Structure-based design and synthesis of small molecule protein-tyrosine phosphatase 1B inhibitors

Article abstract of DOI:10.1016/S0968-0896(98)00140-0

Protein-tyrosine phosphatase (PTP) inhibitors are attractive as potential signal transduction-directed therapeutics which may be useful in the treatment of a variety of diseases. We have previously reported the X-ray structure of 1,1-difluoro-1-(2-naphtha

Full text of DOI:10.1016/S0968-0896(98)00140-0

BIOORGANIC &
MEDICINAL
CHEMISTRY
Bioorganic & Medicinal Chemistry 6 (1998) 1799±1810
Structure-based Design and Synthesis of Small Molecule
Protein±Tyrosine Phosphatase 1B Inhibitorsy
Zhu-Jun Yao,a Bin Ye,a Xiong-Wu Wu,b Shaomeng Wang,b Li Wu,c
Zhong-Yin Zhangc and Terrence R. Burke, Jr.a,*
aLaboratory of Medicinal Chemistry, Division of Basic Sciences, National Cancer Institute, National Institutes of Health,
Bethesda, MD 20892, USA
bInstitute for Cognitive and Computational Sciences, Georgetown University Medical Center, Washington, DC 20007, USA
cDepartment of Molecular Pharmacology, Albert Einstein College of Medicine, Bronx, NY 10461, USA
Received 19 March 1998; accepted 11 May 1998
AbstractÐProtein±tyrosine phosphatase (PTP) inhibitors are attractive as potential signal transduction-directed thera-
peutics which may be useful in the treatment of a variety of diseases. We have previously reported the X-ray structure
of 1,1-di¯uoro-1-(2-naphthalenyl)methyl] phosphonic acid (4) complexed with the human the protein±tyrosine phos-
phatase 1B (PTP1B) and its use in the design of an analogue which binds with higher anity within the catalytic site
(Burke, T. R., Jr. et al. Biochemistry 1996, 35, 15989). In the current study, new naphthyldi¯uoromethyl phosphonic
acids were designed bearing acidic functionality intended to interact with the PTP1B Arg47, which is situated just
outside the catalytic pocket. This residue has been shown previously to provide key interactions with acidic residues of
phosphotyrosyl-containing peptide substrates. Consistent with trends predicted by molecular dynamics calculations,
the new analogues bound with 7- to 14-fold higher anity than the parent 4, in principal validating the design ratio-
nale. # 1998 Elsevier Science Ltd. All rights reserved.
Introduction
been reported in the PTP inhibitor ®eld,8 in spite of the
attractiveness of these latter targets.
Protein±tyrosine kinase (PTK) dependent signal trans-
duction is achieved through modulating the phosphoryl-
ation states of tyrosyl residues (pTyr, 1) in key cellular
proteins. An imbalance in the PTK-mediated generation
of phosphotyrosyl (pTyr) residues versus dephos-
phorylation by protein±tyrosine phosphatases (PTPs)1±6
can contribute to a variety of diseases, making develop-
ment of both PTK and PTP inhibitors of potential
interest. While signi®cant progress has been made in the
preparation of PTK inhibitors,7 considerably less has
Key words: Amino acids and derivatives; mimetics; phosphonic
acids and derivatives; phosphorylation.
*Corresponding author. Tel.: (301) 402-1717; Fax: (301) 402-
2275; E-mail: tburke@helix.nih.gov
yA preliminary account of this work have been reported: Burke,
T. R., Jr.; Yao, Z. J.; Ye, B.; Wang, S.; Zhang, Z. Y. Structure-
based design of protein±tyrosine phosphatase inhibitors. Pre-
sented at the 213th National ACS Meeting, San Francisco, CA,
April 13±17, 1997; MEDI-313.
Utilizing the nonreceptor phosphatase PTP1B as a
model, we have approached PTP inhibitor development
by an iterative process of: (1) inhibitor design and
synthesis; (2) biological evaluation/enzyme±ligand
structure elucidation; and ®nally, (3) redesign, to pro-
gress from peptide-based inhibitors toward progressively
more potent small molecule analogues. Exemplary is
0968-0896/98/$Ðsee front matter # 1998 Elsevier Science Ltd. All rights reserved
PII: S0968-0896(98)00140-0

1800
Z.-J. Yao et al./Bioorg. Med. Chem. 6 (1998) 1799±1810
our initial discovery that replacement of the pTyr resi-
due in the peptide substrate ``D-A-D-E-pY-L''9 with the
hydrolytically stable pTyr analogue, di¯uorophos-
phonomethyl phenylalanine10,11 (F2Pmp, 2), resulted in
an extremely potent PTP1B inhibitor (IC50=100 mM).12
The importance of the di¯uorophosphonate moiety in
PTP binding interactions was dramatically illustrated by
the much reduced inhibitory potency of the corre-
sponding peptide substituted with the non¯uorine-con-
taining phosphonomethyl phenylalanine residue (Pmp,
3; IC50=200 mM). In subsequent studies we showed that
simple aryldi¯uoromethyl phosphonates lacking peptide
component, also retained reasonable inhibitory potency
if a polycyclic ring system, such as naphthyl (for exam-
ple [1,1-di¯uoro-1-(2-naphthalenyl)methyl] phosphonic
acid 4) was employed rather than the single phenyl ring
found in pTyr or F2Pmp.13 In order to understand the
structural basis for the binding of such polycyclic-con-
taining di¯uorophosphonates, we have solved the X-ray
structure of 4 complexed within the PTP1B catalytic
site, and based on the observed binding interactions a
new naphthyl analogue (5) was designed that contains a
hydroxyl group which mimics H2O molecules originally
bound to Arg47 at the lip of the pTyr binding pocket.
Using this in conjunction with our ®nding that F2Pmp
serves as a very potent pTyr mimetic in the context of
PTP inhibitors, and that naphthyl-containing di¯uoro-
methyl phosphonates can also bind with good anity,
we sought to prepare small molecule inhibitors based on
the ``Ac-E-pY-amide'' structure (6), in which either
F2Pmp or naphthyldi¯uoromethyl phosphonic acid
moiety 4 served as pTyr replacements. When using
F2Pmp as a pTyr replacement, dipeptide 7 resulted in a
straight forward fashion. For naphthyl-based analogues
using 4 as a pTyr replacement, it was less obvious how
the parent 4 should be modi®ed in order to allow
attachment of the Glu residue. The most direct
approach was to add a carboxyl group at the naphthyl
6-position to yield analogue 9, which was reminiscent of
a ring-constrained version of F2Pmp. However, since
compound 9 lacked critical a-amino functionality found
in the parent F2Pmp, coupling of 9 with a Glu residue
could not occur in a manner directly analogous to that
seen in parent dipeptide 7. Instead the ``pseudodipep-
tide'' 10 would result, in which the Glu residue e€ec-
tively becomes situated at the carboxyl-terminal side of
the construct, similar to the ``reverse order'' dipeptide 8.
The e€ects of this were unclear, since the di€erential
importance of residues situated amino versus carboxyl
terminal to the pTyr residue had previously been shown
for small phosphopeptide substrates.9 Additionally, the
original structural interaction of the Glu residue of the
parent D-A-D-E-pY-L peptide with Arg47 of the
PTP1B complex, was based on its location to the amino
side of the pTyr residue.16 Further adding to the uncer-
tainty of analogue 9, was that when viewed as a pTyr
mimetic, the embedded phenylpropanoyl side chain was
e€ectively constrained to coplanarity, while in the actual
PTP1B-bound peptide, the side chain of the pTyr resi-
due projected at an angle nearly perpendicular to the
plane of the phenyl ring.16 Because of these ambiguities
in the manner in which 10 would mimic parent 7, we
also designed the isomeric naphthyl-containing dipep-
tide mimetic 11. While 11 e€ectively places the di¯uoro-
phosphonate group in the wrong position relative to
parent F2Pmp-containing 7, our previous studies had
demonstrated that signi®cant latitude is permitted in the
placement of the di¯uorophosphonate group relative to
the naphthyl ring.13,14 In light of these considerations, it
also seemed appropriate to prepare the sequence-
reversed dipeptide 8 for purposes of comparison.
.
bound within the PTP1B 4 catalytic site. This modi®ca-
tion resulted in a twofold increase in inhibitory potency
of 5 relative to parent 4, thereby supporting the princi-
ple of structure-based design as a means of increasing
inhibitory potency.14
While these previous studies focused on enzyme inhib-
itor interactions within the pTyr binding cavity, it
became of equal interest to examine potential inter-
actions outside the actual catalytic site. It is recognized
that in small pTyr-containing substrates, acidic residues
proximal to the pTyr residue enhance binding anity.15
A structural basis for this empirical observation became
evident from the X-ray structure of PTP1B complexed
to the D-A-D-E-pY-L-amide hexapeptide, where it was
seen that a critical arginine residue (Arg47) situated just
external to the pTyr binding pocket, entered into key
ionic interactions with the side chain carboxyl groups
Glu and Asp residues situated in the pTyr-1 and pTyr-2
positions of the substrate peptide.16 Based on the
assumption that similar types of interactions with Arg47
could potentially be taken advantage of by small mole-
cule inhibitors, the present study was undertaken to
prepare new inhibitors of PTP1B using the structure of
parent naphthyldi¯uoromethyl phosphonate 4, which
could potentially interact with Arg47.
Synthesis
Inhibitor design
Central to the synthesis of target dipeptide mimetics 10
and 11 was the regiochemically controlled introduction
of carboxylate functionality onto the parent (di-tert-
butyl) (2-di¯uoromethyl)naphthyl phosphonate struc-
ture. For 10 and 11, this required carboxylation at the 6-
The work by Jia et al.16 highlighted the favorable bind-
ing interactions of the ``E-pY'' portion of the larger D-
A-D-E-pY-L-amide peptide, in which the pTyr residue
bound within the catalytic site and the Glu residue

