The discovery and development of heat shock protein 70 (Hsp70) inhibitors is currently a hot topic in cancer. In the preceding paper in this issue ( 10.1021/jm401551n ), we have described structure-activity relationship studies in the first Hsp70 inhibitor class rationally designed to bind to a novel allosteric pocket located in the N-terminal domain of the protein. These ligands contained an acrylamide to take advantage of an active cysteine embedded in the allosteric pocket and acted as covalent protein modifiers upon binding. Here, we perform chemical modifications around the irreversible inhibitor scaffold to demonstrate that covalent modification is not a requirement for activity within this class of compounds. The study identifies derivative 27c, which mimics the biological effects of the irreversible inhibitors at comparable concentrations. Collectively, the back-to-back manuscripts describe the first pharmacophores that favorably and selectively interact with a never explored pocket in Hsp70 and provide a novel blueprint for a cancer-oriented development of Hsp70-directed ligands.
The discovery and development of heat shock protein 70 (Hsp70) inhibitors is currently a hot topic in cancer. In the preceding paper in this issue ( 10.1021/jm401551n ), we have described structure-activity relationship studies in the first Hsp70 inhibitor class rationally designed to bind to a novel allosteric pocket located in the N-terminal domain of the protein. These ligands contained an acrylamide to take advantage of an active cysteine embedded in the allosteric pocket and acted as covalent protein modifiers upon binding. Here, we perform chemical modifications around the irreversible inhibitor scaffold to demonstrate that covalent modification is not a requirement for activity within this class of compounds. The study identifies derivative 27c, which mimics the biological effects of the irreversible inhibitors at comparable concentrations. Collectively, the back-to-back manuscripts describe the first pharmacophores that favorably and selectively interact with a never explored pocket in Hsp70 and provide a novel blueprint for a cancer-oriented development of Hsp70-directed ligands.
The heat shock protein 70 (Hsp70) is a
molecular chaperone which
plays an important function in protein homeostasis as well as in cell
signaling and survival.[1,2] Some of its functions include
folding newly synthesized peptides, refolding misfolded proteins,
assembling multiprotein complexes, and transporting proteins across
cellular membranes. In addition to these housekeeping functions, Hsp70
is an important regulator of malignant transformation, both through
its role as a powerful antiapoptotic protein and as a cochaperone
of heat shock protein 90 (Hsp90).[3−5] As a cochaperone of Hsp90,
Hsp70 is thought to load client proteins onto the Hsp90 machinery
through the action of another cochaperone, heat shock organizing protein
(HOP).[6,7] The Hsp90 machinery is an important mechanism
by which cancer cells regulate the function of several cancer-driving
proteins, such as those involved in altered signaling, the cell cycle,
and transcriptional regulation. Indeed, it is primarily for this reason
that Hsp90 has been actively pursued as an anticancer target.[8,9] As an antiapoptotic molecule, Hsp70 acts at multiple points in the
apoptotic pathway to prevent cell death.[3−5] Due to these functions,
it is not surprising that Hsp70 is frequently overexpressed in cancer,
where the elevated expression is furthermore believed to be a cause
of or to lead to resistance to chemotherapy and other treatments.[10] These dual roles of Hsp70 in cancer, i.e., cochaperone
of Hsp90 and antiapoptotic molecule, suggest that inhibition of Hsp70
may offer a valuable anticancer strategy, as supported by Hsp70 knockdown
studies.[11] Indeed, Hsp70 is an important
and highly sought after cancer target,[5,12,13] and as such it is of no surprise that the discovery
and development of Hsp70 inhibitors is currently a hot topic.[3−5]To identify druglike Hsp70 inhibitors, we took a structure-based
approach. In the first preceding paper in this issue, we described
the development of inhibitors that target an allosteric pocket of
Hsp70 located in the N-terminal domain of the protein.[14] This pocket, not evident in the available crystal
structures of Hsp70, has been recently identified by us through computational
analyses.[15] Thus, in the absence of an
appropriate X-ray structure of humanHsp70, we used this homology
model to design ligands that could bind to the Hsp70 allosteric pocket.
Because the pocket also harbors a potentially reactive cysteine residue,
the initially designed inhibitors, all built around the 2,5′-thiodipyrimidine
and 5-(phenylthio)pyrimidine scaffolds, also incorporated an acrylamide
moiety suitably positioned to interact with this amino acid upon insertion
into the binding site (Figure 1). This body
of work led to the identification of low micromolar inhibitors of
Hsp70 with a good cell permeability profile and potent and selective
biological activity in several cancer cells through an Hsp70-mediated
mechanism of action. Our data indicated a good fit for these molecules
inside the Hsp70 pocket, suggesting that enthalpy played an important
role in their interaction with the protein.[14,15]
Figure 1
Chemical
structure of acrylamide-containing 2,5′-thiodipyrimidine
and 5-(phenylthio)pyrimidine scaffold Hsp70 inhibitors that were designed
to insert into the Hsp70 allosteric pocket and form a covalent bond
with Cys267 upon binding. The yellow surface shows the geometry of
the allosteric pocket as determined by SiteMap (Schrodinger LLC, New
York).
Chemical
structure of acrylamide-containing 2,5′-thiodipyrimidine
and 5-(phenylthio)pyrimidine scaffold Hsp70 inhibitors that were designed
to insert into the Hsp70 allosteric pocket and form a covalent bond
with Cys267 upon binding. The yellow surface shows the geometry of
the allosteric pocket as determined by SiteMap (Schrodinger LLC, New
York).In addition to being good leads,
these compounds were also useful
in demonstrating the therapeutic relevance of inhibiting the novel
allosteric Hsp70 pocket as a potential anticancer approach.[15] Specifically, by inserting into the allosteric
pocket, these inhibitors alter the oncogenic Hsp70–Hsp90–client
complexes, resulting in degradation of Hsp90–Hsp70–onco-client
proteins and inhibition of cell growth and induction of apoptosis.
They do so without activating a feedback heat shock response,[15] a mechanism believed to be responsible for limiting
the anticancer activity of Hsp90 inhibitors.[16] The Hsp90–Hsp70 machinery is also a known repressor of heat
shock factor 1 (HSF-1).[16] Inhibition of
Hsp90, but not depletion of the Hsp90 cochaperones Hsp70, HOP, HIP,
p23, and CyP40, led to HSF-1 activation, possibly because while these
cochaperones participate with Hsp90 in the regulation of HSF-1, only
Hsp90 plays a nonredundant role in repressing its heat shock activation
ability.[16] Activation of HSF-1 has a protective
effect on the cancer cell as it leads to the upregulation of antiapoptotic
molecules.[5] Thus, Hsp70 allosteric inhibitors
by differentiating between the regulatory activity of the Hsp90–Hsp70
machinery on onco-clients and on HSF-1 may result in more robust apoptosis
in cancer cells when compared to Hsp90 inhibitors.Here we focus
on the identification of inhibitors that act on Hsp70
through reversible binding into the allosteric pocket. Although covalent
inhibition may offer certain advantages (i.e., maintaining activity
against mutations and the ability to inhibit the target in the presence
of millimolar concentrations of ATP within the cell) and can be desirable
in some instances, it may come with some drawbacks.[17] Because of the presence of a reactive moiety, covalent
inhibitors have a greater potential for off-target effects, such as
inherent reactivity toward nonspecific thiols such as glutathione.[18] Therefore, we desired molecules that could also
act in this pocket in a reversible manner. At this early point in
Hsp70 drug discovery, it is not clear which mode is best; therefore,
we thought it wise to pursue both, and in this paper we show our initial
efforts in the elaboration of a covalent inhibitor targeting the allosteric
site into a reversible noncompetitive inhibitor of low micromolar
potency.
Design of Reversible Ligands of the Hsp70 Allosteric Binding
Site
The design of reversible ligands for the allosteric
Hsp70 pocket
took advantage of our knowledge on the geometry of the allosteric
pocket[15] and the structure–activity
relationship (SAR) we have learned from previous studies in the irreversible
ligand series[14] (Figure 2). Our tactic was to first take a localized approach in an
attempt to phase out the covalent binding potential and to maximize
noncovalent binding in this immediate region (modification on X4). Then we took a global molecular perspective to probe and
to enhance interaction in other parts of the molecule (modification
on X3, rings A and B, X5, and X6).
