important to note that, for
proteins with long half-lives, such as Hsp70,[47] irreversible inhibition is mechanistically advantageous as it confers
complete target inhibition until resynthesis of the target protein[48] and thus allows for less frequent dosing, which
may also tip the balance toward a better therapeutic index for such
compounds.In addition to providing both a novel pharmacophore
and medicinal
chemistry for its assembly, in this paper we describe a testing battery
for assessing Hsp70-mediated mechanisms in cancer cells and for evaluating
specific ligand action in cells through Hsp70 inhibition. Altogether,
these findings provide a novel blueprint for a cancer-oriented development
of Hsp70-directed ligands.To conclude, the series of novel
chemical entities reported herein
interact with a yet unexplored pocket in Hsp70. These compounds of
tractable SAR favorably permeate cancer cells and exhibit their biological
activity through inhibition of the tumorHsp70 species, validating
the scaffold as an important starting point for the development of
Hsp70 inhibitors with potential therapeutic applications. The accompanying
paper in this issue describes the synthesis and structure–activity
evaluation of inhibitors of this series that target the same allosteric
pocket of Hsp70 described herein but act through a reversible mechanism
of action.[23]
Experimental
Section
Chemistry
All reagents were purchased from either Aldrich
or Acros Organics and used without purification. All reactions were
performed under argon protection. NMR spectra were recorded on a Bruker
AV-III-500 MHz NMR spectrometer. Chemical shifts are reported in δ
values in parts per million downfield from TMS as the internal standard. 1H data are reported as follows: chemical shift, multiplicity
(s = singlet, d = doublet, t = triplet, q = quartet, br = broad, m
= multiplet), coupling constant (Hz), integration. 13C
chemical shifts are reported in δ values in parts per million
downfield from TMS as the internal standard. High-resolution mass
spectra were recorded on a Waters LCT Premier system. Low-resolution
mass spectra were obtained on a Waters Acquity Ultra Performance LC
instrument with electrospray ionization and an SQ detector. Analytical
HPLC was performed on a Waters Autopurification system with PDA, MicroMass
ZQ, and ELSD detectors. The purity of the title compounds used in
pharmacology testing was determined by HPLC–MS using the following
method: 10–12 min gradient on a Waters2525 binary gradient
pump of increasing concentrations of acetonitrile in water (5% →
95%) containing 0.1% formic acid with a flow rate of 1.2 mL/min and
UV detection at λ = 220 and 254 nm on an XBridge C18 150 mm
× 4.6 mm, 5 μm column. Title compounds used in pharmacology
testing were >95% pure. Analytical thin-layer chromatography was
performed
on 250 μM silica gel F254 plates. Preparative thin-layer
chromatography was performed on 1000 μM silica gel F254 plates. Flash column chromatography was performed employing 230–400
mesh silica gel. Solvents were HPLC grade. Myricetin was purchased
from Indofine Chemical Co. (Hillsborough, NJ). The synthesis of compounds 44a,b and 45a,b is
described in detail in the Supporting Information.
To a solution of 2-amino-4,6-dimethoxypyrimidine
(2.0 g, 12.9 mmol) (8a) in 20 mL of DMF at 0 °C
was added NaH (1.24 g, 51.5 mmol), and the mixture was stirred at
rt for 10 min. 4-Methoxybenzyl chloride (4.03 g, 25.7 mmol) was added,
and the mixture was stirred at rt overnight. The reaction was quenched
with methanol and solvent removed under reduced pressure. The residue
was dissolved in EtOAc, washed with brine, and dried over MgSO4. Solvent was evaporated under reduced pressure, and the residue
was purified by column chromatography (hexane/EtOAc, 4:1) to afford
4.8 g (95%) of 9a. 1H NMR (500 MHz, CDCl3): δ 7.26 (d, J = 8.4 Hz, 4H), 6.89
(d, J = 8.4 Hz, 4H), 5.47 (s, 1H), 4.78 (s, 4H),
3.85 (s, 6H), 3.80 (s, 6H). 13C NMR (125 MHz, CDCl3): δ 171.9, 161.5, 158.7, 130.9, 129.1, 113.8, 78.7,
55.2, 53.4, 48.3. MS (m/z): [M +
H]+ 396.3.
To a solution of 9a (4.8
g, 12.3 mmol) in 50 mL of acetonitrile was added NIS (4.13 g, 18.4
mmol), and the resulting solution was stirred at rt for 1 h. Solvent
was evaporated, and the residue was purified by column chromatography
(hexane/EtOAc, 4:1) to afford 6.3 g (98%) of 10a. 1H NMR (500 MHz, CDCl3): δ 7.18 (d, J = 8.6 Hz, 4H), 6.83 (d, J = 8.6 Hz, 4H),
4.71 (s, 4H), 3.89 (s, 6H), 3.79 (s, 6H). 13C NMR (125
MHz, CDCl3): δ 169.1, 160.9, 158.8, 130.5, 129.0,
113.9, 55.3, 54.7, 48.6, 43.9. MS (m/z): [M + H]+ 522.4.
A mixture of 10a (6.2 g,
11.9 mmol), 4,6-diamino-2-mercaptopyrimidine (1.7 g, 11.9 mmol), neocuproine
(0.538 g, 2.38 mmol), CuI (0.452 g, 2.38 mmol), and K2CO3 (3.3 g, 33.8 mmol) in 100 mL of DMSO was stirred at 120 °C
for 16 h. Solvent was removed under reduced pressure, and the residue
was purified by column chromatography (CH2Cl2/MeOH–NH3 (7 N), 20:1) to afford 4.2 g (65%) of 11a. 1H NMR (500 MHz, DMSO-d6): δ 7.18 (d, J = 8.6 Hz, 4H), 6.80
(d, J = 8.6 Hz, 4H), 5.09 (s, 1H), 4.68 (s, 4H),
4.40 (s, 4H), 3.79 (s, 6H), 3.74 (s, 6H). MS (m/z): [M + H]+ 536.5.
A solution of 11a (3.2 g,
6.0 mmol) and DMAP (0.037 g, 0.3 mmol) in 20 mL of acetic anhydride
was stirred at 110 °C for 2 h. Solvent was removed under reduced
pressure, and the residue was purified by column chromatography (hexane/EtOAc,
1:1) to afford 3.4 g (91%) of 12a. 1H NMR
(500 MHz, CDCl3/DMSO-d6): δ
8.25 (br s, 1H), 7.70 (br s, 2H), 7.18 (d, J = 10.0
Hz, 4H), 6.81 (d, J = 10.0 Hz, 4H), 4.70 (s, 4H),
3.78 (s, 6H), 3.74 (s, 6H), 2.10 (s, 6H). MS (m/z): [M + H]+ 620.4.
A solution of 12a (0.950
g, 1.5 mmol) in 20 mL of TFA/CHCl3 (1:1) was heated at
62 °C for 24 h. Excess TFA and solvent were removed under reduced
pressure, and the residue was purified by column chromatography (CH2Cl2/MeOH, 20:1) to afford 0.550 g (95%) of 13a. 1H NMR (500 MHz, DMSO-d6): δ 10.51 (br s, 2H), 8.37 (br s, 1H), 6.98 (s, 2H),
3.78 (s, 6H), 2.06 (s, 6H). 13C NMR (125 MHz, DMSO-d6): δ 170.9, 170.1, 169.2, 162.5, 158.9,
92.8, 78.5, 53.9, 24.1. MS (m/z):
[M + H]+ 380.2.