Z.-J. Yao et al./Bioorg. Med. Chem. 6 (1998) 1799±1810
1801
and 7-positions of the naphthyl ring, respectively. Our
initial approach toward the synthesis of bis(tert-butyl)
aryl(di¯uoromethyl)phosphonates involved diethylamino-
sulfur tri¯uoride (DAST)-mediated ¯uorination of the
corresponding ketophosphonates, which themselves
were obtained by addition of (di-tert-butyl)phosphite to
the aryl aldehydes, followed by oxidation of the result-
ing a-hydroxyphosphonates.17 Although ketophos-
phonates have also been prepared by addition of
phosphites to acyl chlorides, we have been unsuccessful
at preparing tert-butyl-protected ketophosphonates by
this more direct route. Recently, a number of new
approaches toward the synthesis of aryl(di¯uoro-
methyl)phosphonates have appeared which do not rely
on intermediate ketophosphonates,18,19 including the
preparation of naphthyl bis[(di¯uoromethyl)phos-
phonates].20 However, none of these reported synthesis
provided tert-butyl-protected phosphonates, which we
desired in order to allow facile cleavage of phosphonate
protection under mild acidic conditions. Based on these
considerations, we decided to utilize our original keto-
phosphonate-route.
ketophosphonate approach, aldehydes 15 (Scheme 1)
and 29 (Scheme 2) became intermediate targets. Both 15
and 29 were prepared by oxidation of benzylic alcohols.
In the case of aldehyde 15, commercially available 2,6-
naphthalenedicarboxylic acid 12 was ®rst di€erentially
protected as the monobenzylic ester 13 and then reduced
to the corresponding benzylic alcohol 14, which upon
oxidation, yielded aldehyde 15. A somewhat more
lengthy route was utilized to obtain aldehyde 29.
Carboxymethylation of commercially available 2,7-di-
hydroxynaphthalene 20 was preceded by its activation
as the bistri¯ate 21 followed by exposure to carbon
monoxide under the catalysis of Pd(OAc)2 in the pre-
sence of Ph2P(CH2)3PPh2. Mono-deprotection of the
resulting 2,7-dicarbomethoxynaphthalene 22, which
would be required to di€erentiate the two benzylic
positions, was not readily achieved. However, after
reduction to diol 23 (LiAlH4), selective silation of a sin-
gle alcohol was possible, providing monobenzylic alco-
hol 24. This selective silation was achieved using
TBDMSCl and LiN(SiMe3)2 in THF, but not using
TBDMSCl and imidazole in DMF. Conversion of alco-
hol 24 to benzyl ester 27 was done in a two-step fashion
through aldehyde 25 (oxidation using MnO2) and car-
boxylic acid 26 (treatment with bu€ered NaClO2) fol-
lowed by alkylation with benzyl bromide (DBU in DMF).
Finally, removal of silyl protection (tetrabutylammonium-
Our synthesis of dipeptide mimetics 10 and 11 relied on
the preparation of naphthoic acids 19 and 33, respec-
tively, which would subsequently be coupled under
solid-phase protocol to resin-bound Glu residues
(Schemes 1 and 2). For synthesis of 19 and 33 by our
Scheme 1. Reagents and conditions: (a) DBU, BnBr, DMF, 17%; (b) BH3õTHF, THF, 84%; (c) MnO2, toluene, 100%; (d) (tBuO)2-
P(O)H, basic Al2O3, 91%; (e) Swern oxidation; (f) DAST, 59%; (g) 0.2 N LiOH, THF, 74%; (h) DIPCDI, HOBt, NMP; then TFA,
anisole; HPLC, 44%.

1802
Z.-J. Yao et al./Bioorg. Med. Chem. 6 (1998) 1799±1810
Scheme 2. Reagents and conditions: (a) Tf2O, 2,6-lutidine, DMAP, CH2Cl2-THF, 74%; (b) Pd(OAc)2, dppp, CO, MeOH, TEA,
DMSO, 84%; (c) LiAlH4, THF, 59%; (d) LiHMDS, TBDMSCl, 45%; (e) MnO2, toluene, 77%; (f) NaClO2, KH2PO4, 2-methyl-2-
butene, tBuOH±water, 98%; (g) DBU, BnBr, DMF, 92%; (h) TBAF, THF, 93%; (i) PCC, CH2Cl2, 95%; (j) (tBuO)2P(O)H,
LiHMDS, THF, 98%; (k) Swern oxidation, 71%; (l) DAST, 52%; (m) 0.2 N LiOH, THF; (n) DIPCDI, HOBt, NMP; TFA, anisole;
HPLC, 30%.
¯uoride) and oxidation (PCC) of the resulting benzylic
alcohol 28 gave desired aldehyde 29.
4
14 and the tetrapeptide ``Ac-D-E-pY-L'' (35).16 In both
cases, signi®cant conformational and energetic changes
were observed at higher temperatures (T>298 K), how-
ever at room temperature (T=298 K) structures
remained in low energy conformations throughout the
simulation intervals. Final trajectories were very close to
the actual X-ray structures in both cases, with root-
mean-square (RMS) values for all atoms being 1.03&
and 1.09& for complexes involving 4 and 35, respec-
tively. The most signi®cant conformational di€erences
arose from the phosphate-binding loop region. All
molecular dynamics simulations were conducted using a
stepwise protocol rather than at constant temperature.
In this fashion, local energy barriers were more readily
overcome, resulting in lower energy enzyme±inhibitor
complexes (see Experimental section). This was shown
by comparing results of stepwise dynamics simulations
with those obtained at constant temperature. For the
Isomeric aldehydes 15 and 29 were converted to tert-
butyl protected di¯uorophosphonates 18 and 33 respec-
tively, according to our published three-step proce-
dure.17 Synthesis of dipeptide mimetics 10 and 11 were
then achieved by HOBT active ester solid-phase techni-
ques using (Og-tert-butyl)glutamic acid attached to a
Rink amide resin.21 The importance of tert-butyl pro-
tection was then demonstrated, as treatment with TFA
resulted in concomitant cleavage from the resin and
removal of all protecting groups. Synthesis of F2Pmp-
containing dipeptides 10 and 11 was also achieved by
solid-phase techniques using Rink amide resin and Na-
Fmoc F2Pmp which lacked phosphonate protection.22
.
complex PTP1B 4 after 100 ps of simulation at a con-
Results and Discussion
Molecular modelling
stant temperature of 298 K, an average potential energy
1
of
203 kcal mol higher than the corresponding value of
9974 kcal mol
1
for ®nal trajectories, was
1
To demonstrate the validity of our stepwise protocol,
molecular dynamics simulations were performed on the
X-ray structures of PTP1B in complex alternatively with
10,077 kcal mol obtained by a stepwise simulation.
Similar potential energy di€erences were observed for
.
the complex PTP1B 35, indicating that the stepwise