Specifically, our modifications focused on replacing the acrylamide
(X4) with a favorable noncovalent modifying functionality
and on altering the nature of rings A and B (Figure 2a, highlighted in blue). They then aimed to gain additional
favorable enthalpy by filling out a hydrophobic pocket now only occupied
by small X5,6 groups such as methoxy (Figure 2a, highlighted in blue, and Figure 2b, red circle). Favorable modifications, such as methylpiperazine
at position X7, were left mostly unchanged (Figure 2a).
Figure 2
Design of the reversible Hsp70 inhibitors and their proposed
mode
of interaction with the Hsp70 pocket. (a) General strategy that indicates
modifications that are explored here in the design of reversible inhibitors
(highlighted in blue). (b) Step-by-step process that highlights key
modifications that led to 27c, a reversible Hsp70 binder
of activity rivaling that of the most potent irreversible inhibitor 1e.
Design of the reversible Hsp70 inhibitors and their proposed
mode
of interaction with the Hsp70 pocket. (a) General strategy that indicates
modifications that are explored here in the design of reversible inhibitors
(highlighted in blue). (b) Step-by-step process that highlights key
modifications that led to 27c, a reversible Hsp70 binder
of activity rivaling that of the most potent irreversible inhibitor 1e.Thus, our overall design
strategy entailed (1) determination of
the extent of dependence of covalent modification for activity by
replacing the acrylamide with groups that could provide favorable
but noncovalent interaction with cysteine, (2) exploration of the
SAR to maximize noncovalent interactions within the pocket occupied
by acrylamide, and (3) exploration of the SAR in other parts of the
molecule to maximize binding affinity.
Major Modifications
Modification
X4
Modifying the acrylamide
was based on several observations. As mentioned above, our modeling
studies indicated that the region of Hsp70 occupied by X4 also contained Leu237, Val238, Arg264, Asp234, and Lys271 in addition
to the reactive Cys267 (Figure 2a). While the
ability to form a covalent bond with the cysteine added significantly
to the potency of the irreversible inhibitors, modeling studies indicated
that the alkene moiety of the acrylamide group was positioned to also
form hydrophobic interactions with adjacent protein residues, including
Leu237 and Val238, while its carbonyl was poised to form electrostatic
interactions with Arg264 (Figure 2a). Therefore,
we reasoned that suitable groups substituted at X4 would
be capable of interacting with these residues in a wholly noncovalent
manner. In our initial design of reversible ligands, we kept the amide
functionality of X4 but modified the alkene with a variety
of groups (R, Figure 2a) incapable of interacting
with cysteine covalently. These included alkyl groups of various sizes
as well as aryl groups (R = methyl, ethyl, cyclopropyl, cyclobutyl,
cyclohexyl, n-heptyl, allyl, phenyl, and furanyl)
to probe the extent of modifications allowed by the pocket at this
position. We also replaced the acrylamide with substituents that could
potentially establish an ionic pair interaction with the S– of Cys267 (i.e., ionizableamines such as in 7f (R
= aminomethyl) and 9 (R = 2-(dimethylamino)ethyl)) (Tables 1 and 2).
Table 1
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.
Insoluble.
Table 2
Inhibition
of growth measured in
Kasumi-1 acute myeloid leukemia cells.
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.Insoluble.Inhibition
of growth measured in
Kasumi-1 acute myeloid leukemia cells.Caspase-3,7 activation measured
in MOLM13 acute myeloid leukemia cells. Values are the mean ±
SEM.
Ring B
Ring B
is predicted to occupy the lower part
of the binding site that also contains Arg264. Thus, the ligand’s
interaction with the pocket could potentially be stabilized by cation−π
interactions between the aromatic ring of ring B and the guanidine
group of arginine (Figure 2a).[19] Cation−π interactions play an important role
in protein structure[20,21] and molecular recognition[19,22,23] and can contribute significantly
to the binding energy of a ligand.[24] Such
interactions are more likely to occur between an electron-rich π
system (i.e., phenyl, ethylene, acetylene) and a neighboring cation
and furthermore are generally favored with heterocycles that incorporate
lone pair electrons into the aromatic system (i.e., indole, pyrrole)
and become deactivated when the lone pair does not contribute to aromaticity
(i.e., pyridine, pyrimidine).[25] In the
irreversible series, pyrimidine was favored for ring B; the two nitrogen
atoms of the pyrimidine ring rendered the ring more electron deficient,
thus activating the acrylamide’s Michael acceptor capability.[14] In the reversible series, on the other hand,
electron-rich B rings should be favored over electron-deficient ones
because of their higher propensity toward cation−π interactions
with Arg264 (Figure 2a). To test the hypothesis,
we made compounds in which ring B was either phenyl (i.e., activating)
or pyrimidine (i.e., deactivating) (Tables 1 and 2).
Ring A and Substituents
X5,6
Ring A and
its substituents X5 and X6 are placed midpocket.
Our initial investigations have mainly explored small substituents
at this position such as methoxy, ethoxy, and methyl.[14] This site, however, provides opportunity for substantial
gains in binding affinity. It is lined by amino acids that could provide
hydrophobic (Val59, Tyr41, Phe68, Trp90) and electrostatic (Arg261)
interactions with the ligand (Figure 2a). Therefore,
we explored here the effects of larger substituents on ring A, especially
those poised for interactions within the lipophilic pocket formed
by Tyr41, Phe68, and Trp90, such as derivatives in which X5 contains an aryl (red circle in Figure 2b
and Table 3). We also disrupted the symmetry
of these molecules around ring A by probing monosubstitution of the
ring (X5 ≠ X6 = H, Figure 2b). The representative SAR in this series is presented in
Table 3.
Table 3
Inhibition
of growth measured in
Kasumi-1 acute myeloid leukemia cells.
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.Caspase-3,7 activation measured
in MOLM13 acute myeloid leukemia cells. Values are the mean ±
SEM.
Minor Modifications
Modification
X3
Ring B orients its X3 substituent
toward several important pocket residues such
as Asp234 and Glu268 (X3 is NH2 in 1a, Figure 2a). NH2 at this position
is poised to form hydrogen bonds with their carboxylate groups, and
we retained this functionality from the irreversible series in the
design of new derivatives. The introduction of an −NH2 at this position was synthetically challenging when ring B was phenyl,
and for these, hydrogen was introduced at X3. Thus, all
compounds of this series contain either an NH2 or a H at
position X3.
Modification X7
Our previous
SAR also showed
that N-methylpiperazine was a preferred substituent
on ring A at position X7, presumably because of a proper
fit in the minor groove located at the exit of the binding site and
because it could form a hydrogen bond interaction with the backbone
carbonyl of His89 (Figure 2a for 1a). As a result, X7 was unchanged in the design of the
novel derivatives described here.
Chemistry
The
synthesis of all designed compounds evaluated in this study
is shown in Schemes 1–3. Target compounds 3a–d and 4a–e were prepared by reaction of 2a–d with the appropriate acid chloride
(Scheme 1). Similarly,
target compounds 7a–h were prepared
by reaction of 6a,b with the appropriate
acid chloride (Scheme 1). Thioethers 6a and 6b were prepared by CuI/neocuproine-catalyzed
coupling of iodo derivative 5a or 5b, respectively,
with 3-aminothiophenol (Scheme 1). 9 was prepared by Michael addition of dimethylamine to 8 with DBU (Scheme 1).
Scheme 1
Scheme 3
The synthesis of
pyridine derivatives is shown in Scheme 2.
Bromopyridine 10 was reacted with N-methylpiperazine
at 130 °C for 16 h to give 11 in 88% yield. 12 was obtained in 95% yield
following reaction of 11 with NIS at rt. 12 was coupled to 4-amino-2-mercaptopyrimidine with CuI/neocuproine
to give thioether 13 in 60% yield. Acylation of 13 with propionyl chloride or cyclopropanecarbonyl chloride
resulted in 14a or 14b, respectively. Coupling
of 12 with 3-aminothiophenol resulted in 15, which was acylated with various acid chlorides to give target compounds 16a–d. 17 was obtained from
demethylation of 11 by refluxing in 48% HBr(aq) along
with a catalytic amount of tetrabutylammonium bromide. Following PMB
protection and reaction with NIS, iodo derivative 19 was
obtained, which was coupled with 3-aminothiophenol to give thioether 20. Reaction of 20 with Boc-glycine and subsequent
deprotection with TFA resulted in 21.