13a (2.0 g, 5.3 mmol) was
added to a plastic tube fitted with a stir bar and cooled to 0 °C
followed by addition of a solution of HF/pyridine (3.6 mL, 144 mmol).
NaNO2 (0.545 g, 7.9 mmol) was added in portions over a
period of 20 min with stirring. The resulting solution was vigorously
stirred for an additional 50 min at 0 °C and 2 h at rt. CaCO3 (14.4 g, 144 mmol) was added to quench excess HF. The mixture
was extracted with CH2Cl2 and purified by column
chromatography (CH2Cl2/MeOH–NH3(7N), 20:1) to afford 1.1 g (54%) of 14a. 1H NMR (500 MHz, CDCl3): δ 8.45 (s, 1H), 7.61 (s,
2H), 4.00 (s, 6H), 2.18 (s, 6H). MS (m/z): [M + H]+ 383.2.
To a solution of 14a (30
mg, 0.078 mmol) in 2 mL of DMF was added 1-methylpiperazine (31 mg,
0.31 mmol), and the resulting solution was heated at 90 °C for
1 h. Solvent and excess reagent were removed under reduced pressure,
and the residue was purified by column chromatography (CHCl3/MeOH–NH3 (7 N), 10:1) to yield 32 mg (90%) of 15a. 1H NMR (500 MHz, CDCl3): δ
8.36 (br s, 1H), 8.13 (br s, 2H), 3.88 (s, 6H), 3.87 (m, 4H), 2.46
(m, 4H), 2.35 (s, 3H), 1.99 (s, 6H). 13C NMR (125 MHz,
CDCl3): δ 172.2, 169.4, 168.9, 160.4, 158.9, 96.1,
95.9, 54.7, 54.3, 46.1, 43.8, 24.7. MS (m/z): [M + H]+ 463.2.
A mixture of 15a (50 mg,
0.108 mmol) and 1 N NaOH(aq) (2 mL) in 4 mL of methanol was stirred
at 60 °C for 1 h. Solvents were removed under reduced pressure,
and the residue was purified by preparatory TLC to afford 39 mg (95%)
of 16a. 1H NMR (500 MHz, CDCl3):
δ 5.19 (s, 1H), 4.48 (s, 4H), 3.88 (s, 6H), 3.87 (m, 4H), 2.48
(m, 4H), 2.35 (s, 3H). MS (m/z):
[M + H]+ 378.9.
To a solution of 14a (60
mg, 0.157 mmol) in 2 mL of DMF was added morpholine (54 mg, 0.620
mmol), and the resulting solution was heated at 90 °C for 1 h.
Solvent and excess reagent were removed under reduced pressure, and
the residue was purified by preparatory TLC (CHCl3/MeOH–NH3 (7 N), 10:1) to yield 56 mg (79%) of 15b. 1H NMR (500 MHz, CDCl3): δ 8.41 (br s, 1H),
7.82 (br s, 2H), 3.83 (s, 6H), 3.77 (m, 4H), 3.68 (m, 4H), 2.16 (s,
6H). HRMS (m/z): [M + H]+ calcd for C18H24N7O5S, 450.1560; found, 450.1548. HPLC: (a) H2O + 0.1% TFA,
(b) ACN + 0.1% TFA (5–95% ACN in 10 min), tR = 8.47 min.
A mixture of 15b (45 mg,
0.100 mmol) and 1 N NaOH(aq) (2 mL) in 4 mL of methanol was stirred
at 60 °C for 1 h. Solvents were removed under reduced pressure,
and the residue was purified by preparatory TLC to afford 35 mg (95%)
of 16b. 1H NMR (500 MHz, CDCl3):
δ 5.19 (s, 1H), 4.44 (s, 4H), 3.87 (s, 6H), 3.83 (m, 4H), 3.77
(m, 4H), 2.16 (s, 6H). HRMS (m/z): [M + H]+ calcd for C14H20N7O3S, 366.1348; found, 366.1361. HPLC: (a) H2O + 0.1% TFA, (b) ACN + 0.1% TFA (5–95% ACN in 10 min), tR = 6.45 min.
To a solution of 14a (60
mg, 0.157 mmol) in 2 mL of DMF was added piperidine (52 mg, 0.611
mmol), and the resulting solution was heated at 90 °C for 1 h.
Solvent and excess reagent were removed under reduced pressure, and
the residue was purified by preparatory TLC (CHCl3/MeOH–NH3 (7 N), 10:1) to yield 55 mg (78%) of 15c. 1H NMR (500 MHz, CDCl3): δ 8.32 (br s, 1H),
7.77 (br s, 2H), 3.88 (s, 6H), 2.16 (m, 6H), 1.54–1.70 (m,
6H). HRMS (m/z): [M + H]+ calcd for C19H26N7O4S, 448.1767; found, 448.1766. HPLC: (a) H2O + 0.1% TFA,
(b) ACN + 0.1% TFA (5–95% ACN in 10 min), tR = 10.13 min.
A mixture of 15c (55 mg,
0.123 mmol) and 1 N NaOH(aq) (2 mL) in 4 mL of methanol was stirred
at 60 °C for 1 h. Solvents were removed under reduced pressure,
and the residue was purified by preparatory TLC to afford 43 mg (96%)
of 16c. 1H NMR (500 MHz, CDCl3):
δ 5.18 (s, 1H), 4.42 (s, 4H), 3.88 (s, 6H), 3.80 (m, 4H), 1.60–1.71
(m, 6H). HRMS (m/z): [M + H]+ calcd for C15H22N7O2S, 364.1556; found, 364.1544. HPLC: (a) H2O + 0.1%
TFA, (b) ACN + 0.1% TFA (5–95% ACN in 10 min), tR = 7.43 min.
To a solution of 16c (46
mg, 0.127 mmol) and Et3N (130 mg, 1.30 mmol) in 4 mL of
anhydrous dioxane was added acryloyl chloride (119 mg, 1.30 mmol)
dropwise under a water bath. The resulting mixture was stirred at
rt for 12 h. Solvent was removed under reduced pressure, and the residue
was purified by preparatory TLC (CHCl3/MeOH–NH3 (7 N), 10:1) to afford 28 mg (52%) of 17c. 1H NMR (500 MHz, CDCl3/DMSO-d6): δ 9.33 (s, 1H), 7.09 (s, 1H), 6.47–6.35 (m,
2H), 5.71 (d, J = 9.3 Hz, 1H), 5.31 (br s, 2H), 3.88
(s, 6H), 3.81 (m, 4H), 1.69 (m, 2H), 1.62 (m, 4H). HRMS (m/z): [M + H]+ calcd for C18H24N7O3S, 418.1661; found, 418.1660.
HPLC: (a) H2O + 0.1% TFA, (b) ACN + 0.1% TFA (5–95%
ACN in 10 min), tR = 8.06 min.