Z.-J. Yao et al./Bioorg. Med. Chem. 6 (1998) 1799±1810
1803
simulation is more e€ective in delineating low energy
conformations, and this protocol was therefore
employed for all systems in the present study.
most probably contribute signi®cantly to the higher
binding anity of 10 relative to parent 4.
.
For the complex PTP1B 11, binding interactions of the
As previously described, naphthyl dipeptide mimetics 10
and 11 were designed empirically based on observed
favorable ``E-pY'' interactions in the X-ray structure of
a PTP1B-bound phosphopeptide.16 Once designed, the
ability of these analogues to interact both within the
pTyr binding pocket and with external elements outside
the pocket, noteably the Arg47 residue, were evaluated
by molecular dynamics simulations. Dynamics simula-
phosphonate, di¯uoromethylene, and naphthyl rings are
all quite similar to those observed for 4 and 10, with
major di€erences between 10 and 11 stemming from
hydrogen bonding patterns of inhibitor side chain func-
tionality. Similar to 10, the side chain carboxyl group of
11 could enter into a hydrogen bond with the backbone
amido group of Arg47, however unlike 10, the terminal
carboxamido group of 11 did not interact with the pro-
tein. The loss of these three potential hydrogen bonds
could contribute signi®cantly to the observed lower
binding anity of 11.
.
tions of the complex PTP1B 10 resulted in phosphonate,
di¯uoromethylene, and naphthyl ring orientations
assuming orientations quite similar to those observed
.
for the parent complex PTP1B 4. In both cases, the
phosphonate oxygens all form strong hydrogen bonds
to the protein, with one oxygen bonding to the back-
bone amide hydrogens of Ser216 and Ala217. A second
oxygen forms two hydrogen bonds with backbone
amide hydrogen and the side chain guanidino group of
Arg221. This oxygen also forms a third hydrogen bond
to the side chain hydroxyl of Ser215. The third phos-
phonate oxygen hydrogen bonds to the backbone
amides of Gly218, Ile219, and Gly220, while the two
geminal ¯uorine atoms hydrogen bond to the backbone
amido group of Phe182 and the side chain carboxamide
of Gln266. The naphthyl ring of 10 is sandwiched
between a number of hydrophobic residues, with Tyr46,
Val49, Ala217, and Ile219 lying on one face and Phe182
situated in an oblique fashion on the opposite face.
Dynamics simulations of PTP1B complexes with tri-
peptides 7 and 8, show that the phosphonate and
di¯uoromethylene groups bind with the enzyme in
fashions identical to those observed for 4, 10, and 11.
The phenyl rings of both 7 and 8 interact with the
hydrophobic side chains of Val49 and Ala217. However,
the Glu side chain carboxyl of 7 forms two hydrogen
bonds to the side chain guanidino group of Arg47, while
Glu carboxyl group of 8 forms only one such hydrogen
bond. The C-terminal Glu carboxamido group of 7
hydrogen bonds with the side chain carboxyl of Asp48.
In binding of naphthyl-based inhibitors 4, 10, and 11,
the Tyr46 aryl ring is twisted by at least 30 degrees
relative to that observed for peptide analogues 7, 8, and
35, in which ¯exible phenyl rings are present. Overall,
the improved binding observed for inhibitors correlated
well with formation of new hydrogen bonds, particu-
larly in the loop region between Tyr46 to Val49.
Of special interest in the present study, were potential
interactions between inhibitor functionality with resi-
dues outside the pTyr binding pocket, particularly with
Arg47. By design, it was hoped that the side chain car-
boxyl of 10 would be able to interact with the positively
charged guanidino group of the Arg47 residue. Mole-
cular dynamics simulations indicated that while such
interactions could occur readily, they were highly rever-
sible in nature, with rapid on/o€ exchange occurring
between guanidino, carboxyl and solvent. Such ¯uidity
was not totally unexpected in a surface phenomena of
this nature. In some tragetories of the dynamics simula-
tion, binding of 10 occurred with its side chain carboxyl
group interacting with water. At the same time, three
hydrogen bonds formed between its terminal carbox-
amido group and the enzyme; one between its amido
hydrogens and the side chain carboxyl group of Asp48
and one each between the carbonyl and the backbone
amide groups of Asp48 and Val49. In other trajectories,
the carboxyl group interacts directly with the Arg47
guanidino group. The appended carboxyl group of 10
also interacts well with the protein, forming one hydro-
gen bond with the backbone amido group of Arg47.
Taken together, these newly introduced hydrogen bonds
Biological evaluation
The e€ect of the aryldi¯uromethylphosphonates on the
PTP1B catalyzed p-nitrophenyl phosphate (pNPP)
hydrolysis reaction was examined at 30 ꢀC and pH 7.
The small aryldi¯uromethylphosphonates examined in
this study inhibited the PTP reaction and the mode of
inhibition was competitive with respect to the substrate
(data not shown). The Ki values are summarized in
Table 1. There are several points that are worth dis-
cussing. We have previously found that F2Pmp-con-
taining peptides are potent inhibitors of PTP1B.14,23 It
has been suggested, based on the X-ray structure of a
complex between PTP1B and a peptide substrate, that
the recognition pocket for pTyr provides the dominant
driving force for peptide substrate binding.16 Thus, we
thought that the aryldi¯uromethylphosphonate moiety
might represent a good starting point for designing low
molecular weight, nonpeptide PTP inhibitors. While the
simplest aryldi¯uromethylphosphonate, a,a-di¯uoro-
benzyl phosphonic acid (34),13,17 exhibited a Ki value of

1804
Z.-J. Yao et al./Bioorg. Med. Chem. 6 (1998) 1799±1810
only 2.5 mM, adding a second phenyl ring (compound
4,13,14 Ki=179 mM) enhanced the anity 14-fold. This
suggests that there is considerable plasticity in the pTyr
binding pocket, and that appropriately positioned
hydrophobic functionalities can improve the binding
ecacy. Indeed, the structure of PTP1B complexed with
4 shows that the naphthalene ring forms speci®c hydro-
phobic interactions with side chains of active site resi-
dues.14
PTP1B have shown that Km values for free pTyr or aryl
phosphates are 3- to 4-orders of magnitude higher than
the best peptide substrates, indicating that amino acid
residues ¯anking the pTyr moiety are important for
ecient PTP1B catalysis.9 Furthermore, it has been
shown that acidic residues located proximal to the
amino side of the pTyr residue are important for sub-
strate recognition.15,24 The recently solved crystal struc-
ture of the active site Cys215 to Ser mutant of PTP1B
complexed with D-A-D-E-pTyr-l-amide reveals the
structural basis for the favorable e€ect of acidic resi-
dues.16 There are speci®c interactions between the acidic
side chains (at -1 and -2 positions) of the substrate and
the basic residue, Arg47 of the enzyme. These results
suggest that the binding of peptide substrate/inhibitor is
a cooperative event that involves the recognition of both
pTyr/F2Pmp functional groups as well as the structural
features surrounding residues.
The anity of a,a-di¯uorobenzyl phosphonic acid 34 is
nearly 14,000-fold lower than Ac-D-A-D-E-F2Pmp-l-
amide (Ki=0.18 mM).23 This result also implies that
structural features in addition to the F2Pmp motif are
required for high anity binding. Kinetic studies with
Table 1. Inhibition constants of inhibitors assayed against
PTP1B as described in the Experimental section
Ki‹S.D. (mM)
For high anity small molecule PTP1B inhibitors,
additional molecular features are required beyond the
simple arylmethyl di¯uorophosphonate structure. We
have appended various acidic functionalities onto
F2Pmp (generating 7 and 8) and the naphthyl di¯uoro-
methylphosphonate 4 (generating 9±11) to allow extra
binding interactions with residues outside the catalytic
site, especially Arg47. Relative to a,a-di¯uorobenzyl
phosphonic acid 34, the dipeptide analogue Ac-Glu-
F2Pmp-amide (7) displayed a 1240-fold higher anity
toward PTP1B. This is remarkable, since the hexapep-
tide analogue Ac-Asp-Ala-Asp-Glu-F2Pmp-amide is
only 11-fold better than the dipeptide. In contrast, the
anity of the ``reverse order'' dipeptide analogue Ac-
F2Pmp-Glu-amide (8) is reduced by 22-fold in compar-
ison with Ac-Glu-F2Pmp-amide (7). Thus, it is most
likely that the Glu side chain situated amino-terminal to
F2Pmp in 7 would be able to interact with the side chain
of Arg47, while the Glu side chain situated carboxy-
terminal to F2Pmp in 8 would not be able to make the
same interaction as favorably. This indicates that direc-
tionality is important for substrate/inhibitor recognition
by PTP1B, even for peptides as small as two amino
acids.
2,500‹120
179‹25
2.0‹0.14
43.4‹2.2
22.2‹1.1
12.4‹0.9
25.0‹2.0
Analogue 9 mimics a ring-constrained version of F2Pmp
(2). An eightfold enhancement in potency was observed
for analogue 9 relative to the parent naphthyl di¯uor-
omethyl phosphonate 4. Conceptually similar naphtha-
lene bis-di¯uorophosphonates have also recently been
reported to exhibit potent PTP1B inhibition.25 Since the
carboxyl group in 9 would not be expected to make
direct interaction with the side chain of Arg47, the
increased binding may arise from the interactions of the
carboxylate with other residues in PTP1B or with Arg47
through bridged water molecules. Because 9 lacks the
amino group found in F2Pmp, coupling of 9 with a Glu