Scheme 2
Target compounds 27a–f were synthesized
as shown in Scheme 3. First, 2,4-dichloropyrimidine (22) was reacted with
benzyl alcohol under PTC conditions to result in a regioisomeric mixture
of the desired 4-(benzyloxy)-2-chloropyrimidine (23a)
along with 2-(benzyloxy)-4-chloropyrimidine in a ratio of 75:25, respectively.
This mixture was used directly in the next step and reacted with N-methylpiperazine to give 24a, which was obtained
pure following chromatography. After iodination with NIS to yield 25a and further coupling with 3-aminobenzenethiol, 26a was obtained. 27a–f was then obtained
by reaction of the appropriate acid chloride with 26a. Phenyl-substituted analogue 27g was obtained by a
similar route used for compounds 27a–f; however, the first step in the synthesis utilized a Suzuki coupling
to install a phenyl group selectively onto the 4-position of 22 to yield 23b (Scheme 3).
Biological Evaluation and Structure–Activity Relationship
in the Hsp70 Inhibitor Series
Our target-derived biological
investigation was based on a specific
battery of assays that we designed to probe Hsp70-derived mechanisms
in a cancer cell.[14] These include biochemical
and phenotypic assays that probe biological effects that stem from
alteration of the Hsp70–HOP–Hsp90 megacomplex and its
chaperoning of oncoproteins.[14] These biological
tests were paralleled by computational studies, and each derivative
was docked into the allosteric pocket of Hsp70 to understand the contribution
of each chemical modification to the observed biological activity.Our first step toward the potential realization of reversible inhibitors
involved removing the reactive acrylamide moiety. Rewardingly, elimination
of the covalent mode of binding by isosteric replacement of the acrylamide
to ethylamide diminished the biological activity by only 1 log (i.e., 1a vs 3a, 1b vs 4b, 1c vs 7a, and 1d vs 4c, Tables 1 and 4) while
retaining an Hsp70-driven mechanism of action (Table 1 and Figure 3). Specifically, when
cancer cells were incubated with select derivatives (i.e., 1a and 3a, acrylamide- and ethylamide-containing derivatives,
respectively), we observed dissociation of Hsp70–HOP complexes
at concentrations at which we also noted degradation of the Hsp70
onco-client protein kinase, HER2 (Figure 3).
As previously mentioned, HOP is a cochaperone that bridges Hsp70 and
Hsp90 to form the megachaperone complex that regulates the stability
of several onco-client proteins, such as HER2 and Raf-1. When these
complexes become disrupted, the client proteins (i.e., HER2 and Raf-1)
become unstable and are cleared through the proteasome pathway.[15] As mentioned above, the pocket occupied by the
X4 substituent contains Leu237, Val238, and Cys267, and
the ethyl group of 3a would be well positioned to form
hydrophobic interactions with these residues.
Table 4
a
compd
HER2b
Raf-1b
cPARPb
growth inhibitionb
Hsp90 bindingc
1a
1
2.5
1.5
1.1
>500
1d
4
5
5
1.8
>500
3a
50
70
80
55
>250
7a
75
80
75
55
>500
3c
125
150
180
NA
>500
7b
18
20
18
15
>50
3d
>25
>25
>25
NA
>500
7e
7.5
10
7.5
8
497
7c
70
45
75
70
>250
4d
25
25
30
30
>250
7f
20
20
40
25
>500
9
70
50
75
ND
>500
16c
19
20
25
20
>250
7h
18
15
19
15
>500
27c
1.5
1.2
1.5
2.1
>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 tested
at the maximum concentration allowed by solubility.
Figure 3
Hsp70–HOP dissociation and HER2 degradation
by 1a and 3a. SKBr3 cells were treated for
24 h with vehicle
or indicated concentrations of 1a or 3a.
Hsp70–HOP dissociation: Upon cell lysing, Hsp70 complexes were
isolated with an anti-Hsp70 antibody (IP BB70) and analyzed by Western
blotting (WB). Specificity of binding was tested with a control IgG.
HER2 degradation: Proteins in the lysate were analyzed by immonoblotting
with an anti-HER2 antibody. β-Actin was used to control for
equal loading. Gels were quantified by densitometry, values normalized
to the control (vehicle only treated cells), and data graphed against
the Hsp70 inhibitor concentration. Error bars represent the SD of
the mean (n = 3).
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 tested
at the maximum concentration allowed by solubility.Hsp70–HOP dissociation and HER2 degradation
by 1a and 3a. SKBr3 cells were treated for
24 h with vehicle
or indicated concentrations of 1a or 3a.
Hsp70–HOP dissociation: Upon cell lysing, Hsp70 complexes were
isolated with an anti-Hsp70 antibody (IP BB70) and analyzed by Western
blotting (WB). Specificity of binding was tested with a control IgG.
HER2 degradation: Proteins in the lysate were analyzed by immonoblotting
with an anti-HER2 antibody. β-Actin was used to control for
equal loading. Gels were quantified by densitometry, values normalized
to the control (vehicle only treated cells), and data graphed against
the Hsp70 inhibitor concentration. Error bars represent the SD of
the mean (n = 3).We wanted to explore the further potential of increasing
potency
by exploring this pocket with R groups of various sizes and to take
advantage of potential hydrophobic interactions within the pocket.
Substituting ethyl with methyl (i.e., 3c and 4a) diminished activity, while cyclopropyl at this position slightly
increased activity (i.e., 4c vs 4e and 3a vs 7b, Table 1, X1,2 = N). Substitution of ethyl with groups such as allyl or
cyclobutyl (i.e., 7a vs 7g and 7c) led to no significant change in activity, suggesting that such
groups can be accommodated in this site, but that no additional favorable
interactions were possible. Much larger hydrophobic substituents at
this position abolished both activity (i.e., n-heptyl
in 3d, steric clashes with Asn235, Leu237, Val238, and
Val260[15]) and solubility (i.e., cyclohexyl
in 7d), which in the latter case precluded biological
evaluation.We also found that flat, conformationally restricted
aromatic groups
were well tolerated at this position and in fact significantly improved
activity. Specifically, the 2-furanyl derivative (i.e., 4d) was 2–3-fold more potent than the corresponding ethyl and
cyclopropyl derivatives (i.e., 4c and 4e, respectively) and only 5-fold less potent than the corresponding
acrylamide derivative (i.e., 1d) (Table 1). Even more favored was a phenyl at this position (i.e., 7e), this compound being 8–10-fold more active than
the corresponding ethyl derivative (i.e., 7a) (Tables 1 and 4). Such an increase
in potency for hydrophobic and π-electron-containing R groups
may be explained by enhanced hydrophobic interactions with Leu237,
Val238, Leu263, and Phe241 or a potential S−π interaction
between the sulfur of Cys267 and the R group. Such favorable interactions
between sulfur and π aromatic systems were first suggested by
Morgan et al.,[26] and a number of stable
orientations are possible, including one where the sulfur is positioned
3.5–4.0 Å above the plane of the aromatic ring and another
where the sulfur is positioned slightly above (≤2.5 Å)
and to the edge (4.5–6.0 Å) of the aromatic ring.[22] Both orientations have been found to occur with
cysteine residues in proteins, and recently a S–H/π interaction
was proposed to rationalize the binding affinity of an FLT3 kinase
inhibitor.[27]Our second approach
toward enhancing activity within the pocket
was to take advantage of the native cysteine residue and its potential
to form ionic interactions through its side chain thiolate (S–). We attempted to take advantage of such an interaction
by incorporating an amine group at the same position as the β-carbon
of the Michael acceptor acrylamide in 1a.[28−30] An increase in potency was observed for R substituents that could
potentially be positively ionized and form an ionic pair with S– of cysteine (2-aminomethyl in 7f and 16d). Indeed, 2-aminomethyl appears to favorably position
its amine, which is likely protonated at physiological pH, to interact
with the sulfur of cysteine (i.e., 7f in Table 1 and 16d in Table 2). These derivatives have an activity that is comparable to
that of the corresponding phenyl and/or furanyl compounds (7f vs 7e and 4d, Table 1, and 16d vs 16c, Table 2). Positioning the amine one carbon away, such as in 9 (R = 2-(dimethylamino)ethyl, Table 1), resulted in activity comparable to that of 7f in
the growth inhibition and caspase-3,7 activation assays. In this case, 9 has the potential to undergo bioconversion in cells via
β-elimination to the corresponding alkene,[31] and therefore, in addition to possible interactions described
above, it may also interact with Cys267 covalently.Having determined
these two potential strategies that replace the
acrylamide with minimal loss of activity, i.e., a π-electron-
or positively ionizable functionality-containing R group, we went
on to explore other sites on the molecule that would provide a gain
in binding affinity. As mentioned in our design, changing ring B from
pyrimidine to phenyl is expected to increase potency by enhancing
the interaction of the ring with Arg264. Indeed, unlike in the irreversible
inhibitor series, where pyrimidine was favored over phenyl because
it increased the reactivity of acrylamide for cysteine,[14] in the reversible inhibitor series, we found
the opposite to be true (i.e., 7a vs 4b and 16b vs 14b, Tables 1 and 2, and additional examples in Table 1). Changing ring B from pyrimidine to a phenyl improved the
activity 2-fold. Arg264 has its guanidinium group positioned atop
ring B, thus favorably aligned for cation−π interactions
to occur (Figure 2a).[19] Thus, an electron-rich aromatic ring, such as phenyl, may form a
stronger stacking interaction than pyrimidine with this residue. In
heterocycles, where the lone pair does not contribute to aromaticity,
the electronegativity of the heteroatom weakens the cation−π
binding ability.We next explored modifications to ring A. Having
ring A either
pyrimidine or pyridine could increase the chemistry feasibility for
exploring distinct X5 and X6 combinations on
the molecules, and thus, we first explored the effect on activity
by changing the nature of the ring. Rewardingly, a change from pyrimidine
to pyridine led to no substantial change in activity (i.e., 7e vs 16c and 7f vs 16d, Tables 1 and 2).