2-Amino-4,6-diethoxypyrimidine
(8b)
To
a solution of 2-amino-4,6-dichloropyrimidine (1.0 g, 6.09 mmol) in
20 mL of absolute ethanol was added NaH (0.585 g, 24.39 mmol) at rt.
The mixture was stirred under reflux for 12 h. Solvent was removed
under reduced pressure, and the residue was dissolved in CH2Cl2 and washed with brine. Solvent was evaporated, and
the resulting solid was purified by column chromatography (hexane/EtOAc,
4:1) to afford 1.0 g (89%) of 8b. 1H NMR (500
MHz, CDCl3): δ 5.42 (s, 1H), 4.78 (br s, 2H), 4.24
(q, J = 7.1 Hz, 4H), 1.34 (t, J =
7.1 Hz, 6H). MS (m/z): [M + H]+ 183.9.
To a solution of 8b (1.00
g, 5.46 mmol) in 20 mL of DMF at 0 °C was added NaH (0.524 g,
21.83 mmol), and the resulting solution was stirred at rt for 10 min.
4-Methoxybenzyl chloride (1.88 g, 12.0 mmol) was added, and the mixture
was stirred at rt overnight. The reaction was quenched with ethanol,
and solvent was removed under reduced pressure. The residue was dissolved
in EtOAc, washed with brine, dried over MgSO4, and concentrated
to give a residue that was purified by column chromatography (hexane/EtOAc,
4:1) to afford 2.25 g (97%) of 9b. 1H NMR
(500 MHz, CDCl3): δ 7.18 (d, J =
8.1, 4H), 6.84 (d, J = 8.1, 4H), 5.38 (s, 1H), 4.70
(s, 4H), 4.27 (q, J = 7.1 Hz, 4H), 3.79 (s, 6H),
1.29 (t, J = 7.1 Hz, 6H). 13C NMR (125
MHz, CDCl3): δ 171.5, 161.4, 158.6, 130.9, 129.0,
113.7, 78.6, 61.8, 55.2, 48.1, 14.6. MS (m/z): [M + H]+ 424.2.
To a solution of 9b (2.2
g, 5.2 mmol) in 50 mL of acetonitrile was added NIS (1.7 g, 8 mmol),
and the resulting solution was stirred at rt for 1 h. Solvent was
removed under reduced pressure, and the residue was purified by column
chromatography (hexane/EtOAc, 4:1) to afford 2.75 g (96%) of 10b. 1H NMR (500 MHz, CDCl3): δ
7.17 (m, 4H), 6.84 (m, 4H), 4.68 (s, 4H), 4.34 (q, J = 7.1 Hz, 4H), 3.80 (s, 6H), 1.32 (t, J = 7.1 Hz,
6H). 13C NMR (125 MHz, CDCl3/DMSO-d6): δ 168.3, 160.4, 158.2, 130.1, 128.4, 113.2,
62.6, 54.8, 48.1, 44.2, 14.1. MS (m/z): [M + H]+ 550.1.
A mixture of 10b (2.75 g,
5.0 mmol), 4,6-diamino-2-mercaptopyrimidine (0.71 g, 5.0 mmol), neocuproine
(0.226 g, 1.0 mmol), CuI (0.190 g, 1.0 mmol), and K2CO3 (1.38 g, 10.0 mmol) in 60 mL of DMSO was stirred at 120 °C
for 16 h. Solvent was removed under reduced pressure, and the residue
was partially purified by column chromatography (CH2Cl2/MeOH–NH3 (7 N), 20:1) to afford 2.2 g (80%)
of impure 11b [MS (m/z): [M + H]+ 564.2], which was used without further purification
in the next step.
A solution of 11b (1.2 g,
2.19 mmol) and DMAP (0.013 g, 0.11 mmol) in 20 mL of acetic anhydride
was stirred at 110 °C for 2 h. Solvent was removed under reduced
pressure, and the residue was purified by column chromatography (hexane/EtOAc,
1:1) to afford 1.2 g (89%) of 12b. 1H NMR
(500 MHz, CDCl3): δ 8.22 (s, 2H), 7.21 (d, J = 8.5 Hz, 4H), 6.86 (d, J = 8.5 Hz, 4H),
4.70 (s, 4H) 4.32 (q, J = 7.1 Hz, 4H), 3.80 (s, 6H),
2.16 (s, 6H), 1.20 (t, J = 7.1 Hz, 6H). MS (m/z): [M + H]+ 648.1.
A solution of 12b (2.00 g,
3.09 mmol) in 20 mL of TFA/CHCl3 (1:1) was heated at 62
°C for 24 h. Excess TFA and solvent were removed under reduced
pressure, and the residue was purified by column chromatography (CH2Cl2/MeOH, 20:1) to afford 1.15 g (92%) of 13b. MS (m/z): [M + H]+ 407.8.
13b (1.5 g, 3.68 mmol) was
added to a plastic tube fitted with a stir bar and cooled to 0 °C
followed by addition of a solution of HF/pyridine (3.0 mL, 120 mmol).
After several minutes NaNO2 (0.380 g, 5.52 mmol) was added
in portions over a period of 20 min with stirring. The resulting solution
was vigorously stirred for an additional 50 min at 0 °C. CaCO3 (12.0 g, 120 mmol) was added to quench excess HF. The mixture
was extracted with CH2Cl2 and purified by column
chromatography (CH2Cl2/MeOH–NH3 (7 N), 20:1) to afford 0.76 g (46%) of 14b. 1H NMR (500 MHz, CDCl3): δ 8.47 (br s, 1H), 7.85
(br s, 2H), 4.44 (q, J = 7.1 Hz, 4H), 2.17 (s, 6H),
1.31 (t, J = 7.1 Hz, 6H). MS (m/z): [M + H]+ 411.3.
To a solution of 14b (0.165
g, 0.402 mmol) in 3 mL of DMF was added 1-methylpiperazine (400 mg,
4.4 mmol), and the resulting solution was heated to 90 °C for
1 h. Solvent was removed under reduced pressure, and the residue was
purified by column chromatography (CH2Cl2/MeOH–NH3 (7 N), 10:1) to yield 0.180 g (91%) of 18a. 1H NMR (500 MHz, CDCl3): δ 8.35 (br s, 1H),
8.06 (br s, 2H), 4.35 (q, J = 7.1 Hz, 4H), 3.82 (m,
4H), 2.43 (m, 4H), 2.36 (s, 3H), 2.15 (s, 6H), 1.26 (t, J = 7.1 Hz, 6H). MS (m/z): [M +
H]+ 491.2.
A mixture of 18a (0.130 g,
0.265 mmol) and 1 N NaOH(aq) (2 mL) in 7 mL of methanol was stirred
at 60 °C for 1 h. Solvent was removed under reduced pressure,
and the residue was purified by preparatory TLC (CH2Cl2/MeOH, 10:1) to afford 0.100 g (93%) of 19a. 1H NMR (500 MHz, CDCl3): δ 5.17 (br s, 1H),
4.48 (s, 4H), 4.34 (q, J = 7.1 Hz, 4H), 3.83 (m,
4H), 2.47 (m, 4H), 2.35 (s, 3H), 1.27 (t, J = 7.1
Hz, 6H). MS (m/z): [M + H]+ 407.1.