Z.-J. Yao et al./Bioorg. Med. Chem. 6 (1998) 1799±1810
1805
residue could not occur in a manner similar to that seen
in dipeptide 7. Instead the ``pseudodipeptide'' 10 was
prepared in which the Glu residue e€ectively becomes
situated at the carboxy-terminal side of the naphthyl
di¯uoromethylphosphonate 4, similar to the ``reverse
order'' dipeptide 8. Because the exact interaction of 10
with PTP1B could not be predicted with certainty, the
isomeric naphthyl di¯uoromethyl phosphonate-contain-
ing dipeptide mimetic 11 was also prepared to introduce
variation in the orientation of the acidic side chain
relative to the naphthyl ring. As shown in Table 1,
compound 10 exhibited a Ki value of 12.4 mM, which was
twofold more potent than 11. Since the parent naphthyl
di¯uoromethylphosphonate 4 is 14-fold more potent
than a,a-di¯uorobenzyl phosphonic acid 34, the fact
that compound 10 is sixfold less potent than 7 suggests
that interaction of the Glu side chain in 10 with Arg47
may be less than optimal. Since molecular modelling
predictions indicate that interaction of the inhibitors
side chain carboxyl with the Arg47 guanidino group
should readily occur, it was only logical to actually pre-
pare and test the compound. That the binding anity of
10 turned out to be lower than would be expected based
on optimum interaction with the Arg47 guanidino
group, may be rationalized as resulting from competi-
tion between the carboxyl and solvent for interaction
with the guanidino group as suggested by dynamics
simulations. This provides an example where actual
experimental fact serves to place perspective on model-
ling predictions.
PTP1B was diluted 1:100 into 1 L of 2ÂYT medium
containing 100 g/mL ampicillin. The culture was grown
at 37 ꢀC until absorbance at 600 nm reached 0.6, at
which point the cells were induced with 0.4 mM iso-
propyl ¯-d-thiogalatoside for 6 h. The cells were har-
vested by centrifugation and stored at 20 ꢀC. The
frozen cell pellet was thawed at room temperature and
resuspended in 30 mL of ice-cold bu€er A (100 mM 2-(4-
morpholino)-ethane sulfonic acid, pH 6.5, 1 mM DTT)
and lysed by two passes through a French press at
1300 psi. All of the following steps were then carried out
at 4 ꢀC. The lysate was centrifuged at 15,000 rpm
(Dupont SS-34 rotor) for 30 min. The supernatant was
incubated with 50 mL of CM-Sephadex C50 equili-
brated with bu€er A and shaken gently for 40 min. The
resin was washed three times with the same volume of
bu€er A, loaded onto a column and washed again with
10 bed volumes of bu€er A. The protein was then eluted
from the column by a linear gradient from 0 to 0.5 M
NaCl in 200 L of bu€er A. The positions of the peak
fractions were assessed by Coomassie Blue staining of
SDS-PAGE. Peak fractions which contain homo-
geneous PTP1B were combined into a ®nal pool of
protein, which was then concentrated to 30 mg/mL.
Phosphatase assay. The PTP1B phosphatase activity
was assayed at 30 ꢀC in a reaction mixture (0.2 mL)
containing appropriate concentrations of p-nitrophe-
nylphosphate (pNPP) as substrate. The bu€er used was
pH 7.0, 50 mM 3,3-dimethylglutarate, 1 mM EDTA.
The ionic strength of the solution was kept at 0.15 M
using NaCl. The reaction was initiated by addition of
enzyme and quenched after 2±3 min by addition of 1 mL
of 1 N NaOH. The nonenzymatic hydrolysis of the
substrate was corrected by measuring the control with-
out the addition of enzyme. The amount of product p-
nitrophenol was determined from the absorbance at
Experimental
Biological evaluation
Recombinant human PTP1B. The cDNA encoding the
catalytic domain of human PTP1B (amino acids 1 to
321) was obtained using the polymerase chain reaction
(PCR) from a human fetal brain cDNA library (Strata-
gene). The PCR primers used were 50-AGCTGGATC-
CATATGG AGATGGAAAAGGAGTT (encoding
both a BamHI and a NdeI site), and 30-ACGCGAAT-
TCTTAATTGTGTGGCTCCAGGATTCG (encoding
an EcoRI site). The PCR product was digested with
BamHI and EcoRI and subcloned into a pUC118 vec-
tor. The PTP1B coding sequence was con®rmed by
DNA sequencing. The coding region for PTP1B was
then cut from pUC118-PTP1B with BamHI and EcoRI
and ligated to the corresponding sites of plasmid pT7-7.
The PTP1B coding sequence was placed in frame
downstream of the phage T7 RNA polymerase pro-
moter at the NdeI site of pT7-7 to provide the transla-
tional initiation at Met 1 of PTP1B. The resulting
plasmid pT7-7/PTP1B was used to transform E. coli
BL21(DE3). An overnight culture of BL2(DE3) cells
with the pT7-7 expression vector containing the mutant
405 nm using
a
molar extinction coecient of
18,000 M 1 cm 1. Steady state kinetic parameters were
evaluated by ®tting directly the v versus [S] data to the
Michaelis±Menten equation using KINETASYST
(IntelliKinetics, State College, PA, USA).
Inhibition constant (Ki) determination. Inhibition con-
stants for the small PTP inhibitors was determined for
PTP1B in the following manner. The initial rate at eight
di€erent pNPP concentrations (0.2±5 Km) was measured
at three di€erent ®xed inhibitor concentrations.26 The
inhibition constant and inhibition pattern were eval-
uated using a direct curve-®tting program KINETA-
SYST.
Molecular modelling
All molecular modelling was performed using Insight II
(version 95.5)27 on Silicon Graphics Indigo2 R10000