On the other hand, disrupting the aromaticity of ring A completely
abolished activity (21, Table 2). 2-Hydroxypyridines undergo tautomerism to give pyridones which
can behave like amides and exist mainly as the “amide”
tautomer in most solvents. The dramatic loss of activity observed
upon replacing the methoxy group with a hydroxy group at X5 hinted at the importance of this position for binding energy and
demonstrates that this group makes significant interactions that are
essential for the potency of these compounds.Thus, having the
ability to use both pyridine and pyrimidine as
ring A increased our chemical versatility, and we next explored distinct
X5,6 combinations. Results with the pyridine series (Table 2) showed that disubstitution of X5 and
X6 was not necessarily required as having a single methoxy
group on the molecule performed as well as having two. This was also
observed within the pyrimidine series, as 7h was of potency
similar to that of its disubstituted analogue 7f (Table 1). We therefore sought to explore this position
further, especially by adding aryl substituents that, as indicated
above, would be poised to fill the hydrophobic pocket currently occupied
by methoxy (Figure 2b, red circle). Substitution
of X5 with benzyloxy led to a remarkable 1 log increase
in potency (compare 27a to 7a). In fact,
this derivative was almost as potent as the corresponding acrylamide-containing
derivative (compare 27a to 1c). We further
increased the activity of this compound once the methyl (R on X4; see Figure 2a) was substituted with
a phenyl or aminomethyl (27b and 27c, respectively,
Table 3), both favored X4 substituents
(Tables 1 and 2). With
aminomethyl being favored at this position, we also probed whether
adding hydrophobic bulk to this functionality could lead to a gain
in affinity. As indicated above, Leu237 and Val238 are in the vicinity
of Cys267, and thus, there is potential for minor affinity gain by
increasing hydrophobicity. Replacing the H in 27c with
a methyl (27d) or isopropyl (27e) or switching
the aminomethyl to 2-pyrrolidinyl (27f) resulted in a
minor increase in activity or had no measurable effect on activity
(Table 3).Changing X5 from
benzyloxy to the less flexible phenyl
substituent directly attached to the pyrimidine A ring resulted in
a significant 10-fold loss of activity (27g, Table 3). Furthermore, substituting the phenyl ring of 27c with the relatively more polar pyridine in 27h also decreased affinity 5-fold. These results are very much concordant
with the proposed mode of binding for these ligands. Docking analysis
showed that the benzyloxy group can be accommodated in the left-side
hydrophobic subpocket (red circle, Figure 2b) and form hydrophobic interactions with Tyr41, Phe68, and Trp90.
For 27g, the phenyl substituent is not able to orient
properly in this cavity and, due to its inflexibility, rather orients
toward Asp69 and Glu231, providing an explanation for its loss of
activity. Similarly, replacing the phenyl in 27c with
pyridine as in 27h increases the polarity and thus its
ability to favorably interact with the hydrophobic residue environment.Altogether, and as seen in the irreversible series, our tests demonstrated
a good correlation between the predicted binding mode (Figure 2) and the observed biological activity (Tables 1–4) of the designed
ligands, further consolidating that the biological effect of these
molecules in cancer cells is majorly Hsp70 mediated. No effect was
noted for these compounds on Hsp90 at concentrations as high as 500
μM (Table 4).One of the most active
compounds derived from these studies is 27c. This ligand
combines individual features that this study
identifies as most favorable at each evaluated position; its X4 is glycine, ring B is phenyl, and X5 is benzyloxy
(Figure 2b). This analogue has an activity
in cells comparable to that of the most active reported irreversible
inhibitors of this class,[14] namely, low
micromolar activity in cancer cells (Figure 4). We found that 27c interfered with the formation of
functional Hsp70–HOP–Hsp90 machinery as indicated by
its ability to dose-dependently alter the megacomplex components and
to destabilize an Hsp70–Hsp90 machinery client, Raf-1 (Figure 4a). Hsp90 in concert with Hsp70 maintains the transforming
capacity of several oncoproteins, including HER2, AKT, Raf-1, IGF-IR,
and HIF-1.[4,9] When this chaperone complex becomes pharmacologically
inhibited, these oncoproteins become destabilized and are degraded
mainly by the proteasomal pathway. Indeed, we found that the steady-state
levels and/or the activity of several oncoproteins involved in increased
signaling through a pathogenic pathway were sensitive to Hsp70 inhibition
by 27c (Figure 4b). These signaling
oncoproteins include HER2 and Raf-1 in the HER2-overexpressing SKBr3breast cancer cells, STAT3 and Raf-1 in the triple-negative breast
cancerMDA-MB-468 cells, STAT3, Raf-1, and AKT in the MiaPaCa2 pancreatic
cancer cells, and mutant FLT3 and STAT5 in MOLM13 acute myeloid leukemia
cells. 27c also resulted in induction of apoptosis in
these cancer cells, as indicated by substantial PARP cleavage (cPARP;
Figure 4b).
Figure 4
Allosteric ligands with a reversible mode
of binding mimic the
cellular phenotype observed with the irreversible Hsp70 inhibitors.
Addition of 27c to cancer cells dose-dependently alters
the formation of the Hsp70–HOP complex, a phenomenon associated
with their destabilization and reduction in half-life (a). It is associated
with degradation and/or inhibition of Hsp90–Hsp70 oncoproteins
and induction of apoptosis (b). Collectively, these data indicate
that its mechanism of action is mediated by interaction with the Hsp70
allosteric pocket in a fashion resembling the interaction of 1e.[14,15]
Allosteric ligands with a reversible mode
of binding mimic the
cellular phenotype observed with the irreversible Hsp70 inhibitors.