To a solution of 14b (165
mg, 0.402 mmol) in 3 mL of DMF was added morpholine (350 mg, 4.02
mmol), and the resulting solution was heated at 90 °C for 1 h.
Solvent and excess reagent were removed under reduced pressure, and
the residue was purified by preparatory TLC (CHCl3/MeOH–NH3 (7 N), 10:1) to yield 192 mg (81%) of 18b. MS
(m/z): [M + H]+ 478.1. 1H NMR (500 MHz, CDCl3): δ 8.35 (br s, 1H),
8.06 (br s, 2H), 4.34 (q, J = 7.0 Hz, 4H), 3.82–3.84
(m, 4H), 2.43–2.45 (m, 4H), 2.15 (s, 6H), 1.26 (t, J = 7.0 Hz, 6H).
A mixture of 18b (0.200 g,
0.42 mmol) and 1 N NaOH(aq) (2 mL) in 5 mL of methanol was stirred
at 60 °C for 1 h. Solvent was removed under reduced pressure,
and the residue was purified by chromatography (CH2Cl2/MeOH–NH3 (7 N), 10:1) to afford 0.153 g
(93%) of 19b. 1H NMR (500 MHz, CDCl3): δ 6.39 (br s, 1H), 4.31 (q, J = 6.2 Hz,
4H), 3.51–3.83 (m, 4H), 1.90–1.95 (m, 4H), 1.19 (t, J = 6.2 Hz, 6H). MS (m/z): [M + H]+ 394.1.
To a solution of 14b (165
mg, 0.402 mmol) in 3 mL of DMF was added piperidine (342 mg, 4.02
mmol), and the resulting solution was heated at 90 °C for 1 h.
Solvent and excess reagent were removed under reduced pressure, and
the residue was purified by preparatory TLC (CHCl3/MeOH–NH3 (7 N), 10:1) to yield 167 mg (87%) of 18c. 1H NMR (500 MHz, CDCl3): δ 8.60 (br s, 2H),
8.35 (br s, 1H), 4.34 (q, J = 7.1 Hz, 4H), 3.75 (m,
4H), 2.17 (s, 6H), 1.67 (m, 2H), 1.58 (m, 4H), 1.27 (t, J = 7.1 Hz, 6H). MS (m/z): [M +
H]+ 476.2.
A mixture of 18c (0.200 g,
0.42 mmol) and 1 N NaOH(aq) (2 mL) in 5 mL of methanol was stirred
at 60 °C for 1 h. Solvent was removed under reduced pressure,
and the residue was purified by chromatography (CH2Cl2/MeOH–NH3 (7 N), 10:1) to afford 0.156 g
(95%) of 19c. MS (m/z): [M + H]+ 392.2.
A solution of 21 (7.6 g, 14.6
mmol) in 50 mL of acetic anhydride was stirred at 120 °C for
1 h. Solvent was removed under reduced pressure, and the residue was
purified by column chromatography (CH2Cl2/MeOH,
100:0 to 90:10) to afford 7.37 g (93%) of 22. 1H NMR (500 MHz, CDCl3): δ 9.17 (s, 1H), 8.32 (d, J = 5.5 Hz, 1H), 7.76 (br s, 1H), 7.22 (d, J = 8.3 Hz, 4H), 6.85 (d, J = 8.3 Hz, 4H), 4.73 (s,
4H), 3.84 (s, 6H), 3.79 (s, 6H), 2.16 (s, 3H). MS (m/z): [M + H]+ 563.2.
23 (1.72 g, 5.32 mmol) was
added to a plastic tube fitted with a stir bar and cooled to 0 °C
followed by addition of a solution of HF/pyridine (3.6 mL, 144 mmol).
NaNO2 (0.545 g, 7.9 mmol) was added in portions over a
period of 20 min with stirring. The resulting solution was vigorously
stirred for an additional 50 min at 0 °C and 2 h at rt. CaCO3 (14.4 g, 144 mmol) was added to destroy excess HF. The mixture
was extracted with CH2Cl2 and purified by column
chromatography (CH2Cl2/MeOH, 20:1) to afford
0.99 g (57%) of 24. 1H NMR (500 MHz, CDCl3): δ 8.34 (d, J = 5.6 Hz, 1H), 7.82
(d, J = 5.6 Hz, 1H), 4.00 (s, 6H), 2.21 (s, 3H). 13C NMR (125 MHz, CDCl3): δ 173.2, 169.2,
162.8, 161.1, 159.0, 157.1, 105.9, 90.7, 55.8, 24.8. HRMS (m/z): [M + H]+ calcd for C12H13FN5O3S, 326.0723; found,
326.0732. HPLC: (a) H2O + 0.1% TFA, (b) ACN + 0.1% TFA
(5–95% ACN in 10 min), tR = 7.75
min.
To a solution of 24 (100 mg,
0.307
mmol) in MeOH (5 mL) was added NaOCH3 (36 mg, 0.675 mmol),
and the resulting solution was heated at 90 °C for 1 h. Then
1 N NaOH(aq) (2 mL) was added and heating continued at 60 °C
for 1 h. Solvent and excess reagent were removed under reduced pressure,
and the residue was purified by preparatory TLC (CHCl3/MeOH–NH3 (7 N), 10:1) to yield 68 mg (75%) of 29. 1H NMR (500 MHz, DMSO-d6): δ
7.77 (d, J = 5.8 Hz, 1H), 6.86 (br s, 2H), 6.11 (d, J = 5.8 Hz, 1H), 3.95 (s, 3H), 3.87 (s, 6H). MS (m/z): [M + H]+ 295.9.
To a solution of 11a (50 mg,
0.093 mmol) and Et3N (14 mg, 0.140 mmol) in CH2Cl2 (2 mL) was added acryloyl chloride (12.7 mg, 0.140
mmol) under a water bath. The resulting mixture was stirred at rt
for 2 h. Solvent was removed under reduced pressure, the residue was
dissolved in 1 mL of TFA/CHCl3 (1:1), and the resulting
solution was heated at 62 °C for 12 h. TFA and solvent were removed
under reduced pressure, and the residue was purified by preparatory
TLC (CH2Cl2/MeOH–NH3 (7 N),
10:1) to afford 18 mg (55%) of 31. 1H NMR
(500 MHz, CDCl3/MeOH-d4): δ
7.04 (s, 1H), 6.32–6.41 (m, 2H), 5.80 (dd, J = 10.3, 0.8 Hz, 1H), 3.88 (s, 6H). HRMS (m/z): [M + H]+ calcd for C13H16N7O3S, 350.1035; found, 350.1025. HPLC: (a)
H2O + 0.1% TFA, (b) ACN + 0.1% TFA (5–95% ACN in
12 min), tR = 5.92 min.