1806
Z.-J. Yao et al./Bioorg. Med. Chem. 6 (1998) 1799±1810
work stations running IRIX (version 6.2), with mole-
cular dynamics simulations and energy minimizations
being conducted using the Discover 3 module. C€91
force ®eld and charge templates were employed, as were
cell-multipole methods, for calculation of nonbonded
interactions.28±30 The previously determined X-ray
structures of PTP1B in complex with the phosphopep-
tide ligand, Ac-D-E-pY-L (35)16 or with the naphtha-
lene di¯ourophosphonate 4,14 provided the structural
basis for initial alignment of inhibitors examined in this
study. Speci®cally, the initial structure of PTP1B in
complex with 7 was constructed starting from the
tions at T=398 K; (5) 40 ps of molecular dynamics
simulations at T=348 K; (6) 100 ps of molecular
dynamics simulations at T=298 K. The ®rst round of
minimizations was intended to eliminate high energy
overlapping resulting from initial complex construction,
while the subsequent four rounds of dynamics simula-
tions were conducted as annealing processes to over-
come potential energy barriers between di€erent binding
modes, thereby achieving a low energy state for the sys-
tem. The ®nal simulation at 298 K was undertaken to
achieve an equilibrated distribution of structural and
energetic properties for the enzyme±ligand complex.
.
reported PTP1B 35, by substituting the phosphate ester
oxygen bridge with di¯uoromethylene unit, replacing
the Ac-Asp portion with an acetyl group and deleting
the Leu residue, leaving the C-terminal primary tyrosyl-
General synthetic methods
Melting points were determined on a Mel Temp II
melting point apparatus and are uncorrected. Elemental
analyses were obtained from Atlantic Microlab, Nor-
cross, GA, and fast atom bombardment mass spectra
(FABMS) were acquired with a VG Analytical 7070E
mass spectrometer under the control of a VG 2035 data
.
amide. Similarly, the initial PTP1B 8 complex was
obtained by changing the con®guration of the a-carbon
.
.
.
of PTP1B 7. Starting PTP1B 10 and PTP1B 11 com-
.
plexes were derived from the published PTP1B 4 struc-
ture by appending appropriate functionality to the
naphthyldi¯uorophosphonate ring. Hydrogens were
added to heavy atoms according to their hybridization,
with protonation states of ligand and protein residues
being assigned at pH 7, unless otherwise stated. Under
these conditions, the phosphate group is present in its
di-deprotonated form. Each complex was solvated with
approximately 1500 TIP3P waters contained within a
27& radius sphere centered at the site of phosphate or
phosphonate attachment to the aryl ring.
1
system. H NMR data were obtained on Bruker AC250
(250 MHz) or if indicated, AMX500 (500 MHz) instru-
ments and are reported in ppm relative to TMS and
referenced to the solvent in which they were run. Infra-
red spectral were acquired on a Perkin-Elmer 1600 series
FTIR instrument. Solvent was removed by rotary evap-
oration under reduced pressure and silica gel chroma-
tography was performed using Merck silica gel 60 with a
particle size of 40±63 m. Anhydrous solvents were
obtained commercially and used without further drying.
Preparative HPLC were conducted using a Vydac pre-
parative C18 peptide and protein column. Dipeptides 7
and 8 were prepared by solid-phase techniques.
Since our primary goal was to achieve an understanding
of the interactions between PTP1B and the ligands,
molecular dynamics simulations were focused on the
binding region. In order to avoid boundary e€ects,
boundary regions of PTP1B complexes were excluded
from simulations. Based on the X-ray structures of
PTP1B both free and in complex with 4 and 35, it is
apparent that upon ligand binding, the enzyme does not
undergo signi®cant conformational change outside the
binding region. Based on these considerations, each
inhibitor±ligand complex was divided into both move-
able and ®xed regions. Moveable regions included the
ligand itself as well as all protein residues and water
molecules (approximately 380) within 20& of the
ligand. The remainder of the system was ®xed, meaning
that atoms participated in interactions, but did not
move. In this fashion, computational speed was sig-
ni®cantly enhanced, while focusing attention on the
binding region. For each study, a sequential stepwise
process of minimization and simulation was performed,
consisting of the following: (1) 1000 steps of minimiza-
tion using a conjugate-gradient method (Polak-Ribiere);
(2) 10 ps of molecular dynamics simulations at
T=598 K; (3) 20 ps of molecular dynamics simulations
at T=498 K; (4) 30 ps of molecular dynamics simula-
2,6-Naphthalenedicarboxylic acid monobenzyl ester (13).
A suspension of 2,6-naphthalendicarboxylic acid 12
(10.8 g, 50.0 mmol) in anhydrous DMF (500 mL) with
1,8-diazabicyclo[5.4.0]undec-7-ene
(DBU;
7.5 mL,
50 mmol) was heated to near boiling, then stirred at
room temperature. To the resulting ®ne suspension was
added benzyl bromide (6 mL, 50 mmol), portionwise
over 10 min, giving a solution. After three days the
solution was diluted with 0.5 N HCl/brine (1 L), extrac-
ted with EtOAc and the combined organic extracts were
washed with H2O, dried (MgSO4) and taken to dryness
to yield an o€-white solid. The solid was suspended in
ether, diluted with an equal volume of petroleum ether,
then collected, washed with petroleum ether, and dried,
yielding a white solid (8.11 g). The solid was dissolved in
EtOAc (50 mL) and NEt3 (4 mL, 29 mmol) was added
with slight warming to e€ect total dissolution of all
solid. This was applied to a 6.5-cm sintered glass funnel
containing 200 mL of a 1/1 mixture of 5±25 m silica and
63±200 mesh basic alumina oxide. Initial elution with
EtOAc provided 2,6-naphthalenedicarboxylic acid dibenzyl

Z.-J. Yao et al./Bioorg. Med. Chem. 6 (1998) 1799±1810
1807
ester. Subsequent elution with EtOAc/formic acid (99/1)
brought desired product. Evaporation of solvent and
suspension of the resulting solid in petroleum ether,
followed by collection and drying of solid gave pure 13
as white crystals (2.62 g, 17%): mp 242±243 ꢀC (dec);
1H NMR (DMSO-d6) d 8.75 (s, 1H), 8.69 (s, 1H), 8.21±
8.30 (m, 2H), 8.12±8.03 (m, 2H), 7.56±7.38 (m, 5H), 5.45
Di-(tert-butyl) [(6-(benzyloxycarbonyl)naphth-2-yl)di¯uoro-
methyl]phosphonate (18). To 5 mL of anhydrous
CH2Cl2 in a dry 100 mL rb at 78 ꢀC under argon was
added oxalyl chloride, 2.0 M in CH2Cl2 (1.39 mL,
2.79 mmol). A solution of DMSO (394 L, 5.56 mmol) in
anhydrous CH2Cl2 (3 mL) was then added, followed by
a solution of 16 (674 mg, 1.39 mmol) in anhydrous
CH2Cl2 (5 mL). After stirring at 78 ꢀC (35 min) NEt3
(1.40 g, 13.9 mmol) was added and stirring continued at
78 ꢀC (10 min), then the reaction was then transferred
to an ice bath and stirred at 0 ꢀC (20 min). The mixture
was partitioned between ice-cold 0.2 N HCl/brine
(200 mL) and EtOAc (3Â50 mL), dried (MgSO4), then
taken to dryness to provide ketophosphonate 17 as a
yellow syrup in quantitative yield which was then
¯uorinated directly: 1H NMR (CDCl3) d 9.05 (s, 1H),
8.58 (s, 1H), 8.17±7.92 (m, 4H), 7.47±7.29 (m, 5H), 5.38
(s, 2H), 1.50 (s, 18H).
1
.
(s, 2H). Analysis (C19H14O4 /8H2O): C, H.
6-Hydroxymethyl-2-naphthalenedicarboxylic acid benzyl
ester (14). To a suspension of 13 (2.89 g, 9.4 mmol) in
anhydrous THF (50 mL) under argon at room tem-
.
perature, was added 1.0 M BH3 THF (11.3 mL). The
resulting light brown solution was stirred overnight,
then partitioned between brine (200 mL) and EtOAc
(3Â75 mL), dried (MgSO4), and taken to dryness to
yield a white crystalline solid. Trituration with petro-
leum ether and drying provided 14 suciently pure for
further use (2.31 g, 84%), mp 88±94 ꢀC. An analytical
sample provided mp 98±99 ꢀC: H NMR (DMSO-d6) d
The crude 17 was cooled to 78 ꢀC then DAST (923 L,
1.13 g, 7 mmol) was added and the reaction ¯ask swirled
at room temperature until a solution formed. The reac-
tion mixture was then stirred overnight on an ice bath
which was allowed to come to ambient temperature
gradually. The reaction mixture was cooled to 78 ꢀC,
diluted with CHCl3 (50 mL) and shaken with ice-cold
saturated NaHCO3 (100 mL). The organic layer was
combined with CHCl3 back-extracts of the aqueous
phase and washed with ice-cold saturated NaHCO3
(100 mL), then dried (MgSO4), and taken to dryness to
yield a yellow syrup. Puri®cation by silica gel chroma-
tography (petroleum ether containing ®rst 10%, then
30% EtOAc) provide pure 18 as light yellow crystals
(390 mg, 59%), mp 91.5±92.5 ꢀC: A sample was recrys-
1
8.65 (s, 1H), 8.13±7.93 (m, 4H), 7.59±7.38 (m, 6H), 5.48±
5.43 (m, 2H), 4.71 (d, 2H, J=5.6 Hz). Analysis
(C19H16O3): C, H.
6-Formyl-2-naphthalenedicarboxylic acid benzyl ester (15).
A mixture of 14 (2.0 g, 6.8 mmol) and activated MnO2
(5.91 g, 68 mmol) in toluene (75 mL) was stirred at re¯ux
under argon (30 min), then ®ltered through Celite. The
Celite was washed with EtOAc and combined organics
taken to dryness to yield product 15 as an oil which
crystallized (2.0 g, 100%). An analytical sample was
obtained by recrystallization from ether/petroleum
ether, mp 75.5±78.5 ꢀC: 1H NMR (DMSO-d6) d 10.21 (s,
1H), 8.77 (s, 1H), 8.68 (s, 1H), 8.34±8.29 (m, 2H), 8.04
(dd, 1H, J=1.5, 8.6 Hz), 7.98 (dd, 1H, J=1.5, 8.6 Hz),
7.58±7.39 (m, 5H), 5.45 (s, 2H). Analysis (C19H14O3):
C, H.
tallized for analysis (EtOAc/hexanes) mp 92±93 ꢀC: H
1
NMR (CDCl3) d 8.60 (s, 1H), 8.11±8.04 (m, 2H), 7.95±
7.87 (m, 2H), 7.67 (d, 1H, J=8.6 Hz), 7.46±7.30 (m, 5H),
5.37 (s, 2H), 1.40 (s, 18H). Analysis (C27H31O5PF2):
C, H.
Di-(tert-butyl) [(6-(benzyloxycarbonyl)naphth-2-yl)hydroxy-
methyl]phosphonate (16). A suspension of 15 (580 mg,
2.0 mmol) in di-(tert-butyl) phosphite (776 mg, 4.0 mmol)
was warmed at 75 ꢀC under argon. To the resulting
solution was added basic Woelm aluminum oxide (4.0 g)
and the mixture then maintained at 75 ꢀC (1 h). The
aluminum oxide was then extracted well with EtOAc/
MeOH (1/1) and the combined extracts taken to dryness
to yield a light yellow syrup. Silica gel chromatographic
puri®cation (CHCl3) ®rst eluted starting 15 (213 mg)
followed by product 16, which was obtained as a white
crystalline solid (560 mg, 91% based on recovered start-
Solid-phase synthesis of 10
To 18 (339 mg, 0.72 mmol) in THF (15 mL) at room
temperature was added 0.2 N LiOH (10.8 mL) dropwise
and the reaction mixture then stirred (3.5 h). The mix-
ture was chilled, then partitioned between ice-cold 0.2 N
HCl/brine (200 mL) and EtOAc (2Â100 mL), dried
(MgSO4), then taken to dryness to yield crude 19
(222 mg, 74%): 1H NMR (DMSO-d6) d 8.69 (s, 1H),
8.30±8.15 (m, 3H), 8.06 (dd, 1H, J=1.5, 8.6 Hz), 7.69 (d,
1H, J=8.6 Hz), 1.41 (s, 18 Hz). Solid-phase coupling of
crude 19 was achieved directly using manual techniques.
A total of 0.2 mequiv of N-Fmoc Rink amide resin
(Bachem, 0.46 mequiv/g) was washed well with several
2 mL portions of N-methyl-2-pyrolidoinone (NMP), then
the Fmoc amino protection was removed by treatment
ing material), mp 154 ꢀC soften, 162±165 ꢀC: H NMR
1
(DMSO-d6) d 8.65 (s, 1H), 8.10 (d, 1H J=8.6 Hz), 8.50±
7.97 (m, 3H), 7.68 (d, J=8.6 Hz), 7.56±7.50 (m, 2H),
7.48±7.35 (m, 3H), 6.2±6.1 (m, 1H), 5.43 (s, 2H), 4.95±
4.85 (m, 1H), 1.36 (s, 9H), 1.33 (s, 9H). Analysis
.
(C27H33O6P 0.8H2O): C, H.