Addition of 27c to cancer cells dose-dependently alters
the formation of the Hsp70–HOP complex, a phenomenon associated
with their destabilization and reduction in half-life (a). It is associated
with degradation and/or inhibition of Hsp90–Hsp70 oncoproteins
and induction of apoptosis (b). Collectively, these data indicate
that its mechanism of action is mediated by interaction with the Hsp70
allosteric pocket in a fashion resembling the interaction of 1e.[14,15]Next, we investigated whether binding of 27c to Hsp70
interfered with its main biochemical activities, specifically refolding
of a denatured client protein (Figure 5). Hsp70
activities are stimulated by Hsp40 proteins and nucleotide exchange
factors, such as Hsp110.[1,7,32] Humans have several cytosolic Hsp40’s, including Hdj1, DJA1,
DJA2, and DJA4, and it has recently been reported that DJA2 is most
efficient in promoting the refolding of an Hsp70 polypeptide substrate,
firefly luciferase.[33−35] In cells, the refolding of heat-denatured luciferase
by endogenous as well as transfected Hsp70 was inhibited by 27c. The nonspecific capacity of transfected Hsp70 to maintain
substrate solubility after heat shock was not greatly affected, indicating
that 27c, in a manner we reported for 1e,[15] targeted the specific substrate folding
activity of Hsp70 (Figure 5).
Figure 5
HEK293 cells were transfected
with luciferase and either control
vector or Hsp70. The cells were treated with cycloheximide and either
vehicle or 27c at 10 μM, incubated at 45 °C
for 1 h, and allowed to recover at 37 °C for 2 h (left). Cells
were lysed during and after refolding, and soluble HA-tagged luciferase
and chaperones were detected in the lysates; exogenous transfected
Flag-tagged Hsp70 is visible as a band above endogenous Hsp70 (middle).
Luciferase enzymatic activities in the lysates were measured at 2
h of refolding, unless otherwise indicated, and are represented as
percentages of the initial activity before heat shock (right). 27c significantly inhibited endogenous Hsp70 and transfected
Hsp70 in multiple experiments (n ≥ 3).
HEK293 cells were transfected
with luciferase and either control
vector or Hsp70. The cells were treated with cycloheximide and either
vehicle or 27c at 10 μM, incubated at 45 °C
for 1 h, and allowed to recover at 37 °C for 2 h (left). Cells
were lysed during and after refolding, and soluble HA-tagged luciferase
and chaperones were detected in the lysates; exogenous transfected
Flag-tagged Hsp70 is visible as a band above endogenous Hsp70 (middle).
Luciferase enzymatic activities in the lysates were measured at 2
h of refolding, unless otherwise indicated, and are represented as
percentages of the initial activity before heat shock (right). 27c significantly inhibited endogenous Hsp70 and transfected
Hsp70 in multiple experiments (n ≥ 3).
Conclusions
We have shown that significant
biological activity may be retained
through reversible Hsp70 inhibitors targeting an allosteric pocket
located at the N-terminal domain. Because of the appropriate fit,
and thus good enthalpy of binding of the irreversible inhibitors,
simple modifications such as replacement of the covalent linkage with
an ionic bridge (yellow, Figure 2b) and filling
of the hydrophobic pocket occupied by methoxy in 1e with
a benzyloxy (red, Figure 2b) have led to ligands
of reversible mode of binding that not only mimic the phenotype observed
with 1e, but do so with similar potency (Figure 4).Our data show that the acrylamide group
could be eliminated altogether
by improving the enthalpy of the binding and indicate that significant
binding energy can be attained through additional hydrophobic interactions
of X5 substituents with the Tyr41, Phe68, and Trp90 residues.
This interaction weighs majorly toward the binding of these ligands
to the allosteric pocket of Hsp70.Combined, the SAR data from
this and the accompanying paper[14] validate
the homology model and the proposed
binding of these ligands to the allosteric pocket. First, the model
has allowed for the rational design of a ligand that, when incubated
with the thousands of proteins expressed in a cancer cell, affinity
purified one, Hsp70. Second, the correctness of the binding mode analyses
has allowed the rational design of specific ligands with a tractable
SAR. Some of these specific ligands as we show have a reversible mode
of binding whose potency rivals that of the irreversible ligands,
indicating that favorable enthalpy drives the binding of these compounds
to Hsp70. Third, the concordance in the observed biochemical and phenotypic
effects observed with these agents in cancer cells, such as exemplified
for 27c, suggests that, in the tested concentration range,
the biological activity of these agents is majorly and selectively
channeled through an Hsp70-binding mechanism.In addition to
providing both a novel pharmacophore and medicinal
chemistry for its assembly, we describe in our papers a testing battery
for assessing Hsp70-mediated mechanisms in cancer cells and for evaluating
specific ligand action in cancer cells through Hsp70 inhibition. Therefore,
we provide a novel blueprint for a cancer-oriented development of
Hsp70-directed ligands.In conclusion, our findings propose
the allosteric Hsp70 inhibitors
as important leads toward the development of novel targeted anticancer
therapeutics. 27c serves as a molecule for further development
that can potentially be elaborated into more potent molecules with
in vivo efficacy. We are currently working to further optimize this
class of compounds for potency and in vivo activity and will disclose
our results in due course.
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 or 600 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 verified 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. The syntheses of 1a–d and 8 are described
elsewhere.[14]
To a solution of 6a (20 mg,
0.0552 mmol) in THF (1 mL) were added Boc-glycine (9.7 mg, 0.0552
mmol) and DCC (12 mg, 0.058 mmol), and the resulting solution was
stirred at rt for 5 h. The reaction mixture was concentrated under
reduced pressure, and the residue was purified by preparatory TLC
(EtOAc/MeOH–NH3 (7 N), 20:1) to afford a solid which
was dissolved in 2 mL of CH2Cl2/TFA (4:1). The
resulting solution was stirred at rt for 1 h. The reaction mixture
was concentrated under reduced pressure, and the residue was purified
by preparatory TLC (CH2Cl2/MeOH–NH3 (7 N), 15:1) to afford 13.9 mg (60%) of 7f. 1H NMR (500 MHz, CDCl3/MeOH-d4): δ 7.38 (d, J = 8.2 Hz, 1H), 7.26
(s, 1H), 7.15 (t, J = 7.9 Hz, 1H), 6.79 (d, J = 7.6 Hz, 1H), 3.86–3.97 (m, 10H), 3.52 (s, 2H),
2.58 (m, 4H), 2.41 (s, 3H). HRMS (m/z): [M + H]+ calcd for C19H27N6O3S, 419.1865; found, 419.1862.
A mixture of 5b (0.200 g,
0.600 mmol) and K2CO3 (0.166 g, 1.20 mmol) in
DMF (6 mL) was evacuated and backfilled with argon three times. Copper(I)thiophene-2-carboxylate (0.034 g, 0.180 mmol) was added, and the resulting
mixure was evacuated and backfilled with argon two times. 3-Aminothiophenol
(76 μL, 0.090 g, 0.72 mmol) was added, and the reaction mixture
was heated at 120 °C for 24 h. Solvent was removed under reduced
pressure, and the residue was purified by column chromatography (CH2Cl2/MeOH–NH3 (7 N), 200:1 to
40:1) to afford 0.147 g (74%) of 6b. 1H NMR
(500 MHz, CDCl3): δ 8.22 (s, 1H), 7.00 (t, J = 7.8 Hz, 1H), 6.52 (d, J = 7.7 Hz, 1H),
6.38–6.45 (m, 2H), 3.91 (s, 3H), 3.89 (m, 4H), 3.60 (br s,
2H), 2.49 (m, 4H), 2.36 (s, 3H). 13C NMR (125 MHz, CDCl3): δ 169.4, 164.9, 161.6, 146.9, 138.8, 129.6, 116.9,
112.9, 112.5, 99.2, 54.9, 53.9, 46.2, 43.8. HRMS (ESI) m/z [M + H]+ calcd for C16H22N5OS, 332.1545; found, 332.1532.
To a solution of 6b (15.5
mg, 0.047 mmol) in THF (1 mL) were added Boc-glycine (9.1 mg, 0.052
mmol) and DCC (10.7 mg, 0.052 mmol). After the resulting solution
was stirred overnight at rt, THF was evaporated and 1 mL of CH2Cl2/TFA (4:1) was added. The solution was stirred
for 45 min and then concentrated to dryness under reduced pressure
to give a residue which was purified by preparatory TLC (CH2Cl2/MeOH–NH3 (7 N), 15:1) to afford
13.4 mg (73%) of 7h. 1H NMR (500 MHz, CDCl3): δ 9.29 (br s, 1H), 8.23 (s, 1H), 7.45–7.50
(m, 1H), 7.30–7.34 (m, 1H), 7.17 (t, J = 8.0
Hz, 1H), 6.81–6.87 (m, 1H), 3.91 (s, 3H), 3.86–3.90
(m, 4H), 3.45 (s, 2H), 2.46–2.52 (m, 4H), 2.35 (s, 3H). MS
(m/z): [M + H]+ 389.3.