A 5 g (0.0286 mol) portion of 2-chloro-4,6-dimethoxypyrimidine
(32) and 7.93 mL (7.16 g, 0.0715 mol) of N-methylpiperazine in 22 mL of DMF were heated at 90 °C for 2.5
h. The reaction mixture was concentrated under reduced pressure, and
the residue was taken up into 200 mL of CH2Cl2. This was washed with brine (3 × 50 mL), dried over MgSO4, filtered, and concentrated to give 6.51 g (95%) of 33a. 1H NMR (500 MHz, CDCl3): δ
5.37 (s, 1H), 3.85 (s, 6H), 3.82 (m, 4H), 2.44 (m, 4H), 2.33 (s, 3H). 13C NMR (125 MHz, CDCl3): δ 172.0, 160.8,
77.8, 55.0, 53.4, 46.3, 43.7. MS (m/z): [M + H]+ 239.2.
To 1.59 g (6.67 mmol) of 33a in
40 mL of acetonitrile were added 1.80 g (8.01 mmol) of NIS and 0.771
mL (1.14 g, 10.0 mmol) of TFA, and the resulting solution was stirred
at rt for 1.5 h. The reaction mixture was concentrated to dryness,
and the residue was taken up into 100 mL of CH2Cl2, washed with 10% sodium thiosulfate (50 mL) and 5% NaHCO3 (3 × 50 mL), dried over MgSO4, filtered, and concentrated
to give a residue which was purified by column chromatography (CH2Cl2/MeOH–NH3 (7 N), 1:0 to 20:1)
to yield 2.06 g (86%) of 34a. 1H NMR (500
MHz, CDCl3): δ 3.93 (s, 6H), 3.82 (m, 4H), 2.47 (m,
4H), 2.36 (s, 3H). MS (m/z): [M
+ H]+ 365.1.
To a solution of 35a (50
mg, 0.138 mmol) and Et3N (140 mg, 1.38 mmol) in CH2Cl2 (2 mL) was added methacryloyl chloride (144
mg, 1.38 mmol) dropwise under a water bath. The resulting mixture
was stirred at rt for 16 h. Solvent was removed under reduced pressure,
and the residue was purified by preparatory TLC (CH2Cl2/MeOH, 20:1) to afford 38 mg (65%) of 36b. 1H NMR (500 MHz, CDCl3): δ 7.40–7.47
(m, 2H), 7.12–7.18 (m, 2H), 6.79–6.83 (m, 1H), 5.74
(d, J = 1.2 Hz, 1H), 5.43 (d, J =
1.2 Hz, 1H), 3.99 (m, 4H), 3.91 (s, 6H), 2.65 (m, 4H), 2.47 (s, 3H),
2.03 (s, 3H). 13C NMR (125 MHz, CDCl3): δ
171.8, 166.7, 160.1, 141.2, 139.5, 138.4, 129.4, 121.9, 119.9, 117.1,
116.9, 81.3, 54.72, 54.67, 45.9, 43.3, 19.0. HRMS (m/z): [M + H]+ calcd for C21H28N5O3S, 430.1913; found, 430.1902.
HPLC: (a) H2O + 0.1% TFA, (b) ACN + 0.1% TFA (40–95%
ACN in 10 min), tR = 3.02 min.
4-(4,6-Dimethoxypyrimidin-2-yl)morpholine
(33b)
A 5 g (0.0286 mol) portion of 32 and 6.25 mL (6.23
g, 0.0715 mol) of morpholine in 22 mL of DMF were heated at 90 °C
for 2.5 h. The reaction mixture was concentrated under reduced pressure,
and the residue was taken up into 200 mL of CH2Cl2. This was washed with brine (3 × 50 mL), dried over MgSO4, filtered, and concentrated to give 6.00 g (93%) of 33b. 1H NMR (500 MHz, CDCl3): δ
5.40 (s, 1H), 3.86 (s, 6H), 3.77 (m, 4H), 3.75 (m, 4H). 13C NMR (125 MHz, CDCl3): δ 172.0, 160.9, 78.2, 66.8,
53.5, 44.2. MS (m/z): [M + H]+ 226.2.
To 1.5 g (6.66 mmol) of 33b in 40 mL of acetonitrile
was added 1.80 g (8.00 mmol) of NIS, and the resulting solution was
stirred at rt for 2 h. The reaction mixture was concentrated to dryness,
and the residue was taken up into 100 mL of CH2Cl2, washed with 10% sodium thiosulfate (50 mL) and 5% NaHCO3 (3 × 50 mL), dried over MgSO4, filtered, and concentrated
to give a residue which was purified by column chromatography (hexane/EtOAc,
80:20) to yield 1.99 g (85%) of 34b. 1H NMR
(500 MHz, CDCl3): δ 3.92 (s, 6H), 3.71–3.79
(m, 8H). MS (m/z): 352.1 [M + H]+.
A 1.43 g (10 mmol) portion of 2-chloro-4,6-dimethylpyrimidine
(38) and 2.50 g (25 mmol) of N-methylpiperazine
in 20 mL of DMF were heated at 90 °C for 2.5 h. The reaction
mixture was concentrated under reduced pressure, and the residue was
taken up into 100 mL of CH2Cl2. This was washed
with brine (3 × 25 mL), dried over MgSO4, filtered,
and concentrated to give 1.92 g (93%) of 39a. MS (m/z): [M + H]+ 207.0.
To 1.61 g (7.8 mmol) of 39a in
40 mL of acetonitrile were added 3.7 g (16.4 mmol) of NIS and 2.4
mL (3.56 g, 31.2 mmol) of TFA, and the resulting solution was stirred
at rt for 17 h. The reaction mixture was concentrated to dryness,
and the residue was taken up into 150 mL of CH2Cl2, washed with 10% sodium thiosulfate (50 mL) and 5% NaHCO3 (3 × 50 mL), dried over MgSO4, filtered, and concentrated
to give a residue which was purified by column chromatography (CH2Cl2/MeOH, 1:0 to 10:1) to yield 2.32 g (90%) of 40a. 1H NMR (500 MHz, CDCl3): δ
3.82 (m, 4H), 2.52 (s, 6H), 2.43 (m, 4H), 2.33 (s, 3H). MS (m/z): [M + H]+ 332.9.
A mixture of 40a (350 mg,
1.05 mmol), 4-amino-2-mercaptopyrimidine (147 mg, 1.15 mmol), neocuproine
(47 mg, 0.21 mmol), CuI (40 mg, 0.21 mmol), and K2CO3 (1.46 g, 10.5 mmol) in 20 mL of DMF was stirred at 110 °C
for 16 h. Solvent was removed under reduced pressure, and the residue
was purified by column chromatography (CH2Cl2/MeOH, 100:0 to 90:10) to afford 261 mg (75%) of 41a. MS (m/z): [M + H]+ 332.4.
To a solution of 41a (50
mg, 0.151 mmol) and Et3N (152 mg, 1.51 mmol) in CH2Cl2 (2 mL) was added methacryloyl chloride (158
mg, 1.51 mmol) dropwise under a water bath. The resulting mixture
was stirred at rt for 16 h. Solvent was removed under reduced pressure,
and the residue was purified by preparatory TLC (CH2Cl2/MeOH, 20:1) to afford 60 mg (75%) of 42c. 1H NMR (500 MHz, CDCl3): δ 8.37 (d, J = 5.7 Hz, 1H), 8.08 (br s, 1H), 7.87 (d, J = 5.7 Hz, 1H), 5.85 (s, 1H), 5.60 (s, 1H), 3.95 (m, 4H), 2.52 (m,
4H), 2.45 (s, 6H), 2.38 (s, 3H), 2.04 (s, 3H). HRMS (m/z): [M + H]+ calcd for C19H26N7OS, 400.1920; found, 400.1909.