1808
Z.-J. Yao et al./Bioorg. Med. Chem. 6 (1998) 1799±1810
with 20% piperidine in NMP (2 mL, 20 min). The
deblocked resin was washed well with NMP (10Â2 mL),
then coupled overnight with a solution of active ester
formed by reacting 0.5 mmol each of N-Fmoc-glutamic
acid g-tert-butyl ester, 1-hydroxybenzotriazole (HOBT),
and 1,3-diisopropylcarbodiimide (DIPCDI) in NMP
(2 mL, 10 min). The resin was washed with NMP
(10Â2 mL), and the amino Fmoc-protection was
removed by treatment with 2 mL 20% piperidine in
NMP (20 min). A solution of 19 (83 mg, 0.2 mmol) in
NMP (2 mL) was activated by treatment with HOBT
(27 mg, 0.2 mmol) and DIPCDI (31 L, 0.2 mmol) at
room temperature (10 min), then coupled with the resin
(4 h). The resin was washed (5Â2 mL NMP; 5Â2 mL
CH2Cl2) and dried. A 50% portion of the dried resin
was swollen with CH2Cl2 then cleaved with tri-
¯uoroacetic acid (TFA) containing 5% anisole (5 mL,
30 min). The resulting supernatant was taken to dryness,
dissolved in acetonitrile/H2O (1/1), and puri®ed twice by
HPLC (retention time, 10.2 min; linear gradient 0±50%
B over 30 min) to provide 10 as a white solid (19 mg,
44%): 1H NMR (D2O) d 8.28 (brs, 1H), 8.08 (s, 1H),
7.99 (d, 1H, J=8.6 Hz), 7.98 (d, 1H, J=8.6 Hz), 7.75
(dd, 1H, J=8.6, 1.8 Hz), 7.64 (d, 1H, J=8.6 Hz), 4.45
(m, 1H), 2.49 (m, 2H), 2.1 (m, 2H); FABMS ( VE,
glycerol matrix) m/z=429.059 (calcd M-H, 429.066).
provide crude product. Puri®cation by silica-gel chro-
matography (hexane/EtOAc, 95/5) a€orded 22 as a
white solid (2.64 g, 84%): mp 136.5±137.5 ꢀC; H NMR
1
(CDCl3) d 8.68 (2H, brs), 8.16 (2H, dd, J=8.6, 1.3 Hz),
7.91 (2H, d, J=8.6 Hz), 3.99 (6H, s). Analysis
(C14H12O4): C, H.
2,7-Di(hydroxymethyl)naphthalene (23). To a suspension
of LiAlH4 (3.28 g, 86.4 mmol) in THF (52 mL) was
added 22 (2.64 g, 10.8 mmol) and the reaction mixture
stirred at room temperature (3 h), then quenched with
1 N HCl (30 mL) at 0 ꢀC. The mixture was then extrac-
ted with ether (3Â50 mL), washed with brine
(2Â15 mL), dried (Na2SO4), and taken to dryness to
yield crude product. Puri®cation by silica-gel chroma-
tography (CH2Cl2/MeOH, 98/2) yielded 23 as a white
solid (1.2 g, 59%): mp 158±160 ꢀC; H NMR (CDCl3) d
1
7.83 (2H, d, J=8.3 Hz), 7.79 (2H, brs), 7.67 (2H, dd,
J=8.3, 1.3 Hz), 4.85 (4H, d, J=3.9Hz), 1.54 (2H, s).
Analysis (C12H12O2): C, H.
2,7-Di(hydroxymethyl)naphthalene mono-tert-butyldimethyl-
silyl ether (24). To a solution of compound 23 (2.33 g,
12.4 mmol) in THF (62 mL) at 0 ꢀC was added dropwise
LiN(TMS)2 (12.4 mL, 1.0 M in THF). After 30 min,
DMAP and TBDMSCl was added in one portion and
the mixture stirred at room temperature (overnight),
then quenched with 0.2 N HCl (30 mL), extracted with
EtOAc (3Â50 mL), washed with brine (2Â30 mL), dried
(Na2SO4), and taken to dryness to yield crude product.
Puri®cation by silica gel chromatography (hexane/
EtOAc, 4/1) provided 24 as a white solid (1.7 g, 45%):
mp 41.5±43.5 ꢀC; 1H NMR (CDCl3) d 7.82±7.76 (4H,
m), 7.42 (2H, dt, J=1.6 Hz, 8.2 Hz), 5.28 (2H, s), 5.25
(2H, d, J=5.8 Hz), 1.54 (1H, s), 0.95 (9H, s), 0.11 (6H,
s). Analysis (C18H26SiO2) C, H.
.
Analysis (C17H17N2O7PF2 H2O): C, H, N.
2,7-Dihydroxynaphthalene ditri¯ate (21). To a mixture
of 2,7-dihydroxynaphthalene 20 (13.5 g, 84.4 mmol), 2,6-
lutidine (19.9 g, 186 mmol) and DMAP (2.06 g,
17 mmol) in CH2Cl2 (140 mL) and THF (140 mL) at
78 ꢀC was added Tf2O (50 g, 177 mmol) dropwise over
30 min. The reaction mixture was stirred ®rst at 78 ꢀC
(2 h), then at 0 ꢀC (5 h), and then carefully quenched
with saturated NaHCO3 (250 mL). It was extracted with
CH2Cl2 (3Â200 mL), washed with brine (2Â50 mL),
dried (Na2SO4) and taken to dryness to provide crude
product which was puri®ed by silica-gel chromato-
graphy (hexane/EtOAc, 4/1) to a€ord 21 as a white
solid (26.4 g, 74%). An analytical sample was obtained
by crystallization from hexane/EtOAc (20/1) mp 61.5±
62.5 ꢀC: 1H NMR (CDCl3) d 7.99 (2H, d, J=9.1 Hz),
7.8 (2H, d, J=2.3 Hz), 7.47 (2H, dd, J=9.1, 2.3 Hz).
Analysis (C12H6F6O6S2): C, H.
7-(tert-Butyldimethylsilyoxymethyl)naphthalene-2-carbox-
aldehyde (25). To a solution of 24 (7.01 g, 23.3 mmol) in
toluene (200 mL) was added MnO2 (20.0 g, 230 mmol) at
room temperature. The mixture was heated to re¯ux
(40 min), then brought to room temperature and ®l-
tered. The ®ltrate was taken to dryness to provide crude
product, which was puri®ed by silica gel ¯ash chroma-
tography (hexane/EtOAc, 4/1) to provide 25 as a white
solid (5.34 g, 77%): mp 38±39.5 ꢀC; 1H NMR (CDCl3) d
10.83 (1H, s), 9.00 (1H, brs), 8.62±8.52 (4H, m), 8.26
(1H, dd, J=8.6, 1.5 Hz), 5.6 (2H, s), 1.66 (9H, s), 0.83
(6H, s). Analysis (C18H24SiO2) C, H.
2,7-Naphthalenedicarboxylic acid dimethyl ester (22). A
¯ask containing a mixture of 21 (5.45 g, 12.9 mmol),
Pd(OAc)2 (352 mg, 1.57 mmol,), and Ph2P(CH2)3PPh2
(648 mg, 1.57 mmol) was degassed and recharged three
times with CO: then MeOH (12.6 g, 393 mmol), DMSO
(52 mL), and NEt3 (4.77 g, 47.1 mmol), were successively
added. This mixture was kept at 70 ꢀC (5 h), then cooled
to room temperature, and treated with brine (70 mL),
extracted with EtOAc (3Â80 mL), washed with brine
(2Â10 mL), dried (Na2SO4), and taken to dryness to
7-(tert-Butyldimethylsilyoxymethyl)naphthalene-2-carb-
oxylic acid (26). To a mixture of 25 (5.55 g, 18.5 mmol),
KH2PO4 (3.28 g, 24.1 mmol), 2-methyl-2-butene (3.25 g,
46.3 mmol), t-BuOH (92.5 mL), and H2O (85 mL) at
room temperature, was added NaClO2 (2.18 g,
24.1 mmol) in one portion and the reaction mixture was