To a solution of 8 (23 mg,
0.053 mmol) in CH3CN (1 mL) were added dimethylamine (2
M, THF; 53 μL, 0.106 mmol) and DBU (4 mg, 0.026 mmol) at rt,
and the resulting solution was stirred for 6 h. The reaction mixture
was concentrated under reduced pressure, and the residue was purified
by preparatory TLC (CH2Cl2/MeOH–NH3 (7 N), 10:1) to afford 13.5 mg (53%) of 9. 1H NMR (600 MHz, CDCl3): δ 11.1 (br s, 1H),
6.91 (s, 1H), 4.79 (br s, 2H), 3.84–3.91 (m, 10H), 2.56 (t, J = 5.8 Hz, 2H), 2.45–2.51 (m, 4H), 2.41 (t, J = 5.8 Hz, 2H), 2.37 (s, 3H), 2.22 (s, 6H). 13C NMR (150 MHz, CDCl3): δ 171.9, 171.2, 170.2, 164.3,
160.0, 157.1, 88.4, 80.8, 54.9, 54.4, 54.1, 46.2, 44.1, 43.6, 33.6.
HRMS (m/z): [M + H]+ calcd
for C20H32N9O3S, 478.2349;
found, 478.2343.
1-(6-Methoxypyridin-2-yl)-4-methylpiperazine
(11)
To a solution of 2-bromo-6-methoxypyridine
(10) (150 mg, 0.8 mmol) and 1-methylpiperazine (240 mg,
2.4 mmol) in
2 mL of DMF was added of K2CO3 (220 mg, 1.6
mmol), and the resulting mixture was heated to 130 °C for 16
h. Solvent was evaporated under reduced pressure, and the residue
was purified by column chromatography (5–10% MeOH in CH2Cl2) to afford 145 mg (88%) of 11. 1H NMR (500 MHz, CDCl3): δ 7.41 (t, J = 8.0 Hz, 1H), 6.16 (d, J = 8.0 Hz, 1H),
6.08 (d, J = 8.0 Hz, 1H), 3.87 (s, 3H), 3.54 (m,
4H), 2.51 (m, 4H), 2.54 (s, 3H). 13C NMR (125 MHz, CDCl3): δ 163.1, 158.3, 140.1, 98.2, 98.1, 54.8, 52.9, 46.2,
45.1. MS (m/z): [M + H]+ 208.4.
To a solution of 11 (124 mg,
0.6 mmol) in
5 mL of acetonitrile was added N-iodosuccinimide
(203 mg, 0.9 mmol), and the resulting mixture was stirred at rt for
2 h. The solvent was evaporated under reduced pressure, and the residue
was purified by column chromatography (CH2Cl2/MeOH–NH3 (7 N), 1:0 to 85:15) to afford 190 mg
(95%) of 12. 1H NMR (500 MHz, CDCl3): δ 7.70 (d, J = 8.0 Hz, 1H), 6.02 (d, J = 8.0 Hz, 1H), 3.90 (s, 3H), 3.60 (m, 4H), 2.62 (m, 4H),
2.39 (s, 3H). 13C NMR (125 MHz, CDCl3): δ
177.7, 160.5, 157.8, 148.6, 100.7, 61.7, 54.2, 45.5, 44.5. MS (m/z): [M + H]+ 334.1.
A mixture of 12 (100 mg, 0.3
mmol), 3-aminothiophenol (37 mg, 0.3 mmol), K2CO3 (83 mg, 0.6 mmol), neocuproine (33 mg, 0.15 mmol), and CuI (29 mg,
0.15 mmol) in DMF (3 mL) was heated to 130 °C for 16 h. Solvent
was removed under reduced pressure, and the residue was purified by
column chromatography (CH2Cl2/MeOH–NH3 (7 N), 1:0 to 90:10) to afford 60 mg (60%) of 15. MS (m/z): [M + H]+ 331.2.
To a solution of 17 (1.2 g,
6.22 mmol) in DMF (40 mL) was added NaH (0.596 g, 24.8 mmol) at rt,
and the mixture was stirred for 10 min. Then (PMB)Cl (1.06 g, 6.83
mmol) was added dropwise, and the mixture was stirred at rt for 1
h. The reaction mixture was concentrated under reduced pressure and
purified by column chromatography to give 1.63 g (84%) of 18. MS (m/z): [M + H]+ 314.2.
To a solution of 18 (1.16
g, 3.7 mmol) in acetonitrile (50 mL) were added NIS (1.66 g, 7.4 mmol)
and TFA (1.42 mL, 18.5 mmol), and the solution was stirred for 1 h
at rt. The reaction mixture was concentrated under reduced pressure
and purified by column chromatography to give 1.22 g (75%) of 19. MS (m/z): [M + H]+ 440.0.
A mixture of 19 (452 mg, 1.03
mmol), 3-aminothiophenol (137 mg, 1.10 mmol), K2CO3 (552 mg, 4.0 mmol), neocuproine (45 mg, 0.2 mmol), and CuI
(39 mg, 0.2 mmol) in DMF (10 mL) was heated to 130 °C for 16
h. Solvent was removed under reduced pressure, and the residue was
purified by column chromatography (CH2Cl2/MeOH–NH3 (7 N), 1:0 to 85:15) to afford 224 mg (50%) of 20. MS (m/z): [M + H]+ 437.2.
To a solution of 20 (220 mg,
0.5 mmol) in CH2Cl2 (10 mL) were added Boc-glycine
(96 mg, 0.55 mmol), DMAP (6.1 mg, 0.05 mmol), and EDCI (105 mg, 0.55
mmol). The resulting solution was stirred at rt for 2 h. Solvent was
evaporated under reduced pressure, and the residue was purified by
column chromatography (CH2Cl2/MeOH–NH3 (7 N), 1:0 to 85:15) to afford a residue. To this was added
10 mL of 30% TFA/CH2Cl2, and the resulting solution
was stirred at rt for 2 h. Solvent was evaporated under reduced pressure,
and the residue was purified by column chromatography (CH2Cl2/MeOH–NH3 (7N), 1:0 to 85:15) to
afford 153 mg (82%) of 21. 1H NMR (500 MHz,
CDCl3): δ 9.31 (br s, 1H), 7.62 (d, J = 8.4 Hz, 1H), 7.44 (d, J = 8.0 Hz, 1H), 7.31 (s,
1H), 7.19 (t, J = 8.0 Hz, 1H), 6.93 (d, J = 7.8 Hz, 1H), 5.58 (d, J = 8.4 Hz, 1H), 3.34 (s,
2H), 3.33 (m, 4H), 2.29 (m, 4H), 2.22 (s, 3H). HRMS (m/z): [M + H]+ calcd for C18H24N5O2S, 374.1651; found, 374.1646.
4-(Benzyloxy)-2-chloropyrimidine (23a)
To a
solution of 2,4-dichloropyrimidine (22) (2.0 g,
0.0134 mmol) in toluene (20 mL) were added benzyl alcohol (1.53 mL,
1.59 g, 0.0147 mol), KOH (0.82 g, 0.0147 mol), and 18-crown-6 (0.177
g, 0.00067 mol), and the resulting solution was stirred at rt for
1 h. The reaction mixture was diluted with EtOAc (400 mL), washed
with water (3 × 50 mL), dried over MgSO4, filtered,
and concentrated to give a white solid that was chromatographed (hexane/CH2Cl2, 1:1 to 3:7) to afford 2.09 g (71%) of a mixture
of 23a with regioisomeric 2-(benzyloxy)-4-chloropyrimidine
(relative ratio 75:25 by 1H NMR, respectively). MS (m/z): [M + Na]+ 243.1.