4-(4,6-Dimethylpyrimidin-2-yl)morpholine
(39b)
A 1.43 g (10 mmol) portion of 38 and 2.18 g (25 mmol)
of morpholine in 20 mL of DMF were heated at 90 °C for 2.5 h.
The reaction mixture was concentrated under reduced pressure, and
the residue was taken up into 100 mL of CH2Cl2. This was washed with brine (3 × 25 mL), dried over MgSO4, filtered, and concentrated to give 1.84 g (95%) of 39b. 1H NMR (500 MHz, CDCl3): δ
6.30 (s, 1H), 3.79 (m, 4H), 3.75 (m, 4H), 2.29 (s, 6H). 13C NMR (125 MHz, CDCl3): δ 167.1, 161.9, 109.4, 67.0,
44.3, 24.0. MS (m/z): [M + Na]+ 216.1.
To 1.50 g (7.8 mmol) of 39b in 40 mL of acetonitrile
were added 2.1 g (9.4 mmol) of NIS and 0.898 mL (1.33 g, 11.7 mmol)
of TFA, and the resulting solution was stirred at rt for 2 h. The
reaction mixture was concentrated to dryness, and the residue was
taken up into 100 mL of CH2Cl2, washed with
10% sodium thiosulfate (50 mL) and 5% NaHCO3 (3 ×
50 mL), dried over MgSO4, filtered, and concentrated to
give a residue which was purified by column chromatography (CH2Cl2/MeOH, 1:0 to 20:1) to yield 2.09 g (84%) of 40b. 1H NMR (500 MHz, CDCl3): δ
3.77 (m, 4H), 3.73 (m, 4H), 2.53 (s, 6H). MS (m/z): [M + Na]+ 341.9.
A mixture of 40b (1.0 g,
3.13 mmol), 4-amino-2-mercaptopyrimidine (0.477 g, 3.76 mmol), neocuproine
(0.140 g, 0.626 mmol), CuI (0.122 g, 0.626 mmol), and K2CO3 (0.864 g, 6.26 mmol) in 60 mL of DMF was stirred at
110 °C for 16 h. Solvent was removed under reduced pressure,
and the residue was purified by column chromatography (CH2Cl2/MeOH, 100:0 to 90:10) to afford 0.690 g (69%) of 41b. MS (m/z): [M + H]+ 319.1.
To a solution of 41b (50 mg,
0.157 mmol) and Et3N (79 mg, 0.786 mmol) in CH2Cl2 (2 mL) was added acryloyl chloride (71 mg, 0.786 mmol)
dropwise under a water bath. The resulting mixture was stirred at
rt for 16 h. Solvent was removed under reduced pressure, and the residue
was purified by preparatory TLC (CH2Cl2/MeOH–NH3 (7 N), 10:1) to afford 58 mg (68%) of 43. 1H NMR (500 MHz, CDCl3/MeOH-d4): δ 8.39 (d, J = 5.7 Hz, 1H), 6.49
(d, J = 16.9 Hz, 1H), 6.27 (dd, J = 16.9, 10.3 Hz, 1H), 5.86 (d, J = 10.3 Hz, 1H),
3.89 (m, 4H), 3.78 (m, 4H), 2.46 (s, 6H). 13C NMR (125
MHz, CDCl3): δ 172.2, 170.4, 164.6, 160.4, 159.0,
157.8, 130.1, 130.0, 109.2, 106.2, 66.8, 44.1, 23.7. MS (m/z): [M + H]+ 373.0.
Biological Testing
Cell Lines
SKBr3 cells were a gift from Dr. Neal Rosen
(Memorial Sloan-Kettering Cancer Center, MSKCC) and Kasumi-1 and MOLM-13
from Dr. S. Nimer (MSKCC). Cells were cultured routinely in DME/F12
(SKBr3) or in RPMI (Kasumi-1, MOML-13) supplemented with 10% fetal
bovine serum, 1% l-glutamine, 1% penicillin, and streptomycin.
Western Blotting
Cells were grown to 60–70%
confluence and treated with inhibitor or DMSO vehicle for the indicated
times. Protein lysates were prepared in 50 mM Tris, pH 7.4, 150 mM
NaCl, and 1% NP-40 lysis buffer. Protein concentrations were measured
using the BCA kit (Pierce) according to the manufacturer’s
instructions. Protein lysates (10–50 μg) were resolved
by SDS–PAGE, transferred onto a nitrocellulose membrane, and
incubated with the indicated primary antibodies: anti-HER2 from rabbit
(1:250, 28-0004, Zymed), anti-Hsp70 from mouse (1:500, SPA-810, Stressgen),
anti-Raf-1 from rabbit (1:500, sc-133, Santa Cruz), anti-PARP (p85
fragment) from rabbit (1:500, G7341, Promega), anti-HOP from mouse
(1:500, SRA-1500, Enzo) and anti-β-actin from mouse (1:2500,
A1978, Sigma-Aldrich). Membranes were then incubated with a corresponding
peroxidase-conjugated secondary antibody (1:3000 dilution).
Hsp90
Binding Assay
For the competition studies, fluorescence
polarization (FP) assays were performed as previously reported.[49] Briefly, FP measurements were performed on an
Analyst GT instrument (Molecular Devices, Sunnyvale, CA). Measurements
were taken in black 96-well microtiter plates (Corning no. 3650) where
both the excitation and the emission occurred from the top of the
wells. A stock of 10 μM GM-cy3B was prepared in DMSO and diluted
with Felts buffer (20 mM HEPES (K), pH 7.3, 50 mM KCl, 2 mM DTT, 5
mM MgCl2, 20 mM Na2MoO4, and 0.01%
NP40 with 0.1 mg/mL BGG). To each 96-well plate were added 6 nM fluorescent
GM (GM-cy3B), 3 μg of SKBr3 lysate (total protein), and tested
inhibitor (initial stock in DMSO) in a final volume of 100 μL
of HFB buffer. Drugs were added in triplicate wells. For each assay,
background wells (buffer only), tracer controls (free, fluorescent
GM only), and bound GM controls (fluorescent GM in the presence of
SKBr3 lysate) were included on each assay plate. GM was used as a
positive control. The assay plate was incubated on a shaker at 4 °C
for 24 h, and the FP values (mP) were measured. The fraction of tracer
bound to Hsp90 was correlated to the FP value and plotted against
the values of competitor concentrations. The inhibitor concentration
at which 50% of bound GM was displaced was obtained by fitting the
data. All experimental data were analyzed using SOFTmax Pro 4.3.1
and plotted using Prism 4.0 (Graphpad Software Inc., San Diego, CA).