Z.-J. Yao et al./Bioorg. Med. Chem. 6 (1998) 1799±1810
1809
stirred at room temperature overnight. The t-BuOH was
reduced in volume under reduced pressure, then the pH
was adjusted to 5±6 by addition of 2 N HCl. The mix-
ture was extracted with EtOAc (4Â50 mL), washed with
brine (2Â20 mL), dried (Na2SO4), and taken to dryness
to give crude product which, after puri®cation by silica
gel chromatography (hexane/EtOAc, 1/1) a€orded 26 as
a white solid (5.75 g, 98%): 1H NMR (CDCl3) d 8.7
(1H, brs), 8.1 (1H, dd, J=9.1, 1.6 Hz), 7.92±7.84 (3H,
m), 7.55 (1H, d, J=5.4 Hz), 4.89 (2H, s), 0.98 (9H, s),
0.14 (6H, s). Analysis (C18H24SiO3) C, H.
8.6 Hz), 7.92±7.88 (3H, m), 7.36±7.51 (5H, m), 5.44 (2H,
s). Analysis (C19H14O3): C, H.
Di-(tert-butyl) [(7-(benzyloxycarbonyl)naphth-2-yl)hydroxy-
methyl]phosphonate (30). To a solution of di-(tert-butyl)
phosphite (194 mg, 1.00 mmol) in anhydrous THF
(5 mL) was added a 1.0 M solution of lithium bis(-
trimethylsilyl)amide in THF (1.0 mL, 1.0 mmol) at
78 ꢀC under argon. After 30 min a solution of 29
(260 mg, 0.90 mmol) in THF (3 mL) was added and the
mixture was stirred at 78 ꢀC (2.5 h). The reaction was
quenched by addition of 1 N HCl (1 mL) and extracted
with CHCl3 (3Â10 mL). The combined organics were
washed with brine (2Â10 mL), dried (Na2SO4), and
concentrated. Residue was puri®ed by silica gel ¯ash
column chromatography (hexane/EtOAc, from 10/1 to
1/1) to yield 30 as a white solid (430 mg, 98%):
1H NMR (CDCl3) d 1.40 (9H, s), 1.42 (9H, s), 3.38 (1H,
brs), 8.62 (1H, s), 8.04 (2H, m), 7.84 (2H, m), 7.68
(1H, m), 7.48 (2H, m), 7.39 (3H, m), 5.41 (2H, s),
5.02 (1H, dd, J=10.5, 4.0 Hz). Analysis (C27H33O6P):
C, H.
7-(tert-Butyldimethylsilyoxymethyl)naphthalene-2-carb-
oxylic acid benzyl ester (27). To a solution of 26
(852 mg, 2.7 mmol) in DMF (9 mL) was added DBU
(534 mg, 3.51 mmol) and benzyl bromide (600 mg,
3.51 mmol) and the mixture stirred at room temperature
(20 h). The mixture was quenched with 1 N HCl (4 mL),
diluted with brine (10 mL), extracted with ether
(3Â50 mL), washed with brine (2Â10 mL), dried
(Na2SO4) and taken to dryness to yield crude product
which was puri®ed by silica-gel chromatography (hex-
ane/EtOAc, 99/1) to provide 27 as a white solid (1.01 g,
92%): mp 50.5±52.0 ꢀC; H NMR (CDCl3) d 8.62 (1H,
Di-(tert-butyl) [(7-(benzyloxycarbonyl)naphth-2-yl)oxo-
methyl]phosphonate (31). To a solution of oxalyl chloride
(228 mg, 1.8 mmol) in anhydrous CH2Cl2 (5 mL) was
added DMSO (281 mg, 3.6 mmol) in CH2Cl2 (1 mL) at
78 ꢀC under argon. The mixture was stirred (0.5 h)
then a solution of 30 (428 mg, 0.9 mmol) in CH2Cl2
(5 mL) was added at 78 ꢀC, and the mixture was stir-
red at 78 ꢀC (1 h), then quenched with NEt3 (911 mg,
9.0 mmol) and allowed to warm to room temperature.
The mixture was then diluted with CH2Cl2 (10 mL),
washed with brine (3Â10 mL), dried (Na2SO4), and
concentrated. The crude product was puri®ed by silica-
gel chromatography (hexane/EtOAc, 4/1) to yield 31 as
a clear thick oil (303 mg, 71%): 1H NMR (CDCl3) d
9.19 (1H, s), 8.78 (1H, s), 8.24 (2H, m), 7.92 (2H, m),
1
s), 8.06 (1H, dd, J=1.5, 8.5 Hz), 7.85 (3H, m), 7.41 (3H,
m), 5.43 (2H, s), 4.90 (2H, s), 0.96 (9H, s), 0.13 (6H, s).
Analysis (C25H30SiO3) C, H.
7-(Hydroxymethyl)naphthalene-2-carboxylic acid benzyl
ester (28). To a solution of 27 (6.66 g, 16.4 mmol) in
THF (82 mL) was added tetrabutylammonium ¯uoride,
1.0 M in THF (21.3 mL, 21.3 mmol) and the mixture
stirred at room temperature overnight, then quenched
with brine (50 mL), extracted with CHCl3 (3Â70 mL),
washed with brine (2Â20 mL), (over Na2SO4) and
taken to dryness. Puri®cation by silica gel chromato-
graphy (hexane/EtOAc, 1/1) provided 28 as a white
solid (4.44 g, 93%): mp 91.0±92.0 ꢀC; 1H NMR (CDCl3)
d 8.62 (1H, s), 8.07 (1H, dd, J=8.6, 1.6 Hz), 7.9±
7.82 (3H, m), 7.58 (1H, dd, J=8.6, 1.6 Hz), 7.5±7.34
(5H, m), 5.41 (2H, s), 4.88 (2H, brs). Analysis
(C19H16O3) C, H.
7.35±7.50 (5H, m), 5.43 (2H, s), 1.55 (18H, s). Analysis
3
.
(C27H31PO6 /4H2O): C, H.
Di-(tert-butyl) [(7-(benzyloxycarbonyl)naphth-2-yl)di¯uoro-
methyl]phosphonate (32). To a 25-mL ¯ask containing
31 (320 mg, 0.66 mmol) at 78 ꢀC under argon was
added DAST (535 mg, 3.32 mmol) via syringe, then the
reaction mixture was brought to 0 ꢀC in an ice±water
bath and allowed to come to room temperature with
stirring overnight. The mixture was recooled to 78 ꢀC,
diluted with CHCl3 (50 mL) and washed with 0.5 N
NaOH (40 mL), then extracted with CHCl3 (2Â20 mL)
and the combined organics washed with water
(2Â50 mL), dried (Na2SO4), and concentrated. Puri®ca-
tion using a 1:1 mixture of 5±25 m silica and 63±200
mesh basic alumina oxide (hexane/EtOAc, from 10/1 to
3/1) gave 32 as a solid (173 mg, 52%): mp 48.5±50.5 ꢀC;
1H NMR (CDCl3) d 8.69 (1H, s), 8.18 (1H, s), 8.16 (1H,
7-(Benzyloxycarbonyl)naphthalene-2-carboxaldehyde (29).
To
a
suspension of pyridinium chlorochromate
(634 mg, 2.94 mmol) in anhydrous CH2Cl2 (6 mL) was
added a solution of 28 (286 mg, 0.98 mmol) in anhy-
drous CH2Cl2 (10 mL) via syringe. The mixture was
stirred at room temperature (2 h), then ®ltered through a
pad of celite, and the ®lter cake was washed with
CH2Cl2 (2Â10 mL). The combined organic was con-
centrated under reduced pressure and puri®ed by silica
gel ¯ash column chromatograph, with hexane/EtOAc
mixture from 9/1 to 4/1, to give 29 as a white solid
(269 mg, 95%): mp 67±69 ꢀC; 1H NMR (CDCl3) d 10.17
(1H, s), 8.78 (1H, s), 8.43 (1H, s), 8.24 (1H, dd, J=1.7,