To a solution of 23a (2.09
g, 0.00947 mol;
contains regioisomer) in DMF (34 mL) was added 1-methylpiperazine
(3.15 mL, 2.85 g, 0.0284 mol), and the resulting solution was heated
at 80 °C for 1.75 h. Solvent was removed under reduced pressure,
and the residue was taken up into EtOAc (350 mL) and washed with brine
(3 × 50 mL). The aqueous layer was extracted with EtOAc (2 ×
50 mL), and the combined organic layers were dried over MgSO4, filtered, and concentrated to give an oil that was purified by
column chromatography (EtOAc/MeOH–NH3 (7 N), 1:0
to 25:1) to afford 1.88 g (70%) of 24a. 1H
NMR (500 MHz, CDCl3): δ 8.06 (d, J = 5.6 Hz, 1H), 7.41 (d, J = 7.0 Hz, 2H), 7.35 (t, J = 7.0 Hz, 2H), 7.32 (d, J = 7.0 Hz, 1H),
6.03 (d, J = 5.6 Hz, 1H), 5.35 (s, 2H), 3.83 (m,
4H), 2.45 (m, 4H), 2.33 (s, 3H). MS (m/z): [M + H]+ 284.9.
To 24a (0.937 g, 0.0033 mol)
in
acetonitrile (16 mL) were added TFA (1.02 mL, 1.51 g, 0.0132 mol)
and N-iodosuccinimide (0.965 g, 0.0043 mol), and
the resulting solution was stirred at rt for 1 h. Then 7 mL of 10%
Na2CO3 (0.70 g, 0.066 mol) was added, and the
resulting solution was stirred for 2 min. The reaction mixture was
concentrated to dryness, and the residue was taken up into CH2Cl2 (200 mL) and washed with 10% Na2CO3 (2 × 50 mL), 10% sodium thiosulfate (50 mL),
and brine (50 mL). The organic layer was dried over MgSO4, filtered, and concentrated to give an oil which was purified by
column chromatography (CH2Cl2/MeOH–NH3 (7 N), 50:1) to yield 1.31 g (97%) of 25a. 1H NMR (500 MHz, CDCl3): δ 8.27 (s, 1H), 7.44
(d, J = 7.4 Hz, 2H), 7.37 (t, J =
7.2 Hz, 2H), 7.32 (d, J = 7.3 Hz, 1H), 5.40 (s, 2H),
3.79 (m, 4H), 2.42 (m, 4H), 2.32 (s, 3H). MS (m/z): [M + H]+ 411.0.
To 26a (5.2 mg, 0.013 mmol)
in CH2Cl2 (0.2 mL) was added acetic anhydride
(1.5 μL, 1.6 mg, 0.0156 mmol), and the resulting solution was
stirred at rt for 4 h. The solution was then concentrated to dryness
under reduced pressure to give a residue which was purified by preparatory
TLC (CH2Cl2/MeOH–NH3 (7 N),
20:1) to afford 4.5 mg (79%) of 27a. 1H NMR
(500 MHz, CDCl3/MeOH-d4): δ
8.24 (s, 1H), 7.54 (d, J = 7.7 Hz, 1H), 7.22–7.27
(m, 3H), 7.13–7.20 (m, 3H), 7.09 (s, 1H), 6.87 (d, J = 7.7 Hz, 1H), 5.36 (s, 2H), 3.85 (m, 4H), 2.49 (m, 4H),
2.35 (s, 3H), 2.10 (s, 3H). MS (m/z): [M + H]+ 450.1.
To 26a (30 mg, 0.0736 mmol)
in THF (3 mL) were added Boc-glycine (14.2 mg, 0.081 mmol) and DCC
(16.7 mg, 0.081 mmol), and the resulting solution was stirred at rt
overnight. The reaction mixture was concentrated under reduced pressure,
the residue was purified by preparatory TLC (CH2Cl2/MeOH–NH3 (7 N), 25:1) to afford an oil
which was dissolved in 2.5 mL of CH2Cl2/TFA
(4:1), and the resulting solution was stirred at rt for 45 min. The
reaction mixture was concentrated under reduced pressure, and the
residue was purified by preparatory TLC (CH2Cl2/MeOH–NH3 (7 N), 15:1) to afford 24.6 mg (72%)
of 27c. 1H NMR (500 MHz, CDCl3/MeOH-d4): δ 8.24 (s, 1H), 7.50 (dd, J = 1.2, 8.1 Hz, 1H), 7.31 (m, 1H), 7.22–7.27 (m,
3H), 7.13–7.20 (m, 3H), 6.86 (d, J = 7.9 Hz,
1H), 5.37 (s, 2H), 3.85 (m, 4H), 3.39 (s, 2H), 2.49 (m, 4H), 2.35
(s, 3H). 13C NMR (125 MHz, CDCl3/MeOH-d4): δ 171.4, 168.5, 164.1, 161.2, 138.2,
138.1, 136.4, 129.2, 128.2, 127.6, 127.2, 123.0, 118.2, 117.0, 67.7,
54.5, 45.8, 44.6, 43.5. MS (m/z):
[M + H]+ 465.3.
To 26a (8.4 mg, 0.0206 mmol)
in THF (0.5 mL) were added Boc-l-alanine (5.6 mg, 0.0227
mmol) and DCC (4.7 mg, 0.0227 mmol), and the resulting solution was
stirred at rt overnight. The reaction mixture was concentrated under
reduced pressure, the residue was dissolved in 0.35 mL of CH2Cl2/TFA (4:1), and the resulting solution was stirred
at rt for 45 min. The reaction mixture was concentrated under reduced
pressure, and the residue was purified by preparatory TLC (CH2Cl2/MeOH–NH3 (7 N), 15:1) to
afford 4.1 mg (41%) of 27d. 1H NMR (500 MHz,
CDCl3): δ 9.34 (s, 1H), 8.26 (s, 1H), 7.59 (d, J = 8.2 Hz, 1H), 7.20–7.30 (m, 4H), 7.12–7.19
(m, 3H), 6.85 (m, 1H), 5.37 (s, 2H), 3.81–3.91 (m, 4H), 3.60
(q, J = 7.0 Hz, 1H), 2.43–2.52 (m, 4H), 2.35
(s, 3H), 1.42 (d, J = 7.0 Hz, 3H). MS (m/z): [M + H]+ 479.4.
To 26a (6.6 mg, 0.0162 mmol)
in THF (0.5 mL) were added Boc-l-valine (4.1 mg, 0.0178 mmol)
and DCC (3.7 mg, 0.0178 mmol), and the resulting solution was stirred
at rt overnight. The reaction mixture was concentrated under reduced
pressure, the residue was dissolved in 0.35 mL of CH2Cl2/TFA (4:1), and the resulting solution was stirred at rt for
45 min. The reaction mixture was concentrated under reduced pressure,
and the residue was purified by preparatory TLC (CH2Cl2/MeOH–NH3 (7 N), 15:1) to afford 4.4 mg
(54%) of 27e. 1H NMR (500 MHz, CDCl3): δ 9.39 (s, 1H), 8.26 (s, 1H), 7.60 (d, J = 8.0 Hz, 1H), 7.20–7.34 (m, 4H), 7.12–7.19 (m, 3H),
6.84 (d, J = 7.8 Hz, 1H), 5.37 (s, 2H), 3.81–3.92
(m, 4H), 3.36 (m, 1H), 2.39–2.52 (m, 5H), 2.35 (s, 3H), 1.04
(d, J = 6.8 Hz, 3H), 0.86 (d, J =
6.7 Hz, 3H). MS (m/z): [M + H]+ 507.2.
To 26a (8.6 mg, 0.0211 mmol)
in THF (0.5 mL) were added Boc-l-proline (5 mg, 0.0232 mmol)
and DCC (5 mg, 0.0232 mmol), and the resulting solution was stirred
at rt overnight. The reaction mixture was concentrated under reduced
pressure, the residue was dissolved in 0.35 mL of CH2Cl2/TFA (4:1), and the resulting solution was stirred at rt for
45 min. The reaction mixture was concentrated under reduced pressure,
and the residue was purified by preparatory TLC (CH2Cl2/MeOH–NH3 (7 N), 15:1) to afford 6.0 mg
(57%) of 27f. 1H NMR (500 MHz, CDCl3): δ 9.63 (s, 1H), 8.26 (s, 1H), 7.58 (dd, J = 1.2, 8.1 Hz, 1H), 7.31 (t, J = 1.9 Hz, 1H), 7.12–7.17
(m, 3H), 7.22–7.26 (m, 3H), 6.83 (d, J = 7.9
Hz, 1H), 5.36 (s, 2H), 3.80–3.91 (m, 5H), 3.04–3.11
(m, 1H), 2.94–3.00 (m, 1H), 2.47 (m, 4H), 2.35 (s, 3H), 2.15–2.25
(m, 1H), 1.98–2.06 (m, 1H), 1.70–1.81 (m, 1H). MS (m/z): [M + H]+ 505.2.