Chemical Precipitation in Cell Extracts
Protein lysates
were prepared using 20 mM Tris, pH 7.4, 25 mM NaCl, 0.1% NP-40 lysis
buffer. Streptavidin agarose beads (50 μL) (Thermo Scientific)
were washed three times with lysis buffer, 44b and 45b were added at the indicated concentrations, and the complexes
were incubated at 4 °C for 1 h. Upon a three-time wash with buffer,
the beads were added to the indicated total cellular protein in the
binding buffer. Samples were incubated at 4 °C overnight, washed
five times with the lysis buffer, and applied to SDS–PAGE (see
Figure 4a).
Chemical Precipitation
in a Live Cell
Cancer cells
were treated with 44b (25 μM) or d-biotin
for 6 h and then lysed in a buffer containing 20 mM Tris, pH 7.4,
25 mM NaCl, and 0.1% NP-40. Aliquots (500 μg total protein)
were incubated with streptavidin beads for 1 h at 4 °C. Purified
protein complexes were washed with 20 mM Tris, pH 7.4, 1 M NaCl, 0.1%
NP-40 (high salt) buffer and applied to SDS–PAGE. The gel was
stained with the SilverQuest staining kit (Invitrogen) (see Figure 4b).
Immunoprecipitation
Cancer cells
were collected and
lysed in a buffer containing 20 mM Tris, pH 7.4, 25 mM NaCl, and 0.1%
NP-40. The anti-Hsp70 antibody (BB70; 5 μL) was added to 500
μg of extract together with protein G agarose beads (30 μL)
(Upstate), and the mixture was incubated at 4 °C overnight. Samples
were washed with lysis buffer and applied to SDS–PAGE. The
gel was stained with the SilverQuest staining kit (Invitrogen) (see
Figure 4b).
Competition Assay
Cancer cells were pretreated for
2 h with increasing concentrations of 17a and then lysed
in a buffer containing 20 mM Tris, pH 7.4, 25 mM NaCl, and 0.1% NP-40.
Meanwhile, to prepare the 44b–streptavidin bead
conjugates, 44b (50 μM) was added to high-capacity
streptavidin agarose beads (Thermo Scientific) (30 μL per experimental
sample), incubated for 1 h at 4 °C, and then washed with lysis
buffer. Protein extracts (500 μg of total protein) were then
incubated overnight with the 44b–streptavidin
bead conjugate. The complexes were washed in high-salt buffer, applied
to SDS–PAGE, and stained with the SilverQuest staining kit
(see Figure 4c).
Hsp70–HOP Complex
Analysis
SKBr3 cells were treated with the indicated
concentrations of the inhibitor
for 24 h. Samples were collected and lysed in 20 mM Tris, pH 7.4,
25 mM NaCl, 0.1% NP-40 buffer with protease inhibitors added. Aliquots
of 500 μg of total protein adjusted to 100 μL with the
lysis buffer were prepared. Samples were incubated with 5 μL
of BB70 antibody (Stressmarq) or normal IgG (as a negative control)
and 20 μL of protein G agarose beads (Upstate) at 4 °C
overnight. Samples were washed five times with the lysis buffer and
applied to SDS–PAGE followed by a standard Western blotting
procedure to detect levels of HOP protein in the Hsp70 complexes upon
treatment.
Growth Inhibition Assay
We evaluated
the antiproliferative
effects of inhibitors using the dye Alamar Blue. This reagent offers
a rapid objective measure of cell viability in cell culture, and it
uses the indicator dye resazurin to measure the metabolic capacity
of cells, an indicator of cell viability. Briefly, cells were plated
on Costar 96-well plates. For attached cells (such as SKBr3), 8000
cells/well were used. For suspension cells (such as Kasumi-1), 20000
cells/well were plated. Cells were allowed to incubate for 24 h at
37 °C before drug treatment. Drugs were added in triplicate at
the indicated concentrations, and the plate was incubated for 72 h.
Alamar Blue (50 μM) was added and the plate read 6 h later using
the Analyst GT (fluorescence intensity mode, excitation 530 nm, emission
580 nm, with a 560 nm dichroic mirror). Results were analyzed using
the Softmax Pro software. The percentage cell growth inhibition was
calculated by comparing fluorescence readings obtained from treated
versus control cells, accounting for the initial cell population (time
zero). IC50 was calculated as the drug concentration that
inhibits cell growth by 50%.
Caspase-3,7 Activation.[50]
MOLM-13 cells (30000 cells/well) were plated in
black 96-well plates
(Corning no. 3603) in 40 μL of RPMI medium and left in an incubator
(37 °C, 5% CO2) for up to 24 h. Cells were treated
for 16 h with compounds or DMSO (control) at the desired concentrations
in 50 μL of medium. Drugs were added in triplicate wells. Following
exposure of cells to Hsp70 inhibitors, 50 μL of buffer containing
10 mM HEPES (pH 7.5), 2 mM EDTA, 0.1% CHAPS, and the caspase substrate
Z-DEVD-R110 at 25 μM was added to each well. Plates were incubated
until the signal stabilized, and then the fluorescence signal of each
well was measured in an Analyst GT microplate reader. The percentage
increase in apoptotic cells was calculated by comparison of the fluorescence
reading obtained from treated versus control cells.
Kinase Screen
For most assays, kinase-tagged T7 phage
strains were grown in parallel in 24-well blocks in an Escherichia coli host derived from the BL21 strain. E. coli were grown to log phase and infected with
T7 phage from a frozen stock (multiplicity of infection 0.4) and incubated
with shaking at 32 °C until lysis (90–150 min). The lysates
were centrifuged (6000g) and filtered (0.2 μm)
to remove cell debris. The remaining kinases were produced in HEK-293
cells and subsequently tagged with DNA for qPCR detection. Streptavidin-coated
magnetic beads were treated with biotinylated small-molecule ligands
for 30 min at room temperature to generate affinity resins for kinase
assays. The liganded beads were blocked with excess biotin and washed
with blocking buffer (SeaBlock (Pierce), 1% BSA, 0.05% Tween 20, 1
mM DTT) to remove unbound ligand and to reduce nonspecific phage binding.
Binding reactions were assembled by combining kinases, liganded affinity
beads, and test compounds in 1× binding buffer (20% SeaBlock,
0.17× PBS, 0.05% Tween 20, 6 mM DTT). Test compounds were prepared
as 40× stocks in 100% DMSO and directly diluted into the assay.
All reactions were performed in polypropylene 384-well plates in a
final volume of 0.04 mL. The assay plates were incubated at room temperature
with shaking for 1 h, and the affinity beads were washed with wash
buffer (1× PBS, 0.05% Tween 20). The beads were then resuspended
in elution buffer (1× PBS, 0.05% Tween 20, 0.5 μM nonbiotinylated
affinity ligand) and incubated at room temperature with shaking for
30 min. The kinase concentration in the eluates was measured by qPCR.
KINOMEscan’s selectivity score (S) is a quantitative measure of compound selectivity. It is calculated
by dividing the number of kinases that bind to the compound by the
total number of distinct kinases tested, excluding mutant variants.