1810
Z.-J. Yao et al./Bioorg. Med. Chem. 6 (1998) 1799±1810
dd), 7.92 (2H, dd, J=3.0, 8.4 Hz), 7.77 (1H, d,
J=8.6 Hz), 7.30±7.52 (5H, m), 5.42 (2H, s), 1.44 (18H,
.
s). Analysis (C27H31PO5F2 /2H2O) C, H.
References and Notes
3
1. Barford, D. Curr. Opin. Struct. Biol. 1995, 5, 728.
2. Streuli, M. Curr. Opin. Cell. Biol. 1996, 8, 182.
3. Tonks, N. K.; Neel, B. G. Cell 1996, 87, 365.
4. Neel, B. G.; Tonks, N. K. Curr. Opin. Cell. Biol. 1997, 9, 193.
5. Jia, Z. C. Biochem. Cell. Biol. 1997, 75, 17.
Solid-phase synthesis of 11. To an ice-cooled solution of
32 (160 mg, 0.317 mmol) in THF (8 mL) was added
aqueous LiOH (0.2 N, 4.8 mL, 0.96 mmol) at 0 ꢀC and
the mixture was stirred at room temperature (3 h). The
mixture was diluted with EtOAc (20 mL), washed with
ice-cooled 0.2 N HCl/brine (20 mL) and the aqueous
phase back-extracted with EtOAc (2Â10 mL). The
combined organics were washed with brine (35 mL) and
dried (Na2SO4) and solvent removed under reduced
pressure. The residue was taken up in a small amount of
EtOAc, then diluted with hexane (1 mL) and solvent
decanted o€ the resulting syrup, to yield di-(tert-butyl)
[(7 - (carboxy)naphth - 2 - yl)di¯uoromethyl]phosphonate
(33) which was dried under high vacuum and utilized
directly for solid-phase synthesis. Coupling of 33 was
achieved using manual solid-phase techniques. Na-
Fmoc-l-(Og-tert-butyl)glutamicamide Rink, prepared as
indicated above (200 mg, 0.08 mmol) was washed with
NMP (3Â1 mL), then deblocked with 10% piperidine in
NMP (1 mL, 20 min) and the resin washed with NMP
(5Â1 mL). Prederivatization of 33 (33 mg, 0.08 mmol) as
its active ester was achieved by stirring with HOBT
(11 mg, 0.08 mmol) in NMP (1 mL) in the presence of
DIPCDI (11 mg, 0.08 mmol) at room temperature
(10 min) prior to coupling overnight with the deblocked
Glu-resin. The resulting resin was washed with NMP
(5Â1 mL) then CH2Cl2 (5Â1 mL) and cleaved by treat-
ment at room temperature with 1.6 mL of TFA/anisole
(95/5; 30 min). The resin was then washed with CH2Cl2
(3Â2 mL) and combined ®ltrates concentrated to yield
crude product as a pink syrup, which was 79% pure by
HPLC. This was dissolved in 2 mL of H2O/MeCN (1/1)
and puri®ed by HPLC using a Vydac Peptide and Pro-
tein C18 reverse-phase column to yield 11 as a white
6. Zhang, Z. Y. Curr. Topics Cell. Reg. 1997, 35, 21.
7. Levitzki, A.; Gazit, A. Science 1995, 267, 1782.
8. Burke, Jr. T. R.; Zhang, Z. Y. Pept. Sci., in press.
9. Zhang, Z. Y.; Maclean, D.; Mcnamara, D. J.; Sawyer, T.
K.; Dixon, J. E. Biochemistry 1994, 33, 2285.
10. Burke, Jr. T. R.; Smyth, M.; Nomizu, M.; Otaka, A.;
Roller, P. P. J. Org. Chem. 1993, 58, 1336.
11. Smyth, M. S.; Burke, Jr. T. R.; Tetrahedron Lett. 1994, 35, 551.
12. Burke, Jr. T. R.; Kole, H. K.; Roller, P. P. Biochem. Biophys.
Res. Commun. 1994, 204, 129.
13. Kole, H. K.; Smyth, M. S.; Russ, P. L.; Burke, Jr. T. R.;
Biochemical J. 1995, 311, 1025.
14. Burke, Jr. T. R.; Ye, B.; Yan, X. J.; Wang, S. M.; Jia, Z.
C.; Chen, L.; Zhang, Z. Y.; Barford, D. Biochemistry 1996, 35,
15989.
15. Zhang, Z. Y.; Thiemese¯er, A. M.; Maclean, D.; Mcna-
mara, D. J.; Dobrusin, E. M.; Sawyer, T. K.; Dixon, J. E. Proc.
Natl. Acad. Sci. U. S. A. 1993, 90, 4446.
16. Jia, Z. C.; Barford, D.; Flint, A. J.; Tonks, N. K. Science
1995, 268, 1754.
17. Smyth, M. S.; Ford, Jr. H.; Burke, Jr. T. R.; Tetrahedron
Lett. 1992, 33, 4137.
18. Qiu, W. M.; Burton, D. J. Tetrahedron Lett. 1996, 37, 2745.
19. Blades, K.; Cockerill, G. S.; Easter®eld, H. J.; Lequeux, T.
P.; Percy, J. M. Chem. Commun. 1996, 1615.
20. Taylor, S. D.; Dinaut, A. N.; Thadani, A. N.; Huang, Z.
Tetrahedron Lett. 1996, 37, 8089.
21. Rink, H. Tetrahedron Lett. 1987, 28, 3787.
22. Gordeev, M. F.; Patel , D. V.; Barker, P. L.; Gordon, E.
M. Tetrahedron Lett. 1994, 35, 7585.
1
23. Chen, L.; Wu, L.; Otaka, A.; Smyth, M. S.; Roller, P. P.;
Burke, T. R., Jr.; Denhertog, J.; Zhang, Z. Y. Biochem. Bio-
phys. Res. Commun. 1995, 216, 976.
powder (10.6 mg, 30%): H NMR (D2O) d 8.45 (1H, s),
8.27 (1H, s), 8.09 (1H, s), 8.05 (1H, s), 7.89 (1H, d,
J=8.7 Hz), 7.80 (d, 1H J=8.3 Hz), 4.60 (1H, q,
J=5.2 Hz), 2.61 (2H, t, J=7.2 Hz), 2.28 (1H, m) 2.19
24. Hippen, K. L.; Jakes, S.; Richards, J.; Jena, B. P.; Beck, B. L.;
Tabatabai, L. B.; Ingebritsen, T. S. Biochemistry 1993, 32, 12405.
(1H, m); FABMS (m/z) 429 [M H] . Analysis
5
25. Wang, Q. P.; Huang, Z.; Ramachandran, C.; Dinaut, A.
N.; Taylor, S. D. Bioorg. Med. Chem. Lett. 1998, 8 345.
.
(C17H17N2O7F2P CF3COOH /2H2O): C, H.
.
26. Chen, L.; Montserat, J.; Lawrence, D. S.; Zhang, Z. Y.
Biochemistry 1996, 35, 9349.
Acknowledgements
27. Insight II/Discover molecular modelling package by Mole-
cular Simulations, Inc.
Appreciation is expressed to Dr. James Kelley and Ms
Pamela Russ of the LMC for mass spectral analysis.
This work was partially supported by a Pilot Research
Project Award to Z.-Y. Z. From the Howard Hughes
Medical InstituteÐResearch Resources for Medical
Schools.
28. Greengard, L.; Rokhlin, V. I. J. Comp Phys. 1987, 73, 325.
29. Schmidt, K. E.; Lee, M. A. J. Stat. Phys. 1991, 63, 1223.
30. Ding, H. Q.; Darasawa, N.; Goddard, W. A., III. J. Chem.
Phys. 1992, 97, 4309.