2-Chloro-4-phenylpyrimidine
(23b)
To a
mixture of 22 (50 mg, 0.336 mmol), phenylboronic acid
(41 mg, 0.336 mmol), sodium carbonate (110 mg in 0.5 mL of water),
and DME (2.5 mL) were added palladium acetate (3.8 mg, 0.0168 mmol)
and triphenylphosphine (8.8 mg, 0.0336 mmol). The reaction mixture
was heated at 95 °C for 20 h. Solvent was removed under reduced
pressure and the residue taken up into dichloromethane (20 mL), washed
with water (3 × 5 mL), dried over MgSO4, and concentrated
to give a residue which was purified by preparatory TLC (hexane/EtOAc,
8:2) to yield 41 mg (64%) of 23b. 1H NMR (500
MHz, CDCl3): δ 8.63 (d, J = 5.2
Hz, 1H), 8.06–8.11 (m, 2H), 7.64 (d, J = 5.3
Hz, 1H), 7.47–7.57 (m, 3H). 13C NMR (125 MHz, CDCl3): δ 167.2, 161.9, 159.8, 135.1, 131.9, 129.1, 127.4,
115.2.
To 24b (49 mg, 0.193 mmol)
in acetonitrile
(1.4 mL) were added TFA (59 μL, 88 mg, 0.772 mmol) and N-iodosuccinimide (43 mg, 0.193 mmol), and the resulting
solution was stirred at rt for 1 h. Solvent was evaporated, and the
residue was taken up into dichloromethane (15 mL), washed with 10%
Na2CO3 (2 × 5 mL) and water (5 mL), dried
over MgSO4, and concentrated to give a residue which was
purified by preparatory TLC (CH2Cl2/MeOH, 10:1)
to yield 67 mg (92%) of 25b. 1H NMR (500 MHz,
CDCl3): δ 8.60 (s, 1H), 7.65–7.73 (m, 2H),
7.42–7.49 (m, 3H), 3.86 (m, 4H), 2.45 (m, 4H), 2.33 (s, 3H). 13C NMR (125 MHz, CDCl3): δ 167.5, 165.8,
160.9, 140.3, 129.7, 129.3, 128.1, 76.2, 55.0, 46.4, 43.9. MS (m/z): [M + H]+ 380.9.
To a solution of 26b (5 mg,
0.0133 mmol) in THF (0.5 mL) were added Boc-glycine (2.3 mg, 0.0133
mmol) and DCC (3 mg, 0.0146 mmol). After the resulting solution was
stirred for 2 h at rt, THF was evaporated, and 0.5 mL of CH2Cl2/TFA (4:1) was added. The solution was stirred for
45 min and then concentrated to dryness under reduced pressure to
give a residue which was purified by preparatory TLC (CH2Cl2/MeOH–NH3 (7 N), 10:1) to afford
2 mg (35%) of 27g. 1H NMR (500 MHz, CDCl3/MeOH-d4): δ 8.45 (s, 1H),
7.69 (d, J = 8.2 Hz, 2H), 7.35–7.45 (m, 4H),
7.31 (m, 1H), 7.17 (t, J = 8.0 Hz, 1H), 6.76 (d, J = 7.8 Hz, 1H), 3.96 (m, 4H), 3.36 (s, 2H), 2.56 (m, 4H),
2.38 (s, 3H). MS (m/z): [M + H]+ 435.0.
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. Stephen Nimer (MSKCC). MDA-MB-468 and Mia-PaCa-2 cell lines
were purchased from ATCC. Cells were cultured routinely in DME/F12
(SKBr3, MDA-MB-468, Mia-PaCa-2) 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-Raf-1 from rabbit (1:500, sc-133, Santa
Cruz), anti-PARP (p85 fragment) from rabbit (1:500, G7341, Promega),
anti-β-actin from mouse (1:2500, A1978, Sigma-Aldrich), anti-Akt
from rabbit (1:500, 9272, Cell Signaling), anti-p-STAT3 (Y705) from
rabbit (1:500, 9145, Cell Signaling), anti-p-STAT5 (Y694) from rabbit
(1:500, 9351, Cell Signaling), anti-FLT3 form rabbit (1:500, sc-480,
Santa Cruz), anti-HOP from mouse (1:500, SRA-1500, Enzo), and anti-Hsp/c70
from mouse (1:2000, smc-106B, Stressmarq). 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.[36] 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).
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.[37]
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. 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.
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 Hsp/c70 complexes upon treatment.
Hsp70 Luciferase Refolding in Cells
The refolding of
heat-denatured luciferase was assayed as described.[35] HEK293 cells were transfected with 2 μg of luciferase
plasmid and 8 μg of Hsp70 plasmid or control vector per 60 mm
dish. HumanHsp70, Hsp70-C267S, and Hsc70 were N-terminally Flag-tagged,
and luciferase was C-terminally HA-tagged in pcDNA3.1. Two days after
transfection, cells were treated with cycloheximide (Sigma) at 50
μg/mL to inhibit protein synthesis, transferred to 45 °C
for 1 h, and then brought back to 37 °C for up to 2 h of recovery. 27c was added immediately before heat shock. Cell samples,
taken before heat shock and at 0, 1, and 2 h of recovery, were lysed
on ice with PBS containing 1% Triton X-100, and the insoluble material
was removed by centrifugation at 20000g. Total protein
amounts in the supernatants were determined using the BCA Protein
Assay Kit (Pierce). Luciferase enzymatic activities in the supernatants
were measured using the Luciferase Reporter Assay Kit (Promega) and
activity values normalized to total protein amounts.
Cycloheximide
Treatments
Cells were treated with cycloheximide
(at a final concentration of 100 μg/mL) with added vehicle (DMSO)
or 27c (20 μM) for the indicated times. Cells were
lysed as indicated above, and the resulting samples were analyzed
by Western blotting.
Authors: Melanie K Bhangoo; Stefan Tzankov; Anna C Y Fan; Kurt Dejgaard; David Y Thomas; Jason C Young Journal: Mol Biol Cell Date: 2007-06-27 Impact factor: 4.138
Authors: Allan Wissner; Elsebe Overbeek; Marvin F Reich; M Brawner Floyd; Bernard D Johnson; Nellie Mamuya; Edward C Rosfjord; Carolyn Discafani; Rachel Davis; Xiaoqing Shi; Sridhar K Rabindran; Brian C Gruber; Fei Ye; William A Hallett; Ramaswamy Nilakantan; Ru Shen; Yu-Fen Wang; Lee M Greenberger; Hwei-Ru Tsou Journal: J Med Chem Date: 2003-01-02 Impact factor: 7.446
Authors: Anna Rodina; Maria Vilenchik; Kamalika Moulick; Julia Aguirre; Joungnam Kim; Anne Chiang; Julie Litz; Cristina C Clement; Yanlong Kang; Yuhong She; Nian Wu; Sara Felts; Peter Wipf; Joan Massague; Xuejun Jiang; Jeffrey L Brodsky; Geoffrey W Krystal; Gabriela Chiosis Journal: Nat Chem Biol Date: 2007-07-01 Impact factor: 15.040
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: Mackenzie D Martin; Jeremy D Baker; Amirthaa Suntharalingam; Bryce A Nordhues; Lindsey B Shelton; Dali Zheng; Jonathan J Sabbagh; Timothy A J Haystead; Jason E Gestwicki; Chad A Dickey Journal: ACS Chem Biol Date: 2016-05-26 Impact factor: 5.100
Figure 1
Figure 2
Scheme 1
Scheme 3
Scheme 2
Figure 3
Figure 4
Figure 5