TREEspot is a proprietary data visualization software
tool developed by KINOMEscan. Kinases found to bind
are marked with red circles, where larger circles indicate higher
affinity binding. The kinase dendrogram was adapted and is reproduced
with permission from Science and Cell Signaling Technology, Inc.
Computational Studies
All computations
were carried
out on an HP workstation xw8200 with the Ubuntu 8.10 operating system
using Maestro v8.5 (Schrodinger LLC, New York). Grids were prepared
using the Receptor Grid Generation tool in Glidev4.0. Docking calculations
were run in the standard precision (SP) mode of Glide v4.0. The maxkeep variable, which sets the maximum number of poses
generated during the initial phase of the docking calculation, was
set to 5000, and the keep best variable, which sets
the number of poses per ligand that enter the energy minimization,
was set to 1000. The energy minimization protocol includes a dielectric
constant of 4.0 and 1000 steps of conjugate gradient. Upon completion
of each docking calculation, at most 100 poses per ligand were allowed
to generate. The best docked conformation was chosen considering the
orientation and Glidescore (G-score).[22]
Authors: Kamalika Moulick; James H Ahn; Hongliang Zong; Anna Rodina; Leandro Cerchietti; Erica M Gomes DaGama; Eloisi Caldas-Lopes; Kristin Beebe; Fabiana Perna; Katerina Hatzi; Ly P Vu; Xinyang Zhao; Danuta Zatorska; Tony Taldone; Peter Smith-Jones; Mary Alpaugh; Steven S Gross; Nagavarakishore Pillarsetty; Thomas Ku; Jason S Lewis; Steven M Larson; Ross Levine; Hediye Erdjument-Bromage; Monica L Guzman; Stephen D Nimer; Ari Melnick; Len Neckers; Gabriela Chiosis Journal: Nat Chem Biol Date: 2011-09-25 Impact factor: 15.040
Authors: Lecia V Sequist; Benjamin Besse; Thomas J Lynch; Vincent A Miller; Kwok K Wong; Barbara Gitlitz; Keith Eaton; Charles Zacharchuk; Amy Freyman; Christine Powell; Revathi Ananthakrishnan; Susan Quinn; Jean-Charles Soria Journal: J Clin Oncol Date: 2010-05-17 Impact factor: 44.544
Authors: Aikaterini Rousaki; Yoshinari Miyata; Umesh K Jinwal; Chad A Dickey; Jason E Gestwicki; Erik R P Zuiderweg Journal: J Mol Biol Date: 2011-06-25 Impact factor: 5.469
Authors: Harold J Burstein; Yan Sun; Luc Y Dirix; Zefei Jiang; Robert Paridaens; Antoinette R Tan; Ahmad Awada; Anantbhushan Ranade; Shunchang Jiao; Gary Schwartz; Richat Abbas; Christine Powell; Kathleen Turnbull; Jennifer Vermette; Charles Zacharchuk; Rajendra Badwe Journal: J Clin Oncol Date: 2010-02-08 Impact factor: 44.544
Authors: Andrew J Massey; Douglas S Williamson; Helen Browne; James B Murray; Pawel Dokurno; Terry Shaw; Alba T Macias; Zoe Daniels; Stephanie Geoffroy; Melanie Dopson; Paul Lavan; Natalia Matassova; Geraint L Francis; Christopher J Graham; Rachel Parsons; Yikang Wang; Antony Padfield; Mike Comer; Martin J Drysdale; Mike Wood Journal: Cancer Chemother Pharmacol Date: 2009-12-13 Impact factor: 3.333
Authors: Eloisi Caldas-Lopes; Leandro Cerchietti; James H Ahn; Cristina C Clement; Ana I Robles; Anna Rodina; Kamalika Moulick; Tony Taldone; Alexander Gozman; Yunke Guo; Nian Wu; Elisa de Stanchina; Julie White; Steven S Gross; Yuliang Ma; Lyuba Varticovski; Ari Melnick; Gabriela Chiosis Journal: Proc Natl Acad Sci U S A Date: 2009-05-05 Impact factor: 11.205
Authors: Hao Shao; Xiaokai Li; Michael A Moses; Luke A Gilbert; Chakrapani Kalyanaraman; Zapporah T Young; Margarita Chernova; Sara N Journey; Jonathan S Weissman; Byron Hann; Matthew P Jacobson; Len Neckers; Jason E Gestwicki Journal: J Med Chem Date: 2018-07-13 Impact factor: 7.446
Authors: Tai Wang; Anna Rodina; Mark P Dunphy; Adriana Corben; Shanu Modi; Monica L Guzman; Daniel T Gewirth; Gabriela Chiosis Journal: J Biol Chem Date: 2018-11-08 Impact factor: 5.157
Authors: Sarah N Fontaine; Jennifer N Rauch; Bryce A Nordhues; Victoria A Assimon; Andrew R Stothert; Umesh K Jinwal; Jonathan J Sabbagh; Lyra Chang; Stanley M Stevens; Erik R P Zuiderweg; Jason E Gestwicki; Chad A Dickey Journal: J Biol Chem Date: 2015-04-11 Impact factor: 5.157
Authors: Xiaokai Li; Teresa Colvin; Jennifer N Rauch; Diego Acosta-Alvear; Martin Kampmann; Bryan Dunyak; Byron Hann; Blake T Aftab; Megan Murnane; Min Cho; Peter Walter; Jonathan S Weissman; Michael Y Sherman; Jason E Gestwicki Journal: Mol Cancer Ther Date: 2015-01-06 Impact factor: 6.261
Authors: Anna Rodina; Tai Wang; Pengrong Yan; Erica DaGama Gomes; Mark P S Dunphy; Nagavarakishore Pillarsetty; John Koren; John F Gerecitano; Tony Taldone; Hongliang Zong; Eloisi Caldas-Lopes; Mary Alpaugh; Adriana Corben; Matthew Riolo; Brad Beattie; Christina Pressl; Radu I Peter; Chao Xu; Robert Trondl; Hardik J Patel; Fumiko Shimizu; Alexander Bolaender; Chenghua Yang; Palak Panchal; Mohammad F Farooq; Sarah Kishinevsky; Shanu Modi; Oscar Lin; Feixia Chu; Sujata Patil; Hediye Erdjument-Bromage; Pat Zanzonico; Clifford Hudis; Lorenz Studer; Gail J Roboz; Ethel Cesarman; Leandro Cerchietti; Ross Levine; Ari Melnick; Steven M Larson; Jason S Lewis; Monica L Guzman; Gabriela Chiosis Journal: Nature Date: 2016-10-05 Impact factor: 49.962
Authors: Amit J Sabnis; Christopher J Guerriero; Victor Olivas; Anin Sayana; Jonathan Shue; Jennifer Flanagan; Saurabh Asthana; Adrienne W Paton; James C Paton; Jason E Gestwicki; Peter Walter; Jonathan S Weissman; Peter Wipf; Jeffrey L Brodsky; Trever G Bivona Journal: Proc Natl Acad Sci U S A Date: 2016-07-22 Impact factor: 11.205
Figure 1
Figure 2
Scheme 1
Scheme 3
Scheme 2
Figure 3
Figure 4
Figure 5
Figure 6